diff --git a/.gitattributes b/.gitattributes index 0d685e229fe55f77301292d5a6c06a0002a9c07e..d961776cdf9a2a010b3fc88137aa3371a7242d27 100644 --- a/.gitattributes +++ b/.gitattributes @@ -427,3 +427,20 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text 2026.05.06/04.52.43_train_llmbc_lowdim_box-close-v2/wandb/run-20260506_045327-t6qu9x0g/run-t6qu9x0g.wandb filter=lfs diff=lfs merge=lfs -text 2026.05.06/05.04.12_train_llmbc_lowdim_box-close-v2/logs.json.txt filter=lfs diff=lfs merge=lfs -text 2026.05.06/05.04.12_train_llmbc_lowdim_box-close-v2/wandb/run-20260506_050458-k7lt6rm7/run-k7lt6rm7.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/run-ags56sz9.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/run-2o6jqedz.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt filter=lfs diff=lfs merge=lfs -text +2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094422-68a5glbf/run-68a5glbf.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/09.42.50_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt filter=lfs diff=lfs merge=lfs -text +2026.04.10/09.42.50_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094424-vk0ow3fl/run-vk0ow3fl.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/run-3qme4js3.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/run-1qgdq8b7.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_142250-821w1msn/run-821w1msn.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/14.22.41_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_142416-4mgh6loq/run-4mgh6loq.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/14.22.50_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_142252-g9z2m8kf/run-g9z2m8kf.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/18.16.49_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt filter=lfs diff=lfs merge=lfs -text +2026.04.10/18.16.49_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_181743-g6hz51in/run-g6hz51in.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/18.20.24_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt filter=lfs diff=lfs merge=lfs -text +2026.04.10/18.20.24_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_182424-q1qace9c/run-q1qace9c.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/22.25.42_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_222637-i2nrjl69/run-i2nrjl69.wandb filter=lfs diff=lfs merge=lfs -text +2026.04.10/22.29.11_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_223235-ehh65ppa/run-ehh65ppa.wandb filter=lfs diff=lfs merge=lfs -text diff --git a/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml b/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..47ce5ab8c44bb357555ae15594211533d2aaa393 --- /dev/null +++ b/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml @@ -0,0 +1,121 @@ +name: train_bc_lowdim +_target_: llmbc.workspace.train_bc_lowdim_workspace.TrainBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +policy: + _target_: llmbc.policy.bc_lowdim_policy.BCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} +dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.01 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: cosine + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 1 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: maze2d-medium-v0 + obs_dim: 6 + action_dim: 2 + env_runner: + _target_: llmbc.env_runner.pointmaze_lowdim_runner.PointmazeLowdimRunner + env_name: llf-pointmaze-maze2d-medium-v0 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 250 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.pointmaze_lowdim_dataset.PointmazeLowdimDataset + data_path: datasets/maze2d-medium-v0.pt + data_path2: datasets/maze2d-medium-v0.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.02 + dummy_normalizer: false + instructor: + _target_: llmbc.translator.instructor.pointmaze_instructor.maze2d_medium_v0_instructor.Maze2dMediumV0Instructor diff --git a/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml b/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..240e3bf8b35a7db569ec5a6afd568cd1c91b9948 --- /dev/null +++ b/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train_bc_lowdim_workspace + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: train_bc_lowdim_workspace + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/llmbc/config + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0 + choices: + task: maze2d-medium-v0 + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml b/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log b/2026.04.10/08.02.20_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml b/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..47ce5ab8c44bb357555ae15594211533d2aaa393 --- /dev/null +++ b/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml @@ -0,0 +1,121 @@ +name: train_bc_lowdim +_target_: llmbc.workspace.train_bc_lowdim_workspace.TrainBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +policy: + _target_: llmbc.policy.bc_lowdim_policy.BCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} +dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.01 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: cosine + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 1 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: maze2d-medium-v0 + obs_dim: 6 + action_dim: 2 + env_runner: + _target_: llmbc.env_runner.pointmaze_lowdim_runner.PointmazeLowdimRunner + env_name: llf-pointmaze-maze2d-medium-v0 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 250 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.pointmaze_lowdim_dataset.PointmazeLowdimDataset + data_path: datasets/maze2d-medium-v0.pt + data_path2: datasets/maze2d-medium-v0.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.02 + dummy_normalizer: false + instructor: + _target_: llmbc.translator.instructor.pointmaze_instructor.maze2d_medium_v0_instructor.Maze2dMediumV0Instructor diff --git a/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml b/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8c5fd60151b6b6d5987d41f4705079a43a6d4c11 --- /dev/null +++ b/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train_bc_lowdim_workspace + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: train_bc_lowdim_workspace + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/llmbc/config + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0 + choices: + task: maze2d-medium-v0 + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml b/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log b/2026.04.10/09.02.41_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml b/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..47ce5ab8c44bb357555ae15594211533d2aaa393 --- /dev/null +++ b/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml @@ -0,0 +1,121 @@ +name: train_bc_lowdim +_target_: llmbc.workspace.train_bc_lowdim_workspace.TrainBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +policy: + _target_: llmbc.policy.bc_lowdim_policy.BCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} +dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.01 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: cosine + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 1 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: maze2d-medium-v0 + obs_dim: 6 + action_dim: 2 + env_runner: + _target_: llmbc.env_runner.pointmaze_lowdim_runner.PointmazeLowdimRunner + env_name: llf-pointmaze-maze2d-medium-v0 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 250 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.pointmaze_lowdim_dataset.PointmazeLowdimDataset + data_path: datasets/maze2d-medium-v0.pt + data_path2: datasets/maze2d-medium-v0.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.02 + dummy_normalizer: false + instructor: + _target_: llmbc.translator.instructor.pointmaze_instructor.maze2d_medium_v0_instructor.Maze2dMediumV0Instructor diff --git a/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml b/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..baff41055b0b424f4f73995ac3f3848d446708fc --- /dev/null +++ b/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train_bc_lowdim_workspace + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: train_bc_lowdim_workspace + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/llmbc/config + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0 + choices: + task: maze2d-medium-v0 + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml b/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log b/2026.04.10/09.03.35_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml b/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..47ce5ab8c44bb357555ae15594211533d2aaa393 --- /dev/null +++ b/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml @@ -0,0 +1,121 @@ +name: train_bc_lowdim +_target_: llmbc.workspace.train_bc_lowdim_workspace.TrainBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +policy: + _target_: llmbc.policy.bc_lowdim_policy.BCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} +dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.01 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: cosine + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 1 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: maze2d-medium-v0 + obs_dim: 6 + action_dim: 2 + env_runner: + _target_: llmbc.env_runner.pointmaze_lowdim_runner.PointmazeLowdimRunner + env_name: llf-pointmaze-maze2d-medium-v0 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 250 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.pointmaze_lowdim_dataset.PointmazeLowdimDataset + data_path: datasets/maze2d-medium-v0.pt + data_path2: datasets/maze2d-medium-v0.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.02 + dummy_normalizer: false + instructor: + _target_: llmbc.translator.instructor.pointmaze_instructor.maze2d_medium_v0_instructor.Maze2dMediumV0Instructor diff --git a/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml b/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bdb5ddcb6bfed1ba6097832b78ad4cb3c822eec0 --- /dev/null +++ b/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train_bc_lowdim_workspace + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: train_bc_lowdim_workspace + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/llmbc/config + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0 + choices: + task: maze2d-medium-v0 + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml b/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log b/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log new file mode 100644 index 0000000000000000000000000000000000000000..45313b9577b94d8d18c90ef576842046d943a5af --- /dev/null +++ b/2026.04.10/09.06.09_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log @@ -0,0 +1,8 @@ +[2026-04-10 09:06:15,647][root][INFO] - running build_ext +[2026-04-10 09:06:15,649][root][INFO] - building 'mujoco_py.cymj' extension +[2026-04-10 09:06:15,649][root][INFO] - creating /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py +[2026-04-10 09:06:15,649][root][INFO] - creating /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl +[2026-04-10 09:06:15,650][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o -fopenmp -w +[2026-04-10 09:06:57,299][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -fopenmp -w +[2026-04-10 09:06:57,388][root][INFO] - creating /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/lib.linux-x86_64-cpython-39/mujoco_py +[2026-04-10 09:06:57,389][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -shared -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib -L/venv/llm-bc/lib -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -L/root/.mujoco/mujoco210/bin -Wl,--enable-new-dtags,-rpath,/root/.mujoco/mujoco210/bin -lmujoco210 -lglewosmesa -lOSMesa -lGL -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/lib.linux-x86_64-cpython-39/mujoco_py/cymj.cpython-39-x86_64-linux-gnu.so -fopenmp diff --git a/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml b/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..47ce5ab8c44bb357555ae15594211533d2aaa393 --- /dev/null +++ b/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml @@ -0,0 +1,121 @@ +name: train_bc_lowdim +_target_: llmbc.workspace.train_bc_lowdim_workspace.TrainBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +policy: + _target_: llmbc.policy.bc_lowdim_policy.BCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} +dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.01 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: cosine + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 1 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: maze2d-medium-v0 + obs_dim: 6 + action_dim: 2 + env_runner: + _target_: llmbc.env_runner.pointmaze_lowdim_runner.PointmazeLowdimRunner + env_name: llf-pointmaze-maze2d-medium-v0 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 250 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.pointmaze_lowdim_dataset.PointmazeLowdimDataset + data_path: datasets/maze2d-medium-v0.pt + data_path2: datasets/maze2d-medium-v0.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.02 + dummy_normalizer: false + instructor: + _target_: llmbc.translator.instructor.pointmaze_instructor.maze2d_medium_v0_instructor.Maze2dMediumV0Instructor diff --git a/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml b/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a2210e776c0ff0ca128a22799abae044c571ebb7 --- /dev/null +++ b/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train_bc_lowdim_workspace + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: train_bc_lowdim_workspace + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/llmbc/config + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0 + choices: + task: maze2d-medium-v0 + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml b/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log b/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log new file mode 100644 index 0000000000000000000000000000000000000000..c67ced043bd0b0012cb7e1f3e85012286820f558 --- /dev/null +++ b/2026.04.10/09.12.25_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log @@ -0,0 +1,5 @@ +[2026-04-10 09:12:28,251][root][INFO] - running build_ext +[2026-04-10 09:12:28,253][root][INFO] - building 'mujoco_py.cymj' extension +[2026-04-10 09:12:28,253][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o -fopenmp -w +[2026-04-10 09:13:10,017][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -fopenmp -w +[2026-04-10 09:13:10,101][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -shared -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib -L/venv/llm-bc/lib -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -L/root/.mujoco/mujoco210/bin -Wl,--enable-new-dtags,-rpath,/root/.mujoco/mujoco210/bin -lmujoco210 -lglewosmesa -lOSMesa -lGL -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/lib.linux-x86_64-cpython-39/mujoco_py/cymj.cpython-39-x86_64-linux-gnu.so -fopenmp diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..47ce5ab8c44bb357555ae15594211533d2aaa393 --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/.hydra/config.yaml @@ -0,0 +1,121 @@ +name: train_bc_lowdim +_target_: llmbc.workspace.train_bc_lowdim_workspace.TrainBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +policy: + _target_: llmbc.policy.bc_lowdim_policy.BCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} +dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 128 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.01 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: cosine + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 1 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: maze2d-medium-v0 + obs_dim: 6 + action_dim: 2 + env_runner: + _target_: llmbc.env_runner.pointmaze_lowdim_runner.PointmazeLowdimRunner + env_name: llf-pointmaze-maze2d-medium-v0 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 250 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.pointmaze_lowdim_dataset.PointmazeLowdimDataset + data_path: datasets/maze2d-medium-v0.pt + data_path2: datasets/maze2d-medium-v0.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.02 + dummy_normalizer: false + instructor: + _target_: llmbc.translator.instructor.pointmaze_instructor.maze2d_medium_v0_instructor.Maze2dMediumV0Instructor diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..89b0997060e0cf940d7d5a100244445bf1f76bcf --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train_bc_lowdim_workspace + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: train_bc_lowdim_workspace + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/llmbc/config + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0 + choices: + task: maze2d-medium-v0 + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/logs.json.txt b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/logs.json.txt new file mode 100644 index 0000000000000000000000000000000000000000..7dbdfb71f2eec14afef63a14d7f3f9c351c12cf5 --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/logs.json.txt @@ -0,0 +1,198 @@ +{"train_loss": 0.5742757320404053, "train_loss_bc": 0.5742757320404053, "grad_norm": 0.4610179662704468, "global_step": 0, "epoch": 0, "lr": 0.001} +{"train_loss": 0.5470483899116516, "train_loss_bc": 0.5470483899116516, "grad_norm": 0.3259114921092987, "global_step": 1, "epoch": 0, "lr": 0.002} +{"train_loss": 0.5437748432159424, "train_loss_bc": 0.5437748432159424, "grad_norm": 0.3096829354763031, "global_step": 2, "epoch": 0, "lr": 0.003} +{"train_loss": 0.5680013298988342, "train_loss_bc": 0.5680013298988342, "grad_norm": 0.8777705430984497, "global_step": 3, "epoch": 0, "lr": 0.004} +{"train_loss": 0.5333088636398315, "train_loss_bc": 0.5333088636398315, "grad_norm": 0.5981258749961853, "global_step": 4, "epoch": 0, "lr": 0.005} +{"train_loss": 0.5779598951339722, "train_loss_bc": 0.5779598951339722, "grad_norm": 0.5873940587043762, "global_step": 5, "epoch": 0, "lr": 0.006} +{"train_loss": 0.5605453252792358, "train_loss_bc": 0.5605453252792358, "grad_norm": 0.546937882900238, "global_step": 6, "epoch": 0, "lr": 0.006999999999999999} +{"train_loss": 0.5720486044883728, "train_loss_bc": 0.5720486044883728, "grad_norm": 0.6765738725662231, "global_step": 7, "epoch": 0, "lr": 0.008} +{"train_loss": 0.5544816255569458, "train_loss_bc": 0.5544816255569458, "grad_norm": 0.7473601698875427, "global_step": 8, "epoch": 0, "lr": 0.009000000000000001} +{"train_loss": 0.5382111072540283, "train_loss_bc": 0.5382111072540283, "grad_norm": 0.3417593240737915, "global_step": 9, "epoch": 0, "lr": 0.01} +{"train_loss": 0.5005457997322083, "train_loss_bc": 0.5005457997322083, "grad_norm": 0.2392510324716568, "global_step": 10, "epoch": 0, "lr": 0.009999999999378116} +{"train_loss": 0.5228367447853088, "train_loss_bc": 0.5228367447853088, "grad_norm": 0.3644765317440033, "global_step": 11, "epoch": 0, "lr": 0.009999999997512466} +{"train_loss": 0.5359384417533875, "train_loss_bc": 0.5359384417533875, "grad_norm": 0.2075331062078476, "global_step": 12, "epoch": 0, "lr": 0.009999999994403048} +{"train_loss": 0.46896055340766907, "train_loss_bc": 0.46896055340766907, "grad_norm": 0.1523733139038086, "global_step": 13, "epoch": 0, "lr": 0.009999999990049864} +{"train_loss": 0.48494070768356323, "train_loss_bc": 0.48494070768356323, "grad_norm": 0.22689184546470642, "global_step": 14, "epoch": 0, "lr": 0.009999999984452912} +{"train_loss": 0.5497490167617798, "train_loss_bc": 0.5497490167617798, "grad_norm": 0.4582449495792389, "global_step": 15, "epoch": 0, "lr": 0.009999999977612193} +{"train_loss": 0.5699339509010315, "train_loss_bc": 0.5699339509010315, "grad_norm": 0.680681049823761, "global_step": 16, "epoch": 0, "lr": 0.009999999969527707} +{"train_loss": 0.5408633947372437, "train_loss_bc": 0.5408633947372437, "grad_norm": 0.2640772759914398, "global_step": 17, "epoch": 0, "lr": 0.009999999960199455} +{"train_loss": 0.5184565186500549, "train_loss_bc": 0.5184565186500549, "grad_norm": 0.1633172631263733, "global_step": 18, "epoch": 0, "lr": 0.009999999949627435} +{"train_loss": 0.5436556935310364, "train_loss_bc": 0.5436556935310364, "grad_norm": 0.1845201849937439, "global_step": 19, "epoch": 0, "lr": 0.009999999937811649} +{"train_loss": 0.5068360567092896, "train_loss_bc": 0.5068360567092896, "grad_norm": 0.3228622078895569, "global_step": 20, "epoch": 0, "lr": 0.009999999924752095} +{"train_loss": 0.5451252460479736, "train_loss_bc": 0.5451252460479736, "grad_norm": 0.2506679892539978, "global_step": 21, "epoch": 0, "lr": 0.009999999910448773} +{"train_loss": 0.5045832395553589, "train_loss_bc": 0.5045832395553589, "grad_norm": 0.44697698950767517, "global_step": 22, "epoch": 0, "lr": 0.009999999894901686} +{"train_loss": 0.5218442678451538, "train_loss_bc": 0.5218442678451538, "grad_norm": 0.3040887713432312, "global_step": 23, "epoch": 0, "lr": 0.00999999987811083} +{"train_loss": 0.49454182386398315, "train_loss_bc": 0.49454182386398315, "grad_norm": 0.20201249420642853, "global_step": 24, "epoch": 0, "lr": 0.009999999860076207} +{"train_loss": 0.5165571570396423, "train_loss_bc": 0.5165571570396423, "grad_norm": 0.15043173730373383, "global_step": 25, "epoch": 0, "lr": 0.00999999984079782} +{"train_loss": 0.5213630199432373, "train_loss_bc": 0.5213630199432373, "grad_norm": 0.2970457673072815, "global_step": 26, "epoch": 0, "lr": 0.009999999820275662} +{"train_loss": 0.5392494201660156, "train_loss_bc": 0.5392494201660156, "grad_norm": 0.33821403980255127, "global_step": 27, "epoch": 0, "lr": 0.00999999979850974} +{"train_loss": 0.507509171962738, "train_loss_bc": 0.507509171962738, "grad_norm": 0.16721631586551666, "global_step": 28, "epoch": 0, "lr": 0.009999999775500049} +{"train_loss": 0.5158230662345886, "train_loss_bc": 0.5158230662345886, "grad_norm": 0.2735518217086792, "global_step": 29, "epoch": 0, "lr": 0.009999999751246592} +{"train_loss": 0.5095375180244446, "train_loss_bc": 0.5095375180244446, "grad_norm": 0.23328596353530884, "global_step": 30, "epoch": 0, "lr": 0.009999999725749368} +{"train_loss": 0.4946969449520111, "train_loss_bc": 0.4946969449520111, "grad_norm": 0.2920290231704712, "global_step": 31, "epoch": 0, "lr": 0.009999999699008378} +{"train_loss": 0.5286844968795776, "train_loss_bc": 0.5286844968795776, "grad_norm": 0.26879623532295227, "global_step": 32, "epoch": 0, "lr": 0.00999999967102362} +{"train_loss": 0.5065042972564697, "train_loss_bc": 0.5065042972564697, "grad_norm": 0.3493255376815796, "global_step": 33, "epoch": 0, "lr": 0.009999999641795095} +{"train_loss": 0.5176247954368591, "train_loss_bc": 0.5176247954368591, "grad_norm": 0.32611459493637085, "global_step": 34, "epoch": 0, "lr": 0.009999999611322802} +{"train_loss": 0.5595735907554626, "train_loss_bc": 0.5595735907554626, "grad_norm": 0.5473532676696777, "global_step": 35, "epoch": 0, "lr": 0.009999999579606744} +{"train_loss": 0.5184394121170044, "train_loss_bc": 0.5184394121170044, "grad_norm": 0.33806461095809937, "global_step": 36, "epoch": 0, "lr": 0.009999999546646918} +{"train_loss": 0.4891864061355591, "train_loss_bc": 0.4891864061355591, "grad_norm": 0.2983148396015167, "global_step": 37, "epoch": 0, "lr": 0.009999999512443325} +{"train_loss": 0.5090031623840332, "train_loss_bc": 0.5090031623840332, "grad_norm": 0.374706894159317, "global_step": 38, "epoch": 0, "lr": 0.009999999476995964} +{"train_loss": 0.5417673587799072, "train_loss_bc": 0.5417673587799072, "grad_norm": 0.28812533617019653, "global_step": 39, "epoch": 0, "lr": 0.009999999440304838} +{"train_loss": 0.5432811975479126, "train_loss_bc": 0.5432811975479126, "grad_norm": 0.4054281711578369, "global_step": 40, "epoch": 0, "lr": 0.009999999402369945} +{"train_loss": 0.5155792832374573, "train_loss_bc": 0.5155792832374573, "grad_norm": 0.41070905327796936, "global_step": 41, "epoch": 0, "lr": 0.009999999363191285} +{"train_loss": 0.4796150028705597, "train_loss_bc": 0.4796150028705597, "grad_norm": 0.22966867685317993, "global_step": 42, "epoch": 0, "lr": 0.009999999322768858} +{"train_loss": 0.5101842880249023, "train_loss_bc": 0.5101842880249023, "grad_norm": 0.3219834268093109, "global_step": 43, "epoch": 0, "lr": 0.009999999281102663} +{"train_loss": 0.502321720123291, "train_loss_bc": 0.502321720123291, "grad_norm": 0.3314169645309448, "global_step": 44, "epoch": 0, "lr": 0.009999999238192702} +{"train_loss": 0.4937172830104828, "train_loss_bc": 0.4937172830104828, "grad_norm": 0.17465521395206451, "global_step": 45, "epoch": 0, "lr": 0.009999999194038973} +{"train_loss": 0.5150852799415588, "train_loss_bc": 0.5150852799415588, "grad_norm": 0.30881091952323914, "global_step": 46, "epoch": 0, "lr": 0.009999999148641478} +{"train_loss": 0.46151822805404663, "train_loss_bc": 0.46151822805404663, "grad_norm": 0.19222350418567657, "global_step": 47, "epoch": 0, "lr": 0.009999999102000216} +{"train_loss": 0.5235646963119507, "train_loss_bc": 0.5235646963119507, "grad_norm": 0.17854043841362, "global_step": 48, "epoch": 0, "lr": 0.009999999054115189} +{"train_loss": 0.47330576181411743, "train_loss_bc": 0.47330576181411743, "grad_norm": 0.186621755361557, "global_step": 49, "epoch": 0, "lr": 0.009999999004986394} +{"train_loss": 0.4679132103919983, "train_loss_bc": 0.4679132103919983, "grad_norm": 0.3300033211708069, "global_step": 50, "epoch": 0, "lr": 0.009999998954613832} +{"train_loss": 0.4872625470161438, "train_loss_bc": 0.4872625470161438, "grad_norm": 0.20406848192214966, "global_step": 51, "epoch": 0, "lr": 0.009999998902997503} +{"train_loss": 0.4838597774505615, "train_loss_bc": 0.4838597774505615, "grad_norm": 0.20800356566905975, "global_step": 52, "epoch": 0, "lr": 0.009999998850137406} +{"train_loss": 0.48533451557159424, "train_loss_bc": 0.48533451557159424, "grad_norm": 0.2890176475048065, "global_step": 53, "epoch": 0, "lr": 0.009999998796033544} +{"train_loss": 0.49930649995803833, "train_loss_bc": 0.49930649995803833, "grad_norm": 0.2625446617603302, "global_step": 54, "epoch": 0, "lr": 0.009999998740685917} +{"train_loss": 0.474785178899765, "train_loss_bc": 0.474785178899765, "grad_norm": 0.2095745950937271, "global_step": 55, "epoch": 0, "lr": 0.00999999868409452} +{"train_loss": 0.46426108479499817, "train_loss_bc": 0.46426108479499817, "grad_norm": 0.23595686256885529, "global_step": 56, "epoch": 0, "lr": 0.009999998626259356} +{"train_loss": 0.4779570400714874, "train_loss_bc": 0.4779570400714874, "grad_norm": 0.19150982797145844, "global_step": 57, "epoch": 0, "lr": 0.009999998567180427} +{"train_loss": 0.5089184641838074, "train_loss_bc": 0.5089184641838074, "grad_norm": 0.2511100172996521, "global_step": 58, "epoch": 0, "lr": 0.00999999850685773} +{"train_loss": 0.4668119549751282, "train_loss_bc": 0.4668119549751282, "grad_norm": 0.1843072921037674, "global_step": 59, "epoch": 0, "lr": 0.00999999844529127} +{"train_loss": 0.5078861713409424, "train_loss_bc": 0.5078861713409424, "grad_norm": 0.29048284888267517, "global_step": 60, "epoch": 0, "lr": 0.00999999838248104} +{"train_loss": 0.4471106231212616, "train_loss_bc": 0.4471106231212616, "grad_norm": 0.24524496495723724, "global_step": 61, "epoch": 0, "lr": 0.009999998318427043} +{"train_loss": 0.4805048108100891, "train_loss_bc": 0.4805048108100891, "grad_norm": 0.21890461444854736, "global_step": 62, "epoch": 0, "lr": 0.009999998253129281} +{"train_loss": 0.47960126399993896, "train_loss_bc": 0.47960126399993896, "grad_norm": 0.2846362888813019, "global_step": 63, "epoch": 0, "lr": 0.009999998186587752} +{"train_loss": 0.4453185200691223, "train_loss_bc": 0.4453185200691223, "grad_norm": 0.2093414068222046, "global_step": 64, "epoch": 0, "lr": 0.009999998118802455} +{"train_loss": 0.47915151715278625, "train_loss_bc": 0.47915151715278625, "grad_norm": 0.25503861904144287, "global_step": 65, "epoch": 0, "lr": 0.009999998049773393} +{"train_loss": 0.4890253245830536, "train_loss_bc": 0.4890253245830536, "grad_norm": 0.42124173045158386, "global_step": 66, "epoch": 0, "lr": 0.009999997979500564} +{"train_loss": 0.5161541104316711, "train_loss_bc": 0.5161541104316711, "grad_norm": 0.28259626030921936, "global_step": 67, "epoch": 0, "lr": 0.00999999790798397} +{"train_loss": 0.46412068605422974, "train_loss_bc": 0.46412068605422974, "grad_norm": 0.26055908203125, "global_step": 68, "epoch": 0, "lr": 0.009999997835223607} +{"train_loss": 0.4562651216983795, "train_loss_bc": 0.4562651216983795, "grad_norm": 0.34777477383613586, "global_step": 69, "epoch": 0, "lr": 0.009999997761219478} +{"train_loss": 0.530666708946228, "train_loss_bc": 0.530666708946228, "grad_norm": 0.32242047786712646, "global_step": 70, "epoch": 0, "lr": 0.009999997685971582} +{"train_loss": 0.5017932057380676, "train_loss_bc": 0.5017932057380676, "grad_norm": 0.2552233040332794, "global_step": 71, "epoch": 0, "lr": 0.00999999760947992} +{"train_loss": 0.45941221714019775, "train_loss_bc": 0.45941221714019775, "grad_norm": 0.20706579089164734, "global_step": 72, "epoch": 0, "lr": 0.009999997531744494} +{"train_loss": 0.49711522459983826, "train_loss_bc": 0.49711522459983826, "grad_norm": 0.3599870800971985, "global_step": 73, "epoch": 0, "lr": 0.0099999974527653} +{"train_loss": 0.4859696328639984, "train_loss_bc": 0.4859696328639984, "grad_norm": 0.2239435315132141, "global_step": 74, "epoch": 0, "lr": 0.009999997372542337} +{"train_loss": 0.4816194176673889, "train_loss_bc": 0.4816194176673889, "grad_norm": 0.34854719042778015, "global_step": 75, "epoch": 0, "lr": 0.009999997291075612} +{"train_loss": 0.4330894649028778, "train_loss_bc": 0.4330894649028778, "grad_norm": 0.2513454854488373, "global_step": 76, "epoch": 0, "lr": 0.009999997208365117} +{"train_loss": 0.45905041694641113, "train_loss_bc": 0.45905041694641113, "grad_norm": 0.3416292071342468, "global_step": 77, "epoch": 0, "lr": 0.009999997124410858} +{"train_loss": 0.46681252121925354, "train_loss_bc": 0.46681252121925354, "grad_norm": 0.2973605990409851, "global_step": 78, "epoch": 0, "lr": 0.009999997039212831} +{"train_loss": 0.4290766716003418, "train_loss_bc": 0.4290766716003418, "grad_norm": 0.4387994706630707, "global_step": 79, "epoch": 0, "lr": 0.009999996952771038} +{"train_loss": 0.4637358784675598, "train_loss_bc": 0.4637358784675598, "grad_norm": 0.2826111614704132, "global_step": 80, "epoch": 0, "lr": 0.00999999686508548} +{"train_loss": 0.44887471199035645, "train_loss_bc": 0.44887471199035645, "grad_norm": 0.2915211021900177, "global_step": 81, "epoch": 0, "lr": 0.009999996776156154} +{"train_loss": 0.4132304787635803, "train_loss_bc": 0.4132304787635803, "grad_norm": 0.31090226769447327, "global_step": 82, "epoch": 0, "lr": 0.009999996685983063} +{"train_loss": 0.42577099800109863, "train_loss_bc": 0.42577099800109863, "grad_norm": 0.33208024501800537, "global_step": 83, "epoch": 0, "lr": 0.009999996594566206} +{"train_loss": 0.4459473788738251, "train_loss_bc": 0.4459473788738251, "grad_norm": 0.4792470932006836, "global_step": 84, "epoch": 0, "lr": 0.009999996501905582} +{"train_loss": 0.4580143094062805, "train_loss_bc": 0.4580143094062805, "grad_norm": 0.43946200609207153, "global_step": 85, "epoch": 0, "lr": 0.00999999640800119} +{"train_loss": 0.46910616755485535, "train_loss_bc": 0.46910616755485535, "grad_norm": 0.2916896641254425, "global_step": 86, "epoch": 0, "lr": 0.009999996312853034} +{"train_loss": 0.4336342513561249, "train_loss_bc": 0.4336342513561249, "grad_norm": 0.20134015381336212, "global_step": 87, "epoch": 0, "lr": 0.009999996216461112} +{"train_loss": 0.44419264793395996, "train_loss_bc": 0.44419264793395996, "grad_norm": 0.2640151083469391, "global_step": 88, "epoch": 0, "lr": 0.009999996118825424} +{"train_loss": 0.4430444538593292, "train_loss_bc": 0.4430444538593292, "grad_norm": 0.301547646522522, "global_step": 89, "epoch": 0, "lr": 0.009999996019945969} +{"train_loss": 0.43620699644088745, "train_loss_bc": 0.43620699644088745, "grad_norm": 0.34488314390182495, "global_step": 90, "epoch": 0, "lr": 0.009999995919822749} +{"train_loss": 0.4108070135116577, "train_loss_bc": 0.4108070135116577, "grad_norm": 0.3344149887561798, "global_step": 91, "epoch": 0, "lr": 0.009999995818455763} +{"train_loss": 0.4386339783668518, "train_loss_bc": 0.4386339783668518, "grad_norm": 0.2507685124874115, "global_step": 92, "epoch": 0, "lr": 0.00999999571584501} +{"train_loss": 0.43485045433044434, "train_loss_bc": 0.43485045433044434, "grad_norm": 0.3099294900894165, "global_step": 93, "epoch": 0, "lr": 0.00999999561199049} +{"train_loss": 0.462785005569458, "train_loss_bc": 0.462785005569458, "grad_norm": 0.22741125524044037, "global_step": 94, "epoch": 0, "lr": 0.009999995506892206} +{"train_loss": 0.4141497015953064, "train_loss_bc": 0.4141497015953064, "grad_norm": 0.22992584109306335, "global_step": 95, "epoch": 0, "lr": 0.009999995400550157} +{"train_loss": 0.40021058917045593, "train_loss_bc": 0.40021058917045593, "grad_norm": 0.23979420959949493, "global_step": 96, "epoch": 0, "lr": 0.00999999529296434} +{"train_loss": 0.3920753002166748, "train_loss_bc": 0.3920753002166748, "grad_norm": 0.29057979583740234, "global_step": 97, "epoch": 0, "lr": 0.009999995184134757} +{"train_loss": 0.4283364415168762, "train_loss_bc": 0.4283364415168762, "grad_norm": 0.5014083981513977, "global_step": 98, "epoch": 0, "lr": 0.00999999507406141} +{"train_loss": 0.4140693247318268, "train_loss_bc": 0.4140693247318268, "grad_norm": 0.39413923025131226, "global_step": 99, "epoch": 0, "lr": 0.009999994962744294} +{"train_loss": 0.44777703285217285, "train_loss_bc": 0.44777703285217285, "grad_norm": 0.2801749110221863, "global_step": 100, "epoch": 0, "lr": 0.009999994850183417} +{"train_loss": 0.4148218631744385, "train_loss_bc": 0.4148218631744385, "grad_norm": 0.40548500418663025, "global_step": 101, "epoch": 0, "lr": 0.00999999473637877} +{"train_loss": 0.39296433329582214, "train_loss_bc": 0.39296433329582214, "grad_norm": 0.29740381240844727, "global_step": 102, "epoch": 0, "lr": 0.009999994621330359} +{"train_loss": 0.4314180016517639, "train_loss_bc": 0.4314180016517639, "grad_norm": 0.3907894194126129, "global_step": 103, "epoch": 0, "lr": 0.009999994505038181} +{"train_loss": 0.42784494161605835, "train_loss_bc": 0.42784494161605835, "grad_norm": 0.434723824262619, "global_step": 104, "epoch": 0, "lr": 0.009999994387502239} +{"train_loss": 0.4203360080718994, "train_loss_bc": 0.4203360080718994, "grad_norm": 0.41060370206832886, "global_step": 105, "epoch": 0, "lr": 0.00999999426872253} +{"train_loss": 0.45144379138946533, "train_loss_bc": 0.45144379138946533, "grad_norm": 0.27415886521339417, "global_step": 106, "epoch": 0, "lr": 0.009999994148699056} +{"train_loss": 0.4519392251968384, "train_loss_bc": 0.4519392251968384, "grad_norm": 0.4237353801727295, "global_step": 107, "epoch": 0, "lr": 0.009999994027431817} +{"train_loss": 0.3734853267669678, "train_loss_bc": 0.3734853267669678, "grad_norm": 0.2555156350135803, "global_step": 108, "epoch": 0, "lr": 0.009999993904920812} +{"train_loss": 0.4230925738811493, "train_loss_bc": 0.4230925738811493, "grad_norm": 0.4065951704978943, "global_step": 109, "epoch": 0, "lr": 0.009999993781166042} +{"train_loss": 0.3788265585899353, "train_loss_bc": 0.3788265585899353, "grad_norm": 0.27832919359207153, "global_step": 110, "epoch": 0, "lr": 0.009999993656167506} +{"train_loss": 0.4092004895210266, "train_loss_bc": 0.4092004895210266, "grad_norm": 0.2456955760717392, "global_step": 111, "epoch": 0, "lr": 0.009999993529925205} +{"train_loss": 0.40052908658981323, "train_loss_bc": 0.40052908658981323, "grad_norm": 0.3523780405521393, "global_step": 112, "epoch": 0, "lr": 0.009999993402439136} +{"train_loss": 0.4118815064430237, "train_loss_bc": 0.4118815064430237, "grad_norm": 0.3752630352973938, "global_step": 113, "epoch": 0, "lr": 0.009999993273709304} +{"train_loss": 0.39718833565711975, "train_loss_bc": 0.39718833565711975, "grad_norm": 0.33827918767929077, "global_step": 114, "epoch": 0, "lr": 0.009999993143735706} +{"train_loss": 0.4095107316970825, "train_loss_bc": 0.4095107316970825, "grad_norm": 0.3141377568244934, "global_step": 115, "epoch": 0, "lr": 0.009999993012518343} +{"train_loss": 0.43222084641456604, "train_loss_bc": 0.43222084641456604, "grad_norm": 0.34126782417297363, "global_step": 116, "epoch": 0, "lr": 0.009999992880057214} +{"train_loss": 0.39195510745048523, "train_loss_bc": 0.39195510745048523, "grad_norm": 0.358022540807724, "global_step": 117, "epoch": 0, "lr": 0.00999999274635232} +{"train_loss": 0.4415290355682373, "train_loss_bc": 0.4415290355682373, "grad_norm": 0.3781397342681885, "global_step": 118, "epoch": 0, "lr": 0.009999992611403661} +{"train_loss": 0.40648651123046875, "train_loss_bc": 0.40648651123046875, "grad_norm": 0.35686495900154114, "global_step": 119, "epoch": 0, "lr": 0.009999992475211238} +{"train_loss": 0.4031302332878113, "train_loss_bc": 0.4031302332878113, "grad_norm": 0.3705500662326813, "global_step": 120, "epoch": 0, "lr": 0.009999992337775048} +{"train_loss": 0.45374390482902527, "train_loss_bc": 0.45374390482902527, "grad_norm": 0.2800765037536621, "global_step": 121, "epoch": 0, "lr": 0.009999992199095094} +{"train_loss": 0.41550350189208984, "train_loss_bc": 0.41550350189208984, "grad_norm": 0.3643583357334137, "global_step": 122, "epoch": 0, "lr": 0.009999992059171374} +{"train_loss": 0.42902517318725586, "train_loss_bc": 0.42902517318725586, "grad_norm": 0.2818589210510254, "global_step": 123, "epoch": 0, "lr": 0.00999999191800389} +{"train_loss": 0.3701103925704956, "train_loss_bc": 0.3701103925704956, "grad_norm": 0.28646788001060486, "global_step": 124, "epoch": 0, "lr": 0.00999999177559264} +{"train_loss": 0.4084678888320923, "train_loss_bc": 0.4084678888320923, "grad_norm": 0.3128051459789276, "global_step": 125, "epoch": 0, "lr": 0.009999991631937623} +{"train_loss": 0.43478596210479736, "train_loss_bc": 0.43478596210479736, "grad_norm": 0.2467166632413864, "global_step": 126, "epoch": 0, "lr": 0.009999991487038845} +{"train_loss": 0.4194517135620117, "train_loss_bc": 0.4194517135620117, "grad_norm": 0.1867067962884903, "global_step": 127, "epoch": 0, "lr": 0.009999991340896299} +{"train_loss": 0.40881070494651794, "train_loss_bc": 0.40881070494651794, "grad_norm": 0.4022925794124603, "global_step": 128, "epoch": 0, "lr": 0.00999999119350999} +{"train_loss": 0.41983669996261597, "train_loss_bc": 0.41983669996261597, "grad_norm": 0.27260592579841614, "global_step": 129, "epoch": 0, "lr": 0.009999991044879915} +{"train_loss": 0.4503186345100403, "train_loss_bc": 0.4503186345100403, "grad_norm": 0.40697935223579407, "global_step": 130, "epoch": 0, "lr": 0.009999990895006076} +{"train_loss": 0.4265207052230835, "train_loss_bc": 0.4265207052230835, "grad_norm": 0.2521716356277466, "global_step": 131, "epoch": 0, "lr": 0.009999990743888474} +{"train_loss": 0.4375496506690979, "train_loss_bc": 0.4375496506690979, "grad_norm": 0.27460354566574097, "global_step": 132, "epoch": 0, "lr": 0.009999990591527104} +{"train_loss": 0.4457821846008301, "train_loss_bc": 0.4457821846008301, "grad_norm": 0.24064527451992035, "global_step": 133, "epoch": 0, "lr": 0.009999990437921972} +{"train_loss": 0.4605506360530853, "train_loss_bc": 0.4605506360530853, "grad_norm": 0.2716171145439148, "global_step": 134, "epoch": 0, "lr": 0.009999990283073072} +{"train_loss": 0.38944196701049805, "train_loss_bc": 0.38944196701049805, "grad_norm": 0.30115771293640137, "global_step": 135, "epoch": 0, "lr": 0.009999990126980409} +{"train_loss": 0.41354790329933167, "train_loss_bc": 0.41354790329933167, "grad_norm": 0.33541253209114075, "global_step": 136, "epoch": 0, "lr": 0.009999989969643982} +{"train_loss": 0.37190109491348267, "train_loss_bc": 0.37190109491348267, "grad_norm": 0.19214670360088348, "global_step": 137, "epoch": 0, "lr": 0.00999998981106379} +{"train_loss": 0.4145277738571167, "train_loss_bc": 0.4145277738571167, "grad_norm": 0.3911716043949127, "global_step": 138, "epoch": 0, "lr": 0.009999989651239834} +{"train_loss": 0.43325337767601013, "train_loss_bc": 0.43325337767601013, "grad_norm": 0.3034513592720032, "global_step": 139, "epoch": 0, "lr": 0.009999989490172112} +{"train_loss": 0.36040061712265015, "train_loss_bc": 0.36040061712265015, "grad_norm": 0.2691652178764343, "global_step": 140, "epoch": 0, "lr": 0.009999989327860628} +{"train_loss": 0.3998282551765442, "train_loss_bc": 0.3998282551765442, "grad_norm": 0.2916651666164398, "global_step": 141, "epoch": 0, "lr": 0.009999989164305377} +{"train_loss": 0.4111650586128235, "train_loss_bc": 0.4111650586128235, "grad_norm": 0.27103137969970703, "global_step": 142, "epoch": 0, "lr": 0.009999988999506364} +{"train_loss": 0.36338406801223755, "train_loss_bc": 0.36338406801223755, "grad_norm": 0.2345210313796997, "global_step": 143, "epoch": 0, "lr": 0.009999988833463584} +{"train_loss": 0.3807051479816437, "train_loss_bc": 0.3807051479816437, "grad_norm": 0.2513706684112549, "global_step": 144, "epoch": 0, "lr": 0.009999988666177042} +{"train_loss": 0.4108376204967499, "train_loss_bc": 0.4108376204967499, "grad_norm": 0.323003888130188, "global_step": 145, "epoch": 0, "lr": 0.009999988497646734} +{"train_loss": 0.3920934796333313, "train_loss_bc": 0.3920934796333313, "grad_norm": 0.26234710216522217, "global_step": 146, "epoch": 0, "lr": 0.009999988327872664} +{"train_loss": 0.3924039304256439, "train_loss_bc": 0.3924039304256439, "grad_norm": 0.3997865617275238, "global_step": 147, "epoch": 0, "lr": 0.009999988156854829} +{"train_loss": 0.3870144784450531, "train_loss_bc": 0.3870144784450531, "grad_norm": 0.27861636877059937, "global_step": 148, "epoch": 0, "lr": 0.00999998798459323} +{"train_loss": 0.38959047198295593, "train_loss_bc": 0.38959047198295593, "grad_norm": 0.3393172025680542, "global_step": 149, "epoch": 0, "lr": 0.009999987811087865} +{"train_loss": 0.40116578340530396, "train_loss_bc": 0.40116578340530396, "grad_norm": 0.2975669503211975, "global_step": 150, "epoch": 0, "lr": 0.009999987636338739} +{"train_loss": 0.42044979333877563, "train_loss_bc": 0.42044979333877563, "grad_norm": 0.44693636894226074, "global_step": 151, "epoch": 0, "lr": 0.009999987460345847} +{"train_loss": 0.35583871603012085, "train_loss_bc": 0.35583871603012085, "grad_norm": 0.34289121627807617, "global_step": 152, "epoch": 0, "lr": 0.009999987283109192} +{"train_loss": 0.41687560081481934, "train_loss_bc": 0.41687560081481934, "grad_norm": 0.44672927260398865, "global_step": 153, "epoch": 0, "lr": 0.009999987104628773} +{"train_loss": 0.40855151414871216, "train_loss_bc": 0.40855151414871216, "grad_norm": 0.2894774377346039, "global_step": 154, "epoch": 0, "lr": 0.00999998692490459} +{"train_loss": 0.4261748790740967, "train_loss_bc": 0.4261748790740967, "grad_norm": 0.37578925490379333, "global_step": 155, "epoch": 0, "lr": 0.009999986743936642} +{"train_loss": 0.41958966851234436, "train_loss_bc": 0.41958966851234436, "grad_norm": 0.39029765129089355, "global_step": 156, "epoch": 0, "lr": 0.009999986561724931} +{"train_loss": 0.4157213270664215, "train_loss_bc": 0.4157213270664215, "grad_norm": 0.550552487373352, "global_step": 157, "epoch": 0, "lr": 0.009999986378269458} +{"train_loss": 0.3687168061733246, "train_loss_bc": 0.3687168061733246, "grad_norm": 0.32177433371543884, "global_step": 158, "epoch": 0, "lr": 0.00999998619357022} +{"train_loss": 0.37296295166015625, "train_loss_bc": 0.37296295166015625, "grad_norm": 0.7262616753578186, "global_step": 159, "epoch": 0, "lr": 0.009999986007627217} +{"train_loss": 0.40853071212768555, "train_loss_bc": 0.40853071212768555, "grad_norm": 0.7078126668930054, "global_step": 160, "epoch": 0, "lr": 0.009999985820440453} +{"train_loss": 0.42252951860427856, "train_loss_bc": 0.42252951860427856, "grad_norm": 0.34911733865737915, "global_step": 161, "epoch": 0, "lr": 0.009999985632009925} +{"train_loss": 0.42843812704086304, "train_loss_bc": 0.42843812704086304, "grad_norm": 0.2839755415916443, "global_step": 162, "epoch": 0, "lr": 0.009999985442335633} +{"train_loss": 0.37448471784591675, "train_loss_bc": 0.37448471784591675, "grad_norm": 0.3231787383556366, "global_step": 163, "epoch": 0, "lr": 0.009999985251417575} +{"train_loss": 0.38526156544685364, "train_loss_bc": 0.38526156544685364, "grad_norm": 0.42802849411964417, "global_step": 164, "epoch": 0, "lr": 0.009999985059255758} +{"train_loss": 0.43893128633499146, "train_loss_bc": 0.43893128633499146, "grad_norm": 0.43370962142944336, "global_step": 165, "epoch": 0, "lr": 0.009999984865850176} +{"train_loss": 0.36541861295700073, "train_loss_bc": 0.36541861295700073, "grad_norm": 0.3593718707561493, "global_step": 166, "epoch": 0, "lr": 0.009999984671200831} +{"train_loss": 0.37354975938796997, "train_loss_bc": 0.37354975938796997, "grad_norm": 0.31158536672592163, "global_step": 167, "epoch": 0, "lr": 0.00999998447530772} +{"train_loss": 0.3358023166656494, "train_loss_bc": 0.3358023166656494, "grad_norm": 0.3619420826435089, "global_step": 168, "epoch": 0, "lr": 0.00999998427817085} +{"train_loss": 0.3332926630973816, "train_loss_bc": 0.3332926630973816, "grad_norm": 0.2583524286746979, "global_step": 169, "epoch": 0, "lr": 0.009999984079790214} +{"train_loss": 0.3937195837497711, "train_loss_bc": 0.3937195837497711, "grad_norm": 0.4220624566078186, "global_step": 170, "epoch": 0, "lr": 0.009999983880165816} +{"train_loss": 0.3483230471611023, "train_loss_bc": 0.3483230471611023, "grad_norm": 0.2745370864868164, "global_step": 171, "epoch": 0, "lr": 0.009999983679297653} +{"train_loss": 0.3706860840320587, "train_loss_bc": 0.3706860840320587, "grad_norm": 0.3540709912776947, "global_step": 172, "epoch": 0, "lr": 0.00999998347718573} +{"train_loss": 0.3873794674873352, "train_loss_bc": 0.3873794674873352, "grad_norm": 0.3056386113166809, "global_step": 173, "epoch": 0, "lr": 0.009999983273830042} +{"train_loss": 0.3573632538318634, "train_loss_bc": 0.3573632538318634, "grad_norm": 0.4181464910507202, "global_step": 174, "epoch": 0, "lr": 0.009999983069230592} +{"train_loss": 0.42982178926467896, "train_loss_bc": 0.42982178926467896, "grad_norm": 0.4089113473892212, "global_step": 175, "epoch": 0, "lr": 0.00999998286338738} +{"train_loss": 0.3415180444717407, "train_loss_bc": 0.3415180444717407, "grad_norm": 0.2808316946029663, "global_step": 176, "epoch": 0, "lr": 0.009999982656300404} +{"train_loss": 0.39931759238243103, "train_loss_bc": 0.39931759238243103, "grad_norm": 0.31670093536376953, "global_step": 177, "epoch": 0, "lr": 0.009999982447969664} +{"train_loss": 0.42158377170562744, "train_loss_bc": 0.42158377170562744, "grad_norm": 0.434965580701828, "global_step": 178, "epoch": 0, "lr": 0.009999982238395164} +{"train_loss": 0.35083356499671936, "train_loss_bc": 0.35083356499671936, "grad_norm": 0.4147171676158905, "global_step": 179, "epoch": 0, "lr": 0.0099999820275769} +{"train_loss": 0.34347087144851685, "train_loss_bc": 0.34347087144851685, "grad_norm": 0.317233145236969, "global_step": 180, "epoch": 0, "lr": 0.009999981815514874} +{"train_loss": 0.3805704712867737, "train_loss_bc": 0.3805704712867737, "grad_norm": 0.2734706997871399, "global_step": 181, "epoch": 0, "lr": 0.009999981602209084} +{"train_loss": 0.3702990710735321, "train_loss_bc": 0.3702990710735321, "grad_norm": 0.36316049098968506, "global_step": 182, "epoch": 0, "lr": 0.009999981387659533} +{"train_loss": 0.35707738995552063, "train_loss_bc": 0.35707738995552063, "grad_norm": 0.30813276767730713, "global_step": 183, "epoch": 0, "lr": 0.009999981171866219} +{"train_loss": 0.36711764335632324, "train_loss_bc": 0.36711764335632324, "grad_norm": 0.38487568497657776, "global_step": 184, "epoch": 0, "lr": 0.009999980954829145} +{"train_loss": 0.34280383586883545, "train_loss_bc": 0.34280383586883545, "grad_norm": 0.277854323387146, "global_step": 185, "epoch": 0, "lr": 0.009999980736548305} +{"train_loss": 0.33399534225463867, "train_loss_bc": 0.33399534225463867, "grad_norm": 0.4292750358581543, "global_step": 186, "epoch": 0, "lr": 0.009999980517023705} +{"train_loss": 0.3977174758911133, "train_loss_bc": 0.3977174758911133, "grad_norm": 0.30483928322792053, "global_step": 187, "epoch": 0, "lr": 0.009999980296255342} +{"train_loss": 0.34298792481422424, "train_loss_bc": 0.34298792481422424, "grad_norm": 0.2111782729625702, "global_step": 188, "epoch": 0, "lr": 0.009999980074243216} +{"train_loss": 0.32223907113075256, "train_loss_bc": 0.32223907113075256, "grad_norm": 0.29651424288749695, "global_step": 189, "epoch": 0, "lr": 0.009999979850987328} +{"train_loss": 0.3656037747859955, "train_loss_bc": 0.3656037747859955, "grad_norm": 0.3420378267765045, "global_step": 190, "epoch": 0, "lr": 0.00999997962648768} +{"train_loss": 0.3618483543395996, "train_loss_bc": 0.3618483543395996, "grad_norm": 0.30036675930023193, "global_step": 191, "epoch": 0, "lr": 0.009999979400744268} +{"train_loss": 0.38904881477355957, "train_loss_bc": 0.38904881477355957, "grad_norm": 0.28105607628822327, "global_step": 192, "epoch": 0, "lr": 0.009999979173757096} +{"train_loss": 0.3561739921569824, "train_loss_bc": 0.3561739921569824, "grad_norm": 0.23257242143154144, "global_step": 193, "epoch": 0, "lr": 0.00999997894552616} +{"train_loss": 0.4250050485134125, "train_loss_bc": 0.4250050485134125, "grad_norm": 0.4045962989330292, "global_step": 194, "epoch": 0, "lr": 0.009999978716051463} +{"train_loss": 0.36044198274612427, "train_loss_bc": 0.36044198274612427, "grad_norm": 0.3454419672489166, "global_step": 195, "epoch": 0, "lr": 0.009999978485333006} +{"train_loss": 0.36355459690093994, "train_loss_bc": 0.36355459690093994, "grad_norm": 0.28779804706573486, "global_step": 196, "epoch": 0, "lr": 0.009999978253370786} +{"train_loss": 0.36438101530075073, "train_loss_bc": 0.36438101530075073, "grad_norm": 0.3752126395702362, "global_step": 197, "epoch": 0, "lr": 0.009999978020164803} diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log new file mode 100644 index 0000000000000000000000000000000000000000..7e0c19e55e8fa8feed1a911efcd3f092b0e8580e --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/train_bc_lowdim_workspace.log @@ -0,0 +1,11 @@ +[2026-04-10 09:16:09,258][root][INFO] - running build_ext +[2026-04-10 09:16:09,260][root][INFO] - building 'mujoco_py.cymj' extension +[2026-04-10 09:16:09,261][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o -fopenmp -w +[2026-04-10 09:16:51,053][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -fopenmp -w +[2026-04-10 09:16:51,130][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -shared -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib -L/venv/llm-bc/lib -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -L/root/.mujoco/mujoco210/bin -Wl,--enable-new-dtags,-rpath,/root/.mujoco/mujoco210/bin -lmujoco210 -lglewosmesa -lOSMesa -lGL -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/lib.linux-x86_64-cpython-39/mujoco_py/cymj.cpython-39-x86_64-linux-gnu.so -fopenmp +[2026-04-10 09:16:51,979][matplotlib.font_manager][INFO] - generated new fontManager +[2026-04-10 09:16:56,008][absl][INFO] - MUJOCO_GL=osmesa, attempting to import specified OpenGL backend. +[2026-04-10 09:16:56,014][absl][INFO] - MuJoCo library version is: 2.3.7 +[2026-04-10 09:16:56,122][datasets][INFO] - PyTorch version 2.2.2 available. +[2026-04-10 09:16:56,123][datasets][INFO] - TensorFlow version 2.15.1 available. +[2026-04-10 09:16:56,123][datasets][INFO] - JAX version 0.4.30 available. diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/config.yaml b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b11558d9aff0431d355c9409977bf8bd84f586f6 --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/config.yaml @@ -0,0 +1,199 @@ +_target_: + value: llmbc.workspace.train_bc_lowdim_workspace.TrainBCLowdimWorkspace +_wandb: + value: + cli_version: 0.18.6 + m: [] + python_version: 3.9.25 + t: + "1": + - 1 + - 2 + - 3 + - 5 + - 11 + - 12 + - 41 + - 49 + - 50 + - 51 + - 53 + - 55 + - 71 + - 83 + - 95 + - 100 + - 105 + "2": + - 1 + - 2 + - 3 + - 5 + - 11 + - 12 + - 41 + - 49 + - 50 + - 51 + - 53 + - 55 + - 71 + - 83 + - 95 + - 100 + - 105 + "3": + - 13 + - 15 + - 16 + - 23 + - 55 + - 61 + "4": 3.9.25 + "5": 0.18.6 + "6": 4.47.1 + "8": + - 5 + "12": 0.18.6 + "13": linux-x86_64 +action_dim: + value: 2 +checkpoint: + value: + save_last_ckpt: true + save_last_snapshot: false + topk: + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + k: 5 + mode: max + monitor_key: test_success_rate +dataloader: + value: + batch_size: 128 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true +exp_name: + value: default +horizon: + value: 1 +logging: + value: + group: null + id: null + mode: online + name: 2026.04.10-09.16.56_train_bc_lowdim_maze2d-medium-v0 + project: maze2d-medium-v0-training + resume: true + tags: + - train_bc_lowdim + - maze2d-medium-v0 + - default +multi_run: + value: + run_dir: data/outputs/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0 + wandb_name_base: 2026.04.10-09.16.56_train_bc_lowdim_maze2d-medium-v0 +n_action_steps: + value: 1 +n_latency_steps: + value: 0 +n_obs_steps: + value: 1 +name: + value: train_bc_lowdim +obs_dim: + value: 6 +optimizer: + value: + _target_: torch.optim.AdamW + betas: + - 0.95 + - 0.999 + eps: 1e-08 + lr: 0.01 + weight_decay: 1e-06 +output_dir: + value: /workspace/LLM-BC/data/outputs/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0 +past_action_visible: + value: false +policy: + value: + _target_: llmbc.policy.bc_lowdim_policy.BCLowdimPolicy + action_dim: 2 + horizon: 1 + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + activation: relu + hidden_size: + - 256 + - 256 + input_size: 6 + n_action_steps: 1 + n_obs_steps: 1 + output_size: 2 + n_action_steps: 1 + n_obs_steps: 1 + obs_dim: 6 +task: + value: + action_dim: 2 + dataset: + _target_: llmbc.dataset.pointmaze_lowdim_dataset.PointmazeLowdimDataset + data_path: datasets/maze2d-medium-v0.pt + data_path2: datasets/maze2d-medium-v0.pt + dummy_normalizer: false + horizon: 1 + obs_eef_target: true + pad_after: 0 + pad_before: 0 + use_manual_normalizer: false + val_ratio: 0.02 + env_runner: + _target_: llmbc.env_runner.pointmaze_lowdim_runner.PointmazeLowdimRunner + discount: 0.99 + env_name: llf-pointmaze-maze2d-medium-v0 + feedback_type: + - hp + - hn + - fp + instruction_type: b + max_steps: 250 + n_action_steps: 1 + n_envs: 10 + n_obs_steps: 1 + n_test: 50 + n_train: 10 + visual: false + instructor: + _target_: llmbc.translator.instructor.pointmaze_instructor.maze2d_medium_v0_instructor.Maze2dMediumV0Instructor + name: maze2d-medium-v0 + obs_dim: 6 +task_name: + value: maze2d-medium-v0 +training: + value: + checkpoint_every: 5 + debug: false + device: cuda:0 + grad_norm_clip: 0.5 + gradient_accumulate_every: 1 + lr_scheduler: cosine + lr_warmup_steps: 10 + max_train_steps: null + max_val_steps: null + num_epochs: 1001 + resume: false + rollout_every: 5 + sample_every: 5 + sample_max_batch: 128 + seed: 42 + tqdm_interval_sec: 1 + val_every: 1 +val_dataloader: + value: + batch_size: 128 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/output.log b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/output.log new file mode 100644 index 0000000000000000000000000000000000000000..317badd6032cc4e896af5274388f7326bad2deb5 --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/output.log @@ -0,0 +1,60 @@ +/venv/llm-bc/lib/python3.9/site-packages/gym/envs/registration.py:505: UserWarning: WARN: The environment maze2d-medium-v0 is out of date. You should consider upgrading to version `v1` with the environment ID `maze2d-medium-v1`. + logger.warn( +Eval PointmazeLowdimRunner 1/6: 0%| | 0/250 [00:00 + main() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/main.py", line 90, in decorated_main + _run_hydra( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 389, in _run_hydra + _run_app( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 452, in _run_app + run_and_report( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 213, in run_and_report + return func() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 453, in + lambda: hydra.run( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/hydra.py", line 119, in run + ret = run_job( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/core/utils.py", line 186, in run_job + ret.return_value = task_function(task_cfg) + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 280, in main + workspace.run() + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 199, in run + runner_log = env_runner.run(policy) + File "/workspace/LLM-BC/llmbc/env_runner/pointmaze_lowdim_runner.py", line 149, in run + obs = obs_converter(observation, visual=self.visual) + File "/workspace/LLM-BC/llmbc/common/llfbench_util.py", line 10, in obs_converter + ob = obs_converter(obs_v, visual, obs_image) + File "/workspace/LLM-BC/llmbc/common/llfbench_util.py", line 39, in obs_converter + obs = torch.tensor(obs, dtype=torch.float32).unsqueeze(dim=0).to(device) +KeyboardInterrupt +Traceback (most recent call last): + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 283, in + main() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/main.py", line 90, in decorated_main + _run_hydra( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 389, in _run_hydra + _run_app( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 452, in _run_app + run_and_report( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 213, in run_and_report + return func() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 453, in + lambda: hydra.run( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/hydra.py", line 119, in run + ret = run_job( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/core/utils.py", line 186, in run_job + ret.return_value = task_function(task_cfg) + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 280, in main + workspace.run() + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 199, in run + runner_log = env_runner.run(policy) + File "/workspace/LLM-BC/llmbc/env_runner/pointmaze_lowdim_runner.py", line 149, in run + obs = obs_converter(observation, visual=self.visual) + File "/workspace/LLM-BC/llmbc/common/llfbench_util.py", line 10, in obs_converter + ob = obs_converter(obs_v, visual, obs_image) + File "/workspace/LLM-BC/llmbc/common/llfbench_util.py", line 39, in obs_converter + obs = torch.tensor(obs, dtype=torch.float32).unsqueeze(dim=0).to(device) +KeyboardInterrupt diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/requirements.txt b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/requirements.txt new file mode 100644 index 0000000000000000000000000000000000000000..f419b8629a9eb0bb659f63d4d8c14d64affc8311 --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/requirements.txt @@ -0,0 +1,860 @@ +aiohappyeyeballs==2.6.1 +antlr4-python3-runtime==4.9.3 +asciitree==0.3.3 +async-timeout==5.0.1 +attrs==25.4.0 +certifi==2025.10.5 +charset-normalizer==3.4.4 +colorama==0.4.6 +dill==0.3.8 +docstring-parser==0.16 +eval_type_backport==0.2.2 +fasteners==0.20 +filelock==3.17.0 +frozenlist==1.8.0 +fsspec==2024.3.1 +gmpy2==2.2.1 +hf-xet==1.1.8 +idna==3.11 +llvmlite==0.39.1 +MarkupSafe==3.0.2 +mdurl==0.1.2 +mkl-service==2.4.0 +mpmath==1.3.0 +msgpack==1.1.1 +networkx==3.2.1 +packaging==25.0 +propcache==0.3.1 +psutil==7.0.0 +pycparser==2.23 +Pygments==2.19.1 +PyOpenGL==3.1.10 +PySocks==1.7.1 +tzdata==2025.3 +xxhash==3.5.0 +pytz==2025.2 +PyYAML==6.0.2 +regex==2025.9.1 +safetensors==0.5.3 +setuptools==80.9.0 +shtab==1.7.2 +six==1.17.0 +tqdm==4.67.1 +tensorboard==2.15.2 +uvloop==0.22.1 +wheel==0.45.1 +zipp==3.21.0 +tensorflow-estimator==2.15.0 +tensorboard-data-server==0.7.2 +google-auth-oauthlib==1.3.1 +tensorflow==2.15.1 +Jinja2==3.1.6 +markdown-it-py==2.2.0 +multiprocess==0.70.15 +numpy==1.23.5 +pillow==11.3.0 +pip==23.3.2 +python-dateutil==2.9.0.post0 +aiosignal==1.4.0 +brotlicffi==1.0.9.2 +hydra-core==1.2.0 +multidict==6.7.0 +pycares==4.10.0 +rich==14.2.0 +sympy==1.14.0 +typeguard==4.4.4 +aiodns==3.5.0 +urllib3==2.5.0 +yarl==1.18.0 +aiohttp==3.12.15 +pyarrow==21.0.0 +pyarrow-hotfix==0.7 +tokenizers==0.21.0 +av==10.0.0 +einops==0.4.1 +Bottleneck==1.4.2 +mkl_fft==1.3.11 +mkl_random==1.2.8 +numba==0.56.4 +numcodecs==0.12.1 +numexpr==2.10.1 +pandas==2.3.3 +zarr==2.12.0 +datasets==2.19.0 +transformers==4.47.1 +accelerate==1.0.1 +peft==0.14.0 +trl==0.11.4 +torch==2.2.2 +tabulate==0.9.0 +sentencepiece==0.2.0 +pyperclip==1.11.0 +PyOpenGL-accelerate==3.1.10 +pydub==0.25.1 +py-cpuinfo==9.0.0 +pure_eval==0.2.3 +ptyprocess==0.7.0 +nvidia-ml-py==13.595.45 +hjson==3.1.0 +gym-notices==0.1.0 +glfw==2.0.0 +Farama-Notifications==0.0.4 +websockets==15.0.1 +wcwidth==0.6.0 +uc-micro-py==1.0.3 +typing-inspection==0.4.2 +triton==2.2.0 +trimesh==4.11.5 +transforms3d==0.4.2 +traitlets==5.14.3 +threadpoolctl==3.6.0 +tenacity==9.1.2 +sniffio==1.3.1 +smmap==5.0.3 +setproctitle==1.3.7 +sentry-sdk==2.57.0 +semantic-version==2.10.0 +scipy==1.13.1 +rpds-py==0.27.1 +python-multipart==0.0.20 +pyparsing==3.3.2 +pynvml==13.0.1 +pydantic_core==2.41.5 +pycryptodomex==3.23.0 +pyasn1==0.6.3 +grpcio==1.80.0 +platformdirs==4.4.0 +pexpect==4.9.0 +parso==0.8.6 +orjson==3.11.5 +opencv-python==4.11.0.86 +stack-data==0.6.3 +omegaconf==2.2.1 +nvidia-nvtx-cu12==12.1.105 +nvidia-nvjitlink-cu12==12.9.86 +nvidia-nccl-cu12==2.19.3 +nvidia-curand-cu12==10.3.2.106 +nvidia-cufft-cu12==11.0.2.54 +nvidia-cuda-runtime-cu12==12.1.105 +nvidia-cuda-nvrtc-cu12==12.1.105 +nvidia-cuda-cupti-cu12==12.1.105 +nvidia-cublas-cu12==12.1.3.1 +ninja==1.13.0 +narwhals==2.19.0 +lxml==6.0.3 +kiwisolver==1.4.7 +joblib==1.5.3 +jiter==0.13.0 +imageio-ffmpeg==0.6.0 +ImageIO==2.37.2 +h5py==3.14.0 +h11==0.16.0 +fonttools==4.60.2 +ffmpy==1.0.0 +fast_kinematics==0.2.2 +executing==2.2.1 +exceptiongroup==1.3.1 +docker-pycreds==0.4.0 +distro==1.9.0 +decorator==5.2.1 +dacite==1.9.2 +cycler==0.12.1 +contourpy==1.3.0 +cloudpickle==3.1.2 +click==8.1.8 +asttokens==3.0.1 +annotated-types==0.7.0 +annotated-doc==0.0.4 +aiofiles==25.1.0 +absl-py==2.3.1 +uvicorn==0.39.0 +tiktoken==0.8.0 +scikit-learn==1.6.1 +sapien==3.0.0b1 +referencing==0.36.2 +pydantic==2.12.5 +pyasn1_modules==0.4.2 +prompt_toolkit==3.0.52 +nvidia-cusparse-cu12==12.1.0.106 +nvidia-cudnn-cu12==8.9.2.26 +nltk==3.9.2 +mujoco==2.3.7 +mdit-py-plugins==0.3.3 +matplotlib-inline==0.2.1 +matplotlib==3.7.5 +linkify-it-py==2.0.3 +jedi==0.19.2 +huggingface-hub==0.26.2 +httpcore==1.0.9 +gymnasium==0.29.1 +google-pasta==0.2.0 +gitdb==4.0.12 +cryptography==46.0.7 +blobfile==3.0.0 +anyio==4.12.1 +llmbc==0.0.0 +tyro==0.9.1 +toppra==0.6.6 +starlette==0.49.3 +seaborn==0.13.2 +rouge_score==0.1.2 +pynndescent==0.6.0 +nvidia-cusolver-cu12==11.4.5.107 +jsonschema-specifications==2025.9.1 +ipython==8.18.1 +httpx==0.28.1 +gymnasium-robotics==1.2.0 +google-auth==2.49.2 +GitPython==3.1.46 +diffusers==0.31.0 +wandb==0.18.6 +umap-learn==0.5.9.post2 +requests-oauthlib==2.0.0 +openai==2.8.1 +mplib==0.1.1 +jsonschema==4.25.1 +gradio_client==0.2.9 +google-genai==1.47.0 +fastapi==0.128.8 +torchvision==0.17.2 +torchaudio==2.2.2 +sentence-transformers==3.2.1 +pytorch-seed==0.2.0 +deepspeed==0.16.1 +bitsandbytes==0.45.0 +altair==6.0.0 +timm==1.0.22 +gradio==3.36.1 +evaluate==0.4.3 +arm_pytorch_utilities==0.5.0 +tensorflow-io-gcs-filesystem==0.37.1 +Markdown==3.9 +pybullet==3.2.6 +wrapt==1.14.2 +tensorflow-probability==0.23.0 +parse==1.19.1 +dm-tree==0.1.8 +termcolor==3.1.0 +requests==2.32.0 +cffi==2.0.0 +opt_einsum==3.4.0 +labmaze==1.0.6 +typing_extensions==4.15.0 +importlib-resources==5.13.0 +importlib-metadata==5.2.0 +FLAML==2.3.6 +dm-env==1.6 +diskcache==5.6.3 +Cython==0.29.37 +mujoco-py==2.1.2.14 +jaxlib==0.4.30 +dm-control==1.0.14 +cmudict==1.0.13 +syllables==1.0.9 +metaworld==2.0.0 +jax==0.4.30 +highway-env==1.9.1 +gym_bandits==0.0.2 +d4rl==1.1 +pyautogen==0.1.0 +protobuf==4.25.9 +tf-agents==0.19.0 +oauthlib==3.3.1 +pytorch-kinematics==0.7.5 +pygame==2.6.1 +gin-config==0.5.0 +mani_skill==3.0.0b20 +tifffile==2024.8.30 +Shapely==1.8.4 +PyWavelets==1.6.0 +Werkzeug==3.1.8 +astunparse==1.6.3 +scikit-image==0.19.3 +pymunk==6.2.1 +ml-dtypes==0.3.2 +gast==0.7.0 +stable-baselines3==2.2.1 +keras==2.15.0 +llfbench==0.1.0 +libclang==18.1.1 +flatbuffers==25.12.19 +gym==0.23.1 +aiohappyeyeballs==2.6.1 +antlr4-python3-runtime==4.9.3 +asciitree==0.3.3 +async-timeout==5.0.1 +attrs==25.4.0 +certifi==2025.10.5 +charset-normalizer==3.4.4 +colorama==0.4.6 +dill==0.3.8 +docstring-parser==0.16 +eval_type_backport==0.2.2 +fasteners==0.20 +filelock==3.17.0 +frozenlist==1.8.0 +fsspec==2024.3.1 +gmpy2==2.2.1 +hf-xet==1.1.8 +idna==3.11 +llvmlite==0.39.1 +MarkupSafe==3.0.2 +mdurl==0.1.2 +mkl-service==2.4.0 +mpmath==1.3.0 +msgpack==1.1.1 +networkx==3.2.1 +packaging==25.0 +propcache==0.3.1 +psutil==7.0.0 +pycparser==2.23 +Pygments==2.19.1 +PyOpenGL==3.1.10 +PySocks==1.7.1 +tzdata==2025.3 +xxhash==3.5.0 +pytz==2025.2 +PyYAML==6.0.2 +regex==2025.9.1 +safetensors==0.5.3 +setuptools==80.9.0 +shtab==1.7.2 +six==1.17.0 +tqdm==4.67.1 +tensorboard==2.15.2 +uvloop==0.22.1 +wheel==0.45.1 +zipp==3.21.0 +tensorflow-estimator==2.15.0 +tensorboard-data-server==0.7.2 +google-auth-oauthlib==1.3.1 +tensorflow==2.15.1 +Jinja2==3.1.6 +markdown-it-py==2.2.0 +multiprocess==0.70.15 +numpy==1.23.5 +pillow==11.3.0 +pip==23.3.2 +python-dateutil==2.9.0.post0 +aiosignal==1.4.0 +brotlicffi==1.0.9.2 +hydra-core==1.2.0 +multidict==6.7.0 +pycares==4.10.0 +rich==14.2.0 +sympy==1.14.0 +typeguard==4.4.4 +aiodns==3.5.0 +urllib3==2.5.0 +yarl==1.18.0 +aiohttp==3.12.15 +pyarrow==21.0.0 +pyarrow-hotfix==0.7 +tokenizers==0.21.0 +av==10.0.0 +einops==0.4.1 +Bottleneck==1.4.2 +mkl_fft==1.3.11 +mkl_random==1.2.8 +numba==0.56.4 +numcodecs==0.12.1 +numexpr==2.10.1 +pandas==2.3.3 +zarr==2.12.0 +datasets==2.19.0 +transformers==4.47.1 +accelerate==1.0.1 +peft==0.14.0 +trl==0.11.4 +torch==2.2.2 +tabulate==0.9.0 +sentencepiece==0.2.0 +pyperclip==1.11.0 +PyOpenGL-accelerate==3.1.10 +pydub==0.25.1 +py-cpuinfo==9.0.0 +pure_eval==0.2.3 +ptyprocess==0.7.0 +nvidia-ml-py==13.595.45 +hjson==3.1.0 +gym-notices==0.1.0 +glfw==2.0.0 +Farama-Notifications==0.0.4 +websockets==15.0.1 +wcwidth==0.6.0 +uc-micro-py==1.0.3 +typing-inspection==0.4.2 +triton==2.2.0 +trimesh==4.11.5 +transforms3d==0.4.2 +traitlets==5.14.3 +threadpoolctl==3.6.0 +tenacity==9.1.2 +sniffio==1.3.1 +smmap==5.0.3 +setproctitle==1.3.7 +sentry-sdk==2.57.0 +semantic-version==2.10.0 +scipy==1.13.1 +rpds-py==0.27.1 +python-multipart==0.0.20 +pyparsing==3.3.2 +pynvml==13.0.1 +pydantic_core==2.41.5 +pycryptodomex==3.23.0 +pyasn1==0.6.3 +grpcio==1.80.0 +platformdirs==4.4.0 +pexpect==4.9.0 +parso==0.8.6 +orjson==3.11.5 +opencv-python==4.11.0.86 +stack-data==0.6.3 +omegaconf==2.2.1 +nvidia-nvtx-cu12==12.1.105 +nvidia-nvjitlink-cu12==12.9.86 +nvidia-nccl-cu12==2.19.3 +nvidia-curand-cu12==10.3.2.106 +nvidia-cufft-cu12==11.0.2.54 +nvidia-cuda-runtime-cu12==12.1.105 +nvidia-cuda-nvrtc-cu12==12.1.105 +nvidia-cuda-cupti-cu12==12.1.105 +nvidia-cublas-cu12==12.1.3.1 +ninja==1.13.0 +narwhals==2.19.0 +lxml==6.0.3 +kiwisolver==1.4.7 +joblib==1.5.3 +jiter==0.13.0 +imageio-ffmpeg==0.6.0 +ImageIO==2.37.2 +h5py==3.14.0 +h11==0.16.0 +fonttools==4.60.2 +ffmpy==1.0.0 +fast_kinematics==0.2.2 +executing==2.2.1 +exceptiongroup==1.3.1 +docker-pycreds==0.4.0 +distro==1.9.0 +decorator==5.2.1 +dacite==1.9.2 +cycler==0.12.1 +contourpy==1.3.0 +cloudpickle==3.1.2 +click==8.1.8 +asttokens==3.0.1 +annotated-types==0.7.0 +annotated-doc==0.0.4 +aiofiles==25.1.0 +absl-py==2.3.1 +uvicorn==0.39.0 +tiktoken==0.8.0 +scikit-learn==1.6.1 +sapien==3.0.0b1 +referencing==0.36.2 +pydantic==2.12.5 +pyasn1_modules==0.4.2 +prompt_toolkit==3.0.52 +nvidia-cusparse-cu12==12.1.0.106 +nvidia-cudnn-cu12==8.9.2.26 +nltk==3.9.2 +mujoco==2.3.7 +mdit-py-plugins==0.3.3 +matplotlib-inline==0.2.1 +matplotlib==3.7.5 +linkify-it-py==2.0.3 +jedi==0.19.2 +huggingface-hub==0.26.2 +httpcore==1.0.9 +gymnasium==0.29.1 +google-pasta==0.2.0 +gitdb==4.0.12 +cryptography==46.0.7 +blobfile==3.0.0 +anyio==4.12.1 +llmbc==0.0.0 +tyro==0.9.1 +toppra==0.6.6 +starlette==0.49.3 +seaborn==0.13.2 +rouge_score==0.1.2 +pynndescent==0.6.0 +nvidia-cusolver-cu12==11.4.5.107 +jsonschema-specifications==2025.9.1 +ipython==8.18.1 +httpx==0.28.1 +gymnasium-robotics==1.2.0 +google-auth==2.49.2 +GitPython==3.1.46 +diffusers==0.31.0 +wandb==0.18.6 +umap-learn==0.5.9.post2 +requests-oauthlib==2.0.0 +openai==2.8.1 +mplib==0.1.1 +jsonschema==4.25.1 +gradio_client==0.2.9 +google-genai==1.47.0 +fastapi==0.128.8 +torchvision==0.17.2 +torchaudio==2.2.2 +sentence-transformers==3.2.1 +pytorch-seed==0.2.0 +deepspeed==0.16.1 +bitsandbytes==0.45.0 +altair==6.0.0 +timm==1.0.22 +gradio==3.36.1 +evaluate==0.4.3 +arm_pytorch_utilities==0.5.0 +tensorflow-io-gcs-filesystem==0.37.1 +Markdown==3.9 +pybullet==3.2.6 +wrapt==1.14.2 +tensorflow-probability==0.23.0 +parse==1.19.1 +dm-tree==0.1.8 +termcolor==3.1.0 +requests==2.32.0 +cffi==2.0.0 +opt_einsum==3.4.0 +labmaze==1.0.6 +typing_extensions==4.15.0 +importlib-resources==5.13.0 +importlib-metadata==5.2.0 +FLAML==2.3.6 +dm-env==1.6 +diskcache==5.6.3 +Cython==0.29.37 +mujoco-py==2.1.2.14 +jaxlib==0.4.30 +dm-control==1.0.14 +cmudict==1.0.13 +syllables==1.0.9 +metaworld==2.0.0 +jax==0.4.30 +highway-env==1.9.1 +gym_bandits==0.0.2 +d4rl==1.1 +pyautogen==0.1.0 +protobuf==4.25.9 +tf-agents==0.19.0 +oauthlib==3.3.1 +pytorch-kinematics==0.7.5 +pygame==2.6.1 +gin-config==0.5.0 +mani_skill==3.0.0b20 +tifffile==2024.8.30 +Shapely==1.8.4 +PyWavelets==1.6.0 +Werkzeug==3.1.8 +astunparse==1.6.3 +scikit-image==0.19.3 +pymunk==6.2.1 +ml-dtypes==0.3.2 +gast==0.7.0 +stable-baselines3==2.2.1 +keras==2.15.0 +llfbench==0.1.0 +libclang==18.1.1 +flatbuffers==25.12.19 +gym==0.23.1 +aiohappyeyeballs==2.6.1 +antlr4-python3-runtime==4.9.3 +asciitree==0.3.3 +async-timeout==5.0.1 +attrs==25.4.0 +certifi==2025.10.5 +charset-normalizer==3.4.4 +colorama==0.4.6 +dill==0.3.8 +docstring-parser==0.16 +eval_type_backport==0.2.2 +fasteners==0.20 +filelock==3.17.0 +frozenlist==1.8.0 +fsspec==2024.3.1 +gmpy2==2.2.1 +hf-xet==1.1.8 +idna==3.11 +llvmlite==0.39.1 +MarkupSafe==3.0.2 +mdurl==0.1.2 +mkl-service==2.4.0 +mpmath==1.3.0 +msgpack==1.1.1 +networkx==3.2.1 +packaging==25.0 +propcache==0.3.1 +psutil==7.0.0 +pycparser==2.23 +Pygments==2.19.1 +PyOpenGL==3.1.10 +PySocks==1.7.1 +tzdata==2025.3 +xxhash==3.5.0 +pytz==2025.2 +PyYAML==6.0.2 +regex==2025.9.1 +safetensors==0.5.3 +setuptools==80.9.0 +shtab==1.7.2 +six==1.17.0 +tqdm==4.67.1 +tensorboard==2.15.2 +uvloop==0.22.1 +wheel==0.45.1 +zipp==3.21.0 +tensorflow-estimator==2.15.0 +tensorboard-data-server==0.7.2 +google-auth-oauthlib==1.3.1 +tensorflow==2.15.1 +Jinja2==3.1.6 +markdown-it-py==2.2.0 +multiprocess==0.70.15 +numpy==1.23.5 +pillow==11.3.0 +pip==23.3.2 +python-dateutil==2.9.0.post0 +aiosignal==1.4.0 +brotlicffi==1.0.9.2 +hydra-core==1.2.0 +multidict==6.7.0 +pycares==4.10.0 +rich==14.2.0 +sympy==1.14.0 +typeguard==4.4.4 +aiodns==3.5.0 +urllib3==2.5.0 +yarl==1.18.0 +aiohttp==3.12.15 +pyarrow==21.0.0 +pyarrow-hotfix==0.7 +tokenizers==0.21.0 +av==10.0.0 +einops==0.4.1 +Bottleneck==1.4.2 +mkl_fft==1.3.11 +mkl_random==1.2.8 +numba==0.56.4 +numcodecs==0.12.1 +numexpr==2.10.1 +pandas==2.3.3 +zarr==2.12.0 +datasets==2.19.0 +transformers==4.47.1 +accelerate==1.0.1 +peft==0.14.0 +trl==0.11.4 +torch==2.2.2 +tabulate==0.9.0 +sentencepiece==0.2.0 +pyperclip==1.11.0 +PyOpenGL-accelerate==3.1.10 +pydub==0.25.1 +py-cpuinfo==9.0.0 +pure_eval==0.2.3 +ptyprocess==0.7.0 +nvidia-ml-py==13.595.45 +hjson==3.1.0 +gym-notices==0.1.0 +glfw==2.0.0 +Farama-Notifications==0.0.4 +websockets==15.0.1 +wcwidth==0.6.0 +uc-micro-py==1.0.3 +typing-inspection==0.4.2 +triton==2.2.0 +trimesh==4.11.5 +transforms3d==0.4.2 +traitlets==5.14.3 +threadpoolctl==3.6.0 +tenacity==9.1.2 +sniffio==1.3.1 +smmap==5.0.3 +setproctitle==1.3.7 +sentry-sdk==2.57.0 +semantic-version==2.10.0 +scipy==1.13.1 +rpds-py==0.27.1 +python-multipart==0.0.20 +pyparsing==3.3.2 +pynvml==13.0.1 +pydantic_core==2.41.5 +pycryptodomex==3.23.0 +pyasn1==0.6.3 +grpcio==1.80.0 +platformdirs==4.4.0 +pexpect==4.9.0 +parso==0.8.6 +orjson==3.11.5 +opencv-python==4.11.0.86 +stack-data==0.6.3 +omegaconf==2.2.1 +nvidia-nvtx-cu12==12.1.105 +nvidia-nvjitlink-cu12==12.9.86 +nvidia-nccl-cu12==2.19.3 +nvidia-curand-cu12==10.3.2.106 +nvidia-cufft-cu12==11.0.2.54 +nvidia-cuda-runtime-cu12==12.1.105 +nvidia-cuda-nvrtc-cu12==12.1.105 +nvidia-cuda-cupti-cu12==12.1.105 +nvidia-cublas-cu12==12.1.3.1 +ninja==1.13.0 +narwhals==2.19.0 +lxml==6.0.3 +kiwisolver==1.4.7 +joblib==1.5.3 +jiter==0.13.0 +imageio-ffmpeg==0.6.0 +ImageIO==2.37.2 +h5py==3.14.0 +h11==0.16.0 +fonttools==4.60.2 +ffmpy==1.0.0 +fast_kinematics==0.2.2 +executing==2.2.1 +exceptiongroup==1.3.1 +docker-pycreds==0.4.0 +distro==1.9.0 +decorator==5.2.1 +dacite==1.9.2 +cycler==0.12.1 +contourpy==1.3.0 +cloudpickle==3.1.2 +click==8.1.8 +asttokens==3.0.1 +annotated-types==0.7.0 +annotated-doc==0.0.4 +aiofiles==25.1.0 +absl-py==2.3.1 +uvicorn==0.39.0 +tiktoken==0.8.0 +scikit-learn==1.6.1 +sapien==3.0.0b1 +referencing==0.36.2 +pydantic==2.12.5 +pyasn1_modules==0.4.2 +prompt_toolkit==3.0.52 +nvidia-cusparse-cu12==12.1.0.106 +nvidia-cudnn-cu12==8.9.2.26 +nltk==3.9.2 +mujoco==2.3.7 +mdit-py-plugins==0.3.3 +matplotlib-inline==0.2.1 +matplotlib==3.7.5 +linkify-it-py==2.0.3 +jedi==0.19.2 +huggingface-hub==0.26.2 +httpcore==1.0.9 +gymnasium==0.29.1 +google-pasta==0.2.0 +gitdb==4.0.12 +cryptography==46.0.7 +blobfile==3.0.0 +anyio==4.12.1 +llmbc==0.0.0 +tyro==0.9.1 +toppra==0.6.6 +starlette==0.49.3 +seaborn==0.13.2 +rouge_score==0.1.2 +pynndescent==0.6.0 +nvidia-cusolver-cu12==11.4.5.107 +jsonschema-specifications==2025.9.1 +ipython==8.18.1 +httpx==0.28.1 +gymnasium-robotics==1.2.0 +google-auth==2.49.2 +GitPython==3.1.46 +diffusers==0.31.0 +wandb==0.18.6 +umap-learn==0.5.9.post2 +requests-oauthlib==2.0.0 +openai==2.8.1 +mplib==0.1.1 +jsonschema==4.25.1 +gradio_client==0.2.9 +google-genai==1.47.0 +fastapi==0.128.8 +torchvision==0.17.2 +torchaudio==2.2.2 +sentence-transformers==3.2.1 +pytorch-seed==0.2.0 +deepspeed==0.16.1 +bitsandbytes==0.45.0 +altair==6.0.0 +timm==1.0.22 +gradio==3.36.1 +evaluate==0.4.3 +arm_pytorch_utilities==0.5.0 +tensorflow-io-gcs-filesystem==0.37.1 +Markdown==3.9 +pybullet==3.2.6 +wrapt==1.14.2 +tensorflow-probability==0.23.0 +parse==1.19.1 +dm-tree==0.1.8 +termcolor==3.1.0 +requests==2.32.0 +cffi==2.0.0 +opt_einsum==3.4.0 +labmaze==1.0.6 +typing_extensions==4.15.0 +importlib-resources==5.13.0 +importlib-metadata==5.2.0 +FLAML==2.3.6 +dm-env==1.6 +diskcache==5.6.3 +Cython==0.29.37 +mujoco-py==2.1.2.14 +jaxlib==0.4.30 +dm-control==1.0.14 +cmudict==1.0.13 +syllables==1.0.9 +metaworld==2.0.0 +jax==0.4.30 +highway-env==1.9.1 +gym_bandits==0.0.2 +d4rl==1.1 +pyautogen==0.1.0 +protobuf==4.25.9 +tf-agents==0.19.0 +oauthlib==3.3.1 +pytorch-kinematics==0.7.5 +pygame==2.6.1 +gin-config==0.5.0 +mani_skill==3.0.0b20 +tifffile==2024.8.30 +Shapely==1.8.4 +PyWavelets==1.6.0 +Werkzeug==3.1.8 +astunparse==1.6.3 +scikit-image==0.19.3 +pymunk==6.2.1 +ml-dtypes==0.3.2 +gast==0.7.0 +stable-baselines3==2.2.1 +keras==2.15.0 +llfbench==0.1.0 +libclang==18.1.1 +flatbuffers==25.12.19 +gym==0.23.1 +llmbc==0.0.0 +autocommand==2.2.2 +backports.tarfile==1.2.0 +importlib_metadata==8.0.0 +inflect==7.3.1 +jaraco.collections==5.1.0 +jaraco.context==5.3.0 +jaraco.functools==4.0.1 +jaraco.text==3.12.1 +more-itertools==10.3.0 +packaging==24.2 +platformdirs==4.2.2 +tomli==2.0.1 +typeguard==4.3.0 +typing_extensions==4.12.2 +wheel==0.45.1 +zipp==3.19.2 diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/wandb-metadata.json b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/wandb-metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..208fdb6640afa57a5e848962d8a63ffe365cb19e --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/wandb-metadata.json @@ -0,0 +1,49 @@ +{ + "os": "Linux-5.15.0-168-generic-x86_64-with-glibc2.39", + "python": "3.9.25", + "startedAt": "2026-04-10T09:16:57.612247Z", + "program": "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", + "codePath": "llmbc/workspace/train_bc_lowdim_workspace.py", + "git": { + "remote": "https://github.com/CHYang25/LLM-BC.git", + "commit": "2fa092f9eb4c2ad9435919c661a618071a583be5" + }, + "email": "chris920325@gmail.com", + "root": "/workspace/LLM-BC/data/outputs/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0", + "host": "c8ef52649900", + "username": "root", + "executable": "/venv/llm-bc/bin/python3", + "codePathLocal": "llmbc/workspace/train_bc_lowdim_workspace.py", + "cpu_count": 72, + "cpu_count_logical": 144, + "gpu": "NVIDIA GeForce RTX 4090", + "gpu_count": 2, + "disk": { + "/": { + "total": "545460846592", + "used": "206078439424" + } + }, + "memory": { + "total": "810942898176" + }, + "cpu": { + "count": 72, + "countLogical": 144 + }, + "gpu_nvidia": [ + { + "name": "NVIDIA GeForce RTX 4090", + "memoryTotal": "25757220864", + "cudaCores": 16384, + "architecture": "Ada" + }, + { + "name": "NVIDIA GeForce RTX 4090", + "memoryTotal": "25757220864", + "cudaCores": 16384, + "architecture": "Ada" + } + ], + "cudaVersion": "12.8" +} \ No newline at end of file diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/wandb-summary.json b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/wandb-summary.json new file mode 100644 index 0000000000000000000000000000000000000000..3fada2dc75b88ef95041c5bcaaed042d5d46bf6b --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/files/wandb-summary.json @@ -0,0 +1 @@ +{"_wandb":{"runtime":472},"global_step":197,"lr":0.009999978020164803,"_runtime":472.555626345,"_timestamp":1.7758126203738923e+09,"train_loss":0.36438101530075073,"epoch":0,"_step":197,"train_loss_bc":0.36438101530075073,"grad_norm":0.3752126395702362} \ No newline at end of file diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/run-ags56sz9.wandb b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/run-ags56sz9.wandb new file mode 100644 index 0000000000000000000000000000000000000000..8eeed81f879da450f53ce408a5ad909898662ccf --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_091657-ags56sz9/run-ags56sz9.wandb @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:22f617861eb4d60e1dbaa3f0c287a22618d6896fbe0438b819f9419b852f7172 +size 425984 diff --git a/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/wandb-resume.json b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/wandb-resume.json new file mode 100644 index 0000000000000000000000000000000000000000..2da213692d35b35480174aa0c9889ba755f2ec53 --- /dev/null +++ b/2026.04.10/09.16.06_train_bc_lowdim_maze2d-medium-v0/wandb/wandb-resume.json @@ -0,0 +1 @@ +{"run_id": "ags56sz9"} \ No newline at end of file diff --git a/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/.hydra/config.yaml b/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b2aad840d49c297289633c63d93884b4b398d9d0 --- /dev/null +++ b/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/.hydra/config.yaml @@ -0,0 +1,165 @@ +name: train_llmbc_lowdim +_target_: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +model_name: ${llm.name} +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +llm_orig_expert_feedback: true +llm_do_sample: false +policy: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: tanh + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + task_id: ${task_name} + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + cfg: ${llm} + obs_dim: ${task.obs_dim} + action_dim: ${task.action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + loss_bc_weight: 1.0 + loss_llm_weight: 0.01 + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + normalize_llm_loss: true +dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.0001 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: cosine + lr_warmup_steps: 10 + num_epochs: 801 + gradient_accumulate_every: 8 + grad_norm_clip: 0.5 + rollout_every: 4 + checkpoint_every: 4 + val_every: 1 + sample_every: 4 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training-mix + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: maze2d-medium-v0 + obs_dim: 6 + action_dim: 2 + env_runner: + _target_: llmbc.env_runner.pointmaze_lowdim_runner.PointmazeLowdimRunner + env_name: llf-pointmaze-maze2d-medium-v0 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 250 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.pointmaze_lowdim_dataset.PointmazeLowdimDataset + data_path: datasets/maze2d-medium-v0-mix(csp5).pt + data_path2: datasets/maze2d-medium-v0.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.02 + dummy_normalizer: true + instructor: + _target_: llmbc.translator.instructor.pointmaze_instructor.maze2d_medium_v0_instructor.Maze2dMediumV0Instructor +llm: + name: HuggingFaceTB/SmolLM2-135M-Instruct + model_name: SmolLM2-135M-Instruct + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + use_quantization: false + use_joint_mlp_projector: true + llm_mode: ete-finetuned + finetune_mode: orig + checkpoint: data/outputs/2025.10.16/11.53.22_train_llm_lowdim_maze2d-medium-v0/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-maze2d-medium-v0/checkpoint-11850 + max_length: 100 + lora_config: + r: 32 + lora_alpha: 64 + lora_dropout: 0.05 + bias: none + task_type: CAUSAL_LM + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + hydra: + job: + override_dirname: ${model_name} + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${model_name} diff --git a/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml b/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..18690d7a0d257749c0aff197003cdc847fc88e1f --- /dev/null +++ b/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: + - training.seed=42 + job: + name: train + chdir: null + override_dirname: training.seed=42 + id: ??? + num: ??? + config_name: llmbc_maze2d-medium-v0_mix(csp5).yaml + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/config/mix_data + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0 + choices: + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml b/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b8fcfadfbd1253a3af726bb8f893b41de23943fc --- /dev/null +++ b/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml @@ -0,0 +1 @@ +- training.seed=42 diff --git a/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/train.log b/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/train.log new file mode 100644 index 0000000000000000000000000000000000000000..94f3e5e1afa7e7c4c6735b5f96224bfd465e8609 --- /dev/null +++ b/2026.04.10/09.26.57_train_llmbc_lowdim_maze2d-medium-v0/train.log @@ -0,0 +1,6 @@ +[2026-04-10 09:26:58,688][numexpr.utils][INFO] - Note: detected 144 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable. +[2026-04-10 09:26:58,688][numexpr.utils][INFO] - Note: NumExpr detected 144 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 16. +[2026-04-10 09:26:58,688][numexpr.utils][INFO] - NumExpr defaulting to 16 threads. +[2026-04-10 09:27:02,106][datasets][INFO] - PyTorch version 2.2.2 available. +[2026-04-10 09:27:02,106][datasets][INFO] - TensorFlow version 2.15.1 available. +[2026-04-10 09:27:02,107][datasets][INFO] - JAX version 0.4.30 available. diff --git a/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml b/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..688c0cf11e42e639a2dcc43dbbe166e0b5ff10ea --- /dev/null +++ b/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml @@ -0,0 +1,163 @@ +name: train_llmbc_lowdim +_target_: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +model_name: ${llm.name} +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +llm_orig_expert_feedback: true +llm_do_sample: false +policy: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + task_id: ${task_name} + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + cfg: ${llm} + obs_dim: ${task.obs_dim} + action_dim: ${task.action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + loss_bc_weight: 1.0 + loss_llm_weight: 0.01 + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + normalize_llm_loss: true +dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.001 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: constant + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 8 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training-mix + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: adroit-hand-relocate-v1 + obs_dim: 39 + action_dim: 30 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + env_name: llf-adroit-adroit-hand-relocate-v1 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 150 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-mix(csp100).pt + data_path2: datasets/adroit-hand-relocate-v1.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.1 + dummy_normalizer: true + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor +llm: + name: HuggingFaceTB/SmolLM2-135M-Instruct + model_name: SmolLM2-135M-Instruct + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + use_quantization: false + use_joint_mlp_projector: true + llm_mode: ete-finetuned + finetune_mode: orig + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + max_length: 100 + lora_config: + r: 32 + lora_alpha: 64 + lora_dropout: 0.05 + bias: none + task_type: CAUSAL_LM + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + hydra: + job: + override_dirname: ${model_name} + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${model_name} diff --git a/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml b/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..168e24c1ae86bd8c2cdbe7230a464ee111d4f273 --- /dev/null +++ b/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml @@ -0,0 +1,154 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: llmbc_adroit-hand-relocate-v1_mix(csp100).yaml + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/config/mix_data + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1 + choices: + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml b/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log b/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log new file mode 100644 index 0000000000000000000000000000000000000000..40e65c8ba6d28adca1447bb42cde08ac9cf5d515 --- /dev/null +++ b/2026.04.10/09.36.23_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log @@ -0,0 +1,6 @@ +[2026-04-10 09:36:24,868][numexpr.utils][INFO] - Note: detected 144 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable. +[2026-04-10 09:36:24,868][numexpr.utils][INFO] - Note: NumExpr detected 144 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 16. +[2026-04-10 09:36:24,868][numexpr.utils][INFO] - NumExpr defaulting to 16 threads. +[2026-04-10 09:36:28,715][datasets][INFO] - PyTorch version 2.2.2 available. +[2026-04-10 09:36:28,716][datasets][INFO] - TensorFlow version 2.15.1 available. +[2026-04-10 09:36:28,717][datasets][INFO] - JAX version 0.4.30 available. diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/.hydra/config.yaml b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b2aad840d49c297289633c63d93884b4b398d9d0 --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/.hydra/config.yaml @@ -0,0 +1,165 @@ +name: train_llmbc_lowdim +_target_: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +model_name: ${llm.name} +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +llm_orig_expert_feedback: true +llm_do_sample: false +policy: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: tanh + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + task_id: ${task_name} + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + cfg: ${llm} + obs_dim: ${task.obs_dim} + action_dim: ${task.action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + loss_bc_weight: 1.0 + loss_llm_weight: 0.01 + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + normalize_llm_loss: true +dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.0001 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: cosine + lr_warmup_steps: 10 + num_epochs: 801 + gradient_accumulate_every: 8 + grad_norm_clip: 0.5 + rollout_every: 4 + checkpoint_every: 4 + val_every: 1 + sample_every: 4 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training-mix + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: maze2d-medium-v0 + obs_dim: 6 + action_dim: 2 + env_runner: + _target_: llmbc.env_runner.pointmaze_lowdim_runner.PointmazeLowdimRunner + env_name: llf-pointmaze-maze2d-medium-v0 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 250 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.pointmaze_lowdim_dataset.PointmazeLowdimDataset + data_path: datasets/maze2d-medium-v0-mix(csp5).pt + data_path2: datasets/maze2d-medium-v0.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.02 + dummy_normalizer: true + instructor: + _target_: llmbc.translator.instructor.pointmaze_instructor.maze2d_medium_v0_instructor.Maze2dMediumV0Instructor +llm: + name: HuggingFaceTB/SmolLM2-135M-Instruct + model_name: SmolLM2-135M-Instruct + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + use_quantization: false + use_joint_mlp_projector: true + llm_mode: ete-finetuned + finetune_mode: orig + checkpoint: data/outputs/2025.10.16/11.53.22_train_llm_lowdim_maze2d-medium-v0/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-maze2d-medium-v0/checkpoint-11850 + max_length: 100 + lora_config: + r: 32 + lora_alpha: 64 + lora_dropout: 0.05 + bias: none + task_type: CAUSAL_LM + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + hydra: + job: + override_dirname: ${model_name} + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${model_name} diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..9f1c3fe2393e1c8fa389780606016da96b549075 --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: + - training.seed=42 + job: + name: train + chdir: null + override_dirname: training.seed=42 + id: ??? + num: ??? + config_name: llmbc_maze2d-medium-v0_mix(csp5).yaml + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/config/mix_data + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0 + choices: + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b8fcfadfbd1253a3af726bb8f893b41de23943fc --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/.hydra/overrides.yaml @@ -0,0 +1 @@ +- training.seed=42 diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/logs.json.txt b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/logs.json.txt new file mode 100644 index 0000000000000000000000000000000000000000..153ceea35da610bb9097edb8250eeb0e6ddc161a --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/logs.json.txt @@ -0,0 +1,994 @@ +{"train_loss": 0.42537829279899597, "train_loss_bc": 0.4242355227470398, "train_loss_llm": 0.11427727341651917, "grad_norm": 0.20109279453754425, "global_step": 0, "epoch": 0, "lr": 1e-05} +{"train_loss": 0.5587033629417419, "train_loss_bc": 0.5566513538360596, "train_loss_llm": 0.20520302653312683, "grad_norm": 0.18879377841949463, "global_step": 1, "epoch": 0, "lr": 1e-05} +{"train_loss": 0.4462607204914093, "train_loss_bc": 0.4445344805717468, "train_loss_llm": 0.17262420058250427, "grad_norm": 0.1635759174823761, "global_step": 2, "epoch": 0, "lr": 1e-05} +{"train_loss": 0.41555556654930115, "train_loss_bc": 0.4142303764820099, "train_loss_llm": 0.13251951336860657, "grad_norm": 0.2087690234184265, "global_step": 3, "epoch": 0, "lr": 1e-05} +{"train_loss": 0.44046691060066223, "train_loss_bc": 0.4389246106147766, "train_loss_llm": 0.15422984957695007, "grad_norm": 0.25152289867401123, "global_step": 4, "epoch": 0, "lr": 1e-05} +{"train_loss": 0.4718659222126007, "train_loss_bc": 0.47070080041885376, "train_loss_llm": 0.11651232838630676, "grad_norm": 0.33151760697364807, "global_step": 5, "epoch": 0, "lr": 1e-05} +{"train_loss": 0.3306937515735626, "train_loss_bc": 0.33017823100090027, "train_loss_llm": 0.05155107378959656, "grad_norm": 0.35088518261909485, "global_step": 6, "epoch": 0, "lr": 1e-05} +{"train_loss": 0.35400843620300293, "train_loss_bc": 0.35274773836135864, "train_loss_llm": 0.1260685920715332, "grad_norm": 0.4061644971370697, "global_step": 7, "epoch": 0, "lr": 1e-05} +{"train_loss": 0.5303317904472351, "train_loss_bc": 0.52833092212677, "train_loss_llm": 0.20008507370948792, "grad_norm": 0.572645902633667, "global_step": 8, "epoch": 0, "lr": 2e-05} +{"train_loss": 0.5247105360031128, "train_loss_bc": 0.5230463743209839, "train_loss_llm": 0.16641587018966675, "grad_norm": 0.12925344705581665, "global_step": 9, "epoch": 0, "lr": 2e-05} +{"train_loss": 0.4259011447429657, "train_loss_bc": 0.4245714545249939, "train_loss_llm": 0.13297036290168762, "grad_norm": 0.13075397908687592, "global_step": 10, "epoch": 0, "lr": 2e-05} +{"train_loss": 0.4494341015815735, "train_loss_bc": 0.44796547293663025, "train_loss_llm": 0.14686396718025208, "grad_norm": 0.14756828546524048, "global_step": 11, "epoch": 0, "lr": 2e-05} +{"train_loss": 0.43677598237991333, "train_loss_bc": 0.4352250099182129, "train_loss_llm": 0.15509828925132751, "grad_norm": 0.19899871945381165, "global_step": 12, "epoch": 0, "lr": 2e-05} +{"train_loss": 0.543933629989624, "train_loss_bc": 0.5421090722084045, "train_loss_llm": 0.18245667219161987, "grad_norm": 0.23126213252544403, "global_step": 13, "epoch": 0, "lr": 2e-05} +{"train_loss": 0.45055487751960754, "train_loss_bc": 0.44912266731262207, "train_loss_llm": 0.14322206377983093, "grad_norm": 0.293745756149292, "global_step": 14, "epoch": 0, "lr": 2e-05} +{"train_loss": 0.6138245463371277, "train_loss_bc": 0.612447202205658, "train_loss_llm": 0.13773223757743835, "grad_norm": 0.2887137532234192, "global_step": 15, "epoch": 0, "lr": 2e-05} +{"train_loss": 0.5976467132568359, "train_loss_bc": 0.5962076187133789, "train_loss_llm": 0.14390695095062256, "grad_norm": 0.3522990643978119, "global_step": 16, "epoch": 0, "lr": 3e-05} +{"train_loss": 0.689741849899292, "train_loss_bc": 0.688717246055603, "train_loss_llm": 0.10246020555496216, "grad_norm": 0.19683517515659332, "global_step": 17, "epoch": 0, "lr": 3e-05} +{"train_loss": 0.5476487874984741, "train_loss_bc": 0.5460153222084045, "train_loss_llm": 0.1633484959602356, "grad_norm": 0.20818094909191132, "global_step": 18, "epoch": 0, "lr": 3e-05} +{"train_loss": 0.46317753195762634, "train_loss_bc": 0.46163302659988403, "train_loss_llm": 0.15445014834403992, "grad_norm": 0.23228569328784943, "global_step": 19, "epoch": 0, "lr": 3e-05} +{"train_loss": 0.4844187796115875, "train_loss_bc": 0.48298776149749756, "train_loss_llm": 0.14310038089752197, "grad_norm": 0.35107213258743286, "global_step": 20, "epoch": 0, "lr": 3e-05} +{"train_loss": 0.6067336201667786, "train_loss_bc": 0.6051198244094849, "train_loss_llm": 0.1613824963569641, "grad_norm": 0.45436468720436096, "global_step": 21, "epoch": 0, "lr": 3e-05} +{"train_loss": 0.46505242586135864, "train_loss_bc": 0.46333444118499756, "train_loss_llm": 0.17179954051971436, "grad_norm": 0.5813204646110535, "global_step": 22, "epoch": 0, "lr": 3e-05} +{"train_loss": 0.6188356876373291, "train_loss_bc": 0.6170319318771362, "train_loss_llm": 0.18037381768226624, "grad_norm": 0.6040111184120178, "global_step": 23, "epoch": 0, "lr": 3e-05} +{"train_loss": 0.4598807990550995, "train_loss_bc": 0.45842939615249634, "train_loss_llm": 0.14514175057411194, "grad_norm": 0.7608086466789246, "global_step": 24, "epoch": 0, "lr": 4e-05} +{"train_loss": 0.5262072086334229, "train_loss_bc": 0.5246200561523438, "train_loss_llm": 0.15871602296829224, "grad_norm": 0.15511608123779297, "global_step": 25, "epoch": 0, "lr": 4e-05} +{"train_loss": 0.3689386248588562, "train_loss_bc": 0.36803102493286133, "train_loss_llm": 0.09075862169265747, "grad_norm": 0.2087269276380539, "global_step": 26, "epoch": 0, "lr": 4e-05} +{"train_loss": 0.4542337954044342, "train_loss_bc": 0.45304545760154724, "train_loss_llm": 0.11883461475372314, "grad_norm": 0.2554692327976227, "global_step": 27, "epoch": 0, "lr": 4e-05} +{"train_loss": 0.43385347723960876, "train_loss_bc": 0.4322516918182373, "train_loss_llm": 0.16017770767211914, "grad_norm": 0.2653736174106598, "global_step": 28, "epoch": 0, "lr": 4e-05} +{"train_loss": 0.29679232835769653, "train_loss_bc": 0.29573842883110046, "train_loss_llm": 0.10538855195045471, "grad_norm": 0.2753986716270447, "global_step": 29, "epoch": 0, "lr": 4e-05} +{"train_loss": 0.5480039119720459, "train_loss_bc": 0.5463132858276367, "train_loss_llm": 0.16906005144119263, "grad_norm": 0.2644517719745636, "global_step": 30, "epoch": 0, "lr": 4e-05} +{"train_loss": 0.5770899057388306, "train_loss_bc": 0.5753148198127747, "train_loss_llm": 0.17751061916351318, "grad_norm": 0.31916460394859314, "global_step": 31, "epoch": 0, "lr": 4e-05} +{"train_loss": 0.4671458601951599, "train_loss_bc": 0.46583545207977295, "train_loss_llm": 0.13104140758514404, "grad_norm": 0.4049204885959625, "global_step": 32, "epoch": 0, "lr": 5e-05} +{"train_loss": 0.453065425157547, "train_loss_bc": 0.4515276551246643, "train_loss_llm": 0.15377846360206604, "grad_norm": 0.20320573449134827, "global_step": 33, "epoch": 0, "lr": 5e-05} +{"train_loss": 0.5046079754829407, "train_loss_bc": 0.5035127401351929, "train_loss_llm": 0.10952073335647583, "grad_norm": 0.23440375924110413, "global_step": 34, "epoch": 0, "lr": 5e-05} +{"train_loss": 0.4279428720474243, "train_loss_bc": 0.4265894591808319, "train_loss_llm": 0.1353399157524109, "grad_norm": 0.23843787610530853, "global_step": 35, "epoch": 0, "lr": 5e-05} +{"train_loss": 0.5524616241455078, "train_loss_bc": 0.550616979598999, "train_loss_llm": 0.18446198105812073, "grad_norm": 0.2991805374622345, "global_step": 36, "epoch": 0, "lr": 5e-05} +{"train_loss": 0.42292094230651855, "train_loss_bc": 0.42167216539382935, "train_loss_llm": 0.12487682700157166, "grad_norm": 0.396682471036911, "global_step": 37, "epoch": 0, "lr": 5e-05} +{"train_loss": 0.5053940415382385, "train_loss_bc": 0.504203200340271, "train_loss_llm": 0.1190866231918335, "grad_norm": 0.5061362385749817, "global_step": 38, "epoch": 0, "lr": 5e-05} +{"train_loss": 0.5233940482139587, "train_loss_bc": 0.521477222442627, "train_loss_llm": 0.1916818618774414, "grad_norm": 0.5985338091850281, "global_step": 39, "epoch": 0, "lr": 5e-05} +{"train_loss": 0.6034775376319885, "train_loss_bc": 0.602109432220459, "train_loss_llm": 0.13680776953697205, "grad_norm": 0.6971749663352966, "global_step": 40, "epoch": 0, "lr": 6e-05} +{"train_loss": 0.5834956765174866, "train_loss_bc": 0.581815242767334, "train_loss_llm": 0.16804471611976624, "grad_norm": 0.07088257372379303, "global_step": 41, "epoch": 0, "lr": 6e-05} +{"train_loss": 0.5674425363540649, "train_loss_bc": 0.5662237405776978, "train_loss_llm": 0.1218770444393158, "grad_norm": 0.12487657368183136, "global_step": 42, "epoch": 0, "lr": 6e-05} +{"train_loss": 0.5035828948020935, "train_loss_bc": 0.5017828345298767, "train_loss_llm": 0.18000522255897522, "grad_norm": 0.207634836435318, "global_step": 43, "epoch": 0, "lr": 6e-05} +{"train_loss": 0.5520501136779785, "train_loss_bc": 0.5502804517745972, "train_loss_llm": 0.17696797847747803, "grad_norm": 0.2586645185947418, "global_step": 44, "epoch": 0, "lr": 6e-05} +{"train_loss": 0.435793936252594, "train_loss_bc": 0.4343380331993103, "train_loss_llm": 0.1455911099910736, "grad_norm": 0.3548068106174469, "global_step": 45, "epoch": 0, "lr": 6e-05} +{"train_loss": 0.48491862416267395, "train_loss_bc": 0.4836008846759796, "train_loss_llm": 0.13177356123924255, "grad_norm": 0.34207063913345337, "global_step": 46, "epoch": 0, "lr": 6e-05} +{"train_loss": 0.47173404693603516, "train_loss_bc": 0.47061073780059814, "train_loss_llm": 0.11233055591583252, "grad_norm": 0.3656021058559418, "global_step": 47, "epoch": 0, "lr": 6e-05} +{"train_loss": 0.5558405518531799, "train_loss_bc": 0.5543102025985718, "train_loss_llm": 0.15303555130958557, "grad_norm": 0.4806089401245117, "global_step": 48, "epoch": 0, "lr": 7e-05} +{"train_loss": 0.46238505840301514, "train_loss_bc": 0.46122118830680847, "train_loss_llm": 0.11638659238815308, "grad_norm": 0.08411291241645813, "global_step": 49, "epoch": 0, "lr": 7e-05} +{"train_loss": 0.4940248727798462, "train_loss_bc": 0.4927865266799927, "train_loss_llm": 0.12383323907852173, "grad_norm": 0.23042914271354675, "global_step": 50, "epoch": 0, "lr": 7e-05} +{"train_loss": 0.4324430227279663, "train_loss_bc": 0.4311506152153015, "train_loss_llm": 0.12924090027809143, "grad_norm": 0.3413877487182617, "global_step": 51, "epoch": 0, "lr": 7e-05} +{"train_loss": 0.4552828073501587, "train_loss_bc": 0.4536489248275757, "train_loss_llm": 0.16338947415351868, "grad_norm": 0.42864707112312317, "global_step": 52, "epoch": 0, "lr": 7e-05} +{"train_loss": 0.5363460183143616, "train_loss_bc": 0.5348853468894958, "train_loss_llm": 0.14606910943984985, "grad_norm": 0.49550768733024597, "global_step": 53, "epoch": 0, "lr": 7e-05} +{"train_loss": 0.5482083559036255, "train_loss_bc": 0.5469648838043213, "train_loss_llm": 0.12434685230255127, "grad_norm": 0.4439186453819275, "global_step": 54, "epoch": 0, "lr": 7e-05} +{"train_loss": 0.4959681034088135, "train_loss_bc": 0.4943790137767792, "train_loss_llm": 0.15890923142433167, "grad_norm": 0.5558111071586609, "global_step": 55, "epoch": 0, "lr": 7e-05} +{"train_loss": 0.3806264400482178, "train_loss_bc": 0.37928667664527893, "train_loss_llm": 0.13397592306137085, "grad_norm": 0.5495619773864746, "global_step": 56, "epoch": 0, "lr": 8e-05} +{"train_loss": 0.4922499656677246, "train_loss_bc": 0.49102920293807983, "train_loss_llm": 0.12207609415054321, "grad_norm": 0.11377433687448502, "global_step": 57, "epoch": 0, "lr": 8e-05} +{"train_loss": 0.5370638966560364, "train_loss_bc": 0.5353915095329285, "train_loss_llm": 0.16724076867103577, "grad_norm": 0.16296590864658356, "global_step": 58, "epoch": 0, "lr": 8e-05} +{"train_loss": 0.546135425567627, "train_loss_bc": 0.5442732572555542, "train_loss_llm": 0.18621894717216492, "grad_norm": 0.22496655583381653, "global_step": 59, "epoch": 0, "lr": 8e-05} +{"train_loss": 0.5145379304885864, "train_loss_bc": 0.5133035182952881, "train_loss_llm": 0.1234428882598877, "grad_norm": 0.2609196901321411, "global_step": 60, "epoch": 0, "lr": 8e-05} +{"train_loss": 0.4656980335712433, "train_loss_bc": 0.46409252285957336, "train_loss_llm": 0.16055193543434143, "grad_norm": 0.3485998511314392, "global_step": 61, "epoch": 0, "lr": 8e-05} +{"train_loss": 0.5646767020225525, "train_loss_bc": 0.5628073215484619, "train_loss_llm": 0.18693813681602478, "grad_norm": 0.30050185322761536, "global_step": 62, "epoch": 0, "lr": 8e-05} +{"train_loss": 0.5088056921958923, "train_loss_bc": 0.507488489151001, "train_loss_llm": 0.13171929121017456, "grad_norm": 0.3856169581413269, "global_step": 63, "epoch": 0, "lr": 8e-05} +{"train_loss": 0.35728180408477783, "train_loss_bc": 0.3560566008090973, "train_loss_llm": 0.1225210428237915, "grad_norm": 0.47128307819366455, "global_step": 64, "epoch": 0, "lr": 9e-05} +{"train_loss": 0.40459445118904114, "train_loss_bc": 0.4030808210372925, "train_loss_llm": 0.1513628363609314, "grad_norm": 0.12876716256141663, "global_step": 65, "epoch": 0, "lr": 9e-05} +{"train_loss": 0.46717119216918945, "train_loss_bc": 0.46529215574264526, "train_loss_llm": 0.18790432810783386, "grad_norm": 0.25889140367507935, "global_step": 66, "epoch": 0, "lr": 9e-05} +{"train_loss": 0.4788550138473511, "train_loss_bc": 0.4774649143218994, "train_loss_llm": 0.13901013135910034, "grad_norm": 0.3773616552352905, "global_step": 67, "epoch": 0, "lr": 9e-05} +{"train_loss": 0.35348936915397644, "train_loss_bc": 0.3523732125759125, "train_loss_llm": 0.11161491274833679, "grad_norm": 0.40989989042282104, "global_step": 68, "epoch": 0, "lr": 9e-05} +{"train_loss": 0.4767206013202667, "train_loss_bc": 0.47509491443634033, "train_loss_llm": 0.16256842017173767, "grad_norm": 0.5057125091552734, "global_step": 69, "epoch": 0, "lr": 9e-05} +{"train_loss": 0.33047839999198914, "train_loss_bc": 0.32951927185058594, "train_loss_llm": 0.09591150283813477, "grad_norm": 0.530107855796814, "global_step": 70, "epoch": 0, "lr": 9e-05} +{"train_loss": 0.42009755969047546, "train_loss_bc": 0.41886377334594727, "train_loss_llm": 0.12337765097618103, "grad_norm": 0.45365235209465027, "global_step": 71, "epoch": 0, "lr": 9e-05} +{"train_loss": 0.5082347393035889, "train_loss_bc": 0.5067936182022095, "train_loss_llm": 0.14411038160324097, "grad_norm": 0.5319412350654602, "global_step": 72, "epoch": 0, "lr": 0.0001} +{"train_loss": 0.5480773448944092, "train_loss_bc": 0.5465723276138306, "train_loss_llm": 0.1504993438720703, "grad_norm": 0.08676057308912277, "global_step": 73, "epoch": 0, "lr": 0.0001} +{"train_loss": 0.5218899250030518, "train_loss_bc": 0.520464301109314, "train_loss_llm": 0.14256072044372559, "grad_norm": 0.3363056778907776, "global_step": 74, "epoch": 0, "lr": 0.0001} +{"train_loss": 0.442917138338089, "train_loss_bc": 0.4414409101009369, "train_loss_llm": 0.1476224958896637, "grad_norm": 0.4445405602455139, "global_step": 75, "epoch": 0, "lr": 0.0001} +{"train_loss": 0.4704687297344208, "train_loss_bc": 0.4689462184906006, "train_loss_llm": 0.1522519290447235, "grad_norm": 0.46211710572242737, "global_step": 76, "epoch": 0, "lr": 0.0001} +{"train_loss": 0.38872218132019043, "train_loss_bc": 0.3873938322067261, "train_loss_llm": 0.13283368945121765, "grad_norm": 0.40344831347465515, "global_step": 77, "epoch": 0, "lr": 0.0001} +{"train_loss": 0.4938354790210724, "train_loss_bc": 0.49224162101745605, "train_loss_llm": 0.15938666462898254, "grad_norm": 0.386066734790802, "global_step": 78, "epoch": 0, "lr": 0.0001} +{"train_loss": 0.3551795482635498, "train_loss_bc": 0.35430803894996643, "train_loss_llm": 0.08715137839317322, "grad_norm": 0.41477641463279724, "global_step": 79, "epoch": 0, "lr": 0.0001} +{"train_loss": 0.5461940169334412, "train_loss_bc": 0.5444260239601135, "train_loss_llm": 0.1767999529838562, "grad_norm": 0.46446946263313293, "global_step": 80, "epoch": 0, "lr": 9.999999999440777e-05} +{"train_loss": 0.5846241116523743, "train_loss_bc": 0.5827171802520752, "train_loss_llm": 0.19069188833236694, "grad_norm": 0.09149337559938431, "global_step": 81, "epoch": 0, "lr": 9.999999999440777e-05} +{"train_loss": 0.4963512718677521, "train_loss_bc": 0.4950132369995117, "train_loss_llm": 0.13380387425422668, "grad_norm": 0.1511898636817932, "global_step": 82, "epoch": 0, "lr": 9.999999999440777e-05} +{"train_loss": 0.4979674518108368, "train_loss_bc": 0.49698758125305176, "train_loss_llm": 0.09798750281333923, "grad_norm": 0.21909315884113312, "global_step": 83, "epoch": 0, "lr": 9.999999999440777e-05} +{"train_loss": 0.5414943099021912, "train_loss_bc": 0.5398356318473816, "train_loss_llm": 0.16586947441101074, "grad_norm": 0.20258116722106934, "global_step": 84, "epoch": 0, "lr": 9.999999999440777e-05} +{"train_loss": 0.515712559223175, "train_loss_bc": 0.5144752860069275, "train_loss_llm": 0.12372535467147827, "grad_norm": 0.2562904953956604, "global_step": 85, "epoch": 0, "lr": 9.999999999440777e-05} +{"train_loss": 0.4642152190208435, "train_loss_bc": 0.4627920985221863, "train_loss_llm": 0.1423119604587555, "grad_norm": 0.2340392917394638, "global_step": 86, "epoch": 0, "lr": 9.999999999440777e-05} +{"train_loss": 0.351498007774353, "train_loss_bc": 0.3503400385379791, "train_loss_llm": 0.11579820513725281, "grad_norm": 0.2210949957370758, "global_step": 87, "epoch": 0, "lr": 9.999999999440777e-05} +{"train_loss": 0.47873637080192566, "train_loss_bc": 0.47719547152519226, "train_loss_llm": 0.15409085154533386, "grad_norm": 0.24862340092658997, "global_step": 88, "epoch": 0, "lr": 9.999999997763103e-05} +{"train_loss": 0.5378441214561462, "train_loss_bc": 0.5362796187400818, "train_loss_llm": 0.15644943714141846, "grad_norm": 0.09886565804481506, "global_step": 89, "epoch": 0, "lr": 9.999999997763103e-05} +{"train_loss": 0.45138272643089294, "train_loss_bc": 0.4504159986972809, "train_loss_llm": 0.09667408466339111, "grad_norm": 0.13475902378559113, "global_step": 90, "epoch": 0, "lr": 9.999999997763103e-05} +{"train_loss": 0.5092982053756714, "train_loss_bc": 0.508211612701416, "train_loss_llm": 0.10865724086761475, "grad_norm": 0.16423630714416504, "global_step": 91, "epoch": 0, "lr": 9.999999997763103e-05} +{"train_loss": 0.4703865647315979, "train_loss_bc": 0.4685760736465454, "train_loss_llm": 0.18104958534240723, "grad_norm": 0.13507616519927979, "global_step": 92, "epoch": 0, "lr": 9.999999997763103e-05} +{"train_loss": 0.5042034983634949, "train_loss_bc": 0.5027568340301514, "train_loss_llm": 0.14466816186904907, "grad_norm": 0.19379568099975586, "global_step": 93, "epoch": 0, "lr": 9.999999997763103e-05} +{"train_loss": 0.49900463223457336, "train_loss_bc": 0.4977543354034424, "train_loss_llm": 0.1250307857990265, "grad_norm": 0.3972480595111847, "global_step": 94, "epoch": 0, "lr": 9.999999997763103e-05} +{"train_loss": 0.36963579058647156, "train_loss_bc": 0.3686002492904663, "train_loss_llm": 0.10355502367019653, "grad_norm": 0.4472208321094513, "global_step": 95, "epoch": 0, "lr": 9.999999997763103e-05} +{"train_loss": 0.45268434286117554, "train_loss_bc": 0.4514434337615967, "train_loss_llm": 0.12408989667892456, "grad_norm": 0.502593994140625, "global_step": 96, "epoch": 0, "lr": 9.999999994966981e-05} +{"train_loss": 0.453140527009964, "train_loss_bc": 0.45146286487579346, "train_loss_llm": 0.16776618361473083, "grad_norm": 0.07273141294717789, "global_step": 97, "epoch": 0, "lr": 9.999999994966981e-05} +{"train_loss": 0.3973062038421631, "train_loss_bc": 0.39598485827445984, "train_loss_llm": 0.13213497400283813, "grad_norm": 0.12742798030376434, "global_step": 98, "epoch": 0, "lr": 9.999999994966981e-05} +{"train_loss": 0.5444526672363281, "train_loss_bc": 0.5433326959609985, "train_loss_llm": 0.11199888586997986, "grad_norm": 0.18529370427131653, "global_step": 99, "epoch": 0, "lr": 9.999999994966981e-05} +{"train_loss": 0.4612184166908264, "train_loss_bc": 0.4601033329963684, "train_loss_llm": 0.11150789260864258, "grad_norm": 0.22524675726890564, "global_step": 100, "epoch": 0, "lr": 9.999999994966981e-05} +{"train_loss": 0.4753912091255188, "train_loss_bc": 0.4742206037044525, "train_loss_llm": 0.11706015467643738, "grad_norm": 0.2990623712539673, "global_step": 101, "epoch": 0, "lr": 9.999999994966981e-05} +{"train_loss": 0.5015730261802673, "train_loss_bc": 0.5003759860992432, "train_loss_llm": 0.11970412731170654, "grad_norm": 0.2643897533416748, "global_step": 102, "epoch": 0, "lr": 9.999999994966981e-05} +{"train_loss": 0.6083633303642273, "train_loss_bc": 0.6069818735122681, "train_loss_llm": 0.13814318180084229, "grad_norm": 0.3867564797401428, "global_step": 103, "epoch": 0, "lr": 9.999999994966981e-05} +{"train_loss": 0.504302442073822, "train_loss_bc": 0.502646803855896, "train_loss_llm": 0.16556596755981445, "grad_norm": 0.4619770348072052, "global_step": 104, "epoch": 0, "lr": 9.99999999105241e-05} +{"train_loss": 0.518189013004303, "train_loss_bc": 0.5170090198516846, "train_loss_llm": 0.11799734830856323, "grad_norm": 0.21459585428237915, "global_step": 105, "epoch": 0, "lr": 9.99999999105241e-05} +{"train_loss": 0.4508154094219208, "train_loss_bc": 0.44938838481903076, "train_loss_llm": 0.14270314574241638, "grad_norm": 0.2093033492565155, "global_step": 106, "epoch": 0, "lr": 9.99999999105241e-05} +{"train_loss": 0.4866752326488495, "train_loss_bc": 0.48523634672164917, "train_loss_llm": 0.14389005303382874, "grad_norm": 0.38878968358039856, "global_step": 107, "epoch": 0, "lr": 9.99999999105241e-05} +{"train_loss": 0.5696703791618347, "train_loss_bc": 0.5682628750801086, "train_loss_llm": 0.14074894785881042, "grad_norm": 0.449392169713974, "global_step": 108, "epoch": 0, "lr": 9.99999999105241e-05} +{"train_loss": 0.41269052028656006, "train_loss_bc": 0.41145557165145874, "train_loss_llm": 0.12349450588226318, "grad_norm": 0.3769812285900116, "global_step": 109, "epoch": 0, "lr": 9.99999999105241e-05} +{"train_loss": 0.36800768971443176, "train_loss_bc": 0.36644500494003296, "train_loss_llm": 0.15626752376556396, "grad_norm": 0.37687283754348755, "global_step": 110, "epoch": 0, "lr": 9.99999999105241e-05} +{"train_loss": 0.48692283034324646, "train_loss_bc": 0.48562875390052795, "train_loss_llm": 0.1294083595275879, "grad_norm": 0.5166057348251343, "global_step": 111, "epoch": 0, "lr": 9.99999999105241e-05} +{"train_loss": 0.5153733491897583, "train_loss_bc": 0.5135075449943542, "train_loss_llm": 0.18658113479614258, "grad_norm": 0.5034386515617371, "global_step": 112, "epoch": 0, "lr": 9.999999986019391e-05} +{"train_loss": 0.5139015913009644, "train_loss_bc": 0.5124173164367676, "train_loss_llm": 0.14842838048934937, "grad_norm": 0.11090953648090363, "global_step": 113, "epoch": 0, "lr": 9.999999986019391e-05} +{"train_loss": 0.4305392801761627, "train_loss_bc": 0.42907634377479553, "train_loss_llm": 0.14629414677619934, "grad_norm": 0.13358819484710693, "global_step": 114, "epoch": 0, "lr": 9.999999986019391e-05} +{"train_loss": 0.4820598065853119, "train_loss_bc": 0.4806095063686371, "train_loss_llm": 0.14503106474876404, "grad_norm": 0.1469360738992691, "global_step": 115, "epoch": 0, "lr": 9.999999986019391e-05} +{"train_loss": 0.5452823638916016, "train_loss_bc": 0.5434094667434692, "train_loss_llm": 0.18728885054588318, "grad_norm": 0.3448217213153839, "global_step": 116, "epoch": 0, "lr": 9.999999986019391e-05} +{"train_loss": 0.48445528745651245, "train_loss_bc": 0.4829200208187103, "train_loss_llm": 0.15352600812911987, "grad_norm": 0.31305384635925293, "global_step": 117, "epoch": 0, "lr": 9.999999986019391e-05} +{"train_loss": 0.5596451759338379, "train_loss_bc": 0.5585609674453735, "train_loss_llm": 0.10841885209083557, "grad_norm": 0.2849327325820923, "global_step": 118, "epoch": 0, "lr": 9.999999986019391e-05} +{"train_loss": 0.4792906939983368, "train_loss_bc": 0.47799795866012573, "train_loss_llm": 0.12927451729774475, "grad_norm": 0.39754050970077515, "global_step": 119, "epoch": 0, "lr": 9.999999986019391e-05} +{"train_loss": 0.4401831328868866, "train_loss_bc": 0.43885380029678345, "train_loss_llm": 0.13293355703353882, "grad_norm": 0.384951114654541, "global_step": 120, "epoch": 0, "lr": 9.999999979867922e-05} +{"train_loss": 0.43352052569389343, "train_loss_bc": 0.4322432279586792, "train_loss_llm": 0.12773022055625916, "grad_norm": 0.1289375126361847, "global_step": 121, "epoch": 0, "lr": 9.999999979867922e-05} +{"train_loss": 0.5084827542304993, "train_loss_bc": 0.5069909691810608, "train_loss_llm": 0.1491813361644745, "grad_norm": 0.17626002430915833, "global_step": 122, "epoch": 0, "lr": 9.999999979867922e-05} +{"train_loss": 0.46655505895614624, "train_loss_bc": 0.4648292362689972, "train_loss_llm": 0.17258161306381226, "grad_norm": 0.14814235270023346, "global_step": 123, "epoch": 0, "lr": 9.999999979867922e-05} +{"train_loss": 0.45474323630332947, "train_loss_bc": 0.45358604192733765, "train_loss_llm": 0.11571845412254333, "grad_norm": 0.17808249592781067, "global_step": 124, "epoch": 0, "lr": 9.999999979867922e-05} +{"train_loss": 0.3575454652309418, "train_loss_bc": 0.35649996995925903, "train_loss_llm": 0.1045491099357605, "grad_norm": 0.1687416285276413, "global_step": 125, "epoch": 0, "lr": 9.999999979867922e-05} +{"train_loss": 0.42461392283439636, "train_loss_bc": 0.4232383966445923, "train_loss_llm": 0.13755375146865845, "grad_norm": 0.20393416285514832, "global_step": 126, "epoch": 0, "lr": 9.999999979867922e-05} +{"train_loss": 0.3533098101615906, "train_loss_bc": 0.352429062128067, "train_loss_llm": 0.0880734920501709, "grad_norm": 0.2515350580215454, "global_step": 127, "epoch": 0, "lr": 9.999999979867922e-05} +{"train_loss": 0.547330915927887, "train_loss_bc": 0.5458576083183289, "train_loss_llm": 0.14733082056045532, "grad_norm": 0.27076393365859985, "global_step": 128, "epoch": 0, "lr": 9.999999972598005e-05} +{"train_loss": 0.3951107859611511, "train_loss_bc": 0.39386576414108276, "train_loss_llm": 0.12450191378593445, "grad_norm": 0.14949746429920197, "global_step": 129, "epoch": 0, "lr": 9.999999972598005e-05} +{"train_loss": 0.42244744300842285, "train_loss_bc": 0.42126911878585815, "train_loss_llm": 0.11783313751220703, "grad_norm": 0.18234899640083313, "global_step": 130, "epoch": 0, "lr": 9.999999972598005e-05} +{"train_loss": 0.5798062086105347, "train_loss_bc": 0.5780292749404907, "train_loss_llm": 0.1776927411556244, "grad_norm": 0.14992594718933105, "global_step": 131, "epoch": 0, "lr": 9.999999972598005e-05} +{"train_loss": 0.38929322361946106, "train_loss_bc": 0.38788753747940063, "train_loss_llm": 0.14056798815727234, "grad_norm": 0.3055795431137085, "global_step": 132, "epoch": 0, "lr": 9.999999972598005e-05} +{"train_loss": 0.4500117599964142, "train_loss_bc": 0.4485156536102295, "train_loss_llm": 0.14961206912994385, "grad_norm": 0.3067500591278076, "global_step": 133, "epoch": 0, "lr": 9.999999972598005e-05} +{"train_loss": 0.4285288155078888, "train_loss_bc": 0.4271991550922394, "train_loss_llm": 0.13296663761138916, "grad_norm": 0.34004029631614685, "global_step": 134, "epoch": 0, "lr": 9.999999972598005e-05} +{"train_loss": 0.5130741000175476, "train_loss_bc": 0.5118278861045837, "train_loss_llm": 0.1246216893196106, "grad_norm": 0.35129207372665405, "global_step": 135, "epoch": 0, "lr": 9.999999972598005e-05} +{"train_loss": 0.40861961245536804, "train_loss_bc": 0.4073326289653778, "train_loss_llm": 0.12869763374328613, "grad_norm": 0.2889193296432495, "global_step": 136, "epoch": 0, "lr": 9.999999964209638e-05} +{"train_loss": 0.33554527163505554, "train_loss_bc": 0.33429890871047974, "train_loss_llm": 0.12463775277137756, "grad_norm": 0.10974100977182388, "global_step": 137, "epoch": 0, "lr": 9.999999964209638e-05} +{"train_loss": 0.4965358376502991, "train_loss_bc": 0.495059072971344, "train_loss_llm": 0.14767563343048096, "grad_norm": 0.15823940932750702, "global_step": 138, "epoch": 0, "lr": 9.999999964209638e-05} +{"train_loss": 0.46989020705223083, "train_loss_bc": 0.46816280484199524, "train_loss_llm": 0.17273974418640137, "grad_norm": 0.12684907019138336, "global_step": 139, "epoch": 0, "lr": 9.999999964209638e-05} +{"train_loss": 0.4871211349964142, "train_loss_bc": 0.48584455251693726, "train_loss_llm": 0.12765911221504211, "grad_norm": 0.15189939737319946, "global_step": 140, "epoch": 0, "lr": 9.999999964209638e-05} +{"train_loss": 0.48943379521369934, "train_loss_bc": 0.48809337615966797, "train_loss_llm": 0.13404065370559692, "grad_norm": 0.2438279390335083, "global_step": 141, "epoch": 0, "lr": 9.999999964209638e-05} +{"train_loss": 0.4534691572189331, "train_loss_bc": 0.45203444361686707, "train_loss_llm": 0.14347264170646667, "grad_norm": 0.18031901121139526, "global_step": 142, "epoch": 0, "lr": 9.999999964209638e-05} +{"train_loss": 0.5469819903373718, "train_loss_bc": 0.545139491558075, "train_loss_llm": 0.1842520833015442, "grad_norm": 0.23916636407375336, "global_step": 143, "epoch": 0, "lr": 9.999999964209638e-05} +{"train_loss": 0.5236669778823853, "train_loss_bc": 0.5223870277404785, "train_loss_llm": 0.12799468636512756, "grad_norm": 0.19613394141197205, "global_step": 144, "epoch": 0, "lr": 9.999999954702824e-05} +{"train_loss": 0.5006002187728882, "train_loss_bc": 0.4987196922302246, "train_loss_llm": 0.1880531907081604, "grad_norm": 0.1564287394285202, "global_step": 145, "epoch": 0, "lr": 9.999999954702824e-05} +{"train_loss": 0.39800477027893066, "train_loss_bc": 0.39681944251060486, "train_loss_llm": 0.1185322105884552, "grad_norm": 0.25627344846725464, "global_step": 146, "epoch": 0, "lr": 9.999999954702824e-05} +{"train_loss": 0.3850008547306061, "train_loss_bc": 0.38365018367767334, "train_loss_llm": 0.13506639003753662, "grad_norm": 0.27468013763427734, "global_step": 147, "epoch": 0, "lr": 9.999999954702824e-05} +{"train_loss": 0.44352349638938904, "train_loss_bc": 0.44231343269348145, "train_loss_llm": 0.12100732326507568, "grad_norm": 0.2150159627199173, "global_step": 148, "epoch": 0, "lr": 9.999999954702824e-05} +{"train_loss": 0.44835609197616577, "train_loss_bc": 0.4463629424571991, "train_loss_llm": 0.19931519031524658, "grad_norm": 0.2722063958644867, "global_step": 149, "epoch": 0, "lr": 9.999999954702824e-05} +{"train_loss": 0.5689083337783813, "train_loss_bc": 0.5677956342697144, "train_loss_llm": 0.11127063632011414, "grad_norm": 0.2833058834075928, "global_step": 150, "epoch": 0, "lr": 9.999999954702824e-05} +{"train_loss": 0.47362175583839417, "train_loss_bc": 0.47255629301071167, "train_loss_llm": 0.10654711723327637, "grad_norm": 0.26513800024986267, "global_step": 151, "epoch": 0, "lr": 9.999999954702824e-05} +{"train_loss": 0.46520376205444336, "train_loss_bc": 0.464280903339386, "train_loss_llm": 0.09228649735450745, "grad_norm": 0.3166268467903137, "global_step": 152, "epoch": 0, "lr": 9.99999994407756e-05} +{"train_loss": 0.4090734124183655, "train_loss_bc": 0.40779566764831543, "train_loss_llm": 0.1277731955051422, "grad_norm": 0.04599767178297043, "global_step": 153, "epoch": 0, "lr": 9.99999994407756e-05} +{"train_loss": 0.5307814478874207, "train_loss_bc": 0.5297832489013672, "train_loss_llm": 0.09982085227966309, "grad_norm": 0.18267887830734253, "global_step": 154, "epoch": 0, "lr": 9.99999994407756e-05} +{"train_loss": 0.4541011452674866, "train_loss_bc": 0.4530741274356842, "train_loss_llm": 0.10270175337791443, "grad_norm": 0.17866390943527222, "global_step": 155, "epoch": 0, "lr": 9.99999994407756e-05} +{"train_loss": 0.4424351155757904, "train_loss_bc": 0.44135236740112305, "train_loss_llm": 0.10827392339706421, "grad_norm": 0.26685774326324463, "global_step": 156, "epoch": 0, "lr": 9.99999994407756e-05} +{"train_loss": 0.4392889738082886, "train_loss_bc": 0.4375961124897003, "train_loss_llm": 0.16928544640541077, "grad_norm": 0.23997336626052856, "global_step": 157, "epoch": 0, "lr": 9.99999994407756e-05} +{"train_loss": 0.5245538949966431, "train_loss_bc": 0.5228623747825623, "train_loss_llm": 0.16914978623390198, "grad_norm": 0.25918954610824585, "global_step": 158, "epoch": 0, "lr": 9.99999994407756e-05} +{"train_loss": 0.4532073736190796, "train_loss_bc": 0.45170679688453674, "train_loss_llm": 0.15005803108215332, "grad_norm": 0.2396339327096939, "global_step": 159, "epoch": 0, "lr": 9.99999994407756e-05} +{"train_loss": 0.42831116914749146, "train_loss_bc": 0.42694830894470215, "train_loss_llm": 0.13628679513931274, "grad_norm": 0.3088024854660034, "global_step": 160, "epoch": 0, "lr": 9.999999932333848e-05} +{"train_loss": 0.45796042680740356, "train_loss_bc": 0.45673877000808716, "train_loss_llm": 0.12216591835021973, "grad_norm": 0.07000040262937546, "global_step": 161, "epoch": 0, "lr": 9.999999932333848e-05} +{"train_loss": 0.4128691554069519, "train_loss_bc": 0.4118698239326477, "train_loss_llm": 0.09993374347686768, "grad_norm": 0.1746189296245575, "global_step": 162, "epoch": 0, "lr": 9.999999932333848e-05} +{"train_loss": 0.37696942687034607, "train_loss_bc": 0.3755904734134674, "train_loss_llm": 0.13789677619934082, "grad_norm": 0.23928140103816986, "global_step": 163, "epoch": 0, "lr": 9.999999932333848e-05} +{"train_loss": 0.6614714860916138, "train_loss_bc": 0.6600403785705566, "train_loss_llm": 0.1431114375591278, "grad_norm": 0.3700709342956543, "global_step": 164, "epoch": 0, "lr": 9.999999932333848e-05} +{"train_loss": 0.4790080189704895, "train_loss_bc": 0.4773780107498169, "train_loss_llm": 0.16299965977668762, "grad_norm": 0.49836286902427673, "global_step": 165, "epoch": 0, "lr": 9.999999932333848e-05} +{"train_loss": 0.4560374617576599, "train_loss_bc": 0.4545397460460663, "train_loss_llm": 0.14977195858955383, "grad_norm": 0.5713899731636047, "global_step": 166, "epoch": 0, "lr": 9.999999932333848e-05} +{"train_loss": 0.35940152406692505, "train_loss_bc": 0.35821473598480225, "train_loss_llm": 0.11867982149124146, "grad_norm": 0.5162184238433838, "global_step": 167, "epoch": 0, "lr": 9.999999932333848e-05} +{"train_loss": 0.5046243071556091, "train_loss_bc": 0.5033835172653198, "train_loss_llm": 0.12407755851745605, "grad_norm": 0.5262932777404785, "global_step": 168, "epoch": 0, "lr": 9.999999919471688e-05} +{"train_loss": 0.4371809661388397, "train_loss_bc": 0.43590205907821655, "train_loss_llm": 0.12789174914360046, "grad_norm": 0.14133556187152863, "global_step": 169, "epoch": 0, "lr": 9.999999919471688e-05} +{"train_loss": 0.3440878987312317, "train_loss_bc": 0.34310418367385864, "train_loss_llm": 0.0983729362487793, "grad_norm": 0.15779443085193634, "global_step": 170, "epoch": 0, "lr": 9.999999919471688e-05} +{"train_loss": 0.5430394411087036, "train_loss_bc": 0.5417658090591431, "train_loss_llm": 0.12736579775810242, "grad_norm": 0.2662231922149658, "global_step": 171, "epoch": 0, "lr": 9.999999919471688e-05} +{"train_loss": 0.35604211688041687, "train_loss_bc": 0.35424521565437317, "train_loss_llm": 0.17968913912773132, "grad_norm": 0.35400983691215515, "global_step": 172, "epoch": 0, "lr": 9.999999919471688e-05} +{"train_loss": 0.3418031930923462, "train_loss_bc": 0.34073585271835327, "train_loss_llm": 0.10673323273658752, "grad_norm": 0.5046758651733398, "global_step": 173, "epoch": 0, "lr": 9.999999919471688e-05} +{"train_loss": 0.4296170771121979, "train_loss_bc": 0.42799776792526245, "train_loss_llm": 0.16192969679832458, "grad_norm": 0.48174068331718445, "global_step": 174, "epoch": 0, "lr": 9.999999919471688e-05} +{"train_loss": 0.4443141222000122, "train_loss_bc": 0.4427102208137512, "train_loss_llm": 0.1603909134864807, "grad_norm": 0.45451048016548157, "global_step": 175, "epoch": 0, "lr": 9.999999919471688e-05} +{"train_loss": 0.42721912264823914, "train_loss_bc": 0.4252225160598755, "train_loss_llm": 0.19966188073158264, "grad_norm": 0.5579273104667664, "global_step": 176, "epoch": 0, "lr": 9.999999905491077e-05} +{"train_loss": 0.42219996452331543, "train_loss_bc": 0.42060840129852295, "train_loss_llm": 0.15915673971176147, "grad_norm": 0.11076367646455765, "global_step": 177, "epoch": 0, "lr": 9.999999905491077e-05} +{"train_loss": 0.4523386061191559, "train_loss_bc": 0.4508865475654602, "train_loss_llm": 0.14520549774169922, "grad_norm": 0.2567046582698822, "global_step": 178, "epoch": 0, "lr": 9.999999905491077e-05} +{"train_loss": 0.3691568970680237, "train_loss_bc": 0.36791032552719116, "train_loss_llm": 0.1246580183506012, "grad_norm": 0.3146982491016388, "global_step": 179, "epoch": 0, "lr": 9.999999905491077e-05} +{"train_loss": 0.5310431718826294, "train_loss_bc": 0.529695987701416, "train_loss_llm": 0.13471773266792297, "grad_norm": 0.3284354507923126, "global_step": 180, "epoch": 0, "lr": 9.999999905491077e-05} +{"train_loss": 0.40890252590179443, "train_loss_bc": 0.40761488676071167, "train_loss_llm": 0.12876415252685547, "grad_norm": 0.4036176800727844, "global_step": 181, "epoch": 0, "lr": 9.999999905491077e-05} +{"train_loss": 0.4704955518245697, "train_loss_bc": 0.46891945600509644, "train_loss_llm": 0.1576085090637207, "grad_norm": 0.431083083152771, "global_step": 182, "epoch": 0, "lr": 9.999999905491077e-05} +{"train_loss": 0.4618470072746277, "train_loss_bc": 0.4602696895599365, "train_loss_llm": 0.15773320198059082, "grad_norm": 0.46126362681388855, "global_step": 183, "epoch": 0, "lr": 9.999999905491077e-05} +{"train_loss": 0.5228875875473022, "train_loss_bc": 0.5217981338500977, "train_loss_llm": 0.10894355177879333, "grad_norm": 0.4385491907596588, "global_step": 184, "epoch": 0, "lr": 9.999999890392018e-05} +{"train_loss": 0.38834285736083984, "train_loss_bc": 0.3870711326599121, "train_loss_llm": 0.12717217206954956, "grad_norm": 0.06606757640838623, "global_step": 185, "epoch": 0, "lr": 9.999999890392018e-05} +{"train_loss": 0.5991695523262024, "train_loss_bc": 0.5980571508407593, "train_loss_llm": 0.11124098300933838, "grad_norm": 0.1613786220550537, "global_step": 186, "epoch": 0, "lr": 9.999999890392018e-05} +{"train_loss": 0.47662416100502014, "train_loss_bc": 0.47488272190093994, "train_loss_llm": 0.1741437315940857, "grad_norm": 0.1918288767337799, "global_step": 187, "epoch": 0, "lr": 9.999999890392018e-05} +{"train_loss": 0.5109984278678894, "train_loss_bc": 0.5089355707168579, "train_loss_llm": 0.20628663897514343, "grad_norm": 0.17448517680168152, "global_step": 188, "epoch": 0, "lr": 9.999999890392018e-05} +{"train_loss": 0.44833824038505554, "train_loss_bc": 0.4472258687019348, "train_loss_llm": 0.11123636364936829, "grad_norm": 0.23448139429092407, "global_step": 189, "epoch": 0, "lr": 9.999999890392018e-05} +{"train_loss": 0.2989017069339752, "train_loss_bc": 0.2977512776851654, "train_loss_llm": 0.11504152417182922, "grad_norm": 0.2199392020702362, "global_step": 190, "epoch": 0, "lr": 9.999999890392018e-05} +{"train_loss": 0.5157555937767029, "train_loss_bc": 0.5139880180358887, "train_loss_llm": 0.17675742506980896, "grad_norm": 0.21442601084709167, "global_step": 191, "epoch": 0, "lr": 9.999999890392018e-05} +{"train_loss": 0.33841919898986816, "train_loss_bc": 0.33665046095848083, "train_loss_llm": 0.17687371373176575, "grad_norm": 0.22916430234909058, "global_step": 192, "epoch": 0, "lr": 9.999999874174511e-05} +{"train_loss": 0.4627310633659363, "train_loss_bc": 0.46136975288391113, "train_loss_llm": 0.1361299455165863, "grad_norm": 0.13606008887290955, "global_step": 193, "epoch": 0, "lr": 9.999999874174511e-05} +{"train_loss": 0.3711932301521301, "train_loss_bc": 0.36965829133987427, "train_loss_llm": 0.1534944474697113, "grad_norm": 0.1672300398349762, "global_step": 194, "epoch": 0, "lr": 9.999999874174511e-05} +{"train_loss": 0.4457559287548065, "train_loss_bc": 0.4441356062889099, "train_loss_llm": 0.16203254461288452, "grad_norm": 0.19516238570213318, "global_step": 195, "epoch": 0, "lr": 9.999999874174511e-05} +{"train_loss": 0.3456268906593323, "train_loss_bc": 0.3440917730331421, "train_loss_llm": 0.15351155400276184, "grad_norm": 0.1214531660079956, "global_step": 196, "epoch": 0, "lr": 9.999999874174511e-05} +{"train_loss": 0.49130669236183167, "train_loss_bc": 0.4898536205291748, "train_loss_llm": 0.145308256149292, "grad_norm": 0.12953642010688782, "global_step": 197, "epoch": 0, "lr": 9.999999874174511e-05} +{"train_loss": 0.4904896318912506, "train_loss_bc": 0.4889376163482666, "train_loss_llm": 0.1552022397518158, "grad_norm": 0.24106130003929138, "global_step": 198, "epoch": 0, "lr": 9.999999874174511e-05} +{"train_loss": 0.5558136701583862, "train_loss_bc": 0.5542365908622742, "train_loss_llm": 0.15770575404167175, "grad_norm": 0.23344939947128296, "global_step": 199, "epoch": 0, "lr": 9.999999874174511e-05} +{"train_loss": 0.5036170482635498, "train_loss_bc": 0.5018457770347595, "train_loss_llm": 0.177126944065094, "grad_norm": 0.2206622064113617, "global_step": 200, "epoch": 0, "lr": 9.999999856838556e-05} +{"train_loss": 0.49838924407958984, "train_loss_bc": 0.4970696270465851, "train_loss_llm": 0.13196173310279846, "grad_norm": 0.1284995973110199, "global_step": 201, "epoch": 0, "lr": 9.999999856838556e-05} +{"train_loss": 0.4288214147090912, "train_loss_bc": 0.4277445673942566, "train_loss_llm": 0.10768568515777588, "grad_norm": 0.09020152688026428, "global_step": 202, "epoch": 0, "lr": 9.999999856838556e-05} +{"train_loss": 0.47157788276672363, "train_loss_bc": 0.47020071744918823, "train_loss_llm": 0.13771778345108032, "grad_norm": 0.07819076627492905, "global_step": 203, "epoch": 0, "lr": 9.999999856838556e-05} +{"train_loss": 0.5295555591583252, "train_loss_bc": 0.5277771949768066, "train_loss_llm": 0.17783686518669128, "grad_norm": 0.13658098876476288, "global_step": 204, "epoch": 0, "lr": 9.999999856838556e-05} +{"train_loss": 0.5104283094406128, "train_loss_bc": 0.5089181065559387, "train_loss_llm": 0.15102159976959229, "grad_norm": 0.22971421480178833, "global_step": 205, "epoch": 0, "lr": 9.999999856838556e-05} +{"train_loss": 0.5116090774536133, "train_loss_bc": 0.5101298093795776, "train_loss_llm": 0.14792749285697937, "grad_norm": 0.2055637389421463, "global_step": 206, "epoch": 0, "lr": 9.999999856838556e-05} +{"train_loss": 0.5849693417549133, "train_loss_bc": 0.583702027797699, "train_loss_llm": 0.1267300248146057, "grad_norm": 0.26533693075180054, "global_step": 207, "epoch": 0, "lr": 9.999999856838556e-05} +{"train_loss": 0.40312016010284424, "train_loss_bc": 0.40157195925712585, "train_loss_llm": 0.15481889247894287, "grad_norm": 0.39653781056404114, "global_step": 208, "epoch": 0, "lr": 9.99999983838415e-05} +{"train_loss": 0.3709404766559601, "train_loss_bc": 0.3697349727153778, "train_loss_llm": 0.120551198720932, "grad_norm": 0.10661180317401886, "global_step": 209, "epoch": 0, "lr": 9.99999983838415e-05} +{"train_loss": 0.3831084072589874, "train_loss_bc": 0.38187795877456665, "train_loss_llm": 0.123043954372406, "grad_norm": 0.1655275970697403, "global_step": 210, "epoch": 0, "lr": 9.99999983838415e-05} +{"train_loss": 0.4003806412220001, "train_loss_bc": 0.3987475037574768, "train_loss_llm": 0.16331452131271362, "grad_norm": 0.28441521525382996, "global_step": 211, "epoch": 0, "lr": 9.99999983838415e-05} +{"train_loss": 0.4688017964363098, "train_loss_bc": 0.46736764907836914, "train_loss_llm": 0.14341410994529724, "grad_norm": 0.3773496448993683, "global_step": 212, "epoch": 0, "lr": 9.99999983838415e-05} +{"train_loss": 0.37370193004608154, "train_loss_bc": 0.3722708821296692, "train_loss_llm": 0.14310383796691895, "grad_norm": 0.3253101110458374, "global_step": 213, "epoch": 0, "lr": 9.99999983838415e-05} +{"train_loss": 0.2955486476421356, "train_loss_bc": 0.29423728585243225, "train_loss_llm": 0.131135493516922, "grad_norm": 0.41460081934928894, "global_step": 214, "epoch": 0, "lr": 9.99999983838415e-05} +{"train_loss": 0.524498462677002, "train_loss_bc": 0.5235158205032349, "train_loss_llm": 0.09826689958572388, "grad_norm": 0.4054737091064453, "global_step": 215, "epoch": 0, "lr": 9.99999983838415e-05} +{"train_loss": 0.44162464141845703, "train_loss_bc": 0.44016754627227783, "train_loss_llm": 0.1457107961177826, "grad_norm": 0.39613816142082214, "global_step": 216, "epoch": 0, "lr": 9.999999818811297e-05} +{"train_loss": 0.5350319147109985, "train_loss_bc": 0.533849835395813, "train_loss_llm": 0.11820894479751587, "grad_norm": 0.10832974314689636, "global_step": 217, "epoch": 0, "lr": 9.999999818811297e-05} +{"train_loss": 0.4740488827228546, "train_loss_bc": 0.4726904034614563, "train_loss_llm": 0.13584721088409424, "grad_norm": 0.19226783514022827, "global_step": 218, "epoch": 0, "lr": 9.999999818811297e-05} +{"train_loss": 0.5852010250091553, "train_loss_bc": 0.5834216475486755, "train_loss_llm": 0.17793679237365723, "grad_norm": 0.21397559344768524, "global_step": 219, "epoch": 0, "lr": 9.999999818811297e-05} +{"train_loss": 0.5124484896659851, "train_loss_bc": 0.5110262632369995, "train_loss_llm": 0.14222225546836853, "grad_norm": 0.21491438150405884, "global_step": 220, "epoch": 0, "lr": 9.999999818811297e-05} +{"train_loss": 0.4186595380306244, "train_loss_bc": 0.41726022958755493, "train_loss_llm": 0.13992974162101746, "grad_norm": 0.2644885778427124, "global_step": 221, "epoch": 0, "lr": 9.999999818811297e-05} +{"train_loss": 0.3839101195335388, "train_loss_bc": 0.38252806663513184, "train_loss_llm": 0.13820472359657288, "grad_norm": 0.2239675670862198, "global_step": 222, "epoch": 0, "lr": 9.999999818811297e-05} +{"train_loss": 0.2865701913833618, "train_loss_bc": 0.28537362813949585, "train_loss_llm": 0.11965581774711609, "grad_norm": 0.24009735882282257, "global_step": 223, "epoch": 0, "lr": 9.999999818811297e-05} +{"train_loss": 0.46241065859794617, "train_loss_bc": 0.4612131714820862, "train_loss_llm": 0.11974966526031494, "grad_norm": 0.33678242564201355, "global_step": 224, "epoch": 0, "lr": 9.999999798119993e-05} +{"train_loss": 0.4807482659816742, "train_loss_bc": 0.4795793294906616, "train_loss_llm": 0.11689290404319763, "grad_norm": 0.18909280002117157, "global_step": 225, "epoch": 0, "lr": 9.999999798119993e-05} +{"train_loss": 0.4647921919822693, "train_loss_bc": 0.4633268117904663, "train_loss_llm": 0.14653772115707397, "grad_norm": 0.20287412405014038, "global_step": 226, "epoch": 0, "lr": 9.999999798119993e-05} +{"train_loss": 0.3235853910446167, "train_loss_bc": 0.3223698139190674, "train_loss_llm": 0.12155833840370178, "grad_norm": 0.25691166520118713, "global_step": 227, "epoch": 0, "lr": 9.999999798119993e-05} +{"train_loss": 0.48162052035331726, "train_loss_bc": 0.4802006781101227, "train_loss_llm": 0.1419827938079834, "grad_norm": 0.1927158385515213, "global_step": 228, "epoch": 0, "lr": 9.999999798119993e-05} +{"train_loss": 0.4351486265659332, "train_loss_bc": 0.43379366397857666, "train_loss_llm": 0.13549718260765076, "grad_norm": 0.20294009149074554, "global_step": 229, "epoch": 0, "lr": 9.999999798119993e-05} +{"train_loss": 0.4667045772075653, "train_loss_bc": 0.46539777517318726, "train_loss_llm": 0.13067957758903503, "grad_norm": 0.2676793336868286, "global_step": 230, "epoch": 0, "lr": 9.999999798119993e-05} +{"train_loss": 0.4963558614253998, "train_loss_bc": 0.49491092562675476, "train_loss_llm": 0.14449357986450195, "grad_norm": 0.28663256764411926, "global_step": 231, "epoch": 0, "lr": 9.999999798119993e-05} +{"train_loss": 0.29978886246681213, "train_loss_bc": 0.29905274510383606, "train_loss_llm": 0.07361292839050293, "grad_norm": 0.3045455515384674, "global_step": 232, "epoch": 0, "lr": 9.999999776310242e-05} +{"train_loss": 0.4061849117279053, "train_loss_bc": 0.40459394454956055, "train_loss_llm": 0.15909603238105774, "grad_norm": 0.08643153309822083, "global_step": 233, "epoch": 0, "lr": 9.999999776310242e-05} +{"train_loss": 0.5327658653259277, "train_loss_bc": 0.5314740538597107, "train_loss_llm": 0.12917914986610413, "grad_norm": 0.1096641793847084, "global_step": 234, "epoch": 0, "lr": 9.999999776310242e-05} +{"train_loss": 0.46701255440711975, "train_loss_bc": 0.4656456708908081, "train_loss_llm": 0.13668766617774963, "grad_norm": 0.18364651501178741, "global_step": 235, "epoch": 0, "lr": 9.999999776310242e-05} +{"train_loss": 0.4609335958957672, "train_loss_bc": 0.45971813797950745, "train_loss_llm": 0.12154534459114075, "grad_norm": 0.17089353501796722, "global_step": 236, "epoch": 0, "lr": 9.999999776310242e-05} +{"train_loss": 0.40160396695137024, "train_loss_bc": 0.40031543374061584, "train_loss_llm": 0.12885281443595886, "grad_norm": 0.20633822679519653, "global_step": 237, "epoch": 0, "lr": 9.999999776310242e-05} +{"train_loss": 0.6086274981498718, "train_loss_bc": 0.6072677969932556, "train_loss_llm": 0.13597145676612854, "grad_norm": 0.2718789875507355, "global_step": 238, "epoch": 0, "lr": 9.999999776310242e-05} +{"train_loss": 0.49996188282966614, "train_loss_bc": 0.4983826279640198, "train_loss_llm": 0.1579260230064392, "grad_norm": 0.33263587951660156, "global_step": 239, "epoch": 0, "lr": 9.999999776310242e-05} +{"train_loss": 0.5027485489845276, "train_loss_bc": 0.501329779624939, "train_loss_llm": 0.14187422394752502, "grad_norm": 0.35866373777389526, "global_step": 240, "epoch": 0, "lr": 9.999999753382043e-05} +{"train_loss": 0.4826712906360626, "train_loss_bc": 0.4813128709793091, "train_loss_llm": 0.13584184646606445, "grad_norm": 0.12715676426887512, "global_step": 241, "epoch": 0, "lr": 9.999999753382043e-05} +{"train_loss": 0.35669106245040894, "train_loss_bc": 0.3554719090461731, "train_loss_llm": 0.12191447615623474, "grad_norm": 0.11628498136997223, "global_step": 242, "epoch": 0, "lr": 9.999999753382043e-05} +{"train_loss": 0.4046419560909271, "train_loss_bc": 0.4031323194503784, "train_loss_llm": 0.15096458792686462, "grad_norm": 0.1684313267469406, "global_step": 243, "epoch": 0, "lr": 9.999999753382043e-05} +{"train_loss": 0.5056776404380798, "train_loss_bc": 0.5038183927536011, "train_loss_llm": 0.18592357635498047, "grad_norm": 0.38399288058280945, "global_step": 244, "epoch": 0, "lr": 9.999999753382043e-05} +{"train_loss": 0.37411290407180786, "train_loss_bc": 0.3727595806121826, "train_loss_llm": 0.13533315062522888, "grad_norm": 0.407470703125, "global_step": 245, "epoch": 0, "lr": 9.999999753382043e-05} +{"train_loss": 0.39478400349617004, "train_loss_bc": 0.3932441771030426, "train_loss_llm": 0.15398159623146057, "grad_norm": 0.49406519532203674, "global_step": 246, "epoch": 0, "lr": 9.999999753382043e-05} +{"train_loss": 0.34229400753974915, "train_loss_bc": 0.3409374952316284, "train_loss_llm": 0.13565048575401306, "grad_norm": 0.5884677171707153, "global_step": 247, "epoch": 0, "lr": 9.999999753382043e-05} +{"train_loss": 0.4749721586704254, "train_loss_bc": 0.4737721383571625, "train_loss_llm": 0.12000155448913574, "grad_norm": 0.47413018345832825, "global_step": 248, "epoch": 0, "lr": 9.999999729335394e-05} +{"train_loss": 0.4578547775745392, "train_loss_bc": 0.4564540386199951, "train_loss_llm": 0.14007365703582764, "grad_norm": 0.07279176265001297, "global_step": 249, "epoch": 0, "lr": 9.999999729335394e-05} +{"train_loss": 0.3324807584285736, "train_loss_bc": 0.33115747570991516, "train_loss_llm": 0.1323278844356537, "grad_norm": 0.09465418756008148, "global_step": 250, "epoch": 0, "lr": 9.999999729335394e-05} +{"train_loss": 0.48672300577163696, "train_loss_bc": 0.48546528816223145, "train_loss_llm": 0.12577283382415771, "grad_norm": 0.14441658556461334, "global_step": 251, "epoch": 0, "lr": 9.999999729335394e-05} +{"train_loss": 0.38449743390083313, "train_loss_bc": 0.38310763239860535, "train_loss_llm": 0.13898122310638428, "grad_norm": 0.17639993131160736, "global_step": 252, "epoch": 0, "lr": 9.999999729335394e-05} +{"train_loss": 0.3338666260242462, "train_loss_bc": 0.33266425132751465, "train_loss_llm": 0.12023752927780151, "grad_norm": 0.269575297832489, "global_step": 253, "epoch": 0, "lr": 9.999999729335394e-05} +{"train_loss": 0.3827843964099884, "train_loss_bc": 0.38170522451400757, "train_loss_llm": 0.10791599750518799, "grad_norm": 0.2737504243850708, "global_step": 254, "epoch": 0, "lr": 9.999999729335394e-05} +{"train_loss": 0.31307464838027954, "train_loss_bc": 0.31137993931770325, "train_loss_llm": 0.16947093605995178, "grad_norm": 0.28003835678100586, "global_step": 255, "epoch": 0, "lr": 9.999999729335394e-05} +{"train_loss": 0.4258021414279938, "train_loss_bc": 0.4239563047885895, "train_loss_llm": 0.18458488583564758, "grad_norm": 0.30002522468566895, "global_step": 256, "epoch": 0, "lr": 9.999999704170296e-05} +{"train_loss": 0.41750580072402954, "train_loss_bc": 0.416411817073822, "train_loss_llm": 0.10939714312553406, "grad_norm": 0.07486779242753983, "global_step": 257, "epoch": 0, "lr": 9.999999704170296e-05} +{"train_loss": 0.4414028227329254, "train_loss_bc": 0.4396103620529175, "train_loss_llm": 0.1792447865009308, "grad_norm": 0.12679605185985565, "global_step": 258, "epoch": 0, "lr": 9.999999704170296e-05} +{"train_loss": 0.4474702775478363, "train_loss_bc": 0.44621071219444275, "train_loss_llm": 0.12595561146736145, "grad_norm": 0.21293529868125916, "global_step": 259, "epoch": 0, "lr": 9.999999704170296e-05} +{"train_loss": 0.45398786664009094, "train_loss_bc": 0.4522455036640167, "train_loss_llm": 0.17423567175865173, "grad_norm": 0.19735704362392426, "global_step": 260, "epoch": 0, "lr": 9.999999704170296e-05} +{"train_loss": 0.45205387473106384, "train_loss_bc": 0.45062530040740967, "train_loss_llm": 0.14285743236541748, "grad_norm": 0.221365287899971, "global_step": 261, "epoch": 0, "lr": 9.999999704170296e-05} +{"train_loss": 0.3651925027370453, "train_loss_bc": 0.3635643720626831, "train_loss_llm": 0.16281402111053467, "grad_norm": 0.24405691027641296, "global_step": 262, "epoch": 0, "lr": 9.999999704170296e-05} +{"train_loss": 0.5158424377441406, "train_loss_bc": 0.5147004127502441, "train_loss_llm": 0.1142028272151947, "grad_norm": 0.29550719261169434, "global_step": 263, "epoch": 0, "lr": 9.999999704170296e-05} +{"train_loss": 0.501900315284729, "train_loss_bc": 0.5004952549934387, "train_loss_llm": 0.1405060887336731, "grad_norm": 0.22012563049793243, "global_step": 264, "epoch": 0, "lr": 9.999999677886748e-05} +{"train_loss": 0.5547175407409668, "train_loss_bc": 0.5533242225646973, "train_loss_llm": 0.13933426141738892, "grad_norm": 0.1802000254392624, "global_step": 265, "epoch": 0, "lr": 9.999999677886748e-05} +{"train_loss": 0.4960016906261444, "train_loss_bc": 0.4949193596839905, "train_loss_llm": 0.1082327663898468, "grad_norm": 0.21128597855567932, "global_step": 266, "epoch": 0, "lr": 9.999999677886748e-05} +{"train_loss": 0.38849759101867676, "train_loss_bc": 0.38733458518981934, "train_loss_llm": 0.11630144715309143, "grad_norm": 0.32609131932258606, "global_step": 267, "epoch": 0, "lr": 9.999999677886748e-05} +{"train_loss": 0.5069462060928345, "train_loss_bc": 0.5055183172225952, "train_loss_llm": 0.14278662204742432, "grad_norm": 0.35645654797554016, "global_step": 268, "epoch": 0, "lr": 9.999999677886748e-05} +{"train_loss": 0.3498342037200928, "train_loss_bc": 0.3481409549713135, "train_loss_llm": 0.1693248152732849, "grad_norm": 0.3803897500038147, "global_step": 269, "epoch": 0, "lr": 9.999999677886748e-05} +{"train_loss": 0.4138376712799072, "train_loss_bc": 0.41241973638534546, "train_loss_llm": 0.1417936086654663, "grad_norm": 0.39421358704566956, "global_step": 270, "epoch": 0, "lr": 9.999999677886748e-05} +{"train_loss": 0.40116086602211, "train_loss_bc": 0.3997952342033386, "train_loss_llm": 0.13656291365623474, "grad_norm": 0.4513140916824341, "global_step": 271, "epoch": 0, "lr": 9.999999677886748e-05} +{"train_loss": 0.3253784775733948, "train_loss_bc": 0.32437634468078613, "train_loss_llm": 0.10021468997001648, "grad_norm": 0.4629686176776886, "global_step": 272, "epoch": 0, "lr": 9.999999650484755e-05} +{"train_loss": 0.4688898026943207, "train_loss_bc": 0.4670729637145996, "train_loss_llm": 0.1816847026348114, "grad_norm": 0.19157733023166656, "global_step": 273, "epoch": 0, "lr": 9.999999650484755e-05} +{"train_loss": 0.44128838181495667, "train_loss_bc": 0.4398792088031769, "train_loss_llm": 0.14091730117797852, "grad_norm": 0.29975634813308716, "global_step": 274, "epoch": 0, "lr": 9.999999650484755e-05} +{"train_loss": 0.5226980447769165, "train_loss_bc": 0.5209264159202576, "train_loss_llm": 0.177162766456604, "grad_norm": 0.20200972259044647, "global_step": 275, "epoch": 0, "lr": 9.999999650484755e-05} +{"train_loss": 0.488549143075943, "train_loss_bc": 0.48716187477111816, "train_loss_llm": 0.1387256383895874, "grad_norm": 0.27788445353507996, "global_step": 276, "epoch": 0, "lr": 9.999999650484755e-05} +{"train_loss": 0.5938391089439392, "train_loss_bc": 0.5919771194458008, "train_loss_llm": 0.18619847297668457, "grad_norm": 0.3419470489025116, "global_step": 277, "epoch": 0, "lr": 9.999999650484755e-05} +{"train_loss": 0.34958070516586304, "train_loss_bc": 0.34854161739349365, "train_loss_llm": 0.10390952229499817, "grad_norm": 0.43380007147789, "global_step": 278, "epoch": 0, "lr": 9.999999650484755e-05} +{"train_loss": 0.3573320209980011, "train_loss_bc": 0.35602596402168274, "train_loss_llm": 0.13060519099235535, "grad_norm": 0.41530922055244446, "global_step": 279, "epoch": 0, "lr": 9.999999650484755e-05} +{"train_loss": 0.4187597334384918, "train_loss_bc": 0.417165070772171, "train_loss_llm": 0.1594652235507965, "grad_norm": 0.4791331887245178, "global_step": 280, "epoch": 0, "lr": 9.99999962196431e-05} +{"train_loss": 0.489306777715683, "train_loss_bc": 0.4876519739627838, "train_loss_llm": 0.16548079252243042, "grad_norm": 0.13929283618927002, "global_step": 281, "epoch": 0, "lr": 9.99999962196431e-05} +{"train_loss": 0.45168349146842957, "train_loss_bc": 0.4503023028373718, "train_loss_llm": 0.13811764121055603, "grad_norm": 0.15657302737236023, "global_step": 282, "epoch": 0, "lr": 9.99999962196431e-05} +{"train_loss": 0.46502405405044556, "train_loss_bc": 0.4637814462184906, "train_loss_llm": 0.12426182627677917, "grad_norm": 0.12926176190376282, "global_step": 283, "epoch": 0, "lr": 9.99999962196431e-05} +{"train_loss": 0.5104905962944031, "train_loss_bc": 0.5086830854415894, "train_loss_llm": 0.18075108528137207, "grad_norm": 0.13587743043899536, "global_step": 284, "epoch": 0, "lr": 9.99999962196431e-05} +{"train_loss": 0.39535877108573914, "train_loss_bc": 0.39384230971336365, "train_loss_llm": 0.15164539217948914, "grad_norm": 0.22818033397197723, "global_step": 285, "epoch": 0, "lr": 9.99999962196431e-05} +{"train_loss": 0.5428739190101624, "train_loss_bc": 0.5416086912155151, "train_loss_llm": 0.12652459740638733, "grad_norm": 0.25617656111717224, "global_step": 286, "epoch": 0, "lr": 9.99999962196431e-05} +{"train_loss": 0.470071941614151, "train_loss_bc": 0.46860024333000183, "train_loss_llm": 0.14716997742652893, "grad_norm": 0.35483404994010925, "global_step": 287, "epoch": 0, "lr": 9.99999962196431e-05} +{"train_loss": 0.49615737795829773, "train_loss_bc": 0.49484026432037354, "train_loss_llm": 0.13171163201332092, "grad_norm": 0.38000038266181946, "global_step": 288, "epoch": 0, "lr": 9.999999592325419e-05} +{"train_loss": 0.40782466530799866, "train_loss_bc": 0.4064186215400696, "train_loss_llm": 0.140605628490448, "grad_norm": 0.17697758972644806, "global_step": 289, "epoch": 0, "lr": 9.999999592325419e-05} +{"train_loss": 0.41140982508659363, "train_loss_bc": 0.4102866053581238, "train_loss_llm": 0.1123214066028595, "grad_norm": 0.25229179859161377, "global_step": 290, "epoch": 0, "lr": 9.999999592325419e-05} +{"train_loss": 0.5142717957496643, "train_loss_bc": 0.512864887714386, "train_loss_llm": 0.14068946242332458, "grad_norm": 0.3419853448867798, "global_step": 291, "epoch": 0, "lr": 9.999999592325419e-05} +{"train_loss": 0.32486462593078613, "train_loss_bc": 0.3237345814704895, "train_loss_llm": 0.11300429701805115, "grad_norm": 0.3385412096977234, "global_step": 292, "epoch": 0, "lr": 9.999999592325419e-05} +{"train_loss": 0.38490915298461914, "train_loss_bc": 0.3832823932170868, "train_loss_llm": 0.16267475485801697, "grad_norm": 0.3462136387825012, "global_step": 293, "epoch": 0, "lr": 9.999999592325419e-05} +{"train_loss": 0.5673600435256958, "train_loss_bc": 0.5659370422363281, "train_loss_llm": 0.14229875802993774, "grad_norm": 0.5142638683319092, "global_step": 294, "epoch": 0, "lr": 9.999999592325419e-05} +{"train_loss": 0.3522787392139435, "train_loss_bc": 0.3508474826812744, "train_loss_llm": 0.14312613010406494, "grad_norm": 0.5230362415313721, "global_step": 295, "epoch": 0, "lr": 9.999999592325419e-05} +{"train_loss": 0.536923885345459, "train_loss_bc": 0.5354461073875427, "train_loss_llm": 0.14777785539627075, "grad_norm": 0.3804051876068115, "global_step": 296, "epoch": 0, "lr": 9.999999561568078e-05} +{"train_loss": 0.43360280990600586, "train_loss_bc": 0.43208807706832886, "train_loss_llm": 0.15147429704666138, "grad_norm": 0.12204445898532867, "global_step": 297, "epoch": 0, "lr": 9.999999561568078e-05} +{"train_loss": 0.4311789870262146, "train_loss_bc": 0.43001917004585266, "train_loss_llm": 0.11598166823387146, "grad_norm": 0.11353069543838501, "global_step": 298, "epoch": 0, "lr": 9.999999561568078e-05} +{"train_loss": 0.3661663234233856, "train_loss_bc": 0.3649553954601288, "train_loss_llm": 0.1210927963256836, "grad_norm": 0.15506236255168915, "global_step": 299, "epoch": 0, "lr": 9.999999561568078e-05} +{"train_loss": 0.4559875428676605, "train_loss_bc": 0.45493921637535095, "train_loss_llm": 0.10483124852180481, "grad_norm": 0.18460023403167725, "global_step": 300, "epoch": 0, "lr": 9.999999561568078e-05} +{"train_loss": 0.5285307765007019, "train_loss_bc": 0.5273026823997498, "train_loss_llm": 0.12280848622322083, "grad_norm": 0.24207845330238342, "global_step": 301, "epoch": 0, "lr": 9.999999561568078e-05} +{"train_loss": 0.5566150546073914, "train_loss_bc": 0.5552654266357422, "train_loss_llm": 0.134964257478714, "grad_norm": 0.3185100853443146, "global_step": 302, "epoch": 0, "lr": 9.999999561568078e-05} +{"train_loss": 0.47669094800949097, "train_loss_bc": 0.47508424520492554, "train_loss_llm": 0.16067081689834595, "grad_norm": 0.3561716377735138, "global_step": 303, "epoch": 0, "lr": 9.999999561568078e-05} +{"train_loss": 0.41046255826950073, "train_loss_bc": 0.4091735780239105, "train_loss_llm": 0.1288984715938568, "grad_norm": 0.3982714116573334, "global_step": 304, "epoch": 0, "lr": 9.999999529692287e-05} +{"train_loss": 0.4458627700805664, "train_loss_bc": 0.44482651352882385, "train_loss_llm": 0.10362643003463745, "grad_norm": 0.09240652620792389, "global_step": 305, "epoch": 0, "lr": 9.999999529692287e-05} +{"train_loss": 0.33148279786109924, "train_loss_bc": 0.3304671049118042, "train_loss_llm": 0.10156944394111633, "grad_norm": 0.1285073608160019, "global_step": 306, "epoch": 0, "lr": 9.999999529692287e-05} +{"train_loss": 0.4878723621368408, "train_loss_bc": 0.48639005422592163, "train_loss_llm": 0.14823070168495178, "grad_norm": 0.1883501261472702, "global_step": 307, "epoch": 0, "lr": 9.999999529692287e-05} +{"train_loss": 0.554612398147583, "train_loss_bc": 0.5532375574111938, "train_loss_llm": 0.1374816596508026, "grad_norm": 0.1642327457666397, "global_step": 308, "epoch": 0, "lr": 9.999999529692287e-05} +{"train_loss": 0.28455621004104614, "train_loss_bc": 0.2834703028202057, "train_loss_llm": 0.108591228723526, "grad_norm": 0.16397428512573242, "global_step": 309, "epoch": 0, "lr": 9.999999529692287e-05} +{"train_loss": 0.4632170796394348, "train_loss_bc": 0.4620139002799988, "train_loss_llm": 0.12031927704811096, "grad_norm": 0.16347527503967285, "global_step": 310, "epoch": 0, "lr": 9.999999529692287e-05} +{"train_loss": 0.32091233134269714, "train_loss_bc": 0.31957748532295227, "train_loss_llm": 0.13348472118377686, "grad_norm": 0.1995544135570526, "global_step": 311, "epoch": 0, "lr": 9.999999529692287e-05} +{"train_loss": 0.442105233669281, "train_loss_bc": 0.4413487911224365, "train_loss_llm": 0.0756453275680542, "grad_norm": 0.2321440577507019, "global_step": 312, "epoch": 0, "lr": 9.99999949669805e-05} +{"train_loss": 0.37721845507621765, "train_loss_bc": 0.3763888478279114, "train_loss_llm": 0.08296006917953491, "grad_norm": 0.11543812602758408, "global_step": 313, "epoch": 0, "lr": 9.99999949669805e-05} +{"train_loss": 0.3989948332309723, "train_loss_bc": 0.39769935607910156, "train_loss_llm": 0.12954899668693542, "grad_norm": 0.1061362475156784, "global_step": 314, "epoch": 0, "lr": 9.99999949669805e-05} +{"train_loss": 0.3919609785079956, "train_loss_bc": 0.3908499479293823, "train_loss_llm": 0.11110222339630127, "grad_norm": 0.11855069547891617, "global_step": 315, "epoch": 0, "lr": 9.99999949669805e-05} +{"train_loss": 0.4353964328765869, "train_loss_bc": 0.4343418478965759, "train_loss_llm": 0.10545811057090759, "grad_norm": 0.164484441280365, "global_step": 316, "epoch": 0, "lr": 9.99999949669805e-05} +{"train_loss": 0.48694831132888794, "train_loss_bc": 0.4856909215450287, "train_loss_llm": 0.12574037909507751, "grad_norm": 0.20515409111976624, "global_step": 317, "epoch": 0, "lr": 9.99999949669805e-05} +{"train_loss": 0.38299235701560974, "train_loss_bc": 0.3817113935947418, "train_loss_llm": 0.12809523940086365, "grad_norm": 0.2341000735759735, "global_step": 318, "epoch": 0, "lr": 9.99999949669805e-05} +{"train_loss": 0.4417354166507721, "train_loss_bc": 0.4405592679977417, "train_loss_llm": 0.11761382222175598, "grad_norm": 0.2308935821056366, "global_step": 319, "epoch": 0, "lr": 9.99999949669805e-05} +{"train_loss": 0.5682020783424377, "train_loss_bc": 0.5667705535888672, "train_loss_llm": 0.14315065741539001, "grad_norm": 0.25292038917541504, "global_step": 320, "epoch": 0, "lr": 9.999999462585362e-05} +{"train_loss": 0.4116712808609009, "train_loss_bc": 0.4102035164833069, "train_loss_llm": 0.14677509665489197, "grad_norm": 0.07832875102758408, "global_step": 321, "epoch": 0, "lr": 9.999999462585362e-05} +{"train_loss": 0.37415629625320435, "train_loss_bc": 0.3727145195007324, "train_loss_llm": 0.14417880773544312, "grad_norm": 0.12110793590545654, "global_step": 322, "epoch": 0, "lr": 9.999999462585362e-05} +{"train_loss": 0.4695186913013458, "train_loss_bc": 0.46824967861175537, "train_loss_llm": 0.1269010603427887, "grad_norm": 0.17378371953964233, "global_step": 323, "epoch": 0, "lr": 9.999999462585362e-05} +{"train_loss": 0.5012226104736328, "train_loss_bc": 0.49974721670150757, "train_loss_llm": 0.14754056930541992, "grad_norm": 0.2674083113670349, "global_step": 324, "epoch": 0, "lr": 9.999999462585362e-05} +{"train_loss": 0.3409188389778137, "train_loss_bc": 0.33989137411117554, "train_loss_llm": 0.10274767875671387, "grad_norm": 0.2569374740123749, "global_step": 325, "epoch": 0, "lr": 9.999999462585362e-05} +{"train_loss": 0.36540529131889343, "train_loss_bc": 0.3638814687728882, "train_loss_llm": 0.15238291025161743, "grad_norm": 0.26285725831985474, "global_step": 326, "epoch": 0, "lr": 9.999999462585362e-05} +{"train_loss": 0.4318124055862427, "train_loss_bc": 0.43092700839042664, "train_loss_llm": 0.08854120969772339, "grad_norm": 0.40445253252983093, "global_step": 327, "epoch": 0, "lr": 9.999999462585362e-05} +{"train_loss": 0.4670446217060089, "train_loss_bc": 0.46574825048446655, "train_loss_llm": 0.1296379268169403, "grad_norm": 0.43074831366539, "global_step": 328, "epoch": 0, "lr": 9.999999427354226e-05} +{"train_loss": 0.5307365655899048, "train_loss_bc": 0.5290765762329102, "train_loss_llm": 0.16599786281585693, "grad_norm": 0.06676384806632996, "global_step": 329, "epoch": 0, "lr": 9.999999427354226e-05} +{"train_loss": 0.4938928484916687, "train_loss_bc": 0.49269500374794006, "train_loss_llm": 0.1197856068611145, "grad_norm": 0.16840828955173492, "global_step": 330, "epoch": 0, "lr": 9.999999427354226e-05} +{"train_loss": 0.46303075551986694, "train_loss_bc": 0.46140289306640625, "train_loss_llm": 0.16278541088104248, "grad_norm": 0.21267223358154297, "global_step": 331, "epoch": 0, "lr": 9.999999427354226e-05} +{"train_loss": 0.44752928614616394, "train_loss_bc": 0.44644999504089355, "train_loss_llm": 0.10793039202690125, "grad_norm": 0.38221776485443115, "global_step": 332, "epoch": 0, "lr": 9.999999427354226e-05} +{"train_loss": 0.48328879475593567, "train_loss_bc": 0.4815678000450134, "train_loss_llm": 0.17209839820861816, "grad_norm": 0.5120506286621094, "global_step": 333, "epoch": 0, "lr": 9.999999427354226e-05} +{"train_loss": 0.38235434889793396, "train_loss_bc": 0.3813237249851227, "train_loss_llm": 0.10306346416473389, "grad_norm": 0.5296585559844971, "global_step": 334, "epoch": 0, "lr": 9.999999427354226e-05} +{"train_loss": 0.3251821994781494, "train_loss_bc": 0.3238959312438965, "train_loss_llm": 0.12862813472747803, "grad_norm": 0.5403621792793274, "global_step": 335, "epoch": 0, "lr": 9.999999427354226e-05} +{"train_loss": 0.6832368969917297, "train_loss_bc": 0.6814543604850769, "train_loss_llm": 0.17825376987457275, "grad_norm": 0.6231938004493713, "global_step": 336, "epoch": 0, "lr": 9.999999391004642e-05} +{"train_loss": 0.5653870701789856, "train_loss_bc": 0.5640597939491272, "train_loss_llm": 0.1327296793460846, "grad_norm": 0.1755756288766861, "global_step": 337, "epoch": 0, "lr": 9.999999391004642e-05} +{"train_loss": 0.44427576661109924, "train_loss_bc": 0.44269993901252747, "train_loss_llm": 0.15758416056632996, "grad_norm": 0.23777222633361816, "global_step": 338, "epoch": 0, "lr": 9.999999391004642e-05} +{"train_loss": 0.3795701861381531, "train_loss_bc": 0.3779670298099518, "train_loss_llm": 0.16031423211097717, "grad_norm": 0.2144278734922409, "global_step": 339, "epoch": 0, "lr": 9.999999391004642e-05} +{"train_loss": 0.449429452419281, "train_loss_bc": 0.4485020339488983, "train_loss_llm": 0.09274080395698547, "grad_norm": 0.22950133681297302, "global_step": 340, "epoch": 0, "lr": 9.999999391004642e-05} +{"train_loss": 0.5058606863021851, "train_loss_bc": 0.5039888620376587, "train_loss_llm": 0.1871795356273651, "grad_norm": 0.24388182163238525, "global_step": 341, "epoch": 0, "lr": 9.999999391004642e-05} +{"train_loss": 0.3657059371471405, "train_loss_bc": 0.36405983567237854, "train_loss_llm": 0.16460970044136047, "grad_norm": 0.19768968224525452, "global_step": 342, "epoch": 0, "lr": 9.999999391004642e-05} +{"train_loss": 0.4090924859046936, "train_loss_bc": 0.4076713025569916, "train_loss_llm": 0.1421193778514862, "grad_norm": 0.20467603206634521, "global_step": 343, "epoch": 0, "lr": 9.999999391004642e-05} +{"train_loss": 0.37313520908355713, "train_loss_bc": 0.37163814902305603, "train_loss_llm": 0.14970451593399048, "grad_norm": 0.19586215913295746, "global_step": 344, "epoch": 0, "lr": 9.999999353536608e-05} +{"train_loss": 0.4713234603404999, "train_loss_bc": 0.46961912512779236, "train_loss_llm": 0.17043429613113403, "grad_norm": 0.10007829964160919, "global_step": 345, "epoch": 0, "lr": 9.999999353536608e-05} +{"train_loss": 0.5589625239372253, "train_loss_bc": 0.5577791929244995, "train_loss_llm": 0.11833113431930542, "grad_norm": 0.19543062150478363, "global_step": 346, "epoch": 0, "lr": 9.999999353536608e-05} +{"train_loss": 0.5354010462760925, "train_loss_bc": 0.5336629748344421, "train_loss_llm": 0.1738082468509674, "grad_norm": 0.2150149941444397, "global_step": 347, "epoch": 0, "lr": 9.999999353536608e-05} +{"train_loss": 0.38946643471717834, "train_loss_bc": 0.38801109790802, "train_loss_llm": 0.1455337107181549, "grad_norm": 0.3880280554294586, "global_step": 348, "epoch": 0, "lr": 9.999999353536608e-05} +{"train_loss": 0.526273250579834, "train_loss_bc": 0.5244959592819214, "train_loss_llm": 0.1777302324771881, "grad_norm": 0.42013049125671387, "global_step": 349, "epoch": 0, "lr": 9.999999353536608e-05} +{"train_loss": 0.42479225993156433, "train_loss_bc": 0.4234405755996704, "train_loss_llm": 0.1351698637008667, "grad_norm": 0.4463967978954315, "global_step": 350, "epoch": 0, "lr": 9.999999353536608e-05} +{"train_loss": 0.4110559821128845, "train_loss_bc": 0.4097674489021301, "train_loss_llm": 0.12885206937789917, "grad_norm": 0.35627904534339905, "global_step": 351, "epoch": 0, "lr": 9.999999353536608e-05} +{"train_loss": 0.4681563973426819, "train_loss_bc": 0.4665173888206482, "train_loss_llm": 0.16390177607536316, "grad_norm": 0.3326302468776703, "global_step": 352, "epoch": 0, "lr": 9.999999314950125e-05} +{"train_loss": 0.3621272146701813, "train_loss_bc": 0.36100947856903076, "train_loss_llm": 0.11177214980125427, "grad_norm": 0.16894367337226868, "global_step": 353, "epoch": 0, "lr": 9.999999314950125e-05} +{"train_loss": 0.32545971870422363, "train_loss_bc": 0.32430005073547363, "train_loss_llm": 0.11596694588661194, "grad_norm": 0.22514688968658447, "global_step": 354, "epoch": 0, "lr": 9.999999314950125e-05} +{"train_loss": 0.4466659724712372, "train_loss_bc": 0.4453117549419403, "train_loss_llm": 0.13542085886001587, "grad_norm": 0.2908576428890228, "global_step": 355, "epoch": 0, "lr": 9.999999314950125e-05} +{"train_loss": 0.4447338879108429, "train_loss_bc": 0.4432002902030945, "train_loss_llm": 0.15335869789123535, "grad_norm": 0.33273985981941223, "global_step": 356, "epoch": 0, "lr": 9.999999314950125e-05} +{"train_loss": 0.3028169274330139, "train_loss_bc": 0.30134111642837524, "train_loss_llm": 0.1475812792778015, "grad_norm": 0.35265302658081055, "global_step": 357, "epoch": 0, "lr": 9.999999314950125e-05} +{"train_loss": 0.36328670382499695, "train_loss_bc": 0.3623049557209015, "train_loss_llm": 0.09817513823509216, "grad_norm": 0.3572205901145935, "global_step": 358, "epoch": 0, "lr": 9.999999314950125e-05} +{"train_loss": 0.5039355158805847, "train_loss_bc": 0.5025749206542969, "train_loss_llm": 0.1360584795475006, "grad_norm": 0.37342244386672974, "global_step": 359, "epoch": 0, "lr": 9.999999314950125e-05} +{"train_loss": 0.4145217537879944, "train_loss_bc": 0.41300877928733826, "train_loss_llm": 0.15129798650741577, "grad_norm": 0.3919113874435425, "global_step": 360, "epoch": 0, "lr": 9.999999275245196e-05} +{"train_loss": 0.35467657446861267, "train_loss_bc": 0.35366570949554443, "train_loss_llm": 0.10108518600463867, "grad_norm": 0.22939105331897736, "global_step": 361, "epoch": 0, "lr": 9.999999275245196e-05} +{"train_loss": 0.4531727731227875, "train_loss_bc": 0.4516623616218567, "train_loss_llm": 0.15104034543037415, "grad_norm": 0.18854081630706787, "global_step": 362, "epoch": 0, "lr": 9.999999275245196e-05} +{"train_loss": 0.380286306142807, "train_loss_bc": 0.3787105083465576, "train_loss_llm": 0.15757936239242554, "grad_norm": 0.2528596818447113, "global_step": 363, "epoch": 0, "lr": 9.999999275245196e-05} +{"train_loss": 0.4824751615524292, "train_loss_bc": 0.48106589913368225, "train_loss_llm": 0.14092671871185303, "grad_norm": 0.2946903109550476, "global_step": 364, "epoch": 0, "lr": 9.999999275245196e-05} +{"train_loss": 0.344907283782959, "train_loss_bc": 0.34375351667404175, "train_loss_llm": 0.11537817120552063, "grad_norm": 0.3084697127342224, "global_step": 365, "epoch": 0, "lr": 9.999999275245196e-05} +{"train_loss": 0.346515417098999, "train_loss_bc": 0.34504324197769165, "train_loss_llm": 0.147216796875, "grad_norm": 0.364883691072464, "global_step": 366, "epoch": 0, "lr": 9.999999275245196e-05} +{"train_loss": 0.3734190762042999, "train_loss_bc": 0.372480183839798, "train_loss_llm": 0.09388908743858337, "grad_norm": 0.35551688075065613, "global_step": 367, "epoch": 0, "lr": 9.999999275245196e-05} +{"train_loss": 0.3528594672679901, "train_loss_bc": 0.3514936566352844, "train_loss_llm": 0.13658156991004944, "grad_norm": 0.3454764783382416, "global_step": 368, "epoch": 0, "lr": 9.999999234421818e-05} +{"train_loss": 0.4187875986099243, "train_loss_bc": 0.4172857999801636, "train_loss_llm": 0.15018019080162048, "grad_norm": 0.07556066662073135, "global_step": 369, "epoch": 0, "lr": 9.999999234421818e-05} +{"train_loss": 0.3932890295982361, "train_loss_bc": 0.39230960607528687, "train_loss_llm": 0.09794139862060547, "grad_norm": 0.13089242577552795, "global_step": 370, "epoch": 0, "lr": 9.999999234421818e-05} +{"train_loss": 0.3365221619606018, "train_loss_bc": 0.33495989441871643, "train_loss_llm": 0.15622824430465698, "grad_norm": 0.13683158159255981, "global_step": 371, "epoch": 0, "lr": 9.999999234421818e-05} +{"train_loss": 0.46397870779037476, "train_loss_bc": 0.4623446762561798, "train_loss_llm": 0.16340431571006775, "grad_norm": 0.21105390787124634, "global_step": 372, "epoch": 0, "lr": 9.999999234421818e-05} +{"train_loss": 0.46956387162208557, "train_loss_bc": 0.4683939814567566, "train_loss_llm": 0.11698868870735168, "grad_norm": 0.16431595385074615, "global_step": 373, "epoch": 0, "lr": 9.999999234421818e-05} +{"train_loss": 0.5180339217185974, "train_loss_bc": 0.5169132947921753, "train_loss_llm": 0.11206221580505371, "grad_norm": 0.1885470151901245, "global_step": 374, "epoch": 0, "lr": 9.999999234421818e-05} +{"train_loss": 0.5634371042251587, "train_loss_bc": 0.5622462034225464, "train_loss_llm": 0.11908873915672302, "grad_norm": 0.29724830389022827, "global_step": 375, "epoch": 0, "lr": 9.999999234421818e-05} +{"train_loss": 0.5512478351593018, "train_loss_bc": 0.5498888492584229, "train_loss_llm": 0.1358986794948578, "grad_norm": 0.3737054169178009, "global_step": 376, "epoch": 0, "lr": 9.99999919247999e-05} +{"train_loss": 0.4260881841182709, "train_loss_bc": 0.4246520400047302, "train_loss_llm": 0.14361464977264404, "grad_norm": 0.1925584077835083, "global_step": 377, "epoch": 0, "lr": 9.99999919247999e-05} +{"train_loss": 0.37589308619499207, "train_loss_bc": 0.3741491436958313, "train_loss_llm": 0.1743943989276886, "grad_norm": 0.23194032907485962, "global_step": 378, "epoch": 0, "lr": 9.99999919247999e-05} +{"train_loss": 0.43548619747161865, "train_loss_bc": 0.4341698884963989, "train_loss_llm": 0.13163089752197266, "grad_norm": 0.28228095173835754, "global_step": 379, "epoch": 0, "lr": 9.99999919247999e-05} +{"train_loss": 0.28802403807640076, "train_loss_bc": 0.28748324513435364, "train_loss_llm": 0.05407911539077759, "grad_norm": 0.255656898021698, "global_step": 380, "epoch": 0, "lr": 9.99999919247999e-05} +{"train_loss": 0.439613401889801, "train_loss_bc": 0.43818920850753784, "train_loss_llm": 0.1424199342727661, "grad_norm": 0.38304033875465393, "global_step": 381, "epoch": 0, "lr": 9.99999919247999e-05} +{"train_loss": 0.44420501589775085, "train_loss_bc": 0.44263386726379395, "train_loss_llm": 0.1571148931980133, "grad_norm": 0.5148699283599854, "global_step": 382, "epoch": 0, "lr": 9.99999919247999e-05} +{"train_loss": 0.43274328112602234, "train_loss_bc": 0.4314499497413635, "train_loss_llm": 0.12933233380317688, "grad_norm": 0.5555089712142944, "global_step": 383, "epoch": 0, "lr": 9.99999919247999e-05} +{"train_loss": 0.2951040267944336, "train_loss_bc": 0.29395198822021484, "train_loss_llm": 0.11520266532897949, "grad_norm": 0.5536786913871765, "global_step": 384, "epoch": 0, "lr": 9.999999149419712e-05} +{"train_loss": 0.49426496028900146, "train_loss_bc": 0.4924396872520447, "train_loss_llm": 0.18252822756767273, "grad_norm": 0.12307372689247131, "global_step": 385, "epoch": 0, "lr": 9.999999149419712e-05} +{"train_loss": 0.3756541907787323, "train_loss_bc": 0.3745954632759094, "train_loss_llm": 0.1058732271194458, "grad_norm": 0.2264237254858017, "global_step": 386, "epoch": 0, "lr": 9.999999149419712e-05} +{"train_loss": 0.47481077909469604, "train_loss_bc": 0.4734797179698944, "train_loss_llm": 0.13310620188713074, "grad_norm": 0.22834548354148865, "global_step": 387, "epoch": 0, "lr": 9.999999149419712e-05} +{"train_loss": 0.3801238536834717, "train_loss_bc": 0.37886157631874084, "train_loss_llm": 0.12622791528701782, "grad_norm": 0.22366006672382355, "global_step": 388, "epoch": 0, "lr": 9.999999149419712e-05} +{"train_loss": 0.39786866307258606, "train_loss_bc": 0.3959633409976959, "train_loss_llm": 0.1905331313610077, "grad_norm": 0.37780389189720154, "global_step": 389, "epoch": 0, "lr": 9.999999149419712e-05} +{"train_loss": 0.304758220911026, "train_loss_bc": 0.30368131399154663, "train_loss_llm": 0.1076907217502594, "grad_norm": 0.3794277012348175, "global_step": 390, "epoch": 0, "lr": 9.999999149419712e-05} +{"train_loss": 0.4547385275363922, "train_loss_bc": 0.4532665014266968, "train_loss_llm": 0.1472022831439972, "grad_norm": 0.4795153737068176, "global_step": 391, "epoch": 0, "lr": 9.999999149419712e-05} +{"train_loss": 0.5029798746109009, "train_loss_bc": 0.5014903545379639, "train_loss_llm": 0.1489521861076355, "grad_norm": 0.5293166041374207, "global_step": 392, "epoch": 0, "lr": 9.999999105240987e-05} +{"train_loss": 0.39656582474708557, "train_loss_bc": 0.3948219418525696, "train_loss_llm": 0.1743890941143036, "grad_norm": 0.15729466080665588, "global_step": 393, "epoch": 0, "lr": 9.999999105240987e-05} +{"train_loss": 0.44001615047454834, "train_loss_bc": 0.4387052059173584, "train_loss_llm": 0.13109296560287476, "grad_norm": 0.31271892786026, "global_step": 394, "epoch": 0, "lr": 9.999999105240987e-05} +{"train_loss": 0.39903339743614197, "train_loss_bc": 0.39764049649238586, "train_loss_llm": 0.13929006457328796, "grad_norm": 0.36049574613571167, "global_step": 395, "epoch": 0, "lr": 9.999999105240987e-05} +{"train_loss": 0.4977453649044037, "train_loss_bc": 0.4962042570114136, "train_loss_llm": 0.15410995483398438, "grad_norm": 0.39051616191864014, "global_step": 396, "epoch": 0, "lr": 9.999999105240987e-05} +{"train_loss": 0.43982723355293274, "train_loss_bc": 0.43862128257751465, "train_loss_llm": 0.12059435248374939, "grad_norm": 0.4354154169559479, "global_step": 397, "epoch": 0, "lr": 9.999999105240987e-05} +{"train_loss": 0.5899555087089539, "train_loss_bc": 0.5885719060897827, "train_loss_llm": 0.13835924863815308, "grad_norm": 0.5845114588737488, "global_step": 398, "epoch": 0, "lr": 9.999999105240987e-05} +{"train_loss": 0.43320438265800476, "train_loss_bc": 0.43193840980529785, "train_loss_llm": 0.12659692764282227, "grad_norm": 0.6804840564727783, "global_step": 399, "epoch": 0, "lr": 9.999999105240987e-05} +{"train_loss": 0.4942322373390198, "train_loss_bc": 0.49287840723991394, "train_loss_llm": 0.13538378477096558, "grad_norm": 0.7226712107658386, "global_step": 400, "epoch": 0, "lr": 9.999999059943815e-05} +{"train_loss": 0.48597660660743713, "train_loss_bc": 0.4844684600830078, "train_loss_llm": 0.15081602334976196, "grad_norm": 0.17442671954631805, "global_step": 401, "epoch": 0, "lr": 9.999999059943815e-05} +{"train_loss": 0.5099399089813232, "train_loss_bc": 0.5084052085876465, "train_loss_llm": 0.15346896648406982, "grad_norm": 0.1842973679304123, "global_step": 402, "epoch": 0, "lr": 9.999999059943815e-05} +{"train_loss": 0.5818295478820801, "train_loss_bc": 0.5802335739135742, "train_loss_llm": 0.1595945954322815, "grad_norm": 0.32389625906944275, "global_step": 403, "epoch": 0, "lr": 9.999999059943815e-05} +{"train_loss": 0.3964095115661621, "train_loss_bc": 0.39489465951919556, "train_loss_llm": 0.15148639678955078, "grad_norm": 0.35125210881233215, "global_step": 404, "epoch": 0, "lr": 9.999999059943815e-05} +{"train_loss": 0.3978312313556671, "train_loss_bc": 0.396207332611084, "train_loss_llm": 0.16238966584205627, "grad_norm": 0.37977051734924316, "global_step": 405, "epoch": 0, "lr": 9.999999059943815e-05} +{"train_loss": 0.5158287286758423, "train_loss_bc": 0.514661431312561, "train_loss_llm": 0.1167304515838623, "grad_norm": 0.40595048666000366, "global_step": 406, "epoch": 0, "lr": 9.999999059943815e-05} +{"train_loss": 0.43381360173225403, "train_loss_bc": 0.432359516620636, "train_loss_llm": 0.14540767669677734, "grad_norm": 0.40257078409194946, "global_step": 407, "epoch": 0, "lr": 9.999999059943815e-05} +{"train_loss": 0.49121731519699097, "train_loss_bc": 0.4904581606388092, "train_loss_llm": 0.07591649889945984, "grad_norm": 0.45930102467536926, "global_step": 408, "epoch": 0, "lr": 9.999999013528191e-05} +{"train_loss": 0.4137920141220093, "train_loss_bc": 0.4127359986305237, "train_loss_llm": 0.10560128092765808, "grad_norm": 0.08735445886850357, "global_step": 409, "epoch": 0, "lr": 9.999999013528191e-05} +{"train_loss": 0.4210764765739441, "train_loss_bc": 0.42003607749938965, "train_loss_llm": 0.10404101014137268, "grad_norm": 0.12577006220817566, "global_step": 410, "epoch": 0, "lr": 9.999999013528191e-05} +{"train_loss": 0.3600696623325348, "train_loss_bc": 0.3590986728668213, "train_loss_llm": 0.09709915518760681, "grad_norm": 0.1764354705810547, "global_step": 411, "epoch": 0, "lr": 9.999999013528191e-05} +{"train_loss": 0.3732806444168091, "train_loss_bc": 0.3721208870410919, "train_loss_llm": 0.11597716808319092, "grad_norm": 0.26038920879364014, "global_step": 412, "epoch": 0, "lr": 9.999999013528191e-05} +{"train_loss": 0.3598593473434448, "train_loss_bc": 0.35865819454193115, "train_loss_llm": 0.120115727186203, "grad_norm": 0.27648207545280457, "global_step": 413, "epoch": 0, "lr": 9.999999013528191e-05} +{"train_loss": 0.3415822386741638, "train_loss_bc": 0.340486079454422, "train_loss_llm": 0.1096150279045105, "grad_norm": 0.2826792895793915, "global_step": 414, "epoch": 0, "lr": 9.999999013528191e-05} +{"train_loss": 0.40801647305488586, "train_loss_bc": 0.4063044786453247, "train_loss_llm": 0.17119815945625305, "grad_norm": 0.3926190733909607, "global_step": 415, "epoch": 0, "lr": 9.999999013528191e-05} +{"train_loss": 0.4018140435218811, "train_loss_bc": 0.400858610868454, "train_loss_llm": 0.09554430842399597, "grad_norm": 0.4989631772041321, "global_step": 416, "epoch": 0, "lr": 9.999998965994121e-05} +{"train_loss": 0.430081307888031, "train_loss_bc": 0.4291030764579773, "train_loss_llm": 0.09782415628433228, "grad_norm": 0.14672893285751343, "global_step": 417, "epoch": 0, "lr": 9.999998965994121e-05} +{"train_loss": 0.4517761170864105, "train_loss_bc": 0.45034217834472656, "train_loss_llm": 0.14339518547058105, "grad_norm": 0.19781732559204102, "global_step": 418, "epoch": 0, "lr": 9.999998965994121e-05} +{"train_loss": 0.5285919904708862, "train_loss_bc": 0.5270531177520752, "train_loss_llm": 0.1538901925086975, "grad_norm": 0.24092361330986023, "global_step": 419, "epoch": 0, "lr": 9.999998965994121e-05} +{"train_loss": 0.36633288860321045, "train_loss_bc": 0.36479756236076355, "train_loss_llm": 0.15353232622146606, "grad_norm": 0.2585471570491791, "global_step": 420, "epoch": 0, "lr": 9.999998965994121e-05} +{"train_loss": 0.44682684540748596, "train_loss_bc": 0.44523486495018005, "train_loss_llm": 0.15919902920722961, "grad_norm": 0.30681025981903076, "global_step": 421, "epoch": 0, "lr": 9.999998965994121e-05} +{"train_loss": 0.5571611523628235, "train_loss_bc": 0.5559309720993042, "train_loss_llm": 0.12301701307296753, "grad_norm": 0.2803216278553009, "global_step": 422, "epoch": 0, "lr": 9.999998965994121e-05} +{"train_loss": 0.5000087022781372, "train_loss_bc": 0.49823516607284546, "train_loss_llm": 0.17735281586647034, "grad_norm": 0.3081950545310974, "global_step": 423, "epoch": 0, "lr": 9.999998965994121e-05} +{"train_loss": 0.3700602948665619, "train_loss_bc": 0.3685716986656189, "train_loss_llm": 0.14885857701301575, "grad_norm": 0.3197224736213684, "global_step": 424, "epoch": 0, "lr": 9.999998917341602e-05} +{"train_loss": 0.40064796805381775, "train_loss_bc": 0.39938539266586304, "train_loss_llm": 0.12625622749328613, "grad_norm": 0.14771443605422974, "global_step": 425, "epoch": 0, "lr": 9.999998917341602e-05} +{"train_loss": 0.4279921352863312, "train_loss_bc": 0.426150918006897, "train_loss_llm": 0.184120774269104, "grad_norm": 0.2524369955062866, "global_step": 426, "epoch": 0, "lr": 9.999998917341602e-05} +{"train_loss": 0.4211033880710602, "train_loss_bc": 0.4193189740180969, "train_loss_llm": 0.17844107747077942, "grad_norm": 0.26317286491394043, "global_step": 427, "epoch": 0, "lr": 9.999998917341602e-05} +{"train_loss": 0.6362380981445312, "train_loss_bc": 0.6346979141235352, "train_loss_llm": 0.15401586890220642, "grad_norm": 0.17563782632350922, "global_step": 428, "epoch": 0, "lr": 9.999998917341602e-05} +{"train_loss": 0.4269055128097534, "train_loss_bc": 0.4255620241165161, "train_loss_llm": 0.13434946537017822, "grad_norm": 0.155125230550766, "global_step": 429, "epoch": 0, "lr": 9.999998917341602e-05} +{"train_loss": 0.6000247001647949, "train_loss_bc": 0.5989522933959961, "train_loss_llm": 0.10723856091499329, "grad_norm": 0.2046971321105957, "global_step": 430, "epoch": 0, "lr": 9.999998917341602e-05} +{"train_loss": 0.36251482367515564, "train_loss_bc": 0.36118003726005554, "train_loss_llm": 0.13347837328910828, "grad_norm": 0.14988385140895844, "global_step": 431, "epoch": 0, "lr": 9.999998917341602e-05} +{"train_loss": 0.39190879464149475, "train_loss_bc": 0.390873521566391, "train_loss_llm": 0.10352712869644165, "grad_norm": 0.1398547738790512, "global_step": 432, "epoch": 0, "lr": 9.999998867570633e-05} +{"train_loss": 0.3154737949371338, "train_loss_bc": 0.3138362765312195, "train_loss_llm": 0.16375190019607544, "grad_norm": 0.10454696416854858, "global_step": 433, "epoch": 0, "lr": 9.999998867570633e-05} +{"train_loss": 0.4790007472038269, "train_loss_bc": 0.4780702292919159, "train_loss_llm": 0.0930524468421936, "grad_norm": 0.19803619384765625, "global_step": 434, "epoch": 0, "lr": 9.999998867570633e-05} +{"train_loss": 0.41390570998191833, "train_loss_bc": 0.4127756953239441, "train_loss_llm": 0.11300241947174072, "grad_norm": 0.23788657784461975, "global_step": 435, "epoch": 0, "lr": 9.999998867570633e-05} +{"train_loss": 0.4542067348957062, "train_loss_bc": 0.45255035161972046, "train_loss_llm": 0.16563963890075684, "grad_norm": 0.23237694799900055, "global_step": 436, "epoch": 0, "lr": 9.999998867570633e-05} +{"train_loss": 0.40062129497528076, "train_loss_bc": 0.39910954236984253, "train_loss_llm": 0.1511746346950531, "grad_norm": 0.1750929057598114, "global_step": 437, "epoch": 0, "lr": 9.999998867570633e-05} +{"train_loss": 0.315080463886261, "train_loss_bc": 0.3137435019016266, "train_loss_llm": 0.13369756937026978, "grad_norm": 0.18327240645885468, "global_step": 438, "epoch": 0, "lr": 9.999998867570633e-05} +{"train_loss": 0.4291898012161255, "train_loss_bc": 0.4280950427055359, "train_loss_llm": 0.1094750463962555, "grad_norm": 0.38284456729888916, "global_step": 439, "epoch": 0, "lr": 9.999998867570633e-05} +{"train_loss": 0.48087987303733826, "train_loss_bc": 0.479707270860672, "train_loss_llm": 0.11726036667823792, "grad_norm": 0.47546645998954773, "global_step": 440, "epoch": 0, "lr": 9.999998816681217e-05} +{"train_loss": 0.3898439109325409, "train_loss_bc": 0.3887038826942444, "train_loss_llm": 0.1140018105506897, "grad_norm": 0.17067891359329224, "global_step": 441, "epoch": 0, "lr": 9.999998816681217e-05} +{"train_loss": 0.4663299024105072, "train_loss_bc": 0.4648596942424774, "train_loss_llm": 0.14702022075653076, "grad_norm": 0.10368398576974869, "global_step": 442, "epoch": 0, "lr": 9.999998816681217e-05} +{"train_loss": 0.4274120628833771, "train_loss_bc": 0.42585489153862, "train_loss_llm": 0.1557181477546692, "grad_norm": 0.1276116520166397, "global_step": 443, "epoch": 0, "lr": 9.999998816681217e-05} +{"train_loss": 0.4503438174724579, "train_loss_bc": 0.44907230138778687, "train_loss_llm": 0.12715256214141846, "grad_norm": 0.13852684199810028, "global_step": 444, "epoch": 0, "lr": 9.999998816681217e-05} +{"train_loss": 0.40157538652420044, "train_loss_bc": 0.39994263648986816, "train_loss_llm": 0.16327396035194397, "grad_norm": 0.11996623128652573, "global_step": 445, "epoch": 0, "lr": 9.999998816681217e-05} +{"train_loss": 0.34672412276268005, "train_loss_bc": 0.34560561180114746, "train_loss_llm": 0.11184990406036377, "grad_norm": 0.14760375022888184, "global_step": 446, "epoch": 0, "lr": 9.999998816681217e-05} +{"train_loss": 0.43688446283340454, "train_loss_bc": 0.43577367067337036, "train_loss_llm": 0.11107787489891052, "grad_norm": 0.13797178864479065, "global_step": 447, "epoch": 0, "lr": 9.999998816681217e-05} +{"train_loss": 0.44639456272125244, "train_loss_bc": 0.44505175948143005, "train_loss_llm": 0.13428038358688354, "grad_norm": 0.18448537588119507, "global_step": 448, "epoch": 0, "lr": 9.999998764673352e-05} +{"train_loss": 0.31943270564079285, "train_loss_bc": 0.3184398114681244, "train_loss_llm": 0.09928864240646362, "grad_norm": 0.1288689821958542, "global_step": 449, "epoch": 0, "lr": 9.999998764673352e-05} +{"train_loss": 0.4913393557071686, "train_loss_bc": 0.4899936318397522, "train_loss_llm": 0.13457122445106506, "grad_norm": 0.18942956626415253, "global_step": 450, "epoch": 0, "lr": 9.999998764673352e-05} +{"train_loss": 0.44432005286216736, "train_loss_bc": 0.44301894307136536, "train_loss_llm": 0.13011130690574646, "grad_norm": 0.1983322650194168, "global_step": 451, "epoch": 0, "lr": 9.999998764673352e-05} +{"train_loss": 0.4887709319591522, "train_loss_bc": 0.4871753454208374, "train_loss_llm": 0.15955889225006104, "grad_norm": 0.3274311423301697, "global_step": 452, "epoch": 0, "lr": 9.999998764673352e-05} +{"train_loss": 0.38893312215805054, "train_loss_bc": 0.3876280188560486, "train_loss_llm": 0.13050931692123413, "grad_norm": 0.37997424602508545, "global_step": 453, "epoch": 0, "lr": 9.999998764673352e-05} +{"train_loss": 0.37864962220191956, "train_loss_bc": 0.37733376026153564, "train_loss_llm": 0.1315847635269165, "grad_norm": 0.3214993476867676, "global_step": 454, "epoch": 0, "lr": 9.999998764673352e-05} +{"train_loss": 0.5456466674804688, "train_loss_bc": 0.5444359183311462, "train_loss_llm": 0.12107640504837036, "grad_norm": 0.44003385305404663, "global_step": 455, "epoch": 0, "lr": 9.999998764673352e-05} +{"train_loss": 0.4820876121520996, "train_loss_bc": 0.48043423891067505, "train_loss_llm": 0.16533619165420532, "grad_norm": 0.4430270493030548, "global_step": 456, "epoch": 0, "lr": 9.999998711547038e-05} +{"train_loss": 0.401773065328598, "train_loss_bc": 0.4003927707672119, "train_loss_llm": 0.13802924752235413, "grad_norm": 0.13466723263263702, "global_step": 457, "epoch": 0, "lr": 9.999998711547038e-05} +{"train_loss": 0.2624478340148926, "train_loss_bc": 0.26151204109191895, "train_loss_llm": 0.09357982873916626, "grad_norm": 0.12545832991600037, "global_step": 458, "epoch": 0, "lr": 9.999998711547038e-05} +{"train_loss": 0.5148327350616455, "train_loss_bc": 0.5128586292266846, "train_loss_llm": 0.19740766286849976, "grad_norm": 0.24275615811347961, "global_step": 459, "epoch": 0, "lr": 9.999998711547038e-05} +{"train_loss": 0.31938913464546204, "train_loss_bc": 0.3182697296142578, "train_loss_llm": 0.11194014549255371, "grad_norm": 0.2743895947933197, "global_step": 460, "epoch": 0, "lr": 9.999998711547038e-05} +{"train_loss": 0.33581191301345825, "train_loss_bc": 0.33450591564178467, "train_loss_llm": 0.130598783493042, "grad_norm": 0.2917158305644989, "global_step": 461, "epoch": 0, "lr": 9.999998711547038e-05} +{"train_loss": 0.3085053861141205, "train_loss_bc": 0.30709484219551086, "train_loss_llm": 0.14105501770973206, "grad_norm": 0.3809147775173187, "global_step": 462, "epoch": 0, "lr": 9.999998711547038e-05} +{"train_loss": 0.4393481910228729, "train_loss_bc": 0.438098669052124, "train_loss_llm": 0.12495085597038269, "grad_norm": 0.4167930483818054, "global_step": 463, "epoch": 0, "lr": 9.999998711547038e-05} +{"train_loss": 0.4838191866874695, "train_loss_bc": 0.4822643995285034, "train_loss_llm": 0.15547746419906616, "grad_norm": 0.44134873151779175, "global_step": 464, "epoch": 0, "lr": 9.999998657302277e-05} +{"train_loss": 0.5111101269721985, "train_loss_bc": 0.5095697045326233, "train_loss_llm": 0.15404245257377625, "grad_norm": 0.12208423763513565, "global_step": 465, "epoch": 0, "lr": 9.999998657302277e-05} +{"train_loss": 0.46335914731025696, "train_loss_bc": 0.46191537380218506, "train_loss_llm": 0.1443779468536377, "grad_norm": 0.11516986787319183, "global_step": 466, "epoch": 0, "lr": 9.999998657302277e-05} +{"train_loss": 0.4259379804134369, "train_loss_bc": 0.4243289828300476, "train_loss_llm": 0.16089865565299988, "grad_norm": 0.16940295696258545, "global_step": 467, "epoch": 0, "lr": 9.999998657302277e-05} +{"train_loss": 0.3941490948200226, "train_loss_bc": 0.392690509557724, "train_loss_llm": 0.1458589732646942, "grad_norm": 0.2359917014837265, "global_step": 468, "epoch": 0, "lr": 9.999998657302277e-05} +{"train_loss": 0.37476423382759094, "train_loss_bc": 0.3736567795276642, "train_loss_llm": 0.110745370388031, "grad_norm": 0.30503109097480774, "global_step": 469, "epoch": 0, "lr": 9.999998657302277e-05} +{"train_loss": 0.49577876925468445, "train_loss_bc": 0.49449652433395386, "train_loss_llm": 0.12822502851486206, "grad_norm": 0.38839685916900635, "global_step": 470, "epoch": 0, "lr": 9.999998657302277e-05} +{"train_loss": 0.3916528522968292, "train_loss_bc": 0.3901559114456177, "train_loss_llm": 0.14969471096992493, "grad_norm": 0.3123827576637268, "global_step": 471, "epoch": 0, "lr": 9.999998657302277e-05} +{"train_loss": 0.3310234248638153, "train_loss_bc": 0.3300684690475464, "train_loss_llm": 0.0954965353012085, "grad_norm": 0.34952715039253235, "global_step": 472, "epoch": 0, "lr": 9.999998601939066e-05} +{"train_loss": 0.5453798174858093, "train_loss_bc": 0.543769896030426, "train_loss_llm": 0.160993754863739, "grad_norm": 0.18697309494018555, "global_step": 473, "epoch": 0, "lr": 9.999998601939066e-05} +{"train_loss": 0.41136693954467773, "train_loss_bc": 0.4100807309150696, "train_loss_llm": 0.12862235307693481, "grad_norm": 0.1862008273601532, "global_step": 474, "epoch": 0, "lr": 9.999998601939066e-05} +{"train_loss": 0.3592260479927063, "train_loss_bc": 0.35791534185409546, "train_loss_llm": 0.13107100129127502, "grad_norm": 0.24944868683815002, "global_step": 475, "epoch": 0, "lr": 9.999998601939066e-05} +{"train_loss": 0.6407182216644287, "train_loss_bc": 0.6389790177345276, "train_loss_llm": 0.17392203211784363, "grad_norm": 0.3814625144004822, "global_step": 476, "epoch": 0, "lr": 9.999998601939066e-05} +{"train_loss": 0.4018906056880951, "train_loss_bc": 0.40025702118873596, "train_loss_llm": 0.16335827112197876, "grad_norm": 0.37768566608428955, "global_step": 477, "epoch": 0, "lr": 9.999998601939066e-05} +{"train_loss": 0.418509840965271, "train_loss_bc": 0.41750001907348633, "train_loss_llm": 0.10098099708557129, "grad_norm": 0.43738582730293274, "global_step": 478, "epoch": 0, "lr": 9.999998601939066e-05} +{"train_loss": 0.3062284290790558, "train_loss_bc": 0.30499154329299927, "train_loss_llm": 0.12368762493133545, "grad_norm": 0.5155507922172546, "global_step": 479, "epoch": 0, "lr": 9.999998601939066e-05} +{"train_loss": 0.31277596950531006, "train_loss_bc": 0.3117507994174957, "train_loss_llm": 0.10251766443252563, "grad_norm": 0.6082274913787842, "global_step": 480, "epoch": 0, "lr": 9.999998545457408e-05} +{"train_loss": 0.4614284336566925, "train_loss_bc": 0.4600205719470978, "train_loss_llm": 0.1407873034477234, "grad_norm": 0.15221890807151794, "global_step": 481, "epoch": 0, "lr": 9.999998545457408e-05} +{"train_loss": 0.4230389893054962, "train_loss_bc": 0.421322762966156, "train_loss_llm": 0.1716226041316986, "grad_norm": 0.15240395069122314, "global_step": 482, "epoch": 0, "lr": 9.999998545457408e-05} +{"train_loss": 0.5801178812980652, "train_loss_bc": 0.5780540704727173, "train_loss_llm": 0.20638036727905273, "grad_norm": 0.23938965797424316, "global_step": 483, "epoch": 0, "lr": 9.999998545457408e-05} +{"train_loss": 0.537693202495575, "train_loss_bc": 0.5364312529563904, "train_loss_llm": 0.12619301676750183, "grad_norm": 0.36453065276145935, "global_step": 484, "epoch": 0, "lr": 9.999998545457408e-05} +{"train_loss": 0.34497976303100586, "train_loss_bc": 0.3441852927207947, "train_loss_llm": 0.07944628596305847, "grad_norm": 0.3134423494338989, "global_step": 485, "epoch": 0, "lr": 9.999998545457408e-05} +{"train_loss": 0.35838940739631653, "train_loss_bc": 0.35705840587615967, "train_loss_llm": 0.13309887051582336, "grad_norm": 0.35410037636756897, "global_step": 486, "epoch": 0, "lr": 9.999998545457408e-05} +{"train_loss": 0.47313427925109863, "train_loss_bc": 0.47136640548706055, "train_loss_llm": 0.17678722739219666, "grad_norm": 0.48954063653945923, "global_step": 487, "epoch": 0, "lr": 9.999998545457408e-05} +{"train_loss": 0.48256510496139526, "train_loss_bc": 0.4812368154525757, "train_loss_llm": 0.132828950881958, "grad_norm": 0.5150009989738464, "global_step": 488, "epoch": 0, "lr": 9.9999984878573e-05} +{"train_loss": 0.4850676655769348, "train_loss_bc": 0.4836665689945221, "train_loss_llm": 0.14010915160179138, "grad_norm": 0.11707025021314621, "global_step": 489, "epoch": 0, "lr": 9.9999984878573e-05} +{"train_loss": 0.4168444275856018, "train_loss_bc": 0.41549062728881836, "train_loss_llm": 0.13538148999214172, "grad_norm": 0.20563825964927673, "global_step": 490, "epoch": 0, "lr": 9.9999984878573e-05} +{"train_loss": 0.4963814616203308, "train_loss_bc": 0.4949724078178406, "train_loss_llm": 0.1409062147140503, "grad_norm": 0.2898010015487671, "global_step": 491, "epoch": 0, "lr": 9.9999984878573e-05} +{"train_loss": 0.31517115235328674, "train_loss_bc": 0.3138827681541443, "train_loss_llm": 0.1288374662399292, "grad_norm": 0.35648441314697266, "global_step": 492, "epoch": 0, "lr": 9.9999984878573e-05} +{"train_loss": 0.4574425220489502, "train_loss_bc": 0.4562724828720093, "train_loss_llm": 0.11700320243835449, "grad_norm": 0.3613799512386322, "global_step": 493, "epoch": 0, "lr": 9.9999984878573e-05} +{"train_loss": 0.472825288772583, "train_loss_bc": 0.4712083637714386, "train_loss_llm": 0.16169312596321106, "grad_norm": 0.44376441836357117, "global_step": 494, "epoch": 0, "lr": 9.9999984878573e-05} +{"train_loss": 0.5169175267219543, "train_loss_bc": 0.5153428316116333, "train_loss_llm": 0.1574695110321045, "grad_norm": 0.47124773263931274, "global_step": 495, "epoch": 0, "lr": 9.9999984878573e-05} +{"train_loss": 0.4964981973171234, "train_loss_bc": 0.4946369230747223, "train_loss_llm": 0.18612626194953918, "grad_norm": 0.45836684107780457, "global_step": 496, "epoch": 0, "lr": 9.999998429138744e-05} +{"train_loss": 0.3909382224082947, "train_loss_bc": 0.38937729597091675, "train_loss_llm": 0.15609151124954224, "grad_norm": 0.1459275335073471, "global_step": 497, "epoch": 0, "lr": 9.999998429138744e-05} +{"train_loss": 0.4286676049232483, "train_loss_bc": 0.4274001121520996, "train_loss_llm": 0.12674924731254578, "grad_norm": 0.1715053915977478, "global_step": 498, "epoch": 0, "lr": 9.999998429138744e-05} +{"train_loss": 0.39432257413864136, "train_loss_bc": 0.39305007457733154, "train_loss_llm": 0.12724903225898743, "grad_norm": 0.17844915390014648, "global_step": 499, "epoch": 0, "lr": 9.999998429138744e-05} +{"train_loss": 0.4661673903465271, "train_loss_bc": 0.4649445414543152, "train_loss_llm": 0.1222846508026123, "grad_norm": 0.17432816326618195, "global_step": 500, "epoch": 0, "lr": 9.999998429138744e-05} +{"train_loss": 0.34359753131866455, "train_loss_bc": 0.3425159454345703, "train_loss_llm": 0.10815823078155518, "grad_norm": 0.19458213448524475, "global_step": 501, "epoch": 0, "lr": 9.999998429138744e-05} +{"train_loss": 0.40056681632995605, "train_loss_bc": 0.39885884523391724, "train_loss_llm": 0.17079809308052063, "grad_norm": 0.20834076404571533, "global_step": 502, "epoch": 0, "lr": 9.999998429138744e-05} +{"train_loss": 0.5176026821136475, "train_loss_bc": 0.5161987543106079, "train_loss_llm": 0.14039567112922668, "grad_norm": 0.26620596647262573, "global_step": 503, "epoch": 0, "lr": 9.999998429138744e-05} +{"train_loss": 0.4068608582019806, "train_loss_bc": 0.4058382511138916, "train_loss_llm": 0.10225984454154968, "grad_norm": 0.29349592328071594, "global_step": 504, "epoch": 0, "lr": 9.99999836930174e-05} +{"train_loss": 0.35090312361717224, "train_loss_bc": 0.3500174283981323, "train_loss_llm": 0.08857083320617676, "grad_norm": 0.11117282509803772, "global_step": 505, "epoch": 0, "lr": 9.99999836930174e-05} +{"train_loss": 0.4139205813407898, "train_loss_bc": 0.41290992498397827, "train_loss_llm": 0.10106602311134338, "grad_norm": 0.16088035702705383, "global_step": 506, "epoch": 0, "lr": 9.99999836930174e-05} +{"train_loss": 0.466202974319458, "train_loss_bc": 0.46463456749916077, "train_loss_llm": 0.1568402647972107, "grad_norm": 0.2754596173763275, "global_step": 507, "epoch": 0, "lr": 9.99999836930174e-05} +{"train_loss": 0.4747956097126007, "train_loss_bc": 0.4734051823616028, "train_loss_llm": 0.13904142379760742, "grad_norm": 0.32233870029449463, "global_step": 508, "epoch": 0, "lr": 9.99999836930174e-05} +{"train_loss": 0.3717607259750366, "train_loss_bc": 0.3702400028705597, "train_loss_llm": 0.15207374095916748, "grad_norm": 0.3244297504425049, "global_step": 509, "epoch": 0, "lr": 9.99999836930174e-05} +{"train_loss": 0.3526959717273712, "train_loss_bc": 0.35189297795295715, "train_loss_llm": 0.08029988408088684, "grad_norm": 0.33454468846321106, "global_step": 510, "epoch": 0, "lr": 9.99999836930174e-05} +{"train_loss": 0.3579065799713135, "train_loss_bc": 0.35683122277259827, "train_loss_llm": 0.10753712058067322, "grad_norm": 0.33429887890815735, "global_step": 511, "epoch": 0, "lr": 9.99999836930174e-05} +{"train_loss": 0.45556026697158813, "train_loss_bc": 0.4541212022304535, "train_loss_llm": 0.14390522241592407, "grad_norm": 0.4624151885509491, "global_step": 512, "epoch": 0, "lr": 9.999998308346286e-05} +{"train_loss": 0.5283898711204529, "train_loss_bc": 0.5268588662147522, "train_loss_llm": 0.1531004011631012, "grad_norm": 0.09483818709850311, "global_step": 513, "epoch": 0, "lr": 9.999998308346286e-05} +{"train_loss": 0.6458296775817871, "train_loss_bc": 0.6440832614898682, "train_loss_llm": 0.17464014887809753, "grad_norm": 0.19383709132671356, "global_step": 514, "epoch": 0, "lr": 9.999998308346286e-05} +{"train_loss": 0.5531331300735474, "train_loss_bc": 0.5513787865638733, "train_loss_llm": 0.17543554306030273, "grad_norm": 0.20223230123519897, "global_step": 515, "epoch": 0, "lr": 9.999998308346286e-05} +{"train_loss": 0.5076155662536621, "train_loss_bc": 0.5062960982322693, "train_loss_llm": 0.13194802403450012, "grad_norm": 0.17811913788318634, "global_step": 516, "epoch": 0, "lr": 9.999998308346286e-05} +{"train_loss": 0.3484884798526764, "train_loss_bc": 0.34735366702079773, "train_loss_llm": 0.11348021030426025, "grad_norm": 0.16792768239974976, "global_step": 517, "epoch": 0, "lr": 9.999998308346286e-05} +{"train_loss": 0.395936518907547, "train_loss_bc": 0.3950364291667938, "train_loss_llm": 0.09000906348228455, "grad_norm": 0.2023770958185196, "global_step": 518, "epoch": 0, "lr": 9.999998308346286e-05} +{"train_loss": 0.45047861337661743, "train_loss_bc": 0.449161171913147, "train_loss_llm": 0.1317441463470459, "grad_norm": 0.24776966869831085, "global_step": 519, "epoch": 0, "lr": 9.999998308346286e-05} +{"train_loss": 0.5846872329711914, "train_loss_bc": 0.5833771228790283, "train_loss_llm": 0.13101065158843994, "grad_norm": 0.2631140947341919, "global_step": 520, "epoch": 0, "lr": 9.999998246272385e-05} +{"train_loss": 0.35722488164901733, "train_loss_bc": 0.35609161853790283, "train_loss_llm": 0.11332684755325317, "grad_norm": 0.1942409723997116, "global_step": 521, "epoch": 0, "lr": 9.999998246272385e-05} +{"train_loss": 0.45665645599365234, "train_loss_bc": 0.4553197920322418, "train_loss_llm": 0.1336669921875, "grad_norm": 0.12631982564926147, "global_step": 522, "epoch": 0, "lr": 9.999998246272385e-05} +{"train_loss": 0.33727920055389404, "train_loss_bc": 0.33604696393013, "train_loss_llm": 0.12322363257408142, "grad_norm": 0.14392586052417755, "global_step": 523, "epoch": 0, "lr": 9.999998246272385e-05} +{"train_loss": 0.47541773319244385, "train_loss_bc": 0.4742015600204468, "train_loss_llm": 0.12161818146705627, "grad_norm": 0.2969023883342743, "global_step": 524, "epoch": 0, "lr": 9.999998246272385e-05} +{"train_loss": 0.3839433789253235, "train_loss_bc": 0.38317179679870605, "train_loss_llm": 0.07715818285942078, "grad_norm": 0.31521326303482056, "global_step": 525, "epoch": 0, "lr": 9.999998246272385e-05} +{"train_loss": 0.2998273968696594, "train_loss_bc": 0.2985779643058777, "train_loss_llm": 0.12494179606437683, "grad_norm": 0.32320526242256165, "global_step": 526, "epoch": 0, "lr": 9.999998246272385e-05} +{"train_loss": 0.46597835421562195, "train_loss_bc": 0.4649353623390198, "train_loss_llm": 0.10429981350898743, "grad_norm": 0.33954906463623047, "global_step": 527, "epoch": 0, "lr": 9.999998246272385e-05} +{"train_loss": 0.4422476887702942, "train_loss_bc": 0.44077128171920776, "train_loss_llm": 0.14763975143432617, "grad_norm": 0.38341936469078064, "global_step": 528, "epoch": 0, "lr": 9.999998183080035e-05} +{"train_loss": 0.4010619521141052, "train_loss_bc": 0.39970681071281433, "train_loss_llm": 0.13551461696624756, "grad_norm": 0.15890274941921234, "global_step": 529, "epoch": 0, "lr": 9.999998183080035e-05} +{"train_loss": 0.43727365136146545, "train_loss_bc": 0.43607696890830994, "train_loss_llm": 0.1196679174900055, "grad_norm": 0.1591171771287918, "global_step": 530, "epoch": 0, "lr": 9.999998183080035e-05} +{"train_loss": 0.40885961055755615, "train_loss_bc": 0.4078683853149414, "train_loss_llm": 0.0991223156452179, "grad_norm": 0.10598937422037125, "global_step": 531, "epoch": 0, "lr": 9.999998183080035e-05} +{"train_loss": 0.42146050930023193, "train_loss_bc": 0.4201066493988037, "train_loss_llm": 0.13538533449172974, "grad_norm": 0.1596555858850479, "global_step": 532, "epoch": 0, "lr": 9.999998183080035e-05} +{"train_loss": 0.47013160586357117, "train_loss_bc": 0.46826934814453125, "train_loss_llm": 0.18622556328773499, "grad_norm": 0.3127519190311432, "global_step": 533, "epoch": 0, "lr": 9.999998183080035e-05} +{"train_loss": 0.43455177545547485, "train_loss_bc": 0.43307244777679443, "train_loss_llm": 0.147932231426239, "grad_norm": 0.4107877016067505, "global_step": 534, "epoch": 0, "lr": 9.999998183080035e-05} +{"train_loss": 0.5151340365409851, "train_loss_bc": 0.5133260488510132, "train_loss_llm": 0.18080013990402222, "grad_norm": 0.45147961378097534, "global_step": 535, "epoch": 0, "lr": 9.999998183080035e-05} +{"train_loss": 0.468685507774353, "train_loss_bc": 0.46713870763778687, "train_loss_llm": 0.15468057990074158, "grad_norm": 0.46718934178352356, "global_step": 536, "epoch": 0, "lr": 9.999998118769238e-05} +{"train_loss": 0.33666184544563293, "train_loss_bc": 0.33582043647766113, "train_loss_llm": 0.08413982391357422, "grad_norm": 0.07298891991376877, "global_step": 537, "epoch": 0, "lr": 9.999998118769238e-05} +{"train_loss": 0.3427335321903229, "train_loss_bc": 0.34128645062446594, "train_loss_llm": 0.14470744132995605, "grad_norm": 0.10871179401874542, "global_step": 538, "epoch": 0, "lr": 9.999998118769238e-05} +{"train_loss": 0.5018195509910583, "train_loss_bc": 0.5006521940231323, "train_loss_llm": 0.11673450469970703, "grad_norm": 0.16300837695598602, "global_step": 539, "epoch": 0, "lr": 9.999998118769238e-05} +{"train_loss": 0.5012814998626709, "train_loss_bc": 0.4999098777770996, "train_loss_llm": 0.13716137409210205, "grad_norm": 0.21575038135051727, "global_step": 540, "epoch": 0, "lr": 9.999998118769238e-05} +{"train_loss": 0.489978551864624, "train_loss_bc": 0.48860251903533936, "train_loss_llm": 0.1376034915447235, "grad_norm": 0.2113872766494751, "global_step": 541, "epoch": 0, "lr": 9.999998118769238e-05} +{"train_loss": 0.3948250710964203, "train_loss_bc": 0.3938084840774536, "train_loss_llm": 0.10165795683860779, "grad_norm": 0.3279014825820923, "global_step": 542, "epoch": 0, "lr": 9.999998118769238e-05} +{"train_loss": 0.5285438895225525, "train_loss_bc": 0.5268083214759827, "train_loss_llm": 0.17355450987815857, "grad_norm": 0.4272497594356537, "global_step": 543, "epoch": 0, "lr": 9.999998118769238e-05} +{"train_loss": 0.48978671431541443, "train_loss_bc": 0.48840296268463135, "train_loss_llm": 0.13837400078773499, "grad_norm": 0.5272319912910461, "global_step": 544, "epoch": 0, "lr": 9.999998053339992e-05} +{"train_loss": 0.35207897424697876, "train_loss_bc": 0.3506183326244354, "train_loss_llm": 0.1460632085800171, "grad_norm": 0.1743731051683426, "global_step": 545, "epoch": 0, "lr": 9.999998053339992e-05} +{"train_loss": 0.5272657871246338, "train_loss_bc": 0.5256842970848083, "train_loss_llm": 0.15814724564552307, "grad_norm": 0.295947402715683, "global_step": 546, "epoch": 0, "lr": 9.999998053339992e-05} +{"train_loss": 0.4524272382259369, "train_loss_bc": 0.45136499404907227, "train_loss_llm": 0.10622522234916687, "grad_norm": 0.34490787982940674, "global_step": 547, "epoch": 0, "lr": 9.999998053339992e-05} +{"train_loss": 0.47027266025543213, "train_loss_bc": 0.4689750373363495, "train_loss_llm": 0.1297619640827179, "grad_norm": 0.2702181935310364, "global_step": 548, "epoch": 0, "lr": 9.999998053339992e-05} +{"train_loss": 0.5305567383766174, "train_loss_bc": 0.5286378264427185, "train_loss_llm": 0.19189050793647766, "grad_norm": 0.30008596181869507, "global_step": 549, "epoch": 0, "lr": 9.999998053339992e-05} +{"train_loss": 0.44006362557411194, "train_loss_bc": 0.4386911392211914, "train_loss_llm": 0.13724949955940247, "grad_norm": 0.36235639452934265, "global_step": 550, "epoch": 0, "lr": 9.999998053339992e-05} +{"train_loss": 0.5773160457611084, "train_loss_bc": 0.57586669921875, "train_loss_llm": 0.14493277668952942, "grad_norm": 0.37752899527549744, "global_step": 551, "epoch": 0, "lr": 9.999998053339992e-05} +{"train_loss": 0.4000251591205597, "train_loss_bc": 0.39876028895378113, "train_loss_llm": 0.12648743391036987, "grad_norm": 0.3652367889881134, "global_step": 552, "epoch": 0, "lr": 9.999997986792297e-05} +{"train_loss": 0.4534410834312439, "train_loss_bc": 0.45195817947387695, "train_loss_llm": 0.1482906937599182, "grad_norm": 0.10996562987565994, "global_step": 553, "epoch": 0, "lr": 9.999997986792297e-05} +{"train_loss": 0.5020356178283691, "train_loss_bc": 0.5003174543380737, "train_loss_llm": 0.17181852459907532, "grad_norm": 0.16083915531635284, "global_step": 554, "epoch": 0, "lr": 9.999997986792297e-05} +{"train_loss": 0.40376171469688416, "train_loss_bc": 0.4027332067489624, "train_loss_llm": 0.1028505265712738, "grad_norm": 0.1733359545469284, "global_step": 555, "epoch": 0, "lr": 9.999997986792297e-05} +{"train_loss": 0.30626556277275085, "train_loss_bc": 0.3052656054496765, "train_loss_llm": 0.09999540448188782, "grad_norm": 0.2684016227722168, "global_step": 556, "epoch": 0, "lr": 9.999997986792297e-05} +{"train_loss": 0.4023430347442627, "train_loss_bc": 0.40086281299591064, "train_loss_llm": 0.14802241325378418, "grad_norm": 0.3983711302280426, "global_step": 557, "epoch": 0, "lr": 9.999997986792297e-05} +{"train_loss": 0.37437474727630615, "train_loss_bc": 0.37329378724098206, "train_loss_llm": 0.108096182346344, "grad_norm": 0.36518532037734985, "global_step": 558, "epoch": 0, "lr": 9.999997986792297e-05} +{"train_loss": 0.43426066637039185, "train_loss_bc": 0.4328705072402954, "train_loss_llm": 0.13901570439338684, "grad_norm": 0.3366381525993347, "global_step": 559, "epoch": 0, "lr": 9.999997986792297e-05} +{"train_loss": 0.5288271307945251, "train_loss_bc": 0.5273444056510925, "train_loss_llm": 0.1482728123664856, "grad_norm": 0.4081956446170807, "global_step": 560, "epoch": 0, "lr": 9.999997919126154e-05} +{"train_loss": 0.3179990351200104, "train_loss_bc": 0.317210853099823, "train_loss_llm": 0.07881835103034973, "grad_norm": 0.11407114565372467, "global_step": 561, "epoch": 0, "lr": 9.999997919126154e-05} +{"train_loss": 0.4405460059642792, "train_loss_bc": 0.4392358064651489, "train_loss_llm": 0.13102111220359802, "grad_norm": 0.15511688590049744, "global_step": 562, "epoch": 0, "lr": 9.999997919126154e-05} +{"train_loss": 0.354533314704895, "train_loss_bc": 0.3535911440849304, "train_loss_llm": 0.094217449426651, "grad_norm": 0.21661803126335144, "global_step": 563, "epoch": 0, "lr": 9.999997919126154e-05} +{"train_loss": 0.39877161383628845, "train_loss_bc": 0.3976114094257355, "train_loss_llm": 0.1160200834274292, "grad_norm": 0.24238751828670502, "global_step": 564, "epoch": 0, "lr": 9.999997919126154e-05} +{"train_loss": 0.3421781361103058, "train_loss_bc": 0.34106457233428955, "train_loss_llm": 0.11135563254356384, "grad_norm": 0.2127084732055664, "global_step": 565, "epoch": 0, "lr": 9.999997919126154e-05} +{"train_loss": 0.49832308292388916, "train_loss_bc": 0.49685126543045044, "train_loss_llm": 0.14718025922775269, "grad_norm": 0.18412984907627106, "global_step": 566, "epoch": 0, "lr": 9.999997919126154e-05} +{"train_loss": 0.3954983353614807, "train_loss_bc": 0.39409857988357544, "train_loss_llm": 0.1399749219417572, "grad_norm": 0.2935402989387512, "global_step": 567, "epoch": 0, "lr": 9.999997919126154e-05} +{"train_loss": 0.3805713653564453, "train_loss_bc": 0.3792915344238281, "train_loss_llm": 0.12798207998275757, "grad_norm": 0.40969666838645935, "global_step": 568, "epoch": 0, "lr": 9.999997850341563e-05} +{"train_loss": 0.315425306558609, "train_loss_bc": 0.3146827816963196, "train_loss_llm": 0.07425335049629211, "grad_norm": 0.10903266072273254, "global_step": 569, "epoch": 0, "lr": 9.999997850341563e-05} +{"train_loss": 0.4334687292575836, "train_loss_bc": 0.432287335395813, "train_loss_llm": 0.11813810467720032, "grad_norm": 0.1422020047903061, "global_step": 570, "epoch": 0, "lr": 9.999997850341563e-05} +{"train_loss": 0.50501549243927, "train_loss_bc": 0.5034652948379517, "train_loss_llm": 0.15502256155014038, "grad_norm": 0.19771867990493774, "global_step": 571, "epoch": 0, "lr": 9.999997850341563e-05} +{"train_loss": 0.4925400912761688, "train_loss_bc": 0.4910658597946167, "train_loss_llm": 0.14742416143417358, "grad_norm": 0.27655327320098877, "global_step": 572, "epoch": 0, "lr": 9.999997850341563e-05} +{"train_loss": 0.3185931146144867, "train_loss_bc": 0.3175286650657654, "train_loss_llm": 0.10644495487213135, "grad_norm": 0.25428882241249084, "global_step": 573, "epoch": 0, "lr": 9.999997850341563e-05} +{"train_loss": 0.3887785077095032, "train_loss_bc": 0.38751542568206787, "train_loss_llm": 0.12630781531333923, "grad_norm": 0.27936604619026184, "global_step": 574, "epoch": 0, "lr": 9.999997850341563e-05} +{"train_loss": 0.3779152035713196, "train_loss_bc": 0.37660133838653564, "train_loss_llm": 0.13138729333877563, "grad_norm": 0.47680532932281494, "global_step": 575, "epoch": 0, "lr": 9.999997850341563e-05} +{"train_loss": 0.4860542118549347, "train_loss_bc": 0.48456716537475586, "train_loss_llm": 0.1487049162387848, "grad_norm": 0.6616555452346802, "global_step": 576, "epoch": 0, "lr": 9.999997780438522e-05} +{"train_loss": 0.49950721859931946, "train_loss_bc": 0.4983985424041748, "train_loss_llm": 0.11086836457252502, "grad_norm": 0.18198266625404358, "global_step": 577, "epoch": 0, "lr": 9.999997780438522e-05} +{"train_loss": 0.4686140716075897, "train_loss_bc": 0.46717584133148193, "train_loss_llm": 0.14382407069206238, "grad_norm": 0.3312739133834839, "global_step": 578, "epoch": 0, "lr": 9.999997780438522e-05} +{"train_loss": 0.5470864772796631, "train_loss_bc": 0.5458697080612183, "train_loss_llm": 0.1216752827167511, "grad_norm": 0.47114571928977966, "global_step": 579, "epoch": 0, "lr": 9.999997780438522e-05} +{"train_loss": 0.39431431889533997, "train_loss_bc": 0.3931993842124939, "train_loss_llm": 0.11149218678474426, "grad_norm": 0.45695143938064575, "global_step": 580, "epoch": 0, "lr": 9.999997780438522e-05} +{"train_loss": 0.3757352828979492, "train_loss_bc": 0.3745602071285248, "train_loss_llm": 0.11750820279121399, "grad_norm": 0.508247435092926, "global_step": 581, "epoch": 0, "lr": 9.999997780438522e-05} +{"train_loss": 0.4377124607563019, "train_loss_bc": 0.43647441267967224, "train_loss_llm": 0.12380599975585938, "grad_norm": 0.49301835894584656, "global_step": 582, "epoch": 0, "lr": 9.999997780438522e-05} +{"train_loss": 0.45535722374916077, "train_loss_bc": 0.4538666009902954, "train_loss_llm": 0.14906099438667297, "grad_norm": 0.5854777693748474, "global_step": 583, "epoch": 0, "lr": 9.999997780438522e-05} +{"train_loss": 0.3647003769874573, "train_loss_bc": 0.36371609568595886, "train_loss_llm": 0.09842696785926819, "grad_norm": 0.5401211977005005, "global_step": 584, "epoch": 0, "lr": 9.999997709417033e-05} +{"train_loss": 0.33450859785079956, "train_loss_bc": 0.3331279158592224, "train_loss_llm": 0.1380678415298462, "grad_norm": 0.13530641794204712, "global_step": 585, "epoch": 0, "lr": 9.999997709417033e-05} +{"train_loss": 0.3917991816997528, "train_loss_bc": 0.39066940546035767, "train_loss_llm": 0.11297649145126343, "grad_norm": 0.09318878501653671, "global_step": 586, "epoch": 0, "lr": 9.999997709417033e-05} +{"train_loss": 0.42085719108581543, "train_loss_bc": 0.41950592398643494, "train_loss_llm": 0.13512656092643738, "grad_norm": 0.12827591598033905, "global_step": 587, "epoch": 0, "lr": 9.999997709417033e-05} +{"train_loss": 0.3857162594795227, "train_loss_bc": 0.38425734639167786, "train_loss_llm": 0.14589032530784607, "grad_norm": 0.13327433168888092, "global_step": 588, "epoch": 0, "lr": 9.999997709417033e-05} +{"train_loss": 0.42687976360321045, "train_loss_bc": 0.4254385530948639, "train_loss_llm": 0.1441207230091095, "grad_norm": 0.15169674158096313, "global_step": 589, "epoch": 0, "lr": 9.999997709417033e-05} +{"train_loss": 0.509350597858429, "train_loss_bc": 0.5077565312385559, "train_loss_llm": 0.1594066619873047, "grad_norm": 0.3126141130924225, "global_step": 590, "epoch": 0, "lr": 9.999997709417033e-05} +{"train_loss": 0.2661266326904297, "train_loss_bc": 0.265122652053833, "train_loss_llm": 0.10039719939231873, "grad_norm": 0.32771924138069153, "global_step": 591, "epoch": 0, "lr": 9.999997709417033e-05} +{"train_loss": 0.4261973798274994, "train_loss_bc": 0.4247882068157196, "train_loss_llm": 0.14091673493385315, "grad_norm": 0.357107549905777, "global_step": 592, "epoch": 0, "lr": 9.999997637277098e-05} +{"train_loss": 0.4417702853679657, "train_loss_bc": 0.4405673146247864, "train_loss_llm": 0.12029564380645752, "grad_norm": 0.21018153429031372, "global_step": 593, "epoch": 0, "lr": 9.999997637277098e-05} +{"train_loss": 0.4224977493286133, "train_loss_bc": 0.42130085825920105, "train_loss_llm": 0.1196904182434082, "grad_norm": 0.17576035857200623, "global_step": 594, "epoch": 0, "lr": 9.999997637277098e-05} +{"train_loss": 0.24003392457962036, "train_loss_bc": 0.23908013105392456, "train_loss_llm": 0.09537988901138306, "grad_norm": 0.14014993607997894, "global_step": 595, "epoch": 0, "lr": 9.999997637277098e-05} +{"train_loss": 0.3188736140727997, "train_loss_bc": 0.3178519010543823, "train_loss_llm": 0.10217243432998657, "grad_norm": 0.2083711177110672, "global_step": 596, "epoch": 0, "lr": 9.999997637277098e-05} +{"train_loss": 0.32962101697921753, "train_loss_bc": 0.32892146706581116, "train_loss_llm": 0.06995385885238647, "grad_norm": 0.2619014382362366, "global_step": 597, "epoch": 0, "lr": 9.999997637277098e-05} +{"train_loss": 0.32570961117744446, "train_loss_bc": 0.3247925937175751, "train_loss_llm": 0.09170261025428772, "grad_norm": 0.3052881956100464, "global_step": 598, "epoch": 0, "lr": 9.999997637277098e-05} +{"train_loss": 0.34404316544532776, "train_loss_bc": 0.34271562099456787, "train_loss_llm": 0.1327531933784485, "grad_norm": 0.3279159665107727, "global_step": 599, "epoch": 0, "lr": 9.999997637277098e-05} +{"train_loss": 0.4341554641723633, "train_loss_bc": 0.43255412578582764, "train_loss_llm": 0.16013360023498535, "grad_norm": 0.4468793272972107, "global_step": 600, "epoch": 0, "lr": 9.999997564018713e-05} +{"train_loss": 0.5450310111045837, "train_loss_bc": 0.5434368848800659, "train_loss_llm": 0.15941256284713745, "grad_norm": 0.12080206722021103, "global_step": 601, "epoch": 0, "lr": 9.999997564018713e-05} +{"train_loss": 0.4429208040237427, "train_loss_bc": 0.44145193696022034, "train_loss_llm": 0.14688768982887268, "grad_norm": 0.19700393080711365, "global_step": 602, "epoch": 0, "lr": 9.999997564018713e-05} +{"train_loss": 0.2259754240512848, "train_loss_bc": 0.22511287033557892, "train_loss_llm": 0.08625537157058716, "grad_norm": 0.2877824902534485, "global_step": 603, "epoch": 0, "lr": 9.999997564018713e-05} +{"train_loss": 0.4095430374145508, "train_loss_bc": 0.40821927785873413, "train_loss_llm": 0.13237535953521729, "grad_norm": 0.4067089259624481, "global_step": 604, "epoch": 0, "lr": 9.999997564018713e-05} +{"train_loss": 0.4214470386505127, "train_loss_bc": 0.4202752113342285, "train_loss_llm": 0.11718130111694336, "grad_norm": 0.5375756025314331, "global_step": 605, "epoch": 0, "lr": 9.999997564018713e-05} +{"train_loss": 0.2929988205432892, "train_loss_bc": 0.29193437099456787, "train_loss_llm": 0.1064460277557373, "grad_norm": 0.5106360912322998, "global_step": 606, "epoch": 0, "lr": 9.999997564018713e-05} +{"train_loss": 0.452322393655777, "train_loss_bc": 0.450954794883728, "train_loss_llm": 0.13675937056541443, "grad_norm": 0.612158477306366, "global_step": 607, "epoch": 0, "lr": 9.999997564018713e-05} +{"train_loss": 0.5061517953872681, "train_loss_bc": 0.5048820376396179, "train_loss_llm": 0.12697654962539673, "grad_norm": 0.7311630845069885, "global_step": 608, "epoch": 0, "lr": 9.99999748964188e-05} +{"train_loss": 0.4055657982826233, "train_loss_bc": 0.4043877124786377, "train_loss_llm": 0.1178099513053894, "grad_norm": 0.10408678650856018, "global_step": 609, "epoch": 0, "lr": 9.99999748964188e-05} +{"train_loss": 0.34568390250205994, "train_loss_bc": 0.34470319747924805, "train_loss_llm": 0.09806981682777405, "grad_norm": 0.0900803878903389, "global_step": 610, "epoch": 0, "lr": 9.99999748964188e-05} +{"train_loss": 0.23388712108135223, "train_loss_bc": 0.23302355408668518, "train_loss_llm": 0.08635708689689636, "grad_norm": 0.1710396260023117, "global_step": 611, "epoch": 0, "lr": 9.99999748964188e-05} +{"train_loss": 0.5390656590461731, "train_loss_bc": 0.5379109978675842, "train_loss_llm": 0.11546611785888672, "grad_norm": 0.2550126612186432, "global_step": 612, "epoch": 0, "lr": 9.99999748964188e-05} +{"train_loss": 0.3056194484233856, "train_loss_bc": 0.3048853576183319, "train_loss_llm": 0.07340827584266663, "grad_norm": 0.301627516746521, "global_step": 613, "epoch": 0, "lr": 9.99999748964188e-05} +{"train_loss": 0.42089927196502686, "train_loss_bc": 0.4192892909049988, "train_loss_llm": 0.16099870204925537, "grad_norm": 0.24902242422103882, "global_step": 614, "epoch": 0, "lr": 9.99999748964188e-05} +{"train_loss": 0.49767935276031494, "train_loss_bc": 0.4964262843132019, "train_loss_llm": 0.1253071427345276, "grad_norm": 0.25635117292404175, "global_step": 615, "epoch": 0, "lr": 9.99999748964188e-05} +{"train_loss": 0.3980560302734375, "train_loss_bc": 0.3969253599643707, "train_loss_llm": 0.11306843161582947, "grad_norm": 0.20105355978012085, "global_step": 616, "epoch": 0, "lr": 9.9999974141466e-05} +{"train_loss": 0.5935341119766235, "train_loss_bc": 0.5922272801399231, "train_loss_llm": 0.13068339228630066, "grad_norm": 0.13810065388679504, "global_step": 617, "epoch": 0, "lr": 9.9999974141466e-05} +{"train_loss": 0.4552997946739197, "train_loss_bc": 0.4537251889705658, "train_loss_llm": 0.15746089816093445, "grad_norm": 0.12634815275669098, "global_step": 618, "epoch": 0, "lr": 9.9999974141466e-05} +{"train_loss": 0.4198523759841919, "train_loss_bc": 0.4183838963508606, "train_loss_llm": 0.14684754610061646, "grad_norm": 0.20647352933883667, "global_step": 619, "epoch": 0, "lr": 9.9999974141466e-05} +{"train_loss": 0.3329331576824188, "train_loss_bc": 0.3317468762397766, "train_loss_llm": 0.11862751841545105, "grad_norm": 0.19605395197868347, "global_step": 620, "epoch": 0, "lr": 9.9999974141466e-05} +{"train_loss": 0.5986315608024597, "train_loss_bc": 0.5967831611633301, "train_loss_llm": 0.1848422884941101, "grad_norm": 0.3618508279323578, "global_step": 621, "epoch": 0, "lr": 9.9999974141466e-05} +{"train_loss": 0.3867855668067932, "train_loss_bc": 0.38546088337898254, "train_loss_llm": 0.13246724009513855, "grad_norm": 0.3084125518798828, "global_step": 622, "epoch": 0, "lr": 9.9999974141466e-05} +{"train_loss": 0.38991618156433105, "train_loss_bc": 0.3884229362010956, "train_loss_llm": 0.14932426810264587, "grad_norm": 0.3683825731277466, "global_step": 623, "epoch": 0, "lr": 9.9999974141466e-05} +{"train_loss": 0.4563721716403961, "train_loss_bc": 0.4547320604324341, "train_loss_llm": 0.16401097178459167, "grad_norm": 0.5868925452232361, "global_step": 624, "epoch": 0, "lr": 9.999997337532869e-05} +{"train_loss": 0.5038174986839294, "train_loss_bc": 0.5024147629737854, "train_loss_llm": 0.14027628302574158, "grad_norm": 0.15974006056785583, "global_step": 625, "epoch": 0, "lr": 9.999997337532869e-05} +{"train_loss": 0.4912840723991394, "train_loss_bc": 0.4896485209465027, "train_loss_llm": 0.16355496644973755, "grad_norm": 0.15308967232704163, "global_step": 626, "epoch": 0, "lr": 9.999997337532869e-05} +{"train_loss": 0.43595683574676514, "train_loss_bc": 0.4345545768737793, "train_loss_llm": 0.14022603631019592, "grad_norm": 0.22710762917995453, "global_step": 627, "epoch": 0, "lr": 9.999997337532869e-05} +{"train_loss": 0.3652057349681854, "train_loss_bc": 0.3637992739677429, "train_loss_llm": 0.14064624905586243, "grad_norm": 0.1738707721233368, "global_step": 628, "epoch": 0, "lr": 9.999997337532869e-05} +{"train_loss": 0.508272647857666, "train_loss_bc": 0.5067090392112732, "train_loss_llm": 0.15636023879051208, "grad_norm": 0.2183106243610382, "global_step": 629, "epoch": 0, "lr": 9.999997337532869e-05} +{"train_loss": 0.38012072443962097, "train_loss_bc": 0.37898361682891846, "train_loss_llm": 0.11371111869812012, "grad_norm": 0.24278563261032104, "global_step": 630, "epoch": 0, "lr": 9.999997337532869e-05} +{"train_loss": 0.3614209294319153, "train_loss_bc": 0.35965263843536377, "train_loss_llm": 0.1768282949924469, "grad_norm": 0.3274829685688019, "global_step": 631, "epoch": 0, "lr": 9.999997337532869e-05} +{"train_loss": 0.3611985743045807, "train_loss_bc": 0.3599350154399872, "train_loss_llm": 0.1263551414012909, "grad_norm": 0.4635583162307739, "global_step": 632, "epoch": 0, "lr": 9.999997259800692e-05} +{"train_loss": 0.6066305041313171, "train_loss_bc": 0.6052604913711548, "train_loss_llm": 0.13699951767921448, "grad_norm": 0.2087971270084381, "global_step": 633, "epoch": 0, "lr": 9.999997259800692e-05} +{"train_loss": 0.5504487752914429, "train_loss_bc": 0.5488619804382324, "train_loss_llm": 0.15867656469345093, "grad_norm": 0.36908337473869324, "global_step": 634, "epoch": 0, "lr": 9.999997259800692e-05} +{"train_loss": 0.39519745111465454, "train_loss_bc": 0.3937787711620331, "train_loss_llm": 0.14186912775039673, "grad_norm": 0.4809526801109314, "global_step": 635, "epoch": 0, "lr": 9.999997259800692e-05} +{"train_loss": 0.41608870029449463, "train_loss_bc": 0.4146168828010559, "train_loss_llm": 0.14718031883239746, "grad_norm": 0.32000064849853516, "global_step": 636, "epoch": 0, "lr": 9.999997259800692e-05} +{"train_loss": 0.44419065117836, "train_loss_bc": 0.4432423710823059, "train_loss_llm": 0.09482666850090027, "grad_norm": 0.2030947059392929, "global_step": 637, "epoch": 0, "lr": 9.999997259800692e-05} +{"train_loss": 0.349100798368454, "train_loss_bc": 0.3480339050292969, "train_loss_llm": 0.10668930411338806, "grad_norm": 0.22338126599788666, "global_step": 638, "epoch": 0, "lr": 9.999997259800692e-05} +{"train_loss": 0.5182974934577942, "train_loss_bc": 0.5168792009353638, "train_loss_llm": 0.14182883501052856, "grad_norm": 0.254781037569046, "global_step": 639, "epoch": 0, "lr": 9.999997259800692e-05} +{"train_loss": 0.2992722988128662, "train_loss_bc": 0.29819560050964355, "train_loss_llm": 0.10766971111297607, "grad_norm": 0.27245032787323, "global_step": 640, "epoch": 0, "lr": 9.999997180950067e-05} +{"train_loss": 0.3970007598400116, "train_loss_bc": 0.3957017660140991, "train_loss_llm": 0.12989866733551025, "grad_norm": 0.13837115466594696, "global_step": 641, "epoch": 0, "lr": 9.999997180950067e-05} +{"train_loss": 0.44034236669540405, "train_loss_bc": 0.43897563219070435, "train_loss_llm": 0.13667485117912292, "grad_norm": 0.1540679931640625, "global_step": 642, "epoch": 0, "lr": 9.999997180950067e-05} +{"train_loss": 0.43921220302581787, "train_loss_bc": 0.438190221786499, "train_loss_llm": 0.1021992564201355, "grad_norm": 0.23249521851539612, "global_step": 643, "epoch": 0, "lr": 9.999997180950067e-05} +{"train_loss": 0.33070892095565796, "train_loss_bc": 0.3294546604156494, "train_loss_llm": 0.1254260540008545, "grad_norm": 0.20411910116672516, "global_step": 644, "epoch": 0, "lr": 9.999997180950067e-05} +{"train_loss": 0.4514554440975189, "train_loss_bc": 0.4503100514411926, "train_loss_llm": 0.11453825235366821, "grad_norm": 0.3148227632045746, "global_step": 645, "epoch": 0, "lr": 9.999997180950067e-05} +{"train_loss": 0.45420894026756287, "train_loss_bc": 0.4525565803050995, "train_loss_llm": 0.16523534059524536, "grad_norm": 0.2901774048805237, "global_step": 646, "epoch": 0, "lr": 9.999997180950067e-05} +{"train_loss": 0.34236758947372437, "train_loss_bc": 0.3410881459712982, "train_loss_llm": 0.1279456615447998, "grad_norm": 0.3068084716796875, "global_step": 647, "epoch": 0, "lr": 9.999997180950067e-05} +{"train_loss": 0.36798909306526184, "train_loss_bc": 0.36706113815307617, "train_loss_llm": 0.0927959680557251, "grad_norm": 0.3730047345161438, "global_step": 648, "epoch": 0, "lr": 9.999997100980993e-05} +{"train_loss": 0.5186164975166321, "train_loss_bc": 0.5175862312316895, "train_loss_llm": 0.10302457213401794, "grad_norm": 0.09078611433506012, "global_step": 649, "epoch": 0, "lr": 9.999997100980993e-05} +{"train_loss": 0.3458503186702728, "train_loss_bc": 0.34468936920166016, "train_loss_llm": 0.11609384417533875, "grad_norm": 0.17715536057949066, "global_step": 650, "epoch": 0, "lr": 9.999997100980993e-05} +{"train_loss": 0.5002723932266235, "train_loss_bc": 0.4988064765930176, "train_loss_llm": 0.146589457988739, "grad_norm": 0.2379833161830902, "global_step": 651, "epoch": 0, "lr": 9.999997100980993e-05} +{"train_loss": 0.4527694582939148, "train_loss_bc": 0.4513401985168457, "train_loss_llm": 0.14292606711387634, "grad_norm": 0.32560983300209045, "global_step": 652, "epoch": 0, "lr": 9.999997100980993e-05} +{"train_loss": 0.3723355829715729, "train_loss_bc": 0.3708031475543976, "train_loss_llm": 0.1532444953918457, "grad_norm": 0.337532639503479, "global_step": 653, "epoch": 0, "lr": 9.999997100980993e-05} +{"train_loss": 0.34707722067832947, "train_loss_bc": 0.3456161618232727, "train_loss_llm": 0.14610719680786133, "grad_norm": 0.4738572835922241, "global_step": 654, "epoch": 0, "lr": 9.999997100980993e-05} +{"train_loss": 0.5286421179771423, "train_loss_bc": 0.5275452733039856, "train_loss_llm": 0.10968193411827087, "grad_norm": 0.5108717679977417, "global_step": 655, "epoch": 0, "lr": 9.999997100980993e-05} +{"train_loss": 0.4695184528827667, "train_loss_bc": 0.46837037801742554, "train_loss_llm": 0.11480781435966492, "grad_norm": 0.6493931412696838, "global_step": 656, "epoch": 0, "lr": 9.99999701989347e-05} +{"train_loss": 0.4380721151828766, "train_loss_bc": 0.43665844202041626, "train_loss_llm": 0.14136740565299988, "grad_norm": 0.2012345939874649, "global_step": 657, "epoch": 0, "lr": 9.99999701989347e-05} +{"train_loss": 0.4790930151939392, "train_loss_bc": 0.47738754749298096, "train_loss_llm": 0.1705467700958252, "grad_norm": 0.1765247881412506, "global_step": 658, "epoch": 0, "lr": 9.99999701989347e-05} +{"train_loss": 0.4085133373737335, "train_loss_bc": 0.40731099247932434, "train_loss_llm": 0.12023511528968811, "grad_norm": 0.20196306705474854, "global_step": 659, "epoch": 0, "lr": 9.99999701989347e-05} +{"train_loss": 0.4066547453403473, "train_loss_bc": 0.4055253863334656, "train_loss_llm": 0.11293521523475647, "grad_norm": 0.15905286371707916, "global_step": 660, "epoch": 0, "lr": 9.99999701989347e-05} +{"train_loss": 0.49358707666397095, "train_loss_bc": 0.49217498302459717, "train_loss_llm": 0.14120939373970032, "grad_norm": 0.14518553018569946, "global_step": 661, "epoch": 0, "lr": 9.99999701989347e-05} +{"train_loss": 0.46300771832466125, "train_loss_bc": 0.4616585075855255, "train_loss_llm": 0.13492169976234436, "grad_norm": 0.21999910473823547, "global_step": 662, "epoch": 0, "lr": 9.99999701989347e-05} +{"train_loss": 0.4623633325099945, "train_loss_bc": 0.46140217781066895, "train_loss_llm": 0.09611445665359497, "grad_norm": 0.3314727544784546, "global_step": 663, "epoch": 0, "lr": 9.99999701989347e-05} +{"train_loss": 0.5139443278312683, "train_loss_bc": 0.5121505260467529, "train_loss_llm": 0.17937740683555603, "grad_norm": 0.3732542395591736, "global_step": 664, "epoch": 0, "lr": 9.9999969376875e-05} +{"train_loss": 0.3617967665195465, "train_loss_bc": 0.3607713282108307, "train_loss_llm": 0.10254350304603577, "grad_norm": 0.18659253418445587, "global_step": 665, "epoch": 0, "lr": 9.9999969376875e-05} +{"train_loss": 0.6355041861534119, "train_loss_bc": 0.6337125301361084, "train_loss_llm": 0.1791645884513855, "grad_norm": 0.5422189235687256, "global_step": 666, "epoch": 0, "lr": 9.9999969376875e-05} +{"train_loss": 0.4108755886554718, "train_loss_bc": 0.40987104177474976, "train_loss_llm": 0.10045334696769714, "grad_norm": 0.5765852928161621, "global_step": 667, "epoch": 0, "lr": 9.9999969376875e-05} +{"train_loss": 0.4670461118221283, "train_loss_bc": 0.4652903079986572, "train_loss_llm": 0.17558184266090393, "grad_norm": 0.7307344079017639, "global_step": 668, "epoch": 0, "lr": 9.9999969376875e-05} +{"train_loss": 0.5064515471458435, "train_loss_bc": 0.5050027370452881, "train_loss_llm": 0.14488250017166138, "grad_norm": 0.8418503403663635, "global_step": 669, "epoch": 0, "lr": 9.9999969376875e-05} +{"train_loss": 0.4039197564125061, "train_loss_bc": 0.40287265181541443, "train_loss_llm": 0.10470932722091675, "grad_norm": 0.8778255581855774, "global_step": 670, "epoch": 0, "lr": 9.9999969376875e-05} +{"train_loss": 0.5388345122337341, "train_loss_bc": 0.5375344157218933, "train_loss_llm": 0.13000819087028503, "grad_norm": 0.9715580344200134, "global_step": 671, "epoch": 0, "lr": 9.9999969376875e-05} +{"train_loss": 0.3694583475589752, "train_loss_bc": 0.3681648373603821, "train_loss_llm": 0.12935060262680054, "grad_norm": 1.072373867034912, "global_step": 672, "epoch": 0, "lr": 9.999996854363081e-05} +{"train_loss": 0.3775085210800171, "train_loss_bc": 0.376005083322525, "train_loss_llm": 0.1503431797027588, "grad_norm": 0.1132691279053688, "global_step": 673, "epoch": 0, "lr": 9.999996854363081e-05} +{"train_loss": 0.3032853305339813, "train_loss_bc": 0.3019849359989166, "train_loss_llm": 0.13003897666931152, "grad_norm": 0.14916090667247772, "global_step": 674, "epoch": 0, "lr": 9.999996854363081e-05} +{"train_loss": 0.4819992482662201, "train_loss_bc": 0.480733722448349, "train_loss_llm": 0.12655282020568848, "grad_norm": 0.29269519448280334, "global_step": 675, "epoch": 0, "lr": 9.999996854363081e-05} +{"train_loss": 0.3888154923915863, "train_loss_bc": 0.38713181018829346, "train_loss_llm": 0.1683672070503235, "grad_norm": 0.2845066785812378, "global_step": 676, "epoch": 0, "lr": 9.999996854363081e-05} +{"train_loss": 0.5516088604927063, "train_loss_bc": 0.550236701965332, "train_loss_llm": 0.13721629977226257, "grad_norm": 0.3500043451786041, "global_step": 677, "epoch": 0, "lr": 9.999996854363081e-05} +{"train_loss": 0.40210846066474915, "train_loss_bc": 0.4006125330924988, "train_loss_llm": 0.14959394931793213, "grad_norm": 0.3395308554172516, "global_step": 678, "epoch": 0, "lr": 9.999996854363081e-05} +{"train_loss": 0.4210907816886902, "train_loss_bc": 0.4198663532733917, "train_loss_llm": 0.12244343757629395, "grad_norm": 0.2646319270133972, "global_step": 679, "epoch": 0, "lr": 9.999996854363081e-05} +{"train_loss": 0.38542547821998596, "train_loss_bc": 0.3842034339904785, "train_loss_llm": 0.1222030520439148, "grad_norm": 0.26181671023368835, "global_step": 680, "epoch": 0, "lr": 9.999996769920214e-05} +{"train_loss": 0.36602091789245605, "train_loss_bc": 0.36506766080856323, "train_loss_llm": 0.09532523155212402, "grad_norm": 0.15427228808403015, "global_step": 681, "epoch": 0, "lr": 9.999996769920214e-05} +{"train_loss": 0.4792448878288269, "train_loss_bc": 0.4778333306312561, "train_loss_llm": 0.14115667343139648, "grad_norm": 0.1350840926170349, "global_step": 682, "epoch": 0, "lr": 9.999996769920214e-05} +{"train_loss": 0.47713136672973633, "train_loss_bc": 0.4756928086280823, "train_loss_llm": 0.14385607838630676, "grad_norm": 0.19607211649417877, "global_step": 683, "epoch": 0, "lr": 9.999996769920214e-05} +{"train_loss": 0.5072566866874695, "train_loss_bc": 0.5054144859313965, "train_loss_llm": 0.18421751260757446, "grad_norm": 0.26409026980400085, "global_step": 684, "epoch": 0, "lr": 9.999996769920214e-05} +{"train_loss": 0.3899046778678894, "train_loss_bc": 0.389045774936676, "train_loss_llm": 0.08589011430740356, "grad_norm": 0.3460109233856201, "global_step": 685, "epoch": 0, "lr": 9.999996769920214e-05} +{"train_loss": 0.4416724443435669, "train_loss_bc": 0.44038811326026917, "train_loss_llm": 0.1284339725971222, "grad_norm": 0.3730100393295288, "global_step": 686, "epoch": 0, "lr": 9.999996769920214e-05} +{"train_loss": 0.5536163449287415, "train_loss_bc": 0.5516965389251709, "train_loss_llm": 0.19197824597358704, "grad_norm": 0.44623008370399475, "global_step": 687, "epoch": 0, "lr": 9.999996769920214e-05} +{"train_loss": 0.4815637469291687, "train_loss_bc": 0.47983619570732117, "train_loss_llm": 0.17275595664978027, "grad_norm": 0.5500245690345764, "global_step": 688, "epoch": 0, "lr": 9.999996684358901e-05} +{"train_loss": 0.5841023325920105, "train_loss_bc": 0.5824371576309204, "train_loss_llm": 0.16651517152786255, "grad_norm": 0.234535351395607, "global_step": 689, "epoch": 0, "lr": 9.999996684358901e-05} +{"train_loss": 0.44914403557777405, "train_loss_bc": 0.4475044012069702, "train_loss_llm": 0.16396218538284302, "grad_norm": 0.25557929277420044, "global_step": 690, "epoch": 0, "lr": 9.999996684358901e-05} +{"train_loss": 0.3759511113166809, "train_loss_bc": 0.3744603991508484, "train_loss_llm": 0.14907196164131165, "grad_norm": 0.45482495427131653, "global_step": 691, "epoch": 0, "lr": 9.999996684358901e-05} +{"train_loss": 0.4200955927371979, "train_loss_bc": 0.418731153011322, "train_loss_llm": 0.136442631483078, "grad_norm": 0.3262624740600586, "global_step": 692, "epoch": 0, "lr": 9.999996684358901e-05} +{"train_loss": 0.4043809175491333, "train_loss_bc": 0.4031071066856384, "train_loss_llm": 0.12738072872161865, "grad_norm": 0.4146956503391266, "global_step": 693, "epoch": 0, "lr": 9.999996684358901e-05} +{"train_loss": 0.4611559510231018, "train_loss_bc": 0.4601346552371979, "train_loss_llm": 0.10212811827659607, "grad_norm": 0.30988994240760803, "global_step": 694, "epoch": 0, "lr": 9.999996684358901e-05} +{"train_loss": 0.5045652389526367, "train_loss_bc": 0.5030310750007629, "train_loss_llm": 0.15341421961784363, "grad_norm": 0.42709189653396606, "global_step": 695, "epoch": 0, "lr": 9.999996684358901e-05} +{"train_loss": 0.5231519341468811, "train_loss_bc": 0.5215438604354858, "train_loss_llm": 0.16080820560455322, "grad_norm": 0.42454394698143005, "global_step": 696, "epoch": 0, "lr": 9.999996597679137e-05} +{"train_loss": 0.4410187900066376, "train_loss_bc": 0.4399849474430084, "train_loss_llm": 0.10338541865348816, "grad_norm": 0.0741937905550003, "global_step": 697, "epoch": 0, "lr": 9.999996597679137e-05} +{"train_loss": 0.43824654817581177, "train_loss_bc": 0.4367842376232147, "train_loss_llm": 0.14623045921325684, "grad_norm": 0.154398575425148, "global_step": 698, "epoch": 0, "lr": 9.999996597679137e-05} +{"train_loss": 0.3794785439968109, "train_loss_bc": 0.3783220052719116, "train_loss_llm": 0.11565369367599487, "grad_norm": 0.12712602317333221, "global_step": 699, "epoch": 0, "lr": 9.999996597679137e-05} +{"train_loss": 0.4526432752609253, "train_loss_bc": 0.45120781660079956, "train_loss_llm": 0.14354535937309265, "grad_norm": 0.21711814403533936, "global_step": 700, "epoch": 0, "lr": 9.999996597679137e-05} +{"train_loss": 0.5144056081771851, "train_loss_bc": 0.5123848915100098, "train_loss_llm": 0.2020743191242218, "grad_norm": 0.1846182644367218, "global_step": 701, "epoch": 0, "lr": 9.999996597679137e-05} +{"train_loss": 0.487212210893631, "train_loss_bc": 0.48636239767074585, "train_loss_llm": 0.08498120307922363, "grad_norm": 0.35769122838974, "global_step": 702, "epoch": 0, "lr": 9.999996597679137e-05} +{"train_loss": 0.4268065392971039, "train_loss_bc": 0.42537808418273926, "train_loss_llm": 0.14284411072731018, "grad_norm": 0.24965746700763702, "global_step": 703, "epoch": 0, "lr": 9.999996597679137e-05} +{"train_loss": 0.24769797921180725, "train_loss_bc": 0.2469937652349472, "train_loss_llm": 0.07042080163955688, "grad_norm": 0.32342955470085144, "global_step": 704, "epoch": 0, "lr": 9.999996509880928e-05} +{"train_loss": 0.44799527525901794, "train_loss_bc": 0.4464735686779022, "train_loss_llm": 0.15217021107673645, "grad_norm": 0.1785305291414261, "global_step": 705, "epoch": 0, "lr": 9.999996509880928e-05} +{"train_loss": 0.48176732659339905, "train_loss_bc": 0.4803884029388428, "train_loss_llm": 0.13789373636245728, "grad_norm": 0.3359695374965668, "global_step": 706, "epoch": 0, "lr": 9.999996509880928e-05} +{"train_loss": 0.40119537711143494, "train_loss_bc": 0.39965328574180603, "train_loss_llm": 0.1542084515094757, "grad_norm": 0.3453904390335083, "global_step": 707, "epoch": 0, "lr": 9.999996509880928e-05} +{"train_loss": 0.3725847005844116, "train_loss_bc": 0.3716793358325958, "train_loss_llm": 0.09053558111190796, "grad_norm": 0.3976081311702728, "global_step": 708, "epoch": 0, "lr": 9.999996509880928e-05} +{"train_loss": 0.3447168469429016, "train_loss_bc": 0.3432568907737732, "train_loss_llm": 0.14599433541297913, "grad_norm": 0.364182710647583, "global_step": 709, "epoch": 0, "lr": 9.999996509880928e-05} +{"train_loss": 0.4946703016757965, "train_loss_bc": 0.49284207820892334, "train_loss_llm": 0.18282106518745422, "grad_norm": 0.5332158803939819, "global_step": 710, "epoch": 0, "lr": 9.999996509880928e-05} +{"train_loss": 0.5023349523544312, "train_loss_bc": 0.500494122505188, "train_loss_llm": 0.18408560752868652, "grad_norm": 0.7355331778526306, "global_step": 711, "epoch": 0, "lr": 9.999996509880928e-05} +{"train_loss": 0.5317637920379639, "train_loss_bc": 0.5301767587661743, "train_loss_llm": 0.1587003469467163, "grad_norm": 0.5979422926902771, "global_step": 712, "epoch": 0, "lr": 9.999996420964269e-05} +{"train_loss": 0.3867203891277313, "train_loss_bc": 0.38556772470474243, "train_loss_llm": 0.11526641249656677, "grad_norm": 0.15208011865615845, "global_step": 713, "epoch": 0, "lr": 9.999996420964269e-05} +{"train_loss": 0.324870228767395, "train_loss_bc": 0.3240911662578583, "train_loss_llm": 0.07790669798851013, "grad_norm": 0.27033838629722595, "global_step": 714, "epoch": 0, "lr": 9.999996420964269e-05} +{"train_loss": 0.4077690839767456, "train_loss_bc": 0.4064152240753174, "train_loss_llm": 0.1353853940963745, "grad_norm": 0.2142527550458908, "global_step": 715, "epoch": 0, "lr": 9.999996420964269e-05} +{"train_loss": 0.38206955790519714, "train_loss_bc": 0.3805384039878845, "train_loss_llm": 0.15311622619628906, "grad_norm": 0.22046618163585663, "global_step": 716, "epoch": 0, "lr": 9.999996420964269e-05} +{"train_loss": 0.503631591796875, "train_loss_bc": 0.5022858381271362, "train_loss_llm": 0.13457295298576355, "grad_norm": 0.19850732386112213, "global_step": 717, "epoch": 0, "lr": 9.999996420964269e-05} +{"train_loss": 0.3522709608078003, "train_loss_bc": 0.350716233253479, "train_loss_llm": 0.15547418594360352, "grad_norm": 0.26946842670440674, "global_step": 718, "epoch": 0, "lr": 9.999996420964269e-05} +{"train_loss": 0.39072301983833313, "train_loss_bc": 0.3893094062805176, "train_loss_llm": 0.14136117696762085, "grad_norm": 0.28080469369888306, "global_step": 719, "epoch": 0, "lr": 9.999996420964269e-05} +{"train_loss": 0.4491545557975769, "train_loss_bc": 0.4474574625492096, "train_loss_llm": 0.16971072554588318, "grad_norm": 0.30231422185897827, "global_step": 720, "epoch": 0, "lr": 9.999996330929162e-05} +{"train_loss": 0.3498592674732208, "train_loss_bc": 0.3485373854637146, "train_loss_llm": 0.13218703866004944, "grad_norm": 0.19960981607437134, "global_step": 721, "epoch": 0, "lr": 9.999996330929162e-05} +{"train_loss": 0.3955504298210144, "train_loss_bc": 0.39425837993621826, "train_loss_llm": 0.12920638918876648, "grad_norm": 0.2596740424633026, "global_step": 722, "epoch": 0, "lr": 9.999996330929162e-05} +{"train_loss": 0.4701121151447296, "train_loss_bc": 0.468752384185791, "train_loss_llm": 0.13597255945205688, "grad_norm": 0.31495392322540283, "global_step": 723, "epoch": 0, "lr": 9.999996330929162e-05} +{"train_loss": 0.5052410960197449, "train_loss_bc": 0.5031529664993286, "train_loss_llm": 0.20881444215774536, "grad_norm": 0.2587316036224365, "global_step": 724, "epoch": 0, "lr": 9.999996330929162e-05} +{"train_loss": 0.4714301824569702, "train_loss_bc": 0.4699521064758301, "train_loss_llm": 0.14780807495117188, "grad_norm": 0.37743210792541504, "global_step": 725, "epoch": 0, "lr": 9.999996330929162e-05} +{"train_loss": 0.4981463551521301, "train_loss_bc": 0.49689823389053345, "train_loss_llm": 0.12481123208999634, "grad_norm": 0.5135627388954163, "global_step": 726, "epoch": 0, "lr": 9.999996330929162e-05} +{"train_loss": 0.513901948928833, "train_loss_bc": 0.5121231079101562, "train_loss_llm": 0.17788228392601013, "grad_norm": 0.48381975293159485, "global_step": 727, "epoch": 0, "lr": 9.999996330929162e-05} +{"train_loss": 0.5299534797668457, "train_loss_bc": 0.5286766290664673, "train_loss_llm": 0.12768593430519104, "grad_norm": 0.5247568488121033, "global_step": 728, "epoch": 0, "lr": 9.999996239775608e-05} +{"train_loss": 0.44028180837631226, "train_loss_bc": 0.43916481733322144, "train_loss_llm": 0.1116977334022522, "grad_norm": 0.19921325147151947, "global_step": 729, "epoch": 0, "lr": 9.999996239775608e-05} +{"train_loss": 0.30627939105033875, "train_loss_bc": 0.30522727966308594, "train_loss_llm": 0.10521078109741211, "grad_norm": 0.18406879901885986, "global_step": 730, "epoch": 0, "lr": 9.999996239775608e-05} +{"train_loss": 0.3498186767101288, "train_loss_bc": 0.3485940098762512, "train_loss_llm": 0.12246569991111755, "grad_norm": 0.12964852154254913, "global_step": 731, "epoch": 0, "lr": 9.999996239775608e-05} +{"train_loss": 0.46024417877197266, "train_loss_bc": 0.45908302068710327, "train_loss_llm": 0.11611557006835938, "grad_norm": 0.16968220472335815, "global_step": 732, "epoch": 0, "lr": 9.999996239775608e-05} +{"train_loss": 0.4574472904205322, "train_loss_bc": 0.45637768507003784, "train_loss_llm": 0.1069595217704773, "grad_norm": 0.2518070340156555, "global_step": 733, "epoch": 0, "lr": 9.999996239775608e-05} +{"train_loss": 0.3599821627140045, "train_loss_bc": 0.3585447371006012, "train_loss_llm": 0.14374321699142456, "grad_norm": 0.27374696731567383, "global_step": 734, "epoch": 0, "lr": 9.999996239775608e-05} +{"train_loss": 0.30314844846725464, "train_loss_bc": 0.30214226245880127, "train_loss_llm": 0.1006200909614563, "grad_norm": 0.2712038457393646, "global_step": 735, "epoch": 0, "lr": 9.999996239775608e-05} +{"train_loss": 0.47343429923057556, "train_loss_bc": 0.47231245040893555, "train_loss_llm": 0.11218541860580444, "grad_norm": 0.30330991744995117, "global_step": 736, "epoch": 0, "lr": 9.999996147503605e-05} +{"train_loss": 0.37856918573379517, "train_loss_bc": 0.37734994292259216, "train_loss_llm": 0.12192380428314209, "grad_norm": 0.08694847673177719, "global_step": 737, "epoch": 0, "lr": 9.999996147503605e-05} +{"train_loss": 0.568943440914154, "train_loss_bc": 0.5675009489059448, "train_loss_llm": 0.14425000548362732, "grad_norm": 0.13533131778240204, "global_step": 738, "epoch": 0, "lr": 9.999996147503605e-05} +{"train_loss": 0.546090304851532, "train_loss_bc": 0.5448291301727295, "train_loss_llm": 0.12611711025238037, "grad_norm": 0.1940602958202362, "global_step": 739, "epoch": 0, "lr": 9.999996147503605e-05} +{"train_loss": 0.47889894247055054, "train_loss_bc": 0.4775034189224243, "train_loss_llm": 0.1395530104637146, "grad_norm": 0.18891654908657074, "global_step": 740, "epoch": 0, "lr": 9.999996147503605e-05} +{"train_loss": 0.5004976391792297, "train_loss_bc": 0.49890419840812683, "train_loss_llm": 0.1593429148197174, "grad_norm": 0.22369500994682312, "global_step": 741, "epoch": 0, "lr": 9.999996147503605e-05} +{"train_loss": 0.4289986789226532, "train_loss_bc": 0.42788243293762207, "train_loss_llm": 0.11162453889846802, "grad_norm": 0.339557021856308, "global_step": 742, "epoch": 0, "lr": 9.999996147503605e-05} +{"train_loss": 0.5300840139389038, "train_loss_bc": 0.528813362121582, "train_loss_llm": 0.12706738710403442, "grad_norm": 0.30943524837493896, "global_step": 743, "epoch": 0, "lr": 9.999996147503605e-05} +{"train_loss": 0.47221875190734863, "train_loss_bc": 0.47100862860679626, "train_loss_llm": 0.1210109293460846, "grad_norm": 0.4173502027988434, "global_step": 744, "epoch": 0, "lr": 9.999996054113154e-05} +{"train_loss": 0.3905128538608551, "train_loss_bc": 0.3884217143058777, "train_loss_llm": 0.20911380648612976, "grad_norm": 0.15257810056209564, "global_step": 745, "epoch": 0, "lr": 9.999996054113154e-05} +{"train_loss": 0.5026034712791443, "train_loss_bc": 0.5013154149055481, "train_loss_llm": 0.12880536913871765, "grad_norm": 0.20023779571056366, "global_step": 746, "epoch": 0, "lr": 9.999996054113154e-05} +{"train_loss": 0.47565290331840515, "train_loss_bc": 0.47420042753219604, "train_loss_llm": 0.14524823427200317, "grad_norm": 0.28291016817092896, "global_step": 747, "epoch": 0, "lr": 9.999996054113154e-05} +{"train_loss": 0.48675331473350525, "train_loss_bc": 0.4853453040122986, "train_loss_llm": 0.14080080389976501, "grad_norm": 0.3061433732509613, "global_step": 748, "epoch": 0, "lr": 9.999996054113154e-05} +{"train_loss": 0.49515989422798157, "train_loss_bc": 0.4937002658843994, "train_loss_llm": 0.14596214890480042, "grad_norm": 0.40628647804260254, "global_step": 749, "epoch": 0, "lr": 9.999996054113154e-05} +{"train_loss": 0.2966780662536621, "train_loss_bc": 0.2958890199661255, "train_loss_llm": 0.07890340685844421, "grad_norm": 0.45909613370895386, "global_step": 750, "epoch": 0, "lr": 9.999996054113154e-05} +{"train_loss": 0.2501656413078308, "train_loss_bc": 0.24916397035121918, "train_loss_llm": 0.10016623139381409, "grad_norm": 0.45718130469322205, "global_step": 751, "epoch": 0, "lr": 9.999996054113154e-05} +{"train_loss": 0.46850523352622986, "train_loss_bc": 0.4669528603553772, "train_loss_llm": 0.15523681044578552, "grad_norm": 0.5289716720581055, "global_step": 752, "epoch": 0, "lr": 9.999995959604256e-05} +{"train_loss": 0.359755277633667, "train_loss_bc": 0.3583693504333496, "train_loss_llm": 0.13859277963638306, "grad_norm": 0.10197532922029495, "global_step": 753, "epoch": 0, "lr": 9.999995959604256e-05} +{"train_loss": 0.442962110042572, "train_loss_bc": 0.44167059659957886, "train_loss_llm": 0.12915101647377014, "grad_norm": 0.14329703152179718, "global_step": 754, "epoch": 0, "lr": 9.999995959604256e-05} +{"train_loss": 0.402640163898468, "train_loss_bc": 0.4016171097755432, "train_loss_llm": 0.10230538249015808, "grad_norm": 0.19640891253948212, "global_step": 755, "epoch": 0, "lr": 9.999995959604256e-05} +{"train_loss": 0.5023623108863831, "train_loss_bc": 0.5007985830307007, "train_loss_llm": 0.1563742458820343, "grad_norm": 0.2108265459537506, "global_step": 756, "epoch": 0, "lr": 9.999995959604256e-05} +{"train_loss": 0.3814651668071747, "train_loss_bc": 0.38031840324401855, "train_loss_llm": 0.11467549204826355, "grad_norm": 0.2650246024131775, "global_step": 757, "epoch": 0, "lr": 9.999995959604256e-05} +{"train_loss": 0.4334079623222351, "train_loss_bc": 0.43238040804862976, "train_loss_llm": 0.10275408625602722, "grad_norm": 0.32268771529197693, "global_step": 758, "epoch": 0, "lr": 9.999995959604256e-05} +{"train_loss": 0.5475137233734131, "train_loss_bc": 0.5460928082466125, "train_loss_llm": 0.14209246635437012, "grad_norm": 0.39937904477119446, "global_step": 759, "epoch": 0, "lr": 9.999995959604256e-05} +{"train_loss": 0.3154490292072296, "train_loss_bc": 0.3144824504852295, "train_loss_llm": 0.0966586172580719, "grad_norm": 0.4118819236755371, "global_step": 760, "epoch": 0, "lr": 9.99999586397691e-05} +{"train_loss": 0.4999276399612427, "train_loss_bc": 0.4982565641403198, "train_loss_llm": 0.16710838675498962, "grad_norm": 0.17443329095840454, "global_step": 761, "epoch": 0, "lr": 9.99999586397691e-05} +{"train_loss": 0.3348698019981384, "train_loss_bc": 0.33375442028045654, "train_loss_llm": 0.11153775453567505, "grad_norm": 0.1769266426563263, "global_step": 762, "epoch": 0, "lr": 9.99999586397691e-05} +{"train_loss": 0.46121659874916077, "train_loss_bc": 0.4601248502731323, "train_loss_llm": 0.1091742217540741, "grad_norm": 0.2355940043926239, "global_step": 763, "epoch": 0, "lr": 9.99999586397691e-05} +{"train_loss": 0.5022720694541931, "train_loss_bc": 0.5010197162628174, "train_loss_llm": 0.12523508071899414, "grad_norm": 0.28587213158607483, "global_step": 764, "epoch": 0, "lr": 9.99999586397691e-05} +{"train_loss": 0.3365619480609894, "train_loss_bc": 0.336049884557724, "train_loss_llm": 0.05120629072189331, "grad_norm": 0.2753007113933563, "global_step": 765, "epoch": 0, "lr": 9.99999586397691e-05} +{"train_loss": 0.47622284293174744, "train_loss_bc": 0.4748573899269104, "train_loss_llm": 0.13654515147209167, "grad_norm": 0.43221092224121094, "global_step": 766, "epoch": 0, "lr": 9.99999586397691e-05} +{"train_loss": 0.4418603479862213, "train_loss_bc": 0.440447598695755, "train_loss_llm": 0.14127585291862488, "grad_norm": 0.3395805060863495, "global_step": 767, "epoch": 0, "lr": 9.99999586397691e-05} +{"train_loss": 0.4092864990234375, "train_loss_bc": 0.40789875388145447, "train_loss_llm": 0.13877582550048828, "grad_norm": 0.4346903860569, "global_step": 768, "epoch": 0, "lr": 9.999995767231117e-05} +{"train_loss": 0.37854093313217163, "train_loss_bc": 0.3772284686565399, "train_loss_llm": 0.13124549388885498, "grad_norm": 0.13683906197547913, "global_step": 769, "epoch": 0, "lr": 9.999995767231117e-05} +{"train_loss": 0.3885970413684845, "train_loss_bc": 0.3867207169532776, "train_loss_llm": 0.18763259053230286, "grad_norm": 0.16374289989471436, "global_step": 770, "epoch": 0, "lr": 9.999995767231117e-05} +{"train_loss": 0.509136974811554, "train_loss_bc": 0.5079625248908997, "train_loss_llm": 0.11744439601898193, "grad_norm": 0.19966085255146027, "global_step": 771, "epoch": 0, "lr": 9.999995767231117e-05} +{"train_loss": 0.4480476379394531, "train_loss_bc": 0.4462229013442993, "train_loss_llm": 0.18247252702713013, "grad_norm": 0.2061576247215271, "global_step": 772, "epoch": 0, "lr": 9.999995767231117e-05} +{"train_loss": 0.44607651233673096, "train_loss_bc": 0.444861501455307, "train_loss_llm": 0.12150022387504578, "grad_norm": 0.20870722830295563, "global_step": 773, "epoch": 0, "lr": 9.999995767231117e-05} +{"train_loss": 0.45687711238861084, "train_loss_bc": 0.45550864934921265, "train_loss_llm": 0.13684555888175964, "grad_norm": 0.3540310859680176, "global_step": 774, "epoch": 0, "lr": 9.999995767231117e-05} +{"train_loss": 0.28119978308677673, "train_loss_bc": 0.2800223231315613, "train_loss_llm": 0.11774641275405884, "grad_norm": 0.3708130717277527, "global_step": 775, "epoch": 0, "lr": 9.999995767231117e-05} +{"train_loss": 0.36435937881469727, "train_loss_bc": 0.36323535442352295, "train_loss_llm": 0.1124035120010376, "grad_norm": 0.3589748442173004, "global_step": 776, "epoch": 0, "lr": 9.999995669366873e-05} +{"train_loss": 0.3511892259120941, "train_loss_bc": 0.34960442781448364, "train_loss_llm": 0.15847864747047424, "grad_norm": 0.15811693668365479, "global_step": 777, "epoch": 0, "lr": 9.999995669366873e-05} +{"train_loss": 0.4977981746196747, "train_loss_bc": 0.4965898394584656, "train_loss_llm": 0.12083470821380615, "grad_norm": 0.11093614995479584, "global_step": 778, "epoch": 0, "lr": 9.999995669366873e-05} +{"train_loss": 0.3353201746940613, "train_loss_bc": 0.33430755138397217, "train_loss_llm": 0.10126262903213501, "grad_norm": 0.16895242035388947, "global_step": 779, "epoch": 0, "lr": 9.999995669366873e-05} +{"train_loss": 0.5816384553909302, "train_loss_bc": 0.5798485279083252, "train_loss_llm": 0.17899319529533386, "grad_norm": 0.24064700305461884, "global_step": 780, "epoch": 0, "lr": 9.999995669366873e-05} +{"train_loss": 0.45553719997406006, "train_loss_bc": 0.45428967475891113, "train_loss_llm": 0.12475389242172241, "grad_norm": 0.2814551591873169, "global_step": 781, "epoch": 0, "lr": 9.999995669366873e-05} +{"train_loss": 0.5738955736160278, "train_loss_bc": 0.572656512260437, "train_loss_llm": 0.12390339374542236, "grad_norm": 0.3519173860549927, "global_step": 782, "epoch": 0, "lr": 9.999995669366873e-05} +{"train_loss": 0.55666184425354, "train_loss_bc": 0.5550835132598877, "train_loss_llm": 0.15783312916755676, "grad_norm": 0.3521682918071747, "global_step": 783, "epoch": 0, "lr": 9.999995669366873e-05} +{"train_loss": 0.4659648835659027, "train_loss_bc": 0.4646364748477936, "train_loss_llm": 0.13283956050872803, "grad_norm": 0.45815718173980713, "global_step": 784, "epoch": 0, "lr": 9.999995570384185e-05} +{"train_loss": 0.4600337743759155, "train_loss_bc": 0.4588106572628021, "train_loss_llm": 0.12231138348579407, "grad_norm": 0.20575882494449615, "global_step": 785, "epoch": 0, "lr": 9.999995570384185e-05} +{"train_loss": 0.44163277745246887, "train_loss_bc": 0.4400327801704407, "train_loss_llm": 0.15999844670295715, "grad_norm": 0.21481171250343323, "global_step": 786, "epoch": 0, "lr": 9.999995570384185e-05} +{"train_loss": 0.32197827100753784, "train_loss_bc": 0.32115018367767334, "train_loss_llm": 0.0828092098236084, "grad_norm": 0.32799476385116577, "global_step": 787, "epoch": 0, "lr": 9.999995570384185e-05} +{"train_loss": 0.5294073820114136, "train_loss_bc": 0.5280924439430237, "train_loss_llm": 0.13149195909500122, "grad_norm": 0.47922441363334656, "global_step": 788, "epoch": 0, "lr": 9.999995570384185e-05} +{"train_loss": 0.41958868503570557, "train_loss_bc": 0.4182693362236023, "train_loss_llm": 0.1319340169429779, "grad_norm": 0.5975518822669983, "global_step": 789, "epoch": 0, "lr": 9.999995570384185e-05} +{"train_loss": 0.3610004782676697, "train_loss_bc": 0.3597356975078583, "train_loss_llm": 0.1264781951904297, "grad_norm": 0.698198139667511, "global_step": 790, "epoch": 0, "lr": 9.999995570384185e-05} +{"train_loss": 0.4488995373249054, "train_loss_bc": 0.4473235011100769, "train_loss_llm": 0.1576026976108551, "grad_norm": 0.7349069118499756, "global_step": 791, "epoch": 0, "lr": 9.999995570384185e-05} +{"train_loss": 0.3241168260574341, "train_loss_bc": 0.32292285561561584, "train_loss_llm": 0.11939609050750732, "grad_norm": 0.7909736633300781, "global_step": 792, "epoch": 0, "lr": 9.999995470283045e-05} +{"train_loss": 0.5295166969299316, "train_loss_bc": 0.5278809070587158, "train_loss_llm": 0.1635788381099701, "grad_norm": 0.1566392034292221, "global_step": 793, "epoch": 0, "lr": 9.999995470283045e-05} +{"train_loss": 0.5089865326881409, "train_loss_bc": 0.5071623921394348, "train_loss_llm": 0.18241682648658752, "grad_norm": 0.14472274482250214, "global_step": 794, "epoch": 0, "lr": 9.999995470283045e-05} +{"train_loss": 0.4349845051765442, "train_loss_bc": 0.4335893988609314, "train_loss_llm": 0.13950982689857483, "grad_norm": 0.22686867415905, "global_step": 795, "epoch": 0, "lr": 9.999995470283045e-05} +{"train_loss": 0.31643012166023254, "train_loss_bc": 0.3153582811355591, "train_loss_llm": 0.10718479752540588, "grad_norm": 0.30232539772987366, "global_step": 796, "epoch": 0, "lr": 9.999995470283045e-05} +{"train_loss": 0.28871825337409973, "train_loss_bc": 0.28768211603164673, "train_loss_llm": 0.10361319780349731, "grad_norm": 0.38791507482528687, "global_step": 797, "epoch": 0, "lr": 9.999995470283045e-05} +{"train_loss": 0.41977596282958984, "train_loss_bc": 0.418587863445282, "train_loss_llm": 0.11880886554718018, "grad_norm": 0.47844067215919495, "global_step": 798, "epoch": 0, "lr": 9.999995470283045e-05} +{"train_loss": 0.43506383895874023, "train_loss_bc": 0.4337069094181061, "train_loss_llm": 0.1356930136680603, "grad_norm": 0.5041659474372864, "global_step": 799, "epoch": 0, "lr": 9.999995470283045e-05} +{"train_loss": 0.5048915147781372, "train_loss_bc": 0.5036287903785706, "train_loss_llm": 0.12627306580543518, "grad_norm": 0.554296612739563, "global_step": 800, "epoch": 0, "lr": 9.99999536906346e-05} +{"train_loss": 0.4756609797477722, "train_loss_bc": 0.47453808784484863, "train_loss_llm": 0.11228945851325989, "grad_norm": 0.1021411269903183, "global_step": 801, "epoch": 0, "lr": 9.99999536906346e-05} +{"train_loss": 0.5929127931594849, "train_loss_bc": 0.5913003087043762, "train_loss_llm": 0.16124817728996277, "grad_norm": 0.16123424470424652, "global_step": 802, "epoch": 0, "lr": 9.99999536906346e-05} +{"train_loss": 0.2964675724506378, "train_loss_bc": 0.2958076596260071, "train_loss_llm": 0.06598988175392151, "grad_norm": 0.16209018230438232, "global_step": 803, "epoch": 0, "lr": 9.99999536906346e-05} +{"train_loss": 0.3405551016330719, "train_loss_bc": 0.33945536613464355, "train_loss_llm": 0.10997328162193298, "grad_norm": 0.1590115875005722, "global_step": 804, "epoch": 0, "lr": 9.99999536906346e-05} +{"train_loss": 0.4123091697692871, "train_loss_bc": 0.4110938310623169, "train_loss_llm": 0.12153404951095581, "grad_norm": 0.21888233721256256, "global_step": 805, "epoch": 0, "lr": 9.99999536906346e-05} +{"train_loss": 0.4325730800628662, "train_loss_bc": 0.4309022128582001, "train_loss_llm": 0.16708767414093018, "grad_norm": 0.22386661171913147, "global_step": 806, "epoch": 0, "lr": 9.99999536906346e-05} +{"train_loss": 0.44433650374412537, "train_loss_bc": 0.44288375973701477, "train_loss_llm": 0.14527517557144165, "grad_norm": 0.3503759205341339, "global_step": 807, "epoch": 0, "lr": 9.99999536906346e-05} +{"train_loss": 0.4519703984260559, "train_loss_bc": 0.4506406784057617, "train_loss_llm": 0.13297182321548462, "grad_norm": 0.5202891826629639, "global_step": 808, "epoch": 0, "lr": 9.999995266725428e-05} +{"train_loss": 0.46295109391212463, "train_loss_bc": 0.461530864238739, "train_loss_llm": 0.14202404022216797, "grad_norm": 0.1520911306142807, "global_step": 809, "epoch": 0, "lr": 9.999995266725428e-05} +{"train_loss": 0.5591545104980469, "train_loss_bc": 0.5580669641494751, "train_loss_llm": 0.10875451564788818, "grad_norm": 0.16319619119167328, "global_step": 810, "epoch": 0, "lr": 9.999995266725428e-05} +{"train_loss": 0.26364827156066895, "train_loss_bc": 0.2628573775291443, "train_loss_llm": 0.07908812165260315, "grad_norm": 0.23261579871177673, "global_step": 811, "epoch": 0, "lr": 9.999995266725428e-05} +{"train_loss": 0.3030109107494354, "train_loss_bc": 0.30185097455978394, "train_loss_llm": 0.11599314212799072, "grad_norm": 0.37339770793914795, "global_step": 812, "epoch": 0, "lr": 9.999995266725428e-05} +{"train_loss": 0.4111844599246979, "train_loss_bc": 0.40956294536590576, "train_loss_llm": 0.1621505320072174, "grad_norm": 0.2744762897491455, "global_step": 813, "epoch": 0, "lr": 9.999995266725428e-05} +{"train_loss": 0.37956929206848145, "train_loss_bc": 0.3784160614013672, "train_loss_llm": 0.11532264947891235, "grad_norm": 0.43625593185424805, "global_step": 814, "epoch": 0, "lr": 9.999995266725428e-05} +{"train_loss": 0.4456196427345276, "train_loss_bc": 0.44406354427337646, "train_loss_llm": 0.1556093990802765, "grad_norm": 0.4496365487575531, "global_step": 815, "epoch": 0, "lr": 9.999995266725428e-05} +{"train_loss": 0.38981252908706665, "train_loss_bc": 0.3888477683067322, "train_loss_llm": 0.09647500514984131, "grad_norm": 0.3684163987636566, "global_step": 816, "epoch": 0, "lr": 9.999995163268947e-05} +{"train_loss": 0.3970646858215332, "train_loss_bc": 0.39601749181747437, "train_loss_llm": 0.10472023487091064, "grad_norm": 0.2269795835018158, "global_step": 817, "epoch": 0, "lr": 9.999995163268947e-05} +{"train_loss": 0.2952520549297333, "train_loss_bc": 0.29387107491493225, "train_loss_llm": 0.13809815049171448, "grad_norm": 0.37163665890693665, "global_step": 818, "epoch": 0, "lr": 9.999995163268947e-05} +{"train_loss": 0.49978846311569214, "train_loss_bc": 0.49854010343551636, "train_loss_llm": 0.12483593821525574, "grad_norm": 0.23900480568408966, "global_step": 819, "epoch": 0, "lr": 9.999995163268947e-05} +{"train_loss": 0.5154771208763123, "train_loss_bc": 0.5140070915222168, "train_loss_llm": 0.14700275659561157, "grad_norm": 0.24776293337345123, "global_step": 820, "epoch": 0, "lr": 9.999995163268947e-05} +{"train_loss": 0.4097417891025543, "train_loss_bc": 0.4086484909057617, "train_loss_llm": 0.10932886600494385, "grad_norm": 0.23228804767131805, "global_step": 821, "epoch": 0, "lr": 9.999995163268947e-05} +{"train_loss": 0.48011040687561035, "train_loss_bc": 0.47857582569122314, "train_loss_llm": 0.15345925092697144, "grad_norm": 0.34754064679145813, "global_step": 822, "epoch": 0, "lr": 9.999995163268947e-05} +{"train_loss": 0.36280250549316406, "train_loss_bc": 0.36192649602890015, "train_loss_llm": 0.08760061860084534, "grad_norm": 0.3779703378677368, "global_step": 823, "epoch": 0, "lr": 9.999995163268947e-05} +{"train_loss": 0.3764092028141022, "train_loss_bc": 0.3750823438167572, "train_loss_llm": 0.1326860785484314, "grad_norm": 0.29815050959587097, "global_step": 824, "epoch": 0, "lr": 9.999995058694018e-05} +{"train_loss": 0.5065320134162903, "train_loss_bc": 0.5053824186325073, "train_loss_llm": 0.1149618923664093, "grad_norm": 0.11318471282720566, "global_step": 825, "epoch": 0, "lr": 9.999995058694018e-05} +{"train_loss": 0.42440029978752136, "train_loss_bc": 0.4232295751571655, "train_loss_llm": 0.1170712411403656, "grad_norm": 0.1330985724925995, "global_step": 826, "epoch": 0, "lr": 9.999995058694018e-05} +{"train_loss": 0.4227786064147949, "train_loss_bc": 0.4210730791091919, "train_loss_llm": 0.17055395245552063, "grad_norm": 0.1604996621608734, "global_step": 827, "epoch": 0, "lr": 9.999995058694018e-05} +{"train_loss": 0.49485889077186584, "train_loss_bc": 0.4933386445045471, "train_loss_llm": 0.15202561020851135, "grad_norm": 0.1900646984577179, "global_step": 828, "epoch": 0, "lr": 9.999995058694018e-05} +{"train_loss": 0.40058740973472595, "train_loss_bc": 0.398720920085907, "train_loss_llm": 0.1866493821144104, "grad_norm": 0.2794025242328644, "global_step": 829, "epoch": 0, "lr": 9.999995058694018e-05} +{"train_loss": 0.42090266942977905, "train_loss_bc": 0.41947275400161743, "train_loss_llm": 0.14299049973487854, "grad_norm": 0.31900399923324585, "global_step": 830, "epoch": 0, "lr": 9.999995058694018e-05} +{"train_loss": 0.5276516675949097, "train_loss_bc": 0.5261500477790833, "train_loss_llm": 0.15016040205955505, "grad_norm": 0.42090725898742676, "global_step": 831, "epoch": 0, "lr": 9.999995058694018e-05} +{"train_loss": 0.4433680474758148, "train_loss_bc": 0.4417176842689514, "train_loss_llm": 0.16503658890724182, "grad_norm": 0.47822070121765137, "global_step": 832, "epoch": 0, "lr": 9.999994953000642e-05} +{"train_loss": 0.42253556847572327, "train_loss_bc": 0.42153528332710266, "train_loss_llm": 0.10002976655960083, "grad_norm": 0.2791462242603302, "global_step": 833, "epoch": 0, "lr": 9.999994953000642e-05} +{"train_loss": 0.42078253626823425, "train_loss_bc": 0.41995173692703247, "train_loss_llm": 0.08308050036430359, "grad_norm": 0.296310693025589, "global_step": 834, "epoch": 0, "lr": 9.999994953000642e-05} +{"train_loss": 0.4137524664402008, "train_loss_bc": 0.4122909903526306, "train_loss_llm": 0.14614883065223694, "grad_norm": 0.3143576383590698, "global_step": 835, "epoch": 0, "lr": 9.999994953000642e-05} +{"train_loss": 0.36327284574508667, "train_loss_bc": 0.36239904165267944, "train_loss_llm": 0.08738100528717041, "grad_norm": 0.3195045292377472, "global_step": 836, "epoch": 0, "lr": 9.999994953000642e-05} +{"train_loss": 0.40926259756088257, "train_loss_bc": 0.4079577922821045, "train_loss_llm": 0.13048097491264343, "grad_norm": 0.5069186687469482, "global_step": 837, "epoch": 0, "lr": 9.999994953000642e-05} +{"train_loss": 0.4288123846054077, "train_loss_bc": 0.42769938707351685, "train_loss_llm": 0.11130073666572571, "grad_norm": 0.577712893486023, "global_step": 838, "epoch": 0, "lr": 9.999994953000642e-05} +{"train_loss": 0.5182204246520996, "train_loss_bc": 0.5168851613998413, "train_loss_llm": 0.13352781534194946, "grad_norm": 0.6310889720916748, "global_step": 839, "epoch": 0, "lr": 9.999994953000642e-05} +{"train_loss": 0.37798523902893066, "train_loss_bc": 0.37679821252822876, "train_loss_llm": 0.11870148777961731, "grad_norm": 0.7218919396400452, "global_step": 840, "epoch": 0, "lr": 9.999994846188819e-05} +{"train_loss": 0.4202688932418823, "train_loss_bc": 0.41848820447921753, "train_loss_llm": 0.1780681312084198, "grad_norm": 0.11473484337329865, "global_step": 841, "epoch": 0, "lr": 9.999994846188819e-05} +{"train_loss": 0.4164406359195709, "train_loss_bc": 0.41493093967437744, "train_loss_llm": 0.15097007155418396, "grad_norm": 0.21355359256267548, "global_step": 842, "epoch": 0, "lr": 9.999994846188819e-05} +{"train_loss": 0.44272536039352417, "train_loss_bc": 0.4412243664264679, "train_loss_llm": 0.1500985324382782, "grad_norm": 0.22728338837623596, "global_step": 843, "epoch": 0, "lr": 9.999994846188819e-05} +{"train_loss": 0.4408054053783417, "train_loss_bc": 0.4396691620349884, "train_loss_llm": 0.11362537741661072, "grad_norm": 0.3101786673069, "global_step": 844, "epoch": 0, "lr": 9.999994846188819e-05} +{"train_loss": 0.6227639317512512, "train_loss_bc": 0.6216356158256531, "train_loss_llm": 0.11283424496650696, "grad_norm": 0.34569984674453735, "global_step": 845, "epoch": 0, "lr": 9.999994846188819e-05} +{"train_loss": 0.4020445942878723, "train_loss_bc": 0.40085673332214355, "train_loss_llm": 0.11878734827041626, "grad_norm": 0.5083690285682678, "global_step": 846, "epoch": 0, "lr": 9.999994846188819e-05} +{"train_loss": 0.4856792390346527, "train_loss_bc": 0.48400574922561646, "train_loss_llm": 0.16734924912452698, "grad_norm": 0.5528178215026855, "global_step": 847, "epoch": 0, "lr": 9.999994846188819e-05} +{"train_loss": 0.44932353496551514, "train_loss_bc": 0.4481476843357086, "train_loss_llm": 0.11758485436439514, "grad_norm": 0.48026207089424133, "global_step": 848, "epoch": 0, "lr": 9.999994738258546e-05} +{"train_loss": 0.28494200110435486, "train_loss_bc": 0.28382861614227295, "train_loss_llm": 0.1113387942314148, "grad_norm": 0.1087321788072586, "global_step": 849, "epoch": 0, "lr": 9.999994738258546e-05} +{"train_loss": 0.45408862829208374, "train_loss_bc": 0.4528406262397766, "train_loss_llm": 0.12480074167251587, "grad_norm": 0.22985370457172394, "global_step": 850, "epoch": 0, "lr": 9.999994738258546e-05} +{"train_loss": 0.3384667634963989, "train_loss_bc": 0.3373321294784546, "train_loss_llm": 0.11346229910850525, "grad_norm": 0.19277672469615936, "global_step": 851, "epoch": 0, "lr": 9.999994738258546e-05} +{"train_loss": 0.2939108610153198, "train_loss_bc": 0.29312995076179504, "train_loss_llm": 0.0780918300151825, "grad_norm": 0.1637968271970749, "global_step": 852, "epoch": 0, "lr": 9.999994738258546e-05} +{"train_loss": 0.34464749693870544, "train_loss_bc": 0.3436492681503296, "train_loss_llm": 0.09982168674468994, "grad_norm": 0.18302200734615326, "global_step": 853, "epoch": 0, "lr": 9.999994738258546e-05} +{"train_loss": 0.5617199540138245, "train_loss_bc": 0.5601301193237305, "train_loss_llm": 0.1589830219745636, "grad_norm": 0.32621505856513977, "global_step": 854, "epoch": 0, "lr": 9.999994738258546e-05} +{"train_loss": 0.5973924398422241, "train_loss_bc": 0.5957185626029968, "train_loss_llm": 0.16738846898078918, "grad_norm": 0.49640968441963196, "global_step": 855, "epoch": 0, "lr": 9.999994738258546e-05} +{"train_loss": 0.42385590076446533, "train_loss_bc": 0.4228513240814209, "train_loss_llm": 0.10045793652534485, "grad_norm": 0.5047431588172913, "global_step": 856, "epoch": 0, "lr": 9.999994629209827e-05} +{"train_loss": 0.5222750306129456, "train_loss_bc": 0.5207557678222656, "train_loss_llm": 0.15192639827728271, "grad_norm": 0.20339296758174896, "global_step": 857, "epoch": 0, "lr": 9.999994629209827e-05} +{"train_loss": 0.49918875098228455, "train_loss_bc": 0.4973152279853821, "train_loss_llm": 0.1873524785041809, "grad_norm": 0.37862664461135864, "global_step": 858, "epoch": 0, "lr": 9.999994629209827e-05} +{"train_loss": 0.4883464276790619, "train_loss_bc": 0.4872998595237732, "train_loss_llm": 0.10465636849403381, "grad_norm": 0.5141247510910034, "global_step": 859, "epoch": 0, "lr": 9.999994629209827e-05} +{"train_loss": 0.29056328535079956, "train_loss_bc": 0.289577454328537, "train_loss_llm": 0.09858295321464539, "grad_norm": 0.5172990560531616, "global_step": 860, "epoch": 0, "lr": 9.999994629209827e-05} +{"train_loss": 0.393641859292984, "train_loss_bc": 0.3925539255142212, "train_loss_llm": 0.10879379510879517, "grad_norm": 0.5052634477615356, "global_step": 861, "epoch": 0, "lr": 9.999994629209827e-05} +{"train_loss": 0.3258166015148163, "train_loss_bc": 0.3244849443435669, "train_loss_llm": 0.13316550850868225, "grad_norm": 0.5082921385765076, "global_step": 862, "epoch": 0, "lr": 9.999994629209827e-05} +{"train_loss": 0.4812297821044922, "train_loss_bc": 0.4797605872154236, "train_loss_llm": 0.1469208002090454, "grad_norm": 0.5514684319496155, "global_step": 863, "epoch": 0, "lr": 9.999994629209827e-05} +{"train_loss": 0.5455595850944519, "train_loss_bc": 0.5444753170013428, "train_loss_llm": 0.10842922329902649, "grad_norm": 0.6611815690994263, "global_step": 864, "epoch": 0, "lr": 9.99999451904266e-05} +{"train_loss": 0.3700634837150574, "train_loss_bc": 0.3684004843235016, "train_loss_llm": 0.166298508644104, "grad_norm": 0.1089361384510994, "global_step": 865, "epoch": 0, "lr": 9.99999451904266e-05} +{"train_loss": 0.44800522923469543, "train_loss_bc": 0.44628429412841797, "train_loss_llm": 0.17209327220916748, "grad_norm": 0.1541561335325241, "global_step": 866, "epoch": 0, "lr": 9.99999451904266e-05} +{"train_loss": 0.42328840494155884, "train_loss_bc": 0.4220215678215027, "train_loss_llm": 0.12668293714523315, "grad_norm": 0.20925258100032806, "global_step": 867, "epoch": 0, "lr": 9.99999451904266e-05} +{"train_loss": 0.5185708999633789, "train_loss_bc": 0.5171864032745361, "train_loss_llm": 0.13845232129096985, "grad_norm": 0.1881992369890213, "global_step": 868, "epoch": 0, "lr": 9.99999451904266e-05} +{"train_loss": 0.19765236973762512, "train_loss_bc": 0.19687986373901367, "train_loss_llm": 0.07725125551223755, "grad_norm": 0.2618674635887146, "global_step": 869, "epoch": 0, "lr": 9.99999451904266e-05} +{"train_loss": 0.5704347491264343, "train_loss_bc": 0.5689895153045654, "train_loss_llm": 0.1445261538028717, "grad_norm": 0.36627131700515747, "global_step": 870, "epoch": 0, "lr": 9.99999451904266e-05} +{"train_loss": 0.3305930495262146, "train_loss_bc": 0.3294355869293213, "train_loss_llm": 0.11574667692184448, "grad_norm": 0.3550454378128052, "global_step": 871, "epoch": 0, "lr": 9.99999451904266e-05} +{"train_loss": 0.5014159679412842, "train_loss_bc": 0.4995027184486389, "train_loss_llm": 0.1913270354270935, "grad_norm": 0.47749871015548706, "global_step": 872, "epoch": 0, "lr": 9.999994407757045e-05} +{"train_loss": 0.4831414818763733, "train_loss_bc": 0.481984406709671, "train_loss_llm": 0.11570602655410767, "grad_norm": 0.11637505888938904, "global_step": 873, "epoch": 0, "lr": 9.999994407757045e-05} +{"train_loss": 0.3962070047855377, "train_loss_bc": 0.3948107659816742, "train_loss_llm": 0.13962528109550476, "grad_norm": 0.17879343032836914, "global_step": 874, "epoch": 0, "lr": 9.999994407757045e-05} +{"train_loss": 0.4549310803413391, "train_loss_bc": 0.4538165330886841, "train_loss_llm": 0.11145418882369995, "grad_norm": 0.1287202686071396, "global_step": 875, "epoch": 0, "lr": 9.999994407757045e-05} +{"train_loss": 0.342518150806427, "train_loss_bc": 0.34123674035072327, "train_loss_llm": 0.12814116477966309, "grad_norm": 0.16375166177749634, "global_step": 876, "epoch": 0, "lr": 9.999994407757045e-05} +{"train_loss": 0.513576328754425, "train_loss_bc": 0.5124807357788086, "train_loss_llm": 0.10955944657325745, "grad_norm": 0.143615260720253, "global_step": 877, "epoch": 0, "lr": 9.999994407757045e-05} +{"train_loss": 0.3754693269729614, "train_loss_bc": 0.3743240535259247, "train_loss_llm": 0.11452609300613403, "grad_norm": 0.2990853488445282, "global_step": 878, "epoch": 0, "lr": 9.999994407757045e-05} +{"train_loss": 0.3489813506603241, "train_loss_bc": 0.3479025363922119, "train_loss_llm": 0.10788211226463318, "grad_norm": 0.24397443234920502, "global_step": 879, "epoch": 0, "lr": 9.999994407757045e-05} +{"train_loss": 0.5130777955055237, "train_loss_bc": 0.5116080045700073, "train_loss_llm": 0.14698147773742676, "grad_norm": 0.24227982759475708, "global_step": 880, "epoch": 0, "lr": 9.999994295352984e-05} +{"train_loss": 0.3216426372528076, "train_loss_bc": 0.3204021751880646, "train_loss_llm": 0.1240459680557251, "grad_norm": 0.04466254636645317, "global_step": 881, "epoch": 0, "lr": 9.999994295352984e-05} +{"train_loss": 0.3189554214477539, "train_loss_bc": 0.31763333082199097, "train_loss_llm": 0.13220840692520142, "grad_norm": 0.15020985901355743, "global_step": 882, "epoch": 0, "lr": 9.999994295352984e-05} +{"train_loss": 0.46145445108413696, "train_loss_bc": 0.45978182554244995, "train_loss_llm": 0.16726267337799072, "grad_norm": 0.2338121235370636, "global_step": 883, "epoch": 0, "lr": 9.999994295352984e-05} +{"train_loss": 0.46099480986595154, "train_loss_bc": 0.4593226909637451, "train_loss_llm": 0.16721290349960327, "grad_norm": 0.33087408542633057, "global_step": 884, "epoch": 0, "lr": 9.999994295352984e-05} +{"train_loss": 0.43931055068969727, "train_loss_bc": 0.43819791078567505, "train_loss_llm": 0.11126327514648438, "grad_norm": 0.2931665778160095, "global_step": 885, "epoch": 0, "lr": 9.999994295352984e-05} +{"train_loss": 0.43387913703918457, "train_loss_bc": 0.4326079487800598, "train_loss_llm": 0.12711742520332336, "grad_norm": 0.2820306122303009, "global_step": 886, "epoch": 0, "lr": 9.999994295352984e-05} +{"train_loss": 0.5288122892379761, "train_loss_bc": 0.5273622274398804, "train_loss_llm": 0.14500758051872253, "grad_norm": 0.20907562971115112, "global_step": 887, "epoch": 0, "lr": 9.999994295352984e-05} +{"train_loss": 0.4765452444553375, "train_loss_bc": 0.4749133288860321, "train_loss_llm": 0.16319113969802856, "grad_norm": 0.23047305643558502, "global_step": 888, "epoch": 0, "lr": 9.999994181830474e-05} +{"train_loss": 0.4809669256210327, "train_loss_bc": 0.47949111461639404, "train_loss_llm": 0.1475808322429657, "grad_norm": 0.1613912433385849, "global_step": 889, "epoch": 0, "lr": 9.999994181830474e-05} +{"train_loss": 0.41568759083747864, "train_loss_bc": 0.4145168662071228, "train_loss_llm": 0.11707177758216858, "grad_norm": 0.2331201732158661, "global_step": 890, "epoch": 0, "lr": 9.999994181830474e-05} +{"train_loss": 0.34334221482276917, "train_loss_bc": 0.3422807455062866, "train_loss_llm": 0.1061461865901947, "grad_norm": 0.30210810899734497, "global_step": 891, "epoch": 0, "lr": 9.999994181830474e-05} +{"train_loss": 0.3753444254398346, "train_loss_bc": 0.3736709654331207, "train_loss_llm": 0.16734597086906433, "grad_norm": 0.24136368930339813, "global_step": 892, "epoch": 0, "lr": 9.999994181830474e-05} +{"train_loss": 0.45010149478912354, "train_loss_bc": 0.4486684203147888, "train_loss_llm": 0.14330750703811646, "grad_norm": 0.271028995513916, "global_step": 893, "epoch": 0, "lr": 9.999994181830474e-05} +{"train_loss": 0.38253799080848694, "train_loss_bc": 0.3817411959171295, "train_loss_llm": 0.07968047261238098, "grad_norm": 0.36184075474739075, "global_step": 894, "epoch": 0, "lr": 9.999994181830474e-05} +{"train_loss": 0.3775109052658081, "train_loss_bc": 0.3765418529510498, "train_loss_llm": 0.0969039797782898, "grad_norm": 0.48663678765296936, "global_step": 895, "epoch": 0, "lr": 9.999994181830474e-05} +{"train_loss": 0.44369497895240784, "train_loss_bc": 0.4421958923339844, "train_loss_llm": 0.1499100923538208, "grad_norm": 0.44652509689331055, "global_step": 896, "epoch": 0, "lr": 9.999994067189517e-05} +{"train_loss": 0.32948046922683716, "train_loss_bc": 0.32839131355285645, "train_loss_llm": 0.10891419649124146, "grad_norm": 0.20219655334949493, "global_step": 897, "epoch": 0, "lr": 9.999994067189517e-05} +{"train_loss": 0.42447617650032043, "train_loss_bc": 0.4229239225387573, "train_loss_llm": 0.15522640943527222, "grad_norm": 0.2949985861778259, "global_step": 898, "epoch": 0, "lr": 9.999994067189517e-05} +{"train_loss": 0.4458390176296234, "train_loss_bc": 0.44426536560058594, "train_loss_llm": 0.15736564993858337, "grad_norm": 0.27578264474868774, "global_step": 899, "epoch": 0, "lr": 9.999994067189517e-05} +{"train_loss": 0.37028768658638, "train_loss_bc": 0.3694707155227661, "train_loss_llm": 0.08169576525688171, "grad_norm": 0.313737154006958, "global_step": 900, "epoch": 0, "lr": 9.999994067189517e-05} +{"train_loss": 0.4045461416244507, "train_loss_bc": 0.40314868092536926, "train_loss_llm": 0.139745831489563, "grad_norm": 0.41730961203575134, "global_step": 901, "epoch": 0, "lr": 9.999994067189517e-05} +{"train_loss": 0.36359426379203796, "train_loss_bc": 0.36219683289527893, "train_loss_llm": 0.1397443413734436, "grad_norm": 0.4372030794620514, "global_step": 902, "epoch": 0, "lr": 9.999994067189517e-05} +{"train_loss": 0.4287393391132355, "train_loss_bc": 0.4275266230106354, "train_loss_llm": 0.12127023935317993, "grad_norm": 0.38986799120903015, "global_step": 903, "epoch": 0, "lr": 9.999994067189517e-05} +{"train_loss": 0.47504499554634094, "train_loss_bc": 0.4741963744163513, "train_loss_llm": 0.08486181497573853, "grad_norm": 0.4350925385951996, "global_step": 904, "epoch": 0, "lr": 9.999993951430112e-05} +{"train_loss": 0.38130590319633484, "train_loss_bc": 0.38024812936782837, "train_loss_llm": 0.10577821731567383, "grad_norm": 0.14353474974632263, "global_step": 905, "epoch": 0, "lr": 9.999993951430112e-05} +{"train_loss": 0.3631591796875, "train_loss_bc": 0.3618849515914917, "train_loss_llm": 0.12742313742637634, "grad_norm": 0.2719558775424957, "global_step": 906, "epoch": 0, "lr": 9.999993951430112e-05} +{"train_loss": 0.4201207756996155, "train_loss_bc": 0.41859468817710876, "train_loss_llm": 0.1526075303554535, "grad_norm": 0.45465901494026184, "global_step": 907, "epoch": 0, "lr": 9.999993951430112e-05} +{"train_loss": 0.6030901670455933, "train_loss_bc": 0.6015290021896362, "train_loss_llm": 0.15611857175827026, "grad_norm": 0.3896191120147705, "global_step": 908, "epoch": 0, "lr": 9.999993951430112e-05} +{"train_loss": 0.5976340770721436, "train_loss_bc": 0.5963919162750244, "train_loss_llm": 0.1242172122001648, "grad_norm": 0.5227014422416687, "global_step": 909, "epoch": 0, "lr": 9.999993951430112e-05} +{"train_loss": 0.48432812094688416, "train_loss_bc": 0.4827817678451538, "train_loss_llm": 0.1546359658241272, "grad_norm": 0.5403639078140259, "global_step": 910, "epoch": 0, "lr": 9.999993951430112e-05} +{"train_loss": 0.3975037634372711, "train_loss_bc": 0.39637523889541626, "train_loss_llm": 0.11285299062728882, "grad_norm": 0.4841592609882355, "global_step": 911, "epoch": 0, "lr": 9.999993951430112e-05} +{"train_loss": 0.3206377327442169, "train_loss_bc": 0.3196225166320801, "train_loss_llm": 0.1015220582485199, "grad_norm": 0.5449240803718567, "global_step": 912, "epoch": 0, "lr": 9.999993834552261e-05} +{"train_loss": 0.4160666763782501, "train_loss_bc": 0.4149484634399414, "train_loss_llm": 0.11182042956352234, "grad_norm": 0.061857324093580246, "global_step": 913, "epoch": 0, "lr": 9.999993834552261e-05} +{"train_loss": 0.47550028562545776, "train_loss_bc": 0.473949134349823, "train_loss_llm": 0.15511548519134521, "grad_norm": 0.17099443078041077, "global_step": 914, "epoch": 0, "lr": 9.999993834552261e-05} +{"train_loss": 0.5402200818061829, "train_loss_bc": 0.5386112332344055, "train_loss_llm": 0.16088640689849854, "grad_norm": 0.32569122314453125, "global_step": 915, "epoch": 0, "lr": 9.999993834552261e-05} +{"train_loss": 0.4395882785320282, "train_loss_bc": 0.43817126750946045, "train_loss_llm": 0.1417020857334137, "grad_norm": 0.28916698694229126, "global_step": 916, "epoch": 0, "lr": 9.999993834552261e-05} +{"train_loss": 0.49348247051239014, "train_loss_bc": 0.4918447434902191, "train_loss_llm": 0.16377419233322144, "grad_norm": 0.2031954675912857, "global_step": 917, "epoch": 0, "lr": 9.999993834552261e-05} +{"train_loss": 0.4133864939212799, "train_loss_bc": 0.41203778982162476, "train_loss_llm": 0.13486900925636292, "grad_norm": 0.2717237174510956, "global_step": 918, "epoch": 0, "lr": 9.999993834552261e-05} +{"train_loss": 0.44576001167297363, "train_loss_bc": 0.44451865553855896, "train_loss_llm": 0.12413465976715088, "grad_norm": 0.3041559159755707, "global_step": 919, "epoch": 0, "lr": 9.999993834552261e-05} +{"train_loss": 0.3562610447406769, "train_loss_bc": 0.35521578788757324, "train_loss_llm": 0.10452505946159363, "grad_norm": 0.27864062786102295, "global_step": 920, "epoch": 0, "lr": 9.999993716555961e-05} +{"train_loss": 0.5913837552070618, "train_loss_bc": 0.589896023273468, "train_loss_llm": 0.1487726867198944, "grad_norm": 0.16369028389453888, "global_step": 921, "epoch": 0, "lr": 9.999993716555961e-05} +{"train_loss": 0.45828431844711304, "train_loss_bc": 0.45654210448265076, "train_loss_llm": 0.17422088980674744, "grad_norm": 0.30248767137527466, "global_step": 922, "epoch": 0, "lr": 9.999993716555961e-05} +{"train_loss": 0.34721118211746216, "train_loss_bc": 0.3463204503059387, "train_loss_llm": 0.08907273411750793, "grad_norm": 0.23506149649620056, "global_step": 923, "epoch": 0, "lr": 9.999993716555961e-05} +{"train_loss": 0.3813880980014801, "train_loss_bc": 0.380020409822464, "train_loss_llm": 0.13676825165748596, "grad_norm": 0.440594345331192, "global_step": 924, "epoch": 0, "lr": 9.999993716555961e-05} +{"train_loss": 0.494778573513031, "train_loss_bc": 0.4933704137802124, "train_loss_llm": 0.14081737399101257, "grad_norm": 0.48429983854293823, "global_step": 925, "epoch": 0, "lr": 9.999993716555961e-05} +{"train_loss": 0.41278064250946045, "train_loss_bc": 0.41171562671661377, "train_loss_llm": 0.10650098323822021, "grad_norm": 0.5706457495689392, "global_step": 926, "epoch": 0, "lr": 9.999993716555961e-05} +{"train_loss": 0.3818730115890503, "train_loss_bc": 0.3805137872695923, "train_loss_llm": 0.13592228293418884, "grad_norm": 0.6689505577087402, "global_step": 927, "epoch": 0, "lr": 9.999993716555961e-05} +{"train_loss": 0.29504355788230896, "train_loss_bc": 0.29377466440200806, "train_loss_llm": 0.12688937783241272, "grad_norm": 0.6089324355125427, "global_step": 928, "epoch": 0, "lr": 9.999993597441214e-05} +{"train_loss": 0.5230653882026672, "train_loss_bc": 0.521541953086853, "train_loss_llm": 0.15234127640724182, "grad_norm": 0.12326289713382721, "global_step": 929, "epoch": 0, "lr": 9.999993597441214e-05} +{"train_loss": 0.4340781271457672, "train_loss_bc": 0.4331514835357666, "train_loss_llm": 0.09266352653503418, "grad_norm": 0.1716209352016449, "global_step": 930, "epoch": 0, "lr": 9.999993597441214e-05} +{"train_loss": 0.3551509976387024, "train_loss_bc": 0.3538585305213928, "train_loss_llm": 0.12924587726593018, "grad_norm": 0.1996259242296219, "global_step": 931, "epoch": 0, "lr": 9.999993597441214e-05} +{"train_loss": 0.3795628249645233, "train_loss_bc": 0.37850016355514526, "train_loss_llm": 0.10626718401908875, "grad_norm": 0.2124268263578415, "global_step": 932, "epoch": 0, "lr": 9.999993597441214e-05} +{"train_loss": 0.4625495970249176, "train_loss_bc": 0.460946649312973, "train_loss_llm": 0.1602936089038849, "grad_norm": 0.2961867153644562, "global_step": 933, "epoch": 0, "lr": 9.999993597441214e-05} +{"train_loss": 0.2359141707420349, "train_loss_bc": 0.23467940092086792, "train_loss_llm": 0.12347662448883057, "grad_norm": 0.3200758099555969, "global_step": 934, "epoch": 0, "lr": 9.999993597441214e-05} +{"train_loss": 0.34700679779052734, "train_loss_bc": 0.3456500768661499, "train_loss_llm": 0.1356707215309143, "grad_norm": 0.42867714166641235, "global_step": 935, "epoch": 0, "lr": 9.999993597441214e-05} +{"train_loss": 0.7701971530914307, "train_loss_bc": 0.7685655355453491, "train_loss_llm": 0.16315969824790955, "grad_norm": 0.30987536907196045, "global_step": 936, "epoch": 0, "lr": 9.99999347720802e-05} +{"train_loss": 0.3566986620426178, "train_loss_bc": 0.35551321506500244, "train_loss_llm": 0.11854413151741028, "grad_norm": 0.15181134641170502, "global_step": 937, "epoch": 0, "lr": 9.99999347720802e-05} +{"train_loss": 0.3753644824028015, "train_loss_bc": 0.3743097484111786, "train_loss_llm": 0.10547453165054321, "grad_norm": 0.2639258801937103, "global_step": 938, "epoch": 0, "lr": 9.99999347720802e-05} +{"train_loss": 0.47676321864128113, "train_loss_bc": 0.47507601976394653, "train_loss_llm": 0.16872075200080872, "grad_norm": 0.3367781639099121, "global_step": 939, "epoch": 0, "lr": 9.99999347720802e-05} +{"train_loss": 0.4664318859577179, "train_loss_bc": 0.464763879776001, "train_loss_llm": 0.1668003499507904, "grad_norm": 0.35273051261901855, "global_step": 940, "epoch": 0, "lr": 9.99999347720802e-05} +{"train_loss": 0.3885119557380676, "train_loss_bc": 0.38687068223953247, "train_loss_llm": 0.1641261875629425, "grad_norm": 0.36588603258132935, "global_step": 941, "epoch": 0, "lr": 9.99999347720802e-05} +{"train_loss": 0.5820093750953674, "train_loss_bc": 0.5804073810577393, "train_loss_llm": 0.1601971983909607, "grad_norm": 0.4582238793373108, "global_step": 942, "epoch": 0, "lr": 9.99999347720802e-05} +{"train_loss": 0.4710002541542053, "train_loss_bc": 0.4697147309780121, "train_loss_llm": 0.12855172157287598, "grad_norm": 0.4619675874710083, "global_step": 943, "epoch": 0, "lr": 9.99999347720802e-05} +{"train_loss": 0.508743405342102, "train_loss_bc": 0.5070276260375977, "train_loss_llm": 0.1715766191482544, "grad_norm": 0.4349057972431183, "global_step": 944, "epoch": 0, "lr": 9.999993355856379e-05} +{"train_loss": 0.4097251892089844, "train_loss_bc": 0.408425509929657, "train_loss_llm": 0.12996813654899597, "grad_norm": 0.21188735961914062, "global_step": 945, "epoch": 0, "lr": 9.999993355856379e-05} +{"train_loss": 0.36142587661743164, "train_loss_bc": 0.3603687286376953, "train_loss_llm": 0.1057155430316925, "grad_norm": 0.1576070338487625, "global_step": 946, "epoch": 0, "lr": 9.999993355856379e-05} +{"train_loss": 0.46797996759414673, "train_loss_bc": 0.4663555324077606, "train_loss_llm": 0.16244223713874817, "grad_norm": 0.17794547975063324, "global_step": 947, "epoch": 0, "lr": 9.999993355856379e-05} +{"train_loss": 0.3987632989883423, "train_loss_bc": 0.3974025249481201, "train_loss_llm": 0.1360776722431183, "grad_norm": 0.24680264294147491, "global_step": 948, "epoch": 0, "lr": 9.999993355856379e-05} +{"train_loss": 0.480319082736969, "train_loss_bc": 0.4789217710494995, "train_loss_llm": 0.13972973823547363, "grad_norm": 0.19733159244060516, "global_step": 949, "epoch": 0, "lr": 9.999993355856379e-05} +{"train_loss": 0.4735085964202881, "train_loss_bc": 0.4714745879173279, "train_loss_llm": 0.20340001583099365, "grad_norm": 0.25462454557418823, "global_step": 950, "epoch": 0, "lr": 9.999993355856379e-05} +{"train_loss": 0.4643380045890808, "train_loss_bc": 0.46281957626342773, "train_loss_llm": 0.15184339880943298, "grad_norm": 0.2597101926803589, "global_step": 951, "epoch": 0, "lr": 9.999993355856379e-05} +{"train_loss": 0.3206809163093567, "train_loss_bc": 0.3196219801902771, "train_loss_llm": 0.1058928370475769, "grad_norm": 0.28536492586135864, "global_step": 952, "epoch": 0, "lr": 9.99999323338629e-05} +{"train_loss": 0.49491363763809204, "train_loss_bc": 0.49333426356315613, "train_loss_llm": 0.15793663263320923, "grad_norm": 0.12770074605941772, "global_step": 953, "epoch": 0, "lr": 9.99999323338629e-05} +{"train_loss": 0.45401108264923096, "train_loss_bc": 0.4527237117290497, "train_loss_llm": 0.1287357211112976, "grad_norm": 0.15090590715408325, "global_step": 954, "epoch": 0, "lr": 9.99999323338629e-05} +{"train_loss": 0.4615676999092102, "train_loss_bc": 0.4603375196456909, "train_loss_llm": 0.12301769852638245, "grad_norm": 0.2043774425983429, "global_step": 955, "epoch": 0, "lr": 9.99999323338629e-05} +{"train_loss": 0.4413764178752899, "train_loss_bc": 0.43981802463531494, "train_loss_llm": 0.15583953261375427, "grad_norm": 0.2845759689807892, "global_step": 956, "epoch": 0, "lr": 9.99999323338629e-05} +{"train_loss": 0.48241737484931946, "train_loss_bc": 0.4807056784629822, "train_loss_llm": 0.17116934061050415, "grad_norm": 0.4689476788043976, "global_step": 957, "epoch": 0, "lr": 9.99999323338629e-05} +{"train_loss": 0.2088479995727539, "train_loss_bc": 0.2078133225440979, "train_loss_llm": 0.10346764326095581, "grad_norm": 0.4865969717502594, "global_step": 958, "epoch": 0, "lr": 9.99999323338629e-05} +{"train_loss": 0.3820194900035858, "train_loss_bc": 0.3810354173183441, "train_loss_llm": 0.09840837121009827, "grad_norm": 0.4195979833602905, "global_step": 959, "epoch": 0, "lr": 9.99999323338629e-05} +{"train_loss": 0.32970884442329407, "train_loss_bc": 0.32901647686958313, "train_loss_llm": 0.06923675537109375, "grad_norm": 0.4474729001522064, "global_step": 960, "epoch": 0, "lr": 9.999993109797754e-05} +{"train_loss": 0.45453983545303345, "train_loss_bc": 0.4531002640724182, "train_loss_llm": 0.14395850896835327, "grad_norm": 0.15855643153190613, "global_step": 961, "epoch": 0, "lr": 9.999993109797754e-05} +{"train_loss": 0.3913039267063141, "train_loss_bc": 0.3900826573371887, "train_loss_llm": 0.12212702631950378, "grad_norm": 0.27189525961875916, "global_step": 962, "epoch": 0, "lr": 9.999993109797754e-05} +{"train_loss": 0.46142885088920593, "train_loss_bc": 0.4599838852882385, "train_loss_llm": 0.14449554681777954, "grad_norm": 0.3128124475479126, "global_step": 963, "epoch": 0, "lr": 9.999993109797754e-05} +{"train_loss": 0.4210406541824341, "train_loss_bc": 0.41966408491134644, "train_loss_llm": 0.1376570463180542, "grad_norm": 0.3017627000808716, "global_step": 964, "epoch": 0, "lr": 9.999993109797754e-05} +{"train_loss": 0.5593462586402893, "train_loss_bc": 0.5578506588935852, "train_loss_llm": 0.14956098794937134, "grad_norm": 0.3158896565437317, "global_step": 965, "epoch": 0, "lr": 9.999993109797754e-05} +{"train_loss": 0.32064300775527954, "train_loss_bc": 0.31926125288009644, "train_loss_llm": 0.13817477226257324, "grad_norm": 0.33661696314811707, "global_step": 966, "epoch": 0, "lr": 9.999993109797754e-05} +{"train_loss": 0.3716479539871216, "train_loss_bc": 0.3704668879508972, "train_loss_llm": 0.11810600757598877, "grad_norm": 0.34416380524635315, "global_step": 967, "epoch": 0, "lr": 9.999993109797754e-05} +{"train_loss": 0.4636586606502533, "train_loss_bc": 0.4627121686935425, "train_loss_llm": 0.09464913606643677, "grad_norm": 0.3615496754646301, "global_step": 968, "epoch": 0, "lr": 9.99999298509077e-05} +{"train_loss": 0.45983949303627014, "train_loss_bc": 0.45801088213920593, "train_loss_llm": 0.1828618049621582, "grad_norm": 0.20059309899806976, "global_step": 969, "epoch": 0, "lr": 9.99999298509077e-05} +{"train_loss": 0.47500577569007874, "train_loss_bc": 0.4741523265838623, "train_loss_llm": 0.08534479141235352, "grad_norm": 0.16255320608615875, "global_step": 970, "epoch": 0, "lr": 9.99999298509077e-05} +{"train_loss": 0.5007393956184387, "train_loss_bc": 0.4995819926261902, "train_loss_llm": 0.11574050784111023, "grad_norm": 0.22098475694656372, "global_step": 971, "epoch": 0, "lr": 9.99999298509077e-05} +{"train_loss": 0.31725624203681946, "train_loss_bc": 0.3160168528556824, "train_loss_llm": 0.12393766641616821, "grad_norm": 0.3111163377761841, "global_step": 972, "epoch": 0, "lr": 9.99999298509077e-05} +{"train_loss": 0.4328397512435913, "train_loss_bc": 0.4318913221359253, "train_loss_llm": 0.09484201669692993, "grad_norm": 0.17413409054279327, "global_step": 973, "epoch": 0, "lr": 9.99999298509077e-05} +{"train_loss": 0.38284486532211304, "train_loss_bc": 0.3814420998096466, "train_loss_llm": 0.1402776539325714, "grad_norm": 0.27953779697418213, "global_step": 974, "epoch": 0, "lr": 9.99999298509077e-05} +{"train_loss": 0.32041943073272705, "train_loss_bc": 0.3187214434146881, "train_loss_llm": 0.1697985827922821, "grad_norm": 0.3178544044494629, "global_step": 975, "epoch": 0, "lr": 9.99999298509077e-05} +{"train_loss": 0.49346300959587097, "train_loss_bc": 0.492059588432312, "train_loss_llm": 0.14034298062324524, "grad_norm": 0.3612702190876007, "global_step": 976, "epoch": 0, "lr": 9.999992859265339e-05} +{"train_loss": 0.4420956075191498, "train_loss_bc": 0.44101983308792114, "train_loss_llm": 0.10757714509963989, "grad_norm": 0.18905779719352722, "global_step": 977, "epoch": 0, "lr": 9.999992859265339e-05} +{"train_loss": 0.3752184212207794, "train_loss_bc": 0.3740226626396179, "train_loss_llm": 0.11957621574401855, "grad_norm": 0.28007572889328003, "global_step": 978, "epoch": 0, "lr": 9.999992859265339e-05} +{"train_loss": 0.5372825264930725, "train_loss_bc": 0.5356180667877197, "train_loss_llm": 0.166448175907135, "grad_norm": 0.23705901205539703, "global_step": 979, "epoch": 0, "lr": 9.999992859265339e-05} +{"train_loss": 0.4472073018550873, "train_loss_bc": 0.4458105266094208, "train_loss_llm": 0.1396775245666504, "grad_norm": 0.4167845547199249, "global_step": 980, "epoch": 0, "lr": 9.999992859265339e-05} +{"train_loss": 0.322329044342041, "train_loss_bc": 0.32105445861816406, "train_loss_llm": 0.12745898962020874, "grad_norm": 0.45782527327537537, "global_step": 981, "epoch": 0, "lr": 9.999992859265339e-05} +{"train_loss": 0.5622375011444092, "train_loss_bc": 0.5605005025863647, "train_loss_llm": 0.1737024486064911, "grad_norm": 0.5945319533348083, "global_step": 982, "epoch": 0, "lr": 9.999992859265339e-05} +{"train_loss": 0.4104715883731842, "train_loss_bc": 0.4093075096607208, "train_loss_llm": 0.11640921235084534, "grad_norm": 0.7822667956352234, "global_step": 983, "epoch": 0, "lr": 9.999992859265339e-05} +{"train_loss": 0.45150408148765564, "train_loss_bc": 0.45012420415878296, "train_loss_llm": 0.13798868656158447, "grad_norm": 0.7692086100578308, "global_step": 984, "epoch": 0, "lr": 9.999992732321461e-05} +{"train_loss": 0.35420626401901245, "train_loss_bc": 0.3527546226978302, "train_loss_llm": 0.1451631784439087, "grad_norm": 0.200129896402359, "global_step": 985, "epoch": 0, "lr": 9.999992732321461e-05} +{"train_loss": 0.42388004064559937, "train_loss_bc": 0.42240381240844727, "train_loss_llm": 0.1476237177848816, "grad_norm": 0.24076329171657562, "global_step": 986, "epoch": 0, "lr": 9.999992732321461e-05} +{"train_loss": 0.3590567111968994, "train_loss_bc": 0.358163058757782, "train_loss_llm": 0.08936557173728943, "grad_norm": 0.31469953060150146, "global_step": 987, "epoch": 0, "lr": 9.999992732321461e-05} +{"train_loss": 0.41901734471321106, "train_loss_bc": 0.4181640148162842, "train_loss_llm": 0.08533385396003723, "grad_norm": 0.4831927716732025, "global_step": 988, "epoch": 0, "lr": 9.999992732321461e-05} +{"train_loss": 0.41994404792785645, "train_loss_bc": 0.4182512164115906, "train_loss_llm": 0.16928249597549438, "grad_norm": 0.5612785220146179, "global_step": 989, "epoch": 0, "lr": 9.999992732321461e-05} +{"train_loss": 0.529492974281311, "train_loss_bc": 0.527848482131958, "train_loss_llm": 0.16445034742355347, "grad_norm": 0.7547512054443359, "global_step": 990, "epoch": 0, "lr": 9.999992732321461e-05} +{"train_loss": 0.41025105118751526, "train_loss_bc": 0.40857410430908203, "train_loss_llm": 0.1676936149597168, "grad_norm": 0.7393769025802612, "global_step": 991, "epoch": 0, "lr": 9.999992732321461e-05} +{"train_loss": 0.24451971054077148, "train_loss_bc": 0.2437412142753601, "train_loss_llm": 0.0778496265411377, "grad_norm": 0.7509053349494934, "global_step": 992, "epoch": 0, "lr": 9.999992604259137e-05} +{"train_loss": 0.4309055507183075, "train_loss_bc": 0.429854154586792, "train_loss_llm": 0.1051386296749115, "grad_norm": 0.13102659583091736, "global_step": 993, "epoch": 0, "lr": 9.999992604259137e-05} diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/train.log b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/train.log new file mode 100644 index 0000000000000000000000000000000000000000..8da10e224ee83c81463aca3713b741c54b2adc8a --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/train.log @@ -0,0 +1,13 @@ +[2026-04-10 09:36:50,164][numexpr.utils][INFO] - Note: detected 144 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable. +[2026-04-10 09:36:50,164][numexpr.utils][INFO] - Note: NumExpr detected 144 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 16. +[2026-04-10 09:36:50,164][numexpr.utils][INFO] - NumExpr defaulting to 16 threads. +[2026-04-10 09:36:53,596][datasets][INFO] - PyTorch version 2.2.2 available. +[2026-04-10 09:36:53,597][datasets][INFO] - TensorFlow version 2.15.1 available. +[2026-04-10 09:36:53,597][datasets][INFO] - JAX version 0.4.30 available. +[2026-04-10 09:36:57,574][root][INFO] - running build_ext +[2026-04-10 09:36:57,576][root][INFO] - building 'mujoco_py.cymj' extension +[2026-04-10 09:36:57,576][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o -fopenmp -w +[2026-04-10 09:37:39,277][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -fopenmp -w +[2026-04-10 09:37:39,368][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -shared -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib -L/venv/llm-bc/lib -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -L/root/.mujoco/mujoco210/bin -Wl,--enable-new-dtags,-rpath,/root/.mujoco/mujoco210/bin -lmujoco210 -lglewosmesa -lOSMesa -lGL -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/lib.linux-x86_64-cpython-39/mujoco_py/cymj.cpython-39-x86_64-linux-gnu.so -fopenmp +[2026-04-10 09:37:41,989][absl][INFO] - MUJOCO_GL=osmesa, attempting to import specified OpenGL backend. +[2026-04-10 09:37:41,996][absl][INFO] - MuJoCo library version is: 2.3.7 diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/config.yaml b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..7140baa6ab5c91c6e86204464a2978e14383dff0 --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/config.yaml @@ -0,0 +1,273 @@ +_target_: + value: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +_wandb: + value: + cli_version: 0.18.6 + m: [] + python_version: 3.9.25 + t: + "1": + - 1 + - 2 + - 3 + - 5 + - 11 + - 12 + - 41 + - 49 + - 50 + - 51 + - 53 + - 55 + - 71 + - 83 + - 95 + - 98 + - 100 + - 105 + "2": + - 1 + - 2 + - 3 + - 5 + - 11 + - 12 + - 41 + - 49 + - 50 + - 51 + - 53 + - 55 + - 71 + - 83 + - 95 + - 98 + - 100 + - 105 + "3": + - 13 + - 15 + - 16 + - 23 + - 55 + - 61 + "4": 3.9.25 + "5": 0.18.6 + "6": 4.47.1 + "8": + - 5 + "12": 0.18.6 + "13": linux-x86_64 +action_dim: + value: 2 +checkpoint: + value: + save_last_ckpt: true + save_last_snapshot: false + topk: + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + k: 5 + mode: max + monitor_key: test_success_rate +dataloader: + value: + batch_size: 16 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true +exp_name: + value: default +horizon: + value: 1 +llm: + value: + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + checkpoint: data/outputs/2025.10.16/11.53.22_train_llm_lowdim_maze2d-medium-v0/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-maze2d-medium-v0/checkpoint-11850 + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + finetune_mode: orig + hydra: + job: + override_dirname: HuggingFaceTB/SmolLM2-135M-Instruct + run: + dir: data/outputs/2026.04.10/09.36.48_HuggingFaceTB/SmolLM2-135M-Instruct + llm_mode: ete-finetuned + lora_config: + bias: none + lora_alpha: 64 + lora_dropout: 0.05 + r: 32 + task_type: CAUSAL_LM + max_length: 100 + model_name: SmolLM2-135M-Instruct + name: HuggingFaceTB/SmolLM2-135M-Instruct + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + use_joint_mlp_projector: true + use_quantization: false +llm_do_sample: + value: false +llm_orig_expert_feedback: + value: true +logging: + value: + group: null + id: null + mode: online + name: 2026.04.10-09.36.48_train_llmbc_lowdim_maze2d-medium-v0 + project: maze2d-medium-v0-training-mix + resume: true + tags: + - train_llmbc_lowdim + - maze2d-medium-v0 + - default +model_name: + value: HuggingFaceTB/SmolLM2-135M-Instruct +multi_run: + value: + run_dir: data/outputs/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0 + wandb_name_base: 2026.04.10-09.36.48_train_llmbc_lowdim_maze2d-medium-v0 +n_action_steps: + value: 1 +n_latency_steps: + value: 0 +n_obs_steps: + value: 1 +name: + value: train_llmbc_lowdim +obs_dim: + value: 6 +optimizer: + value: + _target_: torch.optim.AdamW + betas: + - 0.95 + - 0.999 + eps: 1e-08 + lr: 0.0001 + weight_decay: 1e-06 +output_dir: + value: /workspace/LLM-BC/data/outputs/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0 +past_action_visible: + value: false +policy: + value: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + action_dim: 2 + horizon: 1 + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + action_dim: 2 + cfg: + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + checkpoint: data/outputs/2025.10.16/11.53.22_train_llm_lowdim_maze2d-medium-v0/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-maze2d-medium-v0/checkpoint-11850 + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + finetune_mode: orig + hydra: + job: + override_dirname: HuggingFaceTB/SmolLM2-135M-Instruct + run: + dir: data/outputs/2026.04.10/09.36.48_HuggingFaceTB/SmolLM2-135M-Instruct + llm_mode: ete-finetuned + lora_config: + bias: none + lora_alpha: 64 + lora_dropout: 0.05 + r: 32 + task_type: CAUSAL_LM + max_length: 100 + model_name: SmolLM2-135M-Instruct + name: HuggingFaceTB/SmolLM2-135M-Instruct + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + use_joint_mlp_projector: true + use_quantization: false + horizon: 1 + n_action_steps: 1 + n_obs_steps: 1 + obs_dim: 6 + task_id: maze2d-medium-v0 + loss_bc_weight: 1 + loss_llm_weight: 0.01 + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + activation: tanh + hidden_size: + - 256 + - 256 + - 256 + - 256 + input_size: 6 + n_action_steps: 1 + n_obs_steps: 1 + output_size: 2 + n_action_steps: 1 + n_obs_steps: 1 + normalize_llm_loss: true + obs_dim: 6 +task: + value: + action_dim: 2 + dataset: + _target_: llmbc.dataset.pointmaze_lowdim_dataset.PointmazeLowdimDataset + data_path: datasets/maze2d-medium-v0-mix(csp5).pt + data_path2: datasets/maze2d-medium-v0.pt + dummy_normalizer: true + horizon: 1 + obs_eef_target: true + pad_after: 0 + pad_before: 0 + use_manual_normalizer: false + val_ratio: 0.02 + env_runner: + _target_: llmbc.env_runner.pointmaze_lowdim_runner.PointmazeLowdimRunner + discount: 0.99 + env_name: llf-pointmaze-maze2d-medium-v0 + feedback_type: + - hp + - hn + - fp + instruction_type: b + max_steps: 250 + n_action_steps: 1 + n_envs: 10 + n_obs_steps: 1 + n_test: 50 + n_train: 10 + visual: false + instructor: + _target_: llmbc.translator.instructor.pointmaze_instructor.maze2d_medium_v0_instructor.Maze2dMediumV0Instructor + name: maze2d-medium-v0 + obs_dim: 6 +task_name: + value: maze2d-medium-v0 +training: + value: + checkpoint_every: 4 + debug: false + device: cuda:0 + grad_norm_clip: 0.5 + gradient_accumulate_every: 8 + lr_scheduler: cosine + lr_warmup_steps: 10 + max_train_steps: null + max_val_steps: null + num_epochs: 801 + resume: false + rollout_every: 4 + sample_every: 4 + sample_max_batch: 128 + seed: 42 + tqdm_interval_sec: 1 + val_every: 1 +val_dataloader: + value: + batch_size: 16 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/output.log b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/output.log new file mode 100644 index 0000000000000000000000000000000000000000..8d49c2d1162901ee02ccbcf8988d8108741524aa --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/output.log @@ -0,0 +1,52 @@ +Traceback (most recent call last): + File "/workspace/LLM-BC/./train.py", line 35, in + main() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/main.py", line 90, in decorated_main + _run_hydra( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 389, in _run_hydra + _run_app( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 452, in _run_app + run_and_report( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 213, in run_and_report + return func() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 453, in + lambda: hydra.run( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/hydra.py", line 119, in run + ret = run_job( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/core/utils.py", line 186, in run_job + ret.return_value = task_function(task_cfg) + File "/workspace/LLM-BC/./train.py", line 32, in main + workspace.run() + File "/workspace/LLM-BC/llmbc/workspace/train_llmbc_lowdim_workspace.py", line 186, in run + loss.backward() + File "/venv/llm-bc/lib/python3.9/site-packages/torch/_tensor.py", line 522, in backward + torch.autograd.backward( + File "/venv/llm-bc/lib/python3.9/site-packages/torch/autograd/__init__.py", line 266, in backward + Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass +KeyboardInterrupt +Traceback (most recent call last): + File "/workspace/LLM-BC/./train.py", line 35, in + main() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/main.py", line 90, in decorated_main + _run_hydra( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 389, in _run_hydra + _run_app( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 452, in _run_app + run_and_report( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 213, in run_and_report + return func() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 453, in + lambda: hydra.run( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/hydra.py", line 119, in run + ret = run_job( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/core/utils.py", line 186, in run_job + ret.return_value = task_function(task_cfg) + File "/workspace/LLM-BC/./train.py", line 32, in main + workspace.run() + File "/workspace/LLM-BC/llmbc/workspace/train_llmbc_lowdim_workspace.py", line 186, in run + loss.backward() + File "/venv/llm-bc/lib/python3.9/site-packages/torch/_tensor.py", line 522, in backward + torch.autograd.backward( + File "/venv/llm-bc/lib/python3.9/site-packages/torch/autograd/__init__.py", line 266, in backward + Variable._execution_engine.run_backward( # Calls into the C++ engine to run the backward pass +KeyboardInterrupt diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/requirements.txt b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/requirements.txt new file mode 100644 index 0000000000000000000000000000000000000000..7942780cb43aff2ea665c4babf0201d7769b0330 --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/requirements.txt @@ -0,0 +1,861 @@ +aiohappyeyeballs==2.6.1 +antlr4-python3-runtime==4.9.3 +asciitree==0.3.3 +async-timeout==5.0.1 +attrs==25.4.0 +certifi==2025.10.5 +charset-normalizer==3.4.4 +colorama==0.4.6 +dill==0.3.8 +docstring-parser==0.16 +eval_type_backport==0.2.2 +fasteners==0.20 +filelock==3.17.0 +frozenlist==1.8.0 +fsspec==2024.3.1 +gmpy2==2.2.1 +hf-xet==1.1.8 +idna==3.11 +llvmlite==0.39.1 +MarkupSafe==3.0.2 +mdurl==0.1.2 +mkl-service==2.4.0 +mpmath==1.3.0 +msgpack==1.1.1 +networkx==3.2.1 +packaging==25.0 +propcache==0.3.1 +psutil==7.0.0 +pycparser==2.23 +Pygments==2.19.1 +PyOpenGL==3.1.10 +PySocks==1.7.1 +tzdata==2025.3 +xxhash==3.5.0 +pytz==2025.2 +PyYAML==6.0.2 +regex==2025.9.1 +safetensors==0.5.3 +setuptools==80.9.0 +shtab==1.7.2 +six==1.17.0 +tqdm==4.67.1 +tensorboard==2.15.2 +uvloop==0.22.1 +wheel==0.45.1 +zipp==3.21.0 +tensorflow-estimator==2.15.0 +tensorboard-data-server==0.7.2 +google-auth-oauthlib==1.3.1 +tensorflow==2.15.1 +Jinja2==3.1.6 +markdown-it-py==2.2.0 +multiprocess==0.70.15 +numpy==1.23.5 +pillow==11.3.0 +pip==23.3.2 +python-dateutil==2.9.0.post0 +aiosignal==1.4.0 +brotlicffi==1.0.9.2 +hydra-core==1.2.0 +multidict==6.7.0 +pycares==4.10.0 +rich==14.2.0 +sympy==1.14.0 +typeguard==4.4.4 +aiodns==3.5.0 +urllib3==2.5.0 +yarl==1.18.0 +aiohttp==3.12.15 +pyarrow==21.0.0 +pyarrow-hotfix==0.7 +tokenizers==0.21.0 +av==10.0.0 +einops==0.4.1 +Bottleneck==1.4.2 +mkl_fft==1.3.11 +mkl_random==1.2.8 +numba==0.56.4 +numcodecs==0.12.1 +numexpr==2.10.1 +pandas==2.3.3 +zarr==2.12.0 +datasets==2.19.0 +transformers==4.47.1 +accelerate==1.0.1 +peft==0.14.0 +trl==0.11.4 +torch==2.2.2 +tabulate==0.9.0 +sentencepiece==0.2.0 +pyperclip==1.11.0 +PyOpenGL-accelerate==3.1.10 +pydub==0.25.1 +py-cpuinfo==9.0.0 +pure_eval==0.2.3 +ptyprocess==0.7.0 +nvidia-ml-py==13.595.45 +hjson==3.1.0 +gym-notices==0.1.0 +glfw==2.0.0 +Farama-Notifications==0.0.4 +websockets==15.0.1 +wcwidth==0.6.0 +uc-micro-py==1.0.3 +typing-inspection==0.4.2 +triton==2.2.0 +trimesh==4.11.5 +transforms3d==0.4.2 +traitlets==5.14.3 +threadpoolctl==3.6.0 +tenacity==9.1.2 +sniffio==1.3.1 +smmap==5.0.3 +setproctitle==1.3.7 +sentry-sdk==2.57.0 +semantic-version==2.10.0 +scipy==1.13.1 +rpds-py==0.27.1 +python-multipart==0.0.20 +pyparsing==3.3.2 +pynvml==13.0.1 +pydantic_core==2.41.5 +pycryptodomex==3.23.0 +pyasn1==0.6.3 +grpcio==1.80.0 +platformdirs==4.4.0 +pexpect==4.9.0 +parso==0.8.6 +orjson==3.11.5 +opencv-python==4.11.0.86 +stack-data==0.6.3 +omegaconf==2.2.1 +nvidia-nvtx-cu12==12.1.105 +nvidia-nvjitlink-cu12==12.9.86 +nvidia-nccl-cu12==2.19.3 +nvidia-curand-cu12==10.3.2.106 +nvidia-cufft-cu12==11.0.2.54 +nvidia-cuda-runtime-cu12==12.1.105 +nvidia-cuda-nvrtc-cu12==12.1.105 +nvidia-cuda-cupti-cu12==12.1.105 +nvidia-cublas-cu12==12.1.3.1 +ninja==1.13.0 +narwhals==2.19.0 +lxml==6.0.3 +kiwisolver==1.4.7 +joblib==1.5.3 +jiter==0.13.0 +imageio-ffmpeg==0.6.0 +ImageIO==2.37.2 +h5py==3.14.0 +h11==0.16.0 +fonttools==4.60.2 +ffmpy==1.0.0 +fast_kinematics==0.2.2 +executing==2.2.1 +exceptiongroup==1.3.1 +docker-pycreds==0.4.0 +distro==1.9.0 +decorator==5.2.1 +dacite==1.9.2 +cycler==0.12.1 +contourpy==1.3.0 +cloudpickle==3.1.2 +click==8.1.8 +asttokens==3.0.1 +annotated-types==0.7.0 +annotated-doc==0.0.4 +aiofiles==25.1.0 +absl-py==2.3.1 +uvicorn==0.39.0 +tiktoken==0.8.0 +scikit-learn==1.6.1 +sapien==3.0.0b1 +referencing==0.36.2 +pydantic==2.12.5 +pyasn1_modules==0.4.2 +prompt_toolkit==3.0.52 +nvidia-cusparse-cu12==12.1.0.106 +nvidia-cudnn-cu12==8.9.2.26 +nltk==3.9.2 +mujoco==2.3.7 +mdit-py-plugins==0.3.3 +matplotlib-inline==0.2.1 +matplotlib==3.7.5 +linkify-it-py==2.0.3 +jedi==0.19.2 +huggingface-hub==0.26.2 +httpcore==1.0.9 +gymnasium==0.29.1 +google-pasta==0.2.0 +gitdb==4.0.12 +cryptography==46.0.7 +blobfile==3.0.0 +anyio==4.12.1 +tyro==0.9.1 +toppra==0.6.6 +starlette==0.49.3 +seaborn==0.13.2 +rouge_score==0.1.2 +pynndescent==0.6.0 +nvidia-cusolver-cu12==11.4.5.107 +jsonschema-specifications==2025.9.1 +ipython==8.18.1 +httpx==0.28.1 +gymnasium-robotics==1.2.0 +google-auth==2.49.2 +GitPython==3.1.46 +diffusers==0.31.0 +wandb==0.18.6 +umap-learn==0.5.9.post2 +requests-oauthlib==2.0.0 +openai==2.8.1 +mplib==0.1.1 +jsonschema==4.25.1 +gradio_client==0.2.9 +google-genai==1.47.0 +fastapi==0.128.8 +torchvision==0.17.2 +torchaudio==2.2.2 +sentence-transformers==3.2.1 +pytorch-seed==0.2.0 +deepspeed==0.16.1 +bitsandbytes==0.45.0 +altair==6.0.0 +timm==1.0.22 +gradio==3.36.1 +evaluate==0.4.3 +arm_pytorch_utilities==0.5.0 +tensorflow-io-gcs-filesystem==0.37.1 +Markdown==3.9 +pybullet==3.2.6 +wrapt==1.14.2 +tensorflow-probability==0.23.0 +parse==1.19.1 +dm-tree==0.1.8 +termcolor==3.1.0 +requests==2.32.0 +cffi==2.0.0 +opt_einsum==3.4.0 +labmaze==1.0.6 +typing_extensions==4.15.0 +importlib-resources==5.13.0 +importlib-metadata==5.2.0 +FLAML==2.3.6 +dm-env==1.6 +diskcache==5.6.3 +Cython==0.29.37 +mujoco-py==2.1.2.14 +jaxlib==0.4.30 +dm-control==1.0.14 +cmudict==1.0.13 +syllables==1.0.9 +metaworld==2.0.0 +jax==0.4.30 +highway-env==1.9.1 +gym_bandits==0.0.2 +d4rl==1.1 +pyautogen==0.1.0 +protobuf==4.25.9 +tf-agents==0.19.0 +oauthlib==3.3.1 +pytorch-kinematics==0.7.5 +pygame==2.6.1 +gin-config==0.5.0 +mani_skill==3.0.0b20 +tifffile==2024.8.30 +Shapely==1.8.4 +PyWavelets==1.6.0 +Werkzeug==3.1.8 +astunparse==1.6.3 +scikit-image==0.19.3 +pymunk==6.2.1 +ml-dtypes==0.3.2 +gast==0.7.0 +stable-baselines3==2.2.1 +keras==2.15.0 +llfbench==0.1.0 +libclang==18.1.1 +flatbuffers==25.12.19 +gym==0.23.1 +llmbc==0.0.0 +aiohappyeyeballs==2.6.1 +antlr4-python3-runtime==4.9.3 +asciitree==0.3.3 +async-timeout==5.0.1 +attrs==25.4.0 +certifi==2025.10.5 +charset-normalizer==3.4.4 +colorama==0.4.6 +dill==0.3.8 +docstring-parser==0.16 +eval_type_backport==0.2.2 +fasteners==0.20 +filelock==3.17.0 +frozenlist==1.8.0 +fsspec==2024.3.1 +gmpy2==2.2.1 +hf-xet==1.1.8 +idna==3.11 +llvmlite==0.39.1 +MarkupSafe==3.0.2 +mdurl==0.1.2 +mkl-service==2.4.0 +mpmath==1.3.0 +msgpack==1.1.1 +networkx==3.2.1 +packaging==25.0 +propcache==0.3.1 +psutil==7.0.0 +pycparser==2.23 +Pygments==2.19.1 +PyOpenGL==3.1.10 +PySocks==1.7.1 +tzdata==2025.3 +xxhash==3.5.0 +pytz==2025.2 +PyYAML==6.0.2 +regex==2025.9.1 +safetensors==0.5.3 +setuptools==80.9.0 +shtab==1.7.2 +six==1.17.0 +tqdm==4.67.1 +tensorboard==2.15.2 +uvloop==0.22.1 +wheel==0.45.1 +zipp==3.21.0 +tensorflow-estimator==2.15.0 +tensorboard-data-server==0.7.2 +google-auth-oauthlib==1.3.1 +tensorflow==2.15.1 +Jinja2==3.1.6 +markdown-it-py==2.2.0 +multiprocess==0.70.15 +numpy==1.23.5 +pillow==11.3.0 +pip==23.3.2 +python-dateutil==2.9.0.post0 +aiosignal==1.4.0 +brotlicffi==1.0.9.2 +hydra-core==1.2.0 +multidict==6.7.0 +pycares==4.10.0 +rich==14.2.0 +sympy==1.14.0 +typeguard==4.4.4 +aiodns==3.5.0 +urllib3==2.5.0 +yarl==1.18.0 +aiohttp==3.12.15 +pyarrow==21.0.0 +pyarrow-hotfix==0.7 +tokenizers==0.21.0 +av==10.0.0 +einops==0.4.1 +Bottleneck==1.4.2 +mkl_fft==1.3.11 +mkl_random==1.2.8 +numba==0.56.4 +numcodecs==0.12.1 +numexpr==2.10.1 +pandas==2.3.3 +zarr==2.12.0 +datasets==2.19.0 +transformers==4.47.1 +accelerate==1.0.1 +peft==0.14.0 +trl==0.11.4 +torch==2.2.2 +tabulate==0.9.0 +sentencepiece==0.2.0 +pyperclip==1.11.0 +PyOpenGL-accelerate==3.1.10 +pydub==0.25.1 +py-cpuinfo==9.0.0 +pure_eval==0.2.3 +ptyprocess==0.7.0 +nvidia-ml-py==13.595.45 +hjson==3.1.0 +gym-notices==0.1.0 +glfw==2.0.0 +Farama-Notifications==0.0.4 +websockets==15.0.1 +wcwidth==0.6.0 +uc-micro-py==1.0.3 +typing-inspection==0.4.2 +triton==2.2.0 +trimesh==4.11.5 +transforms3d==0.4.2 +traitlets==5.14.3 +threadpoolctl==3.6.0 +tenacity==9.1.2 +sniffio==1.3.1 +smmap==5.0.3 +setproctitle==1.3.7 +sentry-sdk==2.57.0 +semantic-version==2.10.0 +scipy==1.13.1 +rpds-py==0.27.1 +python-multipart==0.0.20 +pyparsing==3.3.2 +pynvml==13.0.1 +pydantic_core==2.41.5 +pycryptodomex==3.23.0 +pyasn1==0.6.3 +grpcio==1.80.0 +platformdirs==4.4.0 +pexpect==4.9.0 +parso==0.8.6 +orjson==3.11.5 +opencv-python==4.11.0.86 +stack-data==0.6.3 +omegaconf==2.2.1 +nvidia-nvtx-cu12==12.1.105 +nvidia-nvjitlink-cu12==12.9.86 +nvidia-nccl-cu12==2.19.3 +nvidia-curand-cu12==10.3.2.106 +nvidia-cufft-cu12==11.0.2.54 +nvidia-cuda-runtime-cu12==12.1.105 +nvidia-cuda-nvrtc-cu12==12.1.105 +nvidia-cuda-cupti-cu12==12.1.105 +nvidia-cublas-cu12==12.1.3.1 +ninja==1.13.0 +narwhals==2.19.0 +lxml==6.0.3 +kiwisolver==1.4.7 +joblib==1.5.3 +jiter==0.13.0 +imageio-ffmpeg==0.6.0 +ImageIO==2.37.2 +h5py==3.14.0 +h11==0.16.0 +fonttools==4.60.2 +ffmpy==1.0.0 +fast_kinematics==0.2.2 +executing==2.2.1 +exceptiongroup==1.3.1 +docker-pycreds==0.4.0 +distro==1.9.0 +decorator==5.2.1 +dacite==1.9.2 +cycler==0.12.1 +contourpy==1.3.0 +cloudpickle==3.1.2 +click==8.1.8 +asttokens==3.0.1 +annotated-types==0.7.0 +annotated-doc==0.0.4 +aiofiles==25.1.0 +absl-py==2.3.1 +uvicorn==0.39.0 +tiktoken==0.8.0 +scikit-learn==1.6.1 +sapien==3.0.0b1 +referencing==0.36.2 +pydantic==2.12.5 +pyasn1_modules==0.4.2 +prompt_toolkit==3.0.52 +nvidia-cusparse-cu12==12.1.0.106 +nvidia-cudnn-cu12==8.9.2.26 +nltk==3.9.2 +mujoco==2.3.7 +mdit-py-plugins==0.3.3 +matplotlib-inline==0.2.1 +matplotlib==3.7.5 +linkify-it-py==2.0.3 +jedi==0.19.2 +huggingface-hub==0.26.2 +httpcore==1.0.9 +gymnasium==0.29.1 +google-pasta==0.2.0 +gitdb==4.0.12 +cryptography==46.0.7 +blobfile==3.0.0 +anyio==4.12.1 +tyro==0.9.1 +toppra==0.6.6 +starlette==0.49.3 +seaborn==0.13.2 +rouge_score==0.1.2 +pynndescent==0.6.0 +nvidia-cusolver-cu12==11.4.5.107 +jsonschema-specifications==2025.9.1 +ipython==8.18.1 +httpx==0.28.1 +gymnasium-robotics==1.2.0 +google-auth==2.49.2 +GitPython==3.1.46 +diffusers==0.31.0 +wandb==0.18.6 +umap-learn==0.5.9.post2 +requests-oauthlib==2.0.0 +openai==2.8.1 +mplib==0.1.1 +jsonschema==4.25.1 +gradio_client==0.2.9 +google-genai==1.47.0 +fastapi==0.128.8 +torchvision==0.17.2 +torchaudio==2.2.2 +sentence-transformers==3.2.1 +pytorch-seed==0.2.0 +deepspeed==0.16.1 +bitsandbytes==0.45.0 +altair==6.0.0 +timm==1.0.22 +gradio==3.36.1 +evaluate==0.4.3 +arm_pytorch_utilities==0.5.0 +tensorflow-io-gcs-filesystem==0.37.1 +Markdown==3.9 +pybullet==3.2.6 +wrapt==1.14.2 +tensorflow-probability==0.23.0 +parse==1.19.1 +dm-tree==0.1.8 +termcolor==3.1.0 +requests==2.32.0 +cffi==2.0.0 +opt_einsum==3.4.0 +labmaze==1.0.6 +typing_extensions==4.15.0 +importlib-resources==5.13.0 +importlib-metadata==5.2.0 +FLAML==2.3.6 +dm-env==1.6 +diskcache==5.6.3 +Cython==0.29.37 +mujoco-py==2.1.2.14 +jaxlib==0.4.30 +dm-control==1.0.14 +cmudict==1.0.13 +syllables==1.0.9 +metaworld==2.0.0 +jax==0.4.30 +highway-env==1.9.1 +gym_bandits==0.0.2 +d4rl==1.1 +pyautogen==0.1.0 +protobuf==4.25.9 +tf-agents==0.19.0 +oauthlib==3.3.1 +pytorch-kinematics==0.7.5 +pygame==2.6.1 +gin-config==0.5.0 +mani_skill==3.0.0b20 +tifffile==2024.8.30 +Shapely==1.8.4 +PyWavelets==1.6.0 +Werkzeug==3.1.8 +astunparse==1.6.3 +scikit-image==0.19.3 +pymunk==6.2.1 +ml-dtypes==0.3.2 +gast==0.7.0 +stable-baselines3==2.2.1 +keras==2.15.0 +llfbench==0.1.0 +libclang==18.1.1 +flatbuffers==25.12.19 +gym==0.23.1 +llmbc==0.0.0 +llmbc==0.0.0 +llmbc==0.0.0 +aiohappyeyeballs==2.6.1 +antlr4-python3-runtime==4.9.3 +asciitree==0.3.3 +async-timeout==5.0.1 +attrs==25.4.0 +certifi==2025.10.5 +charset-normalizer==3.4.4 +colorama==0.4.6 +dill==0.3.8 +docstring-parser==0.16 +eval_type_backport==0.2.2 +fasteners==0.20 +filelock==3.17.0 +frozenlist==1.8.0 +fsspec==2024.3.1 +gmpy2==2.2.1 +hf-xet==1.1.8 +idna==3.11 +llvmlite==0.39.1 +MarkupSafe==3.0.2 +mdurl==0.1.2 +mkl-service==2.4.0 +mpmath==1.3.0 +msgpack==1.1.1 +networkx==3.2.1 +packaging==25.0 +propcache==0.3.1 +psutil==7.0.0 +pycparser==2.23 +Pygments==2.19.1 +PyOpenGL==3.1.10 +PySocks==1.7.1 +tzdata==2025.3 +xxhash==3.5.0 +pytz==2025.2 +PyYAML==6.0.2 +regex==2025.9.1 +safetensors==0.5.3 +setuptools==80.9.0 +shtab==1.7.2 +six==1.17.0 +tqdm==4.67.1 +tensorboard==2.15.2 +uvloop==0.22.1 +wheel==0.45.1 +zipp==3.21.0 +tensorflow-estimator==2.15.0 +tensorboard-data-server==0.7.2 +google-auth-oauthlib==1.3.1 +tensorflow==2.15.1 +Jinja2==3.1.6 +markdown-it-py==2.2.0 +multiprocess==0.70.15 +numpy==1.23.5 +pillow==11.3.0 +pip==23.3.2 +python-dateutil==2.9.0.post0 +aiosignal==1.4.0 +brotlicffi==1.0.9.2 +hydra-core==1.2.0 +multidict==6.7.0 +pycares==4.10.0 +rich==14.2.0 +sympy==1.14.0 +typeguard==4.4.4 +aiodns==3.5.0 +urllib3==2.5.0 +yarl==1.18.0 +aiohttp==3.12.15 +pyarrow==21.0.0 +pyarrow-hotfix==0.7 +tokenizers==0.21.0 +av==10.0.0 +einops==0.4.1 +Bottleneck==1.4.2 +mkl_fft==1.3.11 +mkl_random==1.2.8 +numba==0.56.4 +numcodecs==0.12.1 +numexpr==2.10.1 +pandas==2.3.3 +zarr==2.12.0 +datasets==2.19.0 +transformers==4.47.1 +accelerate==1.0.1 +peft==0.14.0 +trl==0.11.4 +torch==2.2.2 +tabulate==0.9.0 +sentencepiece==0.2.0 +pyperclip==1.11.0 +PyOpenGL-accelerate==3.1.10 +pydub==0.25.1 +py-cpuinfo==9.0.0 +pure_eval==0.2.3 +ptyprocess==0.7.0 +nvidia-ml-py==13.595.45 +hjson==3.1.0 +gym-notices==0.1.0 +glfw==2.0.0 +Farama-Notifications==0.0.4 +websockets==15.0.1 +wcwidth==0.6.0 +uc-micro-py==1.0.3 +typing-inspection==0.4.2 +triton==2.2.0 +trimesh==4.11.5 +transforms3d==0.4.2 +traitlets==5.14.3 +threadpoolctl==3.6.0 +tenacity==9.1.2 +sniffio==1.3.1 +smmap==5.0.3 +setproctitle==1.3.7 +sentry-sdk==2.57.0 +semantic-version==2.10.0 +scipy==1.13.1 +rpds-py==0.27.1 +python-multipart==0.0.20 +pyparsing==3.3.2 +pynvml==13.0.1 +pydantic_core==2.41.5 +pycryptodomex==3.23.0 +pyasn1==0.6.3 +grpcio==1.80.0 +platformdirs==4.4.0 +pexpect==4.9.0 +parso==0.8.6 +orjson==3.11.5 +opencv-python==4.11.0.86 +stack-data==0.6.3 +omegaconf==2.2.1 +nvidia-nvtx-cu12==12.1.105 +nvidia-nvjitlink-cu12==12.9.86 +nvidia-nccl-cu12==2.19.3 +nvidia-curand-cu12==10.3.2.106 +nvidia-cufft-cu12==11.0.2.54 +nvidia-cuda-runtime-cu12==12.1.105 +nvidia-cuda-nvrtc-cu12==12.1.105 +nvidia-cuda-cupti-cu12==12.1.105 +nvidia-cublas-cu12==12.1.3.1 +ninja==1.13.0 +narwhals==2.19.0 +lxml==6.0.3 +kiwisolver==1.4.7 +joblib==1.5.3 +jiter==0.13.0 +imageio-ffmpeg==0.6.0 +ImageIO==2.37.2 +h5py==3.14.0 +h11==0.16.0 +fonttools==4.60.2 +ffmpy==1.0.0 +fast_kinematics==0.2.2 +executing==2.2.1 +exceptiongroup==1.3.1 +docker-pycreds==0.4.0 +distro==1.9.0 +decorator==5.2.1 +dacite==1.9.2 +cycler==0.12.1 +contourpy==1.3.0 +cloudpickle==3.1.2 +click==8.1.8 +asttokens==3.0.1 +annotated-types==0.7.0 +annotated-doc==0.0.4 +aiofiles==25.1.0 +absl-py==2.3.1 +uvicorn==0.39.0 +tiktoken==0.8.0 +scikit-learn==1.6.1 +sapien==3.0.0b1 +referencing==0.36.2 +pydantic==2.12.5 +pyasn1_modules==0.4.2 +prompt_toolkit==3.0.52 +nvidia-cusparse-cu12==12.1.0.106 +nvidia-cudnn-cu12==8.9.2.26 +nltk==3.9.2 +mujoco==2.3.7 +mdit-py-plugins==0.3.3 +matplotlib-inline==0.2.1 +matplotlib==3.7.5 +linkify-it-py==2.0.3 +jedi==0.19.2 +huggingface-hub==0.26.2 +httpcore==1.0.9 +gymnasium==0.29.1 +google-pasta==0.2.0 +gitdb==4.0.12 +cryptography==46.0.7 +blobfile==3.0.0 +anyio==4.12.1 +tyro==0.9.1 +toppra==0.6.6 +starlette==0.49.3 +seaborn==0.13.2 +rouge_score==0.1.2 +pynndescent==0.6.0 +nvidia-cusolver-cu12==11.4.5.107 +jsonschema-specifications==2025.9.1 +ipython==8.18.1 +httpx==0.28.1 +gymnasium-robotics==1.2.0 +google-auth==2.49.2 +GitPython==3.1.46 +diffusers==0.31.0 +wandb==0.18.6 +umap-learn==0.5.9.post2 +requests-oauthlib==2.0.0 +openai==2.8.1 +mplib==0.1.1 +jsonschema==4.25.1 +gradio_client==0.2.9 +google-genai==1.47.0 +fastapi==0.128.8 +torchvision==0.17.2 +torchaudio==2.2.2 +sentence-transformers==3.2.1 +pytorch-seed==0.2.0 +deepspeed==0.16.1 +bitsandbytes==0.45.0 +altair==6.0.0 +timm==1.0.22 +gradio==3.36.1 +evaluate==0.4.3 +arm_pytorch_utilities==0.5.0 +tensorflow-io-gcs-filesystem==0.37.1 +Markdown==3.9 +pybullet==3.2.6 +wrapt==1.14.2 +tensorflow-probability==0.23.0 +parse==1.19.1 +dm-tree==0.1.8 +termcolor==3.1.0 +requests==2.32.0 +cffi==2.0.0 +opt_einsum==3.4.0 +labmaze==1.0.6 +typing_extensions==4.15.0 +importlib-resources==5.13.0 +importlib-metadata==5.2.0 +FLAML==2.3.6 +dm-env==1.6 +diskcache==5.6.3 +Cython==0.29.37 +mujoco-py==2.1.2.14 +jaxlib==0.4.30 +dm-control==1.0.14 +cmudict==1.0.13 +syllables==1.0.9 +metaworld==2.0.0 +jax==0.4.30 +highway-env==1.9.1 +gym_bandits==0.0.2 +d4rl==1.1 +pyautogen==0.1.0 +protobuf==4.25.9 +tf-agents==0.19.0 +oauthlib==3.3.1 +pytorch-kinematics==0.7.5 +pygame==2.6.1 +gin-config==0.5.0 +mani_skill==3.0.0b20 +tifffile==2024.8.30 +Shapely==1.8.4 +PyWavelets==1.6.0 +Werkzeug==3.1.8 +astunparse==1.6.3 +scikit-image==0.19.3 +pymunk==6.2.1 +ml-dtypes==0.3.2 +gast==0.7.0 +stable-baselines3==2.2.1 +keras==2.15.0 +llfbench==0.1.0 +libclang==18.1.1 +flatbuffers==25.12.19 +gym==0.23.1 +llmbc==0.0.0 +autocommand==2.2.2 +backports.tarfile==1.2.0 +importlib_metadata==8.0.0 +inflect==7.3.1 +jaraco.collections==5.1.0 +jaraco.context==5.3.0 +jaraco.functools==4.0.1 +jaraco.text==3.12.1 +more-itertools==10.3.0 +packaging==24.2 +platformdirs==4.2.2 +tomli==2.0.1 +typeguard==4.3.0 +typing_extensions==4.12.2 +wheel==0.45.1 +zipp==3.19.2 diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/wandb-metadata.json b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/wandb-metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..696efd852e6db6d88e754f72090493e54f138fec --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/wandb-metadata.json @@ -0,0 +1,56 @@ +{ + "os": "Linux-5.15.0-168-generic-x86_64-with-glibc2.39", + "python": "3.9.25", + "startedAt": "2026-04-10T09:37:42.923208Z", + "args": [ + "--config-path", + "config/mix_data", + "--config-name", + "llmbc_maze2d-medium-v0_mix(csp5).yaml", + "training.seed=42" + ], + "program": "/workspace/LLM-BC/./train.py", + "codePath": "train.py", + "git": { + "remote": "https://github.com/CHYang25/LLM-BC.git", + "commit": "2fa092f9eb4c2ad9435919c661a618071a583be5" + }, + "email": "chris920325@gmail.com", + "root": "/workspace/LLM-BC/data/outputs/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0", + "host": "c8ef52649900", + "username": "root", + "executable": "/venv/llm-bc/bin/python3", + "codePathLocal": "train.py", + "cpu_count": 72, + "cpu_count_logical": 144, + "gpu": "NVIDIA GeForce RTX 4090", + "gpu_count": 2, + "disk": { + "/": { + "total": "545460846592", + "used": "335971295232" + } + }, + "memory": { + "total": "810942898176" + }, + "cpu": { + "count": 72, + "countLogical": 144 + }, + "gpu_nvidia": [ + { + "name": "NVIDIA GeForce RTX 4090", + "memoryTotal": "25757220864", + "cudaCores": 16384, + "architecture": "Ada" + }, + { + "name": "NVIDIA GeForce RTX 4090", + "memoryTotal": "25757220864", + "cudaCores": 16384, + "architecture": "Ada" + } + ], + "cudaVersion": "12.8" +} \ No newline at end of file diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/wandb-summary.json b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/wandb-summary.json new file mode 100644 index 0000000000000000000000000000000000000000..3dda668bcc36b51e8d30061ebf9bd54601dbbd7b --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/files/wandb-summary.json @@ -0,0 +1 @@ +{"train_loss":0.4309055507183075,"epoch":0,"_wandb":{"runtime":146},"train_loss_llm":0.1051386296749115,"_runtime":146.804598991,"global_step":993,"lr":9.999992604259137e-05,"grad_norm":0.13102659583091736,"_step":993,"_timestamp":1.7758140095915012e+09,"train_loss_bc":0.429854154586792} \ No newline at end of file diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/run-2o6jqedz.wandb b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/run-2o6jqedz.wandb new file mode 100644 index 0000000000000000000000000000000000000000..6096b95ddcf6c699758749e4104c78e82951a7da --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/run-20260410_093742-2o6jqedz/run-2o6jqedz.wandb @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7a19821c4a683fd144338c84eb916c0c11650e775b42b477195d4aa6ae7ec8f5 +size 753664 diff --git a/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/wandb-resume.json b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/wandb-resume.json new file mode 100644 index 0000000000000000000000000000000000000000..65aee273d7ef6ddc47e41ad0e5fee03ca9c7a836 --- /dev/null +++ b/2026.04.10/09.36.48_train_llmbc_lowdim_maze2d-medium-v0/wandb/wandb-resume.json @@ -0,0 +1 @@ +{"run_id": "2o6jqedz"} \ No newline at end of file diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..688c0cf11e42e639a2dcc43dbbe166e0b5ff10ea --- /dev/null +++ b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml @@ -0,0 +1,163 @@ +name: train_llmbc_lowdim +_target_: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +model_name: ${llm.name} +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +llm_orig_expert_feedback: true +llm_do_sample: false +policy: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + task_id: ${task_name} + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + cfg: ${llm} + obs_dim: ${task.obs_dim} + action_dim: ${task.action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + loss_bc_weight: 1.0 + loss_llm_weight: 0.01 + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + normalize_llm_loss: true +dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.001 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: constant + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 8 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training-mix + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: adroit-hand-relocate-v1 + obs_dim: 39 + action_dim: 30 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + env_name: llf-adroit-adroit-hand-relocate-v1 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 150 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-mix(csp100).pt + data_path2: datasets/adroit-hand-relocate-v1.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.1 + dummy_normalizer: true + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor +llm: + name: HuggingFaceTB/SmolLM2-135M-Instruct + model_name: SmolLM2-135M-Instruct + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + use_quantization: false + use_joint_mlp_projector: true + llm_mode: ete-finetuned + finetune_mode: orig + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + max_length: 100 + lora_config: + r: 32 + lora_alpha: 64 + lora_dropout: 0.05 + bias: none + task_type: CAUSAL_LM + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + hydra: + job: + override_dirname: ${model_name} + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${model_name} diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..10fff6c4bd959e7229a81fdf0b10809f6fd63b15 --- /dev/null +++ b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml @@ -0,0 +1,154 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: llmbc_adroit-hand-relocate-v1_mix(csp100).yaml + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/config/mix_data + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1 + choices: + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt new file mode 100644 index 0000000000000000000000000000000000000000..d352d4e2e3931243dd2b2dbf03c98f7fddd50aa4 --- /dev/null +++ b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt @@ -0,0 +1,71 @@ +{"train_loss": 0.31865474581718445, "train_loss_bc": 0.3170997202396393, "train_loss_llm": 0.155502587556839, "grad_norm": 0.037330325692892075, "global_step": 0, "epoch": 0, "lr": 0.001} +{"train_loss": 0.32586202025413513, "train_loss_bc": 0.3245287835597992, "train_loss_llm": 0.13332363963127136, "grad_norm": 0.034294720739126205, "global_step": 1, "epoch": 0, "lr": 0.001} +{"train_loss": 0.3280045688152313, "train_loss_bc": 0.3266527056694031, "train_loss_llm": 0.13518568873405457, "grad_norm": 0.06898238509893417, "global_step": 2, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2669298052787781, "train_loss_bc": 0.2651206851005554, "train_loss_llm": 0.1809120774269104, "grad_norm": 0.1012270525097847, "global_step": 3, "epoch": 0, "lr": 0.001} +{"train_loss": 0.3040700852870941, "train_loss_bc": 0.3034462630748749, "train_loss_llm": 0.06238079071044922, "grad_norm": 0.13301171362400055, "global_step": 4, "epoch": 0, "lr": 0.001} +{"train_loss": 0.31336113810539246, "train_loss_bc": 0.31230518221855164, "train_loss_llm": 0.10559609532356262, "grad_norm": 0.1672557145357132, "global_step": 5, "epoch": 0, "lr": 0.001} +{"train_loss": 0.28599026799201965, "train_loss_bc": 0.28475603461265564, "train_loss_llm": 0.12342292070388794, "grad_norm": 0.19959396123886108, "global_step": 6, "epoch": 0, "lr": 0.001} +{"train_loss": 0.3234570324420929, "train_loss_bc": 0.32263192534446716, "train_loss_llm": 0.08250969648361206, "grad_norm": 0.23234137892723083, "global_step": 7, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2769324481487274, "train_loss_bc": 0.2754666209220886, "train_loss_llm": 0.14658260345458984, "grad_norm": 0.2625921964645386, "global_step": 8, "epoch": 0, "lr": 0.001} +{"train_loss": 0.3109692931175232, "train_loss_bc": 0.3099941611289978, "train_loss_llm": 0.09751325845718384, "grad_norm": 0.039986252784729004, "global_step": 9, "epoch": 0, "lr": 0.001} +{"train_loss": 0.28615450859069824, "train_loss_bc": 0.28488457202911377, "train_loss_llm": 0.12699276208877563, "grad_norm": 0.0715850293636322, "global_step": 10, "epoch": 0, "lr": 0.001} +{"train_loss": 0.24621188640594482, "train_loss_bc": 0.2453380823135376, "train_loss_llm": 0.08738017082214355, "grad_norm": 0.10016963630914688, "global_step": 11, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2773529887199402, "train_loss_bc": 0.27618131041526794, "train_loss_llm": 0.11716768145561218, "grad_norm": 0.1299743503332138, "global_step": 12, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2735079228878021, "train_loss_bc": 0.2726663053035736, "train_loss_llm": 0.08416306972503662, "grad_norm": 0.16233587265014648, "global_step": 13, "epoch": 0, "lr": 0.001} +{"train_loss": 0.30661383271217346, "train_loss_bc": 0.30552756786346436, "train_loss_llm": 0.10862642526626587, "grad_norm": 0.19619083404541016, "global_step": 14, "epoch": 0, "lr": 0.001} +{"train_loss": 0.28366735577583313, "train_loss_bc": 0.28252166509628296, "train_loss_llm": 0.1145695149898529, "grad_norm": 0.22849199175834656, "global_step": 15, "epoch": 0, "lr": 0.001} +{"train_loss": 0.22930900752544403, "train_loss_bc": 0.22819356620311737, "train_loss_llm": 0.11154377460479736, "grad_norm": 0.2590712606906891, "global_step": 16, "epoch": 0, "lr": 0.001} +{"train_loss": 0.22637949883937836, "train_loss_bc": 0.22556377947330475, "train_loss_llm": 0.08157205581665039, "grad_norm": 0.031221352517604828, "global_step": 17, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19987130165100098, "train_loss_bc": 0.19892151653766632, "train_loss_llm": 0.09497779607772827, "grad_norm": 0.062297701835632324, "global_step": 18, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2648264169692993, "train_loss_bc": 0.2638438940048218, "train_loss_llm": 0.09825262427330017, "grad_norm": 0.09701936691999435, "global_step": 19, "epoch": 0, "lr": 0.001} +{"train_loss": 0.26097890734672546, "train_loss_bc": 0.2599954307079315, "train_loss_llm": 0.09834623336791992, "grad_norm": 0.1258426308631897, "global_step": 20, "epoch": 0, "lr": 0.001} +{"train_loss": 0.25044819712638855, "train_loss_bc": 0.2498278021812439, "train_loss_llm": 0.062038302421569824, "grad_norm": 0.15424573421478271, "global_step": 21, "epoch": 0, "lr": 0.001} +{"train_loss": 0.25505244731903076, "train_loss_bc": 0.25415077805519104, "train_loss_llm": 0.09016796946525574, "grad_norm": 0.186727836728096, "global_step": 22, "epoch": 0, "lr": 0.001} +{"train_loss": 0.23085804283618927, "train_loss_bc": 0.23013581335544586, "train_loss_llm": 0.07222336530685425, "grad_norm": 0.21632854640483856, "global_step": 23, "epoch": 0, "lr": 0.001} +{"train_loss": 0.24062438309192657, "train_loss_bc": 0.23985892534255981, "train_loss_llm": 0.0765453577041626, "grad_norm": 0.24554501473903656, "global_step": 24, "epoch": 0, "lr": 0.001} +{"train_loss": 0.23058189451694489, "train_loss_bc": 0.22989396750926971, "train_loss_llm": 0.06879279017448425, "grad_norm": 0.0329669751226902, "global_step": 25, "epoch": 0, "lr": 0.001} +{"train_loss": 0.23450350761413574, "train_loss_bc": 0.2336718738079071, "train_loss_llm": 0.08316299319267273, "grad_norm": 0.06469928473234177, "global_step": 26, "epoch": 0, "lr": 0.001} +{"train_loss": 0.21867641806602478, "train_loss_bc": 0.21810907125473022, "train_loss_llm": 0.05673402547836304, "grad_norm": 0.09001963585615158, "global_step": 27, "epoch": 0, "lr": 0.001} +{"train_loss": 0.23224763572216034, "train_loss_bc": 0.2318083494901657, "train_loss_llm": 0.04392918944358826, "grad_norm": 0.1155921071767807, "global_step": 28, "epoch": 0, "lr": 0.001} +{"train_loss": 0.23083648085594177, "train_loss_bc": 0.23035317659378052, "train_loss_llm": 0.048329949378967285, "grad_norm": 0.14488989114761353, "global_step": 29, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2155812531709671, "train_loss_bc": 0.21481823921203613, "train_loss_llm": 0.07630094885826111, "grad_norm": 0.17692092061042786, "global_step": 30, "epoch": 0, "lr": 0.001} +{"train_loss": 0.22036553919315338, "train_loss_bc": 0.22007495164871216, "train_loss_llm": 0.02905815839767456, "grad_norm": 0.20235739648342133, "global_step": 31, "epoch": 0, "lr": 0.001} +{"train_loss": 0.18274901807308197, "train_loss_bc": 0.1821819394826889, "train_loss_llm": 0.056708335876464844, "grad_norm": 0.2313731461763382, "global_step": 32, "epoch": 0, "lr": 0.001} +{"train_loss": 0.22325503826141357, "train_loss_bc": 0.2225644737482071, "train_loss_llm": 0.06905588507652283, "grad_norm": 0.029686858877539635, "global_step": 33, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2014690786600113, "train_loss_bc": 0.20082609355449677, "train_loss_llm": 0.06429821252822876, "grad_norm": 0.05757380276918411, "global_step": 34, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2072751224040985, "train_loss_bc": 0.20646369457244873, "train_loss_llm": 0.08114317059516907, "grad_norm": 0.08581757545471191, "global_step": 35, "epoch": 0, "lr": 0.001} +{"train_loss": 0.24447759985923767, "train_loss_bc": 0.2439330816268921, "train_loss_llm": 0.05445152521133423, "grad_norm": 0.11434584856033325, "global_step": 36, "epoch": 0, "lr": 0.001} +{"train_loss": 0.18403244018554688, "train_loss_bc": 0.18347524106502533, "train_loss_llm": 0.055720120668411255, "grad_norm": 0.1417362242937088, "global_step": 37, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19022689759731293, "train_loss_bc": 0.18980389833450317, "train_loss_llm": 0.042300403118133545, "grad_norm": 0.164987713098526, "global_step": 38, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19494764506816864, "train_loss_bc": 0.1940891295671463, "train_loss_llm": 0.08585211634635925, "grad_norm": 0.1930799037218094, "global_step": 39, "epoch": 0, "lr": 0.001} +{"train_loss": 0.22805000841617584, "train_loss_bc": 0.22751189768314362, "train_loss_llm": 0.05381166934967041, "grad_norm": 0.2215198278427124, "global_step": 40, "epoch": 0, "lr": 0.001} +{"train_loss": 0.20951256155967712, "train_loss_bc": 0.2090826779603958, "train_loss_llm": 0.042988866567611694, "grad_norm": 0.03214744105935097, "global_step": 41, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19074368476867676, "train_loss_bc": 0.19043195247650146, "train_loss_llm": 0.031173378229141235, "grad_norm": 0.053819525986909866, "global_step": 42, "epoch": 0, "lr": 0.001} +{"train_loss": 0.17628498375415802, "train_loss_bc": 0.1759534776210785, "train_loss_llm": 0.03315111994743347, "grad_norm": 0.07898898422718048, "global_step": 43, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2021668255329132, "train_loss_bc": 0.20168636739253998, "train_loss_llm": 0.048046112060546875, "grad_norm": 0.10661231726408005, "global_step": 44, "epoch": 0, "lr": 0.001} +{"train_loss": 0.1748126894235611, "train_loss_bc": 0.17407286167144775, "train_loss_llm": 0.07398265600204468, "grad_norm": 0.12946173548698425, "global_step": 45, "epoch": 0, "lr": 0.001} +{"train_loss": 0.17575593292713165, "train_loss_bc": 0.17541193962097168, "train_loss_llm": 0.034399181604385376, "grad_norm": 0.15593087673187256, "global_step": 46, "epoch": 0, "lr": 0.001} +{"train_loss": 0.22766298055648804, "train_loss_bc": 0.22704260051250458, "train_loss_llm": 0.062038689851760864, "grad_norm": 0.17924897372722626, "global_step": 47, "epoch": 0, "lr": 0.001} +{"train_loss": 0.17298723757266998, "train_loss_bc": 0.1727081537246704, "train_loss_llm": 0.02790820598602295, "grad_norm": 0.20546390116214752, "global_step": 48, "epoch": 0, "lr": 0.001} +{"train_loss": 0.14932842552661896, "train_loss_bc": 0.1488334685564041, "train_loss_llm": 0.04949536919593811, "grad_norm": 0.025824714452028275, "global_step": 49, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19859372079372406, "train_loss_bc": 0.1980268359184265, "train_loss_llm": 0.056688904762268066, "grad_norm": 0.04908924177289009, "global_step": 50, "epoch": 0, "lr": 0.001} +{"train_loss": 0.1671150028705597, "train_loss_bc": 0.16687263548374176, "train_loss_llm": 0.024236023426055908, "grad_norm": 0.07283542305231094, "global_step": 51, "epoch": 0, "lr": 0.001} +{"train_loss": 0.17893573641777039, "train_loss_bc": 0.17849938571453094, "train_loss_llm": 0.04363524913787842, "grad_norm": 0.09348991513252258, "global_step": 52, "epoch": 0, "lr": 0.001} +{"train_loss": 0.22472478449344635, "train_loss_bc": 0.2246263325214386, "train_loss_llm": 0.00984451174736023, "grad_norm": 0.11573847383260727, "global_step": 53, "epoch": 0, "lr": 0.001} +{"train_loss": 0.17013958096504211, "train_loss_bc": 0.16962361335754395, "train_loss_llm": 0.05159679055213928, "grad_norm": 0.14166922867298126, "global_step": 54, "epoch": 0, "lr": 0.001} +{"train_loss": 0.1889440417289734, "train_loss_bc": 0.18821223080158234, "train_loss_llm": 0.07318049669265747, "grad_norm": 0.162848100066185, "global_step": 55, "epoch": 0, "lr": 0.001} +{"train_loss": 0.21185244619846344, "train_loss_bc": 0.21140828728675842, "train_loss_llm": 0.04441553354263306, "grad_norm": 0.1916223019361496, "global_step": 56, "epoch": 0, "lr": 0.001} +{"train_loss": 0.17594307661056519, "train_loss_bc": 0.1753692328929901, "train_loss_llm": 0.05738368630409241, "grad_norm": 0.021315308287739754, "global_step": 57, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19558878242969513, "train_loss_bc": 0.19496507942676544, "train_loss_llm": 0.062370359897613525, "grad_norm": 0.04522071033716202, "global_step": 58, "epoch": 0, "lr": 0.001} +{"train_loss": 0.20047597587108612, "train_loss_bc": 0.2002260386943817, "train_loss_llm": 0.024993091821670532, "grad_norm": 0.0678013265132904, "global_step": 59, "epoch": 0, "lr": 0.001} +{"train_loss": 0.16332882642745972, "train_loss_bc": 0.16282124817371368, "train_loss_llm": 0.05075785517692566, "grad_norm": 0.08263341337442398, "global_step": 60, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19273340702056885, "train_loss_bc": 0.1921701282262802, "train_loss_llm": 0.056327760219573975, "grad_norm": 0.10481162369251251, "global_step": 61, "epoch": 0, "lr": 0.001} +{"train_loss": 0.15804670751094818, "train_loss_bc": 0.15773825347423553, "train_loss_llm": 0.03084525465965271, "grad_norm": 0.12284409254789352, "global_step": 62, "epoch": 0, "lr": 0.001} +{"train_loss": 0.16953814029693604, "train_loss_bc": 0.16868296265602112, "train_loss_llm": 0.08551749587059021, "grad_norm": 0.14134983718395233, "global_step": 63, "epoch": 0, "lr": 0.001} +{"train_loss": 0.13983431458473206, "train_loss_bc": 0.13944396376609802, "train_loss_llm": 0.03903469443321228, "grad_norm": 0.16004106402397156, "global_step": 64, "epoch": 0, "lr": 0.001} +{"train_loss": 0.15782199800014496, "train_loss_bc": 0.1570679098367691, "train_loss_llm": 0.07540848851203918, "grad_norm": 0.023670973256230354, "global_step": 65, "epoch": 0, "lr": 0.001} +{"train_loss": 0.14042401313781738, "train_loss_bc": 0.13989663124084473, "train_loss_llm": 0.052738070487976074, "grad_norm": 0.048458635807037354, "global_step": 66, "epoch": 0, "lr": 0.001} +{"train_loss": 0.161502406001091, "train_loss_bc": 0.16071049869060516, "train_loss_llm": 0.07919105887413025, "grad_norm": 0.06283467262983322, "global_step": 67, "epoch": 0, "lr": 0.001} +{"train_loss": 0.1589268445968628, "train_loss_bc": 0.15807612240314484, "train_loss_llm": 0.08507251739501953, "grad_norm": 0.08433432877063751, "global_step": 68, "epoch": 0, "lr": 0.001} +{"train_loss": 0.15944941341876984, "train_loss_bc": 0.15855075418949127, "train_loss_llm": 0.08986592292785645, "grad_norm": 0.10458381474018097, "global_step": 69, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19175000488758087, "train_loss_bc": 0.19121704995632172, "train_loss_llm": 0.05329522490501404, "grad_norm": 0.1160665899515152, "global_step": 70, "epoch": 0, "lr": 0.001} diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log new file mode 100644 index 0000000000000000000000000000000000000000..76a390027aed9b4ea3dd16be49faaaa4b147def5 --- /dev/null +++ b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log @@ -0,0 +1,13 @@ +[2026-04-10 09:40:21,840][numexpr.utils][INFO] - Note: detected 144 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable. +[2026-04-10 09:40:21,840][numexpr.utils][INFO] - Note: NumExpr detected 144 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 16. +[2026-04-10 09:40:21,840][numexpr.utils][INFO] - NumExpr defaulting to 16 threads. +[2026-04-10 09:40:25,248][datasets][INFO] - PyTorch version 2.2.2 available. +[2026-04-10 09:40:25,248][datasets][INFO] - TensorFlow version 2.15.1 available. +[2026-04-10 09:40:25,249][datasets][INFO] - JAX version 0.4.30 available. +[2026-04-10 09:40:29,360][root][INFO] - running build_ext +[2026-04-10 09:40:29,362][root][INFO] - building 'mujoco_py.cymj' extension +[2026-04-10 09:40:29,362][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o -fopenmp -w +[2026-04-10 09:41:10,848][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -fopenmp -w +[2026-04-10 09:41:10,947][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -shared -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib -L/venv/llm-bc/lib -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -L/root/.mujoco/mujoco210/bin -Wl,--enable-new-dtags,-rpath,/root/.mujoco/mujoco210/bin -lmujoco210 -lglewosmesa -lOSMesa -lGL -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/lib.linux-x86_64-cpython-39/mujoco_py/cymj.cpython-39-x86_64-linux-gnu.so -fopenmp +[2026-04-10 09:41:13,603][absl][INFO] - MUJOCO_GL=osmesa, attempting to import specified OpenGL backend. +[2026-04-10 09:41:13,608][absl][INFO] - MuJoCo library version is: 2.3.7 diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/config.yaml b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..84360b239f4007c4d46abf4b39dc8abc1d282a7c --- /dev/null +++ b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/config.yaml @@ -0,0 +1,271 @@ +_target_: + value: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +_wandb: + value: + cli_version: 0.18.6 + m: [] + python_version: 3.9.25 + t: + "1": + - 1 + - 2 + - 3 + - 5 + - 11 + - 12 + - 41 + - 49 + - 50 + - 51 + - 53 + - 55 + - 71 + - 83 + - 95 + - 98 + - 100 + - 105 + "2": + - 1 + - 2 + - 3 + - 5 + - 11 + - 12 + - 41 + - 49 + - 50 + - 51 + - 53 + - 55 + - 71 + - 83 + - 95 + - 98 + - 100 + - 105 + "3": + - 13 + - 15 + - 16 + - 23 + - 55 + - 61 + "4": 3.9.25 + "5": 0.18.6 + "6": 4.47.1 + "8": + - 5 + "12": 0.18.6 + "13": linux-x86_64 +action_dim: + value: 30 +checkpoint: + value: + save_last_ckpt: true + save_last_snapshot: false + topk: + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + k: 5 + mode: max + monitor_key: test_success_rate +dataloader: + value: + batch_size: 16 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true +exp_name: + value: default +horizon: + value: 1 +llm: + value: + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + finetune_mode: orig + hydra: + job: + override_dirname: HuggingFaceTB/SmolLM2-135M-Instruct + run: + dir: data/outputs/2026.04.10/09.40.20_HuggingFaceTB/SmolLM2-135M-Instruct + llm_mode: ete-finetuned + lora_config: + bias: none + lora_alpha: 64 + lora_dropout: 0.05 + r: 32 + task_type: CAUSAL_LM + max_length: 100 + model_name: SmolLM2-135M-Instruct + name: HuggingFaceTB/SmolLM2-135M-Instruct + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + use_joint_mlp_projector: true + use_quantization: false +llm_do_sample: + value: false +llm_orig_expert_feedback: + value: true +logging: + value: + group: null + id: null + mode: online + name: 2026.04.10-09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1 + project: adroit-hand-relocate-v1-training-mix + resume: true + tags: + - train_llmbc_lowdim + - adroit-hand-relocate-v1 + - default +model_name: + value: HuggingFaceTB/SmolLM2-135M-Instruct +multi_run: + value: + run_dir: data/outputs/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1 + wandb_name_base: 2026.04.10-09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1 +n_action_steps: + value: 1 +n_latency_steps: + value: 0 +n_obs_steps: + value: 1 +name: + value: train_llmbc_lowdim +obs_dim: + value: 39 +optimizer: + value: + _target_: torch.optim.AdamW + betas: + - 0.95 + - 0.999 + eps: 1e-08 + lr: 0.001 + weight_decay: 1e-06 +output_dir: + value: /workspace/LLM-BC/data/outputs/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1 +past_action_visible: + value: false +policy: + value: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + action_dim: 30 + horizon: 1 + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + action_dim: 30 + cfg: + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + finetune_mode: orig + hydra: + job: + override_dirname: HuggingFaceTB/SmolLM2-135M-Instruct + run: + dir: data/outputs/2026.04.10/09.40.20_HuggingFaceTB/SmolLM2-135M-Instruct + llm_mode: ete-finetuned + lora_config: + bias: none + lora_alpha: 64 + lora_dropout: 0.05 + r: 32 + task_type: CAUSAL_LM + max_length: 100 + model_name: SmolLM2-135M-Instruct + name: HuggingFaceTB/SmolLM2-135M-Instruct + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + use_joint_mlp_projector: true + use_quantization: false + horizon: 1 + n_action_steps: 1 + n_obs_steps: 1 + obs_dim: 39 + task_id: adroit-hand-relocate-v1 + loss_bc_weight: 1 + loss_llm_weight: 0.01 + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + activation: relu + hidden_size: + - 256 + - 256 + input_size: 39 + n_action_steps: 1 + n_obs_steps: 1 + output_size: 30 + n_action_steps: 1 + n_obs_steps: 1 + normalize_llm_loss: true + obs_dim: 39 +task: + value: + action_dim: 30 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-mix(csp100).pt + data_path2: datasets/adroit-hand-relocate-v1.pt + dummy_normalizer: true + horizon: 1 + obs_eef_target: true + pad_after: 0 + pad_before: 0 + use_manual_normalizer: false + val_ratio: 0.1 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + discount: 0.99 + env_name: llf-adroit-adroit-hand-relocate-v1 + feedback_type: + - hp + - hn + - fp + instruction_type: b + max_steps: 150 + n_action_steps: 1 + n_envs: 10 + n_obs_steps: 1 + n_test: 50 + n_train: 10 + visual: false + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor + name: adroit-hand-relocate-v1 + obs_dim: 39 +task_name: + value: adroit-hand-relocate-v1 +training: + value: + checkpoint_every: 5 + debug: false + device: cuda:0 + grad_norm_clip: 0.5 + gradient_accumulate_every: 8 + lr_scheduler: constant + lr_warmup_steps: 10 + max_train_steps: null + max_val_steps: null + num_epochs: 1001 + resume: false + rollout_every: 5 + sample_every: 5 + sample_max_batch: 128 + seed: 42 + tqdm_interval_sec: 1 + val_every: 1 +val_dataloader: + value: + batch_size: 16 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/output.log b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/output.log new file mode 100644 index 0000000000000000000000000000000000000000..952cb977f63c19d7d534cfe526422b776b95b79a --- /dev/null +++ b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/output.log @@ -0,0 +1,23 @@ +Error executing job with overrides: [] +Traceback (most recent call last): + File "/workspace/LLM-BC/./train.py", line 32, in main + workspace.run() + File "/workspace/LLM-BC/llmbc/workspace/train_llmbc_lowdim_workspace.py", line 238, in run + runner_log = env_runner.run(policy) + File "/workspace/LLM-BC/llmbc/env_runner/adroit_lowdim_runner.py", line 112, in run + self.env = AsyncVectorEnv(self.env_fns) + File "/workspace/LLM-BC/llmbc/gym_util/async_vector_env.py", line 103, in __init__ + dummy_env = dummy_env_fn() + File "/workspace/LLM-BC/llmbc/env_runner/adroit_lowdim_runner.py", line 56, in env_fn + gym.make( + File "/workspace/LLM-BC/LLF-Bench/llfbench/__init__.py", line 5, in make + env = gym.make(env_name, instruction_type=instruction_type, feedback_type=feedback_type, visual=visual, seed=seed, warning=warning, **env_kwargs) + File "/venv/llm-bc/lib/python3.9/site-packages/gymnasium/envs/registration.py", line 802, in make + env = env_creator(**env_spec_kwargs) + File "/workspace/LLM-BC/LLF-Bench/llfbench/envs/adroit/__init__.py", line 55, in make_env + return TimeLimit(AdroitWrapper(env, instruction_type=instruction_type, feedback_type=feedback_type), max_episode_steps=max_episode_steps) + File "/workspace/LLM-BC/LLF-Bench/llfbench/envs/adroit/wrapper.py", line 93, in __init__ + self._policy = pickle.load(f) +ModuleNotFoundError: No module named 'mjrl' + +Set the environment variable HYDRA_FULL_ERROR=1 for a complete stack trace. diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/wandb-metadata.json b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/wandb-metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..ab8c394d19868d170ed01ad126a94cd43d847af4 --- /dev/null +++ b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/wandb-metadata.json @@ -0,0 +1,55 @@ +{ + "os": "Linux-5.15.0-168-generic-x86_64-with-glibc2.39", + "python": "3.9.25", + "startedAt": "2026-04-10T09:41:14.532897Z", + "args": [ + "--config-path", + "config/mix_data", + "--config-name", + "llmbc_adroit-hand-relocate-v1_mix(csp100).yaml" + ], + "program": "/workspace/LLM-BC/./train.py", + "codePath": "train.py", + "git": { + "remote": "https://github.com/CHYang25/LLM-BC.git", + "commit": "2fa092f9eb4c2ad9435919c661a618071a583be5" + }, + "email": "chris920325@gmail.com", + "root": "/workspace/LLM-BC/data/outputs/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1", + "host": "c8ef52649900", + "username": "root", + "executable": "/venv/llm-bc/bin/python3", + "codePathLocal": "train.py", + "cpu_count": 72, + "cpu_count_logical": 144, + "gpu": "NVIDIA GeForce RTX 4090", + "gpu_count": 2, + "disk": { + "/": { + "total": "545460846592", + "used": "345529266176" + } + }, + "memory": { + "total": "810942898176" + }, + "cpu": { + "count": 72, + "countLogical": 144 + }, + "gpu_nvidia": [ + { + "name": "NVIDIA GeForce RTX 4090", + "memoryTotal": "25757220864", + "cudaCores": 16384, + "architecture": "Ada" + }, + { + "name": "NVIDIA GeForce RTX 4090", + "memoryTotal": "25757220864", + "cudaCores": 16384, + "architecture": "Ada" + } + ], + "cudaVersion": "12.8" +} \ No newline at end of file diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/wandb-summary.json b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/wandb-summary.json new file mode 100644 index 0000000000000000000000000000000000000000..0388352fd47cdd9caee0da42f7574f7ee280e155 --- /dev/null +++ b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/files/wandb-summary.json @@ -0,0 +1 @@ +{"_runtime":13.063473219,"epoch":0,"train_loss":0.19175000488758087,"_timestamp":1.7758140872613006e+09,"train_loss_llm":0.05329522490501404,"global_step":70,"lr":0.001,"train_loss_bc":0.19121704995632172,"_step":70,"_wandb":{"runtime":13},"grad_norm":0.1160665899515152} \ No newline at end of file diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/run-18cttt3f.wandb b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/run-18cttt3f.wandb new file mode 100644 index 0000000000000000000000000000000000000000..bb781084360c2522eaf5111e9a5a18acdd1e0292 Binary files /dev/null and b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094114-18cttt3f/run-18cttt3f.wandb differ diff --git a/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/wandb-resume.json b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/wandb-resume.json new file mode 100644 index 0000000000000000000000000000000000000000..67746cc96b3a59a44739348bc671e0744ad8fafb --- /dev/null +++ b/2026.04.10/09.40.20_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/wandb-resume.json @@ -0,0 +1 @@ +{"run_id": "18cttt3f"} \ No newline at end of file diff --git a/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml b/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..18aa4f4462d3efb73155313a9027971620c6732b --- /dev/null +++ b/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml @@ -0,0 +1,163 @@ +name: train_llmbc_lowdim +_target_: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +model_name: ${llm.name} +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +llm_orig_expert_feedback: true +llm_do_sample: false +policy: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + task_id: ${task_name} + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + cfg: ${llm} + obs_dim: ${task.obs_dim} + action_dim: ${task.action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + loss_bc_weight: 1.0 + loss_llm_weight: 0.01 + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + normalize_llm_loss: true +dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.001 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 43 + debug: false + resume: false + lr_scheduler: constant + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 8 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training-mix + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: adroit-hand-relocate-v1 + obs_dim: 39 + action_dim: 30 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + env_name: llf-adroit-adroit-hand-relocate-v1 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 150 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-mix(csp100).pt + data_path2: datasets/adroit-hand-relocate-v1.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.1 + dummy_normalizer: true + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor +llm: + name: HuggingFaceTB/SmolLM2-135M-Instruct + model_name: SmolLM2-135M-Instruct + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + use_quantization: false + use_joint_mlp_projector: true + llm_mode: ete-finetuned + finetune_mode: orig + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + max_length: 100 + lora_config: + r: 32 + lora_alpha: 64 + lora_dropout: 0.05 + bias: none + task_type: CAUSAL_LM + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + hydra: + job: + override_dirname: ${model_name} + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${model_name} diff --git a/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml b/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..bd70a445dbafb2ca3be3ea12e5e64beba16da5d0 --- /dev/null +++ b/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: + - training.seed=43 + job: + name: train + chdir: null + override_dirname: training.seed=43 + id: ??? + num: ??? + config_name: llmbc_adroit-hand-relocate-v1_mix(csp100).yaml + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/config/mix_data + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1 + choices: + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml b/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ebf61aeca24956094799b4a1cab9370df040027f --- /dev/null +++ b/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml @@ -0,0 +1 @@ +- training.seed=43 diff --git a/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log b/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log new file mode 100644 index 0000000000000000000000000000000000000000..d01d11270873ace2a9ec50cf7f025e4b8377dea3 --- /dev/null +++ b/2026.04.10/09.41.29_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log @@ -0,0 +1,3 @@ +[2026-04-10 09:41:31,080][numexpr.utils][INFO] - Note: detected 144 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable. +[2026-04-10 09:41:31,081][numexpr.utils][INFO] - Note: NumExpr detected 144 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 16. +[2026-04-10 09:41:31,081][numexpr.utils][INFO] - NumExpr defaulting to 16 threads. diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..18aa4f4462d3efb73155313a9027971620c6732b --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml @@ -0,0 +1,163 @@ +name: train_llmbc_lowdim +_target_: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +model_name: ${llm.name} +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +llm_orig_expert_feedback: true +llm_do_sample: false +policy: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + task_id: ${task_name} + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + cfg: ${llm} + obs_dim: ${task.obs_dim} + action_dim: ${task.action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + loss_bc_weight: 1.0 + loss_llm_weight: 0.01 + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + normalize_llm_loss: true +dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 16 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.001 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 43 + debug: false + resume: false + lr_scheduler: constant + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 8 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training-mix + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: adroit-hand-relocate-v1 + obs_dim: 39 + action_dim: 30 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + env_name: llf-adroit-adroit-hand-relocate-v1 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 150 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-mix(csp100).pt + data_path2: datasets/adroit-hand-relocate-v1.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.1 + dummy_normalizer: true + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor +llm: + name: HuggingFaceTB/SmolLM2-135M-Instruct + model_name: SmolLM2-135M-Instruct + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + use_quantization: false + use_joint_mlp_projector: true + llm_mode: ete-finetuned + finetune_mode: orig + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + max_length: 100 + lora_config: + r: 32 + lora_alpha: 64 + lora_dropout: 0.05 + bias: none + task_type: CAUSAL_LM + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + hydra: + job: + override_dirname: ${model_name} + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${model_name} diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..8a42c1626c41b75385ca0576156549ec9d1ee352 --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: + - training.seed=43 + job: + name: train + chdir: null + override_dirname: training.seed=43 + id: ??? + num: ??? + config_name: llmbc_adroit-hand-relocate-v1_mix(csp100).yaml + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/config/mix_data + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1 + choices: + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ebf61aeca24956094799b4a1cab9370df040027f --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml @@ -0,0 +1 @@ +- training.seed=43 diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0085-test_success_rate=0.320.ckpt b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0085-test_success_rate=0.320.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..6303aeb43a866e7696e23542fe8b6b03ea1b5d84 --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0085-test_success_rate=0.320.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ec9d3d904a79b2ddeb778615053900a8f86109a01065adb01e5ebd00516c0ce5 +size 1036936 diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0235-test_success_rate=0.320.ckpt b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0235-test_success_rate=0.320.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..df45956d18206c95fec76ee9f4fe1d1d9ea04502 --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0235-test_success_rate=0.320.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7dd80653708e2cf07af1bd549ce43ac7f10ebc6dacae0f70b492870fbddc6a8f +size 1036936 diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0320-test_success_rate=0.280.ckpt b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0320-test_success_rate=0.280.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..162cb2a4213e8204c9e6585c35e920e4e028e671 --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0320-test_success_rate=0.280.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:98b1c33783fe7a046e4a2ae06fd3d2dd448f586988a742a90aad73847b2c69f5 +size 1036936 diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0395-test_success_rate=0.320.ckpt b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0395-test_success_rate=0.320.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..ab57697768b200fd2c3646cf07f9461b5b850f1b --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0395-test_success_rate=0.320.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da99901cca58bb121dac33d2b2628bf0e8d9c877c58107345cecc6434cddae1f +size 1036936 diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0630-test_success_rate=0.300.ckpt b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0630-test_success_rate=0.300.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..9c1ec068b307b8f2e4390950db2fadde657432d7 --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0630-test_success_rate=0.300.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9109f94076384983e18fcf663fcc08e037b8bc7faefba240bfd494d4a2d28179 +size 1036936 diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/latest.ckpt b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/latest.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..4207499ba3ed209bb8328d336ff54584cd3ac64d --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/latest.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a1faf9736f848518224084432cfc79629987840e4ad6d23ec61a98ecd41e8e2f +size 1036936 diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt new file mode 100644 index 0000000000000000000000000000000000000000..94e397c97fe446029b01284ad7a14cf39bb192b2 --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:30a46e608a7aa277232196312eb43f44fe59ac749da2e13254862f885a955dec +size 12485162 diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log new file mode 100644 index 0000000000000000000000000000000000000000..37902d5809896ecd7ef9745dc4d82c7777ab3214 --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log @@ -0,0 +1,13 @@ +[2026-04-10 09:42:49,124][numexpr.utils][INFO] - Note: detected 144 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable. +[2026-04-10 09:42:49,124][numexpr.utils][INFO] - Note: NumExpr detected 144 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 16. +[2026-04-10 09:42:49,124][numexpr.utils][INFO] - NumExpr defaulting to 16 threads. +[2026-04-10 09:42:53,195][datasets][INFO] - PyTorch version 2.2.2 available. +[2026-04-10 09:42:53,196][datasets][INFO] - TensorFlow version 2.15.1 available. +[2026-04-10 09:42:53,196][datasets][INFO] - JAX version 0.4.30 available. +[2026-04-10 09:42:57,462][root][INFO] - running build_ext +[2026-04-10 09:42:57,464][root][INFO] - building 'mujoco_py.cymj' extension +[2026-04-10 09:42:57,464][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o -fopenmp -w +[2026-04-10 09:43:38,761][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -fopenmp -w +[2026-04-10 09:43:38,886][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -shared -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib -L/venv/llm-bc/lib -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -L/root/.mujoco/mujoco210/bin -Wl,--enable-new-dtags,-rpath,/root/.mujoco/mujoco210/bin -lmujoco210 -lglewosmesa -lOSMesa -lGL -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/lib.linux-x86_64-cpython-39/mujoco_py/cymj.cpython-39-x86_64-linux-gnu.so -fopenmp +[2026-04-10 09:44:21,911][absl][INFO] - MUJOCO_GL=osmesa, attempting to import specified OpenGL backend. +[2026-04-10 09:44:21,921][absl][INFO] - MuJoCo library version is: 2.3.7 diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094422-68a5glbf/files/config.yaml b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094422-68a5glbf/files/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..c440964d420e1aea7d691d4bf6943e0c263e9879 --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094422-68a5glbf/files/config.yaml @@ -0,0 +1,271 @@ +_target_: + value: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +_wandb: + value: + cli_version: 0.18.6 + m: [] + python_version: 3.9.25 + t: + "1": + - 1 + - 2 + - 3 + - 5 + - 11 + - 12 + - 41 + - 49 + - 50 + - 51 + - 53 + - 55 + - 71 + - 83 + - 95 + - 98 + - 100 + - 105 + "2": + - 1 + - 2 + - 3 + - 5 + - 11 + - 12 + - 41 + - 49 + - 50 + - 51 + - 53 + - 55 + - 71 + - 83 + - 95 + - 98 + - 100 + - 105 + "3": + - 13 + - 15 + - 16 + - 23 + - 55 + - 61 + "4": 3.9.25 + "5": 0.18.6 + "6": 4.47.1 + "8": + - 5 + "12": 0.18.6 + "13": linux-x86_64 +action_dim: + value: 30 +checkpoint: + value: + save_last_ckpt: true + save_last_snapshot: false + topk: + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + k: 5 + mode: max + monitor_key: test_success_rate +dataloader: + value: + batch_size: 16 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true +exp_name: + value: default +horizon: + value: 1 +llm: + value: + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + finetune_mode: orig + hydra: + job: + override_dirname: HuggingFaceTB/SmolLM2-135M-Instruct + run: + dir: data/outputs/2026.04.10/09.42.47_HuggingFaceTB/SmolLM2-135M-Instruct + llm_mode: ete-finetuned + lora_config: + bias: none + lora_alpha: 64 + lora_dropout: 0.05 + r: 32 + task_type: CAUSAL_LM + max_length: 100 + model_name: SmolLM2-135M-Instruct + name: HuggingFaceTB/SmolLM2-135M-Instruct + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + use_joint_mlp_projector: true + use_quantization: false +llm_do_sample: + value: false +llm_orig_expert_feedback: + value: true +logging: + value: + group: null + id: null + mode: online + name: 2026.04.10-09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1 + project: adroit-hand-relocate-v1-training-mix + resume: true + tags: + - train_llmbc_lowdim + - adroit-hand-relocate-v1 + - default +model_name: + value: HuggingFaceTB/SmolLM2-135M-Instruct +multi_run: + value: + run_dir: data/outputs/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1 + wandb_name_base: 2026.04.10-09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1 +n_action_steps: + value: 1 +n_latency_steps: + value: 0 +n_obs_steps: + value: 1 +name: + value: train_llmbc_lowdim +obs_dim: + value: 39 +optimizer: + value: + _target_: torch.optim.AdamW + betas: + - 0.95 + - 0.999 + eps: 1e-08 + lr: 0.001 + weight_decay: 1e-06 +output_dir: + value: /workspace/LLM-BC/data/outputs/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1 +past_action_visible: + value: false +policy: + value: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + action_dim: 30 + horizon: 1 + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + action_dim: 30 + cfg: + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + finetune_mode: orig + hydra: + job: + override_dirname: HuggingFaceTB/SmolLM2-135M-Instruct + run: + dir: data/outputs/2026.04.10/09.42.47_HuggingFaceTB/SmolLM2-135M-Instruct + llm_mode: ete-finetuned + lora_config: + bias: none + lora_alpha: 64 + lora_dropout: 0.05 + r: 32 + task_type: CAUSAL_LM + max_length: 100 + model_name: SmolLM2-135M-Instruct + name: HuggingFaceTB/SmolLM2-135M-Instruct + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + use_joint_mlp_projector: true + use_quantization: false + horizon: 1 + n_action_steps: 1 + n_obs_steps: 1 + obs_dim: 39 + task_id: adroit-hand-relocate-v1 + loss_bc_weight: 1 + loss_llm_weight: 0.01 + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + activation: relu + hidden_size: + - 256 + - 256 + input_size: 39 + n_action_steps: 1 + n_obs_steps: 1 + output_size: 30 + n_action_steps: 1 + n_obs_steps: 1 + normalize_llm_loss: true + obs_dim: 39 +task: + value: + action_dim: 30 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-mix(csp100).pt + data_path2: datasets/adroit-hand-relocate-v1.pt + dummy_normalizer: true + horizon: 1 + obs_eef_target: true + pad_after: 0 + pad_before: 0 + use_manual_normalizer: false + val_ratio: 0.1 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + discount: 0.99 + env_name: llf-adroit-adroit-hand-relocate-v1 + feedback_type: + - hp + - hn + - fp + instruction_type: b + max_steps: 150 + n_action_steps: 1 + n_envs: 10 + n_obs_steps: 1 + n_test: 50 + n_train: 10 + visual: false + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor + name: adroit-hand-relocate-v1 + obs_dim: 39 +task_name: + value: adroit-hand-relocate-v1 +training: + value: + checkpoint_every: 5 + debug: false + device: cuda:0 + grad_norm_clip: 0.5 + gradient_accumulate_every: 8 + lr_scheduler: constant + lr_warmup_steps: 10 + max_train_steps: null + max_val_steps: null + num_epochs: 1001 + resume: false + rollout_every: 5 + sample_every: 5 + sample_max_batch: 128 + seed: 43 + tqdm_interval_sec: 1 + val_every: 1 +val_dataloader: + value: + batch_size: 16 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true diff --git a/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094422-68a5glbf/files/output.log b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094422-68a5glbf/files/output.log new file mode 100644 index 0000000000000000000000000000000000000000..6ae5595360152f3931cf9686ca2f54ef93088ad5 --- /dev/null +++ b/2026.04.10/09.42.47_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_094422-68a5glbf/files/output.log @@ -0,0 +1,3 @@ +Eval AdroitHandLowdimRunner 1/6: 0%| | 0/150 [00:00 + results, successes = zip(*[pipe.recv() for pipe in self.parent_pipes]) + File "/venv/llm-bc/lib/python3.9/multiprocessing/connection.py", line 250, in recv + buf = self._recv_bytes() + File "/venv/llm-bc/lib/python3.9/multiprocessing/connection.py", line 414, in _recv_bytes + buf = self._recv(4) + File "/venv/llm-bc/lib/python3.9/multiprocessing/connection.py", line 379, in _recv + chunk = read(handle, remaining) +KeyboardInterrupt + +During handling of the above exception, another exception occurred: + +Traceback (most recent call last): + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 283, in + main() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/main.py", line 90, in decorated_main + _run_hydra( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 389, in _run_hydra + _run_app( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 452, in _run_app + run_and_report( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 213, in run_and_report + return func() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 453, in + lambda: hydra.run( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/hydra.py", line 119, in run + ret = run_job( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/core/utils.py", line 186, in run_job + ret.return_value = task_function(task_cfg) + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 280, in main + workspace.run() + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 199, in run + runner_log = env_runner.run(policy) +KeyboardInterrupt +Traceback (most recent call last): + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 199, in run + runner_log = env_runner.run(policy) + File "/workspace/LLM-BC/llmbc/env_runner/pointmaze_lowdim_runner.py", line 157, in run + observation, reward, terminated, truncated, info = self.env.step(action) + File "/venv/llm-bc/lib/python3.9/site-packages/gymnasium/vector/vector_env.py", line 204, in step + return self.step_wait() + File "/workspace/LLM-BC/llmbc/gym_util/async_vector_env.py", line 298, in step_wait + results, successes = zip(*[pipe.recv() for pipe in self.parent_pipes]) + File "/workspace/LLM-BC/llmbc/gym_util/async_vector_env.py", line 298, in + results, successes = zip(*[pipe.recv() for pipe in self.parent_pipes]) + File "/venv/llm-bc/lib/python3.9/multiprocessing/connection.py", line 250, in recv + buf = self._recv_bytes() + File "/venv/llm-bc/lib/python3.9/multiprocessing/connection.py", line 414, in _recv_bytes + buf = self._recv(4) + File "/venv/llm-bc/lib/python3.9/multiprocessing/connection.py", line 379, in _recv + chunk = read(handle, remaining) +KeyboardInterrupt + +During handling of the above exception, another exception occurred: + +Traceback (most recent call last): + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 283, in + main() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/main.py", line 90, in decorated_main + _run_hydra( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 389, in _run_hydra + _run_app( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 452, in _run_app + run_and_report( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 213, in run_and_report + return func() + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/utils.py", line 453, in + lambda: hydra.run( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/_internal/hydra.py", line 119, in run + ret = run_job( + File "/venv/llm-bc/lib/python3.9/site-packages/hydra/core/utils.py", line 186, in run_job + ret.return_value = task_function(task_cfg) + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 280, in main + workspace.run() + File "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", line 199, in run + runner_log = env_runner.run(policy) +KeyboardInterrupt diff --git a/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/files/requirements.txt b/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/files/requirements.txt new file mode 100644 index 0000000000000000000000000000000000000000..2bfb2e62ba762bccdc9f965f019ed38cc8805d9c --- /dev/null +++ b/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/files/requirements.txt @@ -0,0 +1,864 @@ +aiohappyeyeballs==2.6.1 +antlr4-python3-runtime==4.9.3 +asciitree==0.3.3 +async-timeout==5.0.1 +attrs==25.4.0 +certifi==2025.10.5 +charset-normalizer==3.4.4 +colorama==0.4.6 +dill==0.3.8 +docstring-parser==0.16 +eval_type_backport==0.2.2 +fasteners==0.20 +filelock==3.17.0 +frozenlist==1.8.0 +fsspec==2024.3.1 +gmpy2==2.2.1 +hf-xet==1.1.8 +idna==3.11 +llvmlite==0.39.1 +MarkupSafe==3.0.2 +mdurl==0.1.2 +mkl-service==2.4.0 +mpmath==1.3.0 +msgpack==1.1.1 +networkx==3.2.1 +packaging==25.0 +propcache==0.3.1 +psutil==7.0.0 +pycparser==2.23 +Pygments==2.19.1 +PyOpenGL==3.1.10 +PySocks==1.7.1 +tzdata==2025.3 +xxhash==3.5.0 +pytz==2025.2 +PyYAML==6.0.2 +regex==2025.9.1 +safetensors==0.5.3 +setuptools==80.9.0 +shtab==1.7.2 +six==1.17.0 +tqdm==4.67.1 +tensorboard==2.15.2 +uvloop==0.22.1 +wheel==0.45.1 +zipp==3.21.0 +tensorflow-estimator==2.15.0 +tensorboard-data-server==0.7.2 +google-auth-oauthlib==1.3.1 +tensorflow==2.15.1 +Jinja2==3.1.6 +markdown-it-py==2.2.0 +multiprocess==0.70.15 +numpy==1.23.5 +pillow==11.3.0 +pip==23.3.2 +python-dateutil==2.9.0.post0 +aiosignal==1.4.0 +brotlicffi==1.0.9.2 +hydra-core==1.2.0 +multidict==6.7.0 +pycares==4.10.0 +rich==14.2.0 +sympy==1.14.0 +typeguard==4.4.4 +aiodns==3.5.0 +urllib3==2.5.0 +yarl==1.18.0 +aiohttp==3.12.15 +pyarrow==21.0.0 +pyarrow-hotfix==0.7 +tokenizers==0.21.0 +av==10.0.0 +einops==0.4.1 +Bottleneck==1.4.2 +mkl_fft==1.3.11 +mkl_random==1.2.8 +numba==0.56.4 +numcodecs==0.12.1 +numexpr==2.10.1 +pandas==2.3.3 +zarr==2.12.0 +datasets==2.19.0 +transformers==4.47.1 +accelerate==1.0.1 +peft==0.14.0 +trl==0.11.4 +torch==2.2.2 +tabulate==0.9.0 +sentencepiece==0.2.0 +pyperclip==1.11.0 +PyOpenGL-accelerate==3.1.10 +pydub==0.25.1 +py-cpuinfo==9.0.0 +pure_eval==0.2.3 +ptyprocess==0.7.0 +nvidia-ml-py==13.595.45 +hjson==3.1.0 +gym-notices==0.1.0 +glfw==2.0.0 +Farama-Notifications==0.0.4 +websockets==15.0.1 +wcwidth==0.6.0 +uc-micro-py==1.0.3 +typing-inspection==0.4.2 +triton==2.2.0 +trimesh==4.11.5 +transforms3d==0.4.2 +traitlets==5.14.3 +threadpoolctl==3.6.0 +tenacity==9.1.2 +sniffio==1.3.1 +smmap==5.0.3 +setproctitle==1.3.7 +sentry-sdk==2.57.0 +semantic-version==2.10.0 +scipy==1.13.1 +rpds-py==0.27.1 +python-multipart==0.0.20 +pyparsing==3.3.2 +pynvml==13.0.1 +pydantic_core==2.41.5 +pycryptodomex==3.23.0 +pyasn1==0.6.3 +grpcio==1.80.0 +platformdirs==4.4.0 +pexpect==4.9.0 +parso==0.8.6 +orjson==3.11.5 +opencv-python==4.11.0.86 +stack-data==0.6.3 +omegaconf==2.2.1 +nvidia-nvtx-cu12==12.1.105 +nvidia-nvjitlink-cu12==12.9.86 +nvidia-nccl-cu12==2.19.3 +nvidia-curand-cu12==10.3.2.106 +nvidia-cufft-cu12==11.0.2.54 +nvidia-cuda-runtime-cu12==12.1.105 +nvidia-cuda-nvrtc-cu12==12.1.105 +nvidia-cuda-cupti-cu12==12.1.105 +nvidia-cublas-cu12==12.1.3.1 +ninja==1.13.0 +narwhals==2.19.0 +lxml==6.0.3 +kiwisolver==1.4.7 +joblib==1.5.3 +jiter==0.13.0 +imageio-ffmpeg==0.6.0 +ImageIO==2.37.2 +h5py==3.14.0 +h11==0.16.0 +fonttools==4.60.2 +ffmpy==1.0.0 +fast_kinematics==0.2.2 +executing==2.2.1 +exceptiongroup==1.3.1 +docker-pycreds==0.4.0 +distro==1.9.0 +decorator==5.2.1 +dacite==1.9.2 +cycler==0.12.1 +contourpy==1.3.0 +cloudpickle==3.1.2 +click==8.1.8 +asttokens==3.0.1 +annotated-types==0.7.0 +annotated-doc==0.0.4 +aiofiles==25.1.0 +absl-py==2.3.1 +uvicorn==0.39.0 +tiktoken==0.8.0 +scikit-learn==1.6.1 +sapien==3.0.0b1 +referencing==0.36.2 +pydantic==2.12.5 +pyasn1_modules==0.4.2 +prompt_toolkit==3.0.52 +nvidia-cusparse-cu12==12.1.0.106 +nvidia-cudnn-cu12==8.9.2.26 +nltk==3.9.2 +mujoco==2.3.7 +mdit-py-plugins==0.3.3 +matplotlib-inline==0.2.1 +matplotlib==3.7.5 +linkify-it-py==2.0.3 +jedi==0.19.2 +huggingface-hub==0.26.2 +httpcore==1.0.9 +gymnasium==0.29.1 +google-pasta==0.2.0 +gitdb==4.0.12 +cryptography==46.0.7 +blobfile==3.0.0 +anyio==4.12.1 +tyro==0.9.1 +toppra==0.6.6 +starlette==0.49.3 +seaborn==0.13.2 +rouge_score==0.1.2 +pynndescent==0.6.0 +nvidia-cusolver-cu12==11.4.5.107 +jsonschema-specifications==2025.9.1 +ipython==8.18.1 +httpx==0.28.1 +gymnasium-robotics==1.2.0 +google-auth==2.49.2 +GitPython==3.1.46 +diffusers==0.31.0 +wandb==0.18.6 +umap-learn==0.5.9.post2 +requests-oauthlib==2.0.0 +openai==2.8.1 +mplib==0.1.1 +jsonschema==4.25.1 +gradio_client==0.2.9 +google-genai==1.47.0 +fastapi==0.128.8 +torchvision==0.17.2 +torchaudio==2.2.2 +sentence-transformers==3.2.1 +pytorch-seed==0.2.0 +deepspeed==0.16.1 +bitsandbytes==0.45.0 +altair==6.0.0 +timm==1.0.22 +gradio==3.36.1 +evaluate==0.4.3 +arm_pytorch_utilities==0.5.0 +tensorflow-io-gcs-filesystem==0.37.1 +Markdown==3.9 +pybullet==3.2.6 +wrapt==1.14.2 +tensorflow-probability==0.23.0 +parse==1.19.1 +dm-tree==0.1.8 +termcolor==3.1.0 +requests==2.32.0 +cffi==2.0.0 +opt_einsum==3.4.0 +labmaze==1.0.6 +typing_extensions==4.15.0 +importlib-resources==5.13.0 +importlib-metadata==5.2.0 +FLAML==2.3.6 +dm-env==1.6 +diskcache==5.6.3 +Cython==0.29.37 +mujoco-py==2.1.2.14 +jaxlib==0.4.30 +dm-control==1.0.14 +cmudict==1.0.13 +syllables==1.0.9 +metaworld==2.0.0 +jax==0.4.30 +highway-env==1.9.1 +gym_bandits==0.0.2 +d4rl==1.1 +pyautogen==0.1.0 +protobuf==4.25.9 +tf-agents==0.19.0 +oauthlib==3.3.1 +pytorch-kinematics==0.7.5 +pygame==2.6.1 +gin-config==0.5.0 +mani_skill==3.0.0b20 +tifffile==2024.8.30 +Shapely==1.8.4 +PyWavelets==1.6.0 +Werkzeug==3.1.8 +astunparse==1.6.3 +scikit-image==0.19.3 +pymunk==6.2.1 +ml-dtypes==0.3.2 +gast==0.7.0 +stable-baselines3==2.2.1 +keras==2.15.0 +libclang==18.1.1 +flatbuffers==25.12.19 +gym==0.23.1 +llmbc==0.0.0 +mjrl==1.0.0 +llfbench==0.1.0 +aiohappyeyeballs==2.6.1 +antlr4-python3-runtime==4.9.3 +asciitree==0.3.3 +async-timeout==5.0.1 +attrs==25.4.0 +certifi==2025.10.5 +charset-normalizer==3.4.4 +colorama==0.4.6 +dill==0.3.8 +docstring-parser==0.16 +eval_type_backport==0.2.2 +fasteners==0.20 +filelock==3.17.0 +frozenlist==1.8.0 +fsspec==2024.3.1 +gmpy2==2.2.1 +hf-xet==1.1.8 +idna==3.11 +llvmlite==0.39.1 +MarkupSafe==3.0.2 +mdurl==0.1.2 +mkl-service==2.4.0 +mpmath==1.3.0 +msgpack==1.1.1 +networkx==3.2.1 +packaging==25.0 +propcache==0.3.1 +psutil==7.0.0 +pycparser==2.23 +Pygments==2.19.1 +PyOpenGL==3.1.10 +PySocks==1.7.1 +tzdata==2025.3 +xxhash==3.5.0 +pytz==2025.2 +PyYAML==6.0.2 +regex==2025.9.1 +safetensors==0.5.3 +setuptools==80.9.0 +shtab==1.7.2 +six==1.17.0 +tqdm==4.67.1 +tensorboard==2.15.2 +uvloop==0.22.1 +wheel==0.45.1 +zipp==3.21.0 +tensorflow-estimator==2.15.0 +tensorboard-data-server==0.7.2 +google-auth-oauthlib==1.3.1 +tensorflow==2.15.1 +Jinja2==3.1.6 +markdown-it-py==2.2.0 +multiprocess==0.70.15 +numpy==1.23.5 +pillow==11.3.0 +pip==23.3.2 +python-dateutil==2.9.0.post0 +aiosignal==1.4.0 +brotlicffi==1.0.9.2 +hydra-core==1.2.0 +multidict==6.7.0 +pycares==4.10.0 +rich==14.2.0 +sympy==1.14.0 +typeguard==4.4.4 +aiodns==3.5.0 +urllib3==2.5.0 +yarl==1.18.0 +aiohttp==3.12.15 +pyarrow==21.0.0 +pyarrow-hotfix==0.7 +tokenizers==0.21.0 +av==10.0.0 +einops==0.4.1 +Bottleneck==1.4.2 +mkl_fft==1.3.11 +mkl_random==1.2.8 +numba==0.56.4 +numcodecs==0.12.1 +numexpr==2.10.1 +pandas==2.3.3 +zarr==2.12.0 +datasets==2.19.0 +transformers==4.47.1 +accelerate==1.0.1 +peft==0.14.0 +trl==0.11.4 +torch==2.2.2 +tabulate==0.9.0 +sentencepiece==0.2.0 +pyperclip==1.11.0 +PyOpenGL-accelerate==3.1.10 +pydub==0.25.1 +py-cpuinfo==9.0.0 +pure_eval==0.2.3 +ptyprocess==0.7.0 +nvidia-ml-py==13.595.45 +hjson==3.1.0 +gym-notices==0.1.0 +glfw==2.0.0 +Farama-Notifications==0.0.4 +websockets==15.0.1 +wcwidth==0.6.0 +uc-micro-py==1.0.3 +typing-inspection==0.4.2 +triton==2.2.0 +trimesh==4.11.5 +transforms3d==0.4.2 +traitlets==5.14.3 +threadpoolctl==3.6.0 +tenacity==9.1.2 +sniffio==1.3.1 +smmap==5.0.3 +setproctitle==1.3.7 +sentry-sdk==2.57.0 +semantic-version==2.10.0 +scipy==1.13.1 +rpds-py==0.27.1 +python-multipart==0.0.20 +pyparsing==3.3.2 +pynvml==13.0.1 +pydantic_core==2.41.5 +pycryptodomex==3.23.0 +pyasn1==0.6.3 +grpcio==1.80.0 +platformdirs==4.4.0 +pexpect==4.9.0 +parso==0.8.6 +orjson==3.11.5 +opencv-python==4.11.0.86 +stack-data==0.6.3 +omegaconf==2.2.1 +nvidia-nvtx-cu12==12.1.105 +nvidia-nvjitlink-cu12==12.9.86 +nvidia-nccl-cu12==2.19.3 +nvidia-curand-cu12==10.3.2.106 +nvidia-cufft-cu12==11.0.2.54 +nvidia-cuda-runtime-cu12==12.1.105 +nvidia-cuda-nvrtc-cu12==12.1.105 +nvidia-cuda-cupti-cu12==12.1.105 +nvidia-cublas-cu12==12.1.3.1 +ninja==1.13.0 +narwhals==2.19.0 +lxml==6.0.3 +kiwisolver==1.4.7 +joblib==1.5.3 +jiter==0.13.0 +imageio-ffmpeg==0.6.0 +ImageIO==2.37.2 +h5py==3.14.0 +h11==0.16.0 +fonttools==4.60.2 +ffmpy==1.0.0 +fast_kinematics==0.2.2 +executing==2.2.1 +exceptiongroup==1.3.1 +docker-pycreds==0.4.0 +distro==1.9.0 +decorator==5.2.1 +dacite==1.9.2 +cycler==0.12.1 +contourpy==1.3.0 +cloudpickle==3.1.2 +click==8.1.8 +asttokens==3.0.1 +annotated-types==0.7.0 +annotated-doc==0.0.4 +aiofiles==25.1.0 +absl-py==2.3.1 +uvicorn==0.39.0 +tiktoken==0.8.0 +scikit-learn==1.6.1 +sapien==3.0.0b1 +referencing==0.36.2 +pydantic==2.12.5 +pyasn1_modules==0.4.2 +prompt_toolkit==3.0.52 +nvidia-cusparse-cu12==12.1.0.106 +nvidia-cudnn-cu12==8.9.2.26 +nltk==3.9.2 +mujoco==2.3.7 +mdit-py-plugins==0.3.3 +matplotlib-inline==0.2.1 +matplotlib==3.7.5 +linkify-it-py==2.0.3 +jedi==0.19.2 +huggingface-hub==0.26.2 +httpcore==1.0.9 +gymnasium==0.29.1 +google-pasta==0.2.0 +gitdb==4.0.12 +cryptography==46.0.7 +blobfile==3.0.0 +anyio==4.12.1 +tyro==0.9.1 +toppra==0.6.6 +starlette==0.49.3 +seaborn==0.13.2 +rouge_score==0.1.2 +pynndescent==0.6.0 +nvidia-cusolver-cu12==11.4.5.107 +jsonschema-specifications==2025.9.1 +ipython==8.18.1 +httpx==0.28.1 +gymnasium-robotics==1.2.0 +google-auth==2.49.2 +GitPython==3.1.46 +diffusers==0.31.0 +wandb==0.18.6 +umap-learn==0.5.9.post2 +requests-oauthlib==2.0.0 +openai==2.8.1 +mplib==0.1.1 +jsonschema==4.25.1 +gradio_client==0.2.9 +google-genai==1.47.0 +fastapi==0.128.8 +torchvision==0.17.2 +torchaudio==2.2.2 +sentence-transformers==3.2.1 +pytorch-seed==0.2.0 +deepspeed==0.16.1 +bitsandbytes==0.45.0 +altair==6.0.0 +timm==1.0.22 +gradio==3.36.1 +evaluate==0.4.3 +arm_pytorch_utilities==0.5.0 +tensorflow-io-gcs-filesystem==0.37.1 +Markdown==3.9 +pybullet==3.2.6 +wrapt==1.14.2 +tensorflow-probability==0.23.0 +parse==1.19.1 +dm-tree==0.1.8 +termcolor==3.1.0 +requests==2.32.0 +cffi==2.0.0 +opt_einsum==3.4.0 +labmaze==1.0.6 +typing_extensions==4.15.0 +importlib-resources==5.13.0 +importlib-metadata==5.2.0 +FLAML==2.3.6 +dm-env==1.6 +diskcache==5.6.3 +Cython==0.29.37 +mujoco-py==2.1.2.14 +jaxlib==0.4.30 +dm-control==1.0.14 +cmudict==1.0.13 +syllables==1.0.9 +metaworld==2.0.0 +jax==0.4.30 +highway-env==1.9.1 +gym_bandits==0.0.2 +d4rl==1.1 +pyautogen==0.1.0 +protobuf==4.25.9 +tf-agents==0.19.0 +oauthlib==3.3.1 +pytorch-kinematics==0.7.5 +pygame==2.6.1 +gin-config==0.5.0 +mani_skill==3.0.0b20 +tifffile==2024.8.30 +Shapely==1.8.4 +PyWavelets==1.6.0 +Werkzeug==3.1.8 +astunparse==1.6.3 +scikit-image==0.19.3 +pymunk==6.2.1 +ml-dtypes==0.3.2 +gast==0.7.0 +stable-baselines3==2.2.1 +keras==2.15.0 +libclang==18.1.1 +flatbuffers==25.12.19 +gym==0.23.1 +llmbc==0.0.0 +mjrl==1.0.0 +llfbench==0.1.0 +llmbc==0.0.0 +aiohappyeyeballs==2.6.1 +antlr4-python3-runtime==4.9.3 +asciitree==0.3.3 +async-timeout==5.0.1 +attrs==25.4.0 +certifi==2025.10.5 +charset-normalizer==3.4.4 +colorama==0.4.6 +dill==0.3.8 +docstring-parser==0.16 +eval_type_backport==0.2.2 +fasteners==0.20 +filelock==3.17.0 +frozenlist==1.8.0 +fsspec==2024.3.1 +gmpy2==2.2.1 +hf-xet==1.1.8 +idna==3.11 +llvmlite==0.39.1 +MarkupSafe==3.0.2 +mdurl==0.1.2 +mkl-service==2.4.0 +mpmath==1.3.0 +msgpack==1.1.1 +networkx==3.2.1 +packaging==25.0 +propcache==0.3.1 +psutil==7.0.0 +pycparser==2.23 +Pygments==2.19.1 +PyOpenGL==3.1.10 +PySocks==1.7.1 +tzdata==2025.3 +xxhash==3.5.0 +pytz==2025.2 +PyYAML==6.0.2 +regex==2025.9.1 +safetensors==0.5.3 +setuptools==80.9.0 +shtab==1.7.2 +six==1.17.0 +tqdm==4.67.1 +tensorboard==2.15.2 +uvloop==0.22.1 +wheel==0.45.1 +zipp==3.21.0 +tensorflow-estimator==2.15.0 +tensorboard-data-server==0.7.2 +google-auth-oauthlib==1.3.1 +tensorflow==2.15.1 +Jinja2==3.1.6 +markdown-it-py==2.2.0 +multiprocess==0.70.15 +numpy==1.23.5 +pillow==11.3.0 +pip==23.3.2 +python-dateutil==2.9.0.post0 +aiosignal==1.4.0 +brotlicffi==1.0.9.2 +hydra-core==1.2.0 +multidict==6.7.0 +pycares==4.10.0 +rich==14.2.0 +sympy==1.14.0 +typeguard==4.4.4 +aiodns==3.5.0 +urllib3==2.5.0 +yarl==1.18.0 +aiohttp==3.12.15 +pyarrow==21.0.0 +pyarrow-hotfix==0.7 +tokenizers==0.21.0 +av==10.0.0 +einops==0.4.1 +Bottleneck==1.4.2 +mkl_fft==1.3.11 +mkl_random==1.2.8 +numba==0.56.4 +numcodecs==0.12.1 +numexpr==2.10.1 +pandas==2.3.3 +zarr==2.12.0 +datasets==2.19.0 +transformers==4.47.1 +accelerate==1.0.1 +peft==0.14.0 +trl==0.11.4 +torch==2.2.2 +tabulate==0.9.0 +sentencepiece==0.2.0 +pyperclip==1.11.0 +PyOpenGL-accelerate==3.1.10 +pydub==0.25.1 +py-cpuinfo==9.0.0 +pure_eval==0.2.3 +ptyprocess==0.7.0 +nvidia-ml-py==13.595.45 +hjson==3.1.0 +gym-notices==0.1.0 +glfw==2.0.0 +Farama-Notifications==0.0.4 +websockets==15.0.1 +wcwidth==0.6.0 +uc-micro-py==1.0.3 +typing-inspection==0.4.2 +triton==2.2.0 +trimesh==4.11.5 +transforms3d==0.4.2 +traitlets==5.14.3 +threadpoolctl==3.6.0 +tenacity==9.1.2 +sniffio==1.3.1 +smmap==5.0.3 +setproctitle==1.3.7 +sentry-sdk==2.57.0 +semantic-version==2.10.0 +scipy==1.13.1 +rpds-py==0.27.1 +python-multipart==0.0.20 +pyparsing==3.3.2 +pynvml==13.0.1 +pydantic_core==2.41.5 +pycryptodomex==3.23.0 +pyasn1==0.6.3 +grpcio==1.80.0 +platformdirs==4.4.0 +pexpect==4.9.0 +parso==0.8.6 +orjson==3.11.5 +opencv-python==4.11.0.86 +stack-data==0.6.3 +omegaconf==2.2.1 +nvidia-nvtx-cu12==12.1.105 +nvidia-nvjitlink-cu12==12.9.86 +nvidia-nccl-cu12==2.19.3 +nvidia-curand-cu12==10.3.2.106 +nvidia-cufft-cu12==11.0.2.54 +nvidia-cuda-runtime-cu12==12.1.105 +nvidia-cuda-nvrtc-cu12==12.1.105 +nvidia-cuda-cupti-cu12==12.1.105 +nvidia-cublas-cu12==12.1.3.1 +ninja==1.13.0 +narwhals==2.19.0 +lxml==6.0.3 +kiwisolver==1.4.7 +joblib==1.5.3 +jiter==0.13.0 +imageio-ffmpeg==0.6.0 +ImageIO==2.37.2 +h5py==3.14.0 +h11==0.16.0 +fonttools==4.60.2 +ffmpy==1.0.0 +fast_kinematics==0.2.2 +executing==2.2.1 +exceptiongroup==1.3.1 +docker-pycreds==0.4.0 +distro==1.9.0 +decorator==5.2.1 +dacite==1.9.2 +cycler==0.12.1 +contourpy==1.3.0 +cloudpickle==3.1.2 +click==8.1.8 +asttokens==3.0.1 +annotated-types==0.7.0 +annotated-doc==0.0.4 +aiofiles==25.1.0 +absl-py==2.3.1 +uvicorn==0.39.0 +tiktoken==0.8.0 +scikit-learn==1.6.1 +sapien==3.0.0b1 +referencing==0.36.2 +pydantic==2.12.5 +pyasn1_modules==0.4.2 +prompt_toolkit==3.0.52 +nvidia-cusparse-cu12==12.1.0.106 +nvidia-cudnn-cu12==8.9.2.26 +nltk==3.9.2 +mujoco==2.3.7 +mdit-py-plugins==0.3.3 +matplotlib-inline==0.2.1 +matplotlib==3.7.5 +linkify-it-py==2.0.3 +jedi==0.19.2 +huggingface-hub==0.26.2 +httpcore==1.0.9 +gymnasium==0.29.1 +google-pasta==0.2.0 +gitdb==4.0.12 +cryptography==46.0.7 +blobfile==3.0.0 +anyio==4.12.1 +tyro==0.9.1 +toppra==0.6.6 +starlette==0.49.3 +seaborn==0.13.2 +rouge_score==0.1.2 +pynndescent==0.6.0 +nvidia-cusolver-cu12==11.4.5.107 +jsonschema-specifications==2025.9.1 +ipython==8.18.1 +httpx==0.28.1 +gymnasium-robotics==1.2.0 +google-auth==2.49.2 +GitPython==3.1.46 +diffusers==0.31.0 +wandb==0.18.6 +umap-learn==0.5.9.post2 +requests-oauthlib==2.0.0 +openai==2.8.1 +mplib==0.1.1 +jsonschema==4.25.1 +gradio_client==0.2.9 +google-genai==1.47.0 +fastapi==0.128.8 +torchvision==0.17.2 +torchaudio==2.2.2 +sentence-transformers==3.2.1 +pytorch-seed==0.2.0 +deepspeed==0.16.1 +bitsandbytes==0.45.0 +altair==6.0.0 +timm==1.0.22 +gradio==3.36.1 +evaluate==0.4.3 +arm_pytorch_utilities==0.5.0 +tensorflow-io-gcs-filesystem==0.37.1 +Markdown==3.9 +pybullet==3.2.6 +wrapt==1.14.2 +tensorflow-probability==0.23.0 +parse==1.19.1 +dm-tree==0.1.8 +termcolor==3.1.0 +requests==2.32.0 +cffi==2.0.0 +opt_einsum==3.4.0 +labmaze==1.0.6 +typing_extensions==4.15.0 +importlib-resources==5.13.0 +importlib-metadata==5.2.0 +FLAML==2.3.6 +dm-env==1.6 +diskcache==5.6.3 +Cython==0.29.37 +mujoco-py==2.1.2.14 +jaxlib==0.4.30 +dm-control==1.0.14 +cmudict==1.0.13 +syllables==1.0.9 +metaworld==2.0.0 +jax==0.4.30 +highway-env==1.9.1 +gym_bandits==0.0.2 +d4rl==1.1 +pyautogen==0.1.0 +protobuf==4.25.9 +tf-agents==0.19.0 +oauthlib==3.3.1 +pytorch-kinematics==0.7.5 +pygame==2.6.1 +gin-config==0.5.0 +mani_skill==3.0.0b20 +tifffile==2024.8.30 +Shapely==1.8.4 +PyWavelets==1.6.0 +Werkzeug==3.1.8 +astunparse==1.6.3 +scikit-image==0.19.3 +pymunk==6.2.1 +ml-dtypes==0.3.2 +gast==0.7.0 +stable-baselines3==2.2.1 +keras==2.15.0 +libclang==18.1.1 +flatbuffers==25.12.19 +gym==0.23.1 +llmbc==0.0.0 +mjrl==1.0.0 +llfbench==0.1.0 +llmbc==0.0.0 +autocommand==2.2.2 +backports.tarfile==1.2.0 +importlib_metadata==8.0.0 +inflect==7.3.1 +jaraco.collections==5.1.0 +jaraco.context==5.3.0 +jaraco.functools==4.0.1 +jaraco.text==3.12.1 +more-itertools==10.3.0 +packaging==24.2 +platformdirs==4.2.2 +tomli==2.0.1 +typeguard==4.3.0 +typing_extensions==4.12.2 +wheel==0.45.1 +zipp==3.19.2 diff --git a/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/files/wandb-metadata.json b/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/files/wandb-metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..9160ae05d80b76f18aae23edbfc08f7db64dcb40 --- /dev/null +++ b/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/files/wandb-metadata.json @@ -0,0 +1,49 @@ +{ + "os": "Linux-5.15.0-168-generic-x86_64-with-glibc2.39", + "python": "3.9.25", + "startedAt": "2026-04-10T09:55:19.313680Z", + "program": "/workspace/LLM-BC/./llmbc/workspace/train_bc_lowdim_workspace.py", + "codePath": "llmbc/workspace/train_bc_lowdim_workspace.py", + "git": { + "remote": "https://github.com/CHYang25/LLM-BC.git", + "commit": "2fa092f9eb4c2ad9435919c661a618071a583be5" + }, + "email": "chris920325@gmail.com", + "root": "/workspace/LLM-BC/data/outputs/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0", + "host": "c8ef52649900", + "username": "root", + "executable": "/venv/llm-bc/bin/python3", + "codePathLocal": "llmbc/workspace/train_bc_lowdim_workspace.py", + "cpu_count": 72, + "cpu_count_logical": 144, + "gpu": "NVIDIA GeForce RTX 4090", + "gpu_count": 2, + "disk": { + "/": { + "total": "545460846592", + "used": "348699824128" + } + }, + "memory": { + "total": "810942898176" + }, + "cpu": { + "count": 72, + "countLogical": 144 + }, + "gpu_nvidia": [ + { + "name": "NVIDIA GeForce RTX 4090", + "memoryTotal": "25757220864", + "cudaCores": 16384, + "architecture": "Ada" + }, + { + "name": "NVIDIA GeForce RTX 4090", + "memoryTotal": "25757220864", + "cudaCores": 16384, + "architecture": "Ada" + } + ], + "cudaVersion": "12.8" +} \ No newline at end of file diff --git a/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/files/wandb-summary.json b/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/files/wandb-summary.json new file mode 100644 index 0000000000000000000000000000000000000000..ed448093136a79aa01b048ebf2ff36972219b599 --- /dev/null +++ b/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/files/wandb-summary.json @@ -0,0 +1 @@ +{"_step":7162,"test/success_rate":0.34,"train/success_rate":0.5,"val_loss":0.16037246584892273,"epoch":35,"lr":0.00996821480892109,"_timestamp":1.775816016129702e+09,"train_loss_bc":0.07907605171203613,"train/mean_score":0.8,"train_loss":0.07907605171203613,"test/mean_score":0.74,"global_step":7162,"train/cumulative_reward":7.082814626201153,"test/cumulative_reward":5.784717055851158,"_wandb":{"runtime":1127},"_runtime":1127.124224606,"grad_norm":0.34789955615997314,"train_action_mse_error":0.07420593500137329} \ No newline at end of file diff --git a/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/run-3qme4js3.wandb b/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/run-3qme4js3.wandb new file mode 100644 index 0000000000000000000000000000000000000000..3cdc2425e7e26448edc071db509a3a2ed173ea9f --- /dev/null +++ b/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/run-20260410_095519-3qme4js3/run-3qme4js3.wandb @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6657e812a28ed696cb20204dce5faebd75e78841b2a88399a7ddb252a8cac10a +size 5570560 diff --git a/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/wandb-resume.json b/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/wandb-resume.json new file mode 100644 index 0000000000000000000000000000000000000000..99b3c1ccd11741bf42c814ce605481e99114e218 --- /dev/null +++ b/2026.04.10/09.46.35_train_bc_lowdim_maze2d-medium-v0/wandb/wandb-resume.json @@ -0,0 +1 @@ +{"run_id": "3qme4js3"} \ No newline at end of file diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/.hydra/config.yaml b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..aa023105f1e5fd07b23aaf2d9cc2a7b74bc7a591 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/.hydra/config.yaml @@ -0,0 +1,112 @@ +name: train_rbc_reward_model +_target_: llmbc.workspace.train_rbc_reward_workspace.TrainRBCRewardWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +reward_model: + _target_: llmbc.policy.rbc_lowdim_policy.RBCLowdimRewardModel + input_size: ${eval:'${n_obs_steps}*${obs_dim}+${n_action_steps}*${action_dim}'} + hidden_size: + - 128 + obs_dim: ${obs_dim} + action_dim: ${action_dim} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} +dataloader: + batch_size: 256 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 256 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.Adam + lr: 0.0001 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda + seed: 42 + debug: false + resume: true + lr_scheduler: cosine + lr_warmup_steps: 0 + num_epochs: 500 + gradient_accumulate_every: 1 + grad_norm_clip: 0.5 + checkpoint_every: 10 + val_every: 1 + sample_every: 10 + sample_max_batch: 1000 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: rbc_module_training + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: val_loss + mode: min + k: 5 + format_str: epoch={epoch:04d}-val_loss={val_loss:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: adroit-hand-relocate-v1 + obs_dim: 39 + action_dim: 30 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + env_name: llf-adroit-adroit-hand-relocate-v1 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 150 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-general-reward.pt + data_path2: datasets/adroit-hand-relocate-v1-reward.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.1 + dummy_normalizer: true + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/.hydra/hydra.yaml b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b9daf1283017f9b9cab7ec50c79d00369965a245 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train_rbc_reward_workspace + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: train_rbc_reward_workspace + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /tmp2/chyang/workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /tmp2/chyang/workspace/LLM-BC/llmbc/config + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1 + choices: + task: adroit-hand-relocate-v1 + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/.hydra/overrides.yaml b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0060-val_loss=0.019.ckpt b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0060-val_loss=0.019.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..7bf5a7ee2c1ecec1ca4ffa854ec2cbffec599126 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0060-val_loss=0.019.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:58a809e5dc076c74f274cf9e9b747c65f85c028742f04aea5c2f5d00acea58c0 +size 129952 diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0070-val_loss=0.019.ckpt b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0070-val_loss=0.019.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..958aa52ab92a2137bb4c2e94a06c0e1f1acef79d --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0070-val_loss=0.019.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3bf71e3e1b87797655a0ff182bc92d881817662287c9dae3e861132bd4daabac +size 129952 diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0090-val_loss=0.019.ckpt b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0090-val_loss=0.019.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..1b4eecb2655465f4572055254529a52dfd6f00c7 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0090-val_loss=0.019.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:52a07c0383607a2a846ff12577ed92b26ad9fd8d0a8eb160f28511677bd33359 +size 129952 diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0100-val_loss=0.019.ckpt b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0100-val_loss=0.019.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..e0c9f0baedc8d38ed1d15168ab78f20997b09a10 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0100-val_loss=0.019.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c2685b627ff23b4513ad74e3c4718b7df781b3f2cec3d9a316293e95dc2fcb79 +size 129952 diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0140-val_loss=0.019.ckpt b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0140-val_loss=0.019.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..e623bbbb0244885ed1d7ccd3e4b65a5ad2e00754 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/epoch=0140-val_loss=0.019.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e589658c574bbcff36105b1bcd527088a33dd11bb272ea0312ab29213a9fee45 +size 129952 diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/latest.ckpt b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/latest.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..842d1b020a4065714c2a9390ff2b8a7ab9f6063b --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/checkpoints/latest.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e75dfb2edf91b21634ab7d2e1718be860a7d1091a8464ea6b9e3b40f064d14f1 +size 129952 diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/logs.json.txt b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/logs.json.txt new file mode 100644 index 0000000000000000000000000000000000000000..ee9afbd199bd2351d312c117a023db4940672fdb --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/logs.json.txt @@ -0,0 +1,53000 @@ +{"train_loss": 0.20690089464187622, "grad_norm": 2.102705717086792, "global_step": 0, "epoch": 0, "lr": 9.999999991216087e-05} +{"train_loss": 0.1948395073413849, "grad_norm": 2.008852958679199, "global_step": 1, "epoch": 0, "lr": 9.999999964864349e-05} +{"train_loss": 0.19189071655273438, "grad_norm": 1.9625327587127686, "global_step": 2, "epoch": 0, "lr": 9.999999920944786e-05} +{"train_loss": 0.18709224462509155, "grad_norm": 1.9426323175430298, "global_step": 3, "epoch": 0, "lr": 9.999999859457396e-05} +{"train_loss": 0.19331859052181244, "grad_norm": 2.0110065937042236, "global_step": 4, "epoch": 0, "lr": 9.999999780402182e-05} +{"train_loss": 0.176039457321167, "grad_norm": 1.8457386493682861, "global_step": 5, "epoch": 0, "lr": 9.999999683779145e-05} +{"train_loss": 0.190693199634552, "grad_norm": 1.9572184085845947, "global_step": 6, "epoch": 0, "lr": 9.99999956958828e-05} +{"train_loss": 0.1541934609413147, "grad_norm": 1.7434414625167847, "global_step": 7, "epoch": 0, "lr": 9.999999437829592e-05} +{"train_loss": 0.15493151545524597, "grad_norm": 1.7817802429199219, "global_step": 8, "epoch": 0, "lr": 9.999999288503081e-05} +{"train_loss": 0.1592177301645279, "grad_norm": 1.7421391010284424, "global_step": 9, "epoch": 0, "lr": 9.999999121608747e-05} +{"train_loss": 0.15629956126213074, "grad_norm": 1.720711350440979, "global_step": 10, "epoch": 0, "lr": 9.999998937146591e-05} +{"train_loss": 0.1439814567565918, "grad_norm": 1.663385272026062, "global_step": 11, "epoch": 0, "lr": 9.999998735116613e-05} +{"train_loss": 0.1470830887556076, "grad_norm": 1.6876306533813477, "global_step": 12, "epoch": 0, "lr": 9.999998515518814e-05} +{"train_loss": 0.13616374135017395, "grad_norm": 1.556946039199829, "global_step": 13, "epoch": 0, "lr": 9.999998278353194e-05} +{"train_loss": 0.15143591165542603, "grad_norm": 1.711565613746643, "global_step": 14, "epoch": 0, "lr": 9.999998023619754e-05} +{"train_loss": 0.13607212901115417, "grad_norm": 1.6124807596206665, "global_step": 15, "epoch": 0, "lr": 9.999997751318497e-05} +{"train_loss": 0.1235513985157013, "grad_norm": 1.4824150800704956, "global_step": 16, "epoch": 0, "lr": 9.999997461449421e-05} +{"train_loss": 0.11342135071754456, "grad_norm": 1.4401867389678955, "global_step": 17, "epoch": 0, "lr": 9.999997154012529e-05} +{"train_loss": 0.12092997878789902, "grad_norm": 1.4769949913024902, "global_step": 18, "epoch": 0, "lr": 9.999996829007822e-05} +{"train_loss": 0.1107790619134903, "grad_norm": 1.3802534341812134, "global_step": 19, "epoch": 0, "lr": 9.9999964864353e-05} +{"train_loss": 0.11148479580879211, "grad_norm": 1.4066178798675537, "global_step": 20, "epoch": 0, "lr": 9.999996126294964e-05} +{"train_loss": 0.11097148060798645, "grad_norm": 1.3715944290161133, "global_step": 21, "epoch": 0, "lr": 9.999995748586817e-05} +{"train_loss": 0.09184426069259644, "grad_norm": 1.1498712301254272, "global_step": 22, "epoch": 0, "lr": 9.999995353310858e-05} +{"train_loss": 0.09512749314308167, "grad_norm": 1.2204712629318237, "global_step": 23, "epoch": 0, "lr": 9.999994940467092e-05} +{"train_loss": 0.0928386002779007, "grad_norm": 1.2465025186538696, "global_step": 24, "epoch": 0, "lr": 9.999994510055516e-05} +{"train_loss": 0.08863574266433716, "grad_norm": 1.1555383205413818, "global_step": 25, "epoch": 0, "lr": 9.999994062076135e-05} +{"train_loss": 0.0804978683590889, "grad_norm": 1.0593129396438599, "global_step": 26, "epoch": 0, "lr": 9.999993596528949e-05} +{"train_loss": 0.08887752145528793, "grad_norm": 1.1375396251678467, "global_step": 27, "epoch": 0, "lr": 9.999993113413962e-05} +{"train_loss": 0.0764143243432045, "grad_norm": 0.9798145294189453, "global_step": 28, "epoch": 0, "lr": 9.999992612731171e-05} +{"train_loss": 0.07680009305477142, "grad_norm": 1.042072057723999, "global_step": 29, "epoch": 0, "lr": 9.99999209448058e-05} +{"train_loss": 0.07033389061689377, "grad_norm": 0.901203453540802, "global_step": 30, "epoch": 0, "lr": 9.999991558662193e-05} +{"train_loss": 0.08599089086055756, "grad_norm": 1.0778385400772095, "global_step": 31, "epoch": 0, "lr": 9.99999100527601e-05} +{"train_loss": 0.06678328663110733, "grad_norm": 0.8999868035316467, "global_step": 32, "epoch": 0, "lr": 9.999990434322032e-05} +{"train_loss": 0.061068594455718994, "grad_norm": 0.85400390625, "global_step": 33, "epoch": 0, "lr": 9.999989845800262e-05} +{"train_loss": 0.07444320619106293, "grad_norm": 0.9256008267402649, "global_step": 34, "epoch": 0, "lr": 9.999989239710703e-05} +{"train_loss": 0.05240153521299362, "grad_norm": 0.6938344836235046, "global_step": 35, "epoch": 0, "lr": 9.999988616053355e-05} +{"train_loss": 0.058511883020401, "grad_norm": 0.7514157295227051, "global_step": 36, "epoch": 0, "lr": 9.999987974828223e-05} +{"train_loss": 0.057643283158540726, "grad_norm": 0.7036154270172119, "global_step": 37, "epoch": 0, "lr": 9.999987316035306e-05} +{"train_loss": 0.059493519365787506, "grad_norm": 0.6432070136070251, "global_step": 38, "epoch": 0, "lr": 9.99998663967461e-05} +{"train_loss": 0.06088825315237045, "grad_norm": 0.6194462776184082, "global_step": 39, "epoch": 0, "lr": 9.999985945746134e-05} +{"train_loss": 0.056644219905138016, "grad_norm": 0.5949135422706604, "global_step": 40, "epoch": 0, "lr": 9.999985234249883e-05} +{"train_loss": 0.053215496242046356, "grad_norm": 0.5646234154701233, "global_step": 41, "epoch": 0, "lr": 9.999984505185857e-05} +{"train_loss": 0.04595933482050896, "grad_norm": 0.41667309403419495, "global_step": 42, "epoch": 0, "lr": 9.999983758554061e-05} +{"train_loss": 0.0429154634475708, "grad_norm": 0.3911772668361664, "global_step": 43, "epoch": 0, "lr": 9.999982994354495e-05} +{"train_loss": 0.042363401502370834, "grad_norm": 0.41747280955314636, "global_step": 44, "epoch": 0, "lr": 9.999982212587164e-05} +{"train_loss": 0.0414576455950737, "grad_norm": 0.3797820210456848, "global_step": 45, "epoch": 0, "lr": 9.99998141325207e-05} +{"train_loss": 0.0451238714158535, "grad_norm": 0.45853304862976074, "global_step": 46, "epoch": 0, "lr": 9.999980596349217e-05} +{"train_loss": 0.04941907152533531, "grad_norm": 0.3574184477329254, "global_step": 47, "epoch": 0, "lr": 9.999979761878604e-05} +{"train_loss": 0.042682550847530365, "grad_norm": 0.21679574251174927, "global_step": 48, "epoch": 0, "lr": 9.99997890984024e-05} +{"train_loss": 0.042043957859277725, "grad_norm": 0.3733653128147125, "global_step": 49, "epoch": 0, "lr": 9.999978040234121e-05} +{"train_loss": 0.044052623212337494, "grad_norm": 0.19809012115001678, "global_step": 50, "epoch": 0, "lr": 9.999977153060256e-05} +{"train_loss": 0.0433746799826622, "grad_norm": 0.20370027422904968, "global_step": 51, "epoch": 0, "lr": 9.999976248318644e-05} +{"train_loss": 0.04225756973028183, "grad_norm": 0.14347293972969055, "global_step": 52, "epoch": 0, "lr": 9.999975326009292e-05} +{"train_loss": 0.03826380521059036, "grad_norm": 0.17007143795490265, "global_step": 53, "epoch": 0, "lr": 9.999974386132199e-05} +{"train_loss": 0.044900137931108475, "grad_norm": 0.1561371386051178, "global_step": 54, "epoch": 0, "lr": 9.999973428687371e-05} +{"train_loss": 0.04411432147026062, "grad_norm": 0.1352076232433319, "global_step": 55, "epoch": 0, "lr": 9.999972453674812e-05} +{"train_loss": 0.04392673447728157, "grad_norm": 0.13720746338367462, "global_step": 56, "epoch": 0, "lr": 9.999971461094522e-05} +{"train_loss": 0.03908829018473625, "grad_norm": 0.1663181036710739, "global_step": 57, "epoch": 0, "lr": 9.999970450946509e-05} +{"train_loss": 0.03880184143781662, "grad_norm": 0.1661773920059204, "global_step": 58, "epoch": 0, "lr": 9.999969423230772e-05} +{"train_loss": 0.03885580971837044, "grad_norm": 0.11981502920389175, "global_step": 59, "epoch": 0, "lr": 9.999968377947319e-05} +{"train_loss": 0.038557883352041245, "grad_norm": 0.15895384550094604, "global_step": 60, "epoch": 0, "lr": 9.999967315096152e-05} +{"train_loss": 0.04067794606089592, "grad_norm": 0.240568608045578, "global_step": 61, "epoch": 0, "lr": 9.999966234677271e-05} +{"train_loss": 0.03875764086842537, "grad_norm": 0.1550970822572708, "global_step": 62, "epoch": 0, "lr": 9.999965136690687e-05} +{"train_loss": 0.04087049514055252, "grad_norm": 0.3180786371231079, "global_step": 63, "epoch": 0, "lr": 9.999964021136397e-05} +{"train_loss": 0.039885278791189194, "grad_norm": 0.20441284775733948, "global_step": 64, "epoch": 0, "lr": 9.999962888014409e-05} +{"train_loss": 0.0441952720284462, "grad_norm": 0.2339756190776825, "global_step": 65, "epoch": 0, "lr": 9.999961737324725e-05} +{"train_loss": 0.04217604547739029, "grad_norm": 0.24067340791225433, "global_step": 66, "epoch": 0, "lr": 9.999960569067352e-05} +{"train_loss": 0.04069565609097481, "grad_norm": 0.34394797682762146, "global_step": 67, "epoch": 0, "lr": 9.99995938324229e-05} +{"train_loss": 0.043418657034635544, "grad_norm": 0.16300319135189056, "global_step": 68, "epoch": 0, "lr": 9.999958179849546e-05} +{"train_loss": 0.044898487627506256, "grad_norm": 0.3336860239505768, "global_step": 69, "epoch": 0, "lr": 9.999956958889123e-05} +{"train_loss": 0.04714639484882355, "grad_norm": 0.3212790787220001, "global_step": 70, "epoch": 0, "lr": 9.999955720361027e-05} +{"train_loss": 0.04204151779413223, "grad_norm": 0.3483557105064392, "global_step": 71, "epoch": 0, "lr": 9.99995446426526e-05} +{"train_loss": 0.04301440343260765, "grad_norm": 0.2435990422964096, "global_step": 72, "epoch": 0, "lr": 9.999953190601826e-05} +{"train_loss": 0.03645404428243637, "grad_norm": 0.3318331837654114, "global_step": 73, "epoch": 0, "lr": 9.999951899370732e-05} +{"train_loss": 0.03599288687109947, "grad_norm": 0.330047607421875, "global_step": 74, "epoch": 0, "lr": 9.999950590571983e-05} +{"train_loss": 0.04013041406869888, "grad_norm": 0.2393750548362732, "global_step": 75, "epoch": 0, "lr": 9.999949264205579e-05} +{"train_loss": 0.039702191948890686, "grad_norm": 0.2533881664276123, "global_step": 76, "epoch": 0, "lr": 9.999947920271531e-05} +{"train_loss": 0.041999608278274536, "grad_norm": 0.22126740217208862, "global_step": 77, "epoch": 0, "lr": 9.999946558769838e-05} +{"train_loss": 0.0411483570933342, "grad_norm": 0.2783517837524414, "global_step": 78, "epoch": 0, "lr": 9.999945179700507e-05} +{"train_loss": 0.03845767676830292, "grad_norm": 0.2557986378669739, "global_step": 79, "epoch": 0, "lr": 9.999943783063545e-05} +{"train_loss": 0.042252205312252045, "grad_norm": 0.2378852367401123, "global_step": 80, "epoch": 0, "lr": 9.999942368858955e-05} +{"train_loss": 0.03959976136684418, "grad_norm": 0.20329870283603668, "global_step": 81, "epoch": 0, "lr": 9.999940937086741e-05} +{"train_loss": 0.04298325628042221, "grad_norm": 0.09557542949914932, "global_step": 82, "epoch": 0, "lr": 9.999939487746908e-05} +{"train_loss": 0.04166339337825775, "grad_norm": 0.20976749062538147, "global_step": 83, "epoch": 0, "lr": 9.999938020839464e-05} +{"train_loss": 0.03748054802417755, "grad_norm": 0.22029954195022583, "global_step": 84, "epoch": 0, "lr": 9.99993653636441e-05} +{"train_loss": 0.042376913130283356, "grad_norm": 0.17042338848114014, "global_step": 85, "epoch": 0, "lr": 9.999935034321755e-05} +{"train_loss": 0.03932422772049904, "grad_norm": 0.20792970061302185, "global_step": 86, "epoch": 0, "lr": 9.999933514711502e-05} +{"train_loss": 0.04103337973356247, "grad_norm": 0.15017449855804443, "global_step": 87, "epoch": 0, "lr": 9.999931977533657e-05} +{"train_loss": 0.0365462563931942, "grad_norm": 0.08021162450313568, "global_step": 88, "epoch": 0, "lr": 9.999930422788227e-05} +{"train_loss": 0.040397919714450836, "grad_norm": 0.25543954968452454, "global_step": 89, "epoch": 0, "lr": 9.999928850475214e-05} +{"train_loss": 0.04046417400240898, "grad_norm": 0.10623904317617416, "global_step": 90, "epoch": 0, "lr": 9.999927260594626e-05} +{"train_loss": 0.03968045860528946, "grad_norm": 0.1907067745923996, "global_step": 91, "epoch": 0, "lr": 9.999925653146469e-05} +{"train_loss": 0.03492370992898941, "grad_norm": 0.10737563669681549, "global_step": 92, "epoch": 0, "lr": 9.999924028130747e-05} +{"train_loss": 0.03471498191356659, "grad_norm": 0.08601723611354828, "global_step": 93, "epoch": 0, "lr": 9.999922385547466e-05} +{"train_loss": 0.03609948232769966, "grad_norm": 0.06602834910154343, "global_step": 94, "epoch": 0, "lr": 9.999920725396632e-05} +{"train_loss": 0.03565610945224762, "grad_norm": 0.08663801848888397, "global_step": 95, "epoch": 0, "lr": 9.999919047678252e-05} +{"train_loss": 0.034851059317588806, "grad_norm": 0.0851728767156601, "global_step": 96, "epoch": 0, "lr": 9.99991735239233e-05} +{"train_loss": 0.03232201933860779, "grad_norm": 0.05234619230031967, "global_step": 97, "epoch": 0, "lr": 9.999915639538872e-05} +{"train_loss": 0.0313965380191803, "grad_norm": 0.06071435287594795, "global_step": 98, "epoch": 0, "lr": 9.999913909117886e-05} +{"train_loss": 0.03440210223197937, "grad_norm": 0.08731284737586975, "global_step": 99, "epoch": 0, "lr": 9.999912161129376e-05} +{"train_loss": 0.03136659786105156, "grad_norm": 0.10282204300165176, "global_step": 100, "epoch": 0, "lr": 9.999910395573351e-05} +{"train_loss": 0.038527268916368484, "grad_norm": 0.14859239757061005, "global_step": 101, "epoch": 0, "lr": 9.999908612449815e-05} +{"train_loss": 0.03294378146529198, "grad_norm": 0.09187758713960648, "global_step": 102, "epoch": 0, "lr": 9.999906811758772e-05} +{"train_loss": 0.03674476593732834, "grad_norm": 0.08441931754350662, "global_step": 103, "epoch": 0, "lr": 9.999904993500233e-05} +{"train_loss": 0.03641355782747269, "grad_norm": 0.12490600347518921, "global_step": 104, "epoch": 0, "lr": 9.999903157674202e-05} +{"train_loss": 0.0695633239481809, "grad_norm": 0.18931855261325836, "global_step": 105, "epoch": 0, "lr": 9.999901304280685e-05, "train_batch_mse_error": 0.03426551818847656, "val_loss": 0.030651092529296875} +{"train_loss": 0.033274095505476, "grad_norm": 0.12605680525302887, "global_step": 106, "epoch": 1, "lr": 9.99989943331969e-05} +{"train_loss": 0.03494269400835037, "grad_norm": 0.12714317440986633, "global_step": 107, "epoch": 1, "lr": 9.999897544791224e-05} +{"train_loss": 0.036344535648822784, "grad_norm": 0.07821591198444366, "global_step": 108, "epoch": 1, "lr": 9.99989563869529e-05} +{"train_loss": 0.03469664230942726, "grad_norm": 0.17626895010471344, "global_step": 109, "epoch": 1, "lr": 9.999893715031898e-05} +{"train_loss": 0.03346690908074379, "grad_norm": 0.09429707378149033, "global_step": 110, "epoch": 1, "lr": 9.999891773801055e-05} +{"train_loss": 0.035281676799058914, "grad_norm": 0.06354650855064392, "global_step": 111, "epoch": 1, "lr": 9.999889815002766e-05} +{"train_loss": 0.03315851837396622, "grad_norm": 0.0860055610537529, "global_step": 112, "epoch": 1, "lr": 9.999887838637038e-05} +{"train_loss": 0.035447653383016586, "grad_norm": 0.08606725186109543, "global_step": 113, "epoch": 1, "lr": 9.999885844703878e-05} +{"train_loss": 0.03930744528770447, "grad_norm": 0.2138584852218628, "global_step": 114, "epoch": 1, "lr": 9.999883833203294e-05} +{"train_loss": 0.035501375794410706, "grad_norm": 0.1064247414469719, "global_step": 115, "epoch": 1, "lr": 9.999881804135293e-05} +{"train_loss": 0.03510626032948494, "grad_norm": 0.23692935705184937, "global_step": 116, "epoch": 1, "lr": 9.999879757499882e-05} +{"train_loss": 0.03507991507649422, "grad_norm": 0.06856951862573624, "global_step": 117, "epoch": 1, "lr": 9.999877693297068e-05} +{"train_loss": 0.03344421833753586, "grad_norm": 0.14508172869682312, "global_step": 118, "epoch": 1, "lr": 9.999875611526857e-05} +{"train_loss": 0.0324566513299942, "grad_norm": 0.14633218944072723, "global_step": 119, "epoch": 1, "lr": 9.999873512189259e-05} +{"train_loss": 0.032923463732004166, "grad_norm": 0.05318470299243927, "global_step": 120, "epoch": 1, "lr": 9.99987139528428e-05} +{"train_loss": 0.037254102528095245, "grad_norm": 0.08906786888837814, "global_step": 121, "epoch": 1, "lr": 9.999869260811925e-05} +{"train_loss": 0.032831281423568726, "grad_norm": 0.1263599991798401, "global_step": 122, "epoch": 1, "lr": 9.999867108772206e-05} +{"train_loss": 0.03438383340835571, "grad_norm": 0.070524662733078, "global_step": 123, "epoch": 1, "lr": 9.999864939165127e-05} +{"train_loss": 0.03109777346253395, "grad_norm": 0.06424808502197266, "global_step": 124, "epoch": 1, "lr": 9.999862751990697e-05} +{"train_loss": 0.03411497920751572, "grad_norm": 0.07326297461986542, "global_step": 125, "epoch": 1, "lr": 9.999860547248924e-05} +{"train_loss": 0.031552236527204514, "grad_norm": 0.12194424867630005, "global_step": 126, "epoch": 1, "lr": 9.999858324939817e-05} +{"train_loss": 0.03452565521001816, "grad_norm": 0.06691677123308182, "global_step": 127, "epoch": 1, "lr": 9.99985608506338e-05} +{"train_loss": 0.04055164009332657, "grad_norm": 0.06835930794477463, "global_step": 128, "epoch": 1, "lr": 9.999853827619624e-05} +{"train_loss": 0.035368308424949646, "grad_norm": 0.09959418326616287, "global_step": 129, "epoch": 1, "lr": 9.999851552608557e-05} +{"train_loss": 0.03338250517845154, "grad_norm": 0.058767203241586685, "global_step": 130, "epoch": 1, "lr": 9.999849260030184e-05} +{"train_loss": 0.03073815256357193, "grad_norm": 0.0500606931746006, "global_step": 131, "epoch": 1, "lr": 9.999846949884517e-05} +{"train_loss": 0.03368256613612175, "grad_norm": 0.08517766743898392, "global_step": 132, "epoch": 1, "lr": 9.999844622171561e-05} +{"train_loss": 0.03460916131734848, "grad_norm": 0.06868886202573776, "global_step": 133, "epoch": 1, "lr": 9.999842276891325e-05} +{"train_loss": 0.03209973871707916, "grad_norm": 0.07716329395771027, "global_step": 134, "epoch": 1, "lr": 9.999839914043818e-05} +{"train_loss": 0.028545279055833817, "grad_norm": 0.05982876196503639, "global_step": 135, "epoch": 1, "lr": 9.999837533629049e-05} +{"train_loss": 0.030043045058846474, "grad_norm": 0.041306160390377045, "global_step": 136, "epoch": 1, "lr": 9.999835135647025e-05} +{"train_loss": 0.03771517425775528, "grad_norm": 0.048825956881046295, "global_step": 137, "epoch": 1, "lr": 9.999832720097754e-05} +{"train_loss": 0.03522423654794693, "grad_norm": 0.06837695091962814, "global_step": 138, "epoch": 1, "lr": 9.999830286981244e-05} +{"train_loss": 0.03131553903222084, "grad_norm": 0.09087679535150528, "global_step": 139, "epoch": 1, "lr": 9.999827836297507e-05} +{"train_loss": 0.030572816729545593, "grad_norm": 0.05311621353030205, "global_step": 140, "epoch": 1, "lr": 9.99982536804655e-05} +{"train_loss": 0.03280899301171303, "grad_norm": 0.06892812252044678, "global_step": 141, "epoch": 1, "lr": 9.999822882228378e-05} +{"train_loss": 0.032127633690834045, "grad_norm": 0.07826043665409088, "global_step": 142, "epoch": 1, "lr": 9.999820378843005e-05} +{"train_loss": 0.03087790310382843, "grad_norm": 0.08545217663049698, "global_step": 143, "epoch": 1, "lr": 9.999817857890436e-05} +{"train_loss": 0.03686977177858353, "grad_norm": 0.05776585265994072, "global_step": 144, "epoch": 1, "lr": 9.999815319370683e-05} +{"train_loss": 0.03563973307609558, "grad_norm": 0.14005069434642792, "global_step": 145, "epoch": 1, "lr": 9.999812763283751e-05} +{"train_loss": 0.03121764212846756, "grad_norm": 0.11447542160749435, "global_step": 146, "epoch": 1, "lr": 9.999810189629654e-05} +{"train_loss": 0.0304276030510664, "grad_norm": 0.1292549967765808, "global_step": 147, "epoch": 1, "lr": 9.999807598408397e-05} +{"train_loss": 0.031911976635456085, "grad_norm": 0.04744856432080269, "global_step": 148, "epoch": 1, "lr": 9.999804989619991e-05} +{"train_loss": 0.03235382214188576, "grad_norm": 0.0786663293838501, "global_step": 149, "epoch": 1, "lr": 9.999802363264444e-05} +{"train_loss": 0.03123030625283718, "grad_norm": 0.07022860646247864, "global_step": 150, "epoch": 1, "lr": 9.999799719341765e-05} +{"train_loss": 0.027906442061066628, "grad_norm": 0.05041176825761795, "global_step": 151, "epoch": 1, "lr": 9.999797057851967e-05} +{"train_loss": 0.030027884989976883, "grad_norm": 0.045919571071863174, "global_step": 152, "epoch": 1, "lr": 9.999794378795054e-05} +{"train_loss": 0.03466428816318512, "grad_norm": 0.06636496633291245, "global_step": 153, "epoch": 1, "lr": 9.999791682171039e-05} +{"train_loss": 0.031886495649814606, "grad_norm": 0.11496098339557648, "global_step": 154, "epoch": 1, "lr": 9.999788967979931e-05} +{"train_loss": 0.03298000991344452, "grad_norm": 0.09521432220935822, "global_step": 155, "epoch": 1, "lr": 9.999786236221737e-05} +{"train_loss": 0.03186787664890289, "grad_norm": 0.09064815193414688, "global_step": 156, "epoch": 1, "lr": 9.999783486896469e-05} +{"train_loss": 0.029194120317697525, "grad_norm": 0.06889694184064865, "global_step": 157, "epoch": 1, "lr": 9.999780720004137e-05} +{"train_loss": 0.03440069407224655, "grad_norm": 0.04539816826581955, "global_step": 158, "epoch": 1, "lr": 9.99977793554475e-05} +{"train_loss": 0.03000236302614212, "grad_norm": 0.05185740441083908, "global_step": 159, "epoch": 1, "lr": 9.999775133518317e-05} +{"train_loss": 0.030348435044288635, "grad_norm": 0.061818480491638184, "global_step": 160, "epoch": 1, "lr": 9.999772313924849e-05} +{"train_loss": 0.03195320814847946, "grad_norm": 0.06125014275312424, "global_step": 161, "epoch": 1, "lr": 9.999769476764355e-05} +{"train_loss": 0.03019505739212036, "grad_norm": 0.05445617809891701, "global_step": 162, "epoch": 1, "lr": 9.999766622036846e-05} +{"train_loss": 0.030944008380174637, "grad_norm": 0.0472399927675724, "global_step": 163, "epoch": 1, "lr": 9.99976374974233e-05} +{"train_loss": 0.03559153899550438, "grad_norm": 0.07264941930770874, "global_step": 164, "epoch": 1, "lr": 9.99976085988082e-05} +{"train_loss": 0.035771310329437256, "grad_norm": 0.04943203181028366, "global_step": 165, "epoch": 1, "lr": 9.999757952452324e-05} +{"train_loss": 0.0334167554974556, "grad_norm": 0.04927501082420349, "global_step": 166, "epoch": 1, "lr": 9.999755027456854e-05} +{"train_loss": 0.03380659967660904, "grad_norm": 0.09534112364053726, "global_step": 167, "epoch": 1, "lr": 9.999752084894419e-05} +{"train_loss": 0.03140408545732498, "grad_norm": 0.1421033889055252, "global_step": 168, "epoch": 1, "lr": 9.999749124765028e-05} +{"train_loss": 0.03498300537467003, "grad_norm": 0.08249147981405258, "global_step": 169, "epoch": 1, "lr": 9.999746147068694e-05} +{"train_loss": 0.03253217414021492, "grad_norm": 0.04241132736206055, "global_step": 170, "epoch": 1, "lr": 9.999743151805425e-05} +{"train_loss": 0.033601313829422, "grad_norm": 0.06230204924941063, "global_step": 171, "epoch": 1, "lr": 9.999740138975234e-05} +{"train_loss": 0.03070656582713127, "grad_norm": 0.05796630308032036, "global_step": 172, "epoch": 1, "lr": 9.999737108578131e-05} +{"train_loss": 0.027913957834243774, "grad_norm": 0.10271865129470825, "global_step": 173, "epoch": 1, "lr": 9.999734060614126e-05} +{"train_loss": 0.03285415843129158, "grad_norm": 0.04805703088641167, "global_step": 174, "epoch": 1, "lr": 9.999730995083229e-05} +{"train_loss": 0.029270123690366745, "grad_norm": 0.05123389884829521, "global_step": 175, "epoch": 1, "lr": 9.999727911985452e-05} +{"train_loss": 0.035264790058135986, "grad_norm": 0.05552799999713898, "global_step": 176, "epoch": 1, "lr": 9.999724811320805e-05} +{"train_loss": 0.028629429638385773, "grad_norm": 0.04803669452667236, "global_step": 177, "epoch": 1, "lr": 9.9997216930893e-05} +{"train_loss": 0.032643888145685196, "grad_norm": 0.10521727055311203, "global_step": 178, "epoch": 1, "lr": 9.999718557290947e-05} +{"train_loss": 0.030180800706148148, "grad_norm": 0.04070200398564339, "global_step": 179, "epoch": 1, "lr": 9.999715403925758e-05} +{"train_loss": 0.034073591232299805, "grad_norm": 0.06954093277454376, "global_step": 180, "epoch": 1, "lr": 9.999712232993744e-05} +{"train_loss": 0.0352698490023613, "grad_norm": 0.13019169867038727, "global_step": 181, "epoch": 1, "lr": 9.999709044494912e-05} +{"train_loss": 0.029383983463048935, "grad_norm": 0.03924454748630524, "global_step": 182, "epoch": 1, "lr": 9.99970583842928e-05} +{"train_loss": 0.03457208350300789, "grad_norm": 0.04473830759525299, "global_step": 183, "epoch": 1, "lr": 9.999702614796855e-05} +{"train_loss": 0.029791120439767838, "grad_norm": 0.10996409505605698, "global_step": 184, "epoch": 1, "lr": 9.999699373597649e-05} +{"train_loss": 0.028682533651590347, "grad_norm": 0.09784140437841415, "global_step": 185, "epoch": 1, "lr": 9.999696114831673e-05} +{"train_loss": 0.027255505323410034, "grad_norm": 0.08718732744455338, "global_step": 186, "epoch": 1, "lr": 9.999692838498942e-05} +{"train_loss": 0.03560596704483032, "grad_norm": 0.0704951360821724, "global_step": 187, "epoch": 1, "lr": 9.999689544599463e-05} +{"train_loss": 0.030578210949897766, "grad_norm": 0.10066770017147064, "global_step": 188, "epoch": 1, "lr": 9.999686233133249e-05} +{"train_loss": 0.03518005460500717, "grad_norm": 0.05139991268515587, "global_step": 189, "epoch": 1, "lr": 9.999682904100312e-05} +{"train_loss": 0.03139428794384003, "grad_norm": 0.05154846981167793, "global_step": 190, "epoch": 1, "lr": 9.999679557500663e-05} +{"train_loss": 0.026063188910484314, "grad_norm": 0.1009453535079956, "global_step": 191, "epoch": 1, "lr": 9.999676193334316e-05} +{"train_loss": 0.03573678433895111, "grad_norm": 0.07883638143539429, "global_step": 192, "epoch": 1, "lr": 9.999672811601279e-05} +{"train_loss": 0.027232984080910683, "grad_norm": 0.06613217294216156, "global_step": 193, "epoch": 1, "lr": 9.999669412301568e-05} +{"train_loss": 0.0322989746928215, "grad_norm": 0.08291692286729813, "global_step": 194, "epoch": 1, "lr": 9.999665995435192e-05} +{"train_loss": 0.03180278092622757, "grad_norm": 0.03854561597108841, "global_step": 195, "epoch": 1, "lr": 9.999662561002164e-05} +{"train_loss": 0.027026360854506493, "grad_norm": 0.08551416546106339, "global_step": 196, "epoch": 1, "lr": 9.999659109002497e-05} +{"train_loss": 0.03339216858148575, "grad_norm": 0.045152921229600906, "global_step": 197, "epoch": 1, "lr": 9.9996556394362e-05} +{"train_loss": 0.03528851270675659, "grad_norm": 0.05384838208556175, "global_step": 198, "epoch": 1, "lr": 9.99965215230329e-05} +{"train_loss": 0.03084539622068405, "grad_norm": 0.09237328916788101, "global_step": 199, "epoch": 1, "lr": 9.999648647603774e-05} +{"train_loss": 0.0338272899389267, "grad_norm": 0.06252285838127136, "global_step": 200, "epoch": 1, "lr": 9.999645125337669e-05} +{"train_loss": 0.038187310099601746, "grad_norm": 0.12108223885297775, "global_step": 201, "epoch": 1, "lr": 9.999641585504985e-05} +{"train_loss": 0.029902657493948936, "grad_norm": 0.05370615795254707, "global_step": 202, "epoch": 1, "lr": 9.999638028105735e-05} +{"train_loss": 0.027714483439922333, "grad_norm": 0.08737368136644363, "global_step": 203, "epoch": 1, "lr": 9.99963445313993e-05} +{"train_loss": 0.028141973540186882, "grad_norm": 0.0635451078414917, "global_step": 204, "epoch": 1, "lr": 9.999630860607586e-05} +{"train_loss": 0.031687863171100616, "grad_norm": 0.047348517924547195, "global_step": 205, "epoch": 1, "lr": 9.999627250508711e-05} +{"train_loss": 0.03312785178422928, "grad_norm": 0.11381201446056366, "global_step": 206, "epoch": 1, "lr": 9.99962362284332e-05} +{"train_loss": 0.03176204115152359, "grad_norm": 0.050777748227119446, "global_step": 207, "epoch": 1, "lr": 9.999619977611428e-05} +{"train_loss": 0.03052201122045517, "grad_norm": 0.07039935141801834, "global_step": 208, "epoch": 1, "lr": 9.999616314813045e-05} +{"train_loss": 0.03365926444530487, "grad_norm": 0.08078213781118393, "global_step": 209, "epoch": 1, "lr": 9.999612634448183e-05} +{"train_loss": 0.02830103039741516, "grad_norm": 0.0780908539891243, "global_step": 210, "epoch": 1, "lr": 9.999608936516858e-05} +{"train_loss": 0.032547101448729354, "grad_norm": 0.07347510010004044, "global_step": 211, "epoch": 1, "lr": 9.999605221019081e-05, "val_loss": 0.027779866009950638} +{"train_loss": 0.03726448863744736, "grad_norm": 0.08971595019102097, "global_step": 212, "epoch": 2, "lr": 9.999601487954867e-05} +{"train_loss": 0.036622676998376846, "grad_norm": 0.06771181523799896, "global_step": 213, "epoch": 2, "lr": 9.999597737324224e-05} +{"train_loss": 0.02904311567544937, "grad_norm": 0.06992073357105255, "global_step": 214, "epoch": 2, "lr": 9.999593969127173e-05} +{"train_loss": 0.02913038246333599, "grad_norm": 0.05797048658132553, "global_step": 215, "epoch": 2, "lr": 9.999590183363721e-05} +{"train_loss": 0.029211997985839844, "grad_norm": 0.05739878863096237, "global_step": 216, "epoch": 2, "lr": 9.999586380033883e-05} +{"train_loss": 0.027663078159093857, "grad_norm": 0.06773938983678818, "global_step": 217, "epoch": 2, "lr": 9.999582559137672e-05} +{"train_loss": 0.0313989594578743, "grad_norm": 0.06477605551481247, "global_step": 218, "epoch": 2, "lr": 9.999578720675105e-05} +{"train_loss": 0.021534837782382965, "grad_norm": 0.04793505370616913, "global_step": 219, "epoch": 2, "lr": 9.99957486464619e-05} +{"train_loss": 0.027983738109469414, "grad_norm": 0.08363333344459534, "global_step": 220, "epoch": 2, "lr": 9.999570991050944e-05} +{"train_loss": 0.03321191295981407, "grad_norm": 0.09989254176616669, "global_step": 221, "epoch": 2, "lr": 9.99956709988938e-05} +{"train_loss": 0.03118385747075081, "grad_norm": 0.058946821838617325, "global_step": 222, "epoch": 2, "lr": 9.999563191161512e-05} +{"train_loss": 0.029719490557909012, "grad_norm": 0.06883183121681213, "global_step": 223, "epoch": 2, "lr": 9.999559264867352e-05} +{"train_loss": 0.033882856369018555, "grad_norm": 0.0647350400686264, "global_step": 224, "epoch": 2, "lr": 9.999555321006917e-05} +{"train_loss": 0.031511832028627396, "grad_norm": 0.07261164486408234, "global_step": 225, "epoch": 2, "lr": 9.999551359580217e-05} +{"train_loss": 0.03150250017642975, "grad_norm": 0.09351082146167755, "global_step": 226, "epoch": 2, "lr": 9.999547380587269e-05} +{"train_loss": 0.02895563468337059, "grad_norm": 0.037106212228536606, "global_step": 227, "epoch": 2, "lr": 9.999543384028083e-05} +{"train_loss": 0.028460877016186714, "grad_norm": 0.04837068170309067, "global_step": 228, "epoch": 2, "lr": 9.999539369902678e-05} +{"train_loss": 0.030721690505743027, "grad_norm": 0.05014389008283615, "global_step": 229, "epoch": 2, "lr": 9.999535338211066e-05} +{"train_loss": 0.030523691326379776, "grad_norm": 0.09702645242214203, "global_step": 230, "epoch": 2, "lr": 9.999531288953261e-05} +{"train_loss": 0.029614411294460297, "grad_norm": 0.03594118356704712, "global_step": 231, "epoch": 2, "lr": 9.999527222129277e-05} +{"train_loss": 0.029630174860358238, "grad_norm": 0.16889266669750214, "global_step": 232, "epoch": 2, "lr": 9.999523137739128e-05} +{"train_loss": 0.03044090047478676, "grad_norm": 0.12057343870401382, "global_step": 233, "epoch": 2, "lr": 9.99951903578283e-05} +{"train_loss": 0.027173299342393875, "grad_norm": 0.12368988245725632, "global_step": 234, "epoch": 2, "lr": 9.999514916260396e-05} +{"train_loss": 0.031055189669132233, "grad_norm": 0.06134612485766411, "global_step": 235, "epoch": 2, "lr": 9.999510779171842e-05} +{"train_loss": 0.030280742794275284, "grad_norm": 0.07358214259147644, "global_step": 236, "epoch": 2, "lr": 9.999506624517179e-05} +{"train_loss": 0.03165556490421295, "grad_norm": 0.048813942819833755, "global_step": 237, "epoch": 2, "lr": 9.999502452296426e-05} +{"train_loss": 0.02709042653441429, "grad_norm": 0.06911061704158783, "global_step": 238, "epoch": 2, "lr": 9.999498262509593e-05} +{"train_loss": 0.03099430538713932, "grad_norm": 0.09296997636556625, "global_step": 239, "epoch": 2, "lr": 9.9994940551567e-05} +{"train_loss": 0.028175663203001022, "grad_norm": 0.060236185789108276, "global_step": 240, "epoch": 2, "lr": 9.999489830237758e-05} +{"train_loss": 0.028911352157592773, "grad_norm": 0.043728698045015335, "global_step": 241, "epoch": 2, "lr": 9.999485587752783e-05} +{"train_loss": 0.034436874091625214, "grad_norm": 0.04691457003355026, "global_step": 242, "epoch": 2, "lr": 9.999481327701789e-05} +{"train_loss": 0.03059268184006214, "grad_norm": 0.06720702350139618, "global_step": 243, "epoch": 2, "lr": 9.999477050084793e-05} +{"train_loss": 0.032667264342308044, "grad_norm": 0.0637330412864685, "global_step": 244, "epoch": 2, "lr": 9.999472754901808e-05} +{"train_loss": 0.029474228620529175, "grad_norm": 0.1526808738708496, "global_step": 245, "epoch": 2, "lr": 9.99946844215285e-05} +{"train_loss": 0.02722255513072014, "grad_norm": 0.04186349734663963, "global_step": 246, "epoch": 2, "lr": 9.999464111837936e-05} +{"train_loss": 0.03495804965496063, "grad_norm": 0.05656510218977928, "global_step": 247, "epoch": 2, "lr": 9.999459763957077e-05} +{"train_loss": 0.030460091307759285, "grad_norm": 0.06412942707538605, "global_step": 248, "epoch": 2, "lr": 9.999455398510293e-05} +{"train_loss": 0.02707902528345585, "grad_norm": 0.05667736753821373, "global_step": 249, "epoch": 2, "lr": 9.999451015497595e-05} +{"train_loss": 0.026811305433511734, "grad_norm": 0.11309488117694855, "global_step": 250, "epoch": 2, "lr": 9.999446614919001e-05} +{"train_loss": 0.025042515248060226, "grad_norm": 0.04836800694465637, "global_step": 251, "epoch": 2, "lr": 9.999442196774526e-05} +{"train_loss": 0.027484916150569916, "grad_norm": 0.04808366671204567, "global_step": 252, "epoch": 2, "lr": 9.999437761064184e-05} +{"train_loss": 0.027972552925348282, "grad_norm": 0.08824358880519867, "global_step": 253, "epoch": 2, "lr": 9.999433307787993e-05} +{"train_loss": 0.03312936797738075, "grad_norm": 0.057353150099515915, "global_step": 254, "epoch": 2, "lr": 9.999428836945968e-05} +{"train_loss": 0.03484576940536499, "grad_norm": 0.06297706067562103, "global_step": 255, "epoch": 2, "lr": 9.999424348538124e-05} +{"train_loss": 0.028897302225232124, "grad_norm": 0.08587954193353653, "global_step": 256, "epoch": 2, "lr": 9.999419842564476e-05} +{"train_loss": 0.03253701701760292, "grad_norm": 0.09829230606555939, "global_step": 257, "epoch": 2, "lr": 9.999415319025042e-05} +{"train_loss": 0.029118023812770844, "grad_norm": 0.03704211860895157, "global_step": 258, "epoch": 2, "lr": 9.999410777919836e-05} +{"train_loss": 0.025333426892757416, "grad_norm": 0.05944732949137688, "global_step": 259, "epoch": 2, "lr": 9.999406219248874e-05} +{"train_loss": 0.030558187514543533, "grad_norm": 0.052015744149684906, "global_step": 260, "epoch": 2, "lr": 9.999401643012174e-05} +{"train_loss": 0.023427320644259453, "grad_norm": 0.07550811767578125, "global_step": 261, "epoch": 2, "lr": 9.999397049209751e-05} +{"train_loss": 0.030250418931245804, "grad_norm": 0.05395721271634102, "global_step": 262, "epoch": 2, "lr": 9.999392437841621e-05} +{"train_loss": 0.028811015188694, "grad_norm": 0.06284771859645844, "global_step": 263, "epoch": 2, "lr": 9.999387808907799e-05} +{"train_loss": 0.028101354837417603, "grad_norm": 0.04579087719321251, "global_step": 264, "epoch": 2, "lr": 9.999383162408304e-05} +{"train_loss": 0.024949291720986366, "grad_norm": 0.06357766687870026, "global_step": 265, "epoch": 2, "lr": 9.999378498343149e-05} +{"train_loss": 0.027228130027651787, "grad_norm": 0.04410289227962494, "global_step": 266, "epoch": 2, "lr": 9.999373816712354e-05} +{"train_loss": 0.028885353356599808, "grad_norm": 0.04328237101435661, "global_step": 267, "epoch": 2, "lr": 9.99936911751593e-05} +{"train_loss": 0.02574840560555458, "grad_norm": 0.08612435311079025, "global_step": 268, "epoch": 2, "lr": 9.999364400753902e-05} +{"train_loss": 0.027916450053453445, "grad_norm": 0.07982773333787918, "global_step": 269, "epoch": 2, "lr": 9.999359666426277e-05} +{"train_loss": 0.032179493457078934, "grad_norm": 0.058261752128601074, "global_step": 270, "epoch": 2, "lr": 9.99935491453308e-05} +{"train_loss": 0.03166324272751808, "grad_norm": 0.05037238448858261, "global_step": 271, "epoch": 2, "lr": 9.999350145074322e-05} +{"train_loss": 0.02902737259864807, "grad_norm": 0.05337261036038399, "global_step": 272, "epoch": 2, "lr": 9.999345358050023e-05} +{"train_loss": 0.030108364298939705, "grad_norm": 0.07833291590213776, "global_step": 273, "epoch": 2, "lr": 9.999340553460198e-05} +{"train_loss": 0.02960631251335144, "grad_norm": 0.052846360951662064, "global_step": 274, "epoch": 2, "lr": 9.999335731304865e-05} +{"train_loss": 0.026346784085035324, "grad_norm": 0.08291283249855042, "global_step": 275, "epoch": 2, "lr": 9.99933089158404e-05} +{"train_loss": 0.02883434295654297, "grad_norm": 0.07300172746181488, "global_step": 276, "epoch": 2, "lr": 9.999326034297739e-05} +{"train_loss": 0.03092651255428791, "grad_norm": 0.11993634700775146, "global_step": 277, "epoch": 2, "lr": 9.999321159445982e-05} +{"train_loss": 0.028911028057336807, "grad_norm": 0.061514999717473984, "global_step": 278, "epoch": 2, "lr": 9.999316267028785e-05} +{"train_loss": 0.02690235897898674, "grad_norm": 0.0507112517952919, "global_step": 279, "epoch": 2, "lr": 9.999311357046163e-05} +{"train_loss": 0.026988867670297623, "grad_norm": 0.1068304255604744, "global_step": 280, "epoch": 2, "lr": 9.999306429498137e-05} +{"train_loss": 0.03078414686024189, "grad_norm": 0.04198962450027466, "global_step": 281, "epoch": 2, "lr": 9.99930148438472e-05} +{"train_loss": 0.022765956819057465, "grad_norm": 0.04877130687236786, "global_step": 282, "epoch": 2, "lr": 9.999296521705934e-05} +{"train_loss": 0.02850252203643322, "grad_norm": 0.05548081547021866, "global_step": 283, "epoch": 2, "lr": 9.999291541461793e-05} +{"train_loss": 0.03142522647976875, "grad_norm": 0.10083369165658951, "global_step": 284, "epoch": 2, "lr": 9.999286543652316e-05} +{"train_loss": 0.027276799082756042, "grad_norm": 0.08954643458127975, "global_step": 285, "epoch": 2, "lr": 9.99928152827752e-05} +{"train_loss": 0.029628153890371323, "grad_norm": 0.11535680294036865, "global_step": 286, "epoch": 2, "lr": 9.999276495337422e-05} +{"train_loss": 0.029925741255283356, "grad_norm": 0.038933880627155304, "global_step": 287, "epoch": 2, "lr": 9.999271444832041e-05} +{"train_loss": 0.028991149738430977, "grad_norm": 0.058069128543138504, "global_step": 288, "epoch": 2, "lr": 9.999266376761395e-05} +{"train_loss": 0.02674322947859764, "grad_norm": 0.05671289563179016, "global_step": 289, "epoch": 2, "lr": 9.999261291125501e-05} +{"train_loss": 0.030671723186969757, "grad_norm": 0.050534430891275406, "global_step": 290, "epoch": 2, "lr": 9.999256187924378e-05} +{"train_loss": 0.02700984664261341, "grad_norm": 0.08043409138917923, "global_step": 291, "epoch": 2, "lr": 9.999251067158041e-05} +{"train_loss": 0.02866969257593155, "grad_norm": 0.041417382657527924, "global_step": 292, "epoch": 2, "lr": 9.999245928826511e-05} +{"train_loss": 0.025927051901817322, "grad_norm": 0.13963976502418518, "global_step": 293, "epoch": 2, "lr": 9.999240772929804e-05} +{"train_loss": 0.031601179391145706, "grad_norm": 0.10442182421684265, "global_step": 294, "epoch": 2, "lr": 9.99923559946794e-05} +{"train_loss": 0.0294304508715868, "grad_norm": 0.040846917778253555, "global_step": 295, "epoch": 2, "lr": 9.999230408440935e-05} +{"train_loss": 0.029513809829950333, "grad_norm": 0.04188467934727669, "global_step": 296, "epoch": 2, "lr": 9.999225199848809e-05} +{"train_loss": 0.02486991137266159, "grad_norm": 0.03958280757069588, "global_step": 297, "epoch": 2, "lr": 9.999219973691581e-05} +{"train_loss": 0.029493043199181557, "grad_norm": 0.08865891396999359, "global_step": 298, "epoch": 2, "lr": 9.999214729969267e-05} +{"train_loss": 0.02713083103299141, "grad_norm": 0.06276105344295502, "global_step": 299, "epoch": 2, "lr": 9.999209468681885e-05} +{"train_loss": 0.030625129118561745, "grad_norm": 0.04139472171664238, "global_step": 300, "epoch": 2, "lr": 9.999204189829457e-05} +{"train_loss": 0.025810543447732925, "grad_norm": 0.049295876175165176, "global_step": 301, "epoch": 2, "lr": 9.999198893411999e-05} +{"train_loss": 0.029209226369857788, "grad_norm": 0.046844787895679474, "global_step": 302, "epoch": 2, "lr": 9.999193579429531e-05} +{"train_loss": 0.028921902179718018, "grad_norm": 0.06705188006162643, "global_step": 303, "epoch": 2, "lr": 9.99918824788207e-05} +{"train_loss": 0.02722805365920067, "grad_norm": 0.044591259211301804, "global_step": 304, "epoch": 2, "lr": 9.999182898769636e-05} +{"train_loss": 0.02770257368683815, "grad_norm": 0.059116579592227936, "global_step": 305, "epoch": 2, "lr": 9.999177532092247e-05} +{"train_loss": 0.029375605285167694, "grad_norm": 0.042371347546577454, "global_step": 306, "epoch": 2, "lr": 9.999172147849922e-05} +{"train_loss": 0.03084706701338291, "grad_norm": 0.071102574467659, "global_step": 307, "epoch": 2, "lr": 9.999166746042681e-05} +{"train_loss": 0.0277730543166399, "grad_norm": 0.04123951122164726, "global_step": 308, "epoch": 2, "lr": 9.99916132667054e-05} +{"train_loss": 0.027110636234283447, "grad_norm": 0.05529164895415306, "global_step": 309, "epoch": 2, "lr": 9.999155889733523e-05} +{"train_loss": 0.026271572336554527, "grad_norm": 0.07326245307922363, "global_step": 310, "epoch": 2, "lr": 9.999150435231646e-05} +{"train_loss": 0.029053371399641037, "grad_norm": 0.10310962796211243, "global_step": 311, "epoch": 2, "lr": 9.999144963164927e-05} +{"train_loss": 0.024600787088274956, "grad_norm": 0.05091644078493118, "global_step": 312, "epoch": 2, "lr": 9.999139473533388e-05} +{"train_loss": 0.02885841578245163, "grad_norm": 0.06260525435209274, "global_step": 313, "epoch": 2, "lr": 9.999133966337045e-05} +{"train_loss": 0.02593705803155899, "grad_norm": 0.03715028986334801, "global_step": 314, "epoch": 2, "lr": 9.99912844157592e-05} +{"train_loss": 0.030749844387173653, "grad_norm": 0.11362279951572418, "global_step": 315, "epoch": 2, "lr": 9.999122899250033e-05} +{"train_loss": 0.02749975584447384, "grad_norm": 0.05121017247438431, "global_step": 316, "epoch": 2, "lr": 9.999117339359401e-05} +{"train_loss": 0.02906797768302121, "grad_norm": 0.1118093952536583, "global_step": 317, "epoch": 2, "lr": 9.999111761904046e-05, "val_loss": 0.026000507175922394} +{"train_loss": 0.026112649589776993, "grad_norm": 0.07152663171291351, "global_step": 318, "epoch": 3, "lr": 9.999106166883985e-05} +{"train_loss": 0.028516311198472977, "grad_norm": 0.11830776184797287, "global_step": 319, "epoch": 3, "lr": 9.999100554299239e-05} +{"train_loss": 0.026250112801790237, "grad_norm": 0.16390953958034515, "global_step": 320, "epoch": 3, "lr": 9.99909492414983e-05} +{"train_loss": 0.030842751264572144, "grad_norm": 0.14899842441082, "global_step": 321, "epoch": 3, "lr": 9.999089276435773e-05} +{"train_loss": 0.025845468044281006, "grad_norm": 0.08190663903951645, "global_step": 322, "epoch": 3, "lr": 9.99908361115709e-05} +{"train_loss": 0.031005367636680603, "grad_norm": 0.0697806254029274, "global_step": 323, "epoch": 3, "lr": 9.999077928313803e-05} +{"train_loss": 0.02595827542245388, "grad_norm": 0.06707490235567093, "global_step": 324, "epoch": 3, "lr": 9.999072227905929e-05} +{"train_loss": 0.029453856870532036, "grad_norm": 0.07257840037345886, "global_step": 325, "epoch": 3, "lr": 9.99906650993349e-05} +{"train_loss": 0.028099477291107178, "grad_norm": 0.03637196496129036, "global_step": 326, "epoch": 3, "lr": 9.999060774396505e-05} +{"train_loss": 0.027844276279211044, "grad_norm": 0.06002700701355934, "global_step": 327, "epoch": 3, "lr": 9.999055021294995e-05} +{"train_loss": 0.02908068522810936, "grad_norm": 0.04253299906849861, "global_step": 328, "epoch": 3, "lr": 9.999049250628978e-05} +{"train_loss": 0.025827374309301376, "grad_norm": 0.1558893471956253, "global_step": 329, "epoch": 3, "lr": 9.999043462398476e-05} +{"train_loss": 0.03174586966633797, "grad_norm": 0.09248242527246475, "global_step": 330, "epoch": 3, "lr": 9.999037656603512e-05} +{"train_loss": 0.02590055763721466, "grad_norm": 0.14006638526916504, "global_step": 331, "epoch": 3, "lr": 9.999031833244101e-05} +{"train_loss": 0.025872379541397095, "grad_norm": 0.04167274013161659, "global_step": 332, "epoch": 3, "lr": 9.999025992320268e-05} +{"train_loss": 0.030756382271647453, "grad_norm": 0.11273862421512604, "global_step": 333, "epoch": 3, "lr": 9.999020133832031e-05} +{"train_loss": 0.02579031139612198, "grad_norm": 0.13009048998355865, "global_step": 334, "epoch": 3, "lr": 9.99901425777941e-05} +{"train_loss": 0.026883689686655998, "grad_norm": 0.07984588295221329, "global_step": 335, "epoch": 3, "lr": 9.999008364162429e-05} +{"train_loss": 0.026279602199792862, "grad_norm": 0.03989680856466293, "global_step": 336, "epoch": 3, "lr": 9.999002452981105e-05} +{"train_loss": 0.028925232589244843, "grad_norm": 0.095634326338768, "global_step": 337, "epoch": 3, "lr": 9.998996524235463e-05} +{"train_loss": 0.028062596917152405, "grad_norm": 0.07112135738134384, "global_step": 338, "epoch": 3, "lr": 9.998990577925518e-05} +{"train_loss": 0.026401467621326447, "grad_norm": 0.08257287740707397, "global_step": 339, "epoch": 3, "lr": 9.998984614051296e-05} +{"train_loss": 0.025263536721467972, "grad_norm": 0.10054187476634979, "global_step": 340, "epoch": 3, "lr": 9.998978632612815e-05} +{"train_loss": 0.02696249820291996, "grad_norm": 0.06772653013467789, "global_step": 341, "epoch": 3, "lr": 9.998972633610097e-05} +{"train_loss": 0.03157633915543556, "grad_norm": 0.0710861086845398, "global_step": 342, "epoch": 3, "lr": 9.998966617043164e-05} +{"train_loss": 0.021940751001238823, "grad_norm": 0.037410810589790344, "global_step": 343, "epoch": 3, "lr": 9.998960582912036e-05} +{"train_loss": 0.02787376567721367, "grad_norm": 0.07077254354953766, "global_step": 344, "epoch": 3, "lr": 9.998954531216735e-05} +{"train_loss": 0.025396864861249924, "grad_norm": 0.06456790119409561, "global_step": 345, "epoch": 3, "lr": 9.99894846195728e-05} +{"train_loss": 0.029296567663550377, "grad_norm": 0.049108393490314484, "global_step": 346, "epoch": 3, "lr": 9.998942375133697e-05} +{"train_loss": 0.02933317795395851, "grad_norm": 0.030918831005692482, "global_step": 347, "epoch": 3, "lr": 9.998936270746002e-05} +{"train_loss": 0.030952611938118935, "grad_norm": 0.08500389009714127, "global_step": 348, "epoch": 3, "lr": 9.998930148794222e-05} +{"train_loss": 0.028707528486847878, "grad_norm": 0.08316099643707275, "global_step": 349, "epoch": 3, "lr": 9.998924009278372e-05} +{"train_loss": 0.0245773084461689, "grad_norm": 0.03313073888421059, "global_step": 350, "epoch": 3, "lr": 9.998917852198477e-05} +{"train_loss": 0.030290408059954643, "grad_norm": 0.110013946890831, "global_step": 351, "epoch": 3, "lr": 9.998911677554561e-05} +{"train_loss": 0.029301991686224937, "grad_norm": 0.08714357763528824, "global_step": 352, "epoch": 3, "lr": 9.998905485346641e-05} +{"train_loss": 0.024172313511371613, "grad_norm": 0.12426746636629105, "global_step": 353, "epoch": 3, "lr": 9.998899275574742e-05} +{"train_loss": 0.024564510211348534, "grad_norm": 0.09570270031690598, "global_step": 354, "epoch": 3, "lr": 9.998893048238887e-05} +{"train_loss": 0.02968817949295044, "grad_norm": 0.08284185081720352, "global_step": 355, "epoch": 3, "lr": 9.998886803339094e-05} +{"train_loss": 0.02695944905281067, "grad_norm": 0.07682517170906067, "global_step": 356, "epoch": 3, "lr": 9.998880540875386e-05} +{"train_loss": 0.02482529729604721, "grad_norm": 0.059486355632543564, "global_step": 357, "epoch": 3, "lr": 9.998874260847786e-05} +{"train_loss": 0.024459194391965866, "grad_norm": 0.056699685752391815, "global_step": 358, "epoch": 3, "lr": 9.998867963256318e-05} +{"train_loss": 0.028653038665652275, "grad_norm": 0.07864575833082199, "global_step": 359, "epoch": 3, "lr": 9.998861648100999e-05} +{"train_loss": 0.023385323584079742, "grad_norm": 0.069581039249897, "global_step": 360, "epoch": 3, "lr": 9.998855315381856e-05} +{"train_loss": 0.028487036004662514, "grad_norm": 0.14467467367649078, "global_step": 361, "epoch": 3, "lr": 9.998848965098909e-05} +{"train_loss": 0.02964450791478157, "grad_norm": 0.04112040624022484, "global_step": 362, "epoch": 3, "lr": 9.998842597252179e-05} +{"train_loss": 0.027058787643909454, "grad_norm": 0.06113028898835182, "global_step": 363, "epoch": 3, "lr": 9.998836211841691e-05} +{"train_loss": 0.02675556391477585, "grad_norm": 0.039426084607839584, "global_step": 364, "epoch": 3, "lr": 9.998829808867467e-05} +{"train_loss": 0.028410430997610092, "grad_norm": 0.07890152931213379, "global_step": 365, "epoch": 3, "lr": 9.99882338832953e-05} +{"train_loss": 0.02984366938471794, "grad_norm": 0.08876091986894608, "global_step": 366, "epoch": 3, "lr": 9.9988169502279e-05} +{"train_loss": 0.027295446023344994, "grad_norm": 0.04839525371789932, "global_step": 367, "epoch": 3, "lr": 9.998810494562602e-05} +{"train_loss": 0.027287054806947708, "grad_norm": 0.11683716624975204, "global_step": 368, "epoch": 3, "lr": 9.998804021333656e-05} +{"train_loss": 0.026199474930763245, "grad_norm": 0.0915127694606781, "global_step": 369, "epoch": 3, "lr": 9.998797530541089e-05} +{"train_loss": 0.02563576027750969, "grad_norm": 0.03282804787158966, "global_step": 370, "epoch": 3, "lr": 9.998791022184922e-05} +{"train_loss": 0.027079947292804718, "grad_norm": 0.04627114534378052, "global_step": 371, "epoch": 3, "lr": 9.998784496265174e-05} +{"train_loss": 0.023514382541179657, "grad_norm": 0.049760233610868454, "global_step": 372, "epoch": 3, "lr": 9.998777952781872e-05} +{"train_loss": 0.02425912395119667, "grad_norm": 0.050354063510894775, "global_step": 373, "epoch": 3, "lr": 9.99877139173504e-05} +{"train_loss": 0.026756644248962402, "grad_norm": 0.0570375919342041, "global_step": 374, "epoch": 3, "lr": 9.998764813124699e-05} +{"train_loss": 0.025179393589496613, "grad_norm": 0.059979185461997986, "global_step": 375, "epoch": 3, "lr": 9.998758216950872e-05} +{"train_loss": 0.023834489285945892, "grad_norm": 0.07383499294519424, "global_step": 376, "epoch": 3, "lr": 9.998751603213581e-05} +{"train_loss": 0.02905569039285183, "grad_norm": 0.07371704280376434, "global_step": 377, "epoch": 3, "lr": 9.998744971912852e-05} +{"train_loss": 0.030300630256533623, "grad_norm": 0.04507659375667572, "global_step": 378, "epoch": 3, "lr": 9.998738323048707e-05} +{"train_loss": 0.0276908241212368, "grad_norm": 0.07850860059261322, "global_step": 379, "epoch": 3, "lr": 9.99873165662117e-05} +{"train_loss": 0.025729328393936157, "grad_norm": 0.03921676054596901, "global_step": 380, "epoch": 3, "lr": 9.998724972630263e-05} +{"train_loss": 0.02721005119383335, "grad_norm": 0.03905080258846283, "global_step": 381, "epoch": 3, "lr": 9.998718271076011e-05} +{"train_loss": 0.02660035341978073, "grad_norm": 0.06860348582267761, "global_step": 382, "epoch": 3, "lr": 9.998711551958437e-05} +{"train_loss": 0.025891657918691635, "grad_norm": 0.06386873871088028, "global_step": 383, "epoch": 3, "lr": 9.998704815277565e-05} +{"train_loss": 0.02830914594233036, "grad_norm": 0.11585406959056854, "global_step": 384, "epoch": 3, "lr": 9.998698061033416e-05} +{"train_loss": 0.028661416843533516, "grad_norm": 0.07035131007432938, "global_step": 385, "epoch": 3, "lr": 9.998691289226016e-05} +{"train_loss": 0.02920670434832573, "grad_norm": 0.09693284332752228, "global_step": 386, "epoch": 3, "lr": 9.99868449985539e-05} +{"train_loss": 0.025883739814162254, "grad_norm": 0.07662685960531235, "global_step": 387, "epoch": 3, "lr": 9.998677692921561e-05} +{"train_loss": 0.024183517321944237, "grad_norm": 0.12103227525949478, "global_step": 388, "epoch": 3, "lr": 9.998670868424551e-05} +{"train_loss": 0.027118142694234848, "grad_norm": 0.08424708992242813, "global_step": 389, "epoch": 3, "lr": 9.998664026364386e-05} +{"train_loss": 0.024510785937309265, "grad_norm": 0.04824148491024971, "global_step": 390, "epoch": 3, "lr": 9.99865716674109e-05} +{"train_loss": 0.024995043873786926, "grad_norm": 0.05446017533540726, "global_step": 391, "epoch": 3, "lr": 9.998650289554686e-05} +{"train_loss": 0.026099059730768204, "grad_norm": 0.05721554905176163, "global_step": 392, "epoch": 3, "lr": 9.9986433948052e-05} +{"train_loss": 0.030459793284535408, "grad_norm": 0.06196407228708267, "global_step": 393, "epoch": 3, "lr": 9.998636482492654e-05} +{"train_loss": 0.02354784496128559, "grad_norm": 0.04773406684398651, "global_step": 394, "epoch": 3, "lr": 9.998629552617073e-05} +{"train_loss": 0.023705974221229553, "grad_norm": 0.03292224928736687, "global_step": 395, "epoch": 3, "lr": 9.998622605178481e-05} +{"train_loss": 0.02769019827246666, "grad_norm": 0.10486340522766113, "global_step": 396, "epoch": 3, "lr": 9.998615640176903e-05} +{"train_loss": 0.027803853154182434, "grad_norm": 0.07180038094520569, "global_step": 397, "epoch": 3, "lr": 9.998608657612364e-05} +{"train_loss": 0.027593206614255905, "grad_norm": 0.07661314308643341, "global_step": 398, "epoch": 3, "lr": 9.998601657484889e-05} +{"train_loss": 0.024608902633190155, "grad_norm": 0.07911572605371475, "global_step": 399, "epoch": 3, "lr": 9.998594639794501e-05} +{"train_loss": 0.0294482484459877, "grad_norm": 0.0990164503455162, "global_step": 400, "epoch": 3, "lr": 9.998587604541225e-05} +{"train_loss": 0.02843649312853813, "grad_norm": 0.13693690299987793, "global_step": 401, "epoch": 3, "lr": 9.998580551725087e-05} +{"train_loss": 0.02716394141316414, "grad_norm": 0.03773903101682663, "global_step": 402, "epoch": 3, "lr": 9.99857348134611e-05} +{"train_loss": 0.02787187322974205, "grad_norm": 0.09915974736213684, "global_step": 403, "epoch": 3, "lr": 9.998566393404319e-05} +{"train_loss": 0.02585464157164097, "grad_norm": 0.054379504173994064, "global_step": 404, "epoch": 3, "lr": 9.99855928789974e-05} +{"train_loss": 0.026743996888399124, "grad_norm": 0.030051490291953087, "global_step": 405, "epoch": 3, "lr": 9.998552164832398e-05} +{"train_loss": 0.02641282230615616, "grad_norm": 0.09956119954586029, "global_step": 406, "epoch": 3, "lr": 9.998545024202317e-05} +{"train_loss": 0.025231095030903816, "grad_norm": 0.04444780573248863, "global_step": 407, "epoch": 3, "lr": 9.998537866009522e-05} +{"train_loss": 0.0299639031291008, "grad_norm": 0.06338421255350113, "global_step": 408, "epoch": 3, "lr": 9.99853069025404e-05} +{"train_loss": 0.026843976229429245, "grad_norm": 0.06598442047834396, "global_step": 409, "epoch": 3, "lr": 9.998523496935894e-05} +{"train_loss": 0.02740049734711647, "grad_norm": 0.043809711933135986, "global_step": 410, "epoch": 3, "lr": 9.998516286055111e-05} +{"train_loss": 0.02766069769859314, "grad_norm": 0.04807659611105919, "global_step": 411, "epoch": 3, "lr": 9.998509057611715e-05} +{"train_loss": 0.024819279089570045, "grad_norm": 0.05285349860787392, "global_step": 412, "epoch": 3, "lr": 9.998501811605732e-05} +{"train_loss": 0.028748642653226852, "grad_norm": 0.08488048613071442, "global_step": 413, "epoch": 3, "lr": 9.998494548037187e-05} +{"train_loss": 0.028081100434064865, "grad_norm": 0.06741122901439667, "global_step": 414, "epoch": 3, "lr": 9.998487266906107e-05} +{"train_loss": 0.025576330721378326, "grad_norm": 0.14409995079040527, "global_step": 415, "epoch": 3, "lr": 9.998479968212515e-05} +{"train_loss": 0.023951251059770584, "grad_norm": 0.09294846653938293, "global_step": 416, "epoch": 3, "lr": 9.998472651956438e-05} +{"train_loss": 0.028378034010529518, "grad_norm": 0.04358884319663048, "global_step": 417, "epoch": 3, "lr": 9.998465318137904e-05} +{"train_loss": 0.024011332541704178, "grad_norm": 0.03709207475185394, "global_step": 418, "epoch": 3, "lr": 9.998457966756935e-05} +{"train_loss": 0.02542746067047119, "grad_norm": 0.08553089201450348, "global_step": 419, "epoch": 3, "lr": 9.998450597813561e-05} +{"train_loss": 0.024313729256391525, "grad_norm": 0.0359191820025444, "global_step": 420, "epoch": 3, "lr": 9.998443211307804e-05} +{"train_loss": 0.026083610951900482, "grad_norm": 0.03866295889019966, "global_step": 421, "epoch": 3, "lr": 9.99843580723969e-05} +{"train_loss": 0.023430367931723595, "grad_norm": 0.07682793587446213, "global_step": 422, "epoch": 3, "lr": 9.998428385609246e-05} +{"train_loss": 0.027035437910905423, "grad_norm": 0.11752782016992569, "global_step": 423, "epoch": 3, "lr": 9.9984209464165e-05, "val_loss": 0.02453131414949894} +{"train_loss": 0.02540021389722824, "grad_norm": 0.13305234909057617, "global_step": 424, "epoch": 4, "lr": 9.998413489661476e-05} +{"train_loss": 0.02541271783411503, "grad_norm": 0.07504552602767944, "global_step": 425, "epoch": 4, "lr": 9.998406015344201e-05} +{"train_loss": 0.02512252703309059, "grad_norm": 0.09271826595067978, "global_step": 426, "epoch": 4, "lr": 9.998398523464702e-05} +{"train_loss": 0.026728350669145584, "grad_norm": 0.11550754308700562, "global_step": 427, "epoch": 4, "lr": 9.998391014023004e-05} +{"train_loss": 0.02412370964884758, "grad_norm": 0.10161762684583664, "global_step": 428, "epoch": 4, "lr": 9.998383487019131e-05} +{"train_loss": 0.024224095046520233, "grad_norm": 0.07350118458271027, "global_step": 429, "epoch": 4, "lr": 9.998375942453115e-05} +{"train_loss": 0.024280674755573273, "grad_norm": 0.03955167159438133, "global_step": 430, "epoch": 4, "lr": 9.998368380324979e-05} +{"train_loss": 0.02730528637766838, "grad_norm": 0.04607017710804939, "global_step": 431, "epoch": 4, "lr": 9.99836080063475e-05} +{"train_loss": 0.027548035606741905, "grad_norm": 0.039617978036403656, "global_step": 432, "epoch": 4, "lr": 9.998353203382454e-05} +{"train_loss": 0.027008701115846634, "grad_norm": 0.0701042115688324, "global_step": 433, "epoch": 4, "lr": 9.99834558856812e-05} +{"train_loss": 0.030224189162254333, "grad_norm": 0.11246369034051895, "global_step": 434, "epoch": 4, "lr": 9.998337956191773e-05} +{"train_loss": 0.02545006386935711, "grad_norm": 0.07281386852264404, "global_step": 435, "epoch": 4, "lr": 9.99833030625344e-05} +{"train_loss": 0.027435526251792908, "grad_norm": 0.07120907306671143, "global_step": 436, "epoch": 4, "lr": 9.998322638753148e-05} +{"train_loss": 0.026825720444321632, "grad_norm": 0.03322931379079819, "global_step": 437, "epoch": 4, "lr": 9.998314953690924e-05} +{"train_loss": 0.029211897403001785, "grad_norm": 0.0813872218132019, "global_step": 438, "epoch": 4, "lr": 9.998307251066794e-05} +{"train_loss": 0.025642767548561096, "grad_norm": 0.10124076902866364, "global_step": 439, "epoch": 4, "lr": 9.998299530880787e-05} +{"train_loss": 0.0257129929959774, "grad_norm": 0.05917460098862648, "global_step": 440, "epoch": 4, "lr": 9.998291793132929e-05} +{"train_loss": 0.02415587566792965, "grad_norm": 0.04432906582951546, "global_step": 441, "epoch": 4, "lr": 9.998284037823248e-05} +{"train_loss": 0.023884985595941544, "grad_norm": 0.12817759811878204, "global_step": 442, "epoch": 4, "lr": 9.99827626495177e-05} +{"train_loss": 0.026769006624817848, "grad_norm": 0.0410134457051754, "global_step": 443, "epoch": 4, "lr": 9.998268474518521e-05} +{"train_loss": 0.02598883956670761, "grad_norm": 0.07304532825946808, "global_step": 444, "epoch": 4, "lr": 9.998260666523533e-05} +{"train_loss": 0.029035670682787895, "grad_norm": 0.07495170831680298, "global_step": 445, "epoch": 4, "lr": 9.998252840966831e-05} +{"train_loss": 0.0257294699549675, "grad_norm": 0.07394834607839584, "global_step": 446, "epoch": 4, "lr": 9.998244997848441e-05} +{"train_loss": 0.026282241567969322, "grad_norm": 0.054805267602205276, "global_step": 447, "epoch": 4, "lr": 9.998237137168391e-05} +{"train_loss": 0.027238084003329277, "grad_norm": 0.08627880364656448, "global_step": 448, "epoch": 4, "lr": 9.99822925892671e-05} +{"train_loss": 0.02572575956583023, "grad_norm": 0.07642001658678055, "global_step": 449, "epoch": 4, "lr": 9.998221363123425e-05} +{"train_loss": 0.022231508046388626, "grad_norm": 0.07131218165159225, "global_step": 450, "epoch": 4, "lr": 9.998213449758564e-05} +{"train_loss": 0.025611357763409615, "grad_norm": 0.11191164702177048, "global_step": 451, "epoch": 4, "lr": 9.998205518832156e-05} +{"train_loss": 0.02249128371477127, "grad_norm": 0.05727686733007431, "global_step": 452, "epoch": 4, "lr": 9.998197570344225e-05} +{"train_loss": 0.02459222450852394, "grad_norm": 0.05347767472267151, "global_step": 453, "epoch": 4, "lr": 9.998189604294803e-05} +{"train_loss": 0.02762339822947979, "grad_norm": 0.0466853603720665, "global_step": 454, "epoch": 4, "lr": 9.998181620683917e-05} +{"train_loss": 0.027624767273664474, "grad_norm": 0.03915196284651756, "global_step": 455, "epoch": 4, "lr": 9.998173619511593e-05} +{"train_loss": 0.02943413332104683, "grad_norm": 0.08398184180259705, "global_step": 456, "epoch": 4, "lr": 9.99816560077786e-05} +{"train_loss": 0.02792191132903099, "grad_norm": 0.04014461860060692, "global_step": 457, "epoch": 4, "lr": 9.998157564482749e-05} +{"train_loss": 0.025851737707853317, "grad_norm": 0.1285729855298996, "global_step": 458, "epoch": 4, "lr": 9.998149510626284e-05} +{"train_loss": 0.026673609390854836, "grad_norm": 0.062158193439245224, "global_step": 459, "epoch": 4, "lr": 9.998141439208495e-05} +{"train_loss": 0.027441056445240974, "grad_norm": 0.04858440160751343, "global_step": 460, "epoch": 4, "lr": 9.998133350229412e-05} +{"train_loss": 0.025501878932118416, "grad_norm": 0.03223281353712082, "global_step": 461, "epoch": 4, "lr": 9.998125243689061e-05} +{"train_loss": 0.0263516902923584, "grad_norm": 0.09051781892776489, "global_step": 462, "epoch": 4, "lr": 9.998117119587473e-05} +{"train_loss": 0.02210017666220665, "grad_norm": 0.06290948390960693, "global_step": 463, "epoch": 4, "lr": 9.998108977924674e-05} +{"train_loss": 0.024425240233540535, "grad_norm": 0.07442675530910492, "global_step": 464, "epoch": 4, "lr": 9.998100818700693e-05} +{"train_loss": 0.02345234528183937, "grad_norm": 0.07786436378955841, "global_step": 465, "epoch": 4, "lr": 9.99809264191556e-05} +{"train_loss": 0.026647116988897324, "grad_norm": 0.09266906976699829, "global_step": 466, "epoch": 4, "lr": 9.998084447569303e-05} +{"train_loss": 0.02645743079483509, "grad_norm": 0.0588143989443779, "global_step": 467, "epoch": 4, "lr": 9.998076235661951e-05} +{"train_loss": 0.026487216353416443, "grad_norm": 0.13065169751644135, "global_step": 468, "epoch": 4, "lr": 9.998068006193533e-05} +{"train_loss": 0.026664670556783676, "grad_norm": 0.08350877463817596, "global_step": 469, "epoch": 4, "lr": 9.998059759164078e-05} +{"train_loss": 0.025562115013599396, "grad_norm": 0.043344058096408844, "global_step": 470, "epoch": 4, "lr": 9.998051494573613e-05} +{"train_loss": 0.025961317121982574, "grad_norm": 0.05436795949935913, "global_step": 471, "epoch": 4, "lr": 9.99804321242217e-05} +{"train_loss": 0.02592601627111435, "grad_norm": 0.06871426850557327, "global_step": 472, "epoch": 4, "lr": 9.998034912709776e-05} +{"train_loss": 0.025967400521039963, "grad_norm": 0.05366189405322075, "global_step": 473, "epoch": 4, "lr": 9.998026595436461e-05} +{"train_loss": 0.026909438893198967, "grad_norm": 0.07950195670127869, "global_step": 474, "epoch": 4, "lr": 9.998018260602255e-05} +{"train_loss": 0.022336076945066452, "grad_norm": 0.07654087245464325, "global_step": 475, "epoch": 4, "lr": 9.998009908207185e-05} +{"train_loss": 0.02591617777943611, "grad_norm": 0.034689515829086304, "global_step": 476, "epoch": 4, "lr": 9.998001538251282e-05} +{"train_loss": 0.028642628341913223, "grad_norm": 0.05013241618871689, "global_step": 477, "epoch": 4, "lr": 9.997993150734577e-05} +{"train_loss": 0.027910511940717697, "grad_norm": 0.08314794301986694, "global_step": 478, "epoch": 4, "lr": 9.997984745657097e-05} +{"train_loss": 0.025721866637468338, "grad_norm": 0.03906261548399925, "global_step": 479, "epoch": 4, "lr": 9.997976323018871e-05} +{"train_loss": 0.02568177878856659, "grad_norm": 0.10261064767837524, "global_step": 480, "epoch": 4, "lr": 9.99796788281993e-05} +{"train_loss": 0.029139390215277672, "grad_norm": 0.03857232630252838, "global_step": 481, "epoch": 4, "lr": 9.997959425060304e-05} +{"train_loss": 0.02888374589383602, "grad_norm": 0.037003982812166214, "global_step": 482, "epoch": 4, "lr": 9.997950949740022e-05} +{"train_loss": 0.028477055951952934, "grad_norm": 0.08879515528678894, "global_step": 483, "epoch": 4, "lr": 9.997942456859113e-05} +{"train_loss": 0.02606218308210373, "grad_norm": 0.032694559544324875, "global_step": 484, "epoch": 4, "lr": 9.997933946417607e-05} +{"train_loss": 0.029216036200523376, "grad_norm": 0.05214531347155571, "global_step": 485, "epoch": 4, "lr": 9.997925418415539e-05} +{"train_loss": 0.025601722300052643, "grad_norm": 0.06513062119483948, "global_step": 486, "epoch": 4, "lr": 9.99791687285293e-05} +{"train_loss": 0.027345359325408936, "grad_norm": 0.05991361290216446, "global_step": 487, "epoch": 4, "lr": 9.997908309729818e-05} +{"train_loss": 0.02412155270576477, "grad_norm": 0.03937182202935219, "global_step": 488, "epoch": 4, "lr": 9.997899729046227e-05} +{"train_loss": 0.02495587058365345, "grad_norm": 0.062255796045064926, "global_step": 489, "epoch": 4, "lr": 9.997891130802192e-05} +{"train_loss": 0.02340950071811676, "grad_norm": 0.043841443955898285, "global_step": 490, "epoch": 4, "lr": 9.99788251499774e-05} +{"train_loss": 0.022146565839648247, "grad_norm": 0.035573627799749374, "global_step": 491, "epoch": 4, "lr": 9.997873881632902e-05} +{"train_loss": 0.02579636685550213, "grad_norm": 0.13474436104297638, "global_step": 492, "epoch": 4, "lr": 9.99786523070771e-05} +{"train_loss": 0.027072399854660034, "grad_norm": 0.07467580586671829, "global_step": 493, "epoch": 4, "lr": 9.997856562222192e-05} +{"train_loss": 0.027334844693541527, "grad_norm": 0.04631876200437546, "global_step": 494, "epoch": 4, "lr": 9.997847876176379e-05} +{"train_loss": 0.02508120983839035, "grad_norm": 0.07951907068490982, "global_step": 495, "epoch": 4, "lr": 9.997839172570303e-05} +{"train_loss": 0.02202780544757843, "grad_norm": 0.12446410953998566, "global_step": 496, "epoch": 4, "lr": 9.997830451403993e-05} +{"train_loss": 0.02221788465976715, "grad_norm": 0.028427820652723312, "global_step": 497, "epoch": 4, "lr": 9.997821712677482e-05} +{"train_loss": 0.02740601822733879, "grad_norm": 0.041825637221336365, "global_step": 498, "epoch": 4, "lr": 9.997812956390798e-05} +{"train_loss": 0.023334024474024773, "grad_norm": 0.06873654574155807, "global_step": 499, "epoch": 4, "lr": 9.997804182543973e-05} +{"train_loss": 0.0226791650056839, "grad_norm": 0.06595759093761444, "global_step": 500, "epoch": 4, "lr": 9.997795391137037e-05} +{"train_loss": 0.026757147163152695, "grad_norm": 0.04759099707007408, "global_step": 501, "epoch": 4, "lr": 9.99778658217002e-05} +{"train_loss": 0.021771838888525963, "grad_norm": 0.13352803885936737, "global_step": 502, "epoch": 4, "lr": 9.997777755642957e-05} +{"train_loss": 0.026614587754011154, "grad_norm": 0.03965076804161072, "global_step": 503, "epoch": 4, "lr": 9.997768911555876e-05} +{"train_loss": 0.02739664912223816, "grad_norm": 0.0472562275826931, "global_step": 504, "epoch": 4, "lr": 9.997760049908809e-05} +{"train_loss": 0.029190562665462494, "grad_norm": 0.07531802356243134, "global_step": 505, "epoch": 4, "lr": 9.997751170701785e-05} +{"train_loss": 0.021899882704019547, "grad_norm": 0.05165518447756767, "global_step": 506, "epoch": 4, "lr": 9.997742273934836e-05} +{"train_loss": 0.026365047320723534, "grad_norm": 0.07223277539014816, "global_step": 507, "epoch": 4, "lr": 9.997733359607997e-05} +{"train_loss": 0.02401481196284294, "grad_norm": 0.11733275651931763, "global_step": 508, "epoch": 4, "lr": 9.997724427721295e-05} +{"train_loss": 0.023594342172145844, "grad_norm": 0.07097987830638885, "global_step": 509, "epoch": 4, "lr": 9.997715478274762e-05} +{"train_loss": 0.026763880625367165, "grad_norm": 0.0483386404812336, "global_step": 510, "epoch": 4, "lr": 9.997706511268432e-05} +{"train_loss": 0.02351195737719536, "grad_norm": 0.06207440420985222, "global_step": 511, "epoch": 4, "lr": 9.997697526702334e-05} +{"train_loss": 0.026157010346651077, "grad_norm": 0.06875003129243851, "global_step": 512, "epoch": 4, "lr": 9.9976885245765e-05} +{"train_loss": 0.02541254833340645, "grad_norm": 0.07802195847034454, "global_step": 513, "epoch": 4, "lr": 9.997679504890962e-05} +{"train_loss": 0.025730669498443604, "grad_norm": 0.11411664634943008, "global_step": 514, "epoch": 4, "lr": 9.997670467645752e-05} +{"train_loss": 0.025120090693235397, "grad_norm": 0.11952058225870132, "global_step": 515, "epoch": 4, "lr": 9.997661412840902e-05} +{"train_loss": 0.026046859100461006, "grad_norm": 0.11699531227350235, "global_step": 516, "epoch": 4, "lr": 9.997652340476443e-05} +{"train_loss": 0.02496388554573059, "grad_norm": 0.04052771255373955, "global_step": 517, "epoch": 4, "lr": 9.997643250552406e-05} +{"train_loss": 0.025227047502994537, "grad_norm": 0.030840329825878143, "global_step": 518, "epoch": 4, "lr": 9.997634143068825e-05} +{"train_loss": 0.02458024024963379, "grad_norm": 0.04155219346284866, "global_step": 519, "epoch": 4, "lr": 9.997625018025731e-05} +{"train_loss": 0.021773584187030792, "grad_norm": 0.058496054261922836, "global_step": 520, "epoch": 4, "lr": 9.997615875423156e-05} +{"train_loss": 0.027887754142284393, "grad_norm": 0.06277579069137573, "global_step": 521, "epoch": 4, "lr": 9.997606715261132e-05} +{"train_loss": 0.025633882731199265, "grad_norm": 0.10276342928409576, "global_step": 522, "epoch": 4, "lr": 9.997597537539691e-05} +{"train_loss": 0.026056133210659027, "grad_norm": 0.04496682435274124, "global_step": 523, "epoch": 4, "lr": 9.997588342258865e-05} +{"train_loss": 0.02550637349486351, "grad_norm": 0.0512939989566803, "global_step": 524, "epoch": 4, "lr": 9.997579129418688e-05} +{"train_loss": 0.024819884449243546, "grad_norm": 0.04390212893486023, "global_step": 525, "epoch": 4, "lr": 9.997569899019193e-05} +{"train_loss": 0.023729104548692703, "grad_norm": 0.0817941427230835, "global_step": 526, "epoch": 4, "lr": 9.997560651060409e-05} +{"train_loss": 0.02231907844543457, "grad_norm": 0.11018239706754684, "global_step": 527, "epoch": 4, "lr": 9.99755138554237e-05} +{"train_loss": 0.02777743525803089, "grad_norm": 0.051406215876340866, "global_step": 528, "epoch": 4, "lr": 9.99754210246511e-05} +{"train_loss": 0.025760771407974215, "grad_norm": 0.10792924463748932, "global_step": 529, "epoch": 4, "lr": 9.997532801828658e-05, "val_loss": 0.023849111050367355} +{"train_loss": 0.02399727888405323, "grad_norm": 0.05972142145037651, "global_step": 530, "epoch": 5, "lr": 9.997523483633051e-05} +{"train_loss": 0.025168489664793015, "grad_norm": 0.03898408263921738, "global_step": 531, "epoch": 5, "lr": 9.99751414787832e-05} +{"train_loss": 0.027199672535061836, "grad_norm": 0.0606030635535717, "global_step": 532, "epoch": 5, "lr": 9.997504794564496e-05} +{"train_loss": 0.02447207272052765, "grad_norm": 0.03175637125968933, "global_step": 533, "epoch": 5, "lr": 9.997495423691616e-05} +{"train_loss": 0.02571311593055725, "grad_norm": 0.03629527986049652, "global_step": 534, "epoch": 5, "lr": 9.997486035259708e-05} +{"train_loss": 0.022157244384288788, "grad_norm": 0.03912577033042908, "global_step": 535, "epoch": 5, "lr": 9.997476629268809e-05} +{"train_loss": 0.027475621551275253, "grad_norm": 0.15960930287837982, "global_step": 536, "epoch": 5, "lr": 9.99746720571895e-05} +{"train_loss": 0.027671054005622864, "grad_norm": 0.09085825830698013, "global_step": 537, "epoch": 5, "lr": 9.997457764610165e-05} +{"train_loss": 0.023947181180119514, "grad_norm": 0.1444002240896225, "global_step": 538, "epoch": 5, "lr": 9.997448305942486e-05} +{"train_loss": 0.027277499437332153, "grad_norm": 0.0481526181101799, "global_step": 539, "epoch": 5, "lr": 9.997438829715948e-05} +{"train_loss": 0.024540377780795097, "grad_norm": 0.1384110450744629, "global_step": 540, "epoch": 5, "lr": 9.997429335930582e-05} +{"train_loss": 0.025868769735097885, "grad_norm": 0.05592754855751991, "global_step": 541, "epoch": 5, "lr": 9.997419824586424e-05} +{"train_loss": 0.02366171032190323, "grad_norm": 0.04424649849534035, "global_step": 542, "epoch": 5, "lr": 9.997410295683505e-05} +{"train_loss": 0.026378054171800613, "grad_norm": 0.036929842084646225, "global_step": 543, "epoch": 5, "lr": 9.997400749221859e-05} +{"train_loss": 0.02384129725396633, "grad_norm": 0.041734568774700165, "global_step": 544, "epoch": 5, "lr": 9.997391185201521e-05} +{"train_loss": 0.02944956161081791, "grad_norm": 0.1019650399684906, "global_step": 545, "epoch": 5, "lr": 9.997381603622525e-05} +{"train_loss": 0.025632459670305252, "grad_norm": 0.03677859902381897, "global_step": 546, "epoch": 5, "lr": 9.997372004484902e-05} +{"train_loss": 0.02545289695262909, "grad_norm": 0.04747525975108147, "global_step": 547, "epoch": 5, "lr": 9.997362387788687e-05} +{"train_loss": 0.027453996241092682, "grad_norm": 0.0456041544675827, "global_step": 548, "epoch": 5, "lr": 9.997352753533913e-05} +{"train_loss": 0.023587770760059357, "grad_norm": 0.06023411452770233, "global_step": 549, "epoch": 5, "lr": 9.997343101720615e-05} +{"train_loss": 0.02371317148208618, "grad_norm": 0.08383660763502121, "global_step": 550, "epoch": 5, "lr": 9.997333432348827e-05} +{"train_loss": 0.027921609580516815, "grad_norm": 0.05643269047141075, "global_step": 551, "epoch": 5, "lr": 9.997323745418583e-05} +{"train_loss": 0.026183325797319412, "grad_norm": 0.05270441249012947, "global_step": 552, "epoch": 5, "lr": 9.997314040929918e-05} +{"train_loss": 0.029507359489798546, "grad_norm": 0.04930200055241585, "global_step": 553, "epoch": 5, "lr": 9.997304318882863e-05} +{"train_loss": 0.024986136704683304, "grad_norm": 0.03962183743715286, "global_step": 554, "epoch": 5, "lr": 9.997294579277454e-05} +{"train_loss": 0.025727931410074234, "grad_norm": 0.03602074086666107, "global_step": 555, "epoch": 5, "lr": 9.997284822113725e-05} +{"train_loss": 0.02636316791176796, "grad_norm": 0.05273842811584473, "global_step": 556, "epoch": 5, "lr": 9.99727504739171e-05} +{"train_loss": 0.026635296642780304, "grad_norm": 0.06744015216827393, "global_step": 557, "epoch": 5, "lr": 9.997265255111446e-05} +{"train_loss": 0.02542310208082199, "grad_norm": 0.07455099374055862, "global_step": 558, "epoch": 5, "lr": 9.997255445272963e-05} +{"train_loss": 0.025965068489313126, "grad_norm": 0.0801808163523674, "global_step": 559, "epoch": 5, "lr": 9.997245617876297e-05} +{"train_loss": 0.024566853418946266, "grad_norm": 0.039363596588373184, "global_step": 560, "epoch": 5, "lr": 9.997235772921485e-05} +{"train_loss": 0.02396615967154503, "grad_norm": 0.06566248834133148, "global_step": 561, "epoch": 5, "lr": 9.997225910408558e-05} +{"train_loss": 0.021461796015501022, "grad_norm": 0.06323274224996567, "global_step": 562, "epoch": 5, "lr": 9.997216030337555e-05} +{"train_loss": 0.02841857075691223, "grad_norm": 0.058585602790117264, "global_step": 563, "epoch": 5, "lr": 9.997206132708505e-05} +{"train_loss": 0.025629373267292976, "grad_norm": 0.03408821299672127, "global_step": 564, "epoch": 5, "lr": 9.997196217521448e-05} +{"train_loss": 0.02459268644452095, "grad_norm": 0.05942723527550697, "global_step": 565, "epoch": 5, "lr": 9.997186284776417e-05} +{"train_loss": 0.024590663611888885, "grad_norm": 0.08111606538295746, "global_step": 566, "epoch": 5, "lr": 9.997176334473445e-05} +{"train_loss": 0.025693850591778755, "grad_norm": 0.10645707696676254, "global_step": 567, "epoch": 5, "lr": 9.997166366612569e-05} +{"train_loss": 0.02544480189681053, "grad_norm": 0.0788666158914566, "global_step": 568, "epoch": 5, "lr": 9.997156381193824e-05} +{"train_loss": 0.023345397785305977, "grad_norm": 0.036952149122953415, "global_step": 569, "epoch": 5, "lr": 9.997146378217245e-05} +{"train_loss": 0.0232224240899086, "grad_norm": 0.0444970428943634, "global_step": 570, "epoch": 5, "lr": 9.997136357682866e-05} +{"train_loss": 0.024206962436437607, "grad_norm": 0.04686804488301277, "global_step": 571, "epoch": 5, "lr": 9.997126319590724e-05} +{"train_loss": 0.02479584515094757, "grad_norm": 0.05130450800061226, "global_step": 572, "epoch": 5, "lr": 9.997116263940853e-05} +{"train_loss": 0.027517391368746758, "grad_norm": 0.08445964753627777, "global_step": 573, "epoch": 5, "lr": 9.997106190733288e-05} +{"train_loss": 0.02264915034174919, "grad_norm": 0.16200925409793854, "global_step": 574, "epoch": 5, "lr": 9.997096099968065e-05} +{"train_loss": 0.025041785091161728, "grad_norm": 0.050624292343854904, "global_step": 575, "epoch": 5, "lr": 9.997085991645219e-05} +{"train_loss": 0.026356074959039688, "grad_norm": 0.048881471157073975, "global_step": 576, "epoch": 5, "lr": 9.997075865764787e-05} +{"train_loss": 0.023208137601614, "grad_norm": 0.03238365426659584, "global_step": 577, "epoch": 5, "lr": 9.997065722326804e-05} +{"train_loss": 0.024995291605591774, "grad_norm": 0.09984396398067474, "global_step": 578, "epoch": 5, "lr": 9.997055561331306e-05} +{"train_loss": 0.024890869855880737, "grad_norm": 0.06039237976074219, "global_step": 579, "epoch": 5, "lr": 9.997045382778325e-05} +{"train_loss": 0.024236124008893967, "grad_norm": 0.13869430124759674, "global_step": 580, "epoch": 5, "lr": 9.997035186667903e-05} +{"train_loss": 0.025973767042160034, "grad_norm": 0.10373516380786896, "global_step": 581, "epoch": 5, "lr": 9.99702497300007e-05} +{"train_loss": 0.026239067316055298, "grad_norm": 0.10610555112361908, "global_step": 582, "epoch": 5, "lr": 9.997014741774866e-05} +{"train_loss": 0.023791326209902763, "grad_norm": 0.11153605580329895, "global_step": 583, "epoch": 5, "lr": 9.997004492992323e-05} +{"train_loss": 0.024715881794691086, "grad_norm": 0.04144347086548805, "global_step": 584, "epoch": 5, "lr": 9.996994226652481e-05} +{"train_loss": 0.02451113983988762, "grad_norm": 0.10122714191675186, "global_step": 585, "epoch": 5, "lr": 9.996983942755374e-05} +{"train_loss": 0.02638302743434906, "grad_norm": 0.05916024371981621, "global_step": 586, "epoch": 5, "lr": 9.99697364130104e-05} +{"train_loss": 0.02317391149699688, "grad_norm": 0.07340096682310104, "global_step": 587, "epoch": 5, "lr": 9.996963322289512e-05} +{"train_loss": 0.027066994458436966, "grad_norm": 0.08199653774499893, "global_step": 588, "epoch": 5, "lr": 9.996952985720828e-05} +{"train_loss": 0.02528628334403038, "grad_norm": 0.04107009992003441, "global_step": 589, "epoch": 5, "lr": 9.996942631595025e-05} +{"train_loss": 0.02559400163590908, "grad_norm": 0.09609483182430267, "global_step": 590, "epoch": 5, "lr": 9.99693225991214e-05} +{"train_loss": 0.024201413616538048, "grad_norm": 0.057534847408533096, "global_step": 591, "epoch": 5, "lr": 9.996921870672206e-05} +{"train_loss": 0.022745560854673386, "grad_norm": 0.09574950486421585, "global_step": 592, "epoch": 5, "lr": 9.996911463875262e-05} +{"train_loss": 0.02573009580373764, "grad_norm": 0.05504102259874344, "global_step": 593, "epoch": 5, "lr": 9.996901039521345e-05} +{"train_loss": 0.025210710242390633, "grad_norm": 0.08970990777015686, "global_step": 594, "epoch": 5, "lr": 9.99689059761049e-05} +{"train_loss": 0.02505221590399742, "grad_norm": 0.039687320590019226, "global_step": 595, "epoch": 5, "lr": 9.996880138142735e-05} +{"train_loss": 0.02544066309928894, "grad_norm": 0.03754715248942375, "global_step": 596, "epoch": 5, "lr": 9.996869661118118e-05} +{"train_loss": 0.02097686380147934, "grad_norm": 0.051264144480228424, "global_step": 597, "epoch": 5, "lr": 9.996859166536671e-05} +{"train_loss": 0.025570884346961975, "grad_norm": 0.1234242394566536, "global_step": 598, "epoch": 5, "lr": 9.996848654398436e-05} +{"train_loss": 0.025530897080898285, "grad_norm": 0.05370388552546501, "global_step": 599, "epoch": 5, "lr": 9.996838124703447e-05} +{"train_loss": 0.02486291155219078, "grad_norm": 0.11398403346538544, "global_step": 600, "epoch": 5, "lr": 9.996827577451741e-05} +{"train_loss": 0.023636721074581146, "grad_norm": 0.09451400488615036, "global_step": 601, "epoch": 5, "lr": 9.996817012643359e-05} +{"train_loss": 0.023235101252794266, "grad_norm": 0.03910832479596138, "global_step": 602, "epoch": 5, "lr": 9.996806430278333e-05} +{"train_loss": 0.02616547793149948, "grad_norm": 0.027684234082698822, "global_step": 603, "epoch": 5, "lr": 9.996795830356702e-05} +{"train_loss": 0.025470420718193054, "grad_norm": 0.051346052438020706, "global_step": 604, "epoch": 5, "lr": 9.996785212878503e-05} +{"train_loss": 0.024955488741397858, "grad_norm": 0.047286417335271835, "global_step": 605, "epoch": 5, "lr": 9.996774577843776e-05} +{"train_loss": 0.022813640534877777, "grad_norm": 0.037962906062603, "global_step": 606, "epoch": 5, "lr": 9.996763925252555e-05} +{"train_loss": 0.027041181921958923, "grad_norm": 0.11020021885633469, "global_step": 607, "epoch": 5, "lr": 9.996753255104878e-05} +{"train_loss": 0.023869343101978302, "grad_norm": 0.08379130810499191, "global_step": 608, "epoch": 5, "lr": 9.996742567400784e-05} +{"train_loss": 0.02458353526890278, "grad_norm": 0.04963703453540802, "global_step": 609, "epoch": 5, "lr": 9.996731862140309e-05} +{"train_loss": 0.024948231875896454, "grad_norm": 0.03455319628119469, "global_step": 610, "epoch": 5, "lr": 9.996721139323492e-05} +{"train_loss": 0.023150049149990082, "grad_norm": 0.09998032450675964, "global_step": 611, "epoch": 5, "lr": 9.996710398950368e-05} +{"train_loss": 0.021478429436683655, "grad_norm": 0.07490620017051697, "global_step": 612, "epoch": 5, "lr": 9.996699641020979e-05} +{"train_loss": 0.024609211832284927, "grad_norm": 0.06789272278547287, "global_step": 613, "epoch": 5, "lr": 9.996688865535359e-05} +{"train_loss": 0.0240766741335392, "grad_norm": 0.053204722702503204, "global_step": 614, "epoch": 5, "lr": 9.996678072493548e-05} +{"train_loss": 0.024519074708223343, "grad_norm": 0.09111741185188293, "global_step": 615, "epoch": 5, "lr": 9.996667261895585e-05} +{"train_loss": 0.02304287627339363, "grad_norm": 0.12833966314792633, "global_step": 616, "epoch": 5, "lr": 9.996656433741504e-05} +{"train_loss": 0.022352037951350212, "grad_norm": 0.07381384074687958, "global_step": 617, "epoch": 5, "lr": 9.996645588031346e-05} +{"train_loss": 0.024711184203624725, "grad_norm": 0.121941477060318, "global_step": 618, "epoch": 5, "lr": 9.996634724765148e-05} +{"train_loss": 0.025836221873760223, "grad_norm": 0.0561634823679924, "global_step": 619, "epoch": 5, "lr": 9.996623843942947e-05} +{"train_loss": 0.022393546998500824, "grad_norm": 0.049074895679950714, "global_step": 620, "epoch": 5, "lr": 9.996612945564785e-05} +{"train_loss": 0.024980051442980766, "grad_norm": 0.1406715214252472, "global_step": 621, "epoch": 5, "lr": 9.996602029630698e-05} +{"train_loss": 0.027057567611336708, "grad_norm": 0.07033990323543549, "global_step": 622, "epoch": 5, "lr": 9.996591096140724e-05} +{"train_loss": 0.02101813815534115, "grad_norm": 0.04235398769378662, "global_step": 623, "epoch": 5, "lr": 9.996580145094902e-05} +{"train_loss": 0.02355080097913742, "grad_norm": 0.07160893082618713, "global_step": 624, "epoch": 5, "lr": 9.996569176493269e-05} +{"train_loss": 0.02257019653916359, "grad_norm": 0.04308738186955452, "global_step": 625, "epoch": 5, "lr": 9.996558190335868e-05} +{"train_loss": 0.02472374215722084, "grad_norm": 0.11165127903223038, "global_step": 626, "epoch": 5, "lr": 9.996547186622731e-05} +{"train_loss": 0.026653483510017395, "grad_norm": 0.09229874610900879, "global_step": 627, "epoch": 5, "lr": 9.996536165353902e-05} +{"train_loss": 0.026204565539956093, "grad_norm": 0.04803215712308884, "global_step": 628, "epoch": 5, "lr": 9.996525126529417e-05} +{"train_loss": 0.028196485713124275, "grad_norm": 0.07015058398246765, "global_step": 629, "epoch": 5, "lr": 9.996514070149316e-05} +{"train_loss": 0.022657915949821472, "grad_norm": 0.05358823388814926, "global_step": 630, "epoch": 5, "lr": 9.996502996213636e-05} +{"train_loss": 0.0254393070936203, "grad_norm": 0.03326752036809921, "global_step": 631, "epoch": 5, "lr": 9.996491904722421e-05} +{"train_loss": 0.021369099617004395, "grad_norm": 0.04058064892888069, "global_step": 632, "epoch": 5, "lr": 9.996480795675705e-05} +{"train_loss": 0.027789045125246048, "grad_norm": 0.05527729541063309, "global_step": 633, "epoch": 5, "lr": 9.996469669073527e-05} +{"train_loss": 0.024270063266158104, "grad_norm": 0.057271555066108704, "global_step": 634, "epoch": 5, "lr": 9.996458524915928e-05} +{"train_loss": 0.02498436640104595, "grad_norm": 0.06411436200141907, "global_step": 635, "epoch": 5, "lr": 9.996447363202946e-05, "val_loss": 0.022283228114247322} +{"train_loss": 0.024485304951667786, "grad_norm": 0.08334246277809143, "global_step": 636, "epoch": 6, "lr": 9.996436183934623e-05} +{"train_loss": 0.022492242977023125, "grad_norm": 0.06693440675735474, "global_step": 637, "epoch": 6, "lr": 9.996424987110995e-05} +{"train_loss": 0.023044832050800323, "grad_norm": 0.04553801938891411, "global_step": 638, "epoch": 6, "lr": 9.996413772732101e-05} +{"train_loss": 0.02216998115181923, "grad_norm": 0.0474940650165081, "global_step": 639, "epoch": 6, "lr": 9.996402540797985e-05} +{"train_loss": 0.02461877465248108, "grad_norm": 0.03551987186074257, "global_step": 640, "epoch": 6, "lr": 9.99639129130868e-05} +{"train_loss": 0.021541977301239967, "grad_norm": 0.07492151111364365, "global_step": 641, "epoch": 6, "lr": 9.996380024264231e-05} +{"train_loss": 0.022442558780312538, "grad_norm": 0.05617701634764671, "global_step": 642, "epoch": 6, "lr": 9.996368739664674e-05} +{"train_loss": 0.024438684806227684, "grad_norm": 0.056649014353752136, "global_step": 643, "epoch": 6, "lr": 9.99635743751005e-05} +{"train_loss": 0.022220810875296593, "grad_norm": 0.08770652115345001, "global_step": 644, "epoch": 6, "lr": 9.996346117800402e-05} +{"train_loss": 0.024774111807346344, "grad_norm": 0.04670124128460884, "global_step": 645, "epoch": 6, "lr": 9.996334780535763e-05} +{"train_loss": 0.021957699209451675, "grad_norm": 0.08767033368349075, "global_step": 646, "epoch": 6, "lr": 9.996323425716178e-05} +{"train_loss": 0.024014504626393318, "grad_norm": 0.09458492696285248, "global_step": 647, "epoch": 6, "lr": 9.996312053341686e-05} +{"train_loss": 0.02427946776151657, "grad_norm": 0.14913442730903625, "global_step": 648, "epoch": 6, "lr": 9.996300663412326e-05} +{"train_loss": 0.023963607847690582, "grad_norm": 0.10051226615905762, "global_step": 649, "epoch": 6, "lr": 9.996289255928137e-05} +{"train_loss": 0.024900006130337715, "grad_norm": 0.06749290972948074, "global_step": 650, "epoch": 6, "lr": 9.996277830889161e-05} +{"train_loss": 0.02409771829843521, "grad_norm": 0.08277945965528488, "global_step": 651, "epoch": 6, "lr": 9.996266388295439e-05} +{"train_loss": 0.02712133526802063, "grad_norm": 0.07047981023788452, "global_step": 652, "epoch": 6, "lr": 9.996254928147007e-05} +{"train_loss": 0.023012056946754456, "grad_norm": 0.07593125104904175, "global_step": 653, "epoch": 6, "lr": 9.996243450443909e-05} +{"train_loss": 0.023357361555099487, "grad_norm": 0.06539614498615265, "global_step": 654, "epoch": 6, "lr": 9.996231955186186e-05} +{"train_loss": 0.027259573340415955, "grad_norm": 0.05359321087598801, "global_step": 655, "epoch": 6, "lr": 9.996220442373875e-05} +{"train_loss": 0.021255141124129295, "grad_norm": 0.04805575683712959, "global_step": 656, "epoch": 6, "lr": 9.996208912007019e-05} +{"train_loss": 0.024535078555345535, "grad_norm": 0.07267870008945465, "global_step": 657, "epoch": 6, "lr": 9.996197364085658e-05} +{"train_loss": 0.023971080780029297, "grad_norm": 0.04144430533051491, "global_step": 658, "epoch": 6, "lr": 9.99618579860983e-05} +{"train_loss": 0.023570410907268524, "grad_norm": 0.057293377816677094, "global_step": 659, "epoch": 6, "lr": 9.99617421557958e-05} +{"train_loss": 0.028711052611470222, "grad_norm": 0.05118595436215401, "global_step": 660, "epoch": 6, "lr": 9.996162614994946e-05} +{"train_loss": 0.022520892322063446, "grad_norm": 0.13114626705646515, "global_step": 661, "epoch": 6, "lr": 9.996150996855969e-05} +{"train_loss": 0.025855397805571556, "grad_norm": 0.040340349078178406, "global_step": 662, "epoch": 6, "lr": 9.996139361162691e-05} +{"train_loss": 0.023441027849912643, "grad_norm": 0.09582053869962692, "global_step": 663, "epoch": 6, "lr": 9.996127707915151e-05} +{"train_loss": 0.02777843177318573, "grad_norm": 0.03408433869481087, "global_step": 664, "epoch": 6, "lr": 9.996116037113392e-05} +{"train_loss": 0.02568722516298294, "grad_norm": 0.13389816880226135, "global_step": 665, "epoch": 6, "lr": 9.996104348757453e-05} +{"train_loss": 0.02342548593878746, "grad_norm": 0.049926917999982834, "global_step": 666, "epoch": 6, "lr": 9.996092642847377e-05} +{"train_loss": 0.025607258081436157, "grad_norm": 0.03576529398560524, "global_step": 667, "epoch": 6, "lr": 9.996080919383204e-05} +{"train_loss": 0.0271445345133543, "grad_norm": 0.0727703869342804, "global_step": 668, "epoch": 6, "lr": 9.996069178364974e-05} +{"train_loss": 0.02386590465903282, "grad_norm": 0.06164316087961197, "global_step": 669, "epoch": 6, "lr": 9.996057419792732e-05} +{"train_loss": 0.02121877670288086, "grad_norm": 0.09130055457353592, "global_step": 670, "epoch": 6, "lr": 9.996045643666517e-05} +{"train_loss": 0.023989375680685043, "grad_norm": 0.1847798228263855, "global_step": 671, "epoch": 6, "lr": 9.996033849986368e-05} +{"train_loss": 0.02913149818778038, "grad_norm": 0.04558717459440231, "global_step": 672, "epoch": 6, "lr": 9.996022038752329e-05} +{"train_loss": 0.027440400794148445, "grad_norm": 0.06339239329099655, "global_step": 673, "epoch": 6, "lr": 9.996010209964444e-05} +{"train_loss": 0.02606699801981449, "grad_norm": 0.03725389391183853, "global_step": 674, "epoch": 6, "lr": 9.995998363622749e-05} +{"train_loss": 0.024851227179169655, "grad_norm": 0.04655861109495163, "global_step": 675, "epoch": 6, "lr": 9.99598649972729e-05} +{"train_loss": 0.02497108094394207, "grad_norm": 0.03763166069984436, "global_step": 676, "epoch": 6, "lr": 9.995974618278106e-05} +{"train_loss": 0.023282106965780258, "grad_norm": 0.04697704687714577, "global_step": 677, "epoch": 6, "lr": 9.99596271927524e-05} +{"train_loss": 0.024665486067533493, "grad_norm": 0.12424133718013763, "global_step": 678, "epoch": 6, "lr": 9.995950802718734e-05} +{"train_loss": 0.023104015737771988, "grad_norm": 0.10491424798965454, "global_step": 679, "epoch": 6, "lr": 9.99593886860863e-05} +{"train_loss": 0.024949388578534126, "grad_norm": 0.06878061592578888, "global_step": 680, "epoch": 6, "lr": 9.995926916944969e-05} +{"train_loss": 0.023861829191446304, "grad_norm": 0.041567035019397736, "global_step": 681, "epoch": 6, "lr": 9.995914947727794e-05} +{"train_loss": 0.03020595945417881, "grad_norm": 0.07516157627105713, "global_step": 682, "epoch": 6, "lr": 9.995902960957146e-05} +{"train_loss": 0.023490648716688156, "grad_norm": 0.04870545491576195, "global_step": 683, "epoch": 6, "lr": 9.995890956633068e-05} +{"train_loss": 0.023493055254220963, "grad_norm": 0.07378873229026794, "global_step": 684, "epoch": 6, "lr": 9.995878934755602e-05} +{"train_loss": 0.02476598508656025, "grad_norm": 0.09183268249034882, "global_step": 685, "epoch": 6, "lr": 9.99586689532479e-05} +{"train_loss": 0.02321111038327217, "grad_norm": 0.05074460804462433, "global_step": 686, "epoch": 6, "lr": 9.995854838340675e-05} +{"train_loss": 0.025729238986968994, "grad_norm": 0.10435143858194351, "global_step": 687, "epoch": 6, "lr": 9.995842763803297e-05} +{"train_loss": 0.02390322834253311, "grad_norm": 0.07673035562038422, "global_step": 688, "epoch": 6, "lr": 9.9958306717127e-05} +{"train_loss": 0.029095489531755447, "grad_norm": 0.1898263394832611, "global_step": 689, "epoch": 6, "lr": 9.995818562068928e-05} +{"train_loss": 0.026811808347702026, "grad_norm": 0.06344617158174515, "global_step": 690, "epoch": 6, "lr": 9.995806434872022e-05} +{"train_loss": 0.02133762836456299, "grad_norm": 0.12307902425527573, "global_step": 691, "epoch": 6, "lr": 9.995794290122024e-05} +{"train_loss": 0.02485920675098896, "grad_norm": 0.033708613365888596, "global_step": 692, "epoch": 6, "lr": 9.99578212781898e-05} +{"train_loss": 0.024659939110279083, "grad_norm": 0.04534398764371872, "global_step": 693, "epoch": 6, "lr": 9.995769947962928e-05} +{"train_loss": 0.025555476546287537, "grad_norm": 0.06130712851881981, "global_step": 694, "epoch": 6, "lr": 9.995757750553914e-05} +{"train_loss": 0.021109193563461304, "grad_norm": 0.06626052409410477, "global_step": 695, "epoch": 6, "lr": 9.995745535591978e-05} +{"train_loss": 0.024539701640605927, "grad_norm": 0.14091700315475464, "global_step": 696, "epoch": 6, "lr": 9.995733303077167e-05} +{"train_loss": 0.02770526148378849, "grad_norm": 0.11695130169391632, "global_step": 697, "epoch": 6, "lr": 9.995721053009521e-05} +{"train_loss": 0.021919410675764084, "grad_norm": 0.05101311579346657, "global_step": 698, "epoch": 6, "lr": 9.995708785389083e-05} +{"train_loss": 0.022696178406476974, "grad_norm": 0.0692916214466095, "global_step": 699, "epoch": 6, "lr": 9.995696500215898e-05} +{"train_loss": 0.022566530853509903, "grad_norm": 0.06914246827363968, "global_step": 700, "epoch": 6, "lr": 9.995684197490007e-05} +{"train_loss": 0.023870212957262993, "grad_norm": 0.052614256739616394, "global_step": 701, "epoch": 6, "lr": 9.995671877211456e-05} +{"train_loss": 0.023531842976808548, "grad_norm": 0.054824408143758774, "global_step": 702, "epoch": 6, "lr": 9.995659539380284e-05} +{"train_loss": 0.027309227734804153, "grad_norm": 0.09394708275794983, "global_step": 703, "epoch": 6, "lr": 9.995647183996539e-05} +{"train_loss": 0.022345159202814102, "grad_norm": 0.03886132687330246, "global_step": 704, "epoch": 6, "lr": 9.995634811060262e-05} +{"train_loss": 0.02837766520678997, "grad_norm": 0.033299192786216736, "global_step": 705, "epoch": 6, "lr": 9.995622420571496e-05} +{"train_loss": 0.02429167367517948, "grad_norm": 0.11006367206573486, "global_step": 706, "epoch": 6, "lr": 9.995610012530287e-05} +{"train_loss": 0.025644365698099136, "grad_norm": 0.16020025312900543, "global_step": 707, "epoch": 6, "lr": 9.995597586936676e-05} +{"train_loss": 0.021053118631243706, "grad_norm": 0.04333270713686943, "global_step": 708, "epoch": 6, "lr": 9.995585143790706e-05} +{"train_loss": 0.024378735572099686, "grad_norm": 0.10698118805885315, "global_step": 709, "epoch": 6, "lr": 9.995572683092425e-05} +{"train_loss": 0.026128673925995827, "grad_norm": 0.13401734828948975, "global_step": 710, "epoch": 6, "lr": 9.995560204841873e-05} +{"train_loss": 0.02389475330710411, "grad_norm": 0.0645332932472229, "global_step": 711, "epoch": 6, "lr": 9.995547709039094e-05} +{"train_loss": 0.027263306081295013, "grad_norm": 0.06046357378363609, "global_step": 712, "epoch": 6, "lr": 9.995535195684133e-05} +{"train_loss": 0.02336573787033558, "grad_norm": 0.06702115386724472, "global_step": 713, "epoch": 6, "lr": 9.995522664777036e-05} +{"train_loss": 0.026052607223391533, "grad_norm": 0.06764865666627884, "global_step": 714, "epoch": 6, "lr": 9.995510116317843e-05} +{"train_loss": 0.023855354636907578, "grad_norm": 0.037354208528995514, "global_step": 715, "epoch": 6, "lr": 9.9954975503066e-05} +{"train_loss": 0.021568527445197105, "grad_norm": 0.11663051694631577, "global_step": 716, "epoch": 6, "lr": 9.995484966743351e-05} +{"train_loss": 0.025869358330965042, "grad_norm": 0.07896867394447327, "global_step": 717, "epoch": 6, "lr": 9.995472365628142e-05} +{"train_loss": 0.024484040215611458, "grad_norm": 0.04866373538970947, "global_step": 718, "epoch": 6, "lr": 9.995459746961013e-05} +{"train_loss": 0.026773083955049515, "grad_norm": 0.04769404977560043, "global_step": 719, "epoch": 6, "lr": 9.995447110742014e-05} +{"train_loss": 0.022862665355205536, "grad_norm": 0.0857553780078888, "global_step": 720, "epoch": 6, "lr": 9.995434456971185e-05} +{"train_loss": 0.025153420865535736, "grad_norm": 0.07732073217630386, "global_step": 721, "epoch": 6, "lr": 9.995421785648571e-05} +{"train_loss": 0.02517075277864933, "grad_norm": 0.050421342253685, "global_step": 722, "epoch": 6, "lr": 9.995409096774218e-05} +{"train_loss": 0.026045937091112137, "grad_norm": 0.044695232063531876, "global_step": 723, "epoch": 6, "lr": 9.995396390348169e-05} +{"train_loss": 0.019952833652496338, "grad_norm": 0.04693105071783066, "global_step": 724, "epoch": 6, "lr": 9.99538366637047e-05} +{"train_loss": 0.021182622760534286, "grad_norm": 0.06767791509628296, "global_step": 725, "epoch": 6, "lr": 9.995370924841165e-05} +{"train_loss": 0.023910220712423325, "grad_norm": 0.05954110622406006, "global_step": 726, "epoch": 6, "lr": 9.995358165760299e-05} +{"train_loss": 0.02666987106204033, "grad_norm": 0.05051102489233017, "global_step": 727, "epoch": 6, "lr": 9.995345389127917e-05} +{"train_loss": 0.023753520101308823, "grad_norm": 0.07563203573226929, "global_step": 728, "epoch": 6, "lr": 9.995332594944065e-05} +{"train_loss": 0.026022765785455704, "grad_norm": 0.06372234225273132, "global_step": 729, "epoch": 6, "lr": 9.995319783208785e-05} +{"train_loss": 0.024379435926675797, "grad_norm": 0.0988188087940216, "global_step": 730, "epoch": 6, "lr": 9.995306953922123e-05} +{"train_loss": 0.021467171609401703, "grad_norm": 0.07156986743211746, "global_step": 731, "epoch": 6, "lr": 9.995294107084127e-05} +{"train_loss": 0.024103395640850067, "grad_norm": 0.13194580376148224, "global_step": 732, "epoch": 6, "lr": 9.995281242694838e-05} +{"train_loss": 0.02269287034869194, "grad_norm": 0.0374593511223793, "global_step": 733, "epoch": 6, "lr": 9.995268360754304e-05} +{"train_loss": 0.02444734424352646, "grad_norm": 0.04462895914912224, "global_step": 734, "epoch": 6, "lr": 9.995255461262569e-05} +{"train_loss": 0.027147486805915833, "grad_norm": 0.12084033340215683, "global_step": 735, "epoch": 6, "lr": 9.995242544219679e-05} +{"train_loss": 0.022073082625865936, "grad_norm": 0.0999116525053978, "global_step": 736, "epoch": 6, "lr": 9.99522960962568e-05} +{"train_loss": 0.027975643053650856, "grad_norm": 0.09929677098989487, "global_step": 737, "epoch": 6, "lr": 9.995216657480615e-05} +{"train_loss": 0.022180506959557533, "grad_norm": 0.03693000227212906, "global_step": 738, "epoch": 6, "lr": 9.995203687784532e-05} +{"train_loss": 0.026181867346167564, "grad_norm": 0.05340823531150818, "global_step": 739, "epoch": 6, "lr": 9.995190700537476e-05} +{"train_loss": 0.02419264242053032, "grad_norm": 0.044764984399080276, "global_step": 740, "epoch": 6, "lr": 9.995177695739491e-05} +{"train_loss": 0.02446374701300882, "grad_norm": 0.10143054276704788, "global_step": 741, "epoch": 6, "lr": 9.995164673390625e-05, "val_loss": 0.022893596440553665} +{"train_loss": 0.024481311440467834, "grad_norm": 0.06385926157236099, "global_step": 742, "epoch": 7, "lr": 9.995151633490921e-05} +{"train_loss": 0.02194778248667717, "grad_norm": 0.03161991015076637, "global_step": 743, "epoch": 7, "lr": 9.995138576040428e-05} +{"train_loss": 0.0239906907081604, "grad_norm": 0.05161505565047264, "global_step": 744, "epoch": 7, "lr": 9.995125501039189e-05} +{"train_loss": 0.025086281821131706, "grad_norm": 0.057432763278484344, "global_step": 745, "epoch": 7, "lr": 9.995112408487252e-05} +{"train_loss": 0.024410802870988846, "grad_norm": 0.04955582693219185, "global_step": 746, "epoch": 7, "lr": 9.995099298384664e-05} +{"train_loss": 0.024383869022130966, "grad_norm": 0.15170647203922272, "global_step": 747, "epoch": 7, "lr": 9.995086170731466e-05} +{"train_loss": 0.024764245375990868, "grad_norm": 0.03778636083006859, "global_step": 748, "epoch": 7, "lr": 9.995073025527711e-05} +{"train_loss": 0.02621445432305336, "grad_norm": 0.10234066843986511, "global_step": 749, "epoch": 7, "lr": 9.99505986277344e-05} +{"train_loss": 0.02371455729007721, "grad_norm": 0.1353064775466919, "global_step": 750, "epoch": 7, "lr": 9.995046682468701e-05} +{"train_loss": 0.02156054601073265, "grad_norm": 0.04420989006757736, "global_step": 751, "epoch": 7, "lr": 9.99503348461354e-05} +{"train_loss": 0.025104589760303497, "grad_norm": 0.09630127996206284, "global_step": 752, "epoch": 7, "lr": 9.995020269208004e-05} +{"train_loss": 0.023319873958826065, "grad_norm": 0.04629501700401306, "global_step": 753, "epoch": 7, "lr": 9.99500703625214e-05} +{"train_loss": 0.02544865570962429, "grad_norm": 0.07573874294757843, "global_step": 754, "epoch": 7, "lr": 9.994993785745992e-05} +{"train_loss": 0.024090342223644257, "grad_norm": 0.10000459104776382, "global_step": 755, "epoch": 7, "lr": 9.994980517689608e-05} +{"train_loss": 0.024991322308778763, "grad_norm": 0.057938121259212494, "global_step": 756, "epoch": 7, "lr": 9.994967232083036e-05} +{"train_loss": 0.022845538333058357, "grad_norm": 0.1331498771905899, "global_step": 757, "epoch": 7, "lr": 9.99495392892632e-05} +{"train_loss": 0.02471877820789814, "grad_norm": 0.06174832209944725, "global_step": 758, "epoch": 7, "lr": 9.99494060821951e-05} +{"train_loss": 0.024655789136886597, "grad_norm": 0.048813942819833755, "global_step": 759, "epoch": 7, "lr": 9.994927269962649e-05} +{"train_loss": 0.026835329830646515, "grad_norm": 0.04774100333452225, "global_step": 760, "epoch": 7, "lr": 9.994913914155787e-05} +{"train_loss": 0.023398449644446373, "grad_norm": 0.12312689423561096, "global_step": 761, "epoch": 7, "lr": 9.994900540798968e-05} +{"train_loss": 0.02453641965985298, "grad_norm": 0.15327024459838867, "global_step": 762, "epoch": 7, "lr": 9.994887149892242e-05} +{"train_loss": 0.02679648995399475, "grad_norm": 0.08779709041118622, "global_step": 763, "epoch": 7, "lr": 9.994873741435655e-05} +{"train_loss": 0.020905155688524246, "grad_norm": 0.032071780413389206, "global_step": 764, "epoch": 7, "lr": 9.994860315429256e-05} +{"train_loss": 0.022310107946395874, "grad_norm": 0.1588813215494156, "global_step": 765, "epoch": 7, "lr": 9.994846871873087e-05} +{"train_loss": 0.03135273978114128, "grad_norm": 0.10416566580533981, "global_step": 766, "epoch": 7, "lr": 9.994833410767199e-05} +{"train_loss": 0.02653493359684944, "grad_norm": 0.06772644817829132, "global_step": 767, "epoch": 7, "lr": 9.994819932111639e-05} +{"train_loss": 0.02009054645895958, "grad_norm": 0.03380943834781647, "global_step": 768, "epoch": 7, "lr": 9.994806435906453e-05} +{"train_loss": 0.01954568736255169, "grad_norm": 0.035109370946884155, "global_step": 769, "epoch": 7, "lr": 9.99479292215169e-05} +{"train_loss": 0.02618616819381714, "grad_norm": 0.05919992923736572, "global_step": 770, "epoch": 7, "lr": 9.994779390847398e-05} +{"train_loss": 0.024165932089090347, "grad_norm": 0.18728934228420258, "global_step": 771, "epoch": 7, "lr": 9.99476584199362e-05} +{"train_loss": 0.022715911269187927, "grad_norm": 0.13077586889266968, "global_step": 772, "epoch": 7, "lr": 9.994752275590411e-05} +{"train_loss": 0.020879942923784256, "grad_norm": 0.036998823285102844, "global_step": 773, "epoch": 7, "lr": 9.994738691637812e-05} +{"train_loss": 0.023414533585309982, "grad_norm": 0.06825239211320877, "global_step": 774, "epoch": 7, "lr": 9.994725090135874e-05} +{"train_loss": 0.019288934767246246, "grad_norm": 0.046281374990940094, "global_step": 775, "epoch": 7, "lr": 9.994711471084645e-05} +{"train_loss": 0.023252766579389572, "grad_norm": 0.09283424913883209, "global_step": 776, "epoch": 7, "lr": 9.994697834484173e-05} +{"train_loss": 0.02343771606683731, "grad_norm": 0.056581173092126846, "global_step": 777, "epoch": 7, "lr": 9.994684180334502e-05} +{"train_loss": 0.02205924317240715, "grad_norm": 0.041034866124391556, "global_step": 778, "epoch": 7, "lr": 9.994670508635684e-05} +{"train_loss": 0.022786933928728104, "grad_norm": 0.13072450459003448, "global_step": 779, "epoch": 7, "lr": 9.994656819387767e-05} +{"train_loss": 0.026078423485159874, "grad_norm": 0.12970030307769775, "global_step": 780, "epoch": 7, "lr": 9.994643112590799e-05} +{"train_loss": 0.028235817328095436, "grad_norm": 0.057276852428913116, "global_step": 781, "epoch": 7, "lr": 9.994629388244824e-05} +{"train_loss": 0.02536897547543049, "grad_norm": 0.06167539954185486, "global_step": 782, "epoch": 7, "lr": 9.994615646349894e-05} +{"train_loss": 0.022319039329886436, "grad_norm": 0.07407258450984955, "global_step": 783, "epoch": 7, "lr": 9.994601886906059e-05} +{"train_loss": 0.02304467186331749, "grad_norm": 0.053421393036842346, "global_step": 784, "epoch": 7, "lr": 9.994588109913363e-05} +{"train_loss": 0.024279078468680382, "grad_norm": 0.03565121442079544, "global_step": 785, "epoch": 7, "lr": 9.994574315371858e-05} +{"train_loss": 0.02152155339717865, "grad_norm": 0.07626047730445862, "global_step": 786, "epoch": 7, "lr": 9.99456050328159e-05} +{"train_loss": 0.023476647213101387, "grad_norm": 0.0775323137640953, "global_step": 787, "epoch": 7, "lr": 9.994546673642608e-05} +{"train_loss": 0.023760909214615822, "grad_norm": 0.031119301915168762, "global_step": 788, "epoch": 7, "lr": 9.994532826454964e-05} +{"train_loss": 0.024070337414741516, "grad_norm": 0.04922770708799362, "global_step": 789, "epoch": 7, "lr": 9.9945189617187e-05} +{"train_loss": 0.022316252812743187, "grad_norm": 0.07276342809200287, "global_step": 790, "epoch": 7, "lr": 9.99450507943387e-05} +{"train_loss": 0.025579329580068588, "grad_norm": 0.10246720165014267, "global_step": 791, "epoch": 7, "lr": 9.99449117960052e-05} +{"train_loss": 0.024382196366786957, "grad_norm": 0.051188357174396515, "global_step": 792, "epoch": 7, "lr": 9.994477262218702e-05} +{"train_loss": 0.023782778531312943, "grad_norm": 0.05120649188756943, "global_step": 793, "epoch": 7, "lr": 9.994463327288463e-05} +{"train_loss": 0.022929370403289795, "grad_norm": 0.09116789698600769, "global_step": 794, "epoch": 7, "lr": 9.994449374809851e-05} +{"train_loss": 0.02462753653526306, "grad_norm": 0.07652758061885834, "global_step": 795, "epoch": 7, "lr": 9.994435404782916e-05} +{"train_loss": 0.028691967949271202, "grad_norm": 0.043570395559072495, "global_step": 796, "epoch": 7, "lr": 9.994421417207706e-05} +{"train_loss": 0.023028291761875153, "grad_norm": 0.04478859901428223, "global_step": 797, "epoch": 7, "lr": 9.994407412084273e-05} +{"train_loss": 0.023072347044944763, "grad_norm": 0.04837901517748833, "global_step": 798, "epoch": 7, "lr": 9.994393389412662e-05} +{"train_loss": 0.02626314014196396, "grad_norm": 0.07031811028718948, "global_step": 799, "epoch": 7, "lr": 9.994379349192926e-05} +{"train_loss": 0.02384466864168644, "grad_norm": 0.07935655862092972, "global_step": 800, "epoch": 7, "lr": 9.994365291425114e-05} +{"train_loss": 0.023721609264612198, "grad_norm": 0.047578733414411545, "global_step": 801, "epoch": 7, "lr": 9.994351216109274e-05} +{"train_loss": 0.021162066608667374, "grad_norm": 0.0688418447971344, "global_step": 802, "epoch": 7, "lr": 9.994337123245454e-05} +{"train_loss": 0.02079099416732788, "grad_norm": 0.038967326283454895, "global_step": 803, "epoch": 7, "lr": 9.994323012833707e-05} +{"train_loss": 0.025493625551462173, "grad_norm": 0.09142347425222397, "global_step": 804, "epoch": 7, "lr": 9.994308884874081e-05} +{"train_loss": 0.02322106808423996, "grad_norm": 0.039460793137550354, "global_step": 805, "epoch": 7, "lr": 9.994294739366626e-05} +{"train_loss": 0.022387705743312836, "grad_norm": 0.07227490097284317, "global_step": 806, "epoch": 7, "lr": 9.994280576311389e-05} +{"train_loss": 0.02417765185236931, "grad_norm": 0.03407652676105499, "global_step": 807, "epoch": 7, "lr": 9.994266395708423e-05} +{"train_loss": 0.025713255628943443, "grad_norm": 0.12109557539224625, "global_step": 808, "epoch": 7, "lr": 9.994252197557778e-05} +{"train_loss": 0.024872735142707825, "grad_norm": 0.057025447487831116, "global_step": 809, "epoch": 7, "lr": 9.9942379818595e-05} +{"train_loss": 0.021056674420833588, "grad_norm": 0.058230262249708176, "global_step": 810, "epoch": 7, "lr": 9.994223748613644e-05} +{"train_loss": 0.023664051666855812, "grad_norm": 0.03624237701296806, "global_step": 811, "epoch": 7, "lr": 9.994209497820258e-05} +{"train_loss": 0.02207186445593834, "grad_norm": 0.04183812811970711, "global_step": 812, "epoch": 7, "lr": 9.994195229479391e-05} +{"train_loss": 0.02619234099984169, "grad_norm": 0.07019264250993729, "global_step": 813, "epoch": 7, "lr": 9.994180943591094e-05} +{"train_loss": 0.023077841848134995, "grad_norm": 0.05290014296770096, "global_step": 814, "epoch": 7, "lr": 9.994166640155416e-05} +{"train_loss": 0.02294785901904106, "grad_norm": 0.08013325929641724, "global_step": 815, "epoch": 7, "lr": 9.99415231917241e-05} +{"train_loss": 0.02452395111322403, "grad_norm": 0.044469885528087616, "global_step": 816, "epoch": 7, "lr": 9.994137980642123e-05} +{"train_loss": 0.024329178035259247, "grad_norm": 0.05431082472205162, "global_step": 817, "epoch": 7, "lr": 9.994123624564609e-05} +{"train_loss": 0.02439882978796959, "grad_norm": 0.057940009981393814, "global_step": 818, "epoch": 7, "lr": 9.994109250939915e-05} +{"train_loss": 0.024809706956148148, "grad_norm": 0.09079520404338837, "global_step": 819, "epoch": 7, "lr": 9.994094859768092e-05} +{"train_loss": 0.025111712515354156, "grad_norm": 0.050921402871608734, "global_step": 820, "epoch": 7, "lr": 9.994080451049194e-05} +{"train_loss": 0.027552209794521332, "grad_norm": 0.08785727620124817, "global_step": 821, "epoch": 7, "lr": 9.994066024783268e-05} +{"train_loss": 0.023275408893823624, "grad_norm": 0.08063554763793945, "global_step": 822, "epoch": 7, "lr": 9.994051580970365e-05} +{"train_loss": 0.024002796038985252, "grad_norm": 0.07104945182800293, "global_step": 823, "epoch": 7, "lr": 9.994037119610537e-05} +{"train_loss": 0.028876150026917458, "grad_norm": 0.09114386141300201, "global_step": 824, "epoch": 7, "lr": 9.994022640703836e-05} +{"train_loss": 0.02182900905609131, "grad_norm": 0.03692994266748428, "global_step": 825, "epoch": 7, "lr": 9.994008144250311e-05} +{"train_loss": 0.023444652557373047, "grad_norm": 0.06209272891283035, "global_step": 826, "epoch": 7, "lr": 9.993993630250011e-05} +{"train_loss": 0.026875915005803108, "grad_norm": 0.03647319972515106, "global_step": 827, "epoch": 7, "lr": 9.993979098702991e-05} +{"train_loss": 0.023892473429441452, "grad_norm": 0.0801064670085907, "global_step": 828, "epoch": 7, "lr": 9.9939645496093e-05} +{"train_loss": 0.022226659581065178, "grad_norm": 0.047145452350378036, "global_step": 829, "epoch": 7, "lr": 9.993949982968989e-05} +{"train_loss": 0.024579806253314018, "grad_norm": 0.10463175177574158, "global_step": 830, "epoch": 7, "lr": 9.99393539878211e-05} +{"train_loss": 0.02699689194560051, "grad_norm": 0.031596019864082336, "global_step": 831, "epoch": 7, "lr": 9.993920797048715e-05} +{"train_loss": 0.024732321500778198, "grad_norm": 0.17781420052051544, "global_step": 832, "epoch": 7, "lr": 9.993906177768852e-05} +{"train_loss": 0.02287539467215538, "grad_norm": 0.03103785589337349, "global_step": 833, "epoch": 7, "lr": 9.993891540942574e-05} +{"train_loss": 0.025563012808561325, "grad_norm": 0.13796274363994598, "global_step": 834, "epoch": 7, "lr": 9.993876886569936e-05} +{"train_loss": 0.02367648109793663, "grad_norm": 0.03795648366212845, "global_step": 835, "epoch": 7, "lr": 9.993862214650984e-05} +{"train_loss": 0.02271764725446701, "grad_norm": 0.07141869515180588, "global_step": 836, "epoch": 7, "lr": 9.993847525185772e-05} +{"train_loss": 0.026050932705402374, "grad_norm": 0.03684188798069954, "global_step": 837, "epoch": 7, "lr": 9.993832818174352e-05} +{"train_loss": 0.02374020405113697, "grad_norm": 0.10811394453048706, "global_step": 838, "epoch": 7, "lr": 9.993818093616776e-05} +{"train_loss": 0.022658228874206543, "grad_norm": 0.05375811830163002, "global_step": 839, "epoch": 7, "lr": 9.993803351513094e-05} +{"train_loss": 0.023312266916036606, "grad_norm": 0.049353472888469696, "global_step": 840, "epoch": 7, "lr": 9.99378859186336e-05} +{"train_loss": 0.018915802240371704, "grad_norm": 0.037961672991514206, "global_step": 841, "epoch": 7, "lr": 9.993773814667623e-05} +{"train_loss": 0.024096667766571045, "grad_norm": 0.03784877806901932, "global_step": 842, "epoch": 7, "lr": 9.993759019925937e-05} +{"train_loss": 0.02628188207745552, "grad_norm": 0.12274733185768127, "global_step": 843, "epoch": 7, "lr": 9.993744207638353e-05} +{"train_loss": 0.02584640495479107, "grad_norm": 0.07710486650466919, "global_step": 844, "epoch": 7, "lr": 9.993729377804925e-05} +{"train_loss": 0.023883264511823654, "grad_norm": 0.07473551481962204, "global_step": 845, "epoch": 7, "lr": 9.993714530425703e-05} +{"train_loss": 0.027837099507451057, "grad_norm": 0.08328834921121597, "global_step": 846, "epoch": 7, "lr": 9.99369966550074e-05} +{"train_loss": 0.02403221592166514, "grad_norm": 0.05165523290634155, "global_step": 847, "epoch": 7, "lr": 9.993684783030088e-05, "val_loss": 0.02230124920606613} +{"train_loss": 0.020257674157619476, "grad_norm": 0.1348002552986145, "global_step": 848, "epoch": 8, "lr": 9.993669883013799e-05} +{"train_loss": 0.024310067296028137, "grad_norm": 0.11286945641040802, "global_step": 849, "epoch": 8, "lr": 9.993654965451926e-05} +{"train_loss": 0.019658230245113373, "grad_norm": 0.04577208310365677, "global_step": 850, "epoch": 8, "lr": 9.993640030344522e-05} +{"train_loss": 0.021440569311380386, "grad_norm": 0.050596125423908234, "global_step": 851, "epoch": 8, "lr": 9.993625077691638e-05} +{"train_loss": 0.025919750332832336, "grad_norm": 0.037702709436416626, "global_step": 852, "epoch": 8, "lr": 9.993610107493327e-05} +{"train_loss": 0.024776093661785126, "grad_norm": 0.060551486909389496, "global_step": 853, "epoch": 8, "lr": 9.993595119749642e-05} +{"train_loss": 0.024531345814466476, "grad_norm": 0.06438316404819489, "global_step": 854, "epoch": 8, "lr": 9.993580114460634e-05} +{"train_loss": 0.02349284291267395, "grad_norm": 0.04247337579727173, "global_step": 855, "epoch": 8, "lr": 9.993565091626358e-05} +{"train_loss": 0.02402130514383316, "grad_norm": 0.040215857326984406, "global_step": 856, "epoch": 8, "lr": 9.993550051246867e-05} +{"train_loss": 0.025228282436728477, "grad_norm": 0.032729800790548325, "global_step": 857, "epoch": 8, "lr": 9.993534993322211e-05} +{"train_loss": 0.02156033366918564, "grad_norm": 0.04231743887066841, "global_step": 858, "epoch": 8, "lr": 9.993519917852445e-05} +{"train_loss": 0.02262689173221588, "grad_norm": 0.03481203317642212, "global_step": 859, "epoch": 8, "lr": 9.993504824837623e-05} +{"train_loss": 0.022422177717089653, "grad_norm": 0.03463790938258171, "global_step": 860, "epoch": 8, "lr": 9.993489714277796e-05} +{"train_loss": 0.025778841227293015, "grad_norm": 0.0454288087785244, "global_step": 861, "epoch": 8, "lr": 9.993474586173018e-05} +{"train_loss": 0.02306818589568138, "grad_norm": 0.06088769808411598, "global_step": 862, "epoch": 8, "lr": 9.99345944052334e-05} +{"train_loss": 0.022902827709913254, "grad_norm": 0.04420963674783707, "global_step": 863, "epoch": 8, "lr": 9.993444277328819e-05} +{"train_loss": 0.023619554936885834, "grad_norm": 0.0623861588537693, "global_step": 864, "epoch": 8, "lr": 9.993429096589506e-05} +{"train_loss": 0.022869568318128586, "grad_norm": 0.05397766828536987, "global_step": 865, "epoch": 8, "lr": 9.993413898305455e-05} +{"train_loss": 0.024947328492999077, "grad_norm": 0.04726733639836311, "global_step": 866, "epoch": 8, "lr": 9.99339868247672e-05} +{"train_loss": 0.02463267184793949, "grad_norm": 0.03630806505680084, "global_step": 867, "epoch": 8, "lr": 9.993383449103352e-05} +{"train_loss": 0.025633554905653, "grad_norm": 0.030818145722150803, "global_step": 868, "epoch": 8, "lr": 9.993368198185407e-05} +{"train_loss": 0.02496933564543724, "grad_norm": 0.13737335801124573, "global_step": 869, "epoch": 8, "lr": 9.993352929722937e-05} +{"train_loss": 0.028043631464242935, "grad_norm": 0.05224386230111122, "global_step": 870, "epoch": 8, "lr": 9.993337643715996e-05} +{"train_loss": 0.02397143468260765, "grad_norm": 0.03779594227671623, "global_step": 871, "epoch": 8, "lr": 9.993322340164639e-05} +{"train_loss": 0.021894508972764015, "grad_norm": 0.10300884395837784, "global_step": 872, "epoch": 8, "lr": 9.993307019068919e-05} +{"train_loss": 0.02644713781774044, "grad_norm": 0.06089048087596893, "global_step": 873, "epoch": 8, "lr": 9.993291680428889e-05} +{"train_loss": 0.023067377507686615, "grad_norm": 0.08793345838785172, "global_step": 874, "epoch": 8, "lr": 9.993276324244604e-05} +{"train_loss": 0.021118253469467163, "grad_norm": 0.0869532898068428, "global_step": 875, "epoch": 8, "lr": 9.993260950516116e-05} +{"train_loss": 0.021371612325310707, "grad_norm": 0.06707099825143814, "global_step": 876, "epoch": 8, "lr": 9.993245559243482e-05} +{"train_loss": 0.024113232269883156, "grad_norm": 0.05387115851044655, "global_step": 877, "epoch": 8, "lr": 9.993230150426756e-05} +{"train_loss": 0.02713455632328987, "grad_norm": 0.05006979778409004, "global_step": 878, "epoch": 8, "lr": 9.99321472406599e-05} +{"train_loss": 0.02451397478580475, "grad_norm": 0.0727498009800911, "global_step": 879, "epoch": 8, "lr": 9.993199280161237e-05} +{"train_loss": 0.025556087493896484, "grad_norm": 0.03781704604625702, "global_step": 880, "epoch": 8, "lr": 9.993183818712554e-05} +{"train_loss": 0.023649360984563828, "grad_norm": 0.046400029212236404, "global_step": 881, "epoch": 8, "lr": 9.993168339719996e-05} +{"train_loss": 0.02366858534514904, "grad_norm": 0.049223996698856354, "global_step": 882, "epoch": 8, "lr": 9.993152843183615e-05} +{"train_loss": 0.026957551017403603, "grad_norm": 0.06409656256437302, "global_step": 883, "epoch": 8, "lr": 9.993137329103466e-05} +{"train_loss": 0.023520704358816147, "grad_norm": 0.030988767743110657, "global_step": 884, "epoch": 8, "lr": 9.993121797479603e-05} +{"train_loss": 0.02433251030743122, "grad_norm": 0.059055864810943604, "global_step": 885, "epoch": 8, "lr": 9.993106248312083e-05} +{"train_loss": 0.021312646567821503, "grad_norm": 0.08575119078159332, "global_step": 886, "epoch": 8, "lr": 9.99309068160096e-05} +{"train_loss": 0.027836062014102936, "grad_norm": 0.11349491775035858, "global_step": 887, "epoch": 8, "lr": 9.993075097346285e-05} +{"train_loss": 0.022617124021053314, "grad_norm": 0.043852198868989944, "global_step": 888, "epoch": 8, "lr": 9.993059495548118e-05} +{"train_loss": 0.02379198744893074, "grad_norm": 0.040885426104068756, "global_step": 889, "epoch": 8, "lr": 9.993043876206509e-05} +{"train_loss": 0.023020297288894653, "grad_norm": 0.04809102043509483, "global_step": 890, "epoch": 8, "lr": 9.993028239321518e-05} +{"train_loss": 0.022095534950494766, "grad_norm": 0.06288284808397293, "global_step": 891, "epoch": 8, "lr": 9.993012584893196e-05} +{"train_loss": 0.024179380387067795, "grad_norm": 0.03032318875193596, "global_step": 892, "epoch": 8, "lr": 9.992996912921599e-05} +{"train_loss": 0.02376997284591198, "grad_norm": 0.040772322565317154, "global_step": 893, "epoch": 8, "lr": 9.99298122340678e-05} +{"train_loss": 0.025965996086597443, "grad_norm": 0.1250365972518921, "global_step": 894, "epoch": 8, "lr": 9.9929655163488e-05} +{"train_loss": 0.0211808979511261, "grad_norm": 0.07259642332792282, "global_step": 895, "epoch": 8, "lr": 9.99294979174771e-05} +{"train_loss": 0.02621389739215374, "grad_norm": 0.07552099972963333, "global_step": 896, "epoch": 8, "lr": 9.992934049603563e-05} +{"train_loss": 0.02423478290438652, "grad_norm": 0.1469060331583023, "global_step": 897, "epoch": 8, "lr": 9.99291828991642e-05} +{"train_loss": 0.02680514194071293, "grad_norm": 0.0615532360970974, "global_step": 898, "epoch": 8, "lr": 9.992902512686333e-05} +{"train_loss": 0.021748099476099014, "grad_norm": 0.11070206016302109, "global_step": 899, "epoch": 8, "lr": 9.992886717913356e-05} +{"train_loss": 0.021923812106251717, "grad_norm": 0.09569307416677475, "global_step": 900, "epoch": 8, "lr": 9.992870905597548e-05} +{"train_loss": 0.024387221783399582, "grad_norm": 0.035566918551921844, "global_step": 901, "epoch": 8, "lr": 9.992855075738964e-05} +{"train_loss": 0.026458866894245148, "grad_norm": 0.05499396100640297, "global_step": 902, "epoch": 8, "lr": 9.992839228337658e-05} +{"train_loss": 0.026774223893880844, "grad_norm": 0.03835749253630638, "global_step": 903, "epoch": 8, "lr": 9.992823363393685e-05} +{"train_loss": 0.02283976972103119, "grad_norm": 0.042717672884464264, "global_step": 904, "epoch": 8, "lr": 9.992807480907103e-05} +{"train_loss": 0.024294378235936165, "grad_norm": 0.04556431248784065, "global_step": 905, "epoch": 8, "lr": 9.992791580877968e-05} +{"train_loss": 0.022504128515720367, "grad_norm": 0.05400913953781128, "global_step": 906, "epoch": 8, "lr": 9.992775663306331e-05} +{"train_loss": 0.02250947803258896, "grad_norm": 0.04284030944108963, "global_step": 907, "epoch": 8, "lr": 9.992759728192255e-05} +{"train_loss": 0.02429153397679329, "grad_norm": 0.09588974714279175, "global_step": 908, "epoch": 8, "lr": 9.992743775535792e-05} +{"train_loss": 0.02356547862291336, "grad_norm": 0.08230867236852646, "global_step": 909, "epoch": 8, "lr": 9.992727805336998e-05} +{"train_loss": 0.023347929120063782, "grad_norm": 0.07684620469808578, "global_step": 910, "epoch": 8, "lr": 9.99271181759593e-05} +{"train_loss": 0.02427683025598526, "grad_norm": 0.07739230245351791, "global_step": 911, "epoch": 8, "lr": 9.992695812312643e-05} +{"train_loss": 0.020671643316745758, "grad_norm": 0.035045165568590164, "global_step": 912, "epoch": 8, "lr": 9.992679789487196e-05} +{"train_loss": 0.021776951849460602, "grad_norm": 0.055356431752443314, "global_step": 913, "epoch": 8, "lr": 9.992663749119642e-05} +{"train_loss": 0.024368293583393097, "grad_norm": 0.10992206633090973, "global_step": 914, "epoch": 8, "lr": 9.992647691210039e-05} +{"train_loss": 0.022994060069322586, "grad_norm": 0.04710695892572403, "global_step": 915, "epoch": 8, "lr": 9.992631615758444e-05} +{"train_loss": 0.02483922429382801, "grad_norm": 0.04374173283576965, "global_step": 916, "epoch": 8, "lr": 9.992615522764913e-05} +{"train_loss": 0.02362971007823944, "grad_norm": 0.09208498895168304, "global_step": 917, "epoch": 8, "lr": 9.992599412229503e-05} +{"train_loss": 0.02436947636306286, "grad_norm": 0.10780512541532516, "global_step": 918, "epoch": 8, "lr": 9.992583284152266e-05} +{"train_loss": 0.023343868553638458, "grad_norm": 0.05131452903151512, "global_step": 919, "epoch": 8, "lr": 9.992567138533267e-05} +{"train_loss": 0.023401333019137383, "grad_norm": 0.038638919591903687, "global_step": 920, "epoch": 8, "lr": 9.992550975372555e-05} +{"train_loss": 0.02265588566660881, "grad_norm": 0.05601770430803299, "global_step": 921, "epoch": 8, "lr": 9.992534794670192e-05} +{"train_loss": 0.02393706887960434, "grad_norm": 0.1332544982433319, "global_step": 922, "epoch": 8, "lr": 9.992518596426233e-05} +{"train_loss": 0.023374106734991074, "grad_norm": 0.16408808529376984, "global_step": 923, "epoch": 8, "lr": 9.992502380640734e-05} +{"train_loss": 0.024296289309859276, "grad_norm": 0.04345828667283058, "global_step": 924, "epoch": 8, "lr": 9.992486147313754e-05} +{"train_loss": 0.02696261927485466, "grad_norm": 0.05764070898294449, "global_step": 925, "epoch": 8, "lr": 9.992469896445347e-05} +{"train_loss": 0.023005641996860504, "grad_norm": 0.07203597575426102, "global_step": 926, "epoch": 8, "lr": 9.992453628035572e-05} +{"train_loss": 0.01935759000480175, "grad_norm": 0.04163389652967453, "global_step": 927, "epoch": 8, "lr": 9.992437342084488e-05} +{"train_loss": 0.0244230218231678, "grad_norm": 0.06032711640000343, "global_step": 928, "epoch": 8, "lr": 9.992421038592148e-05} +{"train_loss": 0.02291637472808361, "grad_norm": 0.06046034395694733, "global_step": 929, "epoch": 8, "lr": 9.992404717558614e-05} +{"train_loss": 0.02288016490638256, "grad_norm": 0.05468236282467842, "global_step": 930, "epoch": 8, "lr": 9.992388378983939e-05} +{"train_loss": 0.024730633944272995, "grad_norm": 0.05852143093943596, "global_step": 931, "epoch": 8, "lr": 9.992372022868184e-05} +{"train_loss": 0.02595534361898899, "grad_norm": 0.10928677022457123, "global_step": 932, "epoch": 8, "lr": 9.992355649211404e-05} +{"train_loss": 0.02729123830795288, "grad_norm": 0.12984882295131683, "global_step": 933, "epoch": 8, "lr": 9.992339258013658e-05} +{"train_loss": 0.024781208485364914, "grad_norm": 0.09405621886253357, "global_step": 934, "epoch": 8, "lr": 9.992322849275001e-05} +{"train_loss": 0.023468445986509323, "grad_norm": 0.03232515975832939, "global_step": 935, "epoch": 8, "lr": 9.992306422995494e-05} +{"train_loss": 0.021378543227910995, "grad_norm": 0.07201552391052246, "global_step": 936, "epoch": 8, "lr": 9.992289979175194e-05} +{"train_loss": 0.02118390053510666, "grad_norm": 0.14125485718250275, "global_step": 937, "epoch": 8, "lr": 9.992273517814157e-05} +{"train_loss": 0.02442728728055954, "grad_norm": 0.0406055711209774, "global_step": 938, "epoch": 8, "lr": 9.992257038912442e-05} +{"train_loss": 0.024213500320911407, "grad_norm": 0.11743393540382385, "global_step": 939, "epoch": 8, "lr": 9.992240542470108e-05} +{"train_loss": 0.024812906980514526, "grad_norm": 0.04222873970866203, "global_step": 940, "epoch": 8, "lr": 9.992224028487211e-05} +{"train_loss": 0.02097054198384285, "grad_norm": 0.09513659030199051, "global_step": 941, "epoch": 8, "lr": 9.99220749696381e-05} +{"train_loss": 0.022553347051143646, "grad_norm": 0.03233823925256729, "global_step": 942, "epoch": 8, "lr": 9.992190947899963e-05} +{"train_loss": 0.02410499006509781, "grad_norm": 0.07958821952342987, "global_step": 943, "epoch": 8, "lr": 9.992174381295728e-05} +{"train_loss": 0.02264607511460781, "grad_norm": 0.07505935430526733, "global_step": 944, "epoch": 8, "lr": 9.992157797151163e-05} +{"train_loss": 0.02273840829730034, "grad_norm": 0.08289269357919693, "global_step": 945, "epoch": 8, "lr": 9.992141195466327e-05} +{"train_loss": 0.02212490513920784, "grad_norm": 0.08456418663263321, "global_step": 946, "epoch": 8, "lr": 9.992124576241277e-05} +{"train_loss": 0.02322084829211235, "grad_norm": 0.08039689064025879, "global_step": 947, "epoch": 8, "lr": 9.992107939476073e-05} +{"train_loss": 0.02515498176217079, "grad_norm": 0.04073395952582359, "global_step": 948, "epoch": 8, "lr": 9.992091285170772e-05} +{"train_loss": 0.02292991429567337, "grad_norm": 0.038847118616104126, "global_step": 949, "epoch": 8, "lr": 9.992074613325434e-05} +{"train_loss": 0.023679591715335846, "grad_norm": 0.15496882796287537, "global_step": 950, "epoch": 8, "lr": 9.992057923940117e-05} +{"train_loss": 0.023074615746736526, "grad_norm": 0.07276999205350876, "global_step": 951, "epoch": 8, "lr": 9.992041217014878e-05} +{"train_loss": 0.021855950355529785, "grad_norm": 0.06232272833585739, "global_step": 952, "epoch": 8, "lr": 9.992024492549779e-05} +{"train_loss": 0.02374222976559738, "grad_norm": 0.15777330100536346, "global_step": 953, "epoch": 8, "lr": 9.992007750544876e-05, "val_loss": 0.021691296249628067} +{"train_loss": 0.02030107192695141, "grad_norm": 0.03637665882706642, "global_step": 954, "epoch": 9, "lr": 9.991990991000229e-05} +{"train_loss": 0.02400142140686512, "grad_norm": 0.10110015422105789, "global_step": 955, "epoch": 9, "lr": 9.991974213915895e-05} +{"train_loss": 0.02441456913948059, "grad_norm": 0.03553793206810951, "global_step": 956, "epoch": 9, "lr": 9.991957419291937e-05} +{"train_loss": 0.020944947376847267, "grad_norm": 0.03590190038084984, "global_step": 957, "epoch": 9, "lr": 9.991940607128409e-05} +{"train_loss": 0.025220390409231186, "grad_norm": 0.1090133935213089, "global_step": 958, "epoch": 9, "lr": 9.991923777425375e-05} +{"train_loss": 0.02085680514574051, "grad_norm": 0.04528505727648735, "global_step": 959, "epoch": 9, "lr": 9.99190693018289e-05} +{"train_loss": 0.024014079943299294, "grad_norm": 0.03551696613430977, "global_step": 960, "epoch": 9, "lr": 9.991890065401017e-05} +{"train_loss": 0.027983151376247406, "grad_norm": 0.08710525929927826, "global_step": 961, "epoch": 9, "lr": 9.991873183079811e-05} +{"train_loss": 0.02091274969279766, "grad_norm": 0.039517369121313095, "global_step": 962, "epoch": 9, "lr": 9.991856283219337e-05} +{"train_loss": 0.019876746460795403, "grad_norm": 0.04400007054209709, "global_step": 963, "epoch": 9, "lr": 9.991839365819647e-05} +{"train_loss": 0.02329144813120365, "grad_norm": 0.10993508994579315, "global_step": 964, "epoch": 9, "lr": 9.991822430880806e-05} +{"train_loss": 0.02219141274690628, "grad_norm": 0.047580309212207794, "global_step": 965, "epoch": 9, "lr": 9.991805478402871e-05} +{"train_loss": 0.021788477897644043, "grad_norm": 0.037884194403886795, "global_step": 966, "epoch": 9, "lr": 9.991788508385904e-05} +{"train_loss": 0.026515211910009384, "grad_norm": 0.03445034846663475, "global_step": 967, "epoch": 9, "lr": 9.991771520829962e-05} +{"train_loss": 0.020589880645275116, "grad_norm": 0.04258904978632927, "global_step": 968, "epoch": 9, "lr": 9.991754515735106e-05} +{"train_loss": 0.02174043282866478, "grad_norm": 0.04002440720796585, "global_step": 969, "epoch": 9, "lr": 9.991737493101397e-05} +{"train_loss": 0.02311088517308235, "grad_norm": 0.03932361304759979, "global_step": 970, "epoch": 9, "lr": 9.99172045292889e-05} +{"train_loss": 0.023611275479197502, "grad_norm": 0.05263686180114746, "global_step": 971, "epoch": 9, "lr": 9.99170339521765e-05} +{"train_loss": 0.025423455983400345, "grad_norm": 0.0784611776471138, "global_step": 972, "epoch": 9, "lr": 9.991686319967736e-05} +{"train_loss": 0.024429690092802048, "grad_norm": 0.0434182845056057, "global_step": 973, "epoch": 9, "lr": 9.991669227179205e-05} +{"train_loss": 0.024482466280460358, "grad_norm": 0.04777205362915993, "global_step": 974, "epoch": 9, "lr": 9.991652116852122e-05} +{"train_loss": 0.025640936568379402, "grad_norm": 0.05949629843235016, "global_step": 975, "epoch": 9, "lr": 9.991634988986542e-05} +{"train_loss": 0.021876970306038857, "grad_norm": 0.0461338572204113, "global_step": 976, "epoch": 9, "lr": 9.991617843582528e-05} +{"train_loss": 0.020566891878843307, "grad_norm": 0.09052138030529022, "global_step": 977, "epoch": 9, "lr": 9.99160068064014e-05} +{"train_loss": 0.023396167904138565, "grad_norm": 0.04730086401104927, "global_step": 978, "epoch": 9, "lr": 9.991583500159438e-05} +{"train_loss": 0.025693248957395554, "grad_norm": 0.04894231632351875, "global_step": 979, "epoch": 9, "lr": 9.991566302140482e-05} +{"train_loss": 0.023689769208431244, "grad_norm": 0.09407670795917511, "global_step": 980, "epoch": 9, "lr": 9.991549086583333e-05} +{"train_loss": 0.023891732096672058, "grad_norm": 0.03393418341875076, "global_step": 981, "epoch": 9, "lr": 9.991531853488051e-05} +{"train_loss": 0.020894335582852364, "grad_norm": 0.09121713787317276, "global_step": 982, "epoch": 9, "lr": 9.991514602854697e-05} +{"train_loss": 0.018863102421164513, "grad_norm": 0.0380987711250782, "global_step": 983, "epoch": 9, "lr": 9.99149733468333e-05} +{"train_loss": 0.027236316353082657, "grad_norm": 0.12428092956542969, "global_step": 984, "epoch": 9, "lr": 9.991480048974016e-05} +{"train_loss": 0.027877245098352432, "grad_norm": 0.08610913157463074, "global_step": 985, "epoch": 9, "lr": 9.991462745726809e-05} +{"train_loss": 0.02499045804142952, "grad_norm": 0.03729109838604927, "global_step": 986, "epoch": 9, "lr": 9.991445424941772e-05} +{"train_loss": 0.022644367069005966, "grad_norm": 0.04101298376917839, "global_step": 987, "epoch": 9, "lr": 9.991428086618969e-05} +{"train_loss": 0.023139873519539833, "grad_norm": 0.09426260739564896, "global_step": 988, "epoch": 9, "lr": 9.991410730758456e-05} +{"train_loss": 0.025584949180483818, "grad_norm": 0.11338808387517929, "global_step": 989, "epoch": 9, "lr": 9.991393357360296e-05} +{"train_loss": 0.021949030458927155, "grad_norm": 0.132223978638649, "global_step": 990, "epoch": 9, "lr": 9.991375966424553e-05} +{"train_loss": 0.024078354239463806, "grad_norm": 0.0558367557823658, "global_step": 991, "epoch": 9, "lr": 9.991358557951285e-05} +{"train_loss": 0.022718917578458786, "grad_norm": 0.04944003000855446, "global_step": 992, "epoch": 9, "lr": 9.991341131940553e-05} +{"train_loss": 0.022147398442029953, "grad_norm": 0.05710693448781967, "global_step": 993, "epoch": 9, "lr": 9.991323688392418e-05} +{"train_loss": 0.023473341017961502, "grad_norm": 0.06563547998666763, "global_step": 994, "epoch": 9, "lr": 9.991306227306944e-05} +{"train_loss": 0.02276807278394699, "grad_norm": 0.03671284765005112, "global_step": 995, "epoch": 9, "lr": 9.991288748684188e-05} +{"train_loss": 0.023455586284399033, "grad_norm": 0.03178045153617859, "global_step": 996, "epoch": 9, "lr": 9.991271252524217e-05} +{"train_loss": 0.022890981286764145, "grad_norm": 0.050480395555496216, "global_step": 997, "epoch": 9, "lr": 9.991253738827089e-05} +{"train_loss": 0.022247135639190674, "grad_norm": 0.04263188689947128, "global_step": 998, "epoch": 9, "lr": 9.991236207592865e-05} +{"train_loss": 0.022630589082837105, "grad_norm": 0.05663970485329628, "global_step": 999, "epoch": 9, "lr": 9.991218658821608e-05} +{"train_loss": 0.022996462881565094, "grad_norm": 0.0478815883398056, "global_step": 1000, "epoch": 9, "lr": 9.991201092513378e-05} +{"train_loss": 0.023216266185045242, "grad_norm": 0.04774592071771622, "global_step": 1001, "epoch": 9, "lr": 9.991183508668241e-05} +{"train_loss": 0.02428809553384781, "grad_norm": 0.08478625118732452, "global_step": 1002, "epoch": 9, "lr": 9.991165907286253e-05} +{"train_loss": 0.022063642740249634, "grad_norm": 0.053086962550878525, "global_step": 1003, "epoch": 9, "lr": 9.991148288367479e-05} +{"train_loss": 0.02388087287545204, "grad_norm": 0.04089364409446716, "global_step": 1004, "epoch": 9, "lr": 9.991130651911982e-05} +{"train_loss": 0.024265389889478683, "grad_norm": 0.05005345121026039, "global_step": 1005, "epoch": 9, "lr": 9.991112997919821e-05} +{"train_loss": 0.023977044969797134, "grad_norm": 0.03516671434044838, "global_step": 1006, "epoch": 9, "lr": 9.99109532639106e-05} +{"train_loss": 0.02614704892039299, "grad_norm": 0.04791339486837387, "global_step": 1007, "epoch": 9, "lr": 9.991077637325759e-05} +{"train_loss": 0.02415839210152626, "grad_norm": 0.07242875546216965, "global_step": 1008, "epoch": 9, "lr": 9.991059930723985e-05} +{"train_loss": 0.02390347793698311, "grad_norm": 0.03385118395090103, "global_step": 1009, "epoch": 9, "lr": 9.991042206585794e-05} +{"train_loss": 0.02552523836493492, "grad_norm": 0.11169091612100601, "global_step": 1010, "epoch": 9, "lr": 9.991024464911253e-05} +{"train_loss": 0.024515733122825623, "grad_norm": 0.05712088942527771, "global_step": 1011, "epoch": 9, "lr": 9.991006705700422e-05} +{"train_loss": 0.02139052003622055, "grad_norm": 0.0433456227183342, "global_step": 1012, "epoch": 9, "lr": 9.990988928953364e-05} +{"train_loss": 0.02719789929687977, "grad_norm": 0.11031026393175125, "global_step": 1013, "epoch": 9, "lr": 9.990971134670141e-05} +{"train_loss": 0.0222521610558033, "grad_norm": 0.04578395560383797, "global_step": 1014, "epoch": 9, "lr": 9.990953322850816e-05} +{"train_loss": 0.023584505543112755, "grad_norm": 0.12388880550861359, "global_step": 1015, "epoch": 9, "lr": 9.990935493495452e-05} +{"train_loss": 0.02581060491502285, "grad_norm": 0.031173357740044594, "global_step": 1016, "epoch": 9, "lr": 9.990917646604112e-05} +{"train_loss": 0.02451026812195778, "grad_norm": 0.09462785720825195, "global_step": 1017, "epoch": 9, "lr": 9.990899782176856e-05} +{"train_loss": 0.02518445812165737, "grad_norm": 0.04627377539873123, "global_step": 1018, "epoch": 9, "lr": 9.99088190021375e-05} +{"train_loss": 0.022960934787988663, "grad_norm": 0.07833505421876907, "global_step": 1019, "epoch": 9, "lr": 9.990864000714855e-05} +{"train_loss": 0.019842669367790222, "grad_norm": 0.07798423618078232, "global_step": 1020, "epoch": 9, "lr": 9.990846083680235e-05} +{"train_loss": 0.027663372457027435, "grad_norm": 0.04373708739876747, "global_step": 1021, "epoch": 9, "lr": 9.990828149109952e-05} +{"train_loss": 0.020218413323163986, "grad_norm": 0.04564322531223297, "global_step": 1022, "epoch": 9, "lr": 9.99081019700407e-05} +{"train_loss": 0.021991468966007233, "grad_norm": 0.10931508988142014, "global_step": 1023, "epoch": 9, "lr": 9.990792227362649e-05} +{"train_loss": 0.025304652750492096, "grad_norm": 0.12254223227500916, "global_step": 1024, "epoch": 9, "lr": 9.990774240185756e-05} +{"train_loss": 0.022289182990789413, "grad_norm": 0.05709828436374664, "global_step": 1025, "epoch": 9, "lr": 9.990756235473453e-05} +{"train_loss": 0.02537040039896965, "grad_norm": 0.07401987165212631, "global_step": 1026, "epoch": 9, "lr": 9.990738213225804e-05} +{"train_loss": 0.024870146065950394, "grad_norm": 0.03711295127868652, "global_step": 1027, "epoch": 9, "lr": 9.990720173442869e-05} +{"train_loss": 0.027683427557349205, "grad_norm": 0.06829559803009033, "global_step": 1028, "epoch": 9, "lr": 9.990702116124714e-05} +{"train_loss": 0.022052276879549026, "grad_norm": 0.09186795353889465, "global_step": 1029, "epoch": 9, "lr": 9.990684041271405e-05} +{"train_loss": 0.0252075158059597, "grad_norm": 0.06985417008399963, "global_step": 1030, "epoch": 9, "lr": 9.990665948883e-05} +{"train_loss": 0.023324640467762947, "grad_norm": 0.036856759339571, "global_step": 1031, "epoch": 9, "lr": 9.990647838959565e-05} +{"train_loss": 0.023813139647245407, "grad_norm": 0.06352673470973969, "global_step": 1032, "epoch": 9, "lr": 9.990629711501166e-05} +{"train_loss": 0.023870617151260376, "grad_norm": 0.10372831672430038, "global_step": 1033, "epoch": 9, "lr": 9.990611566507863e-05} +{"train_loss": 0.021990355104207993, "grad_norm": 0.0404994897544384, "global_step": 1034, "epoch": 9, "lr": 9.990593403979722e-05} +{"train_loss": 0.023574233055114746, "grad_norm": 0.08158314973115921, "global_step": 1035, "epoch": 9, "lr": 9.990575223916806e-05} +{"train_loss": 0.01988646760582924, "grad_norm": 0.07863814383745193, "global_step": 1036, "epoch": 9, "lr": 9.990557026319178e-05} +{"train_loss": 0.027944259345531464, "grad_norm": 0.13018709421157837, "global_step": 1037, "epoch": 9, "lr": 9.990538811186905e-05} +{"train_loss": 0.023539453744888306, "grad_norm": 0.03514864295721054, "global_step": 1038, "epoch": 9, "lr": 9.990520578520047e-05} +{"train_loss": 0.020357750356197357, "grad_norm": 0.08014793694019318, "global_step": 1039, "epoch": 9, "lr": 9.990502328318671e-05} +{"train_loss": 0.021066922694444656, "grad_norm": 0.03841752931475639, "global_step": 1040, "epoch": 9, "lr": 9.990484060582839e-05} +{"train_loss": 0.01984519511461258, "grad_norm": 0.03265665844082832, "global_step": 1041, "epoch": 9, "lr": 9.990465775312619e-05} +{"train_loss": 0.021674830466508865, "grad_norm": 0.05540714040398598, "global_step": 1042, "epoch": 9, "lr": 9.99044747250807e-05} +{"train_loss": 0.021473070606589317, "grad_norm": 0.05405887961387634, "global_step": 1043, "epoch": 9, "lr": 9.99042915216926e-05} +{"train_loss": 0.023697681725025177, "grad_norm": 0.06506624072790146, "global_step": 1044, "epoch": 9, "lr": 9.990410814296252e-05} +{"train_loss": 0.025916006416082382, "grad_norm": 0.10617240518331528, "global_step": 1045, "epoch": 9, "lr": 9.990392458889111e-05} +{"train_loss": 0.02545030228793621, "grad_norm": 0.046853262931108475, "global_step": 1046, "epoch": 9, "lr": 9.9903740859479e-05} +{"train_loss": 0.02504263073205948, "grad_norm": 0.04997033625841141, "global_step": 1047, "epoch": 9, "lr": 9.990355695472685e-05} +{"train_loss": 0.027197031304240227, "grad_norm": 0.12853235006332397, "global_step": 1048, "epoch": 9, "lr": 9.990337287463531e-05} +{"train_loss": 0.021433765068650246, "grad_norm": 0.06760647147893906, "global_step": 1049, "epoch": 9, "lr": 9.990318861920502e-05} +{"train_loss": 0.022008255124092102, "grad_norm": 0.05012283846735954, "global_step": 1050, "epoch": 9, "lr": 9.990300418843662e-05} +{"train_loss": 0.02551092579960823, "grad_norm": 0.04874700680375099, "global_step": 1051, "epoch": 9, "lr": 9.990281958233077e-05} +{"train_loss": 0.021187424659729004, "grad_norm": 0.05152588710188866, "global_step": 1052, "epoch": 9, "lr": 9.990263480088812e-05} +{"train_loss": 0.023612383753061295, "grad_norm": 0.031555477529764175, "global_step": 1053, "epoch": 9, "lr": 9.99024498441093e-05} +{"train_loss": 0.023739933967590332, "grad_norm": 0.07500790059566498, "global_step": 1054, "epoch": 9, "lr": 9.990226471199499e-05} +{"train_loss": 0.024873625487089157, "grad_norm": 0.04009094461798668, "global_step": 1055, "epoch": 9, "lr": 9.990207940454581e-05} +{"train_loss": 0.019682012498378754, "grad_norm": 0.05849150940775871, "global_step": 1056, "epoch": 9, "lr": 9.990189392176244e-05} +{"train_loss": 0.023150654509663582, "grad_norm": 0.06185435131192207, "global_step": 1057, "epoch": 9, "lr": 9.99017082636455e-05} +{"train_loss": 0.022065402939915657, "grad_norm": 0.041625380516052246, "global_step": 1058, "epoch": 9, "lr": 9.990152243019567e-05} +{"train_loss": 0.02346313447814505, "grad_norm": 0.05050399526953697, "global_step": 1059, "epoch": 9, "lr": 9.990133642141359e-05, "val_loss": 0.022195080295205116} +{"train_loss": 0.02549327351152897, "grad_norm": 0.06673415750265121, "global_step": 1060, "epoch": 10, "lr": 9.990115023729991e-05} +{"train_loss": 0.021625656634569168, "grad_norm": 0.1036132425069809, "global_step": 1061, "epoch": 10, "lr": 9.990096387785529e-05} +{"train_loss": 0.02861766517162323, "grad_norm": 0.05399066209793091, "global_step": 1062, "epoch": 10, "lr": 9.990077734308038e-05} +{"train_loss": 0.02299407497048378, "grad_norm": 0.12001270055770874, "global_step": 1063, "epoch": 10, "lr": 9.990059063297585e-05} +{"train_loss": 0.021282516419887543, "grad_norm": 0.07706784456968307, "global_step": 1064, "epoch": 10, "lr": 9.990040374754235e-05} +{"train_loss": 0.023220254108309746, "grad_norm": 0.05325851961970329, "global_step": 1065, "epoch": 10, "lr": 9.990021668678051e-05} +{"train_loss": 0.024664899334311485, "grad_norm": 0.04189462587237358, "global_step": 1066, "epoch": 10, "lr": 9.990002945069103e-05} +{"train_loss": 0.02467224933207035, "grad_norm": 0.030679013580083847, "global_step": 1067, "epoch": 10, "lr": 9.989984203927454e-05} +{"train_loss": 0.024662597104907036, "grad_norm": 0.13015154004096985, "global_step": 1068, "epoch": 10, "lr": 9.98996544525317e-05} +{"train_loss": 0.02239939570426941, "grad_norm": 0.08972137421369553, "global_step": 1069, "epoch": 10, "lr": 9.98994666904632e-05} +{"train_loss": 0.02459130436182022, "grad_norm": 0.11119549721479416, "global_step": 1070, "epoch": 10, "lr": 9.989927875306965e-05} +{"train_loss": 0.01871611922979355, "grad_norm": 0.05187330022454262, "global_step": 1071, "epoch": 10, "lr": 9.989909064035175e-05} +{"train_loss": 0.024741673842072487, "grad_norm": 0.12232550978660583, "global_step": 1072, "epoch": 10, "lr": 9.989890235231013e-05} +{"train_loss": 0.024442313238978386, "grad_norm": 0.04122009128332138, "global_step": 1073, "epoch": 10, "lr": 9.989871388894547e-05} +{"train_loss": 0.021389376372098923, "grad_norm": 0.035993654280900955, "global_step": 1074, "epoch": 10, "lr": 9.989852525025845e-05} +{"train_loss": 0.024529188871383667, "grad_norm": 0.1009051501750946, "global_step": 1075, "epoch": 10, "lr": 9.989833643624968e-05} +{"train_loss": 0.024445775896310806, "grad_norm": 0.08748748898506165, "global_step": 1076, "epoch": 10, "lr": 9.989814744691988e-05} +{"train_loss": 0.024785539135336876, "grad_norm": 0.10567335784435272, "global_step": 1077, "epoch": 10, "lr": 9.98979582822697e-05} +{"train_loss": 0.025324678048491478, "grad_norm": 0.08019020408391953, "global_step": 1078, "epoch": 10, "lr": 9.989776894229976e-05} +{"train_loss": 0.02217652089893818, "grad_norm": 0.04348773509263992, "global_step": 1079, "epoch": 10, "lr": 9.989757942701078e-05} +{"train_loss": 0.0207010917365551, "grad_norm": 0.0626877173781395, "global_step": 1080, "epoch": 10, "lr": 9.989738973640341e-05} +{"train_loss": 0.02487601526081562, "grad_norm": 0.05472157523036003, "global_step": 1081, "epoch": 10, "lr": 9.989719987047829e-05} +{"train_loss": 0.026396162807941437, "grad_norm": 0.10279646515846252, "global_step": 1082, "epoch": 10, "lr": 9.989700982923613e-05} +{"train_loss": 0.021983783692121506, "grad_norm": 0.03363296017050743, "global_step": 1083, "epoch": 10, "lr": 9.989681961267758e-05} +{"train_loss": 0.023449048399925232, "grad_norm": 0.035253413021564484, "global_step": 1084, "epoch": 10, "lr": 9.989662922080329e-05} +{"train_loss": 0.020492348819971085, "grad_norm": 0.04912324994802475, "global_step": 1085, "epoch": 10, "lr": 9.989643865361395e-05} +{"train_loss": 0.023507511243224144, "grad_norm": 0.037203747779130936, "global_step": 1086, "epoch": 10, "lr": 9.989624791111024e-05} +{"train_loss": 0.02259061112999916, "grad_norm": 0.1321083903312683, "global_step": 1087, "epoch": 10, "lr": 9.989605699329281e-05} +{"train_loss": 0.022570591419935226, "grad_norm": 0.05746682360768318, "global_step": 1088, "epoch": 10, "lr": 9.989586590016231e-05} +{"train_loss": 0.021563325077295303, "grad_norm": 0.09203042089939117, "global_step": 1089, "epoch": 10, "lr": 9.989567463171946e-05} +{"train_loss": 0.022056082263588905, "grad_norm": 0.06464357674121857, "global_step": 1090, "epoch": 10, "lr": 9.98954831879649e-05} +{"train_loss": 0.02198072336614132, "grad_norm": 0.07403240352869034, "global_step": 1091, "epoch": 10, "lr": 9.989529156889932e-05} +{"train_loss": 0.025022272020578384, "grad_norm": 0.04427967220544815, "global_step": 1092, "epoch": 10, "lr": 9.989509977452337e-05} +{"train_loss": 0.024007871747016907, "grad_norm": 0.09681414067745209, "global_step": 1093, "epoch": 10, "lr": 9.989490780483775e-05} +{"train_loss": 0.022787615656852722, "grad_norm": 0.03692108765244484, "global_step": 1094, "epoch": 10, "lr": 9.98947156598431e-05} +{"train_loss": 0.02406209520995617, "grad_norm": 0.03810061514377594, "global_step": 1095, "epoch": 10, "lr": 9.989452333954015e-05} +{"train_loss": 0.022382188588380814, "grad_norm": 0.07233069837093353, "global_step": 1096, "epoch": 10, "lr": 9.989433084392951e-05} +{"train_loss": 0.023096485063433647, "grad_norm": 0.03984655812382698, "global_step": 1097, "epoch": 10, "lr": 9.989413817301192e-05} +{"train_loss": 0.02518138289451599, "grad_norm": 0.05689432844519615, "global_step": 1098, "epoch": 10, "lr": 9.989394532678801e-05} +{"train_loss": 0.022496655583381653, "grad_norm": 0.05761587247252464, "global_step": 1099, "epoch": 10, "lr": 9.989375230525848e-05} +{"train_loss": 0.02446383610367775, "grad_norm": 0.049403440207242966, "global_step": 1100, "epoch": 10, "lr": 9.9893559108424e-05} +{"train_loss": 0.01888572797179222, "grad_norm": 0.1971195489168167, "global_step": 1101, "epoch": 10, "lr": 9.989336573628527e-05} +{"train_loss": 0.019001033157110214, "grad_norm": 0.06497874855995178, "global_step": 1102, "epoch": 10, "lr": 9.989317218884292e-05} +{"train_loss": 0.02166082337498665, "grad_norm": 0.04167954996228218, "global_step": 1103, "epoch": 10, "lr": 9.989297846609768e-05} +{"train_loss": 0.02317708358168602, "grad_norm": 0.11817097663879395, "global_step": 1104, "epoch": 10, "lr": 9.98927845680502e-05} +{"train_loss": 0.025207489728927612, "grad_norm": 0.10749762505292892, "global_step": 1105, "epoch": 10, "lr": 9.989259049470119e-05} +{"train_loss": 0.02419496513903141, "grad_norm": 0.050992075353860855, "global_step": 1106, "epoch": 10, "lr": 9.98923962460513e-05} +{"train_loss": 0.02079937793314457, "grad_norm": 0.11400419473648071, "global_step": 1107, "epoch": 10, "lr": 9.989220182210123e-05} +{"train_loss": 0.022470273077487946, "grad_norm": 0.12678778171539307, "global_step": 1108, "epoch": 10, "lr": 9.989200722285166e-05} +{"train_loss": 0.021130356937646866, "grad_norm": 0.03221043571829796, "global_step": 1109, "epoch": 10, "lr": 9.989181244830328e-05} +{"train_loss": 0.021658584475517273, "grad_norm": 0.11185697466135025, "global_step": 1110, "epoch": 10, "lr": 9.989161749845677e-05} +{"train_loss": 0.02630792371928692, "grad_norm": 0.03711799904704094, "global_step": 1111, "epoch": 10, "lr": 9.98914223733128e-05} +{"train_loss": 0.02408767119050026, "grad_norm": 0.10414819419384003, "global_step": 1112, "epoch": 10, "lr": 9.989122707287208e-05} +{"train_loss": 0.025407735258340836, "grad_norm": 0.15097680687904358, "global_step": 1113, "epoch": 10, "lr": 9.989103159713529e-05} +{"train_loss": 0.024745289236307144, "grad_norm": 0.06191626563668251, "global_step": 1114, "epoch": 10, "lr": 9.989083594610309e-05} +{"train_loss": 0.024122154340147972, "grad_norm": 0.04948154091835022, "global_step": 1115, "epoch": 10, "lr": 9.98906401197762e-05} +{"train_loss": 0.02242102101445198, "grad_norm": 0.12416154146194458, "global_step": 1116, "epoch": 10, "lr": 9.989044411815532e-05} +{"train_loss": 0.022876236587762833, "grad_norm": 0.08647066354751587, "global_step": 1117, "epoch": 10, "lr": 9.989024794124109e-05} +{"train_loss": 0.024810176342725754, "grad_norm": 0.1487596333026886, "global_step": 1118, "epoch": 10, "lr": 9.989005158903423e-05} +{"train_loss": 0.023258056491613388, "grad_norm": 0.09625573456287384, "global_step": 1119, "epoch": 10, "lr": 9.988985506153543e-05} +{"train_loss": 0.02304108440876007, "grad_norm": 0.09646840393543243, "global_step": 1120, "epoch": 10, "lr": 9.988965835874537e-05} +{"train_loss": 0.02184459939599037, "grad_norm": 0.050067562609910965, "global_step": 1121, "epoch": 10, "lr": 9.988946148066475e-05} +{"train_loss": 0.02530517801642418, "grad_norm": 0.19500888884067535, "global_step": 1122, "epoch": 10, "lr": 9.988926442729426e-05} +{"train_loss": 0.022104797884821892, "grad_norm": 0.036813054233789444, "global_step": 1123, "epoch": 10, "lr": 9.988906719863459e-05} +{"train_loss": 0.02587497979402542, "grad_norm": 0.09845051914453506, "global_step": 1124, "epoch": 10, "lr": 9.988886979468643e-05} +{"train_loss": 0.022622086107730865, "grad_norm": 0.12597519159317017, "global_step": 1125, "epoch": 10, "lr": 9.988867221545049e-05} +{"train_loss": 0.02314961701631546, "grad_norm": 0.0526847206056118, "global_step": 1126, "epoch": 10, "lr": 9.988847446092745e-05} +{"train_loss": 0.02093445509672165, "grad_norm": 0.04483245685696602, "global_step": 1127, "epoch": 10, "lr": 9.988827653111798e-05} +{"train_loss": 0.02681531384587288, "grad_norm": 0.04753503203392029, "global_step": 1128, "epoch": 10, "lr": 9.988807842602284e-05} +{"train_loss": 0.021824132651090622, "grad_norm": 0.04850444942712784, "global_step": 1129, "epoch": 10, "lr": 9.988788014564266e-05} +{"train_loss": 0.024866238236427307, "grad_norm": 0.08225340396165848, "global_step": 1130, "epoch": 10, "lr": 9.988768168997818e-05} +{"train_loss": 0.023972567170858383, "grad_norm": 0.07998906821012497, "global_step": 1131, "epoch": 10, "lr": 9.988748305903007e-05} +{"train_loss": 0.02059859037399292, "grad_norm": 0.03872138634324074, "global_step": 1132, "epoch": 10, "lr": 9.988728425279905e-05} +{"train_loss": 0.022420862689614296, "grad_norm": 0.09809338301420212, "global_step": 1133, "epoch": 10, "lr": 9.98870852712858e-05} +{"train_loss": 0.023152276873588562, "grad_norm": 0.057722385972738266, "global_step": 1134, "epoch": 10, "lr": 9.988688611449104e-05} +{"train_loss": 0.027112776413559914, "grad_norm": 0.05194025859236717, "global_step": 1135, "epoch": 10, "lr": 9.988668678241546e-05} +{"train_loss": 0.020542342215776443, "grad_norm": 0.03594754636287689, "global_step": 1136, "epoch": 10, "lr": 9.988648727505974e-05} +{"train_loss": 0.022164562717080116, "grad_norm": 0.08722314238548279, "global_step": 1137, "epoch": 10, "lr": 9.988628759242462e-05} +{"train_loss": 0.020644035190343857, "grad_norm": 0.12778982520103455, "global_step": 1138, "epoch": 10, "lr": 9.988608773451076e-05} +{"train_loss": 0.025757474824786186, "grad_norm": 0.14730188250541687, "global_step": 1139, "epoch": 10, "lr": 9.988588770131889e-05} +{"train_loss": 0.021766871213912964, "grad_norm": 0.06348398327827454, "global_step": 1140, "epoch": 10, "lr": 9.988568749284971e-05} +{"train_loss": 0.021269183605909348, "grad_norm": 0.04184272885322571, "global_step": 1141, "epoch": 10, "lr": 9.988548710910392e-05} +{"train_loss": 0.023292459547519684, "grad_norm": 0.07776626944541931, "global_step": 1142, "epoch": 10, "lr": 9.988528655008222e-05} +{"train_loss": 0.023066667839884758, "grad_norm": 0.05485086515545845, "global_step": 1143, "epoch": 10, "lr": 9.98850858157853e-05} +{"train_loss": 0.02404525689780712, "grad_norm": 0.08346197009086609, "global_step": 1144, "epoch": 10, "lr": 9.98848849062139e-05} +{"train_loss": 0.02521519362926483, "grad_norm": 0.0917288213968277, "global_step": 1145, "epoch": 10, "lr": 9.98846838213687e-05} +{"train_loss": 0.022658031433820724, "grad_norm": 0.0734482929110527, "global_step": 1146, "epoch": 10, "lr": 9.988448256125043e-05} +{"train_loss": 0.01982872560620308, "grad_norm": 0.03707370162010193, "global_step": 1147, "epoch": 10, "lr": 9.988428112585978e-05} +{"train_loss": 0.02408737875521183, "grad_norm": 0.078373022377491, "global_step": 1148, "epoch": 10, "lr": 9.988407951519745e-05} +{"train_loss": 0.023064076900482178, "grad_norm": 0.08801670372486115, "global_step": 1149, "epoch": 10, "lr": 9.988387772926418e-05} +{"train_loss": 0.02147279866039753, "grad_norm": 0.03824817016720772, "global_step": 1150, "epoch": 10, "lr": 9.988367576806064e-05} +{"train_loss": 0.022776644676923752, "grad_norm": 0.07264948636293411, "global_step": 1151, "epoch": 10, "lr": 9.988347363158756e-05} +{"train_loss": 0.024350248277187347, "grad_norm": 0.04462587088346481, "global_step": 1152, "epoch": 10, "lr": 9.988327131984564e-05} +{"train_loss": 0.024510618299245834, "grad_norm": 0.05578738823533058, "global_step": 1153, "epoch": 10, "lr": 9.98830688328356e-05} +{"train_loss": 0.02274264767765999, "grad_norm": 0.09747804701328278, "global_step": 1154, "epoch": 10, "lr": 9.988286617055816e-05} +{"train_loss": 0.0224660262465477, "grad_norm": 0.06082437187433243, "global_step": 1155, "epoch": 10, "lr": 9.988266333301401e-05} +{"train_loss": 0.025021396577358246, "grad_norm": 0.12037302553653717, "global_step": 1156, "epoch": 10, "lr": 9.988246032020389e-05} +{"train_loss": 0.026381192728877068, "grad_norm": 0.1882259100675583, "global_step": 1157, "epoch": 10, "lr": 9.988225713212849e-05} +{"train_loss": 0.020267372950911522, "grad_norm": 0.08429181575775146, "global_step": 1158, "epoch": 10, "lr": 9.988205376878852e-05} +{"train_loss": 0.02329612523317337, "grad_norm": 0.038064707070589066, "global_step": 1159, "epoch": 10, "lr": 9.988185023018472e-05} +{"train_loss": 0.02229027822613716, "grad_norm": 0.06285571306943893, "global_step": 1160, "epoch": 10, "lr": 9.988164651631778e-05} +{"train_loss": 0.02486603707075119, "grad_norm": 0.05871783569455147, "global_step": 1161, "epoch": 10, "lr": 9.988144262718844e-05} +{"train_loss": 0.026846986263990402, "grad_norm": 0.05864344537258148, "global_step": 1162, "epoch": 10, "lr": 9.98812385627974e-05} +{"train_loss": 0.021848399192094803, "grad_norm": 0.055460210889577866, "global_step": 1163, "epoch": 10, "lr": 9.988103432314537e-05} +{"train_loss": 0.022944804280996323, "grad_norm": 0.04716232791543007, "global_step": 1164, "epoch": 10, "lr": 9.988082990823309e-05} +{"train_loss": 0.02325687984461492, "grad_norm": 0.042031578719615936, "global_step": 1165, "epoch": 10, "lr": 9.988062531806126e-05, "train_batch_mse_error": 0.02171059139072895, "val_loss": 0.02159876748919487} +{"train_loss": 0.023128025233745575, "grad_norm": 0.1221662312746048, "global_step": 1166, "epoch": 11, "lr": 9.988042055263062e-05} +{"train_loss": 0.022381220012903214, "grad_norm": 0.07850322127342224, "global_step": 1167, "epoch": 11, "lr": 9.988021561194185e-05} +{"train_loss": 0.02145465277135372, "grad_norm": 0.12691126763820648, "global_step": 1168, "epoch": 11, "lr": 9.988001049599571e-05} +{"train_loss": 0.026261821389198303, "grad_norm": 0.038711484521627426, "global_step": 1169, "epoch": 11, "lr": 9.98798052047929e-05} +{"train_loss": 0.023388251662254333, "grad_norm": 0.059725094586610794, "global_step": 1170, "epoch": 11, "lr": 9.987959973833416e-05} +{"train_loss": 0.021805591881275177, "grad_norm": 0.09202609211206436, "global_step": 1171, "epoch": 11, "lr": 9.987939409662019e-05} +{"train_loss": 0.024800019338726997, "grad_norm": 0.06670252978801727, "global_step": 1172, "epoch": 11, "lr": 9.98791882796517e-05} +{"train_loss": 0.023157507181167603, "grad_norm": 0.05054493620991707, "global_step": 1173, "epoch": 11, "lr": 9.987898228742946e-05} +{"train_loss": 0.021727176383137703, "grad_norm": 0.09253628551959991, "global_step": 1174, "epoch": 11, "lr": 9.987877611995416e-05} +{"train_loss": 0.025416254997253418, "grad_norm": 0.0624011792242527, "global_step": 1175, "epoch": 11, "lr": 9.987856977722653e-05} +{"train_loss": 0.021170472726225853, "grad_norm": 0.07944269478321075, "global_step": 1176, "epoch": 11, "lr": 9.98783632592473e-05} +{"train_loss": 0.022340361028909683, "grad_norm": 0.031344443559646606, "global_step": 1177, "epoch": 11, "lr": 9.98781565660172e-05} +{"train_loss": 0.02157479152083397, "grad_norm": 0.0774196907877922, "global_step": 1178, "epoch": 11, "lr": 9.987794969753694e-05} +{"train_loss": 0.020582035183906555, "grad_norm": 0.05949476361274719, "global_step": 1179, "epoch": 11, "lr": 9.987774265380727e-05} +{"train_loss": 0.023479342460632324, "grad_norm": 0.08621011674404144, "global_step": 1180, "epoch": 11, "lr": 9.98775354348289e-05} +{"train_loss": 0.02664307877421379, "grad_norm": 0.03728862851858139, "global_step": 1181, "epoch": 11, "lr": 9.987732804060254e-05} +{"train_loss": 0.018001336604356766, "grad_norm": 0.059589531272649765, "global_step": 1182, "epoch": 11, "lr": 9.987712047112896e-05} +{"train_loss": 0.02737073227763176, "grad_norm": 0.12230002135038376, "global_step": 1183, "epoch": 11, "lr": 9.987691272640886e-05} +{"train_loss": 0.021177949383854866, "grad_norm": 0.048481978476047516, "global_step": 1184, "epoch": 11, "lr": 9.987670480644299e-05} +{"train_loss": 0.021409109234809875, "grad_norm": 0.0378628708422184, "global_step": 1185, "epoch": 11, "lr": 9.987649671123207e-05} +{"train_loss": 0.023158209398388863, "grad_norm": 0.06744761019945145, "global_step": 1186, "epoch": 11, "lr": 9.987628844077681e-05} +{"train_loss": 0.023110492154955864, "grad_norm": 0.0645788162946701, "global_step": 1187, "epoch": 11, "lr": 9.987607999507799e-05} +{"train_loss": 0.024862460792064667, "grad_norm": 0.04631254822015762, "global_step": 1188, "epoch": 11, "lr": 9.98758713741363e-05} +{"train_loss": 0.023683816194534302, "grad_norm": 0.052023664116859436, "global_step": 1189, "epoch": 11, "lr": 9.98756625779525e-05} +{"train_loss": 0.025930488482117653, "grad_norm": 0.07754947990179062, "global_step": 1190, "epoch": 11, "lr": 9.98754536065273e-05} +{"train_loss": 0.02442665956914425, "grad_norm": 0.11517550051212311, "global_step": 1191, "epoch": 11, "lr": 9.987524445986145e-05} +{"train_loss": 0.022921022027730942, "grad_norm": 0.12222692370414734, "global_step": 1192, "epoch": 11, "lr": 9.987503513795568e-05} +{"train_loss": 0.019618988037109375, "grad_norm": 0.05282917246222496, "global_step": 1193, "epoch": 11, "lr": 9.987482564081074e-05} +{"train_loss": 0.024691112339496613, "grad_norm": 0.08237981051206589, "global_step": 1194, "epoch": 11, "lr": 9.987461596842733e-05} +{"train_loss": 0.02278604917228222, "grad_norm": 0.06563480198383331, "global_step": 1195, "epoch": 11, "lr": 9.987440612080623e-05} +{"train_loss": 0.023931309580802917, "grad_norm": 0.06700441241264343, "global_step": 1196, "epoch": 11, "lr": 9.987419609794814e-05} +{"train_loss": 0.021084679290652275, "grad_norm": 0.03629838302731514, "global_step": 1197, "epoch": 11, "lr": 9.987398589985382e-05} +{"train_loss": 0.02137213945388794, "grad_norm": 0.07555458694696426, "global_step": 1198, "epoch": 11, "lr": 9.9873775526524e-05} +{"train_loss": 0.023487407714128494, "grad_norm": 0.11754447966814041, "global_step": 1199, "epoch": 11, "lr": 9.987356497795943e-05} +{"train_loss": 0.02213282696902752, "grad_norm": 0.04030003771185875, "global_step": 1200, "epoch": 11, "lr": 9.987335425416085e-05} +{"train_loss": 0.021956302225589752, "grad_norm": 0.047168388962745667, "global_step": 1201, "epoch": 11, "lr": 9.987314335512898e-05} +{"train_loss": 0.02046963945031166, "grad_norm": 0.04862244799733162, "global_step": 1202, "epoch": 11, "lr": 9.987293228086458e-05} +{"train_loss": 0.023496141657233238, "grad_norm": 0.05551471188664436, "global_step": 1203, "epoch": 11, "lr": 9.987272103136838e-05} +{"train_loss": 0.021686263382434845, "grad_norm": 0.05303441360592842, "global_step": 1204, "epoch": 11, "lr": 9.987250960664111e-05} +{"train_loss": 0.02043602056801319, "grad_norm": 0.06584739685058594, "global_step": 1205, "epoch": 11, "lr": 9.987229800668356e-05} +{"train_loss": 0.020538216456770897, "grad_norm": 0.0741300955414772, "global_step": 1206, "epoch": 11, "lr": 9.987208623149643e-05} +{"train_loss": 0.021665450185537338, "grad_norm": 0.03952623903751373, "global_step": 1207, "epoch": 11, "lr": 9.987187428108048e-05} +{"train_loss": 0.024879779666662216, "grad_norm": 0.03520560264587402, "global_step": 1208, "epoch": 11, "lr": 9.987166215543645e-05} +{"train_loss": 0.02200787141919136, "grad_norm": 0.10129602253437042, "global_step": 1209, "epoch": 11, "lr": 9.987144985456509e-05} +{"train_loss": 0.024370649829506874, "grad_norm": 0.05442315340042114, "global_step": 1210, "epoch": 11, "lr": 9.987123737846715e-05} +{"train_loss": 0.019266024231910706, "grad_norm": 0.08124646544456482, "global_step": 1211, "epoch": 11, "lr": 9.987102472714335e-05} +{"train_loss": 0.02391575463116169, "grad_norm": 0.06690088659524918, "global_step": 1212, "epoch": 11, "lr": 9.987081190059449e-05} +{"train_loss": 0.01816021278500557, "grad_norm": 0.09128185361623764, "global_step": 1213, "epoch": 11, "lr": 9.987059889882126e-05} +{"train_loss": 0.0250430665910244, "grad_norm": 0.04595933482050896, "global_step": 1214, "epoch": 11, "lr": 9.987038572182444e-05} +{"train_loss": 0.026612738147377968, "grad_norm": 0.07727094739675522, "global_step": 1215, "epoch": 11, "lr": 9.987017236960477e-05} +{"train_loss": 0.02133587747812271, "grad_norm": 0.05108967795968056, "global_step": 1216, "epoch": 11, "lr": 9.986995884216302e-05} +{"train_loss": 0.022013705223798752, "grad_norm": 0.040943168103694916, "global_step": 1217, "epoch": 11, "lr": 9.98697451394999e-05} +{"train_loss": 0.0253495741635561, "grad_norm": 0.045668281614780426, "global_step": 1218, "epoch": 11, "lr": 9.986953126161619e-05} +{"train_loss": 0.022142931818962097, "grad_norm": 0.041169483214616776, "global_step": 1219, "epoch": 11, "lr": 9.986931720851262e-05} +{"train_loss": 0.026146501302719116, "grad_norm": 0.03713987022638321, "global_step": 1220, "epoch": 11, "lr": 9.986910298018998e-05} +{"train_loss": 0.021609710529446602, "grad_norm": 0.08503686636686325, "global_step": 1221, "epoch": 11, "lr": 9.9868888576649e-05} +{"train_loss": 0.02788412570953369, "grad_norm": 0.06884793937206268, "global_step": 1222, "epoch": 11, "lr": 9.986867399789042e-05} +{"train_loss": 0.025772763416171074, "grad_norm": 0.10424229502677917, "global_step": 1223, "epoch": 11, "lr": 9.986845924391502e-05} +{"train_loss": 0.019501639530062675, "grad_norm": 0.04988330230116844, "global_step": 1224, "epoch": 11, "lr": 9.986824431472352e-05} +{"train_loss": 0.02320779860019684, "grad_norm": 0.06764303147792816, "global_step": 1225, "epoch": 11, "lr": 9.986802921031673e-05} +{"train_loss": 0.0186476930975914, "grad_norm": 0.04289755970239639, "global_step": 1226, "epoch": 11, "lr": 9.986781393069534e-05} +{"train_loss": 0.024433964863419533, "grad_norm": 0.03686055913567543, "global_step": 1227, "epoch": 11, "lr": 9.986759847586016e-05} +{"train_loss": 0.02084030956029892, "grad_norm": 0.06102593243122101, "global_step": 1228, "epoch": 11, "lr": 9.986738284581192e-05} +{"train_loss": 0.02083166129887104, "grad_norm": 0.07339351624250412, "global_step": 1229, "epoch": 11, "lr": 9.986716704055136e-05} +{"train_loss": 0.022325243800878525, "grad_norm": 0.048895880579948425, "global_step": 1230, "epoch": 11, "lr": 9.986695106007929e-05} +{"train_loss": 0.023506373167037964, "grad_norm": 0.03715876117348671, "global_step": 1231, "epoch": 11, "lr": 9.986673490439645e-05} +{"train_loss": 0.02369699254631996, "grad_norm": 0.0465729646384716, "global_step": 1232, "epoch": 11, "lr": 9.986651857350356e-05} +{"train_loss": 0.022979527711868286, "grad_norm": 0.0491768978536129, "global_step": 1233, "epoch": 11, "lr": 9.986630206740142e-05} +{"train_loss": 0.022556515410542488, "grad_norm": 0.11697512120008469, "global_step": 1234, "epoch": 11, "lr": 9.986608538609079e-05} +{"train_loss": 0.023660633713006973, "grad_norm": 0.04817413166165352, "global_step": 1235, "epoch": 11, "lr": 9.986586852957242e-05} +{"train_loss": 0.023307006806135178, "grad_norm": 0.10433605313301086, "global_step": 1236, "epoch": 11, "lr": 9.986565149784706e-05} +{"train_loss": 0.017630649730563164, "grad_norm": 0.034938469529151917, "global_step": 1237, "epoch": 11, "lr": 9.98654342909155e-05} +{"train_loss": 0.02257654070854187, "grad_norm": 0.04579813405871391, "global_step": 1238, "epoch": 11, "lr": 9.986521690877848e-05} +{"train_loss": 0.02408374473452568, "grad_norm": 0.045555442571640015, "global_step": 1239, "epoch": 11, "lr": 9.986499935143679e-05} +{"train_loss": 0.020465971902012825, "grad_norm": 0.07948852330446243, "global_step": 1240, "epoch": 11, "lr": 9.986478161889117e-05} +{"train_loss": 0.02501579374074936, "grad_norm": 0.03788601979613304, "global_step": 1241, "epoch": 11, "lr": 9.986456371114239e-05} +{"train_loss": 0.026356857270002365, "grad_norm": 0.08638961613178253, "global_step": 1242, "epoch": 11, "lr": 9.986434562819122e-05} +{"train_loss": 0.020397212356328964, "grad_norm": 0.08090600371360779, "global_step": 1243, "epoch": 11, "lr": 9.986412737003843e-05} +{"train_loss": 0.023144174367189407, "grad_norm": 0.044770438224077225, "global_step": 1244, "epoch": 11, "lr": 9.986390893668476e-05} +{"train_loss": 0.02308468520641327, "grad_norm": 0.08498522639274597, "global_step": 1245, "epoch": 11, "lr": 9.986369032813103e-05} +{"train_loss": 0.022220319136977196, "grad_norm": 0.048825740814208984, "global_step": 1246, "epoch": 11, "lr": 9.986347154437796e-05} +{"train_loss": 0.025903552770614624, "grad_norm": 0.04896186664700508, "global_step": 1247, "epoch": 11, "lr": 9.986325258542633e-05} +{"train_loss": 0.02615051344037056, "grad_norm": 0.05182436853647232, "global_step": 1248, "epoch": 11, "lr": 9.986303345127693e-05} +{"train_loss": 0.02384040504693985, "grad_norm": 0.15340007841587067, "global_step": 1249, "epoch": 11, "lr": 9.986281414193051e-05} +{"train_loss": 0.026663782075047493, "grad_norm": 0.05881739407777786, "global_step": 1250, "epoch": 11, "lr": 9.986259465738785e-05} +{"train_loss": 0.020297477021813393, "grad_norm": 0.061208438128232956, "global_step": 1251, "epoch": 11, "lr": 9.986237499764971e-05} +{"train_loss": 0.021158238872885704, "grad_norm": 0.03746294230222702, "global_step": 1252, "epoch": 11, "lr": 9.986215516271686e-05} +{"train_loss": 0.02401936799287796, "grad_norm": 0.03778019919991493, "global_step": 1253, "epoch": 11, "lr": 9.98619351525901e-05} +{"train_loss": 0.020432744175195694, "grad_norm": 0.0418817475438118, "global_step": 1254, "epoch": 11, "lr": 9.986171496727019e-05} +{"train_loss": 0.026519671082496643, "grad_norm": 0.20119406282901764, "global_step": 1255, "epoch": 11, "lr": 9.986149460675788e-05} +{"train_loss": 0.01988774910569191, "grad_norm": 0.07125796377658844, "global_step": 1256, "epoch": 11, "lr": 9.986127407105396e-05} +{"train_loss": 0.024966631084680557, "grad_norm": 0.09259180724620819, "global_step": 1257, "epoch": 11, "lr": 9.986105336015921e-05} +{"train_loss": 0.025139475241303444, "grad_norm": 0.058673251420259476, "global_step": 1258, "epoch": 11, "lr": 9.986083247407441e-05} +{"train_loss": 0.02361561357975006, "grad_norm": 0.08719213306903839, "global_step": 1259, "epoch": 11, "lr": 9.986061141280034e-05} +{"train_loss": 0.020206758752465248, "grad_norm": 0.05225219950079918, "global_step": 1260, "epoch": 11, "lr": 9.986039017633775e-05} +{"train_loss": 0.022532811388373375, "grad_norm": 0.052033931016922, "global_step": 1261, "epoch": 11, "lr": 9.986016876468743e-05} +{"train_loss": 0.023749761283397675, "grad_norm": 0.04118786379694939, "global_step": 1262, "epoch": 11, "lr": 9.985994717785016e-05} +{"train_loss": 0.02745930477976799, "grad_norm": 0.05599234253168106, "global_step": 1263, "epoch": 11, "lr": 9.985972541582673e-05} +{"train_loss": 0.025054218247532845, "grad_norm": 0.07972410321235657, "global_step": 1264, "epoch": 11, "lr": 9.98595034786179e-05} +{"train_loss": 0.021756522357463837, "grad_norm": 0.05194548889994621, "global_step": 1265, "epoch": 11, "lr": 9.985928136622445e-05} +{"train_loss": 0.024568714201450348, "grad_norm": 0.09821509569883347, "global_step": 1266, "epoch": 11, "lr": 9.985905907864717e-05} +{"train_loss": 0.023926496505737305, "grad_norm": 0.07056774199008942, "global_step": 1267, "epoch": 11, "lr": 9.985883661588686e-05} +{"train_loss": 0.02314472571015358, "grad_norm": 0.11810161918401718, "global_step": 1268, "epoch": 11, "lr": 9.985861397794427e-05} +{"train_loss": 0.024605007842183113, "grad_norm": 0.11426763236522675, "global_step": 1269, "epoch": 11, "lr": 9.985839116482019e-05} +{"train_loss": 0.02629193291068077, "grad_norm": 0.09996988624334335, "global_step": 1270, "epoch": 11, "lr": 9.985816817651542e-05} +{"train_loss": 0.02298861188497746, "grad_norm": 0.07136156409978867, "global_step": 1271, "epoch": 11, "lr": 9.98579450130307e-05, "val_loss": 0.021601028740406036} +{"train_loss": 0.02225634828209877, "grad_norm": 0.06577400118112564, "global_step": 1272, "epoch": 12, "lr": 9.985772167436686e-05} +{"train_loss": 0.02309875562787056, "grad_norm": 0.05419095978140831, "global_step": 1273, "epoch": 12, "lr": 9.985749816052467e-05} +{"train_loss": 0.019537217915058136, "grad_norm": 0.09789945930242538, "global_step": 1274, "epoch": 12, "lr": 9.985727447150492e-05} +{"train_loss": 0.024531085044145584, "grad_norm": 0.11273840069770813, "global_step": 1275, "epoch": 12, "lr": 9.985705060730839e-05} +{"train_loss": 0.024960460141301155, "grad_norm": 0.07909242808818817, "global_step": 1276, "epoch": 12, "lr": 9.985682656793584e-05} +{"train_loss": 0.024511389434337616, "grad_norm": 0.0574304535984993, "global_step": 1277, "epoch": 12, "lr": 9.985660235338812e-05} +{"train_loss": 0.023741556331515312, "grad_norm": 0.05816356837749481, "global_step": 1278, "epoch": 12, "lr": 9.985637796366595e-05} +{"train_loss": 0.022192245349287987, "grad_norm": 0.05238812416791916, "global_step": 1279, "epoch": 12, "lr": 9.985615339877018e-05} +{"train_loss": 0.021506519988179207, "grad_norm": 0.12813983857631683, "global_step": 1280, "epoch": 12, "lr": 9.985592865870154e-05} +{"train_loss": 0.022815117612481117, "grad_norm": 0.1023116484284401, "global_step": 1281, "epoch": 12, "lr": 9.985570374346087e-05} +{"train_loss": 0.024330824613571167, "grad_norm": 0.05937973037362099, "global_step": 1282, "epoch": 12, "lr": 9.985547865304894e-05} +{"train_loss": 0.024943329393863678, "grad_norm": 0.03927001357078552, "global_step": 1283, "epoch": 12, "lr": 9.985525338746653e-05} +{"train_loss": 0.021533207967877388, "grad_norm": 0.09050697833299637, "global_step": 1284, "epoch": 12, "lr": 9.985502794671444e-05} +{"train_loss": 0.023463871330022812, "grad_norm": 0.06816253811120987, "global_step": 1285, "epoch": 12, "lr": 9.985480233079346e-05} +{"train_loss": 0.020170902833342552, "grad_norm": 0.10718127340078354, "global_step": 1286, "epoch": 12, "lr": 9.98545765397044e-05} +{"train_loss": 0.01938547194004059, "grad_norm": 0.061649974435567856, "global_step": 1287, "epoch": 12, "lr": 9.985435057344803e-05} +{"train_loss": 0.019801609218120575, "grad_norm": 0.03638465702533722, "global_step": 1288, "epoch": 12, "lr": 9.985412443202516e-05} +{"train_loss": 0.021800238639116287, "grad_norm": 0.09424442797899246, "global_step": 1289, "epoch": 12, "lr": 9.985389811543656e-05} +{"train_loss": 0.025164060294628143, "grad_norm": 0.04505196958780289, "global_step": 1290, "epoch": 12, "lr": 9.985367162368305e-05} +{"train_loss": 0.02117668092250824, "grad_norm": 0.05752706527709961, "global_step": 1291, "epoch": 12, "lr": 9.985344495676542e-05} +{"train_loss": 0.022335924208164215, "grad_norm": 0.0605107843875885, "global_step": 1292, "epoch": 12, "lr": 9.985321811468447e-05} +{"train_loss": 0.02135327458381653, "grad_norm": 0.03221508860588074, "global_step": 1293, "epoch": 12, "lr": 9.985299109744099e-05} +{"train_loss": 0.02344236895442009, "grad_norm": 0.08198581635951996, "global_step": 1294, "epoch": 12, "lr": 9.985276390503579e-05} +{"train_loss": 0.02126813679933548, "grad_norm": 0.055000029504299164, "global_step": 1295, "epoch": 12, "lr": 9.985253653746964e-05} +{"train_loss": 0.02169349230825901, "grad_norm": 0.0633637085556984, "global_step": 1296, "epoch": 12, "lr": 9.985230899474336e-05} +{"train_loss": 0.02097139321267605, "grad_norm": 0.057990752160549164, "global_step": 1297, "epoch": 12, "lr": 9.985208127685775e-05} +{"train_loss": 0.021017789840698242, "grad_norm": 0.028729906305670738, "global_step": 1298, "epoch": 12, "lr": 9.985185338381362e-05} +{"train_loss": 0.022066619247198105, "grad_norm": 0.07883419841527939, "global_step": 1299, "epoch": 12, "lr": 9.985162531561174e-05} +{"train_loss": 0.019737370312213898, "grad_norm": 0.03610322251915932, "global_step": 1300, "epoch": 12, "lr": 9.985139707225293e-05} +{"train_loss": 0.025014694780111313, "grad_norm": 0.04629186540842056, "global_step": 1301, "epoch": 12, "lr": 9.9851168653738e-05} +{"train_loss": 0.02117813378572464, "grad_norm": 0.039751604199409485, "global_step": 1302, "epoch": 12, "lr": 9.985094006006774e-05} +{"train_loss": 0.025122620165348053, "grad_norm": 0.062308963388204575, "global_step": 1303, "epoch": 12, "lr": 9.985071129124295e-05} +{"train_loss": 0.02105940878391266, "grad_norm": 0.05765756219625473, "global_step": 1304, "epoch": 12, "lr": 9.985048234726444e-05} +{"train_loss": 0.021220244467258453, "grad_norm": 0.07906341552734375, "global_step": 1305, "epoch": 12, "lr": 9.985025322813302e-05} +{"train_loss": 0.02394457720220089, "grad_norm": 0.09692676365375519, "global_step": 1306, "epoch": 12, "lr": 9.985002393384948e-05} +{"train_loss": 0.02182985097169876, "grad_norm": 0.05204221233725548, "global_step": 1307, "epoch": 12, "lr": 9.984979446441465e-05} +{"train_loss": 0.023004746064543724, "grad_norm": 0.07015813887119293, "global_step": 1308, "epoch": 12, "lr": 9.984956481982931e-05} +{"train_loss": 0.021526165306568146, "grad_norm": 0.06719871610403061, "global_step": 1309, "epoch": 12, "lr": 9.984933500009427e-05} +{"train_loss": 0.024237802252173424, "grad_norm": 0.08927495032548904, "global_step": 1310, "epoch": 12, "lr": 9.984910500521036e-05} +{"train_loss": 0.022955844178795815, "grad_norm": 0.055822309106588364, "global_step": 1311, "epoch": 12, "lr": 9.984887483517839e-05} +{"train_loss": 0.028040491044521332, "grad_norm": 0.051582787185907364, "global_step": 1312, "epoch": 12, "lr": 9.984864448999913e-05} +{"train_loss": 0.02306010015308857, "grad_norm": 0.09982527792453766, "global_step": 1313, "epoch": 12, "lr": 9.984841396967341e-05} +{"train_loss": 0.02088945358991623, "grad_norm": 0.04888699948787689, "global_step": 1314, "epoch": 12, "lr": 9.984818327420205e-05} +{"train_loss": 0.020420754328370094, "grad_norm": 0.042540065944194794, "global_step": 1315, "epoch": 12, "lr": 9.984795240358584e-05} +{"train_loss": 0.023834029212594032, "grad_norm": 0.04690656438469887, "global_step": 1316, "epoch": 12, "lr": 9.984772135782563e-05} +{"train_loss": 0.0200191717594862, "grad_norm": 0.0399288535118103, "global_step": 1317, "epoch": 12, "lr": 9.984749013692219e-05} +{"train_loss": 0.02409438043832779, "grad_norm": 0.059736911207437515, "global_step": 1318, "epoch": 12, "lr": 9.984725874087633e-05} +{"train_loss": 0.02745509147644043, "grad_norm": 0.11910145729780197, "global_step": 1319, "epoch": 12, "lr": 9.98470271696889e-05} +{"train_loss": 0.022983646020293236, "grad_norm": 0.0352831706404686, "global_step": 1320, "epoch": 12, "lr": 9.98467954233607e-05} +{"train_loss": 0.022071195766329765, "grad_norm": 0.06392346322536469, "global_step": 1321, "epoch": 12, "lr": 9.984656350189252e-05} +{"train_loss": 0.020398706197738647, "grad_norm": 0.06694889813661575, "global_step": 1322, "epoch": 12, "lr": 9.98463314052852e-05} +{"train_loss": 0.02123962715268135, "grad_norm": 0.06656673550605774, "global_step": 1323, "epoch": 12, "lr": 9.984609913353957e-05} +{"train_loss": 0.025893812999129295, "grad_norm": 0.07840122282505035, "global_step": 1324, "epoch": 12, "lr": 9.98458666866564e-05} +{"train_loss": 0.024341320618987083, "grad_norm": 0.034453146159648895, "global_step": 1325, "epoch": 12, "lr": 9.984563406463653e-05} +{"train_loss": 0.027665341272950172, "grad_norm": 0.04472646489739418, "global_step": 1326, "epoch": 12, "lr": 9.984540126748081e-05} +{"train_loss": 0.02157628908753395, "grad_norm": 0.03827042877674103, "global_step": 1327, "epoch": 12, "lr": 9.984516829518999e-05} +{"train_loss": 0.02359214797616005, "grad_norm": 0.07248339802026749, "global_step": 1328, "epoch": 12, "lr": 9.984493514776494e-05} +{"train_loss": 0.026571951806545258, "grad_norm": 0.07030462473630905, "global_step": 1329, "epoch": 12, "lr": 9.984470182520647e-05} +{"train_loss": 0.02320491336286068, "grad_norm": 0.10904933512210846, "global_step": 1330, "epoch": 12, "lr": 9.98444683275154e-05} +{"train_loss": 0.020564112812280655, "grad_norm": 0.062113869935274124, "global_step": 1331, "epoch": 12, "lr": 9.984423465469254e-05} +{"train_loss": 0.022722456604242325, "grad_norm": 0.06599016487598419, "global_step": 1332, "epoch": 12, "lr": 9.98440008067387e-05} +{"train_loss": 0.023450186476111412, "grad_norm": 0.041055381298065186, "global_step": 1333, "epoch": 12, "lr": 9.984376678365475e-05} +{"train_loss": 0.02258290909230709, "grad_norm": 0.031927332282066345, "global_step": 1334, "epoch": 12, "lr": 9.984353258544146e-05} +{"train_loss": 0.020999111235141754, "grad_norm": 0.08915067464113235, "global_step": 1335, "epoch": 12, "lr": 9.984329821209965e-05} +{"train_loss": 0.026154330000281334, "grad_norm": 0.033601514995098114, "global_step": 1336, "epoch": 12, "lr": 9.984306366363021e-05} +{"train_loss": 0.020172111690044403, "grad_norm": 0.10511648654937744, "global_step": 1337, "epoch": 12, "lr": 9.98428289400339e-05} +{"train_loss": 0.021549783647060394, "grad_norm": 0.07136107236146927, "global_step": 1338, "epoch": 12, "lr": 9.984259404131155e-05} +{"train_loss": 0.02625863626599312, "grad_norm": 0.03528154641389847, "global_step": 1339, "epoch": 12, "lr": 9.984235896746403e-05} +{"train_loss": 0.023598812520503998, "grad_norm": 0.041380301117897034, "global_step": 1340, "epoch": 12, "lr": 9.984212371849211e-05} +{"train_loss": 0.021419081836938858, "grad_norm": 0.08144204318523407, "global_step": 1341, "epoch": 12, "lr": 9.984188829439667e-05} +{"train_loss": 0.022979427129030228, "grad_norm": 0.09210558235645294, "global_step": 1342, "epoch": 12, "lr": 9.984165269517849e-05} +{"train_loss": 0.02319204993546009, "grad_norm": 0.04497818276286125, "global_step": 1343, "epoch": 12, "lr": 9.984141692083841e-05} +{"train_loss": 0.02130233123898506, "grad_norm": 0.052918337285518646, "global_step": 1344, "epoch": 12, "lr": 9.98411809713773e-05} +{"train_loss": 0.02318689227104187, "grad_norm": 0.05201902613043785, "global_step": 1345, "epoch": 12, "lr": 9.984094484679593e-05} +{"train_loss": 0.022891368716955185, "grad_norm": 0.05888282507658005, "global_step": 1346, "epoch": 12, "lr": 9.984070854709515e-05} +{"train_loss": 0.019048482179641724, "grad_norm": 0.046290311962366104, "global_step": 1347, "epoch": 12, "lr": 9.984047207227582e-05} +{"train_loss": 0.025366514921188354, "grad_norm": 0.11344014108181, "global_step": 1348, "epoch": 12, "lr": 9.984023542233872e-05} +{"train_loss": 0.02405545860528946, "grad_norm": 0.054462309926748276, "global_step": 1349, "epoch": 12, "lr": 9.983999859728473e-05} +{"train_loss": 0.022693660110235214, "grad_norm": 0.05969010293483734, "global_step": 1350, "epoch": 12, "lr": 9.983976159711464e-05} +{"train_loss": 0.024652982130646706, "grad_norm": 0.07870401442050934, "global_step": 1351, "epoch": 12, "lr": 9.983952442182932e-05} +{"train_loss": 0.02563849836587906, "grad_norm": 0.047557953745126724, "global_step": 1352, "epoch": 12, "lr": 9.983928707142958e-05} +{"train_loss": 0.01815955713391304, "grad_norm": 0.03613174334168434, "global_step": 1353, "epoch": 12, "lr": 9.983904954591626e-05} +{"train_loss": 0.021044814959168434, "grad_norm": 0.08222055435180664, "global_step": 1354, "epoch": 12, "lr": 9.983881184529019e-05} +{"train_loss": 0.02585708349943161, "grad_norm": 0.04025796800851822, "global_step": 1355, "epoch": 12, "lr": 9.983857396955221e-05} +{"train_loss": 0.025125913321971893, "grad_norm": 0.07628018409013748, "global_step": 1356, "epoch": 12, "lr": 9.983833591870315e-05} +{"train_loss": 0.021967250853776932, "grad_norm": 0.05523604154586792, "global_step": 1357, "epoch": 12, "lr": 9.983809769274386e-05} +{"train_loss": 0.020602315664291382, "grad_norm": 0.1679338812828064, "global_step": 1358, "epoch": 12, "lr": 9.983785929167517e-05} +{"train_loss": 0.02319340407848358, "grad_norm": 0.03815389797091484, "global_step": 1359, "epoch": 12, "lr": 9.983762071549792e-05} +{"train_loss": 0.023289088159799576, "grad_norm": 0.05937544256448746, "global_step": 1360, "epoch": 12, "lr": 9.983738196421293e-05} +{"train_loss": 0.023370729759335518, "grad_norm": 0.1255185753107071, "global_step": 1361, "epoch": 12, "lr": 9.983714303782106e-05} +{"train_loss": 0.024577375501394272, "grad_norm": 0.06767179816961288, "global_step": 1362, "epoch": 12, "lr": 9.983690393632314e-05} +{"train_loss": 0.020098714157938957, "grad_norm": 0.045864880084991455, "global_step": 1363, "epoch": 12, "lr": 9.983666465972003e-05} +{"train_loss": 0.022611960768699646, "grad_norm": 0.0691467821598053, "global_step": 1364, "epoch": 12, "lr": 9.983642520801253e-05} +{"train_loss": 0.022805731743574142, "grad_norm": 0.05279456079006195, "global_step": 1365, "epoch": 12, "lr": 9.983618558120151e-05} +{"train_loss": 0.0227158572524786, "grad_norm": 0.07904629409313202, "global_step": 1366, "epoch": 12, "lr": 9.983594577928784e-05} +{"train_loss": 0.02428865060210228, "grad_norm": 0.11839565634727478, "global_step": 1367, "epoch": 12, "lr": 9.983570580227228e-05} +{"train_loss": 0.023514999076724052, "grad_norm": 0.11293194442987442, "global_step": 1368, "epoch": 12, "lr": 9.983546565015576e-05} +{"train_loss": 0.02243712730705738, "grad_norm": 0.049544911831617355, "global_step": 1369, "epoch": 12, "lr": 9.983522532293907e-05} +{"train_loss": 0.02165595069527626, "grad_norm": 0.08349122852087021, "global_step": 1370, "epoch": 12, "lr": 9.983498482062306e-05} +{"train_loss": 0.02890447899699211, "grad_norm": 0.10082948952913284, "global_step": 1371, "epoch": 12, "lr": 9.983474414320861e-05} +{"train_loss": 0.026966193690896034, "grad_norm": 0.08459935337305069, "global_step": 1372, "epoch": 12, "lr": 9.983450329069652e-05} +{"train_loss": 0.02022401988506317, "grad_norm": 0.061956632882356644, "global_step": 1373, "epoch": 12, "lr": 9.983426226308768e-05} +{"train_loss": 0.021734017878770828, "grad_norm": 0.12789927423000336, "global_step": 1374, "epoch": 12, "lr": 9.983402106038291e-05} +{"train_loss": 0.022398395463824272, "grad_norm": 0.1517702341079712, "global_step": 1375, "epoch": 12, "lr": 9.983377968258305e-05} +{"train_loss": 0.021017711609601974, "grad_norm": 0.03343697264790535, "global_step": 1376, "epoch": 12, "lr": 9.983353812968897e-05} +{"train_loss": 0.022808209448209348, "grad_norm": 0.0811709389090538, "global_step": 1377, "epoch": 12, "lr": 9.983329640170149e-05, "val_loss": 0.021536260843276978} +{"train_loss": 0.022214021533727646, "grad_norm": 0.04711104929447174, "global_step": 1378, "epoch": 13, "lr": 9.983305449862151e-05} +{"train_loss": 0.024564940482378006, "grad_norm": 0.08899375796318054, "global_step": 1379, "epoch": 13, "lr": 9.983281242044982e-05} +{"train_loss": 0.021516479551792145, "grad_norm": 0.056396424770355225, "global_step": 1380, "epoch": 13, "lr": 9.983257016718732e-05} +{"train_loss": 0.021549765020608902, "grad_norm": 0.15878374874591827, "global_step": 1381, "epoch": 13, "lr": 9.983232773883483e-05} +{"train_loss": 0.022267024964094162, "grad_norm": 0.03968986123800278, "global_step": 1382, "epoch": 13, "lr": 9.983208513539322e-05} +{"train_loss": 0.02184775471687317, "grad_norm": 0.06924866139888763, "global_step": 1383, "epoch": 13, "lr": 9.983184235686331e-05} +{"train_loss": 0.027949608862400055, "grad_norm": 0.05139557644724846, "global_step": 1384, "epoch": 13, "lr": 9.9831599403246e-05} +{"train_loss": 0.023924315348267555, "grad_norm": 0.04165400564670563, "global_step": 1385, "epoch": 13, "lr": 9.983135627454212e-05} +{"train_loss": 0.02240126207470894, "grad_norm": 0.10903085768222809, "global_step": 1386, "epoch": 13, "lr": 9.983111297075252e-05} +{"train_loss": 0.026705272495746613, "grad_norm": 0.07922074943780899, "global_step": 1387, "epoch": 13, "lr": 9.983086949187806e-05} +{"train_loss": 0.022225800901651382, "grad_norm": 0.06683356314897537, "global_step": 1388, "epoch": 13, "lr": 9.98306258379196e-05} +{"train_loss": 0.022455904632806778, "grad_norm": 0.12457022815942764, "global_step": 1389, "epoch": 13, "lr": 9.983038200887798e-05} +{"train_loss": 0.022908836603164673, "grad_norm": 0.09191206842660904, "global_step": 1390, "epoch": 13, "lr": 9.983013800475407e-05} +{"train_loss": 0.021442241966724396, "grad_norm": 0.1165793165564537, "global_step": 1391, "epoch": 13, "lr": 9.982989382554874e-05} +{"train_loss": 0.023731078952550888, "grad_norm": 0.0714220255613327, "global_step": 1392, "epoch": 13, "lr": 9.982964947126282e-05} +{"train_loss": 0.022381626069545746, "grad_norm": 0.11420343816280365, "global_step": 1393, "epoch": 13, "lr": 9.982940494189719e-05} +{"train_loss": 0.02111325040459633, "grad_norm": 0.04728211462497711, "global_step": 1394, "epoch": 13, "lr": 9.98291602374527e-05} +{"train_loss": 0.024067532271146774, "grad_norm": 0.047474902123212814, "global_step": 1395, "epoch": 13, "lr": 9.98289153579302e-05} +{"train_loss": 0.023096347227692604, "grad_norm": 0.04554164409637451, "global_step": 1396, "epoch": 13, "lr": 9.982867030333058e-05} +{"train_loss": 0.019777260720729828, "grad_norm": 0.04807192459702492, "global_step": 1397, "epoch": 13, "lr": 9.982842507365467e-05} +{"train_loss": 0.020851222798228264, "grad_norm": 0.17968125641345978, "global_step": 1398, "epoch": 13, "lr": 9.982817966890334e-05} +{"train_loss": 0.023313436657190323, "grad_norm": 0.050260696560144424, "global_step": 1399, "epoch": 13, "lr": 9.982793408907747e-05} +{"train_loss": 0.018458310514688492, "grad_norm": 0.03248438611626625, "global_step": 1400, "epoch": 13, "lr": 9.982768833417791e-05} +{"train_loss": 0.022401314228773117, "grad_norm": 0.037953250110149384, "global_step": 1401, "epoch": 13, "lr": 9.98274424042055e-05} +{"train_loss": 0.02153627946972847, "grad_norm": 0.11883430927991867, "global_step": 1402, "epoch": 13, "lr": 9.982719629916115e-05} +{"train_loss": 0.023532573133707047, "grad_norm": 0.12462447583675385, "global_step": 1403, "epoch": 13, "lr": 9.982695001904569e-05} +{"train_loss": 0.02213282324373722, "grad_norm": 0.11814460158348083, "global_step": 1404, "epoch": 13, "lr": 9.982670356386e-05} +{"train_loss": 0.021739976480603218, "grad_norm": 0.03789651393890381, "global_step": 1405, "epoch": 13, "lr": 9.982645693360496e-05} +{"train_loss": 0.024139083921909332, "grad_norm": 0.03513859584927559, "global_step": 1406, "epoch": 13, "lr": 9.98262101282814e-05} +{"train_loss": 0.022326448932290077, "grad_norm": 0.09751474857330322, "global_step": 1407, "epoch": 13, "lr": 9.98259631478902e-05} +{"train_loss": 0.025144506245851517, "grad_norm": 0.12195079028606415, "global_step": 1408, "epoch": 13, "lr": 9.982571599243224e-05} +{"train_loss": 0.02109384723007679, "grad_norm": 0.068515345454216, "global_step": 1409, "epoch": 13, "lr": 9.982546866190838e-05} +{"train_loss": 0.025543056428432465, "grad_norm": 0.04632791131734848, "global_step": 1410, "epoch": 13, "lr": 9.982522115631951e-05} +{"train_loss": 0.021865680813789368, "grad_norm": 0.06021462380886078, "global_step": 1411, "epoch": 13, "lr": 9.982497347566646e-05} +{"train_loss": 0.022412549704313278, "grad_norm": 0.037086889147758484, "global_step": 1412, "epoch": 13, "lr": 9.982472561995014e-05} +{"train_loss": 0.020808551460504532, "grad_norm": 0.12635648250579834, "global_step": 1413, "epoch": 13, "lr": 9.982447758917139e-05} +{"train_loss": 0.02145472913980484, "grad_norm": 0.06379132717847824, "global_step": 1414, "epoch": 13, "lr": 9.98242293833311e-05} +{"train_loss": 0.022240281105041504, "grad_norm": 0.0853339359164238, "global_step": 1415, "epoch": 13, "lr": 9.982398100243014e-05} +{"train_loss": 0.02075236849486828, "grad_norm": 0.043294891715049744, "global_step": 1416, "epoch": 13, "lr": 9.982373244646937e-05} +{"train_loss": 0.02039230987429619, "grad_norm": 0.04776054993271828, "global_step": 1417, "epoch": 13, "lr": 9.982348371544966e-05} +{"train_loss": 0.024468068033456802, "grad_norm": 0.06825361400842667, "global_step": 1418, "epoch": 13, "lr": 9.982323480937191e-05} +{"train_loss": 0.022481843829154968, "grad_norm": 0.03273007273674011, "global_step": 1419, "epoch": 13, "lr": 9.982298572823698e-05} +{"train_loss": 0.021007563918828964, "grad_norm": 0.047438882291316986, "global_step": 1420, "epoch": 13, "lr": 9.982273647204574e-05} +{"train_loss": 0.021728698164224625, "grad_norm": 0.06422977894544601, "global_step": 1421, "epoch": 13, "lr": 9.982248704079908e-05} +{"train_loss": 0.02465057745575905, "grad_norm": 0.037511397153139114, "global_step": 1422, "epoch": 13, "lr": 9.982223743449785e-05} +{"train_loss": 0.020423978567123413, "grad_norm": 0.08597822487354279, "global_step": 1423, "epoch": 13, "lr": 9.982198765314296e-05} +{"train_loss": 0.022069290280342102, "grad_norm": 0.05325532332062721, "global_step": 1424, "epoch": 13, "lr": 9.982173769673529e-05} +{"train_loss": 0.024154221639037132, "grad_norm": 0.054367877542972565, "global_step": 1425, "epoch": 13, "lr": 9.982148756527567e-05} +{"train_loss": 0.02241208404302597, "grad_norm": 0.06978368014097214, "global_step": 1426, "epoch": 13, "lr": 9.982123725876502e-05} +{"train_loss": 0.025125404819846153, "grad_norm": 0.05492646247148514, "global_step": 1427, "epoch": 13, "lr": 9.982098677720422e-05} +{"train_loss": 0.022614868357777596, "grad_norm": 0.17502687871456146, "global_step": 1428, "epoch": 13, "lr": 9.982073612059413e-05} +{"train_loss": 0.024504519999027252, "grad_norm": 0.04707207903265953, "global_step": 1429, "epoch": 13, "lr": 9.982048528893564e-05} +{"train_loss": 0.023033183068037033, "grad_norm": 0.06354553252458572, "global_step": 1430, "epoch": 13, "lr": 9.982023428222962e-05} +{"train_loss": 0.027251355350017548, "grad_norm": 0.05233258008956909, "global_step": 1431, "epoch": 13, "lr": 9.981998310047698e-05} +{"train_loss": 0.02286774292588234, "grad_norm": 0.04266592115163803, "global_step": 1432, "epoch": 13, "lr": 9.981973174367858e-05} +{"train_loss": 0.02840343490242958, "grad_norm": 0.06485360860824585, "global_step": 1433, "epoch": 13, "lr": 9.98194802118353e-05} +{"train_loss": 0.02164469286799431, "grad_norm": 0.08711729198694229, "global_step": 1434, "epoch": 13, "lr": 9.981922850494805e-05} +{"train_loss": 0.02235671505331993, "grad_norm": 0.14895108342170715, "global_step": 1435, "epoch": 13, "lr": 9.981897662301768e-05} +{"train_loss": 0.023273855447769165, "grad_norm": 0.17186276614665985, "global_step": 1436, "epoch": 13, "lr": 9.98187245660451e-05} +{"train_loss": 0.023935725912451744, "grad_norm": 0.04777250438928604, "global_step": 1437, "epoch": 13, "lr": 9.98184723340312e-05} +{"train_loss": 0.022924041375517845, "grad_norm": 0.03953973948955536, "global_step": 1438, "epoch": 13, "lr": 9.981821992697682e-05} +{"train_loss": 0.024783600121736526, "grad_norm": 0.0759970098733902, "global_step": 1439, "epoch": 13, "lr": 9.981796734488292e-05} +{"train_loss": 0.023333609104156494, "grad_norm": 0.051867496222257614, "global_step": 1440, "epoch": 13, "lr": 9.981771458775031e-05} +{"train_loss": 0.018773619085550308, "grad_norm": 0.030931126326322556, "global_step": 1441, "epoch": 13, "lr": 9.981746165557993e-05} +{"train_loss": 0.02198038622736931, "grad_norm": 0.06550133973360062, "global_step": 1442, "epoch": 13, "lr": 9.981720854837266e-05} +{"train_loss": 0.023254886269569397, "grad_norm": 0.05792524665594101, "global_step": 1443, "epoch": 13, "lr": 9.98169552661294e-05} +{"train_loss": 0.021975688636302948, "grad_norm": 0.06830933690071106, "global_step": 1444, "epoch": 13, "lr": 9.9816701808851e-05} +{"train_loss": 0.022275658324360847, "grad_norm": 0.13028937578201294, "global_step": 1445, "epoch": 13, "lr": 9.981644817653838e-05} +{"train_loss": 0.021116962656378746, "grad_norm": 0.030894363299012184, "global_step": 1446, "epoch": 13, "lr": 9.981619436919243e-05} +{"train_loss": 0.02232866920530796, "grad_norm": 0.040155328810214996, "global_step": 1447, "epoch": 13, "lr": 9.981594038681404e-05} +{"train_loss": 0.024556880816817284, "grad_norm": 0.03279934823513031, "global_step": 1448, "epoch": 13, "lr": 9.981568622940409e-05} +{"train_loss": 0.02087903395295143, "grad_norm": 0.04820727929472923, "global_step": 1449, "epoch": 13, "lr": 9.981543189696348e-05} +{"train_loss": 0.02298812009394169, "grad_norm": 0.1178266704082489, "global_step": 1450, "epoch": 13, "lr": 9.981517738949311e-05} +{"train_loss": 0.02066054940223694, "grad_norm": 0.05227063596248627, "global_step": 1451, "epoch": 13, "lr": 9.981492270699389e-05} +{"train_loss": 0.02182086557149887, "grad_norm": 0.07490942627191544, "global_step": 1452, "epoch": 13, "lr": 9.981466784946666e-05} +{"train_loss": 0.02003590762615204, "grad_norm": 0.028933217748999596, "global_step": 1453, "epoch": 13, "lr": 9.981441281691238e-05} +{"train_loss": 0.024167444556951523, "grad_norm": 0.038101840764284134, "global_step": 1454, "epoch": 13, "lr": 9.981415760933191e-05} +{"train_loss": 0.023393455892801285, "grad_norm": 0.036213893443346024, "global_step": 1455, "epoch": 13, "lr": 9.981390222672615e-05} +{"train_loss": 0.023843303322792053, "grad_norm": 0.06029853597283363, "global_step": 1456, "epoch": 13, "lr": 9.9813646669096e-05} +{"train_loss": 0.02198854461312294, "grad_norm": 0.05459365248680115, "global_step": 1457, "epoch": 13, "lr": 9.981339093644237e-05} +{"train_loss": 0.023759905248880386, "grad_norm": 0.0640915110707283, "global_step": 1458, "epoch": 13, "lr": 9.981313502876615e-05} +{"train_loss": 0.022422771900892258, "grad_norm": 0.04635148122906685, "global_step": 1459, "epoch": 13, "lr": 9.981287894606822e-05} +{"train_loss": 0.023236915469169617, "grad_norm": 0.039941392838954926, "global_step": 1460, "epoch": 13, "lr": 9.98126226883495e-05} +{"train_loss": 0.023297583684325218, "grad_norm": 0.05229334905743599, "global_step": 1461, "epoch": 13, "lr": 9.98123662556109e-05} +{"train_loss": 0.02195075713098049, "grad_norm": 0.11375901103019714, "global_step": 1462, "epoch": 13, "lr": 9.981210964785328e-05} +{"train_loss": 0.02301732264459133, "grad_norm": 0.12744532525539398, "global_step": 1463, "epoch": 13, "lr": 9.981185286507759e-05} +{"train_loss": 0.021493688225746155, "grad_norm": 0.09664681553840637, "global_step": 1464, "epoch": 13, "lr": 9.981159590728471e-05} +{"train_loss": 0.021599553525447845, "grad_norm": 0.04883016645908356, "global_step": 1465, "epoch": 13, "lr": 9.981133877447555e-05} +{"train_loss": 0.02205463871359825, "grad_norm": 0.046306025236845016, "global_step": 1466, "epoch": 13, "lr": 9.9811081466651e-05} +{"train_loss": 0.024880100041627884, "grad_norm": 0.060890380293130875, "global_step": 1467, "epoch": 13, "lr": 9.981082398381196e-05} +{"train_loss": 0.024255435913801193, "grad_norm": 0.04414210468530655, "global_step": 1468, "epoch": 13, "lr": 9.981056632595936e-05} +{"train_loss": 0.022093787789344788, "grad_norm": 0.05161961913108826, "global_step": 1469, "epoch": 13, "lr": 9.98103084930941e-05} +{"train_loss": 0.024857811629772186, "grad_norm": 0.0783422663807869, "global_step": 1470, "epoch": 13, "lr": 9.981005048521706e-05} +{"train_loss": 0.02223600447177887, "grad_norm": 0.04441870003938675, "global_step": 1471, "epoch": 13, "lr": 9.980979230232919e-05} +{"train_loss": 0.02250872552394867, "grad_norm": 0.10307317227125168, "global_step": 1472, "epoch": 13, "lr": 9.980953394443136e-05} +{"train_loss": 0.02206040918827057, "grad_norm": 0.12489140778779984, "global_step": 1473, "epoch": 13, "lr": 9.980927541152447e-05} +{"train_loss": 0.021107584238052368, "grad_norm": 0.06314852833747864, "global_step": 1474, "epoch": 13, "lr": 9.980901670360948e-05} +{"train_loss": 0.024427466094493866, "grad_norm": 0.06165657192468643, "global_step": 1475, "epoch": 13, "lr": 9.980875782068724e-05} +{"train_loss": 0.021495379507541656, "grad_norm": 0.050873931497335434, "global_step": 1476, "epoch": 13, "lr": 9.980849876275868e-05} +{"train_loss": 0.02159067988395691, "grad_norm": 0.03867705911397934, "global_step": 1477, "epoch": 13, "lr": 9.980823952982474e-05} +{"train_loss": 0.021567150950431824, "grad_norm": 0.03555049002170563, "global_step": 1478, "epoch": 13, "lr": 9.98079801218863e-05} +{"train_loss": 0.020643215626478195, "grad_norm": 0.05810963734984398, "global_step": 1479, "epoch": 13, "lr": 9.980772053894427e-05} +{"train_loss": 0.02081085368990898, "grad_norm": 0.08228681981563568, "global_step": 1480, "epoch": 13, "lr": 9.980746078099958e-05} +{"train_loss": 0.021778978407382965, "grad_norm": 0.0463578961789608, "global_step": 1481, "epoch": 13, "lr": 9.980720084805314e-05} +{"train_loss": 0.024098187685012817, "grad_norm": 0.04327861964702606, "global_step": 1482, "epoch": 13, "lr": 9.980694074010585e-05} +{"train_loss": 0.02264513825681412, "grad_norm": 0.10062047839164734, "global_step": 1483, "epoch": 13, "lr": 9.980668045715864e-05, "val_loss": 0.020934518426656723} +{"train_loss": 0.023250140249729156, "grad_norm": 0.04508966580033302, "global_step": 1484, "epoch": 14, "lr": 9.980641999921238e-05} +{"train_loss": 0.024055782705545425, "grad_norm": 0.1007465198636055, "global_step": 1485, "epoch": 14, "lr": 9.980615936626807e-05} +{"train_loss": 0.023163694888353348, "grad_norm": 0.045159753412008286, "global_step": 1486, "epoch": 14, "lr": 9.980589855832654e-05} +{"train_loss": 0.02188606560230255, "grad_norm": 0.040355194360017776, "global_step": 1487, "epoch": 14, "lr": 9.980563757538875e-05} +{"train_loss": 0.022992905229330063, "grad_norm": 0.0692940205335617, "global_step": 1488, "epoch": 14, "lr": 9.980537641745562e-05} +{"train_loss": 0.0218384750187397, "grad_norm": 0.06262573599815369, "global_step": 1489, "epoch": 14, "lr": 9.980511508452806e-05} +{"train_loss": 0.023740222677588463, "grad_norm": 0.044073667377233505, "global_step": 1490, "epoch": 14, "lr": 9.980485357660696e-05} +{"train_loss": 0.02497221902012825, "grad_norm": 0.031082268804311752, "global_step": 1491, "epoch": 14, "lr": 9.980459189369329e-05} +{"train_loss": 0.02343442291021347, "grad_norm": 0.13442464172840118, "global_step": 1492, "epoch": 14, "lr": 9.980433003578793e-05} +{"train_loss": 0.02129947394132614, "grad_norm": 0.05810866132378578, "global_step": 1493, "epoch": 14, "lr": 9.980406800289182e-05} +{"train_loss": 0.02266465499997139, "grad_norm": 0.05994570627808571, "global_step": 1494, "epoch": 14, "lr": 9.980380579500587e-05} +{"train_loss": 0.01983380690217018, "grad_norm": 0.08158054202795029, "global_step": 1495, "epoch": 14, "lr": 9.9803543412131e-05} +{"train_loss": 0.023644737899303436, "grad_norm": 0.05470539256930351, "global_step": 1496, "epoch": 14, "lr": 9.980328085426814e-05} +{"train_loss": 0.024264849722385406, "grad_norm": 0.1889009028673172, "global_step": 1497, "epoch": 14, "lr": 9.980301812141822e-05} +{"train_loss": 0.020639870315790176, "grad_norm": 0.07250458002090454, "global_step": 1498, "epoch": 14, "lr": 9.980275521358216e-05} +{"train_loss": 0.024023961275815964, "grad_norm": 0.04793822392821312, "global_step": 1499, "epoch": 14, "lr": 9.980249213076084e-05} +{"train_loss": 0.022588597610592842, "grad_norm": 0.04594345763325691, "global_step": 1500, "epoch": 14, "lr": 9.980222887295525e-05} +{"train_loss": 0.021727992221713066, "grad_norm": 0.08068013936281204, "global_step": 1501, "epoch": 14, "lr": 9.980196544016629e-05} +{"train_loss": 0.021472010761499405, "grad_norm": 0.047319795936346054, "global_step": 1502, "epoch": 14, "lr": 9.980170183239488e-05} +{"train_loss": 0.02247123420238495, "grad_norm": 0.05180756002664566, "global_step": 1503, "epoch": 14, "lr": 9.980143804964193e-05} +{"train_loss": 0.022814467549324036, "grad_norm": 0.06684641540050507, "global_step": 1504, "epoch": 14, "lr": 9.98011740919084e-05} +{"train_loss": 0.01964770257472992, "grad_norm": 0.0641021728515625, "global_step": 1505, "epoch": 14, "lr": 9.98009099591952e-05} +{"train_loss": 0.022669438272714615, "grad_norm": 0.08750231564044952, "global_step": 1506, "epoch": 14, "lr": 9.980064565150325e-05} +{"train_loss": 0.023238714784383774, "grad_norm": 0.11534704267978668, "global_step": 1507, "epoch": 14, "lr": 9.98003811688335e-05} +{"train_loss": 0.024695387110114098, "grad_norm": 0.07288343459367752, "global_step": 1508, "epoch": 14, "lr": 9.980011651118686e-05} +{"train_loss": 0.022900432348251343, "grad_norm": 0.05768652260303497, "global_step": 1509, "epoch": 14, "lr": 9.979985167856429e-05} +{"train_loss": 0.022050131112337112, "grad_norm": 0.11536630988121033, "global_step": 1510, "epoch": 14, "lr": 9.979958667096667e-05} +{"train_loss": 0.02200521156191826, "grad_norm": 0.04438501223921776, "global_step": 1511, "epoch": 14, "lr": 9.979932148839497e-05} +{"train_loss": 0.021107222884893417, "grad_norm": 0.049374476075172424, "global_step": 1512, "epoch": 14, "lr": 9.979905613085011e-05} +{"train_loss": 0.020442143082618713, "grad_norm": 0.04908207058906555, "global_step": 1513, "epoch": 14, "lr": 9.979879059833304e-05} +{"train_loss": 0.02563788741827011, "grad_norm": 0.05436432734131813, "global_step": 1514, "epoch": 14, "lr": 9.979852489084465e-05} +{"train_loss": 0.023020192980766296, "grad_norm": 0.08279725164175034, "global_step": 1515, "epoch": 14, "lr": 9.979825900838591e-05} +{"train_loss": 0.02228661999106407, "grad_norm": 0.12664461135864258, "global_step": 1516, "epoch": 14, "lr": 9.979799295095775e-05} +{"train_loss": 0.023034734651446342, "grad_norm": 0.05153694748878479, "global_step": 1517, "epoch": 14, "lr": 9.979772671856111e-05} +{"train_loss": 0.02021394856274128, "grad_norm": 0.035265952348709106, "global_step": 1518, "epoch": 14, "lr": 9.97974603111969e-05} +{"train_loss": 0.02217940241098404, "grad_norm": 0.03872644901275635, "global_step": 1519, "epoch": 14, "lr": 9.979719372886607e-05} +{"train_loss": 0.024139225482940674, "grad_norm": 0.0834607183933258, "global_step": 1520, "epoch": 14, "lr": 9.979692697156958e-05} +{"train_loss": 0.023757636547088623, "grad_norm": 0.07512663304805756, "global_step": 1521, "epoch": 14, "lr": 9.979666003930832e-05} +{"train_loss": 0.023509828373789787, "grad_norm": 0.12110596150159836, "global_step": 1522, "epoch": 14, "lr": 9.979639293208325e-05} +{"train_loss": 0.024160904809832573, "grad_norm": 0.040362171828746796, "global_step": 1523, "epoch": 14, "lr": 9.97961256498953e-05} +{"train_loss": 0.02028648555278778, "grad_norm": 0.07847896218299866, "global_step": 1524, "epoch": 14, "lr": 9.979585819274547e-05} +{"train_loss": 0.019799411296844482, "grad_norm": 0.057184670120477676, "global_step": 1525, "epoch": 14, "lr": 9.979559056063462e-05} +{"train_loss": 0.023840459063649178, "grad_norm": 0.0530632883310318, "global_step": 1526, "epoch": 14, "lr": 9.979532275356372e-05} +{"train_loss": 0.0221199169754982, "grad_norm": 0.03412040323019028, "global_step": 1527, "epoch": 14, "lr": 9.97950547715337e-05} +{"train_loss": 0.02087138034403324, "grad_norm": 0.07706596702337265, "global_step": 1528, "epoch": 14, "lr": 9.979478661454553e-05} +{"train_loss": 0.02577792853116989, "grad_norm": 0.04882718622684479, "global_step": 1529, "epoch": 14, "lr": 9.979451828260012e-05} +{"train_loss": 0.020010467618703842, "grad_norm": 0.036991849541664124, "global_step": 1530, "epoch": 14, "lr": 9.979424977569844e-05} +{"train_loss": 0.022026946768164635, "grad_norm": 0.11476834863424301, "global_step": 1531, "epoch": 14, "lr": 9.979398109384142e-05} +{"train_loss": 0.021404579281806946, "grad_norm": 0.11998691409826279, "global_step": 1532, "epoch": 14, "lr": 9.979371223703e-05} +{"train_loss": 0.02388952486217022, "grad_norm": 0.05534064397215843, "global_step": 1533, "epoch": 14, "lr": 9.979344320526513e-05} +{"train_loss": 0.02448756992816925, "grad_norm": 0.07869622111320496, "global_step": 1534, "epoch": 14, "lr": 9.979317399854776e-05} +{"train_loss": 0.02182013913989067, "grad_norm": 0.040771692991256714, "global_step": 1535, "epoch": 14, "lr": 9.979290461687882e-05} +{"train_loss": 0.021141834557056427, "grad_norm": 0.08128616958856583, "global_step": 1536, "epoch": 14, "lr": 9.979263506025929e-05} +{"train_loss": 0.02049599587917328, "grad_norm": 0.07922063767910004, "global_step": 1537, "epoch": 14, "lr": 9.979236532869008e-05} +{"train_loss": 0.01854069158434868, "grad_norm": 0.05708588659763336, "global_step": 1538, "epoch": 14, "lr": 9.979209542217217e-05} +{"train_loss": 0.025349190458655357, "grad_norm": 0.04559232294559479, "global_step": 1539, "epoch": 14, "lr": 9.979182534070646e-05} +{"train_loss": 0.024429205805063248, "grad_norm": 0.05634671449661255, "global_step": 1540, "epoch": 14, "lr": 9.979155508429394e-05} +{"train_loss": 0.02229323983192444, "grad_norm": 0.10806314647197723, "global_step": 1541, "epoch": 14, "lr": 9.979128465293556e-05} +{"train_loss": 0.021837692707777023, "grad_norm": 0.08547835797071457, "global_step": 1542, "epoch": 14, "lr": 9.979101404663225e-05} +{"train_loss": 0.020677870139479637, "grad_norm": 0.03470625355839729, "global_step": 1543, "epoch": 14, "lr": 9.979074326538498e-05} +{"train_loss": 0.019404247403144836, "grad_norm": 0.0987231433391571, "global_step": 1544, "epoch": 14, "lr": 9.979047230919468e-05} +{"train_loss": 0.022470636293292046, "grad_norm": 0.030223917216062546, "global_step": 1545, "epoch": 14, "lr": 9.979020117806232e-05} +{"train_loss": 0.021279267966747284, "grad_norm": 0.04892699420452118, "global_step": 1546, "epoch": 14, "lr": 9.978992987198885e-05} +{"train_loss": 0.020107844844460487, "grad_norm": 0.042287591844797134, "global_step": 1547, "epoch": 14, "lr": 9.97896583909752e-05} +{"train_loss": 0.022449232637882233, "grad_norm": 0.09105414152145386, "global_step": 1548, "epoch": 14, "lr": 9.978938673502236e-05} +{"train_loss": 0.021204955875873566, "grad_norm": 0.033599067479372025, "global_step": 1549, "epoch": 14, "lr": 9.978911490413127e-05} +{"train_loss": 0.024311088025569916, "grad_norm": 0.06154730170965195, "global_step": 1550, "epoch": 14, "lr": 9.978884289830289e-05} +{"train_loss": 0.024524904787540436, "grad_norm": 0.0750950425863266, "global_step": 1551, "epoch": 14, "lr": 9.978857071753814e-05} +{"train_loss": 0.0222819484770298, "grad_norm": 0.0890950858592987, "global_step": 1552, "epoch": 14, "lr": 9.978829836183803e-05} +{"train_loss": 0.023656416684389114, "grad_norm": 0.1019781231880188, "global_step": 1553, "epoch": 14, "lr": 9.978802583120348e-05} +{"train_loss": 0.023216083645820618, "grad_norm": 0.053128115832805634, "global_step": 1554, "epoch": 14, "lr": 9.978775312563546e-05} +{"train_loss": 0.022873524576425552, "grad_norm": 0.05938899517059326, "global_step": 1555, "epoch": 14, "lr": 9.978748024513493e-05} +{"train_loss": 0.025205524638295174, "grad_norm": 0.07191074639558792, "global_step": 1556, "epoch": 14, "lr": 9.978720718970285e-05} +{"train_loss": 0.022868921980261803, "grad_norm": 0.08195941895246506, "global_step": 1557, "epoch": 14, "lr": 9.978693395934017e-05} +{"train_loss": 0.021760523319244385, "grad_norm": 0.1255938559770584, "global_step": 1558, "epoch": 14, "lr": 9.978666055404786e-05} +{"train_loss": 0.021347420290112495, "grad_norm": 0.09348344802856445, "global_step": 1559, "epoch": 14, "lr": 9.978638697382687e-05} +{"train_loss": 0.022139068692922592, "grad_norm": 0.05716967210173607, "global_step": 1560, "epoch": 14, "lr": 9.978611321867818e-05} +{"train_loss": 0.021919939666986465, "grad_norm": 0.05374050885438919, "global_step": 1561, "epoch": 14, "lr": 9.978583928860274e-05} +{"train_loss": 0.02428305521607399, "grad_norm": 0.06571225821971893, "global_step": 1562, "epoch": 14, "lr": 9.978556518360149e-05} +{"train_loss": 0.02576269581913948, "grad_norm": 0.21790099143981934, "global_step": 1563, "epoch": 14, "lr": 9.978529090367544e-05} +{"train_loss": 0.022542830556631088, "grad_norm": 0.08864030241966248, "global_step": 1564, "epoch": 14, "lr": 9.978501644882552e-05} +{"train_loss": 0.023030217736959457, "grad_norm": 0.06128954514861107, "global_step": 1565, "epoch": 14, "lr": 9.97847418190527e-05} +{"train_loss": 0.023370862007141113, "grad_norm": 0.040491580963134766, "global_step": 1566, "epoch": 14, "lr": 9.978446701435795e-05} +{"train_loss": 0.02250424586236477, "grad_norm": 0.18044598400592804, "global_step": 1567, "epoch": 14, "lr": 9.978419203474223e-05} +{"train_loss": 0.023891601711511612, "grad_norm": 0.05678635835647583, "global_step": 1568, "epoch": 14, "lr": 9.978391688020651e-05} +{"train_loss": 0.02087513543665409, "grad_norm": 0.09689992666244507, "global_step": 1569, "epoch": 14, "lr": 9.978364155075178e-05} +{"train_loss": 0.024546753615140915, "grad_norm": 0.0803535059094429, "global_step": 1570, "epoch": 14, "lr": 9.978336604637895e-05} +{"train_loss": 0.018952030688524246, "grad_norm": 0.09516139328479767, "global_step": 1571, "epoch": 14, "lr": 9.978309036708903e-05} +{"train_loss": 0.020925886929035187, "grad_norm": 0.028723089024424553, "global_step": 1572, "epoch": 14, "lr": 9.978281451288298e-05} +{"train_loss": 0.02069276198744774, "grad_norm": 0.10550984740257263, "global_step": 1573, "epoch": 14, "lr": 9.978253848376177e-05} +{"train_loss": 0.024661147966980934, "grad_norm": 0.11498662829399109, "global_step": 1574, "epoch": 14, "lr": 9.97822622797264e-05} +{"train_loss": 0.026493936777114868, "grad_norm": 0.04993389546871185, "global_step": 1575, "epoch": 14, "lr": 9.978198590077779e-05} +{"train_loss": 0.024824440479278564, "grad_norm": 0.07172885537147522, "global_step": 1576, "epoch": 14, "lr": 9.978170934691691e-05} +{"train_loss": 0.021455595269799232, "grad_norm": 0.06168758496642113, "global_step": 1577, "epoch": 14, "lr": 9.978143261814477e-05} +{"train_loss": 0.022824907675385475, "grad_norm": 0.054074253886938095, "global_step": 1578, "epoch": 14, "lr": 9.978115571446233e-05} +{"train_loss": 0.02311136946082115, "grad_norm": 0.133219376206398, "global_step": 1579, "epoch": 14, "lr": 9.978087863587056e-05} +{"train_loss": 0.02215491235256195, "grad_norm": 0.05873400345444679, "global_step": 1580, "epoch": 14, "lr": 9.978060138237043e-05} +{"train_loss": 0.02365928888320923, "grad_norm": 0.06429395079612732, "global_step": 1581, "epoch": 14, "lr": 9.97803239539629e-05} +{"train_loss": 0.02074778638780117, "grad_norm": 0.062379539012908936, "global_step": 1582, "epoch": 14, "lr": 9.978004635064897e-05} +{"train_loss": 0.019949140027165413, "grad_norm": 0.09150306135416031, "global_step": 1583, "epoch": 14, "lr": 9.977976857242961e-05} +{"train_loss": 0.020713545382022858, "grad_norm": 0.08658776432275772, "global_step": 1584, "epoch": 14, "lr": 9.977949061930578e-05} +{"train_loss": 0.021276652812957764, "grad_norm": 0.05725586786866188, "global_step": 1585, "epoch": 14, "lr": 9.977921249127849e-05} +{"train_loss": 0.019499192014336586, "grad_norm": 0.07254434376955032, "global_step": 1586, "epoch": 14, "lr": 9.977893418834867e-05} +{"train_loss": 0.020518507808446884, "grad_norm": 0.037080343812704086, "global_step": 1587, "epoch": 14, "lr": 9.977865571051734e-05} +{"train_loss": 0.023592961952090263, "grad_norm": 0.057860903441905975, "global_step": 1588, "epoch": 14, "lr": 9.977837705778546e-05} +{"train_loss": 0.02247980010326741, "grad_norm": 0.17998167872428894, "global_step": 1589, "epoch": 14, "lr": 9.977809823015401e-05, "val_loss": 0.02125772461295128} +{"train_loss": 0.02112302929162979, "grad_norm": 0.04551950842142105, "global_step": 1590, "epoch": 15, "lr": 9.977781922762397e-05} +{"train_loss": 0.02319730818271637, "grad_norm": 0.03918036073446274, "global_step": 1591, "epoch": 15, "lr": 9.977754005019631e-05} +{"train_loss": 0.02245175652205944, "grad_norm": 0.07920479774475098, "global_step": 1592, "epoch": 15, "lr": 9.977726069787204e-05} +{"train_loss": 0.020699311047792435, "grad_norm": 0.05659707263112068, "global_step": 1593, "epoch": 15, "lr": 9.977698117065212e-05} +{"train_loss": 0.021231401711702347, "grad_norm": 0.10652196407318115, "global_step": 1594, "epoch": 15, "lr": 9.977670146853753e-05} +{"train_loss": 0.023217767477035522, "grad_norm": 0.10501687228679657, "global_step": 1595, "epoch": 15, "lr": 9.977642159152925e-05} +{"train_loss": 0.02387366257607937, "grad_norm": 0.05712868645787239, "global_step": 1596, "epoch": 15, "lr": 9.977614153962829e-05} +{"train_loss": 0.020885836333036423, "grad_norm": 0.09543144702911377, "global_step": 1597, "epoch": 15, "lr": 9.97758613128356e-05} +{"train_loss": 0.019105087965726852, "grad_norm": 0.04729966074228287, "global_step": 1598, "epoch": 15, "lr": 9.97755809111522e-05} +{"train_loss": 0.02237749844789505, "grad_norm": 0.06270604580640793, "global_step": 1599, "epoch": 15, "lr": 9.977530033457905e-05} +{"train_loss": 0.022547690197825432, "grad_norm": 0.13432647287845612, "global_step": 1600, "epoch": 15, "lr": 9.977501958311712e-05} +{"train_loss": 0.02172672748565674, "grad_norm": 0.057181719690561295, "global_step": 1601, "epoch": 15, "lr": 9.977473865676744e-05} +{"train_loss": 0.020550299435853958, "grad_norm": 0.0540640614926815, "global_step": 1602, "epoch": 15, "lr": 9.977445755553096e-05} +{"train_loss": 0.020664796233177185, "grad_norm": 0.06442753970623016, "global_step": 1603, "epoch": 15, "lr": 9.977417627940871e-05} +{"train_loss": 0.017800351604819298, "grad_norm": 0.04450558125972748, "global_step": 1604, "epoch": 15, "lr": 9.977389482840161e-05} +{"train_loss": 0.024989530444145203, "grad_norm": 0.05680717155337334, "global_step": 1605, "epoch": 15, "lr": 9.977361320251072e-05} +{"train_loss": 0.023089587688446045, "grad_norm": 0.04799182340502739, "global_step": 1606, "epoch": 15, "lr": 9.9773331401737e-05} +{"train_loss": 0.019018050283193588, "grad_norm": 0.04067850857973099, "global_step": 1607, "epoch": 15, "lr": 9.977304942608141e-05} +{"train_loss": 0.024770542979240417, "grad_norm": 0.03958810865879059, "global_step": 1608, "epoch": 15, "lr": 9.977276727554499e-05} +{"train_loss": 0.021002888679504395, "grad_norm": 0.08036492019891739, "global_step": 1609, "epoch": 15, "lr": 9.977248495012873e-05} +{"train_loss": 0.020357053726911545, "grad_norm": 0.03659331053495407, "global_step": 1610, "epoch": 15, "lr": 9.977220244983357e-05} +{"train_loss": 0.028067389503121376, "grad_norm": 0.09631071984767914, "global_step": 1611, "epoch": 15, "lr": 9.977191977466055e-05} +{"train_loss": 0.020002266392111778, "grad_norm": 0.031374599784612656, "global_step": 1612, "epoch": 15, "lr": 9.977163692461065e-05} +{"train_loss": 0.02531982585787773, "grad_norm": 0.11119439452886581, "global_step": 1613, "epoch": 15, "lr": 9.977135389968486e-05} +{"train_loss": 0.02043682523071766, "grad_norm": 0.04253304749727249, "global_step": 1614, "epoch": 15, "lr": 9.977107069988418e-05} +{"train_loss": 0.020196666941046715, "grad_norm": 0.06220084801316261, "global_step": 1615, "epoch": 15, "lr": 9.97707873252096e-05} +{"train_loss": 0.024852152913808823, "grad_norm": 0.06395718455314636, "global_step": 1616, "epoch": 15, "lr": 9.977050377566213e-05} +{"train_loss": 0.02107531577348709, "grad_norm": 0.04578297212719917, "global_step": 1617, "epoch": 15, "lr": 9.977022005124273e-05} +{"train_loss": 0.022432155907154083, "grad_norm": 0.03943555802106857, "global_step": 1618, "epoch": 15, "lr": 9.976993615195245e-05} +{"train_loss": 0.02359418198466301, "grad_norm": 0.06376904994249344, "global_step": 1619, "epoch": 15, "lr": 9.976965207779223e-05} +{"train_loss": 0.026462551206350327, "grad_norm": 0.055473167449235916, "global_step": 1620, "epoch": 15, "lr": 9.976936782876312e-05} +{"train_loss": 0.019670583307743073, "grad_norm": 0.07960660010576248, "global_step": 1621, "epoch": 15, "lr": 9.976908340486608e-05} +{"train_loss": 0.02220844477415085, "grad_norm": 0.061980951577425, "global_step": 1622, "epoch": 15, "lr": 9.976879880610213e-05} +{"train_loss": 0.02009541727602482, "grad_norm": 0.03346658870577812, "global_step": 1623, "epoch": 15, "lr": 9.976851403247227e-05} +{"train_loss": 0.02552061527967453, "grad_norm": 0.09366227686405182, "global_step": 1624, "epoch": 15, "lr": 9.976822908397749e-05} +{"train_loss": 0.023108238354325294, "grad_norm": 0.06251104921102524, "global_step": 1625, "epoch": 15, "lr": 9.976794396061879e-05} +{"train_loss": 0.024350401014089584, "grad_norm": 0.11626742035150528, "global_step": 1626, "epoch": 15, "lr": 9.976765866239719e-05} +{"train_loss": 0.02269253320991993, "grad_norm": 0.07283515483140945, "global_step": 1627, "epoch": 15, "lr": 9.976737318931369e-05} +{"train_loss": 0.022831236943602562, "grad_norm": 0.08352439105510712, "global_step": 1628, "epoch": 15, "lr": 9.976708754136926e-05} +{"train_loss": 0.023679520934820175, "grad_norm": 0.04682416096329689, "global_step": 1629, "epoch": 15, "lr": 9.976680171856494e-05} +{"train_loss": 0.02235332503914833, "grad_norm": 0.03465481474995613, "global_step": 1630, "epoch": 15, "lr": 9.976651572090173e-05} +{"train_loss": 0.021289605647325516, "grad_norm": 0.0980529710650444, "global_step": 1631, "epoch": 15, "lr": 9.976622954838062e-05} +{"train_loss": 0.020099710673093796, "grad_norm": 0.05714557319879532, "global_step": 1632, "epoch": 15, "lr": 9.97659432010026e-05} +{"train_loss": 0.022500989958643913, "grad_norm": 0.03939797356724739, "global_step": 1633, "epoch": 15, "lr": 9.976565667876874e-05} +{"train_loss": 0.023101724684238434, "grad_norm": 0.09880693256855011, "global_step": 1634, "epoch": 15, "lr": 9.976536998167998e-05} +{"train_loss": 0.02453836053609848, "grad_norm": 0.11726853251457214, "global_step": 1635, "epoch": 15, "lr": 9.976508310973736e-05} +{"train_loss": 0.02250376157462597, "grad_norm": 0.08025427907705307, "global_step": 1636, "epoch": 15, "lr": 9.976479606294187e-05} +{"train_loss": 0.023492947220802307, "grad_norm": 0.07121452689170837, "global_step": 1637, "epoch": 15, "lr": 9.976450884129454e-05} +{"train_loss": 0.018960192799568176, "grad_norm": 0.10785388201475143, "global_step": 1638, "epoch": 15, "lr": 9.976422144479635e-05} +{"train_loss": 0.020569099113345146, "grad_norm": 0.03740467131137848, "global_step": 1639, "epoch": 15, "lr": 9.976393387344834e-05} +{"train_loss": 0.023349769413471222, "grad_norm": 0.03790887072682381, "global_step": 1640, "epoch": 15, "lr": 9.976364612725152e-05} +{"train_loss": 0.023873450234532356, "grad_norm": 0.03805336356163025, "global_step": 1641, "epoch": 15, "lr": 9.976335820620686e-05} +{"train_loss": 0.0223483145236969, "grad_norm": 0.053328737616539, "global_step": 1642, "epoch": 15, "lr": 9.976307011031542e-05} +{"train_loss": 0.020953785628080368, "grad_norm": 0.10383422672748566, "global_step": 1643, "epoch": 15, "lr": 9.976278183957818e-05} +{"train_loss": 0.021503783762454987, "grad_norm": 0.06828935444355011, "global_step": 1644, "epoch": 15, "lr": 9.976249339399619e-05} +{"train_loss": 0.02206522412598133, "grad_norm": 0.054199062287807465, "global_step": 1645, "epoch": 15, "lr": 9.976220477357043e-05} +{"train_loss": 0.02088620886206627, "grad_norm": 0.08734086155891418, "global_step": 1646, "epoch": 15, "lr": 9.976191597830191e-05} +{"train_loss": 0.022801052778959274, "grad_norm": 0.14551284909248352, "global_step": 1647, "epoch": 15, "lr": 9.976162700819167e-05} +{"train_loss": 0.020045991986989975, "grad_norm": 0.044853147119283676, "global_step": 1648, "epoch": 15, "lr": 9.976133786324072e-05} +{"train_loss": 0.019315684214234352, "grad_norm": 0.02878306806087494, "global_step": 1649, "epoch": 15, "lr": 9.976104854345006e-05} +{"train_loss": 0.02050212025642395, "grad_norm": 0.07741989940404892, "global_step": 1650, "epoch": 15, "lr": 9.976075904882071e-05} +{"train_loss": 0.022505607455968857, "grad_norm": 0.05701747164130211, "global_step": 1651, "epoch": 15, "lr": 9.976046937935371e-05} +{"train_loss": 0.02110752835869789, "grad_norm": 0.11842145770788193, "global_step": 1652, "epoch": 15, "lr": 9.976017953505004e-05} +{"train_loss": 0.022703662514686584, "grad_norm": 0.08410868793725967, "global_step": 1653, "epoch": 15, "lr": 9.975988951591075e-05} +{"train_loss": 0.02492033876478672, "grad_norm": 0.04495769739151001, "global_step": 1654, "epoch": 15, "lr": 9.975959932193685e-05} +{"train_loss": 0.021862395107746124, "grad_norm": 0.046285297721624374, "global_step": 1655, "epoch": 15, "lr": 9.975930895312935e-05} +{"train_loss": 0.023664209991693497, "grad_norm": 0.09205058962106705, "global_step": 1656, "epoch": 15, "lr": 9.97590184094893e-05} +{"train_loss": 0.0224299356341362, "grad_norm": 0.06768324971199036, "global_step": 1657, "epoch": 15, "lr": 9.975872769101766e-05} +{"train_loss": 0.02137664332985878, "grad_norm": 0.13401579856872559, "global_step": 1658, "epoch": 15, "lr": 9.975843679771553e-05} +{"train_loss": 0.023225087672472, "grad_norm": 0.14684602618217468, "global_step": 1659, "epoch": 15, "lr": 9.975814572958387e-05} +{"train_loss": 0.020757704973220825, "grad_norm": 0.1435670554637909, "global_step": 1660, "epoch": 15, "lr": 9.975785448662372e-05} +{"train_loss": 0.022641487419605255, "grad_norm": 0.08846277743577957, "global_step": 1661, "epoch": 15, "lr": 9.975756306883612e-05} +{"train_loss": 0.02282140403985977, "grad_norm": 0.1281692534685135, "global_step": 1662, "epoch": 15, "lr": 9.975727147622209e-05} +{"train_loss": 0.023851606994867325, "grad_norm": 0.07399620115756989, "global_step": 1663, "epoch": 15, "lr": 9.975697970878265e-05} +{"train_loss": 0.021776311099529266, "grad_norm": 0.08890806883573532, "global_step": 1664, "epoch": 15, "lr": 9.97566877665188e-05} +{"train_loss": 0.0246506929397583, "grad_norm": 0.044004037976264954, "global_step": 1665, "epoch": 15, "lr": 9.97563956494316e-05} +{"train_loss": 0.021711867302656174, "grad_norm": 0.11741223931312561, "global_step": 1666, "epoch": 15, "lr": 9.975610335752206e-05} +{"train_loss": 0.019176792353391647, "grad_norm": 0.09230121225118637, "global_step": 1667, "epoch": 15, "lr": 9.975581089079121e-05} +{"train_loss": 0.02128371223807335, "grad_norm": 0.03252848610281944, "global_step": 1668, "epoch": 15, "lr": 9.975551824924008e-05} +{"train_loss": 0.02408413030207157, "grad_norm": 0.051523856818675995, "global_step": 1669, "epoch": 15, "lr": 9.97552254328697e-05} +{"train_loss": 0.023536354303359985, "grad_norm": 0.03950534015893936, "global_step": 1670, "epoch": 15, "lr": 9.97549324416811e-05} +{"train_loss": 0.01888800784945488, "grad_norm": 0.058378174901008606, "global_step": 1671, "epoch": 15, "lr": 9.97546392756753e-05} +{"train_loss": 0.023380395025014877, "grad_norm": 0.048862554132938385, "global_step": 1672, "epoch": 15, "lr": 9.975434593485333e-05} +{"train_loss": 0.02072097361087799, "grad_norm": 0.12621228396892548, "global_step": 1673, "epoch": 15, "lr": 9.975405241921623e-05} +{"train_loss": 0.02111370861530304, "grad_norm": 0.06805764138698578, "global_step": 1674, "epoch": 15, "lr": 9.975375872876503e-05} +{"train_loss": 0.02107175812125206, "grad_norm": 0.044219549745321274, "global_step": 1675, "epoch": 15, "lr": 9.975346486350077e-05} +{"train_loss": 0.019605442881584167, "grad_norm": 0.04770579934120178, "global_step": 1676, "epoch": 15, "lr": 9.975317082342446e-05} +{"train_loss": 0.021225597709417343, "grad_norm": 0.16152428090572357, "global_step": 1677, "epoch": 15, "lr": 9.975287660853713e-05} +{"train_loss": 0.02691034972667694, "grad_norm": 0.09219447523355484, "global_step": 1678, "epoch": 15, "lr": 9.975258221883985e-05} +{"train_loss": 0.021809330210089684, "grad_norm": 0.09050553292036057, "global_step": 1679, "epoch": 15, "lr": 9.975228765433362e-05} +{"train_loss": 0.02044498361647129, "grad_norm": 0.09623038023710251, "global_step": 1680, "epoch": 15, "lr": 9.975199291501949e-05} +{"train_loss": 0.026838216930627823, "grad_norm": 0.12192563712596893, "global_step": 1681, "epoch": 15, "lr": 9.975169800089851e-05} +{"train_loss": 0.02188751846551895, "grad_norm": 0.05707237496972084, "global_step": 1682, "epoch": 15, "lr": 9.975140291197168e-05} +{"train_loss": 0.026098541915416718, "grad_norm": 0.05615580826997757, "global_step": 1683, "epoch": 15, "lr": 9.975110764824007e-05} +{"train_loss": 0.021554499864578247, "grad_norm": 0.13267211616039276, "global_step": 1684, "epoch": 15, "lr": 9.97508122097047e-05} +{"train_loss": 0.02405814453959465, "grad_norm": 0.05546044558286667, "global_step": 1685, "epoch": 15, "lr": 9.975051659636661e-05} +{"train_loss": 0.025516685098409653, "grad_norm": 0.1356498748064041, "global_step": 1686, "epoch": 15, "lr": 9.975022080822683e-05} +{"train_loss": 0.020764729008078575, "grad_norm": 0.05840647220611572, "global_step": 1687, "epoch": 15, "lr": 9.974992484528642e-05} +{"train_loss": 0.02227344736456871, "grad_norm": 0.05512261390686035, "global_step": 1688, "epoch": 15, "lr": 9.97496287075464e-05} +{"train_loss": 0.023556934669613838, "grad_norm": 0.14944222569465637, "global_step": 1689, "epoch": 15, "lr": 9.974933239500785e-05} +{"train_loss": 0.025059249252080917, "grad_norm": 0.03058004565536976, "global_step": 1690, "epoch": 15, "lr": 9.974903590767176e-05} +{"train_loss": 0.023948341608047485, "grad_norm": 0.09428051114082336, "global_step": 1691, "epoch": 15, "lr": 9.974873924553918e-05} +{"train_loss": 0.02051331102848053, "grad_norm": 0.04975339397788048, "global_step": 1692, "epoch": 15, "lr": 9.974844240861119e-05} +{"train_loss": 0.022296695038676262, "grad_norm": 0.08145460486412048, "global_step": 1693, "epoch": 15, "lr": 9.974814539688879e-05} +{"train_loss": 0.025677276775240898, "grad_norm": 0.0429205596446991, "global_step": 1694, "epoch": 15, "lr": 9.974784821037307e-05} +{"train_loss": 0.02230349306087449, "grad_norm": 0.07438081502914429, "global_step": 1695, "epoch": 15, "lr": 9.974755084906502e-05, "val_loss": 0.020646173506975174} +{"train_loss": 0.02187231183052063, "grad_norm": 0.03929497301578522, "global_step": 1696, "epoch": 16, "lr": 9.97472533129657e-05} +{"train_loss": 0.02149486541748047, "grad_norm": 0.056956883519887924, "global_step": 1697, "epoch": 16, "lr": 9.974695560207619e-05} +{"train_loss": 0.022265654057264328, "grad_norm": 0.10483744740486145, "global_step": 1698, "epoch": 16, "lr": 9.97466577163975e-05} +{"train_loss": 0.017957579344511032, "grad_norm": 0.18525481224060059, "global_step": 1699, "epoch": 16, "lr": 9.97463596559307e-05} +{"train_loss": 0.022447319701313972, "grad_norm": 0.15275755524635315, "global_step": 1700, "epoch": 16, "lr": 9.974606142067683e-05} +{"train_loss": 0.022380435839295387, "grad_norm": 0.1153365820646286, "global_step": 1701, "epoch": 16, "lr": 9.974576301063692e-05} +{"train_loss": 0.021161116659641266, "grad_norm": 0.12013653665781021, "global_step": 1702, "epoch": 16, "lr": 9.974546442581205e-05} +{"train_loss": 0.022015776485204697, "grad_norm": 0.09807229042053223, "global_step": 1703, "epoch": 16, "lr": 9.974516566620322e-05} +{"train_loss": 0.022547677159309387, "grad_norm": 0.035872895270586014, "global_step": 1704, "epoch": 16, "lr": 9.974486673181154e-05} +{"train_loss": 0.02085084840655327, "grad_norm": 0.09334485977888107, "global_step": 1705, "epoch": 16, "lr": 9.974456762263802e-05} +{"train_loss": 0.023328445851802826, "grad_norm": 0.03637378290295601, "global_step": 1706, "epoch": 16, "lr": 9.974426833868374e-05} +{"train_loss": 0.021483201533555984, "grad_norm": 0.1579562872648239, "global_step": 1707, "epoch": 16, "lr": 9.974396887994971e-05} +{"train_loss": 0.020363084971904755, "grad_norm": 0.06850403547286987, "global_step": 1708, "epoch": 16, "lr": 9.9743669246437e-05} +{"train_loss": 0.02265196293592453, "grad_norm": 0.06258965283632278, "global_step": 1709, "epoch": 16, "lr": 9.974336943814671e-05} +{"train_loss": 0.021007979288697243, "grad_norm": 0.20064237713813782, "global_step": 1710, "epoch": 16, "lr": 9.974306945507981e-05} +{"train_loss": 0.021142736077308655, "grad_norm": 0.03597503527998924, "global_step": 1711, "epoch": 16, "lr": 9.974276929723743e-05} +{"train_loss": 0.02377566322684288, "grad_norm": 0.0528489388525486, "global_step": 1712, "epoch": 16, "lr": 9.974246896462058e-05} +{"train_loss": 0.02202300913631916, "grad_norm": 0.12033210694789886, "global_step": 1713, "epoch": 16, "lr": 9.974216845723031e-05} +{"train_loss": 0.02136046066880226, "grad_norm": 0.05401701480150223, "global_step": 1714, "epoch": 16, "lr": 9.974186777506771e-05} +{"train_loss": 0.02236182987689972, "grad_norm": 0.05980764329433441, "global_step": 1715, "epoch": 16, "lr": 9.974156691813381e-05} +{"train_loss": 0.021381797268986702, "grad_norm": 0.10451337695121765, "global_step": 1716, "epoch": 16, "lr": 9.974126588642969e-05} +{"train_loss": 0.02154386043548584, "grad_norm": 0.06334168463945389, "global_step": 1717, "epoch": 16, "lr": 9.974096467995638e-05} +{"train_loss": 0.018703926354646683, "grad_norm": 0.12063025683164597, "global_step": 1718, "epoch": 16, "lr": 9.974066329871495e-05} +{"train_loss": 0.02405547723174095, "grad_norm": 0.045534897595644, "global_step": 1719, "epoch": 16, "lr": 9.974036174270647e-05} +{"train_loss": 0.021383989602327347, "grad_norm": 0.04540618136525154, "global_step": 1720, "epoch": 16, "lr": 9.974006001193199e-05} +{"train_loss": 0.021082919090986252, "grad_norm": 0.06324054300785065, "global_step": 1721, "epoch": 16, "lr": 9.973975810639255e-05} +{"train_loss": 0.021873172372579575, "grad_norm": 0.12761348485946655, "global_step": 1722, "epoch": 16, "lr": 9.973945602608925e-05} +{"train_loss": 0.02565154619514942, "grad_norm": 0.04105338826775551, "global_step": 1723, "epoch": 16, "lr": 9.973915377102314e-05} +{"train_loss": 0.020070713013410568, "grad_norm": 0.0411204919219017, "global_step": 1724, "epoch": 16, "lr": 9.973885134119527e-05} +{"train_loss": 0.02351175621151924, "grad_norm": 0.10901646316051483, "global_step": 1725, "epoch": 16, "lr": 9.973854873660669e-05} +{"train_loss": 0.02225406840443611, "grad_norm": 0.06471698731184006, "global_step": 1726, "epoch": 16, "lr": 9.97382459572585e-05} +{"train_loss": 0.023993946611881256, "grad_norm": 0.11781519651412964, "global_step": 1727, "epoch": 16, "lr": 9.973794300315173e-05} +{"train_loss": 0.023503176867961884, "grad_norm": 0.06333257257938385, "global_step": 1728, "epoch": 16, "lr": 9.973763987428748e-05} +{"train_loss": 0.023530490696430206, "grad_norm": 0.10472062975168228, "global_step": 1729, "epoch": 16, "lr": 9.973733657066677e-05} +{"train_loss": 0.022018786519765854, "grad_norm": 0.08093152195215225, "global_step": 1730, "epoch": 16, "lr": 9.97370330922907e-05} +{"train_loss": 0.023413527756929398, "grad_norm": 0.09933702647686005, "global_step": 1731, "epoch": 16, "lr": 9.973672943916033e-05} +{"train_loss": 0.020871154963970184, "grad_norm": 0.07894345372915268, "global_step": 1732, "epoch": 16, "lr": 9.973642561127671e-05} +{"train_loss": 0.021101705729961395, "grad_norm": 0.10651635378599167, "global_step": 1733, "epoch": 16, "lr": 9.973612160864094e-05} +{"train_loss": 0.02112102322280407, "grad_norm": 0.03141580522060394, "global_step": 1734, "epoch": 16, "lr": 9.973581743125405e-05} +{"train_loss": 0.02149692364037037, "grad_norm": 0.05492318421602249, "global_step": 1735, "epoch": 16, "lr": 9.973551307911712e-05} +{"train_loss": 0.024943159893155098, "grad_norm": 0.04303247109055519, "global_step": 1736, "epoch": 16, "lr": 9.973520855223126e-05} +{"train_loss": 0.0212409570813179, "grad_norm": 0.16535130143165588, "global_step": 1737, "epoch": 16, "lr": 9.973490385059748e-05} +{"train_loss": 0.020078666508197784, "grad_norm": 0.15892751514911652, "global_step": 1738, "epoch": 16, "lr": 9.973459897421688e-05} +{"train_loss": 0.024399414658546448, "grad_norm": 0.04304486885666847, "global_step": 1739, "epoch": 16, "lr": 9.973429392309054e-05} +{"train_loss": 0.022852696478366852, "grad_norm": 0.07663438469171524, "global_step": 1740, "epoch": 16, "lr": 9.97339886972195e-05} +{"train_loss": 0.02059296704828739, "grad_norm": 0.06693079322576523, "global_step": 1741, "epoch": 16, "lr": 9.973368329660487e-05} +{"train_loss": 0.023958340287208557, "grad_norm": 0.09035485237836838, "global_step": 1742, "epoch": 16, "lr": 9.97333777212477e-05} +{"train_loss": 0.023482684046030045, "grad_norm": 0.03884163871407509, "global_step": 1743, "epoch": 16, "lr": 9.973307197114907e-05} +{"train_loss": 0.0220283605158329, "grad_norm": 0.05495334044098854, "global_step": 1744, "epoch": 16, "lr": 9.973276604631006e-05} +{"train_loss": 0.022770045325160027, "grad_norm": 0.04991631954908371, "global_step": 1745, "epoch": 16, "lr": 9.973245994673173e-05} +{"train_loss": 0.025886541232466698, "grad_norm": 0.03908400982618332, "global_step": 1746, "epoch": 16, "lr": 9.973215367241515e-05} +{"train_loss": 0.023023292422294617, "grad_norm": 0.04414815083146095, "global_step": 1747, "epoch": 16, "lr": 9.973184722336142e-05} +{"train_loss": 0.02330334112048149, "grad_norm": 0.03704676777124405, "global_step": 1748, "epoch": 16, "lr": 9.973154059957162e-05} +{"train_loss": 0.022581597790122032, "grad_norm": 0.0890054926276207, "global_step": 1749, "epoch": 16, "lr": 9.973123380104681e-05} +{"train_loss": 0.021285951137542725, "grad_norm": 0.0846472978591919, "global_step": 1750, "epoch": 16, "lr": 9.973092682778807e-05} +{"train_loss": 0.020145956426858902, "grad_norm": 0.09215917438268661, "global_step": 1751, "epoch": 16, "lr": 9.973061967979647e-05} +{"train_loss": 0.020481830462813377, "grad_norm": 0.06896188110113144, "global_step": 1752, "epoch": 16, "lr": 9.97303123570731e-05} +{"train_loss": 0.023821845650672913, "grad_norm": 0.0599893182516098, "global_step": 1753, "epoch": 16, "lr": 9.973000485961905e-05} +{"train_loss": 0.02299411967396736, "grad_norm": 0.1279447227716446, "global_step": 1754, "epoch": 16, "lr": 9.97296971874354e-05} +{"train_loss": 0.022844212129712105, "grad_norm": 0.0541185587644577, "global_step": 1755, "epoch": 16, "lr": 9.972938934052321e-05} +{"train_loss": 0.02440490946173668, "grad_norm": 0.048805609345436096, "global_step": 1756, "epoch": 16, "lr": 9.972908131888356e-05} +{"train_loss": 0.021382838487625122, "grad_norm": 0.05689244344830513, "global_step": 1757, "epoch": 16, "lr": 9.972877312251756e-05} +{"train_loss": 0.02260161191225052, "grad_norm": 0.07742469012737274, "global_step": 1758, "epoch": 16, "lr": 9.972846475142626e-05} +{"train_loss": 0.02170887216925621, "grad_norm": 0.06705281883478165, "global_step": 1759, "epoch": 16, "lr": 9.972815620561077e-05} +{"train_loss": 0.024431077763438225, "grad_norm": 0.06737358868122101, "global_step": 1760, "epoch": 16, "lr": 9.972784748507218e-05} +{"train_loss": 0.023006128147244453, "grad_norm": 0.06550601124763489, "global_step": 1761, "epoch": 16, "lr": 9.972753858981155e-05} +{"train_loss": 0.025268863886594772, "grad_norm": 0.07564043253660202, "global_step": 1762, "epoch": 16, "lr": 9.972722951982998e-05} +{"train_loss": 0.019099581986665726, "grad_norm": 0.08493967354297638, "global_step": 1763, "epoch": 16, "lr": 9.972692027512854e-05} +{"train_loss": 0.020754968747496605, "grad_norm": 0.08089534193277359, "global_step": 1764, "epoch": 16, "lr": 9.972661085570834e-05} +{"train_loss": 0.02307041548192501, "grad_norm": 0.10105887800455093, "global_step": 1765, "epoch": 16, "lr": 9.972630126157045e-05} +{"train_loss": 0.02660854160785675, "grad_norm": 0.055236879736185074, "global_step": 1766, "epoch": 16, "lr": 9.972599149271596e-05} +{"train_loss": 0.021754760295152664, "grad_norm": 0.07759381830692291, "global_step": 1767, "epoch": 16, "lr": 9.972568154914595e-05} +{"train_loss": 0.02266985923051834, "grad_norm": 0.06112667918205261, "global_step": 1768, "epoch": 16, "lr": 9.972537143086153e-05} +{"train_loss": 0.020156992599368095, "grad_norm": 0.037956319749355316, "global_step": 1769, "epoch": 16, "lr": 9.97250611378638e-05} +{"train_loss": 0.021918606013059616, "grad_norm": 0.10987279564142227, "global_step": 1770, "epoch": 16, "lr": 9.972475067015381e-05} +{"train_loss": 0.020836368203163147, "grad_norm": 0.05242966115474701, "global_step": 1771, "epoch": 16, "lr": 9.972444002773267e-05} +{"train_loss": 0.021935291588306427, "grad_norm": 0.06584233045578003, "global_step": 1772, "epoch": 16, "lr": 9.972412921060148e-05} +{"train_loss": 0.022637616842985153, "grad_norm": 0.05210336297750473, "global_step": 1773, "epoch": 16, "lr": 9.972381821876131e-05} +{"train_loss": 0.021776746958494186, "grad_norm": 0.059031274169683456, "global_step": 1774, "epoch": 16, "lr": 9.972350705221328e-05} +{"train_loss": 0.020384525880217552, "grad_norm": 0.050695326179265976, "global_step": 1775, "epoch": 16, "lr": 9.972319571095847e-05} +{"train_loss": 0.025725258514285088, "grad_norm": 0.051236867904663086, "global_step": 1776, "epoch": 16, "lr": 9.972288419499797e-05} +{"train_loss": 0.022394156083464622, "grad_norm": 0.058372996747493744, "global_step": 1777, "epoch": 16, "lr": 9.972257250433287e-05} +{"train_loss": 0.02209731936454773, "grad_norm": 0.05912986025214195, "global_step": 1778, "epoch": 16, "lr": 9.972226063896429e-05} +{"train_loss": 0.023456895723938942, "grad_norm": 0.04989217221736908, "global_step": 1779, "epoch": 16, "lr": 9.972194859889331e-05} +{"train_loss": 0.018892301246523857, "grad_norm": 0.043755337595939636, "global_step": 1780, "epoch": 16, "lr": 9.972163638412102e-05} +{"train_loss": 0.020670518279075623, "grad_norm": 0.05078853294253349, "global_step": 1781, "epoch": 16, "lr": 9.972132399464854e-05} +{"train_loss": 0.025808649137616158, "grad_norm": 0.08068184554576874, "global_step": 1782, "epoch": 16, "lr": 9.972101143047692e-05} +{"train_loss": 0.02209838293492794, "grad_norm": 0.03502042219042778, "global_step": 1783, "epoch": 16, "lr": 9.97206986916073e-05} +{"train_loss": 0.02179524675011635, "grad_norm": 0.07465074211359024, "global_step": 1784, "epoch": 16, "lr": 9.972038577804077e-05} +{"train_loss": 0.019466858357191086, "grad_norm": 0.07564152777194977, "global_step": 1785, "epoch": 16, "lr": 9.972007268977843e-05} +{"train_loss": 0.024392709136009216, "grad_norm": 0.0655038133263588, "global_step": 1786, "epoch": 16, "lr": 9.971975942682138e-05} +{"train_loss": 0.02231510728597641, "grad_norm": 0.03310448303818703, "global_step": 1787, "epoch": 16, "lr": 9.971944598917071e-05} +{"train_loss": 0.020304016768932343, "grad_norm": 0.061095621436834335, "global_step": 1788, "epoch": 16, "lr": 9.971913237682753e-05} +{"train_loss": 0.02237718179821968, "grad_norm": 0.1185699850320816, "global_step": 1789, "epoch": 16, "lr": 9.971881858979295e-05} +{"train_loss": 0.01714159920811653, "grad_norm": 0.041854117065668106, "global_step": 1790, "epoch": 16, "lr": 9.971850462806806e-05} +{"train_loss": 0.021739374846220016, "grad_norm": 0.044307295233011246, "global_step": 1791, "epoch": 16, "lr": 9.971819049165397e-05} +{"train_loss": 0.02250552922487259, "grad_norm": 0.08169440180063248, "global_step": 1792, "epoch": 16, "lr": 9.971787618055177e-05} +{"train_loss": 0.02050567977130413, "grad_norm": 0.0536934956908226, "global_step": 1793, "epoch": 16, "lr": 9.971756169476257e-05} +{"train_loss": 0.021451415494084358, "grad_norm": 0.09309624880552292, "global_step": 1794, "epoch": 16, "lr": 9.971724703428748e-05} +{"train_loss": 0.021125178784132004, "grad_norm": 0.11331010609865189, "global_step": 1795, "epoch": 16, "lr": 9.971693219912763e-05} +{"train_loss": 0.02567053586244583, "grad_norm": 0.045402269810438156, "global_step": 1796, "epoch": 16, "lr": 9.971661718928409e-05} +{"train_loss": 0.024975869804620743, "grad_norm": 0.07279876619577408, "global_step": 1797, "epoch": 16, "lr": 9.971630200475797e-05} +{"train_loss": 0.020307490602135658, "grad_norm": 0.03724568337202072, "global_step": 1798, "epoch": 16, "lr": 9.971598664555039e-05} +{"train_loss": 0.020597757771611214, "grad_norm": 0.063977912068367, "global_step": 1799, "epoch": 16, "lr": 9.971567111166246e-05} +{"train_loss": 0.021311186254024506, "grad_norm": 0.07447710633277893, "global_step": 1800, "epoch": 16, "lr": 9.971535540309526e-05} +{"train_loss": 0.022177924226336885, "grad_norm": 0.10456109046936035, "global_step": 1801, "epoch": 16, "lr": 9.971503951984995e-05, "val_loss": 0.020486295223236084} +{"train_loss": 0.025918418541550636, "grad_norm": 0.0686689242720604, "global_step": 1802, "epoch": 17, "lr": 9.97147234619276e-05} +{"train_loss": 0.021841347217559814, "grad_norm": 0.050882160663604736, "global_step": 1803, "epoch": 17, "lr": 9.971440722932933e-05} +{"train_loss": 0.02298804745078087, "grad_norm": 0.09718897938728333, "global_step": 1804, "epoch": 17, "lr": 9.971409082205624e-05} +{"train_loss": 0.020933479070663452, "grad_norm": 0.040464483201503754, "global_step": 1805, "epoch": 17, "lr": 9.971377424010948e-05} +{"train_loss": 0.022767415270209312, "grad_norm": 0.12208765745162964, "global_step": 1806, "epoch": 17, "lr": 9.971345748349014e-05} +{"train_loss": 0.022108741104602814, "grad_norm": 0.10942556709051132, "global_step": 1807, "epoch": 17, "lr": 9.971314055219931e-05} +{"train_loss": 0.02123538963496685, "grad_norm": 0.042553629726171494, "global_step": 1808, "epoch": 17, "lr": 9.971282344623814e-05} +{"train_loss": 0.022329118102788925, "grad_norm": 0.0953318178653717, "global_step": 1809, "epoch": 17, "lr": 9.971250616560772e-05} +{"train_loss": 0.022143233567476273, "grad_norm": 0.11804313957691193, "global_step": 1810, "epoch": 17, "lr": 9.971218871030916e-05} +{"train_loss": 0.024151219055056572, "grad_norm": 0.11771628260612488, "global_step": 1811, "epoch": 17, "lr": 9.97118710803436e-05} +{"train_loss": 0.018231578171253204, "grad_norm": 0.049495112150907516, "global_step": 1812, "epoch": 17, "lr": 9.971155327571216e-05} +{"train_loss": 0.023099130019545555, "grad_norm": 0.041329748928546906, "global_step": 1813, "epoch": 17, "lr": 9.971123529641592e-05} +{"train_loss": 0.022806642577052116, "grad_norm": 0.07098706066608429, "global_step": 1814, "epoch": 17, "lr": 9.971091714245602e-05} +{"train_loss": 0.024136334657669067, "grad_norm": 0.05372270941734314, "global_step": 1815, "epoch": 17, "lr": 9.971059881383358e-05} +{"train_loss": 0.021311068907380104, "grad_norm": 0.09950442612171173, "global_step": 1816, "epoch": 17, "lr": 9.971028031054972e-05} +{"train_loss": 0.021050451323390007, "grad_norm": 0.037316884845495224, "global_step": 1817, "epoch": 17, "lr": 9.970996163260554e-05} +{"train_loss": 0.022537071257829666, "grad_norm": 0.04417697712779045, "global_step": 1818, "epoch": 17, "lr": 9.970964278000219e-05} +{"train_loss": 0.02155616134405136, "grad_norm": 0.03365396708250046, "global_step": 1819, "epoch": 17, "lr": 9.970932375274075e-05} +{"train_loss": 0.022584721446037292, "grad_norm": 0.11579025536775589, "global_step": 1820, "epoch": 17, "lr": 9.970900455082238e-05} +{"train_loss": 0.021835409104824066, "grad_norm": 0.10837716609239578, "global_step": 1821, "epoch": 17, "lr": 9.970868517424818e-05} +{"train_loss": 0.022177565842866898, "grad_norm": 0.06435644626617432, "global_step": 1822, "epoch": 17, "lr": 9.970836562301929e-05} +{"train_loss": 0.019898604601621628, "grad_norm": 0.125503271818161, "global_step": 1823, "epoch": 17, "lr": 9.97080458971368e-05} +{"train_loss": 0.01937314309179783, "grad_norm": 0.12337049841880798, "global_step": 1824, "epoch": 17, "lr": 9.970772599660188e-05} +{"train_loss": 0.020987965166568756, "grad_norm": 0.10636449605226517, "global_step": 1825, "epoch": 17, "lr": 9.970740592141561e-05} +{"train_loss": 0.020995059981942177, "grad_norm": 0.04589267820119858, "global_step": 1826, "epoch": 17, "lr": 9.970708567157914e-05} +{"train_loss": 0.01775244250893593, "grad_norm": 0.08672090619802475, "global_step": 1827, "epoch": 17, "lr": 9.97067652470936e-05} +{"train_loss": 0.024455644190311432, "grad_norm": 0.05815039575099945, "global_step": 1828, "epoch": 17, "lr": 9.970644464796009e-05} +{"train_loss": 0.022070154547691345, "grad_norm": 0.08882535994052887, "global_step": 1829, "epoch": 17, "lr": 9.970612387417974e-05} +{"train_loss": 0.021723441779613495, "grad_norm": 0.040552910417318344, "global_step": 1830, "epoch": 17, "lr": 9.970580292575368e-05} +{"train_loss": 0.023185458034276962, "grad_norm": 0.04162833094596863, "global_step": 1831, "epoch": 17, "lr": 9.970548180268307e-05} +{"train_loss": 0.023185912519693375, "grad_norm": 0.03190917521715164, "global_step": 1832, "epoch": 17, "lr": 9.970516050496901e-05} +{"train_loss": 0.023158304393291473, "grad_norm": 0.06290273368358612, "global_step": 1833, "epoch": 17, "lr": 9.970483903261263e-05} +{"train_loss": 0.023068644106388092, "grad_norm": 0.0778404250741005, "global_step": 1834, "epoch": 17, "lr": 9.970451738561506e-05} +{"train_loss": 0.025335095822811127, "grad_norm": 0.055149249732494354, "global_step": 1835, "epoch": 17, "lr": 9.970419556397743e-05} +{"train_loss": 0.01882592961192131, "grad_norm": 0.02816467359662056, "global_step": 1836, "epoch": 17, "lr": 9.970387356770087e-05} +{"train_loss": 0.023268429562449455, "grad_norm": 0.08127864450216293, "global_step": 1837, "epoch": 17, "lr": 9.970355139678652e-05} +{"train_loss": 0.022541392594575882, "grad_norm": 0.08560667932033539, "global_step": 1838, "epoch": 17, "lr": 9.97032290512355e-05} +{"train_loss": 0.02309088595211506, "grad_norm": 0.13724198937416077, "global_step": 1839, "epoch": 17, "lr": 9.970290653104896e-05} +{"train_loss": 0.023211747407913208, "grad_norm": 0.09831954538822174, "global_step": 1840, "epoch": 17, "lr": 9.970258383622801e-05} +{"train_loss": 0.020302873104810715, "grad_norm": 0.08225364238023758, "global_step": 1841, "epoch": 17, "lr": 9.97022609667738e-05} +{"train_loss": 0.02444005385041237, "grad_norm": 0.05639025568962097, "global_step": 1842, "epoch": 17, "lr": 9.970193792268746e-05} +{"train_loss": 0.020454268902540207, "grad_norm": 0.03958814963698387, "global_step": 1843, "epoch": 17, "lr": 9.970161470397013e-05} +{"train_loss": 0.021547574549913406, "grad_norm": 0.04524987190961838, "global_step": 1844, "epoch": 17, "lr": 9.970129131062294e-05} +{"train_loss": 0.024111438542604446, "grad_norm": 0.08597032725811005, "global_step": 1845, "epoch": 17, "lr": 9.970096774264702e-05} +{"train_loss": 0.024119097739458084, "grad_norm": 0.037300173193216324, "global_step": 1846, "epoch": 17, "lr": 9.97006440000435e-05} +{"train_loss": 0.02286429889500141, "grad_norm": 0.09012537449598312, "global_step": 1847, "epoch": 17, "lr": 9.970032008281356e-05} +{"train_loss": 0.02118973806500435, "grad_norm": 0.061278119683265686, "global_step": 1848, "epoch": 17, "lr": 9.96999959909583e-05} +{"train_loss": 0.026655029505491257, "grad_norm": 0.1100383773446083, "global_step": 1849, "epoch": 17, "lr": 9.969967172447885e-05} +{"train_loss": 0.02441907301545143, "grad_norm": 0.08599536865949631, "global_step": 1850, "epoch": 17, "lr": 9.969934728337638e-05} +{"train_loss": 0.023052146658301353, "grad_norm": 0.08245163410902023, "global_step": 1851, "epoch": 17, "lr": 9.969902266765203e-05} +{"train_loss": 0.021438896656036377, "grad_norm": 0.08318032324314117, "global_step": 1852, "epoch": 17, "lr": 9.969869787730691e-05} +{"train_loss": 0.022622521966695786, "grad_norm": 0.08455344289541245, "global_step": 1853, "epoch": 17, "lr": 9.969837291234218e-05} +{"train_loss": 0.02183942124247551, "grad_norm": 0.031585391610860825, "global_step": 1854, "epoch": 17, "lr": 9.9698047772759e-05} +{"train_loss": 0.021961573511362076, "grad_norm": 0.06913965195417404, "global_step": 1855, "epoch": 17, "lr": 9.969772245855846e-05} +{"train_loss": 0.021612033247947693, "grad_norm": 0.08554543554782867, "global_step": 1856, "epoch": 17, "lr": 9.969739696974176e-05} +{"train_loss": 0.02077311836183071, "grad_norm": 0.07330312579870224, "global_step": 1857, "epoch": 17, "lr": 9.969707130631002e-05} +{"train_loss": 0.021797489374876022, "grad_norm": 0.10527852177619934, "global_step": 1858, "epoch": 17, "lr": 9.969674546826438e-05} +{"train_loss": 0.019941262900829315, "grad_norm": 0.16585376858711243, "global_step": 1859, "epoch": 17, "lr": 9.9696419455606e-05} +{"train_loss": 0.024439029395580292, "grad_norm": 0.1293739229440689, "global_step": 1860, "epoch": 17, "lr": 9.9696093268336e-05} +{"train_loss": 0.021052565425634384, "grad_norm": 0.12425214052200317, "global_step": 1861, "epoch": 17, "lr": 9.969576690645554e-05} +{"train_loss": 0.022619951516389847, "grad_norm": 0.08919090777635574, "global_step": 1862, "epoch": 17, "lr": 9.969544036996577e-05} +{"train_loss": 0.02500748820602894, "grad_norm": 0.1378224641084671, "global_step": 1863, "epoch": 17, "lr": 9.969511365886784e-05} +{"train_loss": 0.021777058020234108, "grad_norm": 0.041832614690065384, "global_step": 1864, "epoch": 17, "lr": 9.969478677316288e-05} +{"train_loss": 0.02121911756694317, "grad_norm": 0.13171106576919556, "global_step": 1865, "epoch": 17, "lr": 9.969445971285205e-05} +{"train_loss": 0.021295197308063507, "grad_norm": 0.04082922264933586, "global_step": 1866, "epoch": 17, "lr": 9.969413247793652e-05} +{"train_loss": 0.024128038436174393, "grad_norm": 0.09690313041210175, "global_step": 1867, "epoch": 17, "lr": 9.969380506841742e-05} +{"train_loss": 0.0219167061150074, "grad_norm": 0.07951917499303818, "global_step": 1868, "epoch": 17, "lr": 9.969347748429587e-05} +{"train_loss": 0.020865367725491524, "grad_norm": 0.113462433218956, "global_step": 1869, "epoch": 17, "lr": 9.969314972557308e-05} +{"train_loss": 0.01850615069270134, "grad_norm": 0.029103204607963562, "global_step": 1870, "epoch": 17, "lr": 9.969282179225016e-05} +{"train_loss": 0.023265443742275238, "grad_norm": 0.049174871295690536, "global_step": 1871, "epoch": 17, "lr": 9.969249368432827e-05} +{"train_loss": 0.02034757286310196, "grad_norm": 0.08295225352048874, "global_step": 1872, "epoch": 17, "lr": 9.969216540180859e-05} +{"train_loss": 0.02137509360909462, "grad_norm": 0.13131582736968994, "global_step": 1873, "epoch": 17, "lr": 9.969183694469223e-05} +{"train_loss": 0.019690662622451782, "grad_norm": 0.10442657768726349, "global_step": 1874, "epoch": 17, "lr": 9.969150831298038e-05} +{"train_loss": 0.021913941949605942, "grad_norm": 0.15815526247024536, "global_step": 1875, "epoch": 17, "lr": 9.969117950667416e-05} +{"train_loss": 0.024302076548337936, "grad_norm": 0.0861176922917366, "global_step": 1876, "epoch": 17, "lr": 9.969085052577476e-05} +{"train_loss": 0.022622626274824142, "grad_norm": 0.12229093909263611, "global_step": 1877, "epoch": 17, "lr": 9.969052137028332e-05} +{"train_loss": 0.023186292499303818, "grad_norm": 0.03943875432014465, "global_step": 1878, "epoch": 17, "lr": 9.9690192040201e-05} +{"train_loss": 0.02103331685066223, "grad_norm": 0.06046835333108902, "global_step": 1879, "epoch": 17, "lr": 9.968986253552895e-05} +{"train_loss": 0.025058865547180176, "grad_norm": 0.058917246758937836, "global_step": 1880, "epoch": 17, "lr": 9.968953285626835e-05} +{"train_loss": 0.02400967851281166, "grad_norm": 0.06197478249669075, "global_step": 1881, "epoch": 17, "lr": 9.968920300242031e-05} +{"train_loss": 0.023119159042835236, "grad_norm": 0.04697472229599953, "global_step": 1882, "epoch": 17, "lr": 9.968887297398604e-05} +{"train_loss": 0.022020390257239342, "grad_norm": 0.03745841607451439, "global_step": 1883, "epoch": 17, "lr": 9.968854277096668e-05} +{"train_loss": 0.015581540763378143, "grad_norm": 0.056942168623209, "global_step": 1884, "epoch": 17, "lr": 9.968821239336338e-05} +{"train_loss": 0.02089151181280613, "grad_norm": 0.05263075605034828, "global_step": 1885, "epoch": 17, "lr": 9.968788184117732e-05} +{"train_loss": 0.02092808485031128, "grad_norm": 0.04482723027467728, "global_step": 1886, "epoch": 17, "lr": 9.968755111440964e-05} +{"train_loss": 0.02319353073835373, "grad_norm": 0.07785512506961823, "global_step": 1887, "epoch": 17, "lr": 9.968722021306151e-05} +{"train_loss": 0.025872578844428062, "grad_norm": 0.05205344408750534, "global_step": 1888, "epoch": 17, "lr": 9.968688913713413e-05} +{"train_loss": 0.02219298854470253, "grad_norm": 0.07499632239341736, "global_step": 1889, "epoch": 17, "lr": 9.96865578866286e-05} +{"train_loss": 0.022497527301311493, "grad_norm": 0.046036653220653534, "global_step": 1890, "epoch": 17, "lr": 9.968622646154611e-05} +{"train_loss": 0.021584369242191315, "grad_norm": 0.12409178167581558, "global_step": 1891, "epoch": 17, "lr": 9.968589486188784e-05} +{"train_loss": 0.02075931988656521, "grad_norm": 0.03900504857301712, "global_step": 1892, "epoch": 17, "lr": 9.968556308765496e-05} +{"train_loss": 0.02308804914355278, "grad_norm": 0.08204575628042221, "global_step": 1893, "epoch": 17, "lr": 9.968523113884859e-05} +{"train_loss": 0.02355598658323288, "grad_norm": 0.05276860296726227, "global_step": 1894, "epoch": 17, "lr": 9.968489901546994e-05} +{"train_loss": 0.019819527864456177, "grad_norm": 0.06643117964267731, "global_step": 1895, "epoch": 17, "lr": 9.968456671752015e-05} +{"train_loss": 0.021547239273786545, "grad_norm": 0.12415481358766556, "global_step": 1896, "epoch": 17, "lr": 9.968423424500042e-05} +{"train_loss": 0.02421262115240097, "grad_norm": 0.04486516863107681, "global_step": 1897, "epoch": 17, "lr": 9.968390159791188e-05} +{"train_loss": 0.020267462357878685, "grad_norm": 0.055597878992557526, "global_step": 1898, "epoch": 17, "lr": 9.968356877625574e-05} +{"train_loss": 0.018884645774960518, "grad_norm": 0.046544723212718964, "global_step": 1899, "epoch": 17, "lr": 9.968323578003311e-05} +{"train_loss": 0.020306240767240524, "grad_norm": 0.06411351263523102, "global_step": 1900, "epoch": 17, "lr": 9.968290260924522e-05} +{"train_loss": 0.019571322947740555, "grad_norm": 0.10447544604539871, "global_step": 1901, "epoch": 17, "lr": 9.968256926389321e-05} +{"train_loss": 0.019686050713062286, "grad_norm": 0.05385204032063484, "global_step": 1902, "epoch": 17, "lr": 9.968223574397825e-05} +{"train_loss": 0.02255767583847046, "grad_norm": 0.04023967310786247, "global_step": 1903, "epoch": 17, "lr": 9.968190204950154e-05} +{"train_loss": 0.021880727261304855, "grad_norm": 0.06313126534223557, "global_step": 1904, "epoch": 17, "lr": 9.968156818046424e-05} +{"train_loss": 0.021361177787184715, "grad_norm": 0.0711749717593193, "global_step": 1905, "epoch": 17, "lr": 9.968123413686748e-05} +{"train_loss": 0.01934896409511566, "grad_norm": 0.05317383259534836, "global_step": 1906, "epoch": 17, "lr": 9.96808999187125e-05} +{"train_loss": 0.022032968945941835, "grad_norm": 0.07188470661640167, "global_step": 1907, "epoch": 17, "lr": 9.968056552600043e-05, "val_loss": 0.020268401131033897} +{"train_loss": 0.02326115593314171, "grad_norm": 0.04959192872047424, "global_step": 1908, "epoch": 18, "lr": 9.968023095873244e-05} +{"train_loss": 0.0208063293248415, "grad_norm": 0.03944604471325874, "global_step": 1909, "epoch": 18, "lr": 9.967989621690976e-05} +{"train_loss": 0.022186219692230225, "grad_norm": 0.04638335481286049, "global_step": 1910, "epoch": 18, "lr": 9.967956130053351e-05} +{"train_loss": 0.02185232937335968, "grad_norm": 0.05225805938243866, "global_step": 1911, "epoch": 18, "lr": 9.967922620960489e-05} +{"train_loss": 0.022426757961511612, "grad_norm": 0.10890225321054459, "global_step": 1912, "epoch": 18, "lr": 9.967889094412507e-05} +{"train_loss": 0.025730464607477188, "grad_norm": 0.15223953127861023, "global_step": 1913, "epoch": 18, "lr": 9.967855550409524e-05} +{"train_loss": 0.021691545844078064, "grad_norm": 0.04506572335958481, "global_step": 1914, "epoch": 18, "lr": 9.967821988951655e-05} +{"train_loss": 0.021898215636610985, "grad_norm": 0.06618800759315491, "global_step": 1915, "epoch": 18, "lr": 9.967788410039023e-05} +{"train_loss": 0.0233352929353714, "grad_norm": 0.05911288410425186, "global_step": 1916, "epoch": 18, "lr": 9.967754813671741e-05} +{"train_loss": 0.020888468250632286, "grad_norm": 0.05968122184276581, "global_step": 1917, "epoch": 18, "lr": 9.96772119984993e-05} +{"train_loss": 0.02433761954307556, "grad_norm": 0.09984349459409714, "global_step": 1918, "epoch": 18, "lr": 9.967687568573707e-05} +{"train_loss": 0.021048689261078835, "grad_norm": 0.05658281221985817, "global_step": 1919, "epoch": 18, "lr": 9.96765391984319e-05} +{"train_loss": 0.019343825057148933, "grad_norm": 0.04540446773171425, "global_step": 1920, "epoch": 18, "lr": 9.967620253658496e-05} +{"train_loss": 0.02366894669830799, "grad_norm": 0.05962678790092468, "global_step": 1921, "epoch": 18, "lr": 9.967586570019746e-05} +{"train_loss": 0.02261967584490776, "grad_norm": 0.03922530636191368, "global_step": 1922, "epoch": 18, "lr": 9.967552868927058e-05} +{"train_loss": 0.021896393969655037, "grad_norm": 0.03410005196928978, "global_step": 1923, "epoch": 18, "lr": 9.967519150380549e-05} +{"train_loss": 0.01972593367099762, "grad_norm": 0.08227977156639099, "global_step": 1924, "epoch": 18, "lr": 9.967485414380337e-05} +{"train_loss": 0.021041320636868477, "grad_norm": 0.037373751401901245, "global_step": 1925, "epoch": 18, "lr": 9.967451660926544e-05} +{"train_loss": 0.0242328904569149, "grad_norm": 0.06360060721635818, "global_step": 1926, "epoch": 18, "lr": 9.967417890019283e-05} +{"train_loss": 0.01939113810658455, "grad_norm": 0.05457102879881859, "global_step": 1927, "epoch": 18, "lr": 9.967384101658678e-05} +{"train_loss": 0.021373819559812546, "grad_norm": 0.07725216448307037, "global_step": 1928, "epoch": 18, "lr": 9.967350295844846e-05} +{"train_loss": 0.019054891541600227, "grad_norm": 0.024011077359318733, "global_step": 1929, "epoch": 18, "lr": 9.967316472577903e-05} +{"train_loss": 0.020924564450979233, "grad_norm": 0.07051430642604828, "global_step": 1930, "epoch": 18, "lr": 9.967282631857973e-05} +{"train_loss": 0.022968020290136337, "grad_norm": 0.061986394226551056, "global_step": 1931, "epoch": 18, "lr": 9.96724877368517e-05} +{"train_loss": 0.019428309053182602, "grad_norm": 0.07147131115198135, "global_step": 1932, "epoch": 18, "lr": 9.967214898059615e-05} +{"train_loss": 0.02459000237286091, "grad_norm": 0.031146157532930374, "global_step": 1933, "epoch": 18, "lr": 9.967181004981427e-05} +{"train_loss": 0.023316223174333572, "grad_norm": 0.1990276575088501, "global_step": 1934, "epoch": 18, "lr": 9.967147094450726e-05} +{"train_loss": 0.022708628326654434, "grad_norm": 0.03420382738113403, "global_step": 1935, "epoch": 18, "lr": 9.96711316646763e-05} +{"train_loss": 0.020937005057930946, "grad_norm": 0.038217540830373764, "global_step": 1936, "epoch": 18, "lr": 9.967079221032258e-05} +{"train_loss": 0.023344870656728745, "grad_norm": 0.08707457035779953, "global_step": 1937, "epoch": 18, "lr": 9.967045258144731e-05} +{"train_loss": 0.02272806316614151, "grad_norm": 0.0610225610435009, "global_step": 1938, "epoch": 18, "lr": 9.967011277805165e-05} +{"train_loss": 0.021425839513540268, "grad_norm": 0.0359988808631897, "global_step": 1939, "epoch": 18, "lr": 9.966977280013683e-05} +{"train_loss": 0.02552022598683834, "grad_norm": 0.07291442155838013, "global_step": 1940, "epoch": 18, "lr": 9.966943264770404e-05} +{"train_loss": 0.02095053344964981, "grad_norm": 0.0406835563480854, "global_step": 1941, "epoch": 18, "lr": 9.966909232075443e-05} +{"train_loss": 0.02245776169002056, "grad_norm": 0.03837968409061432, "global_step": 1942, "epoch": 18, "lr": 9.966875181928925e-05} +{"train_loss": 0.0250646211206913, "grad_norm": 0.03894219920039177, "global_step": 1943, "epoch": 18, "lr": 9.966841114330969e-05} +{"train_loss": 0.02208002284169197, "grad_norm": 0.061887335032224655, "global_step": 1944, "epoch": 18, "lr": 9.96680702928169e-05} +{"train_loss": 0.017460878938436508, "grad_norm": 0.07089141011238098, "global_step": 1945, "epoch": 18, "lr": 9.966772926781213e-05} +{"train_loss": 0.022324873134493828, "grad_norm": 0.06516076624393463, "global_step": 1946, "epoch": 18, "lr": 9.966738806829658e-05} +{"train_loss": 0.020233094692230225, "grad_norm": 0.0685393288731575, "global_step": 1947, "epoch": 18, "lr": 9.966704669427141e-05} +{"train_loss": 0.022437293082475662, "grad_norm": 0.08299222588539124, "global_step": 1948, "epoch": 18, "lr": 9.966670514573783e-05} +{"train_loss": 0.017703188583254814, "grad_norm": 0.07995712012052536, "global_step": 1949, "epoch": 18, "lr": 9.966636342269705e-05} +{"train_loss": 0.022953655570745468, "grad_norm": 0.1131051778793335, "global_step": 1950, "epoch": 18, "lr": 9.966602152515028e-05} +{"train_loss": 0.02026313915848732, "grad_norm": 0.08025684952735901, "global_step": 1951, "epoch": 18, "lr": 9.96656794530987e-05} +{"train_loss": 0.021251913160085678, "grad_norm": 0.0709015280008316, "global_step": 1952, "epoch": 18, "lr": 9.966533720654352e-05} +{"train_loss": 0.02164662629365921, "grad_norm": 0.03991319239139557, "global_step": 1953, "epoch": 18, "lr": 9.966499478548595e-05} +{"train_loss": 0.020371418446302414, "grad_norm": 0.0509888231754303, "global_step": 1954, "epoch": 18, "lr": 9.966465218992717e-05} +{"train_loss": 0.022535299882292747, "grad_norm": 0.03909047320485115, "global_step": 1955, "epoch": 18, "lr": 9.966430941986842e-05} +{"train_loss": 0.028065277263522148, "grad_norm": 0.14333251118659973, "global_step": 1956, "epoch": 18, "lr": 9.966396647531087e-05} +{"train_loss": 0.023086121305823326, "grad_norm": 0.06320124119520187, "global_step": 1957, "epoch": 18, "lr": 9.966362335625576e-05} +{"train_loss": 0.02613040804862976, "grad_norm": 0.15327316522598267, "global_step": 1958, "epoch": 18, "lr": 9.966328006270427e-05} +{"train_loss": 0.022199831902980804, "grad_norm": 0.07136747241020203, "global_step": 1959, "epoch": 18, "lr": 9.966293659465759e-05} +{"train_loss": 0.02345997467637062, "grad_norm": 0.032342810183763504, "global_step": 1960, "epoch": 18, "lr": 9.966259295211697e-05} +{"train_loss": 0.020291902124881744, "grad_norm": 0.0586661733686924, "global_step": 1961, "epoch": 18, "lr": 9.96622491350836e-05} +{"train_loss": 0.021128453314304352, "grad_norm": 0.08590071648359299, "global_step": 1962, "epoch": 18, "lr": 9.966190514355866e-05} +{"train_loss": 0.022996412590146065, "grad_norm": 0.1252441704273224, "global_step": 1963, "epoch": 18, "lr": 9.966156097754341e-05} +{"train_loss": 0.018812377005815506, "grad_norm": 0.055004753172397614, "global_step": 1964, "epoch": 18, "lr": 9.966121663703902e-05} +{"train_loss": 0.021978184580802917, "grad_norm": 0.06638254970312119, "global_step": 1965, "epoch": 18, "lr": 9.966087212204672e-05} +{"train_loss": 0.020797213539481163, "grad_norm": 0.11146754771471024, "global_step": 1966, "epoch": 18, "lr": 9.96605274325677e-05} +{"train_loss": 0.0190857145935297, "grad_norm": 0.05375486612319946, "global_step": 1967, "epoch": 18, "lr": 9.966018256860318e-05} +{"train_loss": 0.02123061567544937, "grad_norm": 0.06320661306381226, "global_step": 1968, "epoch": 18, "lr": 9.96598375301544e-05} +{"train_loss": 0.022230733186006546, "grad_norm": 0.13192428648471832, "global_step": 1969, "epoch": 18, "lr": 9.965949231722251e-05} +{"train_loss": 0.0240299254655838, "grad_norm": 0.12201421707868576, "global_step": 1970, "epoch": 18, "lr": 9.96591469298088e-05} +{"train_loss": 0.0237053781747818, "grad_norm": 0.20913289487361908, "global_step": 1971, "epoch": 18, "lr": 9.965880136791441e-05} +{"train_loss": 0.02194540947675705, "grad_norm": 0.09135442972183228, "global_step": 1972, "epoch": 18, "lr": 9.965845563154062e-05} +{"train_loss": 0.021044649183750153, "grad_norm": 0.058045342564582825, "global_step": 1973, "epoch": 18, "lr": 9.965810972068859e-05} +{"train_loss": 0.02132543735206127, "grad_norm": 0.04563812538981438, "global_step": 1974, "epoch": 18, "lr": 9.965776363535956e-05} +{"train_loss": 0.023538630455732346, "grad_norm": 0.036167725920677185, "global_step": 1975, "epoch": 18, "lr": 9.965741737555474e-05} +{"train_loss": 0.020614851266145706, "grad_norm": 0.04435311257839203, "global_step": 1976, "epoch": 18, "lr": 9.965707094127537e-05} +{"train_loss": 0.026330411434173584, "grad_norm": 0.13819560408592224, "global_step": 1977, "epoch": 18, "lr": 9.965672433252263e-05} +{"train_loss": 0.026700343936681747, "grad_norm": 0.15719880163669586, "global_step": 1978, "epoch": 18, "lr": 9.965637754929777e-05} +{"train_loss": 0.020482061430811882, "grad_norm": 0.07732730358839035, "global_step": 1979, "epoch": 18, "lr": 9.965603059160197e-05} +{"train_loss": 0.022731125354766846, "grad_norm": 0.11469238996505737, "global_step": 1980, "epoch": 18, "lr": 9.96556834594365e-05} +{"train_loss": 0.019055219367146492, "grad_norm": 0.03083375282585621, "global_step": 1981, "epoch": 18, "lr": 9.965533615280254e-05} +{"train_loss": 0.01910613663494587, "grad_norm": 0.11885486543178558, "global_step": 1982, "epoch": 18, "lr": 9.965498867170133e-05} +{"train_loss": 0.022767573595046997, "grad_norm": 0.08973155170679092, "global_step": 1983, "epoch": 18, "lr": 9.965464101613409e-05} +{"train_loss": 0.022882012650370598, "grad_norm": 0.11154144257307053, "global_step": 1984, "epoch": 18, "lr": 9.965429318610202e-05} +{"train_loss": 0.021985268220305443, "grad_norm": 0.04248226806521416, "global_step": 1985, "epoch": 18, "lr": 9.965394518160635e-05} +{"train_loss": 0.021908575668931007, "grad_norm": 0.14189203083515167, "global_step": 1986, "epoch": 18, "lr": 9.965359700264833e-05} +{"train_loss": 0.022900670766830444, "grad_norm": 0.05496010184288025, "global_step": 1987, "epoch": 18, "lr": 9.965324864922917e-05} +{"train_loss": 0.02074189856648445, "grad_norm": 0.035827863961458206, "global_step": 1988, "epoch": 18, "lr": 9.965290012135007e-05} +{"train_loss": 0.02066757343709469, "grad_norm": 0.03666515648365021, "global_step": 1989, "epoch": 18, "lr": 9.965255141901229e-05} +{"train_loss": 0.019429229199886322, "grad_norm": 0.07621364295482635, "global_step": 1990, "epoch": 18, "lr": 9.965220254221702e-05} +{"train_loss": 0.021916083991527557, "grad_norm": 0.04838348180055618, "global_step": 1991, "epoch": 18, "lr": 9.965185349096551e-05} +{"train_loss": 0.020240124315023422, "grad_norm": 0.06281547993421555, "global_step": 1992, "epoch": 18, "lr": 9.965150426525898e-05} +{"train_loss": 0.021978814154863358, "grad_norm": 0.10485496371984482, "global_step": 1993, "epoch": 18, "lr": 9.965115486509866e-05} +{"train_loss": 0.025043293833732605, "grad_norm": 0.08146002888679504, "global_step": 1994, "epoch": 18, "lr": 9.965080529048577e-05} +{"train_loss": 0.022317111492156982, "grad_norm": 0.07465233653783798, "global_step": 1995, "epoch": 18, "lr": 9.965045554142154e-05} +{"train_loss": 0.025555063039064407, "grad_norm": 0.06994317471981049, "global_step": 1996, "epoch": 18, "lr": 9.965010561790721e-05} +{"train_loss": 0.022439582273364067, "grad_norm": 0.04366419091820717, "global_step": 1997, "epoch": 18, "lr": 9.9649755519944e-05} +{"train_loss": 0.01848572865128517, "grad_norm": 0.06659569591283798, "global_step": 1998, "epoch": 18, "lr": 9.964940524753313e-05} +{"train_loss": 0.01939782314002514, "grad_norm": 0.0369819737970829, "global_step": 1999, "epoch": 18, "lr": 9.964905480067586e-05} +{"train_loss": 0.020500367507338524, "grad_norm": 0.07040934264659882, "global_step": 2000, "epoch": 18, "lr": 9.964870417937339e-05} +{"train_loss": 0.01819266751408577, "grad_norm": 0.05605906620621681, "global_step": 2001, "epoch": 18, "lr": 9.964835338362697e-05} +{"train_loss": 0.021075397729873657, "grad_norm": 0.05528107285499573, "global_step": 2002, "epoch": 18, "lr": 9.964800241343781e-05} +{"train_loss": 0.020610753446817398, "grad_norm": 0.04295856133103371, "global_step": 2003, "epoch": 18, "lr": 9.964765126880718e-05} +{"train_loss": 0.019733354449272156, "grad_norm": 0.03430803492665291, "global_step": 2004, "epoch": 18, "lr": 9.964729994973629e-05} +{"train_loss": 0.022251751273870468, "grad_norm": 0.038577672094106674, "global_step": 2005, "epoch": 18, "lr": 9.96469484562264e-05} +{"train_loss": 0.022325480356812477, "grad_norm": 0.0381680503487587, "global_step": 2006, "epoch": 18, "lr": 9.96465967882787e-05} +{"train_loss": 0.023856155574321747, "grad_norm": 0.0724189430475235, "global_step": 2007, "epoch": 18, "lr": 9.964624494589445e-05} +{"train_loss": 0.018909480422735214, "grad_norm": 0.06946537643671036, "global_step": 2008, "epoch": 18, "lr": 9.964589292907489e-05} +{"train_loss": 0.020118193700909615, "grad_norm": 0.06865854561328888, "global_step": 2009, "epoch": 18, "lr": 9.964554073782126e-05} +{"train_loss": 0.020388901233673096, "grad_norm": 0.05999627336859703, "global_step": 2010, "epoch": 18, "lr": 9.964518837213477e-05} +{"train_loss": 0.024720696732401848, "grad_norm": 0.04685673117637634, "global_step": 2011, "epoch": 18, "lr": 9.964483583201669e-05} +{"train_loss": 0.02328616939485073, "grad_norm": 0.0520002581179142, "global_step": 2012, "epoch": 18, "lr": 9.964448311746826e-05} +{"train_loss": 0.021885367690251685, "grad_norm": 0.08029678463935852, "global_step": 2013, "epoch": 18, "lr": 9.964413022849068e-05, "val_loss": 0.020473947748541832} +{"train_loss": 0.021448824554681778, "grad_norm": 0.08342362940311432, "global_step": 2014, "epoch": 19, "lr": 9.964377716508524e-05} +{"train_loss": 0.022395601496100426, "grad_norm": 0.049994390457868576, "global_step": 2015, "epoch": 19, "lr": 9.964342392725315e-05} +{"train_loss": 0.022940896451473236, "grad_norm": 0.07135385274887085, "global_step": 2016, "epoch": 19, "lr": 9.964307051499565e-05} +{"train_loss": 0.023402169346809387, "grad_norm": 0.05777280032634735, "global_step": 2017, "epoch": 19, "lr": 9.964271692831399e-05} +{"train_loss": 0.02095954120159149, "grad_norm": 0.0919949933886528, "global_step": 2018, "epoch": 19, "lr": 9.964236316720942e-05} +{"train_loss": 0.024476729333400726, "grad_norm": 0.06349896639585495, "global_step": 2019, "epoch": 19, "lr": 9.964200923168316e-05} +{"train_loss": 0.022828634828329086, "grad_norm": 0.11807458847761154, "global_step": 2020, "epoch": 19, "lr": 9.964165512173649e-05} +{"train_loss": 0.021018972620368004, "grad_norm": 0.07930727303028107, "global_step": 2021, "epoch": 19, "lr": 9.964130083737062e-05} +{"train_loss": 0.02103860303759575, "grad_norm": 0.10339397192001343, "global_step": 2022, "epoch": 19, "lr": 9.964094637858681e-05} +{"train_loss": 0.021010350435972214, "grad_norm": 0.06620623171329498, "global_step": 2023, "epoch": 19, "lr": 9.96405917453863e-05} +{"train_loss": 0.023493792861700058, "grad_norm": 0.03862752765417099, "global_step": 2024, "epoch": 19, "lr": 9.964023693777034e-05} +{"train_loss": 0.019836831837892532, "grad_norm": 0.052696626633405685, "global_step": 2025, "epoch": 19, "lr": 9.963988195574017e-05} +{"train_loss": 0.02129235491156578, "grad_norm": 0.078268863260746, "global_step": 2026, "epoch": 19, "lr": 9.963952679929704e-05} +{"train_loss": 0.020760053768754005, "grad_norm": 0.07671169936656952, "global_step": 2027, "epoch": 19, "lr": 9.963917146844221e-05} +{"train_loss": 0.02102534845471382, "grad_norm": 0.07210490107536316, "global_step": 2028, "epoch": 19, "lr": 9.963881596317692e-05} +{"train_loss": 0.021624824032187462, "grad_norm": 0.05616551265120506, "global_step": 2029, "epoch": 19, "lr": 9.963846028350242e-05} +{"train_loss": 0.024232573807239532, "grad_norm": 0.08781228214502335, "global_step": 2030, "epoch": 19, "lr": 9.963810442941996e-05} +{"train_loss": 0.019628366455435753, "grad_norm": 0.03765133023262024, "global_step": 2031, "epoch": 19, "lr": 9.963774840093077e-05} +{"train_loss": 0.020706932991743088, "grad_norm": 0.04943392053246498, "global_step": 2032, "epoch": 19, "lr": 9.963739219803614e-05} +{"train_loss": 0.021479392424225807, "grad_norm": 0.04890095815062523, "global_step": 2033, "epoch": 19, "lr": 9.96370358207373e-05} +{"train_loss": 0.020027320832014084, "grad_norm": 0.12942180037498474, "global_step": 2034, "epoch": 19, "lr": 9.963667926903548e-05} +{"train_loss": 0.020416520535945892, "grad_norm": 0.03602281212806702, "global_step": 2035, "epoch": 19, "lr": 9.963632254293197e-05} +{"train_loss": 0.02053334005177021, "grad_norm": 0.07955390959978104, "global_step": 2036, "epoch": 19, "lr": 9.9635965642428e-05} +{"train_loss": 0.02007276751101017, "grad_norm": 0.06723150610923767, "global_step": 2037, "epoch": 19, "lr": 9.963560856752486e-05} +{"train_loss": 0.02060074731707573, "grad_norm": 0.03300578147172928, "global_step": 2038, "epoch": 19, "lr": 9.963525131822375e-05} +{"train_loss": 0.017673451453447342, "grad_norm": 0.037817034870386124, "global_step": 2039, "epoch": 19, "lr": 9.963489389452596e-05} +{"train_loss": 0.02049010619521141, "grad_norm": 0.09946834295988083, "global_step": 2040, "epoch": 19, "lr": 9.963453629643273e-05} +{"train_loss": 0.022452963516116142, "grad_norm": 0.04217132553458214, "global_step": 2041, "epoch": 19, "lr": 9.963417852394534e-05} +{"train_loss": 0.019785989075899124, "grad_norm": 0.091635562479496, "global_step": 2042, "epoch": 19, "lr": 9.963382057706502e-05} +{"train_loss": 0.02147115021944046, "grad_norm": 0.057259075343608856, "global_step": 2043, "epoch": 19, "lr": 9.963346245579305e-05} +{"train_loss": 0.02141382545232773, "grad_norm": 0.03034585900604725, "global_step": 2044, "epoch": 19, "lr": 9.963310416013067e-05} +{"train_loss": 0.020592113956809044, "grad_norm": 0.0553542785346508, "global_step": 2045, "epoch": 19, "lr": 9.963274569007915e-05} +{"train_loss": 0.022302860394120216, "grad_norm": 0.08526279032230377, "global_step": 2046, "epoch": 19, "lr": 9.963238704563976e-05} +{"train_loss": 0.02227811887860298, "grad_norm": 0.06162232160568237, "global_step": 2047, "epoch": 19, "lr": 9.963202822681372e-05} +{"train_loss": 0.022012190893292427, "grad_norm": 0.10328997671604156, "global_step": 2048, "epoch": 19, "lr": 9.963166923360234e-05} +{"train_loss": 0.020651543512940407, "grad_norm": 0.08683273941278458, "global_step": 2049, "epoch": 19, "lr": 9.963131006600684e-05} +{"train_loss": 0.023697752505540848, "grad_norm": 0.12104147672653198, "global_step": 2050, "epoch": 19, "lr": 9.963095072402852e-05} +{"train_loss": 0.024246148765087128, "grad_norm": 0.039307840168476105, "global_step": 2051, "epoch": 19, "lr": 9.96305912076686e-05} +{"train_loss": 0.017830412834882736, "grad_norm": 0.03285463526844978, "global_step": 2052, "epoch": 19, "lr": 9.963023151692838e-05} +{"train_loss": 0.022468987852334976, "grad_norm": 0.05089829862117767, "global_step": 2053, "epoch": 19, "lr": 9.962987165180911e-05} +{"train_loss": 0.02428862825036049, "grad_norm": 0.09524413198232651, "global_step": 2054, "epoch": 19, "lr": 9.962951161231206e-05} +{"train_loss": 0.021492522209882736, "grad_norm": 0.07378341257572174, "global_step": 2055, "epoch": 19, "lr": 9.962915139843848e-05} +{"train_loss": 0.02007463574409485, "grad_norm": 0.07633420079946518, "global_step": 2056, "epoch": 19, "lr": 9.962879101018964e-05} +{"train_loss": 0.021770406514406204, "grad_norm": 0.05408133193850517, "global_step": 2057, "epoch": 19, "lr": 9.962843044756682e-05} +{"train_loss": 0.020164333283901215, "grad_norm": 0.039268460124731064, "global_step": 2058, "epoch": 19, "lr": 9.962806971057127e-05} +{"train_loss": 0.024964747950434685, "grad_norm": 0.10687825828790665, "global_step": 2059, "epoch": 19, "lr": 9.96277087992043e-05} +{"train_loss": 0.021377140656113625, "grad_norm": 0.058332111686468124, "global_step": 2060, "epoch": 19, "lr": 9.962734771346711e-05} +{"train_loss": 0.020559562370181084, "grad_norm": 0.045615632086992264, "global_step": 2061, "epoch": 19, "lr": 9.962698645336101e-05} +{"train_loss": 0.017288804054260254, "grad_norm": 0.0373363234102726, "global_step": 2062, "epoch": 19, "lr": 9.962662501888727e-05} +{"train_loss": 0.02037816308438778, "grad_norm": 0.0689130499958992, "global_step": 2063, "epoch": 19, "lr": 9.962626341004712e-05} +{"train_loss": 0.021694030612707138, "grad_norm": 0.10002776980400085, "global_step": 2064, "epoch": 19, "lr": 9.96259016268419e-05} +{"train_loss": 0.021042881533503532, "grad_norm": 0.05714599788188934, "global_step": 2065, "epoch": 19, "lr": 9.962553966927282e-05} +{"train_loss": 0.024056166410446167, "grad_norm": 0.04562560096383095, "global_step": 2066, "epoch": 19, "lr": 9.96251775373412e-05} +{"train_loss": 0.02225741744041443, "grad_norm": 0.06428240239620209, "global_step": 2067, "epoch": 19, "lr": 9.962481523104826e-05} +{"train_loss": 0.02132345736026764, "grad_norm": 0.06586203724145889, "global_step": 2068, "epoch": 19, "lr": 9.962445275039531e-05} +{"train_loss": 0.02303517982363701, "grad_norm": 0.06821732968091965, "global_step": 2069, "epoch": 19, "lr": 9.962409009538362e-05} +{"train_loss": 0.023379677906632423, "grad_norm": 0.0642886608839035, "global_step": 2070, "epoch": 19, "lr": 9.962372726601445e-05} +{"train_loss": 0.018238332122564316, "grad_norm": 0.03475271165370941, "global_step": 2071, "epoch": 19, "lr": 9.962336426228909e-05} +{"train_loss": 0.021873990073800087, "grad_norm": 0.05822402983903885, "global_step": 2072, "epoch": 19, "lr": 9.96230010842088e-05} +{"train_loss": 0.021088778972625732, "grad_norm": 0.055234283208847046, "global_step": 2073, "epoch": 19, "lr": 9.962263773177488e-05} +{"train_loss": 0.02338206209242344, "grad_norm": 0.05389580503106117, "global_step": 2074, "epoch": 19, "lr": 9.962227420498859e-05} +{"train_loss": 0.021752500906586647, "grad_norm": 0.08455260843038559, "global_step": 2075, "epoch": 19, "lr": 9.962191050385119e-05} +{"train_loss": 0.023875530809164047, "grad_norm": 0.05815834179520607, "global_step": 2076, "epoch": 19, "lr": 9.962154662836399e-05} +{"train_loss": 0.026707200333476067, "grad_norm": 0.07963179051876068, "global_step": 2077, "epoch": 19, "lr": 9.962118257852826e-05} +{"train_loss": 0.02406478300690651, "grad_norm": 0.18532152473926544, "global_step": 2078, "epoch": 19, "lr": 9.962081835434527e-05} +{"train_loss": 0.020289473235607147, "grad_norm": 0.10280568897724152, "global_step": 2079, "epoch": 19, "lr": 9.96204539558163e-05} +{"train_loss": 0.02492103911936283, "grad_norm": 0.037882186472415924, "global_step": 2080, "epoch": 19, "lr": 9.962008938294263e-05} +{"train_loss": 0.022491326555609703, "grad_norm": 0.06059687212109566, "global_step": 2081, "epoch": 19, "lr": 9.961972463572555e-05} +{"train_loss": 0.02239811234176159, "grad_norm": 0.0398753359913826, "global_step": 2082, "epoch": 19, "lr": 9.961935971416634e-05} +{"train_loss": 0.023091398179531097, "grad_norm": 0.04024713858962059, "global_step": 2083, "epoch": 19, "lr": 9.961899461826627e-05} +{"train_loss": 0.023643149062991142, "grad_norm": 0.10783113539218903, "global_step": 2084, "epoch": 19, "lr": 9.961862934802664e-05} +{"train_loss": 0.017934288829565048, "grad_norm": 0.041663046926259995, "global_step": 2085, "epoch": 19, "lr": 9.961826390344873e-05} +{"train_loss": 0.0208582803606987, "grad_norm": 0.058864474296569824, "global_step": 2086, "epoch": 19, "lr": 9.961789828453381e-05} +{"train_loss": 0.019995469599962234, "grad_norm": 0.05711562931537628, "global_step": 2087, "epoch": 19, "lr": 9.961753249128316e-05} +{"train_loss": 0.023532990366220474, "grad_norm": 0.055892132222652435, "global_step": 2088, "epoch": 19, "lr": 9.96171665236981e-05} +{"train_loss": 0.019510211423039436, "grad_norm": 0.05233442783355713, "global_step": 2089, "epoch": 19, "lr": 9.96168003817799e-05} +{"train_loss": 0.02384502813220024, "grad_norm": 0.13960003852844238, "global_step": 2090, "epoch": 19, "lr": 9.961643406552982e-05} +{"train_loss": 0.02069927752017975, "grad_norm": 0.038710083812475204, "global_step": 2091, "epoch": 19, "lr": 9.961606757494917e-05} +{"train_loss": 0.020260639488697052, "grad_norm": 0.12211719900369644, "global_step": 2092, "epoch": 19, "lr": 9.961570091003924e-05} +{"train_loss": 0.02201402559876442, "grad_norm": 0.14497138559818268, "global_step": 2093, "epoch": 19, "lr": 9.961533407080132e-05} +{"train_loss": 0.022006195038557053, "grad_norm": 0.04068718105554581, "global_step": 2094, "epoch": 19, "lr": 9.96149670572367e-05} +{"train_loss": 0.021282246336340904, "grad_norm": 0.14991673827171326, "global_step": 2095, "epoch": 19, "lr": 9.961459986934665e-05} +{"train_loss": 0.0204304326325655, "grad_norm": 0.05557052046060562, "global_step": 2096, "epoch": 19, "lr": 9.961423250713248e-05} +{"train_loss": 0.02134021371603012, "grad_norm": 0.07167889922857285, "global_step": 2097, "epoch": 19, "lr": 9.961386497059547e-05} +{"train_loss": 0.02621344104409218, "grad_norm": 0.06518013030290604, "global_step": 2098, "epoch": 19, "lr": 9.961349725973692e-05} +{"train_loss": 0.022167138755321503, "grad_norm": 0.12784110009670258, "global_step": 2099, "epoch": 19, "lr": 9.961312937455811e-05} +{"train_loss": 0.01979992538690567, "grad_norm": 0.06720499694347382, "global_step": 2100, "epoch": 19, "lr": 9.961276131506034e-05} +{"train_loss": 0.022474071010947227, "grad_norm": 0.0845848023891449, "global_step": 2101, "epoch": 19, "lr": 9.96123930812449e-05} +{"train_loss": 0.019685683771967888, "grad_norm": 0.1522713601589203, "global_step": 2102, "epoch": 19, "lr": 9.96120246731131e-05} +{"train_loss": 0.02375253103673458, "grad_norm": 0.0532294325530529, "global_step": 2103, "epoch": 19, "lr": 9.96116560906662e-05} +{"train_loss": 0.023831874132156372, "grad_norm": 0.0897173210978508, "global_step": 2104, "epoch": 19, "lr": 9.961128733390555e-05} +{"train_loss": 0.021118102595210075, "grad_norm": 0.06065119430422783, "global_step": 2105, "epoch": 19, "lr": 9.961091840283238e-05} +{"train_loss": 0.022477395832538605, "grad_norm": 0.092587411403656, "global_step": 2106, "epoch": 19, "lr": 9.961054929744803e-05} +{"train_loss": 0.020687730982899666, "grad_norm": 0.1015513688325882, "global_step": 2107, "epoch": 19, "lr": 9.961018001775379e-05} +{"train_loss": 0.022452037781476974, "grad_norm": 0.05797388777136803, "global_step": 2108, "epoch": 19, "lr": 9.960981056375094e-05} +{"train_loss": 0.021950533613562584, "grad_norm": 0.13912443816661835, "global_step": 2109, "epoch": 19, "lr": 9.96094409354408e-05} +{"train_loss": 0.020404163748025894, "grad_norm": 0.03852914646267891, "global_step": 2110, "epoch": 19, "lr": 9.960907113282465e-05} +{"train_loss": 0.021397866308689117, "grad_norm": 0.0849936380982399, "global_step": 2111, "epoch": 19, "lr": 9.96087011559038e-05} +{"train_loss": 0.021310904994606972, "grad_norm": 0.051044076681137085, "global_step": 2112, "epoch": 19, "lr": 9.960833100467956e-05} +{"train_loss": 0.023514337837696075, "grad_norm": 0.06372283399105072, "global_step": 2113, "epoch": 19, "lr": 9.960796067915321e-05} +{"train_loss": 0.022875357419252396, "grad_norm": 0.04809068888425827, "global_step": 2114, "epoch": 19, "lr": 9.960759017932606e-05} +{"train_loss": 0.02194211818277836, "grad_norm": 0.08437346667051315, "global_step": 2115, "epoch": 19, "lr": 9.960721950519941e-05} +{"train_loss": 0.02005809172987938, "grad_norm": 0.041609395295381546, "global_step": 2116, "epoch": 19, "lr": 9.960684865677457e-05} +{"train_loss": 0.019251976162195206, "grad_norm": 0.05647313594818115, "global_step": 2117, "epoch": 19, "lr": 9.960647763405283e-05} +{"train_loss": 0.02582647278904915, "grad_norm": 0.0634281262755394, "global_step": 2118, "epoch": 19, "lr": 9.96061064370355e-05} +{"train_loss": 0.021737411214073875, "grad_norm": 0.07631850242614746, "global_step": 2119, "epoch": 19, "lr": 9.96057350657239e-05, "val_loss": 0.020076794549822807} +{"train_loss": 0.019156135618686676, "grad_norm": 0.041648685932159424, "global_step": 2120, "epoch": 20, "lr": 9.960536352011931e-05} +{"train_loss": 0.02107103541493416, "grad_norm": 0.06054820120334625, "global_step": 2121, "epoch": 20, "lr": 9.960499180022305e-05} +{"train_loss": 0.020755043253302574, "grad_norm": 0.054660625755786896, "global_step": 2122, "epoch": 20, "lr": 9.96046199060364e-05} +{"train_loss": 0.019416790455579758, "grad_norm": 0.10881643742322922, "global_step": 2123, "epoch": 20, "lr": 9.960424783756071e-05} +{"train_loss": 0.022836266085505486, "grad_norm": 0.0765797346830368, "global_step": 2124, "epoch": 20, "lr": 9.960387559479727e-05} +{"train_loss": 0.022291304543614388, "grad_norm": 0.06898386031389236, "global_step": 2125, "epoch": 20, "lr": 9.960350317774737e-05} +{"train_loss": 0.023307843133807182, "grad_norm": 0.048704661428928375, "global_step": 2126, "epoch": 20, "lr": 9.960313058641233e-05} +{"train_loss": 0.02110596001148224, "grad_norm": 0.04791352152824402, "global_step": 2127, "epoch": 20, "lr": 9.960275782079346e-05} +{"train_loss": 0.020999770611524582, "grad_norm": 0.10761865973472595, "global_step": 2128, "epoch": 20, "lr": 9.960238488089209e-05} +{"train_loss": 0.019744373857975006, "grad_norm": 0.06801900267601013, "global_step": 2129, "epoch": 20, "lr": 9.960201176670949e-05} +{"train_loss": 0.02280402183532715, "grad_norm": 0.06699524074792862, "global_step": 2130, "epoch": 20, "lr": 9.960163847824701e-05} +{"train_loss": 0.02183319628238678, "grad_norm": 0.09026007354259491, "global_step": 2131, "epoch": 20, "lr": 9.960126501550593e-05} +{"train_loss": 0.023442953824996948, "grad_norm": 0.09753577411174774, "global_step": 2132, "epoch": 20, "lr": 9.960089137848758e-05} +{"train_loss": 0.02336297743022442, "grad_norm": 0.0662059336900711, "global_step": 2133, "epoch": 20, "lr": 9.960051756719327e-05} +{"train_loss": 0.017823264002799988, "grad_norm": 0.03842172771692276, "global_step": 2134, "epoch": 20, "lr": 9.96001435816243e-05} +{"train_loss": 0.02248370833694935, "grad_norm": 0.03640315309166908, "global_step": 2135, "epoch": 20, "lr": 9.959976942178202e-05} +{"train_loss": 0.02204647660255432, "grad_norm": 0.07595443725585938, "global_step": 2136, "epoch": 20, "lr": 9.959939508766771e-05} +{"train_loss": 0.017662081867456436, "grad_norm": 0.03134292736649513, "global_step": 2137, "epoch": 20, "lr": 9.959902057928268e-05} +{"train_loss": 0.022826164960861206, "grad_norm": 0.05782289057970047, "global_step": 2138, "epoch": 20, "lr": 9.959864589662829e-05} +{"train_loss": 0.025675278156995773, "grad_norm": 0.038652919232845306, "global_step": 2139, "epoch": 20, "lr": 9.95982710397058e-05} +{"train_loss": 0.02058887667953968, "grad_norm": 0.04488178342580795, "global_step": 2140, "epoch": 20, "lr": 9.959789600851657e-05} +{"train_loss": 0.02366385981440544, "grad_norm": 0.0476810522377491, "global_step": 2141, "epoch": 20, "lr": 9.95975208030619e-05} +{"train_loss": 0.02097179740667343, "grad_norm": 0.07762666046619415, "global_step": 2142, "epoch": 20, "lr": 9.95971454233431e-05} +{"train_loss": 0.02282511070370674, "grad_norm": 0.09350744634866714, "global_step": 2143, "epoch": 20, "lr": 9.959676986936153e-05} +{"train_loss": 0.022328834980726242, "grad_norm": 0.03177107498049736, "global_step": 2144, "epoch": 20, "lr": 9.959639414111845e-05} +{"train_loss": 0.02239402011036873, "grad_norm": 0.07267245650291443, "global_step": 2145, "epoch": 20, "lr": 9.959601823861523e-05} +{"train_loss": 0.022917166352272034, "grad_norm": 0.040863484144210815, "global_step": 2146, "epoch": 20, "lr": 9.959564216185316e-05} +{"train_loss": 0.019846893846988678, "grad_norm": 0.041971929371356964, "global_step": 2147, "epoch": 20, "lr": 9.959526591083358e-05} +{"train_loss": 0.018532197922468185, "grad_norm": 0.05018794536590576, "global_step": 2148, "epoch": 20, "lr": 9.95948894855578e-05} +{"train_loss": 0.021281465888023376, "grad_norm": 0.047703422605991364, "global_step": 2149, "epoch": 20, "lr": 9.959451288602714e-05} +{"train_loss": 0.01870795339345932, "grad_norm": 0.03390570729970932, "global_step": 2150, "epoch": 20, "lr": 9.959413611224293e-05} +{"train_loss": 0.022444330155849457, "grad_norm": 0.03952018544077873, "global_step": 2151, "epoch": 20, "lr": 9.959375916420651e-05} +{"train_loss": 0.020048713311553, "grad_norm": 0.10713184624910355, "global_step": 2152, "epoch": 20, "lr": 9.959338204191917e-05} +{"train_loss": 0.021820779889822006, "grad_norm": 0.045947834849357605, "global_step": 2153, "epoch": 20, "lr": 9.959300474538226e-05} +{"train_loss": 0.021739821881055832, "grad_norm": 0.10923308879137039, "global_step": 2154, "epoch": 20, "lr": 9.959262727459711e-05} +{"train_loss": 0.019506994634866714, "grad_norm": 0.10680153220891953, "global_step": 2155, "epoch": 20, "lr": 9.959224962956502e-05} +{"train_loss": 0.0204276405274868, "grad_norm": 0.11934547871351242, "global_step": 2156, "epoch": 20, "lr": 9.959187181028733e-05} +{"train_loss": 0.021298466250300407, "grad_norm": 0.058394867926836014, "global_step": 2157, "epoch": 20, "lr": 9.959149381676538e-05} +{"train_loss": 0.022321542724967003, "grad_norm": 0.11478926986455917, "global_step": 2158, "epoch": 20, "lr": 9.959111564900048e-05} +{"train_loss": 0.02310919389128685, "grad_norm": 0.07813297212123871, "global_step": 2159, "epoch": 20, "lr": 9.959073730699397e-05} +{"train_loss": 0.021452490240335464, "grad_norm": 0.04078662768006325, "global_step": 2160, "epoch": 20, "lr": 9.959035879074717e-05} +{"train_loss": 0.020952491089701653, "grad_norm": 0.050954073667526245, "global_step": 2161, "epoch": 20, "lr": 9.958998010026142e-05} +{"train_loss": 0.022056587040424347, "grad_norm": 0.05044623091816902, "global_step": 2162, "epoch": 20, "lr": 9.958960123553804e-05} +{"train_loss": 0.017665013670921326, "grad_norm": 0.045012276619672775, "global_step": 2163, "epoch": 20, "lr": 9.958922219657838e-05} +{"train_loss": 0.023951331153512, "grad_norm": 0.061522096395492554, "global_step": 2164, "epoch": 20, "lr": 9.958884298338375e-05} +{"train_loss": 0.021006044000387192, "grad_norm": 0.03856208920478821, "global_step": 2165, "epoch": 20, "lr": 9.95884635959555e-05} +{"train_loss": 0.01927376165986061, "grad_norm": 0.03650094196200371, "global_step": 2166, "epoch": 20, "lr": 9.958808403429494e-05} +{"train_loss": 0.02299521490931511, "grad_norm": 0.054744236171245575, "global_step": 2167, "epoch": 20, "lr": 9.958770429840343e-05} +{"train_loss": 0.020785590633749962, "grad_norm": 0.04136888310313225, "global_step": 2168, "epoch": 20, "lr": 9.958732438828229e-05} +{"train_loss": 0.02070680260658264, "grad_norm": 0.06252428889274597, "global_step": 2169, "epoch": 20, "lr": 9.958694430393285e-05} +{"train_loss": 0.020318562164902687, "grad_norm": 0.10536319017410278, "global_step": 2170, "epoch": 20, "lr": 9.958656404535647e-05} +{"train_loss": 0.021434541791677475, "grad_norm": 0.0416279174387455, "global_step": 2171, "epoch": 20, "lr": 9.958618361255445e-05} +{"train_loss": 0.021307459101080894, "grad_norm": 0.0835084542632103, "global_step": 2172, "epoch": 20, "lr": 9.958580300552815e-05} +{"train_loss": 0.023401502519845963, "grad_norm": 0.07956580817699432, "global_step": 2173, "epoch": 20, "lr": 9.95854222242789e-05} +{"train_loss": 0.023471444845199585, "grad_norm": 0.09964471310377121, "global_step": 2174, "epoch": 20, "lr": 9.958504126880805e-05} +{"train_loss": 0.02285159006714821, "grad_norm": 0.05047067254781723, "global_step": 2175, "epoch": 20, "lr": 9.958466013911691e-05} +{"train_loss": 0.020191529765725136, "grad_norm": 0.04949333891272545, "global_step": 2176, "epoch": 20, "lr": 9.958427883520686e-05} +{"train_loss": 0.023202931508421898, "grad_norm": 0.0353163480758667, "global_step": 2177, "epoch": 20, "lr": 9.95838973570792e-05} +{"train_loss": 0.024921899661421776, "grad_norm": 0.061851855367422104, "global_step": 2178, "epoch": 20, "lr": 9.95835157047353e-05} +{"train_loss": 0.016328496858477592, "grad_norm": 0.03390097618103027, "global_step": 2179, "epoch": 20, "lr": 9.958313387817649e-05} +{"train_loss": 0.020603325217962265, "grad_norm": 0.08186246454715729, "global_step": 2180, "epoch": 20, "lr": 9.958275187740411e-05} +{"train_loss": 0.023925840854644775, "grad_norm": 0.08522399514913559, "global_step": 2181, "epoch": 20, "lr": 9.95823697024195e-05} +{"train_loss": 0.020745238289237022, "grad_norm": 0.0706988126039505, "global_step": 2182, "epoch": 20, "lr": 9.958198735322401e-05} +{"train_loss": 0.01797034777700901, "grad_norm": 0.0937231257557869, "global_step": 2183, "epoch": 20, "lr": 9.958160482981898e-05} +{"train_loss": 0.022170349955558777, "grad_norm": 0.045483529567718506, "global_step": 2184, "epoch": 20, "lr": 9.958122213220573e-05} +{"train_loss": 0.02006189152598381, "grad_norm": 0.07674752920866013, "global_step": 2185, "epoch": 20, "lr": 9.958083926038567e-05} +{"train_loss": 0.021820906549692154, "grad_norm": 0.10549067705869675, "global_step": 2186, "epoch": 20, "lr": 9.958045621436006e-05} +{"train_loss": 0.02133696898818016, "grad_norm": 0.11095022410154343, "global_step": 2187, "epoch": 20, "lr": 9.95800729941303e-05} +{"train_loss": 0.02052493393421173, "grad_norm": 0.043619461357593536, "global_step": 2188, "epoch": 20, "lr": 9.957968959969774e-05} +{"train_loss": 0.021561821922659874, "grad_norm": 0.056696515530347824, "global_step": 2189, "epoch": 20, "lr": 9.957930603106371e-05} +{"train_loss": 0.020499877631664276, "grad_norm": 0.05052398145198822, "global_step": 2190, "epoch": 20, "lr": 9.957892228822955e-05} +{"train_loss": 0.02284371666610241, "grad_norm": 0.060156285762786865, "global_step": 2191, "epoch": 20, "lr": 9.957853837119662e-05} +{"train_loss": 0.021504826843738556, "grad_norm": 0.051310744136571884, "global_step": 2192, "epoch": 20, "lr": 9.957815427996629e-05} +{"train_loss": 0.022115230560302734, "grad_norm": 0.04668546840548515, "global_step": 2193, "epoch": 20, "lr": 9.957777001453985e-05} +{"train_loss": 0.019104402512311935, "grad_norm": 0.12892577052116394, "global_step": 2194, "epoch": 20, "lr": 9.957738557491871e-05} +{"train_loss": 0.02467462792992592, "grad_norm": 0.1263670176267624, "global_step": 2195, "epoch": 20, "lr": 9.957700096110419e-05} +{"train_loss": 0.023324711248278618, "grad_norm": 0.06517069041728973, "global_step": 2196, "epoch": 20, "lr": 9.957661617309766e-05} +{"train_loss": 0.01911267824470997, "grad_norm": 0.11903513967990875, "global_step": 2197, "epoch": 20, "lr": 9.957623121090045e-05} +{"train_loss": 0.02114013209939003, "grad_norm": 0.09212610870599747, "global_step": 2198, "epoch": 20, "lr": 9.957584607451392e-05} +{"train_loss": 0.023723313584923744, "grad_norm": 0.04515041783452034, "global_step": 2199, "epoch": 20, "lr": 9.957546076393943e-05} +{"train_loss": 0.02368120476603508, "grad_norm": 0.08497193455696106, "global_step": 2200, "epoch": 20, "lr": 9.957507527917833e-05} +{"train_loss": 0.020568354055285454, "grad_norm": 0.038747213780879974, "global_step": 2201, "epoch": 20, "lr": 9.957468962023199e-05} +{"train_loss": 0.02352435514330864, "grad_norm": 0.03454838693141937, "global_step": 2202, "epoch": 20, "lr": 9.957430378710174e-05} +{"train_loss": 0.02333982102572918, "grad_norm": 0.08061902970075607, "global_step": 2203, "epoch": 20, "lr": 9.957391777978894e-05} +{"train_loss": 0.01976119540631771, "grad_norm": 0.06312721222639084, "global_step": 2204, "epoch": 20, "lr": 9.957353159829496e-05} +{"train_loss": 0.019878007471561432, "grad_norm": 0.08358169347047806, "global_step": 2205, "epoch": 20, "lr": 9.957314524262113e-05} +{"train_loss": 0.02706047147512436, "grad_norm": 0.14119403064250946, "global_step": 2206, "epoch": 20, "lr": 9.957275871276886e-05} +{"train_loss": 0.019687891006469727, "grad_norm": 0.1072055995464325, "global_step": 2207, "epoch": 20, "lr": 9.957237200873946e-05} +{"train_loss": 0.01946832798421383, "grad_norm": 0.08935627341270447, "global_step": 2208, "epoch": 20, "lr": 9.957198513053429e-05} +{"train_loss": 0.02521035075187683, "grad_norm": 0.05032725632190704, "global_step": 2209, "epoch": 20, "lr": 9.957159807815472e-05} +{"train_loss": 0.021078936755657196, "grad_norm": 0.03689717501401901, "global_step": 2210, "epoch": 20, "lr": 9.957121085160214e-05} +{"train_loss": 0.023208051919937134, "grad_norm": 0.05849588289856911, "global_step": 2211, "epoch": 20, "lr": 9.957082345087788e-05} +{"train_loss": 0.018682915717363358, "grad_norm": 0.05717877671122551, "global_step": 2212, "epoch": 20, "lr": 9.957043587598327e-05} +{"train_loss": 0.0222090482711792, "grad_norm": 0.10018343478441238, "global_step": 2213, "epoch": 20, "lr": 9.957004812691974e-05} +{"train_loss": 0.023610157892107964, "grad_norm": 0.18584756553173065, "global_step": 2214, "epoch": 20, "lr": 9.95696602036886e-05} +{"train_loss": 0.025778379291296005, "grad_norm": 0.06552968919277191, "global_step": 2215, "epoch": 20, "lr": 9.956927210629123e-05} +{"train_loss": 0.020847417414188385, "grad_norm": 0.05103039741516113, "global_step": 2216, "epoch": 20, "lr": 9.956888383472902e-05} +{"train_loss": 0.021232858300209045, "grad_norm": 0.05331649258732796, "global_step": 2217, "epoch": 20, "lr": 9.956849538900328e-05} +{"train_loss": 0.020277855917811394, "grad_norm": 0.04832588508725166, "global_step": 2218, "epoch": 20, "lr": 9.956810676911542e-05} +{"train_loss": 0.02157396450638771, "grad_norm": 0.038512296974658966, "global_step": 2219, "epoch": 20, "lr": 9.956771797506679e-05} +{"train_loss": 0.02239786647260189, "grad_norm": 0.052644677460193634, "global_step": 2220, "epoch": 20, "lr": 9.956732900685875e-05} +{"train_loss": 0.0230107344686985, "grad_norm": 0.060386717319488525, "global_step": 2221, "epoch": 20, "lr": 9.956693986449266e-05} +{"train_loss": 0.020992472767829895, "grad_norm": 0.08563687652349472, "global_step": 2222, "epoch": 20, "lr": 9.956655054796991e-05} +{"train_loss": 0.02349812164902687, "grad_norm": 0.11385931074619293, "global_step": 2223, "epoch": 20, "lr": 9.956616105729187e-05} +{"train_loss": 0.02110932022333145, "grad_norm": 0.0694667175412178, "global_step": 2224, "epoch": 20, "lr": 9.956577139245988e-05} +{"train_loss": 0.02160288635992779, "grad_norm": 0.04509332776069641, "global_step": 2225, "epoch": 20, "lr": 9.956538155347534e-05, "train_batch_mse_error": 0.02003643289208412, "val_loss": 0.020591557025909424} +{"train_loss": 0.021479453891515732, "grad_norm": 0.11317005753517151, "global_step": 2226, "epoch": 21, "lr": 9.956499154033958e-05} +{"train_loss": 0.024078886955976486, "grad_norm": 0.07617577910423279, "global_step": 2227, "epoch": 21, "lr": 9.9564601353054e-05} +{"train_loss": 0.020205745473504066, "grad_norm": 0.061969250440597534, "global_step": 2228, "epoch": 21, "lr": 9.956421099161999e-05} +{"train_loss": 0.023723052814602852, "grad_norm": 0.05761485919356346, "global_step": 2229, "epoch": 21, "lr": 9.956382045603889e-05} +{"train_loss": 0.0216091088950634, "grad_norm": 0.04252634197473526, "global_step": 2230, "epoch": 21, "lr": 9.956342974631207e-05} +{"train_loss": 0.02260936051607132, "grad_norm": 0.03519357368350029, "global_step": 2231, "epoch": 21, "lr": 9.956303886244092e-05} +{"train_loss": 0.02002323418855667, "grad_norm": 0.0376071035861969, "global_step": 2232, "epoch": 21, "lr": 9.95626478044268e-05} +{"train_loss": 0.02022269181907177, "grad_norm": 0.03324330225586891, "global_step": 2233, "epoch": 21, "lr": 9.956225657227109e-05} +{"train_loss": 0.020898092538118362, "grad_norm": 0.13500352203845978, "global_step": 2234, "epoch": 21, "lr": 9.956186516597516e-05} +{"train_loss": 0.02039267309010029, "grad_norm": 0.04189680889248848, "global_step": 2235, "epoch": 21, "lr": 9.95614735855404e-05} +{"train_loss": 0.019360307604074478, "grad_norm": 0.04341401904821396, "global_step": 2236, "epoch": 21, "lr": 9.956108183096817e-05} +{"train_loss": 0.024751167744398117, "grad_norm": 0.11080672591924667, "global_step": 2237, "epoch": 21, "lr": 9.956068990225983e-05} +{"train_loss": 0.020512759685516357, "grad_norm": 0.08530599623918533, "global_step": 2238, "epoch": 21, "lr": 9.956029779941683e-05} +{"train_loss": 0.018563948571681976, "grad_norm": 0.04620065540075302, "global_step": 2239, "epoch": 21, "lr": 9.955990552244045e-05} +{"train_loss": 0.021401479840278625, "grad_norm": 0.05745549127459526, "global_step": 2240, "epoch": 21, "lr": 9.955951307133214e-05} +{"train_loss": 0.020855054259300232, "grad_norm": 0.05461679399013519, "global_step": 2241, "epoch": 21, "lr": 9.955912044609326e-05} +{"train_loss": 0.02024412900209427, "grad_norm": 0.09980982542037964, "global_step": 2242, "epoch": 21, "lr": 9.955872764672515e-05} +{"train_loss": 0.02220500446856022, "grad_norm": 0.07270143181085587, "global_step": 2243, "epoch": 21, "lr": 9.955833467322926e-05} +{"train_loss": 0.02214844524860382, "grad_norm": 0.11669453233480453, "global_step": 2244, "epoch": 21, "lr": 9.955794152560692e-05} +{"train_loss": 0.02148938551545143, "grad_norm": 0.08541758358478546, "global_step": 2245, "epoch": 21, "lr": 9.955754820385951e-05} +{"train_loss": 0.02405177429318428, "grad_norm": 0.04467616602778435, "global_step": 2246, "epoch": 21, "lr": 9.955715470798845e-05} +{"train_loss": 0.023636721074581146, "grad_norm": 0.03984830901026726, "global_step": 2247, "epoch": 21, "lr": 9.95567610379951e-05} +{"train_loss": 0.020589204505085945, "grad_norm": 0.047435861080884933, "global_step": 2248, "epoch": 21, "lr": 9.955636719388083e-05} +{"train_loss": 0.020768722519278526, "grad_norm": 0.058047980070114136, "global_step": 2249, "epoch": 21, "lr": 9.955597317564705e-05} +{"train_loss": 0.019625771790742874, "grad_norm": 0.03431187942624092, "global_step": 2250, "epoch": 21, "lr": 9.955557898329511e-05} +{"train_loss": 0.02415606752038002, "grad_norm": 0.03819416090846062, "global_step": 2251, "epoch": 21, "lr": 9.955518461682643e-05} +{"train_loss": 0.018558280542492867, "grad_norm": 0.04828106239438057, "global_step": 2252, "epoch": 21, "lr": 9.955479007624239e-05} +{"train_loss": 0.025271475315093994, "grad_norm": 0.07843253761529922, "global_step": 2253, "epoch": 21, "lr": 9.955439536154435e-05} +{"train_loss": 0.022387072443962097, "grad_norm": 0.04042108356952667, "global_step": 2254, "epoch": 21, "lr": 9.955400047273374e-05} +{"train_loss": 0.023738538846373558, "grad_norm": 0.02920020930469036, "global_step": 2255, "epoch": 21, "lr": 9.955360540981191e-05} +{"train_loss": 0.022728193551301956, "grad_norm": 0.07104308158159256, "global_step": 2256, "epoch": 21, "lr": 9.955321017278025e-05} +{"train_loss": 0.02245458960533142, "grad_norm": 0.08528048545122147, "global_step": 2257, "epoch": 21, "lr": 9.955281476164018e-05} +{"train_loss": 0.022097833454608917, "grad_norm": 0.08322963863611221, "global_step": 2258, "epoch": 21, "lr": 9.955241917639306e-05} +{"train_loss": 0.019866058602929115, "grad_norm": 0.10848185420036316, "global_step": 2259, "epoch": 21, "lr": 9.955202341704029e-05} +{"train_loss": 0.022026697173714638, "grad_norm": 0.044521111994981766, "global_step": 2260, "epoch": 21, "lr": 9.955162748358324e-05} +{"train_loss": 0.02013494074344635, "grad_norm": 0.057178545743227005, "global_step": 2261, "epoch": 21, "lr": 9.955123137602334e-05} +{"train_loss": 0.020031508058309555, "grad_norm": 0.06860779970884323, "global_step": 2262, "epoch": 21, "lr": 9.955083509436196e-05} +{"train_loss": 0.023126501590013504, "grad_norm": 0.08876451104879379, "global_step": 2263, "epoch": 21, "lr": 9.955043863860049e-05} +{"train_loss": 0.024684857577085495, "grad_norm": 0.1215335875749588, "global_step": 2264, "epoch": 21, "lr": 9.955004200874034e-05} +{"train_loss": 0.021133646368980408, "grad_norm": 0.06405983120203018, "global_step": 2265, "epoch": 21, "lr": 9.954964520478288e-05} +{"train_loss": 0.020731408149003983, "grad_norm": 0.041495874524116516, "global_step": 2266, "epoch": 21, "lr": 9.954924822672953e-05} +{"train_loss": 0.020656786859035492, "grad_norm": 0.09546800702810287, "global_step": 2267, "epoch": 21, "lr": 9.954885107458165e-05} +{"train_loss": 0.021474655717611313, "grad_norm": 0.033265821635723114, "global_step": 2268, "epoch": 21, "lr": 9.954845374834066e-05} +{"train_loss": 0.018294263631105423, "grad_norm": 0.07734593003988266, "global_step": 2269, "epoch": 21, "lr": 9.954805624800796e-05} +{"train_loss": 0.01956474967300892, "grad_norm": 0.10314865410327911, "global_step": 2270, "epoch": 21, "lr": 9.954765857358494e-05} +{"train_loss": 0.023556075990200043, "grad_norm": 0.07411205023527145, "global_step": 2271, "epoch": 21, "lr": 9.954726072507299e-05} +{"train_loss": 0.021326318383216858, "grad_norm": 0.09175600856542587, "global_step": 2272, "epoch": 21, "lr": 9.95468627024735e-05} +{"train_loss": 0.0193791426718235, "grad_norm": 0.038964759558439255, "global_step": 2273, "epoch": 21, "lr": 9.95464645057879e-05} +{"train_loss": 0.02308153733611107, "grad_norm": 0.08914477378129959, "global_step": 2274, "epoch": 21, "lr": 9.954606613501757e-05} +{"train_loss": 0.024151530116796494, "grad_norm": 0.04969104379415512, "global_step": 2275, "epoch": 21, "lr": 9.95456675901639e-05} +{"train_loss": 0.025566885247826576, "grad_norm": 0.11623258888721466, "global_step": 2276, "epoch": 21, "lr": 9.95452688712283e-05} +{"train_loss": 0.020453764125704765, "grad_norm": 0.11225538700819016, "global_step": 2277, "epoch": 21, "lr": 9.954486997821219e-05} +{"train_loss": 0.018992826342582703, "grad_norm": 0.07998500019311905, "global_step": 2278, "epoch": 21, "lr": 9.954447091111694e-05} +{"train_loss": 0.021924618631601334, "grad_norm": 0.07344096153974533, "global_step": 2279, "epoch": 21, "lr": 9.954407166994397e-05} +{"train_loss": 0.019449319690465927, "grad_norm": 0.08091016113758087, "global_step": 2280, "epoch": 21, "lr": 9.954367225469468e-05} +{"train_loss": 0.02210019715130329, "grad_norm": 0.04759472981095314, "global_step": 2281, "epoch": 21, "lr": 9.954327266537047e-05} +{"train_loss": 0.020843423902988434, "grad_norm": 0.06310810148715973, "global_step": 2282, "epoch": 21, "lr": 9.954287290197273e-05} +{"train_loss": 0.021307922899723053, "grad_norm": 0.10098298639059067, "global_step": 2283, "epoch": 21, "lr": 9.954247296450289e-05} +{"train_loss": 0.02173096314072609, "grad_norm": 0.07929706573486328, "global_step": 2284, "epoch": 21, "lr": 9.954207285296235e-05} +{"train_loss": 0.022186879068613052, "grad_norm": 0.050581518560647964, "global_step": 2285, "epoch": 21, "lr": 9.954167256735252e-05} +{"train_loss": 0.020255599170923233, "grad_norm": 0.05408649891614914, "global_step": 2286, "epoch": 21, "lr": 9.954127210767479e-05} +{"train_loss": 0.023863602429628372, "grad_norm": 0.08680124580860138, "global_step": 2287, "epoch": 21, "lr": 9.954087147393056e-05} +{"train_loss": 0.022423334419727325, "grad_norm": 0.09357015788555145, "global_step": 2288, "epoch": 21, "lr": 9.954047066612127e-05} +{"train_loss": 0.020393576472997665, "grad_norm": 0.08261386305093765, "global_step": 2289, "epoch": 21, "lr": 9.954006968424832e-05} +{"train_loss": 0.02176240086555481, "grad_norm": 0.06443318724632263, "global_step": 2290, "epoch": 21, "lr": 9.95396685283131e-05} +{"train_loss": 0.02161400578916073, "grad_norm": 0.06677649170160294, "global_step": 2291, "epoch": 21, "lr": 9.9539267198317e-05} +{"train_loss": 0.020904380828142166, "grad_norm": 0.0775047019124031, "global_step": 2292, "epoch": 21, "lr": 9.95388656942615e-05} +{"train_loss": 0.02157672867178917, "grad_norm": 0.05168421193957329, "global_step": 2293, "epoch": 21, "lr": 9.953846401614794e-05} +{"train_loss": 0.02106522023677826, "grad_norm": 0.05888151749968529, "global_step": 2294, "epoch": 21, "lr": 9.953806216397779e-05} +{"train_loss": 0.02236373908817768, "grad_norm": 0.07255431264638901, "global_step": 2295, "epoch": 21, "lr": 9.953766013775241e-05} +{"train_loss": 0.0236748605966568, "grad_norm": 0.10100382566452026, "global_step": 2296, "epoch": 21, "lr": 9.953725793747324e-05} +{"train_loss": 0.020432578399777412, "grad_norm": 0.08120553195476532, "global_step": 2297, "epoch": 21, "lr": 9.95368555631417e-05} +{"train_loss": 0.02084822952747345, "grad_norm": 0.07725872844457626, "global_step": 2298, "epoch": 21, "lr": 9.953645301475917e-05} +{"train_loss": 0.020735565572977066, "grad_norm": 0.0556473582983017, "global_step": 2299, "epoch": 21, "lr": 9.953605029232711e-05} +{"train_loss": 0.02154623530805111, "grad_norm": 0.14954589307308197, "global_step": 2300, "epoch": 21, "lr": 9.95356473958469e-05} +{"train_loss": 0.02039310336112976, "grad_norm": 0.12117213010787964, "global_step": 2301, "epoch": 21, "lr": 9.953524432531997e-05} +{"train_loss": 0.020930349826812744, "grad_norm": 0.06623899936676025, "global_step": 2302, "epoch": 21, "lr": 9.953484108074773e-05} +{"train_loss": 0.021380243822932243, "grad_norm": 0.09351852536201477, "global_step": 2303, "epoch": 21, "lr": 9.953443766213162e-05} +{"train_loss": 0.0251375213265419, "grad_norm": 0.11496824026107788, "global_step": 2304, "epoch": 21, "lr": 9.953403406947302e-05} +{"train_loss": 0.022006355226039886, "grad_norm": 0.13411177694797516, "global_step": 2305, "epoch": 21, "lr": 9.953363030277336e-05} +{"train_loss": 0.022236190736293793, "grad_norm": 0.08167151361703873, "global_step": 2306, "epoch": 21, "lr": 9.953322636203408e-05} +{"train_loss": 0.022085312753915787, "grad_norm": 0.05069352313876152, "global_step": 2307, "epoch": 21, "lr": 9.953282224725658e-05} +{"train_loss": 0.02135048806667328, "grad_norm": 0.05561409890651703, "global_step": 2308, "epoch": 21, "lr": 9.953241795844228e-05} +{"train_loss": 0.019676733762025833, "grad_norm": 0.05556792765855789, "global_step": 2309, "epoch": 21, "lr": 9.95320134955926e-05} +{"train_loss": 0.021199874579906464, "grad_norm": 0.052491627633571625, "global_step": 2310, "epoch": 21, "lr": 9.953160885870898e-05} +{"train_loss": 0.021016206592321396, "grad_norm": 0.03685469552874565, "global_step": 2311, "epoch": 21, "lr": 9.953120404779282e-05} +{"train_loss": 0.02139001339673996, "grad_norm": 0.09224314987659454, "global_step": 2312, "epoch": 21, "lr": 9.953079906284554e-05} +{"train_loss": 0.022400110960006714, "grad_norm": 0.09836041927337646, "global_step": 2313, "epoch": 21, "lr": 9.953039390386858e-05} +{"train_loss": 0.02067875862121582, "grad_norm": 0.0573597326874733, "global_step": 2314, "epoch": 21, "lr": 9.952998857086337e-05} +{"train_loss": 0.020883547142148018, "grad_norm": 0.06492770463228226, "global_step": 2315, "epoch": 21, "lr": 9.95295830638313e-05} +{"train_loss": 0.02075626514852047, "grad_norm": 0.04195839539170265, "global_step": 2316, "epoch": 21, "lr": 9.952917738277382e-05} +{"train_loss": 0.02429543435573578, "grad_norm": 0.1572030931711197, "global_step": 2317, "epoch": 21, "lr": 9.952877152769236e-05} +{"train_loss": 0.019880421459674835, "grad_norm": 0.04750680923461914, "global_step": 2318, "epoch": 21, "lr": 9.952836549858833e-05} +{"train_loss": 0.02203461155295372, "grad_norm": 0.03883321210741997, "global_step": 2319, "epoch": 21, "lr": 9.952795929546314e-05} +{"train_loss": 0.021035144105553627, "grad_norm": 0.034546926617622375, "global_step": 2320, "epoch": 21, "lr": 9.952755291831828e-05} +{"train_loss": 0.021114371716976166, "grad_norm": 0.03457487002015114, "global_step": 2321, "epoch": 21, "lr": 9.952714636715512e-05} +{"train_loss": 0.020965419709682465, "grad_norm": 0.04961942508816719, "global_step": 2322, "epoch": 21, "lr": 9.95267396419751e-05} +{"train_loss": 0.01946842297911644, "grad_norm": 0.07199393212795258, "global_step": 2323, "epoch": 21, "lr": 9.952633274277967e-05} +{"train_loss": 0.01955379545688629, "grad_norm": 0.049159400165081024, "global_step": 2324, "epoch": 21, "lr": 9.952592566957023e-05} +{"train_loss": 0.020749196410179138, "grad_norm": 0.04682077839970589, "global_step": 2325, "epoch": 21, "lr": 9.952551842234823e-05} +{"train_loss": 0.019175682216882706, "grad_norm": 0.13142190873622894, "global_step": 2326, "epoch": 21, "lr": 9.95251110011151e-05} +{"train_loss": 0.02260805293917656, "grad_norm": 0.03873144090175629, "global_step": 2327, "epoch": 21, "lr": 9.952470340587228e-05} +{"train_loss": 0.02274204045534134, "grad_norm": 0.07743245363235474, "global_step": 2328, "epoch": 21, "lr": 9.952429563662117e-05} +{"train_loss": 0.019245730713009834, "grad_norm": 0.12246976792812347, "global_step": 2329, "epoch": 21, "lr": 9.952388769336322e-05} +{"train_loss": 0.023370465263724327, "grad_norm": 0.03564588725566864, "global_step": 2330, "epoch": 21, "lr": 9.952347957609989e-05} +{"train_loss": 0.021498477719021292, "grad_norm": 0.10899189114570618, "global_step": 2331, "epoch": 21, "lr": 9.952307128483256e-05, "val_loss": 0.02002161741256714} +{"train_loss": 0.021535947918891907, "grad_norm": 0.11746489256620407, "global_step": 2332, "epoch": 22, "lr": 9.952266281956272e-05} +{"train_loss": 0.018476366996765137, "grad_norm": 0.07779590040445328, "global_step": 2333, "epoch": 22, "lr": 9.952225418029176e-05} +{"train_loss": 0.02213764749467373, "grad_norm": 0.04095611348748207, "global_step": 2334, "epoch": 22, "lr": 9.952184536702115e-05} +{"train_loss": 0.01899205893278122, "grad_norm": 0.11259403824806213, "global_step": 2335, "epoch": 22, "lr": 9.95214363797523e-05} +{"train_loss": 0.023865992203354836, "grad_norm": 0.05253357067704201, "global_step": 2336, "epoch": 22, "lr": 9.952102721848667e-05} +{"train_loss": 0.019560053944587708, "grad_norm": 0.09229999035596848, "global_step": 2337, "epoch": 22, "lr": 9.952061788322569e-05} +{"train_loss": 0.02067575231194496, "grad_norm": 0.10988421738147736, "global_step": 2338, "epoch": 22, "lr": 9.952020837397078e-05} +{"train_loss": 0.022050632163882256, "grad_norm": 0.06921028345823288, "global_step": 2339, "epoch": 22, "lr": 9.951979869072342e-05} +{"train_loss": 0.02131480537354946, "grad_norm": 0.053621917963027954, "global_step": 2340, "epoch": 22, "lr": 9.951938883348499e-05} +{"train_loss": 0.024005860090255737, "grad_norm": 0.15875481069087982, "global_step": 2341, "epoch": 22, "lr": 9.951897880225698e-05} +{"train_loss": 0.02104686200618744, "grad_norm": 0.13533726334571838, "global_step": 2342, "epoch": 22, "lr": 9.951856859704081e-05} +{"train_loss": 0.019482744857668877, "grad_norm": 0.11974000185728073, "global_step": 2343, "epoch": 22, "lr": 9.951815821783792e-05} +{"train_loss": 0.022868501022458076, "grad_norm": 0.05920068919658661, "global_step": 2344, "epoch": 22, "lr": 9.951774766464976e-05} +{"train_loss": 0.018834341317415237, "grad_norm": 0.08313986659049988, "global_step": 2345, "epoch": 22, "lr": 9.951733693747777e-05} +{"train_loss": 0.022395581007003784, "grad_norm": 0.05557834729552269, "global_step": 2346, "epoch": 22, "lr": 9.951692603632339e-05} +{"train_loss": 0.020820219069719315, "grad_norm": 0.05057801678776741, "global_step": 2347, "epoch": 22, "lr": 9.951651496118807e-05} +{"train_loss": 0.02409747615456581, "grad_norm": 0.1172969713807106, "global_step": 2348, "epoch": 22, "lr": 9.951610371207325e-05} +{"train_loss": 0.024144835770130157, "grad_norm": 0.13554157316684723, "global_step": 2349, "epoch": 22, "lr": 9.951569228898039e-05} +{"train_loss": 0.021014872938394547, "grad_norm": 0.04230582341551781, "global_step": 2350, "epoch": 22, "lr": 9.951528069191089e-05} +{"train_loss": 0.02020549774169922, "grad_norm": 0.03693293035030365, "global_step": 2351, "epoch": 22, "lr": 9.951486892086626e-05} +{"train_loss": 0.023320525884628296, "grad_norm": 0.11055229604244232, "global_step": 2352, "epoch": 22, "lr": 9.951445697584789e-05} +{"train_loss": 0.021324925124645233, "grad_norm": 0.10850818455219269, "global_step": 2353, "epoch": 22, "lr": 9.951404485685727e-05} +{"train_loss": 0.018857937306165695, "grad_norm": 0.0410800538957119, "global_step": 2354, "epoch": 22, "lr": 9.95136325638958e-05} +{"train_loss": 0.021309923380613327, "grad_norm": 0.08792617917060852, "global_step": 2355, "epoch": 22, "lr": 9.9513220096965e-05} +{"train_loss": 0.02126466855406761, "grad_norm": 0.049625955522060394, "global_step": 2356, "epoch": 22, "lr": 9.951280745606625e-05} +{"train_loss": 0.021939873695373535, "grad_norm": 0.07681997865438461, "global_step": 2357, "epoch": 22, "lr": 9.951239464120103e-05} +{"train_loss": 0.022349629551172256, "grad_norm": 0.04883895441889763, "global_step": 2358, "epoch": 22, "lr": 9.951198165237079e-05} +{"train_loss": 0.019214173778891563, "grad_norm": 0.03952382504940033, "global_step": 2359, "epoch": 22, "lr": 9.951156848957698e-05} +{"train_loss": 0.02041042037308216, "grad_norm": 0.038871388882398605, "global_step": 2360, "epoch": 22, "lr": 9.951115515282104e-05} +{"train_loss": 0.020511282607913017, "grad_norm": 0.07879314571619034, "global_step": 2361, "epoch": 22, "lr": 9.951074164210446e-05} +{"train_loss": 0.02132938802242279, "grad_norm": 0.04057887941598892, "global_step": 2362, "epoch": 22, "lr": 9.951032795742863e-05} +{"train_loss": 0.022016305476427078, "grad_norm": 0.06259641796350479, "global_step": 2363, "epoch": 22, "lr": 9.950991409879507e-05} +{"train_loss": 0.019808277487754822, "grad_norm": 0.13658849895000458, "global_step": 2364, "epoch": 22, "lr": 9.950950006620519e-05} +{"train_loss": 0.023619774729013443, "grad_norm": 0.10347060859203339, "global_step": 2365, "epoch": 22, "lr": 9.950908585966046e-05} +{"train_loss": 0.02233130857348442, "grad_norm": 0.042452871799468994, "global_step": 2366, "epoch": 22, "lr": 9.950867147916232e-05} +{"train_loss": 0.018467646092176437, "grad_norm": 0.08833590894937515, "global_step": 2367, "epoch": 22, "lr": 9.950825692471226e-05} +{"train_loss": 0.023534627631306648, "grad_norm": 0.04939134046435356, "global_step": 2368, "epoch": 22, "lr": 9.95078421963117e-05} +{"train_loss": 0.01964426040649414, "grad_norm": 0.09047399461269379, "global_step": 2369, "epoch": 22, "lr": 9.950742729396212e-05} +{"train_loss": 0.020932309329509735, "grad_norm": 0.03167285397648811, "global_step": 2370, "epoch": 22, "lr": 9.950701221766496e-05} +{"train_loss": 0.022950097918510437, "grad_norm": 0.06931457668542862, "global_step": 2371, "epoch": 22, "lr": 9.950659696742171e-05} +{"train_loss": 0.021022863686084747, "grad_norm": 0.07595048099756241, "global_step": 2372, "epoch": 22, "lr": 9.95061815432338e-05} +{"train_loss": 0.02554735727608204, "grad_norm": 0.06418467313051224, "global_step": 2373, "epoch": 22, "lr": 9.950576594510268e-05} +{"train_loss": 0.02266007289290428, "grad_norm": 0.0850628986954689, "global_step": 2374, "epoch": 22, "lr": 9.950535017302985e-05} +{"train_loss": 0.01981067657470703, "grad_norm": 0.054269079118967056, "global_step": 2375, "epoch": 22, "lr": 9.950493422701674e-05} +{"train_loss": 0.021545160561800003, "grad_norm": 0.035011161118745804, "global_step": 2376, "epoch": 22, "lr": 9.950451810706481e-05} +{"train_loss": 0.020362449809908867, "grad_norm": 0.06232622638344765, "global_step": 2377, "epoch": 22, "lr": 9.950410181317554e-05} +{"train_loss": 0.019337834790349007, "grad_norm": 0.06390857696533203, "global_step": 2378, "epoch": 22, "lr": 9.950368534535039e-05} +{"train_loss": 0.01837928779423237, "grad_norm": 0.0611569881439209, "global_step": 2379, "epoch": 22, "lr": 9.950326870359082e-05} +{"train_loss": 0.019288845360279083, "grad_norm": 0.052435312420129776, "global_step": 2380, "epoch": 22, "lr": 9.950285188789829e-05} +{"train_loss": 0.020670384168624878, "grad_norm": 0.05002478510141373, "global_step": 2381, "epoch": 22, "lr": 9.950243489827426e-05} +{"train_loss": 0.021686777472496033, "grad_norm": 0.037451911717653275, "global_step": 2382, "epoch": 22, "lr": 9.950201773472021e-05} +{"train_loss": 0.024632040411233902, "grad_norm": 0.038424767553806305, "global_step": 2383, "epoch": 22, "lr": 9.950160039723759e-05} +{"train_loss": 0.02383962832391262, "grad_norm": 0.05159126967191696, "global_step": 2384, "epoch": 22, "lr": 9.950118288582788e-05} +{"train_loss": 0.02347722090780735, "grad_norm": 0.04632199555635452, "global_step": 2385, "epoch": 22, "lr": 9.950076520049253e-05} +{"train_loss": 0.018564995378255844, "grad_norm": 0.03829772025346756, "global_step": 2386, "epoch": 22, "lr": 9.950034734123304e-05} +{"train_loss": 0.020717451348900795, "grad_norm": 0.05615729093551636, "global_step": 2387, "epoch": 22, "lr": 9.949992930805084e-05} +{"train_loss": 0.020280543714761734, "grad_norm": 0.04423658922314644, "global_step": 2388, "epoch": 22, "lr": 9.949951110094742e-05} +{"train_loss": 0.022162875160574913, "grad_norm": 0.07293669134378433, "global_step": 2389, "epoch": 22, "lr": 9.949909271992424e-05} +{"train_loss": 0.023106057196855545, "grad_norm": 0.03888897970318794, "global_step": 2390, "epoch": 22, "lr": 9.949867416498278e-05} +{"train_loss": 0.018939102068543434, "grad_norm": 0.06513477861881256, "global_step": 2391, "epoch": 22, "lr": 9.949825543612451e-05} +{"train_loss": 0.019083183258771896, "grad_norm": 0.0610508993268013, "global_step": 2392, "epoch": 22, "lr": 9.949783653335089e-05} +{"train_loss": 0.02582092210650444, "grad_norm": 0.04860829934477806, "global_step": 2393, "epoch": 22, "lr": 9.949741745666341e-05} +{"train_loss": 0.022214729338884354, "grad_norm": 0.06253954023122787, "global_step": 2394, "epoch": 22, "lr": 9.949699820606351e-05} +{"train_loss": 0.021622970700263977, "grad_norm": 0.08069610595703125, "global_step": 2395, "epoch": 22, "lr": 9.94965787815527e-05} +{"train_loss": 0.019249899312853813, "grad_norm": 0.08089891076087952, "global_step": 2396, "epoch": 22, "lr": 9.949615918313243e-05} +{"train_loss": 0.018922336399555206, "grad_norm": 0.048142336308956146, "global_step": 2397, "epoch": 22, "lr": 9.949573941080419e-05} +{"train_loss": 0.018475104123353958, "grad_norm": 0.07819687575101852, "global_step": 2398, "epoch": 22, "lr": 9.949531946456943e-05} +{"train_loss": 0.020283423364162445, "grad_norm": 0.06236148998141289, "global_step": 2399, "epoch": 22, "lr": 9.949489934442966e-05} +{"train_loss": 0.025494877249002457, "grad_norm": 0.046585239470005035, "global_step": 2400, "epoch": 22, "lr": 9.949447905038633e-05} +{"train_loss": 0.019146980717778206, "grad_norm": 0.032492175698280334, "global_step": 2401, "epoch": 22, "lr": 9.949405858244093e-05} +{"train_loss": 0.020387252792716026, "grad_norm": 0.031054968014359474, "global_step": 2402, "epoch": 22, "lr": 9.949363794059493e-05} +{"train_loss": 0.024592477828264236, "grad_norm": 0.05769604444503784, "global_step": 2403, "epoch": 22, "lr": 9.94932171248498e-05} +{"train_loss": 0.02112307772040367, "grad_norm": 0.044397804886102676, "global_step": 2404, "epoch": 22, "lr": 9.949279613520705e-05} +{"train_loss": 0.020884908735752106, "grad_norm": 0.03186195343732834, "global_step": 2405, "epoch": 22, "lr": 9.949237497166812e-05} +{"train_loss": 0.0235214252024889, "grad_norm": 0.0769546777009964, "global_step": 2406, "epoch": 22, "lr": 9.949195363423451e-05} +{"train_loss": 0.0199893768876791, "grad_norm": 0.0769309252500534, "global_step": 2407, "epoch": 22, "lr": 9.949153212290772e-05} +{"train_loss": 0.02336305007338524, "grad_norm": 0.044920992106199265, "global_step": 2408, "epoch": 22, "lr": 9.94911104376892e-05} +{"train_loss": 0.019317347556352615, "grad_norm": 0.05032491311430931, "global_step": 2409, "epoch": 22, "lr": 9.949068857858041e-05} +{"train_loss": 0.023923518136143684, "grad_norm": 0.03863130882382393, "global_step": 2410, "epoch": 22, "lr": 9.94902665455829e-05} +{"train_loss": 0.018964719027280807, "grad_norm": 0.04590725153684616, "global_step": 2411, "epoch": 22, "lr": 9.94898443386981e-05} +{"train_loss": 0.01960773952305317, "grad_norm": 0.05991603061556816, "global_step": 2412, "epoch": 22, "lr": 9.948942195792752e-05} +{"train_loss": 0.017871031537652016, "grad_norm": 0.035362787544727325, "global_step": 2413, "epoch": 22, "lr": 9.948899940327262e-05} +{"train_loss": 0.02154449373483658, "grad_norm": 0.10770215839147568, "global_step": 2414, "epoch": 22, "lr": 9.94885766747349e-05} +{"train_loss": 0.020788202062249184, "grad_norm": 0.04697920382022858, "global_step": 2415, "epoch": 22, "lr": 9.948815377231586e-05} +{"train_loss": 0.025025933980941772, "grad_norm": 0.0748802050948143, "global_step": 2416, "epoch": 22, "lr": 9.948773069601696e-05} +{"train_loss": 0.02358832396566868, "grad_norm": 0.054543618112802505, "global_step": 2417, "epoch": 22, "lr": 9.948730744583969e-05} +{"train_loss": 0.021575216203927994, "grad_norm": 0.04269036278128624, "global_step": 2418, "epoch": 22, "lr": 9.948688402178555e-05} +{"train_loss": 0.02131241001188755, "grad_norm": 0.05706379935145378, "global_step": 2419, "epoch": 22, "lr": 9.948646042385601e-05} +{"train_loss": 0.022499438375234604, "grad_norm": 0.04836146533489227, "global_step": 2420, "epoch": 22, "lr": 9.948603665205258e-05} +{"train_loss": 0.0218706876039505, "grad_norm": 0.04330996796488762, "global_step": 2421, "epoch": 22, "lr": 9.948561270637672e-05} +{"train_loss": 0.018885336816310883, "grad_norm": 0.0613839328289032, "global_step": 2422, "epoch": 22, "lr": 9.948518858682996e-05} +{"train_loss": 0.018438192084431648, "grad_norm": 0.04653596505522728, "global_step": 2423, "epoch": 22, "lr": 9.948476429341377e-05} +{"train_loss": 0.02089637890458107, "grad_norm": 0.06741851568222046, "global_step": 2424, "epoch": 22, "lr": 9.948433982612964e-05} +{"train_loss": 0.025898272171616554, "grad_norm": 0.15826261043548584, "global_step": 2425, "epoch": 22, "lr": 9.948391518497904e-05} +{"train_loss": 0.022632351145148277, "grad_norm": 0.05526113137602806, "global_step": 2426, "epoch": 22, "lr": 9.948349036996349e-05} +{"train_loss": 0.020122140645980835, "grad_norm": 0.05342938005924225, "global_step": 2427, "epoch": 22, "lr": 9.948306538108447e-05} +{"train_loss": 0.02264564111828804, "grad_norm": 0.06917301565408707, "global_step": 2428, "epoch": 22, "lr": 9.948264021834348e-05} +{"train_loss": 0.02159591019153595, "grad_norm": 0.04076047241687775, "global_step": 2429, "epoch": 22, "lr": 9.948221488174203e-05} +{"train_loss": 0.024196622893214226, "grad_norm": 0.05924738198518753, "global_step": 2430, "epoch": 22, "lr": 9.948178937128157e-05} +{"train_loss": 0.019918259233236313, "grad_norm": 0.03881635516881943, "global_step": 2431, "epoch": 22, "lr": 9.948136368696363e-05} +{"train_loss": 0.019917603582143784, "grad_norm": 0.029528293758630753, "global_step": 2432, "epoch": 22, "lr": 9.948093782878971e-05} +{"train_loss": 0.021085578948259354, "grad_norm": 0.0678006038069725, "global_step": 2433, "epoch": 22, "lr": 9.948051179676128e-05} +{"train_loss": 0.020443618297576904, "grad_norm": 0.03850068524479866, "global_step": 2434, "epoch": 22, "lr": 9.948008559087986e-05} +{"train_loss": 0.02055034041404724, "grad_norm": 0.09753697365522385, "global_step": 2435, "epoch": 22, "lr": 9.947965921114692e-05} +{"train_loss": 0.021347109228372574, "grad_norm": 0.03294513374567032, "global_step": 2436, "epoch": 22, "lr": 9.947923265756399e-05} +{"train_loss": 0.02132287530123063, "grad_norm": 0.08854151517152786, "global_step": 2437, "epoch": 22, "lr": 9.947880593013255e-05, "val_loss": 0.02049664966762066} +{"train_loss": 0.022251568734645844, "grad_norm": 0.04877713695168495, "global_step": 2438, "epoch": 23, "lr": 9.94783790288541e-05} +{"train_loss": 0.020783934742212296, "grad_norm": 0.050639644265174866, "global_step": 2439, "epoch": 23, "lr": 9.947795195373016e-05} +{"train_loss": 0.017479751259088516, "grad_norm": 0.09282823652029037, "global_step": 2440, "epoch": 23, "lr": 9.947752470476219e-05} +{"train_loss": 0.02187451347708702, "grad_norm": 0.045803170651197433, "global_step": 2441, "epoch": 23, "lr": 9.947709728195174e-05} +{"train_loss": 0.021226853132247925, "grad_norm": 0.06116924062371254, "global_step": 2442, "epoch": 23, "lr": 9.947666968530028e-05} +{"train_loss": 0.020698949694633484, "grad_norm": 0.03687307983636856, "global_step": 2443, "epoch": 23, "lr": 9.94762419148093e-05} +{"train_loss": 0.022544577717781067, "grad_norm": 0.07985293120145798, "global_step": 2444, "epoch": 23, "lr": 9.947581397048034e-05} +{"train_loss": 0.019696395844221115, "grad_norm": 0.07263512164354324, "global_step": 2445, "epoch": 23, "lr": 9.947538585231488e-05} +{"train_loss": 0.022325482219457626, "grad_norm": 0.1108320951461792, "global_step": 2446, "epoch": 23, "lr": 9.947495756031444e-05} +{"train_loss": 0.02136686071753502, "grad_norm": 0.04143299534916878, "global_step": 2447, "epoch": 23, "lr": 9.947452909448051e-05} +{"train_loss": 0.02353302389383316, "grad_norm": 0.05429229885339737, "global_step": 2448, "epoch": 23, "lr": 9.94741004548146e-05} +{"train_loss": 0.022029850631952286, "grad_norm": 0.0642876923084259, "global_step": 2449, "epoch": 23, "lr": 9.947367164131822e-05} +{"train_loss": 0.02392711490392685, "grad_norm": 0.0602043978869915, "global_step": 2450, "epoch": 23, "lr": 9.947324265399287e-05} +{"train_loss": 0.02162821590900421, "grad_norm": 0.07405795156955719, "global_step": 2451, "epoch": 23, "lr": 9.947281349284005e-05} +{"train_loss": 0.01974708028137684, "grad_norm": 0.13526949286460876, "global_step": 2452, "epoch": 23, "lr": 9.947238415786128e-05} +{"train_loss": 0.024690909311175346, "grad_norm": 0.13778777420520782, "global_step": 2453, "epoch": 23, "lr": 9.94719546490581e-05} +{"train_loss": 0.023135501891374588, "grad_norm": 0.14063721895217896, "global_step": 2454, "epoch": 23, "lr": 9.947152496643195e-05} +{"train_loss": 0.020570840686559677, "grad_norm": 0.1525055468082428, "global_step": 2455, "epoch": 23, "lr": 9.947109510998439e-05} +{"train_loss": 0.02391510270535946, "grad_norm": 0.0424097515642643, "global_step": 2456, "epoch": 23, "lr": 9.947066507971692e-05} +{"train_loss": 0.021110430359840393, "grad_norm": 0.05219574272632599, "global_step": 2457, "epoch": 23, "lr": 9.947023487563102e-05} +{"train_loss": 0.02061743661761284, "grad_norm": 0.06830611079931259, "global_step": 2458, "epoch": 23, "lr": 9.946980449772827e-05} +{"train_loss": 0.021439433097839355, "grad_norm": 0.05375593900680542, "global_step": 2459, "epoch": 23, "lr": 9.946937394601011e-05} +{"train_loss": 0.02022535726428032, "grad_norm": 0.04950535297393799, "global_step": 2460, "epoch": 23, "lr": 9.946894322047811e-05} +{"train_loss": 0.020609576255083084, "grad_norm": 0.16923508048057556, "global_step": 2461, "epoch": 23, "lr": 9.946851232113372e-05} +{"train_loss": 0.02042899653315544, "grad_norm": 0.05284879356622696, "global_step": 2462, "epoch": 23, "lr": 9.946808124797852e-05} +{"train_loss": 0.019094757735729218, "grad_norm": 0.04238734394311905, "global_step": 2463, "epoch": 23, "lr": 9.946765000101399e-05} +{"train_loss": 0.02052253484725952, "grad_norm": 0.03612101823091507, "global_step": 2464, "epoch": 23, "lr": 9.946721858024165e-05} +{"train_loss": 0.01927754655480385, "grad_norm": 0.06899973005056381, "global_step": 2465, "epoch": 23, "lr": 9.946678698566301e-05} +{"train_loss": 0.022124096751213074, "grad_norm": 0.07264725863933563, "global_step": 2466, "epoch": 23, "lr": 9.946635521727961e-05} +{"train_loss": 0.02184511348605156, "grad_norm": 0.05444558709859848, "global_step": 2467, "epoch": 23, "lr": 9.946592327509293e-05} +{"train_loss": 0.020357714965939522, "grad_norm": 0.03622762858867645, "global_step": 2468, "epoch": 23, "lr": 9.946549115910451e-05} +{"train_loss": 0.019717112183570862, "grad_norm": 0.04770452529191971, "global_step": 2469, "epoch": 23, "lr": 9.946505886931587e-05} +{"train_loss": 0.02000681683421135, "grad_norm": 0.04612519592046738, "global_step": 2470, "epoch": 23, "lr": 9.946462640572854e-05} +{"train_loss": 0.02217097207903862, "grad_norm": 0.04529378563165665, "global_step": 2471, "epoch": 23, "lr": 9.946419376834402e-05} +{"train_loss": 0.02175014466047287, "grad_norm": 0.06692340224981308, "global_step": 2472, "epoch": 23, "lr": 9.946376095716381e-05} +{"train_loss": 0.017529163509607315, "grad_norm": 0.04020221158862114, "global_step": 2473, "epoch": 23, "lr": 9.946332797218947e-05} +{"train_loss": 0.023149292916059494, "grad_norm": 0.04276951029896736, "global_step": 2474, "epoch": 23, "lr": 9.946289481342251e-05} +{"train_loss": 0.019259369000792503, "grad_norm": 0.062026288360357285, "global_step": 2475, "epoch": 23, "lr": 9.946246148086444e-05} +{"train_loss": 0.0259847491979599, "grad_norm": 0.05723291635513306, "global_step": 2476, "epoch": 23, "lr": 9.94620279745168e-05} +{"train_loss": 0.018136171624064445, "grad_norm": 0.08432953804731369, "global_step": 2477, "epoch": 23, "lr": 9.946159429438111e-05} +{"train_loss": 0.02125491574406624, "grad_norm": 0.07253145426511765, "global_step": 2478, "epoch": 23, "lr": 9.946116044045888e-05} +{"train_loss": 0.018039710819721222, "grad_norm": 0.057332854717969894, "global_step": 2479, "epoch": 23, "lr": 9.946072641275163e-05} +{"train_loss": 0.020820684731006622, "grad_norm": 0.04188027232885361, "global_step": 2480, "epoch": 23, "lr": 9.946029221126093e-05} +{"train_loss": 0.020372092723846436, "grad_norm": 0.09642012417316437, "global_step": 2481, "epoch": 23, "lr": 9.945985783598824e-05} +{"train_loss": 0.02278401330113411, "grad_norm": 0.09177408367395401, "global_step": 2482, "epoch": 23, "lr": 9.945942328693514e-05} +{"train_loss": 0.02006029710173607, "grad_norm": 0.0809585377573967, "global_step": 2483, "epoch": 23, "lr": 9.945898856410312e-05} +{"train_loss": 0.02086351439356804, "grad_norm": 0.06364530324935913, "global_step": 2484, "epoch": 23, "lr": 9.945855366749372e-05} +{"train_loss": 0.0194068793207407, "grad_norm": 0.05258730426430702, "global_step": 2485, "epoch": 23, "lr": 9.94581185971085e-05} +{"train_loss": 0.0182524211704731, "grad_norm": 0.085689015686512, "global_step": 2486, "epoch": 23, "lr": 9.945768335294894e-05} +{"train_loss": 0.021993126720190048, "grad_norm": 0.0928799957036972, "global_step": 2487, "epoch": 23, "lr": 9.945724793501659e-05} +{"train_loss": 0.022601306438446045, "grad_norm": 0.08055955916643143, "global_step": 2488, "epoch": 23, "lr": 9.945681234331298e-05} +{"train_loss": 0.021045736968517303, "grad_norm": 0.06369929760694504, "global_step": 2489, "epoch": 23, "lr": 9.945637657783964e-05} +{"train_loss": 0.02199692279100418, "grad_norm": 0.03571292385458946, "global_step": 2490, "epoch": 23, "lr": 9.945594063859809e-05} +{"train_loss": 0.021217621862888336, "grad_norm": 0.10038125514984131, "global_step": 2491, "epoch": 23, "lr": 9.945550452558989e-05} +{"train_loss": 0.01992885395884514, "grad_norm": 0.05409487709403038, "global_step": 2492, "epoch": 23, "lr": 9.945506823881654e-05} +{"train_loss": 0.022159487009048462, "grad_norm": 0.11723460257053375, "global_step": 2493, "epoch": 23, "lr": 9.945463177827961e-05} +{"train_loss": 0.02025487646460533, "grad_norm": 0.04647327587008476, "global_step": 2494, "epoch": 23, "lr": 9.945419514398059e-05} +{"train_loss": 0.018989412114024162, "grad_norm": 0.07742714881896973, "global_step": 2495, "epoch": 23, "lr": 9.945375833592103e-05} +{"train_loss": 0.020853642374277115, "grad_norm": 0.05172957107424736, "global_step": 2496, "epoch": 23, "lr": 9.945332135410249e-05} +{"train_loss": 0.021550044417381287, "grad_norm": 0.14419692754745483, "global_step": 2497, "epoch": 23, "lr": 9.945288419852647e-05} +{"train_loss": 0.019455065950751305, "grad_norm": 0.09186358749866486, "global_step": 2498, "epoch": 23, "lr": 9.945244686919454e-05} +{"train_loss": 0.023003239184617996, "grad_norm": 0.09514842927455902, "global_step": 2499, "epoch": 23, "lr": 9.94520093661082e-05} +{"train_loss": 0.019356336444616318, "grad_norm": 0.03659888356924057, "global_step": 2500, "epoch": 23, "lr": 9.945157168926902e-05} +{"train_loss": 0.02030562236905098, "grad_norm": 0.06759433448314667, "global_step": 2501, "epoch": 23, "lr": 9.945113383867851e-05} +{"train_loss": 0.02343769371509552, "grad_norm": 0.08674252778291702, "global_step": 2502, "epoch": 23, "lr": 9.945069581433824e-05} +{"train_loss": 0.01970771886408329, "grad_norm": 0.07122616469860077, "global_step": 2503, "epoch": 23, "lr": 9.945025761624971e-05} +{"train_loss": 0.023520950227975845, "grad_norm": 0.03928777575492859, "global_step": 2504, "epoch": 23, "lr": 9.944981924441448e-05} +{"train_loss": 0.023639651015400887, "grad_norm": 0.04086265712976456, "global_step": 2505, "epoch": 23, "lr": 9.94493806988341e-05} +{"train_loss": 0.01988917402923107, "grad_norm": 0.052833955734968185, "global_step": 2506, "epoch": 23, "lr": 9.944894197951009e-05} +{"train_loss": 0.02082645706832409, "grad_norm": 0.09221156686544418, "global_step": 2507, "epoch": 23, "lr": 9.944850308644402e-05} +{"train_loss": 0.018895331770181656, "grad_norm": 0.028424054384231567, "global_step": 2508, "epoch": 23, "lr": 9.94480640196374e-05} +{"train_loss": 0.018430929630994797, "grad_norm": 0.07558044046163559, "global_step": 2509, "epoch": 23, "lr": 9.94476247790918e-05} +{"train_loss": 0.023549089208245277, "grad_norm": 0.04436073079705238, "global_step": 2510, "epoch": 23, "lr": 9.944718536480875e-05} +{"train_loss": 0.020452246069908142, "grad_norm": 0.037482358515262604, "global_step": 2511, "epoch": 23, "lr": 9.944674577678978e-05} +{"train_loss": 0.022010741755366325, "grad_norm": 0.050979189574718475, "global_step": 2512, "epoch": 23, "lr": 9.944630601503647e-05} +{"train_loss": 0.024159597232937813, "grad_norm": 0.10523123294115067, "global_step": 2513, "epoch": 23, "lr": 9.944586607955032e-05} +{"train_loss": 0.022014815360307693, "grad_norm": 0.07925613969564438, "global_step": 2514, "epoch": 23, "lr": 9.944542597033291e-05} +{"train_loss": 0.019743887707591057, "grad_norm": 0.0553295835852623, "global_step": 2515, "epoch": 23, "lr": 9.944498568738579e-05} +{"train_loss": 0.02100776508450508, "grad_norm": 0.049031008034944534, "global_step": 2516, "epoch": 23, "lr": 9.944454523071047e-05} +{"train_loss": 0.02352210134267807, "grad_norm": 0.04007204622030258, "global_step": 2517, "epoch": 23, "lr": 9.944410460030854e-05} +{"train_loss": 0.02365981601178646, "grad_norm": 0.11499827355146408, "global_step": 2518, "epoch": 23, "lr": 9.94436637961815e-05} +{"train_loss": 0.018642358481884003, "grad_norm": 0.04916512966156006, "global_step": 2519, "epoch": 23, "lr": 9.944322281833096e-05} +{"train_loss": 0.018479984253644943, "grad_norm": 0.06099230423569679, "global_step": 2520, "epoch": 23, "lr": 9.944278166675843e-05} +{"train_loss": 0.019930534064769745, "grad_norm": 0.0520826019346714, "global_step": 2521, "epoch": 23, "lr": 9.944234034146547e-05} +{"train_loss": 0.020189810544252396, "grad_norm": 0.0335635244846344, "global_step": 2522, "epoch": 23, "lr": 9.944189884245362e-05} +{"train_loss": 0.02050022780895233, "grad_norm": 0.037190813571214676, "global_step": 2523, "epoch": 23, "lr": 9.944145716972444e-05} +{"train_loss": 0.021865593269467354, "grad_norm": 0.05339933559298515, "global_step": 2524, "epoch": 23, "lr": 9.944101532327948e-05} +{"train_loss": 0.021102096885442734, "grad_norm": 0.07587388902902603, "global_step": 2525, "epoch": 23, "lr": 9.94405733031203e-05} +{"train_loss": 0.01970050483942032, "grad_norm": 0.05088037624955177, "global_step": 2526, "epoch": 23, "lr": 9.944013110924844e-05} +{"train_loss": 0.024273330345749855, "grad_norm": 0.09992888569831848, "global_step": 2527, "epoch": 23, "lr": 9.943968874166547e-05} +{"train_loss": 0.022408276796340942, "grad_norm": 0.0399755984544754, "global_step": 2528, "epoch": 23, "lr": 9.943924620037292e-05} +{"train_loss": 0.019728511571884155, "grad_norm": 0.04358068108558655, "global_step": 2529, "epoch": 23, "lr": 9.943880348537237e-05} +{"train_loss": 0.018560029566287994, "grad_norm": 0.04989796131849289, "global_step": 2530, "epoch": 23, "lr": 9.943836059666535e-05} +{"train_loss": 0.021265776827931404, "grad_norm": 0.06038470193743706, "global_step": 2531, "epoch": 23, "lr": 9.943791753425344e-05} +{"train_loss": 0.02331918478012085, "grad_norm": 0.07687874883413315, "global_step": 2532, "epoch": 23, "lr": 9.94374742981382e-05} +{"train_loss": 0.021019868552684784, "grad_norm": 0.042250435799360275, "global_step": 2533, "epoch": 23, "lr": 9.943703088832117e-05} +{"train_loss": 0.02677871659398079, "grad_norm": 0.06917628645896912, "global_step": 2534, "epoch": 23, "lr": 9.943658730480391e-05} +{"train_loss": 0.02599567547440529, "grad_norm": 0.1215357780456543, "global_step": 2535, "epoch": 23, "lr": 9.943614354758798e-05} +{"train_loss": 0.019757002592086792, "grad_norm": 0.04419731721282005, "global_step": 2536, "epoch": 23, "lr": 9.943569961667494e-05} +{"train_loss": 0.023719333112239838, "grad_norm": 0.08644542843103409, "global_step": 2537, "epoch": 23, "lr": 9.943525551206635e-05} +{"train_loss": 0.020192882046103477, "grad_norm": 0.030702287331223488, "global_step": 2538, "epoch": 23, "lr": 9.943481123376377e-05} +{"train_loss": 0.021172676235437393, "grad_norm": 0.04737000912427902, "global_step": 2539, "epoch": 23, "lr": 9.943436678176876e-05} +{"train_loss": 0.022021647542715073, "grad_norm": 0.11776676774024963, "global_step": 2540, "epoch": 23, "lr": 9.94339221560829e-05} +{"train_loss": 0.022889429703354836, "grad_norm": 0.03814760223031044, "global_step": 2541, "epoch": 23, "lr": 9.943347735670772e-05} +{"train_loss": 0.01930980384349823, "grad_norm": 0.03816545382142067, "global_step": 2542, "epoch": 23, "lr": 9.94330323836448e-05} +{"train_loss": 0.021215951948795678, "grad_norm": 0.08984411507844925, "global_step": 2543, "epoch": 23, "lr": 9.94325872368957e-05, "val_loss": 0.020069586113095284} +{"train_loss": 0.020631950348615646, "grad_norm": 0.050644487142562866, "global_step": 2544, "epoch": 24, "lr": 9.9432141916462e-05} +{"train_loss": 0.021776828914880753, "grad_norm": 0.06429476290941238, "global_step": 2545, "epoch": 24, "lr": 9.943169642234524e-05} +{"train_loss": 0.020196914672851562, "grad_norm": 0.08185148984193802, "global_step": 2546, "epoch": 24, "lr": 9.9431250754547e-05} +{"train_loss": 0.0197785422205925, "grad_norm": 0.04732302576303482, "global_step": 2547, "epoch": 24, "lr": 9.943080491306884e-05} +{"train_loss": 0.022498641163110733, "grad_norm": 0.06795818358659744, "global_step": 2548, "epoch": 24, "lr": 9.943035889791233e-05} +{"train_loss": 0.023861855268478394, "grad_norm": 0.057313404977321625, "global_step": 2549, "epoch": 24, "lr": 9.942991270907902e-05} +{"train_loss": 0.020814763382077217, "grad_norm": 0.04508647695183754, "global_step": 2550, "epoch": 24, "lr": 9.94294663465705e-05} +{"train_loss": 0.020554194226861, "grad_norm": 0.08549243211746216, "global_step": 2551, "epoch": 24, "lr": 9.942901981038834e-05} +{"train_loss": 0.02006256766617298, "grad_norm": 0.04873901978135109, "global_step": 2552, "epoch": 24, "lr": 9.942857310053411e-05} +{"train_loss": 0.01763547956943512, "grad_norm": 0.10431533306837082, "global_step": 2553, "epoch": 24, "lr": 9.942812621700936e-05} +{"train_loss": 0.024339783936738968, "grad_norm": 0.07380799204111099, "global_step": 2554, "epoch": 24, "lr": 9.942767915981566e-05} +{"train_loss": 0.021582255139946938, "grad_norm": 0.05224146693944931, "global_step": 2555, "epoch": 24, "lr": 9.94272319289546e-05} +{"train_loss": 0.024197984486818314, "grad_norm": 0.13155822455883026, "global_step": 2556, "epoch": 24, "lr": 9.942678452442776e-05} +{"train_loss": 0.02351951226592064, "grad_norm": 0.10440616309642792, "global_step": 2557, "epoch": 24, "lr": 9.942633694623667e-05} +{"train_loss": 0.024194305762648582, "grad_norm": 0.1063152402639389, "global_step": 2558, "epoch": 24, "lr": 9.942588919438292e-05} +{"train_loss": 0.020827453583478928, "grad_norm": 0.07536841183900833, "global_step": 2559, "epoch": 24, "lr": 9.942544126886812e-05} +{"train_loss": 0.023183999583125114, "grad_norm": 0.03283114731311798, "global_step": 2560, "epoch": 24, "lr": 9.942499316969379e-05} +{"train_loss": 0.020885556936264038, "grad_norm": 0.05541755259037018, "global_step": 2561, "epoch": 24, "lr": 9.942454489686152e-05} +{"train_loss": 0.021217379719018936, "grad_norm": 0.04077041894197464, "global_step": 2562, "epoch": 24, "lr": 9.942409645037292e-05} +{"train_loss": 0.021116673946380615, "grad_norm": 0.10019566118717194, "global_step": 2563, "epoch": 24, "lr": 9.942364783022952e-05} +{"train_loss": 0.02236451767385006, "grad_norm": 0.03629158064723015, "global_step": 2564, "epoch": 24, "lr": 9.942319903643294e-05} +{"train_loss": 0.022982832044363022, "grad_norm": 0.060838017612695694, "global_step": 2565, "epoch": 24, "lr": 9.942275006898472e-05} +{"train_loss": 0.02168595790863037, "grad_norm": 0.07521744072437286, "global_step": 2566, "epoch": 24, "lr": 9.942230092788644e-05} +{"train_loss": 0.02115616202354431, "grad_norm": 0.035345058888196945, "global_step": 2567, "epoch": 24, "lr": 9.942185161313968e-05} +{"train_loss": 0.020866066217422485, "grad_norm": 0.07385117560625076, "global_step": 2568, "epoch": 24, "lr": 9.942140212474604e-05} +{"train_loss": 0.022820690646767616, "grad_norm": 0.06354629993438721, "global_step": 2569, "epoch": 24, "lr": 9.94209524627071e-05} +{"train_loss": 0.02188771590590477, "grad_norm": 0.05572985112667084, "global_step": 2570, "epoch": 24, "lr": 9.94205026270244e-05} +{"train_loss": 0.021069588139653206, "grad_norm": 0.03566472604870796, "global_step": 2571, "epoch": 24, "lr": 9.942005261769955e-05} +{"train_loss": 0.02094220742583275, "grad_norm": 0.08541800081729889, "global_step": 2572, "epoch": 24, "lr": 9.941960243473416e-05} +{"train_loss": 0.022034909576177597, "grad_norm": 0.04159340634942055, "global_step": 2573, "epoch": 24, "lr": 9.941915207812974e-05} +{"train_loss": 0.021822545677423477, "grad_norm": 0.07062532007694244, "global_step": 2574, "epoch": 24, "lr": 9.941870154788793e-05} +{"train_loss": 0.018705835565924644, "grad_norm": 0.03965849429368973, "global_step": 2575, "epoch": 24, "lr": 9.941825084401028e-05} +{"train_loss": 0.022134296596050262, "grad_norm": 0.05836202949285507, "global_step": 2576, "epoch": 24, "lr": 9.94177999664984e-05} +{"train_loss": 0.01998041570186615, "grad_norm": 0.04087594896554947, "global_step": 2577, "epoch": 24, "lr": 9.941734891535386e-05} +{"train_loss": 0.019207414239645004, "grad_norm": 0.1113005131483078, "global_step": 2578, "epoch": 24, "lr": 9.941689769057824e-05} +{"train_loss": 0.018549282103776932, "grad_norm": 0.05146195739507675, "global_step": 2579, "epoch": 24, "lr": 9.941644629217315e-05} +{"train_loss": 0.020441334694623947, "grad_norm": 0.04516782984137535, "global_step": 2580, "epoch": 24, "lr": 9.941599472014014e-05} +{"train_loss": 0.02080237865447998, "grad_norm": 0.04368824139237404, "global_step": 2581, "epoch": 24, "lr": 9.941554297448083e-05} +{"train_loss": 0.020928360521793365, "grad_norm": 0.0376245342195034, "global_step": 2582, "epoch": 24, "lr": 9.941509105519677e-05} +{"train_loss": 0.020542331039905548, "grad_norm": 0.05215048789978027, "global_step": 2583, "epoch": 24, "lr": 9.94146389622896e-05} +{"train_loss": 0.02069322019815445, "grad_norm": 0.04342161864042282, "global_step": 2584, "epoch": 24, "lr": 9.941418669576085e-05} +{"train_loss": 0.018773220479488373, "grad_norm": 0.08864770829677582, "global_step": 2585, "epoch": 24, "lr": 9.941373425561215e-05} +{"train_loss": 0.023033438250422478, "grad_norm": 0.049241408705711365, "global_step": 2586, "epoch": 24, "lr": 9.941328164184509e-05} +{"train_loss": 0.022703256458044052, "grad_norm": 0.08554644882678986, "global_step": 2587, "epoch": 24, "lr": 9.941282885446124e-05} +{"train_loss": 0.02022716775536537, "grad_norm": 0.08231782168149948, "global_step": 2588, "epoch": 24, "lr": 9.941237589346218e-05} +{"train_loss": 0.020235102623701096, "grad_norm": 0.11286606639623642, "global_step": 2589, "epoch": 24, "lr": 9.941192275884954e-05} +{"train_loss": 0.020461678504943848, "grad_norm": 0.06014150753617287, "global_step": 2590, "epoch": 24, "lr": 9.941146945062489e-05} +{"train_loss": 0.019027069211006165, "grad_norm": 0.08539839088916779, "global_step": 2591, "epoch": 24, "lr": 9.941101596878981e-05} +{"train_loss": 0.01889597624540329, "grad_norm": 0.04874000325798988, "global_step": 2592, "epoch": 24, "lr": 9.941056231334592e-05} +{"train_loss": 0.023256804794073105, "grad_norm": 0.05670773610472679, "global_step": 2593, "epoch": 24, "lr": 9.94101084842948e-05} +{"train_loss": 0.02018047496676445, "grad_norm": 0.07632419466972351, "global_step": 2594, "epoch": 24, "lr": 9.940965448163806e-05} +{"train_loss": 0.01866849511861801, "grad_norm": 0.07699017971754074, "global_step": 2595, "epoch": 24, "lr": 9.940920030537726e-05} +{"train_loss": 0.0198298878967762, "grad_norm": 0.07043713331222534, "global_step": 2596, "epoch": 24, "lr": 9.940874595551404e-05} +{"train_loss": 0.02205175906419754, "grad_norm": 0.0432845801115036, "global_step": 2597, "epoch": 24, "lr": 9.940829143204995e-05} +{"train_loss": 0.021184757351875305, "grad_norm": 0.052048612385988235, "global_step": 2598, "epoch": 24, "lr": 9.940783673498663e-05} +{"train_loss": 0.019921116530895233, "grad_norm": 0.08670578896999359, "global_step": 2599, "epoch": 24, "lr": 9.940738186432565e-05} +{"train_loss": 0.023092210292816162, "grad_norm": 0.030982745811343193, "global_step": 2600, "epoch": 24, "lr": 9.940692682006861e-05} +{"train_loss": 0.020822763442993164, "grad_norm": 0.13417159020900726, "global_step": 2601, "epoch": 24, "lr": 9.940647160221712e-05} +{"train_loss": 0.020925432443618774, "grad_norm": 0.044909603893756866, "global_step": 2602, "epoch": 24, "lr": 9.940601621077279e-05} +{"train_loss": 0.02062489092350006, "grad_norm": 0.07534687966108322, "global_step": 2603, "epoch": 24, "lr": 9.940556064573718e-05} +{"train_loss": 0.017803382128477097, "grad_norm": 0.05232248455286026, "global_step": 2604, "epoch": 24, "lr": 9.940510490711193e-05} +{"train_loss": 0.0220378115773201, "grad_norm": 0.1435539722442627, "global_step": 2605, "epoch": 24, "lr": 9.940464899489863e-05} +{"train_loss": 0.02260160818696022, "grad_norm": 0.07632801681756973, "global_step": 2606, "epoch": 24, "lr": 9.940419290909888e-05} +{"train_loss": 0.02160009555518627, "grad_norm": 0.06809788942337036, "global_step": 2607, "epoch": 24, "lr": 9.940373664971427e-05} +{"train_loss": 0.019970763474702835, "grad_norm": 0.0401572547852993, "global_step": 2608, "epoch": 24, "lr": 9.940328021674644e-05} +{"train_loss": 0.025349566712975502, "grad_norm": 0.15063610672950745, "global_step": 2609, "epoch": 24, "lr": 9.940282361019694e-05} +{"train_loss": 0.017615484073758125, "grad_norm": 0.03342214599251747, "global_step": 2610, "epoch": 24, "lr": 9.94023668300674e-05} +{"train_loss": 0.01746269501745701, "grad_norm": 0.03051297925412655, "global_step": 2611, "epoch": 24, "lr": 9.940190987635944e-05} +{"train_loss": 0.02137671783566475, "grad_norm": 0.04526775702834129, "global_step": 2612, "epoch": 24, "lr": 9.940145274907466e-05} +{"train_loss": 0.01877463608980179, "grad_norm": 0.035147178918123245, "global_step": 2613, "epoch": 24, "lr": 9.940099544821465e-05} +{"train_loss": 0.02209864929318428, "grad_norm": 0.0685596764087677, "global_step": 2614, "epoch": 24, "lr": 9.940053797378103e-05} +{"train_loss": 0.021300196647644043, "grad_norm": 0.04347088560461998, "global_step": 2615, "epoch": 24, "lr": 9.94000803257754e-05} +{"train_loss": 0.02089594304561615, "grad_norm": 0.08424589782953262, "global_step": 2616, "epoch": 24, "lr": 9.939962250419938e-05} +{"train_loss": 0.020811963826417923, "grad_norm": 0.08625886589288712, "global_step": 2617, "epoch": 24, "lr": 9.939916450905456e-05} +{"train_loss": 0.020829427987337112, "grad_norm": 0.07592508941888809, "global_step": 2618, "epoch": 24, "lr": 9.939870634034257e-05} +{"train_loss": 0.022633716464042664, "grad_norm": 0.052151989191770554, "global_step": 2619, "epoch": 24, "lr": 9.939824799806498e-05} +{"train_loss": 0.019896600395441055, "grad_norm": 0.07031553238630295, "global_step": 2620, "epoch": 24, "lr": 9.939778948222346e-05} +{"train_loss": 0.02183772623538971, "grad_norm": 0.07161197811365128, "global_step": 2621, "epoch": 24, "lr": 9.939733079281958e-05} +{"train_loss": 0.019120782613754272, "grad_norm": 0.07173626869916916, "global_step": 2622, "epoch": 24, "lr": 9.939687192985496e-05} +{"train_loss": 0.025660932064056396, "grad_norm": 0.13439902663230896, "global_step": 2623, "epoch": 24, "lr": 9.939641289333121e-05} +{"train_loss": 0.020029235631227493, "grad_norm": 0.07963728159666061, "global_step": 2624, "epoch": 24, "lr": 9.939595368324996e-05} +{"train_loss": 0.021856022998690605, "grad_norm": 0.08660907298326492, "global_step": 2625, "epoch": 24, "lr": 9.93954942996128e-05} +{"train_loss": 0.02298838272690773, "grad_norm": 0.02871520072221756, "global_step": 2626, "epoch": 24, "lr": 9.939503474242136e-05} +{"train_loss": 0.024166898801922798, "grad_norm": 0.037844978272914886, "global_step": 2627, "epoch": 24, "lr": 9.939457501167723e-05} +{"train_loss": 0.021002091467380524, "grad_norm": 0.05195162072777748, "global_step": 2628, "epoch": 24, "lr": 9.939411510738209e-05} +{"train_loss": 0.02171233296394348, "grad_norm": 0.036876898258924484, "global_step": 2629, "epoch": 24, "lr": 9.939365502953748e-05} +{"train_loss": 0.020625993609428406, "grad_norm": 0.03671358525753021, "global_step": 2630, "epoch": 24, "lr": 9.939319477814504e-05} +{"train_loss": 0.020044786855578423, "grad_norm": 0.04193808138370514, "global_step": 2631, "epoch": 24, "lr": 9.939273435320639e-05} +{"train_loss": 0.02393762767314911, "grad_norm": 0.06182330101728439, "global_step": 2632, "epoch": 24, "lr": 9.939227375472318e-05} +{"train_loss": 0.020855192095041275, "grad_norm": 0.05544376000761986, "global_step": 2633, "epoch": 24, "lr": 9.939181298269698e-05} +{"train_loss": 0.02024919167160988, "grad_norm": 0.10092446953058243, "global_step": 2634, "epoch": 24, "lr": 9.939135203712944e-05} +{"train_loss": 0.02030760981142521, "grad_norm": 0.05266343429684639, "global_step": 2635, "epoch": 24, "lr": 9.939089091802216e-05} +{"train_loss": 0.022533362731337547, "grad_norm": 0.07572395354509354, "global_step": 2636, "epoch": 24, "lr": 9.939042962537677e-05} +{"train_loss": 0.02102716639637947, "grad_norm": 0.08395601063966751, "global_step": 2637, "epoch": 24, "lr": 9.93899681591949e-05} +{"train_loss": 0.023761101067066193, "grad_norm": 0.038565605878829956, "global_step": 2638, "epoch": 24, "lr": 9.938950651947814e-05} +{"train_loss": 0.023702654987573624, "grad_norm": 0.037895187735557556, "global_step": 2639, "epoch": 24, "lr": 9.938904470622815e-05} +{"train_loss": 0.02198963612318039, "grad_norm": 0.042594317346811295, "global_step": 2640, "epoch": 24, "lr": 9.938858271944653e-05} +{"train_loss": 0.019515449181199074, "grad_norm": 0.05247850343585014, "global_step": 2641, "epoch": 24, "lr": 9.938812055913492e-05} +{"train_loss": 0.020501643419265747, "grad_norm": 0.03656495735049248, "global_step": 2642, "epoch": 24, "lr": 9.938765822529491e-05} +{"train_loss": 0.02081644907593727, "grad_norm": 0.04079127684235573, "global_step": 2643, "epoch": 24, "lr": 9.938719571792817e-05} +{"train_loss": 0.01897340640425682, "grad_norm": 0.056605350226163864, "global_step": 2644, "epoch": 24, "lr": 9.938673303703628e-05} +{"train_loss": 0.018993237987160683, "grad_norm": 0.09453289955854416, "global_step": 2645, "epoch": 24, "lr": 9.93862701826209e-05} +{"train_loss": 0.020544521510601044, "grad_norm": 0.1029960885643959, "global_step": 2646, "epoch": 24, "lr": 9.938580715468364e-05} +{"train_loss": 0.018977293744683266, "grad_norm": 0.0407717227935791, "global_step": 2647, "epoch": 24, "lr": 9.938534395322614e-05} +{"train_loss": 0.01911487802863121, "grad_norm": 0.08248802274465561, "global_step": 2648, "epoch": 24, "lr": 9.938488057825002e-05} +{"train_loss": 0.021093982394854976, "grad_norm": 0.06582420319318771, "global_step": 2649, "epoch": 24, "lr": 9.938441702975689e-05, "val_loss": 0.020345814526081085} +{"train_loss": 0.025495115667581558, "grad_norm": 0.05624945089221001, "global_step": 2650, "epoch": 25, "lr": 9.938395330774841e-05} +{"train_loss": 0.019699465483427048, "grad_norm": 0.09759083390235901, "global_step": 2651, "epoch": 25, "lr": 9.93834894122262e-05} +{"train_loss": 0.020606165751814842, "grad_norm": 0.12163794785737991, "global_step": 2652, "epoch": 25, "lr": 9.938302534319185e-05} +{"train_loss": 0.021106433123350143, "grad_norm": 0.06628001481294632, "global_step": 2653, "epoch": 25, "lr": 9.938256110064706e-05} +{"train_loss": 0.01978921890258789, "grad_norm": 0.0441889725625515, "global_step": 2654, "epoch": 25, "lr": 9.938209668459341e-05} +{"train_loss": 0.022140150889754295, "grad_norm": 0.07300882786512375, "global_step": 2655, "epoch": 25, "lr": 9.938163209503255e-05} +{"train_loss": 0.01954321563243866, "grad_norm": 0.08736872673034668, "global_step": 2656, "epoch": 25, "lr": 9.938116733196612e-05} +{"train_loss": 0.019189808517694473, "grad_norm": 0.12473665177822113, "global_step": 2657, "epoch": 25, "lr": 9.938070239539574e-05} +{"train_loss": 0.019858770072460175, "grad_norm": 0.04257821664214134, "global_step": 2658, "epoch": 25, "lr": 9.938023728532304e-05} +{"train_loss": 0.02148374542593956, "grad_norm": 0.08792271465063095, "global_step": 2659, "epoch": 25, "lr": 9.937977200174966e-05} +{"train_loss": 0.01916152983903885, "grad_norm": 0.07427505403757095, "global_step": 2660, "epoch": 25, "lr": 9.937930654467724e-05} +{"train_loss": 0.024316739290952682, "grad_norm": 0.05031261220574379, "global_step": 2661, "epoch": 25, "lr": 9.937884091410741e-05} +{"train_loss": 0.021596282720565796, "grad_norm": 0.06287242472171783, "global_step": 2662, "epoch": 25, "lr": 9.937837511004182e-05} +{"train_loss": 0.021637581288814545, "grad_norm": 0.057369768619537354, "global_step": 2663, "epoch": 25, "lr": 9.937790913248209e-05} +{"train_loss": 0.022364255040884018, "grad_norm": 0.11112291365861893, "global_step": 2664, "epoch": 25, "lr": 9.937744298142985e-05} +{"train_loss": 0.01995149254798889, "grad_norm": 0.060738250613212585, "global_step": 2665, "epoch": 25, "lr": 9.937697665688674e-05} +{"train_loss": 0.021433565765619278, "grad_norm": 0.050097014755010605, "global_step": 2666, "epoch": 25, "lr": 9.937651015885443e-05} +{"train_loss": 0.022113826125860214, "grad_norm": 0.10620871931314468, "global_step": 2667, "epoch": 25, "lr": 9.937604348733453e-05} +{"train_loss": 0.020943965762853622, "grad_norm": 0.05192062258720398, "global_step": 2668, "epoch": 25, "lr": 9.937557664232868e-05} +{"train_loss": 0.019314005970954895, "grad_norm": 0.08355895429849625, "global_step": 2669, "epoch": 25, "lr": 9.937510962383853e-05} +{"train_loss": 0.02161896601319313, "grad_norm": 0.05154917761683464, "global_step": 2670, "epoch": 25, "lr": 9.937464243186572e-05} +{"train_loss": 0.019895115867257118, "grad_norm": 0.0864417776465416, "global_step": 2671, "epoch": 25, "lr": 9.937417506641189e-05} +{"train_loss": 0.022399207577109337, "grad_norm": 0.1214524582028389, "global_step": 2672, "epoch": 25, "lr": 9.937370752747867e-05} +{"train_loss": 0.01841871254146099, "grad_norm": 0.06052530184388161, "global_step": 2673, "epoch": 25, "lr": 9.937323981506772e-05} +{"train_loss": 0.018642963841557503, "grad_norm": 0.04601104557514191, "global_step": 2674, "epoch": 25, "lr": 9.937277192918069e-05} +{"train_loss": 0.02254914492368698, "grad_norm": 0.10870753228664398, "global_step": 2675, "epoch": 25, "lr": 9.93723038698192e-05} +{"train_loss": 0.02074536681175232, "grad_norm": 0.05646523833274841, "global_step": 2676, "epoch": 25, "lr": 9.93718356369849e-05} +{"train_loss": 0.021142836660146713, "grad_norm": 0.07978510111570358, "global_step": 2677, "epoch": 25, "lr": 9.937136723067944e-05} +{"train_loss": 0.020952880382537842, "grad_norm": 0.059272777289152145, "global_step": 2678, "epoch": 25, "lr": 9.937089865090447e-05} +{"train_loss": 0.020083405077457428, "grad_norm": 0.052038565278053284, "global_step": 2679, "epoch": 25, "lr": 9.937042989766165e-05} +{"train_loss": 0.018062619492411613, "grad_norm": 0.04131440818309784, "global_step": 2680, "epoch": 25, "lr": 9.93699609709526e-05} +{"train_loss": 0.02073504962027073, "grad_norm": 0.036840636283159256, "global_step": 2681, "epoch": 25, "lr": 9.936949187077897e-05} +{"train_loss": 0.0236628670245409, "grad_norm": 0.04013144597411156, "global_step": 2682, "epoch": 25, "lr": 9.936902259714241e-05} +{"train_loss": 0.023994050920009613, "grad_norm": 0.06026895344257355, "global_step": 2683, "epoch": 25, "lr": 9.936855315004458e-05} +{"train_loss": 0.022497473284602165, "grad_norm": 0.05574822053313255, "global_step": 2684, "epoch": 25, "lr": 9.936808352948715e-05} +{"train_loss": 0.02028829976916313, "grad_norm": 0.178987056016922, "global_step": 2685, "epoch": 25, "lr": 9.93676137354717e-05} +{"train_loss": 0.020402435213327408, "grad_norm": 0.05350921303033829, "global_step": 2686, "epoch": 25, "lr": 9.936714376799996e-05} +{"train_loss": 0.021129384636878967, "grad_norm": 0.05991942808032036, "global_step": 2687, "epoch": 25, "lr": 9.936667362707353e-05} +{"train_loss": 0.020109038800001144, "grad_norm": 0.04214085265994072, "global_step": 2688, "epoch": 25, "lr": 9.936620331269408e-05} +{"train_loss": 0.020772915333509445, "grad_norm": 0.04136982187628746, "global_step": 2689, "epoch": 25, "lr": 9.936573282486326e-05} +{"train_loss": 0.01948651298880577, "grad_norm": 0.061017949134111404, "global_step": 2690, "epoch": 25, "lr": 9.936526216358273e-05} +{"train_loss": 0.02287178859114647, "grad_norm": 0.08843933790922165, "global_step": 2691, "epoch": 25, "lr": 9.936479132885412e-05} +{"train_loss": 0.01966845989227295, "grad_norm": 0.06708662956953049, "global_step": 2692, "epoch": 25, "lr": 9.936432032067913e-05} +{"train_loss": 0.020019419491291046, "grad_norm": 0.08279361575841904, "global_step": 2693, "epoch": 25, "lr": 9.936384913905936e-05} +{"train_loss": 0.020748578011989594, "grad_norm": 0.036033518612384796, "global_step": 2694, "epoch": 25, "lr": 9.93633777839965e-05} +{"train_loss": 0.02008461393415928, "grad_norm": 0.08766830712556839, "global_step": 2695, "epoch": 25, "lr": 9.93629062554922e-05} +{"train_loss": 0.020236484706401825, "grad_norm": 0.09545419365167618, "global_step": 2696, "epoch": 25, "lr": 9.936243455354812e-05} +{"train_loss": 0.022867992520332336, "grad_norm": 0.1780800223350525, "global_step": 2697, "epoch": 25, "lr": 9.93619626781659e-05} +{"train_loss": 0.026312485337257385, "grad_norm": 0.1942238062620163, "global_step": 2698, "epoch": 25, "lr": 9.936149062934721e-05} +{"train_loss": 0.02013368532061577, "grad_norm": 0.11096794903278351, "global_step": 2699, "epoch": 25, "lr": 9.936101840709372e-05} +{"train_loss": 0.020221248269081116, "grad_norm": 0.08748544752597809, "global_step": 2700, "epoch": 25, "lr": 9.936054601140708e-05} +{"train_loss": 0.021242983639240265, "grad_norm": 0.05794582515954971, "global_step": 2701, "epoch": 25, "lr": 9.936007344228894e-05} +{"train_loss": 0.021560080349445343, "grad_norm": 0.041503217071294785, "global_step": 2702, "epoch": 25, "lr": 9.935960069974096e-05} +{"train_loss": 0.02436016872525215, "grad_norm": 0.08817760646343231, "global_step": 2703, "epoch": 25, "lr": 9.935912778376482e-05} +{"train_loss": 0.02097911760210991, "grad_norm": 0.05850988253951073, "global_step": 2704, "epoch": 25, "lr": 9.935865469436217e-05} +{"train_loss": 0.021624227985739708, "grad_norm": 0.046708375215530396, "global_step": 2705, "epoch": 25, "lr": 9.935818143153467e-05} +{"train_loss": 0.020316841080784798, "grad_norm": 0.041671089828014374, "global_step": 2706, "epoch": 25, "lr": 9.935770799528399e-05} +{"train_loss": 0.02030141092836857, "grad_norm": 0.08446118980646133, "global_step": 2707, "epoch": 25, "lr": 9.935723438561178e-05} +{"train_loss": 0.02226785570383072, "grad_norm": 0.066886305809021, "global_step": 2708, "epoch": 25, "lr": 9.935676060251973e-05} +{"train_loss": 0.019903942942619324, "grad_norm": 0.0615355484187603, "global_step": 2709, "epoch": 25, "lr": 9.935628664600946e-05} +{"train_loss": 0.01638219505548477, "grad_norm": 0.03909826651215553, "global_step": 2710, "epoch": 25, "lr": 9.93558125160827e-05} +{"train_loss": 0.019318901002407074, "grad_norm": 0.04711645096540451, "global_step": 2711, "epoch": 25, "lr": 9.935533821274106e-05} +{"train_loss": 0.02121036686003208, "grad_norm": 0.15243898332118988, "global_step": 2712, "epoch": 25, "lr": 9.935486373598621e-05} +{"train_loss": 0.0208599753677845, "grad_norm": 0.0563490092754364, "global_step": 2713, "epoch": 25, "lr": 9.935438908581985e-05} +{"train_loss": 0.0207061767578125, "grad_norm": 0.050402089953422546, "global_step": 2714, "epoch": 25, "lr": 9.935391426224363e-05} +{"train_loss": 0.019773432984948158, "grad_norm": 0.08603352308273315, "global_step": 2715, "epoch": 25, "lr": 9.935343926525921e-05} +{"train_loss": 0.02193107083439827, "grad_norm": 0.06724291294813156, "global_step": 2716, "epoch": 25, "lr": 9.935296409486829e-05} +{"train_loss": 0.022836998105049133, "grad_norm": 0.053468212485313416, "global_step": 2717, "epoch": 25, "lr": 9.93524887510725e-05} +{"train_loss": 0.02044953964650631, "grad_norm": 0.03652836009860039, "global_step": 2718, "epoch": 25, "lr": 9.935201323387353e-05} +{"train_loss": 0.020041778683662415, "grad_norm": 0.03951774910092354, "global_step": 2719, "epoch": 25, "lr": 9.935153754327304e-05} +{"train_loss": 0.01885037124156952, "grad_norm": 0.04187975451350212, "global_step": 2720, "epoch": 25, "lr": 9.935106167927272e-05} +{"train_loss": 0.019908424466848373, "grad_norm": 0.08847606182098389, "global_step": 2721, "epoch": 25, "lr": 9.935058564187423e-05} +{"train_loss": 0.02157048135995865, "grad_norm": 0.043301548808813095, "global_step": 2722, "epoch": 25, "lr": 9.935010943107925e-05} +{"train_loss": 0.01884857937693596, "grad_norm": 0.043851349502801895, "global_step": 2723, "epoch": 25, "lr": 9.934963304688943e-05} +{"train_loss": 0.022623583674430847, "grad_norm": 0.04947763308882713, "global_step": 2724, "epoch": 25, "lr": 9.934915648930647e-05} +{"train_loss": 0.020275497809052467, "grad_norm": 0.08495616912841797, "global_step": 2725, "epoch": 25, "lr": 9.934867975833205e-05} +{"train_loss": 0.021707305684685707, "grad_norm": 0.057872503995895386, "global_step": 2726, "epoch": 25, "lr": 9.93482028539678e-05} +{"train_loss": 0.02154701203107834, "grad_norm": 0.09997481107711792, "global_step": 2727, "epoch": 25, "lr": 9.934772577621545e-05} +{"train_loss": 0.022279027849435806, "grad_norm": 0.05868053436279297, "global_step": 2728, "epoch": 25, "lr": 9.934724852507664e-05} +{"train_loss": 0.01897309720516205, "grad_norm": 0.04692932218313217, "global_step": 2729, "epoch": 25, "lr": 9.934677110055307e-05} +{"train_loss": 0.021738525480031967, "grad_norm": 0.0812099501490593, "global_step": 2730, "epoch": 25, "lr": 9.93462935026464e-05} +{"train_loss": 0.019816439598798752, "grad_norm": 0.03514238819479942, "global_step": 2731, "epoch": 25, "lr": 9.934581573135832e-05} +{"train_loss": 0.023287851363420486, "grad_norm": 0.09508691728115082, "global_step": 2732, "epoch": 25, "lr": 9.93453377866905e-05} +{"train_loss": 0.02442910149693489, "grad_norm": 0.039757370948791504, "global_step": 2733, "epoch": 25, "lr": 9.934485966864462e-05} +{"train_loss": 0.018100379034876823, "grad_norm": 0.050288185477256775, "global_step": 2734, "epoch": 25, "lr": 9.934438137722237e-05} +{"train_loss": 0.02338431403040886, "grad_norm": 0.14025191962718964, "global_step": 2735, "epoch": 25, "lr": 9.934390291242541e-05} +{"train_loss": 0.020069435238838196, "grad_norm": 0.08747893571853638, "global_step": 2736, "epoch": 25, "lr": 9.934342427425545e-05} +{"train_loss": 0.021814288571476936, "grad_norm": 0.04072989150881767, "global_step": 2737, "epoch": 25, "lr": 9.934294546271414e-05} +{"train_loss": 0.02243461087346077, "grad_norm": 0.09503313153982162, "global_step": 2738, "epoch": 25, "lr": 9.934246647780319e-05} +{"train_loss": 0.018581613898277283, "grad_norm": 0.05421904847025871, "global_step": 2739, "epoch": 25, "lr": 9.934198731952427e-05} +{"train_loss": 0.022234488278627396, "grad_norm": 0.08559095114469528, "global_step": 2740, "epoch": 25, "lr": 9.934150798787905e-05} +{"train_loss": 0.01964367926120758, "grad_norm": 0.056213509291410446, "global_step": 2741, "epoch": 25, "lr": 9.934102848286926e-05} +{"train_loss": 0.02014458179473877, "grad_norm": 0.03214556351304054, "global_step": 2742, "epoch": 25, "lr": 9.934054880449652e-05} +{"train_loss": 0.01960684545338154, "grad_norm": 0.04097241535782814, "global_step": 2743, "epoch": 25, "lr": 9.934006895276257e-05} +{"train_loss": 0.023841332644224167, "grad_norm": 0.07589559257030487, "global_step": 2744, "epoch": 25, "lr": 9.933958892766907e-05} +{"train_loss": 0.022301187738776207, "grad_norm": 0.09855262190103531, "global_step": 2745, "epoch": 25, "lr": 9.93391087292177e-05} +{"train_loss": 0.025408321991562843, "grad_norm": 0.046974897384643555, "global_step": 2746, "epoch": 25, "lr": 9.933862835741017e-05} +{"train_loss": 0.019669800996780396, "grad_norm": 0.04531349614262581, "global_step": 2747, "epoch": 25, "lr": 9.933814781224818e-05} +{"train_loss": 0.022582881152629852, "grad_norm": 0.08477111160755157, "global_step": 2748, "epoch": 25, "lr": 9.933766709373335e-05} +{"train_loss": 0.02077750861644745, "grad_norm": 0.07599738240242004, "global_step": 2749, "epoch": 25, "lr": 9.933718620186744e-05} +{"train_loss": 0.018422670662403107, "grad_norm": 0.0682515874505043, "global_step": 2750, "epoch": 25, "lr": 9.933670513665212e-05} +{"train_loss": 0.023660290986299515, "grad_norm": 0.06888772547245026, "global_step": 2751, "epoch": 25, "lr": 9.933622389808906e-05} +{"train_loss": 0.01869170553982258, "grad_norm": 0.030968965962529182, "global_step": 2752, "epoch": 25, "lr": 9.933574248617997e-05} +{"train_loss": 0.019498080015182495, "grad_norm": 0.06242019310593605, "global_step": 2753, "epoch": 25, "lr": 9.933526090092652e-05} +{"train_loss": 0.020815754309296608, "grad_norm": 0.120256707072258, "global_step": 2754, "epoch": 25, "lr": 9.933477914233044e-05} +{"train_loss": 0.020987393264219445, "grad_norm": 0.08850756287574768, "global_step": 2755, "epoch": 25, "lr": 9.93342972103934e-05, "val_loss": 0.020173970609903336} +{"train_loss": 0.017322132363915443, "grad_norm": 0.03606034442782402, "global_step": 2756, "epoch": 26, "lr": 9.93338151051171e-05} +{"train_loss": 0.022262142971158028, "grad_norm": 0.06706106662750244, "global_step": 2757, "epoch": 26, "lr": 9.933333282650321e-05} +{"train_loss": 0.0221126526594162, "grad_norm": 0.0847100019454956, "global_step": 2758, "epoch": 26, "lr": 9.933285037455347e-05} +{"train_loss": 0.01899573765695095, "grad_norm": 0.03575930371880531, "global_step": 2759, "epoch": 26, "lr": 9.933236774926952e-05} +{"train_loss": 0.022164449095726013, "grad_norm": 0.07617931813001633, "global_step": 2760, "epoch": 26, "lr": 9.933188495065308e-05} +{"train_loss": 0.020276784896850586, "grad_norm": 0.06130911409854889, "global_step": 2761, "epoch": 26, "lr": 9.933140197870587e-05} +{"train_loss": 0.017969155684113503, "grad_norm": 0.04434311017394066, "global_step": 2762, "epoch": 26, "lr": 9.933091883342956e-05} +{"train_loss": 0.02001558430492878, "grad_norm": 0.04530131071805954, "global_step": 2763, "epoch": 26, "lr": 9.933043551482586e-05} +{"train_loss": 0.02019156515598297, "grad_norm": 0.04729028046131134, "global_step": 2764, "epoch": 26, "lr": 9.932995202289645e-05} +{"train_loss": 0.022608214989304543, "grad_norm": 0.17157453298568726, "global_step": 2765, "epoch": 26, "lr": 9.932946835764306e-05} +{"train_loss": 0.01873694732785225, "grad_norm": 0.04782847315073013, "global_step": 2766, "epoch": 26, "lr": 9.932898451906735e-05} +{"train_loss": 0.02351069077849388, "grad_norm": 0.05076166242361069, "global_step": 2767, "epoch": 26, "lr": 9.932850050717106e-05} +{"train_loss": 0.017190169543027878, "grad_norm": 0.05571100115776062, "global_step": 2768, "epoch": 26, "lr": 9.932801632195587e-05} +{"train_loss": 0.01969725638628006, "grad_norm": 0.036146074533462524, "global_step": 2769, "epoch": 26, "lr": 9.932753196342347e-05} +{"train_loss": 0.0190504752099514, "grad_norm": 0.032467082142829895, "global_step": 2770, "epoch": 26, "lr": 9.932704743157559e-05} +{"train_loss": 0.021919678896665573, "grad_norm": 0.04491736739873886, "global_step": 2771, "epoch": 26, "lr": 9.932656272641389e-05} +{"train_loss": 0.022763986140489578, "grad_norm": 0.05620447173714638, "global_step": 2772, "epoch": 26, "lr": 9.932607784794012e-05} +{"train_loss": 0.024807121604681015, "grad_norm": 0.15977619588375092, "global_step": 2773, "epoch": 26, "lr": 9.932559279615597e-05} +{"train_loss": 0.019310466945171356, "grad_norm": 0.07230682671070099, "global_step": 2774, "epoch": 26, "lr": 9.932510757106313e-05} +{"train_loss": 0.02380523830652237, "grad_norm": 0.08379286527633667, "global_step": 2775, "epoch": 26, "lr": 9.932462217266332e-05} +{"train_loss": 0.02159777656197548, "grad_norm": 0.06817162781953812, "global_step": 2776, "epoch": 26, "lr": 9.932413660095823e-05} +{"train_loss": 0.022563468664884567, "grad_norm": 0.1533316820859909, "global_step": 2777, "epoch": 26, "lr": 9.932365085594956e-05} +{"train_loss": 0.022190473973751068, "grad_norm": 0.06227324157953262, "global_step": 2778, "epoch": 26, "lr": 9.932316493763905e-05} +{"train_loss": 0.01956227608025074, "grad_norm": 0.10896153002977371, "global_step": 2779, "epoch": 26, "lr": 9.932267884602839e-05} +{"train_loss": 0.020830828696489334, "grad_norm": 0.04796129837632179, "global_step": 2780, "epoch": 26, "lr": 9.932219258111927e-05} +{"train_loss": 0.021901143714785576, "grad_norm": 0.05460195243358612, "global_step": 2781, "epoch": 26, "lr": 9.932170614291344e-05} +{"train_loss": 0.02097569964826107, "grad_norm": 0.11709558963775635, "global_step": 2782, "epoch": 26, "lr": 9.932121953141257e-05} +{"train_loss": 0.022728191688656807, "grad_norm": 0.10861225426197052, "global_step": 2783, "epoch": 26, "lr": 9.932073274661839e-05} +{"train_loss": 0.018100257962942123, "grad_norm": 0.09782838821411133, "global_step": 2784, "epoch": 26, "lr": 9.93202457885326e-05} +{"train_loss": 0.021325204521417618, "grad_norm": 0.04829318821430206, "global_step": 2785, "epoch": 26, "lr": 9.93197586571569e-05} +{"train_loss": 0.023897189646959305, "grad_norm": 0.03207901865243912, "global_step": 2786, "epoch": 26, "lr": 9.931927135249304e-05} +{"train_loss": 0.01832423359155655, "grad_norm": 0.06166622415184975, "global_step": 2787, "epoch": 26, "lr": 9.93187838745427e-05} +{"train_loss": 0.022835420444607735, "grad_norm": 0.06388280540704727, "global_step": 2788, "epoch": 26, "lr": 9.931829622330759e-05} +{"train_loss": 0.019453277811408043, "grad_norm": 0.036540400236845016, "global_step": 2789, "epoch": 26, "lr": 9.931780839878946e-05} +{"train_loss": 0.020667487755417824, "grad_norm": 0.05853680521249771, "global_step": 2790, "epoch": 26, "lr": 9.931732040098997e-05} +{"train_loss": 0.02214415930211544, "grad_norm": 0.06668689846992493, "global_step": 2791, "epoch": 26, "lr": 9.931683222991089e-05} +{"train_loss": 0.0203053317964077, "grad_norm": 0.07635432481765747, "global_step": 2792, "epoch": 26, "lr": 9.93163438855539e-05} +{"train_loss": 0.01965957321226597, "grad_norm": 0.04277702793478966, "global_step": 2793, "epoch": 26, "lr": 9.931585536792072e-05} +{"train_loss": 0.016436733305454254, "grad_norm": 0.10357745736837387, "global_step": 2794, "epoch": 26, "lr": 9.931536667701306e-05} +{"train_loss": 0.02309860661625862, "grad_norm": 0.04433466121554375, "global_step": 2795, "epoch": 26, "lr": 9.931487781283267e-05} +{"train_loss": 0.020876090973615646, "grad_norm": 0.14605176448822021, "global_step": 2796, "epoch": 26, "lr": 9.931438877538124e-05} +{"train_loss": 0.01941545121371746, "grad_norm": 0.045387011021375656, "global_step": 2797, "epoch": 26, "lr": 9.931389956466049e-05} +{"train_loss": 0.021271927282214165, "grad_norm": 0.1117587685585022, "global_step": 2798, "epoch": 26, "lr": 9.931341018067215e-05} +{"train_loss": 0.021388286724686623, "grad_norm": 0.05804029852151871, "global_step": 2799, "epoch": 26, "lr": 9.931292062341793e-05} +{"train_loss": 0.01932242512702942, "grad_norm": 0.10201481729745865, "global_step": 2800, "epoch": 26, "lr": 9.931243089289954e-05} +{"train_loss": 0.021424129605293274, "grad_norm": 0.09310898929834366, "global_step": 2801, "epoch": 26, "lr": 9.931194098911874e-05} +{"train_loss": 0.0236598402261734, "grad_norm": 0.06595871597528458, "global_step": 2802, "epoch": 26, "lr": 9.931145091207719e-05} +{"train_loss": 0.022432927042245865, "grad_norm": 0.0762971043586731, "global_step": 2803, "epoch": 26, "lr": 9.931096066177667e-05} +{"train_loss": 0.022252928465604782, "grad_norm": 0.0956096276640892, "global_step": 2804, "epoch": 26, "lr": 9.931047023821887e-05} +{"train_loss": 0.02135728858411312, "grad_norm": 0.09666512906551361, "global_step": 2805, "epoch": 26, "lr": 9.930997964140552e-05} +{"train_loss": 0.022837333381175995, "grad_norm": 0.048625972121953964, "global_step": 2806, "epoch": 26, "lr": 9.930948887133836e-05} +{"train_loss": 0.021865960210561752, "grad_norm": 0.04575902596116066, "global_step": 2807, "epoch": 26, "lr": 9.930899792801909e-05} +{"train_loss": 0.021524861454963684, "grad_norm": 0.10332576930522919, "global_step": 2808, "epoch": 26, "lr": 9.930850681144945e-05} +{"train_loss": 0.022246720269322395, "grad_norm": 0.09343307465314865, "global_step": 2809, "epoch": 26, "lr": 9.930801552163116e-05} +{"train_loss": 0.021754687651991844, "grad_norm": 0.19503425061702728, "global_step": 2810, "epoch": 26, "lr": 9.930752405856594e-05} +{"train_loss": 0.02007942646741867, "grad_norm": 0.06057985499501228, "global_step": 2811, "epoch": 26, "lr": 9.930703242225552e-05} +{"train_loss": 0.023587126284837723, "grad_norm": 0.11681216210126877, "global_step": 2812, "epoch": 26, "lr": 9.930654061270163e-05} +{"train_loss": 0.02100522257387638, "grad_norm": 0.046603552997112274, "global_step": 2813, "epoch": 26, "lr": 9.930604862990602e-05} +{"train_loss": 0.01904565468430519, "grad_norm": 0.07175043225288391, "global_step": 2814, "epoch": 26, "lr": 9.930555647387039e-05} +{"train_loss": 0.01831131801009178, "grad_norm": 0.054562538862228394, "global_step": 2815, "epoch": 26, "lr": 9.930506414459647e-05} +{"train_loss": 0.021980609744787216, "grad_norm": 0.054076794534921646, "global_step": 2816, "epoch": 26, "lr": 9.9304571642086e-05} +{"train_loss": 0.02159755304455757, "grad_norm": 0.06714844703674316, "global_step": 2817, "epoch": 26, "lr": 9.93040789663407e-05} +{"train_loss": 0.020886782556772232, "grad_norm": 0.07536175847053528, "global_step": 2818, "epoch": 26, "lr": 9.930358611736232e-05} +{"train_loss": 0.026765653863549232, "grad_norm": 0.056345775723457336, "global_step": 2819, "epoch": 26, "lr": 9.930309309515258e-05} +{"train_loss": 0.020878080278635025, "grad_norm": 0.057968202978372574, "global_step": 2820, "epoch": 26, "lr": 9.93025998997132e-05} +{"train_loss": 0.02302894927561283, "grad_norm": 0.1594306081533432, "global_step": 2821, "epoch": 26, "lr": 9.930210653104594e-05} +{"train_loss": 0.02143351174890995, "grad_norm": 0.08979275077581406, "global_step": 2822, "epoch": 26, "lr": 9.930161298915251e-05} +{"train_loss": 0.01987900212407112, "grad_norm": 0.11612071841955185, "global_step": 2823, "epoch": 26, "lr": 9.930111927403465e-05} +{"train_loss": 0.01878158003091812, "grad_norm": 0.12810464203357697, "global_step": 2824, "epoch": 26, "lr": 9.93006253856941e-05} +{"train_loss": 0.02188553661108017, "grad_norm": 0.0351785309612751, "global_step": 2825, "epoch": 26, "lr": 9.93001313241326e-05} +{"train_loss": 0.019934356212615967, "grad_norm": 0.055690400302410126, "global_step": 2826, "epoch": 26, "lr": 9.929963708935187e-05} +{"train_loss": 0.019302042201161385, "grad_norm": 0.04421134665608406, "global_step": 2827, "epoch": 26, "lr": 9.929914268135366e-05} +{"train_loss": 0.023246146738529205, "grad_norm": 0.06594764441251755, "global_step": 2828, "epoch": 26, "lr": 9.929864810013969e-05} +{"train_loss": 0.018906259909272194, "grad_norm": 0.07303649932146072, "global_step": 2829, "epoch": 26, "lr": 9.929815334571174e-05} +{"train_loss": 0.023865316063165665, "grad_norm": 0.083269864320755, "global_step": 2830, "epoch": 26, "lr": 9.929765841807148e-05} +{"train_loss": 0.018405914306640625, "grad_norm": 0.0616401731967926, "global_step": 2831, "epoch": 26, "lr": 9.92971633172207e-05} +{"train_loss": 0.021589860320091248, "grad_norm": 0.07221464812755585, "global_step": 2832, "epoch": 26, "lr": 9.929666804316113e-05} +{"train_loss": 0.02116427943110466, "grad_norm": 0.05626676231622696, "global_step": 2833, "epoch": 26, "lr": 9.929617259589452e-05} +{"train_loss": 0.022365348413586617, "grad_norm": 0.1136387512087822, "global_step": 2834, "epoch": 26, "lr": 9.92956769754226e-05} +{"train_loss": 0.017512904480099678, "grad_norm": 0.08483146131038666, "global_step": 2835, "epoch": 26, "lr": 9.929518118174707e-05} +{"train_loss": 0.019962474703788757, "grad_norm": 0.09070377796888351, "global_step": 2836, "epoch": 26, "lr": 9.929468521486975e-05} +{"train_loss": 0.01811542734503746, "grad_norm": 0.08890774101018906, "global_step": 2837, "epoch": 26, "lr": 9.929418907479234e-05} +{"train_loss": 0.019934292882680893, "grad_norm": 0.046247754245996475, "global_step": 2838, "epoch": 26, "lr": 9.929369276151657e-05} +{"train_loss": 0.01883714087307453, "grad_norm": 0.05999375507235527, "global_step": 2839, "epoch": 26, "lr": 9.92931962750442e-05} +{"train_loss": 0.02093886397778988, "grad_norm": 0.06351286172866821, "global_step": 2840, "epoch": 26, "lr": 9.9292699615377e-05} +{"train_loss": 0.01811753213405609, "grad_norm": 0.06819980591535568, "global_step": 2841, "epoch": 26, "lr": 9.929220278251667e-05} +{"train_loss": 0.022787529975175858, "grad_norm": 0.10523539036512375, "global_step": 2842, "epoch": 26, "lr": 9.929170577646498e-05} +{"train_loss": 0.022288601845502853, "grad_norm": 0.06710365414619446, "global_step": 2843, "epoch": 26, "lr": 9.929120859722368e-05} +{"train_loss": 0.017430700361728668, "grad_norm": 0.08138164132833481, "global_step": 2844, "epoch": 26, "lr": 9.92907112447945e-05} +{"train_loss": 0.018431277945637703, "grad_norm": 0.07021236419677734, "global_step": 2845, "epoch": 26, "lr": 9.929021371917919e-05} +{"train_loss": 0.020253168419003487, "grad_norm": 0.04541441798210144, "global_step": 2846, "epoch": 26, "lr": 9.928971602037952e-05} +{"train_loss": 0.01978989690542221, "grad_norm": 0.0780322477221489, "global_step": 2847, "epoch": 26, "lr": 9.928921814839722e-05} +{"train_loss": 0.021317368373274803, "grad_norm": 0.050733234733343124, "global_step": 2848, "epoch": 26, "lr": 9.928872010323403e-05} +{"train_loss": 0.02348114177584648, "grad_norm": 0.07458596676588058, "global_step": 2849, "epoch": 26, "lr": 9.928822188489173e-05} +{"train_loss": 0.022457342594861984, "grad_norm": 0.095030777156353, "global_step": 2850, "epoch": 26, "lr": 9.928772349337204e-05} +{"train_loss": 0.019772058352828026, "grad_norm": 0.055070340633392334, "global_step": 2851, "epoch": 26, "lr": 9.928722492867673e-05} +{"train_loss": 0.019081076607108116, "grad_norm": 0.09554455429315567, "global_step": 2852, "epoch": 26, "lr": 9.928672619080754e-05} +{"train_loss": 0.026756450533866882, "grad_norm": 0.11814756691455841, "global_step": 2853, "epoch": 26, "lr": 9.928622727976624e-05} +{"train_loss": 0.021752487868070602, "grad_norm": 0.046012602746486664, "global_step": 2854, "epoch": 26, "lr": 9.928572819555458e-05} +{"train_loss": 0.02048756182193756, "grad_norm": 0.13783730566501617, "global_step": 2855, "epoch": 26, "lr": 9.928522893817428e-05} +{"train_loss": 0.02063390612602234, "grad_norm": 0.05526502802968025, "global_step": 2856, "epoch": 26, "lr": 9.928472950762712e-05} +{"train_loss": 0.02247658371925354, "grad_norm": 0.07353315502405167, "global_step": 2857, "epoch": 26, "lr": 9.928422990391488e-05} +{"train_loss": 0.018570436164736748, "grad_norm": 0.039048172533512115, "global_step": 2858, "epoch": 26, "lr": 9.928373012703926e-05} +{"train_loss": 0.018989499658346176, "grad_norm": 0.057400789111852646, "global_step": 2859, "epoch": 26, "lr": 9.928323017700207e-05} +{"train_loss": 0.021275512874126434, "grad_norm": 0.06357328593730927, "global_step": 2860, "epoch": 26, "lr": 9.928273005380502e-05} +{"train_loss": 0.020907990985883855, "grad_norm": 0.12673398852348328, "global_step": 2861, "epoch": 26, "lr": 9.928222975744991e-05, "val_loss": 0.02010398358106613} +{"train_loss": 0.017950909212231636, "grad_norm": 0.13014833629131317, "global_step": 2862, "epoch": 27, "lr": 9.928172928793846e-05} +{"train_loss": 0.024135753512382507, "grad_norm": 0.1467972695827484, "global_step": 2863, "epoch": 27, "lr": 9.928122864527245e-05} +{"train_loss": 0.02070252038538456, "grad_norm": 0.0703638568520546, "global_step": 2864, "epoch": 27, "lr": 9.928072782945363e-05} +{"train_loss": 0.01776566356420517, "grad_norm": 0.05986907705664635, "global_step": 2865, "epoch": 27, "lr": 9.928022684048379e-05} +{"train_loss": 0.020656904205679893, "grad_norm": 0.052577320486307144, "global_step": 2866, "epoch": 27, "lr": 9.927972567836463e-05} +{"train_loss": 0.020451296120882034, "grad_norm": 0.04950931668281555, "global_step": 2867, "epoch": 27, "lr": 9.927922434309797e-05} +{"train_loss": 0.017649654299020767, "grad_norm": 0.06475874781608582, "global_step": 2868, "epoch": 27, "lr": 9.927872283468553e-05} +{"train_loss": 0.020742306485772133, "grad_norm": 0.04526490345597267, "global_step": 2869, "epoch": 27, "lr": 9.927822115312909e-05} +{"train_loss": 0.021946027874946594, "grad_norm": 0.059132955968379974, "global_step": 2870, "epoch": 27, "lr": 9.927771929843042e-05} +{"train_loss": 0.022026672959327698, "grad_norm": 0.03904605284333229, "global_step": 2871, "epoch": 27, "lr": 9.927721727059127e-05} +{"train_loss": 0.022180182859301567, "grad_norm": 0.05828944966197014, "global_step": 2872, "epoch": 27, "lr": 9.927671506961342e-05} +{"train_loss": 0.020504344254732132, "grad_norm": 0.08413839340209961, "global_step": 2873, "epoch": 27, "lr": 9.92762126954986e-05} +{"train_loss": 0.020257849246263504, "grad_norm": 0.035998471081256866, "global_step": 2874, "epoch": 27, "lr": 9.927571014824861e-05} +{"train_loss": 0.020909830927848816, "grad_norm": 0.056819263845682144, "global_step": 2875, "epoch": 27, "lr": 9.927520742786521e-05} +{"train_loss": 0.02057625725865364, "grad_norm": 0.04836276173591614, "global_step": 2876, "epoch": 27, "lr": 9.927470453435014e-05} +{"train_loss": 0.022049233317375183, "grad_norm": 0.051060326397418976, "global_step": 2877, "epoch": 27, "lr": 9.927420146770522e-05} +{"train_loss": 0.0242219939827919, "grad_norm": 0.050993531942367554, "global_step": 2878, "epoch": 27, "lr": 9.927369822793216e-05} +{"train_loss": 0.019918594509363174, "grad_norm": 0.05031810700893402, "global_step": 2879, "epoch": 27, "lr": 9.927319481503277e-05} +{"train_loss": 0.022200822830200195, "grad_norm": 0.14964745938777924, "global_step": 2880, "epoch": 27, "lr": 9.927269122900879e-05} +{"train_loss": 0.02205955982208252, "grad_norm": 0.05626910552382469, "global_step": 2881, "epoch": 27, "lr": 9.9272187469862e-05} +{"train_loss": 0.023996636271476746, "grad_norm": 0.06592217832803726, "global_step": 2882, "epoch": 27, "lr": 9.927168353759417e-05} +{"train_loss": 0.018938463181257248, "grad_norm": 0.07522331178188324, "global_step": 2883, "epoch": 27, "lr": 9.927117943220707e-05} +{"train_loss": 0.019067255780100822, "grad_norm": 0.08181710541248322, "global_step": 2884, "epoch": 27, "lr": 9.927067515370248e-05} +{"train_loss": 0.02087748423218727, "grad_norm": 0.035031888633966446, "global_step": 2885, "epoch": 27, "lr": 9.927017070208218e-05} +{"train_loss": 0.02132684551179409, "grad_norm": 0.08265630155801773, "global_step": 2886, "epoch": 27, "lr": 9.92696660773479e-05} +{"train_loss": 0.022268736734986305, "grad_norm": 0.09212881326675415, "global_step": 2887, "epoch": 27, "lr": 9.926916127950146e-05} +{"train_loss": 0.019948912784457207, "grad_norm": 0.07856858521699905, "global_step": 2888, "epoch": 27, "lr": 9.92686563085446e-05} +{"train_loss": 0.022172510623931885, "grad_norm": 0.09628555923700333, "global_step": 2889, "epoch": 27, "lr": 9.926815116447911e-05} +{"train_loss": 0.022710580378770828, "grad_norm": 0.17254412174224854, "global_step": 2890, "epoch": 27, "lr": 9.926764584730677e-05} +{"train_loss": 0.023102235049009323, "grad_norm": 0.03428029268980026, "global_step": 2891, "epoch": 27, "lr": 9.926714035702934e-05} +{"train_loss": 0.02188335731625557, "grad_norm": 0.0493021123111248, "global_step": 2892, "epoch": 27, "lr": 9.926663469364861e-05} +{"train_loss": 0.019214104861021042, "grad_norm": 0.05279248580336571, "global_step": 2893, "epoch": 27, "lr": 9.926612885716635e-05} +{"train_loss": 0.01995934173464775, "grad_norm": 0.04504236951470375, "global_step": 2894, "epoch": 27, "lr": 9.926562284758433e-05} +{"train_loss": 0.02097255364060402, "grad_norm": 0.12153533101081848, "global_step": 2895, "epoch": 27, "lr": 9.926511666490436e-05} +{"train_loss": 0.018888551741838455, "grad_norm": 0.06545798480510712, "global_step": 2896, "epoch": 27, "lr": 9.926461030912818e-05} +{"train_loss": 0.02129345014691353, "grad_norm": 0.04606840759515762, "global_step": 2897, "epoch": 27, "lr": 9.926410378025758e-05} +{"train_loss": 0.021563736721873283, "grad_norm": 0.09685644507408142, "global_step": 2898, "epoch": 27, "lr": 9.926359707829435e-05} +{"train_loss": 0.02038571983575821, "grad_norm": 0.05830106511712074, "global_step": 2899, "epoch": 27, "lr": 9.926309020324025e-05} +{"train_loss": 0.021470513194799423, "grad_norm": 0.07355333119630814, "global_step": 2900, "epoch": 27, "lr": 9.926258315509709e-05} +{"train_loss": 0.019347086548805237, "grad_norm": 0.08031143248081207, "global_step": 2901, "epoch": 27, "lr": 9.926207593386664e-05} +{"train_loss": 0.018935635685920715, "grad_norm": 0.04803162440657616, "global_step": 2902, "epoch": 27, "lr": 9.926156853955067e-05} +{"train_loss": 0.021003862842917442, "grad_norm": 0.06314724683761597, "global_step": 2903, "epoch": 27, "lr": 9.926106097215098e-05} +{"train_loss": 0.019649280235171318, "grad_norm": 0.06192860007286072, "global_step": 2904, "epoch": 27, "lr": 9.926055323166933e-05} +{"train_loss": 0.023714961484074593, "grad_norm": 0.04897339269518852, "global_step": 2905, "epoch": 27, "lr": 9.926004531810752e-05} +{"train_loss": 0.022646278142929077, "grad_norm": 0.10105337202548981, "global_step": 2906, "epoch": 27, "lr": 9.925953723146735e-05} +{"train_loss": 0.020896153524518013, "grad_norm": 0.10081300139427185, "global_step": 2907, "epoch": 27, "lr": 9.925902897175057e-05} +{"train_loss": 0.020467843860387802, "grad_norm": 0.04527430608868599, "global_step": 2908, "epoch": 27, "lr": 9.9258520538959e-05} +{"train_loss": 0.02088855765759945, "grad_norm": 0.06803908199071884, "global_step": 2909, "epoch": 27, "lr": 9.925801193309441e-05} +{"train_loss": 0.02197951264679432, "grad_norm": 0.03934085741639137, "global_step": 2910, "epoch": 27, "lr": 9.925750315415858e-05} +{"train_loss": 0.02132614143192768, "grad_norm": 0.05017733573913574, "global_step": 2911, "epoch": 27, "lr": 9.92569942021533e-05} +{"train_loss": 0.018578944727778435, "grad_norm": 0.07243635505437851, "global_step": 2912, "epoch": 27, "lr": 9.925648507708038e-05} +{"train_loss": 0.02021726220846176, "grad_norm": 0.0368172787129879, "global_step": 2913, "epoch": 27, "lr": 9.925597577894157e-05} +{"train_loss": 0.024522170424461365, "grad_norm": 0.10103477537631989, "global_step": 2914, "epoch": 27, "lr": 9.92554663077387e-05} +{"train_loss": 0.02348712459206581, "grad_norm": 0.0649547204375267, "global_step": 2915, "epoch": 27, "lr": 9.925495666347353e-05} +{"train_loss": 0.02166946418583393, "grad_norm": 0.04701126739382744, "global_step": 2916, "epoch": 27, "lr": 9.925444684614789e-05} +{"train_loss": 0.023168381303548813, "grad_norm": 0.08028353750705719, "global_step": 2917, "epoch": 27, "lr": 9.925393685576352e-05} +{"train_loss": 0.01966189406812191, "grad_norm": 0.04451257735490799, "global_step": 2918, "epoch": 27, "lr": 9.925342669232223e-05} +{"train_loss": 0.01837215945124626, "grad_norm": 0.05315810814499855, "global_step": 2919, "epoch": 27, "lr": 9.925291635582583e-05} +{"train_loss": 0.0202928576618433, "grad_norm": 0.11260057985782623, "global_step": 2920, "epoch": 27, "lr": 9.925240584627609e-05} +{"train_loss": 0.017786558717489243, "grad_norm": 0.034641433507204056, "global_step": 2921, "epoch": 27, "lr": 9.925189516367484e-05} +{"train_loss": 0.02214374765753746, "grad_norm": 0.036209143698215485, "global_step": 2922, "epoch": 27, "lr": 9.925138430802383e-05} +{"train_loss": 0.019034523516893387, "grad_norm": 0.06204827129840851, "global_step": 2923, "epoch": 27, "lr": 9.925087327932487e-05} +{"train_loss": 0.01919219270348549, "grad_norm": 0.10033517330884933, "global_step": 2924, "epoch": 27, "lr": 9.925036207757978e-05} +{"train_loss": 0.02432919666171074, "grad_norm": 0.03606952726840973, "global_step": 2925, "epoch": 27, "lr": 9.924985070279031e-05} +{"train_loss": 0.023768065497279167, "grad_norm": 0.05375903472304344, "global_step": 2926, "epoch": 27, "lr": 9.92493391549583e-05} +{"train_loss": 0.017813362181186676, "grad_norm": 0.04170834273099899, "global_step": 2927, "epoch": 27, "lr": 9.924882743408553e-05} +{"train_loss": 0.01766238547861576, "grad_norm": 0.06965835392475128, "global_step": 2928, "epoch": 27, "lr": 9.924831554017379e-05} +{"train_loss": 0.018775567412376404, "grad_norm": 0.051725149154663086, "global_step": 2929, "epoch": 27, "lr": 9.92478034732249e-05} +{"train_loss": 0.01963123306632042, "grad_norm": 0.04147746041417122, "global_step": 2930, "epoch": 27, "lr": 9.924729123324064e-05} +{"train_loss": 0.019742242991924286, "grad_norm": 0.039409007877111435, "global_step": 2931, "epoch": 27, "lr": 9.924677882022282e-05} +{"train_loss": 0.020227301865816116, "grad_norm": 0.0952741876244545, "global_step": 2932, "epoch": 27, "lr": 9.924626623417323e-05} +{"train_loss": 0.01900535821914673, "grad_norm": 0.06460575759410858, "global_step": 2933, "epoch": 27, "lr": 9.924575347509368e-05} +{"train_loss": 0.020615942776203156, "grad_norm": 0.08020409941673279, "global_step": 2934, "epoch": 27, "lr": 9.924524054298596e-05} +{"train_loss": 0.023152539506554604, "grad_norm": 0.0406101755797863, "global_step": 2935, "epoch": 27, "lr": 9.924472743785189e-05} +{"train_loss": 0.020853040739893913, "grad_norm": 0.07237206399440765, "global_step": 2936, "epoch": 27, "lr": 9.924421415969326e-05} +{"train_loss": 0.019932221621274948, "grad_norm": 0.14512157440185547, "global_step": 2937, "epoch": 27, "lr": 9.924370070851188e-05} +{"train_loss": 0.021400317549705505, "grad_norm": 0.05790431424975395, "global_step": 2938, "epoch": 27, "lr": 9.924318708430955e-05} +{"train_loss": 0.02170652151107788, "grad_norm": 0.057634610682725906, "global_step": 2939, "epoch": 27, "lr": 9.924267328708808e-05} +{"train_loss": 0.02156439609825611, "grad_norm": 0.037777338176965714, "global_step": 2940, "epoch": 27, "lr": 9.924215931684926e-05} +{"train_loss": 0.023368071764707565, "grad_norm": 0.08381039649248123, "global_step": 2941, "epoch": 27, "lr": 9.924164517359491e-05} +{"train_loss": 0.02083791419863701, "grad_norm": 0.16908980906009674, "global_step": 2942, "epoch": 27, "lr": 9.924113085732684e-05} +{"train_loss": 0.018366821110248566, "grad_norm": 0.05150218307971954, "global_step": 2943, "epoch": 27, "lr": 9.924061636804684e-05} +{"train_loss": 0.021587533876299858, "grad_norm": 0.15099023282527924, "global_step": 2944, "epoch": 27, "lr": 9.924010170575674e-05} +{"train_loss": 0.020049521699547768, "grad_norm": 0.10801607370376587, "global_step": 2945, "epoch": 27, "lr": 9.923958687045833e-05} +{"train_loss": 0.021755969151854515, "grad_norm": 0.07104571163654327, "global_step": 2946, "epoch": 27, "lr": 9.923907186215342e-05} +{"train_loss": 0.01964152604341507, "grad_norm": 0.0569007433950901, "global_step": 2947, "epoch": 27, "lr": 9.923855668084384e-05} +{"train_loss": 0.018972769379615784, "grad_norm": 0.0782981887459755, "global_step": 2948, "epoch": 27, "lr": 9.923804132653138e-05} +{"train_loss": 0.01835026778280735, "grad_norm": 0.045639533549547195, "global_step": 2949, "epoch": 27, "lr": 9.923752579921786e-05} +{"train_loss": 0.020407147705554962, "grad_norm": 0.11218422651290894, "global_step": 2950, "epoch": 27, "lr": 9.923701009890507e-05} +{"train_loss": 0.02018231526017189, "grad_norm": 0.1224258542060852, "global_step": 2951, "epoch": 27, "lr": 9.923649422559484e-05} +{"train_loss": 0.025046706199645996, "grad_norm": 0.07309737056493759, "global_step": 2952, "epoch": 27, "lr": 9.9235978179289e-05} +{"train_loss": 0.02153708040714264, "grad_norm": 0.050626568496227264, "global_step": 2953, "epoch": 27, "lr": 9.923546195998934e-05} +{"train_loss": 0.020631901919841766, "grad_norm": 0.04031599685549736, "global_step": 2954, "epoch": 27, "lr": 9.923494556769767e-05} +{"train_loss": 0.022493157535791397, "grad_norm": 0.06211173161864281, "global_step": 2955, "epoch": 27, "lr": 9.923442900241582e-05} +{"train_loss": 0.01757313869893551, "grad_norm": 0.05103299021720886, "global_step": 2956, "epoch": 27, "lr": 9.92339122641456e-05} +{"train_loss": 0.021543653681874275, "grad_norm": 0.03863469883799553, "global_step": 2957, "epoch": 27, "lr": 9.923339535288883e-05} +{"train_loss": 0.01778404973447323, "grad_norm": 0.07397370785474777, "global_step": 2958, "epoch": 27, "lr": 9.92328782686473e-05} +{"train_loss": 0.021678302437067032, "grad_norm": 0.06392712891101837, "global_step": 2959, "epoch": 27, "lr": 9.923236101142286e-05} +{"train_loss": 0.021284200251102448, "grad_norm": 0.08505931496620178, "global_step": 2960, "epoch": 27, "lr": 9.92318435812173e-05} +{"train_loss": 0.021648883819580078, "grad_norm": 0.16007372736930847, "global_step": 2961, "epoch": 27, "lr": 9.923132597803246e-05} +{"train_loss": 0.01799217239022255, "grad_norm": 0.0511283241212368, "global_step": 2962, "epoch": 27, "lr": 9.923080820187016e-05} +{"train_loss": 0.02126239985227585, "grad_norm": 0.06759138405323029, "global_step": 2963, "epoch": 27, "lr": 9.92302902527322e-05} +{"train_loss": 0.021162502467632294, "grad_norm": 0.09829036891460419, "global_step": 2964, "epoch": 27, "lr": 9.922977213062041e-05} +{"train_loss": 0.024339398369193077, "grad_norm": 0.08567875623703003, "global_step": 2965, "epoch": 27, "lr": 9.922925383553661e-05} +{"train_loss": 0.01550399512052536, "grad_norm": 0.06614571064710617, "global_step": 2966, "epoch": 27, "lr": 9.922873536748263e-05} +{"train_loss": 0.020788596134703113, "grad_norm": 0.05990981683135033, "global_step": 2967, "epoch": 27, "lr": 9.922821672646027e-05, "val_loss": 0.019112946465611458} +{"train_loss": 0.023590087890625, "grad_norm": 0.059111230075359344, "global_step": 2968, "epoch": 28, "lr": 9.922769791247138e-05} +{"train_loss": 0.021164104342460632, "grad_norm": 0.07498568296432495, "global_step": 2969, "epoch": 28, "lr": 9.922717892551776e-05} +{"train_loss": 0.018632646650075912, "grad_norm": 0.030667254701256752, "global_step": 2970, "epoch": 28, "lr": 9.922665976560124e-05} +{"train_loss": 0.01995275728404522, "grad_norm": 0.09188144654035568, "global_step": 2971, "epoch": 28, "lr": 9.922614043272365e-05} +{"train_loss": 0.023826103657484055, "grad_norm": 0.06766090542078018, "global_step": 2972, "epoch": 28, "lr": 9.922562092688681e-05} +{"train_loss": 0.017927542328834534, "grad_norm": 0.09464462846517563, "global_step": 2973, "epoch": 28, "lr": 9.922510124809254e-05} +{"train_loss": 0.021349895745515823, "grad_norm": 0.052801162004470825, "global_step": 2974, "epoch": 28, "lr": 9.922458139634267e-05} +{"train_loss": 0.016226019710302353, "grad_norm": 0.035473402589559555, "global_step": 2975, "epoch": 28, "lr": 9.922406137163904e-05} +{"train_loss": 0.021593524143099785, "grad_norm": 0.17133112251758575, "global_step": 2976, "epoch": 28, "lr": 9.922354117398344e-05} +{"train_loss": 0.021954037249088287, "grad_norm": 0.08400789648294449, "global_step": 2977, "epoch": 28, "lr": 9.922302080337775e-05} +{"train_loss": 0.020842794328927994, "grad_norm": 0.043087661266326904, "global_step": 2978, "epoch": 28, "lr": 9.922250025982375e-05} +{"train_loss": 0.018881160765886307, "grad_norm": 0.06937394291162491, "global_step": 2979, "epoch": 28, "lr": 9.922197954332332e-05} +{"train_loss": 0.020342953503131866, "grad_norm": 0.04435574263334274, "global_step": 2980, "epoch": 28, "lr": 9.922145865387824e-05} +{"train_loss": 0.0221755001693964, "grad_norm": 0.0622255802154541, "global_step": 2981, "epoch": 28, "lr": 9.922093759149036e-05} +{"train_loss": 0.020661354064941406, "grad_norm": 0.03356168046593666, "global_step": 2982, "epoch": 28, "lr": 9.922041635616152e-05} +{"train_loss": 0.020443931221961975, "grad_norm": 0.07795190811157227, "global_step": 2983, "epoch": 28, "lr": 9.921989494789353e-05} +{"train_loss": 0.02003473788499832, "grad_norm": 0.1034214198589325, "global_step": 2984, "epoch": 28, "lr": 9.921937336668825e-05} +{"train_loss": 0.02070716768503189, "grad_norm": 0.08221350610256195, "global_step": 2985, "epoch": 28, "lr": 9.921885161254749e-05} +{"train_loss": 0.020297307521104813, "grad_norm": 0.09260528534650803, "global_step": 2986, "epoch": 28, "lr": 9.921832968547309e-05} +{"train_loss": 0.023196473717689514, "grad_norm": 0.13325677812099457, "global_step": 2987, "epoch": 28, "lr": 9.921780758546687e-05} +{"train_loss": 0.021026872098445892, "grad_norm": 0.06774882972240448, "global_step": 2988, "epoch": 28, "lr": 9.921728531253069e-05} +{"train_loss": 0.021750111132860184, "grad_norm": 0.061830028891563416, "global_step": 2989, "epoch": 28, "lr": 9.921676286666637e-05} +{"train_loss": 0.018254688009619713, "grad_norm": 0.0830782949924469, "global_step": 2990, "epoch": 28, "lr": 9.921624024787576e-05} +{"train_loss": 0.018829280510544777, "grad_norm": 0.04941172897815704, "global_step": 2991, "epoch": 28, "lr": 9.921571745616068e-05} +{"train_loss": 0.018878670409321785, "grad_norm": 0.044053610414266586, "global_step": 2992, "epoch": 28, "lr": 9.921519449152296e-05} +{"train_loss": 0.02044299989938736, "grad_norm": 0.059194907546043396, "global_step": 2993, "epoch": 28, "lr": 9.921467135396448e-05} +{"train_loss": 0.01695910096168518, "grad_norm": 0.05553105100989342, "global_step": 2994, "epoch": 28, "lr": 9.9214148043487e-05} +{"train_loss": 0.021759508177638054, "grad_norm": 0.04598462954163551, "global_step": 2995, "epoch": 28, "lr": 9.921362456009245e-05} +{"train_loss": 0.021011022850871086, "grad_norm": 0.07274147868156433, "global_step": 2996, "epoch": 28, "lr": 9.92131009037826e-05} +{"train_loss": 0.022451601922512054, "grad_norm": 0.045306093990802765, "global_step": 2997, "epoch": 28, "lr": 9.921257707455933e-05} +{"train_loss": 0.020476490259170532, "grad_norm": 0.07836043834686279, "global_step": 2998, "epoch": 28, "lr": 9.921205307242448e-05} +{"train_loss": 0.02299083210527897, "grad_norm": 0.07698310911655426, "global_step": 2999, "epoch": 28, "lr": 9.921152889737984e-05} +{"train_loss": 0.016368165612220764, "grad_norm": 0.08630837500095367, "global_step": 3000, "epoch": 28, "lr": 9.92110045494273e-05} +{"train_loss": 0.023241642862558365, "grad_norm": 0.05385631322860718, "global_step": 3001, "epoch": 28, "lr": 9.921048002856871e-05} +{"train_loss": 0.022083349525928497, "grad_norm": 0.06715825200080872, "global_step": 3002, "epoch": 28, "lr": 9.920995533480588e-05} +{"train_loss": 0.022879278287291527, "grad_norm": 0.042551226913928986, "global_step": 3003, "epoch": 28, "lr": 9.920943046814068e-05} +{"train_loss": 0.02052272856235504, "grad_norm": 0.1571170538663864, "global_step": 3004, "epoch": 28, "lr": 9.920890542857493e-05} +{"train_loss": 0.018794681876897812, "grad_norm": 0.035624150186777115, "global_step": 3005, "epoch": 28, "lr": 9.92083802161105e-05} +{"train_loss": 0.019130952656269073, "grad_norm": 0.04802993685007095, "global_step": 3006, "epoch": 28, "lr": 9.92078548307492e-05} +{"train_loss": 0.019907571375370026, "grad_norm": 0.07963153719902039, "global_step": 3007, "epoch": 28, "lr": 9.920732927249291e-05} +{"train_loss": 0.020381223410367966, "grad_norm": 0.03672827407717705, "global_step": 3008, "epoch": 28, "lr": 9.920680354134346e-05} +{"train_loss": 0.01789720170199871, "grad_norm": 0.07709691673517227, "global_step": 3009, "epoch": 28, "lr": 9.920627763730271e-05} +{"train_loss": 0.02056717313826084, "grad_norm": 0.10834310203790665, "global_step": 3010, "epoch": 28, "lr": 9.920575156037251e-05} +{"train_loss": 0.02510462887585163, "grad_norm": 0.04032658785581589, "global_step": 3011, "epoch": 28, "lr": 9.920522531055467e-05} +{"train_loss": 0.01822758838534355, "grad_norm": 0.07612846791744232, "global_step": 3012, "epoch": 28, "lr": 9.920469888785108e-05} +{"train_loss": 0.018186498433351517, "grad_norm": 0.03670167922973633, "global_step": 3013, "epoch": 28, "lr": 9.920417229226358e-05} +{"train_loss": 0.018455250188708305, "grad_norm": 0.0509709008038044, "global_step": 3014, "epoch": 28, "lr": 9.920364552379402e-05} +{"train_loss": 0.020996205508708954, "grad_norm": 0.040018271654844284, "global_step": 3015, "epoch": 28, "lr": 9.920311858244424e-05} +{"train_loss": 0.018363531678915024, "grad_norm": 0.09611144661903381, "global_step": 3016, "epoch": 28, "lr": 9.920259146821611e-05} +{"train_loss": 0.024750370532274246, "grad_norm": 0.05873879790306091, "global_step": 3017, "epoch": 28, "lr": 9.920206418111145e-05} +{"train_loss": 0.02175804041326046, "grad_norm": 0.03885522484779358, "global_step": 3018, "epoch": 28, "lr": 9.920153672113215e-05} +{"train_loss": 0.019751051440835, "grad_norm": 0.12153742462396622, "global_step": 3019, "epoch": 28, "lr": 9.920100908828006e-05} +{"train_loss": 0.01935291476547718, "grad_norm": 0.04760155454277992, "global_step": 3020, "epoch": 28, "lr": 9.920048128255699e-05} +{"train_loss": 0.01934235170483589, "grad_norm": 0.05583391711115837, "global_step": 3021, "epoch": 28, "lr": 9.919995330396484e-05} +{"train_loss": 0.018856674432754517, "grad_norm": 0.06029646843671799, "global_step": 3022, "epoch": 28, "lr": 9.919942515250546e-05} +{"train_loss": 0.021587084978818893, "grad_norm": 0.16312561929225922, "global_step": 3023, "epoch": 28, "lr": 9.919889682818068e-05} +{"train_loss": 0.019876904785633087, "grad_norm": 0.06616492569446564, "global_step": 3024, "epoch": 28, "lr": 9.919836833099238e-05} +{"train_loss": 0.019018402323126793, "grad_norm": 0.12458830326795578, "global_step": 3025, "epoch": 28, "lr": 9.91978396609424e-05} +{"train_loss": 0.020613983273506165, "grad_norm": 0.07195013016462326, "global_step": 3026, "epoch": 28, "lr": 9.919731081803263e-05} +{"train_loss": 0.022427529096603394, "grad_norm": 0.041563794016838074, "global_step": 3027, "epoch": 28, "lr": 9.919678180226488e-05} +{"train_loss": 0.020638292655348778, "grad_norm": 0.07574954628944397, "global_step": 3028, "epoch": 28, "lr": 9.919625261364105e-05} +{"train_loss": 0.01804334670305252, "grad_norm": 0.08560917526483536, "global_step": 3029, "epoch": 28, "lr": 9.919572325216298e-05} +{"train_loss": 0.01957191340625286, "grad_norm": 0.06419482827186584, "global_step": 3030, "epoch": 28, "lr": 9.919519371783253e-05} +{"train_loss": 0.01940033584833145, "grad_norm": 0.09954959154129028, "global_step": 3031, "epoch": 28, "lr": 9.919466401065156e-05} +{"train_loss": 0.02421833761036396, "grad_norm": 0.08547867834568024, "global_step": 3032, "epoch": 28, "lr": 9.919413413062195e-05} +{"train_loss": 0.018195178359746933, "grad_norm": 0.06673601269721985, "global_step": 3033, "epoch": 28, "lr": 9.919360407774555e-05} +{"train_loss": 0.022578321397304535, "grad_norm": 0.13449923694133759, "global_step": 3034, "epoch": 28, "lr": 9.919307385202418e-05} +{"train_loss": 0.023562870919704437, "grad_norm": 0.1660195142030716, "global_step": 3035, "epoch": 28, "lr": 9.919254345345979e-05} +{"train_loss": 0.025825290009379387, "grad_norm": 0.13228446245193481, "global_step": 3036, "epoch": 28, "lr": 9.919201288205417e-05} +{"train_loss": 0.02485446259379387, "grad_norm": 0.14524100720882416, "global_step": 3037, "epoch": 28, "lr": 9.91914821378092e-05} +{"train_loss": 0.020443923771381378, "grad_norm": 0.03909892961382866, "global_step": 3038, "epoch": 28, "lr": 9.919095122072677e-05} +{"train_loss": 0.020441200584173203, "grad_norm": 0.035832732915878296, "global_step": 3039, "epoch": 28, "lr": 9.919042013080873e-05} +{"train_loss": 0.01905195042490959, "grad_norm": 0.06444631516933441, "global_step": 3040, "epoch": 28, "lr": 9.918988886805695e-05} +{"train_loss": 0.02059614285826683, "grad_norm": 0.04076751694083214, "global_step": 3041, "epoch": 28, "lr": 9.918935743247328e-05} +{"train_loss": 0.02117939665913582, "grad_norm": 0.09178074449300766, "global_step": 3042, "epoch": 28, "lr": 9.918882582405962e-05} +{"train_loss": 0.021255269646644592, "grad_norm": 0.05138933286070824, "global_step": 3043, "epoch": 28, "lr": 9.91882940428178e-05} +{"train_loss": 0.019055910408496857, "grad_norm": 0.07890752702951431, "global_step": 3044, "epoch": 28, "lr": 9.91877620887497e-05} +{"train_loss": 0.02181241661310196, "grad_norm": 0.045474737882614136, "global_step": 3045, "epoch": 28, "lr": 9.918722996185721e-05} +{"train_loss": 0.0203925222158432, "grad_norm": 0.07467302680015564, "global_step": 3046, "epoch": 28, "lr": 9.918669766214218e-05} +{"train_loss": 0.02171958237886429, "grad_norm": 0.09205245226621628, "global_step": 3047, "epoch": 28, "lr": 9.918616518960648e-05} +{"train_loss": 0.021303167566657066, "grad_norm": 0.038363415747880936, "global_step": 3048, "epoch": 28, "lr": 9.9185632544252e-05} +{"train_loss": 0.023355424404144287, "grad_norm": 0.12543955445289612, "global_step": 3049, "epoch": 28, "lr": 9.918509972608059e-05} +{"train_loss": 0.018425099551677704, "grad_norm": 0.03794325888156891, "global_step": 3050, "epoch": 28, "lr": 9.918456673509413e-05} +{"train_loss": 0.023754049092531204, "grad_norm": 0.06589769572019577, "global_step": 3051, "epoch": 28, "lr": 9.918403357129448e-05} +{"train_loss": 0.019362596794962883, "grad_norm": 0.04818977415561676, "global_step": 3052, "epoch": 28, "lr": 9.918350023468355e-05} +{"train_loss": 0.022085024043917656, "grad_norm": 0.074797123670578, "global_step": 3053, "epoch": 28, "lr": 9.918296672526317e-05} +{"train_loss": 0.022225338965654373, "grad_norm": 0.07469302415847778, "global_step": 3054, "epoch": 28, "lr": 9.918243304303524e-05} +{"train_loss": 0.02221580594778061, "grad_norm": 0.08625751733779907, "global_step": 3055, "epoch": 28, "lr": 9.918189918800163e-05} +{"train_loss": 0.023414399474859238, "grad_norm": 0.06645453721284866, "global_step": 3056, "epoch": 28, "lr": 9.91813651601642e-05} +{"train_loss": 0.019300004467368126, "grad_norm": 0.06752899289131165, "global_step": 3057, "epoch": 28, "lr": 9.918083095952486e-05} +{"train_loss": 0.02073788456618786, "grad_norm": 0.07012233138084412, "global_step": 3058, "epoch": 28, "lr": 9.918029658608548e-05} +{"train_loss": 0.024972161278128624, "grad_norm": 0.09157383441925049, "global_step": 3059, "epoch": 28, "lr": 9.91797620398479e-05} +{"train_loss": 0.01994413137435913, "grad_norm": 0.06869574636220932, "global_step": 3060, "epoch": 28, "lr": 9.917922732081403e-05} +{"train_loss": 0.019114013761281967, "grad_norm": 0.05560527741909027, "global_step": 3061, "epoch": 28, "lr": 9.917869242898574e-05} +{"train_loss": 0.019714592024683952, "grad_norm": 0.039221763610839844, "global_step": 3062, "epoch": 28, "lr": 9.917815736436494e-05} +{"train_loss": 0.021067384630441666, "grad_norm": 0.12345432490110397, "global_step": 3063, "epoch": 28, "lr": 9.917762212695344e-05} +{"train_loss": 0.022343827411532402, "grad_norm": 0.05765571445226669, "global_step": 3064, "epoch": 28, "lr": 9.917708671675319e-05} +{"train_loss": 0.02109953761100769, "grad_norm": 0.04875463619828224, "global_step": 3065, "epoch": 28, "lr": 9.917655113376604e-05} +{"train_loss": 0.019871395081281662, "grad_norm": 0.0357547327876091, "global_step": 3066, "epoch": 28, "lr": 9.917601537799387e-05} +{"train_loss": 0.019794408231973648, "grad_norm": 0.06008252128958702, "global_step": 3067, "epoch": 28, "lr": 9.917547944943856e-05} +{"train_loss": 0.020704973489046097, "grad_norm": 0.0683378353714943, "global_step": 3068, "epoch": 28, "lr": 9.917494334810202e-05} +{"train_loss": 0.019469965249300003, "grad_norm": 0.07178512960672379, "global_step": 3069, "epoch": 28, "lr": 9.917440707398611e-05} +{"train_loss": 0.022438915446400642, "grad_norm": 0.04183098301291466, "global_step": 3070, "epoch": 28, "lr": 9.917387062709271e-05} +{"train_loss": 0.019179536029696465, "grad_norm": 0.03840434551239014, "global_step": 3071, "epoch": 28, "lr": 9.917333400742372e-05} +{"train_loss": 0.019085008651018143, "grad_norm": 0.08722531795501709, "global_step": 3072, "epoch": 28, "lr": 9.917279721498102e-05} +{"train_loss": 0.02069442471454166, "grad_norm": 0.06724988669157028, "global_step": 3073, "epoch": 28, "lr": 9.917226024976649e-05, "val_loss": 0.019881339743733406} +{"train_loss": 0.018390031531453133, "grad_norm": 0.06510378420352936, "global_step": 3074, "epoch": 29, "lr": 9.917172311178202e-05} +{"train_loss": 0.019791439175605774, "grad_norm": 0.044970590621232986, "global_step": 3075, "epoch": 29, "lr": 9.917118580102951e-05} +{"train_loss": 0.021758802235126495, "grad_norm": 0.036843132227659225, "global_step": 3076, "epoch": 29, "lr": 9.917064831751083e-05} +{"train_loss": 0.019236672669649124, "grad_norm": 0.03951190784573555, "global_step": 3077, "epoch": 29, "lr": 9.917011066122788e-05} +{"train_loss": 0.0227702297270298, "grad_norm": 0.08891212195158005, "global_step": 3078, "epoch": 29, "lr": 9.916957283218254e-05} +{"train_loss": 0.022091522812843323, "grad_norm": 0.037614915519952774, "global_step": 3079, "epoch": 29, "lr": 9.91690348303767e-05} +{"train_loss": 0.019208136945962906, "grad_norm": 0.052545417100191116, "global_step": 3080, "epoch": 29, "lr": 9.916849665581227e-05} +{"train_loss": 0.018225159496068954, "grad_norm": 0.04944999888539314, "global_step": 3081, "epoch": 29, "lr": 9.916795830849111e-05} +{"train_loss": 0.02028530463576317, "grad_norm": 0.05363676697015762, "global_step": 3082, "epoch": 29, "lr": 9.916741978841512e-05} +{"train_loss": 0.02007589489221573, "grad_norm": 0.08575926721096039, "global_step": 3083, "epoch": 29, "lr": 9.916688109558623e-05} +{"train_loss": 0.018397239968180656, "grad_norm": 0.11767305433750153, "global_step": 3084, "epoch": 29, "lr": 9.916634223000628e-05} +{"train_loss": 0.022696295753121376, "grad_norm": 0.06443207710981369, "global_step": 3085, "epoch": 29, "lr": 9.916580319167719e-05} +{"train_loss": 0.01891094073653221, "grad_norm": 0.09187324345111847, "global_step": 3086, "epoch": 29, "lr": 9.916526398060084e-05} +{"train_loss": 0.018812164664268494, "grad_norm": 0.07836944609880447, "global_step": 3087, "epoch": 29, "lr": 9.916472459677915e-05} +{"train_loss": 0.020822543650865555, "grad_norm": 0.07149286568164825, "global_step": 3088, "epoch": 29, "lr": 9.916418504021399e-05} +{"train_loss": 0.02105092443525791, "grad_norm": 0.059249717742204666, "global_step": 3089, "epoch": 29, "lr": 9.916364531090728e-05} +{"train_loss": 0.01924647018313408, "grad_norm": 0.07997260987758636, "global_step": 3090, "epoch": 29, "lr": 9.916310540886087e-05} +{"train_loss": 0.02188429795205593, "grad_norm": 0.10432577133178711, "global_step": 3091, "epoch": 29, "lr": 9.916256533407671e-05} +{"train_loss": 0.023585332557559013, "grad_norm": 0.12648805975914001, "global_step": 3092, "epoch": 29, "lr": 9.916202508655668e-05} +{"train_loss": 0.019043702632188797, "grad_norm": 0.042687442153692245, "global_step": 3093, "epoch": 29, "lr": 9.916148466630266e-05} +{"train_loss": 0.021009070798754692, "grad_norm": 0.10293291509151459, "global_step": 3094, "epoch": 29, "lr": 9.916094407331656e-05} +{"train_loss": 0.020912226289510727, "grad_norm": 0.08433117717504501, "global_step": 3095, "epoch": 29, "lr": 9.91604033076003e-05} +{"train_loss": 0.02226369082927704, "grad_norm": 0.11675220727920532, "global_step": 3096, "epoch": 29, "lr": 9.915986236915574e-05} +{"train_loss": 0.017026497051119804, "grad_norm": 0.08044414222240448, "global_step": 3097, "epoch": 29, "lr": 9.915932125798482e-05} +{"train_loss": 0.022444695234298706, "grad_norm": 0.07301385700702667, "global_step": 3098, "epoch": 29, "lr": 9.915877997408943e-05} +{"train_loss": 0.020553812384605408, "grad_norm": 0.09035904705524445, "global_step": 3099, "epoch": 29, "lr": 9.915823851747144e-05} +{"train_loss": 0.02166936732828617, "grad_norm": 0.05804343521595001, "global_step": 3100, "epoch": 29, "lr": 9.915769688813279e-05} +{"train_loss": 0.018328050151467323, "grad_norm": 0.07266714423894882, "global_step": 3101, "epoch": 29, "lr": 9.915715508607537e-05} +{"train_loss": 0.01899084635078907, "grad_norm": 0.12933209538459778, "global_step": 3102, "epoch": 29, "lr": 9.915661311130109e-05} +{"train_loss": 0.020263876765966415, "grad_norm": 0.09731384366750717, "global_step": 3103, "epoch": 29, "lr": 9.915607096381185e-05} +{"train_loss": 0.020722754299640656, "grad_norm": 0.08169025182723999, "global_step": 3104, "epoch": 29, "lr": 9.915552864360955e-05} +{"train_loss": 0.022457582876086235, "grad_norm": 0.03630679100751877, "global_step": 3105, "epoch": 29, "lr": 9.91549861506961e-05} +{"train_loss": 0.022849980741739273, "grad_norm": 0.06994353979825974, "global_step": 3106, "epoch": 29, "lr": 9.91544434850734e-05} +{"train_loss": 0.025436701253056526, "grad_norm": 0.046034667640924454, "global_step": 3107, "epoch": 29, "lr": 9.915390064674339e-05} +{"train_loss": 0.02041812241077423, "grad_norm": 0.0663706362247467, "global_step": 3108, "epoch": 29, "lr": 9.915335763570792e-05} +{"train_loss": 0.020576808601617813, "grad_norm": 0.06641724705696106, "global_step": 3109, "epoch": 29, "lr": 9.915281445196894e-05} +{"train_loss": 0.020898817107081413, "grad_norm": 0.03552168235182762, "global_step": 3110, "epoch": 29, "lr": 9.915227109552834e-05} +{"train_loss": 0.020492680370807648, "grad_norm": 0.08029231429100037, "global_step": 3111, "epoch": 29, "lr": 9.915172756638804e-05} +{"train_loss": 0.019565017893910408, "grad_norm": 0.04967207461595535, "global_step": 3112, "epoch": 29, "lr": 9.915118386454994e-05} +{"train_loss": 0.01917722448706627, "grad_norm": 0.03748832643032074, "global_step": 3113, "epoch": 29, "lr": 9.915063999001595e-05} +{"train_loss": 0.02070481702685356, "grad_norm": 0.050698619335889816, "global_step": 3114, "epoch": 29, "lr": 9.915009594278799e-05} +{"train_loss": 0.020384779199957848, "grad_norm": 0.09216028451919556, "global_step": 3115, "epoch": 29, "lr": 9.914955172286798e-05} +{"train_loss": 0.01988249644637108, "grad_norm": 0.08268100768327713, "global_step": 3116, "epoch": 29, "lr": 9.91490073302578e-05} +{"train_loss": 0.020719699561595917, "grad_norm": 0.039927296340465546, "global_step": 3117, "epoch": 29, "lr": 9.91484627649594e-05} +{"train_loss": 0.01957063004374504, "grad_norm": 0.06645160913467407, "global_step": 3118, "epoch": 29, "lr": 9.914791802697467e-05} +{"train_loss": 0.021388141438364983, "grad_norm": 0.047425877302885056, "global_step": 3119, "epoch": 29, "lr": 9.914737311630553e-05} +{"train_loss": 0.021857861429452896, "grad_norm": 0.04493491351604462, "global_step": 3120, "epoch": 29, "lr": 9.914682803295389e-05} +{"train_loss": 0.01993791200220585, "grad_norm": 0.10093113034963608, "global_step": 3121, "epoch": 29, "lr": 9.914628277692169e-05} +{"train_loss": 0.02149803936481476, "grad_norm": 0.05619879812002182, "global_step": 3122, "epoch": 29, "lr": 9.91457373482108e-05} +{"train_loss": 0.01915653981268406, "grad_norm": 0.0871187224984169, "global_step": 3123, "epoch": 29, "lr": 9.914519174682317e-05} +{"train_loss": 0.019796961918473244, "grad_norm": 0.055430252104997635, "global_step": 3124, "epoch": 29, "lr": 9.914464597276071e-05} +{"train_loss": 0.02007206343114376, "grad_norm": 0.04646870866417885, "global_step": 3125, "epoch": 29, "lr": 9.914410002602533e-05} +{"train_loss": 0.02032678946852684, "grad_norm": 0.15335004031658173, "global_step": 3126, "epoch": 29, "lr": 9.914355390661896e-05} +{"train_loss": 0.019764117896556854, "grad_norm": 0.0735526755452156, "global_step": 3127, "epoch": 29, "lr": 9.914300761454351e-05} +{"train_loss": 0.02123141847550869, "grad_norm": 0.05647887662053108, "global_step": 3128, "epoch": 29, "lr": 9.914246114980093e-05} +{"train_loss": 0.022742584347724915, "grad_norm": 0.09702441096305847, "global_step": 3129, "epoch": 29, "lr": 9.914191451239308e-05} +{"train_loss": 0.0167899951338768, "grad_norm": 0.0964093878865242, "global_step": 3130, "epoch": 29, "lr": 9.914136770232192e-05} +{"train_loss": 0.01965353637933731, "grad_norm": 0.043857086449861526, "global_step": 3131, "epoch": 29, "lr": 9.914082071958937e-05} +{"train_loss": 0.021057549864053726, "grad_norm": 0.076544851064682, "global_step": 3132, "epoch": 29, "lr": 9.914027356419735e-05} +{"train_loss": 0.021883860230445862, "grad_norm": 0.08618303388357162, "global_step": 3133, "epoch": 29, "lr": 9.913972623614778e-05} +{"train_loss": 0.022230442613363266, "grad_norm": 0.06104916334152222, "global_step": 3134, "epoch": 29, "lr": 9.913917873544256e-05} +{"train_loss": 0.02056538686156273, "grad_norm": 0.053031884133815765, "global_step": 3135, "epoch": 29, "lr": 9.913863106208366e-05} +{"train_loss": 0.020873622968792915, "grad_norm": 0.08009715378284454, "global_step": 3136, "epoch": 29, "lr": 9.913808321607297e-05} +{"train_loss": 0.018558666110038757, "grad_norm": 0.0534493662416935, "global_step": 3137, "epoch": 29, "lr": 9.913753519741244e-05} +{"train_loss": 0.02257244847714901, "grad_norm": 0.09247537702322006, "global_step": 3138, "epoch": 29, "lr": 9.913698700610397e-05} +{"train_loss": 0.021156297996640205, "grad_norm": 0.06984546780586243, "global_step": 3139, "epoch": 29, "lr": 9.913643864214949e-05} +{"train_loss": 0.019374823197722435, "grad_norm": 0.05473703891038895, "global_step": 3140, "epoch": 29, "lr": 9.913589010555093e-05} +{"train_loss": 0.02170487865805626, "grad_norm": 0.09025948494672775, "global_step": 3141, "epoch": 29, "lr": 9.913534139631025e-05} +{"train_loss": 0.020312894135713577, "grad_norm": 0.09916025400161743, "global_step": 3142, "epoch": 29, "lr": 9.913479251442932e-05} +{"train_loss": 0.021025003865361214, "grad_norm": 0.08135613799095154, "global_step": 3143, "epoch": 29, "lr": 9.91342434599101e-05} +{"train_loss": 0.018687844276428223, "grad_norm": 0.06824854761362076, "global_step": 3144, "epoch": 29, "lr": 9.913369423275452e-05} +{"train_loss": 0.017528917640447617, "grad_norm": 0.05694318935275078, "global_step": 3145, "epoch": 29, "lr": 9.913314483296452e-05} +{"train_loss": 0.0187237448990345, "grad_norm": 0.03822660073637962, "global_step": 3146, "epoch": 29, "lr": 9.9132595260542e-05} +{"train_loss": 0.02410784550011158, "grad_norm": 0.06892739981412888, "global_step": 3147, "epoch": 29, "lr": 9.913204551548892e-05} +{"train_loss": 0.01906711608171463, "grad_norm": 0.050644535571336746, "global_step": 3148, "epoch": 29, "lr": 9.913149559780719e-05} +{"train_loss": 0.019856449216604233, "grad_norm": 0.06899891793727875, "global_step": 3149, "epoch": 29, "lr": 9.913094550749875e-05} +{"train_loss": 0.022126005962491035, "grad_norm": 0.05406244471669197, "global_step": 3150, "epoch": 29, "lr": 9.913039524456553e-05} +{"train_loss": 0.021005382761359215, "grad_norm": 0.09732562303543091, "global_step": 3151, "epoch": 29, "lr": 9.912984480900948e-05} +{"train_loss": 0.020194292068481445, "grad_norm": 0.0953492820262909, "global_step": 3152, "epoch": 29, "lr": 9.912929420083252e-05} +{"train_loss": 0.020732147619128227, "grad_norm": 0.06964100897312164, "global_step": 3153, "epoch": 29, "lr": 9.912874342003658e-05} +{"train_loss": 0.019856180995702744, "grad_norm": 0.05736979469656944, "global_step": 3154, "epoch": 29, "lr": 9.91281924666236e-05} +{"train_loss": 0.01977608911693096, "grad_norm": 0.09650513529777527, "global_step": 3155, "epoch": 29, "lr": 9.912764134059553e-05} +{"train_loss": 0.01963767781853676, "grad_norm": 0.062398556619882584, "global_step": 3156, "epoch": 29, "lr": 9.912709004195428e-05} +{"train_loss": 0.020548708736896515, "grad_norm": 0.05401143804192543, "global_step": 3157, "epoch": 29, "lr": 9.91265385707018e-05} +{"train_loss": 0.021727854385972023, "grad_norm": 0.09437859058380127, "global_step": 3158, "epoch": 29, "lr": 9.912598692684003e-05} +{"train_loss": 0.020239952951669693, "grad_norm": 0.10582850873470306, "global_step": 3159, "epoch": 29, "lr": 9.912543511037091e-05} +{"train_loss": 0.02615952491760254, "grad_norm": 0.04317118972539902, "global_step": 3160, "epoch": 29, "lr": 9.912488312129637e-05} +{"train_loss": 0.02561201900243759, "grad_norm": 0.13102558255195618, "global_step": 3161, "epoch": 29, "lr": 9.912433095961836e-05} +{"train_loss": 0.018420446664094925, "grad_norm": 0.08393602818250656, "global_step": 3162, "epoch": 29, "lr": 9.912377862533881e-05} +{"train_loss": 0.02070348523557186, "grad_norm": 0.03527577593922615, "global_step": 3163, "epoch": 29, "lr": 9.912322611845968e-05} +{"train_loss": 0.0222773440182209, "grad_norm": 0.0796484500169754, "global_step": 3164, "epoch": 29, "lr": 9.912267343898287e-05} +{"train_loss": 0.021329190582036972, "grad_norm": 0.07644373923540115, "global_step": 3165, "epoch": 29, "lr": 9.912212058691037e-05} +{"train_loss": 0.01838529482483864, "grad_norm": 0.07288996130228043, "global_step": 3166, "epoch": 29, "lr": 9.912156756224408e-05} +{"train_loss": 0.019152913242578506, "grad_norm": 0.04406018555164337, "global_step": 3167, "epoch": 29, "lr": 9.912101436498598e-05} +{"train_loss": 0.02236252650618553, "grad_norm": 0.05819476768374443, "global_step": 3168, "epoch": 29, "lr": 9.9120460995138e-05} +{"train_loss": 0.01881086826324463, "grad_norm": 0.07994596660137177, "global_step": 3169, "epoch": 29, "lr": 9.911990745270208e-05} +{"train_loss": 0.021931473165750504, "grad_norm": 0.07591958343982697, "global_step": 3170, "epoch": 29, "lr": 9.911935373768016e-05} +{"train_loss": 0.022227060049772263, "grad_norm": 0.0422712080180645, "global_step": 3171, "epoch": 29, "lr": 9.911879985007419e-05} +{"train_loss": 0.023280225694179535, "grad_norm": 0.041914865374565125, "global_step": 3172, "epoch": 29, "lr": 9.911824578988613e-05} +{"train_loss": 0.023550039157271385, "grad_norm": 0.13737253844738007, "global_step": 3173, "epoch": 29, "lr": 9.911769155711792e-05} +{"train_loss": 0.01705743372440338, "grad_norm": 0.07776522636413574, "global_step": 3174, "epoch": 29, "lr": 9.911713715177149e-05} +{"train_loss": 0.018524184823036194, "grad_norm": 0.07007826119661331, "global_step": 3175, "epoch": 29, "lr": 9.91165825738488e-05} +{"train_loss": 0.020175980404019356, "grad_norm": 0.06375216692686081, "global_step": 3176, "epoch": 29, "lr": 9.911602782335181e-05} +{"train_loss": 0.021514514461159706, "grad_norm": 0.10972446203231812, "global_step": 3177, "epoch": 29, "lr": 9.911547290028245e-05} +{"train_loss": 0.020917333662509918, "grad_norm": 0.06464816629886627, "global_step": 3178, "epoch": 29, "lr": 9.911491780464267e-05} +{"train_loss": 0.02060921101848472, "grad_norm": 0.0752539411187172, "global_step": 3179, "epoch": 29, "lr": 9.911436253643445e-05, "val_loss": 0.019335784018039703} +{"train_loss": 0.02083258330821991, "grad_norm": 0.08939088881015778, "global_step": 3180, "epoch": 30, "lr": 9.91138070956597e-05} +{"train_loss": 0.01817382499575615, "grad_norm": 0.037294428795576096, "global_step": 3181, "epoch": 30, "lr": 9.911325148232039e-05} +{"train_loss": 0.01827286183834076, "grad_norm": 0.06683021038770676, "global_step": 3182, "epoch": 30, "lr": 9.911269569641848e-05} +{"train_loss": 0.023328781127929688, "grad_norm": 0.07089824229478836, "global_step": 3183, "epoch": 30, "lr": 9.911213973795592e-05} +{"train_loss": 0.01899039000272751, "grad_norm": 0.04790060222148895, "global_step": 3184, "epoch": 30, "lr": 9.911158360693465e-05} +{"train_loss": 0.02093679830431938, "grad_norm": 0.09319771081209183, "global_step": 3185, "epoch": 30, "lr": 9.911102730335664e-05} +{"train_loss": 0.018284058198332787, "grad_norm": 0.04429972916841507, "global_step": 3186, "epoch": 30, "lr": 9.911047082722384e-05} +{"train_loss": 0.020888349041342735, "grad_norm": 0.12689407169818878, "global_step": 3187, "epoch": 30, "lr": 9.91099141785382e-05} +{"train_loss": 0.020109601318836212, "grad_norm": 0.053152091801166534, "global_step": 3188, "epoch": 30, "lr": 9.910935735730167e-05} +{"train_loss": 0.019102299585938454, "grad_norm": 0.057595979422330856, "global_step": 3189, "epoch": 30, "lr": 9.910880036351623e-05} +{"train_loss": 0.020125826820731163, "grad_norm": 0.09327531605958939, "global_step": 3190, "epoch": 30, "lr": 9.910824319718381e-05} +{"train_loss": 0.01767098158597946, "grad_norm": 0.06744962930679321, "global_step": 3191, "epoch": 30, "lr": 9.910768585830639e-05} +{"train_loss": 0.02182425931096077, "grad_norm": 0.04894200339913368, "global_step": 3192, "epoch": 30, "lr": 9.910712834688592e-05} +{"train_loss": 0.021660584956407547, "grad_norm": 0.08617792278528214, "global_step": 3193, "epoch": 30, "lr": 9.910657066292434e-05} +{"train_loss": 0.018616847693920135, "grad_norm": 0.10428449511528015, "global_step": 3194, "epoch": 30, "lr": 9.910601280642364e-05} +{"train_loss": 0.016830366104841232, "grad_norm": 0.06240934878587723, "global_step": 3195, "epoch": 30, "lr": 9.910545477738577e-05} +{"train_loss": 0.019373882561922073, "grad_norm": 0.050530463457107544, "global_step": 3196, "epoch": 30, "lr": 9.910489657581268e-05} +{"train_loss": 0.021527379751205444, "grad_norm": 0.04448417201638222, "global_step": 3197, "epoch": 30, "lr": 9.910433820170635e-05} +{"train_loss": 0.019951704889535904, "grad_norm": 0.04900217056274414, "global_step": 3198, "epoch": 30, "lr": 9.910377965506871e-05} +{"train_loss": 0.02049766108393669, "grad_norm": 0.047800157219171524, "global_step": 3199, "epoch": 30, "lr": 9.910322093590177e-05} +{"train_loss": 0.01972726359963417, "grad_norm": 0.055365703999996185, "global_step": 3200, "epoch": 30, "lr": 9.910266204420743e-05} +{"train_loss": 0.02045060321688652, "grad_norm": 0.10798980295658112, "global_step": 3201, "epoch": 30, "lr": 9.910210297998772e-05} +{"train_loss": 0.02288561873137951, "grad_norm": 0.12126195430755615, "global_step": 3202, "epoch": 30, "lr": 9.910154374324456e-05} +{"train_loss": 0.019506340846419334, "grad_norm": 0.05213884264230728, "global_step": 3203, "epoch": 30, "lr": 9.910098433397994e-05} +{"train_loss": 0.020129714161157608, "grad_norm": 0.048889387398958206, "global_step": 3204, "epoch": 30, "lr": 9.910042475219581e-05} +{"train_loss": 0.021343383938074112, "grad_norm": 0.06427138298749924, "global_step": 3205, "epoch": 30, "lr": 9.909986499789413e-05} +{"train_loss": 0.021263664588332176, "grad_norm": 0.1032581627368927, "global_step": 3206, "epoch": 30, "lr": 9.909930507107689e-05} +{"train_loss": 0.017859332263469696, "grad_norm": 0.05490065738558769, "global_step": 3207, "epoch": 30, "lr": 9.909874497174604e-05} +{"train_loss": 0.01713603548705578, "grad_norm": 0.06550886482000351, "global_step": 3208, "epoch": 30, "lr": 9.909818469990356e-05} +{"train_loss": 0.019698020070791245, "grad_norm": 0.0816376656293869, "global_step": 3209, "epoch": 30, "lr": 9.90976242555514e-05} +{"train_loss": 0.020752325654029846, "grad_norm": 0.045856017619371414, "global_step": 3210, "epoch": 30, "lr": 9.909706363869153e-05} +{"train_loss": 0.020765697583556175, "grad_norm": 0.06325510889291763, "global_step": 3211, "epoch": 30, "lr": 9.909650284932595e-05} +{"train_loss": 0.018656786531209946, "grad_norm": 0.04830297455191612, "global_step": 3212, "epoch": 30, "lr": 9.90959418874566e-05} +{"train_loss": 0.021930847316980362, "grad_norm": 0.04131472855806351, "global_step": 3213, "epoch": 30, "lr": 9.909538075308547e-05} +{"train_loss": 0.02117760479450226, "grad_norm": 0.06105712056159973, "global_step": 3214, "epoch": 30, "lr": 9.909481944621452e-05} +{"train_loss": 0.019264668226242065, "grad_norm": 0.08635301887989044, "global_step": 3215, "epoch": 30, "lr": 9.909425796684573e-05} +{"train_loss": 0.02249019965529442, "grad_norm": 0.06355714052915573, "global_step": 3216, "epoch": 30, "lr": 9.909369631498107e-05} +{"train_loss": 0.020534932613372803, "grad_norm": 0.0439179427921772, "global_step": 3217, "epoch": 30, "lr": 9.909313449062249e-05} +{"train_loss": 0.018697217106819153, "grad_norm": 0.05774655565619469, "global_step": 3218, "epoch": 30, "lr": 9.909257249377201e-05} +{"train_loss": 0.02161804400384426, "grad_norm": 0.056853193789720535, "global_step": 3219, "epoch": 30, "lr": 9.909201032443156e-05} +{"train_loss": 0.021957600489258766, "grad_norm": 0.09449145197868347, "global_step": 3220, "epoch": 30, "lr": 9.909144798260315e-05} +{"train_loss": 0.019763147458434105, "grad_norm": 0.08339165151119232, "global_step": 3221, "epoch": 30, "lr": 9.909088546828873e-05} +{"train_loss": 0.019275832921266556, "grad_norm": 0.05038946121931076, "global_step": 3222, "epoch": 30, "lr": 9.909032278149031e-05} +{"train_loss": 0.02443738281726837, "grad_norm": 0.16477122902870178, "global_step": 3223, "epoch": 30, "lr": 9.908975992220981e-05} +{"train_loss": 0.021107666194438934, "grad_norm": 0.056347865611314774, "global_step": 3224, "epoch": 30, "lr": 9.908919689044927e-05} +{"train_loss": 0.017992373555898666, "grad_norm": 0.08462744206190109, "global_step": 3225, "epoch": 30, "lr": 9.908863368621063e-05} +{"train_loss": 0.02040877193212509, "grad_norm": 0.04845217615365982, "global_step": 3226, "epoch": 30, "lr": 9.908807030949587e-05} +{"train_loss": 0.02022462710738182, "grad_norm": 0.05282500758767128, "global_step": 3227, "epoch": 30, "lr": 9.9087506760307e-05} +{"train_loss": 0.021044515073299408, "grad_norm": 0.07338004559278488, "global_step": 3228, "epoch": 30, "lr": 9.908694303864596e-05} +{"train_loss": 0.018260672688484192, "grad_norm": 0.06167741119861603, "global_step": 3229, "epoch": 30, "lr": 9.908637914451477e-05} +{"train_loss": 0.022260330617427826, "grad_norm": 0.038378067314624786, "global_step": 3230, "epoch": 30, "lr": 9.908581507791536e-05} +{"train_loss": 0.02227003127336502, "grad_norm": 0.05595250427722931, "global_step": 3231, "epoch": 30, "lr": 9.908525083884978e-05} +{"train_loss": 0.021163035184144974, "grad_norm": 0.04669836536049843, "global_step": 3232, "epoch": 30, "lr": 9.908468642731995e-05} +{"train_loss": 0.021984999999403954, "grad_norm": 0.0468856617808342, "global_step": 3233, "epoch": 30, "lr": 9.908412184332788e-05} +{"train_loss": 0.01992787793278694, "grad_norm": 0.05092359334230423, "global_step": 3234, "epoch": 30, "lr": 9.908355708687556e-05} +{"train_loss": 0.020112626254558563, "grad_norm": 0.12174816429615021, "global_step": 3235, "epoch": 30, "lr": 9.908299215796495e-05} +{"train_loss": 0.024727713316679, "grad_norm": 0.04811057820916176, "global_step": 3236, "epoch": 30, "lr": 9.908242705659808e-05} +{"train_loss": 0.021449558436870575, "grad_norm": 0.056412409991025925, "global_step": 3237, "epoch": 30, "lr": 9.908186178277689e-05} +{"train_loss": 0.01816784404218197, "grad_norm": 0.038910873234272, "global_step": 3238, "epoch": 30, "lr": 9.908129633650339e-05} +{"train_loss": 0.021449368447065353, "grad_norm": 0.08505883812904358, "global_step": 3239, "epoch": 30, "lr": 9.908073071777954e-05} +{"train_loss": 0.017747309058904648, "grad_norm": 0.036573395133018494, "global_step": 3240, "epoch": 30, "lr": 9.908016492660737e-05} +{"train_loss": 0.020664192736148834, "grad_norm": 0.0938166007399559, "global_step": 3241, "epoch": 30, "lr": 9.907959896298882e-05} +{"train_loss": 0.022071409970521927, "grad_norm": 0.04085058718919754, "global_step": 3242, "epoch": 30, "lr": 9.907903282692592e-05} +{"train_loss": 0.019777879118919373, "grad_norm": 0.1434216946363449, "global_step": 3243, "epoch": 30, "lr": 9.907846651842064e-05} +{"train_loss": 0.020435981452465057, "grad_norm": 0.06844686716794968, "global_step": 3244, "epoch": 30, "lr": 9.907790003747497e-05} +{"train_loss": 0.020504625514149666, "grad_norm": 0.074381984770298, "global_step": 3245, "epoch": 30, "lr": 9.907733338409091e-05} +{"train_loss": 0.02118041180074215, "grad_norm": 0.10196506232023239, "global_step": 3246, "epoch": 30, "lr": 9.907676655827044e-05} +{"train_loss": 0.02039968967437744, "grad_norm": 0.06274823099374771, "global_step": 3247, "epoch": 30, "lr": 9.907619956001555e-05} +{"train_loss": 0.02359590120613575, "grad_norm": 0.09218623489141464, "global_step": 3248, "epoch": 30, "lr": 9.907563238932825e-05} +{"train_loss": 0.01786261610686779, "grad_norm": 0.04802633821964264, "global_step": 3249, "epoch": 30, "lr": 9.907506504621052e-05} +{"train_loss": 0.021281570196151733, "grad_norm": 0.05363650247454643, "global_step": 3250, "epoch": 30, "lr": 9.907449753066434e-05} +{"train_loss": 0.020869795233011246, "grad_norm": 0.11466092616319656, "global_step": 3251, "epoch": 30, "lr": 9.907392984269171e-05} +{"train_loss": 0.020979568362236023, "grad_norm": 0.09619896858930588, "global_step": 3252, "epoch": 30, "lr": 9.907336198229465e-05} +{"train_loss": 0.020390726625919342, "grad_norm": 0.05119442567229271, "global_step": 3253, "epoch": 30, "lr": 9.907279394947512e-05} +{"train_loss": 0.01993625983595848, "grad_norm": 0.05715202912688255, "global_step": 3254, "epoch": 30, "lr": 9.907222574423516e-05} +{"train_loss": 0.02308768406510353, "grad_norm": 0.08994985371828079, "global_step": 3255, "epoch": 30, "lr": 9.907165736657673e-05} +{"train_loss": 0.01956753246486187, "grad_norm": 0.09282293170690536, "global_step": 3256, "epoch": 30, "lr": 9.907108881650183e-05} +{"train_loss": 0.02448960766196251, "grad_norm": 0.05513688176870346, "global_step": 3257, "epoch": 30, "lr": 9.907052009401247e-05} +{"train_loss": 0.019043289124965668, "grad_norm": 0.13087405264377594, "global_step": 3258, "epoch": 30, "lr": 9.906995119911063e-05} +{"train_loss": 0.02271411567926407, "grad_norm": 0.04312197491526604, "global_step": 3259, "epoch": 30, "lr": 9.906938213179833e-05} +{"train_loss": 0.022501982748508453, "grad_norm": 0.04279733821749687, "global_step": 3260, "epoch": 30, "lr": 9.906881289207757e-05} +{"train_loss": 0.0188538059592247, "grad_norm": 0.045863792300224304, "global_step": 3261, "epoch": 30, "lr": 9.906824347995033e-05} +{"train_loss": 0.019619733095169067, "grad_norm": 0.07068983465433121, "global_step": 3262, "epoch": 30, "lr": 9.906767389541863e-05} +{"train_loss": 0.020014559850096703, "grad_norm": 0.06336627155542374, "global_step": 3263, "epoch": 30, "lr": 9.906710413848446e-05} +{"train_loss": 0.018657736480236053, "grad_norm": 0.06396156549453735, "global_step": 3264, "epoch": 30, "lr": 9.906653420914982e-05} +{"train_loss": 0.018466277047991753, "grad_norm": 0.06093573570251465, "global_step": 3265, "epoch": 30, "lr": 9.90659641074167e-05} +{"train_loss": 0.019363220781087875, "grad_norm": 0.06131705641746521, "global_step": 3266, "epoch": 30, "lr": 9.906539383328715e-05} +{"train_loss": 0.021961607038974762, "grad_norm": 0.0339675173163414, "global_step": 3267, "epoch": 30, "lr": 9.906482338676313e-05} +{"train_loss": 0.025596365332603455, "grad_norm": 0.06520602107048035, "global_step": 3268, "epoch": 30, "lr": 9.906425276784666e-05} +{"train_loss": 0.019936949014663696, "grad_norm": 0.057285718619823456, "global_step": 3269, "epoch": 30, "lr": 9.906368197653974e-05} +{"train_loss": 0.02148433029651642, "grad_norm": 0.05172949284315109, "global_step": 3270, "epoch": 30, "lr": 9.90631110128444e-05} +{"train_loss": 0.022907473146915436, "grad_norm": 0.06368543952703476, "global_step": 3271, "epoch": 30, "lr": 9.90625398767626e-05} +{"train_loss": 0.025935811921954155, "grad_norm": 0.05555539205670357, "global_step": 3272, "epoch": 30, "lr": 9.906196856829639e-05} +{"train_loss": 0.019716542214155197, "grad_norm": 0.09602795541286469, "global_step": 3273, "epoch": 30, "lr": 9.906139708744775e-05} +{"train_loss": 0.018860578536987305, "grad_norm": 0.08953012526035309, "global_step": 3274, "epoch": 30, "lr": 9.906082543421869e-05} +{"train_loss": 0.021470634266734123, "grad_norm": 0.04817565903067589, "global_step": 3275, "epoch": 30, "lr": 9.906025360861123e-05} +{"train_loss": 0.021023044362664223, "grad_norm": 0.1408270299434662, "global_step": 3276, "epoch": 30, "lr": 9.905968161062737e-05} +{"train_loss": 0.022394053637981415, "grad_norm": 0.09881946444511414, "global_step": 3277, "epoch": 30, "lr": 9.905910944026913e-05} +{"train_loss": 0.019177939742803574, "grad_norm": 0.03184397891163826, "global_step": 3278, "epoch": 30, "lr": 9.905853709753852e-05} +{"train_loss": 0.020801132544875145, "grad_norm": 0.03730081394314766, "global_step": 3279, "epoch": 30, "lr": 9.905796458243755e-05} +{"train_loss": 0.019902199506759644, "grad_norm": 0.07068603485822678, "global_step": 3280, "epoch": 30, "lr": 9.905739189496822e-05} +{"train_loss": 0.01946665719151497, "grad_norm": 0.04533208906650543, "global_step": 3281, "epoch": 30, "lr": 9.905681903513254e-05} +{"train_loss": 0.020858582109212875, "grad_norm": 0.11890362203121185, "global_step": 3282, "epoch": 30, "lr": 9.905624600293254e-05} +{"train_loss": 0.02054523304104805, "grad_norm": 0.11483965069055557, "global_step": 3283, "epoch": 30, "lr": 9.905567279837023e-05} +{"train_loss": 0.018950391560792923, "grad_norm": 0.1274765133857727, "global_step": 3284, "epoch": 30, "lr": 9.905509942144762e-05} +{"train_loss": 0.020514507285969437, "grad_norm": 0.05335740000009537, "global_step": 3285, "epoch": 30, "lr": 9.90545258721667e-05, "train_batch_mse_error": 0.019273489713668823, "val_loss": 0.020149080082774162} +{"train_loss": 0.0215495266020298, "grad_norm": 0.04058241471648216, "global_step": 3286, "epoch": 31, "lr": 9.905395215052954e-05} +{"train_loss": 0.021615760400891304, "grad_norm": 0.11372960358858109, "global_step": 3287, "epoch": 31, "lr": 9.905337825653811e-05} +{"train_loss": 0.021942541003227234, "grad_norm": 0.13152191042900085, "global_step": 3288, "epoch": 31, "lr": 9.905280419019444e-05} +{"train_loss": 0.02284085936844349, "grad_norm": 0.04770190268754959, "global_step": 3289, "epoch": 31, "lr": 9.905222995150056e-05} +{"train_loss": 0.02127465046942234, "grad_norm": 0.04659094661474228, "global_step": 3290, "epoch": 31, "lr": 9.905165554045848e-05} +{"train_loss": 0.020099161192774773, "grad_norm": 0.0848003402352333, "global_step": 3291, "epoch": 31, "lr": 9.905108095707019e-05} +{"train_loss": 0.021334385499358177, "grad_norm": 0.06984314322471619, "global_step": 3292, "epoch": 31, "lr": 9.905050620133776e-05} +{"train_loss": 0.022607658058404922, "grad_norm": 0.10610458254814148, "global_step": 3293, "epoch": 31, "lr": 9.904993127326317e-05} +{"train_loss": 0.017868647351861, "grad_norm": 0.07517058402299881, "global_step": 3294, "epoch": 31, "lr": 9.904935617284844e-05} +{"train_loss": 0.02029147371649742, "grad_norm": 0.05095576494932175, "global_step": 3295, "epoch": 31, "lr": 9.904878090009562e-05} +{"train_loss": 0.019773326814174652, "grad_norm": 0.09818822145462036, "global_step": 3296, "epoch": 31, "lr": 9.90482054550067e-05} +{"train_loss": 0.02238382399082184, "grad_norm": 0.04479283094406128, "global_step": 3297, "epoch": 31, "lr": 9.904762983758373e-05} +{"train_loss": 0.01869136467576027, "grad_norm": 0.06185842677950859, "global_step": 3298, "epoch": 31, "lr": 9.904705404782873e-05} +{"train_loss": 0.019820068031549454, "grad_norm": 0.04671727493405342, "global_step": 3299, "epoch": 31, "lr": 9.90464780857437e-05} +{"train_loss": 0.020568877458572388, "grad_norm": 0.05198172107338905, "global_step": 3300, "epoch": 31, "lr": 9.904590195133066e-05} +{"train_loss": 0.01920769363641739, "grad_norm": 0.05087737366557121, "global_step": 3301, "epoch": 31, "lr": 9.904532564459167e-05} +{"train_loss": 0.02158306911587715, "grad_norm": 0.06422478705644608, "global_step": 3302, "epoch": 31, "lr": 9.904474916552872e-05} +{"train_loss": 0.021280936896800995, "grad_norm": 0.08912383019924164, "global_step": 3303, "epoch": 31, "lr": 9.904417251414386e-05} +{"train_loss": 0.01886477693915367, "grad_norm": 0.048997025936841965, "global_step": 3304, "epoch": 31, "lr": 9.904359569043912e-05} +{"train_loss": 0.02195119857788086, "grad_norm": 0.05085575208067894, "global_step": 3305, "epoch": 31, "lr": 9.90430186944165e-05} +{"train_loss": 0.02001461386680603, "grad_norm": 0.04419524967670441, "global_step": 3306, "epoch": 31, "lr": 9.904244152607803e-05} +{"train_loss": 0.022310655564069748, "grad_norm": 0.08660826832056046, "global_step": 3307, "epoch": 31, "lr": 9.904186418542576e-05} +{"train_loss": 0.01900564879179001, "grad_norm": 0.06586828827857971, "global_step": 3308, "epoch": 31, "lr": 9.90412866724617e-05} +{"train_loss": 0.019911041483283043, "grad_norm": 0.04313618317246437, "global_step": 3309, "epoch": 31, "lr": 9.90407089871879e-05} +{"train_loss": 0.01869122125208378, "grad_norm": 0.07031287252902985, "global_step": 3310, "epoch": 31, "lr": 9.904013112960637e-05} +{"train_loss": 0.02225659228861332, "grad_norm": 0.05864483118057251, "global_step": 3311, "epoch": 31, "lr": 9.903955309971914e-05} +{"train_loss": 0.023169241845607758, "grad_norm": 0.04306577146053314, "global_step": 3312, "epoch": 31, "lr": 9.903897489752825e-05} +{"train_loss": 0.021871495991945267, "grad_norm": 0.06661121547222137, "global_step": 3313, "epoch": 31, "lr": 9.903839652303573e-05} +{"train_loss": 0.01917228475213051, "grad_norm": 0.050305817276239395, "global_step": 3314, "epoch": 31, "lr": 9.90378179762436e-05} +{"train_loss": 0.025866003707051277, "grad_norm": 0.10326536744832993, "global_step": 3315, "epoch": 31, "lr": 9.903723925715392e-05} +{"train_loss": 0.02093394100666046, "grad_norm": 0.09798693656921387, "global_step": 3316, "epoch": 31, "lr": 9.90366603657687e-05} +{"train_loss": 0.022121213376522064, "grad_norm": 0.09423286467790604, "global_step": 3317, "epoch": 31, "lr": 9.903608130208998e-05} +{"train_loss": 0.022467724978923798, "grad_norm": 0.13864226639270782, "global_step": 3318, "epoch": 31, "lr": 9.90355020661198e-05} +{"train_loss": 0.02158505655825138, "grad_norm": 0.04812759906053543, "global_step": 3319, "epoch": 31, "lr": 9.903492265786019e-05} +{"train_loss": 0.022137798368930817, "grad_norm": 0.04388962313532829, "global_step": 3320, "epoch": 31, "lr": 9.903434307731318e-05} +{"train_loss": 0.023702461272478104, "grad_norm": 0.057282183319330215, "global_step": 3321, "epoch": 31, "lr": 9.903376332448082e-05} +{"train_loss": 0.022663595154881477, "grad_norm": 0.0647834911942482, "global_step": 3322, "epoch": 31, "lr": 9.903318339936514e-05} +{"train_loss": 0.021175425499677658, "grad_norm": 0.0663788840174675, "global_step": 3323, "epoch": 31, "lr": 9.903260330196818e-05} +{"train_loss": 0.021919604390859604, "grad_norm": 0.08251651376485825, "global_step": 3324, "epoch": 31, "lr": 9.903202303229197e-05} +{"train_loss": 0.02089221589267254, "grad_norm": 0.139033243060112, "global_step": 3325, "epoch": 31, "lr": 9.903144259033856e-05} +{"train_loss": 0.02019486203789711, "grad_norm": 0.07342518866062164, "global_step": 3326, "epoch": 31, "lr": 9.903086197610997e-05} +{"train_loss": 0.019965549930930138, "grad_norm": 0.08543965965509415, "global_step": 3327, "epoch": 31, "lr": 9.903028118960827e-05} +{"train_loss": 0.020161692053079605, "grad_norm": 0.04993142932653427, "global_step": 3328, "epoch": 31, "lr": 9.902970023083548e-05} +{"train_loss": 0.02051122486591339, "grad_norm": 0.09913063049316406, "global_step": 3329, "epoch": 31, "lr": 9.902911909979364e-05} +{"train_loss": 0.01920141652226448, "grad_norm": 0.03131478279829025, "global_step": 3330, "epoch": 31, "lr": 9.90285377964848e-05} +{"train_loss": 0.0214151032269001, "grad_norm": 0.04648243263363838, "global_step": 3331, "epoch": 31, "lr": 9.9027956320911e-05} +{"train_loss": 0.019977383315563202, "grad_norm": 0.08278229832649231, "global_step": 3332, "epoch": 31, "lr": 9.902737467307428e-05} +{"train_loss": 0.01580018922686577, "grad_norm": 0.03356318548321724, "global_step": 3333, "epoch": 31, "lr": 9.90267928529767e-05} +{"train_loss": 0.021168602630496025, "grad_norm": 0.05859636887907982, "global_step": 3334, "epoch": 31, "lr": 9.902621086062028e-05} +{"train_loss": 0.02135055512189865, "grad_norm": 0.045777495950460434, "global_step": 3335, "epoch": 31, "lr": 9.902562869600707e-05} +{"train_loss": 0.021403390914201736, "grad_norm": 0.05474860221147537, "global_step": 3336, "epoch": 31, "lr": 9.902504635913914e-05} +{"train_loss": 0.019021308049559593, "grad_norm": 0.08048581331968307, "global_step": 3337, "epoch": 31, "lr": 9.902446385001851e-05} +{"train_loss": 0.020212016999721527, "grad_norm": 0.052494242787361145, "global_step": 3338, "epoch": 31, "lr": 9.902388116864722e-05} +{"train_loss": 0.018232762813568115, "grad_norm": 0.08723720908164978, "global_step": 3339, "epoch": 31, "lr": 9.902329831502735e-05} +{"train_loss": 0.020781416445970535, "grad_norm": 0.053958918899297714, "global_step": 3340, "epoch": 31, "lr": 9.902271528916091e-05} +{"train_loss": 0.017846550792455673, "grad_norm": 0.06262892484664917, "global_step": 3341, "epoch": 31, "lr": 9.902213209104999e-05} +{"train_loss": 0.019507519900798798, "grad_norm": 0.06377727538347244, "global_step": 3342, "epoch": 31, "lr": 9.90215487206966e-05} +{"train_loss": 0.020541079342365265, "grad_norm": 0.08869203925132751, "global_step": 3343, "epoch": 31, "lr": 9.902096517810282e-05} +{"train_loss": 0.016582198441028595, "grad_norm": 0.06727220118045807, "global_step": 3344, "epoch": 31, "lr": 9.902038146327069e-05} +{"train_loss": 0.019290562719106674, "grad_norm": 0.08917616307735443, "global_step": 3345, "epoch": 31, "lr": 9.901979757620225e-05} +{"train_loss": 0.02073999121785164, "grad_norm": 0.11409249156713486, "global_step": 3346, "epoch": 31, "lr": 9.901921351689955e-05} +{"train_loss": 0.02248748391866684, "grad_norm": 0.04195254296064377, "global_step": 3347, "epoch": 31, "lr": 9.901862928536465e-05} +{"train_loss": 0.018073026090860367, "grad_norm": 0.03727275878190994, "global_step": 3348, "epoch": 31, "lr": 9.901804488159962e-05} +{"train_loss": 0.021156955510377884, "grad_norm": 0.1722547709941864, "global_step": 3349, "epoch": 31, "lr": 9.90174603056065e-05} +{"train_loss": 0.01824360340833664, "grad_norm": 0.06533331423997879, "global_step": 3350, "epoch": 31, "lr": 9.901687555738732e-05} +{"train_loss": 0.021483875811100006, "grad_norm": 0.09449989348649979, "global_step": 3351, "epoch": 31, "lr": 9.901629063694417e-05} +{"train_loss": 0.020391374826431274, "grad_norm": 0.037355195730924606, "global_step": 3352, "epoch": 31, "lr": 9.90157055442791e-05} +{"train_loss": 0.01755410060286522, "grad_norm": 0.07445856183767319, "global_step": 3353, "epoch": 31, "lr": 9.901512027939415e-05} +{"train_loss": 0.021247154101729393, "grad_norm": 0.043890196830034256, "global_step": 3354, "epoch": 31, "lr": 9.901453484229138e-05} +{"train_loss": 0.020630326122045517, "grad_norm": 0.07115163654088974, "global_step": 3355, "epoch": 31, "lr": 9.901394923297285e-05} +{"train_loss": 0.02032153680920601, "grad_norm": 0.0631997138261795, "global_step": 3356, "epoch": 31, "lr": 9.901336345144064e-05} +{"train_loss": 0.020666630938649178, "grad_norm": 0.12105144560337067, "global_step": 3357, "epoch": 31, "lr": 9.901277749769675e-05} +{"train_loss": 0.019912559539079666, "grad_norm": 0.09454082697629929, "global_step": 3358, "epoch": 31, "lr": 9.901219137174329e-05} +{"train_loss": 0.019104136154055595, "grad_norm": 0.031358711421489716, "global_step": 3359, "epoch": 31, "lr": 9.901160507358232e-05} +{"train_loss": 0.02122785709798336, "grad_norm": 0.06082718446850777, "global_step": 3360, "epoch": 31, "lr": 9.901101860321587e-05} +{"train_loss": 0.018418917432427406, "grad_norm": 0.07175000756978989, "global_step": 3361, "epoch": 31, "lr": 9.9010431960646e-05} +{"train_loss": 0.019463161006569862, "grad_norm": 0.14202870428562164, "global_step": 3362, "epoch": 31, "lr": 9.90098451458748e-05} +{"train_loss": 0.021889137104153633, "grad_norm": 0.05518961697816849, "global_step": 3363, "epoch": 31, "lr": 9.900925815890432e-05} +{"train_loss": 0.02182590588927269, "grad_norm": 0.0688922330737114, "global_step": 3364, "epoch": 31, "lr": 9.900867099973663e-05} +{"train_loss": 0.01836797222495079, "grad_norm": 0.12948475778102875, "global_step": 3365, "epoch": 31, "lr": 9.900808366837377e-05} +{"train_loss": 0.017308535054326057, "grad_norm": 0.03998970240354538, "global_step": 3366, "epoch": 31, "lr": 9.900749616481782e-05} +{"train_loss": 0.021231092512607574, "grad_norm": 0.06984076648950577, "global_step": 3367, "epoch": 31, "lr": 9.900690848907083e-05} +{"train_loss": 0.018138986080884933, "grad_norm": 0.05605487525463104, "global_step": 3368, "epoch": 31, "lr": 9.900632064113488e-05} +{"train_loss": 0.022959712892770767, "grad_norm": 0.08805231750011444, "global_step": 3369, "epoch": 31, "lr": 9.900573262101205e-05} +{"train_loss": 0.01706140860915184, "grad_norm": 0.05926113948225975, "global_step": 3370, "epoch": 31, "lr": 9.900514442870438e-05} +{"train_loss": 0.0207752026617527, "grad_norm": 0.08101797848939896, "global_step": 3371, "epoch": 31, "lr": 9.900455606421394e-05} +{"train_loss": 0.019409585744142532, "grad_norm": 0.06759221106767654, "global_step": 3372, "epoch": 31, "lr": 9.90039675275428e-05} +{"train_loss": 0.018808651715517044, "grad_norm": 0.08847729861736298, "global_step": 3373, "epoch": 31, "lr": 9.900337881869304e-05} +{"train_loss": 0.017199676483869553, "grad_norm": 0.07631717622280121, "global_step": 3374, "epoch": 31, "lr": 9.90027899376667e-05} +{"train_loss": 0.02110639587044716, "grad_norm": 0.1484934687614441, "global_step": 3375, "epoch": 31, "lr": 9.900220088446588e-05} +{"train_loss": 0.024457117542624474, "grad_norm": 0.11298208683729172, "global_step": 3376, "epoch": 31, "lr": 9.900161165909263e-05} +{"train_loss": 0.0172625370323658, "grad_norm": 0.1040988340973854, "global_step": 3377, "epoch": 31, "lr": 9.900102226154903e-05} +{"train_loss": 0.022226719185709953, "grad_norm": 0.042392797768116, "global_step": 3378, "epoch": 31, "lr": 9.900043269183715e-05} +{"train_loss": 0.021903153508901596, "grad_norm": 0.11216073483228683, "global_step": 3379, "epoch": 31, "lr": 9.899984294995905e-05} +{"train_loss": 0.01962439902126789, "grad_norm": 0.05136485397815704, "global_step": 3380, "epoch": 31, "lr": 9.899925303591682e-05} +{"train_loss": 0.02172994241118431, "grad_norm": 0.05073394253849983, "global_step": 3381, "epoch": 31, "lr": 9.899866294971253e-05} +{"train_loss": 0.02084699086844921, "grad_norm": 0.06916381418704987, "global_step": 3382, "epoch": 31, "lr": 9.899807269134824e-05} +{"train_loss": 0.020386379212141037, "grad_norm": 0.07759898900985718, "global_step": 3383, "epoch": 31, "lr": 9.899748226082604e-05} +{"train_loss": 0.02045528218150139, "grad_norm": 0.10460882633924484, "global_step": 3384, "epoch": 31, "lr": 9.899689165814798e-05} +{"train_loss": 0.017974313348531723, "grad_norm": 0.07889486849308014, "global_step": 3385, "epoch": 31, "lr": 9.899630088331615e-05} +{"train_loss": 0.022297531366348267, "grad_norm": 0.16444897651672363, "global_step": 3386, "epoch": 31, "lr": 9.899570993633263e-05} +{"train_loss": 0.02181561291217804, "grad_norm": 0.12188770622015, "global_step": 3387, "epoch": 31, "lr": 9.89951188171995e-05} +{"train_loss": 0.018966522067785263, "grad_norm": 0.11910798400640488, "global_step": 3388, "epoch": 31, "lr": 9.899452752591883e-05} +{"train_loss": 0.020824745297431946, "grad_norm": 0.07062046229839325, "global_step": 3389, "epoch": 31, "lr": 9.89939360624927e-05} +{"train_loss": 0.02226102538406849, "grad_norm": 0.08250555396080017, "global_step": 3390, "epoch": 31, "lr": 9.899334442692317e-05} +{"train_loss": 0.020484150053476386, "grad_norm": 0.07033263146877289, "global_step": 3391, "epoch": 31, "lr": 9.899275261921234e-05, "val_loss": 0.019675301387906075} +{"train_loss": 0.021576859056949615, "grad_norm": 0.06600889563560486, "global_step": 3392, "epoch": 32, "lr": 9.89921606393623e-05} +{"train_loss": 0.0199977308511734, "grad_norm": 0.06254516541957855, "global_step": 3393, "epoch": 32, "lr": 9.89915684873751e-05} +{"train_loss": 0.019885100424289703, "grad_norm": 0.05824147164821625, "global_step": 3394, "epoch": 32, "lr": 9.899097616325284e-05} +{"train_loss": 0.023468807339668274, "grad_norm": 0.03867167606949806, "global_step": 3395, "epoch": 32, "lr": 9.899038366699757e-05} +{"train_loss": 0.01678909733891487, "grad_norm": 0.1486688107252121, "global_step": 3396, "epoch": 32, "lr": 9.898979099861143e-05} +{"train_loss": 0.01853526011109352, "grad_norm": 0.044529497623443604, "global_step": 3397, "epoch": 32, "lr": 9.898919815809647e-05} +{"train_loss": 0.019780989736318588, "grad_norm": 0.044444676488637924, "global_step": 3398, "epoch": 32, "lr": 9.898860514545475e-05} +{"train_loss": 0.019322097301483154, "grad_norm": 0.03384382650256157, "global_step": 3399, "epoch": 32, "lr": 9.898801196068839e-05} +{"train_loss": 0.021117959171533585, "grad_norm": 0.0683918297290802, "global_step": 3400, "epoch": 32, "lr": 9.898741860379946e-05} +{"train_loss": 0.021706677973270416, "grad_norm": 0.09553788602352142, "global_step": 3401, "epoch": 32, "lr": 9.898682507479003e-05} +{"train_loss": 0.02299598976969719, "grad_norm": 0.15492600202560425, "global_step": 3402, "epoch": 32, "lr": 9.898623137366222e-05} +{"train_loss": 0.01971205323934555, "grad_norm": 0.09095335751771927, "global_step": 3403, "epoch": 32, "lr": 9.89856375004181e-05} +{"train_loss": 0.01955614611506462, "grad_norm": 0.07452179491519928, "global_step": 3404, "epoch": 32, "lr": 9.898504345505973e-05} +{"train_loss": 0.019697703421115875, "grad_norm": 0.06415878981351852, "global_step": 3405, "epoch": 32, "lr": 9.898444923758923e-05} +{"train_loss": 0.02003299817442894, "grad_norm": 0.04335411638021469, "global_step": 3406, "epoch": 32, "lr": 9.898385484800869e-05} +{"train_loss": 0.020746033638715744, "grad_norm": 0.08336449414491653, "global_step": 3407, "epoch": 32, "lr": 9.898326028632016e-05} +{"train_loss": 0.019679436460137367, "grad_norm": 0.0425085723400116, "global_step": 3408, "epoch": 32, "lr": 9.898266555252578e-05} +{"train_loss": 0.022475818172097206, "grad_norm": 0.06843531876802444, "global_step": 3409, "epoch": 32, "lr": 9.898207064662761e-05} +{"train_loss": 0.024377893656492233, "grad_norm": 0.09257511049509048, "global_step": 3410, "epoch": 32, "lr": 9.898147556862774e-05} +{"train_loss": 0.020054344087839127, "grad_norm": 0.037678830325603485, "global_step": 3411, "epoch": 32, "lr": 9.898088031852828e-05} +{"train_loss": 0.021514281630516052, "grad_norm": 0.04815370589494705, "global_step": 3412, "epoch": 32, "lr": 9.898028489633129e-05} +{"train_loss": 0.022244758903980255, "grad_norm": 0.06983619183301926, "global_step": 3413, "epoch": 32, "lr": 9.89796893020389e-05} +{"train_loss": 0.020595990121364594, "grad_norm": 0.1091855838894844, "global_step": 3414, "epoch": 32, "lr": 9.897909353565316e-05} +{"train_loss": 0.019218305125832558, "grad_norm": 0.07107660919427872, "global_step": 3415, "epoch": 32, "lr": 9.897849759717619e-05} +{"train_loss": 0.02155635878443718, "grad_norm": 0.0661490261554718, "global_step": 3416, "epoch": 32, "lr": 9.897790148661008e-05} +{"train_loss": 0.01796046271920204, "grad_norm": 0.07457053661346436, "global_step": 3417, "epoch": 32, "lr": 9.897730520395694e-05} +{"train_loss": 0.018457084894180298, "grad_norm": 0.08826492726802826, "global_step": 3418, "epoch": 32, "lr": 9.897670874921884e-05} +{"train_loss": 0.017585843801498413, "grad_norm": 0.04020881652832031, "global_step": 3419, "epoch": 32, "lr": 9.897611212239788e-05} +{"train_loss": 0.02378177084028721, "grad_norm": 0.05391087010502815, "global_step": 3420, "epoch": 32, "lr": 9.897551532349616e-05} +{"train_loss": 0.02378346398472786, "grad_norm": 0.04619704931974411, "global_step": 3421, "epoch": 32, "lr": 9.897491835251578e-05} +{"train_loss": 0.02045297436416149, "grad_norm": 0.05242233723402023, "global_step": 3422, "epoch": 32, "lr": 9.897432120945884e-05} +{"train_loss": 0.01945183426141739, "grad_norm": 0.05199572816491127, "global_step": 3423, "epoch": 32, "lr": 9.897372389432745e-05} +{"train_loss": 0.020625077188014984, "grad_norm": 0.04204310104250908, "global_step": 3424, "epoch": 32, "lr": 9.897312640712366e-05} +{"train_loss": 0.018590716645121574, "grad_norm": 0.11130937933921814, "global_step": 3425, "epoch": 32, "lr": 9.897252874784961e-05} +{"train_loss": 0.019235920161008835, "grad_norm": 0.041139330714941025, "global_step": 3426, "epoch": 32, "lr": 9.89719309165074e-05} +{"train_loss": 0.024578668177127838, "grad_norm": 0.1084270253777504, "global_step": 3427, "epoch": 32, "lr": 9.897133291309912e-05} +{"train_loss": 0.020944908261299133, "grad_norm": 0.0947425588965416, "global_step": 3428, "epoch": 32, "lr": 9.897073473762688e-05} +{"train_loss": 0.021402418613433838, "grad_norm": 0.10636916011571884, "global_step": 3429, "epoch": 32, "lr": 9.897013639009277e-05} +{"train_loss": 0.017643924802541733, "grad_norm": 0.04308449476957321, "global_step": 3430, "epoch": 32, "lr": 9.896953787049888e-05} +{"train_loss": 0.02088494598865509, "grad_norm": 0.07646790146827698, "global_step": 3431, "epoch": 32, "lr": 9.896893917884735e-05} +{"train_loss": 0.02360599860548973, "grad_norm": 0.07791732996702194, "global_step": 3432, "epoch": 32, "lr": 9.896834031514026e-05} +{"train_loss": 0.022043390199542046, "grad_norm": 0.14283427596092224, "global_step": 3433, "epoch": 32, "lr": 9.89677412793797e-05} +{"train_loss": 0.01886424422264099, "grad_norm": 0.05095933750271797, "global_step": 3434, "epoch": 32, "lr": 9.896714207156782e-05} +{"train_loss": 0.019069546833634377, "grad_norm": 0.049000684171915054, "global_step": 3435, "epoch": 32, "lr": 9.896654269170669e-05} +{"train_loss": 0.01828289031982422, "grad_norm": 0.049188122153282166, "global_step": 3436, "epoch": 32, "lr": 9.896594313979844e-05} +{"train_loss": 0.019935443997383118, "grad_norm": 0.059616535902023315, "global_step": 3437, "epoch": 32, "lr": 9.896534341584513e-05} +{"train_loss": 0.019174806773662567, "grad_norm": 0.043183647096157074, "global_step": 3438, "epoch": 32, "lr": 9.89647435198489e-05} +{"train_loss": 0.020621605217456818, "grad_norm": 0.13753515481948853, "global_step": 3439, "epoch": 32, "lr": 9.896414345181188e-05} +{"train_loss": 0.021100014448165894, "grad_norm": 0.04207991063594818, "global_step": 3440, "epoch": 32, "lr": 9.896354321173613e-05} +{"train_loss": 0.019073158502578735, "grad_norm": 0.0729021281003952, "global_step": 3441, "epoch": 32, "lr": 9.896294279962379e-05} +{"train_loss": 0.021396499127149582, "grad_norm": 0.050601840019226074, "global_step": 3442, "epoch": 32, "lr": 9.896234221547696e-05} +{"train_loss": 0.022257009521126747, "grad_norm": 0.04460333660244942, "global_step": 3443, "epoch": 32, "lr": 9.896174145929776e-05} +{"train_loss": 0.020293641835451126, "grad_norm": 0.050824124366045, "global_step": 3444, "epoch": 32, "lr": 9.896114053108829e-05} +{"train_loss": 0.019973115995526314, "grad_norm": 0.08008904755115509, "global_step": 3445, "epoch": 32, "lr": 9.896053943085066e-05} +{"train_loss": 0.022154375910758972, "grad_norm": 0.09173069149255753, "global_step": 3446, "epoch": 32, "lr": 9.8959938158587e-05} +{"train_loss": 0.019978424534201622, "grad_norm": 0.04698394238948822, "global_step": 3447, "epoch": 32, "lr": 9.89593367142994e-05} +{"train_loss": 0.0211747158318758, "grad_norm": 0.04451361671090126, "global_step": 3448, "epoch": 32, "lr": 9.895873509798999e-05} +{"train_loss": 0.022106945514678955, "grad_norm": 0.055704791098833084, "global_step": 3449, "epoch": 32, "lr": 9.895813330966086e-05} +{"train_loss": 0.022804681211709976, "grad_norm": 0.06760012358427048, "global_step": 3450, "epoch": 32, "lr": 9.895753134931417e-05} +{"train_loss": 0.01924300566315651, "grad_norm": 0.05683930218219757, "global_step": 3451, "epoch": 32, "lr": 9.895692921695198e-05} +{"train_loss": 0.021095668897032738, "grad_norm": 0.042958054691553116, "global_step": 3452, "epoch": 32, "lr": 9.895632691257644e-05} +{"train_loss": 0.02046786993741989, "grad_norm": 0.04340435564517975, "global_step": 3453, "epoch": 32, "lr": 9.895572443618966e-05} +{"train_loss": 0.01896921545267105, "grad_norm": 0.04848947376012802, "global_step": 3454, "epoch": 32, "lr": 9.895512178779373e-05} +{"train_loss": 0.021091993898153305, "grad_norm": 0.08600237220525742, "global_step": 3455, "epoch": 32, "lr": 9.895451896739083e-05} +{"train_loss": 0.017514195293188095, "grad_norm": 0.09878426790237427, "global_step": 3456, "epoch": 32, "lr": 9.895391597498301e-05} +{"train_loss": 0.01905914954841137, "grad_norm": 0.07258781790733337, "global_step": 3457, "epoch": 32, "lr": 9.895331281057243e-05} +{"train_loss": 0.018615227192640305, "grad_norm": 0.059687793254852295, "global_step": 3458, "epoch": 32, "lr": 9.895270947416121e-05} +{"train_loss": 0.021760907024145126, "grad_norm": 0.05036012455821037, "global_step": 3459, "epoch": 32, "lr": 9.895210596575144e-05} +{"train_loss": 0.019846897572278976, "grad_norm": 0.05954165384173393, "global_step": 3460, "epoch": 32, "lr": 9.895150228534527e-05} +{"train_loss": 0.0164769496768713, "grad_norm": 0.08906806260347366, "global_step": 3461, "epoch": 32, "lr": 9.89508984329448e-05} +{"train_loss": 0.021403871476650238, "grad_norm": 0.04335678368806839, "global_step": 3462, "epoch": 32, "lr": 9.895029440855216e-05} +{"train_loss": 0.018641307950019836, "grad_norm": 0.12530669569969177, "global_step": 3463, "epoch": 32, "lr": 9.894969021216949e-05} +{"train_loss": 0.019136618822813034, "grad_norm": 0.11757766455411911, "global_step": 3464, "epoch": 32, "lr": 9.89490858437989e-05} +{"train_loss": 0.019473496824502945, "grad_norm": 0.052634838968515396, "global_step": 3465, "epoch": 32, "lr": 9.894848130344248e-05} +{"train_loss": 0.02013552188873291, "grad_norm": 0.051615092903375626, "global_step": 3466, "epoch": 32, "lr": 9.894787659110241e-05} +{"train_loss": 0.018372507765889168, "grad_norm": 0.046742651611566544, "global_step": 3467, "epoch": 32, "lr": 9.894727170678077e-05} +{"train_loss": 0.020156344398856163, "grad_norm": 0.05237725377082825, "global_step": 3468, "epoch": 32, "lr": 9.894666665047972e-05} +{"train_loss": 0.01855970360338688, "grad_norm": 0.13848090171813965, "global_step": 3469, "epoch": 32, "lr": 9.894606142220135e-05} +{"train_loss": 0.019873492419719696, "grad_norm": 0.04239396005868912, "global_step": 3470, "epoch": 32, "lr": 9.894545602194782e-05} +{"train_loss": 0.02197810262441635, "grad_norm": 0.04707003012299538, "global_step": 3471, "epoch": 32, "lr": 9.894485044972125e-05} +{"train_loss": 0.024201873689889908, "grad_norm": 0.058660462498664856, "global_step": 3472, "epoch": 32, "lr": 9.894424470552374e-05} +{"train_loss": 0.017865188419818878, "grad_norm": 0.04305487498641014, "global_step": 3473, "epoch": 32, "lr": 9.894363878935746e-05} +{"train_loss": 0.01959816925227642, "grad_norm": 0.08389416337013245, "global_step": 3474, "epoch": 32, "lr": 9.894303270122452e-05} +{"train_loss": 0.02132597751915455, "grad_norm": 0.052422139793634415, "global_step": 3475, "epoch": 32, "lr": 9.894242644112703e-05} +{"train_loss": 0.02075512707233429, "grad_norm": 0.13612620532512665, "global_step": 3476, "epoch": 32, "lr": 9.894182000906716e-05} +{"train_loss": 0.022359855473041534, "grad_norm": 0.05604520067572594, "global_step": 3477, "epoch": 32, "lr": 9.8941213405047e-05} +{"train_loss": 0.020300261676311493, "grad_norm": 0.03858174756169319, "global_step": 3478, "epoch": 32, "lr": 9.89406066290687e-05} +{"train_loss": 0.01868681237101555, "grad_norm": 0.057983189821243286, "global_step": 3479, "epoch": 32, "lr": 9.893999968113439e-05} +{"train_loss": 0.020105469971895218, "grad_norm": 0.09510360658168793, "global_step": 3480, "epoch": 32, "lr": 9.893939256124623e-05} +{"train_loss": 0.02332509309053421, "grad_norm": 0.0833035558462143, "global_step": 3481, "epoch": 32, "lr": 9.89387852694063e-05} +{"train_loss": 0.016661256551742554, "grad_norm": 0.054140519350767136, "global_step": 3482, "epoch": 32, "lr": 9.893817780561675e-05} +{"train_loss": 0.017504507675766945, "grad_norm": 0.040108710527420044, "global_step": 3483, "epoch": 32, "lr": 9.893757016987977e-05} +{"train_loss": 0.019819140434265137, "grad_norm": 0.03753112629055977, "global_step": 3484, "epoch": 32, "lr": 9.893696236219742e-05} +{"train_loss": 0.0243915393948555, "grad_norm": 0.046757977455854416, "global_step": 3485, "epoch": 32, "lr": 9.893635438257187e-05} +{"train_loss": 0.018183831125497818, "grad_norm": 0.06617245823144913, "global_step": 3486, "epoch": 32, "lr": 9.893574623100525e-05} +{"train_loss": 0.017270375043153763, "grad_norm": 0.03242633491754532, "global_step": 3487, "epoch": 32, "lr": 9.893513790749971e-05} +{"train_loss": 0.020389951765537262, "grad_norm": 0.10499145835638046, "global_step": 3488, "epoch": 32, "lr": 9.893452941205739e-05} +{"train_loss": 0.021658143028616905, "grad_norm": 0.047887638211250305, "global_step": 3489, "epoch": 32, "lr": 9.893392074468039e-05} +{"train_loss": 0.02002369612455368, "grad_norm": 0.07344606518745422, "global_step": 3490, "epoch": 32, "lr": 9.893331190537087e-05} +{"train_loss": 0.020169802010059357, "grad_norm": 0.04188496991991997, "global_step": 3491, "epoch": 32, "lr": 9.8932702894131e-05} +{"train_loss": 0.02090674266219139, "grad_norm": 0.042216792702674866, "global_step": 3492, "epoch": 32, "lr": 9.893209371096285e-05} +{"train_loss": 0.022138334810733795, "grad_norm": 0.09937575459480286, "global_step": 3493, "epoch": 32, "lr": 9.893148435586864e-05} +{"train_loss": 0.021431252360343933, "grad_norm": 0.06332401931285858, "global_step": 3494, "epoch": 32, "lr": 9.893087482885046e-05} +{"train_loss": 0.020851144567131996, "grad_norm": 0.06467103213071823, "global_step": 3495, "epoch": 32, "lr": 9.893026512991047e-05} +{"train_loss": 0.019858093932271004, "grad_norm": 0.06527644395828247, "global_step": 3496, "epoch": 32, "lr": 9.89296552590508e-05} +{"train_loss": 0.020337132332881667, "grad_norm": 0.12243897467851639, "global_step": 3497, "epoch": 32, "lr": 9.892904521627361e-05, "val_loss": 0.019116291776299477} +{"train_loss": 0.020159028470516205, "grad_norm": 0.03640691563487053, "global_step": 3498, "epoch": 33, "lr": 9.892843500158102e-05} +{"train_loss": 0.02127266302704811, "grad_norm": 0.042082760483026505, "global_step": 3499, "epoch": 33, "lr": 9.89278246149752e-05} +{"train_loss": 0.020578786730766296, "grad_norm": 0.08742024004459381, "global_step": 3500, "epoch": 33, "lr": 9.892721405645828e-05} +{"train_loss": 0.023517102003097534, "grad_norm": 0.06663530319929123, "global_step": 3501, "epoch": 33, "lr": 9.892660332603241e-05} +{"train_loss": 0.022454844787716866, "grad_norm": 0.06323880702257156, "global_step": 3502, "epoch": 33, "lr": 9.892599242369973e-05} +{"train_loss": 0.020705081522464752, "grad_norm": 0.04578695073723793, "global_step": 3503, "epoch": 33, "lr": 9.892538134946239e-05} +{"train_loss": 0.021186623722314835, "grad_norm": 0.04983426630496979, "global_step": 3504, "epoch": 33, "lr": 9.892477010332255e-05} +{"train_loss": 0.020109916105866432, "grad_norm": 0.05772429704666138, "global_step": 3505, "epoch": 33, "lr": 9.892415868528233e-05} +{"train_loss": 0.016891643404960632, "grad_norm": 0.06647311896085739, "global_step": 3506, "epoch": 33, "lr": 9.892354709534388e-05} +{"train_loss": 0.019131481647491455, "grad_norm": 0.0770539939403534, "global_step": 3507, "epoch": 33, "lr": 9.892293533350939e-05} +{"train_loss": 0.021148143336176872, "grad_norm": 0.05167299136519432, "global_step": 3508, "epoch": 33, "lr": 9.892232339978095e-05} +{"train_loss": 0.018645457923412323, "grad_norm": 0.06570611894130707, "global_step": 3509, "epoch": 33, "lr": 9.892171129416076e-05} +{"train_loss": 0.021495800465345383, "grad_norm": 0.047925688326358795, "global_step": 3510, "epoch": 33, "lr": 9.892109901665097e-05} +{"train_loss": 0.02052563801407814, "grad_norm": 0.08759783953428268, "global_step": 3511, "epoch": 33, "lr": 9.892048656725367e-05} +{"train_loss": 0.021761763840913773, "grad_norm": 0.09529387950897217, "global_step": 3512, "epoch": 33, "lr": 9.891987394597109e-05} +{"train_loss": 0.01914493925869465, "grad_norm": 0.09824196994304657, "global_step": 3513, "epoch": 33, "lr": 9.891926115280535e-05} +{"train_loss": 0.01991194114089012, "grad_norm": 0.144190713763237, "global_step": 3514, "epoch": 33, "lr": 9.891864818775858e-05} +{"train_loss": 0.020413924008607864, "grad_norm": 0.1039157435297966, "global_step": 3515, "epoch": 33, "lr": 9.891803505083295e-05} +{"train_loss": 0.024191921576857567, "grad_norm": 0.07826925814151764, "global_step": 3516, "epoch": 33, "lr": 9.891742174203065e-05} +{"train_loss": 0.020617514848709106, "grad_norm": 0.06937692314386368, "global_step": 3517, "epoch": 33, "lr": 9.891680826135377e-05} +{"train_loss": 0.018438681960105896, "grad_norm": 0.06679384410381317, "global_step": 3518, "epoch": 33, "lr": 9.891619460880452e-05} +{"train_loss": 0.019331194460392, "grad_norm": 0.07612353563308716, "global_step": 3519, "epoch": 33, "lr": 9.891558078438503e-05} +{"train_loss": 0.0201447494328022, "grad_norm": 0.124636709690094, "global_step": 3520, "epoch": 33, "lr": 9.891496678809746e-05} +{"train_loss": 0.019341904670000076, "grad_norm": 0.10782468318939209, "global_step": 3521, "epoch": 33, "lr": 9.891435261994397e-05} +{"train_loss": 0.018931785598397255, "grad_norm": 0.04536306485533714, "global_step": 3522, "epoch": 33, "lr": 9.891373827992672e-05} +{"train_loss": 0.016989044845104218, "grad_norm": 0.04284663498401642, "global_step": 3523, "epoch": 33, "lr": 9.891312376804786e-05} +{"train_loss": 0.02008630707859993, "grad_norm": 0.07654856145381927, "global_step": 3524, "epoch": 33, "lr": 9.891250908430956e-05} +{"train_loss": 0.01879604160785675, "grad_norm": 0.0369652695953846, "global_step": 3525, "epoch": 33, "lr": 9.891189422871396e-05} +{"train_loss": 0.022619854658842087, "grad_norm": 0.0516178272664547, "global_step": 3526, "epoch": 33, "lr": 9.891127920126327e-05} +{"train_loss": 0.02006400004029274, "grad_norm": 0.06581638008356094, "global_step": 3527, "epoch": 33, "lr": 9.891066400195959e-05} +{"train_loss": 0.019768409430980682, "grad_norm": 0.07062706351280212, "global_step": 3528, "epoch": 33, "lr": 9.891004863080511e-05} +{"train_loss": 0.018844371661543846, "grad_norm": 0.03896189108490944, "global_step": 3529, "epoch": 33, "lr": 9.890943308780199e-05} +{"train_loss": 0.019385362043976784, "grad_norm": 0.14247599244117737, "global_step": 3530, "epoch": 33, "lr": 9.890881737295239e-05} +{"train_loss": 0.019711285829544067, "grad_norm": 0.06060163304209709, "global_step": 3531, "epoch": 33, "lr": 9.890820148625847e-05} +{"train_loss": 0.0225532203912735, "grad_norm": 0.05198383703827858, "global_step": 3532, "epoch": 33, "lr": 9.89075854277224e-05} +{"train_loss": 0.018382515758275986, "grad_norm": 0.04821041226387024, "global_step": 3533, "epoch": 33, "lr": 9.890696919734637e-05} +{"train_loss": 0.016782179474830627, "grad_norm": 0.04988914728164673, "global_step": 3534, "epoch": 33, "lr": 9.89063527951325e-05} +{"train_loss": 0.02039380930364132, "grad_norm": 0.1452757567167282, "global_step": 3535, "epoch": 33, "lr": 9.890573622108296e-05} +{"train_loss": 0.020032383501529694, "grad_norm": 0.046113427728414536, "global_step": 3536, "epoch": 33, "lr": 9.890511947519996e-05} +{"train_loss": 0.020526781678199768, "grad_norm": 0.047123536467552185, "global_step": 3537, "epoch": 33, "lr": 9.890450255748562e-05} +{"train_loss": 0.021577537059783936, "grad_norm": 0.14562863111495972, "global_step": 3538, "epoch": 33, "lr": 9.890388546794213e-05} +{"train_loss": 0.0220428928732872, "grad_norm": 0.06081291288137436, "global_step": 3539, "epoch": 33, "lr": 9.890326820657167e-05} +{"train_loss": 0.018579792231321335, "grad_norm": 0.06268630176782608, "global_step": 3540, "epoch": 33, "lr": 9.890265077337637e-05} +{"train_loss": 0.020145202055573463, "grad_norm": 0.04625314101576805, "global_step": 3541, "epoch": 33, "lr": 9.890203316835844e-05} +{"train_loss": 0.02244332805275917, "grad_norm": 0.09658168256282806, "global_step": 3542, "epoch": 33, "lr": 9.890141539152003e-05} +{"train_loss": 0.019264908507466316, "grad_norm": 0.061830319464206696, "global_step": 3543, "epoch": 33, "lr": 9.89007974428633e-05} +{"train_loss": 0.01924942620098591, "grad_norm": 0.05101272836327553, "global_step": 3544, "epoch": 33, "lr": 9.890017932239044e-05} +{"train_loss": 0.019576720893383026, "grad_norm": 0.06355306506156921, "global_step": 3545, "epoch": 33, "lr": 9.889956103010362e-05} +{"train_loss": 0.0221342034637928, "grad_norm": 0.060733988881111145, "global_step": 3546, "epoch": 33, "lr": 9.8898942566005e-05} +{"train_loss": 0.01940212957561016, "grad_norm": 0.05036099627614021, "global_step": 3547, "epoch": 33, "lr": 9.889832393009678e-05} +{"train_loss": 0.018174614757299423, "grad_norm": 0.04700436443090439, "global_step": 3548, "epoch": 33, "lr": 9.88977051223811e-05} +{"train_loss": 0.021547682583332062, "grad_norm": 0.05794716998934746, "global_step": 3549, "epoch": 33, "lr": 9.889708614286014e-05} +{"train_loss": 0.02034628391265869, "grad_norm": 0.08989774435758591, "global_step": 3550, "epoch": 33, "lr": 9.88964669915361e-05} +{"train_loss": 0.019758909940719604, "grad_norm": 0.05801023542881012, "global_step": 3551, "epoch": 33, "lr": 9.889584766841114e-05} +{"train_loss": 0.0218433178961277, "grad_norm": 0.08988353610038757, "global_step": 3552, "epoch": 33, "lr": 9.889522817348742e-05} +{"train_loss": 0.022291850298643112, "grad_norm": 0.1347416788339615, "global_step": 3553, "epoch": 33, "lr": 9.889460850676716e-05} +{"train_loss": 0.019971244037151337, "grad_norm": 0.08879128098487854, "global_step": 3554, "epoch": 33, "lr": 9.889398866825248e-05} +{"train_loss": 0.019342275336384773, "grad_norm": 0.05249989032745361, "global_step": 3555, "epoch": 33, "lr": 9.889336865794559e-05} +{"train_loss": 0.019682176411151886, "grad_norm": 0.06422818452119827, "global_step": 3556, "epoch": 33, "lr": 9.889274847584868e-05} +{"train_loss": 0.018564455211162567, "grad_norm": 0.046807438135147095, "global_step": 3557, "epoch": 33, "lr": 9.889212812196391e-05} +{"train_loss": 0.019254937767982483, "grad_norm": 0.11937227845191956, "global_step": 3558, "epoch": 33, "lr": 9.889150759629346e-05} +{"train_loss": 0.019931774586439133, "grad_norm": 0.07067004591226578, "global_step": 3559, "epoch": 33, "lr": 9.889088689883952e-05} +{"train_loss": 0.020591553300619125, "grad_norm": 0.03704793378710747, "global_step": 3560, "epoch": 33, "lr": 9.889026602960427e-05} +{"train_loss": 0.021241353824734688, "grad_norm": 0.05367303267121315, "global_step": 3561, "epoch": 33, "lr": 9.888964498858988e-05} +{"train_loss": 0.01914110779762268, "grad_norm": 0.03543165326118469, "global_step": 3562, "epoch": 33, "lr": 9.888902377579854e-05} +{"train_loss": 0.0181418564170599, "grad_norm": 0.031255342066287994, "global_step": 3563, "epoch": 33, "lr": 9.888840239123243e-05} +{"train_loss": 0.020571669563651085, "grad_norm": 0.04561442881822586, "global_step": 3564, "epoch": 33, "lr": 9.888778083489374e-05} +{"train_loss": 0.018667707219719887, "grad_norm": 0.08491016179323196, "global_step": 3565, "epoch": 33, "lr": 9.888715910678465e-05} +{"train_loss": 0.022083435207605362, "grad_norm": 0.10378492623567581, "global_step": 3566, "epoch": 33, "lr": 9.888653720690732e-05} +{"train_loss": 0.019027430564165115, "grad_norm": 0.1063753068447113, "global_step": 3567, "epoch": 33, "lr": 9.8885915135264e-05} +{"train_loss": 0.019315974786877632, "grad_norm": 0.11109165847301483, "global_step": 3568, "epoch": 33, "lr": 9.888529289185679e-05} +{"train_loss": 0.019066665321588516, "grad_norm": 0.0516556091606617, "global_step": 3569, "epoch": 33, "lr": 9.888467047668796e-05} +{"train_loss": 0.020933669060468674, "grad_norm": 0.10034450143575668, "global_step": 3570, "epoch": 33, "lr": 9.888404788975963e-05} +{"train_loss": 0.02189222350716591, "grad_norm": 0.04193507879972458, "global_step": 3571, "epoch": 33, "lr": 9.888342513107403e-05} +{"train_loss": 0.019701149314641953, "grad_norm": 0.07475119829177856, "global_step": 3572, "epoch": 33, "lr": 9.888280220063333e-05} +{"train_loss": 0.01820559985935688, "grad_norm": 0.03674089163541794, "global_step": 3573, "epoch": 33, "lr": 9.888217909843973e-05} +{"train_loss": 0.0173305906355381, "grad_norm": 0.03256711736321449, "global_step": 3574, "epoch": 33, "lr": 9.888155582449539e-05} +{"train_loss": 0.020754802972078323, "grad_norm": 0.04770846664905548, "global_step": 3575, "epoch": 33, "lr": 9.888093237880253e-05} +{"train_loss": 0.02028077095746994, "grad_norm": 0.07126226276159286, "global_step": 3576, "epoch": 33, "lr": 9.888030876136333e-05} +{"train_loss": 0.025794897228479385, "grad_norm": 0.1405705362558365, "global_step": 3577, "epoch": 33, "lr": 9.887968497217998e-05} +{"train_loss": 0.020335305482149124, "grad_norm": 0.044813934713602066, "global_step": 3578, "epoch": 33, "lr": 9.887906101125468e-05} +{"train_loss": 0.019062303006649017, "grad_norm": 0.06270298361778259, "global_step": 3579, "epoch": 33, "lr": 9.887843687858962e-05} +{"train_loss": 0.021919820457696915, "grad_norm": 0.08949014544487, "global_step": 3580, "epoch": 33, "lr": 9.887781257418698e-05} +{"train_loss": 0.01948467828333378, "grad_norm": 0.0797845795750618, "global_step": 3581, "epoch": 33, "lr": 9.887718809804897e-05} +{"train_loss": 0.021757453680038452, "grad_norm": 0.06915387511253357, "global_step": 3582, "epoch": 33, "lr": 9.887656345017777e-05} +{"train_loss": 0.019643452018499374, "grad_norm": 0.07402089983224869, "global_step": 3583, "epoch": 33, "lr": 9.88759386305756e-05} +{"train_loss": 0.019215736538171768, "grad_norm": 0.07428194582462311, "global_step": 3584, "epoch": 33, "lr": 9.887531363924463e-05} +{"train_loss": 0.018926596269011497, "grad_norm": 0.07304259389638901, "global_step": 3585, "epoch": 33, "lr": 9.887468847618705e-05} +{"train_loss": 0.020440928637981415, "grad_norm": 0.04809287562966347, "global_step": 3586, "epoch": 33, "lr": 9.887406314140508e-05} +{"train_loss": 0.018040360882878304, "grad_norm": 0.03129059448838234, "global_step": 3587, "epoch": 33, "lr": 9.887343763490091e-05} +{"train_loss": 0.018993664532899857, "grad_norm": 0.1165197491645813, "global_step": 3588, "epoch": 33, "lr": 9.887281195667672e-05} +{"train_loss": 0.023597734048962593, "grad_norm": 0.09543176740407944, "global_step": 3589, "epoch": 33, "lr": 9.887218610673472e-05} +{"train_loss": 0.022964345291256905, "grad_norm": 0.04654580354690552, "global_step": 3590, "epoch": 33, "lr": 9.887156008507714e-05} +{"train_loss": 0.019877146929502487, "grad_norm": 0.03748944774270058, "global_step": 3591, "epoch": 33, "lr": 9.887093389170613e-05} +{"train_loss": 0.020727887749671936, "grad_norm": 0.053732313215732574, "global_step": 3592, "epoch": 33, "lr": 9.887030752662392e-05} +{"train_loss": 0.021036935970187187, "grad_norm": 0.05094074830412865, "global_step": 3593, "epoch": 33, "lr": 9.886968098983269e-05} +{"train_loss": 0.019205160439014435, "grad_norm": 0.06137879565358162, "global_step": 3594, "epoch": 33, "lr": 9.886905428133467e-05} +{"train_loss": 0.021924762055277824, "grad_norm": 0.09591945260763168, "global_step": 3595, "epoch": 33, "lr": 9.886842740113204e-05} +{"train_loss": 0.019151680171489716, "grad_norm": 0.04853193089365959, "global_step": 3596, "epoch": 33, "lr": 9.8867800349227e-05} +{"train_loss": 0.021046124398708344, "grad_norm": 0.06543147563934326, "global_step": 3597, "epoch": 33, "lr": 9.886717312562178e-05} +{"train_loss": 0.019711809232831, "grad_norm": 0.10103566944599152, "global_step": 3598, "epoch": 33, "lr": 9.886654573031855e-05} +{"train_loss": 0.022562116384506226, "grad_norm": 0.1415826678276062, "global_step": 3599, "epoch": 33, "lr": 9.886591816331954e-05} +{"train_loss": 0.019486023113131523, "grad_norm": 0.08126229047775269, "global_step": 3600, "epoch": 33, "lr": 9.886529042462693e-05} +{"train_loss": 0.021817583590745926, "grad_norm": 0.05168798193335533, "global_step": 3601, "epoch": 33, "lr": 9.886466251424295e-05} +{"train_loss": 0.022360369563102722, "grad_norm": 0.0943354070186615, "global_step": 3602, "epoch": 33, "lr": 9.886403443216981e-05} +{"train_loss": 0.02028497953389613, "grad_norm": 0.1448129415512085, "global_step": 3603, "epoch": 33, "lr": 9.886340617840968e-05, "val_loss": 0.018937084823846817} +{"train_loss": 0.02074461430311203, "grad_norm": 0.053959764540195465, "global_step": 3604, "epoch": 34, "lr": 9.886277775296479e-05} +{"train_loss": 0.019094711169600487, "grad_norm": 0.05708331987261772, "global_step": 3605, "epoch": 34, "lr": 9.886214915583737e-05} +{"train_loss": 0.020545002073049545, "grad_norm": 0.11054974049329758, "global_step": 3606, "epoch": 34, "lr": 9.88615203870296e-05} +{"train_loss": 0.018085001036524773, "grad_norm": 0.06261561065912247, "global_step": 3607, "epoch": 34, "lr": 9.88608914465437e-05} +{"train_loss": 0.01992003619670868, "grad_norm": 0.08208263665437698, "global_step": 3608, "epoch": 34, "lr": 9.886026233438186e-05} +{"train_loss": 0.020414074882864952, "grad_norm": 0.07974307239055634, "global_step": 3609, "epoch": 34, "lr": 9.88596330505463e-05} +{"train_loss": 0.021270843222737312, "grad_norm": 0.05194227397441864, "global_step": 3610, "epoch": 34, "lr": 9.885900359503926e-05} +{"train_loss": 0.021928131580352783, "grad_norm": 0.04566020146012306, "global_step": 3611, "epoch": 34, "lr": 9.885837396786292e-05} +{"train_loss": 0.020941201597452164, "grad_norm": 0.03634864464402199, "global_step": 3612, "epoch": 34, "lr": 9.885774416901952e-05} +{"train_loss": 0.02244492620229721, "grad_norm": 0.08303403109312057, "global_step": 3613, "epoch": 34, "lr": 9.885711419851125e-05} +{"train_loss": 0.02011600136756897, "grad_norm": 0.07354745268821716, "global_step": 3614, "epoch": 34, "lr": 9.885648405634032e-05} +{"train_loss": 0.019244730472564697, "grad_norm": 0.08432256430387497, "global_step": 3615, "epoch": 34, "lr": 9.885585374250894e-05} +{"train_loss": 0.019105223938822746, "grad_norm": 0.04417872801423073, "global_step": 3616, "epoch": 34, "lr": 9.885522325701935e-05} +{"train_loss": 0.02140812575817108, "grad_norm": 0.1368262767791748, "global_step": 3617, "epoch": 34, "lr": 9.885459259987375e-05} +{"train_loss": 0.01869579218327999, "grad_norm": 0.05684000253677368, "global_step": 3618, "epoch": 34, "lr": 9.885396177107436e-05} +{"train_loss": 0.01995834708213806, "grad_norm": 0.12386087328195572, "global_step": 3619, "epoch": 34, "lr": 9.885333077062339e-05} +{"train_loss": 0.02145252749323845, "grad_norm": 0.12618377804756165, "global_step": 3620, "epoch": 34, "lr": 9.885269959852306e-05} +{"train_loss": 0.02057674340903759, "grad_norm": 0.05223634093999863, "global_step": 3621, "epoch": 34, "lr": 9.885206825477559e-05} +{"train_loss": 0.02352844551205635, "grad_norm": 0.08627176284790039, "global_step": 3622, "epoch": 34, "lr": 9.88514367393832e-05} +{"train_loss": 0.02308390848338604, "grad_norm": 0.07209868729114532, "global_step": 3623, "epoch": 34, "lr": 9.885080505234811e-05} +{"train_loss": 0.018615394830703735, "grad_norm": 0.1011306568980217, "global_step": 3624, "epoch": 34, "lr": 9.885017319367253e-05} +{"train_loss": 0.018214451149106026, "grad_norm": 0.048660047352313995, "global_step": 3625, "epoch": 34, "lr": 9.884954116335869e-05} +{"train_loss": 0.022594312205910683, "grad_norm": 0.10032783448696136, "global_step": 3626, "epoch": 34, "lr": 9.884890896140881e-05} +{"train_loss": 0.01988375000655651, "grad_norm": 0.07601359486579895, "global_step": 3627, "epoch": 34, "lr": 9.88482765878251e-05} +{"train_loss": 0.019052287563681602, "grad_norm": 0.10620632022619247, "global_step": 3628, "epoch": 34, "lr": 9.88476440426098e-05} +{"train_loss": 0.019040968269109726, "grad_norm": 0.07553136348724365, "global_step": 3629, "epoch": 34, "lr": 9.88470113257651e-05} +{"train_loss": 0.019014393910765648, "grad_norm": 0.08644465357065201, "global_step": 3630, "epoch": 34, "lr": 9.884637843729326e-05} +{"train_loss": 0.018800007179379463, "grad_norm": 0.08826430886983871, "global_step": 3631, "epoch": 34, "lr": 9.884574537719648e-05} +{"train_loss": 0.0172707661986351, "grad_norm": 0.042116496711969376, "global_step": 3632, "epoch": 34, "lr": 9.884511214547702e-05} +{"train_loss": 0.01816265657544136, "grad_norm": 0.10051285475492477, "global_step": 3633, "epoch": 34, "lr": 9.884447874213704e-05} +{"train_loss": 0.02299535647034645, "grad_norm": 0.06972657889127731, "global_step": 3634, "epoch": 34, "lr": 9.884384516717884e-05} +{"train_loss": 0.01903371885418892, "grad_norm": 0.057710278779268265, "global_step": 3635, "epoch": 34, "lr": 9.884321142060459e-05} +{"train_loss": 0.018253689631819725, "grad_norm": 0.08298811316490173, "global_step": 3636, "epoch": 34, "lr": 9.884257750241653e-05} +{"train_loss": 0.018368976190686226, "grad_norm": 0.07307536900043488, "global_step": 3637, "epoch": 34, "lr": 9.88419434126169e-05} +{"train_loss": 0.02153194509446621, "grad_norm": 0.05266937240958214, "global_step": 3638, "epoch": 34, "lr": 9.884130915120793e-05} +{"train_loss": 0.01592528074979782, "grad_norm": 0.06246035918593407, "global_step": 3639, "epoch": 34, "lr": 9.884067471819184e-05} +{"train_loss": 0.023525003343820572, "grad_norm": 0.08305768668651581, "global_step": 3640, "epoch": 34, "lr": 9.884004011357085e-05} +{"train_loss": 0.020351592451334, "grad_norm": 0.09935080260038376, "global_step": 3641, "epoch": 34, "lr": 9.88394053373472e-05} +{"train_loss": 0.02112261950969696, "grad_norm": 0.06104739010334015, "global_step": 3642, "epoch": 34, "lr": 9.883877038952312e-05} +{"train_loss": 0.019548308104276657, "grad_norm": 0.04664069041609764, "global_step": 3643, "epoch": 34, "lr": 9.883813527010084e-05} +{"train_loss": 0.01961693912744522, "grad_norm": 0.04653262719511986, "global_step": 3644, "epoch": 34, "lr": 9.88374999790826e-05} +{"train_loss": 0.020983727648854256, "grad_norm": 0.06201261281967163, "global_step": 3645, "epoch": 34, "lr": 9.883686451647062e-05} +{"train_loss": 0.021325552836060524, "grad_norm": 0.05955370143055916, "global_step": 3646, "epoch": 34, "lr": 9.883622888226713e-05} +{"train_loss": 0.021543383598327637, "grad_norm": 0.09236010909080505, "global_step": 3647, "epoch": 34, "lr": 9.883559307647439e-05} +{"train_loss": 0.020341306924819946, "grad_norm": 0.04158018156886101, "global_step": 3648, "epoch": 34, "lr": 9.883495709909458e-05} +{"train_loss": 0.022420503199100494, "grad_norm": 0.040379978716373444, "global_step": 3649, "epoch": 34, "lr": 9.883432095013e-05} +{"train_loss": 0.02429061010479927, "grad_norm": 0.07996749132871628, "global_step": 3650, "epoch": 34, "lr": 9.883368462958284e-05} +{"train_loss": 0.020732475444674492, "grad_norm": 0.05065930634737015, "global_step": 3651, "epoch": 34, "lr": 9.883304813745535e-05} +{"train_loss": 0.025410674512386322, "grad_norm": 0.047381430864334106, "global_step": 3652, "epoch": 34, "lr": 9.883241147374976e-05} +{"train_loss": 0.02005183696746826, "grad_norm": 0.04993641376495361, "global_step": 3653, "epoch": 34, "lr": 9.883177463846832e-05} +{"train_loss": 0.020053785294294357, "grad_norm": 0.05033905804157257, "global_step": 3654, "epoch": 34, "lr": 9.883113763161326e-05} +{"train_loss": 0.020965486764907837, "grad_norm": 0.058685336261987686, "global_step": 3655, "epoch": 34, "lr": 9.883050045318683e-05} +{"train_loss": 0.019266411662101746, "grad_norm": 0.06416407227516174, "global_step": 3656, "epoch": 34, "lr": 9.882986310319124e-05} +{"train_loss": 0.0194278284907341, "grad_norm": 0.1819148063659668, "global_step": 3657, "epoch": 34, "lr": 9.882922558162875e-05} +{"train_loss": 0.01855020597577095, "grad_norm": 0.04279875010251999, "global_step": 3658, "epoch": 34, "lr": 9.88285878885016e-05} +{"train_loss": 0.018617281690239906, "grad_norm": 0.06628809869289398, "global_step": 3659, "epoch": 34, "lr": 9.882795002381203e-05} +{"train_loss": 0.020527714863419533, "grad_norm": 0.0523524172604084, "global_step": 3660, "epoch": 34, "lr": 9.882731198756226e-05} +{"train_loss": 0.020270535722374916, "grad_norm": 0.13237103819847107, "global_step": 3661, "epoch": 34, "lr": 9.882667377975457e-05} +{"train_loss": 0.02122349664568901, "grad_norm": 0.05641805753111839, "global_step": 3662, "epoch": 34, "lr": 9.882603540039117e-05} +{"train_loss": 0.01991811767220497, "grad_norm": 0.08549479395151138, "global_step": 3663, "epoch": 34, "lr": 9.882539684947432e-05} +{"train_loss": 0.02334209531545639, "grad_norm": 0.05225636437535286, "global_step": 3664, "epoch": 34, "lr": 9.882475812700627e-05} +{"train_loss": 0.018071483820676804, "grad_norm": 0.058350443840026855, "global_step": 3665, "epoch": 34, "lr": 9.882411923298925e-05} +{"train_loss": 0.01975499838590622, "grad_norm": 0.060379430651664734, "global_step": 3666, "epoch": 34, "lr": 9.882348016742549e-05} +{"train_loss": 0.020716644823551178, "grad_norm": 0.10497738420963287, "global_step": 3667, "epoch": 34, "lr": 9.882284093031727e-05} +{"train_loss": 0.020957911387085915, "grad_norm": 0.06865319609642029, "global_step": 3668, "epoch": 34, "lr": 9.882220152166681e-05} +{"train_loss": 0.0193789042532444, "grad_norm": 0.1262681782245636, "global_step": 3669, "epoch": 34, "lr": 9.882156194147636e-05} +{"train_loss": 0.02310071885585785, "grad_norm": 0.17446044087409973, "global_step": 3670, "epoch": 34, "lr": 9.882092218974818e-05} +{"train_loss": 0.023164108395576477, "grad_norm": 0.05080787092447281, "global_step": 3671, "epoch": 34, "lr": 9.882028226648454e-05} +{"train_loss": 0.020727263763546944, "grad_norm": 0.07834272086620331, "global_step": 3672, "epoch": 34, "lr": 9.881964217168762e-05} +{"train_loss": 0.019713066518306732, "grad_norm": 0.04001418128609657, "global_step": 3673, "epoch": 34, "lr": 9.881900190535973e-05} +{"train_loss": 0.01977996900677681, "grad_norm": 0.08420401811599731, "global_step": 3674, "epoch": 34, "lr": 9.88183614675031e-05} +{"train_loss": 0.016863921657204628, "grad_norm": 0.04580328240990639, "global_step": 3675, "epoch": 34, "lr": 9.881772085811995e-05} +{"train_loss": 0.020803995430469513, "grad_norm": 0.0682055726647377, "global_step": 3676, "epoch": 34, "lr": 9.88170800772126e-05} +{"train_loss": 0.020646100863814354, "grad_norm": 0.04710821807384491, "global_step": 3677, "epoch": 34, "lr": 9.881643912478322e-05} +{"train_loss": 0.015759970992803574, "grad_norm": 0.08194447308778763, "global_step": 3678, "epoch": 34, "lr": 9.881579800083414e-05} +{"train_loss": 0.022731488570570946, "grad_norm": 0.0634930208325386, "global_step": 3679, "epoch": 34, "lr": 9.881515670536755e-05} +{"train_loss": 0.01629672385752201, "grad_norm": 0.08031115680932999, "global_step": 3680, "epoch": 34, "lr": 9.881451523838574e-05} +{"train_loss": 0.019174864515662193, "grad_norm": 0.03817782178521156, "global_step": 3681, "epoch": 34, "lr": 9.881387359989094e-05} +{"train_loss": 0.02077363431453705, "grad_norm": 0.06218644231557846, "global_step": 3682, "epoch": 34, "lr": 9.881323178988543e-05} +{"train_loss": 0.022841619327664375, "grad_norm": 0.10367138683795929, "global_step": 3683, "epoch": 34, "lr": 9.881258980837144e-05} +{"train_loss": 0.02140870876610279, "grad_norm": 0.07366403937339783, "global_step": 3684, "epoch": 34, "lr": 9.881194765535124e-05} +{"train_loss": 0.019298281520605087, "grad_norm": 0.05916256830096245, "global_step": 3685, "epoch": 34, "lr": 9.88113053308271e-05} +{"train_loss": 0.016749732196331024, "grad_norm": 0.09808838367462158, "global_step": 3686, "epoch": 34, "lr": 9.881066283480124e-05} +{"train_loss": 0.018376462161540985, "grad_norm": 0.09365294873714447, "global_step": 3687, "epoch": 34, "lr": 9.881002016727595e-05} +{"train_loss": 0.019910044968128204, "grad_norm": 0.11243801563978195, "global_step": 3688, "epoch": 34, "lr": 9.880937732825348e-05} +{"train_loss": 0.02125474065542221, "grad_norm": 0.07057496160268784, "global_step": 3689, "epoch": 34, "lr": 9.880873431773607e-05} +{"train_loss": 0.020793315023183823, "grad_norm": 0.08386348187923431, "global_step": 3690, "epoch": 34, "lr": 9.8808091135726e-05} +{"train_loss": 0.020651500672101974, "grad_norm": 0.03927567973732948, "global_step": 3691, "epoch": 34, "lr": 9.880744778222553e-05} +{"train_loss": 0.01799279823899269, "grad_norm": 0.05174800381064415, "global_step": 3692, "epoch": 34, "lr": 9.88068042572369e-05} +{"train_loss": 0.01970488205552101, "grad_norm": 0.08628343790769577, "global_step": 3693, "epoch": 34, "lr": 9.88061605607624e-05} +{"train_loss": 0.01668788678944111, "grad_norm": 0.09046745300292969, "global_step": 3694, "epoch": 34, "lr": 9.880551669280426e-05} +{"train_loss": 0.0199073925614357, "grad_norm": 0.05230945721268654, "global_step": 3695, "epoch": 34, "lr": 9.880487265336478e-05} +{"train_loss": 0.020358573645353317, "grad_norm": 0.11549008637666702, "global_step": 3696, "epoch": 34, "lr": 9.88042284424462e-05} +{"train_loss": 0.020757712423801422, "grad_norm": 0.06310813874006271, "global_step": 3697, "epoch": 34, "lr": 9.880358406005078e-05} +{"train_loss": 0.0202143806964159, "grad_norm": 0.1120661124587059, "global_step": 3698, "epoch": 34, "lr": 9.880293950618079e-05} +{"train_loss": 0.019727617502212524, "grad_norm": 0.05644623562693596, "global_step": 3699, "epoch": 34, "lr": 9.880229478083849e-05} +{"train_loss": 0.02163258194923401, "grad_norm": 0.036694929003715515, "global_step": 3700, "epoch": 34, "lr": 9.880164988402616e-05} +{"train_loss": 0.020283058285713196, "grad_norm": 0.09177155047655106, "global_step": 3701, "epoch": 34, "lr": 9.880100481574603e-05} +{"train_loss": 0.01804560422897339, "grad_norm": 0.043709803372621536, "global_step": 3702, "epoch": 34, "lr": 9.880035957600042e-05} +{"train_loss": 0.020917100831866264, "grad_norm": 0.08169058710336685, "global_step": 3703, "epoch": 34, "lr": 9.879971416479157e-05} +{"train_loss": 0.02170693874359131, "grad_norm": 0.046418868005275726, "global_step": 3704, "epoch": 34, "lr": 9.879906858212174e-05} +{"train_loss": 0.01847527176141739, "grad_norm": 0.07137469947338104, "global_step": 3705, "epoch": 34, "lr": 9.879842282799318e-05} +{"train_loss": 0.020949969068169594, "grad_norm": 0.09281962364912033, "global_step": 3706, "epoch": 34, "lr": 9.87977769024082e-05} +{"train_loss": 0.017319206148386, "grad_norm": 0.13138335943222046, "global_step": 3707, "epoch": 34, "lr": 9.879713080536908e-05} +{"train_loss": 0.022145193070173264, "grad_norm": 0.04867343604564667, "global_step": 3708, "epoch": 34, "lr": 9.879648453687805e-05} +{"train_loss": 0.0202042661115246, "grad_norm": 0.07651370763778687, "global_step": 3709, "epoch": 34, "lr": 9.879583809693738e-05, "val_loss": 0.01954018697142601} +{"train_loss": 0.019455548375844955, "grad_norm": 0.03753499314188957, "global_step": 3710, "epoch": 35, "lr": 9.879519148554936e-05} +{"train_loss": 0.021383941173553467, "grad_norm": 0.04539206996560097, "global_step": 3711, "epoch": 35, "lr": 9.879454470271626e-05} +{"train_loss": 0.02163485810160637, "grad_norm": 0.04008800536394119, "global_step": 3712, "epoch": 35, "lr": 9.879389774844034e-05} +{"train_loss": 0.021782170981168747, "grad_norm": 0.06623690575361252, "global_step": 3713, "epoch": 35, "lr": 9.879325062272391e-05} +{"train_loss": 0.016512393951416016, "grad_norm": 0.05398103967308998, "global_step": 3714, "epoch": 35, "lr": 9.879260332556919e-05} +{"train_loss": 0.017805740237236023, "grad_norm": 0.0389527752995491, "global_step": 3715, "epoch": 35, "lr": 9.879195585697849e-05} +{"train_loss": 0.018085692077875137, "grad_norm": 0.05770595371723175, "global_step": 3716, "epoch": 35, "lr": 9.879130821695409e-05} +{"train_loss": 0.020229987800121307, "grad_norm": 0.07661664485931396, "global_step": 3717, "epoch": 35, "lr": 9.879066040549823e-05} +{"train_loss": 0.022637013345956802, "grad_norm": 0.08918789029121399, "global_step": 3718, "epoch": 35, "lr": 9.879001242261323e-05} +{"train_loss": 0.020006082952022552, "grad_norm": 0.04786005988717079, "global_step": 3719, "epoch": 35, "lr": 9.878936426830132e-05} +{"train_loss": 0.01837141625583172, "grad_norm": 0.04286794364452362, "global_step": 3720, "epoch": 35, "lr": 9.878871594256483e-05} +{"train_loss": 0.020121026784181595, "grad_norm": 0.056916605681180954, "global_step": 3721, "epoch": 35, "lr": 9.878806744540598e-05} +{"train_loss": 0.020472541451454163, "grad_norm": 0.06592024862766266, "global_step": 3722, "epoch": 35, "lr": 9.878741877682709e-05} +{"train_loss": 0.018756885081529617, "grad_norm": 0.14627815783023834, "global_step": 3723, "epoch": 35, "lr": 9.878676993683044e-05} +{"train_loss": 0.02253756858408451, "grad_norm": 0.07593713700771332, "global_step": 3724, "epoch": 35, "lr": 9.878612092541828e-05} +{"train_loss": 0.019775059074163437, "grad_norm": 0.07164569944143295, "global_step": 3725, "epoch": 35, "lr": 9.878547174259292e-05} +{"train_loss": 0.01901182159781456, "grad_norm": 0.07037583738565445, "global_step": 3726, "epoch": 35, "lr": 9.878482238835663e-05} +{"train_loss": 0.019451353698968887, "grad_norm": 0.06832965463399887, "global_step": 3727, "epoch": 35, "lr": 9.878417286271166e-05} +{"train_loss": 0.019724447280168533, "grad_norm": 0.04395594447851181, "global_step": 3728, "epoch": 35, "lr": 9.878352316566035e-05} +{"train_loss": 0.020798541605472565, "grad_norm": 0.05282500758767128, "global_step": 3729, "epoch": 35, "lr": 9.878287329720495e-05} +{"train_loss": 0.02092652954161167, "grad_norm": 0.1498725265264511, "global_step": 3730, "epoch": 35, "lr": 9.878222325734777e-05} +{"train_loss": 0.019621076062321663, "grad_norm": 0.12023090571165085, "global_step": 3731, "epoch": 35, "lr": 9.878157304609104e-05} +{"train_loss": 0.018618082627654076, "grad_norm": 0.05412072315812111, "global_step": 3732, "epoch": 35, "lr": 9.878092266343708e-05} +{"train_loss": 0.01915881223976612, "grad_norm": 0.0385873056948185, "global_step": 3733, "epoch": 35, "lr": 9.878027210938819e-05} +{"train_loss": 0.02185080572962761, "grad_norm": 0.07114475965499878, "global_step": 3734, "epoch": 35, "lr": 9.877962138394663e-05} +{"train_loss": 0.021722130477428436, "grad_norm": 0.10296078771352768, "global_step": 3735, "epoch": 35, "lr": 9.877897048711468e-05} +{"train_loss": 0.02164752595126629, "grad_norm": 0.08824755251407623, "global_step": 3736, "epoch": 35, "lr": 9.877831941889466e-05} +{"train_loss": 0.020055513828992844, "grad_norm": 0.07575641572475433, "global_step": 3737, "epoch": 35, "lr": 9.877766817928884e-05} +{"train_loss": 0.023167770355939865, "grad_norm": 0.0853862389922142, "global_step": 3738, "epoch": 35, "lr": 9.87770167682995e-05} +{"train_loss": 0.02058095484972, "grad_norm": 0.09355957061052322, "global_step": 3739, "epoch": 35, "lr": 9.877636518592892e-05} +{"train_loss": 0.022630859166383743, "grad_norm": 0.0733264610171318, "global_step": 3740, "epoch": 35, "lr": 9.877571343217941e-05} +{"train_loss": 0.018387138843536377, "grad_norm": 0.10617299377918243, "global_step": 3741, "epoch": 35, "lr": 9.877506150705328e-05} +{"train_loss": 0.020248521119356155, "grad_norm": 0.07522442191839218, "global_step": 3742, "epoch": 35, "lr": 9.877440941055278e-05} +{"train_loss": 0.020901009440422058, "grad_norm": 0.04008784517645836, "global_step": 3743, "epoch": 35, "lr": 9.877375714268023e-05} +{"train_loss": 0.019883550703525543, "grad_norm": 0.0514364019036293, "global_step": 3744, "epoch": 35, "lr": 9.877310470343788e-05} +{"train_loss": 0.018767066299915314, "grad_norm": 0.06356139481067657, "global_step": 3745, "epoch": 35, "lr": 9.877245209282805e-05} +{"train_loss": 0.017633214592933655, "grad_norm": 0.07330654561519623, "global_step": 3746, "epoch": 35, "lr": 9.877179931085308e-05} +{"train_loss": 0.021051686257123947, "grad_norm": 0.1012929305434227, "global_step": 3747, "epoch": 35, "lr": 9.877114635751517e-05} +{"train_loss": 0.0187118798494339, "grad_norm": 0.0460764542222023, "global_step": 3748, "epoch": 35, "lr": 9.877049323281668e-05} +{"train_loss": 0.018979016691446304, "grad_norm": 0.06954506784677505, "global_step": 3749, "epoch": 35, "lr": 9.876983993675989e-05} +{"train_loss": 0.021603699773550034, "grad_norm": 0.04543037340044975, "global_step": 3750, "epoch": 35, "lr": 9.876918646934709e-05} +{"train_loss": 0.022024601697921753, "grad_norm": 0.06783559173345566, "global_step": 3751, "epoch": 35, "lr": 9.876853283058057e-05} +{"train_loss": 0.021386560052633286, "grad_norm": 0.16820141673088074, "global_step": 3752, "epoch": 35, "lr": 9.876787902046264e-05} +{"train_loss": 0.018817681819200516, "grad_norm": 0.044125329703092575, "global_step": 3753, "epoch": 35, "lr": 9.876722503899559e-05} +{"train_loss": 0.020973853766918182, "grad_norm": 0.046024784445762634, "global_step": 3754, "epoch": 35, "lr": 9.876657088618173e-05} +{"train_loss": 0.01865256391465664, "grad_norm": 0.04525119811296463, "global_step": 3755, "epoch": 35, "lr": 9.876591656202335e-05} +{"train_loss": 0.02103237248957157, "grad_norm": 0.07026646286249161, "global_step": 3756, "epoch": 35, "lr": 9.876526206652272e-05} +{"train_loss": 0.0183994323015213, "grad_norm": 0.11505110561847687, "global_step": 3757, "epoch": 35, "lr": 9.876460739968219e-05} +{"train_loss": 0.018495339900255203, "grad_norm": 0.04448448494076729, "global_step": 3758, "epoch": 35, "lr": 9.876395256150403e-05} +{"train_loss": 0.020965207368135452, "grad_norm": 0.10040045529603958, "global_step": 3759, "epoch": 35, "lr": 9.876329755199054e-05} +{"train_loss": 0.02104969322681427, "grad_norm": 0.04396245628595352, "global_step": 3760, "epoch": 35, "lr": 9.876264237114403e-05} +{"train_loss": 0.021134767681360245, "grad_norm": 0.08998648822307587, "global_step": 3761, "epoch": 35, "lr": 9.876198701896681e-05} +{"train_loss": 0.021366599947214127, "grad_norm": 0.04683356732130051, "global_step": 3762, "epoch": 35, "lr": 9.876133149546118e-05} +{"train_loss": 0.016942359507083893, "grad_norm": 0.08457236737012863, "global_step": 3763, "epoch": 35, "lr": 9.876067580062942e-05} +{"train_loss": 0.019330691546201706, "grad_norm": 0.06046948954463005, "global_step": 3764, "epoch": 35, "lr": 9.876001993447385e-05} +{"train_loss": 0.020388634875416756, "grad_norm": 0.050708021968603134, "global_step": 3765, "epoch": 35, "lr": 9.87593638969968e-05} +{"train_loss": 0.02311849221587181, "grad_norm": 0.05125291272997856, "global_step": 3766, "epoch": 35, "lr": 9.875870768820053e-05} +{"train_loss": 0.020203597843647003, "grad_norm": 0.041440706700086594, "global_step": 3767, "epoch": 35, "lr": 9.875805130808737e-05} +{"train_loss": 0.020115792751312256, "grad_norm": 0.04199349135160446, "global_step": 3768, "epoch": 35, "lr": 9.875739475665962e-05} +{"train_loss": 0.020658206194639206, "grad_norm": 0.04585333168506622, "global_step": 3769, "epoch": 35, "lr": 9.875673803391958e-05} +{"train_loss": 0.01993165910243988, "grad_norm": 0.043370697647333145, "global_step": 3770, "epoch": 35, "lr": 9.875608113986958e-05} +{"train_loss": 0.018473777920007706, "grad_norm": 0.07933526486158371, "global_step": 3771, "epoch": 35, "lr": 9.875542407451192e-05} +{"train_loss": 0.02127302624285221, "grad_norm": 0.0822337418794632, "global_step": 3772, "epoch": 35, "lr": 9.875476683784889e-05} +{"train_loss": 0.02144802361726761, "grad_norm": 0.06660538166761398, "global_step": 3773, "epoch": 35, "lr": 9.875410942988281e-05} +{"train_loss": 0.021001458168029785, "grad_norm": 0.06786856800317764, "global_step": 3774, "epoch": 35, "lr": 9.875345185061601e-05} +{"train_loss": 0.01896038092672825, "grad_norm": 0.056960467249155045, "global_step": 3775, "epoch": 35, "lr": 9.875279410005076e-05} +{"train_loss": 0.020158974453806877, "grad_norm": 0.0655578151345253, "global_step": 3776, "epoch": 35, "lr": 9.875213617818942e-05} +{"train_loss": 0.020358799025416374, "grad_norm": 0.07060182839632034, "global_step": 3777, "epoch": 35, "lr": 9.875147808503424e-05} +{"train_loss": 0.02104804664850235, "grad_norm": 0.07173220068216324, "global_step": 3778, "epoch": 35, "lr": 9.87508198205876e-05} +{"train_loss": 0.019549626857042313, "grad_norm": 0.051068708300590515, "global_step": 3779, "epoch": 35, "lr": 9.875016138485176e-05} +{"train_loss": 0.02087780088186264, "grad_norm": 0.07602802664041519, "global_step": 3780, "epoch": 35, "lr": 9.874950277782907e-05} +{"train_loss": 0.02155008167028427, "grad_norm": 0.04740820452570915, "global_step": 3781, "epoch": 35, "lr": 9.874884399952181e-05} +{"train_loss": 0.018719855695962906, "grad_norm": 0.04259411245584488, "global_step": 3782, "epoch": 35, "lr": 9.874818504993232e-05} +{"train_loss": 0.023007074370980263, "grad_norm": 0.03910039737820625, "global_step": 3783, "epoch": 35, "lr": 9.874752592906291e-05} +{"train_loss": 0.020339110866189003, "grad_norm": 0.06068691983819008, "global_step": 3784, "epoch": 35, "lr": 9.87468666369159e-05} +{"train_loss": 0.020488178357481956, "grad_norm": 0.07960093766450882, "global_step": 3785, "epoch": 35, "lr": 9.874620717349359e-05} +{"train_loss": 0.02013542130589485, "grad_norm": 0.04444055259227753, "global_step": 3786, "epoch": 35, "lr": 9.874554753879831e-05} +{"train_loss": 0.022547166794538498, "grad_norm": 0.07832223176956177, "global_step": 3787, "epoch": 35, "lr": 9.874488773283237e-05} +{"train_loss": 0.01880880817770958, "grad_norm": 0.04083091393113136, "global_step": 3788, "epoch": 35, "lr": 9.87442277555981e-05} +{"train_loss": 0.016790196299552917, "grad_norm": 0.046145468950271606, "global_step": 3789, "epoch": 35, "lr": 9.874356760709782e-05} +{"train_loss": 0.021084623411297798, "grad_norm": 0.05006655305624008, "global_step": 3790, "epoch": 35, "lr": 9.874290728733382e-05} +{"train_loss": 0.024004077538847923, "grad_norm": 0.11758776009082794, "global_step": 3791, "epoch": 35, "lr": 9.874224679630846e-05} +{"train_loss": 0.02071595937013626, "grad_norm": 0.04007948189973831, "global_step": 3792, "epoch": 35, "lr": 9.874158613402403e-05} +{"train_loss": 0.01956689916551113, "grad_norm": 0.06753065437078476, "global_step": 3793, "epoch": 35, "lr": 9.874092530048286e-05} +{"train_loss": 0.02135620266199112, "grad_norm": 0.04279809445142746, "global_step": 3794, "epoch": 35, "lr": 9.87402642956873e-05} +{"train_loss": 0.021512024104595184, "grad_norm": 0.05375145003199577, "global_step": 3795, "epoch": 35, "lr": 9.873960311963963e-05} +{"train_loss": 0.02081144228577614, "grad_norm": 0.036992575973272324, "global_step": 3796, "epoch": 35, "lr": 9.87389417723422e-05} +{"train_loss": 0.020157698541879654, "grad_norm": 0.05382901057600975, "global_step": 3797, "epoch": 35, "lr": 9.87382802537973e-05} +{"train_loss": 0.019053779542446136, "grad_norm": 0.05953783914446831, "global_step": 3798, "epoch": 35, "lr": 9.87376185640073e-05} +{"train_loss": 0.022219037637114525, "grad_norm": 0.05750526115298271, "global_step": 3799, "epoch": 35, "lr": 9.87369567029745e-05} +{"train_loss": 0.023362766951322556, "grad_norm": 0.0543036051094532, "global_step": 3800, "epoch": 35, "lr": 9.873629467070124e-05} +{"train_loss": 0.017505984753370285, "grad_norm": 0.08050905913114548, "global_step": 3801, "epoch": 35, "lr": 9.873563246718982e-05} +{"train_loss": 0.019462943077087402, "grad_norm": 0.05303576588630676, "global_step": 3802, "epoch": 35, "lr": 9.87349700924426e-05} +{"train_loss": 0.020024549216032028, "grad_norm": 0.08537879586219788, "global_step": 3803, "epoch": 35, "lr": 9.873430754646187e-05} +{"train_loss": 0.020458705723285675, "grad_norm": 0.11896130442619324, "global_step": 3804, "epoch": 35, "lr": 9.873364482924998e-05} +{"train_loss": 0.02051413431763649, "grad_norm": 0.04229070246219635, "global_step": 3805, "epoch": 35, "lr": 9.873298194080925e-05} +{"train_loss": 0.020244769752025604, "grad_norm": 0.054945699870586395, "global_step": 3806, "epoch": 35, "lr": 9.873231888114202e-05} +{"train_loss": 0.018602915108203888, "grad_norm": 0.04842270165681839, "global_step": 3807, "epoch": 35, "lr": 9.873165565025062e-05} +{"train_loss": 0.019380461424589157, "grad_norm": 0.04749251529574394, "global_step": 3808, "epoch": 35, "lr": 9.873099224813737e-05} +{"train_loss": 0.016701769083738327, "grad_norm": 0.07907330989837646, "global_step": 3809, "epoch": 35, "lr": 9.873032867480462e-05} +{"train_loss": 0.017086295410990715, "grad_norm": 0.05032595619559288, "global_step": 3810, "epoch": 35, "lr": 9.872966493025468e-05} +{"train_loss": 0.016643594950437546, "grad_norm": 0.0559999980032444, "global_step": 3811, "epoch": 35, "lr": 9.872900101448988e-05} +{"train_loss": 0.018421683460474014, "grad_norm": 0.038147397339344025, "global_step": 3812, "epoch": 35, "lr": 9.872833692751258e-05} +{"train_loss": 0.016805017367005348, "grad_norm": 0.03466992452740669, "global_step": 3813, "epoch": 35, "lr": 9.872767266932507e-05} +{"train_loss": 0.020302653312683105, "grad_norm": 0.08836193382740021, "global_step": 3814, "epoch": 35, "lr": 9.872700823992973e-05} +{"train_loss": 0.02008819177676484, "grad_norm": 0.11438125371932983, "global_step": 3815, "epoch": 35, "lr": 9.872634363932887e-05, "val_loss": 0.020460816100239754} +{"train_loss": 0.021586094051599503, "grad_norm": 0.08336933702230453, "global_step": 3816, "epoch": 36, "lr": 9.872567886752482e-05} +{"train_loss": 0.020417187362909317, "grad_norm": 0.04732836037874222, "global_step": 3817, "epoch": 36, "lr": 9.872501392451993e-05} +{"train_loss": 0.0205227043479681, "grad_norm": 0.05715975537896156, "global_step": 3818, "epoch": 36, "lr": 9.872434881031654e-05} +{"train_loss": 0.021320389583706856, "grad_norm": 0.06845400482416153, "global_step": 3819, "epoch": 36, "lr": 9.872368352491696e-05} +{"train_loss": 0.01900787279009819, "grad_norm": 0.07203646004199982, "global_step": 3820, "epoch": 36, "lr": 9.872301806832355e-05} +{"train_loss": 0.01887676864862442, "grad_norm": 0.0466826893389225, "global_step": 3821, "epoch": 36, "lr": 9.872235244053865e-05} +{"train_loss": 0.019076058641076088, "grad_norm": 0.06108894944190979, "global_step": 3822, "epoch": 36, "lr": 9.872168664156459e-05} +{"train_loss": 0.02042728289961815, "grad_norm": 0.080431267619133, "global_step": 3823, "epoch": 36, "lr": 9.872102067140372e-05} +{"train_loss": 0.018293065950274467, "grad_norm": 0.06863762438297272, "global_step": 3824, "epoch": 36, "lr": 9.872035453005836e-05} +{"train_loss": 0.020104650408029556, "grad_norm": 0.07828022539615631, "global_step": 3825, "epoch": 36, "lr": 9.871968821753086e-05} +{"train_loss": 0.017880596220493317, "grad_norm": 0.04396980628371239, "global_step": 3826, "epoch": 36, "lr": 9.871902173382356e-05} +{"train_loss": 0.022458121180534363, "grad_norm": 0.06682822108268738, "global_step": 3827, "epoch": 36, "lr": 9.871835507893883e-05} +{"train_loss": 0.016840681433677673, "grad_norm": 0.06050052121281624, "global_step": 3828, "epoch": 36, "lr": 9.871768825287897e-05} +{"train_loss": 0.019865110516548157, "grad_norm": 0.09901051223278046, "global_step": 3829, "epoch": 36, "lr": 9.871702125564634e-05} +{"train_loss": 0.017146412283182144, "grad_norm": 0.04197998344898224, "global_step": 3830, "epoch": 36, "lr": 9.871635408724329e-05} +{"train_loss": 0.021539853885769844, "grad_norm": 0.10253044962882996, "global_step": 3831, "epoch": 36, "lr": 9.871568674767215e-05} +{"train_loss": 0.01890156790614128, "grad_norm": 0.03526660054922104, "global_step": 3832, "epoch": 36, "lr": 9.871501923693528e-05} +{"train_loss": 0.02123202383518219, "grad_norm": 0.0672096535563469, "global_step": 3833, "epoch": 36, "lr": 9.8714351555035e-05} +{"train_loss": 0.025223692879080772, "grad_norm": 0.12226615846157074, "global_step": 3834, "epoch": 36, "lr": 9.87136837019737e-05} +{"train_loss": 0.018951714038848877, "grad_norm": 0.04947619140148163, "global_step": 3835, "epoch": 36, "lr": 9.871301567775368e-05} +{"train_loss": 0.020189788192510605, "grad_norm": 0.04594184830784798, "global_step": 3836, "epoch": 36, "lr": 9.871234748237732e-05} +{"train_loss": 0.0193074531853199, "grad_norm": 0.06329935044050217, "global_step": 3837, "epoch": 36, "lr": 9.871167911584696e-05} +{"train_loss": 0.018196478486061096, "grad_norm": 0.054347701370716095, "global_step": 3838, "epoch": 36, "lr": 9.871101057816493e-05} +{"train_loss": 0.02085493505001068, "grad_norm": 0.04435791075229645, "global_step": 3839, "epoch": 36, "lr": 9.87103418693336e-05} +{"train_loss": 0.021193383261561394, "grad_norm": 0.06433113664388657, "global_step": 3840, "epoch": 36, "lr": 9.870967298935532e-05} +{"train_loss": 0.021060187369585037, "grad_norm": 0.1065928041934967, "global_step": 3841, "epoch": 36, "lr": 9.870900393823241e-05} +{"train_loss": 0.023267481476068497, "grad_norm": 0.049824971705675125, "global_step": 3842, "epoch": 36, "lr": 9.870833471596725e-05} +{"train_loss": 0.019952382892370224, "grad_norm": 0.04783457890152931, "global_step": 3843, "epoch": 36, "lr": 9.87076653225622e-05} +{"train_loss": 0.02032313123345375, "grad_norm": 0.04626528546214104, "global_step": 3844, "epoch": 36, "lr": 9.870699575801958e-05} +{"train_loss": 0.019557515159249306, "grad_norm": 0.09062240272760391, "global_step": 3845, "epoch": 36, "lr": 9.870632602234177e-05} +{"train_loss": 0.018460817635059357, "grad_norm": 0.061200615018606186, "global_step": 3846, "epoch": 36, "lr": 9.870565611553111e-05} +{"train_loss": 0.018236450850963593, "grad_norm": 0.046868085861206055, "global_step": 3847, "epoch": 36, "lr": 9.870498603758995e-05} +{"train_loss": 0.01975148543715477, "grad_norm": 0.138118714094162, "global_step": 3848, "epoch": 36, "lr": 9.870431578852067e-05} +{"train_loss": 0.018021389842033386, "grad_norm": 0.058062925934791565, "global_step": 3849, "epoch": 36, "lr": 9.870364536832557e-05} +{"train_loss": 0.019452117383480072, "grad_norm": 0.047851864248514175, "global_step": 3850, "epoch": 36, "lr": 9.870297477700707e-05} +{"train_loss": 0.02010728418827057, "grad_norm": 0.039412274956703186, "global_step": 3851, "epoch": 36, "lr": 9.870230401456749e-05} +{"train_loss": 0.020005308091640472, "grad_norm": 0.04912950098514557, "global_step": 3852, "epoch": 36, "lr": 9.87016330810092e-05} +{"train_loss": 0.02119087427854538, "grad_norm": 0.04804685339331627, "global_step": 3853, "epoch": 36, "lr": 9.870096197633453e-05} +{"train_loss": 0.018935522064566612, "grad_norm": 0.05688133463263512, "global_step": 3854, "epoch": 36, "lr": 9.870029070054587e-05} +{"train_loss": 0.018699154257774353, "grad_norm": 0.09541741013526917, "global_step": 3855, "epoch": 36, "lr": 9.869961925364557e-05} +{"train_loss": 0.021262679249048233, "grad_norm": 0.05417018383741379, "global_step": 3856, "epoch": 36, "lr": 9.869894763563599e-05} +{"train_loss": 0.021469123661518097, "grad_norm": 0.12590022385120392, "global_step": 3857, "epoch": 36, "lr": 9.869827584651948e-05} +{"train_loss": 0.018212182447314262, "grad_norm": 0.04807957634329796, "global_step": 3858, "epoch": 36, "lr": 9.869760388629841e-05} +{"train_loss": 0.02022898942232132, "grad_norm": 0.08891718089580536, "global_step": 3859, "epoch": 36, "lr": 9.869693175497514e-05} +{"train_loss": 0.019510071724653244, "grad_norm": 0.05235026031732559, "global_step": 3860, "epoch": 36, "lr": 9.869625945255202e-05} +{"train_loss": 0.021053552627563477, "grad_norm": 0.10974088311195374, "global_step": 3861, "epoch": 36, "lr": 9.869558697903142e-05} +{"train_loss": 0.01886444166302681, "grad_norm": 0.05881734564900398, "global_step": 3862, "epoch": 36, "lr": 9.869491433441572e-05} +{"train_loss": 0.018793966621160507, "grad_norm": 0.0555950365960598, "global_step": 3863, "epoch": 36, "lr": 9.869424151870727e-05} +{"train_loss": 0.02052289992570877, "grad_norm": 0.06132890656590462, "global_step": 3864, "epoch": 36, "lr": 9.869356853190843e-05} +{"train_loss": 0.019704662263393402, "grad_norm": 0.03438631072640419, "global_step": 3865, "epoch": 36, "lr": 9.869289537402156e-05} +{"train_loss": 0.02105872333049774, "grad_norm": 0.06260838359594345, "global_step": 3866, "epoch": 36, "lr": 9.869222204504902e-05} +{"train_loss": 0.019339557737112045, "grad_norm": 0.11945455521345139, "global_step": 3867, "epoch": 36, "lr": 9.86915485449932e-05} +{"train_loss": 0.02176194079220295, "grad_norm": 0.049086473882198334, "global_step": 3868, "epoch": 36, "lr": 9.869087487385644e-05} +{"train_loss": 0.020169632509350777, "grad_norm": 0.04331263527274132, "global_step": 3869, "epoch": 36, "lr": 9.869020103164115e-05} +{"train_loss": 0.02167699858546257, "grad_norm": 0.17693233489990234, "global_step": 3870, "epoch": 36, "lr": 9.868952701834963e-05} +{"train_loss": 0.02161252498626709, "grad_norm": 0.049913857132196426, "global_step": 3871, "epoch": 36, "lr": 9.868885283398433e-05} +{"train_loss": 0.019539592787623405, "grad_norm": 0.047323063015937805, "global_step": 3872, "epoch": 36, "lr": 9.868817847854755e-05} +{"train_loss": 0.020593535155057907, "grad_norm": 0.05401575192809105, "global_step": 3873, "epoch": 36, "lr": 9.868750395204168e-05} +{"train_loss": 0.02010723575949669, "grad_norm": 0.10848487168550491, "global_step": 3874, "epoch": 36, "lr": 9.86868292544691e-05} +{"train_loss": 0.019827954471111298, "grad_norm": 0.08106237649917603, "global_step": 3875, "epoch": 36, "lr": 9.868615438583219e-05} +{"train_loss": 0.021488619968295097, "grad_norm": 0.09798675030469894, "global_step": 3876, "epoch": 36, "lr": 9.86854793461333e-05} +{"train_loss": 0.019782189279794693, "grad_norm": 0.03955470770597458, "global_step": 3877, "epoch": 36, "lr": 9.86848041353748e-05} +{"train_loss": 0.019450925290584564, "grad_norm": 0.03988850861787796, "global_step": 3878, "epoch": 36, "lr": 9.868412875355906e-05} +{"train_loss": 0.019814975559711456, "grad_norm": 0.07750806957483292, "global_step": 3879, "epoch": 36, "lr": 9.868345320068847e-05} +{"train_loss": 0.018689561635255814, "grad_norm": 0.10605564713478088, "global_step": 3880, "epoch": 36, "lr": 9.868277747676542e-05} +{"train_loss": 0.02003161609172821, "grad_norm": 0.07361116260290146, "global_step": 3881, "epoch": 36, "lr": 9.868210158179226e-05} +{"train_loss": 0.018166586756706238, "grad_norm": 0.059268802404403687, "global_step": 3882, "epoch": 36, "lr": 9.868142551577133e-05} +{"train_loss": 0.01740190014243126, "grad_norm": 0.07628852128982544, "global_step": 3883, "epoch": 36, "lr": 9.868074927870508e-05} +{"train_loss": 0.020271427929401398, "grad_norm": 0.09615388512611389, "global_step": 3884, "epoch": 36, "lr": 9.868007287059581e-05} +{"train_loss": 0.021839864552021027, "grad_norm": 0.05333133786916733, "global_step": 3885, "epoch": 36, "lr": 9.867939629144597e-05} +{"train_loss": 0.017300216481089592, "grad_norm": 0.08819755166769028, "global_step": 3886, "epoch": 36, "lr": 9.86787195412579e-05} +{"train_loss": 0.023459941148757935, "grad_norm": 0.08396083861589432, "global_step": 3887, "epoch": 36, "lr": 9.867804262003395e-05} +{"train_loss": 0.018349135294556618, "grad_norm": 0.1093289852142334, "global_step": 3888, "epoch": 36, "lr": 9.867736552777655e-05} +{"train_loss": 0.020086530596017838, "grad_norm": 0.07806573808193207, "global_step": 3889, "epoch": 36, "lr": 9.867668826448806e-05} +{"train_loss": 0.01983891800045967, "grad_norm": 0.15140168368816376, "global_step": 3890, "epoch": 36, "lr": 9.867601083017084e-05} +{"train_loss": 0.017949337139725685, "grad_norm": 0.0899534523487091, "global_step": 3891, "epoch": 36, "lr": 9.867533322482729e-05} +{"train_loss": 0.020384889096021652, "grad_norm": 0.08951084315776825, "global_step": 3892, "epoch": 36, "lr": 9.86746554484598e-05} +{"train_loss": 0.021145189180970192, "grad_norm": 0.09247687458992004, "global_step": 3893, "epoch": 36, "lr": 9.867397750107072e-05} +{"train_loss": 0.018631402403116226, "grad_norm": 0.07275231182575226, "global_step": 3894, "epoch": 36, "lr": 9.867329938266248e-05} +{"train_loss": 0.020335711538791656, "grad_norm": 0.08016518503427505, "global_step": 3895, "epoch": 36, "lr": 9.867262109323741e-05} +{"train_loss": 0.01868651621043682, "grad_norm": 0.07282546162605286, "global_step": 3896, "epoch": 36, "lr": 9.867194263279793e-05} +{"train_loss": 0.01710657775402069, "grad_norm": 0.05293353646993637, "global_step": 3897, "epoch": 36, "lr": 9.867126400134638e-05} +{"train_loss": 0.02419138327240944, "grad_norm": 0.09852445870637894, "global_step": 3898, "epoch": 36, "lr": 9.867058519888522e-05} +{"train_loss": 0.020314030349254608, "grad_norm": 0.04888340085744858, "global_step": 3899, "epoch": 36, "lr": 9.866990622541677e-05} +{"train_loss": 0.018586965277791023, "grad_norm": 0.037752676755189896, "global_step": 3900, "epoch": 36, "lr": 9.866922708094342e-05} +{"train_loss": 0.01980298012495041, "grad_norm": 0.0642656609416008, "global_step": 3901, "epoch": 36, "lr": 9.866854776546759e-05} +{"train_loss": 0.02190944366157055, "grad_norm": 0.06117327883839607, "global_step": 3902, "epoch": 36, "lr": 9.866786827899164e-05} +{"train_loss": 0.01966368779540062, "grad_norm": 0.06944669783115387, "global_step": 3903, "epoch": 36, "lr": 9.866718862151796e-05} +{"train_loss": 0.017267931252717972, "grad_norm": 0.06898298114538193, "global_step": 3904, "epoch": 36, "lr": 9.866650879304896e-05} +{"train_loss": 0.021975629031658173, "grad_norm": 0.12840913236141205, "global_step": 3905, "epoch": 36, "lr": 9.866582879358701e-05} +{"train_loss": 0.017556333914399147, "grad_norm": 0.058632828295230865, "global_step": 3906, "epoch": 36, "lr": 9.86651486231345e-05} +{"train_loss": 0.021073445677757263, "grad_norm": 0.08501466363668442, "global_step": 3907, "epoch": 36, "lr": 9.86644682816938e-05} +{"train_loss": 0.02055458351969719, "grad_norm": 0.10667113959789276, "global_step": 3908, "epoch": 36, "lr": 9.866378776926734e-05} +{"train_loss": 0.020489903166890144, "grad_norm": 0.08348605781793594, "global_step": 3909, "epoch": 36, "lr": 9.86631070858575e-05} +{"train_loss": 0.02157573215663433, "grad_norm": 0.04853817820549011, "global_step": 3910, "epoch": 36, "lr": 9.866242623146666e-05} +{"train_loss": 0.021791331470012665, "grad_norm": 0.03947778791189194, "global_step": 3911, "epoch": 36, "lr": 9.866174520609723e-05} +{"train_loss": 0.019218450412154198, "grad_norm": 0.05401703715324402, "global_step": 3912, "epoch": 36, "lr": 9.866106400975157e-05} +{"train_loss": 0.019765250384807587, "grad_norm": 0.08513882756233215, "global_step": 3913, "epoch": 36, "lr": 9.86603826424321e-05} +{"train_loss": 0.02243867516517639, "grad_norm": 0.059919483959674835, "global_step": 3914, "epoch": 36, "lr": 9.86597011041412e-05} +{"train_loss": 0.01939019188284874, "grad_norm": 0.12211404740810394, "global_step": 3915, "epoch": 36, "lr": 9.865901939488128e-05} +{"train_loss": 0.024557359516620636, "grad_norm": 0.06379154324531555, "global_step": 3916, "epoch": 36, "lr": 9.865833751465472e-05} +{"train_loss": 0.021962489932775497, "grad_norm": 0.18576623499393463, "global_step": 3917, "epoch": 36, "lr": 9.865765546346393e-05} +{"train_loss": 0.019501786679029465, "grad_norm": 0.0853085070848465, "global_step": 3918, "epoch": 36, "lr": 9.86569732413113e-05} +{"train_loss": 0.020569458603858948, "grad_norm": 0.07900606095790863, "global_step": 3919, "epoch": 36, "lr": 9.865629084819923e-05} +{"train_loss": 0.017895586788654327, "grad_norm": 0.07040068507194519, "global_step": 3920, "epoch": 36, "lr": 9.86556082841301e-05} +{"train_loss": 0.020044126020709298, "grad_norm": 0.060356080532073975, "global_step": 3921, "epoch": 36, "lr": 9.865492554910633e-05, "val_loss": 0.01943170092999935} +{"train_loss": 0.020494477823376656, "grad_norm": 0.04663139581680298, "global_step": 3922, "epoch": 37, "lr": 9.865424264313031e-05} +{"train_loss": 0.01684933714568615, "grad_norm": 0.04919252544641495, "global_step": 3923, "epoch": 37, "lr": 9.865355956620444e-05} +{"train_loss": 0.019590938463807106, "grad_norm": 0.04785085469484329, "global_step": 3924, "epoch": 37, "lr": 9.865287631833114e-05} +{"train_loss": 0.020815227180719376, "grad_norm": 0.1067952960729599, "global_step": 3925, "epoch": 37, "lr": 9.865219289951276e-05} +{"train_loss": 0.01820526458323002, "grad_norm": 0.0682956799864769, "global_step": 3926, "epoch": 37, "lr": 9.865150930975175e-05} +{"train_loss": 0.020714346319437027, "grad_norm": 0.1100984439253807, "global_step": 3927, "epoch": 37, "lr": 9.865082554905048e-05} +{"train_loss": 0.0204567089676857, "grad_norm": 0.0403495654463768, "global_step": 3928, "epoch": 37, "lr": 9.865014161741139e-05} +{"train_loss": 0.01796375960111618, "grad_norm": 0.08459191024303436, "global_step": 3929, "epoch": 37, "lr": 9.864945751483684e-05} +{"train_loss": 0.018246907740831375, "grad_norm": 0.04754983261227608, "global_step": 3930, "epoch": 37, "lr": 9.864877324132926e-05} +{"train_loss": 0.020284902304410934, "grad_norm": 0.1359485238790512, "global_step": 3931, "epoch": 37, "lr": 9.864808879689105e-05} +{"train_loss": 0.020021647214889526, "grad_norm": 0.04268619790673256, "global_step": 3932, "epoch": 37, "lr": 9.86474041815246e-05} +{"train_loss": 0.020566176623106003, "grad_norm": 0.04768955707550049, "global_step": 3933, "epoch": 37, "lr": 9.864671939523234e-05} +{"train_loss": 0.019910229369997978, "grad_norm": 0.06113428249955177, "global_step": 3934, "epoch": 37, "lr": 9.864603443801666e-05} +{"train_loss": 0.021109536290168762, "grad_norm": 0.08726479858160019, "global_step": 3935, "epoch": 37, "lr": 9.864534930987999e-05} +{"train_loss": 0.017791401594877243, "grad_norm": 0.08273836970329285, "global_step": 3936, "epoch": 37, "lr": 9.864466401082468e-05} +{"train_loss": 0.019957490265369415, "grad_norm": 0.09557228535413742, "global_step": 3937, "epoch": 37, "lr": 9.864397854085322e-05} +{"train_loss": 0.020294813439249992, "grad_norm": 0.11270294338464737, "global_step": 3938, "epoch": 37, "lr": 9.864329289996795e-05} +{"train_loss": 0.02175191044807434, "grad_norm": 0.09175746142864227, "global_step": 3939, "epoch": 37, "lr": 9.864260708817129e-05} +{"train_loss": 0.01976502314209938, "grad_norm": 0.08154374361038208, "global_step": 3940, "epoch": 37, "lr": 9.864192110546567e-05} +{"train_loss": 0.020087461918592453, "grad_norm": 0.0716090127825737, "global_step": 3941, "epoch": 37, "lr": 9.864123495185351e-05} +{"train_loss": 0.02123512700200081, "grad_norm": 0.131888285279274, "global_step": 3942, "epoch": 37, "lr": 9.864054862733718e-05} +{"train_loss": 0.022782500833272934, "grad_norm": 0.05422618240118027, "global_step": 3943, "epoch": 37, "lr": 9.863986213191913e-05} +{"train_loss": 0.019166570156812668, "grad_norm": 0.05898984521627426, "global_step": 3944, "epoch": 37, "lr": 9.863917546560177e-05} +{"train_loss": 0.02316390722990036, "grad_norm": 0.13220080733299255, "global_step": 3945, "epoch": 37, "lr": 9.863848862838747e-05} +{"train_loss": 0.019587410613894463, "grad_norm": 0.04677741602063179, "global_step": 3946, "epoch": 37, "lr": 9.863780162027869e-05} +{"train_loss": 0.021229146048426628, "grad_norm": 0.05873091518878937, "global_step": 3947, "epoch": 37, "lr": 9.863711444127784e-05} +{"train_loss": 0.018018659204244614, "grad_norm": 0.06936030089855194, "global_step": 3948, "epoch": 37, "lr": 9.863642709138729e-05} +{"train_loss": 0.017913661897182465, "grad_norm": 0.07266531884670258, "global_step": 3949, "epoch": 37, "lr": 9.863573957060953e-05} +{"train_loss": 0.01859474554657936, "grad_norm": 0.0806092843413353, "global_step": 3950, "epoch": 37, "lr": 9.863505187894689e-05} +{"train_loss": 0.021844830363988876, "grad_norm": 0.054303739219903946, "global_step": 3951, "epoch": 37, "lr": 9.863436401640185e-05} +{"train_loss": 0.02035379782319069, "grad_norm": 0.059121597558259964, "global_step": 3952, "epoch": 37, "lr": 9.863367598297679e-05} +{"train_loss": 0.02040718123316765, "grad_norm": 0.10566627979278564, "global_step": 3953, "epoch": 37, "lr": 9.863298777867416e-05} +{"train_loss": 0.018632542341947556, "grad_norm": 0.043335556983947754, "global_step": 3954, "epoch": 37, "lr": 9.863229940349635e-05} +{"train_loss": 0.018571168184280396, "grad_norm": 0.03577504679560661, "global_step": 3955, "epoch": 37, "lr": 9.86316108574458e-05} +{"train_loss": 0.02127840183675289, "grad_norm": 0.05084497481584549, "global_step": 3956, "epoch": 37, "lr": 9.863092214052492e-05} +{"train_loss": 0.016641080379486084, "grad_norm": 0.05495775118470192, "global_step": 3957, "epoch": 37, "lr": 9.86302332527361e-05} +{"train_loss": 0.020314320921897888, "grad_norm": 0.10198254883289337, "global_step": 3958, "epoch": 37, "lr": 9.862954419408181e-05} +{"train_loss": 0.01878063566982746, "grad_norm": 0.13192519545555115, "global_step": 3959, "epoch": 37, "lr": 9.862885496456447e-05} +{"train_loss": 0.020196225494146347, "grad_norm": 0.1877368986606598, "global_step": 3960, "epoch": 37, "lr": 9.862816556418646e-05} +{"train_loss": 0.02027752995491028, "grad_norm": 0.09464628994464874, "global_step": 3961, "epoch": 37, "lr": 9.862747599295022e-05} +{"train_loss": 0.02106720581650734, "grad_norm": 0.06606598943471909, "global_step": 3962, "epoch": 37, "lr": 9.862678625085818e-05} +{"train_loss": 0.021171223372220993, "grad_norm": 0.0365423820912838, "global_step": 3963, "epoch": 37, "lr": 9.862609633791276e-05} +{"train_loss": 0.02051946148276329, "grad_norm": 0.08931899070739746, "global_step": 3964, "epoch": 37, "lr": 9.862540625411639e-05} +{"train_loss": 0.020099759101867676, "grad_norm": 0.06149454787373543, "global_step": 3965, "epoch": 37, "lr": 9.86247159994715e-05} +{"train_loss": 0.017213746905326843, "grad_norm": 0.03671223297715187, "global_step": 3966, "epoch": 37, "lr": 9.862402557398048e-05} +{"train_loss": 0.02255219593644142, "grad_norm": 0.05448739603161812, "global_step": 3967, "epoch": 37, "lr": 9.86233349776458e-05} +{"train_loss": 0.01820915937423706, "grad_norm": 0.07776801288127899, "global_step": 3968, "epoch": 37, "lr": 9.862264421046985e-05} +{"train_loss": 0.022293314337730408, "grad_norm": 0.06396829336881638, "global_step": 3969, "epoch": 37, "lr": 9.862195327245508e-05} +{"train_loss": 0.021132178604602814, "grad_norm": 0.08895143866539001, "global_step": 3970, "epoch": 37, "lr": 9.862126216360393e-05} +{"train_loss": 0.022063663229346275, "grad_norm": 0.1411997228860855, "global_step": 3971, "epoch": 37, "lr": 9.862057088391878e-05} +{"train_loss": 0.021966585889458656, "grad_norm": 0.20550644397735596, "global_step": 3972, "epoch": 37, "lr": 9.861987943340209e-05} +{"train_loss": 0.02229481004178524, "grad_norm": 0.1349506676197052, "global_step": 3973, "epoch": 37, "lr": 9.86191878120563e-05} +{"train_loss": 0.020175499841570854, "grad_norm": 0.04560284689068794, "global_step": 3974, "epoch": 37, "lr": 9.861849601988383e-05} +{"train_loss": 0.018467135727405548, "grad_norm": 0.05574779585003853, "global_step": 3975, "epoch": 37, "lr": 9.86178040568871e-05} +{"train_loss": 0.021672498434782028, "grad_norm": 0.05329800397157669, "global_step": 3976, "epoch": 37, "lr": 9.861711192306857e-05} +{"train_loss": 0.01973172463476658, "grad_norm": 0.05452597141265869, "global_step": 3977, "epoch": 37, "lr": 9.861641961843064e-05} +{"train_loss": 0.02047606185078621, "grad_norm": 0.05770625174045563, "global_step": 3978, "epoch": 37, "lr": 9.861572714297575e-05} +{"train_loss": 0.019710417836904526, "grad_norm": 0.06084269657731056, "global_step": 3979, "epoch": 37, "lr": 9.861503449670633e-05} +{"train_loss": 0.021779974922537804, "grad_norm": 0.15484723448753357, "global_step": 3980, "epoch": 37, "lr": 9.861434167962483e-05} +{"train_loss": 0.021723339334130287, "grad_norm": 0.08045098930597305, "global_step": 3981, "epoch": 37, "lr": 9.861364869173367e-05} +{"train_loss": 0.021168943494558334, "grad_norm": 0.08015645295381546, "global_step": 3982, "epoch": 37, "lr": 9.86129555330353e-05} +{"train_loss": 0.02066543512046337, "grad_norm": 0.21814851462841034, "global_step": 3983, "epoch": 37, "lr": 9.861226220353214e-05} +{"train_loss": 0.020903021097183228, "grad_norm": 0.10471101850271225, "global_step": 3984, "epoch": 37, "lr": 9.861156870322664e-05} +{"train_loss": 0.019187383353710175, "grad_norm": 0.06644878536462784, "global_step": 3985, "epoch": 37, "lr": 9.86108750321212e-05} +{"train_loss": 0.018852083012461662, "grad_norm": 0.08825120329856873, "global_step": 3986, "epoch": 37, "lr": 9.861018119021833e-05} +{"train_loss": 0.018761251121759415, "grad_norm": 0.1100839301943779, "global_step": 3987, "epoch": 37, "lr": 9.860948717752039e-05} +{"train_loss": 0.02008073590695858, "grad_norm": 0.13336017727851868, "global_step": 3988, "epoch": 37, "lr": 9.860879299402986e-05} +{"train_loss": 0.01839277520775795, "grad_norm": 0.15937289595603943, "global_step": 3989, "epoch": 37, "lr": 9.860809863974916e-05} +{"train_loss": 0.021601159125566483, "grad_norm": 0.08091054856777191, "global_step": 3990, "epoch": 37, "lr": 9.860740411468075e-05} +{"train_loss": 0.02059057727456093, "grad_norm": 0.03767906874418259, "global_step": 3991, "epoch": 37, "lr": 9.860670941882707e-05} +{"train_loss": 0.020557068288326263, "grad_norm": 0.04755222797393799, "global_step": 3992, "epoch": 37, "lr": 9.860601455219055e-05} +{"train_loss": 0.016749288886785507, "grad_norm": 0.04598446190357208, "global_step": 3993, "epoch": 37, "lr": 9.860531951477362e-05} +{"train_loss": 0.021484196186065674, "grad_norm": 0.055654946714639664, "global_step": 3994, "epoch": 37, "lr": 9.860462430657874e-05} +{"train_loss": 0.020145611837506294, "grad_norm": 0.08498136699199677, "global_step": 3995, "epoch": 37, "lr": 9.860392892760834e-05} +{"train_loss": 0.019742868840694427, "grad_norm": 0.039706215262413025, "global_step": 3996, "epoch": 37, "lr": 9.86032333778649e-05} +{"train_loss": 0.01833355985581875, "grad_norm": 0.09287963807582855, "global_step": 3997, "epoch": 37, "lr": 9.860253765735081e-05} +{"train_loss": 0.01969894766807556, "grad_norm": 0.1919563114643097, "global_step": 3998, "epoch": 37, "lr": 9.860184176606855e-05} +{"train_loss": 0.018796231597661972, "grad_norm": 0.20656803250312805, "global_step": 3999, "epoch": 37, "lr": 9.860114570402054e-05} +{"train_loss": 0.01791749894618988, "grad_norm": 0.08081765472888947, "global_step": 4000, "epoch": 37, "lr": 9.860044947120926e-05} +{"train_loss": 0.021220717579126358, "grad_norm": 0.05819444730877876, "global_step": 4001, "epoch": 37, "lr": 9.859975306763711e-05} +{"train_loss": 0.02162948064506054, "grad_norm": 0.044484857469797134, "global_step": 4002, "epoch": 37, "lr": 9.859905649330658e-05} +{"train_loss": 0.019014425575733185, "grad_norm": 0.04542595148086548, "global_step": 4003, "epoch": 37, "lr": 9.859835974822011e-05} +{"train_loss": 0.022381724789738655, "grad_norm": 0.0771757960319519, "global_step": 4004, "epoch": 37, "lr": 9.859766283238012e-05} +{"train_loss": 0.02214030921459198, "grad_norm": 0.05944971367716789, "global_step": 4005, "epoch": 37, "lr": 9.85969657457891e-05} +{"train_loss": 0.02033115178346634, "grad_norm": 0.04488300532102585, "global_step": 4006, "epoch": 37, "lr": 9.859626848844945e-05} +{"train_loss": 0.020698655396699905, "grad_norm": 0.049512118101119995, "global_step": 4007, "epoch": 37, "lr": 9.859557106036366e-05} +{"train_loss": 0.018924031406641006, "grad_norm": 0.07625195384025574, "global_step": 4008, "epoch": 37, "lr": 9.859487346153418e-05} +{"train_loss": 0.021410886198282242, "grad_norm": 0.07574960589408875, "global_step": 4009, "epoch": 37, "lr": 9.859417569196343e-05} +{"train_loss": 0.019025545567274094, "grad_norm": 0.0590907484292984, "global_step": 4010, "epoch": 37, "lr": 9.859347775165388e-05} +{"train_loss": 0.019545191898941994, "grad_norm": 0.0812426432967186, "global_step": 4011, "epoch": 37, "lr": 9.859277964060798e-05} +{"train_loss": 0.01797669567167759, "grad_norm": 0.0579390823841095, "global_step": 4012, "epoch": 37, "lr": 9.85920813588282e-05} +{"train_loss": 0.019682254642248154, "grad_norm": 0.04955556243658066, "global_step": 4013, "epoch": 37, "lr": 9.859138290631696e-05} +{"train_loss": 0.021143220365047455, "grad_norm": 0.08585680276155472, "global_step": 4014, "epoch": 37, "lr": 9.859068428307674e-05} +{"train_loss": 0.01819927617907524, "grad_norm": 0.06588935852050781, "global_step": 4015, "epoch": 37, "lr": 9.858998548911e-05} +{"train_loss": 0.01867704465985298, "grad_norm": 0.06133623048663139, "global_step": 4016, "epoch": 37, "lr": 9.858928652441917e-05} +{"train_loss": 0.0195626150816679, "grad_norm": 0.0912913978099823, "global_step": 4017, "epoch": 37, "lr": 9.858858738900671e-05} +{"train_loss": 0.02058461867272854, "grad_norm": 0.049719203263521194, "global_step": 4018, "epoch": 37, "lr": 9.858788808287508e-05} +{"train_loss": 0.01987181417644024, "grad_norm": 0.08352504670619965, "global_step": 4019, "epoch": 37, "lr": 9.858718860602676e-05} +{"train_loss": 0.018483147025108337, "grad_norm": 0.1158878430724144, "global_step": 4020, "epoch": 37, "lr": 9.858648895846418e-05} +{"train_loss": 0.019139934331178665, "grad_norm": 0.05210946872830391, "global_step": 4021, "epoch": 37, "lr": 9.858578914018982e-05} +{"train_loss": 0.02363728918135166, "grad_norm": 0.05954568833112717, "global_step": 4022, "epoch": 37, "lr": 9.858508915120612e-05} +{"train_loss": 0.018309656530618668, "grad_norm": 0.07470732927322388, "global_step": 4023, "epoch": 37, "lr": 9.858438899151554e-05} +{"train_loss": 0.019759008660912514, "grad_norm": 0.04378499835729599, "global_step": 4024, "epoch": 37, "lr": 9.858368866112055e-05} +{"train_loss": 0.01831144094467163, "grad_norm": 0.13939212262630463, "global_step": 4025, "epoch": 37, "lr": 9.85829881600236e-05} +{"train_loss": 0.018888309597969055, "grad_norm": 0.06205325573682785, "global_step": 4026, "epoch": 37, "lr": 9.858228748822718e-05} +{"train_loss": 0.020005688036106667, "grad_norm": 0.05482451245188713, "global_step": 4027, "epoch": 37, "lr": 9.85815866457337e-05, "val_loss": 0.01903378963470459} +{"train_loss": 0.020885903388261795, "grad_norm": 0.16857674717903137, "global_step": 4028, "epoch": 38, "lr": 9.858088563254566e-05} +{"train_loss": 0.021681584417819977, "grad_norm": 0.1354241669178009, "global_step": 4029, "epoch": 38, "lr": 9.858018444866552e-05} +{"train_loss": 0.020047727972269058, "grad_norm": 0.08431782573461533, "global_step": 4030, "epoch": 38, "lr": 9.857948309409575e-05} +{"train_loss": 0.018784957006573677, "grad_norm": 0.09411269426345825, "global_step": 4031, "epoch": 38, "lr": 9.85787815688388e-05} +{"train_loss": 0.019012585282325745, "grad_norm": 0.09070680290460587, "global_step": 4032, "epoch": 38, "lr": 9.85780798728971e-05} +{"train_loss": 0.02010420709848404, "grad_norm": 0.05547862872481346, "global_step": 4033, "epoch": 38, "lr": 9.85773780062732e-05} +{"train_loss": 0.019731992855668068, "grad_norm": 0.045675911009311676, "global_step": 4034, "epoch": 38, "lr": 9.85766759689695e-05} +{"train_loss": 0.02037280797958374, "grad_norm": 0.038511741906404495, "global_step": 4035, "epoch": 38, "lr": 9.85759737609885e-05} +{"train_loss": 0.019852856174111366, "grad_norm": 0.1463354229927063, "global_step": 4036, "epoch": 38, "lr": 9.857527138233263e-05} +{"train_loss": 0.020192785188555717, "grad_norm": 0.07807746529579163, "global_step": 4037, "epoch": 38, "lr": 9.85745688330044e-05} +{"train_loss": 0.019308215007185936, "grad_norm": 0.12223874032497406, "global_step": 4038, "epoch": 38, "lr": 9.857386611300626e-05} +{"train_loss": 0.020179379731416702, "grad_norm": 0.11332062631845474, "global_step": 4039, "epoch": 38, "lr": 9.857316322234067e-05} +{"train_loss": 0.018991883844137192, "grad_norm": 0.14178553223609924, "global_step": 4040, "epoch": 38, "lr": 9.85724601610101e-05} +{"train_loss": 0.02136564999818802, "grad_norm": 0.18844300508499146, "global_step": 4041, "epoch": 38, "lr": 9.857175692901704e-05} +{"train_loss": 0.022492079064249992, "grad_norm": 0.1258881837129593, "global_step": 4042, "epoch": 38, "lr": 9.857105352636395e-05} +{"train_loss": 0.017667099833488464, "grad_norm": 0.13933584094047546, "global_step": 4043, "epoch": 38, "lr": 9.85703499530533e-05} +{"train_loss": 0.019362377002835274, "grad_norm": 0.0640667974948883, "global_step": 4044, "epoch": 38, "lr": 9.856964620908757e-05} +{"train_loss": 0.02144072949886322, "grad_norm": 0.060407526791095734, "global_step": 4045, "epoch": 38, "lr": 9.85689422944692e-05} +{"train_loss": 0.021896563470363617, "grad_norm": 0.041120246052742004, "global_step": 4046, "epoch": 38, "lr": 9.856823820920072e-05} +{"train_loss": 0.0195388812571764, "grad_norm": 0.0853581577539444, "global_step": 4047, "epoch": 38, "lr": 9.856753395328455e-05} +{"train_loss": 0.019297435879707336, "grad_norm": 0.14813044667243958, "global_step": 4048, "epoch": 38, "lr": 9.856682952672321e-05} +{"train_loss": 0.01729552447795868, "grad_norm": 0.058612823486328125, "global_step": 4049, "epoch": 38, "lr": 9.856612492951913e-05} +{"train_loss": 0.018085576593875885, "grad_norm": 0.06549863517284393, "global_step": 4050, "epoch": 38, "lr": 9.856542016167481e-05} +{"train_loss": 0.017013803124427795, "grad_norm": 0.05517089366912842, "global_step": 4051, "epoch": 38, "lr": 9.856471522319274e-05} +{"train_loss": 0.0207875557243824, "grad_norm": 0.11952733248472214, "global_step": 4052, "epoch": 38, "lr": 9.856401011407536e-05} +{"train_loss": 0.018678847700357437, "grad_norm": 0.06882599741220474, "global_step": 4053, "epoch": 38, "lr": 9.856330483432519e-05} +{"train_loss": 0.018050096929073334, "grad_norm": 0.0649518370628357, "global_step": 4054, "epoch": 38, "lr": 9.856259938394468e-05} +{"train_loss": 0.02074052393436432, "grad_norm": 0.14268723130226135, "global_step": 4055, "epoch": 38, "lr": 9.856189376293631e-05} +{"train_loss": 0.02069028466939926, "grad_norm": 0.1021576076745987, "global_step": 4056, "epoch": 38, "lr": 9.856118797130257e-05} +{"train_loss": 0.021994391456246376, "grad_norm": 0.06806612759828568, "global_step": 4057, "epoch": 38, "lr": 9.856048200904592e-05} +{"train_loss": 0.019528277218341827, "grad_norm": 0.05779339373111725, "global_step": 4058, "epoch": 38, "lr": 9.855977587616887e-05} +{"train_loss": 0.018793273717164993, "grad_norm": 0.1475822478532791, "global_step": 4059, "epoch": 38, "lr": 9.855906957267388e-05} +{"train_loss": 0.023433895781636238, "grad_norm": 0.15396563708782196, "global_step": 4060, "epoch": 38, "lr": 9.855836309856344e-05} +{"train_loss": 0.018504232168197632, "grad_norm": 0.12507271766662598, "global_step": 4061, "epoch": 38, "lr": 9.855765645384002e-05} +{"train_loss": 0.01952563226222992, "grad_norm": 0.04619736969470978, "global_step": 4062, "epoch": 38, "lr": 9.855694963850612e-05} +{"train_loss": 0.020215937867760658, "grad_norm": 0.03916659578680992, "global_step": 4063, "epoch": 38, "lr": 9.855624265256423e-05} +{"train_loss": 0.020294617861509323, "grad_norm": 0.07472580671310425, "global_step": 4064, "epoch": 38, "lr": 9.85555354960168e-05} +{"train_loss": 0.018400084227323532, "grad_norm": 0.09224437177181244, "global_step": 4065, "epoch": 38, "lr": 9.855482816886634e-05} +{"train_loss": 0.01865854114294052, "grad_norm": 0.14032815396785736, "global_step": 4066, "epoch": 38, "lr": 9.855412067111532e-05} +{"train_loss": 0.01937917061150074, "grad_norm": 0.13894952833652496, "global_step": 4067, "epoch": 38, "lr": 9.855341300276624e-05} +{"train_loss": 0.019713563844561577, "grad_norm": 0.139932319521904, "global_step": 4068, "epoch": 38, "lr": 9.85527051638216e-05} +{"train_loss": 0.020473279058933258, "grad_norm": 0.11279213428497314, "global_step": 4069, "epoch": 38, "lr": 9.855199715428385e-05} +{"train_loss": 0.021146243438124657, "grad_norm": 0.11902019381523132, "global_step": 4070, "epoch": 38, "lr": 9.85512889741555e-05} +{"train_loss": 0.021717417985200882, "grad_norm": 0.06823604553937912, "global_step": 4071, "epoch": 38, "lr": 9.855058062343904e-05} +{"train_loss": 0.020173590630292892, "grad_norm": 0.03750492259860039, "global_step": 4072, "epoch": 38, "lr": 9.854987210213695e-05} +{"train_loss": 0.020094620063900948, "grad_norm": 0.12802082300186157, "global_step": 4073, "epoch": 38, "lr": 9.854916341025173e-05} +{"train_loss": 0.019172580912709236, "grad_norm": 0.08829960227012634, "global_step": 4074, "epoch": 38, "lr": 9.854845454778585e-05} +{"train_loss": 0.018255572766065598, "grad_norm": 0.050172168761491776, "global_step": 4075, "epoch": 38, "lr": 9.854774551474182e-05} +{"train_loss": 0.02346208319067955, "grad_norm": 0.10072843730449677, "global_step": 4076, "epoch": 38, "lr": 9.854703631112214e-05} +{"train_loss": 0.016222083941102028, "grad_norm": 0.05420231446623802, "global_step": 4077, "epoch": 38, "lr": 9.854632693692927e-05} +{"train_loss": 0.020158058032393456, "grad_norm": 0.08029314130544662, "global_step": 4078, "epoch": 38, "lr": 9.854561739216572e-05} +{"train_loss": 0.018437640741467476, "grad_norm": 0.13922178745269775, "global_step": 4079, "epoch": 38, "lr": 9.854490767683397e-05} +{"train_loss": 0.019317397847771645, "grad_norm": 0.08084402978420258, "global_step": 4080, "epoch": 38, "lr": 9.854419779093655e-05} +{"train_loss": 0.022747695446014404, "grad_norm": 0.05968998745083809, "global_step": 4081, "epoch": 38, "lr": 9.854348773447593e-05} +{"train_loss": 0.021587979048490524, "grad_norm": 0.05653757229447365, "global_step": 4082, "epoch": 38, "lr": 9.85427775074546e-05} +{"train_loss": 0.020397713407874107, "grad_norm": 0.11371997743844986, "global_step": 4083, "epoch": 38, "lr": 9.854206710987506e-05} +{"train_loss": 0.02213316410779953, "grad_norm": 0.04510756954550743, "global_step": 4084, "epoch": 38, "lr": 9.85413565417398e-05} +{"train_loss": 0.019756587222218513, "grad_norm": 0.042558878660202026, "global_step": 4085, "epoch": 38, "lr": 9.854064580305132e-05} +{"train_loss": 0.017322931438684464, "grad_norm": 0.045603059232234955, "global_step": 4086, "epoch": 38, "lr": 9.853993489381212e-05} +{"train_loss": 0.01774654909968376, "grad_norm": 0.03727934509515762, "global_step": 4087, "epoch": 38, "lr": 9.853922381402471e-05} +{"train_loss": 0.020643431693315506, "grad_norm": 0.04378663748502731, "global_step": 4088, "epoch": 38, "lr": 9.853851256369157e-05} +{"train_loss": 0.016920581459999084, "grad_norm": 0.04100966826081276, "global_step": 4089, "epoch": 38, "lr": 9.853780114281521e-05} +{"train_loss": 0.0199117474257946, "grad_norm": 0.12578563392162323, "global_step": 4090, "epoch": 38, "lr": 9.853708955139813e-05} +{"train_loss": 0.0204666368663311, "grad_norm": 0.03079993650317192, "global_step": 4091, "epoch": 38, "lr": 9.853637778944283e-05} +{"train_loss": 0.020769011229276657, "grad_norm": 0.09510179609060287, "global_step": 4092, "epoch": 38, "lr": 9.853566585695179e-05} +{"train_loss": 0.022565284743905067, "grad_norm": 0.05010146647691727, "global_step": 4093, "epoch": 38, "lr": 9.853495375392752e-05} +{"train_loss": 0.020857468247413635, "grad_norm": 0.038772158324718475, "global_step": 4094, "epoch": 38, "lr": 9.853424148037255e-05} +{"train_loss": 0.01933549903333187, "grad_norm": 0.04527546837925911, "global_step": 4095, "epoch": 38, "lr": 9.853352903628936e-05} +{"train_loss": 0.020359277725219727, "grad_norm": 0.07058955729007721, "global_step": 4096, "epoch": 38, "lr": 9.853281642168045e-05} +{"train_loss": 0.01968418061733246, "grad_norm": 0.06502404063940048, "global_step": 4097, "epoch": 38, "lr": 9.853210363654833e-05} +{"train_loss": 0.017412498593330383, "grad_norm": 0.07869020104408264, "global_step": 4098, "epoch": 38, "lr": 9.85313906808955e-05} +{"train_loss": 0.016469605267047882, "grad_norm": 0.045432694256305695, "global_step": 4099, "epoch": 38, "lr": 9.853067755472446e-05} +{"train_loss": 0.019062571227550507, "grad_norm": 0.05478885769844055, "global_step": 4100, "epoch": 38, "lr": 9.852996425803776e-05} +{"train_loss": 0.018269795924425125, "grad_norm": 0.07807821780443192, "global_step": 4101, "epoch": 38, "lr": 9.852925079083785e-05} +{"train_loss": 0.019947685301303864, "grad_norm": 0.0405525341629982, "global_step": 4102, "epoch": 38, "lr": 9.852853715312725e-05} +{"train_loss": 0.021149858832359314, "grad_norm": 0.06365640461444855, "global_step": 4103, "epoch": 38, "lr": 9.852782334490849e-05} +{"train_loss": 0.021674590185284615, "grad_norm": 0.04959690943360329, "global_step": 4104, "epoch": 38, "lr": 9.852710936618405e-05} +{"train_loss": 0.02138664573431015, "grad_norm": 0.05236997827887535, "global_step": 4105, "epoch": 38, "lr": 9.852639521695646e-05} +{"train_loss": 0.02064795233309269, "grad_norm": 0.09451808035373688, "global_step": 4106, "epoch": 38, "lr": 9.85256808972282e-05} +{"train_loss": 0.022883065044879913, "grad_norm": 0.1184270977973938, "global_step": 4107, "epoch": 38, "lr": 9.852496640700183e-05} +{"train_loss": 0.017446989193558693, "grad_norm": 0.03218759968876839, "global_step": 4108, "epoch": 38, "lr": 9.852425174627982e-05} +{"train_loss": 0.023075388744473457, "grad_norm": 0.12283935397863388, "global_step": 4109, "epoch": 38, "lr": 9.852353691506469e-05} +{"train_loss": 0.021473918110132217, "grad_norm": 0.08449801802635193, "global_step": 4110, "epoch": 38, "lr": 9.852282191335895e-05} +{"train_loss": 0.020537981763482094, "grad_norm": 0.09640596061944962, "global_step": 4111, "epoch": 38, "lr": 9.85221067411651e-05} +{"train_loss": 0.021886605769395828, "grad_norm": 0.07393207401037216, "global_step": 4112, "epoch": 38, "lr": 9.852139139848571e-05} +{"train_loss": 0.020474784076213837, "grad_norm": 0.04831325262784958, "global_step": 4113, "epoch": 38, "lr": 9.852067588532323e-05} +{"train_loss": 0.018379662185907364, "grad_norm": 0.10660304129123688, "global_step": 4114, "epoch": 38, "lr": 9.85199602016802e-05} +{"train_loss": 0.017613163217902184, "grad_norm": 0.12926562130451202, "global_step": 4115, "epoch": 38, "lr": 9.851924434755912e-05} +{"train_loss": 0.022215373814105988, "grad_norm": 0.06482598185539246, "global_step": 4116, "epoch": 38, "lr": 9.851852832296252e-05} +{"train_loss": 0.02284107729792595, "grad_norm": 0.06389828026294708, "global_step": 4117, "epoch": 38, "lr": 9.851781212789292e-05} +{"train_loss": 0.020211592316627502, "grad_norm": 0.07473436743021011, "global_step": 4118, "epoch": 38, "lr": 9.851709576235282e-05} +{"train_loss": 0.02022087201476097, "grad_norm": 0.047247420996427536, "global_step": 4119, "epoch": 38, "lr": 9.851637922634475e-05} +{"train_loss": 0.023460811004042625, "grad_norm": 0.07322023808956146, "global_step": 4120, "epoch": 38, "lr": 9.851566251987121e-05} +{"train_loss": 0.018860219046473503, "grad_norm": 0.1546178013086319, "global_step": 4121, "epoch": 38, "lr": 9.851494564293475e-05} +{"train_loss": 0.017260733991861343, "grad_norm": 0.0422278456389904, "global_step": 4122, "epoch": 38, "lr": 9.851422859553784e-05} +{"train_loss": 0.018478846177458763, "grad_norm": 0.05378321185708046, "global_step": 4123, "epoch": 38, "lr": 9.851351137768306e-05} +{"train_loss": 0.0167524591088295, "grad_norm": 0.03536424785852432, "global_step": 4124, "epoch": 38, "lr": 9.85127939893729e-05} +{"train_loss": 0.022116079926490784, "grad_norm": 0.08191639930009842, "global_step": 4125, "epoch": 38, "lr": 9.851207643060987e-05} +{"train_loss": 0.01758795604109764, "grad_norm": 0.07354916632175446, "global_step": 4126, "epoch": 38, "lr": 9.85113587013965e-05} +{"train_loss": 0.019127242267131805, "grad_norm": 0.0605708584189415, "global_step": 4127, "epoch": 38, "lr": 9.85106408017353e-05} +{"train_loss": 0.02056572213768959, "grad_norm": 0.06553341448307037, "global_step": 4128, "epoch": 38, "lr": 9.850992273162882e-05} +{"train_loss": 0.023434165865182877, "grad_norm": 0.07224033772945404, "global_step": 4129, "epoch": 38, "lr": 9.850920449107956e-05} +{"train_loss": 0.018844882026314735, "grad_norm": 0.0534420870244503, "global_step": 4130, "epoch": 38, "lr": 9.850848608009005e-05} +{"train_loss": 0.01924210786819458, "grad_norm": 0.07148697972297668, "global_step": 4131, "epoch": 38, "lr": 9.850776749866283e-05} +{"train_loss": 0.019446510821580887, "grad_norm": 0.08153098076581955, "global_step": 4132, "epoch": 38, "lr": 9.85070487468004e-05} +{"train_loss": 0.019933904855037637, "grad_norm": 0.08214343339204788, "global_step": 4133, "epoch": 38, "lr": 9.85063298245053e-05, "val_loss": 0.019393088296055794} +{"train_loss": 0.020753327757120132, "grad_norm": 0.040575411170721054, "global_step": 4134, "epoch": 39, "lr": 9.850561073178004e-05} +{"train_loss": 0.017791390419006348, "grad_norm": 0.05633338540792465, "global_step": 4135, "epoch": 39, "lr": 9.850489146862718e-05} +{"train_loss": 0.020866824313998222, "grad_norm": 0.062400370836257935, "global_step": 4136, "epoch": 39, "lr": 9.85041720350492e-05} +{"train_loss": 0.018540041521191597, "grad_norm": 0.11686024814844131, "global_step": 4137, "epoch": 39, "lr": 9.850345243104868e-05} +{"train_loss": 0.020680181682109833, "grad_norm": 0.08086961507797241, "global_step": 4138, "epoch": 39, "lr": 9.85027326566281e-05} +{"train_loss": 0.019066186621785164, "grad_norm": 0.049989014863967896, "global_step": 4139, "epoch": 39, "lr": 9.850201271179002e-05} +{"train_loss": 0.018802279606461525, "grad_norm": 0.044499728828668594, "global_step": 4140, "epoch": 39, "lr": 9.850129259653696e-05} +{"train_loss": 0.018593067303299904, "grad_norm": 0.10855789482593536, "global_step": 4141, "epoch": 39, "lr": 9.850057231087146e-05} +{"train_loss": 0.020785370841622353, "grad_norm": 0.03776935115456581, "global_step": 4142, "epoch": 39, "lr": 9.849985185479602e-05} +{"train_loss": 0.02017783932387829, "grad_norm": 0.05241646245121956, "global_step": 4143, "epoch": 39, "lr": 9.84991312283132e-05} +{"train_loss": 0.019630173221230507, "grad_norm": 0.07518386095762253, "global_step": 4144, "epoch": 39, "lr": 9.849841043142553e-05} +{"train_loss": 0.01981838047504425, "grad_norm": 0.1212216168642044, "global_step": 4145, "epoch": 39, "lr": 9.849768946413554e-05} +{"train_loss": 0.01940171793103218, "grad_norm": 0.051747407764196396, "global_step": 4146, "epoch": 39, "lr": 9.849696832644576e-05} +{"train_loss": 0.018119430169463158, "grad_norm": 0.05745508149266243, "global_step": 4147, "epoch": 39, "lr": 9.849624701835872e-05} +{"train_loss": 0.01774430088698864, "grad_norm": 0.08709154278039932, "global_step": 4148, "epoch": 39, "lr": 9.849552553987695e-05} +{"train_loss": 0.019807923585176468, "grad_norm": 0.038947802037000656, "global_step": 4149, "epoch": 39, "lr": 9.849480389100301e-05} +{"train_loss": 0.01977447420358658, "grad_norm": 0.10311437398195267, "global_step": 4150, "epoch": 39, "lr": 9.84940820717394e-05} +{"train_loss": 0.019792266190052032, "grad_norm": 0.06468972563743591, "global_step": 4151, "epoch": 39, "lr": 9.849336008208869e-05} +{"train_loss": 0.01913996785879135, "grad_norm": 0.049235060811042786, "global_step": 4152, "epoch": 39, "lr": 9.849263792205342e-05} +{"train_loss": 0.020681653171777725, "grad_norm": 0.05915823206305504, "global_step": 4153, "epoch": 39, "lr": 9.849191559163607e-05} +{"train_loss": 0.02025391161441803, "grad_norm": 0.0791422501206398, "global_step": 4154, "epoch": 39, "lr": 9.849119309083923e-05} +{"train_loss": 0.019092943519353867, "grad_norm": 0.10723290592432022, "global_step": 4155, "epoch": 39, "lr": 9.849047041966543e-05} +{"train_loss": 0.01902293786406517, "grad_norm": 0.0457516685128212, "global_step": 4156, "epoch": 39, "lr": 9.848974757811721e-05} +{"train_loss": 0.018857717514038086, "grad_norm": 0.1139710396528244, "global_step": 4157, "epoch": 39, "lr": 9.84890245661971e-05} +{"train_loss": 0.019704949110746384, "grad_norm": 0.06322997808456421, "global_step": 4158, "epoch": 39, "lr": 9.848830138390764e-05} +{"train_loss": 0.01962585747241974, "grad_norm": 0.05861985683441162, "global_step": 4159, "epoch": 39, "lr": 9.848757803125138e-05} +{"train_loss": 0.02287287265062332, "grad_norm": 0.0471152663230896, "global_step": 4160, "epoch": 39, "lr": 9.848685450823086e-05} +{"train_loss": 0.019733212888240814, "grad_norm": 0.1069888100028038, "global_step": 4161, "epoch": 39, "lr": 9.848613081484862e-05} +{"train_loss": 0.021586980670690536, "grad_norm": 0.043924301862716675, "global_step": 4162, "epoch": 39, "lr": 9.84854069511072e-05} +{"train_loss": 0.017046358436346054, "grad_norm": 0.05072489008307457, "global_step": 4163, "epoch": 39, "lr": 9.848468291700914e-05} +{"train_loss": 0.01975804939866066, "grad_norm": 0.06095203757286072, "global_step": 4164, "epoch": 39, "lr": 9.8483958712557e-05} +{"train_loss": 0.018906202167272568, "grad_norm": 0.07021143287420273, "global_step": 4165, "epoch": 39, "lr": 9.84832343377533e-05} +{"train_loss": 0.02157553657889366, "grad_norm": 0.0974750816822052, "global_step": 4166, "epoch": 39, "lr": 9.848250979260061e-05} +{"train_loss": 0.019562944769859314, "grad_norm": 0.06522703915834427, "global_step": 4167, "epoch": 39, "lr": 9.848178507710147e-05} +{"train_loss": 0.01892169937491417, "grad_norm": 0.04744253680109978, "global_step": 4168, "epoch": 39, "lr": 9.848106019125843e-05} +{"train_loss": 0.020946837961673737, "grad_norm": 0.10124179720878601, "global_step": 4169, "epoch": 39, "lr": 9.8480335135074e-05} +{"train_loss": 0.02055465430021286, "grad_norm": 0.10292242467403412, "global_step": 4170, "epoch": 39, "lr": 9.847960990855076e-05} +{"train_loss": 0.01914394460618496, "grad_norm": 0.08945703506469727, "global_step": 4171, "epoch": 39, "lr": 9.847888451169127e-05} +{"train_loss": 0.022142883390188217, "grad_norm": 0.06632133573293686, "global_step": 4172, "epoch": 39, "lr": 9.847815894449805e-05} +{"train_loss": 0.01866888999938965, "grad_norm": 0.03713637962937355, "global_step": 4173, "epoch": 39, "lr": 9.847743320697368e-05} +{"train_loss": 0.020733168348670006, "grad_norm": 0.07595372200012207, "global_step": 4174, "epoch": 39, "lr": 9.847670729912069e-05} +{"train_loss": 0.02097943425178528, "grad_norm": 0.07396286725997925, "global_step": 4175, "epoch": 39, "lr": 9.847598122094161e-05} +{"train_loss": 0.02185109257698059, "grad_norm": 0.050933580845594406, "global_step": 4176, "epoch": 39, "lr": 9.847525497243903e-05} +{"train_loss": 0.021985765546560287, "grad_norm": 0.05213760584592819, "global_step": 4177, "epoch": 39, "lr": 9.847452855361547e-05} +{"train_loss": 0.01967180147767067, "grad_norm": 0.0518304705619812, "global_step": 4178, "epoch": 39, "lr": 9.847380196447353e-05} +{"train_loss": 0.018811799585819244, "grad_norm": 0.11755657196044922, "global_step": 4179, "epoch": 39, "lr": 9.84730752050157e-05} +{"train_loss": 0.020972391590476036, "grad_norm": 0.10434235632419586, "global_step": 4180, "epoch": 39, "lr": 9.847234827524457e-05} +{"train_loss": 0.018093183636665344, "grad_norm": 0.04662341624498367, "global_step": 4181, "epoch": 39, "lr": 9.84716211751627e-05} +{"train_loss": 0.020713960751891136, "grad_norm": 0.06569044291973114, "global_step": 4182, "epoch": 39, "lr": 9.847089390477262e-05} +{"train_loss": 0.021365247666835785, "grad_norm": 0.06808988004922867, "global_step": 4183, "epoch": 39, "lr": 9.847016646407691e-05} +{"train_loss": 0.02046830952167511, "grad_norm": 0.044731255620718, "global_step": 4184, "epoch": 39, "lr": 9.846943885307811e-05} +{"train_loss": 0.01918569952249527, "grad_norm": 0.05550941824913025, "global_step": 4185, "epoch": 39, "lr": 9.846871107177878e-05} +{"train_loss": 0.021020323038101196, "grad_norm": 0.1065320074558258, "global_step": 4186, "epoch": 39, "lr": 9.846798312018146e-05} +{"train_loss": 0.019213618710637093, "grad_norm": 0.08395994454622269, "global_step": 4187, "epoch": 39, "lr": 9.846725499828875e-05} +{"train_loss": 0.019819222390651703, "grad_norm": 0.04452737048268318, "global_step": 4188, "epoch": 39, "lr": 9.846652670610317e-05} +{"train_loss": 0.018787547945976257, "grad_norm": 0.06020190194249153, "global_step": 4189, "epoch": 39, "lr": 9.84657982436273e-05} +{"train_loss": 0.021046537905931473, "grad_norm": 0.06900794059038162, "global_step": 4190, "epoch": 39, "lr": 9.84650696108637e-05} +{"train_loss": 0.020269084721803665, "grad_norm": 0.04762011766433716, "global_step": 4191, "epoch": 39, "lr": 9.846434080781491e-05} +{"train_loss": 0.021754316985607147, "grad_norm": 0.1270742118358612, "global_step": 4192, "epoch": 39, "lr": 9.84636118344835e-05} +{"train_loss": 0.022540010511875153, "grad_norm": 0.044611841440200806, "global_step": 4193, "epoch": 39, "lr": 9.846288269087205e-05} +{"train_loss": 0.018186960369348526, "grad_norm": 0.04613730311393738, "global_step": 4194, "epoch": 39, "lr": 9.846215337698309e-05} +{"train_loss": 0.018876250833272934, "grad_norm": 0.03758787736296654, "global_step": 4195, "epoch": 39, "lr": 9.84614238928192e-05} +{"train_loss": 0.018402226269245148, "grad_norm": 0.12477422505617142, "global_step": 4196, "epoch": 39, "lr": 9.846069423838296e-05} +{"train_loss": 0.023045087233185768, "grad_norm": 0.06521344184875488, "global_step": 4197, "epoch": 39, "lr": 9.845996441367689e-05} +{"train_loss": 0.020792581140995026, "grad_norm": 0.07799466699361801, "global_step": 4198, "epoch": 39, "lr": 9.845923441870359e-05} +{"train_loss": 0.019253447651863098, "grad_norm": 0.0563468411564827, "global_step": 4199, "epoch": 39, "lr": 9.845850425346563e-05} +{"train_loss": 0.022334851324558258, "grad_norm": 0.06158948689699173, "global_step": 4200, "epoch": 39, "lr": 9.845777391796553e-05} +{"train_loss": 0.02064870111644268, "grad_norm": 0.05315687879920006, "global_step": 4201, "epoch": 39, "lr": 9.845704341220591e-05} +{"train_loss": 0.0192929208278656, "grad_norm": 0.06479749083518982, "global_step": 4202, "epoch": 39, "lr": 9.84563127361893e-05} +{"train_loss": 0.023299722000956535, "grad_norm": 0.048781655728816986, "global_step": 4203, "epoch": 39, "lr": 9.84555818899183e-05} +{"train_loss": 0.018945002928376198, "grad_norm": 0.06625284999608994, "global_step": 4204, "epoch": 39, "lr": 9.845485087339543e-05} +{"train_loss": 0.02156035602092743, "grad_norm": 0.08440151065587997, "global_step": 4205, "epoch": 39, "lr": 9.84541196866233e-05} +{"train_loss": 0.020270809531211853, "grad_norm": 0.09226932376623154, "global_step": 4206, "epoch": 39, "lr": 9.845338832960446e-05} +{"train_loss": 0.018832148984074593, "grad_norm": 0.10139786452054977, "global_step": 4207, "epoch": 39, "lr": 9.84526568023415e-05} +{"train_loss": 0.018641598522663116, "grad_norm": 0.0974382758140564, "global_step": 4208, "epoch": 39, "lr": 9.845192510483696e-05} +{"train_loss": 0.02067868411540985, "grad_norm": 0.03946414589881897, "global_step": 4209, "epoch": 39, "lr": 9.845119323709344e-05} +{"train_loss": 0.016625773161649704, "grad_norm": 0.06965918838977814, "global_step": 4210, "epoch": 39, "lr": 9.84504611991135e-05} +{"train_loss": 0.021077662706375122, "grad_norm": 0.11225038766860962, "global_step": 4211, "epoch": 39, "lr": 9.844972899089969e-05} +{"train_loss": 0.02189449965953827, "grad_norm": 0.053386278450489044, "global_step": 4212, "epoch": 39, "lr": 9.844899661245461e-05} +{"train_loss": 0.01854914054274559, "grad_norm": 0.1450001448392868, "global_step": 4213, "epoch": 39, "lr": 9.844826406378084e-05} +{"train_loss": 0.018580295145511627, "grad_norm": 0.047301284968853, "global_step": 4214, "epoch": 39, "lr": 9.844753134488091e-05} +{"train_loss": 0.019449016079306602, "grad_norm": 0.0343443900346756, "global_step": 4215, "epoch": 39, "lr": 9.844679845575746e-05} +{"train_loss": 0.021153181791305542, "grad_norm": 0.07076713442802429, "global_step": 4216, "epoch": 39, "lr": 9.844606539641302e-05} +{"train_loss": 0.01923983171582222, "grad_norm": 0.03756209835410118, "global_step": 4217, "epoch": 39, "lr": 9.844533216685016e-05} +{"train_loss": 0.019289931282401085, "grad_norm": 0.09111784398555756, "global_step": 4218, "epoch": 39, "lr": 9.844459876707147e-05} +{"train_loss": 0.0174972265958786, "grad_norm": 0.03465821221470833, "global_step": 4219, "epoch": 39, "lr": 9.844386519707953e-05} +{"train_loss": 0.021370774134993553, "grad_norm": 0.15004174411296844, "global_step": 4220, "epoch": 39, "lr": 9.844313145687694e-05} +{"train_loss": 0.02032879739999771, "grad_norm": 0.04097335413098335, "global_step": 4221, "epoch": 39, "lr": 9.844239754646622e-05} +{"train_loss": 0.021128274500370026, "grad_norm": 0.06587647646665573, "global_step": 4222, "epoch": 39, "lr": 9.844166346585e-05} +{"train_loss": 0.017958441749215126, "grad_norm": 0.1203775629401207, "global_step": 4223, "epoch": 39, "lr": 9.844092921503085e-05} +{"train_loss": 0.020653430372476578, "grad_norm": 0.10448232293128967, "global_step": 4224, "epoch": 39, "lr": 9.844019479401132e-05} +{"train_loss": 0.019568664953112602, "grad_norm": 0.05992424115538597, "global_step": 4225, "epoch": 39, "lr": 9.843946020279401e-05} +{"train_loss": 0.020681235939264297, "grad_norm": 0.059038735926151276, "global_step": 4226, "epoch": 39, "lr": 9.843872544138154e-05} +{"train_loss": 0.01784619130194187, "grad_norm": 0.036783747375011444, "global_step": 4227, "epoch": 39, "lr": 9.84379905097764e-05} +{"train_loss": 0.01799611747264862, "grad_norm": 0.053494539111852646, "global_step": 4228, "epoch": 39, "lr": 9.843725540798126e-05} +{"train_loss": 0.022885821759700775, "grad_norm": 0.055228691548109055, "global_step": 4229, "epoch": 39, "lr": 9.843652013599866e-05} +{"train_loss": 0.019101649522781372, "grad_norm": 0.0998576208949089, "global_step": 4230, "epoch": 39, "lr": 9.84357846938312e-05} +{"train_loss": 0.017162704840302467, "grad_norm": 0.04733920097351074, "global_step": 4231, "epoch": 39, "lr": 9.843504908148146e-05} +{"train_loss": 0.020367469638586044, "grad_norm": 0.04402288794517517, "global_step": 4232, "epoch": 39, "lr": 9.843431329895201e-05} +{"train_loss": 0.022342074662446976, "grad_norm": 0.07585278898477554, "global_step": 4233, "epoch": 39, "lr": 9.843357734624545e-05} +{"train_loss": 0.01835484430193901, "grad_norm": 0.06388799846172333, "global_step": 4234, "epoch": 39, "lr": 9.843284122336436e-05} +{"train_loss": 0.016975102946162224, "grad_norm": 0.12250732630491257, "global_step": 4235, "epoch": 39, "lr": 9.843210493031134e-05} +{"train_loss": 0.017931319773197174, "grad_norm": 0.037491653114557266, "global_step": 4236, "epoch": 39, "lr": 9.843136846708896e-05} +{"train_loss": 0.020503118634223938, "grad_norm": 0.13041412830352783, "global_step": 4237, "epoch": 39, "lr": 9.84306318336998e-05} +{"train_loss": 0.01814393326640129, "grad_norm": 0.10559264570474625, "global_step": 4238, "epoch": 39, "lr": 9.842989503014648e-05} +{"train_loss": 0.019845455287481256, "grad_norm": 0.11876573413610458, "global_step": 4239, "epoch": 39, "lr": 9.842915805643155e-05, "val_loss": 0.01963530108332634} +{"train_loss": 0.02002742327749729, "grad_norm": 0.1659451723098755, "global_step": 4240, "epoch": 40, "lr": 9.842842091255764e-05} +{"train_loss": 0.02394852787256241, "grad_norm": 0.06921017169952393, "global_step": 4241, "epoch": 40, "lr": 9.842768359852732e-05} +{"train_loss": 0.019242167472839355, "grad_norm": 0.07031071931123734, "global_step": 4242, "epoch": 40, "lr": 9.842694611434318e-05} +{"train_loss": 0.017743730917572975, "grad_norm": 0.04039347916841507, "global_step": 4243, "epoch": 40, "lr": 9.84262084600078e-05} +{"train_loss": 0.020878545939922333, "grad_norm": 0.09747623652219772, "global_step": 4244, "epoch": 40, "lr": 9.842547063552379e-05} +{"train_loss": 0.020860105752944946, "grad_norm": 0.05892474949359894, "global_step": 4245, "epoch": 40, "lr": 9.842473264089373e-05} +{"train_loss": 0.01926422119140625, "grad_norm": 0.06347931176424026, "global_step": 4246, "epoch": 40, "lr": 9.842399447612023e-05} +{"train_loss": 0.01769465208053589, "grad_norm": 0.053368061780929565, "global_step": 4247, "epoch": 40, "lr": 9.842325614120586e-05} +{"train_loss": 0.020185202360153198, "grad_norm": 0.07503221929073334, "global_step": 4248, "epoch": 40, "lr": 9.842251763615323e-05} +{"train_loss": 0.0235951729118824, "grad_norm": 0.050791893154382706, "global_step": 4249, "epoch": 40, "lr": 9.842177896096494e-05} +{"train_loss": 0.016263101249933243, "grad_norm": 0.14633414149284363, "global_step": 4250, "epoch": 40, "lr": 9.842104011564357e-05} +{"train_loss": 0.020405150949954987, "grad_norm": 0.11341243237257004, "global_step": 4251, "epoch": 40, "lr": 9.842030110019172e-05} +{"train_loss": 0.018929142504930496, "grad_norm": 0.09738544374704361, "global_step": 4252, "epoch": 40, "lr": 9.841956191461199e-05} +{"train_loss": 0.022215303033590317, "grad_norm": 0.10231010615825653, "global_step": 4253, "epoch": 40, "lr": 9.841882255890698e-05} +{"train_loss": 0.022442901507019997, "grad_norm": 0.04592300206422806, "global_step": 4254, "epoch": 40, "lr": 9.841808303307929e-05} +{"train_loss": 0.022343849763274193, "grad_norm": 0.09581974893808365, "global_step": 4255, "epoch": 40, "lr": 9.84173433371315e-05} +{"train_loss": 0.020716989412903786, "grad_norm": 0.062234144657850266, "global_step": 4256, "epoch": 40, "lr": 9.841660347106622e-05} +{"train_loss": 0.020722828805446625, "grad_norm": 0.0761529952287674, "global_step": 4257, "epoch": 40, "lr": 9.841586343488607e-05} +{"train_loss": 0.01998290792107582, "grad_norm": 0.0674748569726944, "global_step": 4258, "epoch": 40, "lr": 9.841512322859361e-05} +{"train_loss": 0.021329127252101898, "grad_norm": 0.042358264327049255, "global_step": 4259, "epoch": 40, "lr": 9.841438285219149e-05} +{"train_loss": 0.019870735704898834, "grad_norm": 0.10204579681158066, "global_step": 4260, "epoch": 40, "lr": 9.841364230568227e-05} +{"train_loss": 0.0179741308093071, "grad_norm": 0.08687490969896317, "global_step": 4261, "epoch": 40, "lr": 9.841290158906855e-05} +{"train_loss": 0.020087433978915215, "grad_norm": 0.07072658091783524, "global_step": 4262, "epoch": 40, "lr": 9.841216070235296e-05} +{"train_loss": 0.020955795422196388, "grad_norm": 0.18513205647468567, "global_step": 4263, "epoch": 40, "lr": 9.841141964553809e-05} +{"train_loss": 0.020976472645998, "grad_norm": 0.1018616110086441, "global_step": 4264, "epoch": 40, "lr": 9.841067841862655e-05} +{"train_loss": 0.018585234880447388, "grad_norm": 0.06341543793678284, "global_step": 4265, "epoch": 40, "lr": 9.840993702162094e-05} +{"train_loss": 0.01693139225244522, "grad_norm": 0.1498887538909912, "global_step": 4266, "epoch": 40, "lr": 9.840919545452385e-05} +{"train_loss": 0.023898489773273468, "grad_norm": 0.13317076861858368, "global_step": 4267, "epoch": 40, "lr": 9.840845371733791e-05} +{"train_loss": 0.02058331109583378, "grad_norm": 0.047450434416532516, "global_step": 4268, "epoch": 40, "lr": 9.840771181006572e-05} +{"train_loss": 0.0191043633967638, "grad_norm": 0.09965145587921143, "global_step": 4269, "epoch": 40, "lr": 9.840696973270987e-05} +{"train_loss": 0.019456081092357635, "grad_norm": 0.07599349319934845, "global_step": 4270, "epoch": 40, "lr": 9.840622748527299e-05} +{"train_loss": 0.020407147705554962, "grad_norm": 0.09423798322677612, "global_step": 4271, "epoch": 40, "lr": 9.840548506775768e-05} +{"train_loss": 0.016381900757551193, "grad_norm": 0.15050768852233887, "global_step": 4272, "epoch": 40, "lr": 9.840474248016654e-05} +{"train_loss": 0.020636364817619324, "grad_norm": 0.10827910900115967, "global_step": 4273, "epoch": 40, "lr": 9.84039997225022e-05} +{"train_loss": 0.021510131657123566, "grad_norm": 0.07831810414791107, "global_step": 4274, "epoch": 40, "lr": 9.840325679476723e-05} +{"train_loss": 0.020804956555366516, "grad_norm": 0.07841622084379196, "global_step": 4275, "epoch": 40, "lr": 9.840251369696428e-05} +{"train_loss": 0.020068613812327385, "grad_norm": 0.1334409862756729, "global_step": 4276, "epoch": 40, "lr": 9.840177042909595e-05} +{"train_loss": 0.020700091496109962, "grad_norm": 0.08083054423332214, "global_step": 4277, "epoch": 40, "lr": 9.840102699116485e-05} +{"train_loss": 0.018550340086221695, "grad_norm": 0.11374380439519882, "global_step": 4278, "epoch": 40, "lr": 9.840028338317357e-05} +{"train_loss": 0.016754671931266785, "grad_norm": 0.036296576261520386, "global_step": 4279, "epoch": 40, "lr": 9.839953960512476e-05} +{"train_loss": 0.01861022599041462, "grad_norm": 0.06897950172424316, "global_step": 4280, "epoch": 40, "lr": 9.839879565702101e-05} +{"train_loss": 0.02159935235977173, "grad_norm": 0.07770586758852005, "global_step": 4281, "epoch": 40, "lr": 9.839805153886495e-05} +{"train_loss": 0.018535684794187546, "grad_norm": 0.03295518830418587, "global_step": 4282, "epoch": 40, "lr": 9.839730725065917e-05} +{"train_loss": 0.019420750439167023, "grad_norm": 0.06935993582010269, "global_step": 4283, "epoch": 40, "lr": 9.839656279240632e-05} +{"train_loss": 0.02395307645201683, "grad_norm": 0.10932344198226929, "global_step": 4284, "epoch": 40, "lr": 9.839581816410898e-05} +{"train_loss": 0.02128266915678978, "grad_norm": 0.1633501946926117, "global_step": 4285, "epoch": 40, "lr": 9.839507336576978e-05} +{"train_loss": 0.01863979548215866, "grad_norm": 0.10927660018205643, "global_step": 4286, "epoch": 40, "lr": 9.839432839739135e-05} +{"train_loss": 0.020560510456562042, "grad_norm": 0.06371582299470901, "global_step": 4287, "epoch": 40, "lr": 9.839358325897629e-05} +{"train_loss": 0.02140161581337452, "grad_norm": 0.10364406555891037, "global_step": 4288, "epoch": 40, "lr": 9.839283795052722e-05} +{"train_loss": 0.019005514681339264, "grad_norm": 0.07357405126094818, "global_step": 4289, "epoch": 40, "lr": 9.839209247204677e-05} +{"train_loss": 0.021807607263326645, "grad_norm": 0.07227052003145218, "global_step": 4290, "epoch": 40, "lr": 9.839134682353754e-05} +{"train_loss": 0.018898028880357742, "grad_norm": 0.053666677325963974, "global_step": 4291, "epoch": 40, "lr": 9.839060100500217e-05} +{"train_loss": 0.019040394574403763, "grad_norm": 0.05961298197507858, "global_step": 4292, "epoch": 40, "lr": 9.838985501644328e-05} +{"train_loss": 0.01992257684469223, "grad_norm": 0.07381080090999603, "global_step": 4293, "epoch": 40, "lr": 9.838910885786347e-05} +{"train_loss": 0.018671246245503426, "grad_norm": 0.045726824551820755, "global_step": 4294, "epoch": 40, "lr": 9.838836252926539e-05} +{"train_loss": 0.01854413002729416, "grad_norm": 0.1322937309741974, "global_step": 4295, "epoch": 40, "lr": 9.838761603065163e-05} +{"train_loss": 0.020013093948364258, "grad_norm": 0.08088259398937225, "global_step": 4296, "epoch": 40, "lr": 9.838686936202485e-05} +{"train_loss": 0.019480109214782715, "grad_norm": 0.047402601689100266, "global_step": 4297, "epoch": 40, "lr": 9.838612252338764e-05} +{"train_loss": 0.019464343786239624, "grad_norm": 0.057042770087718964, "global_step": 4298, "epoch": 40, "lr": 9.838537551474265e-05} +{"train_loss": 0.01896747574210167, "grad_norm": 0.11070306599140167, "global_step": 4299, "epoch": 40, "lr": 9.838462833609248e-05} +{"train_loss": 0.01876278966665268, "grad_norm": 0.07514560967683792, "global_step": 4300, "epoch": 40, "lr": 9.838388098743978e-05} +{"train_loss": 0.01807299070060253, "grad_norm": 0.05197702348232269, "global_step": 4301, "epoch": 40, "lr": 9.838313346878715e-05} +{"train_loss": 0.018593335524201393, "grad_norm": 0.06342439353466034, "global_step": 4302, "epoch": 40, "lr": 9.838238578013723e-05} +{"train_loss": 0.022517133504152298, "grad_norm": 0.05152983218431473, "global_step": 4303, "epoch": 40, "lr": 9.838163792149267e-05} +{"train_loss": 0.019195744767785072, "grad_norm": 0.11976120620965958, "global_step": 4304, "epoch": 40, "lr": 9.838088989285605e-05} +{"train_loss": 0.021493539214134216, "grad_norm": 0.04102964326739311, "global_step": 4305, "epoch": 40, "lr": 9.838014169423004e-05} +{"train_loss": 0.01828969456255436, "grad_norm": 0.06295166909694672, "global_step": 4306, "epoch": 40, "lr": 9.837939332561724e-05} +{"train_loss": 0.019302653148770332, "grad_norm": 0.03718831390142441, "global_step": 4307, "epoch": 40, "lr": 9.83786447870203e-05} +{"train_loss": 0.018477309495210648, "grad_norm": 0.04598795622587204, "global_step": 4308, "epoch": 40, "lr": 9.837789607844183e-05} +{"train_loss": 0.022386737167835236, "grad_norm": 0.08298656344413757, "global_step": 4309, "epoch": 40, "lr": 9.83771471998845e-05} +{"train_loss": 0.02152915671467781, "grad_norm": 0.08309400081634521, "global_step": 4310, "epoch": 40, "lr": 9.837639815135087e-05} +{"train_loss": 0.020727772265672684, "grad_norm": 0.0739501565694809, "global_step": 4311, "epoch": 40, "lr": 9.837564893284365e-05} +{"train_loss": 0.01717275381088257, "grad_norm": 0.04236481711268425, "global_step": 4312, "epoch": 40, "lr": 9.837489954436543e-05} +{"train_loss": 0.021474115550518036, "grad_norm": 0.0900777205824852, "global_step": 4313, "epoch": 40, "lr": 9.837414998591884e-05} +{"train_loss": 0.01989533007144928, "grad_norm": 0.04615391418337822, "global_step": 4314, "epoch": 40, "lr": 9.837340025750653e-05} +{"train_loss": 0.021637987345457077, "grad_norm": 0.04383685439825058, "global_step": 4315, "epoch": 40, "lr": 9.837265035913112e-05} +{"train_loss": 0.01938600279390812, "grad_norm": 0.06459932774305344, "global_step": 4316, "epoch": 40, "lr": 9.837190029079528e-05} +{"train_loss": 0.01903311163187027, "grad_norm": 0.042858775705099106, "global_step": 4317, "epoch": 40, "lr": 9.837115005250159e-05} +{"train_loss": 0.018551304936408997, "grad_norm": 0.04492467641830444, "global_step": 4318, "epoch": 40, "lr": 9.837039964425273e-05} +{"train_loss": 0.018853869289159775, "grad_norm": 0.07778280228376389, "global_step": 4319, "epoch": 40, "lr": 9.836964906605133e-05} +{"train_loss": 0.017094811424613, "grad_norm": 0.050371408462524414, "global_step": 4320, "epoch": 40, "lr": 9.83688983179e-05} +{"train_loss": 0.02010098472237587, "grad_norm": 0.05026392266154289, "global_step": 4321, "epoch": 40, "lr": 9.836814739980141e-05} +{"train_loss": 0.019288934767246246, "grad_norm": 0.07483828812837601, "global_step": 4322, "epoch": 40, "lr": 9.836739631175818e-05} +{"train_loss": 0.01915895938873291, "grad_norm": 0.04202887415885925, "global_step": 4323, "epoch": 40, "lr": 9.836664505377294e-05} +{"train_loss": 0.019016150385141373, "grad_norm": 0.060867857187986374, "global_step": 4324, "epoch": 40, "lr": 9.836589362584837e-05} +{"train_loss": 0.016993975266814232, "grad_norm": 0.056734684854745865, "global_step": 4325, "epoch": 40, "lr": 9.836514202798705e-05} +{"train_loss": 0.01927844062447548, "grad_norm": 0.04426262527704239, "global_step": 4326, "epoch": 40, "lr": 9.836439026019168e-05} +{"train_loss": 0.014997566118836403, "grad_norm": 0.024344170466065407, "global_step": 4327, "epoch": 40, "lr": 9.836363832246487e-05} +{"train_loss": 0.020400768145918846, "grad_norm": 0.04559189826250076, "global_step": 4328, "epoch": 40, "lr": 9.836288621480926e-05} +{"train_loss": 0.020836565643548965, "grad_norm": 0.09010366350412369, "global_step": 4329, "epoch": 40, "lr": 9.836213393722751e-05} +{"train_loss": 0.02015024423599243, "grad_norm": 0.10231182724237442, "global_step": 4330, "epoch": 40, "lr": 9.836138148972225e-05} +{"train_loss": 0.018320459872484207, "grad_norm": 0.10732117295265198, "global_step": 4331, "epoch": 40, "lr": 9.836062887229613e-05} +{"train_loss": 0.022883132100105286, "grad_norm": 0.05355452746152878, "global_step": 4332, "epoch": 40, "lr": 9.835987608495179e-05} +{"train_loss": 0.01705196686089039, "grad_norm": 0.0333467498421669, "global_step": 4333, "epoch": 40, "lr": 9.835912312769186e-05} +{"train_loss": 0.021367503330111504, "grad_norm": 0.04357626289129257, "global_step": 4334, "epoch": 40, "lr": 9.835837000051903e-05} +{"train_loss": 0.01776411198079586, "grad_norm": 0.06057475879788399, "global_step": 4335, "epoch": 40, "lr": 9.835761670343591e-05} +{"train_loss": 0.020157452672719955, "grad_norm": 0.06509888917207718, "global_step": 4336, "epoch": 40, "lr": 9.835686323644515e-05} +{"train_loss": 0.01929328590631485, "grad_norm": 0.04201490059494972, "global_step": 4337, "epoch": 40, "lr": 9.83561095995494e-05} +{"train_loss": 0.019846521317958832, "grad_norm": 0.04157672077417374, "global_step": 4338, "epoch": 40, "lr": 9.835535579275132e-05} +{"train_loss": 0.01984269730746746, "grad_norm": 0.13916802406311035, "global_step": 4339, "epoch": 40, "lr": 9.835460181605354e-05} +{"train_loss": 0.02156216651201248, "grad_norm": 0.059770453721284866, "global_step": 4340, "epoch": 40, "lr": 9.835384766945873e-05} +{"train_loss": 0.01967473328113556, "grad_norm": 0.03695099055767059, "global_step": 4341, "epoch": 40, "lr": 9.835309335296952e-05} +{"train_loss": 0.017901279032230377, "grad_norm": 0.043440427631139755, "global_step": 4342, "epoch": 40, "lr": 9.835233886658855e-05} +{"train_loss": 0.020226657390594482, "grad_norm": 0.05426213890314102, "global_step": 4343, "epoch": 40, "lr": 9.835158421031852e-05} +{"train_loss": 0.019709933549165726, "grad_norm": 0.05428801849484444, "global_step": 4344, "epoch": 40, "lr": 9.835082938416202e-05} +{"train_loss": 0.019797915223775046, "grad_norm": 0.05934056267142296, "global_step": 4345, "epoch": 40, "lr": 9.835007438812177e-05, "train_batch_mse_error": 0.01790257915854454, "val_loss": 0.020013924688100815} +{"train_loss": 0.018561426550149918, "grad_norm": 0.0828714668750763, "global_step": 4346, "epoch": 41, "lr": 9.834931922220035e-05} +{"train_loss": 0.019964549690485, "grad_norm": 0.0607273131608963, "global_step": 4347, "epoch": 41, "lr": 9.834856388640046e-05} +{"train_loss": 0.017880015075206757, "grad_norm": 0.051826879382133484, "global_step": 4348, "epoch": 41, "lr": 9.834780838072474e-05} +{"train_loss": 0.02248665690422058, "grad_norm": 0.10746211558580399, "global_step": 4349, "epoch": 41, "lr": 9.834705270517588e-05} +{"train_loss": 0.017628703266382217, "grad_norm": 0.050025854259729385, "global_step": 4350, "epoch": 41, "lr": 9.834629685975647e-05} +{"train_loss": 0.0181424543261528, "grad_norm": 0.07553126662969589, "global_step": 4351, "epoch": 41, "lr": 9.83455408444692e-05} +{"train_loss": 0.02033621072769165, "grad_norm": 0.03485402837395668, "global_step": 4352, "epoch": 41, "lr": 9.834478465931672e-05} +{"train_loss": 0.017379481345415115, "grad_norm": 0.04287347570061684, "global_step": 4353, "epoch": 41, "lr": 9.834402830430171e-05} +{"train_loss": 0.018079981207847595, "grad_norm": 0.09188107401132584, "global_step": 4354, "epoch": 41, "lr": 9.83432717794268e-05} +{"train_loss": 0.019171161577105522, "grad_norm": 0.05623732507228851, "global_step": 4355, "epoch": 41, "lr": 9.834251508469465e-05} +{"train_loss": 0.021607931703329086, "grad_norm": 0.06524482369422913, "global_step": 4356, "epoch": 41, "lr": 9.834175822010794e-05} +{"train_loss": 0.02098991721868515, "grad_norm": 0.045738834887742996, "global_step": 4357, "epoch": 41, "lr": 9.834100118566929e-05} +{"train_loss": 0.020636260509490967, "grad_norm": 0.06732749193906784, "global_step": 4358, "epoch": 41, "lr": 9.834024398138141e-05} +{"train_loss": 0.019426550716161728, "grad_norm": 0.04195345938205719, "global_step": 4359, "epoch": 41, "lr": 9.833948660724692e-05} +{"train_loss": 0.021235879510641098, "grad_norm": 0.0425475537776947, "global_step": 4360, "epoch": 41, "lr": 9.83387290632685e-05} +{"train_loss": 0.02043977752327919, "grad_norm": 0.08357218652963638, "global_step": 4361, "epoch": 41, "lr": 9.833797134944882e-05} +{"train_loss": 0.019000284373760223, "grad_norm": 0.07332297414541245, "global_step": 4362, "epoch": 41, "lr": 9.833721346579053e-05} +{"train_loss": 0.01908170059323311, "grad_norm": 0.04775036498904228, "global_step": 4363, "epoch": 41, "lr": 9.833645541229628e-05} +{"train_loss": 0.023187264800071716, "grad_norm": 0.1764109581708908, "global_step": 4364, "epoch": 41, "lr": 9.833569718896875e-05} +{"train_loss": 0.020107291638851166, "grad_norm": 0.0387684628367424, "global_step": 4365, "epoch": 41, "lr": 9.833493879581059e-05} +{"train_loss": 0.019994651898741722, "grad_norm": 0.07471305876970291, "global_step": 4366, "epoch": 41, "lr": 9.833418023282448e-05} +{"train_loss": 0.019843313843011856, "grad_norm": 0.04478438198566437, "global_step": 4367, "epoch": 41, "lr": 9.83334215000131e-05} +{"train_loss": 0.020570004358887672, "grad_norm": 0.05932558327913284, "global_step": 4368, "epoch": 41, "lr": 9.833266259737908e-05} +{"train_loss": 0.0192138459533453, "grad_norm": 0.06778205186128616, "global_step": 4369, "epoch": 41, "lr": 9.833190352492512e-05} +{"train_loss": 0.017515871673822403, "grad_norm": 0.06277729570865631, "global_step": 4370, "epoch": 41, "lr": 9.833114428265384e-05} +{"train_loss": 0.021384572610259056, "grad_norm": 0.19637222588062286, "global_step": 4371, "epoch": 41, "lr": 9.833038487056795e-05} +{"train_loss": 0.017351560294628143, "grad_norm": 0.0971183180809021, "global_step": 4372, "epoch": 41, "lr": 9.832962528867013e-05} +{"train_loss": 0.01871192455291748, "grad_norm": 0.07816696912050247, "global_step": 4373, "epoch": 41, "lr": 9.8328865536963e-05} +{"train_loss": 0.019765887409448624, "grad_norm": 0.07433511316776276, "global_step": 4374, "epoch": 41, "lr": 9.832810561544925e-05} +{"train_loss": 0.022057663649320602, "grad_norm": 0.05587208643555641, "global_step": 4375, "epoch": 41, "lr": 9.832734552413156e-05} +{"train_loss": 0.021732958033680916, "grad_norm": 0.06207806617021561, "global_step": 4376, "epoch": 41, "lr": 9.832658526301262e-05} +{"train_loss": 0.021168194711208344, "grad_norm": 0.05324447154998779, "global_step": 4377, "epoch": 41, "lr": 9.832582483209505e-05} +{"train_loss": 0.020710811018943787, "grad_norm": 0.0857819989323616, "global_step": 4378, "epoch": 41, "lr": 9.832506423138154e-05} +{"train_loss": 0.0196390300989151, "grad_norm": 0.043353091925382614, "global_step": 4379, "epoch": 41, "lr": 9.832430346087479e-05} +{"train_loss": 0.02195390872657299, "grad_norm": 0.11793139576911926, "global_step": 4380, "epoch": 41, "lr": 9.832354252057745e-05} +{"train_loss": 0.020932774990797043, "grad_norm": 0.038054611533880234, "global_step": 4381, "epoch": 41, "lr": 9.832278141049218e-05} +{"train_loss": 0.020859098061919212, "grad_norm": 0.05711885541677475, "global_step": 4382, "epoch": 41, "lr": 9.83220201306217e-05} +{"train_loss": 0.018313854932785034, "grad_norm": 0.14118212461471558, "global_step": 4383, "epoch": 41, "lr": 9.832125868096864e-05} +{"train_loss": 0.0195111446082592, "grad_norm": 0.06534720212221146, "global_step": 4384, "epoch": 41, "lr": 9.832049706153568e-05} +{"train_loss": 0.020276188850402832, "grad_norm": 0.1138104572892189, "global_step": 4385, "epoch": 41, "lr": 9.831973527232552e-05} +{"train_loss": 0.02233012765645981, "grad_norm": 0.07515814155340195, "global_step": 4386, "epoch": 41, "lr": 9.831897331334083e-05} +{"train_loss": 0.019088976085186005, "grad_norm": 0.03874539956450462, "global_step": 4387, "epoch": 41, "lr": 9.831821118458427e-05} +{"train_loss": 0.018915729597210884, "grad_norm": 0.07823731750249863, "global_step": 4388, "epoch": 41, "lr": 9.831744888605853e-05} +{"train_loss": 0.01950782909989357, "grad_norm": 0.040890809148550034, "global_step": 4389, "epoch": 41, "lr": 9.83166864177663e-05} +{"train_loss": 0.019501693546772003, "grad_norm": 0.08425131440162659, "global_step": 4390, "epoch": 41, "lr": 9.831592377971023e-05} +{"train_loss": 0.020201660692691803, "grad_norm": 0.05220252647995949, "global_step": 4391, "epoch": 41, "lr": 9.831516097189301e-05} +{"train_loss": 0.021919801831245422, "grad_norm": 0.06795139610767365, "global_step": 4392, "epoch": 41, "lr": 9.831439799431735e-05} +{"train_loss": 0.01859942451119423, "grad_norm": 0.0873093381524086, "global_step": 4393, "epoch": 41, "lr": 9.83136348469859e-05} +{"train_loss": 0.020714614540338516, "grad_norm": 0.12823931872844696, "global_step": 4394, "epoch": 41, "lr": 9.831287152990134e-05} +{"train_loss": 0.02088422328233719, "grad_norm": 0.1256081759929657, "global_step": 4395, "epoch": 41, "lr": 9.831210804306636e-05} +{"train_loss": 0.019466612488031387, "grad_norm": 0.04367813467979431, "global_step": 4396, "epoch": 41, "lr": 9.831134438648365e-05} +{"train_loss": 0.017390422523021698, "grad_norm": 0.04539267346262932, "global_step": 4397, "epoch": 41, "lr": 9.831058056015587e-05} +{"train_loss": 0.022079814225435257, "grad_norm": 0.10202135890722275, "global_step": 4398, "epoch": 41, "lr": 9.830981656408574e-05} +{"train_loss": 0.01951167732477188, "grad_norm": 0.04747461527585983, "global_step": 4399, "epoch": 41, "lr": 9.830905239827593e-05} +{"train_loss": 0.018717655912041664, "grad_norm": 0.0447450615465641, "global_step": 4400, "epoch": 41, "lr": 9.830828806272909e-05} +{"train_loss": 0.018174495548009872, "grad_norm": 0.10408960282802582, "global_step": 4401, "epoch": 41, "lr": 9.830752355744794e-05} +{"train_loss": 0.019125565886497498, "grad_norm": 0.04422266036272049, "global_step": 4402, "epoch": 41, "lr": 9.830675888243518e-05} +{"train_loss": 0.018651796504855156, "grad_norm": 0.04348083212971687, "global_step": 4403, "epoch": 41, "lr": 9.830599403769347e-05} +{"train_loss": 0.019362475723028183, "grad_norm": 0.095340795814991, "global_step": 4404, "epoch": 41, "lr": 9.83052290232255e-05} +{"train_loss": 0.019784601405262947, "grad_norm": 0.05163508281111717, "global_step": 4405, "epoch": 41, "lr": 9.830446383903395e-05} +{"train_loss": 0.01788315922021866, "grad_norm": 0.043647438287734985, "global_step": 4406, "epoch": 41, "lr": 9.830369848512156e-05} +{"train_loss": 0.020420534536242485, "grad_norm": 0.047577034682035446, "global_step": 4407, "epoch": 41, "lr": 9.830293296149095e-05} +{"train_loss": 0.020916365087032318, "grad_norm": 0.10201749950647354, "global_step": 4408, "epoch": 41, "lr": 9.830216726814485e-05} +{"train_loss": 0.019336650148034096, "grad_norm": 0.09311825782060623, "global_step": 4409, "epoch": 41, "lr": 9.830140140508592e-05} +{"train_loss": 0.0186366718262434, "grad_norm": 0.064677894115448, "global_step": 4410, "epoch": 41, "lr": 9.83006353723169e-05} +{"train_loss": 0.02155187539756298, "grad_norm": 0.13440664112567902, "global_step": 4411, "epoch": 41, "lr": 9.829986916984043e-05} +{"train_loss": 0.019187089055776596, "grad_norm": 0.04870965704321861, "global_step": 4412, "epoch": 41, "lr": 9.829910279765924e-05} +{"train_loss": 0.018315766006708145, "grad_norm": 0.106076180934906, "global_step": 4413, "epoch": 41, "lr": 9.8298336255776e-05} +{"train_loss": 0.021345414221286774, "grad_norm": 0.10483164340257645, "global_step": 4414, "epoch": 41, "lr": 9.829756954419341e-05} +{"train_loss": 0.021686404943466187, "grad_norm": 0.06951645016670227, "global_step": 4415, "epoch": 41, "lr": 9.829680266291417e-05} +{"train_loss": 0.02176651731133461, "grad_norm": 0.11933496594429016, "global_step": 4416, "epoch": 41, "lr": 9.829603561194096e-05} +{"train_loss": 0.018902156502008438, "grad_norm": 0.09802145510911942, "global_step": 4417, "epoch": 41, "lr": 9.82952683912765e-05} +{"train_loss": 0.016848638653755188, "grad_norm": 0.0384889654815197, "global_step": 4418, "epoch": 41, "lr": 9.829450100092345e-05} +{"train_loss": 0.021559035405516624, "grad_norm": 0.11259764432907104, "global_step": 4419, "epoch": 41, "lr": 9.829373344088455e-05} +{"train_loss": 0.0192544087767601, "grad_norm": 0.06668972969055176, "global_step": 4420, "epoch": 41, "lr": 9.829296571116246e-05} +{"train_loss": 0.019975747913122177, "grad_norm": 0.06156362220644951, "global_step": 4421, "epoch": 41, "lr": 9.829219781175989e-05} +{"train_loss": 0.018880873918533325, "grad_norm": 0.1325346976518631, "global_step": 4422, "epoch": 41, "lr": 9.829142974267954e-05} +{"train_loss": 0.01967022940516472, "grad_norm": 0.13716623187065125, "global_step": 4423, "epoch": 41, "lr": 9.82906615039241e-05} +{"train_loss": 0.020917896181344986, "grad_norm": 0.08457092195749283, "global_step": 4424, "epoch": 41, "lr": 9.828989309549627e-05} +{"train_loss": 0.015220660716295242, "grad_norm": 0.04067320004105568, "global_step": 4425, "epoch": 41, "lr": 9.828912451739877e-05} +{"train_loss": 0.02012273296713829, "grad_norm": 0.0537608303129673, "global_step": 4426, "epoch": 41, "lr": 9.82883557696343e-05} +{"train_loss": 0.018297182396054268, "grad_norm": 0.05949754640460014, "global_step": 4427, "epoch": 41, "lr": 9.828758685220553e-05} +{"train_loss": 0.018880721181631088, "grad_norm": 0.07399123907089233, "global_step": 4428, "epoch": 41, "lr": 9.828681776511519e-05} +{"train_loss": 0.0200762078166008, "grad_norm": 0.0745575875043869, "global_step": 4429, "epoch": 41, "lr": 9.828604850836596e-05} +{"train_loss": 0.021665826439857483, "grad_norm": 0.04376609995961189, "global_step": 4430, "epoch": 41, "lr": 9.828527908196057e-05} +{"train_loss": 0.01981033757328987, "grad_norm": 0.08575460314750671, "global_step": 4431, "epoch": 41, "lr": 9.828450948590169e-05} +{"train_loss": 0.017070673406124115, "grad_norm": 0.05469416081905365, "global_step": 4432, "epoch": 41, "lr": 9.828373972019205e-05} +{"train_loss": 0.018831446766853333, "grad_norm": 0.0784384161233902, "global_step": 4433, "epoch": 41, "lr": 9.828296978483435e-05} +{"train_loss": 0.019662296399474144, "grad_norm": 0.0947459414601326, "global_step": 4434, "epoch": 41, "lr": 9.828219967983128e-05} +{"train_loss": 0.019329778850078583, "grad_norm": 0.09356376528739929, "global_step": 4435, "epoch": 41, "lr": 9.828142940518558e-05} +{"train_loss": 0.0217870082706213, "grad_norm": 0.09998553991317749, "global_step": 4436, "epoch": 41, "lr": 9.828065896089991e-05} +{"train_loss": 0.020031580701470375, "grad_norm": 0.05857372656464577, "global_step": 4437, "epoch": 41, "lr": 9.8279888346977e-05} +{"train_loss": 0.021063541993498802, "grad_norm": 0.10715451091527939, "global_step": 4438, "epoch": 41, "lr": 9.827911756341958e-05} +{"train_loss": 0.01633099652826786, "grad_norm": 0.04568713158369064, "global_step": 4439, "epoch": 41, "lr": 9.827834661023034e-05} +{"train_loss": 0.019972344860434532, "grad_norm": 0.05357092618942261, "global_step": 4440, "epoch": 41, "lr": 9.827757548741196e-05} +{"train_loss": 0.01849938929080963, "grad_norm": 0.061768196523189545, "global_step": 4441, "epoch": 41, "lr": 9.82768041949672e-05} +{"train_loss": 0.018857145681977272, "grad_norm": 0.04059197008609772, "global_step": 4442, "epoch": 41, "lr": 9.827603273289874e-05} +{"train_loss": 0.01977081596851349, "grad_norm": 0.08318982273340225, "global_step": 4443, "epoch": 41, "lr": 9.827526110120929e-05} +{"train_loss": 0.018037477508187294, "grad_norm": 0.04669158533215523, "global_step": 4444, "epoch": 41, "lr": 9.827448929990155e-05} +{"train_loss": 0.019705018028616905, "grad_norm": 0.15506744384765625, "global_step": 4445, "epoch": 41, "lr": 9.827371732897827e-05} +{"train_loss": 0.01988472416996956, "grad_norm": 0.10273468494415283, "global_step": 4446, "epoch": 41, "lr": 9.827294518844213e-05} +{"train_loss": 0.021569062024354935, "grad_norm": 0.05306382104754448, "global_step": 4447, "epoch": 41, "lr": 9.827217287829586e-05} +{"train_loss": 0.019205793738365173, "grad_norm": 0.09287821501493454, "global_step": 4448, "epoch": 41, "lr": 9.827140039854217e-05} +{"train_loss": 0.018615249544382095, "grad_norm": 0.07913381606340408, "global_step": 4449, "epoch": 41, "lr": 9.827062774918376e-05} +{"train_loss": 0.020515572279691696, "grad_norm": 0.040355049073696136, "global_step": 4450, "epoch": 41, "lr": 9.826985493022337e-05} +{"train_loss": 0.019697148651587514, "grad_norm": 0.10330253839492798, "global_step": 4451, "epoch": 41, "lr": 9.82690819416637e-05, "val_loss": 0.01901359297335148} +{"train_loss": 0.022163115441799164, "grad_norm": 0.05633356794714928, "global_step": 4452, "epoch": 42, "lr": 9.826830878350746e-05} +{"train_loss": 0.02212674170732498, "grad_norm": 0.06853441894054413, "global_step": 4453, "epoch": 42, "lr": 9.826753545575737e-05} +{"train_loss": 0.020679153501987457, "grad_norm": 0.09462320804595947, "global_step": 4454, "epoch": 42, "lr": 9.826676195841617e-05} +{"train_loss": 0.01796356774866581, "grad_norm": 0.035298388451337814, "global_step": 4455, "epoch": 42, "lr": 9.826598829148654e-05} +{"train_loss": 0.02317580208182335, "grad_norm": 0.08973378688097, "global_step": 4456, "epoch": 42, "lr": 9.826521445497123e-05} +{"train_loss": 0.01911090314388275, "grad_norm": 0.036785174161195755, "global_step": 4457, "epoch": 42, "lr": 9.826444044887295e-05} +{"train_loss": 0.020540393888950348, "grad_norm": 0.0480879545211792, "global_step": 4458, "epoch": 42, "lr": 9.826366627319442e-05} +{"train_loss": 0.01973038539290428, "grad_norm": 0.05700519308447838, "global_step": 4459, "epoch": 42, "lr": 9.826289192793834e-05} +{"train_loss": 0.01794271171092987, "grad_norm": 0.08420246094465256, "global_step": 4460, "epoch": 42, "lr": 9.826211741310745e-05} +{"train_loss": 0.020345233380794525, "grad_norm": 0.08853745460510254, "global_step": 4461, "epoch": 42, "lr": 9.826134272870447e-05} +{"train_loss": 0.019429925829172134, "grad_norm": 0.06201391667127609, "global_step": 4462, "epoch": 42, "lr": 9.826056787473213e-05} +{"train_loss": 0.019250212237238884, "grad_norm": 0.0946061834692955, "global_step": 4463, "epoch": 42, "lr": 9.825979285119311e-05} +{"train_loss": 0.02296888269484043, "grad_norm": 0.12306997925043106, "global_step": 4464, "epoch": 42, "lr": 9.82590176580902e-05} +{"train_loss": 0.0164361409842968, "grad_norm": 0.043008849024772644, "global_step": 4465, "epoch": 42, "lr": 9.825824229542607e-05} +{"train_loss": 0.01601482927799225, "grad_norm": 0.06729044020175934, "global_step": 4466, "epoch": 42, "lr": 9.825746676320348e-05} +{"train_loss": 0.021837852895259857, "grad_norm": 0.09883757680654526, "global_step": 4467, "epoch": 42, "lr": 9.825669106142513e-05} +{"train_loss": 0.016925156116485596, "grad_norm": 0.04214784875512123, "global_step": 4468, "epoch": 42, "lr": 9.825591519009374e-05} +{"train_loss": 0.01849253475666046, "grad_norm": 0.03413984924554825, "global_step": 4469, "epoch": 42, "lr": 9.825513914921208e-05} +{"train_loss": 0.01852819323539734, "grad_norm": 0.08331094682216644, "global_step": 4470, "epoch": 42, "lr": 9.825436293878281e-05} +{"train_loss": 0.021772531792521477, "grad_norm": 0.1248810887336731, "global_step": 4471, "epoch": 42, "lr": 9.825358655880872e-05} +{"train_loss": 0.022300563752651215, "grad_norm": 0.09001593291759491, "global_step": 4472, "epoch": 42, "lr": 9.82528100092925e-05} +{"train_loss": 0.020176012068986893, "grad_norm": 0.11433175951242447, "global_step": 4473, "epoch": 42, "lr": 9.82520332902369e-05} +{"train_loss": 0.020090624690055847, "grad_norm": 0.09380137175321579, "global_step": 4474, "epoch": 42, "lr": 9.825125640164463e-05} +{"train_loss": 0.016996722668409348, "grad_norm": 0.055852025747299194, "global_step": 4475, "epoch": 42, "lr": 9.825047934351843e-05} +{"train_loss": 0.018763892352581024, "grad_norm": 0.07375456392765045, "global_step": 4476, "epoch": 42, "lr": 9.824970211586104e-05} +{"train_loss": 0.020945768803358078, "grad_norm": 0.07336581498384476, "global_step": 4477, "epoch": 42, "lr": 9.824892471867517e-05} +{"train_loss": 0.02115444466471672, "grad_norm": 0.12298355996608734, "global_step": 4478, "epoch": 42, "lr": 9.824814715196356e-05} +{"train_loss": 0.01718936115503311, "grad_norm": 0.04642948508262634, "global_step": 4479, "epoch": 42, "lr": 9.824736941572896e-05} +{"train_loss": 0.01615184172987938, "grad_norm": 0.039278920739889145, "global_step": 4480, "epoch": 42, "lr": 9.824659150997407e-05} +{"train_loss": 0.022248676046729088, "grad_norm": 0.09808012843132019, "global_step": 4481, "epoch": 42, "lr": 9.824581343470165e-05} +{"train_loss": 0.01584002748131752, "grad_norm": 0.05421731621026993, "global_step": 4482, "epoch": 42, "lr": 9.824503518991441e-05} +{"train_loss": 0.020558223128318787, "grad_norm": 0.07474420964717865, "global_step": 4483, "epoch": 42, "lr": 9.824425677561511e-05} +{"train_loss": 0.020518040284514427, "grad_norm": 0.0769554153084755, "global_step": 4484, "epoch": 42, "lr": 9.824347819180647e-05} +{"train_loss": 0.02124417945742607, "grad_norm": 0.06443830579519272, "global_step": 4485, "epoch": 42, "lr": 9.824269943849121e-05} +{"train_loss": 0.017296940088272095, "grad_norm": 0.05053306370973587, "global_step": 4486, "epoch": 42, "lr": 9.82419205156721e-05} +{"train_loss": 0.017747249454259872, "grad_norm": 0.0477764867246151, "global_step": 4487, "epoch": 42, "lr": 9.824114142335185e-05} +{"train_loss": 0.019675305113196373, "grad_norm": 0.033421386033296585, "global_step": 4488, "epoch": 42, "lr": 9.824036216153323e-05} +{"train_loss": 0.01942029967904091, "grad_norm": 0.06418760120868683, "global_step": 4489, "epoch": 42, "lr": 9.823958273021894e-05} +{"train_loss": 0.016938120126724243, "grad_norm": 0.03919462114572525, "global_step": 4490, "epoch": 42, "lr": 9.823880312941173e-05} +{"train_loss": 0.020546160638332367, "grad_norm": 0.07887133955955505, "global_step": 4491, "epoch": 42, "lr": 9.823802335911437e-05} +{"train_loss": 0.017970886081457138, "grad_norm": 0.051688119769096375, "global_step": 4492, "epoch": 42, "lr": 9.823724341932956e-05} +{"train_loss": 0.023191355168819427, "grad_norm": 0.05341508239507675, "global_step": 4493, "epoch": 42, "lr": 9.823646331006005e-05} +{"train_loss": 0.020604053512215614, "grad_norm": 0.08732138574123383, "global_step": 4494, "epoch": 42, "lr": 9.823568303130859e-05} +{"train_loss": 0.025859761983156204, "grad_norm": 0.14813975989818573, "global_step": 4495, "epoch": 42, "lr": 9.823490258307792e-05} +{"train_loss": 0.018602602183818817, "grad_norm": 0.06659013032913208, "global_step": 4496, "epoch": 42, "lr": 9.823412196537077e-05} +{"train_loss": 0.023895425722002983, "grad_norm": 0.07050064206123352, "global_step": 4497, "epoch": 42, "lr": 9.82333411781899e-05} +{"train_loss": 0.019038015976548195, "grad_norm": 0.10375938564538956, "global_step": 4498, "epoch": 42, "lr": 9.823256022153804e-05} +{"train_loss": 0.020635757595300674, "grad_norm": 0.05893674120306969, "global_step": 4499, "epoch": 42, "lr": 9.823177909541794e-05} +{"train_loss": 0.01766440086066723, "grad_norm": 0.05272562429308891, "global_step": 4500, "epoch": 42, "lr": 9.823099779983234e-05} +{"train_loss": 0.020759450271725655, "grad_norm": 0.15598730742931366, "global_step": 4501, "epoch": 42, "lr": 9.8230216334784e-05} +{"train_loss": 0.016348108649253845, "grad_norm": 0.05447111278772354, "global_step": 4502, "epoch": 42, "lr": 9.822943470027565e-05} +{"train_loss": 0.021196424961090088, "grad_norm": 0.05827372148633003, "global_step": 4503, "epoch": 42, "lr": 9.822865289631002e-05} +{"train_loss": 0.019976921379566193, "grad_norm": 0.052357904613018036, "global_step": 4504, "epoch": 42, "lr": 9.822787092288991e-05} +{"train_loss": 0.018464695662260056, "grad_norm": 0.05548390746116638, "global_step": 4505, "epoch": 42, "lr": 9.822708878001803e-05} +{"train_loss": 0.019980231299996376, "grad_norm": 0.05440312996506691, "global_step": 4506, "epoch": 42, "lr": 9.822630646769712e-05} +{"train_loss": 0.019621513783931732, "grad_norm": 0.039747513830661774, "global_step": 4507, "epoch": 42, "lr": 9.822552398592995e-05} +{"train_loss": 0.019158072769641876, "grad_norm": 0.042342957109212875, "global_step": 4508, "epoch": 42, "lr": 9.822474133471927e-05} +{"train_loss": 0.019512390717864037, "grad_norm": 0.13384799659252167, "global_step": 4509, "epoch": 42, "lr": 9.82239585140678e-05} +{"train_loss": 0.019690949469804764, "grad_norm": 0.044883303344249725, "global_step": 4510, "epoch": 42, "lr": 9.822317552397831e-05} +{"train_loss": 0.022357556968927383, "grad_norm": 0.07044393569231033, "global_step": 4511, "epoch": 42, "lr": 9.822239236445356e-05} +{"train_loss": 0.01832318678498268, "grad_norm": 0.0455290712416172, "global_step": 4512, "epoch": 42, "lr": 9.82216090354963e-05} +{"train_loss": 0.021608203649520874, "grad_norm": 0.05692732706665993, "global_step": 4513, "epoch": 42, "lr": 9.822082553710929e-05} +{"train_loss": 0.020404402166604996, "grad_norm": 0.06955531239509583, "global_step": 4514, "epoch": 42, "lr": 9.822004186929525e-05} +{"train_loss": 0.018616940826177597, "grad_norm": 0.054434552788734436, "global_step": 4515, "epoch": 42, "lr": 9.821925803205696e-05} +{"train_loss": 0.016619157046079636, "grad_norm": 0.09405319392681122, "global_step": 4516, "epoch": 42, "lr": 9.821847402539717e-05} +{"train_loss": 0.017401807010173798, "grad_norm": 0.0539921373128891, "global_step": 4517, "epoch": 42, "lr": 9.821768984931863e-05} +{"train_loss": 0.021649131551384926, "grad_norm": 0.05108465999364853, "global_step": 4518, "epoch": 42, "lr": 9.82169055038241e-05} +{"train_loss": 0.017230458557605743, "grad_norm": 0.04607509449124336, "global_step": 4519, "epoch": 42, "lr": 9.821612098891632e-05} +{"train_loss": 0.02049509808421135, "grad_norm": 0.04435870796442032, "global_step": 4520, "epoch": 42, "lr": 9.821533630459807e-05} +{"train_loss": 0.01897994987666607, "grad_norm": 0.07295297831296921, "global_step": 4521, "epoch": 42, "lr": 9.82145514508721e-05} +{"train_loss": 0.0204514991492033, "grad_norm": 0.11810766160488129, "global_step": 4522, "epoch": 42, "lr": 9.821376642774115e-05} +{"train_loss": 0.020433293655514717, "grad_norm": 0.06080982834100723, "global_step": 4523, "epoch": 42, "lr": 9.821298123520801e-05} +{"train_loss": 0.01875920407474041, "grad_norm": 0.051417797803878784, "global_step": 4524, "epoch": 42, "lr": 9.821219587327542e-05} +{"train_loss": 0.017406223341822624, "grad_norm": 0.10854791849851608, "global_step": 4525, "epoch": 42, "lr": 9.821141034194612e-05} +{"train_loss": 0.019395766779780388, "grad_norm": 0.052283450961112976, "global_step": 4526, "epoch": 42, "lr": 9.82106246412229e-05} +{"train_loss": 0.020886674523353577, "grad_norm": 0.10358784347772598, "global_step": 4527, "epoch": 42, "lr": 9.820983877110853e-05} +{"train_loss": 0.017527323216199875, "grad_norm": 0.03895486891269684, "global_step": 4528, "epoch": 42, "lr": 9.820905273160572e-05} +{"train_loss": 0.019014466553926468, "grad_norm": 0.05693056434392929, "global_step": 4529, "epoch": 42, "lr": 9.820826652271729e-05} +{"train_loss": 0.019104106351733208, "grad_norm": 0.06268633902072906, "global_step": 4530, "epoch": 42, "lr": 9.820748014444596e-05} +{"train_loss": 0.01961587183177471, "grad_norm": 0.10173020511865616, "global_step": 4531, "epoch": 42, "lr": 9.820669359679453e-05} +{"train_loss": 0.01878441870212555, "grad_norm": 0.04127362370491028, "global_step": 4532, "epoch": 42, "lr": 9.820590687976573e-05} +{"train_loss": 0.01803186535835266, "grad_norm": 0.049946099519729614, "global_step": 4533, "epoch": 42, "lr": 9.820511999336233e-05} +{"train_loss": 0.015377998352050781, "grad_norm": 0.04157622531056404, "global_step": 4534, "epoch": 42, "lr": 9.820433293758712e-05} +{"train_loss": 0.018517956137657166, "grad_norm": 0.05112335830926895, "global_step": 4535, "epoch": 42, "lr": 9.820354571244284e-05} +{"train_loss": 0.02041623182594776, "grad_norm": 0.0785183236002922, "global_step": 4536, "epoch": 42, "lr": 9.820275831793226e-05} +{"train_loss": 0.019959934055805206, "grad_norm": 0.058745212852954865, "global_step": 4537, "epoch": 42, "lr": 9.820197075405814e-05} +{"train_loss": 0.021544218063354492, "grad_norm": 0.053855787962675095, "global_step": 4538, "epoch": 42, "lr": 9.820118302082329e-05} +{"train_loss": 0.020619789138436317, "grad_norm": 0.09745308756828308, "global_step": 4539, "epoch": 42, "lr": 9.820039511823041e-05} +{"train_loss": 0.018942348659038544, "grad_norm": 0.0479779876768589, "global_step": 4540, "epoch": 42, "lr": 9.819960704628233e-05} +{"train_loss": 0.019666597247123718, "grad_norm": 0.05308493971824646, "global_step": 4541, "epoch": 42, "lr": 9.819881880498177e-05} +{"train_loss": 0.01665802299976349, "grad_norm": 0.13377059996128082, "global_step": 4542, "epoch": 42, "lr": 9.819803039433153e-05} +{"train_loss": 0.020680628716945648, "grad_norm": 0.09738174080848694, "global_step": 4543, "epoch": 42, "lr": 9.819724181433438e-05} +{"train_loss": 0.020440077409148216, "grad_norm": 0.046877432614564896, "global_step": 4544, "epoch": 42, "lr": 9.819645306499308e-05} +{"train_loss": 0.020946016535162926, "grad_norm": 0.08504630625247955, "global_step": 4545, "epoch": 42, "lr": 9.81956641463104e-05} +{"train_loss": 0.019129551947116852, "grad_norm": 0.049269914627075195, "global_step": 4546, "epoch": 42, "lr": 9.819487505828911e-05} +{"train_loss": 0.021931126713752747, "grad_norm": 0.05293640121817589, "global_step": 4547, "epoch": 42, "lr": 9.8194085800932e-05} +{"train_loss": 0.02181769348680973, "grad_norm": 0.04772025719285011, "global_step": 4548, "epoch": 42, "lr": 9.819329637424181e-05} +{"train_loss": 0.021892787888646126, "grad_norm": 0.06860458105802536, "global_step": 4549, "epoch": 42, "lr": 9.819250677822135e-05} +{"train_loss": 0.019956964999437332, "grad_norm": 0.0887700766324997, "global_step": 4550, "epoch": 42, "lr": 9.81917170128734e-05} +{"train_loss": 0.018628565594553947, "grad_norm": 0.1016840785741806, "global_step": 4551, "epoch": 42, "lr": 9.81909270782007e-05} +{"train_loss": 0.018795574083924294, "grad_norm": 0.06131863594055176, "global_step": 4552, "epoch": 42, "lr": 9.819013697420603e-05} +{"train_loss": 0.019879113882780075, "grad_norm": 0.10881324112415314, "global_step": 4553, "epoch": 42, "lr": 9.818934670089218e-05} +{"train_loss": 0.017933689057826996, "grad_norm": 0.07907639443874359, "global_step": 4554, "epoch": 42, "lr": 9.818855625826193e-05} +{"train_loss": 0.021092785522341728, "grad_norm": 0.11243464797735214, "global_step": 4555, "epoch": 42, "lr": 9.818776564631805e-05} +{"train_loss": 0.021603211760520935, "grad_norm": 0.055039696395397186, "global_step": 4556, "epoch": 42, "lr": 9.818697486506329e-05} +{"train_loss": 0.019670984321186004, "grad_norm": 0.1498079150915146, "global_step": 4557, "epoch": 42, "lr": 9.81861839145005e-05, "val_loss": 0.019327519461512566} +{"train_loss": 0.022237088531255722, "grad_norm": 0.04251689836382866, "global_step": 4558, "epoch": 43, "lr": 9.818539279463239e-05} +{"train_loss": 0.01841002143919468, "grad_norm": 0.07366374880075455, "global_step": 4559, "epoch": 43, "lr": 9.818460150546177e-05} +{"train_loss": 0.020550718531012535, "grad_norm": 0.03260994702577591, "global_step": 4560, "epoch": 43, "lr": 9.81838100469914e-05} +{"train_loss": 0.017838679254055023, "grad_norm": 0.03265458345413208, "global_step": 4561, "epoch": 43, "lr": 9.81830184192241e-05} +{"train_loss": 0.018777946010231972, "grad_norm": 0.11417786031961441, "global_step": 4562, "epoch": 43, "lr": 9.818222662216261e-05} +{"train_loss": 0.019810982048511505, "grad_norm": 0.06683948636054993, "global_step": 4563, "epoch": 43, "lr": 9.818143465580973e-05} +{"train_loss": 0.020259536802768707, "grad_norm": 0.06383491307497025, "global_step": 4564, "epoch": 43, "lr": 9.818064252016825e-05} +{"train_loss": 0.019848590716719627, "grad_norm": 0.04001747444272041, "global_step": 4565, "epoch": 43, "lr": 9.817985021524095e-05} +{"train_loss": 0.02167685329914093, "grad_norm": 0.05878562852740288, "global_step": 4566, "epoch": 43, "lr": 9.817905774103061e-05} +{"train_loss": 0.018769845366477966, "grad_norm": 0.10154101252555847, "global_step": 4567, "epoch": 43, "lr": 9.817826509754e-05} +{"train_loss": 0.019113648682832718, "grad_norm": 0.0949370414018631, "global_step": 4568, "epoch": 43, "lr": 9.817747228477194e-05} +{"train_loss": 0.022588122636079788, "grad_norm": 0.08385913074016571, "global_step": 4569, "epoch": 43, "lr": 9.817667930272918e-05} +{"train_loss": 0.01826060190796852, "grad_norm": 0.04137592390179634, "global_step": 4570, "epoch": 43, "lr": 9.817588615141451e-05} +{"train_loss": 0.0234760083258152, "grad_norm": 0.16860076785087585, "global_step": 4571, "epoch": 43, "lr": 9.817509283083075e-05} +{"train_loss": 0.01745527610182762, "grad_norm": 0.03866344690322876, "global_step": 4572, "epoch": 43, "lr": 9.817429934098064e-05} +{"train_loss": 0.01710221916437149, "grad_norm": 0.038404665887355804, "global_step": 4573, "epoch": 43, "lr": 9.817350568186701e-05} +{"train_loss": 0.01781516522169113, "grad_norm": 0.047380708158016205, "global_step": 4574, "epoch": 43, "lr": 9.817271185349262e-05} +{"train_loss": 0.019164839759469032, "grad_norm": 0.06309859454631805, "global_step": 4575, "epoch": 43, "lr": 9.817191785586028e-05} +{"train_loss": 0.01848778873682022, "grad_norm": 0.052118800580501556, "global_step": 4576, "epoch": 43, "lr": 9.817112368897277e-05} +{"train_loss": 0.021218813955783844, "grad_norm": 0.09313719719648361, "global_step": 4577, "epoch": 43, "lr": 9.817032935283288e-05} +{"train_loss": 0.02429318055510521, "grad_norm": 0.0868186503648758, "global_step": 4578, "epoch": 43, "lr": 9.81695348474434e-05} +{"train_loss": 0.02047542855143547, "grad_norm": 0.04829048365354538, "global_step": 4579, "epoch": 43, "lr": 9.816874017280712e-05} +{"train_loss": 0.016429632902145386, "grad_norm": 0.04985743388533592, "global_step": 4580, "epoch": 43, "lr": 9.816794532892683e-05} +{"train_loss": 0.020790502429008484, "grad_norm": 0.08726400882005692, "global_step": 4581, "epoch": 43, "lr": 9.816715031580535e-05} +{"train_loss": 0.021708719432353973, "grad_norm": 0.1149882972240448, "global_step": 4582, "epoch": 43, "lr": 9.816635513344543e-05} +{"train_loss": 0.01892523467540741, "grad_norm": 0.04619196057319641, "global_step": 4583, "epoch": 43, "lr": 9.81655597818499e-05} +{"train_loss": 0.019472790881991386, "grad_norm": 0.05332496389746666, "global_step": 4584, "epoch": 43, "lr": 9.816476426102151e-05} +{"train_loss": 0.01925526186823845, "grad_norm": 0.07360667735338211, "global_step": 4585, "epoch": 43, "lr": 9.81639685709631e-05} +{"train_loss": 0.017398662865161896, "grad_norm": 0.032003775238990784, "global_step": 4586, "epoch": 43, "lr": 9.816317271167747e-05} +{"train_loss": 0.017376534640789032, "grad_norm": 0.0877477377653122, "global_step": 4587, "epoch": 43, "lr": 9.816237668316738e-05} +{"train_loss": 0.015202552080154419, "grad_norm": 0.035392459481954575, "global_step": 4588, "epoch": 43, "lr": 9.816158048543565e-05} +{"train_loss": 0.019617419689893723, "grad_norm": 0.04604785889387131, "global_step": 4589, "epoch": 43, "lr": 9.816078411848506e-05} +{"train_loss": 0.01564217358827591, "grad_norm": 0.08107483386993408, "global_step": 4590, "epoch": 43, "lr": 9.815998758231842e-05} +{"train_loss": 0.019611850380897522, "grad_norm": 0.10875263810157776, "global_step": 4591, "epoch": 43, "lr": 9.815919087693854e-05} +{"train_loss": 0.01964190974831581, "grad_norm": 0.05487575754523277, "global_step": 4592, "epoch": 43, "lr": 9.81583940023482e-05} +{"train_loss": 0.018673807382583618, "grad_norm": 0.0716136246919632, "global_step": 4593, "epoch": 43, "lr": 9.815759695855021e-05} +{"train_loss": 0.020011870190501213, "grad_norm": 0.06793282926082611, "global_step": 4594, "epoch": 43, "lr": 9.815679974554737e-05} +{"train_loss": 0.018526986241340637, "grad_norm": 0.08227896690368652, "global_step": 4595, "epoch": 43, "lr": 9.815600236334246e-05} +{"train_loss": 0.021917592734098434, "grad_norm": 0.06708875298500061, "global_step": 4596, "epoch": 43, "lr": 9.815520481193832e-05} +{"train_loss": 0.018469063565135002, "grad_norm": 0.0448019914329052, "global_step": 4597, "epoch": 43, "lr": 9.815440709133771e-05} +{"train_loss": 0.020305756479501724, "grad_norm": 0.0705999955534935, "global_step": 4598, "epoch": 43, "lr": 9.815360920154348e-05} +{"train_loss": 0.020669762045145035, "grad_norm": 0.04974432289600372, "global_step": 4599, "epoch": 43, "lr": 9.815281114255841e-05} +{"train_loss": 0.01841907575726509, "grad_norm": 0.04769175499677658, "global_step": 4600, "epoch": 43, "lr": 9.815201291438529e-05} +{"train_loss": 0.020876754075288773, "grad_norm": 0.11502202600240707, "global_step": 4601, "epoch": 43, "lr": 9.815121451702693e-05} +{"train_loss": 0.01873183622956276, "grad_norm": 0.060539208352565765, "global_step": 4602, "epoch": 43, "lr": 9.815041595048616e-05} +{"train_loss": 0.017992591485381126, "grad_norm": 0.058642223477363586, "global_step": 4603, "epoch": 43, "lr": 9.814961721476576e-05} +{"train_loss": 0.021627867594361305, "grad_norm": 0.04975905269384384, "global_step": 4604, "epoch": 43, "lr": 9.814881830986855e-05} +{"train_loss": 0.01570531725883484, "grad_norm": 0.0413997620344162, "global_step": 4605, "epoch": 43, "lr": 9.814801923579733e-05} +{"train_loss": 0.017551403492689133, "grad_norm": 0.038413409143686295, "global_step": 4606, "epoch": 43, "lr": 9.81472199925549e-05} +{"train_loss": 0.01840405911207199, "grad_norm": 0.06824890524148941, "global_step": 4607, "epoch": 43, "lr": 9.81464205801441e-05} +{"train_loss": 0.018124762922525406, "grad_norm": 0.09678426384925842, "global_step": 4608, "epoch": 43, "lr": 9.81456209985677e-05} +{"train_loss": 0.01793140359222889, "grad_norm": 0.06963249295949936, "global_step": 4609, "epoch": 43, "lr": 9.814482124782852e-05} +{"train_loss": 0.01741148717701435, "grad_norm": 0.06283587217330933, "global_step": 4610, "epoch": 43, "lr": 9.814402132792939e-05} +{"train_loss": 0.02087198570370674, "grad_norm": 0.0892796739935875, "global_step": 4611, "epoch": 43, "lr": 9.81432212388731e-05} +{"train_loss": 0.021025400608778, "grad_norm": 0.09551283717155457, "global_step": 4612, "epoch": 43, "lr": 9.814242098066246e-05} +{"train_loss": 0.021468691527843475, "grad_norm": 0.04355707764625549, "global_step": 4613, "epoch": 43, "lr": 9.81416205533003e-05} +{"train_loss": 0.017857328057289124, "grad_norm": 0.03778231143951416, "global_step": 4614, "epoch": 43, "lr": 9.814081995678942e-05} +{"train_loss": 0.01956849731504917, "grad_norm": 0.04648078978061676, "global_step": 4615, "epoch": 43, "lr": 9.814001919113263e-05} +{"train_loss": 0.021459702402353287, "grad_norm": 0.08377379924058914, "global_step": 4616, "epoch": 43, "lr": 9.813921825633274e-05} +{"train_loss": 0.01725720427930355, "grad_norm": 0.054470714181661606, "global_step": 4617, "epoch": 43, "lr": 9.813841715239259e-05} +{"train_loss": 0.019547225907444954, "grad_norm": 0.06765251606702805, "global_step": 4618, "epoch": 43, "lr": 9.813761587931497e-05} +{"train_loss": 0.018859142437577248, "grad_norm": 0.040605902671813965, "global_step": 4619, "epoch": 43, "lr": 9.81368144371027e-05} +{"train_loss": 0.021140191704034805, "grad_norm": 0.04282218590378761, "global_step": 4620, "epoch": 43, "lr": 9.81360128257586e-05} +{"train_loss": 0.019664738327264786, "grad_norm": 0.04040404036641121, "global_step": 4621, "epoch": 43, "lr": 9.813521104528549e-05} +{"train_loss": 0.01910088211297989, "grad_norm": 0.05723624303936958, "global_step": 4622, "epoch": 43, "lr": 9.813440909568618e-05} +{"train_loss": 0.017748437821865082, "grad_norm": 0.03593812137842178, "global_step": 4623, "epoch": 43, "lr": 9.813360697696347e-05} +{"train_loss": 0.021174879744648933, "grad_norm": 0.15324899554252625, "global_step": 4624, "epoch": 43, "lr": 9.813280468912023e-05} +{"train_loss": 0.01971139758825302, "grad_norm": 0.09153912961483002, "global_step": 4625, "epoch": 43, "lr": 9.813200223215923e-05} +{"train_loss": 0.018099043518304825, "grad_norm": 0.04978175088763237, "global_step": 4626, "epoch": 43, "lr": 9.81311996060833e-05} +{"train_loss": 0.021559104323387146, "grad_norm": 0.045645490288734436, "global_step": 4627, "epoch": 43, "lr": 9.813039681089527e-05} +{"train_loss": 0.018613699823617935, "grad_norm": 0.10307148098945618, "global_step": 4628, "epoch": 43, "lr": 9.812959384659796e-05} +{"train_loss": 0.020048508420586586, "grad_norm": 0.12471361458301544, "global_step": 4629, "epoch": 43, "lr": 9.81287907131942e-05} +{"train_loss": 0.020118722692131996, "grad_norm": 0.0601736418902874, "global_step": 4630, "epoch": 43, "lr": 9.812798741068679e-05} +{"train_loss": 0.01887880079448223, "grad_norm": 0.03995141014456749, "global_step": 4631, "epoch": 43, "lr": 9.812718393907856e-05} +{"train_loss": 0.020318076014518738, "grad_norm": 0.1434377133846283, "global_step": 4632, "epoch": 43, "lr": 9.812638029837233e-05} +{"train_loss": 0.017025036737322807, "grad_norm": 0.049387700855731964, "global_step": 4633, "epoch": 43, "lr": 9.812557648857094e-05} +{"train_loss": 0.022830406203866005, "grad_norm": 0.05515366047620773, "global_step": 4634, "epoch": 43, "lr": 9.81247725096772e-05} +{"train_loss": 0.01983368769288063, "grad_norm": 0.10176631063222885, "global_step": 4635, "epoch": 43, "lr": 9.812396836169393e-05} +{"train_loss": 0.021305032074451447, "grad_norm": 0.15018953382968903, "global_step": 4636, "epoch": 43, "lr": 9.812316404462398e-05} +{"train_loss": 0.018933426588773727, "grad_norm": 0.0734957605600357, "global_step": 4637, "epoch": 43, "lr": 9.812235955847016e-05} +{"train_loss": 0.02417745813727379, "grad_norm": 0.11423468589782715, "global_step": 4638, "epoch": 43, "lr": 9.812155490323526e-05} +{"train_loss": 0.020371148362755775, "grad_norm": 0.05135974660515785, "global_step": 4639, "epoch": 43, "lr": 9.812075007892218e-05} +{"train_loss": 0.018426742404699326, "grad_norm": 0.037724513560533524, "global_step": 4640, "epoch": 43, "lr": 9.81199450855337e-05} +{"train_loss": 0.02021165005862713, "grad_norm": 0.07017511874437332, "global_step": 4641, "epoch": 43, "lr": 9.811913992307267e-05} +{"train_loss": 0.01922222040593624, "grad_norm": 0.09153155982494354, "global_step": 4642, "epoch": 43, "lr": 9.81183345915419e-05} +{"train_loss": 0.020570242777466774, "grad_norm": 0.07238303124904633, "global_step": 4643, "epoch": 43, "lr": 9.811752909094423e-05} +{"train_loss": 0.02107592485845089, "grad_norm": 0.09255103021860123, "global_step": 4644, "epoch": 43, "lr": 9.811672342128248e-05} +{"train_loss": 0.021985448896884918, "grad_norm": 0.07530185580253601, "global_step": 4645, "epoch": 43, "lr": 9.811591758255951e-05} +{"train_loss": 0.020143842324614525, "grad_norm": 0.04242775961756706, "global_step": 4646, "epoch": 43, "lr": 9.811511157477812e-05} +{"train_loss": 0.01981028914451599, "grad_norm": 0.09030862152576447, "global_step": 4647, "epoch": 43, "lr": 9.811430539794115e-05} +{"train_loss": 0.018808044493198395, "grad_norm": 0.05930457264184952, "global_step": 4648, "epoch": 43, "lr": 9.811349905205143e-05} +{"train_loss": 0.01885155960917473, "grad_norm": 0.03670233115553856, "global_step": 4649, "epoch": 43, "lr": 9.81126925371118e-05} +{"train_loss": 0.02366260066628456, "grad_norm": 0.06005100905895233, "global_step": 4650, "epoch": 43, "lr": 9.811188585312509e-05} +{"train_loss": 0.020120825618505478, "grad_norm": 0.06493053585290909, "global_step": 4651, "epoch": 43, "lr": 9.811107900009416e-05} +{"train_loss": 0.01828991062939167, "grad_norm": 0.08921175450086594, "global_step": 4652, "epoch": 43, "lr": 9.81102719780218e-05} +{"train_loss": 0.021339185535907745, "grad_norm": 0.049129776656627655, "global_step": 4653, "epoch": 43, "lr": 9.810946478691088e-05} +{"train_loss": 0.020135756582021713, "grad_norm": 0.05278702452778816, "global_step": 4654, "epoch": 43, "lr": 9.810865742676421e-05} +{"train_loss": 0.019399369135499, "grad_norm": 0.061630167067050934, "global_step": 4655, "epoch": 43, "lr": 9.810784989758464e-05} +{"train_loss": 0.017606079578399658, "grad_norm": 0.06374496221542358, "global_step": 4656, "epoch": 43, "lr": 9.8107042199375e-05} +{"train_loss": 0.021798327565193176, "grad_norm": 0.07671386748552322, "global_step": 4657, "epoch": 43, "lr": 9.810623433213815e-05} +{"train_loss": 0.020420372486114502, "grad_norm": 0.07450200617313385, "global_step": 4658, "epoch": 43, "lr": 9.810542629587692e-05} +{"train_loss": 0.019866757094860077, "grad_norm": 0.0399646982550621, "global_step": 4659, "epoch": 43, "lr": 9.810461809059412e-05} +{"train_loss": 0.01873926818370819, "grad_norm": 0.06203238293528557, "global_step": 4660, "epoch": 43, "lr": 9.810380971629262e-05} +{"train_loss": 0.019033201038837433, "grad_norm": 0.05616431310772896, "global_step": 4661, "epoch": 43, "lr": 9.810300117297526e-05} +{"train_loss": 0.020138826221227646, "grad_norm": 0.07144542038440704, "global_step": 4662, "epoch": 43, "lr": 9.810219246064486e-05} +{"train_loss": 0.019589950717142167, "grad_norm": 0.07658183574676514, "global_step": 4663, "epoch": 43, "lr": 9.81013835793043e-05, "val_loss": 0.019588222727179527} +{"train_loss": 0.019724838435649872, "grad_norm": 0.10336192697286606, "global_step": 4664, "epoch": 44, "lr": 9.810057452895638e-05} +{"train_loss": 0.018115941435098648, "grad_norm": 0.06723179668188095, "global_step": 4665, "epoch": 44, "lr": 9.809976530960396e-05} +{"train_loss": 0.01763010025024414, "grad_norm": 0.07288998365402222, "global_step": 4666, "epoch": 44, "lr": 9.80989559212499e-05} +{"train_loss": 0.019665854051709175, "grad_norm": 0.049494389444589615, "global_step": 4667, "epoch": 44, "lr": 9.809814636389699e-05} +{"train_loss": 0.01970365084707737, "grad_norm": 0.04743795841932297, "global_step": 4668, "epoch": 44, "lr": 9.809733663754813e-05} +{"train_loss": 0.02192743681371212, "grad_norm": 0.06206731125712395, "global_step": 4669, "epoch": 44, "lr": 9.809652674220616e-05} +{"train_loss": 0.017089227214455605, "grad_norm": 0.05975630506873131, "global_step": 4670, "epoch": 44, "lr": 9.80957166778739e-05} +{"train_loss": 0.0198429673910141, "grad_norm": 0.09924205392599106, "global_step": 4671, "epoch": 44, "lr": 9.80949064445542e-05} +{"train_loss": 0.022491300478577614, "grad_norm": 0.0825742557644844, "global_step": 4672, "epoch": 44, "lr": 9.809409604224992e-05} +{"train_loss": 0.01748751848936081, "grad_norm": 0.14549632370471954, "global_step": 4673, "epoch": 44, "lr": 9.809328547096391e-05} +{"train_loss": 0.01836787536740303, "grad_norm": 0.09305594116449356, "global_step": 4674, "epoch": 44, "lr": 9.8092474730699e-05} +{"train_loss": 0.020430386066436768, "grad_norm": 0.04714151844382286, "global_step": 4675, "epoch": 44, "lr": 9.809166382145804e-05} +{"train_loss": 0.019142042845487595, "grad_norm": 0.061490558087825775, "global_step": 4676, "epoch": 44, "lr": 9.80908527432439e-05} +{"train_loss": 0.020397141575813293, "grad_norm": 0.054448094218969345, "global_step": 4677, "epoch": 44, "lr": 9.809004149605943e-05} +{"train_loss": 0.018749844282865524, "grad_norm": 0.03942853957414627, "global_step": 4678, "epoch": 44, "lr": 9.808923007990744e-05} +{"train_loss": 0.01746424287557602, "grad_norm": 0.11558035761117935, "global_step": 4679, "epoch": 44, "lr": 9.808841849479084e-05} +{"train_loss": 0.02104063332080841, "grad_norm": 0.11456745862960815, "global_step": 4680, "epoch": 44, "lr": 9.808760674071243e-05} +{"train_loss": 0.01929960958659649, "grad_norm": 0.05281859636306763, "global_step": 4681, "epoch": 44, "lr": 9.808679481767509e-05} +{"train_loss": 0.018500961363315582, "grad_norm": 0.06078468635678291, "global_step": 4682, "epoch": 44, "lr": 9.808598272568165e-05} +{"train_loss": 0.016402598470449448, "grad_norm": 0.045929014682769775, "global_step": 4683, "epoch": 44, "lr": 9.8085170464735e-05} +{"train_loss": 0.0186658576130867, "grad_norm": 0.09431982040405273, "global_step": 4684, "epoch": 44, "lr": 9.808435803483796e-05} +{"train_loss": 0.02088095061480999, "grad_norm": 0.08951426297426224, "global_step": 4685, "epoch": 44, "lr": 9.808354543599341e-05} +{"train_loss": 0.021881215274333954, "grad_norm": 0.11593032628297806, "global_step": 4686, "epoch": 44, "lr": 9.808273266820418e-05} +{"train_loss": 0.023014970123767853, "grad_norm": 0.06590613722801208, "global_step": 4687, "epoch": 44, "lr": 9.808191973147314e-05} +{"train_loss": 0.02252396196126938, "grad_norm": 0.07611612230539322, "global_step": 4688, "epoch": 44, "lr": 9.808110662580314e-05} +{"train_loss": 0.019919149577617645, "grad_norm": 0.08197890967130661, "global_step": 4689, "epoch": 44, "lr": 9.808029335119706e-05} +{"train_loss": 0.02233036421239376, "grad_norm": 0.044241853058338165, "global_step": 4690, "epoch": 44, "lr": 9.807947990765773e-05} +{"train_loss": 0.019871190190315247, "grad_norm": 0.040151141583919525, "global_step": 4691, "epoch": 44, "lr": 9.8078666295188e-05} +{"train_loss": 0.02056356891989708, "grad_norm": 0.035221390426158905, "global_step": 4692, "epoch": 44, "lr": 9.807785251379075e-05} +{"train_loss": 0.01772485487163067, "grad_norm": 0.04625706374645233, "global_step": 4693, "epoch": 44, "lr": 9.807703856346885e-05} +{"train_loss": 0.015591878443956375, "grad_norm": 0.045500535517930984, "global_step": 4694, "epoch": 44, "lr": 9.807622444422516e-05} +{"train_loss": 0.017876122146844864, "grad_norm": 0.037889912724494934, "global_step": 4695, "epoch": 44, "lr": 9.807541015606249e-05} +{"train_loss": 0.019177662208676338, "grad_norm": 0.07275764644145966, "global_step": 4696, "epoch": 44, "lr": 9.807459569898374e-05} +{"train_loss": 0.019432485103607178, "grad_norm": 0.04839228093624115, "global_step": 4697, "epoch": 44, "lr": 9.807378107299179e-05} +{"train_loss": 0.019036579877138138, "grad_norm": 0.06568186730146408, "global_step": 4698, "epoch": 44, "lr": 9.807296627808946e-05} +{"train_loss": 0.019941166043281555, "grad_norm": 0.054199811071157455, "global_step": 4699, "epoch": 44, "lr": 9.807215131427965e-05} +{"train_loss": 0.024688448756933212, "grad_norm": 0.0462101548910141, "global_step": 4700, "epoch": 44, "lr": 9.807133618156519e-05} +{"train_loss": 0.021471604704856873, "grad_norm": 0.0488816499710083, "global_step": 4701, "epoch": 44, "lr": 9.807052087994897e-05} +{"train_loss": 0.01839582249522209, "grad_norm": 0.04855601489543915, "global_step": 4702, "epoch": 44, "lr": 9.806970540943383e-05} +{"train_loss": 0.0174392219632864, "grad_norm": 0.05528857186436653, "global_step": 4703, "epoch": 44, "lr": 9.806888977002266e-05} +{"train_loss": 0.018105164170265198, "grad_norm": 0.06142927706241608, "global_step": 4704, "epoch": 44, "lr": 9.806807396171832e-05} +{"train_loss": 0.017224494367837906, "grad_norm": 0.0667300894856453, "global_step": 4705, "epoch": 44, "lr": 9.806725798452367e-05} +{"train_loss": 0.020850900560617447, "grad_norm": 0.039315540343523026, "global_step": 4706, "epoch": 44, "lr": 9.806644183844159e-05} +{"train_loss": 0.020641574636101723, "grad_norm": 0.058475494384765625, "global_step": 4707, "epoch": 44, "lr": 9.806562552347491e-05} +{"train_loss": 0.019285976886749268, "grad_norm": 0.07334814220666885, "global_step": 4708, "epoch": 44, "lr": 9.806480903962655e-05} +{"train_loss": 0.019086312502622604, "grad_norm": 0.06841664016246796, "global_step": 4709, "epoch": 44, "lr": 9.806399238689934e-05} +{"train_loss": 0.019398897886276245, "grad_norm": 0.046667926013469696, "global_step": 4710, "epoch": 44, "lr": 9.806317556529617e-05} +{"train_loss": 0.021150212734937668, "grad_norm": 0.11851076781749725, "global_step": 4711, "epoch": 44, "lr": 9.806235857481991e-05} +{"train_loss": 0.019500356167554855, "grad_norm": 0.10569898039102554, "global_step": 4712, "epoch": 44, "lr": 9.806154141547342e-05} +{"train_loss": 0.020837608724832535, "grad_norm": 0.057352643460035324, "global_step": 4713, "epoch": 44, "lr": 9.806072408725956e-05} +{"train_loss": 0.018883462995290756, "grad_norm": 0.08412756770849228, "global_step": 4714, "epoch": 44, "lr": 9.805990659018123e-05} +{"train_loss": 0.022692449390888214, "grad_norm": 0.04318659380078316, "global_step": 4715, "epoch": 44, "lr": 9.805908892424128e-05} +{"train_loss": 0.018819883465766907, "grad_norm": 0.05691729485988617, "global_step": 4716, "epoch": 44, "lr": 9.80582710894426e-05} +{"train_loss": 0.01796690747141838, "grad_norm": 0.04663700982928276, "global_step": 4717, "epoch": 44, "lr": 9.805745308578806e-05} +{"train_loss": 0.018310073763132095, "grad_norm": 0.14575062692165375, "global_step": 4718, "epoch": 44, "lr": 9.805663491328052e-05} +{"train_loss": 0.01953992247581482, "grad_norm": 0.055024970322847366, "global_step": 4719, "epoch": 44, "lr": 9.805581657192287e-05} +{"train_loss": 0.02226230688393116, "grad_norm": 0.08642307668924332, "global_step": 4720, "epoch": 44, "lr": 9.805499806171797e-05} +{"train_loss": 0.01861000433564186, "grad_norm": 0.11634598672389984, "global_step": 4721, "epoch": 44, "lr": 9.80541793826687e-05} +{"train_loss": 0.023087378591299057, "grad_norm": 0.07145823538303375, "global_step": 4722, "epoch": 44, "lr": 9.805336053477795e-05} +{"train_loss": 0.01734323985874653, "grad_norm": 0.07335108518600464, "global_step": 4723, "epoch": 44, "lr": 9.805254151804861e-05} +{"train_loss": 0.022656485438346863, "grad_norm": 0.05629940703511238, "global_step": 4724, "epoch": 44, "lr": 9.805172233248351e-05} +{"train_loss": 0.018426241353154182, "grad_norm": 0.1026206910610199, "global_step": 4725, "epoch": 44, "lr": 9.805090297808557e-05} +{"train_loss": 0.018420511856675148, "grad_norm": 0.08865704387426376, "global_step": 4726, "epoch": 44, "lr": 9.805008345485765e-05} +{"train_loss": 0.015001763589680195, "grad_norm": 0.033378928899765015, "global_step": 4727, "epoch": 44, "lr": 9.804926376280264e-05} +{"train_loss": 0.018576372414827347, "grad_norm": 0.0572352334856987, "global_step": 4728, "epoch": 44, "lr": 9.804844390192339e-05} +{"train_loss": 0.01731722615659237, "grad_norm": 0.1264689564704895, "global_step": 4729, "epoch": 44, "lr": 9.804762387222282e-05} +{"train_loss": 0.019823916256427765, "grad_norm": 0.07731291651725769, "global_step": 4730, "epoch": 44, "lr": 9.804680367370379e-05} +{"train_loss": 0.02137148007750511, "grad_norm": 0.0916869267821312, "global_step": 4731, "epoch": 44, "lr": 9.804598330636919e-05} +{"train_loss": 0.019919313490390778, "grad_norm": 0.04178501293063164, "global_step": 4732, "epoch": 44, "lr": 9.80451627702219e-05} +{"train_loss": 0.01854061707854271, "grad_norm": 0.06697895377874374, "global_step": 4733, "epoch": 44, "lr": 9.804434206526481e-05} +{"train_loss": 0.021630220115184784, "grad_norm": 0.0519610196352005, "global_step": 4734, "epoch": 44, "lr": 9.804352119150077e-05} +{"train_loss": 0.019364479929208755, "grad_norm": 0.04480719566345215, "global_step": 4735, "epoch": 44, "lr": 9.804270014893271e-05} +{"train_loss": 0.01833009161055088, "grad_norm": 0.05398121848702431, "global_step": 4736, "epoch": 44, "lr": 9.80418789375635e-05} +{"train_loss": 0.01670733094215393, "grad_norm": 0.09237097948789597, "global_step": 4737, "epoch": 44, "lr": 9.804105755739602e-05} +{"train_loss": 0.01985086128115654, "grad_norm": 0.05175527185201645, "global_step": 4738, "epoch": 44, "lr": 9.804023600843314e-05} +{"train_loss": 0.02228458970785141, "grad_norm": 0.07252278923988342, "global_step": 4739, "epoch": 44, "lr": 9.803941429067777e-05} +{"train_loss": 0.023082830011844635, "grad_norm": 0.04926950857043266, "global_step": 4740, "epoch": 44, "lr": 9.803859240413279e-05} +{"train_loss": 0.021336747333407402, "grad_norm": 0.07260959595441818, "global_step": 4741, "epoch": 44, "lr": 9.803777034880109e-05} +{"train_loss": 0.019942359998822212, "grad_norm": 0.12596967816352844, "global_step": 4742, "epoch": 44, "lr": 9.803694812468555e-05} +{"train_loss": 0.01797797530889511, "grad_norm": 0.07841861993074417, "global_step": 4743, "epoch": 44, "lr": 9.803612573178907e-05} +{"train_loss": 0.017284410074353218, "grad_norm": 0.09543964266777039, "global_step": 4744, "epoch": 44, "lr": 9.803530317011455e-05} +{"train_loss": 0.018148355185985565, "grad_norm": 0.0606221929192543, "global_step": 4745, "epoch": 44, "lr": 9.803448043966485e-05} +{"train_loss": 0.015614112839102745, "grad_norm": 0.049674633890390396, "global_step": 4746, "epoch": 44, "lr": 9.803365754044287e-05} +{"train_loss": 0.01954195275902748, "grad_norm": 0.05301865562796593, "global_step": 4747, "epoch": 44, "lr": 9.80328344724515e-05} +{"train_loss": 0.01989361271262169, "grad_norm": 0.09493757784366608, "global_step": 4748, "epoch": 44, "lr": 9.803201123569365e-05} +{"train_loss": 0.021853869780898094, "grad_norm": 0.09838373214006424, "global_step": 4749, "epoch": 44, "lr": 9.803118783017222e-05} +{"train_loss": 0.01776300184428692, "grad_norm": 0.034517962485551834, "global_step": 4750, "epoch": 44, "lr": 9.803036425589005e-05} +{"train_loss": 0.021894536912441254, "grad_norm": 0.10782354325056076, "global_step": 4751, "epoch": 44, "lr": 9.802954051285009e-05} +{"train_loss": 0.0197172649204731, "grad_norm": 0.04528620466589928, "global_step": 4752, "epoch": 44, "lr": 9.80287166010552e-05} +{"train_loss": 0.01878797449171543, "grad_norm": 0.0652010440826416, "global_step": 4753, "epoch": 44, "lr": 9.802789252050831e-05} +{"train_loss": 0.018794480711221695, "grad_norm": 0.10869449377059937, "global_step": 4754, "epoch": 44, "lr": 9.802706827121227e-05} +{"train_loss": 0.019226394593715668, "grad_norm": 0.06742081046104431, "global_step": 4755, "epoch": 44, "lr": 9.802624385317e-05} +{"train_loss": 0.018302153795957565, "grad_norm": 0.05257115885615349, "global_step": 4756, "epoch": 44, "lr": 9.80254192663844e-05} +{"train_loss": 0.019595332443714142, "grad_norm": 0.05493790656328201, "global_step": 4757, "epoch": 44, "lr": 9.802459451085834e-05} +{"train_loss": 0.019551698118448257, "grad_norm": 0.04437994584441185, "global_step": 4758, "epoch": 44, "lr": 9.802376958659479e-05} +{"train_loss": 0.0177917517721653, "grad_norm": 0.08683954179286957, "global_step": 4759, "epoch": 44, "lr": 9.802294449359656e-05} +{"train_loss": 0.018812086433172226, "grad_norm": 0.03488216921687126, "global_step": 4760, "epoch": 44, "lr": 9.80221192318666e-05} +{"train_loss": 0.02172621712088585, "grad_norm": 0.044413138180971146, "global_step": 4761, "epoch": 44, "lr": 9.802129380140779e-05} +{"train_loss": 0.019130151718854904, "grad_norm": 0.054693348705768585, "global_step": 4762, "epoch": 44, "lr": 9.802046820222303e-05} +{"train_loss": 0.020589902997016907, "grad_norm": 0.07993669807910919, "global_step": 4763, "epoch": 44, "lr": 9.801964243431526e-05} +{"train_loss": 0.018598828464746475, "grad_norm": 0.042391031980514526, "global_step": 4764, "epoch": 44, "lr": 9.801881649768731e-05} +{"train_loss": 0.021819492802023888, "grad_norm": 0.08046356588602066, "global_step": 4765, "epoch": 44, "lr": 9.801799039234215e-05} +{"train_loss": 0.017158370465040207, "grad_norm": 0.09783624857664108, "global_step": 4766, "epoch": 44, "lr": 9.801716411828265e-05} +{"train_loss": 0.02185957320034504, "grad_norm": 0.15431369841098785, "global_step": 4767, "epoch": 44, "lr": 9.801633767551172e-05} +{"train_loss": 0.019484560936689377, "grad_norm": 0.12361438572406769, "global_step": 4768, "epoch": 44, "lr": 9.801551106403226e-05} +{"train_loss": 0.019517333554279693, "grad_norm": 0.05278649553656578, "global_step": 4769, "epoch": 44, "lr": 9.801468428384716e-05, "val_loss": 0.019057320430874825} +{"train_loss": 0.01881599798798561, "grad_norm": 0.06013207510113716, "global_step": 4770, "epoch": 45, "lr": 9.801385733495934e-05} +{"train_loss": 0.021872282028198242, "grad_norm": 0.06124984100461006, "global_step": 4771, "epoch": 45, "lr": 9.801303021737172e-05} +{"train_loss": 0.019935615360736847, "grad_norm": 0.11491848528385162, "global_step": 4772, "epoch": 45, "lr": 9.80122029310872e-05} +{"train_loss": 0.022160604596138, "grad_norm": 0.10291687399148941, "global_step": 4773, "epoch": 45, "lr": 9.801137547610865e-05} +{"train_loss": 0.01774260401725769, "grad_norm": 0.06645067781209946, "global_step": 4774, "epoch": 45, "lr": 9.801054785243904e-05} +{"train_loss": 0.02101992815732956, "grad_norm": 0.04482536017894745, "global_step": 4775, "epoch": 45, "lr": 9.800972006008123e-05} +{"train_loss": 0.016135472804307938, "grad_norm": 0.12742742896080017, "global_step": 4776, "epoch": 45, "lr": 9.800889209903814e-05} +{"train_loss": 0.018603459000587463, "grad_norm": 0.04030679911375046, "global_step": 4777, "epoch": 45, "lr": 9.800806396931268e-05} +{"train_loss": 0.0184215996414423, "grad_norm": 0.03331223130226135, "global_step": 4778, "epoch": 45, "lr": 9.800723567090775e-05} +{"train_loss": 0.01610308699309826, "grad_norm": 0.06083519756793976, "global_step": 4779, "epoch": 45, "lr": 9.80064072038263e-05} +{"train_loss": 0.018177680671215057, "grad_norm": 0.04859928786754608, "global_step": 4780, "epoch": 45, "lr": 9.80055785680712e-05} +{"train_loss": 0.019093504175543785, "grad_norm": 0.05151579529047012, "global_step": 4781, "epoch": 45, "lr": 9.800474976364537e-05} +{"train_loss": 0.023119602352380753, "grad_norm": 0.04569912701845169, "global_step": 4782, "epoch": 45, "lr": 9.800392079055172e-05} +{"train_loss": 0.019808951765298843, "grad_norm": 0.06363067775964737, "global_step": 4783, "epoch": 45, "lr": 9.800309164879317e-05} +{"train_loss": 0.018656142055988312, "grad_norm": 0.08296843618154526, "global_step": 4784, "epoch": 45, "lr": 9.800226233837264e-05} +{"train_loss": 0.019893428310751915, "grad_norm": 0.14292003214359283, "global_step": 4785, "epoch": 45, "lr": 9.800143285929305e-05} +{"train_loss": 0.01761651784181595, "grad_norm": 0.05935266613960266, "global_step": 4786, "epoch": 45, "lr": 9.800060321155728e-05} +{"train_loss": 0.01845843903720379, "grad_norm": 0.08321160078048706, "global_step": 4787, "epoch": 45, "lr": 9.799977339516826e-05} +{"train_loss": 0.01896856166422367, "grad_norm": 0.11767792701721191, "global_step": 4788, "epoch": 45, "lr": 9.799894341012893e-05} +{"train_loss": 0.015897922217845917, "grad_norm": 0.06326337158679962, "global_step": 4789, "epoch": 45, "lr": 9.799811325644217e-05} +{"train_loss": 0.021785881370306015, "grad_norm": 0.07136040180921555, "global_step": 4790, "epoch": 45, "lr": 9.799728293411092e-05} +{"train_loss": 0.017340194433927536, "grad_norm": 0.0400785394012928, "global_step": 4791, "epoch": 45, "lr": 9.799645244313809e-05} +{"train_loss": 0.019055083394050598, "grad_norm": 0.04372832179069519, "global_step": 4792, "epoch": 45, "lr": 9.799562178352662e-05} +{"train_loss": 0.017355237156152725, "grad_norm": 0.03487984091043472, "global_step": 4793, "epoch": 45, "lr": 9.799479095527938e-05} +{"train_loss": 0.020535284653306007, "grad_norm": 0.09330274909734726, "global_step": 4794, "epoch": 45, "lr": 9.799395995839933e-05} +{"train_loss": 0.018699930980801582, "grad_norm": 0.05411022529006004, "global_step": 4795, "epoch": 45, "lr": 9.799312879288938e-05} +{"train_loss": 0.01811652071774006, "grad_norm": 0.05542576685547829, "global_step": 4796, "epoch": 45, "lr": 9.799229745875242e-05} +{"train_loss": 0.018637556582689285, "grad_norm": 0.0668172538280487, "global_step": 4797, "epoch": 45, "lr": 9.799146595599142e-05} +{"train_loss": 0.021323688328266144, "grad_norm": 0.14437182247638702, "global_step": 4798, "epoch": 45, "lr": 9.799063428460928e-05} +{"train_loss": 0.018475530669093132, "grad_norm": 0.10633959621191025, "global_step": 4799, "epoch": 45, "lr": 9.798980244460893e-05} +{"train_loss": 0.021086812019348145, "grad_norm": 0.09163937717676163, "global_step": 4800, "epoch": 45, "lr": 9.798897043599327e-05} +{"train_loss": 0.02079026773571968, "grad_norm": 0.07507684826850891, "global_step": 4801, "epoch": 45, "lr": 9.798813825876525e-05} +{"train_loss": 0.02023998647928238, "grad_norm": 0.122392438352108, "global_step": 4802, "epoch": 45, "lr": 9.798730591292777e-05} +{"train_loss": 0.020300252363085747, "grad_norm": 0.044828176498413086, "global_step": 4803, "epoch": 45, "lr": 9.798647339848376e-05} +{"train_loss": 0.017906345427036285, "grad_norm": 0.08298886567354202, "global_step": 4804, "epoch": 45, "lr": 9.798564071543617e-05} +{"train_loss": 0.018992405384778976, "grad_norm": 0.05445731058716774, "global_step": 4805, "epoch": 45, "lr": 9.79848078637879e-05} +{"train_loss": 0.019516456872224808, "grad_norm": 0.040728598833084106, "global_step": 4806, "epoch": 45, "lr": 9.798397484354187e-05} +{"train_loss": 0.02030296064913273, "grad_norm": 0.04521554708480835, "global_step": 4807, "epoch": 45, "lr": 9.798314165470104e-05} +{"train_loss": 0.021989421918988228, "grad_norm": 0.11628440022468567, "global_step": 4808, "epoch": 45, "lr": 9.798230829726831e-05} +{"train_loss": 0.018761824816465378, "grad_norm": 0.06177941709756851, "global_step": 4809, "epoch": 45, "lr": 9.79814747712466e-05} +{"train_loss": 0.017164509743452072, "grad_norm": 0.0339372456073761, "global_step": 4810, "epoch": 45, "lr": 9.798064107663888e-05} +{"train_loss": 0.021646223962306976, "grad_norm": 0.04688802734017372, "global_step": 4811, "epoch": 45, "lr": 9.797980721344804e-05} +{"train_loss": 0.015896204859018326, "grad_norm": 0.15018370747566223, "global_step": 4812, "epoch": 45, "lr": 9.797897318167702e-05} +{"train_loss": 0.016923105344176292, "grad_norm": 0.05109921097755432, "global_step": 4813, "epoch": 45, "lr": 9.797813898132877e-05} +{"train_loss": 0.018659725785255432, "grad_norm": 0.05778292194008827, "global_step": 4814, "epoch": 45, "lr": 9.79773046124062e-05} +{"train_loss": 0.02029145136475563, "grad_norm": 0.1310388743877411, "global_step": 4815, "epoch": 45, "lr": 9.797647007491222e-05} +{"train_loss": 0.01949061080813408, "grad_norm": 0.06467929482460022, "global_step": 4816, "epoch": 45, "lr": 9.797563536884981e-05} +{"train_loss": 0.02081238478422165, "grad_norm": 0.06231497600674629, "global_step": 4817, "epoch": 45, "lr": 9.797480049422189e-05} +{"train_loss": 0.01771780289709568, "grad_norm": 0.07994020730257034, "global_step": 4818, "epoch": 45, "lr": 9.797396545103137e-05} +{"train_loss": 0.017322562634944916, "grad_norm": 0.09364993125200272, "global_step": 4819, "epoch": 45, "lr": 9.797313023928121e-05} +{"train_loss": 0.017029110342264175, "grad_norm": 0.03633221611380577, "global_step": 4820, "epoch": 45, "lr": 9.797229485897432e-05} +{"train_loss": 0.017736729234457016, "grad_norm": 0.05628214776515961, "global_step": 4821, "epoch": 45, "lr": 9.797145931011365e-05} +{"train_loss": 0.018332337960600853, "grad_norm": 0.036232877522706985, "global_step": 4822, "epoch": 45, "lr": 9.797062359270215e-05} +{"train_loss": 0.02053084969520569, "grad_norm": 0.0402788408100605, "global_step": 4823, "epoch": 45, "lr": 9.796978770674273e-05} +{"train_loss": 0.0219956636428833, "grad_norm": 0.08755150437355042, "global_step": 4824, "epoch": 45, "lr": 9.796895165223834e-05} +{"train_loss": 0.021064985543489456, "grad_norm": 0.04845491796731949, "global_step": 4825, "epoch": 45, "lr": 9.796811542919192e-05} +{"train_loss": 0.020118387416005135, "grad_norm": 0.04955975338816643, "global_step": 4826, "epoch": 45, "lr": 9.796727903760638e-05} +{"train_loss": 0.01770900748670101, "grad_norm": 0.06477587670087814, "global_step": 4827, "epoch": 45, "lr": 9.79664424774847e-05} +{"train_loss": 0.02034037373960018, "grad_norm": 0.07626347243785858, "global_step": 4828, "epoch": 45, "lr": 9.79656057488298e-05} +{"train_loss": 0.018680082634091377, "grad_norm": 0.11795707046985626, "global_step": 4829, "epoch": 45, "lr": 9.796476885164462e-05} +{"train_loss": 0.022111261263489723, "grad_norm": 0.11775021255016327, "global_step": 4830, "epoch": 45, "lr": 9.79639317859321e-05} +{"train_loss": 0.019070854410529137, "grad_norm": 0.09818217158317566, "global_step": 4831, "epoch": 45, "lr": 9.796309455169519e-05} +{"train_loss": 0.016235068440437317, "grad_norm": 0.09425468742847443, "global_step": 4832, "epoch": 45, "lr": 9.796225714893681e-05} +{"train_loss": 0.01718379557132721, "grad_norm": 0.057358451187610626, "global_step": 4833, "epoch": 45, "lr": 9.796141957765993e-05} +{"train_loss": 0.017800873145461082, "grad_norm": 0.09171567857265472, "global_step": 4834, "epoch": 45, "lr": 9.796058183786747e-05} +{"train_loss": 0.018118947744369507, "grad_norm": 0.08916065841913223, "global_step": 4835, "epoch": 45, "lr": 9.79597439295624e-05} +{"train_loss": 0.020257452502846718, "grad_norm": 0.03800428658723831, "global_step": 4836, "epoch": 45, "lr": 9.795890585274763e-05} +{"train_loss": 0.019574647769331932, "grad_norm": 0.12183312326669693, "global_step": 4837, "epoch": 45, "lr": 9.795806760742611e-05} +{"train_loss": 0.020993083715438843, "grad_norm": 0.04207543283700943, "global_step": 4838, "epoch": 45, "lr": 9.795722919360082e-05} +{"train_loss": 0.022231925278902054, "grad_norm": 0.06071479991078377, "global_step": 4839, "epoch": 45, "lr": 9.795639061127468e-05} +{"train_loss": 0.016912542283535004, "grad_norm": 0.06169964373111725, "global_step": 4840, "epoch": 45, "lr": 9.795555186045063e-05} +{"train_loss": 0.016118042171001434, "grad_norm": 0.062312908470630646, "global_step": 4841, "epoch": 45, "lr": 9.795471294113163e-05} +{"train_loss": 0.019740939140319824, "grad_norm": 0.1073371097445488, "global_step": 4842, "epoch": 45, "lr": 9.795387385332064e-05} +{"train_loss": 0.018945373594760895, "grad_norm": 0.07115764915943146, "global_step": 4843, "epoch": 45, "lr": 9.795303459702056e-05} +{"train_loss": 0.021856365725398064, "grad_norm": 0.06118491664528847, "global_step": 4844, "epoch": 45, "lr": 9.79521951722344e-05} +{"train_loss": 0.015748344361782074, "grad_norm": 0.05609841272234917, "global_step": 4845, "epoch": 45, "lr": 9.795135557896506e-05} +{"train_loss": 0.019517602398991585, "grad_norm": 0.07302815467119217, "global_step": 4846, "epoch": 45, "lr": 9.795051581721552e-05} +{"train_loss": 0.022073950618505478, "grad_norm": 0.14159491658210754, "global_step": 4847, "epoch": 45, "lr": 9.794967588698872e-05} +{"train_loss": 0.019277645274996758, "grad_norm": 0.08363852649927139, "global_step": 4848, "epoch": 45, "lr": 9.794883578828761e-05} +{"train_loss": 0.020489659160375595, "grad_norm": 0.09649508446455002, "global_step": 4849, "epoch": 45, "lr": 9.794799552111513e-05} +{"train_loss": 0.0209323950111866, "grad_norm": 0.08039145916700363, "global_step": 4850, "epoch": 45, "lr": 9.794715508547428e-05} +{"train_loss": 0.021105732768774033, "grad_norm": 0.1169358417391777, "global_step": 4851, "epoch": 45, "lr": 9.794631448136795e-05} +{"train_loss": 0.023116013035178185, "grad_norm": 0.06819453835487366, "global_step": 4852, "epoch": 45, "lr": 9.794547370879911e-05} +{"train_loss": 0.020642656832933426, "grad_norm": 0.0579083114862442, "global_step": 4853, "epoch": 45, "lr": 9.794463276777077e-05} +{"train_loss": 0.020721081644296646, "grad_norm": 0.10054180026054382, "global_step": 4854, "epoch": 45, "lr": 9.79437916582858e-05} +{"train_loss": 0.019166836515069008, "grad_norm": 0.045626744627952576, "global_step": 4855, "epoch": 45, "lr": 9.794295038034722e-05} +{"train_loss": 0.020306304097175598, "grad_norm": 0.05671550706028938, "global_step": 4856, "epoch": 45, "lr": 9.794210893395796e-05} +{"train_loss": 0.020164377987384796, "grad_norm": 0.04793926328420639, "global_step": 4857, "epoch": 45, "lr": 9.794126731912097e-05} +{"train_loss": 0.020768392831087112, "grad_norm": 0.07880206406116486, "global_step": 4858, "epoch": 45, "lr": 9.794042553583922e-05} +{"train_loss": 0.021960988640785217, "grad_norm": 0.17903392016887665, "global_step": 4859, "epoch": 45, "lr": 9.793958358411566e-05} +{"train_loss": 0.02244170941412449, "grad_norm": 0.05914711207151413, "global_step": 4860, "epoch": 45, "lr": 9.793874146395326e-05} +{"train_loss": 0.0194348506629467, "grad_norm": 0.04282615706324577, "global_step": 4861, "epoch": 45, "lr": 9.793789917535496e-05} +{"train_loss": 0.021166307851672173, "grad_norm": 0.04598046839237213, "global_step": 4862, "epoch": 45, "lr": 9.793705671832374e-05} +{"train_loss": 0.018573977053165436, "grad_norm": 0.0389183908700943, "global_step": 4863, "epoch": 45, "lr": 9.793621409286254e-05} +{"train_loss": 0.020361006259918213, "grad_norm": 0.08917287737131119, "global_step": 4864, "epoch": 45, "lr": 9.793537129897434e-05} +{"train_loss": 0.023836087435483932, "grad_norm": 0.05492116138339043, "global_step": 4865, "epoch": 45, "lr": 9.793452833666207e-05} +{"train_loss": 0.021320592612028122, "grad_norm": 0.07675319910049438, "global_step": 4866, "epoch": 45, "lr": 9.793368520592875e-05} +{"train_loss": 0.019583823159337044, "grad_norm": 0.09117141366004944, "global_step": 4867, "epoch": 45, "lr": 9.793284190677728e-05} +{"train_loss": 0.019613116979599, "grad_norm": 0.17941462993621826, "global_step": 4868, "epoch": 45, "lr": 9.793199843921066e-05} +{"train_loss": 0.018693432211875916, "grad_norm": 0.07895966619253159, "global_step": 4869, "epoch": 45, "lr": 9.793115480323183e-05} +{"train_loss": 0.019823240116238594, "grad_norm": 0.04156699776649475, "global_step": 4870, "epoch": 45, "lr": 9.793031099884378e-05} +{"train_loss": 0.01915106549859047, "grad_norm": 0.07104088366031647, "global_step": 4871, "epoch": 45, "lr": 9.792946702604946e-05} +{"train_loss": 0.019301068037748337, "grad_norm": 0.07670196145772934, "global_step": 4872, "epoch": 45, "lr": 9.792862288485183e-05} +{"train_loss": 0.022294245660305023, "grad_norm": 0.06680218130350113, "global_step": 4873, "epoch": 45, "lr": 9.792777857525387e-05} +{"train_loss": 0.018698081374168396, "grad_norm": 0.039167456328868866, "global_step": 4874, "epoch": 45, "lr": 9.792693409725854e-05} +{"train_loss": 0.01947638452193647, "grad_norm": 0.04521949216723442, "global_step": 4875, "epoch": 45, "lr": 9.79260894508688e-05, "val_loss": 0.01932705193758011} +{"train_loss": 0.02214054949581623, "grad_norm": 0.05563848465681076, "global_step": 4876, "epoch": 46, "lr": 9.792524463608763e-05} +{"train_loss": 0.018494093790650368, "grad_norm": 0.07006634771823883, "global_step": 4877, "epoch": 46, "lr": 9.792439965291798e-05} +{"train_loss": 0.01972457766532898, "grad_norm": 0.15657000243663788, "global_step": 4878, "epoch": 46, "lr": 9.792355450136284e-05} +{"train_loss": 0.018476184457540512, "grad_norm": 0.06495185196399689, "global_step": 4879, "epoch": 46, "lr": 9.792270918142518e-05} +{"train_loss": 0.019117653369903564, "grad_norm": 0.13155107200145721, "global_step": 4880, "epoch": 46, "lr": 9.792186369310795e-05} +{"train_loss": 0.01820242777466774, "grad_norm": 0.038674384355545044, "global_step": 4881, "epoch": 46, "lr": 9.792101803641413e-05} +{"train_loss": 0.018967993557453156, "grad_norm": 0.06825581192970276, "global_step": 4882, "epoch": 46, "lr": 9.79201722113467e-05} +{"train_loss": 0.019995659589767456, "grad_norm": 0.045422330498695374, "global_step": 4883, "epoch": 46, "lr": 9.791932621790862e-05} +{"train_loss": 0.0180289875715971, "grad_norm": 0.040724508464336395, "global_step": 4884, "epoch": 46, "lr": 9.791848005610287e-05} +{"train_loss": 0.018031351268291473, "grad_norm": 0.06537749618291855, "global_step": 4885, "epoch": 46, "lr": 9.791763372593241e-05} +{"train_loss": 0.019422948360443115, "grad_norm": 0.1038990393280983, "global_step": 4886, "epoch": 46, "lr": 9.791678722740024e-05} +{"train_loss": 0.01916111633181572, "grad_norm": 0.046058882027864456, "global_step": 4887, "epoch": 46, "lr": 9.79159405605093e-05} +{"train_loss": 0.01683966815471649, "grad_norm": 0.03677160292863846, "global_step": 4888, "epoch": 46, "lr": 9.79150937252626e-05} +{"train_loss": 0.018755203112959862, "grad_norm": 0.053347405046224594, "global_step": 4889, "epoch": 46, "lr": 9.791424672166309e-05} +{"train_loss": 0.02052146941423416, "grad_norm": 0.0712190717458725, "global_step": 4890, "epoch": 46, "lr": 9.791339954971377e-05} +{"train_loss": 0.019790105521678925, "grad_norm": 0.0675382912158966, "global_step": 4891, "epoch": 46, "lr": 9.791255220941759e-05} +{"train_loss": 0.01887085661292076, "grad_norm": 0.08100493997335434, "global_step": 4892, "epoch": 46, "lr": 9.791170470077753e-05} +{"train_loss": 0.021200228482484818, "grad_norm": 0.12869475781917572, "global_step": 4893, "epoch": 46, "lr": 9.791085702379659e-05} +{"train_loss": 0.01673147641122341, "grad_norm": 0.06049885228276253, "global_step": 4894, "epoch": 46, "lr": 9.791000917847772e-05} +{"train_loss": 0.018291257321834564, "grad_norm": 0.07221342623233795, "global_step": 4895, "epoch": 46, "lr": 9.790916116482392e-05} +{"train_loss": 0.021168531849980354, "grad_norm": 0.04628828912973404, "global_step": 4896, "epoch": 46, "lr": 9.790831298283816e-05} +{"train_loss": 0.0205213725566864, "grad_norm": 0.04563508182764053, "global_step": 4897, "epoch": 46, "lr": 9.790746463252342e-05} +{"train_loss": 0.017883144319057465, "grad_norm": 0.05618498846888542, "global_step": 4898, "epoch": 46, "lr": 9.79066161138827e-05} +{"train_loss": 0.02306189201772213, "grad_norm": 0.12722927331924438, "global_step": 4899, "epoch": 46, "lr": 9.790576742691895e-05} +{"train_loss": 0.019598983228206635, "grad_norm": 0.0649850070476532, "global_step": 4900, "epoch": 46, "lr": 9.790491857163517e-05} +{"train_loss": 0.01917083188891411, "grad_norm": 0.06422269344329834, "global_step": 4901, "epoch": 46, "lr": 9.790406954803434e-05} +{"train_loss": 0.020005077123641968, "grad_norm": 0.0720815509557724, "global_step": 4902, "epoch": 46, "lr": 9.790322035611945e-05} +{"train_loss": 0.018660008907318115, "grad_norm": 0.08413955569267273, "global_step": 4903, "epoch": 46, "lr": 9.790237099589346e-05} +{"train_loss": 0.01820089854300022, "grad_norm": 0.07601413875818253, "global_step": 4904, "epoch": 46, "lr": 9.790152146735938e-05} +{"train_loss": 0.02063480019569397, "grad_norm": 0.08502674847841263, "global_step": 4905, "epoch": 46, "lr": 9.790067177052019e-05} +{"train_loss": 0.016925547271966934, "grad_norm": 0.08122483640909195, "global_step": 4906, "epoch": 46, "lr": 9.789982190537886e-05} +{"train_loss": 0.018163926899433136, "grad_norm": 0.06301283836364746, "global_step": 4907, "epoch": 46, "lr": 9.78989718719384e-05} +{"train_loss": 0.02099691331386566, "grad_norm": 0.06989731639623642, "global_step": 4908, "epoch": 46, "lr": 9.789812167020178e-05} +{"train_loss": 0.020480982959270477, "grad_norm": 0.0663171112537384, "global_step": 4909, "epoch": 46, "lr": 9.789727130017197e-05} +{"train_loss": 0.01998535916209221, "grad_norm": 0.08186180889606476, "global_step": 4910, "epoch": 46, "lr": 9.789642076185201e-05} +{"train_loss": 0.018800387158989906, "grad_norm": 0.0904931053519249, "global_step": 4911, "epoch": 46, "lr": 9.789557005524484e-05} +{"train_loss": 0.01981264352798462, "grad_norm": 0.04192427173256874, "global_step": 4912, "epoch": 46, "lr": 9.789471918035346e-05} +{"train_loss": 0.020213767886161804, "grad_norm": 0.04621991887688637, "global_step": 4913, "epoch": 46, "lr": 9.789386813718088e-05} +{"train_loss": 0.01967904530465603, "grad_norm": 0.0623408779501915, "global_step": 4914, "epoch": 46, "lr": 9.789301692573007e-05} +{"train_loss": 0.017124222591519356, "grad_norm": 0.0552872009575367, "global_step": 4915, "epoch": 46, "lr": 9.789216554600402e-05} +{"train_loss": 0.015785597264766693, "grad_norm": 0.08567021042108536, "global_step": 4916, "epoch": 46, "lr": 9.789131399800574e-05} +{"train_loss": 0.01769997552037239, "grad_norm": 0.04751694202423096, "global_step": 4917, "epoch": 46, "lr": 9.789046228173821e-05} +{"train_loss": 0.02000495418906212, "grad_norm": 0.06404382735490799, "global_step": 4918, "epoch": 46, "lr": 9.788961039720442e-05} +{"train_loss": 0.020766548812389374, "grad_norm": 0.039284516125917435, "global_step": 4919, "epoch": 46, "lr": 9.788875834440737e-05} +{"train_loss": 0.01980961300432682, "grad_norm": 0.092249296605587, "global_step": 4920, "epoch": 46, "lr": 9.788790612335004e-05} +{"train_loss": 0.021615032106637955, "grad_norm": 0.04328199848532677, "global_step": 4921, "epoch": 46, "lr": 9.788705373403543e-05} +{"train_loss": 0.019082961603999138, "grad_norm": 0.06067890673875809, "global_step": 4922, "epoch": 46, "lr": 9.788620117646654e-05} +{"train_loss": 0.0167097058147192, "grad_norm": 0.03428139537572861, "global_step": 4923, "epoch": 46, "lr": 9.788534845064636e-05} +{"train_loss": 0.019533667713403702, "grad_norm": 0.040160100907087326, "global_step": 4924, "epoch": 46, "lr": 9.788449555657791e-05} +{"train_loss": 0.018053269013762474, "grad_norm": 0.1255914270877838, "global_step": 4925, "epoch": 46, "lr": 9.788364249426416e-05} +{"train_loss": 0.01918201893568039, "grad_norm": 0.12241119146347046, "global_step": 4926, "epoch": 46, "lr": 9.788278926370811e-05} +{"train_loss": 0.01776396483182907, "grad_norm": 0.044576242566108704, "global_step": 4927, "epoch": 46, "lr": 9.788193586491277e-05} +{"train_loss": 0.02134256809949875, "grad_norm": 0.10054148733615875, "global_step": 4928, "epoch": 46, "lr": 9.788108229788111e-05} +{"train_loss": 0.01787957176566124, "grad_norm": 0.04010360687971115, "global_step": 4929, "epoch": 46, "lr": 9.788022856261617e-05} +{"train_loss": 0.018312204629182816, "grad_norm": 0.0988074317574501, "global_step": 4930, "epoch": 46, "lr": 9.787937465912092e-05} +{"train_loss": 0.021197577938437462, "grad_norm": 0.07311142981052399, "global_step": 4931, "epoch": 46, "lr": 9.787852058739837e-05} +{"train_loss": 0.01951078325510025, "grad_norm": 0.10121052712202072, "global_step": 4932, "epoch": 46, "lr": 9.787766634745153e-05} +{"train_loss": 0.01954210177063942, "grad_norm": 0.06257868558168411, "global_step": 4933, "epoch": 46, "lr": 9.787681193928338e-05} +{"train_loss": 0.017454735934734344, "grad_norm": 0.09703486412763596, "global_step": 4934, "epoch": 46, "lr": 9.787595736289695e-05} +{"train_loss": 0.016506997868418694, "grad_norm": 0.10607032477855682, "global_step": 4935, "epoch": 46, "lr": 9.787510261829522e-05} +{"train_loss": 0.021056970581412315, "grad_norm": 0.10995694994926453, "global_step": 4936, "epoch": 46, "lr": 9.787424770548119e-05} +{"train_loss": 0.01916750892996788, "grad_norm": 0.09570500999689102, "global_step": 4937, "epoch": 46, "lr": 9.787339262445788e-05} +{"train_loss": 0.020381171256303787, "grad_norm": 0.06320624053478241, "global_step": 4938, "epoch": 46, "lr": 9.787253737522827e-05} +{"train_loss": 0.017700092867016792, "grad_norm": 0.07148393243551254, "global_step": 4939, "epoch": 46, "lr": 9.787168195779541e-05} +{"train_loss": 0.019651412963867188, "grad_norm": 0.10641872882843018, "global_step": 4940, "epoch": 46, "lr": 9.787082637216228e-05} +{"train_loss": 0.01981293596327305, "grad_norm": 0.05219583958387375, "global_step": 4941, "epoch": 46, "lr": 9.786997061833186e-05} +{"train_loss": 0.01836496964097023, "grad_norm": 0.05065202713012695, "global_step": 4942, "epoch": 46, "lr": 9.78691146963072e-05} +{"train_loss": 0.017763860523700714, "grad_norm": 0.06330157816410065, "global_step": 4943, "epoch": 46, "lr": 9.786825860609127e-05} +{"train_loss": 0.02048294059932232, "grad_norm": 0.045286040753126144, "global_step": 4944, "epoch": 46, "lr": 9.786740234768713e-05} +{"train_loss": 0.019678805023431778, "grad_norm": 0.10593877732753754, "global_step": 4945, "epoch": 46, "lr": 9.786654592109772e-05} +{"train_loss": 0.016994772478938103, "grad_norm": 0.04982144013047218, "global_step": 4946, "epoch": 46, "lr": 9.78656893263261e-05} +{"train_loss": 0.020686421543359756, "grad_norm": 0.060128625482320786, "global_step": 4947, "epoch": 46, "lr": 9.786483256337527e-05} +{"train_loss": 0.017400864511728287, "grad_norm": 0.08778964728116989, "global_step": 4948, "epoch": 46, "lr": 9.786397563224822e-05} +{"train_loss": 0.01844806596636772, "grad_norm": 0.10099145025014877, "global_step": 4949, "epoch": 46, "lr": 9.786311853294798e-05} +{"train_loss": 0.017575642094016075, "grad_norm": 0.05723297968506813, "global_step": 4950, "epoch": 46, "lr": 9.786226126547756e-05} +{"train_loss": 0.020400069653987885, "grad_norm": 0.10008145868778229, "global_step": 4951, "epoch": 46, "lr": 9.786140382983997e-05} +{"train_loss": 0.01816667802631855, "grad_norm": 0.07576397061347961, "global_step": 4952, "epoch": 46, "lr": 9.786054622603822e-05} +{"train_loss": 0.021681390702724457, "grad_norm": 0.09985875338315964, "global_step": 4953, "epoch": 46, "lr": 9.785968845407531e-05} +{"train_loss": 0.019543766975402832, "grad_norm": 0.05770660191774368, "global_step": 4954, "epoch": 46, "lr": 9.785883051395429e-05} +{"train_loss": 0.02204200252890587, "grad_norm": 0.04183013364672661, "global_step": 4955, "epoch": 46, "lr": 9.785797240567815e-05} +{"train_loss": 0.0224359892308712, "grad_norm": 0.04124828428030014, "global_step": 4956, "epoch": 46, "lr": 9.78571141292499e-05} +{"train_loss": 0.021305475383996964, "grad_norm": 0.06081872060894966, "global_step": 4957, "epoch": 46, "lr": 9.785625568467256e-05} +{"train_loss": 0.023364398628473282, "grad_norm": 0.0686546117067337, "global_step": 4958, "epoch": 46, "lr": 9.785539707194915e-05} +{"train_loss": 0.019060488790273666, "grad_norm": 0.03643878549337387, "global_step": 4959, "epoch": 46, "lr": 9.78545382910827e-05} +{"train_loss": 0.024924185127019882, "grad_norm": 0.05003175511956215, "global_step": 4960, "epoch": 46, "lr": 9.78536793420762e-05} +{"train_loss": 0.015747979283332825, "grad_norm": 0.08681557327508926, "global_step": 4961, "epoch": 46, "lr": 9.785282022493269e-05} +{"train_loss": 0.02183975651860237, "grad_norm": 0.08919309079647064, "global_step": 4962, "epoch": 46, "lr": 9.785196093965518e-05} +{"train_loss": 0.018689647316932678, "grad_norm": 0.051132600754499435, "global_step": 4963, "epoch": 46, "lr": 9.785110148624669e-05} +{"train_loss": 0.021220549941062927, "grad_norm": 0.10179980844259262, "global_step": 4964, "epoch": 46, "lr": 9.785024186471023e-05} +{"train_loss": 0.021086666733026505, "grad_norm": 0.04110943526029587, "global_step": 4965, "epoch": 46, "lr": 9.784938207504884e-05} +{"train_loss": 0.0173010416328907, "grad_norm": 0.09001569449901581, "global_step": 4966, "epoch": 46, "lr": 9.784852211726554e-05} +{"train_loss": 0.020034216344356537, "grad_norm": 0.06952625513076782, "global_step": 4967, "epoch": 46, "lr": 9.784766199136331e-05} +{"train_loss": 0.018194425851106644, "grad_norm": 0.03569713234901428, "global_step": 4968, "epoch": 46, "lr": 9.784680169734524e-05} +{"train_loss": 0.020846987143158913, "grad_norm": 0.08930792659521103, "global_step": 4969, "epoch": 46, "lr": 9.784594123521429e-05} +{"train_loss": 0.02283995971083641, "grad_norm": 0.11449062824249268, "global_step": 4970, "epoch": 46, "lr": 9.784508060497351e-05} +{"train_loss": 0.019128818064928055, "grad_norm": 0.05973636731505394, "global_step": 4971, "epoch": 46, "lr": 9.784421980662595e-05} +{"train_loss": 0.021634023636579514, "grad_norm": 0.04264383390545845, "global_step": 4972, "epoch": 46, "lr": 9.78433588401746e-05} +{"train_loss": 0.019465859979391098, "grad_norm": 0.04111287370324135, "global_step": 4973, "epoch": 46, "lr": 9.784249770562247e-05} +{"train_loss": 0.01866455376148224, "grad_norm": 0.11655741930007935, "global_step": 4974, "epoch": 46, "lr": 9.784163640297262e-05} +{"train_loss": 0.019177500158548355, "grad_norm": 0.09625521302223206, "global_step": 4975, "epoch": 46, "lr": 9.784077493222807e-05} +{"train_loss": 0.01780775561928749, "grad_norm": 0.10749376565217972, "global_step": 4976, "epoch": 46, "lr": 9.783991329339185e-05} +{"train_loss": 0.02232842892408371, "grad_norm": 0.12482237815856934, "global_step": 4977, "epoch": 46, "lr": 9.783905148646697e-05} +{"train_loss": 0.016933484002947807, "grad_norm": 0.07738203555345535, "global_step": 4978, "epoch": 46, "lr": 9.783818951145646e-05} +{"train_loss": 0.02060990408062935, "grad_norm": 0.057540521025657654, "global_step": 4979, "epoch": 46, "lr": 9.783732736836338e-05} +{"train_loss": 0.019728701561689377, "grad_norm": 0.09169794619083405, "global_step": 4980, "epoch": 46, "lr": 9.783646505719073e-05} +{"train_loss": 0.01944729128747054, "grad_norm": 0.09185246378183365, "global_step": 4981, "epoch": 46, "lr": 9.783560257794154e-05, "val_loss": 0.018704380840063095} +{"train_loss": 0.01878124289214611, "grad_norm": 0.04916787147521973, "global_step": 4982, "epoch": 47, "lr": 9.783473993061884e-05} +{"train_loss": 0.018902722746133804, "grad_norm": 0.054203953593969345, "global_step": 4983, "epoch": 47, "lr": 9.783387711522568e-05} +{"train_loss": 0.01666618511080742, "grad_norm": 0.06793135404586792, "global_step": 4984, "epoch": 47, "lr": 9.783301413176505e-05} +{"train_loss": 0.019361546263098717, "grad_norm": 0.08580280095338821, "global_step": 4985, "epoch": 47, "lr": 9.783215098024004e-05} +{"train_loss": 0.018630273640155792, "grad_norm": 0.08837692439556122, "global_step": 4986, "epoch": 47, "lr": 9.783128766065363e-05} +{"train_loss": 0.020322442054748535, "grad_norm": 0.0978088304400444, "global_step": 4987, "epoch": 47, "lr": 9.78304241730089e-05} +{"train_loss": 0.020937837660312653, "grad_norm": 0.12926515936851501, "global_step": 4988, "epoch": 47, "lr": 9.782956051730884e-05} +{"train_loss": 0.018477920442819595, "grad_norm": 0.052029382437467575, "global_step": 4989, "epoch": 47, "lr": 9.782869669355651e-05} +{"train_loss": 0.021718500182032585, "grad_norm": 0.12697380781173706, "global_step": 4990, "epoch": 47, "lr": 9.782783270175494e-05} +{"train_loss": 0.01924295909702778, "grad_norm": 0.06649509817361832, "global_step": 4991, "epoch": 47, "lr": 9.782696854190717e-05} +{"train_loss": 0.016061969101428986, "grad_norm": 0.053933706134557724, "global_step": 4992, "epoch": 47, "lr": 9.782610421401624e-05} +{"train_loss": 0.01796991378068924, "grad_norm": 0.09838810563087463, "global_step": 4993, "epoch": 47, "lr": 9.782523971808518e-05} +{"train_loss": 0.020595595240592957, "grad_norm": 0.058043986558914185, "global_step": 4994, "epoch": 47, "lr": 9.7824375054117e-05} +{"train_loss": 0.018804412335157394, "grad_norm": 0.04660497605800629, "global_step": 4995, "epoch": 47, "lr": 9.782351022211478e-05} +{"train_loss": 0.021188758313655853, "grad_norm": 0.11919775605201721, "global_step": 4996, "epoch": 47, "lr": 9.782264522208155e-05} +{"train_loss": 0.019768426194787025, "grad_norm": 0.05788789689540863, "global_step": 4997, "epoch": 47, "lr": 9.782178005402033e-05} +{"train_loss": 0.018626436591148376, "grad_norm": 0.05800740420818329, "global_step": 4998, "epoch": 47, "lr": 9.782091471793418e-05} +{"train_loss": 0.019519852474331856, "grad_norm": 0.06024065613746643, "global_step": 4999, "epoch": 47, "lr": 9.782004921382612e-05} +{"train_loss": 0.021007314324378967, "grad_norm": 0.08186092227697372, "global_step": 5000, "epoch": 47, "lr": 9.781918354169922e-05} +{"train_loss": 0.01647837460041046, "grad_norm": 0.05375800281763077, "global_step": 5001, "epoch": 47, "lr": 9.78183177015565e-05} +{"train_loss": 0.016230741515755653, "grad_norm": 0.10417605191469193, "global_step": 5002, "epoch": 47, "lr": 9.7817451693401e-05} +{"train_loss": 0.019928134977817535, "grad_norm": 0.04299590736627579, "global_step": 5003, "epoch": 47, "lr": 9.781658551723578e-05} +{"train_loss": 0.01654740981757641, "grad_norm": 0.036204416304826736, "global_step": 5004, "epoch": 47, "lr": 9.781571917306387e-05} +{"train_loss": 0.020643942058086395, "grad_norm": 0.14139032363891602, "global_step": 5005, "epoch": 47, "lr": 9.781485266088833e-05} +{"train_loss": 0.022007150575518608, "grad_norm": 0.14275440573692322, "global_step": 5006, "epoch": 47, "lr": 9.781398598071216e-05} +{"train_loss": 0.019861876964569092, "grad_norm": 0.09975354373455048, "global_step": 5007, "epoch": 47, "lr": 9.781311913253845e-05} +{"train_loss": 0.016572225838899612, "grad_norm": 0.03937116637825966, "global_step": 5008, "epoch": 47, "lr": 9.781225211637024e-05} +{"train_loss": 0.016160443425178528, "grad_norm": 0.08200964331626892, "global_step": 5009, "epoch": 47, "lr": 9.781138493221058e-05} +{"train_loss": 0.016485821455717087, "grad_norm": 0.03989766165614128, "global_step": 5010, "epoch": 47, "lr": 9.781051758006248e-05} +{"train_loss": 0.017729712650179863, "grad_norm": 0.09230991452932358, "global_step": 5011, "epoch": 47, "lr": 9.780965005992903e-05} +{"train_loss": 0.020881421864032745, "grad_norm": 0.07797598838806152, "global_step": 5012, "epoch": 47, "lr": 9.780878237181326e-05} +{"train_loss": 0.017605585977435112, "grad_norm": 0.060437433421611786, "global_step": 5013, "epoch": 47, "lr": 9.780791451571822e-05} +{"train_loss": 0.018935732543468475, "grad_norm": 0.10978913307189941, "global_step": 5014, "epoch": 47, "lr": 9.780704649164697e-05} +{"train_loss": 0.017262406647205353, "grad_norm": 0.05822379142045975, "global_step": 5015, "epoch": 47, "lr": 9.780617829960252e-05} +{"train_loss": 0.020056428387761116, "grad_norm": 0.06477367877960205, "global_step": 5016, "epoch": 47, "lr": 9.780530993958797e-05} +{"train_loss": 0.019170250743627548, "grad_norm": 0.0726543590426445, "global_step": 5017, "epoch": 47, "lr": 9.780444141160635e-05} +{"train_loss": 0.020373329520225525, "grad_norm": 0.11647876352071762, "global_step": 5018, "epoch": 47, "lr": 9.780357271566072e-05} +{"train_loss": 0.01969616487622261, "grad_norm": 0.061652492731809616, "global_step": 5019, "epoch": 47, "lr": 9.780270385175411e-05} +{"train_loss": 0.019349507987499237, "grad_norm": 0.07037397474050522, "global_step": 5020, "epoch": 47, "lr": 9.780183481988958e-05} +{"train_loss": 0.019669894129037857, "grad_norm": 0.08824966847896576, "global_step": 5021, "epoch": 47, "lr": 9.780096562007022e-05} +{"train_loss": 0.01976999267935753, "grad_norm": 0.05984528735280037, "global_step": 5022, "epoch": 47, "lr": 9.780009625229902e-05} +{"train_loss": 0.020263206213712692, "grad_norm": 0.08419068157672882, "global_step": 5023, "epoch": 47, "lr": 9.77992267165791e-05} +{"train_loss": 0.02004935033619404, "grad_norm": 0.04296674206852913, "global_step": 5024, "epoch": 47, "lr": 9.779835701291347e-05} +{"train_loss": 0.020357780158519745, "grad_norm": 0.0391426682472229, "global_step": 5025, "epoch": 47, "lr": 9.779748714130521e-05} +{"train_loss": 0.017140299081802368, "grad_norm": 0.07762850821018219, "global_step": 5026, "epoch": 47, "lr": 9.779661710175736e-05} +{"train_loss": 0.019810693338513374, "grad_norm": 0.05996103957295418, "global_step": 5027, "epoch": 47, "lr": 9.779574689427297e-05} +{"train_loss": 0.01834496110677719, "grad_norm": 0.056984852999448776, "global_step": 5028, "epoch": 47, "lr": 9.779487651885513e-05} +{"train_loss": 0.02231934480369091, "grad_norm": 0.06293406337499619, "global_step": 5029, "epoch": 47, "lr": 9.779400597550688e-05} +{"train_loss": 0.01896640844643116, "grad_norm": 0.03555937111377716, "global_step": 5030, "epoch": 47, "lr": 9.779313526423126e-05} +{"train_loss": 0.016896791756153107, "grad_norm": 0.10432644933462143, "global_step": 5031, "epoch": 47, "lr": 9.779226438503138e-05} +{"train_loss": 0.018637264147400856, "grad_norm": 0.09594040364027023, "global_step": 5032, "epoch": 47, "lr": 9.779139333791023e-05} +{"train_loss": 0.0170274805277586, "grad_norm": 0.0396338626742363, "global_step": 5033, "epoch": 47, "lr": 9.779052212287092e-05} +{"train_loss": 0.020790915936231613, "grad_norm": 0.05477853864431381, "global_step": 5034, "epoch": 47, "lr": 9.778965073991651e-05} +{"train_loss": 0.019234674051404, "grad_norm": 0.08013356477022171, "global_step": 5035, "epoch": 47, "lr": 9.778877918905005e-05} +{"train_loss": 0.019970759749412537, "grad_norm": 0.051854413002729416, "global_step": 5036, "epoch": 47, "lr": 9.77879074702746e-05} +{"train_loss": 0.017602097243070602, "grad_norm": 0.042888522148132324, "global_step": 5037, "epoch": 47, "lr": 9.77870355835932e-05} +{"train_loss": 0.020105663686990738, "grad_norm": 0.15757212042808533, "global_step": 5038, "epoch": 47, "lr": 9.778616352900898e-05} +{"train_loss": 0.01555545348674059, "grad_norm": 0.050513602793216705, "global_step": 5039, "epoch": 47, "lr": 9.778529130652494e-05} +{"train_loss": 0.01806899905204773, "grad_norm": 0.05113457515835762, "global_step": 5040, "epoch": 47, "lr": 9.778441891614418e-05} +{"train_loss": 0.017999399453401566, "grad_norm": 0.03513383865356445, "global_step": 5041, "epoch": 47, "lr": 9.778354635786974e-05} +{"train_loss": 0.021461348980665207, "grad_norm": 0.06369844824075699, "global_step": 5042, "epoch": 47, "lr": 9.77826736317047e-05} +{"train_loss": 0.021212801337242126, "grad_norm": 0.08830565959215164, "global_step": 5043, "epoch": 47, "lr": 9.778180073765213e-05} +{"train_loss": 0.018701542168855667, "grad_norm": 0.08630512654781342, "global_step": 5044, "epoch": 47, "lr": 9.778092767571508e-05} +{"train_loss": 0.019962279126048088, "grad_norm": 0.07732092589139938, "global_step": 5045, "epoch": 47, "lr": 9.778005444589665e-05} +{"train_loss": 0.018556267023086548, "grad_norm": 0.05188950151205063, "global_step": 5046, "epoch": 47, "lr": 9.777918104819987e-05} +{"train_loss": 0.02129724994301796, "grad_norm": 0.054454974830150604, "global_step": 5047, "epoch": 47, "lr": 9.777830748262784e-05} +{"train_loss": 0.021377310156822205, "grad_norm": 0.05805637687444687, "global_step": 5048, "epoch": 47, "lr": 9.77774337491836e-05} +{"train_loss": 0.020353328436613083, "grad_norm": 0.11350934952497482, "global_step": 5049, "epoch": 47, "lr": 9.777655984787026e-05} +{"train_loss": 0.018800552934408188, "grad_norm": 0.04489927738904953, "global_step": 5050, "epoch": 47, "lr": 9.777568577869084e-05} +{"train_loss": 0.020748089998960495, "grad_norm": 0.13712754845619202, "global_step": 5051, "epoch": 47, "lr": 9.777481154164845e-05} +{"train_loss": 0.023058131337165833, "grad_norm": 0.10734701156616211, "global_step": 5052, "epoch": 47, "lr": 9.777393713674613e-05} +{"train_loss": 0.02191201224923134, "grad_norm": 0.10738426446914673, "global_step": 5053, "epoch": 47, "lr": 9.7773062563987e-05} +{"train_loss": 0.018881317228078842, "grad_norm": 0.044260717928409576, "global_step": 5054, "epoch": 47, "lr": 9.77721878233741e-05} +{"train_loss": 0.01859283819794655, "grad_norm": 0.07758677750825882, "global_step": 5055, "epoch": 47, "lr": 9.77713129149105e-05} +{"train_loss": 0.019606515765190125, "grad_norm": 0.0608198419213295, "global_step": 5056, "epoch": 47, "lr": 9.777043783859929e-05} +{"train_loss": 0.020225640386343002, "grad_norm": 0.08829326927661896, "global_step": 5057, "epoch": 47, "lr": 9.776956259444352e-05} +{"train_loss": 0.020271971821784973, "grad_norm": 0.05921579524874687, "global_step": 5058, "epoch": 47, "lr": 9.77686871824463e-05} +{"train_loss": 0.02036353200674057, "grad_norm": 0.09840376675128937, "global_step": 5059, "epoch": 47, "lr": 9.776781160261065e-05} +{"train_loss": 0.01761534810066223, "grad_norm": 0.046182163059711456, "global_step": 5060, "epoch": 47, "lr": 9.776693585493973e-05} +{"train_loss": 0.023063398897647858, "grad_norm": 0.11010231822729111, "global_step": 5061, "epoch": 47, "lr": 9.776605993943655e-05} +{"train_loss": 0.01954202726483345, "grad_norm": 0.051634885370731354, "global_step": 5062, "epoch": 47, "lr": 9.77651838561042e-05} +{"train_loss": 0.018046023324131966, "grad_norm": 0.04109548032283783, "global_step": 5063, "epoch": 47, "lr": 9.776430760494578e-05} +{"train_loss": 0.018295718356966972, "grad_norm": 0.05735878646373749, "global_step": 5064, "epoch": 47, "lr": 9.776343118596437e-05} +{"train_loss": 0.01735476404428482, "grad_norm": 0.09654678404331207, "global_step": 5065, "epoch": 47, "lr": 9.7762554599163e-05} +{"train_loss": 0.020479945465922356, "grad_norm": 0.09608829021453857, "global_step": 5066, "epoch": 47, "lr": 9.776167784454479e-05} +{"train_loss": 0.01870010793209076, "grad_norm": 0.030149755999445915, "global_step": 5067, "epoch": 47, "lr": 9.776080092211284e-05} +{"train_loss": 0.015812834724783897, "grad_norm": 0.06047138571739197, "global_step": 5068, "epoch": 47, "lr": 9.775992383187019e-05} +{"train_loss": 0.01841667667031288, "grad_norm": 0.036580607295036316, "global_step": 5069, "epoch": 47, "lr": 9.775904657381994e-05} +{"train_loss": 0.019775860011577606, "grad_norm": 0.054171718657016754, "global_step": 5070, "epoch": 47, "lr": 9.775816914796516e-05} +{"train_loss": 0.020912665873765945, "grad_norm": 0.1867404729127884, "global_step": 5071, "epoch": 47, "lr": 9.775729155430895e-05} +{"train_loss": 0.018753230571746826, "grad_norm": 0.06649361550807953, "global_step": 5072, "epoch": 47, "lr": 9.775641379285441e-05} +{"train_loss": 0.018768247216939926, "grad_norm": 0.04578901827335358, "global_step": 5073, "epoch": 47, "lr": 9.775553586360458e-05} +{"train_loss": 0.021341077983379364, "grad_norm": 0.07958205044269562, "global_step": 5074, "epoch": 47, "lr": 9.775465776656256e-05} +{"train_loss": 0.020568320527672768, "grad_norm": 0.08695543557405472, "global_step": 5075, "epoch": 47, "lr": 9.775377950173143e-05} +{"train_loss": 0.024927085265517235, "grad_norm": 0.07569108158349991, "global_step": 5076, "epoch": 47, "lr": 9.775290106911431e-05} +{"train_loss": 0.02068212814629078, "grad_norm": 0.11709168553352356, "global_step": 5077, "epoch": 47, "lr": 9.775202246871426e-05} +{"train_loss": 0.020785002037882805, "grad_norm": 0.04465920478105545, "global_step": 5078, "epoch": 47, "lr": 9.775114370053436e-05} +{"train_loss": 0.02314598858356476, "grad_norm": 0.09490326792001724, "global_step": 5079, "epoch": 47, "lr": 9.775026476457771e-05} +{"train_loss": 0.02077959105372429, "grad_norm": 0.06547466665506363, "global_step": 5080, "epoch": 47, "lr": 9.77493856608474e-05} +{"train_loss": 0.017944788560271263, "grad_norm": 0.03917952999472618, "global_step": 5081, "epoch": 47, "lr": 9.774850638934651e-05} +{"train_loss": 0.017773877829313278, "grad_norm": 0.07873114943504333, "global_step": 5082, "epoch": 47, "lr": 9.774762695007814e-05} +{"train_loss": 0.019001871347427368, "grad_norm": 0.04078608751296997, "global_step": 5083, "epoch": 47, "lr": 9.774674734304538e-05} +{"train_loss": 0.02263200469315052, "grad_norm": 0.07401454448699951, "global_step": 5084, "epoch": 47, "lr": 9.774586756825131e-05} +{"train_loss": 0.01735084131360054, "grad_norm": 0.05175265669822693, "global_step": 5085, "epoch": 47, "lr": 9.774498762569901e-05} +{"train_loss": 0.021180246025323868, "grad_norm": 0.05740991234779358, "global_step": 5086, "epoch": 47, "lr": 9.77441075153916e-05} +{"train_loss": 0.01939367832405106, "grad_norm": 0.12234766036272049, "global_step": 5087, "epoch": 47, "lr": 9.774322723733216e-05, "val_loss": 0.01958371140062809} +{"train_loss": 0.020532090216875076, "grad_norm": 0.06606116890907288, "global_step": 5088, "epoch": 48, "lr": 9.774234679152376e-05} +{"train_loss": 0.015922553837299347, "grad_norm": 0.044152967631816864, "global_step": 5089, "epoch": 48, "lr": 9.774146617796956e-05} +{"train_loss": 0.02055269479751587, "grad_norm": 0.06128721311688423, "global_step": 5090, "epoch": 48, "lr": 9.774058539667257e-05} +{"train_loss": 0.018329516053199768, "grad_norm": 0.07017257064580917, "global_step": 5091, "epoch": 48, "lr": 9.773970444763593e-05} +{"train_loss": 0.019825011491775513, "grad_norm": 0.05242234468460083, "global_step": 5092, "epoch": 48, "lr": 9.773882333086275e-05} +{"train_loss": 0.015630871057510376, "grad_norm": 0.04299231246113777, "global_step": 5093, "epoch": 48, "lr": 9.77379420463561e-05} +{"train_loss": 0.024476688355207443, "grad_norm": 0.15068501234054565, "global_step": 5094, "epoch": 48, "lr": 9.773706059411908e-05} +{"train_loss": 0.02030917815864086, "grad_norm": 0.06226212531328201, "global_step": 5095, "epoch": 48, "lr": 9.773617897415479e-05} +{"train_loss": 0.017741840332746506, "grad_norm": 0.09444138407707214, "global_step": 5096, "epoch": 48, "lr": 9.773529718646633e-05} +{"train_loss": 0.019603624939918518, "grad_norm": 0.10640255361795425, "global_step": 5097, "epoch": 48, "lr": 9.773441523105677e-05} +{"train_loss": 0.020870991051197052, "grad_norm": 0.07420306652784348, "global_step": 5098, "epoch": 48, "lr": 9.773353310792927e-05} +{"train_loss": 0.015563474968075752, "grad_norm": 0.08738569170236588, "global_step": 5099, "epoch": 48, "lr": 9.773265081708687e-05} +{"train_loss": 0.01837996020913124, "grad_norm": 0.05209391191601753, "global_step": 5100, "epoch": 48, "lr": 9.77317683585327e-05} +{"train_loss": 0.02056032232940197, "grad_norm": 0.07241269201040268, "global_step": 5101, "epoch": 48, "lr": 9.773088573226985e-05} +{"train_loss": 0.016982439905405045, "grad_norm": 0.054607853293418884, "global_step": 5102, "epoch": 48, "lr": 9.773000293830143e-05} +{"train_loss": 0.021752920001745224, "grad_norm": 0.05110412463545799, "global_step": 5103, "epoch": 48, "lr": 9.772911997663054e-05} +{"train_loss": 0.019738852977752686, "grad_norm": 0.054037608206272125, "global_step": 5104, "epoch": 48, "lr": 9.772823684726028e-05} +{"train_loss": 0.01927403174340725, "grad_norm": 0.09310150146484375, "global_step": 5105, "epoch": 48, "lr": 9.772735355019374e-05} +{"train_loss": 0.020526468753814697, "grad_norm": 0.1363431066274643, "global_step": 5106, "epoch": 48, "lr": 9.772647008543406e-05} +{"train_loss": 0.01908351294696331, "grad_norm": 0.06809616833925247, "global_step": 5107, "epoch": 48, "lr": 9.77255864529843e-05} +{"train_loss": 0.01993292011320591, "grad_norm": 0.05335535109043121, "global_step": 5108, "epoch": 48, "lr": 9.772470265284758e-05} +{"train_loss": 0.021150996908545494, "grad_norm": 0.05408085137605667, "global_step": 5109, "epoch": 48, "lr": 9.772381868502702e-05} +{"train_loss": 0.01841924339532852, "grad_norm": 0.06056598946452141, "global_step": 5110, "epoch": 48, "lr": 9.772293454952571e-05} +{"train_loss": 0.021309085190296173, "grad_norm": 0.07926245778799057, "global_step": 5111, "epoch": 48, "lr": 9.772205024634678e-05} +{"train_loss": 0.019578779116272926, "grad_norm": 0.08470296859741211, "global_step": 5112, "epoch": 48, "lr": 9.772116577549329e-05} +{"train_loss": 0.021866414695978165, "grad_norm": 0.07467547059059143, "global_step": 5113, "epoch": 48, "lr": 9.772028113696841e-05} +{"train_loss": 0.018788857385516167, "grad_norm": 0.055680010467767715, "global_step": 5114, "epoch": 48, "lr": 9.771939633077518e-05} +{"train_loss": 0.017334621399641037, "grad_norm": 0.06048131361603737, "global_step": 5115, "epoch": 48, "lr": 9.771851135691676e-05} +{"train_loss": 0.019070785492658615, "grad_norm": 0.08530401438474655, "global_step": 5116, "epoch": 48, "lr": 9.771762621539626e-05} +{"train_loss": 0.01853605918586254, "grad_norm": 0.06485350430011749, "global_step": 5117, "epoch": 48, "lr": 9.771674090621675e-05} +{"train_loss": 0.02241658978164196, "grad_norm": 0.0785093903541565, "global_step": 5118, "epoch": 48, "lr": 9.771585542938138e-05} +{"train_loss": 0.020833900198340416, "grad_norm": 0.09185785800218582, "global_step": 5119, "epoch": 48, "lr": 9.771496978489323e-05} +{"train_loss": 0.02134387195110321, "grad_norm": 0.0861092209815979, "global_step": 5120, "epoch": 48, "lr": 9.771408397275543e-05} +{"train_loss": 0.01988890767097473, "grad_norm": 0.05284174904227257, "global_step": 5121, "epoch": 48, "lr": 9.771319799297109e-05} +{"train_loss": 0.021069485694169998, "grad_norm": 0.06478048115968704, "global_step": 5122, "epoch": 48, "lr": 9.771231184554333e-05} +{"train_loss": 0.020368125289678574, "grad_norm": 0.057743750512599945, "global_step": 5123, "epoch": 48, "lr": 9.771142553047525e-05} +{"train_loss": 0.018679851666092873, "grad_norm": 0.0655556246638298, "global_step": 5124, "epoch": 48, "lr": 9.771053904776996e-05} +{"train_loss": 0.01821383461356163, "grad_norm": 0.04742204397916794, "global_step": 5125, "epoch": 48, "lr": 9.770965239743059e-05} +{"train_loss": 0.02227569930255413, "grad_norm": 0.06513000279664993, "global_step": 5126, "epoch": 48, "lr": 9.770876557946027e-05} +{"train_loss": 0.019589096307754517, "grad_norm": 0.0966886356472969, "global_step": 5127, "epoch": 48, "lr": 9.770787859386208e-05} +{"train_loss": 0.01993442326784134, "grad_norm": 0.057315342128276825, "global_step": 5128, "epoch": 48, "lr": 9.770699144063916e-05} +{"train_loss": 0.02066151425242424, "grad_norm": 0.09146855026483536, "global_step": 5129, "epoch": 48, "lr": 9.77061041197946e-05} +{"train_loss": 0.018025396391749382, "grad_norm": 0.11567603051662445, "global_step": 5130, "epoch": 48, "lr": 9.770521663133155e-05} +{"train_loss": 0.01766904443502426, "grad_norm": 0.05621171370148659, "global_step": 5131, "epoch": 48, "lr": 9.770432897525312e-05} +{"train_loss": 0.019565150141716003, "grad_norm": 0.10012057423591614, "global_step": 5132, "epoch": 48, "lr": 9.770344115156242e-05} +{"train_loss": 0.018300583586096764, "grad_norm": 0.07095121592283249, "global_step": 5133, "epoch": 48, "lr": 9.770255316026257e-05} +{"train_loss": 0.018280118703842163, "grad_norm": 0.031104091554880142, "global_step": 5134, "epoch": 48, "lr": 9.77016650013567e-05} +{"train_loss": 0.018424388021230698, "grad_norm": 0.05471155047416687, "global_step": 5135, "epoch": 48, "lr": 9.770077667484792e-05} +{"train_loss": 0.019613511860370636, "grad_norm": 0.12492096424102783, "global_step": 5136, "epoch": 48, "lr": 9.769988818073934e-05} +{"train_loss": 0.015392439439892769, "grad_norm": 0.08788035809993744, "global_step": 5137, "epoch": 48, "lr": 9.769899951903412e-05} +{"train_loss": 0.01855221763253212, "grad_norm": 0.06320787221193314, "global_step": 5138, "epoch": 48, "lr": 9.769811068973536e-05} +{"train_loss": 0.02127343788743019, "grad_norm": 0.05294272303581238, "global_step": 5139, "epoch": 48, "lr": 9.769722169284617e-05} +{"train_loss": 0.019605355337262154, "grad_norm": 0.07997755706310272, "global_step": 5140, "epoch": 48, "lr": 9.769633252836969e-05} +{"train_loss": 0.01883111521601677, "grad_norm": 0.07615139335393906, "global_step": 5141, "epoch": 48, "lr": 9.769544319630903e-05} +{"train_loss": 0.015787603333592415, "grad_norm": 0.08882302790880203, "global_step": 5142, "epoch": 48, "lr": 9.769455369666734e-05} +{"train_loss": 0.019877459853887558, "grad_norm": 0.045821662992239, "global_step": 5143, "epoch": 48, "lr": 9.769366402944772e-05} +{"train_loss": 0.017544306814670563, "grad_norm": 0.044870052486658096, "global_step": 5144, "epoch": 48, "lr": 9.769277419465332e-05} +{"train_loss": 0.018275171518325806, "grad_norm": 0.07990584522485733, "global_step": 5145, "epoch": 48, "lr": 9.769188419228723e-05} +{"train_loss": 0.019131168723106384, "grad_norm": 0.05569033324718475, "global_step": 5146, "epoch": 48, "lr": 9.769099402235262e-05} +{"train_loss": 0.020052101463079453, "grad_norm": 0.06860514730215073, "global_step": 5147, "epoch": 48, "lr": 9.769010368485258e-05} +{"train_loss": 0.019465317949652672, "grad_norm": 0.07266926765441895, "global_step": 5148, "epoch": 48, "lr": 9.768921317979027e-05} +{"train_loss": 0.017287207767367363, "grad_norm": 0.1253337562084198, "global_step": 5149, "epoch": 48, "lr": 9.768832250716879e-05} +{"train_loss": 0.01851964369416237, "grad_norm": 0.10661055892705917, "global_step": 5150, "epoch": 48, "lr": 9.76874316669913e-05} +{"train_loss": 0.018709693104028702, "grad_norm": 0.05038975551724434, "global_step": 5151, "epoch": 48, "lr": 9.76865406592609e-05} +{"train_loss": 0.022611772641539574, "grad_norm": 0.06562139838933945, "global_step": 5152, "epoch": 48, "lr": 9.768564948398075e-05} +{"train_loss": 0.018735717982053757, "grad_norm": 0.0552716962993145, "global_step": 5153, "epoch": 48, "lr": 9.768475814115396e-05} +{"train_loss": 0.022866517305374146, "grad_norm": 0.057661041617393494, "global_step": 5154, "epoch": 48, "lr": 9.768386663078366e-05} +{"train_loss": 0.019652944058179855, "grad_norm": 0.07416286319494247, "global_step": 5155, "epoch": 48, "lr": 9.7682974952873e-05} +{"train_loss": 0.019245188683271408, "grad_norm": 0.08956234902143478, "global_step": 5156, "epoch": 48, "lr": 9.768208310742511e-05} +{"train_loss": 0.02361183613538742, "grad_norm": 0.07404907792806625, "global_step": 5157, "epoch": 48, "lr": 9.768119109444309e-05} +{"train_loss": 0.01962445303797722, "grad_norm": 0.05199478194117546, "global_step": 5158, "epoch": 48, "lr": 9.768029891393011e-05} +{"train_loss": 0.020463641732931137, "grad_norm": 0.15430548787117004, "global_step": 5159, "epoch": 48, "lr": 9.767940656588931e-05} +{"train_loss": 0.019187863916158676, "grad_norm": 0.13310813903808594, "global_step": 5160, "epoch": 48, "lr": 9.76785140503238e-05} +{"train_loss": 0.0186864472925663, "grad_norm": 0.043996572494506836, "global_step": 5161, "epoch": 48, "lr": 9.767762136723672e-05} +{"train_loss": 0.01882125437259674, "grad_norm": 0.040666673332452774, "global_step": 5162, "epoch": 48, "lr": 9.767672851663124e-05} +{"train_loss": 0.01910354569554329, "grad_norm": 0.06610354036092758, "global_step": 5163, "epoch": 48, "lr": 9.767583549851044e-05} +{"train_loss": 0.018436729907989502, "grad_norm": 0.048764992505311966, "global_step": 5164, "epoch": 48, "lr": 9.76749423128775e-05} +{"train_loss": 0.02183368429541588, "grad_norm": 0.05339340493083, "global_step": 5165, "epoch": 48, "lr": 9.767404895973553e-05} +{"train_loss": 0.018892891705036163, "grad_norm": 0.056477081030607224, "global_step": 5166, "epoch": 48, "lr": 9.76731554390877e-05} +{"train_loss": 0.017321869730949402, "grad_norm": 0.05509570240974426, "global_step": 5167, "epoch": 48, "lr": 9.767226175093714e-05} +{"train_loss": 0.0198109969496727, "grad_norm": 0.04230102524161339, "global_step": 5168, "epoch": 48, "lr": 9.767136789528697e-05} +{"train_loss": 0.020107697695493698, "grad_norm": 0.05093294754624367, "global_step": 5169, "epoch": 48, "lr": 9.767047387214036e-05} +{"train_loss": 0.017504893243312836, "grad_norm": 0.07169358432292938, "global_step": 5170, "epoch": 48, "lr": 9.766957968150042e-05} +{"train_loss": 0.020344041287899017, "grad_norm": 0.07718724012374878, "global_step": 5171, "epoch": 48, "lr": 9.766868532337032e-05} +{"train_loss": 0.01836406998336315, "grad_norm": 0.07992244511842728, "global_step": 5172, "epoch": 48, "lr": 9.76677907977532e-05} +{"train_loss": 0.019631020724773407, "grad_norm": 0.07583316415548325, "global_step": 5173, "epoch": 48, "lr": 9.766689610465216e-05} +{"train_loss": 0.019759297370910645, "grad_norm": 0.12838467955589294, "global_step": 5174, "epoch": 48, "lr": 9.766600124407041e-05} +{"train_loss": 0.020004983991384506, "grad_norm": 0.04800308495759964, "global_step": 5175, "epoch": 48, "lr": 9.766510621601104e-05} +{"train_loss": 0.018253268674016, "grad_norm": 0.07065124064683914, "global_step": 5176, "epoch": 48, "lr": 9.766421102047722e-05} +{"train_loss": 0.01790209859609604, "grad_norm": 0.13714073598384857, "global_step": 5177, "epoch": 48, "lr": 9.76633156574721e-05} +{"train_loss": 0.01740005426108837, "grad_norm": 0.04207833483815193, "global_step": 5178, "epoch": 48, "lr": 9.766242012699881e-05} +{"train_loss": 0.01578683964908123, "grad_norm": 0.048073410987854004, "global_step": 5179, "epoch": 48, "lr": 9.76615244290605e-05} +{"train_loss": 0.018916606903076172, "grad_norm": 0.04231695458292961, "global_step": 5180, "epoch": 48, "lr": 9.766062856366033e-05} +{"train_loss": 0.01994422823190689, "grad_norm": 0.048244111239910126, "global_step": 5181, "epoch": 48, "lr": 9.765973253080143e-05} +{"train_loss": 0.01829412579536438, "grad_norm": 0.06276077032089233, "global_step": 5182, "epoch": 48, "lr": 9.765883633048695e-05} +{"train_loss": 0.01744125783443451, "grad_norm": 0.07380269467830658, "global_step": 5183, "epoch": 48, "lr": 9.765793996272006e-05} +{"train_loss": 0.018745452165603638, "grad_norm": 0.22579729557037354, "global_step": 5184, "epoch": 48, "lr": 9.76570434275039e-05} +{"train_loss": 0.01883307285606861, "grad_norm": 0.11205779761075974, "global_step": 5185, "epoch": 48, "lr": 9.76561467248416e-05} +{"train_loss": 0.019261375069618225, "grad_norm": 0.1110973134636879, "global_step": 5186, "epoch": 48, "lr": 9.765524985473633e-05} +{"train_loss": 0.01745503768324852, "grad_norm": 0.044452160596847534, "global_step": 5187, "epoch": 48, "lr": 9.765435281719124e-05} +{"train_loss": 0.01775992289185524, "grad_norm": 0.04913192242383957, "global_step": 5188, "epoch": 48, "lr": 9.765345561220948e-05} +{"train_loss": 0.020752444863319397, "grad_norm": 0.05612106993794441, "global_step": 5189, "epoch": 48, "lr": 9.76525582397942e-05} +{"train_loss": 0.023331232368946075, "grad_norm": 0.043137211352586746, "global_step": 5190, "epoch": 48, "lr": 9.765166069994856e-05} +{"train_loss": 0.019538866356015205, "grad_norm": 0.056161344051361084, "global_step": 5191, "epoch": 48, "lr": 9.765076299267569e-05} +{"train_loss": 0.02073066495358944, "grad_norm": 0.09256891161203384, "global_step": 5192, "epoch": 48, "lr": 9.764986511797877e-05} +{"train_loss": 0.01934016490671432, "grad_norm": 0.09092263877391815, "global_step": 5193, "epoch": 48, "lr": 9.764896707586096e-05, "val_loss": 0.01925990730524063} +{"train_loss": 0.02131488360464573, "grad_norm": 0.06303809583187103, "global_step": 5194, "epoch": 49, "lr": 9.764806886632539e-05} +{"train_loss": 0.014918288215994835, "grad_norm": 0.12688875198364258, "global_step": 5195, "epoch": 49, "lr": 9.764717048937522e-05} +{"train_loss": 0.020052410662174225, "grad_norm": 0.07537280768156052, "global_step": 5196, "epoch": 49, "lr": 9.76462719450136e-05} +{"train_loss": 0.019102156162261963, "grad_norm": 0.10894408077001572, "global_step": 5197, "epoch": 49, "lr": 9.764537323324373e-05} +{"train_loss": 0.01691780611872673, "grad_norm": 0.058804214000701904, "global_step": 5198, "epoch": 49, "lr": 9.764447435406874e-05} +{"train_loss": 0.019984101876616478, "grad_norm": 0.05999438092112541, "global_step": 5199, "epoch": 49, "lr": 9.764357530749178e-05} +{"train_loss": 0.018504362553358078, "grad_norm": 0.10836835950613022, "global_step": 5200, "epoch": 49, "lr": 9.764267609351601e-05} +{"train_loss": 0.01910681277513504, "grad_norm": 0.06257917732000351, "global_step": 5201, "epoch": 49, "lr": 9.764177671214459e-05} +{"train_loss": 0.019495464861392975, "grad_norm": 0.04251166060566902, "global_step": 5202, "epoch": 49, "lr": 9.76408771633807e-05} +{"train_loss": 0.018420033156871796, "grad_norm": 0.09859570860862732, "global_step": 5203, "epoch": 49, "lr": 9.763997744722748e-05} +{"train_loss": 0.01730048656463623, "grad_norm": 0.09231308102607727, "global_step": 5204, "epoch": 49, "lr": 9.763907756368809e-05} +{"train_loss": 0.021008722484111786, "grad_norm": 0.03586757928133011, "global_step": 5205, "epoch": 49, "lr": 9.763817751276572e-05} +{"train_loss": 0.019798951223492622, "grad_norm": 0.07073573768138885, "global_step": 5206, "epoch": 49, "lr": 9.763727729446347e-05} +{"train_loss": 0.01862391270697117, "grad_norm": 0.13288384675979614, "global_step": 5207, "epoch": 49, "lr": 9.763637690878459e-05} +{"train_loss": 0.01867079734802246, "grad_norm": 0.14464396238327026, "global_step": 5208, "epoch": 49, "lr": 9.763547635573219e-05} +{"train_loss": 0.016048524528741837, "grad_norm": 0.07487789541482925, "global_step": 5209, "epoch": 49, "lr": 9.763457563530942e-05} +{"train_loss": 0.02105427160859108, "grad_norm": 0.13233539462089539, "global_step": 5210, "epoch": 49, "lr": 9.763367474751947e-05} +{"train_loss": 0.017447903752326965, "grad_norm": 0.03994514420628548, "global_step": 5211, "epoch": 49, "lr": 9.763277369236552e-05} +{"train_loss": 0.01575428806245327, "grad_norm": 0.045217741280794144, "global_step": 5212, "epoch": 49, "lr": 9.763187246985072e-05} +{"train_loss": 0.018555250018835068, "grad_norm": 0.08992932736873627, "global_step": 5213, "epoch": 49, "lr": 9.763097107997823e-05} +{"train_loss": 0.017436467111110687, "grad_norm": 0.052507586777210236, "global_step": 5214, "epoch": 49, "lr": 9.763006952275122e-05} +{"train_loss": 0.01979469135403633, "grad_norm": 0.0699700340628624, "global_step": 5215, "epoch": 49, "lr": 9.762916779817284e-05} +{"train_loss": 0.02001553773880005, "grad_norm": 0.0482136607170105, "global_step": 5216, "epoch": 49, "lr": 9.76282659062463e-05} +{"train_loss": 0.01843617483973503, "grad_norm": 0.07120590656995773, "global_step": 5217, "epoch": 49, "lr": 9.762736384697474e-05} +{"train_loss": 0.01647159643471241, "grad_norm": 0.05574316903948784, "global_step": 5218, "epoch": 49, "lr": 9.762646162036134e-05} +{"train_loss": 0.020912431180477142, "grad_norm": 0.06111613288521767, "global_step": 5219, "epoch": 49, "lr": 9.762555922640925e-05} +{"train_loss": 0.019905539229512215, "grad_norm": 0.04017164558172226, "global_step": 5220, "epoch": 49, "lr": 9.762465666512168e-05} +{"train_loss": 0.01961451582610607, "grad_norm": 0.042935796082019806, "global_step": 5221, "epoch": 49, "lr": 9.762375393650175e-05} +{"train_loss": 0.020651210099458694, "grad_norm": 0.06635811924934387, "global_step": 5222, "epoch": 49, "lr": 9.762285104055268e-05} +{"train_loss": 0.021165305748581886, "grad_norm": 0.05300161615014076, "global_step": 5223, "epoch": 49, "lr": 9.762194797727764e-05} +{"train_loss": 0.019994208589196205, "grad_norm": 0.06352321803569794, "global_step": 5224, "epoch": 49, "lr": 9.762104474667976e-05} +{"train_loss": 0.019220396876335144, "grad_norm": 0.06305144727230072, "global_step": 5225, "epoch": 49, "lr": 9.762014134876223e-05} +{"train_loss": 0.019841762259602547, "grad_norm": 0.1280815303325653, "global_step": 5226, "epoch": 49, "lr": 9.761923778352826e-05} +{"train_loss": 0.015830151736736298, "grad_norm": 0.07604535669088364, "global_step": 5227, "epoch": 49, "lr": 9.761833405098098e-05} +{"train_loss": 0.018779169768095016, "grad_norm": 0.04950831085443497, "global_step": 5228, "epoch": 49, "lr": 9.76174301511236e-05} +{"train_loss": 0.01920459233224392, "grad_norm": 0.0676843449473381, "global_step": 5229, "epoch": 49, "lr": 9.761652608395924e-05} +{"train_loss": 0.017757870256900787, "grad_norm": 0.05660717561841011, "global_step": 5230, "epoch": 49, "lr": 9.761562184949115e-05} +{"train_loss": 0.021154187619686127, "grad_norm": 0.06229902431368828, "global_step": 5231, "epoch": 49, "lr": 9.761471744772247e-05} +{"train_loss": 0.024489471688866615, "grad_norm": 0.05650130659341812, "global_step": 5232, "epoch": 49, "lr": 9.761381287865637e-05} +{"train_loss": 0.020001811906695366, "grad_norm": 0.04661183059215546, "global_step": 5233, "epoch": 49, "lr": 9.761290814229604e-05} +{"train_loss": 0.01537657342851162, "grad_norm": 0.03353680670261383, "global_step": 5234, "epoch": 49, "lr": 9.761200323864465e-05} +{"train_loss": 0.019904883578419685, "grad_norm": 0.05755945295095444, "global_step": 5235, "epoch": 49, "lr": 9.76110981677054e-05} +{"train_loss": 0.01932891458272934, "grad_norm": 0.12167688459157944, "global_step": 5236, "epoch": 49, "lr": 9.761019292948144e-05} +{"train_loss": 0.017741061747074127, "grad_norm": 0.07555318623781204, "global_step": 5237, "epoch": 49, "lr": 9.760928752397598e-05} +{"train_loss": 0.016524706035852432, "grad_norm": 0.057931650429964066, "global_step": 5238, "epoch": 49, "lr": 9.76083819511922e-05} +{"train_loss": 0.01914578303694725, "grad_norm": 0.07588215917348862, "global_step": 5239, "epoch": 49, "lr": 9.760747621113325e-05} +{"train_loss": 0.016700811684131622, "grad_norm": 0.052347905933856964, "global_step": 5240, "epoch": 49, "lr": 9.760657030380234e-05} +{"train_loss": 0.018768485635519028, "grad_norm": 0.1003536507487297, "global_step": 5241, "epoch": 49, "lr": 9.760566422920263e-05} +{"train_loss": 0.02037736028432846, "grad_norm": 0.05880308523774147, "global_step": 5242, "epoch": 49, "lr": 9.760475798733735e-05} +{"train_loss": 0.01822632923722267, "grad_norm": 0.07706236094236374, "global_step": 5243, "epoch": 49, "lr": 9.760385157820962e-05} +{"train_loss": 0.016487905755639076, "grad_norm": 0.04223831370472908, "global_step": 5244, "epoch": 49, "lr": 9.760294500182268e-05} +{"train_loss": 0.017881665378808975, "grad_norm": 0.045176684856414795, "global_step": 5245, "epoch": 49, "lr": 9.760203825817968e-05} +{"train_loss": 0.020387688651680946, "grad_norm": 0.03733144327998161, "global_step": 5246, "epoch": 49, "lr": 9.760113134728384e-05} +{"train_loss": 0.022376492619514465, "grad_norm": 0.07027503848075867, "global_step": 5247, "epoch": 49, "lr": 9.760022426913831e-05} +{"train_loss": 0.020228900015354156, "grad_norm": 0.05202243849635124, "global_step": 5248, "epoch": 49, "lr": 9.75993170237463e-05} +{"train_loss": 0.020913556218147278, "grad_norm": 0.060949768871068954, "global_step": 5249, "epoch": 49, "lr": 9.759840961111098e-05} +{"train_loss": 0.018915001302957535, "grad_norm": 0.04476574435830116, "global_step": 5250, "epoch": 49, "lr": 9.759750203123556e-05} +{"train_loss": 0.021771706640720367, "grad_norm": 0.07464677840471268, "global_step": 5251, "epoch": 49, "lr": 9.75965942841232e-05} +{"train_loss": 0.0155210942029953, "grad_norm": 0.036164551973342896, "global_step": 5252, "epoch": 49, "lr": 9.759568636977711e-05} +{"train_loss": 0.02042674459517002, "grad_norm": 0.056838639080524445, "global_step": 5253, "epoch": 49, "lr": 9.759477828820049e-05} +{"train_loss": 0.019684160128235817, "grad_norm": 0.05679358169436455, "global_step": 5254, "epoch": 49, "lr": 9.759387003939652e-05} +{"train_loss": 0.019095562398433685, "grad_norm": 0.07153381407260895, "global_step": 5255, "epoch": 49, "lr": 9.759296162336839e-05} +{"train_loss": 0.018612219020724297, "grad_norm": 0.051797837018966675, "global_step": 5256, "epoch": 49, "lr": 9.759205304011926e-05} +{"train_loss": 0.0184246264398098, "grad_norm": 0.06018383800983429, "global_step": 5257, "epoch": 49, "lr": 9.759114428965239e-05} +{"train_loss": 0.018742620944976807, "grad_norm": 0.06982216984033585, "global_step": 5258, "epoch": 49, "lr": 9.759023537197092e-05} +{"train_loss": 0.020883679389953613, "grad_norm": 0.07920734584331512, "global_step": 5259, "epoch": 49, "lr": 9.758932628707806e-05} +{"train_loss": 0.021621625870466232, "grad_norm": 0.11058531701564789, "global_step": 5260, "epoch": 49, "lr": 9.758841703497701e-05} +{"train_loss": 0.019056441262364388, "grad_norm": 0.07596848160028458, "global_step": 5261, "epoch": 49, "lr": 9.758750761567095e-05} +{"train_loss": 0.018850186839699745, "grad_norm": 0.07762551307678223, "global_step": 5262, "epoch": 49, "lr": 9.758659802916308e-05} +{"train_loss": 0.019489509984850883, "grad_norm": 0.07649268954992294, "global_step": 5263, "epoch": 49, "lr": 9.75856882754566e-05} +{"train_loss": 0.020982874557375908, "grad_norm": 0.06858658045530319, "global_step": 5264, "epoch": 49, "lr": 9.758477835455471e-05} +{"train_loss": 0.020440489053726196, "grad_norm": 0.12069711834192276, "global_step": 5265, "epoch": 49, "lr": 9.75838682664606e-05} +{"train_loss": 0.02074391022324562, "grad_norm": 0.10421190410852432, "global_step": 5266, "epoch": 49, "lr": 9.758295801117748e-05} +{"train_loss": 0.018462881445884705, "grad_norm": 0.0696183294057846, "global_step": 5267, "epoch": 49, "lr": 9.758204758870854e-05} +{"train_loss": 0.019221141934394836, "grad_norm": 0.1289498209953308, "global_step": 5268, "epoch": 49, "lr": 9.758113699905698e-05} +{"train_loss": 0.01744088903069496, "grad_norm": 0.0992751345038414, "global_step": 5269, "epoch": 49, "lr": 9.758022624222599e-05} +{"train_loss": 0.023358847945928574, "grad_norm": 0.08802266418933868, "global_step": 5270, "epoch": 49, "lr": 9.757931531821876e-05} +{"train_loss": 0.017343174666166306, "grad_norm": 0.03676438704133034, "global_step": 5271, "epoch": 49, "lr": 9.757840422703853e-05} +{"train_loss": 0.016528282314538956, "grad_norm": 0.050459299236536026, "global_step": 5272, "epoch": 49, "lr": 9.757749296868847e-05} +{"train_loss": 0.01850949227809906, "grad_norm": 0.07195243239402771, "global_step": 5273, "epoch": 49, "lr": 9.757658154317179e-05} +{"train_loss": 0.01837274245917797, "grad_norm": 0.06726694107055664, "global_step": 5274, "epoch": 49, "lr": 9.75756699504917e-05} +{"train_loss": 0.02089395374059677, "grad_norm": 0.17448315024375916, "global_step": 5275, "epoch": 49, "lr": 9.757475819065139e-05} +{"train_loss": 0.01786171831190586, "grad_norm": 0.11744899302721024, "global_step": 5276, "epoch": 49, "lr": 9.757384626365407e-05} +{"train_loss": 0.020275644958019257, "grad_norm": 0.06088770553469658, "global_step": 5277, "epoch": 49, "lr": 9.757293416950294e-05} +{"train_loss": 0.020547989755868912, "grad_norm": 0.08935420215129852, "global_step": 5278, "epoch": 49, "lr": 9.75720219082012e-05} +{"train_loss": 0.017533425241708755, "grad_norm": 0.06829290091991425, "global_step": 5279, "epoch": 49, "lr": 9.757110947975208e-05} +{"train_loss": 0.019107306376099586, "grad_norm": 0.08020797371864319, "global_step": 5280, "epoch": 49, "lr": 9.757019688415877e-05} +{"train_loss": 0.01582631841301918, "grad_norm": 0.0363757386803627, "global_step": 5281, "epoch": 49, "lr": 9.756928412142448e-05} +{"train_loss": 0.021227717399597168, "grad_norm": 0.09299822151660919, "global_step": 5282, "epoch": 49, "lr": 9.75683711915524e-05} +{"train_loss": 0.02032172679901123, "grad_norm": 0.08038035780191422, "global_step": 5283, "epoch": 49, "lr": 9.756745809454575e-05} +{"train_loss": 0.016132179647684097, "grad_norm": 0.04712986201047897, "global_step": 5284, "epoch": 49, "lr": 9.756654483040775e-05} +{"train_loss": 0.02020067162811756, "grad_norm": 0.06187986582517624, "global_step": 5285, "epoch": 49, "lr": 9.75656313991416e-05} +{"train_loss": 0.022965332493185997, "grad_norm": 0.045720815658569336, "global_step": 5286, "epoch": 49, "lr": 9.756471780075047e-05} +{"train_loss": 0.02092830464243889, "grad_norm": 0.06737880408763885, "global_step": 5287, "epoch": 49, "lr": 9.756380403523764e-05} +{"train_loss": 0.020402561873197556, "grad_norm": 0.07755335420370102, "global_step": 5288, "epoch": 49, "lr": 9.756289010260628e-05} +{"train_loss": 0.01852821558713913, "grad_norm": 0.05058877542614937, "global_step": 5289, "epoch": 49, "lr": 9.756197600285961e-05} +{"train_loss": 0.018924251198768616, "grad_norm": 0.050309404730796814, "global_step": 5290, "epoch": 49, "lr": 9.756106173600084e-05} +{"train_loss": 0.0203157439827919, "grad_norm": 0.09206322580575943, "global_step": 5291, "epoch": 49, "lr": 9.75601473020332e-05} +{"train_loss": 0.025055402889847755, "grad_norm": 0.0820041298866272, "global_step": 5292, "epoch": 49, "lr": 9.755923270095985e-05} +{"train_loss": 0.021417278796434402, "grad_norm": 0.04368676617741585, "global_step": 5293, "epoch": 49, "lr": 9.755831793278407e-05} +{"train_loss": 0.022772353142499924, "grad_norm": 0.07338279485702515, "global_step": 5294, "epoch": 49, "lr": 9.755740299750904e-05} +{"train_loss": 0.01902764104306698, "grad_norm": 0.09293972700834274, "global_step": 5295, "epoch": 49, "lr": 9.755648789513796e-05} +{"train_loss": 0.018101446330547333, "grad_norm": 0.04856562614440918, "global_step": 5296, "epoch": 49, "lr": 9.755557262567407e-05} +{"train_loss": 0.020886268466711044, "grad_norm": 0.060258060693740845, "global_step": 5297, "epoch": 49, "lr": 9.75546571891206e-05} +{"train_loss": 0.021169690415263176, "grad_norm": 0.06254765391349792, "global_step": 5298, "epoch": 49, "lr": 9.755374158548072e-05} +{"train_loss": 0.019266815434368152, "grad_norm": 0.11980441957712173, "global_step": 5299, "epoch": 49, "lr": 9.755282581475769e-05, "val_loss": 0.019091559574007988} +{"train_loss": 0.018409166485071182, "grad_norm": 0.04704724997282028, "global_step": 5300, "epoch": 50, "lr": 9.755190987695469e-05} +{"train_loss": 0.02229115553200245, "grad_norm": 0.06192195415496826, "global_step": 5301, "epoch": 50, "lr": 9.755099377207497e-05} +{"train_loss": 0.019497517496347427, "grad_norm": 0.08142463862895966, "global_step": 5302, "epoch": 50, "lr": 9.755007750012174e-05} +{"train_loss": 0.019562670961022377, "grad_norm": 0.10051074624061584, "global_step": 5303, "epoch": 50, "lr": 9.754916106109821e-05} +{"train_loss": 0.020620698109269142, "grad_norm": 0.07209274172782898, "global_step": 5304, "epoch": 50, "lr": 9.754824445500761e-05} +{"train_loss": 0.01982506364583969, "grad_norm": 0.04490812122821808, "global_step": 5305, "epoch": 50, "lr": 9.754732768185318e-05} +{"train_loss": 0.019504232332110405, "grad_norm": 0.05764934420585632, "global_step": 5306, "epoch": 50, "lr": 9.754641074163808e-05} +{"train_loss": 0.01909099519252777, "grad_norm": 0.059627071022987366, "global_step": 5307, "epoch": 50, "lr": 9.754549363436558e-05} +{"train_loss": 0.02181997150182724, "grad_norm": 0.06845873594284058, "global_step": 5308, "epoch": 50, "lr": 9.75445763600389e-05} +{"train_loss": 0.017850983887910843, "grad_norm": 0.12544554471969604, "global_step": 5309, "epoch": 50, "lr": 9.754365891866126e-05} +{"train_loss": 0.021560942754149437, "grad_norm": 0.07640966027975082, "global_step": 5310, "epoch": 50, "lr": 9.754274131023588e-05} +{"train_loss": 0.020827826112508774, "grad_norm": 0.05771040543913841, "global_step": 5311, "epoch": 50, "lr": 9.754182353476597e-05} +{"train_loss": 0.0173978041857481, "grad_norm": 0.11364667117595673, "global_step": 5312, "epoch": 50, "lr": 9.754090559225477e-05} +{"train_loss": 0.017115574330091476, "grad_norm": 0.06290680170059204, "global_step": 5313, "epoch": 50, "lr": 9.753998748270551e-05} +{"train_loss": 0.01944010518491268, "grad_norm": 0.05221197381615639, "global_step": 5314, "epoch": 50, "lr": 9.75390692061214e-05} +{"train_loss": 0.017564471811056137, "grad_norm": 0.13235868513584137, "global_step": 5315, "epoch": 50, "lr": 9.753815076250567e-05} +{"train_loss": 0.016879193484783173, "grad_norm": 0.03945748880505562, "global_step": 5316, "epoch": 50, "lr": 9.753723215186156e-05} +{"train_loss": 0.01614869385957718, "grad_norm": 0.0735747367143631, "global_step": 5317, "epoch": 50, "lr": 9.75363133741923e-05} +{"train_loss": 0.01873847097158432, "grad_norm": 0.04155499488115311, "global_step": 5318, "epoch": 50, "lr": 9.75353944295011e-05} +{"train_loss": 0.018746918067336082, "grad_norm": 0.04770105332136154, "global_step": 5319, "epoch": 50, "lr": 9.75344753177912e-05} +{"train_loss": 0.019940001890063286, "grad_norm": 0.06249241158366203, "global_step": 5320, "epoch": 50, "lr": 9.753355603906583e-05} +{"train_loss": 0.018512777984142303, "grad_norm": 0.0774390920996666, "global_step": 5321, "epoch": 50, "lr": 9.753263659332818e-05} +{"train_loss": 0.02088896557688713, "grad_norm": 0.05549519136548042, "global_step": 5322, "epoch": 50, "lr": 9.753171698058157e-05} +{"train_loss": 0.019360553473234177, "grad_norm": 0.059719350188970566, "global_step": 5323, "epoch": 50, "lr": 9.753079720082914e-05} +{"train_loss": 0.01818789727985859, "grad_norm": 0.10666534304618835, "global_step": 5324, "epoch": 50, "lr": 9.752987725407416e-05} +{"train_loss": 0.022290976718068123, "grad_norm": 0.05899515375494957, "global_step": 5325, "epoch": 50, "lr": 9.752895714031989e-05} +{"train_loss": 0.018353287130594254, "grad_norm": 0.059878453612327576, "global_step": 5326, "epoch": 50, "lr": 9.75280368595695e-05} +{"train_loss": 0.018803833052515984, "grad_norm": 0.03487779572606087, "global_step": 5327, "epoch": 50, "lr": 9.752711641182628e-05} +{"train_loss": 0.01889791525900364, "grad_norm": 0.16382315754890442, "global_step": 5328, "epoch": 50, "lr": 9.752619579709344e-05} +{"train_loss": 0.01936652883887291, "grad_norm": 0.037082113325595856, "global_step": 5329, "epoch": 50, "lr": 9.75252750153742e-05} +{"train_loss": 0.015405267477035522, "grad_norm": 0.07944578677415848, "global_step": 5330, "epoch": 50, "lr": 9.752435406667184e-05} +{"train_loss": 0.01680327020585537, "grad_norm": 0.08881092071533203, "global_step": 5331, "epoch": 50, "lr": 9.752343295098955e-05} +{"train_loss": 0.017827734351158142, "grad_norm": 0.06260348856449127, "global_step": 5332, "epoch": 50, "lr": 9.752251166833057e-05} +{"train_loss": 0.017963092774152756, "grad_norm": 0.05926552414894104, "global_step": 5333, "epoch": 50, "lr": 9.752159021869818e-05} +{"train_loss": 0.01694362983107567, "grad_norm": 0.10237010568380356, "global_step": 5334, "epoch": 50, "lr": 9.752066860209556e-05} +{"train_loss": 0.015225195325911045, "grad_norm": 0.043465226888656616, "global_step": 5335, "epoch": 50, "lr": 9.7519746818526e-05} +{"train_loss": 0.01823301613330841, "grad_norm": 0.05910147354006767, "global_step": 5336, "epoch": 50, "lr": 9.751882486799271e-05} +{"train_loss": 0.01790892332792282, "grad_norm": 0.08578396588563919, "global_step": 5337, "epoch": 50, "lr": 9.751790275049893e-05} +{"train_loss": 0.01977650076150894, "grad_norm": 0.07708221673965454, "global_step": 5338, "epoch": 50, "lr": 9.751698046604792e-05} +{"train_loss": 0.02360275760293007, "grad_norm": 0.09811948239803314, "global_step": 5339, "epoch": 50, "lr": 9.751605801464289e-05} +{"train_loss": 0.017779728397727013, "grad_norm": 0.03708922863006592, "global_step": 5340, "epoch": 50, "lr": 9.75151353962871e-05} +{"train_loss": 0.01895594596862793, "grad_norm": 0.04855528473854065, "global_step": 5341, "epoch": 50, "lr": 9.751421261098378e-05} +{"train_loss": 0.01759466528892517, "grad_norm": 0.04833073541522026, "global_step": 5342, "epoch": 50, "lr": 9.751328965873618e-05} +{"train_loss": 0.017208198085427284, "grad_norm": 0.05783567950129509, "global_step": 5343, "epoch": 50, "lr": 9.751236653954756e-05} +{"train_loss": 0.019155273213982582, "grad_norm": 0.056088224053382874, "global_step": 5344, "epoch": 50, "lr": 9.751144325342115e-05} +{"train_loss": 0.020653096958994865, "grad_norm": 0.08170028030872345, "global_step": 5345, "epoch": 50, "lr": 9.751051980036016e-05} +{"train_loss": 0.01845390722155571, "grad_norm": 0.05001615360379219, "global_step": 5346, "epoch": 50, "lr": 9.750959618036788e-05} +{"train_loss": 0.020442871376872063, "grad_norm": 0.04425458237528801, "global_step": 5347, "epoch": 50, "lr": 9.750867239344754e-05} +{"train_loss": 0.017162511125206947, "grad_norm": 0.04752257838845253, "global_step": 5348, "epoch": 50, "lr": 9.75077484396024e-05} +{"train_loss": 0.0205145925283432, "grad_norm": 0.11584684252738953, "global_step": 5349, "epoch": 50, "lr": 9.750682431883568e-05} +{"train_loss": 0.018711812794208527, "grad_norm": 0.06037197634577751, "global_step": 5350, "epoch": 50, "lr": 9.750590003115064e-05} +{"train_loss": 0.02127465046942234, "grad_norm": 0.0653463676571846, "global_step": 5351, "epoch": 50, "lr": 9.750497557655053e-05} +{"train_loss": 0.017740588635206223, "grad_norm": 0.051491886377334595, "global_step": 5352, "epoch": 50, "lr": 9.750405095503859e-05} +{"train_loss": 0.019634071737527847, "grad_norm": 0.04369219392538071, "global_step": 5353, "epoch": 50, "lr": 9.750312616661808e-05} +{"train_loss": 0.017752595245838165, "grad_norm": 0.06977296620607376, "global_step": 5354, "epoch": 50, "lr": 9.750220121129223e-05} +{"train_loss": 0.01770145073533058, "grad_norm": 0.057979486882686615, "global_step": 5355, "epoch": 50, "lr": 9.750127608906432e-05} +{"train_loss": 0.01883118972182274, "grad_norm": 0.05892065912485123, "global_step": 5356, "epoch": 50, "lr": 9.750035079993757e-05} +{"train_loss": 0.01799086667597294, "grad_norm": 0.06502529233694077, "global_step": 5357, "epoch": 50, "lr": 9.749942534391526e-05} +{"train_loss": 0.019409045577049255, "grad_norm": 0.04580426588654518, "global_step": 5358, "epoch": 50, "lr": 9.749849972100063e-05} +{"train_loss": 0.01936332881450653, "grad_norm": 0.06831417977809906, "global_step": 5359, "epoch": 50, "lr": 9.749757393119691e-05} +{"train_loss": 0.021665498614311218, "grad_norm": 0.11526241898536682, "global_step": 5360, "epoch": 50, "lr": 9.749664797450737e-05} +{"train_loss": 0.01996587961912155, "grad_norm": 0.06263037770986557, "global_step": 5361, "epoch": 50, "lr": 9.749572185093527e-05} +{"train_loss": 0.01648915559053421, "grad_norm": 0.06338945776224136, "global_step": 5362, "epoch": 50, "lr": 9.749479556048386e-05} +{"train_loss": 0.020204246044158936, "grad_norm": 0.06865178048610687, "global_step": 5363, "epoch": 50, "lr": 9.749386910315641e-05} +{"train_loss": 0.01959027536213398, "grad_norm": 0.08267362415790558, "global_step": 5364, "epoch": 50, "lr": 9.749294247895613e-05} +{"train_loss": 0.021667417138814926, "grad_norm": 0.05248458683490753, "global_step": 5365, "epoch": 50, "lr": 9.749201568788633e-05} +{"train_loss": 0.02166716568171978, "grad_norm": 0.04494927451014519, "global_step": 5366, "epoch": 50, "lr": 9.749108872995023e-05} +{"train_loss": 0.02118917927145958, "grad_norm": 0.045880429446697235, "global_step": 5367, "epoch": 50, "lr": 9.749016160515109e-05} +{"train_loss": 0.01897434890270233, "grad_norm": 0.06889042258262634, "global_step": 5368, "epoch": 50, "lr": 9.748923431349217e-05} +{"train_loss": 0.023602211847901344, "grad_norm": 0.10712557286024094, "global_step": 5369, "epoch": 50, "lr": 9.748830685497676e-05} +{"train_loss": 0.018828004598617554, "grad_norm": 0.05419012904167175, "global_step": 5370, "epoch": 50, "lr": 9.748737922960807e-05} +{"train_loss": 0.019214050844311714, "grad_norm": 0.09794656187295914, "global_step": 5371, "epoch": 50, "lr": 9.748645143738939e-05} +{"train_loss": 0.019926365464925766, "grad_norm": 0.05545886978507042, "global_step": 5372, "epoch": 50, "lr": 9.748552347832395e-05} +{"train_loss": 0.01896955631673336, "grad_norm": 0.05059077963232994, "global_step": 5373, "epoch": 50, "lr": 9.748459535241506e-05} +{"train_loss": 0.019711419939994812, "grad_norm": 0.04350106418132782, "global_step": 5374, "epoch": 50, "lr": 9.748366705966594e-05} +{"train_loss": 0.018176382407546043, "grad_norm": 0.05181783437728882, "global_step": 5375, "epoch": 50, "lr": 9.748273860007987e-05} +{"train_loss": 0.01887108013033867, "grad_norm": 0.07661959528923035, "global_step": 5376, "epoch": 50, "lr": 9.74818099736601e-05} +{"train_loss": 0.019130494445562363, "grad_norm": 0.09915821999311447, "global_step": 5377, "epoch": 50, "lr": 9.74808811804099e-05} +{"train_loss": 0.0198037251830101, "grad_norm": 0.08327027410268784, "global_step": 5378, "epoch": 50, "lr": 9.747995222033254e-05} +{"train_loss": 0.021567517891526222, "grad_norm": 0.13649626076221466, "global_step": 5379, "epoch": 50, "lr": 9.747902309343127e-05} +{"train_loss": 0.018089475110173225, "grad_norm": 0.0423918142914772, "global_step": 5380, "epoch": 50, "lr": 9.747809379970936e-05} +{"train_loss": 0.017549799755215645, "grad_norm": 0.11471287161111832, "global_step": 5381, "epoch": 50, "lr": 9.747716433917007e-05} +{"train_loss": 0.01976368948817253, "grad_norm": 0.11522872745990753, "global_step": 5382, "epoch": 50, "lr": 9.747623471181669e-05} +{"train_loss": 0.018816597759723663, "grad_norm": 0.04826843738555908, "global_step": 5383, "epoch": 50, "lr": 9.747530491765244e-05} +{"train_loss": 0.017445098608732224, "grad_norm": 0.03498051315546036, "global_step": 5384, "epoch": 50, "lr": 9.747437495668064e-05} +{"train_loss": 0.019420048221945763, "grad_norm": 0.06303694099187851, "global_step": 5385, "epoch": 50, "lr": 9.747344482890453e-05} +{"train_loss": 0.02242147922515869, "grad_norm": 0.08542323857545853, "global_step": 5386, "epoch": 50, "lr": 9.747251453432738e-05} +{"train_loss": 0.017723118886351585, "grad_norm": 0.06072070449590683, "global_step": 5387, "epoch": 50, "lr": 9.747158407295245e-05} +{"train_loss": 0.01930215209722519, "grad_norm": 0.058134812861680984, "global_step": 5388, "epoch": 50, "lr": 9.747065344478301e-05} +{"train_loss": 0.01856832392513752, "grad_norm": 0.042314134538173676, "global_step": 5389, "epoch": 50, "lr": 9.746972264982236e-05} +{"train_loss": 0.020416632294654846, "grad_norm": 0.05063045024871826, "global_step": 5390, "epoch": 50, "lr": 9.746879168807374e-05} +{"train_loss": 0.018014784902334213, "grad_norm": 0.06192445009946823, "global_step": 5391, "epoch": 50, "lr": 9.746786055954042e-05} +{"train_loss": 0.016720611602067947, "grad_norm": 0.059946510940790176, "global_step": 5392, "epoch": 50, "lr": 9.746692926422569e-05} +{"train_loss": 0.019329313188791275, "grad_norm": 0.04259965196251869, "global_step": 5393, "epoch": 50, "lr": 9.746599780213282e-05} +{"train_loss": 0.02190312370657921, "grad_norm": 0.06130364164710045, "global_step": 5394, "epoch": 50, "lr": 9.746506617326507e-05} +{"train_loss": 0.018306642770767212, "grad_norm": 0.05424802005290985, "global_step": 5395, "epoch": 50, "lr": 9.746413437762571e-05} +{"train_loss": 0.01901955157518387, "grad_norm": 0.08165300637483597, "global_step": 5396, "epoch": 50, "lr": 9.746320241521805e-05} +{"train_loss": 0.022914156317710876, "grad_norm": 0.08039464056491852, "global_step": 5397, "epoch": 50, "lr": 9.74622702860453e-05} +{"train_loss": 0.018695402890443802, "grad_norm": 0.05766250193119049, "global_step": 5398, "epoch": 50, "lr": 9.746133799011081e-05} +{"train_loss": 0.021363556385040283, "grad_norm": 0.040725111961364746, "global_step": 5399, "epoch": 50, "lr": 9.74604055274178e-05} +{"train_loss": 0.018374990671873093, "grad_norm": 0.09211322665214539, "global_step": 5400, "epoch": 50, "lr": 9.745947289796957e-05} +{"train_loss": 0.01971173658967018, "grad_norm": 0.07953701913356781, "global_step": 5401, "epoch": 50, "lr": 9.745854010176938e-05} +{"train_loss": 0.017087440937757492, "grad_norm": 0.057347673922777176, "global_step": 5402, "epoch": 50, "lr": 9.745760713882053e-05} +{"train_loss": 0.0202246755361557, "grad_norm": 0.05243725702166557, "global_step": 5403, "epoch": 50, "lr": 9.745667400912629e-05} +{"train_loss": 0.023387659341096878, "grad_norm": 0.0458054393529892, "global_step": 5404, "epoch": 50, "lr": 9.745574071268992e-05} +{"train_loss": 0.01921471741647934, "grad_norm": 0.08521045744419098, "global_step": 5405, "epoch": 50, "lr": 9.745480724951473e-05, "train_batch_mse_error": 0.017448045313358307, "val_loss": 0.019660722464323044} +{"train_loss": 0.0181538425385952, "grad_norm": 0.04635307192802429, "global_step": 5406, "epoch": 51, "lr": 9.745387361960399e-05} +{"train_loss": 0.019849304109811783, "grad_norm": 0.044907763600349426, "global_step": 5407, "epoch": 51, "lr": 9.745293982296095e-05} +{"train_loss": 0.0193844772875309, "grad_norm": 0.055997565388679504, "global_step": 5408, "epoch": 51, "lr": 9.745200585958895e-05} +{"train_loss": 0.019433042034506798, "grad_norm": 0.08182326704263687, "global_step": 5409, "epoch": 51, "lr": 9.745107172949122e-05} +{"train_loss": 0.020600732415914536, "grad_norm": 0.09224904328584671, "global_step": 5410, "epoch": 51, "lr": 9.745013743267106e-05} +{"train_loss": 0.018887773156166077, "grad_norm": 0.0900159701704979, "global_step": 5411, "epoch": 51, "lr": 9.744920296913174e-05} +{"train_loss": 0.018097516149282455, "grad_norm": 0.04794104024767876, "global_step": 5412, "epoch": 51, "lr": 9.744826833887657e-05} +{"train_loss": 0.018575100228190422, "grad_norm": 0.05157582834362984, "global_step": 5413, "epoch": 51, "lr": 9.744733354190883e-05} +{"train_loss": 0.019406750798225403, "grad_norm": 0.05033323913812637, "global_step": 5414, "epoch": 51, "lr": 9.744639857823178e-05} +{"train_loss": 0.019434891641139984, "grad_norm": 0.05771442502737045, "global_step": 5415, "epoch": 51, "lr": 9.744546344784873e-05} +{"train_loss": 0.016274530440568924, "grad_norm": 0.06757813692092896, "global_step": 5416, "epoch": 51, "lr": 9.744452815076294e-05} +{"train_loss": 0.020077422261238098, "grad_norm": 0.0815819799900055, "global_step": 5417, "epoch": 51, "lr": 9.744359268697771e-05} +{"train_loss": 0.01634906977415085, "grad_norm": 0.14908620715141296, "global_step": 5418, "epoch": 51, "lr": 9.744265705649634e-05} +{"train_loss": 0.01646924577653408, "grad_norm": 0.04807096719741821, "global_step": 5419, "epoch": 51, "lr": 9.74417212593221e-05} +{"train_loss": 0.01973569765686989, "grad_norm": 0.0720042809844017, "global_step": 5420, "epoch": 51, "lr": 9.744078529545827e-05} +{"train_loss": 0.01704932190477848, "grad_norm": 0.04196786507964134, "global_step": 5421, "epoch": 51, "lr": 9.743984916490818e-05} +{"train_loss": 0.01701296493411064, "grad_norm": 0.037625718861818314, "global_step": 5422, "epoch": 51, "lr": 9.743891286767509e-05} +{"train_loss": 0.02121860906481743, "grad_norm": 0.10888595134019852, "global_step": 5423, "epoch": 51, "lr": 9.743797640376227e-05} +{"train_loss": 0.0184114221483469, "grad_norm": 0.04790353775024414, "global_step": 5424, "epoch": 51, "lr": 9.743703977317305e-05} +{"train_loss": 0.019551653414964676, "grad_norm": 0.0691513940691948, "global_step": 5425, "epoch": 51, "lr": 9.743610297591068e-05} +{"train_loss": 0.018119417130947113, "grad_norm": 0.05928615853190422, "global_step": 5426, "epoch": 51, "lr": 9.743516601197849e-05} +{"train_loss": 0.022406775504350662, "grad_norm": 0.04502944275736809, "global_step": 5427, "epoch": 51, "lr": 9.743422888137974e-05} +{"train_loss": 0.020092105492949486, "grad_norm": 0.06483332812786102, "global_step": 5428, "epoch": 51, "lr": 9.743329158411776e-05} +{"train_loss": 0.019399624317884445, "grad_norm": 0.07609006017446518, "global_step": 5429, "epoch": 51, "lr": 9.743235412019581e-05} +{"train_loss": 0.019118964672088623, "grad_norm": 0.03439735993742943, "global_step": 5430, "epoch": 51, "lr": 9.74314164896172e-05} +{"train_loss": 0.020349852740764618, "grad_norm": 0.055921200662851334, "global_step": 5431, "epoch": 51, "lr": 9.743047869238522e-05} +{"train_loss": 0.01960628852248192, "grad_norm": 0.13810555636882782, "global_step": 5432, "epoch": 51, "lr": 9.742954072850316e-05} +{"train_loss": 0.02029828540980816, "grad_norm": 0.0650775358080864, "global_step": 5433, "epoch": 51, "lr": 9.742860259797432e-05} +{"train_loss": 0.01702570728957653, "grad_norm": 0.07530898600816727, "global_step": 5434, "epoch": 51, "lr": 9.742766430080199e-05} +{"train_loss": 0.018394026905298233, "grad_norm": 0.03791850805282593, "global_step": 5435, "epoch": 51, "lr": 9.742672583698948e-05} +{"train_loss": 0.019922848790884018, "grad_norm": 0.055148106068372726, "global_step": 5436, "epoch": 51, "lr": 9.742578720654009e-05} +{"train_loss": 0.01782344840466976, "grad_norm": 0.12047448754310608, "global_step": 5437, "epoch": 51, "lr": 9.742484840945709e-05} +{"train_loss": 0.020264318212866783, "grad_norm": 0.06887854635715485, "global_step": 5438, "epoch": 51, "lr": 9.742390944574382e-05} +{"train_loss": 0.018466033041477203, "grad_norm": 0.049173448234796524, "global_step": 5439, "epoch": 51, "lr": 9.742297031540354e-05} +{"train_loss": 0.017830220982432365, "grad_norm": 0.0374906025826931, "global_step": 5440, "epoch": 51, "lr": 9.742203101843957e-05} +{"train_loss": 0.01936880685389042, "grad_norm": 0.041976604610681534, "global_step": 5441, "epoch": 51, "lr": 9.742109155485521e-05} +{"train_loss": 0.019830022007226944, "grad_norm": 0.0828951969742775, "global_step": 5442, "epoch": 51, "lr": 9.742015192465375e-05} +{"train_loss": 0.020938334986567497, "grad_norm": 0.06125658005475998, "global_step": 5443, "epoch": 51, "lr": 9.741921212783851e-05} +{"train_loss": 0.01902160421013832, "grad_norm": 0.04704632610082626, "global_step": 5444, "epoch": 51, "lr": 9.741827216441277e-05} +{"train_loss": 0.017527662217617035, "grad_norm": 0.06820657849311829, "global_step": 5445, "epoch": 51, "lr": 9.741733203437984e-05} +{"train_loss": 0.017802461981773376, "grad_norm": 0.040474653244018555, "global_step": 5446, "epoch": 51, "lr": 9.741639173774304e-05} +{"train_loss": 0.0205649696290493, "grad_norm": 0.05021196976304054, "global_step": 5447, "epoch": 51, "lr": 9.741545127450564e-05} +{"train_loss": 0.019297733902931213, "grad_norm": 0.04924215376377106, "global_step": 5448, "epoch": 51, "lr": 9.741451064467098e-05} +{"train_loss": 0.01715315878391266, "grad_norm": 0.112430140376091, "global_step": 5449, "epoch": 51, "lr": 9.741356984824235e-05} +{"train_loss": 0.020266246050596237, "grad_norm": 0.08290916681289673, "global_step": 5450, "epoch": 51, "lr": 9.741262888522305e-05} +{"train_loss": 0.019429771229624748, "grad_norm": 0.09556373208761215, "global_step": 5451, "epoch": 51, "lr": 9.74116877556164e-05} +{"train_loss": 0.02064797654747963, "grad_norm": 0.06202556937932968, "global_step": 5452, "epoch": 51, "lr": 9.741074645942569e-05} +{"train_loss": 0.01860426366329193, "grad_norm": 0.06454335153102875, "global_step": 5453, "epoch": 51, "lr": 9.740980499665424e-05} +{"train_loss": 0.02181859314441681, "grad_norm": 0.06988780200481415, "global_step": 5454, "epoch": 51, "lr": 9.740886336730534e-05} +{"train_loss": 0.016738228499889374, "grad_norm": 0.09170626103878021, "global_step": 5455, "epoch": 51, "lr": 9.740792157138234e-05} +{"train_loss": 0.02089869976043701, "grad_norm": 0.07376204431056976, "global_step": 5456, "epoch": 51, "lr": 9.74069796088885e-05} +{"train_loss": 0.022585207596421242, "grad_norm": 0.09619389474391937, "global_step": 5457, "epoch": 51, "lr": 9.740603747982715e-05} +{"train_loss": 0.021506164222955704, "grad_norm": 0.0511227585375309, "global_step": 5458, "epoch": 51, "lr": 9.74050951842016e-05} +{"train_loss": 0.01719951257109642, "grad_norm": 0.047526147216558456, "global_step": 5459, "epoch": 51, "lr": 9.740415272201517e-05} +{"train_loss": 0.015546243637800217, "grad_norm": 0.03473161160945892, "global_step": 5460, "epoch": 51, "lr": 9.740321009327115e-05} +{"train_loss": 0.018936049193143845, "grad_norm": 0.04574728384613991, "global_step": 5461, "epoch": 51, "lr": 9.740226729797289e-05} +{"train_loss": 0.0212286077439785, "grad_norm": 0.04756509140133858, "global_step": 5462, "epoch": 51, "lr": 9.740132433612366e-05} +{"train_loss": 0.020101597532629967, "grad_norm": 0.1300811916589737, "global_step": 5463, "epoch": 51, "lr": 9.740038120772678e-05} +{"train_loss": 0.022076716646552086, "grad_norm": 0.1632262021303177, "global_step": 5464, "epoch": 51, "lr": 9.739943791278559e-05} +{"train_loss": 0.018875068053603172, "grad_norm": 0.04899308830499649, "global_step": 5465, "epoch": 51, "lr": 9.739849445130338e-05} +{"train_loss": 0.021132657304406166, "grad_norm": 0.1131996214389801, "global_step": 5466, "epoch": 51, "lr": 9.739755082328347e-05} +{"train_loss": 0.02113485522568226, "grad_norm": 0.06963244080543518, "global_step": 5467, "epoch": 51, "lr": 9.739660702872919e-05} +{"train_loss": 0.015698911622166634, "grad_norm": 0.04406336322426796, "global_step": 5468, "epoch": 51, "lr": 9.739566306764385e-05} +{"train_loss": 0.017547260969877243, "grad_norm": 0.06705302000045776, "global_step": 5469, "epoch": 51, "lr": 9.739471894003075e-05} +{"train_loss": 0.02029762789607048, "grad_norm": 0.04409060627222061, "global_step": 5470, "epoch": 51, "lr": 9.739377464589323e-05} +{"train_loss": 0.017767546698451042, "grad_norm": 0.100678950548172, "global_step": 5471, "epoch": 51, "lr": 9.739283018523458e-05} +{"train_loss": 0.019164469093084335, "grad_norm": 0.07542269676923752, "global_step": 5472, "epoch": 51, "lr": 9.739188555805816e-05} +{"train_loss": 0.01790536381304264, "grad_norm": 0.057796310633420944, "global_step": 5473, "epoch": 51, "lr": 9.739094076436723e-05} +{"train_loss": 0.017476214095950127, "grad_norm": 0.04363303259015083, "global_step": 5474, "epoch": 51, "lr": 9.738999580416517e-05} +{"train_loss": 0.018059872090816498, "grad_norm": 0.06046787276864052, "global_step": 5475, "epoch": 51, "lr": 9.738905067745527e-05} +{"train_loss": 0.021959759294986725, "grad_norm": 0.06280580908060074, "global_step": 5476, "epoch": 51, "lr": 9.738810538424085e-05} +{"train_loss": 0.018330886960029602, "grad_norm": 0.045609891414642334, "global_step": 5477, "epoch": 51, "lr": 9.738715992452524e-05} +{"train_loss": 0.01652093045413494, "grad_norm": 0.13584886491298676, "global_step": 5478, "epoch": 51, "lr": 9.738621429831177e-05} +{"train_loss": 0.020129937678575516, "grad_norm": 0.05931205675005913, "global_step": 5479, "epoch": 51, "lr": 9.738526850560373e-05} +{"train_loss": 0.020384255796670914, "grad_norm": 0.06748080253601074, "global_step": 5480, "epoch": 51, "lr": 9.738432254640446e-05} +{"train_loss": 0.01760556921362877, "grad_norm": 0.05728720873594284, "global_step": 5481, "epoch": 51, "lr": 9.738337642071731e-05} +{"train_loss": 0.017604751512408257, "grad_norm": 0.131373792886734, "global_step": 5482, "epoch": 51, "lr": 9.738243012854557e-05} +{"train_loss": 0.02185158245265484, "grad_norm": 0.09025729447603226, "global_step": 5483, "epoch": 51, "lr": 9.738148366989257e-05} +{"train_loss": 0.01885409653186798, "grad_norm": 0.07190974801778793, "global_step": 5484, "epoch": 51, "lr": 9.738053704476165e-05} +{"train_loss": 0.02269297093153, "grad_norm": 0.06615281850099564, "global_step": 5485, "epoch": 51, "lr": 9.737959025315613e-05} +{"train_loss": 0.0201921034604311, "grad_norm": 0.0659077912569046, "global_step": 5486, "epoch": 51, "lr": 9.737864329507933e-05} +{"train_loss": 0.02014956995844841, "grad_norm": 0.0455506332218647, "global_step": 5487, "epoch": 51, "lr": 9.737769617053458e-05} +{"train_loss": 0.019017118960618973, "grad_norm": 0.11724870651960373, "global_step": 5488, "epoch": 51, "lr": 9.737674887952522e-05} +{"train_loss": 0.018587786704301834, "grad_norm": 0.11426886171102524, "global_step": 5489, "epoch": 51, "lr": 9.737580142205456e-05} +{"train_loss": 0.021216394379734993, "grad_norm": 0.13282132148742676, "global_step": 5490, "epoch": 51, "lr": 9.737485379812594e-05} +{"train_loss": 0.020910628139972687, "grad_norm": 0.062188174575567245, "global_step": 5491, "epoch": 51, "lr": 9.737390600774268e-05} +{"train_loss": 0.020088031888008118, "grad_norm": 0.06931555271148682, "global_step": 5492, "epoch": 51, "lr": 9.737295805090813e-05} +{"train_loss": 0.02033364027738571, "grad_norm": 0.08789709210395813, "global_step": 5493, "epoch": 51, "lr": 9.737200992762559e-05} +{"train_loss": 0.0155267845839262, "grad_norm": 0.06722845882177353, "global_step": 5494, "epoch": 51, "lr": 9.737106163789843e-05} +{"train_loss": 0.018986068665981293, "grad_norm": 0.050540607422590256, "global_step": 5495, "epoch": 51, "lr": 9.737011318172994e-05} +{"train_loss": 0.01923324540257454, "grad_norm": 0.08830292522907257, "global_step": 5496, "epoch": 51, "lr": 9.736916455912348e-05} +{"train_loss": 0.018915273249149323, "grad_norm": 0.07529759407043457, "global_step": 5497, "epoch": 51, "lr": 9.736821577008238e-05} +{"train_loss": 0.021482348442077637, "grad_norm": 0.12355645000934601, "global_step": 5498, "epoch": 51, "lr": 9.736726681460998e-05} +{"train_loss": 0.018997807055711746, "grad_norm": 0.055111393332481384, "global_step": 5499, "epoch": 51, "lr": 9.736631769270957e-05} +{"train_loss": 0.020615071058273315, "grad_norm": 0.11014042794704437, "global_step": 5500, "epoch": 51, "lr": 9.736536840438455e-05} +{"train_loss": 0.020499806851148605, "grad_norm": 0.05711815878748894, "global_step": 5501, "epoch": 51, "lr": 9.736441894963822e-05} +{"train_loss": 0.021709300577640533, "grad_norm": 0.09531643986701965, "global_step": 5502, "epoch": 51, "lr": 9.73634693284739e-05} +{"train_loss": 0.01684633269906044, "grad_norm": 0.05698102340102196, "global_step": 5503, "epoch": 51, "lr": 9.736251954089496e-05} +{"train_loss": 0.0198289155960083, "grad_norm": 0.0869651660323143, "global_step": 5504, "epoch": 51, "lr": 9.736156958690472e-05} +{"train_loss": 0.01855124719440937, "grad_norm": 0.0765913799405098, "global_step": 5505, "epoch": 51, "lr": 9.736061946650654e-05} +{"train_loss": 0.018007131293416023, "grad_norm": 0.04784611612558365, "global_step": 5506, "epoch": 51, "lr": 9.735966917970372e-05} +{"train_loss": 0.0158180333673954, "grad_norm": 0.03626801073551178, "global_step": 5507, "epoch": 51, "lr": 9.735871872649962e-05} +{"train_loss": 0.01834065094590187, "grad_norm": 0.0577748604118824, "global_step": 5508, "epoch": 51, "lr": 9.735776810689758e-05} +{"train_loss": 0.016300402581691742, "grad_norm": 0.062244515866041183, "global_step": 5509, "epoch": 51, "lr": 9.735681732090094e-05} +{"train_loss": 0.01712387055158615, "grad_norm": 0.06782589107751846, "global_step": 5510, "epoch": 51, "lr": 9.735586636851304e-05} +{"train_loss": 0.01917304543180848, "grad_norm": 0.08102760463953018, "global_step": 5511, "epoch": 51, "lr": 9.735491524973722e-05, "val_loss": 0.018966306000947952} +{"train_loss": 0.017897384241223335, "grad_norm": 0.04391728341579437, "global_step": 5512, "epoch": 52, "lr": 9.735396396457682e-05} +{"train_loss": 0.01789606548845768, "grad_norm": 0.05941465497016907, "global_step": 5513, "epoch": 52, "lr": 9.735301251303519e-05} +{"train_loss": 0.01969139464199543, "grad_norm": 0.043730203062295914, "global_step": 5514, "epoch": 52, "lr": 9.735206089511566e-05} +{"train_loss": 0.018861979246139526, "grad_norm": 0.0807400494813919, "global_step": 5515, "epoch": 52, "lr": 9.735110911082158e-05} +{"train_loss": 0.019303500652313232, "grad_norm": 0.06866324692964554, "global_step": 5516, "epoch": 52, "lr": 9.73501571601563e-05} +{"train_loss": 0.018513452261686325, "grad_norm": 0.05519097298383713, "global_step": 5517, "epoch": 52, "lr": 9.734920504312317e-05} +{"train_loss": 0.021085543558001518, "grad_norm": 0.08428268879652023, "global_step": 5518, "epoch": 52, "lr": 9.73482527597255e-05} +{"train_loss": 0.01997009851038456, "grad_norm": 0.046293556690216064, "global_step": 5519, "epoch": 52, "lr": 9.734730030996669e-05} +{"train_loss": 0.017258185893297195, "grad_norm": 0.05819564685225487, "global_step": 5520, "epoch": 52, "lr": 9.734634769385003e-05} +{"train_loss": 0.01929062232375145, "grad_norm": 0.05869263410568237, "global_step": 5521, "epoch": 52, "lr": 9.73453949113789e-05} +{"train_loss": 0.017861254513263702, "grad_norm": 0.05980081856250763, "global_step": 5522, "epoch": 52, "lr": 9.734444196255665e-05} +{"train_loss": 0.01525837555527687, "grad_norm": 0.03301987797021866, "global_step": 5523, "epoch": 52, "lr": 9.734348884738662e-05} +{"train_loss": 0.01699334755539894, "grad_norm": 0.06817609816789627, "global_step": 5524, "epoch": 52, "lr": 9.734253556587216e-05} +{"train_loss": 0.016996469348669052, "grad_norm": 0.09040768444538116, "global_step": 5525, "epoch": 52, "lr": 9.734158211801663e-05} +{"train_loss": 0.01848558709025383, "grad_norm": 0.09425133466720581, "global_step": 5526, "epoch": 52, "lr": 9.734062850382335e-05} +{"train_loss": 0.023678751662373543, "grad_norm": 0.059534382075071335, "global_step": 5527, "epoch": 52, "lr": 9.733967472329569e-05} +{"train_loss": 0.02008635550737381, "grad_norm": 0.08684906363487244, "global_step": 5528, "epoch": 52, "lr": 9.733872077643701e-05} +{"train_loss": 0.018228109925985336, "grad_norm": 0.07539215683937073, "global_step": 5529, "epoch": 52, "lr": 9.733776666325065e-05} +{"train_loss": 0.018485713750123978, "grad_norm": 0.06365851312875748, "global_step": 5530, "epoch": 52, "lr": 9.733681238373995e-05} +{"train_loss": 0.01860463246703148, "grad_norm": 0.051042769104242325, "global_step": 5531, "epoch": 52, "lr": 9.733585793790829e-05} +{"train_loss": 0.01565542072057724, "grad_norm": 0.04678111895918846, "global_step": 5532, "epoch": 52, "lr": 9.733490332575902e-05} +{"train_loss": 0.019817233085632324, "grad_norm": 0.053254127502441406, "global_step": 5533, "epoch": 52, "lr": 9.733394854729548e-05} +{"train_loss": 0.017231080681085587, "grad_norm": 0.05059674754738808, "global_step": 5534, "epoch": 52, "lr": 9.733299360252102e-05} +{"train_loss": 0.019312676042318344, "grad_norm": 0.057117871940135956, "global_step": 5535, "epoch": 52, "lr": 9.733203849143901e-05} +{"train_loss": 0.01835746318101883, "grad_norm": 0.09531227499246597, "global_step": 5536, "epoch": 52, "lr": 9.733108321405282e-05} +{"train_loss": 0.017479009926319122, "grad_norm": 0.05255131423473358, "global_step": 5537, "epoch": 52, "lr": 9.733012777036576e-05} +{"train_loss": 0.018792221322655678, "grad_norm": 0.06523865461349487, "global_step": 5538, "epoch": 52, "lr": 9.732917216038122e-05} +{"train_loss": 0.022923409938812256, "grad_norm": 0.052801523357629776, "global_step": 5539, "epoch": 52, "lr": 9.732821638410256e-05} +{"train_loss": 0.018098704516887665, "grad_norm": 0.07531414180994034, "global_step": 5540, "epoch": 52, "lr": 9.732726044153312e-05} +{"train_loss": 0.01663762703537941, "grad_norm": 0.04588913917541504, "global_step": 5541, "epoch": 52, "lr": 9.732630433267627e-05} +{"train_loss": 0.02066454291343689, "grad_norm": 0.09063004702329636, "global_step": 5542, "epoch": 52, "lr": 9.732534805753538e-05} +{"train_loss": 0.019731678068637848, "grad_norm": 0.07975508272647858, "global_step": 5543, "epoch": 52, "lr": 9.73243916161138e-05} +{"train_loss": 0.01786588504910469, "grad_norm": 0.05134342983365059, "global_step": 5544, "epoch": 52, "lr": 9.732343500841487e-05} +{"train_loss": 0.017029235139489174, "grad_norm": 0.05600232258439064, "global_step": 5545, "epoch": 52, "lr": 9.732247823444198e-05} +{"train_loss": 0.021772602573037148, "grad_norm": 0.10508034378290176, "global_step": 5546, "epoch": 52, "lr": 9.732152129419848e-05} +{"train_loss": 0.020169224590063095, "grad_norm": 0.045662932097911835, "global_step": 5547, "epoch": 52, "lr": 9.732056418768774e-05} +{"train_loss": 0.018055066466331482, "grad_norm": 0.12401458621025085, "global_step": 5548, "epoch": 52, "lr": 9.731960691491311e-05} +{"train_loss": 0.02170727029442787, "grad_norm": 0.09521714597940445, "global_step": 5549, "epoch": 52, "lr": 9.731864947587796e-05} +{"train_loss": 0.019253764301538467, "grad_norm": 0.0861755758523941, "global_step": 5550, "epoch": 52, "lr": 9.731769187058566e-05} +{"train_loss": 0.01642078533768654, "grad_norm": 0.05706994608044624, "global_step": 5551, "epoch": 52, "lr": 9.731673409903956e-05} +{"train_loss": 0.019778171554207802, "grad_norm": 0.04959392920136452, "global_step": 5552, "epoch": 52, "lr": 9.731577616124304e-05} +{"train_loss": 0.0209316685795784, "grad_norm": 0.08502685278654099, "global_step": 5553, "epoch": 52, "lr": 9.731481805719947e-05} +{"train_loss": 0.020430011674761772, "grad_norm": 0.09672446548938751, "global_step": 5554, "epoch": 52, "lr": 9.73138597869122e-05} +{"train_loss": 0.020539620891213417, "grad_norm": 0.04465137794613838, "global_step": 5555, "epoch": 52, "lr": 9.731290135038458e-05} +{"train_loss": 0.018811151385307312, "grad_norm": 0.0632103830575943, "global_step": 5556, "epoch": 52, "lr": 9.731194274762e-05} +{"train_loss": 0.019249381497502327, "grad_norm": 0.0926641896367073, "global_step": 5557, "epoch": 52, "lr": 9.731098397862186e-05} +{"train_loss": 0.01623779535293579, "grad_norm": 0.04600159823894501, "global_step": 5558, "epoch": 52, "lr": 9.731002504339348e-05} +{"train_loss": 0.02153095044195652, "grad_norm": 0.045689262449741364, "global_step": 5559, "epoch": 52, "lr": 9.730906594193824e-05} +{"train_loss": 0.01714976131916046, "grad_norm": 0.09044233709573746, "global_step": 5560, "epoch": 52, "lr": 9.730810667425954e-05} +{"train_loss": 0.017273113131523132, "grad_norm": 0.11551510542631149, "global_step": 5561, "epoch": 52, "lr": 9.730714724036071e-05} +{"train_loss": 0.0197373628616333, "grad_norm": 0.05272124335169792, "global_step": 5562, "epoch": 52, "lr": 9.730618764024515e-05} +{"train_loss": 0.020379208028316498, "grad_norm": 0.037569429725408554, "global_step": 5563, "epoch": 52, "lr": 9.73052278739162e-05} +{"train_loss": 0.020939147099852562, "grad_norm": 0.04631701484322548, "global_step": 5564, "epoch": 52, "lr": 9.730426794137727e-05} +{"train_loss": 0.017742719501256943, "grad_norm": 0.10198306292295456, "global_step": 5565, "epoch": 52, "lr": 9.73033078426317e-05} +{"train_loss": 0.02008136361837387, "grad_norm": 0.04163925722241402, "global_step": 5566, "epoch": 52, "lr": 9.730234757768289e-05} +{"train_loss": 0.021958569064736366, "grad_norm": 0.06944218277931213, "global_step": 5567, "epoch": 52, "lr": 9.73013871465342e-05} +{"train_loss": 0.018315263092517853, "grad_norm": 0.05536092817783356, "global_step": 5568, "epoch": 52, "lr": 9.7300426549189e-05} +{"train_loss": 0.017239587381482124, "grad_norm": 0.06631835550069809, "global_step": 5569, "epoch": 52, "lr": 9.729946578565068e-05} +{"train_loss": 0.02186707779765129, "grad_norm": 0.06862858682870865, "global_step": 5570, "epoch": 52, "lr": 9.72985048559226e-05} +{"train_loss": 0.016030577942728996, "grad_norm": 0.05101378634572029, "global_step": 5571, "epoch": 52, "lr": 9.729754376000814e-05} +{"train_loss": 0.01845378801226616, "grad_norm": 0.04306786507368088, "global_step": 5572, "epoch": 52, "lr": 9.729658249791069e-05} +{"train_loss": 0.02016475424170494, "grad_norm": 0.06320242583751678, "global_step": 5573, "epoch": 52, "lr": 9.729562106963361e-05} +{"train_loss": 0.022164516150951385, "grad_norm": 0.051658257842063904, "global_step": 5574, "epoch": 52, "lr": 9.72946594751803e-05} +{"train_loss": 0.017425958067178726, "grad_norm": 0.06825288385152817, "global_step": 5575, "epoch": 52, "lr": 9.72936977145541e-05} +{"train_loss": 0.016835961490869522, "grad_norm": 0.0719015821814537, "global_step": 5576, "epoch": 52, "lr": 9.729273578775844e-05} +{"train_loss": 0.016634875908493996, "grad_norm": 0.06458039581775665, "global_step": 5577, "epoch": 52, "lr": 9.729177369479665e-05} +{"train_loss": 0.017199818044900894, "grad_norm": 0.13315658271312714, "global_step": 5578, "epoch": 52, "lr": 9.729081143567215e-05} +{"train_loss": 0.01803569123148918, "grad_norm": 0.04751673340797424, "global_step": 5579, "epoch": 52, "lr": 9.72898490103883e-05} +{"train_loss": 0.021598178893327713, "grad_norm": 0.04351335018873215, "global_step": 5580, "epoch": 52, "lr": 9.728888641894849e-05} +{"train_loss": 0.01994885504245758, "grad_norm": 0.12038125842809677, "global_step": 5581, "epoch": 52, "lr": 9.728792366135609e-05} +{"train_loss": 0.01894945278763771, "grad_norm": 0.07529709488153458, "global_step": 5582, "epoch": 52, "lr": 9.72869607376145e-05} +{"train_loss": 0.0205125343054533, "grad_norm": 0.15591493248939514, "global_step": 5583, "epoch": 52, "lr": 9.728599764772709e-05} +{"train_loss": 0.021626442670822144, "grad_norm": 0.07732747495174408, "global_step": 5584, "epoch": 52, "lr": 9.728503439169725e-05} +{"train_loss": 0.01865791343152523, "grad_norm": 0.11248365789651871, "global_step": 5585, "epoch": 52, "lr": 9.728407096952836e-05} +{"train_loss": 0.02068173512816429, "grad_norm": 0.06996501237154007, "global_step": 5586, "epoch": 52, "lr": 9.728310738122381e-05} +{"train_loss": 0.016098499298095703, "grad_norm": 0.041085418313741684, "global_step": 5587, "epoch": 52, "lr": 9.728214362678699e-05} +{"train_loss": 0.020146910101175308, "grad_norm": 0.07782606780529022, "global_step": 5588, "epoch": 52, "lr": 9.728117970622126e-05} +{"train_loss": 0.019766857847571373, "grad_norm": 0.05236120894551277, "global_step": 5589, "epoch": 52, "lr": 9.728021561953005e-05} +{"train_loss": 0.018436703830957413, "grad_norm": 0.04400397092103958, "global_step": 5590, "epoch": 52, "lr": 9.727925136671672e-05} +{"train_loss": 0.01960095763206482, "grad_norm": 0.06149657070636749, "global_step": 5591, "epoch": 52, "lr": 9.727828694778465e-05} +{"train_loss": 0.017350034788250923, "grad_norm": 0.053991712629795074, "global_step": 5592, "epoch": 52, "lr": 9.727732236273725e-05} +{"train_loss": 0.019844001159071922, "grad_norm": 0.042358219623565674, "global_step": 5593, "epoch": 52, "lr": 9.727635761157789e-05} +{"train_loss": 0.021886292845010757, "grad_norm": 0.10206055641174316, "global_step": 5594, "epoch": 52, "lr": 9.727539269430998e-05} +{"train_loss": 0.016851652413606644, "grad_norm": 0.07347778230905533, "global_step": 5595, "epoch": 52, "lr": 9.72744276109369e-05} +{"train_loss": 0.018928684294223785, "grad_norm": 0.05529823154211044, "global_step": 5596, "epoch": 52, "lr": 9.727346236146203e-05} +{"train_loss": 0.02008115127682686, "grad_norm": 0.05909145250916481, "global_step": 5597, "epoch": 52, "lr": 9.727249694588878e-05} +{"train_loss": 0.0170147605240345, "grad_norm": 0.044820647686719894, "global_step": 5598, "epoch": 52, "lr": 9.727153136422053e-05} +{"train_loss": 0.01908775605261326, "grad_norm": 0.10272464901208878, "global_step": 5599, "epoch": 52, "lr": 9.727056561646066e-05} +{"train_loss": 0.02306874468922615, "grad_norm": 0.08240950107574463, "global_step": 5600, "epoch": 52, "lr": 9.72695997026126e-05} +{"train_loss": 0.02046727016568184, "grad_norm": 0.055244967341423035, "global_step": 5601, "epoch": 52, "lr": 9.726863362267973e-05} +{"train_loss": 0.017484799027442932, "grad_norm": 0.05085720494389534, "global_step": 5602, "epoch": 52, "lr": 9.726766737666542e-05} +{"train_loss": 0.022758539766073227, "grad_norm": 0.14112070202827454, "global_step": 5603, "epoch": 52, "lr": 9.72667009645731e-05} +{"train_loss": 0.01888241246342659, "grad_norm": 0.09076668322086334, "global_step": 5604, "epoch": 52, "lr": 9.726573438640614e-05} +{"train_loss": 0.019970398396253586, "grad_norm": 0.11307114362716675, "global_step": 5605, "epoch": 52, "lr": 9.726476764216794e-05} +{"train_loss": 0.018282562494277954, "grad_norm": 0.04542531073093414, "global_step": 5606, "epoch": 52, "lr": 9.726380073186193e-05} +{"train_loss": 0.019213244318962097, "grad_norm": 0.13592961430549622, "global_step": 5607, "epoch": 52, "lr": 9.726283365549143e-05} +{"train_loss": 0.019637364894151688, "grad_norm": 0.051760438829660416, "global_step": 5608, "epoch": 52, "lr": 9.726186641305992e-05} +{"train_loss": 0.017552491277456284, "grad_norm": 0.05842963978648186, "global_step": 5609, "epoch": 52, "lr": 9.726089900457077e-05} +{"train_loss": 0.018733281642198563, "grad_norm": 0.04601140692830086, "global_step": 5610, "epoch": 52, "lr": 9.725993143002736e-05} +{"train_loss": 0.018776841461658478, "grad_norm": 0.04371950030326843, "global_step": 5611, "epoch": 52, "lr": 9.72589636894331e-05} +{"train_loss": 0.019782062619924545, "grad_norm": 0.1068195104598999, "global_step": 5612, "epoch": 52, "lr": 9.72579957827914e-05} +{"train_loss": 0.017679065465927124, "grad_norm": 0.07888076454401016, "global_step": 5613, "epoch": 52, "lr": 9.725702771010565e-05} +{"train_loss": 0.021451059728860855, "grad_norm": 0.1296459585428238, "global_step": 5614, "epoch": 52, "lr": 9.725605947137926e-05} +{"train_loss": 0.021069280803203583, "grad_norm": 0.07902831584215164, "global_step": 5615, "epoch": 52, "lr": 9.725509106661562e-05} +{"train_loss": 0.020651333034038544, "grad_norm": 0.07404948025941849, "global_step": 5616, "epoch": 52, "lr": 9.725412249581815e-05} +{"train_loss": 0.01912253086437594, "grad_norm": 0.0790371373295784, "global_step": 5617, "epoch": 52, "lr": 9.725315375899024e-05, "val_loss": 0.01937042362987995} +{"train_loss": 0.01915358006954193, "grad_norm": 0.03657930716872215, "global_step": 5618, "epoch": 53, "lr": 9.72521848561353e-05} +{"train_loss": 0.020135963335633278, "grad_norm": 0.039918240159749985, "global_step": 5619, "epoch": 53, "lr": 9.725121578725674e-05} +{"train_loss": 0.018550649285316467, "grad_norm": 0.08160994201898575, "global_step": 5620, "epoch": 53, "lr": 9.725024655235795e-05} +{"train_loss": 0.015840478241443634, "grad_norm": 0.03568285331130028, "global_step": 5621, "epoch": 53, "lr": 9.724927715144234e-05} +{"train_loss": 0.019155971705913544, "grad_norm": 0.06376581639051437, "global_step": 5622, "epoch": 53, "lr": 9.724830758451332e-05} +{"train_loss": 0.022072235122323036, "grad_norm": 0.05044049024581909, "global_step": 5623, "epoch": 53, "lr": 9.72473378515743e-05} +{"train_loss": 0.01751689612865448, "grad_norm": 0.062186501920223236, "global_step": 5624, "epoch": 53, "lr": 9.724636795262868e-05} +{"train_loss": 0.021117787808179855, "grad_norm": 0.1212683767080307, "global_step": 5625, "epoch": 53, "lr": 9.724539788767986e-05} +{"train_loss": 0.01951412856578827, "grad_norm": 0.1334187239408493, "global_step": 5626, "epoch": 53, "lr": 9.724442765673126e-05} +{"train_loss": 0.0178922638297081, "grad_norm": 0.05755645036697388, "global_step": 5627, "epoch": 53, "lr": 9.72434572597863e-05} +{"train_loss": 0.018285365775227547, "grad_norm": 0.0743604451417923, "global_step": 5628, "epoch": 53, "lr": 9.724248669684837e-05} +{"train_loss": 0.020603494718670845, "grad_norm": 0.07410501688718796, "global_step": 5629, "epoch": 53, "lr": 9.724151596792088e-05} +{"train_loss": 0.017050713300704956, "grad_norm": 0.1288260668516159, "global_step": 5630, "epoch": 53, "lr": 9.724054507300725e-05} +{"train_loss": 0.018482308834791183, "grad_norm": 0.04559149965643883, "global_step": 5631, "epoch": 53, "lr": 9.72395740121109e-05} +{"train_loss": 0.01921696402132511, "grad_norm": 0.0727643147110939, "global_step": 5632, "epoch": 53, "lr": 9.723860278523521e-05} +{"train_loss": 0.021648060530424118, "grad_norm": 0.05468958243727684, "global_step": 5633, "epoch": 53, "lr": 9.723763139238364e-05} +{"train_loss": 0.018933534622192383, "grad_norm": 0.058333415538072586, "global_step": 5634, "epoch": 53, "lr": 9.723665983355955e-05} +{"train_loss": 0.015732552856206894, "grad_norm": 0.05211437866091728, "global_step": 5635, "epoch": 53, "lr": 9.72356881087664e-05} +{"train_loss": 0.01826251484453678, "grad_norm": 0.03812882676720619, "global_step": 5636, "epoch": 53, "lr": 9.723471621800757e-05} +{"train_loss": 0.016715720295906067, "grad_norm": 0.07703552395105362, "global_step": 5637, "epoch": 53, "lr": 9.72337441612865e-05} +{"train_loss": 0.021095281466841698, "grad_norm": 0.11768697947263718, "global_step": 5638, "epoch": 53, "lr": 9.723277193860658e-05} +{"train_loss": 0.019528448581695557, "grad_norm": 0.10380508750677109, "global_step": 5639, "epoch": 53, "lr": 9.723179954997125e-05} +{"train_loss": 0.01925233006477356, "grad_norm": 0.06371384114027023, "global_step": 5640, "epoch": 53, "lr": 9.723082699538393e-05} +{"train_loss": 0.018827494233846664, "grad_norm": 0.11888684332370758, "global_step": 5641, "epoch": 53, "lr": 9.7229854274848e-05} +{"train_loss": 0.019088566303253174, "grad_norm": 0.06005572900176048, "global_step": 5642, "epoch": 53, "lr": 9.722888138836693e-05} +{"train_loss": 0.01490158773958683, "grad_norm": 0.05601396784186363, "global_step": 5643, "epoch": 53, "lr": 9.722790833594408e-05} +{"train_loss": 0.02018408477306366, "grad_norm": 0.12183254212141037, "global_step": 5644, "epoch": 53, "lr": 9.722693511758292e-05} +{"train_loss": 0.017781447619199753, "grad_norm": 0.09324566274881363, "global_step": 5645, "epoch": 53, "lr": 9.722596173328684e-05} +{"train_loss": 0.017318228259682655, "grad_norm": 0.06444285809993744, "global_step": 5646, "epoch": 53, "lr": 9.722498818305927e-05} +{"train_loss": 0.020010866224765778, "grad_norm": 0.04057511314749718, "global_step": 5647, "epoch": 53, "lr": 9.722401446690363e-05} +{"train_loss": 0.021094724535942078, "grad_norm": 0.046246349811553955, "global_step": 5648, "epoch": 53, "lr": 9.722304058482334e-05} +{"train_loss": 0.017039600759744644, "grad_norm": 0.04351923614740372, "global_step": 5649, "epoch": 53, "lr": 9.722206653682183e-05} +{"train_loss": 0.020088233053684235, "grad_norm": 0.06358054280281067, "global_step": 5650, "epoch": 53, "lr": 9.722109232290249e-05} +{"train_loss": 0.019053416326642036, "grad_norm": 0.040409285575151443, "global_step": 5651, "epoch": 53, "lr": 9.722011794306878e-05} +{"train_loss": 0.020495988428592682, "grad_norm": 0.09894797950983047, "global_step": 5652, "epoch": 53, "lr": 9.721914339732412e-05} +{"train_loss": 0.017665646970272064, "grad_norm": 0.065547414124012, "global_step": 5653, "epoch": 53, "lr": 9.721816868567193e-05} +{"train_loss": 0.016824116930365562, "grad_norm": 0.05605421960353851, "global_step": 5654, "epoch": 53, "lr": 9.721719380811561e-05} +{"train_loss": 0.016667256131768227, "grad_norm": 0.04423098266124725, "global_step": 5655, "epoch": 53, "lr": 9.721621876465862e-05} +{"train_loss": 0.016765138134360313, "grad_norm": 0.06336236745119095, "global_step": 5656, "epoch": 53, "lr": 9.721524355530435e-05} +{"train_loss": 0.018784012645483017, "grad_norm": 0.06544245779514313, "global_step": 5657, "epoch": 53, "lr": 9.721426818005628e-05} +{"train_loss": 0.01890987530350685, "grad_norm": 0.06375382095575333, "global_step": 5658, "epoch": 53, "lr": 9.721329263891778e-05} +{"train_loss": 0.022331256419420242, "grad_norm": 0.09039614349603653, "global_step": 5659, "epoch": 53, "lr": 9.721231693189231e-05} +{"train_loss": 0.020622191950678825, "grad_norm": 0.13384701311588287, "global_step": 5660, "epoch": 53, "lr": 9.721134105898329e-05} +{"train_loss": 0.01864374428987503, "grad_norm": 0.129500612616539, "global_step": 5661, "epoch": 53, "lr": 9.721036502019416e-05} +{"train_loss": 0.01676538772881031, "grad_norm": 0.042895495891571045, "global_step": 5662, "epoch": 53, "lr": 9.720938881552833e-05} +{"train_loss": 0.019306205213069916, "grad_norm": 0.054717160761356354, "global_step": 5663, "epoch": 53, "lr": 9.720841244498922e-05} +{"train_loss": 0.01667271926999092, "grad_norm": 0.045837461948394775, "global_step": 5664, "epoch": 53, "lr": 9.72074359085803e-05} +{"train_loss": 0.017661895602941513, "grad_norm": 0.06053756922483444, "global_step": 5665, "epoch": 53, "lr": 9.720645920630498e-05} +{"train_loss": 0.020215846598148346, "grad_norm": 0.06418151408433914, "global_step": 5666, "epoch": 53, "lr": 9.720548233816668e-05} +{"train_loss": 0.019568759948015213, "grad_norm": 0.051488541066646576, "global_step": 5667, "epoch": 53, "lr": 9.720450530416885e-05} +{"train_loss": 0.021974440664052963, "grad_norm": 0.13342484831809998, "global_step": 5668, "epoch": 53, "lr": 9.720352810431493e-05} +{"train_loss": 0.0179462693631649, "grad_norm": 0.0681440681219101, "global_step": 5669, "epoch": 53, "lr": 9.720255073860833e-05} +{"train_loss": 0.017218954861164093, "grad_norm": 0.038751792162656784, "global_step": 5670, "epoch": 53, "lr": 9.72015732070525e-05} +{"train_loss": 0.017257271334528923, "grad_norm": 0.09592118859291077, "global_step": 5671, "epoch": 53, "lr": 9.720059550965086e-05} +{"train_loss": 0.021482836455106735, "grad_norm": 0.09074898809194565, "global_step": 5672, "epoch": 53, "lr": 9.719961764640685e-05} +{"train_loss": 0.017988581210374832, "grad_norm": 0.05340507999062538, "global_step": 5673, "epoch": 53, "lr": 9.719863961732392e-05} +{"train_loss": 0.01879248209297657, "grad_norm": 0.05724955350160599, "global_step": 5674, "epoch": 53, "lr": 9.719766142240549e-05} +{"train_loss": 0.01998990960419178, "grad_norm": 0.04212931916117668, "global_step": 5675, "epoch": 53, "lr": 9.719668306165502e-05} +{"train_loss": 0.01862672157585621, "grad_norm": 0.060642387717962265, "global_step": 5676, "epoch": 53, "lr": 9.719570453507592e-05} +{"train_loss": 0.01997760683298111, "grad_norm": 0.1370193064212799, "global_step": 5677, "epoch": 53, "lr": 9.719472584267163e-05} +{"train_loss": 0.02053101733326912, "grad_norm": 0.057656195014715195, "global_step": 5678, "epoch": 53, "lr": 9.719374698444561e-05} +{"train_loss": 0.018630720674991608, "grad_norm": 0.06776872277259827, "global_step": 5679, "epoch": 53, "lr": 9.719276796040127e-05} +{"train_loss": 0.019343670457601547, "grad_norm": 0.09923461824655533, "global_step": 5680, "epoch": 53, "lr": 9.719178877054208e-05} +{"train_loss": 0.019477633759379387, "grad_norm": 0.07429034262895584, "global_step": 5681, "epoch": 53, "lr": 9.719080941487148e-05} +{"train_loss": 0.01667175069451332, "grad_norm": 0.08639921247959137, "global_step": 5682, "epoch": 53, "lr": 9.718982989339288e-05} +{"train_loss": 0.017371544614434242, "grad_norm": 0.05257846787571907, "global_step": 5683, "epoch": 53, "lr": 9.718885020610974e-05} +{"train_loss": 0.019779935479164124, "grad_norm": 0.1118515357375145, "global_step": 5684, "epoch": 53, "lr": 9.718787035302551e-05} +{"train_loss": 0.020055223256349564, "grad_norm": 0.04228256270289421, "global_step": 5685, "epoch": 53, "lr": 9.718689033414364e-05} +{"train_loss": 0.01857307180762291, "grad_norm": 0.07727770507335663, "global_step": 5686, "epoch": 53, "lr": 9.718591014946752e-05} +{"train_loss": 0.01894809678196907, "grad_norm": 0.04761635512113571, "global_step": 5687, "epoch": 53, "lr": 9.718492979900067e-05} +{"train_loss": 0.018254883587360382, "grad_norm": 0.06397942453622818, "global_step": 5688, "epoch": 53, "lr": 9.718394928274648e-05} +{"train_loss": 0.016788311302661896, "grad_norm": 0.06362737715244293, "global_step": 5689, "epoch": 53, "lr": 9.718296860070841e-05} +{"train_loss": 0.021026182919740677, "grad_norm": 0.11662691831588745, "global_step": 5690, "epoch": 53, "lr": 9.718198775288991e-05} +{"train_loss": 0.02026364766061306, "grad_norm": 0.05311315506696701, "global_step": 5691, "epoch": 53, "lr": 9.718100673929443e-05} +{"train_loss": 0.019770152866840363, "grad_norm": 0.10649578273296356, "global_step": 5692, "epoch": 53, "lr": 9.718002555992541e-05} +{"train_loss": 0.017886994406580925, "grad_norm": 0.04946229234337807, "global_step": 5693, "epoch": 53, "lr": 9.717904421478628e-05} +{"train_loss": 0.01923072524368763, "grad_norm": 0.053980253636837006, "global_step": 5694, "epoch": 53, "lr": 9.717806270388052e-05} +{"train_loss": 0.017835300415754318, "grad_norm": 0.05173151567578316, "global_step": 5695, "epoch": 53, "lr": 9.717708102721157e-05} +{"train_loss": 0.01998974196612835, "grad_norm": 0.1053357720375061, "global_step": 5696, "epoch": 53, "lr": 9.717609918478286e-05} +{"train_loss": 0.021522281691432, "grad_norm": 0.05733579397201538, "global_step": 5697, "epoch": 53, "lr": 9.717511717659787e-05} +{"train_loss": 0.020021503791213036, "grad_norm": 0.11535181850194931, "global_step": 5698, "epoch": 53, "lr": 9.717413500266001e-05} +{"train_loss": 0.018206439912319183, "grad_norm": 0.07304514199495316, "global_step": 5699, "epoch": 53, "lr": 9.717315266297277e-05} +{"train_loss": 0.01944442093372345, "grad_norm": 0.11715510487556458, "global_step": 5700, "epoch": 53, "lr": 9.71721701575396e-05} +{"train_loss": 0.020394636318087578, "grad_norm": 0.07987091690301895, "global_step": 5701, "epoch": 53, "lr": 9.717118748636393e-05} +{"train_loss": 0.01902918703854084, "grad_norm": 0.08529023081064224, "global_step": 5702, "epoch": 53, "lr": 9.717020464944921e-05} +{"train_loss": 0.016827747225761414, "grad_norm": 0.04322179779410362, "global_step": 5703, "epoch": 53, "lr": 9.716922164679891e-05} +{"train_loss": 0.021023232489824295, "grad_norm": 0.05590255185961723, "global_step": 5704, "epoch": 53, "lr": 9.71682384784165e-05} +{"train_loss": 0.02290089800953865, "grad_norm": 0.04797283187508583, "global_step": 5705, "epoch": 53, "lr": 9.716725514430538e-05} +{"train_loss": 0.018770014867186546, "grad_norm": 0.0494714230298996, "global_step": 5706, "epoch": 53, "lr": 9.716627164446906e-05} +{"train_loss": 0.015864670276641846, "grad_norm": 0.04313823953270912, "global_step": 5707, "epoch": 53, "lr": 9.716528797891096e-05} +{"train_loss": 0.022638941183686256, "grad_norm": 0.05057789012789726, "global_step": 5708, "epoch": 53, "lr": 9.716430414763456e-05} +{"train_loss": 0.026822678744792938, "grad_norm": 0.08013888448476791, "global_step": 5709, "epoch": 53, "lr": 9.71633201506433e-05} +{"train_loss": 0.017972592264413834, "grad_norm": 0.08385339379310608, "global_step": 5710, "epoch": 53, "lr": 9.716233598794065e-05} +{"train_loss": 0.019779939204454422, "grad_norm": 0.05054435878992081, "global_step": 5711, "epoch": 53, "lr": 9.716135165953005e-05} +{"train_loss": 0.020353101193904877, "grad_norm": 0.0972064658999443, "global_step": 5712, "epoch": 53, "lr": 9.716036716541497e-05} +{"train_loss": 0.019970525056123734, "grad_norm": 0.04874725639820099, "global_step": 5713, "epoch": 53, "lr": 9.715938250559889e-05} +{"train_loss": 0.01995827443897724, "grad_norm": 0.05510871857404709, "global_step": 5714, "epoch": 53, "lr": 9.715839768008524e-05} +{"train_loss": 0.017958303913474083, "grad_norm": 0.07443038374185562, "global_step": 5715, "epoch": 53, "lr": 9.715741268887747e-05} +{"train_loss": 0.017725247889757156, "grad_norm": 0.047248657792806625, "global_step": 5716, "epoch": 53, "lr": 9.715642753197909e-05} +{"train_loss": 0.019927702844142914, "grad_norm": 0.12829241156578064, "global_step": 5717, "epoch": 53, "lr": 9.715544220939351e-05} +{"train_loss": 0.019067496061325073, "grad_norm": 0.0869106575846672, "global_step": 5718, "epoch": 53, "lr": 9.715445672112422e-05} +{"train_loss": 0.017647817730903625, "grad_norm": 0.04980034753680229, "global_step": 5719, "epoch": 53, "lr": 9.715347106717467e-05} +{"train_loss": 0.022675639018416405, "grad_norm": 0.06665030121803284, "global_step": 5720, "epoch": 53, "lr": 9.715248524754834e-05} +{"train_loss": 0.01777833327651024, "grad_norm": 0.060502853244543076, "global_step": 5721, "epoch": 53, "lr": 9.715149926224867e-05} +{"train_loss": 0.019391212612390518, "grad_norm": 0.0557413175702095, "global_step": 5722, "epoch": 53, "lr": 9.715051311127915e-05} +{"train_loss": 0.019069671894441236, "grad_norm": 0.051925793290138245, "global_step": 5723, "epoch": 53, "lr": 9.714952679464323e-05, "val_loss": 0.01889149844646454} +{"train_loss": 0.017448417842388153, "grad_norm": 0.07521219551563263, "global_step": 5724, "epoch": 54, "lr": 9.714854031234437e-05} +{"train_loss": 0.019131729379296303, "grad_norm": 0.08268827944993973, "global_step": 5725, "epoch": 54, "lr": 9.714755366438606e-05} +{"train_loss": 0.01987975463271141, "grad_norm": 0.0736490860581398, "global_step": 5726, "epoch": 54, "lr": 9.714656685077172e-05} +{"train_loss": 0.019698765128850937, "grad_norm": 0.11906842142343521, "global_step": 5727, "epoch": 54, "lr": 9.714557987150488e-05} +{"train_loss": 0.02055598422884941, "grad_norm": 0.07297408580780029, "global_step": 5728, "epoch": 54, "lr": 9.714459272658895e-05} +{"train_loss": 0.017145445570349693, "grad_norm": 0.063054159283638, "global_step": 5729, "epoch": 54, "lr": 9.714360541602745e-05} +{"train_loss": 0.017597230151295662, "grad_norm": 0.05956123769283295, "global_step": 5730, "epoch": 54, "lr": 9.71426179398238e-05} +{"train_loss": 0.017386095598340034, "grad_norm": 0.07135012745857239, "global_step": 5731, "epoch": 54, "lr": 9.714163029798152e-05} +{"train_loss": 0.018678849563002586, "grad_norm": 0.06630104780197144, "global_step": 5732, "epoch": 54, "lr": 9.714064249050402e-05} +{"train_loss": 0.017922772094607353, "grad_norm": 0.05558227375149727, "global_step": 5733, "epoch": 54, "lr": 9.713965451739482e-05} +{"train_loss": 0.017576150596141815, "grad_norm": 0.04649040848016739, "global_step": 5734, "epoch": 54, "lr": 9.713866637865736e-05} +{"train_loss": 0.019428638741374016, "grad_norm": 0.07233530282974243, "global_step": 5735, "epoch": 54, "lr": 9.713767807429515e-05} +{"train_loss": 0.018199292942881584, "grad_norm": 0.04687352851033211, "global_step": 5736, "epoch": 54, "lr": 9.713668960431162e-05} +{"train_loss": 0.021301044151186943, "grad_norm": 0.06647317111492157, "global_step": 5737, "epoch": 54, "lr": 9.713570096871028e-05} +{"train_loss": 0.02184738591313362, "grad_norm": 0.1010855883359909, "global_step": 5738, "epoch": 54, "lr": 9.713471216749457e-05} +{"train_loss": 0.01945525035262108, "grad_norm": 0.12081439048051834, "global_step": 5739, "epoch": 54, "lr": 9.713372320066798e-05} +{"train_loss": 0.01927030086517334, "grad_norm": 0.14020107686519623, "global_step": 5740, "epoch": 54, "lr": 9.713273406823398e-05} +{"train_loss": 0.018414340913295746, "grad_norm": 0.0708606168627739, "global_step": 5741, "epoch": 54, "lr": 9.713174477019607e-05} +{"train_loss": 0.018215380609035492, "grad_norm": 0.04497915506362915, "global_step": 5742, "epoch": 54, "lr": 9.713075530655768e-05} +{"train_loss": 0.01806851662695408, "grad_norm": 0.0682009607553482, "global_step": 5743, "epoch": 54, "lr": 9.712976567732232e-05} +{"train_loss": 0.01633613184094429, "grad_norm": 0.059693291783332825, "global_step": 5744, "epoch": 54, "lr": 9.712877588249348e-05} +{"train_loss": 0.01886875368654728, "grad_norm": 0.06512337177991867, "global_step": 5745, "epoch": 54, "lr": 9.71277859220746e-05} +{"train_loss": 0.019792046397924423, "grad_norm": 0.06543691456317902, "global_step": 5746, "epoch": 54, "lr": 9.712679579606918e-05} +{"train_loss": 0.022269677370786667, "grad_norm": 0.06515912711620331, "global_step": 5747, "epoch": 54, "lr": 9.712580550448069e-05} +{"train_loss": 0.01973751187324524, "grad_norm": 0.07829819619655609, "global_step": 5748, "epoch": 54, "lr": 9.712481504731263e-05} +{"train_loss": 0.018523992970585823, "grad_norm": 0.04989691078662872, "global_step": 5749, "epoch": 54, "lr": 9.712382442456845e-05} +{"train_loss": 0.017984768375754356, "grad_norm": 0.05956316739320755, "global_step": 5750, "epoch": 54, "lr": 9.712283363625165e-05} +{"train_loss": 0.017646482214331627, "grad_norm": 0.06597020477056503, "global_step": 5751, "epoch": 54, "lr": 9.71218426823657e-05} +{"train_loss": 0.016537010669708252, "grad_norm": 0.05079900845885277, "global_step": 5752, "epoch": 54, "lr": 9.712085156291409e-05} +{"train_loss": 0.017555322498083115, "grad_norm": 0.04072559252381325, "global_step": 5753, "epoch": 54, "lr": 9.71198602779003e-05} +{"train_loss": 0.0172392800450325, "grad_norm": 0.05622159317135811, "global_step": 5754, "epoch": 54, "lr": 9.71188688273278e-05} +{"train_loss": 0.01912717893719673, "grad_norm": 0.06182398647069931, "global_step": 5755, "epoch": 54, "lr": 9.711787721120011e-05} +{"train_loss": 0.019332358613610268, "grad_norm": 0.06616095453500748, "global_step": 5756, "epoch": 54, "lr": 9.711688542952066e-05} +{"train_loss": 0.019127342849969864, "grad_norm": 0.1310577392578125, "global_step": 5757, "epoch": 54, "lr": 9.7115893482293e-05} +{"train_loss": 0.01652679406106472, "grad_norm": 0.0478765144944191, "global_step": 5758, "epoch": 54, "lr": 9.711490136952056e-05} +{"train_loss": 0.017026487737894058, "grad_norm": 0.06223431974649429, "global_step": 5759, "epoch": 54, "lr": 9.711390909120686e-05} +{"train_loss": 0.018617069348692894, "grad_norm": 0.0697145164012909, "global_step": 5760, "epoch": 54, "lr": 9.711291664735535e-05} +{"train_loss": 0.0189959779381752, "grad_norm": 0.054392293095588684, "global_step": 5761, "epoch": 54, "lr": 9.711192403796955e-05} +{"train_loss": 0.015529362484812737, "grad_norm": 0.04797976091504097, "global_step": 5762, "epoch": 54, "lr": 9.711093126305296e-05} +{"train_loss": 0.018573980778455734, "grad_norm": 0.043276552110910416, "global_step": 5763, "epoch": 54, "lr": 9.710993832260902e-05} +{"train_loss": 0.018759271129965782, "grad_norm": 0.04929089918732643, "global_step": 5764, "epoch": 54, "lr": 9.710894521664125e-05} +{"train_loss": 0.019370652735233307, "grad_norm": 0.0645834431052208, "global_step": 5765, "epoch": 54, "lr": 9.710795194515313e-05} +{"train_loss": 0.017093053087592125, "grad_norm": 0.10030204802751541, "global_step": 5766, "epoch": 54, "lr": 9.710695850814817e-05} +{"train_loss": 0.01681019738316536, "grad_norm": 0.08786504715681076, "global_step": 5767, "epoch": 54, "lr": 9.710596490562982e-05} +{"train_loss": 0.02192908525466919, "grad_norm": 0.055886343121528625, "global_step": 5768, "epoch": 54, "lr": 9.710497113760161e-05} +{"train_loss": 0.0200475063174963, "grad_norm": 0.04937295615673065, "global_step": 5769, "epoch": 54, "lr": 9.710397720406701e-05} +{"train_loss": 0.020361140370368958, "grad_norm": 0.061696261167526245, "global_step": 5770, "epoch": 54, "lr": 9.710298310502952e-05} +{"train_loss": 0.01887233555316925, "grad_norm": 0.052274227142333984, "global_step": 5771, "epoch": 54, "lr": 9.710198884049265e-05} +{"train_loss": 0.019350256770849228, "grad_norm": 0.06728245317935944, "global_step": 5772, "epoch": 54, "lr": 9.710099441045986e-05} +{"train_loss": 0.019561147317290306, "grad_norm": 0.04385965317487717, "global_step": 5773, "epoch": 54, "lr": 9.709999981493465e-05} +{"train_loss": 0.0170341357588768, "grad_norm": 0.07993227988481522, "global_step": 5774, "epoch": 54, "lr": 9.709900505392054e-05} +{"train_loss": 0.01615668460726738, "grad_norm": 0.03714468330144882, "global_step": 5775, "epoch": 54, "lr": 9.7098010127421e-05} +{"train_loss": 0.02114572562277317, "grad_norm": 0.10808759927749634, "global_step": 5776, "epoch": 54, "lr": 9.709701503543954e-05} +{"train_loss": 0.020057130604982376, "grad_norm": 0.09642315655946732, "global_step": 5777, "epoch": 54, "lr": 9.709601977797964e-05} +{"train_loss": 0.018747253343462944, "grad_norm": 0.037299443036317825, "global_step": 5778, "epoch": 54, "lr": 9.709502435504483e-05} +{"train_loss": 0.01753799058496952, "grad_norm": 0.06783398985862732, "global_step": 5779, "epoch": 54, "lr": 9.709402876663856e-05} +{"train_loss": 0.019393134862184525, "grad_norm": 0.08112791180610657, "global_step": 5780, "epoch": 54, "lr": 9.709303301276437e-05} +{"train_loss": 0.017504338175058365, "grad_norm": 0.0749242752790451, "global_step": 5781, "epoch": 54, "lr": 9.709203709342574e-05} +{"train_loss": 0.020141005516052246, "grad_norm": 0.060245081782341, "global_step": 5782, "epoch": 54, "lr": 9.709104100862616e-05} +{"train_loss": 0.021258728578686714, "grad_norm": 0.1070936843752861, "global_step": 5783, "epoch": 54, "lr": 9.709004475836916e-05} +{"train_loss": 0.017543360590934753, "grad_norm": 0.047366317361593246, "global_step": 5784, "epoch": 54, "lr": 9.708904834265822e-05} +{"train_loss": 0.01975182071328163, "grad_norm": 0.08451418578624725, "global_step": 5785, "epoch": 54, "lr": 9.708805176149682e-05} +{"train_loss": 0.02087307535111904, "grad_norm": 0.06717269867658615, "global_step": 5786, "epoch": 54, "lr": 9.70870550148885e-05} +{"train_loss": 0.020120620727539062, "grad_norm": 0.060198910534381866, "global_step": 5787, "epoch": 54, "lr": 9.708605810283675e-05} +{"train_loss": 0.0194895938038826, "grad_norm": 0.05113288387656212, "global_step": 5788, "epoch": 54, "lr": 9.708506102534507e-05} +{"train_loss": 0.02020891383290291, "grad_norm": 0.05907835811376572, "global_step": 5789, "epoch": 54, "lr": 9.708406378241695e-05} +{"train_loss": 0.019253097474575043, "grad_norm": 0.04446324333548546, "global_step": 5790, "epoch": 54, "lr": 9.708306637405592e-05} +{"train_loss": 0.021380888298153877, "grad_norm": 0.12990613281726837, "global_step": 5791, "epoch": 54, "lr": 9.708206880026546e-05} +{"train_loss": 0.017689576372504234, "grad_norm": 0.05112609639763832, "global_step": 5792, "epoch": 54, "lr": 9.70810710610491e-05} +{"train_loss": 0.022266272455453873, "grad_norm": 0.09249582886695862, "global_step": 5793, "epoch": 54, "lr": 9.70800731564103e-05} +{"train_loss": 0.020303498953580856, "grad_norm": 0.08840174973011017, "global_step": 5794, "epoch": 54, "lr": 9.707907508635262e-05} +{"train_loss": 0.0185160581022501, "grad_norm": 0.07056280225515366, "global_step": 5795, "epoch": 54, "lr": 9.707807685087955e-05} +{"train_loss": 0.01984025351703167, "grad_norm": 0.044616568833589554, "global_step": 5796, "epoch": 54, "lr": 9.707707844999458e-05} +{"train_loss": 0.020230788737535477, "grad_norm": 0.057818081229925156, "global_step": 5797, "epoch": 54, "lr": 9.707607988370123e-05} +{"train_loss": 0.01902824267745018, "grad_norm": 0.07231997698545456, "global_step": 5798, "epoch": 54, "lr": 9.7075081152003e-05} +{"train_loss": 0.018596073612570763, "grad_norm": 0.06581965833902359, "global_step": 5799, "epoch": 54, "lr": 9.707408225490344e-05} +{"train_loss": 0.01712142303586006, "grad_norm": 0.047319553792476654, "global_step": 5800, "epoch": 54, "lr": 9.7073083192406e-05} +{"train_loss": 0.016589727252721786, "grad_norm": 0.10060913115739822, "global_step": 5801, "epoch": 54, "lr": 9.707208396451421e-05} +{"train_loss": 0.01767226681113243, "grad_norm": 0.08636001497507095, "global_step": 5802, "epoch": 54, "lr": 9.707108457123162e-05} +{"train_loss": 0.01981884241104126, "grad_norm": 0.09788738191127777, "global_step": 5803, "epoch": 54, "lr": 9.707008501256168e-05} +{"train_loss": 0.018819700926542282, "grad_norm": 0.14216065406799316, "global_step": 5804, "epoch": 54, "lr": 9.706908528850796e-05} +{"train_loss": 0.019111786037683487, "grad_norm": 0.07117863744497299, "global_step": 5805, "epoch": 54, "lr": 9.706808539907392e-05} +{"train_loss": 0.019986504688858986, "grad_norm": 0.12841357290744781, "global_step": 5806, "epoch": 54, "lr": 9.70670853442631e-05} +{"train_loss": 0.017258431762456894, "grad_norm": 0.045612357556819916, "global_step": 5807, "epoch": 54, "lr": 9.7066085124079e-05} +{"train_loss": 0.01633055880665779, "grad_norm": 0.10630978643894196, "global_step": 5808, "epoch": 54, "lr": 9.706508473852516e-05} +{"train_loss": 0.018986791372299194, "grad_norm": 0.1443607062101364, "global_step": 5809, "epoch": 54, "lr": 9.706408418760509e-05} +{"train_loss": 0.017204649746418, "grad_norm": 0.047427479177713394, "global_step": 5810, "epoch": 54, "lr": 9.706308347132229e-05} +{"train_loss": 0.01799265667796135, "grad_norm": 0.12245628982782364, "global_step": 5811, "epoch": 54, "lr": 9.706208258968028e-05} +{"train_loss": 0.021576618775725365, "grad_norm": 0.08447769284248352, "global_step": 5812, "epoch": 54, "lr": 9.706108154268257e-05} +{"train_loss": 0.021039718762040138, "grad_norm": 0.07330559194087982, "global_step": 5813, "epoch": 54, "lr": 9.706008033033269e-05} +{"train_loss": 0.016676848754286766, "grad_norm": 0.05947209149599075, "global_step": 5814, "epoch": 54, "lr": 9.705907895263415e-05} +{"train_loss": 0.02078501507639885, "grad_norm": 0.07381560653448105, "global_step": 5815, "epoch": 54, "lr": 9.705807740959049e-05} +{"train_loss": 0.021848758682608604, "grad_norm": 0.051483359187841415, "global_step": 5816, "epoch": 54, "lr": 9.70570757012052e-05} +{"train_loss": 0.023101966828107834, "grad_norm": 0.11194559186697006, "global_step": 5817, "epoch": 54, "lr": 9.705607382748181e-05} +{"train_loss": 0.020261278375983238, "grad_norm": 0.11091812700033188, "global_step": 5818, "epoch": 54, "lr": 9.705507178842384e-05} +{"train_loss": 0.01988391950726509, "grad_norm": 0.10373416543006897, "global_step": 5819, "epoch": 54, "lr": 9.70540695840348e-05} +{"train_loss": 0.021416382864117622, "grad_norm": 0.069272980093956, "global_step": 5820, "epoch": 54, "lr": 9.705306721431824e-05} +{"train_loss": 0.015409264713525772, "grad_norm": 0.04249633476138115, "global_step": 5821, "epoch": 54, "lr": 9.705206467927766e-05} +{"train_loss": 0.019858989864587784, "grad_norm": 0.07495283335447311, "global_step": 5822, "epoch": 54, "lr": 9.705106197891658e-05} +{"train_loss": 0.02011488378047943, "grad_norm": 0.04325573891401291, "global_step": 5823, "epoch": 54, "lr": 9.705005911323854e-05} +{"train_loss": 0.020492039620876312, "grad_norm": 0.04367424547672272, "global_step": 5824, "epoch": 54, "lr": 9.704905608224705e-05} +{"train_loss": 0.018552035093307495, "grad_norm": 0.05301180109381676, "global_step": 5825, "epoch": 54, "lr": 9.704805288594563e-05} +{"train_loss": 0.020447513088583946, "grad_norm": 0.07094468921422958, "global_step": 5826, "epoch": 54, "lr": 9.704704952433783e-05} +{"train_loss": 0.020027048885822296, "grad_norm": 0.061214692890644073, "global_step": 5827, "epoch": 54, "lr": 9.704604599742715e-05} +{"train_loss": 0.020668288692831993, "grad_norm": 0.07041871547698975, "global_step": 5828, "epoch": 54, "lr": 9.704504230521713e-05} +{"train_loss": 0.01904870877218134, "grad_norm": 0.061245545744895935, "global_step": 5829, "epoch": 54, "lr": 9.704403844771128e-05, "val_loss": 0.019507594406604767} +{"train_loss": 0.017298754304647446, "grad_norm": 0.05332154035568237, "global_step": 5830, "epoch": 55, "lr": 9.704303442491314e-05} +{"train_loss": 0.02175995707511902, "grad_norm": 0.05414905026555061, "global_step": 5831, "epoch": 55, "lr": 9.704203023682624e-05} +{"train_loss": 0.018409166485071182, "grad_norm": 0.10425341874361038, "global_step": 5832, "epoch": 55, "lr": 9.70410258834541e-05} +{"train_loss": 0.01929156854748726, "grad_norm": 0.11008641123771667, "global_step": 5833, "epoch": 55, "lr": 9.704002136480025e-05} +{"train_loss": 0.01717589795589447, "grad_norm": 0.1029597669839859, "global_step": 5834, "epoch": 55, "lr": 9.703901668086824e-05} +{"train_loss": 0.021779369562864304, "grad_norm": 0.0673934668302536, "global_step": 5835, "epoch": 55, "lr": 9.703801183166156e-05} +{"train_loss": 0.016676276922225952, "grad_norm": 0.09325801581144333, "global_step": 5836, "epoch": 55, "lr": 9.703700681718378e-05} +{"train_loss": 0.017260286957025528, "grad_norm": 0.06985383480787277, "global_step": 5837, "epoch": 55, "lr": 9.703600163743839e-05} +{"train_loss": 0.01636444218456745, "grad_norm": 0.044599179178476334, "global_step": 5838, "epoch": 55, "lr": 9.703499629242896e-05} +{"train_loss": 0.021668920293450356, "grad_norm": 0.05031083896756172, "global_step": 5839, "epoch": 55, "lr": 9.7033990782159e-05} +{"train_loss": 0.019534863531589508, "grad_norm": 0.15402328968048096, "global_step": 5840, "epoch": 55, "lr": 9.703298510663207e-05} +{"train_loss": 0.021711882203817368, "grad_norm": 0.09406576305627823, "global_step": 5841, "epoch": 55, "lr": 9.703197926585166e-05} +{"train_loss": 0.01947437785565853, "grad_norm": 0.0467393733561039, "global_step": 5842, "epoch": 55, "lr": 9.703097325982135e-05} +{"train_loss": 0.018583714962005615, "grad_norm": 0.05549582466483116, "global_step": 5843, "epoch": 55, "lr": 9.702996708854464e-05} +{"train_loss": 0.01932898536324501, "grad_norm": 0.0962778627872467, "global_step": 5844, "epoch": 55, "lr": 9.70289607520251e-05} +{"train_loss": 0.017983518540859222, "grad_norm": 0.109725221991539, "global_step": 5845, "epoch": 55, "lr": 9.702795425026623e-05} +{"train_loss": 0.017573900520801544, "grad_norm": 0.06464876234531403, "global_step": 5846, "epoch": 55, "lr": 9.702694758327158e-05} +{"train_loss": 0.02073105052113533, "grad_norm": 0.09740879386663437, "global_step": 5847, "epoch": 55, "lr": 9.702594075104468e-05} +{"train_loss": 0.02075628563761711, "grad_norm": 0.05071711540222168, "global_step": 5848, "epoch": 55, "lr": 9.702493375358907e-05} +{"train_loss": 0.01749381050467491, "grad_norm": 0.06522034853696823, "global_step": 5849, "epoch": 55, "lr": 9.702392659090831e-05} +{"train_loss": 0.019096579402685165, "grad_norm": 0.10954920202493668, "global_step": 5850, "epoch": 55, "lr": 9.702291926300592e-05} +{"train_loss": 0.019003331661224365, "grad_norm": 0.08228651434183121, "global_step": 5851, "epoch": 55, "lr": 9.702191176988544e-05} +{"train_loss": 0.02037126198410988, "grad_norm": 0.05915749445557594, "global_step": 5852, "epoch": 55, "lr": 9.70209041115504e-05} +{"train_loss": 0.015384901314973831, "grad_norm": 0.07551160454750061, "global_step": 5853, "epoch": 55, "lr": 9.701989628800436e-05} +{"train_loss": 0.018475476652383804, "grad_norm": 0.06433162838220596, "global_step": 5854, "epoch": 55, "lr": 9.701888829925086e-05} +{"train_loss": 0.0173353124409914, "grad_norm": 0.11761859804391861, "global_step": 5855, "epoch": 55, "lr": 9.701788014529342e-05} +{"train_loss": 0.02036655694246292, "grad_norm": 0.06140362471342087, "global_step": 5856, "epoch": 55, "lr": 9.701687182613561e-05} +{"train_loss": 0.018806181848049164, "grad_norm": 0.07095644623041153, "global_step": 5857, "epoch": 55, "lr": 9.701586334178096e-05} +{"train_loss": 0.019914060831069946, "grad_norm": 0.07768607139587402, "global_step": 5858, "epoch": 55, "lr": 9.701485469223302e-05} +{"train_loss": 0.018834486603736877, "grad_norm": 0.059361644089221954, "global_step": 5859, "epoch": 55, "lr": 9.701384587749531e-05} +{"train_loss": 0.01906413957476616, "grad_norm": 0.051376815885305405, "global_step": 5860, "epoch": 55, "lr": 9.70128368975714e-05} +{"train_loss": 0.02098741941154003, "grad_norm": 0.04945058375597, "global_step": 5861, "epoch": 55, "lr": 9.701182775246482e-05} +{"train_loss": 0.017697419971227646, "grad_norm": 0.05000624805688858, "global_step": 5862, "epoch": 55, "lr": 9.701081844217912e-05} +{"train_loss": 0.01781102642416954, "grad_norm": 0.11399610340595245, "global_step": 5863, "epoch": 55, "lr": 9.700980896671785e-05} +{"train_loss": 0.016947370022535324, "grad_norm": 0.09003538638353348, "global_step": 5864, "epoch": 55, "lr": 9.700879932608456e-05} +{"train_loss": 0.01836479641497135, "grad_norm": 0.11408483982086182, "global_step": 5865, "epoch": 55, "lr": 9.70077895202828e-05} +{"train_loss": 0.01951914094388485, "grad_norm": 0.04583338275551796, "global_step": 5866, "epoch": 55, "lr": 9.700677954931609e-05} +{"train_loss": 0.017024755477905273, "grad_norm": 0.04011427238583565, "global_step": 5867, "epoch": 55, "lr": 9.700576941318803e-05} +{"train_loss": 0.0171523280441761, "grad_norm": 0.07076121121644974, "global_step": 5868, "epoch": 55, "lr": 9.70047591119021e-05} +{"train_loss": 0.01671072095632553, "grad_norm": 0.06842376291751862, "global_step": 5869, "epoch": 55, "lr": 9.700374864546191e-05} +{"train_loss": 0.020829100161790848, "grad_norm": 0.06652573496103287, "global_step": 5870, "epoch": 55, "lr": 9.700273801387098e-05} +{"train_loss": 0.02192169800400734, "grad_norm": 0.13566388189792633, "global_step": 5871, "epoch": 55, "lr": 9.700172721713288e-05} +{"train_loss": 0.019081389531493187, "grad_norm": 0.05145643651485443, "global_step": 5872, "epoch": 55, "lr": 9.700071625525115e-05} +{"train_loss": 0.019061118364334106, "grad_norm": 0.04247962310910225, "global_step": 5873, "epoch": 55, "lr": 9.699970512822935e-05} +{"train_loss": 0.02045043744146824, "grad_norm": 0.05178442969918251, "global_step": 5874, "epoch": 55, "lr": 9.6998693836071e-05} +{"train_loss": 0.018849557265639305, "grad_norm": 0.05125708505511284, "global_step": 5875, "epoch": 55, "lr": 9.699768237877971e-05} +{"train_loss": 0.02315508760511875, "grad_norm": 0.0433158241212368, "global_step": 5876, "epoch": 55, "lr": 9.699667075635899e-05} +{"train_loss": 0.02060449868440628, "grad_norm": 0.11457709223031998, "global_step": 5877, "epoch": 55, "lr": 9.69956589688124e-05} +{"train_loss": 0.018646854907274246, "grad_norm": 0.0881483182311058, "global_step": 5878, "epoch": 55, "lr": 9.699464701614352e-05} +{"train_loss": 0.02173466607928276, "grad_norm": 0.048197947442531586, "global_step": 5879, "epoch": 55, "lr": 9.699363489835586e-05} +{"train_loss": 0.017612919211387634, "grad_norm": 0.038992367684841156, "global_step": 5880, "epoch": 55, "lr": 9.699262261545303e-05} +{"train_loss": 0.01804683357477188, "grad_norm": 0.07253886759281158, "global_step": 5881, "epoch": 55, "lr": 9.699161016743855e-05} +{"train_loss": 0.01700364053249359, "grad_norm": 0.08632564544677734, "global_step": 5882, "epoch": 55, "lr": 9.699059755431598e-05} +{"train_loss": 0.019016757607460022, "grad_norm": 0.074436716735363, "global_step": 5883, "epoch": 55, "lr": 9.698958477608891e-05} +{"train_loss": 0.01846363954246044, "grad_norm": 0.05143541097640991, "global_step": 5884, "epoch": 55, "lr": 9.698857183276084e-05} +{"train_loss": 0.021859576925635338, "grad_norm": 0.08749381452798843, "global_step": 5885, "epoch": 55, "lr": 9.698755872433539e-05} +{"train_loss": 0.016021905466914177, "grad_norm": 0.05772217735648155, "global_step": 5886, "epoch": 55, "lr": 9.698654545081609e-05} +{"train_loss": 0.019737601280212402, "grad_norm": 0.05929954722523689, "global_step": 5887, "epoch": 55, "lr": 9.69855320122065e-05} +{"train_loss": 0.018647950142621994, "grad_norm": 0.06534293293952942, "global_step": 5888, "epoch": 55, "lr": 9.698451840851018e-05} +{"train_loss": 0.01879776641726494, "grad_norm": 0.12723876535892487, "global_step": 5889, "epoch": 55, "lr": 9.69835046397307e-05} +{"train_loss": 0.018605422228574753, "grad_norm": 0.10413327813148499, "global_step": 5890, "epoch": 55, "lr": 9.698249070587162e-05} +{"train_loss": 0.018539972603321075, "grad_norm": 0.041832007467746735, "global_step": 5891, "epoch": 55, "lr": 9.698147660693648e-05} +{"train_loss": 0.017263337969779968, "grad_norm": 0.04981238767504692, "global_step": 5892, "epoch": 55, "lr": 9.698046234292889e-05} +{"train_loss": 0.017371874302625656, "grad_norm": 0.12477480620145798, "global_step": 5893, "epoch": 55, "lr": 9.697944791385236e-05} +{"train_loss": 0.01762908697128296, "grad_norm": 0.03958272933959961, "global_step": 5894, "epoch": 55, "lr": 9.697843331971051e-05} +{"train_loss": 0.021167226135730743, "grad_norm": 0.07815674692392349, "global_step": 5895, "epoch": 55, "lr": 9.697741856050685e-05} +{"train_loss": 0.015904050320386887, "grad_norm": 0.05999062582850456, "global_step": 5896, "epoch": 55, "lr": 9.697640363624498e-05} +{"train_loss": 0.020308393985033035, "grad_norm": 0.07569468766450882, "global_step": 5897, "epoch": 55, "lr": 9.697538854692846e-05} +{"train_loss": 0.01862466335296631, "grad_norm": 0.06975933164358139, "global_step": 5898, "epoch": 55, "lr": 9.697437329256085e-05} +{"train_loss": 0.016263559460639954, "grad_norm": 0.0564001202583313, "global_step": 5899, "epoch": 55, "lr": 9.697335787314573e-05} +{"train_loss": 0.018196336925029755, "grad_norm": 0.05943724140524864, "global_step": 5900, "epoch": 55, "lr": 9.697234228868666e-05} +{"train_loss": 0.018049225211143494, "grad_norm": 0.05855352804064751, "global_step": 5901, "epoch": 55, "lr": 9.697132653918719e-05} +{"train_loss": 0.021043749526143074, "grad_norm": 0.09198451787233353, "global_step": 5902, "epoch": 55, "lr": 9.697031062465092e-05} +{"train_loss": 0.017016571015119553, "grad_norm": 0.049988068640232086, "global_step": 5903, "epoch": 55, "lr": 9.696929454508139e-05} +{"train_loss": 0.022439228370785713, "grad_norm": 0.05835379287600517, "global_step": 5904, "epoch": 55, "lr": 9.696827830048221e-05} +{"train_loss": 0.021120203658938408, "grad_norm": 0.05507032200694084, "global_step": 5905, "epoch": 55, "lr": 9.69672618908569e-05} +{"train_loss": 0.017900139093399048, "grad_norm": 0.12038726359605789, "global_step": 5906, "epoch": 55, "lr": 9.696624531620908e-05} +{"train_loss": 0.02007407695055008, "grad_norm": 0.06091729924082756, "global_step": 5907, "epoch": 55, "lr": 9.696522857654229e-05} +{"train_loss": 0.019411344081163406, "grad_norm": 0.05994746461510658, "global_step": 5908, "epoch": 55, "lr": 9.69642116718601e-05} +{"train_loss": 0.020856115967035294, "grad_norm": 0.04520028084516525, "global_step": 5909, "epoch": 55, "lr": 9.696319460216611e-05} +{"train_loss": 0.016023902222514153, "grad_norm": 0.045000314712524414, "global_step": 5910, "epoch": 55, "lr": 9.696217736746387e-05} +{"train_loss": 0.017154674977064133, "grad_norm": 0.08053027093410492, "global_step": 5911, "epoch": 55, "lr": 9.696115996775696e-05} +{"train_loss": 0.015925638377666473, "grad_norm": 0.052294954657554626, "global_step": 5912, "epoch": 55, "lr": 9.696014240304895e-05} +{"train_loss": 0.01702863723039627, "grad_norm": 0.04527709633111954, "global_step": 5913, "epoch": 55, "lr": 9.695912467334344e-05} +{"train_loss": 0.01887720823287964, "grad_norm": 0.05555684491991997, "global_step": 5914, "epoch": 55, "lr": 9.695810677864398e-05} +{"train_loss": 0.01850074902176857, "grad_norm": 0.04209155961871147, "global_step": 5915, "epoch": 55, "lr": 9.695708871895415e-05} +{"train_loss": 0.02081749588251114, "grad_norm": 0.0677511990070343, "global_step": 5916, "epoch": 55, "lr": 9.695607049427755e-05} +{"train_loss": 0.019863806664943695, "grad_norm": 0.05824156105518341, "global_step": 5917, "epoch": 55, "lr": 9.695505210461772e-05} +{"train_loss": 0.019176965579390526, "grad_norm": 0.07817375659942627, "global_step": 5918, "epoch": 55, "lr": 9.695403354997826e-05} +{"train_loss": 0.02057565189898014, "grad_norm": 0.0750107541680336, "global_step": 5919, "epoch": 55, "lr": 9.695301483036275e-05} +{"train_loss": 0.022427257150411606, "grad_norm": 0.04286942631006241, "global_step": 5920, "epoch": 55, "lr": 9.695199594577476e-05} +{"train_loss": 0.019259624183177948, "grad_norm": 0.04808934032917023, "global_step": 5921, "epoch": 55, "lr": 9.695097689621789e-05} +{"train_loss": 0.017660029232501984, "grad_norm": 0.055224012583494186, "global_step": 5922, "epoch": 55, "lr": 9.694995768169567e-05} +{"train_loss": 0.02148028276860714, "grad_norm": 0.14618255198001862, "global_step": 5923, "epoch": 55, "lr": 9.694893830221176e-05} +{"train_loss": 0.017478391528129578, "grad_norm": 0.06573653966188431, "global_step": 5924, "epoch": 55, "lr": 9.694791875776967e-05} +{"train_loss": 0.018625061959028244, "grad_norm": 0.047335490584373474, "global_step": 5925, "epoch": 55, "lr": 9.694689904837302e-05} +{"train_loss": 0.01780106872320175, "grad_norm": 0.05464727431535721, "global_step": 5926, "epoch": 55, "lr": 9.694587917402538e-05} +{"train_loss": 0.018579203635454178, "grad_norm": 0.09173685312271118, "global_step": 5927, "epoch": 55, "lr": 9.694485913473034e-05} +{"train_loss": 0.01929582469165325, "grad_norm": 0.10454408824443817, "global_step": 5928, "epoch": 55, "lr": 9.694383893049148e-05} +{"train_loss": 0.01989767700433731, "grad_norm": 0.04635413736104965, "global_step": 5929, "epoch": 55, "lr": 9.69428185613124e-05} +{"train_loss": 0.02028181403875351, "grad_norm": 0.06344441324472427, "global_step": 5930, "epoch": 55, "lr": 9.694179802719664e-05} +{"train_loss": 0.0209937933832407, "grad_norm": 0.06911289691925049, "global_step": 5931, "epoch": 55, "lr": 9.694077732814783e-05} +{"train_loss": 0.019306859001517296, "grad_norm": 0.06527149677276611, "global_step": 5932, "epoch": 55, "lr": 9.693975646416954e-05} +{"train_loss": 0.020177878439426422, "grad_norm": 0.116989865899086, "global_step": 5933, "epoch": 55, "lr": 9.693873543526538e-05} +{"train_loss": 0.018558520823717117, "grad_norm": 0.05731919780373573, "global_step": 5934, "epoch": 55, "lr": 9.69377142414389e-05} +{"train_loss": 0.01899146346621356, "grad_norm": 0.07951261103153229, "global_step": 5935, "epoch": 55, "lr": 9.693669288269372e-05, "val_loss": 0.01937384344637394} +{"train_loss": 0.015676572918891907, "grad_norm": 0.07284142822027206, "global_step": 5936, "epoch": 56, "lr": 9.693567135903338e-05} +{"train_loss": 0.01956970989704132, "grad_norm": 0.06458653509616852, "global_step": 5937, "epoch": 56, "lr": 9.693464967046153e-05} +{"train_loss": 0.01975363679230213, "grad_norm": 0.04268263280391693, "global_step": 5938, "epoch": 56, "lr": 9.693362781698173e-05} +{"train_loss": 0.018816497176885605, "grad_norm": 0.11967488378286362, "global_step": 5939, "epoch": 56, "lr": 9.693260579859756e-05} +{"train_loss": 0.017099887132644653, "grad_norm": 0.035321854054927826, "global_step": 5940, "epoch": 56, "lr": 9.693158361531262e-05} +{"train_loss": 0.01741916313767433, "grad_norm": 0.11863978207111359, "global_step": 5941, "epoch": 56, "lr": 9.693056126713053e-05} +{"train_loss": 0.021605245769023895, "grad_norm": 0.06114621460437775, "global_step": 5942, "epoch": 56, "lr": 9.692953875405484e-05} +{"train_loss": 0.018546201288700104, "grad_norm": 0.12930341064929962, "global_step": 5943, "epoch": 56, "lr": 9.692851607608918e-05} +{"train_loss": 0.01898106560111046, "grad_norm": 0.11698947101831436, "global_step": 5944, "epoch": 56, "lr": 9.692749323323709e-05} +{"train_loss": 0.01873709447681904, "grad_norm": 0.06844813376665115, "global_step": 5945, "epoch": 56, "lr": 9.692647022550221e-05} +{"train_loss": 0.016873160377144814, "grad_norm": 0.0703815296292305, "global_step": 5946, "epoch": 56, "lr": 9.692544705288813e-05} +{"train_loss": 0.0204066913574934, "grad_norm": 0.0490284189581871, "global_step": 5947, "epoch": 56, "lr": 9.692442371539843e-05} +{"train_loss": 0.018200047314167023, "grad_norm": 0.08250156790018082, "global_step": 5948, "epoch": 56, "lr": 9.692340021303672e-05} +{"train_loss": 0.020599715411663055, "grad_norm": 0.06920193880796432, "global_step": 5949, "epoch": 56, "lr": 9.692237654580658e-05} +{"train_loss": 0.021011359989643097, "grad_norm": 0.07400965690612793, "global_step": 5950, "epoch": 56, "lr": 9.69213527137116e-05} +{"train_loss": 0.018713101744651794, "grad_norm": 0.07035378366708755, "global_step": 5951, "epoch": 56, "lr": 9.692032871675542e-05} +{"train_loss": 0.02108171209692955, "grad_norm": 0.08275686949491501, "global_step": 5952, "epoch": 56, "lr": 9.691930455494161e-05} +{"train_loss": 0.017370538786053658, "grad_norm": 0.15111112594604492, "global_step": 5953, "epoch": 56, "lr": 9.691828022827374e-05} +{"train_loss": 0.01911672204732895, "grad_norm": 0.050575558096170425, "global_step": 5954, "epoch": 56, "lr": 9.691725573675546e-05} +{"train_loss": 0.016227178275585175, "grad_norm": 0.05270844325423241, "global_step": 5955, "epoch": 56, "lr": 9.691623108039034e-05} +{"train_loss": 0.018518663942813873, "grad_norm": 0.10505720973014832, "global_step": 5956, "epoch": 56, "lr": 9.691520625918199e-05} +{"train_loss": 0.02060660719871521, "grad_norm": 0.1440633088350296, "global_step": 5957, "epoch": 56, "lr": 9.691418127313401e-05} +{"train_loss": 0.021781837567687035, "grad_norm": 0.11017812043428421, "global_step": 5958, "epoch": 56, "lr": 9.691315612224999e-05} +{"train_loss": 0.018792930990457535, "grad_norm": 0.07280521094799042, "global_step": 5959, "epoch": 56, "lr": 9.691213080653356e-05} +{"train_loss": 0.01985468715429306, "grad_norm": 0.06833621859550476, "global_step": 5960, "epoch": 56, "lr": 9.691110532598828e-05} +{"train_loss": 0.02034934051334858, "grad_norm": 0.04739700257778168, "global_step": 5961, "epoch": 56, "lr": 9.691007968061778e-05} +{"train_loss": 0.020312082022428513, "grad_norm": 0.10213956981897354, "global_step": 5962, "epoch": 56, "lr": 9.690905387042567e-05} +{"train_loss": 0.01887730322778225, "grad_norm": 0.050198979675769806, "global_step": 5963, "epoch": 56, "lr": 9.690802789541555e-05} +{"train_loss": 0.019585585221648216, "grad_norm": 0.08122893422842026, "global_step": 5964, "epoch": 56, "lr": 9.690700175559102e-05} +{"train_loss": 0.02055191993713379, "grad_norm": 0.10849344730377197, "global_step": 5965, "epoch": 56, "lr": 9.690597545095568e-05} +{"train_loss": 0.019191831350326538, "grad_norm": 0.05344656854867935, "global_step": 5966, "epoch": 56, "lr": 9.690494898151313e-05} +{"train_loss": 0.017780793830752373, "grad_norm": 0.06212669983506203, "global_step": 5967, "epoch": 56, "lr": 9.690392234726701e-05} +{"train_loss": 0.016952134668827057, "grad_norm": 0.04070400446653366, "global_step": 5968, "epoch": 56, "lr": 9.690289554822088e-05} +{"train_loss": 0.017470339313149452, "grad_norm": 0.04303428530693054, "global_step": 5969, "epoch": 56, "lr": 9.69018685843784e-05} +{"train_loss": 0.018920641392469406, "grad_norm": 0.06412999331951141, "global_step": 5970, "epoch": 56, "lr": 9.690084145574313e-05} +{"train_loss": 0.019169922918081284, "grad_norm": 0.09361172467470169, "global_step": 5971, "epoch": 56, "lr": 9.689981416231871e-05} +{"train_loss": 0.016634825617074966, "grad_norm": 0.04919137433171272, "global_step": 5972, "epoch": 56, "lr": 9.689878670410872e-05} +{"train_loss": 0.02042490616440773, "grad_norm": 0.09651286154985428, "global_step": 5973, "epoch": 56, "lr": 9.689775908111682e-05} +{"train_loss": 0.0175009872764349, "grad_norm": 0.07272662222385406, "global_step": 5974, "epoch": 56, "lr": 9.689673129334656e-05} +{"train_loss": 0.021330781280994415, "grad_norm": 0.0633910596370697, "global_step": 5975, "epoch": 56, "lr": 9.68957033408016e-05} +{"train_loss": 0.02136359177529812, "grad_norm": 0.10209063440561295, "global_step": 5976, "epoch": 56, "lr": 9.689467522348553e-05} +{"train_loss": 0.014960935339331627, "grad_norm": 0.05231071263551712, "global_step": 5977, "epoch": 56, "lr": 9.689364694140196e-05} +{"train_loss": 0.0188730601221323, "grad_norm": 0.05951278284192085, "global_step": 5978, "epoch": 56, "lr": 9.689261849455452e-05} +{"train_loss": 0.01944544166326523, "grad_norm": 0.06831394135951996, "global_step": 5979, "epoch": 56, "lr": 9.68915898829468e-05} +{"train_loss": 0.018182961270213127, "grad_norm": 0.04207531735301018, "global_step": 5980, "epoch": 56, "lr": 9.689056110658244e-05} +{"train_loss": 0.01734074205160141, "grad_norm": 0.052351806312799454, "global_step": 5981, "epoch": 56, "lr": 9.688953216546502e-05} +{"train_loss": 0.018488138914108276, "grad_norm": 0.1109732910990715, "global_step": 5982, "epoch": 56, "lr": 9.688850305959818e-05} +{"train_loss": 0.016324762254953384, "grad_norm": 0.10038546472787857, "global_step": 5983, "epoch": 56, "lr": 9.688747378898554e-05} +{"train_loss": 0.017064664512872696, "grad_norm": 0.06442280113697052, "global_step": 5984, "epoch": 56, "lr": 9.688644435363071e-05} +{"train_loss": 0.018763627856969833, "grad_norm": 0.0741826519370079, "global_step": 5985, "epoch": 56, "lr": 9.688541475353729e-05} +{"train_loss": 0.018204333260655403, "grad_norm": 0.06742068380117416, "global_step": 5986, "epoch": 56, "lr": 9.688438498870892e-05} +{"train_loss": 0.02078864350914955, "grad_norm": 0.057053301483392715, "global_step": 5987, "epoch": 56, "lr": 9.688335505914921e-05} +{"train_loss": 0.015200117602944374, "grad_norm": 0.08703377842903137, "global_step": 5988, "epoch": 56, "lr": 9.688232496486178e-05} +{"train_loss": 0.01745907962322235, "grad_norm": 0.0832756906747818, "global_step": 5989, "epoch": 56, "lr": 9.688129470585026e-05} +{"train_loss": 0.01899784430861473, "grad_norm": 0.044429514557123184, "global_step": 5990, "epoch": 56, "lr": 9.688026428211825e-05} +{"train_loss": 0.019847556948661804, "grad_norm": 0.04945012181997299, "global_step": 5991, "epoch": 56, "lr": 9.687923369366937e-05} +{"train_loss": 0.0189545676112175, "grad_norm": 0.16063982248306274, "global_step": 5992, "epoch": 56, "lr": 9.687820294050727e-05} +{"train_loss": 0.0193449929356575, "grad_norm": 0.05764440447092056, "global_step": 5993, "epoch": 56, "lr": 9.687717202263553e-05} +{"train_loss": 0.022261813282966614, "grad_norm": 0.1169828474521637, "global_step": 5994, "epoch": 56, "lr": 9.687614094005781e-05} +{"train_loss": 0.02034815028309822, "grad_norm": 0.0673835277557373, "global_step": 5995, "epoch": 56, "lr": 9.687510969277771e-05} +{"train_loss": 0.018434613943099976, "grad_norm": 0.0606335885822773, "global_step": 5996, "epoch": 56, "lr": 9.687407828079886e-05} +{"train_loss": 0.019400710240006447, "grad_norm": 0.058962542563676834, "global_step": 5997, "epoch": 56, "lr": 9.687304670412488e-05} +{"train_loss": 0.018790945410728455, "grad_norm": 0.1253974586725235, "global_step": 5998, "epoch": 56, "lr": 9.68720149627594e-05} +{"train_loss": 0.016634132713079453, "grad_norm": 0.07423177361488342, "global_step": 5999, "epoch": 56, "lr": 9.687098305670605e-05} +{"train_loss": 0.018872413784265518, "grad_norm": 0.04368823021650314, "global_step": 6000, "epoch": 56, "lr": 9.686995098596845e-05} +{"train_loss": 0.019495956599712372, "grad_norm": 0.07409341633319855, "global_step": 6001, "epoch": 56, "lr": 9.686891875055022e-05} +{"train_loss": 0.018962368369102478, "grad_norm": 0.04759100452065468, "global_step": 6002, "epoch": 56, "lr": 9.686788635045498e-05} +{"train_loss": 0.0183084849268198, "grad_norm": 0.046223681420087814, "global_step": 6003, "epoch": 56, "lr": 9.686685378568639e-05} +{"train_loss": 0.021456323564052582, "grad_norm": 0.13809345662593842, "global_step": 6004, "epoch": 56, "lr": 9.686582105624806e-05} +{"train_loss": 0.021135441958904266, "grad_norm": 0.10572952777147293, "global_step": 6005, "epoch": 56, "lr": 9.68647881621436e-05} +{"train_loss": 0.019222280010581017, "grad_norm": 0.0728922039270401, "global_step": 6006, "epoch": 56, "lr": 9.686375510337666e-05} +{"train_loss": 0.0199868343770504, "grad_norm": 0.1272294968366623, "global_step": 6007, "epoch": 56, "lr": 9.686272187995087e-05} +{"train_loss": 0.020675867795944214, "grad_norm": 0.11142061650753021, "global_step": 6008, "epoch": 56, "lr": 9.686168849186985e-05} +{"train_loss": 0.020436348393559456, "grad_norm": 0.05150572583079338, "global_step": 6009, "epoch": 56, "lr": 9.686065493913723e-05} +{"train_loss": 0.017397023737430573, "grad_norm": 0.07510972768068314, "global_step": 6010, "epoch": 56, "lr": 9.685962122175667e-05} +{"train_loss": 0.020262904465198517, "grad_norm": 0.05415957421064377, "global_step": 6011, "epoch": 56, "lr": 9.685858733973177e-05} +{"train_loss": 0.01925678923726082, "grad_norm": 0.05488564819097519, "global_step": 6012, "epoch": 56, "lr": 9.685755329306616e-05} +{"train_loss": 0.01860787346959114, "grad_norm": 0.04012824594974518, "global_step": 6013, "epoch": 56, "lr": 9.68565190817635e-05} +{"train_loss": 0.020478103309869766, "grad_norm": 0.05883077532052994, "global_step": 6014, "epoch": 56, "lr": 9.685548470582739e-05} +{"train_loss": 0.023239562287926674, "grad_norm": 0.08470679819583893, "global_step": 6015, "epoch": 56, "lr": 9.68544501652615e-05} +{"train_loss": 0.019328083842992783, "grad_norm": 0.15188293159008026, "global_step": 6016, "epoch": 56, "lr": 9.685341546006945e-05} +{"train_loss": 0.017673371359705925, "grad_norm": 0.07712282985448837, "global_step": 6017, "epoch": 56, "lr": 9.685238059025487e-05} +{"train_loss": 0.020070627331733704, "grad_norm": 0.0803709626197815, "global_step": 6018, "epoch": 56, "lr": 9.685134555582141e-05} +{"train_loss": 0.018620409071445465, "grad_norm": 0.11448514461517334, "global_step": 6019, "epoch": 56, "lr": 9.685031035677268e-05} +{"train_loss": 0.01798078790307045, "grad_norm": 0.11062286794185638, "global_step": 6020, "epoch": 56, "lr": 9.684927499311234e-05} +{"train_loss": 0.018527738749980927, "grad_norm": 0.09486309438943863, "global_step": 6021, "epoch": 56, "lr": 9.684823946484401e-05} +{"train_loss": 0.017563272267580032, "grad_norm": 0.04312824085354805, "global_step": 6022, "epoch": 56, "lr": 9.684720377197136e-05} +{"train_loss": 0.016689330339431763, "grad_norm": 0.04778742045164108, "global_step": 6023, "epoch": 56, "lr": 9.684616791449801e-05} +{"train_loss": 0.017187535762786865, "grad_norm": 0.059640977531671524, "global_step": 6024, "epoch": 56, "lr": 9.684513189242759e-05} +{"train_loss": 0.019691042602062225, "grad_norm": 0.04514021426439285, "global_step": 6025, "epoch": 56, "lr": 9.684409570576373e-05} +{"train_loss": 0.019331149756908417, "grad_norm": 0.07764549553394318, "global_step": 6026, "epoch": 56, "lr": 9.684305935451012e-05} +{"train_loss": 0.021906152367591858, "grad_norm": 0.06893053650856018, "global_step": 6027, "epoch": 56, "lr": 9.684202283867036e-05} +{"train_loss": 0.017437517642974854, "grad_norm": 0.06993388384580612, "global_step": 6028, "epoch": 56, "lr": 9.684098615824811e-05} +{"train_loss": 0.019633077085018158, "grad_norm": 0.05160481855273247, "global_step": 6029, "epoch": 56, "lr": 9.683994931324699e-05} +{"train_loss": 0.01833062246441841, "grad_norm": 0.05509946867823601, "global_step": 6030, "epoch": 56, "lr": 9.683891230367067e-05} +{"train_loss": 0.019053807482123375, "grad_norm": 0.14193958044052124, "global_step": 6031, "epoch": 56, "lr": 9.683787512952278e-05} +{"train_loss": 0.020448319613933563, "grad_norm": 0.13746701180934906, "global_step": 6032, "epoch": 56, "lr": 9.683683779080696e-05} +{"train_loss": 0.01908480003476143, "grad_norm": 0.12281308323144913, "global_step": 6033, "epoch": 56, "lr": 9.683580028752687e-05} +{"train_loss": 0.0169847309589386, "grad_norm": 0.04980432242155075, "global_step": 6034, "epoch": 56, "lr": 9.683476261968614e-05} +{"train_loss": 0.020537789911031723, "grad_norm": 0.10178396105766296, "global_step": 6035, "epoch": 56, "lr": 9.683372478728843e-05} +{"train_loss": 0.019840102642774582, "grad_norm": 0.04833362251520157, "global_step": 6036, "epoch": 56, "lr": 9.683268679033737e-05} +{"train_loss": 0.017814746126532555, "grad_norm": 0.03972472622990608, "global_step": 6037, "epoch": 56, "lr": 9.683164862883662e-05} +{"train_loss": 0.01674267277121544, "grad_norm": 0.052656032145023346, "global_step": 6038, "epoch": 56, "lr": 9.683061030278981e-05} +{"train_loss": 0.017286794260144234, "grad_norm": 0.05870983377099037, "global_step": 6039, "epoch": 56, "lr": 9.682957181220062e-05} +{"train_loss": 0.019976697862148285, "grad_norm": 0.10471250116825104, "global_step": 6040, "epoch": 56, "lr": 9.682853315707268e-05} +{"train_loss": 0.018950185879080924, "grad_norm": 0.13346563279628754, "global_step": 6041, "epoch": 56, "lr": 9.682749433740962e-05, "val_loss": 0.0191297959536314} +{"train_loss": 0.018888115882873535, "grad_norm": 0.061812423169612885, "global_step": 6042, "epoch": 57, "lr": 9.682645535321513e-05} +{"train_loss": 0.01782340556383133, "grad_norm": 0.04217321798205376, "global_step": 6043, "epoch": 57, "lr": 9.682541620449282e-05} +{"train_loss": 0.018993550911545753, "grad_norm": 0.05800565704703331, "global_step": 6044, "epoch": 57, "lr": 9.682437689124638e-05} +{"train_loss": 0.019983358681201935, "grad_norm": 0.0614156648516655, "global_step": 6045, "epoch": 57, "lr": 9.682333741347944e-05} +{"train_loss": 0.016400478780269623, "grad_norm": 0.08661142736673355, "global_step": 6046, "epoch": 57, "lr": 9.682229777119564e-05} +{"train_loss": 0.01973990723490715, "grad_norm": 0.07722834497690201, "global_step": 6047, "epoch": 57, "lr": 9.682125796439868e-05} +{"train_loss": 0.018693244084715843, "grad_norm": 0.09421040862798691, "global_step": 6048, "epoch": 57, "lr": 9.682021799309215e-05} +{"train_loss": 0.01953566074371338, "grad_norm": 0.054069846868515015, "global_step": 6049, "epoch": 57, "lr": 9.681917785727974e-05} +{"train_loss": 0.021373048424720764, "grad_norm": 0.041531339287757874, "global_step": 6050, "epoch": 57, "lr": 9.681813755696512e-05} +{"train_loss": 0.016087476164102554, "grad_norm": 0.1391700953245163, "global_step": 6051, "epoch": 57, "lr": 9.681709709215191e-05} +{"train_loss": 0.01812056452035904, "grad_norm": 0.06698665767908096, "global_step": 6052, "epoch": 57, "lr": 9.681605646284378e-05} +{"train_loss": 0.019493721425533295, "grad_norm": 0.07569077610969543, "global_step": 6053, "epoch": 57, "lr": 9.681501566904439e-05} +{"train_loss": 0.02095627598464489, "grad_norm": 0.05942912772297859, "global_step": 6054, "epoch": 57, "lr": 9.681397471075739e-05} +{"train_loss": 0.018996911123394966, "grad_norm": 0.06444589048624039, "global_step": 6055, "epoch": 57, "lr": 9.681293358798645e-05} +{"train_loss": 0.016121890395879745, "grad_norm": 0.039719197899103165, "global_step": 6056, "epoch": 57, "lr": 9.681189230073522e-05} +{"train_loss": 0.021569404751062393, "grad_norm": 0.04793903976678848, "global_step": 6057, "epoch": 57, "lr": 9.681085084900736e-05} +{"train_loss": 0.01847059838473797, "grad_norm": 0.11031930893659592, "global_step": 6058, "epoch": 57, "lr": 9.680980923280651e-05} +{"train_loss": 0.020674776285886765, "grad_norm": 0.09311436861753464, "global_step": 6059, "epoch": 57, "lr": 9.680876745213636e-05} +{"train_loss": 0.01641189306974411, "grad_norm": 0.06040685251355171, "global_step": 6060, "epoch": 57, "lr": 9.680772550700057e-05} +{"train_loss": 0.014991628006100655, "grad_norm": 0.07978601008653641, "global_step": 6061, "epoch": 57, "lr": 9.680668339740278e-05} +{"train_loss": 0.01805710420012474, "grad_norm": 0.09248795360326767, "global_step": 6062, "epoch": 57, "lr": 9.680564112334665e-05} +{"train_loss": 0.021040603518486023, "grad_norm": 0.0470007099211216, "global_step": 6063, "epoch": 57, "lr": 9.680459868483585e-05} +{"train_loss": 0.015807412564754486, "grad_norm": 0.05988626927137375, "global_step": 6064, "epoch": 57, "lr": 9.680355608187406e-05} +{"train_loss": 0.019037757068872452, "grad_norm": 0.04966985806822777, "global_step": 6065, "epoch": 57, "lr": 9.680251331446492e-05} +{"train_loss": 0.018516264855861664, "grad_norm": 0.04868732765316963, "global_step": 6066, "epoch": 57, "lr": 9.680147038261212e-05} +{"train_loss": 0.020164016634225845, "grad_norm": 0.057916004210710526, "global_step": 6067, "epoch": 57, "lr": 9.680042728631929e-05} +{"train_loss": 0.02118583768606186, "grad_norm": 0.039937086403369904, "global_step": 6068, "epoch": 57, "lr": 9.67993840255901e-05} +{"train_loss": 0.020860545337200165, "grad_norm": 0.07598858326673508, "global_step": 6069, "epoch": 57, "lr": 9.679834060042824e-05} +{"train_loss": 0.01974097266793251, "grad_norm": 0.07380449771881104, "global_step": 6070, "epoch": 57, "lr": 9.679729701083737e-05} +{"train_loss": 0.017255356535315514, "grad_norm": 0.046641040593385696, "global_step": 6071, "epoch": 57, "lr": 9.679625325682114e-05} +{"train_loss": 0.01872926950454712, "grad_norm": 0.06025981158018112, "global_step": 6072, "epoch": 57, "lr": 9.679520933838321e-05} +{"train_loss": 0.017593834549188614, "grad_norm": 0.05476619303226471, "global_step": 6073, "epoch": 57, "lr": 9.67941652555273e-05} +{"train_loss": 0.020302774384617805, "grad_norm": 0.0541972890496254, "global_step": 6074, "epoch": 57, "lr": 9.679312100825702e-05} +{"train_loss": 0.019416015595197678, "grad_norm": 0.11785095930099487, "global_step": 6075, "epoch": 57, "lr": 9.679207659657608e-05} +{"train_loss": 0.017896335572004318, "grad_norm": 0.07136590778827667, "global_step": 6076, "epoch": 57, "lr": 9.679103202048811e-05} +{"train_loss": 0.01960841938853264, "grad_norm": 0.07644553482532501, "global_step": 6077, "epoch": 57, "lr": 9.678998727999684e-05} +{"train_loss": 0.018634675070643425, "grad_norm": 0.05693206563591957, "global_step": 6078, "epoch": 57, "lr": 9.678894237510587e-05} +{"train_loss": 0.019269753247499466, "grad_norm": 0.06172860413789749, "global_step": 6079, "epoch": 57, "lr": 9.678789730581892e-05} +{"train_loss": 0.018721871078014374, "grad_norm": 0.06601755321025848, "global_step": 6080, "epoch": 57, "lr": 9.678685207213962e-05} +{"train_loss": 0.0216005127876997, "grad_norm": 0.13150255382061005, "global_step": 6081, "epoch": 57, "lr": 9.67858066740717e-05} +{"train_loss": 0.021803278475999832, "grad_norm": 0.11601272970438004, "global_step": 6082, "epoch": 57, "lr": 9.678476111161879e-05} +{"train_loss": 0.016777589917182922, "grad_norm": 0.12124371528625488, "global_step": 6083, "epoch": 57, "lr": 9.678371538478457e-05} +{"train_loss": 0.01805039495229721, "grad_norm": 0.11089775711297989, "global_step": 6084, "epoch": 57, "lr": 9.678266949357273e-05} +{"train_loss": 0.016880406066775322, "grad_norm": 0.1250941902399063, "global_step": 6085, "epoch": 57, "lr": 9.678162343798694e-05} +{"train_loss": 0.01738973706960678, "grad_norm": 0.05188009515404701, "global_step": 6086, "epoch": 57, "lr": 9.678057721803084e-05} +{"train_loss": 0.019329655915498734, "grad_norm": 0.05362961068749428, "global_step": 6087, "epoch": 57, "lr": 9.677953083370817e-05} +{"train_loss": 0.017521612346172333, "grad_norm": 0.07649274170398712, "global_step": 6088, "epoch": 57, "lr": 9.677848428502257e-05} +{"train_loss": 0.020253030583262444, "grad_norm": 0.06301556527614594, "global_step": 6089, "epoch": 57, "lr": 9.67774375719777e-05} +{"train_loss": 0.02074388414621353, "grad_norm": 0.07742282748222351, "global_step": 6090, "epoch": 57, "lr": 9.677639069457726e-05} +{"train_loss": 0.02063354291021824, "grad_norm": 0.06506379693746567, "global_step": 6091, "epoch": 57, "lr": 9.677534365282493e-05} +{"train_loss": 0.018261753022670746, "grad_norm": 0.09938664734363556, "global_step": 6092, "epoch": 57, "lr": 9.677429644672439e-05} +{"train_loss": 0.017692651599645615, "grad_norm": 0.1795981377363205, "global_step": 6093, "epoch": 57, "lr": 9.67732490762793e-05} +{"train_loss": 0.019698189571499825, "grad_norm": 0.054604172706604004, "global_step": 6094, "epoch": 57, "lr": 9.677220154149337e-05} +{"train_loss": 0.017289357259869576, "grad_norm": 0.048160240054130554, "global_step": 6095, "epoch": 57, "lr": 9.677115384237025e-05} +{"train_loss": 0.021193699911236763, "grad_norm": 0.13549596071243286, "global_step": 6096, "epoch": 57, "lr": 9.677010597891364e-05} +{"train_loss": 0.018678415566682816, "grad_norm": 0.10260235518217087, "global_step": 6097, "epoch": 57, "lr": 9.676905795112722e-05} +{"train_loss": 0.018406439572572708, "grad_norm": 0.07803617417812347, "global_step": 6098, "epoch": 57, "lr": 9.676800975901467e-05} +{"train_loss": 0.016701769083738327, "grad_norm": 0.06426841765642166, "global_step": 6099, "epoch": 57, "lr": 9.676696140257969e-05} +{"train_loss": 0.017729494720697403, "grad_norm": 0.05818486958742142, "global_step": 6100, "epoch": 57, "lr": 9.676591288182593e-05} +{"train_loss": 0.021295970305800438, "grad_norm": 0.04769670590758324, "global_step": 6101, "epoch": 57, "lr": 9.67648641967571e-05} +{"train_loss": 0.019326170906424522, "grad_norm": 0.0627952516078949, "global_step": 6102, "epoch": 57, "lr": 9.676381534737685e-05} +{"train_loss": 0.02193240262567997, "grad_norm": 0.05621454119682312, "global_step": 6103, "epoch": 57, "lr": 9.676276633368891e-05} +{"train_loss": 0.020648006349802017, "grad_norm": 0.09961649030447006, "global_step": 6104, "epoch": 57, "lr": 9.676171715569693e-05} +{"train_loss": 0.019944339990615845, "grad_norm": 0.13672757148742676, "global_step": 6105, "epoch": 57, "lr": 9.676066781340462e-05} +{"train_loss": 0.022532587870955467, "grad_norm": 0.05778178572654724, "global_step": 6106, "epoch": 57, "lr": 9.675961830681567e-05} +{"train_loss": 0.017185527831315994, "grad_norm": 0.06375911086797714, "global_step": 6107, "epoch": 57, "lr": 9.675856863593376e-05} +{"train_loss": 0.018167031928896904, "grad_norm": 0.054929085075855255, "global_step": 6108, "epoch": 57, "lr": 9.675751880076255e-05} +{"train_loss": 0.02097042091190815, "grad_norm": 0.12934139370918274, "global_step": 6109, "epoch": 57, "lr": 9.675646880130577e-05} +{"train_loss": 0.018692784011363983, "grad_norm": 0.050477348268032074, "global_step": 6110, "epoch": 57, "lr": 9.675541863756708e-05} +{"train_loss": 0.01845654286444187, "grad_norm": 0.09562156349420547, "global_step": 6111, "epoch": 57, "lr": 9.675436830955019e-05} +{"train_loss": 0.018629973754286766, "grad_norm": 0.06331784278154373, "global_step": 6112, "epoch": 57, "lr": 9.675331781725878e-05} +{"train_loss": 0.01830407977104187, "grad_norm": 0.13052445650100708, "global_step": 6113, "epoch": 57, "lr": 9.675226716069656e-05} +{"train_loss": 0.019515525549650192, "grad_norm": 0.10612578690052032, "global_step": 6114, "epoch": 57, "lr": 9.675121633986719e-05} +{"train_loss": 0.021629052236676216, "grad_norm": 0.0788089856505394, "global_step": 6115, "epoch": 57, "lr": 9.675016535477437e-05} +{"train_loss": 0.017318252474069595, "grad_norm": 0.04893484711647034, "global_step": 6116, "epoch": 57, "lr": 9.67491142054218e-05} +{"train_loss": 0.019742172211408615, "grad_norm": 0.07511939853429794, "global_step": 6117, "epoch": 57, "lr": 9.674806289181319e-05} +{"train_loss": 0.017197221517562866, "grad_norm": 0.04888130724430084, "global_step": 6118, "epoch": 57, "lr": 9.674701141395222e-05} +{"train_loss": 0.017260100692510605, "grad_norm": 0.04182014986872673, "global_step": 6119, "epoch": 57, "lr": 9.674595977184256e-05} +{"train_loss": 0.017793430015444756, "grad_norm": 0.041046272963285446, "global_step": 6120, "epoch": 57, "lr": 9.674490796548795e-05} +{"train_loss": 0.01787080056965351, "grad_norm": 0.062492191791534424, "global_step": 6121, "epoch": 57, "lr": 9.674385599489204e-05} +{"train_loss": 0.020408499985933304, "grad_norm": 0.06350879371166229, "global_step": 6122, "epoch": 57, "lr": 9.674280386005856e-05} +{"train_loss": 0.018752411007881165, "grad_norm": 0.10921620577573776, "global_step": 6123, "epoch": 57, "lr": 9.674175156099119e-05} +{"train_loss": 0.019029945135116577, "grad_norm": 0.12055694311857224, "global_step": 6124, "epoch": 57, "lr": 9.674069909769365e-05} +{"train_loss": 0.019932592287659645, "grad_norm": 0.15344399213790894, "global_step": 6125, "epoch": 57, "lr": 9.67396464701696e-05} +{"train_loss": 0.018295347690582275, "grad_norm": 0.0845826119184494, "global_step": 6126, "epoch": 57, "lr": 9.673859367842277e-05} +{"train_loss": 0.020230527967214584, "grad_norm": 0.10670458525419235, "global_step": 6127, "epoch": 57, "lr": 9.673754072245685e-05} +{"train_loss": 0.020149219781160355, "grad_norm": 0.0620107427239418, "global_step": 6128, "epoch": 57, "lr": 9.673648760227552e-05} +{"train_loss": 0.01957569271326065, "grad_norm": 0.08203701674938202, "global_step": 6129, "epoch": 57, "lr": 9.67354343178825e-05} +{"train_loss": 0.016032252460718155, "grad_norm": 0.048771291971206665, "global_step": 6130, "epoch": 57, "lr": 9.67343808692815e-05} +{"train_loss": 0.019530199468135834, "grad_norm": 0.08943495899438858, "global_step": 6131, "epoch": 57, "lr": 9.673332725647621e-05} +{"train_loss": 0.018495425581932068, "grad_norm": 0.04883735626935959, "global_step": 6132, "epoch": 57, "lr": 9.673227347947032e-05} +{"train_loss": 0.018288932740688324, "grad_norm": 0.06114727258682251, "global_step": 6133, "epoch": 57, "lr": 9.673121953826756e-05} +{"train_loss": 0.02090204879641533, "grad_norm": 0.10049799084663391, "global_step": 6134, "epoch": 57, "lr": 9.673016543287162e-05} +{"train_loss": 0.01882515847682953, "grad_norm": 0.13878051936626434, "global_step": 6135, "epoch": 57, "lr": 9.672911116328618e-05} +{"train_loss": 0.01879075914621353, "grad_norm": 0.06843023002147675, "global_step": 6136, "epoch": 57, "lr": 9.672805672951498e-05} +{"train_loss": 0.02047603204846382, "grad_norm": 0.12856292724609375, "global_step": 6137, "epoch": 57, "lr": 9.67270021315617e-05} +{"train_loss": 0.01868579164147377, "grad_norm": 0.05085213854908943, "global_step": 6138, "epoch": 57, "lr": 9.672594736943008e-05} +{"train_loss": 0.018816810101270676, "grad_norm": 0.13509351015090942, "global_step": 6139, "epoch": 57, "lr": 9.672489244312377e-05} +{"train_loss": 0.020088529214262962, "grad_norm": 0.0406482070684433, "global_step": 6140, "epoch": 57, "lr": 9.672383735264652e-05} +{"train_loss": 0.01603403314948082, "grad_norm": 0.09199225902557373, "global_step": 6141, "epoch": 57, "lr": 9.672278209800203e-05} +{"train_loss": 0.019893676042556763, "grad_norm": 0.05502885952591896, "global_step": 6142, "epoch": 57, "lr": 9.6721726679194e-05} +{"train_loss": 0.02074313536286354, "grad_norm": 0.07443853467702866, "global_step": 6143, "epoch": 57, "lr": 9.672067109622614e-05} +{"train_loss": 0.018173672258853912, "grad_norm": 0.17532649636268616, "global_step": 6144, "epoch": 57, "lr": 9.671961534910217e-05} +{"train_loss": 0.01787630282342434, "grad_norm": 0.0962183028459549, "global_step": 6145, "epoch": 57, "lr": 9.671855943782577e-05} +{"train_loss": 0.016165636479854584, "grad_norm": 0.05392591282725334, "global_step": 6146, "epoch": 57, "lr": 9.671750336240069e-05} +{"train_loss": 0.018946573682972846, "grad_norm": 0.07607556879520416, "global_step": 6147, "epoch": 57, "lr": 9.671644712283061e-05, "val_loss": 0.01905697211623192} +{"train_loss": 0.020389340817928314, "grad_norm": 0.060317136347293854, "global_step": 6148, "epoch": 58, "lr": 9.671539071911925e-05} +{"train_loss": 0.018260408192873, "grad_norm": 0.044141098856925964, "global_step": 6149, "epoch": 58, "lr": 9.671433415127033e-05} +{"train_loss": 0.019714683294296265, "grad_norm": 0.059271086007356644, "global_step": 6150, "epoch": 58, "lr": 9.671327741928755e-05} +{"train_loss": 0.01775011606514454, "grad_norm": 0.12442876398563385, "global_step": 6151, "epoch": 58, "lr": 9.671222052317462e-05} +{"train_loss": 0.019400935620069504, "grad_norm": 0.044949404895305634, "global_step": 6152, "epoch": 58, "lr": 9.671116346293527e-05} +{"train_loss": 0.019202101975679398, "grad_norm": 0.05011086165904999, "global_step": 6153, "epoch": 58, "lr": 9.671010623857321e-05} +{"train_loss": 0.01807546615600586, "grad_norm": 0.08811599761247635, "global_step": 6154, "epoch": 58, "lr": 9.670904885009214e-05} +{"train_loss": 0.01975630410015583, "grad_norm": 0.1254679560661316, "global_step": 6155, "epoch": 58, "lr": 9.67079912974958e-05} +{"train_loss": 0.018808990716934204, "grad_norm": 0.07373150438070297, "global_step": 6156, "epoch": 58, "lr": 9.670693358078789e-05} +{"train_loss": 0.021307243034243584, "grad_norm": 0.043549709022045135, "global_step": 6157, "epoch": 58, "lr": 9.670587569997212e-05} +{"train_loss": 0.01914167031645775, "grad_norm": 0.07638055831193924, "global_step": 6158, "epoch": 58, "lr": 9.67048176550522e-05} +{"train_loss": 0.020722685381770134, "grad_norm": 0.06995873898267746, "global_step": 6159, "epoch": 58, "lr": 9.670375944603189e-05} +{"train_loss": 0.0219686608761549, "grad_norm": 0.06630342453718185, "global_step": 6160, "epoch": 58, "lr": 9.670270107291486e-05} +{"train_loss": 0.019957702606916428, "grad_norm": 0.04615335166454315, "global_step": 6161, "epoch": 58, "lr": 9.670164253570487e-05} +{"train_loss": 0.01803339645266533, "grad_norm": 0.10397780686616898, "global_step": 6162, "epoch": 58, "lr": 9.670058383440559e-05} +{"train_loss": 0.02026463858783245, "grad_norm": 0.05163677781820297, "global_step": 6163, "epoch": 58, "lr": 9.669952496902079e-05} +{"train_loss": 0.018158167600631714, "grad_norm": 0.03772797808051109, "global_step": 6164, "epoch": 58, "lr": 9.669846593955417e-05} +{"train_loss": 0.018785551190376282, "grad_norm": 0.056301411241292953, "global_step": 6165, "epoch": 58, "lr": 9.669740674600942e-05} +{"train_loss": 0.020613571628928185, "grad_norm": 0.13326486945152283, "global_step": 6166, "epoch": 58, "lr": 9.669634738839032e-05} +{"train_loss": 0.018474990501999855, "grad_norm": 0.06314762681722641, "global_step": 6167, "epoch": 58, "lr": 9.669528786670055e-05} +{"train_loss": 0.019781969487667084, "grad_norm": 0.08518151938915253, "global_step": 6168, "epoch": 58, "lr": 9.669422818094385e-05} +{"train_loss": 0.018598001450300217, "grad_norm": 0.10137870907783508, "global_step": 6169, "epoch": 58, "lr": 9.669316833112395e-05} +{"train_loss": 0.01894020289182663, "grad_norm": 0.06844906508922577, "global_step": 6170, "epoch": 58, "lr": 9.669210831724454e-05} +{"train_loss": 0.018537145107984543, "grad_norm": 0.100272998213768, "global_step": 6171, "epoch": 58, "lr": 9.669104813930938e-05} +{"train_loss": 0.021385053172707558, "grad_norm": 0.08488099277019501, "global_step": 6172, "epoch": 58, "lr": 9.668998779732218e-05} +{"train_loss": 0.01752265729010105, "grad_norm": 0.05027924105525017, "global_step": 6173, "epoch": 58, "lr": 9.668892729128666e-05} +{"train_loss": 0.019133009016513824, "grad_norm": 0.0611405111849308, "global_step": 6174, "epoch": 58, "lr": 9.668786662120656e-05} +{"train_loss": 0.018006183207035065, "grad_norm": 0.0943136140704155, "global_step": 6175, "epoch": 58, "lr": 9.668680578708562e-05} +{"train_loss": 0.01947222463786602, "grad_norm": 0.057982075959444046, "global_step": 6176, "epoch": 58, "lr": 9.668574478892754e-05} +{"train_loss": 0.020539939403533936, "grad_norm": 0.08419860154390335, "global_step": 6177, "epoch": 58, "lr": 9.668468362673604e-05} +{"train_loss": 0.017265552654862404, "grad_norm": 0.09941205382347107, "global_step": 6178, "epoch": 58, "lr": 9.668362230051487e-05} +{"train_loss": 0.015404443256556988, "grad_norm": 0.04692278429865837, "global_step": 6179, "epoch": 58, "lr": 9.668256081026775e-05} +{"train_loss": 0.017680402845144272, "grad_norm": 0.03925420343875885, "global_step": 6180, "epoch": 58, "lr": 9.668149915599843e-05} +{"train_loss": 0.019733652472496033, "grad_norm": 0.04188511520624161, "global_step": 6181, "epoch": 58, "lr": 9.66804373377106e-05} +{"train_loss": 0.021072443574666977, "grad_norm": 0.08877693861722946, "global_step": 6182, "epoch": 58, "lr": 9.667937535540802e-05} +{"train_loss": 0.019373498857021332, "grad_norm": 0.05685795471072197, "global_step": 6183, "epoch": 58, "lr": 9.667831320909443e-05} +{"train_loss": 0.01947208121418953, "grad_norm": 0.07590309530496597, "global_step": 6184, "epoch": 58, "lr": 9.667725089877353e-05} +{"train_loss": 0.0173155777156353, "grad_norm": 0.13852031528949738, "global_step": 6185, "epoch": 58, "lr": 9.667618842444909e-05} +{"train_loss": 0.016512151807546616, "grad_norm": 0.06748370081186295, "global_step": 6186, "epoch": 58, "lr": 9.667512578612481e-05} +{"train_loss": 0.01951003260910511, "grad_norm": 0.13453799486160278, "global_step": 6187, "epoch": 58, "lr": 9.667406298380444e-05} +{"train_loss": 0.017492402344942093, "grad_norm": 0.05082429200410843, "global_step": 6188, "epoch": 58, "lr": 9.66730000174917e-05} +{"train_loss": 0.017341721802949905, "grad_norm": 0.07777293026447296, "global_step": 6189, "epoch": 58, "lr": 9.667193688719034e-05} +{"train_loss": 0.01697954535484314, "grad_norm": 0.09648332744836807, "global_step": 6190, "epoch": 58, "lr": 9.667087359290409e-05} +{"train_loss": 0.02190548926591873, "grad_norm": 0.09963272511959076, "global_step": 6191, "epoch": 58, "lr": 9.66698101346367e-05} +{"train_loss": 0.019432472065091133, "grad_norm": 0.1009429320693016, "global_step": 6192, "epoch": 58, "lr": 9.666874651239188e-05} +{"train_loss": 0.019235413521528244, "grad_norm": 0.04208042100071907, "global_step": 6193, "epoch": 58, "lr": 9.66676827261734e-05} +{"train_loss": 0.020345522090792656, "grad_norm": 0.054073065519332886, "global_step": 6194, "epoch": 58, "lr": 9.666661877598496e-05} +{"train_loss": 0.016776736825704575, "grad_norm": 0.07412603497505188, "global_step": 6195, "epoch": 58, "lr": 9.666555466183031e-05} +{"train_loss": 0.021200373768806458, "grad_norm": 0.09237419068813324, "global_step": 6196, "epoch": 58, "lr": 9.666449038371322e-05} +{"train_loss": 0.018511876463890076, "grad_norm": 0.11277934163808823, "global_step": 6197, "epoch": 58, "lr": 9.666342594163738e-05} +{"train_loss": 0.01987285166978836, "grad_norm": 0.04481808468699455, "global_step": 6198, "epoch": 58, "lr": 9.666236133560655e-05} +{"train_loss": 0.015875142067670822, "grad_norm": 0.04751746729016304, "global_step": 6199, "epoch": 58, "lr": 9.66612965656245e-05} +{"train_loss": 0.01837727427482605, "grad_norm": 0.08229244500398636, "global_step": 6200, "epoch": 58, "lr": 9.666023163169493e-05} +{"train_loss": 0.02014295570552349, "grad_norm": 0.09352036565542221, "global_step": 6201, "epoch": 58, "lr": 9.665916653382162e-05} +{"train_loss": 0.01936265453696251, "grad_norm": 0.0625058263540268, "global_step": 6202, "epoch": 58, "lr": 9.665810127200826e-05} +{"train_loss": 0.019355230033397675, "grad_norm": 0.09368361532688141, "global_step": 6203, "epoch": 58, "lr": 9.665703584625864e-05} +{"train_loss": 0.017814109101891518, "grad_norm": 0.13294285535812378, "global_step": 6204, "epoch": 58, "lr": 9.665597025657649e-05} +{"train_loss": 0.02113024890422821, "grad_norm": 0.055496010929346085, "global_step": 6205, "epoch": 58, "lr": 9.665490450296554e-05} +{"train_loss": 0.019370298832654953, "grad_norm": 0.053431436419487, "global_step": 6206, "epoch": 58, "lr": 9.665383858542954e-05} +{"train_loss": 0.01757545955479145, "grad_norm": 0.12511223554611206, "global_step": 6207, "epoch": 58, "lr": 9.665277250397226e-05} +{"train_loss": 0.021435029804706573, "grad_norm": 0.046571120619773865, "global_step": 6208, "epoch": 58, "lr": 9.66517062585974e-05} +{"train_loss": 0.018106631934642792, "grad_norm": 0.10920815914869308, "global_step": 6209, "epoch": 58, "lr": 9.665063984930875e-05} +{"train_loss": 0.01739918813109398, "grad_norm": 0.06345284730195999, "global_step": 6210, "epoch": 58, "lr": 9.664957327611004e-05} +{"train_loss": 0.018160425126552582, "grad_norm": 0.05456649884581566, "global_step": 6211, "epoch": 58, "lr": 9.664850653900502e-05} +{"train_loss": 0.02006460167467594, "grad_norm": 0.13035793602466583, "global_step": 6212, "epoch": 58, "lr": 9.664743963799743e-05} +{"train_loss": 0.018792789429426193, "grad_norm": 0.08635314553976059, "global_step": 6213, "epoch": 58, "lr": 9.664637257309103e-05} +{"train_loss": 0.021178901195526123, "grad_norm": 0.05658789351582527, "global_step": 6214, "epoch": 58, "lr": 9.664530534428955e-05} +{"train_loss": 0.018114149570465088, "grad_norm": 0.04856778681278229, "global_step": 6215, "epoch": 58, "lr": 9.664423795159676e-05} +{"train_loss": 0.016761809587478638, "grad_norm": 0.09065905958414078, "global_step": 6216, "epoch": 58, "lr": 9.66431703950164e-05} +{"train_loss": 0.016749761998653412, "grad_norm": 0.052693936973810196, "global_step": 6217, "epoch": 58, "lr": 9.664210267455222e-05} +{"train_loss": 0.017245756462216377, "grad_norm": 0.08431493490934372, "global_step": 6218, "epoch": 58, "lr": 9.664103479020799e-05} +{"train_loss": 0.01767737604677677, "grad_norm": 0.14981456100940704, "global_step": 6219, "epoch": 58, "lr": 9.663996674198743e-05} +{"train_loss": 0.018399611115455627, "grad_norm": 0.20041270554065704, "global_step": 6220, "epoch": 58, "lr": 9.663889852989432e-05} +{"train_loss": 0.017471734434366226, "grad_norm": 0.09271134436130524, "global_step": 6221, "epoch": 58, "lr": 9.66378301539324e-05} +{"train_loss": 0.01826236955821514, "grad_norm": 0.08076855540275574, "global_step": 6222, "epoch": 58, "lr": 9.663676161410543e-05} +{"train_loss": 0.019116999581456184, "grad_norm": 0.05915793403983116, "global_step": 6223, "epoch": 58, "lr": 9.663569291041717e-05} +{"train_loss": 0.02023247256875038, "grad_norm": 0.04525191709399223, "global_step": 6224, "epoch": 58, "lr": 9.663462404287136e-05} +{"train_loss": 0.018217846751213074, "grad_norm": 0.0761793702840805, "global_step": 6225, "epoch": 58, "lr": 9.663355501147175e-05} +{"train_loss": 0.01951981894671917, "grad_norm": 0.08546889573335648, "global_step": 6226, "epoch": 58, "lr": 9.663248581622212e-05} +{"train_loss": 0.018531309440732002, "grad_norm": 0.050538670271635056, "global_step": 6227, "epoch": 58, "lr": 9.663141645712621e-05} +{"train_loss": 0.01879311352968216, "grad_norm": 0.041678428649902344, "global_step": 6228, "epoch": 58, "lr": 9.663034693418777e-05} +{"train_loss": 0.017694659531116486, "grad_norm": 0.06317058950662613, "global_step": 6229, "epoch": 58, "lr": 9.66292772474106e-05} +{"train_loss": 0.02076617442071438, "grad_norm": 0.18151071667671204, "global_step": 6230, "epoch": 58, "lr": 9.662820739679841e-05} +{"train_loss": 0.01917530596256256, "grad_norm": 0.13920602202415466, "global_step": 6231, "epoch": 58, "lr": 9.662713738235497e-05} +{"train_loss": 0.016396336257457733, "grad_norm": 0.1411190778017044, "global_step": 6232, "epoch": 58, "lr": 9.662606720408406e-05} +{"train_loss": 0.01931339129805565, "grad_norm": 0.10625357925891876, "global_step": 6233, "epoch": 58, "lr": 9.662499686198943e-05} +{"train_loss": 0.018359126523137093, "grad_norm": 0.07279107719659805, "global_step": 6234, "epoch": 58, "lr": 9.662392635607483e-05} +{"train_loss": 0.020398948341608047, "grad_norm": 0.19730466604232788, "global_step": 6235, "epoch": 58, "lr": 9.662285568634402e-05} +{"train_loss": 0.01898244582116604, "grad_norm": 0.22217391431331635, "global_step": 6236, "epoch": 58, "lr": 9.662178485280077e-05} +{"train_loss": 0.018802227452397346, "grad_norm": 0.046396348625421524, "global_step": 6237, "epoch": 58, "lr": 9.662071385544887e-05} +{"train_loss": 0.019479690119624138, "grad_norm": 0.04835841432213783, "global_step": 6238, "epoch": 58, "lr": 9.661964269429204e-05} +{"train_loss": 0.020004916936159134, "grad_norm": 0.05939529091119766, "global_step": 6239, "epoch": 58, "lr": 9.661857136933405e-05} +{"train_loss": 0.017647311091423035, "grad_norm": 0.09306101500988007, "global_step": 6240, "epoch": 58, "lr": 9.661749988057869e-05} +{"train_loss": 0.01894926279783249, "grad_norm": 0.2061295360326767, "global_step": 6241, "epoch": 58, "lr": 9.66164282280297e-05} +{"train_loss": 0.01918991655111313, "grad_norm": 0.17372572422027588, "global_step": 6242, "epoch": 58, "lr": 9.661535641169086e-05} +{"train_loss": 0.016870884224772453, "grad_norm": 0.08439885824918747, "global_step": 6243, "epoch": 58, "lr": 9.661428443156593e-05} +{"train_loss": 0.016552213579416275, "grad_norm": 0.08280948549509048, "global_step": 6244, "epoch": 58, "lr": 9.661321228765868e-05} +{"train_loss": 0.01850089803338051, "grad_norm": 0.05513910949230194, "global_step": 6245, "epoch": 58, "lr": 9.661213997997285e-05} +{"train_loss": 0.018791012465953827, "grad_norm": 0.09611967206001282, "global_step": 6246, "epoch": 58, "lr": 9.661106750851225e-05} +{"train_loss": 0.021131914108991623, "grad_norm": 0.15027080476284027, "global_step": 6247, "epoch": 58, "lr": 9.660999487328062e-05} +{"train_loss": 0.020795632153749466, "grad_norm": 0.07294675707817078, "global_step": 6248, "epoch": 58, "lr": 9.660892207428176e-05} +{"train_loss": 0.02050577662885189, "grad_norm": 0.050861917436122894, "global_step": 6249, "epoch": 58, "lr": 9.66078491115194e-05} +{"train_loss": 0.020576870068907738, "grad_norm": 0.04419810324907303, "global_step": 6250, "epoch": 58, "lr": 9.660677598499732e-05} +{"train_loss": 0.01605920121073723, "grad_norm": 0.06638849526643753, "global_step": 6251, "epoch": 58, "lr": 9.660570269471931e-05} +{"train_loss": 0.019633885473012924, "grad_norm": 0.06345030665397644, "global_step": 6252, "epoch": 58, "lr": 9.660462924068913e-05} +{"train_loss": 0.018884832856579208, "grad_norm": 0.07370153069496155, "global_step": 6253, "epoch": 58, "lr": 9.660355562291055e-05, "val_loss": 0.01916937343776226} +{"train_loss": 0.019148312509059906, "grad_norm": 0.044893112033605576, "global_step": 6254, "epoch": 59, "lr": 9.660248184138734e-05} +{"train_loss": 0.019971733912825584, "grad_norm": 0.04221239313483238, "global_step": 6255, "epoch": 59, "lr": 9.660140789612328e-05} +{"train_loss": 0.020783834159374237, "grad_norm": 0.12787215411663055, "global_step": 6256, "epoch": 59, "lr": 9.660033378712215e-05} +{"train_loss": 0.019553204998373985, "grad_norm": 0.06075034663081169, "global_step": 6257, "epoch": 59, "lr": 9.659925951438768e-05} +{"train_loss": 0.018635224550962448, "grad_norm": 0.046818509697914124, "global_step": 6258, "epoch": 59, "lr": 9.65981850779237e-05} +{"train_loss": 0.0182182714343071, "grad_norm": 0.06478181481361389, "global_step": 6259, "epoch": 59, "lr": 9.659711047773396e-05} +{"train_loss": 0.022027619183063507, "grad_norm": 0.0936707928776741, "global_step": 6260, "epoch": 59, "lr": 9.659603571382225e-05} +{"train_loss": 0.01993817836046219, "grad_norm": 0.10480169206857681, "global_step": 6261, "epoch": 59, "lr": 9.659496078619231e-05} +{"train_loss": 0.019332967698574066, "grad_norm": 0.07360917329788208, "global_step": 6262, "epoch": 59, "lr": 9.659388569484795e-05} +{"train_loss": 0.01917242258787155, "grad_norm": 0.08250298351049423, "global_step": 6263, "epoch": 59, "lr": 9.659281043979295e-05} +{"train_loss": 0.019372424110770226, "grad_norm": 0.07572288066148758, "global_step": 6264, "epoch": 59, "lr": 9.659173502103107e-05} +{"train_loss": 0.017687618732452393, "grad_norm": 0.06882049143314362, "global_step": 6265, "epoch": 59, "lr": 9.659065943856609e-05} +{"train_loss": 0.015924232080578804, "grad_norm": 0.04192164167761803, "global_step": 6266, "epoch": 59, "lr": 9.658958369240179e-05} +{"train_loss": 0.01797666773200035, "grad_norm": 0.07876800000667572, "global_step": 6267, "epoch": 59, "lr": 9.658850778254197e-05} +{"train_loss": 0.019154466688632965, "grad_norm": 0.05018099769949913, "global_step": 6268, "epoch": 59, "lr": 9.658743170899039e-05} +{"train_loss": 0.02014322206377983, "grad_norm": 0.07990439236164093, "global_step": 6269, "epoch": 59, "lr": 9.658635547175083e-05} +{"train_loss": 0.01750265061855316, "grad_norm": 0.04916989430785179, "global_step": 6270, "epoch": 59, "lr": 9.658527907082709e-05} +{"train_loss": 0.01713547855615616, "grad_norm": 0.1161675974726677, "global_step": 6271, "epoch": 59, "lr": 9.658420250622293e-05} +{"train_loss": 0.020173562690615654, "grad_norm": 0.06928851455450058, "global_step": 6272, "epoch": 59, "lr": 9.658312577794213e-05} +{"train_loss": 0.017795413732528687, "grad_norm": 0.14949853718280792, "global_step": 6273, "epoch": 59, "lr": 9.65820488859885e-05} +{"train_loss": 0.01687554270029068, "grad_norm": 0.07041876018047333, "global_step": 6274, "epoch": 59, "lr": 9.65809718303658e-05} +{"train_loss": 0.01856626570224762, "grad_norm": 0.08600728213787079, "global_step": 6275, "epoch": 59, "lr": 9.657989461107782e-05} +{"train_loss": 0.01595252752304077, "grad_norm": 0.04205522686243057, "global_step": 6276, "epoch": 59, "lr": 9.657881722812836e-05} +{"train_loss": 0.019059747457504272, "grad_norm": 0.06387237459421158, "global_step": 6277, "epoch": 59, "lr": 9.657773968152117e-05} +{"train_loss": 0.0174734927713871, "grad_norm": 0.06044723466038704, "global_step": 6278, "epoch": 59, "lr": 9.657666197126008e-05} +{"train_loss": 0.017118722200393677, "grad_norm": 0.06316495686769485, "global_step": 6279, "epoch": 59, "lr": 9.657558409734886e-05} +{"train_loss": 0.018455082550644875, "grad_norm": 0.05709384009242058, "global_step": 6280, "epoch": 59, "lr": 9.657450605979128e-05} +{"train_loss": 0.017158880829811096, "grad_norm": 0.06895329803228378, "global_step": 6281, "epoch": 59, "lr": 9.657342785859114e-05} +{"train_loss": 0.017081625759601593, "grad_norm": 0.04886772483587265, "global_step": 6282, "epoch": 59, "lr": 9.657234949375223e-05} +{"train_loss": 0.017685718834400177, "grad_norm": 0.04854264855384827, "global_step": 6283, "epoch": 59, "lr": 9.657127096527834e-05} +{"train_loss": 0.02099122479557991, "grad_norm": 0.051045652478933334, "global_step": 6284, "epoch": 59, "lr": 9.657019227317326e-05} +{"train_loss": 0.019460033625364304, "grad_norm": 0.047084543853998184, "global_step": 6285, "epoch": 59, "lr": 9.656911341744077e-05} +{"train_loss": 0.019543930888175964, "grad_norm": 0.05941863730549812, "global_step": 6286, "epoch": 59, "lr": 9.656803439808466e-05} +{"train_loss": 0.019387369975447655, "grad_norm": 0.05973055958747864, "global_step": 6287, "epoch": 59, "lr": 9.656695521510875e-05} +{"train_loss": 0.0202749352902174, "grad_norm": 0.07885783165693283, "global_step": 6288, "epoch": 59, "lr": 9.65658758685168e-05} +{"train_loss": 0.01995808631181717, "grad_norm": 0.1481252759695053, "global_step": 6289, "epoch": 59, "lr": 9.656479635831262e-05} +{"train_loss": 0.02053162083029747, "grad_norm": 0.04278777539730072, "global_step": 6290, "epoch": 59, "lr": 9.656371668449998e-05} +{"train_loss": 0.019184168428182602, "grad_norm": 0.05463740602135658, "global_step": 6291, "epoch": 59, "lr": 9.656263684708269e-05} +{"train_loss": 0.018741168081760406, "grad_norm": 0.03932289034128189, "global_step": 6292, "epoch": 59, "lr": 9.656155684606457e-05} +{"train_loss": 0.020809482783079147, "grad_norm": 0.07609745115041733, "global_step": 6293, "epoch": 59, "lr": 9.656047668144937e-05} +{"train_loss": 0.019336901605129242, "grad_norm": 0.11871693283319473, "global_step": 6294, "epoch": 59, "lr": 9.65593963532409e-05} +{"train_loss": 0.02173655852675438, "grad_norm": 0.12982328236103058, "global_step": 6295, "epoch": 59, "lr": 9.655831586144298e-05} +{"train_loss": 0.019194986671209335, "grad_norm": 0.10658984631299973, "global_step": 6296, "epoch": 59, "lr": 9.655723520605935e-05} +{"train_loss": 0.01707892119884491, "grad_norm": 0.060441285371780396, "global_step": 6297, "epoch": 59, "lr": 9.655615438709386e-05} +{"train_loss": 0.01760891079902649, "grad_norm": 0.05928569287061691, "global_step": 6298, "epoch": 59, "lr": 9.65550734045503e-05} +{"train_loss": 0.01919633522629738, "grad_norm": 0.10577905178070068, "global_step": 6299, "epoch": 59, "lr": 9.655399225843245e-05} +{"train_loss": 0.017324505373835564, "grad_norm": 0.05500185117125511, "global_step": 6300, "epoch": 59, "lr": 9.655291094874412e-05} +{"train_loss": 0.01859883964061737, "grad_norm": 0.05972639471292496, "global_step": 6301, "epoch": 59, "lr": 9.655182947548911e-05} +{"train_loss": 0.018885252997279167, "grad_norm": 0.057066090404987335, "global_step": 6302, "epoch": 59, "lr": 9.655074783867121e-05} +{"train_loss": 0.019523970782756805, "grad_norm": 0.10871735960245132, "global_step": 6303, "epoch": 59, "lr": 9.654966603829423e-05} +{"train_loss": 0.019311705604195595, "grad_norm": 0.08318538963794708, "global_step": 6304, "epoch": 59, "lr": 9.654858407436195e-05} +{"train_loss": 0.019847765564918518, "grad_norm": 0.06425198912620544, "global_step": 6305, "epoch": 59, "lr": 9.654750194687822e-05} +{"train_loss": 0.020918896421790123, "grad_norm": 0.13245421648025513, "global_step": 6306, "epoch": 59, "lr": 9.654641965584678e-05} +{"train_loss": 0.017645908519625664, "grad_norm": 0.04838518425822258, "global_step": 6307, "epoch": 59, "lr": 9.65453372012715e-05} +{"train_loss": 0.018500622361898422, "grad_norm": 0.0440608374774456, "global_step": 6308, "epoch": 59, "lr": 9.654425458315611e-05} +{"train_loss": 0.0188443586230278, "grad_norm": 0.05928847938776016, "global_step": 6309, "epoch": 59, "lr": 9.654317180150447e-05} +{"train_loss": 0.017439164221286774, "grad_norm": 0.06210404261946678, "global_step": 6310, "epoch": 59, "lr": 9.654208885632037e-05} +{"train_loss": 0.018364790827035904, "grad_norm": 0.06073734909296036, "global_step": 6311, "epoch": 59, "lr": 9.65410057476076e-05} +{"train_loss": 0.018744703382253647, "grad_norm": 0.07308518141508102, "global_step": 6312, "epoch": 59, "lr": 9.653992247536998e-05} +{"train_loss": 0.021685075014829636, "grad_norm": 0.0987151563167572, "global_step": 6313, "epoch": 59, "lr": 9.653883903961131e-05} +{"train_loss": 0.018012160435318947, "grad_norm": 0.10797141492366791, "global_step": 6314, "epoch": 59, "lr": 9.65377554403354e-05} +{"train_loss": 0.02307107299566269, "grad_norm": 0.05894690006971359, "global_step": 6315, "epoch": 59, "lr": 9.653667167754606e-05} +{"train_loss": 0.018552355468273163, "grad_norm": 0.04764028266072273, "global_step": 6316, "epoch": 59, "lr": 9.653558775124708e-05} +{"train_loss": 0.019647693261504173, "grad_norm": 0.045095279812812805, "global_step": 6317, "epoch": 59, "lr": 9.65345036614423e-05} +{"train_loss": 0.01769379898905754, "grad_norm": 0.04795161262154579, "global_step": 6318, "epoch": 59, "lr": 9.653341940813549e-05} +{"train_loss": 0.022280387580394745, "grad_norm": 0.07429434359073639, "global_step": 6319, "epoch": 59, "lr": 9.65323349913305e-05} +{"train_loss": 0.018888071179389954, "grad_norm": 0.0502534955739975, "global_step": 6320, "epoch": 59, "lr": 9.653125041103112e-05} +{"train_loss": 0.016874799504876137, "grad_norm": 0.04947422072291374, "global_step": 6321, "epoch": 59, "lr": 9.653016566724115e-05} +{"train_loss": 0.019662510603666306, "grad_norm": 0.07085662335157394, "global_step": 6322, "epoch": 59, "lr": 9.652908075996443e-05} +{"train_loss": 0.018254123628139496, "grad_norm": 0.08579292893409729, "global_step": 6323, "epoch": 59, "lr": 9.652799568920474e-05} +{"train_loss": 0.01876690797507763, "grad_norm": 0.05776110664010048, "global_step": 6324, "epoch": 59, "lr": 9.652691045496591e-05} +{"train_loss": 0.01774151250720024, "grad_norm": 0.057907406240701675, "global_step": 6325, "epoch": 59, "lr": 9.652582505725177e-05} +{"train_loss": 0.018410684540867805, "grad_norm": 0.1033547967672348, "global_step": 6326, "epoch": 59, "lr": 9.652473949606608e-05} +{"train_loss": 0.02017255686223507, "grad_norm": 0.09211454540491104, "global_step": 6327, "epoch": 59, "lr": 9.652365377141272e-05} +{"train_loss": 0.02084217220544815, "grad_norm": 0.23728784918785095, "global_step": 6328, "epoch": 59, "lr": 9.652256788329545e-05} +{"train_loss": 0.019390640780329704, "grad_norm": 0.054107602685689926, "global_step": 6329, "epoch": 59, "lr": 9.652148183171813e-05} +{"train_loss": 0.02044367790222168, "grad_norm": 0.17910516262054443, "global_step": 6330, "epoch": 59, "lr": 9.652039561668453e-05} +{"train_loss": 0.016062501817941666, "grad_norm": 0.05683908239006996, "global_step": 6331, "epoch": 59, "lr": 9.651930923819851e-05} +{"train_loss": 0.021152596920728683, "grad_norm": 0.03916151449084282, "global_step": 6332, "epoch": 59, "lr": 9.651822269626387e-05} +{"train_loss": 0.015488396398723125, "grad_norm": 0.07468795776367188, "global_step": 6333, "epoch": 59, "lr": 9.651713599088442e-05} +{"train_loss": 0.017739716917276382, "grad_norm": 0.1141941174864769, "global_step": 6334, "epoch": 59, "lr": 9.651604912206398e-05} +{"train_loss": 0.021095840260386467, "grad_norm": 0.08348574489355087, "global_step": 6335, "epoch": 59, "lr": 9.651496208980637e-05} +{"train_loss": 0.01930181123316288, "grad_norm": 0.10652963817119598, "global_step": 6336, "epoch": 59, "lr": 9.651387489411542e-05} +{"train_loss": 0.017445232719182968, "grad_norm": 0.038185250014066696, "global_step": 6337, "epoch": 59, "lr": 9.651278753499494e-05} +{"train_loss": 0.01640150137245655, "grad_norm": 0.061155449599027634, "global_step": 6338, "epoch": 59, "lr": 9.651170001244876e-05} +{"train_loss": 0.022567331790924072, "grad_norm": 0.1213078498840332, "global_step": 6339, "epoch": 59, "lr": 9.651061232648068e-05} +{"train_loss": 0.019115526229143143, "grad_norm": 0.09914936870336533, "global_step": 6340, "epoch": 59, "lr": 9.650952447709453e-05} +{"train_loss": 0.018519986420869827, "grad_norm": 0.10418283939361572, "global_step": 6341, "epoch": 59, "lr": 9.650843646429416e-05} +{"train_loss": 0.019508030265569687, "grad_norm": 0.1008390560746193, "global_step": 6342, "epoch": 59, "lr": 9.650734828808335e-05} +{"train_loss": 0.017101718112826347, "grad_norm": 0.07205434143543243, "global_step": 6343, "epoch": 59, "lr": 9.650625994846597e-05} +{"train_loss": 0.017062462866306305, "grad_norm": 0.09638567268848419, "global_step": 6344, "epoch": 59, "lr": 9.650517144544578e-05} +{"train_loss": 0.01570555567741394, "grad_norm": 0.05169457197189331, "global_step": 6345, "epoch": 59, "lr": 9.650408277902667e-05} +{"train_loss": 0.016974210739135742, "grad_norm": 0.06549529731273651, "global_step": 6346, "epoch": 59, "lr": 9.650299394921242e-05} +{"train_loss": 0.018321415409445763, "grad_norm": 0.09706075489521027, "global_step": 6347, "epoch": 59, "lr": 9.650190495600689e-05} +{"train_loss": 0.01900586113333702, "grad_norm": 0.12261872738599777, "global_step": 6348, "epoch": 59, "lr": 9.650081579941388e-05} +{"train_loss": 0.01834416203200817, "grad_norm": 0.08726949989795685, "global_step": 6349, "epoch": 59, "lr": 9.649972647943721e-05} +{"train_loss": 0.017024699598550797, "grad_norm": 0.05870191380381584, "global_step": 6350, "epoch": 59, "lr": 9.649863699608074e-05} +{"train_loss": 0.019926153123378754, "grad_norm": 0.08529188483953476, "global_step": 6351, "epoch": 59, "lr": 9.64975473493483e-05} +{"train_loss": 0.020636070519685745, "grad_norm": 0.04652557522058487, "global_step": 6352, "epoch": 59, "lr": 9.649645753924368e-05} +{"train_loss": 0.019574474543333054, "grad_norm": 0.04627557471394539, "global_step": 6353, "epoch": 59, "lr": 9.649536756577072e-05} +{"train_loss": 0.018746091052889824, "grad_norm": 0.108465775847435, "global_step": 6354, "epoch": 59, "lr": 9.649427742893326e-05} +{"train_loss": 0.017664050683379173, "grad_norm": 0.058444343507289886, "global_step": 6355, "epoch": 59, "lr": 9.649318712873515e-05} +{"train_loss": 0.015407216735184193, "grad_norm": 0.05378919839859009, "global_step": 6356, "epoch": 59, "lr": 9.649209666518018e-05} +{"train_loss": 0.018153224140405655, "grad_norm": 0.047311920672655106, "global_step": 6357, "epoch": 59, "lr": 9.649100603827221e-05} +{"train_loss": 0.0196811743080616, "grad_norm": 0.043371282517910004, "global_step": 6358, "epoch": 59, "lr": 9.648991524801508e-05} +{"train_loss": 0.018834193711573223, "grad_norm": 0.07448741048574448, "global_step": 6359, "epoch": 59, "lr": 9.648882429441257e-05, "val_loss": 0.01872766762971878} +{"train_loss": 0.019169485196471214, "grad_norm": 0.05465255677700043, "global_step": 6360, "epoch": 60, "lr": 9.648773317746858e-05} +{"train_loss": 0.01859825849533081, "grad_norm": 0.044428449124097824, "global_step": 6361, "epoch": 60, "lr": 9.648664189718689e-05} +{"train_loss": 0.018976902589201927, "grad_norm": 0.10244778543710709, "global_step": 6362, "epoch": 60, "lr": 9.648555045357138e-05} +{"train_loss": 0.019423235207796097, "grad_norm": 0.06159462407231331, "global_step": 6363, "epoch": 60, "lr": 9.648445884662584e-05} +{"train_loss": 0.016233358532190323, "grad_norm": 0.05781333148479462, "global_step": 6364, "epoch": 60, "lr": 9.648336707635413e-05} +{"train_loss": 0.01744154468178749, "grad_norm": 0.06309371441602707, "global_step": 6365, "epoch": 60, "lr": 9.648227514276009e-05} +{"train_loss": 0.018187399953603745, "grad_norm": 0.0591137669980526, "global_step": 6366, "epoch": 60, "lr": 9.648118304584755e-05} +{"train_loss": 0.018781697377562523, "grad_norm": 0.04147664085030556, "global_step": 6367, "epoch": 60, "lr": 9.648009078562035e-05} +{"train_loss": 0.01680791936814785, "grad_norm": 0.13703399896621704, "global_step": 6368, "epoch": 60, "lr": 9.64789983620823e-05} +{"train_loss": 0.017970170825719833, "grad_norm": 0.07460180670022964, "global_step": 6369, "epoch": 60, "lr": 9.647790577523729e-05} +{"train_loss": 0.01788623258471489, "grad_norm": 0.04669160023331642, "global_step": 6370, "epoch": 60, "lr": 9.647681302508914e-05} +{"train_loss": 0.019190631806850433, "grad_norm": 0.04727378487586975, "global_step": 6371, "epoch": 60, "lr": 9.647572011164165e-05} +{"train_loss": 0.0211813822388649, "grad_norm": 0.08778420835733414, "global_step": 6372, "epoch": 60, "lr": 9.64746270348987e-05} +{"train_loss": 0.021263789385557175, "grad_norm": 0.10177424550056458, "global_step": 6373, "epoch": 60, "lr": 9.647353379486414e-05} +{"train_loss": 0.022991135716438293, "grad_norm": 0.09807008504867554, "global_step": 6374, "epoch": 60, "lr": 9.647244039154178e-05} +{"train_loss": 0.016619477421045303, "grad_norm": 0.046684686094522476, "global_step": 6375, "epoch": 60, "lr": 9.647134682493548e-05} +{"train_loss": 0.018364455550909042, "grad_norm": 0.0933942124247551, "global_step": 6376, "epoch": 60, "lr": 9.647025309504907e-05} +{"train_loss": 0.019596826285123825, "grad_norm": 0.13589714467525482, "global_step": 6377, "epoch": 60, "lr": 9.646915920188639e-05} +{"train_loss": 0.01698172837495804, "grad_norm": 0.05776064470410347, "global_step": 6378, "epoch": 60, "lr": 9.646806514545132e-05} +{"train_loss": 0.016223035752773285, "grad_norm": 0.07593190670013428, "global_step": 6379, "epoch": 60, "lr": 9.646697092574766e-05} +{"train_loss": 0.018055830150842667, "grad_norm": 0.12584149837493896, "global_step": 6380, "epoch": 60, "lr": 9.646587654277927e-05} +{"train_loss": 0.017362909391522408, "grad_norm": 0.062455371022224426, "global_step": 6381, "epoch": 60, "lr": 9.646478199655e-05} +{"train_loss": 0.020041972398757935, "grad_norm": 0.06860342621803284, "global_step": 6382, "epoch": 60, "lr": 9.646368728706372e-05} +{"train_loss": 0.021719427779316902, "grad_norm": 0.08050201833248138, "global_step": 6383, "epoch": 60, "lr": 9.64625924143242e-05} +{"train_loss": 0.018698371946811676, "grad_norm": 0.07295811921358109, "global_step": 6384, "epoch": 60, "lr": 9.646149737833537e-05} +{"train_loss": 0.017454177141189575, "grad_norm": 0.04381842911243439, "global_step": 6385, "epoch": 60, "lr": 9.646040217910104e-05} +{"train_loss": 0.02065110206604004, "grad_norm": 0.11797674745321274, "global_step": 6386, "epoch": 60, "lr": 9.645930681662506e-05} +{"train_loss": 0.0194377638399601, "grad_norm": 0.07048813998699188, "global_step": 6387, "epoch": 60, "lr": 9.645821129091128e-05} +{"train_loss": 0.017265096306800842, "grad_norm": 0.046493250876665115, "global_step": 6388, "epoch": 60, "lr": 9.645711560196355e-05} +{"train_loss": 0.01993476413190365, "grad_norm": 0.04971899464726448, "global_step": 6389, "epoch": 60, "lr": 9.645601974978571e-05} +{"train_loss": 0.01819509081542492, "grad_norm": 0.042582135647535324, "global_step": 6390, "epoch": 60, "lr": 9.645492373438162e-05} +{"train_loss": 0.019236549735069275, "grad_norm": 0.07180647552013397, "global_step": 6391, "epoch": 60, "lr": 9.645382755575514e-05} +{"train_loss": 0.018137626349925995, "grad_norm": 0.0595991276204586, "global_step": 6392, "epoch": 60, "lr": 9.645273121391012e-05} +{"train_loss": 0.01638166606426239, "grad_norm": 0.05152827873826027, "global_step": 6393, "epoch": 60, "lr": 9.64516347088504e-05} +{"train_loss": 0.01907969079911709, "grad_norm": 0.09747003763914108, "global_step": 6394, "epoch": 60, "lr": 9.645053804057982e-05} +{"train_loss": 0.02075010910630226, "grad_norm": 0.05034545063972473, "global_step": 6395, "epoch": 60, "lr": 9.644944120910226e-05} +{"train_loss": 0.01738087460398674, "grad_norm": 0.059574756771326065, "global_step": 6396, "epoch": 60, "lr": 9.644834421442157e-05} +{"train_loss": 0.01803462952375412, "grad_norm": 0.05549801141023636, "global_step": 6397, "epoch": 60, "lr": 9.64472470565416e-05} +{"train_loss": 0.022926820442080498, "grad_norm": 0.12440195679664612, "global_step": 6398, "epoch": 60, "lr": 9.644614973546619e-05} +{"train_loss": 0.018497157841920853, "grad_norm": 0.10563384741544724, "global_step": 6399, "epoch": 60, "lr": 9.644505225119922e-05} +{"train_loss": 0.016116857528686523, "grad_norm": 0.111955426633358, "global_step": 6400, "epoch": 60, "lr": 9.644395460374453e-05} +{"train_loss": 0.019209127873182297, "grad_norm": 0.10197214037179947, "global_step": 6401, "epoch": 60, "lr": 9.644285679310598e-05} +{"train_loss": 0.018773406744003296, "grad_norm": 0.065802201628685, "global_step": 6402, "epoch": 60, "lr": 9.644175881928743e-05} +{"train_loss": 0.018678730353713036, "grad_norm": 0.08283249288797379, "global_step": 6403, "epoch": 60, "lr": 9.644066068229273e-05} +{"train_loss": 0.018996134400367737, "grad_norm": 0.06105264648795128, "global_step": 6404, "epoch": 60, "lr": 9.643956238212578e-05} +{"train_loss": 0.017566192895174026, "grad_norm": 0.10307928174734116, "global_step": 6405, "epoch": 60, "lr": 9.643846391879037e-05} +{"train_loss": 0.018435271456837654, "grad_norm": 0.13216903805732727, "global_step": 6406, "epoch": 60, "lr": 9.643736529229039e-05} +{"train_loss": 0.018076108768582344, "grad_norm": 0.06179654970765114, "global_step": 6407, "epoch": 60, "lr": 9.643626650262971e-05} +{"train_loss": 0.014692729339003563, "grad_norm": 0.03768276050686836, "global_step": 6408, "epoch": 60, "lr": 9.64351675498122e-05} +{"train_loss": 0.018499314785003662, "grad_norm": 0.10453138500452042, "global_step": 6409, "epoch": 60, "lr": 9.643406843384169e-05} +{"train_loss": 0.017923593521118164, "grad_norm": 0.03737622871994972, "global_step": 6410, "epoch": 60, "lr": 9.643296915472205e-05} +{"train_loss": 0.018413478508591652, "grad_norm": 0.10290324687957764, "global_step": 6411, "epoch": 60, "lr": 9.643186971245716e-05} +{"train_loss": 0.020163755863904953, "grad_norm": 0.05422350764274597, "global_step": 6412, "epoch": 60, "lr": 9.643077010705087e-05} +{"train_loss": 0.01808542013168335, "grad_norm": 0.04800593852996826, "global_step": 6413, "epoch": 60, "lr": 9.642967033850706e-05} +{"train_loss": 0.019170694053173065, "grad_norm": 0.11401347815990448, "global_step": 6414, "epoch": 60, "lr": 9.642857040682955e-05} +{"train_loss": 0.02141295187175274, "grad_norm": 0.08733081817626953, "global_step": 6415, "epoch": 60, "lr": 9.642747031202225e-05} +{"train_loss": 0.017962969839572906, "grad_norm": 0.06354191154241562, "global_step": 6416, "epoch": 60, "lr": 9.642637005408902e-05} +{"train_loss": 0.01691308245062828, "grad_norm": 0.04288407042622566, "global_step": 6417, "epoch": 60, "lr": 9.642526963303371e-05} +{"train_loss": 0.018928291276097298, "grad_norm": 0.05757385492324829, "global_step": 6418, "epoch": 60, "lr": 9.64241690488602e-05} +{"train_loss": 0.018676795065402985, "grad_norm": 0.07310531288385391, "global_step": 6419, "epoch": 60, "lr": 9.642306830157233e-05} +{"train_loss": 0.02268221601843834, "grad_norm": 0.08394521474838257, "global_step": 6420, "epoch": 60, "lr": 9.6421967391174e-05} +{"train_loss": 0.019977333024144173, "grad_norm": 0.07436731457710266, "global_step": 6421, "epoch": 60, "lr": 9.642086631766907e-05} +{"train_loss": 0.019575919955968857, "grad_norm": 0.07674168795347214, "global_step": 6422, "epoch": 60, "lr": 9.641976508106138e-05} +{"train_loss": 0.017298122867941856, "grad_norm": 0.050526779145002365, "global_step": 6423, "epoch": 60, "lr": 9.641866368135484e-05} +{"train_loss": 0.018913280218839645, "grad_norm": 0.046628549695014954, "global_step": 6424, "epoch": 60, "lr": 9.64175621185533e-05} +{"train_loss": 0.01669568382203579, "grad_norm": 0.04490179568529129, "global_step": 6425, "epoch": 60, "lr": 9.641646039266063e-05} +{"train_loss": 0.020530829206109047, "grad_norm": 0.09524671733379364, "global_step": 6426, "epoch": 60, "lr": 9.641535850368073e-05} +{"train_loss": 0.019519809633493423, "grad_norm": 0.077332504093647, "global_step": 6427, "epoch": 60, "lr": 9.641425645161742e-05} +{"train_loss": 0.02343900129199028, "grad_norm": 0.05373810976743698, "global_step": 6428, "epoch": 60, "lr": 9.641315423647458e-05} +{"train_loss": 0.016098588705062866, "grad_norm": 0.07201576232910156, "global_step": 6429, "epoch": 60, "lr": 9.641205185825613e-05} +{"train_loss": 0.018740659579634666, "grad_norm": 0.09416024386882782, "global_step": 6430, "epoch": 60, "lr": 9.641094931696592e-05} +{"train_loss": 0.01713353954255581, "grad_norm": 0.0524788424372673, "global_step": 6431, "epoch": 60, "lr": 9.64098466126078e-05} +{"train_loss": 0.01875581592321396, "grad_norm": 0.07423289120197296, "global_step": 6432, "epoch": 60, "lr": 9.640874374518567e-05} +{"train_loss": 0.01860496960580349, "grad_norm": 0.07044892013072968, "global_step": 6433, "epoch": 60, "lr": 9.640764071470338e-05} +{"train_loss": 0.020546117797493935, "grad_norm": 0.05811222642660141, "global_step": 6434, "epoch": 60, "lr": 9.640653752116484e-05} +{"train_loss": 0.01944647915661335, "grad_norm": 0.050739068537950516, "global_step": 6435, "epoch": 60, "lr": 9.64054341645739e-05} +{"train_loss": 0.019986117258667946, "grad_norm": 0.09077946841716766, "global_step": 6436, "epoch": 60, "lr": 9.640433064493445e-05} +{"train_loss": 0.016863131895661354, "grad_norm": 0.060198232531547546, "global_step": 6437, "epoch": 60, "lr": 9.640322696225038e-05} +{"train_loss": 0.01967979595065117, "grad_norm": 0.060734570026397705, "global_step": 6438, "epoch": 60, "lr": 9.640212311652552e-05} +{"train_loss": 0.017305733636021614, "grad_norm": 0.07571326196193695, "global_step": 6439, "epoch": 60, "lr": 9.640101910776381e-05} +{"train_loss": 0.019193440675735474, "grad_norm": 0.0993603765964508, "global_step": 6440, "epoch": 60, "lr": 9.639991493596907e-05} +{"train_loss": 0.01947346329689026, "grad_norm": 0.061417873948812485, "global_step": 6441, "epoch": 60, "lr": 9.639881060114522e-05} +{"train_loss": 0.017366506159305573, "grad_norm": 0.03880733251571655, "global_step": 6442, "epoch": 60, "lr": 9.639770610329613e-05} +{"train_loss": 0.018405871465802193, "grad_norm": 0.08900033682584763, "global_step": 6443, "epoch": 60, "lr": 9.639660144242568e-05} +{"train_loss": 0.018068991601467133, "grad_norm": 0.06902912259101868, "global_step": 6444, "epoch": 60, "lr": 9.639549661853774e-05} +{"train_loss": 0.018338646739721298, "grad_norm": 0.05764647200703621, "global_step": 6445, "epoch": 60, "lr": 9.639439163163622e-05} +{"train_loss": 0.018651919439435005, "grad_norm": 0.08030535280704498, "global_step": 6446, "epoch": 60, "lr": 9.639328648172496e-05} +{"train_loss": 0.017243657261133194, "grad_norm": 0.03779584914445877, "global_step": 6447, "epoch": 60, "lr": 9.639218116880788e-05} +{"train_loss": 0.016670163720846176, "grad_norm": 0.04598074406385422, "global_step": 6448, "epoch": 60, "lr": 9.639107569288885e-05} +{"train_loss": 0.02237882651388645, "grad_norm": 0.09668316692113876, "global_step": 6449, "epoch": 60, "lr": 9.638997005397175e-05} +{"train_loss": 0.020946325734257698, "grad_norm": 0.06176844239234924, "global_step": 6450, "epoch": 60, "lr": 9.638886425206047e-05} +{"train_loss": 0.017086613923311234, "grad_norm": 0.06985772401094437, "global_step": 6451, "epoch": 60, "lr": 9.63877582871589e-05} +{"train_loss": 0.018941335380077362, "grad_norm": 0.09710939228534698, "global_step": 6452, "epoch": 60, "lr": 9.638665215927092e-05} +{"train_loss": 0.017629854381084442, "grad_norm": 0.12194499373435974, "global_step": 6453, "epoch": 60, "lr": 9.638554586840043e-05} +{"train_loss": 0.019145023077726364, "grad_norm": 0.07981608062982559, "global_step": 6454, "epoch": 60, "lr": 9.638443941455129e-05} +{"train_loss": 0.017910242080688477, "grad_norm": 0.050193991512060165, "global_step": 6455, "epoch": 60, "lr": 9.63833327977274e-05} +{"train_loss": 0.020826615393161774, "grad_norm": 0.10122615098953247, "global_step": 6456, "epoch": 60, "lr": 9.638222601793265e-05} +{"train_loss": 0.021815285086631775, "grad_norm": 0.09365925937891006, "global_step": 6457, "epoch": 60, "lr": 9.638111907517093e-05} +{"train_loss": 0.02051345258951187, "grad_norm": 0.10373245924711227, "global_step": 6458, "epoch": 60, "lr": 9.638001196944613e-05} +{"train_loss": 0.01737554743885994, "grad_norm": 0.03584945946931839, "global_step": 6459, "epoch": 60, "lr": 9.637890470076215e-05} +{"train_loss": 0.01896040514111519, "grad_norm": 0.06424061954021454, "global_step": 6460, "epoch": 60, "lr": 9.637779726912285e-05} +{"train_loss": 0.017908308655023575, "grad_norm": 0.03985299542546272, "global_step": 6461, "epoch": 60, "lr": 9.637668967453215e-05} +{"train_loss": 0.016205281019210815, "grad_norm": 0.06314989179372787, "global_step": 6462, "epoch": 60, "lr": 9.637558191699392e-05} +{"train_loss": 0.020994730293750763, "grad_norm": 0.10339873284101486, "global_step": 6463, "epoch": 60, "lr": 9.637447399651207e-05} +{"train_loss": 0.01931428164243698, "grad_norm": 0.11305618286132812, "global_step": 6464, "epoch": 60, "lr": 9.63733659130905e-05} +{"train_loss": 0.018778340908575733, "grad_norm": 0.09137596189975739, "global_step": 6465, "epoch": 60, "lr": 9.637225766673307e-05, "train_batch_mse_error": 0.016862619668245316, "val_loss": 0.018549608066678047} +{"train_loss": 0.014111118391156197, "grad_norm": 0.0459953136742115, "global_step": 6466, "epoch": 61, "lr": 9.63711492574437e-05} +{"train_loss": 0.019140351563692093, "grad_norm": 0.07969444990158081, "global_step": 6467, "epoch": 61, "lr": 9.637004068522628e-05} +{"train_loss": 0.019529666751623154, "grad_norm": 0.04562891274690628, "global_step": 6468, "epoch": 61, "lr": 9.63689319500847e-05} +{"train_loss": 0.022123582661151886, "grad_norm": 0.07955916225910187, "global_step": 6469, "epoch": 61, "lr": 9.636782305202285e-05} +{"train_loss": 0.01776963844895363, "grad_norm": 0.06187742203474045, "global_step": 6470, "epoch": 61, "lr": 9.636671399104465e-05} +{"train_loss": 0.01941782236099243, "grad_norm": 0.0718010663986206, "global_step": 6471, "epoch": 61, "lr": 9.636560476715397e-05} +{"train_loss": 0.01969253458082676, "grad_norm": 0.0591205395758152, "global_step": 6472, "epoch": 61, "lr": 9.636449538035473e-05} +{"train_loss": 0.020166940987110138, "grad_norm": 0.18499161303043365, "global_step": 6473, "epoch": 61, "lr": 9.636338583065081e-05} +{"train_loss": 0.020775631070137024, "grad_norm": 0.06416536122560501, "global_step": 6474, "epoch": 61, "lr": 9.636227611804611e-05} +{"train_loss": 0.018911249935626984, "grad_norm": 0.04926133528351784, "global_step": 6475, "epoch": 61, "lr": 9.636116624254455e-05} +{"train_loss": 0.019682729616761208, "grad_norm": 0.08726657927036285, "global_step": 6476, "epoch": 61, "lr": 9.636005620415001e-05} +{"train_loss": 0.020330408588051796, "grad_norm": 0.07807931303977966, "global_step": 6477, "epoch": 61, "lr": 9.635894600286638e-05} +{"train_loss": 0.02149169333279133, "grad_norm": 0.12710939347743988, "global_step": 6478, "epoch": 61, "lr": 9.635783563869757e-05} +{"train_loss": 0.01795041374862194, "grad_norm": 0.14237193763256073, "global_step": 6479, "epoch": 61, "lr": 9.635672511164751e-05} +{"train_loss": 0.01771836169064045, "grad_norm": 0.061604082584381104, "global_step": 6480, "epoch": 61, "lr": 9.635561442172008e-05} +{"train_loss": 0.01908911019563675, "grad_norm": 0.10934703052043915, "global_step": 6481, "epoch": 61, "lr": 9.635450356891915e-05} +{"train_loss": 0.01682206057012081, "grad_norm": 0.07825745642185211, "global_step": 6482, "epoch": 61, "lr": 9.635339255324867e-05} +{"train_loss": 0.01954137533903122, "grad_norm": 0.09693337231874466, "global_step": 6483, "epoch": 61, "lr": 9.635228137471253e-05} +{"train_loss": 0.020892389118671417, "grad_norm": 0.05301973968744278, "global_step": 6484, "epoch": 61, "lr": 9.635117003331463e-05} +{"train_loss": 0.01839524134993553, "grad_norm": 0.09405592083930969, "global_step": 6485, "epoch": 61, "lr": 9.635005852905888e-05} +{"train_loss": 0.020702391862869263, "grad_norm": 0.05824087932705879, "global_step": 6486, "epoch": 61, "lr": 9.634894686194918e-05} +{"train_loss": 0.019729794934391975, "grad_norm": 0.05329121649265289, "global_step": 6487, "epoch": 61, "lr": 9.634783503198943e-05} +{"train_loss": 0.01595434732735157, "grad_norm": 0.09222887456417084, "global_step": 6488, "epoch": 61, "lr": 9.634672303918355e-05} +{"train_loss": 0.01747305504977703, "grad_norm": 0.10261820256710052, "global_step": 6489, "epoch": 61, "lr": 9.634561088353543e-05} +{"train_loss": 0.01688484102487564, "grad_norm": 0.12611126899719238, "global_step": 6490, "epoch": 61, "lr": 9.634449856504898e-05} +{"train_loss": 0.018730364739894867, "grad_norm": 0.04816165938973427, "global_step": 6491, "epoch": 61, "lr": 9.634338608372815e-05} +{"train_loss": 0.01968989335000515, "grad_norm": 0.052940983325242996, "global_step": 6492, "epoch": 61, "lr": 9.63422734395768e-05} +{"train_loss": 0.01762179657816887, "grad_norm": 0.045313797891139984, "global_step": 6493, "epoch": 61, "lr": 9.634116063259886e-05} +{"train_loss": 0.01780925877392292, "grad_norm": 0.05708736553788185, "global_step": 6494, "epoch": 61, "lr": 9.634004766279823e-05} +{"train_loss": 0.01915678381919861, "grad_norm": 0.04400131106376648, "global_step": 6495, "epoch": 61, "lr": 9.633893453017881e-05} +{"train_loss": 0.01776738651096821, "grad_norm": 0.07151314616203308, "global_step": 6496, "epoch": 61, "lr": 9.633782123474455e-05} +{"train_loss": 0.018348345533013344, "grad_norm": 0.06457309424877167, "global_step": 6497, "epoch": 61, "lr": 9.633670777649932e-05} +{"train_loss": 0.01870334893465042, "grad_norm": 0.07688631117343903, "global_step": 6498, "epoch": 61, "lr": 9.633559415544706e-05} +{"train_loss": 0.016472887247800827, "grad_norm": 0.06014081835746765, "global_step": 6499, "epoch": 61, "lr": 9.633448037159167e-05} +{"train_loss": 0.016026590019464493, "grad_norm": 0.05451830103993416, "global_step": 6500, "epoch": 61, "lr": 9.633336642493707e-05} +{"train_loss": 0.020914247259497643, "grad_norm": 0.06194407865405083, "global_step": 6501, "epoch": 61, "lr": 9.633225231548717e-05} +{"train_loss": 0.019283026456832886, "grad_norm": 0.052972499281167984, "global_step": 6502, "epoch": 61, "lr": 9.633113804324588e-05} +{"train_loss": 0.01838313415646553, "grad_norm": 0.04730365052819252, "global_step": 6503, "epoch": 61, "lr": 9.633002360821713e-05} +{"train_loss": 0.016233045607805252, "grad_norm": 0.050501082092523575, "global_step": 6504, "epoch": 61, "lr": 9.632890901040481e-05} +{"train_loss": 0.02171744778752327, "grad_norm": 0.09331609308719635, "global_step": 6505, "epoch": 61, "lr": 9.632779424981287e-05} +{"train_loss": 0.02025153487920761, "grad_norm": 0.124052494764328, "global_step": 6506, "epoch": 61, "lr": 9.632667932644519e-05} +{"train_loss": 0.017384134232997894, "grad_norm": 0.06155192852020264, "global_step": 6507, "epoch": 61, "lr": 9.632556424030572e-05} +{"train_loss": 0.01573116146028042, "grad_norm": 0.0840459018945694, "global_step": 6508, "epoch": 61, "lr": 9.632444899139835e-05} +{"train_loss": 0.019927240908145905, "grad_norm": 0.047537144273519516, "global_step": 6509, "epoch": 61, "lr": 9.632333357972703e-05} +{"train_loss": 0.018522704020142555, "grad_norm": 0.05592479184269905, "global_step": 6510, "epoch": 61, "lr": 9.632221800529564e-05} +{"train_loss": 0.014818661846220493, "grad_norm": 0.03907551243901253, "global_step": 6511, "epoch": 61, "lr": 9.632110226810813e-05} +{"train_loss": 0.02086396887898445, "grad_norm": 0.08996693044900894, "global_step": 6512, "epoch": 61, "lr": 9.631998636816841e-05} +{"train_loss": 0.024718355387449265, "grad_norm": 0.09935442358255386, "global_step": 6513, "epoch": 61, "lr": 9.631887030548041e-05} +{"train_loss": 0.016009580343961716, "grad_norm": 0.07365478575229645, "global_step": 6514, "epoch": 61, "lr": 9.631775408004804e-05} +{"train_loss": 0.01733725517988205, "grad_norm": 0.06355840712785721, "global_step": 6515, "epoch": 61, "lr": 9.631663769187522e-05} +{"train_loss": 0.022248761728405952, "grad_norm": 0.10256354510784149, "global_step": 6516, "epoch": 61, "lr": 9.631552114096589e-05} +{"train_loss": 0.018715638667345047, "grad_norm": 0.04576772823929787, "global_step": 6517, "epoch": 61, "lr": 9.631440442732396e-05} +{"train_loss": 0.019072795286774635, "grad_norm": 0.05542654171586037, "global_step": 6518, "epoch": 61, "lr": 9.631328755095333e-05} +{"train_loss": 0.019032079726457596, "grad_norm": 0.057086728513240814, "global_step": 6519, "epoch": 61, "lr": 9.631217051185797e-05} +{"train_loss": 0.018898073583841324, "grad_norm": 0.11063504964113235, "global_step": 6520, "epoch": 61, "lr": 9.631105331004177e-05} +{"train_loss": 0.017475763335824013, "grad_norm": 0.06334955245256424, "global_step": 6521, "epoch": 61, "lr": 9.630993594550869e-05} +{"train_loss": 0.017222164198756218, "grad_norm": 0.05094088613986969, "global_step": 6522, "epoch": 61, "lr": 9.630881841826263e-05} +{"train_loss": 0.018079932779073715, "grad_norm": 0.056520726531744, "global_step": 6523, "epoch": 61, "lr": 9.63077007283075e-05} +{"train_loss": 0.01821896806359291, "grad_norm": 0.07929825782775879, "global_step": 6524, "epoch": 61, "lr": 9.630658287564727e-05} +{"train_loss": 0.019695641472935677, "grad_norm": 0.06554204225540161, "global_step": 6525, "epoch": 61, "lr": 9.630546486028584e-05} +{"train_loss": 0.020158469676971436, "grad_norm": 0.13446594774723053, "global_step": 6526, "epoch": 61, "lr": 9.630434668222715e-05} +{"train_loss": 0.015852730721235275, "grad_norm": 0.0531473234295845, "global_step": 6527, "epoch": 61, "lr": 9.63032283414751e-05} +{"train_loss": 0.018560132011771202, "grad_norm": 0.09573730081319809, "global_step": 6528, "epoch": 61, "lr": 9.630210983803366e-05} +{"train_loss": 0.020499348640441895, "grad_norm": 0.0710606798529625, "global_step": 6529, "epoch": 61, "lr": 9.630099117190674e-05} +{"train_loss": 0.020216509699821472, "grad_norm": 0.0864994004368782, "global_step": 6530, "epoch": 61, "lr": 9.629987234309828e-05} +{"train_loss": 0.0175161249935627, "grad_norm": 0.12762075662612915, "global_step": 6531, "epoch": 61, "lr": 9.629875335161218e-05} +{"train_loss": 0.018903251737356186, "grad_norm": 0.06614705920219421, "global_step": 6532, "epoch": 61, "lr": 9.629763419745241e-05} +{"train_loss": 0.019149068742990494, "grad_norm": 0.08484061062335968, "global_step": 6533, "epoch": 61, "lr": 9.62965148806229e-05} +{"train_loss": 0.01499453280121088, "grad_norm": 0.051059409976005554, "global_step": 6534, "epoch": 61, "lr": 9.629539540112756e-05} +{"train_loss": 0.019075030460953712, "grad_norm": 0.0631660670042038, "global_step": 6535, "epoch": 61, "lr": 9.629427575897033e-05} +{"train_loss": 0.020282048732042313, "grad_norm": 0.049571264535188675, "global_step": 6536, "epoch": 61, "lr": 9.629315595415516e-05} +{"train_loss": 0.01881524920463562, "grad_norm": 0.06286369264125824, "global_step": 6537, "epoch": 61, "lr": 9.629203598668595e-05} +{"train_loss": 0.019542070105671883, "grad_norm": 0.0651119202375412, "global_step": 6538, "epoch": 61, "lr": 9.629091585656668e-05} +{"train_loss": 0.01924552395939827, "grad_norm": 0.055176980793476105, "global_step": 6539, "epoch": 61, "lr": 9.628979556380125e-05} +{"train_loss": 0.015956513583660126, "grad_norm": 0.1376534253358841, "global_step": 6540, "epoch": 61, "lr": 9.628867510839361e-05} +{"train_loss": 0.017727874219417572, "grad_norm": 0.18788371980190277, "global_step": 6541, "epoch": 61, "lr": 9.628755449034772e-05} +{"train_loss": 0.01997552439570427, "grad_norm": 0.09507603198289871, "global_step": 6542, "epoch": 61, "lr": 9.628643370966746e-05} +{"train_loss": 0.0192475076764822, "grad_norm": 0.14994865655899048, "global_step": 6543, "epoch": 61, "lr": 9.628531276635682e-05} +{"train_loss": 0.018502265214920044, "grad_norm": 0.09286689758300781, "global_step": 6544, "epoch": 61, "lr": 9.628419166041971e-05} +{"train_loss": 0.021118590608239174, "grad_norm": 0.21106202900409698, "global_step": 6545, "epoch": 61, "lr": 9.628307039186009e-05} +{"train_loss": 0.017106343060731888, "grad_norm": 0.05428578332066536, "global_step": 6546, "epoch": 61, "lr": 9.628194896068188e-05} +{"train_loss": 0.02056504786014557, "grad_norm": 0.06413830071687698, "global_step": 6547, "epoch": 61, "lr": 9.628082736688902e-05} +{"train_loss": 0.017717253416776657, "grad_norm": 0.04768446087837219, "global_step": 6548, "epoch": 61, "lr": 9.627970561048548e-05} +{"train_loss": 0.015999557450413704, "grad_norm": 0.038309283554553986, "global_step": 6549, "epoch": 61, "lr": 9.627858369147517e-05} +{"train_loss": 0.018744757398962975, "grad_norm": 0.11954889446496964, "global_step": 6550, "epoch": 61, "lr": 9.627746160986204e-05} +{"train_loss": 0.020679375156760216, "grad_norm": 0.1311396211385727, "global_step": 6551, "epoch": 61, "lr": 9.627633936565004e-05} +{"train_loss": 0.020663127303123474, "grad_norm": 0.050370123237371445, "global_step": 6552, "epoch": 61, "lr": 9.627521695884311e-05} +{"train_loss": 0.020348452031612396, "grad_norm": 0.05864904448390007, "global_step": 6553, "epoch": 61, "lr": 9.627409438944519e-05} +{"train_loss": 0.017993135377764702, "grad_norm": 0.16188612580299377, "global_step": 6554, "epoch": 61, "lr": 9.627297165746024e-05} +{"train_loss": 0.020444724708795547, "grad_norm": 0.12360965460538864, "global_step": 6555, "epoch": 61, "lr": 9.627184876289216e-05} +{"train_loss": 0.018605072051286697, "grad_norm": 0.0420726016163826, "global_step": 6556, "epoch": 61, "lr": 9.627072570574495e-05} +{"train_loss": 0.018778717145323753, "grad_norm": 0.0946325957775116, "global_step": 6557, "epoch": 61, "lr": 9.626960248602252e-05} +{"train_loss": 0.02053602784872055, "grad_norm": 0.0816524550318718, "global_step": 6558, "epoch": 61, "lr": 9.626847910372885e-05} +{"train_loss": 0.019250711426138878, "grad_norm": 0.1014605164527893, "global_step": 6559, "epoch": 61, "lr": 9.626735555886785e-05} +{"train_loss": 0.015853559598326683, "grad_norm": 0.04750112071633339, "global_step": 6560, "epoch": 61, "lr": 9.626623185144348e-05} +{"train_loss": 0.018178550526499748, "grad_norm": 0.052788008004426956, "global_step": 6561, "epoch": 61, "lr": 9.62651079814597e-05} +{"train_loss": 0.017137493938207626, "grad_norm": 0.13385219871997833, "global_step": 6562, "epoch": 61, "lr": 9.626398394892045e-05} +{"train_loss": 0.018420884385704994, "grad_norm": 0.08669101446866989, "global_step": 6563, "epoch": 61, "lr": 9.626285975382969e-05} +{"train_loss": 0.01842944324016571, "grad_norm": 0.1395418345928192, "global_step": 6564, "epoch": 61, "lr": 9.626173539619135e-05} +{"train_loss": 0.02085334062576294, "grad_norm": 0.06373021006584167, "global_step": 6565, "epoch": 61, "lr": 9.626061087600937e-05} +{"train_loss": 0.01720849797129631, "grad_norm": 0.09161526709794998, "global_step": 6566, "epoch": 61, "lr": 9.625948619328775e-05} +{"train_loss": 0.021573428064584732, "grad_norm": 0.09016437083482742, "global_step": 6567, "epoch": 61, "lr": 9.625836134803041e-05} +{"train_loss": 0.01801963523030281, "grad_norm": 0.07746615260839462, "global_step": 6568, "epoch": 61, "lr": 9.625723634024129e-05} +{"train_loss": 0.017818816006183624, "grad_norm": 0.08533845841884613, "global_step": 6569, "epoch": 61, "lr": 9.625611116992436e-05} +{"train_loss": 0.018063602969050407, "grad_norm": 0.05466551333665848, "global_step": 6570, "epoch": 61, "lr": 9.625498583708359e-05} +{"train_loss": 0.018770324259574683, "grad_norm": 0.14528028666973114, "global_step": 6571, "epoch": 61, "lr": 9.62538603417229e-05, "val_loss": 0.018937405198812485} +{"train_loss": 0.017945073544979095, "grad_norm": 0.12098293006420135, "global_step": 6572, "epoch": 62, "lr": 9.625273468384628e-05} +{"train_loss": 0.015825629234313965, "grad_norm": 0.0693235993385315, "global_step": 6573, "epoch": 62, "lr": 9.625160886345764e-05} +{"train_loss": 0.020175427198410034, "grad_norm": 0.0515855997800827, "global_step": 6574, "epoch": 62, "lr": 9.625048288056097e-05} +{"train_loss": 0.018086742609739304, "grad_norm": 0.06337066739797592, "global_step": 6575, "epoch": 62, "lr": 9.624935673516022e-05} +{"train_loss": 0.01874002069234848, "grad_norm": 0.06342878937721252, "global_step": 6576, "epoch": 62, "lr": 9.624823042725935e-05} +{"train_loss": 0.017547890543937683, "grad_norm": 0.05173628404736519, "global_step": 6577, "epoch": 62, "lr": 9.62471039568623e-05} +{"train_loss": 0.01857539266347885, "grad_norm": 0.056735944002866745, "global_step": 6578, "epoch": 62, "lr": 9.624597732397304e-05} +{"train_loss": 0.018218182027339935, "grad_norm": 0.18083547055721283, "global_step": 6579, "epoch": 62, "lr": 9.624485052859553e-05} +{"train_loss": 0.018574386835098267, "grad_norm": 0.09000198543071747, "global_step": 6580, "epoch": 62, "lr": 9.624372357073372e-05} +{"train_loss": 0.016397416591644287, "grad_norm": 0.04620646312832832, "global_step": 6581, "epoch": 62, "lr": 9.624259645039159e-05} +{"train_loss": 0.018255766481161118, "grad_norm": 0.12814056873321533, "global_step": 6582, "epoch": 62, "lr": 9.624146916757307e-05} +{"train_loss": 0.018631167709827423, "grad_norm": 0.13959550857543945, "global_step": 6583, "epoch": 62, "lr": 9.624034172228214e-05} +{"train_loss": 0.019827568903565407, "grad_norm": 0.06095941364765167, "global_step": 6584, "epoch": 62, "lr": 9.623921411452277e-05} +{"train_loss": 0.02001061663031578, "grad_norm": 0.0356527604162693, "global_step": 6585, "epoch": 62, "lr": 9.623808634429891e-05} +{"train_loss": 0.01644020900130272, "grad_norm": 0.0813906118273735, "global_step": 6586, "epoch": 62, "lr": 9.623695841161451e-05} +{"train_loss": 0.01925765536725521, "grad_norm": 0.0700584352016449, "global_step": 6587, "epoch": 62, "lr": 9.623583031647357e-05} +{"train_loss": 0.018475688993930817, "grad_norm": 0.05147354677319527, "global_step": 6588, "epoch": 62, "lr": 9.623470205888001e-05} +{"train_loss": 0.020055564120411873, "grad_norm": 0.05815687030553818, "global_step": 6589, "epoch": 62, "lr": 9.62335736388378e-05} +{"train_loss": 0.018842246383428574, "grad_norm": 0.0488431379199028, "global_step": 6590, "epoch": 62, "lr": 9.623244505635094e-05} +{"train_loss": 0.01841190829873085, "grad_norm": 0.15012305974960327, "global_step": 6591, "epoch": 62, "lr": 9.623131631142336e-05} +{"train_loss": 0.020234471186995506, "grad_norm": 0.04338902235031128, "global_step": 6592, "epoch": 62, "lr": 9.623018740405907e-05} +{"train_loss": 0.02381640113890171, "grad_norm": 0.07899383455514908, "global_step": 6593, "epoch": 62, "lr": 9.622905833426198e-05} +{"train_loss": 0.020931169390678406, "grad_norm": 0.11360007524490356, "global_step": 6594, "epoch": 62, "lr": 9.62279291020361e-05} +{"train_loss": 0.020327236503362656, "grad_norm": 0.09184049814939499, "global_step": 6595, "epoch": 62, "lr": 9.622679970738538e-05} +{"train_loss": 0.01650565303862095, "grad_norm": 0.11914398521184921, "global_step": 6596, "epoch": 62, "lr": 9.622567015031376e-05} +{"train_loss": 0.02050258219242096, "grad_norm": 0.040233127772808075, "global_step": 6597, "epoch": 62, "lr": 9.622454043082527e-05} +{"train_loss": 0.022613361477851868, "grad_norm": 0.1094561517238617, "global_step": 6598, "epoch": 62, "lr": 9.622341054892383e-05} +{"train_loss": 0.019538458436727524, "grad_norm": 0.05937140807509422, "global_step": 6599, "epoch": 62, "lr": 9.622228050461343e-05} +{"train_loss": 0.017582599073648453, "grad_norm": 0.06020713597536087, "global_step": 6600, "epoch": 62, "lr": 9.622115029789804e-05} +{"train_loss": 0.017842642962932587, "grad_norm": 0.044555600732564926, "global_step": 6601, "epoch": 62, "lr": 9.622001992878163e-05} +{"train_loss": 0.016440991312265396, "grad_norm": 0.08468371629714966, "global_step": 6602, "epoch": 62, "lr": 9.621888939726817e-05} +{"train_loss": 0.016775324940681458, "grad_norm": 0.10657306015491486, "global_step": 6603, "epoch": 62, "lr": 9.621775870336164e-05} +{"train_loss": 0.016677243635058403, "grad_norm": 0.047122757881879807, "global_step": 6604, "epoch": 62, "lr": 9.6216627847066e-05} +{"train_loss": 0.018164247274398804, "grad_norm": 0.17239898443222046, "global_step": 6605, "epoch": 62, "lr": 9.621549682838522e-05} +{"train_loss": 0.021745257079601288, "grad_norm": 0.1524818241596222, "global_step": 6606, "epoch": 62, "lr": 9.621436564732329e-05} +{"train_loss": 0.019050482660531998, "grad_norm": 0.07234548777341843, "global_step": 6607, "epoch": 62, "lr": 9.621323430388418e-05} +{"train_loss": 0.019703183323144913, "grad_norm": 0.16561776399612427, "global_step": 6608, "epoch": 62, "lr": 9.621210279807187e-05} +{"train_loss": 0.016500238329172134, "grad_norm": 0.14254572987556458, "global_step": 6609, "epoch": 62, "lr": 9.621097112989032e-05} +{"train_loss": 0.02017088606953621, "grad_norm": 0.08706783503293991, "global_step": 6610, "epoch": 62, "lr": 9.620983929934351e-05} +{"train_loss": 0.017479712143540382, "grad_norm": 0.07093735039234161, "global_step": 6611, "epoch": 62, "lr": 9.620870730643542e-05} +{"train_loss": 0.018879543989896774, "grad_norm": 0.06508102267980576, "global_step": 6612, "epoch": 62, "lr": 9.620757515117003e-05} +{"train_loss": 0.019269950687885284, "grad_norm": 0.07218310981988907, "global_step": 6613, "epoch": 62, "lr": 9.620644283355132e-05} +{"train_loss": 0.017629671841859818, "grad_norm": 0.041189584881067276, "global_step": 6614, "epoch": 62, "lr": 9.620531035358325e-05} +{"train_loss": 0.017956756055355072, "grad_norm": 0.04341127723455429, "global_step": 6615, "epoch": 62, "lr": 9.620417771126984e-05} +{"train_loss": 0.01860319823026657, "grad_norm": 0.15594260394573212, "global_step": 6616, "epoch": 62, "lr": 9.620304490661502e-05} +{"train_loss": 0.019373886287212372, "grad_norm": 0.1311245709657669, "global_step": 6617, "epoch": 62, "lr": 9.62019119396228e-05} +{"train_loss": 0.021727154031395912, "grad_norm": 0.09367450326681137, "global_step": 6618, "epoch": 62, "lr": 9.620077881029717e-05} +{"train_loss": 0.019298192113637924, "grad_norm": 0.057577718049287796, "global_step": 6619, "epoch": 62, "lr": 9.619964551864207e-05} +{"train_loss": 0.016132820397615433, "grad_norm": 0.09062883257865906, "global_step": 6620, "epoch": 62, "lr": 9.619851206466152e-05} +{"train_loss": 0.01984081044793129, "grad_norm": 0.11961288005113602, "global_step": 6621, "epoch": 62, "lr": 9.619737844835948e-05} +{"train_loss": 0.016696013510227203, "grad_norm": 0.08810438960790634, "global_step": 6622, "epoch": 62, "lr": 9.619624466973995e-05} +{"train_loss": 0.0159407090395689, "grad_norm": 0.03990377113223076, "global_step": 6623, "epoch": 62, "lr": 9.619511072880691e-05} +{"train_loss": 0.020903177559375763, "grad_norm": 0.08228249102830887, "global_step": 6624, "epoch": 62, "lr": 9.619397662556435e-05} +{"train_loss": 0.01767963543534279, "grad_norm": 0.08826788514852524, "global_step": 6625, "epoch": 62, "lr": 9.619284236001622e-05} +{"train_loss": 0.016429178416728973, "grad_norm": 0.09761009365320206, "global_step": 6626, "epoch": 62, "lr": 9.619170793216656e-05} +{"train_loss": 0.018157701939344406, "grad_norm": 0.07320450991392136, "global_step": 6627, "epoch": 62, "lr": 9.619057334201931e-05} +{"train_loss": 0.01565895602107048, "grad_norm": 0.05144572630524635, "global_step": 6628, "epoch": 62, "lr": 9.618943858957847e-05} +{"train_loss": 0.01877414993941784, "grad_norm": 0.057198815047740936, "global_step": 6629, "epoch": 62, "lr": 9.618830367484804e-05} +{"train_loss": 0.018854059278964996, "grad_norm": 0.1271265298128128, "global_step": 6630, "epoch": 62, "lr": 9.618716859783199e-05} +{"train_loss": 0.020868929103016853, "grad_norm": 0.1408771574497223, "global_step": 6631, "epoch": 62, "lr": 9.618603335853431e-05} +{"train_loss": 0.01932234689593315, "grad_norm": 0.11518342792987823, "global_step": 6632, "epoch": 62, "lr": 9.618489795695899e-05} +{"train_loss": 0.017630036920309067, "grad_norm": 0.10120677202939987, "global_step": 6633, "epoch": 62, "lr": 9.618376239311004e-05} +{"train_loss": 0.018969513475894928, "grad_norm": 0.051549866795539856, "global_step": 6634, "epoch": 62, "lr": 9.618262666699143e-05} +{"train_loss": 0.017961634323000908, "grad_norm": 0.07802153378725052, "global_step": 6635, "epoch": 62, "lr": 9.618149077860716e-05} +{"train_loss": 0.019773010164499283, "grad_norm": 0.06152229383587837, "global_step": 6636, "epoch": 62, "lr": 9.61803547279612e-05} +{"train_loss": 0.020026901736855507, "grad_norm": 0.06580911576747894, "global_step": 6637, "epoch": 62, "lr": 9.617921851505757e-05} +{"train_loss": 0.016742747277021408, "grad_norm": 0.09635636955499649, "global_step": 6638, "epoch": 62, "lr": 9.617808213990026e-05} +{"train_loss": 0.01936645805835724, "grad_norm": 0.050919078290462494, "global_step": 6639, "epoch": 62, "lr": 9.617694560249322e-05} +{"train_loss": 0.02103113755583763, "grad_norm": 0.12263642251491547, "global_step": 6640, "epoch": 62, "lr": 9.617580890284051e-05} +{"train_loss": 0.01755421608686447, "grad_norm": 0.040273066610097885, "global_step": 6641, "epoch": 62, "lr": 9.617467204094606e-05} +{"train_loss": 0.019685976207256317, "grad_norm": 0.06957405060529709, "global_step": 6642, "epoch": 62, "lr": 9.61735350168139e-05} +{"train_loss": 0.01638985238969326, "grad_norm": 0.0860997885465622, "global_step": 6643, "epoch": 62, "lr": 9.617239783044803e-05} +{"train_loss": 0.019172461703419685, "grad_norm": 0.06469894200563431, "global_step": 6644, "epoch": 62, "lr": 9.617126048185242e-05} +{"train_loss": 0.01741177588701248, "grad_norm": 0.05120827630162239, "global_step": 6645, "epoch": 62, "lr": 9.617012297103109e-05} +{"train_loss": 0.019341878592967987, "grad_norm": 0.05273328348994255, "global_step": 6646, "epoch": 62, "lr": 9.6168985297988e-05} +{"train_loss": 0.02023717761039734, "grad_norm": 0.05331352353096008, "global_step": 6647, "epoch": 62, "lr": 9.616784746272721e-05} +{"train_loss": 0.017651841044425964, "grad_norm": 0.12505711615085602, "global_step": 6648, "epoch": 62, "lr": 9.616670946525267e-05} +{"train_loss": 0.020479969680309296, "grad_norm": 0.08361616730690002, "global_step": 6649, "epoch": 62, "lr": 9.616557130556839e-05} +{"train_loss": 0.02013682946562767, "grad_norm": 0.18144801259040833, "global_step": 6650, "epoch": 62, "lr": 9.616443298367837e-05} +{"train_loss": 0.022755861282348633, "grad_norm": 0.13630293309688568, "global_step": 6651, "epoch": 62, "lr": 9.61632944995866e-05} +{"train_loss": 0.018565446138381958, "grad_norm": 0.07677804678678513, "global_step": 6652, "epoch": 62, "lr": 9.616215585329711e-05} +{"train_loss": 0.01883840560913086, "grad_norm": 0.03203345462679863, "global_step": 6653, "epoch": 62, "lr": 9.616101704481386e-05} +{"train_loss": 0.01764703169465065, "grad_norm": 0.05743566155433655, "global_step": 6654, "epoch": 62, "lr": 9.615987807414088e-05} +{"train_loss": 0.01757577434182167, "grad_norm": 0.04532046988606453, "global_step": 6655, "epoch": 62, "lr": 9.615873894128216e-05} +{"train_loss": 0.017225656658411026, "grad_norm": 0.05566585436463356, "global_step": 6656, "epoch": 62, "lr": 9.615759964624169e-05} +{"train_loss": 0.019524695351719856, "grad_norm": 0.11807309091091156, "global_step": 6657, "epoch": 62, "lr": 9.615646018902352e-05} +{"train_loss": 0.02036573924124241, "grad_norm": 0.12733735144138336, "global_step": 6658, "epoch": 62, "lr": 9.61553205696316e-05} +{"train_loss": 0.021104533225297928, "grad_norm": 0.14078055322170258, "global_step": 6659, "epoch": 62, "lr": 9.615418078806995e-05} +{"train_loss": 0.01751500926911831, "grad_norm": 0.05132855847477913, "global_step": 6660, "epoch": 62, "lr": 9.61530408443426e-05} +{"train_loss": 0.020708858966827393, "grad_norm": 0.1348126083612442, "global_step": 6661, "epoch": 62, "lr": 9.615190073845353e-05} +{"train_loss": 0.018548093736171722, "grad_norm": 0.10195139795541763, "global_step": 6662, "epoch": 62, "lr": 9.615076047040675e-05} +{"train_loss": 0.022122813388705254, "grad_norm": 0.1369413286447525, "global_step": 6663, "epoch": 62, "lr": 9.614962004020626e-05} +{"train_loss": 0.018900040537118912, "grad_norm": 0.14320987462997437, "global_step": 6664, "epoch": 62, "lr": 9.614847944785608e-05} +{"train_loss": 0.016495762392878532, "grad_norm": 0.047643449157476425, "global_step": 6665, "epoch": 62, "lr": 9.614733869336021e-05} +{"train_loss": 0.018802020698785782, "grad_norm": 0.05606301873922348, "global_step": 6666, "epoch": 62, "lr": 9.614619777672266e-05} +{"train_loss": 0.014866596087813377, "grad_norm": 0.04968439042568207, "global_step": 6667, "epoch": 62, "lr": 9.614505669794745e-05} +{"train_loss": 0.022546641528606415, "grad_norm": 0.05188586190342903, "global_step": 6668, "epoch": 62, "lr": 9.614391545703858e-05} +{"train_loss": 0.0162986908107996, "grad_norm": 0.05837947875261307, "global_step": 6669, "epoch": 62, "lr": 9.614277405400006e-05} +{"train_loss": 0.017407353967428207, "grad_norm": 0.049028392881155014, "global_step": 6670, "epoch": 62, "lr": 9.614163248883588e-05} +{"train_loss": 0.019049987196922302, "grad_norm": 0.0849953293800354, "global_step": 6671, "epoch": 62, "lr": 9.614049076155009e-05} +{"train_loss": 0.02045457251369953, "grad_norm": 0.06636486947536469, "global_step": 6672, "epoch": 62, "lr": 9.613934887214666e-05} +{"train_loss": 0.0173172764480114, "grad_norm": 0.1348096877336502, "global_step": 6673, "epoch": 62, "lr": 9.613820682062964e-05} +{"train_loss": 0.0161368940025568, "grad_norm": 0.07411523163318634, "global_step": 6674, "epoch": 62, "lr": 9.613706460700302e-05} +{"train_loss": 0.02028578892350197, "grad_norm": 0.040919478982686996, "global_step": 6675, "epoch": 62, "lr": 9.613592223127084e-05} +{"train_loss": 0.02009882777929306, "grad_norm": 0.06932396441698074, "global_step": 6676, "epoch": 62, "lr": 9.613477969343708e-05} +{"train_loss": 0.018768269305099856, "grad_norm": 0.20412446558475494, "global_step": 6677, "epoch": 62, "lr": 9.613363699350575e-05, "val_loss": 0.01850212551653385} +{"train_loss": 0.017914924770593643, "grad_norm": 0.04795211926102638, "global_step": 6678, "epoch": 63, "lr": 9.613249413148088e-05} +{"train_loss": 0.019039830192923546, "grad_norm": 0.12910382449626923, "global_step": 6679, "epoch": 63, "lr": 9.61313511073665e-05} +{"train_loss": 0.020970987156033516, "grad_norm": 0.06358864903450012, "global_step": 6680, "epoch": 63, "lr": 9.613020792116663e-05} +{"train_loss": 0.018964851275086403, "grad_norm": 0.0899902880191803, "global_step": 6681, "epoch": 63, "lr": 9.612906457288526e-05} +{"train_loss": 0.020773153752088547, "grad_norm": 0.06957992166280746, "global_step": 6682, "epoch": 63, "lr": 9.612792106252642e-05} +{"train_loss": 0.020128950476646423, "grad_norm": 0.04379482567310333, "global_step": 6683, "epoch": 63, "lr": 9.612677739009412e-05} +{"train_loss": 0.021405545994639397, "grad_norm": 0.09343844652175903, "global_step": 6684, "epoch": 63, "lr": 9.61256335555924e-05} +{"train_loss": 0.02016005665063858, "grad_norm": 0.11783923208713531, "global_step": 6685, "epoch": 63, "lr": 9.612448955902524e-05} +{"train_loss": 0.016016928479075432, "grad_norm": 0.0704290121793747, "global_step": 6686, "epoch": 63, "lr": 9.61233454003967e-05} +{"train_loss": 0.017135819420218468, "grad_norm": 0.04452192410826683, "global_step": 6687, "epoch": 63, "lr": 9.612220107971077e-05} +{"train_loss": 0.019526749849319458, "grad_norm": 0.04123479872941971, "global_step": 6688, "epoch": 63, "lr": 9.612105659697148e-05} +{"train_loss": 0.017651550471782684, "grad_norm": 0.051040198653936386, "global_step": 6689, "epoch": 63, "lr": 9.611991195218286e-05} +{"train_loss": 0.018097765743732452, "grad_norm": 0.1327710747718811, "global_step": 6690, "epoch": 63, "lr": 9.611876714534893e-05} +{"train_loss": 0.017701778560876846, "grad_norm": 0.10025469213724136, "global_step": 6691, "epoch": 63, "lr": 9.611762217647371e-05} +{"train_loss": 0.014990163967013359, "grad_norm": 0.08684508502483368, "global_step": 6692, "epoch": 63, "lr": 9.611647704556123e-05} +{"train_loss": 0.020635083317756653, "grad_norm": 0.08999871462583542, "global_step": 6693, "epoch": 63, "lr": 9.611533175261549e-05} +{"train_loss": 0.01976809650659561, "grad_norm": 0.10417716205120087, "global_step": 6694, "epoch": 63, "lr": 9.611418629764053e-05} +{"train_loss": 0.018502522259950638, "grad_norm": 0.05596266686916351, "global_step": 6695, "epoch": 63, "lr": 9.611304068064038e-05} +{"train_loss": 0.017929602414369583, "grad_norm": 0.05461002513766289, "global_step": 6696, "epoch": 63, "lr": 9.611189490161906e-05} +{"train_loss": 0.01767057552933693, "grad_norm": 0.08341784030199051, "global_step": 6697, "epoch": 63, "lr": 9.611074896058059e-05} +{"train_loss": 0.019994065165519714, "grad_norm": 0.10369416326284409, "global_step": 6698, "epoch": 63, "lr": 9.610960285752902e-05} +{"train_loss": 0.01763099431991577, "grad_norm": 0.07819100469350815, "global_step": 6699, "epoch": 63, "lr": 9.610845659246834e-05} +{"train_loss": 0.018295112997293472, "grad_norm": 0.0613449364900589, "global_step": 6700, "epoch": 63, "lr": 9.610731016540259e-05} +{"train_loss": 0.018672708421945572, "grad_norm": 0.05622818320989609, "global_step": 6701, "epoch": 63, "lr": 9.610616357633582e-05} +{"train_loss": 0.01577124558389187, "grad_norm": 0.053108781576156616, "global_step": 6702, "epoch": 63, "lr": 9.610501682527203e-05} +{"train_loss": 0.019115498289465904, "grad_norm": 0.0401034839451313, "global_step": 6703, "epoch": 63, "lr": 9.610386991221526e-05} +{"train_loss": 0.020013488829135895, "grad_norm": 0.0953650176525116, "global_step": 6704, "epoch": 63, "lr": 9.610272283716955e-05} +{"train_loss": 0.018515467643737793, "grad_norm": 0.1690671741962433, "global_step": 6705, "epoch": 63, "lr": 9.61015756001389e-05} +{"train_loss": 0.01621580868959427, "grad_norm": 0.04533722996711731, "global_step": 6706, "epoch": 63, "lr": 9.610042820112739e-05} +{"train_loss": 0.018102072179317474, "grad_norm": 0.09123480319976807, "global_step": 6707, "epoch": 63, "lr": 9.609928064013902e-05} +{"train_loss": 0.018823742866516113, "grad_norm": 0.07640323787927628, "global_step": 6708, "epoch": 63, "lr": 9.609813291717782e-05} +{"train_loss": 0.01652989722788334, "grad_norm": 0.07069429755210876, "global_step": 6709, "epoch": 63, "lr": 9.609698503224782e-05} +{"train_loss": 0.017611850053071976, "grad_norm": 0.09422910213470459, "global_step": 6710, "epoch": 63, "lr": 9.609583698535308e-05} +{"train_loss": 0.017762035131454468, "grad_norm": 0.09749893844127655, "global_step": 6711, "epoch": 63, "lr": 9.60946887764976e-05} +{"train_loss": 0.01607457362115383, "grad_norm": 0.05227409303188324, "global_step": 6712, "epoch": 63, "lr": 9.609354040568543e-05} +{"train_loss": 0.019339632242918015, "grad_norm": 0.04267444834113121, "global_step": 6713, "epoch": 63, "lr": 9.60923918729206e-05} +{"train_loss": 0.019811589270830154, "grad_norm": 0.050686970353126526, "global_step": 6714, "epoch": 63, "lr": 9.609124317820717e-05} +{"train_loss": 0.018297215923666954, "grad_norm": 0.13364215195178986, "global_step": 6715, "epoch": 63, "lr": 9.609009432154914e-05} +{"train_loss": 0.016901269555091858, "grad_norm": 0.053152699023485184, "global_step": 6716, "epoch": 63, "lr": 9.608894530295057e-05} +{"train_loss": 0.018437113612890244, "grad_norm": 0.08928178250789642, "global_step": 6717, "epoch": 63, "lr": 9.60877961224155e-05} +{"train_loss": 0.02104838192462921, "grad_norm": 0.07020311057567596, "global_step": 6718, "epoch": 63, "lr": 9.608664677994795e-05} +{"train_loss": 0.019005270674824715, "grad_norm": 0.05064985901117325, "global_step": 6719, "epoch": 63, "lr": 9.608549727555195e-05} +{"train_loss": 0.019168414175510406, "grad_norm": 0.09633084386587143, "global_step": 6720, "epoch": 63, "lr": 9.608434760923156e-05} +{"train_loss": 0.018012210726737976, "grad_norm": 0.08152022212743759, "global_step": 6721, "epoch": 63, "lr": 9.608319778099083e-05} +{"train_loss": 0.020517997443675995, "grad_norm": 0.08882594108581543, "global_step": 6722, "epoch": 63, "lr": 9.608204779083376e-05} +{"train_loss": 0.017861831933259964, "grad_norm": 0.05370175093412399, "global_step": 6723, "epoch": 63, "lr": 9.608089763876443e-05} +{"train_loss": 0.018033288419246674, "grad_norm": 0.04554841294884682, "global_step": 6724, "epoch": 63, "lr": 9.607974732478687e-05} +{"train_loss": 0.019741229712963104, "grad_norm": 0.08622507750988007, "global_step": 6725, "epoch": 63, "lr": 9.607859684890511e-05} +{"train_loss": 0.01823514699935913, "grad_norm": 0.05186721682548523, "global_step": 6726, "epoch": 63, "lr": 9.60774462111232e-05} +{"train_loss": 0.02135438844561577, "grad_norm": 0.08954735100269318, "global_step": 6727, "epoch": 63, "lr": 9.607629541144518e-05} +{"train_loss": 0.019729578867554665, "grad_norm": 0.09211260080337524, "global_step": 6728, "epoch": 63, "lr": 9.607514444987511e-05} +{"train_loss": 0.017735658213496208, "grad_norm": 0.0712219849228859, "global_step": 6729, "epoch": 63, "lr": 9.6073993326417e-05} +{"train_loss": 0.016508450731635094, "grad_norm": 0.07093651592731476, "global_step": 6730, "epoch": 63, "lr": 9.607284204107493e-05} +{"train_loss": 0.018346600234508514, "grad_norm": 0.0554681234061718, "global_step": 6731, "epoch": 63, "lr": 9.607169059385293e-05} +{"train_loss": 0.01877547986805439, "grad_norm": 0.05504164099693298, "global_step": 6732, "epoch": 63, "lr": 9.607053898475503e-05} +{"train_loss": 0.018118485808372498, "grad_norm": 0.04794079810380936, "global_step": 6733, "epoch": 63, "lr": 9.60693872137853e-05} +{"train_loss": 0.02015196532011032, "grad_norm": 0.13432766497135162, "global_step": 6734, "epoch": 63, "lr": 9.606823528094778e-05} +{"train_loss": 0.0181779395788908, "grad_norm": 0.10719478875398636, "global_step": 6735, "epoch": 63, "lr": 9.60670831862465e-05} +{"train_loss": 0.018963374197483063, "grad_norm": 0.09256229549646378, "global_step": 6736, "epoch": 63, "lr": 9.606593092968555e-05} +{"train_loss": 0.017794977873563766, "grad_norm": 0.041909340769052505, "global_step": 6737, "epoch": 63, "lr": 9.606477851126894e-05} +{"train_loss": 0.018655598163604736, "grad_norm": 0.05948563665151596, "global_step": 6738, "epoch": 63, "lr": 9.606362593100073e-05} +{"train_loss": 0.017540551722049713, "grad_norm": 0.040769290179014206, "global_step": 6739, "epoch": 63, "lr": 9.606247318888497e-05} +{"train_loss": 0.019810382276773453, "grad_norm": 0.12926527857780457, "global_step": 6740, "epoch": 63, "lr": 9.606132028492572e-05} +{"train_loss": 0.01886967197060585, "grad_norm": 0.058640025556087494, "global_step": 6741, "epoch": 63, "lr": 9.606016721912703e-05} +{"train_loss": 0.0167132206261158, "grad_norm": 0.06992499530315399, "global_step": 6742, "epoch": 63, "lr": 9.605901399149292e-05} +{"train_loss": 0.01733751967549324, "grad_norm": 0.10538794100284576, "global_step": 6743, "epoch": 63, "lr": 9.605786060202749e-05} +{"train_loss": 0.017861682921648026, "grad_norm": 0.05272507295012474, "global_step": 6744, "epoch": 63, "lr": 9.605670705073474e-05} +{"train_loss": 0.019593823701143265, "grad_norm": 0.08170844614505768, "global_step": 6745, "epoch": 63, "lr": 9.605555333761877e-05} +{"train_loss": 0.01888948678970337, "grad_norm": 0.08096733689308167, "global_step": 6746, "epoch": 63, "lr": 9.605439946268359e-05} +{"train_loss": 0.017763815820217133, "grad_norm": 0.06341823190450668, "global_step": 6747, "epoch": 63, "lr": 9.60532454259333e-05} +{"train_loss": 0.019818797707557678, "grad_norm": 0.10089951008558273, "global_step": 6748, "epoch": 63, "lr": 9.605209122737194e-05} +{"train_loss": 0.021889839321374893, "grad_norm": 0.13368570804595947, "global_step": 6749, "epoch": 63, "lr": 9.605093686700355e-05} +{"train_loss": 0.017285294830799103, "grad_norm": 0.10157792270183563, "global_step": 6750, "epoch": 63, "lr": 9.604978234483219e-05} +{"train_loss": 0.018109887838363647, "grad_norm": 0.049700669944286346, "global_step": 6751, "epoch": 63, "lr": 9.604862766086194e-05} +{"train_loss": 0.021533634513616562, "grad_norm": 0.0987936481833458, "global_step": 6752, "epoch": 63, "lr": 9.604747281509682e-05} +{"train_loss": 0.01858176477253437, "grad_norm": 0.14418824017047882, "global_step": 6753, "epoch": 63, "lr": 9.604631780754091e-05} +{"train_loss": 0.018633395433425903, "grad_norm": 0.048918191343545914, "global_step": 6754, "epoch": 63, "lr": 9.604516263819826e-05} +{"train_loss": 0.016972944140434265, "grad_norm": 0.054823096841573715, "global_step": 6755, "epoch": 63, "lr": 9.604400730707295e-05} +{"train_loss": 0.019242089241743088, "grad_norm": 0.13735072314739227, "global_step": 6756, "epoch": 63, "lr": 9.6042851814169e-05} +{"train_loss": 0.01648479886353016, "grad_norm": 0.047892797738313675, "global_step": 6757, "epoch": 63, "lr": 9.60416961594905e-05} +{"train_loss": 0.016213754191994667, "grad_norm": 0.08642823249101639, "global_step": 6758, "epoch": 63, "lr": 9.60405403430415e-05} +{"train_loss": 0.01775796338915825, "grad_norm": 0.0677894577383995, "global_step": 6759, "epoch": 63, "lr": 9.603938436482608e-05} +{"train_loss": 0.019936829805374146, "grad_norm": 0.10878583788871765, "global_step": 6760, "epoch": 63, "lr": 9.603822822484827e-05} +{"train_loss": 0.015959398820996284, "grad_norm": 0.04426221176981926, "global_step": 6761, "epoch": 63, "lr": 9.603707192311215e-05} +{"train_loss": 0.021572785452008247, "grad_norm": 0.06800762563943863, "global_step": 6762, "epoch": 63, "lr": 9.603591545962179e-05} +{"train_loss": 0.01587291806936264, "grad_norm": 0.09293069690465927, "global_step": 6763, "epoch": 63, "lr": 9.603475883438122e-05} +{"train_loss": 0.017417147755622864, "grad_norm": 0.038679659366607666, "global_step": 6764, "epoch": 63, "lr": 9.603360204739456e-05} +{"train_loss": 0.019225016236305237, "grad_norm": 0.05272795632481575, "global_step": 6765, "epoch": 63, "lr": 9.603244509866583e-05} +{"train_loss": 0.019796457141637802, "grad_norm": 0.06519535928964615, "global_step": 6766, "epoch": 63, "lr": 9.603128798819909e-05} +{"train_loss": 0.021836038678884506, "grad_norm": 0.08292744308710098, "global_step": 6767, "epoch": 63, "lr": 9.603013071599844e-05} +{"train_loss": 0.017798475921154022, "grad_norm": 0.05761643499135971, "global_step": 6768, "epoch": 63, "lr": 9.602897328206792e-05} +{"train_loss": 0.019676124677062035, "grad_norm": 0.1371626853942871, "global_step": 6769, "epoch": 63, "lr": 9.602781568641161e-05} +{"train_loss": 0.01905875653028488, "grad_norm": 0.044258490204811096, "global_step": 6770, "epoch": 63, "lr": 9.602665792903357e-05} +{"train_loss": 0.021091708913445473, "grad_norm": 0.06943968683481216, "global_step": 6771, "epoch": 63, "lr": 9.602550000993786e-05} +{"train_loss": 0.018596511334180832, "grad_norm": 0.04816750809550285, "global_step": 6772, "epoch": 63, "lr": 9.602434192912858e-05} +{"train_loss": 0.02019399404525757, "grad_norm": 0.0777641162276268, "global_step": 6773, "epoch": 63, "lr": 9.602318368660976e-05} +{"train_loss": 0.017911329865455627, "grad_norm": 0.055081676691770554, "global_step": 6774, "epoch": 63, "lr": 9.60220252823855e-05} +{"train_loss": 0.018858149647712708, "grad_norm": 0.06337001919746399, "global_step": 6775, "epoch": 63, "lr": 9.602086671645985e-05} +{"train_loss": 0.02154596708714962, "grad_norm": 0.07333359867334366, "global_step": 6776, "epoch": 63, "lr": 9.60197079888369e-05} +{"train_loss": 0.01972145214676857, "grad_norm": 0.06995406746864319, "global_step": 6777, "epoch": 63, "lr": 9.601854909952071e-05} +{"train_loss": 0.019607581198215485, "grad_norm": 0.04160333424806595, "global_step": 6778, "epoch": 63, "lr": 9.601739004851534e-05} +{"train_loss": 0.018336322158575058, "grad_norm": 0.0633709654211998, "global_step": 6779, "epoch": 63, "lr": 9.601623083582487e-05} +{"train_loss": 0.018457498401403427, "grad_norm": 0.08155667781829834, "global_step": 6780, "epoch": 63, "lr": 9.601507146145338e-05} +{"train_loss": 0.019035285338759422, "grad_norm": 0.06383886933326721, "global_step": 6781, "epoch": 63, "lr": 9.601391192540494e-05} +{"train_loss": 0.018075574189424515, "grad_norm": 0.05463818088173866, "global_step": 6782, "epoch": 63, "lr": 9.601275222768363e-05} +{"train_loss": 0.01867891644250672, "grad_norm": 0.10108362138271332, "global_step": 6783, "epoch": 63, "lr": 9.601159236829352e-05, "val_loss": 0.018976110965013504} +{"train_loss": 0.019216306507587433, "grad_norm": 0.09683885425329208, "global_step": 6784, "epoch": 64, "lr": 9.601043234723868e-05} +{"train_loss": 0.018443256616592407, "grad_norm": 0.09478723257780075, "global_step": 6785, "epoch": 64, "lr": 9.60092721645232e-05} +{"train_loss": 0.020011339336633682, "grad_norm": 0.04024389758706093, "global_step": 6786, "epoch": 64, "lr": 9.600811182015113e-05} +{"train_loss": 0.02201123535633087, "grad_norm": 0.05545373260974884, "global_step": 6787, "epoch": 64, "lr": 9.600695131412658e-05} +{"train_loss": 0.016198869794607162, "grad_norm": 0.08684048801660538, "global_step": 6788, "epoch": 64, "lr": 9.60057906464536e-05} +{"train_loss": 0.021269120275974274, "grad_norm": 0.06387102603912354, "global_step": 6789, "epoch": 64, "lr": 9.600462981713627e-05} +{"train_loss": 0.01607605069875717, "grad_norm": 0.062036316841840744, "global_step": 6790, "epoch": 64, "lr": 9.60034688261787e-05} +{"train_loss": 0.017583895474672318, "grad_norm": 0.05682713910937309, "global_step": 6791, "epoch": 64, "lr": 9.600230767358494e-05} +{"train_loss": 0.01844123750925064, "grad_norm": 0.10166557133197784, "global_step": 6792, "epoch": 64, "lr": 9.600114635935906e-05} +{"train_loss": 0.019887421280145645, "grad_norm": 0.08609151840209961, "global_step": 6793, "epoch": 64, "lr": 9.599998488350518e-05} +{"train_loss": 0.017738360911607742, "grad_norm": 0.12984012067317963, "global_step": 6794, "epoch": 64, "lr": 9.599882324602735e-05} +{"train_loss": 0.01697506755590439, "grad_norm": 0.0761883482336998, "global_step": 6795, "epoch": 64, "lr": 9.599766144692965e-05} +{"train_loss": 0.01563442498445511, "grad_norm": 0.12103985995054245, "global_step": 6796, "epoch": 64, "lr": 9.599649948621619e-05} +{"train_loss": 0.021092690527439117, "grad_norm": 0.043607693165540695, "global_step": 6797, "epoch": 64, "lr": 9.599533736389102e-05} +{"train_loss": 0.016267763450741768, "grad_norm": 0.04339923337101936, "global_step": 6798, "epoch": 64, "lr": 9.599417507995824e-05} +{"train_loss": 0.019052615389227867, "grad_norm": 0.06515716761350632, "global_step": 6799, "epoch": 64, "lr": 9.599301263442192e-05} +{"train_loss": 0.020694095641374588, "grad_norm": 0.16206814348697662, "global_step": 6800, "epoch": 64, "lr": 9.599185002728617e-05} +{"train_loss": 0.017923589795827866, "grad_norm": 0.0931793674826622, "global_step": 6801, "epoch": 64, "lr": 9.599068725855507e-05} +{"train_loss": 0.018806535750627518, "grad_norm": 0.10473687946796417, "global_step": 6802, "epoch": 64, "lr": 9.598952432823267e-05} +{"train_loss": 0.019485261291265488, "grad_norm": 0.09174890071153641, "global_step": 6803, "epoch": 64, "lr": 9.59883612363231e-05} +{"train_loss": 0.01774372160434723, "grad_norm": 0.12685230374336243, "global_step": 6804, "epoch": 64, "lr": 9.598719798283042e-05} +{"train_loss": 0.01805330440402031, "grad_norm": 0.08649152517318726, "global_step": 6805, "epoch": 64, "lr": 9.598603456775873e-05} +{"train_loss": 0.019983015954494476, "grad_norm": 0.07113033533096313, "global_step": 6806, "epoch": 64, "lr": 9.598487099111211e-05} +{"train_loss": 0.019627705216407776, "grad_norm": 0.10217924416065216, "global_step": 6807, "epoch": 64, "lr": 9.598370725289466e-05} +{"train_loss": 0.01905878074467182, "grad_norm": 0.05399177595973015, "global_step": 6808, "epoch": 64, "lr": 9.598254335311046e-05} +{"train_loss": 0.015922367572784424, "grad_norm": 0.07511766999959946, "global_step": 6809, "epoch": 64, "lr": 9.598137929176358e-05} +{"train_loss": 0.017467480152845383, "grad_norm": 0.12425335496664047, "global_step": 6810, "epoch": 64, "lr": 9.598021506885815e-05} +{"train_loss": 0.019741889089345932, "grad_norm": 0.0613342709839344, "global_step": 6811, "epoch": 64, "lr": 9.597905068439823e-05} +{"train_loss": 0.023360038176178932, "grad_norm": 0.0816199854016304, "global_step": 6812, "epoch": 64, "lr": 9.597788613838793e-05} +{"train_loss": 0.019649159163236618, "grad_norm": 0.12057460844516754, "global_step": 6813, "epoch": 64, "lr": 9.597672143083134e-05} +{"train_loss": 0.018128428608179092, "grad_norm": 0.06272051483392715, "global_step": 6814, "epoch": 64, "lr": 9.597555656173252e-05} +{"train_loss": 0.01858150213956833, "grad_norm": 0.10847001522779465, "global_step": 6815, "epoch": 64, "lr": 9.597439153109562e-05} +{"train_loss": 0.015909090638160706, "grad_norm": 0.08008965849876404, "global_step": 6816, "epoch": 64, "lr": 9.597322633892468e-05} +{"train_loss": 0.018555644899606705, "grad_norm": 0.06340506672859192, "global_step": 6817, "epoch": 64, "lr": 9.597206098522382e-05} +{"train_loss": 0.016037482768297195, "grad_norm": 0.05542173981666565, "global_step": 6818, "epoch": 64, "lr": 9.597089546999713e-05} +{"train_loss": 0.019045647233724594, "grad_norm": 0.05574030801653862, "global_step": 6819, "epoch": 64, "lr": 9.596972979324872e-05} +{"train_loss": 0.020565848797559738, "grad_norm": 0.1163703054189682, "global_step": 6820, "epoch": 64, "lr": 9.596856395498265e-05} +{"train_loss": 0.02248571068048477, "grad_norm": 0.1930074244737625, "global_step": 6821, "epoch": 64, "lr": 9.596739795520304e-05} +{"train_loss": 0.019117526710033417, "grad_norm": 0.11412162333726883, "global_step": 6822, "epoch": 64, "lr": 9.5966231793914e-05} +{"train_loss": 0.01661837473511696, "grad_norm": 0.05398695915937424, "global_step": 6823, "epoch": 64, "lr": 9.59650654711196e-05} +{"train_loss": 0.017269477248191833, "grad_norm": 0.09100815653800964, "global_step": 6824, "epoch": 64, "lr": 9.596389898682395e-05} +{"train_loss": 0.016273258253932, "grad_norm": 0.05056202411651611, "global_step": 6825, "epoch": 64, "lr": 9.596273234103115e-05} +{"train_loss": 0.018611472100019455, "grad_norm": 0.11998261511325836, "global_step": 6826, "epoch": 64, "lr": 9.596156553374529e-05} +{"train_loss": 0.019223153591156006, "grad_norm": 0.07699674367904663, "global_step": 6827, "epoch": 64, "lr": 9.596039856497049e-05} +{"train_loss": 0.020928865298628807, "grad_norm": 0.06723054498434067, "global_step": 6828, "epoch": 64, "lr": 9.595923143471084e-05} +{"train_loss": 0.016155458986759186, "grad_norm": 0.09292497485876083, "global_step": 6829, "epoch": 64, "lr": 9.595806414297041e-05} +{"train_loss": 0.020609470084309578, "grad_norm": 0.11516715586185455, "global_step": 6830, "epoch": 64, "lr": 9.595689668975336e-05} +{"train_loss": 0.01785343512892723, "grad_norm": 0.13930541276931763, "global_step": 6831, "epoch": 64, "lr": 9.595572907506375e-05} +{"train_loss": 0.018401382490992546, "grad_norm": 0.11384499818086624, "global_step": 6832, "epoch": 64, "lr": 9.59545612989057e-05} +{"train_loss": 0.020201491191983223, "grad_norm": 0.07519371062517166, "global_step": 6833, "epoch": 64, "lr": 9.595339336128329e-05} +{"train_loss": 0.020919589325785637, "grad_norm": 0.051818158477544785, "global_step": 6834, "epoch": 64, "lr": 9.595222526220066e-05} +{"train_loss": 0.020915132015943527, "grad_norm": 0.063829205930233, "global_step": 6835, "epoch": 64, "lr": 9.595105700166187e-05} +{"train_loss": 0.01902170106768608, "grad_norm": 0.07397941499948502, "global_step": 6836, "epoch": 64, "lr": 9.594988857967106e-05} +{"train_loss": 0.016690336167812347, "grad_norm": 0.1274404227733612, "global_step": 6837, "epoch": 64, "lr": 9.594871999623234e-05} +{"train_loss": 0.02019762061536312, "grad_norm": 0.08009625226259232, "global_step": 6838, "epoch": 64, "lr": 9.594755125134979e-05} +{"train_loss": 0.016961008310317993, "grad_norm": 0.05762726441025734, "global_step": 6839, "epoch": 64, "lr": 9.594638234502753e-05} +{"train_loss": 0.020100906491279602, "grad_norm": 0.08279169350862503, "global_step": 6840, "epoch": 64, "lr": 9.594521327726967e-05} +{"train_loss": 0.016660217195749283, "grad_norm": 0.11724857985973358, "global_step": 6841, "epoch": 64, "lr": 9.594404404808029e-05} +{"train_loss": 0.02182336337864399, "grad_norm": 0.1547357141971588, "global_step": 6842, "epoch": 64, "lr": 9.594287465746352e-05} +{"train_loss": 0.018488187342882156, "grad_norm": 0.06872976571321487, "global_step": 6843, "epoch": 64, "lr": 9.594170510542348e-05} +{"train_loss": 0.01936391368508339, "grad_norm": 0.06275003403425217, "global_step": 6844, "epoch": 64, "lr": 9.594053539196428e-05} +{"train_loss": 0.018841836601495743, "grad_norm": 0.17992331087589264, "global_step": 6845, "epoch": 64, "lr": 9.593936551709e-05} +{"train_loss": 0.018139392137527466, "grad_norm": 0.05869534984230995, "global_step": 6846, "epoch": 64, "lr": 9.593819548080478e-05} +{"train_loss": 0.018604587763547897, "grad_norm": 0.0647122710943222, "global_step": 6847, "epoch": 64, "lr": 9.593702528311271e-05} +{"train_loss": 0.01748712547123432, "grad_norm": 0.1399977207183838, "global_step": 6848, "epoch": 64, "lr": 9.593585492401791e-05} +{"train_loss": 0.021950645372271538, "grad_norm": 0.06732958555221558, "global_step": 6849, "epoch": 64, "lr": 9.59346844035245e-05} +{"train_loss": 0.021353207528591156, "grad_norm": 0.07602179050445557, "global_step": 6850, "epoch": 64, "lr": 9.593351372163658e-05} +{"train_loss": 0.016323335468769073, "grad_norm": 0.08180078119039536, "global_step": 6851, "epoch": 64, "lr": 9.593234287835826e-05} +{"train_loss": 0.01827850565314293, "grad_norm": 0.08757588267326355, "global_step": 6852, "epoch": 64, "lr": 9.593117187369368e-05} +{"train_loss": 0.02426929771900177, "grad_norm": 0.06996989250183105, "global_step": 6853, "epoch": 64, "lr": 9.593000070764692e-05} +{"train_loss": 0.020066581666469574, "grad_norm": 0.045861244201660156, "global_step": 6854, "epoch": 64, "lr": 9.592882938022213e-05} +{"train_loss": 0.019809283316135406, "grad_norm": 0.08280269056558609, "global_step": 6855, "epoch": 64, "lr": 9.592765789142339e-05} +{"train_loss": 0.015546341426670551, "grad_norm": 0.044272832572460175, "global_step": 6856, "epoch": 64, "lr": 9.592648624125485e-05} +{"train_loss": 0.019969774410128593, "grad_norm": 0.08835093677043915, "global_step": 6857, "epoch": 64, "lr": 9.592531442972061e-05} +{"train_loss": 0.01700728014111519, "grad_norm": 0.23264776170253754, "global_step": 6858, "epoch": 64, "lr": 9.592414245682478e-05} +{"train_loss": 0.020503634586930275, "grad_norm": 0.17372813820838928, "global_step": 6859, "epoch": 64, "lr": 9.59229703225715e-05} +{"train_loss": 0.01820870116353035, "grad_norm": 0.1185840517282486, "global_step": 6860, "epoch": 64, "lr": 9.592179802696484e-05} +{"train_loss": 0.017892036586999893, "grad_norm": 0.17428244650363922, "global_step": 6861, "epoch": 64, "lr": 9.592062557000898e-05} +{"train_loss": 0.018302638083696365, "grad_norm": 0.07992380112409592, "global_step": 6862, "epoch": 64, "lr": 9.5919452951708e-05} +{"train_loss": 0.017794303596019745, "grad_norm": 0.11313895881175995, "global_step": 6863, "epoch": 64, "lr": 9.591828017206603e-05} +{"train_loss": 0.020292392000555992, "grad_norm": 0.044745009392499924, "global_step": 6864, "epoch": 64, "lr": 9.59171072310872e-05} +{"train_loss": 0.016861334443092346, "grad_norm": 0.04690917953848839, "global_step": 6865, "epoch": 64, "lr": 9.591593412877563e-05} +{"train_loss": 0.016909444704651833, "grad_norm": 0.05723126605153084, "global_step": 6866, "epoch": 64, "lr": 9.591476086513543e-05} +{"train_loss": 0.018527470529079437, "grad_norm": 0.06292939931154251, "global_step": 6867, "epoch": 64, "lr": 9.591358744017071e-05} +{"train_loss": 0.0165170356631279, "grad_norm": 0.07762306183576584, "global_step": 6868, "epoch": 64, "lr": 9.591241385388564e-05} +{"train_loss": 0.017241451889276505, "grad_norm": 0.09516394138336182, "global_step": 6869, "epoch": 64, "lr": 9.591124010628431e-05} +{"train_loss": 0.022396046668291092, "grad_norm": 0.14023683965206146, "global_step": 6870, "epoch": 64, "lr": 9.591006619737084e-05} +{"train_loss": 0.017804084345698357, "grad_norm": 0.1484813690185547, "global_step": 6871, "epoch": 64, "lr": 9.590889212714935e-05} +{"train_loss": 0.018681030720472336, "grad_norm": 0.1490359604358673, "global_step": 6872, "epoch": 64, "lr": 9.590771789562401e-05} +{"train_loss": 0.01928473822772503, "grad_norm": 0.12097262591123581, "global_step": 6873, "epoch": 64, "lr": 9.590654350279889e-05} +{"train_loss": 0.017881684005260468, "grad_norm": 0.1135006919503212, "global_step": 6874, "epoch": 64, "lr": 9.590536894867814e-05} +{"train_loss": 0.017570234835147858, "grad_norm": 0.17451782524585724, "global_step": 6875, "epoch": 64, "lr": 9.59041942332659e-05} +{"train_loss": 0.016170945018529892, "grad_norm": 0.10423938184976578, "global_step": 6876, "epoch": 64, "lr": 9.590301935656627e-05} +{"train_loss": 0.015599381178617477, "grad_norm": 0.03920338675379753, "global_step": 6877, "epoch": 64, "lr": 9.59018443185834e-05} +{"train_loss": 0.01822138950228691, "grad_norm": 0.05724890157580376, "global_step": 6878, "epoch": 64, "lr": 9.590066911932143e-05} +{"train_loss": 0.019615113735198975, "grad_norm": 0.06377284973859787, "global_step": 6879, "epoch": 64, "lr": 9.589949375878444e-05} +{"train_loss": 0.01956445351243019, "grad_norm": 0.07580289989709854, "global_step": 6880, "epoch": 64, "lr": 9.589831823697662e-05} +{"train_loss": 0.01767018809914589, "grad_norm": 0.08727376908063889, "global_step": 6881, "epoch": 64, "lr": 9.589714255390204e-05} +{"train_loss": 0.01913902349770069, "grad_norm": 0.05866007134318352, "global_step": 6882, "epoch": 64, "lr": 9.589596670956487e-05} +{"train_loss": 0.018822671845555305, "grad_norm": 0.20719245076179504, "global_step": 6883, "epoch": 64, "lr": 9.589479070396925e-05} +{"train_loss": 0.016294807195663452, "grad_norm": 0.12461799383163452, "global_step": 6884, "epoch": 64, "lr": 9.589361453711927e-05} +{"train_loss": 0.018900495022535324, "grad_norm": 0.0588345006108284, "global_step": 6885, "epoch": 64, "lr": 9.58924382090191e-05} +{"train_loss": 0.020672589540481567, "grad_norm": 0.1278584748506546, "global_step": 6886, "epoch": 64, "lr": 9.589126171967286e-05} +{"train_loss": 0.018349844962358475, "grad_norm": 0.044045694172382355, "global_step": 6887, "epoch": 64, "lr": 9.589008506908468e-05} +{"train_loss": 0.01758585497736931, "grad_norm": 0.10336198657751083, "global_step": 6888, "epoch": 64, "lr": 9.588890825725871e-05} +{"train_loss": 0.018704698623140465, "grad_norm": 0.12131750583648682, "global_step": 6889, "epoch": 64, "lr": 9.588773128419906e-05, "val_loss": 0.01873721368610859} +{"train_loss": 0.021220043301582336, "grad_norm": 0.07045092433691025, "global_step": 6890, "epoch": 65, "lr": 9.588655414990988e-05} +{"train_loss": 0.020635373890399933, "grad_norm": 0.07625553011894226, "global_step": 6891, "epoch": 65, "lr": 9.588537685439531e-05} +{"train_loss": 0.018068693578243256, "grad_norm": 0.05820940062403679, "global_step": 6892, "epoch": 65, "lr": 9.588419939765948e-05} +{"train_loss": 0.01999250426888466, "grad_norm": 0.08460217714309692, "global_step": 6893, "epoch": 65, "lr": 9.588302177970651e-05} +{"train_loss": 0.018776174634695053, "grad_norm": 0.1876230090856552, "global_step": 6894, "epoch": 65, "lr": 9.588184400054058e-05} +{"train_loss": 0.016980765387415886, "grad_norm": 0.048171669244766235, "global_step": 6895, "epoch": 65, "lr": 9.588066606016579e-05} +{"train_loss": 0.016337081789970398, "grad_norm": 0.042687300592660904, "global_step": 6896, "epoch": 65, "lr": 9.58794879585863e-05} +{"train_loss": 0.018007559701800346, "grad_norm": 0.08247945457696915, "global_step": 6897, "epoch": 65, "lr": 9.587830969580625e-05} +{"train_loss": 0.01926560327410698, "grad_norm": 0.04845425486564636, "global_step": 6898, "epoch": 65, "lr": 9.587713127182975e-05} +{"train_loss": 0.0181471835821867, "grad_norm": 0.1245906725525856, "global_step": 6899, "epoch": 65, "lr": 9.587595268666099e-05} +{"train_loss": 0.01790722645819187, "grad_norm": 0.08579752594232559, "global_step": 6900, "epoch": 65, "lr": 9.587477394030407e-05} +{"train_loss": 0.01926490291953087, "grad_norm": 0.05933172628283501, "global_step": 6901, "epoch": 65, "lr": 9.587359503276312e-05} +{"train_loss": 0.01818123646080494, "grad_norm": 0.12342435866594315, "global_step": 6902, "epoch": 65, "lr": 9.587241596404232e-05} +{"train_loss": 0.020323891192674637, "grad_norm": 0.10812787711620331, "global_step": 6903, "epoch": 65, "lr": 9.587123673414581e-05} +{"train_loss": 0.019878048449754715, "grad_norm": 0.1398411989212036, "global_step": 6904, "epoch": 65, "lr": 9.587005734307773e-05} +{"train_loss": 0.020324602723121643, "grad_norm": 0.048576463013887405, "global_step": 6905, "epoch": 65, "lr": 9.586887779084222e-05} +{"train_loss": 0.017444614320993423, "grad_norm": 0.052298251539468765, "global_step": 6906, "epoch": 65, "lr": 9.586769807744341e-05} +{"train_loss": 0.017237193882465363, "grad_norm": 0.0636119693517685, "global_step": 6907, "epoch": 65, "lr": 9.586651820288543e-05} +{"train_loss": 0.016275277361273766, "grad_norm": 0.050017379224300385, "global_step": 6908, "epoch": 65, "lr": 9.586533816717249e-05} +{"train_loss": 0.016687951982021332, "grad_norm": 0.09838757663965225, "global_step": 6909, "epoch": 65, "lr": 9.586415797030867e-05} +{"train_loss": 0.017431404441595078, "grad_norm": 0.09923022985458374, "global_step": 6910, "epoch": 65, "lr": 9.586297761229816e-05} +{"train_loss": 0.017309458926320076, "grad_norm": 0.07091093063354492, "global_step": 6911, "epoch": 65, "lr": 9.58617970931451e-05} +{"train_loss": 0.01610967144370079, "grad_norm": 0.055256400257349014, "global_step": 6912, "epoch": 65, "lr": 9.586061641285362e-05} +{"train_loss": 0.022071514278650284, "grad_norm": 0.16060176491737366, "global_step": 6913, "epoch": 65, "lr": 9.585943557142788e-05} +{"train_loss": 0.019190030172467232, "grad_norm": 0.05550847947597504, "global_step": 6914, "epoch": 65, "lr": 9.585825456887203e-05} +{"train_loss": 0.018875012174248695, "grad_norm": 0.06416724622249603, "global_step": 6915, "epoch": 65, "lr": 9.585707340519021e-05} +{"train_loss": 0.02239280194044113, "grad_norm": 0.08279650658369064, "global_step": 6916, "epoch": 65, "lr": 9.585589208038658e-05} +{"train_loss": 0.020630553364753723, "grad_norm": 0.056981220841407776, "global_step": 6917, "epoch": 65, "lr": 9.585471059446529e-05} +{"train_loss": 0.018853582441806793, "grad_norm": 0.13038377463817596, "global_step": 6918, "epoch": 65, "lr": 9.585352894743048e-05} +{"train_loss": 0.021593162789940834, "grad_norm": 0.07388288527727127, "global_step": 6919, "epoch": 65, "lr": 9.585234713928633e-05} +{"train_loss": 0.01709221675992012, "grad_norm": 0.08682502806186676, "global_step": 6920, "epoch": 65, "lr": 9.585116517003695e-05} +{"train_loss": 0.02028411254286766, "grad_norm": 0.053640641272068024, "global_step": 6921, "epoch": 65, "lr": 9.584998303968653e-05} +{"train_loss": 0.017973685637116432, "grad_norm": 0.047264136373996735, "global_step": 6922, "epoch": 65, "lr": 9.58488007482392e-05} +{"train_loss": 0.018106788396835327, "grad_norm": 0.04740032181143761, "global_step": 6923, "epoch": 65, "lr": 9.584761829569912e-05} +{"train_loss": 0.016607563942670822, "grad_norm": 0.03718895837664604, "global_step": 6924, "epoch": 65, "lr": 9.584643568207046e-05} +{"train_loss": 0.018511496484279633, "grad_norm": 0.05249422416090965, "global_step": 6925, "epoch": 65, "lr": 9.584525290735736e-05} +{"train_loss": 0.018094809725880623, "grad_norm": 0.0714641734957695, "global_step": 6926, "epoch": 65, "lr": 9.584406997156397e-05} +{"train_loss": 0.01954372227191925, "grad_norm": 0.046589139848947525, "global_step": 6927, "epoch": 65, "lr": 9.584288687469447e-05} +{"train_loss": 0.021485544741153717, "grad_norm": 0.10049236565828323, "global_step": 6928, "epoch": 65, "lr": 9.5841703616753e-05} +{"train_loss": 0.01611984148621559, "grad_norm": 0.09468864649534225, "global_step": 6929, "epoch": 65, "lr": 9.58405201977437e-05} +{"train_loss": 0.023689361289143562, "grad_norm": 0.14108654856681824, "global_step": 6930, "epoch": 65, "lr": 9.583933661767078e-05} +{"train_loss": 0.017523715272545815, "grad_norm": 0.044670384377241135, "global_step": 6931, "epoch": 65, "lr": 9.583815287653833e-05} +{"train_loss": 0.015126589685678482, "grad_norm": 0.06901991367340088, "global_step": 6932, "epoch": 65, "lr": 9.583696897435057e-05} +{"train_loss": 0.017358306795358658, "grad_norm": 0.056835755705833435, "global_step": 6933, "epoch": 65, "lr": 9.583578491111163e-05} +{"train_loss": 0.020616406574845314, "grad_norm": 0.057838212698698044, "global_step": 6934, "epoch": 65, "lr": 9.583460068682567e-05} +{"train_loss": 0.017035232856869698, "grad_norm": 0.05816717818379402, "global_step": 6935, "epoch": 65, "lr": 9.583341630149684e-05} +{"train_loss": 0.02093815803527832, "grad_norm": 0.1058560311794281, "global_step": 6936, "epoch": 65, "lr": 9.583223175512936e-05} +{"train_loss": 0.01724688708782196, "grad_norm": 0.08124880492687225, "global_step": 6937, "epoch": 65, "lr": 9.583104704772732e-05} +{"train_loss": 0.0183588657528162, "grad_norm": 0.0718340277671814, "global_step": 6938, "epoch": 65, "lr": 9.582986217929491e-05} +{"train_loss": 0.018196001648902893, "grad_norm": 0.046128176152706146, "global_step": 6939, "epoch": 65, "lr": 9.582867714983632e-05} +{"train_loss": 0.01793374866247177, "grad_norm": 0.04062221944332123, "global_step": 6940, "epoch": 65, "lr": 9.582749195935566e-05} +{"train_loss": 0.01658482663333416, "grad_norm": 0.11949367076158524, "global_step": 6941, "epoch": 65, "lr": 9.582630660785715e-05} +{"train_loss": 0.01811041496694088, "grad_norm": 0.06425350159406662, "global_step": 6942, "epoch": 65, "lr": 9.58251210953449e-05} +{"train_loss": 0.01852092705667019, "grad_norm": 0.08793367445468903, "global_step": 6943, "epoch": 65, "lr": 9.582393542182312e-05} +{"train_loss": 0.017643531784415245, "grad_norm": 0.06309455633163452, "global_step": 6944, "epoch": 65, "lr": 9.582274958729595e-05} +{"train_loss": 0.01632853038609028, "grad_norm": 0.04803203418850899, "global_step": 6945, "epoch": 65, "lr": 9.582156359176758e-05} +{"train_loss": 0.01813787966966629, "grad_norm": 0.04885442554950714, "global_step": 6946, "epoch": 65, "lr": 9.582037743524215e-05} +{"train_loss": 0.01844296045601368, "grad_norm": 0.06857716292142868, "global_step": 6947, "epoch": 65, "lr": 9.581919111772385e-05} +{"train_loss": 0.01722269505262375, "grad_norm": 0.12015915662050247, "global_step": 6948, "epoch": 65, "lr": 9.581800463921682e-05} +{"train_loss": 0.01984502747654915, "grad_norm": 0.062149111181497574, "global_step": 6949, "epoch": 65, "lr": 9.581681799972527e-05} +{"train_loss": 0.018444523215293884, "grad_norm": 0.040862955152988434, "global_step": 6950, "epoch": 65, "lr": 9.581563119925334e-05} +{"train_loss": 0.022136598825454712, "grad_norm": 0.1115606427192688, "global_step": 6951, "epoch": 65, "lr": 9.581444423780521e-05} +{"train_loss": 0.018543075770139694, "grad_norm": 0.05544458329677582, "global_step": 6952, "epoch": 65, "lr": 9.581325711538504e-05} +{"train_loss": 0.01973005011677742, "grad_norm": 0.05680304020643234, "global_step": 6953, "epoch": 65, "lr": 9.5812069831997e-05} +{"train_loss": 0.016941770911216736, "grad_norm": 0.051007337868213654, "global_step": 6954, "epoch": 65, "lr": 9.581088238764528e-05} +{"train_loss": 0.015336068347096443, "grad_norm": 0.05508594214916229, "global_step": 6955, "epoch": 65, "lr": 9.580969478233405e-05} +{"train_loss": 0.016966313123703003, "grad_norm": 0.11320586502552032, "global_step": 6956, "epoch": 65, "lr": 9.580850701606746e-05} +{"train_loss": 0.019430212676525116, "grad_norm": 0.16141366958618164, "global_step": 6957, "epoch": 65, "lr": 9.58073190888497e-05} +{"train_loss": 0.020881228148937225, "grad_norm": 0.06999107450246811, "global_step": 6958, "epoch": 65, "lr": 9.580613100068495e-05} +{"train_loss": 0.0172266848385334, "grad_norm": 0.05812588334083557, "global_step": 6959, "epoch": 65, "lr": 9.580494275157737e-05} +{"train_loss": 0.018086932599544525, "grad_norm": 0.05577275529503822, "global_step": 6960, "epoch": 65, "lr": 9.580375434153115e-05} +{"train_loss": 0.01970638707280159, "grad_norm": 0.0976128950715065, "global_step": 6961, "epoch": 65, "lr": 9.580256577055045e-05} +{"train_loss": 0.01680024154484272, "grad_norm": 0.06920521706342697, "global_step": 6962, "epoch": 65, "lr": 9.580137703863945e-05} +{"train_loss": 0.019464239478111267, "grad_norm": 0.079900823533535, "global_step": 6963, "epoch": 65, "lr": 9.580018814580233e-05} +{"train_loss": 0.017896533012390137, "grad_norm": 0.04685644432902336, "global_step": 6964, "epoch": 65, "lr": 9.579899909204328e-05} +{"train_loss": 0.019129056483507156, "grad_norm": 0.05528465285897255, "global_step": 6965, "epoch": 65, "lr": 9.579780987736644e-05} +{"train_loss": 0.018990391865372658, "grad_norm": 0.07895992696285248, "global_step": 6966, "epoch": 65, "lr": 9.579662050177603e-05} +{"train_loss": 0.016548609361052513, "grad_norm": 0.07171456515789032, "global_step": 6967, "epoch": 65, "lr": 9.579543096527622e-05} +{"train_loss": 0.017426669597625732, "grad_norm": 0.11457375437021255, "global_step": 6968, "epoch": 65, "lr": 9.579424126787116e-05} +{"train_loss": 0.017573736608028412, "grad_norm": 0.07694157212972641, "global_step": 6969, "epoch": 65, "lr": 9.579305140956508e-05} +{"train_loss": 0.019925791770219803, "grad_norm": 0.06764781475067139, "global_step": 6970, "epoch": 65, "lr": 9.579186139036209e-05} +{"train_loss": 0.01772618293762207, "grad_norm": 0.03443457931280136, "global_step": 6971, "epoch": 65, "lr": 9.579067121026645e-05} +{"train_loss": 0.016787391155958176, "grad_norm": 0.046194422990083694, "global_step": 6972, "epoch": 65, "lr": 9.578948086928229e-05} +{"train_loss": 0.015378067269921303, "grad_norm": 0.06201159209012985, "global_step": 6973, "epoch": 65, "lr": 9.578829036741381e-05} +{"train_loss": 0.015972193330526352, "grad_norm": 0.08681824058294296, "global_step": 6974, "epoch": 65, "lr": 9.57870997046652e-05} +{"train_loss": 0.019458886235952377, "grad_norm": 0.049550849944353104, "global_step": 6975, "epoch": 65, "lr": 9.578590888104062e-05} +{"train_loss": 0.02111717127263546, "grad_norm": 0.13033296167850494, "global_step": 6976, "epoch": 65, "lr": 9.578471789654426e-05} +{"train_loss": 0.018691862002015114, "grad_norm": 0.065072201192379, "global_step": 6977, "epoch": 65, "lr": 9.578352675118033e-05} +{"train_loss": 0.019192099571228027, "grad_norm": 0.06852211803197861, "global_step": 6978, "epoch": 65, "lr": 9.5782335444953e-05} +{"train_loss": 0.017688188701868057, "grad_norm": 0.04143216833472252, "global_step": 6979, "epoch": 65, "lr": 9.578114397786644e-05} +{"train_loss": 0.019979022443294525, "grad_norm": 0.07910194247961044, "global_step": 6980, "epoch": 65, "lr": 9.577995234992485e-05} +{"train_loss": 0.02109963446855545, "grad_norm": 0.08614552021026611, "global_step": 6981, "epoch": 65, "lr": 9.577876056113243e-05} +{"train_loss": 0.019940264523029327, "grad_norm": 0.10107197612524033, "global_step": 6982, "epoch": 65, "lr": 9.577756861149334e-05} +{"train_loss": 0.017201019451022148, "grad_norm": 0.10227544605731964, "global_step": 6983, "epoch": 65, "lr": 9.577637650101179e-05} +{"train_loss": 0.01854601874947548, "grad_norm": 0.04888424649834633, "global_step": 6984, "epoch": 65, "lr": 9.577518422969196e-05} +{"train_loss": 0.0198868066072464, "grad_norm": 0.07908408343791962, "global_step": 6985, "epoch": 65, "lr": 9.577399179753804e-05} +{"train_loss": 0.018982799723744392, "grad_norm": 0.07195524126291275, "global_step": 6986, "epoch": 65, "lr": 9.577279920455422e-05} +{"train_loss": 0.017070509493350983, "grad_norm": 0.07387259602546692, "global_step": 6987, "epoch": 65, "lr": 9.57716064507447e-05} +{"train_loss": 0.016988223418593407, "grad_norm": 0.07290450483560562, "global_step": 6988, "epoch": 65, "lr": 9.577041353611365e-05} +{"train_loss": 0.017940577119588852, "grad_norm": 0.08600696921348572, "global_step": 6989, "epoch": 65, "lr": 9.576922046066525e-05} +{"train_loss": 0.02092479169368744, "grad_norm": 0.06825011223554611, "global_step": 6990, "epoch": 65, "lr": 9.576802722440374e-05} +{"train_loss": 0.020405856892466545, "grad_norm": 0.07393541932106018, "global_step": 6991, "epoch": 65, "lr": 9.576683382733329e-05} +{"train_loss": 0.021093958988785744, "grad_norm": 0.06242187321186066, "global_step": 6992, "epoch": 65, "lr": 9.576564026945807e-05} +{"train_loss": 0.01959899067878723, "grad_norm": 0.08001551032066345, "global_step": 6993, "epoch": 65, "lr": 9.576444655078231e-05} +{"train_loss": 0.017286043614149094, "grad_norm": 0.10298244655132294, "global_step": 6994, "epoch": 65, "lr": 9.576325267131018e-05} +{"train_loss": 0.018586107055252453, "grad_norm": 0.09386289119720459, "global_step": 6995, "epoch": 65, "lr": 9.576205863104588e-05, "val_loss": 0.018975473940372467} +{"train_loss": 0.018496155738830566, "grad_norm": 0.19698691368103027, "global_step": 6996, "epoch": 66, "lr": 9.576086442999361e-05} +{"train_loss": 0.02015964314341545, "grad_norm": 0.055843181908130646, "global_step": 6997, "epoch": 66, "lr": 9.575967006815757e-05} +{"train_loss": 0.018782079219818115, "grad_norm": 0.04893554374575615, "global_step": 6998, "epoch": 66, "lr": 9.575847554554194e-05} +{"train_loss": 0.01847606524825096, "grad_norm": 0.1480480283498764, "global_step": 6999, "epoch": 66, "lr": 9.575728086215092e-05} +{"train_loss": 0.018853390589356422, "grad_norm": 0.08867130428552628, "global_step": 7000, "epoch": 66, "lr": 9.575608601798873e-05} +{"train_loss": 0.018439946696162224, "grad_norm": 0.06840426474809647, "global_step": 7001, "epoch": 66, "lr": 9.575489101305955e-05} +{"train_loss": 0.015669871121644974, "grad_norm": 0.18932944536209106, "global_step": 7002, "epoch": 66, "lr": 9.575369584736758e-05} +{"train_loss": 0.016509806737303734, "grad_norm": 0.058729685842990875, "global_step": 7003, "epoch": 66, "lr": 9.575250052091702e-05} +{"train_loss": 0.018640276044607162, "grad_norm": 0.08054690062999725, "global_step": 7004, "epoch": 66, "lr": 9.575130503371206e-05} +{"train_loss": 0.018327701836824417, "grad_norm": 0.058808065950870514, "global_step": 7005, "epoch": 66, "lr": 9.575010938575693e-05} +{"train_loss": 0.018101345747709274, "grad_norm": 0.047604721039533615, "global_step": 7006, "epoch": 66, "lr": 9.57489135770558e-05} +{"train_loss": 0.016359437257051468, "grad_norm": 0.044238340109586716, "global_step": 7007, "epoch": 66, "lr": 9.574771760761288e-05} +{"train_loss": 0.01830640807747841, "grad_norm": 0.062403373420238495, "global_step": 7008, "epoch": 66, "lr": 9.574652147743238e-05} +{"train_loss": 0.016391173005104065, "grad_norm": 0.16578003764152527, "global_step": 7009, "epoch": 66, "lr": 9.57453251865185e-05} +{"train_loss": 0.020446233451366425, "grad_norm": 0.05956276133656502, "global_step": 7010, "epoch": 66, "lr": 9.574412873487543e-05} +{"train_loss": 0.01845639944076538, "grad_norm": 0.1352182924747467, "global_step": 7011, "epoch": 66, "lr": 9.57429321225074e-05} +{"train_loss": 0.017961006611585617, "grad_norm": 0.12357280403375626, "global_step": 7012, "epoch": 66, "lr": 9.57417353494186e-05} +{"train_loss": 0.019911356270313263, "grad_norm": 0.059404078871011734, "global_step": 7013, "epoch": 66, "lr": 9.574053841561323e-05} +{"train_loss": 0.01659272611141205, "grad_norm": 0.04568847268819809, "global_step": 7014, "epoch": 66, "lr": 9.57393413210955e-05} +{"train_loss": 0.018864808604121208, "grad_norm": 0.06142789125442505, "global_step": 7015, "epoch": 66, "lr": 9.573814406586961e-05} +{"train_loss": 0.020548444241285324, "grad_norm": 0.06368295103311539, "global_step": 7016, "epoch": 66, "lr": 9.573694664993977e-05} +{"train_loss": 0.017563603818416595, "grad_norm": 0.07975407689809799, "global_step": 7017, "epoch": 66, "lr": 9.573574907331019e-05} +{"train_loss": 0.02016643062233925, "grad_norm": 0.05432466045022011, "global_step": 7018, "epoch": 66, "lr": 9.57345513359851e-05} +{"train_loss": 0.014771023765206337, "grad_norm": 0.056933965533971786, "global_step": 7019, "epoch": 66, "lr": 9.573335343796867e-05} +{"train_loss": 0.019238894805312157, "grad_norm": 0.0996749997138977, "global_step": 7020, "epoch": 66, "lr": 9.573215537926513e-05} +{"train_loss": 0.01695859059691429, "grad_norm": 0.04900914803147316, "global_step": 7021, "epoch": 66, "lr": 9.573095715987868e-05} +{"train_loss": 0.019680960103869438, "grad_norm": 0.09637442976236343, "global_step": 7022, "epoch": 66, "lr": 9.572975877981354e-05} +{"train_loss": 0.015432136133313179, "grad_norm": 0.05810019001364708, "global_step": 7023, "epoch": 66, "lr": 9.572856023907391e-05} +{"train_loss": 0.016783829778432846, "grad_norm": 0.10824882984161377, "global_step": 7024, "epoch": 66, "lr": 9.572736153766401e-05} +{"train_loss": 0.01666952669620514, "grad_norm": 0.04150279238820076, "global_step": 7025, "epoch": 66, "lr": 9.572616267558803e-05} +{"train_loss": 0.018910333514213562, "grad_norm": 0.09996367990970612, "global_step": 7026, "epoch": 66, "lr": 9.572496365285022e-05} +{"train_loss": 0.016758419573307037, "grad_norm": 0.04992469400167465, "global_step": 7027, "epoch": 66, "lr": 9.572376446945477e-05} +{"train_loss": 0.013888142071664333, "grad_norm": 0.06803138554096222, "global_step": 7028, "epoch": 66, "lr": 9.572256512540589e-05} +{"train_loss": 0.019658654928207397, "grad_norm": 0.04948519542813301, "global_step": 7029, "epoch": 66, "lr": 9.572136562070781e-05} +{"train_loss": 0.016461657360196114, "grad_norm": 0.051669951528310776, "global_step": 7030, "epoch": 66, "lr": 9.572016595536472e-05} +{"train_loss": 0.01706869527697563, "grad_norm": 0.04166765511035919, "global_step": 7031, "epoch": 66, "lr": 9.571896612938088e-05} +{"train_loss": 0.020094875246286392, "grad_norm": 0.15653467178344727, "global_step": 7032, "epoch": 66, "lr": 9.571776614276044e-05} +{"train_loss": 0.020210042595863342, "grad_norm": 0.18425434827804565, "global_step": 7033, "epoch": 66, "lr": 9.571656599550767e-05} +{"train_loss": 0.01886861026287079, "grad_norm": 0.08173316717147827, "global_step": 7034, "epoch": 66, "lr": 9.571536568762677e-05} +{"train_loss": 0.018289867788553238, "grad_norm": 0.046864524483680725, "global_step": 7035, "epoch": 66, "lr": 9.571416521912195e-05} +{"train_loss": 0.01613425835967064, "grad_norm": 0.15685886144638062, "global_step": 7036, "epoch": 66, "lr": 9.571296458999743e-05} +{"train_loss": 0.019536960870027542, "grad_norm": 0.06479791551828384, "global_step": 7037, "epoch": 66, "lr": 9.571176380025744e-05} +{"train_loss": 0.016598397865891457, "grad_norm": 0.07294189929962158, "global_step": 7038, "epoch": 66, "lr": 9.571056284990619e-05} +{"train_loss": 0.019538037478923798, "grad_norm": 0.06909173727035522, "global_step": 7039, "epoch": 66, "lr": 9.57093617389479e-05} +{"train_loss": 0.021253865212202072, "grad_norm": 0.06063295528292656, "global_step": 7040, "epoch": 66, "lr": 9.57081604673868e-05} +{"train_loss": 0.016592051833868027, "grad_norm": 0.12428714334964752, "global_step": 7041, "epoch": 66, "lr": 9.570695903522708e-05} +{"train_loss": 0.01882099360227585, "grad_norm": 0.12545795738697052, "global_step": 7042, "epoch": 66, "lr": 9.5705757442473e-05} +{"train_loss": 0.020801736041903496, "grad_norm": 0.12659479677677155, "global_step": 7043, "epoch": 66, "lr": 9.570455568912878e-05} +{"train_loss": 0.01728057488799095, "grad_norm": 0.0675310418009758, "global_step": 7044, "epoch": 66, "lr": 9.57033537751986e-05} +{"train_loss": 0.020963337272405624, "grad_norm": 0.21888795495033264, "global_step": 7045, "epoch": 66, "lr": 9.570215170068672e-05} +{"train_loss": 0.018914129585027695, "grad_norm": 0.05522245541214943, "global_step": 7046, "epoch": 66, "lr": 9.570094946559735e-05} +{"train_loss": 0.016453087329864502, "grad_norm": 0.14548905193805695, "global_step": 7047, "epoch": 66, "lr": 9.569974706993473e-05} +{"train_loss": 0.020151417702436447, "grad_norm": 0.057964567095041275, "global_step": 7048, "epoch": 66, "lr": 9.569854451370307e-05} +{"train_loss": 0.019709942862391472, "grad_norm": 0.06670696288347244, "global_step": 7049, "epoch": 66, "lr": 9.569734179690659e-05} +{"train_loss": 0.02027740702033043, "grad_norm": 0.11465102434158325, "global_step": 7050, "epoch": 66, "lr": 9.569613891954954e-05} +{"train_loss": 0.018400121480226517, "grad_norm": 0.1486525982618332, "global_step": 7051, "epoch": 66, "lr": 9.569493588163611e-05} +{"train_loss": 0.018266886472702026, "grad_norm": 0.10155686736106873, "global_step": 7052, "epoch": 66, "lr": 9.569373268317056e-05} +{"train_loss": 0.019068799912929535, "grad_norm": 0.06429428607225418, "global_step": 7053, "epoch": 66, "lr": 9.56925293241571e-05} +{"train_loss": 0.0180340763181448, "grad_norm": 0.1308472752571106, "global_step": 7054, "epoch": 66, "lr": 9.569132580459996e-05} +{"train_loss": 0.01847691461443901, "grad_norm": 0.07637854665517807, "global_step": 7055, "epoch": 66, "lr": 9.56901221245034e-05} +{"train_loss": 0.019083624705672264, "grad_norm": 0.08746682852506638, "global_step": 7056, "epoch": 66, "lr": 9.568891828387159e-05} +{"train_loss": 0.0190767589956522, "grad_norm": 0.0794968381524086, "global_step": 7057, "epoch": 66, "lr": 9.568771428270879e-05} +{"train_loss": 0.018941102549433708, "grad_norm": 0.12507538497447968, "global_step": 7058, "epoch": 66, "lr": 9.568651012101923e-05} +{"train_loss": 0.022234627977013588, "grad_norm": 0.10409197956323624, "global_step": 7059, "epoch": 66, "lr": 9.568530579880716e-05} +{"train_loss": 0.019090035930275917, "grad_norm": 0.054424602538347244, "global_step": 7060, "epoch": 66, "lr": 9.568410131607679e-05} +{"train_loss": 0.019588757306337357, "grad_norm": 0.09775755554437637, "global_step": 7061, "epoch": 66, "lr": 9.568289667283235e-05} +{"train_loss": 0.016122184693813324, "grad_norm": 0.06409654766321182, "global_step": 7062, "epoch": 66, "lr": 9.568169186907807e-05} +{"train_loss": 0.017665920779109, "grad_norm": 0.06581378728151321, "global_step": 7063, "epoch": 66, "lr": 9.568048690481821e-05} +{"train_loss": 0.021688144654035568, "grad_norm": 0.12078958749771118, "global_step": 7064, "epoch": 66, "lr": 9.567928178005697e-05} +{"train_loss": 0.02024712786078453, "grad_norm": 0.05564865469932556, "global_step": 7065, "epoch": 66, "lr": 9.56780764947986e-05} +{"train_loss": 0.018242856487631798, "grad_norm": 0.08237984031438828, "global_step": 7066, "epoch": 66, "lr": 9.567687104904734e-05} +{"train_loss": 0.01825391873717308, "grad_norm": 0.05492835491895676, "global_step": 7067, "epoch": 66, "lr": 9.567566544280742e-05} +{"train_loss": 0.019596856087446213, "grad_norm": 0.13393142819404602, "global_step": 7068, "epoch": 66, "lr": 9.567445967608308e-05} +{"train_loss": 0.021382201462984085, "grad_norm": 0.11269896477460861, "global_step": 7069, "epoch": 66, "lr": 9.567325374887855e-05} +{"train_loss": 0.0184970460832119, "grad_norm": 0.10229674726724625, "global_step": 7070, "epoch": 66, "lr": 9.567204766119806e-05} +{"train_loss": 0.019975315779447556, "grad_norm": 0.05099525302648544, "global_step": 7071, "epoch": 66, "lr": 9.567084141304586e-05} +{"train_loss": 0.020490147173404694, "grad_norm": 0.06678089499473572, "global_step": 7072, "epoch": 66, "lr": 9.566963500442618e-05} +{"train_loss": 0.021772416308522224, "grad_norm": 0.05987217649817467, "global_step": 7073, "epoch": 66, "lr": 9.566842843534327e-05} +{"train_loss": 0.0180564746260643, "grad_norm": 0.038496170192956924, "global_step": 7074, "epoch": 66, "lr": 9.566722170580137e-05} +{"train_loss": 0.017963843420147896, "grad_norm": 0.08577267080545425, "global_step": 7075, "epoch": 66, "lr": 9.56660148158047e-05} +{"train_loss": 0.01945190690457821, "grad_norm": 0.06581466645002365, "global_step": 7076, "epoch": 66, "lr": 9.566480776535752e-05} +{"train_loss": 0.018115531653165817, "grad_norm": 0.04604385420680046, "global_step": 7077, "epoch": 66, "lr": 9.566360055446408e-05} +{"train_loss": 0.018918385729193687, "grad_norm": 0.160613015294075, "global_step": 7078, "epoch": 66, "lr": 9.566239318312857e-05} +{"train_loss": 0.01825682632625103, "grad_norm": 0.0587691031396389, "global_step": 7079, "epoch": 66, "lr": 9.56611856513553e-05} +{"train_loss": 0.016450557857751846, "grad_norm": 0.0522729717195034, "global_step": 7080, "epoch": 66, "lr": 9.565997795914848e-05} +{"train_loss": 0.01699219085276127, "grad_norm": 0.0381334125995636, "global_step": 7081, "epoch": 66, "lr": 9.565877010651234e-05} +{"train_loss": 0.020321514457464218, "grad_norm": 0.045195166021585464, "global_step": 7082, "epoch": 66, "lr": 9.565756209345114e-05} +{"train_loss": 0.01642463542521, "grad_norm": 0.055477745831012726, "global_step": 7083, "epoch": 66, "lr": 9.565635391996912e-05} +{"train_loss": 0.01767406053841114, "grad_norm": 0.050248585641384125, "global_step": 7084, "epoch": 66, "lr": 9.565514558607053e-05} +{"train_loss": 0.017865773290395737, "grad_norm": 0.0745180994272232, "global_step": 7085, "epoch": 66, "lr": 9.565393709175963e-05} +{"train_loss": 0.01700810343027115, "grad_norm": 0.08608900755643845, "global_step": 7086, "epoch": 66, "lr": 9.565272843704062e-05} +{"train_loss": 0.020352743566036224, "grad_norm": 0.1264125555753708, "global_step": 7087, "epoch": 66, "lr": 9.565151962191781e-05} +{"train_loss": 0.017288990318775177, "grad_norm": 0.15156079828739166, "global_step": 7088, "epoch": 66, "lr": 9.565031064639537e-05} +{"train_loss": 0.019834434613585472, "grad_norm": 0.07383811473846436, "global_step": 7089, "epoch": 66, "lr": 9.564910151047764e-05} +{"train_loss": 0.015641042962670326, "grad_norm": 0.059977684170007706, "global_step": 7090, "epoch": 66, "lr": 9.564789221416879e-05} +{"train_loss": 0.019046153873205185, "grad_norm": 0.053853631019592285, "global_step": 7091, "epoch": 66, "lr": 9.56466827574731e-05} +{"train_loss": 0.019813664257526398, "grad_norm": 0.04826226085424423, "global_step": 7092, "epoch": 66, "lr": 9.564547314039482e-05} +{"train_loss": 0.017777420580387115, "grad_norm": 0.13345319032669067, "global_step": 7093, "epoch": 66, "lr": 9.564426336293821e-05} +{"train_loss": 0.023047950118780136, "grad_norm": 0.10936484485864639, "global_step": 7094, "epoch": 66, "lr": 9.564305342510751e-05} +{"train_loss": 0.01947259157896042, "grad_norm": 0.061568062752485275, "global_step": 7095, "epoch": 66, "lr": 9.564184332690695e-05} +{"train_loss": 0.019286135211586952, "grad_norm": 0.10509146004915237, "global_step": 7096, "epoch": 66, "lr": 9.564063306834082e-05} +{"train_loss": 0.022830266505479813, "grad_norm": 0.04871567338705063, "global_step": 7097, "epoch": 66, "lr": 9.563942264941335e-05} +{"train_loss": 0.017639808356761932, "grad_norm": 0.0817992240190506, "global_step": 7098, "epoch": 66, "lr": 9.563821207012877e-05} +{"train_loss": 0.017815526574850082, "grad_norm": 0.045889973640441895, "global_step": 7099, "epoch": 66, "lr": 9.563700133049139e-05} +{"train_loss": 0.01885472796857357, "grad_norm": 0.06256387382745743, "global_step": 7100, "epoch": 66, "lr": 9.563579043050543e-05} +{"train_loss": 0.01857926235270669, "grad_norm": 0.08811797201633453, "global_step": 7101, "epoch": 66, "lr": 9.563457937017515e-05, "val_loss": 0.019376488402485847} +{"train_loss": 0.015879789367318153, "grad_norm": 0.06435494869947433, "global_step": 7102, "epoch": 67, "lr": 9.563336814950479e-05} +{"train_loss": 0.022622209042310715, "grad_norm": 0.052786458283662796, "global_step": 7103, "epoch": 67, "lr": 9.563215676849864e-05} +{"train_loss": 0.019690800458192825, "grad_norm": 0.06304869055747986, "global_step": 7104, "epoch": 67, "lr": 9.563094522716092e-05} +{"train_loss": 0.015629850327968597, "grad_norm": 0.09231331944465637, "global_step": 7105, "epoch": 67, "lr": 9.562973352549591e-05} +{"train_loss": 0.016770128160715103, "grad_norm": 0.055314626544713974, "global_step": 7106, "epoch": 67, "lr": 9.562852166350787e-05} +{"train_loss": 0.01863027736544609, "grad_norm": 0.043641090393066406, "global_step": 7107, "epoch": 67, "lr": 9.562730964120102e-05} +{"train_loss": 0.01718101277947426, "grad_norm": 0.1271647959947586, "global_step": 7108, "epoch": 67, "lr": 9.562609745857967e-05} +{"train_loss": 0.02391618862748146, "grad_norm": 0.06911271065473557, "global_step": 7109, "epoch": 67, "lr": 9.562488511564804e-05} +{"train_loss": 0.020834458991885185, "grad_norm": 0.16468563675880432, "global_step": 7110, "epoch": 67, "lr": 9.562367261241042e-05} +{"train_loss": 0.01905023865401745, "grad_norm": 0.04619238153100014, "global_step": 7111, "epoch": 67, "lr": 9.562245994887105e-05} +{"train_loss": 0.019747063517570496, "grad_norm": 0.08008377999067307, "global_step": 7112, "epoch": 67, "lr": 9.562124712503419e-05} +{"train_loss": 0.018780579790472984, "grad_norm": 0.08992855995893478, "global_step": 7113, "epoch": 67, "lr": 9.562003414090413e-05} +{"train_loss": 0.014778378419578075, "grad_norm": 0.05179988592863083, "global_step": 7114, "epoch": 67, "lr": 9.561882099648509e-05} +{"train_loss": 0.019399229437112808, "grad_norm": 0.07346334308385849, "global_step": 7115, "epoch": 67, "lr": 9.561760769178137e-05} +{"train_loss": 0.019503707066178322, "grad_norm": 0.05407533049583435, "global_step": 7116, "epoch": 67, "lr": 9.561639422679718e-05} +{"train_loss": 0.01749049872159958, "grad_norm": 0.12780915200710297, "global_step": 7117, "epoch": 67, "lr": 9.561518060153684e-05} +{"train_loss": 0.017341256141662598, "grad_norm": 0.05960690602660179, "global_step": 7118, "epoch": 67, "lr": 9.561396681600461e-05} +{"train_loss": 0.02006310224533081, "grad_norm": 0.1351460963487625, "global_step": 7119, "epoch": 67, "lr": 9.561275287020472e-05} +{"train_loss": 0.018892647698521614, "grad_norm": 0.09017496556043625, "global_step": 7120, "epoch": 67, "lr": 9.561153876414144e-05} +{"train_loss": 0.01699768379330635, "grad_norm": 0.09842678159475327, "global_step": 7121, "epoch": 67, "lr": 9.561032449781907e-05} +{"train_loss": 0.016236014664173126, "grad_norm": 0.084182970225811, "global_step": 7122, "epoch": 67, "lr": 9.560911007124185e-05} +{"train_loss": 0.021672695875167847, "grad_norm": 0.06465456634759903, "global_step": 7123, "epoch": 67, "lr": 9.560789548441405e-05} +{"train_loss": 0.020717568695545197, "grad_norm": 0.07292324304580688, "global_step": 7124, "epoch": 67, "lr": 9.560668073733994e-05} +{"train_loss": 0.019297394901514053, "grad_norm": 0.06820739060640335, "global_step": 7125, "epoch": 67, "lr": 9.560546583002378e-05} +{"train_loss": 0.01846836507320404, "grad_norm": 0.06356671452522278, "global_step": 7126, "epoch": 67, "lr": 9.560425076246986e-05} +{"train_loss": 0.020106932148337364, "grad_norm": 0.08392729610204697, "global_step": 7127, "epoch": 67, "lr": 9.560303553468241e-05} +{"train_loss": 0.01646294817328453, "grad_norm": 0.045991867780685425, "global_step": 7128, "epoch": 67, "lr": 9.560182014666577e-05} +{"train_loss": 0.016698122024536133, "grad_norm": 0.036541953682899475, "global_step": 7129, "epoch": 67, "lr": 9.560060459842412e-05} +{"train_loss": 0.017213264480233192, "grad_norm": 0.04411663860082626, "global_step": 7130, "epoch": 67, "lr": 9.559938888996181e-05} +{"train_loss": 0.018951401114463806, "grad_norm": 0.07304495573043823, "global_step": 7131, "epoch": 67, "lr": 9.559817302128307e-05} +{"train_loss": 0.018431751057505608, "grad_norm": 0.04703960195183754, "global_step": 7132, "epoch": 67, "lr": 9.559695699239216e-05} +{"train_loss": 0.020132210105657578, "grad_norm": 0.10891776531934738, "global_step": 7133, "epoch": 67, "lr": 9.559574080329339e-05} +{"train_loss": 0.017218220978975296, "grad_norm": 0.08526542037725449, "global_step": 7134, "epoch": 67, "lr": 9.559452445399102e-05} +{"train_loss": 0.01865381747484207, "grad_norm": 0.10797155648469925, "global_step": 7135, "epoch": 67, "lr": 9.55933079444893e-05} +{"train_loss": 0.017539620399475098, "grad_norm": 0.09034592658281326, "global_step": 7136, "epoch": 67, "lr": 9.559209127479254e-05} +{"train_loss": 0.01822495087981224, "grad_norm": 0.10257365554571152, "global_step": 7137, "epoch": 67, "lr": 9.559087444490499e-05} +{"train_loss": 0.020250894129276276, "grad_norm": 0.12105927616357803, "global_step": 7138, "epoch": 67, "lr": 9.558965745483096e-05} +{"train_loss": 0.01822344958782196, "grad_norm": 0.06335482001304626, "global_step": 7139, "epoch": 67, "lr": 9.558844030457467e-05} +{"train_loss": 0.01899876445531845, "grad_norm": 0.043612152338027954, "global_step": 7140, "epoch": 67, "lr": 9.558722299414044e-05} +{"train_loss": 0.015833750367164612, "grad_norm": 0.09884312003850937, "global_step": 7141, "epoch": 67, "lr": 9.558600552353252e-05} +{"train_loss": 0.017242152243852615, "grad_norm": 0.08518005162477493, "global_step": 7142, "epoch": 67, "lr": 9.558478789275523e-05} +{"train_loss": 0.01878521218895912, "grad_norm": 0.11860845237970352, "global_step": 7143, "epoch": 67, "lr": 9.558357010181279e-05} +{"train_loss": 0.02019406668841839, "grad_norm": 0.08966955542564392, "global_step": 7144, "epoch": 67, "lr": 9.558235215070952e-05} +{"train_loss": 0.018383871763944626, "grad_norm": 0.07663984596729279, "global_step": 7145, "epoch": 67, "lr": 9.558113403944968e-05} +{"train_loss": 0.017238296568393707, "grad_norm": 0.07854151725769043, "global_step": 7146, "epoch": 67, "lr": 9.557991576803757e-05} +{"train_loss": 0.018605882301926613, "grad_norm": 0.12485011667013168, "global_step": 7147, "epoch": 67, "lr": 9.557869733647746e-05} +{"train_loss": 0.021505843847990036, "grad_norm": 0.1574583500623703, "global_step": 7148, "epoch": 67, "lr": 9.557747874477362e-05} +{"train_loss": 0.01973271183669567, "grad_norm": 0.06824326515197754, "global_step": 7149, "epoch": 67, "lr": 9.557625999293034e-05} +{"train_loss": 0.022067468613386154, "grad_norm": 0.0997060164809227, "global_step": 7150, "epoch": 67, "lr": 9.55750410809519e-05} +{"train_loss": 0.016524363309144974, "grad_norm": 0.09715339541435242, "global_step": 7151, "epoch": 67, "lr": 9.55738220088426e-05} +{"train_loss": 0.018140822649002075, "grad_norm": 0.12646764516830444, "global_step": 7152, "epoch": 67, "lr": 9.55726027766067e-05} +{"train_loss": 0.019281776621937752, "grad_norm": 0.06420403718948364, "global_step": 7153, "epoch": 67, "lr": 9.55713833842485e-05} +{"train_loss": 0.019243549555540085, "grad_norm": 0.14823786914348602, "global_step": 7154, "epoch": 67, "lr": 9.557016383177227e-05} +{"train_loss": 0.018288183957338333, "grad_norm": 0.04026167467236519, "global_step": 7155, "epoch": 67, "lr": 9.55689441191823e-05} +{"train_loss": 0.018177758902311325, "grad_norm": 0.13651810586452484, "global_step": 7156, "epoch": 67, "lr": 9.556772424648288e-05} +{"train_loss": 0.01935391128063202, "grad_norm": 0.12588319182395935, "global_step": 7157, "epoch": 67, "lr": 9.55665042136783e-05} +{"train_loss": 0.01710617169737816, "grad_norm": 0.06142295524477959, "global_step": 7158, "epoch": 67, "lr": 9.556528402077283e-05} +{"train_loss": 0.01695127785205841, "grad_norm": 0.07466486096382141, "global_step": 7159, "epoch": 67, "lr": 9.556406366777077e-05} +{"train_loss": 0.018032889813184738, "grad_norm": 0.06102290004491806, "global_step": 7160, "epoch": 67, "lr": 9.556284315467641e-05} +{"train_loss": 0.019179653376340866, "grad_norm": 0.10864543914794922, "global_step": 7161, "epoch": 67, "lr": 9.556162248149405e-05} +{"train_loss": 0.01927526667714119, "grad_norm": 0.06098340079188347, "global_step": 7162, "epoch": 67, "lr": 9.556040164822794e-05} +{"train_loss": 0.02058691717684269, "grad_norm": 0.055533718317747116, "global_step": 7163, "epoch": 67, "lr": 9.555918065488238e-05} +{"train_loss": 0.01936221681535244, "grad_norm": 0.05371444299817085, "global_step": 7164, "epoch": 67, "lr": 9.55579595014617e-05} +{"train_loss": 0.015449056401848793, "grad_norm": 0.0590963289141655, "global_step": 7165, "epoch": 67, "lr": 9.555673818797015e-05} +{"train_loss": 0.01591642014682293, "grad_norm": 0.05213506892323494, "global_step": 7166, "epoch": 67, "lr": 9.555551671441204e-05} +{"train_loss": 0.020069681107997894, "grad_norm": 0.08866836130619049, "global_step": 7167, "epoch": 67, "lr": 9.555429508079166e-05} +{"train_loss": 0.01965043693780899, "grad_norm": 0.042902447283267975, "global_step": 7168, "epoch": 67, "lr": 9.555307328711329e-05} +{"train_loss": 0.019103076308965683, "grad_norm": 0.06098547950387001, "global_step": 7169, "epoch": 67, "lr": 9.555185133338123e-05} +{"train_loss": 0.018675711005926132, "grad_norm": 0.07713472098112106, "global_step": 7170, "epoch": 67, "lr": 9.555062921959978e-05} +{"train_loss": 0.018192004412412643, "grad_norm": 0.059911150485277176, "global_step": 7171, "epoch": 67, "lr": 9.554940694577322e-05} +{"train_loss": 0.017398308962583542, "grad_norm": 0.03860687091946602, "global_step": 7172, "epoch": 67, "lr": 9.554818451190586e-05} +{"train_loss": 0.015270071104168892, "grad_norm": 0.06126764416694641, "global_step": 7173, "epoch": 67, "lr": 9.554696191800198e-05} +{"train_loss": 0.017650151625275612, "grad_norm": 0.15804722905158997, "global_step": 7174, "epoch": 67, "lr": 9.55457391640659e-05} +{"train_loss": 0.018454764038324356, "grad_norm": 0.07065382599830627, "global_step": 7175, "epoch": 67, "lr": 9.554451625010188e-05} +{"train_loss": 0.01843445934355259, "grad_norm": 0.09380657225847244, "global_step": 7176, "epoch": 67, "lr": 9.554329317611426e-05} +{"train_loss": 0.015466606244444847, "grad_norm": 0.04454898089170456, "global_step": 7177, "epoch": 67, "lr": 9.55420699421073e-05} +{"train_loss": 0.0158260278403759, "grad_norm": 0.04550483077764511, "global_step": 7178, "epoch": 67, "lr": 9.55408465480853e-05} +{"train_loss": 0.017974672839045525, "grad_norm": 0.11709202080965042, "global_step": 7179, "epoch": 67, "lr": 9.55396229940526e-05} +{"train_loss": 0.019184954464435577, "grad_norm": 0.07101975381374359, "global_step": 7180, "epoch": 67, "lr": 9.553839928001345e-05} +{"train_loss": 0.0199583750218153, "grad_norm": 0.05419979616999626, "global_step": 7181, "epoch": 67, "lr": 9.553717540597215e-05} +{"train_loss": 0.01813993975520134, "grad_norm": 0.052476320415735245, "global_step": 7182, "epoch": 67, "lr": 9.553595137193306e-05} +{"train_loss": 0.018284283578395844, "grad_norm": 0.08919751644134521, "global_step": 7183, "epoch": 67, "lr": 9.553472717790042e-05} +{"train_loss": 0.0195069070905447, "grad_norm": 0.0545249804854393, "global_step": 7184, "epoch": 67, "lr": 9.553350282387854e-05} +{"train_loss": 0.018751375377178192, "grad_norm": 0.05200541764497757, "global_step": 7185, "epoch": 67, "lr": 9.553227830987175e-05} +{"train_loss": 0.019810296595096588, "grad_norm": 0.057798054069280624, "global_step": 7186, "epoch": 67, "lr": 9.553105363588433e-05} +{"train_loss": 0.018584080040454865, "grad_norm": 0.06069422885775566, "global_step": 7187, "epoch": 67, "lr": 9.552982880192058e-05} +{"train_loss": 0.019718322902917862, "grad_norm": 0.1055314689874649, "global_step": 7188, "epoch": 67, "lr": 9.552860380798482e-05} +{"train_loss": 0.018245749175548553, "grad_norm": 0.061212439090013504, "global_step": 7189, "epoch": 67, "lr": 9.552737865408135e-05} +{"train_loss": 0.020349707454442978, "grad_norm": 0.13440988957881927, "global_step": 7190, "epoch": 67, "lr": 9.552615334021447e-05} +{"train_loss": 0.01967044174671173, "grad_norm": 0.058719005435705185, "global_step": 7191, "epoch": 67, "lr": 9.552492786638849e-05} +{"train_loss": 0.020460598170757294, "grad_norm": 0.06844277679920197, "global_step": 7192, "epoch": 67, "lr": 9.552370223260769e-05} +{"train_loss": 0.017681708559393883, "grad_norm": 0.08014918863773346, "global_step": 7193, "epoch": 67, "lr": 9.552247643887641e-05} +{"train_loss": 0.01968671940267086, "grad_norm": 0.054457876831293106, "global_step": 7194, "epoch": 67, "lr": 9.552125048519897e-05} +{"train_loss": 0.017613699659705162, "grad_norm": 0.052112091332674026, "global_step": 7195, "epoch": 67, "lr": 9.552002437157962e-05} +{"train_loss": 0.017557896673679352, "grad_norm": 0.07488185912370682, "global_step": 7196, "epoch": 67, "lr": 9.55187980980227e-05} +{"train_loss": 0.01951967366039753, "grad_norm": 0.06564809381961823, "global_step": 7197, "epoch": 67, "lr": 9.551757166453255e-05} +{"train_loss": 0.018283888697624207, "grad_norm": 0.0467582568526268, "global_step": 7198, "epoch": 67, "lr": 9.551634507111342e-05} +{"train_loss": 0.017387814819812775, "grad_norm": 0.05246642231941223, "global_step": 7199, "epoch": 67, "lr": 9.551511831776965e-05} +{"train_loss": 0.016472820192575455, "grad_norm": 0.07145871967077255, "global_step": 7200, "epoch": 67, "lr": 9.551389140450557e-05} +{"train_loss": 0.01750325784087181, "grad_norm": 0.053337208926677704, "global_step": 7201, "epoch": 67, "lr": 9.551266433132544e-05} +{"train_loss": 0.019236141815781593, "grad_norm": 0.06439977139234543, "global_step": 7202, "epoch": 67, "lr": 9.551143709823363e-05} +{"train_loss": 0.017765773460268974, "grad_norm": 0.06192966550588608, "global_step": 7203, "epoch": 67, "lr": 9.55102097052344e-05} +{"train_loss": 0.019246554002165794, "grad_norm": 0.06499865651130676, "global_step": 7204, "epoch": 67, "lr": 9.550898215233209e-05} +{"train_loss": 0.018661409616470337, "grad_norm": 0.10315530747175217, "global_step": 7205, "epoch": 67, "lr": 9.550775443953102e-05} +{"train_loss": 0.018854212015867233, "grad_norm": 0.06989095360040665, "global_step": 7206, "epoch": 67, "lr": 9.550652656683548e-05} +{"train_loss": 0.01852291450783048, "grad_norm": 0.15432989597320557, "global_step": 7207, "epoch": 67, "lr": 9.550529853424979e-05, "val_loss": 0.018866028636693954} +{"train_loss": 0.017023909837007523, "grad_norm": 0.03913518413901329, "global_step": 7208, "epoch": 68, "lr": 9.550407034177829e-05} +{"train_loss": 0.018281344324350357, "grad_norm": 0.057435035705566406, "global_step": 7209, "epoch": 68, "lr": 9.550284198942525e-05} +{"train_loss": 0.017321698367595673, "grad_norm": 0.09311985969543457, "global_step": 7210, "epoch": 68, "lr": 9.550161347719503e-05} +{"train_loss": 0.01798943243920803, "grad_norm": 0.051547128707170486, "global_step": 7211, "epoch": 68, "lr": 9.550038480509192e-05} +{"train_loss": 0.017522161826491356, "grad_norm": 0.06477772444486618, "global_step": 7212, "epoch": 68, "lr": 9.549915597312024e-05} +{"train_loss": 0.017624933272600174, "grad_norm": 0.07301714271306992, "global_step": 7213, "epoch": 68, "lr": 9.549792698128432e-05} +{"train_loss": 0.022560661658644676, "grad_norm": 0.06058622896671295, "global_step": 7214, "epoch": 68, "lr": 9.549669782958849e-05} +{"train_loss": 0.020080827176570892, "grad_norm": 0.0724833756685257, "global_step": 7215, "epoch": 68, "lr": 9.549546851803704e-05} +{"train_loss": 0.019334785640239716, "grad_norm": 0.04026621952652931, "global_step": 7216, "epoch": 68, "lr": 9.549423904663428e-05} +{"train_loss": 0.01741223782300949, "grad_norm": 0.048370443284511566, "global_step": 7217, "epoch": 68, "lr": 9.549300941538457e-05} +{"train_loss": 0.02189207263290882, "grad_norm": 0.047642938792705536, "global_step": 7218, "epoch": 68, "lr": 9.549177962429219e-05} +{"train_loss": 0.016662370413541794, "grad_norm": 0.06462907046079636, "global_step": 7219, "epoch": 68, "lr": 9.549054967336151e-05} +{"train_loss": 0.017643285915255547, "grad_norm": 0.05286848172545433, "global_step": 7220, "epoch": 68, "lr": 9.54893195625968e-05} +{"train_loss": 0.019780278205871582, "grad_norm": 0.06381610780954361, "global_step": 7221, "epoch": 68, "lr": 9.548808929200239e-05} +{"train_loss": 0.02203362062573433, "grad_norm": 0.11643335968255997, "global_step": 7222, "epoch": 68, "lr": 9.548685886158264e-05} +{"train_loss": 0.019836846739053726, "grad_norm": 0.05852390080690384, "global_step": 7223, "epoch": 68, "lr": 9.548562827134184e-05} +{"train_loss": 0.01923687383532524, "grad_norm": 0.0639212504029274, "global_step": 7224, "epoch": 68, "lr": 9.548439752128433e-05} +{"train_loss": 0.01878274232149124, "grad_norm": 0.14698442816734314, "global_step": 7225, "epoch": 68, "lr": 9.548316661141444e-05} +{"train_loss": 0.01688830554485321, "grad_norm": 0.03983528912067413, "global_step": 7226, "epoch": 68, "lr": 9.548193554173645e-05} +{"train_loss": 0.016348976641893387, "grad_norm": 0.045996420085430145, "global_step": 7227, "epoch": 68, "lr": 9.548070431225473e-05} +{"train_loss": 0.018299201503396034, "grad_norm": 0.11933404952287674, "global_step": 7228, "epoch": 68, "lr": 9.547947292297361e-05} +{"train_loss": 0.020405089482665062, "grad_norm": 0.042497821152210236, "global_step": 7229, "epoch": 68, "lr": 9.547824137389739e-05} +{"train_loss": 0.017860066145658493, "grad_norm": 0.06221736595034599, "global_step": 7230, "epoch": 68, "lr": 9.547700966503041e-05} +{"train_loss": 0.02067875862121582, "grad_norm": 0.06573429703712463, "global_step": 7231, "epoch": 68, "lr": 9.547577779637699e-05} +{"train_loss": 0.020235002040863037, "grad_norm": 0.04673650860786438, "global_step": 7232, "epoch": 68, "lr": 9.547454576794146e-05} +{"train_loss": 0.019446246325969696, "grad_norm": 0.06689827144145966, "global_step": 7233, "epoch": 68, "lr": 9.547331357972815e-05} +{"train_loss": 0.019387276843190193, "grad_norm": 0.044303394854068756, "global_step": 7234, "epoch": 68, "lr": 9.547208123174141e-05} +{"train_loss": 0.019169125705957413, "grad_norm": 0.04825904965400696, "global_step": 7235, "epoch": 68, "lr": 9.547084872398553e-05} +{"train_loss": 0.013533875346183777, "grad_norm": 0.05569633096456528, "global_step": 7236, "epoch": 68, "lr": 9.546961605646489e-05} +{"train_loss": 0.015497727319598198, "grad_norm": 0.12219713628292084, "global_step": 7237, "epoch": 68, "lr": 9.546838322918378e-05} +{"train_loss": 0.018330585211515427, "grad_norm": 0.05920546501874924, "global_step": 7238, "epoch": 68, "lr": 9.546715024214654e-05} +{"train_loss": 0.015496303327381611, "grad_norm": 0.06614094227552414, "global_step": 7239, "epoch": 68, "lr": 9.546591709535751e-05} +{"train_loss": 0.016435129567980766, "grad_norm": 0.07271557301282883, "global_step": 7240, "epoch": 68, "lr": 9.546468378882102e-05} +{"train_loss": 0.022161714732646942, "grad_norm": 0.055991195142269135, "global_step": 7241, "epoch": 68, "lr": 9.54634503225414e-05} +{"train_loss": 0.020510055124759674, "grad_norm": 0.11522608995437622, "global_step": 7242, "epoch": 68, "lr": 9.5462216696523e-05} +{"train_loss": 0.02006646804511547, "grad_norm": 0.06000322476029396, "global_step": 7243, "epoch": 68, "lr": 9.546098291077012e-05} +{"train_loss": 0.018081914633512497, "grad_norm": 0.03857661411166191, "global_step": 7244, "epoch": 68, "lr": 9.545974896528714e-05} +{"train_loss": 0.01917608641088009, "grad_norm": 0.06674719601869583, "global_step": 7245, "epoch": 68, "lr": 9.545851486007835e-05} +{"train_loss": 0.019000018015503883, "grad_norm": 0.0507848858833313, "global_step": 7246, "epoch": 68, "lr": 9.545728059514814e-05} +{"train_loss": 0.017633594572544098, "grad_norm": 0.04452870041131973, "global_step": 7247, "epoch": 68, "lr": 9.545604617050077e-05} +{"train_loss": 0.019408557564020157, "grad_norm": 0.08479341864585876, "global_step": 7248, "epoch": 68, "lr": 9.545481158614066e-05} +{"train_loss": 0.01810421608388424, "grad_norm": 0.048716455698013306, "global_step": 7249, "epoch": 68, "lr": 9.54535768420721e-05} +{"train_loss": 0.017999716103076935, "grad_norm": 0.08075794577598572, "global_step": 7250, "epoch": 68, "lr": 9.545234193829943e-05} +{"train_loss": 0.01939784735441208, "grad_norm": 0.07616722583770752, "global_step": 7251, "epoch": 68, "lr": 9.545110687482699e-05} +{"train_loss": 0.017936721444129944, "grad_norm": 0.066545270383358, "global_step": 7252, "epoch": 68, "lr": 9.544987165165915e-05} +{"train_loss": 0.018143579363822937, "grad_norm": 0.05580534413456917, "global_step": 7253, "epoch": 68, "lr": 9.54486362688002e-05} +{"train_loss": 0.014437062665820122, "grad_norm": 0.13857750594615936, "global_step": 7254, "epoch": 68, "lr": 9.544740072625453e-05} +{"train_loss": 0.020322244614362717, "grad_norm": 0.13241755962371826, "global_step": 7255, "epoch": 68, "lr": 9.544616502402646e-05} +{"train_loss": 0.017639044672250748, "grad_norm": 0.05166683718562126, "global_step": 7256, "epoch": 68, "lr": 9.544492916212031e-05} +{"train_loss": 0.016079291701316833, "grad_norm": 0.03466375917196274, "global_step": 7257, "epoch": 68, "lr": 9.544369314054046e-05} +{"train_loss": 0.021679237484931946, "grad_norm": 0.10117815434932709, "global_step": 7258, "epoch": 68, "lr": 9.544245695929122e-05} +{"train_loss": 0.019466686993837357, "grad_norm": 0.044824860990047455, "global_step": 7259, "epoch": 68, "lr": 9.544122061837697e-05} +{"train_loss": 0.019235864281654358, "grad_norm": 0.051529739052057266, "global_step": 7260, "epoch": 68, "lr": 9.543998411780201e-05} +{"train_loss": 0.01939740777015686, "grad_norm": 0.05746838077902794, "global_step": 7261, "epoch": 68, "lr": 9.543874745757071e-05} +{"train_loss": 0.01921648159623146, "grad_norm": 0.07522491365671158, "global_step": 7262, "epoch": 68, "lr": 9.543751063768742e-05} +{"train_loss": 0.020427042618393898, "grad_norm": 0.07732976227998734, "global_step": 7263, "epoch": 68, "lr": 9.543627365815648e-05} +{"train_loss": 0.01788080483675003, "grad_norm": 0.07625937461853027, "global_step": 7264, "epoch": 68, "lr": 9.543503651898223e-05} +{"train_loss": 0.020181018859148026, "grad_norm": 0.16548418998718262, "global_step": 7265, "epoch": 68, "lr": 9.543379922016901e-05} +{"train_loss": 0.017505697906017303, "grad_norm": 0.11599098145961761, "global_step": 7266, "epoch": 68, "lr": 9.54325617617212e-05} +{"train_loss": 0.01673164591193199, "grad_norm": 0.08222328126430511, "global_step": 7267, "epoch": 68, "lr": 9.543132414364312e-05} +{"train_loss": 0.017840174958109856, "grad_norm": 0.057773202657699585, "global_step": 7268, "epoch": 68, "lr": 9.543008636593913e-05} +{"train_loss": 0.016545603051781654, "grad_norm": 0.060234688222408295, "global_step": 7269, "epoch": 68, "lr": 9.542884842861355e-05} +{"train_loss": 0.016879014670848846, "grad_norm": 0.062417227774858475, "global_step": 7270, "epoch": 68, "lr": 9.542761033167078e-05} +{"train_loss": 0.020621681585907936, "grad_norm": 0.05752040818333626, "global_step": 7271, "epoch": 68, "lr": 9.542637207511511e-05} +{"train_loss": 0.018876317888498306, "grad_norm": 0.12172973901033401, "global_step": 7272, "epoch": 68, "lr": 9.542513365895096e-05} +{"train_loss": 0.021349523216485977, "grad_norm": 0.15061555802822113, "global_step": 7273, "epoch": 68, "lr": 9.542389508318262e-05} +{"train_loss": 0.02155071683228016, "grad_norm": 0.05944392830133438, "global_step": 7274, "epoch": 68, "lr": 9.542265634781447e-05} +{"train_loss": 0.016077663749456406, "grad_norm": 0.042599719017744064, "global_step": 7275, "epoch": 68, "lr": 9.542141745285087e-05} +{"train_loss": 0.017794989049434662, "grad_norm": 0.07972527295351028, "global_step": 7276, "epoch": 68, "lr": 9.542017839829615e-05} +{"train_loss": 0.019605852663517, "grad_norm": 0.0563192255795002, "global_step": 7277, "epoch": 68, "lr": 9.54189391841547e-05} +{"train_loss": 0.020304199308156967, "grad_norm": 0.04957854002714157, "global_step": 7278, "epoch": 68, "lr": 9.541769981043082e-05} +{"train_loss": 0.01748213917016983, "grad_norm": 0.04001011326909065, "global_step": 7279, "epoch": 68, "lr": 9.541646027712892e-05} +{"train_loss": 0.017498798668384552, "grad_norm": 0.050701238214969635, "global_step": 7280, "epoch": 68, "lr": 9.541522058425332e-05} +{"train_loss": 0.01965523511171341, "grad_norm": 0.15287017822265625, "global_step": 7281, "epoch": 68, "lr": 9.541398073180839e-05} +{"train_loss": 0.016258234158158302, "grad_norm": 0.10249510407447815, "global_step": 7282, "epoch": 68, "lr": 9.541274071979846e-05} +{"train_loss": 0.021733291447162628, "grad_norm": 0.07277729362249374, "global_step": 7283, "epoch": 68, "lr": 9.541150054822793e-05} +{"train_loss": 0.015941280871629715, "grad_norm": 0.0470384918153286, "global_step": 7284, "epoch": 68, "lr": 9.541026021710113e-05} +{"train_loss": 0.016894249245524406, "grad_norm": 0.05563534051179886, "global_step": 7285, "epoch": 68, "lr": 9.540901972642242e-05} +{"train_loss": 0.016955869272351265, "grad_norm": 0.09108491986989975, "global_step": 7286, "epoch": 68, "lr": 9.540777907619616e-05} +{"train_loss": 0.017966197803616524, "grad_norm": 0.0522901676595211, "global_step": 7287, "epoch": 68, "lr": 9.540653826642671e-05} +{"train_loss": 0.017336580902338028, "grad_norm": 0.13027682900428772, "global_step": 7288, "epoch": 68, "lr": 9.540529729711844e-05} +{"train_loss": 0.018387414515018463, "grad_norm": 0.04536690190434456, "global_step": 7289, "epoch": 68, "lr": 9.540405616827572e-05} +{"train_loss": 0.016131892800331116, "grad_norm": 0.054067518562078476, "global_step": 7290, "epoch": 68, "lr": 9.540281487990286e-05} +{"train_loss": 0.01804836466908455, "grad_norm": 0.11125335097312927, "global_step": 7291, "epoch": 68, "lr": 9.540157343200426e-05} +{"train_loss": 0.015671556815505028, "grad_norm": 0.08401288837194443, "global_step": 7292, "epoch": 68, "lr": 9.54003318245843e-05} +{"train_loss": 0.018636098131537437, "grad_norm": 0.07365113496780396, "global_step": 7293, "epoch": 68, "lr": 9.539909005764729e-05} +{"train_loss": 0.01816558465361595, "grad_norm": 0.14497217535972595, "global_step": 7294, "epoch": 68, "lr": 9.539784813119763e-05} +{"train_loss": 0.0163415614515543, "grad_norm": 0.048628710210323334, "global_step": 7295, "epoch": 68, "lr": 9.539660604523967e-05} +{"train_loss": 0.019795678555965424, "grad_norm": 0.06422259658575058, "global_step": 7296, "epoch": 68, "lr": 9.53953637997778e-05} +{"train_loss": 0.01905050128698349, "grad_norm": 0.08315573632717133, "global_step": 7297, "epoch": 68, "lr": 9.539412139481633e-05} +{"train_loss": 0.018707670271396637, "grad_norm": 0.0375128909945488, "global_step": 7298, "epoch": 68, "lr": 9.539287883035969e-05} +{"train_loss": 0.016497449949383736, "grad_norm": 0.04800565913319588, "global_step": 7299, "epoch": 68, "lr": 9.53916361064122e-05} +{"train_loss": 0.016743509098887444, "grad_norm": 0.044427886605262756, "global_step": 7300, "epoch": 68, "lr": 9.539039322297825e-05} +{"train_loss": 0.02035464718937874, "grad_norm": 0.07624997943639755, "global_step": 7301, "epoch": 68, "lr": 9.538915018006218e-05} +{"train_loss": 0.016478877514600754, "grad_norm": 0.0792628824710846, "global_step": 7302, "epoch": 68, "lr": 9.53879069776684e-05} +{"train_loss": 0.017589470371603966, "grad_norm": 0.05803943797945976, "global_step": 7303, "epoch": 68, "lr": 9.538666361580122e-05} +{"train_loss": 0.019204797223210335, "grad_norm": 0.167131245136261, "global_step": 7304, "epoch": 68, "lr": 9.538542009446507e-05} +{"train_loss": 0.018728237599134445, "grad_norm": 0.10164891183376312, "global_step": 7305, "epoch": 68, "lr": 9.538417641366428e-05} +{"train_loss": 0.01603800803422928, "grad_norm": 0.06868330389261246, "global_step": 7306, "epoch": 68, "lr": 9.538293257340326e-05} +{"train_loss": 0.02092103473842144, "grad_norm": 0.07211372256278992, "global_step": 7307, "epoch": 68, "lr": 9.53816885736863e-05} +{"train_loss": 0.0217842198908329, "grad_norm": 0.18062502145767212, "global_step": 7308, "epoch": 68, "lr": 9.538044441451785e-05} +{"train_loss": 0.016376879066228867, "grad_norm": 0.06997210532426834, "global_step": 7309, "epoch": 68, "lr": 9.537920009590226e-05} +{"train_loss": 0.017285630106925964, "grad_norm": 0.11192533373832703, "global_step": 7310, "epoch": 68, "lr": 9.537795561784388e-05} +{"train_loss": 0.01952054724097252, "grad_norm": 0.05215418338775635, "global_step": 7311, "epoch": 68, "lr": 9.537671098034711e-05} +{"train_loss": 0.022078227251768112, "grad_norm": 0.05188335105776787, "global_step": 7312, "epoch": 68, "lr": 9.53754661834163e-05} +{"train_loss": 0.018473016388081718, "grad_norm": 0.07692909240722656, "global_step": 7313, "epoch": 68, "lr": 9.537422122705585e-05, "val_loss": 0.018729113042354584} +{"train_loss": 0.020225469022989273, "grad_norm": 0.06427549570798874, "global_step": 7314, "epoch": 69, "lr": 9.537297611127012e-05} +{"train_loss": 0.018171170726418495, "grad_norm": 0.05398006737232208, "global_step": 7315, "epoch": 69, "lr": 9.537173083606348e-05} +{"train_loss": 0.015596522018313408, "grad_norm": 0.09159756451845169, "global_step": 7316, "epoch": 69, "lr": 9.537048540144029e-05} +{"train_loss": 0.019035305827856064, "grad_norm": 0.0746852308511734, "global_step": 7317, "epoch": 69, "lr": 9.536923980740496e-05} +{"train_loss": 0.022007262334227562, "grad_norm": 0.20548321306705475, "global_step": 7318, "epoch": 69, "lr": 9.536799405396186e-05} +{"train_loss": 0.019296817481517792, "grad_norm": 0.0575045570731163, "global_step": 7319, "epoch": 69, "lr": 9.536674814111535e-05} +{"train_loss": 0.016155501827597618, "grad_norm": 0.08219262212514877, "global_step": 7320, "epoch": 69, "lr": 9.536550206886982e-05} +{"train_loss": 0.018268607556819916, "grad_norm": 0.06134994700551033, "global_step": 7321, "epoch": 69, "lr": 9.536425583722965e-05} +{"train_loss": 0.019173303619027138, "grad_norm": 0.11425936222076416, "global_step": 7322, "epoch": 69, "lr": 9.536300944619921e-05} +{"train_loss": 0.01810559071600437, "grad_norm": 0.06344027072191238, "global_step": 7323, "epoch": 69, "lr": 9.536176289578287e-05} +{"train_loss": 0.018669720739126205, "grad_norm": 0.057702675461769104, "global_step": 7324, "epoch": 69, "lr": 9.536051618598502e-05} +{"train_loss": 0.020851925015449524, "grad_norm": 0.08019109815359116, "global_step": 7325, "epoch": 69, "lr": 9.535926931681006e-05} +{"train_loss": 0.01886928454041481, "grad_norm": 0.08596490323543549, "global_step": 7326, "epoch": 69, "lr": 9.535802228826235e-05} +{"train_loss": 0.017670493572950363, "grad_norm": 0.07034996151924133, "global_step": 7327, "epoch": 69, "lr": 9.535677510034628e-05} +{"train_loss": 0.02040233090519905, "grad_norm": 0.07610966265201569, "global_step": 7328, "epoch": 69, "lr": 9.535552775306622e-05} +{"train_loss": 0.01913752406835556, "grad_norm": 0.07855945825576782, "global_step": 7329, "epoch": 69, "lr": 9.535428024642656e-05} +{"train_loss": 0.023153109475970268, "grad_norm": 0.12490083277225494, "global_step": 7330, "epoch": 69, "lr": 9.535303258043167e-05} +{"train_loss": 0.019320141524076462, "grad_norm": 0.10727518796920776, "global_step": 7331, "epoch": 69, "lr": 9.535178475508598e-05} +{"train_loss": 0.017534825950860977, "grad_norm": 0.0850415825843811, "global_step": 7332, "epoch": 69, "lr": 9.535053677039381e-05} +{"train_loss": 0.01742379181087017, "grad_norm": 0.1013268455862999, "global_step": 7333, "epoch": 69, "lr": 9.534928862635959e-05} +{"train_loss": 0.01710759848356247, "grad_norm": 0.05194173753261566, "global_step": 7334, "epoch": 69, "lr": 9.53480403229877e-05} +{"train_loss": 0.019155383110046387, "grad_norm": 0.06495825946331024, "global_step": 7335, "epoch": 69, "lr": 9.53467918602825e-05} +{"train_loss": 0.0204765684902668, "grad_norm": 0.09635837376117706, "global_step": 7336, "epoch": 69, "lr": 9.53455432382484e-05} +{"train_loss": 0.01818033680319786, "grad_norm": 0.061842888593673706, "global_step": 7337, "epoch": 69, "lr": 9.53442944568898e-05} +{"train_loss": 0.020575599744915962, "grad_norm": 0.0642966479063034, "global_step": 7338, "epoch": 69, "lr": 9.534304551621104e-05} +{"train_loss": 0.019286181777715683, "grad_norm": 0.17013585567474365, "global_step": 7339, "epoch": 69, "lr": 9.534179641621654e-05} +{"train_loss": 0.01986793801188469, "grad_norm": 0.07002557069063187, "global_step": 7340, "epoch": 69, "lr": 9.534054715691071e-05} +{"train_loss": 0.018937934190034866, "grad_norm": 0.06212368234992027, "global_step": 7341, "epoch": 69, "lr": 9.53392977382979e-05} +{"train_loss": 0.019327394664287567, "grad_norm": 0.066404789686203, "global_step": 7342, "epoch": 69, "lr": 9.533804816038252e-05} +{"train_loss": 0.02020428329706192, "grad_norm": 0.06154809892177582, "global_step": 7343, "epoch": 69, "lr": 9.533679842316895e-05} +{"train_loss": 0.0170506089925766, "grad_norm": 0.08576826006174088, "global_step": 7344, "epoch": 69, "lr": 9.533554852666157e-05} +{"train_loss": 0.016859183087944984, "grad_norm": 0.07009156048297882, "global_step": 7345, "epoch": 69, "lr": 9.533429847086482e-05} +{"train_loss": 0.017686424776911736, "grad_norm": 0.04614840820431709, "global_step": 7346, "epoch": 69, "lr": 9.533304825578305e-05} +{"train_loss": 0.015846461057662964, "grad_norm": 0.08090709894895554, "global_step": 7347, "epoch": 69, "lr": 9.533179788142065e-05} +{"train_loss": 0.01928977482020855, "grad_norm": 0.06907989084720612, "global_step": 7348, "epoch": 69, "lr": 9.533054734778204e-05} +{"train_loss": 0.016295738518238068, "grad_norm": 0.07748182862997055, "global_step": 7349, "epoch": 69, "lr": 9.53292966548716e-05} +{"train_loss": 0.019303344190120697, "grad_norm": 0.11828623712062836, "global_step": 7350, "epoch": 69, "lr": 9.532804580269373e-05} +{"train_loss": 0.01790868118405342, "grad_norm": 0.10089520364999771, "global_step": 7351, "epoch": 69, "lr": 9.532679479125282e-05} +{"train_loss": 0.016585567966103554, "grad_norm": 0.09009946137666702, "global_step": 7352, "epoch": 69, "lr": 9.532554362055326e-05} +{"train_loss": 0.018717240542173386, "grad_norm": 0.08441364765167236, "global_step": 7353, "epoch": 69, "lr": 9.532429229059945e-05} +{"train_loss": 0.01928420178592205, "grad_norm": 0.09538348019123077, "global_step": 7354, "epoch": 69, "lr": 9.532304080139579e-05} +{"train_loss": 0.018575316295027733, "grad_norm": 0.0775286927819252, "global_step": 7355, "epoch": 69, "lr": 9.532178915294666e-05} +{"train_loss": 0.01698688231408596, "grad_norm": 0.07830286771059036, "global_step": 7356, "epoch": 69, "lr": 9.53205373452565e-05} +{"train_loss": 0.017886096611618996, "grad_norm": 0.07816122472286224, "global_step": 7357, "epoch": 69, "lr": 9.531928537832966e-05} +{"train_loss": 0.018223926424980164, "grad_norm": 0.21573157608509064, "global_step": 7358, "epoch": 69, "lr": 9.531803325217056e-05} +{"train_loss": 0.01804361492395401, "grad_norm": 0.05534583330154419, "global_step": 7359, "epoch": 69, "lr": 9.53167809667836e-05} +{"train_loss": 0.01779698207974434, "grad_norm": 0.04102441668510437, "global_step": 7360, "epoch": 69, "lr": 9.531552852217317e-05} +{"train_loss": 0.01731458306312561, "grad_norm": 0.045088991522789, "global_step": 7361, "epoch": 69, "lr": 9.531427591834372e-05} +{"train_loss": 0.01729324832558632, "grad_norm": 0.08095457404851913, "global_step": 7362, "epoch": 69, "lr": 9.531302315529958e-05} +{"train_loss": 0.01822914555668831, "grad_norm": 0.05254335328936577, "global_step": 7363, "epoch": 69, "lr": 9.531177023304519e-05} +{"train_loss": 0.019972369074821472, "grad_norm": 0.09144295752048492, "global_step": 7364, "epoch": 69, "lr": 9.531051715158495e-05} +{"train_loss": 0.01874624565243721, "grad_norm": 0.05446499586105347, "global_step": 7365, "epoch": 69, "lr": 9.530926391092326e-05} +{"train_loss": 0.02124784328043461, "grad_norm": 0.11006812006235123, "global_step": 7366, "epoch": 69, "lr": 9.530801051106449e-05} +{"train_loss": 0.020427200943231583, "grad_norm": 0.0663275495171547, "global_step": 7367, "epoch": 69, "lr": 9.53067569520131e-05} +{"train_loss": 0.019699081778526306, "grad_norm": 0.04235352203249931, "global_step": 7368, "epoch": 69, "lr": 9.530550323377347e-05} +{"train_loss": 0.01855684444308281, "grad_norm": 0.0907769724726677, "global_step": 7369, "epoch": 69, "lr": 9.530424935635e-05} +{"train_loss": 0.016641834750771523, "grad_norm": 0.07870092242956161, "global_step": 7370, "epoch": 69, "lr": 9.53029953197471e-05} +{"train_loss": 0.017702855169773102, "grad_norm": 0.04692734405398369, "global_step": 7371, "epoch": 69, "lr": 9.530174112396918e-05} +{"train_loss": 0.018844276666641235, "grad_norm": 0.2251378297805786, "global_step": 7372, "epoch": 69, "lr": 9.530048676902064e-05} +{"train_loss": 0.01603037118911743, "grad_norm": 0.12442368268966675, "global_step": 7373, "epoch": 69, "lr": 9.52992322549059e-05} +{"train_loss": 0.016366377472877502, "grad_norm": 0.1010427176952362, "global_step": 7374, "epoch": 69, "lr": 9.529797758162935e-05} +{"train_loss": 0.017021525651216507, "grad_norm": 0.0721949115395546, "global_step": 7375, "epoch": 69, "lr": 9.52967227491954e-05} +{"train_loss": 0.01549856923520565, "grad_norm": 0.0793607085943222, "global_step": 7376, "epoch": 69, "lr": 9.529546775760847e-05} +{"train_loss": 0.01916646957397461, "grad_norm": 0.04318326711654663, "global_step": 7377, "epoch": 69, "lr": 9.529421260687297e-05} +{"train_loss": 0.01886267215013504, "grad_norm": 0.04109950363636017, "global_step": 7378, "epoch": 69, "lr": 9.52929572969933e-05} +{"train_loss": 0.018314670771360397, "grad_norm": 0.07660908997058868, "global_step": 7379, "epoch": 69, "lr": 9.529170182797386e-05} +{"train_loss": 0.017241813242435455, "grad_norm": 0.0462784469127655, "global_step": 7380, "epoch": 69, "lr": 9.529044619981911e-05} +{"train_loss": 0.016418475657701492, "grad_norm": 0.05980677530169487, "global_step": 7381, "epoch": 69, "lr": 9.52891904125334e-05} +{"train_loss": 0.018183009698987007, "grad_norm": 0.09243810176849365, "global_step": 7382, "epoch": 69, "lr": 9.52879344661212e-05} +{"train_loss": 0.01832379400730133, "grad_norm": 0.05348915606737137, "global_step": 7383, "epoch": 69, "lr": 9.528667836058686e-05} +{"train_loss": 0.01802312582731247, "grad_norm": 0.0843426063656807, "global_step": 7384, "epoch": 69, "lr": 9.528542209593484e-05} +{"train_loss": 0.016949614509940147, "grad_norm": 0.04721125215291977, "global_step": 7385, "epoch": 69, "lr": 9.528416567216954e-05} +{"train_loss": 0.01918240636587143, "grad_norm": 0.12445051223039627, "global_step": 7386, "epoch": 69, "lr": 9.528290908929538e-05} +{"train_loss": 0.02056531235575676, "grad_norm": 0.14932593703269958, "global_step": 7387, "epoch": 69, "lr": 9.528165234731676e-05} +{"train_loss": 0.02207682467997074, "grad_norm": 0.1310097575187683, "global_step": 7388, "epoch": 69, "lr": 9.528039544623813e-05} +{"train_loss": 0.019544970244169235, "grad_norm": 0.10613694041967392, "global_step": 7389, "epoch": 69, "lr": 9.527913838606386e-05} +{"train_loss": 0.019527871161699295, "grad_norm": 0.18744626641273499, "global_step": 7390, "epoch": 69, "lr": 9.527788116679838e-05} +{"train_loss": 0.018741153180599213, "grad_norm": 0.10162221640348434, "global_step": 7391, "epoch": 69, "lr": 9.527662378844615e-05} +{"train_loss": 0.02118537202477455, "grad_norm": 0.05251985043287277, "global_step": 7392, "epoch": 69, "lr": 9.527536625101153e-05} +{"train_loss": 0.017456889152526855, "grad_norm": 0.042509760707616806, "global_step": 7393, "epoch": 69, "lr": 9.527410855449896e-05} +{"train_loss": 0.01964050903916359, "grad_norm": 0.12500838935375214, "global_step": 7394, "epoch": 69, "lr": 9.527285069891288e-05} +{"train_loss": 0.017345599830150604, "grad_norm": 0.04266709089279175, "global_step": 7395, "epoch": 69, "lr": 9.527159268425769e-05} +{"train_loss": 0.018889423459768295, "grad_norm": 0.08684593439102173, "global_step": 7396, "epoch": 69, "lr": 9.52703345105378e-05} +{"train_loss": 0.014669058844447136, "grad_norm": 0.040775272995233536, "global_step": 7397, "epoch": 69, "lr": 9.526907617775765e-05} +{"train_loss": 0.02056189253926277, "grad_norm": 0.1441037952899933, "global_step": 7398, "epoch": 69, "lr": 9.526781768592164e-05} +{"train_loss": 0.018441004678606987, "grad_norm": 0.08383125066757202, "global_step": 7399, "epoch": 69, "lr": 9.526655903503423e-05} +{"train_loss": 0.01795385219156742, "grad_norm": 0.054761603474617004, "global_step": 7400, "epoch": 69, "lr": 9.526530022509979e-05} +{"train_loss": 0.016608018428087234, "grad_norm": 0.08181101083755493, "global_step": 7401, "epoch": 69, "lr": 9.526404125612279e-05} +{"train_loss": 0.01685420610010624, "grad_norm": 0.12535390257835388, "global_step": 7402, "epoch": 69, "lr": 9.526278212810762e-05} +{"train_loss": 0.01788068190217018, "grad_norm": 0.13023050129413605, "global_step": 7403, "epoch": 69, "lr": 9.526152284105872e-05} +{"train_loss": 0.015195515938103199, "grad_norm": 0.1218641921877861, "global_step": 7404, "epoch": 69, "lr": 9.526026339498052e-05} +{"train_loss": 0.019274502992630005, "grad_norm": 0.21567319333553314, "global_step": 7405, "epoch": 69, "lr": 9.525900378987743e-05} +{"train_loss": 0.02001137100160122, "grad_norm": 0.09385058283805847, "global_step": 7406, "epoch": 69, "lr": 9.525774402575389e-05} +{"train_loss": 0.01615520380437374, "grad_norm": 0.07727693766355515, "global_step": 7407, "epoch": 69, "lr": 9.525648410261432e-05} +{"train_loss": 0.018183792009949684, "grad_norm": 0.10221406817436218, "global_step": 7408, "epoch": 69, "lr": 9.525522402046315e-05} +{"train_loss": 0.021694397553801537, "grad_norm": 0.056097690016031265, "global_step": 7409, "epoch": 69, "lr": 9.525396377930478e-05} +{"train_loss": 0.01865810714662075, "grad_norm": 0.05316027253866196, "global_step": 7410, "epoch": 69, "lr": 9.525270337914368e-05} +{"train_loss": 0.018566301092505455, "grad_norm": 0.0771447941660881, "global_step": 7411, "epoch": 69, "lr": 9.525144281998425e-05} +{"train_loss": 0.020566968247294426, "grad_norm": 0.11325717717409134, "global_step": 7412, "epoch": 69, "lr": 9.525018210183094e-05} +{"train_loss": 0.02038400247693062, "grad_norm": 0.05604095011949539, "global_step": 7413, "epoch": 69, "lr": 9.524892122468819e-05} +{"train_loss": 0.015267450362443924, "grad_norm": 0.12715353071689606, "global_step": 7414, "epoch": 69, "lr": 9.524766018856037e-05} +{"train_loss": 0.01758355274796486, "grad_norm": 0.07495956867933273, "global_step": 7415, "epoch": 69, "lr": 9.524639899345197e-05} +{"train_loss": 0.01956142485141754, "grad_norm": 0.04248524084687233, "global_step": 7416, "epoch": 69, "lr": 9.524513763936739e-05} +{"train_loss": 0.0163477323949337, "grad_norm": 0.054375264793634415, "global_step": 7417, "epoch": 69, "lr": 9.524387612631108e-05} +{"train_loss": 0.017222322523593903, "grad_norm": 0.07413415610790253, "global_step": 7418, "epoch": 69, "lr": 9.524261445428747e-05} +{"train_loss": 0.01845747034750738, "grad_norm": 0.12884202599525452, "global_step": 7419, "epoch": 69, "lr": 9.524135262330098e-05, "val_loss": 0.018897423520684242} +{"train_loss": 0.018202947452664375, "grad_norm": 0.08873341232538223, "global_step": 7420, "epoch": 70, "lr": 9.524009063335606e-05} +{"train_loss": 0.01906108483672142, "grad_norm": 0.11140385270118713, "global_step": 7421, "epoch": 70, "lr": 9.523882848445713e-05} +{"train_loss": 0.019569266587495804, "grad_norm": 0.07719770073890686, "global_step": 7422, "epoch": 70, "lr": 9.523756617660863e-05} +{"train_loss": 0.018825124949216843, "grad_norm": 0.08368872851133347, "global_step": 7423, "epoch": 70, "lr": 9.5236303709815e-05} +{"train_loss": 0.017354803159832954, "grad_norm": 0.04737456515431404, "global_step": 7424, "epoch": 70, "lr": 9.523504108408067e-05} +{"train_loss": 0.018621036782860756, "grad_norm": 0.07039303332567215, "global_step": 7425, "epoch": 70, "lr": 9.523377829941007e-05} +{"train_loss": 0.017203569412231445, "grad_norm": 0.08328808099031448, "global_step": 7426, "epoch": 70, "lr": 9.523251535580766e-05} +{"train_loss": 0.016390973702073097, "grad_norm": 0.10909456759691238, "global_step": 7427, "epoch": 70, "lr": 9.523125225327785e-05} +{"train_loss": 0.018571149557828903, "grad_norm": 0.06027046591043472, "global_step": 7428, "epoch": 70, "lr": 9.522998899182509e-05} +{"train_loss": 0.01975330524146557, "grad_norm": 0.08797463774681091, "global_step": 7429, "epoch": 70, "lr": 9.522872557145382e-05} +{"train_loss": 0.01501805055886507, "grad_norm": 0.06222454831004143, "global_step": 7430, "epoch": 70, "lr": 9.522746199216848e-05} +{"train_loss": 0.018428657203912735, "grad_norm": 0.06875424087047577, "global_step": 7431, "epoch": 70, "lr": 9.522619825397351e-05} +{"train_loss": 0.017837125808000565, "grad_norm": 0.07023125141859055, "global_step": 7432, "epoch": 70, "lr": 9.522493435687335e-05} +{"train_loss": 0.018631070852279663, "grad_norm": 0.09790017455816269, "global_step": 7433, "epoch": 70, "lr": 9.522367030087243e-05} +{"train_loss": 0.01860722526907921, "grad_norm": 0.10136798769235611, "global_step": 7434, "epoch": 70, "lr": 9.522240608597521e-05} +{"train_loss": 0.01984289661049843, "grad_norm": 0.03961482271552086, "global_step": 7435, "epoch": 70, "lr": 9.52211417121861e-05} +{"train_loss": 0.017590001225471497, "grad_norm": 0.11307729780673981, "global_step": 7436, "epoch": 70, "lr": 9.521987717950958e-05} +{"train_loss": 0.016330121085047722, "grad_norm": 0.07655858993530273, "global_step": 7437, "epoch": 70, "lr": 9.521861248795007e-05} +{"train_loss": 0.01632164977490902, "grad_norm": 0.0673358365893364, "global_step": 7438, "epoch": 70, "lr": 9.521734763751205e-05} +{"train_loss": 0.023630548268556595, "grad_norm": 0.05467076599597931, "global_step": 7439, "epoch": 70, "lr": 9.52160826281999e-05} +{"train_loss": 0.02136784978210926, "grad_norm": 0.0702005922794342, "global_step": 7440, "epoch": 70, "lr": 9.52148174600181e-05} +{"train_loss": 0.01945517770946026, "grad_norm": 0.07106516510248184, "global_step": 7441, "epoch": 70, "lr": 9.521355213297112e-05} +{"train_loss": 0.02041744813323021, "grad_norm": 0.09303552657365799, "global_step": 7442, "epoch": 70, "lr": 9.521228664706337e-05} +{"train_loss": 0.01861528679728508, "grad_norm": 0.06807879358530045, "global_step": 7443, "epoch": 70, "lr": 9.52110210022993e-05} +{"train_loss": 0.01867154985666275, "grad_norm": 0.10229185223579407, "global_step": 7444, "epoch": 70, "lr": 9.520975519868338e-05} +{"train_loss": 0.020180102437734604, "grad_norm": 0.10103832185268402, "global_step": 7445, "epoch": 70, "lr": 9.520848923622002e-05} +{"train_loss": 0.02120297960937023, "grad_norm": 0.09725934267044067, "global_step": 7446, "epoch": 70, "lr": 9.52072231149137e-05} +{"train_loss": 0.016815895214676857, "grad_norm": 0.07295428961515427, "global_step": 7447, "epoch": 70, "lr": 9.520595683476885e-05} +{"train_loss": 0.020187413319945335, "grad_norm": 0.0625515878200531, "global_step": 7448, "epoch": 70, "lr": 9.520469039578994e-05} +{"train_loss": 0.018707146868109703, "grad_norm": 0.05372724309563637, "global_step": 7449, "epoch": 70, "lr": 9.520342379798141e-05} +{"train_loss": 0.01821918413043022, "grad_norm": 0.0658237561583519, "global_step": 7450, "epoch": 70, "lr": 9.52021570413477e-05} +{"train_loss": 0.021267017349600792, "grad_norm": 0.04346712306141853, "global_step": 7451, "epoch": 70, "lr": 9.520089012589326e-05} +{"train_loss": 0.01894140988588333, "grad_norm": 0.08530106395483017, "global_step": 7452, "epoch": 70, "lr": 9.519962305162255e-05} +{"train_loss": 0.022440295666456223, "grad_norm": 0.06185042858123779, "global_step": 7453, "epoch": 70, "lr": 9.519835581854004e-05} +{"train_loss": 0.019508250057697296, "grad_norm": 0.06006130576133728, "global_step": 7454, "epoch": 70, "lr": 9.519708842665013e-05} +{"train_loss": 0.017446380108594894, "grad_norm": 0.07117298245429993, "global_step": 7455, "epoch": 70, "lr": 9.519582087595734e-05} +{"train_loss": 0.015502176247537136, "grad_norm": 0.0686267837882042, "global_step": 7456, "epoch": 70, "lr": 9.519455316646608e-05} +{"train_loss": 0.015778563916683197, "grad_norm": 0.06500335782766342, "global_step": 7457, "epoch": 70, "lr": 9.519328529818081e-05} +{"train_loss": 0.017423536628484726, "grad_norm": 0.04906231164932251, "global_step": 7458, "epoch": 70, "lr": 9.5192017271106e-05} +{"train_loss": 0.01715017296373844, "grad_norm": 0.08565663546323776, "global_step": 7459, "epoch": 70, "lr": 9.519074908524609e-05} +{"train_loss": 0.021406838670372963, "grad_norm": 0.08131498843431473, "global_step": 7460, "epoch": 70, "lr": 9.518948074060554e-05} +{"train_loss": 0.018593428656458855, "grad_norm": 0.0391215942800045, "global_step": 7461, "epoch": 70, "lr": 9.518821223718881e-05} +{"train_loss": 0.019388604909181595, "grad_norm": 0.06501807272434235, "global_step": 7462, "epoch": 70, "lr": 9.518694357500034e-05} +{"train_loss": 0.019558396190404892, "grad_norm": 0.07676316797733307, "global_step": 7463, "epoch": 70, "lr": 9.518567475404461e-05} +{"train_loss": 0.019149091094732285, "grad_norm": 0.09154469519853592, "global_step": 7464, "epoch": 70, "lr": 9.518440577432608e-05} +{"train_loss": 0.019873259589076042, "grad_norm": 0.11008746176958084, "global_step": 7465, "epoch": 70, "lr": 9.518313663584919e-05} +{"train_loss": 0.019015531986951828, "grad_norm": 0.04110480844974518, "global_step": 7466, "epoch": 70, "lr": 9.51818673386184e-05} +{"train_loss": 0.016764702275395393, "grad_norm": 0.07529769837856293, "global_step": 7467, "epoch": 70, "lr": 9.51805978826382e-05} +{"train_loss": 0.014268557541072369, "grad_norm": 0.05364331603050232, "global_step": 7468, "epoch": 70, "lr": 9.5179328267913e-05} +{"train_loss": 0.01685948111116886, "grad_norm": 0.10851570963859558, "global_step": 7469, "epoch": 70, "lr": 9.51780584944473e-05} +{"train_loss": 0.0215715654194355, "grad_norm": 0.08292610198259354, "global_step": 7470, "epoch": 70, "lr": 9.517678856224554e-05} +{"train_loss": 0.017287518829107285, "grad_norm": 0.157149076461792, "global_step": 7471, "epoch": 70, "lr": 9.517551847131221e-05} +{"train_loss": 0.018673870712518692, "grad_norm": 0.08044986426830292, "global_step": 7472, "epoch": 70, "lr": 9.517424822165175e-05} +{"train_loss": 0.02156030759215355, "grad_norm": 0.10047238320112228, "global_step": 7473, "epoch": 70, "lr": 9.517297781326862e-05} +{"train_loss": 0.01975702866911888, "grad_norm": 0.08456117659807205, "global_step": 7474, "epoch": 70, "lr": 9.51717072461673e-05} +{"train_loss": 0.01882275380194187, "grad_norm": 0.06506463885307312, "global_step": 7475, "epoch": 70, "lr": 9.517043652035224e-05} +{"train_loss": 0.015305398032069206, "grad_norm": 0.05958088859915733, "global_step": 7476, "epoch": 70, "lr": 9.516916563582791e-05} +{"train_loss": 0.01949363201856613, "grad_norm": 0.17066335678100586, "global_step": 7477, "epoch": 70, "lr": 9.516789459259878e-05} +{"train_loss": 0.016633950173854828, "grad_norm": 0.11000671237707138, "global_step": 7478, "epoch": 70, "lr": 9.51666233906693e-05} +{"train_loss": 0.017741389572620392, "grad_norm": 0.0635829046368599, "global_step": 7479, "epoch": 70, "lr": 9.516535203004395e-05} +{"train_loss": 0.0165406446903944, "grad_norm": 0.06876610219478607, "global_step": 7480, "epoch": 70, "lr": 9.516408051072721e-05} +{"train_loss": 0.018621966242790222, "grad_norm": 0.07376264035701752, "global_step": 7481, "epoch": 70, "lr": 9.516280883272353e-05} +{"train_loss": 0.018946141004562378, "grad_norm": 0.049361180514097214, "global_step": 7482, "epoch": 70, "lr": 9.516153699603736e-05} +{"train_loss": 0.017451588064432144, "grad_norm": 0.08130890130996704, "global_step": 7483, "epoch": 70, "lr": 9.516026500067321e-05} +{"train_loss": 0.017101561650633812, "grad_norm": 0.07589545845985413, "global_step": 7484, "epoch": 70, "lr": 9.515899284663553e-05} +{"train_loss": 0.018343327566981316, "grad_norm": 0.04789840802550316, "global_step": 7485, "epoch": 70, "lr": 9.515772053392876e-05} +{"train_loss": 0.014542479068040848, "grad_norm": 0.049569021910429, "global_step": 7486, "epoch": 70, "lr": 9.515644806255743e-05} +{"train_loss": 0.017277564853429794, "grad_norm": 0.06426200270652771, "global_step": 7487, "epoch": 70, "lr": 9.515517543252596e-05} +{"train_loss": 0.018249841406941414, "grad_norm": 0.08645960688591003, "global_step": 7488, "epoch": 70, "lr": 9.515390264383884e-05} +{"train_loss": 0.016965020447969437, "grad_norm": 0.12173182517290115, "global_step": 7489, "epoch": 70, "lr": 9.515262969650056e-05} +{"train_loss": 0.018300995230674744, "grad_norm": 0.08540698885917664, "global_step": 7490, "epoch": 70, "lr": 9.515135659051558e-05} +{"train_loss": 0.018210336565971375, "grad_norm": 0.05702938511967659, "global_step": 7491, "epoch": 70, "lr": 9.515008332588834e-05} +{"train_loss": 0.02162114530801773, "grad_norm": 0.05846830829977989, "global_step": 7492, "epoch": 70, "lr": 9.514880990262336e-05} +{"train_loss": 0.018514420837163925, "grad_norm": 0.08848106861114502, "global_step": 7493, "epoch": 70, "lr": 9.51475363207251e-05} +{"train_loss": 0.020457087084650993, "grad_norm": 0.048097386956214905, "global_step": 7494, "epoch": 70, "lr": 9.514626258019802e-05} +{"train_loss": 0.016427550464868546, "grad_norm": 0.052971579134464264, "global_step": 7495, "epoch": 70, "lr": 9.514498868104661e-05} +{"train_loss": 0.01975138485431671, "grad_norm": 0.08127866685390472, "global_step": 7496, "epoch": 70, "lr": 9.514371462327534e-05} +{"train_loss": 0.019973989576101303, "grad_norm": 0.08271809667348862, "global_step": 7497, "epoch": 70, "lr": 9.514244040688871e-05} +{"train_loss": 0.015440662391483784, "grad_norm": 0.07684038579463959, "global_step": 7498, "epoch": 70, "lr": 9.514116603189116e-05} +{"train_loss": 0.018332790583372116, "grad_norm": 0.049904752522706985, "global_step": 7499, "epoch": 70, "lr": 9.513989149828718e-05} +{"train_loss": 0.017273807898163795, "grad_norm": 0.040442902594804764, "global_step": 7500, "epoch": 70, "lr": 9.513861680608126e-05} +{"train_loss": 0.019448380917310715, "grad_norm": 0.09556582570075989, "global_step": 7501, "epoch": 70, "lr": 9.513734195527786e-05} +{"train_loss": 0.018627634271979332, "grad_norm": 0.04967343434691429, "global_step": 7502, "epoch": 70, "lr": 9.513606694588149e-05} +{"train_loss": 0.016167722642421722, "grad_norm": 0.04351399838924408, "global_step": 7503, "epoch": 70, "lr": 9.51347917778966e-05} +{"train_loss": 0.017093388363718987, "grad_norm": 0.05914038419723511, "global_step": 7504, "epoch": 70, "lr": 9.513351645132767e-05} +{"train_loss": 0.016228584572672844, "grad_norm": 0.058996111154556274, "global_step": 7505, "epoch": 70, "lr": 9.51322409661792e-05} +{"train_loss": 0.015267495065927505, "grad_norm": 0.06508654356002808, "global_step": 7506, "epoch": 70, "lr": 9.513096532245567e-05} +{"train_loss": 0.017259914427995682, "grad_norm": 0.07412869483232498, "global_step": 7507, "epoch": 70, "lr": 9.512968952016155e-05} +{"train_loss": 0.018907995894551277, "grad_norm": 0.10016962885856628, "global_step": 7508, "epoch": 70, "lr": 9.512841355930131e-05} +{"train_loss": 0.016401037573814392, "grad_norm": 0.07306954264640808, "global_step": 7509, "epoch": 70, "lr": 9.512713743987946e-05} +{"train_loss": 0.01933591067790985, "grad_norm": 0.14038977026939392, "global_step": 7510, "epoch": 70, "lr": 9.512586116190049e-05} +{"train_loss": 0.019310466945171356, "grad_norm": 0.04902740567922592, "global_step": 7511, "epoch": 70, "lr": 9.512458472536887e-05} +{"train_loss": 0.016385558992624283, "grad_norm": 0.05885104089975357, "global_step": 7512, "epoch": 70, "lr": 9.512330813028907e-05} +{"train_loss": 0.019657764583826065, "grad_norm": 0.0499330535531044, "global_step": 7513, "epoch": 70, "lr": 9.512203137666558e-05} +{"train_loss": 0.016606885939836502, "grad_norm": 0.06672412902116776, "global_step": 7514, "epoch": 70, "lr": 9.51207544645029e-05} +{"train_loss": 0.021275967359542847, "grad_norm": 0.1414663940668106, "global_step": 7515, "epoch": 70, "lr": 9.511947739380553e-05} +{"train_loss": 0.017535533756017685, "grad_norm": 0.07492085546255112, "global_step": 7516, "epoch": 70, "lr": 9.511820016457793e-05} +{"train_loss": 0.0179300494492054, "grad_norm": 0.1526288390159607, "global_step": 7517, "epoch": 70, "lr": 9.511692277682459e-05} +{"train_loss": 0.017585519701242447, "grad_norm": 0.1049591451883316, "global_step": 7518, "epoch": 70, "lr": 9.511564523054999e-05} +{"train_loss": 0.01918773539364338, "grad_norm": 0.140010267496109, "global_step": 7519, "epoch": 70, "lr": 9.511436752575866e-05} +{"train_loss": 0.01965087652206421, "grad_norm": 0.06584472954273224, "global_step": 7520, "epoch": 70, "lr": 9.511308966245504e-05} +{"train_loss": 0.019633488729596138, "grad_norm": 0.07948344200849533, "global_step": 7521, "epoch": 70, "lr": 9.511181164064367e-05} +{"train_loss": 0.019591832533478737, "grad_norm": 0.0487484410405159, "global_step": 7522, "epoch": 70, "lr": 9.511053346032899e-05} +{"train_loss": 0.01749749854207039, "grad_norm": 0.08896218240261078, "global_step": 7523, "epoch": 70, "lr": 9.510925512151552e-05} +{"train_loss": 0.018984545022249222, "grad_norm": 0.09969523549079895, "global_step": 7524, "epoch": 70, "lr": 9.510797662420774e-05} +{"train_loss": 0.018434751550403406, "grad_norm": 0.06031890958547592, "global_step": 7525, "epoch": 70, "lr": 9.510669796841014e-05, "train_batch_mse_error": 0.01666896417737007, "val_loss": 0.01883651502430439} +{"train_loss": 0.0177285335958004, "grad_norm": 0.07544708997011185, "global_step": 7526, "epoch": 71, "lr": 9.510541915412722e-05} +{"train_loss": 0.019335120916366577, "grad_norm": 0.0741531029343605, "global_step": 7527, "epoch": 71, "lr": 9.51041401813635e-05} +{"train_loss": 0.018909649923443794, "grad_norm": 0.0459706112742424, "global_step": 7528, "epoch": 71, "lr": 9.510286105012343e-05} +{"train_loss": 0.01603289507329464, "grad_norm": 0.05828410014510155, "global_step": 7529, "epoch": 71, "lr": 9.510158176041153e-05} +{"train_loss": 0.01811152696609497, "grad_norm": 0.10863185673952103, "global_step": 7530, "epoch": 71, "lr": 9.510030231223228e-05} +{"train_loss": 0.021425414830446243, "grad_norm": 0.11774078756570816, "global_step": 7531, "epoch": 71, "lr": 9.509902270559017e-05} +{"train_loss": 0.01876162178814411, "grad_norm": 0.12412481755018234, "global_step": 7532, "epoch": 71, "lr": 9.509774294048972e-05} +{"train_loss": 0.021095339208841324, "grad_norm": 0.09355854243040085, "global_step": 7533, "epoch": 71, "lr": 9.50964630169354e-05} +{"train_loss": 0.021897131577134132, "grad_norm": 0.0472545325756073, "global_step": 7534, "epoch": 71, "lr": 9.509518293493174e-05} +{"train_loss": 0.01700657419860363, "grad_norm": 0.13041891157627106, "global_step": 7535, "epoch": 71, "lr": 9.50939026944832e-05} +{"train_loss": 0.019633401185274124, "grad_norm": 0.12196966260671616, "global_step": 7536, "epoch": 71, "lr": 9.50926222955943e-05} +{"train_loss": 0.020194601267576218, "grad_norm": 0.1034713089466095, "global_step": 7537, "epoch": 71, "lr": 9.509134173826955e-05} +{"train_loss": 0.01767249032855034, "grad_norm": 0.05505947023630142, "global_step": 7538, "epoch": 71, "lr": 9.509006102251342e-05} +{"train_loss": 0.018024927005171776, "grad_norm": 0.09463607519865036, "global_step": 7539, "epoch": 71, "lr": 9.508878014833043e-05} +{"train_loss": 0.0187561996281147, "grad_norm": 0.05950497090816498, "global_step": 7540, "epoch": 71, "lr": 9.508749911572507e-05} +{"train_loss": 0.020114634186029434, "grad_norm": 0.050677068531513214, "global_step": 7541, "epoch": 71, "lr": 9.508621792470186e-05} +{"train_loss": 0.017033714801073074, "grad_norm": 0.03648068755865097, "global_step": 7542, "epoch": 71, "lr": 9.508493657526527e-05} +{"train_loss": 0.0195065438747406, "grad_norm": 0.08986607193946838, "global_step": 7543, "epoch": 71, "lr": 9.508365506741983e-05} +{"train_loss": 0.021535363048315048, "grad_norm": 0.04558153450489044, "global_step": 7544, "epoch": 71, "lr": 9.508237340117003e-05} +{"train_loss": 0.01847928762435913, "grad_norm": 0.09428760409355164, "global_step": 7545, "epoch": 71, "lr": 9.508109157652038e-05} +{"train_loss": 0.016826041042804718, "grad_norm": 0.19557365775108337, "global_step": 7546, "epoch": 71, "lr": 9.507980959347538e-05} +{"train_loss": 0.018906019628047943, "grad_norm": 0.13855372369289398, "global_step": 7547, "epoch": 71, "lr": 9.507852745203954e-05} +{"train_loss": 0.014691264368593693, "grad_norm": 0.09813839942216873, "global_step": 7548, "epoch": 71, "lr": 9.507724515221734e-05} +{"train_loss": 0.020867273211479187, "grad_norm": 0.11752522736787796, "global_step": 7549, "epoch": 71, "lr": 9.507596269401331e-05} +{"train_loss": 0.017269261181354523, "grad_norm": 0.09268756210803986, "global_step": 7550, "epoch": 71, "lr": 9.507468007743197e-05} +{"train_loss": 0.02038761042058468, "grad_norm": 0.18375550210475922, "global_step": 7551, "epoch": 71, "lr": 9.507339730247777e-05} +{"train_loss": 0.017817627638578415, "grad_norm": 0.10826145857572556, "global_step": 7552, "epoch": 71, "lr": 9.507211436915529e-05} +{"train_loss": 0.017671333625912666, "grad_norm": 0.040298983454704285, "global_step": 7553, "epoch": 71, "lr": 9.507083127746899e-05} +{"train_loss": 0.018766429275274277, "grad_norm": 0.09771167486906052, "global_step": 7554, "epoch": 71, "lr": 9.506954802742338e-05} +{"train_loss": 0.020784763619303703, "grad_norm": 0.06155845895409584, "global_step": 7555, "epoch": 71, "lr": 9.506826461902299e-05} +{"train_loss": 0.015216313302516937, "grad_norm": 0.07197543233633041, "global_step": 7556, "epoch": 71, "lr": 9.50669810522723e-05} +{"train_loss": 0.021157801151275635, "grad_norm": 0.06220231205224991, "global_step": 7557, "epoch": 71, "lr": 9.506569732717585e-05} +{"train_loss": 0.018050413578748703, "grad_norm": 0.06856116652488708, "global_step": 7558, "epoch": 71, "lr": 9.506441344373813e-05} +{"train_loss": 0.019624751061201096, "grad_norm": 0.10461648553609848, "global_step": 7559, "epoch": 71, "lr": 9.506312940196367e-05} +{"train_loss": 0.016636330634355545, "grad_norm": 0.045814331620931625, "global_step": 7560, "epoch": 71, "lr": 9.506184520185697e-05} +{"train_loss": 0.018242310732603073, "grad_norm": 0.23668169975280762, "global_step": 7561, "epoch": 71, "lr": 9.506056084342254e-05} +{"train_loss": 0.017020387575030327, "grad_norm": 0.12985603511333466, "global_step": 7562, "epoch": 71, "lr": 9.505927632666489e-05} +{"train_loss": 0.017822565510869026, "grad_norm": 0.17394737899303436, "global_step": 7563, "epoch": 71, "lr": 9.505799165158856e-05} +{"train_loss": 0.017935164272785187, "grad_norm": 0.08856765180826187, "global_step": 7564, "epoch": 71, "lr": 9.505670681819801e-05} +{"train_loss": 0.018815405666828156, "grad_norm": 0.07088547199964523, "global_step": 7565, "epoch": 71, "lr": 9.505542182649781e-05} +{"train_loss": 0.021133888512849808, "grad_norm": 0.18309924006462097, "global_step": 7566, "epoch": 71, "lr": 9.505413667649243e-05} +{"train_loss": 0.021096473559737206, "grad_norm": 0.06785190850496292, "global_step": 7567, "epoch": 71, "lr": 9.505285136818644e-05} +{"train_loss": 0.01724325120449066, "grad_norm": 0.06721264123916626, "global_step": 7568, "epoch": 71, "lr": 9.505156590158431e-05} +{"train_loss": 0.018941374495625496, "grad_norm": 0.0393909253180027, "global_step": 7569, "epoch": 71, "lr": 9.505028027669056e-05} +{"train_loss": 0.016608478501439095, "grad_norm": 0.0533713735640049, "global_step": 7570, "epoch": 71, "lr": 9.504899449350972e-05} +{"train_loss": 0.02002793364226818, "grad_norm": 0.04980429634451866, "global_step": 7571, "epoch": 71, "lr": 9.50477085520463e-05} +{"train_loss": 0.01831865683197975, "grad_norm": 0.07180982083082199, "global_step": 7572, "epoch": 71, "lr": 9.504642245230484e-05} +{"train_loss": 0.01648656092584133, "grad_norm": 0.043059494346380234, "global_step": 7573, "epoch": 71, "lr": 9.504513619428984e-05} +{"train_loss": 0.019686536863446236, "grad_norm": 0.06367866694927216, "global_step": 7574, "epoch": 71, "lr": 9.504384977800581e-05} +{"train_loss": 0.020667865872383118, "grad_norm": 0.05507173016667366, "global_step": 7575, "epoch": 71, "lr": 9.504256320345728e-05} +{"train_loss": 0.016254965215921402, "grad_norm": 0.07024702429771423, "global_step": 7576, "epoch": 71, "lr": 9.504127647064877e-05} +{"train_loss": 0.01699329912662506, "grad_norm": 0.15986375510692596, "global_step": 7577, "epoch": 71, "lr": 9.50399895795848e-05} +{"train_loss": 0.01794603280723095, "grad_norm": 0.10758398473262787, "global_step": 7578, "epoch": 71, "lr": 9.503870253026991e-05} +{"train_loss": 0.0197712704539299, "grad_norm": 0.23501451313495636, "global_step": 7579, "epoch": 71, "lr": 9.50374153227086e-05} +{"train_loss": 0.02075551450252533, "grad_norm": 0.08508665859699249, "global_step": 7580, "epoch": 71, "lr": 9.50361279569054e-05} +{"train_loss": 0.0166337713599205, "grad_norm": 0.14656856656074524, "global_step": 7581, "epoch": 71, "lr": 9.503484043286482e-05} +{"train_loss": 0.017516393214464188, "grad_norm": 0.11715802550315857, "global_step": 7582, "epoch": 71, "lr": 9.50335527505914e-05} +{"train_loss": 0.018267132341861725, "grad_norm": 0.10159096866846085, "global_step": 7583, "epoch": 71, "lr": 9.503226491008968e-05} +{"train_loss": 0.014686819165945053, "grad_norm": 0.05236329138278961, "global_step": 7584, "epoch": 71, "lr": 9.503097691136413e-05} +{"train_loss": 0.017021507024765015, "grad_norm": 0.05907677114009857, "global_step": 7585, "epoch": 71, "lr": 9.502968875441933e-05} +{"train_loss": 0.017152391374111176, "grad_norm": 0.07109314948320389, "global_step": 7586, "epoch": 71, "lr": 9.502840043925981e-05} +{"train_loss": 0.01620742306113243, "grad_norm": 0.14852969348430634, "global_step": 7587, "epoch": 71, "lr": 9.502711196589003e-05} +{"train_loss": 0.019997598603367805, "grad_norm": 0.11671816557645798, "global_step": 7588, "epoch": 71, "lr": 9.502582333431458e-05} +{"train_loss": 0.015636347234249115, "grad_norm": 0.09608805179595947, "global_step": 7589, "epoch": 71, "lr": 9.502453454453796e-05} +{"train_loss": 0.020210804417729378, "grad_norm": 0.06461402028799057, "global_step": 7590, "epoch": 71, "lr": 9.502324559656472e-05} +{"train_loss": 0.019859839230775833, "grad_norm": 0.14032913744449615, "global_step": 7591, "epoch": 71, "lr": 9.502195649039937e-05} +{"train_loss": 0.01941945031285286, "grad_norm": 0.08950933068990707, "global_step": 7592, "epoch": 71, "lr": 9.502066722604646e-05} +{"train_loss": 0.01870875060558319, "grad_norm": 0.05168928578495979, "global_step": 7593, "epoch": 71, "lr": 9.501937780351049e-05} +{"train_loss": 0.01895114779472351, "grad_norm": 0.1707846075296402, "global_step": 7594, "epoch": 71, "lr": 9.5018088222796e-05} +{"train_loss": 0.018408767879009247, "grad_norm": 0.06601563841104507, "global_step": 7595, "epoch": 71, "lr": 9.501679848390752e-05} +{"train_loss": 0.016697699204087257, "grad_norm": 0.0665891170501709, "global_step": 7596, "epoch": 71, "lr": 9.50155085868496e-05} +{"train_loss": 0.018719185143709183, "grad_norm": 0.07592606544494629, "global_step": 7597, "epoch": 71, "lr": 9.501421853162675e-05} +{"train_loss": 0.02076975256204605, "grad_norm": 0.05883260816335678, "global_step": 7598, "epoch": 71, "lr": 9.501292831824353e-05} +{"train_loss": 0.019002262502908707, "grad_norm": 0.07292324304580688, "global_step": 7599, "epoch": 71, "lr": 9.501163794670444e-05} +{"train_loss": 0.02063436061143875, "grad_norm": 0.07985340803861618, "global_step": 7600, "epoch": 71, "lr": 9.501034741701403e-05} +{"train_loss": 0.01784837804734707, "grad_norm": 0.12077397108078003, "global_step": 7601, "epoch": 71, "lr": 9.500905672917684e-05} +{"train_loss": 0.020198779180645943, "grad_norm": 0.05376068875193596, "global_step": 7602, "epoch": 71, "lr": 9.500776588319739e-05} +{"train_loss": 0.018405750393867493, "grad_norm": 0.11441221088171005, "global_step": 7603, "epoch": 71, "lr": 9.500647487908024e-05} +{"train_loss": 0.0179080031812191, "grad_norm": 0.05217328667640686, "global_step": 7604, "epoch": 71, "lr": 9.500518371682989e-05} +{"train_loss": 0.017239291220903397, "grad_norm": 0.1628785878419876, "global_step": 7605, "epoch": 71, "lr": 9.500389239645091e-05} +{"train_loss": 0.020622847601771355, "grad_norm": 0.039262689650058746, "global_step": 7606, "epoch": 71, "lr": 9.500260091794782e-05} +{"train_loss": 0.01791227236390114, "grad_norm": 0.07517903298139572, "global_step": 7607, "epoch": 71, "lr": 9.500130928132515e-05} +{"train_loss": 0.01984577625989914, "grad_norm": 0.12426313012838364, "global_step": 7608, "epoch": 71, "lr": 9.500001748658747e-05} +{"train_loss": 0.01827474683523178, "grad_norm": 0.059088777750730515, "global_step": 7609, "epoch": 71, "lr": 9.49987255337393e-05} +{"train_loss": 0.01941421627998352, "grad_norm": 0.13356153666973114, "global_step": 7610, "epoch": 71, "lr": 9.499743342278516e-05} +{"train_loss": 0.015939457342028618, "grad_norm": 0.04092475399374962, "global_step": 7611, "epoch": 71, "lr": 9.499614115372961e-05} +{"train_loss": 0.017104800790548325, "grad_norm": 0.05266899615526199, "global_step": 7612, "epoch": 71, "lr": 9.49948487265772e-05} +{"train_loss": 0.017307385802268982, "grad_norm": 0.056057706475257874, "global_step": 7613, "epoch": 71, "lr": 9.499355614133245e-05} +{"train_loss": 0.01565162092447281, "grad_norm": 0.048517271876335144, "global_step": 7614, "epoch": 71, "lr": 9.499226339799992e-05} +{"train_loss": 0.021236956119537354, "grad_norm": 0.07095657289028168, "global_step": 7615, "epoch": 71, "lr": 9.499097049658415e-05} +{"train_loss": 0.018543388694524765, "grad_norm": 0.12780910730361938, "global_step": 7616, "epoch": 71, "lr": 9.498967743708964e-05} +{"train_loss": 0.016977334395051003, "grad_norm": 0.05293354019522667, "global_step": 7617, "epoch": 71, "lr": 9.498838421952099e-05} +{"train_loss": 0.01824065111577511, "grad_norm": 0.1624070703983307, "global_step": 7618, "epoch": 71, "lr": 9.498709084388273e-05} +{"train_loss": 0.016983075067400932, "grad_norm": 0.17142866551876068, "global_step": 7619, "epoch": 71, "lr": 9.498579731017939e-05} +{"train_loss": 0.0200950987637043, "grad_norm": 0.1401565819978714, "global_step": 7620, "epoch": 71, "lr": 9.498450361841554e-05} +{"train_loss": 0.016824129968881607, "grad_norm": 0.04444888234138489, "global_step": 7621, "epoch": 71, "lr": 9.498320976859568e-05} +{"train_loss": 0.01848440431058407, "grad_norm": 0.08735327422618866, "global_step": 7622, "epoch": 71, "lr": 9.498191576072439e-05} +{"train_loss": 0.018642088398337364, "grad_norm": 0.047427739948034286, "global_step": 7623, "epoch": 71, "lr": 9.498062159480621e-05} +{"train_loss": 0.01636829599738121, "grad_norm": 0.057102516293525696, "global_step": 7624, "epoch": 71, "lr": 9.49793272708457e-05} +{"train_loss": 0.019513007253408432, "grad_norm": 0.048530176281929016, "global_step": 7625, "epoch": 71, "lr": 9.49780327888474e-05} +{"train_loss": 0.01852869801223278, "grad_norm": 0.05357607081532478, "global_step": 7626, "epoch": 71, "lr": 9.497673814881583e-05} +{"train_loss": 0.01869051530957222, "grad_norm": 0.05433264374732971, "global_step": 7627, "epoch": 71, "lr": 9.497544335075557e-05} +{"train_loss": 0.018200067803263664, "grad_norm": 0.06183112412691116, "global_step": 7628, "epoch": 71, "lr": 9.497414839467116e-05} +{"train_loss": 0.017526928335428238, "grad_norm": 0.047406770288944244, "global_step": 7629, "epoch": 71, "lr": 9.497285328056716e-05} +{"train_loss": 0.01922815479338169, "grad_norm": 0.05264938622713089, "global_step": 7630, "epoch": 71, "lr": 9.49715580084481e-05} +{"train_loss": 0.01846301090850864, "grad_norm": 0.0922430008649826, "global_step": 7631, "epoch": 71, "lr": 9.497026257831855e-05, "val_loss": 0.018965614959597588} +{"train_loss": 0.023332346230745316, "grad_norm": 0.07491911947727203, "global_step": 7632, "epoch": 72, "lr": 9.496896699018307e-05} +{"train_loss": 0.018447592854499817, "grad_norm": 0.1100459173321724, "global_step": 7633, "epoch": 72, "lr": 9.496767124404617e-05} +{"train_loss": 0.021123358979821205, "grad_norm": 0.08540317416191101, "global_step": 7634, "epoch": 72, "lr": 9.496637533991244e-05} +{"train_loss": 0.01704874075949192, "grad_norm": 0.0972735658288002, "global_step": 7635, "epoch": 72, "lr": 9.496507927778641e-05} +{"train_loss": 0.01781631074845791, "grad_norm": 0.08527760207653046, "global_step": 7636, "epoch": 72, "lr": 9.496378305767268e-05} +{"train_loss": 0.019729401916265488, "grad_norm": 0.11781422048807144, "global_step": 7637, "epoch": 72, "lr": 9.496248667957575e-05} +{"train_loss": 0.015828300267457962, "grad_norm": 0.05347519367933273, "global_step": 7638, "epoch": 72, "lr": 9.496119014350018e-05} +{"train_loss": 0.017477137967944145, "grad_norm": 0.05595432594418526, "global_step": 7639, "epoch": 72, "lr": 9.495989344945056e-05} +{"train_loss": 0.01654389128088951, "grad_norm": 0.04811794310808182, "global_step": 7640, "epoch": 72, "lr": 9.495859659743143e-05} +{"train_loss": 0.018171072006225586, "grad_norm": 0.043348900973796844, "global_step": 7641, "epoch": 72, "lr": 9.495729958744732e-05} +{"train_loss": 0.015472063794732094, "grad_norm": 0.0958857461810112, "global_step": 7642, "epoch": 72, "lr": 9.495600241950283e-05} +{"train_loss": 0.018499936908483505, "grad_norm": 0.07556528598070145, "global_step": 7643, "epoch": 72, "lr": 9.49547050936025e-05} +{"train_loss": 0.017485585063695908, "grad_norm": 0.04023472219705582, "global_step": 7644, "epoch": 72, "lr": 9.495340760975088e-05} +{"train_loss": 0.021112672984600067, "grad_norm": 0.07667846232652664, "global_step": 7645, "epoch": 72, "lr": 9.495210996795254e-05} +{"train_loss": 0.01813700422644615, "grad_norm": 0.1135227158665657, "global_step": 7646, "epoch": 72, "lr": 9.495081216821203e-05} +{"train_loss": 0.01782795786857605, "grad_norm": 0.1167842447757721, "global_step": 7647, "epoch": 72, "lr": 9.494951421053391e-05} +{"train_loss": 0.018291868269443512, "grad_norm": 0.0629325658082962, "global_step": 7648, "epoch": 72, "lr": 9.494821609492274e-05} +{"train_loss": 0.0204540453851223, "grad_norm": 0.08273551613092422, "global_step": 7649, "epoch": 72, "lr": 9.494691782138311e-05} +{"train_loss": 0.018685221672058105, "grad_norm": 0.0798577219247818, "global_step": 7650, "epoch": 72, "lr": 9.494561938991956e-05} +{"train_loss": 0.0227959044277668, "grad_norm": 0.17571523785591125, "global_step": 7651, "epoch": 72, "lr": 9.494432080053661e-05} +{"train_loss": 0.023033834993839264, "grad_norm": 0.11417984217405319, "global_step": 7652, "epoch": 72, "lr": 9.49430220532389e-05} +{"train_loss": 0.018955916166305542, "grad_norm": 0.06349417567253113, "global_step": 7653, "epoch": 72, "lr": 9.494172314803094e-05} +{"train_loss": 0.018222590908408165, "grad_norm": 0.04944474995136261, "global_step": 7654, "epoch": 72, "lr": 9.494042408491731e-05} +{"train_loss": 0.015681907534599304, "grad_norm": 0.04699951782822609, "global_step": 7655, "epoch": 72, "lr": 9.493912486390258e-05} +{"train_loss": 0.018093474209308624, "grad_norm": 0.10830904543399811, "global_step": 7656, "epoch": 72, "lr": 9.49378254849913e-05} +{"train_loss": 0.017066296190023422, "grad_norm": 0.11598972231149673, "global_step": 7657, "epoch": 72, "lr": 9.493652594818805e-05} +{"train_loss": 0.020162804052233696, "grad_norm": 0.09077568352222443, "global_step": 7658, "epoch": 72, "lr": 9.49352262534974e-05} +{"train_loss": 0.019013598561286926, "grad_norm": 0.07708630710840225, "global_step": 7659, "epoch": 72, "lr": 9.493392640092389e-05} +{"train_loss": 0.02024083212018013, "grad_norm": 0.06223750859498978, "global_step": 7660, "epoch": 72, "lr": 9.493262639047212e-05} +{"train_loss": 0.017792604863643646, "grad_norm": 0.05603041127324104, "global_step": 7661, "epoch": 72, "lr": 9.493132622214662e-05} +{"train_loss": 0.015976006165146828, "grad_norm": 0.0540422685444355, "global_step": 7662, "epoch": 72, "lr": 9.4930025895952e-05} +{"train_loss": 0.02136501669883728, "grad_norm": 0.12252656370401382, "global_step": 7663, "epoch": 72, "lr": 9.49287254118928e-05} +{"train_loss": 0.019879689440131187, "grad_norm": 0.07525849342346191, "global_step": 7664, "epoch": 72, "lr": 9.492742476997358e-05} +{"train_loss": 0.015739168971776962, "grad_norm": 0.045335181057453156, "global_step": 7665, "epoch": 72, "lr": 9.492612397019895e-05} +{"train_loss": 0.01850169710814953, "grad_norm": 0.09907509386539459, "global_step": 7666, "epoch": 72, "lr": 9.492482301257346e-05} +{"train_loss": 0.017382711172103882, "grad_norm": 0.07563610374927521, "global_step": 7667, "epoch": 72, "lr": 9.492352189710168e-05} +{"train_loss": 0.01727786287665367, "grad_norm": 0.050727806985378265, "global_step": 7668, "epoch": 72, "lr": 9.492222062378817e-05} +{"train_loss": 0.01904319040477276, "grad_norm": 0.1404428482055664, "global_step": 7669, "epoch": 72, "lr": 9.49209191926375e-05} +{"train_loss": 0.016628200188279152, "grad_norm": 0.06934566795825958, "global_step": 7670, "epoch": 72, "lr": 9.491961760365428e-05} +{"train_loss": 0.016911445185542107, "grad_norm": 0.04813094809651375, "global_step": 7671, "epoch": 72, "lr": 9.491831585684306e-05} +{"train_loss": 0.015528728254139423, "grad_norm": 0.05813449248671532, "global_step": 7672, "epoch": 72, "lr": 9.49170139522084e-05} +{"train_loss": 0.018690630793571472, "grad_norm": 0.08022750914096832, "global_step": 7673, "epoch": 72, "lr": 9.491571188975489e-05} +{"train_loss": 0.019205596297979355, "grad_norm": 0.04861428588628769, "global_step": 7674, "epoch": 72, "lr": 9.49144096694871e-05} +{"train_loss": 0.01563696190714836, "grad_norm": 0.11223331093788147, "global_step": 7675, "epoch": 72, "lr": 9.491310729140961e-05} +{"train_loss": 0.01797705516219139, "grad_norm": 0.05082621052861214, "global_step": 7676, "epoch": 72, "lr": 9.491180475552699e-05} +{"train_loss": 0.016963262110948563, "grad_norm": 0.06681621819734573, "global_step": 7677, "epoch": 72, "lr": 9.491050206184382e-05} +{"train_loss": 0.017526740208268166, "grad_norm": 0.10029775649309158, "global_step": 7678, "epoch": 72, "lr": 9.490919921036468e-05} +{"train_loss": 0.018961116671562195, "grad_norm": 0.13094952702522278, "global_step": 7679, "epoch": 72, "lr": 9.490789620109415e-05} +{"train_loss": 0.016036463901400566, "grad_norm": 0.0427531898021698, "global_step": 7680, "epoch": 72, "lr": 9.49065930340368e-05} +{"train_loss": 0.020124021917581558, "grad_norm": 0.06175614148378372, "global_step": 7681, "epoch": 72, "lr": 9.49052897091972e-05} +{"train_loss": 0.018252022564411163, "grad_norm": 0.08241277933120728, "global_step": 7682, "epoch": 72, "lr": 9.490398622657994e-05} +{"train_loss": 0.01935773715376854, "grad_norm": 0.10150955617427826, "global_step": 7683, "epoch": 72, "lr": 9.490268258618962e-05} +{"train_loss": 0.019465390592813492, "grad_norm": 0.047770749777555466, "global_step": 7684, "epoch": 72, "lr": 9.490137878803079e-05} +{"train_loss": 0.016033658757805824, "grad_norm": 0.1065206453204155, "global_step": 7685, "epoch": 72, "lr": 9.490007483210803e-05} +{"train_loss": 0.01633480004966259, "grad_norm": 0.06587373465299606, "global_step": 7686, "epoch": 72, "lr": 9.489877071842595e-05} +{"train_loss": 0.018717097118496895, "grad_norm": 0.06515361368656158, "global_step": 7687, "epoch": 72, "lr": 9.489746644698912e-05} +{"train_loss": 0.017768297344446182, "grad_norm": 0.05433950573205948, "global_step": 7688, "epoch": 72, "lr": 9.489616201780211e-05} +{"train_loss": 0.01839285157620907, "grad_norm": 0.06996627151966095, "global_step": 7689, "epoch": 72, "lr": 9.48948574308695e-05} +{"train_loss": 0.01968439668416977, "grad_norm": 0.04779072105884552, "global_step": 7690, "epoch": 72, "lr": 9.489355268619591e-05} +{"train_loss": 0.018263408914208412, "grad_norm": 0.03398119658231735, "global_step": 7691, "epoch": 72, "lr": 9.489224778378587e-05} +{"train_loss": 0.020051894709467888, "grad_norm": 0.06590840220451355, "global_step": 7692, "epoch": 72, "lr": 9.489094272364402e-05} +{"train_loss": 0.018512940034270287, "grad_norm": 0.10157272964715958, "global_step": 7693, "epoch": 72, "lr": 9.488963750577491e-05} +{"train_loss": 0.018054887652397156, "grad_norm": 0.05565081909298897, "global_step": 7694, "epoch": 72, "lr": 9.488833213018313e-05} +{"train_loss": 0.017726434394717216, "grad_norm": 0.07481883466243744, "global_step": 7695, "epoch": 72, "lr": 9.488702659687328e-05} +{"train_loss": 0.018046077340841293, "grad_norm": 0.09902148693799973, "global_step": 7696, "epoch": 72, "lr": 9.488572090584995e-05} +{"train_loss": 0.016425296664237976, "grad_norm": 0.0730612650513649, "global_step": 7697, "epoch": 72, "lr": 9.48844150571177e-05} +{"train_loss": 0.018110990524291992, "grad_norm": 0.08094920963048935, "global_step": 7698, "epoch": 72, "lr": 9.488310905068114e-05} +{"train_loss": 0.018275802955031395, "grad_norm": 0.053017839789390564, "global_step": 7699, "epoch": 72, "lr": 9.488180288654485e-05} +{"train_loss": 0.019073575735092163, "grad_norm": 0.13908728957176208, "global_step": 7700, "epoch": 72, "lr": 9.488049656471343e-05} +{"train_loss": 0.017703011631965637, "grad_norm": 0.0918084904551506, "global_step": 7701, "epoch": 72, "lr": 9.487919008519145e-05} +{"train_loss": 0.019901569932699203, "grad_norm": 0.0628291442990303, "global_step": 7702, "epoch": 72, "lr": 9.487788344798353e-05} +{"train_loss": 0.019532879814505577, "grad_norm": 0.05698668956756592, "global_step": 7703, "epoch": 72, "lr": 9.487657665309424e-05} +{"train_loss": 0.016542036086320877, "grad_norm": 0.06473924219608307, "global_step": 7704, "epoch": 72, "lr": 9.487526970052818e-05} +{"train_loss": 0.0178002268075943, "grad_norm": 0.08742410689592361, "global_step": 7705, "epoch": 72, "lr": 9.487396259028992e-05} +{"train_loss": 0.01685279607772827, "grad_norm": 0.12226103246212006, "global_step": 7706, "epoch": 72, "lr": 9.487265532238408e-05} +{"train_loss": 0.02039818838238716, "grad_norm": 0.08192863315343857, "global_step": 7707, "epoch": 72, "lr": 9.487134789681524e-05} +{"train_loss": 0.020983058959245682, "grad_norm": 0.1097254827618599, "global_step": 7708, "epoch": 72, "lr": 9.487004031358802e-05} +{"train_loss": 0.018957694992423058, "grad_norm": 0.050404999405145645, "global_step": 7709, "epoch": 72, "lr": 9.486873257270696e-05} +{"train_loss": 0.01614684984087944, "grad_norm": 0.0853026732802391, "global_step": 7710, "epoch": 72, "lr": 9.486742467417671e-05} +{"train_loss": 0.019664447754621506, "grad_norm": 0.08320574462413788, "global_step": 7711, "epoch": 72, "lr": 9.486611661800185e-05} +{"train_loss": 0.019315805286169052, "grad_norm": 0.08500318229198456, "global_step": 7712, "epoch": 72, "lr": 9.486480840418694e-05} +{"train_loss": 0.01743747480213642, "grad_norm": 0.07022106647491455, "global_step": 7713, "epoch": 72, "lr": 9.48635000327366e-05} +{"train_loss": 0.019555170089006424, "grad_norm": 0.10868846625089645, "global_step": 7714, "epoch": 72, "lr": 9.486219150365546e-05} +{"train_loss": 0.018186453729867935, "grad_norm": 0.08054781705141068, "global_step": 7715, "epoch": 72, "lr": 9.486088281694808e-05} +{"train_loss": 0.02125205472111702, "grad_norm": 0.09964863955974579, "global_step": 7716, "epoch": 72, "lr": 9.485957397261906e-05} +{"train_loss": 0.018619854003190994, "grad_norm": 0.07062628120183945, "global_step": 7717, "epoch": 72, "lr": 9.485826497067301e-05} +{"train_loss": 0.020194847136735916, "grad_norm": 0.08261831849813461, "global_step": 7718, "epoch": 72, "lr": 9.485695581111453e-05} +{"train_loss": 0.018815362825989723, "grad_norm": 0.13116052746772766, "global_step": 7719, "epoch": 72, "lr": 9.485564649394822e-05} +{"train_loss": 0.01711210422217846, "grad_norm": 0.04758796468377113, "global_step": 7720, "epoch": 72, "lr": 9.485433701917865e-05} +{"train_loss": 0.020105725154280663, "grad_norm": 0.1263352632522583, "global_step": 7721, "epoch": 72, "lr": 9.485302738681048e-05} +{"train_loss": 0.01948583871126175, "grad_norm": 0.1698296219110489, "global_step": 7722, "epoch": 72, "lr": 9.485171759684825e-05} +{"train_loss": 0.016631226986646652, "grad_norm": 0.13860957324504852, "global_step": 7723, "epoch": 72, "lr": 9.48504076492966e-05} +{"train_loss": 0.016723131760954857, "grad_norm": 0.11048327386379242, "global_step": 7724, "epoch": 72, "lr": 9.484909754416013e-05} +{"train_loss": 0.01726209744811058, "grad_norm": 0.042119551450014114, "global_step": 7725, "epoch": 72, "lr": 9.484778728144343e-05} +{"train_loss": 0.016464587301015854, "grad_norm": 0.12852174043655396, "global_step": 7726, "epoch": 72, "lr": 9.48464768611511e-05} +{"train_loss": 0.018675196915864944, "grad_norm": 0.0934547707438469, "global_step": 7727, "epoch": 72, "lr": 9.484516628328776e-05} +{"train_loss": 0.02010979689657688, "grad_norm": 0.05718918517231941, "global_step": 7728, "epoch": 72, "lr": 9.484385554785802e-05} +{"train_loss": 0.018022876232862473, "grad_norm": 0.05948828533291817, "global_step": 7729, "epoch": 72, "lr": 9.484254465486647e-05} +{"train_loss": 0.01901363953948021, "grad_norm": 0.062788225710392, "global_step": 7730, "epoch": 72, "lr": 9.484123360431769e-05} +{"train_loss": 0.015925168991088867, "grad_norm": 0.12222979217767715, "global_step": 7731, "epoch": 72, "lr": 9.483992239621635e-05} +{"train_loss": 0.01803022064268589, "grad_norm": 0.04793030768632889, "global_step": 7732, "epoch": 72, "lr": 9.483861103056702e-05} +{"train_loss": 0.018541840836405754, "grad_norm": 0.05053732544183731, "global_step": 7733, "epoch": 72, "lr": 9.483729950737429e-05} +{"train_loss": 0.01855223998427391, "grad_norm": 0.12678086757659912, "global_step": 7734, "epoch": 72, "lr": 9.48359878266428e-05} +{"train_loss": 0.018382124602794647, "grad_norm": 0.04025368019938469, "global_step": 7735, "epoch": 72, "lr": 9.483467598837715e-05} +{"train_loss": 0.01856115087866783, "grad_norm": 0.08416031301021576, "global_step": 7736, "epoch": 72, "lr": 9.483336399258194e-05} +{"train_loss": 0.01835175388369639, "grad_norm": 0.12346525490283966, "global_step": 7737, "epoch": 72, "lr": 9.483205183926181e-05, "val_loss": 0.019017981365323067} +{"train_loss": 0.017169754952192307, "grad_norm": 0.11767960339784622, "global_step": 7738, "epoch": 73, "lr": 9.483073952842131e-05} +{"train_loss": 0.01867436245083809, "grad_norm": 0.1313929259777069, "global_step": 7739, "epoch": 73, "lr": 9.482942706006511e-05} +{"train_loss": 0.017389347776770592, "grad_norm": 0.06593125313520432, "global_step": 7740, "epoch": 73, "lr": 9.48281144341978e-05} +{"train_loss": 0.014743639156222343, "grad_norm": 0.05560534447431564, "global_step": 7741, "epoch": 73, "lr": 9.482680165082399e-05} +{"train_loss": 0.020438147708773613, "grad_norm": 0.08910571783781052, "global_step": 7742, "epoch": 73, "lr": 9.482548870994828e-05} +{"train_loss": 0.01688927412033081, "grad_norm": 0.040260083973407745, "global_step": 7743, "epoch": 73, "lr": 9.48241756115753e-05} +{"train_loss": 0.019968129694461823, "grad_norm": 0.042723268270492554, "global_step": 7744, "epoch": 73, "lr": 9.482286235570968e-05} +{"train_loss": 0.01813345216214657, "grad_norm": 0.052924592047929764, "global_step": 7745, "epoch": 73, "lr": 9.4821548942356e-05} +{"train_loss": 0.020024536177515984, "grad_norm": 0.049939535558223724, "global_step": 7746, "epoch": 73, "lr": 9.48202353715189e-05} +{"train_loss": 0.01982869952917099, "grad_norm": 0.11504258215427399, "global_step": 7747, "epoch": 73, "lr": 9.481892164320296e-05} +{"train_loss": 0.01605338044464588, "grad_norm": 0.054507240653038025, "global_step": 7748, "epoch": 73, "lr": 9.481760775741285e-05} +{"train_loss": 0.016841623932123184, "grad_norm": 0.04873877018690109, "global_step": 7749, "epoch": 73, "lr": 9.481629371415313e-05} +{"train_loss": 0.016213400289416313, "grad_norm": 0.059306006878614426, "global_step": 7750, "epoch": 73, "lr": 9.481497951342847e-05} +{"train_loss": 0.019800230860710144, "grad_norm": 0.04869470000267029, "global_step": 7751, "epoch": 73, "lr": 9.481366515524345e-05} +{"train_loss": 0.018700331449508667, "grad_norm": 0.07149285078048706, "global_step": 7752, "epoch": 73, "lr": 9.48123506396027e-05} +{"train_loss": 0.01771485060453415, "grad_norm": 0.06308324635028839, "global_step": 7753, "epoch": 73, "lr": 9.481103596651083e-05} +{"train_loss": 0.016514889895915985, "grad_norm": 0.04890233278274536, "global_step": 7754, "epoch": 73, "lr": 9.480972113597248e-05} +{"train_loss": 0.0181729719042778, "grad_norm": 0.07143735885620117, "global_step": 7755, "epoch": 73, "lr": 9.480840614799225e-05} +{"train_loss": 0.020671548321843147, "grad_norm": 0.0753268450498581, "global_step": 7756, "epoch": 73, "lr": 9.480709100257477e-05} +{"train_loss": 0.02002602629363537, "grad_norm": 0.12349475175142288, "global_step": 7757, "epoch": 73, "lr": 9.480577569972465e-05} +{"train_loss": 0.018498435616493225, "grad_norm": 0.07530547678470612, "global_step": 7758, "epoch": 73, "lr": 9.480446023944654e-05} +{"train_loss": 0.015657074749469757, "grad_norm": 0.07878387719392776, "global_step": 7759, "epoch": 73, "lr": 9.480314462174502e-05} +{"train_loss": 0.018376220017671585, "grad_norm": 0.0954999029636383, "global_step": 7760, "epoch": 73, "lr": 9.480182884662475e-05} +{"train_loss": 0.020301971584558487, "grad_norm": 0.06776285916566849, "global_step": 7761, "epoch": 73, "lr": 9.480051291409032e-05} +{"train_loss": 0.018874391913414, "grad_norm": 0.0930662751197815, "global_step": 7762, "epoch": 73, "lr": 9.479919682414639e-05} +{"train_loss": 0.018696872517466545, "grad_norm": 0.056242477148771286, "global_step": 7763, "epoch": 73, "lr": 9.479788057679755e-05} +{"train_loss": 0.02089586667716503, "grad_norm": 0.0921880453824997, "global_step": 7764, "epoch": 73, "lr": 9.479656417204845e-05} +{"train_loss": 0.016531813889741898, "grad_norm": 0.05570722371339798, "global_step": 7765, "epoch": 73, "lr": 9.479524760990371e-05} +{"train_loss": 0.02006702870130539, "grad_norm": 0.044263023883104324, "global_step": 7766, "epoch": 73, "lr": 9.479393089036792e-05} +{"train_loss": 0.015627097338438034, "grad_norm": 0.05288389325141907, "global_step": 7767, "epoch": 73, "lr": 9.479261401344577e-05} +{"train_loss": 0.0176580548286438, "grad_norm": 0.05035121366381645, "global_step": 7768, "epoch": 73, "lr": 9.479129697914185e-05} +{"train_loss": 0.01996382139623165, "grad_norm": 0.06869649142026901, "global_step": 7769, "epoch": 73, "lr": 9.478997978746078e-05} +{"train_loss": 0.01800990104675293, "grad_norm": 0.0561053529381752, "global_step": 7770, "epoch": 73, "lr": 9.478866243840719e-05} +{"train_loss": 0.01822766661643982, "grad_norm": 0.08210020512342453, "global_step": 7771, "epoch": 73, "lr": 9.478734493198574e-05} +{"train_loss": 0.01844683662056923, "grad_norm": 0.057548295706510544, "global_step": 7772, "epoch": 73, "lr": 9.478602726820103e-05} +{"train_loss": 0.01710658147931099, "grad_norm": 0.11437094956636429, "global_step": 7773, "epoch": 73, "lr": 9.478470944705769e-05} +{"train_loss": 0.01852099969983101, "grad_norm": 0.1355162411928177, "global_step": 7774, "epoch": 73, "lr": 9.478339146856037e-05} +{"train_loss": 0.01443480420857668, "grad_norm": 0.07423275709152222, "global_step": 7775, "epoch": 73, "lr": 9.478207333271368e-05} +{"train_loss": 0.01479613408446312, "grad_norm": 0.03535803034901619, "global_step": 7776, "epoch": 73, "lr": 9.478075503952226e-05} +{"train_loss": 0.019393838942050934, "grad_norm": 0.06881385296583176, "global_step": 7777, "epoch": 73, "lr": 9.477943658899076e-05} +{"train_loss": 0.01703285612165928, "grad_norm": 0.05315900221467018, "global_step": 7778, "epoch": 73, "lr": 9.477811798112376e-05} +{"train_loss": 0.016630031168460846, "grad_norm": 0.05737332999706268, "global_step": 7779, "epoch": 73, "lr": 9.477679921592594e-05} +{"train_loss": 0.01785782352089882, "grad_norm": 0.11415006965398788, "global_step": 7780, "epoch": 73, "lr": 9.477548029340193e-05} +{"train_loss": 0.01845472678542137, "grad_norm": 0.062204111367464066, "global_step": 7781, "epoch": 73, "lr": 9.477416121355635e-05} +{"train_loss": 0.020224906504154205, "grad_norm": 0.06841793656349182, "global_step": 7782, "epoch": 73, "lr": 9.477284197639385e-05} +{"train_loss": 0.01653973013162613, "grad_norm": 0.07446511089801788, "global_step": 7783, "epoch": 73, "lr": 9.477152258191904e-05} +{"train_loss": 0.01890602707862854, "grad_norm": 0.05392683669924736, "global_step": 7784, "epoch": 73, "lr": 9.477020303013657e-05} +{"train_loss": 0.018650293350219727, "grad_norm": 0.06816399097442627, "global_step": 7785, "epoch": 73, "lr": 9.476888332105107e-05} +{"train_loss": 0.0165339857339859, "grad_norm": 0.04324532672762871, "global_step": 7786, "epoch": 73, "lr": 9.47675634546672e-05} +{"train_loss": 0.017034918069839478, "grad_norm": 0.04583899304270744, "global_step": 7787, "epoch": 73, "lr": 9.476624343098958e-05} +{"train_loss": 0.018421035259962082, "grad_norm": 0.14170078933238983, "global_step": 7788, "epoch": 73, "lr": 9.476492325002284e-05} +{"train_loss": 0.019595803692936897, "grad_norm": 0.058056171983480453, "global_step": 7789, "epoch": 73, "lr": 9.476360291177163e-05} +{"train_loss": 0.01921910047531128, "grad_norm": 0.07912642508745193, "global_step": 7790, "epoch": 73, "lr": 9.476228241624059e-05} +{"train_loss": 0.020944278687238693, "grad_norm": 0.049617331475019455, "global_step": 7791, "epoch": 73, "lr": 9.476096176343435e-05} +{"train_loss": 0.015504940412938595, "grad_norm": 0.1031901091337204, "global_step": 7792, "epoch": 73, "lr": 9.475964095335756e-05} +{"train_loss": 0.01895267143845558, "grad_norm": 0.05134754255414009, "global_step": 7793, "epoch": 73, "lr": 9.475831998601485e-05} +{"train_loss": 0.0179718267172575, "grad_norm": 0.053943734616041183, "global_step": 7794, "epoch": 73, "lr": 9.475699886141089e-05} +{"train_loss": 0.018638145178556442, "grad_norm": 0.07566093653440475, "global_step": 7795, "epoch": 73, "lr": 9.475567757955029e-05} +{"train_loss": 0.020241612568497658, "grad_norm": 0.097417451441288, "global_step": 7796, "epoch": 73, "lr": 9.475435614043768e-05} +{"train_loss": 0.017054975032806396, "grad_norm": 0.06944837421178818, "global_step": 7797, "epoch": 73, "lr": 9.475303454407775e-05} +{"train_loss": 0.01737803965806961, "grad_norm": 0.08895391970872879, "global_step": 7798, "epoch": 73, "lr": 9.475171279047512e-05} +{"train_loss": 0.017445240169763565, "grad_norm": 0.13588353991508484, "global_step": 7799, "epoch": 73, "lr": 9.475039087963442e-05} +{"train_loss": 0.02031286060810089, "grad_norm": 0.0780830979347229, "global_step": 7800, "epoch": 73, "lr": 9.474906881156032e-05} +{"train_loss": 0.017275018617510796, "grad_norm": 0.07721764594316483, "global_step": 7801, "epoch": 73, "lr": 9.474774658625745e-05} +{"train_loss": 0.014608247205615044, "grad_norm": 0.07886306941509247, "global_step": 7802, "epoch": 73, "lr": 9.474642420373046e-05} +{"train_loss": 0.016071366146206856, "grad_norm": 0.0646699070930481, "global_step": 7803, "epoch": 73, "lr": 9.474510166398398e-05} +{"train_loss": 0.018583396449685097, "grad_norm": 0.09073175489902496, "global_step": 7804, "epoch": 73, "lr": 9.474377896702269e-05} +{"train_loss": 0.018008220940828323, "grad_norm": 0.0682155042886734, "global_step": 7805, "epoch": 73, "lr": 9.47424561128512e-05} +{"train_loss": 0.018155651167035103, "grad_norm": 0.10842312127351761, "global_step": 7806, "epoch": 73, "lr": 9.47411331014742e-05} +{"train_loss": 0.01880929060280323, "grad_norm": 0.07157517224550247, "global_step": 7807, "epoch": 73, "lr": 9.47398099328963e-05} +{"train_loss": 0.01965463161468506, "grad_norm": 0.06670460104942322, "global_step": 7808, "epoch": 73, "lr": 9.473848660712216e-05} +{"train_loss": 0.020492883399128914, "grad_norm": 0.07480437308549881, "global_step": 7809, "epoch": 73, "lr": 9.473716312415646e-05} +{"train_loss": 0.016918238252401352, "grad_norm": 0.04768362641334534, "global_step": 7810, "epoch": 73, "lr": 9.47358394840038e-05} +{"train_loss": 0.018417788669466972, "grad_norm": 0.08340831845998764, "global_step": 7811, "epoch": 73, "lr": 9.473451568666887e-05} +{"train_loss": 0.019350247457623482, "grad_norm": 0.0612342432141304, "global_step": 7812, "epoch": 73, "lr": 9.47331917321563e-05} +{"train_loss": 0.017237532883882523, "grad_norm": 0.05670271813869476, "global_step": 7813, "epoch": 73, "lr": 9.473186762047075e-05} +{"train_loss": 0.019430607557296753, "grad_norm": 0.0794982835650444, "global_step": 7814, "epoch": 73, "lr": 9.473054335161687e-05} +{"train_loss": 0.018206201493740082, "grad_norm": 0.06495697051286697, "global_step": 7815, "epoch": 73, "lr": 9.472921892559932e-05} +{"train_loss": 0.017815500497817993, "grad_norm": 0.09067830443382263, "global_step": 7816, "epoch": 73, "lr": 9.472789434242273e-05} +{"train_loss": 0.017097419127821922, "grad_norm": 0.05332854017615318, "global_step": 7817, "epoch": 73, "lr": 9.472656960209179e-05} +{"train_loss": 0.018338728696107864, "grad_norm": 0.05452612042427063, "global_step": 7818, "epoch": 73, "lr": 9.472524470461112e-05} +{"train_loss": 0.01878552883863449, "grad_norm": 0.09062957018613815, "global_step": 7819, "epoch": 73, "lr": 9.472391964998539e-05} +{"train_loss": 0.01858283206820488, "grad_norm": 0.06153713911771774, "global_step": 7820, "epoch": 73, "lr": 9.472259443821928e-05} +{"train_loss": 0.01956181600689888, "grad_norm": 0.04716336354613304, "global_step": 7821, "epoch": 73, "lr": 9.47212690693174e-05} +{"train_loss": 0.0174286849796772, "grad_norm": 0.06301212310791016, "global_step": 7822, "epoch": 73, "lr": 9.471994354328442e-05} +{"train_loss": 0.02058723196387291, "grad_norm": 0.13164371252059937, "global_step": 7823, "epoch": 73, "lr": 9.471861786012503e-05} +{"train_loss": 0.01913607306778431, "grad_norm": 0.07456064969301224, "global_step": 7824, "epoch": 73, "lr": 9.471729201984385e-05} +{"train_loss": 0.020706452429294586, "grad_norm": 0.07104047387838364, "global_step": 7825, "epoch": 73, "lr": 9.471596602244555e-05} +{"train_loss": 0.020588085055351257, "grad_norm": 0.10516860336065292, "global_step": 7826, "epoch": 73, "lr": 9.47146398679348e-05} +{"train_loss": 0.021587729454040527, "grad_norm": 0.06140096113085747, "global_step": 7827, "epoch": 73, "lr": 9.471331355631626e-05} +{"train_loss": 0.015611594542860985, "grad_norm": 0.043716173619031906, "global_step": 7828, "epoch": 73, "lr": 9.471198708759457e-05} +{"train_loss": 0.020702047273516655, "grad_norm": 0.07130618393421173, "global_step": 7829, "epoch": 73, "lr": 9.471066046177442e-05} +{"train_loss": 0.018219372257590294, "grad_norm": 0.054893240332603455, "global_step": 7830, "epoch": 73, "lr": 9.470933367886044e-05} +{"train_loss": 0.02030022442340851, "grad_norm": 0.07564257085323334, "global_step": 7831, "epoch": 73, "lr": 9.470800673885729e-05} +{"train_loss": 0.018857939168810844, "grad_norm": 0.07218720018863678, "global_step": 7832, "epoch": 73, "lr": 9.470667964176967e-05} +{"train_loss": 0.0197049118578434, "grad_norm": 0.1485746204853058, "global_step": 7833, "epoch": 73, "lr": 9.470535238760221e-05} +{"train_loss": 0.019090769812464714, "grad_norm": 0.05545971542596817, "global_step": 7834, "epoch": 73, "lr": 9.470402497635957e-05} +{"train_loss": 0.01923997886478901, "grad_norm": 0.14019525051116943, "global_step": 7835, "epoch": 73, "lr": 9.470269740804645e-05} +{"train_loss": 0.01903907209634781, "grad_norm": 0.0829503983259201, "global_step": 7836, "epoch": 73, "lr": 9.470136968266748e-05} +{"train_loss": 0.017840053886175156, "grad_norm": 0.04766940325498581, "global_step": 7837, "epoch": 73, "lr": 9.470004180022734e-05} +{"train_loss": 0.020162764936685562, "grad_norm": 0.09005239605903625, "global_step": 7838, "epoch": 73, "lr": 9.469871376073069e-05} +{"train_loss": 0.017450960353016853, "grad_norm": 0.043062321841716766, "global_step": 7839, "epoch": 73, "lr": 9.469738556418222e-05} +{"train_loss": 0.01556827686727047, "grad_norm": 0.05907783657312393, "global_step": 7840, "epoch": 73, "lr": 9.469605721058656e-05} +{"train_loss": 0.01721605658531189, "grad_norm": 0.04607417434453964, "global_step": 7841, "epoch": 73, "lr": 9.469472869994838e-05} +{"train_loss": 0.01996275968849659, "grad_norm": 0.07667622715234756, "global_step": 7842, "epoch": 73, "lr": 9.469340003227237e-05} +{"train_loss": 0.01830606062386958, "grad_norm": 0.08734684437513351, "global_step": 7843, "epoch": 73, "lr": 9.46920712075632e-05, "val_loss": 0.018542125821113586} +{"train_loss": 0.018566934391856194, "grad_norm": 0.06653156131505966, "global_step": 7844, "epoch": 74, "lr": 9.469074222582551e-05} +{"train_loss": 0.021769119426608086, "grad_norm": 0.11066871136426926, "global_step": 7845, "epoch": 74, "lr": 9.4689413087064e-05} +{"train_loss": 0.01800835132598877, "grad_norm": 0.051498088985681534, "global_step": 7846, "epoch": 74, "lr": 9.468808379128332e-05} +{"train_loss": 0.01874256506562233, "grad_norm": 0.16579918563365936, "global_step": 7847, "epoch": 74, "lr": 9.468675433848815e-05} +{"train_loss": 0.01817617192864418, "grad_norm": 0.07878050953149796, "global_step": 7848, "epoch": 74, "lr": 9.468542472868316e-05} +{"train_loss": 0.018801704049110413, "grad_norm": 0.08185514062643051, "global_step": 7849, "epoch": 74, "lr": 9.468409496187303e-05} +{"train_loss": 0.01982072740793228, "grad_norm": 0.04676447808742523, "global_step": 7850, "epoch": 74, "lr": 9.46827650380624e-05} +{"train_loss": 0.018154051154851913, "grad_norm": 0.1229977235198021, "global_step": 7851, "epoch": 74, "lr": 9.468143495725599e-05} +{"train_loss": 0.017528174445033073, "grad_norm": 0.09396246075630188, "global_step": 7852, "epoch": 74, "lr": 9.468010471945843e-05} +{"train_loss": 0.019115805625915527, "grad_norm": 0.08244988322257996, "global_step": 7853, "epoch": 74, "lr": 9.467877432467442e-05} +{"train_loss": 0.019159391522407532, "grad_norm": 0.1006661057472229, "global_step": 7854, "epoch": 74, "lr": 9.467744377290864e-05} +{"train_loss": 0.01894718036055565, "grad_norm": 0.05675981938838959, "global_step": 7855, "epoch": 74, "lr": 9.467611306416572e-05} +{"train_loss": 0.01589026302099228, "grad_norm": 0.05017229914665222, "global_step": 7856, "epoch": 74, "lr": 9.46747821984504e-05} +{"train_loss": 0.015908394008874893, "grad_norm": 0.10334762930870056, "global_step": 7857, "epoch": 74, "lr": 9.467345117576731e-05} +{"train_loss": 0.02100497856736183, "grad_norm": 0.09999212622642517, "global_step": 7858, "epoch": 74, "lr": 9.467211999612114e-05} +{"train_loss": 0.01789320819079876, "grad_norm": 0.04167065769433975, "global_step": 7859, "epoch": 74, "lr": 9.467078865951657e-05} +{"train_loss": 0.015388148836791515, "grad_norm": 0.03662712499499321, "global_step": 7860, "epoch": 74, "lr": 9.466945716595827e-05} +{"train_loss": 0.01922549307346344, "grad_norm": 0.04957538843154907, "global_step": 7861, "epoch": 74, "lr": 9.466812551545095e-05} +{"train_loss": 0.019014641642570496, "grad_norm": 0.07048913836479187, "global_step": 7862, "epoch": 74, "lr": 9.466679370799923e-05} +{"train_loss": 0.017204968258738518, "grad_norm": 0.07395070046186447, "global_step": 7863, "epoch": 74, "lr": 9.466546174360784e-05} +{"train_loss": 0.019273193553090096, "grad_norm": 0.05474376678466797, "global_step": 7864, "epoch": 74, "lr": 9.466412962228144e-05} +{"train_loss": 0.01801883615553379, "grad_norm": 0.051468536257743835, "global_step": 7865, "epoch": 74, "lr": 9.46627973440247e-05} +{"train_loss": 0.01703602820634842, "grad_norm": 0.08712445199489594, "global_step": 7866, "epoch": 74, "lr": 9.466146490884232e-05} +{"train_loss": 0.016437899321317673, "grad_norm": 0.05082770809531212, "global_step": 7867, "epoch": 74, "lr": 9.466013231673898e-05} +{"train_loss": 0.018216855823993683, "grad_norm": 0.042277056723833084, "global_step": 7868, "epoch": 74, "lr": 9.465879956771935e-05} +{"train_loss": 0.018680142238736153, "grad_norm": 0.0689789280295372, "global_step": 7869, "epoch": 74, "lr": 9.465746666178814e-05} +{"train_loss": 0.017278030514717102, "grad_norm": 0.04159845411777496, "global_step": 7870, "epoch": 74, "lr": 9.465613359894999e-05} +{"train_loss": 0.017499826848506927, "grad_norm": 0.03917841985821724, "global_step": 7871, "epoch": 74, "lr": 9.465480037920962e-05} +{"train_loss": 0.02496921271085739, "grad_norm": 0.12060749530792236, "global_step": 7872, "epoch": 74, "lr": 9.46534670025717e-05} +{"train_loss": 0.01822151616215706, "grad_norm": 0.08572991192340851, "global_step": 7873, "epoch": 74, "lr": 9.465213346904089e-05} +{"train_loss": 0.01642414927482605, "grad_norm": 0.0461830236017704, "global_step": 7874, "epoch": 74, "lr": 9.465079977862193e-05} +{"train_loss": 0.01587606407701969, "grad_norm": 0.06650484353303909, "global_step": 7875, "epoch": 74, "lr": 9.464946593131947e-05} +{"train_loss": 0.01695508323609829, "grad_norm": 0.05155850201845169, "global_step": 7876, "epoch": 74, "lr": 9.464813192713822e-05} +{"train_loss": 0.016961034387350082, "grad_norm": 0.044815290719270706, "global_step": 7877, "epoch": 74, "lr": 9.464679776608282e-05} +{"train_loss": 0.019875656813383102, "grad_norm": 0.05244462564587593, "global_step": 7878, "epoch": 74, "lr": 9.464546344815801e-05} +{"train_loss": 0.01904354989528656, "grad_norm": 0.07299523055553436, "global_step": 7879, "epoch": 74, "lr": 9.464412897336845e-05} +{"train_loss": 0.021007541567087173, "grad_norm": 0.1127934381365776, "global_step": 7880, "epoch": 74, "lr": 9.464279434171886e-05} +{"train_loss": 0.020040512084960938, "grad_norm": 0.07212857902050018, "global_step": 7881, "epoch": 74, "lr": 9.464145955321387e-05} +{"train_loss": 0.01923326775431633, "grad_norm": 0.08109858632087708, "global_step": 7882, "epoch": 74, "lr": 9.464012460785823e-05} +{"train_loss": 0.018714603036642075, "grad_norm": 0.04442014917731285, "global_step": 7883, "epoch": 74, "lr": 9.46387895056566e-05} +{"train_loss": 0.01857738196849823, "grad_norm": 0.06483745574951172, "global_step": 7884, "epoch": 74, "lr": 9.463745424661365e-05} +{"train_loss": 0.016398049890995026, "grad_norm": 0.09629815816879272, "global_step": 7885, "epoch": 74, "lr": 9.463611883073413e-05} +{"train_loss": 0.02361515909433365, "grad_norm": 0.08688708394765854, "global_step": 7886, "epoch": 74, "lr": 9.463478325802268e-05} +{"train_loss": 0.01705102063715458, "grad_norm": 0.056881971657276154, "global_step": 7887, "epoch": 74, "lr": 9.463344752848402e-05} +{"train_loss": 0.019219359382987022, "grad_norm": 0.1057715117931366, "global_step": 7888, "epoch": 74, "lr": 9.463211164212284e-05} +{"train_loss": 0.017936505377292633, "grad_norm": 0.0808517262339592, "global_step": 7889, "epoch": 74, "lr": 9.463077559894384e-05} +{"train_loss": 0.020954661071300507, "grad_norm": 0.06669522821903229, "global_step": 7890, "epoch": 74, "lr": 9.462943939895169e-05} +{"train_loss": 0.021975062787532806, "grad_norm": 0.0934242531657219, "global_step": 7891, "epoch": 74, "lr": 9.462810304215111e-05} +{"train_loss": 0.01632896065711975, "grad_norm": 0.049696143716573715, "global_step": 7892, "epoch": 74, "lr": 9.462676652854676e-05} +{"train_loss": 0.01700034737586975, "grad_norm": 0.0756494551897049, "global_step": 7893, "epoch": 74, "lr": 9.46254298581434e-05} +{"train_loss": 0.017361655831336975, "grad_norm": 0.06533636152744293, "global_step": 7894, "epoch": 74, "lr": 9.462409303094565e-05} +{"train_loss": 0.017549047246575356, "grad_norm": 0.13098129630088806, "global_step": 7895, "epoch": 74, "lr": 9.462275604695825e-05} +{"train_loss": 0.02022075466811657, "grad_norm": 0.04241160303354263, "global_step": 7896, "epoch": 74, "lr": 9.46214189061859e-05} +{"train_loss": 0.0184965580701828, "grad_norm": 0.06049155443906784, "global_step": 7897, "epoch": 74, "lr": 9.462008160863329e-05} +{"train_loss": 0.01765787973999977, "grad_norm": 0.03940384089946747, "global_step": 7898, "epoch": 74, "lr": 9.461874415430512e-05} +{"train_loss": 0.017222262918949127, "grad_norm": 0.04598120227456093, "global_step": 7899, "epoch": 74, "lr": 9.461740654320607e-05} +{"train_loss": 0.017835818231105804, "grad_norm": 0.10687005519866943, "global_step": 7900, "epoch": 74, "lr": 9.461606877534087e-05} +{"train_loss": 0.02003692463040352, "grad_norm": 0.05211779102683067, "global_step": 7901, "epoch": 74, "lr": 9.46147308507142e-05} +{"train_loss": 0.01893538236618042, "grad_norm": 0.056594107300043106, "global_step": 7902, "epoch": 74, "lr": 9.461339276933078e-05} +{"train_loss": 0.01792539283633232, "grad_norm": 0.05478762835264206, "global_step": 7903, "epoch": 74, "lr": 9.46120545311953e-05} +{"train_loss": 0.013734899461269379, "grad_norm": 0.07584390044212341, "global_step": 7904, "epoch": 74, "lr": 9.461071613631245e-05} +{"train_loss": 0.017466951161623, "grad_norm": 0.0565858855843544, "global_step": 7905, "epoch": 74, "lr": 9.460937758468695e-05} +{"train_loss": 0.02000855654478073, "grad_norm": 0.11195754259824753, "global_step": 7906, "epoch": 74, "lr": 9.46080388763235e-05} +{"train_loss": 0.016357241198420525, "grad_norm": 0.09060998260974884, "global_step": 7907, "epoch": 74, "lr": 9.46067000112268e-05} +{"train_loss": 0.018214087933301926, "grad_norm": 0.06778574734926224, "global_step": 7908, "epoch": 74, "lr": 9.460536098940154e-05} +{"train_loss": 0.017428046092391014, "grad_norm": 0.060649324208498, "global_step": 7909, "epoch": 74, "lr": 9.460402181085246e-05} +{"train_loss": 0.018506770953536034, "grad_norm": 0.11281462758779526, "global_step": 7910, "epoch": 74, "lr": 9.460268247558424e-05} +{"train_loss": 0.01600697636604309, "grad_norm": 0.07275306433439255, "global_step": 7911, "epoch": 74, "lr": 9.460134298360159e-05} +{"train_loss": 0.016625145450234413, "grad_norm": 0.07687656581401825, "global_step": 7912, "epoch": 74, "lr": 9.460000333490922e-05} +{"train_loss": 0.020199373364448547, "grad_norm": 0.09748366475105286, "global_step": 7913, "epoch": 74, "lr": 9.459866352951182e-05} +{"train_loss": 0.018679827451705933, "grad_norm": 0.049978069961071014, "global_step": 7914, "epoch": 74, "lr": 9.459732356741413e-05} +{"train_loss": 0.01757405512034893, "grad_norm": 0.09754763543605804, "global_step": 7915, "epoch": 74, "lr": 9.459598344862082e-05} +{"train_loss": 0.021297991275787354, "grad_norm": 0.0819799080491066, "global_step": 7916, "epoch": 74, "lr": 9.459464317313664e-05} +{"train_loss": 0.021001864224672318, "grad_norm": 0.07038503140211105, "global_step": 7917, "epoch": 74, "lr": 9.459330274096627e-05} +{"train_loss": 0.01529812440276146, "grad_norm": 0.05060864984989166, "global_step": 7918, "epoch": 74, "lr": 9.459196215211443e-05} +{"train_loss": 0.01884584128856659, "grad_norm": 0.08641789853572845, "global_step": 7919, "epoch": 74, "lr": 9.459062140658582e-05} +{"train_loss": 0.01586850732564926, "grad_norm": 0.0732373595237732, "global_step": 7920, "epoch": 74, "lr": 9.458928050438515e-05} +{"train_loss": 0.01996094547212124, "grad_norm": 0.0441494844853878, "global_step": 7921, "epoch": 74, "lr": 9.458793944551715e-05} +{"train_loss": 0.017480425536632538, "grad_norm": 0.05704357102513313, "global_step": 7922, "epoch": 74, "lr": 9.458659822998652e-05} +{"train_loss": 0.01879344880580902, "grad_norm": 0.11623463779687881, "global_step": 7923, "epoch": 74, "lr": 9.458525685779797e-05} +{"train_loss": 0.01826375722885132, "grad_norm": 0.03991701081395149, "global_step": 7924, "epoch": 74, "lr": 9.458391532895624e-05} +{"train_loss": 0.015278765931725502, "grad_norm": 0.03791766241192818, "global_step": 7925, "epoch": 74, "lr": 9.4582573643466e-05} +{"train_loss": 0.021164022386074066, "grad_norm": 0.050510160624980927, "global_step": 7926, "epoch": 74, "lr": 9.458123180133199e-05} +{"train_loss": 0.01737864501774311, "grad_norm": 0.04700343683362007, "global_step": 7927, "epoch": 74, "lr": 9.457988980255892e-05} +{"train_loss": 0.016124360263347626, "grad_norm": 0.06031263247132301, "global_step": 7928, "epoch": 74, "lr": 9.45785476471515e-05} +{"train_loss": 0.017277460545301437, "grad_norm": 0.06810273975133896, "global_step": 7929, "epoch": 74, "lr": 9.457720533511444e-05} +{"train_loss": 0.017908671870827675, "grad_norm": 0.055271703749895096, "global_step": 7930, "epoch": 74, "lr": 9.45758628664525e-05} +{"train_loss": 0.020502720028162003, "grad_norm": 0.04815554618835449, "global_step": 7931, "epoch": 74, "lr": 9.457452024117034e-05} +{"train_loss": 0.017787007614970207, "grad_norm": 0.056089773774147034, "global_step": 7932, "epoch": 74, "lr": 9.45731774592727e-05} +{"train_loss": 0.0194525346159935, "grad_norm": 0.10073699802160263, "global_step": 7933, "epoch": 74, "lr": 9.45718345207643e-05} +{"train_loss": 0.016238288953900337, "grad_norm": 0.056264687329530716, "global_step": 7934, "epoch": 74, "lr": 9.457049142564986e-05} +{"train_loss": 0.016242213547229767, "grad_norm": 0.11384907364845276, "global_step": 7935, "epoch": 74, "lr": 9.45691481739341e-05} +{"train_loss": 0.021247945725917816, "grad_norm": 0.050351332873106, "global_step": 7936, "epoch": 74, "lr": 9.456780476562173e-05} +{"train_loss": 0.017655275762081146, "grad_norm": 0.0843716412782669, "global_step": 7937, "epoch": 74, "lr": 9.456646120071748e-05} +{"train_loss": 0.018910914659500122, "grad_norm": 0.11462458223104477, "global_step": 7938, "epoch": 74, "lr": 9.456511747922607e-05} +{"train_loss": 0.01929311454296112, "grad_norm": 0.10281645506620407, "global_step": 7939, "epoch": 74, "lr": 9.456377360115221e-05} +{"train_loss": 0.018145855516195297, "grad_norm": 0.08072422444820404, "global_step": 7940, "epoch": 74, "lr": 9.456242956650064e-05} +{"train_loss": 0.014531731605529785, "grad_norm": 0.06665660440921783, "global_step": 7941, "epoch": 74, "lr": 9.456108537527605e-05} +{"train_loss": 0.017683744430541992, "grad_norm": 0.15785570442676544, "global_step": 7942, "epoch": 74, "lr": 9.45597410274832e-05} +{"train_loss": 0.020171038806438446, "grad_norm": 0.10717002302408218, "global_step": 7943, "epoch": 74, "lr": 9.45583965231268e-05} +{"train_loss": 0.018757615238428116, "grad_norm": 0.14017997682094574, "global_step": 7944, "epoch": 74, "lr": 9.455705186221157e-05} +{"train_loss": 0.016757231205701828, "grad_norm": 0.0984848216176033, "global_step": 7945, "epoch": 74, "lr": 9.455570704474223e-05} +{"train_loss": 0.018233101814985275, "grad_norm": 0.04648875817656517, "global_step": 7946, "epoch": 74, "lr": 9.455436207072353e-05} +{"train_loss": 0.017700523138046265, "grad_norm": 0.08722559362649918, "global_step": 7947, "epoch": 74, "lr": 9.455301694016017e-05} +{"train_loss": 0.017608823254704475, "grad_norm": 0.07506129145622253, "global_step": 7948, "epoch": 74, "lr": 9.455167165305689e-05} +{"train_loss": 0.01827814587268908, "grad_norm": 0.18762989342212677, "global_step": 7949, "epoch": 74, "lr": 9.45503262094184e-05, "val_loss": 0.01915058307349682} +{"train_loss": 0.017722483724355698, "grad_norm": 0.051528315991163254, "global_step": 7950, "epoch": 75, "lr": 9.454898060924944e-05} +{"train_loss": 0.016809049993753433, "grad_norm": 0.05651551112532616, "global_step": 7951, "epoch": 75, "lr": 9.454763485255474e-05} +{"train_loss": 0.01783105358481407, "grad_norm": 0.041641831398010254, "global_step": 7952, "epoch": 75, "lr": 9.4546288939339e-05} +{"train_loss": 0.021532833576202393, "grad_norm": 0.15047957003116608, "global_step": 7953, "epoch": 75, "lr": 9.454494286960702e-05} +{"train_loss": 0.016064632683992386, "grad_norm": 0.1044556051492691, "global_step": 7954, "epoch": 75, "lr": 9.454359664336345e-05} +{"train_loss": 0.018837880343198776, "grad_norm": 0.08833039551973343, "global_step": 7955, "epoch": 75, "lr": 9.454225026061306e-05} +{"train_loss": 0.018643585965037346, "grad_norm": 0.08077746629714966, "global_step": 7956, "epoch": 75, "lr": 9.454090372136058e-05} +{"train_loss": 0.020743275061249733, "grad_norm": 0.07027503103017807, "global_step": 7957, "epoch": 75, "lr": 9.453955702561072e-05} +{"train_loss": 0.01784023828804493, "grad_norm": 0.045883070677518845, "global_step": 7958, "epoch": 75, "lr": 9.453821017336823e-05} +{"train_loss": 0.020932093262672424, "grad_norm": 0.07465910911560059, "global_step": 7959, "epoch": 75, "lr": 9.453686316463786e-05} +{"train_loss": 0.017468810081481934, "grad_norm": 0.06308387219905853, "global_step": 7960, "epoch": 75, "lr": 9.453551599942429e-05} +{"train_loss": 0.01839536800980568, "grad_norm": 0.05630932375788689, "global_step": 7961, "epoch": 75, "lr": 9.45341686777323e-05} +{"train_loss": 0.016340650618076324, "grad_norm": 0.09829345345497131, "global_step": 7962, "epoch": 75, "lr": 9.453282119956661e-05} +{"train_loss": 0.017765656113624573, "grad_norm": 0.07968004792928696, "global_step": 7963, "epoch": 75, "lr": 9.453147356493196e-05} +{"train_loss": 0.0186676736921072, "grad_norm": 0.08967693895101547, "global_step": 7964, "epoch": 75, "lr": 9.453012577383307e-05} +{"train_loss": 0.01960214413702488, "grad_norm": 0.09790261834859848, "global_step": 7965, "epoch": 75, "lr": 9.452877782627469e-05} +{"train_loss": 0.016664013266563416, "grad_norm": 0.12307186424732208, "global_step": 7966, "epoch": 75, "lr": 9.452742972226154e-05} +{"train_loss": 0.02048490196466446, "grad_norm": 0.04708293080329895, "global_step": 7967, "epoch": 75, "lr": 9.452608146179836e-05} +{"train_loss": 0.01847713254392147, "grad_norm": 0.07324844598770142, "global_step": 7968, "epoch": 75, "lr": 9.452473304488992e-05} +{"train_loss": 0.018389342352747917, "grad_norm": 0.06973572075366974, "global_step": 7969, "epoch": 75, "lr": 9.45233844715409e-05} +{"train_loss": 0.019765060395002365, "grad_norm": 0.06048651039600372, "global_step": 7970, "epoch": 75, "lr": 9.452203574175609e-05} +{"train_loss": 0.017799124121665955, "grad_norm": 0.1444881558418274, "global_step": 7971, "epoch": 75, "lr": 9.452068685554021e-05} +{"train_loss": 0.01726614683866501, "grad_norm": 0.11451251804828644, "global_step": 7972, "epoch": 75, "lr": 9.4519337812898e-05} +{"train_loss": 0.017929457128047943, "grad_norm": 0.04391322284936905, "global_step": 7973, "epoch": 75, "lr": 9.451798861383418e-05} +{"train_loss": 0.019128577783703804, "grad_norm": 0.10320349037647247, "global_step": 7974, "epoch": 75, "lr": 9.451663925835351e-05} +{"train_loss": 0.018198862671852112, "grad_norm": 0.11878413707017899, "global_step": 7975, "epoch": 75, "lr": 9.451528974646074e-05} +{"train_loss": 0.020411601290106773, "grad_norm": 0.05700153484940529, "global_step": 7976, "epoch": 75, "lr": 9.451394007816059e-05} +{"train_loss": 0.01866469345986843, "grad_norm": 0.07278639823198318, "global_step": 7977, "epoch": 75, "lr": 9.451259025345782e-05} +{"train_loss": 0.01730743981897831, "grad_norm": 0.044948969036340714, "global_step": 7978, "epoch": 75, "lr": 9.451124027235718e-05} +{"train_loss": 0.018242087215185165, "grad_norm": 0.06853313744068146, "global_step": 7979, "epoch": 75, "lr": 9.450989013486337e-05} +{"train_loss": 0.013979672454297543, "grad_norm": 0.07186102867126465, "global_step": 7980, "epoch": 75, "lr": 9.450853984098119e-05} +{"train_loss": 0.01781011000275612, "grad_norm": 0.05304930359125137, "global_step": 7981, "epoch": 75, "lr": 9.450718939071534e-05} +{"train_loss": 0.018216796219348907, "grad_norm": 0.0626056119799614, "global_step": 7982, "epoch": 75, "lr": 9.450583878407059e-05} +{"train_loss": 0.019594723358750343, "grad_norm": 0.056995976716279984, "global_step": 7983, "epoch": 75, "lr": 9.450448802105167e-05} +{"train_loss": 0.01896785944700241, "grad_norm": 0.06618006527423859, "global_step": 7984, "epoch": 75, "lr": 9.450313710166334e-05} +{"train_loss": 0.017846642062067986, "grad_norm": 0.06442499160766602, "global_step": 7985, "epoch": 75, "lr": 9.450178602591034e-05} +{"train_loss": 0.01869715005159378, "grad_norm": 0.04929369315505028, "global_step": 7986, "epoch": 75, "lr": 9.450043479379742e-05} +{"train_loss": 0.01737338863313198, "grad_norm": 0.05104398727416992, "global_step": 7987, "epoch": 75, "lr": 9.449908340532932e-05} +{"train_loss": 0.02043789252638817, "grad_norm": 0.08952509611845016, "global_step": 7988, "epoch": 75, "lr": 9.449773186051079e-05} +{"train_loss": 0.016365066170692444, "grad_norm": 0.05794178694486618, "global_step": 7989, "epoch": 75, "lr": 9.449638015934659e-05} +{"train_loss": 0.017462056130170822, "grad_norm": 0.08666634559631348, "global_step": 7990, "epoch": 75, "lr": 9.449502830184146e-05} +{"train_loss": 0.0189921036362648, "grad_norm": 0.04513914883136749, "global_step": 7991, "epoch": 75, "lr": 9.449367628800015e-05} +{"train_loss": 0.01632373407483101, "grad_norm": 0.05474985018372536, "global_step": 7992, "epoch": 75, "lr": 9.44923241178274e-05} +{"train_loss": 0.019013971090316772, "grad_norm": 0.15283887088298798, "global_step": 7993, "epoch": 75, "lr": 9.4490971791328e-05} +{"train_loss": 0.016482722014188766, "grad_norm": 0.1650271713733673, "global_step": 7994, "epoch": 75, "lr": 9.448961930850664e-05} +{"train_loss": 0.020982861518859863, "grad_norm": 0.08962804079055786, "global_step": 7995, "epoch": 75, "lr": 9.448826666936811e-05} +{"train_loss": 0.017321908846497536, "grad_norm": 0.055635906755924225, "global_step": 7996, "epoch": 75, "lr": 9.448691387391719e-05} +{"train_loss": 0.018783072009682655, "grad_norm": 0.10719749331474304, "global_step": 7997, "epoch": 75, "lr": 9.448556092215858e-05} +{"train_loss": 0.02058316394686699, "grad_norm": 0.09372426569461823, "global_step": 7998, "epoch": 75, "lr": 9.448420781409707e-05} +{"train_loss": 0.018273042514920235, "grad_norm": 0.06909451633691788, "global_step": 7999, "epoch": 75, "lr": 9.448285454973738e-05} +{"train_loss": 0.018472928553819656, "grad_norm": 0.07956184446811676, "global_step": 8000, "epoch": 75, "lr": 9.448150112908427e-05} +{"train_loss": 0.018483245745301247, "grad_norm": 0.1054559201002121, "global_step": 8001, "epoch": 75, "lr": 9.448014755214255e-05} +{"train_loss": 0.01693980023264885, "grad_norm": 0.12468380481004715, "global_step": 8002, "epoch": 75, "lr": 9.447879381891692e-05} +{"train_loss": 0.01798497885465622, "grad_norm": 0.08715613186359406, "global_step": 8003, "epoch": 75, "lr": 9.447743992941214e-05} +{"train_loss": 0.017120491713285446, "grad_norm": 0.06974749267101288, "global_step": 8004, "epoch": 75, "lr": 9.447608588363296e-05} +{"train_loss": 0.01850443333387375, "grad_norm": 0.09176867455244064, "global_step": 8005, "epoch": 75, "lr": 9.447473168158421e-05} +{"train_loss": 0.01757603883743286, "grad_norm": 0.0739150196313858, "global_step": 8006, "epoch": 75, "lr": 9.447337732327056e-05} +{"train_loss": 0.01911396160721779, "grad_norm": 0.11637677997350693, "global_step": 8007, "epoch": 75, "lr": 9.44720228086968e-05} +{"train_loss": 0.0171004980802536, "grad_norm": 0.07578112930059433, "global_step": 8008, "epoch": 75, "lr": 9.447066813786771e-05} +{"train_loss": 0.01632489636540413, "grad_norm": 0.1103505939245224, "global_step": 8009, "epoch": 75, "lr": 9.446931331078802e-05} +{"train_loss": 0.01882270723581314, "grad_norm": 0.08118010312318802, "global_step": 8010, "epoch": 75, "lr": 9.44679583274625e-05} +{"train_loss": 0.020529218018054962, "grad_norm": 0.109037384390831, "global_step": 8011, "epoch": 75, "lr": 9.446660318789593e-05} +{"train_loss": 0.01785443350672722, "grad_norm": 0.061840977519750595, "global_step": 8012, "epoch": 75, "lr": 9.446524789209304e-05} +{"train_loss": 0.020691467449069023, "grad_norm": 0.04733653739094734, "global_step": 8013, "epoch": 75, "lr": 9.44638924400586e-05} +{"train_loss": 0.015845786780118942, "grad_norm": 0.06225572153925896, "global_step": 8014, "epoch": 75, "lr": 9.44625368317974e-05} +{"train_loss": 0.0179001335054636, "grad_norm": 0.07587576657533646, "global_step": 8015, "epoch": 75, "lr": 9.446118106731417e-05} +{"train_loss": 0.015188492834568024, "grad_norm": 0.08930496871471405, "global_step": 8016, "epoch": 75, "lr": 9.445982514661367e-05} +{"train_loss": 0.020555950701236725, "grad_norm": 0.06384005397558212, "global_step": 8017, "epoch": 75, "lr": 9.445846906970071e-05} +{"train_loss": 0.01785176619887352, "grad_norm": 0.0410432405769825, "global_step": 8018, "epoch": 75, "lr": 9.445711283658001e-05} +{"train_loss": 0.01755186915397644, "grad_norm": 0.094316765666008, "global_step": 8019, "epoch": 75, "lr": 9.445575644725634e-05} +{"train_loss": 0.02079875022172928, "grad_norm": 0.08914481103420258, "global_step": 8020, "epoch": 75, "lr": 9.445439990173448e-05} +{"train_loss": 0.019466904923319817, "grad_norm": 0.0783439576625824, "global_step": 8021, "epoch": 75, "lr": 9.44530432000192e-05} +{"train_loss": 0.01630956493318081, "grad_norm": 0.0499383918941021, "global_step": 8022, "epoch": 75, "lr": 9.445168634211526e-05} +{"train_loss": 0.018746349960565567, "grad_norm": 0.11372001469135284, "global_step": 8023, "epoch": 75, "lr": 9.445032932802742e-05} +{"train_loss": 0.0193852037191391, "grad_norm": 0.14236024022102356, "global_step": 8024, "epoch": 75, "lr": 9.444897215776045e-05} +{"train_loss": 0.01549482997506857, "grad_norm": 0.046087075024843216, "global_step": 8025, "epoch": 75, "lr": 9.444761483131912e-05} +{"train_loss": 0.01606827974319458, "grad_norm": 0.05926486477255821, "global_step": 8026, "epoch": 75, "lr": 9.44462573487082e-05} +{"train_loss": 0.01817910559475422, "grad_norm": 0.1002858579158783, "global_step": 8027, "epoch": 75, "lr": 9.444489970993248e-05} +{"train_loss": 0.020847158506512642, "grad_norm": 0.06320405006408691, "global_step": 8028, "epoch": 75, "lr": 9.444354191499669e-05} +{"train_loss": 0.017608828842639923, "grad_norm": 0.045336488634347916, "global_step": 8029, "epoch": 75, "lr": 9.444218396390563e-05} +{"train_loss": 0.018888209015130997, "grad_norm": 0.11727188527584076, "global_step": 8030, "epoch": 75, "lr": 9.444082585666407e-05} +{"train_loss": 0.017402568832039833, "grad_norm": 0.16139575839042664, "global_step": 8031, "epoch": 75, "lr": 9.443946759327677e-05} +{"train_loss": 0.019608305767178535, "grad_norm": 0.06263913959264755, "global_step": 8032, "epoch": 75, "lr": 9.443810917374848e-05} +{"train_loss": 0.019475871697068214, "grad_norm": 0.05432378500699997, "global_step": 8033, "epoch": 75, "lr": 9.443675059808404e-05} +{"train_loss": 0.0188013706356287, "grad_norm": 0.07434717565774918, "global_step": 8034, "epoch": 75, "lr": 9.443539186628818e-05} +{"train_loss": 0.016850676387548447, "grad_norm": 0.041935995221138, "global_step": 8035, "epoch": 75, "lr": 9.443403297836565e-05} +{"train_loss": 0.017174864187836647, "grad_norm": 0.06376122683286667, "global_step": 8036, "epoch": 75, "lr": 9.443267393432127e-05} +{"train_loss": 0.015965159982442856, "grad_norm": 0.09584669023752213, "global_step": 8037, "epoch": 75, "lr": 9.44313147341598e-05} +{"train_loss": 0.020186347886919975, "grad_norm": 0.09781753271818161, "global_step": 8038, "epoch": 75, "lr": 9.4429955377886e-05} +{"train_loss": 0.01699652336537838, "grad_norm": 0.06405645608901978, "global_step": 8039, "epoch": 75, "lr": 9.442859586550468e-05} +{"train_loss": 0.018966108560562134, "grad_norm": 0.062331728637218475, "global_step": 8040, "epoch": 75, "lr": 9.442723619702057e-05} +{"train_loss": 0.01775955595076084, "grad_norm": 0.06250306963920593, "global_step": 8041, "epoch": 75, "lr": 9.442587637243849e-05} +{"train_loss": 0.016195621341466904, "grad_norm": 0.07862606644630432, "global_step": 8042, "epoch": 75, "lr": 9.44245163917632e-05} +{"train_loss": 0.019936416298151016, "grad_norm": 0.07769939303398132, "global_step": 8043, "epoch": 75, "lr": 9.442315625499947e-05} +{"train_loss": 0.017560845240950584, "grad_norm": 0.09874829649925232, "global_step": 8044, "epoch": 75, "lr": 9.442179596215209e-05} +{"train_loss": 0.02164492942392826, "grad_norm": 0.09064574539661407, "global_step": 8045, "epoch": 75, "lr": 9.442043551322583e-05} +{"train_loss": 0.021131223067641258, "grad_norm": 0.041292790323495865, "global_step": 8046, "epoch": 75, "lr": 9.44190749082255e-05} +{"train_loss": 0.01656285673379898, "grad_norm": 0.09492047131061554, "global_step": 8047, "epoch": 75, "lr": 9.441771414715584e-05} +{"train_loss": 0.014876480214297771, "grad_norm": 0.06564296782016754, "global_step": 8048, "epoch": 75, "lr": 9.441635323002166e-05} +{"train_loss": 0.017431238666176796, "grad_norm": 0.04659591615200043, "global_step": 8049, "epoch": 75, "lr": 9.441499215682772e-05} +{"train_loss": 0.01942494511604309, "grad_norm": 0.17078900337219238, "global_step": 8050, "epoch": 75, "lr": 9.441363092757882e-05} +{"train_loss": 0.017949476838111877, "grad_norm": 0.0910462886095047, "global_step": 8051, "epoch": 75, "lr": 9.441226954227973e-05} +{"train_loss": 0.020266011357307434, "grad_norm": 0.09212601184844971, "global_step": 8052, "epoch": 75, "lr": 9.441090800093526e-05} +{"train_loss": 0.01749373786151409, "grad_norm": 0.06058383360505104, "global_step": 8053, "epoch": 75, "lr": 9.440954630355015e-05} +{"train_loss": 0.020392635837197304, "grad_norm": 0.06419170647859573, "global_step": 8054, "epoch": 75, "lr": 9.440818445012921e-05} +{"train_loss": 0.01826472243926716, "grad_norm": 0.08248049765825272, "global_step": 8055, "epoch": 75, "lr": 9.440682244067724e-05, "val_loss": 0.019029932096600533} +{"train_loss": 0.015581651590764523, "grad_norm": 0.06982304155826569, "global_step": 8056, "epoch": 76, "lr": 9.440546027519899e-05} +{"train_loss": 0.018769990652799606, "grad_norm": 0.11817360669374466, "global_step": 8057, "epoch": 76, "lr": 9.440409795369927e-05} +{"train_loss": 0.018170662224292755, "grad_norm": 0.08636105060577393, "global_step": 8058, "epoch": 76, "lr": 9.440273547618285e-05} +{"train_loss": 0.01947149634361267, "grad_norm": 0.07225455343723297, "global_step": 8059, "epoch": 76, "lr": 9.440137284265454e-05} +{"train_loss": 0.015124807134270668, "grad_norm": 0.06838305294513702, "global_step": 8060, "epoch": 76, "lr": 9.440001005311911e-05} +{"train_loss": 0.017648618668317795, "grad_norm": 0.048196226358413696, "global_step": 8061, "epoch": 76, "lr": 9.439864710758136e-05} +{"train_loss": 0.014829279854893684, "grad_norm": 0.04739018902182579, "global_step": 8062, "epoch": 76, "lr": 9.439728400604607e-05} +{"train_loss": 0.018567703664302826, "grad_norm": 0.079454205930233, "global_step": 8063, "epoch": 76, "lr": 9.439592074851803e-05} +{"train_loss": 0.01898229867219925, "grad_norm": 0.08885370194911957, "global_step": 8064, "epoch": 76, "lr": 9.439455733500203e-05} +{"train_loss": 0.022927220910787582, "grad_norm": 0.05241024121642113, "global_step": 8065, "epoch": 76, "lr": 9.439319376550286e-05} +{"train_loss": 0.017617974430322647, "grad_norm": 0.053397808223962784, "global_step": 8066, "epoch": 76, "lr": 9.439183004002532e-05} +{"train_loss": 0.015207404270768166, "grad_norm": 0.06233528256416321, "global_step": 8067, "epoch": 76, "lr": 9.43904661585742e-05} +{"train_loss": 0.015933414921164513, "grad_norm": 0.06624654680490494, "global_step": 8068, "epoch": 76, "lr": 9.438910212115429e-05} +{"train_loss": 0.018545100465416908, "grad_norm": 0.10642077028751373, "global_step": 8069, "epoch": 76, "lr": 9.438773792777036e-05} +{"train_loss": 0.016795096918940544, "grad_norm": 0.049232784658670425, "global_step": 8070, "epoch": 76, "lr": 9.438637357842724e-05} +{"train_loss": 0.021515488624572754, "grad_norm": 0.1134515330195427, "global_step": 8071, "epoch": 76, "lr": 9.43850090731297e-05} +{"train_loss": 0.018734224140644073, "grad_norm": 0.05657517537474632, "global_step": 8072, "epoch": 76, "lr": 9.438364441188252e-05} +{"train_loss": 0.01884959638118744, "grad_norm": 0.07080601155757904, "global_step": 8073, "epoch": 76, "lr": 9.438227959469054e-05} +{"train_loss": 0.017248179763555527, "grad_norm": 0.09306752681732178, "global_step": 8074, "epoch": 76, "lr": 9.438091462155853e-05} +{"train_loss": 0.01885182596743107, "grad_norm": 0.0766325443983078, "global_step": 8075, "epoch": 76, "lr": 9.437954949249128e-05} +{"train_loss": 0.0173199363052845, "grad_norm": 0.060351964086294174, "global_step": 8076, "epoch": 76, "lr": 9.43781842074936e-05} +{"train_loss": 0.01958133466541767, "grad_norm": 0.07426296174526215, "global_step": 8077, "epoch": 76, "lr": 9.437681876657027e-05} +{"train_loss": 0.01680900901556015, "grad_norm": 0.06870066374540329, "global_step": 8078, "epoch": 76, "lr": 9.437545316972611e-05} +{"train_loss": 0.019102230668067932, "grad_norm": 0.07422282546758652, "global_step": 8079, "epoch": 76, "lr": 9.437408741696589e-05} +{"train_loss": 0.01862560398876667, "grad_norm": 0.0838322788476944, "global_step": 8080, "epoch": 76, "lr": 9.437272150829443e-05} +{"train_loss": 0.015093695372343063, "grad_norm": 0.03662508726119995, "global_step": 8081, "epoch": 76, "lr": 9.437135544371654e-05} +{"train_loss": 0.019437268376350403, "grad_norm": 0.04617840424180031, "global_step": 8082, "epoch": 76, "lr": 9.436998922323699e-05} +{"train_loss": 0.017437200993299484, "grad_norm": 0.09214287251234055, "global_step": 8083, "epoch": 76, "lr": 9.436862284686058e-05} +{"train_loss": 0.019123293459415436, "grad_norm": 0.12726205587387085, "global_step": 8084, "epoch": 76, "lr": 9.436725631459215e-05} +{"train_loss": 0.019523780792951584, "grad_norm": 0.1217564195394516, "global_step": 8085, "epoch": 76, "lr": 9.436588962643646e-05} +{"train_loss": 0.017381716519594193, "grad_norm": 0.07443313300609589, "global_step": 8086, "epoch": 76, "lr": 9.436452278239834e-05} +{"train_loss": 0.017982371151447296, "grad_norm": 0.06074714660644531, "global_step": 8087, "epoch": 76, "lr": 9.436315578248258e-05} +{"train_loss": 0.018875647336244583, "grad_norm": 0.06110469624400139, "global_step": 8088, "epoch": 76, "lr": 9.436178862669398e-05} +{"train_loss": 0.016800832003355026, "grad_norm": 0.14530406892299652, "global_step": 8089, "epoch": 76, "lr": 9.436042131503734e-05} +{"train_loss": 0.01630803942680359, "grad_norm": 0.11841776967048645, "global_step": 8090, "epoch": 76, "lr": 9.435905384751748e-05} +{"train_loss": 0.016658784821629524, "grad_norm": 0.10461914539337158, "global_step": 8091, "epoch": 76, "lr": 9.435768622413919e-05} +{"train_loss": 0.020035361871123314, "grad_norm": 0.07525786757469177, "global_step": 8092, "epoch": 76, "lr": 9.435631844490729e-05} +{"train_loss": 0.019100187346339226, "grad_norm": 0.10445742309093475, "global_step": 8093, "epoch": 76, "lr": 9.435495050982657e-05} +{"train_loss": 0.0186570156365633, "grad_norm": 0.055814724415540695, "global_step": 8094, "epoch": 76, "lr": 9.435358241890184e-05} +{"train_loss": 0.019985275343060493, "grad_norm": 0.1282416433095932, "global_step": 8095, "epoch": 76, "lr": 9.43522141721379e-05} +{"train_loss": 0.017372742295265198, "grad_norm": 0.05447198823094368, "global_step": 8096, "epoch": 76, "lr": 9.435084576953958e-05} +{"train_loss": 0.016587868332862854, "grad_norm": 0.1579328328371048, "global_step": 8097, "epoch": 76, "lr": 9.434947721111169e-05} +{"train_loss": 0.01905161887407303, "grad_norm": 0.09490806609392166, "global_step": 8098, "epoch": 76, "lr": 9.434810849685901e-05} +{"train_loss": 0.01955721154808998, "grad_norm": 0.07607197016477585, "global_step": 8099, "epoch": 76, "lr": 9.434673962678638e-05} +{"train_loss": 0.022321537137031555, "grad_norm": 0.062063127756118774, "global_step": 8100, "epoch": 76, "lr": 9.434537060089857e-05} +{"train_loss": 0.01839253678917885, "grad_norm": 0.20436014235019684, "global_step": 8101, "epoch": 76, "lr": 9.434400141920041e-05} +{"train_loss": 0.019413961097598076, "grad_norm": 0.17412599921226501, "global_step": 8102, "epoch": 76, "lr": 9.434263208169674e-05} +{"train_loss": 0.017844952642917633, "grad_norm": 0.0575936995446682, "global_step": 8103, "epoch": 76, "lr": 9.434126258839232e-05} +{"train_loss": 0.016957715153694153, "grad_norm": 0.05792474001646042, "global_step": 8104, "epoch": 76, "lr": 9.4339892939292e-05} +{"train_loss": 0.016209444031119347, "grad_norm": 0.10504305362701416, "global_step": 8105, "epoch": 76, "lr": 9.433852313440058e-05} +{"train_loss": 0.020630644634366035, "grad_norm": 0.10815135389566422, "global_step": 8106, "epoch": 76, "lr": 9.433715317372287e-05} +{"train_loss": 0.019903168082237244, "grad_norm": 0.0960434153676033, "global_step": 8107, "epoch": 76, "lr": 9.433578305726367e-05} +{"train_loss": 0.016957517713308334, "grad_norm": 0.11764078587293625, "global_step": 8108, "epoch": 76, "lr": 9.433441278502783e-05} +{"train_loss": 0.018224533647298813, "grad_norm": 0.07587359845638275, "global_step": 8109, "epoch": 76, "lr": 9.433304235702015e-05} +{"train_loss": 0.017251553013920784, "grad_norm": 0.1147579476237297, "global_step": 8110, "epoch": 76, "lr": 9.433167177324542e-05} +{"train_loss": 0.016901127994060516, "grad_norm": 0.09275182336568832, "global_step": 8111, "epoch": 76, "lr": 9.433030103370849e-05} +{"train_loss": 0.01894202083349228, "grad_norm": 0.08748868852853775, "global_step": 8112, "epoch": 76, "lr": 9.432893013841413e-05} +{"train_loss": 0.0183065477758646, "grad_norm": 0.12708942592144012, "global_step": 8113, "epoch": 76, "lr": 9.432755908736722e-05} +{"train_loss": 0.017578694969415665, "grad_norm": 0.05760352313518524, "global_step": 8114, "epoch": 76, "lr": 9.432618788057253e-05} +{"train_loss": 0.018826598301529884, "grad_norm": 0.22848154604434967, "global_step": 8115, "epoch": 76, "lr": 9.432481651803488e-05} +{"train_loss": 0.016043219715356827, "grad_norm": 0.06746363639831543, "global_step": 8116, "epoch": 76, "lr": 9.432344499975914e-05} +{"train_loss": 0.020493265241384506, "grad_norm": 0.1059894859790802, "global_step": 8117, "epoch": 76, "lr": 9.432207332575006e-05} +{"train_loss": 0.01754409447312355, "grad_norm": 0.06894148886203766, "global_step": 8118, "epoch": 76, "lr": 9.43207014960125e-05} +{"train_loss": 0.018561942502856255, "grad_norm": 0.13934041559696198, "global_step": 8119, "epoch": 76, "lr": 9.431932951055127e-05} +{"train_loss": 0.01832585409283638, "grad_norm": 0.06959804147481918, "global_step": 8120, "epoch": 76, "lr": 9.431795736937118e-05} +{"train_loss": 0.020384304225444794, "grad_norm": 0.1263989359140396, "global_step": 8121, "epoch": 76, "lr": 9.431658507247708e-05} +{"train_loss": 0.019124172627925873, "grad_norm": 0.12925194203853607, "global_step": 8122, "epoch": 76, "lr": 9.431521261987375e-05} +{"train_loss": 0.01938742585480213, "grad_norm": 0.10372993350028992, "global_step": 8123, "epoch": 76, "lr": 9.431384001156605e-05} +{"train_loss": 0.018065115436911583, "grad_norm": 0.08160481601953506, "global_step": 8124, "epoch": 76, "lr": 9.431246724755877e-05} +{"train_loss": 0.01795918121933937, "grad_norm": 0.136376291513443, "global_step": 8125, "epoch": 76, "lr": 9.431109432785676e-05} +{"train_loss": 0.019139017909765244, "grad_norm": 0.10044457763433456, "global_step": 8126, "epoch": 76, "lr": 9.430972125246484e-05} +{"train_loss": 0.01609245501458645, "grad_norm": 0.11498995125293732, "global_step": 8127, "epoch": 76, "lr": 9.430834802138784e-05} +{"train_loss": 0.01886012777686119, "grad_norm": 0.08696180582046509, "global_step": 8128, "epoch": 76, "lr": 9.430697463463056e-05} +{"train_loss": 0.017244260758161545, "grad_norm": 0.07179585099220276, "global_step": 8129, "epoch": 76, "lr": 9.430560109219785e-05} +{"train_loss": 0.01672593131661415, "grad_norm": 0.07331538945436478, "global_step": 8130, "epoch": 76, "lr": 9.430422739409452e-05} +{"train_loss": 0.017429642379283905, "grad_norm": 0.10374613851308823, "global_step": 8131, "epoch": 76, "lr": 9.430285354032539e-05} +{"train_loss": 0.016880061477422714, "grad_norm": 0.1135285422205925, "global_step": 8132, "epoch": 76, "lr": 9.430147953089532e-05} +{"train_loss": 0.01636189967393875, "grad_norm": 0.04283878207206726, "global_step": 8133, "epoch": 76, "lr": 9.430010536580911e-05} +{"train_loss": 0.019509674981236458, "grad_norm": 0.09633650630712509, "global_step": 8134, "epoch": 76, "lr": 9.429873104507159e-05} +{"train_loss": 0.017265859991312027, "grad_norm": 0.059892527759075165, "global_step": 8135, "epoch": 76, "lr": 9.429735656868759e-05} +{"train_loss": 0.016532208770513535, "grad_norm": 0.051308389753103256, "global_step": 8136, "epoch": 76, "lr": 9.429598193666196e-05} +{"train_loss": 0.02076657861471176, "grad_norm": 0.11946956068277359, "global_step": 8137, "epoch": 76, "lr": 9.429460714899951e-05} +{"train_loss": 0.01682109385728836, "grad_norm": 0.09796655923128128, "global_step": 8138, "epoch": 76, "lr": 9.429323220570505e-05} +{"train_loss": 0.020466666668653488, "grad_norm": 0.08664793521165848, "global_step": 8139, "epoch": 76, "lr": 9.429185710678345e-05} +{"train_loss": 0.018537122756242752, "grad_norm": 0.09476357698440552, "global_step": 8140, "epoch": 76, "lr": 9.429048185223954e-05} +{"train_loss": 0.016707777976989746, "grad_norm": 0.09439750760793686, "global_step": 8141, "epoch": 76, "lr": 9.428910644207812e-05} +{"train_loss": 0.01720748096704483, "grad_norm": 0.047774787992239, "global_step": 8142, "epoch": 76, "lr": 9.428773087630404e-05} +{"train_loss": 0.020274320617318153, "grad_norm": 0.14565876126289368, "global_step": 8143, "epoch": 76, "lr": 9.428635515492214e-05} +{"train_loss": 0.019505823031067848, "grad_norm": 0.20077328383922577, "global_step": 8144, "epoch": 76, "lr": 9.428497927793726e-05} +{"train_loss": 0.02210431545972824, "grad_norm": 0.07517806440591812, "global_step": 8145, "epoch": 76, "lr": 9.428360324535421e-05} +{"train_loss": 0.01737901195883751, "grad_norm": 0.054901815950870514, "global_step": 8146, "epoch": 76, "lr": 9.428222705717784e-05} +{"train_loss": 0.018732909113168716, "grad_norm": 0.04997418448328972, "global_step": 8147, "epoch": 76, "lr": 9.428085071341298e-05} +{"train_loss": 0.01687546819448471, "grad_norm": 0.048860978335142136, "global_step": 8148, "epoch": 76, "lr": 9.427947421406447e-05} +{"train_loss": 0.020348507910966873, "grad_norm": 0.10388635843992233, "global_step": 8149, "epoch": 76, "lr": 9.427809755913714e-05} +{"train_loss": 0.017346657812595367, "grad_norm": 0.054743442684412, "global_step": 8150, "epoch": 76, "lr": 9.427672074863583e-05} +{"train_loss": 0.018643248826265335, "grad_norm": 0.05896696820855141, "global_step": 8151, "epoch": 76, "lr": 9.427534378256538e-05} +{"train_loss": 0.021081898361444473, "grad_norm": 0.05753907188773155, "global_step": 8152, "epoch": 76, "lr": 9.427396666093062e-05} +{"train_loss": 0.017437150701880455, "grad_norm": 0.08337545394897461, "global_step": 8153, "epoch": 76, "lr": 9.42725893837364e-05} +{"train_loss": 0.01850508712232113, "grad_norm": 0.09065453708171844, "global_step": 8154, "epoch": 76, "lr": 9.427121195098757e-05} +{"train_loss": 0.01816677860915661, "grad_norm": 0.06550745666027069, "global_step": 8155, "epoch": 76, "lr": 9.426983436268895e-05} +{"train_loss": 0.018611829727888107, "grad_norm": 0.08289957791566849, "global_step": 8156, "epoch": 76, "lr": 9.426845661884537e-05} +{"train_loss": 0.017729632556438446, "grad_norm": 0.04599343612790108, "global_step": 8157, "epoch": 76, "lr": 9.426707871946169e-05} +{"train_loss": 0.019506685435771942, "grad_norm": 0.09517639875411987, "global_step": 8158, "epoch": 76, "lr": 9.426570066454275e-05} +{"train_loss": 0.018284188583493233, "grad_norm": 0.047442782670259476, "global_step": 8159, "epoch": 76, "lr": 9.426432245409338e-05} +{"train_loss": 0.020034506916999817, "grad_norm": 0.05923457071185112, "global_step": 8160, "epoch": 76, "lr": 9.426294408811845e-05} +{"train_loss": 0.01827473433266552, "grad_norm": 0.11781005561351776, "global_step": 8161, "epoch": 76, "lr": 9.426156556662276e-05, "val_loss": 0.019010435789823532} +{"train_loss": 0.018985528498888016, "grad_norm": 0.054813116788864136, "global_step": 8162, "epoch": 77, "lr": 9.42601868896112e-05} +{"train_loss": 0.018220964819192886, "grad_norm": 0.06610973924398422, "global_step": 8163, "epoch": 77, "lr": 9.425880805708857e-05} +{"train_loss": 0.0173516683280468, "grad_norm": 0.06113418564200401, "global_step": 8164, "epoch": 77, "lr": 9.425742906905976e-05} +{"train_loss": 0.019855301827192307, "grad_norm": 0.08138187229633331, "global_step": 8165, "epoch": 77, "lr": 9.425604992552957e-05} +{"train_loss": 0.020523246377706528, "grad_norm": 0.080337293446064, "global_step": 8166, "epoch": 77, "lr": 9.425467062650288e-05} +{"train_loss": 0.020189544185996056, "grad_norm": 0.07914035022258759, "global_step": 8167, "epoch": 77, "lr": 9.425329117198452e-05} +{"train_loss": 0.01758194901049137, "grad_norm": 0.13731487095355988, "global_step": 8168, "epoch": 77, "lr": 9.425191156197933e-05} +{"train_loss": 0.01959436573088169, "grad_norm": 0.10529439151287079, "global_step": 8169, "epoch": 77, "lr": 9.425053179649218e-05} +{"train_loss": 0.01658305712044239, "grad_norm": 0.0756247267127037, "global_step": 8170, "epoch": 77, "lr": 9.42491518755279e-05} +{"train_loss": 0.019340060651302338, "grad_norm": 0.04749315604567528, "global_step": 8171, "epoch": 77, "lr": 9.424777179909135e-05} +{"train_loss": 0.017009533941745758, "grad_norm": 0.0399964340031147, "global_step": 8172, "epoch": 77, "lr": 9.424639156718736e-05} +{"train_loss": 0.01757640577852726, "grad_norm": 0.058408480137586594, "global_step": 8173, "epoch": 77, "lr": 9.424501117982081e-05} +{"train_loss": 0.017567187547683716, "grad_norm": 0.052907176315784454, "global_step": 8174, "epoch": 77, "lr": 9.424363063699652e-05} +{"train_loss": 0.014802795834839344, "grad_norm": 0.05377805233001709, "global_step": 8175, "epoch": 77, "lr": 9.424224993871935e-05} +{"train_loss": 0.017528295516967773, "grad_norm": 0.11105620861053467, "global_step": 8176, "epoch": 77, "lr": 9.424086908499416e-05} +{"train_loss": 0.01657130755484104, "grad_norm": 0.053465552628040314, "global_step": 8177, "epoch": 77, "lr": 9.42394880758258e-05} +{"train_loss": 0.018467005342245102, "grad_norm": 0.11168520897626877, "global_step": 8178, "epoch": 77, "lr": 9.423810691121911e-05} +{"train_loss": 0.01580684632062912, "grad_norm": 0.046784088015556335, "global_step": 8179, "epoch": 77, "lr": 9.423672559117895e-05} +{"train_loss": 0.017745859920978546, "grad_norm": 0.08258910477161407, "global_step": 8180, "epoch": 77, "lr": 9.423534411571019e-05} +{"train_loss": 0.020894646644592285, "grad_norm": 0.05719015747308731, "global_step": 8181, "epoch": 77, "lr": 9.423396248481767e-05} +{"train_loss": 0.018768660724163055, "grad_norm": 0.07891235500574112, "global_step": 8182, "epoch": 77, "lr": 9.423258069850621e-05} +{"train_loss": 0.01762695424258709, "grad_norm": 0.05083616450428963, "global_step": 8183, "epoch": 77, "lr": 9.423119875678073e-05} +{"train_loss": 0.017175544053316116, "grad_norm": 0.04389280453324318, "global_step": 8184, "epoch": 77, "lr": 9.422981665964603e-05} +{"train_loss": 0.017000563442707062, "grad_norm": 0.05722927302122116, "global_step": 8185, "epoch": 77, "lr": 9.4228434407107e-05} +{"train_loss": 0.02078889310359955, "grad_norm": 0.09959030151367188, "global_step": 8186, "epoch": 77, "lr": 9.422705199916848e-05} +{"train_loss": 0.017933228984475136, "grad_norm": 0.05908135697245598, "global_step": 8187, "epoch": 77, "lr": 9.422566943583535e-05} +{"train_loss": 0.015684733167290688, "grad_norm": 0.05753166601061821, "global_step": 8188, "epoch": 77, "lr": 9.422428671711243e-05} +{"train_loss": 0.017221080139279366, "grad_norm": 0.048710208386182785, "global_step": 8189, "epoch": 77, "lr": 9.422290384300461e-05} +{"train_loss": 0.018439196050167084, "grad_norm": 0.07375577092170715, "global_step": 8190, "epoch": 77, "lr": 9.422152081351673e-05} +{"train_loss": 0.0201706700026989, "grad_norm": 0.04884364455938339, "global_step": 8191, "epoch": 77, "lr": 9.422013762865367e-05} +{"train_loss": 0.017496075481176376, "grad_norm": 0.07375779747962952, "global_step": 8192, "epoch": 77, "lr": 9.421875428842028e-05} +{"train_loss": 0.019264549016952515, "grad_norm": 0.04486886411905289, "global_step": 8193, "epoch": 77, "lr": 9.42173707928214e-05} +{"train_loss": 0.020712291821837425, "grad_norm": 0.11324144154787064, "global_step": 8194, "epoch": 77, "lr": 9.421598714186192e-05} +{"train_loss": 0.01689109019935131, "grad_norm": 0.058351002633571625, "global_step": 8195, "epoch": 77, "lr": 9.421460333554669e-05} +{"train_loss": 0.020277833566069603, "grad_norm": 0.06973839551210403, "global_step": 8196, "epoch": 77, "lr": 9.421321937388057e-05} +{"train_loss": 0.016659073531627655, "grad_norm": 0.07503189891576767, "global_step": 8197, "epoch": 77, "lr": 9.421183525686843e-05} +{"train_loss": 0.01643889769911766, "grad_norm": 0.04665606841444969, "global_step": 8198, "epoch": 77, "lr": 9.421045098451512e-05} +{"train_loss": 0.017292272299528122, "grad_norm": 0.05039821192622185, "global_step": 8199, "epoch": 77, "lr": 9.420906655682553e-05} +{"train_loss": 0.015078732743859291, "grad_norm": 0.10588574409484863, "global_step": 8200, "epoch": 77, "lr": 9.420768197380447e-05} +{"train_loss": 0.015787875279784203, "grad_norm": 0.1322847455739975, "global_step": 8201, "epoch": 77, "lr": 9.420629723545687e-05} +{"train_loss": 0.019016806036233902, "grad_norm": 0.054308947175741196, "global_step": 8202, "epoch": 77, "lr": 9.420491234178756e-05} +{"train_loss": 0.015429952181875706, "grad_norm": 0.10572032630443573, "global_step": 8203, "epoch": 77, "lr": 9.42035272928014e-05} +{"train_loss": 0.01861738972365856, "grad_norm": 0.05606762319803238, "global_step": 8204, "epoch": 77, "lr": 9.42021420885033e-05} +{"train_loss": 0.01709456369280815, "grad_norm": 0.049669746309518814, "global_step": 8205, "epoch": 77, "lr": 9.420075672889806e-05} +{"train_loss": 0.019292598590254784, "grad_norm": 0.07993326336145401, "global_step": 8206, "epoch": 77, "lr": 9.41993712139906e-05} +{"train_loss": 0.02011418528854847, "grad_norm": 0.11024682968854904, "global_step": 8207, "epoch": 77, "lr": 9.419798554378577e-05} +{"train_loss": 0.021154649555683136, "grad_norm": 0.08279994875192642, "global_step": 8208, "epoch": 77, "lr": 9.419659971828845e-05} +{"train_loss": 0.02045491337776184, "grad_norm": 0.0703386515378952, "global_step": 8209, "epoch": 77, "lr": 9.419521373750349e-05} +{"train_loss": 0.020435405895113945, "grad_norm": 0.07493613660335541, "global_step": 8210, "epoch": 77, "lr": 9.419382760143576e-05} +{"train_loss": 0.019120875746011734, "grad_norm": 0.1485053151845932, "global_step": 8211, "epoch": 77, "lr": 9.419244131009015e-05} +{"train_loss": 0.017631947994232178, "grad_norm": 0.07699036598205566, "global_step": 8212, "epoch": 77, "lr": 9.419105486347152e-05} +{"train_loss": 0.01855424791574478, "grad_norm": 0.09241564571857452, "global_step": 8213, "epoch": 77, "lr": 9.418966826158473e-05} +{"train_loss": 0.021779745817184448, "grad_norm": 0.10980140417814255, "global_step": 8214, "epoch": 77, "lr": 9.418828150443469e-05} +{"train_loss": 0.017403852194547653, "grad_norm": 0.06768172979354858, "global_step": 8215, "epoch": 77, "lr": 9.418689459202621e-05} +{"train_loss": 0.01749706268310547, "grad_norm": 0.036721136420965195, "global_step": 8216, "epoch": 77, "lr": 9.418550752436422e-05} +{"train_loss": 0.018158607184886932, "grad_norm": 0.07370109856128693, "global_step": 8217, "epoch": 77, "lr": 9.418412030145358e-05} +{"train_loss": 0.018502559512853622, "grad_norm": 0.0660240426659584, "global_step": 8218, "epoch": 77, "lr": 9.418273292329913e-05} +{"train_loss": 0.01928327977657318, "grad_norm": 0.07726190239191055, "global_step": 8219, "epoch": 77, "lr": 9.41813453899058e-05} +{"train_loss": 0.016484566032886505, "grad_norm": 0.053854286670684814, "global_step": 8220, "epoch": 77, "lr": 9.417995770127841e-05} +{"train_loss": 0.018810462206602097, "grad_norm": 0.04748104140162468, "global_step": 8221, "epoch": 77, "lr": 9.417856985742189e-05} +{"train_loss": 0.018616968765854836, "grad_norm": 0.09771620482206345, "global_step": 8222, "epoch": 77, "lr": 9.417718185834105e-05} +{"train_loss": 0.01723698526620865, "grad_norm": 0.08622590452432632, "global_step": 8223, "epoch": 77, "lr": 9.417579370404084e-05} +{"train_loss": 0.016062254086136818, "grad_norm": 0.07215898483991623, "global_step": 8224, "epoch": 77, "lr": 9.417440539452609e-05} +{"train_loss": 0.017015084624290466, "grad_norm": 0.04233429208397865, "global_step": 8225, "epoch": 77, "lr": 9.41730169298017e-05} +{"train_loss": 0.016318056732416153, "grad_norm": 0.05839519575238228, "global_step": 8226, "epoch": 77, "lr": 9.417162830987252e-05} +{"train_loss": 0.017189480364322662, "grad_norm": 0.0525931790471077, "global_step": 8227, "epoch": 77, "lr": 9.417023953474347e-05} +{"train_loss": 0.01588025875389576, "grad_norm": 0.09049010276794434, "global_step": 8228, "epoch": 77, "lr": 9.416885060441938e-05} +{"train_loss": 0.018988559022545815, "grad_norm": 0.06795128434896469, "global_step": 8229, "epoch": 77, "lr": 9.416746151890517e-05} +{"train_loss": 0.016226675361394882, "grad_norm": 0.055851276963949203, "global_step": 8230, "epoch": 77, "lr": 9.416607227820571e-05} +{"train_loss": 0.019346006214618683, "grad_norm": 0.08317248523235321, "global_step": 8231, "epoch": 77, "lr": 9.416468288232588e-05} +{"train_loss": 0.017311042174696922, "grad_norm": 0.0913388803601265, "global_step": 8232, "epoch": 77, "lr": 9.416329333127057e-05} +{"train_loss": 0.01747121661901474, "grad_norm": 0.05438810959458351, "global_step": 8233, "epoch": 77, "lr": 9.416190362504464e-05} +{"train_loss": 0.018387049436569214, "grad_norm": 0.051918964833021164, "global_step": 8234, "epoch": 77, "lr": 9.416051376365299e-05} +{"train_loss": 0.019939657300710678, "grad_norm": 0.04567630589008331, "global_step": 8235, "epoch": 77, "lr": 9.41591237471005e-05} +{"train_loss": 0.018609143793582916, "grad_norm": 0.08161542564630508, "global_step": 8236, "epoch": 77, "lr": 9.415773357539205e-05} +{"train_loss": 0.01625974103808403, "grad_norm": 0.06074310466647148, "global_step": 8237, "epoch": 77, "lr": 9.415634324853253e-05} +{"train_loss": 0.01779388263821602, "grad_norm": 0.04466790333390236, "global_step": 8238, "epoch": 77, "lr": 9.415495276652683e-05} +{"train_loss": 0.019706768915057182, "grad_norm": 0.07525910437107086, "global_step": 8239, "epoch": 77, "lr": 9.415356212937982e-05} +{"train_loss": 0.017380941659212112, "grad_norm": 0.06081318110227585, "global_step": 8240, "epoch": 77, "lr": 9.41521713370964e-05} +{"train_loss": 0.018653888255357742, "grad_norm": 0.06329978257417679, "global_step": 8241, "epoch": 77, "lr": 9.415078038968145e-05} +{"train_loss": 0.019496027380228043, "grad_norm": 0.04549098759889603, "global_step": 8242, "epoch": 77, "lr": 9.414938928713986e-05} +{"train_loss": 0.017920859158039093, "grad_norm": 0.06584988534450531, "global_step": 8243, "epoch": 77, "lr": 9.414799802947652e-05} +{"train_loss": 0.017521338537335396, "grad_norm": 0.048862993717193604, "global_step": 8244, "epoch": 77, "lr": 9.41466066166963e-05} +{"train_loss": 0.018367305397987366, "grad_norm": 0.06970249861478806, "global_step": 8245, "epoch": 77, "lr": 9.414521504880411e-05} +{"train_loss": 0.019051846116781235, "grad_norm": 0.10227324813604355, "global_step": 8246, "epoch": 77, "lr": 9.414382332580483e-05} +{"train_loss": 0.01673917844891548, "grad_norm": 0.0625922903418541, "global_step": 8247, "epoch": 77, "lr": 9.414243144770335e-05} +{"train_loss": 0.019458431750535965, "grad_norm": 0.07333646714687347, "global_step": 8248, "epoch": 77, "lr": 9.414103941450457e-05} +{"train_loss": 0.019739262759685516, "grad_norm": 0.05424318462610245, "global_step": 8249, "epoch": 77, "lr": 9.413964722621338e-05} +{"train_loss": 0.01964736357331276, "grad_norm": 0.055102359503507614, "global_step": 8250, "epoch": 77, "lr": 9.413825488283466e-05} +{"train_loss": 0.01804163306951523, "grad_norm": 0.09608225524425507, "global_step": 8251, "epoch": 77, "lr": 9.41368623843733e-05} +{"train_loss": 0.017979009076952934, "grad_norm": 0.13079649209976196, "global_step": 8252, "epoch": 77, "lr": 9.41354697308342e-05} +{"train_loss": 0.017147157341241837, "grad_norm": 0.06727737933397293, "global_step": 8253, "epoch": 77, "lr": 9.413407692222226e-05} +{"train_loss": 0.016346164047718048, "grad_norm": 0.07977229356765747, "global_step": 8254, "epoch": 77, "lr": 9.413268395854236e-05} +{"train_loss": 0.016943754628300667, "grad_norm": 0.04702137038111687, "global_step": 8255, "epoch": 77, "lr": 9.413129083979941e-05} +{"train_loss": 0.020309001207351685, "grad_norm": 0.059833116829395294, "global_step": 8256, "epoch": 77, "lr": 9.41298975659983e-05} +{"train_loss": 0.017838848754763603, "grad_norm": 0.0660693570971489, "global_step": 8257, "epoch": 77, "lr": 9.412850413714391e-05} +{"train_loss": 0.020190196111798286, "grad_norm": 0.10031155496835709, "global_step": 8258, "epoch": 77, "lr": 9.412711055324116e-05} +{"train_loss": 0.020578451454639435, "grad_norm": 0.09469667822122574, "global_step": 8259, "epoch": 77, "lr": 9.412571681429492e-05} +{"train_loss": 0.01787731610238552, "grad_norm": 0.05829055607318878, "global_step": 8260, "epoch": 77, "lr": 9.41243229203101e-05} +{"train_loss": 0.018053151667118073, "grad_norm": 0.05771249905228615, "global_step": 8261, "epoch": 77, "lr": 9.412292887129161e-05} +{"train_loss": 0.018879054114222527, "grad_norm": 0.05556405335664749, "global_step": 8262, "epoch": 77, "lr": 9.412153466724433e-05} +{"train_loss": 0.017927777022123337, "grad_norm": 0.06089603528380394, "global_step": 8263, "epoch": 77, "lr": 9.412014030817317e-05} +{"train_loss": 0.016793696209788322, "grad_norm": 0.044843800365924835, "global_step": 8264, "epoch": 77, "lr": 9.411874579408303e-05} +{"train_loss": 0.020871642976999283, "grad_norm": 0.10164899379014969, "global_step": 8265, "epoch": 77, "lr": 9.411735112497879e-05} +{"train_loss": 0.016737161204218864, "grad_norm": 0.06799472868442535, "global_step": 8266, "epoch": 77, "lr": 9.411595630086538e-05} +{"train_loss": 0.018168600240968308, "grad_norm": 0.1523093581199646, "global_step": 8267, "epoch": 77, "lr": 9.411456132174767e-05, "val_loss": 0.018684353679418564} +{"train_loss": 0.01656625233590603, "grad_norm": 0.14371992647647858, "global_step": 8268, "epoch": 78, "lr": 9.411316618763058e-05} +{"train_loss": 0.01773224025964737, "grad_norm": 0.13455310463905334, "global_step": 8269, "epoch": 78, "lr": 9.411177089851902e-05} +{"train_loss": 0.016618328168988228, "grad_norm": 0.08131422847509384, "global_step": 8270, "epoch": 78, "lr": 9.411037545441786e-05} +{"train_loss": 0.018510742112994194, "grad_norm": 0.04679566249251366, "global_step": 8271, "epoch": 78, "lr": 9.410897985533203e-05} +{"train_loss": 0.01713968813419342, "grad_norm": 0.04889223724603653, "global_step": 8272, "epoch": 78, "lr": 9.410758410126643e-05} +{"train_loss": 0.017118193209171295, "grad_norm": 0.08976138383150101, "global_step": 8273, "epoch": 78, "lr": 9.410618819222596e-05} +{"train_loss": 0.017741084098815918, "grad_norm": 0.09183262288570404, "global_step": 8274, "epoch": 78, "lr": 9.410479212821555e-05} +{"train_loss": 0.017862599343061447, "grad_norm": 0.05012405291199684, "global_step": 8275, "epoch": 78, "lr": 9.410339590924005e-05} +{"train_loss": 0.016665400937199593, "grad_norm": 0.056509729474782944, "global_step": 8276, "epoch": 78, "lr": 9.41019995353044e-05} +{"train_loss": 0.016767408698797226, "grad_norm": 0.1209532767534256, "global_step": 8277, "epoch": 78, "lr": 9.410060300641353e-05} +{"train_loss": 0.018275408074259758, "grad_norm": 0.0928848460316658, "global_step": 8278, "epoch": 78, "lr": 9.409920632257229e-05} +{"train_loss": 0.019265182316303253, "grad_norm": 0.08198755979537964, "global_step": 8279, "epoch": 78, "lr": 9.409780948378562e-05} +{"train_loss": 0.016168706119060516, "grad_norm": 0.041194070130586624, "global_step": 8280, "epoch": 78, "lr": 9.409641249005843e-05} +{"train_loss": 0.019876820966601372, "grad_norm": 0.0965924933552742, "global_step": 8281, "epoch": 78, "lr": 9.409501534139563e-05} +{"train_loss": 0.017233960330486298, "grad_norm": 0.12965743243694305, "global_step": 8282, "epoch": 78, "lr": 9.409361803780212e-05} +{"train_loss": 0.01755356416106224, "grad_norm": 0.06538417190313339, "global_step": 8283, "epoch": 78, "lr": 9.40922205792828e-05} +{"train_loss": 0.017521822825074196, "grad_norm": 0.06575401872396469, "global_step": 8284, "epoch": 78, "lr": 9.409082296584262e-05} +{"train_loss": 0.01816129870712757, "grad_norm": 0.06582359224557877, "global_step": 8285, "epoch": 78, "lr": 9.408942519748643e-05} +{"train_loss": 0.018456459045410156, "grad_norm": 0.08222366869449615, "global_step": 8286, "epoch": 78, "lr": 9.408802727421919e-05} +{"train_loss": 0.019289463758468628, "grad_norm": 0.050046149641275406, "global_step": 8287, "epoch": 78, "lr": 9.40866291960458e-05} +{"train_loss": 0.02038036659359932, "grad_norm": 0.053687360137701035, "global_step": 8288, "epoch": 78, "lr": 9.408523096297117e-05} +{"train_loss": 0.02087494730949402, "grad_norm": 0.0630062073469162, "global_step": 8289, "epoch": 78, "lr": 9.40838325750002e-05} +{"train_loss": 0.020643753930926323, "grad_norm": 0.054575011134147644, "global_step": 8290, "epoch": 78, "lr": 9.408243403213781e-05} +{"train_loss": 0.01609020307660103, "grad_norm": 0.05162515118718147, "global_step": 8291, "epoch": 78, "lr": 9.408103533438892e-05} +{"train_loss": 0.01738094538450241, "grad_norm": 0.08932093530893326, "global_step": 8292, "epoch": 78, "lr": 9.407963648175846e-05} +{"train_loss": 0.019130438566207886, "grad_norm": 0.11864088475704193, "global_step": 8293, "epoch": 78, "lr": 9.407823747425131e-05} +{"train_loss": 0.018306026235222816, "grad_norm": 0.08548995852470398, "global_step": 8294, "epoch": 78, "lr": 9.407683831187239e-05} +{"train_loss": 0.019072934985160828, "grad_norm": 0.08249060809612274, "global_step": 8295, "epoch": 78, "lr": 9.407543899462666e-05} +{"train_loss": 0.018753571435809135, "grad_norm": 0.05595838651061058, "global_step": 8296, "epoch": 78, "lr": 9.407403952251899e-05} +{"train_loss": 0.018111582845449448, "grad_norm": 0.0618545226752758, "global_step": 8297, "epoch": 78, "lr": 9.40726398955543e-05} +{"train_loss": 0.015356761403381824, "grad_norm": 0.1305140107870102, "global_step": 8298, "epoch": 78, "lr": 9.407124011373754e-05} +{"train_loss": 0.018073584884405136, "grad_norm": 0.07379454374313354, "global_step": 8299, "epoch": 78, "lr": 9.406984017707361e-05} +{"train_loss": 0.019375495612621307, "grad_norm": 0.046971842646598816, "global_step": 8300, "epoch": 78, "lr": 9.406844008556742e-05} +{"train_loss": 0.017592113465070724, "grad_norm": 0.09022314101457596, "global_step": 8301, "epoch": 78, "lr": 9.406703983922389e-05} +{"train_loss": 0.01848769374191761, "grad_norm": 0.04807063564658165, "global_step": 8302, "epoch": 78, "lr": 9.406563943804795e-05} +{"train_loss": 0.021729949861764908, "grad_norm": 0.049384262412786484, "global_step": 8303, "epoch": 78, "lr": 9.406423888204452e-05} +{"train_loss": 0.01959073171019554, "grad_norm": 0.04421602562069893, "global_step": 8304, "epoch": 78, "lr": 9.40628381712185e-05} +{"train_loss": 0.017576050013303757, "grad_norm": 0.08326195925474167, "global_step": 8305, "epoch": 78, "lr": 9.406143730557486e-05} +{"train_loss": 0.02009989693760872, "grad_norm": 0.09141354262828827, "global_step": 8306, "epoch": 78, "lr": 9.406003628511847e-05} +{"train_loss": 0.016562625765800476, "grad_norm": 0.05747590214014053, "global_step": 8307, "epoch": 78, "lr": 9.405863510985429e-05} +{"train_loss": 0.018956683576107025, "grad_norm": 0.04390452802181244, "global_step": 8308, "epoch": 78, "lr": 9.40572337797872e-05} +{"train_loss": 0.018625231459736824, "grad_norm": 0.06099493429064751, "global_step": 8309, "epoch": 78, "lr": 9.405583229492217e-05} +{"train_loss": 0.021387668326497078, "grad_norm": 0.05312899872660637, "global_step": 8310, "epoch": 78, "lr": 9.40544306552641e-05} +{"train_loss": 0.016757210716605186, "grad_norm": 0.045206520706415176, "global_step": 8311, "epoch": 78, "lr": 9.405302886081793e-05} +{"train_loss": 0.021984416991472244, "grad_norm": 0.05160287395119667, "global_step": 8312, "epoch": 78, "lr": 9.405162691158856e-05} +{"train_loss": 0.01802019029855728, "grad_norm": 0.039303284138441086, "global_step": 8313, "epoch": 78, "lr": 9.405022480758094e-05} +{"train_loss": 0.017776373773813248, "grad_norm": 0.10907111316919327, "global_step": 8314, "epoch": 78, "lr": 9.404882254879998e-05} +{"train_loss": 0.02129673957824707, "grad_norm": 0.11260338872671127, "global_step": 8315, "epoch": 78, "lr": 9.404742013525061e-05} +{"train_loss": 0.02046232670545578, "grad_norm": 0.09300009161233902, "global_step": 8316, "epoch": 78, "lr": 9.404601756693777e-05} +{"train_loss": 0.019422978162765503, "grad_norm": 0.07331673055887222, "global_step": 8317, "epoch": 78, "lr": 9.404461484386637e-05} +{"train_loss": 0.01751847192645073, "grad_norm": 0.04603048041462898, "global_step": 8318, "epoch": 78, "lr": 9.404321196604135e-05} +{"train_loss": 0.022134210914373398, "grad_norm": 0.05458681285381317, "global_step": 8319, "epoch": 78, "lr": 9.404180893346766e-05} +{"train_loss": 0.0186927430331707, "grad_norm": 0.05858515948057175, "global_step": 8320, "epoch": 78, "lr": 9.404040574615018e-05} +{"train_loss": 0.018564946949481964, "grad_norm": 0.05303261801600456, "global_step": 8321, "epoch": 78, "lr": 9.403900240409389e-05} +{"train_loss": 0.015566764399409294, "grad_norm": 0.040343042463064194, "global_step": 8322, "epoch": 78, "lr": 9.403759890730368e-05} +{"train_loss": 0.0164417065680027, "grad_norm": 0.04957788810133934, "global_step": 8323, "epoch": 78, "lr": 9.40361952557845e-05} +{"train_loss": 0.02024213597178459, "grad_norm": 0.10587197542190552, "global_step": 8324, "epoch": 78, "lr": 9.403479144954128e-05} +{"train_loss": 0.016932815313339233, "grad_norm": 0.07571399956941605, "global_step": 8325, "epoch": 78, "lr": 9.403338748857896e-05} +{"train_loss": 0.01654541864991188, "grad_norm": 0.04431037977337837, "global_step": 8326, "epoch": 78, "lr": 9.403198337290246e-05} +{"train_loss": 0.019567258656024933, "grad_norm": 0.1682286262512207, "global_step": 8327, "epoch": 78, "lr": 9.403057910251672e-05} +{"train_loss": 0.018308348953723907, "grad_norm": 0.0639965608716011, "global_step": 8328, "epoch": 78, "lr": 9.402917467742667e-05} +{"train_loss": 0.017272571101784706, "grad_norm": 0.05168949067592621, "global_step": 8329, "epoch": 78, "lr": 9.402777009763726e-05} +{"train_loss": 0.01970992609858513, "grad_norm": 0.0755632221698761, "global_step": 8330, "epoch": 78, "lr": 9.402636536315341e-05} +{"train_loss": 0.01676136627793312, "grad_norm": 0.04357482120394707, "global_step": 8331, "epoch": 78, "lr": 9.402496047398005e-05} +{"train_loss": 0.018258629366755486, "grad_norm": 0.1127474457025528, "global_step": 8332, "epoch": 78, "lr": 9.402355543012213e-05} +{"train_loss": 0.02191871777176857, "grad_norm": 0.0808948203921318, "global_step": 8333, "epoch": 78, "lr": 9.402215023158458e-05} +{"train_loss": 0.017900148406624794, "grad_norm": 0.09977614134550095, "global_step": 8334, "epoch": 78, "lr": 9.402074487837233e-05} +{"train_loss": 0.015984417870640755, "grad_norm": 0.08655271679162979, "global_step": 8335, "epoch": 78, "lr": 9.401933937049035e-05} +{"train_loss": 0.018183309584856033, "grad_norm": 0.06978829205036163, "global_step": 8336, "epoch": 78, "lr": 9.401793370794354e-05} +{"train_loss": 0.015792030841112137, "grad_norm": 0.04351135343313217, "global_step": 8337, "epoch": 78, "lr": 9.401652789073685e-05} +{"train_loss": 0.015754248946905136, "grad_norm": 0.047786541283130646, "global_step": 8338, "epoch": 78, "lr": 9.401512191887522e-05} +{"train_loss": 0.019069228321313858, "grad_norm": 0.09470590204000473, "global_step": 8339, "epoch": 78, "lr": 9.401371579236361e-05} +{"train_loss": 0.019303947687149048, "grad_norm": 0.0489792637526989, "global_step": 8340, "epoch": 78, "lr": 9.401230951120694e-05} +{"train_loss": 0.018619544804096222, "grad_norm": 0.06095382198691368, "global_step": 8341, "epoch": 78, "lr": 9.401090307541015e-05} +{"train_loss": 0.01923113316297531, "grad_norm": 0.0803481787443161, "global_step": 8342, "epoch": 78, "lr": 9.400949648497818e-05} +{"train_loss": 0.017099233344197273, "grad_norm": 0.07366184890270233, "global_step": 8343, "epoch": 78, "lr": 9.400808973991599e-05} +{"train_loss": 0.01734587550163269, "grad_norm": 0.07754684239625931, "global_step": 8344, "epoch": 78, "lr": 9.400668284022851e-05} +{"train_loss": 0.018584588542580605, "grad_norm": 0.10552562773227692, "global_step": 8345, "epoch": 78, "lr": 9.400527578592068e-05} +{"train_loss": 0.017412666231393814, "grad_norm": 0.04030786454677582, "global_step": 8346, "epoch": 78, "lr": 9.400386857699745e-05} +{"train_loss": 0.016294173896312714, "grad_norm": 0.06547557562589645, "global_step": 8347, "epoch": 78, "lr": 9.400246121346377e-05} +{"train_loss": 0.016364652663469315, "grad_norm": 0.10837645083665848, "global_step": 8348, "epoch": 78, "lr": 9.400105369532456e-05} +{"train_loss": 0.016226807609200478, "grad_norm": 0.045701440423727036, "global_step": 8349, "epoch": 78, "lr": 9.39996460225848e-05} +{"train_loss": 0.016715548932552338, "grad_norm": 0.09548506140708923, "global_step": 8350, "epoch": 78, "lr": 9.39982381952494e-05} +{"train_loss": 0.016173657029867172, "grad_norm": 0.04915929213166237, "global_step": 8351, "epoch": 78, "lr": 9.399683021332334e-05} +{"train_loss": 0.018865959718823433, "grad_norm": 0.04377733916044235, "global_step": 8352, "epoch": 78, "lr": 9.399542207681154e-05} +{"train_loss": 0.020434297621250153, "grad_norm": 0.07039928436279297, "global_step": 8353, "epoch": 78, "lr": 9.399401378571897e-05} +{"train_loss": 0.0167177002876997, "grad_norm": 0.07588961720466614, "global_step": 8354, "epoch": 78, "lr": 9.399260534005056e-05} +{"train_loss": 0.020815763622522354, "grad_norm": 0.05318768694996834, "global_step": 8355, "epoch": 78, "lr": 9.399119673981129e-05} +{"train_loss": 0.017227165400981903, "grad_norm": 0.05213486775755882, "global_step": 8356, "epoch": 78, "lr": 9.398978798500606e-05} +{"train_loss": 0.020804986357688904, "grad_norm": 0.06322043389081955, "global_step": 8357, "epoch": 78, "lr": 9.398837907563986e-05} +{"train_loss": 0.019571375101804733, "grad_norm": 0.04995725676417351, "global_step": 8358, "epoch": 78, "lr": 9.398697001171763e-05} +{"train_loss": 0.017736729234457016, "grad_norm": 0.05647719278931618, "global_step": 8359, "epoch": 78, "lr": 9.39855607932443e-05} +{"train_loss": 0.01847386173903942, "grad_norm": 0.07043870538473129, "global_step": 8360, "epoch": 78, "lr": 9.398415142022486e-05} +{"train_loss": 0.018219400197267532, "grad_norm": 0.05589469149708748, "global_step": 8361, "epoch": 78, "lr": 9.398274189266421e-05} +{"train_loss": 0.017360787838697433, "grad_norm": 0.04610222950577736, "global_step": 8362, "epoch": 78, "lr": 9.398133221056736e-05} +{"train_loss": 0.018865443766117096, "grad_norm": 0.07095949351787567, "global_step": 8363, "epoch": 78, "lr": 9.397992237393922e-05} +{"train_loss": 0.01630229502916336, "grad_norm": 0.11825299263000488, "global_step": 8364, "epoch": 78, "lr": 9.397851238278476e-05} +{"train_loss": 0.01632685959339142, "grad_norm": 0.07143717259168625, "global_step": 8365, "epoch": 78, "lr": 9.397710223710894e-05} +{"train_loss": 0.01794867217540741, "grad_norm": 0.07494165748357773, "global_step": 8366, "epoch": 78, "lr": 9.397569193691669e-05} +{"train_loss": 0.01762019470334053, "grad_norm": 0.04855581372976303, "global_step": 8367, "epoch": 78, "lr": 9.3974281482213e-05} +{"train_loss": 0.015441827476024628, "grad_norm": 0.0738239660859108, "global_step": 8368, "epoch": 78, "lr": 9.397287087300282e-05} +{"train_loss": 0.01649715006351471, "grad_norm": 0.09698925167322159, "global_step": 8369, "epoch": 78, "lr": 9.397146010929108e-05} +{"train_loss": 0.01606929674744606, "grad_norm": 0.07676206529140472, "global_step": 8370, "epoch": 78, "lr": 9.397004919108275e-05} +{"train_loss": 0.016990870237350464, "grad_norm": 0.06279031187295914, "global_step": 8371, "epoch": 78, "lr": 9.396863811838279e-05} +{"train_loss": 0.018750280141830444, "grad_norm": 0.06168309226632118, "global_step": 8372, "epoch": 78, "lr": 9.396722689119617e-05} +{"train_loss": 0.018150884034287818, "grad_norm": 0.0814322754740715, "global_step": 8373, "epoch": 78, "lr": 9.396581550952781e-05, "val_loss": 0.019276274368166924} +{"train_loss": 0.020810142159461975, "grad_norm": 0.1544864922761917, "global_step": 8374, "epoch": 79, "lr": 9.396440397338272e-05} +{"train_loss": 0.01598879136145115, "grad_norm": 0.07868766039609909, "global_step": 8375, "epoch": 79, "lr": 9.396299228276582e-05} +{"train_loss": 0.016097908839583397, "grad_norm": 0.05904658883810043, "global_step": 8376, "epoch": 79, "lr": 9.396158043768208e-05} +{"train_loss": 0.01815083809196949, "grad_norm": 0.048973239958286285, "global_step": 8377, "epoch": 79, "lr": 9.396016843813649e-05} +{"train_loss": 0.015101850032806396, "grad_norm": 0.06543806195259094, "global_step": 8378, "epoch": 79, "lr": 9.395875628413395e-05} +{"train_loss": 0.019194286316633224, "grad_norm": 0.04995962604880333, "global_step": 8379, "epoch": 79, "lr": 9.395734397567948e-05} +{"train_loss": 0.0181039460003376, "grad_norm": 0.09901466965675354, "global_step": 8380, "epoch": 79, "lr": 9.395593151277802e-05} +{"train_loss": 0.017683371901512146, "grad_norm": 0.114659883081913, "global_step": 8381, "epoch": 79, "lr": 9.395451889543453e-05} +{"train_loss": 0.015961095690727234, "grad_norm": 0.08980470150709152, "global_step": 8382, "epoch": 79, "lr": 9.395310612365397e-05} +{"train_loss": 0.018931066617369652, "grad_norm": 0.05998630076646805, "global_step": 8383, "epoch": 79, "lr": 9.395169319744132e-05} +{"train_loss": 0.017111968249082565, "grad_norm": 0.1094503179192543, "global_step": 8384, "epoch": 79, "lr": 9.395028011680152e-05} +{"train_loss": 0.01733647845685482, "grad_norm": 0.0921788364648819, "global_step": 8385, "epoch": 79, "lr": 9.394886688173956e-05} +{"train_loss": 0.01658451184630394, "grad_norm": 0.13100332021713257, "global_step": 8386, "epoch": 79, "lr": 9.39474534922604e-05} +{"train_loss": 0.017737364396452904, "grad_norm": 0.09517141431570053, "global_step": 8387, "epoch": 79, "lr": 9.394603994836899e-05} +{"train_loss": 0.020159699022769928, "grad_norm": 0.05195870250463486, "global_step": 8388, "epoch": 79, "lr": 9.394462625007031e-05} +{"train_loss": 0.01762271486222744, "grad_norm": 0.04634591192007065, "global_step": 8389, "epoch": 79, "lr": 9.394321239736932e-05} +{"train_loss": 0.01767168752849102, "grad_norm": 0.0745922327041626, "global_step": 8390, "epoch": 79, "lr": 9.3941798390271e-05} +{"train_loss": 0.0176185742020607, "grad_norm": 0.07672488689422607, "global_step": 8391, "epoch": 79, "lr": 9.394038422878031e-05} +{"train_loss": 0.017792725935578346, "grad_norm": 0.0822710394859314, "global_step": 8392, "epoch": 79, "lr": 9.393896991290221e-05} +{"train_loss": 0.019413333386182785, "grad_norm": 0.06506640464067459, "global_step": 8393, "epoch": 79, "lr": 9.393755544264169e-05} +{"train_loss": 0.017991814762353897, "grad_norm": 0.09964616596698761, "global_step": 8394, "epoch": 79, "lr": 9.39361408180037e-05} +{"train_loss": 0.020814619958400726, "grad_norm": 0.12622720003128052, "global_step": 8395, "epoch": 79, "lr": 9.393472603899322e-05} +{"train_loss": 0.017031967639923096, "grad_norm": 0.06699401885271072, "global_step": 8396, "epoch": 79, "lr": 9.393331110561522e-05} +{"train_loss": 0.019188761711120605, "grad_norm": 0.07700490951538086, "global_step": 8397, "epoch": 79, "lr": 9.393189601787466e-05} +{"train_loss": 0.02311413548886776, "grad_norm": 0.09236150979995728, "global_step": 8398, "epoch": 79, "lr": 9.393048077577655e-05} +{"train_loss": 0.019615957513451576, "grad_norm": 0.08063480257987976, "global_step": 8399, "epoch": 79, "lr": 9.392906537932582e-05} +{"train_loss": 0.01881498098373413, "grad_norm": 0.08871272206306458, "global_step": 8400, "epoch": 79, "lr": 9.392764982852745e-05} +{"train_loss": 0.017611242830753326, "grad_norm": 0.0632929801940918, "global_step": 8401, "epoch": 79, "lr": 9.392623412338643e-05} +{"train_loss": 0.01849343441426754, "grad_norm": 0.09412519633769989, "global_step": 8402, "epoch": 79, "lr": 9.392481826390773e-05} +{"train_loss": 0.01822451315820217, "grad_norm": 0.10698479413986206, "global_step": 8403, "epoch": 79, "lr": 9.392340225009631e-05} +{"train_loss": 0.018295112997293472, "grad_norm": 0.06685883551836014, "global_step": 8404, "epoch": 79, "lr": 9.392198608195717e-05} +{"train_loss": 0.015663431957364082, "grad_norm": 0.03767399489879608, "global_step": 8405, "epoch": 79, "lr": 9.392056975949526e-05} +{"train_loss": 0.01610356941819191, "grad_norm": 0.0951055958867073, "global_step": 8406, "epoch": 79, "lr": 9.391915328271557e-05} +{"train_loss": 0.015681259334087372, "grad_norm": 0.09827335178852081, "global_step": 8407, "epoch": 79, "lr": 9.391773665162309e-05} +{"train_loss": 0.01933388039469719, "grad_norm": 0.1071801632642746, "global_step": 8408, "epoch": 79, "lr": 9.391631986622277e-05} +{"train_loss": 0.019867319613695145, "grad_norm": 0.11242833733558655, "global_step": 8409, "epoch": 79, "lr": 9.391490292651961e-05} +{"train_loss": 0.017585521563887596, "grad_norm": 0.06935132294893265, "global_step": 8410, "epoch": 79, "lr": 9.391348583251855e-05} +{"train_loss": 0.017861925065517426, "grad_norm": 0.1603899449110031, "global_step": 8411, "epoch": 79, "lr": 9.391206858422464e-05} +{"train_loss": 0.01781216636300087, "grad_norm": 0.12620361149311066, "global_step": 8412, "epoch": 79, "lr": 9.391065118164279e-05} +{"train_loss": 0.018354522064328194, "grad_norm": 0.0749102309346199, "global_step": 8413, "epoch": 79, "lr": 9.390923362477801e-05} +{"train_loss": 0.01607365906238556, "grad_norm": 0.12040814757347107, "global_step": 8414, "epoch": 79, "lr": 9.390781591363529e-05} +{"train_loss": 0.018548913300037384, "grad_norm": 0.06861378997564316, "global_step": 8415, "epoch": 79, "lr": 9.39063980482196e-05} +{"train_loss": 0.016081329435110092, "grad_norm": 0.0424867644906044, "global_step": 8416, "epoch": 79, "lr": 9.390498002853592e-05} +{"train_loss": 0.01937926933169365, "grad_norm": 0.12959609925746918, "global_step": 8417, "epoch": 79, "lr": 9.390356185458922e-05} +{"train_loss": 0.018613889813423157, "grad_norm": 0.0841759666800499, "global_step": 8418, "epoch": 79, "lr": 9.390214352638451e-05} +{"train_loss": 0.019579706713557243, "grad_norm": 0.05667976662516594, "global_step": 8419, "epoch": 79, "lr": 9.390072504392676e-05} +{"train_loss": 0.01874556764960289, "grad_norm": 0.09635627269744873, "global_step": 8420, "epoch": 79, "lr": 9.389930640722096e-05} +{"train_loss": 0.017782706767320633, "grad_norm": 0.15988512337207794, "global_step": 8421, "epoch": 79, "lr": 9.389788761627207e-05} +{"train_loss": 0.016305260360240936, "grad_norm": 0.08760777860879898, "global_step": 8422, "epoch": 79, "lr": 9.38964686710851e-05} +{"train_loss": 0.01622980460524559, "grad_norm": 0.057094868272542953, "global_step": 8423, "epoch": 79, "lr": 9.389504957166505e-05} +{"train_loss": 0.01726738177239895, "grad_norm": 0.10646667331457138, "global_step": 8424, "epoch": 79, "lr": 9.389363031801686e-05} +{"train_loss": 0.019776079803705215, "grad_norm": 0.11698324233293533, "global_step": 8425, "epoch": 79, "lr": 9.389221091014554e-05} +{"train_loss": 0.01791202276945114, "grad_norm": 0.08660347014665604, "global_step": 8426, "epoch": 79, "lr": 9.389079134805609e-05} +{"train_loss": 0.019309619441628456, "grad_norm": 0.058153197169303894, "global_step": 8427, "epoch": 79, "lr": 9.388937163175348e-05} +{"train_loss": 0.01888415217399597, "grad_norm": 0.0592745803296566, "global_step": 8428, "epoch": 79, "lr": 9.388795176124273e-05} +{"train_loss": 0.015997227281332016, "grad_norm": 0.05454493314027786, "global_step": 8429, "epoch": 79, "lr": 9.388653173652877e-05} +{"train_loss": 0.01890389434993267, "grad_norm": 0.15781667828559875, "global_step": 8430, "epoch": 79, "lr": 9.388511155761663e-05} +{"train_loss": 0.018014738336205482, "grad_norm": 0.08712788671255112, "global_step": 8431, "epoch": 79, "lr": 9.38836912245113e-05} +{"train_loss": 0.017081405967473984, "grad_norm": 0.05838276818394661, "global_step": 8432, "epoch": 79, "lr": 9.388227073721777e-05} +{"train_loss": 0.01923508010804653, "grad_norm": 0.08345134556293488, "global_step": 8433, "epoch": 79, "lr": 9.388085009574102e-05} +{"train_loss": 0.016520056873559952, "grad_norm": 0.0682934895157814, "global_step": 8434, "epoch": 79, "lr": 9.387942930008604e-05} +{"train_loss": 0.019338389858603477, "grad_norm": 0.05567670240998268, "global_step": 8435, "epoch": 79, "lr": 9.387800835025783e-05} +{"train_loss": 0.018749209120869637, "grad_norm": 0.15276286005973816, "global_step": 8436, "epoch": 79, "lr": 9.387658724626139e-05} +{"train_loss": 0.0190016720443964, "grad_norm": 0.10507635027170181, "global_step": 8437, "epoch": 79, "lr": 9.387516598810171e-05} +{"train_loss": 0.01965935528278351, "grad_norm": 0.06889154762029648, "global_step": 8438, "epoch": 79, "lr": 9.387374457578375e-05} +{"train_loss": 0.01849975436925888, "grad_norm": 0.054022327065467834, "global_step": 8439, "epoch": 79, "lr": 9.387232300931255e-05} +{"train_loss": 0.018354523926973343, "grad_norm": 0.09646003693342209, "global_step": 8440, "epoch": 79, "lr": 9.387090128869309e-05} +{"train_loss": 0.0178663469851017, "grad_norm": 0.08058527857065201, "global_step": 8441, "epoch": 79, "lr": 9.386947941393035e-05} +{"train_loss": 0.016376320272684097, "grad_norm": 0.07503139972686768, "global_step": 8442, "epoch": 79, "lr": 9.386805738502935e-05} +{"train_loss": 0.02203909121453762, "grad_norm": 0.05539527162909508, "global_step": 8443, "epoch": 79, "lr": 9.386663520199506e-05} +{"train_loss": 0.020182989537715912, "grad_norm": 0.061147239059209824, "global_step": 8444, "epoch": 79, "lr": 9.38652128648325e-05} +{"train_loss": 0.019625822082161903, "grad_norm": 0.04385019466280937, "global_step": 8445, "epoch": 79, "lr": 9.386379037354666e-05} +{"train_loss": 0.020142745226621628, "grad_norm": 0.07215786725282669, "global_step": 8446, "epoch": 79, "lr": 9.386236772814252e-05} +{"train_loss": 0.018439646810293198, "grad_norm": 0.0661548301577568, "global_step": 8447, "epoch": 79, "lr": 9.386094492862512e-05} +{"train_loss": 0.018798507750034332, "grad_norm": 0.06255868077278137, "global_step": 8448, "epoch": 79, "lr": 9.385952197499942e-05} +{"train_loss": 0.016701504588127136, "grad_norm": 0.050204452127218246, "global_step": 8449, "epoch": 79, "lr": 9.385809886727044e-05} +{"train_loss": 0.016799665987491608, "grad_norm": 0.04135956987738609, "global_step": 8450, "epoch": 79, "lr": 9.385667560544316e-05} +{"train_loss": 0.018649637699127197, "grad_norm": 0.04102707654237747, "global_step": 8451, "epoch": 79, "lr": 9.38552521895226e-05} +{"train_loss": 0.01828598603606224, "grad_norm": 0.08979272097349167, "global_step": 8452, "epoch": 79, "lr": 9.385382861951377e-05} +{"train_loss": 0.015664799138903618, "grad_norm": 0.04764882102608681, "global_step": 8453, "epoch": 79, "lr": 9.385240489542164e-05} +{"train_loss": 0.01998225226998329, "grad_norm": 0.10579950362443924, "global_step": 8454, "epoch": 79, "lr": 9.385098101725124e-05} +{"train_loss": 0.01959589496254921, "grad_norm": 0.07204775512218475, "global_step": 8455, "epoch": 79, "lr": 9.384955698500757e-05} +{"train_loss": 0.01973181962966919, "grad_norm": 0.09783195704221725, "global_step": 8456, "epoch": 79, "lr": 9.384813279869561e-05} +{"train_loss": 0.01812904141843319, "grad_norm": 0.05264505743980408, "global_step": 8457, "epoch": 79, "lr": 9.384670845832039e-05} +{"train_loss": 0.01788623258471489, "grad_norm": 0.12571747601032257, "global_step": 8458, "epoch": 79, "lr": 9.38452839638869e-05} +{"train_loss": 0.018078599125146866, "grad_norm": 0.04894918203353882, "global_step": 8459, "epoch": 79, "lr": 9.384385931540016e-05} +{"train_loss": 0.015190914273262024, "grad_norm": 0.08439535647630692, "global_step": 8460, "epoch": 79, "lr": 9.384243451286517e-05} +{"train_loss": 0.020793087780475616, "grad_norm": 0.07138112187385559, "global_step": 8461, "epoch": 79, "lr": 9.384100955628691e-05} +{"train_loss": 0.01769348978996277, "grad_norm": 0.05481618642807007, "global_step": 8462, "epoch": 79, "lr": 9.383958444567042e-05} +{"train_loss": 0.01681380718946457, "grad_norm": 0.05272841081023216, "global_step": 8463, "epoch": 79, "lr": 9.383815918102069e-05} +{"train_loss": 0.017706135287880898, "grad_norm": 0.10107484459877014, "global_step": 8464, "epoch": 79, "lr": 9.383673376234274e-05} +{"train_loss": 0.017182402312755585, "grad_norm": 0.05609507858753204, "global_step": 8465, "epoch": 79, "lr": 9.383530818964157e-05} +{"train_loss": 0.01757647655904293, "grad_norm": 0.1536223143339157, "global_step": 8466, "epoch": 79, "lr": 9.383388246292219e-05} +{"train_loss": 0.019765077158808708, "grad_norm": 0.10047547519207001, "global_step": 8467, "epoch": 79, "lr": 9.38324565821896e-05} +{"train_loss": 0.020116101950407028, "grad_norm": 0.06255492568016052, "global_step": 8468, "epoch": 79, "lr": 9.383103054744884e-05} +{"train_loss": 0.018060175701975822, "grad_norm": 0.07608644664287567, "global_step": 8469, "epoch": 79, "lr": 9.38296043587049e-05} +{"train_loss": 0.017737384885549545, "grad_norm": 0.06374695897102356, "global_step": 8470, "epoch": 79, "lr": 9.382817801596277e-05} +{"train_loss": 0.020475536584854126, "grad_norm": 0.06121223047375679, "global_step": 8471, "epoch": 79, "lr": 9.382675151922748e-05} +{"train_loss": 0.017579710111021996, "grad_norm": 0.07110434025526047, "global_step": 8472, "epoch": 79, "lr": 9.382532486850406e-05} +{"train_loss": 0.017080068588256836, "grad_norm": 0.10155688971281052, "global_step": 8473, "epoch": 79, "lr": 9.382389806379751e-05} +{"train_loss": 0.018181104212999344, "grad_norm": 0.06287608295679092, "global_step": 8474, "epoch": 79, "lr": 9.382247110511284e-05} +{"train_loss": 0.018127702176570892, "grad_norm": 0.054091960191726685, "global_step": 8475, "epoch": 79, "lr": 9.382104399245506e-05} +{"train_loss": 0.016308516263961792, "grad_norm": 0.08054791390895844, "global_step": 8476, "epoch": 79, "lr": 9.38196167258292e-05} +{"train_loss": 0.016730107367038727, "grad_norm": 0.10147279500961304, "global_step": 8477, "epoch": 79, "lr": 9.381818930524024e-05} +{"train_loss": 0.017630435526371002, "grad_norm": 0.09341636300086975, "global_step": 8478, "epoch": 79, "lr": 9.381676173069324e-05} +{"train_loss": 0.018134525783781736, "grad_norm": 0.06769830733537674, "global_step": 8479, "epoch": 79, "lr": 9.381533400219318e-05, "val_loss": 0.018909774720668793} +{"train_loss": 0.017460552975535393, "grad_norm": 0.04189024493098259, "global_step": 8480, "epoch": 80, "lr": 9.38139061197451e-05} +{"train_loss": 0.01813199557363987, "grad_norm": 0.050650857388973236, "global_step": 8481, "epoch": 80, "lr": 9.381247808335401e-05} +{"train_loss": 0.017718840390443802, "grad_norm": 0.05373218283057213, "global_step": 8482, "epoch": 80, "lr": 9.381104989302492e-05} +{"train_loss": 0.017554722726345062, "grad_norm": 0.09459683299064636, "global_step": 8483, "epoch": 80, "lr": 9.380962154876285e-05} +{"train_loss": 0.018192321062088013, "grad_norm": 0.06576017290353775, "global_step": 8484, "epoch": 80, "lr": 9.380819305057284e-05} +{"train_loss": 0.017512787133455276, "grad_norm": 0.10194964706897736, "global_step": 8485, "epoch": 80, "lr": 9.380676439845987e-05} +{"train_loss": 0.02231248840689659, "grad_norm": 0.08412392437458038, "global_step": 8486, "epoch": 80, "lr": 9.380533559242898e-05} +{"train_loss": 0.016658414155244827, "grad_norm": 0.04383796080946922, "global_step": 8487, "epoch": 80, "lr": 9.380390663248522e-05} +{"train_loss": 0.015113046392798424, "grad_norm": 0.05448043346405029, "global_step": 8488, "epoch": 80, "lr": 9.380247751863356e-05} +{"train_loss": 0.020096393302083015, "grad_norm": 0.079445019364357, "global_step": 8489, "epoch": 80, "lr": 9.380104825087905e-05} +{"train_loss": 0.015962522476911545, "grad_norm": 0.07947826385498047, "global_step": 8490, "epoch": 80, "lr": 9.37996188292267e-05} +{"train_loss": 0.017779089510440826, "grad_norm": 0.0676540657877922, "global_step": 8491, "epoch": 80, "lr": 9.379818925368153e-05} +{"train_loss": 0.021514181047677994, "grad_norm": 0.06139549985527992, "global_step": 8492, "epoch": 80, "lr": 9.379675952424858e-05} +{"train_loss": 0.01731705479323864, "grad_norm": 0.06759298592805862, "global_step": 8493, "epoch": 80, "lr": 9.379532964093286e-05} +{"train_loss": 0.01932281255722046, "grad_norm": 0.057031482458114624, "global_step": 8494, "epoch": 80, "lr": 9.37938996037394e-05} +{"train_loss": 0.017638543620705605, "grad_norm": 0.14018003642559052, "global_step": 8495, "epoch": 80, "lr": 9.379246941267323e-05} +{"train_loss": 0.01760639064013958, "grad_norm": 0.07648254930973053, "global_step": 8496, "epoch": 80, "lr": 9.379103906773936e-05} +{"train_loss": 0.02095022052526474, "grad_norm": 0.11757019907236099, "global_step": 8497, "epoch": 80, "lr": 9.37896085689428e-05} +{"train_loss": 0.019142311066389084, "grad_norm": 0.06290510296821594, "global_step": 8498, "epoch": 80, "lr": 9.378817791628864e-05} +{"train_loss": 0.017096836119890213, "grad_norm": 0.06245414540171623, "global_step": 8499, "epoch": 80, "lr": 9.378674710978185e-05} +{"train_loss": 0.01672174409031868, "grad_norm": 0.1190832331776619, "global_step": 8500, "epoch": 80, "lr": 9.378531614942748e-05} +{"train_loss": 0.015131289139389992, "grad_norm": 0.11494588106870651, "global_step": 8501, "epoch": 80, "lr": 9.378388503523052e-05} +{"train_loss": 0.020126190036535263, "grad_norm": 0.14829829335212708, "global_step": 8502, "epoch": 80, "lr": 9.378245376719605e-05} +{"train_loss": 0.015404878184199333, "grad_norm": 0.08378945291042328, "global_step": 8503, "epoch": 80, "lr": 9.378102234532908e-05} +{"train_loss": 0.016355177387595177, "grad_norm": 0.1223752573132515, "global_step": 8504, "epoch": 80, "lr": 9.377959076963464e-05} +{"train_loss": 0.01956712082028389, "grad_norm": 0.059268519282341, "global_step": 8505, "epoch": 80, "lr": 9.377815904011775e-05} +{"train_loss": 0.015891792252659798, "grad_norm": 0.06405763328075409, "global_step": 8506, "epoch": 80, "lr": 9.377672715678345e-05} +{"train_loss": 0.015568342059850693, "grad_norm": 0.06555071473121643, "global_step": 8507, "epoch": 80, "lr": 9.377529511963678e-05} +{"train_loss": 0.017490867525339127, "grad_norm": 0.09501943737268448, "global_step": 8508, "epoch": 80, "lr": 9.377386292868275e-05} +{"train_loss": 0.017872963100671768, "grad_norm": 0.07708297669887543, "global_step": 8509, "epoch": 80, "lr": 9.377243058392639e-05} +{"train_loss": 0.01691444031894207, "grad_norm": 0.0849621444940567, "global_step": 8510, "epoch": 80, "lr": 9.377099808537276e-05} +{"train_loss": 0.01952098123729229, "grad_norm": 0.060504354536533356, "global_step": 8511, "epoch": 80, "lr": 9.376956543302689e-05} +{"train_loss": 0.01858118548989296, "grad_norm": 0.04309941455721855, "global_step": 8512, "epoch": 80, "lr": 9.376813262689378e-05} +{"train_loss": 0.02001851424574852, "grad_norm": 0.08387843519449234, "global_step": 8513, "epoch": 80, "lr": 9.376669966697848e-05} +{"train_loss": 0.018067751079797745, "grad_norm": 0.06783134490251541, "global_step": 8514, "epoch": 80, "lr": 9.376526655328606e-05} +{"train_loss": 0.018378086388111115, "grad_norm": 0.09947273135185242, "global_step": 8515, "epoch": 80, "lr": 9.37638332858215e-05} +{"train_loss": 0.019416876137256622, "grad_norm": 0.1439390480518341, "global_step": 8516, "epoch": 80, "lr": 9.376239986458987e-05} +{"train_loss": 0.01730583794414997, "grad_norm": 0.0617198720574379, "global_step": 8517, "epoch": 80, "lr": 9.376096628959621e-05} +{"train_loss": 0.01736576482653618, "grad_norm": 0.0672251507639885, "global_step": 8518, "epoch": 80, "lr": 9.375953256084554e-05} +{"train_loss": 0.0209085363894701, "grad_norm": 0.14170928299427032, "global_step": 8519, "epoch": 80, "lr": 9.37580986783429e-05} +{"train_loss": 0.020848644897341728, "grad_norm": 0.10137306153774261, "global_step": 8520, "epoch": 80, "lr": 9.375666464209333e-05} +{"train_loss": 0.01790546253323555, "grad_norm": 0.09391409903764725, "global_step": 8521, "epoch": 80, "lr": 9.375523045210187e-05} +{"train_loss": 0.01893727108836174, "grad_norm": 0.08170843869447708, "global_step": 8522, "epoch": 80, "lr": 9.375379610837356e-05} +{"train_loss": 0.018163559958338737, "grad_norm": 0.07156392931938171, "global_step": 8523, "epoch": 80, "lr": 9.375236161091345e-05} +{"train_loss": 0.016017839312553406, "grad_norm": 0.08132018148899078, "global_step": 8524, "epoch": 80, "lr": 9.375092695972655e-05} +{"train_loss": 0.020405031740665436, "grad_norm": 0.13612200319766998, "global_step": 8525, "epoch": 80, "lr": 9.374949215481792e-05} +{"train_loss": 0.017014916986227036, "grad_norm": 0.08416015654802322, "global_step": 8526, "epoch": 80, "lr": 9.374805719619262e-05} +{"train_loss": 0.01776479370892048, "grad_norm": 0.05473366007208824, "global_step": 8527, "epoch": 80, "lr": 9.374662208385565e-05} +{"train_loss": 0.017262838780879974, "grad_norm": 0.10513493418693542, "global_step": 8528, "epoch": 80, "lr": 9.374518681781208e-05} +{"train_loss": 0.018425367772579193, "grad_norm": 0.09160467237234116, "global_step": 8529, "epoch": 80, "lr": 9.374375139806696e-05} +{"train_loss": 0.016030747443437576, "grad_norm": 0.05001460760831833, "global_step": 8530, "epoch": 80, "lr": 9.374231582462532e-05} +{"train_loss": 0.016620302572846413, "grad_norm": 0.06690748035907745, "global_step": 8531, "epoch": 80, "lr": 9.37408800974922e-05} +{"train_loss": 0.016696112230420113, "grad_norm": 0.130464568734169, "global_step": 8532, "epoch": 80, "lr": 9.373944421667265e-05} +{"train_loss": 0.020364288240671158, "grad_norm": 0.06313705444335938, "global_step": 8533, "epoch": 80, "lr": 9.373800818217171e-05} +{"train_loss": 0.016280248761177063, "grad_norm": 0.15173563361167908, "global_step": 8534, "epoch": 80, "lr": 9.373657199399443e-05} +{"train_loss": 0.022120650857686996, "grad_norm": 0.058739110827445984, "global_step": 8535, "epoch": 80, "lr": 9.373513565214587e-05} +{"train_loss": 0.019131017848849297, "grad_norm": 0.07028491050004959, "global_step": 8536, "epoch": 80, "lr": 9.373369915663103e-05} +{"train_loss": 0.01644914783537388, "grad_norm": 0.045862723141908646, "global_step": 8537, "epoch": 80, "lr": 9.373226250745504e-05} +{"train_loss": 0.019015926867723465, "grad_norm": 0.0835656002163887, "global_step": 8538, "epoch": 80, "lr": 9.373082570462287e-05} +{"train_loss": 0.018133262172341347, "grad_norm": 0.05383116379380226, "global_step": 8539, "epoch": 80, "lr": 9.372938874813959e-05} +{"train_loss": 0.016017550602555275, "grad_norm": 0.08211600035429001, "global_step": 8540, "epoch": 80, "lr": 9.372795163801027e-05} +{"train_loss": 0.018536336719989777, "grad_norm": 0.12057787925004959, "global_step": 8541, "epoch": 80, "lr": 9.372651437423994e-05} +{"train_loss": 0.017259173095226288, "grad_norm": 0.10058414191007614, "global_step": 8542, "epoch": 80, "lr": 9.372507695683366e-05} +{"train_loss": 0.017364922910928726, "grad_norm": 0.04651946574449539, "global_step": 8543, "epoch": 80, "lr": 9.372363938579646e-05} +{"train_loss": 0.017263710498809814, "grad_norm": 0.07125330716371536, "global_step": 8544, "epoch": 80, "lr": 9.372220166113341e-05} +{"train_loss": 0.017316143959760666, "grad_norm": 0.07241223752498627, "global_step": 8545, "epoch": 80, "lr": 9.372076378284958e-05} +{"train_loss": 0.02092389203608036, "grad_norm": 0.08058782666921616, "global_step": 8546, "epoch": 80, "lr": 9.371932575094997e-05} +{"train_loss": 0.019766472280025482, "grad_norm": 0.1799928992986679, "global_step": 8547, "epoch": 80, "lr": 9.371788756543967e-05} +{"train_loss": 0.01597313955426216, "grad_norm": 0.05879030376672745, "global_step": 8548, "epoch": 80, "lr": 9.371644922632374e-05} +{"train_loss": 0.02125648595392704, "grad_norm": 0.07271850109100342, "global_step": 8549, "epoch": 80, "lr": 9.37150107336072e-05} +{"train_loss": 0.019789811223745346, "grad_norm": 0.07687867432832718, "global_step": 8550, "epoch": 80, "lr": 9.371357208729514e-05} +{"train_loss": 0.021995704621076584, "grad_norm": 0.05282425880432129, "global_step": 8551, "epoch": 80, "lr": 9.371213328739258e-05} +{"train_loss": 0.01927132159471512, "grad_norm": 0.07865213602781296, "global_step": 8552, "epoch": 80, "lr": 9.371069433390462e-05} +{"train_loss": 0.020330829545855522, "grad_norm": 0.057309336960315704, "global_step": 8553, "epoch": 80, "lr": 9.370925522683627e-05} +{"train_loss": 0.017639925703406334, "grad_norm": 0.05637001246213913, "global_step": 8554, "epoch": 80, "lr": 9.37078159661926e-05} +{"train_loss": 0.016939107328653336, "grad_norm": 0.06090957671403885, "global_step": 8555, "epoch": 80, "lr": 9.370637655197869e-05} +{"train_loss": 0.018307620659470558, "grad_norm": 0.09257979691028595, "global_step": 8556, "epoch": 80, "lr": 9.370493698419958e-05} +{"train_loss": 0.01822931319475174, "grad_norm": 0.13345904648303986, "global_step": 8557, "epoch": 80, "lr": 9.370349726286032e-05} +{"train_loss": 0.01686117798089981, "grad_norm": 0.0922941192984581, "global_step": 8558, "epoch": 80, "lr": 9.370205738796598e-05} +{"train_loss": 0.017584167420864105, "grad_norm": 0.045987728983163834, "global_step": 8559, "epoch": 80, "lr": 9.37006173595216e-05} +{"train_loss": 0.016490843147039413, "grad_norm": 0.09974781423807144, "global_step": 8560, "epoch": 80, "lr": 9.369917717753227e-05} +{"train_loss": 0.017354100942611694, "grad_norm": 0.04324401170015335, "global_step": 8561, "epoch": 80, "lr": 9.369773684200304e-05} +{"train_loss": 0.016529887914657593, "grad_norm": 0.037850044667720795, "global_step": 8562, "epoch": 80, "lr": 9.369629635293897e-05} +{"train_loss": 0.017971694469451904, "grad_norm": 0.08140315860509872, "global_step": 8563, "epoch": 80, "lr": 9.36948557103451e-05} +{"train_loss": 0.017666980624198914, "grad_norm": 0.1294904500246048, "global_step": 8564, "epoch": 80, "lr": 9.369341491422652e-05} +{"train_loss": 0.017047161236405373, "grad_norm": 0.04686414450407028, "global_step": 8565, "epoch": 80, "lr": 9.369197396458828e-05} +{"train_loss": 0.016962997615337372, "grad_norm": 0.07007133215665817, "global_step": 8566, "epoch": 80, "lr": 9.369053286143544e-05} +{"train_loss": 0.018742162734270096, "grad_norm": 0.11788057535886765, "global_step": 8567, "epoch": 80, "lr": 9.368909160477309e-05} +{"train_loss": 0.01985129341483116, "grad_norm": 0.07349827140569687, "global_step": 8568, "epoch": 80, "lr": 9.368765019460623e-05} +{"train_loss": 0.01970532536506653, "grad_norm": 0.1392677277326584, "global_step": 8569, "epoch": 80, "lr": 9.368620863094e-05} +{"train_loss": 0.017776265740394592, "grad_norm": 0.0715470090508461, "global_step": 8570, "epoch": 80, "lr": 9.368476691377944e-05} +{"train_loss": 0.016481302678585052, "grad_norm": 0.06939703226089478, "global_step": 8571, "epoch": 80, "lr": 9.368332504312958e-05} +{"train_loss": 0.017650729045271873, "grad_norm": 0.12743514776229858, "global_step": 8572, "epoch": 80, "lr": 9.368188301899552e-05} +{"train_loss": 0.020378507673740387, "grad_norm": 0.15413615107536316, "global_step": 8573, "epoch": 80, "lr": 9.368044084138232e-05} +{"train_loss": 0.01919371634721756, "grad_norm": 0.08111044019460678, "global_step": 8574, "epoch": 80, "lr": 9.367899851029504e-05} +{"train_loss": 0.016751522198319435, "grad_norm": 0.11121424287557602, "global_step": 8575, "epoch": 80, "lr": 9.367755602573878e-05} +{"train_loss": 0.020147033035755157, "grad_norm": 0.10775403678417206, "global_step": 8576, "epoch": 80, "lr": 9.367611338771856e-05} +{"train_loss": 0.020513644441962242, "grad_norm": 0.05434146896004677, "global_step": 8577, "epoch": 80, "lr": 9.367467059623948e-05} +{"train_loss": 0.016112470999360085, "grad_norm": 0.07118260860443115, "global_step": 8578, "epoch": 80, "lr": 9.367322765130659e-05} +{"train_loss": 0.018412981182336807, "grad_norm": 0.04437660425901413, "global_step": 8579, "epoch": 80, "lr": 9.367178455292499e-05} +{"train_loss": 0.014610077254474163, "grad_norm": 0.07124806940555573, "global_step": 8580, "epoch": 80, "lr": 9.367034130109973e-05} +{"train_loss": 0.017258180305361748, "grad_norm": 0.05613844096660614, "global_step": 8581, "epoch": 80, "lr": 9.366889789583586e-05} +{"train_loss": 0.018382105976343155, "grad_norm": 0.06508033722639084, "global_step": 8582, "epoch": 80, "lr": 9.36674543371385e-05} +{"train_loss": 0.01648862287402153, "grad_norm": 0.13952286541461945, "global_step": 8583, "epoch": 80, "lr": 9.366601062501268e-05} +{"train_loss": 0.018745645880699158, "grad_norm": 0.05180612951517105, "global_step": 8584, "epoch": 80, "lr": 9.366456675946349e-05} +{"train_loss": 0.01810980478371933, "grad_norm": 0.08345523476600647, "global_step": 8585, "epoch": 80, "lr": 9.366312274049602e-05, "train_batch_mse_error": 0.016529936343431473, "val_loss": 0.019443988800048828} +{"train_loss": 0.018503814935684204, "grad_norm": 0.08626557886600494, "global_step": 8586, "epoch": 81, "lr": 9.36616785681153e-05} +{"train_loss": 0.017230067402124405, "grad_norm": 0.07917240262031555, "global_step": 8587, "epoch": 81, "lr": 9.366023424232644e-05} +{"train_loss": 0.021209800615906715, "grad_norm": 0.049483902752399445, "global_step": 8588, "epoch": 81, "lr": 9.365878976313452e-05} +{"train_loss": 0.017419960349798203, "grad_norm": 0.05254818871617317, "global_step": 8589, "epoch": 81, "lr": 9.365734513054457e-05} +{"train_loss": 0.01796528697013855, "grad_norm": 0.05704701319336891, "global_step": 8590, "epoch": 81, "lr": 9.36559003445617e-05} +{"train_loss": 0.019711777567863464, "grad_norm": 0.13280177116394043, "global_step": 8591, "epoch": 81, "lr": 9.3654455405191e-05} +{"train_loss": 0.01699669286608696, "grad_norm": 0.050850484520196915, "global_step": 8592, "epoch": 81, "lr": 9.365301031243751e-05} +{"train_loss": 0.01842658780515194, "grad_norm": 0.10744025558233261, "global_step": 8593, "epoch": 81, "lr": 9.365156506630634e-05} +{"train_loss": 0.018985815346240997, "grad_norm": 0.0811866819858551, "global_step": 8594, "epoch": 81, "lr": 9.365011966680255e-05} +{"train_loss": 0.019244052469730377, "grad_norm": 0.05747111141681671, "global_step": 8595, "epoch": 81, "lr": 9.364867411393123e-05} +{"train_loss": 0.021007373929023743, "grad_norm": 0.07633857429027557, "global_step": 8596, "epoch": 81, "lr": 9.364722840769744e-05} +{"train_loss": 0.016432736068964005, "grad_norm": 0.08813043683767319, "global_step": 8597, "epoch": 81, "lr": 9.364578254810627e-05} +{"train_loss": 0.01562560722231865, "grad_norm": 0.04339755326509476, "global_step": 8598, "epoch": 81, "lr": 9.36443365351628e-05} +{"train_loss": 0.01726420223712921, "grad_norm": 0.06862349808216095, "global_step": 8599, "epoch": 81, "lr": 9.364289036887213e-05} +{"train_loss": 0.018502380698919296, "grad_norm": 0.12681375443935394, "global_step": 8600, "epoch": 81, "lr": 9.364144404923931e-05} +{"train_loss": 0.01738777756690979, "grad_norm": 0.04506449028849602, "global_step": 8601, "epoch": 81, "lr": 9.363999757626942e-05} +{"train_loss": 0.01913420855998993, "grad_norm": 0.06647591292858124, "global_step": 8602, "epoch": 81, "lr": 9.363855094996759e-05} +{"train_loss": 0.019769173115491867, "grad_norm": 0.06682708114385605, "global_step": 8603, "epoch": 81, "lr": 9.363710417033884e-05} +{"train_loss": 0.017950378358364105, "grad_norm": 0.14040391147136688, "global_step": 8604, "epoch": 81, "lr": 9.36356572373883e-05} +{"train_loss": 0.021127695217728615, "grad_norm": 0.10958299785852432, "global_step": 8605, "epoch": 81, "lr": 9.363421015112102e-05} +{"train_loss": 0.016964605078101158, "grad_norm": 0.05065613240003586, "global_step": 8606, "epoch": 81, "lr": 9.363276291154213e-05} +{"train_loss": 0.014511221088469028, "grad_norm": 0.07318764179944992, "global_step": 8607, "epoch": 81, "lr": 9.363131551865666e-05} +{"train_loss": 0.018902961164712906, "grad_norm": 0.05116911977529526, "global_step": 8608, "epoch": 81, "lr": 9.362986797246973e-05} +{"train_loss": 0.01749958097934723, "grad_norm": 0.05094250291585922, "global_step": 8609, "epoch": 81, "lr": 9.362842027298643e-05} +{"train_loss": 0.017169106751680374, "grad_norm": 0.06873390078544617, "global_step": 8610, "epoch": 81, "lr": 9.36269724202118e-05} +{"train_loss": 0.019584618508815765, "grad_norm": 0.05565248057246208, "global_step": 8611, "epoch": 81, "lr": 9.3625524414151e-05} +{"train_loss": 0.016814742237329483, "grad_norm": 0.09189432114362717, "global_step": 8612, "epoch": 81, "lr": 9.362407625480906e-05} +{"train_loss": 0.01641969569027424, "grad_norm": 0.04378453269600868, "global_step": 8613, "epoch": 81, "lr": 9.36226279421911e-05} +{"train_loss": 0.019925914704799652, "grad_norm": 0.056487780064344406, "global_step": 8614, "epoch": 81, "lr": 9.36211794763022e-05} +{"train_loss": 0.01867862045764923, "grad_norm": 0.12628842890262604, "global_step": 8615, "epoch": 81, "lr": 9.361973085714744e-05} +{"train_loss": 0.018722454085946083, "grad_norm": 0.05391343683004379, "global_step": 8616, "epoch": 81, "lr": 9.36182820847319e-05} +{"train_loss": 0.01889219880104065, "grad_norm": 0.1046392098069191, "global_step": 8617, "epoch": 81, "lr": 9.36168331590607e-05} +{"train_loss": 0.018224798142910004, "grad_norm": 0.13455741107463837, "global_step": 8618, "epoch": 81, "lr": 9.361538408013891e-05} +{"train_loss": 0.016972100362181664, "grad_norm": 0.12212319672107697, "global_step": 8619, "epoch": 81, "lr": 9.361393484797163e-05} +{"train_loss": 0.01989365927875042, "grad_norm": 0.07474149018526077, "global_step": 8620, "epoch": 81, "lr": 9.361248546256396e-05} +{"train_loss": 0.016557840630412102, "grad_norm": 0.06119345501065254, "global_step": 8621, "epoch": 81, "lr": 9.361103592392098e-05} +{"train_loss": 0.018593572080135345, "grad_norm": 0.09124014526605606, "global_step": 8622, "epoch": 81, "lr": 9.360958623204776e-05} +{"train_loss": 0.017672210931777954, "grad_norm": 0.05699002370238304, "global_step": 8623, "epoch": 81, "lr": 9.360813638694945e-05} +{"train_loss": 0.019806180149316788, "grad_norm": 0.12438029795885086, "global_step": 8624, "epoch": 81, "lr": 9.36066863886311e-05} +{"train_loss": 0.0168672576546669, "grad_norm": 0.04438711330294609, "global_step": 8625, "epoch": 81, "lr": 9.360523623709783e-05} +{"train_loss": 0.019749145954847336, "grad_norm": 0.06415686756372452, "global_step": 8626, "epoch": 81, "lr": 9.360378593235471e-05} +{"train_loss": 0.019742920994758606, "grad_norm": 0.10572215169668198, "global_step": 8627, "epoch": 81, "lr": 9.360233547440686e-05} +{"train_loss": 0.017888404428958893, "grad_norm": 0.08572668582201004, "global_step": 8628, "epoch": 81, "lr": 9.360088486325935e-05} +{"train_loss": 0.017332255840301514, "grad_norm": 0.09029381722211838, "global_step": 8629, "epoch": 81, "lr": 9.35994340989173e-05} +{"train_loss": 0.016268804669380188, "grad_norm": 0.13424170017242432, "global_step": 8630, "epoch": 81, "lr": 9.359798318138581e-05} +{"train_loss": 0.016366250813007355, "grad_norm": 0.08140596002340317, "global_step": 8631, "epoch": 81, "lr": 9.359653211066996e-05} +{"train_loss": 0.01614418253302574, "grad_norm": 0.04883173853158951, "global_step": 8632, "epoch": 81, "lr": 9.359508088677485e-05} +{"train_loss": 0.018367283046245575, "grad_norm": 0.04312954097986221, "global_step": 8633, "epoch": 81, "lr": 9.359362950970558e-05} +{"train_loss": 0.01822858676314354, "grad_norm": 0.10416494309902191, "global_step": 8634, "epoch": 81, "lr": 9.359217797946727e-05} +{"train_loss": 0.019368212670087814, "grad_norm": 0.12248461693525314, "global_step": 8635, "epoch": 81, "lr": 9.359072629606499e-05} +{"train_loss": 0.015682272613048553, "grad_norm": 0.04731360077857971, "global_step": 8636, "epoch": 81, "lr": 9.358927445950387e-05} +{"train_loss": 0.015650799497961998, "grad_norm": 0.03601217269897461, "global_step": 8637, "epoch": 81, "lr": 9.3587822469789e-05} +{"train_loss": 0.019182970747351646, "grad_norm": 0.10000874102115631, "global_step": 8638, "epoch": 81, "lr": 9.358637032692545e-05} +{"train_loss": 0.017933400347828865, "grad_norm": 0.045568302273750305, "global_step": 8639, "epoch": 81, "lr": 9.358491803091836e-05} +{"train_loss": 0.015531623736023903, "grad_norm": 0.049664054065942764, "global_step": 8640, "epoch": 81, "lr": 9.358346558177283e-05} +{"train_loss": 0.016809049993753433, "grad_norm": 0.038383323699235916, "global_step": 8641, "epoch": 81, "lr": 9.358201297949396e-05} +{"train_loss": 0.019821567460894585, "grad_norm": 0.07764515280723572, "global_step": 8642, "epoch": 81, "lr": 9.358056022408684e-05} +{"train_loss": 0.016269585117697716, "grad_norm": 0.08488986641168594, "global_step": 8643, "epoch": 81, "lr": 9.35791073155566e-05} +{"train_loss": 0.01690601371228695, "grad_norm": 0.06623873859643936, "global_step": 8644, "epoch": 81, "lr": 9.35776542539083e-05} +{"train_loss": 0.018976783379912376, "grad_norm": 0.1302628517150879, "global_step": 8645, "epoch": 81, "lr": 9.35762010391471e-05} +{"train_loss": 0.019334083423018456, "grad_norm": 0.11102475225925446, "global_step": 8646, "epoch": 81, "lr": 9.357474767127808e-05} +{"train_loss": 0.018523383885622025, "grad_norm": 0.07847140729427338, "global_step": 8647, "epoch": 81, "lr": 9.357329415030632e-05} +{"train_loss": 0.02108481153845787, "grad_norm": 0.08898574858903885, "global_step": 8648, "epoch": 81, "lr": 9.357184047623698e-05} +{"train_loss": 0.01960125006735325, "grad_norm": 0.05825083330273628, "global_step": 8649, "epoch": 81, "lr": 9.357038664907514e-05} +{"train_loss": 0.018139634281396866, "grad_norm": 0.06102880463004112, "global_step": 8650, "epoch": 81, "lr": 9.356893266882589e-05} +{"train_loss": 0.019574418663978577, "grad_norm": 0.15948385000228882, "global_step": 8651, "epoch": 81, "lr": 9.356747853549437e-05} +{"train_loss": 0.018451275303959846, "grad_norm": 0.06247705593705177, "global_step": 8652, "epoch": 81, "lr": 9.356602424908568e-05} +{"train_loss": 0.01795594021677971, "grad_norm": 0.11882218718528748, "global_step": 8653, "epoch": 81, "lr": 9.356456980960492e-05} +{"train_loss": 0.018397845327854156, "grad_norm": 0.04732166975736618, "global_step": 8654, "epoch": 81, "lr": 9.356311521705723e-05} +{"train_loss": 0.017750363796949387, "grad_norm": 0.0641668289899826, "global_step": 8655, "epoch": 81, "lr": 9.356166047144768e-05} +{"train_loss": 0.01795724406838417, "grad_norm": 0.07020308077335358, "global_step": 8656, "epoch": 81, "lr": 9.35602055727814e-05} +{"train_loss": 0.01633503846824169, "grad_norm": 0.042740464210510254, "global_step": 8657, "epoch": 81, "lr": 9.35587505210635e-05} +{"train_loss": 0.019202718511223793, "grad_norm": 0.0778779610991478, "global_step": 8658, "epoch": 81, "lr": 9.355729531629908e-05} +{"train_loss": 0.016530903056263924, "grad_norm": 0.14352497458457947, "global_step": 8659, "epoch": 81, "lr": 9.35558399584933e-05} +{"train_loss": 0.015911364927887917, "grad_norm": 0.13489890098571777, "global_step": 8660, "epoch": 81, "lr": 9.355438444765122e-05} +{"train_loss": 0.015726327896118164, "grad_norm": 0.069477379322052, "global_step": 8661, "epoch": 81, "lr": 9.355292878377798e-05} +{"train_loss": 0.018786504864692688, "grad_norm": 0.08556890487670898, "global_step": 8662, "epoch": 81, "lr": 9.35514729668787e-05} +{"train_loss": 0.018119189888238907, "grad_norm": 0.055439677089452744, "global_step": 8663, "epoch": 81, "lr": 9.355001699695846e-05} +{"train_loss": 0.01765715703368187, "grad_norm": 0.0859474241733551, "global_step": 8664, "epoch": 81, "lr": 9.35485608740224e-05} +{"train_loss": 0.0147128626704216, "grad_norm": 0.055512066930532455, "global_step": 8665, "epoch": 81, "lr": 9.354710459807565e-05} +{"train_loss": 0.017242299392819405, "grad_norm": 0.07095874845981598, "global_step": 8666, "epoch": 81, "lr": 9.354564816912331e-05} +{"train_loss": 0.019035544246435165, "grad_norm": 0.043036021292209625, "global_step": 8667, "epoch": 81, "lr": 9.354419158717051e-05} +{"train_loss": 0.019539203494787216, "grad_norm": 0.11280302703380585, "global_step": 8668, "epoch": 81, "lr": 9.354273485222233e-05} +{"train_loss": 0.018163379281759262, "grad_norm": 0.10782419145107269, "global_step": 8669, "epoch": 81, "lr": 9.354127796428393e-05} +{"train_loss": 0.019054165109992027, "grad_norm": 0.05981568247079849, "global_step": 8670, "epoch": 81, "lr": 9.353982092336042e-05} +{"train_loss": 0.018342096358537674, "grad_norm": 0.10097387433052063, "global_step": 8671, "epoch": 81, "lr": 9.353836372945692e-05} +{"train_loss": 0.019901106134057045, "grad_norm": 0.10975702106952667, "global_step": 8672, "epoch": 81, "lr": 9.353690638257854e-05} +{"train_loss": 0.020037930458784103, "grad_norm": 0.08199914544820786, "global_step": 8673, "epoch": 81, "lr": 9.35354488827304e-05} +{"train_loss": 0.018677886575460434, "grad_norm": 0.07723304629325867, "global_step": 8674, "epoch": 81, "lr": 9.353399122991762e-05} +{"train_loss": 0.018918734043836594, "grad_norm": 0.11829233169555664, "global_step": 8675, "epoch": 81, "lr": 9.353253342414535e-05} +{"train_loss": 0.016273757442831993, "grad_norm": 0.047764647752046585, "global_step": 8676, "epoch": 81, "lr": 9.353107546541866e-05} +{"train_loss": 0.018077179789543152, "grad_norm": 0.123142771422863, "global_step": 8677, "epoch": 81, "lr": 9.352961735374272e-05} +{"train_loss": 0.01595265418291092, "grad_norm": 0.09560734778642654, "global_step": 8678, "epoch": 81, "lr": 9.352815908912263e-05} +{"train_loss": 0.01694348081946373, "grad_norm": 0.06538266688585281, "global_step": 8679, "epoch": 81, "lr": 9.35267006715635e-05} +{"train_loss": 0.01942049339413643, "grad_norm": 0.07674584537744522, "global_step": 8680, "epoch": 81, "lr": 9.35252421010705e-05} +{"train_loss": 0.018534433096647263, "grad_norm": 0.05737835913896561, "global_step": 8681, "epoch": 81, "lr": 9.352378337764873e-05} +{"train_loss": 0.01919749192893505, "grad_norm": 0.06576621532440186, "global_step": 8682, "epoch": 81, "lr": 9.352232450130329e-05} +{"train_loss": 0.017713099718093872, "grad_norm": 0.10698892176151276, "global_step": 8683, "epoch": 81, "lr": 9.352086547203936e-05} +{"train_loss": 0.01960722915828228, "grad_norm": 0.06930474191904068, "global_step": 8684, "epoch": 81, "lr": 9.3519406289862e-05} +{"train_loss": 0.019841685891151428, "grad_norm": 0.14294016361236572, "global_step": 8685, "epoch": 81, "lr": 9.351794695477639e-05} +{"train_loss": 0.018823791295289993, "grad_norm": 0.09307777881622314, "global_step": 8686, "epoch": 81, "lr": 9.351648746678766e-05} +{"train_loss": 0.021425681188702583, "grad_norm": 0.16351750493049622, "global_step": 8687, "epoch": 81, "lr": 9.351502782590089e-05} +{"train_loss": 0.01557814422994852, "grad_norm": 0.06341489404439926, "global_step": 8688, "epoch": 81, "lr": 9.351356803212123e-05} +{"train_loss": 0.01438859198242426, "grad_norm": 0.09774632751941681, "global_step": 8689, "epoch": 81, "lr": 9.351210808545383e-05} +{"train_loss": 0.019822634756565094, "grad_norm": 0.044725291430950165, "global_step": 8690, "epoch": 81, "lr": 9.35106479859038e-05} +{"train_loss": 0.01808365287680952, "grad_norm": 0.07050932198762894, "global_step": 8691, "epoch": 81, "lr": 9.35091877334763e-05, "val_loss": 0.01943952962756157} +{"train_loss": 0.016431070864200592, "grad_norm": 0.081156425178051, "global_step": 8692, "epoch": 82, "lr": 9.35077273281764e-05} +{"train_loss": 0.018525905907154083, "grad_norm": 0.09641759842634201, "global_step": 8693, "epoch": 82, "lr": 9.350626677000929e-05} +{"train_loss": 0.016619350761175156, "grad_norm": 0.04745585471391678, "global_step": 8694, "epoch": 82, "lr": 9.350480605898007e-05} +{"train_loss": 0.018227703869342804, "grad_norm": 0.07403639703989029, "global_step": 8695, "epoch": 82, "lr": 9.350334519509388e-05} +{"train_loss": 0.017600474879145622, "grad_norm": 0.0425676591694355, "global_step": 8696, "epoch": 82, "lr": 9.350188417835585e-05} +{"train_loss": 0.01640966534614563, "grad_norm": 0.04973907023668289, "global_step": 8697, "epoch": 82, "lr": 9.350042300877113e-05} +{"train_loss": 0.016598382964730263, "grad_norm": 0.040306515991687775, "global_step": 8698, "epoch": 82, "lr": 9.349896168634485e-05} +{"train_loss": 0.017930272966623306, "grad_norm": 0.0688471794128418, "global_step": 8699, "epoch": 82, "lr": 9.349750021108211e-05} +{"train_loss": 0.01647249422967434, "grad_norm": 0.05469398945569992, "global_step": 8700, "epoch": 82, "lr": 9.349603858298808e-05} +{"train_loss": 0.017213135957717896, "grad_norm": 0.10709095746278763, "global_step": 8701, "epoch": 82, "lr": 9.34945768020679e-05} +{"train_loss": 0.01802721992135048, "grad_norm": 0.0858844444155693, "global_step": 8702, "epoch": 82, "lr": 9.349311486832668e-05} +{"train_loss": 0.01697973534464836, "grad_norm": 0.07686656713485718, "global_step": 8703, "epoch": 82, "lr": 9.349165278176957e-05} +{"train_loss": 0.01438838429749012, "grad_norm": 0.04369799792766571, "global_step": 8704, "epoch": 82, "lr": 9.34901905424017e-05} +{"train_loss": 0.016882237046957016, "grad_norm": 0.061106935143470764, "global_step": 8705, "epoch": 82, "lr": 9.348872815022822e-05} +{"train_loss": 0.01688890904188156, "grad_norm": 0.0906844288110733, "global_step": 8706, "epoch": 82, "lr": 9.348726560525427e-05} +{"train_loss": 0.0181525107473135, "grad_norm": 0.05006048083305359, "global_step": 8707, "epoch": 82, "lr": 9.348580290748497e-05} +{"train_loss": 0.014149399474263191, "grad_norm": 0.058174166828393936, "global_step": 8708, "epoch": 82, "lr": 9.348434005692546e-05} +{"train_loss": 0.01763227954506874, "grad_norm": 0.04663773253560066, "global_step": 8709, "epoch": 82, "lr": 9.348287705358091e-05} +{"train_loss": 0.020622309297323227, "grad_norm": 0.07148735225200653, "global_step": 8710, "epoch": 82, "lr": 9.348141389745644e-05} +{"train_loss": 0.018635254353284836, "grad_norm": 0.14063319563865662, "global_step": 8711, "epoch": 82, "lr": 9.347995058855719e-05} +{"train_loss": 0.017626682296395302, "grad_norm": 0.09624557942152023, "global_step": 8712, "epoch": 82, "lr": 9.34784871268883e-05} +{"train_loss": 0.018775027245283127, "grad_norm": 0.08251836150884628, "global_step": 8713, "epoch": 82, "lr": 9.34770235124549e-05} +{"train_loss": 0.020267818123102188, "grad_norm": 0.06886907666921616, "global_step": 8714, "epoch": 82, "lr": 9.347555974526217e-05} +{"train_loss": 0.0165458545088768, "grad_norm": 0.049001045525074005, "global_step": 8715, "epoch": 82, "lr": 9.34740958253152e-05} +{"train_loss": 0.019355125725269318, "grad_norm": 0.052025023847818375, "global_step": 8716, "epoch": 82, "lr": 9.347263175261919e-05} +{"train_loss": 0.018637316301465034, "grad_norm": 0.08609148859977722, "global_step": 8717, "epoch": 82, "lr": 9.347116752717925e-05} +{"train_loss": 0.018560895696282387, "grad_norm": 0.05810889974236488, "global_step": 8718, "epoch": 82, "lr": 9.346970314900053e-05} +{"train_loss": 0.016287554055452347, "grad_norm": 0.07808374613523483, "global_step": 8719, "epoch": 82, "lr": 9.346823861808817e-05} +{"train_loss": 0.016450421884655952, "grad_norm": 0.05151980370283127, "global_step": 8720, "epoch": 82, "lr": 9.346677393444733e-05} +{"train_loss": 0.01730409264564514, "grad_norm": 0.12440904229879379, "global_step": 8721, "epoch": 82, "lr": 9.346530909808318e-05} +{"train_loss": 0.01840120553970337, "grad_norm": 0.05236855894327164, "global_step": 8722, "epoch": 82, "lr": 9.346384410900079e-05} +{"train_loss": 0.01983371376991272, "grad_norm": 0.07142625004053116, "global_step": 8723, "epoch": 82, "lr": 9.346237896720538e-05} +{"train_loss": 0.018317030742764473, "grad_norm": 0.05347564443945885, "global_step": 8724, "epoch": 82, "lr": 9.346091367270206e-05} +{"train_loss": 0.018396764993667603, "grad_norm": 0.09661365300416946, "global_step": 8725, "epoch": 82, "lr": 9.3459448225496e-05} +{"train_loss": 0.018289968371391296, "grad_norm": 0.11656196415424347, "global_step": 8726, "epoch": 82, "lr": 9.345798262559233e-05} +{"train_loss": 0.01679603010416031, "grad_norm": 0.05385001376271248, "global_step": 8727, "epoch": 82, "lr": 9.345651687299621e-05} +{"train_loss": 0.01810237392783165, "grad_norm": 0.0776805579662323, "global_step": 8728, "epoch": 82, "lr": 9.34550509677128e-05} +{"train_loss": 0.02186943218111992, "grad_norm": 0.049608975648880005, "global_step": 8729, "epoch": 82, "lr": 9.345358490974722e-05} +{"train_loss": 0.01794067770242691, "grad_norm": 0.07126324623823166, "global_step": 8730, "epoch": 82, "lr": 9.345211869910466e-05} +{"train_loss": 0.01734904944896698, "grad_norm": 0.050894834101200104, "global_step": 8731, "epoch": 82, "lr": 9.345065233579024e-05} +{"train_loss": 0.01641746237874031, "grad_norm": 0.09395851194858551, "global_step": 8732, "epoch": 82, "lr": 9.344918581980913e-05} +{"train_loss": 0.01891544833779335, "grad_norm": 0.06354348361492157, "global_step": 8733, "epoch": 82, "lr": 9.344771915116646e-05} +{"train_loss": 0.01865975931286812, "grad_norm": 0.05824980139732361, "global_step": 8734, "epoch": 82, "lr": 9.344625232986741e-05} +{"train_loss": 0.019044073298573494, "grad_norm": 0.09595420211553574, "global_step": 8735, "epoch": 82, "lr": 9.344478535591713e-05} +{"train_loss": 0.019192185252904892, "grad_norm": 0.12451454997062683, "global_step": 8736, "epoch": 82, "lr": 9.344331822932076e-05} +{"train_loss": 0.021083246916532516, "grad_norm": 0.15167774260044098, "global_step": 8737, "epoch": 82, "lr": 9.344185095008347e-05} +{"train_loss": 0.019556939601898193, "grad_norm": 0.06162630394101143, "global_step": 8738, "epoch": 82, "lr": 9.344038351821041e-05} +{"train_loss": 0.019211426377296448, "grad_norm": 0.0549306757748127, "global_step": 8739, "epoch": 82, "lr": 9.343891593370672e-05} +{"train_loss": 0.01859705150127411, "grad_norm": 0.08968017250299454, "global_step": 8740, "epoch": 82, "lr": 9.343744819657758e-05} +{"train_loss": 0.017814133316278458, "grad_norm": 0.059300243854522705, "global_step": 8741, "epoch": 82, "lr": 9.343598030682815e-05} +{"train_loss": 0.02058841846883297, "grad_norm": 0.09207624197006226, "global_step": 8742, "epoch": 82, "lr": 9.343451226446354e-05} +{"train_loss": 0.01781412959098816, "grad_norm": 0.053357917815446854, "global_step": 8743, "epoch": 82, "lr": 9.343304406948897e-05} +{"train_loss": 0.017283014953136444, "grad_norm": 0.0564785934984684, "global_step": 8744, "epoch": 82, "lr": 9.343157572190957e-05} +{"train_loss": 0.017514444887638092, "grad_norm": 0.04456273466348648, "global_step": 8745, "epoch": 82, "lr": 9.34301072217305e-05} +{"train_loss": 0.015398291870951653, "grad_norm": 0.08214101195335388, "global_step": 8746, "epoch": 82, "lr": 9.342863856895692e-05} +{"train_loss": 0.017070233821868896, "grad_norm": 0.045518908649683, "global_step": 8747, "epoch": 82, "lr": 9.342716976359398e-05} +{"train_loss": 0.01813463494181633, "grad_norm": 0.07430003583431244, "global_step": 8748, "epoch": 82, "lr": 9.342570080564685e-05} +{"train_loss": 0.016420044004917145, "grad_norm": 0.10004771500825882, "global_step": 8749, "epoch": 82, "lr": 9.342423169512071e-05} +{"train_loss": 0.01930084079504013, "grad_norm": 0.051864881068468094, "global_step": 8750, "epoch": 82, "lr": 9.342276243202069e-05} +{"train_loss": 0.02030620351433754, "grad_norm": 0.09092108905315399, "global_step": 8751, "epoch": 82, "lr": 9.342129301635198e-05} +{"train_loss": 0.017232248559594154, "grad_norm": 0.08404341340065002, "global_step": 8752, "epoch": 82, "lr": 9.341982344811973e-05} +{"train_loss": 0.015512129291892052, "grad_norm": 0.0826202929019928, "global_step": 8753, "epoch": 82, "lr": 9.341835372732909e-05} +{"train_loss": 0.02103554643690586, "grad_norm": 0.07827309519052505, "global_step": 8754, "epoch": 82, "lr": 9.341688385398524e-05} +{"train_loss": 0.018578868359327316, "grad_norm": 0.0563037171959877, "global_step": 8755, "epoch": 82, "lr": 9.341541382809335e-05} +{"train_loss": 0.019123125821352005, "grad_norm": 0.0618215836584568, "global_step": 8756, "epoch": 82, "lr": 9.341394364965856e-05} +{"train_loss": 0.021021811291575432, "grad_norm": 0.04005257785320282, "global_step": 8757, "epoch": 82, "lr": 9.341247331868608e-05} +{"train_loss": 0.018965715542435646, "grad_norm": 0.08335676044225693, "global_step": 8758, "epoch": 82, "lr": 9.341100283518103e-05} +{"train_loss": 0.020278219133615494, "grad_norm": 0.07770195603370667, "global_step": 8759, "epoch": 82, "lr": 9.340953219914859e-05} +{"train_loss": 0.01915721595287323, "grad_norm": 0.1643275022506714, "global_step": 8760, "epoch": 82, "lr": 9.340806141059393e-05} +{"train_loss": 0.02042418345808983, "grad_norm": 0.12422902137041092, "global_step": 8761, "epoch": 82, "lr": 9.340659046952224e-05} +{"train_loss": 0.01816748082637787, "grad_norm": 0.10155386477708817, "global_step": 8762, "epoch": 82, "lr": 9.340511937593865e-05} +{"train_loss": 0.016236618161201477, "grad_norm": 0.09365449100732803, "global_step": 8763, "epoch": 82, "lr": 9.340364812984836e-05} +{"train_loss": 0.017226528376340866, "grad_norm": 0.05113760381937027, "global_step": 8764, "epoch": 82, "lr": 9.340217673125653e-05} +{"train_loss": 0.017010724171996117, "grad_norm": 0.05812017247080803, "global_step": 8765, "epoch": 82, "lr": 9.340070518016831e-05} +{"train_loss": 0.01760714314877987, "grad_norm": 0.06484127044677734, "global_step": 8766, "epoch": 82, "lr": 9.33992334765889e-05} +{"train_loss": 0.019560977816581726, "grad_norm": 0.08381511270999908, "global_step": 8767, "epoch": 82, "lr": 9.339776162052344e-05} +{"train_loss": 0.018177885562181473, "grad_norm": 0.06584542989730835, "global_step": 8768, "epoch": 82, "lr": 9.339628961197713e-05} +{"train_loss": 0.02060551568865776, "grad_norm": 0.05721108987927437, "global_step": 8769, "epoch": 82, "lr": 9.339481745095513e-05} +{"train_loss": 0.017045151442289352, "grad_norm": 0.10970029979944229, "global_step": 8770, "epoch": 82, "lr": 9.339334513746261e-05} +{"train_loss": 0.016697879880666733, "grad_norm": 0.06393585354089737, "global_step": 8771, "epoch": 82, "lr": 9.339187267150475e-05} +{"train_loss": 0.015936557203531265, "grad_norm": 0.14665836095809937, "global_step": 8772, "epoch": 82, "lr": 9.339040005308673e-05} +{"train_loss": 0.018174082040786743, "grad_norm": 0.06700091809034348, "global_step": 8773, "epoch": 82, "lr": 9.33889272822137e-05} +{"train_loss": 0.015565318986773491, "grad_norm": 0.08965244889259338, "global_step": 8774, "epoch": 82, "lr": 9.338745435889085e-05} +{"train_loss": 0.01788208819925785, "grad_norm": 0.11994659155607224, "global_step": 8775, "epoch": 82, "lr": 9.338598128312336e-05} +{"train_loss": 0.017506279051303864, "grad_norm": 0.06732018291950226, "global_step": 8776, "epoch": 82, "lr": 9.33845080549164e-05} +{"train_loss": 0.017637286335229874, "grad_norm": 0.0710001066327095, "global_step": 8777, "epoch": 82, "lr": 9.338303467427514e-05} +{"train_loss": 0.020280934870243073, "grad_norm": 0.04311428591609001, "global_step": 8778, "epoch": 82, "lr": 9.338156114120477e-05} +{"train_loss": 0.01716068759560585, "grad_norm": 0.09206822514533997, "global_step": 8779, "epoch": 82, "lr": 9.338008745571046e-05} +{"train_loss": 0.017626427114009857, "grad_norm": 0.12701070308685303, "global_step": 8780, "epoch": 82, "lr": 9.337861361779739e-05} +{"train_loss": 0.01483979169279337, "grad_norm": 0.04321994259953499, "global_step": 8781, "epoch": 82, "lr": 9.337713962747072e-05} +{"train_loss": 0.016991419717669487, "grad_norm": 0.06488700211048126, "global_step": 8782, "epoch": 82, "lr": 9.337566548473567e-05} +{"train_loss": 0.018040861934423447, "grad_norm": 0.04454093426465988, "global_step": 8783, "epoch": 82, "lr": 9.337419118959737e-05} +{"train_loss": 0.014539173804223537, "grad_norm": 0.04676708206534386, "global_step": 8784, "epoch": 82, "lr": 9.337271674206103e-05} +{"train_loss": 0.018641872331500053, "grad_norm": 0.116684190928936, "global_step": 8785, "epoch": 82, "lr": 9.337124214213184e-05} +{"train_loss": 0.018964659422636032, "grad_norm": 0.06107670068740845, "global_step": 8786, "epoch": 82, "lr": 9.336976738981495e-05} +{"train_loss": 0.019924797117710114, "grad_norm": 0.07286136597394943, "global_step": 8787, "epoch": 82, "lr": 9.336829248511559e-05} +{"train_loss": 0.017542585730552673, "grad_norm": 0.07645268738269806, "global_step": 8788, "epoch": 82, "lr": 9.336681742803887e-05} +{"train_loss": 0.018382659181952477, "grad_norm": 0.05254892632365227, "global_step": 8789, "epoch": 82, "lr": 9.336534221859004e-05} +{"train_loss": 0.01674843020737171, "grad_norm": 0.05932234227657318, "global_step": 8790, "epoch": 82, "lr": 9.336386685677425e-05} +{"train_loss": 0.02006867341697216, "grad_norm": 0.07443518191576004, "global_step": 8791, "epoch": 82, "lr": 9.336239134259668e-05} +{"train_loss": 0.018954619765281677, "grad_norm": 0.05547531694173813, "global_step": 8792, "epoch": 82, "lr": 9.336091567606254e-05} +{"train_loss": 0.018284378573298454, "grad_norm": 0.08309201896190643, "global_step": 8793, "epoch": 82, "lr": 9.335943985717699e-05} +{"train_loss": 0.018120236694812775, "grad_norm": 0.05045878142118454, "global_step": 8794, "epoch": 82, "lr": 9.335796388594522e-05} +{"train_loss": 0.018429171293973923, "grad_norm": 0.053555186837911606, "global_step": 8795, "epoch": 82, "lr": 9.335648776237242e-05} +{"train_loss": 0.020202841609716415, "grad_norm": 0.07978297770023346, "global_step": 8796, "epoch": 82, "lr": 9.33550114864638e-05} +{"train_loss": 0.01801693015995453, "grad_norm": 0.10744772106409073, "global_step": 8797, "epoch": 82, "lr": 9.33535350582245e-05, "val_loss": 0.01921941712498665} +{"train_loss": 0.018857114017009735, "grad_norm": 0.05783172324299812, "global_step": 8798, "epoch": 83, "lr": 9.335205847765975e-05} +{"train_loss": 0.018112950026988983, "grad_norm": 0.12235428392887115, "global_step": 8799, "epoch": 83, "lr": 9.335058174477471e-05} +{"train_loss": 0.017572324723005295, "grad_norm": 0.10831370949745178, "global_step": 8800, "epoch": 83, "lr": 9.334910485957457e-05} +{"train_loss": 0.0165577195584774, "grad_norm": 0.0758386105298996, "global_step": 8801, "epoch": 83, "lr": 9.334762782206454e-05} +{"train_loss": 0.01965527981519699, "grad_norm": 0.1481063961982727, "global_step": 8802, "epoch": 83, "lr": 9.334615063224981e-05} +{"train_loss": 0.020030489191412926, "grad_norm": 0.07188571244478226, "global_step": 8803, "epoch": 83, "lr": 9.334467329013554e-05} +{"train_loss": 0.017626479268074036, "grad_norm": 0.0846778079867363, "global_step": 8804, "epoch": 83, "lr": 9.334319579572694e-05} +{"train_loss": 0.016642889007925987, "grad_norm": 0.04549630358815193, "global_step": 8805, "epoch": 83, "lr": 9.334171814902919e-05} +{"train_loss": 0.018352003768086433, "grad_norm": 0.07850181311368942, "global_step": 8806, "epoch": 83, "lr": 9.33402403500475e-05} +{"train_loss": 0.017186377197504044, "grad_norm": 0.04340336471796036, "global_step": 8807, "epoch": 83, "lr": 9.333876239878705e-05} +{"train_loss": 0.017734238877892494, "grad_norm": 0.048801399767398834, "global_step": 8808, "epoch": 83, "lr": 9.333728429525305e-05} +{"train_loss": 0.02130420319736004, "grad_norm": 0.04796730726957321, "global_step": 8809, "epoch": 83, "lr": 9.333580603945067e-05} +{"train_loss": 0.019356772303581238, "grad_norm": 0.05274689942598343, "global_step": 8810, "epoch": 83, "lr": 9.333432763138513e-05} +{"train_loss": 0.017392123118042946, "grad_norm": 0.06398051232099533, "global_step": 8811, "epoch": 83, "lr": 9.333284907106158e-05} +{"train_loss": 0.0177018940448761, "grad_norm": 0.1278047263622284, "global_step": 8812, "epoch": 83, "lr": 9.333137035848524e-05} +{"train_loss": 0.018374625593423843, "grad_norm": 0.0712035745382309, "global_step": 8813, "epoch": 83, "lr": 9.332989149366134e-05} +{"train_loss": 0.01579897664487362, "grad_norm": 0.06606409698724747, "global_step": 8814, "epoch": 83, "lr": 9.332841247659503e-05} +{"train_loss": 0.020375041291117668, "grad_norm": 0.11374730616807938, "global_step": 8815, "epoch": 83, "lr": 9.332693330729152e-05} +{"train_loss": 0.021297693252563477, "grad_norm": 0.09889603406190872, "global_step": 8816, "epoch": 83, "lr": 9.3325453985756e-05} +{"train_loss": 0.01506540272384882, "grad_norm": 0.14600719511508942, "global_step": 8817, "epoch": 83, "lr": 9.33239745119937e-05} +{"train_loss": 0.019285308197140694, "grad_norm": 0.08801740407943726, "global_step": 8818, "epoch": 83, "lr": 9.332249488600977e-05} +{"train_loss": 0.0193089097738266, "grad_norm": 0.09333273768424988, "global_step": 8819, "epoch": 83, "lr": 9.332101510780945e-05} +{"train_loss": 0.01813673786818981, "grad_norm": 0.0654422715306282, "global_step": 8820, "epoch": 83, "lr": 9.331953517739793e-05} +{"train_loss": 0.016646120697259903, "grad_norm": 0.11402281373739243, "global_step": 8821, "epoch": 83, "lr": 9.331805509478037e-05} +{"train_loss": 0.01807311549782753, "grad_norm": 0.09043353796005249, "global_step": 8822, "epoch": 83, "lr": 9.331657485996204e-05} +{"train_loss": 0.017594441771507263, "grad_norm": 0.07224971055984497, "global_step": 8823, "epoch": 83, "lr": 9.331509447294808e-05} +{"train_loss": 0.01639389805495739, "grad_norm": 0.09656578302383423, "global_step": 8824, "epoch": 83, "lr": 9.331361393374372e-05} +{"train_loss": 0.014798134565353394, "grad_norm": 0.06816896051168442, "global_step": 8825, "epoch": 83, "lr": 9.331213324235417e-05} +{"train_loss": 0.01791904866695404, "grad_norm": 0.0815758928656578, "global_step": 8826, "epoch": 83, "lr": 9.33106523987846e-05} +{"train_loss": 0.015923017635941505, "grad_norm": 0.044146474450826645, "global_step": 8827, "epoch": 83, "lr": 9.330917140304024e-05} +{"train_loss": 0.0201827734708786, "grad_norm": 0.0700719952583313, "global_step": 8828, "epoch": 83, "lr": 9.330769025512628e-05} +{"train_loss": 0.018994130194187164, "grad_norm": 0.0562206506729126, "global_step": 8829, "epoch": 83, "lr": 9.330620895504794e-05} +{"train_loss": 0.016654174774885178, "grad_norm": 0.04940369352698326, "global_step": 8830, "epoch": 83, "lr": 9.330472750281043e-05} +{"train_loss": 0.020119106397032738, "grad_norm": 0.04729587212204933, "global_step": 8831, "epoch": 83, "lr": 9.330324589841892e-05} +{"train_loss": 0.017611000686883926, "grad_norm": 0.10684310644865036, "global_step": 8832, "epoch": 83, "lr": 9.330176414187863e-05} +{"train_loss": 0.017392843961715698, "grad_norm": 0.07354363054037094, "global_step": 8833, "epoch": 83, "lr": 9.330028223319479e-05} +{"train_loss": 0.017042074352502823, "grad_norm": 0.07417827099561691, "global_step": 8834, "epoch": 83, "lr": 9.329880017237258e-05} +{"train_loss": 0.016885358840227127, "grad_norm": 0.1214616671204567, "global_step": 8835, "epoch": 83, "lr": 9.329731795941722e-05} +{"train_loss": 0.01781165972352028, "grad_norm": 0.04637501388788223, "global_step": 8836, "epoch": 83, "lr": 9.32958355943339e-05} +{"train_loss": 0.017402078956365585, "grad_norm": 0.07175257802009583, "global_step": 8837, "epoch": 83, "lr": 9.329435307712786e-05} +{"train_loss": 0.02102508768439293, "grad_norm": 0.07513727992773056, "global_step": 8838, "epoch": 83, "lr": 9.329287040780429e-05} +{"train_loss": 0.018210284411907196, "grad_norm": 0.10651930421590805, "global_step": 8839, "epoch": 83, "lr": 9.32913875863684e-05} +{"train_loss": 0.02217291295528412, "grad_norm": 0.07880178093910217, "global_step": 8840, "epoch": 83, "lr": 9.32899046128254e-05} +{"train_loss": 0.01795770227909088, "grad_norm": 0.16634593904018402, "global_step": 8841, "epoch": 83, "lr": 9.328842148718049e-05} +{"train_loss": 0.020619800314307213, "grad_norm": 0.10170434415340424, "global_step": 8842, "epoch": 83, "lr": 9.32869382094389e-05} +{"train_loss": 0.01829349994659424, "grad_norm": 0.04617447406053543, "global_step": 8843, "epoch": 83, "lr": 9.328545477960585e-05} +{"train_loss": 0.016492754220962524, "grad_norm": 0.04297728091478348, "global_step": 8844, "epoch": 83, "lr": 9.328397119768652e-05} +{"train_loss": 0.01815558224916458, "grad_norm": 0.10024736076593399, "global_step": 8845, "epoch": 83, "lr": 9.328248746368613e-05} +{"train_loss": 0.016337957233190536, "grad_norm": 0.15335822105407715, "global_step": 8846, "epoch": 83, "lr": 9.328100357760992e-05} +{"train_loss": 0.017076579853892326, "grad_norm": 0.10194167494773865, "global_step": 8847, "epoch": 83, "lr": 9.327951953946307e-05} +{"train_loss": 0.019524497911334038, "grad_norm": 0.08664140105247498, "global_step": 8848, "epoch": 83, "lr": 9.327803534925083e-05} +{"train_loss": 0.017163939774036407, "grad_norm": 0.07120765000581741, "global_step": 8849, "epoch": 83, "lr": 9.327655100697839e-05} +{"train_loss": 0.015918880701065063, "grad_norm": 0.0537237785756588, "global_step": 8850, "epoch": 83, "lr": 9.327506651265095e-05} +{"train_loss": 0.016259483993053436, "grad_norm": 0.09766758233308792, "global_step": 8851, "epoch": 83, "lr": 9.327358186627376e-05} +{"train_loss": 0.016952473670244217, "grad_norm": 0.1005675196647644, "global_step": 8852, "epoch": 83, "lr": 9.327209706785203e-05} +{"train_loss": 0.01848345436155796, "grad_norm": 0.10754555463790894, "global_step": 8853, "epoch": 83, "lr": 9.327061211739096e-05} +{"train_loss": 0.015765173360705376, "grad_norm": 0.05927322432398796, "global_step": 8854, "epoch": 83, "lr": 9.326912701489578e-05} +{"train_loss": 0.01895136572420597, "grad_norm": 0.07012516260147095, "global_step": 8855, "epoch": 83, "lr": 9.32676417603717e-05} +{"train_loss": 0.016934573650360107, "grad_norm": 0.052418533712625504, "global_step": 8856, "epoch": 83, "lr": 9.326615635382396e-05} +{"train_loss": 0.01630459725856781, "grad_norm": 0.11105263978242874, "global_step": 8857, "epoch": 83, "lr": 9.326467079525774e-05} +{"train_loss": 0.021917160600423813, "grad_norm": 0.0806654617190361, "global_step": 8858, "epoch": 83, "lr": 9.32631850846783e-05} +{"train_loss": 0.01681620255112648, "grad_norm": 0.0654427707195282, "global_step": 8859, "epoch": 83, "lr": 9.326169922209083e-05} +{"train_loss": 0.01618301309645176, "grad_norm": 0.04988541454076767, "global_step": 8860, "epoch": 83, "lr": 9.326021320750056e-05} +{"train_loss": 0.01584663614630699, "grad_norm": 0.05646902322769165, "global_step": 8861, "epoch": 83, "lr": 9.325872704091272e-05} +{"train_loss": 0.02014348655939102, "grad_norm": 0.049390748143196106, "global_step": 8862, "epoch": 83, "lr": 9.325724072233253e-05} +{"train_loss": 0.019711390137672424, "grad_norm": 0.06514013558626175, "global_step": 8863, "epoch": 83, "lr": 9.32557542517652e-05} +{"train_loss": 0.01790832355618477, "grad_norm": 0.13262678682804108, "global_step": 8864, "epoch": 83, "lr": 9.325426762921597e-05} +{"train_loss": 0.018446024507284164, "grad_norm": 0.061814840883016586, "global_step": 8865, "epoch": 83, "lr": 9.325278085469005e-05} +{"train_loss": 0.017239663749933243, "grad_norm": 0.06439635157585144, "global_step": 8866, "epoch": 83, "lr": 9.325129392819268e-05} +{"train_loss": 0.017530083656311035, "grad_norm": 0.04776311665773392, "global_step": 8867, "epoch": 83, "lr": 9.324980684972907e-05} +{"train_loss": 0.0188845656812191, "grad_norm": 0.08697252720594406, "global_step": 8868, "epoch": 83, "lr": 9.324831961930442e-05} +{"train_loss": 0.02009391039609909, "grad_norm": 0.07189837098121643, "global_step": 8869, "epoch": 83, "lr": 9.324683223692401e-05} +{"train_loss": 0.018736787140369415, "grad_norm": 0.06857877224683762, "global_step": 8870, "epoch": 83, "lr": 9.324534470259303e-05} +{"train_loss": 0.016319438815116882, "grad_norm": 0.08249776065349579, "global_step": 8871, "epoch": 83, "lr": 9.324385701631673e-05} +{"train_loss": 0.017477257177233696, "grad_norm": 0.07360991090536118, "global_step": 8872, "epoch": 83, "lr": 9.32423691781003e-05} +{"train_loss": 0.015224490314722061, "grad_norm": 0.06717570871114731, "global_step": 8873, "epoch": 83, "lr": 9.3240881187949e-05} +{"train_loss": 0.019888589158654213, "grad_norm": 0.08010401576757431, "global_step": 8874, "epoch": 83, "lr": 9.323939304586805e-05} +{"train_loss": 0.015525389462709427, "grad_norm": 0.058272309601306915, "global_step": 8875, "epoch": 83, "lr": 9.323790475186266e-05} +{"train_loss": 0.01692701131105423, "grad_norm": 0.1622350513935089, "global_step": 8876, "epoch": 83, "lr": 9.323641630593811e-05} +{"train_loss": 0.01762206293642521, "grad_norm": 0.09107574075460434, "global_step": 8877, "epoch": 83, "lr": 9.323492770809957e-05} +{"train_loss": 0.01543061900883913, "grad_norm": 0.03702997416257858, "global_step": 8878, "epoch": 83, "lr": 9.32334389583523e-05} +{"train_loss": 0.01762979105114937, "grad_norm": 0.05522080138325691, "global_step": 8879, "epoch": 83, "lr": 9.323195005670154e-05} +{"train_loss": 0.019423747435212135, "grad_norm": 0.051789313554763794, "global_step": 8880, "epoch": 83, "lr": 9.323046100315251e-05} +{"train_loss": 0.019997742027044296, "grad_norm": 0.08180659264326096, "global_step": 8881, "epoch": 83, "lr": 9.322897179771041e-05} +{"train_loss": 0.018807098269462585, "grad_norm": 0.0676206648349762, "global_step": 8882, "epoch": 83, "lr": 9.322748244038053e-05} +{"train_loss": 0.017778949812054634, "grad_norm": 0.04175450652837753, "global_step": 8883, "epoch": 83, "lr": 9.322599293116808e-05} +{"train_loss": 0.02136164903640747, "grad_norm": 0.06628985702991486, "global_step": 8884, "epoch": 83, "lr": 9.322450327007828e-05} +{"train_loss": 0.02023240365087986, "grad_norm": 0.06644098460674286, "global_step": 8885, "epoch": 83, "lr": 9.322301345711637e-05} +{"train_loss": 0.0171753391623497, "grad_norm": 0.058354251086711884, "global_step": 8886, "epoch": 83, "lr": 9.322152349228758e-05} +{"train_loss": 0.017367448657751083, "grad_norm": 0.059647608548402786, "global_step": 8887, "epoch": 83, "lr": 9.322003337559717e-05} +{"train_loss": 0.017831183969974518, "grad_norm": 0.11673974990844727, "global_step": 8888, "epoch": 83, "lr": 9.321854310705035e-05} +{"train_loss": 0.016945920884609222, "grad_norm": 0.12634553015232086, "global_step": 8889, "epoch": 83, "lr": 9.321705268665238e-05} +{"train_loss": 0.018395211547613144, "grad_norm": 0.06803562492132187, "global_step": 8890, "epoch": 83, "lr": 9.321556211440847e-05} +{"train_loss": 0.019914986565709114, "grad_norm": 0.07191152125597, "global_step": 8891, "epoch": 83, "lr": 9.321407139032385e-05} +{"train_loss": 0.016569174826145172, "grad_norm": 0.04478922113776207, "global_step": 8892, "epoch": 83, "lr": 9.32125805144038e-05} +{"train_loss": 0.019540484994649887, "grad_norm": 0.052365582436323166, "global_step": 8893, "epoch": 83, "lr": 9.321108948665352e-05} +{"train_loss": 0.015552529133856297, "grad_norm": 0.04500919207930565, "global_step": 8894, "epoch": 83, "lr": 9.320959830707827e-05} +{"train_loss": 0.017862586304545403, "grad_norm": 0.0654449388384819, "global_step": 8895, "epoch": 83, "lr": 9.32081069756833e-05} +{"train_loss": 0.02037706971168518, "grad_norm": 0.045705948024988174, "global_step": 8896, "epoch": 83, "lr": 9.32066154924738e-05} +{"train_loss": 0.018452884629368782, "grad_norm": 0.11170446127653122, "global_step": 8897, "epoch": 83, "lr": 9.320512385745507e-05} +{"train_loss": 0.01902421936392784, "grad_norm": 0.05521496385335922, "global_step": 8898, "epoch": 83, "lr": 9.320363207063232e-05} +{"train_loss": 0.017811372876167297, "grad_norm": 0.04996778815984726, "global_step": 8899, "epoch": 83, "lr": 9.320214013201078e-05} +{"train_loss": 0.01875271275639534, "grad_norm": 0.10550763458013535, "global_step": 8900, "epoch": 83, "lr": 9.320064804159572e-05} +{"train_loss": 0.018557554110884666, "grad_norm": 0.11170326918363571, "global_step": 8901, "epoch": 83, "lr": 9.319915579939238e-05} +{"train_loss": 0.015000146813690662, "grad_norm": 0.06465644389390945, "global_step": 8902, "epoch": 83, "lr": 9.319766340540596e-05} +{"train_loss": 0.01803003938622632, "grad_norm": 0.15474624931812286, "global_step": 8903, "epoch": 83, "lr": 9.319617085964176e-05, "val_loss": 0.01875499077141285} +{"train_loss": 0.018667928874492645, "grad_norm": 0.08500601351261139, "global_step": 8904, "epoch": 84, "lr": 9.3194678162105e-05} +{"train_loss": 0.01830967143177986, "grad_norm": 0.14344415068626404, "global_step": 8905, "epoch": 84, "lr": 9.319318531280094e-05} +{"train_loss": 0.01760084182024002, "grad_norm": 0.04741564393043518, "global_step": 8906, "epoch": 84, "lr": 9.319169231173479e-05} +{"train_loss": 0.020667895674705505, "grad_norm": 0.06990869343280792, "global_step": 8907, "epoch": 84, "lr": 9.319019915891182e-05} +{"train_loss": 0.01915569044649601, "grad_norm": 0.05800311639904976, "global_step": 8908, "epoch": 84, "lr": 9.318870585433728e-05} +{"train_loss": 0.017585383728146553, "grad_norm": 0.040547069162130356, "global_step": 8909, "epoch": 84, "lr": 9.31872123980164e-05} +{"train_loss": 0.017728496342897415, "grad_norm": 0.1297627091407776, "global_step": 8910, "epoch": 84, "lr": 9.318571878995446e-05} +{"train_loss": 0.019361212849617004, "grad_norm": 0.07869809865951538, "global_step": 8911, "epoch": 84, "lr": 9.318422503015665e-05} +{"train_loss": 0.01685146614909172, "grad_norm": 0.08505290746688843, "global_step": 8912, "epoch": 84, "lr": 9.318273111862828e-05} +{"train_loss": 0.019109295681118965, "grad_norm": 0.08857990801334381, "global_step": 8913, "epoch": 84, "lr": 9.318123705537457e-05} +{"train_loss": 0.01635008305311203, "grad_norm": 0.05801418423652649, "global_step": 8914, "epoch": 84, "lr": 9.317974284040076e-05} +{"train_loss": 0.018548280000686646, "grad_norm": 0.147576704621315, "global_step": 8915, "epoch": 84, "lr": 9.317824847371213e-05} +{"train_loss": 0.01719854585826397, "grad_norm": 0.07704256474971771, "global_step": 8916, "epoch": 84, "lr": 9.317675395531389e-05} +{"train_loss": 0.01797361485660076, "grad_norm": 0.07108655571937561, "global_step": 8917, "epoch": 84, "lr": 9.317525928521133e-05} +{"train_loss": 0.017843499779701233, "grad_norm": 0.04075508937239647, "global_step": 8918, "epoch": 84, "lr": 9.317376446340969e-05} +{"train_loss": 0.01791505515575409, "grad_norm": 0.05277344956994057, "global_step": 8919, "epoch": 84, "lr": 9.31722694899142e-05} +{"train_loss": 0.020513560622930527, "grad_norm": 0.09099505841732025, "global_step": 8920, "epoch": 84, "lr": 9.317077436473014e-05} +{"train_loss": 0.01981068216264248, "grad_norm": 0.08932704478502274, "global_step": 8921, "epoch": 84, "lr": 9.316927908786275e-05} +{"train_loss": 0.01894477568566799, "grad_norm": 0.06039394438266754, "global_step": 8922, "epoch": 84, "lr": 9.316778365931729e-05} +{"train_loss": 0.01686977781355381, "grad_norm": 0.058302558958530426, "global_step": 8923, "epoch": 84, "lr": 9.3166288079099e-05} +{"train_loss": 0.021203260868787766, "grad_norm": 0.06968669593334198, "global_step": 8924, "epoch": 84, "lr": 9.316479234721315e-05} +{"train_loss": 0.018089786171913147, "grad_norm": 0.08567609637975693, "global_step": 8925, "epoch": 84, "lr": 9.316329646366501e-05} +{"train_loss": 0.019101494923233986, "grad_norm": 0.056703612208366394, "global_step": 8926, "epoch": 84, "lr": 9.316180042845981e-05} +{"train_loss": 0.017648035660386086, "grad_norm": 0.0710078626871109, "global_step": 8927, "epoch": 84, "lr": 9.31603042416028e-05} +{"train_loss": 0.018426094204187393, "grad_norm": 0.0599275566637516, "global_step": 8928, "epoch": 84, "lr": 9.315880790309926e-05} +{"train_loss": 0.01833864115178585, "grad_norm": 0.053203485906124115, "global_step": 8929, "epoch": 84, "lr": 9.315731141295444e-05} +{"train_loss": 0.01690453663468361, "grad_norm": 0.11418737471103668, "global_step": 8930, "epoch": 84, "lr": 9.315581477117358e-05} +{"train_loss": 0.016957463696599007, "grad_norm": 0.05046870559453964, "global_step": 8931, "epoch": 84, "lr": 9.315431797776198e-05} +{"train_loss": 0.017128489911556244, "grad_norm": 0.06729397177696228, "global_step": 8932, "epoch": 84, "lr": 9.315282103272487e-05} +{"train_loss": 0.01866689696907997, "grad_norm": 0.052183009684085846, "global_step": 8933, "epoch": 84, "lr": 9.315132393606751e-05} +{"train_loss": 0.018030688166618347, "grad_norm": 0.06728216260671616, "global_step": 8934, "epoch": 84, "lr": 9.314982668779515e-05} +{"train_loss": 0.01728714257478714, "grad_norm": 0.04239271581172943, "global_step": 8935, "epoch": 84, "lr": 9.314832928791308e-05} +{"train_loss": 0.017794327810406685, "grad_norm": 0.05772973597049713, "global_step": 8936, "epoch": 84, "lr": 9.314683173642654e-05} +{"train_loss": 0.018693383783102036, "grad_norm": 0.14256379008293152, "global_step": 8937, "epoch": 84, "lr": 9.31453340333408e-05} +{"train_loss": 0.01692012883722782, "grad_norm": 0.14493682980537415, "global_step": 8938, "epoch": 84, "lr": 9.314383617866111e-05} +{"train_loss": 0.016358844935894012, "grad_norm": 0.06829584389925003, "global_step": 8939, "epoch": 84, "lr": 9.314233817239275e-05} +{"train_loss": 0.01610662043094635, "grad_norm": 0.05007988214492798, "global_step": 8940, "epoch": 84, "lr": 9.314084001454098e-05} +{"train_loss": 0.014353265054523945, "grad_norm": 0.054511480033397675, "global_step": 8941, "epoch": 84, "lr": 9.313934170511106e-05} +{"train_loss": 0.018305150792002678, "grad_norm": 0.0909915417432785, "global_step": 8942, "epoch": 84, "lr": 9.313784324410825e-05} +{"train_loss": 0.018501365557312965, "grad_norm": 0.05880984663963318, "global_step": 8943, "epoch": 84, "lr": 9.313634463153782e-05} +{"train_loss": 0.01474351342767477, "grad_norm": 0.11367952078580856, "global_step": 8944, "epoch": 84, "lr": 9.313484586740502e-05} +{"train_loss": 0.015358397737145424, "grad_norm": 0.07611309736967087, "global_step": 8945, "epoch": 84, "lr": 9.313334695171514e-05} +{"train_loss": 0.01800435781478882, "grad_norm": 0.0840212032198906, "global_step": 8946, "epoch": 84, "lr": 9.313184788447345e-05} +{"train_loss": 0.019316505640745163, "grad_norm": 0.14003078639507294, "global_step": 8947, "epoch": 84, "lr": 9.31303486656852e-05} +{"train_loss": 0.020463792607188225, "grad_norm": 0.06791650503873825, "global_step": 8948, "epoch": 84, "lr": 9.312884929535565e-05} +{"train_loss": 0.016489818692207336, "grad_norm": 0.09697970002889633, "global_step": 8949, "epoch": 84, "lr": 9.31273497734901e-05} +{"train_loss": 0.01881520077586174, "grad_norm": 0.12141535431146622, "global_step": 8950, "epoch": 84, "lr": 9.312585010009375e-05} +{"train_loss": 0.01999516412615776, "grad_norm": 0.13123999536037445, "global_step": 8951, "epoch": 84, "lr": 9.312435027517197e-05} +{"train_loss": 0.01618093065917492, "grad_norm": 0.08540765941143036, "global_step": 8952, "epoch": 84, "lr": 9.312285029872995e-05} +{"train_loss": 0.018019042909145355, "grad_norm": 0.05936284735798836, "global_step": 8953, "epoch": 84, "lr": 9.3121350170773e-05} +{"train_loss": 0.0200292207300663, "grad_norm": 0.07137439399957657, "global_step": 8954, "epoch": 84, "lr": 9.311984989130638e-05} +{"train_loss": 0.016552435234189034, "grad_norm": 0.09665558487176895, "global_step": 8955, "epoch": 84, "lr": 9.311834946033536e-05} +{"train_loss": 0.016890088096261024, "grad_norm": 0.07603716105222702, "global_step": 8956, "epoch": 84, "lr": 9.31168488778652e-05} +{"train_loss": 0.019247323274612427, "grad_norm": 0.05258389189839363, "global_step": 8957, "epoch": 84, "lr": 9.311534814390119e-05} +{"train_loss": 0.016752704977989197, "grad_norm": 0.041085582226514816, "global_step": 8958, "epoch": 84, "lr": 9.311384725844861e-05} +{"train_loss": 0.016552720218896866, "grad_norm": 0.0767107903957367, "global_step": 8959, "epoch": 84, "lr": 9.311234622151271e-05} +{"train_loss": 0.018510617315769196, "grad_norm": 0.05121462419629097, "global_step": 8960, "epoch": 84, "lr": 9.311084503309877e-05} +{"train_loss": 0.01570637710392475, "grad_norm": 0.09799447655677795, "global_step": 8961, "epoch": 84, "lr": 9.310934369321207e-05} +{"train_loss": 0.01774497516453266, "grad_norm": 0.06774678081274033, "global_step": 8962, "epoch": 84, "lr": 9.310784220185788e-05} +{"train_loss": 0.019092682749032974, "grad_norm": 0.05377107113599777, "global_step": 8963, "epoch": 84, "lr": 9.310634055904148e-05} +{"train_loss": 0.01666055992245674, "grad_norm": 0.05228777229785919, "global_step": 8964, "epoch": 84, "lr": 9.310483876476817e-05} +{"train_loss": 0.019027046859264374, "grad_norm": 0.048100363463163376, "global_step": 8965, "epoch": 84, "lr": 9.310333681904318e-05} +{"train_loss": 0.01616288721561432, "grad_norm": 0.08958656340837479, "global_step": 8966, "epoch": 84, "lr": 9.310183472187181e-05} +{"train_loss": 0.01705370470881462, "grad_norm": 0.051577892154455185, "global_step": 8967, "epoch": 84, "lr": 9.310033247325933e-05} +{"train_loss": 0.017624162137508392, "grad_norm": 0.059530261904001236, "global_step": 8968, "epoch": 84, "lr": 9.309883007321104e-05} +{"train_loss": 0.017243005335330963, "grad_norm": 0.058891285210847855, "global_step": 8969, "epoch": 84, "lr": 9.30973275217322e-05} +{"train_loss": 0.01783069781959057, "grad_norm": 0.06357855349779129, "global_step": 8970, "epoch": 84, "lr": 9.309582481882811e-05} +{"train_loss": 0.01818731427192688, "grad_norm": 0.05971192568540573, "global_step": 8971, "epoch": 84, "lr": 9.309432196450401e-05} +{"train_loss": 0.016458217054605484, "grad_norm": 0.10065186768770218, "global_step": 8972, "epoch": 84, "lr": 9.30928189587652e-05} +{"train_loss": 0.0150067787617445, "grad_norm": 0.08497034758329391, "global_step": 8973, "epoch": 84, "lr": 9.309131580161699e-05} +{"train_loss": 0.01835743524134159, "grad_norm": 0.045565828680992126, "global_step": 8974, "epoch": 84, "lr": 9.308981249306461e-05} +{"train_loss": 0.01566135138273239, "grad_norm": 0.04910745471715927, "global_step": 8975, "epoch": 84, "lr": 9.308830903311339e-05} +{"train_loss": 0.020333541557192802, "grad_norm": 0.07543694972991943, "global_step": 8976, "epoch": 84, "lr": 9.308680542176857e-05} +{"train_loss": 0.02059372514486313, "grad_norm": 0.06470336019992828, "global_step": 8977, "epoch": 84, "lr": 9.308530165903546e-05} +{"train_loss": 0.020772092044353485, "grad_norm": 0.044047873467206955, "global_step": 8978, "epoch": 84, "lr": 9.308379774491935e-05} +{"train_loss": 0.021165356040000916, "grad_norm": 0.09892945736646652, "global_step": 8979, "epoch": 84, "lr": 9.30822936794255e-05} +{"train_loss": 0.017476964741945267, "grad_norm": 0.0591830313205719, "global_step": 8980, "epoch": 84, "lr": 9.30807894625592e-05} +{"train_loss": 0.019680576398968697, "grad_norm": 0.15409569442272186, "global_step": 8981, "epoch": 84, "lr": 9.307928509432576e-05} +{"train_loss": 0.02074463479220867, "grad_norm": 0.08045287430286407, "global_step": 8982, "epoch": 84, "lr": 9.307778057473044e-05} +{"train_loss": 0.01608511619269848, "grad_norm": 0.04844595491886139, "global_step": 8983, "epoch": 84, "lr": 9.307627590377853e-05} +{"train_loss": 0.01807922124862671, "grad_norm": 0.050221920013427734, "global_step": 8984, "epoch": 84, "lr": 9.307477108147532e-05} +{"train_loss": 0.01793394796550274, "grad_norm": 0.08666115999221802, "global_step": 8985, "epoch": 84, "lr": 9.307326610782609e-05} +{"train_loss": 0.015867559239268303, "grad_norm": 0.06839684396982193, "global_step": 8986, "epoch": 84, "lr": 9.307176098283614e-05} +{"train_loss": 0.01837535947561264, "grad_norm": 0.04775972664356232, "global_step": 8987, "epoch": 84, "lr": 9.307025570651076e-05} +{"train_loss": 0.01605064608156681, "grad_norm": 0.08599825203418732, "global_step": 8988, "epoch": 84, "lr": 9.306875027885524e-05} +{"train_loss": 0.01709398627281189, "grad_norm": 0.05846943333745003, "global_step": 8989, "epoch": 84, "lr": 9.306724469987485e-05} +{"train_loss": 0.02291288413107395, "grad_norm": 0.059848807752132416, "global_step": 8990, "epoch": 84, "lr": 9.30657389695749e-05} +{"train_loss": 0.01686064898967743, "grad_norm": 0.10918516665697098, "global_step": 8991, "epoch": 84, "lr": 9.306423308796064e-05} +{"train_loss": 0.015279795043170452, "grad_norm": 0.12054341286420822, "global_step": 8992, "epoch": 84, "lr": 9.306272705503743e-05} +{"train_loss": 0.017635708674788475, "grad_norm": 0.06417988985776901, "global_step": 8993, "epoch": 84, "lr": 9.30612208708105e-05} +{"train_loss": 0.017218278720974922, "grad_norm": 0.051668938249349594, "global_step": 8994, "epoch": 84, "lr": 9.305971453528518e-05} +{"train_loss": 0.017982251942157745, "grad_norm": 0.052468959242105484, "global_step": 8995, "epoch": 84, "lr": 9.305820804846675e-05} +{"train_loss": 0.018091294914484024, "grad_norm": 0.10902630537748337, "global_step": 8996, "epoch": 84, "lr": 9.30567014103605e-05} +{"train_loss": 0.015536605380475521, "grad_norm": 0.05604378134012222, "global_step": 8997, "epoch": 84, "lr": 9.305519462097172e-05} +{"train_loss": 0.020386971533298492, "grad_norm": 0.0546402633190155, "global_step": 8998, "epoch": 84, "lr": 9.305368768030572e-05} +{"train_loss": 0.01772356778383255, "grad_norm": 0.14082524180412292, "global_step": 8999, "epoch": 84, "lr": 9.305218058836778e-05} +{"train_loss": 0.017065726220607758, "grad_norm": 0.06900892406702042, "global_step": 9000, "epoch": 84, "lr": 9.30506733451632e-05} +{"train_loss": 0.01862495392560959, "grad_norm": 0.04812861233949661, "global_step": 9001, "epoch": 84, "lr": 9.304916595069729e-05} +{"train_loss": 0.01953211985528469, "grad_norm": 0.06597250699996948, "global_step": 9002, "epoch": 84, "lr": 9.304765840497531e-05} +{"train_loss": 0.018519319593906403, "grad_norm": 0.12104828655719757, "global_step": 9003, "epoch": 84, "lr": 9.304615070800257e-05} +{"train_loss": 0.020468343049287796, "grad_norm": 0.15378384292125702, "global_step": 9004, "epoch": 84, "lr": 9.304464285978442e-05} +{"train_loss": 0.01948920637369156, "grad_norm": 0.06041278317570686, "global_step": 9005, "epoch": 84, "lr": 9.304313486032608e-05} +{"train_loss": 0.01962207444012165, "grad_norm": 0.08362217247486115, "global_step": 9006, "epoch": 84, "lr": 9.30416267096329e-05} +{"train_loss": 0.016081027686595917, "grad_norm": 0.042268943041563034, "global_step": 9007, "epoch": 84, "lr": 9.304011840771015e-05} +{"train_loss": 0.01678188517689705, "grad_norm": 0.05511980131268501, "global_step": 9008, "epoch": 84, "lr": 9.303860995456315e-05} +{"train_loss": 0.017956798070303676, "grad_norm": 0.09502030909061432, "global_step": 9009, "epoch": 84, "lr": 9.30371013501972e-05, "val_loss": 0.019831348210573196} +{"train_loss": 0.016407329589128494, "grad_norm": 0.09625191241502762, "global_step": 9010, "epoch": 85, "lr": 9.303559259461758e-05} +{"train_loss": 0.016653526574373245, "grad_norm": 0.03810827434062958, "global_step": 9011, "epoch": 85, "lr": 9.303408368782959e-05} +{"train_loss": 0.018089814111590385, "grad_norm": 0.054369453340768814, "global_step": 9012, "epoch": 85, "lr": 9.303257462983857e-05} +{"train_loss": 0.018599236384034157, "grad_norm": 0.04344805330038071, "global_step": 9013, "epoch": 85, "lr": 9.303106542064977e-05} +{"train_loss": 0.018175514414906502, "grad_norm": 0.17005471885204315, "global_step": 9014, "epoch": 85, "lr": 9.302955606026854e-05} +{"train_loss": 0.01858803629875183, "grad_norm": 0.1373967081308365, "global_step": 9015, "epoch": 85, "lr": 9.302804654870016e-05} +{"train_loss": 0.017064733430743217, "grad_norm": 0.06888477504253387, "global_step": 9016, "epoch": 85, "lr": 9.302653688594994e-05} +{"train_loss": 0.01648796908557415, "grad_norm": 0.07272136211395264, "global_step": 9017, "epoch": 85, "lr": 9.302502707202317e-05} +{"train_loss": 0.018980154767632484, "grad_norm": 0.08368092775344849, "global_step": 9018, "epoch": 85, "lr": 9.302351710692518e-05} +{"train_loss": 0.016020558774471283, "grad_norm": 0.050764333456754684, "global_step": 9019, "epoch": 85, "lr": 9.302200699066124e-05} +{"train_loss": 0.023234715685248375, "grad_norm": 0.052336059510707855, "global_step": 9020, "epoch": 85, "lr": 9.30204967232367e-05} +{"train_loss": 0.018067331984639168, "grad_norm": 0.05776891112327576, "global_step": 9021, "epoch": 85, "lr": 9.301898630465684e-05} +{"train_loss": 0.017425918951630592, "grad_norm": 0.051690585911273956, "global_step": 9022, "epoch": 85, "lr": 9.301747573492696e-05} +{"train_loss": 0.016875900328159332, "grad_norm": 0.06721625477075577, "global_step": 9023, "epoch": 85, "lr": 9.301596501405238e-05} +{"train_loss": 0.017662543803453445, "grad_norm": 0.07616745680570602, "global_step": 9024, "epoch": 85, "lr": 9.301445414203842e-05} +{"train_loss": 0.018597066402435303, "grad_norm": 0.11126411706209183, "global_step": 9025, "epoch": 85, "lr": 9.301294311889037e-05} +{"train_loss": 0.01664668507874012, "grad_norm": 0.06955817341804504, "global_step": 9026, "epoch": 85, "lr": 9.301143194461352e-05} +{"train_loss": 0.018648885190486908, "grad_norm": 0.10480675101280212, "global_step": 9027, "epoch": 85, "lr": 9.300992061921323e-05} +{"train_loss": 0.015162465162575245, "grad_norm": 0.06623263657093048, "global_step": 9028, "epoch": 85, "lr": 9.300840914269477e-05} +{"train_loss": 0.0169733427464962, "grad_norm": 0.09915987402200699, "global_step": 9029, "epoch": 85, "lr": 9.300689751506347e-05} +{"train_loss": 0.01568041369318962, "grad_norm": 0.06954687833786011, "global_step": 9030, "epoch": 85, "lr": 9.300538573632464e-05} +{"train_loss": 0.019173290580511093, "grad_norm": 0.11162353307008743, "global_step": 9031, "epoch": 85, "lr": 9.300387380648358e-05} +{"train_loss": 0.01603444665670395, "grad_norm": 0.04204583913087845, "global_step": 9032, "epoch": 85, "lr": 9.300236172554562e-05} +{"train_loss": 0.018122930079698563, "grad_norm": 0.05021394416689873, "global_step": 9033, "epoch": 85, "lr": 9.300084949351604e-05} +{"train_loss": 0.018136952072381973, "grad_norm": 0.04964759945869446, "global_step": 9034, "epoch": 85, "lr": 9.299933711040019e-05} +{"train_loss": 0.015260016545653343, "grad_norm": 0.037653084844350815, "global_step": 9035, "epoch": 85, "lr": 9.299782457620337e-05} +{"train_loss": 0.019580619409680367, "grad_norm": 0.048313580453395844, "global_step": 9036, "epoch": 85, "lr": 9.29963118909309e-05} +{"train_loss": 0.017397280782461166, "grad_norm": 0.07140064984560013, "global_step": 9037, "epoch": 85, "lr": 9.299479905458808e-05} +{"train_loss": 0.021371223032474518, "grad_norm": 0.12451812624931335, "global_step": 9038, "epoch": 85, "lr": 9.299328606718023e-05} +{"train_loss": 0.016258107498288155, "grad_norm": 0.045625776052474976, "global_step": 9039, "epoch": 85, "lr": 9.299177292871267e-05} +{"train_loss": 0.01877022162079811, "grad_norm": 0.043944213539361954, "global_step": 9040, "epoch": 85, "lr": 9.299025963919072e-05} +{"train_loss": 0.01737792044878006, "grad_norm": 0.06849230080842972, "global_step": 9041, "epoch": 85, "lr": 9.29887461986197e-05} +{"train_loss": 0.01939805969595909, "grad_norm": 0.10515087097883224, "global_step": 9042, "epoch": 85, "lr": 9.29872326070049e-05} +{"train_loss": 0.017024433240294456, "grad_norm": 0.07542292773723602, "global_step": 9043, "epoch": 85, "lr": 9.298571886435168e-05} +{"train_loss": 0.020563539117574692, "grad_norm": 0.04892859607934952, "global_step": 9044, "epoch": 85, "lr": 9.298420497066533e-05} +{"train_loss": 0.018014343455433846, "grad_norm": 0.061239708214998245, "global_step": 9045, "epoch": 85, "lr": 9.298269092595119e-05} +{"train_loss": 0.018298612907528877, "grad_norm": 0.07752559334039688, "global_step": 9046, "epoch": 85, "lr": 9.298117673021455e-05} +{"train_loss": 0.018427932634949684, "grad_norm": 0.12704023718833923, "global_step": 9047, "epoch": 85, "lr": 9.297966238346076e-05} +{"train_loss": 0.018822431564331055, "grad_norm": 0.08342515677213669, "global_step": 9048, "epoch": 85, "lr": 9.297814788569512e-05} +{"train_loss": 0.02049747109413147, "grad_norm": 0.07733616977930069, "global_step": 9049, "epoch": 85, "lr": 9.297663323692297e-05} +{"train_loss": 0.019305776804685593, "grad_norm": 0.07054618746042252, "global_step": 9050, "epoch": 85, "lr": 9.29751184371496e-05} +{"train_loss": 0.020025456324219704, "grad_norm": 0.0638972595334053, "global_step": 9051, "epoch": 85, "lr": 9.297360348638037e-05} +{"train_loss": 0.02094154618680477, "grad_norm": 0.1460147649049759, "global_step": 9052, "epoch": 85, "lr": 9.297208838462058e-05} +{"train_loss": 0.01834816113114357, "grad_norm": 0.0738232433795929, "global_step": 9053, "epoch": 85, "lr": 9.297057313187556e-05} +{"train_loss": 0.017032790929079056, "grad_norm": 0.08542782068252563, "global_step": 9054, "epoch": 85, "lr": 9.296905772815063e-05} +{"train_loss": 0.017331179231405258, "grad_norm": 0.06050214543938637, "global_step": 9055, "epoch": 85, "lr": 9.296754217345112e-05} +{"train_loss": 0.018787991255521774, "grad_norm": 0.1176268458366394, "global_step": 9056, "epoch": 85, "lr": 9.296602646778236e-05} +{"train_loss": 0.018929149955511093, "grad_norm": 0.04743009805679321, "global_step": 9057, "epoch": 85, "lr": 9.296451061114966e-05} +{"train_loss": 0.01561313308775425, "grad_norm": 0.08904397487640381, "global_step": 9058, "epoch": 85, "lr": 9.296299460355835e-05} +{"train_loss": 0.02015802264213562, "grad_norm": 0.0904574990272522, "global_step": 9059, "epoch": 85, "lr": 9.296147844501377e-05} +{"train_loss": 0.018480796366930008, "grad_norm": 0.08492349833250046, "global_step": 9060, "epoch": 85, "lr": 9.295996213552123e-05} +{"train_loss": 0.017201606184244156, "grad_norm": 0.044296253472566605, "global_step": 9061, "epoch": 85, "lr": 9.295844567508607e-05} +{"train_loss": 0.020400766283273697, "grad_norm": 0.10411586612462997, "global_step": 9062, "epoch": 85, "lr": 9.295692906371363e-05} +{"train_loss": 0.018201248720288277, "grad_norm": 0.0642441064119339, "global_step": 9063, "epoch": 85, "lr": 9.295541230140921e-05} +{"train_loss": 0.01884835585951805, "grad_norm": 0.10729281604290009, "global_step": 9064, "epoch": 85, "lr": 9.295389538817814e-05} +{"train_loss": 0.015419606119394302, "grad_norm": 0.040379904210567474, "global_step": 9065, "epoch": 85, "lr": 9.295237832402578e-05} +{"train_loss": 0.018238656222820282, "grad_norm": 0.12432146817445755, "global_step": 9066, "epoch": 85, "lr": 9.295086110895744e-05} +{"train_loss": 0.01960800401866436, "grad_norm": 0.05414857342839241, "global_step": 9067, "epoch": 85, "lr": 9.294934374297844e-05} +{"train_loss": 0.015280241146683693, "grad_norm": 0.16852891445159912, "global_step": 9068, "epoch": 85, "lr": 9.294782622609412e-05} +{"train_loss": 0.017246101051568985, "grad_norm": 0.11413697153329849, "global_step": 9069, "epoch": 85, "lr": 9.294630855830983e-05} +{"train_loss": 0.017393507063388824, "grad_norm": 0.0825161337852478, "global_step": 9070, "epoch": 85, "lr": 9.294479073963089e-05} +{"train_loss": 0.01734398677945137, "grad_norm": 0.0973578616976738, "global_step": 9071, "epoch": 85, "lr": 9.294327277006263e-05} +{"train_loss": 0.018902156502008438, "grad_norm": 0.08502647280693054, "global_step": 9072, "epoch": 85, "lr": 9.29417546496104e-05} +{"train_loss": 0.01823076233267784, "grad_norm": 0.07426870614290237, "global_step": 9073, "epoch": 85, "lr": 9.294023637827949e-05} +{"train_loss": 0.015783702954649925, "grad_norm": 0.11728860437870026, "global_step": 9074, "epoch": 85, "lr": 9.293871795607527e-05} +{"train_loss": 0.01588115096092224, "grad_norm": 0.06920686364173889, "global_step": 9075, "epoch": 85, "lr": 9.293719938300308e-05} +{"train_loss": 0.017882108688354492, "grad_norm": 0.05830540135502815, "global_step": 9076, "epoch": 85, "lr": 9.293568065906824e-05} +{"train_loss": 0.017832957208156586, "grad_norm": 0.09046930074691772, "global_step": 9077, "epoch": 85, "lr": 9.293416178427608e-05} +{"train_loss": 0.019008580595254898, "grad_norm": 0.06503831595182419, "global_step": 9078, "epoch": 85, "lr": 9.293264275863196e-05} +{"train_loss": 0.016550473868846893, "grad_norm": 0.11250419169664383, "global_step": 9079, "epoch": 85, "lr": 9.29311235821412e-05} +{"train_loss": 0.021901434287428856, "grad_norm": 0.0471043698489666, "global_step": 9080, "epoch": 85, "lr": 9.292960425480914e-05} +{"train_loss": 0.01663041114807129, "grad_norm": 0.05832240357995033, "global_step": 9081, "epoch": 85, "lr": 9.292808477664113e-05} +{"train_loss": 0.019268780946731567, "grad_norm": 0.09962823241949081, "global_step": 9082, "epoch": 85, "lr": 9.292656514764248e-05} +{"train_loss": 0.01604713499546051, "grad_norm": 0.05388086661696434, "global_step": 9083, "epoch": 85, "lr": 9.292504536781856e-05} +{"train_loss": 0.017581118270754814, "grad_norm": 0.060897521674633026, "global_step": 9084, "epoch": 85, "lr": 9.29235254371747e-05} +{"train_loss": 0.016638461500406265, "grad_norm": 0.059480153024196625, "global_step": 9085, "epoch": 85, "lr": 9.292200535571625e-05} +{"train_loss": 0.018442178145051003, "grad_norm": 0.04871896654367447, "global_step": 9086, "epoch": 85, "lr": 9.292048512344853e-05} +{"train_loss": 0.018471578136086464, "grad_norm": 0.08028832077980042, "global_step": 9087, "epoch": 85, "lr": 9.291896474037688e-05} +{"train_loss": 0.01799796149134636, "grad_norm": 0.056066349148750305, "global_step": 9088, "epoch": 85, "lr": 9.291744420650665e-05} +{"train_loss": 0.017497729510068893, "grad_norm": 0.09425521641969681, "global_step": 9089, "epoch": 85, "lr": 9.29159235218432e-05} +{"train_loss": 0.018071172758936882, "grad_norm": 0.058648284524679184, "global_step": 9090, "epoch": 85, "lr": 9.291440268639185e-05} +{"train_loss": 0.014424538239836693, "grad_norm": 0.06883938610553741, "global_step": 9091, "epoch": 85, "lr": 9.291288170015796e-05} +{"train_loss": 0.018307674676179886, "grad_norm": 0.058932121843099594, "global_step": 9092, "epoch": 85, "lr": 9.291136056314686e-05} +{"train_loss": 0.01785130798816681, "grad_norm": 0.058394525200128555, "global_step": 9093, "epoch": 85, "lr": 9.29098392753639e-05} +{"train_loss": 0.018648382276296616, "grad_norm": 0.055082157254219055, "global_step": 9094, "epoch": 85, "lr": 9.290831783681444e-05} +{"train_loss": 0.018199462443590164, "grad_norm": 0.11636017262935638, "global_step": 9095, "epoch": 85, "lr": 9.290679624750378e-05} +{"train_loss": 0.0176534466445446, "grad_norm": 0.04240762069821358, "global_step": 9096, "epoch": 85, "lr": 9.290527450743734e-05} +{"train_loss": 0.019655782729387283, "grad_norm": 0.09285546094179153, "global_step": 9097, "epoch": 85, "lr": 9.29037526166204e-05} +{"train_loss": 0.019814983010292053, "grad_norm": 0.09348152577877045, "global_step": 9098, "epoch": 85, "lr": 9.290223057505833e-05} +{"train_loss": 0.01589176431298256, "grad_norm": 0.07649344950914383, "global_step": 9099, "epoch": 85, "lr": 9.290070838275649e-05} +{"train_loss": 0.018510133028030396, "grad_norm": 0.06839696317911148, "global_step": 9100, "epoch": 85, "lr": 9.28991860397202e-05} +{"train_loss": 0.020020978525280952, "grad_norm": 0.10524150729179382, "global_step": 9101, "epoch": 85, "lr": 9.289766354595484e-05} +{"train_loss": 0.0179985910654068, "grad_norm": 0.0445544570684433, "global_step": 9102, "epoch": 85, "lr": 9.289614090146575e-05} +{"train_loss": 0.016824845224618912, "grad_norm": 0.12094403058290482, "global_step": 9103, "epoch": 85, "lr": 9.289461810625826e-05} +{"train_loss": 0.016653941944241524, "grad_norm": 0.06967926025390625, "global_step": 9104, "epoch": 85, "lr": 9.289309516033775e-05} +{"train_loss": 0.017507733777165413, "grad_norm": 0.04911985620856285, "global_step": 9105, "epoch": 85, "lr": 9.289157206370956e-05} +{"train_loss": 0.015648595988750458, "grad_norm": 0.07650578022003174, "global_step": 9106, "epoch": 85, "lr": 9.289004881637901e-05} +{"train_loss": 0.015359947457909584, "grad_norm": 0.06923605501651764, "global_step": 9107, "epoch": 85, "lr": 9.288852541835151e-05} +{"train_loss": 0.017061816528439522, "grad_norm": 0.05025103688240051, "global_step": 9108, "epoch": 85, "lr": 9.288700186963237e-05} +{"train_loss": 0.020142417401075363, "grad_norm": 0.10321073234081268, "global_step": 9109, "epoch": 85, "lr": 9.288547817022697e-05} +{"train_loss": 0.01740485429763794, "grad_norm": 0.056776780635118484, "global_step": 9110, "epoch": 85, "lr": 9.288395432014064e-05} +{"train_loss": 0.016779018566012383, "grad_norm": 0.0803234651684761, "global_step": 9111, "epoch": 85, "lr": 9.288243031937876e-05} +{"train_loss": 0.017895236611366272, "grad_norm": 0.04989120736718178, "global_step": 9112, "epoch": 85, "lr": 9.288090616794665e-05} +{"train_loss": 0.017320524901151657, "grad_norm": 0.060498788952827454, "global_step": 9113, "epoch": 85, "lr": 9.287938186584968e-05} +{"train_loss": 0.02002939209342003, "grad_norm": 0.08218356221914291, "global_step": 9114, "epoch": 85, "lr": 9.287785741309323e-05} +{"train_loss": 0.017947939619913977, "grad_norm": 0.18981264531612396, "global_step": 9115, "epoch": 85, "lr": 9.287633280968261e-05, "val_loss": 0.018704159185290337} +{"train_loss": 0.01777750253677368, "grad_norm": 0.06157199665904045, "global_step": 9116, "epoch": 86, "lr": 9.287480805562323e-05} +{"train_loss": 0.02139749377965927, "grad_norm": 0.050210002809762955, "global_step": 9117, "epoch": 86, "lr": 9.287328315092041e-05} +{"train_loss": 0.016517076641321182, "grad_norm": 0.06705238670110703, "global_step": 9118, "epoch": 86, "lr": 9.28717580955795e-05} +{"train_loss": 0.020675795152783394, "grad_norm": 0.11834004521369934, "global_step": 9119, "epoch": 86, "lr": 9.28702328896059e-05} +{"train_loss": 0.0182674340903759, "grad_norm": 0.08634887635707855, "global_step": 9120, "epoch": 86, "lr": 9.286870753300494e-05} +{"train_loss": 0.015698056668043137, "grad_norm": 0.11362745612859726, "global_step": 9121, "epoch": 86, "lr": 9.2867182025782e-05} +{"train_loss": 0.022385085001587868, "grad_norm": 0.0908963605761528, "global_step": 9122, "epoch": 86, "lr": 9.28656563679424e-05} +{"train_loss": 0.015716727823019028, "grad_norm": 0.11813023686408997, "global_step": 9123, "epoch": 86, "lr": 9.286413055949152e-05} +{"train_loss": 0.020273357629776, "grad_norm": 0.10323646664619446, "global_step": 9124, "epoch": 86, "lr": 9.286260460043475e-05} +{"train_loss": 0.0181942917406559, "grad_norm": 0.1344653069972992, "global_step": 9125, "epoch": 86, "lr": 9.286107849077743e-05} +{"train_loss": 0.018170705065131187, "grad_norm": 0.07461334764957428, "global_step": 9126, "epoch": 86, "lr": 9.28595522305249e-05} +{"train_loss": 0.015687569975852966, "grad_norm": 0.06587571650743484, "global_step": 9127, "epoch": 86, "lr": 9.285802581968254e-05} +{"train_loss": 0.01568717323243618, "grad_norm": 0.07214018702507019, "global_step": 9128, "epoch": 86, "lr": 9.285649925825573e-05} +{"train_loss": 0.017634112387895584, "grad_norm": 0.06232307851314545, "global_step": 9129, "epoch": 86, "lr": 9.285497254624982e-05} +{"train_loss": 0.016997184604406357, "grad_norm": 0.05138459429144859, "global_step": 9130, "epoch": 86, "lr": 9.285344568367017e-05} +{"train_loss": 0.018134763464331627, "grad_norm": 0.05201338231563568, "global_step": 9131, "epoch": 86, "lr": 9.285191867052214e-05} +{"train_loss": 0.015969600528478622, "grad_norm": 0.07335424423217773, "global_step": 9132, "epoch": 86, "lr": 9.285039150681112e-05} +{"train_loss": 0.015819747000932693, "grad_norm": 0.1501401662826538, "global_step": 9133, "epoch": 86, "lr": 9.284886419254245e-05} +{"train_loss": 0.015452839434146881, "grad_norm": 0.08465327322483063, "global_step": 9134, "epoch": 86, "lr": 9.28473367277215e-05} +{"train_loss": 0.015616519376635551, "grad_norm": 0.0522250235080719, "global_step": 9135, "epoch": 86, "lr": 9.284580911235365e-05} +{"train_loss": 0.016267139464616776, "grad_norm": 0.15324805676937103, "global_step": 9136, "epoch": 86, "lr": 9.284428134644426e-05} +{"train_loss": 0.01961371675133705, "grad_norm": 0.05255482718348503, "global_step": 9137, "epoch": 86, "lr": 9.284275342999869e-05} +{"train_loss": 0.013951409608125687, "grad_norm": 0.04174429550766945, "global_step": 9138, "epoch": 86, "lr": 9.284122536302234e-05} +{"train_loss": 0.019577210769057274, "grad_norm": 0.0481349341571331, "global_step": 9139, "epoch": 86, "lr": 9.283969714552053e-05} +{"train_loss": 0.020511480048298836, "grad_norm": 0.05449267104268074, "global_step": 9140, "epoch": 86, "lr": 9.283816877749867e-05} +{"train_loss": 0.01656370609998703, "grad_norm": 0.11234846711158752, "global_step": 9141, "epoch": 86, "lr": 9.28366402589621e-05} +{"train_loss": 0.020276600494980812, "grad_norm": 0.08324024081230164, "global_step": 9142, "epoch": 86, "lr": 9.283511158991621e-05} +{"train_loss": 0.019769202917814255, "grad_norm": 0.07140243053436279, "global_step": 9143, "epoch": 86, "lr": 9.283358277036638e-05} +{"train_loss": 0.01699146442115307, "grad_norm": 0.04121930152177811, "global_step": 9144, "epoch": 86, "lr": 9.283205380031794e-05} +{"train_loss": 0.019517982378602028, "grad_norm": 0.07784503698348999, "global_step": 9145, "epoch": 86, "lr": 9.283052467977631e-05} +{"train_loss": 0.01751737669110298, "grad_norm": 0.16639405488967896, "global_step": 9146, "epoch": 86, "lr": 9.282899540874686e-05} +{"train_loss": 0.01941627264022827, "grad_norm": 0.04830864071846008, "global_step": 9147, "epoch": 86, "lr": 9.282746598723491e-05} +{"train_loss": 0.01721080206334591, "grad_norm": 0.07593005895614624, "global_step": 9148, "epoch": 86, "lr": 9.282593641524589e-05} +{"train_loss": 0.019011681899428368, "grad_norm": 0.0473729744553566, "global_step": 9149, "epoch": 86, "lr": 9.282440669278517e-05} +{"train_loss": 0.022863158956170082, "grad_norm": 0.06467581540346146, "global_step": 9150, "epoch": 86, "lr": 9.282287681985807e-05} +{"train_loss": 0.017631901428103447, "grad_norm": 0.0665857344865799, "global_step": 9151, "epoch": 86, "lr": 9.282134679647002e-05} +{"train_loss": 0.017579756677150726, "grad_norm": 0.09131938219070435, "global_step": 9152, "epoch": 86, "lr": 9.281981662262638e-05} +{"train_loss": 0.01575188711285591, "grad_norm": 0.06955739110708237, "global_step": 9153, "epoch": 86, "lr": 9.281828629833254e-05} +{"train_loss": 0.016673555597662926, "grad_norm": 0.058698005974292755, "global_step": 9154, "epoch": 86, "lr": 9.281675582359384e-05} +{"train_loss": 0.014923257753252983, "grad_norm": 0.09356839954853058, "global_step": 9155, "epoch": 86, "lr": 9.28152251984157e-05} +{"train_loss": 0.017623960971832275, "grad_norm": 0.09419400990009308, "global_step": 9156, "epoch": 86, "lr": 9.281369442280346e-05} +{"train_loss": 0.019267600029706955, "grad_norm": 0.10728120803833008, "global_step": 9157, "epoch": 86, "lr": 9.281216349676253e-05} +{"train_loss": 0.0180356465280056, "grad_norm": 0.05444946885108948, "global_step": 9158, "epoch": 86, "lr": 9.281063242029829e-05} +{"train_loss": 0.01905059441924095, "grad_norm": 0.09350976347923279, "global_step": 9159, "epoch": 86, "lr": 9.280910119341609e-05} +{"train_loss": 0.018017800524830818, "grad_norm": 0.11538862437009811, "global_step": 9160, "epoch": 86, "lr": 9.280756981612131e-05} +{"train_loss": 0.016780376434326172, "grad_norm": 0.04546121135354042, "global_step": 9161, "epoch": 86, "lr": 9.280603828841936e-05} +{"train_loss": 0.01826365664601326, "grad_norm": 0.10552983731031418, "global_step": 9162, "epoch": 86, "lr": 9.280450661031562e-05} +{"train_loss": 0.020063206553459167, "grad_norm": 0.07199843972921371, "global_step": 9163, "epoch": 86, "lr": 9.280297478181543e-05} +{"train_loss": 0.01907661184668541, "grad_norm": 0.07339738309383392, "global_step": 9164, "epoch": 86, "lr": 9.280144280292422e-05} +{"train_loss": 0.01937514916062355, "grad_norm": 0.15184849500656128, "global_step": 9165, "epoch": 86, "lr": 9.279991067364736e-05} +{"train_loss": 0.0204213447868824, "grad_norm": 0.07792475819587708, "global_step": 9166, "epoch": 86, "lr": 9.279837839399022e-05} +{"train_loss": 0.0207059383392334, "grad_norm": 0.1793026477098465, "global_step": 9167, "epoch": 86, "lr": 9.279684596395818e-05} +{"train_loss": 0.017347872257232666, "grad_norm": 0.135390967130661, "global_step": 9168, "epoch": 86, "lr": 9.279531338355666e-05} +{"train_loss": 0.01937086507678032, "grad_norm": 0.17692428827285767, "global_step": 9169, "epoch": 86, "lr": 9.2793780652791e-05} +{"train_loss": 0.019871648401021957, "grad_norm": 0.11511658132076263, "global_step": 9170, "epoch": 86, "lr": 9.279224777166662e-05} +{"train_loss": 0.02031558007001877, "grad_norm": 0.09939489513635635, "global_step": 9171, "epoch": 86, "lr": 9.279071474018888e-05} +{"train_loss": 0.01833045296370983, "grad_norm": 0.06133625656366348, "global_step": 9172, "epoch": 86, "lr": 9.278918155836317e-05} +{"train_loss": 0.017921676859259605, "grad_norm": 0.055964577943086624, "global_step": 9173, "epoch": 86, "lr": 9.278764822619491e-05} +{"train_loss": 0.01715957000851631, "grad_norm": 0.04278530925512314, "global_step": 9174, "epoch": 86, "lr": 9.278611474368943e-05} +{"train_loss": 0.019173242151737213, "grad_norm": 0.13665401935577393, "global_step": 9175, "epoch": 86, "lr": 9.278458111085216e-05} +{"train_loss": 0.016051143407821655, "grad_norm": 0.11361997574567795, "global_step": 9176, "epoch": 86, "lr": 9.27830473276885e-05} +{"train_loss": 0.018764803186058998, "grad_norm": 0.07129513472318649, "global_step": 9177, "epoch": 86, "lr": 9.27815133942038e-05} +{"train_loss": 0.017411913722753525, "grad_norm": 0.11104694753885269, "global_step": 9178, "epoch": 86, "lr": 9.277997931040347e-05} +{"train_loss": 0.01744561642408371, "grad_norm": 0.06393275409936905, "global_step": 9179, "epoch": 86, "lr": 9.27784450762929e-05} +{"train_loss": 0.018357496708631516, "grad_norm": 0.06464052200317383, "global_step": 9180, "epoch": 86, "lr": 9.277691069187747e-05} +{"train_loss": 0.019037436693906784, "grad_norm": 0.07549992948770523, "global_step": 9181, "epoch": 86, "lr": 9.277537615716258e-05} +{"train_loss": 0.01699548214673996, "grad_norm": 0.08843135833740234, "global_step": 9182, "epoch": 86, "lr": 9.277384147215362e-05} +{"train_loss": 0.02044486254453659, "grad_norm": 0.0728568434715271, "global_step": 9183, "epoch": 86, "lr": 9.277230663685599e-05} +{"train_loss": 0.01704481989145279, "grad_norm": 0.06160661205649376, "global_step": 9184, "epoch": 86, "lr": 9.277077165127507e-05} +{"train_loss": 0.018977483734488487, "grad_norm": 0.11926430463790894, "global_step": 9185, "epoch": 86, "lr": 9.276923651541627e-05} +{"train_loss": 0.01694195345044136, "grad_norm": 0.13575898110866547, "global_step": 9186, "epoch": 86, "lr": 9.276770122928495e-05} +{"train_loss": 0.020336449146270752, "grad_norm": 0.07038673758506775, "global_step": 9187, "epoch": 86, "lr": 9.276616579288655e-05} +{"train_loss": 0.016097214072942734, "grad_norm": 0.03785180673003197, "global_step": 9188, "epoch": 86, "lr": 9.276463020622643e-05} +{"train_loss": 0.018650200217962265, "grad_norm": 0.13339437544345856, "global_step": 9189, "epoch": 86, "lr": 9.276309446931e-05} +{"train_loss": 0.016702167689800262, "grad_norm": 0.07419731467962265, "global_step": 9190, "epoch": 86, "lr": 9.276155858214266e-05} +{"train_loss": 0.01778065413236618, "grad_norm": 0.05580003187060356, "global_step": 9191, "epoch": 86, "lr": 9.276002254472978e-05} +{"train_loss": 0.018134023994207382, "grad_norm": 0.05660619959235191, "global_step": 9192, "epoch": 86, "lr": 9.27584863570768e-05} +{"train_loss": 0.019184302538633347, "grad_norm": 0.0715843215584755, "global_step": 9193, "epoch": 86, "lr": 9.275695001918908e-05} +{"train_loss": 0.016580600291490555, "grad_norm": 0.0359325110912323, "global_step": 9194, "epoch": 86, "lr": 9.275541353107204e-05} +{"train_loss": 0.01684403046965599, "grad_norm": 0.0691981315612793, "global_step": 9195, "epoch": 86, "lr": 9.275387689273107e-05} +{"train_loss": 0.015712065622210503, "grad_norm": 0.05892665311694145, "global_step": 9196, "epoch": 86, "lr": 9.275234010417156e-05} +{"train_loss": 0.016284717246890068, "grad_norm": 0.04790399596095085, "global_step": 9197, "epoch": 86, "lr": 9.275080316539893e-05} +{"train_loss": 0.016391385346651077, "grad_norm": 0.07673736661672592, "global_step": 9198, "epoch": 86, "lr": 9.274926607641856e-05} +{"train_loss": 0.016923371702432632, "grad_norm": 0.05103321745991707, "global_step": 9199, "epoch": 86, "lr": 9.274772883723587e-05} +{"train_loss": 0.015957089141011238, "grad_norm": 0.05169108510017395, "global_step": 9200, "epoch": 86, "lr": 9.274619144785625e-05} +{"train_loss": 0.014612870290875435, "grad_norm": 0.04494733363389969, "global_step": 9201, "epoch": 86, "lr": 9.27446539082851e-05} +{"train_loss": 0.01691782847046852, "grad_norm": 0.04733257368206978, "global_step": 9202, "epoch": 86, "lr": 9.274311621852782e-05} +{"train_loss": 0.01592588983476162, "grad_norm": 0.08654051274061203, "global_step": 9203, "epoch": 86, "lr": 9.274157837858983e-05} +{"train_loss": 0.019855951890349388, "grad_norm": 0.061685990542173386, "global_step": 9204, "epoch": 86, "lr": 9.274004038847651e-05} +{"train_loss": 0.01596890017390251, "grad_norm": 0.08638413995504379, "global_step": 9205, "epoch": 86, "lr": 9.273850224819328e-05} +{"train_loss": 0.016861513257026672, "grad_norm": 0.048965103924274445, "global_step": 9206, "epoch": 86, "lr": 9.273696395774554e-05} +{"train_loss": 0.018493616953492165, "grad_norm": 0.1306067705154419, "global_step": 9207, "epoch": 86, "lr": 9.27354255171387e-05} +{"train_loss": 0.01766480691730976, "grad_norm": 0.06150280684232712, "global_step": 9208, "epoch": 86, "lr": 9.273388692637815e-05} +{"train_loss": 0.016716022044420242, "grad_norm": 0.04801229014992714, "global_step": 9209, "epoch": 86, "lr": 9.273234818546931e-05} +{"train_loss": 0.017161909490823746, "grad_norm": 0.08872700482606888, "global_step": 9210, "epoch": 86, "lr": 9.273080929441757e-05} +{"train_loss": 0.017777202650904655, "grad_norm": 0.04653673991560936, "global_step": 9211, "epoch": 86, "lr": 9.272927025322838e-05} +{"train_loss": 0.014514023438096046, "grad_norm": 0.06978391110897064, "global_step": 9212, "epoch": 86, "lr": 9.272773106190709e-05} +{"train_loss": 0.019347617402672768, "grad_norm": 0.0881311297416687, "global_step": 9213, "epoch": 86, "lr": 9.272619172045915e-05} +{"train_loss": 0.017388710752129555, "grad_norm": 0.0677451640367508, "global_step": 9214, "epoch": 86, "lr": 9.272465222888994e-05} +{"train_loss": 0.02318892627954483, "grad_norm": 0.1320285052061081, "global_step": 9215, "epoch": 86, "lr": 9.272311258720489e-05} +{"train_loss": 0.018095137551426888, "grad_norm": 0.08639680594205856, "global_step": 9216, "epoch": 86, "lr": 9.27215727954094e-05} +{"train_loss": 0.018768252804875374, "grad_norm": 0.0472455732524395, "global_step": 9217, "epoch": 86, "lr": 9.272003285350887e-05} +{"train_loss": 0.01905006915330887, "grad_norm": 0.06679479777812958, "global_step": 9218, "epoch": 86, "lr": 9.271849276150875e-05} +{"train_loss": 0.018994703888893127, "grad_norm": 0.16626697778701782, "global_step": 9219, "epoch": 86, "lr": 9.271695251941442e-05} +{"train_loss": 0.017786767333745956, "grad_norm": 0.07608965784311295, "global_step": 9220, "epoch": 86, "lr": 9.271541212723128e-05} +{"train_loss": 0.017909842543303967, "grad_norm": 0.09745481610298157, "global_step": 9221, "epoch": 86, "lr": 9.271387158496476e-05, "val_loss": 0.019711939617991447} +{"train_loss": 0.017955897375941277, "grad_norm": 0.10532176494598389, "global_step": 9222, "epoch": 87, "lr": 9.271233089262029e-05} +{"train_loss": 0.019243212416768074, "grad_norm": 0.06579508632421494, "global_step": 9223, "epoch": 87, "lr": 9.271079005020325e-05} +{"train_loss": 0.02044156938791275, "grad_norm": 0.09106577187776566, "global_step": 9224, "epoch": 87, "lr": 9.270924905771907e-05} +{"train_loss": 0.01826399564743042, "grad_norm": 0.09893042594194412, "global_step": 9225, "epoch": 87, "lr": 9.270770791517314e-05} +{"train_loss": 0.01831493154168129, "grad_norm": 0.0868794396519661, "global_step": 9226, "epoch": 87, "lr": 9.270616662257093e-05} +{"train_loss": 0.017538800835609436, "grad_norm": 0.09073954820632935, "global_step": 9227, "epoch": 87, "lr": 9.27046251799178e-05} +{"train_loss": 0.017232153564691544, "grad_norm": 0.055764488875865936, "global_step": 9228, "epoch": 87, "lr": 9.270308358721921e-05} +{"train_loss": 0.019320864230394363, "grad_norm": 0.09966246038675308, "global_step": 9229, "epoch": 87, "lr": 9.270154184448054e-05} +{"train_loss": 0.017058808356523514, "grad_norm": 0.04679124057292938, "global_step": 9230, "epoch": 87, "lr": 9.269999995170724e-05} +{"train_loss": 0.01778092235326767, "grad_norm": 0.04808218032121658, "global_step": 9231, "epoch": 87, "lr": 9.26984579089047e-05} +{"train_loss": 0.019450275227427483, "grad_norm": 0.06838368624448776, "global_step": 9232, "epoch": 87, "lr": 9.269691571607833e-05} +{"train_loss": 0.0192398764193058, "grad_norm": 0.093755804002285, "global_step": 9233, "epoch": 87, "lr": 9.26953733732336e-05} +{"train_loss": 0.01807565428316593, "grad_norm": 0.05053408816456795, "global_step": 9234, "epoch": 87, "lr": 9.269383088037587e-05} +{"train_loss": 0.01530156284570694, "grad_norm": 0.056893762201070786, "global_step": 9235, "epoch": 87, "lr": 9.26922882375106e-05} +{"train_loss": 0.015412718988955021, "grad_norm": 0.04775240644812584, "global_step": 9236, "epoch": 87, "lr": 9.269074544464318e-05} +{"train_loss": 0.01574881747364998, "grad_norm": 0.0940094143152237, "global_step": 9237, "epoch": 87, "lr": 9.268920250177906e-05} +{"train_loss": 0.02008848637342453, "grad_norm": 0.08481791615486145, "global_step": 9238, "epoch": 87, "lr": 9.268765940892364e-05} +{"train_loss": 0.02023719809949398, "grad_norm": 0.0647129938006401, "global_step": 9239, "epoch": 87, "lr": 9.268611616608236e-05} +{"train_loss": 0.01563788577914238, "grad_norm": 0.04503946006298065, "global_step": 9240, "epoch": 87, "lr": 9.268457277326062e-05} +{"train_loss": 0.018504468724131584, "grad_norm": 0.04129107668995857, "global_step": 9241, "epoch": 87, "lr": 9.268302923046385e-05} +{"train_loss": 0.01912526786327362, "grad_norm": 0.051084209233522415, "global_step": 9242, "epoch": 87, "lr": 9.26814855376975e-05} +{"train_loss": 0.016865406185388565, "grad_norm": 0.05128001421689987, "global_step": 9243, "epoch": 87, "lr": 9.267994169496694e-05} +{"train_loss": 0.0173978041857481, "grad_norm": 0.09687253832817078, "global_step": 9244, "epoch": 87, "lr": 9.267839770227764e-05} +{"train_loss": 0.020427541807293892, "grad_norm": 0.0493164099752903, "global_step": 9245, "epoch": 87, "lr": 9.267685355963502e-05} +{"train_loss": 0.018651925027370453, "grad_norm": 0.051634419709444046, "global_step": 9246, "epoch": 87, "lr": 9.267530926704449e-05} +{"train_loss": 0.016920294612646103, "grad_norm": 0.04749605059623718, "global_step": 9247, "epoch": 87, "lr": 9.267376482451149e-05} +{"train_loss": 0.020096169784665108, "grad_norm": 0.05822545289993286, "global_step": 9248, "epoch": 87, "lr": 9.267222023204144e-05} +{"train_loss": 0.017441630363464355, "grad_norm": 0.11526024341583252, "global_step": 9249, "epoch": 87, "lr": 9.267067548963975e-05} +{"train_loss": 0.01797567866742611, "grad_norm": 0.06854943931102753, "global_step": 9250, "epoch": 87, "lr": 9.266913059731186e-05} +{"train_loss": 0.014555957168340683, "grad_norm": 0.11180514097213745, "global_step": 9251, "epoch": 87, "lr": 9.26675855550632e-05} +{"train_loss": 0.017145942896604538, "grad_norm": 0.06036478653550148, "global_step": 9252, "epoch": 87, "lr": 9.266604036289923e-05} +{"train_loss": 0.016229484230279922, "grad_norm": 0.056378401815891266, "global_step": 9253, "epoch": 87, "lr": 9.266449502082532e-05} +{"train_loss": 0.017177574336528778, "grad_norm": 0.048828985542058945, "global_step": 9254, "epoch": 87, "lr": 9.266294952884695e-05} +{"train_loss": 0.016807273030281067, "grad_norm": 0.06524693220853806, "global_step": 9255, "epoch": 87, "lr": 9.266140388696951e-05} +{"train_loss": 0.017555788159370422, "grad_norm": 0.06196371838450432, "global_step": 9256, "epoch": 87, "lr": 9.265985809519846e-05} +{"train_loss": 0.013829430565237999, "grad_norm": 0.08589868247509003, "global_step": 9257, "epoch": 87, "lr": 9.265831215353922e-05} +{"train_loss": 0.018017493188381195, "grad_norm": 0.09832388907670975, "global_step": 9258, "epoch": 87, "lr": 9.265676606199722e-05} +{"train_loss": 0.017643868923187256, "grad_norm": 0.061440885066986084, "global_step": 9259, "epoch": 87, "lr": 9.26552198205779e-05} +{"train_loss": 0.01823923923075199, "grad_norm": 0.054896801710128784, "global_step": 9260, "epoch": 87, "lr": 9.265367342928668e-05} +{"train_loss": 0.015420813113451004, "grad_norm": 0.05885927751660347, "global_step": 9261, "epoch": 87, "lr": 9.265212688812901e-05} +{"train_loss": 0.01960870623588562, "grad_norm": 0.07724907249212265, "global_step": 9262, "epoch": 87, "lr": 9.26505801971103e-05} +{"train_loss": 0.018588094040751457, "grad_norm": 0.07359308004379272, "global_step": 9263, "epoch": 87, "lr": 9.264903335623603e-05} +{"train_loss": 0.017811980098485947, "grad_norm": 0.11131900548934937, "global_step": 9264, "epoch": 87, "lr": 9.264748636551157e-05} +{"train_loss": 0.018002355471253395, "grad_norm": 0.06137295812368393, "global_step": 9265, "epoch": 87, "lr": 9.26459392249424e-05} +{"train_loss": 0.019933804869651794, "grad_norm": 0.07145949453115463, "global_step": 9266, "epoch": 87, "lr": 9.264439193453396e-05} +{"train_loss": 0.01587742380797863, "grad_norm": 0.06544118374586105, "global_step": 9267, "epoch": 87, "lr": 9.264284449429166e-05} +{"train_loss": 0.018045995384454727, "grad_norm": 0.06802553683519363, "global_step": 9268, "epoch": 87, "lr": 9.264129690422095e-05} +{"train_loss": 0.016990896314382553, "grad_norm": 0.08722250908613205, "global_step": 9269, "epoch": 87, "lr": 9.263974916432727e-05} +{"train_loss": 0.017167851328849792, "grad_norm": 0.08711112290620804, "global_step": 9270, "epoch": 87, "lr": 9.263820127461605e-05} +{"train_loss": 0.018356066197156906, "grad_norm": 0.1078808605670929, "global_step": 9271, "epoch": 87, "lr": 9.263665323509274e-05} +{"train_loss": 0.018754253163933754, "grad_norm": 0.11397350579500198, "global_step": 9272, "epoch": 87, "lr": 9.263510504576276e-05} +{"train_loss": 0.018909208476543427, "grad_norm": 0.06308332830667496, "global_step": 9273, "epoch": 87, "lr": 9.263355670663159e-05} +{"train_loss": 0.018275799229741096, "grad_norm": 0.07623959332704544, "global_step": 9274, "epoch": 87, "lr": 9.263200821770461e-05} +{"train_loss": 0.019193336367607117, "grad_norm": 0.11924518644809723, "global_step": 9275, "epoch": 87, "lr": 9.263045957898732e-05} +{"train_loss": 0.0170915387570858, "grad_norm": 0.04881574213504791, "global_step": 9276, "epoch": 87, "lr": 9.262891079048513e-05} +{"train_loss": 0.01650945469737053, "grad_norm": 0.048214200884103775, "global_step": 9277, "epoch": 87, "lr": 9.262736185220347e-05} +{"train_loss": 0.018189426511526108, "grad_norm": 0.07596508413553238, "global_step": 9278, "epoch": 87, "lr": 9.26258127641478e-05} +{"train_loss": 0.01660262420773506, "grad_norm": 0.09727732092142105, "global_step": 9279, "epoch": 87, "lr": 9.262426352632358e-05} +{"train_loss": 0.01859196275472641, "grad_norm": 0.08313041180372238, "global_step": 9280, "epoch": 87, "lr": 9.262271413873622e-05} +{"train_loss": 0.017517901957035065, "grad_norm": 0.08122362196445465, "global_step": 9281, "epoch": 87, "lr": 9.262116460139119e-05} +{"train_loss": 0.018106691539287567, "grad_norm": 0.07532622665166855, "global_step": 9282, "epoch": 87, "lr": 9.261961491429391e-05} +{"train_loss": 0.018017729744315147, "grad_norm": 0.09575819224119186, "global_step": 9283, "epoch": 87, "lr": 9.261806507744983e-05} +{"train_loss": 0.018048452213406563, "grad_norm": 0.05137175694108009, "global_step": 9284, "epoch": 87, "lr": 9.261651509086442e-05} +{"train_loss": 0.01694651134312153, "grad_norm": 0.1836467683315277, "global_step": 9285, "epoch": 87, "lr": 9.261496495454311e-05} +{"train_loss": 0.016785267740488052, "grad_norm": 0.04587535187602043, "global_step": 9286, "epoch": 87, "lr": 9.261341466849133e-05} +{"train_loss": 0.017518164590001106, "grad_norm": 0.08619848638772964, "global_step": 9287, "epoch": 87, "lr": 9.261186423271458e-05} +{"train_loss": 0.019362162798643112, "grad_norm": 0.06918417662382126, "global_step": 9288, "epoch": 87, "lr": 9.261031364721823e-05} +{"train_loss": 0.015424218028783798, "grad_norm": 0.04899240657687187, "global_step": 9289, "epoch": 87, "lr": 9.260876291200778e-05} +{"train_loss": 0.017095502465963364, "grad_norm": 0.054880913347005844, "global_step": 9290, "epoch": 87, "lr": 9.260721202708868e-05} +{"train_loss": 0.01694648340344429, "grad_norm": 0.11946139484643936, "global_step": 9291, "epoch": 87, "lr": 9.260566099246635e-05} +{"train_loss": 0.016710998490452766, "grad_norm": 0.05198270082473755, "global_step": 9292, "epoch": 87, "lr": 9.260410980814627e-05} +{"train_loss": 0.017423002049326897, "grad_norm": 0.06899436563253403, "global_step": 9293, "epoch": 87, "lr": 9.260255847413387e-05} +{"train_loss": 0.017067886888980865, "grad_norm": 0.047714121639728546, "global_step": 9294, "epoch": 87, "lr": 9.260100699043462e-05} +{"train_loss": 0.022083288058638573, "grad_norm": 0.05598251894116402, "global_step": 9295, "epoch": 87, "lr": 9.259945535705393e-05} +{"train_loss": 0.01760626584291458, "grad_norm": 0.06682783365249634, "global_step": 9296, "epoch": 87, "lr": 9.259790357399729e-05} +{"train_loss": 0.01914484053850174, "grad_norm": 0.07536397874355316, "global_step": 9297, "epoch": 87, "lr": 9.259635164127016e-05} +{"train_loss": 0.018364224582910538, "grad_norm": 0.05492682382464409, "global_step": 9298, "epoch": 87, "lr": 9.259479955887796e-05} +{"train_loss": 0.018709607422351837, "grad_norm": 0.07888999581336975, "global_step": 9299, "epoch": 87, "lr": 9.259324732682615e-05} +{"train_loss": 0.018043402582406998, "grad_norm": 0.08209255337715149, "global_step": 9300, "epoch": 87, "lr": 9.259169494512022e-05} +{"train_loss": 0.015916286036372185, "grad_norm": 0.14691123366355896, "global_step": 9301, "epoch": 87, "lr": 9.259014241376556e-05} +{"train_loss": 0.01790650188922882, "grad_norm": 0.1172720342874527, "global_step": 9302, "epoch": 87, "lr": 9.25885897327677e-05} +{"train_loss": 0.02190311625599861, "grad_norm": 0.060387421399354935, "global_step": 9303, "epoch": 87, "lr": 9.258703690213201e-05} +{"train_loss": 0.01740424707531929, "grad_norm": 0.0764760673046112, "global_step": 9304, "epoch": 87, "lr": 9.258548392186404e-05} +{"train_loss": 0.020807329565286636, "grad_norm": 0.062488459050655365, "global_step": 9305, "epoch": 87, "lr": 9.258393079196918e-05} +{"train_loss": 0.016998643055558205, "grad_norm": 0.09928303956985474, "global_step": 9306, "epoch": 87, "lr": 9.25823775124529e-05} +{"train_loss": 0.015610450878739357, "grad_norm": 0.13764292001724243, "global_step": 9307, "epoch": 87, "lr": 9.258082408332068e-05} +{"train_loss": 0.014914924278855324, "grad_norm": 0.0588708221912384, "global_step": 9308, "epoch": 87, "lr": 9.257927050457795e-05} +{"train_loss": 0.01740962266921997, "grad_norm": 0.08696787804365158, "global_step": 9309, "epoch": 87, "lr": 9.257771677623017e-05} +{"train_loss": 0.01970072276890278, "grad_norm": 0.1398455798625946, "global_step": 9310, "epoch": 87, "lr": 9.257616289828284e-05} +{"train_loss": 0.01930232159793377, "grad_norm": 0.0627487376332283, "global_step": 9311, "epoch": 87, "lr": 9.257460887074137e-05} +{"train_loss": 0.02072196826338768, "grad_norm": 0.0814751461148262, "global_step": 9312, "epoch": 87, "lr": 9.257305469361124e-05} +{"train_loss": 0.01669776625931263, "grad_norm": 0.09497891366481781, "global_step": 9313, "epoch": 87, "lr": 9.25715003668979e-05} +{"train_loss": 0.016414929181337357, "grad_norm": 0.09030883014202118, "global_step": 9314, "epoch": 87, "lr": 9.256994589060684e-05} +{"train_loss": 0.015751102939248085, "grad_norm": 0.07127681374549866, "global_step": 9315, "epoch": 87, "lr": 9.25683912647435e-05} +{"train_loss": 0.016684530302882195, "grad_norm": 0.0841970220208168, "global_step": 9316, "epoch": 87, "lr": 9.256683648931334e-05} +{"train_loss": 0.018930671736598015, "grad_norm": 0.0625043734908104, "global_step": 9317, "epoch": 87, "lr": 9.256528156432183e-05} +{"train_loss": 0.019429443404078484, "grad_norm": 0.06106140837073326, "global_step": 9318, "epoch": 87, "lr": 9.256372648977443e-05} +{"train_loss": 0.018474498763680458, "grad_norm": 0.07037440687417984, "global_step": 9319, "epoch": 87, "lr": 9.256217126567661e-05} +{"train_loss": 0.017374567687511444, "grad_norm": 0.0563342459499836, "global_step": 9320, "epoch": 87, "lr": 9.256061589203383e-05} +{"train_loss": 0.014971570111811161, "grad_norm": 0.049298178404569626, "global_step": 9321, "epoch": 87, "lr": 9.255906036885155e-05} +{"train_loss": 0.019519992172718048, "grad_norm": 0.13407492637634277, "global_step": 9322, "epoch": 87, "lr": 9.255750469613524e-05} +{"train_loss": 0.01856633648276329, "grad_norm": 0.12396680563688278, "global_step": 9323, "epoch": 87, "lr": 9.255594887389039e-05} +{"train_loss": 0.017318740487098694, "grad_norm": 0.09362117201089859, "global_step": 9324, "epoch": 87, "lr": 9.255439290212241e-05} +{"train_loss": 0.017562385648489, "grad_norm": 0.08734231442213058, "global_step": 9325, "epoch": 87, "lr": 9.255283678083682e-05} +{"train_loss": 0.022588331252336502, "grad_norm": 0.07869242131710052, "global_step": 9326, "epoch": 87, "lr": 9.255128051003905e-05} +{"train_loss": 0.0178688165011271, "grad_norm": 0.08083084225654602, "global_step": 9327, "epoch": 87, "lr": 9.254972408973461e-05, "val_loss": 0.01865505240857601} +{"train_loss": 0.01802457869052887, "grad_norm": 0.10891270637512207, "global_step": 9328, "epoch": 88, "lr": 9.254816751992891e-05} +{"train_loss": 0.020045042037963867, "grad_norm": 0.04982989281415939, "global_step": 9329, "epoch": 88, "lr": 9.254661080062747e-05} +{"train_loss": 0.018207315355539322, "grad_norm": 0.09010646492242813, "global_step": 9330, "epoch": 88, "lr": 9.254505393183575e-05} +{"train_loss": 0.01756107807159424, "grad_norm": 0.05650893598794937, "global_step": 9331, "epoch": 88, "lr": 9.254349691355921e-05} +{"train_loss": 0.018119322136044502, "grad_norm": 0.14675907790660858, "global_step": 9332, "epoch": 88, "lr": 9.254193974580332e-05} +{"train_loss": 0.015833033248782158, "grad_norm": 0.061936650425195694, "global_step": 9333, "epoch": 88, "lr": 9.254038242857356e-05} +{"train_loss": 0.01752496510744095, "grad_norm": 0.07478594034910202, "global_step": 9334, "epoch": 88, "lr": 9.25388249618754e-05} +{"train_loss": 0.018875405192375183, "grad_norm": 0.0919366106390953, "global_step": 9335, "epoch": 88, "lr": 9.253726734571429e-05} +{"train_loss": 0.01762770116329193, "grad_norm": 0.04505318030714989, "global_step": 9336, "epoch": 88, "lr": 9.253570958009574e-05} +{"train_loss": 0.018438993021845818, "grad_norm": 0.061930522322654724, "global_step": 9337, "epoch": 88, "lr": 9.253415166502519e-05} +{"train_loss": 0.01640217751264572, "grad_norm": 0.07369507849216461, "global_step": 9338, "epoch": 88, "lr": 9.253259360050814e-05} +{"train_loss": 0.018199294805526733, "grad_norm": 0.06663765758275986, "global_step": 9339, "epoch": 88, "lr": 9.253103538655005e-05} +{"train_loss": 0.01857656240463257, "grad_norm": 0.1450992077589035, "global_step": 9340, "epoch": 88, "lr": 9.25294770231564e-05} +{"train_loss": 0.019596848636865616, "grad_norm": 0.06900012493133545, "global_step": 9341, "epoch": 88, "lr": 9.252791851033266e-05} +{"train_loss": 0.016590725630521774, "grad_norm": 0.07251245528459549, "global_step": 9342, "epoch": 88, "lr": 9.252635984808432e-05} +{"train_loss": 0.018285611644387245, "grad_norm": 0.10442653298377991, "global_step": 9343, "epoch": 88, "lr": 9.252480103641681e-05} +{"train_loss": 0.01806071773171425, "grad_norm": 0.11017826199531555, "global_step": 9344, "epoch": 88, "lr": 9.252324207533567e-05} +{"train_loss": 0.019215762615203857, "grad_norm": 0.0611809678375721, "global_step": 9345, "epoch": 88, "lr": 9.252168296484636e-05} +{"train_loss": 0.017175855115056038, "grad_norm": 0.07019215822219849, "global_step": 9346, "epoch": 88, "lr": 9.252012370495434e-05} +{"train_loss": 0.018590962514281273, "grad_norm": 0.11996110528707504, "global_step": 9347, "epoch": 88, "lr": 9.251856429566508e-05} +{"train_loss": 0.01777220144867897, "grad_norm": 0.061043139547109604, "global_step": 9348, "epoch": 88, "lr": 9.251700473698409e-05} +{"train_loss": 0.016619421541690826, "grad_norm": 0.06425672769546509, "global_step": 9349, "epoch": 88, "lr": 9.251544502891685e-05} +{"train_loss": 0.02103988267481327, "grad_norm": 0.07978994399309158, "global_step": 9350, "epoch": 88, "lr": 9.25138851714688e-05} +{"train_loss": 0.016965240240097046, "grad_norm": 0.054683294147253036, "global_step": 9351, "epoch": 88, "lr": 9.251232516464545e-05} +{"train_loss": 0.02221624180674553, "grad_norm": 0.08909782767295837, "global_step": 9352, "epoch": 88, "lr": 9.251076500845229e-05} +{"train_loss": 0.019873391836881638, "grad_norm": 0.18135669827461243, "global_step": 9353, "epoch": 88, "lr": 9.250920470289478e-05} +{"train_loss": 0.017992058768868446, "grad_norm": 0.05613742396235466, "global_step": 9354, "epoch": 88, "lr": 9.250764424797842e-05} +{"train_loss": 0.015911288559436798, "grad_norm": 0.058757979422807693, "global_step": 9355, "epoch": 88, "lr": 9.250608364370867e-05} +{"train_loss": 0.016627898439764977, "grad_norm": 0.07805366069078445, "global_step": 9356, "epoch": 88, "lr": 9.250452289009104e-05} +{"train_loss": 0.017894472926855087, "grad_norm": 0.04783390834927559, "global_step": 9357, "epoch": 88, "lr": 9.2502961987131e-05} +{"train_loss": 0.018237095326185226, "grad_norm": 0.056183021515607834, "global_step": 9358, "epoch": 88, "lr": 9.250140093483404e-05} +{"train_loss": 0.021734070032835007, "grad_norm": 0.11535809189081192, "global_step": 9359, "epoch": 88, "lr": 9.249983973320563e-05} +{"train_loss": 0.01746022328734398, "grad_norm": 0.05931266397237778, "global_step": 9360, "epoch": 88, "lr": 9.249827838225127e-05} +{"train_loss": 0.016270212829113007, "grad_norm": 0.09588101506233215, "global_step": 9361, "epoch": 88, "lr": 9.249671688197645e-05} +{"train_loss": 0.0170911718159914, "grad_norm": 0.09867379069328308, "global_step": 9362, "epoch": 88, "lr": 9.249515523238664e-05} +{"train_loss": 0.021002985537052155, "grad_norm": 0.050860561430454254, "global_step": 9363, "epoch": 88, "lr": 9.249359343348732e-05} +{"train_loss": 0.01686634123325348, "grad_norm": 0.06421998888254166, "global_step": 9364, "epoch": 88, "lr": 9.249203148528402e-05} +{"train_loss": 0.01782383769750595, "grad_norm": 0.10442592948675156, "global_step": 9365, "epoch": 88, "lr": 9.249046938778219e-05} +{"train_loss": 0.0187862329185009, "grad_norm": 0.07061097025871277, "global_step": 9366, "epoch": 88, "lr": 9.248890714098733e-05} +{"train_loss": 0.017891759052872658, "grad_norm": 0.08926153182983398, "global_step": 9367, "epoch": 88, "lr": 9.248734474490492e-05} +{"train_loss": 0.017887435853481293, "grad_norm": 0.10158409923315048, "global_step": 9368, "epoch": 88, "lr": 9.248578219954046e-05} +{"train_loss": 0.02074325457215309, "grad_norm": 0.15420427918434143, "global_step": 9369, "epoch": 88, "lr": 9.248421950489944e-05} +{"train_loss": 0.01589658297598362, "grad_norm": 0.05563287436962128, "global_step": 9370, "epoch": 88, "lr": 9.248265666098735e-05} +{"train_loss": 0.016972700133919716, "grad_norm": 0.11182954907417297, "global_step": 9371, "epoch": 88, "lr": 9.248109366780969e-05} +{"train_loss": 0.01714302971959114, "grad_norm": 0.08204665780067444, "global_step": 9372, "epoch": 88, "lr": 9.247953052537192e-05} +{"train_loss": 0.01446540467441082, "grad_norm": 0.0456034392118454, "global_step": 9373, "epoch": 88, "lr": 9.247796723367956e-05} +{"train_loss": 0.018136972561478615, "grad_norm": 0.13926303386688232, "global_step": 9374, "epoch": 88, "lr": 9.247640379273811e-05} +{"train_loss": 0.01580893248319626, "grad_norm": 0.10913535952568054, "global_step": 9375, "epoch": 88, "lr": 9.247484020255304e-05} +{"train_loss": 0.018887817859649658, "grad_norm": 0.11983507126569748, "global_step": 9376, "epoch": 88, "lr": 9.247327646312985e-05} +{"train_loss": 0.021300293505191803, "grad_norm": 0.16698594391345978, "global_step": 9377, "epoch": 88, "lr": 9.247171257447404e-05} +{"train_loss": 0.02152072638273239, "grad_norm": 0.06261048465967178, "global_step": 9378, "epoch": 88, "lr": 9.247014853659111e-05} +{"train_loss": 0.01799161173403263, "grad_norm": 0.1164376363158226, "global_step": 9379, "epoch": 88, "lr": 9.246858434948653e-05} +{"train_loss": 0.018429867923259735, "grad_norm": 0.11085992306470871, "global_step": 9380, "epoch": 88, "lr": 9.246702001316583e-05} +{"train_loss": 0.01679489016532898, "grad_norm": 0.08557401597499847, "global_step": 9381, "epoch": 88, "lr": 9.246545552763448e-05} +{"train_loss": 0.019147582352161407, "grad_norm": 0.06707125157117844, "global_step": 9382, "epoch": 88, "lr": 9.2463890892898e-05} +{"train_loss": 0.01891261525452137, "grad_norm": 0.06728742271661758, "global_step": 9383, "epoch": 88, "lr": 9.246232610896187e-05} +{"train_loss": 0.02021341398358345, "grad_norm": 0.09622680395841599, "global_step": 9384, "epoch": 88, "lr": 9.246076117583159e-05} +{"train_loss": 0.0165335014462471, "grad_norm": 0.10844453424215317, "global_step": 9385, "epoch": 88, "lr": 9.245919609351266e-05} +{"train_loss": 0.018367694690823555, "grad_norm": 0.061511002480983734, "global_step": 9386, "epoch": 88, "lr": 9.245763086201058e-05} +{"train_loss": 0.019916368648409843, "grad_norm": 0.11814802885055542, "global_step": 9387, "epoch": 88, "lr": 9.245606548133084e-05} +{"train_loss": 0.020117830485105515, "grad_norm": 0.08933036774396896, "global_step": 9388, "epoch": 88, "lr": 9.245449995147897e-05} +{"train_loss": 0.017095472663640976, "grad_norm": 0.06101938337087631, "global_step": 9389, "epoch": 88, "lr": 9.245293427246044e-05} +{"train_loss": 0.01614922657608986, "grad_norm": 0.056400131434202194, "global_step": 9390, "epoch": 88, "lr": 9.245136844428075e-05} +{"train_loss": 0.014074563980102539, "grad_norm": 0.041188690811395645, "global_step": 9391, "epoch": 88, "lr": 9.244980246694544e-05} +{"train_loss": 0.01535848155617714, "grad_norm": 0.044288959354162216, "global_step": 9392, "epoch": 88, "lr": 9.244823634045996e-05} +{"train_loss": 0.016576766967773438, "grad_norm": 0.1211911216378212, "global_step": 9393, "epoch": 88, "lr": 9.244667006482985e-05} +{"train_loss": 0.020021945238113403, "grad_norm": 0.05613381788134575, "global_step": 9394, "epoch": 88, "lr": 9.244510364006059e-05} +{"train_loss": 0.014811082743108273, "grad_norm": 0.08404609560966492, "global_step": 9395, "epoch": 88, "lr": 9.244353706615771e-05} +{"train_loss": 0.016270117834210396, "grad_norm": 0.06585628539323807, "global_step": 9396, "epoch": 88, "lr": 9.24419703431267e-05} +{"train_loss": 0.018350178375840187, "grad_norm": 0.04376957565546036, "global_step": 9397, "epoch": 88, "lr": 9.244040347097305e-05} +{"train_loss": 0.01845981366932392, "grad_norm": 0.09571453183889389, "global_step": 9398, "epoch": 88, "lr": 9.243883644970228e-05} +{"train_loss": 0.0173113364726305, "grad_norm": 0.05218236893415451, "global_step": 9399, "epoch": 88, "lr": 9.243726927931991e-05} +{"train_loss": 0.01653391122817993, "grad_norm": 0.08678887039422989, "global_step": 9400, "epoch": 88, "lr": 9.24357019598314e-05} +{"train_loss": 0.020072603598237038, "grad_norm": 0.04453287273645401, "global_step": 9401, "epoch": 88, "lr": 9.243413449124232e-05} +{"train_loss": 0.01932668313384056, "grad_norm": 0.08043846487998962, "global_step": 9402, "epoch": 88, "lr": 9.243256687355813e-05} +{"train_loss": 0.015536151826381683, "grad_norm": 0.050358571112155914, "global_step": 9403, "epoch": 88, "lr": 9.243099910678436e-05} +{"train_loss": 0.018268216401338577, "grad_norm": 0.08961006253957748, "global_step": 9404, "epoch": 88, "lr": 9.24294311909265e-05} +{"train_loss": 0.018278874456882477, "grad_norm": 0.09883073717355728, "global_step": 9405, "epoch": 88, "lr": 9.242786312599007e-05} +{"train_loss": 0.017439354211091995, "grad_norm": 0.15296334028244019, "global_step": 9406, "epoch": 88, "lr": 9.24262949119806e-05} +{"train_loss": 0.018064294010400772, "grad_norm": 0.062284260988235474, "global_step": 9407, "epoch": 88, "lr": 9.242472654890357e-05} +{"train_loss": 0.019565168768167496, "grad_norm": 0.06094789132475853, "global_step": 9408, "epoch": 88, "lr": 9.242315803676451e-05} +{"train_loss": 0.01768484152853489, "grad_norm": 0.09024178236722946, "global_step": 9409, "epoch": 88, "lr": 9.242158937556891e-05} +{"train_loss": 0.018500633537769318, "grad_norm": 0.07554308325052261, "global_step": 9410, "epoch": 88, "lr": 9.24200205653223e-05} +{"train_loss": 0.017190944403409958, "grad_norm": 0.04556624963879585, "global_step": 9411, "epoch": 88, "lr": 9.241845160603018e-05} +{"train_loss": 0.018367299810051918, "grad_norm": 0.1397126168012619, "global_step": 9412, "epoch": 88, "lr": 9.241688249769806e-05} +{"train_loss": 0.01918676309287548, "grad_norm": 0.097714863717556, "global_step": 9413, "epoch": 88, "lr": 9.241531324033148e-05} +{"train_loss": 0.018332894891500473, "grad_norm": 0.09784231334924698, "global_step": 9414, "epoch": 88, "lr": 9.241374383393593e-05} +{"train_loss": 0.014910120517015457, "grad_norm": 0.05913928523659706, "global_step": 9415, "epoch": 88, "lr": 9.241217427851693e-05} +{"train_loss": 0.01722949743270874, "grad_norm": 0.05804428458213806, "global_step": 9416, "epoch": 88, "lr": 9.241060457408e-05} +{"train_loss": 0.020812803879380226, "grad_norm": 0.07558958232402802, "global_step": 9417, "epoch": 88, "lr": 9.240903472063063e-05} +{"train_loss": 0.017011960968375206, "grad_norm": 0.06739423424005508, "global_step": 9418, "epoch": 88, "lr": 9.240746471817437e-05} +{"train_loss": 0.016985656693577766, "grad_norm": 0.18722283840179443, "global_step": 9419, "epoch": 88, "lr": 9.240589456671672e-05} +{"train_loss": 0.01599641516804695, "grad_norm": 0.0908861979842186, "global_step": 9420, "epoch": 88, "lr": 9.240432426626319e-05} +{"train_loss": 0.01753564551472664, "grad_norm": 0.07415688037872314, "global_step": 9421, "epoch": 88, "lr": 9.240275381681931e-05} +{"train_loss": 0.02148691564798355, "grad_norm": 0.0580785758793354, "global_step": 9422, "epoch": 88, "lr": 9.24011832183906e-05} +{"train_loss": 0.016112618148326874, "grad_norm": 0.0838824212551117, "global_step": 9423, "epoch": 88, "lr": 9.239961247098256e-05} +{"train_loss": 0.016563517972826958, "grad_norm": 0.07853341102600098, "global_step": 9424, "epoch": 88, "lr": 9.239804157460074e-05} +{"train_loss": 0.018929049372673035, "grad_norm": 0.05122888833284378, "global_step": 9425, "epoch": 88, "lr": 9.239647052925063e-05} +{"train_loss": 0.015614108182489872, "grad_norm": 0.09839366376399994, "global_step": 9426, "epoch": 88, "lr": 9.239489933493775e-05} +{"train_loss": 0.015392018482089043, "grad_norm": 0.1154295951128006, "global_step": 9427, "epoch": 88, "lr": 9.239332799166765e-05} +{"train_loss": 0.01651051454246044, "grad_norm": 0.057301703840494156, "global_step": 9428, "epoch": 88, "lr": 9.23917564994458e-05} +{"train_loss": 0.017438339069485664, "grad_norm": 0.05856668949127197, "global_step": 9429, "epoch": 88, "lr": 9.239018485827779e-05} +{"train_loss": 0.015616277232766151, "grad_norm": 0.08399506658315659, "global_step": 9430, "epoch": 88, "lr": 9.238861306816908e-05} +{"train_loss": 0.019685054197907448, "grad_norm": 0.04888283461332321, "global_step": 9431, "epoch": 88, "lr": 9.238704112912525e-05} +{"train_loss": 0.017340190708637238, "grad_norm": 0.09489445388317108, "global_step": 9432, "epoch": 88, "lr": 9.238546904115175e-05} +{"train_loss": 0.017871632469150255, "grad_norm": 0.055005092173814774, "global_step": 9433, "epoch": 88, "lr": 9.238389680425416e-05, "val_loss": 0.01885913498699665} +{"train_loss": 0.01576421782374382, "grad_norm": 0.04045704007148743, "global_step": 9434, "epoch": 89, "lr": 9.238232441843799e-05} +{"train_loss": 0.01929854229092598, "grad_norm": 0.16993923485279083, "global_step": 9435, "epoch": 89, "lr": 9.238075188370878e-05} +{"train_loss": 0.018153000622987747, "grad_norm": 0.08405428379774094, "global_step": 9436, "epoch": 89, "lr": 9.237917920007202e-05} +{"train_loss": 0.017961949110031128, "grad_norm": 0.09059524536132812, "global_step": 9437, "epoch": 89, "lr": 9.237760636753325e-05} +{"train_loss": 0.019939569756388664, "grad_norm": 0.07738090306520462, "global_step": 9438, "epoch": 89, "lr": 9.237603338609802e-05} +{"train_loss": 0.016888730227947235, "grad_norm": 0.045884642750024796, "global_step": 9439, "epoch": 89, "lr": 9.237446025577183e-05} +{"train_loss": 0.017551656812429428, "grad_norm": 0.12216030061244965, "global_step": 9440, "epoch": 89, "lr": 9.237288697656021e-05} +{"train_loss": 0.015408564358949661, "grad_norm": 0.08526851236820221, "global_step": 9441, "epoch": 89, "lr": 9.23713135484687e-05} +{"train_loss": 0.017178602516651154, "grad_norm": 0.040805183351039886, "global_step": 9442, "epoch": 89, "lr": 9.236973997150282e-05} +{"train_loss": 0.014740202575922012, "grad_norm": 0.08005298674106598, "global_step": 9443, "epoch": 89, "lr": 9.236816624566809e-05} +{"train_loss": 0.017731737345457077, "grad_norm": 0.09731131047010422, "global_step": 9444, "epoch": 89, "lr": 9.236659237097006e-05} +{"train_loss": 0.018366655334830284, "grad_norm": 0.04767885059118271, "global_step": 9445, "epoch": 89, "lr": 9.236501834741425e-05} +{"train_loss": 0.018968358635902405, "grad_norm": 0.13168694078922272, "global_step": 9446, "epoch": 89, "lr": 9.236344417500618e-05} +{"train_loss": 0.019406069070100784, "grad_norm": 0.06530357897281647, "global_step": 9447, "epoch": 89, "lr": 9.23618698537514e-05} +{"train_loss": 0.016320865601301193, "grad_norm": 0.08770007640123367, "global_step": 9448, "epoch": 89, "lr": 9.236029538365542e-05} +{"train_loss": 0.019257457926869392, "grad_norm": 0.05656201392412186, "global_step": 9449, "epoch": 89, "lr": 9.235872076472378e-05} +{"train_loss": 0.01733255758881569, "grad_norm": 0.05436857044696808, "global_step": 9450, "epoch": 89, "lr": 9.235714599696203e-05} +{"train_loss": 0.015021933242678642, "grad_norm": 0.07579494267702103, "global_step": 9451, "epoch": 89, "lr": 9.235557108037569e-05} +{"train_loss": 0.019961275160312653, "grad_norm": 0.07781562954187393, "global_step": 9452, "epoch": 89, "lr": 9.235399601497029e-05} +{"train_loss": 0.021390821784734726, "grad_norm": 0.11401993036270142, "global_step": 9453, "epoch": 89, "lr": 9.235242080075137e-05} +{"train_loss": 0.02178766578435898, "grad_norm": 0.05253095179796219, "global_step": 9454, "epoch": 89, "lr": 9.235084543772445e-05} +{"train_loss": 0.019129689782857895, "grad_norm": 0.06826955825090408, "global_step": 9455, "epoch": 89, "lr": 9.23492699258951e-05} +{"train_loss": 0.017267491668462753, "grad_norm": 0.11776988208293915, "global_step": 9456, "epoch": 89, "lr": 9.234769426526881e-05} +{"train_loss": 0.017770908772945404, "grad_norm": 0.1912507712841034, "global_step": 9457, "epoch": 89, "lr": 9.234611845585115e-05} +{"train_loss": 0.018832817673683167, "grad_norm": 0.10507849603891373, "global_step": 9458, "epoch": 89, "lr": 9.234454249764764e-05} +{"train_loss": 0.0186434555798769, "grad_norm": 0.04972481355071068, "global_step": 9459, "epoch": 89, "lr": 9.234296639066383e-05} +{"train_loss": 0.018516674637794495, "grad_norm": 0.06717897951602936, "global_step": 9460, "epoch": 89, "lr": 9.234139013490525e-05} +{"train_loss": 0.020647723227739334, "grad_norm": 0.1018744483590126, "global_step": 9461, "epoch": 89, "lr": 9.233981373037743e-05} +{"train_loss": 0.015682172030210495, "grad_norm": 0.04089782014489174, "global_step": 9462, "epoch": 89, "lr": 9.233823717708592e-05} +{"train_loss": 0.016540780663490295, "grad_norm": 0.08365540951490402, "global_step": 9463, "epoch": 89, "lr": 9.233666047503627e-05} +{"train_loss": 0.016958802938461304, "grad_norm": 0.051519881933927536, "global_step": 9464, "epoch": 89, "lr": 9.233508362423399e-05} +{"train_loss": 0.01586393266916275, "grad_norm": 0.05452289804816246, "global_step": 9465, "epoch": 89, "lr": 9.233350662468465e-05} +{"train_loss": 0.016893763095140457, "grad_norm": 0.08510778844356537, "global_step": 9466, "epoch": 89, "lr": 9.233192947639378e-05} +{"train_loss": 0.01725195348262787, "grad_norm": 0.09104365110397339, "global_step": 9467, "epoch": 89, "lr": 9.233035217936689e-05} +{"train_loss": 0.019442006945610046, "grad_norm": 0.13499607145786285, "global_step": 9468, "epoch": 89, "lr": 9.23287747336096e-05} +{"train_loss": 0.020255841314792633, "grad_norm": 0.11411124467849731, "global_step": 9469, "epoch": 89, "lr": 9.232719713912737e-05} +{"train_loss": 0.019149068742990494, "grad_norm": 0.08491576462984085, "global_step": 9470, "epoch": 89, "lr": 9.232561939592578e-05} +{"train_loss": 0.015727832913398743, "grad_norm": 0.06369129568338394, "global_step": 9471, "epoch": 89, "lr": 9.232404150401037e-05} +{"train_loss": 0.01735251396894455, "grad_norm": 0.054187141358852386, "global_step": 9472, "epoch": 89, "lr": 9.232246346338669e-05} +{"train_loss": 0.017288438975811005, "grad_norm": 0.052910223603248596, "global_step": 9473, "epoch": 89, "lr": 9.232088527406029e-05} +{"train_loss": 0.017591256648302078, "grad_norm": 0.05126752331852913, "global_step": 9474, "epoch": 89, "lr": 9.231930693603668e-05} +{"train_loss": 0.018033411353826523, "grad_norm": 0.06453534215688705, "global_step": 9475, "epoch": 89, "lr": 9.231772844932144e-05} +{"train_loss": 0.018306229263544083, "grad_norm": 0.07782012969255447, "global_step": 9476, "epoch": 89, "lr": 9.23161498139201e-05} +{"train_loss": 0.016558784991502762, "grad_norm": 0.1180686503648758, "global_step": 9477, "epoch": 89, "lr": 9.231457102983823e-05} +{"train_loss": 0.01774250902235508, "grad_norm": 0.057394497096538544, "global_step": 9478, "epoch": 89, "lr": 9.231299209708135e-05} +{"train_loss": 0.014800578355789185, "grad_norm": 0.11461897194385529, "global_step": 9479, "epoch": 89, "lr": 9.231141301565502e-05} +{"train_loss": 0.02000940963625908, "grad_norm": 0.058602165430784225, "global_step": 9480, "epoch": 89, "lr": 9.230983378556477e-05} +{"train_loss": 0.018994972109794617, "grad_norm": 0.06490722298622131, "global_step": 9481, "epoch": 89, "lr": 9.230825440681618e-05} +{"train_loss": 0.017819803208112717, "grad_norm": 0.06040240079164505, "global_step": 9482, "epoch": 89, "lr": 9.230667487941478e-05} +{"train_loss": 0.01808352582156658, "grad_norm": 0.0469166524708271, "global_step": 9483, "epoch": 89, "lr": 9.230509520336612e-05} +{"train_loss": 0.01737728714942932, "grad_norm": 0.15611790120601654, "global_step": 9484, "epoch": 89, "lr": 9.230351537867576e-05} +{"train_loss": 0.019598467275500298, "grad_norm": 0.12005636841058731, "global_step": 9485, "epoch": 89, "lr": 9.230193540534924e-05} +{"train_loss": 0.02038307674229145, "grad_norm": 0.05408186838030815, "global_step": 9486, "epoch": 89, "lr": 9.230035528339211e-05} +{"train_loss": 0.018343660980463028, "grad_norm": 0.08021474629640579, "global_step": 9487, "epoch": 89, "lr": 9.229877501280994e-05} +{"train_loss": 0.018706129863858223, "grad_norm": 0.10352714359760284, "global_step": 9488, "epoch": 89, "lr": 9.229719459360826e-05} +{"train_loss": 0.01722627691924572, "grad_norm": 0.0631895586848259, "global_step": 9489, "epoch": 89, "lr": 9.229561402579264e-05} +{"train_loss": 0.01646510511636734, "grad_norm": 0.08239684998989105, "global_step": 9490, "epoch": 89, "lr": 9.22940333093686e-05} +{"train_loss": 0.01623121276497841, "grad_norm": 0.05389454960823059, "global_step": 9491, "epoch": 89, "lr": 9.229245244434176e-05} +{"train_loss": 0.016007758677005768, "grad_norm": 0.097325898706913, "global_step": 9492, "epoch": 89, "lr": 9.229087143071762e-05} +{"train_loss": 0.01681065745651722, "grad_norm": 0.06548362970352173, "global_step": 9493, "epoch": 89, "lr": 9.228929026850175e-05} +{"train_loss": 0.014962615445256233, "grad_norm": 0.05917840823531151, "global_step": 9494, "epoch": 89, "lr": 9.22877089576997e-05} +{"train_loss": 0.021375056356191635, "grad_norm": 0.10802537202835083, "global_step": 9495, "epoch": 89, "lr": 9.228612749831702e-05} +{"train_loss": 0.017493639141321182, "grad_norm": 0.13271546363830566, "global_step": 9496, "epoch": 89, "lr": 9.22845458903593e-05} +{"train_loss": 0.015955306589603424, "grad_norm": 0.1137772798538208, "global_step": 9497, "epoch": 89, "lr": 9.228296413383206e-05} +{"train_loss": 0.02244003489613533, "grad_norm": 0.08215484768152237, "global_step": 9498, "epoch": 89, "lr": 9.228138222874088e-05} +{"train_loss": 0.021580779924988747, "grad_norm": 0.14191395044326782, "global_step": 9499, "epoch": 89, "lr": 9.22798001750913e-05} +{"train_loss": 0.02031969651579857, "grad_norm": 0.1213451400399208, "global_step": 9500, "epoch": 89, "lr": 9.22782179728889e-05} +{"train_loss": 0.018848873674869537, "grad_norm": 0.07769095152616501, "global_step": 9501, "epoch": 89, "lr": 9.227663562213923e-05} +{"train_loss": 0.01750103011727333, "grad_norm": 0.07360664755105972, "global_step": 9502, "epoch": 89, "lr": 9.227505312284784e-05} +{"train_loss": 0.018874485045671463, "grad_norm": 0.0793662741780281, "global_step": 9503, "epoch": 89, "lr": 9.227347047502031e-05} +{"train_loss": 0.01652776077389717, "grad_norm": 0.09327170252799988, "global_step": 9504, "epoch": 89, "lr": 9.227188767866217e-05} +{"train_loss": 0.019753199070692062, "grad_norm": 0.08213599771261215, "global_step": 9505, "epoch": 89, "lr": 9.227030473377902e-05} +{"train_loss": 0.015787895768880844, "grad_norm": 0.038283564150333405, "global_step": 9506, "epoch": 89, "lr": 9.22687216403764e-05} +{"train_loss": 0.01722010411322117, "grad_norm": 0.05402251332998276, "global_step": 9507, "epoch": 89, "lr": 9.226713839845986e-05} +{"train_loss": 0.015832822769880295, "grad_norm": 0.05265453830361366, "global_step": 9508, "epoch": 89, "lr": 9.226555500803498e-05} +{"train_loss": 0.015686385333538055, "grad_norm": 0.13487131893634796, "global_step": 9509, "epoch": 89, "lr": 9.226397146910733e-05} +{"train_loss": 0.01551432441920042, "grad_norm": 0.05763344466686249, "global_step": 9510, "epoch": 89, "lr": 9.226238778168246e-05} +{"train_loss": 0.015264075249433517, "grad_norm": 0.08483494073152542, "global_step": 9511, "epoch": 89, "lr": 9.226080394576594e-05} +{"train_loss": 0.019933082163333893, "grad_norm": 0.10344922542572021, "global_step": 9512, "epoch": 89, "lr": 9.225921996136333e-05} +{"train_loss": 0.016295813024044037, "grad_norm": 0.059319011867046356, "global_step": 9513, "epoch": 89, "lr": 9.225763582848019e-05} +{"train_loss": 0.01921878382563591, "grad_norm": 0.059027496725320816, "global_step": 9514, "epoch": 89, "lr": 9.225605154712211e-05} +{"train_loss": 0.017974989488720894, "grad_norm": 0.06152607500553131, "global_step": 9515, "epoch": 89, "lr": 9.225446711729463e-05} +{"train_loss": 0.018185442313551903, "grad_norm": 0.12949725985527039, "global_step": 9516, "epoch": 89, "lr": 9.225288253900335e-05} +{"train_loss": 0.017952878028154373, "grad_norm": 0.08232153952121735, "global_step": 9517, "epoch": 89, "lr": 9.225129781225382e-05} +{"train_loss": 0.017792485654354095, "grad_norm": 0.10972711443901062, "global_step": 9518, "epoch": 89, "lr": 9.224971293705158e-05} +{"train_loss": 0.016137588769197464, "grad_norm": 0.06738777458667755, "global_step": 9519, "epoch": 89, "lr": 9.224812791340222e-05} +{"train_loss": 0.017716538161039352, "grad_norm": 0.07652521878480911, "global_step": 9520, "epoch": 89, "lr": 9.224654274131132e-05} +{"train_loss": 0.016321107745170593, "grad_norm": 0.06199830025434494, "global_step": 9521, "epoch": 89, "lr": 9.224495742078444e-05} +{"train_loss": 0.017763745039701462, "grad_norm": 0.1261514127254486, "global_step": 9522, "epoch": 89, "lr": 9.224337195182714e-05} +{"train_loss": 0.015932662412524223, "grad_norm": 0.04515761882066727, "global_step": 9523, "epoch": 89, "lr": 9.224178633444503e-05} +{"train_loss": 0.019747808575630188, "grad_norm": 0.08018855005502701, "global_step": 9524, "epoch": 89, "lr": 9.224020056864364e-05} +{"train_loss": 0.018717754632234573, "grad_norm": 0.06094047799706459, "global_step": 9525, "epoch": 89, "lr": 9.223861465442856e-05} +{"train_loss": 0.023079847916960716, "grad_norm": 0.23281806707382202, "global_step": 9526, "epoch": 89, "lr": 9.223702859180535e-05} +{"train_loss": 0.017949499189853668, "grad_norm": 0.06432795524597168, "global_step": 9527, "epoch": 89, "lr": 9.223544238077959e-05} +{"train_loss": 0.015542736276984215, "grad_norm": 0.07662186026573181, "global_step": 9528, "epoch": 89, "lr": 9.223385602135686e-05} +{"train_loss": 0.01721961610019207, "grad_norm": 0.06039796397089958, "global_step": 9529, "epoch": 89, "lr": 9.223226951354272e-05} +{"train_loss": 0.016337737441062927, "grad_norm": 0.07939150929450989, "global_step": 9530, "epoch": 89, "lr": 9.223068285734275e-05} +{"train_loss": 0.018054772168397903, "grad_norm": 0.0453776940703392, "global_step": 9531, "epoch": 89, "lr": 9.222909605276252e-05} +{"train_loss": 0.01686546392738819, "grad_norm": 0.061015982180833817, "global_step": 9532, "epoch": 89, "lr": 9.222750909980762e-05} +{"train_loss": 0.015751654282212257, "grad_norm": 0.05768553167581558, "global_step": 9533, "epoch": 89, "lr": 9.222592199848362e-05} +{"train_loss": 0.01963641867041588, "grad_norm": 0.06514197587966919, "global_step": 9534, "epoch": 89, "lr": 9.222433474879608e-05} +{"train_loss": 0.018465518951416016, "grad_norm": 0.07648419588804245, "global_step": 9535, "epoch": 89, "lr": 9.22227473507506e-05} +{"train_loss": 0.01682191528379917, "grad_norm": 0.07582557201385498, "global_step": 9536, "epoch": 89, "lr": 9.222115980435276e-05} +{"train_loss": 0.017980538308620453, "grad_norm": 0.06846288591623306, "global_step": 9537, "epoch": 89, "lr": 9.221957210960812e-05} +{"train_loss": 0.01762479916214943, "grad_norm": 0.0774640440940857, "global_step": 9538, "epoch": 89, "lr": 9.221798426652226e-05} +{"train_loss": 0.017849675335763197, "grad_norm": 0.12915246188640594, "global_step": 9539, "epoch": 89, "lr": 9.221639627510076e-05, "val_loss": 0.018986649811267853} +{"train_loss": 0.016699710860848427, "grad_norm": 0.06654652208089828, "global_step": 9540, "epoch": 90, "lr": 9.22148081353492e-05} +{"train_loss": 0.01647503301501274, "grad_norm": 0.06441191583871841, "global_step": 9541, "epoch": 90, "lr": 9.221321984727318e-05} +{"train_loss": 0.017894797027111053, "grad_norm": 0.1195061132311821, "global_step": 9542, "epoch": 90, "lr": 9.221163141087826e-05} +{"train_loss": 0.017180465161800385, "grad_norm": 0.1580473631620407, "global_step": 9543, "epoch": 90, "lr": 9.221004282617e-05} +{"train_loss": 0.017596159130334854, "grad_norm": 0.14728306233882904, "global_step": 9544, "epoch": 90, "lr": 9.220845409315402e-05} +{"train_loss": 0.018075857311487198, "grad_norm": 0.15242736041545868, "global_step": 9545, "epoch": 90, "lr": 9.22068652118359e-05} +{"train_loss": 0.015673307701945305, "grad_norm": 0.050275400280952454, "global_step": 9546, "epoch": 90, "lr": 9.22052761822212e-05} +{"train_loss": 0.015798814594745636, "grad_norm": 0.06648392975330353, "global_step": 9547, "epoch": 90, "lr": 9.220368700431552e-05} +{"train_loss": 0.017917633056640625, "grad_norm": 0.0469994880259037, "global_step": 9548, "epoch": 90, "lr": 9.220209767812443e-05} +{"train_loss": 0.0185154527425766, "grad_norm": 0.11489663273096085, "global_step": 9549, "epoch": 90, "lr": 9.220050820365354e-05} +{"train_loss": 0.018987219780683517, "grad_norm": 0.08855597674846649, "global_step": 9550, "epoch": 90, "lr": 9.219891858090839e-05} +{"train_loss": 0.021892424672842026, "grad_norm": 0.05107127130031586, "global_step": 9551, "epoch": 90, "lr": 9.219732880989461e-05} +{"train_loss": 0.019116666167974472, "grad_norm": 0.07537931203842163, "global_step": 9552, "epoch": 90, "lr": 9.219573889061777e-05} +{"train_loss": 0.01663018763065338, "grad_norm": 0.11703074723482132, "global_step": 9553, "epoch": 90, "lr": 9.219414882308343e-05} +{"train_loss": 0.01895039528608322, "grad_norm": 0.09607644379138947, "global_step": 9554, "epoch": 90, "lr": 9.219255860729723e-05} +{"train_loss": 0.017478320747613907, "grad_norm": 0.10513336956501007, "global_step": 9555, "epoch": 90, "lr": 9.219096824326471e-05} +{"train_loss": 0.015949096530675888, "grad_norm": 0.06363203376531601, "global_step": 9556, "epoch": 90, "lr": 9.218937773099147e-05} +{"train_loss": 0.01635032147169113, "grad_norm": 0.05244498699903488, "global_step": 9557, "epoch": 90, "lr": 9.218778707048312e-05} +{"train_loss": 0.01711096614599228, "grad_norm": 0.07214025408029556, "global_step": 9558, "epoch": 90, "lr": 9.218619626174523e-05} +{"train_loss": 0.018368728458881378, "grad_norm": 0.05139409750699997, "global_step": 9559, "epoch": 90, "lr": 9.218460530478338e-05} +{"train_loss": 0.017854414880275726, "grad_norm": 0.11993050575256348, "global_step": 9560, "epoch": 90, "lr": 9.21830141996032e-05} +{"train_loss": 0.01899034157395363, "grad_norm": 0.1620674580335617, "global_step": 9561, "epoch": 90, "lr": 9.218142294621024e-05} +{"train_loss": 0.020357098430395126, "grad_norm": 0.044612716883420944, "global_step": 9562, "epoch": 90, "lr": 9.217983154461009e-05} +{"train_loss": 0.01678755320608616, "grad_norm": 0.19575057923793793, "global_step": 9563, "epoch": 90, "lr": 9.217823999480838e-05} +{"train_loss": 0.01620439812541008, "grad_norm": 0.08295080810785294, "global_step": 9564, "epoch": 90, "lr": 9.217664829681066e-05} +{"train_loss": 0.01799117401242256, "grad_norm": 0.06585119664669037, "global_step": 9565, "epoch": 90, "lr": 9.217505645062254e-05} +{"train_loss": 0.018373727798461914, "grad_norm": 0.11380504816770554, "global_step": 9566, "epoch": 90, "lr": 9.217346445624961e-05} +{"train_loss": 0.014767615124583244, "grad_norm": 0.05760614201426506, "global_step": 9567, "epoch": 90, "lr": 9.217187231369748e-05} +{"train_loss": 0.017968056723475456, "grad_norm": 0.10718904435634613, "global_step": 9568, "epoch": 90, "lr": 9.217028002297174e-05} +{"train_loss": 0.016039103269577026, "grad_norm": 0.06828049570322037, "global_step": 9569, "epoch": 90, "lr": 9.216868758407795e-05} +{"train_loss": 0.017590787261724472, "grad_norm": 0.056274496018886566, "global_step": 9570, "epoch": 90, "lr": 9.216709499702174e-05} +{"train_loss": 0.019093748182058334, "grad_norm": 0.09709617495536804, "global_step": 9571, "epoch": 90, "lr": 9.21655022618087e-05} +{"train_loss": 0.016253706067800522, "grad_norm": 0.07577347755432129, "global_step": 9572, "epoch": 90, "lr": 9.216390937844442e-05} +{"train_loss": 0.017053037881851196, "grad_norm": 0.07876647263765335, "global_step": 9573, "epoch": 90, "lr": 9.216231634693449e-05} +{"train_loss": 0.01796715334057808, "grad_norm": 0.07249890267848969, "global_step": 9574, "epoch": 90, "lr": 9.216072316728452e-05} +{"train_loss": 0.01644490659236908, "grad_norm": 0.07434037327766418, "global_step": 9575, "epoch": 90, "lr": 9.215912983950012e-05} +{"train_loss": 0.019337045028805733, "grad_norm": 0.10655441880226135, "global_step": 9576, "epoch": 90, "lr": 9.215753636358686e-05} +{"train_loss": 0.017808787524700165, "grad_norm": 0.1302560716867447, "global_step": 9577, "epoch": 90, "lr": 9.215594273955035e-05} +{"train_loss": 0.017403846606612206, "grad_norm": 0.14306768774986267, "global_step": 9578, "epoch": 90, "lr": 9.215434896739618e-05} +{"train_loss": 0.016019612550735474, "grad_norm": 0.06083207204937935, "global_step": 9579, "epoch": 90, "lr": 9.215275504712998e-05} +{"train_loss": 0.017087899148464203, "grad_norm": 0.061293087899684906, "global_step": 9580, "epoch": 90, "lr": 9.215116097875733e-05} +{"train_loss": 0.016066186130046844, "grad_norm": 0.04730597138404846, "global_step": 9581, "epoch": 90, "lr": 9.214956676228382e-05} +{"train_loss": 0.01962169073522091, "grad_norm": 0.060420118272304535, "global_step": 9582, "epoch": 90, "lr": 9.214797239771507e-05} +{"train_loss": 0.019619464874267578, "grad_norm": 0.07040417939424515, "global_step": 9583, "epoch": 90, "lr": 9.214637788505667e-05} +{"train_loss": 0.01899293251335621, "grad_norm": 0.11992143094539642, "global_step": 9584, "epoch": 90, "lr": 9.214478322431424e-05} +{"train_loss": 0.02139848656952381, "grad_norm": 0.06361696869134903, "global_step": 9585, "epoch": 90, "lr": 9.214318841549335e-05} +{"train_loss": 0.017487499862909317, "grad_norm": 0.04825780540704727, "global_step": 9586, "epoch": 90, "lr": 9.214159345859963e-05} +{"train_loss": 0.021542895585298538, "grad_norm": 0.13680262863636017, "global_step": 9587, "epoch": 90, "lr": 9.213999835363868e-05} +{"train_loss": 0.021762680262327194, "grad_norm": 0.061383966356515884, "global_step": 9588, "epoch": 90, "lr": 9.21384031006161e-05} +{"train_loss": 0.017519231885671616, "grad_norm": 0.13024301826953888, "global_step": 9589, "epoch": 90, "lr": 9.21368076995375e-05} +{"train_loss": 0.018199775367975235, "grad_norm": 0.1614648550748825, "global_step": 9590, "epoch": 90, "lr": 9.213521215040849e-05} +{"train_loss": 0.01944121904671192, "grad_norm": 0.1348261833190918, "global_step": 9591, "epoch": 90, "lr": 9.213361645323464e-05} +{"train_loss": 0.018084684386849403, "grad_norm": 0.05930202081799507, "global_step": 9592, "epoch": 90, "lr": 9.213202060802161e-05} +{"train_loss": 0.01615871675312519, "grad_norm": 0.10433008521795273, "global_step": 9593, "epoch": 90, "lr": 9.2130424614775e-05} +{"train_loss": 0.01464775763452053, "grad_norm": 0.052722178399562836, "global_step": 9594, "epoch": 90, "lr": 9.212882847350036e-05} +{"train_loss": 0.021099213510751724, "grad_norm": 0.05783095210790634, "global_step": 9595, "epoch": 90, "lr": 9.212723218420336e-05} +{"train_loss": 0.016517821699380875, "grad_norm": 0.12192682921886444, "global_step": 9596, "epoch": 90, "lr": 9.212563574688958e-05} +{"train_loss": 0.01639365404844284, "grad_norm": 0.050023093819618225, "global_step": 9597, "epoch": 90, "lr": 9.212403916156464e-05} +{"train_loss": 0.0174776129424572, "grad_norm": 0.05857686698436737, "global_step": 9598, "epoch": 90, "lr": 9.212244242823414e-05} +{"train_loss": 0.01915169507265091, "grad_norm": 0.11077795177698135, "global_step": 9599, "epoch": 90, "lr": 9.21208455469037e-05} +{"train_loss": 0.014883461408317089, "grad_norm": 0.048773013055324554, "global_step": 9600, "epoch": 90, "lr": 9.211924851757893e-05} +{"train_loss": 0.017594899982213974, "grad_norm": 0.08180567622184753, "global_step": 9601, "epoch": 90, "lr": 9.211765134026543e-05} +{"train_loss": 0.015648145228624344, "grad_norm": 0.08224097639322281, "global_step": 9602, "epoch": 90, "lr": 9.211605401496882e-05} +{"train_loss": 0.016554515808820724, "grad_norm": 0.0516960583627224, "global_step": 9603, "epoch": 90, "lr": 9.211445654169473e-05} +{"train_loss": 0.018227769061923027, "grad_norm": 0.05644512176513672, "global_step": 9604, "epoch": 90, "lr": 9.211285892044872e-05} +{"train_loss": 0.0166911818087101, "grad_norm": 0.08492550998926163, "global_step": 9605, "epoch": 90, "lr": 9.211126115123647e-05} +{"train_loss": 0.015786217525601387, "grad_norm": 0.04677562788128853, "global_step": 9606, "epoch": 90, "lr": 9.210966323406354e-05} +{"train_loss": 0.01711699739098549, "grad_norm": 0.05231361836194992, "global_step": 9607, "epoch": 90, "lr": 9.210806516893559e-05} +{"train_loss": 0.018895171582698822, "grad_norm": 0.10852131992578506, "global_step": 9608, "epoch": 90, "lr": 9.210646695585819e-05} +{"train_loss": 0.02147248573601246, "grad_norm": 0.13272543251514435, "global_step": 9609, "epoch": 90, "lr": 9.210486859483699e-05} +{"train_loss": 0.01641884632408619, "grad_norm": 0.1757696568965912, "global_step": 9610, "epoch": 90, "lr": 9.21032700858776e-05} +{"train_loss": 0.018358860164880753, "grad_norm": 0.11618796736001968, "global_step": 9611, "epoch": 90, "lr": 9.21016714289856e-05} +{"train_loss": 0.016661889851093292, "grad_norm": 0.1159469410777092, "global_step": 9612, "epoch": 90, "lr": 9.210007262416665e-05} +{"train_loss": 0.020899934694170952, "grad_norm": 0.08469784259796143, "global_step": 9613, "epoch": 90, "lr": 9.209847367142637e-05} +{"train_loss": 0.01943918503820896, "grad_norm": 0.10397832840681076, "global_step": 9614, "epoch": 90, "lr": 9.209687457077034e-05} +{"train_loss": 0.020109467208385468, "grad_norm": 0.10421034693717957, "global_step": 9615, "epoch": 90, "lr": 9.209527532220421e-05} +{"train_loss": 0.016926860436797142, "grad_norm": 0.09831415861845016, "global_step": 9616, "epoch": 90, "lr": 9.20936759257336e-05} +{"train_loss": 0.01922393962740898, "grad_norm": 0.07556232810020447, "global_step": 9617, "epoch": 90, "lr": 9.20920763813641e-05} +{"train_loss": 0.01905202865600586, "grad_norm": 0.05972709879279137, "global_step": 9618, "epoch": 90, "lr": 9.209047668910135e-05} +{"train_loss": 0.020375654101371765, "grad_norm": 0.058236073702573776, "global_step": 9619, "epoch": 90, "lr": 9.208887684895098e-05} +{"train_loss": 0.01821741834282875, "grad_norm": 0.09081673622131348, "global_step": 9620, "epoch": 90, "lr": 9.208727686091859e-05} +{"train_loss": 0.017027033492922783, "grad_norm": 0.20155513286590576, "global_step": 9621, "epoch": 90, "lr": 9.208567672500983e-05} +{"train_loss": 0.017576437443494797, "grad_norm": 0.1444227546453476, "global_step": 9622, "epoch": 90, "lr": 9.20840764412303e-05} +{"train_loss": 0.019669579342007637, "grad_norm": 0.14931444823741913, "global_step": 9623, "epoch": 90, "lr": 9.20824760095856e-05} +{"train_loss": 0.017625991255044937, "grad_norm": 0.14377467334270477, "global_step": 9624, "epoch": 90, "lr": 9.208087543008141e-05} +{"train_loss": 0.017474129796028137, "grad_norm": 0.11374187469482422, "global_step": 9625, "epoch": 90, "lr": 9.207927470272331e-05} +{"train_loss": 0.017245635390281677, "grad_norm": 0.17611216008663177, "global_step": 9626, "epoch": 90, "lr": 9.207767382751696e-05} +{"train_loss": 0.016705377027392387, "grad_norm": 0.14339713752269745, "global_step": 9627, "epoch": 90, "lr": 9.207607280446793e-05} +{"train_loss": 0.018535921350121498, "grad_norm": 0.1093861311674118, "global_step": 9628, "epoch": 90, "lr": 9.207447163358189e-05} +{"train_loss": 0.015486597083508968, "grad_norm": 0.09938408434391022, "global_step": 9629, "epoch": 90, "lr": 9.207287031486447e-05} +{"train_loss": 0.016079189255833626, "grad_norm": 0.07151564210653305, "global_step": 9630, "epoch": 90, "lr": 9.207126884832127e-05} +{"train_loss": 0.016383491456508636, "grad_norm": 0.09362892806529999, "global_step": 9631, "epoch": 90, "lr": 9.206966723395792e-05} +{"train_loss": 0.017218612134456635, "grad_norm": 0.18628819286823273, "global_step": 9632, "epoch": 90, "lr": 9.206806547178006e-05} +{"train_loss": 0.017458731308579445, "grad_norm": 0.16146616637706757, "global_step": 9633, "epoch": 90, "lr": 9.206646356179333e-05} +{"train_loss": 0.01885239966213703, "grad_norm": 0.07281554490327835, "global_step": 9634, "epoch": 90, "lr": 9.206486150400331e-05} +{"train_loss": 0.017325539141893387, "grad_norm": 0.11462782323360443, "global_step": 9635, "epoch": 90, "lr": 9.206325929841568e-05} +{"train_loss": 0.02024916186928749, "grad_norm": 0.20332133769989014, "global_step": 9636, "epoch": 90, "lr": 9.206165694503606e-05} +{"train_loss": 0.01751318946480751, "grad_norm": 0.06358078122138977, "global_step": 9637, "epoch": 90, "lr": 9.206005444387005e-05} +{"train_loss": 0.01976637914776802, "grad_norm": 0.12440119683742523, "global_step": 9638, "epoch": 90, "lr": 9.205845179492332e-05} +{"train_loss": 0.018874065950512886, "grad_norm": 0.06209966167807579, "global_step": 9639, "epoch": 90, "lr": 9.205684899820147e-05} +{"train_loss": 0.014995833858847618, "grad_norm": 0.21057355403900146, "global_step": 9640, "epoch": 90, "lr": 9.205524605371012e-05} +{"train_loss": 0.019108664244413376, "grad_norm": 0.06679760664701462, "global_step": 9641, "epoch": 90, "lr": 9.205364296145496e-05} +{"train_loss": 0.018062196671962738, "grad_norm": 0.04760820046067238, "global_step": 9642, "epoch": 90, "lr": 9.205203972144157e-05} +{"train_loss": 0.019306525588035583, "grad_norm": 0.045976318418979645, "global_step": 9643, "epoch": 90, "lr": 9.20504363336756e-05} +{"train_loss": 0.018506014719605446, "grad_norm": 0.07633460313081741, "global_step": 9644, "epoch": 90, "lr": 9.204883279816268e-05} +{"train_loss": 0.017897451068011095, "grad_norm": 0.07992402464151382, "global_step": 9645, "epoch": 90, "lr": 9.204722911490846e-05, "train_batch_mse_error": 0.016099367290735245, "val_loss": 0.018686579540371895} +{"train_loss": 0.018032288178801537, "grad_norm": 0.08961748331785202, "global_step": 9646, "epoch": 91, "lr": 9.204562528391856e-05} +{"train_loss": 0.01871884986758232, "grad_norm": 0.11046301573514938, "global_step": 9647, "epoch": 91, "lr": 9.204402130519861e-05} +{"train_loss": 0.01791444420814514, "grad_norm": 0.06863290071487427, "global_step": 9648, "epoch": 91, "lr": 9.204241717875425e-05} +{"train_loss": 0.01509188860654831, "grad_norm": 0.09283041954040527, "global_step": 9649, "epoch": 91, "lr": 9.204081290459114e-05} +{"train_loss": 0.018909886479377747, "grad_norm": 0.07985486835241318, "global_step": 9650, "epoch": 91, "lr": 9.203920848271487e-05} +{"train_loss": 0.020615871995687485, "grad_norm": 0.08541101962327957, "global_step": 9651, "epoch": 91, "lr": 9.203760391313115e-05} +{"train_loss": 0.018886849284172058, "grad_norm": 0.0442093126475811, "global_step": 9652, "epoch": 91, "lr": 9.203599919584553e-05} +{"train_loss": 0.015849895775318146, "grad_norm": 0.08264101296663284, "global_step": 9653, "epoch": 91, "lr": 9.203439433086369e-05} +{"train_loss": 0.01846475712954998, "grad_norm": 0.0652797669172287, "global_step": 9654, "epoch": 91, "lr": 9.203278931819128e-05} +{"train_loss": 0.01698608696460724, "grad_norm": 0.1311766356229782, "global_step": 9655, "epoch": 91, "lr": 9.203118415783393e-05} +{"train_loss": 0.016458837315440178, "grad_norm": 0.053331539034843445, "global_step": 9656, "epoch": 91, "lr": 9.202957884979729e-05} +{"train_loss": 0.017163148149847984, "grad_norm": 0.07129842787981033, "global_step": 9657, "epoch": 91, "lr": 9.202797339408695e-05} +{"train_loss": 0.015942834317684174, "grad_norm": 0.052607420831918716, "global_step": 9658, "epoch": 91, "lr": 9.202636779070862e-05} +{"train_loss": 0.016344621777534485, "grad_norm": 0.08543950319290161, "global_step": 9659, "epoch": 91, "lr": 9.20247620396679e-05} +{"train_loss": 0.01858890801668167, "grad_norm": 0.09221627563238144, "global_step": 9660, "epoch": 91, "lr": 9.202315614097044e-05} +{"train_loss": 0.01678234525024891, "grad_norm": 0.059013813734054565, "global_step": 9661, "epoch": 91, "lr": 9.20215500946219e-05} +{"train_loss": 0.018820034340023994, "grad_norm": 0.07078401744365692, "global_step": 9662, "epoch": 91, "lr": 9.20199439006279e-05} +{"train_loss": 0.018075184896588326, "grad_norm": 0.0796588733792305, "global_step": 9663, "epoch": 91, "lr": 9.201833755899409e-05} +{"train_loss": 0.018339337781071663, "grad_norm": 0.07188886404037476, "global_step": 9664, "epoch": 91, "lr": 9.20167310697261e-05} +{"train_loss": 0.019947722554206848, "grad_norm": 0.11465584486722946, "global_step": 9665, "epoch": 91, "lr": 9.20151244328296e-05} +{"train_loss": 0.01823665387928486, "grad_norm": 0.1034863144159317, "global_step": 9666, "epoch": 91, "lr": 9.201351764831023e-05} +{"train_loss": 0.018735511228442192, "grad_norm": 0.04641951620578766, "global_step": 9667, "epoch": 91, "lr": 9.201191071617363e-05} +{"train_loss": 0.017955230548977852, "grad_norm": 0.11695259809494019, "global_step": 9668, "epoch": 91, "lr": 9.201030363642545e-05} +{"train_loss": 0.01888304203748703, "grad_norm": 0.0665353536605835, "global_step": 9669, "epoch": 91, "lr": 9.200869640907132e-05} +{"train_loss": 0.017407391220331192, "grad_norm": 0.06543231755495071, "global_step": 9670, "epoch": 91, "lr": 9.200708903411691e-05} +{"train_loss": 0.018616124987602234, "grad_norm": 0.07272300124168396, "global_step": 9671, "epoch": 91, "lr": 9.200548151156785e-05} +{"train_loss": 0.01802624762058258, "grad_norm": 0.08352360874414444, "global_step": 9672, "epoch": 91, "lr": 9.20038738414298e-05} +{"train_loss": 0.017838023602962494, "grad_norm": 0.10531650483608246, "global_step": 9673, "epoch": 91, "lr": 9.200226602370841e-05} +{"train_loss": 0.019953984767198563, "grad_norm": 0.04884329438209534, "global_step": 9674, "epoch": 91, "lr": 9.200065805840932e-05} +{"train_loss": 0.01627151295542717, "grad_norm": 0.08909840136766434, "global_step": 9675, "epoch": 91, "lr": 9.199904994553817e-05} +{"train_loss": 0.01889215037226677, "grad_norm": 0.04995082691311836, "global_step": 9676, "epoch": 91, "lr": 9.199744168510064e-05} +{"train_loss": 0.01820549927651882, "grad_norm": 0.057378362864255905, "global_step": 9677, "epoch": 91, "lr": 9.199583327710235e-05} +{"train_loss": 0.018257567659020424, "grad_norm": 0.07786033302545547, "global_step": 9678, "epoch": 91, "lr": 9.199422472154899e-05} +{"train_loss": 0.019528783857822418, "grad_norm": 0.06833871454000473, "global_step": 9679, "epoch": 91, "lr": 9.199261601844617e-05} +{"train_loss": 0.0175882987678051, "grad_norm": 0.05905622988939285, "global_step": 9680, "epoch": 91, "lr": 9.199100716779956e-05} +{"train_loss": 0.017622828483581543, "grad_norm": 0.15219682455062866, "global_step": 9681, "epoch": 91, "lr": 9.198939816961481e-05} +{"train_loss": 0.017909128218889236, "grad_norm": 0.05640765652060509, "global_step": 9682, "epoch": 91, "lr": 9.198778902389758e-05} +{"train_loss": 0.020010875537991524, "grad_norm": 0.06444679945707321, "global_step": 9683, "epoch": 91, "lr": 9.19861797306535e-05} +{"train_loss": 0.01634419895708561, "grad_norm": 0.0687180608510971, "global_step": 9684, "epoch": 91, "lr": 9.198457028988828e-05} +{"train_loss": 0.01607249490916729, "grad_norm": 0.1268635392189026, "global_step": 9685, "epoch": 91, "lr": 9.198296070160751e-05} +{"train_loss": 0.016481202095746994, "grad_norm": 0.07561290264129639, "global_step": 9686, "epoch": 91, "lr": 9.198135096581689e-05} +{"train_loss": 0.018816858530044556, "grad_norm": 0.08654481172561646, "global_step": 9687, "epoch": 91, "lr": 9.197974108252206e-05} +{"train_loss": 0.018990837037563324, "grad_norm": 0.1220996305346489, "global_step": 9688, "epoch": 91, "lr": 9.197813105172865e-05} +{"train_loss": 0.01935518905520439, "grad_norm": 0.06900274753570557, "global_step": 9689, "epoch": 91, "lr": 9.197652087344237e-05} +{"train_loss": 0.02039080113172531, "grad_norm": 0.07738278061151505, "global_step": 9690, "epoch": 91, "lr": 9.197491054766885e-05} +{"train_loss": 0.016918526962399483, "grad_norm": 0.05360494554042816, "global_step": 9691, "epoch": 91, "lr": 9.197330007441373e-05} +{"train_loss": 0.01663358509540558, "grad_norm": 0.06321178376674652, "global_step": 9692, "epoch": 91, "lr": 9.197168945368271e-05} +{"train_loss": 0.020003922283649445, "grad_norm": 0.17339426279067993, "global_step": 9693, "epoch": 91, "lr": 9.19700786854814e-05} +{"train_loss": 0.014674974605441093, "grad_norm": 0.11878180503845215, "global_step": 9694, "epoch": 91, "lr": 9.196846776981551e-05} +{"train_loss": 0.016959408298134804, "grad_norm": 0.053860295563936234, "global_step": 9695, "epoch": 91, "lr": 9.196685670669068e-05} +{"train_loss": 0.01617821492254734, "grad_norm": 0.04934642091393471, "global_step": 9696, "epoch": 91, "lr": 9.196524549611254e-05} +{"train_loss": 0.019686467945575714, "grad_norm": 0.062233347445726395, "global_step": 9697, "epoch": 91, "lr": 9.19636341380868e-05} +{"train_loss": 0.017833244055509567, "grad_norm": 0.050479963421821594, "global_step": 9698, "epoch": 91, "lr": 9.196202263261908e-05} +{"train_loss": 0.01910492032766342, "grad_norm": 0.1458243876695633, "global_step": 9699, "epoch": 91, "lr": 9.196041097971508e-05} +{"train_loss": 0.016621991991996765, "grad_norm": 0.11038097739219666, "global_step": 9700, "epoch": 91, "lr": 9.195879917938044e-05} +{"train_loss": 0.017485059797763824, "grad_norm": 0.06091420352458954, "global_step": 9701, "epoch": 91, "lr": 9.195718723162082e-05} +{"train_loss": 0.018134338781237602, "grad_norm": 0.04701041430234909, "global_step": 9702, "epoch": 91, "lr": 9.19555751364419e-05} +{"train_loss": 0.018994655460119247, "grad_norm": 0.09048214554786682, "global_step": 9703, "epoch": 91, "lr": 9.195396289384932e-05} +{"train_loss": 0.01771293394267559, "grad_norm": 0.0724455788731575, "global_step": 9704, "epoch": 91, "lr": 9.195235050384877e-05} +{"train_loss": 0.01776830479502678, "grad_norm": 0.04646114259958267, "global_step": 9705, "epoch": 91, "lr": 9.195073796644591e-05} +{"train_loss": 0.019836727529764175, "grad_norm": 0.04383396357297897, "global_step": 9706, "epoch": 91, "lr": 9.194912528164638e-05} +{"train_loss": 0.020258087664842606, "grad_norm": 0.045216768980026245, "global_step": 9707, "epoch": 91, "lr": 9.19475124494559e-05} +{"train_loss": 0.016280675306916237, "grad_norm": 0.08071039617061615, "global_step": 9708, "epoch": 91, "lr": 9.194589946988008e-05} +{"train_loss": 0.017735769972205162, "grad_norm": 0.057433389127254486, "global_step": 9709, "epoch": 91, "lr": 9.194428634292462e-05} +{"train_loss": 0.019509481266140938, "grad_norm": 0.05316011980175972, "global_step": 9710, "epoch": 91, "lr": 9.194267306859517e-05} +{"train_loss": 0.017445366829633713, "grad_norm": 0.058499351143836975, "global_step": 9711, "epoch": 91, "lr": 9.194105964689741e-05} +{"train_loss": 0.014367060735821724, "grad_norm": 0.06336335837841034, "global_step": 9712, "epoch": 91, "lr": 9.193944607783701e-05} +{"train_loss": 0.015343077480793, "grad_norm": 0.10089439153671265, "global_step": 9713, "epoch": 91, "lr": 9.193783236141964e-05} +{"train_loss": 0.016288986429572105, "grad_norm": 0.059739936143159866, "global_step": 9714, "epoch": 91, "lr": 9.193621849765094e-05} +{"train_loss": 0.016911789774894714, "grad_norm": 0.09240320324897766, "global_step": 9715, "epoch": 91, "lr": 9.193460448653664e-05} +{"train_loss": 0.019158368930220604, "grad_norm": 0.08280748128890991, "global_step": 9716, "epoch": 91, "lr": 9.193299032808236e-05} +{"train_loss": 0.016945727169513702, "grad_norm": 0.04771508648991585, "global_step": 9717, "epoch": 91, "lr": 9.193137602229378e-05} +{"train_loss": 0.020504675805568695, "grad_norm": 0.0687665119767189, "global_step": 9718, "epoch": 91, "lr": 9.19297615691766e-05} +{"train_loss": 0.021250007674098015, "grad_norm": 0.13418062031269073, "global_step": 9719, "epoch": 91, "lr": 9.192814696873646e-05} +{"train_loss": 0.015850186347961426, "grad_norm": 0.0396990031003952, "global_step": 9720, "epoch": 91, "lr": 9.192653222097904e-05} +{"train_loss": 0.018254362046718597, "grad_norm": 0.07416372001171112, "global_step": 9721, "epoch": 91, "lr": 9.192491732591003e-05} +{"train_loss": 0.016635868698358536, "grad_norm": 0.14167582988739014, "global_step": 9722, "epoch": 91, "lr": 9.19233022835351e-05} +{"train_loss": 0.016559891402721405, "grad_norm": 0.10792873799800873, "global_step": 9723, "epoch": 91, "lr": 9.19216870938599e-05} +{"train_loss": 0.01698753982782364, "grad_norm": 0.12967157363891602, "global_step": 9724, "epoch": 91, "lr": 9.192007175689013e-05} +{"train_loss": 0.01827356033027172, "grad_norm": 0.08627928793430328, "global_step": 9725, "epoch": 91, "lr": 9.191845627263148e-05} +{"train_loss": 0.018361900001764297, "grad_norm": 0.08655442297458649, "global_step": 9726, "epoch": 91, "lr": 9.191684064108957e-05} +{"train_loss": 0.016100715845823288, "grad_norm": 0.07099609076976776, "global_step": 9727, "epoch": 91, "lr": 9.191522486227012e-05} +{"train_loss": 0.017107482999563217, "grad_norm": 0.0696740373969078, "global_step": 9728, "epoch": 91, "lr": 9.19136089361788e-05} +{"train_loss": 0.016535364091396332, "grad_norm": 0.07381895929574966, "global_step": 9729, "epoch": 91, "lr": 9.191199286282129e-05} +{"train_loss": 0.015004107728600502, "grad_norm": 0.06482493132352829, "global_step": 9730, "epoch": 91, "lr": 9.191037664220327e-05} +{"train_loss": 0.016461893916130066, "grad_norm": 0.0741148367524147, "global_step": 9731, "epoch": 91, "lr": 9.190876027433039e-05} +{"train_loss": 0.017762748524546623, "grad_norm": 0.0523185133934021, "global_step": 9732, "epoch": 91, "lr": 9.190714375920838e-05} +{"train_loss": 0.01756705343723297, "grad_norm": 0.07591472566127777, "global_step": 9733, "epoch": 91, "lr": 9.190552709684288e-05} +{"train_loss": 0.01905255764722824, "grad_norm": 0.04993452504277229, "global_step": 9734, "epoch": 91, "lr": 9.190391028723958e-05} +{"train_loss": 0.016002636402845383, "grad_norm": 0.07935141772031784, "global_step": 9735, "epoch": 91, "lr": 9.190229333040417e-05} +{"train_loss": 0.01366572454571724, "grad_norm": 0.09810863435268402, "global_step": 9736, "epoch": 91, "lr": 9.190067622634232e-05} +{"train_loss": 0.018563248217105865, "grad_norm": 0.06228132173418999, "global_step": 9737, "epoch": 91, "lr": 9.189905897505971e-05} +{"train_loss": 0.016120530664920807, "grad_norm": 0.06556487828493118, "global_step": 9738, "epoch": 91, "lr": 9.189744157656204e-05} +{"train_loss": 0.01746746152639389, "grad_norm": 0.08380859345197678, "global_step": 9739, "epoch": 91, "lr": 9.189582403085497e-05} +{"train_loss": 0.016348781064152718, "grad_norm": 0.07582894712686539, "global_step": 9740, "epoch": 91, "lr": 9.189420633794419e-05} +{"train_loss": 0.01740402728319168, "grad_norm": 0.06513561308383942, "global_step": 9741, "epoch": 91, "lr": 9.189258849783539e-05} +{"train_loss": 0.01573081500828266, "grad_norm": 0.10927223414182663, "global_step": 9742, "epoch": 91, "lr": 9.189097051053429e-05} +{"train_loss": 0.019467554986476898, "grad_norm": 0.04592002555727959, "global_step": 9743, "epoch": 91, "lr": 9.188935237604649e-05} +{"train_loss": 0.021056465804576874, "grad_norm": 0.0592905730009079, "global_step": 9744, "epoch": 91, "lr": 9.188773409437774e-05} +{"train_loss": 0.018168512731790543, "grad_norm": 0.051283080130815506, "global_step": 9745, "epoch": 91, "lr": 9.188611566553372e-05} +{"train_loss": 0.017007239162921906, "grad_norm": 0.07447878271341324, "global_step": 9746, "epoch": 91, "lr": 9.188449708952008e-05} +{"train_loss": 0.02131982147693634, "grad_norm": 0.049305349588394165, "global_step": 9747, "epoch": 91, "lr": 9.188287836634255e-05} +{"train_loss": 0.01783822849392891, "grad_norm": 0.061904426664114, "global_step": 9748, "epoch": 91, "lr": 9.188125949600679e-05} +{"train_loss": 0.02120238170027733, "grad_norm": 0.10559193789958954, "global_step": 9749, "epoch": 91, "lr": 9.187964047851852e-05} +{"train_loss": 0.016544342041015625, "grad_norm": 0.05563506856560707, "global_step": 9750, "epoch": 91, "lr": 9.187802131388336e-05} +{"train_loss": 0.01778353378176689, "grad_norm": 0.08278346061706543, "global_step": 9751, "epoch": 91, "lr": 9.18764020021071e-05, "val_loss": 0.018898190930485725} +{"train_loss": 0.019635271281003952, "grad_norm": 0.09432917088270187, "global_step": 9752, "epoch": 92, "lr": 9.187478254319535e-05} +{"train_loss": 0.017071131616830826, "grad_norm": 0.05395805835723877, "global_step": 9753, "epoch": 92, "lr": 9.187316293715382e-05} +{"train_loss": 0.016904860734939575, "grad_norm": 0.06909641623497009, "global_step": 9754, "epoch": 92, "lr": 9.18715431839882e-05} +{"train_loss": 0.016636310145258904, "grad_norm": 0.04557010903954506, "global_step": 9755, "epoch": 92, "lr": 9.186992328370421e-05} +{"train_loss": 0.019900130107998848, "grad_norm": 0.06718821078538895, "global_step": 9756, "epoch": 92, "lr": 9.18683032363075e-05} +{"train_loss": 0.01791498437523842, "grad_norm": 0.06665357947349548, "global_step": 9757, "epoch": 92, "lr": 9.186668304180379e-05} +{"train_loss": 0.016651831567287445, "grad_norm": 0.05772148445248604, "global_step": 9758, "epoch": 92, "lr": 9.186506270019875e-05} +{"train_loss": 0.018267003819346428, "grad_norm": 0.07301051169633865, "global_step": 9759, "epoch": 92, "lr": 9.18634422114981e-05} +{"train_loss": 0.016337275505065918, "grad_norm": 0.05125829577445984, "global_step": 9760, "epoch": 92, "lr": 9.186182157570752e-05} +{"train_loss": 0.015545446425676346, "grad_norm": 0.06383050978183746, "global_step": 9761, "epoch": 92, "lr": 9.186020079283271e-05} +{"train_loss": 0.018308589234948158, "grad_norm": 0.0602034255862236, "global_step": 9762, "epoch": 92, "lr": 9.185857986287934e-05} +{"train_loss": 0.02255602926015854, "grad_norm": 0.04359809309244156, "global_step": 9763, "epoch": 92, "lr": 9.185695878585313e-05} +{"train_loss": 0.016155656427145004, "grad_norm": 0.054355476051568985, "global_step": 9764, "epoch": 92, "lr": 9.185533756175977e-05} +{"train_loss": 0.016272835433483124, "grad_norm": 0.053365662693977356, "global_step": 9765, "epoch": 92, "lr": 9.185371619060497e-05} +{"train_loss": 0.016385339200496674, "grad_norm": 0.048621077090501785, "global_step": 9766, "epoch": 92, "lr": 9.18520946723944e-05} +{"train_loss": 0.01654699072241783, "grad_norm": 0.05891402065753937, "global_step": 9767, "epoch": 92, "lr": 9.185047300713377e-05} +{"train_loss": 0.015754086896777153, "grad_norm": 0.05621432885527611, "global_step": 9768, "epoch": 92, "lr": 9.184885119482878e-05} +{"train_loss": 0.017889108508825302, "grad_norm": 0.05231890454888344, "global_step": 9769, "epoch": 92, "lr": 9.184722923548512e-05} +{"train_loss": 0.017159290611743927, "grad_norm": 0.13837991654872894, "global_step": 9770, "epoch": 92, "lr": 9.184560712910852e-05} +{"train_loss": 0.017864227294921875, "grad_norm": 0.1469186395406723, "global_step": 9771, "epoch": 92, "lr": 9.184398487570463e-05} +{"train_loss": 0.017304230481386185, "grad_norm": 0.10554980486631393, "global_step": 9772, "epoch": 92, "lr": 9.184236247527918e-05} +{"train_loss": 0.018142983317375183, "grad_norm": 0.07084483653306961, "global_step": 9773, "epoch": 92, "lr": 9.184073992783786e-05} +{"train_loss": 0.01595081388950348, "grad_norm": 0.09076600521802902, "global_step": 9774, "epoch": 92, "lr": 9.183911723338638e-05} +{"train_loss": 0.017683640122413635, "grad_norm": 0.05269210413098335, "global_step": 9775, "epoch": 92, "lr": 9.183749439193044e-05} +{"train_loss": 0.019639071077108383, "grad_norm": 0.07148416340351105, "global_step": 9776, "epoch": 92, "lr": 9.183587140347575e-05} +{"train_loss": 0.01920422911643982, "grad_norm": 0.13650673627853394, "global_step": 9777, "epoch": 92, "lr": 9.183424826802799e-05} +{"train_loss": 0.016845054924488068, "grad_norm": 0.06815541535615921, "global_step": 9778, "epoch": 92, "lr": 9.183262498559287e-05} +{"train_loss": 0.019391607493162155, "grad_norm": 0.08774466812610626, "global_step": 9779, "epoch": 92, "lr": 9.183100155617611e-05} +{"train_loss": 0.021099230274558067, "grad_norm": 0.08214139938354492, "global_step": 9780, "epoch": 92, "lr": 9.18293779797834e-05} +{"train_loss": 0.017664693295955658, "grad_norm": 0.12328159809112549, "global_step": 9781, "epoch": 92, "lr": 9.182775425642044e-05} +{"train_loss": 0.016099918633699417, "grad_norm": 0.05795199051499367, "global_step": 9782, "epoch": 92, "lr": 9.182613038609295e-05} +{"train_loss": 0.015739355236291885, "grad_norm": 0.07427198439836502, "global_step": 9783, "epoch": 92, "lr": 9.182450636880662e-05} +{"train_loss": 0.01872992143034935, "grad_norm": 0.10380787402391434, "global_step": 9784, "epoch": 92, "lr": 9.182288220456716e-05} +{"train_loss": 0.017855416983366013, "grad_norm": 0.07951518893241882, "global_step": 9785, "epoch": 92, "lr": 9.18212578933803e-05} +{"train_loss": 0.016833899542689323, "grad_norm": 0.07222683727741241, "global_step": 9786, "epoch": 92, "lr": 9.18196334352517e-05} +{"train_loss": 0.01866208389401436, "grad_norm": 0.06880486011505127, "global_step": 9787, "epoch": 92, "lr": 9.181800883018712e-05} +{"train_loss": 0.018318939954042435, "grad_norm": 0.09281592816114426, "global_step": 9788, "epoch": 92, "lr": 9.181638407819223e-05} +{"train_loss": 0.018140826374292374, "grad_norm": 0.08671535551548004, "global_step": 9789, "epoch": 92, "lr": 9.181475917927276e-05} +{"train_loss": 0.020213736221194267, "grad_norm": 0.11845574527978897, "global_step": 9790, "epoch": 92, "lr": 9.181313413343439e-05} +{"train_loss": 0.017642103135585785, "grad_norm": 0.08578725904226303, "global_step": 9791, "epoch": 92, "lr": 9.181150894068288e-05} +{"train_loss": 0.016426436603069305, "grad_norm": 0.07057863473892212, "global_step": 9792, "epoch": 92, "lr": 9.180988360102389e-05} +{"train_loss": 0.019097600132226944, "grad_norm": 0.14365196228027344, "global_step": 9793, "epoch": 92, "lr": 9.180825811446317e-05} +{"train_loss": 0.017987310886383057, "grad_norm": 0.05365706980228424, "global_step": 9794, "epoch": 92, "lr": 9.180663248100639e-05} +{"train_loss": 0.017501043155789375, "grad_norm": 0.10368750244379044, "global_step": 9795, "epoch": 92, "lr": 9.180500670065931e-05} +{"train_loss": 0.0178348608314991, "grad_norm": 0.06362385302782059, "global_step": 9796, "epoch": 92, "lr": 9.180338077342759e-05} +{"train_loss": 0.020207762718200684, "grad_norm": 0.21742725372314453, "global_step": 9797, "epoch": 92, "lr": 9.1801754699317e-05} +{"train_loss": 0.016753893345594406, "grad_norm": 0.049994755536317825, "global_step": 9798, "epoch": 92, "lr": 9.180012847833319e-05} +{"train_loss": 0.02253042533993721, "grad_norm": 0.15965566039085388, "global_step": 9799, "epoch": 92, "lr": 9.179850211048193e-05} +{"train_loss": 0.018564879894256592, "grad_norm": 0.062343012541532516, "global_step": 9800, "epoch": 92, "lr": 9.17968755957689e-05} +{"train_loss": 0.01755344495177269, "grad_norm": 0.06893038004636765, "global_step": 9801, "epoch": 92, "lr": 9.179524893419984e-05} +{"train_loss": 0.019013818353414536, "grad_norm": 0.11936374753713608, "global_step": 9802, "epoch": 92, "lr": 9.179362212578044e-05} +{"train_loss": 0.016832876950502396, "grad_norm": 0.08448473364114761, "global_step": 9803, "epoch": 92, "lr": 9.179199517051642e-05} +{"train_loss": 0.020580671727657318, "grad_norm": 0.15535767376422882, "global_step": 9804, "epoch": 92, "lr": 9.179036806841353e-05} +{"train_loss": 0.017867401242256165, "grad_norm": 0.08933762460947037, "global_step": 9805, "epoch": 92, "lr": 9.178874081947744e-05} +{"train_loss": 0.01782011054456234, "grad_norm": 0.12887383997440338, "global_step": 9806, "epoch": 92, "lr": 9.178711342371388e-05} +{"train_loss": 0.016791775822639465, "grad_norm": 0.08673086017370224, "global_step": 9807, "epoch": 92, "lr": 9.178548588112859e-05} +{"train_loss": 0.01772325485944748, "grad_norm": 0.08539385348558426, "global_step": 9808, "epoch": 92, "lr": 9.178385819172728e-05} +{"train_loss": 0.016376061365008354, "grad_norm": 0.09085961431264877, "global_step": 9809, "epoch": 92, "lr": 9.178223035551565e-05} +{"train_loss": 0.015340391546487808, "grad_norm": 0.04760522022843361, "global_step": 9810, "epoch": 92, "lr": 9.178060237249943e-05} +{"train_loss": 0.01591123640537262, "grad_norm": 0.0776820182800293, "global_step": 9811, "epoch": 92, "lr": 9.177897424268437e-05} +{"train_loss": 0.021866105496883392, "grad_norm": 0.06278736144304276, "global_step": 9812, "epoch": 92, "lr": 9.177734596607614e-05} +{"train_loss": 0.0186908058822155, "grad_norm": 0.06374970823526382, "global_step": 9813, "epoch": 92, "lr": 9.17757175426805e-05} +{"train_loss": 0.016550760716199875, "grad_norm": 0.1474260687828064, "global_step": 9814, "epoch": 92, "lr": 9.177408897250314e-05} +{"train_loss": 0.01742985099554062, "grad_norm": 0.0678546205163002, "global_step": 9815, "epoch": 92, "lr": 9.17724602555498e-05} +{"train_loss": 0.016477832570672035, "grad_norm": 0.06196768581867218, "global_step": 9816, "epoch": 92, "lr": 9.177083139182622e-05} +{"train_loss": 0.01728716678917408, "grad_norm": 0.07528214156627655, "global_step": 9817, "epoch": 92, "lr": 9.176920238133808e-05} +{"train_loss": 0.020730840042233467, "grad_norm": 0.060146041214466095, "global_step": 9818, "epoch": 92, "lr": 9.176757322409115e-05} +{"train_loss": 0.01966654509305954, "grad_norm": 0.12026304006576538, "global_step": 9819, "epoch": 92, "lr": 9.176594392009112e-05} +{"train_loss": 0.017532404512166977, "grad_norm": 0.09480942040681839, "global_step": 9820, "epoch": 92, "lr": 9.176431446934372e-05} +{"train_loss": 0.01682925596833229, "grad_norm": 0.12453075498342514, "global_step": 9821, "epoch": 92, "lr": 9.176268487185468e-05} +{"train_loss": 0.018420575186610222, "grad_norm": 0.10184250771999359, "global_step": 9822, "epoch": 92, "lr": 9.176105512762976e-05} +{"train_loss": 0.01633896864950657, "grad_norm": 0.056124214082956314, "global_step": 9823, "epoch": 92, "lr": 9.175942523667462e-05} +{"train_loss": 0.017606960609555244, "grad_norm": 0.05772218108177185, "global_step": 9824, "epoch": 92, "lr": 9.175779519899504e-05} +{"train_loss": 0.018136873841285706, "grad_norm": 0.0552234873175621, "global_step": 9825, "epoch": 92, "lr": 9.175616501459672e-05} +{"train_loss": 0.01762464828789234, "grad_norm": 0.05335621163249016, "global_step": 9826, "epoch": 92, "lr": 9.175453468348538e-05} +{"train_loss": 0.017424147576093674, "grad_norm": 0.04108595475554466, "global_step": 9827, "epoch": 92, "lr": 9.175290420566677e-05} +{"train_loss": 0.0168495811522007, "grad_norm": 0.07311125099658966, "global_step": 9828, "epoch": 92, "lr": 9.175127358114664e-05} +{"train_loss": 0.01648668944835663, "grad_norm": 0.04087017476558685, "global_step": 9829, "epoch": 92, "lr": 9.174964280993067e-05} +{"train_loss": 0.016119584441184998, "grad_norm": 0.06178329885005951, "global_step": 9830, "epoch": 92, "lr": 9.174801189202461e-05} +{"train_loss": 0.01594209484755993, "grad_norm": 0.08088980615139008, "global_step": 9831, "epoch": 92, "lr": 9.174638082743419e-05} +{"train_loss": 0.016356205567717552, "grad_norm": 0.08852898329496384, "global_step": 9832, "epoch": 92, "lr": 9.174474961616514e-05} +{"train_loss": 0.016266625374555588, "grad_norm": 0.07117144018411636, "global_step": 9833, "epoch": 92, "lr": 9.174311825822322e-05} +{"train_loss": 0.017443345859646797, "grad_norm": 0.129214346408844, "global_step": 9834, "epoch": 92, "lr": 9.174148675361411e-05} +{"train_loss": 0.017446326091885567, "grad_norm": 0.07536362111568451, "global_step": 9835, "epoch": 92, "lr": 9.173985510234357e-05} +{"train_loss": 0.016863947734236717, "grad_norm": 0.08081585168838501, "global_step": 9836, "epoch": 92, "lr": 9.173822330441735e-05} +{"train_loss": 0.017348088324069977, "grad_norm": 0.044197145849466324, "global_step": 9837, "epoch": 92, "lr": 9.173659135984115e-05} +{"train_loss": 0.017586352303624153, "grad_norm": 0.07976953685283661, "global_step": 9838, "epoch": 92, "lr": 9.173495926862073e-05} +{"train_loss": 0.017292305827140808, "grad_norm": 0.06845840811729431, "global_step": 9839, "epoch": 92, "lr": 9.17333270307618e-05} +{"train_loss": 0.017316535115242004, "grad_norm": 0.04584995657205582, "global_step": 9840, "epoch": 92, "lr": 9.173169464627012e-05} +{"train_loss": 0.020283253863453865, "grad_norm": 0.05482669174671173, "global_step": 9841, "epoch": 92, "lr": 9.17300621151514e-05} +{"train_loss": 0.01791244000196457, "grad_norm": 0.054376695305109024, "global_step": 9842, "epoch": 92, "lr": 9.17284294374114e-05} +{"train_loss": 0.02095939591526985, "grad_norm": 0.1338575929403305, "global_step": 9843, "epoch": 92, "lr": 9.172679661305583e-05} +{"train_loss": 0.0164379570633173, "grad_norm": 0.07763571292161942, "global_step": 9844, "epoch": 92, "lr": 9.172516364209047e-05} +{"train_loss": 0.015452548861503601, "grad_norm": 0.06147979199886322, "global_step": 9845, "epoch": 92, "lr": 9.172353052452102e-05} +{"train_loss": 0.019094988703727722, "grad_norm": 0.07883942127227783, "global_step": 9846, "epoch": 92, "lr": 9.172189726035321e-05} +{"train_loss": 0.01780511438846588, "grad_norm": 0.12807875871658325, "global_step": 9847, "epoch": 92, "lr": 9.172026384959282e-05} +{"train_loss": 0.017211901023983955, "grad_norm": 0.04237915202975273, "global_step": 9848, "epoch": 92, "lr": 9.171863029224557e-05} +{"train_loss": 0.017049919813871384, "grad_norm": 0.06839632242918015, "global_step": 9849, "epoch": 92, "lr": 9.171699658831719e-05} +{"train_loss": 0.019929731264710426, "grad_norm": 0.16107119619846344, "global_step": 9850, "epoch": 92, "lr": 9.17153627378134e-05} +{"train_loss": 0.019029028713703156, "grad_norm": 0.06353098899126053, "global_step": 9851, "epoch": 92, "lr": 9.171372874074e-05} +{"train_loss": 0.01909317448735237, "grad_norm": 0.059540752321481705, "global_step": 9852, "epoch": 92, "lr": 9.171209459710269e-05} +{"train_loss": 0.019988754764199257, "grad_norm": 0.060270581394433975, "global_step": 9853, "epoch": 92, "lr": 9.17104603069072e-05} +{"train_loss": 0.01546977087855339, "grad_norm": 0.08521458506584167, "global_step": 9854, "epoch": 92, "lr": 9.170882587015933e-05} +{"train_loss": 0.018560832366347313, "grad_norm": 0.08108000457286835, "global_step": 9855, "epoch": 92, "lr": 9.170719128686476e-05} +{"train_loss": 0.016917286440730095, "grad_norm": 0.052075836807489395, "global_step": 9856, "epoch": 92, "lr": 9.170555655702925e-05} +{"train_loss": 0.01776278751708989, "grad_norm": 0.08368268609046936, "global_step": 9857, "epoch": 92, "lr": 9.170392168065857e-05, "val_loss": 0.019564636051654816} +{"train_loss": 0.016534004360437393, "grad_norm": 0.058314427733421326, "global_step": 9858, "epoch": 93, "lr": 9.170228665775844e-05} +{"train_loss": 0.018629174679517746, "grad_norm": 0.08849070221185684, "global_step": 9859, "epoch": 93, "lr": 9.17006514883346e-05} +{"train_loss": 0.018281184136867523, "grad_norm": 0.13222959637641907, "global_step": 9860, "epoch": 93, "lr": 9.169901617239282e-05} +{"train_loss": 0.018539080396294594, "grad_norm": 0.05125083029270172, "global_step": 9861, "epoch": 93, "lr": 9.169738070993882e-05} +{"train_loss": 0.01623164303600788, "grad_norm": 0.06682313978672028, "global_step": 9862, "epoch": 93, "lr": 9.169574510097836e-05} +{"train_loss": 0.01723688468337059, "grad_norm": 0.05939226970076561, "global_step": 9863, "epoch": 93, "lr": 9.169410934551719e-05} +{"train_loss": 0.013915931805968285, "grad_norm": 0.04318061098456383, "global_step": 9864, "epoch": 93, "lr": 9.169247344356105e-05} +{"train_loss": 0.019865259528160095, "grad_norm": 0.08103527128696442, "global_step": 9865, "epoch": 93, "lr": 9.169083739511569e-05} +{"train_loss": 0.01821107044816017, "grad_norm": 0.09520404785871506, "global_step": 9866, "epoch": 93, "lr": 9.168920120018686e-05} +{"train_loss": 0.01993010938167572, "grad_norm": 0.06537047028541565, "global_step": 9867, "epoch": 93, "lr": 9.16875648587803e-05} +{"train_loss": 0.013964381068944931, "grad_norm": 0.08936107158660889, "global_step": 9868, "epoch": 93, "lr": 9.168592837090177e-05} +{"train_loss": 0.016263006255030632, "grad_norm": 0.07843027263879776, "global_step": 9869, "epoch": 93, "lr": 9.168429173655704e-05} +{"train_loss": 0.017755012959241867, "grad_norm": 0.07998359203338623, "global_step": 9870, "epoch": 93, "lr": 9.168265495575182e-05} +{"train_loss": 0.018905749544501305, "grad_norm": 0.09339083731174469, "global_step": 9871, "epoch": 93, "lr": 9.168101802849184e-05} +{"train_loss": 0.017191343009471893, "grad_norm": 0.047249916940927505, "global_step": 9872, "epoch": 93, "lr": 9.167938095478295e-05} +{"train_loss": 0.017535880208015442, "grad_norm": 0.09357518702745438, "global_step": 9873, "epoch": 93, "lr": 9.16777437346308e-05} +{"train_loss": 0.017373789101839066, "grad_norm": 0.08798600733280182, "global_step": 9874, "epoch": 93, "lr": 9.16761063680412e-05} +{"train_loss": 0.01778201013803482, "grad_norm": 0.11030714958906174, "global_step": 9875, "epoch": 93, "lr": 9.167446885501989e-05} +{"train_loss": 0.015982262790203094, "grad_norm": 0.050799090415239334, "global_step": 9876, "epoch": 93, "lr": 9.167283119557262e-05} +{"train_loss": 0.016284653916954994, "grad_norm": 0.07707604020833969, "global_step": 9877, "epoch": 93, "lr": 9.167119338970514e-05} +{"train_loss": 0.015298384241759777, "grad_norm": 0.07550760358572006, "global_step": 9878, "epoch": 93, "lr": 9.16695554374232e-05} +{"train_loss": 0.01794644072651863, "grad_norm": 0.11041657626628876, "global_step": 9879, "epoch": 93, "lr": 9.166791733873257e-05} +{"train_loss": 0.019800426438450813, "grad_norm": 0.13596241176128387, "global_step": 9880, "epoch": 93, "lr": 9.1666279093639e-05} +{"train_loss": 0.014942115172743797, "grad_norm": 0.06710407137870789, "global_step": 9881, "epoch": 93, "lr": 9.166464070214825e-05} +{"train_loss": 0.015077410265803337, "grad_norm": 0.05675192177295685, "global_step": 9882, "epoch": 93, "lr": 9.166300216426606e-05} +{"train_loss": 0.017588423565030098, "grad_norm": 0.09727033972740173, "global_step": 9883, "epoch": 93, "lr": 9.16613634799982e-05} +{"train_loss": 0.01747921295464039, "grad_norm": 0.06296200305223465, "global_step": 9884, "epoch": 93, "lr": 9.165972464935044e-05} +{"train_loss": 0.01987055316567421, "grad_norm": 0.052697066217660904, "global_step": 9885, "epoch": 93, "lr": 9.165808567232852e-05} +{"train_loss": 0.017646383494138718, "grad_norm": 0.08498018980026245, "global_step": 9886, "epoch": 93, "lr": 9.165644654893821e-05} +{"train_loss": 0.01913321577012539, "grad_norm": 0.055805906653404236, "global_step": 9887, "epoch": 93, "lr": 9.165480727918525e-05} +{"train_loss": 0.019354935735464096, "grad_norm": 0.05174165964126587, "global_step": 9888, "epoch": 93, "lr": 9.165316786307542e-05} +{"train_loss": 0.016375282779335976, "grad_norm": 0.09507237374782562, "global_step": 9889, "epoch": 93, "lr": 9.165152830061447e-05} +{"train_loss": 0.0154716270044446, "grad_norm": 0.07093993574380875, "global_step": 9890, "epoch": 93, "lr": 9.164988859180817e-05} +{"train_loss": 0.015139320865273476, "grad_norm": 0.07108710706233978, "global_step": 9891, "epoch": 93, "lr": 9.164824873666226e-05} +{"train_loss": 0.01724240928888321, "grad_norm": 0.07451973110437393, "global_step": 9892, "epoch": 93, "lr": 9.164660873518252e-05} +{"train_loss": 0.01684107631444931, "grad_norm": 0.07992769032716751, "global_step": 9893, "epoch": 93, "lr": 9.164496858737473e-05} +{"train_loss": 0.018732281401753426, "grad_norm": 0.10967928171157837, "global_step": 9894, "epoch": 93, "lr": 9.16433282932446e-05} +{"train_loss": 0.017367977648973465, "grad_norm": 0.06475446373224258, "global_step": 9895, "epoch": 93, "lr": 9.164168785279794e-05} +{"train_loss": 0.01733076199889183, "grad_norm": 0.18919415771961212, "global_step": 9896, "epoch": 93, "lr": 9.16400472660405e-05} +{"train_loss": 0.018377210944890976, "grad_norm": 0.09363040328025818, "global_step": 9897, "epoch": 93, "lr": 9.163840653297804e-05} +{"train_loss": 0.020715657621622086, "grad_norm": 0.053126536309719086, "global_step": 9898, "epoch": 93, "lr": 9.163676565361632e-05} +{"train_loss": 0.016947345808148384, "grad_norm": 0.04978087171912193, "global_step": 9899, "epoch": 93, "lr": 9.163512462796114e-05} +{"train_loss": 0.018532801419496536, "grad_norm": 0.05251079797744751, "global_step": 9900, "epoch": 93, "lr": 9.163348345601822e-05} +{"train_loss": 0.020239146426320076, "grad_norm": 0.0835236981511116, "global_step": 9901, "epoch": 93, "lr": 9.163184213779333e-05} +{"train_loss": 0.019450902938842773, "grad_norm": 0.11616659164428711, "global_step": 9902, "epoch": 93, "lr": 9.163020067329226e-05} +{"train_loss": 0.01668551191687584, "grad_norm": 0.05620073527097702, "global_step": 9903, "epoch": 93, "lr": 9.162855906252078e-05} +{"train_loss": 0.02123882994055748, "grad_norm": 0.1668289303779602, "global_step": 9904, "epoch": 93, "lr": 9.162691730548464e-05} +{"train_loss": 0.016463827341794968, "grad_norm": 0.08437662571668625, "global_step": 9905, "epoch": 93, "lr": 9.162527540218961e-05} +{"train_loss": 0.018865082412958145, "grad_norm": 0.06086498126387596, "global_step": 9906, "epoch": 93, "lr": 9.162363335264148e-05} +{"train_loss": 0.016837023198604584, "grad_norm": 0.14861147105693817, "global_step": 9907, "epoch": 93, "lr": 9.162199115684599e-05} +{"train_loss": 0.0184878371655941, "grad_norm": 0.1408734768629074, "global_step": 9908, "epoch": 93, "lr": 9.162034881480894e-05} +{"train_loss": 0.019314125180244446, "grad_norm": 0.06517545878887177, "global_step": 9909, "epoch": 93, "lr": 9.161870632653608e-05} +{"train_loss": 0.01831674948334694, "grad_norm": 0.07060806453227997, "global_step": 9910, "epoch": 93, "lr": 9.161706369203317e-05} +{"train_loss": 0.01745205745100975, "grad_norm": 0.09124607592821121, "global_step": 9911, "epoch": 93, "lr": 9.161542091130602e-05} +{"train_loss": 0.01726287044584751, "grad_norm": 0.1113726943731308, "global_step": 9912, "epoch": 93, "lr": 9.161377798436036e-05} +{"train_loss": 0.016974814236164093, "grad_norm": 0.058849725872278214, "global_step": 9913, "epoch": 93, "lr": 9.161213491120199e-05} +{"train_loss": 0.01741289347410202, "grad_norm": 0.045957885682582855, "global_step": 9914, "epoch": 93, "lr": 9.161049169183667e-05} +{"train_loss": 0.021379180252552032, "grad_norm": 0.10861147195100784, "global_step": 9915, "epoch": 93, "lr": 9.160884832627017e-05} +{"train_loss": 0.020397484302520752, "grad_norm": 0.13140590488910675, "global_step": 9916, "epoch": 93, "lr": 9.16072048145083e-05} +{"train_loss": 0.013832773081958294, "grad_norm": 0.059846725314855576, "global_step": 9917, "epoch": 93, "lr": 9.160556115655678e-05} +{"train_loss": 0.01573706977069378, "grad_norm": 0.11421240121126175, "global_step": 9918, "epoch": 93, "lr": 9.160391735242143e-05} +{"train_loss": 0.01814970374107361, "grad_norm": 0.09369461238384247, "global_step": 9919, "epoch": 93, "lr": 9.1602273402108e-05} +{"train_loss": 0.018751364201307297, "grad_norm": 0.11880843341350555, "global_step": 9920, "epoch": 93, "lr": 9.160062930562228e-05} +{"train_loss": 0.019758161157369614, "grad_norm": 0.06344752758741379, "global_step": 9921, "epoch": 93, "lr": 9.159898506297002e-05} +{"train_loss": 0.01730496995151043, "grad_norm": 0.06214016675949097, "global_step": 9922, "epoch": 93, "lr": 9.159734067415704e-05} +{"train_loss": 0.018296588212251663, "grad_norm": 0.053053051233291626, "global_step": 9923, "epoch": 93, "lr": 9.159569613918908e-05} +{"train_loss": 0.018345944583415985, "grad_norm": 0.05626988783478737, "global_step": 9924, "epoch": 93, "lr": 9.159405145807194e-05} +{"train_loss": 0.01880430430173874, "grad_norm": 0.10168619453907013, "global_step": 9925, "epoch": 93, "lr": 9.159240663081139e-05} +{"train_loss": 0.01727980561554432, "grad_norm": 0.09298672527074814, "global_step": 9926, "epoch": 93, "lr": 9.15907616574132e-05} +{"train_loss": 0.018441926687955856, "grad_norm": 0.14345216751098633, "global_step": 9927, "epoch": 93, "lr": 9.158911653788318e-05} +{"train_loss": 0.017766743898391724, "grad_norm": 0.08880174160003662, "global_step": 9928, "epoch": 93, "lr": 9.158747127222708e-05} +{"train_loss": 0.016323573887348175, "grad_norm": 0.10118992626667023, "global_step": 9929, "epoch": 93, "lr": 9.15858258604507e-05} +{"train_loss": 0.022229477763175964, "grad_norm": 0.07863199710845947, "global_step": 9930, "epoch": 93, "lr": 9.158418030255982e-05} +{"train_loss": 0.014956356957554817, "grad_norm": 0.06440876424312592, "global_step": 9931, "epoch": 93, "lr": 9.158253459856018e-05} +{"train_loss": 0.017780490219593048, "grad_norm": 0.055033352226018906, "global_step": 9932, "epoch": 93, "lr": 9.158088874845764e-05} +{"train_loss": 0.016704823821783066, "grad_norm": 0.05188773199915886, "global_step": 9933, "epoch": 93, "lr": 9.157924275225792e-05} +{"train_loss": 0.01835463009774685, "grad_norm": 0.1113305389881134, "global_step": 9934, "epoch": 93, "lr": 9.157759660996681e-05} +{"train_loss": 0.01776326820254326, "grad_norm": 0.11793965846300125, "global_step": 9935, "epoch": 93, "lr": 9.157595032159013e-05} +{"train_loss": 0.017926858738064766, "grad_norm": 0.06957928836345673, "global_step": 9936, "epoch": 93, "lr": 9.157430388713364e-05} +{"train_loss": 0.021214986220002174, "grad_norm": 0.10612467676401138, "global_step": 9937, "epoch": 93, "lr": 9.157265730660312e-05} +{"train_loss": 0.01604810543358326, "grad_norm": 0.068559929728508, "global_step": 9938, "epoch": 93, "lr": 9.157101058000436e-05} +{"train_loss": 0.015898488461971283, "grad_norm": 0.04363367334008217, "global_step": 9939, "epoch": 93, "lr": 9.156936370734315e-05} +{"train_loss": 0.019826222211122513, "grad_norm": 0.07711608707904816, "global_step": 9940, "epoch": 93, "lr": 9.156771668862528e-05} +{"train_loss": 0.02033679559826851, "grad_norm": 0.07224836945533752, "global_step": 9941, "epoch": 93, "lr": 9.156606952385652e-05} +{"train_loss": 0.019587470218539238, "grad_norm": 0.05155833065509796, "global_step": 9942, "epoch": 93, "lr": 9.156442221304266e-05} +{"train_loss": 0.01811330020427704, "grad_norm": 0.06158822402358055, "global_step": 9943, "epoch": 93, "lr": 9.15627747561895e-05} +{"train_loss": 0.015270231291651726, "grad_norm": 0.09285697340965271, "global_step": 9944, "epoch": 93, "lr": 9.156112715330285e-05} +{"train_loss": 0.018034208565950394, "grad_norm": 0.11651010811328888, "global_step": 9945, "epoch": 93, "lr": 9.155947940438844e-05} +{"train_loss": 0.017268039286136627, "grad_norm": 0.04231605678796768, "global_step": 9946, "epoch": 93, "lr": 9.155783150945212e-05} +{"train_loss": 0.015464316122233868, "grad_norm": 0.061124395579099655, "global_step": 9947, "epoch": 93, "lr": 9.155618346849963e-05} +{"train_loss": 0.015544707886874676, "grad_norm": 0.050691962242126465, "global_step": 9948, "epoch": 93, "lr": 9.155453528153681e-05} +{"train_loss": 0.016861388459801674, "grad_norm": 0.0713597983121872, "global_step": 9949, "epoch": 93, "lr": 9.15528869485694e-05} +{"train_loss": 0.01911923661828041, "grad_norm": 0.08070104569196701, "global_step": 9950, "epoch": 93, "lr": 9.155123846960323e-05} +{"train_loss": 0.01755884289741516, "grad_norm": 0.04891461506485939, "global_step": 9951, "epoch": 93, "lr": 9.154958984464408e-05} +{"train_loss": 0.01587948389351368, "grad_norm": 0.053972359746694565, "global_step": 9952, "epoch": 93, "lr": 9.154794107369772e-05} +{"train_loss": 0.019580084830522537, "grad_norm": 0.06605780124664307, "global_step": 9953, "epoch": 93, "lr": 9.154629215677e-05} +{"train_loss": 0.0165933258831501, "grad_norm": 0.04894734546542168, "global_step": 9954, "epoch": 93, "lr": 9.154464309386665e-05} +{"train_loss": 0.013369514606893063, "grad_norm": 0.04493781551718712, "global_step": 9955, "epoch": 93, "lr": 9.15429938849935e-05} +{"train_loss": 0.01831069029867649, "grad_norm": 0.11848880350589752, "global_step": 9956, "epoch": 93, "lr": 9.154134453015633e-05} +{"train_loss": 0.017406906932592392, "grad_norm": 0.13570696115493774, "global_step": 9957, "epoch": 93, "lr": 9.153969502936095e-05} +{"train_loss": 0.01928112655878067, "grad_norm": 0.04209602251648903, "global_step": 9958, "epoch": 93, "lr": 9.153804538261314e-05} +{"train_loss": 0.020049121230840683, "grad_norm": 0.09956859052181244, "global_step": 9959, "epoch": 93, "lr": 9.15363955899187e-05} +{"train_loss": 0.01793663576245308, "grad_norm": 0.06675881892442703, "global_step": 9960, "epoch": 93, "lr": 9.153474565128344e-05} +{"train_loss": 0.017677882686257362, "grad_norm": 0.04657939076423645, "global_step": 9961, "epoch": 93, "lr": 9.153309556671315e-05} +{"train_loss": 0.019414491951465607, "grad_norm": 0.16177532076835632, "global_step": 9962, "epoch": 93, "lr": 9.15314453362136e-05} +{"train_loss": 0.017756493621558515, "grad_norm": 0.06671275943517685, "global_step": 9963, "epoch": 93, "lr": 9.152979495979063e-05, "val_loss": 0.018821261823177338} +{"train_loss": 0.015991859138011932, "grad_norm": 0.061767518520355225, "global_step": 9964, "epoch": 94, "lr": 9.152814443745002e-05} +{"train_loss": 0.013727502897381783, "grad_norm": 0.12070954591035843, "global_step": 9965, "epoch": 94, "lr": 9.152649376919757e-05} +{"train_loss": 0.01597648486495018, "grad_norm": 0.0413682721555233, "global_step": 9966, "epoch": 94, "lr": 9.152484295503909e-05} +{"train_loss": 0.020568490028381348, "grad_norm": 0.0575716495513916, "global_step": 9967, "epoch": 94, "lr": 9.152319199498036e-05} +{"train_loss": 0.01754695549607277, "grad_norm": 0.056655626744031906, "global_step": 9968, "epoch": 94, "lr": 9.152154088902719e-05} +{"train_loss": 0.017702467739582062, "grad_norm": 0.09462641179561615, "global_step": 9969, "epoch": 94, "lr": 9.151988963718539e-05} +{"train_loss": 0.020689915865659714, "grad_norm": 0.10485765337944031, "global_step": 9970, "epoch": 94, "lr": 9.151823823946075e-05} +{"train_loss": 0.015456901863217354, "grad_norm": 0.038728728890419006, "global_step": 9971, "epoch": 94, "lr": 9.151658669585907e-05} +{"train_loss": 0.015153611078858376, "grad_norm": 0.08821621537208557, "global_step": 9972, "epoch": 94, "lr": 9.151493500638618e-05} +{"train_loss": 0.017726462334394455, "grad_norm": 0.07283933460712433, "global_step": 9973, "epoch": 94, "lr": 9.151328317104784e-05} +{"train_loss": 0.019788026809692383, "grad_norm": 0.1670100837945938, "global_step": 9974, "epoch": 94, "lr": 9.151163118984988e-05} +{"train_loss": 0.017602551728487015, "grad_norm": 0.050405193120241165, "global_step": 9975, "epoch": 94, "lr": 9.15099790627981e-05} +{"train_loss": 0.01826253905892372, "grad_norm": 0.057038869708776474, "global_step": 9976, "epoch": 94, "lr": 9.150832678989832e-05} +{"train_loss": 0.01613933965563774, "grad_norm": 0.1571602076292038, "global_step": 9977, "epoch": 94, "lr": 9.150667437115632e-05} +{"train_loss": 0.018893834203481674, "grad_norm": 0.04839230328798294, "global_step": 9978, "epoch": 94, "lr": 9.150502180657792e-05} +{"train_loss": 0.01610436849296093, "grad_norm": 0.054745495319366455, "global_step": 9979, "epoch": 94, "lr": 9.150336909616892e-05} +{"train_loss": 0.01838698796927929, "grad_norm": 0.07709644734859467, "global_step": 9980, "epoch": 94, "lr": 9.150171623993514e-05} +{"train_loss": 0.014654306694865227, "grad_norm": 0.09084603935480118, "global_step": 9981, "epoch": 94, "lr": 9.150006323788237e-05} +{"train_loss": 0.017713651061058044, "grad_norm": 0.09139899164438248, "global_step": 9982, "epoch": 94, "lr": 9.149841009001643e-05} +{"train_loss": 0.02151395007967949, "grad_norm": 0.04689152538776398, "global_step": 9983, "epoch": 94, "lr": 9.149675679634312e-05} +{"train_loss": 0.01676594838500023, "grad_norm": 0.051228757947683334, "global_step": 9984, "epoch": 94, "lr": 9.149510335686826e-05} +{"train_loss": 0.017952557653188705, "grad_norm": 0.0721639096736908, "global_step": 9985, "epoch": 94, "lr": 9.149344977159764e-05} +{"train_loss": 0.016540292650461197, "grad_norm": 0.10706688463687897, "global_step": 9986, "epoch": 94, "lr": 9.149179604053709e-05} +{"train_loss": 0.019562385976314545, "grad_norm": 0.039789557456970215, "global_step": 9987, "epoch": 94, "lr": 9.14901421636924e-05} +{"train_loss": 0.017479073256254196, "grad_norm": 0.1027354896068573, "global_step": 9988, "epoch": 94, "lr": 9.148848814106941e-05} +{"train_loss": 0.01837427169084549, "grad_norm": 0.05255645141005516, "global_step": 9989, "epoch": 94, "lr": 9.148683397267393e-05} +{"train_loss": 0.016489515081048012, "grad_norm": 0.061917345970869064, "global_step": 9990, "epoch": 94, "lr": 9.148517965851174e-05} +{"train_loss": 0.02052181400358677, "grad_norm": 0.08595620095729828, "global_step": 9991, "epoch": 94, "lr": 9.148352519858866e-05} +{"train_loss": 0.016497423872351646, "grad_norm": 0.049388110637664795, "global_step": 9992, "epoch": 94, "lr": 9.148187059291052e-05} +{"train_loss": 0.01621674932539463, "grad_norm": 0.08386491984128952, "global_step": 9993, "epoch": 94, "lr": 9.148021584148313e-05} +{"train_loss": 0.01960993930697441, "grad_norm": 0.05235196277499199, "global_step": 9994, "epoch": 94, "lr": 9.14785609443123e-05} +{"train_loss": 0.01782415434718132, "grad_norm": 0.068598173558712, "global_step": 9995, "epoch": 94, "lr": 9.147690590140385e-05} +{"train_loss": 0.02118760719895363, "grad_norm": 0.06049994006752968, "global_step": 9996, "epoch": 94, "lr": 9.147525071276358e-05} +{"train_loss": 0.021873392164707184, "grad_norm": 0.06337013095617294, "global_step": 9997, "epoch": 94, "lr": 9.147359537839732e-05} +{"train_loss": 0.018111534416675568, "grad_norm": 0.05555381998419762, "global_step": 9998, "epoch": 94, "lr": 9.147193989831089e-05} +{"train_loss": 0.019262343645095825, "grad_norm": 0.0611715130507946, "global_step": 9999, "epoch": 94, "lr": 9.14702842725101e-05} +{"train_loss": 0.018873251974582672, "grad_norm": 0.05690259486436844, "global_step": 10000, "epoch": 94, "lr": 9.146862850100075e-05} +{"train_loss": 0.018209408968687057, "grad_norm": 0.05941404402256012, "global_step": 10001, "epoch": 94, "lr": 9.146697258378869e-05} +{"train_loss": 0.017059415578842163, "grad_norm": 0.08699454367160797, "global_step": 10002, "epoch": 94, "lr": 9.14653165208797e-05} +{"train_loss": 0.021371468901634216, "grad_norm": 0.07996721565723419, "global_step": 10003, "epoch": 94, "lr": 9.146366031227964e-05} +{"train_loss": 0.01524817943572998, "grad_norm": 0.08489108085632324, "global_step": 10004, "epoch": 94, "lr": 9.146200395799432e-05} +{"train_loss": 0.015310205519199371, "grad_norm": 0.04085297882556915, "global_step": 10005, "epoch": 94, "lr": 9.146034745802953e-05} +{"train_loss": 0.020095961168408394, "grad_norm": 0.07361046969890594, "global_step": 10006, "epoch": 94, "lr": 9.145869081239111e-05} +{"train_loss": 0.018680624663829803, "grad_norm": 0.0732005313038826, "global_step": 10007, "epoch": 94, "lr": 9.145703402108489e-05} +{"train_loss": 0.01643206924200058, "grad_norm": 0.045632049441337585, "global_step": 10008, "epoch": 94, "lr": 9.145537708411667e-05} +{"train_loss": 0.014848649501800537, "grad_norm": 0.07136819511651993, "global_step": 10009, "epoch": 94, "lr": 9.14537200014923e-05} +{"train_loss": 0.015586494468152523, "grad_norm": 0.0897441953420639, "global_step": 10010, "epoch": 94, "lr": 9.145206277321756e-05} +{"train_loss": 0.017723456025123596, "grad_norm": 0.07150720804929733, "global_step": 10011, "epoch": 94, "lr": 9.145040539929832e-05} +{"train_loss": 0.016432736068964005, "grad_norm": 0.06092316284775734, "global_step": 10012, "epoch": 94, "lr": 9.144874787974038e-05} +{"train_loss": 0.019448943436145782, "grad_norm": 0.13682770729064941, "global_step": 10013, "epoch": 94, "lr": 9.144709021454956e-05} +{"train_loss": 0.016628984361886978, "grad_norm": 0.08154275268316269, "global_step": 10014, "epoch": 94, "lr": 9.144543240373169e-05} +{"train_loss": 0.01718861050903797, "grad_norm": 0.05986225605010986, "global_step": 10015, "epoch": 94, "lr": 9.14437744472926e-05} +{"train_loss": 0.015174336731433868, "grad_norm": 0.07512373477220535, "global_step": 10016, "epoch": 94, "lr": 9.14421163452381e-05} +{"train_loss": 0.019026588648557663, "grad_norm": 0.23342041671276093, "global_step": 10017, "epoch": 94, "lr": 9.144045809757404e-05} +{"train_loss": 0.0182931050658226, "grad_norm": 0.05181152746081352, "global_step": 10018, "epoch": 94, "lr": 9.143879970430622e-05} +{"train_loss": 0.018854010850191116, "grad_norm": 0.09097466617822647, "global_step": 10019, "epoch": 94, "lr": 9.143714116544047e-05} +{"train_loss": 0.017443455755710602, "grad_norm": 0.08514390140771866, "global_step": 10020, "epoch": 94, "lr": 9.143548248098265e-05} +{"train_loss": 0.01587972790002823, "grad_norm": 0.05778315290808678, "global_step": 10021, "epoch": 94, "lr": 9.143382365093855e-05} +{"train_loss": 0.01780269294977188, "grad_norm": 0.17956528067588806, "global_step": 10022, "epoch": 94, "lr": 9.143216467531401e-05} +{"train_loss": 0.019854113459587097, "grad_norm": 0.07376779615879059, "global_step": 10023, "epoch": 94, "lr": 9.143050555411487e-05} +{"train_loss": 0.01499126572161913, "grad_norm": 0.05947815626859665, "global_step": 10024, "epoch": 94, "lr": 9.142884628734696e-05} +{"train_loss": 0.01863032951951027, "grad_norm": 0.08022667467594147, "global_step": 10025, "epoch": 94, "lr": 9.142718687501608e-05} +{"train_loss": 0.01837637834250927, "grad_norm": 0.061910904943943024, "global_step": 10026, "epoch": 94, "lr": 9.14255273171281e-05} +{"train_loss": 0.01707925647497177, "grad_norm": 0.06068886071443558, "global_step": 10027, "epoch": 94, "lr": 9.142386761368881e-05} +{"train_loss": 0.019460897892713547, "grad_norm": 0.07153719663619995, "global_step": 10028, "epoch": 94, "lr": 9.142220776470409e-05} +{"train_loss": 0.01806693896651268, "grad_norm": 0.047503236681222916, "global_step": 10029, "epoch": 94, "lr": 9.142054777017975e-05} +{"train_loss": 0.017262104898691177, "grad_norm": 0.10895438492298126, "global_step": 10030, "epoch": 94, "lr": 9.141888763012158e-05} +{"train_loss": 0.016914699226617813, "grad_norm": 0.14925606548786163, "global_step": 10031, "epoch": 94, "lr": 9.141722734453549e-05} +{"train_loss": 0.01757689192891121, "grad_norm": 0.09594777971506119, "global_step": 10032, "epoch": 94, "lr": 9.141556691342726e-05} +{"train_loss": 0.017774662002921104, "grad_norm": 0.16993770003318787, "global_step": 10033, "epoch": 94, "lr": 9.141390633680274e-05} +{"train_loss": 0.020055431872606277, "grad_norm": 0.06935828924179077, "global_step": 10034, "epoch": 94, "lr": 9.141224561466775e-05} +{"train_loss": 0.017733410000801086, "grad_norm": 0.059260107576847076, "global_step": 10035, "epoch": 94, "lr": 9.141058474702815e-05} +{"train_loss": 0.015307062305510044, "grad_norm": 0.050622258335351944, "global_step": 10036, "epoch": 94, "lr": 9.140892373388976e-05} +{"train_loss": 0.019348256289958954, "grad_norm": 0.10098663717508316, "global_step": 10037, "epoch": 94, "lr": 9.140726257525843e-05} +{"train_loss": 0.016290239989757538, "grad_norm": 0.06159576401114464, "global_step": 10038, "epoch": 94, "lr": 9.140560127113998e-05} +{"train_loss": 0.02014973573386669, "grad_norm": 0.058019768446683884, "global_step": 10039, "epoch": 94, "lr": 9.140393982154026e-05} +{"train_loss": 0.018784265965223312, "grad_norm": 0.05503562465310097, "global_step": 10040, "epoch": 94, "lr": 9.14022782264651e-05} +{"train_loss": 0.018462015315890312, "grad_norm": 0.06644287705421448, "global_step": 10041, "epoch": 94, "lr": 9.140061648592034e-05} +{"train_loss": 0.01748623326420784, "grad_norm": 0.052303846925497055, "global_step": 10042, "epoch": 94, "lr": 9.139895459991181e-05} +{"train_loss": 0.017489373683929443, "grad_norm": 0.08044274151325226, "global_step": 10043, "epoch": 94, "lr": 9.139729256844536e-05} +{"train_loss": 0.018896374851465225, "grad_norm": 0.09817839413881302, "global_step": 10044, "epoch": 94, "lr": 9.139563039152684e-05} +{"train_loss": 0.014690464362502098, "grad_norm": 0.05222892016172409, "global_step": 10045, "epoch": 94, "lr": 9.139396806916205e-05} +{"train_loss": 0.01812562718987465, "grad_norm": 0.1006440594792366, "global_step": 10046, "epoch": 94, "lr": 9.139230560135688e-05} +{"train_loss": 0.01756448671221733, "grad_norm": 0.07293231040239334, "global_step": 10047, "epoch": 94, "lr": 9.139064298811715e-05} +{"train_loss": 0.017392218112945557, "grad_norm": 0.08147004246711731, "global_step": 10048, "epoch": 94, "lr": 9.138898022944869e-05} +{"train_loss": 0.01909848302602768, "grad_norm": 0.09258464723825455, "global_step": 10049, "epoch": 94, "lr": 9.138731732535735e-05} +{"train_loss": 0.017434390261769295, "grad_norm": 0.11161237210035324, "global_step": 10050, "epoch": 94, "lr": 9.1385654275849e-05} +{"train_loss": 0.016535552218556404, "grad_norm": 0.07601901888847351, "global_step": 10051, "epoch": 94, "lr": 9.138399108092943e-05} +{"train_loss": 0.015453667379915714, "grad_norm": 0.0903100073337555, "global_step": 10052, "epoch": 94, "lr": 9.138232774060454e-05} +{"train_loss": 0.018544338643550873, "grad_norm": 0.05361039936542511, "global_step": 10053, "epoch": 94, "lr": 9.138066425488013e-05} +{"train_loss": 0.01386137306690216, "grad_norm": 0.10066071152687073, "global_step": 10054, "epoch": 94, "lr": 9.137900062376206e-05} +{"train_loss": 0.016432058066129684, "grad_norm": 0.07255541533231735, "global_step": 10055, "epoch": 94, "lr": 9.137733684725617e-05} +{"train_loss": 0.02020554058253765, "grad_norm": 0.06653203070163727, "global_step": 10056, "epoch": 94, "lr": 9.137567292536833e-05} +{"train_loss": 0.019001992419362068, "grad_norm": 0.09463956952095032, "global_step": 10057, "epoch": 94, "lr": 9.137400885810436e-05} +{"train_loss": 0.015874428674578667, "grad_norm": 0.04188636690378189, "global_step": 10058, "epoch": 94, "lr": 9.137234464547013e-05} +{"train_loss": 0.020069321617484093, "grad_norm": 0.06529345363378525, "global_step": 10059, "epoch": 94, "lr": 9.137068028747147e-05} +{"train_loss": 0.017001140862703323, "grad_norm": 0.11041226238012314, "global_step": 10060, "epoch": 94, "lr": 9.136901578411421e-05} +{"train_loss": 0.015566382557153702, "grad_norm": 0.08858475089073181, "global_step": 10061, "epoch": 94, "lr": 9.136735113540425e-05} +{"train_loss": 0.017371805384755135, "grad_norm": 0.04174167662858963, "global_step": 10062, "epoch": 94, "lr": 9.136568634134738e-05} +{"train_loss": 0.019353419542312622, "grad_norm": 0.05081290751695633, "global_step": 10063, "epoch": 94, "lr": 9.136402140194948e-05} +{"train_loss": 0.01904922164976597, "grad_norm": 0.09062273055315018, "global_step": 10064, "epoch": 94, "lr": 9.136235631721641e-05} +{"train_loss": 0.016690297052264214, "grad_norm": 0.07964684814214706, "global_step": 10065, "epoch": 94, "lr": 9.136069108715402e-05} +{"train_loss": 0.017877530306577682, "grad_norm": 0.0865921825170517, "global_step": 10066, "epoch": 94, "lr": 9.135902571176813e-05} +{"train_loss": 0.018013106659054756, "grad_norm": 0.06676716357469559, "global_step": 10067, "epoch": 94, "lr": 9.135736019106461e-05} +{"train_loss": 0.016929101198911667, "grad_norm": 0.050799041986465454, "global_step": 10068, "epoch": 94, "lr": 9.135569452504931e-05} +{"train_loss": 0.017704619428599782, "grad_norm": 0.11770796775817871, "global_step": 10069, "epoch": 94, "lr": 9.135402871372808e-05, "val_loss": 0.019469980150461197} +{"train_loss": 0.016633832827210426, "grad_norm": 0.11021985113620758, "global_step": 10070, "epoch": 95, "lr": 9.135236275710681e-05} +{"train_loss": 0.018285341560840607, "grad_norm": 0.07394725829362869, "global_step": 10071, "epoch": 95, "lr": 9.13506966551913e-05} +{"train_loss": 0.017254192382097244, "grad_norm": 0.07172925770282745, "global_step": 10072, "epoch": 95, "lr": 9.134903040798742e-05} +{"train_loss": 0.016346145421266556, "grad_norm": 0.07288068532943726, "global_step": 10073, "epoch": 95, "lr": 9.134736401550104e-05} +{"train_loss": 0.016010411083698273, "grad_norm": 0.07534690946340561, "global_step": 10074, "epoch": 95, "lr": 9.134569747773797e-05} +{"train_loss": 0.01621304266154766, "grad_norm": 0.08176504820585251, "global_step": 10075, "epoch": 95, "lr": 9.134403079470415e-05} +{"train_loss": 0.017459305003285408, "grad_norm": 0.07823550701141357, "global_step": 10076, "epoch": 95, "lr": 9.134236396640536e-05} +{"train_loss": 0.016260290518403053, "grad_norm": 0.06497324258089066, "global_step": 10077, "epoch": 95, "lr": 9.134069699284749e-05} +{"train_loss": 0.01910063438117504, "grad_norm": 0.10827228426933289, "global_step": 10078, "epoch": 95, "lr": 9.133902987403638e-05} +{"train_loss": 0.017441250383853912, "grad_norm": 0.08103839308023453, "global_step": 10079, "epoch": 95, "lr": 9.133736260997789e-05} +{"train_loss": 0.01752687245607376, "grad_norm": 0.10421672463417053, "global_step": 10080, "epoch": 95, "lr": 9.13356952006779e-05} +{"train_loss": 0.016151465475559235, "grad_norm": 0.07778039574623108, "global_step": 10081, "epoch": 95, "lr": 9.133402764614226e-05} +{"train_loss": 0.01773514598608017, "grad_norm": 0.07770702242851257, "global_step": 10082, "epoch": 95, "lr": 9.133235994637681e-05} +{"train_loss": 0.015079612843692303, "grad_norm": 0.06475276499986649, "global_step": 10083, "epoch": 95, "lr": 9.133069210138743e-05} +{"train_loss": 0.015796802937984467, "grad_norm": 0.09829585254192352, "global_step": 10084, "epoch": 95, "lr": 9.132902411117996e-05} +{"train_loss": 0.01988856866955757, "grad_norm": 0.09385735541582108, "global_step": 10085, "epoch": 95, "lr": 9.132735597576028e-05} +{"train_loss": 0.01640057936310768, "grad_norm": 0.043455325067043304, "global_step": 10086, "epoch": 95, "lr": 9.132568769513425e-05} +{"train_loss": 0.015783710405230522, "grad_norm": 0.08726832270622253, "global_step": 10087, "epoch": 95, "lr": 9.132401926930772e-05} +{"train_loss": 0.018472060561180115, "grad_norm": 0.06268660724163055, "global_step": 10088, "epoch": 95, "lr": 9.132235069828655e-05} +{"train_loss": 0.01980578526854515, "grad_norm": 0.15158692002296448, "global_step": 10089, "epoch": 95, "lr": 9.132068198207662e-05} +{"train_loss": 0.019136184826493263, "grad_norm": 0.06641855090856552, "global_step": 10090, "epoch": 95, "lr": 9.131901312068378e-05} +{"train_loss": 0.01919456571340561, "grad_norm": 0.09089077264070511, "global_step": 10091, "epoch": 95, "lr": 9.13173441141139e-05} +{"train_loss": 0.02056625299155712, "grad_norm": 0.0689987763762474, "global_step": 10092, "epoch": 95, "lr": 9.131567496237283e-05} +{"train_loss": 0.017106443643569946, "grad_norm": 0.11356162279844284, "global_step": 10093, "epoch": 95, "lr": 9.131400566546647e-05} +{"train_loss": 0.01840949058532715, "grad_norm": 0.05469920486211777, "global_step": 10094, "epoch": 95, "lr": 9.131233622340063e-05} +{"train_loss": 0.018177654594182968, "grad_norm": 0.1060287207365036, "global_step": 10095, "epoch": 95, "lr": 9.131066663618123e-05} +{"train_loss": 0.019385792315006256, "grad_norm": 0.07510139793157578, "global_step": 10096, "epoch": 95, "lr": 9.13089969038141e-05} +{"train_loss": 0.016483470797538757, "grad_norm": 0.06498126685619354, "global_step": 10097, "epoch": 95, "lr": 9.130732702630514e-05} +{"train_loss": 0.015907078981399536, "grad_norm": 0.06802483648061752, "global_step": 10098, "epoch": 95, "lr": 9.130565700366017e-05} +{"train_loss": 0.01688878983259201, "grad_norm": 0.06648911535739899, "global_step": 10099, "epoch": 95, "lr": 9.13039868358851e-05} +{"train_loss": 0.018737969920039177, "grad_norm": 0.05461535602807999, "global_step": 10100, "epoch": 95, "lr": 9.130231652298577e-05} +{"train_loss": 0.016444675624370575, "grad_norm": 0.0722871869802475, "global_step": 10101, "epoch": 95, "lr": 9.130064606496808e-05} +{"train_loss": 0.017619676887989044, "grad_norm": 0.09000398218631744, "global_step": 10102, "epoch": 95, "lr": 9.129897546183788e-05} +{"train_loss": 0.018535595387220383, "grad_norm": 0.07005849480628967, "global_step": 10103, "epoch": 95, "lr": 9.129730471360103e-05} +{"train_loss": 0.017927829176187515, "grad_norm": 0.07424729317426682, "global_step": 10104, "epoch": 95, "lr": 9.12956338202634e-05} +{"train_loss": 0.01644270122051239, "grad_norm": 0.07304883748292923, "global_step": 10105, "epoch": 95, "lr": 9.129396278183089e-05} +{"train_loss": 0.016836602240800858, "grad_norm": 0.1358346790075302, "global_step": 10106, "epoch": 95, "lr": 9.129229159830935e-05} +{"train_loss": 0.016575735062360764, "grad_norm": 0.04155195504426956, "global_step": 10107, "epoch": 95, "lr": 9.129062026970465e-05} +{"train_loss": 0.018519023433327675, "grad_norm": 0.1666061282157898, "global_step": 10108, "epoch": 95, "lr": 9.128894879602268e-05} +{"train_loss": 0.02147754654288292, "grad_norm": 0.10915680974721909, "global_step": 10109, "epoch": 95, "lr": 9.128727717726928e-05} +{"train_loss": 0.019226130098104477, "grad_norm": 0.09249462932348251, "global_step": 10110, "epoch": 95, "lr": 9.128560541345037e-05} +{"train_loss": 0.01997385546565056, "grad_norm": 0.06878761947154999, "global_step": 10111, "epoch": 95, "lr": 9.128393350457178e-05} +{"train_loss": 0.01949302852153778, "grad_norm": 0.05593061447143555, "global_step": 10112, "epoch": 95, "lr": 9.128226145063939e-05} +{"train_loss": 0.018342111259698868, "grad_norm": 0.19798678159713745, "global_step": 10113, "epoch": 95, "lr": 9.12805892516591e-05} +{"train_loss": 0.01659877598285675, "grad_norm": 0.06438068300485611, "global_step": 10114, "epoch": 95, "lr": 9.127891690763677e-05} +{"train_loss": 0.019126644358038902, "grad_norm": 0.05706339702010155, "global_step": 10115, "epoch": 95, "lr": 9.127724441857827e-05} +{"train_loss": 0.015195055864751339, "grad_norm": 0.05622744560241699, "global_step": 10116, "epoch": 95, "lr": 9.127557178448952e-05} +{"train_loss": 0.01685204729437828, "grad_norm": 0.1365470588207245, "global_step": 10117, "epoch": 95, "lr": 9.127389900537633e-05} +{"train_loss": 0.01839521899819374, "grad_norm": 0.09524476528167725, "global_step": 10118, "epoch": 95, "lr": 9.12722260812446e-05} +{"train_loss": 0.0166892372071743, "grad_norm": 0.0575549490749836, "global_step": 10119, "epoch": 95, "lr": 9.127055301210024e-05} +{"train_loss": 0.019771385937929153, "grad_norm": 0.06393497437238693, "global_step": 10120, "epoch": 95, "lr": 9.126887979794909e-05} +{"train_loss": 0.016165681183338165, "grad_norm": 0.0441935732960701, "global_step": 10121, "epoch": 95, "lr": 9.126720643879705e-05} +{"train_loss": 0.014883708208799362, "grad_norm": 0.09722260385751724, "global_step": 10122, "epoch": 95, "lr": 9.126553293464998e-05} +{"train_loss": 0.016405779868364334, "grad_norm": 0.04908798635005951, "global_step": 10123, "epoch": 95, "lr": 9.126385928551378e-05} +{"train_loss": 0.019055938348174095, "grad_norm": 0.10213618725538254, "global_step": 10124, "epoch": 95, "lr": 9.126218549139433e-05} +{"train_loss": 0.0207468681037426, "grad_norm": 0.19340984523296356, "global_step": 10125, "epoch": 95, "lr": 9.126051155229751e-05} +{"train_loss": 0.017480315640568733, "grad_norm": 0.06819310039281845, "global_step": 10126, "epoch": 95, "lr": 9.125883746822919e-05} +{"train_loss": 0.017297327518463135, "grad_norm": 0.058469776064157486, "global_step": 10127, "epoch": 95, "lr": 9.125716323919526e-05} +{"train_loss": 0.016249127686023712, "grad_norm": 0.0825313925743103, "global_step": 10128, "epoch": 95, "lr": 9.125548886520158e-05} +{"train_loss": 0.017872411757707596, "grad_norm": 0.047490935772657394, "global_step": 10129, "epoch": 95, "lr": 9.125381434625409e-05} +{"train_loss": 0.017265405505895615, "grad_norm": 0.06534571945667267, "global_step": 10130, "epoch": 95, "lr": 9.12521396823586e-05} +{"train_loss": 0.021151965484023094, "grad_norm": 0.10999330133199692, "global_step": 10131, "epoch": 95, "lr": 9.125046487352105e-05} +{"train_loss": 0.014764931984245777, "grad_norm": 0.13235804438591003, "global_step": 10132, "epoch": 95, "lr": 9.12487899197473e-05} +{"train_loss": 0.016224456951022148, "grad_norm": 0.05302823707461357, "global_step": 10133, "epoch": 95, "lr": 9.124711482104326e-05} +{"train_loss": 0.01823991723358631, "grad_norm": 0.09260746091604233, "global_step": 10134, "epoch": 95, "lr": 9.124543957741478e-05} +{"train_loss": 0.01663367822766304, "grad_norm": 0.09239400923252106, "global_step": 10135, "epoch": 95, "lr": 9.124376418886776e-05} +{"train_loss": 0.015957720577716827, "grad_norm": 0.06014082208275795, "global_step": 10136, "epoch": 95, "lr": 9.124208865540809e-05} +{"train_loss": 0.01844644546508789, "grad_norm": 0.07083345204591751, "global_step": 10137, "epoch": 95, "lr": 9.124041297704167e-05} +{"train_loss": 0.019410453736782074, "grad_norm": 0.0769495740532875, "global_step": 10138, "epoch": 95, "lr": 9.123873715377435e-05} +{"train_loss": 0.017902888357639313, "grad_norm": 0.06604050099849701, "global_step": 10139, "epoch": 95, "lr": 9.123706118561206e-05} +{"train_loss": 0.01766168884932995, "grad_norm": 0.06913133710622787, "global_step": 10140, "epoch": 95, "lr": 9.123538507256066e-05} +{"train_loss": 0.018064768984913826, "grad_norm": 0.06701420992612839, "global_step": 10141, "epoch": 95, "lr": 9.123370881462608e-05} +{"train_loss": 0.018298715353012085, "grad_norm": 0.07299840450286865, "global_step": 10142, "epoch": 95, "lr": 9.123203241181416e-05} +{"train_loss": 0.018315214663743973, "grad_norm": 0.09272141009569168, "global_step": 10143, "epoch": 95, "lr": 9.123035586413079e-05} +{"train_loss": 0.018256301060318947, "grad_norm": 0.07712578773498535, "global_step": 10144, "epoch": 95, "lr": 9.122867917158189e-05} +{"train_loss": 0.01664184033870697, "grad_norm": 0.05963599309325218, "global_step": 10145, "epoch": 95, "lr": 9.122700233417335e-05} +{"train_loss": 0.019282478839159012, "grad_norm": 0.047823596745729446, "global_step": 10146, "epoch": 95, "lr": 9.122532535191104e-05} +{"train_loss": 0.014943665824830532, "grad_norm": 0.044619444757699966, "global_step": 10147, "epoch": 95, "lr": 9.122364822480087e-05} +{"train_loss": 0.01834823191165924, "grad_norm": 0.06804375350475311, "global_step": 10148, "epoch": 95, "lr": 9.122197095284875e-05} +{"train_loss": 0.01534662302583456, "grad_norm": 0.09435633569955826, "global_step": 10149, "epoch": 95, "lr": 9.122029353606051e-05} +{"train_loss": 0.0186271071434021, "grad_norm": 0.07361655682325363, "global_step": 10150, "epoch": 95, "lr": 9.121861597444213e-05} +{"train_loss": 0.01731243170797825, "grad_norm": 0.12431430071592331, "global_step": 10151, "epoch": 95, "lr": 9.121693826799943e-05} +{"train_loss": 0.017280178144574165, "grad_norm": 0.07665713876485825, "global_step": 10152, "epoch": 95, "lr": 9.121526041673833e-05} +{"train_loss": 0.018970612436532974, "grad_norm": 0.1109451875090599, "global_step": 10153, "epoch": 95, "lr": 9.121358242066474e-05} +{"train_loss": 0.015000054612755775, "grad_norm": 0.10712028294801712, "global_step": 10154, "epoch": 95, "lr": 9.121190427978455e-05} +{"train_loss": 0.015641940757632256, "grad_norm": 0.09708660840988159, "global_step": 10155, "epoch": 95, "lr": 9.121022599410366e-05} +{"train_loss": 0.01967000588774681, "grad_norm": 0.06163822114467621, "global_step": 10156, "epoch": 95, "lr": 9.120854756362793e-05} +{"train_loss": 0.019348273053765297, "grad_norm": 0.04951652139425278, "global_step": 10157, "epoch": 95, "lr": 9.12068689883633e-05} +{"train_loss": 0.018524667248129845, "grad_norm": 0.05811168625950813, "global_step": 10158, "epoch": 95, "lr": 9.120519026831565e-05} +{"train_loss": 0.020764300599694252, "grad_norm": 0.09708309173583984, "global_step": 10159, "epoch": 95, "lr": 9.120351140349089e-05} +{"train_loss": 0.017346173524856567, "grad_norm": 0.06442560255527496, "global_step": 10160, "epoch": 95, "lr": 9.120183239389488e-05} +{"train_loss": 0.018020877614617348, "grad_norm": 0.07419704645872116, "global_step": 10161, "epoch": 95, "lr": 9.120015323953356e-05} +{"train_loss": 0.018796615302562714, "grad_norm": 0.05485647916793823, "global_step": 10162, "epoch": 95, "lr": 9.119847394041285e-05} +{"train_loss": 0.017845463007688522, "grad_norm": 0.05322956293821335, "global_step": 10163, "epoch": 95, "lr": 9.11967944965386e-05} +{"train_loss": 0.018631193786859512, "grad_norm": 0.06585952639579773, "global_step": 10164, "epoch": 95, "lr": 9.119511490791673e-05} +{"train_loss": 0.017754502594470978, "grad_norm": 0.11291948705911636, "global_step": 10165, "epoch": 95, "lr": 9.119343517455314e-05} +{"train_loss": 0.02068483829498291, "grad_norm": 0.046863287687301636, "global_step": 10166, "epoch": 95, "lr": 9.119175529645372e-05} +{"train_loss": 0.016933724284172058, "grad_norm": 0.05230748653411865, "global_step": 10167, "epoch": 95, "lr": 9.11900752736244e-05} +{"train_loss": 0.015818899497389793, "grad_norm": 0.052530404180288315, "global_step": 10168, "epoch": 95, "lr": 9.118839510607107e-05} +{"train_loss": 0.017179027199745178, "grad_norm": 0.09225527197122574, "global_step": 10169, "epoch": 95, "lr": 9.118671479379963e-05} +{"train_loss": 0.01926206797361374, "grad_norm": 0.12719392776489258, "global_step": 10170, "epoch": 95, "lr": 9.118503433681599e-05} +{"train_loss": 0.017651278525590897, "grad_norm": 0.041743095964193344, "global_step": 10171, "epoch": 95, "lr": 9.118335373512604e-05} +{"train_loss": 0.01760079339146614, "grad_norm": 0.142114520072937, "global_step": 10172, "epoch": 95, "lr": 9.11816729887357e-05} +{"train_loss": 0.01691199280321598, "grad_norm": 0.09894821792840958, "global_step": 10173, "epoch": 95, "lr": 9.117999209765087e-05} +{"train_loss": 0.013706406578421593, "grad_norm": 0.056726422160863876, "global_step": 10174, "epoch": 95, "lr": 9.117831106187747e-05} +{"train_loss": 0.01767230001366082, "grad_norm": 0.10524898022413254, "global_step": 10175, "epoch": 95, "lr": 9.117662988142138e-05, "val_loss": 0.018904542550444603} +{"train_loss": 0.01707572676241398, "grad_norm": 0.05495868995785713, "global_step": 10176, "epoch": 96, "lr": 9.117494855628852e-05} +{"train_loss": 0.019562294706702232, "grad_norm": 0.08337265253067017, "global_step": 10177, "epoch": 96, "lr": 9.11732670864848e-05} +{"train_loss": 0.015335151925683022, "grad_norm": 0.07751021534204483, "global_step": 10178, "epoch": 96, "lr": 9.117158547201612e-05} +{"train_loss": 0.021918073296546936, "grad_norm": 0.10567525774240494, "global_step": 10179, "epoch": 96, "lr": 9.116990371288841e-05} +{"train_loss": 0.019301097840070724, "grad_norm": 0.06172189116477966, "global_step": 10180, "epoch": 96, "lr": 9.116822180910753e-05} +{"train_loss": 0.016985580325126648, "grad_norm": 0.09703516960144043, "global_step": 10181, "epoch": 96, "lr": 9.116653976067944e-05} +{"train_loss": 0.018561743199825287, "grad_norm": 0.05080147832632065, "global_step": 10182, "epoch": 96, "lr": 9.116485756761005e-05} +{"train_loss": 0.01833156868815422, "grad_norm": 0.05136878043413162, "global_step": 10183, "epoch": 96, "lr": 9.116317522990523e-05} +{"train_loss": 0.015854867175221443, "grad_norm": 0.056030817329883575, "global_step": 10184, "epoch": 96, "lr": 9.116149274757093e-05} +{"train_loss": 0.015209682285785675, "grad_norm": 0.11391345411539078, "global_step": 10185, "epoch": 96, "lr": 9.115981012061303e-05} +{"train_loss": 0.017011627554893494, "grad_norm": 0.05736767128109932, "global_step": 10186, "epoch": 96, "lr": 9.115812734903745e-05} +{"train_loss": 0.015344665385782719, "grad_norm": 0.09750435501337051, "global_step": 10187, "epoch": 96, "lr": 9.115644443285012e-05} +{"train_loss": 0.01850125938653946, "grad_norm": 0.07362663000822067, "global_step": 10188, "epoch": 96, "lr": 9.115476137205696e-05} +{"train_loss": 0.018289513885974884, "grad_norm": 0.10022524744272232, "global_step": 10189, "epoch": 96, "lr": 9.115307816666384e-05} +{"train_loss": 0.016947556287050247, "grad_norm": 0.05033630132675171, "global_step": 10190, "epoch": 96, "lr": 9.115139481667672e-05} +{"train_loss": 0.016850052401423454, "grad_norm": 0.057821281254291534, "global_step": 10191, "epoch": 96, "lr": 9.114971132210149e-05} +{"train_loss": 0.016433008015155792, "grad_norm": 0.10113300383090973, "global_step": 10192, "epoch": 96, "lr": 9.114802768294406e-05} +{"train_loss": 0.016384892165660858, "grad_norm": 0.0471547394990921, "global_step": 10193, "epoch": 96, "lr": 9.114634389921035e-05} +{"train_loss": 0.016072850674390793, "grad_norm": 0.03968077525496483, "global_step": 10194, "epoch": 96, "lr": 9.114465997090631e-05} +{"train_loss": 0.01876576989889145, "grad_norm": 0.07432803511619568, "global_step": 10195, "epoch": 96, "lr": 9.11429758980378e-05} +{"train_loss": 0.020361438393592834, "grad_norm": 0.15565505623817444, "global_step": 10196, "epoch": 96, "lr": 9.114129168061079e-05} +{"train_loss": 0.0155895184725523, "grad_norm": 0.04584387317299843, "global_step": 10197, "epoch": 96, "lr": 9.113960731863116e-05} +{"train_loss": 0.017749793827533722, "grad_norm": 0.05867486447095871, "global_step": 10198, "epoch": 96, "lr": 9.113792281210483e-05} +{"train_loss": 0.015824250876903534, "grad_norm": 0.0649072676897049, "global_step": 10199, "epoch": 96, "lr": 9.113623816103773e-05} +{"train_loss": 0.01786976307630539, "grad_norm": 0.06396663933992386, "global_step": 10200, "epoch": 96, "lr": 9.11345533654358e-05} +{"train_loss": 0.018491230905056, "grad_norm": 0.05203009769320488, "global_step": 10201, "epoch": 96, "lr": 9.113286842530493e-05} +{"train_loss": 0.015403971076011658, "grad_norm": 0.05397043377161026, "global_step": 10202, "epoch": 96, "lr": 9.113118334065104e-05} +{"train_loss": 0.01689116656780243, "grad_norm": 0.05206524580717087, "global_step": 10203, "epoch": 96, "lr": 9.112949811148007e-05} +{"train_loss": 0.016248153522610664, "grad_norm": 0.04442177712917328, "global_step": 10204, "epoch": 96, "lr": 9.112781273779793e-05} +{"train_loss": 0.015531790442764759, "grad_norm": 0.03933417424559593, "global_step": 10205, "epoch": 96, "lr": 9.112612721961053e-05} +{"train_loss": 0.019878972321748734, "grad_norm": 0.09116382896900177, "global_step": 10206, "epoch": 96, "lr": 9.112444155692381e-05} +{"train_loss": 0.020625941455364227, "grad_norm": 0.054149333387613297, "global_step": 10207, "epoch": 96, "lr": 9.11227557497437e-05} +{"train_loss": 0.017451424151659012, "grad_norm": 0.13757842779159546, "global_step": 10208, "epoch": 96, "lr": 9.11210697980761e-05} +{"train_loss": 0.019536171108484268, "grad_norm": 0.06947068125009537, "global_step": 10209, "epoch": 96, "lr": 9.111938370192694e-05} +{"train_loss": 0.017969077453017235, "grad_norm": 0.07126917690038681, "global_step": 10210, "epoch": 96, "lr": 9.111769746130215e-05} +{"train_loss": 0.01485821045935154, "grad_norm": 0.0612938217818737, "global_step": 10211, "epoch": 96, "lr": 9.111601107620766e-05} +{"train_loss": 0.01613975316286087, "grad_norm": 0.11396485567092896, "global_step": 10212, "epoch": 96, "lr": 9.111432454664937e-05} +{"train_loss": 0.014976562932133675, "grad_norm": 0.09316278994083405, "global_step": 10213, "epoch": 96, "lr": 9.111263787263324e-05} +{"train_loss": 0.020863156765699387, "grad_norm": 0.05311138555407524, "global_step": 10214, "epoch": 96, "lr": 9.111095105416518e-05} +{"train_loss": 0.01967669278383255, "grad_norm": 0.06803882122039795, "global_step": 10215, "epoch": 96, "lr": 9.110926409125112e-05} +{"train_loss": 0.017465349286794662, "grad_norm": 0.1401652842760086, "global_step": 10216, "epoch": 96, "lr": 9.110757698389696e-05} +{"train_loss": 0.016365844756364822, "grad_norm": 0.04672449454665184, "global_step": 10217, "epoch": 96, "lr": 9.110588973210867e-05} +{"train_loss": 0.017961930483579636, "grad_norm": 0.061380136758089066, "global_step": 10218, "epoch": 96, "lr": 9.110420233589217e-05} +{"train_loss": 0.020024806261062622, "grad_norm": 0.08822381496429443, "global_step": 10219, "epoch": 96, "lr": 9.110251479525335e-05} +{"train_loss": 0.016788233071565628, "grad_norm": 0.05329443886876106, "global_step": 10220, "epoch": 96, "lr": 9.11008271101982e-05} +{"train_loss": 0.017910877242684364, "grad_norm": 0.06576089560985565, "global_step": 10221, "epoch": 96, "lr": 9.10991392807326e-05} +{"train_loss": 0.016073182225227356, "grad_norm": 0.08197794109582901, "global_step": 10222, "epoch": 96, "lr": 9.109745130686249e-05} +{"train_loss": 0.016843318939208984, "grad_norm": 0.13034184277057648, "global_step": 10223, "epoch": 96, "lr": 9.109576318859383e-05} +{"train_loss": 0.017943784594535828, "grad_norm": 0.054660629481077194, "global_step": 10224, "epoch": 96, "lr": 9.109407492593252e-05} +{"train_loss": 0.01811336725950241, "grad_norm": 0.073996402323246, "global_step": 10225, "epoch": 96, "lr": 9.10923865188845e-05} +{"train_loss": 0.0178380124270916, "grad_norm": 0.14559069275856018, "global_step": 10226, "epoch": 96, "lr": 9.10906979674557e-05} +{"train_loss": 0.01977596804499626, "grad_norm": 0.15189699828624725, "global_step": 10227, "epoch": 96, "lr": 9.108900927165206e-05} +{"train_loss": 0.019890196621418, "grad_norm": 0.05541999638080597, "global_step": 10228, "epoch": 96, "lr": 9.108732043147952e-05} +{"train_loss": 0.017949415370821953, "grad_norm": 0.08679739385843277, "global_step": 10229, "epoch": 96, "lr": 9.1085631446944e-05} +{"train_loss": 0.01757832244038582, "grad_norm": 0.10314910113811493, "global_step": 10230, "epoch": 96, "lr": 9.108394231805142e-05} +{"train_loss": 0.016603225842118263, "grad_norm": 0.05308046564459801, "global_step": 10231, "epoch": 96, "lr": 9.108225304480777e-05} +{"train_loss": 0.017849285155534744, "grad_norm": 0.056587181985378265, "global_step": 10232, "epoch": 96, "lr": 9.108056362721893e-05} +{"train_loss": 0.016372906044125557, "grad_norm": 0.08276264369487762, "global_step": 10233, "epoch": 96, "lr": 9.107887406529086e-05} +{"train_loss": 0.018619490787386894, "grad_norm": 0.07929348200559616, "global_step": 10234, "epoch": 96, "lr": 9.107718435902948e-05} +{"train_loss": 0.01850030943751335, "grad_norm": 0.08196638524532318, "global_step": 10235, "epoch": 96, "lr": 9.107549450844072e-05} +{"train_loss": 0.017982617020606995, "grad_norm": 0.0686924010515213, "global_step": 10236, "epoch": 96, "lr": 9.107380451353058e-05} +{"train_loss": 0.01685751974582672, "grad_norm": 0.04348482936620712, "global_step": 10237, "epoch": 96, "lr": 9.107211437430493e-05} +{"train_loss": 0.016155699267983437, "grad_norm": 0.0786367803812027, "global_step": 10238, "epoch": 96, "lr": 9.107042409076974e-05} +{"train_loss": 0.015770960599184036, "grad_norm": 0.056052688509225845, "global_step": 10239, "epoch": 96, "lr": 9.106873366293093e-05} +{"train_loss": 0.01959158107638359, "grad_norm": 0.06832320988178253, "global_step": 10240, "epoch": 96, "lr": 9.106704309079446e-05} +{"train_loss": 0.01777523383498192, "grad_norm": 0.12877242267131805, "global_step": 10241, "epoch": 96, "lr": 9.106535237436624e-05} +{"train_loss": 0.01825902797281742, "grad_norm": 0.08349882811307907, "global_step": 10242, "epoch": 96, "lr": 9.106366151365227e-05} +{"train_loss": 0.017859358340501785, "grad_norm": 0.07721775025129318, "global_step": 10243, "epoch": 96, "lr": 9.106197050865841e-05} +{"train_loss": 0.019630298018455505, "grad_norm": 0.09571774303913116, "global_step": 10244, "epoch": 96, "lr": 9.106027935939067e-05} +{"train_loss": 0.02222702093422413, "grad_norm": 0.13971559703350067, "global_step": 10245, "epoch": 96, "lr": 9.105858806585494e-05} +{"train_loss": 0.017063375562429428, "grad_norm": 0.07458998262882233, "global_step": 10246, "epoch": 96, "lr": 9.10568966280572e-05} +{"train_loss": 0.019553570076823235, "grad_norm": 0.10854988545179367, "global_step": 10247, "epoch": 96, "lr": 9.10552050460034e-05} +{"train_loss": 0.018137939274311066, "grad_norm": 0.042137376964092255, "global_step": 10248, "epoch": 96, "lr": 9.105351331969945e-05} +{"train_loss": 0.017472321167588234, "grad_norm": 0.10544091463088989, "global_step": 10249, "epoch": 96, "lr": 9.10518214491513e-05} +{"train_loss": 0.017441822215914726, "grad_norm": 0.10062497854232788, "global_step": 10250, "epoch": 96, "lr": 9.105012943436491e-05} +{"train_loss": 0.013812955468893051, "grad_norm": 0.05153641104698181, "global_step": 10251, "epoch": 96, "lr": 9.104843727534622e-05} +{"train_loss": 0.01638752967119217, "grad_norm": 0.1019572839140892, "global_step": 10252, "epoch": 96, "lr": 9.104674497210116e-05} +{"train_loss": 0.018330397084355354, "grad_norm": 0.05668673664331436, "global_step": 10253, "epoch": 96, "lr": 9.104505252463572e-05} +{"train_loss": 0.019631098955869675, "grad_norm": 0.07830530405044556, "global_step": 10254, "epoch": 96, "lr": 9.104335993295577e-05} +{"train_loss": 0.017716875299811363, "grad_norm": 0.15185876190662384, "global_step": 10255, "epoch": 96, "lr": 9.104166719706733e-05} +{"train_loss": 0.017822125926613808, "grad_norm": 0.04940762370824814, "global_step": 10256, "epoch": 96, "lr": 9.103997431697631e-05} +{"train_loss": 0.019495997577905655, "grad_norm": 0.12601134181022644, "global_step": 10257, "epoch": 96, "lr": 9.103828129268868e-05} +{"train_loss": 0.017414012923836708, "grad_norm": 0.14076393842697144, "global_step": 10258, "epoch": 96, "lr": 9.103658812421037e-05} +{"train_loss": 0.01790030673146248, "grad_norm": 0.13262459635734558, "global_step": 10259, "epoch": 96, "lr": 9.103489481154734e-05} +{"train_loss": 0.016032299026846886, "grad_norm": 0.07794643193483353, "global_step": 10260, "epoch": 96, "lr": 9.103320135470553e-05} +{"train_loss": 0.01842198520898819, "grad_norm": 0.055277321487665176, "global_step": 10261, "epoch": 96, "lr": 9.103150775369089e-05} +{"train_loss": 0.018206408247351646, "grad_norm": 0.07925368845462799, "global_step": 10262, "epoch": 96, "lr": 9.102981400850939e-05} +{"train_loss": 0.017242958769202232, "grad_norm": 0.06876950711011887, "global_step": 10263, "epoch": 96, "lr": 9.102812011916694e-05} +{"train_loss": 0.01675930991768837, "grad_norm": 0.04020535573363304, "global_step": 10264, "epoch": 96, "lr": 9.102642608566955e-05} +{"train_loss": 0.016641691327095032, "grad_norm": 0.04652579873800278, "global_step": 10265, "epoch": 96, "lr": 9.102473190802312e-05} +{"train_loss": 0.01598621904850006, "grad_norm": 0.0834062248468399, "global_step": 10266, "epoch": 96, "lr": 9.102303758623364e-05} +{"train_loss": 0.01836094632744789, "grad_norm": 0.15369661152362823, "global_step": 10267, "epoch": 96, "lr": 9.102134312030704e-05} +{"train_loss": 0.016270361840724945, "grad_norm": 0.0877050831913948, "global_step": 10268, "epoch": 96, "lr": 9.101964851024926e-05} +{"train_loss": 0.019881371408700943, "grad_norm": 0.08683644980192184, "global_step": 10269, "epoch": 96, "lr": 9.10179537560663e-05} +{"train_loss": 0.017428576946258545, "grad_norm": 0.06926237791776657, "global_step": 10270, "epoch": 96, "lr": 9.101625885776408e-05} +{"train_loss": 0.015027394518256187, "grad_norm": 0.08306818455457687, "global_step": 10271, "epoch": 96, "lr": 9.101456381534857e-05} +{"train_loss": 0.016796525567770004, "grad_norm": 0.09501847624778748, "global_step": 10272, "epoch": 96, "lr": 9.101286862882572e-05} +{"train_loss": 0.017536228522658348, "grad_norm": 0.1247669905424118, "global_step": 10273, "epoch": 96, "lr": 9.101117329820149e-05} +{"train_loss": 0.019823148846626282, "grad_norm": 0.055550653487443924, "global_step": 10274, "epoch": 96, "lr": 9.10094778234818e-05} +{"train_loss": 0.019857024773955345, "grad_norm": 0.05204400420188904, "global_step": 10275, "epoch": 96, "lr": 9.100778220467266e-05} +{"train_loss": 0.016149796545505524, "grad_norm": 0.08446022868156433, "global_step": 10276, "epoch": 96, "lr": 9.100608644178002e-05} +{"train_loss": 0.017329204827547073, "grad_norm": 0.12138217687606812, "global_step": 10277, "epoch": 96, "lr": 9.10043905348098e-05} +{"train_loss": 0.018949728459119797, "grad_norm": 0.06122996285557747, "global_step": 10278, "epoch": 96, "lr": 9.100269448376801e-05} +{"train_loss": 0.014749249443411827, "grad_norm": 0.05920197069644928, "global_step": 10279, "epoch": 96, "lr": 9.100099828866057e-05} +{"train_loss": 0.017150241881608963, "grad_norm": 0.12860925495624542, "global_step": 10280, "epoch": 96, "lr": 9.099930194949346e-05} +{"train_loss": 0.017639851531470723, "grad_norm": 0.06139041855931282, "global_step": 10281, "epoch": 96, "lr": 9.099760546627261e-05, "val_loss": 0.018747029826045036} +{"train_loss": 0.015991443768143654, "grad_norm": 0.04698174446821213, "global_step": 10282, "epoch": 97, "lr": 9.099590883900405e-05} +{"train_loss": 0.0168484915047884, "grad_norm": 0.08438262343406677, "global_step": 10283, "epoch": 97, "lr": 9.099421206769365e-05} +{"train_loss": 0.0177523884922266, "grad_norm": 0.07183706760406494, "global_step": 10284, "epoch": 97, "lr": 9.099251515234744e-05} +{"train_loss": 0.016362052410840988, "grad_norm": 0.0866817757487297, "global_step": 10285, "epoch": 97, "lr": 9.099081809297134e-05} +{"train_loss": 0.01716623082756996, "grad_norm": 0.07910258322954178, "global_step": 10286, "epoch": 97, "lr": 9.098912088957135e-05} +{"train_loss": 0.01889902725815773, "grad_norm": 0.09707453101873398, "global_step": 10287, "epoch": 97, "lr": 9.098742354215339e-05} +{"train_loss": 0.017702050507068634, "grad_norm": 0.08056696504354477, "global_step": 10288, "epoch": 97, "lr": 9.098572605072349e-05} +{"train_loss": 0.01673547737300396, "grad_norm": 0.07882213592529297, "global_step": 10289, "epoch": 97, "lr": 9.098402841528755e-05} +{"train_loss": 0.01834297552704811, "grad_norm": 0.08341233432292938, "global_step": 10290, "epoch": 97, "lr": 9.098233063585156e-05} +{"train_loss": 0.016184275969862938, "grad_norm": 0.05577459558844566, "global_step": 10291, "epoch": 97, "lr": 9.098063271242146e-05} +{"train_loss": 0.014669641852378845, "grad_norm": 0.0728209838271141, "global_step": 10292, "epoch": 97, "lr": 9.097893464500326e-05} +{"train_loss": 0.01842625066637993, "grad_norm": 0.07007908821105957, "global_step": 10293, "epoch": 97, "lr": 9.097723643360291e-05} +{"train_loss": 0.018670566380023956, "grad_norm": 0.12149342894554138, "global_step": 10294, "epoch": 97, "lr": 9.097553807822636e-05} +{"train_loss": 0.017552031204104424, "grad_norm": 0.08671554923057556, "global_step": 10295, "epoch": 97, "lr": 9.097383957887959e-05} +{"train_loss": 0.02052132785320282, "grad_norm": 0.06772436201572418, "global_step": 10296, "epoch": 97, "lr": 9.097214093556859e-05} +{"train_loss": 0.019630465656518936, "grad_norm": 0.05535659193992615, "global_step": 10297, "epoch": 97, "lr": 9.097044214829929e-05} +{"train_loss": 0.018643300980329514, "grad_norm": 0.14956419169902802, "global_step": 10298, "epoch": 97, "lr": 9.096874321707766e-05} +{"train_loss": 0.018408145755529404, "grad_norm": 0.07897907495498657, "global_step": 10299, "epoch": 97, "lr": 9.096704414190969e-05} +{"train_loss": 0.02229841984808445, "grad_norm": 0.0945286825299263, "global_step": 10300, "epoch": 97, "lr": 9.096534492280136e-05} +{"train_loss": 0.020528502762317657, "grad_norm": 0.0481966994702816, "global_step": 10301, "epoch": 97, "lr": 9.096364555975861e-05} +{"train_loss": 0.019404646009206772, "grad_norm": 0.1081988513469696, "global_step": 10302, "epoch": 97, "lr": 9.09619460527874e-05} +{"train_loss": 0.016785118728876114, "grad_norm": 0.10794384777545929, "global_step": 10303, "epoch": 97, "lr": 9.096024640189376e-05} +{"train_loss": 0.017574157565832138, "grad_norm": 0.04610194265842438, "global_step": 10304, "epoch": 97, "lr": 9.095854660708362e-05} +{"train_loss": 0.02129669487476349, "grad_norm": 0.06630104780197144, "global_step": 10305, "epoch": 97, "lr": 9.095684666836294e-05} +{"train_loss": 0.016202401369810104, "grad_norm": 0.047467220574617386, "global_step": 10306, "epoch": 97, "lr": 9.095514658573774e-05} +{"train_loss": 0.020538829267024994, "grad_norm": 0.11622390151023865, "global_step": 10307, "epoch": 97, "lr": 9.095344635921396e-05} +{"train_loss": 0.014758666045963764, "grad_norm": 0.07057861238718033, "global_step": 10308, "epoch": 97, "lr": 9.095174598879757e-05} +{"train_loss": 0.01741066761314869, "grad_norm": 0.07345043867826462, "global_step": 10309, "epoch": 97, "lr": 9.095004547449455e-05} +{"train_loss": 0.01838253065943718, "grad_norm": 0.04776269942522049, "global_step": 10310, "epoch": 97, "lr": 9.094834481631089e-05} +{"train_loss": 0.01708604209125042, "grad_norm": 0.07081291824579239, "global_step": 10311, "epoch": 97, "lr": 9.094664401425255e-05} +{"train_loss": 0.02057012729346752, "grad_norm": 0.08368620276451111, "global_step": 10312, "epoch": 97, "lr": 9.094494306832551e-05} +{"train_loss": 0.016880065202713013, "grad_norm": 0.06711401790380478, "global_step": 10313, "epoch": 97, "lr": 9.094324197853576e-05} +{"train_loss": 0.017721105366945267, "grad_norm": 0.07100702077150345, "global_step": 10314, "epoch": 97, "lr": 9.094154074488925e-05} +{"train_loss": 0.016803860664367676, "grad_norm": 0.04386324808001518, "global_step": 10315, "epoch": 97, "lr": 9.093983936739196e-05} +{"train_loss": 0.017709583044052124, "grad_norm": 0.11651154607534409, "global_step": 10316, "epoch": 97, "lr": 9.09381378460499e-05} +{"train_loss": 0.01877656579017639, "grad_norm": 0.06686518341302872, "global_step": 10317, "epoch": 97, "lr": 9.093643618086902e-05} +{"train_loss": 0.019347406923770905, "grad_norm": 0.11945019662380219, "global_step": 10318, "epoch": 97, "lr": 9.09347343718553e-05} +{"train_loss": 0.01905181258916855, "grad_norm": 0.07317628711462021, "global_step": 10319, "epoch": 97, "lr": 9.093303241901473e-05} +{"train_loss": 0.01868513599038124, "grad_norm": 0.05994752049446106, "global_step": 10320, "epoch": 97, "lr": 9.093133032235328e-05} +{"train_loss": 0.016502846032381058, "grad_norm": 0.08371203392744064, "global_step": 10321, "epoch": 97, "lr": 9.092962808187695e-05} +{"train_loss": 0.017734786495566368, "grad_norm": 0.04695707932114601, "global_step": 10322, "epoch": 97, "lr": 9.092792569759169e-05} +{"train_loss": 0.01522210706025362, "grad_norm": 0.1048949658870697, "global_step": 10323, "epoch": 97, "lr": 9.092622316950351e-05} +{"train_loss": 0.016285378485918045, "grad_norm": 0.05618554353713989, "global_step": 10324, "epoch": 97, "lr": 9.092452049761837e-05} +{"train_loss": 0.01602691411972046, "grad_norm": 0.10323024541139603, "global_step": 10325, "epoch": 97, "lr": 9.092281768194228e-05} +{"train_loss": 0.017117593437433243, "grad_norm": 0.08474680781364441, "global_step": 10326, "epoch": 97, "lr": 9.09211147224812e-05} +{"train_loss": 0.015093990601599216, "grad_norm": 0.09941123425960541, "global_step": 10327, "epoch": 97, "lr": 9.09194116192411e-05} +{"train_loss": 0.017651639878749847, "grad_norm": 0.07430016249418259, "global_step": 10328, "epoch": 97, "lr": 9.091770837222799e-05} +{"train_loss": 0.017036855220794678, "grad_norm": 0.07693532854318619, "global_step": 10329, "epoch": 97, "lr": 9.091600498144785e-05} +{"train_loss": 0.017206508666276932, "grad_norm": 0.060233086347579956, "global_step": 10330, "epoch": 97, "lr": 9.091430144690667e-05} +{"train_loss": 0.019239453598856926, "grad_norm": 0.05708581581711769, "global_step": 10331, "epoch": 97, "lr": 9.091259776861044e-05} +{"train_loss": 0.01968216337263584, "grad_norm": 0.07068202644586563, "global_step": 10332, "epoch": 97, "lr": 9.09108939465651e-05} +{"train_loss": 0.01700131967663765, "grad_norm": 0.09056124836206436, "global_step": 10333, "epoch": 97, "lr": 9.090918998077668e-05} +{"train_loss": 0.015723876655101776, "grad_norm": 0.06629661470651627, "global_step": 10334, "epoch": 97, "lr": 9.090748587125118e-05} +{"train_loss": 0.01878679171204567, "grad_norm": 0.11284369975328445, "global_step": 10335, "epoch": 97, "lr": 9.090578161799454e-05} +{"train_loss": 0.018107343465089798, "grad_norm": 0.06427749991416931, "global_step": 10336, "epoch": 97, "lr": 9.09040772210128e-05} +{"train_loss": 0.01674516685307026, "grad_norm": 0.08919785171747208, "global_step": 10337, "epoch": 97, "lr": 9.09023726803119e-05} +{"train_loss": 0.01569586805999279, "grad_norm": 0.09004854410886765, "global_step": 10338, "epoch": 97, "lr": 9.090066799589785e-05} +{"train_loss": 0.01873692497611046, "grad_norm": 0.061023909598588943, "global_step": 10339, "epoch": 97, "lr": 9.089896316777664e-05} +{"train_loss": 0.01619480550289154, "grad_norm": 0.08065710216760635, "global_step": 10340, "epoch": 97, "lr": 9.089725819595428e-05} +{"train_loss": 0.0179445780813694, "grad_norm": 0.07144904881715775, "global_step": 10341, "epoch": 97, "lr": 9.089555308043672e-05} +{"train_loss": 0.015844270586967468, "grad_norm": 0.06733542680740356, "global_step": 10342, "epoch": 97, "lr": 9.089384782122995e-05} +{"train_loss": 0.019636908546090126, "grad_norm": 0.0882602259516716, "global_step": 10343, "epoch": 97, "lr": 9.089214241834003e-05} +{"train_loss": 0.015183729119598866, "grad_norm": 0.05630834773182869, "global_step": 10344, "epoch": 97, "lr": 9.089043687177288e-05} +{"train_loss": 0.015217382460832596, "grad_norm": 0.183845654129982, "global_step": 10345, "epoch": 97, "lr": 9.088873118153451e-05} +{"train_loss": 0.01723068580031395, "grad_norm": 0.10228651016950607, "global_step": 10346, "epoch": 97, "lr": 9.088702534763093e-05} +{"train_loss": 0.017206046730279922, "grad_norm": 0.09059131890535355, "global_step": 10347, "epoch": 97, "lr": 9.088531937006815e-05} +{"train_loss": 0.018063370138406754, "grad_norm": 0.0601503811776638, "global_step": 10348, "epoch": 97, "lr": 9.088361324885211e-05} +{"train_loss": 0.01617058366537094, "grad_norm": 0.07786432653665543, "global_step": 10349, "epoch": 97, "lr": 9.088190698398883e-05} +{"train_loss": 0.019563129171729088, "grad_norm": 0.09075777232646942, "global_step": 10350, "epoch": 97, "lr": 9.088020057548432e-05} +{"train_loss": 0.01752208173274994, "grad_norm": 0.06040245667099953, "global_step": 10351, "epoch": 97, "lr": 9.087849402334456e-05} +{"train_loss": 0.01782594993710518, "grad_norm": 0.14803242683410645, "global_step": 10352, "epoch": 97, "lr": 9.087678732757555e-05} +{"train_loss": 0.016365494579076767, "grad_norm": 0.051810480654239655, "global_step": 10353, "epoch": 97, "lr": 9.087508048818328e-05} +{"train_loss": 0.016328932717442513, "grad_norm": 0.06696565449237823, "global_step": 10354, "epoch": 97, "lr": 9.087337350517377e-05} +{"train_loss": 0.020223159343004227, "grad_norm": 0.10764669626951218, "global_step": 10355, "epoch": 97, "lr": 9.087166637855298e-05} +{"train_loss": 0.01929587684571743, "grad_norm": 0.07606774568557739, "global_step": 10356, "epoch": 97, "lr": 9.086995910832694e-05} +{"train_loss": 0.01868688501417637, "grad_norm": 0.13277976214885712, "global_step": 10357, "epoch": 97, "lr": 9.086825169450164e-05} +{"train_loss": 0.02011866681277752, "grad_norm": 0.058931753039360046, "global_step": 10358, "epoch": 97, "lr": 9.086654413708307e-05} +{"train_loss": 0.018259646371006966, "grad_norm": 0.05005991831421852, "global_step": 10359, "epoch": 97, "lr": 9.086483643607724e-05} +{"train_loss": 0.019058316946029663, "grad_norm": 0.0947326049208641, "global_step": 10360, "epoch": 97, "lr": 9.086312859149014e-05} +{"train_loss": 0.01885046623647213, "grad_norm": 0.06732034683227539, "global_step": 10361, "epoch": 97, "lr": 9.086142060332778e-05} +{"train_loss": 0.017606932669878006, "grad_norm": 0.04928528517484665, "global_step": 10362, "epoch": 97, "lr": 9.085971247159617e-05} +{"train_loss": 0.01929609104990959, "grad_norm": 0.06588058918714523, "global_step": 10363, "epoch": 97, "lr": 9.085800419630128e-05} +{"train_loss": 0.01725747622549534, "grad_norm": 0.04898706078529358, "global_step": 10364, "epoch": 97, "lr": 9.085629577744915e-05} +{"train_loss": 0.016031108796596527, "grad_norm": 0.05800091102719307, "global_step": 10365, "epoch": 97, "lr": 9.085458721504576e-05} +{"train_loss": 0.016980784013867378, "grad_norm": 0.13800367712974548, "global_step": 10366, "epoch": 97, "lr": 9.085287850909711e-05} +{"train_loss": 0.016593879088759422, "grad_norm": 0.10087937116622925, "global_step": 10367, "epoch": 97, "lr": 9.085116965960921e-05} +{"train_loss": 0.017666371539235115, "grad_norm": 0.045924682170152664, "global_step": 10368, "epoch": 97, "lr": 9.084946066658808e-05} +{"train_loss": 0.016327528282999992, "grad_norm": 0.07832246273756027, "global_step": 10369, "epoch": 97, "lr": 9.08477515300397e-05} +{"train_loss": 0.019368976354599, "grad_norm": 0.0716794803738594, "global_step": 10370, "epoch": 97, "lr": 9.084604224997009e-05} +{"train_loss": 0.018130691722035408, "grad_norm": 0.05172006040811539, "global_step": 10371, "epoch": 97, "lr": 9.084433282638525e-05} +{"train_loss": 0.016668764874339104, "grad_norm": 0.06938854604959488, "global_step": 10372, "epoch": 97, "lr": 9.084262325929118e-05} +{"train_loss": 0.015436469577252865, "grad_norm": 0.09704164415597916, "global_step": 10373, "epoch": 97, "lr": 9.084091354869389e-05} +{"train_loss": 0.014754435047507286, "grad_norm": 0.04440602660179138, "global_step": 10374, "epoch": 97, "lr": 9.083920369459941e-05} +{"train_loss": 0.017809387296438217, "grad_norm": 0.04840095713734627, "global_step": 10375, "epoch": 97, "lr": 9.083749369701372e-05} +{"train_loss": 0.018177030608057976, "grad_norm": 0.04662406072020531, "global_step": 10376, "epoch": 97, "lr": 9.083578355594283e-05} +{"train_loss": 0.016844026744365692, "grad_norm": 0.14474338293075562, "global_step": 10377, "epoch": 97, "lr": 9.083407327139276e-05} +{"train_loss": 0.01683776080608368, "grad_norm": 0.0950356125831604, "global_step": 10378, "epoch": 97, "lr": 9.083236284336951e-05} +{"train_loss": 0.017977990210056305, "grad_norm": 0.05007362365722656, "global_step": 10379, "epoch": 97, "lr": 9.08306522718791e-05} +{"train_loss": 0.01605130359530449, "grad_norm": 0.07926547527313232, "global_step": 10380, "epoch": 97, "lr": 9.082894155692754e-05} +{"train_loss": 0.0182261411100626, "grad_norm": 0.07277029007673264, "global_step": 10381, "epoch": 97, "lr": 9.082723069852084e-05} +{"train_loss": 0.01675533317029476, "grad_norm": 0.06559862941503525, "global_step": 10382, "epoch": 97, "lr": 9.0825519696665e-05} +{"train_loss": 0.01598411425948143, "grad_norm": 0.08491748571395874, "global_step": 10383, "epoch": 97, "lr": 9.082380855136603e-05} +{"train_loss": 0.01656639762222767, "grad_norm": 0.10372324287891388, "global_step": 10384, "epoch": 97, "lr": 9.082209726262996e-05} +{"train_loss": 0.01687082089483738, "grad_norm": 0.049778033047914505, "global_step": 10385, "epoch": 97, "lr": 9.082038583046277e-05} +{"train_loss": 0.018176421523094177, "grad_norm": 0.05605009198188782, "global_step": 10386, "epoch": 97, "lr": 9.081867425487053e-05} +{"train_loss": 0.01760302088662701, "grad_norm": 0.07431337982416153, "global_step": 10387, "epoch": 97, "lr": 9.081696253585921e-05, "val_loss": 0.01944701001048088} +{"train_loss": 0.013866105116903782, "grad_norm": 0.05036468058824539, "global_step": 10388, "epoch": 98, "lr": 9.081525067343483e-05} +{"train_loss": 0.01557615865021944, "grad_norm": 0.062260743230581284, "global_step": 10389, "epoch": 98, "lr": 9.08135386676034e-05} +{"train_loss": 0.016040107235312462, "grad_norm": 0.052509330213069916, "global_step": 10390, "epoch": 98, "lr": 9.081182651837095e-05} +{"train_loss": 0.018144264817237854, "grad_norm": 0.06269063800573349, "global_step": 10391, "epoch": 98, "lr": 9.081011422574351e-05} +{"train_loss": 0.01654576137661934, "grad_norm": 0.07581659406423569, "global_step": 10392, "epoch": 98, "lr": 9.080840178972705e-05} +{"train_loss": 0.018455130979418755, "grad_norm": 0.06747247278690338, "global_step": 10393, "epoch": 98, "lr": 9.080668921032763e-05} +{"train_loss": 0.018079515546560287, "grad_norm": 0.0659310594201088, "global_step": 10394, "epoch": 98, "lr": 9.080497648755124e-05} +{"train_loss": 0.017811138182878494, "grad_norm": 0.0654095709323883, "global_step": 10395, "epoch": 98, "lr": 9.080326362140393e-05} +{"train_loss": 0.021748041734099388, "grad_norm": 0.09657826274633408, "global_step": 10396, "epoch": 98, "lr": 9.080155061189167e-05} +{"train_loss": 0.019470401108264923, "grad_norm": 0.04464033618569374, "global_step": 10397, "epoch": 98, "lr": 9.079983745902052e-05} +{"train_loss": 0.01702754944562912, "grad_norm": 0.041920918971300125, "global_step": 10398, "epoch": 98, "lr": 9.079812416279647e-05} +{"train_loss": 0.018405431881546974, "grad_norm": 0.06472282111644745, "global_step": 10399, "epoch": 98, "lr": 9.079641072322556e-05} +{"train_loss": 0.01635502651333809, "grad_norm": 0.051191024482250214, "global_step": 10400, "epoch": 98, "lr": 9.07946971403138e-05} +{"train_loss": 0.01632472313940525, "grad_norm": 0.05539403855800629, "global_step": 10401, "epoch": 98, "lr": 9.079298341406723e-05} +{"train_loss": 0.020033951848745346, "grad_norm": 0.12695685029029846, "global_step": 10402, "epoch": 98, "lr": 9.079126954449185e-05} +{"train_loss": 0.017427369952201843, "grad_norm": 0.07436061650514603, "global_step": 10403, "epoch": 98, "lr": 9.07895555315937e-05} +{"train_loss": 0.016659922897815704, "grad_norm": 0.056303720921278, "global_step": 10404, "epoch": 98, "lr": 9.078784137537876e-05} +{"train_loss": 0.01951029896736145, "grad_norm": 0.05753707513213158, "global_step": 10405, "epoch": 98, "lr": 9.078612707585312e-05} +{"train_loss": 0.017537422478199005, "grad_norm": 0.1671212613582611, "global_step": 10406, "epoch": 98, "lr": 9.078441263302274e-05} +{"train_loss": 0.013857007026672363, "grad_norm": 0.041051022708415985, "global_step": 10407, "epoch": 98, "lr": 9.078269804689368e-05} +{"train_loss": 0.017916353419423103, "grad_norm": 0.04559851437807083, "global_step": 10408, "epoch": 98, "lr": 9.078098331747196e-05} +{"train_loss": 0.017795024439692497, "grad_norm": 0.08126933127641678, "global_step": 10409, "epoch": 98, "lr": 9.077926844476359e-05} +{"train_loss": 0.017351092770695686, "grad_norm": 0.08628971129655838, "global_step": 10410, "epoch": 98, "lr": 9.077755342877461e-05} +{"train_loss": 0.016325626522302628, "grad_norm": 0.0569990798830986, "global_step": 10411, "epoch": 98, "lr": 9.077583826951105e-05} +{"train_loss": 0.015623586252331734, "grad_norm": 0.07564745098352432, "global_step": 10412, "epoch": 98, "lr": 9.077412296697892e-05} +{"train_loss": 0.016245819628238678, "grad_norm": 0.0622430257499218, "global_step": 10413, "epoch": 98, "lr": 9.077240752118424e-05} +{"train_loss": 0.014637397602200508, "grad_norm": 0.0810421034693718, "global_step": 10414, "epoch": 98, "lr": 9.077069193213308e-05} +{"train_loss": 0.01721980795264244, "grad_norm": 0.05122914910316467, "global_step": 10415, "epoch": 98, "lr": 9.076897619983142e-05} +{"train_loss": 0.02022693306207657, "grad_norm": 0.05645879730582237, "global_step": 10416, "epoch": 98, "lr": 9.07672603242853e-05} +{"train_loss": 0.017702829092741013, "grad_norm": 0.07994353026151657, "global_step": 10417, "epoch": 98, "lr": 9.076554430550077e-05} +{"train_loss": 0.018860887736082077, "grad_norm": 0.07617145776748657, "global_step": 10418, "epoch": 98, "lr": 9.076382814348384e-05} +{"train_loss": 0.014715767465531826, "grad_norm": 0.048195600509643555, "global_step": 10419, "epoch": 98, "lr": 9.076211183824056e-05} +{"train_loss": 0.016938120126724243, "grad_norm": 0.06848322600126266, "global_step": 10420, "epoch": 98, "lr": 9.076039538977693e-05} +{"train_loss": 0.01886921003460884, "grad_norm": 0.08946141600608826, "global_step": 10421, "epoch": 98, "lr": 9.0758678798099e-05} +{"train_loss": 0.018964648246765137, "grad_norm": 0.055129848420619965, "global_step": 10422, "epoch": 98, "lr": 9.07569620632128e-05} +{"train_loss": 0.02236713096499443, "grad_norm": 0.0545433945953846, "global_step": 10423, "epoch": 98, "lr": 9.075524518512435e-05} +{"train_loss": 0.01961642876267433, "grad_norm": 0.11345534026622772, "global_step": 10424, "epoch": 98, "lr": 9.075352816383971e-05} +{"train_loss": 0.01690811663866043, "grad_norm": 0.08404721319675446, "global_step": 10425, "epoch": 98, "lr": 9.07518109993649e-05} +{"train_loss": 0.01886489987373352, "grad_norm": 0.04852715879678726, "global_step": 10426, "epoch": 98, "lr": 9.075009369170593e-05} +{"train_loss": 0.016199618577957153, "grad_norm": 0.08895689249038696, "global_step": 10427, "epoch": 98, "lr": 9.074837624086884e-05} +{"train_loss": 0.01671648770570755, "grad_norm": 0.06966736167669296, "global_step": 10428, "epoch": 98, "lr": 9.074665864685971e-05} +{"train_loss": 0.018138911575078964, "grad_norm": 0.10146412998437881, "global_step": 10429, "epoch": 98, "lr": 9.074494090968453e-05} +{"train_loss": 0.01756419986486435, "grad_norm": 0.04789126291871071, "global_step": 10430, "epoch": 98, "lr": 9.074322302934935e-05} +{"train_loss": 0.016487952321767807, "grad_norm": 0.06095980107784271, "global_step": 10431, "epoch": 98, "lr": 9.074150500586019e-05} +{"train_loss": 0.01737164705991745, "grad_norm": 0.05460105463862419, "global_step": 10432, "epoch": 98, "lr": 9.07397868392231e-05} +{"train_loss": 0.020588930696249008, "grad_norm": 0.08024038374423981, "global_step": 10433, "epoch": 98, "lr": 9.073806852944413e-05} +{"train_loss": 0.01625608280301094, "grad_norm": 0.07122812420129776, "global_step": 10434, "epoch": 98, "lr": 9.07363500765293e-05} +{"train_loss": 0.018505040556192398, "grad_norm": 0.053543489426374435, "global_step": 10435, "epoch": 98, "lr": 9.073463148048465e-05} +{"train_loss": 0.018730519339442253, "grad_norm": 0.06759972125291824, "global_step": 10436, "epoch": 98, "lr": 9.073291274131621e-05} +{"train_loss": 0.019151490181684494, "grad_norm": 0.05486501008272171, "global_step": 10437, "epoch": 98, "lr": 9.073119385903003e-05} +{"train_loss": 0.01737864874303341, "grad_norm": 0.10783224552869797, "global_step": 10438, "epoch": 98, "lr": 9.072947483363216e-05} +{"train_loss": 0.0177911389619112, "grad_norm": 0.04843701049685478, "global_step": 10439, "epoch": 98, "lr": 9.072775566512863e-05} +{"train_loss": 0.015543203800916672, "grad_norm": 0.05624772235751152, "global_step": 10440, "epoch": 98, "lr": 9.072603635352548e-05} +{"train_loss": 0.016210058704018593, "grad_norm": 0.049768660217523575, "global_step": 10441, "epoch": 98, "lr": 9.072431689882875e-05} +{"train_loss": 0.020886048674583435, "grad_norm": 0.07949432730674744, "global_step": 10442, "epoch": 98, "lr": 9.072259730104447e-05} +{"train_loss": 0.015990126878023148, "grad_norm": 0.10566453635692596, "global_step": 10443, "epoch": 98, "lr": 9.072087756017868e-05} +{"train_loss": 0.02093787118792534, "grad_norm": 0.05539563670754433, "global_step": 10444, "epoch": 98, "lr": 9.071915767623746e-05} +{"train_loss": 0.020189793780446053, "grad_norm": 0.06228101626038551, "global_step": 10445, "epoch": 98, "lr": 9.071743764922683e-05} +{"train_loss": 0.01499750092625618, "grad_norm": 0.05360529571771622, "global_step": 10446, "epoch": 98, "lr": 9.07157174791528e-05} +{"train_loss": 0.015024688094854355, "grad_norm": 0.058268073946237564, "global_step": 10447, "epoch": 98, "lr": 9.071399716602146e-05} +{"train_loss": 0.0162799134850502, "grad_norm": 0.09972792118787766, "global_step": 10448, "epoch": 98, "lr": 9.071227670983886e-05} +{"train_loss": 0.01756366342306137, "grad_norm": 0.08745966851711273, "global_step": 10449, "epoch": 98, "lr": 9.071055611061102e-05} +{"train_loss": 0.016855310648679733, "grad_norm": 0.0545407310128212, "global_step": 10450, "epoch": 98, "lr": 9.070883536834398e-05} +{"train_loss": 0.018102634698152542, "grad_norm": 0.06267343461513519, "global_step": 10451, "epoch": 98, "lr": 9.070711448304378e-05} +{"train_loss": 0.016897860914468765, "grad_norm": 0.06664040684700012, "global_step": 10452, "epoch": 98, "lr": 9.070539345471651e-05} +{"train_loss": 0.018674850463867188, "grad_norm": 0.07552418857812881, "global_step": 10453, "epoch": 98, "lr": 9.070367228336818e-05} +{"train_loss": 0.016417691484093666, "grad_norm": 0.062172528356313705, "global_step": 10454, "epoch": 98, "lr": 9.070195096900485e-05} +{"train_loss": 0.01938142441213131, "grad_norm": 0.11068777740001678, "global_step": 10455, "epoch": 98, "lr": 9.070022951163257e-05} +{"train_loss": 0.01660788804292679, "grad_norm": 0.06236279010772705, "global_step": 10456, "epoch": 98, "lr": 9.069850791125738e-05} +{"train_loss": 0.017238982021808624, "grad_norm": 0.0500052347779274, "global_step": 10457, "epoch": 98, "lr": 9.069678616788532e-05} +{"train_loss": 0.01803336665034294, "grad_norm": 0.052474480122327805, "global_step": 10458, "epoch": 98, "lr": 9.069506428152246e-05} +{"train_loss": 0.017444036900997162, "grad_norm": 0.04868937283754349, "global_step": 10459, "epoch": 98, "lr": 9.069334225217485e-05} +{"train_loss": 0.01774270087480545, "grad_norm": 0.0806138888001442, "global_step": 10460, "epoch": 98, "lr": 9.069162007984853e-05} +{"train_loss": 0.017020391300320625, "grad_norm": 0.0751214474439621, "global_step": 10461, "epoch": 98, "lr": 9.068989776454955e-05} +{"train_loss": 0.01616712473332882, "grad_norm": 0.056226544082164764, "global_step": 10462, "epoch": 98, "lr": 9.068817530628396e-05} +{"train_loss": 0.017699364572763443, "grad_norm": 0.045850444585084915, "global_step": 10463, "epoch": 98, "lr": 9.068645270505781e-05} +{"train_loss": 0.018074143677949905, "grad_norm": 0.07695327699184418, "global_step": 10464, "epoch": 98, "lr": 9.068472996087719e-05} +{"train_loss": 0.019325964152812958, "grad_norm": 0.09813527762889862, "global_step": 10465, "epoch": 98, "lr": 9.068300707374808e-05} +{"train_loss": 0.01435651071369648, "grad_norm": 0.09173358976840973, "global_step": 10466, "epoch": 98, "lr": 9.06812840436766e-05} +{"train_loss": 0.015551946125924587, "grad_norm": 0.07543237507343292, "global_step": 10467, "epoch": 98, "lr": 9.067956087066877e-05} +{"train_loss": 0.017941037192940712, "grad_norm": 0.06010887399315834, "global_step": 10468, "epoch": 98, "lr": 9.067783755473068e-05} +{"train_loss": 0.018098697066307068, "grad_norm": 0.07311941683292389, "global_step": 10469, "epoch": 98, "lr": 9.067611409586833e-05} +{"train_loss": 0.017029542475938797, "grad_norm": 0.05332513153553009, "global_step": 10470, "epoch": 98, "lr": 9.067439049408781e-05} +{"train_loss": 0.017594752833247185, "grad_norm": 0.08465588837862015, "global_step": 10471, "epoch": 98, "lr": 9.067266674939519e-05} +{"train_loss": 0.015881596133112907, "grad_norm": 0.10687743127346039, "global_step": 10472, "epoch": 98, "lr": 9.067094286179649e-05} +{"train_loss": 0.021268673241138458, "grad_norm": 0.06244860216975212, "global_step": 10473, "epoch": 98, "lr": 9.066921883129779e-05} +{"train_loss": 0.01778198592364788, "grad_norm": 0.15098172426223755, "global_step": 10474, "epoch": 98, "lr": 9.066749465790514e-05} +{"train_loss": 0.017730794847011566, "grad_norm": 0.08836504071950912, "global_step": 10475, "epoch": 98, "lr": 9.06657703416246e-05} +{"train_loss": 0.016667433083057404, "grad_norm": 0.06057732552289963, "global_step": 10476, "epoch": 98, "lr": 9.066404588246223e-05} +{"train_loss": 0.016272326931357384, "grad_norm": 0.06895335763692856, "global_step": 10477, "epoch": 98, "lr": 9.066232128042407e-05} +{"train_loss": 0.02005448192358017, "grad_norm": 0.050463177263736725, "global_step": 10478, "epoch": 98, "lr": 9.066059653551622e-05} +{"train_loss": 0.015587128698825836, "grad_norm": 0.09209088981151581, "global_step": 10479, "epoch": 98, "lr": 9.065887164774469e-05} +{"train_loss": 0.014829875901341438, "grad_norm": 0.07294722646474838, "global_step": 10480, "epoch": 98, "lr": 9.06571466171156e-05} +{"train_loss": 0.01603674702346325, "grad_norm": 0.1537911742925644, "global_step": 10481, "epoch": 98, "lr": 9.065542144363495e-05} +{"train_loss": 0.017275623977184296, "grad_norm": 0.05355781316757202, "global_step": 10482, "epoch": 98, "lr": 9.065369612730885e-05} +{"train_loss": 0.017803367227315903, "grad_norm": 0.04663899913430214, "global_step": 10483, "epoch": 98, "lr": 9.065197066814333e-05} +{"train_loss": 0.01676822081208229, "grad_norm": 0.13120421767234802, "global_step": 10484, "epoch": 98, "lr": 9.065024506614447e-05} +{"train_loss": 0.0190960094332695, "grad_norm": 0.08581804484128952, "global_step": 10485, "epoch": 98, "lr": 9.064851932131832e-05} +{"train_loss": 0.017941730096936226, "grad_norm": 0.0595286563038826, "global_step": 10486, "epoch": 98, "lr": 9.064679343367095e-05} +{"train_loss": 0.017527444288134575, "grad_norm": 0.08814573287963867, "global_step": 10487, "epoch": 98, "lr": 9.064506740320845e-05} +{"train_loss": 0.02012249454855919, "grad_norm": 0.09011069685220718, "global_step": 10488, "epoch": 98, "lr": 9.064334122993682e-05} +{"train_loss": 0.016972383484244347, "grad_norm": 0.04878922924399376, "global_step": 10489, "epoch": 98, "lr": 9.064161491386218e-05} +{"train_loss": 0.02101587876677513, "grad_norm": 0.05949706584215164, "global_step": 10490, "epoch": 98, "lr": 9.063988845499058e-05} +{"train_loss": 0.018915174528956413, "grad_norm": 0.09758783876895905, "global_step": 10491, "epoch": 98, "lr": 9.06381618533281e-05} +{"train_loss": 0.015754595398902893, "grad_norm": 0.12486226111650467, "global_step": 10492, "epoch": 98, "lr": 9.063643510888077e-05} +{"train_loss": 0.017578602135884313, "grad_norm": 0.13337767124176025, "global_step": 10493, "epoch": 98, "lr": 9.06347082216547e-05, "val_loss": 0.019092019647359848} +{"train_loss": 0.018233973532915115, "grad_norm": 0.07581724971532822, "global_step": 10494, "epoch": 99, "lr": 9.063298119165593e-05} +{"train_loss": 0.01779637299478054, "grad_norm": 0.08262369781732559, "global_step": 10495, "epoch": 99, "lr": 9.063125401889054e-05} +{"train_loss": 0.015629909932613373, "grad_norm": 0.05696159601211548, "global_step": 10496, "epoch": 99, "lr": 9.062952670336457e-05} +{"train_loss": 0.018110375851392746, "grad_norm": 0.05256100744009018, "global_step": 10497, "epoch": 99, "lr": 9.062779924508414e-05} +{"train_loss": 0.018557023257017136, "grad_norm": 0.07619598507881165, "global_step": 10498, "epoch": 99, "lr": 9.062607164405528e-05} +{"train_loss": 0.015235202386975288, "grad_norm": 0.11528737097978592, "global_step": 10499, "epoch": 99, "lr": 9.062434390028407e-05} +{"train_loss": 0.019706575199961662, "grad_norm": 0.06088050827383995, "global_step": 10500, "epoch": 99, "lr": 9.062261601377658e-05} +{"train_loss": 0.016879664734005928, "grad_norm": 0.04476376250386238, "global_step": 10501, "epoch": 99, "lr": 9.06208879845389e-05} +{"train_loss": 0.015784351155161858, "grad_norm": 0.06438182294368744, "global_step": 10502, "epoch": 99, "lr": 9.061915981257707e-05} +{"train_loss": 0.016135193407535553, "grad_norm": 0.08312720060348511, "global_step": 10503, "epoch": 99, "lr": 9.061743149789718e-05} +{"train_loss": 0.01638225093483925, "grad_norm": 0.08942528069019318, "global_step": 10504, "epoch": 99, "lr": 9.06157030405053e-05} +{"train_loss": 0.016010258346796036, "grad_norm": 0.07927939295768738, "global_step": 10505, "epoch": 99, "lr": 9.06139744404075e-05} +{"train_loss": 0.018381208181381226, "grad_norm": 0.1175348162651062, "global_step": 10506, "epoch": 99, "lr": 9.061224569760986e-05} +{"train_loss": 0.018538275733590126, "grad_norm": 0.04774732515215874, "global_step": 10507, "epoch": 99, "lr": 9.061051681211845e-05} +{"train_loss": 0.016691904515028, "grad_norm": 0.07267002761363983, "global_step": 10508, "epoch": 99, "lr": 9.060878778393936e-05} +{"train_loss": 0.01822727359831333, "grad_norm": 0.11531340330839157, "global_step": 10509, "epoch": 99, "lr": 9.060705861307862e-05} +{"train_loss": 0.014914426952600479, "grad_norm": 0.07829350978136063, "global_step": 10510, "epoch": 99, "lr": 9.060532929954235e-05} +{"train_loss": 0.019323859363794327, "grad_norm": 0.1611110121011734, "global_step": 10511, "epoch": 99, "lr": 9.060359984333662e-05} +{"train_loss": 0.018533850088715553, "grad_norm": 0.06999283283948898, "global_step": 10512, "epoch": 99, "lr": 9.06018702444675e-05} +{"train_loss": 0.017692431807518005, "grad_norm": 0.05473357439041138, "global_step": 10513, "epoch": 99, "lr": 9.060014050294105e-05} +{"train_loss": 0.017947902902960777, "grad_norm": 0.05325199291110039, "global_step": 10514, "epoch": 99, "lr": 9.059841061876336e-05} +{"train_loss": 0.018454862758517265, "grad_norm": 0.09146849811077118, "global_step": 10515, "epoch": 99, "lr": 9.059668059194052e-05} +{"train_loss": 0.019724726676940918, "grad_norm": 0.08536418527364731, "global_step": 10516, "epoch": 99, "lr": 9.059495042247861e-05} +{"train_loss": 0.017158813774585724, "grad_norm": 0.057507313787937164, "global_step": 10517, "epoch": 99, "lr": 9.059322011038369e-05} +{"train_loss": 0.01781240850687027, "grad_norm": 0.065010204911232, "global_step": 10518, "epoch": 99, "lr": 9.059148965566183e-05} +{"train_loss": 0.01844564452767372, "grad_norm": 0.06028306484222412, "global_step": 10519, "epoch": 99, "lr": 9.058975905831914e-05} +{"train_loss": 0.01698845624923706, "grad_norm": 0.06971368938684464, "global_step": 10520, "epoch": 99, "lr": 9.05880283183617e-05} +{"train_loss": 0.015458626672625542, "grad_norm": 0.07386735826730728, "global_step": 10521, "epoch": 99, "lr": 9.058629743579556e-05} +{"train_loss": 0.017481572926044464, "grad_norm": 0.050833966583013535, "global_step": 10522, "epoch": 99, "lr": 9.058456641062684e-05} +{"train_loss": 0.016978450119495392, "grad_norm": 0.05022251233458519, "global_step": 10523, "epoch": 99, "lr": 9.058283524286159e-05} +{"train_loss": 0.01705385372042656, "grad_norm": 0.06591625511646271, "global_step": 10524, "epoch": 99, "lr": 9.058110393250593e-05} +{"train_loss": 0.01802397333085537, "grad_norm": 0.10649000108242035, "global_step": 10525, "epoch": 99, "lr": 9.05793724795659e-05} +{"train_loss": 0.016897063702344894, "grad_norm": 0.07414939254522324, "global_step": 10526, "epoch": 99, "lr": 9.057764088404761e-05} +{"train_loss": 0.018947836011648178, "grad_norm": 0.0568668507039547, "global_step": 10527, "epoch": 99, "lr": 9.057590914595713e-05} +{"train_loss": 0.016372360289096832, "grad_norm": 0.06965528428554535, "global_step": 10528, "epoch": 99, "lr": 9.057417726530055e-05} +{"train_loss": 0.01637880876660347, "grad_norm": 0.04963740333914757, "global_step": 10529, "epoch": 99, "lr": 9.057244524208395e-05} +{"train_loss": 0.02045420929789543, "grad_norm": 0.07322833687067032, "global_step": 10530, "epoch": 99, "lr": 9.057071307631345e-05} +{"train_loss": 0.018640143796801567, "grad_norm": 0.13564445078372955, "global_step": 10531, "epoch": 99, "lr": 9.056898076799507e-05} +{"train_loss": 0.018448567017912865, "grad_norm": 0.09086443483829498, "global_step": 10532, "epoch": 99, "lr": 9.056724831713495e-05} +{"train_loss": 0.016973445191979408, "grad_norm": 0.0543978177011013, "global_step": 10533, "epoch": 99, "lr": 9.056551572373917e-05} +{"train_loss": 0.017773479223251343, "grad_norm": 0.07724016159772873, "global_step": 10534, "epoch": 99, "lr": 9.056378298781381e-05} +{"train_loss": 0.017276447266340256, "grad_norm": 0.06265658885240555, "global_step": 10535, "epoch": 99, "lr": 9.056205010936495e-05} +{"train_loss": 0.018536213785409927, "grad_norm": 0.05377131700515747, "global_step": 10536, "epoch": 99, "lr": 9.05603170883987e-05} +{"train_loss": 0.016952652484178543, "grad_norm": 0.051587872207164764, "global_step": 10537, "epoch": 99, "lr": 9.055858392492112e-05} +{"train_loss": 0.0194343701004982, "grad_norm": 0.14658766984939575, "global_step": 10538, "epoch": 99, "lr": 9.05568506189383e-05} +{"train_loss": 0.018999096006155014, "grad_norm": 0.04594544321298599, "global_step": 10539, "epoch": 99, "lr": 9.055511717045638e-05} +{"train_loss": 0.017454765737056732, "grad_norm": 0.15162448585033417, "global_step": 10540, "epoch": 99, "lr": 9.055338357948138e-05} +{"train_loss": 0.016557376831769943, "grad_norm": 0.08648046106100082, "global_step": 10541, "epoch": 99, "lr": 9.055164984601945e-05} +{"train_loss": 0.014313642866909504, "grad_norm": 0.049627628177404404, "global_step": 10542, "epoch": 99, "lr": 9.054991597007665e-05} +{"train_loss": 0.01706182211637497, "grad_norm": 0.04983799532055855, "global_step": 10543, "epoch": 99, "lr": 9.054818195165909e-05} +{"train_loss": 0.015569099225103855, "grad_norm": 0.06402241438627243, "global_step": 10544, "epoch": 99, "lr": 9.054644779077284e-05} +{"train_loss": 0.016937291249632835, "grad_norm": 0.12731166183948517, "global_step": 10545, "epoch": 99, "lr": 9.054471348742399e-05} +{"train_loss": 0.018222561106085777, "grad_norm": 0.04894723370671272, "global_step": 10546, "epoch": 99, "lr": 9.054297904161868e-05} +{"train_loss": 0.02062731236219406, "grad_norm": 0.07585655152797699, "global_step": 10547, "epoch": 99, "lr": 9.054124445336296e-05} +{"train_loss": 0.01655004173517227, "grad_norm": 0.04917798563838005, "global_step": 10548, "epoch": 99, "lr": 9.053950972266294e-05} +{"train_loss": 0.01840389147400856, "grad_norm": 0.08210211247205734, "global_step": 10549, "epoch": 99, "lr": 9.05377748495247e-05} +{"train_loss": 0.016444850713014603, "grad_norm": 0.03754015639424324, "global_step": 10550, "epoch": 99, "lr": 9.053603983395437e-05} +{"train_loss": 0.01690797507762909, "grad_norm": 0.05591559037566185, "global_step": 10551, "epoch": 99, "lr": 9.053430467595803e-05} +{"train_loss": 0.01682342402637005, "grad_norm": 0.045233577489852905, "global_step": 10552, "epoch": 99, "lr": 9.053256937554176e-05} +{"train_loss": 0.01775314286351204, "grad_norm": 0.073126882314682, "global_step": 10553, "epoch": 99, "lr": 9.053083393271164e-05} +{"train_loss": 0.015685617923736572, "grad_norm": 0.054455894976854324, "global_step": 10554, "epoch": 99, "lr": 9.052909834747383e-05} +{"train_loss": 0.01737136021256447, "grad_norm": 0.08563865721225739, "global_step": 10555, "epoch": 99, "lr": 9.052736261983438e-05} +{"train_loss": 0.01723572239279747, "grad_norm": 0.07188355922698975, "global_step": 10556, "epoch": 99, "lr": 9.05256267497994e-05} +{"train_loss": 0.017050351947546005, "grad_norm": 0.05874974653124809, "global_step": 10557, "epoch": 99, "lr": 9.052389073737501e-05} +{"train_loss": 0.01538270153105259, "grad_norm": 0.09054090827703476, "global_step": 10558, "epoch": 99, "lr": 9.052215458256729e-05} +{"train_loss": 0.017381396144628525, "grad_norm": 0.11364543437957764, "global_step": 10559, "epoch": 99, "lr": 9.052041828538233e-05} +{"train_loss": 0.020167343318462372, "grad_norm": 0.08857591450214386, "global_step": 10560, "epoch": 99, "lr": 9.051868184582625e-05} +{"train_loss": 0.020459463819861412, "grad_norm": 0.0926685482263565, "global_step": 10561, "epoch": 99, "lr": 9.051694526390513e-05} +{"train_loss": 0.016211682930588722, "grad_norm": 0.04468623921275139, "global_step": 10562, "epoch": 99, "lr": 9.05152085396251e-05} +{"train_loss": 0.01888657733798027, "grad_norm": 0.05975252017378807, "global_step": 10563, "epoch": 99, "lr": 9.051347167299223e-05} +{"train_loss": 0.017379730939865112, "grad_norm": 0.10389424860477448, "global_step": 10564, "epoch": 99, "lr": 9.051173466401266e-05} +{"train_loss": 0.014401462860405445, "grad_norm": 0.07416459172964096, "global_step": 10565, "epoch": 99, "lr": 9.050999751269245e-05} +{"train_loss": 0.020544255152344704, "grad_norm": 0.08005163073539734, "global_step": 10566, "epoch": 99, "lr": 9.050826021903775e-05} +{"train_loss": 0.014985006302595139, "grad_norm": 0.0732024684548378, "global_step": 10567, "epoch": 99, "lr": 9.050652278305463e-05} +{"train_loss": 0.018410664051771164, "grad_norm": 0.10181893408298492, "global_step": 10568, "epoch": 99, "lr": 9.050478520474919e-05} +{"train_loss": 0.019253570586442947, "grad_norm": 0.11702901124954224, "global_step": 10569, "epoch": 99, "lr": 9.050304748412756e-05} +{"train_loss": 0.01599276065826416, "grad_norm": 0.14112050831317902, "global_step": 10570, "epoch": 99, "lr": 9.050130962119584e-05} +{"train_loss": 0.019677091389894485, "grad_norm": 0.05854266136884689, "global_step": 10571, "epoch": 99, "lr": 9.049957161596012e-05} +{"train_loss": 0.015051327645778656, "grad_norm": 0.1348785161972046, "global_step": 10572, "epoch": 99, "lr": 9.049783346842653e-05} +{"train_loss": 0.017167720943689346, "grad_norm": 0.04828078672289848, "global_step": 10573, "epoch": 99, "lr": 9.049609517860117e-05} +{"train_loss": 0.01742802932858467, "grad_norm": 0.18769633769989014, "global_step": 10574, "epoch": 99, "lr": 9.049435674649013e-05} +{"train_loss": 0.014570298604667187, "grad_norm": 0.053376104682683945, "global_step": 10575, "epoch": 99, "lr": 9.049261817209953e-05} +{"train_loss": 0.016592074185609818, "grad_norm": 0.08802670240402222, "global_step": 10576, "epoch": 99, "lr": 9.04908794554355e-05} +{"train_loss": 0.016662541776895523, "grad_norm": 0.052912548184394836, "global_step": 10577, "epoch": 99, "lr": 9.04891405965041e-05} +{"train_loss": 0.020586630329489708, "grad_norm": 0.07649661600589752, "global_step": 10578, "epoch": 99, "lr": 9.048740159531148e-05} +{"train_loss": 0.01879202388226986, "grad_norm": 0.06354794651269913, "global_step": 10579, "epoch": 99, "lr": 9.048566245186375e-05} +{"train_loss": 0.017832405865192413, "grad_norm": 0.0515131875872612, "global_step": 10580, "epoch": 99, "lr": 9.048392316616701e-05} +{"train_loss": 0.018872488290071487, "grad_norm": 0.11027829349040985, "global_step": 10581, "epoch": 99, "lr": 9.048218373822734e-05} +{"train_loss": 0.01741742715239525, "grad_norm": 0.1371670812368393, "global_step": 10582, "epoch": 99, "lr": 9.048044416805092e-05} +{"train_loss": 0.01705101691186428, "grad_norm": 0.04842839017510414, "global_step": 10583, "epoch": 99, "lr": 9.04787044556438e-05} +{"train_loss": 0.01944763958454132, "grad_norm": 0.08327656984329224, "global_step": 10584, "epoch": 99, "lr": 9.047696460101211e-05} +{"train_loss": 0.017456375062465668, "grad_norm": 0.08775673061609268, "global_step": 10585, "epoch": 99, "lr": 9.047522460416197e-05} +{"train_loss": 0.017405156046152115, "grad_norm": 0.07443089783191681, "global_step": 10586, "epoch": 99, "lr": 9.047348446509952e-05} +{"train_loss": 0.015854407101869583, "grad_norm": 0.07001940160989761, "global_step": 10587, "epoch": 99, "lr": 9.047174418383083e-05} +{"train_loss": 0.017400365322828293, "grad_norm": 0.07486627995967865, "global_step": 10588, "epoch": 99, "lr": 9.047000376036204e-05} +{"train_loss": 0.01641286164522171, "grad_norm": 0.060394737869501114, "global_step": 10589, "epoch": 99, "lr": 9.046826319469925e-05} +{"train_loss": 0.01615084335207939, "grad_norm": 0.05699927732348442, "global_step": 10590, "epoch": 99, "lr": 9.046652248684857e-05} +{"train_loss": 0.017791589722037315, "grad_norm": 0.06478238850831985, "global_step": 10591, "epoch": 99, "lr": 9.046478163681614e-05} +{"train_loss": 0.016158033162355423, "grad_norm": 0.1050705537199974, "global_step": 10592, "epoch": 99, "lr": 9.046304064460807e-05} +{"train_loss": 0.019258614629507065, "grad_norm": 0.10384564846754074, "global_step": 10593, "epoch": 99, "lr": 9.046129951023046e-05} +{"train_loss": 0.020059464499354362, "grad_norm": 0.058580439537763596, "global_step": 10594, "epoch": 99, "lr": 9.045955823368945e-05} +{"train_loss": 0.02031812071800232, "grad_norm": 0.06671367585659027, "global_step": 10595, "epoch": 99, "lr": 9.045781681499115e-05} +{"train_loss": 0.01686442829668522, "grad_norm": 0.08903802186250687, "global_step": 10596, "epoch": 99, "lr": 9.045607525414165e-05} +{"train_loss": 0.01726234331727028, "grad_norm": 0.06044655293226242, "global_step": 10597, "epoch": 99, "lr": 9.045433355114714e-05} +{"train_loss": 0.02078455127775669, "grad_norm": 0.0634392723441124, "global_step": 10598, "epoch": 99, "lr": 9.045259170601366e-05} +{"train_loss": 0.017598253903242777, "grad_norm": 0.1105266585946083, "global_step": 10599, "epoch": 99, "lr": 9.045084971874738e-05, "val_loss": 0.019108187407255173} +{"train_loss": 0.018972009420394897, "grad_norm": 0.07573944330215454, "global_step": 10600, "epoch": 100, "lr": 9.044910758935439e-05} +{"train_loss": 0.017242908477783203, "grad_norm": 0.09520050883293152, "global_step": 10601, "epoch": 100, "lr": 9.044736531784085e-05} +{"train_loss": 0.017462242394685745, "grad_norm": 0.06965655833482742, "global_step": 10602, "epoch": 100, "lr": 9.044562290421284e-05} +{"train_loss": 0.020093286409974098, "grad_norm": 0.07488837838172913, "global_step": 10603, "epoch": 100, "lr": 9.04438803484765e-05} +{"train_loss": 0.018474537879228592, "grad_norm": 0.062324900180101395, "global_step": 10604, "epoch": 100, "lr": 9.044213765063797e-05} +{"train_loss": 0.017100537195801735, "grad_norm": 0.11209350824356079, "global_step": 10605, "epoch": 100, "lr": 9.044039481070334e-05} +{"train_loss": 0.017400264739990234, "grad_norm": 0.07317159324884415, "global_step": 10606, "epoch": 100, "lr": 9.043865182867876e-05} +{"train_loss": 0.01590121164917946, "grad_norm": 0.1411670744419098, "global_step": 10607, "epoch": 100, "lr": 9.043690870457034e-05} +{"train_loss": 0.01859925501048565, "grad_norm": 0.07768566161394119, "global_step": 10608, "epoch": 100, "lr": 9.043516543838422e-05} +{"train_loss": 0.020962286740541458, "grad_norm": 0.07181579619646072, "global_step": 10609, "epoch": 100, "lr": 9.04334220301265e-05} +{"train_loss": 0.017470013350248337, "grad_norm": 0.08951748162508011, "global_step": 10610, "epoch": 100, "lr": 9.043167847980332e-05} +{"train_loss": 0.015547639690339565, "grad_norm": 0.08495571464300156, "global_step": 10611, "epoch": 100, "lr": 9.04299347874208e-05} +{"train_loss": 0.017053432762622833, "grad_norm": 0.11590021103620529, "global_step": 10612, "epoch": 100, "lr": 9.042819095298507e-05} +{"train_loss": 0.01728944107890129, "grad_norm": 0.21321052312850952, "global_step": 10613, "epoch": 100, "lr": 9.042644697650227e-05} +{"train_loss": 0.015646275132894516, "grad_norm": 0.07925581187009811, "global_step": 10614, "epoch": 100, "lr": 9.042470285797851e-05} +{"train_loss": 0.01751403883099556, "grad_norm": 0.0705363005399704, "global_step": 10615, "epoch": 100, "lr": 9.042295859741994e-05} +{"train_loss": 0.01870688796043396, "grad_norm": 0.059899333864450455, "global_step": 10616, "epoch": 100, "lr": 9.042121419483265e-05} +{"train_loss": 0.017735913395881653, "grad_norm": 0.0703396126627922, "global_step": 10617, "epoch": 100, "lr": 9.041946965022281e-05} +{"train_loss": 0.015308869071304798, "grad_norm": 0.04651958495378494, "global_step": 10618, "epoch": 100, "lr": 9.041772496359652e-05} +{"train_loss": 0.014130694791674614, "grad_norm": 0.06711149215698242, "global_step": 10619, "epoch": 100, "lr": 9.041598013495992e-05} +{"train_loss": 0.01524281408637762, "grad_norm": 0.06946473568677902, "global_step": 10620, "epoch": 100, "lr": 9.041423516431915e-05} +{"train_loss": 0.01825866475701332, "grad_norm": 0.1553100347518921, "global_step": 10621, "epoch": 100, "lr": 9.041249005168033e-05} +{"train_loss": 0.017293857410550117, "grad_norm": 0.14843371510505676, "global_step": 10622, "epoch": 100, "lr": 9.041074479704959e-05} +{"train_loss": 0.01706281676888466, "grad_norm": 0.1329973042011261, "global_step": 10623, "epoch": 100, "lr": 9.040899940043308e-05} +{"train_loss": 0.017452605068683624, "grad_norm": 0.0725894346833229, "global_step": 10624, "epoch": 100, "lr": 9.040725386183691e-05} +{"train_loss": 0.016237590461969376, "grad_norm": 0.07099924981594086, "global_step": 10625, "epoch": 100, "lr": 9.040550818126723e-05} +{"train_loss": 0.016256485134363174, "grad_norm": 0.06233024224638939, "global_step": 10626, "epoch": 100, "lr": 9.040376235873016e-05} +{"train_loss": 0.016848210245370865, "grad_norm": 0.08670184016227722, "global_step": 10627, "epoch": 100, "lr": 9.040201639423183e-05} +{"train_loss": 0.017003975808620453, "grad_norm": 0.06950867921113968, "global_step": 10628, "epoch": 100, "lr": 9.04002702877784e-05} +{"train_loss": 0.02021271362900734, "grad_norm": 0.057659950107336044, "global_step": 10629, "epoch": 100, "lr": 9.039852403937599e-05} +{"train_loss": 0.01624452695250511, "grad_norm": 0.0934017077088356, "global_step": 10630, "epoch": 100, "lr": 9.039677764903073e-05} +{"train_loss": 0.021033775061368942, "grad_norm": 0.1547129899263382, "global_step": 10631, "epoch": 100, "lr": 9.039503111674876e-05} +{"train_loss": 0.015727320685982704, "grad_norm": 0.07219402492046356, "global_step": 10632, "epoch": 100, "lr": 9.039328444253623e-05} +{"train_loss": 0.017024226486682892, "grad_norm": 0.0428348109126091, "global_step": 10633, "epoch": 100, "lr": 9.039153762639925e-05} +{"train_loss": 0.017413489520549774, "grad_norm": 0.10994940996170044, "global_step": 10634, "epoch": 100, "lr": 9.038979066834398e-05} +{"train_loss": 0.01941172033548355, "grad_norm": 0.07172854989767075, "global_step": 10635, "epoch": 100, "lr": 9.038804356837654e-05} +{"train_loss": 0.018318697810173035, "grad_norm": 0.056688301265239716, "global_step": 10636, "epoch": 100, "lr": 9.03862963265031e-05} +{"train_loss": 0.01914234086871147, "grad_norm": 0.05029788985848427, "global_step": 10637, "epoch": 100, "lr": 9.038454894272977e-05} +{"train_loss": 0.01754130981862545, "grad_norm": 0.0952380895614624, "global_step": 10638, "epoch": 100, "lr": 9.038280141706268e-05} +{"train_loss": 0.015992864966392517, "grad_norm": 0.12988391518592834, "global_step": 10639, "epoch": 100, "lr": 9.0381053749508e-05} +{"train_loss": 0.015740633010864258, "grad_norm": 0.12105844169855118, "global_step": 10640, "epoch": 100, "lr": 9.037930594007187e-05} +{"train_loss": 0.016249841079115868, "grad_norm": 0.053460847586393356, "global_step": 10641, "epoch": 100, "lr": 9.03775579887604e-05} +{"train_loss": 0.01744813099503517, "grad_norm": 0.10962320864200592, "global_step": 10642, "epoch": 100, "lr": 9.037580989557976e-05} +{"train_loss": 0.019306344911456108, "grad_norm": 0.0554368756711483, "global_step": 10643, "epoch": 100, "lr": 9.037406166053608e-05} +{"train_loss": 0.014234930276870728, "grad_norm": 0.0779503881931305, "global_step": 10644, "epoch": 100, "lr": 9.037231328363552e-05} +{"train_loss": 0.019045762717723846, "grad_norm": 0.08279867470264435, "global_step": 10645, "epoch": 100, "lr": 9.03705647648842e-05} +{"train_loss": 0.01805761456489563, "grad_norm": 0.06877100467681885, "global_step": 10646, "epoch": 100, "lr": 9.036881610428827e-05} +{"train_loss": 0.017177991569042206, "grad_norm": 0.07983162254095078, "global_step": 10647, "epoch": 100, "lr": 9.036706730185387e-05} +{"train_loss": 0.019218647852540016, "grad_norm": 0.14021159708499908, "global_step": 10648, "epoch": 100, "lr": 9.036531835758715e-05} +{"train_loss": 0.019543621689081192, "grad_norm": 0.10633206367492676, "global_step": 10649, "epoch": 100, "lr": 9.036356927149427e-05} +{"train_loss": 0.018359098583459854, "grad_norm": 0.05834193527698517, "global_step": 10650, "epoch": 100, "lr": 9.036182004358136e-05} +{"train_loss": 0.016083959490060806, "grad_norm": 0.06321825832128525, "global_step": 10651, "epoch": 100, "lr": 9.036007067385455e-05} +{"train_loss": 0.017549894750118256, "grad_norm": 0.05344837158918381, "global_step": 10652, "epoch": 100, "lr": 9.035832116232001e-05} +{"train_loss": 0.01966070756316185, "grad_norm": 0.10414295643568039, "global_step": 10653, "epoch": 100, "lr": 9.03565715089839e-05} +{"train_loss": 0.01817299611866474, "grad_norm": 0.07849925756454468, "global_step": 10654, "epoch": 100, "lr": 9.035482171385233e-05} +{"train_loss": 0.01576588861644268, "grad_norm": 0.07309217751026154, "global_step": 10655, "epoch": 100, "lr": 9.035307177693147e-05} +{"train_loss": 0.018997646868228912, "grad_norm": 0.05060398206114769, "global_step": 10656, "epoch": 100, "lr": 9.035132169822747e-05} +{"train_loss": 0.01729845441877842, "grad_norm": 0.08062789589166641, "global_step": 10657, "epoch": 100, "lr": 9.034957147774646e-05} +{"train_loss": 0.018058648332953453, "grad_norm": 0.06484945863485336, "global_step": 10658, "epoch": 100, "lr": 9.034782111549463e-05} +{"train_loss": 0.018255038186907768, "grad_norm": 0.05606989935040474, "global_step": 10659, "epoch": 100, "lr": 9.034607061147807e-05} +{"train_loss": 0.016061628237366676, "grad_norm": 0.10728735476732254, "global_step": 10660, "epoch": 100, "lr": 9.034431996570298e-05} +{"train_loss": 0.020853357389569283, "grad_norm": 0.049308378249406815, "global_step": 10661, "epoch": 100, "lr": 9.03425691781755e-05} +{"train_loss": 0.020234733819961548, "grad_norm": 0.050573982298374176, "global_step": 10662, "epoch": 100, "lr": 9.034081824890177e-05} +{"train_loss": 0.017868373543024063, "grad_norm": 0.06031257286667824, "global_step": 10663, "epoch": 100, "lr": 9.033906717788793e-05} +{"train_loss": 0.017829133197665215, "grad_norm": 0.09643959999084473, "global_step": 10664, "epoch": 100, "lr": 9.033731596514019e-05} +{"train_loss": 0.015878567472100258, "grad_norm": 0.10605908930301666, "global_step": 10665, "epoch": 100, "lr": 9.033556461066464e-05} +{"train_loss": 0.01566784828901291, "grad_norm": 0.07008516788482666, "global_step": 10666, "epoch": 100, "lr": 9.033381311446746e-05} +{"train_loss": 0.016651663929224014, "grad_norm": 0.05885755270719528, "global_step": 10667, "epoch": 100, "lr": 9.033206147655479e-05} +{"train_loss": 0.01857425458729267, "grad_norm": 0.053197525441646576, "global_step": 10668, "epoch": 100, "lr": 9.033030969693281e-05} +{"train_loss": 0.01773078739643097, "grad_norm": 0.059064529836177826, "global_step": 10669, "epoch": 100, "lr": 9.032855777560766e-05} +{"train_loss": 0.019329827278852463, "grad_norm": 0.06550238281488419, "global_step": 10670, "epoch": 100, "lr": 9.032680571258552e-05} +{"train_loss": 0.015019447542726994, "grad_norm": 0.08567003905773163, "global_step": 10671, "epoch": 100, "lr": 9.032505350787248e-05} +{"train_loss": 0.01810755953192711, "grad_norm": 0.056974973529577255, "global_step": 10672, "epoch": 100, "lr": 9.032330116147476e-05} +{"train_loss": 0.01940934732556343, "grad_norm": 0.053900815546512604, "global_step": 10673, "epoch": 100, "lr": 9.032154867339851e-05} +{"train_loss": 0.017794180661439896, "grad_norm": 0.07834000140428543, "global_step": 10674, "epoch": 100, "lr": 9.031979604364983e-05} +{"train_loss": 0.019176840782165527, "grad_norm": 0.050186313688755035, "global_step": 10675, "epoch": 100, "lr": 9.031804327223497e-05} +{"train_loss": 0.016999267041683197, "grad_norm": 0.04826708510518074, "global_step": 10676, "epoch": 100, "lr": 9.031629035916001e-05} +{"train_loss": 0.0171158779412508, "grad_norm": 0.09781854599714279, "global_step": 10677, "epoch": 100, "lr": 9.031453730443115e-05} +{"train_loss": 0.021827975288033485, "grad_norm": 0.07073532044887543, "global_step": 10678, "epoch": 100, "lr": 9.031278410805453e-05} +{"train_loss": 0.01644408330321312, "grad_norm": 0.057333510369062424, "global_step": 10679, "epoch": 100, "lr": 9.031103077003634e-05} +{"train_loss": 0.016780264675617218, "grad_norm": 0.04526742547750473, "global_step": 10680, "epoch": 100, "lr": 9.030927729038269e-05} +{"train_loss": 0.019024159759283066, "grad_norm": 0.08696770668029785, "global_step": 10681, "epoch": 100, "lr": 9.030752366909978e-05} +{"train_loss": 0.02097190171480179, "grad_norm": 0.07007143646478653, "global_step": 10682, "epoch": 100, "lr": 9.030576990619375e-05} +{"train_loss": 0.016232779249548912, "grad_norm": 0.056137826293706894, "global_step": 10683, "epoch": 100, "lr": 9.03040160016708e-05} +{"train_loss": 0.01626008376479149, "grad_norm": 0.09451857209205627, "global_step": 10684, "epoch": 100, "lr": 9.030226195553703e-05} +{"train_loss": 0.015887804329395294, "grad_norm": 0.0891202762722969, "global_step": 10685, "epoch": 100, "lr": 9.030050776779869e-05} +{"train_loss": 0.01807418093085289, "grad_norm": 0.06208624318242073, "global_step": 10686, "epoch": 100, "lr": 9.029875343846184e-05} +{"train_loss": 0.01711425930261612, "grad_norm": 0.06448179483413696, "global_step": 10687, "epoch": 100, "lr": 9.029699896753272e-05} +{"train_loss": 0.019489992409944534, "grad_norm": 0.08596652746200562, "global_step": 10688, "epoch": 100, "lr": 9.029524435501748e-05} +{"train_loss": 0.018271012231707573, "grad_norm": 0.07048610597848892, "global_step": 10689, "epoch": 100, "lr": 9.029348960092225e-05} +{"train_loss": 0.018792325630784035, "grad_norm": 0.10541265457868576, "global_step": 10690, "epoch": 100, "lr": 9.029173470525323e-05} +{"train_loss": 0.014779942110180855, "grad_norm": 0.05196040868759155, "global_step": 10691, "epoch": 100, "lr": 9.02899796680166e-05} +{"train_loss": 0.01939171925187111, "grad_norm": 0.07538534700870514, "global_step": 10692, "epoch": 100, "lr": 9.028822448921847e-05} +{"train_loss": 0.01817432977259159, "grad_norm": 0.054596249014139175, "global_step": 10693, "epoch": 100, "lr": 9.028646916886505e-05} +{"train_loss": 0.013514549471437931, "grad_norm": 0.04173268377780914, "global_step": 10694, "epoch": 100, "lr": 9.02847137069625e-05} +{"train_loss": 0.019336294382810593, "grad_norm": 0.13362428545951843, "global_step": 10695, "epoch": 100, "lr": 9.028295810351699e-05} +{"train_loss": 0.01591235212981701, "grad_norm": 0.04468696936964989, "global_step": 10696, "epoch": 100, "lr": 9.028120235853467e-05} +{"train_loss": 0.017665797844529152, "grad_norm": 0.06113269925117493, "global_step": 10697, "epoch": 100, "lr": 9.027944647202174e-05} +{"train_loss": 0.01697612553834915, "grad_norm": 0.11858809739351273, "global_step": 10698, "epoch": 100, "lr": 9.027769044398435e-05} +{"train_loss": 0.017273124307394028, "grad_norm": 0.04492713883519173, "global_step": 10699, "epoch": 100, "lr": 9.027593427442867e-05} +{"train_loss": 0.015627726912498474, "grad_norm": 0.06365466862916946, "global_step": 10700, "epoch": 100, "lr": 9.027417796336087e-05} +{"train_loss": 0.019191814586520195, "grad_norm": 0.11000366508960724, "global_step": 10701, "epoch": 100, "lr": 9.027242151078712e-05} +{"train_loss": 0.016353115439414978, "grad_norm": 0.08711344748735428, "global_step": 10702, "epoch": 100, "lr": 9.02706649167136e-05} +{"train_loss": 0.016430355608463287, "grad_norm": 0.08891980350017548, "global_step": 10703, "epoch": 100, "lr": 9.026890818114647e-05} +{"train_loss": 0.017368938773870468, "grad_norm": 0.06846166402101517, "global_step": 10704, "epoch": 100, "lr": 9.026715130409192e-05} +{"train_loss": 0.017547554995442898, "grad_norm": 0.10972302407026291, "global_step": 10705, "epoch": 100, "lr": 9.02653942855561e-05, "train_batch_mse_error": 0.015626195818185806, "val_loss": 0.018522679805755615} +{"train_loss": 0.015684697777032852, "grad_norm": 0.0745643824338913, "global_step": 10706, "epoch": 101, "lr": 9.026363712554519e-05} +{"train_loss": 0.016511084511876106, "grad_norm": 0.06304090470075607, "global_step": 10707, "epoch": 101, "lr": 9.02618798240654e-05} +{"train_loss": 0.018363922834396362, "grad_norm": 0.09104689955711365, "global_step": 10708, "epoch": 101, "lr": 9.026012238112285e-05} +{"train_loss": 0.018581705167889595, "grad_norm": 0.06832117587327957, "global_step": 10709, "epoch": 101, "lr": 9.025836479672374e-05} +{"train_loss": 0.015992972999811172, "grad_norm": 0.05516266077756882, "global_step": 10710, "epoch": 101, "lr": 9.025660707087424e-05} +{"train_loss": 0.01836632564663887, "grad_norm": 0.053252074867486954, "global_step": 10711, "epoch": 101, "lr": 9.025484920358054e-05} +{"train_loss": 0.01946420967578888, "grad_norm": 0.09514057636260986, "global_step": 10712, "epoch": 101, "lr": 9.025309119484881e-05} +{"train_loss": 0.018177110701799393, "grad_norm": 0.05371605604887009, "global_step": 10713, "epoch": 101, "lr": 9.025133304468521e-05} +{"train_loss": 0.014875977300107479, "grad_norm": 0.071814626455307, "global_step": 10714, "epoch": 101, "lr": 9.024957475309594e-05} +{"train_loss": 0.015235455706715584, "grad_norm": 0.04632784053683281, "global_step": 10715, "epoch": 101, "lr": 9.024781632008716e-05} +{"train_loss": 0.019388146698474884, "grad_norm": 0.0648157000541687, "global_step": 10716, "epoch": 101, "lr": 9.024605774566506e-05} +{"train_loss": 0.01727369800209999, "grad_norm": 0.049177348613739014, "global_step": 10717, "epoch": 101, "lr": 9.024429902983582e-05} +{"train_loss": 0.018302712589502335, "grad_norm": 0.0871671587228775, "global_step": 10718, "epoch": 101, "lr": 9.024254017260561e-05} +{"train_loss": 0.016493558883666992, "grad_norm": 0.07951583713293076, "global_step": 10719, "epoch": 101, "lr": 9.024078117398062e-05} +{"train_loss": 0.017728757113218307, "grad_norm": 0.05032740905880928, "global_step": 10720, "epoch": 101, "lr": 9.023902203396703e-05} +{"train_loss": 0.016408022493124008, "grad_norm": 0.1197282150387764, "global_step": 10721, "epoch": 101, "lr": 9.0237262752571e-05} +{"train_loss": 0.016791004687547684, "grad_norm": 0.11569707840681076, "global_step": 10722, "epoch": 101, "lr": 9.023550332979874e-05} +{"train_loss": 0.01794823817908764, "grad_norm": 0.06980857998132706, "global_step": 10723, "epoch": 101, "lr": 9.023374376565642e-05} +{"train_loss": 0.016335703432559967, "grad_norm": 0.06184406578540802, "global_step": 10724, "epoch": 101, "lr": 9.02319840601502e-05} +{"train_loss": 0.015026671811938286, "grad_norm": 0.11246058344841003, "global_step": 10725, "epoch": 101, "lr": 9.023022421328632e-05} +{"train_loss": 0.018348319455981255, "grad_norm": 0.0449528731405735, "global_step": 10726, "epoch": 101, "lr": 9.02284642250709e-05} +{"train_loss": 0.017180679365992546, "grad_norm": 0.14846132695674896, "global_step": 10727, "epoch": 101, "lr": 9.022670409551017e-05} +{"train_loss": 0.020060019567608833, "grad_norm": 0.11595011502504349, "global_step": 10728, "epoch": 101, "lr": 9.022494382461029e-05} +{"train_loss": 0.01654946058988571, "grad_norm": 0.06354474276304245, "global_step": 10729, "epoch": 101, "lr": 9.022318341237743e-05} +{"train_loss": 0.01638391613960266, "grad_norm": 0.05622049421072006, "global_step": 10730, "epoch": 101, "lr": 9.022142285881783e-05} +{"train_loss": 0.016945069655776024, "grad_norm": 0.04474299028515816, "global_step": 10731, "epoch": 101, "lr": 9.021966216393763e-05} +{"train_loss": 0.019597403705120087, "grad_norm": 0.1471930891275406, "global_step": 10732, "epoch": 101, "lr": 9.021790132774301e-05} +{"train_loss": 0.01849047653377056, "grad_norm": 0.07392747700214386, "global_step": 10733, "epoch": 101, "lr": 9.02161403502402e-05} +{"train_loss": 0.018368210643529892, "grad_norm": 0.10918143391609192, "global_step": 10734, "epoch": 101, "lr": 9.021437923143534e-05} +{"train_loss": 0.01778888702392578, "grad_norm": 0.11920137703418732, "global_step": 10735, "epoch": 101, "lr": 9.021261797133465e-05} +{"train_loss": 0.01824241131544113, "grad_norm": 0.06879041343927383, "global_step": 10736, "epoch": 101, "lr": 9.021085656994432e-05} +{"train_loss": 0.01692105084657669, "grad_norm": 0.10451507568359375, "global_step": 10737, "epoch": 101, "lr": 9.02090950272705e-05} +{"train_loss": 0.016500890254974365, "grad_norm": 0.16065064072608948, "global_step": 10738, "epoch": 101, "lr": 9.020733334331943e-05} +{"train_loss": 0.01839740015566349, "grad_norm": 0.05402227118611336, "global_step": 10739, "epoch": 101, "lr": 9.020557151809727e-05} +{"train_loss": 0.017461996525526047, "grad_norm": 0.09869799762964249, "global_step": 10740, "epoch": 101, "lr": 9.020380955161021e-05} +{"train_loss": 0.01710815727710724, "grad_norm": 0.13949863612651825, "global_step": 10741, "epoch": 101, "lr": 9.020204744386447e-05} +{"train_loss": 0.013260003179311752, "grad_norm": 0.08891741186380386, "global_step": 10742, "epoch": 101, "lr": 9.02002851948662e-05} +{"train_loss": 0.0157378651201725, "grad_norm": 0.09078366309404373, "global_step": 10743, "epoch": 101, "lr": 9.019852280462161e-05} +{"train_loss": 0.016637124121189117, "grad_norm": 0.10857239365577698, "global_step": 10744, "epoch": 101, "lr": 9.019676027313689e-05} +{"train_loss": 0.018564186990261078, "grad_norm": 0.06079646199941635, "global_step": 10745, "epoch": 101, "lr": 9.019499760041825e-05} +{"train_loss": 0.01724908873438835, "grad_norm": 0.0945533961057663, "global_step": 10746, "epoch": 101, "lr": 9.019323478647186e-05} +{"train_loss": 0.016615111380815506, "grad_norm": 0.04227347671985626, "global_step": 10747, "epoch": 101, "lr": 9.019147183130392e-05} +{"train_loss": 0.02101000025868416, "grad_norm": 0.08704295754432678, "global_step": 10748, "epoch": 101, "lr": 9.018970873492063e-05} +{"train_loss": 0.01850958541035652, "grad_norm": 0.044830333441495895, "global_step": 10749, "epoch": 101, "lr": 9.018794549732819e-05} +{"train_loss": 0.017675161361694336, "grad_norm": 0.12410235404968262, "global_step": 10750, "epoch": 101, "lr": 9.018618211853276e-05} +{"train_loss": 0.018858734518289566, "grad_norm": 0.05597124621272087, "global_step": 10751, "epoch": 101, "lr": 9.018441859854058e-05} +{"train_loss": 0.0170453954488039, "grad_norm": 0.147241473197937, "global_step": 10752, "epoch": 101, "lr": 9.018265493735783e-05} +{"train_loss": 0.01680624857544899, "grad_norm": 0.07112070918083191, "global_step": 10753, "epoch": 101, "lr": 9.01808911349907e-05} +{"train_loss": 0.018971282988786697, "grad_norm": 0.046395592391490936, "global_step": 10754, "epoch": 101, "lr": 9.017912719144541e-05} +{"train_loss": 0.017251841723918915, "grad_norm": 0.10737822949886322, "global_step": 10755, "epoch": 101, "lr": 9.017736310672813e-05} +{"train_loss": 0.019372355192899704, "grad_norm": 0.07071874290704727, "global_step": 10756, "epoch": 101, "lr": 9.017559888084506e-05} +{"train_loss": 0.019541552290320396, "grad_norm": 0.0622430182993412, "global_step": 10757, "epoch": 101, "lr": 9.017383451380241e-05} +{"train_loss": 0.019429046660661697, "grad_norm": 0.11242549866437912, "global_step": 10758, "epoch": 101, "lr": 9.017207000560639e-05} +{"train_loss": 0.015159370377659798, "grad_norm": 0.08915925770998001, "global_step": 10759, "epoch": 101, "lr": 9.017030535626317e-05} +{"train_loss": 0.018159642815589905, "grad_norm": 0.08749737590551376, "global_step": 10760, "epoch": 101, "lr": 9.016854056577897e-05} +{"train_loss": 0.01788049191236496, "grad_norm": 0.07208724319934845, "global_step": 10761, "epoch": 101, "lr": 9.016677563416001e-05} +{"train_loss": 0.01775658316910267, "grad_norm": 0.07224321365356445, "global_step": 10762, "epoch": 101, "lr": 9.016501056141244e-05} +{"train_loss": 0.01600736752152443, "grad_norm": 0.0485236831009388, "global_step": 10763, "epoch": 101, "lr": 9.01632453475425e-05} +{"train_loss": 0.0174383707344532, "grad_norm": 0.10062114894390106, "global_step": 10764, "epoch": 101, "lr": 9.016147999255637e-05} +{"train_loss": 0.018468214198946953, "grad_norm": 0.09281208366155624, "global_step": 10765, "epoch": 101, "lr": 9.015971449646028e-05} +{"train_loss": 0.018410736694931984, "grad_norm": 0.05807948112487793, "global_step": 10766, "epoch": 101, "lr": 9.015794885926042e-05} +{"train_loss": 0.01880558393895626, "grad_norm": 0.09555260837078094, "global_step": 10767, "epoch": 101, "lr": 9.015618308096298e-05} +{"train_loss": 0.01496121659874916, "grad_norm": 0.07767993956804276, "global_step": 10768, "epoch": 101, "lr": 9.01544171615742e-05} +{"train_loss": 0.019173935055732727, "grad_norm": 0.04757092893123627, "global_step": 10769, "epoch": 101, "lr": 9.015265110110023e-05} +{"train_loss": 0.021370364353060722, "grad_norm": 0.06144460663199425, "global_step": 10770, "epoch": 101, "lr": 9.015088489954732e-05} +{"train_loss": 0.017802469432353973, "grad_norm": 0.04958276450634003, "global_step": 10771, "epoch": 101, "lr": 9.014911855692167e-05} +{"train_loss": 0.016503389924764633, "grad_norm": 0.10574223101139069, "global_step": 10772, "epoch": 101, "lr": 9.014735207322946e-05} +{"train_loss": 0.01781979203224182, "grad_norm": 0.12034349888563156, "global_step": 10773, "epoch": 101, "lr": 9.014558544847693e-05} +{"train_loss": 0.01647369936108589, "grad_norm": 0.07180067151784897, "global_step": 10774, "epoch": 101, "lr": 9.014381868267026e-05} +{"train_loss": 0.017667654901742935, "grad_norm": 0.05444910004734993, "global_step": 10775, "epoch": 101, "lr": 9.014205177581567e-05} +{"train_loss": 0.01651342585682869, "grad_norm": 0.044806525111198425, "global_step": 10776, "epoch": 101, "lr": 9.014028472791938e-05} +{"train_loss": 0.017676809802651405, "grad_norm": 0.061797596514225006, "global_step": 10777, "epoch": 101, "lr": 9.013851753898757e-05} +{"train_loss": 0.017093658447265625, "grad_norm": 0.059461452066898346, "global_step": 10778, "epoch": 101, "lr": 9.013675020902647e-05} +{"train_loss": 0.017106592655181885, "grad_norm": 0.10463260114192963, "global_step": 10779, "epoch": 101, "lr": 9.013498273804231e-05} +{"train_loss": 0.01806996390223503, "grad_norm": 0.11522668600082397, "global_step": 10780, "epoch": 101, "lr": 9.013321512604125e-05} +{"train_loss": 0.014221281744539738, "grad_norm": 0.041196275502443314, "global_step": 10781, "epoch": 101, "lr": 9.013144737302953e-05} +{"train_loss": 0.01635390892624855, "grad_norm": 0.048355840146541595, "global_step": 10782, "epoch": 101, "lr": 9.012967947901334e-05} +{"train_loss": 0.01709119975566864, "grad_norm": 0.04374358057975769, "global_step": 10783, "epoch": 101, "lr": 9.012791144399893e-05} +{"train_loss": 0.016975965350866318, "grad_norm": 0.11091113835573196, "global_step": 10784, "epoch": 101, "lr": 9.012614326799249e-05} +{"train_loss": 0.017809007316827774, "grad_norm": 0.09103113412857056, "global_step": 10785, "epoch": 101, "lr": 9.012437495100023e-05} +{"train_loss": 0.020193252712488174, "grad_norm": 0.05538703873753548, "global_step": 10786, "epoch": 101, "lr": 9.012260649302835e-05} +{"train_loss": 0.01771245151758194, "grad_norm": 0.0595061369240284, "global_step": 10787, "epoch": 101, "lr": 9.01208378940831e-05} +{"train_loss": 0.016002243384718895, "grad_norm": 0.0990884080529213, "global_step": 10788, "epoch": 101, "lr": 9.011906915417065e-05} +{"train_loss": 0.015366957522928715, "grad_norm": 0.15025988221168518, "global_step": 10789, "epoch": 101, "lr": 9.011730027329727e-05} +{"train_loss": 0.0193958543241024, "grad_norm": 0.07170727103948593, "global_step": 10790, "epoch": 101, "lr": 9.011553125146913e-05} +{"train_loss": 0.01774480566382408, "grad_norm": 0.05985647812485695, "global_step": 10791, "epoch": 101, "lr": 9.011376208869244e-05} +{"train_loss": 0.02014836110174656, "grad_norm": 0.09296421706676483, "global_step": 10792, "epoch": 101, "lr": 9.011199278497347e-05} +{"train_loss": 0.0159471333026886, "grad_norm": 0.058212101459503174, "global_step": 10793, "epoch": 101, "lr": 9.011022334031838e-05} +{"train_loss": 0.016336524859070778, "grad_norm": 0.06927242875099182, "global_step": 10794, "epoch": 101, "lr": 9.01084537547334e-05} +{"train_loss": 0.018363699316978455, "grad_norm": 0.08610682189464569, "global_step": 10795, "epoch": 101, "lr": 9.010668402822478e-05} +{"train_loss": 0.019833650439977646, "grad_norm": 0.06313382089138031, "global_step": 10796, "epoch": 101, "lr": 9.01049141607987e-05} +{"train_loss": 0.01768716797232628, "grad_norm": 0.05832880362868309, "global_step": 10797, "epoch": 101, "lr": 9.01031441524614e-05} +{"train_loss": 0.015936879441142082, "grad_norm": 0.06195244565606117, "global_step": 10798, "epoch": 101, "lr": 9.010137400321907e-05} +{"train_loss": 0.018411828204989433, "grad_norm": 0.04935989901423454, "global_step": 10799, "epoch": 101, "lr": 9.009960371307798e-05} +{"train_loss": 0.019420871511101723, "grad_norm": 0.0814502090215683, "global_step": 10800, "epoch": 101, "lr": 9.00978332820443e-05} +{"train_loss": 0.01709539070725441, "grad_norm": 0.053565312176942825, "global_step": 10801, "epoch": 101, "lr": 9.00960627101243e-05} +{"train_loss": 0.0164540633559227, "grad_norm": 0.06435193866491318, "global_step": 10802, "epoch": 101, "lr": 9.009429199732414e-05} +{"train_loss": 0.01665458455681801, "grad_norm": 0.09527099877595901, "global_step": 10803, "epoch": 101, "lr": 9.00925211436501e-05} +{"train_loss": 0.019567634910345078, "grad_norm": 0.06712202727794647, "global_step": 10804, "epoch": 101, "lr": 9.009075014910836e-05} +{"train_loss": 0.01706596277654171, "grad_norm": 0.04290525242686272, "global_step": 10805, "epoch": 101, "lr": 9.008897901370517e-05} +{"train_loss": 0.015478156507015228, "grad_norm": 0.049098312854766846, "global_step": 10806, "epoch": 101, "lr": 9.008720773744673e-05} +{"train_loss": 0.016331778839230537, "grad_norm": 0.03900523856282234, "global_step": 10807, "epoch": 101, "lr": 9.008543632033928e-05} +{"train_loss": 0.020760387182235718, "grad_norm": 0.05392187833786011, "global_step": 10808, "epoch": 101, "lr": 9.008366476238904e-05} +{"train_loss": 0.017890378832817078, "grad_norm": 0.05403919145464897, "global_step": 10809, "epoch": 101, "lr": 9.008189306360224e-05} +{"train_loss": 0.016506507992744446, "grad_norm": 0.08743645995855331, "global_step": 10810, "epoch": 101, "lr": 9.008012122398509e-05} +{"train_loss": 0.017509788195690455, "grad_norm": 0.06797154992818832, "global_step": 10811, "epoch": 101, "lr": 9.007834924354383e-05, "val_loss": 0.018493305891752243} +{"train_loss": 0.017692456021904945, "grad_norm": 0.059440046548843384, "global_step": 10812, "epoch": 102, "lr": 9.007657712228468e-05} +{"train_loss": 0.01660797744989395, "grad_norm": 0.061168089509010315, "global_step": 10813, "epoch": 102, "lr": 9.007480486021387e-05} +{"train_loss": 0.018192797899246216, "grad_norm": 0.10804997384548187, "global_step": 10814, "epoch": 102, "lr": 9.007303245733763e-05} +{"train_loss": 0.016715653240680695, "grad_norm": 0.12224239856004715, "global_step": 10815, "epoch": 102, "lr": 9.007125991366217e-05} +{"train_loss": 0.014944436959922314, "grad_norm": 0.05097849294543266, "global_step": 10816, "epoch": 102, "lr": 9.006948722919373e-05} +{"train_loss": 0.020627461373806, "grad_norm": 0.05832776427268982, "global_step": 10817, "epoch": 102, "lr": 9.006771440393853e-05} +{"train_loss": 0.017386283725500107, "grad_norm": 0.06913477182388306, "global_step": 10818, "epoch": 102, "lr": 9.006594143790281e-05} +{"train_loss": 0.015980180352926254, "grad_norm": 0.08172739297151566, "global_step": 10819, "epoch": 102, "lr": 9.00641683310928e-05} +{"train_loss": 0.015327355824410915, "grad_norm": 0.07541164010763168, "global_step": 10820, "epoch": 102, "lr": 9.006239508351472e-05} +{"train_loss": 0.01590132527053356, "grad_norm": 0.07025494426488876, "global_step": 10821, "epoch": 102, "lr": 9.006062169517482e-05} +{"train_loss": 0.019161736592650414, "grad_norm": 0.06482244282960892, "global_step": 10822, "epoch": 102, "lr": 9.005884816607932e-05} +{"train_loss": 0.019208919256925583, "grad_norm": 0.059828732162714005, "global_step": 10823, "epoch": 102, "lr": 9.005707449623443e-05} +{"train_loss": 0.022302348166704178, "grad_norm": 0.1075281873345375, "global_step": 10824, "epoch": 102, "lr": 9.00553006856464e-05} +{"train_loss": 0.017351416870951653, "grad_norm": 0.05446373671293259, "global_step": 10825, "epoch": 102, "lr": 9.005352673432148e-05} +{"train_loss": 0.01965518668293953, "grad_norm": 0.1250111162662506, "global_step": 10826, "epoch": 102, "lr": 9.005175264226589e-05} +{"train_loss": 0.016113800927996635, "grad_norm": 0.06012587994337082, "global_step": 10827, "epoch": 102, "lr": 9.004997840948584e-05} +{"train_loss": 0.019175970926880836, "grad_norm": 0.05068662762641907, "global_step": 10828, "epoch": 102, "lr": 9.00482040359876e-05} +{"train_loss": 0.013832906261086464, "grad_norm": 0.049755655229091644, "global_step": 10829, "epoch": 102, "lr": 9.004642952177737e-05} +{"train_loss": 0.01764724962413311, "grad_norm": 0.051270823925733566, "global_step": 10830, "epoch": 102, "lr": 9.00446548668614e-05} +{"train_loss": 0.020082669332623482, "grad_norm": 0.08226202428340912, "global_step": 10831, "epoch": 102, "lr": 9.004288007124594e-05} +{"train_loss": 0.020298663526773453, "grad_norm": 0.06590744853019714, "global_step": 10832, "epoch": 102, "lr": 9.004110513493722e-05} +{"train_loss": 0.019902965053915977, "grad_norm": 0.0556214265525341, "global_step": 10833, "epoch": 102, "lr": 9.003933005794145e-05} +{"train_loss": 0.014966960996389389, "grad_norm": 0.0731140524148941, "global_step": 10834, "epoch": 102, "lr": 9.00375548402649e-05} +{"train_loss": 0.018854573369026184, "grad_norm": 0.10550683736801147, "global_step": 10835, "epoch": 102, "lr": 9.003577948191379e-05} +{"train_loss": 0.017313342541456223, "grad_norm": 0.05608157441020012, "global_step": 10836, "epoch": 102, "lr": 9.003400398289438e-05} +{"train_loss": 0.01669282838702202, "grad_norm": 0.09430348128080368, "global_step": 10837, "epoch": 102, "lr": 9.003222834321288e-05} +{"train_loss": 0.018606772646307945, "grad_norm": 0.06722227483987808, "global_step": 10838, "epoch": 102, "lr": 9.003045256287552e-05} +{"train_loss": 0.017621349543333054, "grad_norm": 0.12472210079431534, "global_step": 10839, "epoch": 102, "lr": 9.002867664188858e-05} +{"train_loss": 0.017019927501678467, "grad_norm": 0.05235935002565384, "global_step": 10840, "epoch": 102, "lr": 9.002690058025827e-05} +{"train_loss": 0.015747936442494392, "grad_norm": 0.108841672539711, "global_step": 10841, "epoch": 102, "lr": 9.002512437799085e-05} +{"train_loss": 0.0167718306183815, "grad_norm": 0.10768327862024307, "global_step": 10842, "epoch": 102, "lr": 9.002334803509254e-05} +{"train_loss": 0.01724960282444954, "grad_norm": 0.0630660429596901, "global_step": 10843, "epoch": 102, "lr": 9.002157155156958e-05} +{"train_loss": 0.01709512248635292, "grad_norm": 0.08088214695453644, "global_step": 10844, "epoch": 102, "lr": 9.001979492742822e-05} +{"train_loss": 0.013919569551944733, "grad_norm": 0.09372968971729279, "global_step": 10845, "epoch": 102, "lr": 9.001801816267472e-05} +{"train_loss": 0.017171038314700127, "grad_norm": 0.10411220788955688, "global_step": 10846, "epoch": 102, "lr": 9.001624125731531e-05} +{"train_loss": 0.017274176701903343, "grad_norm": 0.04691212251782417, "global_step": 10847, "epoch": 102, "lr": 9.001446421135623e-05} +{"train_loss": 0.015947701409459114, "grad_norm": 0.10754767805337906, "global_step": 10848, "epoch": 102, "lr": 9.001268702480372e-05} +{"train_loss": 0.01825696788728237, "grad_norm": 0.055472344160079956, "global_step": 10849, "epoch": 102, "lr": 9.001090969766403e-05} +{"train_loss": 0.016722138971090317, "grad_norm": 0.05865878239274025, "global_step": 10850, "epoch": 102, "lr": 9.000913222994341e-05} +{"train_loss": 0.018181247636675835, "grad_norm": 0.0756523385643959, "global_step": 10851, "epoch": 102, "lr": 9.00073546216481e-05} +{"train_loss": 0.017856977880001068, "grad_norm": 0.1414794921875, "global_step": 10852, "epoch": 102, "lr": 9.000557687278433e-05} +{"train_loss": 0.01880611479282379, "grad_norm": 0.11064387857913971, "global_step": 10853, "epoch": 102, "lr": 9.000379898335835e-05} +{"train_loss": 0.017074882984161377, "grad_norm": 0.05190441384911537, "global_step": 10854, "epoch": 102, "lr": 9.000202095337644e-05} +{"train_loss": 0.017643822357058525, "grad_norm": 0.07458370923995972, "global_step": 10855, "epoch": 102, "lr": 9.000024278284482e-05} +{"train_loss": 0.019845303148031235, "grad_norm": 0.10243427008390427, "global_step": 10856, "epoch": 102, "lr": 8.999846447176974e-05} +{"train_loss": 0.01715288683772087, "grad_norm": 0.050943441689014435, "global_step": 10857, "epoch": 102, "lr": 8.999668602015745e-05} +{"train_loss": 0.016657810658216476, "grad_norm": 0.07352261245250702, "global_step": 10858, "epoch": 102, "lr": 8.999490742801419e-05} +{"train_loss": 0.017461296170949936, "grad_norm": 0.12169556319713593, "global_step": 10859, "epoch": 102, "lr": 8.999312869534624e-05} +{"train_loss": 0.020773258060216904, "grad_norm": 0.1056215912103653, "global_step": 10860, "epoch": 102, "lr": 8.999134982215982e-05} +{"train_loss": 0.019574299454689026, "grad_norm": 0.05158589780330658, "global_step": 10861, "epoch": 102, "lr": 8.99895708084612e-05} +{"train_loss": 0.020455989986658096, "grad_norm": 0.1252681165933609, "global_step": 10862, "epoch": 102, "lr": 8.99877916542566e-05} +{"train_loss": 0.01925159990787506, "grad_norm": 0.07253807783126831, "global_step": 10863, "epoch": 102, "lr": 8.998601235955228e-05} +{"train_loss": 0.016704823821783066, "grad_norm": 0.053379278630018234, "global_step": 10864, "epoch": 102, "lr": 8.998423292435454e-05} +{"train_loss": 0.01566239446401596, "grad_norm": 0.07251686602830887, "global_step": 10865, "epoch": 102, "lr": 8.998245334866957e-05} +{"train_loss": 0.016883764415979385, "grad_norm": 0.08158419281244278, "global_step": 10866, "epoch": 102, "lr": 8.998067363250365e-05} +{"train_loss": 0.01697550155222416, "grad_norm": 0.048341214656829834, "global_step": 10867, "epoch": 102, "lr": 8.997889377586302e-05} +{"train_loss": 0.01676785945892334, "grad_norm": 0.12508608400821686, "global_step": 10868, "epoch": 102, "lr": 8.997711377875395e-05} +{"train_loss": 0.016753308475017548, "grad_norm": 0.09686389565467834, "global_step": 10869, "epoch": 102, "lr": 8.99753336411827e-05} +{"train_loss": 0.01643526554107666, "grad_norm": 0.10661904513835907, "global_step": 10870, "epoch": 102, "lr": 8.997355336315551e-05} +{"train_loss": 0.01644255220890045, "grad_norm": 0.06054367870092392, "global_step": 10871, "epoch": 102, "lr": 8.997177294467862e-05} +{"train_loss": 0.017978355288505554, "grad_norm": 0.0441727377474308, "global_step": 10872, "epoch": 102, "lr": 8.996999238575832e-05} +{"train_loss": 0.018010880798101425, "grad_norm": 0.09999750554561615, "global_step": 10873, "epoch": 102, "lr": 8.996821168640083e-05} +{"train_loss": 0.019441023468971252, "grad_norm": 0.07074513286352158, "global_step": 10874, "epoch": 102, "lr": 8.996643084661245e-05} +{"train_loss": 0.018802553415298462, "grad_norm": 0.07483319193124771, "global_step": 10875, "epoch": 102, "lr": 8.99646498663994e-05} +{"train_loss": 0.01661285012960434, "grad_norm": 0.10388053953647614, "global_step": 10876, "epoch": 102, "lr": 8.996286874576795e-05} +{"train_loss": 0.014736761339008808, "grad_norm": 0.045073069632053375, "global_step": 10877, "epoch": 102, "lr": 8.996108748472435e-05} +{"train_loss": 0.019402487203478813, "grad_norm": 0.05644155293703079, "global_step": 10878, "epoch": 102, "lr": 8.995930608327489e-05} +{"train_loss": 0.01916421949863434, "grad_norm": 0.10185451805591583, "global_step": 10879, "epoch": 102, "lr": 8.99575245414258e-05} +{"train_loss": 0.017911545932292938, "grad_norm": 0.05113161727786064, "global_step": 10880, "epoch": 102, "lr": 8.995574285918335e-05} +{"train_loss": 0.0198332741856575, "grad_norm": 0.05351058021187782, "global_step": 10881, "epoch": 102, "lr": 8.995396103655377e-05} +{"train_loss": 0.015048468485474586, "grad_norm": 0.07834409177303314, "global_step": 10882, "epoch": 102, "lr": 8.995217907354337e-05} +{"train_loss": 0.015806201845407486, "grad_norm": 0.0669131800532341, "global_step": 10883, "epoch": 102, "lr": 8.995039697015838e-05} +{"train_loss": 0.017268255352973938, "grad_norm": 0.1551680862903595, "global_step": 10884, "epoch": 102, "lr": 8.994861472640507e-05} +{"train_loss": 0.014306178316473961, "grad_norm": 0.05646400526165962, "global_step": 10885, "epoch": 102, "lr": 8.99468323422897e-05} +{"train_loss": 0.018228862434625626, "grad_norm": 0.07446014881134033, "global_step": 10886, "epoch": 102, "lr": 8.994504981781854e-05} +{"train_loss": 0.017487658187747, "grad_norm": 0.04359585791826248, "global_step": 10887, "epoch": 102, "lr": 8.994326715299783e-05} +{"train_loss": 0.016886811703443527, "grad_norm": 0.1245550811290741, "global_step": 10888, "epoch": 102, "lr": 8.994148434783387e-05} +{"train_loss": 0.01795671507716179, "grad_norm": 0.104519322514534, "global_step": 10889, "epoch": 102, "lr": 8.99397014023329e-05} +{"train_loss": 0.016039449721574783, "grad_norm": 0.06229672208428383, "global_step": 10890, "epoch": 102, "lr": 8.993791831650117e-05} +{"train_loss": 0.017163340002298355, "grad_norm": 0.08980275690555573, "global_step": 10891, "epoch": 102, "lr": 8.993613509034498e-05} +{"train_loss": 0.0183935035020113, "grad_norm": 0.1415644735097885, "global_step": 10892, "epoch": 102, "lr": 8.993435172387058e-05} +{"train_loss": 0.01674780435860157, "grad_norm": 0.1110808327794075, "global_step": 10893, "epoch": 102, "lr": 8.993256821708423e-05} +{"train_loss": 0.019190022721886635, "grad_norm": 0.10067766904830933, "global_step": 10894, "epoch": 102, "lr": 8.99307845699922e-05} +{"train_loss": 0.018417809158563614, "grad_norm": 0.058586135506629944, "global_step": 10895, "epoch": 102, "lr": 8.992900078260078e-05} +{"train_loss": 0.018796537071466446, "grad_norm": 0.05434715375304222, "global_step": 10896, "epoch": 102, "lr": 8.992721685491618e-05} +{"train_loss": 0.01983068883419037, "grad_norm": 0.07813116908073425, "global_step": 10897, "epoch": 102, "lr": 8.992543278694471e-05} +{"train_loss": 0.01510524470359087, "grad_norm": 0.055614493787288666, "global_step": 10898, "epoch": 102, "lr": 8.992364857869265e-05} +{"train_loss": 0.01698104292154312, "grad_norm": 0.056473467499017715, "global_step": 10899, "epoch": 102, "lr": 8.992186423016625e-05} +{"train_loss": 0.019473981112241745, "grad_norm": 0.07848314940929413, "global_step": 10900, "epoch": 102, "lr": 8.992007974137178e-05} +{"train_loss": 0.016863111406564713, "grad_norm": 0.11302059143781662, "global_step": 10901, "epoch": 102, "lr": 8.991829511231551e-05} +{"train_loss": 0.014873887412250042, "grad_norm": 0.10391834378242493, "global_step": 10902, "epoch": 102, "lr": 8.991651034300371e-05} +{"train_loss": 0.019041966646909714, "grad_norm": 0.11552788317203522, "global_step": 10903, "epoch": 102, "lr": 8.991472543344264e-05} +{"train_loss": 0.01591476984322071, "grad_norm": 0.0947491005063057, "global_step": 10904, "epoch": 102, "lr": 8.99129403836386e-05} +{"train_loss": 0.016693413257598877, "grad_norm": 0.051939401775598526, "global_step": 10905, "epoch": 102, "lr": 8.991115519359783e-05} +{"train_loss": 0.018976274877786636, "grad_norm": 0.11659099906682968, "global_step": 10906, "epoch": 102, "lr": 8.990936986332662e-05} +{"train_loss": 0.018490750342607498, "grad_norm": 0.05998995527625084, "global_step": 10907, "epoch": 102, "lr": 8.990758439283125e-05} +{"train_loss": 0.01950574666261673, "grad_norm": 0.11419633030891418, "global_step": 10908, "epoch": 102, "lr": 8.990579878211798e-05} +{"train_loss": 0.016657479107379913, "grad_norm": 0.07068859785795212, "global_step": 10909, "epoch": 102, "lr": 8.990401303119308e-05} +{"train_loss": 0.014810672029852867, "grad_norm": 0.05843853950500488, "global_step": 10910, "epoch": 102, "lr": 8.990222714006284e-05} +{"train_loss": 0.01710212044417858, "grad_norm": 0.08828216046094894, "global_step": 10911, "epoch": 102, "lr": 8.990044110873353e-05} +{"train_loss": 0.01649506948888302, "grad_norm": 0.10200183838605881, "global_step": 10912, "epoch": 102, "lr": 8.989865493721141e-05} +{"train_loss": 0.015993375331163406, "grad_norm": 0.06296564638614655, "global_step": 10913, "epoch": 102, "lr": 8.989686862550278e-05} +{"train_loss": 0.01767413131892681, "grad_norm": 0.04775345325469971, "global_step": 10914, "epoch": 102, "lr": 8.98950821736139e-05} +{"train_loss": 0.018139658495783806, "grad_norm": 0.060944490134716034, "global_step": 10915, "epoch": 102, "lr": 8.989329558155106e-05} +{"train_loss": 0.01557801105082035, "grad_norm": 0.1306549310684204, "global_step": 10916, "epoch": 102, "lr": 8.989150884932052e-05} +{"train_loss": 0.01751271098465571, "grad_norm": 0.10530288517475128, "global_step": 10917, "epoch": 102, "lr": 8.988972197692855e-05, "val_loss": 0.01895180158317089} +{"train_loss": 0.016559014096856117, "grad_norm": 0.05492180585861206, "global_step": 10918, "epoch": 103, "lr": 8.988793496438146e-05} +{"train_loss": 0.016520151868462563, "grad_norm": 0.04418821260333061, "global_step": 10919, "epoch": 103, "lr": 8.988614781168551e-05} +{"train_loss": 0.019094636663794518, "grad_norm": 0.0634581595659256, "global_step": 10920, "epoch": 103, "lr": 8.988436051884699e-05} +{"train_loss": 0.016738303005695343, "grad_norm": 0.05052908882498741, "global_step": 10921, "epoch": 103, "lr": 8.988257308587218e-05} +{"train_loss": 0.01839398220181465, "grad_norm": 0.04840155318379402, "global_step": 10922, "epoch": 103, "lr": 8.988078551276733e-05} +{"train_loss": 0.016925331205129623, "grad_norm": 0.05204157903790474, "global_step": 10923, "epoch": 103, "lr": 8.987899779953876e-05} +{"train_loss": 0.016842693090438843, "grad_norm": 0.05373551324009895, "global_step": 10924, "epoch": 103, "lr": 8.987720994619273e-05} +{"train_loss": 0.01737641915678978, "grad_norm": 0.0900653749704361, "global_step": 10925, "epoch": 103, "lr": 8.987542195273553e-05} +{"train_loss": 0.01528174802660942, "grad_norm": 0.050280194729566574, "global_step": 10926, "epoch": 103, "lr": 8.987363381917343e-05} +{"train_loss": 0.01637491211295128, "grad_norm": 0.08328378945589066, "global_step": 10927, "epoch": 103, "lr": 8.987184554551272e-05} +{"train_loss": 0.01594126783311367, "grad_norm": 0.07037170976400375, "global_step": 10928, "epoch": 103, "lr": 8.987005713175968e-05} +{"train_loss": 0.017583627253770828, "grad_norm": 0.09989578276872635, "global_step": 10929, "epoch": 103, "lr": 8.986826857792062e-05} +{"train_loss": 0.016428153961896896, "grad_norm": 0.07406565546989441, "global_step": 10930, "epoch": 103, "lr": 8.986647988400178e-05} +{"train_loss": 0.01622508466243744, "grad_norm": 0.055793892592191696, "global_step": 10931, "epoch": 103, "lr": 8.98646910500095e-05} +{"train_loss": 0.01615496352314949, "grad_norm": 0.07120072096586227, "global_step": 10932, "epoch": 103, "lr": 8.986290207595e-05} +{"train_loss": 0.01928129978477955, "grad_norm": 0.08785481005907059, "global_step": 10933, "epoch": 103, "lr": 8.986111296182961e-05} +{"train_loss": 0.01913621462881565, "grad_norm": 0.04604696109890938, "global_step": 10934, "epoch": 103, "lr": 8.98593237076546e-05} +{"train_loss": 0.01871452108025551, "grad_norm": 0.11134102195501328, "global_step": 10935, "epoch": 103, "lr": 8.985753431343127e-05} +{"train_loss": 0.015266721136868, "grad_norm": 0.0940377488732338, "global_step": 10936, "epoch": 103, "lr": 8.98557447791659e-05} +{"train_loss": 0.01937068998813629, "grad_norm": 0.08221910893917084, "global_step": 10937, "epoch": 103, "lr": 8.985395510486477e-05} +{"train_loss": 0.019272461533546448, "grad_norm": 0.048406489193439484, "global_step": 10938, "epoch": 103, "lr": 8.985216529053416e-05} +{"train_loss": 0.018270833417773247, "grad_norm": 0.05816028267145157, "global_step": 10939, "epoch": 103, "lr": 8.985037533618039e-05} +{"train_loss": 0.01828584633767605, "grad_norm": 0.06133456900715828, "global_step": 10940, "epoch": 103, "lr": 8.984858524180972e-05} +{"train_loss": 0.022115034982562065, "grad_norm": 0.07332780957221985, "global_step": 10941, "epoch": 103, "lr": 8.984679500742845e-05} +{"train_loss": 0.01775268092751503, "grad_norm": 0.09714481234550476, "global_step": 10942, "epoch": 103, "lr": 8.984500463304288e-05} +{"train_loss": 0.018373122438788414, "grad_norm": 0.14250671863555908, "global_step": 10943, "epoch": 103, "lr": 8.984321411865929e-05} +{"train_loss": 0.017640551552176476, "grad_norm": 0.04699242487549782, "global_step": 10944, "epoch": 103, "lr": 8.984142346428397e-05} +{"train_loss": 0.01934923604130745, "grad_norm": 0.04529007524251938, "global_step": 10945, "epoch": 103, "lr": 8.983963266992321e-05} +{"train_loss": 0.015067515894770622, "grad_norm": 0.08314992487430573, "global_step": 10946, "epoch": 103, "lr": 8.983784173558332e-05} +{"train_loss": 0.01622566021978855, "grad_norm": 0.051648396998643875, "global_step": 10947, "epoch": 103, "lr": 8.983605066127056e-05} +{"train_loss": 0.017197856679558754, "grad_norm": 0.11093435436487198, "global_step": 10948, "epoch": 103, "lr": 8.983425944699127e-05} +{"train_loss": 0.01721295714378357, "grad_norm": 0.11762681603431702, "global_step": 10949, "epoch": 103, "lr": 8.98324680927517e-05} +{"train_loss": 0.017084306105971336, "grad_norm": 0.049343451857566833, "global_step": 10950, "epoch": 103, "lr": 8.983067659855816e-05} +{"train_loss": 0.01737404614686966, "grad_norm": 0.0769810825586319, "global_step": 10951, "epoch": 103, "lr": 8.982888496441695e-05} +{"train_loss": 0.015184538438916206, "grad_norm": 0.0460997000336647, "global_step": 10952, "epoch": 103, "lr": 8.982709319033436e-05} +{"train_loss": 0.014450520277023315, "grad_norm": 0.06444037705659866, "global_step": 10953, "epoch": 103, "lr": 8.982530127631667e-05} +{"train_loss": 0.017152685672044754, "grad_norm": 0.05650944635272026, "global_step": 10954, "epoch": 103, "lr": 8.98235092223702e-05} +{"train_loss": 0.015243779867887497, "grad_norm": 0.04709932580590248, "global_step": 10955, "epoch": 103, "lr": 8.982171702850124e-05} +{"train_loss": 0.018093694001436234, "grad_norm": 0.055188558995723724, "global_step": 10956, "epoch": 103, "lr": 8.981992469471608e-05} +{"train_loss": 0.019386643543839455, "grad_norm": 0.05865062028169632, "global_step": 10957, "epoch": 103, "lr": 8.981813222102102e-05} +{"train_loss": 0.019177082926034927, "grad_norm": 0.05595120042562485, "global_step": 10958, "epoch": 103, "lr": 8.981633960742237e-05} +{"train_loss": 0.016784071922302246, "grad_norm": 0.11527678370475769, "global_step": 10959, "epoch": 103, "lr": 8.98145468539264e-05} +{"train_loss": 0.017483586445450783, "grad_norm": 0.0593242347240448, "global_step": 10960, "epoch": 103, "lr": 8.981275396053944e-05} +{"train_loss": 0.01914840191602707, "grad_norm": 0.06434955447912216, "global_step": 10961, "epoch": 103, "lr": 8.981096092726778e-05} +{"train_loss": 0.01755478046834469, "grad_norm": 0.05670137330889702, "global_step": 10962, "epoch": 103, "lr": 8.98091677541177e-05} +{"train_loss": 0.014509016647934914, "grad_norm": 0.07046039402484894, "global_step": 10963, "epoch": 103, "lr": 8.980737444109554e-05} +{"train_loss": 0.01935471221804619, "grad_norm": 0.07529003918170929, "global_step": 10964, "epoch": 103, "lr": 8.980558098820757e-05} +{"train_loss": 0.013256128877401352, "grad_norm": 0.05630480498075485, "global_step": 10965, "epoch": 103, "lr": 8.980378739546009e-05} +{"train_loss": 0.01699651964008808, "grad_norm": 0.06766282767057419, "global_step": 10966, "epoch": 103, "lr": 8.980199366285942e-05} +{"train_loss": 0.016806725412607193, "grad_norm": 0.08432626724243164, "global_step": 10967, "epoch": 103, "lr": 8.980019979041185e-05} +{"train_loss": 0.016777832061052322, "grad_norm": 0.04932870715856552, "global_step": 10968, "epoch": 103, "lr": 8.979840577812369e-05} +{"train_loss": 0.018890339881181717, "grad_norm": 0.12026001513004303, "global_step": 10969, "epoch": 103, "lr": 8.979661162600123e-05} +{"train_loss": 0.016421325504779816, "grad_norm": 0.06309735774993896, "global_step": 10970, "epoch": 103, "lr": 8.97948173340508e-05} +{"train_loss": 0.019449826329946518, "grad_norm": 0.05359154939651489, "global_step": 10971, "epoch": 103, "lr": 8.979302290227868e-05} +{"train_loss": 0.01688128337264061, "grad_norm": 0.10319273918867111, "global_step": 10972, "epoch": 103, "lr": 8.979122833069118e-05} +{"train_loss": 0.019052013754844666, "grad_norm": 0.10627919435501099, "global_step": 10973, "epoch": 103, "lr": 8.978943361929462e-05} +{"train_loss": 0.016966570168733597, "grad_norm": 0.0759912058711052, "global_step": 10974, "epoch": 103, "lr": 8.978763876809529e-05} +{"train_loss": 0.021567711606621742, "grad_norm": 0.06159369274973869, "global_step": 10975, "epoch": 103, "lr": 8.978584377709948e-05} +{"train_loss": 0.01849871501326561, "grad_norm": 0.08425919711589813, "global_step": 10976, "epoch": 103, "lr": 8.978404864631353e-05} +{"train_loss": 0.018051020801067352, "grad_norm": 0.053466252982616425, "global_step": 10977, "epoch": 103, "lr": 8.978225337574376e-05} +{"train_loss": 0.017074106261134148, "grad_norm": 0.12086324393749237, "global_step": 10978, "epoch": 103, "lr": 8.978045796539641e-05} +{"train_loss": 0.014989599585533142, "grad_norm": 0.09305252879858017, "global_step": 10979, "epoch": 103, "lr": 8.977866241527786e-05} +{"train_loss": 0.01655496284365654, "grad_norm": 0.12369097024202347, "global_step": 10980, "epoch": 103, "lr": 8.977686672539438e-05} +{"train_loss": 0.016606371849775314, "grad_norm": 0.10417414456605911, "global_step": 10981, "epoch": 103, "lr": 8.97750708957523e-05} +{"train_loss": 0.015593945980072021, "grad_norm": 0.08115795999765396, "global_step": 10982, "epoch": 103, "lr": 8.977327492635789e-05} +{"train_loss": 0.01698870025575161, "grad_norm": 0.1757495403289795, "global_step": 10983, "epoch": 103, "lr": 8.977147881721752e-05} +{"train_loss": 0.01907915621995926, "grad_norm": 0.05649587884545326, "global_step": 10984, "epoch": 103, "lr": 8.976968256833745e-05} +{"train_loss": 0.01860327646136284, "grad_norm": 0.09154393523931503, "global_step": 10985, "epoch": 103, "lr": 8.976788617972403e-05} +{"train_loss": 0.021053630858659744, "grad_norm": 0.1009315773844719, "global_step": 10986, "epoch": 103, "lr": 8.976608965138353e-05} +{"train_loss": 0.016537735238671303, "grad_norm": 0.13315103948116302, "global_step": 10987, "epoch": 103, "lr": 8.97642929833223e-05} +{"train_loss": 0.017505692318081856, "grad_norm": 0.08048026263713837, "global_step": 10988, "epoch": 103, "lr": 8.976249617554663e-05} +{"train_loss": 0.018727196380496025, "grad_norm": 0.12014255672693253, "global_step": 10989, "epoch": 103, "lr": 8.976069922806286e-05} +{"train_loss": 0.017835251986980438, "grad_norm": 0.18325766921043396, "global_step": 10990, "epoch": 103, "lr": 8.975890214087725e-05} +{"train_loss": 0.014380810782313347, "grad_norm": 0.14461582899093628, "global_step": 10991, "epoch": 103, "lr": 8.975710491399617e-05} +{"train_loss": 0.017641209065914154, "grad_norm": 0.07454045116901398, "global_step": 10992, "epoch": 103, "lr": 8.97553075474259e-05} +{"train_loss": 0.018338292837142944, "grad_norm": 0.06798092275857925, "global_step": 10993, "epoch": 103, "lr": 8.975351004117278e-05} +{"train_loss": 0.015161043033003807, "grad_norm": 0.072640061378479, "global_step": 10994, "epoch": 103, "lr": 8.975171239524311e-05} +{"train_loss": 0.017577949911355972, "grad_norm": 0.0463847853243351, "global_step": 10995, "epoch": 103, "lr": 8.974991460964321e-05} +{"train_loss": 0.01810767874121666, "grad_norm": 0.0985645279288292, "global_step": 10996, "epoch": 103, "lr": 8.974811668437941e-05} +{"train_loss": 0.016022656112909317, "grad_norm": 0.05821974575519562, "global_step": 10997, "epoch": 103, "lr": 8.9746318619458e-05} +{"train_loss": 0.01745649054646492, "grad_norm": 0.06912436336278915, "global_step": 10998, "epoch": 103, "lr": 8.974452041488531e-05} +{"train_loss": 0.015283912420272827, "grad_norm": 0.0365159697830677, "global_step": 10999, "epoch": 103, "lr": 8.974272207066767e-05} +{"train_loss": 0.01858590357005596, "grad_norm": 0.07335029542446136, "global_step": 11000, "epoch": 103, "lr": 8.974092358681139e-05} +{"train_loss": 0.015381598845124245, "grad_norm": 0.10127329081296921, "global_step": 11001, "epoch": 103, "lr": 8.973912496332277e-05} +{"train_loss": 0.019415613263845444, "grad_norm": 0.05388980358839035, "global_step": 11002, "epoch": 103, "lr": 8.973732620020816e-05} +{"train_loss": 0.01591619849205017, "grad_norm": 0.0916428193449974, "global_step": 11003, "epoch": 103, "lr": 8.973552729747385e-05} +{"train_loss": 0.01704188995063305, "grad_norm": 0.11490552127361298, "global_step": 11004, "epoch": 103, "lr": 8.973372825512621e-05} +{"train_loss": 0.018914151936769485, "grad_norm": 0.06292752176523209, "global_step": 11005, "epoch": 103, "lr": 8.973192907317149e-05} +{"train_loss": 0.017709722742438316, "grad_norm": 0.13868528604507446, "global_step": 11006, "epoch": 103, "lr": 8.973012975161608e-05} +{"train_loss": 0.019941985607147217, "grad_norm": 0.05782487988471985, "global_step": 11007, "epoch": 103, "lr": 8.972833029046626e-05} +{"train_loss": 0.01748530939221382, "grad_norm": 0.1437084972858429, "global_step": 11008, "epoch": 103, "lr": 8.972653068972837e-05} +{"train_loss": 0.016011349856853485, "grad_norm": 0.0829143300652504, "global_step": 11009, "epoch": 103, "lr": 8.972473094940873e-05} +{"train_loss": 0.01888321340084076, "grad_norm": 0.06358864903450012, "global_step": 11010, "epoch": 103, "lr": 8.972293106951366e-05} +{"train_loss": 0.017878111451864243, "grad_norm": 0.05374486744403839, "global_step": 11011, "epoch": 103, "lr": 8.972113105004946e-05} +{"train_loss": 0.01776745542883873, "grad_norm": 0.1252961903810501, "global_step": 11012, "epoch": 103, "lr": 8.971933089102251e-05} +{"train_loss": 0.017092380672693253, "grad_norm": 0.06010768190026283, "global_step": 11013, "epoch": 103, "lr": 8.971753059243911e-05} +{"train_loss": 0.02049127221107483, "grad_norm": 0.07235483080148697, "global_step": 11014, "epoch": 103, "lr": 8.971573015430557e-05} +{"train_loss": 0.020184822380542755, "grad_norm": 0.09263606369495392, "global_step": 11015, "epoch": 103, "lr": 8.971392957662821e-05} +{"train_loss": 0.016595354303717613, "grad_norm": 0.04652828723192215, "global_step": 11016, "epoch": 103, "lr": 8.971212885941339e-05} +{"train_loss": 0.020378436893224716, "grad_norm": 0.06041649729013443, "global_step": 11017, "epoch": 103, "lr": 8.97103280026674e-05} +{"train_loss": 0.016265198588371277, "grad_norm": 0.15241067111492157, "global_step": 11018, "epoch": 103, "lr": 8.970852700639661e-05} +{"train_loss": 0.014100709930062294, "grad_norm": 0.08707038313150406, "global_step": 11019, "epoch": 103, "lr": 8.970672587060733e-05} +{"train_loss": 0.01668040081858635, "grad_norm": 0.06217256560921669, "global_step": 11020, "epoch": 103, "lr": 8.970492459530586e-05} +{"train_loss": 0.017994392663240433, "grad_norm": 0.08261246979236603, "global_step": 11021, "epoch": 103, "lr": 8.970312318049856e-05} +{"train_loss": 0.017914116382598877, "grad_norm": 0.05870942771434784, "global_step": 11022, "epoch": 103, "lr": 8.970132162619176e-05} +{"train_loss": 0.017468707275770184, "grad_norm": 0.09090565145015717, "global_step": 11023, "epoch": 103, "lr": 8.969951993239177e-05, "val_loss": 0.018835244700312614} +{"train_loss": 0.015971234068274498, "grad_norm": 0.04613722860813141, "global_step": 11024, "epoch": 104, "lr": 8.969771809910494e-05} +{"train_loss": 0.01699257269501686, "grad_norm": 0.05959167331457138, "global_step": 11025, "epoch": 104, "lr": 8.96959161263376e-05} +{"train_loss": 0.016584578901529312, "grad_norm": 0.08368776738643646, "global_step": 11026, "epoch": 104, "lr": 8.969411401409606e-05} +{"train_loss": 0.0213441401720047, "grad_norm": 0.0512058325111866, "global_step": 11027, "epoch": 104, "lr": 8.969231176238667e-05} +{"train_loss": 0.018641836941242218, "grad_norm": 0.06364060938358307, "global_step": 11028, "epoch": 104, "lr": 8.969050937121574e-05} +{"train_loss": 0.019499454647302628, "grad_norm": 0.1406048685312271, "global_step": 11029, "epoch": 104, "lr": 8.968870684058965e-05} +{"train_loss": 0.01720818690955639, "grad_norm": 0.06361345201730728, "global_step": 11030, "epoch": 104, "lr": 8.968690417051469e-05} +{"train_loss": 0.01747574657201767, "grad_norm": 0.1396593153476715, "global_step": 11031, "epoch": 104, "lr": 8.968510136099721e-05} +{"train_loss": 0.017693830654025078, "grad_norm": 0.06357026100158691, "global_step": 11032, "epoch": 104, "lr": 8.968329841204354e-05} +{"train_loss": 0.018859192728996277, "grad_norm": 0.09756648540496826, "global_step": 11033, "epoch": 104, "lr": 8.968149532366003e-05} +{"train_loss": 0.0177910216152668, "grad_norm": 0.05938266962766647, "global_step": 11034, "epoch": 104, "lr": 8.967969209585298e-05} +{"train_loss": 0.01721908338367939, "grad_norm": 0.07576853036880493, "global_step": 11035, "epoch": 104, "lr": 8.967788872862876e-05} +{"train_loss": 0.018320826813578606, "grad_norm": 0.08039896935224533, "global_step": 11036, "epoch": 104, "lr": 8.967608522199371e-05} +{"train_loss": 0.017468716949224472, "grad_norm": 0.07756014168262482, "global_step": 11037, "epoch": 104, "lr": 8.967428157595413e-05} +{"train_loss": 0.017240939661860466, "grad_norm": 0.06243176385760307, "global_step": 11038, "epoch": 104, "lr": 8.96724777905164e-05} +{"train_loss": 0.013380401767790318, "grad_norm": 0.059910960495471954, "global_step": 11039, "epoch": 104, "lr": 8.967067386568682e-05} +{"train_loss": 0.0161028653383255, "grad_norm": 0.06620276719331741, "global_step": 11040, "epoch": 104, "lr": 8.966886980147173e-05} +{"train_loss": 0.017221566289663315, "grad_norm": 0.0774993747472763, "global_step": 11041, "epoch": 104, "lr": 8.966706559787749e-05} +{"train_loss": 0.018703510984778404, "grad_norm": 0.07963832467794418, "global_step": 11042, "epoch": 104, "lr": 8.966526125491046e-05} +{"train_loss": 0.02033444307744503, "grad_norm": 0.09454607963562012, "global_step": 11043, "epoch": 104, "lr": 8.966345677257693e-05} +{"train_loss": 0.019587328657507896, "grad_norm": 0.09558704495429993, "global_step": 11044, "epoch": 104, "lr": 8.966165215088327e-05} +{"train_loss": 0.019487835466861725, "grad_norm": 0.11662263423204422, "global_step": 11045, "epoch": 104, "lr": 8.965984738983581e-05} +{"train_loss": 0.01612882688641548, "grad_norm": 0.07011300325393677, "global_step": 11046, "epoch": 104, "lr": 8.965804248944087e-05} +{"train_loss": 0.016319816932082176, "grad_norm": 0.10440657287836075, "global_step": 11047, "epoch": 104, "lr": 8.965623744970485e-05} +{"train_loss": 0.015786422416567802, "grad_norm": 0.07732866704463959, "global_step": 11048, "epoch": 104, "lr": 8.965443227063405e-05} +{"train_loss": 0.017777085304260254, "grad_norm": 0.043140534311532974, "global_step": 11049, "epoch": 104, "lr": 8.965262695223483e-05} +{"train_loss": 0.014985652640461922, "grad_norm": 0.0430774912238121, "global_step": 11050, "epoch": 104, "lr": 8.96508214945135e-05} +{"train_loss": 0.02056177332997322, "grad_norm": 0.09295695275068283, "global_step": 11051, "epoch": 104, "lr": 8.964901589747643e-05} +{"train_loss": 0.019600538536906242, "grad_norm": 0.08977467566728592, "global_step": 11052, "epoch": 104, "lr": 8.964721016112996e-05} +{"train_loss": 0.017494527623057365, "grad_norm": 0.07572421431541443, "global_step": 11053, "epoch": 104, "lr": 8.964540428548045e-05} +{"train_loss": 0.01834997534751892, "grad_norm": 0.05522332713007927, "global_step": 11054, "epoch": 104, "lr": 8.964359827053421e-05} +{"train_loss": 0.01832885853946209, "grad_norm": 0.13968972861766815, "global_step": 11055, "epoch": 104, "lr": 8.964179211629762e-05} +{"train_loss": 0.015565373934805393, "grad_norm": 0.13086077570915222, "global_step": 11056, "epoch": 104, "lr": 8.963998582277702e-05} +{"train_loss": 0.018619731068611145, "grad_norm": 0.06501593440771103, "global_step": 11057, "epoch": 104, "lr": 8.963817938997874e-05} +{"train_loss": 0.018067989498376846, "grad_norm": 0.16183461248874664, "global_step": 11058, "epoch": 104, "lr": 8.963637281790913e-05} +{"train_loss": 0.01587606966495514, "grad_norm": 0.04334796965122223, "global_step": 11059, "epoch": 104, "lr": 8.963456610657456e-05} +{"train_loss": 0.015364427119493484, "grad_norm": 0.05573559179902077, "global_step": 11060, "epoch": 104, "lr": 8.963275925598134e-05} +{"train_loss": 0.01892250031232834, "grad_norm": 0.1002405658364296, "global_step": 11061, "epoch": 104, "lr": 8.963095226613585e-05} +{"train_loss": 0.01913289539515972, "grad_norm": 0.07795676589012146, "global_step": 11062, "epoch": 104, "lr": 8.962914513704444e-05} +{"train_loss": 0.017543230205774307, "grad_norm": 0.08709590882062912, "global_step": 11063, "epoch": 104, "lr": 8.962733786871342e-05} +{"train_loss": 0.016978897154331207, "grad_norm": 0.059661615639925, "global_step": 11064, "epoch": 104, "lr": 8.96255304611492e-05} +{"train_loss": 0.018860019743442535, "grad_norm": 0.05899330973625183, "global_step": 11065, "epoch": 104, "lr": 8.962372291435807e-05} +{"train_loss": 0.016222244128584862, "grad_norm": 0.058634739369153976, "global_step": 11066, "epoch": 104, "lr": 8.962191522834644e-05} +{"train_loss": 0.01891758292913437, "grad_norm": 0.12005565315485, "global_step": 11067, "epoch": 104, "lr": 8.962010740312062e-05} +{"train_loss": 0.016112979501485825, "grad_norm": 0.0558246485888958, "global_step": 11068, "epoch": 104, "lr": 8.961829943868696e-05} +{"train_loss": 0.015431864187121391, "grad_norm": 0.09605345875024796, "global_step": 11069, "epoch": 104, "lr": 8.961649133505183e-05} +{"train_loss": 0.01644296571612358, "grad_norm": 0.03922322392463684, "global_step": 11070, "epoch": 104, "lr": 8.961468309222158e-05} +{"train_loss": 0.01922057755291462, "grad_norm": 0.06260304152965546, "global_step": 11071, "epoch": 104, "lr": 8.961287471020257e-05} +{"train_loss": 0.014887132681906223, "grad_norm": 0.10953901708126068, "global_step": 11072, "epoch": 104, "lr": 8.961106618900113e-05} +{"train_loss": 0.016923846676945686, "grad_norm": 0.06834231317043304, "global_step": 11073, "epoch": 104, "lr": 8.960925752862363e-05} +{"train_loss": 0.018668171018362045, "grad_norm": 0.08895960450172424, "global_step": 11074, "epoch": 104, "lr": 8.960744872907645e-05} +{"train_loss": 0.016967052593827248, "grad_norm": 0.11732479929924011, "global_step": 11075, "epoch": 104, "lr": 8.960563979036589e-05} +{"train_loss": 0.016002502292394638, "grad_norm": 0.06895159929990768, "global_step": 11076, "epoch": 104, "lr": 8.960383071249836e-05} +{"train_loss": 0.017340505495667458, "grad_norm": 0.055847689509391785, "global_step": 11077, "epoch": 104, "lr": 8.960202149548019e-05} +{"train_loss": 0.018387317657470703, "grad_norm": 0.0713304728269577, "global_step": 11078, "epoch": 104, "lr": 8.960021213931772e-05} +{"train_loss": 0.016977403312921524, "grad_norm": 0.07464398443698883, "global_step": 11079, "epoch": 104, "lr": 8.959840264401734e-05} +{"train_loss": 0.01637306995689869, "grad_norm": 0.05204213038086891, "global_step": 11080, "epoch": 104, "lr": 8.959659300958538e-05} +{"train_loss": 0.015441123396158218, "grad_norm": 0.11417023837566376, "global_step": 11081, "epoch": 104, "lr": 8.959478323602822e-05} +{"train_loss": 0.015949267894029617, "grad_norm": 0.06366196274757385, "global_step": 11082, "epoch": 104, "lr": 8.959297332335222e-05} +{"train_loss": 0.017750687897205353, "grad_norm": 0.05531851574778557, "global_step": 11083, "epoch": 104, "lr": 8.959116327156371e-05} +{"train_loss": 0.017524056136608124, "grad_norm": 0.0707450732588768, "global_step": 11084, "epoch": 104, "lr": 8.95893530806691e-05} +{"train_loss": 0.018449030816555023, "grad_norm": 0.18149146437644958, "global_step": 11085, "epoch": 104, "lr": 8.958754275067471e-05} +{"train_loss": 0.01913169026374817, "grad_norm": 0.06240015849471092, "global_step": 11086, "epoch": 104, "lr": 8.95857322815869e-05} +{"train_loss": 0.015567418187856674, "grad_norm": 0.12010893225669861, "global_step": 11087, "epoch": 104, "lr": 8.958392167341204e-05} +{"train_loss": 0.01646854355931282, "grad_norm": 0.06901724636554718, "global_step": 11088, "epoch": 104, "lr": 8.958211092615651e-05} +{"train_loss": 0.018941674381494522, "grad_norm": 0.12661609053611755, "global_step": 11089, "epoch": 104, "lr": 8.958030003982665e-05} +{"train_loss": 0.018827008083462715, "grad_norm": 0.055690910667181015, "global_step": 11090, "epoch": 104, "lr": 8.95784890144288e-05} +{"train_loss": 0.021197475492954254, "grad_norm": 0.10058990865945816, "global_step": 11091, "epoch": 104, "lr": 8.957667784996939e-05} +{"train_loss": 0.018093304708600044, "grad_norm": 0.08334039896726608, "global_step": 11092, "epoch": 104, "lr": 8.957486654645473e-05} +{"train_loss": 0.014448889531195164, "grad_norm": 0.045597948133945465, "global_step": 11093, "epoch": 104, "lr": 8.95730551038912e-05} +{"train_loss": 0.018472623080015182, "grad_norm": 0.11120539903640747, "global_step": 11094, "epoch": 104, "lr": 8.957124352228517e-05} +{"train_loss": 0.01729823835194111, "grad_norm": 0.10540889203548431, "global_step": 11095, "epoch": 104, "lr": 8.9569431801643e-05} +{"train_loss": 0.016189923509955406, "grad_norm": 0.0721597746014595, "global_step": 11096, "epoch": 104, "lr": 8.956761994197105e-05} +{"train_loss": 0.019947446882724762, "grad_norm": 0.20790141820907593, "global_step": 11097, "epoch": 104, "lr": 8.956580794327569e-05} +{"train_loss": 0.018386248499155045, "grad_norm": 0.05340050905942917, "global_step": 11098, "epoch": 104, "lr": 8.956399580556329e-05} +{"train_loss": 0.019081395119428635, "grad_norm": 0.10825970768928528, "global_step": 11099, "epoch": 104, "lr": 8.956218352884022e-05} +{"train_loss": 0.018119145184755325, "grad_norm": 0.05882405489683151, "global_step": 11100, "epoch": 104, "lr": 8.956037111311285e-05} +{"train_loss": 0.017286863178014755, "grad_norm": 0.17605820298194885, "global_step": 11101, "epoch": 104, "lr": 8.95585585583875e-05} +{"train_loss": 0.01631252095103264, "grad_norm": 0.048573751002550125, "global_step": 11102, "epoch": 104, "lr": 8.955674586467061e-05} +{"train_loss": 0.020438212901353836, "grad_norm": 0.05388170853257179, "global_step": 11103, "epoch": 104, "lr": 8.955493303196852e-05} +{"train_loss": 0.019108016043901443, "grad_norm": 0.10178886353969574, "global_step": 11104, "epoch": 104, "lr": 8.955312006028759e-05} +{"train_loss": 0.01733258366584778, "grad_norm": 0.055797282606363297, "global_step": 11105, "epoch": 104, "lr": 8.955130694963419e-05} +{"train_loss": 0.018315576016902924, "grad_norm": 0.054564546793699265, "global_step": 11106, "epoch": 104, "lr": 8.954949370001471e-05} +{"train_loss": 0.015049749985337257, "grad_norm": 0.05550868809223175, "global_step": 11107, "epoch": 104, "lr": 8.95476803114355e-05} +{"train_loss": 0.018471939489245415, "grad_norm": 0.044353339821100235, "global_step": 11108, "epoch": 104, "lr": 8.954586678390295e-05} +{"train_loss": 0.014675896614789963, "grad_norm": 0.06503163278102875, "global_step": 11109, "epoch": 104, "lr": 8.954405311742342e-05} +{"train_loss": 0.017272554337978363, "grad_norm": 0.11157112568616867, "global_step": 11110, "epoch": 104, "lr": 8.954223931200327e-05} +{"train_loss": 0.017604902386665344, "grad_norm": 0.11506341397762299, "global_step": 11111, "epoch": 104, "lr": 8.95404253676489e-05} +{"train_loss": 0.018358629196882248, "grad_norm": 0.0747419223189354, "global_step": 11112, "epoch": 104, "lr": 8.953861128436667e-05} +{"train_loss": 0.020081207156181335, "grad_norm": 0.08092708885669708, "global_step": 11113, "epoch": 104, "lr": 8.953679706216295e-05} +{"train_loss": 0.01751483604311943, "grad_norm": 0.13876967132091522, "global_step": 11114, "epoch": 104, "lr": 8.953498270104413e-05} +{"train_loss": 0.01288013905286789, "grad_norm": 0.0617871917784214, "global_step": 11115, "epoch": 104, "lr": 8.953316820101658e-05} +{"train_loss": 0.01534462533891201, "grad_norm": 0.09152162075042725, "global_step": 11116, "epoch": 104, "lr": 8.953135356208664e-05} +{"train_loss": 0.015563305467367172, "grad_norm": 0.04434202238917351, "global_step": 11117, "epoch": 104, "lr": 8.952953878426074e-05} +{"train_loss": 0.014218934811651707, "grad_norm": 0.049852050840854645, "global_step": 11118, "epoch": 104, "lr": 8.952772386754522e-05} +{"train_loss": 0.018420163542032242, "grad_norm": 0.12295094132423401, "global_step": 11119, "epoch": 104, "lr": 8.952590881194648e-05} +{"train_loss": 0.018684303387999535, "grad_norm": 0.06885629892349243, "global_step": 11120, "epoch": 104, "lr": 8.952409361747088e-05} +{"train_loss": 0.017085108906030655, "grad_norm": 0.12349886447191238, "global_step": 11121, "epoch": 104, "lr": 8.952227828412479e-05} +{"train_loss": 0.014820119366049767, "grad_norm": 0.08027327060699463, "global_step": 11122, "epoch": 104, "lr": 8.952046281191462e-05} +{"train_loss": 0.01530005969107151, "grad_norm": 0.07452607154846191, "global_step": 11123, "epoch": 104, "lr": 8.951864720084673e-05} +{"train_loss": 0.01694110967218876, "grad_norm": 0.04687023162841797, "global_step": 11124, "epoch": 104, "lr": 8.951683145092749e-05} +{"train_loss": 0.017835088074207306, "grad_norm": 0.07823266834020615, "global_step": 11125, "epoch": 104, "lr": 8.951501556216329e-05} +{"train_loss": 0.01612680032849312, "grad_norm": 0.051164042204618454, "global_step": 11126, "epoch": 104, "lr": 8.951319953456053e-05} +{"train_loss": 0.019136128947138786, "grad_norm": 0.1205207034945488, "global_step": 11127, "epoch": 104, "lr": 8.951138336812555e-05} +{"train_loss": 0.018840927630662918, "grad_norm": 0.08788769692182541, "global_step": 11128, "epoch": 104, "lr": 8.950956706286475e-05} +{"train_loss": 0.017459077207055293, "grad_norm": 0.058331605046987534, "global_step": 11129, "epoch": 104, "lr": 8.950775061878453e-05, "val_loss": 0.019762365147471428} +{"train_loss": 0.017669621855020523, "grad_norm": 0.0460662804543972, "global_step": 11130, "epoch": 105, "lr": 8.950593403589124e-05} +{"train_loss": 0.0159706249833107, "grad_norm": 0.049213990569114685, "global_step": 11131, "epoch": 105, "lr": 8.950411731419128e-05} +{"train_loss": 0.014620763249695301, "grad_norm": 0.04951555281877518, "global_step": 11132, "epoch": 105, "lr": 8.950230045369103e-05} +{"train_loss": 0.01738046109676361, "grad_norm": 0.07476099580526352, "global_step": 11133, "epoch": 105, "lr": 8.950048345439688e-05} +{"train_loss": 0.018277153372764587, "grad_norm": 0.0539790578186512, "global_step": 11134, "epoch": 105, "lr": 8.949866631631522e-05} +{"train_loss": 0.01657262071967125, "grad_norm": 0.09774333238601685, "global_step": 11135, "epoch": 105, "lr": 8.949684903945242e-05} +{"train_loss": 0.015186591073870659, "grad_norm": 0.057343557476997375, "global_step": 11136, "epoch": 105, "lr": 8.949503162381486e-05} +{"train_loss": 0.01924966834485531, "grad_norm": 0.09847819805145264, "global_step": 11137, "epoch": 105, "lr": 8.949321406940893e-05} +{"train_loss": 0.015063032507896423, "grad_norm": 0.06683866679668427, "global_step": 11138, "epoch": 105, "lr": 8.949139637624105e-05} +{"train_loss": 0.016505565494298935, "grad_norm": 0.05166633054614067, "global_step": 11139, "epoch": 105, "lr": 8.948957854431754e-05} +{"train_loss": 0.016676409170031548, "grad_norm": 0.04737208038568497, "global_step": 11140, "epoch": 105, "lr": 8.948776057364486e-05} +{"train_loss": 0.017265407368540764, "grad_norm": 0.07453920692205429, "global_step": 11141, "epoch": 105, "lr": 8.948594246422934e-05} +{"train_loss": 0.017682552337646484, "grad_norm": 0.056888166815042496, "global_step": 11142, "epoch": 105, "lr": 8.948412421607739e-05} +{"train_loss": 0.01979883760213852, "grad_norm": 0.05627547949552536, "global_step": 11143, "epoch": 105, "lr": 8.94823058291954e-05} +{"train_loss": 0.01767655834555626, "grad_norm": 0.05501873791217804, "global_step": 11144, "epoch": 105, "lr": 8.948048730358976e-05} +{"train_loss": 0.018668629229068756, "grad_norm": 0.05240406095981598, "global_step": 11145, "epoch": 105, "lr": 8.947866863926686e-05} +{"train_loss": 0.016717806458473206, "grad_norm": 0.12766924500465393, "global_step": 11146, "epoch": 105, "lr": 8.947684983623309e-05} +{"train_loss": 0.014149394817650318, "grad_norm": 0.044513583183288574, "global_step": 11147, "epoch": 105, "lr": 8.947503089449483e-05} +{"train_loss": 0.01720505580306053, "grad_norm": 0.13181908428668976, "global_step": 11148, "epoch": 105, "lr": 8.947321181405848e-05} +{"train_loss": 0.015066483989357948, "grad_norm": 0.04773718863725662, "global_step": 11149, "epoch": 105, "lr": 8.947139259493044e-05} +{"train_loss": 0.020290430635213852, "grad_norm": 0.1399746984243393, "global_step": 11150, "epoch": 105, "lr": 8.946957323711708e-05} +{"train_loss": 0.018997926265001297, "grad_norm": 0.15675236284732819, "global_step": 11151, "epoch": 105, "lr": 8.946775374062482e-05} +{"train_loss": 0.01857960969209671, "grad_norm": 0.10473719239234924, "global_step": 11152, "epoch": 105, "lr": 8.946593410546001e-05} +{"train_loss": 0.016035504639148712, "grad_norm": 0.06980814784765244, "global_step": 11153, "epoch": 105, "lr": 8.94641143316291e-05} +{"train_loss": 0.01840784028172493, "grad_norm": 0.1316831260919571, "global_step": 11154, "epoch": 105, "lr": 8.946229441913842e-05} +{"train_loss": 0.017474975436925888, "grad_norm": 0.12538763880729675, "global_step": 11155, "epoch": 105, "lr": 8.946047436799443e-05} +{"train_loss": 0.019799765199422836, "grad_norm": 0.06354561448097229, "global_step": 11156, "epoch": 105, "lr": 8.945865417820349e-05} +{"train_loss": 0.0198188629001379, "grad_norm": 0.06835994869470596, "global_step": 11157, "epoch": 105, "lr": 8.945683384977199e-05} +{"train_loss": 0.01739221066236496, "grad_norm": 0.09048665314912796, "global_step": 11158, "epoch": 105, "lr": 8.945501338270634e-05} +{"train_loss": 0.015580026432871819, "grad_norm": 0.05695515498518944, "global_step": 11159, "epoch": 105, "lr": 8.945319277701293e-05} +{"train_loss": 0.020387589931488037, "grad_norm": 0.10424038022756577, "global_step": 11160, "epoch": 105, "lr": 8.945137203269815e-05} +{"train_loss": 0.018563535064458847, "grad_norm": 0.13114404678344727, "global_step": 11161, "epoch": 105, "lr": 8.944955114976843e-05} +{"train_loss": 0.01634354703128338, "grad_norm": 0.19836905598640442, "global_step": 11162, "epoch": 105, "lr": 8.94477301282301e-05} +{"train_loss": 0.016261713579297066, "grad_norm": 0.09102316200733185, "global_step": 11163, "epoch": 105, "lr": 8.944590896808965e-05} +{"train_loss": 0.016792837530374527, "grad_norm": 0.15946462750434875, "global_step": 11164, "epoch": 105, "lr": 8.944408766935342e-05} +{"train_loss": 0.01692427136003971, "grad_norm": 0.0621689036488533, "global_step": 11165, "epoch": 105, "lr": 8.94422662320278e-05} +{"train_loss": 0.016288436949253082, "grad_norm": 0.08509401232004166, "global_step": 11166, "epoch": 105, "lr": 8.944044465611923e-05} +{"train_loss": 0.018188919872045517, "grad_norm": 0.07283999770879745, "global_step": 11167, "epoch": 105, "lr": 8.943862294163407e-05} +{"train_loss": 0.014829626306891441, "grad_norm": 0.06511346995830536, "global_step": 11168, "epoch": 105, "lr": 8.943680108857875e-05} +{"train_loss": 0.015270239673554897, "grad_norm": 0.06707725673913956, "global_step": 11169, "epoch": 105, "lr": 8.943497909695966e-05} +{"train_loss": 0.014208095148205757, "grad_norm": 0.06515025347471237, "global_step": 11170, "epoch": 105, "lr": 8.943315696678321e-05} +{"train_loss": 0.01847289688885212, "grad_norm": 0.07898017019033432, "global_step": 11171, "epoch": 105, "lr": 8.943133469805577e-05} +{"train_loss": 0.017309710383415222, "grad_norm": 0.06910564005374908, "global_step": 11172, "epoch": 105, "lr": 8.94295122907838e-05} +{"train_loss": 0.01853126659989357, "grad_norm": 0.11323028802871704, "global_step": 11173, "epoch": 105, "lr": 8.942768974497367e-05} +{"train_loss": 0.018501056358218193, "grad_norm": 0.08396010845899582, "global_step": 11174, "epoch": 105, "lr": 8.942586706063177e-05} +{"train_loss": 0.01671207696199417, "grad_norm": 0.0827719196677208, "global_step": 11175, "epoch": 105, "lr": 8.942404423776451e-05} +{"train_loss": 0.01857163943350315, "grad_norm": 0.10410217195749283, "global_step": 11176, "epoch": 105, "lr": 8.942222127637832e-05} +{"train_loss": 0.017463229596614838, "grad_norm": 0.12895052134990692, "global_step": 11177, "epoch": 105, "lr": 8.942039817647958e-05} +{"train_loss": 0.014834032393991947, "grad_norm": 0.08787330240011215, "global_step": 11178, "epoch": 105, "lr": 8.941857493807472e-05} +{"train_loss": 0.014982113614678383, "grad_norm": 0.052020084112882614, "global_step": 11179, "epoch": 105, "lr": 8.941675156117011e-05} +{"train_loss": 0.015847498551011086, "grad_norm": 0.08520293235778809, "global_step": 11180, "epoch": 105, "lr": 8.94149280457722e-05} +{"train_loss": 0.015640508383512497, "grad_norm": 0.0686364471912384, "global_step": 11181, "epoch": 105, "lr": 8.941310439188735e-05} +{"train_loss": 0.01651580259203911, "grad_norm": 0.04582905024290085, "global_step": 11182, "epoch": 105, "lr": 8.941128059952201e-05} +{"train_loss": 0.016931327059864998, "grad_norm": 0.06114684045314789, "global_step": 11183, "epoch": 105, "lr": 8.940945666868256e-05} +{"train_loss": 0.016184136271476746, "grad_norm": 0.051929254084825516, "global_step": 11184, "epoch": 105, "lr": 8.940763259937541e-05} +{"train_loss": 0.016237689182162285, "grad_norm": 0.04737727344036102, "global_step": 11185, "epoch": 105, "lr": 8.940580839160699e-05} +{"train_loss": 0.017453491687774658, "grad_norm": 0.11514440923929214, "global_step": 11186, "epoch": 105, "lr": 8.940398404538371e-05} +{"train_loss": 0.020961139351129532, "grad_norm": 0.10683188587427139, "global_step": 11187, "epoch": 105, "lr": 8.940215956071194e-05} +{"train_loss": 0.018775634467601776, "grad_norm": 0.07546873390674591, "global_step": 11188, "epoch": 105, "lr": 8.940033493759811e-05} +{"train_loss": 0.01763557828962803, "grad_norm": 0.11415722221136093, "global_step": 11189, "epoch": 105, "lr": 8.939851017604866e-05} +{"train_loss": 0.020505152642726898, "grad_norm": 0.11131615936756134, "global_step": 11190, "epoch": 105, "lr": 8.939668527606998e-05} +{"train_loss": 0.01820063591003418, "grad_norm": 0.07352523505687714, "global_step": 11191, "epoch": 105, "lr": 8.939486023766847e-05} +{"train_loss": 0.017848540097475052, "grad_norm": 0.09175420552492142, "global_step": 11192, "epoch": 105, "lr": 8.939303506085057e-05} +{"train_loss": 0.019073517993092537, "grad_norm": 0.07750523090362549, "global_step": 11193, "epoch": 105, "lr": 8.939120974562266e-05} +{"train_loss": 0.014865877106785774, "grad_norm": 0.07021960616111755, "global_step": 11194, "epoch": 105, "lr": 8.938938429199117e-05} +{"train_loss": 0.01652533933520317, "grad_norm": 0.052627190947532654, "global_step": 11195, "epoch": 105, "lr": 8.938755869996252e-05} +{"train_loss": 0.020153826102614403, "grad_norm": 0.1434563547372818, "global_step": 11196, "epoch": 105, "lr": 8.938573296954312e-05} +{"train_loss": 0.018048882484436035, "grad_norm": 0.047660280019044876, "global_step": 11197, "epoch": 105, "lr": 8.938390710073938e-05} +{"train_loss": 0.018930189311504364, "grad_norm": 0.05097713693976402, "global_step": 11198, "epoch": 105, "lr": 8.938208109355771e-05} +{"train_loss": 0.015707755461335182, "grad_norm": 0.12595519423484802, "global_step": 11199, "epoch": 105, "lr": 8.938025494800454e-05} +{"train_loss": 0.01691889390349388, "grad_norm": 0.10864754021167755, "global_step": 11200, "epoch": 105, "lr": 8.937842866408628e-05} +{"train_loss": 0.01764356717467308, "grad_norm": 0.07026796042919159, "global_step": 11201, "epoch": 105, "lr": 8.937660224180935e-05} +{"train_loss": 0.01786935329437256, "grad_norm": 0.17746353149414062, "global_step": 11202, "epoch": 105, "lr": 8.937477568118015e-05} +{"train_loss": 0.017506707459688187, "grad_norm": 0.06917121261358261, "global_step": 11203, "epoch": 105, "lr": 8.937294898220513e-05} +{"train_loss": 0.01909155398607254, "grad_norm": 0.11973389983177185, "global_step": 11204, "epoch": 105, "lr": 8.937112214489069e-05} +{"train_loss": 0.019087569788098335, "grad_norm": 0.13566750288009644, "global_step": 11205, "epoch": 105, "lr": 8.936929516924325e-05} +{"train_loss": 0.01743105985224247, "grad_norm": 0.06983140856027603, "global_step": 11206, "epoch": 105, "lr": 8.936746805526921e-05} +{"train_loss": 0.017835937440395355, "grad_norm": 0.05031339079141617, "global_step": 11207, "epoch": 105, "lr": 8.936564080297501e-05} +{"train_loss": 0.015987461432814598, "grad_norm": 0.14811749756336212, "global_step": 11208, "epoch": 105, "lr": 8.936381341236708e-05} +{"train_loss": 0.017961006611585617, "grad_norm": 0.06238703802227974, "global_step": 11209, "epoch": 105, "lr": 8.936198588345182e-05} +{"train_loss": 0.017642639577388763, "grad_norm": 0.06305167078971863, "global_step": 11210, "epoch": 105, "lr": 8.936015821623567e-05} +{"train_loss": 0.019696127623319626, "grad_norm": 0.058809880167245865, "global_step": 11211, "epoch": 105, "lr": 8.935833041072503e-05} +{"train_loss": 0.014585521072149277, "grad_norm": 0.07316488772630692, "global_step": 11212, "epoch": 105, "lr": 8.935650246692635e-05} +{"train_loss": 0.019785653799772263, "grad_norm": 0.0844271257519722, "global_step": 11213, "epoch": 105, "lr": 8.935467438484602e-05} +{"train_loss": 0.016346201300621033, "grad_norm": 0.053449176251888275, "global_step": 11214, "epoch": 105, "lr": 8.935284616449048e-05} +{"train_loss": 0.019078701734542847, "grad_norm": 0.06892434507608414, "global_step": 11215, "epoch": 105, "lr": 8.935101780586615e-05} +{"train_loss": 0.0174938403069973, "grad_norm": 0.1135919839143753, "global_step": 11216, "epoch": 105, "lr": 8.934918930897946e-05} +{"train_loss": 0.019214771687984467, "grad_norm": 0.0788426548242569, "global_step": 11217, "epoch": 105, "lr": 8.934736067383685e-05} +{"train_loss": 0.01796247996389866, "grad_norm": 0.10732275247573853, "global_step": 11218, "epoch": 105, "lr": 8.93455319004447e-05} +{"train_loss": 0.015815962105989456, "grad_norm": 0.06322292238473892, "global_step": 11219, "epoch": 105, "lr": 8.934370298880947e-05} +{"train_loss": 0.016493752598762512, "grad_norm": 0.08280438929796219, "global_step": 11220, "epoch": 105, "lr": 8.934187393893759e-05} +{"train_loss": 0.019562937319278717, "grad_norm": 0.06973092257976532, "global_step": 11221, "epoch": 105, "lr": 8.934004475083547e-05} +{"train_loss": 0.016987361013889313, "grad_norm": 0.05892500281333923, "global_step": 11222, "epoch": 105, "lr": 8.933821542450953e-05} +{"train_loss": 0.019102510064840317, "grad_norm": 0.0801728144288063, "global_step": 11223, "epoch": 105, "lr": 8.93363859599662e-05} +{"train_loss": 0.017355652526021004, "grad_norm": 0.09629514813423157, "global_step": 11224, "epoch": 105, "lr": 8.933455635721195e-05} +{"train_loss": 0.018261509016156197, "grad_norm": 0.10831195116043091, "global_step": 11225, "epoch": 105, "lr": 8.933272661625315e-05} +{"train_loss": 0.015055079013109207, "grad_norm": 0.09483204036951065, "global_step": 11226, "epoch": 105, "lr": 8.933089673709625e-05} +{"train_loss": 0.01606622524559498, "grad_norm": 0.06990392506122589, "global_step": 11227, "epoch": 105, "lr": 8.93290667197477e-05} +{"train_loss": 0.01673733815550804, "grad_norm": 0.06542127579450607, "global_step": 11228, "epoch": 105, "lr": 8.932723656421392e-05} +{"train_loss": 0.017199845984578133, "grad_norm": 0.06709788739681244, "global_step": 11229, "epoch": 105, "lr": 8.932540627050132e-05} +{"train_loss": 0.016894863918423653, "grad_norm": 0.060006026178598404, "global_step": 11230, "epoch": 105, "lr": 8.932357583861633e-05} +{"train_loss": 0.019835541024804115, "grad_norm": 0.06558572500944138, "global_step": 11231, "epoch": 105, "lr": 8.932174526856543e-05} +{"train_loss": 0.0209474116563797, "grad_norm": 0.07317374646663666, "global_step": 11232, "epoch": 105, "lr": 8.9319914560355e-05} +{"train_loss": 0.018490534275770187, "grad_norm": 0.11225997656583786, "global_step": 11233, "epoch": 105, "lr": 8.931808371399148e-05} +{"train_loss": 0.01677917316555977, "grad_norm": 0.09406805783510208, "global_step": 11234, "epoch": 105, "lr": 8.931625272948131e-05} +{"train_loss": 0.01744542970269356, "grad_norm": 0.08409769833087921, "global_step": 11235, "epoch": 105, "lr": 8.931442160683094e-05, "val_loss": 0.019238445907831192} +{"train_loss": 0.016912633553147316, "grad_norm": 0.08709187805652618, "global_step": 11236, "epoch": 106, "lr": 8.93125903460468e-05} +{"train_loss": 0.019163066521286964, "grad_norm": 0.09284494072198868, "global_step": 11237, "epoch": 106, "lr": 8.931075894713529e-05} +{"train_loss": 0.019863849505782127, "grad_norm": 0.09038680046796799, "global_step": 11238, "epoch": 106, "lr": 8.930892741010289e-05} +{"train_loss": 0.013987855985760689, "grad_norm": 0.06996293365955353, "global_step": 11239, "epoch": 106, "lr": 8.930709573495602e-05} +{"train_loss": 0.02074640989303589, "grad_norm": 0.06065657362341881, "global_step": 11240, "epoch": 106, "lr": 8.930526392170109e-05} +{"train_loss": 0.016394436359405518, "grad_norm": 0.06155817583203316, "global_step": 11241, "epoch": 106, "lr": 8.930343197034455e-05} +{"train_loss": 0.017192792147397995, "grad_norm": 0.0683981329202652, "global_step": 11242, "epoch": 106, "lr": 8.930159988089286e-05} +{"train_loss": 0.013298253528773785, "grad_norm": 0.08004040271043777, "global_step": 11243, "epoch": 106, "lr": 8.929976765335242e-05} +{"train_loss": 0.01588837429881096, "grad_norm": 0.08033902198076248, "global_step": 11244, "epoch": 106, "lr": 8.929793528772972e-05} +{"train_loss": 0.018228329718112946, "grad_norm": 0.11622152477502823, "global_step": 11245, "epoch": 106, "lr": 8.929610278403113e-05} +{"train_loss": 0.016951916739344597, "grad_norm": 0.05351293832063675, "global_step": 11246, "epoch": 106, "lr": 8.929427014226315e-05} +{"train_loss": 0.016198094934225082, "grad_norm": 0.0653938427567482, "global_step": 11247, "epoch": 106, "lr": 8.929243736243218e-05} +{"train_loss": 0.016327761113643646, "grad_norm": 0.045868270099163055, "global_step": 11248, "epoch": 106, "lr": 8.929060444454468e-05} +{"train_loss": 0.019832009449601173, "grad_norm": 0.07329176366329193, "global_step": 11249, "epoch": 106, "lr": 8.928877138860707e-05} +{"train_loss": 0.017118334770202637, "grad_norm": 0.04761126637458801, "global_step": 11250, "epoch": 106, "lr": 8.928693819462581e-05} +{"train_loss": 0.017666906118392944, "grad_norm": 0.06423026323318481, "global_step": 11251, "epoch": 106, "lr": 8.928510486260735e-05} +{"train_loss": 0.01998939737677574, "grad_norm": 0.08747687935829163, "global_step": 11252, "epoch": 106, "lr": 8.92832713925581e-05} +{"train_loss": 0.016832556575536728, "grad_norm": 0.07589620351791382, "global_step": 11253, "epoch": 106, "lr": 8.928143778448453e-05} +{"train_loss": 0.01827809400856495, "grad_norm": 0.08008837699890137, "global_step": 11254, "epoch": 106, "lr": 8.927960403839306e-05} +{"train_loss": 0.015445244498550892, "grad_norm": 0.0937531515955925, "global_step": 11255, "epoch": 106, "lr": 8.927777015429015e-05} +{"train_loss": 0.014132048934698105, "grad_norm": 0.06805542856454849, "global_step": 11256, "epoch": 106, "lr": 8.927593613218223e-05} +{"train_loss": 0.019082752987742424, "grad_norm": 0.05477959290146828, "global_step": 11257, "epoch": 106, "lr": 8.927410197207575e-05} +{"train_loss": 0.021357791498303413, "grad_norm": 0.06693539768457413, "global_step": 11258, "epoch": 106, "lr": 8.927226767397716e-05} +{"train_loss": 0.020114650949835777, "grad_norm": 0.09830735623836517, "global_step": 11259, "epoch": 106, "lr": 8.92704332378929e-05} +{"train_loss": 0.016866009682416916, "grad_norm": 0.05794061720371246, "global_step": 11260, "epoch": 106, "lr": 8.926859866382942e-05} +{"train_loss": 0.017178304493427277, "grad_norm": 0.13990451395511627, "global_step": 11261, "epoch": 106, "lr": 8.926676395179316e-05} +{"train_loss": 0.019948244094848633, "grad_norm": 0.12015670537948608, "global_step": 11262, "epoch": 106, "lr": 8.926492910179056e-05} +{"train_loss": 0.01881362311542034, "grad_norm": 0.08811476081609726, "global_step": 11263, "epoch": 106, "lr": 8.926309411382809e-05} +{"train_loss": 0.019691428169608116, "grad_norm": 0.06666644662618637, "global_step": 11264, "epoch": 106, "lr": 8.926125898791218e-05} +{"train_loss": 0.016879357397556305, "grad_norm": 0.06282996386289597, "global_step": 11265, "epoch": 106, "lr": 8.925942372404926e-05} +{"train_loss": 0.014661489054560661, "grad_norm": 0.06389935314655304, "global_step": 11266, "epoch": 106, "lr": 8.925758832224582e-05} +{"train_loss": 0.017484527081251144, "grad_norm": 0.11814191192388535, "global_step": 11267, "epoch": 106, "lr": 8.925575278250828e-05} +{"train_loss": 0.01580718345940113, "grad_norm": 0.04832931235432625, "global_step": 11268, "epoch": 106, "lr": 8.92539171048431e-05} +{"train_loss": 0.015747904777526855, "grad_norm": 0.09036006778478622, "global_step": 11269, "epoch": 106, "lr": 8.925208128925672e-05} +{"train_loss": 0.016931356862187386, "grad_norm": 0.04205559194087982, "global_step": 11270, "epoch": 106, "lr": 8.92502453357556e-05} +{"train_loss": 0.015872396528720856, "grad_norm": 0.05432786047458649, "global_step": 11271, "epoch": 106, "lr": 8.924840924434619e-05} +{"train_loss": 0.017503123730421066, "grad_norm": 0.058947235345840454, "global_step": 11272, "epoch": 106, "lr": 8.924657301503494e-05} +{"train_loss": 0.018188921734690666, "grad_norm": 0.07062309980392456, "global_step": 11273, "epoch": 106, "lr": 8.924473664782828e-05} +{"train_loss": 0.01855282485485077, "grad_norm": 0.05419331043958664, "global_step": 11274, "epoch": 106, "lr": 8.924290014273271e-05} +{"train_loss": 0.01683044247329235, "grad_norm": 0.0825599953532219, "global_step": 11275, "epoch": 106, "lr": 8.924106349975464e-05} +{"train_loss": 0.015525104478001595, "grad_norm": 0.0866544172167778, "global_step": 11276, "epoch": 106, "lr": 8.923922671890055e-05} +{"train_loss": 0.01682184636592865, "grad_norm": 0.05323211848735809, "global_step": 11277, "epoch": 106, "lr": 8.923738980017686e-05} +{"train_loss": 0.0191661287099123, "grad_norm": 0.07372076064348221, "global_step": 11278, "epoch": 106, "lr": 8.923555274359007e-05} +{"train_loss": 0.016043152660131454, "grad_norm": 0.1285245418548584, "global_step": 11279, "epoch": 106, "lr": 8.923371554914659e-05} +{"train_loss": 0.01725943014025688, "grad_norm": 0.06484906375408173, "global_step": 11280, "epoch": 106, "lr": 8.92318782168529e-05} +{"train_loss": 0.018507247790694237, "grad_norm": 0.06966304779052734, "global_step": 11281, "epoch": 106, "lr": 8.923004074671547e-05} +{"train_loss": 0.017264651134610176, "grad_norm": 0.04274861514568329, "global_step": 11282, "epoch": 106, "lr": 8.922820313874071e-05} +{"train_loss": 0.021188264712691307, "grad_norm": 0.05056669935584068, "global_step": 11283, "epoch": 106, "lr": 8.922636539293514e-05} +{"train_loss": 0.015939947217702866, "grad_norm": 0.12836849689483643, "global_step": 11284, "epoch": 106, "lr": 8.922452750930516e-05} +{"train_loss": 0.016934797167778015, "grad_norm": 0.07047092914581299, "global_step": 11285, "epoch": 106, "lr": 8.922268948785725e-05} +{"train_loss": 0.016672194004058838, "grad_norm": 0.05046903342008591, "global_step": 11286, "epoch": 106, "lr": 8.922085132859786e-05} +{"train_loss": 0.020543407648801804, "grad_norm": 0.05167686566710472, "global_step": 11287, "epoch": 106, "lr": 8.921901303153345e-05} +{"train_loss": 0.01634262129664421, "grad_norm": 0.12295927107334137, "global_step": 11288, "epoch": 106, "lr": 8.92171745966705e-05} +{"train_loss": 0.016102109104394913, "grad_norm": 0.11217650026082993, "global_step": 11289, "epoch": 106, "lr": 8.921533602401546e-05} +{"train_loss": 0.014162934385240078, "grad_norm": 0.11997637152671814, "global_step": 11290, "epoch": 106, "lr": 8.921349731357476e-05} +{"train_loss": 0.018073290586471558, "grad_norm": 0.12012916058301926, "global_step": 11291, "epoch": 106, "lr": 8.921165846535491e-05} +{"train_loss": 0.018267931416630745, "grad_norm": 0.05754632130265236, "global_step": 11292, "epoch": 106, "lr": 8.920981947936233e-05} +{"train_loss": 0.018764348700642586, "grad_norm": 0.07157736271619797, "global_step": 11293, "epoch": 106, "lr": 8.920798035560349e-05} +{"train_loss": 0.01709115505218506, "grad_norm": 0.05809453874826431, "global_step": 11294, "epoch": 106, "lr": 8.920614109408487e-05} +{"train_loss": 0.01560527179390192, "grad_norm": 0.0808861032128334, "global_step": 11295, "epoch": 106, "lr": 8.920430169481291e-05} +{"train_loss": 0.017306683585047722, "grad_norm": 0.11457251012325287, "global_step": 11296, "epoch": 106, "lr": 8.92024621577941e-05} +{"train_loss": 0.017775369808077812, "grad_norm": 0.11962258070707321, "global_step": 11297, "epoch": 106, "lr": 8.920062248303487e-05} +{"train_loss": 0.017918188124895096, "grad_norm": 0.0668928325176239, "global_step": 11298, "epoch": 106, "lr": 8.919878267054171e-05} +{"train_loss": 0.01884504035115242, "grad_norm": 0.1173444390296936, "global_step": 11299, "epoch": 106, "lr": 8.919694272032107e-05} +{"train_loss": 0.014756531454622746, "grad_norm": 0.07419762015342712, "global_step": 11300, "epoch": 106, "lr": 8.919510263237942e-05} +{"train_loss": 0.016914913430809975, "grad_norm": 0.053348358720541, "global_step": 11301, "epoch": 106, "lr": 8.919326240672323e-05} +{"train_loss": 0.015388377010822296, "grad_norm": 0.18223488330841064, "global_step": 11302, "epoch": 106, "lr": 8.919142204335895e-05} +{"train_loss": 0.017879653722047806, "grad_norm": 0.05593724548816681, "global_step": 11303, "epoch": 106, "lr": 8.918958154229307e-05} +{"train_loss": 0.018003717064857483, "grad_norm": 0.09508544951677322, "global_step": 11304, "epoch": 106, "lr": 8.918774090353203e-05} +{"train_loss": 0.016726311296224594, "grad_norm": 0.05565951392054558, "global_step": 11305, "epoch": 106, "lr": 8.918590012708232e-05} +{"train_loss": 0.014307561330497265, "grad_norm": 0.07765305787324905, "global_step": 11306, "epoch": 106, "lr": 8.918405921295038e-05} +{"train_loss": 0.017718199640512466, "grad_norm": 0.05570945516228676, "global_step": 11307, "epoch": 106, "lr": 8.918221816114274e-05} +{"train_loss": 0.016100652515888214, "grad_norm": 0.08264368027448654, "global_step": 11308, "epoch": 106, "lr": 8.91803769716658e-05} +{"train_loss": 0.017673926427960396, "grad_norm": 0.06809362024068832, "global_step": 11309, "epoch": 106, "lr": 8.917853564452604e-05} +{"train_loss": 0.019407937303185463, "grad_norm": 0.053790222853422165, "global_step": 11310, "epoch": 106, "lr": 8.917669417972996e-05} +{"train_loss": 0.014752808958292007, "grad_norm": 0.09827833622694016, "global_step": 11311, "epoch": 106, "lr": 8.9174852577284e-05} +{"train_loss": 0.017436020076274872, "grad_norm": 0.08201617002487183, "global_step": 11312, "epoch": 106, "lr": 8.917301083719466e-05} +{"train_loss": 0.018397435545921326, "grad_norm": 0.05207562819123268, "global_step": 11313, "epoch": 106, "lr": 8.91711689594684e-05} +{"train_loss": 0.018074963241815567, "grad_norm": 0.12190835922956467, "global_step": 11314, "epoch": 106, "lr": 8.916932694411166e-05} +{"train_loss": 0.017054306343197823, "grad_norm": 0.06963624060153961, "global_step": 11315, "epoch": 106, "lr": 8.916748479113097e-05} +{"train_loss": 0.018849140033125877, "grad_norm": 0.05344916135072708, "global_step": 11316, "epoch": 106, "lr": 8.916564250053276e-05} +{"train_loss": 0.01868637278676033, "grad_norm": 0.06383779644966125, "global_step": 11317, "epoch": 106, "lr": 8.916380007232351e-05} +{"train_loss": 0.019132953137159348, "grad_norm": 0.07824385911226273, "global_step": 11318, "epoch": 106, "lr": 8.91619575065097e-05} +{"train_loss": 0.01811572164297104, "grad_norm": 0.1580805778503418, "global_step": 11319, "epoch": 106, "lr": 8.916011480309781e-05} +{"train_loss": 0.018746422603726387, "grad_norm": 0.0643601343035698, "global_step": 11320, "epoch": 106, "lr": 8.91582719620943e-05} +{"train_loss": 0.015728045254945755, "grad_norm": 0.1040879562497139, "global_step": 11321, "epoch": 106, "lr": 8.915642898350566e-05} +{"train_loss": 0.017565272748470306, "grad_norm": 0.08304638415575027, "global_step": 11322, "epoch": 106, "lr": 8.915458586733835e-05} +{"train_loss": 0.016637861728668213, "grad_norm": 0.055449966341257095, "global_step": 11323, "epoch": 106, "lr": 8.915274261359885e-05} +{"train_loss": 0.0190355833619833, "grad_norm": 0.11979995667934418, "global_step": 11324, "epoch": 106, "lr": 8.915089922229364e-05} +{"train_loss": 0.015808552503585815, "grad_norm": 0.04878978803753853, "global_step": 11325, "epoch": 106, "lr": 8.914905569342921e-05} +{"train_loss": 0.016184749081730843, "grad_norm": 0.11877255141735077, "global_step": 11326, "epoch": 106, "lr": 8.914721202701201e-05} +{"train_loss": 0.015871882438659668, "grad_norm": 0.06301954388618469, "global_step": 11327, "epoch": 106, "lr": 8.914536822304853e-05} +{"train_loss": 0.014762726612389088, "grad_norm": 0.04328535497188568, "global_step": 11328, "epoch": 106, "lr": 8.914352428154527e-05} +{"train_loss": 0.018710117787122726, "grad_norm": 0.05309347063302994, "global_step": 11329, "epoch": 106, "lr": 8.914168020250868e-05} +{"train_loss": 0.022291190922260284, "grad_norm": 0.06632425636053085, "global_step": 11330, "epoch": 106, "lr": 8.913983598594522e-05} +{"train_loss": 0.01675090193748474, "grad_norm": 0.06288222968578339, "global_step": 11331, "epoch": 106, "lr": 8.913799163186143e-05} +{"train_loss": 0.017544085159897804, "grad_norm": 0.08729273080825806, "global_step": 11332, "epoch": 106, "lr": 8.913614714026375e-05} +{"train_loss": 0.01954643428325653, "grad_norm": 0.10604263097047806, "global_step": 11333, "epoch": 106, "lr": 8.913430251115866e-05} +{"train_loss": 0.01801646500825882, "grad_norm": 0.07198407500982285, "global_step": 11334, "epoch": 106, "lr": 8.913245774455266e-05} +{"train_loss": 0.01805824600160122, "grad_norm": 0.10154148936271667, "global_step": 11335, "epoch": 106, "lr": 8.913061284045222e-05} +{"train_loss": 0.016145678237080574, "grad_norm": 0.12477727234363556, "global_step": 11336, "epoch": 106, "lr": 8.912876779886383e-05} +{"train_loss": 0.019416842609643936, "grad_norm": 0.1273874193429947, "global_step": 11337, "epoch": 106, "lr": 8.912692261979396e-05} +{"train_loss": 0.017577754333615303, "grad_norm": 0.10852351039648056, "global_step": 11338, "epoch": 106, "lr": 8.91250773032491e-05} +{"train_loss": 0.016326483339071274, "grad_norm": 0.054970882833004, "global_step": 11339, "epoch": 106, "lr": 8.912323184923573e-05} +{"train_loss": 0.014528420753777027, "grad_norm": 0.052174169570207596, "global_step": 11340, "epoch": 106, "lr": 8.912138625776034e-05} +{"train_loss": 0.017400187580793532, "grad_norm": 0.062351323664188385, "global_step": 11341, "epoch": 106, "lr": 8.91195405288294e-05, "val_loss": 0.019080905243754387} +{"train_loss": 0.019605446606874466, "grad_norm": 0.10639555752277374, "global_step": 11342, "epoch": 107, "lr": 8.911769466244944e-05} +{"train_loss": 0.01723434403538704, "grad_norm": 0.05616573989391327, "global_step": 11343, "epoch": 107, "lr": 8.911584865862688e-05} +{"train_loss": 0.016937218606472015, "grad_norm": 0.0937928780913353, "global_step": 11344, "epoch": 107, "lr": 8.911400251736826e-05} +{"train_loss": 0.014222956262528896, "grad_norm": 0.14829088747501373, "global_step": 11345, "epoch": 107, "lr": 8.911215623868003e-05} +{"train_loss": 0.016407117247581482, "grad_norm": 0.05859731137752533, "global_step": 11346, "epoch": 107, "lr": 8.911030982256869e-05} +{"train_loss": 0.015057286247611046, "grad_norm": 0.06310518085956573, "global_step": 11347, "epoch": 107, "lr": 8.910846326904074e-05} +{"train_loss": 0.017236478626728058, "grad_norm": 0.07487861812114716, "global_step": 11348, "epoch": 107, "lr": 8.910661657810266e-05} +{"train_loss": 0.0169102493673563, "grad_norm": 0.0802462175488472, "global_step": 11349, "epoch": 107, "lr": 8.910476974976094e-05} +{"train_loss": 0.016221370548009872, "grad_norm": 0.06827817112207413, "global_step": 11350, "epoch": 107, "lr": 8.910292278402206e-05} +{"train_loss": 0.01953408680856228, "grad_norm": 0.0905497819185257, "global_step": 11351, "epoch": 107, "lr": 8.910107568089251e-05} +{"train_loss": 0.016961468383669853, "grad_norm": 0.05780133232474327, "global_step": 11352, "epoch": 107, "lr": 8.909922844037879e-05} +{"train_loss": 0.016295621171593666, "grad_norm": 0.1485164314508438, "global_step": 11353, "epoch": 107, "lr": 8.909738106248738e-05} +{"train_loss": 0.017858996987342834, "grad_norm": 0.053395383059978485, "global_step": 11354, "epoch": 107, "lr": 8.909553354722478e-05} +{"train_loss": 0.016954850405454636, "grad_norm": 0.08540012687444687, "global_step": 11355, "epoch": 107, "lr": 8.909368589459748e-05} +{"train_loss": 0.01657155342400074, "grad_norm": 0.08178554475307465, "global_step": 11356, "epoch": 107, "lr": 8.909183810461197e-05} +{"train_loss": 0.017054766416549683, "grad_norm": 0.12126610428094864, "global_step": 11357, "epoch": 107, "lr": 8.908999017727473e-05} +{"train_loss": 0.01840868778526783, "grad_norm": 0.06397170573472977, "global_step": 11358, "epoch": 107, "lr": 8.908814211259227e-05} +{"train_loss": 0.019170213490724564, "grad_norm": 0.05624160170555115, "global_step": 11359, "epoch": 107, "lr": 8.908629391057109e-05} +{"train_loss": 0.01815444603562355, "grad_norm": 0.10360973328351974, "global_step": 11360, "epoch": 107, "lr": 8.908444557121765e-05} +{"train_loss": 0.018777796998620033, "grad_norm": 0.13920171558856964, "global_step": 11361, "epoch": 107, "lr": 8.908259709453848e-05} +{"train_loss": 0.015037705190479755, "grad_norm": 0.06329328566789627, "global_step": 11362, "epoch": 107, "lr": 8.908074848054005e-05} +{"train_loss": 0.01898491382598877, "grad_norm": 0.07030049711465836, "global_step": 11363, "epoch": 107, "lr": 8.907889972922887e-05} +{"train_loss": 0.0180109441280365, "grad_norm": 0.06590526551008224, "global_step": 11364, "epoch": 107, "lr": 8.907705084061144e-05} +{"train_loss": 0.018713783472776413, "grad_norm": 0.041920825839042664, "global_step": 11365, "epoch": 107, "lr": 8.907520181469424e-05} +{"train_loss": 0.01689828373491764, "grad_norm": 0.05347958207130432, "global_step": 11366, "epoch": 107, "lr": 8.907335265148377e-05} +{"train_loss": 0.017841944471001625, "grad_norm": 0.074722521007061, "global_step": 11367, "epoch": 107, "lr": 8.907150335098654e-05} +{"train_loss": 0.016738232225179672, "grad_norm": 0.05686792731285095, "global_step": 11368, "epoch": 107, "lr": 8.906965391320903e-05} +{"train_loss": 0.015110929496586323, "grad_norm": 0.0703597366809845, "global_step": 11369, "epoch": 107, "lr": 8.906780433815776e-05} +{"train_loss": 0.015013828873634338, "grad_norm": 0.062278009951114655, "global_step": 11370, "epoch": 107, "lr": 8.90659546258392e-05} +{"train_loss": 0.016531692817807198, "grad_norm": 0.07352373749017715, "global_step": 11371, "epoch": 107, "lr": 8.906410477625988e-05} +{"train_loss": 0.017807791009545326, "grad_norm": 0.10399899631738663, "global_step": 11372, "epoch": 107, "lr": 8.906225478942627e-05} +{"train_loss": 0.017819844186306, "grad_norm": 0.05461866781115532, "global_step": 11373, "epoch": 107, "lr": 8.90604046653449e-05} +{"train_loss": 0.01952754333615303, "grad_norm": 0.04616633057594299, "global_step": 11374, "epoch": 107, "lr": 8.905855440402224e-05} +{"train_loss": 0.015689905732870102, "grad_norm": 0.04414905980229378, "global_step": 11375, "epoch": 107, "lr": 8.905670400546481e-05} +{"train_loss": 0.01999688148498535, "grad_norm": 0.06037641689181328, "global_step": 11376, "epoch": 107, "lr": 8.905485346967912e-05} +{"train_loss": 0.01430508028715849, "grad_norm": 0.04214401915669441, "global_step": 11377, "epoch": 107, "lr": 8.905300279667164e-05} +{"train_loss": 0.017606019973754883, "grad_norm": 0.04579230025410652, "global_step": 11378, "epoch": 107, "lr": 8.905115198644891e-05} +{"train_loss": 0.017419803887605667, "grad_norm": 0.04996693506836891, "global_step": 11379, "epoch": 107, "lr": 8.904930103901741e-05} +{"train_loss": 0.01806069165468216, "grad_norm": 0.06218579411506653, "global_step": 11380, "epoch": 107, "lr": 8.904744995438364e-05} +{"train_loss": 0.015406088903546333, "grad_norm": 0.09913279861211777, "global_step": 11381, "epoch": 107, "lr": 8.904559873255411e-05} +{"train_loss": 0.01686902344226837, "grad_norm": 0.06967034935951233, "global_step": 11382, "epoch": 107, "lr": 8.904374737353534e-05} +{"train_loss": 0.01631198637187481, "grad_norm": 0.06931710243225098, "global_step": 11383, "epoch": 107, "lr": 8.90418958773338e-05} +{"train_loss": 0.016553983092308044, "grad_norm": 0.05681605637073517, "global_step": 11384, "epoch": 107, "lr": 8.904004424395605e-05} +{"train_loss": 0.019756559282541275, "grad_norm": 0.0921824723482132, "global_step": 11385, "epoch": 107, "lr": 8.903819247340853e-05} +{"train_loss": 0.01912013627588749, "grad_norm": 0.08135878294706345, "global_step": 11386, "epoch": 107, "lr": 8.903634056569779e-05} +{"train_loss": 0.01780623197555542, "grad_norm": 0.0811237245798111, "global_step": 11387, "epoch": 107, "lr": 8.903448852083034e-05} +{"train_loss": 0.01706906408071518, "grad_norm": 0.10168258845806122, "global_step": 11388, "epoch": 107, "lr": 8.903263633881265e-05} +{"train_loss": 0.017190568149089813, "grad_norm": 0.04659649357199669, "global_step": 11389, "epoch": 107, "lr": 8.903078401965126e-05} +{"train_loss": 0.020008686929941177, "grad_norm": 0.07263477146625519, "global_step": 11390, "epoch": 107, "lr": 8.902893156335268e-05} +{"train_loss": 0.018414413556456566, "grad_norm": 0.049452781677246094, "global_step": 11391, "epoch": 107, "lr": 8.90270789699234e-05} +{"train_loss": 0.01805778592824936, "grad_norm": 0.05022107809782028, "global_step": 11392, "epoch": 107, "lr": 8.902522623936993e-05} +{"train_loss": 0.01905122771859169, "grad_norm": 0.10057033598423004, "global_step": 11393, "epoch": 107, "lr": 8.90233733716988e-05} +{"train_loss": 0.01642589643597603, "grad_norm": 0.07871789485216141, "global_step": 11394, "epoch": 107, "lr": 8.90215203669165e-05} +{"train_loss": 0.016976891085505486, "grad_norm": 0.060102395713329315, "global_step": 11395, "epoch": 107, "lr": 8.901966722502955e-05} +{"train_loss": 0.01969631016254425, "grad_norm": 0.0662170946598053, "global_step": 11396, "epoch": 107, "lr": 8.901781394604445e-05} +{"train_loss": 0.016578052192926407, "grad_norm": 0.10431289672851562, "global_step": 11397, "epoch": 107, "lr": 8.901596052996773e-05} +{"train_loss": 0.017421362921595573, "grad_norm": 0.05632204934954643, "global_step": 11398, "epoch": 107, "lr": 8.901410697680588e-05} +{"train_loss": 0.016219329088926315, "grad_norm": 0.0767282173037529, "global_step": 11399, "epoch": 107, "lr": 8.901225328656542e-05} +{"train_loss": 0.020289797335863113, "grad_norm": 0.12944483757019043, "global_step": 11400, "epoch": 107, "lr": 8.90103994592529e-05} +{"train_loss": 0.018668487668037415, "grad_norm": 0.07173248380422592, "global_step": 11401, "epoch": 107, "lr": 8.900854549487477e-05} +{"train_loss": 0.017065802589058876, "grad_norm": 0.09564072638750076, "global_step": 11402, "epoch": 107, "lr": 8.900669139343759e-05} +{"train_loss": 0.01599527895450592, "grad_norm": 0.04763907194137573, "global_step": 11403, "epoch": 107, "lr": 8.900483715494785e-05} +{"train_loss": 0.01704106107354164, "grad_norm": 0.04949650913476944, "global_step": 11404, "epoch": 107, "lr": 8.900298277941208e-05} +{"train_loss": 0.01663045398890972, "grad_norm": 0.05515770614147186, "global_step": 11405, "epoch": 107, "lr": 8.900112826683679e-05} +{"train_loss": 0.016367098316550255, "grad_norm": 0.04533843323588371, "global_step": 11406, "epoch": 107, "lr": 8.89992736172285e-05} +{"train_loss": 0.016627952456474304, "grad_norm": 0.06987316161394119, "global_step": 11407, "epoch": 107, "lr": 8.899741883059374e-05} +{"train_loss": 0.014959404245018959, "grad_norm": 0.08141844719648361, "global_step": 11408, "epoch": 107, "lr": 8.899556390693898e-05} +{"train_loss": 0.018870946019887924, "grad_norm": 0.048079852014780045, "global_step": 11409, "epoch": 107, "lr": 8.899370884627079e-05} +{"train_loss": 0.014635024592280388, "grad_norm": 0.07006067782640457, "global_step": 11410, "epoch": 107, "lr": 8.899185364859565e-05} +{"train_loss": 0.01744401827454567, "grad_norm": 0.08479920774698257, "global_step": 11411, "epoch": 107, "lr": 8.89899983139201e-05} +{"train_loss": 0.016741687431931496, "grad_norm": 0.047984302043914795, "global_step": 11412, "epoch": 107, "lr": 8.898814284225064e-05} +{"train_loss": 0.018302524462342262, "grad_norm": 0.08523698896169662, "global_step": 11413, "epoch": 107, "lr": 8.898628723359381e-05} +{"train_loss": 0.017523588612675667, "grad_norm": 0.06795578449964523, "global_step": 11414, "epoch": 107, "lr": 8.898443148795612e-05} +{"train_loss": 0.01821380853652954, "grad_norm": 0.06417156755924225, "global_step": 11415, "epoch": 107, "lr": 8.898257560534409e-05} +{"train_loss": 0.019510220736265182, "grad_norm": 0.05922572687268257, "global_step": 11416, "epoch": 107, "lr": 8.898071958576426e-05} +{"train_loss": 0.020581882447004318, "grad_norm": 0.08608607947826385, "global_step": 11417, "epoch": 107, "lr": 8.897886342922313e-05} +{"train_loss": 0.016701750457286835, "grad_norm": 0.0410933643579483, "global_step": 11418, "epoch": 107, "lr": 8.89770071357272e-05} +{"train_loss": 0.015234498307108879, "grad_norm": 0.05030252784490585, "global_step": 11419, "epoch": 107, "lr": 8.897515070528305e-05} +{"train_loss": 0.013832040131092072, "grad_norm": 0.04735885187983513, "global_step": 11420, "epoch": 107, "lr": 8.897329413789715e-05} +{"train_loss": 0.01609719730913639, "grad_norm": 0.05060998350381851, "global_step": 11421, "epoch": 107, "lr": 8.897143743357603e-05} +{"train_loss": 0.014895965345203876, "grad_norm": 0.046428266912698746, "global_step": 11422, "epoch": 107, "lr": 8.896958059232625e-05} +{"train_loss": 0.018061650916934013, "grad_norm": 0.0777975395321846, "global_step": 11423, "epoch": 107, "lr": 8.89677236141543e-05} +{"train_loss": 0.014765586704015732, "grad_norm": 0.0707675889134407, "global_step": 11424, "epoch": 107, "lr": 8.896586649906673e-05} +{"train_loss": 0.020086657255887985, "grad_norm": 0.07213763892650604, "global_step": 11425, "epoch": 107, "lr": 8.896400924707004e-05} +{"train_loss": 0.01669604703783989, "grad_norm": 0.051762793213129044, "global_step": 11426, "epoch": 107, "lr": 8.896215185817077e-05} +{"train_loss": 0.017830368131399155, "grad_norm": 0.048269547522068024, "global_step": 11427, "epoch": 107, "lr": 8.896029433237544e-05} +{"train_loss": 0.015669509768486023, "grad_norm": 0.04614541307091713, "global_step": 11428, "epoch": 107, "lr": 8.895843666969057e-05} +{"train_loss": 0.017948124557733536, "grad_norm": 0.08763305842876434, "global_step": 11429, "epoch": 107, "lr": 8.895657887012271e-05} +{"train_loss": 0.01886739954352379, "grad_norm": 0.1389709860086441, "global_step": 11430, "epoch": 107, "lr": 8.895472093367837e-05} +{"train_loss": 0.02057424746453762, "grad_norm": 0.07945113629102707, "global_step": 11431, "epoch": 107, "lr": 8.895286286036409e-05} +{"train_loss": 0.01820947229862213, "grad_norm": 0.06777453422546387, "global_step": 11432, "epoch": 107, "lr": 8.895100465018637e-05} +{"train_loss": 0.016798965632915497, "grad_norm": 0.07339426130056381, "global_step": 11433, "epoch": 107, "lr": 8.894914630315178e-05} +{"train_loss": 0.017046675086021423, "grad_norm": 0.051089148968458176, "global_step": 11434, "epoch": 107, "lr": 8.89472878192668e-05} +{"train_loss": 0.016439639031887054, "grad_norm": 0.09959056973457336, "global_step": 11435, "epoch": 107, "lr": 8.894542919853802e-05} +{"train_loss": 0.015710080042481422, "grad_norm": 0.04944910854101181, "global_step": 11436, "epoch": 107, "lr": 8.894357044097193e-05} +{"train_loss": 0.016359712928533554, "grad_norm": 0.04896707460284233, "global_step": 11437, "epoch": 107, "lr": 8.894171154657505e-05} +{"train_loss": 0.018213285133242607, "grad_norm": 0.11010730266571045, "global_step": 11438, "epoch": 107, "lr": 8.893985251535395e-05} +{"train_loss": 0.019849691540002823, "grad_norm": 0.04469013959169388, "global_step": 11439, "epoch": 107, "lr": 8.893799334731514e-05} +{"train_loss": 0.018546637147665024, "grad_norm": 0.044065844267606735, "global_step": 11440, "epoch": 107, "lr": 8.893613404246517e-05} +{"train_loss": 0.017121199518442154, "grad_norm": 0.09670735895633698, "global_step": 11441, "epoch": 107, "lr": 8.893427460081054e-05} +{"train_loss": 0.017271585762500763, "grad_norm": 0.06871174275875092, "global_step": 11442, "epoch": 107, "lr": 8.89324150223578e-05} +{"train_loss": 0.02022489532828331, "grad_norm": 0.06871306896209717, "global_step": 11443, "epoch": 107, "lr": 8.893055530711349e-05} +{"train_loss": 0.016204938292503357, "grad_norm": 0.08617445081472397, "global_step": 11444, "epoch": 107, "lr": 8.892869545508413e-05} +{"train_loss": 0.018900949507951736, "grad_norm": 0.13420400023460388, "global_step": 11445, "epoch": 107, "lr": 8.892683546627628e-05} +{"train_loss": 0.016270466148853302, "grad_norm": 0.09279052913188934, "global_step": 11446, "epoch": 107, "lr": 8.892497534069644e-05} +{"train_loss": 0.017371088099437503, "grad_norm": 0.08304708451032639, "global_step": 11447, "epoch": 107, "lr": 8.892311507835119e-05, "val_loss": 0.018772412091493607} +{"train_loss": 0.016359372064471245, "grad_norm": 0.0667332261800766, "global_step": 11448, "epoch": 108, "lr": 8.892125467924701e-05} +{"train_loss": 0.014274435117840767, "grad_norm": 0.09573860466480255, "global_step": 11449, "epoch": 108, "lr": 8.891939414339047e-05} +{"train_loss": 0.01459463033825159, "grad_norm": 0.06804788112640381, "global_step": 11450, "epoch": 108, "lr": 8.891753347078812e-05} +{"train_loss": 0.019108761101961136, "grad_norm": 0.06967472285032272, "global_step": 11451, "epoch": 108, "lr": 8.89156726614465e-05} +{"train_loss": 0.019526047632098198, "grad_norm": 0.10223585367202759, "global_step": 11452, "epoch": 108, "lr": 8.89138117153721e-05} +{"train_loss": 0.015742329880595207, "grad_norm": 0.05837418511509895, "global_step": 11453, "epoch": 108, "lr": 8.89119506325715e-05} +{"train_loss": 0.017438534647226334, "grad_norm": 0.06712620705366135, "global_step": 11454, "epoch": 108, "lr": 8.891008941305124e-05} +{"train_loss": 0.01873015984892845, "grad_norm": 0.09368851780891418, "global_step": 11455, "epoch": 108, "lr": 8.890822805681783e-05} +{"train_loss": 0.01706327125430107, "grad_norm": 0.10092659294605255, "global_step": 11456, "epoch": 108, "lr": 8.890636656387782e-05} +{"train_loss": 0.016613036394119263, "grad_norm": 0.07300511747598648, "global_step": 11457, "epoch": 108, "lr": 8.890450493423777e-05} +{"train_loss": 0.0173482708632946, "grad_norm": 0.13823220133781433, "global_step": 11458, "epoch": 108, "lr": 8.890264316790421e-05} +{"train_loss": 0.021665304899215698, "grad_norm": 0.21839214861392975, "global_step": 11459, "epoch": 108, "lr": 8.890078126488368e-05} +{"train_loss": 0.017860978841781616, "grad_norm": 0.1210959404706955, "global_step": 11460, "epoch": 108, "lr": 8.889891922518272e-05} +{"train_loss": 0.016863180324435234, "grad_norm": 0.0999007374048233, "global_step": 11461, "epoch": 108, "lr": 8.889705704880787e-05} +{"train_loss": 0.017443247139453888, "grad_norm": 0.06849279999732971, "global_step": 11462, "epoch": 108, "lr": 8.889519473576569e-05} +{"train_loss": 0.019814498722553253, "grad_norm": 0.05951507389545441, "global_step": 11463, "epoch": 108, "lr": 8.88933322860627e-05} +{"train_loss": 0.01572551392018795, "grad_norm": 0.07474076747894287, "global_step": 11464, "epoch": 108, "lr": 8.889146969970545e-05} +{"train_loss": 0.01715002954006195, "grad_norm": 0.12160138785839081, "global_step": 11465, "epoch": 108, "lr": 8.88896069767005e-05} +{"train_loss": 0.01596486195921898, "grad_norm": 0.10062525421380997, "global_step": 11466, "epoch": 108, "lr": 8.888774411705438e-05} +{"train_loss": 0.01729142665863037, "grad_norm": 0.10873717814683914, "global_step": 11467, "epoch": 108, "lr": 8.888588112077363e-05} +{"train_loss": 0.018499789759516716, "grad_norm": 0.11000020802021027, "global_step": 11468, "epoch": 108, "lr": 8.888401798786483e-05} +{"train_loss": 0.016556154936552048, "grad_norm": 0.1544441431760788, "global_step": 11469, "epoch": 108, "lr": 8.888215471833448e-05} +{"train_loss": 0.015944313257932663, "grad_norm": 0.14401637017726898, "global_step": 11470, "epoch": 108, "lr": 8.888029131218915e-05} +{"train_loss": 0.01977478712797165, "grad_norm": 0.055077750235795975, "global_step": 11471, "epoch": 108, "lr": 8.887842776943539e-05} +{"train_loss": 0.0171898752450943, "grad_norm": 0.06772962212562561, "global_step": 11472, "epoch": 108, "lr": 8.887656409007974e-05} +{"train_loss": 0.019857876002788544, "grad_norm": 0.07462470978498459, "global_step": 11473, "epoch": 108, "lr": 8.887470027412875e-05} +{"train_loss": 0.014887651428580284, "grad_norm": 0.06863434612751007, "global_step": 11474, "epoch": 108, "lr": 8.887283632158897e-05} +{"train_loss": 0.01867961883544922, "grad_norm": 0.06817936152219772, "global_step": 11475, "epoch": 108, "lr": 8.887097223246696e-05} +{"train_loss": 0.018073463812470436, "grad_norm": 0.064410001039505, "global_step": 11476, "epoch": 108, "lr": 8.886910800676924e-05} +{"train_loss": 0.018076978623867035, "grad_norm": 0.060535844415426254, "global_step": 11477, "epoch": 108, "lr": 8.886724364450239e-05} +{"train_loss": 0.018359770998358727, "grad_norm": 0.04677099734544754, "global_step": 11478, "epoch": 108, "lr": 8.886537914567293e-05} +{"train_loss": 0.01335622277110815, "grad_norm": 0.048165541142225266, "global_step": 11479, "epoch": 108, "lr": 8.886351451028745e-05} +{"train_loss": 0.016224395483732224, "grad_norm": 0.05456935241818428, "global_step": 11480, "epoch": 108, "lr": 8.886164973835246e-05} +{"train_loss": 0.017728568986058235, "grad_norm": 0.05430300533771515, "global_step": 11481, "epoch": 108, "lr": 8.885978482987456e-05} +{"train_loss": 0.016991566866636276, "grad_norm": 0.10804355144500732, "global_step": 11482, "epoch": 108, "lr": 8.885791978486026e-05} +{"train_loss": 0.015596890822052956, "grad_norm": 0.16820412874221802, "global_step": 11483, "epoch": 108, "lr": 8.885605460331614e-05} +{"train_loss": 0.015420090407133102, "grad_norm": 0.0429980605840683, "global_step": 11484, "epoch": 108, "lr": 8.885418928524873e-05} +{"train_loss": 0.01928531564772129, "grad_norm": 0.10028107464313507, "global_step": 11485, "epoch": 108, "lr": 8.88523238306646e-05} +{"train_loss": 0.02032424509525299, "grad_norm": 0.13414913415908813, "global_step": 11486, "epoch": 108, "lr": 8.88504582395703e-05} +{"train_loss": 0.017835348844528198, "grad_norm": 0.1070237010717392, "global_step": 11487, "epoch": 108, "lr": 8.884859251197239e-05} +{"train_loss": 0.015380915254354477, "grad_norm": 0.059595510363578796, "global_step": 11488, "epoch": 108, "lr": 8.884672664787742e-05} +{"train_loss": 0.01789034903049469, "grad_norm": 0.0706263855099678, "global_step": 11489, "epoch": 108, "lr": 8.884486064729194e-05} +{"train_loss": 0.018704771995544434, "grad_norm": 0.06406667083501816, "global_step": 11490, "epoch": 108, "lr": 8.884299451022252e-05} +{"train_loss": 0.018757816404104233, "grad_norm": 0.051642533391714096, "global_step": 11491, "epoch": 108, "lr": 8.884112823667573e-05} +{"train_loss": 0.018040452152490616, "grad_norm": 0.0590025894343853, "global_step": 11492, "epoch": 108, "lr": 8.883926182665807e-05} +{"train_loss": 0.020450133830308914, "grad_norm": 0.05461191385984421, "global_step": 11493, "epoch": 108, "lr": 8.883739528017613e-05} +{"train_loss": 0.018015649169683456, "grad_norm": 0.05618659406900406, "global_step": 11494, "epoch": 108, "lr": 8.883552859723651e-05} +{"train_loss": 0.01807529665529728, "grad_norm": 0.10348827391862869, "global_step": 11495, "epoch": 108, "lr": 8.88336617778457e-05} +{"train_loss": 0.01711641252040863, "grad_norm": 0.12956322729587555, "global_step": 11496, "epoch": 108, "lr": 8.883179482201031e-05} +{"train_loss": 0.01791532151401043, "grad_norm": 0.12406360357999802, "global_step": 11497, "epoch": 108, "lr": 8.882992772973688e-05} +{"train_loss": 0.018042581155896187, "grad_norm": 0.05630249157547951, "global_step": 11498, "epoch": 108, "lr": 8.882806050103196e-05} +{"train_loss": 0.018324660137295723, "grad_norm": 0.10853803157806396, "global_step": 11499, "epoch": 108, "lr": 8.882619313590212e-05} +{"train_loss": 0.019181344658136368, "grad_norm": 0.0824287086725235, "global_step": 11500, "epoch": 108, "lr": 8.882432563435393e-05} +{"train_loss": 0.01577731966972351, "grad_norm": 0.08318005502223969, "global_step": 11501, "epoch": 108, "lr": 8.882245799639395e-05} +{"train_loss": 0.02040668949484825, "grad_norm": 0.08267202228307724, "global_step": 11502, "epoch": 108, "lr": 8.882059022202873e-05} +{"train_loss": 0.017499100416898727, "grad_norm": 0.059046629816293716, "global_step": 11503, "epoch": 108, "lr": 8.881872231126484e-05} +{"train_loss": 0.021112844347953796, "grad_norm": 0.0917704775929451, "global_step": 11504, "epoch": 108, "lr": 8.881685426410884e-05} +{"train_loss": 0.01641940325498581, "grad_norm": 0.07822015136480331, "global_step": 11505, "epoch": 108, "lr": 8.881498608056729e-05} +{"train_loss": 0.018067603930830956, "grad_norm": 0.06533992290496826, "global_step": 11506, "epoch": 108, "lr": 8.881311776064675e-05} +{"train_loss": 0.015971258282661438, "grad_norm": 0.11278960108757019, "global_step": 11507, "epoch": 108, "lr": 8.881124930435381e-05} +{"train_loss": 0.018515385687351227, "grad_norm": 0.07813595235347748, "global_step": 11508, "epoch": 108, "lr": 8.880938071169502e-05} +{"train_loss": 0.01824824884533882, "grad_norm": 0.07491809874773026, "global_step": 11509, "epoch": 108, "lr": 8.880751198267694e-05} +{"train_loss": 0.0179103072732687, "grad_norm": 0.12142836302518845, "global_step": 11510, "epoch": 108, "lr": 8.880564311730614e-05} +{"train_loss": 0.016813747584819794, "grad_norm": 0.05561504885554314, "global_step": 11511, "epoch": 108, "lr": 8.880377411558917e-05} +{"train_loss": 0.014348549768328667, "grad_norm": 0.03980707749724388, "global_step": 11512, "epoch": 108, "lr": 8.880190497753264e-05} +{"train_loss": 0.014392239972949028, "grad_norm": 0.09934010356664658, "global_step": 11513, "epoch": 108, "lr": 8.880003570314308e-05} +{"train_loss": 0.01653537154197693, "grad_norm": 0.1080787405371666, "global_step": 11514, "epoch": 108, "lr": 8.879816629242707e-05} +{"train_loss": 0.017408162355422974, "grad_norm": 0.07151361554861069, "global_step": 11515, "epoch": 108, "lr": 8.879629674539117e-05} +{"train_loss": 0.01841333508491516, "grad_norm": 0.13152529299259186, "global_step": 11516, "epoch": 108, "lr": 8.879442706204196e-05} +{"train_loss": 0.01899120584130287, "grad_norm": 0.11450760811567307, "global_step": 11517, "epoch": 108, "lr": 8.8792557242386e-05} +{"train_loss": 0.018016887828707695, "grad_norm": 0.1060846820473671, "global_step": 11518, "epoch": 108, "lr": 8.879068728642987e-05} +{"train_loss": 0.01812644489109516, "grad_norm": 0.10404371470212936, "global_step": 11519, "epoch": 108, "lr": 8.878881719418013e-05} +{"train_loss": 0.015300579369068146, "grad_norm": 0.08065042644739151, "global_step": 11520, "epoch": 108, "lr": 8.878694696564337e-05} +{"train_loss": 0.016579020768404007, "grad_norm": 0.06907252222299576, "global_step": 11521, "epoch": 108, "lr": 8.878507660082612e-05} +{"train_loss": 0.016233615577220917, "grad_norm": 0.12659355998039246, "global_step": 11522, "epoch": 108, "lr": 8.8783206099735e-05} +{"train_loss": 0.014611034654080868, "grad_norm": 0.09089744836091995, "global_step": 11523, "epoch": 108, "lr": 8.878133546237655e-05} +{"train_loss": 0.017621247097849846, "grad_norm": 0.058300092816352844, "global_step": 11524, "epoch": 108, "lr": 8.877946468875735e-05} +{"train_loss": 0.015581576153635979, "grad_norm": 0.050481218844652176, "global_step": 11525, "epoch": 108, "lr": 8.877759377888397e-05} +{"train_loss": 0.016576003283262253, "grad_norm": 0.09690291434526443, "global_step": 11526, "epoch": 108, "lr": 8.877572273276302e-05} +{"train_loss": 0.01783076673746109, "grad_norm": 0.06328266113996506, "global_step": 11527, "epoch": 108, "lr": 8.8773851550401e-05} +{"train_loss": 0.016820799559354782, "grad_norm": 0.1904568076133728, "global_step": 11528, "epoch": 108, "lr": 8.877198023180456e-05} +{"train_loss": 0.013915576040744781, "grad_norm": 0.11786959320306778, "global_step": 11529, "epoch": 108, "lr": 8.877010877698024e-05} +{"train_loss": 0.017172913998365402, "grad_norm": 0.10422264784574509, "global_step": 11530, "epoch": 108, "lr": 8.87682371859346e-05} +{"train_loss": 0.016195334494113922, "grad_norm": 0.05776168033480644, "global_step": 11531, "epoch": 108, "lr": 8.876636545867427e-05} +{"train_loss": 0.019633591175079346, "grad_norm": 0.08629477769136429, "global_step": 11532, "epoch": 108, "lr": 8.876449359520576e-05} +{"train_loss": 0.01625010557472706, "grad_norm": 0.11930317431688309, "global_step": 11533, "epoch": 108, "lr": 8.876262159553568e-05} +{"train_loss": 0.01629185490310192, "grad_norm": 0.054709769785404205, "global_step": 11534, "epoch": 108, "lr": 8.876074945967062e-05} +{"train_loss": 0.01807710900902748, "grad_norm": 0.10706663131713867, "global_step": 11535, "epoch": 108, "lr": 8.875887718761713e-05} +{"train_loss": 0.014385193586349487, "grad_norm": 0.08064243942499161, "global_step": 11536, "epoch": 108, "lr": 8.87570047793818e-05} +{"train_loss": 0.015712721273303032, "grad_norm": 0.06780451536178589, "global_step": 11537, "epoch": 108, "lr": 8.875513223497122e-05} +{"train_loss": 0.01731896586716175, "grad_norm": 0.10224448889493942, "global_step": 11538, "epoch": 108, "lr": 8.875325955439196e-05} +{"train_loss": 0.01948513463139534, "grad_norm": 0.15427638590335846, "global_step": 11539, "epoch": 108, "lr": 8.875138673765058e-05} +{"train_loss": 0.018129833042621613, "grad_norm": 0.08798182755708694, "global_step": 11540, "epoch": 108, "lr": 8.874951378475369e-05} +{"train_loss": 0.02109701558947563, "grad_norm": 0.19557321071624756, "global_step": 11541, "epoch": 108, "lr": 8.874764069570787e-05} +{"train_loss": 0.01609921082854271, "grad_norm": 0.11126948893070221, "global_step": 11542, "epoch": 108, "lr": 8.874576747051968e-05} +{"train_loss": 0.01910765841603279, "grad_norm": 0.14883027970790863, "global_step": 11543, "epoch": 108, "lr": 8.874389410919573e-05} +{"train_loss": 0.018126696348190308, "grad_norm": 0.0770624577999115, "global_step": 11544, "epoch": 108, "lr": 8.874202061174258e-05} +{"train_loss": 0.017044391483068466, "grad_norm": 0.10337965935468674, "global_step": 11545, "epoch": 108, "lr": 8.87401469781668e-05} +{"train_loss": 0.015247941017150879, "grad_norm": 0.06360263377428055, "global_step": 11546, "epoch": 108, "lr": 8.8738273208475e-05} +{"train_loss": 0.01621565967798233, "grad_norm": 0.052666857838630676, "global_step": 11547, "epoch": 108, "lr": 8.873639930267376e-05} +{"train_loss": 0.01929691806435585, "grad_norm": 0.05600063502788544, "global_step": 11548, "epoch": 108, "lr": 8.873452526076965e-05} +{"train_loss": 0.017294596880674362, "grad_norm": 0.059504274278879166, "global_step": 11549, "epoch": 108, "lr": 8.873265108276927e-05} +{"train_loss": 0.01946020871400833, "grad_norm": 0.06720392405986786, "global_step": 11550, "epoch": 108, "lr": 8.87307767686792e-05} +{"train_loss": 0.017504381015896797, "grad_norm": 0.05660942196846008, "global_step": 11551, "epoch": 108, "lr": 8.872890231850603e-05} +{"train_loss": 0.018251005560159683, "grad_norm": 0.21857334673404694, "global_step": 11552, "epoch": 108, "lr": 8.872702773225632e-05} +{"train_loss": 0.017426739420460642, "grad_norm": 0.135620579123497, "global_step": 11553, "epoch": 108, "lr": 8.872515300993669e-05, "val_loss": 0.02011914923787117} +{"train_loss": 0.01788315922021866, "grad_norm": 0.09991995990276337, "global_step": 11554, "epoch": 109, "lr": 8.872327815155372e-05} +{"train_loss": 0.0165775828063488, "grad_norm": 0.07038421183824539, "global_step": 11555, "epoch": 109, "lr": 8.872140315711398e-05} +{"train_loss": 0.017259197309613228, "grad_norm": 0.15614858269691467, "global_step": 11556, "epoch": 109, "lr": 8.871952802662408e-05} +{"train_loss": 0.015970606356859207, "grad_norm": 0.09531986713409424, "global_step": 11557, "epoch": 109, "lr": 8.871765276009058e-05} +{"train_loss": 0.018411196768283844, "grad_norm": 0.17342278361320496, "global_step": 11558, "epoch": 109, "lr": 8.87157773575201e-05} +{"train_loss": 0.018257729709148407, "grad_norm": 0.07290733605623245, "global_step": 11559, "epoch": 109, "lr": 8.871390181891921e-05} +{"train_loss": 0.019225571304559708, "grad_norm": 0.09085375815629959, "global_step": 11560, "epoch": 109, "lr": 8.871202614429451e-05} +{"train_loss": 0.01770508661866188, "grad_norm": 0.08530909568071365, "global_step": 11561, "epoch": 109, "lr": 8.871015033365257e-05} +{"train_loss": 0.01701464131474495, "grad_norm": 0.13074509799480438, "global_step": 11562, "epoch": 109, "lr": 8.870827438700001e-05} +{"train_loss": 0.01856464147567749, "grad_norm": 0.13782329857349396, "global_step": 11563, "epoch": 109, "lr": 8.87063983043434e-05} +{"train_loss": 0.018486006185412407, "grad_norm": 0.07961592823266983, "global_step": 11564, "epoch": 109, "lr": 8.870452208568936e-05} +{"train_loss": 0.016132544726133347, "grad_norm": 0.09495144337415695, "global_step": 11565, "epoch": 109, "lr": 8.870264573104443e-05} +{"train_loss": 0.016816169023513794, "grad_norm": 0.05154063552618027, "global_step": 11566, "epoch": 109, "lr": 8.870076924041526e-05} +{"train_loss": 0.01822461001574993, "grad_norm": 0.046872448176145554, "global_step": 11567, "epoch": 109, "lr": 8.869889261380841e-05} +{"train_loss": 0.01608208753168583, "grad_norm": 0.07233714312314987, "global_step": 11568, "epoch": 109, "lr": 8.869701585123047e-05} +{"train_loss": 0.021444397047162056, "grad_norm": 0.15743155777454376, "global_step": 11569, "epoch": 109, "lr": 8.869513895268806e-05} +{"train_loss": 0.014472346752882004, "grad_norm": 0.05608358234167099, "global_step": 11570, "epoch": 109, "lr": 8.869326191818777e-05} +{"train_loss": 0.01687069982290268, "grad_norm": 0.08113934099674225, "global_step": 11571, "epoch": 109, "lr": 8.869138474773615e-05} +{"train_loss": 0.01625819504261017, "grad_norm": 0.06560272723436356, "global_step": 11572, "epoch": 109, "lr": 8.868950744133986e-05} +{"train_loss": 0.01765032857656479, "grad_norm": 0.059624359011650085, "global_step": 11573, "epoch": 109, "lr": 8.868762999900545e-05} +{"train_loss": 0.017593054100871086, "grad_norm": 0.051378000527620316, "global_step": 11574, "epoch": 109, "lr": 8.868575242073953e-05} +{"train_loss": 0.02022000402212143, "grad_norm": 0.11057891696691513, "global_step": 11575, "epoch": 109, "lr": 8.868387470654872e-05} +{"train_loss": 0.017552293837070465, "grad_norm": 0.07874289155006409, "global_step": 11576, "epoch": 109, "lr": 8.868199685643958e-05} +{"train_loss": 0.018009396269917488, "grad_norm": 0.05191139131784439, "global_step": 11577, "epoch": 109, "lr": 8.868011887041874e-05} +{"train_loss": 0.013068951666355133, "grad_norm": 0.07826419174671173, "global_step": 11578, "epoch": 109, "lr": 8.867824074849276e-05} +{"train_loss": 0.01621086150407791, "grad_norm": 0.06167549267411232, "global_step": 11579, "epoch": 109, "lr": 8.867636249066829e-05} +{"train_loss": 0.01640591211616993, "grad_norm": 0.13444791734218597, "global_step": 11580, "epoch": 109, "lr": 8.86744840969519e-05} +{"train_loss": 0.01880495250225067, "grad_norm": 0.08363684266805649, "global_step": 11581, "epoch": 109, "lr": 8.867260556735018e-05} +{"train_loss": 0.01690686121582985, "grad_norm": 0.10376854240894318, "global_step": 11582, "epoch": 109, "lr": 8.867072690186973e-05} +{"train_loss": 0.01812024973332882, "grad_norm": 0.06713183969259262, "global_step": 11583, "epoch": 109, "lr": 8.866884810051718e-05} +{"train_loss": 0.015191474929451942, "grad_norm": 0.08096859604120255, "global_step": 11584, "epoch": 109, "lr": 8.866696916329911e-05} +{"train_loss": 0.015891512855887413, "grad_norm": 0.07816023379564285, "global_step": 11585, "epoch": 109, "lr": 8.866509009022214e-05} +{"train_loss": 0.019006455317139626, "grad_norm": 0.06168677657842636, "global_step": 11586, "epoch": 109, "lr": 8.866321088129283e-05} +{"train_loss": 0.015826374292373657, "grad_norm": 0.05183286592364311, "global_step": 11587, "epoch": 109, "lr": 8.866133153651783e-05} +{"train_loss": 0.01716296374797821, "grad_norm": 0.06534820795059204, "global_step": 11588, "epoch": 109, "lr": 8.865945205590373e-05} +{"train_loss": 0.01709447056055069, "grad_norm": 0.0777694508433342, "global_step": 11589, "epoch": 109, "lr": 8.865757243945711e-05} +{"train_loss": 0.019866814836859703, "grad_norm": 0.09242124855518341, "global_step": 11590, "epoch": 109, "lr": 8.865569268718461e-05} +{"train_loss": 0.018703756853938103, "grad_norm": 0.08082424849271774, "global_step": 11591, "epoch": 109, "lr": 8.865381279909282e-05} +{"train_loss": 0.014696628786623478, "grad_norm": 0.06292374432086945, "global_step": 11592, "epoch": 109, "lr": 8.865193277518832e-05} +{"train_loss": 0.01627197675406933, "grad_norm": 0.07093308120965958, "global_step": 11593, "epoch": 109, "lr": 8.865005261547776e-05} +{"train_loss": 0.01715521700680256, "grad_norm": 0.12319470196962357, "global_step": 11594, "epoch": 109, "lr": 8.86481723199677e-05} +{"train_loss": 0.01865396648645401, "grad_norm": 0.05090932548046112, "global_step": 11595, "epoch": 109, "lr": 8.864629188866479e-05} +{"train_loss": 0.01794370636343956, "grad_norm": 0.11027806252241135, "global_step": 11596, "epoch": 109, "lr": 8.864441132157561e-05} +{"train_loss": 0.019805220887064934, "grad_norm": 0.1318710297346115, "global_step": 11597, "epoch": 109, "lr": 8.864253061870678e-05} +{"train_loss": 0.0181625597178936, "grad_norm": 0.062406156212091446, "global_step": 11598, "epoch": 109, "lr": 8.864064978006488e-05} +{"train_loss": 0.018068324774503708, "grad_norm": 0.06926047056913376, "global_step": 11599, "epoch": 109, "lr": 8.863876880565656e-05} +{"train_loss": 0.013740912079811096, "grad_norm": 0.07071824371814728, "global_step": 11600, "epoch": 109, "lr": 8.863688769548842e-05} +{"train_loss": 0.017880521714687347, "grad_norm": 0.08891072869300842, "global_step": 11601, "epoch": 109, "lr": 8.863500644956704e-05} +{"train_loss": 0.01753918081521988, "grad_norm": 0.09362316131591797, "global_step": 11602, "epoch": 109, "lr": 8.863312506789905e-05} +{"train_loss": 0.016307543963193893, "grad_norm": 0.08595683425664902, "global_step": 11603, "epoch": 109, "lr": 8.863124355049107e-05} +{"train_loss": 0.019082263112068176, "grad_norm": 0.05315341427922249, "global_step": 11604, "epoch": 109, "lr": 8.862936189734969e-05} +{"train_loss": 0.016400527209043503, "grad_norm": 0.08553845435380936, "global_step": 11605, "epoch": 109, "lr": 8.862748010848154e-05} +{"train_loss": 0.019870979711413383, "grad_norm": 0.06430937349796295, "global_step": 11606, "epoch": 109, "lr": 8.862559818389322e-05} +{"train_loss": 0.01815672591328621, "grad_norm": 0.1252911537885666, "global_step": 11607, "epoch": 109, "lr": 8.862371612359134e-05} +{"train_loss": 0.02029666118323803, "grad_norm": 0.10391031205654144, "global_step": 11608, "epoch": 109, "lr": 8.862183392758254e-05} +{"train_loss": 0.016351601108908653, "grad_norm": 0.08277233690023422, "global_step": 11609, "epoch": 109, "lr": 8.86199515958734e-05} +{"train_loss": 0.017225686460733414, "grad_norm": 0.0762082189321518, "global_step": 11610, "epoch": 109, "lr": 8.861806912847055e-05} +{"train_loss": 0.01429063268005848, "grad_norm": 0.06231270730495453, "global_step": 11611, "epoch": 109, "lr": 8.861618652538058e-05} +{"train_loss": 0.019899271428585052, "grad_norm": 0.05459306389093399, "global_step": 11612, "epoch": 109, "lr": 8.861430378661014e-05} +{"train_loss": 0.01882822997868061, "grad_norm": 0.1276477575302124, "global_step": 11613, "epoch": 109, "lr": 8.861242091216582e-05} +{"train_loss": 0.018938755616545677, "grad_norm": 0.1695505678653717, "global_step": 11614, "epoch": 109, "lr": 8.861053790205426e-05} +{"train_loss": 0.01785099133849144, "grad_norm": 0.14340543746948242, "global_step": 11615, "epoch": 109, "lr": 8.860865475628205e-05} +{"train_loss": 0.014125637710094452, "grad_norm": 0.1092081069946289, "global_step": 11616, "epoch": 109, "lr": 8.860677147485583e-05} +{"train_loss": 0.017763923853635788, "grad_norm": 0.17880214750766754, "global_step": 11617, "epoch": 109, "lr": 8.860488805778219e-05} +{"train_loss": 0.020031478255987167, "grad_norm": 0.0719631090760231, "global_step": 11618, "epoch": 109, "lr": 8.860300450506777e-05} +{"train_loss": 0.017907992005348206, "grad_norm": 0.07096748799085617, "global_step": 11619, "epoch": 109, "lr": 8.860112081671918e-05} +{"train_loss": 0.016727197915315628, "grad_norm": 0.07731539011001587, "global_step": 11620, "epoch": 109, "lr": 8.859923699274303e-05} +{"train_loss": 0.01678554341197014, "grad_norm": 0.1109679639339447, "global_step": 11621, "epoch": 109, "lr": 8.859735303314596e-05} +{"train_loss": 0.017042245715856552, "grad_norm": 0.09640586376190186, "global_step": 11622, "epoch": 109, "lr": 8.859546893793456e-05} +{"train_loss": 0.015974057838320732, "grad_norm": 0.12250661104917526, "global_step": 11623, "epoch": 109, "lr": 8.859358470711547e-05} +{"train_loss": 0.016714509576559067, "grad_norm": 0.1878608614206314, "global_step": 11624, "epoch": 109, "lr": 8.859170034069533e-05} +{"train_loss": 0.01700192503631115, "grad_norm": 0.07769644260406494, "global_step": 11625, "epoch": 109, "lr": 8.858981583868071e-05} +{"train_loss": 0.017215093597769737, "grad_norm": 0.0802028477191925, "global_step": 11626, "epoch": 109, "lr": 8.858793120107828e-05} +{"train_loss": 0.017069950699806213, "grad_norm": 0.07805021107196808, "global_step": 11627, "epoch": 109, "lr": 8.858604642789463e-05} +{"train_loss": 0.01860717125236988, "grad_norm": 0.17299462854862213, "global_step": 11628, "epoch": 109, "lr": 8.85841615191364e-05} +{"train_loss": 0.01621975004673004, "grad_norm": 0.04732523113489151, "global_step": 11629, "epoch": 109, "lr": 8.85822764748102e-05} +{"train_loss": 0.015537474304437637, "grad_norm": 0.052260007709264755, "global_step": 11630, "epoch": 109, "lr": 8.858039129492266e-05} +{"train_loss": 0.020214131101965904, "grad_norm": 0.1257781833410263, "global_step": 11631, "epoch": 109, "lr": 8.85785059794804e-05} +{"train_loss": 0.017122915014624596, "grad_norm": 0.048177991062402725, "global_step": 11632, "epoch": 109, "lr": 8.857662052849007e-05} +{"train_loss": 0.01863001100718975, "grad_norm": 0.05684790760278702, "global_step": 11633, "epoch": 109, "lr": 8.857473494195825e-05} +{"train_loss": 0.017994049936532974, "grad_norm": 0.048609621822834015, "global_step": 11634, "epoch": 109, "lr": 8.85728492198916e-05} +{"train_loss": 0.018049482256174088, "grad_norm": 0.07012750953435898, "global_step": 11635, "epoch": 109, "lr": 8.857096336229672e-05} +{"train_loss": 0.015781685709953308, "grad_norm": 0.11447776854038239, "global_step": 11636, "epoch": 109, "lr": 8.856907736918025e-05} +{"train_loss": 0.01873779296875, "grad_norm": 0.10775334388017654, "global_step": 11637, "epoch": 109, "lr": 8.856719124054882e-05} +{"train_loss": 0.01892329752445221, "grad_norm": 0.08815880119800568, "global_step": 11638, "epoch": 109, "lr": 8.856530497640905e-05} +{"train_loss": 0.01627330295741558, "grad_norm": 0.06945917755365372, "global_step": 11639, "epoch": 109, "lr": 8.856341857676758e-05} +{"train_loss": 0.016529694199562073, "grad_norm": 0.10641457885503769, "global_step": 11640, "epoch": 109, "lr": 8.8561532041631e-05} +{"train_loss": 0.016501054167747498, "grad_norm": 0.09653263539075851, "global_step": 11641, "epoch": 109, "lr": 8.8559645371006e-05} +{"train_loss": 0.019191410392522812, "grad_norm": 0.05517827346920967, "global_step": 11642, "epoch": 109, "lr": 8.855775856489917e-05} +{"train_loss": 0.01758693903684616, "grad_norm": 0.06066778302192688, "global_step": 11643, "epoch": 109, "lr": 8.855587162331712e-05} +{"train_loss": 0.019800826907157898, "grad_norm": 0.09299930185079575, "global_step": 11644, "epoch": 109, "lr": 8.855398454626653e-05} +{"train_loss": 0.016429051756858826, "grad_norm": 0.06746123731136322, "global_step": 11645, "epoch": 109, "lr": 8.855209733375399e-05} +{"train_loss": 0.017526816576719284, "grad_norm": 0.08701370656490326, "global_step": 11646, "epoch": 109, "lr": 8.855020998578616e-05} +{"train_loss": 0.018675019964575768, "grad_norm": 0.06435307115316391, "global_step": 11647, "epoch": 109, "lr": 8.854832250236964e-05} +{"train_loss": 0.014957033097743988, "grad_norm": 0.10130773484706879, "global_step": 11648, "epoch": 109, "lr": 8.854643488351108e-05} +{"train_loss": 0.017160722985863686, "grad_norm": 0.08823344856500626, "global_step": 11649, "epoch": 109, "lr": 8.854454712921712e-05} +{"train_loss": 0.01652967929840088, "grad_norm": 0.05482475832104683, "global_step": 11650, "epoch": 109, "lr": 8.854265923949438e-05} +{"train_loss": 0.019000250846147537, "grad_norm": 0.09139085561037064, "global_step": 11651, "epoch": 109, "lr": 8.85407712143495e-05} +{"train_loss": 0.01550644263625145, "grad_norm": 0.04352511465549469, "global_step": 11652, "epoch": 109, "lr": 8.853888305378912e-05} +{"train_loss": 0.01613810658454895, "grad_norm": 0.12134190648794174, "global_step": 11653, "epoch": 109, "lr": 8.853699475781985e-05} +{"train_loss": 0.01728561893105507, "grad_norm": 0.12278661876916885, "global_step": 11654, "epoch": 109, "lr": 8.853510632644834e-05} +{"train_loss": 0.015426929108798504, "grad_norm": 0.08352687954902649, "global_step": 11655, "epoch": 109, "lr": 8.853321775968125e-05} +{"train_loss": 0.01682605966925621, "grad_norm": 0.15418250858783722, "global_step": 11656, "epoch": 109, "lr": 8.853132905752516e-05} +{"train_loss": 0.018271181732416153, "grad_norm": 0.09288378059864044, "global_step": 11657, "epoch": 109, "lr": 8.852944021998674e-05} +{"train_loss": 0.015673477202653885, "grad_norm": 0.047975409775972366, "global_step": 11658, "epoch": 109, "lr": 8.852755124707263e-05} +{"train_loss": 0.017389196761936513, "grad_norm": 0.09225842356681824, "global_step": 11659, "epoch": 109, "lr": 8.852566213878947e-05, "val_loss": 0.019008873030543327} +{"train_loss": 0.014140412211418152, "grad_norm": 0.0559680201113224, "global_step": 11660, "epoch": 110, "lr": 8.852377289514387e-05} +{"train_loss": 0.016432741656899452, "grad_norm": 0.06354090571403503, "global_step": 11661, "epoch": 110, "lr": 8.85218835161425e-05} +{"train_loss": 0.01482801791280508, "grad_norm": 0.04069092497229576, "global_step": 11662, "epoch": 110, "lr": 8.851999400179198e-05} +{"train_loss": 0.019702168181538582, "grad_norm": 0.05214860662817955, "global_step": 11663, "epoch": 110, "lr": 8.851810435209895e-05} +{"train_loss": 0.019234202802181244, "grad_norm": 0.05872706323862076, "global_step": 11664, "epoch": 110, "lr": 8.851621456707005e-05} +{"train_loss": 0.017700452357530594, "grad_norm": 0.0529322549700737, "global_step": 11665, "epoch": 110, "lr": 8.851432464671192e-05} +{"train_loss": 0.016123343259096146, "grad_norm": 0.05445634946227074, "global_step": 11666, "epoch": 110, "lr": 8.851243459103121e-05} +{"train_loss": 0.015962641686201096, "grad_norm": 0.12677118182182312, "global_step": 11667, "epoch": 110, "lr": 8.851054440003455e-05} +{"train_loss": 0.018792325630784035, "grad_norm": 0.14167256653308868, "global_step": 11668, "epoch": 110, "lr": 8.85086540737286e-05} +{"train_loss": 0.017791427671909332, "grad_norm": 0.17015109956264496, "global_step": 11669, "epoch": 110, "lr": 8.850676361211996e-05} +{"train_loss": 0.017925439402461052, "grad_norm": 0.05130475014448166, "global_step": 11670, "epoch": 110, "lr": 8.850487301521531e-05} +{"train_loss": 0.017093058675527573, "grad_norm": 0.11702114343643188, "global_step": 11671, "epoch": 110, "lr": 8.850298228302128e-05} +{"train_loss": 0.019178153946995735, "grad_norm": 0.057540956884622574, "global_step": 11672, "epoch": 110, "lr": 8.850109141554453e-05} +{"train_loss": 0.015210680663585663, "grad_norm": 0.08408103138208389, "global_step": 11673, "epoch": 110, "lr": 8.849920041279168e-05} +{"train_loss": 0.015924034640192986, "grad_norm": 0.05110457167029381, "global_step": 11674, "epoch": 110, "lr": 8.849730927476937e-05} +{"train_loss": 0.014207052066922188, "grad_norm": 0.07775268703699112, "global_step": 11675, "epoch": 110, "lr": 8.849541800148427e-05} +{"train_loss": 0.014593026600778103, "grad_norm": 0.051400359719991684, "global_step": 11676, "epoch": 110, "lr": 8.8493526592943e-05} +{"train_loss": 0.016321692615747452, "grad_norm": 0.08802549540996552, "global_step": 11677, "epoch": 110, "lr": 8.849163504915224e-05} +{"train_loss": 0.015231668949127197, "grad_norm": 0.1486636847257614, "global_step": 11678, "epoch": 110, "lr": 8.84897433701186e-05} +{"train_loss": 0.016012301668524742, "grad_norm": 0.047222573310136795, "global_step": 11679, "epoch": 110, "lr": 8.848785155584874e-05} +{"train_loss": 0.017085518687963486, "grad_norm": 0.06452207267284393, "global_step": 11680, "epoch": 110, "lr": 8.84859596063493e-05} +{"train_loss": 0.01594841480255127, "grad_norm": 0.04164605960249901, "global_step": 11681, "epoch": 110, "lr": 8.848406752162695e-05} +{"train_loss": 0.017801858484745026, "grad_norm": 0.07986699789762497, "global_step": 11682, "epoch": 110, "lr": 8.848217530168832e-05} +{"train_loss": 0.017026713117957115, "grad_norm": 0.10372531414031982, "global_step": 11683, "epoch": 110, "lr": 8.848028294654005e-05} +{"train_loss": 0.017410989850759506, "grad_norm": 0.048918962478637695, "global_step": 11684, "epoch": 110, "lr": 8.847839045618881e-05} +{"train_loss": 0.01609996147453785, "grad_norm": 0.07010059058666229, "global_step": 11685, "epoch": 110, "lr": 8.847649783064124e-05} +{"train_loss": 0.018412448465824127, "grad_norm": 0.05265146121382713, "global_step": 11686, "epoch": 110, "lr": 8.847460506990398e-05} +{"train_loss": 0.014702566899359226, "grad_norm": 0.12131760269403458, "global_step": 11687, "epoch": 110, "lr": 8.84727121739837e-05} +{"train_loss": 0.017736833542585373, "grad_norm": 0.0841357558965683, "global_step": 11688, "epoch": 110, "lr": 8.847081914288704e-05} +{"train_loss": 0.017289571464061737, "grad_norm": 0.06442524492740631, "global_step": 11689, "epoch": 110, "lr": 8.846892597662064e-05} +{"train_loss": 0.019117213785648346, "grad_norm": 0.056938305497169495, "global_step": 11690, "epoch": 110, "lr": 8.846703267519118e-05} +{"train_loss": 0.016859589144587517, "grad_norm": 0.06339847296476364, "global_step": 11691, "epoch": 110, "lr": 8.846513923860529e-05} +{"train_loss": 0.01682935282588005, "grad_norm": 0.1200900599360466, "global_step": 11692, "epoch": 110, "lr": 8.846324566686963e-05} +{"train_loss": 0.01750493422150612, "grad_norm": 0.08157286047935486, "global_step": 11693, "epoch": 110, "lr": 8.846135195999086e-05} +{"train_loss": 0.01724110171198845, "grad_norm": 0.07058454304933548, "global_step": 11694, "epoch": 110, "lr": 8.845945811797562e-05} +{"train_loss": 0.0199359692633152, "grad_norm": 0.06025925278663635, "global_step": 11695, "epoch": 110, "lr": 8.845756414083055e-05} +{"train_loss": 0.01740497536957264, "grad_norm": 0.06287810951471329, "global_step": 11696, "epoch": 110, "lr": 8.845567002856234e-05} +{"train_loss": 0.01947360299527645, "grad_norm": 0.08493704348802567, "global_step": 11697, "epoch": 110, "lr": 8.845377578117763e-05} +{"train_loss": 0.01754678785800934, "grad_norm": 0.06603547930717468, "global_step": 11698, "epoch": 110, "lr": 8.845188139868308e-05} +{"train_loss": 0.016212906688451767, "grad_norm": 0.07249380648136139, "global_step": 11699, "epoch": 110, "lr": 8.844998688108535e-05} +{"train_loss": 0.01813841052353382, "grad_norm": 0.07780305296182632, "global_step": 11700, "epoch": 110, "lr": 8.844809222839107e-05} +{"train_loss": 0.015593219548463821, "grad_norm": 0.11386179178953171, "global_step": 11701, "epoch": 110, "lr": 8.844619744060691e-05} +{"train_loss": 0.015701312571763992, "grad_norm": 0.13014273345470428, "global_step": 11702, "epoch": 110, "lr": 8.844430251773955e-05} +{"train_loss": 0.01589328981935978, "grad_norm": 0.08711502701044083, "global_step": 11703, "epoch": 110, "lr": 8.844240745979562e-05} +{"train_loss": 0.0178165715187788, "grad_norm": 0.058691006153821945, "global_step": 11704, "epoch": 110, "lr": 8.844051226678178e-05} +{"train_loss": 0.01942422054708004, "grad_norm": 0.05154519900679588, "global_step": 11705, "epoch": 110, "lr": 8.84386169387047e-05} +{"train_loss": 0.014957117848098278, "grad_norm": 0.06409654021263123, "global_step": 11706, "epoch": 110, "lr": 8.843672147557105e-05} +{"train_loss": 0.018703307956457138, "grad_norm": 0.08913664519786835, "global_step": 11707, "epoch": 110, "lr": 8.843482587738747e-05} +{"train_loss": 0.015351582318544388, "grad_norm": 0.09106242656707764, "global_step": 11708, "epoch": 110, "lr": 8.843293014416062e-05} +{"train_loss": 0.017072875052690506, "grad_norm": 0.06517777591943741, "global_step": 11709, "epoch": 110, "lr": 8.843103427589718e-05} +{"train_loss": 0.016156718134880066, "grad_norm": 0.06927751004695892, "global_step": 11710, "epoch": 110, "lr": 8.842913827260379e-05} +{"train_loss": 0.01742703467607498, "grad_norm": 0.05295272544026375, "global_step": 11711, "epoch": 110, "lr": 8.842724213428712e-05} +{"train_loss": 0.01445925422012806, "grad_norm": 0.1170729398727417, "global_step": 11712, "epoch": 110, "lr": 8.842534586095383e-05} +{"train_loss": 0.01648813486099243, "grad_norm": 0.12243646383285522, "global_step": 11713, "epoch": 110, "lr": 8.842344945261059e-05} +{"train_loss": 0.014186648651957512, "grad_norm": 0.06248882785439491, "global_step": 11714, "epoch": 110, "lr": 8.842155290926405e-05} +{"train_loss": 0.016904424875974655, "grad_norm": 0.05854456499218941, "global_step": 11715, "epoch": 110, "lr": 8.84196562309209e-05} +{"train_loss": 0.017649518325924873, "grad_norm": 0.13714942336082458, "global_step": 11716, "epoch": 110, "lr": 8.841775941758779e-05} +{"train_loss": 0.020677758380770683, "grad_norm": 0.048413023352622986, "global_step": 11717, "epoch": 110, "lr": 8.841586246927136e-05} +{"train_loss": 0.01604044809937477, "grad_norm": 0.11586155742406845, "global_step": 11718, "epoch": 110, "lr": 8.84139653859783e-05} +{"train_loss": 0.020164266228675842, "grad_norm": 0.07063209265470505, "global_step": 11719, "epoch": 110, "lr": 8.841206816771528e-05} +{"train_loss": 0.01753191649913788, "grad_norm": 0.04781360924243927, "global_step": 11720, "epoch": 110, "lr": 8.841017081448895e-05} +{"train_loss": 0.01770513877272606, "grad_norm": 0.09431078284978867, "global_step": 11721, "epoch": 110, "lr": 8.840827332630599e-05} +{"train_loss": 0.016932412981987, "grad_norm": 0.12776359915733337, "global_step": 11722, "epoch": 110, "lr": 8.840637570317306e-05} +{"train_loss": 0.01832094043493271, "grad_norm": 0.08767752349376678, "global_step": 11723, "epoch": 110, "lr": 8.840447794509683e-05} +{"train_loss": 0.016176367178559303, "grad_norm": 0.07136339694261551, "global_step": 11724, "epoch": 110, "lr": 8.840258005208395e-05} +{"train_loss": 0.01740984059870243, "grad_norm": 0.10988275706768036, "global_step": 11725, "epoch": 110, "lr": 8.840068202414112e-05} +{"train_loss": 0.020850222557783127, "grad_norm": 0.17503558099269867, "global_step": 11726, "epoch": 110, "lr": 8.8398783861275e-05} +{"train_loss": 0.018001198768615723, "grad_norm": 0.0821099728345871, "global_step": 11727, "epoch": 110, "lr": 8.839688556349223e-05} +{"train_loss": 0.01828014850616455, "grad_norm": 0.07228967547416687, "global_step": 11728, "epoch": 110, "lr": 8.839498713079953e-05} +{"train_loss": 0.019849952310323715, "grad_norm": 0.05020338296890259, "global_step": 11729, "epoch": 110, "lr": 8.839308856320352e-05} +{"train_loss": 0.01723150908946991, "grad_norm": 0.11699050664901733, "global_step": 11730, "epoch": 110, "lr": 8.839118986071091e-05} +{"train_loss": 0.01821187138557434, "grad_norm": 0.06443633884191513, "global_step": 11731, "epoch": 110, "lr": 8.838929102332833e-05} +{"train_loss": 0.01597735472023487, "grad_norm": 0.10296986252069473, "global_step": 11732, "epoch": 110, "lr": 8.83873920510625e-05} +{"train_loss": 0.020823078230023384, "grad_norm": 0.09467706084251404, "global_step": 11733, "epoch": 110, "lr": 8.838549294392007e-05} +{"train_loss": 0.017456550151109695, "grad_norm": 0.04138515889644623, "global_step": 11734, "epoch": 110, "lr": 8.83835937019077e-05} +{"train_loss": 0.014607689343392849, "grad_norm": 0.08765743672847748, "global_step": 11735, "epoch": 110, "lr": 8.838169432503207e-05} +{"train_loss": 0.017715314403176308, "grad_norm": 0.06222357600927353, "global_step": 11736, "epoch": 110, "lr": 8.837979481329987e-05} +{"train_loss": 0.02028582990169525, "grad_norm": 0.059936363250017166, "global_step": 11737, "epoch": 110, "lr": 8.837789516671775e-05} +{"train_loss": 0.016542701050639153, "grad_norm": 0.07829445600509644, "global_step": 11738, "epoch": 110, "lr": 8.837599538529241e-05} +{"train_loss": 0.019536318257451057, "grad_norm": 0.10053142160177231, "global_step": 11739, "epoch": 110, "lr": 8.83740954690305e-05} +{"train_loss": 0.016502827405929565, "grad_norm": 0.1700344979763031, "global_step": 11740, "epoch": 110, "lr": 8.837219541793871e-05} +{"train_loss": 0.01673472858965397, "grad_norm": 0.11646149307489395, "global_step": 11741, "epoch": 110, "lr": 8.837029523202372e-05} +{"train_loss": 0.019084572792053223, "grad_norm": 0.12984757125377655, "global_step": 11742, "epoch": 110, "lr": 8.836839491129219e-05} +{"train_loss": 0.01689804345369339, "grad_norm": 0.06584449857473373, "global_step": 11743, "epoch": 110, "lr": 8.83664944557508e-05} +{"train_loss": 0.016874592751264572, "grad_norm": 0.10658694803714752, "global_step": 11744, "epoch": 110, "lr": 8.836459386540626e-05} +{"train_loss": 0.016158640384674072, "grad_norm": 0.05229422450065613, "global_step": 11745, "epoch": 110, "lr": 8.83626931402652e-05} +{"train_loss": 0.019781701266765594, "grad_norm": 0.1518198400735855, "global_step": 11746, "epoch": 110, "lr": 8.836079228033433e-05} +{"train_loss": 0.017096761614084244, "grad_norm": 0.1044103279709816, "global_step": 11747, "epoch": 110, "lr": 8.835889128562031e-05} +{"train_loss": 0.01704278215765953, "grad_norm": 0.11683674901723862, "global_step": 11748, "epoch": 110, "lr": 8.835699015612984e-05} +{"train_loss": 0.020332425832748413, "grad_norm": 0.06444251537322998, "global_step": 11749, "epoch": 110, "lr": 8.835508889186956e-05} +{"train_loss": 0.016105469316244125, "grad_norm": 0.07501254230737686, "global_step": 11750, "epoch": 110, "lr": 8.835318749284621e-05} +{"train_loss": 0.018107879906892776, "grad_norm": 0.08501435071229935, "global_step": 11751, "epoch": 110, "lr": 8.835128595906644e-05} +{"train_loss": 0.016378749161958694, "grad_norm": 0.2395593822002411, "global_step": 11752, "epoch": 110, "lr": 8.83493842905369e-05} +{"train_loss": 0.016141369938850403, "grad_norm": 0.24241235852241516, "global_step": 11753, "epoch": 110, "lr": 8.834748248726432e-05} +{"train_loss": 0.020570814609527588, "grad_norm": 0.14756745100021362, "global_step": 11754, "epoch": 110, "lr": 8.834558054925535e-05} +{"train_loss": 0.017337488010525703, "grad_norm": 0.09874904155731201, "global_step": 11755, "epoch": 110, "lr": 8.83436784765167e-05} +{"train_loss": 0.017732810229063034, "grad_norm": 0.07757484167814255, "global_step": 11756, "epoch": 110, "lr": 8.834177626905504e-05} +{"train_loss": 0.017822692170739174, "grad_norm": 0.13835817575454712, "global_step": 11757, "epoch": 110, "lr": 8.833987392687704e-05} +{"train_loss": 0.018591156229376793, "grad_norm": 0.07050534337759018, "global_step": 11758, "epoch": 110, "lr": 8.833797144998941e-05} +{"train_loss": 0.017690088599920273, "grad_norm": 0.06615351885557175, "global_step": 11759, "epoch": 110, "lr": 8.833606883839881e-05} +{"train_loss": 0.018382880836725235, "grad_norm": 0.0800793319940567, "global_step": 11760, "epoch": 110, "lr": 8.833416609211193e-05} +{"train_loss": 0.0166617501527071, "grad_norm": 0.06352435797452927, "global_step": 11761, "epoch": 110, "lr": 8.833226321113548e-05} +{"train_loss": 0.016578858718276024, "grad_norm": 0.05375640466809273, "global_step": 11762, "epoch": 110, "lr": 8.833036019547614e-05} +{"train_loss": 0.018697839230298996, "grad_norm": 0.0498347245156765, "global_step": 11763, "epoch": 110, "lr": 8.832845704514054e-05} +{"train_loss": 0.021479375660419464, "grad_norm": 0.057591188699007034, "global_step": 11764, "epoch": 110, "lr": 8.832655376013544e-05} +{"train_loss": 0.01732889176258501, "grad_norm": 0.058975037187337875, "global_step": 11765, "epoch": 110, "lr": 8.832465034046749e-05, "train_batch_mse_error": 0.01562616229057312, "val_loss": 0.019211843609809875} +{"train_loss": 0.01863139122724533, "grad_norm": 0.05432457476854324, "global_step": 11766, "epoch": 111, "lr": 8.83227467861434e-05} +{"train_loss": 0.015794096514582634, "grad_norm": 0.08735714107751846, "global_step": 11767, "epoch": 111, "lr": 8.832084309716983e-05} +{"train_loss": 0.017485640943050385, "grad_norm": 0.10238001495599747, "global_step": 11768, "epoch": 111, "lr": 8.831893927355348e-05} +{"train_loss": 0.02120772935450077, "grad_norm": 0.08413432538509369, "global_step": 11769, "epoch": 111, "lr": 8.831703531530105e-05} +{"train_loss": 0.0159501601010561, "grad_norm": 0.06911633163690567, "global_step": 11770, "epoch": 111, "lr": 8.831513122241921e-05} +{"train_loss": 0.018141429871320724, "grad_norm": 0.11283010244369507, "global_step": 11771, "epoch": 111, "lr": 8.831322699491467e-05} +{"train_loss": 0.016813956201076508, "grad_norm": 0.12241977453231812, "global_step": 11772, "epoch": 111, "lr": 8.831132263279411e-05} +{"train_loss": 0.015613645315170288, "grad_norm": 0.07133826613426208, "global_step": 11773, "epoch": 111, "lr": 8.830941813606424e-05} +{"train_loss": 0.016953691840171814, "grad_norm": 0.1007152646780014, "global_step": 11774, "epoch": 111, "lr": 8.83075135047317e-05} +{"train_loss": 0.01937401294708252, "grad_norm": 0.05013243854045868, "global_step": 11775, "epoch": 111, "lr": 8.830560873880325e-05} +{"train_loss": 0.01772666722536087, "grad_norm": 0.06695841252803802, "global_step": 11776, "epoch": 111, "lr": 8.830370383828553e-05} +{"train_loss": 0.017284266650676727, "grad_norm": 0.14366301894187927, "global_step": 11777, "epoch": 111, "lr": 8.830179880318527e-05} +{"train_loss": 0.017281735315918922, "grad_norm": 0.09122230112552643, "global_step": 11778, "epoch": 111, "lr": 8.829989363350913e-05} +{"train_loss": 0.01918504387140274, "grad_norm": 0.05923350900411606, "global_step": 11779, "epoch": 111, "lr": 8.829798832926383e-05} +{"train_loss": 0.017833162099123, "grad_norm": 0.12275710701942444, "global_step": 11780, "epoch": 111, "lr": 8.829608289045604e-05} +{"train_loss": 0.015455623157322407, "grad_norm": 0.06771291047334671, "global_step": 11781, "epoch": 111, "lr": 8.829417731709249e-05} +{"train_loss": 0.015924084931612015, "grad_norm": 0.040807463228702545, "global_step": 11782, "epoch": 111, "lr": 8.829227160917984e-05} +{"train_loss": 0.01686139963567257, "grad_norm": 0.07373897731304169, "global_step": 11783, "epoch": 111, "lr": 8.82903657667248e-05} +{"train_loss": 0.016584403812885284, "grad_norm": 0.05535321682691574, "global_step": 11784, "epoch": 111, "lr": 8.828845978973407e-05} +{"train_loss": 0.015916965901851654, "grad_norm": 0.15059134364128113, "global_step": 11785, "epoch": 111, "lr": 8.828655367821435e-05} +{"train_loss": 0.01551819872111082, "grad_norm": 0.11003945767879486, "global_step": 11786, "epoch": 111, "lr": 8.828464743217232e-05} +{"train_loss": 0.016548091545701027, "grad_norm": 0.05750695988535881, "global_step": 11787, "epoch": 111, "lr": 8.828274105161469e-05} +{"train_loss": 0.017417551949620247, "grad_norm": 0.1316293627023697, "global_step": 11788, "epoch": 111, "lr": 8.828083453654816e-05} +{"train_loss": 0.01795710250735283, "grad_norm": 0.14057688415050507, "global_step": 11789, "epoch": 111, "lr": 8.827892788697942e-05} +{"train_loss": 0.016878390684723854, "grad_norm": 0.06827076524496078, "global_step": 11790, "epoch": 111, "lr": 8.827702110291519e-05} +{"train_loss": 0.017788855358958244, "grad_norm": 0.10256914794445038, "global_step": 11791, "epoch": 111, "lr": 8.827511418436214e-05} +{"train_loss": 0.017717394977808, "grad_norm": 0.18824945390224457, "global_step": 11792, "epoch": 111, "lr": 8.827320713132698e-05} +{"train_loss": 0.017645670101046562, "grad_norm": 0.06351030617952347, "global_step": 11793, "epoch": 111, "lr": 8.827129994381643e-05} +{"train_loss": 0.01765940897166729, "grad_norm": 0.10156441479921341, "global_step": 11794, "epoch": 111, "lr": 8.826939262183717e-05} +{"train_loss": 0.01442989706993103, "grad_norm": 0.08384905755519867, "global_step": 11795, "epoch": 111, "lr": 8.826748516539591e-05} +{"train_loss": 0.018868740648031235, "grad_norm": 0.09303715825080872, "global_step": 11796, "epoch": 111, "lr": 8.826557757449935e-05} +{"train_loss": 0.01883062534034252, "grad_norm": 0.08844298869371414, "global_step": 11797, "epoch": 111, "lr": 8.826366984915419e-05} +{"train_loss": 0.018369050696492195, "grad_norm": 0.07059987634420395, "global_step": 11798, "epoch": 111, "lr": 8.826176198936713e-05} +{"train_loss": 0.01646866463124752, "grad_norm": 0.05103262886404991, "global_step": 11799, "epoch": 111, "lr": 8.825985399514487e-05} +{"train_loss": 0.016496948897838593, "grad_norm": 0.09585655480623245, "global_step": 11800, "epoch": 111, "lr": 8.825794586649414e-05} +{"train_loss": 0.017703820019960403, "grad_norm": 0.06840166449546814, "global_step": 11801, "epoch": 111, "lr": 8.82560376034216e-05} +{"train_loss": 0.015565835870802402, "grad_norm": 0.08663346618413925, "global_step": 11802, "epoch": 111, "lr": 8.825412920593401e-05} +{"train_loss": 0.017577677965164185, "grad_norm": 0.08680901676416397, "global_step": 11803, "epoch": 111, "lr": 8.825222067403804e-05} +{"train_loss": 0.015789944678544998, "grad_norm": 0.12395045161247253, "global_step": 11804, "epoch": 111, "lr": 8.82503120077404e-05} +{"train_loss": 0.014804387465119362, "grad_norm": 0.08034609258174896, "global_step": 11805, "epoch": 111, "lr": 8.824840320704779e-05} +{"train_loss": 0.017636911943554878, "grad_norm": 0.07090511173009872, "global_step": 11806, "epoch": 111, "lr": 8.824649427196692e-05} +{"train_loss": 0.018495304509997368, "grad_norm": 0.08828919380903244, "global_step": 11807, "epoch": 111, "lr": 8.824458520250451e-05} +{"train_loss": 0.017141971737146378, "grad_norm": 0.07778842747211456, "global_step": 11808, "epoch": 111, "lr": 8.824267599866727e-05} +{"train_loss": 0.014428106136620045, "grad_norm": 0.0656077042222023, "global_step": 11809, "epoch": 111, "lr": 8.824076666046188e-05} +{"train_loss": 0.017774712294340134, "grad_norm": 0.091944120824337, "global_step": 11810, "epoch": 111, "lr": 8.823885718789509e-05} +{"train_loss": 0.018304862082004547, "grad_norm": 0.0633530542254448, "global_step": 11811, "epoch": 111, "lr": 8.823694758097357e-05} +{"train_loss": 0.01688513346016407, "grad_norm": 0.05868193134665489, "global_step": 11812, "epoch": 111, "lr": 8.823503783970404e-05} +{"train_loss": 0.01675102859735489, "grad_norm": 0.05659352242946625, "global_step": 11813, "epoch": 111, "lr": 8.823312796409323e-05} +{"train_loss": 0.016072183847427368, "grad_norm": 0.07715370506048203, "global_step": 11814, "epoch": 111, "lr": 8.823121795414782e-05} +{"train_loss": 0.015047435648739338, "grad_norm": 0.08882305771112442, "global_step": 11815, "epoch": 111, "lr": 8.822930780987454e-05} +{"train_loss": 0.017426617443561554, "grad_norm": 0.07202401012182236, "global_step": 11816, "epoch": 111, "lr": 8.82273975312801e-05} +{"train_loss": 0.014231503941118717, "grad_norm": 0.09257643669843674, "global_step": 11817, "epoch": 111, "lr": 8.822548711837121e-05} +{"train_loss": 0.019948963075876236, "grad_norm": 0.12051110714673996, "global_step": 11818, "epoch": 111, "lr": 8.822357657115459e-05} +{"train_loss": 0.016755761578679085, "grad_norm": 0.20869703590869904, "global_step": 11819, "epoch": 111, "lr": 8.822166588963694e-05} +{"train_loss": 0.013894444331526756, "grad_norm": 0.07459938526153564, "global_step": 11820, "epoch": 111, "lr": 8.821975507382496e-05} +{"train_loss": 0.017113463953137398, "grad_norm": 0.05834883451461792, "global_step": 11821, "epoch": 111, "lr": 8.82178441237254e-05} +{"train_loss": 0.015586039051413536, "grad_norm": 0.05843157693743706, "global_step": 11822, "epoch": 111, "lr": 8.821593303934497e-05} +{"train_loss": 0.01819584146142006, "grad_norm": 0.09252557158470154, "global_step": 11823, "epoch": 111, "lr": 8.821402182069033e-05} +{"train_loss": 0.01566489413380623, "grad_norm": 0.06360404938459396, "global_step": 11824, "epoch": 111, "lr": 8.821211046776828e-05} +{"train_loss": 0.01646379940211773, "grad_norm": 0.07176057994365692, "global_step": 11825, "epoch": 111, "lr": 8.821019898058546e-05} +{"train_loss": 0.019282706081867218, "grad_norm": 0.10333910584449768, "global_step": 11826, "epoch": 111, "lr": 8.820828735914865e-05} +{"train_loss": 0.01797252520918846, "grad_norm": 0.04966193810105324, "global_step": 11827, "epoch": 111, "lr": 8.820637560346451e-05} +{"train_loss": 0.0193558968603611, "grad_norm": 0.12566618621349335, "global_step": 11828, "epoch": 111, "lr": 8.820446371353979e-05} +{"train_loss": 0.017719823867082596, "grad_norm": 0.050928082317113876, "global_step": 11829, "epoch": 111, "lr": 8.820255168938118e-05} +{"train_loss": 0.017698444426059723, "grad_norm": 0.05733036622405052, "global_step": 11830, "epoch": 111, "lr": 8.820063953099544e-05} +{"train_loss": 0.016919340938329697, "grad_norm": 0.04711344465613365, "global_step": 11831, "epoch": 111, "lr": 8.819872723838927e-05} +{"train_loss": 0.018177347257733345, "grad_norm": 0.06381876766681671, "global_step": 11832, "epoch": 111, "lr": 8.819681481156938e-05} +{"train_loss": 0.017999621108174324, "grad_norm": 0.05136759206652641, "global_step": 11833, "epoch": 111, "lr": 8.819490225054248e-05} +{"train_loss": 0.016623571515083313, "grad_norm": 0.04019440338015556, "global_step": 11834, "epoch": 111, "lr": 8.81929895553153e-05} +{"train_loss": 0.019594240933656693, "grad_norm": 0.060820892453193665, "global_step": 11835, "epoch": 111, "lr": 8.819107672589458e-05} +{"train_loss": 0.018342019990086555, "grad_norm": 0.0723971351981163, "global_step": 11836, "epoch": 111, "lr": 8.818916376228703e-05} +{"train_loss": 0.019272126257419586, "grad_norm": 0.07154986262321472, "global_step": 11837, "epoch": 111, "lr": 8.818725066449936e-05} +{"train_loss": 0.016404878348112106, "grad_norm": 0.06476528197526932, "global_step": 11838, "epoch": 111, "lr": 8.81853374325383e-05} +{"train_loss": 0.01636580377817154, "grad_norm": 0.12680433690547943, "global_step": 11839, "epoch": 111, "lr": 8.818342406641056e-05} +{"train_loss": 0.015218068845570087, "grad_norm": 0.04687540605664253, "global_step": 11840, "epoch": 111, "lr": 8.818151056612288e-05} +{"train_loss": 0.017003685235977173, "grad_norm": 0.0669928789138794, "global_step": 11841, "epoch": 111, "lr": 8.817959693168198e-05} +{"train_loss": 0.015466907061636448, "grad_norm": 0.0640367791056633, "global_step": 11842, "epoch": 111, "lr": 8.817768316309458e-05} +{"train_loss": 0.014382051303982735, "grad_norm": 0.049514371901750565, "global_step": 11843, "epoch": 111, "lr": 8.817576926036739e-05} +{"train_loss": 0.016881033778190613, "grad_norm": 0.06365741789340973, "global_step": 11844, "epoch": 111, "lr": 8.817385522350716e-05} +{"train_loss": 0.015820112079381943, "grad_norm": 0.05806073546409607, "global_step": 11845, "epoch": 111, "lr": 8.817194105252061e-05} +{"train_loss": 0.01818564347922802, "grad_norm": 0.06196990981698036, "global_step": 11846, "epoch": 111, "lr": 8.817002674741445e-05} +{"train_loss": 0.015366666950285435, "grad_norm": 0.06978847831487656, "global_step": 11847, "epoch": 111, "lr": 8.816811230819543e-05} +{"train_loss": 0.019967660307884216, "grad_norm": 0.05804334580898285, "global_step": 11848, "epoch": 111, "lr": 8.816619773487025e-05} +{"train_loss": 0.01912182942032814, "grad_norm": 0.06101037189364433, "global_step": 11849, "epoch": 111, "lr": 8.816428302744566e-05} +{"train_loss": 0.017221933230757713, "grad_norm": 0.06643126159906387, "global_step": 11850, "epoch": 111, "lr": 8.816236818592836e-05} +{"train_loss": 0.017100928351283073, "grad_norm": 0.07370372861623764, "global_step": 11851, "epoch": 111, "lr": 8.816045321032511e-05} +{"train_loss": 0.019618326798081398, "grad_norm": 0.06899136304855347, "global_step": 11852, "epoch": 111, "lr": 8.815853810064261e-05} +{"train_loss": 0.017929445952177048, "grad_norm": 0.14059606194496155, "global_step": 11853, "epoch": 111, "lr": 8.815662285688762e-05} +{"train_loss": 0.019454501569271088, "grad_norm": 0.10698676854372025, "global_step": 11854, "epoch": 111, "lr": 8.815470747906684e-05} +{"train_loss": 0.018386077135801315, "grad_norm": 0.0968688577413559, "global_step": 11855, "epoch": 111, "lr": 8.8152791967187e-05} +{"train_loss": 0.01688912883400917, "grad_norm": 0.06896517425775528, "global_step": 11856, "epoch": 111, "lr": 8.815087632125485e-05} +{"train_loss": 0.017254360020160675, "grad_norm": 0.11187022179365158, "global_step": 11857, "epoch": 111, "lr": 8.814896054127712e-05} +{"train_loss": 0.017440099269151688, "grad_norm": 0.09805730730295181, "global_step": 11858, "epoch": 111, "lr": 8.814704462726051e-05} +{"train_loss": 0.017621904611587524, "grad_norm": 0.06442781537771225, "global_step": 11859, "epoch": 111, "lr": 8.81451285792118e-05} +{"train_loss": 0.01894105225801468, "grad_norm": 0.05657913163304329, "global_step": 11860, "epoch": 111, "lr": 8.814321239713769e-05} +{"train_loss": 0.019108900800347328, "grad_norm": 0.05918418616056442, "global_step": 11861, "epoch": 111, "lr": 8.814129608104492e-05} +{"train_loss": 0.01781369373202324, "grad_norm": 0.07152462005615234, "global_step": 11862, "epoch": 111, "lr": 8.813937963094023e-05} +{"train_loss": 0.01699705794453621, "grad_norm": 0.057806383818387985, "global_step": 11863, "epoch": 111, "lr": 8.813746304683033e-05} +{"train_loss": 0.02064455673098564, "grad_norm": 0.11590885370969772, "global_step": 11864, "epoch": 111, "lr": 8.813554632872198e-05} +{"train_loss": 0.01531388983130455, "grad_norm": 0.10997198522090912, "global_step": 11865, "epoch": 111, "lr": 8.813362947662189e-05} +{"train_loss": 0.017339371144771576, "grad_norm": 0.08357682824134827, "global_step": 11866, "epoch": 111, "lr": 8.813171249053684e-05} +{"train_loss": 0.019882328808307648, "grad_norm": 0.054331887513399124, "global_step": 11867, "epoch": 111, "lr": 8.81297953704735e-05} +{"train_loss": 0.01644100621342659, "grad_norm": 0.06462319940328598, "global_step": 11868, "epoch": 111, "lr": 8.812787811643867e-05} +{"train_loss": 0.017283041030168533, "grad_norm": 0.0514204204082489, "global_step": 11869, "epoch": 111, "lr": 8.812596072843904e-05} +{"train_loss": 0.01696312427520752, "grad_norm": 0.06016017496585846, "global_step": 11870, "epoch": 111, "lr": 8.812404320648137e-05} +{"train_loss": 0.01730290734437558, "grad_norm": 0.0830627828836441, "global_step": 11871, "epoch": 111, "lr": 8.81221255505724e-05, "val_loss": 0.018961625173687935} +{"train_loss": 0.01856105402112007, "grad_norm": 0.05546054244041443, "global_step": 11872, "epoch": 112, "lr": 8.812020776071886e-05} +{"train_loss": 0.01609150320291519, "grad_norm": 0.05624591186642647, "global_step": 11873, "epoch": 112, "lr": 8.811828983692747e-05} +{"train_loss": 0.01630520075559616, "grad_norm": 0.0720238983631134, "global_step": 11874, "epoch": 112, "lr": 8.8116371779205e-05} +{"train_loss": 0.017973676323890686, "grad_norm": 0.10862714797258377, "global_step": 11875, "epoch": 112, "lr": 8.811445358755817e-05} +{"train_loss": 0.0158524326980114, "grad_norm": 0.05685168132185936, "global_step": 11876, "epoch": 112, "lr": 8.811253526199373e-05} +{"train_loss": 0.015464727766811848, "grad_norm": 0.056934282183647156, "global_step": 11877, "epoch": 112, "lr": 8.811061680251843e-05} +{"train_loss": 0.019838696345686913, "grad_norm": 0.0713067352771759, "global_step": 11878, "epoch": 112, "lr": 8.810869820913896e-05} +{"train_loss": 0.018641119822859764, "grad_norm": 0.04352620989084244, "global_step": 11879, "epoch": 112, "lr": 8.810677948186213e-05} +{"train_loss": 0.017145300284028053, "grad_norm": 0.06620986014604568, "global_step": 11880, "epoch": 112, "lr": 8.810486062069463e-05} +{"train_loss": 0.017898080870509148, "grad_norm": 0.08265942335128784, "global_step": 11881, "epoch": 112, "lr": 8.810294162564323e-05} +{"train_loss": 0.01580491103231907, "grad_norm": 0.1322658509016037, "global_step": 11882, "epoch": 112, "lr": 8.810102249671467e-05} +{"train_loss": 0.014202828519046307, "grad_norm": 0.0632685050368309, "global_step": 11883, "epoch": 112, "lr": 8.809910323391568e-05} +{"train_loss": 0.019168712198734283, "grad_norm": 0.11286729574203491, "global_step": 11884, "epoch": 112, "lr": 8.8097183837253e-05} +{"train_loss": 0.017114635556936264, "grad_norm": 0.10571377724409103, "global_step": 11885, "epoch": 112, "lr": 8.80952643067334e-05} +{"train_loss": 0.01838824525475502, "grad_norm": 0.05680615082383156, "global_step": 11886, "epoch": 112, "lr": 8.809334464236361e-05} +{"train_loss": 0.01604284718632698, "grad_norm": 0.06759275496006012, "global_step": 11887, "epoch": 112, "lr": 8.809142484415036e-05} +{"train_loss": 0.014369556680321693, "grad_norm": 0.08405156433582306, "global_step": 11888, "epoch": 112, "lr": 8.808950491210041e-05} +{"train_loss": 0.016380172222852707, "grad_norm": 0.05957961454987526, "global_step": 11889, "epoch": 112, "lr": 8.808758484622051e-05} +{"train_loss": 0.01929878629744053, "grad_norm": 0.10744446516036987, "global_step": 11890, "epoch": 112, "lr": 8.80856646465174e-05} +{"train_loss": 0.015604699030518532, "grad_norm": 0.074243925511837, "global_step": 11891, "epoch": 112, "lr": 8.808374431299782e-05} +{"train_loss": 0.017379002645611763, "grad_norm": 0.0807795524597168, "global_step": 11892, "epoch": 112, "lr": 8.808182384566854e-05} +{"train_loss": 0.017354026436805725, "grad_norm": 0.05463191494345665, "global_step": 11893, "epoch": 112, "lr": 8.807990324453628e-05} +{"train_loss": 0.019929978996515274, "grad_norm": 0.1236456111073494, "global_step": 11894, "epoch": 112, "lr": 8.80779825096078e-05} +{"train_loss": 0.020413503050804138, "grad_norm": 0.1590641587972641, "global_step": 11895, "epoch": 112, "lr": 8.807606164088986e-05} +{"train_loss": 0.017053592950105667, "grad_norm": 0.05941641330718994, "global_step": 11896, "epoch": 112, "lr": 8.80741406383892e-05} +{"train_loss": 0.018400825560092926, "grad_norm": 0.14781850576400757, "global_step": 11897, "epoch": 112, "lr": 8.807221950211255e-05} +{"train_loss": 0.016228551045060158, "grad_norm": 0.06892138719558716, "global_step": 11898, "epoch": 112, "lr": 8.807029823206669e-05} +{"train_loss": 0.015537729486823082, "grad_norm": 0.061730653047561646, "global_step": 11899, "epoch": 112, "lr": 8.806837682825835e-05} +{"train_loss": 0.017506230622529984, "grad_norm": 0.07953976094722748, "global_step": 11900, "epoch": 112, "lr": 8.80664552906943e-05} +{"train_loss": 0.01772996038198471, "grad_norm": 0.15135252475738525, "global_step": 11901, "epoch": 112, "lr": 8.806453361938127e-05} +{"train_loss": 0.01588955894112587, "grad_norm": 0.09982209652662277, "global_step": 11902, "epoch": 112, "lr": 8.806261181432602e-05} +{"train_loss": 0.01825110614299774, "grad_norm": 0.06490633636713028, "global_step": 11903, "epoch": 112, "lr": 8.806068987553533e-05} +{"train_loss": 0.01972406730055809, "grad_norm": 0.08579153567552567, "global_step": 11904, "epoch": 112, "lr": 8.80587678030159e-05} +{"train_loss": 0.019676584750413895, "grad_norm": 0.10177470743656158, "global_step": 11905, "epoch": 112, "lr": 8.805684559677453e-05} +{"train_loss": 0.0167248472571373, "grad_norm": 0.07339135557413101, "global_step": 11906, "epoch": 112, "lr": 8.805492325681794e-05} +{"train_loss": 0.015249228104948997, "grad_norm": 0.12945428490638733, "global_step": 11907, "epoch": 112, "lr": 8.80530007831529e-05} +{"train_loss": 0.018035857006907463, "grad_norm": 0.15030710399150848, "global_step": 11908, "epoch": 112, "lr": 8.805107817578617e-05} +{"train_loss": 0.017833279445767403, "grad_norm": 0.12281692773103714, "global_step": 11909, "epoch": 112, "lr": 8.804915543472452e-05} +{"train_loss": 0.018236655741930008, "grad_norm": 0.07283671200275421, "global_step": 11910, "epoch": 112, "lr": 8.804723255997466e-05} +{"train_loss": 0.019464708864688873, "grad_norm": 0.05424292013049126, "global_step": 11911, "epoch": 112, "lr": 8.804530955154338e-05} +{"train_loss": 0.016484476625919342, "grad_norm": 0.15957011282444, "global_step": 11912, "epoch": 112, "lr": 8.804338640943742e-05} +{"train_loss": 0.01760992966592312, "grad_norm": 0.07483460009098053, "global_step": 11913, "epoch": 112, "lr": 8.804146313366355e-05} +{"train_loss": 0.015819530934095383, "grad_norm": 0.07115131616592407, "global_step": 11914, "epoch": 112, "lr": 8.803953972422851e-05} +{"train_loss": 0.016023259609937668, "grad_norm": 0.05087819695472717, "global_step": 11915, "epoch": 112, "lr": 8.803761618113908e-05} +{"train_loss": 0.01914234831929207, "grad_norm": 0.0507245771586895, "global_step": 11916, "epoch": 112, "lr": 8.803569250440203e-05} +{"train_loss": 0.0165341105312109, "grad_norm": 0.08781717717647552, "global_step": 11917, "epoch": 112, "lr": 8.803376869402407e-05} +{"train_loss": 0.02063705027103424, "grad_norm": 0.05949828773736954, "global_step": 11918, "epoch": 112, "lr": 8.803184475001198e-05} +{"train_loss": 0.017288092523813248, "grad_norm": 0.14214323461055756, "global_step": 11919, "epoch": 112, "lr": 8.802992067237255e-05} +{"train_loss": 0.01801973395049572, "grad_norm": 0.07698580622673035, "global_step": 11920, "epoch": 112, "lr": 8.80279964611125e-05} +{"train_loss": 0.01664474420249462, "grad_norm": 0.06031782552599907, "global_step": 11921, "epoch": 112, "lr": 8.802607211623864e-05} +{"train_loss": 0.019408922642469406, "grad_norm": 0.08137374371290207, "global_step": 11922, "epoch": 112, "lr": 8.802414763775767e-05} +{"train_loss": 0.014390742406249046, "grad_norm": 0.11553173512220383, "global_step": 11923, "epoch": 112, "lr": 8.80222230256764e-05} +{"train_loss": 0.01835871860384941, "grad_norm": 0.0703100636601448, "global_step": 11924, "epoch": 112, "lr": 8.802029828000156e-05} +{"train_loss": 0.017595680430531502, "grad_norm": 0.05951875075697899, "global_step": 11925, "epoch": 112, "lr": 8.801837340073993e-05} +{"train_loss": 0.018411358818411827, "grad_norm": 0.054895952343940735, "global_step": 11926, "epoch": 112, "lr": 8.801644838789827e-05} +{"train_loss": 0.018235042691230774, "grad_norm": 0.051724500954151154, "global_step": 11927, "epoch": 112, "lr": 8.801452324148333e-05} +{"train_loss": 0.01680736616253853, "grad_norm": 0.08029661327600479, "global_step": 11928, "epoch": 112, "lr": 8.80125979615019e-05} +{"train_loss": 0.019552676007151604, "grad_norm": 0.0535464771091938, "global_step": 11929, "epoch": 112, "lr": 8.801067254796074e-05} +{"train_loss": 0.015811491757631302, "grad_norm": 0.0526922382414341, "global_step": 11930, "epoch": 112, "lr": 8.800874700086659e-05} +{"train_loss": 0.018209638074040413, "grad_norm": 0.09148450195789337, "global_step": 11931, "epoch": 112, "lr": 8.800682132022624e-05} +{"train_loss": 0.01460055448114872, "grad_norm": 0.053389810025691986, "global_step": 11932, "epoch": 112, "lr": 8.800489550604645e-05} +{"train_loss": 0.015737365931272507, "grad_norm": 0.04797609895467758, "global_step": 11933, "epoch": 112, "lr": 8.8002969558334e-05} +{"train_loss": 0.01775810495018959, "grad_norm": 0.056755587458610535, "global_step": 11934, "epoch": 112, "lr": 8.800104347709562e-05} +{"train_loss": 0.017360981553792953, "grad_norm": 0.06296361237764359, "global_step": 11935, "epoch": 112, "lr": 8.799911726233811e-05} +{"train_loss": 0.019321952015161514, "grad_norm": 0.06258013844490051, "global_step": 11936, "epoch": 112, "lr": 8.799719091406822e-05} +{"train_loss": 0.017244333401322365, "grad_norm": 0.05728661268949509, "global_step": 11937, "epoch": 112, "lr": 8.799526443229274e-05} +{"train_loss": 0.016221296042203903, "grad_norm": 0.10006703436374664, "global_step": 11938, "epoch": 112, "lr": 8.79933378170184e-05} +{"train_loss": 0.018808769062161446, "grad_norm": 0.06546474248170853, "global_step": 11939, "epoch": 112, "lr": 8.799141106825202e-05} +{"train_loss": 0.015709586441516876, "grad_norm": 0.0907067134976387, "global_step": 11940, "epoch": 112, "lr": 8.798948418600032e-05} +{"train_loss": 0.018555356189608574, "grad_norm": 0.0660841092467308, "global_step": 11941, "epoch": 112, "lr": 8.798755717027012e-05} +{"train_loss": 0.020508768036961555, "grad_norm": 0.07214560359716415, "global_step": 11942, "epoch": 112, "lr": 8.798563002106814e-05} +{"train_loss": 0.015577059239149094, "grad_norm": 0.06422404944896698, "global_step": 11943, "epoch": 112, "lr": 8.798370273840119e-05} +{"train_loss": 0.015616784803569317, "grad_norm": 0.04669513180851936, "global_step": 11944, "epoch": 112, "lr": 8.798177532227602e-05} +{"train_loss": 0.0180610790848732, "grad_norm": 0.07930321246385574, "global_step": 11945, "epoch": 112, "lr": 8.79798477726994e-05} +{"train_loss": 0.015505999326705933, "grad_norm": 0.07784086465835571, "global_step": 11946, "epoch": 112, "lr": 8.797792008967813e-05} +{"train_loss": 0.015329009853303432, "grad_norm": 0.03744617849588394, "global_step": 11947, "epoch": 112, "lr": 8.797599227321897e-05} +{"train_loss": 0.020563311874866486, "grad_norm": 0.05630039796233177, "global_step": 11948, "epoch": 112, "lr": 8.797406432332867e-05} +{"train_loss": 0.01801767200231552, "grad_norm": 0.05141390487551689, "global_step": 11949, "epoch": 112, "lr": 8.797213624001404e-05} +{"train_loss": 0.014489928260445595, "grad_norm": 0.04278378188610077, "global_step": 11950, "epoch": 112, "lr": 8.797020802328181e-05} +{"train_loss": 0.016684815287590027, "grad_norm": 0.0772240161895752, "global_step": 11951, "epoch": 112, "lr": 8.796827967313881e-05} +{"train_loss": 0.016597624868154526, "grad_norm": 0.06473518162965775, "global_step": 11952, "epoch": 112, "lr": 8.796635118959178e-05} +{"train_loss": 0.016025131568312645, "grad_norm": 0.07241728901863098, "global_step": 11953, "epoch": 112, "lr": 8.79644225726475e-05} +{"train_loss": 0.015298891812562943, "grad_norm": 0.08941961079835892, "global_step": 11954, "epoch": 112, "lr": 8.796249382231274e-05} +{"train_loss": 0.017182696610689163, "grad_norm": 0.10336761176586151, "global_step": 11955, "epoch": 112, "lr": 8.79605649385943e-05} +{"train_loss": 0.01504123117774725, "grad_norm": 0.049533866345882416, "global_step": 11956, "epoch": 112, "lr": 8.795863592149893e-05} +{"train_loss": 0.018870580941438675, "grad_norm": 0.05712591111660004, "global_step": 11957, "epoch": 112, "lr": 8.795670677103343e-05} +{"train_loss": 0.020015742629766464, "grad_norm": 0.10813429206609726, "global_step": 11958, "epoch": 112, "lr": 8.795477748720456e-05} +{"train_loss": 0.021273337304592133, "grad_norm": 0.18574456870555878, "global_step": 11959, "epoch": 112, "lr": 8.795284807001912e-05} +{"train_loss": 0.016598660498857498, "grad_norm": 0.0621911883354187, "global_step": 11960, "epoch": 112, "lr": 8.795091851948387e-05} +{"train_loss": 0.017046894878149033, "grad_norm": 0.06844490766525269, "global_step": 11961, "epoch": 112, "lr": 8.794898883560561e-05} +{"train_loss": 0.0163070410490036, "grad_norm": 0.07360533624887466, "global_step": 11962, "epoch": 112, "lr": 8.794705901839109e-05} +{"train_loss": 0.01591923087835312, "grad_norm": 0.07463255524635315, "global_step": 11963, "epoch": 112, "lr": 8.794512906784712e-05} +{"train_loss": 0.015908394008874893, "grad_norm": 0.06073988974094391, "global_step": 11964, "epoch": 112, "lr": 8.794319898398046e-05} +{"train_loss": 0.015311554074287415, "grad_norm": 0.05698927864432335, "global_step": 11965, "epoch": 112, "lr": 8.79412687667979e-05} +{"train_loss": 0.01690356247127056, "grad_norm": 0.08595900982618332, "global_step": 11966, "epoch": 112, "lr": 8.793933841630622e-05} +{"train_loss": 0.01721063442528248, "grad_norm": 0.08925296366214752, "global_step": 11967, "epoch": 112, "lr": 8.793740793251221e-05} +{"train_loss": 0.016306063160300255, "grad_norm": 0.07283726334571838, "global_step": 11968, "epoch": 112, "lr": 8.793547731542264e-05} +{"train_loss": 0.01684214361011982, "grad_norm": 0.04977257922291756, "global_step": 11969, "epoch": 112, "lr": 8.793354656504432e-05} +{"train_loss": 0.016275063157081604, "grad_norm": 0.0706665962934494, "global_step": 11970, "epoch": 112, "lr": 8.793161568138399e-05} +{"train_loss": 0.018723925575613976, "grad_norm": 0.06685435771942139, "global_step": 11971, "epoch": 112, "lr": 8.792968466444847e-05} +{"train_loss": 0.016389306634664536, "grad_norm": 0.10565363615751266, "global_step": 11972, "epoch": 112, "lr": 8.792775351424453e-05} +{"train_loss": 0.017199348658323288, "grad_norm": 0.06380479782819748, "global_step": 11973, "epoch": 112, "lr": 8.792582223077897e-05} +{"train_loss": 0.01523183286190033, "grad_norm": 0.08578379452228546, "global_step": 11974, "epoch": 112, "lr": 8.792389081405856e-05} +{"train_loss": 0.017959201708436012, "grad_norm": 0.06639043241739273, "global_step": 11975, "epoch": 112, "lr": 8.792195926409008e-05} +{"train_loss": 0.018758755177259445, "grad_norm": 0.19400253891944885, "global_step": 11976, "epoch": 112, "lr": 8.792002758088034e-05} +{"train_loss": 0.017276730379616877, "grad_norm": 0.11080498993396759, "global_step": 11977, "epoch": 112, "lr": 8.79180957644361e-05, "val_loss": 0.019412456080317497} +{"train_loss": 0.018950240686535835, "grad_norm": 0.059051476418972015, "global_step": 11978, "epoch": 113, "lr": 8.791616381476417e-05} +{"train_loss": 0.015709729865193367, "grad_norm": 0.08991829305887222, "global_step": 11979, "epoch": 113, "lr": 8.791423173187132e-05} +{"train_loss": 0.018857017159461975, "grad_norm": 0.05357680842280388, "global_step": 11980, "epoch": 113, "lr": 8.791229951576436e-05} +{"train_loss": 0.018176471814513206, "grad_norm": 0.09512357413768768, "global_step": 11981, "epoch": 113, "lr": 8.791036716645007e-05} +{"train_loss": 0.01590687222778797, "grad_norm": 0.05124574527144432, "global_step": 11982, "epoch": 113, "lr": 8.790843468393521e-05} +{"train_loss": 0.0170535147190094, "grad_norm": 0.08498670905828476, "global_step": 11983, "epoch": 113, "lr": 8.790650206822663e-05} +{"train_loss": 0.015537083148956299, "grad_norm": 0.0583801232278347, "global_step": 11984, "epoch": 113, "lr": 8.790456931933106e-05} +{"train_loss": 0.015406529419124126, "grad_norm": 0.16319289803504944, "global_step": 11985, "epoch": 113, "lr": 8.790263643725534e-05} +{"train_loss": 0.018921345472335815, "grad_norm": 0.07976959645748138, "global_step": 11986, "epoch": 113, "lr": 8.790070342200623e-05} +{"train_loss": 0.019345201551914215, "grad_norm": 0.0441478006541729, "global_step": 11987, "epoch": 113, "lr": 8.789877027359052e-05} +{"train_loss": 0.015150289982557297, "grad_norm": 0.11006033420562744, "global_step": 11988, "epoch": 113, "lr": 8.789683699201503e-05} +{"train_loss": 0.017507892102003098, "grad_norm": 0.09927354753017426, "global_step": 11989, "epoch": 113, "lr": 8.789490357728653e-05} +{"train_loss": 0.016138043254613876, "grad_norm": 0.04556593671441078, "global_step": 11990, "epoch": 113, "lr": 8.789297002941181e-05} +{"train_loss": 0.014984345063567162, "grad_norm": 0.08246149122714996, "global_step": 11991, "epoch": 113, "lr": 8.789103634839768e-05} +{"train_loss": 0.015445226803421974, "grad_norm": 0.06420078873634338, "global_step": 11992, "epoch": 113, "lr": 8.788910253425093e-05} +{"train_loss": 0.01644377037882805, "grad_norm": 0.0522526316344738, "global_step": 11993, "epoch": 113, "lr": 8.788716858697836e-05} +{"train_loss": 0.016078684478998184, "grad_norm": 0.04915834218263626, "global_step": 11994, "epoch": 113, "lr": 8.788523450658673e-05} +{"train_loss": 0.016018375754356384, "grad_norm": 0.08831256628036499, "global_step": 11995, "epoch": 113, "lr": 8.788330029308287e-05} +{"train_loss": 0.019145173951983452, "grad_norm": 0.11733521521091461, "global_step": 11996, "epoch": 113, "lr": 8.788136594647358e-05} +{"train_loss": 0.01627751812338829, "grad_norm": 0.06343056261539459, "global_step": 11997, "epoch": 113, "lr": 8.787943146676565e-05} +{"train_loss": 0.018581077456474304, "grad_norm": 0.09021957963705063, "global_step": 11998, "epoch": 113, "lr": 8.787749685396585e-05} +{"train_loss": 0.01781126856803894, "grad_norm": 0.1089562475681305, "global_step": 11999, "epoch": 113, "lr": 8.787556210808101e-05} +{"train_loss": 0.017065338790416718, "grad_norm": 0.07773634046316147, "global_step": 12000, "epoch": 113, "lr": 8.787362722911792e-05} +{"train_loss": 0.019265349954366684, "grad_norm": 0.09838467091321945, "global_step": 12001, "epoch": 113, "lr": 8.787169221708336e-05} +{"train_loss": 0.020159848034381866, "grad_norm": 0.07672569155693054, "global_step": 12002, "epoch": 113, "lr": 8.786975707198415e-05} +{"train_loss": 0.018877176567912102, "grad_norm": 0.05060867965221405, "global_step": 12003, "epoch": 113, "lr": 8.786782179382708e-05} +{"train_loss": 0.016698841005563736, "grad_norm": 0.08996745944023132, "global_step": 12004, "epoch": 113, "lr": 8.786588638261897e-05} +{"train_loss": 0.01831606589257717, "grad_norm": 0.07265329360961914, "global_step": 12005, "epoch": 113, "lr": 8.786395083836659e-05} +{"train_loss": 0.017041660845279694, "grad_norm": 0.047861501574516296, "global_step": 12006, "epoch": 113, "lr": 8.786201516107675e-05} +{"train_loss": 0.018147757276892662, "grad_norm": 0.10018707811832428, "global_step": 12007, "epoch": 113, "lr": 8.786007935075626e-05} +{"train_loss": 0.017013147473335266, "grad_norm": 0.11258872598409653, "global_step": 12008, "epoch": 113, "lr": 8.78581434074119e-05} +{"train_loss": 0.018094873055815697, "grad_norm": 0.06844856590032578, "global_step": 12009, "epoch": 113, "lr": 8.785620733105051e-05} +{"train_loss": 0.017827844247221947, "grad_norm": 0.07362790405750275, "global_step": 12010, "epoch": 113, "lr": 8.785427112167888e-05} +{"train_loss": 0.017523381859064102, "grad_norm": 0.10123573243618011, "global_step": 12011, "epoch": 113, "lr": 8.785233477930376e-05} +{"train_loss": 0.017799757421016693, "grad_norm": 0.04338214918971062, "global_step": 12012, "epoch": 113, "lr": 8.785039830393204e-05} +{"train_loss": 0.01734154485166073, "grad_norm": 0.06644134968519211, "global_step": 12013, "epoch": 113, "lr": 8.784846169557046e-05} +{"train_loss": 0.015432186424732208, "grad_norm": 0.07862262427806854, "global_step": 12014, "epoch": 113, "lr": 8.784652495422585e-05} +{"train_loss": 0.017749346792697906, "grad_norm": 0.10650631785392761, "global_step": 12015, "epoch": 113, "lr": 8.784458807990501e-05} +{"train_loss": 0.016833653673529625, "grad_norm": 0.06740925461053848, "global_step": 12016, "epoch": 113, "lr": 8.784265107261476e-05} +{"train_loss": 0.0180665235966444, "grad_norm": 0.07095669209957123, "global_step": 12017, "epoch": 113, "lr": 8.784071393236186e-05} +{"train_loss": 0.01741647720336914, "grad_norm": 0.054927367717027664, "global_step": 12018, "epoch": 113, "lr": 8.783877665915316e-05} +{"train_loss": 0.017056960612535477, "grad_norm": 0.05512015148997307, "global_step": 12019, "epoch": 113, "lr": 8.783683925299547e-05} +{"train_loss": 0.016580333933234215, "grad_norm": 0.0619259737432003, "global_step": 12020, "epoch": 113, "lr": 8.783490171389557e-05} +{"train_loss": 0.016538919880986214, "grad_norm": 0.06384793668985367, "global_step": 12021, "epoch": 113, "lr": 8.783296404186029e-05} +{"train_loss": 0.018358688801527023, "grad_norm": 0.08449926227331161, "global_step": 12022, "epoch": 113, "lr": 8.783102623689642e-05} +{"train_loss": 0.019520580768585205, "grad_norm": 0.08950749039649963, "global_step": 12023, "epoch": 113, "lr": 8.782908829901078e-05} +{"train_loss": 0.016789663583040237, "grad_norm": 0.058885421603918076, "global_step": 12024, "epoch": 113, "lr": 8.782715022821016e-05} +{"train_loss": 0.015773870050907135, "grad_norm": 0.09434422850608826, "global_step": 12025, "epoch": 113, "lr": 8.78252120245014e-05} +{"train_loss": 0.019146690145134926, "grad_norm": 0.1575813591480255, "global_step": 12026, "epoch": 113, "lr": 8.782327368789128e-05} +{"train_loss": 0.016244105994701385, "grad_norm": 0.08235710859298706, "global_step": 12027, "epoch": 113, "lr": 8.782133521838663e-05} +{"train_loss": 0.01689520850777626, "grad_norm": 0.09245163947343826, "global_step": 12028, "epoch": 113, "lr": 8.781939661599427e-05} +{"train_loss": 0.017771193757653236, "grad_norm": 0.06505591422319412, "global_step": 12029, "epoch": 113, "lr": 8.781745788072099e-05} +{"train_loss": 0.014737298712134361, "grad_norm": 0.06005201116204262, "global_step": 12030, "epoch": 113, "lr": 8.78155190125736e-05} +{"train_loss": 0.017381997779011726, "grad_norm": 0.05228572338819504, "global_step": 12031, "epoch": 113, "lr": 8.781358001155893e-05} +{"train_loss": 0.0178125761449337, "grad_norm": 0.06451299041509628, "global_step": 12032, "epoch": 113, "lr": 8.781164087768378e-05} +{"train_loss": 0.01707516610622406, "grad_norm": 0.06475464254617691, "global_step": 12033, "epoch": 113, "lr": 8.780970161095497e-05} +{"train_loss": 0.017040956765413284, "grad_norm": 0.045851364731788635, "global_step": 12034, "epoch": 113, "lr": 8.780776221137932e-05} +{"train_loss": 0.01631801575422287, "grad_norm": 0.049319326877593994, "global_step": 12035, "epoch": 113, "lr": 8.780582267896363e-05} +{"train_loss": 0.01604331098496914, "grad_norm": 0.051756225526332855, "global_step": 12036, "epoch": 113, "lr": 8.780388301371472e-05} +{"train_loss": 0.014187444001436234, "grad_norm": 0.05629834905266762, "global_step": 12037, "epoch": 113, "lr": 8.780194321563939e-05} +{"train_loss": 0.01670887880027294, "grad_norm": 0.12097499519586563, "global_step": 12038, "epoch": 113, "lr": 8.780000328474449e-05} +{"train_loss": 0.01604985073208809, "grad_norm": 0.14874699711799622, "global_step": 12039, "epoch": 113, "lr": 8.779806322103682e-05} +{"train_loss": 0.016993219032883644, "grad_norm": 0.0756075531244278, "global_step": 12040, "epoch": 113, "lr": 8.779612302452318e-05} +{"train_loss": 0.017557570710778236, "grad_norm": 0.05962483212351799, "global_step": 12041, "epoch": 113, "lr": 8.779418269521042e-05} +{"train_loss": 0.018456585705280304, "grad_norm": 0.11497477442026138, "global_step": 12042, "epoch": 113, "lr": 8.779224223310532e-05} +{"train_loss": 0.015357112511992455, "grad_norm": 0.07018442451953888, "global_step": 12043, "epoch": 113, "lr": 8.779030163821472e-05} +{"train_loss": 0.017010033130645752, "grad_norm": 0.08317041397094727, "global_step": 12044, "epoch": 113, "lr": 8.778836091054545e-05} +{"train_loss": 0.016549337655305862, "grad_norm": 0.08209927380084991, "global_step": 12045, "epoch": 113, "lr": 8.77864200501043e-05} +{"train_loss": 0.015096425078809261, "grad_norm": 0.05350485444068909, "global_step": 12046, "epoch": 113, "lr": 8.778447905689809e-05} +{"train_loss": 0.016219064593315125, "grad_norm": 0.055325914174318314, "global_step": 12047, "epoch": 113, "lr": 8.778253793093368e-05} +{"train_loss": 0.01677054539322853, "grad_norm": 0.057302169501781464, "global_step": 12048, "epoch": 113, "lr": 8.778059667221784e-05} +{"train_loss": 0.017069928348064423, "grad_norm": 0.06797114759683609, "global_step": 12049, "epoch": 113, "lr": 8.777865528075742e-05} +{"train_loss": 0.017602767795324326, "grad_norm": 0.055231377482414246, "global_step": 12050, "epoch": 113, "lr": 8.777671375655924e-05} +{"train_loss": 0.01779724285006523, "grad_norm": 0.05339726060628891, "global_step": 12051, "epoch": 113, "lr": 8.777477209963011e-05} +{"train_loss": 0.017473649233579636, "grad_norm": 0.11395309120416641, "global_step": 12052, "epoch": 113, "lr": 8.777283030997686e-05} +{"train_loss": 0.019259124994277954, "grad_norm": 0.1311016082763672, "global_step": 12053, "epoch": 113, "lr": 8.777088838760633e-05} +{"train_loss": 0.018432103097438812, "grad_norm": 0.05660249665379524, "global_step": 12054, "epoch": 113, "lr": 8.776894633252529e-05} +{"train_loss": 0.016001295298337936, "grad_norm": 0.05198344215750694, "global_step": 12055, "epoch": 113, "lr": 8.776700414474061e-05} +{"train_loss": 0.01663617044687271, "grad_norm": 0.04566009342670441, "global_step": 12056, "epoch": 113, "lr": 8.776506182425912e-05} +{"train_loss": 0.018075376749038696, "grad_norm": 0.07100290060043335, "global_step": 12057, "epoch": 113, "lr": 8.77631193710876e-05} +{"train_loss": 0.015949072316288948, "grad_norm": 0.10681749880313873, "global_step": 12058, "epoch": 113, "lr": 8.776117678523293e-05} +{"train_loss": 0.017654409632086754, "grad_norm": 0.11213912069797516, "global_step": 12059, "epoch": 113, "lr": 8.775923406670188e-05} +{"train_loss": 0.018212927505373955, "grad_norm": 0.11224137991666794, "global_step": 12060, "epoch": 113, "lr": 8.775729121550132e-05} +{"train_loss": 0.018322715535759926, "grad_norm": 0.05203421041369438, "global_step": 12061, "epoch": 113, "lr": 8.775534823163804e-05} +{"train_loss": 0.02069864794611931, "grad_norm": 0.0630699023604393, "global_step": 12062, "epoch": 113, "lr": 8.775340511511891e-05} +{"train_loss": 0.01906123012304306, "grad_norm": 0.05974680930376053, "global_step": 12063, "epoch": 113, "lr": 8.775146186595073e-05} +{"train_loss": 0.01721355691552162, "grad_norm": 0.04611966386437416, "global_step": 12064, "epoch": 113, "lr": 8.774951848414031e-05} +{"train_loss": 0.017508096992969513, "grad_norm": 0.057546116411685944, "global_step": 12065, "epoch": 113, "lr": 8.774757496969452e-05} +{"train_loss": 0.018396727740764618, "grad_norm": 0.06512945890426636, "global_step": 12066, "epoch": 113, "lr": 8.774563132262015e-05} +{"train_loss": 0.018274981528520584, "grad_norm": 0.08768387883901596, "global_step": 12067, "epoch": 113, "lr": 8.774368754292406e-05} +{"train_loss": 0.01673825830221176, "grad_norm": 0.06726663559675217, "global_step": 12068, "epoch": 113, "lr": 8.774174363061307e-05} +{"train_loss": 0.01648453250527382, "grad_norm": 0.07407673448324203, "global_step": 12069, "epoch": 113, "lr": 8.7739799585694e-05} +{"train_loss": 0.016730811446905136, "grad_norm": 0.10883145034313202, "global_step": 12070, "epoch": 113, "lr": 8.773785540817367e-05} +{"train_loss": 0.019249657168984413, "grad_norm": 0.06548947095870972, "global_step": 12071, "epoch": 113, "lr": 8.773591109805895e-05} +{"train_loss": 0.017909571528434753, "grad_norm": 0.1370767056941986, "global_step": 12072, "epoch": 113, "lr": 8.773396665535665e-05} +{"train_loss": 0.017081666737794876, "grad_norm": 0.0722782164812088, "global_step": 12073, "epoch": 113, "lr": 8.773202208007359e-05} +{"train_loss": 0.017257334664463997, "grad_norm": 0.0628567710518837, "global_step": 12074, "epoch": 113, "lr": 8.773007737221661e-05} +{"train_loss": 0.015786221250891685, "grad_norm": 0.1006065160036087, "global_step": 12075, "epoch": 113, "lr": 8.772813253179254e-05} +{"train_loss": 0.016144735738635063, "grad_norm": 0.08585818111896515, "global_step": 12076, "epoch": 113, "lr": 8.772618755880825e-05} +{"train_loss": 0.016446180641651154, "grad_norm": 0.06080378219485283, "global_step": 12077, "epoch": 113, "lr": 8.772424245327053e-05} +{"train_loss": 0.01795642077922821, "grad_norm": 0.12798051536083221, "global_step": 12078, "epoch": 113, "lr": 8.772229721518622e-05} +{"train_loss": 0.019179973751306534, "grad_norm": 0.08084939420223236, "global_step": 12079, "epoch": 113, "lr": 8.772035184456217e-05} +{"train_loss": 0.017955146729946136, "grad_norm": 0.051527395844459534, "global_step": 12080, "epoch": 113, "lr": 8.77184063414052e-05} +{"train_loss": 0.014855284243822098, "grad_norm": 0.08531644940376282, "global_step": 12081, "epoch": 113, "lr": 8.771646070572217e-05} +{"train_loss": 0.018323395401239395, "grad_norm": 0.08214714378118515, "global_step": 12082, "epoch": 113, "lr": 8.771451493751988e-05} +{"train_loss": 0.01725058607265072, "grad_norm": 0.11582683026790619, "global_step": 12083, "epoch": 113, "lr": 8.771256903680519e-05, "val_loss": 0.019307462498545647} +{"train_loss": 0.015169658698141575, "grad_norm": 0.061214886605739594, "global_step": 12084, "epoch": 114, "lr": 8.771062300358494e-05} +{"train_loss": 0.016841471195220947, "grad_norm": 0.13436554372310638, "global_step": 12085, "epoch": 114, "lr": 8.770867683786597e-05} +{"train_loss": 0.01721157878637314, "grad_norm": 0.05054756999015808, "global_step": 12086, "epoch": 114, "lr": 8.770673053965509e-05} +{"train_loss": 0.017189733684062958, "grad_norm": 0.05099423974752426, "global_step": 12087, "epoch": 114, "lr": 8.770478410895916e-05} +{"train_loss": 0.014746209606528282, "grad_norm": 0.05384545773267746, "global_step": 12088, "epoch": 114, "lr": 8.770283754578503e-05} +{"train_loss": 0.019433122128248215, "grad_norm": 0.14981403946876526, "global_step": 12089, "epoch": 114, "lr": 8.770089085013951e-05} +{"train_loss": 0.017064781859517097, "grad_norm": 0.08267490565776825, "global_step": 12090, "epoch": 114, "lr": 8.769894402202947e-05} +{"train_loss": 0.017808647826313972, "grad_norm": 0.06049752235412598, "global_step": 12091, "epoch": 114, "lr": 8.769699706146172e-05} +{"train_loss": 0.015111985616385937, "grad_norm": 0.12717346847057343, "global_step": 12092, "epoch": 114, "lr": 8.769504996844311e-05} +{"train_loss": 0.0175132155418396, "grad_norm": 0.07249034941196442, "global_step": 12093, "epoch": 114, "lr": 8.769310274298051e-05} +{"train_loss": 0.01807596907019615, "grad_norm": 0.06189749389886856, "global_step": 12094, "epoch": 114, "lr": 8.769115538508072e-05} +{"train_loss": 0.018138490617275238, "grad_norm": 0.11888677626848221, "global_step": 12095, "epoch": 114, "lr": 8.768920789475062e-05} +{"train_loss": 0.018660008907318115, "grad_norm": 0.10284112393856049, "global_step": 12096, "epoch": 114, "lr": 8.768726027199703e-05} +{"train_loss": 0.017342351377010345, "grad_norm": 0.05247744545340538, "global_step": 12097, "epoch": 114, "lr": 8.768531251682679e-05} +{"train_loss": 0.0168689526617527, "grad_norm": 0.0777345821261406, "global_step": 12098, "epoch": 114, "lr": 8.768336462924674e-05} +{"train_loss": 0.016396211460232735, "grad_norm": 0.054776471108198166, "global_step": 12099, "epoch": 114, "lr": 8.768141660926375e-05} +{"train_loss": 0.016879651695489883, "grad_norm": 0.0817926824092865, "global_step": 12100, "epoch": 114, "lr": 8.767946845688465e-05} +{"train_loss": 0.018525445833802223, "grad_norm": 0.10795670002698898, "global_step": 12101, "epoch": 114, "lr": 8.767752017211629e-05} +{"train_loss": 0.019418274983763695, "grad_norm": 0.08209773898124695, "global_step": 12102, "epoch": 114, "lr": 8.767557175496549e-05} +{"train_loss": 0.015363337472081184, "grad_norm": 0.09395149350166321, "global_step": 12103, "epoch": 114, "lr": 8.767362320543913e-05} +{"train_loss": 0.01817639172077179, "grad_norm": 0.06521712243556976, "global_step": 12104, "epoch": 114, "lr": 8.767167452354403e-05} +{"train_loss": 0.017025627195835114, "grad_norm": 0.07465925067663193, "global_step": 12105, "epoch": 114, "lr": 8.766972570928706e-05} +{"train_loss": 0.017290137708187103, "grad_norm": 0.10219533741474152, "global_step": 12106, "epoch": 114, "lr": 8.766777676267505e-05} +{"train_loss": 0.017922215163707733, "grad_norm": 0.06562718003988266, "global_step": 12107, "epoch": 114, "lr": 8.766582768371487e-05} +{"train_loss": 0.017896758392453194, "grad_norm": 0.05695493519306183, "global_step": 12108, "epoch": 114, "lr": 8.766387847241333e-05} +{"train_loss": 0.016631869599223137, "grad_norm": 0.05988423526287079, "global_step": 12109, "epoch": 114, "lr": 8.76619291287773e-05} +{"train_loss": 0.015773441642522812, "grad_norm": 0.08087433874607086, "global_step": 12110, "epoch": 114, "lr": 8.765997965281365e-05} +{"train_loss": 0.019696246832609177, "grad_norm": 0.05817710980772972, "global_step": 12111, "epoch": 114, "lr": 8.76580300445292e-05} +{"train_loss": 0.017287999391555786, "grad_norm": 0.1148814707994461, "global_step": 12112, "epoch": 114, "lr": 8.76560803039308e-05} +{"train_loss": 0.01797110214829445, "grad_norm": 0.08173248171806335, "global_step": 12113, "epoch": 114, "lr": 8.765413043102533e-05} +{"train_loss": 0.01794556900858879, "grad_norm": 0.06375137716531754, "global_step": 12114, "epoch": 114, "lr": 8.76521804258196e-05} +{"train_loss": 0.01753532513976097, "grad_norm": 0.16200928390026093, "global_step": 12115, "epoch": 114, "lr": 8.765023028832049e-05} +{"train_loss": 0.018045742064714432, "grad_norm": 0.061230506747961044, "global_step": 12116, "epoch": 114, "lr": 8.764828001853485e-05} +{"train_loss": 0.017142387107014656, "grad_norm": 0.0802469477057457, "global_step": 12117, "epoch": 114, "lr": 8.764632961646952e-05} +{"train_loss": 0.014739593490958214, "grad_norm": 0.05951358377933502, "global_step": 12118, "epoch": 114, "lr": 8.764437908213137e-05} +{"train_loss": 0.018713422119617462, "grad_norm": 0.07359033823013306, "global_step": 12119, "epoch": 114, "lr": 8.764242841552723e-05} +{"train_loss": 0.014113996177911758, "grad_norm": 0.06961262971162796, "global_step": 12120, "epoch": 114, "lr": 8.764047761666396e-05} +{"train_loss": 0.014046719297766685, "grad_norm": 0.07650476694107056, "global_step": 12121, "epoch": 114, "lr": 8.763852668554844e-05} +{"train_loss": 0.01779807358980179, "grad_norm": 0.06067129597067833, "global_step": 12122, "epoch": 114, "lr": 8.76365756221875e-05} +{"train_loss": 0.017612330615520477, "grad_norm": 0.10614476352930069, "global_step": 12123, "epoch": 114, "lr": 8.763462442658799e-05} +{"train_loss": 0.017757121473550797, "grad_norm": 0.0535200759768486, "global_step": 12124, "epoch": 114, "lr": 8.76326730987568e-05} +{"train_loss": 0.016144314780831337, "grad_norm": 0.06234065815806389, "global_step": 12125, "epoch": 114, "lr": 8.763072163870074e-05} +{"train_loss": 0.019073182716965675, "grad_norm": 0.2263646125793457, "global_step": 12126, "epoch": 114, "lr": 8.76287700464267e-05} +{"train_loss": 0.018494393676519394, "grad_norm": 0.1347709745168686, "global_step": 12127, "epoch": 114, "lr": 8.762681832194153e-05} +{"train_loss": 0.01636120118200779, "grad_norm": 0.15018230676651, "global_step": 12128, "epoch": 114, "lr": 8.762486646525207e-05} +{"train_loss": 0.01757744699716568, "grad_norm": 0.07120433449745178, "global_step": 12129, "epoch": 114, "lr": 8.76229144763652e-05} +{"train_loss": 0.016597479581832886, "grad_norm": 0.053362973034381866, "global_step": 12130, "epoch": 114, "lr": 8.762096235528778e-05} +{"train_loss": 0.017236467450857162, "grad_norm": 0.06830807775259018, "global_step": 12131, "epoch": 114, "lr": 8.761901010202664e-05} +{"train_loss": 0.0163872167468071, "grad_norm": 0.07680866122245789, "global_step": 12132, "epoch": 114, "lr": 8.761705771658868e-05} +{"train_loss": 0.018939487636089325, "grad_norm": 0.07520189881324768, "global_step": 12133, "epoch": 114, "lr": 8.761510519898072e-05} +{"train_loss": 0.01694430038332939, "grad_norm": 0.06863320618867874, "global_step": 12134, "epoch": 114, "lr": 8.761315254920964e-05} +{"train_loss": 0.01705125719308853, "grad_norm": 0.1765279620885849, "global_step": 12135, "epoch": 114, "lr": 8.761119976728232e-05} +{"train_loss": 0.01775621436536312, "grad_norm": 0.1279011368751526, "global_step": 12136, "epoch": 114, "lr": 8.760924685320557e-05} +{"train_loss": 0.017977558076381683, "grad_norm": 0.17151367664337158, "global_step": 12137, "epoch": 114, "lr": 8.760729380698631e-05} +{"train_loss": 0.01597335934638977, "grad_norm": 0.1674177348613739, "global_step": 12138, "epoch": 114, "lr": 8.760534062863137e-05} +{"train_loss": 0.020710906013846397, "grad_norm": 0.16301217675209045, "global_step": 12139, "epoch": 114, "lr": 8.760338731814759e-05} +{"train_loss": 0.015440347604453564, "grad_norm": 0.12045972794294357, "global_step": 12140, "epoch": 114, "lr": 8.760143387554189e-05} +{"train_loss": 0.01552267000079155, "grad_norm": 0.07560479640960693, "global_step": 12141, "epoch": 114, "lr": 8.759948030082109e-05} +{"train_loss": 0.017708653584122658, "grad_norm": 0.06778382509946823, "global_step": 12142, "epoch": 114, "lr": 8.759752659399208e-05} +{"train_loss": 0.015202511101961136, "grad_norm": 0.09102266281843185, "global_step": 12143, "epoch": 114, "lr": 8.75955727550617e-05} +{"train_loss": 0.015770073980093002, "grad_norm": 0.0633433610200882, "global_step": 12144, "epoch": 114, "lr": 8.759361878403683e-05} +{"train_loss": 0.01965533196926117, "grad_norm": 0.08489721268415451, "global_step": 12145, "epoch": 114, "lr": 8.759166468092432e-05} +{"train_loss": 0.016339998692274094, "grad_norm": 0.047249823808670044, "global_step": 12146, "epoch": 114, "lr": 8.758971044573107e-05} +{"train_loss": 0.016971256583929062, "grad_norm": 0.07630976289510727, "global_step": 12147, "epoch": 114, "lr": 8.758775607846393e-05} +{"train_loss": 0.016876207664608955, "grad_norm": 0.13793958723545074, "global_step": 12148, "epoch": 114, "lr": 8.758580157912974e-05} +{"train_loss": 0.019886691123247147, "grad_norm": 0.09044501930475235, "global_step": 12149, "epoch": 114, "lr": 8.75838469477354e-05} +{"train_loss": 0.016964755952358246, "grad_norm": 0.06838298588991165, "global_step": 12150, "epoch": 114, "lr": 8.758189218428777e-05} +{"train_loss": 0.019194236025214195, "grad_norm": 0.09640702605247498, "global_step": 12151, "epoch": 114, "lr": 8.757993728879371e-05} +{"train_loss": 0.018136542290449142, "grad_norm": 0.10220450907945633, "global_step": 12152, "epoch": 114, "lr": 8.757798226126009e-05} +{"train_loss": 0.016086790710687637, "grad_norm": 0.0483342781662941, "global_step": 12153, "epoch": 114, "lr": 8.757602710169378e-05} +{"train_loss": 0.01861349679529667, "grad_norm": 0.07575734704732895, "global_step": 12154, "epoch": 114, "lr": 8.757407181010166e-05} +{"train_loss": 0.015827149152755737, "grad_norm": 0.10632969439029694, "global_step": 12155, "epoch": 114, "lr": 8.757211638649058e-05} +{"train_loss": 0.01637905277311802, "grad_norm": 0.08104940503835678, "global_step": 12156, "epoch": 114, "lr": 8.757016083086744e-05} +{"train_loss": 0.017056340351700783, "grad_norm": 0.08573778718709946, "global_step": 12157, "epoch": 114, "lr": 8.756820514323909e-05} +{"train_loss": 0.015413276851177216, "grad_norm": 0.08450780808925629, "global_step": 12158, "epoch": 114, "lr": 8.75662493236124e-05} +{"train_loss": 0.01839529164135456, "grad_norm": 0.09866150468587875, "global_step": 12159, "epoch": 114, "lr": 8.756429337199425e-05} +{"train_loss": 0.01869552582502365, "grad_norm": 0.08155298978090286, "global_step": 12160, "epoch": 114, "lr": 8.75623372883915e-05} +{"train_loss": 0.017053134739398956, "grad_norm": 0.05441557615995407, "global_step": 12161, "epoch": 114, "lr": 8.756038107281104e-05} +{"train_loss": 0.015160130336880684, "grad_norm": 0.09835202991962433, "global_step": 12162, "epoch": 114, "lr": 8.755842472525974e-05} +{"train_loss": 0.016881156712770462, "grad_norm": 0.06084974482655525, "global_step": 12163, "epoch": 114, "lr": 8.755646824574446e-05} +{"train_loss": 0.018310533836483955, "grad_norm": 0.06628485769033432, "global_step": 12164, "epoch": 114, "lr": 8.755451163427208e-05} +{"train_loss": 0.015538612380623817, "grad_norm": 0.07592108845710754, "global_step": 12165, "epoch": 114, "lr": 8.755255489084949e-05} +{"train_loss": 0.018351271748542786, "grad_norm": 0.05796711891889572, "global_step": 12166, "epoch": 114, "lr": 8.755059801548356e-05} +{"train_loss": 0.016345331445336342, "grad_norm": 0.07386542111635208, "global_step": 12167, "epoch": 114, "lr": 8.754864100818114e-05} +{"train_loss": 0.01687789522111416, "grad_norm": 0.09275947511196136, "global_step": 12168, "epoch": 114, "lr": 8.754668386894913e-05} +{"train_loss": 0.017789077013731003, "grad_norm": 0.060575395822525024, "global_step": 12169, "epoch": 114, "lr": 8.75447265977944e-05} +{"train_loss": 0.016080576926469803, "grad_norm": 0.06682214885950089, "global_step": 12170, "epoch": 114, "lr": 8.754276919472384e-05} +{"train_loss": 0.01688903570175171, "grad_norm": 0.11235836893320084, "global_step": 12171, "epoch": 114, "lr": 8.75408116597443e-05} +{"train_loss": 0.01599556766450405, "grad_norm": 0.08178006112575531, "global_step": 12172, "epoch": 114, "lr": 8.753885399286269e-05} +{"train_loss": 0.016485275700688362, "grad_norm": 0.07391750067472458, "global_step": 12173, "epoch": 114, "lr": 8.753689619408587e-05} +{"train_loss": 0.019358107820153236, "grad_norm": 0.054525308310985565, "global_step": 12174, "epoch": 114, "lr": 8.753493826342071e-05} +{"train_loss": 0.01713520660996437, "grad_norm": 0.057475097477436066, "global_step": 12175, "epoch": 114, "lr": 8.753298020087411e-05} +{"train_loss": 0.017573539167642593, "grad_norm": 0.07335884124040604, "global_step": 12176, "epoch": 114, "lr": 8.753102200645293e-05} +{"train_loss": 0.0151614248752594, "grad_norm": 0.06498383730649948, "global_step": 12177, "epoch": 114, "lr": 8.752906368016407e-05} +{"train_loss": 0.01769500970840454, "grad_norm": 0.06500750780105591, "global_step": 12178, "epoch": 114, "lr": 8.75271052220144e-05} +{"train_loss": 0.017790092155337334, "grad_norm": 0.09782863408327103, "global_step": 12179, "epoch": 114, "lr": 8.752514663201081e-05} +{"train_loss": 0.019022133201360703, "grad_norm": 0.07907260209321976, "global_step": 12180, "epoch": 114, "lr": 8.752318791016017e-05} +{"train_loss": 0.016360167413949966, "grad_norm": 0.0501910038292408, "global_step": 12181, "epoch": 114, "lr": 8.752122905646935e-05} +{"train_loss": 0.01732178032398224, "grad_norm": 0.04208371415734291, "global_step": 12182, "epoch": 114, "lr": 8.751927007094528e-05} +{"train_loss": 0.022504650056362152, "grad_norm": 0.07904141396284103, "global_step": 12183, "epoch": 114, "lr": 8.751731095359476e-05} +{"train_loss": 0.016579803079366684, "grad_norm": 0.06531862169504166, "global_step": 12184, "epoch": 114, "lr": 8.751535170442477e-05} +{"train_loss": 0.015867654234170914, "grad_norm": 0.05302049592137337, "global_step": 12185, "epoch": 114, "lr": 8.751339232344214e-05} +{"train_loss": 0.01694224774837494, "grad_norm": 0.06985052675008774, "global_step": 12186, "epoch": 114, "lr": 8.751143281065375e-05} +{"train_loss": 0.02057807892560959, "grad_norm": 0.0840277373790741, "global_step": 12187, "epoch": 114, "lr": 8.75094731660665e-05} +{"train_loss": 0.01495795976370573, "grad_norm": 0.052065685391426086, "global_step": 12188, "epoch": 114, "lr": 8.75075133896873e-05} +{"train_loss": 0.017255512540632824, "grad_norm": 0.07047568261623383, "global_step": 12189, "epoch": 114, "lr": 8.750555348152298e-05, "val_loss": 0.019088206812739372} +{"train_loss": 0.015442851930856705, "grad_norm": 0.04851361736655235, "global_step": 12190, "epoch": 115, "lr": 8.750359344158047e-05} +{"train_loss": 0.01857592910528183, "grad_norm": 0.06017519161105156, "global_step": 12191, "epoch": 115, "lr": 8.750163326986664e-05} +{"train_loss": 0.017502818256616592, "grad_norm": 0.06889347732067108, "global_step": 12192, "epoch": 115, "lr": 8.749967296638837e-05} +{"train_loss": 0.014243749901652336, "grad_norm": 0.046708956360816956, "global_step": 12193, "epoch": 115, "lr": 8.749771253115258e-05} +{"train_loss": 0.015703551471233368, "grad_norm": 0.06389227509498596, "global_step": 12194, "epoch": 115, "lr": 8.749575196416612e-05} +{"train_loss": 0.01814568042755127, "grad_norm": 0.069523386657238, "global_step": 12195, "epoch": 115, "lr": 8.749379126543589e-05} +{"train_loss": 0.016307003796100616, "grad_norm": 0.083305224776268, "global_step": 12196, "epoch": 115, "lr": 8.749183043496879e-05} +{"train_loss": 0.016822196543216705, "grad_norm": 0.06781890988349915, "global_step": 12197, "epoch": 115, "lr": 8.74898694727717e-05} +{"train_loss": 0.019860338419675827, "grad_norm": 0.11144616454839706, "global_step": 12198, "epoch": 115, "lr": 8.74879083788515e-05} +{"train_loss": 0.017358455806970596, "grad_norm": 0.08126119524240494, "global_step": 12199, "epoch": 115, "lr": 8.748594715321512e-05} +{"train_loss": 0.014660600572824478, "grad_norm": 0.05769896134734154, "global_step": 12200, "epoch": 115, "lr": 8.748398579586942e-05} +{"train_loss": 0.019318826496601105, "grad_norm": 0.08250122517347336, "global_step": 12201, "epoch": 115, "lr": 8.748202430682129e-05} +{"train_loss": 0.018709320574998856, "grad_norm": 0.08306286484003067, "global_step": 12202, "epoch": 115, "lr": 8.748006268607761e-05} +{"train_loss": 0.01294388435781002, "grad_norm": 0.08544772863388062, "global_step": 12203, "epoch": 115, "lr": 8.74781009336453e-05} +{"train_loss": 0.017950821667909622, "grad_norm": 0.10117824375629425, "global_step": 12204, "epoch": 115, "lr": 8.747613904953125e-05} +{"train_loss": 0.01606045663356781, "grad_norm": 0.05053192749619484, "global_step": 12205, "epoch": 115, "lr": 8.747417703374234e-05} +{"train_loss": 0.018288055434823036, "grad_norm": 0.04854762926697731, "global_step": 12206, "epoch": 115, "lr": 8.747221488628547e-05} +{"train_loss": 0.016306990757584572, "grad_norm": 0.0675281211733818, "global_step": 12207, "epoch": 115, "lr": 8.747025260716754e-05} +{"train_loss": 0.018910378217697144, "grad_norm": 0.06053931266069412, "global_step": 12208, "epoch": 115, "lr": 8.746829019639542e-05} +{"train_loss": 0.01684929057955742, "grad_norm": 0.08409678190946579, "global_step": 12209, "epoch": 115, "lr": 8.746632765397603e-05} +{"train_loss": 0.019351061433553696, "grad_norm": 0.06893712282180786, "global_step": 12210, "epoch": 115, "lr": 8.746436497991627e-05} +{"train_loss": 0.014462302438914776, "grad_norm": 0.07501082122325897, "global_step": 12211, "epoch": 115, "lr": 8.746240217422301e-05} +{"train_loss": 0.01696709170937538, "grad_norm": 0.05268317461013794, "global_step": 12212, "epoch": 115, "lr": 8.746043923690316e-05} +{"train_loss": 0.01605764403939247, "grad_norm": 0.08117368817329407, "global_step": 12213, "epoch": 115, "lr": 8.745847616796364e-05} +{"train_loss": 0.0186869315803051, "grad_norm": 0.08855823427438736, "global_step": 12214, "epoch": 115, "lr": 8.74565129674113e-05} +{"train_loss": 0.016919085755944252, "grad_norm": 0.11439712345600128, "global_step": 12215, "epoch": 115, "lr": 8.745454963525308e-05} +{"train_loss": 0.015096291899681091, "grad_norm": 0.055455531924963, "global_step": 12216, "epoch": 115, "lr": 8.745258617149584e-05} +{"train_loss": 0.017977554351091385, "grad_norm": 0.0501142181456089, "global_step": 12217, "epoch": 115, "lr": 8.745062257614652e-05} +{"train_loss": 0.015860915184020996, "grad_norm": 0.06936119496822357, "global_step": 12218, "epoch": 115, "lr": 8.744865884921198e-05} +{"train_loss": 0.017608966678380966, "grad_norm": 0.11895313113927841, "global_step": 12219, "epoch": 115, "lr": 8.744669499069916e-05} +{"train_loss": 0.017688188701868057, "grad_norm": 0.09576895087957382, "global_step": 12220, "epoch": 115, "lr": 8.744473100061492e-05} +{"train_loss": 0.018387075513601303, "grad_norm": 0.048469677567481995, "global_step": 12221, "epoch": 115, "lr": 8.74427668789662e-05} +{"train_loss": 0.017150534316897392, "grad_norm": 0.05668194219470024, "global_step": 12222, "epoch": 115, "lr": 8.744080262575987e-05} +{"train_loss": 0.01840728335082531, "grad_norm": 0.06026790291070938, "global_step": 12223, "epoch": 115, "lr": 8.743883824100284e-05} +{"train_loss": 0.016433004289865494, "grad_norm": 0.06778419762849808, "global_step": 12224, "epoch": 115, "lr": 8.743687372470201e-05} +{"train_loss": 0.015020745806396008, "grad_norm": 0.05442505329847336, "global_step": 12225, "epoch": 115, "lr": 8.743490907686429e-05} +{"train_loss": 0.015408170409500599, "grad_norm": 0.0922362208366394, "global_step": 12226, "epoch": 115, "lr": 8.74329442974966e-05} +{"train_loss": 0.013243936002254486, "grad_norm": 0.059234220534563065, "global_step": 12227, "epoch": 115, "lr": 8.74309793866058e-05} +{"train_loss": 0.016442911699414253, "grad_norm": 0.09782830625772476, "global_step": 12228, "epoch": 115, "lr": 8.742901434419881e-05} +{"train_loss": 0.01716838777065277, "grad_norm": 0.0528203621506691, "global_step": 12229, "epoch": 115, "lr": 8.742704917028257e-05} +{"train_loss": 0.01564181037247181, "grad_norm": 0.08814508467912674, "global_step": 12230, "epoch": 115, "lr": 8.742508386486392e-05} +{"train_loss": 0.018067482858896255, "grad_norm": 0.09236671030521393, "global_step": 12231, "epoch": 115, "lr": 8.742311842794984e-05} +{"train_loss": 0.013874001801013947, "grad_norm": 0.05754192918539047, "global_step": 12232, "epoch": 115, "lr": 8.742115285954718e-05} +{"train_loss": 0.016237592324614525, "grad_norm": 0.045691780745983124, "global_step": 12233, "epoch": 115, "lr": 8.741918715966285e-05} +{"train_loss": 0.016305914148688316, "grad_norm": 0.06779184192419052, "global_step": 12234, "epoch": 115, "lr": 8.741722132830377e-05} +{"train_loss": 0.015367233194410801, "grad_norm": 0.06802591681480408, "global_step": 12235, "epoch": 115, "lr": 8.741525536547685e-05} +{"train_loss": 0.017185386270284653, "grad_norm": 0.06766402721405029, "global_step": 12236, "epoch": 115, "lr": 8.741328927118902e-05} +{"train_loss": 0.017435912042856216, "grad_norm": 0.06564606726169586, "global_step": 12237, "epoch": 115, "lr": 8.741132304544713e-05} +{"train_loss": 0.017474573105573654, "grad_norm": 0.04284941032528877, "global_step": 12238, "epoch": 115, "lr": 8.740935668825814e-05} +{"train_loss": 0.01885809935629368, "grad_norm": 0.06358501315116882, "global_step": 12239, "epoch": 115, "lr": 8.740739019962893e-05} +{"train_loss": 0.015544815920293331, "grad_norm": 0.04938612878322601, "global_step": 12240, "epoch": 115, "lr": 8.740542357956641e-05} +{"train_loss": 0.015139509923756123, "grad_norm": 0.07619035243988037, "global_step": 12241, "epoch": 115, "lr": 8.740345682807751e-05} +{"train_loss": 0.018863873556256294, "grad_norm": 0.06778592616319656, "global_step": 12242, "epoch": 115, "lr": 8.740148994516912e-05} +{"train_loss": 0.016884295269846916, "grad_norm": 0.07917042076587677, "global_step": 12243, "epoch": 115, "lr": 8.739952293084818e-05} +{"train_loss": 0.016973644495010376, "grad_norm": 0.04573105275630951, "global_step": 12244, "epoch": 115, "lr": 8.739755578512157e-05} +{"train_loss": 0.017737887799739838, "grad_norm": 0.09009793400764465, "global_step": 12245, "epoch": 115, "lr": 8.73955885079962e-05} +{"train_loss": 0.016941340640187263, "grad_norm": 0.0706760361790657, "global_step": 12246, "epoch": 115, "lr": 8.739362109947901e-05} +{"train_loss": 0.017014428973197937, "grad_norm": 0.07509288936853409, "global_step": 12247, "epoch": 115, "lr": 8.73916535595769e-05} +{"train_loss": 0.016881488263607025, "grad_norm": 0.07699146866798401, "global_step": 12248, "epoch": 115, "lr": 8.738968588829677e-05} +{"train_loss": 0.01690148562192917, "grad_norm": 0.10980833321809769, "global_step": 12249, "epoch": 115, "lr": 8.738771808564555e-05} +{"train_loss": 0.018545027822256088, "grad_norm": 0.07506243139505386, "global_step": 12250, "epoch": 115, "lr": 8.738575015163015e-05} +{"train_loss": 0.014462852850556374, "grad_norm": 0.05630839988589287, "global_step": 12251, "epoch": 115, "lr": 8.738378208625748e-05} +{"train_loss": 0.018747957423329353, "grad_norm": 0.049201324582099915, "global_step": 12252, "epoch": 115, "lr": 8.738181388953446e-05} +{"train_loss": 0.016244936734437943, "grad_norm": 0.06988398730754852, "global_step": 12253, "epoch": 115, "lr": 8.7379845561468e-05} +{"train_loss": 0.018034031614661217, "grad_norm": 0.05887749418616295, "global_step": 12254, "epoch": 115, "lr": 8.737787710206502e-05} +{"train_loss": 0.016687646508216858, "grad_norm": 0.0728357806801796, "global_step": 12255, "epoch": 115, "lr": 8.737590851133243e-05} +{"train_loss": 0.018853453919291496, "grad_norm": 0.07434720546007156, "global_step": 12256, "epoch": 115, "lr": 8.737393978927715e-05} +{"train_loss": 0.016685504466295242, "grad_norm": 0.11475465446710587, "global_step": 12257, "epoch": 115, "lr": 8.737197093590613e-05} +{"train_loss": 0.02029142901301384, "grad_norm": 0.08722993731498718, "global_step": 12258, "epoch": 115, "lr": 8.737000195122621e-05} +{"train_loss": 0.01736420765519142, "grad_norm": 0.04749234765768051, "global_step": 12259, "epoch": 115, "lr": 8.736803283524438e-05} +{"train_loss": 0.016055840998888016, "grad_norm": 0.07906614243984222, "global_step": 12260, "epoch": 115, "lr": 8.736606358796752e-05} +{"train_loss": 0.01889406144618988, "grad_norm": 0.05850658938288689, "global_step": 12261, "epoch": 115, "lr": 8.736409420940258e-05} +{"train_loss": 0.018209882080554962, "grad_norm": 0.059573303908109665, "global_step": 12262, "epoch": 115, "lr": 8.736212469955646e-05} +{"train_loss": 0.018474871292710304, "grad_norm": 0.19469618797302246, "global_step": 12263, "epoch": 115, "lr": 8.736015505843608e-05} +{"train_loss": 0.01858110912144184, "grad_norm": 0.09847189486026764, "global_step": 12264, "epoch": 115, "lr": 8.735818528604836e-05} +{"train_loss": 0.017148684710264206, "grad_norm": 0.05863482505083084, "global_step": 12265, "epoch": 115, "lr": 8.735621538240022e-05} +{"train_loss": 0.016933564096689224, "grad_norm": 0.10085377842187881, "global_step": 12266, "epoch": 115, "lr": 8.735424534749859e-05} +{"train_loss": 0.018842846155166626, "grad_norm": 0.04558737203478813, "global_step": 12267, "epoch": 115, "lr": 8.735227518135039e-05} +{"train_loss": 0.019128790125250816, "grad_norm": 0.06074390560388565, "global_step": 12268, "epoch": 115, "lr": 8.735030488396251e-05} +{"train_loss": 0.01908530294895172, "grad_norm": 0.07614614814519882, "global_step": 12269, "epoch": 115, "lr": 8.734833445534194e-05} +{"train_loss": 0.01759968511760235, "grad_norm": 0.13778901100158691, "global_step": 12270, "epoch": 115, "lr": 8.734636389549555e-05} +{"train_loss": 0.019341738894581795, "grad_norm": 0.062808558344841, "global_step": 12271, "epoch": 115, "lr": 8.734439320443029e-05} +{"train_loss": 0.015849102288484573, "grad_norm": 0.04618509113788605, "global_step": 12272, "epoch": 115, "lr": 8.734242238215305e-05} +{"train_loss": 0.01634405367076397, "grad_norm": 0.0696028620004654, "global_step": 12273, "epoch": 115, "lr": 8.73404514286708e-05} +{"train_loss": 0.0175950787961483, "grad_norm": 0.08649829775094986, "global_step": 12274, "epoch": 115, "lr": 8.733848034399042e-05} +{"train_loss": 0.022486887872219086, "grad_norm": 0.08328370749950409, "global_step": 12275, "epoch": 115, "lr": 8.733650912811889e-05} +{"train_loss": 0.02012571692466736, "grad_norm": 0.09184595942497253, "global_step": 12276, "epoch": 115, "lr": 8.733453778106308e-05} +{"train_loss": 0.015339158475399017, "grad_norm": 0.04454445093870163, "global_step": 12277, "epoch": 115, "lr": 8.733256630282993e-05} +{"train_loss": 0.01597854122519493, "grad_norm": 0.07924820482730865, "global_step": 12278, "epoch": 115, "lr": 8.73305946934264e-05} +{"train_loss": 0.01755852997303009, "grad_norm": 0.09899482131004333, "global_step": 12279, "epoch": 115, "lr": 8.73286229528594e-05} +{"train_loss": 0.018937241286039352, "grad_norm": 0.06160367652773857, "global_step": 12280, "epoch": 115, "lr": 8.732665108113582e-05} +{"train_loss": 0.01703113690018654, "grad_norm": 0.06530656665563583, "global_step": 12281, "epoch": 115, "lr": 8.732467907826265e-05} +{"train_loss": 0.016630955040454865, "grad_norm": 0.08443761616945267, "global_step": 12282, "epoch": 115, "lr": 8.732270694424678e-05} +{"train_loss": 0.017661387100815773, "grad_norm": 0.2030550092458725, "global_step": 12283, "epoch": 115, "lr": 8.732073467909515e-05} +{"train_loss": 0.01831880584359169, "grad_norm": 0.10294688493013382, "global_step": 12284, "epoch": 115, "lr": 8.731876228281469e-05} +{"train_loss": 0.015657465904951096, "grad_norm": 0.12049433588981628, "global_step": 12285, "epoch": 115, "lr": 8.731678975541233e-05} +{"train_loss": 0.017785649746656418, "grad_norm": 0.059911035001277924, "global_step": 12286, "epoch": 115, "lr": 8.7314817096895e-05} +{"train_loss": 0.018525928258895874, "grad_norm": 0.06671634316444397, "global_step": 12287, "epoch": 115, "lr": 8.731284430726963e-05} +{"train_loss": 0.015513187274336815, "grad_norm": 0.046526551246643066, "global_step": 12288, "epoch": 115, "lr": 8.731087138654315e-05} +{"train_loss": 0.01666928455233574, "grad_norm": 0.07104044407606125, "global_step": 12289, "epoch": 115, "lr": 8.730889833472249e-05} +{"train_loss": 0.017157617956399918, "grad_norm": 0.08438485860824585, "global_step": 12290, "epoch": 115, "lr": 8.73069251518146e-05} +{"train_loss": 0.017367742955684662, "grad_norm": 0.0511719174683094, "global_step": 12291, "epoch": 115, "lr": 8.730495183782639e-05} +{"train_loss": 0.015380950644612312, "grad_norm": 0.06549853831529617, "global_step": 12292, "epoch": 115, "lr": 8.73029783927648e-05} +{"train_loss": 0.017728354781866074, "grad_norm": 0.050552669912576675, "global_step": 12293, "epoch": 115, "lr": 8.730100481663676e-05} +{"train_loss": 0.019164450466632843, "grad_norm": 0.07474514842033386, "global_step": 12294, "epoch": 115, "lr": 8.729903110944922e-05} +{"train_loss": 0.01717684231698513, "grad_norm": 0.12129269540309906, "global_step": 12295, "epoch": 115, "lr": 8.729705727120911e-05, "val_loss": 0.01907738856971264} +{"train_loss": 0.018929917365312576, "grad_norm": 0.08165692538022995, "global_step": 12296, "epoch": 116, "lr": 8.729508330192335e-05} +{"train_loss": 0.015834931284189224, "grad_norm": 0.05753571540117264, "global_step": 12297, "epoch": 116, "lr": 8.72931092015989e-05} +{"train_loss": 0.019707828760147095, "grad_norm": 0.085346519947052, "global_step": 12298, "epoch": 116, "lr": 8.729113497024268e-05} +{"train_loss": 0.01821705885231495, "grad_norm": 0.0758269727230072, "global_step": 12299, "epoch": 116, "lr": 8.728916060786162e-05} +{"train_loss": 0.01511156652122736, "grad_norm": 0.11085449904203415, "global_step": 12300, "epoch": 116, "lr": 8.728718611446267e-05} +{"train_loss": 0.01814301125705242, "grad_norm": 0.08470538258552551, "global_step": 12301, "epoch": 116, "lr": 8.728521149005278e-05} +{"train_loss": 0.016316523775458336, "grad_norm": 0.054410770535469055, "global_step": 12302, "epoch": 116, "lr": 8.728323673463885e-05} +{"train_loss": 0.014478553086519241, "grad_norm": 0.06281623989343643, "global_step": 12303, "epoch": 116, "lr": 8.728126184822784e-05} +{"train_loss": 0.017057664692401886, "grad_norm": 0.07379403710365295, "global_step": 12304, "epoch": 116, "lr": 8.72792868308267e-05} +{"train_loss": 0.018234532326459885, "grad_norm": 0.09157740324735641, "global_step": 12305, "epoch": 116, "lr": 8.727731168244236e-05} +{"train_loss": 0.017691846936941147, "grad_norm": 0.11676228791475296, "global_step": 12306, "epoch": 116, "lr": 8.727533640308174e-05} +{"train_loss": 0.01650119200348854, "grad_norm": 0.06781711429357529, "global_step": 12307, "epoch": 116, "lr": 8.727336099275182e-05} +{"train_loss": 0.016352638602256775, "grad_norm": 0.10273722559213638, "global_step": 12308, "epoch": 116, "lr": 8.727138545145951e-05} +{"train_loss": 0.017959147691726685, "grad_norm": 0.056641001254320145, "global_step": 12309, "epoch": 116, "lr": 8.726940977921176e-05} +{"train_loss": 0.016197971999645233, "grad_norm": 0.15914995968341827, "global_step": 12310, "epoch": 116, "lr": 8.726743397601551e-05} +{"train_loss": 0.018147222697734833, "grad_norm": 0.06012052297592163, "global_step": 12311, "epoch": 116, "lr": 8.726545804187771e-05} +{"train_loss": 0.018431726843118668, "grad_norm": 0.08329133689403534, "global_step": 12312, "epoch": 116, "lr": 8.726348197680529e-05} +{"train_loss": 0.017660457640886307, "grad_norm": 0.08715220540761948, "global_step": 12313, "epoch": 116, "lr": 8.726150578080521e-05} +{"train_loss": 0.01917880028486252, "grad_norm": 0.08145851641893387, "global_step": 12314, "epoch": 116, "lr": 8.72595294538844e-05} +{"train_loss": 0.01770903170108795, "grad_norm": 0.05474630370736122, "global_step": 12315, "epoch": 116, "lr": 8.72575529960498e-05} +{"train_loss": 0.017174547538161278, "grad_norm": 0.052869733422994614, "global_step": 12316, "epoch": 116, "lr": 8.725557640730836e-05} +{"train_loss": 0.017203127965331078, "grad_norm": 0.04749562218785286, "global_step": 12317, "epoch": 116, "lr": 8.725359968766704e-05} +{"train_loss": 0.017399275675415993, "grad_norm": 0.1167249009013176, "global_step": 12318, "epoch": 116, "lr": 8.725162283713275e-05} +{"train_loss": 0.014992034994065762, "grad_norm": 0.06104693189263344, "global_step": 12319, "epoch": 116, "lr": 8.724964585571248e-05} +{"train_loss": 0.01946437358856201, "grad_norm": 0.10965386033058167, "global_step": 12320, "epoch": 116, "lr": 8.724766874341314e-05} +{"train_loss": 0.016909101977944374, "grad_norm": 0.11689770221710205, "global_step": 12321, "epoch": 116, "lr": 8.724569150024169e-05} +{"train_loss": 0.01650194451212883, "grad_norm": 0.05261331424117088, "global_step": 12322, "epoch": 116, "lr": 8.724371412620509e-05} +{"train_loss": 0.019588500261306763, "grad_norm": 0.19325414299964905, "global_step": 12323, "epoch": 116, "lr": 8.724173662131028e-05} +{"train_loss": 0.01662643812596798, "grad_norm": 0.05635170638561249, "global_step": 12324, "epoch": 116, "lr": 8.723975898556419e-05} +{"train_loss": 0.018820300698280334, "grad_norm": 0.07087104022502899, "global_step": 12325, "epoch": 116, "lr": 8.723778121897378e-05} +{"train_loss": 0.014936949126422405, "grad_norm": 0.1435302495956421, "global_step": 12326, "epoch": 116, "lr": 8.7235803321546e-05} +{"train_loss": 0.016860313713550568, "grad_norm": 0.090992771089077, "global_step": 12327, "epoch": 116, "lr": 8.723382529328781e-05} +{"train_loss": 0.015258138999342918, "grad_norm": 0.07007063925266266, "global_step": 12328, "epoch": 116, "lr": 8.723184713420616e-05} +{"train_loss": 0.018507089465856552, "grad_norm": 0.06319648772478104, "global_step": 12329, "epoch": 116, "lr": 8.722986884430797e-05} +{"train_loss": 0.016792703419923782, "grad_norm": 0.1554967761039734, "global_step": 12330, "epoch": 116, "lr": 8.722789042360022e-05} +{"train_loss": 0.015168147161602974, "grad_norm": 0.049748968333005905, "global_step": 12331, "epoch": 116, "lr": 8.722591187208985e-05} +{"train_loss": 0.018866468220949173, "grad_norm": 0.10826730728149414, "global_step": 12332, "epoch": 116, "lr": 8.722393318978382e-05} +{"train_loss": 0.01599622145295143, "grad_norm": 0.12363680452108383, "global_step": 12333, "epoch": 116, "lr": 8.722195437668907e-05} +{"train_loss": 0.017224110662937164, "grad_norm": 0.08668754994869232, "global_step": 12334, "epoch": 116, "lr": 8.721997543281257e-05} +{"train_loss": 0.016744837164878845, "grad_norm": 0.11329542845487595, "global_step": 12335, "epoch": 116, "lr": 8.721799635816126e-05} +{"train_loss": 0.01693926565349102, "grad_norm": 0.0972914770245552, "global_step": 12336, "epoch": 116, "lr": 8.721601715274208e-05} +{"train_loss": 0.017613697797060013, "grad_norm": 0.10648172348737717, "global_step": 12337, "epoch": 116, "lr": 8.721403781656202e-05} +{"train_loss": 0.017194556072354317, "grad_norm": 0.06522958725690842, "global_step": 12338, "epoch": 116, "lr": 8.721205834962802e-05} +{"train_loss": 0.015972062945365906, "grad_norm": 0.10338401794433594, "global_step": 12339, "epoch": 116, "lr": 8.721007875194701e-05} +{"train_loss": 0.01586301438510418, "grad_norm": 0.06058149412274361, "global_step": 12340, "epoch": 116, "lr": 8.720809902352597e-05} +{"train_loss": 0.018833409994840622, "grad_norm": 0.06378988176584244, "global_step": 12341, "epoch": 116, "lr": 8.720611916437187e-05} +{"train_loss": 0.019264761358499527, "grad_norm": 0.05714336410164833, "global_step": 12342, "epoch": 116, "lr": 8.720413917449163e-05} +{"train_loss": 0.016683848574757576, "grad_norm": 0.07446408271789551, "global_step": 12343, "epoch": 116, "lr": 8.720215905389222e-05} +{"train_loss": 0.015747370198369026, "grad_norm": 0.0600387342274189, "global_step": 12344, "epoch": 116, "lr": 8.720017880258063e-05} +{"train_loss": 0.01630948670208454, "grad_norm": 0.08822224289178848, "global_step": 12345, "epoch": 116, "lr": 8.719819842056375e-05} +{"train_loss": 0.019074559211730957, "grad_norm": 0.05675419420003891, "global_step": 12346, "epoch": 116, "lr": 8.71962179078486e-05} +{"train_loss": 0.015537865459918976, "grad_norm": 0.0820324644446373, "global_step": 12347, "epoch": 116, "lr": 8.719423726444211e-05} +{"train_loss": 0.018078554421663284, "grad_norm": 0.052049312740564346, "global_step": 12348, "epoch": 116, "lr": 8.719225649035126e-05} +{"train_loss": 0.017122307792305946, "grad_norm": 0.05199343338608742, "global_step": 12349, "epoch": 116, "lr": 8.7190275585583e-05} +{"train_loss": 0.015884852036833763, "grad_norm": 0.055921606719493866, "global_step": 12350, "epoch": 116, "lr": 8.718829455014425e-05} +{"train_loss": 0.01811455935239792, "grad_norm": 0.1542152762413025, "global_step": 12351, "epoch": 116, "lr": 8.718631338404203e-05} +{"train_loss": 0.018556229770183563, "grad_norm": 0.11326557397842407, "global_step": 12352, "epoch": 116, "lr": 8.718433208728327e-05} +{"train_loss": 0.017716120928525925, "grad_norm": 0.1261867731809616, "global_step": 12353, "epoch": 116, "lr": 8.718235065987495e-05} +{"train_loss": 0.016116898506879807, "grad_norm": 0.1966598778963089, "global_step": 12354, "epoch": 116, "lr": 8.718036910182402e-05} +{"train_loss": 0.017437241971492767, "grad_norm": 0.05873831361532211, "global_step": 12355, "epoch": 116, "lr": 8.717838741313742e-05} +{"train_loss": 0.0176609605550766, "grad_norm": 0.08713236451148987, "global_step": 12356, "epoch": 116, "lr": 8.717640559382214e-05} +{"train_loss": 0.015990646556019783, "grad_norm": 0.07747343927621841, "global_step": 12357, "epoch": 116, "lr": 8.717442364388515e-05} +{"train_loss": 0.019040007144212723, "grad_norm": 0.11288702487945557, "global_step": 12358, "epoch": 116, "lr": 8.71724415633334e-05} +{"train_loss": 0.015195381827652454, "grad_norm": 0.06933547556400299, "global_step": 12359, "epoch": 116, "lr": 8.717045935217385e-05} +{"train_loss": 0.019164826720952988, "grad_norm": 0.08210372924804688, "global_step": 12360, "epoch": 116, "lr": 8.716847701041349e-05} +{"train_loss": 0.017070535570383072, "grad_norm": 0.1182292178273201, "global_step": 12361, "epoch": 116, "lr": 8.716649453805924e-05} +{"train_loss": 0.016088586300611496, "grad_norm": 0.08424320071935654, "global_step": 12362, "epoch": 116, "lr": 8.71645119351181e-05} +{"train_loss": 0.015969693660736084, "grad_norm": 0.06132782995700836, "global_step": 12363, "epoch": 116, "lr": 8.716252920159704e-05} +{"train_loss": 0.018410298973321915, "grad_norm": 0.13655641674995422, "global_step": 12364, "epoch": 116, "lr": 8.716054633750299e-05} +{"train_loss": 0.015503006055951118, "grad_norm": 0.17798644304275513, "global_step": 12365, "epoch": 116, "lr": 8.715856334284296e-05} +{"train_loss": 0.016945891082286835, "grad_norm": 0.1419173926115036, "global_step": 12366, "epoch": 116, "lr": 8.71565802176239e-05} +{"train_loss": 0.016610244289040565, "grad_norm": 0.07764773070812225, "global_step": 12367, "epoch": 116, "lr": 8.715459696185278e-05} +{"train_loss": 0.017600782215595245, "grad_norm": 0.06961943209171295, "global_step": 12368, "epoch": 116, "lr": 8.715261357553655e-05} +{"train_loss": 0.018981266766786575, "grad_norm": 0.0882386788725853, "global_step": 12369, "epoch": 116, "lr": 8.715063005868219e-05} +{"train_loss": 0.01609675958752632, "grad_norm": 0.06450820714235306, "global_step": 12370, "epoch": 116, "lr": 8.714864641129667e-05} +{"train_loss": 0.016489151865243912, "grad_norm": 0.06857701390981674, "global_step": 12371, "epoch": 116, "lr": 8.714666263338697e-05} +{"train_loss": 0.01677502691745758, "grad_norm": 0.12240809202194214, "global_step": 12372, "epoch": 116, "lr": 8.714467872496006e-05} +{"train_loss": 0.017949771136045456, "grad_norm": 0.06079063192009926, "global_step": 12373, "epoch": 116, "lr": 8.71426946860229e-05} +{"train_loss": 0.014616112224757671, "grad_norm": 0.13124214112758636, "global_step": 12374, "epoch": 116, "lr": 8.714071051658245e-05} +{"train_loss": 0.017463358119130135, "grad_norm": 0.17185530066490173, "global_step": 12375, "epoch": 116, "lr": 8.713872621664572e-05} +{"train_loss": 0.014713658019900322, "grad_norm": 0.05572263151407242, "global_step": 12376, "epoch": 116, "lr": 8.713674178621965e-05} +{"train_loss": 0.01786152459681034, "grad_norm": 0.0772947445511818, "global_step": 12377, "epoch": 116, "lr": 8.71347572253112e-05} +{"train_loss": 0.016051052138209343, "grad_norm": 0.05703042820096016, "global_step": 12378, "epoch": 116, "lr": 8.713277253392738e-05} +{"train_loss": 0.017161794006824493, "grad_norm": 0.15173423290252686, "global_step": 12379, "epoch": 116, "lr": 8.713078771207515e-05} +{"train_loss": 0.019535358995199203, "grad_norm": 0.07194526493549347, "global_step": 12380, "epoch": 116, "lr": 8.712880275976145e-05} +{"train_loss": 0.015665274113416672, "grad_norm": 0.10145024210214615, "global_step": 12381, "epoch": 116, "lr": 8.712681767699331e-05} +{"train_loss": 0.01949179917573929, "grad_norm": 0.07473615556955338, "global_step": 12382, "epoch": 116, "lr": 8.712483246377767e-05} +{"train_loss": 0.01878034882247448, "grad_norm": 0.05279521271586418, "global_step": 12383, "epoch": 116, "lr": 8.712284712012154e-05} +{"train_loss": 0.015850737690925598, "grad_norm": 0.06278755515813828, "global_step": 12384, "epoch": 116, "lr": 8.712086164603183e-05} +{"train_loss": 0.016405537724494934, "grad_norm": 0.05815871059894562, "global_step": 12385, "epoch": 116, "lr": 8.711887604151559e-05} +{"train_loss": 0.017905447632074356, "grad_norm": 0.09872853755950928, "global_step": 12386, "epoch": 116, "lr": 8.711689030657972e-05} +{"train_loss": 0.016935035586357117, "grad_norm": 0.17330250144004822, "global_step": 12387, "epoch": 116, "lr": 8.711490444123127e-05} +{"train_loss": 0.02019677869975567, "grad_norm": 0.10058128088712692, "global_step": 12388, "epoch": 116, "lr": 8.711291844547717e-05} +{"train_loss": 0.018023192882537842, "grad_norm": 0.07152354717254639, "global_step": 12389, "epoch": 116, "lr": 8.711093231932443e-05} +{"train_loss": 0.01391138881444931, "grad_norm": 0.11631260067224503, "global_step": 12390, "epoch": 116, "lr": 8.710894606278002e-05} +{"train_loss": 0.0172843337059021, "grad_norm": 0.13897690176963806, "global_step": 12391, "epoch": 116, "lr": 8.71069596758509e-05} +{"train_loss": 0.01981009915471077, "grad_norm": 0.08286131918430328, "global_step": 12392, "epoch": 116, "lr": 8.710497315854406e-05} +{"train_loss": 0.016688324511051178, "grad_norm": 0.0806417316198349, "global_step": 12393, "epoch": 116, "lr": 8.710298651086648e-05} +{"train_loss": 0.0173905361443758, "grad_norm": 0.05250968411564827, "global_step": 12394, "epoch": 116, "lr": 8.710099973282515e-05} +{"train_loss": 0.018519770354032516, "grad_norm": 0.04682442173361778, "global_step": 12395, "epoch": 116, "lr": 8.709901282442703e-05} +{"train_loss": 0.017224367707967758, "grad_norm": 0.06853315234184265, "global_step": 12396, "epoch": 116, "lr": 8.70970257856791e-05} +{"train_loss": 0.01870325393974781, "grad_norm": 0.0641130730509758, "global_step": 12397, "epoch": 116, "lr": 8.709503861658838e-05} +{"train_loss": 0.02004566229879856, "grad_norm": 0.08792631328105927, "global_step": 12398, "epoch": 116, "lr": 8.709305131716182e-05} +{"train_loss": 0.014713393524289131, "grad_norm": 0.07310860604047775, "global_step": 12399, "epoch": 116, "lr": 8.709106388740642e-05} +{"train_loss": 0.018008427694439888, "grad_norm": 0.07705813646316528, "global_step": 12400, "epoch": 116, "lr": 8.708907632732912e-05} +{"train_loss": 0.017247039464495656, "grad_norm": 0.10758586972951889, "global_step": 12401, "epoch": 116, "lr": 8.708708863693697e-05, "val_loss": 0.01942635327577591} +{"train_loss": 0.015293230302631855, "grad_norm": 0.06401865184307098, "global_step": 12402, "epoch": 117, "lr": 8.70851008162369e-05} +{"train_loss": 0.014697302132844925, "grad_norm": 0.06633750349283218, "global_step": 12403, "epoch": 117, "lr": 8.708311286523592e-05} +{"train_loss": 0.015157168731093407, "grad_norm": 0.13876517117023468, "global_step": 12404, "epoch": 117, "lr": 8.708112478394101e-05} +{"train_loss": 0.01784929819405079, "grad_norm": 0.05666075646877289, "global_step": 12405, "epoch": 117, "lr": 8.707913657235915e-05} +{"train_loss": 0.01532820239663124, "grad_norm": 0.07290168106555939, "global_step": 12406, "epoch": 117, "lr": 8.707714823049733e-05} +{"train_loss": 0.017888028174638748, "grad_norm": 0.049300603568553925, "global_step": 12407, "epoch": 117, "lr": 8.707515975836253e-05} +{"train_loss": 0.01805264875292778, "grad_norm": 0.0756937563419342, "global_step": 12408, "epoch": 117, "lr": 8.707317115596174e-05} +{"train_loss": 0.016903944313526154, "grad_norm": 0.1033003032207489, "global_step": 12409, "epoch": 117, "lr": 8.707118242330196e-05} +{"train_loss": 0.018677644431591034, "grad_norm": 0.07144372165203094, "global_step": 12410, "epoch": 117, "lr": 8.706919356039016e-05} +{"train_loss": 0.017101900652050972, "grad_norm": 0.09938283264636993, "global_step": 12411, "epoch": 117, "lr": 8.706720456723334e-05} +{"train_loss": 0.017457973212003708, "grad_norm": 0.08204330503940582, "global_step": 12412, "epoch": 117, "lr": 8.706521544383848e-05} +{"train_loss": 0.018542911857366562, "grad_norm": 0.11432845890522003, "global_step": 12413, "epoch": 117, "lr": 8.706322619021257e-05} +{"train_loss": 0.01566249504685402, "grad_norm": 0.05112133175134659, "global_step": 12414, "epoch": 117, "lr": 8.70612368063626e-05} +{"train_loss": 0.017061296850442886, "grad_norm": 0.13420288264751434, "global_step": 12415, "epoch": 117, "lr": 8.705924729229558e-05} +{"train_loss": 0.017827246338129044, "grad_norm": 0.07490010559558868, "global_step": 12416, "epoch": 117, "lr": 8.705725764801847e-05} +{"train_loss": 0.016667203977704048, "grad_norm": 0.06317628175020218, "global_step": 12417, "epoch": 117, "lr": 8.705526787353826e-05} +{"train_loss": 0.01776827499270439, "grad_norm": 0.06153438985347748, "global_step": 12418, "epoch": 117, "lr": 8.705327796886197e-05} +{"train_loss": 0.016710512340068817, "grad_norm": 0.07775948941707611, "global_step": 12419, "epoch": 117, "lr": 8.705128793399657e-05} +{"train_loss": 0.016623906791210175, "grad_norm": 0.09069197624921799, "global_step": 12420, "epoch": 117, "lr": 8.704929776894905e-05} +{"train_loss": 0.015260996297001839, "grad_norm": 0.0678606629371643, "global_step": 12421, "epoch": 117, "lr": 8.704730747372642e-05} +{"train_loss": 0.01907072588801384, "grad_norm": 0.102995365858078, "global_step": 12422, "epoch": 117, "lr": 8.704531704833566e-05} +{"train_loss": 0.016714755445718765, "grad_norm": 0.05582714453339577, "global_step": 12423, "epoch": 117, "lr": 8.704332649278376e-05} +{"train_loss": 0.015254633501172066, "grad_norm": 0.042216043919324875, "global_step": 12424, "epoch": 117, "lr": 8.704133580707772e-05} +{"train_loss": 0.01795719563961029, "grad_norm": 0.09544622898101807, "global_step": 12425, "epoch": 117, "lr": 8.703934499122454e-05} +{"train_loss": 0.01774045079946518, "grad_norm": 0.05783762410283089, "global_step": 12426, "epoch": 117, "lr": 8.703735404523122e-05} +{"train_loss": 0.01897614449262619, "grad_norm": 0.08629865199327469, "global_step": 12427, "epoch": 117, "lr": 8.703536296910473e-05} +{"train_loss": 0.01961784064769745, "grad_norm": 0.09135254472494125, "global_step": 12428, "epoch": 117, "lr": 8.703337176285209e-05} +{"train_loss": 0.018715400248765945, "grad_norm": 0.10041240602731705, "global_step": 12429, "epoch": 117, "lr": 8.703138042648029e-05} +{"train_loss": 0.01516290195286274, "grad_norm": 0.10004201531410217, "global_step": 12430, "epoch": 117, "lr": 8.70293889599963e-05} +{"train_loss": 0.01976744830608368, "grad_norm": 0.15442931652069092, "global_step": 12431, "epoch": 117, "lr": 8.702739736340717e-05} +{"train_loss": 0.019732151180505753, "grad_norm": 0.05965188145637512, "global_step": 12432, "epoch": 117, "lr": 8.702540563671985e-05} +{"train_loss": 0.015300079248845577, "grad_norm": 0.1651650220155716, "global_step": 12433, "epoch": 117, "lr": 8.702341377994136e-05} +{"train_loss": 0.017746511846780777, "grad_norm": 0.05307527631521225, "global_step": 12434, "epoch": 117, "lr": 8.70214217930787e-05} +{"train_loss": 0.019371986389160156, "grad_norm": 0.07381534576416016, "global_step": 12435, "epoch": 117, "lr": 8.701942967613888e-05} +{"train_loss": 0.020881768316030502, "grad_norm": 0.07609540969133377, "global_step": 12436, "epoch": 117, "lr": 8.701743742912886e-05} +{"train_loss": 0.016795288771390915, "grad_norm": 0.06284351646900177, "global_step": 12437, "epoch": 117, "lr": 8.701544505205567e-05} +{"train_loss": 0.015885023400187492, "grad_norm": 0.10385691374540329, "global_step": 12438, "epoch": 117, "lr": 8.70134525449263e-05} +{"train_loss": 0.01753290370106697, "grad_norm": 0.08104958385229111, "global_step": 12439, "epoch": 117, "lr": 8.701145990774775e-05} +{"train_loss": 0.019686494022607803, "grad_norm": 0.14642630517482758, "global_step": 12440, "epoch": 117, "lr": 8.700946714052703e-05} +{"train_loss": 0.019453197717666626, "grad_norm": 0.07989346235990524, "global_step": 12441, "epoch": 117, "lr": 8.700747424327115e-05} +{"train_loss": 0.018418120220303535, "grad_norm": 0.14779825508594513, "global_step": 12442, "epoch": 117, "lr": 8.700548121598709e-05} +{"train_loss": 0.017687205225229263, "grad_norm": 0.14229585230350494, "global_step": 12443, "epoch": 117, "lr": 8.700348805868185e-05} +{"train_loss": 0.01672140136361122, "grad_norm": 0.0649610161781311, "global_step": 12444, "epoch": 117, "lr": 8.700149477136247e-05} +{"train_loss": 0.017010558396577835, "grad_norm": 0.07550591230392456, "global_step": 12445, "epoch": 117, "lr": 8.699950135403592e-05} +{"train_loss": 0.018123526126146317, "grad_norm": 0.0795903429389, "global_step": 12446, "epoch": 117, "lr": 8.69975078067092e-05} +{"train_loss": 0.014910629950463772, "grad_norm": 0.05021924152970314, "global_step": 12447, "epoch": 117, "lr": 8.699551412938933e-05} +{"train_loss": 0.01672990620136261, "grad_norm": 0.09210081398487091, "global_step": 12448, "epoch": 117, "lr": 8.699352032208332e-05} +{"train_loss": 0.01677776873111725, "grad_norm": 0.06162245571613312, "global_step": 12449, "epoch": 117, "lr": 8.699152638479817e-05} +{"train_loss": 0.018122190609574318, "grad_norm": 0.05962088331580162, "global_step": 12450, "epoch": 117, "lr": 8.698953231754088e-05} +{"train_loss": 0.014560001902282238, "grad_norm": 0.07446738332509995, "global_step": 12451, "epoch": 117, "lr": 8.698753812031846e-05} +{"train_loss": 0.017683181911706924, "grad_norm": 0.056295063346624374, "global_step": 12452, "epoch": 117, "lr": 8.698554379313793e-05} +{"train_loss": 0.017523696646094322, "grad_norm": 0.07020632922649384, "global_step": 12453, "epoch": 117, "lr": 8.698354933600625e-05} +{"train_loss": 0.015680477023124695, "grad_norm": 0.12245079129934311, "global_step": 12454, "epoch": 117, "lr": 8.69815547489305e-05} +{"train_loss": 0.017249081283807755, "grad_norm": 0.05819665640592575, "global_step": 12455, "epoch": 117, "lr": 8.697956003191762e-05} +{"train_loss": 0.0166095569729805, "grad_norm": 0.054359208792448044, "global_step": 12456, "epoch": 117, "lr": 8.697756518497466e-05} +{"train_loss": 0.01649303175508976, "grad_norm": 0.09658188372850418, "global_step": 12457, "epoch": 117, "lr": 8.697557020810861e-05} +{"train_loss": 0.015198204666376114, "grad_norm": 0.11668065190315247, "global_step": 12458, "epoch": 117, "lr": 8.69735751013265e-05} +{"train_loss": 0.017552198842167854, "grad_norm": 0.1032191812992096, "global_step": 12459, "epoch": 117, "lr": 8.697157986463532e-05} +{"train_loss": 0.01539330743253231, "grad_norm": 0.1188134253025055, "global_step": 12460, "epoch": 117, "lr": 8.696958449804209e-05} +{"train_loss": 0.01705138199031353, "grad_norm": 0.0791199654340744, "global_step": 12461, "epoch": 117, "lr": 8.696758900155381e-05} +{"train_loss": 0.015097012743353844, "grad_norm": 0.06139836087822914, "global_step": 12462, "epoch": 117, "lr": 8.696559337517749e-05} +{"train_loss": 0.019473567605018616, "grad_norm": 0.08998657017946243, "global_step": 12463, "epoch": 117, "lr": 8.696359761892017e-05} +{"train_loss": 0.01607939973473549, "grad_norm": 0.04922930523753166, "global_step": 12464, "epoch": 117, "lr": 8.696160173278885e-05} +{"train_loss": 0.01737738773226738, "grad_norm": 0.09385792165994644, "global_step": 12465, "epoch": 117, "lr": 8.695960571679052e-05} +{"train_loss": 0.016016153618693352, "grad_norm": 0.07309630513191223, "global_step": 12466, "epoch": 117, "lr": 8.69576095709322e-05} +{"train_loss": 0.015969518572092056, "grad_norm": 0.07554783672094345, "global_step": 12467, "epoch": 117, "lr": 8.695561329522093e-05} +{"train_loss": 0.018789539113640785, "grad_norm": 0.05602855607867241, "global_step": 12468, "epoch": 117, "lr": 8.695361688966369e-05} +{"train_loss": 0.01725752465426922, "grad_norm": 0.07398587465286255, "global_step": 12469, "epoch": 117, "lr": 8.695162035426751e-05} +{"train_loss": 0.01716422103345394, "grad_norm": 0.11704517900943756, "global_step": 12470, "epoch": 117, "lr": 8.694962368903943e-05} +{"train_loss": 0.018948931246995926, "grad_norm": 0.1371377855539322, "global_step": 12471, "epoch": 117, "lr": 8.694762689398644e-05} +{"train_loss": 0.016981756314635277, "grad_norm": 0.0743698924779892, "global_step": 12472, "epoch": 117, "lr": 8.694562996911553e-05} +{"train_loss": 0.015554752200841904, "grad_norm": 0.09351043403148651, "global_step": 12473, "epoch": 117, "lr": 8.694363291443376e-05} +{"train_loss": 0.014595345593988895, "grad_norm": 0.05671701952815056, "global_step": 12474, "epoch": 117, "lr": 8.694163572994813e-05} +{"train_loss": 0.015543131157755852, "grad_norm": 0.06149708479642868, "global_step": 12475, "epoch": 117, "lr": 8.693963841566566e-05} +{"train_loss": 0.01756289228796959, "grad_norm": 0.11941487342119217, "global_step": 12476, "epoch": 117, "lr": 8.693764097159336e-05} +{"train_loss": 0.019564848393201828, "grad_norm": 0.14306911826133728, "global_step": 12477, "epoch": 117, "lr": 8.693564339773825e-05} +{"train_loss": 0.017327945679426193, "grad_norm": 0.09145238250494003, "global_step": 12478, "epoch": 117, "lr": 8.693364569410736e-05} +{"train_loss": 0.015364836901426315, "grad_norm": 0.08823343366384506, "global_step": 12479, "epoch": 117, "lr": 8.693164786070769e-05} +{"train_loss": 0.016091102734208107, "grad_norm": 0.0571255162358284, "global_step": 12480, "epoch": 117, "lr": 8.692964989754628e-05} +{"train_loss": 0.016317307949066162, "grad_norm": 0.1057671532034874, "global_step": 12481, "epoch": 117, "lr": 8.692765180463014e-05} +{"train_loss": 0.016569610685110092, "grad_norm": 0.059268612414598465, "global_step": 12482, "epoch": 117, "lr": 8.69256535819663e-05} +{"train_loss": 0.018471043556928635, "grad_norm": 0.06146659329533577, "global_step": 12483, "epoch": 117, "lr": 8.692365522956176e-05} +{"train_loss": 0.018242210149765015, "grad_norm": 0.1035069078207016, "global_step": 12484, "epoch": 117, "lr": 8.692165674742355e-05} +{"train_loss": 0.017693109810352325, "grad_norm": 0.11202798783779144, "global_step": 12485, "epoch": 117, "lr": 8.69196581355587e-05} +{"train_loss": 0.01828928105533123, "grad_norm": 0.05714711174368858, "global_step": 12486, "epoch": 117, "lr": 8.691765939397423e-05} +{"train_loss": 0.01838676817715168, "grad_norm": 0.10005442053079605, "global_step": 12487, "epoch": 117, "lr": 8.691566052267717e-05} +{"train_loss": 0.017433561384677887, "grad_norm": 0.07216369360685349, "global_step": 12488, "epoch": 117, "lr": 8.69136615216745e-05} +{"train_loss": 0.017335275188088417, "grad_norm": 0.08884216099977493, "global_step": 12489, "epoch": 117, "lr": 8.69116623909733e-05} +{"train_loss": 0.016491040587425232, "grad_norm": 0.09562492370605469, "global_step": 12490, "epoch": 117, "lr": 8.690966313058058e-05} +{"train_loss": 0.016334448009729385, "grad_norm": 0.14617298543453217, "global_step": 12491, "epoch": 117, "lr": 8.690766374050334e-05} +{"train_loss": 0.020249150693416595, "grad_norm": 0.05315719172358513, "global_step": 12492, "epoch": 117, "lr": 8.690566422074862e-05} +{"train_loss": 0.01683533936738968, "grad_norm": 0.07782968878746033, "global_step": 12493, "epoch": 117, "lr": 8.690366457132345e-05} +{"train_loss": 0.019522089511156082, "grad_norm": 0.07524776458740234, "global_step": 12494, "epoch": 117, "lr": 8.690166479223485e-05} +{"train_loss": 0.017140258103609085, "grad_norm": 0.05688372999429703, "global_step": 12495, "epoch": 117, "lr": 8.689966488348984e-05} +{"train_loss": 0.01702607050538063, "grad_norm": 0.09835480898618698, "global_step": 12496, "epoch": 117, "lr": 8.689766484509547e-05} +{"train_loss": 0.017769156023859978, "grad_norm": 0.07422901690006256, "global_step": 12497, "epoch": 117, "lr": 8.689566467705874e-05} +{"train_loss": 0.01619541272521019, "grad_norm": 0.10352862626314163, "global_step": 12498, "epoch": 117, "lr": 8.689366437938671e-05} +{"train_loss": 0.01859200932085514, "grad_norm": 0.05208205804228783, "global_step": 12499, "epoch": 117, "lr": 8.689166395208636e-05} +{"train_loss": 0.017931289970874786, "grad_norm": 0.12788096070289612, "global_step": 12500, "epoch": 117, "lr": 8.688966339516477e-05} +{"train_loss": 0.0158110111951828, "grad_norm": 0.11864018440246582, "global_step": 12501, "epoch": 117, "lr": 8.688766270862894e-05} +{"train_loss": 0.017029516398906708, "grad_norm": 0.09127907454967499, "global_step": 12502, "epoch": 117, "lr": 8.688566189248591e-05} +{"train_loss": 0.016767723485827446, "grad_norm": 0.11066965013742447, "global_step": 12503, "epoch": 117, "lr": 8.688366094674268e-05} +{"train_loss": 0.01581060327589512, "grad_norm": 0.09564303606748581, "global_step": 12504, "epoch": 117, "lr": 8.688165987140633e-05} +{"train_loss": 0.01643357053399086, "grad_norm": 0.10182607918977737, "global_step": 12505, "epoch": 117, "lr": 8.687965866648386e-05} +{"train_loss": 0.017913872376084328, "grad_norm": 0.06763339042663574, "global_step": 12506, "epoch": 117, "lr": 8.68776573319823e-05} +{"train_loss": 0.01717814740742434, "grad_norm": 0.10104606300592422, "global_step": 12507, "epoch": 117, "lr": 8.68756558679087e-05, "val_loss": 0.018585098907351494} +{"train_loss": 0.020019255578517914, "grad_norm": 0.1342548131942749, "global_step": 12508, "epoch": 118, "lr": 8.687365427427008e-05} +{"train_loss": 0.016314171254634857, "grad_norm": 0.07912597805261612, "global_step": 12509, "epoch": 118, "lr": 8.687165255107348e-05} +{"train_loss": 0.013990083709359169, "grad_norm": 0.0814826488494873, "global_step": 12510, "epoch": 118, "lr": 8.686965069832591e-05} +{"train_loss": 0.014956280589103699, "grad_norm": 0.07707462459802628, "global_step": 12511, "epoch": 118, "lr": 8.686764871603443e-05} +{"train_loss": 0.01634306088089943, "grad_norm": 0.07830077409744263, "global_step": 12512, "epoch": 118, "lr": 8.686564660420608e-05} +{"train_loss": 0.014654919505119324, "grad_norm": 0.10506133735179901, "global_step": 12513, "epoch": 118, "lr": 8.686364436284787e-05} +{"train_loss": 0.014041140675544739, "grad_norm": 0.11201515793800354, "global_step": 12514, "epoch": 118, "lr": 8.686164199196685e-05} +{"train_loss": 0.018253356218338013, "grad_norm": 0.05814236029982567, "global_step": 12515, "epoch": 118, "lr": 8.685963949157004e-05} +{"train_loss": 0.017097108066082, "grad_norm": 0.07737564295530319, "global_step": 12516, "epoch": 118, "lr": 8.685763686166449e-05} +{"train_loss": 0.01966102421283722, "grad_norm": 0.05886939540505409, "global_step": 12517, "epoch": 118, "lr": 8.685563410225723e-05} +{"train_loss": 0.014563565142452717, "grad_norm": 0.06015494465827942, "global_step": 12518, "epoch": 118, "lr": 8.68536312133553e-05} +{"train_loss": 0.01671708934009075, "grad_norm": 0.05443524569272995, "global_step": 12519, "epoch": 118, "lr": 8.685162819496574e-05} +{"train_loss": 0.0163768008351326, "grad_norm": 0.056731898337602615, "global_step": 12520, "epoch": 118, "lr": 8.684962504709559e-05} +{"train_loss": 0.018922654911875725, "grad_norm": 0.06436797976493835, "global_step": 12521, "epoch": 118, "lr": 8.68476217697519e-05} +{"train_loss": 0.017991267144680023, "grad_norm": 0.153849259018898, "global_step": 12522, "epoch": 118, "lr": 8.684561836294166e-05} +{"train_loss": 0.016313128173351288, "grad_norm": 0.08012416958808899, "global_step": 12523, "epoch": 118, "lr": 8.684361482667196e-05} +{"train_loss": 0.018669329583644867, "grad_norm": 0.0933646410703659, "global_step": 12524, "epoch": 118, "lr": 8.68416111609498e-05} +{"train_loss": 0.01969420537352562, "grad_norm": 0.06164280325174332, "global_step": 12525, "epoch": 118, "lr": 8.683960736578226e-05} +{"train_loss": 0.019041623920202255, "grad_norm": 0.09031342715024948, "global_step": 12526, "epoch": 118, "lr": 8.683760344117635e-05} +{"train_loss": 0.017347557470202446, "grad_norm": 0.05135366693139076, "global_step": 12527, "epoch": 118, "lr": 8.683559938713913e-05} +{"train_loss": 0.017498070374131203, "grad_norm": 0.06391499936580658, "global_step": 12528, "epoch": 118, "lr": 8.683359520367763e-05} +{"train_loss": 0.015918809920549393, "grad_norm": 0.05781083554029465, "global_step": 12529, "epoch": 118, "lr": 8.683159089079889e-05} +{"train_loss": 0.01709683984518051, "grad_norm": 0.15919910371303558, "global_step": 12530, "epoch": 118, "lr": 8.682958644850996e-05} +{"train_loss": 0.0201419647783041, "grad_norm": 0.0656762346625328, "global_step": 12531, "epoch": 118, "lr": 8.682758187681787e-05} +{"train_loss": 0.017120283097028732, "grad_norm": 0.045002613216638565, "global_step": 12532, "epoch": 118, "lr": 8.68255771757297e-05} +{"train_loss": 0.015074393711984158, "grad_norm": 0.051303815096616745, "global_step": 12533, "epoch": 118, "lr": 8.682357234525244e-05} +{"train_loss": 0.015618458390235901, "grad_norm": 0.1062467023730278, "global_step": 12534, "epoch": 118, "lr": 8.682156738539317e-05} +{"train_loss": 0.0166408009827137, "grad_norm": 0.05940523371100426, "global_step": 12535, "epoch": 118, "lr": 8.681956229615894e-05} +{"train_loss": 0.019761288538575172, "grad_norm": 0.061558667570352554, "global_step": 12536, "epoch": 118, "lr": 8.681755707755677e-05} +{"train_loss": 0.017115041613578796, "grad_norm": 0.06549215316772461, "global_step": 12537, "epoch": 118, "lr": 8.681555172959371e-05} +{"train_loss": 0.0202677920460701, "grad_norm": 0.11458677053451538, "global_step": 12538, "epoch": 118, "lr": 8.681354625227682e-05} +{"train_loss": 0.0188334658741951, "grad_norm": 0.13037744164466858, "global_step": 12539, "epoch": 118, "lr": 8.681154064561314e-05} +{"train_loss": 0.016745511442422867, "grad_norm": 0.069371797144413, "global_step": 12540, "epoch": 118, "lr": 8.68095349096097e-05} +{"train_loss": 0.015439266338944435, "grad_norm": 0.12775342166423798, "global_step": 12541, "epoch": 118, "lr": 8.680752904427358e-05} +{"train_loss": 0.018195290118455887, "grad_norm": 0.04836075380444527, "global_step": 12542, "epoch": 118, "lr": 8.68055230496118e-05} +{"train_loss": 0.01735624298453331, "grad_norm": 0.10885310918092728, "global_step": 12543, "epoch": 118, "lr": 8.680351692563143e-05} +{"train_loss": 0.015353478491306305, "grad_norm": 0.05488085374236107, "global_step": 12544, "epoch": 118, "lr": 8.680151067233948e-05} +{"train_loss": 0.01582927443087101, "grad_norm": 0.08746762573719025, "global_step": 12545, "epoch": 118, "lr": 8.679950428974304e-05} +{"train_loss": 0.016643159091472626, "grad_norm": 0.04194120317697525, "global_step": 12546, "epoch": 118, "lr": 8.679749777784916e-05} +{"train_loss": 0.019854972139000893, "grad_norm": 0.060993533581495285, "global_step": 12547, "epoch": 118, "lr": 8.679549113666485e-05} +{"train_loss": 0.018378229811787605, "grad_norm": 0.0688926950097084, "global_step": 12548, "epoch": 118, "lr": 8.679348436619721e-05} +{"train_loss": 0.017592668533325195, "grad_norm": 0.07002755999565125, "global_step": 12549, "epoch": 118, "lr": 8.679147746645326e-05} +{"train_loss": 0.014419017359614372, "grad_norm": 0.09420747309923172, "global_step": 12550, "epoch": 118, "lr": 8.678947043744006e-05} +{"train_loss": 0.017374148592352867, "grad_norm": 0.11094058305025101, "global_step": 12551, "epoch": 118, "lr": 8.678746327916465e-05} +{"train_loss": 0.018051641061902046, "grad_norm": 0.04798980802297592, "global_step": 12552, "epoch": 118, "lr": 8.67854559916341e-05} +{"train_loss": 0.015136678703129292, "grad_norm": 0.06893826276063919, "global_step": 12553, "epoch": 118, "lr": 8.678344857485546e-05} +{"train_loss": 0.016445200890302658, "grad_norm": 0.07694867253303528, "global_step": 12554, "epoch": 118, "lr": 8.678144102883576e-05} +{"train_loss": 0.021417127922177315, "grad_norm": 0.09561434388160706, "global_step": 12555, "epoch": 118, "lr": 8.677943335358208e-05} +{"train_loss": 0.017912674695253372, "grad_norm": 0.11230168491601944, "global_step": 12556, "epoch": 118, "lr": 8.677742554910149e-05} +{"train_loss": 0.01676330901682377, "grad_norm": 0.0770844966173172, "global_step": 12557, "epoch": 118, "lr": 8.6775417615401e-05} +{"train_loss": 0.015497272834181786, "grad_norm": 0.09222359955310822, "global_step": 12558, "epoch": 118, "lr": 8.677340955248767e-05} +{"train_loss": 0.01581151969730854, "grad_norm": 0.042411744594573975, "global_step": 12559, "epoch": 118, "lr": 8.677140136036859e-05} +{"train_loss": 0.02082035318017006, "grad_norm": 0.04529967159032822, "global_step": 12560, "epoch": 118, "lr": 8.67693930390508e-05} +{"train_loss": 0.018754344433546066, "grad_norm": 0.08114472031593323, "global_step": 12561, "epoch": 118, "lr": 8.676738458854134e-05} +{"train_loss": 0.014821222051978111, "grad_norm": 0.07746200263500214, "global_step": 12562, "epoch": 118, "lr": 8.67653760088473e-05} +{"train_loss": 0.01625562086701393, "grad_norm": 0.07990451902151108, "global_step": 12563, "epoch": 118, "lr": 8.676336729997571e-05} +{"train_loss": 0.013436906971037388, "grad_norm": 0.08552323281764984, "global_step": 12564, "epoch": 118, "lr": 8.676135846193363e-05} +{"train_loss": 0.015125672332942486, "grad_norm": 0.09618573635816574, "global_step": 12565, "epoch": 118, "lr": 8.675934949472812e-05} +{"train_loss": 0.016393564641475677, "grad_norm": 0.08467421680688858, "global_step": 12566, "epoch": 118, "lr": 8.675734039836626e-05} +{"train_loss": 0.019512120634317398, "grad_norm": 0.10290994495153427, "global_step": 12567, "epoch": 118, "lr": 8.675533117285505e-05} +{"train_loss": 0.017825238406658173, "grad_norm": 0.12226197123527527, "global_step": 12568, "epoch": 118, "lr": 8.675332181820163e-05} +{"train_loss": 0.013551799580454826, "grad_norm": 0.057006631046533585, "global_step": 12569, "epoch": 118, "lr": 8.675131233441301e-05} +{"train_loss": 0.01728590950369835, "grad_norm": 0.07286674529314041, "global_step": 12570, "epoch": 118, "lr": 8.674930272149626e-05} +{"train_loss": 0.019619755446910858, "grad_norm": 0.09634377807378769, "global_step": 12571, "epoch": 118, "lr": 8.674729297945844e-05} +{"train_loss": 0.017893705517053604, "grad_norm": 0.12391026318073273, "global_step": 12572, "epoch": 118, "lr": 8.674528310830662e-05} +{"train_loss": 0.019544020295143127, "grad_norm": 0.07557791471481323, "global_step": 12573, "epoch": 118, "lr": 8.674327310804786e-05} +{"train_loss": 0.015989240258932114, "grad_norm": 0.0740947499871254, "global_step": 12574, "epoch": 118, "lr": 8.674126297868921e-05} +{"train_loss": 0.017220942303538322, "grad_norm": 0.07434812933206558, "global_step": 12575, "epoch": 118, "lr": 8.673925272023773e-05} +{"train_loss": 0.01791948825120926, "grad_norm": 0.05347244441509247, "global_step": 12576, "epoch": 118, "lr": 8.67372423327005e-05} +{"train_loss": 0.018113158643245697, "grad_norm": 0.07616725564002991, "global_step": 12577, "epoch": 118, "lr": 8.673523181608459e-05} +{"train_loss": 0.019071493297815323, "grad_norm": 0.08140280097723007, "global_step": 12578, "epoch": 118, "lr": 8.673322117039704e-05} +{"train_loss": 0.016266323626041412, "grad_norm": 0.06109248101711273, "global_step": 12579, "epoch": 118, "lr": 8.673121039564492e-05} +{"train_loss": 0.016015563160181046, "grad_norm": 0.07913978397846222, "global_step": 12580, "epoch": 118, "lr": 8.672919949183531e-05} +{"train_loss": 0.017881857231259346, "grad_norm": 0.055229417979717255, "global_step": 12581, "epoch": 118, "lr": 8.672718845897527e-05} +{"train_loss": 0.01786019653081894, "grad_norm": 0.06301136314868927, "global_step": 12582, "epoch": 118, "lr": 8.672517729707186e-05} +{"train_loss": 0.015331042930483818, "grad_norm": 0.04441758245229721, "global_step": 12583, "epoch": 118, "lr": 8.672316600613213e-05} +{"train_loss": 0.01532029826194048, "grad_norm": 0.09388168156147003, "global_step": 12584, "epoch": 118, "lr": 8.672115458616319e-05} +{"train_loss": 0.016435842961072922, "grad_norm": 0.05095687508583069, "global_step": 12585, "epoch": 118, "lr": 8.671914303717207e-05} +{"train_loss": 0.015643293038010597, "grad_norm": 0.09347601979970932, "global_step": 12586, "epoch": 118, "lr": 8.671713135916584e-05} +{"train_loss": 0.02082386612892151, "grad_norm": 0.17005227506160736, "global_step": 12587, "epoch": 118, "lr": 8.67151195521516e-05} +{"train_loss": 0.01799267716705799, "grad_norm": 0.06783562898635864, "global_step": 12588, "epoch": 118, "lr": 8.671310761613637e-05} +{"train_loss": 0.01569461077451706, "grad_norm": 0.07061474770307541, "global_step": 12589, "epoch": 118, "lr": 8.671109555112726e-05} +{"train_loss": 0.0170328076928854, "grad_norm": 0.06123446300625801, "global_step": 12590, "epoch": 118, "lr": 8.670908335713135e-05} +{"train_loss": 0.017575372010469437, "grad_norm": 0.0801604613661766, "global_step": 12591, "epoch": 118, "lr": 8.670707103415566e-05} +{"train_loss": 0.016081128269433975, "grad_norm": 0.09041287004947662, "global_step": 12592, "epoch": 118, "lr": 8.670505858220727e-05} +{"train_loss": 0.015766959637403488, "grad_norm": 0.05278339982032776, "global_step": 12593, "epoch": 118, "lr": 8.67030460012933e-05} +{"train_loss": 0.017698541283607483, "grad_norm": 0.06152604520320892, "global_step": 12594, "epoch": 118, "lr": 8.670103329142078e-05} +{"train_loss": 0.018564509227871895, "grad_norm": 0.08704890310764313, "global_step": 12595, "epoch": 118, "lr": 8.669902045259677e-05} +{"train_loss": 0.01820831187069416, "grad_norm": 0.14154046773910522, "global_step": 12596, "epoch": 118, "lr": 8.669700748482839e-05} +{"train_loss": 0.016641966998577118, "grad_norm": 0.10551292449235916, "global_step": 12597, "epoch": 118, "lr": 8.669499438812267e-05} +{"train_loss": 0.016612498089671135, "grad_norm": 0.07913528382778168, "global_step": 12598, "epoch": 118, "lr": 8.66929811624867e-05} +{"train_loss": 0.015157163143157959, "grad_norm": 0.10197584331035614, "global_step": 12599, "epoch": 118, "lr": 8.669096780792753e-05} +{"train_loss": 0.01927308551967144, "grad_norm": 0.1207224577665329, "global_step": 12600, "epoch": 118, "lr": 8.66889543244523e-05} +{"train_loss": 0.01594126969575882, "grad_norm": 0.10381244868040085, "global_step": 12601, "epoch": 118, "lr": 8.668694071206799e-05} +{"train_loss": 0.015443189069628716, "grad_norm": 0.08033861964941025, "global_step": 12602, "epoch": 118, "lr": 8.668492697078176e-05} +{"train_loss": 0.01785437762737274, "grad_norm": 0.05959387123584747, "global_step": 12603, "epoch": 118, "lr": 8.668291310060064e-05} +{"train_loss": 0.01691826805472374, "grad_norm": 0.06116219982504845, "global_step": 12604, "epoch": 118, "lr": 8.668089910153171e-05} +{"train_loss": 0.01770823262631893, "grad_norm": 0.09564074873924255, "global_step": 12605, "epoch": 118, "lr": 8.667888497358207e-05} +{"train_loss": 0.018567703664302826, "grad_norm": 0.07909069210290909, "global_step": 12606, "epoch": 118, "lr": 8.667687071675878e-05} +{"train_loss": 0.01870962232351303, "grad_norm": 0.08661969006061554, "global_step": 12607, "epoch": 118, "lr": 8.667485633106891e-05} +{"train_loss": 0.01648189313709736, "grad_norm": 0.08924957364797592, "global_step": 12608, "epoch": 118, "lr": 8.667284181651954e-05} +{"train_loss": 0.018519382923841476, "grad_norm": 0.1175059899687767, "global_step": 12609, "epoch": 118, "lr": 8.667082717311776e-05} +{"train_loss": 0.01686740852892399, "grad_norm": 0.04773756489157677, "global_step": 12610, "epoch": 118, "lr": 8.666881240087062e-05} +{"train_loss": 0.01469537615776062, "grad_norm": 0.07802318036556244, "global_step": 12611, "epoch": 118, "lr": 8.666679749978523e-05} +{"train_loss": 0.018446147441864014, "grad_norm": 0.07230190187692642, "global_step": 12612, "epoch": 118, "lr": 8.666478246986867e-05} +{"train_loss": 0.017153716294692374, "grad_norm": 0.07053134590387344, "global_step": 12613, "epoch": 118, "lr": 8.666276731112801e-05, "val_loss": 0.019178612157702446} +{"train_loss": 0.0159246027469635, "grad_norm": 0.06010235473513603, "global_step": 12614, "epoch": 119, "lr": 8.666075202357032e-05} +{"train_loss": 0.014150841161608696, "grad_norm": 0.07841340452432632, "global_step": 12615, "epoch": 119, "lr": 8.665873660720269e-05} +{"train_loss": 0.016828075051307678, "grad_norm": 0.07228425145149231, "global_step": 12616, "epoch": 119, "lr": 8.66567210620322e-05} +{"train_loss": 0.014123395085334778, "grad_norm": 0.05866658315062523, "global_step": 12617, "epoch": 119, "lr": 8.665470538806595e-05} +{"train_loss": 0.01843583583831787, "grad_norm": 0.11422496289014816, "global_step": 12618, "epoch": 119, "lr": 8.665268958531098e-05} +{"train_loss": 0.017252501100301743, "grad_norm": 0.13038697838783264, "global_step": 12619, "epoch": 119, "lr": 8.665067365377442e-05} +{"train_loss": 0.018051788210868835, "grad_norm": 0.0713554173707962, "global_step": 12620, "epoch": 119, "lr": 8.664865759346334e-05} +{"train_loss": 0.01653207093477249, "grad_norm": 0.1581547111272812, "global_step": 12621, "epoch": 119, "lr": 8.66466414043848e-05} +{"train_loss": 0.018510904163122177, "grad_norm": 0.12558937072753906, "global_step": 12622, "epoch": 119, "lr": 8.66446250865459e-05} +{"train_loss": 0.017001280561089516, "grad_norm": 0.0906243845820427, "global_step": 12623, "epoch": 119, "lr": 8.664260863995373e-05} +{"train_loss": 0.015451311133801937, "grad_norm": 0.044909294694662094, "global_step": 12624, "epoch": 119, "lr": 8.664059206461535e-05} +{"train_loss": 0.018553968518972397, "grad_norm": 0.0568317249417305, "global_step": 12625, "epoch": 119, "lr": 8.663857536053789e-05} +{"train_loss": 0.015995733439922333, "grad_norm": 0.05177512764930725, "global_step": 12626, "epoch": 119, "lr": 8.663655852772838e-05} +{"train_loss": 0.017825502902269363, "grad_norm": 0.1176113411784172, "global_step": 12627, "epoch": 119, "lr": 8.663454156619397e-05} +{"train_loss": 0.01830475777387619, "grad_norm": 0.08722760528326035, "global_step": 12628, "epoch": 119, "lr": 8.66325244759417e-05} +{"train_loss": 0.014912848360836506, "grad_norm": 0.06640974432229996, "global_step": 12629, "epoch": 119, "lr": 8.663050725697865e-05} +{"train_loss": 0.01854681596159935, "grad_norm": 0.06842228770256042, "global_step": 12630, "epoch": 119, "lr": 8.662848990931196e-05} +{"train_loss": 0.015440474264323711, "grad_norm": 0.0678756833076477, "global_step": 12631, "epoch": 119, "lr": 8.662647243294866e-05} +{"train_loss": 0.02001778408885002, "grad_norm": 0.05892503261566162, "global_step": 12632, "epoch": 119, "lr": 8.662445482789588e-05} +{"train_loss": 0.0163053497672081, "grad_norm": 0.08403187245130539, "global_step": 12633, "epoch": 119, "lr": 8.66224370941607e-05} +{"train_loss": 0.016186576336622238, "grad_norm": 0.07344288378953934, "global_step": 12634, "epoch": 119, "lr": 8.662041923175017e-05} +{"train_loss": 0.01874002069234848, "grad_norm": 0.07038553804159164, "global_step": 12635, "epoch": 119, "lr": 8.661840124067144e-05} +{"train_loss": 0.015831707045435905, "grad_norm": 0.10590210556983948, "global_step": 12636, "epoch": 119, "lr": 8.661638312093157e-05} +{"train_loss": 0.018300365656614304, "grad_norm": 0.08141721040010452, "global_step": 12637, "epoch": 119, "lr": 8.661436487253764e-05} +{"train_loss": 0.019301768392324448, "grad_norm": 0.12397865206003189, "global_step": 12638, "epoch": 119, "lr": 8.661234649549677e-05} +{"train_loss": 0.0152175547555089, "grad_norm": 0.09722325205802917, "global_step": 12639, "epoch": 119, "lr": 8.661032798981604e-05} +{"train_loss": 0.017655009403824806, "grad_norm": 0.07494413107633591, "global_step": 12640, "epoch": 119, "lr": 8.660830935550253e-05} +{"train_loss": 0.01683277077972889, "grad_norm": 0.06612218171358109, "global_step": 12641, "epoch": 119, "lr": 8.660629059256334e-05} +{"train_loss": 0.016614245250821114, "grad_norm": 0.1321219652891159, "global_step": 12642, "epoch": 119, "lr": 8.660427170100555e-05} +{"train_loss": 0.016592787578701973, "grad_norm": 0.07692576944828033, "global_step": 12643, "epoch": 119, "lr": 8.660225268083629e-05} +{"train_loss": 0.017879821360111237, "grad_norm": 0.07406303286552429, "global_step": 12644, "epoch": 119, "lr": 8.660023353206263e-05} +{"train_loss": 0.015297573059797287, "grad_norm": 0.07635867595672607, "global_step": 12645, "epoch": 119, "lr": 8.659821425469165e-05} +{"train_loss": 0.019575946033000946, "grad_norm": 0.08373802900314331, "global_step": 12646, "epoch": 119, "lr": 8.659619484873049e-05} +{"train_loss": 0.018503818660974503, "grad_norm": 0.13974153995513916, "global_step": 12647, "epoch": 119, "lr": 8.659417531418619e-05} +{"train_loss": 0.017163056880235672, "grad_norm": 0.07018907368183136, "global_step": 12648, "epoch": 119, "lr": 8.659215565106589e-05} +{"train_loss": 0.016243139281868935, "grad_norm": 0.10374105721712112, "global_step": 12649, "epoch": 119, "lr": 8.659013585937666e-05} +{"train_loss": 0.01697554998099804, "grad_norm": 0.08456332981586456, "global_step": 12650, "epoch": 119, "lr": 8.658811593912561e-05} +{"train_loss": 0.015240387991070747, "grad_norm": 0.07967586070299149, "global_step": 12651, "epoch": 119, "lr": 8.658609589031983e-05} +{"train_loss": 0.018661018460989, "grad_norm": 0.08310355246067047, "global_step": 12652, "epoch": 119, "lr": 8.658407571296641e-05} +{"train_loss": 0.01718239113688469, "grad_norm": 0.10887075215578079, "global_step": 12653, "epoch": 119, "lr": 8.658205540707248e-05} +{"train_loss": 0.018464867025613785, "grad_norm": 0.11124590039253235, "global_step": 12654, "epoch": 119, "lr": 8.658003497264509e-05} +{"train_loss": 0.016853751614689827, "grad_norm": 0.05663888528943062, "global_step": 12655, "epoch": 119, "lr": 8.657801440969137e-05} +{"train_loss": 0.016493264585733414, "grad_norm": 0.18053285777568817, "global_step": 12656, "epoch": 119, "lr": 8.657599371821843e-05} +{"train_loss": 0.015792712569236755, "grad_norm": 0.06289152055978775, "global_step": 12657, "epoch": 119, "lr": 8.657397289823334e-05} +{"train_loss": 0.019437888637185097, "grad_norm": 0.1238284707069397, "global_step": 12658, "epoch": 119, "lr": 8.657195194974323e-05} +{"train_loss": 0.0158699844032526, "grad_norm": 0.07878509908914566, "global_step": 12659, "epoch": 119, "lr": 8.656993087275518e-05} +{"train_loss": 0.018996357917785645, "grad_norm": 0.07063248008489609, "global_step": 12660, "epoch": 119, "lr": 8.65679096672763e-05} +{"train_loss": 0.016216717660427094, "grad_norm": 0.09244101494550705, "global_step": 12661, "epoch": 119, "lr": 8.656588833331367e-05} +{"train_loss": 0.014100223779678345, "grad_norm": 0.17112162709236145, "global_step": 12662, "epoch": 119, "lr": 8.656386687087443e-05} +{"train_loss": 0.016879156231880188, "grad_norm": 0.08890075236558914, "global_step": 12663, "epoch": 119, "lr": 8.656184527996566e-05} +{"train_loss": 0.018240686506032944, "grad_norm": 0.062245987355709076, "global_step": 12664, "epoch": 119, "lr": 8.655982356059447e-05} +{"train_loss": 0.01954915188252926, "grad_norm": 0.08848633617162704, "global_step": 12665, "epoch": 119, "lr": 8.655780171276795e-05} +{"train_loss": 0.017110999673604965, "grad_norm": 0.06268182396888733, "global_step": 12666, "epoch": 119, "lr": 8.655577973649321e-05} +{"train_loss": 0.014734398573637009, "grad_norm": 0.06108059361577034, "global_step": 12667, "epoch": 119, "lr": 8.655375763177737e-05} +{"train_loss": 0.01667144149541855, "grad_norm": 0.05787636712193489, "global_step": 12668, "epoch": 119, "lr": 8.655173539862751e-05} +{"train_loss": 0.015669258311390877, "grad_norm": 0.10762156546115875, "global_step": 12669, "epoch": 119, "lr": 8.654971303705075e-05} +{"train_loss": 0.02026265487074852, "grad_norm": 0.1000654473900795, "global_step": 12670, "epoch": 119, "lr": 8.65476905470542e-05} +{"train_loss": 0.018955830484628677, "grad_norm": 0.11499211192131042, "global_step": 12671, "epoch": 119, "lr": 8.654566792864496e-05} +{"train_loss": 0.017074860632419586, "grad_norm": 0.09335137903690338, "global_step": 12672, "epoch": 119, "lr": 8.654364518183014e-05} +{"train_loss": 0.01639533042907715, "grad_norm": 0.12668702006340027, "global_step": 12673, "epoch": 119, "lr": 8.654162230661682e-05} +{"train_loss": 0.019100766628980637, "grad_norm": 0.13134616613388062, "global_step": 12674, "epoch": 119, "lr": 8.653959930301216e-05} +{"train_loss": 0.015156292356550694, "grad_norm": 0.10924352705478668, "global_step": 12675, "epoch": 119, "lr": 8.653757617102322e-05} +{"train_loss": 0.019679460674524307, "grad_norm": 0.11228054761886597, "global_step": 12676, "epoch": 119, "lr": 8.653555291065713e-05} +{"train_loss": 0.01664786972105503, "grad_norm": 0.07814449071884155, "global_step": 12677, "epoch": 119, "lr": 8.6533529521921e-05} +{"train_loss": 0.019471121951937675, "grad_norm": 0.09844639897346497, "global_step": 12678, "epoch": 119, "lr": 8.653150600482193e-05} +{"train_loss": 0.015060149133205414, "grad_norm": 0.08525457978248596, "global_step": 12679, "epoch": 119, "lr": 8.652948235936705e-05} +{"train_loss": 0.018523328006267548, "grad_norm": 0.05692392587661743, "global_step": 12680, "epoch": 119, "lr": 8.652745858556344e-05} +{"train_loss": 0.01667226105928421, "grad_norm": 0.05261143669486046, "global_step": 12681, "epoch": 119, "lr": 8.652543468341822e-05} +{"train_loss": 0.019166897982358932, "grad_norm": 0.07273941487073898, "global_step": 12682, "epoch": 119, "lr": 8.652341065293852e-05} +{"train_loss": 0.016443410888314247, "grad_norm": 0.10465758293867111, "global_step": 12683, "epoch": 119, "lr": 8.652138649413144e-05} +{"train_loss": 0.01691797748208046, "grad_norm": 0.06867655366659164, "global_step": 12684, "epoch": 119, "lr": 8.651936220700408e-05} +{"train_loss": 0.01648583635687828, "grad_norm": 0.055940426886081696, "global_step": 12685, "epoch": 119, "lr": 8.651733779156357e-05} +{"train_loss": 0.01830505207180977, "grad_norm": 0.09523621201515198, "global_step": 12686, "epoch": 119, "lr": 8.6515313247817e-05} +{"train_loss": 0.018216121941804886, "grad_norm": 0.13902461528778076, "global_step": 12687, "epoch": 119, "lr": 8.651328857577152e-05} +{"train_loss": 0.01655217632651329, "grad_norm": 0.07203846424818039, "global_step": 12688, "epoch": 119, "lr": 8.651126377543422e-05} +{"train_loss": 0.016481615602970123, "grad_norm": 0.10844920575618744, "global_step": 12689, "epoch": 119, "lr": 8.650923884681222e-05} +{"train_loss": 0.01639316789805889, "grad_norm": 0.10290679335594177, "global_step": 12690, "epoch": 119, "lr": 8.650721378991263e-05} +{"train_loss": 0.017078423872590065, "grad_norm": 0.05541741102933884, "global_step": 12691, "epoch": 119, "lr": 8.650518860474257e-05} +{"train_loss": 0.017275605350732803, "grad_norm": 0.09327734261751175, "global_step": 12692, "epoch": 119, "lr": 8.650316329130914e-05} +{"train_loss": 0.017046954482793808, "grad_norm": 0.05869027599692345, "global_step": 12693, "epoch": 119, "lr": 8.650113784961949e-05} +{"train_loss": 0.0168134868144989, "grad_norm": 0.09627331793308258, "global_step": 12694, "epoch": 119, "lr": 8.64991122796807e-05} +{"train_loss": 0.01768871396780014, "grad_norm": 0.13041828572750092, "global_step": 12695, "epoch": 119, "lr": 8.64970865814999e-05} +{"train_loss": 0.01856349967420101, "grad_norm": 0.06904091686010361, "global_step": 12696, "epoch": 119, "lr": 8.649506075508421e-05} +{"train_loss": 0.015027198940515518, "grad_norm": 0.047234829515218735, "global_step": 12697, "epoch": 119, "lr": 8.649303480044075e-05} +{"train_loss": 0.016905196011066437, "grad_norm": 0.057787004858255386, "global_step": 12698, "epoch": 119, "lr": 8.649100871757664e-05} +{"train_loss": 0.0188608318567276, "grad_norm": 0.06849753111600876, "global_step": 12699, "epoch": 119, "lr": 8.6488982506499e-05} +{"train_loss": 0.016566399484872818, "grad_norm": 0.12280144542455673, "global_step": 12700, "epoch": 119, "lr": 8.648695616721494e-05} +{"train_loss": 0.016643237322568893, "grad_norm": 0.06178486719727516, "global_step": 12701, "epoch": 119, "lr": 8.648492969973157e-05} +{"train_loss": 0.017734814435243607, "grad_norm": 0.05223008245229721, "global_step": 12702, "epoch": 119, "lr": 8.648290310405603e-05} +{"train_loss": 0.015264282934367657, "grad_norm": 0.05900151655077934, "global_step": 12703, "epoch": 119, "lr": 8.648087638019544e-05} +{"train_loss": 0.017062652856111526, "grad_norm": 0.09135618060827255, "global_step": 12704, "epoch": 119, "lr": 8.647884952815692e-05} +{"train_loss": 0.020026184618473053, "grad_norm": 0.0932481586933136, "global_step": 12705, "epoch": 119, "lr": 8.647682254794758e-05} +{"train_loss": 0.017597343772649765, "grad_norm": 0.11575060337781906, "global_step": 12706, "epoch": 119, "lr": 8.647479543957456e-05} +{"train_loss": 0.016000639647245407, "grad_norm": 0.0952938124537468, "global_step": 12707, "epoch": 119, "lr": 8.647276820304495e-05} +{"train_loss": 0.019742067903280258, "grad_norm": 0.06611461937427521, "global_step": 12708, "epoch": 119, "lr": 8.64707408383659e-05} +{"train_loss": 0.015717940405011177, "grad_norm": 0.06371969729661942, "global_step": 12709, "epoch": 119, "lr": 8.646871334554454e-05} +{"train_loss": 0.016224540770053864, "grad_norm": 0.07130936533212662, "global_step": 12710, "epoch": 119, "lr": 8.646668572458799e-05} +{"train_loss": 0.0172603540122509, "grad_norm": 0.10815230011940002, "global_step": 12711, "epoch": 119, "lr": 8.646465797550334e-05} +{"train_loss": 0.017359349876642227, "grad_norm": 0.07093491405248642, "global_step": 12712, "epoch": 119, "lr": 8.646263009829776e-05} +{"train_loss": 0.02011743187904358, "grad_norm": 0.15305404365062714, "global_step": 12713, "epoch": 119, "lr": 8.646060209297833e-05} +{"train_loss": 0.015365447849035263, "grad_norm": 0.15551653504371643, "global_step": 12714, "epoch": 119, "lr": 8.645857395955221e-05} +{"train_loss": 0.018664155155420303, "grad_norm": 0.05539051815867424, "global_step": 12715, "epoch": 119, "lr": 8.645654569802652e-05} +{"train_loss": 0.015403518453240395, "grad_norm": 0.05670291557908058, "global_step": 12716, "epoch": 119, "lr": 8.645451730840839e-05} +{"train_loss": 0.01676793023943901, "grad_norm": 0.12044010311365128, "global_step": 12717, "epoch": 119, "lr": 8.645248879070494e-05} +{"train_loss": 0.01640428602695465, "grad_norm": 0.06693962216377258, "global_step": 12718, "epoch": 119, "lr": 8.645046014492329e-05} +{"train_loss": 0.017141746304366948, "grad_norm": 0.08693954348564148, "global_step": 12719, "epoch": 119, "lr": 8.644843137107059e-05, "val_loss": 0.01910575106739998} +{"train_loss": 0.018770873546600342, "grad_norm": 0.05862022191286087, "global_step": 12720, "epoch": 120, "lr": 8.644640246915393e-05} +{"train_loss": 0.017304135486483574, "grad_norm": 0.11805751919746399, "global_step": 12721, "epoch": 120, "lr": 8.644437343918047e-05} +{"train_loss": 0.0179440937936306, "grad_norm": 0.15234775841236115, "global_step": 12722, "epoch": 120, "lr": 8.644234428115734e-05} +{"train_loss": 0.020044244825839996, "grad_norm": 0.06324250251054764, "global_step": 12723, "epoch": 120, "lr": 8.644031499509165e-05} +{"train_loss": 0.017191870138049126, "grad_norm": 0.09043629467487335, "global_step": 12724, "epoch": 120, "lr": 8.643828558099054e-05} +{"train_loss": 0.018010016530752182, "grad_norm": 0.077341228723526, "global_step": 12725, "epoch": 120, "lr": 8.643625603886116e-05} +{"train_loss": 0.01678358018398285, "grad_norm": 0.11836633831262589, "global_step": 12726, "epoch": 120, "lr": 8.64342263687106e-05} +{"train_loss": 0.015490598045289516, "grad_norm": 0.0663256123661995, "global_step": 12727, "epoch": 120, "lr": 8.643219657054602e-05} +{"train_loss": 0.01604044809937477, "grad_norm": 0.05864887312054634, "global_step": 12728, "epoch": 120, "lr": 8.643016664437454e-05} +{"train_loss": 0.015595467761158943, "grad_norm": 0.06304475665092468, "global_step": 12729, "epoch": 120, "lr": 8.64281365902033e-05} +{"train_loss": 0.015009457245469093, "grad_norm": 0.08262375742197037, "global_step": 12730, "epoch": 120, "lr": 8.642610640803943e-05} +{"train_loss": 0.013889927417039871, "grad_norm": 0.07441285252571106, "global_step": 12731, "epoch": 120, "lr": 8.642407609789007e-05} +{"train_loss": 0.01775147020816803, "grad_norm": 0.055591896176338196, "global_step": 12732, "epoch": 120, "lr": 8.642204565976236e-05} +{"train_loss": 0.017635488882660866, "grad_norm": 0.07601560652256012, "global_step": 12733, "epoch": 120, "lr": 8.64200150936634e-05} +{"train_loss": 0.01837686076760292, "grad_norm": 0.062028538435697556, "global_step": 12734, "epoch": 120, "lr": 8.641798439960034e-05} +{"train_loss": 0.017361285164952278, "grad_norm": 0.06562279909849167, "global_step": 12735, "epoch": 120, "lr": 8.641595357758035e-05} +{"train_loss": 0.017798807471990585, "grad_norm": 0.13193432986736298, "global_step": 12736, "epoch": 120, "lr": 8.64139226276105e-05} +{"train_loss": 0.018584616482257843, "grad_norm": 0.10652679949998856, "global_step": 12737, "epoch": 120, "lr": 8.641189154969798e-05} +{"train_loss": 0.01576555147767067, "grad_norm": 0.056936390697956085, "global_step": 12738, "epoch": 120, "lr": 8.64098603438499e-05} +{"train_loss": 0.019378934055566788, "grad_norm": 0.09046373516321182, "global_step": 12739, "epoch": 120, "lr": 8.640782901007341e-05} +{"train_loss": 0.018226951360702515, "grad_norm": 0.08602147549390793, "global_step": 12740, "epoch": 120, "lr": 8.640579754837566e-05} +{"train_loss": 0.015970265492796898, "grad_norm": 0.05431517958641052, "global_step": 12741, "epoch": 120, "lr": 8.640376595876375e-05} +{"train_loss": 0.017116503790020943, "grad_norm": 0.06434817612171173, "global_step": 12742, "epoch": 120, "lr": 8.640173424124485e-05} +{"train_loss": 0.016672801226377487, "grad_norm": 0.08141583949327469, "global_step": 12743, "epoch": 120, "lr": 8.639970239582607e-05} +{"train_loss": 0.01587039791047573, "grad_norm": 0.06653741002082825, "global_step": 12744, "epoch": 120, "lr": 8.639767042251458e-05} +{"train_loss": 0.015824366360902786, "grad_norm": 0.08436515182256699, "global_step": 12745, "epoch": 120, "lr": 8.639563832131749e-05} +{"train_loss": 0.01714572124183178, "grad_norm": 0.14893849194049835, "global_step": 12746, "epoch": 120, "lr": 8.639360609224196e-05} +{"train_loss": 0.01573369652032852, "grad_norm": 0.04945800453424454, "global_step": 12747, "epoch": 120, "lr": 8.639157373529513e-05} +{"train_loss": 0.01689714565873146, "grad_norm": 0.09851142764091492, "global_step": 12748, "epoch": 120, "lr": 8.638954125048415e-05} +{"train_loss": 0.01588164083659649, "grad_norm": 0.13561475276947021, "global_step": 12749, "epoch": 120, "lr": 8.638750863781612e-05} +{"train_loss": 0.015384543687105179, "grad_norm": 0.07355799525976181, "global_step": 12750, "epoch": 120, "lr": 8.638547589729823e-05} +{"train_loss": 0.017915215343236923, "grad_norm": 0.07960306853055954, "global_step": 12751, "epoch": 120, "lr": 8.63834430289376e-05} +{"train_loss": 0.016068432480096817, "grad_norm": 0.0734013095498085, "global_step": 12752, "epoch": 120, "lr": 8.638141003274136e-05} +{"train_loss": 0.017412938177585602, "grad_norm": 0.12116286158561707, "global_step": 12753, "epoch": 120, "lr": 8.637937690871667e-05} +{"train_loss": 0.01626671478152275, "grad_norm": 0.07139591872692108, "global_step": 12754, "epoch": 120, "lr": 8.637734365687068e-05} +{"train_loss": 0.015738390386104584, "grad_norm": 0.05768255889415741, "global_step": 12755, "epoch": 120, "lr": 8.637531027721053e-05} +{"train_loss": 0.016301020979881287, "grad_norm": 0.06845880299806595, "global_step": 12756, "epoch": 120, "lr": 8.637327676974334e-05} +{"train_loss": 0.019141612574458122, "grad_norm": 0.05493147671222687, "global_step": 12757, "epoch": 120, "lr": 8.637124313447629e-05} +{"train_loss": 0.01915094628930092, "grad_norm": 0.04535098373889923, "global_step": 12758, "epoch": 120, "lr": 8.63692093714165e-05} +{"train_loss": 0.015440467745065689, "grad_norm": 0.06333606690168381, "global_step": 12759, "epoch": 120, "lr": 8.636717548057112e-05} +{"train_loss": 0.019317977130413055, "grad_norm": 0.07175060361623764, "global_step": 12760, "epoch": 120, "lr": 8.63651414619473e-05} +{"train_loss": 0.018817666918039322, "grad_norm": 0.1095399484038353, "global_step": 12761, "epoch": 120, "lr": 8.63631073155522e-05} +{"train_loss": 0.015611573122441769, "grad_norm": 0.05828683078289032, "global_step": 12762, "epoch": 120, "lr": 8.636107304139294e-05} +{"train_loss": 0.019696786999702454, "grad_norm": 0.10811614990234375, "global_step": 12763, "epoch": 120, "lr": 8.63590386394767e-05} +{"train_loss": 0.01579314097762108, "grad_norm": 0.1328526735305786, "global_step": 12764, "epoch": 120, "lr": 8.63570041098106e-05} +{"train_loss": 0.017774401232600212, "grad_norm": 0.05677896738052368, "global_step": 12765, "epoch": 120, "lr": 8.63549694524018e-05} +{"train_loss": 0.01699364185333252, "grad_norm": 0.17425240576267242, "global_step": 12766, "epoch": 120, "lr": 8.635293466725745e-05} +{"train_loss": 0.017032459378242493, "grad_norm": 0.09283260256052017, "global_step": 12767, "epoch": 120, "lr": 8.635089975438471e-05} +{"train_loss": 0.01870650053024292, "grad_norm": 0.0722963809967041, "global_step": 12768, "epoch": 120, "lr": 8.634886471379069e-05} +{"train_loss": 0.016991494223475456, "grad_norm": 0.09761916100978851, "global_step": 12769, "epoch": 120, "lr": 8.634682954548259e-05} +{"train_loss": 0.017552096396684647, "grad_norm": 0.06959787756204605, "global_step": 12770, "epoch": 120, "lr": 8.634479424946752e-05} +{"train_loss": 0.01817578822374344, "grad_norm": 0.07721743732690811, "global_step": 12771, "epoch": 120, "lr": 8.634275882575266e-05} +{"train_loss": 0.015734564512968063, "grad_norm": 0.05221609026193619, "global_step": 12772, "epoch": 120, "lr": 8.634072327434515e-05} +{"train_loss": 0.018708331510424614, "grad_norm": 0.07458823919296265, "global_step": 12773, "epoch": 120, "lr": 8.633868759525214e-05} +{"train_loss": 0.016660630702972412, "grad_norm": 0.11506268382072449, "global_step": 12774, "epoch": 120, "lr": 8.633665178848078e-05} +{"train_loss": 0.018457230180501938, "grad_norm": 0.08138525485992432, "global_step": 12775, "epoch": 120, "lr": 8.633461585403823e-05} +{"train_loss": 0.02046283707022667, "grad_norm": 0.06598585844039917, "global_step": 12776, "epoch": 120, "lr": 8.633257979193164e-05} +{"train_loss": 0.014806609600782394, "grad_norm": 0.07774171978235245, "global_step": 12777, "epoch": 120, "lr": 8.633054360216818e-05} +{"train_loss": 0.01822345145046711, "grad_norm": 0.061615101993083954, "global_step": 12778, "epoch": 120, "lr": 8.632850728475498e-05} +{"train_loss": 0.01693195104598999, "grad_norm": 0.05717229098081589, "global_step": 12779, "epoch": 120, "lr": 8.63264708396992e-05} +{"train_loss": 0.015666203573346138, "grad_norm": 0.05822769179940224, "global_step": 12780, "epoch": 120, "lr": 8.6324434267008e-05} +{"train_loss": 0.018361033871769905, "grad_norm": 0.06716803461313248, "global_step": 12781, "epoch": 120, "lr": 8.632239756668855e-05} +{"train_loss": 0.014847257174551487, "grad_norm": 0.056149568408727646, "global_step": 12782, "epoch": 120, "lr": 8.632036073874797e-05} +{"train_loss": 0.015568113885819912, "grad_norm": 0.07539542019367218, "global_step": 12783, "epoch": 120, "lr": 8.631832378319345e-05} +{"train_loss": 0.017613429576158524, "grad_norm": 0.09320774674415588, "global_step": 12784, "epoch": 120, "lr": 8.631628670003213e-05} +{"train_loss": 0.01628497615456581, "grad_norm": 0.15043243765830994, "global_step": 12785, "epoch": 120, "lr": 8.631424948927118e-05} +{"train_loss": 0.02228870615363121, "grad_norm": 0.07695572078227997, "global_step": 12786, "epoch": 120, "lr": 8.631221215091773e-05} +{"train_loss": 0.01825893297791481, "grad_norm": 0.06456220149993896, "global_step": 12787, "epoch": 120, "lr": 8.631017468497897e-05} +{"train_loss": 0.01901748776435852, "grad_norm": 0.07805830985307693, "global_step": 12788, "epoch": 120, "lr": 8.630813709146206e-05} +{"train_loss": 0.016437958925962448, "grad_norm": 0.050165604799985886, "global_step": 12789, "epoch": 120, "lr": 8.630609937037412e-05} +{"train_loss": 0.016866741701960564, "grad_norm": 0.06289330869913101, "global_step": 12790, "epoch": 120, "lr": 8.630406152172236e-05} +{"train_loss": 0.0184708833694458, "grad_norm": 0.06680452078580856, "global_step": 12791, "epoch": 120, "lr": 8.630202354551392e-05} +{"train_loss": 0.014736898243427277, "grad_norm": 0.0846339538693428, "global_step": 12792, "epoch": 120, "lr": 8.629998544175593e-05} +{"train_loss": 0.017202455550432205, "grad_norm": 0.06304661929607391, "global_step": 12793, "epoch": 120, "lr": 8.629794721045558e-05} +{"train_loss": 0.017003782093524933, "grad_norm": 0.05377545952796936, "global_step": 12794, "epoch": 120, "lr": 8.629590885162005e-05} +{"train_loss": 0.01517568714916706, "grad_norm": 0.08393028378486633, "global_step": 12795, "epoch": 120, "lr": 8.629387036525647e-05} +{"train_loss": 0.01731511950492859, "grad_norm": 0.17595013976097107, "global_step": 12796, "epoch": 120, "lr": 8.6291831751372e-05} +{"train_loss": 0.01607528328895569, "grad_norm": 0.047879692167043686, "global_step": 12797, "epoch": 120, "lr": 8.628979300997384e-05} +{"train_loss": 0.01583855226635933, "grad_norm": 0.05776003375649452, "global_step": 12798, "epoch": 120, "lr": 8.628775414106913e-05} +{"train_loss": 0.015999244526028633, "grad_norm": 0.08106520026922226, "global_step": 12799, "epoch": 120, "lr": 8.628571514466501e-05} +{"train_loss": 0.016266118735074997, "grad_norm": 0.044726964086294174, "global_step": 12800, "epoch": 120, "lr": 8.628367602076869e-05} +{"train_loss": 0.01604279689490795, "grad_norm": 0.13420230150222778, "global_step": 12801, "epoch": 120, "lr": 8.62816367693873e-05} +{"train_loss": 0.017157413065433502, "grad_norm": 0.06670761108398438, "global_step": 12802, "epoch": 120, "lr": 8.627959739052803e-05} +{"train_loss": 0.019114119932055473, "grad_norm": 0.06309137493371964, "global_step": 12803, "epoch": 120, "lr": 8.627755788419801e-05} +{"train_loss": 0.017770584672689438, "grad_norm": 0.08518461138010025, "global_step": 12804, "epoch": 120, "lr": 8.627551825040445e-05} +{"train_loss": 0.01736585423350334, "grad_norm": 0.09648501873016357, "global_step": 12805, "epoch": 120, "lr": 8.627347848915449e-05} +{"train_loss": 0.015471838414669037, "grad_norm": 0.11275143176317215, "global_step": 12806, "epoch": 120, "lr": 8.627143860045527e-05} +{"train_loss": 0.016773952171206474, "grad_norm": 0.08228664845228195, "global_step": 12807, "epoch": 120, "lr": 8.626939858431402e-05} +{"train_loss": 0.014729602262377739, "grad_norm": 0.07277480512857437, "global_step": 12808, "epoch": 120, "lr": 8.626735844073787e-05} +{"train_loss": 0.01594747230410576, "grad_norm": 0.059375837445259094, "global_step": 12809, "epoch": 120, "lr": 8.6265318169734e-05} +{"train_loss": 0.017645178362727165, "grad_norm": 0.061253637075424194, "global_step": 12810, "epoch": 120, "lr": 8.626327777130959e-05} +{"train_loss": 0.015673290938138962, "grad_norm": 0.06193416193127632, "global_step": 12811, "epoch": 120, "lr": 8.626123724547176e-05} +{"train_loss": 0.013755418360233307, "grad_norm": 0.0576012097299099, "global_step": 12812, "epoch": 120, "lr": 8.625919659222771e-05} +{"train_loss": 0.019472889602184296, "grad_norm": 0.14322878420352936, "global_step": 12813, "epoch": 120, "lr": 8.625715581158462e-05} +{"train_loss": 0.016844533383846283, "grad_norm": 0.07266341149806976, "global_step": 12814, "epoch": 120, "lr": 8.625511490354967e-05} +{"train_loss": 0.018378518521785736, "grad_norm": 0.07761815190315247, "global_step": 12815, "epoch": 120, "lr": 8.625307386813e-05} +{"train_loss": 0.01777695119380951, "grad_norm": 0.10219237208366394, "global_step": 12816, "epoch": 120, "lr": 8.625103270533278e-05} +{"train_loss": 0.016935037449002266, "grad_norm": 0.06804272532463074, "global_step": 12817, "epoch": 120, "lr": 8.624899141516522e-05} +{"train_loss": 0.019522925838828087, "grad_norm": 0.12834608554840088, "global_step": 12818, "epoch": 120, "lr": 8.624694999763446e-05} +{"train_loss": 0.01680568978190422, "grad_norm": 0.0683368518948555, "global_step": 12819, "epoch": 120, "lr": 8.624490845274769e-05} +{"train_loss": 0.017559368163347244, "grad_norm": 0.04915416240692139, "global_step": 12820, "epoch": 120, "lr": 8.624286678051205e-05} +{"train_loss": 0.018480360507965088, "grad_norm": 0.05476808175444603, "global_step": 12821, "epoch": 120, "lr": 8.624082498093476e-05} +{"train_loss": 0.01690555363893509, "grad_norm": 0.0848129615187645, "global_step": 12822, "epoch": 120, "lr": 8.623878305402297e-05} +{"train_loss": 0.01545683667063713, "grad_norm": 0.0674847960472107, "global_step": 12823, "epoch": 120, "lr": 8.623674099978385e-05} +{"train_loss": 0.018931496888399124, "grad_norm": 0.07200758159160614, "global_step": 12824, "epoch": 120, "lr": 8.623469881822457e-05} +{"train_loss": 0.017103881417018064, "grad_norm": 0.08865097910165787, "global_step": 12825, "epoch": 120, "lr": 8.623265650935234e-05, "train_batch_mse_error": 0.01537026185542345, "val_loss": 0.019131936132907867} +{"train_loss": 0.01508563756942749, "grad_norm": 0.0732179656624794, "global_step": 12826, "epoch": 121, "lr": 8.62306140731743e-05} +{"train_loss": 0.01662885583937168, "grad_norm": 0.05781140550971031, "global_step": 12827, "epoch": 121, "lr": 8.622857150969763e-05} +{"train_loss": 0.01968139410018921, "grad_norm": 0.09509273618459702, "global_step": 12828, "epoch": 121, "lr": 8.622652881892953e-05} +{"train_loss": 0.01569976657629013, "grad_norm": 0.057219248265028, "global_step": 12829, "epoch": 121, "lr": 8.622448600087716e-05} +{"train_loss": 0.01949070766568184, "grad_norm": 0.15778617560863495, "global_step": 12830, "epoch": 121, "lr": 8.62224430555477e-05} +{"train_loss": 0.014790793880820274, "grad_norm": 0.05142158642411232, "global_step": 12831, "epoch": 121, "lr": 8.62203999829483e-05} +{"train_loss": 0.015666726976633072, "grad_norm": 0.08068925887346268, "global_step": 12832, "epoch": 121, "lr": 8.621835678308621e-05} +{"train_loss": 0.018406357616186142, "grad_norm": 0.13182556629180908, "global_step": 12833, "epoch": 121, "lr": 8.621631345596854e-05} +{"train_loss": 0.016586890444159508, "grad_norm": 0.07284419983625412, "global_step": 12834, "epoch": 121, "lr": 8.621427000160251e-05} +{"train_loss": 0.01723303459584713, "grad_norm": 0.0699658989906311, "global_step": 12835, "epoch": 121, "lr": 8.621222641999527e-05} +{"train_loss": 0.017308063805103302, "grad_norm": 0.10393799096345901, "global_step": 12836, "epoch": 121, "lr": 8.621018271115402e-05} +{"train_loss": 0.015294982120394707, "grad_norm": 0.060339074581861496, "global_step": 12837, "epoch": 121, "lr": 8.620813887508592e-05} +{"train_loss": 0.02036898583173752, "grad_norm": 0.05968126282095909, "global_step": 12838, "epoch": 121, "lr": 8.620609491179818e-05} +{"train_loss": 0.01572907716035843, "grad_norm": 0.0656690001487732, "global_step": 12839, "epoch": 121, "lr": 8.620405082129798e-05} +{"train_loss": 0.01871311478316784, "grad_norm": 0.06673692166805267, "global_step": 12840, "epoch": 121, "lr": 8.620200660359248e-05} +{"train_loss": 0.015983091667294502, "grad_norm": 0.07342309504747391, "global_step": 12841, "epoch": 121, "lr": 8.619996225868887e-05} +{"train_loss": 0.017134513705968857, "grad_norm": 0.05353900417685509, "global_step": 12842, "epoch": 121, "lr": 8.619791778659434e-05} +{"train_loss": 0.017209891229867935, "grad_norm": 0.07817597687244415, "global_step": 12843, "epoch": 121, "lr": 8.619587318731606e-05} +{"train_loss": 0.015941504389047623, "grad_norm": 0.08094872534275055, "global_step": 12844, "epoch": 121, "lr": 8.619382846086124e-05} +{"train_loss": 0.014909096993505955, "grad_norm": 0.08624652028083801, "global_step": 12845, "epoch": 121, "lr": 8.619178360723703e-05} +{"train_loss": 0.01936202123761177, "grad_norm": 0.1410229355096817, "global_step": 12846, "epoch": 121, "lr": 8.618973862645064e-05} +{"train_loss": 0.019838914275169373, "grad_norm": 0.0802178904414177, "global_step": 12847, "epoch": 121, "lr": 8.618769351850924e-05} +{"train_loss": 0.018853887915611267, "grad_norm": 0.1364634931087494, "global_step": 12848, "epoch": 121, "lr": 8.618564828342002e-05} +{"train_loss": 0.01847880892455578, "grad_norm": 0.11257225275039673, "global_step": 12849, "epoch": 121, "lr": 8.618360292119019e-05} +{"train_loss": 0.01495030801743269, "grad_norm": 0.08026587963104248, "global_step": 12850, "epoch": 121, "lr": 8.61815574318269e-05} +{"train_loss": 0.01882803440093994, "grad_norm": 0.08048702776432037, "global_step": 12851, "epoch": 121, "lr": 8.617951181533735e-05} +{"train_loss": 0.015053150244057178, "grad_norm": 0.08132229745388031, "global_step": 12852, "epoch": 121, "lr": 8.617746607172874e-05} +{"train_loss": 0.016547147184610367, "grad_norm": 0.06072576344013214, "global_step": 12853, "epoch": 121, "lr": 8.617542020100823e-05} +{"train_loss": 0.017055831849575043, "grad_norm": 0.1138375774025917, "global_step": 12854, "epoch": 121, "lr": 8.617337420318304e-05} +{"train_loss": 0.013900712132453918, "grad_norm": 0.054429780691862106, "global_step": 12855, "epoch": 121, "lr": 8.617132807826034e-05} +{"train_loss": 0.019072871655225754, "grad_norm": 0.05014542490243912, "global_step": 12856, "epoch": 121, "lr": 8.616928182624732e-05} +{"train_loss": 0.015861067920923233, "grad_norm": 0.09908473491668701, "global_step": 12857, "epoch": 121, "lr": 8.616723544715117e-05} +{"train_loss": 0.01772303506731987, "grad_norm": 0.07510820031166077, "global_step": 12858, "epoch": 121, "lr": 8.616518894097909e-05} +{"train_loss": 0.016335980966687202, "grad_norm": 0.14630445837974548, "global_step": 12859, "epoch": 121, "lr": 8.616314230773827e-05} +{"train_loss": 0.01684945449233055, "grad_norm": 0.04892256483435631, "global_step": 12860, "epoch": 121, "lr": 8.61610955474359e-05} +{"train_loss": 0.016092106699943542, "grad_norm": 0.09150690585374832, "global_step": 12861, "epoch": 121, "lr": 8.615904866007914e-05} +{"train_loss": 0.0165615975856781, "grad_norm": 0.08818814158439636, "global_step": 12862, "epoch": 121, "lr": 8.615700164567522e-05} +{"train_loss": 0.01772972196340561, "grad_norm": 0.08430638909339905, "global_step": 12863, "epoch": 121, "lr": 8.615495450423134e-05} +{"train_loss": 0.01506507396697998, "grad_norm": 0.05050544813275337, "global_step": 12864, "epoch": 121, "lr": 8.615290723575465e-05} +{"train_loss": 0.016920745372772217, "grad_norm": 0.10874751210212708, "global_step": 12865, "epoch": 121, "lr": 8.615085984025238e-05} +{"train_loss": 0.0165904238820076, "grad_norm": 0.08185835927724838, "global_step": 12866, "epoch": 121, "lr": 8.614881231773169e-05} +{"train_loss": 0.019010329619050026, "grad_norm": 0.06639672070741653, "global_step": 12867, "epoch": 121, "lr": 8.614676466819982e-05} +{"train_loss": 0.015971442684531212, "grad_norm": 0.04786328598856926, "global_step": 12868, "epoch": 121, "lr": 8.614471689166392e-05} +{"train_loss": 0.017049258574843407, "grad_norm": 0.04378638416528702, "global_step": 12869, "epoch": 121, "lr": 8.614266898813121e-05} +{"train_loss": 0.018427379429340363, "grad_norm": 0.08898889273405075, "global_step": 12870, "epoch": 121, "lr": 8.614062095760888e-05} +{"train_loss": 0.016585733741521835, "grad_norm": 0.07666612416505814, "global_step": 12871, "epoch": 121, "lr": 8.613857280010414e-05} +{"train_loss": 0.015838805586099625, "grad_norm": 0.09028467535972595, "global_step": 12872, "epoch": 121, "lr": 8.613652451562415e-05} +{"train_loss": 0.015733901411294937, "grad_norm": 0.08133108913898468, "global_step": 12873, "epoch": 121, "lr": 8.613447610417613e-05} +{"train_loss": 0.01707829162478447, "grad_norm": 0.11998141556978226, "global_step": 12874, "epoch": 121, "lr": 8.613242756576728e-05} +{"train_loss": 0.020178981125354767, "grad_norm": 0.11556906998157501, "global_step": 12875, "epoch": 121, "lr": 8.613037890040479e-05} +{"train_loss": 0.014554502442479134, "grad_norm": 0.0637277290225029, "global_step": 12876, "epoch": 121, "lr": 8.612833010809587e-05} +{"train_loss": 0.018366344273090363, "grad_norm": 0.10827258974313736, "global_step": 12877, "epoch": 121, "lr": 8.61262811888477e-05} +{"train_loss": 0.016290079802274704, "grad_norm": 0.0677855834364891, "global_step": 12878, "epoch": 121, "lr": 8.612423214266749e-05} +{"train_loss": 0.01859929971396923, "grad_norm": 0.11598775535821915, "global_step": 12879, "epoch": 121, "lr": 8.612218296956245e-05} +{"train_loss": 0.01736712083220482, "grad_norm": 0.059867650270462036, "global_step": 12880, "epoch": 121, "lr": 8.612013366953976e-05} +{"train_loss": 0.017847009003162384, "grad_norm": 0.06682151556015015, "global_step": 12881, "epoch": 121, "lr": 8.611808424260663e-05} +{"train_loss": 0.01463807001709938, "grad_norm": 0.08282025158405304, "global_step": 12882, "epoch": 121, "lr": 8.611603468877025e-05} +{"train_loss": 0.017633652314543724, "grad_norm": 0.054484594613313675, "global_step": 12883, "epoch": 121, "lr": 8.611398500803784e-05} +{"train_loss": 0.01599784754216671, "grad_norm": 0.09553948789834976, "global_step": 12884, "epoch": 121, "lr": 8.61119352004166e-05} +{"train_loss": 0.015887072309851646, "grad_norm": 0.13724003732204437, "global_step": 12885, "epoch": 121, "lr": 8.61098852659137e-05} +{"train_loss": 0.016901925206184387, "grad_norm": 0.06926613301038742, "global_step": 12886, "epoch": 121, "lr": 8.610783520453638e-05} +{"train_loss": 0.017291948199272156, "grad_norm": 0.06922533363103867, "global_step": 12887, "epoch": 121, "lr": 8.610578501629184e-05} +{"train_loss": 0.015840888023376465, "grad_norm": 0.08584557473659515, "global_step": 12888, "epoch": 121, "lr": 8.610373470118727e-05} +{"train_loss": 0.016634875908493996, "grad_norm": 0.12113781273365021, "global_step": 12889, "epoch": 121, "lr": 8.610168425922987e-05} +{"train_loss": 0.01809033378958702, "grad_norm": 0.062026578933000565, "global_step": 12890, "epoch": 121, "lr": 8.609963369042685e-05} +{"train_loss": 0.019076813012361526, "grad_norm": 0.04386045038700104, "global_step": 12891, "epoch": 121, "lr": 8.609758299478542e-05} +{"train_loss": 0.018719611689448357, "grad_norm": 0.06797950714826584, "global_step": 12892, "epoch": 121, "lr": 8.609553217231279e-05} +{"train_loss": 0.015217933803796768, "grad_norm": 0.05871490016579628, "global_step": 12893, "epoch": 121, "lr": 8.609348122301615e-05} +{"train_loss": 0.01822732388973236, "grad_norm": 0.07210146635770798, "global_step": 12894, "epoch": 121, "lr": 8.60914301469027e-05} +{"train_loss": 0.01806856319308281, "grad_norm": 0.06728500872850418, "global_step": 12895, "epoch": 121, "lr": 8.608937894397968e-05} +{"train_loss": 0.017171332612633705, "grad_norm": 0.12508128583431244, "global_step": 12896, "epoch": 121, "lr": 8.608732761425427e-05} +{"train_loss": 0.015165582299232483, "grad_norm": 0.10409726947546005, "global_step": 12897, "epoch": 121, "lr": 8.608527615773368e-05} +{"train_loss": 0.020615288987755775, "grad_norm": 0.12733475863933563, "global_step": 12898, "epoch": 121, "lr": 8.608322457442512e-05} +{"train_loss": 0.01688269153237343, "grad_norm": 0.0667872428894043, "global_step": 12899, "epoch": 121, "lr": 8.608117286433582e-05} +{"train_loss": 0.015025553293526173, "grad_norm": 0.049925755709409714, "global_step": 12900, "epoch": 121, "lr": 8.607912102747296e-05} +{"train_loss": 0.01652616634964943, "grad_norm": 0.09424237906932831, "global_step": 12901, "epoch": 121, "lr": 8.607706906384376e-05} +{"train_loss": 0.01725875586271286, "grad_norm": 0.07589206844568253, "global_step": 12902, "epoch": 121, "lr": 8.607501697345543e-05} +{"train_loss": 0.015913955867290497, "grad_norm": 0.07264649868011475, "global_step": 12903, "epoch": 121, "lr": 8.607296475631517e-05} +{"train_loss": 0.01671159453690052, "grad_norm": 0.061269860714673996, "global_step": 12904, "epoch": 121, "lr": 8.60709124124302e-05} +{"train_loss": 0.015233448706567287, "grad_norm": 0.0854213759303093, "global_step": 12905, "epoch": 121, "lr": 8.606885994180775e-05} +{"train_loss": 0.017051275819540024, "grad_norm": 0.10525171458721161, "global_step": 12906, "epoch": 121, "lr": 8.606680734445499e-05} +{"train_loss": 0.016514766961336136, "grad_norm": 0.07451676577329636, "global_step": 12907, "epoch": 121, "lr": 8.606475462037916e-05} +{"train_loss": 0.01704968512058258, "grad_norm": 0.07457752525806427, "global_step": 12908, "epoch": 121, "lr": 8.606270176958747e-05} +{"train_loss": 0.018984220921993256, "grad_norm": 0.08201082795858383, "global_step": 12909, "epoch": 121, "lr": 8.606064879208712e-05} +{"train_loss": 0.015726007521152496, "grad_norm": 0.05213360860943794, "global_step": 12910, "epoch": 121, "lr": 8.605859568788535e-05} +{"train_loss": 0.016362939029932022, "grad_norm": 0.05379895120859146, "global_step": 12911, "epoch": 121, "lr": 8.605654245698935e-05} +{"train_loss": 0.01625974103808403, "grad_norm": 0.08362237364053726, "global_step": 12912, "epoch": 121, "lr": 8.605448909940633e-05} +{"train_loss": 0.020377930253744125, "grad_norm": 0.0956730917096138, "global_step": 12913, "epoch": 121, "lr": 8.605243561514353e-05} +{"train_loss": 0.020172622054815292, "grad_norm": 0.06429202854633331, "global_step": 12914, "epoch": 121, "lr": 8.605038200420814e-05} +{"train_loss": 0.017532650381326675, "grad_norm": 0.058727867901325226, "global_step": 12915, "epoch": 121, "lr": 8.604832826660738e-05} +{"train_loss": 0.018181826919317245, "grad_norm": 0.08976943790912628, "global_step": 12916, "epoch": 121, "lr": 8.60462744023485e-05} +{"train_loss": 0.01525900512933731, "grad_norm": 0.04334819316864014, "global_step": 12917, "epoch": 121, "lr": 8.604422041143864e-05} +{"train_loss": 0.016612790524959564, "grad_norm": 0.05597059428691864, "global_step": 12918, "epoch": 121, "lr": 8.60421662938851e-05} +{"train_loss": 0.015948563814163208, "grad_norm": 0.06287723779678345, "global_step": 12919, "epoch": 121, "lr": 8.604011204969507e-05} +{"train_loss": 0.01720169559121132, "grad_norm": 0.14530712366104126, "global_step": 12920, "epoch": 121, "lr": 8.603805767887573e-05} +{"train_loss": 0.016726678237318993, "grad_norm": 0.1147763729095459, "global_step": 12921, "epoch": 121, "lr": 8.603600318143434e-05} +{"train_loss": 0.01953889988362789, "grad_norm": 0.05317755043506622, "global_step": 12922, "epoch": 121, "lr": 8.60339485573781e-05} +{"train_loss": 0.017252683639526367, "grad_norm": 0.06534866988658905, "global_step": 12923, "epoch": 121, "lr": 8.603189380671425e-05} +{"train_loss": 0.019052598625421524, "grad_norm": 0.13837753236293793, "global_step": 12924, "epoch": 121, "lr": 8.602983892945e-05} +{"train_loss": 0.01836492121219635, "grad_norm": 0.09690830856561661, "global_step": 12925, "epoch": 121, "lr": 8.602778392559254e-05} +{"train_loss": 0.015731066465377808, "grad_norm": 0.06951874494552612, "global_step": 12926, "epoch": 121, "lr": 8.602572879514913e-05} +{"train_loss": 0.01931915245950222, "grad_norm": 0.06449299305677414, "global_step": 12927, "epoch": 121, "lr": 8.602367353812696e-05} +{"train_loss": 0.01627018302679062, "grad_norm": 0.04824979603290558, "global_step": 12928, "epoch": 121, "lr": 8.60216181545333e-05} +{"train_loss": 0.018766969442367554, "grad_norm": 0.06950363516807556, "global_step": 12929, "epoch": 121, "lr": 8.601956264437532e-05} +{"train_loss": 0.014664696529507637, "grad_norm": 0.05046918988227844, "global_step": 12930, "epoch": 121, "lr": 8.601750700766027e-05} +{"train_loss": 0.017106474368428846, "grad_norm": 0.08975960314273834, "global_step": 12931, "epoch": 121, "lr": 8.601545124439535e-05, "val_loss": 0.01852184534072876} +{"train_loss": 0.01793229952454567, "grad_norm": 0.07098830491304398, "global_step": 12932, "epoch": 122, "lr": 8.60133953545878e-05} +{"train_loss": 0.015507807023823261, "grad_norm": 0.09484041482210159, "global_step": 12933, "epoch": 122, "lr": 8.601133933824484e-05} +{"train_loss": 0.01788075640797615, "grad_norm": 0.08105707168579102, "global_step": 12934, "epoch": 122, "lr": 8.600928319537371e-05} +{"train_loss": 0.014484122395515442, "grad_norm": 0.11286315321922302, "global_step": 12935, "epoch": 122, "lr": 8.60072269259816e-05} +{"train_loss": 0.020597193390130997, "grad_norm": 0.14711794257164001, "global_step": 12936, "epoch": 122, "lr": 8.600517053007578e-05} +{"train_loss": 0.01659514382481575, "grad_norm": 0.06058010831475258, "global_step": 12937, "epoch": 122, "lr": 8.600311400766341e-05} +{"train_loss": 0.017108306288719177, "grad_norm": 0.04869937151670456, "global_step": 12938, "epoch": 122, "lr": 8.600105735875177e-05} +{"train_loss": 0.023166131228208542, "grad_norm": 0.08003479987382889, "global_step": 12939, "epoch": 122, "lr": 8.599900058334808e-05} +{"train_loss": 0.017122343182563782, "grad_norm": 0.11648549884557724, "global_step": 12940, "epoch": 122, "lr": 8.599694368145956e-05} +{"train_loss": 0.014140295796096325, "grad_norm": 0.06490880995988846, "global_step": 12941, "epoch": 122, "lr": 8.599488665309344e-05} +{"train_loss": 0.015287719666957855, "grad_norm": 0.05237048491835594, "global_step": 12942, "epoch": 122, "lr": 8.59928294982569e-05} +{"train_loss": 0.01588587462902069, "grad_norm": 0.08433782309293747, "global_step": 12943, "epoch": 122, "lr": 8.599077221695726e-05} +{"train_loss": 0.020819304510951042, "grad_norm": 0.06734854727983475, "global_step": 12944, "epoch": 122, "lr": 8.598871480920167e-05} +{"train_loss": 0.017633676528930664, "grad_norm": 0.044736120849847794, "global_step": 12945, "epoch": 122, "lr": 8.59866572749974e-05} +{"train_loss": 0.01432187668979168, "grad_norm": 0.10855954140424728, "global_step": 12946, "epoch": 122, "lr": 8.598459961435166e-05} +{"train_loss": 0.015886716544628143, "grad_norm": 0.05988481268286705, "global_step": 12947, "epoch": 122, "lr": 8.59825418272717e-05} +{"train_loss": 0.016345400363206863, "grad_norm": 0.07323019206523895, "global_step": 12948, "epoch": 122, "lr": 8.598048391376472e-05} +{"train_loss": 0.016741015017032623, "grad_norm": 0.09014910459518433, "global_step": 12949, "epoch": 122, "lr": 8.597842587383798e-05} +{"train_loss": 0.01923968642950058, "grad_norm": 0.08290093392133713, "global_step": 12950, "epoch": 122, "lr": 8.597636770749868e-05} +{"train_loss": 0.015051053836941719, "grad_norm": 0.04967382550239563, "global_step": 12951, "epoch": 122, "lr": 8.597430941475407e-05} +{"train_loss": 0.01739473082125187, "grad_norm": 0.06915391981601715, "global_step": 12952, "epoch": 122, "lr": 8.597225099561139e-05} +{"train_loss": 0.0168946273624897, "grad_norm": 0.06482145935297012, "global_step": 12953, "epoch": 122, "lr": 8.597019245007787e-05} +{"train_loss": 0.01413477212190628, "grad_norm": 0.07410436868667603, "global_step": 12954, "epoch": 122, "lr": 8.596813377816074e-05} +{"train_loss": 0.01897536590695381, "grad_norm": 0.0903775617480278, "global_step": 12955, "epoch": 122, "lr": 8.596607497986721e-05} +{"train_loss": 0.01825520768761635, "grad_norm": 0.047327905893325806, "global_step": 12956, "epoch": 122, "lr": 8.596401605520455e-05} +{"train_loss": 0.017624234780669212, "grad_norm": 0.09073456376791, "global_step": 12957, "epoch": 122, "lr": 8.596195700417998e-05} +{"train_loss": 0.017875339835882187, "grad_norm": 0.05611369013786316, "global_step": 12958, "epoch": 122, "lr": 8.595989782680072e-05} +{"train_loss": 0.01594981551170349, "grad_norm": 0.06121915951371193, "global_step": 12959, "epoch": 122, "lr": 8.595783852307403e-05} +{"train_loss": 0.014887817203998566, "grad_norm": 0.04641639068722725, "global_step": 12960, "epoch": 122, "lr": 8.595577909300712e-05} +{"train_loss": 0.017116863280534744, "grad_norm": 0.1307211071252823, "global_step": 12961, "epoch": 122, "lr": 8.595371953660725e-05} +{"train_loss": 0.016892969608306885, "grad_norm": 0.0825975090265274, "global_step": 12962, "epoch": 122, "lr": 8.595165985388164e-05} +{"train_loss": 0.014269312843680382, "grad_norm": 0.0613420195877552, "global_step": 12963, "epoch": 122, "lr": 8.594960004483753e-05} +{"train_loss": 0.018021980300545692, "grad_norm": 0.07215046137571335, "global_step": 12964, "epoch": 122, "lr": 8.594754010948216e-05} +{"train_loss": 0.017470702528953552, "grad_norm": 0.110293909907341, "global_step": 12965, "epoch": 122, "lr": 8.594548004782277e-05} +{"train_loss": 0.016228467226028442, "grad_norm": 0.13246747851371765, "global_step": 12966, "epoch": 122, "lr": 8.59434198598666e-05} +{"train_loss": 0.01542888954281807, "grad_norm": 0.14250490069389343, "global_step": 12967, "epoch": 122, "lr": 8.594135954562088e-05} +{"train_loss": 0.01675877347588539, "grad_norm": 0.07639271765947342, "global_step": 12968, "epoch": 122, "lr": 8.593929910509286e-05} +{"train_loss": 0.0178071279078722, "grad_norm": 0.053425777703523636, "global_step": 12969, "epoch": 122, "lr": 8.593723853828976e-05} +{"train_loss": 0.016394352540373802, "grad_norm": 0.04296442121267319, "global_step": 12970, "epoch": 122, "lr": 8.593517784521883e-05} +{"train_loss": 0.016036050394177437, "grad_norm": 0.10606177896261215, "global_step": 12971, "epoch": 122, "lr": 8.593311702588733e-05} +{"train_loss": 0.016752086579799652, "grad_norm": 0.10576848685741425, "global_step": 12972, "epoch": 122, "lr": 8.593105608030247e-05} +{"train_loss": 0.016260884702205658, "grad_norm": 0.05774274095892906, "global_step": 12973, "epoch": 122, "lr": 8.592899500847151e-05} +{"train_loss": 0.01819855347275734, "grad_norm": 0.07138585299253464, "global_step": 12974, "epoch": 122, "lr": 8.592693381040169e-05} +{"train_loss": 0.017536023631691933, "grad_norm": 0.05653015524148941, "global_step": 12975, "epoch": 122, "lr": 8.592487248610024e-05} +{"train_loss": 0.014989001676440239, "grad_norm": 0.05651455000042915, "global_step": 12976, "epoch": 122, "lr": 8.592281103557441e-05} +{"train_loss": 0.016871564090251923, "grad_norm": 0.06003929302096367, "global_step": 12977, "epoch": 122, "lr": 8.592074945883145e-05} +{"train_loss": 0.01864471286535263, "grad_norm": 0.07534082233905792, "global_step": 12978, "epoch": 122, "lr": 8.59186877558786e-05} +{"train_loss": 0.019999809563159943, "grad_norm": 0.05374620854854584, "global_step": 12979, "epoch": 122, "lr": 8.59166259267231e-05} +{"train_loss": 0.019571594893932343, "grad_norm": 0.06090182811021805, "global_step": 12980, "epoch": 122, "lr": 8.59145639713722e-05} +{"train_loss": 0.01441875658929348, "grad_norm": 0.06340261548757553, "global_step": 12981, "epoch": 122, "lr": 8.591250188983313e-05} +{"train_loss": 0.017209608107805252, "grad_norm": 0.07117445766925812, "global_step": 12982, "epoch": 122, "lr": 8.591043968211317e-05} +{"train_loss": 0.016138218343257904, "grad_norm": 0.058631159365177155, "global_step": 12983, "epoch": 122, "lr": 8.590837734821952e-05} +{"train_loss": 0.013028180226683617, "grad_norm": 0.09312655031681061, "global_step": 12984, "epoch": 122, "lr": 8.590631488815944e-05} +{"train_loss": 0.017833085730671883, "grad_norm": 0.0701540857553482, "global_step": 12985, "epoch": 122, "lr": 8.590425230194021e-05} +{"train_loss": 0.0171365924179554, "grad_norm": 0.18207032978534698, "global_step": 12986, "epoch": 122, "lr": 8.590218958956904e-05} +{"train_loss": 0.01789642497897148, "grad_norm": 0.10443679988384247, "global_step": 12987, "epoch": 122, "lr": 8.590012675105319e-05} +{"train_loss": 0.016775978729128838, "grad_norm": 0.15545052289962769, "global_step": 12988, "epoch": 122, "lr": 8.58980637863999e-05} +{"train_loss": 0.01597091555595398, "grad_norm": 0.054969288408756256, "global_step": 12989, "epoch": 122, "lr": 8.589600069561644e-05} +{"train_loss": 0.016766151413321495, "grad_norm": 0.05291565880179405, "global_step": 12990, "epoch": 122, "lr": 8.589393747871003e-05} +{"train_loss": 0.014889921993017197, "grad_norm": 0.05815005302429199, "global_step": 12991, "epoch": 122, "lr": 8.589187413568793e-05} +{"train_loss": 0.01543418224900961, "grad_norm": 0.05483732372522354, "global_step": 12992, "epoch": 122, "lr": 8.588981066655742e-05} +{"train_loss": 0.02041957899928093, "grad_norm": 0.11841803789138794, "global_step": 12993, "epoch": 122, "lr": 8.58877470713257e-05} +{"train_loss": 0.017571426928043365, "grad_norm": 0.0586329884827137, "global_step": 12994, "epoch": 122, "lr": 8.588568335000005e-05} +{"train_loss": 0.015844859182834625, "grad_norm": 0.042312417179346085, "global_step": 12995, "epoch": 122, "lr": 8.588361950258771e-05} +{"train_loss": 0.01959836110472679, "grad_norm": 0.07910680770874023, "global_step": 12996, "epoch": 122, "lr": 8.588155552909595e-05} +{"train_loss": 0.015311077237129211, "grad_norm": 0.1691301465034485, "global_step": 12997, "epoch": 122, "lr": 8.5879491429532e-05} +{"train_loss": 0.015976889058947563, "grad_norm": 0.08744462579488754, "global_step": 12998, "epoch": 122, "lr": 8.587742720390312e-05} +{"train_loss": 0.012844990938901901, "grad_norm": 0.05228715389966965, "global_step": 12999, "epoch": 122, "lr": 8.587536285221656e-05} +{"train_loss": 0.018528874963521957, "grad_norm": 0.1509564071893692, "global_step": 13000, "epoch": 122, "lr": 8.587329837447958e-05} +{"train_loss": 0.019247200340032578, "grad_norm": 0.12307704240083694, "global_step": 13001, "epoch": 122, "lr": 8.587123377069943e-05} +{"train_loss": 0.015868976712226868, "grad_norm": 0.06218600645661354, "global_step": 13002, "epoch": 122, "lr": 8.586916904088337e-05} +{"train_loss": 0.01804717630147934, "grad_norm": 0.06205255165696144, "global_step": 13003, "epoch": 122, "lr": 8.586710418503864e-05} +{"train_loss": 0.01928819715976715, "grad_norm": 0.06294648349285126, "global_step": 13004, "epoch": 122, "lr": 8.586503920317252e-05} +{"train_loss": 0.012615292333066463, "grad_norm": 0.06277415156364441, "global_step": 13005, "epoch": 122, "lr": 8.586297409529223e-05} +{"train_loss": 0.01859895884990692, "grad_norm": 0.051927175372838974, "global_step": 13006, "epoch": 122, "lr": 8.586090886140505e-05} +{"train_loss": 0.017864471301436424, "grad_norm": 0.09218377619981766, "global_step": 13007, "epoch": 122, "lr": 8.585884350151824e-05} +{"train_loss": 0.017119944095611572, "grad_norm": 0.05872360244393349, "global_step": 13008, "epoch": 122, "lr": 8.585677801563905e-05} +{"train_loss": 0.01510101929306984, "grad_norm": 0.10024494677782059, "global_step": 13009, "epoch": 122, "lr": 8.585471240377473e-05} +{"train_loss": 0.018395844846963882, "grad_norm": 0.0959063321352005, "global_step": 13010, "epoch": 122, "lr": 8.585264666593253e-05} +{"train_loss": 0.018591269850730896, "grad_norm": 0.110782690346241, "global_step": 13011, "epoch": 122, "lr": 8.585058080211974e-05} +{"train_loss": 0.01862221583724022, "grad_norm": 0.17533472180366516, "global_step": 13012, "epoch": 122, "lr": 8.58485148123436e-05} +{"train_loss": 0.01927386224269867, "grad_norm": 0.07539425790309906, "global_step": 13013, "epoch": 122, "lr": 8.584644869661136e-05} +{"train_loss": 0.017561236396431923, "grad_norm": 0.052417077124118805, "global_step": 13014, "epoch": 122, "lr": 8.584438245493029e-05} +{"train_loss": 0.015960559248924255, "grad_norm": 0.07216187566518784, "global_step": 13015, "epoch": 122, "lr": 8.584231608730766e-05} +{"train_loss": 0.01672722026705742, "grad_norm": 0.08755337446928024, "global_step": 13016, "epoch": 122, "lr": 8.584024959375072e-05} +{"train_loss": 0.017635274678468704, "grad_norm": 0.10011296719312668, "global_step": 13017, "epoch": 122, "lr": 8.58381829742667e-05} +{"train_loss": 0.01747380569577217, "grad_norm": 0.10304269939661026, "global_step": 13018, "epoch": 122, "lr": 8.583611622886292e-05} +{"train_loss": 0.017038170248270035, "grad_norm": 0.04382999241352081, "global_step": 13019, "epoch": 122, "lr": 8.583404935754659e-05} +{"train_loss": 0.020479297265410423, "grad_norm": 0.12926918268203735, "global_step": 13020, "epoch": 122, "lr": 8.583198236032502e-05} +{"train_loss": 0.017418865114450455, "grad_norm": 0.09145356714725494, "global_step": 13021, "epoch": 122, "lr": 8.582991523720542e-05} +{"train_loss": 0.01575033739209175, "grad_norm": 0.07769117504358292, "global_step": 13022, "epoch": 122, "lr": 8.58278479881951e-05} +{"train_loss": 0.01907147839665413, "grad_norm": 0.07254938781261444, "global_step": 13023, "epoch": 122, "lr": 8.582578061330129e-05} +{"train_loss": 0.017131511121988297, "grad_norm": 0.07184122502803802, "global_step": 13024, "epoch": 122, "lr": 8.582371311253127e-05} +{"train_loss": 0.01650090515613556, "grad_norm": 0.1110091432929039, "global_step": 13025, "epoch": 122, "lr": 8.582164548589232e-05} +{"train_loss": 0.017275091260671616, "grad_norm": 0.0773908868432045, "global_step": 13026, "epoch": 122, "lr": 8.581957773339167e-05} +{"train_loss": 0.018448416143655777, "grad_norm": 0.05041739344596863, "global_step": 13027, "epoch": 122, "lr": 8.58175098550366e-05} +{"train_loss": 0.018397513777017593, "grad_norm": 0.08580996841192245, "global_step": 13028, "epoch": 122, "lr": 8.58154418508344e-05} +{"train_loss": 0.016363020986318588, "grad_norm": 0.1034175232052803, "global_step": 13029, "epoch": 122, "lr": 8.581337372079228e-05} +{"train_loss": 0.01741727441549301, "grad_norm": 0.06400016695261002, "global_step": 13030, "epoch": 122, "lr": 8.581130546491754e-05} +{"train_loss": 0.016494063660502434, "grad_norm": 0.05231871455907822, "global_step": 13031, "epoch": 122, "lr": 8.580923708321747e-05} +{"train_loss": 0.019097931683063507, "grad_norm": 0.0733131468296051, "global_step": 13032, "epoch": 122, "lr": 8.58071685756993e-05} +{"train_loss": 0.016101881861686707, "grad_norm": 0.08072429895401001, "global_step": 13033, "epoch": 122, "lr": 8.580509994237032e-05} +{"train_loss": 0.017741529271006584, "grad_norm": 0.09583744406700134, "global_step": 13034, "epoch": 122, "lr": 8.580303118323779e-05} +{"train_loss": 0.014793949201703072, "grad_norm": 0.06525995582342148, "global_step": 13035, "epoch": 122, "lr": 8.580096229830896e-05} +{"train_loss": 0.017365101724863052, "grad_norm": 0.0763525664806366, "global_step": 13036, "epoch": 122, "lr": 8.579889328759114e-05} +{"train_loss": 0.017057895677972515, "grad_norm": 0.08912461996078491, "global_step": 13037, "epoch": 122, "lr": 8.579682415109156e-05, "val_loss": 0.019446028396487236} +{"train_loss": 0.016237959265708923, "grad_norm": 0.05433487892150879, "global_step": 13038, "epoch": 123, "lr": 8.579475488881752e-05} +{"train_loss": 0.015186341479420662, "grad_norm": 0.04860086366534233, "global_step": 13039, "epoch": 123, "lr": 8.579268550077628e-05} +{"train_loss": 0.015934064984321594, "grad_norm": 0.08504556119441986, "global_step": 13040, "epoch": 123, "lr": 8.57906159869751e-05} +{"train_loss": 0.021046146750450134, "grad_norm": 0.16041703522205353, "global_step": 13041, "epoch": 123, "lr": 8.578854634742126e-05} +{"train_loss": 0.018515409901738167, "grad_norm": 0.10852892696857452, "global_step": 13042, "epoch": 123, "lr": 8.578647658212203e-05} +{"train_loss": 0.02106032520532608, "grad_norm": 0.11331383138895035, "global_step": 13043, "epoch": 123, "lr": 8.578440669108469e-05} +{"train_loss": 0.018025968223810196, "grad_norm": 0.15544700622558594, "global_step": 13044, "epoch": 123, "lr": 8.578233667431652e-05} +{"train_loss": 0.017290962859988213, "grad_norm": 0.06316050887107849, "global_step": 13045, "epoch": 123, "lr": 8.578026653182476e-05} +{"train_loss": 0.015368957072496414, "grad_norm": 0.04872123524546623, "global_step": 13046, "epoch": 123, "lr": 8.577819626361672e-05} +{"train_loss": 0.01691216602921486, "grad_norm": 0.15257316827774048, "global_step": 13047, "epoch": 123, "lr": 8.577612586969966e-05} +{"train_loss": 0.017323357984423637, "grad_norm": 0.10346506536006927, "global_step": 13048, "epoch": 123, "lr": 8.577405535008082e-05} +{"train_loss": 0.017423465847969055, "grad_norm": 0.0902688279747963, "global_step": 13049, "epoch": 123, "lr": 8.577198470476753e-05} +{"train_loss": 0.014977339655160904, "grad_norm": 0.10805059224367142, "global_step": 13050, "epoch": 123, "lr": 8.576991393376704e-05} +{"train_loss": 0.018846023827791214, "grad_norm": 0.0535537451505661, "global_step": 13051, "epoch": 123, "lr": 8.576784303708664e-05} +{"train_loss": 0.015332363545894623, "grad_norm": 0.09283121675252914, "global_step": 13052, "epoch": 123, "lr": 8.576577201473358e-05} +{"train_loss": 0.013550572097301483, "grad_norm": 0.059827979654073715, "global_step": 13053, "epoch": 123, "lr": 8.576370086671515e-05} +{"train_loss": 0.018810369074344635, "grad_norm": 0.06922847777605057, "global_step": 13054, "epoch": 123, "lr": 8.576162959303864e-05} +{"train_loss": 0.016874395310878754, "grad_norm": 0.0658426433801651, "global_step": 13055, "epoch": 123, "lr": 8.575955819371131e-05} +{"train_loss": 0.01743912324309349, "grad_norm": 0.07535069435834885, "global_step": 13056, "epoch": 123, "lr": 8.575748666874044e-05} +{"train_loss": 0.01295593660324812, "grad_norm": 0.06079432740807533, "global_step": 13057, "epoch": 123, "lr": 8.575541501813331e-05} +{"train_loss": 0.01879299059510231, "grad_norm": 0.10061740130186081, "global_step": 13058, "epoch": 123, "lr": 8.575334324189722e-05} +{"train_loss": 0.016922954469919205, "grad_norm": 0.07900802046060562, "global_step": 13059, "epoch": 123, "lr": 8.575127134003941e-05} +{"train_loss": 0.015340873971581459, "grad_norm": 0.04500902444124222, "global_step": 13060, "epoch": 123, "lr": 8.574919931256719e-05} +{"train_loss": 0.01805718243122101, "grad_norm": 0.07983529567718506, "global_step": 13061, "epoch": 123, "lr": 8.574712715948783e-05} +{"train_loss": 0.018209604546427727, "grad_norm": 0.10615979880094528, "global_step": 13062, "epoch": 123, "lr": 8.574505488080862e-05} +{"train_loss": 0.016020001843571663, "grad_norm": 0.06266291439533234, "global_step": 13063, "epoch": 123, "lr": 8.574298247653682e-05} +{"train_loss": 0.016728967428207397, "grad_norm": 0.06458307057619095, "global_step": 13064, "epoch": 123, "lr": 8.574090994667972e-05} +{"train_loss": 0.01627518981695175, "grad_norm": 0.14893046021461487, "global_step": 13065, "epoch": 123, "lr": 8.573883729124461e-05} +{"train_loss": 0.019332285970449448, "grad_norm": 0.10010752826929092, "global_step": 13066, "epoch": 123, "lr": 8.573676451023877e-05} +{"train_loss": 0.016598476096987724, "grad_norm": 0.14551524817943573, "global_step": 13067, "epoch": 123, "lr": 8.573469160366948e-05} +{"train_loss": 0.014622123911976814, "grad_norm": 0.05046163126826286, "global_step": 13068, "epoch": 123, "lr": 8.573261857154403e-05} +{"train_loss": 0.017495503649115562, "grad_norm": 0.05134817957878113, "global_step": 13069, "epoch": 123, "lr": 8.573054541386969e-05} +{"train_loss": 0.01666930317878723, "grad_norm": 0.0452142134308815, "global_step": 13070, "epoch": 123, "lr": 8.572847213065376e-05} +{"train_loss": 0.015781939029693604, "grad_norm": 0.038241006433963776, "global_step": 13071, "epoch": 123, "lr": 8.572639872190352e-05} +{"train_loss": 0.015169670805335045, "grad_norm": 0.06304775923490524, "global_step": 13072, "epoch": 123, "lr": 8.572432518762622e-05} +{"train_loss": 0.018753258511424065, "grad_norm": 0.10818924009799957, "global_step": 13073, "epoch": 123, "lr": 8.572225152782921e-05} +{"train_loss": 0.01811515912413597, "grad_norm": 0.08359375596046448, "global_step": 13074, "epoch": 123, "lr": 8.572017774251974e-05} +{"train_loss": 0.01851159706711769, "grad_norm": 0.05130963400006294, "global_step": 13075, "epoch": 123, "lr": 8.57181038317051e-05} +{"train_loss": 0.01603078842163086, "grad_norm": 0.08515844494104385, "global_step": 13076, "epoch": 123, "lr": 8.571602979539257e-05} +{"train_loss": 0.019964691251516342, "grad_norm": 0.08096448332071304, "global_step": 13077, "epoch": 123, "lr": 8.571395563358944e-05} +{"train_loss": 0.01512145809829235, "grad_norm": 0.044716350734233856, "global_step": 13078, "epoch": 123, "lr": 8.571188134630302e-05} +{"train_loss": 0.016912637278437614, "grad_norm": 0.0531904436647892, "global_step": 13079, "epoch": 123, "lr": 8.570980693354056e-05} +{"train_loss": 0.01731998845934868, "grad_norm": 0.05942618101835251, "global_step": 13080, "epoch": 123, "lr": 8.570773239530937e-05} +{"train_loss": 0.02040979638695717, "grad_norm": 0.07038405537605286, "global_step": 13081, "epoch": 123, "lr": 8.570565773161674e-05} +{"train_loss": 0.019558629021048546, "grad_norm": 0.08409589529037476, "global_step": 13082, "epoch": 123, "lr": 8.570358294246996e-05} +{"train_loss": 0.016748085618019104, "grad_norm": 0.08973988145589828, "global_step": 13083, "epoch": 123, "lr": 8.57015080278763e-05} +{"train_loss": 0.0191980991512537, "grad_norm": 0.06619055569171906, "global_step": 13084, "epoch": 123, "lr": 8.56994329878431e-05} +{"train_loss": 0.015762489289045334, "grad_norm": 0.08323665708303452, "global_step": 13085, "epoch": 123, "lr": 8.569735782237759e-05} +{"train_loss": 0.017771095037460327, "grad_norm": 0.08515600860118866, "global_step": 13086, "epoch": 123, "lr": 8.569528253148711e-05} +{"train_loss": 0.017010346055030823, "grad_norm": 0.09269411116838455, "global_step": 13087, "epoch": 123, "lr": 8.569320711517892e-05} +{"train_loss": 0.01726897805929184, "grad_norm": 0.07692751288414001, "global_step": 13088, "epoch": 123, "lr": 8.569113157346031e-05} +{"train_loss": 0.015612390823662281, "grad_norm": 0.056002773344516754, "global_step": 13089, "epoch": 123, "lr": 8.56890559063386e-05} +{"train_loss": 0.01962793618440628, "grad_norm": 0.10353464633226395, "global_step": 13090, "epoch": 123, "lr": 8.568698011382107e-05} +{"train_loss": 0.018703782930970192, "grad_norm": 0.17099995911121368, "global_step": 13091, "epoch": 123, "lr": 8.5684904195915e-05} +{"train_loss": 0.015587533824145794, "grad_norm": 0.05437188968062401, "global_step": 13092, "epoch": 123, "lr": 8.568282815262772e-05} +{"train_loss": 0.01695091649889946, "grad_norm": 0.06772059947252274, "global_step": 13093, "epoch": 123, "lr": 8.568075198396648e-05} +{"train_loss": 0.017755135893821716, "grad_norm": 0.10731228440999985, "global_step": 13094, "epoch": 123, "lr": 8.567867568993859e-05} +{"train_loss": 0.014934257604181767, "grad_norm": 0.06680463254451752, "global_step": 13095, "epoch": 123, "lr": 8.567659927055137e-05} +{"train_loss": 0.01563432440161705, "grad_norm": 0.07102683931589127, "global_step": 13096, "epoch": 123, "lr": 8.567452272581206e-05} +{"train_loss": 0.0178369227796793, "grad_norm": 0.10737700760364532, "global_step": 13097, "epoch": 123, "lr": 8.567244605572803e-05} +{"train_loss": 0.016833946108818054, "grad_norm": 0.08834593743085861, "global_step": 13098, "epoch": 123, "lr": 8.567036926030652e-05} +{"train_loss": 0.015140964649617672, "grad_norm": 0.060766108334064484, "global_step": 13099, "epoch": 123, "lr": 8.566829233955485e-05} +{"train_loss": 0.017635539174079895, "grad_norm": 0.10510949790477753, "global_step": 13100, "epoch": 123, "lr": 8.566621529348031e-05} +{"train_loss": 0.016830122098326683, "grad_norm": 0.11263762414455414, "global_step": 13101, "epoch": 123, "lr": 8.56641381220902e-05} +{"train_loss": 0.015970785170793533, "grad_norm": 0.05157223716378212, "global_step": 13102, "epoch": 123, "lr": 8.566206082539181e-05} +{"train_loss": 0.01350492238998413, "grad_norm": 0.05639563128352165, "global_step": 13103, "epoch": 123, "lr": 8.565998340339248e-05} +{"train_loss": 0.02047373354434967, "grad_norm": 0.061506662517786026, "global_step": 13104, "epoch": 123, "lr": 8.565790585609942e-05} +{"train_loss": 0.016802629455924034, "grad_norm": 0.08882440626621246, "global_step": 13105, "epoch": 123, "lr": 8.565582818352003e-05} +{"train_loss": 0.014623607508838177, "grad_norm": 0.13481707870960236, "global_step": 13106, "epoch": 123, "lr": 8.565375038566157e-05} +{"train_loss": 0.017439017072319984, "grad_norm": 0.050181277096271515, "global_step": 13107, "epoch": 123, "lr": 8.565167246253132e-05} +{"train_loss": 0.017358830198645592, "grad_norm": 0.07814576476812363, "global_step": 13108, "epoch": 123, "lr": 8.56495944141366e-05} +{"train_loss": 0.017442137002944946, "grad_norm": 0.08791863173246384, "global_step": 13109, "epoch": 123, "lr": 8.56475162404847e-05} +{"train_loss": 0.014448761940002441, "grad_norm": 0.0753597691655159, "global_step": 13110, "epoch": 123, "lr": 8.564543794158294e-05} +{"train_loss": 0.013780899345874786, "grad_norm": 0.06225605681538582, "global_step": 13111, "epoch": 123, "lr": 8.56433595174386e-05} +{"train_loss": 0.015027115121483803, "grad_norm": 0.0881679356098175, "global_step": 13112, "epoch": 123, "lr": 8.564128096805902e-05} +{"train_loss": 0.01549556478857994, "grad_norm": 0.0623999647796154, "global_step": 13113, "epoch": 123, "lr": 8.563920229345146e-05} +{"train_loss": 0.016060298308730125, "grad_norm": 0.08816149085760117, "global_step": 13114, "epoch": 123, "lr": 8.563712349362325e-05} +{"train_loss": 0.017990605905652046, "grad_norm": 0.05069667473435402, "global_step": 13115, "epoch": 123, "lr": 8.563504456858168e-05} +{"train_loss": 0.01890556886792183, "grad_norm": 0.0637255534529686, "global_step": 13116, "epoch": 123, "lr": 8.563296551833406e-05} +{"train_loss": 0.015038931742310524, "grad_norm": 0.06561630964279175, "global_step": 13117, "epoch": 123, "lr": 8.563088634288771e-05} +{"train_loss": 0.015504921786487103, "grad_norm": 0.0928410142660141, "global_step": 13118, "epoch": 123, "lr": 8.562880704224991e-05} +{"train_loss": 0.015340648591518402, "grad_norm": 0.04724334925413132, "global_step": 13119, "epoch": 123, "lr": 8.562672761642799e-05} +{"train_loss": 0.01869259774684906, "grad_norm": 0.12215796113014221, "global_step": 13120, "epoch": 123, "lr": 8.562464806542922e-05} +{"train_loss": 0.02087489515542984, "grad_norm": 0.08407317847013474, "global_step": 13121, "epoch": 123, "lr": 8.562256838926094e-05} +{"train_loss": 0.020104248076677322, "grad_norm": 0.11537840962409973, "global_step": 13122, "epoch": 123, "lr": 8.562048858793045e-05} +{"train_loss": 0.015741217881441116, "grad_norm": 0.05073152109980583, "global_step": 13123, "epoch": 123, "lr": 8.561840866144504e-05} +{"train_loss": 0.01513225119560957, "grad_norm": 0.05661763250827789, "global_step": 13124, "epoch": 123, "lr": 8.561632860981205e-05} +{"train_loss": 0.0157514326274395, "grad_norm": 0.08870270103216171, "global_step": 13125, "epoch": 123, "lr": 8.561424843303876e-05} +{"train_loss": 0.01846049353480339, "grad_norm": 0.05261954292654991, "global_step": 13126, "epoch": 123, "lr": 8.561216813113249e-05} +{"train_loss": 0.018077023327350616, "grad_norm": 0.07553496211767197, "global_step": 13127, "epoch": 123, "lr": 8.561008770410055e-05} +{"train_loss": 0.019243456423282623, "grad_norm": 0.08996430039405823, "global_step": 13128, "epoch": 123, "lr": 8.560800715195025e-05} +{"train_loss": 0.015066981315612793, "grad_norm": 0.07024221122264862, "global_step": 13129, "epoch": 123, "lr": 8.56059264746889e-05} +{"train_loss": 0.020219601690769196, "grad_norm": 0.06794925034046173, "global_step": 13130, "epoch": 123, "lr": 8.560384567232381e-05} +{"train_loss": 0.016829397529363632, "grad_norm": 0.049900393933057785, "global_step": 13131, "epoch": 123, "lr": 8.560176474486228e-05} +{"train_loss": 0.01779376156628132, "grad_norm": 0.06035333499312401, "global_step": 13132, "epoch": 123, "lr": 8.559968369231163e-05} +{"train_loss": 0.017527248710393906, "grad_norm": 0.10761351138353348, "global_step": 13133, "epoch": 123, "lr": 8.559760251467918e-05} +{"train_loss": 0.020510178059339523, "grad_norm": 0.13648346066474915, "global_step": 13134, "epoch": 123, "lr": 8.559552121197225e-05} +{"train_loss": 0.015370499342679977, "grad_norm": 0.06123430281877518, "global_step": 13135, "epoch": 123, "lr": 8.559343978419812e-05} +{"train_loss": 0.015103934332728386, "grad_norm": 0.06697387248277664, "global_step": 13136, "epoch": 123, "lr": 8.559135823136413e-05} +{"train_loss": 0.019283989444375038, "grad_norm": 0.09852565079927444, "global_step": 13137, "epoch": 123, "lr": 8.558927655347758e-05} +{"train_loss": 0.016674460843205452, "grad_norm": 0.08833609521389008, "global_step": 13138, "epoch": 123, "lr": 8.55871947505458e-05} +{"train_loss": 0.017611952498555183, "grad_norm": 0.05391734093427658, "global_step": 13139, "epoch": 123, "lr": 8.558511282257608e-05} +{"train_loss": 0.017195817083120346, "grad_norm": 0.11482126265764236, "global_step": 13140, "epoch": 123, "lr": 8.558303076957575e-05} +{"train_loss": 0.014689658768475056, "grad_norm": 0.08684471994638443, "global_step": 13141, "epoch": 123, "lr": 8.558094859155214e-05} +{"train_loss": 0.01798538863658905, "grad_norm": 0.07979393750429153, "global_step": 13142, "epoch": 123, "lr": 8.557886628851252e-05} +{"train_loss": 0.017044055442554207, "grad_norm": 0.07771290093660355, "global_step": 13143, "epoch": 123, "lr": 8.557678386046428e-05, "val_loss": 0.01926986314356327} +{"train_loss": 0.016848279163241386, "grad_norm": 0.1147625744342804, "global_step": 13144, "epoch": 124, "lr": 8.557470130741468e-05} +{"train_loss": 0.015865158289670944, "grad_norm": 0.06709901243448257, "global_step": 13145, "epoch": 124, "lr": 8.557261862937102e-05} +{"train_loss": 0.016223488375544548, "grad_norm": 0.06735733151435852, "global_step": 13146, "epoch": 124, "lr": 8.557053582634066e-05} +{"train_loss": 0.016036922112107277, "grad_norm": 0.09612035006284714, "global_step": 13147, "epoch": 124, "lr": 8.556845289833093e-05} +{"train_loss": 0.01591489464044571, "grad_norm": 0.06779530644416809, "global_step": 13148, "epoch": 124, "lr": 8.556636984534912e-05} +{"train_loss": 0.015158915892243385, "grad_norm": 0.08936218172311783, "global_step": 13149, "epoch": 124, "lr": 8.556428666740253e-05} +{"train_loss": 0.0182323195040226, "grad_norm": 0.05783193185925484, "global_step": 13150, "epoch": 124, "lr": 8.556220336449852e-05} +{"train_loss": 0.015798036009073257, "grad_norm": 0.11473026126623154, "global_step": 13151, "epoch": 124, "lr": 8.556011993664438e-05} +{"train_loss": 0.01729412190616131, "grad_norm": 0.047745149582624435, "global_step": 13152, "epoch": 124, "lr": 8.555803638384746e-05} +{"train_loss": 0.016060229390859604, "grad_norm": 0.11188406497240067, "global_step": 13153, "epoch": 124, "lr": 8.555595270611506e-05} +{"train_loss": 0.015124663710594177, "grad_norm": 0.06303976476192474, "global_step": 13154, "epoch": 124, "lr": 8.55538689034545e-05} +{"train_loss": 0.01754353940486908, "grad_norm": 0.05322335660457611, "global_step": 13155, "epoch": 124, "lr": 8.555178497587312e-05} +{"train_loss": 0.013882072642445564, "grad_norm": 0.10025540739297867, "global_step": 13156, "epoch": 124, "lr": 8.554970092337822e-05} +{"train_loss": 0.01860222965478897, "grad_norm": 0.06747402995824814, "global_step": 13157, "epoch": 124, "lr": 8.554761674597712e-05} +{"train_loss": 0.017223473638296127, "grad_norm": 0.05690951272845268, "global_step": 13158, "epoch": 124, "lr": 8.554553244367716e-05} +{"train_loss": 0.016384771093726158, "grad_norm": 0.07265632599592209, "global_step": 13159, "epoch": 124, "lr": 8.554344801648566e-05} +{"train_loss": 0.016075020655989647, "grad_norm": 0.06120041757822037, "global_step": 13160, "epoch": 124, "lr": 8.554136346440995e-05} +{"train_loss": 0.019408196210861206, "grad_norm": 0.09697354584932327, "global_step": 13161, "epoch": 124, "lr": 8.553927878745733e-05} +{"train_loss": 0.018041566014289856, "grad_norm": 0.08208046853542328, "global_step": 13162, "epoch": 124, "lr": 8.553719398563515e-05} +{"train_loss": 0.016397688537836075, "grad_norm": 0.07320388406515121, "global_step": 13163, "epoch": 124, "lr": 8.553510905895073e-05} +{"train_loss": 0.018980486318469048, "grad_norm": 0.09689575433731079, "global_step": 13164, "epoch": 124, "lr": 8.553302400741137e-05} +{"train_loss": 0.018575426191091537, "grad_norm": 0.05384752154350281, "global_step": 13165, "epoch": 124, "lr": 8.553093883102443e-05} +{"train_loss": 0.016914047300815582, "grad_norm": 0.07872579991817474, "global_step": 13166, "epoch": 124, "lr": 8.552885352979723e-05} +{"train_loss": 0.014576910063624382, "grad_norm": 0.07826472818851471, "global_step": 13167, "epoch": 124, "lr": 8.552676810373708e-05} +{"train_loss": 0.018819663673639297, "grad_norm": 0.05094124376773834, "global_step": 13168, "epoch": 124, "lr": 8.55246825528513e-05} +{"train_loss": 0.015130538493394852, "grad_norm": 0.06809528917074203, "global_step": 13169, "epoch": 124, "lr": 8.552259687714727e-05} +{"train_loss": 0.01681400090456009, "grad_norm": 0.06485329568386078, "global_step": 13170, "epoch": 124, "lr": 8.552051107663226e-05} +{"train_loss": 0.016328107565641403, "grad_norm": 0.16485336422920227, "global_step": 13171, "epoch": 124, "lr": 8.551842515131363e-05} +{"train_loss": 0.01706468127667904, "grad_norm": 0.08951383829116821, "global_step": 13172, "epoch": 124, "lr": 8.551633910119871e-05} +{"train_loss": 0.01542880479246378, "grad_norm": 0.058717913925647736, "global_step": 13173, "epoch": 124, "lr": 8.55142529262948e-05} +{"train_loss": 0.01777717098593712, "grad_norm": 0.11231028288602829, "global_step": 13174, "epoch": 124, "lr": 8.551216662660926e-05} +{"train_loss": 0.01772645115852356, "grad_norm": 0.08295945078134537, "global_step": 13175, "epoch": 124, "lr": 8.55100802021494e-05} +{"train_loss": 0.0204248558729887, "grad_norm": 0.10527677834033966, "global_step": 13176, "epoch": 124, "lr": 8.550799365292256e-05} +{"train_loss": 0.01791262999176979, "grad_norm": 0.1651323437690735, "global_step": 13177, "epoch": 124, "lr": 8.55059069789361e-05} +{"train_loss": 0.016589682549238205, "grad_norm": 0.09406766295433044, "global_step": 13178, "epoch": 124, "lr": 8.550382018019732e-05} +{"train_loss": 0.019049787893891335, "grad_norm": 0.07460872828960419, "global_step": 13179, "epoch": 124, "lr": 8.550173325671353e-05} +{"train_loss": 0.018633529543876648, "grad_norm": 0.0770832821726799, "global_step": 13180, "epoch": 124, "lr": 8.549964620849209e-05} +{"train_loss": 0.01747322827577591, "grad_norm": 0.049747202545404434, "global_step": 13181, "epoch": 124, "lr": 8.549755903554035e-05} +{"train_loss": 0.0186782144010067, "grad_norm": 0.11953011155128479, "global_step": 13182, "epoch": 124, "lr": 8.54954717378656e-05} +{"train_loss": 0.017444457858800888, "grad_norm": 0.08517869561910629, "global_step": 13183, "epoch": 124, "lr": 8.549338431547524e-05} +{"train_loss": 0.015729453414678574, "grad_norm": 0.06073460727930069, "global_step": 13184, "epoch": 124, "lr": 8.549129676837654e-05} +{"train_loss": 0.016367726027965546, "grad_norm": 0.05068831145763397, "global_step": 13185, "epoch": 124, "lr": 8.548920909657685e-05} +{"train_loss": 0.019013691693544388, "grad_norm": 0.09993907809257507, "global_step": 13186, "epoch": 124, "lr": 8.548712130008352e-05} +{"train_loss": 0.01864805445075035, "grad_norm": 0.14449244737625122, "global_step": 13187, "epoch": 124, "lr": 8.548503337890387e-05} +{"train_loss": 0.015095868147909641, "grad_norm": 0.10376698523759842, "global_step": 13188, "epoch": 124, "lr": 8.548294533304526e-05} +{"train_loss": 0.01407682429999113, "grad_norm": 0.08811216801404953, "global_step": 13189, "epoch": 124, "lr": 8.548085716251501e-05} +{"train_loss": 0.01554807461798191, "grad_norm": 0.06581982970237732, "global_step": 13190, "epoch": 124, "lr": 8.547876886732046e-05} +{"train_loss": 0.01770690083503723, "grad_norm": 0.11971455812454224, "global_step": 13191, "epoch": 124, "lr": 8.547668044746894e-05} +{"train_loss": 0.016786610707640648, "grad_norm": 0.07955102622509003, "global_step": 13192, "epoch": 124, "lr": 8.547459190296778e-05} +{"train_loss": 0.01751621812582016, "grad_norm": 0.06468287855386734, "global_step": 13193, "epoch": 124, "lr": 8.547250323382435e-05} +{"train_loss": 0.01605462096631527, "grad_norm": 0.09226801991462708, "global_step": 13194, "epoch": 124, "lr": 8.547041444004597e-05} +{"train_loss": 0.018360815942287445, "grad_norm": 0.10519707947969437, "global_step": 13195, "epoch": 124, "lr": 8.546832552163997e-05} +{"train_loss": 0.017447348684072495, "grad_norm": 0.06321986019611359, "global_step": 13196, "epoch": 124, "lr": 8.54662364786137e-05} +{"train_loss": 0.015220070257782936, "grad_norm": 0.08073924481868744, "global_step": 13197, "epoch": 124, "lr": 8.54641473109745e-05} +{"train_loss": 0.016912413761019707, "grad_norm": 0.041490163654088974, "global_step": 13198, "epoch": 124, "lr": 8.54620580187297e-05} +{"train_loss": 0.01658201962709427, "grad_norm": 0.05928808078169823, "global_step": 13199, "epoch": 124, "lr": 8.545996860188668e-05} +{"train_loss": 0.016811039298772812, "grad_norm": 0.07111402601003647, "global_step": 13200, "epoch": 124, "lr": 8.545787906045273e-05} +{"train_loss": 0.018619049340486526, "grad_norm": 0.13846050202846527, "global_step": 13201, "epoch": 124, "lr": 8.545578939443521e-05} +{"train_loss": 0.022222047671675682, "grad_norm": 0.08395466953516006, "global_step": 13202, "epoch": 124, "lr": 8.545369960384146e-05} +{"train_loss": 0.01746407523751259, "grad_norm": 0.10751071572303772, "global_step": 13203, "epoch": 124, "lr": 8.545160968867885e-05} +{"train_loss": 0.018465831875801086, "grad_norm": 0.09176052361726761, "global_step": 13204, "epoch": 124, "lr": 8.544951964895469e-05} +{"train_loss": 0.017184657976031303, "grad_norm": 0.05145198851823807, "global_step": 13205, "epoch": 124, "lr": 8.544742948467632e-05} +{"train_loss": 0.01689022220671177, "grad_norm": 0.11943984031677246, "global_step": 13206, "epoch": 124, "lr": 8.544533919585113e-05} +{"train_loss": 0.015629757195711136, "grad_norm": 0.0478731207549572, "global_step": 13207, "epoch": 124, "lr": 8.544324878248641e-05} +{"train_loss": 0.018910374492406845, "grad_norm": 0.13694356381893158, "global_step": 13208, "epoch": 124, "lr": 8.544115824458953e-05} +{"train_loss": 0.019071737304329872, "grad_norm": 0.06551236659288406, "global_step": 13209, "epoch": 124, "lr": 8.543906758216784e-05} +{"train_loss": 0.01476012822240591, "grad_norm": 0.04420473426580429, "global_step": 13210, "epoch": 124, "lr": 8.543697679522867e-05} +{"train_loss": 0.014595395885407925, "grad_norm": 0.05427674949169159, "global_step": 13211, "epoch": 124, "lr": 8.543488588377937e-05} +{"train_loss": 0.015549051575362682, "grad_norm": 0.058632127940654755, "global_step": 13212, "epoch": 124, "lr": 8.543279484782732e-05} +{"train_loss": 0.016505220904946327, "grad_norm": 0.1611591875553131, "global_step": 13213, "epoch": 124, "lr": 8.543070368737982e-05} +{"train_loss": 0.01775967888534069, "grad_norm": 0.06490467488765717, "global_step": 13214, "epoch": 124, "lr": 8.542861240244422e-05} +{"train_loss": 0.01627221889793873, "grad_norm": 0.08746504038572311, "global_step": 13215, "epoch": 124, "lr": 8.54265209930279e-05} +{"train_loss": 0.020364370197057724, "grad_norm": 0.09035934507846832, "global_step": 13216, "epoch": 124, "lr": 8.54244294591382e-05} +{"train_loss": 0.015386058948934078, "grad_norm": 0.08422958850860596, "global_step": 13217, "epoch": 124, "lr": 8.542233780078243e-05} +{"train_loss": 0.018856406211853027, "grad_norm": 0.15506252646446228, "global_step": 13218, "epoch": 124, "lr": 8.5420246017968e-05} +{"train_loss": 0.018054174259305, "grad_norm": 0.05992569029331207, "global_step": 13219, "epoch": 124, "lr": 8.541815411070221e-05} +{"train_loss": 0.014727331697940826, "grad_norm": 0.050199005752801895, "global_step": 13220, "epoch": 124, "lr": 8.541606207899246e-05} +{"train_loss": 0.017400093376636505, "grad_norm": 0.11897902190685272, "global_step": 13221, "epoch": 124, "lr": 8.541396992284605e-05} +{"train_loss": 0.016605447977781296, "grad_norm": 0.04588083550333977, "global_step": 13222, "epoch": 124, "lr": 8.541187764227034e-05} +{"train_loss": 0.01690497249364853, "grad_norm": 0.04466128349304199, "global_step": 13223, "epoch": 124, "lr": 8.54097852372727e-05} +{"train_loss": 0.01728716492652893, "grad_norm": 0.10951139032840729, "global_step": 13224, "epoch": 124, "lr": 8.540769270786049e-05} +{"train_loss": 0.015206078998744488, "grad_norm": 0.046193912625312805, "global_step": 13225, "epoch": 124, "lr": 8.540560005404103e-05} +{"train_loss": 0.01413450576364994, "grad_norm": 0.1012750118970871, "global_step": 13226, "epoch": 124, "lr": 8.540350727582169e-05} +{"train_loss": 0.020015774294734, "grad_norm": 0.05508435145020485, "global_step": 13227, "epoch": 124, "lr": 8.540141437320981e-05} +{"train_loss": 0.01858518086373806, "grad_norm": 0.05002034455537796, "global_step": 13228, "epoch": 124, "lr": 8.539932134621278e-05} +{"train_loss": 0.01671297661960125, "grad_norm": 0.06237839534878731, "global_step": 13229, "epoch": 124, "lr": 8.539722819483791e-05} +{"train_loss": 0.016420088708400726, "grad_norm": 0.11908435821533203, "global_step": 13230, "epoch": 124, "lr": 8.539513491909259e-05} +{"train_loss": 0.01763552613556385, "grad_norm": 0.11826340854167938, "global_step": 13231, "epoch": 124, "lr": 8.539304151898415e-05} +{"train_loss": 0.018790582194924355, "grad_norm": 0.08694731444120407, "global_step": 13232, "epoch": 124, "lr": 8.539094799451994e-05} +{"train_loss": 0.014720196835696697, "grad_norm": 0.06617898494005203, "global_step": 13233, "epoch": 124, "lr": 8.538885434570736e-05} +{"train_loss": 0.021093670278787613, "grad_norm": 0.08022553473711014, "global_step": 13234, "epoch": 124, "lr": 8.538676057255372e-05} +{"train_loss": 0.01650281250476837, "grad_norm": 0.07911866158246994, "global_step": 13235, "epoch": 124, "lr": 8.53846666750664e-05} +{"train_loss": 0.013920342549681664, "grad_norm": 0.04342592880129814, "global_step": 13236, "epoch": 124, "lr": 8.538257265325273e-05} +{"train_loss": 0.016739165410399437, "grad_norm": 0.1353246122598648, "global_step": 13237, "epoch": 124, "lr": 8.53804785071201e-05} +{"train_loss": 0.015658367425203323, "grad_norm": 0.052602723240852356, "global_step": 13238, "epoch": 124, "lr": 8.537838423667585e-05} +{"train_loss": 0.015078584663569927, "grad_norm": 0.06915008276700974, "global_step": 13239, "epoch": 124, "lr": 8.537628984192736e-05} +{"train_loss": 0.016955051571130753, "grad_norm": 0.07144765555858612, "global_step": 13240, "epoch": 124, "lr": 8.537419532288196e-05} +{"train_loss": 0.016016770154237747, "grad_norm": 0.07346492260694504, "global_step": 13241, "epoch": 124, "lr": 8.537210067954702e-05} +{"train_loss": 0.016637323424220085, "grad_norm": 0.10159705579280853, "global_step": 13242, "epoch": 124, "lr": 8.53700059119299e-05} +{"train_loss": 0.017603209242224693, "grad_norm": 0.051904309540987015, "global_step": 13243, "epoch": 124, "lr": 8.536791102003796e-05} +{"train_loss": 0.01582079380750656, "grad_norm": 0.10252729803323746, "global_step": 13244, "epoch": 124, "lr": 8.536581600387857e-05} +{"train_loss": 0.0169810988008976, "grad_norm": 0.0909116268157959, "global_step": 13245, "epoch": 124, "lr": 8.536372086345907e-05} +{"train_loss": 0.0182275902479887, "grad_norm": 0.05353236198425293, "global_step": 13246, "epoch": 124, "lr": 8.536162559878686e-05} +{"train_loss": 0.016879526898264885, "grad_norm": 0.08195271342992783, "global_step": 13247, "epoch": 124, "lr": 8.535953020986926e-05} +{"train_loss": 0.01961572840809822, "grad_norm": 0.04663129523396492, "global_step": 13248, "epoch": 124, "lr": 8.535743469671365e-05} +{"train_loss": 0.017037352037457924, "grad_norm": 0.09048093110322952, "global_step": 13249, "epoch": 124, "lr": 8.535533905932738e-05, "val_loss": 0.018833139911293983} +{"train_loss": 0.017790738493204117, "grad_norm": 0.12446877360343933, "global_step": 13250, "epoch": 125, "lr": 8.535324329771784e-05} +{"train_loss": 0.01856575347483158, "grad_norm": 0.16976244747638702, "global_step": 13251, "epoch": 125, "lr": 8.535114741189236e-05} +{"train_loss": 0.01773615926504135, "grad_norm": 0.054359301924705505, "global_step": 13252, "epoch": 125, "lr": 8.534905140185834e-05} +{"train_loss": 0.015516847372055054, "grad_norm": 0.08293940871953964, "global_step": 13253, "epoch": 125, "lr": 8.534695526762311e-05} +{"train_loss": 0.02072383090853691, "grad_norm": 0.07645494490861893, "global_step": 13254, "epoch": 125, "lr": 8.534485900919407e-05} +{"train_loss": 0.01711905002593994, "grad_norm": 0.07133317738771439, "global_step": 13255, "epoch": 125, "lr": 8.534276262657856e-05} +{"train_loss": 0.017510943114757538, "grad_norm": 0.11449149250984192, "global_step": 13256, "epoch": 125, "lr": 8.534066611978394e-05} +{"train_loss": 0.016443580389022827, "grad_norm": 0.04969707876443863, "global_step": 13257, "epoch": 125, "lr": 8.53385694888176e-05} +{"train_loss": 0.016586489975452423, "grad_norm": 0.11305844783782959, "global_step": 13258, "epoch": 125, "lr": 8.53364727336869e-05} +{"train_loss": 0.017265621572732925, "grad_norm": 0.10348961502313614, "global_step": 13259, "epoch": 125, "lr": 8.533437585439918e-05} +{"train_loss": 0.016549978405237198, "grad_norm": 0.0850970521569252, "global_step": 13260, "epoch": 125, "lr": 8.533227885096185e-05} +{"train_loss": 0.016961120069026947, "grad_norm": 0.08728498965501785, "global_step": 13261, "epoch": 125, "lr": 8.533018172338226e-05} +{"train_loss": 0.017865367233753204, "grad_norm": 0.05067981034517288, "global_step": 13262, "epoch": 125, "lr": 8.532808447166776e-05} +{"train_loss": 0.02029820904135704, "grad_norm": 0.07607992738485336, "global_step": 13263, "epoch": 125, "lr": 8.532598709582575e-05} +{"train_loss": 0.015551755204796791, "grad_norm": 0.05624190345406532, "global_step": 13264, "epoch": 125, "lr": 8.532388959586358e-05} +{"train_loss": 0.018712472170591354, "grad_norm": 0.07532978057861328, "global_step": 13265, "epoch": 125, "lr": 8.532179197178861e-05} +{"train_loss": 0.0155206099152565, "grad_norm": 0.08443424105644226, "global_step": 13266, "epoch": 125, "lr": 8.531969422360823e-05} +{"train_loss": 0.014380792155861855, "grad_norm": 0.10221322625875473, "global_step": 13267, "epoch": 125, "lr": 8.531759635132982e-05} +{"train_loss": 0.015508085489273071, "grad_norm": 0.05927980691194534, "global_step": 13268, "epoch": 125, "lr": 8.531549835496072e-05} +{"train_loss": 0.019778799265623093, "grad_norm": 0.12286491692066193, "global_step": 13269, "epoch": 125, "lr": 8.531340023450834e-05} +{"train_loss": 0.013301976025104523, "grad_norm": 0.102834053337574, "global_step": 13270, "epoch": 125, "lr": 8.531130198998e-05} +{"train_loss": 0.016262216493487358, "grad_norm": 0.09927216917276382, "global_step": 13271, "epoch": 125, "lr": 8.530920362138311e-05} +{"train_loss": 0.01869838312268257, "grad_norm": 0.11172369867563248, "global_step": 13272, "epoch": 125, "lr": 8.530710512872504e-05} +{"train_loss": 0.015779392793774605, "grad_norm": 0.09985258430242538, "global_step": 13273, "epoch": 125, "lr": 8.530500651201315e-05} +{"train_loss": 0.018040407449007034, "grad_norm": 0.08283303678035736, "global_step": 13274, "epoch": 125, "lr": 8.530290777125483e-05} +{"train_loss": 0.01852967031300068, "grad_norm": 0.1195288896560669, "global_step": 13275, "epoch": 125, "lr": 8.530080890645744e-05} +{"train_loss": 0.017643213272094727, "grad_norm": 0.08766809850931168, "global_step": 13276, "epoch": 125, "lr": 8.529870991762837e-05} +{"train_loss": 0.01641913130879402, "grad_norm": 0.10128967463970184, "global_step": 13277, "epoch": 125, "lr": 8.529661080477497e-05} +{"train_loss": 0.01827423833310604, "grad_norm": 0.05447474494576454, "global_step": 13278, "epoch": 125, "lr": 8.529451156790463e-05} +{"train_loss": 0.018601316958665848, "grad_norm": 0.08224906027317047, "global_step": 13279, "epoch": 125, "lr": 8.529241220702474e-05} +{"train_loss": 0.016306240111589432, "grad_norm": 0.06185684725642204, "global_step": 13280, "epoch": 125, "lr": 8.529031272214263e-05} +{"train_loss": 0.01584896445274353, "grad_norm": 0.06627098470926285, "global_step": 13281, "epoch": 125, "lr": 8.528821311326574e-05} +{"train_loss": 0.015999387949705124, "grad_norm": 0.07385881245136261, "global_step": 13282, "epoch": 125, "lr": 8.52861133804014e-05} +{"train_loss": 0.014369962736964226, "grad_norm": 0.07188744097948074, "global_step": 13283, "epoch": 125, "lr": 8.528401352355702e-05} +{"train_loss": 0.016917787492275238, "grad_norm": 0.0872662141919136, "global_step": 13284, "epoch": 125, "lr": 8.528191354273994e-05} +{"train_loss": 0.016080733388662338, "grad_norm": 0.045622777193784714, "global_step": 13285, "epoch": 125, "lr": 8.527981343795757e-05} +{"train_loss": 0.017702728509902954, "grad_norm": 0.13908635079860687, "global_step": 13286, "epoch": 125, "lr": 8.527771320921727e-05} +{"train_loss": 0.018896708264946938, "grad_norm": 0.11779309064149857, "global_step": 13287, "epoch": 125, "lr": 8.527561285652643e-05} +{"train_loss": 0.016013253480196, "grad_norm": 0.06165293604135513, "global_step": 13288, "epoch": 125, "lr": 8.527351237989244e-05} +{"train_loss": 0.018199782818555832, "grad_norm": 0.08787688612937927, "global_step": 13289, "epoch": 125, "lr": 8.527141177932267e-05} +{"train_loss": 0.01495665404945612, "grad_norm": 0.055225174874067307, "global_step": 13290, "epoch": 125, "lr": 8.526931105482448e-05} +{"train_loss": 0.017079338431358337, "grad_norm": 0.05823894962668419, "global_step": 13291, "epoch": 125, "lr": 8.526721020640527e-05} +{"train_loss": 0.020239029079675674, "grad_norm": 0.12781968712806702, "global_step": 13292, "epoch": 125, "lr": 8.526510923407242e-05} +{"train_loss": 0.019262712448835373, "grad_norm": 0.076337069272995, "global_step": 13293, "epoch": 125, "lr": 8.526300813783333e-05} +{"train_loss": 0.016863910481333733, "grad_norm": 0.09898039698600769, "global_step": 13294, "epoch": 125, "lr": 8.526090691769535e-05} +{"train_loss": 0.018895499408245087, "grad_norm": 0.058427222073078156, "global_step": 13295, "epoch": 125, "lr": 8.52588055736659e-05} +{"train_loss": 0.01624818705022335, "grad_norm": 0.05452452227473259, "global_step": 13296, "epoch": 125, "lr": 8.525670410575231e-05} +{"train_loss": 0.01796942576766014, "grad_norm": 0.13552448153495789, "global_step": 13297, "epoch": 125, "lr": 8.5254602513962e-05} +{"train_loss": 0.015955103561282158, "grad_norm": 0.0700794979929924, "global_step": 13298, "epoch": 125, "lr": 8.525250079830235e-05} +{"train_loss": 0.018137916922569275, "grad_norm": 0.07208427041769028, "global_step": 13299, "epoch": 125, "lr": 8.525039895878078e-05} +{"train_loss": 0.019197635352611542, "grad_norm": 0.10134554654359818, "global_step": 13300, "epoch": 125, "lr": 8.52482969954046e-05} +{"train_loss": 0.019380757585167885, "grad_norm": 0.07934366911649704, "global_step": 13301, "epoch": 125, "lr": 8.524619490818125e-05} +{"train_loss": 0.017940662801265717, "grad_norm": 0.06961570680141449, "global_step": 13302, "epoch": 125, "lr": 8.524409269711807e-05} +{"train_loss": 0.014301993884146214, "grad_norm": 0.06630823016166687, "global_step": 13303, "epoch": 125, "lr": 8.524199036222251e-05} +{"train_loss": 0.017516184598207474, "grad_norm": 0.09368295967578888, "global_step": 13304, "epoch": 125, "lr": 8.52398879035019e-05} +{"train_loss": 0.017817510291934013, "grad_norm": 0.10375810414552689, "global_step": 13305, "epoch": 125, "lr": 8.523778532096367e-05} +{"train_loss": 0.015872925519943237, "grad_norm": 0.04592270776629448, "global_step": 13306, "epoch": 125, "lr": 8.523568261461516e-05} +{"train_loss": 0.014592870138585567, "grad_norm": 0.09522589296102524, "global_step": 13307, "epoch": 125, "lr": 8.523357978446381e-05} +{"train_loss": 0.01663418672978878, "grad_norm": 0.052821796387434006, "global_step": 13308, "epoch": 125, "lr": 8.523147683051695e-05} +{"train_loss": 0.018152251839637756, "grad_norm": 0.07543638348579407, "global_step": 13309, "epoch": 125, "lr": 8.522937375278204e-05} +{"train_loss": 0.015918496996164322, "grad_norm": 0.06086926534771919, "global_step": 13310, "epoch": 125, "lr": 8.522727055126641e-05} +{"train_loss": 0.015087200328707695, "grad_norm": 0.07741004228591919, "global_step": 13311, "epoch": 125, "lr": 8.522516722597749e-05} +{"train_loss": 0.01833941973745823, "grad_norm": 0.07873926311731339, "global_step": 13312, "epoch": 125, "lr": 8.522306377692263e-05} +{"train_loss": 0.01519377063959837, "grad_norm": 0.08858322352170944, "global_step": 13313, "epoch": 125, "lr": 8.522096020410926e-05} +{"train_loss": 0.01663818582892418, "grad_norm": 0.06174543872475624, "global_step": 13314, "epoch": 125, "lr": 8.521885650754474e-05} +{"train_loss": 0.014203113503754139, "grad_norm": 0.07751143723726273, "global_step": 13315, "epoch": 125, "lr": 8.521675268723648e-05} +{"train_loss": 0.018077753484249115, "grad_norm": 0.07617715746164322, "global_step": 13316, "epoch": 125, "lr": 8.521464874319186e-05} +{"train_loss": 0.01569024845957756, "grad_norm": 0.052704907953739166, "global_step": 13317, "epoch": 125, "lr": 8.521254467541828e-05} +{"train_loss": 0.013473321683704853, "grad_norm": 0.0762057974934578, "global_step": 13318, "epoch": 125, "lr": 8.521044048392313e-05} +{"train_loss": 0.016463764011859894, "grad_norm": 0.051095958799123764, "global_step": 13319, "epoch": 125, "lr": 8.520833616871383e-05} +{"train_loss": 0.015726223587989807, "grad_norm": 0.09243744611740112, "global_step": 13320, "epoch": 125, "lr": 8.520623172979771e-05} +{"train_loss": 0.01688145473599434, "grad_norm": 0.10728191584348679, "global_step": 13321, "epoch": 125, "lr": 8.520412716718223e-05} +{"train_loss": 0.01689295656979084, "grad_norm": 0.06887827813625336, "global_step": 13322, "epoch": 125, "lr": 8.520202248087475e-05} +{"train_loss": 0.015229595825076103, "grad_norm": 0.05039821192622185, "global_step": 13323, "epoch": 125, "lr": 8.519991767088266e-05} +{"train_loss": 0.018513433635234833, "grad_norm": 0.060262441635131836, "global_step": 13324, "epoch": 125, "lr": 8.519781273721336e-05} +{"train_loss": 0.01736355572938919, "grad_norm": 0.0654812604188919, "global_step": 13325, "epoch": 125, "lr": 8.51957076798743e-05} +{"train_loss": 0.019007563591003418, "grad_norm": 0.09524884819984436, "global_step": 13326, "epoch": 125, "lr": 8.519360249887278e-05} +{"train_loss": 0.016144035384058952, "grad_norm": 0.12714961171150208, "global_step": 13327, "epoch": 125, "lr": 8.519149719421628e-05} +{"train_loss": 0.018704790621995926, "grad_norm": 0.05256713554263115, "global_step": 13328, "epoch": 125, "lr": 8.518939176591214e-05} +{"train_loss": 0.019599897786974907, "grad_norm": 0.0817892774939537, "global_step": 13329, "epoch": 125, "lr": 8.518728621396779e-05} +{"train_loss": 0.015991084277629852, "grad_norm": 0.050365373492240906, "global_step": 13330, "epoch": 125, "lr": 8.518518053839063e-05} +{"train_loss": 0.016017481684684753, "grad_norm": 0.0926075279712677, "global_step": 13331, "epoch": 125, "lr": 8.518307473918802e-05} +{"train_loss": 0.016456443816423416, "grad_norm": 0.04967544600367546, "global_step": 13332, "epoch": 125, "lr": 8.518096881636742e-05} +{"train_loss": 0.01673516444861889, "grad_norm": 0.04068407043814659, "global_step": 13333, "epoch": 125, "lr": 8.517886276993616e-05} +{"train_loss": 0.01731184497475624, "grad_norm": 0.0654618963599205, "global_step": 13334, "epoch": 125, "lr": 8.51767565999017e-05} +{"train_loss": 0.015608621761202812, "grad_norm": 0.1064954474568367, "global_step": 13335, "epoch": 125, "lr": 8.51746503062714e-05} +{"train_loss": 0.018294882029294968, "grad_norm": 0.07064328342676163, "global_step": 13336, "epoch": 125, "lr": 8.51725438890527e-05} +{"train_loss": 0.016954386606812477, "grad_norm": 0.07621115446090698, "global_step": 13337, "epoch": 125, "lr": 8.517043734825295e-05} +{"train_loss": 0.018596062436699867, "grad_norm": 0.05163956061005592, "global_step": 13338, "epoch": 125, "lr": 8.51683306838796e-05} +{"train_loss": 0.017269713804125786, "grad_norm": 0.045474156737327576, "global_step": 13339, "epoch": 125, "lr": 8.516622389594001e-05} +{"train_loss": 0.016930188983678818, "grad_norm": 0.09699717164039612, "global_step": 13340, "epoch": 125, "lr": 8.516411698444161e-05} +{"train_loss": 0.01601603254675865, "grad_norm": 0.0696847215294838, "global_step": 13341, "epoch": 125, "lr": 8.51620099493918e-05} +{"train_loss": 0.016650330275297165, "grad_norm": 0.05910171568393707, "global_step": 13342, "epoch": 125, "lr": 8.515990279079797e-05} +{"train_loss": 0.0165710486471653, "grad_norm": 0.05721621215343475, "global_step": 13343, "epoch": 125, "lr": 8.515779550866754e-05} +{"train_loss": 0.01492173783481121, "grad_norm": 0.06577270478010178, "global_step": 13344, "epoch": 125, "lr": 8.51556881030079e-05} +{"train_loss": 0.01700543239712715, "grad_norm": 0.08852961659431458, "global_step": 13345, "epoch": 125, "lr": 8.515358057382647e-05} +{"train_loss": 0.01980973780155182, "grad_norm": 0.04565191641449928, "global_step": 13346, "epoch": 125, "lr": 8.515147292113064e-05} +{"train_loss": 0.018015621230006218, "grad_norm": 0.0575304739177227, "global_step": 13347, "epoch": 125, "lr": 8.514936514492781e-05} +{"train_loss": 0.01860285922884941, "grad_norm": 0.058828387409448624, "global_step": 13348, "epoch": 125, "lr": 8.514725724522542e-05} +{"train_loss": 0.017008747905492783, "grad_norm": 0.04780484735965729, "global_step": 13349, "epoch": 125, "lr": 8.514514922203084e-05} +{"train_loss": 0.014093192294239998, "grad_norm": 0.053000155836343765, "global_step": 13350, "epoch": 125, "lr": 8.514304107535149e-05} +{"train_loss": 0.018906140699982643, "grad_norm": 0.06420952826738358, "global_step": 13351, "epoch": 125, "lr": 8.514093280519478e-05} +{"train_loss": 0.016978934407234192, "grad_norm": 0.05209002271294594, "global_step": 13352, "epoch": 125, "lr": 8.513882441156812e-05} +{"train_loss": 0.017792552709579468, "grad_norm": 0.14610867202281952, "global_step": 13353, "epoch": 125, "lr": 8.513671589447889e-05} +{"train_loss": 0.014803066849708557, "grad_norm": 0.05135750025510788, "global_step": 13354, "epoch": 125, "lr": 8.513460725393456e-05} +{"train_loss": 0.01700666324534225, "grad_norm": 0.08173646032810211, "global_step": 13355, "epoch": 125, "lr": 8.513249848994246e-05, "val_loss": 0.018631093204021454} +{"train_loss": 0.016048017889261246, "grad_norm": 0.053969986736774445, "global_step": 13356, "epoch": 126, "lr": 8.513038960251005e-05} +{"train_loss": 0.01672825962305069, "grad_norm": 0.06289047002792358, "global_step": 13357, "epoch": 126, "lr": 8.512828059164474e-05} +{"train_loss": 0.016683515161275864, "grad_norm": 0.0525759682059288, "global_step": 13358, "epoch": 126, "lr": 8.512617145735393e-05} +{"train_loss": 0.01997605338692665, "grad_norm": 0.09549877792596817, "global_step": 13359, "epoch": 126, "lr": 8.512406219964502e-05} +{"train_loss": 0.014170490205287933, "grad_norm": 0.07385992258787155, "global_step": 13360, "epoch": 126, "lr": 8.512195281852544e-05} +{"train_loss": 0.018286865204572678, "grad_norm": 0.05593346431851387, "global_step": 13361, "epoch": 126, "lr": 8.511984331400258e-05} +{"train_loss": 0.015858471393585205, "grad_norm": 0.06881077587604523, "global_step": 13362, "epoch": 126, "lr": 8.511773368608387e-05} +{"train_loss": 0.01822901889681816, "grad_norm": 0.07657522708177567, "global_step": 13363, "epoch": 126, "lr": 8.511562393477672e-05} +{"train_loss": 0.015545235015451908, "grad_norm": 0.06760404258966446, "global_step": 13364, "epoch": 126, "lr": 8.511351406008852e-05} +{"train_loss": 0.017004072666168213, "grad_norm": 0.09635496884584427, "global_step": 13365, "epoch": 126, "lr": 8.511140406202671e-05} +{"train_loss": 0.018392350524663925, "grad_norm": 0.04774249717593193, "global_step": 13366, "epoch": 126, "lr": 8.510929394059871e-05} +{"train_loss": 0.01761533133685589, "grad_norm": 0.09114474803209305, "global_step": 13367, "epoch": 126, "lr": 8.510718369581192e-05} +{"train_loss": 0.015341667458415031, "grad_norm": 0.08830253779888153, "global_step": 13368, "epoch": 126, "lr": 8.510507332767375e-05} +{"train_loss": 0.015078479424118996, "grad_norm": 0.07423358410596848, "global_step": 13369, "epoch": 126, "lr": 8.510296283619161e-05} +{"train_loss": 0.01819937862455845, "grad_norm": 0.049999937415122986, "global_step": 13370, "epoch": 126, "lr": 8.510085222137292e-05} +{"train_loss": 0.018327821046113968, "grad_norm": 0.056771185249090195, "global_step": 13371, "epoch": 126, "lr": 8.509874148322511e-05} +{"train_loss": 0.016642460599541664, "grad_norm": 0.05963820219039917, "global_step": 13372, "epoch": 126, "lr": 8.50966306217556e-05} +{"train_loss": 0.01950250193476677, "grad_norm": 0.05803993344306946, "global_step": 13373, "epoch": 126, "lr": 8.509451963697177e-05} +{"train_loss": 0.013916019350290298, "grad_norm": 0.04515328258275986, "global_step": 13374, "epoch": 126, "lr": 8.509240852888106e-05} +{"train_loss": 0.018375258892774582, "grad_norm": 0.06097022071480751, "global_step": 13375, "epoch": 126, "lr": 8.50902972974909e-05} +{"train_loss": 0.01543642207980156, "grad_norm": 0.054886527359485626, "global_step": 13376, "epoch": 126, "lr": 8.50881859428087e-05} +{"train_loss": 0.015986260026693344, "grad_norm": 0.1024322658777237, "global_step": 13377, "epoch": 126, "lr": 8.508607446484186e-05} +{"train_loss": 0.01659369096159935, "grad_norm": 0.04657240957021713, "global_step": 13378, "epoch": 126, "lr": 8.508396286359783e-05} +{"train_loss": 0.016684330999851227, "grad_norm": 0.06480967998504639, "global_step": 13379, "epoch": 126, "lr": 8.508185113908398e-05} +{"train_loss": 0.016279924660921097, "grad_norm": 0.04527625814080238, "global_step": 13380, "epoch": 126, "lr": 8.507973929130781e-05} +{"train_loss": 0.017540592700242996, "grad_norm": 0.05965523421764374, "global_step": 13381, "epoch": 126, "lr": 8.507762732027666e-05} +{"train_loss": 0.019671570509672165, "grad_norm": 0.06921613216400146, "global_step": 13382, "epoch": 126, "lr": 8.507551522599797e-05} +{"train_loss": 0.017389465123414993, "grad_norm": 0.09492021799087524, "global_step": 13383, "epoch": 126, "lr": 8.50734030084792e-05} +{"train_loss": 0.016841616481542587, "grad_norm": 0.04836903512477875, "global_step": 13384, "epoch": 126, "lr": 8.507129066772773e-05} +{"train_loss": 0.018708571791648865, "grad_norm": 0.06870520114898682, "global_step": 13385, "epoch": 126, "lr": 8.506917820375101e-05} +{"train_loss": 0.015460800379514694, "grad_norm": 0.05250628665089607, "global_step": 13386, "epoch": 126, "lr": 8.506706561655644e-05} +{"train_loss": 0.016686836257576942, "grad_norm": 0.05200802907347679, "global_step": 13387, "epoch": 126, "lr": 8.506495290615146e-05} +{"train_loss": 0.01689833402633667, "grad_norm": 0.05382690951228142, "global_step": 13388, "epoch": 126, "lr": 8.506284007254348e-05} +{"train_loss": 0.01856369897723198, "grad_norm": 0.10184454172849655, "global_step": 13389, "epoch": 126, "lr": 8.506072711573991e-05} +{"train_loss": 0.012670012190937996, "grad_norm": 0.04557947814464569, "global_step": 13390, "epoch": 126, "lr": 8.505861403574822e-05} +{"train_loss": 0.014940645545721054, "grad_norm": 0.04852573946118355, "global_step": 13391, "epoch": 126, "lr": 8.505650083257579e-05} +{"train_loss": 0.016538895666599274, "grad_norm": 0.05736006796360016, "global_step": 13392, "epoch": 126, "lr": 8.505438750623007e-05} +{"train_loss": 0.01855659857392311, "grad_norm": 0.06454269587993622, "global_step": 13393, "epoch": 126, "lr": 8.505227405671848e-05} +{"train_loss": 0.0170159749686718, "grad_norm": 0.13985629379749298, "global_step": 13394, "epoch": 126, "lr": 8.505016048404843e-05} +{"train_loss": 0.016620654612779617, "grad_norm": 0.10311305522918701, "global_step": 13395, "epoch": 126, "lr": 8.504804678822738e-05} +{"train_loss": 0.016058973968029022, "grad_norm": 0.051891788840293884, "global_step": 13396, "epoch": 126, "lr": 8.504593296926271e-05} +{"train_loss": 0.016282351687550545, "grad_norm": 0.07365545630455017, "global_step": 13397, "epoch": 126, "lr": 8.504381902716188e-05} +{"train_loss": 0.01824026182293892, "grad_norm": 0.08703343570232391, "global_step": 13398, "epoch": 126, "lr": 8.504170496193232e-05} +{"train_loss": 0.018592465668916702, "grad_norm": 0.08099780976772308, "global_step": 13399, "epoch": 126, "lr": 8.503959077358143e-05} +{"train_loss": 0.016896316781640053, "grad_norm": 0.06678877025842667, "global_step": 13400, "epoch": 126, "lr": 8.503747646211669e-05} +{"train_loss": 0.018363002687692642, "grad_norm": 0.09895304590463638, "global_step": 13401, "epoch": 126, "lr": 8.503536202754546e-05} +{"train_loss": 0.019939230754971504, "grad_norm": 0.08349534869194031, "global_step": 13402, "epoch": 126, "lr": 8.503324746987522e-05} +{"train_loss": 0.017566803842782974, "grad_norm": 0.05786871165037155, "global_step": 13403, "epoch": 126, "lr": 8.503113278911338e-05} +{"train_loss": 0.015259156934916973, "grad_norm": 0.06059088185429573, "global_step": 13404, "epoch": 126, "lr": 8.502901798526738e-05} +{"train_loss": 0.014531135559082031, "grad_norm": 0.06805159151554108, "global_step": 13405, "epoch": 126, "lr": 8.502690305834464e-05} +{"train_loss": 0.019500965252518654, "grad_norm": 0.07641591876745224, "global_step": 13406, "epoch": 126, "lr": 8.50247880083526e-05} +{"train_loss": 0.01636432856321335, "grad_norm": 0.12838298082351685, "global_step": 13407, "epoch": 126, "lr": 8.502267283529868e-05} +{"train_loss": 0.017449626699090004, "grad_norm": 0.08183340728282928, "global_step": 13408, "epoch": 126, "lr": 8.502055753919032e-05} +{"train_loss": 0.015293245203793049, "grad_norm": 0.09017004817724228, "global_step": 13409, "epoch": 126, "lr": 8.501844212003496e-05} +{"train_loss": 0.01592627912759781, "grad_norm": 0.07234882563352585, "global_step": 13410, "epoch": 126, "lr": 8.501632657784001e-05} +{"train_loss": 0.020049098879098892, "grad_norm": 0.06281106919050217, "global_step": 13411, "epoch": 126, "lr": 8.501421091261292e-05} +{"train_loss": 0.018883105367422104, "grad_norm": 0.08029697835445404, "global_step": 13412, "epoch": 126, "lr": 8.501209512436113e-05} +{"train_loss": 0.017163336277008057, "grad_norm": 0.06919530779123306, "global_step": 13413, "epoch": 126, "lr": 8.500997921309207e-05} +{"train_loss": 0.0195765420794487, "grad_norm": 0.15168605744838715, "global_step": 13414, "epoch": 126, "lr": 8.500786317881314e-05} +{"train_loss": 0.018483776599168777, "grad_norm": 0.09141577780246735, "global_step": 13415, "epoch": 126, "lr": 8.500574702153181e-05} +{"train_loss": 0.017191149294376373, "grad_norm": 0.08005557209253311, "global_step": 13416, "epoch": 126, "lr": 8.500363074125553e-05} +{"train_loss": 0.019534524530172348, "grad_norm": 0.0646468847990036, "global_step": 13417, "epoch": 126, "lr": 8.50015143379917e-05} +{"train_loss": 0.019663915038108826, "grad_norm": 0.0642070323228836, "global_step": 13418, "epoch": 126, "lr": 8.499939781174777e-05} +{"train_loss": 0.012978538870811462, "grad_norm": 0.06678293645381927, "global_step": 13419, "epoch": 126, "lr": 8.499728116253118e-05} +{"train_loss": 0.01842460222542286, "grad_norm": 0.10081735253334045, "global_step": 13420, "epoch": 126, "lr": 8.499516439034936e-05} +{"train_loss": 0.014997554942965508, "grad_norm": 0.046811122447252274, "global_step": 13421, "epoch": 126, "lr": 8.499304749520976e-05} +{"train_loss": 0.016185956075787544, "grad_norm": 0.10070564597845078, "global_step": 13422, "epoch": 126, "lr": 8.499093047711979e-05} +{"train_loss": 0.015043973922729492, "grad_norm": 0.07909934967756271, "global_step": 13423, "epoch": 126, "lr": 8.498881333608692e-05} +{"train_loss": 0.017816126346588135, "grad_norm": 0.06392029672861099, "global_step": 13424, "epoch": 126, "lr": 8.498669607211858e-05} +{"train_loss": 0.016110967844724655, "grad_norm": 0.09168773144483566, "global_step": 13425, "epoch": 126, "lr": 8.49845786852222e-05} +{"train_loss": 0.01573401503264904, "grad_norm": 0.0651220977306366, "global_step": 13426, "epoch": 126, "lr": 8.498246117540523e-05} +{"train_loss": 0.018045123666524887, "grad_norm": 0.07512740790843964, "global_step": 13427, "epoch": 126, "lr": 8.498034354267512e-05} +{"train_loss": 0.015947245061397552, "grad_norm": 0.11689615249633789, "global_step": 13428, "epoch": 126, "lr": 8.497822578703926e-05} +{"train_loss": 0.0164625346660614, "grad_norm": 0.07421622425317764, "global_step": 13429, "epoch": 126, "lr": 8.497610790850514e-05} +{"train_loss": 0.015500390902161598, "grad_norm": 0.05215569585561752, "global_step": 13430, "epoch": 126, "lr": 8.49739899070802e-05} +{"train_loss": 0.01630311831831932, "grad_norm": 0.10171521455049515, "global_step": 13431, "epoch": 126, "lr": 8.497187178277186e-05} +{"train_loss": 0.017206206917762756, "grad_norm": 0.12626057863235474, "global_step": 13432, "epoch": 126, "lr": 8.496975353558757e-05} +{"train_loss": 0.019799016416072845, "grad_norm": 0.12164358794689178, "global_step": 13433, "epoch": 126, "lr": 8.49676351655348e-05} +{"train_loss": 0.016604149714112282, "grad_norm": 0.06061803922057152, "global_step": 13434, "epoch": 126, "lr": 8.496551667262092e-05} +{"train_loss": 0.017293132841587067, "grad_norm": 0.0638357549905777, "global_step": 13435, "epoch": 126, "lr": 8.496339805685344e-05} +{"train_loss": 0.015868261456489563, "grad_norm": 0.06669499725103378, "global_step": 13436, "epoch": 126, "lr": 8.496127931823978e-05} +{"train_loss": 0.016001977026462555, "grad_norm": 0.06400972604751587, "global_step": 13437, "epoch": 126, "lr": 8.49591604567874e-05} +{"train_loss": 0.01641715317964554, "grad_norm": 0.09704943746328354, "global_step": 13438, "epoch": 126, "lr": 8.495704147250374e-05} +{"train_loss": 0.015360974706709385, "grad_norm": 0.09974349290132523, "global_step": 13439, "epoch": 126, "lr": 8.495492236539623e-05} +{"train_loss": 0.01778068020939827, "grad_norm": 0.04716014862060547, "global_step": 13440, "epoch": 126, "lr": 8.495280313547232e-05} +{"train_loss": 0.015138667076826096, "grad_norm": 0.04238565266132355, "global_step": 13441, "epoch": 126, "lr": 8.495068378273945e-05} +{"train_loss": 0.017991023138165474, "grad_norm": 0.056597743183374405, "global_step": 13442, "epoch": 126, "lr": 8.49485643072051e-05} +{"train_loss": 0.01667320355772972, "grad_norm": 0.054943863302469254, "global_step": 13443, "epoch": 126, "lr": 8.494644470887668e-05} +{"train_loss": 0.017893461510539055, "grad_norm": 0.05583519861102104, "global_step": 13444, "epoch": 126, "lr": 8.494432498776166e-05} +{"train_loss": 0.01825578138232231, "grad_norm": 0.09225699305534363, "global_step": 13445, "epoch": 126, "lr": 8.494220514386748e-05} +{"train_loss": 0.014469613321125507, "grad_norm": 0.06354611366987228, "global_step": 13446, "epoch": 126, "lr": 8.494008517720158e-05} +{"train_loss": 0.015651769936084747, "grad_norm": 0.1273326426744461, "global_step": 13447, "epoch": 126, "lr": 8.493796508777141e-05} +{"train_loss": 0.01330795232206583, "grad_norm": 0.18900172412395477, "global_step": 13448, "epoch": 126, "lr": 8.493584487558444e-05} +{"train_loss": 0.016620030626654625, "grad_norm": 0.16854998469352722, "global_step": 13449, "epoch": 126, "lr": 8.493372454064808e-05} +{"train_loss": 0.016990074887871742, "grad_norm": 0.09904977679252625, "global_step": 13450, "epoch": 126, "lr": 8.493160408296983e-05} +{"train_loss": 0.017363213002681732, "grad_norm": 0.06149549409747124, "global_step": 13451, "epoch": 126, "lr": 8.492948350255711e-05} +{"train_loss": 0.01642969623208046, "grad_norm": 0.056456420570611954, "global_step": 13452, "epoch": 126, "lr": 8.492736279941735e-05} +{"train_loss": 0.017728399485349655, "grad_norm": 0.073405422270298, "global_step": 13453, "epoch": 126, "lr": 8.492524197355806e-05} +{"train_loss": 0.015950942412018776, "grad_norm": 0.06872207671403885, "global_step": 13454, "epoch": 126, "lr": 8.492312102498664e-05} +{"train_loss": 0.01849474012851715, "grad_norm": 0.07552990317344666, "global_step": 13455, "epoch": 126, "lr": 8.492099995371057e-05} +{"train_loss": 0.017944959923624992, "grad_norm": 0.08953574299812317, "global_step": 13456, "epoch": 126, "lr": 8.491887875973729e-05} +{"train_loss": 0.018349166959524155, "grad_norm": 0.12473400682210922, "global_step": 13457, "epoch": 126, "lr": 8.491675744307426e-05} +{"train_loss": 0.015845797955989838, "grad_norm": 0.0811111330986023, "global_step": 13458, "epoch": 126, "lr": 8.491463600372891e-05} +{"train_loss": 0.01662977784872055, "grad_norm": 0.0834355279803276, "global_step": 13459, "epoch": 126, "lr": 8.491251444170873e-05} +{"train_loss": 0.020143946632742882, "grad_norm": 0.08914899826049805, "global_step": 13460, "epoch": 126, "lr": 8.491039275702115e-05} +{"train_loss": 0.016994300315964897, "grad_norm": 0.08723443001508713, "global_step": 13461, "epoch": 126, "lr": 8.490827094967363e-05, "val_loss": 0.019139491021633148} +{"train_loss": 0.014078909531235695, "grad_norm": 0.04856070876121521, "global_step": 13462, "epoch": 127, "lr": 8.490614901967364e-05} +{"train_loss": 0.017399495467543602, "grad_norm": 0.08763761818408966, "global_step": 13463, "epoch": 127, "lr": 8.490402696702861e-05} +{"train_loss": 0.014184406027197838, "grad_norm": 0.11550670862197876, "global_step": 13464, "epoch": 127, "lr": 8.490190479174602e-05} +{"train_loss": 0.021112728863954544, "grad_norm": 0.07118471711874008, "global_step": 13465, "epoch": 127, "lr": 8.489978249383331e-05} +{"train_loss": 0.01574650965631008, "grad_norm": 0.059035565704107285, "global_step": 13466, "epoch": 127, "lr": 8.489766007329795e-05} +{"train_loss": 0.016777489334344864, "grad_norm": 0.09068147093057632, "global_step": 13467, "epoch": 127, "lr": 8.489553753014737e-05} +{"train_loss": 0.015344370156526566, "grad_norm": 0.11403099447488785, "global_step": 13468, "epoch": 127, "lr": 8.489341486438906e-05} +{"train_loss": 0.015612972900271416, "grad_norm": 0.08982618153095245, "global_step": 13469, "epoch": 127, "lr": 8.489129207603046e-05} +{"train_loss": 0.016647130250930786, "grad_norm": 0.1107732355594635, "global_step": 13470, "epoch": 127, "lr": 8.488916916507903e-05} +{"train_loss": 0.01597769744694233, "grad_norm": 0.05086556822061539, "global_step": 13471, "epoch": 127, "lr": 8.488704613154223e-05} +{"train_loss": 0.019570760428905487, "grad_norm": 0.10962233692407608, "global_step": 13472, "epoch": 127, "lr": 8.488492297542754e-05} +{"train_loss": 0.017810210585594177, "grad_norm": 0.05442984774708748, "global_step": 13473, "epoch": 127, "lr": 8.48827996967424e-05} +{"train_loss": 0.01718287728726864, "grad_norm": 0.13127149641513824, "global_step": 13474, "epoch": 127, "lr": 8.488067629549426e-05} +{"train_loss": 0.016476862132549286, "grad_norm": 0.07336331158876419, "global_step": 13475, "epoch": 127, "lr": 8.487855277169059e-05} +{"train_loss": 0.01848759688436985, "grad_norm": 0.10550861805677414, "global_step": 13476, "epoch": 127, "lr": 8.487642912533887e-05} +{"train_loss": 0.015880020335316658, "grad_norm": 0.1464894413948059, "global_step": 13477, "epoch": 127, "lr": 8.487430535644652e-05} +{"train_loss": 0.016799498349428177, "grad_norm": 0.06048799678683281, "global_step": 13478, "epoch": 127, "lr": 8.487218146502105e-05} +{"train_loss": 0.018801718950271606, "grad_norm": 0.18565720319747925, "global_step": 13479, "epoch": 127, "lr": 8.487005745106988e-05} +{"train_loss": 0.01622481644153595, "grad_norm": 0.06601770222187042, "global_step": 13480, "epoch": 127, "lr": 8.486793331460052e-05} +{"train_loss": 0.014248968102037907, "grad_norm": 0.06560266762971878, "global_step": 13481, "epoch": 127, "lr": 8.486580905562039e-05} +{"train_loss": 0.016517026349902153, "grad_norm": 0.10331416130065918, "global_step": 13482, "epoch": 127, "lr": 8.486368467413697e-05} +{"train_loss": 0.015271407552063465, "grad_norm": 0.18207301199436188, "global_step": 13483, "epoch": 127, "lr": 8.486156017015772e-05} +{"train_loss": 0.01583370566368103, "grad_norm": 0.14342892169952393, "global_step": 13484, "epoch": 127, "lr": 8.485943554369011e-05} +{"train_loss": 0.01902632787823677, "grad_norm": 0.1399080455303192, "global_step": 13485, "epoch": 127, "lr": 8.48573107947416e-05} +{"train_loss": 0.01730000413954258, "grad_norm": 0.10976620018482208, "global_step": 13486, "epoch": 127, "lr": 8.485518592331968e-05} +{"train_loss": 0.018501147627830505, "grad_norm": 0.06310488283634186, "global_step": 13487, "epoch": 127, "lr": 8.485306092943177e-05} +{"train_loss": 0.01799609325826168, "grad_norm": 0.06856068223714828, "global_step": 13488, "epoch": 127, "lr": 8.485093581308538e-05} +{"train_loss": 0.015627827495336533, "grad_norm": 0.1555023491382599, "global_step": 13489, "epoch": 127, "lr": 8.484881057428795e-05} +{"train_loss": 0.017047204077243805, "grad_norm": 0.07885627448558807, "global_step": 13490, "epoch": 127, "lr": 8.484668521304695e-05} +{"train_loss": 0.01806480810046196, "grad_norm": 0.07236170768737793, "global_step": 13491, "epoch": 127, "lr": 8.484455972936986e-05} +{"train_loss": 0.016459863632917404, "grad_norm": 0.046692442148923874, "global_step": 13492, "epoch": 127, "lr": 8.484243412326414e-05} +{"train_loss": 0.013226684182882309, "grad_norm": 0.054707057774066925, "global_step": 13493, "epoch": 127, "lr": 8.484030839473725e-05} +{"train_loss": 0.019245723262429237, "grad_norm": 0.046566449105739594, "global_step": 13494, "epoch": 127, "lr": 8.483818254379668e-05} +{"train_loss": 0.01734241098165512, "grad_norm": 0.08628126978874207, "global_step": 13495, "epoch": 127, "lr": 8.483605657044989e-05} +{"train_loss": 0.015534639358520508, "grad_norm": 0.1178019568324089, "global_step": 13496, "epoch": 127, "lr": 8.483393047470433e-05} +{"train_loss": 0.01657136157155037, "grad_norm": 0.11209533363580704, "global_step": 13497, "epoch": 127, "lr": 8.483180425656749e-05} +{"train_loss": 0.017510533332824707, "grad_norm": 0.05876622721552849, "global_step": 13498, "epoch": 127, "lr": 8.482967791604685e-05} +{"train_loss": 0.01709366962313652, "grad_norm": 0.17111824452877045, "global_step": 13499, "epoch": 127, "lr": 8.482755145314986e-05} +{"train_loss": 0.014931531623005867, "grad_norm": 0.07004561275243759, "global_step": 13500, "epoch": 127, "lr": 8.4825424867884e-05} +{"train_loss": 0.016190359368920326, "grad_norm": 0.10534042865037918, "global_step": 13501, "epoch": 127, "lr": 8.482329816025675e-05} +{"train_loss": 0.01643127202987671, "grad_norm": 0.0750376358628273, "global_step": 13502, "epoch": 127, "lr": 8.482117133027556e-05} +{"train_loss": 0.017982710152864456, "grad_norm": 0.07466773688793182, "global_step": 13503, "epoch": 127, "lr": 8.481904437794792e-05} +{"train_loss": 0.018317878246307373, "grad_norm": 0.05684376507997513, "global_step": 13504, "epoch": 127, "lr": 8.481691730328132e-05} +{"train_loss": 0.02139141783118248, "grad_norm": 0.12409211695194244, "global_step": 13505, "epoch": 127, "lr": 8.481479010628319e-05} +{"train_loss": 0.01645669713616371, "grad_norm": 0.0503726527094841, "global_step": 13506, "epoch": 127, "lr": 8.481266278696106e-05} +{"train_loss": 0.017637193202972412, "grad_norm": 0.059625495225191116, "global_step": 13507, "epoch": 127, "lr": 8.481053534532235e-05} +{"train_loss": 0.01785881817340851, "grad_norm": 0.08553660660982132, "global_step": 13508, "epoch": 127, "lr": 8.480840778137455e-05} +{"train_loss": 0.016497965902090073, "grad_norm": 0.05599847435951233, "global_step": 13509, "epoch": 127, "lr": 8.480628009512514e-05} +{"train_loss": 0.017218904569745064, "grad_norm": 0.10322349518537521, "global_step": 13510, "epoch": 127, "lr": 8.480415228658163e-05} +{"train_loss": 0.01551247388124466, "grad_norm": 0.04478207975625992, "global_step": 13511, "epoch": 127, "lr": 8.480202435575144e-05} +{"train_loss": 0.018428528681397438, "grad_norm": 0.050686608999967575, "global_step": 13512, "epoch": 127, "lr": 8.479989630264206e-05} +{"train_loss": 0.019440025091171265, "grad_norm": 0.09940215945243835, "global_step": 13513, "epoch": 127, "lr": 8.479776812726101e-05} +{"train_loss": 0.01670423150062561, "grad_norm": 0.14507189393043518, "global_step": 13514, "epoch": 127, "lr": 8.479563982961571e-05} +{"train_loss": 0.02056572586297989, "grad_norm": 0.1596580594778061, "global_step": 13515, "epoch": 127, "lr": 8.479351140971368e-05} +{"train_loss": 0.016369225457310677, "grad_norm": 0.1183236762881279, "global_step": 13516, "epoch": 127, "lr": 8.479138286756238e-05} +{"train_loss": 0.01932825893163681, "grad_norm": 0.090961754322052, "global_step": 13517, "epoch": 127, "lr": 8.47892542031693e-05} +{"train_loss": 0.017934035509824753, "grad_norm": 0.0822482481598854, "global_step": 13518, "epoch": 127, "lr": 8.478712541654188e-05} +{"train_loss": 0.016298163682222366, "grad_norm": 0.1140470877289772, "global_step": 13519, "epoch": 127, "lr": 8.478499650768767e-05} +{"train_loss": 0.01414581574499607, "grad_norm": 0.05596182867884636, "global_step": 13520, "epoch": 127, "lr": 8.478286747661408e-05} +{"train_loss": 0.016214285045862198, "grad_norm": 0.08664663881063461, "global_step": 13521, "epoch": 127, "lr": 8.478073832332862e-05} +{"train_loss": 0.018951814621686935, "grad_norm": 0.06309258937835693, "global_step": 13522, "epoch": 127, "lr": 8.477860904783878e-05} +{"train_loss": 0.01977185159921646, "grad_norm": 0.052441537380218506, "global_step": 13523, "epoch": 127, "lr": 8.477647965015204e-05} +{"train_loss": 0.018376518040895462, "grad_norm": 0.08562850207090378, "global_step": 13524, "epoch": 127, "lr": 8.477435013027586e-05} +{"train_loss": 0.01934939995408058, "grad_norm": 0.21810483932495117, "global_step": 13525, "epoch": 127, "lr": 8.477222048821775e-05} +{"train_loss": 0.016298655420541763, "grad_norm": 0.06688615679740906, "global_step": 13526, "epoch": 127, "lr": 8.477009072398516e-05} +{"train_loss": 0.018169865012168884, "grad_norm": 0.1060873344540596, "global_step": 13527, "epoch": 127, "lr": 8.476796083758563e-05} +{"train_loss": 0.01725766994059086, "grad_norm": 0.11380886286497116, "global_step": 13528, "epoch": 127, "lr": 8.476583082902655e-05} +{"train_loss": 0.01936214789748192, "grad_norm": 0.14750760793685913, "global_step": 13529, "epoch": 127, "lr": 8.47637006983155e-05} +{"train_loss": 0.016718393191695213, "grad_norm": 0.18282882869243622, "global_step": 13530, "epoch": 127, "lr": 8.476157044545994e-05} +{"train_loss": 0.016881555318832397, "grad_norm": 0.07357128709554672, "global_step": 13531, "epoch": 127, "lr": 8.47594400704673e-05} +{"train_loss": 0.017044398933649063, "grad_norm": 0.13049261271953583, "global_step": 13532, "epoch": 127, "lr": 8.475730957334511e-05} +{"train_loss": 0.015376142226159573, "grad_norm": 0.07906921952962875, "global_step": 13533, "epoch": 127, "lr": 8.475517895410087e-05} +{"train_loss": 0.0179138146340847, "grad_norm": 0.06954437494277954, "global_step": 13534, "epoch": 127, "lr": 8.475304821274202e-05} +{"train_loss": 0.017794538289308548, "grad_norm": 0.07679793238639832, "global_step": 13535, "epoch": 127, "lr": 8.475091734927609e-05} +{"train_loss": 0.01866808533668518, "grad_norm": 0.08338046073913574, "global_step": 13536, "epoch": 127, "lr": 8.474878636371055e-05} +{"train_loss": 0.014217110350728035, "grad_norm": 0.10437273979187012, "global_step": 13537, "epoch": 127, "lr": 8.474665525605287e-05} +{"train_loss": 0.018369946628808975, "grad_norm": 0.0817980095744133, "global_step": 13538, "epoch": 127, "lr": 8.474452402631057e-05} +{"train_loss": 0.018953246995806694, "grad_norm": 0.0512571781873703, "global_step": 13539, "epoch": 127, "lr": 8.474239267449112e-05} +{"train_loss": 0.01676807552576065, "grad_norm": 0.22487987577915192, "global_step": 13540, "epoch": 127, "lr": 8.4740261200602e-05} +{"train_loss": 0.015883684158325195, "grad_norm": 0.09788171201944351, "global_step": 13541, "epoch": 127, "lr": 8.473812960465073e-05} +{"train_loss": 0.01587630622088909, "grad_norm": 0.11319256573915482, "global_step": 13542, "epoch": 127, "lr": 8.473599788664477e-05} +{"train_loss": 0.015071383677423, "grad_norm": 0.09175477921962738, "global_step": 13543, "epoch": 127, "lr": 8.473386604659161e-05} +{"train_loss": 0.016936156898736954, "grad_norm": 0.11722473055124283, "global_step": 13544, "epoch": 127, "lr": 8.473173408449877e-05} +{"train_loss": 0.01600252091884613, "grad_norm": 0.09947460889816284, "global_step": 13545, "epoch": 127, "lr": 8.472960200037371e-05} +{"train_loss": 0.0171040166169405, "grad_norm": 0.06184129789471626, "global_step": 13546, "epoch": 127, "lr": 8.472746979422394e-05} +{"train_loss": 0.01700391247868538, "grad_norm": 0.04876091331243515, "global_step": 13547, "epoch": 127, "lr": 8.472533746605693e-05} +{"train_loss": 0.01726377010345459, "grad_norm": 0.1088346466422081, "global_step": 13548, "epoch": 127, "lr": 8.472320501588019e-05} +{"train_loss": 0.015699317678809166, "grad_norm": 0.07958318293094635, "global_step": 13549, "epoch": 127, "lr": 8.47210724437012e-05} +{"train_loss": 0.01591860130429268, "grad_norm": 0.05755093693733215, "global_step": 13550, "epoch": 127, "lr": 8.471893974952747e-05} +{"train_loss": 0.016182707622647285, "grad_norm": 0.10988237708806992, "global_step": 13551, "epoch": 127, "lr": 8.47168069333665e-05} +{"train_loss": 0.019848378375172615, "grad_norm": 0.06320235878229141, "global_step": 13552, "epoch": 127, "lr": 8.471467399522575e-05} +{"train_loss": 0.01926841400563717, "grad_norm": 0.1140674278140068, "global_step": 13553, "epoch": 127, "lr": 8.471254093511274e-05} +{"train_loss": 0.01679530180990696, "grad_norm": 0.06119564548134804, "global_step": 13554, "epoch": 127, "lr": 8.471040775303496e-05} +{"train_loss": 0.0149298757314682, "grad_norm": 0.08261321485042572, "global_step": 13555, "epoch": 127, "lr": 8.470827444899989e-05} +{"train_loss": 0.015211316756904125, "grad_norm": 0.11859073489904404, "global_step": 13556, "epoch": 127, "lr": 8.470614102301504e-05} +{"train_loss": 0.01571476086974144, "grad_norm": 0.0750737190246582, "global_step": 13557, "epoch": 127, "lr": 8.470400747508791e-05} +{"train_loss": 0.017075739800930023, "grad_norm": 0.04915585368871689, "global_step": 13558, "epoch": 127, "lr": 8.470187380522599e-05} +{"train_loss": 0.016950340941548347, "grad_norm": 0.07507680356502533, "global_step": 13559, "epoch": 127, "lr": 8.469974001343677e-05} +{"train_loss": 0.01865381747484207, "grad_norm": 0.06709308177232742, "global_step": 13560, "epoch": 127, "lr": 8.469760609972777e-05} +{"train_loss": 0.01824279874563217, "grad_norm": 0.07177651673555374, "global_step": 13561, "epoch": 127, "lr": 8.469547206410645e-05} +{"train_loss": 0.01746523752808571, "grad_norm": 0.11363113671541214, "global_step": 13562, "epoch": 127, "lr": 8.469333790658034e-05} +{"train_loss": 0.013805509544909, "grad_norm": 0.06343180686235428, "global_step": 13563, "epoch": 127, "lr": 8.469120362715692e-05} +{"train_loss": 0.015467002056539059, "grad_norm": 0.07023818045854568, "global_step": 13564, "epoch": 127, "lr": 8.468906922584371e-05} +{"train_loss": 0.015661008656024933, "grad_norm": 0.09490849822759628, "global_step": 13565, "epoch": 127, "lr": 8.468693470264819e-05} +{"train_loss": 0.015315206721425056, "grad_norm": 0.05551715940237045, "global_step": 13566, "epoch": 127, "lr": 8.468480005757789e-05} +{"train_loss": 0.017017764030551573, "grad_norm": 0.0622946135699749, "global_step": 13567, "epoch": 127, "lr": 8.468266529064025e-05, "val_loss": 0.019081532955169678} +{"train_loss": 0.016774151474237442, "grad_norm": 0.0770292654633522, "global_step": 13568, "epoch": 128, "lr": 8.468053040184283e-05} +{"train_loss": 0.017260562628507614, "grad_norm": 0.1770033985376358, "global_step": 13569, "epoch": 128, "lr": 8.46783953911931e-05} +{"train_loss": 0.015485579147934914, "grad_norm": 0.10223978757858276, "global_step": 13570, "epoch": 128, "lr": 8.467626025869857e-05} +{"train_loss": 0.017414579167962074, "grad_norm": 0.06552328169345856, "global_step": 13571, "epoch": 128, "lr": 8.467412500436673e-05} +{"train_loss": 0.019200539216399193, "grad_norm": 0.05334949120879173, "global_step": 13572, "epoch": 128, "lr": 8.467198962820512e-05} +{"train_loss": 0.019240282475948334, "grad_norm": 0.12913279235363007, "global_step": 13573, "epoch": 128, "lr": 8.46698541302212e-05} +{"train_loss": 0.018058300018310547, "grad_norm": 0.15792062878608704, "global_step": 13574, "epoch": 128, "lr": 8.46677185104225e-05} +{"train_loss": 0.01738169603049755, "grad_norm": 0.11641626805067062, "global_step": 13575, "epoch": 128, "lr": 8.466558276881651e-05} +{"train_loss": 0.01682499423623085, "grad_norm": 0.06464727967977524, "global_step": 13576, "epoch": 128, "lr": 8.466344690541072e-05} +{"train_loss": 0.01618560031056404, "grad_norm": 0.08567413687705994, "global_step": 13577, "epoch": 128, "lr": 8.466131092021268e-05} +{"train_loss": 0.013863072730600834, "grad_norm": 0.1408708542585373, "global_step": 13578, "epoch": 128, "lr": 8.465917481322984e-05} +{"train_loss": 0.015248949639499187, "grad_norm": 0.15666572749614716, "global_step": 13579, "epoch": 128, "lr": 8.465703858446977e-05} +{"train_loss": 0.019450686872005463, "grad_norm": 0.14733201265335083, "global_step": 13580, "epoch": 128, "lr": 8.465490223393991e-05} +{"train_loss": 0.01741338148713112, "grad_norm": 0.09914586693048477, "global_step": 13581, "epoch": 128, "lr": 8.46527657616478e-05} +{"train_loss": 0.01819498836994171, "grad_norm": 0.04372492432594299, "global_step": 13582, "epoch": 128, "lr": 8.465062916760094e-05} +{"train_loss": 0.015813248232007027, "grad_norm": 0.09971786290407181, "global_step": 13583, "epoch": 128, "lr": 8.464849245180684e-05} +{"train_loss": 0.017216036096215248, "grad_norm": 0.08715394139289856, "global_step": 13584, "epoch": 128, "lr": 8.4646355614273e-05} +{"train_loss": 0.016928423196077347, "grad_norm": 0.07288695126771927, "global_step": 13585, "epoch": 128, "lr": 8.464421865500694e-05} +{"train_loss": 0.018637381494045258, "grad_norm": 0.049223341047763824, "global_step": 13586, "epoch": 128, "lr": 8.464208157401616e-05} +{"train_loss": 0.017965029925107956, "grad_norm": 0.10113845020532608, "global_step": 13587, "epoch": 128, "lr": 8.463994437130818e-05} +{"train_loss": 0.013892442919313908, "grad_norm": 0.09379580616950989, "global_step": 13588, "epoch": 128, "lr": 8.463780704689048e-05} +{"train_loss": 0.01945486292243004, "grad_norm": 0.0894780158996582, "global_step": 13589, "epoch": 128, "lr": 8.463566960077062e-05} +{"train_loss": 0.01827087439596653, "grad_norm": 0.06478515267372131, "global_step": 13590, "epoch": 128, "lr": 8.463353203295605e-05} +{"train_loss": 0.015985071659088135, "grad_norm": 0.04863285273313522, "global_step": 13591, "epoch": 128, "lr": 8.463139434345433e-05} +{"train_loss": 0.01646268367767334, "grad_norm": 0.07532847672700882, "global_step": 13592, "epoch": 128, "lr": 8.462925653227294e-05} +{"train_loss": 0.018031595274806023, "grad_norm": 0.09147977083921432, "global_step": 13593, "epoch": 128, "lr": 8.46271185994194e-05} +{"train_loss": 0.01967303827404976, "grad_norm": 0.05268000811338425, "global_step": 13594, "epoch": 128, "lr": 8.462498054490124e-05} +{"train_loss": 0.016589513048529625, "grad_norm": 0.11184220761060715, "global_step": 13595, "epoch": 128, "lr": 8.462284236872595e-05} +{"train_loss": 0.017941830679774284, "grad_norm": 0.12718619406223297, "global_step": 13596, "epoch": 128, "lr": 8.462070407090104e-05} +{"train_loss": 0.014026314951479435, "grad_norm": 0.06435729563236237, "global_step": 13597, "epoch": 128, "lr": 8.461856565143404e-05} +{"train_loss": 0.013595831580460072, "grad_norm": 0.1312502771615982, "global_step": 13598, "epoch": 128, "lr": 8.461642711033244e-05} +{"train_loss": 0.015142299234867096, "grad_norm": 0.08042589575052261, "global_step": 13599, "epoch": 128, "lr": 8.46142884476038e-05} +{"train_loss": 0.016693003475666046, "grad_norm": 0.09827380627393723, "global_step": 13600, "epoch": 128, "lr": 8.461214966325557e-05} +{"train_loss": 0.017939027398824692, "grad_norm": 0.06204019859433174, "global_step": 13601, "epoch": 128, "lr": 8.461001075729533e-05} +{"train_loss": 0.01882830262184143, "grad_norm": 0.186541348695755, "global_step": 13602, "epoch": 128, "lr": 8.460787172973053e-05} +{"train_loss": 0.017238225787878036, "grad_norm": 0.08602862805128098, "global_step": 13603, "epoch": 128, "lr": 8.460573258056874e-05} +{"train_loss": 0.01476890780031681, "grad_norm": 0.0540526807308197, "global_step": 13604, "epoch": 128, "lr": 8.460359330981746e-05} +{"train_loss": 0.018079165369272232, "grad_norm": 0.08572390675544739, "global_step": 13605, "epoch": 128, "lr": 8.460145391748419e-05} +{"train_loss": 0.015760865062475204, "grad_norm": 0.06096706911921501, "global_step": 13606, "epoch": 128, "lr": 8.459931440357645e-05} +{"train_loss": 0.020044494420289993, "grad_norm": 0.11281406879425049, "global_step": 13607, "epoch": 128, "lr": 8.45971747681018e-05} +{"train_loss": 0.017042387276887894, "grad_norm": 0.08227268606424332, "global_step": 13608, "epoch": 128, "lr": 8.459503501106769e-05} +{"train_loss": 0.019976360723376274, "grad_norm": 0.057341884821653366, "global_step": 13609, "epoch": 128, "lr": 8.45928951324817e-05} +{"train_loss": 0.014539666473865509, "grad_norm": 0.08112142235040665, "global_step": 13610, "epoch": 128, "lr": 8.459075513235129e-05} +{"train_loss": 0.017635319381952286, "grad_norm": 0.11266931146383286, "global_step": 13611, "epoch": 128, "lr": 8.458861501068402e-05} +{"train_loss": 0.017084352672100067, "grad_norm": 0.19821859896183014, "global_step": 13612, "epoch": 128, "lr": 8.458647476748742e-05} +{"train_loss": 0.017865031957626343, "grad_norm": 0.07877936214208603, "global_step": 13613, "epoch": 128, "lr": 8.458433440276896e-05} +{"train_loss": 0.014065107330679893, "grad_norm": 0.13683894276618958, "global_step": 13614, "epoch": 128, "lr": 8.458219391653621e-05} +{"train_loss": 0.014833728782832623, "grad_norm": 0.167999267578125, "global_step": 13615, "epoch": 128, "lr": 8.458005330879666e-05} +{"train_loss": 0.014621933922171593, "grad_norm": 0.0382363423705101, "global_step": 13616, "epoch": 128, "lr": 8.457791257955785e-05} +{"train_loss": 0.017825592309236526, "grad_norm": 0.08755723387002945, "global_step": 13617, "epoch": 128, "lr": 8.457577172882728e-05} +{"train_loss": 0.017783626914024353, "grad_norm": 0.06401418149471283, "global_step": 13618, "epoch": 128, "lr": 8.457363075661249e-05} +{"train_loss": 0.019242700189352036, "grad_norm": 0.07559490203857422, "global_step": 13619, "epoch": 128, "lr": 8.4571489662921e-05} +{"train_loss": 0.014140728861093521, "grad_norm": 0.09862823039293289, "global_step": 13620, "epoch": 128, "lr": 8.456934844776032e-05} +{"train_loss": 0.015656035393476486, "grad_norm": 0.10992542654275894, "global_step": 13621, "epoch": 128, "lr": 8.456720711113799e-05} +{"train_loss": 0.015332967042922974, "grad_norm": 0.1221952736377716, "global_step": 13622, "epoch": 128, "lr": 8.456506565306153e-05} +{"train_loss": 0.018931224942207336, "grad_norm": 0.14888115227222443, "global_step": 13623, "epoch": 128, "lr": 8.456292407353847e-05} +{"train_loss": 0.015374119393527508, "grad_norm": 0.09215424209833145, "global_step": 13624, "epoch": 128, "lr": 8.456078237257631e-05} +{"train_loss": 0.018223488703370094, "grad_norm": 0.17751342058181763, "global_step": 13625, "epoch": 128, "lr": 8.45586405501826e-05} +{"train_loss": 0.016771087422966957, "grad_norm": 0.12931975722312927, "global_step": 13626, "epoch": 128, "lr": 8.455649860636485e-05} +{"train_loss": 0.0175330713391304, "grad_norm": 0.08531998097896576, "global_step": 13627, "epoch": 128, "lr": 8.45543565411306e-05} +{"train_loss": 0.01633937656879425, "grad_norm": 0.10473626852035522, "global_step": 13628, "epoch": 128, "lr": 8.455221435448735e-05} +{"train_loss": 0.01794741302728653, "grad_norm": 0.060018088668584824, "global_step": 13629, "epoch": 128, "lr": 8.455007204644266e-05} +{"train_loss": 0.016626445576548576, "grad_norm": 0.044086650013923645, "global_step": 13630, "epoch": 128, "lr": 8.454792961700405e-05} +{"train_loss": 0.018860362470149994, "grad_norm": 0.13832122087478638, "global_step": 13631, "epoch": 128, "lr": 8.454578706617903e-05} +{"train_loss": 0.014789929613471031, "grad_norm": 0.08334287256002426, "global_step": 13632, "epoch": 128, "lr": 8.454364439397513e-05} +{"train_loss": 0.01675356552004814, "grad_norm": 0.12984691560268402, "global_step": 13633, "epoch": 128, "lr": 8.45415016003999e-05} +{"train_loss": 0.0199505016207695, "grad_norm": 0.09070315212011337, "global_step": 13634, "epoch": 128, "lr": 8.453935868546086e-05} +{"train_loss": 0.019755352288484573, "grad_norm": 0.1096034049987793, "global_step": 13635, "epoch": 128, "lr": 8.45372156491655e-05} +{"train_loss": 0.018978048115968704, "grad_norm": 0.09529615938663483, "global_step": 13636, "epoch": 128, "lr": 8.453507249152141e-05} +{"train_loss": 0.014770402573049068, "grad_norm": 0.06604854762554169, "global_step": 13637, "epoch": 128, "lr": 8.453292921253611e-05} +{"train_loss": 0.013281506486237049, "grad_norm": 0.045441944152116776, "global_step": 13638, "epoch": 128, "lr": 8.45307858122171e-05} +{"train_loss": 0.016412481665611267, "grad_norm": 0.04578539356589317, "global_step": 13639, "epoch": 128, "lr": 8.452864229057191e-05} +{"train_loss": 0.01721012592315674, "grad_norm": 0.06340812891721725, "global_step": 13640, "epoch": 128, "lr": 8.45264986476081e-05} +{"train_loss": 0.017607854679226875, "grad_norm": 0.06451765447854996, "global_step": 13641, "epoch": 128, "lr": 8.452435488333319e-05} +{"train_loss": 0.017473336309194565, "grad_norm": 0.05986107513308525, "global_step": 13642, "epoch": 128, "lr": 8.45222109977547e-05} +{"train_loss": 0.01532834954559803, "grad_norm": 0.09779269248247147, "global_step": 13643, "epoch": 128, "lr": 8.45200669908802e-05} +{"train_loss": 0.01789136976003647, "grad_norm": 0.14690400660037994, "global_step": 13644, "epoch": 128, "lr": 8.451792286271717e-05} +{"train_loss": 0.017549991607666016, "grad_norm": 0.05233755707740784, "global_step": 13645, "epoch": 128, "lr": 8.45157786132732e-05} +{"train_loss": 0.01785862445831299, "grad_norm": 0.06277250498533249, "global_step": 13646, "epoch": 128, "lr": 8.451363424255575e-05} +{"train_loss": 0.0175941064953804, "grad_norm": 0.07313825190067291, "global_step": 13647, "epoch": 128, "lr": 8.451148975057243e-05} +{"train_loss": 0.01683713123202324, "grad_norm": 0.1073305681347847, "global_step": 13648, "epoch": 128, "lr": 8.450934513733074e-05} +{"train_loss": 0.01798625849187374, "grad_norm": 0.05626266077160835, "global_step": 13649, "epoch": 128, "lr": 8.450720040283821e-05} +{"train_loss": 0.018909044563770294, "grad_norm": 0.10017750412225723, "global_step": 13650, "epoch": 128, "lr": 8.450505554710239e-05} +{"train_loss": 0.017134662717580795, "grad_norm": 0.11527077108621597, "global_step": 13651, "epoch": 128, "lr": 8.450291057013082e-05} +{"train_loss": 0.019435446709394455, "grad_norm": 0.1629827469587326, "global_step": 13652, "epoch": 128, "lr": 8.450076547193102e-05} +{"train_loss": 0.01814780756831169, "grad_norm": 0.05267014726996422, "global_step": 13653, "epoch": 128, "lr": 8.449862025251053e-05} +{"train_loss": 0.01725952699780464, "grad_norm": 0.1732790172100067, "global_step": 13654, "epoch": 128, "lr": 8.44964749118769e-05} +{"train_loss": 0.017786964774131775, "grad_norm": 0.06351955235004425, "global_step": 13655, "epoch": 128, "lr": 8.449432945003765e-05} +{"train_loss": 0.01831357553601265, "grad_norm": 0.0788927897810936, "global_step": 13656, "epoch": 128, "lr": 8.449218386700032e-05} +{"train_loss": 0.016341524198651314, "grad_norm": 0.06416890770196915, "global_step": 13657, "epoch": 128, "lr": 8.449003816277247e-05} +{"train_loss": 0.021108539775013924, "grad_norm": 0.08471842110157013, "global_step": 13658, "epoch": 128, "lr": 8.448789233736163e-05} +{"train_loss": 0.016698192805051804, "grad_norm": 0.05962250009179115, "global_step": 13659, "epoch": 128, "lr": 8.448574639077533e-05} +{"train_loss": 0.018862789496779442, "grad_norm": 0.05828431621193886, "global_step": 13660, "epoch": 128, "lr": 8.448360032302112e-05} +{"train_loss": 0.016216179355978966, "grad_norm": 0.05028245970606804, "global_step": 13661, "epoch": 128, "lr": 8.448145413410653e-05} +{"train_loss": 0.016858335584402084, "grad_norm": 0.0774683877825737, "global_step": 13662, "epoch": 128, "lr": 8.44793078240391e-05} +{"train_loss": 0.01467371080070734, "grad_norm": 0.07782663404941559, "global_step": 13663, "epoch": 128, "lr": 8.447716139282637e-05} +{"train_loss": 0.018527250736951828, "grad_norm": 0.07199869304895401, "global_step": 13664, "epoch": 128, "lr": 8.447501484047591e-05} +{"train_loss": 0.015086867846548557, "grad_norm": 0.06235004588961601, "global_step": 13665, "epoch": 128, "lr": 8.447286816699522e-05} +{"train_loss": 0.014726299792528152, "grad_norm": 0.05189989134669304, "global_step": 13666, "epoch": 128, "lr": 8.447072137239187e-05} +{"train_loss": 0.01747041568160057, "grad_norm": 0.07987450808286667, "global_step": 13667, "epoch": 128, "lr": 8.44685744566734e-05} +{"train_loss": 0.017832044512033463, "grad_norm": 0.12964008748531342, "global_step": 13668, "epoch": 128, "lr": 8.446642741984735e-05} +{"train_loss": 0.013766247779130936, "grad_norm": 0.04757765680551529, "global_step": 13669, "epoch": 128, "lr": 8.446428026192125e-05} +{"train_loss": 0.014223464764654636, "grad_norm": 0.07219967246055603, "global_step": 13670, "epoch": 128, "lr": 8.446213298290269e-05} +{"train_loss": 0.016167394816875458, "grad_norm": 0.09033820778131485, "global_step": 13671, "epoch": 128, "lr": 8.445998558279916e-05} +{"train_loss": 0.014034957624971867, "grad_norm": 0.07357285171747208, "global_step": 13672, "epoch": 128, "lr": 8.445783806161823e-05} +{"train_loss": 0.016980217875174758, "grad_norm": 0.19929474592208862, "global_step": 13673, "epoch": 128, "lr": 8.445569041936743e-05, "val_loss": 0.019205886870622635} +{"train_loss": 0.014589606784284115, "grad_norm": 0.10897298902273178, "global_step": 13674, "epoch": 129, "lr": 8.445354265605431e-05} +{"train_loss": 0.018641222268342972, "grad_norm": 0.08809374272823334, "global_step": 13675, "epoch": 129, "lr": 8.445139477168646e-05} +{"train_loss": 0.014629509299993515, "grad_norm": 0.06279091536998749, "global_step": 13676, "epoch": 129, "lr": 8.444924676627137e-05} +{"train_loss": 0.01809670403599739, "grad_norm": 0.07433319091796875, "global_step": 13677, "epoch": 129, "lr": 8.444709863981661e-05} +{"train_loss": 0.01514335349202156, "grad_norm": 0.04424430429935455, "global_step": 13678, "epoch": 129, "lr": 8.444495039232974e-05} +{"train_loss": 0.015107252635061741, "grad_norm": 0.051771778613328934, "global_step": 13679, "epoch": 129, "lr": 8.444280202381826e-05} +{"train_loss": 0.01527322642505169, "grad_norm": 0.10279420763254166, "global_step": 13680, "epoch": 129, "lr": 8.44406535342898e-05} +{"train_loss": 0.020207062363624573, "grad_norm": 0.08216218650341034, "global_step": 13681, "epoch": 129, "lr": 8.443850492375182e-05} +{"train_loss": 0.01885165274143219, "grad_norm": 0.16302701830863953, "global_step": 13682, "epoch": 129, "lr": 8.443635619221193e-05} +{"train_loss": 0.016795475035905838, "grad_norm": 0.08506829291582108, "global_step": 13683, "epoch": 129, "lr": 8.443420733967767e-05} +{"train_loss": 0.01802051067352295, "grad_norm": 0.16532036662101746, "global_step": 13684, "epoch": 129, "lr": 8.443205836615655e-05} +{"train_loss": 0.019500328227877617, "grad_norm": 0.09479380398988724, "global_step": 13685, "epoch": 129, "lr": 8.442990927165617e-05} +{"train_loss": 0.01767107844352722, "grad_norm": 0.05705986171960831, "global_step": 13686, "epoch": 129, "lr": 8.442776005618407e-05} +{"train_loss": 0.017362024635076523, "grad_norm": 0.054181572049856186, "global_step": 13687, "epoch": 129, "lr": 8.442561071974777e-05} +{"train_loss": 0.017394375056028366, "grad_norm": 0.04972098767757416, "global_step": 13688, "epoch": 129, "lr": 8.442346126235486e-05} +{"train_loss": 0.01782592013478279, "grad_norm": 0.06091151386499405, "global_step": 13689, "epoch": 129, "lr": 8.442131168401288e-05} +{"train_loss": 0.017443105578422546, "grad_norm": 0.055100128054618835, "global_step": 13690, "epoch": 129, "lr": 8.441916198472937e-05} +{"train_loss": 0.016149815171957016, "grad_norm": 0.09320332854986191, "global_step": 13691, "epoch": 129, "lr": 8.44170121645119e-05} +{"train_loss": 0.019304577261209488, "grad_norm": 0.0714358538389206, "global_step": 13692, "epoch": 129, "lr": 8.441486222336803e-05} +{"train_loss": 0.01695956662297249, "grad_norm": 0.056654129177331924, "global_step": 13693, "epoch": 129, "lr": 8.441271216130529e-05} +{"train_loss": 0.016078494489192963, "grad_norm": 0.1480504870414734, "global_step": 13694, "epoch": 129, "lr": 8.441056197833123e-05} +{"train_loss": 0.017684724181890488, "grad_norm": 0.10198227316141129, "global_step": 13695, "epoch": 129, "lr": 8.440841167445343e-05} +{"train_loss": 0.015315867960453033, "grad_norm": 0.05764245241880417, "global_step": 13696, "epoch": 129, "lr": 8.440626124967945e-05} +{"train_loss": 0.018012018874287605, "grad_norm": 0.0765543058514595, "global_step": 13697, "epoch": 129, "lr": 8.440411070401681e-05} +{"train_loss": 0.016843169927597046, "grad_norm": 0.12695768475532532, "global_step": 13698, "epoch": 129, "lr": 8.44019600374731e-05} +{"train_loss": 0.01653452403843403, "grad_norm": 0.04375159740447998, "global_step": 13699, "epoch": 129, "lr": 8.439980925005586e-05} +{"train_loss": 0.01824597641825676, "grad_norm": 0.04736431688070297, "global_step": 13700, "epoch": 129, "lr": 8.439765834177264e-05} +{"train_loss": 0.015942201018333435, "grad_norm": 0.054617758840322495, "global_step": 13701, "epoch": 129, "lr": 8.439550731263101e-05} +{"train_loss": 0.01800137758255005, "grad_norm": 0.12269856035709381, "global_step": 13702, "epoch": 129, "lr": 8.439335616263854e-05} +{"train_loss": 0.01542324386537075, "grad_norm": 0.04877379909157753, "global_step": 13703, "epoch": 129, "lr": 8.439120489180277e-05} +{"train_loss": 0.016300145536661148, "grad_norm": 0.05220926180481911, "global_step": 13704, "epoch": 129, "lr": 8.438905350013127e-05} +{"train_loss": 0.017275430262088776, "grad_norm": 0.06908721476793289, "global_step": 13705, "epoch": 129, "lr": 8.438690198763157e-05} +{"train_loss": 0.015007449313998222, "grad_norm": 0.08073583245277405, "global_step": 13706, "epoch": 129, "lr": 8.438475035431126e-05} +{"train_loss": 0.017820434644818306, "grad_norm": 0.11633078753948212, "global_step": 13707, "epoch": 129, "lr": 8.43825986001779e-05} +{"train_loss": 0.016143247485160828, "grad_norm": 0.0493759885430336, "global_step": 13708, "epoch": 129, "lr": 8.438044672523902e-05} +{"train_loss": 0.01743922382593155, "grad_norm": 0.14323930442333221, "global_step": 13709, "epoch": 129, "lr": 8.437829472950222e-05} +{"train_loss": 0.01577100157737732, "grad_norm": 0.06787624210119247, "global_step": 13710, "epoch": 129, "lr": 8.437614261297504e-05} +{"train_loss": 0.01869232766330242, "grad_norm": 0.2144802212715149, "global_step": 13711, "epoch": 129, "lr": 8.437399037566504e-05} +{"train_loss": 0.014635298401117325, "grad_norm": 0.04555321857333183, "global_step": 13712, "epoch": 129, "lr": 8.437183801757979e-05} +{"train_loss": 0.014990398660302162, "grad_norm": 0.06754368543624878, "global_step": 13713, "epoch": 129, "lr": 8.436968553872684e-05} +{"train_loss": 0.01800098642706871, "grad_norm": 0.06750637292861938, "global_step": 13714, "epoch": 129, "lr": 8.436753293911377e-05} +{"train_loss": 0.01590602472424507, "grad_norm": 0.1273992955684662, "global_step": 13715, "epoch": 129, "lr": 8.436538021874814e-05} +{"train_loss": 0.016298554837703705, "grad_norm": 0.048551395535469055, "global_step": 13716, "epoch": 129, "lr": 8.436322737763752e-05} +{"train_loss": 0.017022471874952316, "grad_norm": 0.05325513705611229, "global_step": 13717, "epoch": 129, "lr": 8.436107441578944e-05} +{"train_loss": 0.016388755291700363, "grad_norm": 0.1071716845035553, "global_step": 13718, "epoch": 129, "lr": 8.435892133321149e-05} +{"train_loss": 0.014761196449398994, "grad_norm": 0.08908877521753311, "global_step": 13719, "epoch": 129, "lr": 8.435676812991124e-05} +{"train_loss": 0.017441648989915848, "grad_norm": 0.0731462612748146, "global_step": 13720, "epoch": 129, "lr": 8.435461480589624e-05} +{"train_loss": 0.01732538267970085, "grad_norm": 0.07247474044561386, "global_step": 13721, "epoch": 129, "lr": 8.435246136117407e-05} +{"train_loss": 0.017338527366518974, "grad_norm": 0.16612711548805237, "global_step": 13722, "epoch": 129, "lr": 8.43503077957523e-05} +{"train_loss": 0.015121011063456535, "grad_norm": 0.06698932498693466, "global_step": 13723, "epoch": 129, "lr": 8.434815410963846e-05} +{"train_loss": 0.018376704305410385, "grad_norm": 0.07121416181325912, "global_step": 13724, "epoch": 129, "lr": 8.434600030284017e-05} +{"train_loss": 0.019901271909475327, "grad_norm": 0.08330439776182175, "global_step": 13725, "epoch": 129, "lr": 8.434384637536496e-05} +{"train_loss": 0.0169875118881464, "grad_norm": 0.05803236365318298, "global_step": 13726, "epoch": 129, "lr": 8.434169232722043e-05} +{"train_loss": 0.016795501112937927, "grad_norm": 0.10431733727455139, "global_step": 13727, "epoch": 129, "lr": 8.433953815841409e-05} +{"train_loss": 0.01497658621519804, "grad_norm": 0.06366104632616043, "global_step": 13728, "epoch": 129, "lr": 8.433738386895358e-05} +{"train_loss": 0.01791364885866642, "grad_norm": 0.09291521459817886, "global_step": 13729, "epoch": 129, "lr": 8.433522945884643e-05} +{"train_loss": 0.01731177419424057, "grad_norm": 0.0724215880036354, "global_step": 13730, "epoch": 129, "lr": 8.43330749281002e-05} +{"train_loss": 0.01587291620671749, "grad_norm": 0.09603641927242279, "global_step": 13731, "epoch": 129, "lr": 8.433092027672249e-05} +{"train_loss": 0.015435870736837387, "grad_norm": 0.11278729885816574, "global_step": 13732, "epoch": 129, "lr": 8.432876550472085e-05} +{"train_loss": 0.017987914383411407, "grad_norm": 0.10421213507652283, "global_step": 13733, "epoch": 129, "lr": 8.432661061210285e-05} +{"train_loss": 0.016373392194509506, "grad_norm": 0.06865010410547256, "global_step": 13734, "epoch": 129, "lr": 8.432445559887608e-05} +{"train_loss": 0.016252592206001282, "grad_norm": 0.07779684662818909, "global_step": 13735, "epoch": 129, "lr": 8.43223004650481e-05} +{"train_loss": 0.01865573599934578, "grad_norm": 0.06806646287441254, "global_step": 13736, "epoch": 129, "lr": 8.432014521062647e-05} +{"train_loss": 0.01846887916326523, "grad_norm": 0.08059656620025635, "global_step": 13737, "epoch": 129, "lr": 8.43179898356188e-05} +{"train_loss": 0.01605292037129402, "grad_norm": 0.05539092421531677, "global_step": 13738, "epoch": 129, "lr": 8.43158343400326e-05} +{"train_loss": 0.015088140033185482, "grad_norm": 0.04886393994092941, "global_step": 13739, "epoch": 129, "lr": 8.431367872387553e-05} +{"train_loss": 0.018957611173391342, "grad_norm": 0.07010738551616669, "global_step": 13740, "epoch": 129, "lr": 8.431152298715507e-05} +{"train_loss": 0.0167505145072937, "grad_norm": 0.058917928487062454, "global_step": 13741, "epoch": 129, "lr": 8.430936712987887e-05} +{"train_loss": 0.014675811864435673, "grad_norm": 0.12572170794010162, "global_step": 13742, "epoch": 129, "lr": 8.430721115205446e-05} +{"train_loss": 0.0173969566822052, "grad_norm": 0.11537014693021774, "global_step": 13743, "epoch": 129, "lr": 8.430505505368944e-05} +{"train_loss": 0.02009465917944908, "grad_norm": 0.09861435741186142, "global_step": 13744, "epoch": 129, "lr": 8.430289883479137e-05} +{"train_loss": 0.018101077526807785, "grad_norm": 0.09730944782495499, "global_step": 13745, "epoch": 129, "lr": 8.430074249536783e-05} +{"train_loss": 0.01631094701588154, "grad_norm": 0.07186035066843033, "global_step": 13746, "epoch": 129, "lr": 8.42985860354264e-05} +{"train_loss": 0.0169008057564497, "grad_norm": 0.05815597251057625, "global_step": 13747, "epoch": 129, "lr": 8.429642945497465e-05} +{"train_loss": 0.01956896483898163, "grad_norm": 0.057305824011564255, "global_step": 13748, "epoch": 129, "lr": 8.429427275402018e-05} +{"train_loss": 0.016334502026438713, "grad_norm": 0.12394315749406815, "global_step": 13749, "epoch": 129, "lr": 8.429211593257054e-05} +{"train_loss": 0.01869015395641327, "grad_norm": 0.13208027184009552, "global_step": 13750, "epoch": 129, "lr": 8.42899589906333e-05} +{"train_loss": 0.015499763190746307, "grad_norm": 0.05635173246264458, "global_step": 13751, "epoch": 129, "lr": 8.428780192821608e-05} +{"train_loss": 0.018176771700382233, "grad_norm": 0.057422611862421036, "global_step": 13752, "epoch": 129, "lr": 8.428564474532642e-05} +{"train_loss": 0.018081482499837875, "grad_norm": 0.09960025548934937, "global_step": 13753, "epoch": 129, "lr": 8.428348744197192e-05} +{"train_loss": 0.01670493185520172, "grad_norm": 0.07149706780910492, "global_step": 13754, "epoch": 129, "lr": 8.428133001816017e-05} +{"train_loss": 0.016915401443839073, "grad_norm": 0.07462578266859055, "global_step": 13755, "epoch": 129, "lr": 8.42791724738987e-05} +{"train_loss": 0.014653140679001808, "grad_norm": 0.0940735712647438, "global_step": 13756, "epoch": 129, "lr": 8.427701480919514e-05} +{"train_loss": 0.01764259859919548, "grad_norm": 0.07874911278486252, "global_step": 13757, "epoch": 129, "lr": 8.427485702405707e-05} +{"train_loss": 0.016186270862817764, "grad_norm": 0.11582045257091522, "global_step": 13758, "epoch": 129, "lr": 8.427269911849204e-05} +{"train_loss": 0.014552371576428413, "grad_norm": 0.11106007546186447, "global_step": 13759, "epoch": 129, "lr": 8.427054109250766e-05} +{"train_loss": 0.018157560378313065, "grad_norm": 0.06349436193704605, "global_step": 13760, "epoch": 129, "lr": 8.426838294611152e-05} +{"train_loss": 0.01651589572429657, "grad_norm": 0.06009986251592636, "global_step": 13761, "epoch": 129, "lr": 8.426622467931116e-05} +{"train_loss": 0.01855897530913353, "grad_norm": 0.112030029296875, "global_step": 13762, "epoch": 129, "lr": 8.426406629211419e-05} +{"train_loss": 0.014522179029881954, "grad_norm": 0.05884812772274017, "global_step": 13763, "epoch": 129, "lr": 8.426190778452821e-05} +{"train_loss": 0.016489621251821518, "grad_norm": 0.07803916186094284, "global_step": 13764, "epoch": 129, "lr": 8.425974915656077e-05} +{"train_loss": 0.014875904656946659, "grad_norm": 0.059116464108228683, "global_step": 13765, "epoch": 129, "lr": 8.425759040821948e-05} +{"train_loss": 0.01328339148312807, "grad_norm": 0.056656837463378906, "global_step": 13766, "epoch": 129, "lr": 8.425543153951191e-05} +{"train_loss": 0.016021620482206345, "grad_norm": 0.08032805472612381, "global_step": 13767, "epoch": 129, "lr": 8.425327255044565e-05} +{"train_loss": 0.016992222517728806, "grad_norm": 0.09515306353569031, "global_step": 13768, "epoch": 129, "lr": 8.425111344102829e-05} +{"train_loss": 0.01781523786485195, "grad_norm": 0.0812644362449646, "global_step": 13769, "epoch": 129, "lr": 8.424895421126741e-05} +{"train_loss": 0.016553113237023354, "grad_norm": 0.057485874742269516, "global_step": 13770, "epoch": 129, "lr": 8.42467948611706e-05} +{"train_loss": 0.021871674805879593, "grad_norm": 0.11507520824670792, "global_step": 13771, "epoch": 129, "lr": 8.424463539074546e-05} +{"train_loss": 0.019601956009864807, "grad_norm": 0.11630574613809586, "global_step": 13772, "epoch": 129, "lr": 8.424247579999955e-05} +{"train_loss": 0.018293574452400208, "grad_norm": 0.08693449944257736, "global_step": 13773, "epoch": 129, "lr": 8.424031608894047e-05} +{"train_loss": 0.016539424657821655, "grad_norm": 0.06937382370233536, "global_step": 13774, "epoch": 129, "lr": 8.423815625757581e-05} +{"train_loss": 0.013909299857914448, "grad_norm": 0.05584999918937683, "global_step": 13775, "epoch": 129, "lr": 8.423599630591316e-05} +{"train_loss": 0.019086366519331932, "grad_norm": 0.15121306478977203, "global_step": 13776, "epoch": 129, "lr": 8.423383623396011e-05} +{"train_loss": 0.016581039875745773, "grad_norm": 0.06123347207903862, "global_step": 13777, "epoch": 129, "lr": 8.423167604172424e-05} +{"train_loss": 0.017898160964250565, "grad_norm": 0.049368228763341904, "global_step": 13778, "epoch": 129, "lr": 8.422951572921316e-05} +{"train_loss": 0.01693602809506767, "grad_norm": 0.10460896044969559, "global_step": 13779, "epoch": 129, "lr": 8.422735529643444e-05, "val_loss": 0.020595591515302658} +{"train_loss": 0.018739286810159683, "grad_norm": 0.1595541387796402, "global_step": 13780, "epoch": 130, "lr": 8.422519474339567e-05} +{"train_loss": 0.017548469826579094, "grad_norm": 0.04846213385462761, "global_step": 13781, "epoch": 130, "lr": 8.422303407010447e-05} +{"train_loss": 0.016674652695655823, "grad_norm": 0.050771474838256836, "global_step": 13782, "epoch": 130, "lr": 8.42208732765684e-05} +{"train_loss": 0.017003322020173073, "grad_norm": 0.04899255186319351, "global_step": 13783, "epoch": 130, "lr": 8.421871236279507e-05} +{"train_loss": 0.016446553170681, "grad_norm": 0.11912935227155685, "global_step": 13784, "epoch": 130, "lr": 8.421655132879206e-05} +{"train_loss": 0.014046978205442429, "grad_norm": 0.06892793625593185, "global_step": 13785, "epoch": 130, "lr": 8.421439017456696e-05} +{"train_loss": 0.017256036400794983, "grad_norm": 0.08079838752746582, "global_step": 13786, "epoch": 130, "lr": 8.421222890012738e-05} +{"train_loss": 0.01688362844288349, "grad_norm": 0.06288415938615799, "global_step": 13787, "epoch": 130, "lr": 8.421006750548092e-05} +{"train_loss": 0.016016367822885513, "grad_norm": 0.10922789573669434, "global_step": 13788, "epoch": 130, "lr": 8.420790599063515e-05} +{"train_loss": 0.017285596579313278, "grad_norm": 0.06657387316226959, "global_step": 13789, "epoch": 130, "lr": 8.420574435559766e-05} +{"train_loss": 0.016165951266884804, "grad_norm": 0.07355719804763794, "global_step": 13790, "epoch": 130, "lr": 8.420358260037608e-05} +{"train_loss": 0.017542464658617973, "grad_norm": 0.09363220632076263, "global_step": 13791, "epoch": 130, "lr": 8.420142072497798e-05} +{"train_loss": 0.014748521149158478, "grad_norm": 0.08407655358314514, "global_step": 13792, "epoch": 130, "lr": 8.419925872941097e-05} +{"train_loss": 0.016543462872505188, "grad_norm": 0.08179795742034912, "global_step": 13793, "epoch": 130, "lr": 8.419709661368262e-05} +{"train_loss": 0.017950130626559258, "grad_norm": 0.05119328945875168, "global_step": 13794, "epoch": 130, "lr": 8.419493437780056e-05} +{"train_loss": 0.019945263862609863, "grad_norm": 0.09399687498807907, "global_step": 13795, "epoch": 130, "lr": 8.419277202177235e-05} +{"train_loss": 0.016853567212820053, "grad_norm": 0.16989979147911072, "global_step": 13796, "epoch": 130, "lr": 8.419060954560563e-05} +{"train_loss": 0.0143332090228796, "grad_norm": 0.147539421916008, "global_step": 13797, "epoch": 130, "lr": 8.418844694930797e-05} +{"train_loss": 0.017639830708503723, "grad_norm": 0.05795538052916527, "global_step": 13798, "epoch": 130, "lr": 8.418628423288699e-05} +{"train_loss": 0.019178466871380806, "grad_norm": 0.10359570384025574, "global_step": 13799, "epoch": 130, "lr": 8.418412139635025e-05} +{"train_loss": 0.017410416156053543, "grad_norm": 0.08067475259304047, "global_step": 13800, "epoch": 130, "lr": 8.41819584397054e-05} +{"train_loss": 0.017517700791358948, "grad_norm": 0.058571189641952515, "global_step": 13801, "epoch": 130, "lr": 8.417979536295999e-05} +{"train_loss": 0.015507274307310581, "grad_norm": 0.04897066205739975, "global_step": 13802, "epoch": 130, "lr": 8.417763216612165e-05} +{"train_loss": 0.014921765774488449, "grad_norm": 0.0651402547955513, "global_step": 13803, "epoch": 130, "lr": 8.417546884919796e-05} +{"train_loss": 0.018066994845867157, "grad_norm": 0.12110409885644913, "global_step": 13804, "epoch": 130, "lr": 8.417330541219657e-05} +{"train_loss": 0.015511468052864075, "grad_norm": 0.05348127335309982, "global_step": 13805, "epoch": 130, "lr": 8.417114185512502e-05} +{"train_loss": 0.019840415567159653, "grad_norm": 0.0740261971950531, "global_step": 13806, "epoch": 130, "lr": 8.416897817799094e-05} +{"train_loss": 0.014544602483510971, "grad_norm": 0.10008495301008224, "global_step": 13807, "epoch": 130, "lr": 8.416681438080194e-05} +{"train_loss": 0.015304626896977425, "grad_norm": 0.06351728737354279, "global_step": 13808, "epoch": 130, "lr": 8.416465046356561e-05} +{"train_loss": 0.0165903028100729, "grad_norm": 0.05515376478433609, "global_step": 13809, "epoch": 130, "lr": 8.416248642628955e-05} +{"train_loss": 0.016201701015233994, "grad_norm": 0.05392054468393326, "global_step": 13810, "epoch": 130, "lr": 8.416032226898138e-05} +{"train_loss": 0.016661647707223892, "grad_norm": 0.060095980763435364, "global_step": 13811, "epoch": 130, "lr": 8.415815799164868e-05} +{"train_loss": 0.018063271418213844, "grad_norm": 0.06876620650291443, "global_step": 13812, "epoch": 130, "lr": 8.415599359429908e-05} +{"train_loss": 0.0162340197712183, "grad_norm": 0.11900614947080612, "global_step": 13813, "epoch": 130, "lr": 8.415382907694018e-05} +{"train_loss": 0.01646919921040535, "grad_norm": 0.056759320199489594, "global_step": 13814, "epoch": 130, "lr": 8.415166443957956e-05} +{"train_loss": 0.01866872049868107, "grad_norm": 0.09969902783632278, "global_step": 13815, "epoch": 130, "lr": 8.414949968222485e-05} +{"train_loss": 0.017929397523403168, "grad_norm": 0.05037691816687584, "global_step": 13816, "epoch": 130, "lr": 8.414733480488364e-05} +{"train_loss": 0.017008017748594284, "grad_norm": 0.06231037154793739, "global_step": 13817, "epoch": 130, "lr": 8.414516980756355e-05} +{"train_loss": 0.016086865216493607, "grad_norm": 0.08277972042560577, "global_step": 13818, "epoch": 130, "lr": 8.414300469027219e-05} +{"train_loss": 0.01515248604118824, "grad_norm": 0.053020503371953964, "global_step": 13819, "epoch": 130, "lr": 8.414083945301717e-05} +{"train_loss": 0.020718008279800415, "grad_norm": 0.11075334250926971, "global_step": 13820, "epoch": 130, "lr": 8.413867409580607e-05} +{"train_loss": 0.0179275032132864, "grad_norm": 0.06489845365285873, "global_step": 13821, "epoch": 130, "lr": 8.41365086186465e-05} +{"train_loss": 0.01689145341515541, "grad_norm": 0.08271411061286926, "global_step": 13822, "epoch": 130, "lr": 8.413434302154612e-05} +{"train_loss": 0.017137670889496803, "grad_norm": 0.07438594102859497, "global_step": 13823, "epoch": 130, "lr": 8.413217730451249e-05} +{"train_loss": 0.015832938253879547, "grad_norm": 0.05552612245082855, "global_step": 13824, "epoch": 130, "lr": 8.413001146755322e-05} +{"train_loss": 0.01953006349503994, "grad_norm": 0.0761464312672615, "global_step": 13825, "epoch": 130, "lr": 8.412784551067595e-05} +{"train_loss": 0.017486348748207092, "grad_norm": 0.08332333713769913, "global_step": 13826, "epoch": 130, "lr": 8.412567943388826e-05} +{"train_loss": 0.018927130848169327, "grad_norm": 0.05671841278672218, "global_step": 13827, "epoch": 130, "lr": 8.412351323719778e-05} +{"train_loss": 0.018694616854190826, "grad_norm": 0.06813500821590424, "global_step": 13828, "epoch": 130, "lr": 8.412134692061211e-05} +{"train_loss": 0.01754000596702099, "grad_norm": 0.11040513962507248, "global_step": 13829, "epoch": 130, "lr": 8.411918048413888e-05} +{"train_loss": 0.01696932315826416, "grad_norm": 0.08384136855602264, "global_step": 13830, "epoch": 130, "lr": 8.411701392778566e-05} +{"train_loss": 0.018747922033071518, "grad_norm": 0.10057754814624786, "global_step": 13831, "epoch": 130, "lr": 8.411484725156012e-05} +{"train_loss": 0.015496544539928436, "grad_norm": 0.09986057877540588, "global_step": 13832, "epoch": 130, "lr": 8.411268045546983e-05} +{"train_loss": 0.01830064132809639, "grad_norm": 0.09819821268320084, "global_step": 13833, "epoch": 130, "lr": 8.411051353952242e-05} +{"train_loss": 0.016211654990911484, "grad_norm": 0.05982489138841629, "global_step": 13834, "epoch": 130, "lr": 8.41083465037255e-05} +{"train_loss": 0.013639147393405437, "grad_norm": 0.07808312028646469, "global_step": 13835, "epoch": 130, "lr": 8.410617934808667e-05} +{"train_loss": 0.016532612964510918, "grad_norm": 0.0648849755525589, "global_step": 13836, "epoch": 130, "lr": 8.410401207261357e-05} +{"train_loss": 0.015575789846479893, "grad_norm": 0.07399734854698181, "global_step": 13837, "epoch": 130, "lr": 8.410184467731378e-05} +{"train_loss": 0.016591565683484077, "grad_norm": 0.07090576738119125, "global_step": 13838, "epoch": 130, "lr": 8.409967716219498e-05} +{"train_loss": 0.017378240823745728, "grad_norm": 0.07738771289587021, "global_step": 13839, "epoch": 130, "lr": 8.409750952726472e-05} +{"train_loss": 0.014128731563687325, "grad_norm": 0.059716105461120605, "global_step": 13840, "epoch": 130, "lr": 8.409534177253065e-05} +{"train_loss": 0.01648688316345215, "grad_norm": 0.05492496117949486, "global_step": 13841, "epoch": 130, "lr": 8.409317389800035e-05} +{"train_loss": 0.015348158776760101, "grad_norm": 0.08344036340713501, "global_step": 13842, "epoch": 130, "lr": 8.409100590368149e-05} +{"train_loss": 0.01986224576830864, "grad_norm": 0.11967736482620239, "global_step": 13843, "epoch": 130, "lr": 8.408883778958166e-05} +{"train_loss": 0.015294233337044716, "grad_norm": 0.05210353806614876, "global_step": 13844, "epoch": 130, "lr": 8.408666955570848e-05} +{"train_loss": 0.01742541231215, "grad_norm": 0.05108574777841568, "global_step": 13845, "epoch": 130, "lr": 8.408450120206956e-05} +{"train_loss": 0.01972290128469467, "grad_norm": 0.07386249303817749, "global_step": 13846, "epoch": 130, "lr": 8.408233272867252e-05} +{"train_loss": 0.015719447284936905, "grad_norm": 0.05926678702235222, "global_step": 13847, "epoch": 130, "lr": 8.408016413552499e-05} +{"train_loss": 0.01586516946554184, "grad_norm": 0.07341378182172775, "global_step": 13848, "epoch": 130, "lr": 8.40779954226346e-05} +{"train_loss": 0.016205884516239166, "grad_norm": 0.09817254543304443, "global_step": 13849, "epoch": 130, "lr": 8.407582659000892e-05} +{"train_loss": 0.015425856225192547, "grad_norm": 0.052116818726062775, "global_step": 13850, "epoch": 130, "lr": 8.407365763765563e-05} +{"train_loss": 0.01988944411277771, "grad_norm": 0.05578594654798508, "global_step": 13851, "epoch": 130, "lr": 8.407148856558231e-05} +{"train_loss": 0.013381827622652054, "grad_norm": 0.04912935197353363, "global_step": 13852, "epoch": 130, "lr": 8.406931937379661e-05} +{"train_loss": 0.016045214608311653, "grad_norm": 0.0973508283495903, "global_step": 13853, "epoch": 130, "lr": 8.406715006230613e-05} +{"train_loss": 0.015372579917311668, "grad_norm": 0.0573277585208416, "global_step": 13854, "epoch": 130, "lr": 8.406498063111852e-05} +{"train_loss": 0.01767037808895111, "grad_norm": 0.08624207228422165, "global_step": 13855, "epoch": 130, "lr": 8.406281108024135e-05} +{"train_loss": 0.020607158541679382, "grad_norm": 0.06151289865374565, "global_step": 13856, "epoch": 130, "lr": 8.406064140968229e-05} +{"train_loss": 0.020784294232726097, "grad_norm": 0.061992332339286804, "global_step": 13857, "epoch": 130, "lr": 8.405847161944895e-05} +{"train_loss": 0.016513345763087273, "grad_norm": 0.06483600288629532, "global_step": 13858, "epoch": 130, "lr": 8.405630170954897e-05} +{"train_loss": 0.017469527199864388, "grad_norm": 0.08308174461126328, "global_step": 13859, "epoch": 130, "lr": 8.405413167998993e-05} +{"train_loss": 0.016216307878494263, "grad_norm": 0.05868421122431755, "global_step": 13860, "epoch": 130, "lr": 8.40519615307795e-05} +{"train_loss": 0.014814727008342743, "grad_norm": 0.05133488401770592, "global_step": 13861, "epoch": 130, "lr": 8.404979126192527e-05} +{"train_loss": 0.016995063051581383, "grad_norm": 0.0884307250380516, "global_step": 13862, "epoch": 130, "lr": 8.404762087343489e-05} +{"train_loss": 0.017059847712516785, "grad_norm": 0.06964833289384842, "global_step": 13863, "epoch": 130, "lr": 8.404545036531598e-05} +{"train_loss": 0.015657734125852585, "grad_norm": 0.08735933154821396, "global_step": 13864, "epoch": 130, "lr": 8.404327973757616e-05} +{"train_loss": 0.01752736233174801, "grad_norm": 0.07397793978452682, "global_step": 13865, "epoch": 130, "lr": 8.404110899022303e-05} +{"train_loss": 0.014524087309837341, "grad_norm": 0.10818430781364441, "global_step": 13866, "epoch": 130, "lr": 8.403893812326429e-05} +{"train_loss": 0.017003998160362244, "grad_norm": 0.05526438355445862, "global_step": 13867, "epoch": 130, "lr": 8.40367671367075e-05} +{"train_loss": 0.019107569009065628, "grad_norm": 0.11938222497701645, "global_step": 13868, "epoch": 130, "lr": 8.403459603056033e-05} +{"train_loss": 0.01601199060678482, "grad_norm": 0.05551037937402725, "global_step": 13869, "epoch": 130, "lr": 8.403242480483039e-05} +{"train_loss": 0.017137249931693077, "grad_norm": 0.10952359437942505, "global_step": 13870, "epoch": 130, "lr": 8.403025345952529e-05} +{"train_loss": 0.016198644414544106, "grad_norm": 0.04661119356751442, "global_step": 13871, "epoch": 130, "lr": 8.40280819946527e-05} +{"train_loss": 0.016225501894950867, "grad_norm": 0.09152977168560028, "global_step": 13872, "epoch": 130, "lr": 8.402591041022022e-05} +{"train_loss": 0.016180992126464844, "grad_norm": 0.09743886440992355, "global_step": 13873, "epoch": 130, "lr": 8.402373870623548e-05} +{"train_loss": 0.01840389519929886, "grad_norm": 0.06960345804691315, "global_step": 13874, "epoch": 130, "lr": 8.402156688270613e-05} +{"train_loss": 0.014929628930985928, "grad_norm": 0.08778958022594452, "global_step": 13875, "epoch": 130, "lr": 8.40193949396398e-05} +{"train_loss": 0.015944186598062515, "grad_norm": 0.04995475709438324, "global_step": 13876, "epoch": 130, "lr": 8.401722287704408e-05} +{"train_loss": 0.015431482344865799, "grad_norm": 0.08306325227022171, "global_step": 13877, "epoch": 130, "lr": 8.401505069492665e-05} +{"train_loss": 0.015803813934326172, "grad_norm": 0.11332743614912033, "global_step": 13878, "epoch": 130, "lr": 8.401287839329512e-05} +{"train_loss": 0.017424121499061584, "grad_norm": 0.05232128128409386, "global_step": 13879, "epoch": 130, "lr": 8.401070597215715e-05} +{"train_loss": 0.018407829105854034, "grad_norm": 0.08604326844215393, "global_step": 13880, "epoch": 130, "lr": 8.400853343152032e-05} +{"train_loss": 0.01657278463244438, "grad_norm": 0.052496977150440216, "global_step": 13881, "epoch": 130, "lr": 8.400636077139232e-05} +{"train_loss": 0.017673533409833908, "grad_norm": 0.048522256314754486, "global_step": 13882, "epoch": 130, "lr": 8.400418799178074e-05} +{"train_loss": 0.01525149680674076, "grad_norm": 0.10389591008424759, "global_step": 13883, "epoch": 130, "lr": 8.400201509269324e-05} +{"train_loss": 0.02033461444079876, "grad_norm": 0.05532756820321083, "global_step": 13884, "epoch": 130, "lr": 8.399984207413742e-05} +{"train_loss": 0.016924285567102005, "grad_norm": 0.09898628294467926, "global_step": 13885, "epoch": 130, "lr": 8.399766893612096e-05, "train_batch_mse_error": 0.015315184369683266, "val_loss": 0.01905949041247368} +{"train_loss": 0.015194093808531761, "grad_norm": 0.07734058052301407, "global_step": 13886, "epoch": 131, "lr": 8.399549567865148e-05} +{"train_loss": 0.015576496720314026, "grad_norm": 0.048988621681928635, "global_step": 13887, "epoch": 131, "lr": 8.399332230173662e-05} +{"train_loss": 0.014983815141022205, "grad_norm": 0.057422928512096405, "global_step": 13888, "epoch": 131, "lr": 8.399114880538399e-05} +{"train_loss": 0.018120981752872467, "grad_norm": 0.08253110945224762, "global_step": 13889, "epoch": 131, "lr": 8.398897518960125e-05} +{"train_loss": 0.015784787014126778, "grad_norm": 0.08184944093227386, "global_step": 13890, "epoch": 131, "lr": 8.398680145439603e-05} +{"train_loss": 0.016489598900079727, "grad_norm": 0.06856007874011993, "global_step": 13891, "epoch": 131, "lr": 8.398462759977598e-05} +{"train_loss": 0.019342564046382904, "grad_norm": 0.06326055526733398, "global_step": 13892, "epoch": 131, "lr": 8.398245362574874e-05} +{"train_loss": 0.01508314162492752, "grad_norm": 0.07125867903232574, "global_step": 13893, "epoch": 131, "lr": 8.398027953232191e-05} +{"train_loss": 0.015323040075600147, "grad_norm": 0.1451437771320343, "global_step": 13894, "epoch": 131, "lr": 8.397810531950316e-05} +{"train_loss": 0.017765669152140617, "grad_norm": 0.06338155269622803, "global_step": 13895, "epoch": 131, "lr": 8.397593098730015e-05} +{"train_loss": 0.017597148194909096, "grad_norm": 0.05839034169912338, "global_step": 13896, "epoch": 131, "lr": 8.397375653572047e-05} +{"train_loss": 0.014763642102479935, "grad_norm": 0.0915183275938034, "global_step": 13897, "epoch": 131, "lr": 8.39715819647718e-05} +{"train_loss": 0.01673165336251259, "grad_norm": 0.06340259313583374, "global_step": 13898, "epoch": 131, "lr": 8.396940727446174e-05} +{"train_loss": 0.01504561211913824, "grad_norm": 0.11716407537460327, "global_step": 13899, "epoch": 131, "lr": 8.396723246479797e-05} +{"train_loss": 0.012470319867134094, "grad_norm": 0.07766783237457275, "global_step": 13900, "epoch": 131, "lr": 8.396505753578812e-05} +{"train_loss": 0.017454899847507477, "grad_norm": 0.10727041214704514, "global_step": 13901, "epoch": 131, "lr": 8.396288248743982e-05} +{"train_loss": 0.01618579402565956, "grad_norm": 0.07878097146749496, "global_step": 13902, "epoch": 131, "lr": 8.396070731976073e-05} +{"train_loss": 0.016187939792871475, "grad_norm": 0.08039390295743942, "global_step": 13903, "epoch": 131, "lr": 8.395853203275849e-05} +{"train_loss": 0.015832144767045975, "grad_norm": 0.04419427737593651, "global_step": 13904, "epoch": 131, "lr": 8.395635662644072e-05} +{"train_loss": 0.01727285608649254, "grad_norm": 0.07409535348415375, "global_step": 13905, "epoch": 131, "lr": 8.395418110081509e-05} +{"train_loss": 0.01756395772099495, "grad_norm": 0.06757991760969162, "global_step": 13906, "epoch": 131, "lr": 8.395200545588922e-05} +{"train_loss": 0.016248514875769615, "grad_norm": 0.060588959604501724, "global_step": 13907, "epoch": 131, "lr": 8.39498296916708e-05} +{"train_loss": 0.014312097802758217, "grad_norm": 0.06071449816226959, "global_step": 13908, "epoch": 131, "lr": 8.39476538081674e-05} +{"train_loss": 0.015300096943974495, "grad_norm": 0.06339775025844574, "global_step": 13909, "epoch": 131, "lr": 8.394547780538674e-05} +{"train_loss": 0.01757904142141342, "grad_norm": 0.12995488941669464, "global_step": 13910, "epoch": 131, "lr": 8.394330168333643e-05} +{"train_loss": 0.018246516585350037, "grad_norm": 0.11476462334394455, "global_step": 13911, "epoch": 131, "lr": 8.39411254420241e-05} +{"train_loss": 0.018775880336761475, "grad_norm": 0.060472097247838974, "global_step": 13912, "epoch": 131, "lr": 8.393894908145743e-05} +{"train_loss": 0.01743861474096775, "grad_norm": 0.06982146948575974, "global_step": 13913, "epoch": 131, "lr": 8.393677260164407e-05} +{"train_loss": 0.01813412830233574, "grad_norm": 0.08597934991121292, "global_step": 13914, "epoch": 131, "lr": 8.393459600259163e-05} +{"train_loss": 0.01985316351056099, "grad_norm": 0.1097075343132019, "global_step": 13915, "epoch": 131, "lr": 8.393241928430778e-05} +{"train_loss": 0.018055737018585205, "grad_norm": 0.05482149496674538, "global_step": 13916, "epoch": 131, "lr": 8.393024244680016e-05} +{"train_loss": 0.017768047749996185, "grad_norm": 0.11358409374952316, "global_step": 13917, "epoch": 131, "lr": 8.392806549007643e-05} +{"train_loss": 0.020257458090782166, "grad_norm": 0.08779961615800858, "global_step": 13918, "epoch": 131, "lr": 8.392588841414422e-05} +{"train_loss": 0.019055193290114403, "grad_norm": 0.17566396296024323, "global_step": 13919, "epoch": 131, "lr": 8.392371121901122e-05} +{"train_loss": 0.017361339181661606, "grad_norm": 0.0664396584033966, "global_step": 13920, "epoch": 131, "lr": 8.392153390468503e-05} +{"train_loss": 0.015955850481987, "grad_norm": 0.11446157097816467, "global_step": 13921, "epoch": 131, "lr": 8.391935647117333e-05} +{"train_loss": 0.0157453790307045, "grad_norm": 0.08358518779277802, "global_step": 13922, "epoch": 131, "lr": 8.391717891848375e-05} +{"train_loss": 0.016027064993977547, "grad_norm": 0.04685898497700691, "global_step": 13923, "epoch": 131, "lr": 8.391500124662396e-05} +{"train_loss": 0.015781812369823456, "grad_norm": 0.15902820229530334, "global_step": 13924, "epoch": 131, "lr": 8.39128234556016e-05} +{"train_loss": 0.016078371554613113, "grad_norm": 0.1512330174446106, "global_step": 13925, "epoch": 131, "lr": 8.391064554542435e-05} +{"train_loss": 0.01753884367644787, "grad_norm": 0.0680619478225708, "global_step": 13926, "epoch": 131, "lr": 8.390846751609982e-05} +{"train_loss": 0.0162801593542099, "grad_norm": 0.046628743410110474, "global_step": 13927, "epoch": 131, "lr": 8.390628936763568e-05} +{"train_loss": 0.017742585390806198, "grad_norm": 0.05037117749452591, "global_step": 13928, "epoch": 131, "lr": 8.39041111000396e-05} +{"train_loss": 0.016284508630633354, "grad_norm": 0.09636475890874863, "global_step": 13929, "epoch": 131, "lr": 8.390193271331921e-05} +{"train_loss": 0.021014245226979256, "grad_norm": 0.13027982413768768, "global_step": 13930, "epoch": 131, "lr": 8.389975420748216e-05} +{"train_loss": 0.017352337017655373, "grad_norm": 0.053925979882478714, "global_step": 13931, "epoch": 131, "lr": 8.389757558253613e-05} +{"train_loss": 0.01578262820839882, "grad_norm": 0.0662810429930687, "global_step": 13932, "epoch": 131, "lr": 8.389539683848876e-05} +{"train_loss": 0.018215840682387352, "grad_norm": 0.06537684053182602, "global_step": 13933, "epoch": 131, "lr": 8.389321797534771e-05} +{"train_loss": 0.017020732164382935, "grad_norm": 0.06773126870393753, "global_step": 13934, "epoch": 131, "lr": 8.389103899312061e-05} +{"train_loss": 0.016737114638090134, "grad_norm": 0.11357879638671875, "global_step": 13935, "epoch": 131, "lr": 8.388885989181515e-05} +{"train_loss": 0.016031436622142792, "grad_norm": 0.07017809897661209, "global_step": 13936, "epoch": 131, "lr": 8.388668067143899e-05} +{"train_loss": 0.015967845916748047, "grad_norm": 0.11489822715520859, "global_step": 13937, "epoch": 131, "lr": 8.388450133199974e-05} +{"train_loss": 0.018018813803792, "grad_norm": 0.10226376354694366, "global_step": 13938, "epoch": 131, "lr": 8.388232187350512e-05} +{"train_loss": 0.021738488227128983, "grad_norm": 0.1263156235218048, "global_step": 13939, "epoch": 131, "lr": 8.388014229596275e-05} +{"train_loss": 0.019972138106822968, "grad_norm": 0.05946063622832298, "global_step": 13940, "epoch": 131, "lr": 8.387796259938027e-05} +{"train_loss": 0.01626937836408615, "grad_norm": 0.0717197060585022, "global_step": 13941, "epoch": 131, "lr": 8.387578278376537e-05} +{"train_loss": 0.017074044793844223, "grad_norm": 0.09442013502120972, "global_step": 13942, "epoch": 131, "lr": 8.387360284912574e-05} +{"train_loss": 0.015383216552436352, "grad_norm": 0.0926777794957161, "global_step": 13943, "epoch": 131, "lr": 8.387142279546896e-05} +{"train_loss": 0.013300543650984764, "grad_norm": 0.05289473384618759, "global_step": 13944, "epoch": 131, "lr": 8.386924262280274e-05} +{"train_loss": 0.015901384875178337, "grad_norm": 0.06102795898914337, "global_step": 13945, "epoch": 131, "lr": 8.386706233113474e-05} +{"train_loss": 0.018303008750081062, "grad_norm": 0.06647071987390518, "global_step": 13946, "epoch": 131, "lr": 8.38648819204726e-05} +{"train_loss": 0.01510272640734911, "grad_norm": 0.051179591566324234, "global_step": 13947, "epoch": 131, "lr": 8.386270139082399e-05} +{"train_loss": 0.01651160977780819, "grad_norm": 0.053261954337358475, "global_step": 13948, "epoch": 131, "lr": 8.38605207421966e-05} +{"train_loss": 0.013874152675271034, "grad_norm": 0.060096241533756256, "global_step": 13949, "epoch": 131, "lr": 8.385833997459803e-05} +{"train_loss": 0.017607320100069046, "grad_norm": 0.08868042379617691, "global_step": 13950, "epoch": 131, "lr": 8.385615908803599e-05} +{"train_loss": 0.01765214279294014, "grad_norm": 0.13209138810634613, "global_step": 13951, "epoch": 131, "lr": 8.385397808251813e-05} +{"train_loss": 0.014802577905356884, "grad_norm": 0.12496228516101837, "global_step": 13952, "epoch": 131, "lr": 8.385179695805212e-05} +{"train_loss": 0.017351847141981125, "grad_norm": 0.05938832834362984, "global_step": 13953, "epoch": 131, "lr": 8.384961571464563e-05} +{"train_loss": 0.01456221379339695, "grad_norm": 0.10319647938013077, "global_step": 13954, "epoch": 131, "lr": 8.384743435230629e-05} +{"train_loss": 0.018086224794387817, "grad_norm": 0.05718264356255531, "global_step": 13955, "epoch": 131, "lr": 8.38452528710418e-05} +{"train_loss": 0.019266579300165176, "grad_norm": 0.06236974522471428, "global_step": 13956, "epoch": 131, "lr": 8.38430712708598e-05} +{"train_loss": 0.018658515065908432, "grad_norm": 0.06579829752445221, "global_step": 13957, "epoch": 131, "lr": 8.384088955176796e-05} +{"train_loss": 0.01953154057264328, "grad_norm": 0.07210850715637207, "global_step": 13958, "epoch": 131, "lr": 8.383870771377396e-05} +{"train_loss": 0.01917269080877304, "grad_norm": 0.10008952766656876, "global_step": 13959, "epoch": 131, "lr": 8.383652575688546e-05} +{"train_loss": 0.014384896494448185, "grad_norm": 0.061314020305871964, "global_step": 13960, "epoch": 131, "lr": 8.383434368111011e-05} +{"train_loss": 0.016970131546258926, "grad_norm": 0.09046566486358643, "global_step": 13961, "epoch": 131, "lr": 8.38321614864556e-05} +{"train_loss": 0.020223010331392288, "grad_norm": 0.08633407205343246, "global_step": 13962, "epoch": 131, "lr": 8.38299791729296e-05} +{"train_loss": 0.01566830277442932, "grad_norm": 0.0785384401679039, "global_step": 13963, "epoch": 131, "lr": 8.382779674053975e-05} +{"train_loss": 0.01748640649020672, "grad_norm": 0.10738710314035416, "global_step": 13964, "epoch": 131, "lr": 8.382561418929373e-05} +{"train_loss": 0.018915491178631783, "grad_norm": 0.05497550964355469, "global_step": 13965, "epoch": 131, "lr": 8.382343151919923e-05} +{"train_loss": 0.017535559833049774, "grad_norm": 0.0756189152598381, "global_step": 13966, "epoch": 131, "lr": 8.382124873026389e-05} +{"train_loss": 0.01796003431081772, "grad_norm": 0.19833539426326752, "global_step": 13967, "epoch": 131, "lr": 8.38190658224954e-05} +{"train_loss": 0.015523762442171574, "grad_norm": 0.07925235480070114, "global_step": 13968, "epoch": 131, "lr": 8.38168827959014e-05} +{"train_loss": 0.01552512589842081, "grad_norm": 0.07289974391460419, "global_step": 13969, "epoch": 131, "lr": 8.38146996504896e-05} +{"train_loss": 0.018738431856036186, "grad_norm": 0.07470839470624924, "global_step": 13970, "epoch": 131, "lr": 8.381251638626764e-05} +{"train_loss": 0.015425462275743484, "grad_norm": 0.08839777112007141, "global_step": 13971, "epoch": 131, "lr": 8.38103330032432e-05} +{"train_loss": 0.014440733939409256, "grad_norm": 0.10243050754070282, "global_step": 13972, "epoch": 131, "lr": 8.380814950142396e-05} +{"train_loss": 0.0181162990629673, "grad_norm": 0.11013345420360565, "global_step": 13973, "epoch": 131, "lr": 8.380596588081758e-05} +{"train_loss": 0.018213972449302673, "grad_norm": 0.15750828385353088, "global_step": 13974, "epoch": 131, "lr": 8.380378214143174e-05} +{"train_loss": 0.017669225111603737, "grad_norm": 0.07740584015846252, "global_step": 13975, "epoch": 131, "lr": 8.380159828327413e-05} +{"train_loss": 0.016406774520874023, "grad_norm": 0.060482364147901535, "global_step": 13976, "epoch": 131, "lr": 8.379941430635237e-05} +{"train_loss": 0.016506316140294075, "grad_norm": 0.11450498551130295, "global_step": 13977, "epoch": 131, "lr": 8.379723021067418e-05} +{"train_loss": 0.015361307188868523, "grad_norm": 0.13197830319404602, "global_step": 13978, "epoch": 131, "lr": 8.379504599624722e-05} +{"train_loss": 0.017559967935085297, "grad_norm": 0.06683333218097687, "global_step": 13979, "epoch": 131, "lr": 8.379286166307917e-05} +{"train_loss": 0.018288936465978622, "grad_norm": 0.07020121067762375, "global_step": 13980, "epoch": 131, "lr": 8.37906772111777e-05} +{"train_loss": 0.014920398592948914, "grad_norm": 0.1302403062582016, "global_step": 13981, "epoch": 131, "lr": 8.378849264055048e-05} +{"train_loss": 0.017064113169908524, "grad_norm": 0.09662748128175735, "global_step": 13982, "epoch": 131, "lr": 8.37863079512052e-05} +{"train_loss": 0.016305670142173767, "grad_norm": 0.04375804588198662, "global_step": 13983, "epoch": 131, "lr": 8.378412314314952e-05} +{"train_loss": 0.020030898973345757, "grad_norm": 0.07433784008026123, "global_step": 13984, "epoch": 131, "lr": 8.378193821639114e-05} +{"train_loss": 0.012768927961587906, "grad_norm": 0.07334870845079422, "global_step": 13985, "epoch": 131, "lr": 8.37797531709377e-05} +{"train_loss": 0.021532950922846794, "grad_norm": 0.07455824315547943, "global_step": 13986, "epoch": 131, "lr": 8.37775680067969e-05} +{"train_loss": 0.013937745243310928, "grad_norm": 0.1005139872431755, "global_step": 13987, "epoch": 131, "lr": 8.377538272397643e-05} +{"train_loss": 0.018124636262655258, "grad_norm": 0.07118973135948181, "global_step": 13988, "epoch": 131, "lr": 8.377319732248393e-05} +{"train_loss": 0.017246894538402557, "grad_norm": 0.08596042543649673, "global_step": 13989, "epoch": 131, "lr": 8.377101180232712e-05} +{"train_loss": 0.01587887853384018, "grad_norm": 0.04460327327251434, "global_step": 13990, "epoch": 131, "lr": 8.376882616351367e-05} +{"train_loss": 0.01690327207434852, "grad_norm": 0.06943884491920471, "global_step": 13991, "epoch": 131, "lr": 8.376664040605122e-05, "val_loss": 0.0196016114205122} +{"train_loss": 0.019519031047821045, "grad_norm": 0.06201939657330513, "global_step": 13992, "epoch": 132, "lr": 8.37644545299475e-05} +{"train_loss": 0.017404647544026375, "grad_norm": 0.053266916424036026, "global_step": 13993, "epoch": 132, "lr": 8.376226853521017e-05} +{"train_loss": 0.017710300162434578, "grad_norm": 0.0828772634267807, "global_step": 13994, "epoch": 132, "lr": 8.376008242184692e-05} +{"train_loss": 0.01726333051919937, "grad_norm": 0.053296640515327454, "global_step": 13995, "epoch": 132, "lr": 8.37578961898654e-05} +{"train_loss": 0.0156613290309906, "grad_norm": 0.06798050552606583, "global_step": 13996, "epoch": 132, "lr": 8.375570983927334e-05} +{"train_loss": 0.021036602556705475, "grad_norm": 0.124159075319767, "global_step": 13997, "epoch": 132, "lr": 8.375352337007836e-05} +{"train_loss": 0.022281857207417488, "grad_norm": 0.13955891132354736, "global_step": 13998, "epoch": 132, "lr": 8.375133678228822e-05} +{"train_loss": 0.015335219912230968, "grad_norm": 0.054328955709934235, "global_step": 13999, "epoch": 132, "lr": 8.374915007591053e-05} +{"train_loss": 0.01597226783633232, "grad_norm": 0.14735980331897736, "global_step": 14000, "epoch": 132, "lr": 8.374696325095301e-05} +{"train_loss": 0.014926927164196968, "grad_norm": 0.04977243393659592, "global_step": 14001, "epoch": 132, "lr": 8.374477630742335e-05} +{"train_loss": 0.015493499115109444, "grad_norm": 0.052954308688640594, "global_step": 14002, "epoch": 132, "lr": 8.37425892453292e-05} +{"train_loss": 0.016498249024152756, "grad_norm": 0.061529431492090225, "global_step": 14003, "epoch": 132, "lr": 8.374040206467828e-05} +{"train_loss": 0.016216212883591652, "grad_norm": 0.08023587614297867, "global_step": 14004, "epoch": 132, "lr": 8.373821476547827e-05} +{"train_loss": 0.01505520474165678, "grad_norm": 0.06119821220636368, "global_step": 14005, "epoch": 132, "lr": 8.373602734773683e-05} +{"train_loss": 0.017057474702596664, "grad_norm": 0.09453413635492325, "global_step": 14006, "epoch": 132, "lr": 8.373383981146166e-05} +{"train_loss": 0.015619097277522087, "grad_norm": 0.07089310884475708, "global_step": 14007, "epoch": 132, "lr": 8.373165215666045e-05} +{"train_loss": 0.01611950248479843, "grad_norm": 0.05451677739620209, "global_step": 14008, "epoch": 132, "lr": 8.372946438334089e-05} +{"train_loss": 0.016457054764032364, "grad_norm": 0.04511244222521782, "global_step": 14009, "epoch": 132, "lr": 8.372727649151067e-05} +{"train_loss": 0.01599789224565029, "grad_norm": 0.0741191953420639, "global_step": 14010, "epoch": 132, "lr": 8.372508848117745e-05} +{"train_loss": 0.018870016559958458, "grad_norm": 0.050871316343545914, "global_step": 14011, "epoch": 132, "lr": 8.372290035234893e-05} +{"train_loss": 0.01599806547164917, "grad_norm": 0.08334814012050629, "global_step": 14012, "epoch": 132, "lr": 8.372071210503282e-05} +{"train_loss": 0.01647239178419113, "grad_norm": 0.061315689235925674, "global_step": 14013, "epoch": 132, "lr": 8.371852373923678e-05} +{"train_loss": 0.01681635156273842, "grad_norm": 0.07988882064819336, "global_step": 14014, "epoch": 132, "lr": 8.371633525496853e-05} +{"train_loss": 0.016714002937078476, "grad_norm": 0.0848853662610054, "global_step": 14015, "epoch": 132, "lr": 8.371414665223573e-05} +{"train_loss": 0.018279176205396652, "grad_norm": 0.07012264430522919, "global_step": 14016, "epoch": 132, "lr": 8.371195793104608e-05} +{"train_loss": 0.018268615007400513, "grad_norm": 0.09831615537405014, "global_step": 14017, "epoch": 132, "lr": 8.370976909140727e-05} +{"train_loss": 0.0205826498568058, "grad_norm": 0.057785261422395706, "global_step": 14018, "epoch": 132, "lr": 8.370758013332697e-05} +{"train_loss": 0.01667981967329979, "grad_norm": 0.081479012966156, "global_step": 14019, "epoch": 132, "lr": 8.370539105681295e-05} +{"train_loss": 0.018455177545547485, "grad_norm": 0.08520098030567169, "global_step": 14020, "epoch": 132, "lr": 8.370320186187279e-05} +{"train_loss": 0.016914207488298416, "grad_norm": 0.05637253075838089, "global_step": 14021, "epoch": 132, "lr": 8.370101254851426e-05} +{"train_loss": 0.015170411206781864, "grad_norm": 0.06387005001306534, "global_step": 14022, "epoch": 132, "lr": 8.369882311674502e-05} +{"train_loss": 0.015323806554079056, "grad_norm": 0.04299028590321541, "global_step": 14023, "epoch": 132, "lr": 8.369663356657277e-05} +{"train_loss": 0.015315850265324116, "grad_norm": 0.13053441047668457, "global_step": 14024, "epoch": 132, "lr": 8.36944438980052e-05} +{"train_loss": 0.015741970390081406, "grad_norm": 0.13823600113391876, "global_step": 14025, "epoch": 132, "lr": 8.369225411105003e-05} +{"train_loss": 0.017905343323946, "grad_norm": 0.0673619955778122, "global_step": 14026, "epoch": 132, "lr": 8.36900642057149e-05} +{"train_loss": 0.01802002266049385, "grad_norm": 0.07344730198383331, "global_step": 14027, "epoch": 132, "lr": 8.368787418200756e-05} +{"train_loss": 0.015174685046076775, "grad_norm": 0.05441755801439285, "global_step": 14028, "epoch": 132, "lr": 8.368568403993568e-05} +{"train_loss": 0.014758787117898464, "grad_norm": 0.06072840839624405, "global_step": 14029, "epoch": 132, "lr": 8.368349377950695e-05} +{"train_loss": 0.015621266327798367, "grad_norm": 0.08169553428888321, "global_step": 14030, "epoch": 132, "lr": 8.368130340072906e-05} +{"train_loss": 0.017233096063137054, "grad_norm": 0.08172856271266937, "global_step": 14031, "epoch": 132, "lr": 8.367911290360973e-05} +{"train_loss": 0.015411545522511005, "grad_norm": 0.07083942741155624, "global_step": 14032, "epoch": 132, "lr": 8.367692228815664e-05} +{"train_loss": 0.019823646172881126, "grad_norm": 0.0786370113492012, "global_step": 14033, "epoch": 132, "lr": 8.36747315543775e-05} +{"train_loss": 0.015421545132994652, "grad_norm": 0.06928879022598267, "global_step": 14034, "epoch": 132, "lr": 8.367254070227998e-05} +{"train_loss": 0.015520141460001469, "grad_norm": 0.0754946693778038, "global_step": 14035, "epoch": 132, "lr": 8.36703497318718e-05} +{"train_loss": 0.016687406226992607, "grad_norm": 0.04784328117966652, "global_step": 14036, "epoch": 132, "lr": 8.366815864316066e-05} +{"train_loss": 0.014837514609098434, "grad_norm": 0.08139455318450928, "global_step": 14037, "epoch": 132, "lr": 8.366596743615426e-05} +{"train_loss": 0.01487836241722107, "grad_norm": 0.05754224956035614, "global_step": 14038, "epoch": 132, "lr": 8.366377611086027e-05} +{"train_loss": 0.01660698652267456, "grad_norm": 0.09818468987941742, "global_step": 14039, "epoch": 132, "lr": 8.366158466728644e-05} +{"train_loss": 0.015074866823852062, "grad_norm": 0.12120236456394196, "global_step": 14040, "epoch": 132, "lr": 8.365939310544042e-05} +{"train_loss": 0.013865621760487556, "grad_norm": 0.05261393263936043, "global_step": 14041, "epoch": 132, "lr": 8.365720142532994e-05} +{"train_loss": 0.01803918182849884, "grad_norm": 0.10057536512613297, "global_step": 14042, "epoch": 132, "lr": 8.365500962696269e-05} +{"train_loss": 0.016174038872122765, "grad_norm": 0.1283479928970337, "global_step": 14043, "epoch": 132, "lr": 8.365281771034637e-05} +{"train_loss": 0.018543805927038193, "grad_norm": 0.11892879009246826, "global_step": 14044, "epoch": 132, "lr": 8.365062567548867e-05} +{"train_loss": 0.01717177778482437, "grad_norm": 0.05180491507053375, "global_step": 14045, "epoch": 132, "lr": 8.364843352239733e-05} +{"train_loss": 0.015697428956627846, "grad_norm": 0.050137296319007874, "global_step": 14046, "epoch": 132, "lr": 8.364624125108e-05} +{"train_loss": 0.018710967153310776, "grad_norm": 0.0846097394824028, "global_step": 14047, "epoch": 132, "lr": 8.364404886154444e-05} +{"train_loss": 0.015138791874051094, "grad_norm": 0.05085830017924309, "global_step": 14048, "epoch": 132, "lr": 8.36418563537983e-05} +{"train_loss": 0.017091255635023117, "grad_norm": 0.05092820152640343, "global_step": 14049, "epoch": 132, "lr": 8.363966372784932e-05} +{"train_loss": 0.020082267001271248, "grad_norm": 0.09608618170022964, "global_step": 14050, "epoch": 132, "lr": 8.363747098370519e-05} +{"train_loss": 0.017069585621356964, "grad_norm": 0.06717061996459961, "global_step": 14051, "epoch": 132, "lr": 8.363527812137362e-05} +{"train_loss": 0.01691592112183571, "grad_norm": 0.06081942468881607, "global_step": 14052, "epoch": 132, "lr": 8.36330851408623e-05} +{"train_loss": 0.017443832010030746, "grad_norm": 0.07228363305330276, "global_step": 14053, "epoch": 132, "lr": 8.363089204217896e-05} +{"train_loss": 0.017201017588377, "grad_norm": 0.056916676461696625, "global_step": 14054, "epoch": 132, "lr": 8.362869882533128e-05} +{"train_loss": 0.017610276117920876, "grad_norm": 0.057477518916130066, "global_step": 14055, "epoch": 132, "lr": 8.362650549032697e-05} +{"train_loss": 0.016234973445534706, "grad_norm": 0.09808868914842606, "global_step": 14056, "epoch": 132, "lr": 8.362431203717375e-05} +{"train_loss": 0.016494646668434143, "grad_norm": 0.10388578474521637, "global_step": 14057, "epoch": 132, "lr": 8.362211846587932e-05} +{"train_loss": 0.01918155327439308, "grad_norm": 0.06817406415939331, "global_step": 14058, "epoch": 132, "lr": 8.36199247764514e-05} +{"train_loss": 0.02096322551369667, "grad_norm": 0.0696430504322052, "global_step": 14059, "epoch": 132, "lr": 8.361773096889767e-05} +{"train_loss": 0.019013140350580215, "grad_norm": 0.10325577110052109, "global_step": 14060, "epoch": 132, "lr": 8.361553704322587e-05} +{"train_loss": 0.018037598580121994, "grad_norm": 0.09570102393627167, "global_step": 14061, "epoch": 132, "lr": 8.361334299944367e-05} +{"train_loss": 0.018830912187695503, "grad_norm": 0.09994838386774063, "global_step": 14062, "epoch": 132, "lr": 8.361114883755881e-05} +{"train_loss": 0.0182807557284832, "grad_norm": 0.045282259583473206, "global_step": 14063, "epoch": 132, "lr": 8.3608954557579e-05} +{"train_loss": 0.018985923379659653, "grad_norm": 0.0751623660326004, "global_step": 14064, "epoch": 132, "lr": 8.360676015951191e-05} +{"train_loss": 0.015070737339556217, "grad_norm": 0.06126679480075836, "global_step": 14065, "epoch": 132, "lr": 8.360456564336531e-05} +{"train_loss": 0.016154468059539795, "grad_norm": 0.1480422466993332, "global_step": 14066, "epoch": 132, "lr": 8.360237100914687e-05} +{"train_loss": 0.018483877182006836, "grad_norm": 0.08297183364629745, "global_step": 14067, "epoch": 132, "lr": 8.360017625686431e-05} +{"train_loss": 0.015720639377832413, "grad_norm": 0.05907579883933067, "global_step": 14068, "epoch": 132, "lr": 8.359798138652534e-05} +{"train_loss": 0.01695079915225506, "grad_norm": 0.06565488874912262, "global_step": 14069, "epoch": 132, "lr": 8.35957863981377e-05} +{"train_loss": 0.01400047354400158, "grad_norm": 0.11774086952209473, "global_step": 14070, "epoch": 132, "lr": 8.359359129170905e-05} +{"train_loss": 0.018646633252501488, "grad_norm": 0.08753525465726852, "global_step": 14071, "epoch": 132, "lr": 8.359139606724713e-05} +{"train_loss": 0.01609894260764122, "grad_norm": 0.089415542781353, "global_step": 14072, "epoch": 132, "lr": 8.358920072475966e-05} +{"train_loss": 0.014982729218900204, "grad_norm": 0.06558630615472794, "global_step": 14073, "epoch": 132, "lr": 8.358700526425436e-05} +{"train_loss": 0.016672857105731964, "grad_norm": 0.051588959991931915, "global_step": 14074, "epoch": 132, "lr": 8.358480968573891e-05} +{"train_loss": 0.017855390906333923, "grad_norm": 0.07357344776391983, "global_step": 14075, "epoch": 132, "lr": 8.358261398922105e-05} +{"train_loss": 0.015207016840577126, "grad_norm": 0.07833562046289444, "global_step": 14076, "epoch": 132, "lr": 8.35804181747085e-05} +{"train_loss": 0.01771748811006546, "grad_norm": 0.08749787509441376, "global_step": 14077, "epoch": 132, "lr": 8.357822224220895e-05} +{"train_loss": 0.018094979226589203, "grad_norm": 0.07353304326534271, "global_step": 14078, "epoch": 132, "lr": 8.357602619173013e-05} +{"train_loss": 0.017917079851031303, "grad_norm": 0.17360858619213104, "global_step": 14079, "epoch": 132, "lr": 8.357383002327976e-05} +{"train_loss": 0.016540957614779472, "grad_norm": 0.14688944816589355, "global_step": 14080, "epoch": 132, "lr": 8.357163373686555e-05} +{"train_loss": 0.016126928851008415, "grad_norm": 0.047314949333667755, "global_step": 14081, "epoch": 132, "lr": 8.356943733249522e-05} +{"train_loss": 0.017856042832136154, "grad_norm": 0.07617572695016861, "global_step": 14082, "epoch": 132, "lr": 8.356724081017648e-05} +{"train_loss": 0.018466893583536148, "grad_norm": 0.07135312259197235, "global_step": 14083, "epoch": 132, "lr": 8.356504416991706e-05} +{"train_loss": 0.01691272109746933, "grad_norm": 0.05770464614033699, "global_step": 14084, "epoch": 132, "lr": 8.356284741172469e-05} +{"train_loss": 0.017884615808725357, "grad_norm": 0.07654130458831787, "global_step": 14085, "epoch": 132, "lr": 8.356065053560704e-05} +{"train_loss": 0.017778418958187103, "grad_norm": 0.08803346008062363, "global_step": 14086, "epoch": 132, "lr": 8.355845354157188e-05} +{"train_loss": 0.01593789830803871, "grad_norm": 0.07560016214847565, "global_step": 14087, "epoch": 132, "lr": 8.35562564296269e-05} +{"train_loss": 0.016524329781532288, "grad_norm": 0.07859309762716293, "global_step": 14088, "epoch": 132, "lr": 8.355405919977984e-05} +{"train_loss": 0.015566494315862656, "grad_norm": 0.05870726704597473, "global_step": 14089, "epoch": 132, "lr": 8.355186185203839e-05} +{"train_loss": 0.016077900305390358, "grad_norm": 0.05648590251803398, "global_step": 14090, "epoch": 132, "lr": 8.35496643864103e-05} +{"train_loss": 0.01796717196702957, "grad_norm": 0.0614919438958168, "global_step": 14091, "epoch": 132, "lr": 8.354746680290326e-05} +{"train_loss": 0.014149216935038567, "grad_norm": 0.06083616986870766, "global_step": 14092, "epoch": 132, "lr": 8.354526910152503e-05} +{"train_loss": 0.014591973274946213, "grad_norm": 0.1304168403148651, "global_step": 14093, "epoch": 132, "lr": 8.354307128228331e-05} +{"train_loss": 0.014730464667081833, "grad_norm": 0.12154871225357056, "global_step": 14094, "epoch": 132, "lr": 8.354087334518583e-05} +{"train_loss": 0.017543120309710503, "grad_norm": 0.0663534551858902, "global_step": 14095, "epoch": 132, "lr": 8.353867529024031e-05} +{"train_loss": 0.015164557844400406, "grad_norm": 0.053933560848236084, "global_step": 14096, "epoch": 132, "lr": 8.353647711745445e-05} +{"train_loss": 0.016875725107724376, "grad_norm": 0.09098747372627258, "global_step": 14097, "epoch": 132, "lr": 8.3534278826836e-05, "val_loss": 0.018675697967410088} +{"train_loss": 0.018274180591106415, "grad_norm": 0.12964971363544464, "global_step": 14098, "epoch": 133, "lr": 8.353208041839269e-05} +{"train_loss": 0.01774301379919052, "grad_norm": 0.12050191313028336, "global_step": 14099, "epoch": 133, "lr": 8.352988189213222e-05} +{"train_loss": 0.016493480652570724, "grad_norm": 0.07039251178503036, "global_step": 14100, "epoch": 133, "lr": 8.352768324806234e-05} +{"train_loss": 0.01708979532122612, "grad_norm": 0.06830063462257385, "global_step": 14101, "epoch": 133, "lr": 8.352548448619075e-05} +{"train_loss": 0.016595879569649696, "grad_norm": 0.06883908808231354, "global_step": 14102, "epoch": 133, "lr": 8.35232856065252e-05} +{"train_loss": 0.016496481373906136, "grad_norm": 0.05234977602958679, "global_step": 14103, "epoch": 133, "lr": 8.352108660907339e-05} +{"train_loss": 0.01645169034600258, "grad_norm": 0.14302514493465424, "global_step": 14104, "epoch": 133, "lr": 8.351888749384307e-05} +{"train_loss": 0.019973743706941605, "grad_norm": 0.1323029100894928, "global_step": 14105, "epoch": 133, "lr": 8.351668826084196e-05} +{"train_loss": 0.01438689511269331, "grad_norm": 0.10473550856113434, "global_step": 14106, "epoch": 133, "lr": 8.351448891007776e-05} +{"train_loss": 0.016424141824245453, "grad_norm": 0.124597929418087, "global_step": 14107, "epoch": 133, "lr": 8.351228944155824e-05} +{"train_loss": 0.013992059975862503, "grad_norm": 0.05279233679175377, "global_step": 14108, "epoch": 133, "lr": 8.351008985529111e-05} +{"train_loss": 0.015401216223835945, "grad_norm": 0.08091702312231064, "global_step": 14109, "epoch": 133, "lr": 8.35078901512841e-05} +{"train_loss": 0.015435772016644478, "grad_norm": 0.05706261470913887, "global_step": 14110, "epoch": 133, "lr": 8.350569032954492e-05} +{"train_loss": 0.016047146171331406, "grad_norm": 0.065839022397995, "global_step": 14111, "epoch": 133, "lr": 8.350349039008133e-05} +{"train_loss": 0.01588229276239872, "grad_norm": 0.1424618363380432, "global_step": 14112, "epoch": 133, "lr": 8.350129033290104e-05} +{"train_loss": 0.018663007766008377, "grad_norm": 0.09967995434999466, "global_step": 14113, "epoch": 133, "lr": 8.349909015801179e-05} +{"train_loss": 0.015943292528390884, "grad_norm": 0.1103750541806221, "global_step": 14114, "epoch": 133, "lr": 8.349688986542129e-05} +{"train_loss": 0.019165512174367905, "grad_norm": 0.1076352596282959, "global_step": 14115, "epoch": 133, "lr": 8.349468945513729e-05} +{"train_loss": 0.017029214650392532, "grad_norm": 0.1762973517179489, "global_step": 14116, "epoch": 133, "lr": 8.349248892716753e-05} +{"train_loss": 0.015679722651839256, "grad_norm": 0.1188572496175766, "global_step": 14117, "epoch": 133, "lr": 8.349028828151972e-05} +{"train_loss": 0.016568820923566818, "grad_norm": 0.09091070294380188, "global_step": 14118, "epoch": 133, "lr": 8.348808751820161e-05} +{"train_loss": 0.017430944368243217, "grad_norm": 0.06821231544017792, "global_step": 14119, "epoch": 133, "lr": 8.348588663722092e-05} +{"train_loss": 0.014000574126839638, "grad_norm": 0.07194779068231583, "global_step": 14120, "epoch": 133, "lr": 8.348368563858539e-05} +{"train_loss": 0.016379037871956825, "grad_norm": 0.06294810771942139, "global_step": 14121, "epoch": 133, "lr": 8.348148452230274e-05} +{"train_loss": 0.014414602890610695, "grad_norm": 0.04690004885196686, "global_step": 14122, "epoch": 133, "lr": 8.347928328838073e-05} +{"train_loss": 0.01595776155591011, "grad_norm": 0.07192656397819519, "global_step": 14123, "epoch": 133, "lr": 8.347708193682706e-05} +{"train_loss": 0.01609000563621521, "grad_norm": 0.05372614040970802, "global_step": 14124, "epoch": 133, "lr": 8.347488046764948e-05} +{"train_loss": 0.01889963075518608, "grad_norm": 0.07781343162059784, "global_step": 14125, "epoch": 133, "lr": 8.347267888085575e-05} +{"train_loss": 0.018312644213438034, "grad_norm": 0.06054751202464104, "global_step": 14126, "epoch": 133, "lr": 8.347047717645358e-05} +{"train_loss": 0.01834116503596306, "grad_norm": 0.0917687937617302, "global_step": 14127, "epoch": 133, "lr": 8.34682753544507e-05} +{"train_loss": 0.018070926889777184, "grad_norm": 0.043552812188863754, "global_step": 14128, "epoch": 133, "lr": 8.346607341485486e-05} +{"train_loss": 0.015562161803245544, "grad_norm": 0.06985783576965332, "global_step": 14129, "epoch": 133, "lr": 8.346387135767378e-05} +{"train_loss": 0.01669291779398918, "grad_norm": 0.14780978858470917, "global_step": 14130, "epoch": 133, "lr": 8.346166918291522e-05} +{"train_loss": 0.017707781866192818, "grad_norm": 0.06103741377592087, "global_step": 14131, "epoch": 133, "lr": 8.345946689058692e-05} +{"train_loss": 0.017104800790548325, "grad_norm": 0.0660105049610138, "global_step": 14132, "epoch": 133, "lr": 8.345726448069658e-05} +{"train_loss": 0.02234506979584694, "grad_norm": 0.0971757099032402, "global_step": 14133, "epoch": 133, "lr": 8.345506195325197e-05} +{"train_loss": 0.017298690974712372, "grad_norm": 0.07876235991716385, "global_step": 14134, "epoch": 133, "lr": 8.345285930826083e-05} +{"train_loss": 0.018463145941495895, "grad_norm": 0.06359375268220901, "global_step": 14135, "epoch": 133, "lr": 8.345065654573088e-05} +{"train_loss": 0.016315216198563576, "grad_norm": 0.08365190774202347, "global_step": 14136, "epoch": 133, "lr": 8.344845366566988e-05} +{"train_loss": 0.01811554841697216, "grad_norm": 0.05394511669874191, "global_step": 14137, "epoch": 133, "lr": 8.344625066808556e-05} +{"train_loss": 0.014744298532605171, "grad_norm": 0.07896248996257782, "global_step": 14138, "epoch": 133, "lr": 8.344404755298565e-05} +{"train_loss": 0.017972715198993683, "grad_norm": 0.050199855118989944, "global_step": 14139, "epoch": 133, "lr": 8.34418443203779e-05} +{"train_loss": 0.018741820007562637, "grad_norm": 0.16026538610458374, "global_step": 14140, "epoch": 133, "lr": 8.343964097027006e-05} +{"train_loss": 0.016775067895650864, "grad_norm": 0.07734548300504684, "global_step": 14141, "epoch": 133, "lr": 8.343743750266987e-05} +{"train_loss": 0.016782548278570175, "grad_norm": 0.048002202063798904, "global_step": 14142, "epoch": 133, "lr": 8.343523391758505e-05} +{"train_loss": 0.018342390656471252, "grad_norm": 0.07856707274913788, "global_step": 14143, "epoch": 133, "lr": 8.343303021502338e-05} +{"train_loss": 0.015089038759469986, "grad_norm": 0.053483035415410995, "global_step": 14144, "epoch": 133, "lr": 8.343082639499256e-05} +{"train_loss": 0.017650136724114418, "grad_norm": 0.12156255543231964, "global_step": 14145, "epoch": 133, "lr": 8.342862245750037e-05} +{"train_loss": 0.01615828275680542, "grad_norm": 0.049960020929574966, "global_step": 14146, "epoch": 133, "lr": 8.342641840255454e-05} +{"train_loss": 0.01637696474790573, "grad_norm": 0.11353281885385513, "global_step": 14147, "epoch": 133, "lr": 8.34242142301628e-05} +{"train_loss": 0.016808897256851196, "grad_norm": 0.05910521373152733, "global_step": 14148, "epoch": 133, "lr": 8.342200994033291e-05} +{"train_loss": 0.018772320821881294, "grad_norm": 0.16598643362522125, "global_step": 14149, "epoch": 133, "lr": 8.341980553307261e-05} +{"train_loss": 0.015023444779217243, "grad_norm": 0.05748865380883217, "global_step": 14150, "epoch": 133, "lr": 8.341760100838965e-05} +{"train_loss": 0.016105055809020996, "grad_norm": 0.12573587894439697, "global_step": 14151, "epoch": 133, "lr": 8.341539636629178e-05} +{"train_loss": 0.017255522310733795, "grad_norm": 0.0654054805636406, "global_step": 14152, "epoch": 133, "lr": 8.341319160678673e-05} +{"train_loss": 0.01827164739370346, "grad_norm": 0.08518055826425552, "global_step": 14153, "epoch": 133, "lr": 8.341098672988225e-05} +{"train_loss": 0.014501400291919708, "grad_norm": 0.09336656332015991, "global_step": 14154, "epoch": 133, "lr": 8.340878173558613e-05} +{"train_loss": 0.018123731017112732, "grad_norm": 0.0845082625746727, "global_step": 14155, "epoch": 133, "lr": 8.340657662390605e-05} +{"train_loss": 0.014144946821033955, "grad_norm": 0.0647525042295456, "global_step": 14156, "epoch": 133, "lr": 8.340437139484977e-05} +{"train_loss": 0.016700271517038345, "grad_norm": 0.056035347282886505, "global_step": 14157, "epoch": 133, "lr": 8.340216604842509e-05} +{"train_loss": 0.016904868185520172, "grad_norm": 0.07261830568313599, "global_step": 14158, "epoch": 133, "lr": 8.33999605846397e-05} +{"train_loss": 0.017729856073856354, "grad_norm": 0.06266964972019196, "global_step": 14159, "epoch": 133, "lr": 8.339775500350138e-05} +{"train_loss": 0.01680426299571991, "grad_norm": 0.08818845450878143, "global_step": 14160, "epoch": 133, "lr": 8.339554930501789e-05} +{"train_loss": 0.016238903626799583, "grad_norm": 0.07020783424377441, "global_step": 14161, "epoch": 133, "lr": 8.339334348919696e-05} +{"train_loss": 0.018776141107082367, "grad_norm": 0.11621006578207016, "global_step": 14162, "epoch": 133, "lr": 8.339113755604633e-05} +{"train_loss": 0.016053419560194016, "grad_norm": 0.20129653811454773, "global_step": 14163, "epoch": 133, "lr": 8.338893150557376e-05} +{"train_loss": 0.015003941021859646, "grad_norm": 0.09831405431032181, "global_step": 14164, "epoch": 133, "lr": 8.338672533778702e-05} +{"train_loss": 0.01635066047310829, "grad_norm": 0.09879546612501144, "global_step": 14165, "epoch": 133, "lr": 8.338451905269384e-05} +{"train_loss": 0.017003022134304047, "grad_norm": 0.06420488655567169, "global_step": 14166, "epoch": 133, "lr": 8.338231265030199e-05} +{"train_loss": 0.015745004639029503, "grad_norm": 0.048440441489219666, "global_step": 14167, "epoch": 133, "lr": 8.338010613061919e-05} +{"train_loss": 0.01787637546658516, "grad_norm": 0.07946138828992844, "global_step": 14168, "epoch": 133, "lr": 8.337789949365322e-05} +{"train_loss": 0.017283406108617783, "grad_norm": 0.056108225136995316, "global_step": 14169, "epoch": 133, "lr": 8.337569273941183e-05} +{"train_loss": 0.015815023332834244, "grad_norm": 0.04968629032373428, "global_step": 14170, "epoch": 133, "lr": 8.337348586790277e-05} +{"train_loss": 0.017375288531184196, "grad_norm": 0.09167438745498657, "global_step": 14171, "epoch": 133, "lr": 8.337127887913379e-05} +{"train_loss": 0.019603148102760315, "grad_norm": 0.046862322837114334, "global_step": 14172, "epoch": 133, "lr": 8.336907177311266e-05} +{"train_loss": 0.014724744483828545, "grad_norm": 0.0574946403503418, "global_step": 14173, "epoch": 133, "lr": 8.336686454984712e-05} +{"train_loss": 0.017837096005678177, "grad_norm": 0.0779743567109108, "global_step": 14174, "epoch": 133, "lr": 8.336465720934493e-05} +{"train_loss": 0.017587505280971527, "grad_norm": 0.10491441935300827, "global_step": 14175, "epoch": 133, "lr": 8.336244975161383e-05} +{"train_loss": 0.016775082796812057, "grad_norm": 0.06072551757097244, "global_step": 14176, "epoch": 133, "lr": 8.33602421766616e-05} +{"train_loss": 0.018395669758319855, "grad_norm": 0.09942036867141724, "global_step": 14177, "epoch": 133, "lr": 8.3358034484496e-05} +{"train_loss": 0.017428314313292503, "grad_norm": 0.09276718646287918, "global_step": 14178, "epoch": 133, "lr": 8.335582667512476e-05} +{"train_loss": 0.016360510140657425, "grad_norm": 0.0502132922410965, "global_step": 14179, "epoch": 133, "lr": 8.335361874855565e-05} +{"train_loss": 0.01702187955379486, "grad_norm": 0.170484721660614, "global_step": 14180, "epoch": 133, "lr": 8.335141070479643e-05} +{"train_loss": 0.015451271086931229, "grad_norm": 0.05003535747528076, "global_step": 14181, "epoch": 133, "lr": 8.334920254385486e-05} +{"train_loss": 0.018269944936037064, "grad_norm": 0.11507487297058105, "global_step": 14182, "epoch": 133, "lr": 8.334699426573868e-05} +{"train_loss": 0.014389161951839924, "grad_norm": 0.08033803850412369, "global_step": 14183, "epoch": 133, "lr": 8.334478587045569e-05} +{"train_loss": 0.017704814672470093, "grad_norm": 0.06600409746170044, "global_step": 14184, "epoch": 133, "lr": 8.334257735801361e-05} +{"train_loss": 0.01807141676545143, "grad_norm": 0.06643439829349518, "global_step": 14185, "epoch": 133, "lr": 8.334036872842023e-05} +{"train_loss": 0.015850428491830826, "grad_norm": 0.09622030705213547, "global_step": 14186, "epoch": 133, "lr": 8.333815998168329e-05} +{"train_loss": 0.022489627823233604, "grad_norm": 0.0721227154135704, "global_step": 14187, "epoch": 133, "lr": 8.333595111781053e-05} +{"train_loss": 0.01586546003818512, "grad_norm": 0.0959319993853569, "global_step": 14188, "epoch": 133, "lr": 8.333374213680974e-05} +{"train_loss": 0.015362275764346123, "grad_norm": 0.08810006827116013, "global_step": 14189, "epoch": 133, "lr": 8.333153303868869e-05} +{"train_loss": 0.015872826799750328, "grad_norm": 0.048345956951379776, "global_step": 14190, "epoch": 133, "lr": 8.332932382345513e-05} +{"train_loss": 0.01601945236325264, "grad_norm": 0.11772987991571426, "global_step": 14191, "epoch": 133, "lr": 8.33271144911168e-05} +{"train_loss": 0.018271105363965034, "grad_norm": 0.05815138295292854, "global_step": 14192, "epoch": 133, "lr": 8.332490504168151e-05} +{"train_loss": 0.013797224499285221, "grad_norm": 0.07577688246965408, "global_step": 14193, "epoch": 133, "lr": 8.332269547515698e-05} +{"train_loss": 0.015135270543396473, "grad_norm": 0.05146002024412155, "global_step": 14194, "epoch": 133, "lr": 8.3320485791551e-05} +{"train_loss": 0.019276583567261696, "grad_norm": 0.0892711952328682, "global_step": 14195, "epoch": 133, "lr": 8.331827599087132e-05} +{"train_loss": 0.016271235421299934, "grad_norm": 0.05916724354028702, "global_step": 14196, "epoch": 133, "lr": 8.33160660731257e-05} +{"train_loss": 0.016639895737171173, "grad_norm": 0.0597168393433094, "global_step": 14197, "epoch": 133, "lr": 8.331385603832191e-05} +{"train_loss": 0.016868092119693756, "grad_norm": 0.06124040111899376, "global_step": 14198, "epoch": 133, "lr": 8.331164588646774e-05} +{"train_loss": 0.01688176579773426, "grad_norm": 0.0685696229338646, "global_step": 14199, "epoch": 133, "lr": 8.330943561757092e-05} +{"train_loss": 0.01761351153254509, "grad_norm": 0.0659908577799797, "global_step": 14200, "epoch": 133, "lr": 8.330722523163921e-05} +{"train_loss": 0.014501412399113178, "grad_norm": 0.10735644400119781, "global_step": 14201, "epoch": 133, "lr": 8.330501472868042e-05} +{"train_loss": 0.01634562388062477, "grad_norm": 0.10026012361049652, "global_step": 14202, "epoch": 133, "lr": 8.33028041087023e-05} +{"train_loss": 0.01687589493351725, "grad_norm": 0.10590256005525589, "global_step": 14203, "epoch": 133, "lr": 8.330059337171258e-05, "val_loss": 0.019050970673561096} +{"train_loss": 0.01633438654243946, "grad_norm": 0.0900074690580368, "global_step": 14204, "epoch": 134, "lr": 8.329838251771909e-05} +{"train_loss": 0.018164344131946564, "grad_norm": 0.09913341701030731, "global_step": 14205, "epoch": 134, "lr": 8.329617154672952e-05} +{"train_loss": 0.017909599468111992, "grad_norm": 0.07221506536006927, "global_step": 14206, "epoch": 134, "lr": 8.329396045875173e-05} +{"train_loss": 0.019402628764510155, "grad_norm": 0.0640554279088974, "global_step": 14207, "epoch": 134, "lr": 8.329174925379342e-05} +{"train_loss": 0.01841799169778824, "grad_norm": 0.0713917687535286, "global_step": 14208, "epoch": 134, "lr": 8.328953793186238e-05} +{"train_loss": 0.01570480316877365, "grad_norm": 0.053894780576229095, "global_step": 14209, "epoch": 134, "lr": 8.328732649296637e-05} +{"train_loss": 0.016138117760419846, "grad_norm": 0.08899490535259247, "global_step": 14210, "epoch": 134, "lr": 8.328511493711321e-05} +{"train_loss": 0.018560335040092468, "grad_norm": 0.07488131523132324, "global_step": 14211, "epoch": 134, "lr": 8.32829032643106e-05} +{"train_loss": 0.01746802218258381, "grad_norm": 0.07369624823331833, "global_step": 14212, "epoch": 134, "lr": 8.328069147456634e-05} +{"train_loss": 0.017834335565567017, "grad_norm": 0.10071615129709244, "global_step": 14213, "epoch": 134, "lr": 8.327847956788822e-05} +{"train_loss": 0.01593146100640297, "grad_norm": 0.0546179823577404, "global_step": 14214, "epoch": 134, "lr": 8.327626754428399e-05} +{"train_loss": 0.01852179691195488, "grad_norm": 0.08876974880695343, "global_step": 14215, "epoch": 134, "lr": 8.327405540376143e-05} +{"train_loss": 0.01719595491886139, "grad_norm": 0.10562384873628616, "global_step": 14216, "epoch": 134, "lr": 8.32718431463283e-05} +{"train_loss": 0.018335014581680298, "grad_norm": 0.09855582565069199, "global_step": 14217, "epoch": 134, "lr": 8.32696307719924e-05} +{"train_loss": 0.01641935482621193, "grad_norm": 0.11637863516807556, "global_step": 14218, "epoch": 134, "lr": 8.326741828076145e-05} +{"train_loss": 0.018152926117181778, "grad_norm": 0.06999263167381287, "global_step": 14219, "epoch": 134, "lr": 8.32652056726433e-05} +{"train_loss": 0.014813111163675785, "grad_norm": 0.10987081378698349, "global_step": 14220, "epoch": 134, "lr": 8.326299294764567e-05} +{"train_loss": 0.017010893672704697, "grad_norm": 0.07404092699289322, "global_step": 14221, "epoch": 134, "lr": 8.326078010577634e-05} +{"train_loss": 0.015923727303743362, "grad_norm": 0.1140870526432991, "global_step": 14222, "epoch": 134, "lr": 8.32585671470431e-05} +{"train_loss": 0.015398175455629826, "grad_norm": 0.041444987058639526, "global_step": 14223, "epoch": 134, "lr": 8.32563540714537e-05} +{"train_loss": 0.01872628554701805, "grad_norm": 0.07491565495729446, "global_step": 14224, "epoch": 134, "lr": 8.325414087901597e-05} +{"train_loss": 0.01610923558473587, "grad_norm": 0.04695899784564972, "global_step": 14225, "epoch": 134, "lr": 8.325192756973763e-05} +{"train_loss": 0.015952425077557564, "grad_norm": 0.0905962586402893, "global_step": 14226, "epoch": 134, "lr": 8.324971414362649e-05} +{"train_loss": 0.014616207219660282, "grad_norm": 0.06292315572500229, "global_step": 14227, "epoch": 134, "lr": 8.32475006006903e-05} +{"train_loss": 0.017217691987752914, "grad_norm": 0.09175661951303482, "global_step": 14228, "epoch": 134, "lr": 8.324528694093687e-05} +{"train_loss": 0.01748284511268139, "grad_norm": 0.09604815393686295, "global_step": 14229, "epoch": 134, "lr": 8.324307316437394e-05} +{"train_loss": 0.017758462578058243, "grad_norm": 0.0663478821516037, "global_step": 14230, "epoch": 134, "lr": 8.324085927100931e-05} +{"train_loss": 0.01695813424885273, "grad_norm": 0.08257628977298737, "global_step": 14231, "epoch": 134, "lr": 8.323864526085076e-05} +{"train_loss": 0.016917970031499863, "grad_norm": 0.0903344452381134, "global_step": 14232, "epoch": 134, "lr": 8.323643113390607e-05} +{"train_loss": 0.01772983744740486, "grad_norm": 0.069178506731987, "global_step": 14233, "epoch": 134, "lr": 8.323421689018302e-05} +{"train_loss": 0.017025668174028397, "grad_norm": 0.059473831206560135, "global_step": 14234, "epoch": 134, "lr": 8.323200252968937e-05} +{"train_loss": 0.015954872593283653, "grad_norm": 0.1162315383553505, "global_step": 14235, "epoch": 134, "lr": 8.322978805243293e-05} +{"train_loss": 0.01672673225402832, "grad_norm": 0.14704997837543488, "global_step": 14236, "epoch": 134, "lr": 8.322757345842146e-05} +{"train_loss": 0.01767289638519287, "grad_norm": 0.09198276698589325, "global_step": 14237, "epoch": 134, "lr": 8.322535874766275e-05} +{"train_loss": 0.01725584641098976, "grad_norm": 0.05256663262844086, "global_step": 14238, "epoch": 134, "lr": 8.322314392016458e-05} +{"train_loss": 0.013829350471496582, "grad_norm": 0.06629684567451477, "global_step": 14239, "epoch": 134, "lr": 8.322092897593472e-05} +{"train_loss": 0.018846698105335236, "grad_norm": 0.08218412101268768, "global_step": 14240, "epoch": 134, "lr": 8.321871391498097e-05} +{"train_loss": 0.017353147268295288, "grad_norm": 0.11948306113481522, "global_step": 14241, "epoch": 134, "lr": 8.32164987373111e-05} +{"train_loss": 0.017957068979740143, "grad_norm": 0.10767128318548203, "global_step": 14242, "epoch": 134, "lr": 8.32142834429329e-05} +{"train_loss": 0.01681097224354744, "grad_norm": 0.09447053074836731, "global_step": 14243, "epoch": 134, "lr": 8.321206803185415e-05} +{"train_loss": 0.014878003858029842, "grad_norm": 0.05980617552995682, "global_step": 14244, "epoch": 134, "lr": 8.320985250408264e-05} +{"train_loss": 0.015118122100830078, "grad_norm": 0.11386489868164062, "global_step": 14245, "epoch": 134, "lr": 8.320763685962616e-05} +{"train_loss": 0.016270404681563377, "grad_norm": 0.09755373001098633, "global_step": 14246, "epoch": 134, "lr": 8.320542109849247e-05} +{"train_loss": 0.016080711036920547, "grad_norm": 0.07861077040433884, "global_step": 14247, "epoch": 134, "lr": 8.320320522068938e-05} +{"train_loss": 0.01570970192551613, "grad_norm": 0.06893807649612427, "global_step": 14248, "epoch": 134, "lr": 8.320098922622467e-05} +{"train_loss": 0.016984283924102783, "grad_norm": 0.13693654537200928, "global_step": 14249, "epoch": 134, "lr": 8.319877311510613e-05} +{"train_loss": 0.013848016038537025, "grad_norm": 0.0846080407500267, "global_step": 14250, "epoch": 134, "lr": 8.319655688734152e-05} +{"train_loss": 0.01729414239525795, "grad_norm": 0.1079496368765831, "global_step": 14251, "epoch": 134, "lr": 8.319434054293866e-05} +{"train_loss": 0.015453414060175419, "grad_norm": 0.14190271496772766, "global_step": 14252, "epoch": 134, "lr": 8.319212408190531e-05} +{"train_loss": 0.017636844888329506, "grad_norm": 0.08164223283529282, "global_step": 14253, "epoch": 134, "lr": 8.318990750424927e-05} +{"train_loss": 0.016807444393634796, "grad_norm": 0.1553875058889389, "global_step": 14254, "epoch": 134, "lr": 8.318769080997834e-05} +{"train_loss": 0.016536656767129898, "grad_norm": 0.11970340460538864, "global_step": 14255, "epoch": 134, "lr": 8.31854739991003e-05} +{"train_loss": 0.01667030155658722, "grad_norm": 0.0691554918885231, "global_step": 14256, "epoch": 134, "lr": 8.318325707162293e-05} +{"train_loss": 0.015733987092971802, "grad_norm": 0.06965024024248123, "global_step": 14257, "epoch": 134, "lr": 8.318104002755403e-05} +{"train_loss": 0.016160784289240837, "grad_norm": 0.05916907638311386, "global_step": 14258, "epoch": 134, "lr": 8.317882286690138e-05} +{"train_loss": 0.018105976283550262, "grad_norm": 0.12493974715471268, "global_step": 14259, "epoch": 134, "lr": 8.317660558967277e-05} +{"train_loss": 0.017297476530075073, "grad_norm": 0.07626849412918091, "global_step": 14260, "epoch": 134, "lr": 8.317438819587603e-05} +{"train_loss": 0.017914051190018654, "grad_norm": 0.08249935507774353, "global_step": 14261, "epoch": 134, "lr": 8.317217068551887e-05} +{"train_loss": 0.015293143689632416, "grad_norm": 0.11606308817863464, "global_step": 14262, "epoch": 134, "lr": 8.316995305860915e-05} +{"train_loss": 0.014179649762809277, "grad_norm": 0.16392087936401367, "global_step": 14263, "epoch": 134, "lr": 8.316773531515464e-05} +{"train_loss": 0.016676895320415497, "grad_norm": 0.13382500410079956, "global_step": 14264, "epoch": 134, "lr": 8.316551745516312e-05} +{"train_loss": 0.015446613542735577, "grad_norm": 0.1045822724699974, "global_step": 14265, "epoch": 134, "lr": 8.316329947864241e-05} +{"train_loss": 0.018162094056606293, "grad_norm": 0.0809185728430748, "global_step": 14266, "epoch": 134, "lr": 8.31610813856003e-05} +{"train_loss": 0.016596045345067978, "grad_norm": 0.06033731997013092, "global_step": 14267, "epoch": 134, "lr": 8.315886317604455e-05} +{"train_loss": 0.017571352422237396, "grad_norm": 0.1343357115983963, "global_step": 14268, "epoch": 134, "lr": 8.315664484998297e-05} +{"train_loss": 0.014904385432600975, "grad_norm": 0.09284349530935287, "global_step": 14269, "epoch": 134, "lr": 8.315442640742337e-05} +{"train_loss": 0.018437374383211136, "grad_norm": 0.06387033313512802, "global_step": 14270, "epoch": 134, "lr": 8.315220784837353e-05} +{"train_loss": 0.016191966831684113, "grad_norm": 0.07458624243736267, "global_step": 14271, "epoch": 134, "lr": 8.314998917284125e-05} +{"train_loss": 0.019606418907642365, "grad_norm": 0.12336456030607224, "global_step": 14272, "epoch": 134, "lr": 8.314777038083434e-05} +{"train_loss": 0.018760263919830322, "grad_norm": 0.09098351746797562, "global_step": 14273, "epoch": 134, "lr": 8.314555147236057e-05} +{"train_loss": 0.01658829115331173, "grad_norm": 0.10694103688001633, "global_step": 14274, "epoch": 134, "lr": 8.314333244742774e-05} +{"train_loss": 0.016726849600672722, "grad_norm": 0.06506429612636566, "global_step": 14275, "epoch": 134, "lr": 8.314111330604366e-05} +{"train_loss": 0.01905585080385208, "grad_norm": 0.17953024804592133, "global_step": 14276, "epoch": 134, "lr": 8.31388940482161e-05} +{"train_loss": 0.018742794170975685, "grad_norm": 0.06916486471891403, "global_step": 14277, "epoch": 134, "lr": 8.313667467395292e-05} +{"train_loss": 0.018317947164177895, "grad_norm": 0.06323740631341934, "global_step": 14278, "epoch": 134, "lr": 8.313445518326184e-05} +{"train_loss": 0.018807249143719673, "grad_norm": 0.08626944571733475, "global_step": 14279, "epoch": 134, "lr": 8.313223557615071e-05} +{"train_loss": 0.017810991033911705, "grad_norm": 0.107444629073143, "global_step": 14280, "epoch": 134, "lr": 8.313001585262731e-05} +{"train_loss": 0.016368016600608826, "grad_norm": 0.09328246861696243, "global_step": 14281, "epoch": 134, "lr": 8.312779601269945e-05} +{"train_loss": 0.01543058268725872, "grad_norm": 0.08965042978525162, "global_step": 14282, "epoch": 134, "lr": 8.312557605637493e-05} +{"train_loss": 0.019111022353172302, "grad_norm": 0.07666349411010742, "global_step": 14283, "epoch": 134, "lr": 8.31233559836615e-05} +{"train_loss": 0.01653837040066719, "grad_norm": 0.11227767169475555, "global_step": 14284, "epoch": 134, "lr": 8.312113579456704e-05} +{"train_loss": 0.015611651353538036, "grad_norm": 0.07376446574926376, "global_step": 14285, "epoch": 134, "lr": 8.31189154890993e-05} +{"train_loss": 0.017217490822076797, "grad_norm": 0.13001057505607605, "global_step": 14286, "epoch": 134, "lr": 8.31166950672661e-05} +{"train_loss": 0.017405111342668533, "grad_norm": 0.08987468481063843, "global_step": 14287, "epoch": 134, "lr": 8.311447452907522e-05} +{"train_loss": 0.017340950667858124, "grad_norm": 0.127666175365448, "global_step": 14288, "epoch": 134, "lr": 8.31122538745345e-05} +{"train_loss": 0.01726257987320423, "grad_norm": 0.044050183147192, "global_step": 14289, "epoch": 134, "lr": 8.311003310365169e-05} +{"train_loss": 0.020984135568141937, "grad_norm": 0.19351933896541595, "global_step": 14290, "epoch": 134, "lr": 8.310781221643464e-05} +{"train_loss": 0.01733051985502243, "grad_norm": 0.09102944284677505, "global_step": 14291, "epoch": 134, "lr": 8.310559121289114e-05} +{"train_loss": 0.0147979985922575, "grad_norm": 0.08611924201250076, "global_step": 14292, "epoch": 134, "lr": 8.3103370093029e-05} +{"train_loss": 0.017122477293014526, "grad_norm": 0.12456850707530975, "global_step": 14293, "epoch": 134, "lr": 8.310114885685599e-05} +{"train_loss": 0.017301175743341446, "grad_norm": 0.06375549733638763, "global_step": 14294, "epoch": 134, "lr": 8.309892750437995e-05} +{"train_loss": 0.017412148416042328, "grad_norm": 0.08597200363874435, "global_step": 14295, "epoch": 134, "lr": 8.309670603560869e-05} +{"train_loss": 0.015653889626264572, "grad_norm": 0.0679163783788681, "global_step": 14296, "epoch": 134, "lr": 8.309448445054998e-05} +{"train_loss": 0.017015574499964714, "grad_norm": 0.07406803220510483, "global_step": 14297, "epoch": 134, "lr": 8.309226274921163e-05} +{"train_loss": 0.015032131224870682, "grad_norm": 0.0682341456413269, "global_step": 14298, "epoch": 134, "lr": 8.309004093160147e-05} +{"train_loss": 0.016196269541978836, "grad_norm": 0.06269576400518417, "global_step": 14299, "epoch": 134, "lr": 8.30878189977273e-05} +{"train_loss": 0.01658637821674347, "grad_norm": 0.04745478183031082, "global_step": 14300, "epoch": 134, "lr": 8.308559694759693e-05} +{"train_loss": 0.015438162721693516, "grad_norm": 0.05458087474107742, "global_step": 14301, "epoch": 134, "lr": 8.308337478121815e-05} +{"train_loss": 0.01717551052570343, "grad_norm": 0.14849112927913666, "global_step": 14302, "epoch": 134, "lr": 8.308115249859878e-05} +{"train_loss": 0.01679183728992939, "grad_norm": 0.05370136722922325, "global_step": 14303, "epoch": 134, "lr": 8.307893009974663e-05} +{"train_loss": 0.014810904860496521, "grad_norm": 0.06574271619319916, "global_step": 14304, "epoch": 134, "lr": 8.307670758466952e-05} +{"train_loss": 0.014886368997395039, "grad_norm": 0.11142943799495697, "global_step": 14305, "epoch": 134, "lr": 8.307448495337523e-05} +{"train_loss": 0.018581373617053032, "grad_norm": 0.11814131587743759, "global_step": 14306, "epoch": 134, "lr": 8.307226220587158e-05} +{"train_loss": 0.01693197526037693, "grad_norm": 0.04912150278687477, "global_step": 14307, "epoch": 134, "lr": 8.30700393421664e-05} +{"train_loss": 0.01655917428433895, "grad_norm": 0.05690409243106842, "global_step": 14308, "epoch": 134, "lr": 8.306781636226746e-05} +{"train_loss": 0.016877929618546984, "grad_norm": 0.09247533977031708, "global_step": 14309, "epoch": 134, "lr": 8.306559326618259e-05, "val_loss": 0.019525356590747833} +{"train_loss": 0.01711966097354889, "grad_norm": 0.048083651810884476, "global_step": 14310, "epoch": 135, "lr": 8.306337005391964e-05} +{"train_loss": 0.017966613173484802, "grad_norm": 0.12158714979887009, "global_step": 14311, "epoch": 135, "lr": 8.306114672548634e-05} +{"train_loss": 0.015906093642115593, "grad_norm": 0.06518333405256271, "global_step": 14312, "epoch": 135, "lr": 8.305892328089059e-05} +{"train_loss": 0.015024278312921524, "grad_norm": 0.06833431869745255, "global_step": 14313, "epoch": 135, "lr": 8.305669972014015e-05} +{"train_loss": 0.01850605383515358, "grad_norm": 0.07676558941602707, "global_step": 14314, "epoch": 135, "lr": 8.305447604324284e-05} +{"train_loss": 0.016446884721517563, "grad_norm": 0.07703959941864014, "global_step": 14315, "epoch": 135, "lr": 8.305225225020646e-05} +{"train_loss": 0.014391790144145489, "grad_norm": 0.04893091693520546, "global_step": 14316, "epoch": 135, "lr": 8.305002834103887e-05} +{"train_loss": 0.01664132997393608, "grad_norm": 0.049867767840623856, "global_step": 14317, "epoch": 135, "lr": 8.304780431574784e-05} +{"train_loss": 0.01821928657591343, "grad_norm": 0.09302306175231934, "global_step": 14318, "epoch": 135, "lr": 8.304558017434119e-05} +{"train_loss": 0.017172839492559433, "grad_norm": 0.0649557113647461, "global_step": 14319, "epoch": 135, "lr": 8.304335591682674e-05} +{"train_loss": 0.01857835054397583, "grad_norm": 0.09352429211139679, "global_step": 14320, "epoch": 135, "lr": 8.304113154321231e-05} +{"train_loss": 0.015205970034003258, "grad_norm": 0.06988664716482162, "global_step": 14321, "epoch": 135, "lr": 8.303890705350572e-05} +{"train_loss": 0.017946751788258553, "grad_norm": 0.06320077925920486, "global_step": 14322, "epoch": 135, "lr": 8.303668244771479e-05} +{"train_loss": 0.016970768570899963, "grad_norm": 0.14587759971618652, "global_step": 14323, "epoch": 135, "lr": 8.30344577258473e-05} +{"train_loss": 0.016410691663622856, "grad_norm": 0.12406518310308456, "global_step": 14324, "epoch": 135, "lr": 8.30322328879111e-05} +{"train_loss": 0.01572331413626671, "grad_norm": 0.06972424685955048, "global_step": 14325, "epoch": 135, "lr": 8.303000793391402e-05} +{"train_loss": 0.01763106882572174, "grad_norm": 0.06970274448394775, "global_step": 14326, "epoch": 135, "lr": 8.302778286386384e-05} +{"train_loss": 0.0180966779589653, "grad_norm": 0.07661351561546326, "global_step": 14327, "epoch": 135, "lr": 8.302555767776838e-05} +{"train_loss": 0.01944216527044773, "grad_norm": 0.08142855763435364, "global_step": 14328, "epoch": 135, "lr": 8.302333237563549e-05} +{"train_loss": 0.01748707890510559, "grad_norm": 0.06782136112451553, "global_step": 14329, "epoch": 135, "lr": 8.302110695747297e-05} +{"train_loss": 0.01811216026544571, "grad_norm": 0.06256883591413498, "global_step": 14330, "epoch": 135, "lr": 8.301888142328864e-05} +{"train_loss": 0.016762226819992065, "grad_norm": 0.11378920823335648, "global_step": 14331, "epoch": 135, "lr": 8.301665577309032e-05} +{"train_loss": 0.018914900720119476, "grad_norm": 0.05873016640543938, "global_step": 14332, "epoch": 135, "lr": 8.301443000688583e-05} +{"train_loss": 0.01566316746175289, "grad_norm": 0.06533269584178925, "global_step": 14333, "epoch": 135, "lr": 8.301220412468299e-05} +{"train_loss": 0.016276512295007706, "grad_norm": 0.09356460720300674, "global_step": 14334, "epoch": 135, "lr": 8.300997812648961e-05} +{"train_loss": 0.016786346212029457, "grad_norm": 0.0494607649743557, "global_step": 14335, "epoch": 135, "lr": 8.300775201231355e-05} +{"train_loss": 0.01625911146402359, "grad_norm": 0.09650938957929611, "global_step": 14336, "epoch": 135, "lr": 8.300552578216258e-05} +{"train_loss": 0.0181427039206028, "grad_norm": 0.06722372770309448, "global_step": 14337, "epoch": 135, "lr": 8.300329943604455e-05} +{"train_loss": 0.01433749683201313, "grad_norm": 0.10569373518228531, "global_step": 14338, "epoch": 135, "lr": 8.300107297396728e-05} +{"train_loss": 0.01675514131784439, "grad_norm": 0.09725837409496307, "global_step": 14339, "epoch": 135, "lr": 8.29988463959386e-05} +{"train_loss": 0.01651897467672825, "grad_norm": 0.057110629975795746, "global_step": 14340, "epoch": 135, "lr": 8.299661970196633e-05} +{"train_loss": 0.01632944867014885, "grad_norm": 0.08972974121570587, "global_step": 14341, "epoch": 135, "lr": 8.299439289205827e-05} +{"train_loss": 0.01694643497467041, "grad_norm": 0.09774385392665863, "global_step": 14342, "epoch": 135, "lr": 8.299216596622229e-05} +{"train_loss": 0.01469849981367588, "grad_norm": 0.0740227997303009, "global_step": 14343, "epoch": 135, "lr": 8.298993892446613e-05} +{"train_loss": 0.017067616805434227, "grad_norm": 0.11014128476381302, "global_step": 14344, "epoch": 135, "lr": 8.298771176679772e-05} +{"train_loss": 0.01703052967786789, "grad_norm": 0.07695993036031723, "global_step": 14345, "epoch": 135, "lr": 8.298548449322483e-05} +{"train_loss": 0.015628106892108917, "grad_norm": 0.07799883931875229, "global_step": 14346, "epoch": 135, "lr": 8.298325710375528e-05} +{"train_loss": 0.0214737206697464, "grad_norm": 0.1195223480463028, "global_step": 14347, "epoch": 135, "lr": 8.29810295983969e-05} +{"train_loss": 0.01824530027806759, "grad_norm": 0.08465801924467087, "global_step": 14348, "epoch": 135, "lr": 8.297880197715757e-05} +{"train_loss": 0.01728612370789051, "grad_norm": 0.13224482536315918, "global_step": 14349, "epoch": 135, "lr": 8.297657424004504e-05} +{"train_loss": 0.014907442964613438, "grad_norm": 0.06899388134479523, "global_step": 14350, "epoch": 135, "lr": 8.297434638706716e-05} +{"train_loss": 0.01929435506463051, "grad_norm": 0.06876226514577866, "global_step": 14351, "epoch": 135, "lr": 8.297211841823178e-05} +{"train_loss": 0.018404928967356682, "grad_norm": 0.0740300714969635, "global_step": 14352, "epoch": 135, "lr": 8.296989033354671e-05} +{"train_loss": 0.0154543686658144, "grad_norm": 0.07626023143529892, "global_step": 14353, "epoch": 135, "lr": 8.296766213301978e-05} +{"train_loss": 0.018717870116233826, "grad_norm": 0.08500055223703384, "global_step": 14354, "epoch": 135, "lr": 8.296543381665884e-05} +{"train_loss": 0.016954127699136734, "grad_norm": 0.1370420753955841, "global_step": 14355, "epoch": 135, "lr": 8.296320538447169e-05} +{"train_loss": 0.016214575618505478, "grad_norm": 0.05347191169857979, "global_step": 14356, "epoch": 135, "lr": 8.296097683646617e-05} +{"train_loss": 0.014358889311552048, "grad_norm": 0.07696686685085297, "global_step": 14357, "epoch": 135, "lr": 8.295874817265012e-05} +{"train_loss": 0.016781451180577278, "grad_norm": 0.1993391066789627, "global_step": 14358, "epoch": 135, "lr": 8.295651939303135e-05} +{"train_loss": 0.01708405464887619, "grad_norm": 0.06172271445393562, "global_step": 14359, "epoch": 135, "lr": 8.295429049761772e-05} +{"train_loss": 0.02048874832689762, "grad_norm": 0.0530681386590004, "global_step": 14360, "epoch": 135, "lr": 8.295206148641703e-05} +{"train_loss": 0.017352566123008728, "grad_norm": 0.06861110776662827, "global_step": 14361, "epoch": 135, "lr": 8.294983235943713e-05} +{"train_loss": 0.018988700583577156, "grad_norm": 0.09936254471540451, "global_step": 14362, "epoch": 135, "lr": 8.294760311668586e-05} +{"train_loss": 0.017286837100982666, "grad_norm": 0.06868455559015274, "global_step": 14363, "epoch": 135, "lr": 8.294537375817102e-05} +{"train_loss": 0.01633041352033615, "grad_norm": 0.0842704102396965, "global_step": 14364, "epoch": 135, "lr": 8.294314428390048e-05} +{"train_loss": 0.017696117982268333, "grad_norm": 0.11025384813547134, "global_step": 14365, "epoch": 135, "lr": 8.294091469388205e-05} +{"train_loss": 0.018132459372282028, "grad_norm": 0.0563884861767292, "global_step": 14366, "epoch": 135, "lr": 8.293868498812358e-05} +{"train_loss": 0.01816006749868393, "grad_norm": 0.11266959458589554, "global_step": 14367, "epoch": 135, "lr": 8.29364551666329e-05} +{"train_loss": 0.014357836917042732, "grad_norm": 0.08626891672611237, "global_step": 14368, "epoch": 135, "lr": 8.293422522941781e-05} +{"train_loss": 0.017464179545640945, "grad_norm": 0.07859033346176147, "global_step": 14369, "epoch": 135, "lr": 8.29319951764862e-05} +{"train_loss": 0.014408254995942116, "grad_norm": 0.07214507460594177, "global_step": 14370, "epoch": 135, "lr": 8.292976500784588e-05} +{"train_loss": 0.01734810322523117, "grad_norm": 0.06117184832692146, "global_step": 14371, "epoch": 135, "lr": 8.292753472350468e-05} +{"train_loss": 0.01514516118913889, "grad_norm": 0.046246569603681564, "global_step": 14372, "epoch": 135, "lr": 8.292530432347046e-05} +{"train_loss": 0.018144138157367706, "grad_norm": 0.07817842811346054, "global_step": 14373, "epoch": 135, "lr": 8.2923073807751e-05} +{"train_loss": 0.01756843365728855, "grad_norm": 0.048653312027454376, "global_step": 14374, "epoch": 135, "lr": 8.29208431763542e-05} +{"train_loss": 0.016097847372293472, "grad_norm": 0.06957507133483887, "global_step": 14375, "epoch": 135, "lr": 8.291861242928789e-05} +{"train_loss": 0.017961235716938972, "grad_norm": 0.061643511056900024, "global_step": 14376, "epoch": 135, "lr": 8.291638156655987e-05} +{"train_loss": 0.014535008929669857, "grad_norm": 0.0820329338312149, "global_step": 14377, "epoch": 135, "lr": 8.291415058817801e-05} +{"train_loss": 0.01973641663789749, "grad_norm": 0.08384166657924652, "global_step": 14378, "epoch": 135, "lr": 8.291191949415014e-05} +{"train_loss": 0.016771260648965836, "grad_norm": 0.08406148850917816, "global_step": 14379, "epoch": 135, "lr": 8.290968828448408e-05} +{"train_loss": 0.01985366828739643, "grad_norm": 0.09274746477603912, "global_step": 14380, "epoch": 135, "lr": 8.290745695918772e-05} +{"train_loss": 0.015015894547104836, "grad_norm": 0.0682939887046814, "global_step": 14381, "epoch": 135, "lr": 8.290522551826885e-05} +{"train_loss": 0.015005572699010372, "grad_norm": 0.052772242575883865, "global_step": 14382, "epoch": 135, "lr": 8.29029939617353e-05} +{"train_loss": 0.019252261146903038, "grad_norm": 0.07290846109390259, "global_step": 14383, "epoch": 135, "lr": 8.290076228959498e-05} +{"train_loss": 0.015960849821567535, "grad_norm": 0.08220139145851135, "global_step": 14384, "epoch": 135, "lr": 8.289853050185568e-05} +{"train_loss": 0.01676875725388527, "grad_norm": 0.07455441355705261, "global_step": 14385, "epoch": 135, "lr": 8.289629859852523e-05} +{"train_loss": 0.013891562819480896, "grad_norm": 0.08286651223897934, "global_step": 14386, "epoch": 135, "lr": 8.289406657961152e-05} +{"train_loss": 0.017666228115558624, "grad_norm": 0.07506996393203735, "global_step": 14387, "epoch": 135, "lr": 8.289183444512236e-05} +{"train_loss": 0.0162825807929039, "grad_norm": 0.10799049586057663, "global_step": 14388, "epoch": 135, "lr": 8.288960219506558e-05} +{"train_loss": 0.0169542096555233, "grad_norm": 0.07249660044908524, "global_step": 14389, "epoch": 135, "lr": 8.288736982944907e-05} +{"train_loss": 0.01426362618803978, "grad_norm": 0.12016070634126663, "global_step": 14390, "epoch": 135, "lr": 8.288513734828062e-05} +{"train_loss": 0.015145713463425636, "grad_norm": 0.06656280905008316, "global_step": 14391, "epoch": 135, "lr": 8.288290475156811e-05} +{"train_loss": 0.014764862135052681, "grad_norm": 0.06491696834564209, "global_step": 14392, "epoch": 135, "lr": 8.288067203931937e-05} +{"train_loss": 0.014613761566579342, "grad_norm": 0.04550061747431755, "global_step": 14393, "epoch": 135, "lr": 8.287843921154223e-05} +{"train_loss": 0.016908083111047745, "grad_norm": 0.09797514975070953, "global_step": 14394, "epoch": 135, "lr": 8.287620626824457e-05} +{"train_loss": 0.01881798356771469, "grad_norm": 0.07325848937034607, "global_step": 14395, "epoch": 135, "lr": 8.287397320943423e-05} +{"train_loss": 0.01833864487707615, "grad_norm": 0.05436833202838898, "global_step": 14396, "epoch": 135, "lr": 8.287174003511903e-05} +{"train_loss": 0.014624515548348427, "grad_norm": 0.09879569709300995, "global_step": 14397, "epoch": 135, "lr": 8.286950674530683e-05} +{"train_loss": 0.014456566423177719, "grad_norm": 0.05110497772693634, "global_step": 14398, "epoch": 135, "lr": 8.286727334000547e-05} +{"train_loss": 0.016480356454849243, "grad_norm": 0.054200511425733566, "global_step": 14399, "epoch": 135, "lr": 8.286503981922283e-05} +{"train_loss": 0.01570316031575203, "grad_norm": 0.0786813348531723, "global_step": 14400, "epoch": 135, "lr": 8.286280618296672e-05} +{"train_loss": 0.019796697422862053, "grad_norm": 0.04817509278655052, "global_step": 14401, "epoch": 135, "lr": 8.286057243124501e-05} +{"train_loss": 0.01807459071278572, "grad_norm": 0.15956802666187286, "global_step": 14402, "epoch": 135, "lr": 8.285833856406553e-05} +{"train_loss": 0.016068771481513977, "grad_norm": 0.09005032479763031, "global_step": 14403, "epoch": 135, "lr": 8.285610458143613e-05} +{"train_loss": 0.0168765839189291, "grad_norm": 0.10408560186624527, "global_step": 14404, "epoch": 135, "lr": 8.285387048336469e-05} +{"train_loss": 0.01651349477469921, "grad_norm": 0.06321084499359131, "global_step": 14405, "epoch": 135, "lr": 8.285163626985901e-05} +{"train_loss": 0.015490452758967876, "grad_norm": 0.06542796641588211, "global_step": 14406, "epoch": 135, "lr": 8.284940194092699e-05} +{"train_loss": 0.018527857959270477, "grad_norm": 0.054291315376758575, "global_step": 14407, "epoch": 135, "lr": 8.284716749657644e-05} +{"train_loss": 0.01594860479235649, "grad_norm": 0.09985779970884323, "global_step": 14408, "epoch": 135, "lr": 8.284493293681524e-05} +{"train_loss": 0.015112364664673805, "grad_norm": 0.056711357086896896, "global_step": 14409, "epoch": 135, "lr": 8.284269826165122e-05} +{"train_loss": 0.01727260835468769, "grad_norm": 0.10391315817832947, "global_step": 14410, "epoch": 135, "lr": 8.284046347109225e-05} +{"train_loss": 0.016934920102357864, "grad_norm": 0.057401906698942184, "global_step": 14411, "epoch": 135, "lr": 8.283822856514616e-05} +{"train_loss": 0.014283246360719204, "grad_norm": 0.15019568800926208, "global_step": 14412, "epoch": 135, "lr": 8.283599354382083e-05} +{"train_loss": 0.017280254513025284, "grad_norm": 0.0856269896030426, "global_step": 14413, "epoch": 135, "lr": 8.283375840712409e-05} +{"train_loss": 0.016706909984350204, "grad_norm": 0.07198848575353622, "global_step": 14414, "epoch": 135, "lr": 8.28315231550638e-05} +{"train_loss": 0.016819628649654816, "grad_norm": 0.08650898933410645, "global_step": 14415, "epoch": 135, "lr": 8.282928778764783e-05, "val_loss": 0.01915396936237812} +{"train_loss": 0.017000580206513405, "grad_norm": 0.08631140738725662, "global_step": 14416, "epoch": 136, "lr": 8.282705230488401e-05} +{"train_loss": 0.014296289533376694, "grad_norm": 0.05530057102441788, "global_step": 14417, "epoch": 136, "lr": 8.282481670678021e-05} +{"train_loss": 0.017270883545279503, "grad_norm": 0.06909917294979095, "global_step": 14418, "epoch": 136, "lr": 8.282258099334426e-05} +{"train_loss": 0.015526136383414268, "grad_norm": 0.06995489448308945, "global_step": 14419, "epoch": 136, "lr": 8.282034516458406e-05} +{"train_loss": 0.01778470166027546, "grad_norm": 0.059194620698690414, "global_step": 14420, "epoch": 136, "lr": 8.281810922050743e-05} +{"train_loss": 0.017267195507884026, "grad_norm": 0.053175188601017, "global_step": 14421, "epoch": 136, "lr": 8.281587316112222e-05} +{"train_loss": 0.01813669316470623, "grad_norm": 0.09164059162139893, "global_step": 14422, "epoch": 136, "lr": 8.281363698643634e-05} +{"train_loss": 0.021585747599601746, "grad_norm": 0.12765759229660034, "global_step": 14423, "epoch": 136, "lr": 8.281140069645759e-05} +{"train_loss": 0.014716489240527153, "grad_norm": 0.06395565718412399, "global_step": 14424, "epoch": 136, "lr": 8.280916429119384e-05} +{"train_loss": 0.018874146044254303, "grad_norm": 0.11472167074680328, "global_step": 14425, "epoch": 136, "lr": 8.280692777065295e-05} +{"train_loss": 0.014701971784234047, "grad_norm": 0.07334534823894501, "global_step": 14426, "epoch": 136, "lr": 8.28046911348428e-05} +{"train_loss": 0.01631646789610386, "grad_norm": 0.07139702141284943, "global_step": 14427, "epoch": 136, "lr": 8.280245438377122e-05} +{"train_loss": 0.01820487529039383, "grad_norm": 0.08882514387369156, "global_step": 14428, "epoch": 136, "lr": 8.280021751744609e-05} +{"train_loss": 0.01418742723762989, "grad_norm": 0.04725624993443489, "global_step": 14429, "epoch": 136, "lr": 8.279798053587526e-05} +{"train_loss": 0.015278532169759274, "grad_norm": 0.05755201354622841, "global_step": 14430, "epoch": 136, "lr": 8.279574343906657e-05} +{"train_loss": 0.0177355594933033, "grad_norm": 0.06691861152648926, "global_step": 14431, "epoch": 136, "lr": 8.279350622702793e-05} +{"train_loss": 0.018409909680485725, "grad_norm": 0.07788346707820892, "global_step": 14432, "epoch": 136, "lr": 8.279126889976716e-05} +{"train_loss": 0.016320394352078438, "grad_norm": 0.11294888705015182, "global_step": 14433, "epoch": 136, "lr": 8.278903145729213e-05} +{"train_loss": 0.016210932284593582, "grad_norm": 0.13161686062812805, "global_step": 14434, "epoch": 136, "lr": 8.27867938996107e-05} +{"train_loss": 0.017732741311192513, "grad_norm": 0.0912691205739975, "global_step": 14435, "epoch": 136, "lr": 8.278455622673073e-05} +{"train_loss": 0.015371150337159634, "grad_norm": 0.05611199140548706, "global_step": 14436, "epoch": 136, "lr": 8.278231843866009e-05} +{"train_loss": 0.016325421631336212, "grad_norm": 0.12353847920894623, "global_step": 14437, "epoch": 136, "lr": 8.278008053540665e-05} +{"train_loss": 0.016569117084145546, "grad_norm": 0.061229486018419266, "global_step": 14438, "epoch": 136, "lr": 8.277784251697827e-05} +{"train_loss": 0.018581748008728027, "grad_norm": 0.07079536467790604, "global_step": 14439, "epoch": 136, "lr": 8.277560438338278e-05} +{"train_loss": 0.020371809601783752, "grad_norm": 0.11531306803226471, "global_step": 14440, "epoch": 136, "lr": 8.277336613462808e-05} +{"train_loss": 0.017613526433706284, "grad_norm": 0.08836162835359573, "global_step": 14441, "epoch": 136, "lr": 8.277112777072202e-05} +{"train_loss": 0.01826673001050949, "grad_norm": 0.06959351897239685, "global_step": 14442, "epoch": 136, "lr": 8.276888929167248e-05} +{"train_loss": 0.01564660295844078, "grad_norm": 0.06005090847611427, "global_step": 14443, "epoch": 136, "lr": 8.276665069748732e-05} +{"train_loss": 0.0156448632478714, "grad_norm": 0.050021182745695114, "global_step": 14444, "epoch": 136, "lr": 8.276441198817438e-05} +{"train_loss": 0.015318796038627625, "grad_norm": 0.056439027190208435, "global_step": 14445, "epoch": 136, "lr": 8.276217316374156e-05} +{"train_loss": 0.017715033143758774, "grad_norm": 0.11489231139421463, "global_step": 14446, "epoch": 136, "lr": 8.27599342241967e-05} +{"train_loss": 0.01926678977906704, "grad_norm": 0.20627142488956451, "global_step": 14447, "epoch": 136, "lr": 8.275769516954768e-05} +{"train_loss": 0.015405569225549698, "grad_norm": 0.054701898247003555, "global_step": 14448, "epoch": 136, "lr": 8.275545599980236e-05} +{"train_loss": 0.015145426616072655, "grad_norm": 0.05671844631433487, "global_step": 14449, "epoch": 136, "lr": 8.275321671496863e-05} +{"train_loss": 0.015879254788160324, "grad_norm": 0.08149556070566177, "global_step": 14450, "epoch": 136, "lr": 8.275097731505433e-05} +{"train_loss": 0.016002830117940903, "grad_norm": 0.08245977759361267, "global_step": 14451, "epoch": 136, "lr": 8.274873780006734e-05} +{"train_loss": 0.01720307394862175, "grad_norm": 0.11137815564870834, "global_step": 14452, "epoch": 136, "lr": 8.274649817001553e-05} +{"train_loss": 0.016964828595519066, "grad_norm": 0.08940818160772324, "global_step": 14453, "epoch": 136, "lr": 8.274425842490676e-05} +{"train_loss": 0.016324296593666077, "grad_norm": 0.06322444975376129, "global_step": 14454, "epoch": 136, "lr": 8.274201856474892e-05} +{"train_loss": 0.01793377473950386, "grad_norm": 0.059356819838285446, "global_step": 14455, "epoch": 136, "lr": 8.273977858954984e-05} +{"train_loss": 0.014546271413564682, "grad_norm": 0.07317329198122025, "global_step": 14456, "epoch": 136, "lr": 8.273753849931743e-05} +{"train_loss": 0.013682099990546703, "grad_norm": 0.06246340274810791, "global_step": 14457, "epoch": 136, "lr": 8.273529829405956e-05} +{"train_loss": 0.01862126961350441, "grad_norm": 0.08500361442565918, "global_step": 14458, "epoch": 136, "lr": 8.273305797378408e-05} +{"train_loss": 0.017999187111854553, "grad_norm": 0.08152972161769867, "global_step": 14459, "epoch": 136, "lr": 8.273081753849886e-05} +{"train_loss": 0.015771791338920593, "grad_norm": 0.09754399210214615, "global_step": 14460, "epoch": 136, "lr": 8.272857698821179e-05} +{"train_loss": 0.018464425578713417, "grad_norm": 0.2161470651626587, "global_step": 14461, "epoch": 136, "lr": 8.272633632293072e-05} +{"train_loss": 0.01631731539964676, "grad_norm": 0.0553104430437088, "global_step": 14462, "epoch": 136, "lr": 8.272409554266356e-05} +{"train_loss": 0.014919666573405266, "grad_norm": 0.09916340559720993, "global_step": 14463, "epoch": 136, "lr": 8.272185464741815e-05} +{"train_loss": 0.016544844955205917, "grad_norm": 0.10284553468227386, "global_step": 14464, "epoch": 136, "lr": 8.271961363720235e-05} +{"train_loss": 0.01528865098953247, "grad_norm": 0.0446639284491539, "global_step": 14465, "epoch": 136, "lr": 8.271737251202408e-05} +{"train_loss": 0.014598020352423191, "grad_norm": 0.057867344468832016, "global_step": 14466, "epoch": 136, "lr": 8.27151312718912e-05} +{"train_loss": 0.016677457839250565, "grad_norm": 0.10378257930278778, "global_step": 14467, "epoch": 136, "lr": 8.271288991681156e-05} +{"train_loss": 0.016916627064347267, "grad_norm": 0.10263583809137344, "global_step": 14468, "epoch": 136, "lr": 8.271064844679306e-05} +{"train_loss": 0.01797344535589218, "grad_norm": 0.09767709672451019, "global_step": 14469, "epoch": 136, "lr": 8.270840686184355e-05} +{"train_loss": 0.01728767529129982, "grad_norm": 0.12959346175193787, "global_step": 14470, "epoch": 136, "lr": 8.270616516197094e-05} +{"train_loss": 0.019191350787878036, "grad_norm": 0.07078433781862259, "global_step": 14471, "epoch": 136, "lr": 8.27039233471831e-05} +{"train_loss": 0.01580416038632393, "grad_norm": 0.09148041903972626, "global_step": 14472, "epoch": 136, "lr": 8.270168141748787e-05} +{"train_loss": 0.016473092138767242, "grad_norm": 0.12254686653614044, "global_step": 14473, "epoch": 136, "lr": 8.269943937289317e-05} +{"train_loss": 0.0183403380215168, "grad_norm": 0.06499911099672318, "global_step": 14474, "epoch": 136, "lr": 8.269719721340686e-05} +{"train_loss": 0.017267927527427673, "grad_norm": 0.049365803599357605, "global_step": 14475, "epoch": 136, "lr": 8.26949549390368e-05} +{"train_loss": 0.014775248244404793, "grad_norm": 0.07443203032016754, "global_step": 14476, "epoch": 136, "lr": 8.269271254979091e-05} +{"train_loss": 0.018130145967006683, "grad_norm": 0.06782716512680054, "global_step": 14477, "epoch": 136, "lr": 8.269047004567703e-05} +{"train_loss": 0.017522703856229782, "grad_norm": 0.128097265958786, "global_step": 14478, "epoch": 136, "lr": 8.268822742670306e-05} +{"train_loss": 0.01842642016708851, "grad_norm": 0.05524854362010956, "global_step": 14479, "epoch": 136, "lr": 8.268598469287688e-05} +{"train_loss": 0.01525621023029089, "grad_norm": 0.0789121761918068, "global_step": 14480, "epoch": 136, "lr": 8.268374184420635e-05} +{"train_loss": 0.016399001702666283, "grad_norm": 0.05109060928225517, "global_step": 14481, "epoch": 136, "lr": 8.268149888069938e-05} +{"train_loss": 0.017569363117218018, "grad_norm": 0.07407847046852112, "global_step": 14482, "epoch": 136, "lr": 8.267925580236383e-05} +{"train_loss": 0.018619002774357796, "grad_norm": 0.11146927624940872, "global_step": 14483, "epoch": 136, "lr": 8.267701260920759e-05} +{"train_loss": 0.0174361951649189, "grad_norm": 0.0883563682436943, "global_step": 14484, "epoch": 136, "lr": 8.267476930123855e-05} +{"train_loss": 0.018254198133945465, "grad_norm": 0.1007503941655159, "global_step": 14485, "epoch": 136, "lr": 8.267252587846455e-05} +{"train_loss": 0.017007628455758095, "grad_norm": 0.07899780571460724, "global_step": 14486, "epoch": 136, "lr": 8.267028234089352e-05} +{"train_loss": 0.01716802828013897, "grad_norm": 0.09862282872200012, "global_step": 14487, "epoch": 136, "lr": 8.266803868853334e-05} +{"train_loss": 0.01557224988937378, "grad_norm": 0.07854767143726349, "global_step": 14488, "epoch": 136, "lr": 8.266579492139186e-05} +{"train_loss": 0.01667116768658161, "grad_norm": 0.05316159874200821, "global_step": 14489, "epoch": 136, "lr": 8.266355103947698e-05} +{"train_loss": 0.016487568616867065, "grad_norm": 0.09474822133779526, "global_step": 14490, "epoch": 136, "lr": 8.26613070427966e-05} +{"train_loss": 0.01612546667456627, "grad_norm": 0.04682632163167, "global_step": 14491, "epoch": 136, "lr": 8.265906293135859e-05} +{"train_loss": 0.016018135473132133, "grad_norm": 0.11109748482704163, "global_step": 14492, "epoch": 136, "lr": 8.265681870517084e-05} +{"train_loss": 0.01833243854343891, "grad_norm": 0.06917913258075714, "global_step": 14493, "epoch": 136, "lr": 8.26545743642412e-05} +{"train_loss": 0.017114292830228806, "grad_norm": 0.0603129006922245, "global_step": 14494, "epoch": 136, "lr": 8.265232990857761e-05} +{"train_loss": 0.016176246106624603, "grad_norm": 0.14805486798286438, "global_step": 14495, "epoch": 136, "lr": 8.265008533818795e-05} +{"train_loss": 0.016062185168266296, "grad_norm": 0.05161527171730995, "global_step": 14496, "epoch": 136, "lr": 8.264784065308006e-05} +{"train_loss": 0.01709623634815216, "grad_norm": 0.12138315290212631, "global_step": 14497, "epoch": 136, "lr": 8.264559585326186e-05} +{"train_loss": 0.0173240564763546, "grad_norm": 0.08744633942842484, "global_step": 14498, "epoch": 136, "lr": 8.264335093874124e-05} +{"train_loss": 0.015504714101552963, "grad_norm": 0.05859842151403427, "global_step": 14499, "epoch": 136, "lr": 8.264110590952609e-05} +{"train_loss": 0.01888221688568592, "grad_norm": 0.07339227199554443, "global_step": 14500, "epoch": 136, "lr": 8.263886076562427e-05} +{"train_loss": 0.016083138063549995, "grad_norm": 0.07659268379211426, "global_step": 14501, "epoch": 136, "lr": 8.263661550704372e-05} +{"train_loss": 0.016271015629172325, "grad_norm": 0.06355500221252441, "global_step": 14502, "epoch": 136, "lr": 8.263437013379226e-05} +{"train_loss": 0.016900334507226944, "grad_norm": 0.07062869518995285, "global_step": 14503, "epoch": 136, "lr": 8.263212464587783e-05} +{"train_loss": 0.014897387474775314, "grad_norm": 0.09738215059041977, "global_step": 14504, "epoch": 136, "lr": 8.26298790433083e-05} +{"train_loss": 0.014115260913968086, "grad_norm": 0.06265600025653839, "global_step": 14505, "epoch": 136, "lr": 8.262763332609158e-05} +{"train_loss": 0.015500922687351704, "grad_norm": 0.12050219625234604, "global_step": 14506, "epoch": 136, "lr": 8.262538749423554e-05} +{"train_loss": 0.019538432359695435, "grad_norm": 0.062171924859285355, "global_step": 14507, "epoch": 136, "lr": 8.262314154774806e-05} +{"train_loss": 0.017367666587233543, "grad_norm": 0.052752282470464706, "global_step": 14508, "epoch": 136, "lr": 8.262089548663708e-05} +{"train_loss": 0.015354898758232594, "grad_norm": 0.0661197379231453, "global_step": 14509, "epoch": 136, "lr": 8.261864931091043e-05} +{"train_loss": 0.017475148662924767, "grad_norm": 0.0694359540939331, "global_step": 14510, "epoch": 136, "lr": 8.261640302057606e-05} +{"train_loss": 0.014479943551123142, "grad_norm": 0.04589589685201645, "global_step": 14511, "epoch": 136, "lr": 8.261415661564181e-05} +{"train_loss": 0.019619908183813095, "grad_norm": 0.11754715442657471, "global_step": 14512, "epoch": 136, "lr": 8.26119100961156e-05} +{"train_loss": 0.01635369099676609, "grad_norm": 0.11258931457996368, "global_step": 14513, "epoch": 136, "lr": 8.260966346200531e-05} +{"train_loss": 0.014259057119488716, "grad_norm": 0.05417076498270035, "global_step": 14514, "epoch": 136, "lr": 8.260741671331886e-05} +{"train_loss": 0.015899550169706345, "grad_norm": 0.04996350780129433, "global_step": 14515, "epoch": 136, "lr": 8.260516985006413e-05} +{"train_loss": 0.017174119129776955, "grad_norm": 0.06402222812175751, "global_step": 14516, "epoch": 136, "lr": 8.260292287224902e-05} +{"train_loss": 0.01806562766432762, "grad_norm": 0.08397940546274185, "global_step": 14517, "epoch": 136, "lr": 8.26006757798814e-05} +{"train_loss": 0.015151980333030224, "grad_norm": 0.06151464954018593, "global_step": 14518, "epoch": 136, "lr": 8.25984285729692e-05} +{"train_loss": 0.020792923867702484, "grad_norm": 0.07367832958698273, "global_step": 14519, "epoch": 136, "lr": 8.259618125152028e-05} +{"train_loss": 0.01817840337753296, "grad_norm": 0.09806916862726212, "global_step": 14520, "epoch": 136, "lr": 8.259393381554257e-05} +{"train_loss": 0.016813043179869088, "grad_norm": 0.10998403280973434, "global_step": 14521, "epoch": 136, "lr": 8.259168626504395e-05, "val_loss": 0.020100148394703865} +{"train_loss": 0.01714547723531723, "grad_norm": 0.062294770032167435, "global_step": 14522, "epoch": 137, "lr": 8.258943860003232e-05} +{"train_loss": 0.01765391230583191, "grad_norm": 0.06836124509572983, "global_step": 14523, "epoch": 137, "lr": 8.258719082051556e-05} +{"train_loss": 0.015789411962032318, "grad_norm": 0.0848463773727417, "global_step": 14524, "epoch": 137, "lr": 8.258494292650159e-05} +{"train_loss": 0.019466426223516464, "grad_norm": 0.1152263656258583, "global_step": 14525, "epoch": 137, "lr": 8.258269491799831e-05} +{"train_loss": 0.01791560649871826, "grad_norm": 0.10713152587413788, "global_step": 14526, "epoch": 137, "lr": 8.25804467950136e-05} +{"train_loss": 0.0164271779358387, "grad_norm": 0.055056750774383545, "global_step": 14527, "epoch": 137, "lr": 8.257819855755536e-05} +{"train_loss": 0.017673434689641, "grad_norm": 0.06297197937965393, "global_step": 14528, "epoch": 137, "lr": 8.257595020563152e-05} +{"train_loss": 0.01582358032464981, "grad_norm": 0.060796160250902176, "global_step": 14529, "epoch": 137, "lr": 8.257370173924994e-05} +{"train_loss": 0.017357641831040382, "grad_norm": 0.08535070717334747, "global_step": 14530, "epoch": 137, "lr": 8.257145315841855e-05} +{"train_loss": 0.017074020579457283, "grad_norm": 0.07450834661722183, "global_step": 14531, "epoch": 137, "lr": 8.256920446314524e-05} +{"train_loss": 0.015457358211278915, "grad_norm": 0.059454258531332016, "global_step": 14532, "epoch": 137, "lr": 8.25669556534379e-05} +{"train_loss": 0.018984317779541016, "grad_norm": 0.09769567847251892, "global_step": 14533, "epoch": 137, "lr": 8.256470672930445e-05} +{"train_loss": 0.01809229701757431, "grad_norm": 0.1013302132487297, "global_step": 14534, "epoch": 137, "lr": 8.256245769075278e-05} +{"train_loss": 0.015395785681903362, "grad_norm": 0.09334103018045425, "global_step": 14535, "epoch": 137, "lr": 8.256020853779078e-05} +{"train_loss": 0.015673905611038208, "grad_norm": 0.0479583702981472, "global_step": 14536, "epoch": 137, "lr": 8.255795927042639e-05} +{"train_loss": 0.01763172820210457, "grad_norm": 0.07302998751401901, "global_step": 14537, "epoch": 137, "lr": 8.255570988866749e-05} +{"train_loss": 0.019759012386202812, "grad_norm": 0.05366884544491768, "global_step": 14538, "epoch": 137, "lr": 8.255346039252196e-05} +{"train_loss": 0.01748889498412609, "grad_norm": 0.06066444516181946, "global_step": 14539, "epoch": 137, "lr": 8.255121078199775e-05} +{"train_loss": 0.015705397352576256, "grad_norm": 0.06514210999011993, "global_step": 14540, "epoch": 137, "lr": 8.254896105710272e-05} +{"train_loss": 0.01595660112798214, "grad_norm": 0.04844663292169571, "global_step": 14541, "epoch": 137, "lr": 8.254671121784482e-05} +{"train_loss": 0.013353416696190834, "grad_norm": 0.09920178353786469, "global_step": 14542, "epoch": 137, "lr": 8.254446126423192e-05} +{"train_loss": 0.01695690117776394, "grad_norm": 0.06222013384103775, "global_step": 14543, "epoch": 137, "lr": 8.254221119627193e-05} +{"train_loss": 0.016344934701919556, "grad_norm": 0.05894914269447327, "global_step": 14544, "epoch": 137, "lr": 8.253996101397276e-05} +{"train_loss": 0.014438267797231674, "grad_norm": 0.0524272546172142, "global_step": 14545, "epoch": 137, "lr": 8.253771071734235e-05} +{"train_loss": 0.017130456864833832, "grad_norm": 0.08946220576763153, "global_step": 14546, "epoch": 137, "lr": 8.253546030638854e-05} +{"train_loss": 0.015432866290211678, "grad_norm": 0.1589621901512146, "global_step": 14547, "epoch": 137, "lr": 8.253320978111928e-05} +{"train_loss": 0.01509260106831789, "grad_norm": 0.12102849781513214, "global_step": 14548, "epoch": 137, "lr": 8.253095914154247e-05} +{"train_loss": 0.01721245050430298, "grad_norm": 0.1661279946565628, "global_step": 14549, "epoch": 137, "lr": 8.252870838766602e-05} +{"train_loss": 0.017592905089259148, "grad_norm": 0.1288347691297531, "global_step": 14550, "epoch": 137, "lr": 8.252645751949783e-05} +{"train_loss": 0.017346076667308807, "grad_norm": 0.07861945033073425, "global_step": 14551, "epoch": 137, "lr": 8.252420653704582e-05} +{"train_loss": 0.019352713599801064, "grad_norm": 0.07030996680259705, "global_step": 14552, "epoch": 137, "lr": 8.252195544031789e-05} +{"train_loss": 0.01507413387298584, "grad_norm": 0.11919758468866348, "global_step": 14553, "epoch": 137, "lr": 8.251970422932195e-05} +{"train_loss": 0.017559081315994263, "grad_norm": 0.13583843410015106, "global_step": 14554, "epoch": 137, "lr": 8.251745290406593e-05} +{"train_loss": 0.016558917239308357, "grad_norm": 0.05647164583206177, "global_step": 14555, "epoch": 137, "lr": 8.25152014645577e-05} +{"train_loss": 0.015830088406801224, "grad_norm": 0.15251095592975616, "global_step": 14556, "epoch": 137, "lr": 8.251294991080519e-05} +{"train_loss": 0.017341887578368187, "grad_norm": 0.19285701215267181, "global_step": 14557, "epoch": 137, "lr": 8.251069824281633e-05} +{"train_loss": 0.0179562009871006, "grad_norm": 0.18871565163135529, "global_step": 14558, "epoch": 137, "lr": 8.2508446460599e-05} +{"train_loss": 0.019758131355047226, "grad_norm": 0.09542101621627808, "global_step": 14559, "epoch": 137, "lr": 8.250619456416113e-05} +{"train_loss": 0.01343968790024519, "grad_norm": 0.06727010011672974, "global_step": 14560, "epoch": 137, "lr": 8.250394255351064e-05} +{"train_loss": 0.016446517780423164, "grad_norm": 0.06309054791927338, "global_step": 14561, "epoch": 137, "lr": 8.250169042865543e-05} +{"train_loss": 0.016804294660687447, "grad_norm": 0.0496828556060791, "global_step": 14562, "epoch": 137, "lr": 8.24994381896034e-05} +{"train_loss": 0.016535837203264236, "grad_norm": 0.08979081362485886, "global_step": 14563, "epoch": 137, "lr": 8.24971858363625e-05} +{"train_loss": 0.017023134976625443, "grad_norm": 0.048677463084459305, "global_step": 14564, "epoch": 137, "lr": 8.24949333689406e-05} +{"train_loss": 0.016111548990011215, "grad_norm": 0.058481812477111816, "global_step": 14565, "epoch": 137, "lr": 8.249268078734564e-05} +{"train_loss": 0.015672560781240463, "grad_norm": 0.15421713888645172, "global_step": 14566, "epoch": 137, "lr": 8.249042809158554e-05} +{"train_loss": 0.01593644917011261, "grad_norm": 0.1258537769317627, "global_step": 14567, "epoch": 137, "lr": 8.24881752816682e-05} +{"train_loss": 0.0188540481030941, "grad_norm": 0.10705353319644928, "global_step": 14568, "epoch": 137, "lr": 8.248592235760153e-05} +{"train_loss": 0.015895415097475052, "grad_norm": 0.06975631415843964, "global_step": 14569, "epoch": 137, "lr": 8.248366931939348e-05} +{"train_loss": 0.016024203971028328, "grad_norm": 0.09028918296098709, "global_step": 14570, "epoch": 137, "lr": 8.248141616705192e-05} +{"train_loss": 0.016226202249526978, "grad_norm": 0.05505509302020073, "global_step": 14571, "epoch": 137, "lr": 8.24791629005848e-05} +{"train_loss": 0.015524733811616898, "grad_norm": 0.05367333069443703, "global_step": 14572, "epoch": 137, "lr": 8.247690952000002e-05} +{"train_loss": 0.01732419803738594, "grad_norm": 0.0671505481004715, "global_step": 14573, "epoch": 137, "lr": 8.247465602530551e-05} +{"train_loss": 0.01575690321624279, "grad_norm": 0.06728121638298035, "global_step": 14574, "epoch": 137, "lr": 8.247240241650918e-05} +{"train_loss": 0.015778787434101105, "grad_norm": 0.0690254271030426, "global_step": 14575, "epoch": 137, "lr": 8.247014869361896e-05} +{"train_loss": 0.01639082096517086, "grad_norm": 0.07812511920928955, "global_step": 14576, "epoch": 137, "lr": 8.246789485664275e-05} +{"train_loss": 0.017958372831344604, "grad_norm": 0.06388655304908752, "global_step": 14577, "epoch": 137, "lr": 8.246564090558847e-05} +{"train_loss": 0.015276504680514336, "grad_norm": 0.1012847051024437, "global_step": 14578, "epoch": 137, "lr": 8.246338684046407e-05} +{"train_loss": 0.015562957152724266, "grad_norm": 0.06813708692789078, "global_step": 14579, "epoch": 137, "lr": 8.246113266127741e-05} +{"train_loss": 0.01846908964216709, "grad_norm": 0.06151537969708443, "global_step": 14580, "epoch": 137, "lr": 8.245887836803647e-05} +{"train_loss": 0.015411832369863987, "grad_norm": 0.0749506726861, "global_step": 14581, "epoch": 137, "lr": 8.245662396074914e-05} +{"train_loss": 0.01712225191295147, "grad_norm": 0.1247432753443718, "global_step": 14582, "epoch": 137, "lr": 8.245436943942336e-05} +{"train_loss": 0.017914539203047752, "grad_norm": 0.04860679432749748, "global_step": 14583, "epoch": 137, "lr": 8.245211480406703e-05} +{"train_loss": 0.015474075451493263, "grad_norm": 0.04856186360120773, "global_step": 14584, "epoch": 137, "lr": 8.24498600546881e-05} +{"train_loss": 0.019843365997076035, "grad_norm": 0.08325942605733871, "global_step": 14585, "epoch": 137, "lr": 8.244760519129444e-05} +{"train_loss": 0.017871025949716568, "grad_norm": 0.05029873177409172, "global_step": 14586, "epoch": 137, "lr": 8.244535021389402e-05} +{"train_loss": 0.015388313680887222, "grad_norm": 0.05983031541109085, "global_step": 14587, "epoch": 137, "lr": 8.244309512249476e-05} +{"train_loss": 0.01644214428961277, "grad_norm": 0.07818121463060379, "global_step": 14588, "epoch": 137, "lr": 8.244083991710457e-05} +{"train_loss": 0.01837392896413803, "grad_norm": 0.08997368067502975, "global_step": 14589, "epoch": 137, "lr": 8.243858459773137e-05} +{"train_loss": 0.017945565283298492, "grad_norm": 0.06119829788804054, "global_step": 14590, "epoch": 137, "lr": 8.24363291643831e-05} +{"train_loss": 0.015106448903679848, "grad_norm": 0.061399415135383606, "global_step": 14591, "epoch": 137, "lr": 8.243407361706766e-05} +{"train_loss": 0.014862770214676857, "grad_norm": 0.12134301662445068, "global_step": 14592, "epoch": 137, "lr": 8.2431817955793e-05} +{"train_loss": 0.016399726271629333, "grad_norm": 0.05142370983958244, "global_step": 14593, "epoch": 137, "lr": 8.242956218056704e-05} +{"train_loss": 0.015414953231811523, "grad_norm": 0.10470800846815109, "global_step": 14594, "epoch": 137, "lr": 8.24273062913977e-05} +{"train_loss": 0.014156213961541653, "grad_norm": 0.10611581057310104, "global_step": 14595, "epoch": 137, "lr": 8.242505028829291e-05} +{"train_loss": 0.01843699812889099, "grad_norm": 0.0819186270236969, "global_step": 14596, "epoch": 137, "lr": 8.242279417126059e-05} +{"train_loss": 0.01902535930275917, "grad_norm": 0.05336017161607742, "global_step": 14597, "epoch": 137, "lr": 8.242053794030866e-05} +{"train_loss": 0.016211431473493576, "grad_norm": 0.05103255435824394, "global_step": 14598, "epoch": 137, "lr": 8.241828159544508e-05} +{"train_loss": 0.017555247992277145, "grad_norm": 0.06547815352678299, "global_step": 14599, "epoch": 137, "lr": 8.241602513667774e-05} +{"train_loss": 0.015014134347438812, "grad_norm": 0.0690447986125946, "global_step": 14600, "epoch": 137, "lr": 8.24137685640146e-05} +{"train_loss": 0.017736420035362244, "grad_norm": 0.13421256840229034, "global_step": 14601, "epoch": 137, "lr": 8.241151187746356e-05} +{"train_loss": 0.018257111310958862, "grad_norm": 0.07034237682819366, "global_step": 14602, "epoch": 137, "lr": 8.240925507703258e-05} +{"train_loss": 0.016461487859487534, "grad_norm": 0.07119947671890259, "global_step": 14603, "epoch": 137, "lr": 8.240699816272955e-05} +{"train_loss": 0.018610917031764984, "grad_norm": 0.08685168623924255, "global_step": 14604, "epoch": 137, "lr": 8.240474113456245e-05} +{"train_loss": 0.015639206394553185, "grad_norm": 0.05797291547060013, "global_step": 14605, "epoch": 137, "lr": 8.240248399253916e-05} +{"train_loss": 0.0172070674598217, "grad_norm": 0.10695503652095795, "global_step": 14606, "epoch": 137, "lr": 8.240022673666763e-05} +{"train_loss": 0.018648117780685425, "grad_norm": 0.10392674058675766, "global_step": 14607, "epoch": 137, "lr": 8.239796936695581e-05} +{"train_loss": 0.014473692514002323, "grad_norm": 0.05721409246325493, "global_step": 14608, "epoch": 137, "lr": 8.23957118834116e-05} +{"train_loss": 0.01694047451019287, "grad_norm": 0.06081207096576691, "global_step": 14609, "epoch": 137, "lr": 8.239345428604294e-05} +{"train_loss": 0.016510039567947388, "grad_norm": 0.15498173236846924, "global_step": 14610, "epoch": 137, "lr": 8.23911965748578e-05} +{"train_loss": 0.017613738775253296, "grad_norm": 0.05160185694694519, "global_step": 14611, "epoch": 137, "lr": 8.238893874986407e-05} +{"train_loss": 0.015230342745780945, "grad_norm": 0.15272732079029083, "global_step": 14612, "epoch": 137, "lr": 8.238668081106968e-05} +{"train_loss": 0.01754641905426979, "grad_norm": 0.06007194146513939, "global_step": 14613, "epoch": 137, "lr": 8.238442275848259e-05} +{"train_loss": 0.01811995729804039, "grad_norm": 0.07629053294658661, "global_step": 14614, "epoch": 137, "lr": 8.238216459211072e-05} +{"train_loss": 0.018437519669532776, "grad_norm": 0.07651401311159134, "global_step": 14615, "epoch": 137, "lr": 8.2379906311962e-05} +{"train_loss": 0.016230300068855286, "grad_norm": 0.11276504397392273, "global_step": 14616, "epoch": 137, "lr": 8.237764791804439e-05} +{"train_loss": 0.015412929467856884, "grad_norm": 0.10725601017475128, "global_step": 14617, "epoch": 137, "lr": 8.237538941036578e-05} +{"train_loss": 0.015513977035880089, "grad_norm": 0.05503618344664574, "global_step": 14618, "epoch": 137, "lr": 8.237313078893413e-05} +{"train_loss": 0.017338240519165993, "grad_norm": 0.05059340223670006, "global_step": 14619, "epoch": 137, "lr": 8.237087205375741e-05} +{"train_loss": 0.016886930912733078, "grad_norm": 0.0631137564778328, "global_step": 14620, "epoch": 137, "lr": 8.23686132048435e-05} +{"train_loss": 0.021265357732772827, "grad_norm": 0.09398605674505234, "global_step": 14621, "epoch": 137, "lr": 8.236635424220036e-05} +{"train_loss": 0.017115706577897072, "grad_norm": 0.07644982635974884, "global_step": 14622, "epoch": 137, "lr": 8.236409516583595e-05} +{"train_loss": 0.01880868896842003, "grad_norm": 0.058841973543167114, "global_step": 14623, "epoch": 137, "lr": 8.236183597575815e-05} +{"train_loss": 0.016024576500058174, "grad_norm": 0.08458118885755539, "global_step": 14624, "epoch": 137, "lr": 8.235957667197494e-05} +{"train_loss": 0.019090816378593445, "grad_norm": 0.061506565660238266, "global_step": 14625, "epoch": 137, "lr": 8.235731725449427e-05} +{"train_loss": 0.013919491320848465, "grad_norm": 0.0687641128897667, "global_step": 14626, "epoch": 137, "lr": 8.235505772332404e-05} +{"train_loss": 0.016792733011380682, "grad_norm": 0.0766458511352539, "global_step": 14627, "epoch": 137, "lr": 8.235279807847223e-05, "val_loss": 0.018645521253347397} +{"train_loss": 0.01838713139295578, "grad_norm": 0.05405103787779808, "global_step": 14628, "epoch": 138, "lr": 8.235053831994673e-05} +{"train_loss": 0.01806602254509926, "grad_norm": 0.046018291264772415, "global_step": 14629, "epoch": 138, "lr": 8.234827844775553e-05} +{"train_loss": 0.014413423836231232, "grad_norm": 0.058791209012269974, "global_step": 14630, "epoch": 138, "lr": 8.234601846190652e-05} +{"train_loss": 0.015352442860603333, "grad_norm": 0.139888197183609, "global_step": 14631, "epoch": 138, "lr": 8.23437583624077e-05} +{"train_loss": 0.013946863822638988, "grad_norm": 0.10856132209300995, "global_step": 14632, "epoch": 138, "lr": 8.234149814926696e-05} +{"train_loss": 0.013511575758457184, "grad_norm": 0.08552727848291397, "global_step": 14633, "epoch": 138, "lr": 8.233923782249226e-05} +{"train_loss": 0.018221870064735413, "grad_norm": 0.05208185687661171, "global_step": 14634, "epoch": 138, "lr": 8.233697738209157e-05} +{"train_loss": 0.019246213138103485, "grad_norm": 0.06395279616117477, "global_step": 14635, "epoch": 138, "lr": 8.233471682807277e-05} +{"train_loss": 0.016258206218481064, "grad_norm": 0.05285953730344772, "global_step": 14636, "epoch": 138, "lr": 8.233245616044383e-05} +{"train_loss": 0.018454167991876602, "grad_norm": 0.06796106696128845, "global_step": 14637, "epoch": 138, "lr": 8.233019537921273e-05} +{"train_loss": 0.016458213329315186, "grad_norm": 0.09051496535539627, "global_step": 14638, "epoch": 138, "lr": 8.232793448438737e-05} +{"train_loss": 0.01635543443262577, "grad_norm": 0.0708678811788559, "global_step": 14639, "epoch": 138, "lr": 8.232567347597571e-05} +{"train_loss": 0.016449466347694397, "grad_norm": 0.14074291288852692, "global_step": 14640, "epoch": 138, "lr": 8.232341235398569e-05} +{"train_loss": 0.01643294095993042, "grad_norm": 0.05883801355957985, "global_step": 14641, "epoch": 138, "lr": 8.232115111842527e-05} +{"train_loss": 0.016436483711004257, "grad_norm": 0.14533205330371857, "global_step": 14642, "epoch": 138, "lr": 8.231888976930235e-05} +{"train_loss": 0.01720273867249489, "grad_norm": 0.08198633044958115, "global_step": 14643, "epoch": 138, "lr": 8.231662830662494e-05} +{"train_loss": 0.017102643847465515, "grad_norm": 0.09724175930023193, "global_step": 14644, "epoch": 138, "lr": 8.231436673040093e-05} +{"train_loss": 0.015933100134134293, "grad_norm": 0.07752520591020584, "global_step": 14645, "epoch": 138, "lr": 8.23121050406383e-05} +{"train_loss": 0.0161312073469162, "grad_norm": 0.10579792410135269, "global_step": 14646, "epoch": 138, "lr": 8.230984323734497e-05} +{"train_loss": 0.0177781842648983, "grad_norm": 0.1071091964840889, "global_step": 14647, "epoch": 138, "lr": 8.230758132052891e-05} +{"train_loss": 0.018527880311012268, "grad_norm": 0.06455649435520172, "global_step": 14648, "epoch": 138, "lr": 8.230531929019807e-05} +{"train_loss": 0.017859986051917076, "grad_norm": 0.13625402748584747, "global_step": 14649, "epoch": 138, "lr": 8.230305714636037e-05} +{"train_loss": 0.017203427851200104, "grad_norm": 0.0920243114233017, "global_step": 14650, "epoch": 138, "lr": 8.230079488902379e-05} +{"train_loss": 0.013547874987125397, "grad_norm": 0.07379955053329468, "global_step": 14651, "epoch": 138, "lr": 8.229853251819626e-05} +{"train_loss": 0.01854771375656128, "grad_norm": 0.09038088470697403, "global_step": 14652, "epoch": 138, "lr": 8.229627003388574e-05} +{"train_loss": 0.017002757638692856, "grad_norm": 0.10788153111934662, "global_step": 14653, "epoch": 138, "lr": 8.229400743610015e-05} +{"train_loss": 0.01591269113123417, "grad_norm": 0.049099355936050415, "global_step": 14654, "epoch": 138, "lr": 8.229174472484748e-05} +{"train_loss": 0.01529804989695549, "grad_norm": 0.052576497197151184, "global_step": 14655, "epoch": 138, "lr": 8.228948190013565e-05} +{"train_loss": 0.016369177028536797, "grad_norm": 0.05684877932071686, "global_step": 14656, "epoch": 138, "lr": 8.228721896197265e-05} +{"train_loss": 0.0187535360455513, "grad_norm": 0.10464080423116684, "global_step": 14657, "epoch": 138, "lr": 8.228495591036638e-05} +{"train_loss": 0.01609184592962265, "grad_norm": 0.08612069487571716, "global_step": 14658, "epoch": 138, "lr": 8.228269274532482e-05} +{"train_loss": 0.015822479501366615, "grad_norm": 0.08174832910299301, "global_step": 14659, "epoch": 138, "lr": 8.228042946685591e-05} +{"train_loss": 0.014749657362699509, "grad_norm": 0.05882661044597626, "global_step": 14660, "epoch": 138, "lr": 8.227816607496762e-05} +{"train_loss": 0.01761467009782791, "grad_norm": 0.05475129559636116, "global_step": 14661, "epoch": 138, "lr": 8.227590256966788e-05} +{"train_loss": 0.016190003603696823, "grad_norm": 0.07287407666444778, "global_step": 14662, "epoch": 138, "lr": 8.227363895096468e-05} +{"train_loss": 0.014422038570046425, "grad_norm": 0.06349857151508331, "global_step": 14663, "epoch": 138, "lr": 8.227137521886592e-05} +{"train_loss": 0.015329298563301563, "grad_norm": 0.11091140657663345, "global_step": 14664, "epoch": 138, "lr": 8.22691113733796e-05} +{"train_loss": 0.020290642976760864, "grad_norm": 0.08467282354831696, "global_step": 14665, "epoch": 138, "lr": 8.226684741451364e-05} +{"train_loss": 0.01804133504629135, "grad_norm": 0.12470446527004242, "global_step": 14666, "epoch": 138, "lr": 8.226458334227602e-05} +{"train_loss": 0.01632176712155342, "grad_norm": 0.06466346234083176, "global_step": 14667, "epoch": 138, "lr": 8.226231915667469e-05} +{"train_loss": 0.01560034230351448, "grad_norm": 0.04974719136953354, "global_step": 14668, "epoch": 138, "lr": 8.22600548577176e-05} +{"train_loss": 0.016992514953017235, "grad_norm": 0.050223466008901596, "global_step": 14669, "epoch": 138, "lr": 8.22577904454127e-05} +{"train_loss": 0.01645801216363907, "grad_norm": 0.07569951564073563, "global_step": 14670, "epoch": 138, "lr": 8.225552591976794e-05} +{"train_loss": 0.017414668574929237, "grad_norm": 0.05956381931900978, "global_step": 14671, "epoch": 138, "lr": 8.225326128079131e-05} +{"train_loss": 0.01877042092382908, "grad_norm": 0.07036875188350677, "global_step": 14672, "epoch": 138, "lr": 8.225099652849076e-05} +{"train_loss": 0.016156461089849472, "grad_norm": 0.08853967487812042, "global_step": 14673, "epoch": 138, "lr": 8.224873166287421e-05} +{"train_loss": 0.021092811599373817, "grad_norm": 0.07429075241088867, "global_step": 14674, "epoch": 138, "lr": 8.224646668394963e-05} +{"train_loss": 0.018497254699468613, "grad_norm": 0.06183556094765663, "global_step": 14675, "epoch": 138, "lr": 8.224420159172504e-05} +{"train_loss": 0.01646370440721512, "grad_norm": 0.07567435503005981, "global_step": 14676, "epoch": 138, "lr": 8.224193638620829e-05} +{"train_loss": 0.01303605642169714, "grad_norm": 0.04884392395615578, "global_step": 14677, "epoch": 138, "lr": 8.223967106740744e-05} +{"train_loss": 0.018638137727975845, "grad_norm": 0.09986836463212967, "global_step": 14678, "epoch": 138, "lr": 8.223740563533038e-05} +{"train_loss": 0.015399823896586895, "grad_norm": 0.05534377321600914, "global_step": 14679, "epoch": 138, "lr": 8.22351400899851e-05} +{"train_loss": 0.016796130686998367, "grad_norm": 0.07023132592439651, "global_step": 14680, "epoch": 138, "lr": 8.223287443137957e-05} +{"train_loss": 0.019106006249785423, "grad_norm": 0.07707783579826355, "global_step": 14681, "epoch": 138, "lr": 8.223060865952174e-05} +{"train_loss": 0.01706613041460514, "grad_norm": 0.04769103601574898, "global_step": 14682, "epoch": 138, "lr": 8.222834277441955e-05} +{"train_loss": 0.015745390206575394, "grad_norm": 0.05349842458963394, "global_step": 14683, "epoch": 138, "lr": 8.222607677608098e-05} +{"train_loss": 0.017582640051841736, "grad_norm": 0.06668878346681595, "global_step": 14684, "epoch": 138, "lr": 8.2223810664514e-05} +{"train_loss": 0.019219445064663887, "grad_norm": 0.06414090096950531, "global_step": 14685, "epoch": 138, "lr": 8.222154443972656e-05} +{"train_loss": 0.017449643462896347, "grad_norm": 0.07628770172595978, "global_step": 14686, "epoch": 138, "lr": 8.221927810172663e-05} +{"train_loss": 0.016322582960128784, "grad_norm": 0.12834453582763672, "global_step": 14687, "epoch": 138, "lr": 8.221701165052216e-05} +{"train_loss": 0.017697203904390335, "grad_norm": 0.07263309508562088, "global_step": 14688, "epoch": 138, "lr": 8.221474508612112e-05} +{"train_loss": 0.017541468143463135, "grad_norm": 0.10125025361776352, "global_step": 14689, "epoch": 138, "lr": 8.221247840853148e-05} +{"train_loss": 0.01436547003686428, "grad_norm": 0.0506582148373127, "global_step": 14690, "epoch": 138, "lr": 8.221021161776121e-05} +{"train_loss": 0.01920037344098091, "grad_norm": 0.05758068338036537, "global_step": 14691, "epoch": 138, "lr": 8.220794471381825e-05} +{"train_loss": 0.01646297425031662, "grad_norm": 0.06763514131307602, "global_step": 14692, "epoch": 138, "lr": 8.220567769671059e-05} +{"train_loss": 0.014138253405690193, "grad_norm": 0.06921529769897461, "global_step": 14693, "epoch": 138, "lr": 8.220341056644618e-05} +{"train_loss": 0.016705859452486038, "grad_norm": 0.08199255168437958, "global_step": 14694, "epoch": 138, "lr": 8.220114332303299e-05} +{"train_loss": 0.01856101304292679, "grad_norm": 0.10331742465496063, "global_step": 14695, "epoch": 138, "lr": 8.2198875966479e-05} +{"train_loss": 0.01395491324365139, "grad_norm": 0.12631653249263763, "global_step": 14696, "epoch": 138, "lr": 8.219660849679214e-05} +{"train_loss": 0.018905291333794594, "grad_norm": 0.05822570249438286, "global_step": 14697, "epoch": 138, "lr": 8.219434091398039e-05} +{"train_loss": 0.016832508146762848, "grad_norm": 0.055648110806941986, "global_step": 14698, "epoch": 138, "lr": 8.219207321805175e-05} +{"train_loss": 0.016269516199827194, "grad_norm": 0.12382438033819199, "global_step": 14699, "epoch": 138, "lr": 8.218980540901418e-05} +{"train_loss": 0.017571937292814255, "grad_norm": 0.05912506952881813, "global_step": 14700, "epoch": 138, "lr": 8.21875374868756e-05} +{"train_loss": 0.015981990844011307, "grad_norm": 0.10422217845916748, "global_step": 14701, "epoch": 138, "lr": 8.218526945164402e-05} +{"train_loss": 0.018517181277275085, "grad_norm": 0.0742100179195404, "global_step": 14702, "epoch": 138, "lr": 8.218300130332741e-05} +{"train_loss": 0.018322894349694252, "grad_norm": 0.06349976360797882, "global_step": 14703, "epoch": 138, "lr": 8.218073304193372e-05} +{"train_loss": 0.014798307791352272, "grad_norm": 0.0748581662774086, "global_step": 14704, "epoch": 138, "lr": 8.217846466747095e-05} +{"train_loss": 0.016035888344049454, "grad_norm": 0.07022316008806229, "global_step": 14705, "epoch": 138, "lr": 8.217619617994702e-05} +{"train_loss": 0.01810322143137455, "grad_norm": 0.05281223729252815, "global_step": 14706, "epoch": 138, "lr": 8.217392757936994e-05} +{"train_loss": 0.014323815703392029, "grad_norm": 0.055692218244075775, "global_step": 14707, "epoch": 138, "lr": 8.217165886574767e-05} +{"train_loss": 0.01657235249876976, "grad_norm": 0.05080118030309677, "global_step": 14708, "epoch": 138, "lr": 8.216939003908818e-05} +{"train_loss": 0.016126584261655807, "grad_norm": 0.08855914324522018, "global_step": 14709, "epoch": 138, "lr": 8.216712109939944e-05} +{"train_loss": 0.015555812045931816, "grad_norm": 0.06182115897536278, "global_step": 14710, "epoch": 138, "lr": 8.216485204668945e-05} +{"train_loss": 0.017908833920955658, "grad_norm": 0.0523064061999321, "global_step": 14711, "epoch": 138, "lr": 8.216258288096613e-05} +{"train_loss": 0.01445708330720663, "grad_norm": 0.07497356832027435, "global_step": 14712, "epoch": 138, "lr": 8.21603136022375e-05} +{"train_loss": 0.021369140595197678, "grad_norm": 0.060907475650310516, "global_step": 14713, "epoch": 138, "lr": 8.21580442105115e-05} +{"train_loss": 0.018722597509622574, "grad_norm": 0.06008793041110039, "global_step": 14714, "epoch": 138, "lr": 8.215577470579612e-05} +{"train_loss": 0.015910208225250244, "grad_norm": 0.06737291812896729, "global_step": 14715, "epoch": 138, "lr": 8.215350508809934e-05} +{"train_loss": 0.017958657816052437, "grad_norm": 0.0672973170876503, "global_step": 14716, "epoch": 138, "lr": 8.215123535742912e-05} +{"train_loss": 0.015411749482154846, "grad_norm": 0.06036462262272835, "global_step": 14717, "epoch": 138, "lr": 8.214896551379346e-05} +{"train_loss": 0.016174308955669403, "grad_norm": 0.1087035983800888, "global_step": 14718, "epoch": 138, "lr": 8.214669555720028e-05} +{"train_loss": 0.01717701368033886, "grad_norm": 0.08164626359939575, "global_step": 14719, "epoch": 138, "lr": 8.214442548765762e-05} +{"train_loss": 0.016389135271310806, "grad_norm": 0.11556696146726608, "global_step": 14720, "epoch": 138, "lr": 8.214215530517341e-05} +{"train_loss": 0.018306251615285873, "grad_norm": 0.0695648193359375, "global_step": 14721, "epoch": 138, "lr": 8.213988500975566e-05} +{"train_loss": 0.017664682120084763, "grad_norm": 0.07144930213689804, "global_step": 14722, "epoch": 138, "lr": 8.213761460141233e-05} +{"train_loss": 0.016501668840646744, "grad_norm": 0.07556574791669846, "global_step": 14723, "epoch": 138, "lr": 8.213534408015139e-05} +{"train_loss": 0.015498470515012741, "grad_norm": 0.057775650173425674, "global_step": 14724, "epoch": 138, "lr": 8.213307344598082e-05} +{"train_loss": 0.016488056629896164, "grad_norm": 0.11589401960372925, "global_step": 14725, "epoch": 138, "lr": 8.213080269890863e-05} +{"train_loss": 0.01711529865860939, "grad_norm": 0.12633490562438965, "global_step": 14726, "epoch": 138, "lr": 8.212853183894276e-05} +{"train_loss": 0.015293560922145844, "grad_norm": 0.08529315143823624, "global_step": 14727, "epoch": 138, "lr": 8.212626086609119e-05} +{"train_loss": 0.016821041703224182, "grad_norm": 0.07727470248937607, "global_step": 14728, "epoch": 138, "lr": 8.212398978036192e-05} +{"train_loss": 0.013482902199029922, "grad_norm": 0.1265697479248047, "global_step": 14729, "epoch": 138, "lr": 8.212171858176292e-05} +{"train_loss": 0.01681647077202797, "grad_norm": 0.13057921826839447, "global_step": 14730, "epoch": 138, "lr": 8.211944727030217e-05} +{"train_loss": 0.014691202901303768, "grad_norm": 0.1014060527086258, "global_step": 14731, "epoch": 138, "lr": 8.211717584598765e-05} +{"train_loss": 0.016778333112597466, "grad_norm": 0.0992630198597908, "global_step": 14732, "epoch": 138, "lr": 8.211490430882733e-05} +{"train_loss": 0.01677073779041475, "grad_norm": 0.10424355417490005, "global_step": 14733, "epoch": 138, "lr": 8.211263265882923e-05, "val_loss": 0.019247930496931076} +{"train_loss": 0.01835261844098568, "grad_norm": 0.058732420206069946, "global_step": 14734, "epoch": 139, "lr": 8.211036089600129e-05} +{"train_loss": 0.017594417557120323, "grad_norm": 0.051894985139369965, "global_step": 14735, "epoch": 139, "lr": 8.21080890203515e-05} +{"train_loss": 0.016413981094956398, "grad_norm": 0.07740587741136551, "global_step": 14736, "epoch": 139, "lr": 8.210581703188784e-05} +{"train_loss": 0.015868468210101128, "grad_norm": 0.05359819903969765, "global_step": 14737, "epoch": 139, "lr": 8.210354493061833e-05} +{"train_loss": 0.016988638788461685, "grad_norm": 0.0540069043636322, "global_step": 14738, "epoch": 139, "lr": 8.21012727165509e-05} +{"train_loss": 0.014219441451132298, "grad_norm": 0.05064855515956879, "global_step": 14739, "epoch": 139, "lr": 8.209900038969357e-05} +{"train_loss": 0.01563538983464241, "grad_norm": 0.06002989411354065, "global_step": 14740, "epoch": 139, "lr": 8.209672795005432e-05} +{"train_loss": 0.01931976154446602, "grad_norm": 0.093703992664814, "global_step": 14741, "epoch": 139, "lr": 8.209445539764111e-05} +{"train_loss": 0.017606079578399658, "grad_norm": 0.188058540225029, "global_step": 14742, "epoch": 139, "lr": 8.209218273246195e-05} +{"train_loss": 0.014820521697402, "grad_norm": 0.04644976928830147, "global_step": 14743, "epoch": 139, "lr": 8.208990995452481e-05} +{"train_loss": 0.01730269566178322, "grad_norm": 0.05658353492617607, "global_step": 14744, "epoch": 139, "lr": 8.208763706383768e-05} +{"train_loss": 0.01592542789876461, "grad_norm": 0.14191925525665283, "global_step": 14745, "epoch": 139, "lr": 8.208536406040857e-05} +{"train_loss": 0.01825416088104248, "grad_norm": 0.07716328650712967, "global_step": 14746, "epoch": 139, "lr": 8.208309094424543e-05} +{"train_loss": 0.0180269293487072, "grad_norm": 0.06535372138023376, "global_step": 14747, "epoch": 139, "lr": 8.208081771535626e-05} +{"train_loss": 0.019289886578917503, "grad_norm": 0.1009310781955719, "global_step": 14748, "epoch": 139, "lr": 8.207854437374905e-05} +{"train_loss": 0.01734568551182747, "grad_norm": 0.10148820281028748, "global_step": 14749, "epoch": 139, "lr": 8.207627091943178e-05} +{"train_loss": 0.013485264033079147, "grad_norm": 0.05176601931452751, "global_step": 14750, "epoch": 139, "lr": 8.207399735241243e-05} +{"train_loss": 0.01805911213159561, "grad_norm": 0.11032194644212723, "global_step": 14751, "epoch": 139, "lr": 8.207172367269902e-05} +{"train_loss": 0.016839001327753067, "grad_norm": 0.07901544123888016, "global_step": 14752, "epoch": 139, "lr": 8.206944988029952e-05} +{"train_loss": 0.01686154119670391, "grad_norm": 0.11379458010196686, "global_step": 14753, "epoch": 139, "lr": 8.206717597522191e-05} +{"train_loss": 0.015642322599887848, "grad_norm": 0.10763027518987656, "global_step": 14754, "epoch": 139, "lr": 8.206490195747421e-05} +{"train_loss": 0.016368014737963676, "grad_norm": 0.1665540486574173, "global_step": 14755, "epoch": 139, "lr": 8.206262782706437e-05} +{"train_loss": 0.01592939905822277, "grad_norm": 0.13117562234401703, "global_step": 14756, "epoch": 139, "lr": 8.206035358400041e-05} +{"train_loss": 0.01803169958293438, "grad_norm": 0.0742528960108757, "global_step": 14757, "epoch": 139, "lr": 8.205807922829032e-05} +{"train_loss": 0.016002103686332703, "grad_norm": 0.09412451088428497, "global_step": 14758, "epoch": 139, "lr": 8.205580475994206e-05} +{"train_loss": 0.018473541364073753, "grad_norm": 0.12063486874103546, "global_step": 14759, "epoch": 139, "lr": 8.205353017896365e-05} +{"train_loss": 0.01799321547150612, "grad_norm": 0.07028141617774963, "global_step": 14760, "epoch": 139, "lr": 8.205125548536308e-05} +{"train_loss": 0.017991583794355392, "grad_norm": 0.07659293711185455, "global_step": 14761, "epoch": 139, "lr": 8.204898067914832e-05} +{"train_loss": 0.016825363039970398, "grad_norm": 0.16825298964977264, "global_step": 14762, "epoch": 139, "lr": 8.20467057603274e-05} +{"train_loss": 0.016689404845237732, "grad_norm": 0.1366215944290161, "global_step": 14763, "epoch": 139, "lr": 8.204443072890829e-05} +{"train_loss": 0.018551046028733253, "grad_norm": 0.08418842405080795, "global_step": 14764, "epoch": 139, "lr": 8.204215558489897e-05} +{"train_loss": 0.016083456575870514, "grad_norm": 0.0978783592581749, "global_step": 14765, "epoch": 139, "lr": 8.203988032830748e-05} +{"train_loss": 0.017046937718987465, "grad_norm": 0.05457128956913948, "global_step": 14766, "epoch": 139, "lr": 8.203760495914174e-05} +{"train_loss": 0.016758941113948822, "grad_norm": 0.10980632901191711, "global_step": 14767, "epoch": 139, "lr": 8.203532947740983e-05} +{"train_loss": 0.020723367109894753, "grad_norm": 0.06742192059755325, "global_step": 14768, "epoch": 139, "lr": 8.203305388311968e-05} +{"train_loss": 0.018692240118980408, "grad_norm": 0.07903546839952469, "global_step": 14769, "epoch": 139, "lr": 8.203077817627933e-05} +{"train_loss": 0.01627739518880844, "grad_norm": 0.07489319890737534, "global_step": 14770, "epoch": 139, "lr": 8.202850235689674e-05} +{"train_loss": 0.01575700379908085, "grad_norm": 0.0816296935081482, "global_step": 14771, "epoch": 139, "lr": 8.202622642497992e-05} +{"train_loss": 0.014812730252742767, "grad_norm": 0.04630427435040474, "global_step": 14772, "epoch": 139, "lr": 8.202395038053688e-05} +{"train_loss": 0.01647653430700302, "grad_norm": 0.0689968690276146, "global_step": 14773, "epoch": 139, "lr": 8.202167422357559e-05} +{"train_loss": 0.017369456589221954, "grad_norm": 0.060400452464818954, "global_step": 14774, "epoch": 139, "lr": 8.201939795410407e-05} +{"train_loss": 0.01683644764125347, "grad_norm": 0.08161377906799316, "global_step": 14775, "epoch": 139, "lr": 8.20171215721303e-05} +{"train_loss": 0.014887852594256401, "grad_norm": 0.05975271388888359, "global_step": 14776, "epoch": 139, "lr": 8.201484507766231e-05} +{"train_loss": 0.01574397087097168, "grad_norm": 0.11185768246650696, "global_step": 14777, "epoch": 139, "lr": 8.201256847070806e-05} +{"train_loss": 0.0160911176353693, "grad_norm": 0.07260598987340927, "global_step": 14778, "epoch": 139, "lr": 8.201029175127559e-05} +{"train_loss": 0.015954967588186264, "grad_norm": 0.05906028300523758, "global_step": 14779, "epoch": 139, "lr": 8.200801491937284e-05} +{"train_loss": 0.01761506497859955, "grad_norm": 0.12593461573123932, "global_step": 14780, "epoch": 139, "lr": 8.200573797500788e-05} +{"train_loss": 0.015023555606603622, "grad_norm": 0.10425777733325958, "global_step": 14781, "epoch": 139, "lr": 8.200346091818866e-05} +{"train_loss": 0.014291338622570038, "grad_norm": 0.07100332528352737, "global_step": 14782, "epoch": 139, "lr": 8.200118374892318e-05} +{"train_loss": 0.016809659078717232, "grad_norm": 0.09176868945360184, "global_step": 14783, "epoch": 139, "lr": 8.199890646721948e-05} +{"train_loss": 0.01789221167564392, "grad_norm": 0.13761702179908752, "global_step": 14784, "epoch": 139, "lr": 8.199662907308554e-05} +{"train_loss": 0.016197532415390015, "grad_norm": 0.10325917601585388, "global_step": 14785, "epoch": 139, "lr": 8.199435156652935e-05} +{"train_loss": 0.015041351318359375, "grad_norm": 0.04440217465162277, "global_step": 14786, "epoch": 139, "lr": 8.199207394755893e-05} +{"train_loss": 0.01744016632437706, "grad_norm": 0.08594387024641037, "global_step": 14787, "epoch": 139, "lr": 8.198979621618226e-05} +{"train_loss": 0.01727377250790596, "grad_norm": 0.05612506717443466, "global_step": 14788, "epoch": 139, "lr": 8.198751837240736e-05} +{"train_loss": 0.016440019011497498, "grad_norm": 0.06797758489847183, "global_step": 14789, "epoch": 139, "lr": 8.198524041624225e-05} +{"train_loss": 0.01634812541306019, "grad_norm": 0.06648477911949158, "global_step": 14790, "epoch": 139, "lr": 8.198296234769492e-05} +{"train_loss": 0.0199093259871006, "grad_norm": 0.14326177537441254, "global_step": 14791, "epoch": 139, "lr": 8.198068416677335e-05} +{"train_loss": 0.01641238108277321, "grad_norm": 0.05216946825385094, "global_step": 14792, "epoch": 139, "lr": 8.197840587348557e-05} +{"train_loss": 0.016474133357405663, "grad_norm": 0.05800168216228485, "global_step": 14793, "epoch": 139, "lr": 8.197612746783956e-05} +{"train_loss": 0.015951871871948242, "grad_norm": 0.06048442795872688, "global_step": 14794, "epoch": 139, "lr": 8.197384894984336e-05} +{"train_loss": 0.014494551345705986, "grad_norm": 0.05436202511191368, "global_step": 14795, "epoch": 139, "lr": 8.197157031950498e-05} +{"train_loss": 0.01774117723107338, "grad_norm": 0.048008039593696594, "global_step": 14796, "epoch": 139, "lr": 8.196929157683238e-05} +{"train_loss": 0.014624256640672684, "grad_norm": 0.11287713050842285, "global_step": 14797, "epoch": 139, "lr": 8.19670127218336e-05} +{"train_loss": 0.01779700070619583, "grad_norm": 0.06000183895230293, "global_step": 14798, "epoch": 139, "lr": 8.196473375451665e-05} +{"train_loss": 0.014356471598148346, "grad_norm": 0.09038825333118439, "global_step": 14799, "epoch": 139, "lr": 8.19624546748895e-05} +{"train_loss": 0.017049310728907585, "grad_norm": 0.06672094762325287, "global_step": 14800, "epoch": 139, "lr": 8.19601754829602e-05} +{"train_loss": 0.017969854176044464, "grad_norm": 0.1277398020029068, "global_step": 14801, "epoch": 139, "lr": 8.195789617873674e-05} +{"train_loss": 0.016071079298853874, "grad_norm": 0.05426224321126938, "global_step": 14802, "epoch": 139, "lr": 8.195561676222714e-05} +{"train_loss": 0.015796754509210587, "grad_norm": 0.06328684091567993, "global_step": 14803, "epoch": 139, "lr": 8.195333723343938e-05} +{"train_loss": 0.017013953998684883, "grad_norm": 0.07258988916873932, "global_step": 14804, "epoch": 139, "lr": 8.195105759238152e-05} +{"train_loss": 0.016182387247681618, "grad_norm": 0.07449859380722046, "global_step": 14805, "epoch": 139, "lr": 8.194877783906152e-05} +{"train_loss": 0.016918504610657692, "grad_norm": 0.08391391485929489, "global_step": 14806, "epoch": 139, "lr": 8.194649797348741e-05} +{"train_loss": 0.014642116613686085, "grad_norm": 0.09395674616098404, "global_step": 14807, "epoch": 139, "lr": 8.19442179956672e-05} +{"train_loss": 0.016011405736207962, "grad_norm": 0.10700652003288269, "global_step": 14808, "epoch": 139, "lr": 8.19419379056089e-05} +{"train_loss": 0.016376014798879623, "grad_norm": 0.07916353642940521, "global_step": 14809, "epoch": 139, "lr": 8.193965770332052e-05} +{"train_loss": 0.01568593457341194, "grad_norm": 0.08409450203180313, "global_step": 14810, "epoch": 139, "lr": 8.193737738881007e-05} +{"train_loss": 0.016544228419661522, "grad_norm": 0.08579675853252411, "global_step": 14811, "epoch": 139, "lr": 8.193509696208557e-05} +{"train_loss": 0.017258208245038986, "grad_norm": 0.08523406088352203, "global_step": 14812, "epoch": 139, "lr": 8.193281642315502e-05} +{"train_loss": 0.01687292382121086, "grad_norm": 0.11503898352384567, "global_step": 14813, "epoch": 139, "lr": 8.193053577202645e-05} +{"train_loss": 0.015469473786652088, "grad_norm": 0.06942590326070786, "global_step": 14814, "epoch": 139, "lr": 8.192825500870787e-05} +{"train_loss": 0.01480034738779068, "grad_norm": 0.09385959059000015, "global_step": 14815, "epoch": 139, "lr": 8.192597413320726e-05} +{"train_loss": 0.019832219928503036, "grad_norm": 0.08988790214061737, "global_step": 14816, "epoch": 139, "lr": 8.192369314553269e-05} +{"train_loss": 0.01712273433804512, "grad_norm": 0.0659896731376648, "global_step": 14817, "epoch": 139, "lr": 8.192141204569213e-05} +{"train_loss": 0.016420690342783928, "grad_norm": 0.06925436854362488, "global_step": 14818, "epoch": 139, "lr": 8.191913083369362e-05} +{"train_loss": 0.01719493791460991, "grad_norm": 0.09392987191677094, "global_step": 14819, "epoch": 139, "lr": 8.191684950954517e-05} +{"train_loss": 0.01689598709344864, "grad_norm": 0.09828969091176987, "global_step": 14820, "epoch": 139, "lr": 8.191456807325477e-05} +{"train_loss": 0.016523156315088272, "grad_norm": 0.06622427701950073, "global_step": 14821, "epoch": 139, "lr": 8.191228652483047e-05} +{"train_loss": 0.01815876178443432, "grad_norm": 0.1398809403181076, "global_step": 14822, "epoch": 139, "lr": 8.191000486428026e-05} +{"train_loss": 0.01670297235250473, "grad_norm": 0.08009644597768784, "global_step": 14823, "epoch": 139, "lr": 8.190772309161218e-05} +{"train_loss": 0.01622694730758667, "grad_norm": 0.0867735892534256, "global_step": 14824, "epoch": 139, "lr": 8.190544120683426e-05} +{"train_loss": 0.02044796757400036, "grad_norm": 0.06937889009714127, "global_step": 14825, "epoch": 139, "lr": 8.190315920995446e-05} +{"train_loss": 0.01727926731109619, "grad_norm": 0.06114296242594719, "global_step": 14826, "epoch": 139, "lr": 8.190087710098085e-05} +{"train_loss": 0.018854884430766106, "grad_norm": 0.07468131184577942, "global_step": 14827, "epoch": 139, "lr": 8.189859487992143e-05} +{"train_loss": 0.016368549317121506, "grad_norm": 0.11828519403934479, "global_step": 14828, "epoch": 139, "lr": 8.189631254678422e-05} +{"train_loss": 0.015150895342230797, "grad_norm": 0.10482233762741089, "global_step": 14829, "epoch": 139, "lr": 8.189403010157724e-05} +{"train_loss": 0.01976609416306019, "grad_norm": 0.07974236458539963, "global_step": 14830, "epoch": 139, "lr": 8.18917475443085e-05} +{"train_loss": 0.017191506922245026, "grad_norm": 0.06010320037603378, "global_step": 14831, "epoch": 139, "lr": 8.188946487498603e-05} +{"train_loss": 0.014987580478191376, "grad_norm": 0.09901382029056549, "global_step": 14832, "epoch": 139, "lr": 8.188718209361785e-05} +{"train_loss": 0.016559509560465813, "grad_norm": 0.0842488557100296, "global_step": 14833, "epoch": 139, "lr": 8.188489920021199e-05} +{"train_loss": 0.016496330499649048, "grad_norm": 0.08218029141426086, "global_step": 14834, "epoch": 139, "lr": 8.188261619477645e-05} +{"train_loss": 0.016259942203760147, "grad_norm": 0.056358154863119125, "global_step": 14835, "epoch": 139, "lr": 8.188033307731926e-05} +{"train_loss": 0.016118992120027542, "grad_norm": 0.04976813495159149, "global_step": 14836, "epoch": 139, "lr": 8.187804984784846e-05} +{"train_loss": 0.01981164701282978, "grad_norm": 0.0860893502831459, "global_step": 14837, "epoch": 139, "lr": 8.187576650637204e-05} +{"train_loss": 0.01671683043241501, "grad_norm": 0.07548143714666367, "global_step": 14838, "epoch": 139, "lr": 8.187348305289804e-05} +{"train_loss": 0.016778746874616394, "grad_norm": 0.08535473793745041, "global_step": 14839, "epoch": 139, "lr": 8.18711994874345e-05, "val_loss": 0.01918559893965721} +{"train_loss": 0.01572350412607193, "grad_norm": 0.07227670401334763, "global_step": 14840, "epoch": 140, "lr": 8.18689158099894e-05} +{"train_loss": 0.013640666380524635, "grad_norm": 0.057206545025110245, "global_step": 14841, "epoch": 140, "lr": 8.186663202057081e-05} +{"train_loss": 0.01751692220568657, "grad_norm": 0.15853361785411835, "global_step": 14842, "epoch": 140, "lr": 8.18643481191867e-05} +{"train_loss": 0.015795987099409103, "grad_norm": 0.0850156620144844, "global_step": 14843, "epoch": 140, "lr": 8.186206410584517e-05} +{"train_loss": 0.01632682979106903, "grad_norm": 0.057166021317243576, "global_step": 14844, "epoch": 140, "lr": 8.185977998055419e-05} +{"train_loss": 0.01579425297677517, "grad_norm": 0.05077652260661125, "global_step": 14845, "epoch": 140, "lr": 8.18574957433218e-05} +{"train_loss": 0.013724265620112419, "grad_norm": 0.12822358310222626, "global_step": 14846, "epoch": 140, "lr": 8.1855211394156e-05} +{"train_loss": 0.023150380700826645, "grad_norm": 0.06397003680467606, "global_step": 14847, "epoch": 140, "lr": 8.185292693306487e-05} +{"train_loss": 0.019821692258119583, "grad_norm": 0.1626724898815155, "global_step": 14848, "epoch": 140, "lr": 8.185064236005639e-05} +{"train_loss": 0.017286423593759537, "grad_norm": 0.06993280351161957, "global_step": 14849, "epoch": 140, "lr": 8.18483576751386e-05} +{"train_loss": 0.016731835901737213, "grad_norm": 0.05694231018424034, "global_step": 14850, "epoch": 140, "lr": 8.184607287831955e-05} +{"train_loss": 0.01976625993847847, "grad_norm": 0.0859479010105133, "global_step": 14851, "epoch": 140, "lr": 8.184378796960726e-05} +{"train_loss": 0.015809476375579834, "grad_norm": 0.09859886020421982, "global_step": 14852, "epoch": 140, "lr": 8.184150294900971e-05} +{"train_loss": 0.016138620674610138, "grad_norm": 0.057949211448431015, "global_step": 14853, "epoch": 140, "lr": 8.183921781653499e-05} +{"train_loss": 0.014169663190841675, "grad_norm": 0.06778176873922348, "global_step": 14854, "epoch": 140, "lr": 8.183693257219111e-05} +{"train_loss": 0.016505977138876915, "grad_norm": 0.050550028681755066, "global_step": 14855, "epoch": 140, "lr": 8.183464721598608e-05} +{"train_loss": 0.017572538927197456, "grad_norm": 0.0803925096988678, "global_step": 14856, "epoch": 140, "lr": 8.183236174792794e-05} +{"train_loss": 0.017845463007688522, "grad_norm": 0.17830383777618408, "global_step": 14857, "epoch": 140, "lr": 8.183007616802474e-05} +{"train_loss": 0.0193779319524765, "grad_norm": 0.06057681143283844, "global_step": 14858, "epoch": 140, "lr": 8.182779047628449e-05} +{"train_loss": 0.01814866065979004, "grad_norm": 0.0599852055311203, "global_step": 14859, "epoch": 140, "lr": 8.182550467271521e-05} +{"train_loss": 0.016612038016319275, "grad_norm": 0.05802037939429283, "global_step": 14860, "epoch": 140, "lr": 8.182321875732498e-05} +{"train_loss": 0.013798386789858341, "grad_norm": 0.054204218089580536, "global_step": 14861, "epoch": 140, "lr": 8.182093273012176e-05} +{"train_loss": 0.016384733840823174, "grad_norm": 0.0958491861820221, "global_step": 14862, "epoch": 140, "lr": 8.181864659111364e-05} +{"train_loss": 0.014977019280195236, "grad_norm": 0.1036088615655899, "global_step": 14863, "epoch": 140, "lr": 8.181636034030865e-05} +{"train_loss": 0.01768200658261776, "grad_norm": 0.07081130892038345, "global_step": 14864, "epoch": 140, "lr": 8.18140739777148e-05} +{"train_loss": 0.016412321478128433, "grad_norm": 0.09768261015415192, "global_step": 14865, "epoch": 140, "lr": 8.18117875033401e-05} +{"train_loss": 0.020017102360725403, "grad_norm": 0.09464716911315918, "global_step": 14866, "epoch": 140, "lr": 8.180950091719266e-05} +{"train_loss": 0.016126461327075958, "grad_norm": 0.08306054025888443, "global_step": 14867, "epoch": 140, "lr": 8.180721421928043e-05} +{"train_loss": 0.019352881237864494, "grad_norm": 0.054577309638261795, "global_step": 14868, "epoch": 140, "lr": 8.180492740961149e-05} +{"train_loss": 0.017438586801290512, "grad_norm": 0.0651753842830658, "global_step": 14869, "epoch": 140, "lr": 8.180264048819387e-05} +{"train_loss": 0.015622243285179138, "grad_norm": 0.07465498149394989, "global_step": 14870, "epoch": 140, "lr": 8.180035345503561e-05} +{"train_loss": 0.017389915883541107, "grad_norm": 0.05778102949261665, "global_step": 14871, "epoch": 140, "lr": 8.179806631014472e-05} +{"train_loss": 0.015197298489511013, "grad_norm": 0.10331286489963531, "global_step": 14872, "epoch": 140, "lr": 8.179577905352926e-05} +{"train_loss": 0.019459791481494904, "grad_norm": 0.10152961313724518, "global_step": 14873, "epoch": 140, "lr": 8.179349168519727e-05} +{"train_loss": 0.017458457499742508, "grad_norm": 0.0478624664247036, "global_step": 14874, "epoch": 140, "lr": 8.179120420515676e-05} +{"train_loss": 0.015512961894273758, "grad_norm": 0.06429398059844971, "global_step": 14875, "epoch": 140, "lr": 8.17889166134158e-05} +{"train_loss": 0.01682228222489357, "grad_norm": 0.08232782036066055, "global_step": 14876, "epoch": 140, "lr": 8.178662890998241e-05} +{"train_loss": 0.016902150586247444, "grad_norm": 0.11559198796749115, "global_step": 14877, "epoch": 140, "lr": 8.178434109486461e-05} +{"train_loss": 0.016472360119223595, "grad_norm": 0.0752299502491951, "global_step": 14878, "epoch": 140, "lr": 8.178205316807049e-05} +{"train_loss": 0.01603466272354126, "grad_norm": 0.08950913697481155, "global_step": 14879, "epoch": 140, "lr": 8.177976512960803e-05} +{"train_loss": 0.015922490507364273, "grad_norm": 0.0494704395532608, "global_step": 14880, "epoch": 140, "lr": 8.17774769794853e-05} +{"train_loss": 0.016588954254984856, "grad_norm": 0.06291060149669647, "global_step": 14881, "epoch": 140, "lr": 8.177518871771035e-05} +{"train_loss": 0.016033736988902092, "grad_norm": 0.06363165378570557, "global_step": 14882, "epoch": 140, "lr": 8.177290034429119e-05} +{"train_loss": 0.015718838199973106, "grad_norm": 0.17348028719425201, "global_step": 14883, "epoch": 140, "lr": 8.17706118592359e-05} +{"train_loss": 0.01664598658680916, "grad_norm": 0.15654794871807098, "global_step": 14884, "epoch": 140, "lr": 8.176832326255248e-05} +{"train_loss": 0.01896580494940281, "grad_norm": 0.1083502247929573, "global_step": 14885, "epoch": 140, "lr": 8.176603455424898e-05} +{"train_loss": 0.01624748855829239, "grad_norm": 0.1323866844177246, "global_step": 14886, "epoch": 140, "lr": 8.176374573433345e-05} +{"train_loss": 0.019237587228417397, "grad_norm": 0.06420278549194336, "global_step": 14887, "epoch": 140, "lr": 8.176145680281392e-05} +{"train_loss": 0.014178193174302578, "grad_norm": 0.04803037643432617, "global_step": 14888, "epoch": 140, "lr": 8.175916775969847e-05} +{"train_loss": 0.01956518553197384, "grad_norm": 0.05053635314106941, "global_step": 14889, "epoch": 140, "lr": 8.175687860499511e-05} +{"train_loss": 0.015913071110844612, "grad_norm": 0.05832921341061592, "global_step": 14890, "epoch": 140, "lr": 8.17545893387119e-05} +{"train_loss": 0.015767591074109077, "grad_norm": 0.08860135078430176, "global_step": 14891, "epoch": 140, "lr": 8.175229996085685e-05} +{"train_loss": 0.016972769051790237, "grad_norm": 0.05691482499241829, "global_step": 14892, "epoch": 140, "lr": 8.175001047143804e-05} +{"train_loss": 0.017432983964681625, "grad_norm": 0.05123168230056763, "global_step": 14893, "epoch": 140, "lr": 8.174772087046349e-05} +{"train_loss": 0.015918850898742676, "grad_norm": 0.14855042099952698, "global_step": 14894, "epoch": 140, "lr": 8.174543115794127e-05} +{"train_loss": 0.01979319378733635, "grad_norm": 0.07308509945869446, "global_step": 14895, "epoch": 140, "lr": 8.174314133387941e-05} +{"train_loss": 0.017915917560458183, "grad_norm": 0.12085834890604019, "global_step": 14896, "epoch": 140, "lr": 8.174085139828594e-05} +{"train_loss": 0.014806803315877914, "grad_norm": 0.061316538602113724, "global_step": 14897, "epoch": 140, "lr": 8.173856135116893e-05} +{"train_loss": 0.015603726729750633, "grad_norm": 0.04781056195497513, "global_step": 14898, "epoch": 140, "lr": 8.173627119253642e-05} +{"train_loss": 0.019168926402926445, "grad_norm": 0.12900424003601074, "global_step": 14899, "epoch": 140, "lr": 8.173398092239646e-05} +{"train_loss": 0.019452866166830063, "grad_norm": 0.06800569593906403, "global_step": 14900, "epoch": 140, "lr": 8.173169054075708e-05} +{"train_loss": 0.014113922603428364, "grad_norm": 0.06970231235027313, "global_step": 14901, "epoch": 140, "lr": 8.172940004762635e-05} +{"train_loss": 0.017745807766914368, "grad_norm": 0.11492163687944412, "global_step": 14902, "epoch": 140, "lr": 8.17271094430123e-05} +{"train_loss": 0.015372470021247864, "grad_norm": 0.07145125418901443, "global_step": 14903, "epoch": 140, "lr": 8.172481872692297e-05} +{"train_loss": 0.017854951322078705, "grad_norm": 0.07140188664197922, "global_step": 14904, "epoch": 140, "lr": 8.172252789936645e-05} +{"train_loss": 0.013972611166536808, "grad_norm": 0.08362241089344025, "global_step": 14905, "epoch": 140, "lr": 8.172023696035077e-05} +{"train_loss": 0.014973370358347893, "grad_norm": 0.07784946262836456, "global_step": 14906, "epoch": 140, "lr": 8.171794590988394e-05} +{"train_loss": 0.015427540987730026, "grad_norm": 0.05098900943994522, "global_step": 14907, "epoch": 140, "lr": 8.171565474797407e-05} +{"train_loss": 0.014728691428899765, "grad_norm": 0.04766624793410301, "global_step": 14908, "epoch": 140, "lr": 8.171336347462916e-05} +{"train_loss": 0.021631240844726562, "grad_norm": 0.08197588473558426, "global_step": 14909, "epoch": 140, "lr": 8.171107208985728e-05} +{"train_loss": 0.015555155463516712, "grad_norm": 0.1351064294576645, "global_step": 14910, "epoch": 140, "lr": 8.170878059366651e-05} +{"train_loss": 0.01879783719778061, "grad_norm": 0.12251771241426468, "global_step": 14911, "epoch": 140, "lr": 8.170648898606485e-05} +{"train_loss": 0.017868001013994217, "grad_norm": 0.051629867404699326, "global_step": 14912, "epoch": 140, "lr": 8.17041972670604e-05} +{"train_loss": 0.018166184425354004, "grad_norm": 0.10702018439769745, "global_step": 14913, "epoch": 140, "lr": 8.170190543666116e-05} +{"train_loss": 0.01767362840473652, "grad_norm": 0.12582828104496002, "global_step": 14914, "epoch": 140, "lr": 8.169961349487523e-05} +{"train_loss": 0.01791463792324066, "grad_norm": 0.0966734066605568, "global_step": 14915, "epoch": 140, "lr": 8.169732144171065e-05} +{"train_loss": 0.015985732898116112, "grad_norm": 0.09444377571344376, "global_step": 14916, "epoch": 140, "lr": 8.169502927717545e-05} +{"train_loss": 0.01609131321310997, "grad_norm": 0.13607388734817505, "global_step": 14917, "epoch": 140, "lr": 8.16927370012777e-05} +{"train_loss": 0.016734251752495766, "grad_norm": 0.1317705363035202, "global_step": 14918, "epoch": 140, "lr": 8.169044461402546e-05} +{"train_loss": 0.018953509628772736, "grad_norm": 0.07519065588712692, "global_step": 14919, "epoch": 140, "lr": 8.16881521154268e-05} +{"train_loss": 0.01583021879196167, "grad_norm": 0.07467316836118698, "global_step": 14920, "epoch": 140, "lr": 8.168585950548973e-05} +{"train_loss": 0.015960609540343285, "grad_norm": 0.05219089239835739, "global_step": 14921, "epoch": 140, "lr": 8.168356678422234e-05} +{"train_loss": 0.01552346907556057, "grad_norm": 0.08688453584909439, "global_step": 14922, "epoch": 140, "lr": 8.168127395163268e-05} +{"train_loss": 0.01505257748067379, "grad_norm": 0.10034405440092087, "global_step": 14923, "epoch": 140, "lr": 8.16789810077288e-05} +{"train_loss": 0.01644960045814514, "grad_norm": 0.1810741275548935, "global_step": 14924, "epoch": 140, "lr": 8.167668795251873e-05} +{"train_loss": 0.018357861787080765, "grad_norm": 0.04789576306939125, "global_step": 14925, "epoch": 140, "lr": 8.167439478601058e-05} +{"train_loss": 0.01733945682644844, "grad_norm": 0.04509551823139191, "global_step": 14926, "epoch": 140, "lr": 8.167210150821237e-05} +{"train_loss": 0.01437339186668396, "grad_norm": 0.06180299073457718, "global_step": 14927, "epoch": 140, "lr": 8.166980811913218e-05} +{"train_loss": 0.015664970502257347, "grad_norm": 0.0907071977853775, "global_step": 14928, "epoch": 140, "lr": 8.166751461877807e-05} +{"train_loss": 0.016215544193983078, "grad_norm": 0.13661403954029083, "global_step": 14929, "epoch": 140, "lr": 8.166522100715806e-05} +{"train_loss": 0.018046751618385315, "grad_norm": 0.07087677717208862, "global_step": 14930, "epoch": 140, "lr": 8.166292728428024e-05} +{"train_loss": 0.015759767964482307, "grad_norm": 0.05782247707247734, "global_step": 14931, "epoch": 140, "lr": 8.166063345015267e-05} +{"train_loss": 0.014899937435984612, "grad_norm": 0.060152098536491394, "global_step": 14932, "epoch": 140, "lr": 8.165833950478339e-05} +{"train_loss": 0.015283734537661076, "grad_norm": 0.06573841720819473, "global_step": 14933, "epoch": 140, "lr": 8.165604544818049e-05} +{"train_loss": 0.015981003642082214, "grad_norm": 0.0672244280576706, "global_step": 14934, "epoch": 140, "lr": 8.1653751280352e-05} +{"train_loss": 0.01621343567967415, "grad_norm": 0.07886891067028046, "global_step": 14935, "epoch": 140, "lr": 8.1651457001306e-05} +{"train_loss": 0.016354849562048912, "grad_norm": 0.11171625554561615, "global_step": 14936, "epoch": 140, "lr": 8.164916261105054e-05} +{"train_loss": 0.01882762648165226, "grad_norm": 0.11949866265058517, "global_step": 14937, "epoch": 140, "lr": 8.164686810959369e-05} +{"train_loss": 0.01567557454109192, "grad_norm": 0.07514812797307968, "global_step": 14938, "epoch": 140, "lr": 8.16445734969435e-05} +{"train_loss": 0.017665892839431763, "grad_norm": 0.1389264464378357, "global_step": 14939, "epoch": 140, "lr": 8.164227877310806e-05} +{"train_loss": 0.01887400820851326, "grad_norm": 0.12428750842809677, "global_step": 14940, "epoch": 140, "lr": 8.16399839380954e-05} +{"train_loss": 0.01356570329517126, "grad_norm": 0.10972695797681808, "global_step": 14941, "epoch": 140, "lr": 8.16376889919136e-05} +{"train_loss": 0.01475745439529419, "grad_norm": 0.13961294293403625, "global_step": 14942, "epoch": 140, "lr": 8.163539393457071e-05} +{"train_loss": 0.016977181658148766, "grad_norm": 0.0931946262717247, "global_step": 14943, "epoch": 140, "lr": 8.163309876607481e-05} +{"train_loss": 0.015573694370687008, "grad_norm": 0.07916519790887833, "global_step": 14944, "epoch": 140, "lr": 8.163080348643397e-05} +{"train_loss": 0.016774338292274274, "grad_norm": 0.09830079227685928, "global_step": 14945, "epoch": 140, "lr": 8.162850809565623e-05, "train_batch_mse_error": 0.014913965947926044, "val_loss": 0.0185556598007679} +{"train_loss": 0.015436303801834583, "grad_norm": 0.061332523822784424, "global_step": 14946, "epoch": 141, "lr": 8.162621259374967e-05} +{"train_loss": 0.01623392477631569, "grad_norm": 0.11403922736644745, "global_step": 14947, "epoch": 141, "lr": 8.162391698072235e-05} +{"train_loss": 0.016026947647333145, "grad_norm": 0.0974748358130455, "global_step": 14948, "epoch": 141, "lr": 8.162162125658235e-05} +{"train_loss": 0.014796475879848003, "grad_norm": 0.11545376479625702, "global_step": 14949, "epoch": 141, "lr": 8.16193254213377e-05} +{"train_loss": 0.01731863245368004, "grad_norm": 0.16203272342681885, "global_step": 14950, "epoch": 141, "lr": 8.161702947499651e-05} +{"train_loss": 0.01685251295566559, "grad_norm": 0.12946969270706177, "global_step": 14951, "epoch": 141, "lr": 8.161473341756684e-05} +{"train_loss": 0.018129494041204453, "grad_norm": 0.08096577227115631, "global_step": 14952, "epoch": 141, "lr": 8.161243724905672e-05} +{"train_loss": 0.014101386070251465, "grad_norm": 0.06961144506931305, "global_step": 14953, "epoch": 141, "lr": 8.161014096947425e-05} +{"train_loss": 0.016087427735328674, "grad_norm": 0.061971161514520645, "global_step": 14954, "epoch": 141, "lr": 8.160784457882751e-05} +{"train_loss": 0.017891235649585724, "grad_norm": 0.07601875066757202, "global_step": 14955, "epoch": 141, "lr": 8.160554807712453e-05} +{"train_loss": 0.01593654416501522, "grad_norm": 0.048855010420084, "global_step": 14956, "epoch": 141, "lr": 8.16032514643734e-05} +{"train_loss": 0.017647117376327515, "grad_norm": 0.07000292092561722, "global_step": 14957, "epoch": 141, "lr": 8.16009547405822e-05} +{"train_loss": 0.017254697158932686, "grad_norm": 0.05219476670026779, "global_step": 14958, "epoch": 141, "lr": 8.159865790575898e-05} +{"train_loss": 0.019340697675943375, "grad_norm": 0.11279020458459854, "global_step": 14959, "epoch": 141, "lr": 8.15963609599118e-05} +{"train_loss": 0.017353877425193787, "grad_norm": 0.12195377051830292, "global_step": 14960, "epoch": 141, "lr": 8.159406390304878e-05} +{"train_loss": 0.0168243870139122, "grad_norm": 0.06157473474740982, "global_step": 14961, "epoch": 141, "lr": 8.159176673517794e-05} +{"train_loss": 0.015921102836728096, "grad_norm": 0.0784299299120903, "global_step": 14962, "epoch": 141, "lr": 8.158946945630739e-05} +{"train_loss": 0.020340023562312126, "grad_norm": 0.055465515702962875, "global_step": 14963, "epoch": 141, "lr": 8.158717206644517e-05} +{"train_loss": 0.016193188726902008, "grad_norm": 0.1107708141207695, "global_step": 14964, "epoch": 141, "lr": 8.158487456559935e-05} +{"train_loss": 0.016099225729703903, "grad_norm": 0.11056345701217651, "global_step": 14965, "epoch": 141, "lr": 8.158257695377804e-05} +{"train_loss": 0.018357019871473312, "grad_norm": 0.08737613260746002, "global_step": 14966, "epoch": 141, "lr": 8.158027923098927e-05} +{"train_loss": 0.018112054094672203, "grad_norm": 0.08497241884469986, "global_step": 14967, "epoch": 141, "lr": 8.157798139724115e-05} +{"train_loss": 0.016337377950549126, "grad_norm": 0.06202589347958565, "global_step": 14968, "epoch": 141, "lr": 8.157568345254173e-05} +{"train_loss": 0.01664106920361519, "grad_norm": 0.09758491069078445, "global_step": 14969, "epoch": 141, "lr": 8.15733853968991e-05} +{"train_loss": 0.019907552748918533, "grad_norm": 0.08064345270395279, "global_step": 14970, "epoch": 141, "lr": 8.15710872303213e-05} +{"train_loss": 0.015599241480231285, "grad_norm": 0.06658162921667099, "global_step": 14971, "epoch": 141, "lr": 8.156878895281645e-05} +{"train_loss": 0.016857586801052094, "grad_norm": 0.05266896262764931, "global_step": 14972, "epoch": 141, "lr": 8.15664905643926e-05} +{"train_loss": 0.017547566443681717, "grad_norm": 0.05394173040986061, "global_step": 14973, "epoch": 141, "lr": 8.156419206505783e-05} +{"train_loss": 0.016610383987426758, "grad_norm": 0.06067453324794769, "global_step": 14974, "epoch": 141, "lr": 8.15618934548202e-05} +{"train_loss": 0.013891344889998436, "grad_norm": 0.10679522901773453, "global_step": 14975, "epoch": 141, "lr": 8.155959473368783e-05} +{"train_loss": 0.01635599508881569, "grad_norm": 0.09700589627027512, "global_step": 14976, "epoch": 141, "lr": 8.155729590166875e-05} +{"train_loss": 0.01683715358376503, "grad_norm": 0.1006753146648407, "global_step": 14977, "epoch": 141, "lr": 8.155499695877104e-05} +{"train_loss": 0.01649424061179161, "grad_norm": 0.05063745751976967, "global_step": 14978, "epoch": 141, "lr": 8.155269790500284e-05} +{"train_loss": 0.01568763703107834, "grad_norm": 0.10804322361946106, "global_step": 14979, "epoch": 141, "lr": 8.155039874037216e-05} +{"train_loss": 0.01635483279824257, "grad_norm": 0.08527225255966187, "global_step": 14980, "epoch": 141, "lr": 8.154809946488709e-05} +{"train_loss": 0.016567161306738853, "grad_norm": 0.06515473127365112, "global_step": 14981, "epoch": 141, "lr": 8.154580007855573e-05} +{"train_loss": 0.014546087011694908, "grad_norm": 0.06600837409496307, "global_step": 14982, "epoch": 141, "lr": 8.154350058138613e-05} +{"train_loss": 0.016122499480843544, "grad_norm": 0.11225879937410355, "global_step": 14983, "epoch": 141, "lr": 8.15412009733864e-05} +{"train_loss": 0.016879379749298096, "grad_norm": 0.054221075028181076, "global_step": 14984, "epoch": 141, "lr": 8.153890125456461e-05} +{"train_loss": 0.014355028048157692, "grad_norm": 0.059862397611141205, "global_step": 14985, "epoch": 141, "lr": 8.153660142492882e-05} +{"train_loss": 0.020376496016979218, "grad_norm": 0.05321774631738663, "global_step": 14986, "epoch": 141, "lr": 8.153430148448715e-05} +{"train_loss": 0.01452039834111929, "grad_norm": 0.08862655609846115, "global_step": 14987, "epoch": 141, "lr": 8.153200143324764e-05} +{"train_loss": 0.015621041879057884, "grad_norm": 0.05372617766261101, "global_step": 14988, "epoch": 141, "lr": 8.15297012712184e-05} +{"train_loss": 0.015569724142551422, "grad_norm": 0.045908402651548386, "global_step": 14989, "epoch": 141, "lr": 8.15274009984075e-05} +{"train_loss": 0.013045895844697952, "grad_norm": 0.05546799302101135, "global_step": 14990, "epoch": 141, "lr": 8.152510061482302e-05} +{"train_loss": 0.013481694273650646, "grad_norm": 0.05652007460594177, "global_step": 14991, "epoch": 141, "lr": 8.152280012047306e-05} +{"train_loss": 0.015195240266621113, "grad_norm": 0.08483302593231201, "global_step": 14992, "epoch": 141, "lr": 8.152049951536567e-05} +{"train_loss": 0.01586238667368889, "grad_norm": 0.04423172026872635, "global_step": 14993, "epoch": 141, "lr": 8.151819879950897e-05} +{"train_loss": 0.01788114756345749, "grad_norm": 0.06803125888109207, "global_step": 14994, "epoch": 141, "lr": 8.151589797291102e-05} +{"train_loss": 0.014886055141687393, "grad_norm": 0.06657545268535614, "global_step": 14995, "epoch": 141, "lr": 8.15135970355799e-05} +{"train_loss": 0.017200160771608353, "grad_norm": 0.07778991758823395, "global_step": 14996, "epoch": 141, "lr": 8.151129598752372e-05} +{"train_loss": 0.02004244551062584, "grad_norm": 0.05764880031347275, "global_step": 14997, "epoch": 141, "lr": 8.150899482875054e-05} +{"train_loss": 0.01913958042860031, "grad_norm": 0.11625391989946365, "global_step": 14998, "epoch": 141, "lr": 8.150669355926846e-05} +{"train_loss": 0.017979346215724945, "grad_norm": 0.07041242718696594, "global_step": 14999, "epoch": 141, "lr": 8.150439217908556e-05} +{"train_loss": 0.018019188195466995, "grad_norm": 0.044107452034950256, "global_step": 15000, "epoch": 141, "lr": 8.150209068820992e-05} +{"train_loss": 0.014657474122941494, "grad_norm": 0.08726687729358673, "global_step": 15001, "epoch": 141, "lr": 8.149978908664965e-05} +{"train_loss": 0.016127075999975204, "grad_norm": 0.05577909201383591, "global_step": 15002, "epoch": 141, "lr": 8.149748737441281e-05} +{"train_loss": 0.015288792550563812, "grad_norm": 0.06442555040121078, "global_step": 15003, "epoch": 141, "lr": 8.149518555150749e-05} +{"train_loss": 0.014410417526960373, "grad_norm": 0.05458812415599823, "global_step": 15004, "epoch": 141, "lr": 8.14928836179418e-05} +{"train_loss": 0.018717389553785324, "grad_norm": 0.1346433013677597, "global_step": 15005, "epoch": 141, "lr": 8.14905815737238e-05} +{"train_loss": 0.018942074850201607, "grad_norm": 0.12649215757846832, "global_step": 15006, "epoch": 141, "lr": 8.148827941886159e-05} +{"train_loss": 0.017802421003580093, "grad_norm": 0.06342149525880814, "global_step": 15007, "epoch": 141, "lr": 8.148597715336327e-05} +{"train_loss": 0.017264578491449356, "grad_norm": 0.05732148885726929, "global_step": 15008, "epoch": 141, "lr": 8.148367477723691e-05} +{"train_loss": 0.018515173345804214, "grad_norm": 0.05634918808937073, "global_step": 15009, "epoch": 141, "lr": 8.14813722904906e-05} +{"train_loss": 0.017054948955774307, "grad_norm": 0.1269475817680359, "global_step": 15010, "epoch": 141, "lr": 8.147906969313247e-05} +{"train_loss": 0.01857295259833336, "grad_norm": 0.06512322276830673, "global_step": 15011, "epoch": 141, "lr": 8.147676698517056e-05} +{"train_loss": 0.017079712823033333, "grad_norm": 0.07725299894809723, "global_step": 15012, "epoch": 141, "lr": 8.147446416661297e-05} +{"train_loss": 0.01744060404598713, "grad_norm": 0.10082615166902542, "global_step": 15013, "epoch": 141, "lr": 8.147216123746782e-05} +{"train_loss": 0.015938829630613327, "grad_norm": 0.059325024485588074, "global_step": 15014, "epoch": 141, "lr": 8.146985819774318e-05} +{"train_loss": 0.0177355594933033, "grad_norm": 0.07408653199672699, "global_step": 15015, "epoch": 141, "lr": 8.146755504744713e-05} +{"train_loss": 0.018624495714902878, "grad_norm": 0.05320921912789345, "global_step": 15016, "epoch": 141, "lr": 8.146525178658779e-05} +{"train_loss": 0.015938565135002136, "grad_norm": 0.06806284189224243, "global_step": 15017, "epoch": 141, "lr": 8.146294841517323e-05} +{"train_loss": 0.017522744834423065, "grad_norm": 0.08091051131486893, "global_step": 15018, "epoch": 141, "lr": 8.146064493321155e-05} +{"train_loss": 0.016514737159013748, "grad_norm": 0.06641083210706711, "global_step": 15019, "epoch": 141, "lr": 8.145834134071086e-05} +{"train_loss": 0.01965993642807007, "grad_norm": 0.15212829411029816, "global_step": 15020, "epoch": 141, "lr": 8.145603763767923e-05} +{"train_loss": 0.01803215965628624, "grad_norm": 0.09656977653503418, "global_step": 15021, "epoch": 141, "lr": 8.145373382412475e-05} +{"train_loss": 0.01954932138323784, "grad_norm": 0.08657094836235046, "global_step": 15022, "epoch": 141, "lr": 8.145142990005555e-05} +{"train_loss": 0.015989867970347404, "grad_norm": 0.06239344924688339, "global_step": 15023, "epoch": 141, "lr": 8.14491258654797e-05} +{"train_loss": 0.014871545135974884, "grad_norm": 0.07422356307506561, "global_step": 15024, "epoch": 141, "lr": 8.14468217204053e-05} +{"train_loss": 0.017682712525129318, "grad_norm": 0.08746428787708282, "global_step": 15025, "epoch": 141, "lr": 8.144451746484043e-05} +{"train_loss": 0.017320968210697174, "grad_norm": 0.07923558354377747, "global_step": 15026, "epoch": 141, "lr": 8.14422130987932e-05} +{"train_loss": 0.01566280424594879, "grad_norm": 0.08130650222301483, "global_step": 15027, "epoch": 141, "lr": 8.143990862227172e-05} +{"train_loss": 0.016035055741667747, "grad_norm": 0.09882114082574844, "global_step": 15028, "epoch": 141, "lr": 8.143760403528407e-05} +{"train_loss": 0.015532422810792923, "grad_norm": 0.09206975996494293, "global_step": 15029, "epoch": 141, "lr": 8.143529933783835e-05} +{"train_loss": 0.016346190124750137, "grad_norm": 0.0510411262512207, "global_step": 15030, "epoch": 141, "lr": 8.143299452994267e-05} +{"train_loss": 0.013948895037174225, "grad_norm": 0.04979545250535011, "global_step": 15031, "epoch": 141, "lr": 8.143068961160511e-05} +{"train_loss": 0.015371047891676426, "grad_norm": 0.04505280777812004, "global_step": 15032, "epoch": 141, "lr": 8.142838458283377e-05} +{"train_loss": 0.016740012913942337, "grad_norm": 0.11621562391519547, "global_step": 15033, "epoch": 141, "lr": 8.142607944363674e-05} +{"train_loss": 0.01682267151772976, "grad_norm": 0.15158198773860931, "global_step": 15034, "epoch": 141, "lr": 8.142377419402216e-05} +{"train_loss": 0.015328681096434593, "grad_norm": 0.05480217561125755, "global_step": 15035, "epoch": 141, "lr": 8.142146883399809e-05} +{"train_loss": 0.022029342129826546, "grad_norm": 0.23830503225326538, "global_step": 15036, "epoch": 141, "lr": 8.141916336357264e-05} +{"train_loss": 0.020288314670324326, "grad_norm": 0.08396004140377045, "global_step": 15037, "epoch": 141, "lr": 8.141685778275395e-05} +{"train_loss": 0.017407789826393127, "grad_norm": 0.07834375649690628, "global_step": 15038, "epoch": 141, "lr": 8.141455209155004e-05} +{"train_loss": 0.017300046980381012, "grad_norm": 0.047263093292713165, "global_step": 15039, "epoch": 141, "lr": 8.141224628996906e-05} +{"train_loss": 0.019259244203567505, "grad_norm": 0.07245655357837677, "global_step": 15040, "epoch": 141, "lr": 8.140994037801914e-05} +{"train_loss": 0.015410319902002811, "grad_norm": 0.05294150114059448, "global_step": 15041, "epoch": 141, "lr": 8.140763435570832e-05} +{"train_loss": 0.014760961756110191, "grad_norm": 0.08929397165775299, "global_step": 15042, "epoch": 141, "lr": 8.140532822304474e-05} +{"train_loss": 0.01467540767043829, "grad_norm": 0.05657845735549927, "global_step": 15043, "epoch": 141, "lr": 8.14030219800365e-05} +{"train_loss": 0.017518281936645508, "grad_norm": 0.0630461648106575, "global_step": 15044, "epoch": 141, "lr": 8.14007156266917e-05} +{"train_loss": 0.0176813006401062, "grad_norm": 0.05457992106676102, "global_step": 15045, "epoch": 141, "lr": 8.139840916301842e-05} +{"train_loss": 0.016255296766757965, "grad_norm": 0.07657904177904129, "global_step": 15046, "epoch": 141, "lr": 8.139610258902479e-05} +{"train_loss": 0.015848547220230103, "grad_norm": 0.0681113675236702, "global_step": 15047, "epoch": 141, "lr": 8.139379590471892e-05} +{"train_loss": 0.01361058745533228, "grad_norm": 0.04336104169487953, "global_step": 15048, "epoch": 141, "lr": 8.139148911010891e-05} +{"train_loss": 0.016576392576098442, "grad_norm": 0.12988609075546265, "global_step": 15049, "epoch": 141, "lr": 8.138918220520285e-05} +{"train_loss": 0.01655983366072178, "grad_norm": 0.05593188479542732, "global_step": 15050, "epoch": 141, "lr": 8.138687519000885e-05} +{"train_loss": 0.016728858024163067, "grad_norm": 0.13543888926506042, "global_step": 15051, "epoch": 141, "lr": 8.138456806453503e-05, "val_loss": 0.01891852170228958} +{"train_loss": 0.017207905650138855, "grad_norm": 0.04699813947081566, "global_step": 15052, "epoch": 142, "lr": 8.138226082878948e-05} +{"train_loss": 0.016990192234516144, "grad_norm": 0.06274699419736862, "global_step": 15053, "epoch": 142, "lr": 8.137995348278031e-05} +{"train_loss": 0.015610510483384132, "grad_norm": 0.05198432132601738, "global_step": 15054, "epoch": 142, "lr": 8.137764602651564e-05} +{"train_loss": 0.01777026057243347, "grad_norm": 0.08582893013954163, "global_step": 15055, "epoch": 142, "lr": 8.137533846000357e-05} +{"train_loss": 0.016948943957686424, "grad_norm": 0.04967907816171646, "global_step": 15056, "epoch": 142, "lr": 8.137303078325219e-05} +{"train_loss": 0.014835149049758911, "grad_norm": 0.11225732415914536, "global_step": 15057, "epoch": 142, "lr": 8.137072299626964e-05} +{"train_loss": 0.01660929061472416, "grad_norm": 0.11182717233896255, "global_step": 15058, "epoch": 142, "lr": 8.1368415099064e-05} +{"train_loss": 0.016631506383419037, "grad_norm": 0.10506853461265564, "global_step": 15059, "epoch": 142, "lr": 8.13661070916434e-05} +{"train_loss": 0.016473308205604553, "grad_norm": 0.06768691539764404, "global_step": 15060, "epoch": 142, "lr": 8.136379897401594e-05} +{"train_loss": 0.017947975546121597, "grad_norm": 0.08343648165464401, "global_step": 15061, "epoch": 142, "lr": 8.136149074618972e-05} +{"train_loss": 0.015226908028125763, "grad_norm": 0.09986024349927902, "global_step": 15062, "epoch": 142, "lr": 8.135918240817287e-05} +{"train_loss": 0.02025812491774559, "grad_norm": 0.07431437075138092, "global_step": 15063, "epoch": 142, "lr": 8.13568739599735e-05} +{"train_loss": 0.01652097888290882, "grad_norm": 0.1263679713010788, "global_step": 15064, "epoch": 142, "lr": 8.135456540159968e-05} +{"train_loss": 0.015146690420806408, "grad_norm": 0.050070784986019135, "global_step": 15065, "epoch": 142, "lr": 8.135225673305958e-05} +{"train_loss": 0.015887577086687088, "grad_norm": 0.059150733053684235, "global_step": 15066, "epoch": 142, "lr": 8.134994795436127e-05} +{"train_loss": 0.01808904856443405, "grad_norm": 0.11954386532306671, "global_step": 15067, "epoch": 142, "lr": 8.134763906551288e-05} +{"train_loss": 0.01920345053076744, "grad_norm": 0.08317579329013824, "global_step": 15068, "epoch": 142, "lr": 8.134533006652252e-05} +{"train_loss": 0.014471745118498802, "grad_norm": 0.050083279609680176, "global_step": 15069, "epoch": 142, "lr": 8.13430209573983e-05} +{"train_loss": 0.01729457825422287, "grad_norm": 0.07216008007526398, "global_step": 15070, "epoch": 142, "lr": 8.134071173814835e-05} +{"train_loss": 0.01709996722638607, "grad_norm": 0.0657634437084198, "global_step": 15071, "epoch": 142, "lr": 8.133840240878075e-05} +{"train_loss": 0.01536645833402872, "grad_norm": 0.07874802500009537, "global_step": 15072, "epoch": 142, "lr": 8.133609296930365e-05} +{"train_loss": 0.020547110587358475, "grad_norm": 0.06295294314622879, "global_step": 15073, "epoch": 142, "lr": 8.133378341972513e-05} +{"train_loss": 0.017024973407387733, "grad_norm": 0.06037620082497597, "global_step": 15074, "epoch": 142, "lr": 8.133147376005332e-05} +{"train_loss": 0.01534724235534668, "grad_norm": 0.061761245131492615, "global_step": 15075, "epoch": 142, "lr": 8.132916399029635e-05} +{"train_loss": 0.018601316958665848, "grad_norm": 0.09291118383407593, "global_step": 15076, "epoch": 142, "lr": 8.132685411046233e-05} +{"train_loss": 0.01583816669881344, "grad_norm": 0.10090209543704987, "global_step": 15077, "epoch": 142, "lr": 8.132454412055934e-05} +{"train_loss": 0.017133187502622604, "grad_norm": 0.05518640950322151, "global_step": 15078, "epoch": 142, "lr": 8.132223402059555e-05} +{"train_loss": 0.02043512463569641, "grad_norm": 0.06332997232675552, "global_step": 15079, "epoch": 142, "lr": 8.131992381057904e-05} +{"train_loss": 0.016286170110106468, "grad_norm": 0.06957356631755829, "global_step": 15080, "epoch": 142, "lr": 8.131761349051794e-05} +{"train_loss": 0.01805325224995613, "grad_norm": 0.1481279730796814, "global_step": 15081, "epoch": 142, "lr": 8.131530306042036e-05} +{"train_loss": 0.015218844637274742, "grad_norm": 0.1045958548784256, "global_step": 15082, "epoch": 142, "lr": 8.131299252029444e-05} +{"train_loss": 0.015701234340667725, "grad_norm": 0.05617637559771538, "global_step": 15083, "epoch": 142, "lr": 8.131068187014828e-05} +{"train_loss": 0.016724254935979843, "grad_norm": 0.0707593634724617, "global_step": 15084, "epoch": 142, "lr": 8.130837110998999e-05} +{"train_loss": 0.01595912128686905, "grad_norm": 0.10579380393028259, "global_step": 15085, "epoch": 142, "lr": 8.13060602398277e-05} +{"train_loss": 0.012909214943647385, "grad_norm": 0.06445670872926712, "global_step": 15086, "epoch": 142, "lr": 8.130374925966953e-05} +{"train_loss": 0.01668161153793335, "grad_norm": 0.07745786011219025, "global_step": 15087, "epoch": 142, "lr": 8.13014381695236e-05} +{"train_loss": 0.018315788358449936, "grad_norm": 0.07924988120794296, "global_step": 15088, "epoch": 142, "lr": 8.129912696939803e-05} +{"train_loss": 0.014629293233156204, "grad_norm": 0.08489379286766052, "global_step": 15089, "epoch": 142, "lr": 8.129681565930094e-05} +{"train_loss": 0.01684867963194847, "grad_norm": 0.11125992983579636, "global_step": 15090, "epoch": 142, "lr": 8.129450423924045e-05} +{"train_loss": 0.01661929115653038, "grad_norm": 0.12051395326852798, "global_step": 15091, "epoch": 142, "lr": 8.129219270922468e-05} +{"train_loss": 0.01677822135388851, "grad_norm": 0.0836130902171135, "global_step": 15092, "epoch": 142, "lr": 8.128988106926174e-05} +{"train_loss": 0.01682557910680771, "grad_norm": 0.04963226616382599, "global_step": 15093, "epoch": 142, "lr": 8.128756931935979e-05} +{"train_loss": 0.017207365483045578, "grad_norm": 0.06639691442251205, "global_step": 15094, "epoch": 142, "lr": 8.128525745952692e-05} +{"train_loss": 0.017720460891723633, "grad_norm": 0.07775193452835083, "global_step": 15095, "epoch": 142, "lr": 8.128294548977127e-05} +{"train_loss": 0.017156818881630898, "grad_norm": 0.05423560366034508, "global_step": 15096, "epoch": 142, "lr": 8.128063341010093e-05} +{"train_loss": 0.01646680012345314, "grad_norm": 0.08258017897605896, "global_step": 15097, "epoch": 142, "lr": 8.127832122052406e-05} +{"train_loss": 0.017783239483833313, "grad_norm": 0.08955353498458862, "global_step": 15098, "epoch": 142, "lr": 8.127600892104878e-05} +{"train_loss": 0.021252654492855072, "grad_norm": 0.0921667069196701, "global_step": 15099, "epoch": 142, "lr": 8.12736965116832e-05} +{"train_loss": 0.01698610559105873, "grad_norm": 0.05991218239068985, "global_step": 15100, "epoch": 142, "lr": 8.127138399243545e-05} +{"train_loss": 0.014896491542458534, "grad_norm": 0.04625039920210838, "global_step": 15101, "epoch": 142, "lr": 8.126907136331365e-05} +{"train_loss": 0.018182283267378807, "grad_norm": 0.05391625314950943, "global_step": 15102, "epoch": 142, "lr": 8.126675862432593e-05} +{"train_loss": 0.017796289175748825, "grad_norm": 0.06177322939038277, "global_step": 15103, "epoch": 142, "lr": 8.126444577548043e-05} +{"train_loss": 0.016666097566485405, "grad_norm": 0.06118373945355415, "global_step": 15104, "epoch": 142, "lr": 8.126213281678528e-05} +{"train_loss": 0.014825789257884026, "grad_norm": 0.05036202818155289, "global_step": 15105, "epoch": 142, "lr": 8.125981974824856e-05} +{"train_loss": 0.017639733850955963, "grad_norm": 0.16514357924461365, "global_step": 15106, "epoch": 142, "lr": 8.125750656987842e-05} +{"train_loss": 0.014843454584479332, "grad_norm": 0.064350426197052, "global_step": 15107, "epoch": 142, "lr": 8.125519328168302e-05} +{"train_loss": 0.014245513826608658, "grad_norm": 0.06343646347522736, "global_step": 15108, "epoch": 142, "lr": 8.125287988367047e-05} +{"train_loss": 0.019650984555482864, "grad_norm": 0.07908746600151062, "global_step": 15109, "epoch": 142, "lr": 8.125056637584887e-05} +{"train_loss": 0.01607588492333889, "grad_norm": 0.1220092698931694, "global_step": 15110, "epoch": 142, "lr": 8.124825275822638e-05} +{"train_loss": 0.017833400517702103, "grad_norm": 0.049993257969617844, "global_step": 15111, "epoch": 142, "lr": 8.124593903081113e-05} +{"train_loss": 0.016084104776382446, "grad_norm": 0.05746736004948616, "global_step": 15112, "epoch": 142, "lr": 8.124362519361122e-05} +{"train_loss": 0.016628818586468697, "grad_norm": 0.06072280555963516, "global_step": 15113, "epoch": 142, "lr": 8.12413112466348e-05} +{"train_loss": 0.01693117618560791, "grad_norm": 0.08231091499328613, "global_step": 15114, "epoch": 142, "lr": 8.123899718989001e-05} +{"train_loss": 0.01913212426006794, "grad_norm": 0.06925208121538162, "global_step": 15115, "epoch": 142, "lr": 8.123668302338497e-05} +{"train_loss": 0.017239181324839592, "grad_norm": 0.07815539836883545, "global_step": 15116, "epoch": 142, "lr": 8.123436874712779e-05} +{"train_loss": 0.016302570700645447, "grad_norm": 0.05833660066127777, "global_step": 15117, "epoch": 142, "lr": 8.123205436112664e-05} +{"train_loss": 0.018037181347608566, "grad_norm": 0.05553535744547844, "global_step": 15118, "epoch": 142, "lr": 8.122973986538962e-05} +{"train_loss": 0.013727247714996338, "grad_norm": 0.06430909782648087, "global_step": 15119, "epoch": 142, "lr": 8.12274252599249e-05} +{"train_loss": 0.018278885632753372, "grad_norm": 0.08580721914768219, "global_step": 15120, "epoch": 142, "lr": 8.122511054474057e-05} +{"train_loss": 0.016607515513896942, "grad_norm": 0.09584525972604752, "global_step": 15121, "epoch": 142, "lr": 8.122279571984477e-05} +{"train_loss": 0.016979657113552094, "grad_norm": 0.13525015115737915, "global_step": 15122, "epoch": 142, "lr": 8.122048078524567e-05} +{"train_loss": 0.018400948494672775, "grad_norm": 0.11951081454753876, "global_step": 15123, "epoch": 142, "lr": 8.121816574095137e-05} +{"train_loss": 0.015852922573685646, "grad_norm": 0.07981730997562408, "global_step": 15124, "epoch": 142, "lr": 8.121585058697e-05} +{"train_loss": 0.012922451831400394, "grad_norm": 0.046984005719423294, "global_step": 15125, "epoch": 142, "lr": 8.121353532330972e-05} +{"train_loss": 0.015031544491648674, "grad_norm": 0.055554717779159546, "global_step": 15126, "epoch": 142, "lr": 8.121121994997865e-05} +{"train_loss": 0.015482185408473015, "grad_norm": 0.14503931999206543, "global_step": 15127, "epoch": 142, "lr": 8.120890446698491e-05} +{"train_loss": 0.016083193942904472, "grad_norm": 0.07085742056369781, "global_step": 15128, "epoch": 142, "lr": 8.120658887433667e-05} +{"train_loss": 0.01712670922279358, "grad_norm": 0.04805484414100647, "global_step": 15129, "epoch": 142, "lr": 8.120427317204204e-05} +{"train_loss": 0.016733873635530472, "grad_norm": 0.04714677482843399, "global_step": 15130, "epoch": 142, "lr": 8.120195736010915e-05} +{"train_loss": 0.012808175757527351, "grad_norm": 0.04038548842072487, "global_step": 15131, "epoch": 142, "lr": 8.119964143854619e-05} +{"train_loss": 0.016318589448928833, "grad_norm": 0.08389520645141602, "global_step": 15132, "epoch": 142, "lr": 8.119732540736123e-05} +{"train_loss": 0.015340256504714489, "grad_norm": 0.05024710297584534, "global_step": 15133, "epoch": 142, "lr": 8.119500926656244e-05} +{"train_loss": 0.014357704669237137, "grad_norm": 0.08849485963582993, "global_step": 15134, "epoch": 142, "lr": 8.119269301615796e-05} +{"train_loss": 0.013978412374854088, "grad_norm": 0.05185079947113991, "global_step": 15135, "epoch": 142, "lr": 8.119037665615592e-05} +{"train_loss": 0.016288964077830315, "grad_norm": 0.05240790173411369, "global_step": 15136, "epoch": 142, "lr": 8.118806018656447e-05} +{"train_loss": 0.015501304529607296, "grad_norm": 0.049651969224214554, "global_step": 15137, "epoch": 142, "lr": 8.118574360739172e-05} +{"train_loss": 0.01902303844690323, "grad_norm": 0.06854329258203506, "global_step": 15138, "epoch": 142, "lr": 8.118342691864583e-05} +{"train_loss": 0.018136264756321907, "grad_norm": 0.10955209285020828, "global_step": 15139, "epoch": 142, "lr": 8.118111012033496e-05} +{"train_loss": 0.017017973586916924, "grad_norm": 0.0934360921382904, "global_step": 15140, "epoch": 142, "lr": 8.117879321246721e-05} +{"train_loss": 0.01750471070408821, "grad_norm": 0.06321953237056732, "global_step": 15141, "epoch": 142, "lr": 8.117647619505075e-05} +{"train_loss": 0.015960901975631714, "grad_norm": 0.05805344879627228, "global_step": 15142, "epoch": 142, "lr": 8.11741590680937e-05} +{"train_loss": 0.015701672062277794, "grad_norm": 0.0772612988948822, "global_step": 15143, "epoch": 142, "lr": 8.117184183160422e-05} +{"train_loss": 0.019517865031957626, "grad_norm": 0.06453487277030945, "global_step": 15144, "epoch": 142, "lr": 8.116952448559044e-05} +{"train_loss": 0.01805865578353405, "grad_norm": 0.09148195385932922, "global_step": 15145, "epoch": 142, "lr": 8.116720703006051e-05} +{"train_loss": 0.013010580092668533, "grad_norm": 0.08499238640069962, "global_step": 15146, "epoch": 142, "lr": 8.116488946502259e-05} +{"train_loss": 0.013324359431862831, "grad_norm": 0.10801905393600464, "global_step": 15147, "epoch": 142, "lr": 8.116257179048476e-05} +{"train_loss": 0.018240388482809067, "grad_norm": 0.06285671889781952, "global_step": 15148, "epoch": 142, "lr": 8.116025400645524e-05} +{"train_loss": 0.018898211419582367, "grad_norm": 0.17860382795333862, "global_step": 15149, "epoch": 142, "lr": 8.115793611294211e-05} +{"train_loss": 0.016756892204284668, "grad_norm": 0.056022536009550095, "global_step": 15150, "epoch": 142, "lr": 8.115561810995355e-05} +{"train_loss": 0.018470559269189835, "grad_norm": 0.11766979098320007, "global_step": 15151, "epoch": 142, "lr": 8.115329999749772e-05} +{"train_loss": 0.018028926104307175, "grad_norm": 0.12981387972831726, "global_step": 15152, "epoch": 142, "lr": 8.115098177558273e-05} +{"train_loss": 0.015199152752757072, "grad_norm": 0.053653694689273834, "global_step": 15153, "epoch": 142, "lr": 8.114866344421672e-05} +{"train_loss": 0.019347861409187317, "grad_norm": 0.07142533361911774, "global_step": 15154, "epoch": 142, "lr": 8.114634500340788e-05} +{"train_loss": 0.016282398253679276, "grad_norm": 0.05778611823916435, "global_step": 15155, "epoch": 142, "lr": 8.11440264531643e-05} +{"train_loss": 0.015507861040532589, "grad_norm": 0.052205003798007965, "global_step": 15156, "epoch": 142, "lr": 8.114170779349417e-05} +{"train_loss": 0.01672477457883223, "grad_norm": 0.08741654455661774, "global_step": 15157, "epoch": 142, "lr": 8.113938902440564e-05, "val_loss": 0.0192252229899168} +{"train_loss": 0.016682999208569527, "grad_norm": 0.12502095103263855, "global_step": 15158, "epoch": 143, "lr": 8.113707014590682e-05} +{"train_loss": 0.01790931448340416, "grad_norm": 0.09433504939079285, "global_step": 15159, "epoch": 143, "lr": 8.113475115800586e-05} +{"train_loss": 0.017503414303064346, "grad_norm": 0.13326983153820038, "global_step": 15160, "epoch": 143, "lr": 8.113243206071095e-05} +{"train_loss": 0.01592719554901123, "grad_norm": 0.1337641179561615, "global_step": 15161, "epoch": 143, "lr": 8.11301128540302e-05} +{"train_loss": 0.01530234981328249, "grad_norm": 0.05248234421014786, "global_step": 15162, "epoch": 143, "lr": 8.112779353797178e-05} +{"train_loss": 0.014031359925866127, "grad_norm": 0.1401946097612381, "global_step": 15163, "epoch": 143, "lr": 8.112547411254383e-05} +{"train_loss": 0.017792370170354843, "grad_norm": 0.05655622109770775, "global_step": 15164, "epoch": 143, "lr": 8.11231545777545e-05} +{"train_loss": 0.018338114023208618, "grad_norm": 0.11782047897577286, "global_step": 15165, "epoch": 143, "lr": 8.112083493361194e-05} +{"train_loss": 0.014803078956902027, "grad_norm": 0.05714075267314911, "global_step": 15166, "epoch": 143, "lr": 8.11185151801243e-05} +{"train_loss": 0.018505025655031204, "grad_norm": 0.053836170583963394, "global_step": 15167, "epoch": 143, "lr": 8.111619531729973e-05} +{"train_loss": 0.01667601987719536, "grad_norm": 0.07084852457046509, "global_step": 15168, "epoch": 143, "lr": 8.111387534514638e-05} +{"train_loss": 0.015584046021103859, "grad_norm": 0.061076294630765915, "global_step": 15169, "epoch": 143, "lr": 8.11115552636724e-05} +{"train_loss": 0.01569557934999466, "grad_norm": 0.1490851193666458, "global_step": 15170, "epoch": 143, "lr": 8.110923507288595e-05} +{"train_loss": 0.017574677243828773, "grad_norm": 0.08854663372039795, "global_step": 15171, "epoch": 143, "lr": 8.110691477279517e-05} +{"train_loss": 0.01684301719069481, "grad_norm": 0.05689891427755356, "global_step": 15172, "epoch": 143, "lr": 8.110459436340822e-05} +{"train_loss": 0.017413124442100525, "grad_norm": 0.11362402886152267, "global_step": 15173, "epoch": 143, "lr": 8.110227384473325e-05} +{"train_loss": 0.016409847885370255, "grad_norm": 0.051985856145620346, "global_step": 15174, "epoch": 143, "lr": 8.109995321677844e-05} +{"train_loss": 0.017636893317103386, "grad_norm": 0.05376564338803291, "global_step": 15175, "epoch": 143, "lr": 8.10976324795519e-05} +{"train_loss": 0.01953822188079357, "grad_norm": 0.060182590037584305, "global_step": 15176, "epoch": 143, "lr": 8.109531163306181e-05} +{"train_loss": 0.012912211939692497, "grad_norm": 0.07516548782587051, "global_step": 15177, "epoch": 143, "lr": 8.10929906773163e-05} +{"train_loss": 0.014414668083190918, "grad_norm": 0.10617455095052719, "global_step": 15178, "epoch": 143, "lr": 8.109066961232357e-05} +{"train_loss": 0.014184797182679176, "grad_norm": 0.05705105885863304, "global_step": 15179, "epoch": 143, "lr": 8.108834843809173e-05} +{"train_loss": 0.018437514081597328, "grad_norm": 0.055726394057273865, "global_step": 15180, "epoch": 143, "lr": 8.108602715462896e-05} +{"train_loss": 0.018257588148117065, "grad_norm": 0.0528188981115818, "global_step": 15181, "epoch": 143, "lr": 8.10837057619434e-05} +{"train_loss": 0.016627129167318344, "grad_norm": 0.059913430362939835, "global_step": 15182, "epoch": 143, "lr": 8.108138426004324e-05} +{"train_loss": 0.01903974637389183, "grad_norm": 0.057081274688243866, "global_step": 15183, "epoch": 143, "lr": 8.107906264893659e-05} +{"train_loss": 0.016768718138337135, "grad_norm": 0.13062472641468048, "global_step": 15184, "epoch": 143, "lr": 8.107674092863162e-05} +{"train_loss": 0.016531068831682205, "grad_norm": 0.07538719475269318, "global_step": 15185, "epoch": 143, "lr": 8.107441909913653e-05} +{"train_loss": 0.019394923001527786, "grad_norm": 0.07602808624505997, "global_step": 15186, "epoch": 143, "lr": 8.107209716045943e-05} +{"train_loss": 0.017834480851888657, "grad_norm": 0.07654768228530884, "global_step": 15187, "epoch": 143, "lr": 8.106977511260849e-05} +{"train_loss": 0.0171679500490427, "grad_norm": 0.05873607099056244, "global_step": 15188, "epoch": 143, "lr": 8.106745295559186e-05} +{"train_loss": 0.017148802056908607, "grad_norm": 0.06508108228445053, "global_step": 15189, "epoch": 143, "lr": 8.106513068941773e-05} +{"train_loss": 0.01681477762758732, "grad_norm": 0.09360285103321075, "global_step": 15190, "epoch": 143, "lr": 8.106280831409424e-05} +{"train_loss": 0.017475660890340805, "grad_norm": 0.09474954009056091, "global_step": 15191, "epoch": 143, "lr": 8.106048582962954e-05} +{"train_loss": 0.01795707270503044, "grad_norm": 0.06519634276628494, "global_step": 15192, "epoch": 143, "lr": 8.105816323603181e-05} +{"train_loss": 0.01698565110564232, "grad_norm": 0.06318460404872894, "global_step": 15193, "epoch": 143, "lr": 8.10558405333092e-05} +{"train_loss": 0.015422812663018703, "grad_norm": 0.07636265456676483, "global_step": 15194, "epoch": 143, "lr": 8.105351772146987e-05} +{"train_loss": 0.017181159928441048, "grad_norm": 0.12617188692092896, "global_step": 15195, "epoch": 143, "lr": 8.105119480052198e-05} +{"train_loss": 0.015127013437449932, "grad_norm": 0.06466767191886902, "global_step": 15196, "epoch": 143, "lr": 8.10488717704737e-05} +{"train_loss": 0.01610959693789482, "grad_norm": 0.11849696189165115, "global_step": 15197, "epoch": 143, "lr": 8.104654863133319e-05} +{"train_loss": 0.015626810491085052, "grad_norm": 0.054092396050691605, "global_step": 15198, "epoch": 143, "lr": 8.104422538310861e-05} +{"train_loss": 0.020334508270025253, "grad_norm": 0.07284745573997498, "global_step": 15199, "epoch": 143, "lr": 8.104190202580812e-05} +{"train_loss": 0.015119572170078754, "grad_norm": 0.05407680943608284, "global_step": 15200, "epoch": 143, "lr": 8.103957855943988e-05} +{"train_loss": 0.020827017724514008, "grad_norm": 0.10283099859952927, "global_step": 15201, "epoch": 143, "lr": 8.103725498401207e-05} +{"train_loss": 0.013623742386698723, "grad_norm": 0.04976559057831764, "global_step": 15202, "epoch": 143, "lr": 8.103493129953284e-05} +{"train_loss": 0.01702037826180458, "grad_norm": 0.08209805935621262, "global_step": 15203, "epoch": 143, "lr": 8.103260750601035e-05} +{"train_loss": 0.01798309199512005, "grad_norm": 0.08188436180353165, "global_step": 15204, "epoch": 143, "lr": 8.103028360345278e-05} +{"train_loss": 0.018282614648342133, "grad_norm": 0.12328118085861206, "global_step": 15205, "epoch": 143, "lr": 8.102795959186828e-05} +{"train_loss": 0.017470797523856163, "grad_norm": 0.0720263347029686, "global_step": 15206, "epoch": 143, "lr": 8.102563547126504e-05} +{"train_loss": 0.01753002032637596, "grad_norm": 0.08397754281759262, "global_step": 15207, "epoch": 143, "lr": 8.10233112416512e-05} +{"train_loss": 0.01734292134642601, "grad_norm": 0.06412889808416367, "global_step": 15208, "epoch": 143, "lr": 8.102098690303493e-05} +{"train_loss": 0.015498867258429527, "grad_norm": 0.05505149066448212, "global_step": 15209, "epoch": 143, "lr": 8.101866245542442e-05} +{"train_loss": 0.01898631453514099, "grad_norm": 0.11229346692562103, "global_step": 15210, "epoch": 143, "lr": 8.101633789882781e-05} +{"train_loss": 0.017256546765565872, "grad_norm": 0.06810228526592255, "global_step": 15211, "epoch": 143, "lr": 8.101401323325327e-05} +{"train_loss": 0.016045669093728065, "grad_norm": 0.06482847779989243, "global_step": 15212, "epoch": 143, "lr": 8.101168845870899e-05} +{"train_loss": 0.015528864227235317, "grad_norm": 0.05546895042061806, "global_step": 15213, "epoch": 143, "lr": 8.100936357520312e-05} +{"train_loss": 0.016748588532209396, "grad_norm": 0.0859605222940445, "global_step": 15214, "epoch": 143, "lr": 8.100703858274382e-05} +{"train_loss": 0.017423368990421295, "grad_norm": 0.05171208828687668, "global_step": 15215, "epoch": 143, "lr": 8.100471348133928e-05} +{"train_loss": 0.015036186203360558, "grad_norm": 0.09199560433626175, "global_step": 15216, "epoch": 143, "lr": 8.100238827099766e-05} +{"train_loss": 0.017120109871029854, "grad_norm": 0.0666065365076065, "global_step": 15217, "epoch": 143, "lr": 8.100006295172713e-05} +{"train_loss": 0.01722298189997673, "grad_norm": 0.09518436342477798, "global_step": 15218, "epoch": 143, "lr": 8.099773752353585e-05} +{"train_loss": 0.01508413441479206, "grad_norm": 0.08792059868574142, "global_step": 15219, "epoch": 143, "lr": 8.099541198643201e-05} +{"train_loss": 0.0165044367313385, "grad_norm": 0.08694721758365631, "global_step": 15220, "epoch": 143, "lr": 8.099308634042377e-05} +{"train_loss": 0.016058767214417458, "grad_norm": 0.05723399668931961, "global_step": 15221, "epoch": 143, "lr": 8.099076058551931e-05} +{"train_loss": 0.014682089909911156, "grad_norm": 0.1538110226392746, "global_step": 15222, "epoch": 143, "lr": 8.098843472172679e-05} +{"train_loss": 0.015287088230252266, "grad_norm": 0.07732370495796204, "global_step": 15223, "epoch": 143, "lr": 8.098610874905439e-05} +{"train_loss": 0.01659555919468403, "grad_norm": 0.11274483799934387, "global_step": 15224, "epoch": 143, "lr": 8.098378266751028e-05} +{"train_loss": 0.015057126991450787, "grad_norm": 0.057396382093429565, "global_step": 15225, "epoch": 143, "lr": 8.098145647710262e-05} +{"train_loss": 0.018464140594005585, "grad_norm": 0.11464463919401169, "global_step": 15226, "epoch": 143, "lr": 8.09791301778396e-05} +{"train_loss": 0.01652340777218342, "grad_norm": 0.0832144170999527, "global_step": 15227, "epoch": 143, "lr": 8.097680376972941e-05} +{"train_loss": 0.017058691009879112, "grad_norm": 0.08023548871278763, "global_step": 15228, "epoch": 143, "lr": 8.097447725278018e-05} +{"train_loss": 0.019973957911133766, "grad_norm": 0.07135007530450821, "global_step": 15229, "epoch": 143, "lr": 8.097215062700011e-05} +{"train_loss": 0.017807012423872948, "grad_norm": 0.06312087923288345, "global_step": 15230, "epoch": 143, "lr": 8.096982389239738e-05} +{"train_loss": 0.0167495496571064, "grad_norm": 0.07119665294885635, "global_step": 15231, "epoch": 143, "lr": 8.096749704898014e-05} +{"train_loss": 0.016738872975111008, "grad_norm": 0.05303873121738434, "global_step": 15232, "epoch": 143, "lr": 8.096517009675661e-05} +{"train_loss": 0.013729849830269814, "grad_norm": 0.09493183344602585, "global_step": 15233, "epoch": 143, "lr": 8.096284303573492e-05} +{"train_loss": 0.016042347997426987, "grad_norm": 0.06700107455253601, "global_step": 15234, "epoch": 143, "lr": 8.096051586592327e-05} +{"train_loss": 0.01660020276904106, "grad_norm": 0.050876863300800323, "global_step": 15235, "epoch": 143, "lr": 8.095818858732983e-05} +{"train_loss": 0.013067614287137985, "grad_norm": 0.042965397238731384, "global_step": 15236, "epoch": 143, "lr": 8.095586119996277e-05} +{"train_loss": 0.014680054038763046, "grad_norm": 0.06927823275327682, "global_step": 15237, "epoch": 143, "lr": 8.09535337038303e-05} +{"train_loss": 0.01855379343032837, "grad_norm": 0.09200678020715714, "global_step": 15238, "epoch": 143, "lr": 8.095120609894056e-05} +{"train_loss": 0.015070952475070953, "grad_norm": 0.16155658662319183, "global_step": 15239, "epoch": 143, "lr": 8.094887838530174e-05} +{"train_loss": 0.01663210242986679, "grad_norm": 0.06526122987270355, "global_step": 15240, "epoch": 143, "lr": 8.094655056292202e-05} +{"train_loss": 0.01598266512155533, "grad_norm": 0.11634811013936996, "global_step": 15241, "epoch": 143, "lr": 8.094422263180958e-05} +{"train_loss": 0.01796814426779747, "grad_norm": 0.100099578499794, "global_step": 15242, "epoch": 143, "lr": 8.094189459197259e-05} +{"train_loss": 0.01601717621088028, "grad_norm": 0.07071878015995026, "global_step": 15243, "epoch": 143, "lr": 8.093956644341926e-05} +{"train_loss": 0.01825840398669243, "grad_norm": 0.05786500871181488, "global_step": 15244, "epoch": 143, "lr": 8.093723818615774e-05} +{"train_loss": 0.015188641846179962, "grad_norm": 0.05126900598406792, "global_step": 15245, "epoch": 143, "lr": 8.093490982019621e-05} +{"train_loss": 0.015760283917188644, "grad_norm": 0.1122894287109375, "global_step": 15246, "epoch": 143, "lr": 8.093258134554286e-05} +{"train_loss": 0.01748558320105076, "grad_norm": 0.1112167239189148, "global_step": 15247, "epoch": 143, "lr": 8.09302527622059e-05} +{"train_loss": 0.016098780557513237, "grad_norm": 0.20876188576221466, "global_step": 15248, "epoch": 143, "lr": 8.092792407019344e-05} +{"train_loss": 0.019894201308488846, "grad_norm": 0.14976416528224945, "global_step": 15249, "epoch": 143, "lr": 8.092559526951374e-05} +{"train_loss": 0.017802296206355095, "grad_norm": 0.14072483777999878, "global_step": 15250, "epoch": 143, "lr": 8.092326636017494e-05} +{"train_loss": 0.014889707788825035, "grad_norm": 0.05465106666088104, "global_step": 15251, "epoch": 143, "lr": 8.092093734218522e-05} +{"train_loss": 0.01744728535413742, "grad_norm": 0.07349865883588791, "global_step": 15252, "epoch": 143, "lr": 8.091860821555278e-05} +{"train_loss": 0.0148916682228446, "grad_norm": 0.0721183493733406, "global_step": 15253, "epoch": 143, "lr": 8.09162789802858e-05} +{"train_loss": 0.018165510147809982, "grad_norm": 0.13650375604629517, "global_step": 15254, "epoch": 143, "lr": 8.091394963639245e-05} +{"train_loss": 0.014302698895335197, "grad_norm": 0.10204090923070908, "global_step": 15255, "epoch": 143, "lr": 8.091162018388093e-05} +{"train_loss": 0.015033559873700142, "grad_norm": 0.05465281009674072, "global_step": 15256, "epoch": 143, "lr": 8.090929062275943e-05} +{"train_loss": 0.01863662339746952, "grad_norm": 0.14641670882701874, "global_step": 15257, "epoch": 143, "lr": 8.090696095303611e-05} +{"train_loss": 0.01712695136666298, "grad_norm": 0.060463547706604004, "global_step": 15258, "epoch": 143, "lr": 8.090463117471918e-05} +{"train_loss": 0.017682034522294998, "grad_norm": 0.061905719339847565, "global_step": 15259, "epoch": 143, "lr": 8.090230128781679e-05} +{"train_loss": 0.016091953963041306, "grad_norm": 0.07078167051076889, "global_step": 15260, "epoch": 143, "lr": 8.08999712923372e-05} +{"train_loss": 0.01617579162120819, "grad_norm": 0.13844221830368042, "global_step": 15261, "epoch": 143, "lr": 8.089764118828851e-05} +{"train_loss": 0.018450802192091942, "grad_norm": 0.0638810470700264, "global_step": 15262, "epoch": 143, "lr": 8.089531097567897e-05} +{"train_loss": 0.01671938262528406, "grad_norm": 0.16746576130390167, "global_step": 15263, "epoch": 143, "lr": 8.089298065451672e-05, "val_loss": 0.01861761510372162} +{"train_loss": 0.017763420939445496, "grad_norm": 0.07562362402677536, "global_step": 15264, "epoch": 144, "lr": 8.089065022481e-05} +{"train_loss": 0.01701272465288639, "grad_norm": 0.18522779643535614, "global_step": 15265, "epoch": 144, "lr": 8.088831968656693e-05} +{"train_loss": 0.012940120883286, "grad_norm": 0.047767866402864456, "global_step": 15266, "epoch": 144, "lr": 8.088598903979576e-05} +{"train_loss": 0.016566000878810883, "grad_norm": 0.051120396703481674, "global_step": 15267, "epoch": 144, "lr": 8.088365828450466e-05} +{"train_loss": 0.016063883900642395, "grad_norm": 0.07899786531925201, "global_step": 15268, "epoch": 144, "lr": 8.088132742070181e-05} +{"train_loss": 0.014390457421541214, "grad_norm": 0.08752086013555527, "global_step": 15269, "epoch": 144, "lr": 8.087899644839542e-05} +{"train_loss": 0.016964852809906006, "grad_norm": 0.0967063382267952, "global_step": 15270, "epoch": 144, "lr": 8.087666536759364e-05} +{"train_loss": 0.015982910990715027, "grad_norm": 0.07363259792327881, "global_step": 15271, "epoch": 144, "lr": 8.08743341783047e-05} +{"train_loss": 0.016305383294820786, "grad_norm": 0.13037370145320892, "global_step": 15272, "epoch": 144, "lr": 8.087200288053677e-05} +{"train_loss": 0.016553062945604324, "grad_norm": 0.08885636925697327, "global_step": 15273, "epoch": 144, "lr": 8.086967147429807e-05} +{"train_loss": 0.017107868567109108, "grad_norm": 0.07842092216014862, "global_step": 15274, "epoch": 144, "lr": 8.086733995959674e-05} +{"train_loss": 0.018948188051581383, "grad_norm": 0.09242575615644455, "global_step": 15275, "epoch": 144, "lr": 8.086500833644101e-05} +{"train_loss": 0.015940455719828606, "grad_norm": 0.053819529712200165, "global_step": 15276, "epoch": 144, "lr": 8.086267660483908e-05} +{"train_loss": 0.01681596226990223, "grad_norm": 0.06272953003644943, "global_step": 15277, "epoch": 144, "lr": 8.086034476479912e-05} +{"train_loss": 0.016429465264081955, "grad_norm": 0.06432060152292252, "global_step": 15278, "epoch": 144, "lr": 8.085801281632931e-05} +{"train_loss": 0.01541276928037405, "grad_norm": 0.147091343998909, "global_step": 15279, "epoch": 144, "lr": 8.085568075943787e-05} +{"train_loss": 0.01614810898900032, "grad_norm": 0.06541646271944046, "global_step": 15280, "epoch": 144, "lr": 8.0853348594133e-05} +{"train_loss": 0.018890073522925377, "grad_norm": 0.048844076693058014, "global_step": 15281, "epoch": 144, "lr": 8.085101632042287e-05} +{"train_loss": 0.01779997907578945, "grad_norm": 0.0969022661447525, "global_step": 15282, "epoch": 144, "lr": 8.084868393831569e-05} +{"train_loss": 0.018083592876791954, "grad_norm": 0.07192962616682053, "global_step": 15283, "epoch": 144, "lr": 8.084635144781965e-05} +{"train_loss": 0.018597010523080826, "grad_norm": 0.07776790112257004, "global_step": 15284, "epoch": 144, "lr": 8.084401884894294e-05} +{"train_loss": 0.016503361985087395, "grad_norm": 0.1053261086344719, "global_step": 15285, "epoch": 144, "lr": 8.084168614169378e-05} +{"train_loss": 0.015647193416953087, "grad_norm": 0.054499004036188126, "global_step": 15286, "epoch": 144, "lr": 8.083935332608032e-05} +{"train_loss": 0.013844541274011135, "grad_norm": 0.05222175642848015, "global_step": 15287, "epoch": 144, "lr": 8.083702040211081e-05} +{"train_loss": 0.017031259834766388, "grad_norm": 0.0817718431353569, "global_step": 15288, "epoch": 144, "lr": 8.08346873697934e-05} +{"train_loss": 0.018898552283644676, "grad_norm": 0.0638929232954979, "global_step": 15289, "epoch": 144, "lr": 8.083235422913631e-05} +{"train_loss": 0.01938130520284176, "grad_norm": 0.06962011009454727, "global_step": 15290, "epoch": 144, "lr": 8.083002098014776e-05} +{"train_loss": 0.01968976855278015, "grad_norm": 0.0771631971001625, "global_step": 15291, "epoch": 144, "lr": 8.08276876228359e-05} +{"train_loss": 0.01592968963086605, "grad_norm": 0.10284986346960068, "global_step": 15292, "epoch": 144, "lr": 8.082535415720896e-05} +{"train_loss": 0.015088589861989021, "grad_norm": 0.08986084163188934, "global_step": 15293, "epoch": 144, "lr": 8.082302058327514e-05} +{"train_loss": 0.017914244905114174, "grad_norm": 0.1817825883626938, "global_step": 15294, "epoch": 144, "lr": 8.082068690104261e-05} +{"train_loss": 0.01680004596710205, "grad_norm": 0.06326603144407272, "global_step": 15295, "epoch": 144, "lr": 8.08183531105196e-05} +{"train_loss": 0.019282329827547073, "grad_norm": 0.126246377825737, "global_step": 15296, "epoch": 144, "lr": 8.08160192117143e-05} +{"train_loss": 0.016489803791046143, "grad_norm": 0.049117714166641235, "global_step": 15297, "epoch": 144, "lr": 8.081368520463491e-05} +{"train_loss": 0.018731681630015373, "grad_norm": 0.07173559814691544, "global_step": 15298, "epoch": 144, "lr": 8.081135108928964e-05} +{"train_loss": 0.017800915986299515, "grad_norm": 0.0746169462800026, "global_step": 15299, "epoch": 144, "lr": 8.080901686568665e-05} +{"train_loss": 0.016141587868332863, "grad_norm": 0.04577753320336342, "global_step": 15300, "epoch": 144, "lr": 8.08066825338342e-05} +{"train_loss": 0.0177154503762722, "grad_norm": 0.07812856137752533, "global_step": 15301, "epoch": 144, "lr": 8.080434809374047e-05} +{"train_loss": 0.017436210066080093, "grad_norm": 0.18514446914196014, "global_step": 15302, "epoch": 144, "lr": 8.080201354541364e-05} +{"train_loss": 0.016582109034061432, "grad_norm": 0.1046765148639679, "global_step": 15303, "epoch": 144, "lr": 8.079967888886193e-05} +{"train_loss": 0.017269041389226913, "grad_norm": 0.10284312814474106, "global_step": 15304, "epoch": 144, "lr": 8.079734412409357e-05} +{"train_loss": 0.017388299107551575, "grad_norm": 0.08170149475336075, "global_step": 15305, "epoch": 144, "lr": 8.07950092511167e-05} +{"train_loss": 0.015059816651046276, "grad_norm": 0.055320173501968384, "global_step": 15306, "epoch": 144, "lr": 8.079267426993957e-05} +{"train_loss": 0.01621975749731064, "grad_norm": 0.16533251106739044, "global_step": 15307, "epoch": 144, "lr": 8.07903391805704e-05} +{"train_loss": 0.016949662938714027, "grad_norm": 0.06294865906238556, "global_step": 15308, "epoch": 144, "lr": 8.078800398301734e-05} +{"train_loss": 0.016639115288853645, "grad_norm": 0.08611004054546356, "global_step": 15309, "epoch": 144, "lr": 8.078566867728862e-05} +{"train_loss": 0.014850901439785957, "grad_norm": 0.07898344844579697, "global_step": 15310, "epoch": 144, "lr": 8.078333326339247e-05} +{"train_loss": 0.016509506851434708, "grad_norm": 0.06734553724527359, "global_step": 15311, "epoch": 144, "lr": 8.078099774133707e-05} +{"train_loss": 0.021287307143211365, "grad_norm": 0.06557731330394745, "global_step": 15312, "epoch": 144, "lr": 8.077866211113062e-05} +{"train_loss": 0.015857189893722534, "grad_norm": 0.0784856528043747, "global_step": 15313, "epoch": 144, "lr": 8.077632637278134e-05} +{"train_loss": 0.016662152484059334, "grad_norm": 0.11190050095319748, "global_step": 15314, "epoch": 144, "lr": 8.077399052629744e-05} +{"train_loss": 0.0172426737844944, "grad_norm": 0.07713300734758377, "global_step": 15315, "epoch": 144, "lr": 8.077165457168711e-05} +{"train_loss": 0.014863929711282253, "grad_norm": 0.11568332463502884, "global_step": 15316, "epoch": 144, "lr": 8.076931850895859e-05} +{"train_loss": 0.01976407691836357, "grad_norm": 0.06509311497211456, "global_step": 15317, "epoch": 144, "lr": 8.076698233812003e-05} +{"train_loss": 0.015707693994045258, "grad_norm": 0.09751396626234055, "global_step": 15318, "epoch": 144, "lr": 8.076464605917971e-05} +{"train_loss": 0.015565803274512291, "grad_norm": 0.08775240927934647, "global_step": 15319, "epoch": 144, "lr": 8.076230967214578e-05} +{"train_loss": 0.014119918458163738, "grad_norm": 0.04988805577158928, "global_step": 15320, "epoch": 144, "lr": 8.075997317702648e-05} +{"train_loss": 0.0158421341329813, "grad_norm": 0.0641644150018692, "global_step": 15321, "epoch": 144, "lr": 8.075763657383001e-05} +{"train_loss": 0.01622695103287697, "grad_norm": 0.0979473888874054, "global_step": 15322, "epoch": 144, "lr": 8.075529986256458e-05} +{"train_loss": 0.01751193031668663, "grad_norm": 0.10130950808525085, "global_step": 15323, "epoch": 144, "lr": 8.07529630432384e-05} +{"train_loss": 0.01928599551320076, "grad_norm": 0.08786962926387787, "global_step": 15324, "epoch": 144, "lr": 8.075062611585969e-05} +{"train_loss": 0.017066724598407745, "grad_norm": 0.09273023158311844, "global_step": 15325, "epoch": 144, "lr": 8.074828908043665e-05} +{"train_loss": 0.015979547053575516, "grad_norm": 0.05850418284535408, "global_step": 15326, "epoch": 144, "lr": 8.074595193697749e-05} +{"train_loss": 0.014558505266904831, "grad_norm": 0.052755337208509445, "global_step": 15327, "epoch": 144, "lr": 8.074361468549044e-05} +{"train_loss": 0.015540556982159615, "grad_norm": 0.09706509858369827, "global_step": 15328, "epoch": 144, "lr": 8.074127732598368e-05} +{"train_loss": 0.01661154441535473, "grad_norm": 0.07062515616416931, "global_step": 15329, "epoch": 144, "lr": 8.073893985846544e-05} +{"train_loss": 0.014489378780126572, "grad_norm": 0.06553616374731064, "global_step": 15330, "epoch": 144, "lr": 8.073660228294394e-05} +{"train_loss": 0.013614624738693237, "grad_norm": 0.059054769575595856, "global_step": 15331, "epoch": 144, "lr": 8.073426459942739e-05} +{"train_loss": 0.017195405438542366, "grad_norm": 0.07073069363832474, "global_step": 15332, "epoch": 144, "lr": 8.0731926807924e-05} +{"train_loss": 0.017058754339814186, "grad_norm": 0.08779000490903854, "global_step": 15333, "epoch": 144, "lr": 8.072958890844198e-05} +{"train_loss": 0.015446104109287262, "grad_norm": 0.09280569851398468, "global_step": 15334, "epoch": 144, "lr": 8.072725090098955e-05} +{"train_loss": 0.01502948347479105, "grad_norm": 0.10464835911989212, "global_step": 15335, "epoch": 144, "lr": 8.072491278557493e-05} +{"train_loss": 0.016772393137216568, "grad_norm": 0.08118731528520584, "global_step": 15336, "epoch": 144, "lr": 8.072257456220632e-05} +{"train_loss": 0.014716321602463722, "grad_norm": 0.0668310821056366, "global_step": 15337, "epoch": 144, "lr": 8.072023623089193e-05} +{"train_loss": 0.020390063524246216, "grad_norm": 0.14669737219810486, "global_step": 15338, "epoch": 144, "lr": 8.071789779164e-05} +{"train_loss": 0.01669169031083584, "grad_norm": 0.08221185207366943, "global_step": 15339, "epoch": 144, "lr": 8.071555924445875e-05} +{"train_loss": 0.016951357945799828, "grad_norm": 0.07005990296602249, "global_step": 15340, "epoch": 144, "lr": 8.071322058935638e-05} +{"train_loss": 0.01677979901432991, "grad_norm": 0.07403342425823212, "global_step": 15341, "epoch": 144, "lr": 8.07108818263411e-05} +{"train_loss": 0.01638052985072136, "grad_norm": 0.0669165626168251, "global_step": 15342, "epoch": 144, "lr": 8.070854295542114e-05} +{"train_loss": 0.01727117970585823, "grad_norm": 0.07425109297037125, "global_step": 15343, "epoch": 144, "lr": 8.070620397660471e-05} +{"train_loss": 0.016518743708729744, "grad_norm": 0.06259810924530029, "global_step": 15344, "epoch": 144, "lr": 8.070386488990004e-05} +{"train_loss": 0.0156625397503376, "grad_norm": 0.12581029534339905, "global_step": 15345, "epoch": 144, "lr": 8.070152569531533e-05} +{"train_loss": 0.01707303524017334, "grad_norm": 0.13895174860954285, "global_step": 15346, "epoch": 144, "lr": 8.069918639285882e-05} +{"train_loss": 0.0145052345469594, "grad_norm": 0.06421076506376266, "global_step": 15347, "epoch": 144, "lr": 8.069684698253873e-05} +{"train_loss": 0.018686063587665558, "grad_norm": 0.15407870709896088, "global_step": 15348, "epoch": 144, "lr": 8.069450746436327e-05} +{"train_loss": 0.01535182073712349, "grad_norm": 0.08702026307582855, "global_step": 15349, "epoch": 144, "lr": 8.069216783834064e-05} +{"train_loss": 0.01642524264752865, "grad_norm": 0.0945330262184143, "global_step": 15350, "epoch": 144, "lr": 8.06898281044791e-05} +{"train_loss": 0.015257567167282104, "grad_norm": 0.06304170191287994, "global_step": 15351, "epoch": 144, "lr": 8.068748826278683e-05} +{"train_loss": 0.02054530568420887, "grad_norm": 0.06056331470608711, "global_step": 15352, "epoch": 144, "lr": 8.068514831327207e-05} +{"train_loss": 0.015268923714756966, "grad_norm": 0.1088176742196083, "global_step": 15353, "epoch": 144, "lr": 8.068280825594306e-05} +{"train_loss": 0.01636098138988018, "grad_norm": 0.10743063688278198, "global_step": 15354, "epoch": 144, "lr": 8.0680468090808e-05} +{"train_loss": 0.016899224370718002, "grad_norm": 0.09465520083904266, "global_step": 15355, "epoch": 144, "lr": 8.067812781787512e-05} +{"train_loss": 0.01495295763015747, "grad_norm": 0.05379534140229225, "global_step": 15356, "epoch": 144, "lr": 8.067578743715264e-05} +{"train_loss": 0.0168637465685606, "grad_norm": 0.1109045073390007, "global_step": 15357, "epoch": 144, "lr": 8.067344694864878e-05} +{"train_loss": 0.01692100800573826, "grad_norm": 0.09256315976381302, "global_step": 15358, "epoch": 144, "lr": 8.067110635237176e-05} +{"train_loss": 0.01671501062810421, "grad_norm": 0.10685032606124878, "global_step": 15359, "epoch": 144, "lr": 8.066876564832983e-05} +{"train_loss": 0.018511775881052017, "grad_norm": 0.08133586496114731, "global_step": 15360, "epoch": 144, "lr": 8.066642483653118e-05} +{"train_loss": 0.015870586037635803, "grad_norm": 0.08020635694265366, "global_step": 15361, "epoch": 144, "lr": 8.066408391698405e-05} +{"train_loss": 0.019578151404857635, "grad_norm": 0.13158859312534332, "global_step": 15362, "epoch": 144, "lr": 8.066174288969668e-05} +{"train_loss": 0.016696017235517502, "grad_norm": 0.06314737349748611, "global_step": 15363, "epoch": 144, "lr": 8.065940175467727e-05} +{"train_loss": 0.01410524733364582, "grad_norm": 0.05980480834841728, "global_step": 15364, "epoch": 144, "lr": 8.065706051193404e-05} +{"train_loss": 0.01673031598329544, "grad_norm": 0.05346858128905296, "global_step": 15365, "epoch": 144, "lr": 8.065471916147525e-05} +{"train_loss": 0.014119384810328484, "grad_norm": 0.1101006343960762, "global_step": 15366, "epoch": 144, "lr": 8.065237770330911e-05} +{"train_loss": 0.0192379429936409, "grad_norm": 0.07384390383958817, "global_step": 15367, "epoch": 144, "lr": 8.065003613744384e-05} +{"train_loss": 0.01782422512769699, "grad_norm": 0.06191875413060188, "global_step": 15368, "epoch": 144, "lr": 8.064769446388767e-05} +{"train_loss": 0.01671064796172223, "grad_norm": 0.13876788318157196, "global_step": 15369, "epoch": 144, "lr": 8.064535268264883e-05, "val_loss": 0.018992142751812935} +{"train_loss": 0.013156124390661716, "grad_norm": 0.09334687888622284, "global_step": 15370, "epoch": 145, "lr": 8.064301079373555e-05} +{"train_loss": 0.015528149902820587, "grad_norm": 0.0692170038819313, "global_step": 15371, "epoch": 145, "lr": 8.064066879715606e-05} +{"train_loss": 0.01940527744591236, "grad_norm": 0.06967226415872574, "global_step": 15372, "epoch": 145, "lr": 8.063832669291858e-05} +{"train_loss": 0.015043321065604687, "grad_norm": 0.11584550142288208, "global_step": 15373, "epoch": 145, "lr": 8.063598448103134e-05} +{"train_loss": 0.016264451667666435, "grad_norm": 0.0694388896226883, "global_step": 15374, "epoch": 145, "lr": 8.063364216150257e-05} +{"train_loss": 0.015436960384249687, "grad_norm": 0.08495903015136719, "global_step": 15375, "epoch": 145, "lr": 8.063129973434053e-05} +{"train_loss": 0.01613331213593483, "grad_norm": 0.12638221681118011, "global_step": 15376, "epoch": 145, "lr": 8.06289571995534e-05} +{"train_loss": 0.01816934533417225, "grad_norm": 0.0640093982219696, "global_step": 15377, "epoch": 145, "lr": 8.062661455714944e-05} +{"train_loss": 0.015849974006414413, "grad_norm": 0.05109298974275589, "global_step": 15378, "epoch": 145, "lr": 8.062427180713688e-05} +{"train_loss": 0.01891757920384407, "grad_norm": 0.0763479471206665, "global_step": 15379, "epoch": 145, "lr": 8.062192894952394e-05} +{"train_loss": 0.016636313870549202, "grad_norm": 0.14121033251285553, "global_step": 15380, "epoch": 145, "lr": 8.061958598431887e-05} +{"train_loss": 0.01710652746260166, "grad_norm": 0.06260941177606583, "global_step": 15381, "epoch": 145, "lr": 8.061724291152988e-05} +{"train_loss": 0.01475595310330391, "grad_norm": 0.06161554157733917, "global_step": 15382, "epoch": 145, "lr": 8.061489973116522e-05} +{"train_loss": 0.01727950945496559, "grad_norm": 0.11952371895313263, "global_step": 15383, "epoch": 145, "lr": 8.061255644323312e-05} +{"train_loss": 0.01606636866927147, "grad_norm": 0.049470480531454086, "global_step": 15384, "epoch": 145, "lr": 8.061021304774179e-05} +{"train_loss": 0.017575832083821297, "grad_norm": 0.06849820911884308, "global_step": 15385, "epoch": 145, "lr": 8.06078695446995e-05} +{"train_loss": 0.01793830841779709, "grad_norm": 0.059045448899269104, "global_step": 15386, "epoch": 145, "lr": 8.060552593411446e-05} +{"train_loss": 0.016145914793014526, "grad_norm": 0.10092990100383759, "global_step": 15387, "epoch": 145, "lr": 8.060318221599492e-05} +{"train_loss": 0.01714431308209896, "grad_norm": 0.05254116654396057, "global_step": 15388, "epoch": 145, "lr": 8.060083839034911e-05} +{"train_loss": 0.01804789900779724, "grad_norm": 0.06464449316263199, "global_step": 15389, "epoch": 145, "lr": 8.059849445718524e-05} +{"train_loss": 0.016010232269763947, "grad_norm": 0.06855196505784988, "global_step": 15390, "epoch": 145, "lr": 8.059615041651159e-05} +{"train_loss": 0.014965251088142395, "grad_norm": 0.12884818017482758, "global_step": 15391, "epoch": 145, "lr": 8.059380626833638e-05} +{"train_loss": 0.016704080626368523, "grad_norm": 0.1333436816930771, "global_step": 15392, "epoch": 145, "lr": 8.05914620126678e-05} +{"train_loss": 0.015984054654836655, "grad_norm": 0.11004338413476944, "global_step": 15393, "epoch": 145, "lr": 8.058911764951415e-05} +{"train_loss": 0.017079632729291916, "grad_norm": 0.05510878562927246, "global_step": 15394, "epoch": 145, "lr": 8.058677317888366e-05} +{"train_loss": 0.015655022114515305, "grad_norm": 0.044362518936395645, "global_step": 15395, "epoch": 145, "lr": 8.058442860078453e-05} +{"train_loss": 0.015871766954660416, "grad_norm": 0.0525437593460083, "global_step": 15396, "epoch": 145, "lr": 8.058208391522502e-05} +{"train_loss": 0.01637556031346321, "grad_norm": 0.07384205609560013, "global_step": 15397, "epoch": 145, "lr": 8.057973912221337e-05} +{"train_loss": 0.017077015712857246, "grad_norm": 0.06877975165843964, "global_step": 15398, "epoch": 145, "lr": 8.057739422175782e-05} +{"train_loss": 0.0179685577750206, "grad_norm": 0.13128678500652313, "global_step": 15399, "epoch": 145, "lr": 8.05750492138666e-05} +{"train_loss": 0.015910513699054718, "grad_norm": 0.05497629567980766, "global_step": 15400, "epoch": 145, "lr": 8.057270409854795e-05} +{"train_loss": 0.01735774241387844, "grad_norm": 0.06371341645717621, "global_step": 15401, "epoch": 145, "lr": 8.057035887581013e-05} +{"train_loss": 0.014251274056732655, "grad_norm": 0.055795177817344666, "global_step": 15402, "epoch": 145, "lr": 8.056801354566134e-05} +{"train_loss": 0.019261715933680534, "grad_norm": 0.06564098596572876, "global_step": 15403, "epoch": 145, "lr": 8.056566810810986e-05} +{"train_loss": 0.016238927841186523, "grad_norm": 0.06224532425403595, "global_step": 15404, "epoch": 145, "lr": 8.056332256316392e-05} +{"train_loss": 0.015694547444581985, "grad_norm": 0.14199434220790863, "global_step": 15405, "epoch": 145, "lr": 8.056097691083176e-05} +{"train_loss": 0.018154654651880264, "grad_norm": 0.060309652239084244, "global_step": 15406, "epoch": 145, "lr": 8.055863115112158e-05} +{"train_loss": 0.01651313528418541, "grad_norm": 0.20441441237926483, "global_step": 15407, "epoch": 145, "lr": 8.05562852840417e-05} +{"train_loss": 0.017523834481835365, "grad_norm": 0.10139790177345276, "global_step": 15408, "epoch": 145, "lr": 8.05539393096003e-05} +{"train_loss": 0.014969720505177975, "grad_norm": 0.05462124943733215, "global_step": 15409, "epoch": 145, "lr": 8.055159322780566e-05} +{"train_loss": 0.016305899247527122, "grad_norm": 0.05375533178448677, "global_step": 15410, "epoch": 145, "lr": 8.0549247038666e-05} +{"train_loss": 0.016332820057868958, "grad_norm": 0.06313338130712509, "global_step": 15411, "epoch": 145, "lr": 8.054690074218956e-05} +{"train_loss": 0.01587516814470291, "grad_norm": 0.058205679059028625, "global_step": 15412, "epoch": 145, "lr": 8.054455433838461e-05} +{"train_loss": 0.017811039462685585, "grad_norm": 0.09056475013494492, "global_step": 15413, "epoch": 145, "lr": 8.054220782725938e-05} +{"train_loss": 0.014313159510493279, "grad_norm": 0.06557890772819519, "global_step": 15414, "epoch": 145, "lr": 8.05398612088221e-05} +{"train_loss": 0.015542108565568924, "grad_norm": 0.097285695374012, "global_step": 15415, "epoch": 145, "lr": 8.053751448308105e-05} +{"train_loss": 0.016852037981152534, "grad_norm": 0.12709444761276245, "global_step": 15416, "epoch": 145, "lr": 8.053516765004443e-05} +{"train_loss": 0.015435341745615005, "grad_norm": 0.0779753029346466, "global_step": 15417, "epoch": 145, "lr": 8.053282070972053e-05} +{"train_loss": 0.015070594847202301, "grad_norm": 0.08870057761669159, "global_step": 15418, "epoch": 145, "lr": 8.053047366211756e-05} +{"train_loss": 0.01673394814133644, "grad_norm": 0.06026427075266838, "global_step": 15419, "epoch": 145, "lr": 8.052812650724381e-05} +{"train_loss": 0.017392314970493317, "grad_norm": 0.11167538166046143, "global_step": 15420, "epoch": 145, "lr": 8.052577924510747e-05} +{"train_loss": 0.017427612096071243, "grad_norm": 0.07437324523925781, "global_step": 15421, "epoch": 145, "lr": 8.052343187571683e-05} +{"train_loss": 0.018262650817632675, "grad_norm": 0.0838322639465332, "global_step": 15422, "epoch": 145, "lr": 8.052108439908013e-05} +{"train_loss": 0.016873784363269806, "grad_norm": 0.1002555564045906, "global_step": 15423, "epoch": 145, "lr": 8.051873681520561e-05} +{"train_loss": 0.016865406185388565, "grad_norm": 0.07227057963609695, "global_step": 15424, "epoch": 145, "lr": 8.051638912410151e-05} +{"train_loss": 0.01582123525440693, "grad_norm": 0.06830725818872452, "global_step": 15425, "epoch": 145, "lr": 8.051404132577612e-05} +{"train_loss": 0.01641532965004444, "grad_norm": 0.06070886552333832, "global_step": 15426, "epoch": 145, "lr": 8.051169342023763e-05} +{"train_loss": 0.016000552102923393, "grad_norm": 0.07263000309467316, "global_step": 15427, "epoch": 145, "lr": 8.050934540749433e-05} +{"train_loss": 0.017135173082351685, "grad_norm": 0.06391820311546326, "global_step": 15428, "epoch": 145, "lr": 8.050699728755444e-05} +{"train_loss": 0.01576126739382744, "grad_norm": 0.12748663127422333, "global_step": 15429, "epoch": 145, "lr": 8.050464906042626e-05} +{"train_loss": 0.01787736639380455, "grad_norm": 0.06701719015836716, "global_step": 15430, "epoch": 145, "lr": 8.050230072611797e-05} +{"train_loss": 0.02220269665122032, "grad_norm": 0.19377855956554413, "global_step": 15431, "epoch": 145, "lr": 8.04999522846379e-05} +{"train_loss": 0.0154641792178154, "grad_norm": 0.04900108650326729, "global_step": 15432, "epoch": 145, "lr": 8.049760373599423e-05} +{"train_loss": 0.016844261437654495, "grad_norm": 0.17095907032489777, "global_step": 15433, "epoch": 145, "lr": 8.049525508019527e-05} +{"train_loss": 0.015441415831446648, "grad_norm": 0.04492110386490822, "global_step": 15434, "epoch": 145, "lr": 8.049290631724922e-05} +{"train_loss": 0.015593553893268108, "grad_norm": 0.11134142428636551, "global_step": 15435, "epoch": 145, "lr": 8.049055744716438e-05} +{"train_loss": 0.01693512499332428, "grad_norm": 0.0677829459309578, "global_step": 15436, "epoch": 145, "lr": 8.048820846994896e-05} +{"train_loss": 0.015975773334503174, "grad_norm": 0.09885914623737335, "global_step": 15437, "epoch": 145, "lr": 8.048585938561126e-05} +{"train_loss": 0.016239838674664497, "grad_norm": 0.05581996589899063, "global_step": 15438, "epoch": 145, "lr": 8.04835101941595e-05} +{"train_loss": 0.01564459130167961, "grad_norm": 0.0795353576540947, "global_step": 15439, "epoch": 145, "lr": 8.048116089560194e-05} +{"train_loss": 0.015976710245013237, "grad_norm": 0.0888851135969162, "global_step": 15440, "epoch": 145, "lr": 8.047881148994682e-05} +{"train_loss": 0.017069067806005478, "grad_norm": 0.05553048476576805, "global_step": 15441, "epoch": 145, "lr": 8.047646197720242e-05} +{"train_loss": 0.017598984763026237, "grad_norm": 0.10487052798271179, "global_step": 15442, "epoch": 145, "lr": 8.0474112357377e-05} +{"train_loss": 0.017380308359861374, "grad_norm": 0.06908303499221802, "global_step": 15443, "epoch": 145, "lr": 8.04717626304788e-05} +{"train_loss": 0.018161531537771225, "grad_norm": 0.08559072017669678, "global_step": 15444, "epoch": 145, "lr": 8.046941279651606e-05} +{"train_loss": 0.016494477167725563, "grad_norm": 0.07061420381069183, "global_step": 15445, "epoch": 145, "lr": 8.046706285549708e-05} +{"train_loss": 0.013995690271258354, "grad_norm": 0.060142286121845245, "global_step": 15446, "epoch": 145, "lr": 8.046471280743007e-05} +{"train_loss": 0.017461184412240982, "grad_norm": 0.11135432869195938, "global_step": 15447, "epoch": 145, "lr": 8.046236265232332e-05} +{"train_loss": 0.019526902586221695, "grad_norm": 0.22824163734912872, "global_step": 15448, "epoch": 145, "lr": 8.046001239018505e-05} +{"train_loss": 0.015592566691339016, "grad_norm": 0.06399591267108917, "global_step": 15449, "epoch": 145, "lr": 8.045766202102358e-05} +{"train_loss": 0.015016915276646614, "grad_norm": 0.06703504920005798, "global_step": 15450, "epoch": 145, "lr": 8.045531154484712e-05} +{"train_loss": 0.01876254566013813, "grad_norm": 0.09118007123470306, "global_step": 15451, "epoch": 145, "lr": 8.045296096166393e-05} +{"train_loss": 0.01673639565706253, "grad_norm": 0.06731411814689636, "global_step": 15452, "epoch": 145, "lr": 8.045061027148228e-05} +{"train_loss": 0.016270333901047707, "grad_norm": 0.06065516173839569, "global_step": 15453, "epoch": 145, "lr": 8.044825947431043e-05} +{"train_loss": 0.01569792442023754, "grad_norm": 0.07895486801862717, "global_step": 15454, "epoch": 145, "lr": 8.044590857015664e-05} +{"train_loss": 0.01682080700993538, "grad_norm": 0.08575186133384705, "global_step": 15455, "epoch": 145, "lr": 8.044355755902917e-05} +{"train_loss": 0.017396587878465652, "grad_norm": 0.09822903573513031, "global_step": 15456, "epoch": 145, "lr": 8.044120644093627e-05} +{"train_loss": 0.017383821308612823, "grad_norm": 0.19749478995800018, "global_step": 15457, "epoch": 145, "lr": 8.043885521588622e-05} +{"train_loss": 0.014273782260715961, "grad_norm": 0.12753213942050934, "global_step": 15458, "epoch": 145, "lr": 8.043650388388727e-05} +{"train_loss": 0.018185079097747803, "grad_norm": 0.09151731431484222, "global_step": 15459, "epoch": 145, "lr": 8.043415244494768e-05} +{"train_loss": 0.018770357593894005, "grad_norm": 0.0963345617055893, "global_step": 15460, "epoch": 145, "lr": 8.04318008990757e-05} +{"train_loss": 0.016452660784125328, "grad_norm": 0.10003601759672165, "global_step": 15461, "epoch": 145, "lr": 8.042944924627962e-05} +{"train_loss": 0.014866857789456844, "grad_norm": 0.06812188029289246, "global_step": 15462, "epoch": 145, "lr": 8.042709748656768e-05} +{"train_loss": 0.018875591456890106, "grad_norm": 0.1334533393383026, "global_step": 15463, "epoch": 145, "lr": 8.042474561994816e-05} +{"train_loss": 0.014895849861204624, "grad_norm": 0.05480519309639931, "global_step": 15464, "epoch": 145, "lr": 8.042239364642932e-05} +{"train_loss": 0.019009152427315712, "grad_norm": 0.060698166489601135, "global_step": 15465, "epoch": 145, "lr": 8.04200415660194e-05} +{"train_loss": 0.01610785722732544, "grad_norm": 0.1422027051448822, "global_step": 15466, "epoch": 145, "lr": 8.04176893787267e-05} +{"train_loss": 0.016379736363887787, "grad_norm": 0.08095388859510422, "global_step": 15467, "epoch": 145, "lr": 8.041533708455947e-05} +{"train_loss": 0.015800703316926956, "grad_norm": 0.059112872928380966, "global_step": 15468, "epoch": 145, "lr": 8.041298468352596e-05} +{"train_loss": 0.01716977171599865, "grad_norm": 0.06506402045488358, "global_step": 15469, "epoch": 145, "lr": 8.041063217563444e-05} +{"train_loss": 0.01692265458405018, "grad_norm": 0.07196654379367828, "global_step": 15470, "epoch": 145, "lr": 8.040827956089321e-05} +{"train_loss": 0.018403958529233932, "grad_norm": 0.07091089338064194, "global_step": 15471, "epoch": 145, "lr": 8.04059268393105e-05} +{"train_loss": 0.01721843332052231, "grad_norm": 0.061675067991018295, "global_step": 15472, "epoch": 145, "lr": 8.040357401089458e-05} +{"train_loss": 0.01808244362473488, "grad_norm": 0.0637395977973938, "global_step": 15473, "epoch": 145, "lr": 8.040122107565372e-05} +{"train_loss": 0.016813479363918304, "grad_norm": 0.05990796163678169, "global_step": 15474, "epoch": 145, "lr": 8.03988680335962e-05} +{"train_loss": 0.016660486427047906, "grad_norm": 0.07555560022592545, "global_step": 15475, "epoch": 145, "lr": 8.039651488473028e-05, "val_loss": 0.01914229802787304} +{"train_loss": 0.01627582684159279, "grad_norm": 0.06076763942837715, "global_step": 15476, "epoch": 146, "lr": 8.039416162906421e-05} +{"train_loss": 0.015204467810690403, "grad_norm": 0.10695794969797134, "global_step": 15477, "epoch": 146, "lr": 8.039180826660628e-05} +{"train_loss": 0.018772101029753685, "grad_norm": 0.059421490877866745, "global_step": 15478, "epoch": 146, "lr": 8.038945479736477e-05} +{"train_loss": 0.015577467158436775, "grad_norm": 0.054537344723939896, "global_step": 15479, "epoch": 146, "lr": 8.038710122134791e-05} +{"train_loss": 0.016824238002300262, "grad_norm": 0.158869668841362, "global_step": 15480, "epoch": 146, "lr": 8.0384747538564e-05} +{"train_loss": 0.018077192828059196, "grad_norm": 0.10355950146913528, "global_step": 15481, "epoch": 146, "lr": 8.038239374902131e-05} +{"train_loss": 0.018812065944075584, "grad_norm": 0.1065472960472107, "global_step": 15482, "epoch": 146, "lr": 8.038003985272808e-05} +{"train_loss": 0.02032003551721573, "grad_norm": 0.08981267362833023, "global_step": 15483, "epoch": 146, "lr": 8.037768584969262e-05} +{"train_loss": 0.016297321766614914, "grad_norm": 0.09271939098834991, "global_step": 15484, "epoch": 146, "lr": 8.037533173992317e-05} +{"train_loss": 0.015939701348543167, "grad_norm": 0.07396744936704636, "global_step": 15485, "epoch": 146, "lr": 8.037297752342803e-05} +{"train_loss": 0.017297087237238884, "grad_norm": 0.05324089527130127, "global_step": 15486, "epoch": 146, "lr": 8.037062320021545e-05} +{"train_loss": 0.01666608266532421, "grad_norm": 0.06451080739498138, "global_step": 15487, "epoch": 146, "lr": 8.036826877029371e-05} +{"train_loss": 0.014331160113215446, "grad_norm": 0.158096045255661, "global_step": 15488, "epoch": 146, "lr": 8.036591423367107e-05} +{"train_loss": 0.01666756346821785, "grad_norm": 0.1467357575893402, "global_step": 15489, "epoch": 146, "lr": 8.036355959035581e-05} +{"train_loss": 0.015414206311106682, "grad_norm": 0.13183081150054932, "global_step": 15490, "epoch": 146, "lr": 8.036120484035622e-05} +{"train_loss": 0.016186360269784927, "grad_norm": 0.08578059077262878, "global_step": 15491, "epoch": 146, "lr": 8.035884998368056e-05} +{"train_loss": 0.016477834433317184, "grad_norm": 0.07436636835336685, "global_step": 15492, "epoch": 146, "lr": 8.03564950203371e-05} +{"train_loss": 0.014912998303771019, "grad_norm": 0.05444258078932762, "global_step": 15493, "epoch": 146, "lr": 8.035413995033411e-05} +{"train_loss": 0.018559912219643593, "grad_norm": 0.05136033520102501, "global_step": 15494, "epoch": 146, "lr": 8.035178477367988e-05} +{"train_loss": 0.015071203000843525, "grad_norm": 0.059631723910570145, "global_step": 15495, "epoch": 146, "lr": 8.034942949038268e-05} +{"train_loss": 0.014439880847930908, "grad_norm": 0.058641739189624786, "global_step": 15496, "epoch": 146, "lr": 8.034707410045079e-05} +{"train_loss": 0.01721278391778469, "grad_norm": 0.09583348780870438, "global_step": 15497, "epoch": 146, "lr": 8.034471860389247e-05} +{"train_loss": 0.013626869767904282, "grad_norm": 0.05782118812203407, "global_step": 15498, "epoch": 146, "lr": 8.0342363000716e-05} +{"train_loss": 0.014137310907244682, "grad_norm": 0.0538225881755352, "global_step": 15499, "epoch": 146, "lr": 8.034000729092968e-05} +{"train_loss": 0.015944935381412506, "grad_norm": 0.05945109203457832, "global_step": 15500, "epoch": 146, "lr": 8.033765147454175e-05} +{"train_loss": 0.016457993537187576, "grad_norm": 0.06663449853658676, "global_step": 15501, "epoch": 146, "lr": 8.033529555156052e-05} +{"train_loss": 0.0161119531840086, "grad_norm": 0.09348223358392715, "global_step": 15502, "epoch": 146, "lr": 8.033293952199424e-05} +{"train_loss": 0.01915024034678936, "grad_norm": 0.0725129246711731, "global_step": 15503, "epoch": 146, "lr": 8.033058338585121e-05} +{"train_loss": 0.019449444487690926, "grad_norm": 0.06594249606132507, "global_step": 15504, "epoch": 146, "lr": 8.03282271431397e-05} +{"train_loss": 0.01710846647620201, "grad_norm": 0.08910256624221802, "global_step": 15505, "epoch": 146, "lr": 8.032587079386799e-05} +{"train_loss": 0.016661997884511948, "grad_norm": 0.10604061931371689, "global_step": 15506, "epoch": 146, "lr": 8.032351433804435e-05} +{"train_loss": 0.015737559646368027, "grad_norm": 0.045391883701086044, "global_step": 15507, "epoch": 146, "lr": 8.032115777567708e-05} +{"train_loss": 0.01790039986371994, "grad_norm": 0.0789375752210617, "global_step": 15508, "epoch": 146, "lr": 8.031880110677445e-05} +{"train_loss": 0.016383714973926544, "grad_norm": 0.09894535690546036, "global_step": 15509, "epoch": 146, "lr": 8.031644433134473e-05} +{"train_loss": 0.017252009361982346, "grad_norm": 0.08592355996370316, "global_step": 15510, "epoch": 146, "lr": 8.031408744939621e-05} +{"train_loss": 0.014273170381784439, "grad_norm": 0.06833946704864502, "global_step": 15511, "epoch": 146, "lr": 8.031173046093716e-05} +{"train_loss": 0.016725100576877594, "grad_norm": 0.05308645963668823, "global_step": 15512, "epoch": 146, "lr": 8.030937336597588e-05} +{"train_loss": 0.01783391460776329, "grad_norm": 0.06524267047643661, "global_step": 15513, "epoch": 146, "lr": 8.030701616452065e-05} +{"train_loss": 0.016787532716989517, "grad_norm": 0.04949796572327614, "global_step": 15514, "epoch": 146, "lr": 8.030465885657974e-05} +{"train_loss": 0.019560344517230988, "grad_norm": 0.17070740461349487, "global_step": 15515, "epoch": 146, "lr": 8.030230144216144e-05} +{"train_loss": 0.015414774417877197, "grad_norm": 0.09224913269281387, "global_step": 15516, "epoch": 146, "lr": 8.029994392127403e-05} +{"train_loss": 0.016276121139526367, "grad_norm": 0.09216393530368805, "global_step": 15517, "epoch": 146, "lr": 8.029758629392578e-05} +{"train_loss": 0.014053702354431152, "grad_norm": 0.14038927853107452, "global_step": 15518, "epoch": 146, "lr": 8.029522856012501e-05} +{"train_loss": 0.017664290964603424, "grad_norm": 0.08697162568569183, "global_step": 15519, "epoch": 146, "lr": 8.029287071987997e-05} +{"train_loss": 0.015316958539187908, "grad_norm": 0.16275812685489655, "global_step": 15520, "epoch": 146, "lr": 8.029051277319896e-05} +{"train_loss": 0.016705838963389397, "grad_norm": 0.11445505917072296, "global_step": 15521, "epoch": 146, "lr": 8.028815472009026e-05} +{"train_loss": 0.01637003943324089, "grad_norm": 0.07519207149744034, "global_step": 15522, "epoch": 146, "lr": 8.028579656056216e-05} +{"train_loss": 0.01389114186167717, "grad_norm": 0.07989022880792618, "global_step": 15523, "epoch": 146, "lr": 8.028343829462294e-05} +{"train_loss": 0.016546839848160744, "grad_norm": 0.07530678808689117, "global_step": 15524, "epoch": 146, "lr": 8.028107992228088e-05} +{"train_loss": 0.016731616109609604, "grad_norm": 0.1136549860239029, "global_step": 15525, "epoch": 146, "lr": 8.027872144354429e-05} +{"train_loss": 0.01521420106291771, "grad_norm": 0.0700506940484047, "global_step": 15526, "epoch": 146, "lr": 8.027636285842141e-05} +{"train_loss": 0.015367429703474045, "grad_norm": 0.12271519005298615, "global_step": 15527, "epoch": 146, "lr": 8.027400416692059e-05} +{"train_loss": 0.01753063313663006, "grad_norm": 0.06588970124721527, "global_step": 15528, "epoch": 146, "lr": 8.027164536905008e-05} +{"train_loss": 0.01521623507142067, "grad_norm": 0.06209533289074898, "global_step": 15529, "epoch": 146, "lr": 8.026928646481816e-05} +{"train_loss": 0.015179292298853397, "grad_norm": 0.11482520401477814, "global_step": 15530, "epoch": 146, "lr": 8.026692745423314e-05} +{"train_loss": 0.019182566553354263, "grad_norm": 0.054671429097652435, "global_step": 15531, "epoch": 146, "lr": 8.026456833730329e-05} +{"train_loss": 0.014719633385539055, "grad_norm": 0.06348305195569992, "global_step": 15532, "epoch": 146, "lr": 8.026220911403693e-05} +{"train_loss": 0.01696675829589367, "grad_norm": 0.07200899720191956, "global_step": 15533, "epoch": 146, "lr": 8.02598497844423e-05} +{"train_loss": 0.013690058141946793, "grad_norm": 0.0985323041677475, "global_step": 15534, "epoch": 146, "lr": 8.025749034852775e-05} +{"train_loss": 0.01598632149398327, "grad_norm": 0.0805637389421463, "global_step": 15535, "epoch": 146, "lr": 8.025513080630151e-05} +{"train_loss": 0.016362760215997696, "grad_norm": 0.09011167287826538, "global_step": 15536, "epoch": 146, "lr": 8.02527711577719e-05} +{"train_loss": 0.015409494750201702, "grad_norm": 0.05352948233485222, "global_step": 15537, "epoch": 146, "lr": 8.025041140294722e-05} +{"train_loss": 0.0194032434374094, "grad_norm": 0.07620945572853088, "global_step": 15538, "epoch": 146, "lr": 8.024805154183575e-05} +{"train_loss": 0.015803178772330284, "grad_norm": 0.07633653283119202, "global_step": 15539, "epoch": 146, "lr": 8.024569157444577e-05} +{"train_loss": 0.02020477131009102, "grad_norm": 0.0893973857164383, "global_step": 15540, "epoch": 146, "lr": 8.02433315007856e-05} +{"train_loss": 0.01910541206598282, "grad_norm": 0.10032039880752563, "global_step": 15541, "epoch": 146, "lr": 8.02409713208635e-05} +{"train_loss": 0.017369195818901062, "grad_norm": 0.10038214176893234, "global_step": 15542, "epoch": 146, "lr": 8.023861103468777e-05} +{"train_loss": 0.017506718635559082, "grad_norm": 0.08463817834854126, "global_step": 15543, "epoch": 146, "lr": 8.023625064226674e-05} +{"train_loss": 0.017330095171928406, "grad_norm": 0.07088688760995865, "global_step": 15544, "epoch": 146, "lr": 8.023389014360866e-05} +{"train_loss": 0.022043561562895775, "grad_norm": 0.0872805118560791, "global_step": 15545, "epoch": 146, "lr": 8.023152953872183e-05} +{"train_loss": 0.01638081669807434, "grad_norm": 0.07150658965110779, "global_step": 15546, "epoch": 146, "lr": 8.022916882761456e-05} +{"train_loss": 0.016698256134986877, "grad_norm": 0.071920245885849, "global_step": 15547, "epoch": 146, "lr": 8.022680801029513e-05} +{"train_loss": 0.017293747514486313, "grad_norm": 0.09433269500732422, "global_step": 15548, "epoch": 146, "lr": 8.022444708677186e-05} +{"train_loss": 0.01499854400753975, "grad_norm": 0.05470560863614082, "global_step": 15549, "epoch": 146, "lr": 8.022208605705302e-05} +{"train_loss": 0.01584579423069954, "grad_norm": 0.05649244040250778, "global_step": 15550, "epoch": 146, "lr": 8.021972492114691e-05} +{"train_loss": 0.016035333275794983, "grad_norm": 0.06763635575771332, "global_step": 15551, "epoch": 146, "lr": 8.021736367906184e-05} +{"train_loss": 0.016117487102746964, "grad_norm": 0.05813201889395714, "global_step": 15552, "epoch": 146, "lr": 8.021500233080607e-05} +{"train_loss": 0.018187010660767555, "grad_norm": 0.10766960680484772, "global_step": 15553, "epoch": 146, "lr": 8.021264087638794e-05} +{"train_loss": 0.01593656837940216, "grad_norm": 0.07800617814064026, "global_step": 15554, "epoch": 146, "lr": 8.021027931581572e-05} +{"train_loss": 0.017702767625451088, "grad_norm": 0.1083851009607315, "global_step": 15555, "epoch": 146, "lr": 8.020791764909771e-05} +{"train_loss": 0.01627245917916298, "grad_norm": 0.06727831065654755, "global_step": 15556, "epoch": 146, "lr": 8.020555587624223e-05} +{"train_loss": 0.012757092714309692, "grad_norm": 0.07942125201225281, "global_step": 15557, "epoch": 146, "lr": 8.020319399725758e-05} +{"train_loss": 0.015700746327638626, "grad_norm": 0.08032218366861343, "global_step": 15558, "epoch": 146, "lr": 8.0200832012152e-05} +{"train_loss": 0.013979778625071049, "grad_norm": 0.06607367843389511, "global_step": 15559, "epoch": 146, "lr": 8.019846992093385e-05} +{"train_loss": 0.018578898161649704, "grad_norm": 0.07164609432220459, "global_step": 15560, "epoch": 146, "lr": 8.019610772361142e-05} +{"train_loss": 0.018060967326164246, "grad_norm": 0.07891794294118881, "global_step": 15561, "epoch": 146, "lr": 8.019374542019299e-05} +{"train_loss": 0.017837822437286377, "grad_norm": 0.10158240795135498, "global_step": 15562, "epoch": 146, "lr": 8.019138301068687e-05} +{"train_loss": 0.018635621294379234, "grad_norm": 0.10717473924160004, "global_step": 15563, "epoch": 146, "lr": 8.018902049510136e-05} +{"train_loss": 0.0181594081223011, "grad_norm": 0.07423606514930725, "global_step": 15564, "epoch": 146, "lr": 8.018665787344475e-05} +{"train_loss": 0.016783226281404495, "grad_norm": 0.06720635294914246, "global_step": 15565, "epoch": 146, "lr": 8.018429514572537e-05} +{"train_loss": 0.0169549398124218, "grad_norm": 0.11799056082963943, "global_step": 15566, "epoch": 146, "lr": 8.01819323119515e-05} +{"train_loss": 0.016009120270609856, "grad_norm": 0.0449003130197525, "global_step": 15567, "epoch": 146, "lr": 8.017956937213144e-05} +{"train_loss": 0.015153984539210796, "grad_norm": 0.07232243567705154, "global_step": 15568, "epoch": 146, "lr": 8.017720632627349e-05} +{"train_loss": 0.01742824912071228, "grad_norm": 0.07711741328239441, "global_step": 15569, "epoch": 146, "lr": 8.017484317438597e-05} +{"train_loss": 0.0209711454808712, "grad_norm": 0.08191907405853271, "global_step": 15570, "epoch": 146, "lr": 8.017247991647718e-05} +{"train_loss": 0.015230422839522362, "grad_norm": 0.10739496350288391, "global_step": 15571, "epoch": 146, "lr": 8.017011655255541e-05} +{"train_loss": 0.017671607434749603, "grad_norm": 0.07052511721849442, "global_step": 15572, "epoch": 146, "lr": 8.016775308262897e-05} +{"train_loss": 0.014258133247494698, "grad_norm": 0.06007767841219902, "global_step": 15573, "epoch": 146, "lr": 8.016538950670617e-05} +{"train_loss": 0.01475728489458561, "grad_norm": 0.07704613357782364, "global_step": 15574, "epoch": 146, "lr": 8.01630258247953e-05} +{"train_loss": 0.014881014823913574, "grad_norm": 0.04699060320854187, "global_step": 15575, "epoch": 146, "lr": 8.016066203690469e-05} +{"train_loss": 0.018899019807577133, "grad_norm": 0.09478437155485153, "global_step": 15576, "epoch": 146, "lr": 8.01582981430426e-05} +{"train_loss": 0.017470939084887505, "grad_norm": 0.04935599863529205, "global_step": 15577, "epoch": 146, "lr": 8.01559341432174e-05} +{"train_loss": 0.01727524772286415, "grad_norm": 0.05097786337137222, "global_step": 15578, "epoch": 146, "lr": 8.015357003743735e-05} +{"train_loss": 0.01693974994122982, "grad_norm": 0.10639877617359161, "global_step": 15579, "epoch": 146, "lr": 8.015120582571078e-05} +{"train_loss": 0.017523353919386864, "grad_norm": 0.08027303963899612, "global_step": 15580, "epoch": 146, "lr": 8.014884150804596e-05} +{"train_loss": 0.016661018282526505, "grad_norm": 0.14852091670036316, "global_step": 15581, "epoch": 146, "lr": 8.014647708445124e-05, "val_loss": 0.019225982949137688} +{"train_loss": 0.016898950561881065, "grad_norm": 0.06351899355649948, "global_step": 15582, "epoch": 147, "lr": 8.01441125549349e-05} +{"train_loss": 0.017547395080327988, "grad_norm": 0.07418924570083618, "global_step": 15583, "epoch": 147, "lr": 8.014174791950526e-05} +{"train_loss": 0.01783651113510132, "grad_norm": 0.08926481753587723, "global_step": 15584, "epoch": 147, "lr": 8.013938317817064e-05} +{"train_loss": 0.01657668873667717, "grad_norm": 0.07151161134243011, "global_step": 15585, "epoch": 147, "lr": 8.013701833093931e-05} +{"train_loss": 0.01655818521976471, "grad_norm": 0.10199287533760071, "global_step": 15586, "epoch": 147, "lr": 8.013465337781962e-05} +{"train_loss": 0.015036196447908878, "grad_norm": 0.05376719310879707, "global_step": 15587, "epoch": 147, "lr": 8.013228831881987e-05} +{"train_loss": 0.01445761602371931, "grad_norm": 0.08952057361602783, "global_step": 15588, "epoch": 147, "lr": 8.012992315394836e-05} +{"train_loss": 0.016429290175437927, "grad_norm": 0.06623832881450653, "global_step": 15589, "epoch": 147, "lr": 8.01275578832134e-05} +{"train_loss": 0.0154584851115942, "grad_norm": 0.06145423278212547, "global_step": 15590, "epoch": 147, "lr": 8.01251925066233e-05} +{"train_loss": 0.014805050566792488, "grad_norm": 0.043290264904499054, "global_step": 15591, "epoch": 147, "lr": 8.012282702418636e-05} +{"train_loss": 0.01771380752325058, "grad_norm": 0.0759417861700058, "global_step": 15592, "epoch": 147, "lr": 8.012046143591092e-05} +{"train_loss": 0.017729710787534714, "grad_norm": 0.06080564856529236, "global_step": 15593, "epoch": 147, "lr": 8.011809574180529e-05} +{"train_loss": 0.018262788653373718, "grad_norm": 0.07842434197664261, "global_step": 15594, "epoch": 147, "lr": 8.011572994187775e-05} +{"train_loss": 0.015346875414252281, "grad_norm": 0.05320318043231964, "global_step": 15595, "epoch": 147, "lr": 8.011336403613665e-05} +{"train_loss": 0.016996150836348534, "grad_norm": 0.11267668753862381, "global_step": 15596, "epoch": 147, "lr": 8.011099802459027e-05} +{"train_loss": 0.013662343844771385, "grad_norm": 0.05709392577409744, "global_step": 15597, "epoch": 147, "lr": 8.010863190724695e-05} +{"train_loss": 0.014503426849842072, "grad_norm": 0.08062230050563812, "global_step": 15598, "epoch": 147, "lr": 8.010626568411498e-05} +{"train_loss": 0.016971856355667114, "grad_norm": 0.08143087476491928, "global_step": 15599, "epoch": 147, "lr": 8.01038993552027e-05} +{"train_loss": 0.01850416511297226, "grad_norm": 0.06250488013029099, "global_step": 15600, "epoch": 147, "lr": 8.010153292051838e-05} +{"train_loss": 0.01606946624815464, "grad_norm": 0.1195383071899414, "global_step": 15601, "epoch": 147, "lr": 8.00991663800704e-05} +{"train_loss": 0.01947728544473648, "grad_norm": 0.09583029896020889, "global_step": 15602, "epoch": 147, "lr": 8.009679973386701e-05} +{"train_loss": 0.016673961654305458, "grad_norm": 0.09630414843559265, "global_step": 15603, "epoch": 147, "lr": 8.009443298191657e-05} +{"train_loss": 0.01702629029750824, "grad_norm": 0.05599723011255264, "global_step": 15604, "epoch": 147, "lr": 8.009206612422738e-05} +{"train_loss": 0.01680946536362171, "grad_norm": 0.0997367575764656, "global_step": 15605, "epoch": 147, "lr": 8.008969916080775e-05} +{"train_loss": 0.016112932935357094, "grad_norm": 0.06540703028440475, "global_step": 15606, "epoch": 147, "lr": 8.008733209166601e-05} +{"train_loss": 0.01747872307896614, "grad_norm": 0.11014381796121597, "global_step": 15607, "epoch": 147, "lr": 8.008496491681046e-05} +{"train_loss": 0.015471633523702621, "grad_norm": 0.0684114545583725, "global_step": 15608, "epoch": 147, "lr": 8.008259763624945e-05} +{"train_loss": 0.014462505467236042, "grad_norm": 0.08963026106357574, "global_step": 15609, "epoch": 147, "lr": 8.008023024999125e-05} +{"train_loss": 0.0152774378657341, "grad_norm": 0.07247758656740189, "global_step": 15610, "epoch": 147, "lr": 8.00778627580442e-05} +{"train_loss": 0.017947819083929062, "grad_norm": 0.06948196142911911, "global_step": 15611, "epoch": 147, "lr": 8.007549516041664e-05} +{"train_loss": 0.01581803523004055, "grad_norm": 0.09632899612188339, "global_step": 15612, "epoch": 147, "lr": 8.007312745711686e-05} +{"train_loss": 0.018648941069841385, "grad_norm": 0.08167733252048492, "global_step": 15613, "epoch": 147, "lr": 8.007075964815319e-05} +{"train_loss": 0.01680862158536911, "grad_norm": 0.05870034918189049, "global_step": 15614, "epoch": 147, "lr": 8.006839173353395e-05} +{"train_loss": 0.019288647919893265, "grad_norm": 0.11080824583768845, "global_step": 15615, "epoch": 147, "lr": 8.006602371326745e-05} +{"train_loss": 0.015154117718338966, "grad_norm": 0.09636737406253815, "global_step": 15616, "epoch": 147, "lr": 8.006365558736202e-05} +{"train_loss": 0.01871703565120697, "grad_norm": 0.14216262102127075, "global_step": 15617, "epoch": 147, "lr": 8.006128735582599e-05} +{"train_loss": 0.018012989312410355, "grad_norm": 0.16452357172966003, "global_step": 15618, "epoch": 147, "lr": 8.005891901866766e-05} +{"train_loss": 0.015300039201974869, "grad_norm": 0.05924573540687561, "global_step": 15619, "epoch": 147, "lr": 8.005655057589536e-05} +{"train_loss": 0.020127611234784126, "grad_norm": 0.1474229097366333, "global_step": 15620, "epoch": 147, "lr": 8.005418202751741e-05} +{"train_loss": 0.017183836549520493, "grad_norm": 0.10782667249441147, "global_step": 15621, "epoch": 147, "lr": 8.005181337354213e-05} +{"train_loss": 0.018743440508842468, "grad_norm": 0.09683491289615631, "global_step": 15622, "epoch": 147, "lr": 8.004944461397787e-05} +{"train_loss": 0.014811186119914055, "grad_norm": 0.10238410532474518, "global_step": 15623, "epoch": 147, "lr": 8.004707574883291e-05} +{"train_loss": 0.01680378057062626, "grad_norm": 0.1420169174671173, "global_step": 15624, "epoch": 147, "lr": 8.00447067781156e-05} +{"train_loss": 0.015045436099171638, "grad_norm": 0.1362963318824768, "global_step": 15625, "epoch": 147, "lr": 8.004233770183426e-05} +{"train_loss": 0.017918245866894722, "grad_norm": 0.09975625574588776, "global_step": 15626, "epoch": 147, "lr": 8.00399685199972e-05} +{"train_loss": 0.015381970442831516, "grad_norm": 0.10973221808671951, "global_step": 15627, "epoch": 147, "lr": 8.003759923261276e-05} +{"train_loss": 0.017448347061872482, "grad_norm": 0.15381227433681488, "global_step": 15628, "epoch": 147, "lr": 8.003522983968926e-05} +{"train_loss": 0.014826776459813118, "grad_norm": 0.07794557511806488, "global_step": 15629, "epoch": 147, "lr": 8.003286034123503e-05} +{"train_loss": 0.017845366150140762, "grad_norm": 0.05331264063715935, "global_step": 15630, "epoch": 147, "lr": 8.003049073725837e-05} +{"train_loss": 0.0179767943918705, "grad_norm": 0.0845504030585289, "global_step": 15631, "epoch": 147, "lr": 8.002812102776764e-05} +{"train_loss": 0.017355002462863922, "grad_norm": 0.05119920149445534, "global_step": 15632, "epoch": 147, "lr": 8.002575121277115e-05} +{"train_loss": 0.01762661337852478, "grad_norm": 0.08483079075813293, "global_step": 15633, "epoch": 147, "lr": 8.002338129227723e-05} +{"train_loss": 0.013551026582717896, "grad_norm": 0.06577683985233307, "global_step": 15634, "epoch": 147, "lr": 8.002101126629421e-05} +{"train_loss": 0.01574193313717842, "grad_norm": 0.0845702737569809, "global_step": 15635, "epoch": 147, "lr": 8.001864113483041e-05} +{"train_loss": 0.016275443136692047, "grad_norm": 0.06778516620397568, "global_step": 15636, "epoch": 147, "lr": 8.001627089789414e-05} +{"train_loss": 0.014726550318300724, "grad_norm": 0.07287853211164474, "global_step": 15637, "epoch": 147, "lr": 8.001390055549378e-05} +{"train_loss": 0.015442229807376862, "grad_norm": 0.07271910458803177, "global_step": 15638, "epoch": 147, "lr": 8.00115301076376e-05} +{"train_loss": 0.015749886631965637, "grad_norm": 0.12353227287530899, "global_step": 15639, "epoch": 147, "lr": 8.000915955433396e-05} +{"train_loss": 0.01472575031220913, "grad_norm": 0.08387961238622665, "global_step": 15640, "epoch": 147, "lr": 8.000678889559119e-05} +{"train_loss": 0.018404237926006317, "grad_norm": 0.1964273303747177, "global_step": 15641, "epoch": 147, "lr": 8.000441813141761e-05} +{"train_loss": 0.02029721811413765, "grad_norm": 0.09861452877521515, "global_step": 15642, "epoch": 147, "lr": 8.000204726182156e-05} +{"train_loss": 0.016099117696285248, "grad_norm": 0.08718427270650864, "global_step": 15643, "epoch": 147, "lr": 7.999967628681136e-05} +{"train_loss": 0.0163861270993948, "grad_norm": 0.07334358990192413, "global_step": 15644, "epoch": 147, "lr": 7.999730520639534e-05} +{"train_loss": 0.016746748238801956, "grad_norm": 0.11816688627004623, "global_step": 15645, "epoch": 147, "lr": 7.999493402058183e-05} +{"train_loss": 0.015247877687215805, "grad_norm": 0.06762465834617615, "global_step": 15646, "epoch": 147, "lr": 7.999256272937918e-05} +{"train_loss": 0.016755368560552597, "grad_norm": 0.07699494063854218, "global_step": 15647, "epoch": 147, "lr": 7.99901913327957e-05} +{"train_loss": 0.016148582100868225, "grad_norm": 0.08974692225456238, "global_step": 15648, "epoch": 147, "lr": 7.998781983083974e-05} +{"train_loss": 0.01965336501598358, "grad_norm": 0.13030681014060974, "global_step": 15649, "epoch": 147, "lr": 7.998544822351961e-05} +{"train_loss": 0.01601402461528778, "grad_norm": 0.09940244257450104, "global_step": 15650, "epoch": 147, "lr": 7.998307651084366e-05} +{"train_loss": 0.019219782203435898, "grad_norm": 0.15867066383361816, "global_step": 15651, "epoch": 147, "lr": 7.998070469282021e-05} +{"train_loss": 0.015616350807249546, "grad_norm": 0.12425385415554047, "global_step": 15652, "epoch": 147, "lr": 7.997833276945762e-05} +{"train_loss": 0.017256056889891624, "grad_norm": 0.09783095121383667, "global_step": 15653, "epoch": 147, "lr": 7.997596074076419e-05} +{"train_loss": 0.01686650700867176, "grad_norm": 0.05648622289299965, "global_step": 15654, "epoch": 147, "lr": 7.997358860674829e-05} +{"train_loss": 0.017691515386104584, "grad_norm": 0.08925256878137589, "global_step": 15655, "epoch": 147, "lr": 7.997121636741821e-05} +{"train_loss": 0.01894209161400795, "grad_norm": 0.11066841334104538, "global_step": 15656, "epoch": 147, "lr": 7.996884402278232e-05} +{"train_loss": 0.017796020954847336, "grad_norm": 0.15664267539978027, "global_step": 15657, "epoch": 147, "lr": 7.996647157284895e-05} +{"train_loss": 0.016751321032643318, "grad_norm": 0.07987847179174423, "global_step": 15658, "epoch": 147, "lr": 7.996409901762642e-05} +{"train_loss": 0.01633625105023384, "grad_norm": 0.07219519466161728, "global_step": 15659, "epoch": 147, "lr": 7.996172635712309e-05} +{"train_loss": 0.017641820013523102, "grad_norm": 0.06260816752910614, "global_step": 15660, "epoch": 147, "lr": 7.995935359134728e-05} +{"train_loss": 0.016947589814662933, "grad_norm": 0.11269330978393555, "global_step": 15661, "epoch": 147, "lr": 7.995698072030732e-05} +{"train_loss": 0.015690218657255173, "grad_norm": 0.06321538239717484, "global_step": 15662, "epoch": 147, "lr": 7.995460774401157e-05} +{"train_loss": 0.014730757102370262, "grad_norm": 0.054318636655807495, "global_step": 15663, "epoch": 147, "lr": 7.995223466246835e-05} +{"train_loss": 0.015816381201148033, "grad_norm": 0.05658268183469772, "global_step": 15664, "epoch": 147, "lr": 7.994986147568601e-05} +{"train_loss": 0.01608487218618393, "grad_norm": 0.08107619732618332, "global_step": 15665, "epoch": 147, "lr": 7.994748818367287e-05} +{"train_loss": 0.015005585737526417, "grad_norm": 0.06046341732144356, "global_step": 15666, "epoch": 147, "lr": 7.99451147864373e-05} +{"train_loss": 0.01577185094356537, "grad_norm": 0.11554092168807983, "global_step": 15667, "epoch": 147, "lr": 7.99427412839876e-05} +{"train_loss": 0.016433052718639374, "grad_norm": 0.13087861239910126, "global_step": 15668, "epoch": 147, "lr": 7.994036767633213e-05} +{"train_loss": 0.016020677983760834, "grad_norm": 0.056545112282037735, "global_step": 15669, "epoch": 147, "lr": 7.993799396347924e-05} +{"train_loss": 0.017846528440713882, "grad_norm": 0.10185468941926956, "global_step": 15670, "epoch": 147, "lr": 7.993562014543725e-05} +{"train_loss": 0.016702458262443542, "grad_norm": 0.07616417855024338, "global_step": 15671, "epoch": 147, "lr": 7.99332462222145e-05} +{"train_loss": 0.01576709747314453, "grad_norm": 0.07274270802736282, "global_step": 15672, "epoch": 147, "lr": 7.993087219381937e-05} +{"train_loss": 0.01543334685266018, "grad_norm": 0.05657606199383736, "global_step": 15673, "epoch": 147, "lr": 7.992849806026013e-05} +{"train_loss": 0.01815083995461464, "grad_norm": 0.06098625063896179, "global_step": 15674, "epoch": 147, "lr": 7.992612382154519e-05} +{"train_loss": 0.014310995116829872, "grad_norm": 0.048706185072660446, "global_step": 15675, "epoch": 147, "lr": 7.992374947768286e-05} +{"train_loss": 0.015832819044589996, "grad_norm": 0.10292350500822067, "global_step": 15676, "epoch": 147, "lr": 7.99213750286815e-05} +{"train_loss": 0.016450686380267143, "grad_norm": 0.1374068409204483, "global_step": 15677, "epoch": 147, "lr": 7.991900047454943e-05} +{"train_loss": 0.016743095591664314, "grad_norm": 0.051537152379751205, "global_step": 15678, "epoch": 147, "lr": 7.991662581529501e-05} +{"train_loss": 0.015135915949940681, "grad_norm": 0.07380975782871246, "global_step": 15679, "epoch": 147, "lr": 7.991425105092656e-05} +{"train_loss": 0.015264838002622128, "grad_norm": 0.09935985505580902, "global_step": 15680, "epoch": 147, "lr": 7.991187618145247e-05} +{"train_loss": 0.016824036836624146, "grad_norm": 0.09557457268238068, "global_step": 15681, "epoch": 147, "lr": 7.990950120688104e-05} +{"train_loss": 0.017569273710250854, "grad_norm": 0.09641817212104797, "global_step": 15682, "epoch": 147, "lr": 7.990712612722063e-05} +{"train_loss": 0.017282679677009583, "grad_norm": 0.07457853853702545, "global_step": 15683, "epoch": 147, "lr": 7.990475094247957e-05} +{"train_loss": 0.020177818834781647, "grad_norm": 0.06806067377328873, "global_step": 15684, "epoch": 147, "lr": 7.990237565266624e-05} +{"train_loss": 0.017799096181988716, "grad_norm": 0.11589888483285904, "global_step": 15685, "epoch": 147, "lr": 7.990000025778895e-05} +{"train_loss": 0.019239269196987152, "grad_norm": 0.11547740548849106, "global_step": 15686, "epoch": 147, "lr": 7.989762475785607e-05} +{"train_loss": 0.01666376355789461, "grad_norm": 0.10720840096473694, "global_step": 15687, "epoch": 147, "lr": 7.989524915287595e-05, "val_loss": 0.019265033304691315} +{"train_loss": 0.01902102865278721, "grad_norm": 0.10143858939409256, "global_step": 15688, "epoch": 148, "lr": 7.989287344285691e-05} +{"train_loss": 0.0165104903280735, "grad_norm": 0.07710065692663193, "global_step": 15689, "epoch": 148, "lr": 7.989049762780731e-05} +{"train_loss": 0.01737123727798462, "grad_norm": 0.09254027903079987, "global_step": 15690, "epoch": 148, "lr": 7.988812170773552e-05} +{"train_loss": 0.0203544981777668, "grad_norm": 0.1966373473405838, "global_step": 15691, "epoch": 148, "lr": 7.988574568264986e-05} +{"train_loss": 0.017287734895944595, "grad_norm": 0.07743139564990997, "global_step": 15692, "epoch": 148, "lr": 7.988336955255869e-05} +{"train_loss": 0.017466586083173752, "grad_norm": 0.09045074880123138, "global_step": 15693, "epoch": 148, "lr": 7.988099331747035e-05} +{"train_loss": 0.017446555197238922, "grad_norm": 0.11738025397062302, "global_step": 15694, "epoch": 148, "lr": 7.987861697739319e-05} +{"train_loss": 0.015623836778104305, "grad_norm": 0.051777180284261703, "global_step": 15695, "epoch": 148, "lr": 7.987624053233556e-05} +{"train_loss": 0.01683134213089943, "grad_norm": 0.050742655992507935, "global_step": 15696, "epoch": 148, "lr": 7.987386398230582e-05} +{"train_loss": 0.01472637988626957, "grad_norm": 0.10515166819095612, "global_step": 15697, "epoch": 148, "lr": 7.987148732731233e-05} +{"train_loss": 0.01652257703244686, "grad_norm": 0.0970190167427063, "global_step": 15698, "epoch": 148, "lr": 7.986911056736342e-05} +{"train_loss": 0.015214892104268074, "grad_norm": 0.0705285519361496, "global_step": 15699, "epoch": 148, "lr": 7.986673370246742e-05} +{"train_loss": 0.01636502519249916, "grad_norm": 0.08206724375486374, "global_step": 15700, "epoch": 148, "lr": 7.986435673263273e-05} +{"train_loss": 0.01803780533373356, "grad_norm": 0.11499584466218948, "global_step": 15701, "epoch": 148, "lr": 7.986197965786766e-05} +{"train_loss": 0.01721128076314926, "grad_norm": 0.09031068533658981, "global_step": 15702, "epoch": 148, "lr": 7.98596024781806e-05} +{"train_loss": 0.016560621559619904, "grad_norm": 0.10157442837953568, "global_step": 15703, "epoch": 148, "lr": 7.985722519357987e-05} +{"train_loss": 0.016686052083969116, "grad_norm": 0.055869210511446, "global_step": 15704, "epoch": 148, "lr": 7.985484780407384e-05} +{"train_loss": 0.014745371416211128, "grad_norm": 0.06979519873857498, "global_step": 15705, "epoch": 148, "lr": 7.985247030967085e-05} +{"train_loss": 0.015807785093784332, "grad_norm": 0.05234724283218384, "global_step": 15706, "epoch": 148, "lr": 7.985009271037927e-05} +{"train_loss": 0.016331305727362633, "grad_norm": 0.04623784124851227, "global_step": 15707, "epoch": 148, "lr": 7.984771500620745e-05} +{"train_loss": 0.0164184533059597, "grad_norm": 0.07655329257249832, "global_step": 15708, "epoch": 148, "lr": 7.984533719716372e-05} +{"train_loss": 0.017707891762256622, "grad_norm": 0.0787978395819664, "global_step": 15709, "epoch": 148, "lr": 7.984295928325647e-05} +{"train_loss": 0.015020757913589478, "grad_norm": 0.07683157920837402, "global_step": 15710, "epoch": 148, "lr": 7.984058126449405e-05} +{"train_loss": 0.016967378556728363, "grad_norm": 0.06642245501279831, "global_step": 15711, "epoch": 148, "lr": 7.983820314088479e-05} +{"train_loss": 0.016011685132980347, "grad_norm": 0.05119161680340767, "global_step": 15712, "epoch": 148, "lr": 7.983582491243706e-05} +{"train_loss": 0.015860825777053833, "grad_norm": 0.061268698424100876, "global_step": 15713, "epoch": 148, "lr": 7.983344657915922e-05} +{"train_loss": 0.018442846834659576, "grad_norm": 0.0630432665348053, "global_step": 15714, "epoch": 148, "lr": 7.983106814105962e-05} +{"train_loss": 0.01856771484017372, "grad_norm": 0.0896182581782341, "global_step": 15715, "epoch": 148, "lr": 7.982868959814662e-05} +{"train_loss": 0.016335658729076385, "grad_norm": 0.07187357544898987, "global_step": 15716, "epoch": 148, "lr": 7.982631095042858e-05} +{"train_loss": 0.016529353335499763, "grad_norm": 0.08119196444749832, "global_step": 15717, "epoch": 148, "lr": 7.982393219791386e-05} +{"train_loss": 0.01647469401359558, "grad_norm": 0.1342247873544693, "global_step": 15718, "epoch": 148, "lr": 7.982155334061079e-05} +{"train_loss": 0.01770149916410446, "grad_norm": 0.09155658632516861, "global_step": 15719, "epoch": 148, "lr": 7.981917437852777e-05} +{"train_loss": 0.018201828002929688, "grad_norm": 0.05539615824818611, "global_step": 15720, "epoch": 148, "lr": 7.981679531167314e-05} +{"train_loss": 0.017829865217208862, "grad_norm": 0.07816137373447418, "global_step": 15721, "epoch": 148, "lr": 7.981441614005525e-05} +{"train_loss": 0.017903074622154236, "grad_norm": 0.1603674590587616, "global_step": 15722, "epoch": 148, "lr": 7.981203686368248e-05} +{"train_loss": 0.016783669590950012, "grad_norm": 0.08156316727399826, "global_step": 15723, "epoch": 148, "lr": 7.980965748256317e-05} +{"train_loss": 0.01728004775941372, "grad_norm": 0.12296666949987411, "global_step": 15724, "epoch": 148, "lr": 7.980727799670568e-05} +{"train_loss": 0.01422218419611454, "grad_norm": 0.07980311661958694, "global_step": 15725, "epoch": 148, "lr": 7.980489840611839e-05} +{"train_loss": 0.017174845561385155, "grad_norm": 0.05737518519163132, "global_step": 15726, "epoch": 148, "lr": 7.980251871080964e-05} +{"train_loss": 0.014407024718821049, "grad_norm": 0.0688343197107315, "global_step": 15727, "epoch": 148, "lr": 7.98001389107878e-05} +{"train_loss": 0.015042532235383987, "grad_norm": 0.0762384906411171, "global_step": 15728, "epoch": 148, "lr": 7.979775900606123e-05} +{"train_loss": 0.016049962490797043, "grad_norm": 0.09808067232370377, "global_step": 15729, "epoch": 148, "lr": 7.97953789966383e-05} +{"train_loss": 0.016937002539634705, "grad_norm": 0.08198869228363037, "global_step": 15730, "epoch": 148, "lr": 7.979299888252736e-05} +{"train_loss": 0.014510247856378555, "grad_norm": 0.06343754380941391, "global_step": 15731, "epoch": 148, "lr": 7.979061866373678e-05} +{"train_loss": 0.01394526194781065, "grad_norm": 0.10460725426673889, "global_step": 15732, "epoch": 148, "lr": 7.978823834027494e-05} +{"train_loss": 0.01586190052330494, "grad_norm": 0.08809950947761536, "global_step": 15733, "epoch": 148, "lr": 7.978585791215016e-05} +{"train_loss": 0.01493915356695652, "grad_norm": 0.1093914732336998, "global_step": 15734, "epoch": 148, "lr": 7.978347737937085e-05} +{"train_loss": 0.0172332301735878, "grad_norm": 0.05270195007324219, "global_step": 15735, "epoch": 148, "lr": 7.978109674194533e-05} +{"train_loss": 0.016537250950932503, "grad_norm": 0.10613227635622025, "global_step": 15736, "epoch": 148, "lr": 7.9778715999882e-05} +{"train_loss": 0.015929920598864555, "grad_norm": 0.06675267964601517, "global_step": 15737, "epoch": 148, "lr": 7.977633515318921e-05} +{"train_loss": 0.014218633994460106, "grad_norm": 0.10101719200611115, "global_step": 15738, "epoch": 148, "lr": 7.977395420187532e-05} +{"train_loss": 0.018480690196156502, "grad_norm": 0.08669839799404144, "global_step": 15739, "epoch": 148, "lr": 7.977157314594872e-05} +{"train_loss": 0.01647438295185566, "grad_norm": 0.1528545469045639, "global_step": 15740, "epoch": 148, "lr": 7.976919198541776e-05} +{"train_loss": 0.01763613522052765, "grad_norm": 0.09815099090337753, "global_step": 15741, "epoch": 148, "lr": 7.976681072029078e-05} +{"train_loss": 0.01706644520163536, "grad_norm": 0.06564874202013016, "global_step": 15742, "epoch": 148, "lr": 7.97644293505762e-05} +{"train_loss": 0.016584746539592743, "grad_norm": 0.06120327115058899, "global_step": 15743, "epoch": 148, "lr": 7.976204787628234e-05} +{"train_loss": 0.017815660685300827, "grad_norm": 0.1680392622947693, "global_step": 15744, "epoch": 148, "lr": 7.97596662974176e-05} +{"train_loss": 0.017216041684150696, "grad_norm": 0.05946305766701698, "global_step": 15745, "epoch": 148, "lr": 7.975728461399032e-05} +{"train_loss": 0.015285776928067207, "grad_norm": 0.07768474519252777, "global_step": 15746, "epoch": 148, "lr": 7.975490282600891e-05} +{"train_loss": 0.016642140224575996, "grad_norm": 0.22796663641929626, "global_step": 15747, "epoch": 148, "lr": 7.975252093348169e-05} +{"train_loss": 0.019745320081710815, "grad_norm": 0.13380704820156097, "global_step": 15748, "epoch": 148, "lr": 7.975013893641705e-05} +{"train_loss": 0.016625160351395607, "grad_norm": 0.06834046542644501, "global_step": 15749, "epoch": 148, "lr": 7.974775683482338e-05} +{"train_loss": 0.016326334327459335, "grad_norm": 0.12218380719423294, "global_step": 15750, "epoch": 148, "lr": 7.974537462870901e-05} +{"train_loss": 0.015805214643478394, "grad_norm": 0.09257301688194275, "global_step": 15751, "epoch": 148, "lr": 7.974299231808233e-05} +{"train_loss": 0.018259476870298386, "grad_norm": 0.0705493912100792, "global_step": 15752, "epoch": 148, "lr": 7.974060990295172e-05} +{"train_loss": 0.017459778115153313, "grad_norm": 0.09365754574537277, "global_step": 15753, "epoch": 148, "lr": 7.973822738332552e-05} +{"train_loss": 0.01740063540637493, "grad_norm": 0.07715996354818344, "global_step": 15754, "epoch": 148, "lr": 7.973584475921215e-05} +{"train_loss": 0.016635123640298843, "grad_norm": 0.0886465385556221, "global_step": 15755, "epoch": 148, "lr": 7.973346203061993e-05} +{"train_loss": 0.018014896661043167, "grad_norm": 0.05274483188986778, "global_step": 15756, "epoch": 148, "lr": 7.973107919755727e-05} +{"train_loss": 0.016583818942308426, "grad_norm": 0.07596081495285034, "global_step": 15757, "epoch": 148, "lr": 7.972869626003252e-05} +{"train_loss": 0.016625002026557922, "grad_norm": 0.11980164051055908, "global_step": 15758, "epoch": 148, "lr": 7.972631321805406e-05} +{"train_loss": 0.016993127763271332, "grad_norm": 0.0708322674036026, "global_step": 15759, "epoch": 148, "lr": 7.972393007163026e-05} +{"train_loss": 0.0171226616948843, "grad_norm": 0.10708621144294739, "global_step": 15760, "epoch": 148, "lr": 7.972154682076951e-05} +{"train_loss": 0.015920743346214294, "grad_norm": 0.0775115117430687, "global_step": 15761, "epoch": 148, "lr": 7.971916346548017e-05} +{"train_loss": 0.017471790313720703, "grad_norm": 0.05318010225892067, "global_step": 15762, "epoch": 148, "lr": 7.971678000577059e-05} +{"train_loss": 0.015589477494359016, "grad_norm": 0.1210395023226738, "global_step": 15763, "epoch": 148, "lr": 7.971439644164919e-05} +{"train_loss": 0.01656441017985344, "grad_norm": 0.06550317257642746, "global_step": 15764, "epoch": 148, "lr": 7.971201277312433e-05} +{"train_loss": 0.018085666000843048, "grad_norm": 0.05977893993258476, "global_step": 15765, "epoch": 148, "lr": 7.970962900020436e-05} +{"train_loss": 0.01722690835595131, "grad_norm": 0.07925709336996078, "global_step": 15766, "epoch": 148, "lr": 7.970724512289766e-05} +{"train_loss": 0.015855520963668823, "grad_norm": 0.0621330589056015, "global_step": 15767, "epoch": 148, "lr": 7.970486114121265e-05} +{"train_loss": 0.01563296839594841, "grad_norm": 0.04834207519888878, "global_step": 15768, "epoch": 148, "lr": 7.970247705515768e-05} +{"train_loss": 0.01674165017902851, "grad_norm": 0.08129909634590149, "global_step": 15769, "epoch": 148, "lr": 7.970009286474111e-05} +{"train_loss": 0.015867503359913826, "grad_norm": 0.06526760756969452, "global_step": 15770, "epoch": 148, "lr": 7.969770856997131e-05} +{"train_loss": 0.016244664788246155, "grad_norm": 0.0626128762960434, "global_step": 15771, "epoch": 148, "lr": 7.96953241708567e-05} +{"train_loss": 0.017652660608291626, "grad_norm": 0.07470609247684479, "global_step": 15772, "epoch": 148, "lr": 7.969293966740564e-05} +{"train_loss": 0.016762131825089455, "grad_norm": 0.11045926809310913, "global_step": 15773, "epoch": 148, "lr": 7.969055505962648e-05} +{"train_loss": 0.01659705489873886, "grad_norm": 0.08187375217676163, "global_step": 15774, "epoch": 148, "lr": 7.968817034752764e-05} +{"train_loss": 0.016637522727251053, "grad_norm": 0.1098766028881073, "global_step": 15775, "epoch": 148, "lr": 7.968578553111748e-05} +{"train_loss": 0.018592655658721924, "grad_norm": 0.08991248160600662, "global_step": 15776, "epoch": 148, "lr": 7.968340061040438e-05} +{"train_loss": 0.018149208277463913, "grad_norm": 0.06847076863050461, "global_step": 15777, "epoch": 148, "lr": 7.968101558539671e-05} +{"train_loss": 0.0167971421033144, "grad_norm": 0.05651039630174637, "global_step": 15778, "epoch": 148, "lr": 7.967863045610287e-05} +{"train_loss": 0.01686648279428482, "grad_norm": 0.08052786439657211, "global_step": 15779, "epoch": 148, "lr": 7.967624522253123e-05} +{"train_loss": 0.015459971502423286, "grad_norm": 0.07137526571750641, "global_step": 15780, "epoch": 148, "lr": 7.967385988469017e-05} +{"train_loss": 0.01766340434551239, "grad_norm": 0.07189508527517319, "global_step": 15781, "epoch": 148, "lr": 7.967147444258806e-05} +{"train_loss": 0.01632852479815483, "grad_norm": 0.1341068148612976, "global_step": 15782, "epoch": 148, "lr": 7.96690888962333e-05} +{"train_loss": 0.015879835933446884, "grad_norm": 0.10470206290483475, "global_step": 15783, "epoch": 148, "lr": 7.966670324563426e-05} +{"train_loss": 0.015278605744242668, "grad_norm": 0.06755311787128448, "global_step": 15784, "epoch": 148, "lr": 7.966431749079932e-05} +{"train_loss": 0.013767341151833534, "grad_norm": 0.07041477411985397, "global_step": 15785, "epoch": 148, "lr": 7.966193163173688e-05} +{"train_loss": 0.01562625914812088, "grad_norm": 0.0799115002155304, "global_step": 15786, "epoch": 148, "lr": 7.96595456684553e-05} +{"train_loss": 0.014821413904428482, "grad_norm": 0.04890444129705429, "global_step": 15787, "epoch": 148, "lr": 7.965715960096299e-05} +{"train_loss": 0.01642492413520813, "grad_norm": 0.0782897025346756, "global_step": 15788, "epoch": 148, "lr": 7.965477342926831e-05} +{"train_loss": 0.01764512062072754, "grad_norm": 0.04980111122131348, "global_step": 15789, "epoch": 148, "lr": 7.965238715337966e-05} +{"train_loss": 0.014818515628576279, "grad_norm": 0.06895334273576736, "global_step": 15790, "epoch": 148, "lr": 7.965000077330542e-05} +{"train_loss": 0.018772482872009277, "grad_norm": 0.1306227296590805, "global_step": 15791, "epoch": 148, "lr": 7.964761428905397e-05} +{"train_loss": 0.018324509263038635, "grad_norm": 0.054404910653829575, "global_step": 15792, "epoch": 148, "lr": 7.964522770063367e-05} +{"train_loss": 0.016651218414376927, "grad_norm": 0.0839763879776001, "global_step": 15793, "epoch": 148, "lr": 7.964284100805297e-05, "val_loss": 0.019038794562220573} +{"train_loss": 0.01984090358018875, "grad_norm": 0.06434720754623413, "global_step": 15794, "epoch": 149, "lr": 7.96404542113202e-05} +{"train_loss": 0.016549980267882347, "grad_norm": 0.1151772290468216, "global_step": 15795, "epoch": 149, "lr": 7.963806731044377e-05} +{"train_loss": 0.017840759828686714, "grad_norm": 0.1290939301252365, "global_step": 15796, "epoch": 149, "lr": 7.963568030543204e-05} +{"train_loss": 0.01570216938853264, "grad_norm": 0.057500336319208145, "global_step": 15797, "epoch": 149, "lr": 7.963329319629344e-05} +{"train_loss": 0.018905743956565857, "grad_norm": 0.05790140479803085, "global_step": 15798, "epoch": 149, "lr": 7.963090598303634e-05} +{"train_loss": 0.016821105033159256, "grad_norm": 0.08606906235218048, "global_step": 15799, "epoch": 149, "lr": 7.962851866566912e-05} +{"train_loss": 0.01982717774808407, "grad_norm": 0.07570794224739075, "global_step": 15800, "epoch": 149, "lr": 7.962613124420016e-05} +{"train_loss": 0.016508448868989944, "grad_norm": 0.0839989110827446, "global_step": 15801, "epoch": 149, "lr": 7.962374371863786e-05} +{"train_loss": 0.01415087841451168, "grad_norm": 0.08741623163223267, "global_step": 15802, "epoch": 149, "lr": 7.96213560889906e-05} +{"train_loss": 0.014908072538673878, "grad_norm": 0.11241975426673889, "global_step": 15803, "epoch": 149, "lr": 7.961896835526679e-05} +{"train_loss": 0.017524924129247665, "grad_norm": 0.17349779605865479, "global_step": 15804, "epoch": 149, "lr": 7.961658051747481e-05} +{"train_loss": 0.01660195179283619, "grad_norm": 0.0695396438241005, "global_step": 15805, "epoch": 149, "lr": 7.961419257562302e-05} +{"train_loss": 0.015626192092895508, "grad_norm": 0.059119608253240585, "global_step": 15806, "epoch": 149, "lr": 7.961180452971986e-05} +{"train_loss": 0.013955144211649895, "grad_norm": 0.11078803986310959, "global_step": 15807, "epoch": 149, "lr": 7.96094163797737e-05} +{"train_loss": 0.014036353677511215, "grad_norm": 0.09761036187410355, "global_step": 15808, "epoch": 149, "lr": 7.960702812579291e-05} +{"train_loss": 0.01580268144607544, "grad_norm": 0.05250837281346321, "global_step": 15809, "epoch": 149, "lr": 7.96046397677859e-05} +{"train_loss": 0.015755735337734222, "grad_norm": 0.06609786301851273, "global_step": 15810, "epoch": 149, "lr": 7.960225130576107e-05} +{"train_loss": 0.020219922065734863, "grad_norm": 0.06808433681726456, "global_step": 15811, "epoch": 149, "lr": 7.959986273972681e-05} +{"train_loss": 0.0154111934825778, "grad_norm": 0.07518815994262695, "global_step": 15812, "epoch": 149, "lr": 7.959747406969149e-05} +{"train_loss": 0.017959225922822952, "grad_norm": 0.08165746927261353, "global_step": 15813, "epoch": 149, "lr": 7.959508529566353e-05} +{"train_loss": 0.01508263684809208, "grad_norm": 0.05411854758858681, "global_step": 15814, "epoch": 149, "lr": 7.95926964176513e-05} +{"train_loss": 0.01818181946873665, "grad_norm": 0.08199724555015564, "global_step": 15815, "epoch": 149, "lr": 7.95903074356632e-05} +{"train_loss": 0.014892254024744034, "grad_norm": 0.12615354359149933, "global_step": 15816, "epoch": 149, "lr": 7.958791834970765e-05} +{"train_loss": 0.01621291972696781, "grad_norm": 0.08562182635068893, "global_step": 15817, "epoch": 149, "lr": 7.9585529159793e-05} +{"train_loss": 0.016683679074048996, "grad_norm": 0.1467093676328659, "global_step": 15818, "epoch": 149, "lr": 7.958313986592766e-05} +{"train_loss": 0.017658382654190063, "grad_norm": 0.049417492002248764, "global_step": 15819, "epoch": 149, "lr": 7.958075046812005e-05} +{"train_loss": 0.015606855042278767, "grad_norm": 0.06615152955055237, "global_step": 15820, "epoch": 149, "lr": 7.957836096637856e-05} +{"train_loss": 0.015776462852954865, "grad_norm": 0.07647343724966049, "global_step": 15821, "epoch": 149, "lr": 7.957597136071155e-05} +{"train_loss": 0.017304737120866776, "grad_norm": 0.06644460558891296, "global_step": 15822, "epoch": 149, "lr": 7.957358165112746e-05} +{"train_loss": 0.017070647329092026, "grad_norm": 0.05897718667984009, "global_step": 15823, "epoch": 149, "lr": 7.957119183763465e-05} +{"train_loss": 0.01674739271402359, "grad_norm": 0.052141278982162476, "global_step": 15824, "epoch": 149, "lr": 7.956880192024154e-05} +{"train_loss": 0.015888657420873642, "grad_norm": 0.059445254504680634, "global_step": 15825, "epoch": 149, "lr": 7.95664118989565e-05} +{"train_loss": 0.01743953488767147, "grad_norm": 0.05907637998461723, "global_step": 15826, "epoch": 149, "lr": 7.956402177378797e-05} +{"train_loss": 0.0171666257083416, "grad_norm": 0.05832933634519577, "global_step": 15827, "epoch": 149, "lr": 7.95616315447443e-05} +{"train_loss": 0.016972288489341736, "grad_norm": 0.07023866474628448, "global_step": 15828, "epoch": 149, "lr": 7.955924121183393e-05} +{"train_loss": 0.01777631603181362, "grad_norm": 0.04251303896307945, "global_step": 15829, "epoch": 149, "lr": 7.955685077506523e-05} +{"train_loss": 0.017166275531053543, "grad_norm": 0.07764230668544769, "global_step": 15830, "epoch": 149, "lr": 7.955446023444664e-05} +{"train_loss": 0.016166632995009422, "grad_norm": 0.16875426471233368, "global_step": 15831, "epoch": 149, "lr": 7.95520695899865e-05} +{"train_loss": 0.01882350444793701, "grad_norm": 0.12352903187274933, "global_step": 15832, "epoch": 149, "lr": 7.954967884169325e-05} +{"train_loss": 0.017520323395729065, "grad_norm": 0.098590187728405, "global_step": 15833, "epoch": 149, "lr": 7.954728798957526e-05} +{"train_loss": 0.017922870814800262, "grad_norm": 0.11739017069339752, "global_step": 15834, "epoch": 149, "lr": 7.954489703364098e-05} +{"train_loss": 0.014890830963850021, "grad_norm": 0.1119265928864479, "global_step": 15835, "epoch": 149, "lr": 7.954250597389876e-05} +{"train_loss": 0.017063088715076447, "grad_norm": 0.07283028960227966, "global_step": 15836, "epoch": 149, "lr": 7.954011481035704e-05} +{"train_loss": 0.018845926970243454, "grad_norm": 0.0829939916729927, "global_step": 15837, "epoch": 149, "lr": 7.95377235430242e-05} +{"train_loss": 0.01684042438864708, "grad_norm": 0.07837430387735367, "global_step": 15838, "epoch": 149, "lr": 7.953533217190862e-05} +{"train_loss": 0.016223525628447533, "grad_norm": 0.051381174474954605, "global_step": 15839, "epoch": 149, "lr": 7.953294069701876e-05} +{"train_loss": 0.0161452479660511, "grad_norm": 0.0651664063334465, "global_step": 15840, "epoch": 149, "lr": 7.953054911836297e-05} +{"train_loss": 0.01765131950378418, "grad_norm": 0.13802702724933624, "global_step": 15841, "epoch": 149, "lr": 7.952815743594967e-05} +{"train_loss": 0.015426769852638245, "grad_norm": 0.12689000368118286, "global_step": 15842, "epoch": 149, "lr": 7.952576564978727e-05} +{"train_loss": 0.019454877823591232, "grad_norm": 0.1373690515756607, "global_step": 15843, "epoch": 149, "lr": 7.952337375988418e-05} +{"train_loss": 0.014879552647471428, "grad_norm": 0.1355920284986496, "global_step": 15844, "epoch": 149, "lr": 7.952098176624878e-05} +{"train_loss": 0.01536103617399931, "grad_norm": 0.2326546460390091, "global_step": 15845, "epoch": 149, "lr": 7.95185896688895e-05} +{"train_loss": 0.017938390374183655, "grad_norm": 0.14646905660629272, "global_step": 15846, "epoch": 149, "lr": 7.951619746781474e-05} +{"train_loss": 0.015894975513219833, "grad_norm": 0.10994382202625275, "global_step": 15847, "epoch": 149, "lr": 7.951380516303288e-05} +{"train_loss": 0.018161434680223465, "grad_norm": 0.05867842212319374, "global_step": 15848, "epoch": 149, "lr": 7.951141275455234e-05} +{"train_loss": 0.017547521740198135, "grad_norm": 0.17123375833034515, "global_step": 15849, "epoch": 149, "lr": 7.950902024238156e-05} +{"train_loss": 0.013988243415951729, "grad_norm": 0.08503568917512894, "global_step": 15850, "epoch": 149, "lr": 7.95066276265289e-05} +{"train_loss": 0.01515333354473114, "grad_norm": 0.10956937074661255, "global_step": 15851, "epoch": 149, "lr": 7.950423490700276e-05} +{"train_loss": 0.017710909247398376, "grad_norm": 0.14362362027168274, "global_step": 15852, "epoch": 149, "lr": 7.95018420838116e-05} +{"train_loss": 0.016611695289611816, "grad_norm": 0.07832341641187668, "global_step": 15853, "epoch": 149, "lr": 7.949944915696378e-05} +{"train_loss": 0.01772044599056244, "grad_norm": 0.07221738994121552, "global_step": 15854, "epoch": 149, "lr": 7.949705612646773e-05} +{"train_loss": 0.015864547342061996, "grad_norm": 0.05368373170495033, "global_step": 15855, "epoch": 149, "lr": 7.949466299233186e-05} +{"train_loss": 0.013343947008252144, "grad_norm": 0.061163924634456635, "global_step": 15856, "epoch": 149, "lr": 7.949226975456457e-05} +{"train_loss": 0.016692159697413445, "grad_norm": 0.10574477910995483, "global_step": 15857, "epoch": 149, "lr": 7.948987641317426e-05} +{"train_loss": 0.013902678154408932, "grad_norm": 0.04513666033744812, "global_step": 15858, "epoch": 149, "lr": 7.948748296816934e-05} +{"train_loss": 0.020658625289797783, "grad_norm": 0.08522357046604156, "global_step": 15859, "epoch": 149, "lr": 7.948508941955825e-05} +{"train_loss": 0.015412175096571445, "grad_norm": 0.06023137643933296, "global_step": 15860, "epoch": 149, "lr": 7.948269576734938e-05} +{"train_loss": 0.016613293439149857, "grad_norm": 0.06028330326080322, "global_step": 15861, "epoch": 149, "lr": 7.948030201155112e-05} +{"train_loss": 0.015710115432739258, "grad_norm": 0.13782146573066711, "global_step": 15862, "epoch": 149, "lr": 7.947790815217191e-05} +{"train_loss": 0.014982002787292004, "grad_norm": 0.06631067395210266, "global_step": 15863, "epoch": 149, "lr": 7.947551418922015e-05} +{"train_loss": 0.017248572781682014, "grad_norm": 0.0788876935839653, "global_step": 15864, "epoch": 149, "lr": 7.947312012270425e-05} +{"train_loss": 0.014295912347733974, "grad_norm": 0.07505512982606888, "global_step": 15865, "epoch": 149, "lr": 7.947072595263263e-05} +{"train_loss": 0.018132522702217102, "grad_norm": 0.10303477942943573, "global_step": 15866, "epoch": 149, "lr": 7.946833167901369e-05} +{"train_loss": 0.020291779190301895, "grad_norm": 0.1195981577038765, "global_step": 15867, "epoch": 149, "lr": 7.946593730185585e-05} +{"train_loss": 0.015887191519141197, "grad_norm": 0.05634082481265068, "global_step": 15868, "epoch": 149, "lr": 7.946354282116751e-05} +{"train_loss": 0.015612978488206863, "grad_norm": 0.05198762193322182, "global_step": 15869, "epoch": 149, "lr": 7.946114823695712e-05} +{"train_loss": 0.01807757467031479, "grad_norm": 0.10135039687156677, "global_step": 15870, "epoch": 149, "lr": 7.945875354923305e-05} +{"train_loss": 0.017655588686466217, "grad_norm": 0.1336306631565094, "global_step": 15871, "epoch": 149, "lr": 7.945635875800373e-05} +{"train_loss": 0.015109308063983917, "grad_norm": 0.05342321842908859, "global_step": 15872, "epoch": 149, "lr": 7.945396386327758e-05} +{"train_loss": 0.01544990111142397, "grad_norm": 0.16563354432582855, "global_step": 15873, "epoch": 149, "lr": 7.9451568865063e-05} +{"train_loss": 0.014992504380643368, "grad_norm": 0.09005855023860931, "global_step": 15874, "epoch": 149, "lr": 7.944917376336842e-05} +{"train_loss": 0.016908789053559303, "grad_norm": 0.07739376276731491, "global_step": 15875, "epoch": 149, "lr": 7.944677855820225e-05} +{"train_loss": 0.017317062243819237, "grad_norm": 0.14562121033668518, "global_step": 15876, "epoch": 149, "lr": 7.944438324957291e-05} +{"train_loss": 0.016413342207670212, "grad_norm": 0.08122996985912323, "global_step": 15877, "epoch": 149, "lr": 7.944198783748881e-05} +{"train_loss": 0.01460764929652214, "grad_norm": 0.1364918351173401, "global_step": 15878, "epoch": 149, "lr": 7.943959232195836e-05} +{"train_loss": 0.017990389838814735, "grad_norm": 0.0700472742319107, "global_step": 15879, "epoch": 149, "lr": 7.943719670299002e-05} +{"train_loss": 0.016170095652341843, "grad_norm": 0.05937344208359718, "global_step": 15880, "epoch": 149, "lr": 7.943480098059213e-05} +{"train_loss": 0.014441650360822678, "grad_norm": 0.06117934733629227, "global_step": 15881, "epoch": 149, "lr": 7.943240515477317e-05} +{"train_loss": 0.01616167649626732, "grad_norm": 0.07758428156375885, "global_step": 15882, "epoch": 149, "lr": 7.943000922554153e-05} +{"train_loss": 0.01589914597570896, "grad_norm": 0.058443583548069, "global_step": 15883, "epoch": 149, "lr": 7.942761319290564e-05} +{"train_loss": 0.01823880337178707, "grad_norm": 0.06323850154876709, "global_step": 15884, "epoch": 149, "lr": 7.942521705687393e-05} +{"train_loss": 0.01661158725619316, "grad_norm": 0.054400619119405746, "global_step": 15885, "epoch": 149, "lr": 7.942282081745479e-05} +{"train_loss": 0.019594989717006683, "grad_norm": 0.20999331772327423, "global_step": 15886, "epoch": 149, "lr": 7.942042447465665e-05} +{"train_loss": 0.015304725617170334, "grad_norm": 0.0421888530254364, "global_step": 15887, "epoch": 149, "lr": 7.941802802848794e-05} +{"train_loss": 0.018080102279782295, "grad_norm": 0.08877506852149963, "global_step": 15888, "epoch": 149, "lr": 7.941563147895708e-05} +{"train_loss": 0.01583603397011757, "grad_norm": 0.06578280031681061, "global_step": 15889, "epoch": 149, "lr": 7.941323482607247e-05} +{"train_loss": 0.014325099065899849, "grad_norm": 0.0569206103682518, "global_step": 15890, "epoch": 149, "lr": 7.941083806984256e-05} +{"train_loss": 0.018628714606165886, "grad_norm": 0.06387020647525787, "global_step": 15891, "epoch": 149, "lr": 7.940844121027574e-05} +{"train_loss": 0.015752239152789116, "grad_norm": 0.06721898168325424, "global_step": 15892, "epoch": 149, "lr": 7.940604424738044e-05} +{"train_loss": 0.013804487884044647, "grad_norm": 0.08628324419260025, "global_step": 15893, "epoch": 149, "lr": 7.94036471811651e-05} +{"train_loss": 0.017470210790634155, "grad_norm": 0.07157038897275925, "global_step": 15894, "epoch": 149, "lr": 7.940125001163814e-05} +{"train_loss": 0.01885955035686493, "grad_norm": 0.159134179353714, "global_step": 15895, "epoch": 149, "lr": 7.939885273880797e-05} +{"train_loss": 0.016763832420110703, "grad_norm": 0.06942916661500931, "global_step": 15896, "epoch": 149, "lr": 7.939645536268301e-05} +{"train_loss": 0.01750786229968071, "grad_norm": 0.08631964027881622, "global_step": 15897, "epoch": 149, "lr": 7.939405788327168e-05} +{"train_loss": 0.014856096357107162, "grad_norm": 0.10421308130025864, "global_step": 15898, "epoch": 149, "lr": 7.939166030058243e-05} +{"train_loss": 0.01661098281027011, "grad_norm": 0.08463706076145172, "global_step": 15899, "epoch": 149, "lr": 7.938926261462366e-05, "val_loss": 0.01912287063896656} +{"train_loss": 0.017352141439914703, "grad_norm": 0.12090378999710083, "global_step": 15900, "epoch": 150, "lr": 7.938686482540381e-05} +{"train_loss": 0.015496132895350456, "grad_norm": 0.08266468346118927, "global_step": 15901, "epoch": 150, "lr": 7.93844669329313e-05} +{"train_loss": 0.01794036105275154, "grad_norm": 0.08827130496501923, "global_step": 15902, "epoch": 150, "lr": 7.938206893721453e-05} +{"train_loss": 0.015981316566467285, "grad_norm": 0.06354987621307373, "global_step": 15903, "epoch": 150, "lr": 7.937967083826196e-05} +{"train_loss": 0.014117725193500519, "grad_norm": 0.061013635247945786, "global_step": 15904, "epoch": 150, "lr": 7.937727263608201e-05} +{"train_loss": 0.015102801844477654, "grad_norm": 0.06456788629293442, "global_step": 15905, "epoch": 150, "lr": 7.937487433068309e-05} +{"train_loss": 0.014240719377994537, "grad_norm": 0.07724017649888992, "global_step": 15906, "epoch": 150, "lr": 7.937247592207365e-05} +{"train_loss": 0.016290584579110146, "grad_norm": 0.05355500802397728, "global_step": 15907, "epoch": 150, "lr": 7.93700774102621e-05} +{"train_loss": 0.015299681574106216, "grad_norm": 0.11427157372236252, "global_step": 15908, "epoch": 150, "lr": 7.936767879525686e-05} +{"train_loss": 0.01600959524512291, "grad_norm": 0.1286180019378662, "global_step": 15909, "epoch": 150, "lr": 7.936528007706637e-05} +{"train_loss": 0.015300830826163292, "grad_norm": 0.11896481364965439, "global_step": 15910, "epoch": 150, "lr": 7.936288125569906e-05} +{"train_loss": 0.018600523471832275, "grad_norm": 0.13016609847545624, "global_step": 15911, "epoch": 150, "lr": 7.936048233116336e-05} +{"train_loss": 0.0167239923030138, "grad_norm": 0.06237422302365303, "global_step": 15912, "epoch": 150, "lr": 7.935808330346769e-05} +{"train_loss": 0.01721920818090439, "grad_norm": 0.06250700354576111, "global_step": 15913, "epoch": 150, "lr": 7.93556841726205e-05} +{"train_loss": 0.014197110198438168, "grad_norm": 0.06461308151483536, "global_step": 15914, "epoch": 150, "lr": 7.935328493863017e-05} +{"train_loss": 0.016024205833673477, "grad_norm": 0.06317409127950668, "global_step": 15915, "epoch": 150, "lr": 7.935088560150517e-05} +{"train_loss": 0.01728963851928711, "grad_norm": 0.05811559781432152, "global_step": 15916, "epoch": 150, "lr": 7.934848616125394e-05} +{"train_loss": 0.015913700684905052, "grad_norm": 0.06317491084337234, "global_step": 15917, "epoch": 150, "lr": 7.934608661788487e-05} +{"train_loss": 0.013373693451285362, "grad_norm": 0.09634526818990707, "global_step": 15918, "epoch": 150, "lr": 7.934368697140644e-05} +{"train_loss": 0.018755994737148285, "grad_norm": 0.10739414393901825, "global_step": 15919, "epoch": 150, "lr": 7.934128722182704e-05} +{"train_loss": 0.0188252292573452, "grad_norm": 0.13069888949394226, "global_step": 15920, "epoch": 150, "lr": 7.933888736915512e-05} +{"train_loss": 0.015009051188826561, "grad_norm": 0.13475604355335236, "global_step": 15921, "epoch": 150, "lr": 7.93364874133991e-05} +{"train_loss": 0.020024240016937256, "grad_norm": 0.15869736671447754, "global_step": 15922, "epoch": 150, "lr": 7.933408735456742e-05} +{"train_loss": 0.015736669301986694, "grad_norm": 0.09085045754909515, "global_step": 15923, "epoch": 150, "lr": 7.933168719266852e-05} +{"train_loss": 0.016385424882173538, "grad_norm": 0.07963984459638596, "global_step": 15924, "epoch": 150, "lr": 7.932928692771084e-05} +{"train_loss": 0.014765679836273193, "grad_norm": 0.1351587325334549, "global_step": 15925, "epoch": 150, "lr": 7.93268865597028e-05} +{"train_loss": 0.015871047973632812, "grad_norm": 0.10372991859912872, "global_step": 15926, "epoch": 150, "lr": 7.932448608865281e-05} +{"train_loss": 0.016067352145910263, "grad_norm": 0.06317263096570969, "global_step": 15927, "epoch": 150, "lr": 7.932208551456934e-05} +{"train_loss": 0.017336364835500717, "grad_norm": 0.051155127584934235, "global_step": 15928, "epoch": 150, "lr": 7.931968483746082e-05} +{"train_loss": 0.018348269164562225, "grad_norm": 0.10235118865966797, "global_step": 15929, "epoch": 150, "lr": 7.931728405733568e-05} +{"train_loss": 0.016021646559238434, "grad_norm": 0.06514134258031845, "global_step": 15930, "epoch": 150, "lr": 7.931488317420234e-05} +{"train_loss": 0.015270266681909561, "grad_norm": 0.061963461339473724, "global_step": 15931, "epoch": 150, "lr": 7.931248218806927e-05} +{"train_loss": 0.018208956345915794, "grad_norm": 0.05941561982035637, "global_step": 15932, "epoch": 150, "lr": 7.931008109894487e-05} +{"train_loss": 0.014798101037740707, "grad_norm": 0.08678886294364929, "global_step": 15933, "epoch": 150, "lr": 7.930767990683758e-05} +{"train_loss": 0.018360009416937828, "grad_norm": 0.06373231112957001, "global_step": 15934, "epoch": 150, "lr": 7.930527861175587e-05} +{"train_loss": 0.015569929033517838, "grad_norm": 0.06551872938871384, "global_step": 15935, "epoch": 150, "lr": 7.930287721370815e-05} +{"train_loss": 0.018648436293005943, "grad_norm": 0.17268823087215424, "global_step": 15936, "epoch": 150, "lr": 7.930047571270288e-05} +{"train_loss": 0.015953429043293, "grad_norm": 0.13047051429748535, "global_step": 15937, "epoch": 150, "lr": 7.929807410874846e-05} +{"train_loss": 0.015411429107189178, "grad_norm": 0.05019752308726311, "global_step": 15938, "epoch": 150, "lr": 7.929567240185333e-05} +{"train_loss": 0.015874845907092094, "grad_norm": 0.06627301871776581, "global_step": 15939, "epoch": 150, "lr": 7.929327059202599e-05} +{"train_loss": 0.020647482946515083, "grad_norm": 0.17369017004966736, "global_step": 15940, "epoch": 150, "lr": 7.92908686792748e-05} +{"train_loss": 0.01627441495656967, "grad_norm": 0.13971230387687683, "global_step": 15941, "epoch": 150, "lr": 7.928846666360825e-05} +{"train_loss": 0.015622870996594429, "grad_norm": 0.06427495181560516, "global_step": 15942, "epoch": 150, "lr": 7.928606454503476e-05} +{"train_loss": 0.016688231378793716, "grad_norm": 0.07731201499700546, "global_step": 15943, "epoch": 150, "lr": 7.92836623235628e-05} +{"train_loss": 0.017711594700813293, "grad_norm": 0.08596212416887283, "global_step": 15944, "epoch": 150, "lr": 7.928125999920075e-05} +{"train_loss": 0.01566343382000923, "grad_norm": 0.045627471059560776, "global_step": 15945, "epoch": 150, "lr": 7.927885757195713e-05} +{"train_loss": 0.018504051491618156, "grad_norm": 0.055058833211660385, "global_step": 15946, "epoch": 150, "lr": 7.927645504184029e-05} +{"train_loss": 0.01730111613869667, "grad_norm": 0.09262937307357788, "global_step": 15947, "epoch": 150, "lr": 7.927405240885875e-05} +{"train_loss": 0.01575496420264244, "grad_norm": 0.06847973167896271, "global_step": 15948, "epoch": 150, "lr": 7.92716496730209e-05} +{"train_loss": 0.018753662705421448, "grad_norm": 0.08860165625810623, "global_step": 15949, "epoch": 150, "lr": 7.926924683433523e-05} +{"train_loss": 0.015753205865621567, "grad_norm": 0.11041653901338577, "global_step": 15950, "epoch": 150, "lr": 7.926684389281013e-05} +{"train_loss": 0.01590333878993988, "grad_norm": 0.06753582507371902, "global_step": 15951, "epoch": 150, "lr": 7.926444084845407e-05} +{"train_loss": 0.017928119748830795, "grad_norm": 0.11519672721624374, "global_step": 15952, "epoch": 150, "lr": 7.926203770127552e-05} +{"train_loss": 0.019029807299375534, "grad_norm": 0.24213382601737976, "global_step": 15953, "epoch": 150, "lr": 7.925963445128286e-05} +{"train_loss": 0.01915733516216278, "grad_norm": 0.07225301116704941, "global_step": 15954, "epoch": 150, "lr": 7.925723109848459e-05} +{"train_loss": 0.019237548112869263, "grad_norm": 0.058861665427684784, "global_step": 15955, "epoch": 150, "lr": 7.925482764288913e-05} +{"train_loss": 0.017337383702397346, "grad_norm": 0.0499458983540535, "global_step": 15956, "epoch": 150, "lr": 7.925242408450492e-05} +{"train_loss": 0.014802491292357445, "grad_norm": 0.11041522026062012, "global_step": 15957, "epoch": 150, "lr": 7.925002042334042e-05} +{"train_loss": 0.017912359908223152, "grad_norm": 0.06890522688627243, "global_step": 15958, "epoch": 150, "lr": 7.924761665940408e-05} +{"train_loss": 0.016943760216236115, "grad_norm": 0.15214379131793976, "global_step": 15959, "epoch": 150, "lr": 7.924521279270432e-05} +{"train_loss": 0.015107063576579094, "grad_norm": 0.09527136385440826, "global_step": 15960, "epoch": 150, "lr": 7.92428088232496e-05} +{"train_loss": 0.01760711707174778, "grad_norm": 0.15395259857177734, "global_step": 15961, "epoch": 150, "lr": 7.924040475104837e-05} +{"train_loss": 0.016226498410105705, "grad_norm": 0.07630211114883423, "global_step": 15962, "epoch": 150, "lr": 7.923800057610906e-05} +{"train_loss": 0.016100887209177017, "grad_norm": 0.08461310714483261, "global_step": 15963, "epoch": 150, "lr": 7.923559629844015e-05} +{"train_loss": 0.013614925555884838, "grad_norm": 0.059787675738334656, "global_step": 15964, "epoch": 150, "lr": 7.923319191805007e-05} +{"train_loss": 0.01607266068458557, "grad_norm": 0.05509425327181816, "global_step": 15965, "epoch": 150, "lr": 7.923078743494726e-05} +{"train_loss": 0.015107985585927963, "grad_norm": 0.09733528643846512, "global_step": 15966, "epoch": 150, "lr": 7.922838284914017e-05} +{"train_loss": 0.014915840700268745, "grad_norm": 0.07593023777008057, "global_step": 15967, "epoch": 150, "lr": 7.922597816063726e-05} +{"train_loss": 0.016595927998423576, "grad_norm": 0.07405632734298706, "global_step": 15968, "epoch": 150, "lr": 7.922357336944697e-05} +{"train_loss": 0.015873562544584274, "grad_norm": 0.0728171095252037, "global_step": 15969, "epoch": 150, "lr": 7.922116847557777e-05} +{"train_loss": 0.014644350856542587, "grad_norm": 0.06344177573919296, "global_step": 15970, "epoch": 150, "lr": 7.921876347903807e-05} +{"train_loss": 0.016643457114696503, "grad_norm": 0.10417237132787704, "global_step": 15971, "epoch": 150, "lr": 7.921635837983634e-05} +{"train_loss": 0.017403097823262215, "grad_norm": 0.1305711269378662, "global_step": 15972, "epoch": 150, "lr": 7.921395317798106e-05} +{"train_loss": 0.013312235474586487, "grad_norm": 0.0929780974984169, "global_step": 15973, "epoch": 150, "lr": 7.921154787348063e-05} +{"train_loss": 0.015675632283091545, "grad_norm": 0.05873344466090202, "global_step": 15974, "epoch": 150, "lr": 7.920914246634353e-05} +{"train_loss": 0.018121354281902313, "grad_norm": 0.09073230624198914, "global_step": 15975, "epoch": 150, "lr": 7.920673695657821e-05} +{"train_loss": 0.01796516217291355, "grad_norm": 0.10539557784795761, "global_step": 15976, "epoch": 150, "lr": 7.92043313441931e-05} +{"train_loss": 0.01609266921877861, "grad_norm": 0.07542980462312698, "global_step": 15977, "epoch": 150, "lr": 7.920192562919668e-05} +{"train_loss": 0.014583799988031387, "grad_norm": 0.13468457758426666, "global_step": 15978, "epoch": 150, "lr": 7.919951981159741e-05} +{"train_loss": 0.016838936135172844, "grad_norm": 0.068378746509552, "global_step": 15979, "epoch": 150, "lr": 7.919711389140371e-05} +{"train_loss": 0.017056694254279137, "grad_norm": 0.051189836114645004, "global_step": 15980, "epoch": 150, "lr": 7.919470786862406e-05} +{"train_loss": 0.017526676878333092, "grad_norm": 0.10295840352773666, "global_step": 15981, "epoch": 150, "lr": 7.919230174326688e-05} +{"train_loss": 0.017087094485759735, "grad_norm": 0.12982824444770813, "global_step": 15982, "epoch": 150, "lr": 7.918989551534066e-05} +{"train_loss": 0.01521739736199379, "grad_norm": 0.11055649071931839, "global_step": 15983, "epoch": 150, "lr": 7.918748918485385e-05} +{"train_loss": 0.015782497823238373, "grad_norm": 0.11146382987499237, "global_step": 15984, "epoch": 150, "lr": 7.918508275181489e-05} +{"train_loss": 0.013336185365915298, "grad_norm": 0.04990207403898239, "global_step": 15985, "epoch": 150, "lr": 7.918267621623224e-05} +{"train_loss": 0.017369408160448074, "grad_norm": 0.06406361609697342, "global_step": 15986, "epoch": 150, "lr": 7.918026957811436e-05} +{"train_loss": 0.016179271042346954, "grad_norm": 0.08117120712995529, "global_step": 15987, "epoch": 150, "lr": 7.91778628374697e-05} +{"train_loss": 0.019170818850398064, "grad_norm": 0.06854742020368576, "global_step": 15988, "epoch": 150, "lr": 7.917545599430671e-05} +{"train_loss": 0.016590699553489685, "grad_norm": 0.07217554748058319, "global_step": 15989, "epoch": 150, "lr": 7.917304904863387e-05} +{"train_loss": 0.017489532008767128, "grad_norm": 0.09040592610836029, "global_step": 15990, "epoch": 150, "lr": 7.917064200045963e-05} +{"train_loss": 0.01912672072649002, "grad_norm": 0.09057233482599258, "global_step": 15991, "epoch": 150, "lr": 7.916823484979242e-05} +{"train_loss": 0.016759535297751427, "grad_norm": 0.05791878327727318, "global_step": 15992, "epoch": 150, "lr": 7.916582759664073e-05} +{"train_loss": 0.018666956573724747, "grad_norm": 0.09881441295146942, "global_step": 15993, "epoch": 150, "lr": 7.9163420241013e-05} +{"train_loss": 0.018086563795804977, "grad_norm": 0.10541315376758575, "global_step": 15994, "epoch": 150, "lr": 7.916101278291769e-05} +{"train_loss": 0.018336504697799683, "grad_norm": 0.05749259889125824, "global_step": 15995, "epoch": 150, "lr": 7.915860522236326e-05} +{"train_loss": 0.017600562423467636, "grad_norm": 0.07924985885620117, "global_step": 15996, "epoch": 150, "lr": 7.91561975593582e-05} +{"train_loss": 0.014967766590416431, "grad_norm": 0.05206665024161339, "global_step": 15997, "epoch": 150, "lr": 7.915378979391091e-05} +{"train_loss": 0.0154497679322958, "grad_norm": 0.06534536182880402, "global_step": 15998, "epoch": 150, "lr": 7.91513819260299e-05} +{"train_loss": 0.01634797640144825, "grad_norm": 0.06985964626073837, "global_step": 15999, "epoch": 150, "lr": 7.91489739557236e-05} +{"train_loss": 0.015361460857093334, "grad_norm": 0.050339747220277786, "global_step": 16000, "epoch": 150, "lr": 7.914656588300048e-05} +{"train_loss": 0.017821207642555237, "grad_norm": 0.10613031685352325, "global_step": 16001, "epoch": 150, "lr": 7.914415770786902e-05} +{"train_loss": 0.015599481761455536, "grad_norm": 0.13475286960601807, "global_step": 16002, "epoch": 150, "lr": 7.914174943033766e-05} +{"train_loss": 0.021915948018431664, "grad_norm": 0.1001172661781311, "global_step": 16003, "epoch": 150, "lr": 7.913934105041487e-05} +{"train_loss": 0.017733417451381683, "grad_norm": 0.1101122498512268, "global_step": 16004, "epoch": 150, "lr": 7.913693256810908e-05} +{"train_loss": 0.01660347291496846, "grad_norm": 0.06278180330991745, "global_step": 16005, "epoch": 150, "lr": 7.913452398342881e-05, "train_batch_mse_error": 0.014863270334899426, "val_loss": 0.0195670984685421} +{"train_loss": 0.016384750604629517, "grad_norm": 0.05289982259273529, "global_step": 16006, "epoch": 151, "lr": 7.913211529638248e-05} +{"train_loss": 0.017020490020513535, "grad_norm": 0.07308749854564667, "global_step": 16007, "epoch": 151, "lr": 7.912970650697858e-05} +{"train_loss": 0.017969705164432526, "grad_norm": 0.0829109326004982, "global_step": 16008, "epoch": 151, "lr": 7.912729761522553e-05} +{"train_loss": 0.013852037489414215, "grad_norm": 0.07447976619005203, "global_step": 16009, "epoch": 151, "lr": 7.912488862113185e-05} +{"train_loss": 0.015739642083644867, "grad_norm": 0.10388052463531494, "global_step": 16010, "epoch": 151, "lr": 7.912247952470597e-05} +{"train_loss": 0.01595558598637581, "grad_norm": 0.15494504570960999, "global_step": 16011, "epoch": 151, "lr": 7.912007032595635e-05} +{"train_loss": 0.014066032133996487, "grad_norm": 0.055062104016542435, "global_step": 16012, "epoch": 151, "lr": 7.911766102489148e-05} +{"train_loss": 0.017613139003515244, "grad_norm": 0.059511780738830566, "global_step": 16013, "epoch": 151, "lr": 7.911525162151982e-05} +{"train_loss": 0.01797621324658394, "grad_norm": 0.07232798635959625, "global_step": 16014, "epoch": 151, "lr": 7.91128421158498e-05} +{"train_loss": 0.01618923991918564, "grad_norm": 0.07660368084907532, "global_step": 16015, "epoch": 151, "lr": 7.911043250788994e-05} +{"train_loss": 0.016794662922620773, "grad_norm": 0.06734161078929901, "global_step": 16016, "epoch": 151, "lr": 7.910802279764867e-05} +{"train_loss": 0.015828490257263184, "grad_norm": 0.12615351378917694, "global_step": 16017, "epoch": 151, "lr": 7.910561298513446e-05} +{"train_loss": 0.01688069850206375, "grad_norm": 0.06698238849639893, "global_step": 16018, "epoch": 151, "lr": 7.91032030703558e-05} +{"train_loss": 0.01610589772462845, "grad_norm": 0.07306000590324402, "global_step": 16019, "epoch": 151, "lr": 7.910079305332112e-05} +{"train_loss": 0.016008321195840836, "grad_norm": 0.09595135599374771, "global_step": 16020, "epoch": 151, "lr": 7.909838293403892e-05} +{"train_loss": 0.014420852065086365, "grad_norm": 0.08483121544122696, "global_step": 16021, "epoch": 151, "lr": 7.909597271251766e-05} +{"train_loss": 0.01636826992034912, "grad_norm": 0.06547154486179352, "global_step": 16022, "epoch": 151, "lr": 7.90935623887658e-05} +{"train_loss": 0.018257498741149902, "grad_norm": 0.05513398349285126, "global_step": 16023, "epoch": 151, "lr": 7.909115196279182e-05} +{"train_loss": 0.016851646825671196, "grad_norm": 0.05036857724189758, "global_step": 16024, "epoch": 151, "lr": 7.908874143460418e-05} +{"train_loss": 0.014673285186290741, "grad_norm": 0.05944833904504776, "global_step": 16025, "epoch": 151, "lr": 7.908633080421136e-05} +{"train_loss": 0.01718834787607193, "grad_norm": 0.0611993744969368, "global_step": 16026, "epoch": 151, "lr": 7.908392007162181e-05} +{"train_loss": 0.01706628128886223, "grad_norm": 0.06171248480677605, "global_step": 16027, "epoch": 151, "lr": 7.908150923684402e-05} +{"train_loss": 0.017879117280244827, "grad_norm": 0.06379808485507965, "global_step": 16028, "epoch": 151, "lr": 7.907909829988645e-05} +{"train_loss": 0.016706576570868492, "grad_norm": 0.07128839194774628, "global_step": 16029, "epoch": 151, "lr": 7.907668726075758e-05} +{"train_loss": 0.015404711477458477, "grad_norm": 0.0824480876326561, "global_step": 16030, "epoch": 151, "lr": 7.907427611946587e-05} +{"train_loss": 0.018103819340467453, "grad_norm": 0.054250385612249374, "global_step": 16031, "epoch": 151, "lr": 7.90718648760198e-05} +{"train_loss": 0.015621744096279144, "grad_norm": 0.05395512655377388, "global_step": 16032, "epoch": 151, "lr": 7.906945353042785e-05} +{"train_loss": 0.01584317907691002, "grad_norm": 0.06042784824967384, "global_step": 16033, "epoch": 151, "lr": 7.906704208269847e-05} +{"train_loss": 0.015100669115781784, "grad_norm": 0.10319340974092484, "global_step": 16034, "epoch": 151, "lr": 7.906463053284016e-05} +{"train_loss": 0.01750548928976059, "grad_norm": 0.10272851586341858, "global_step": 16035, "epoch": 151, "lr": 7.906221888086136e-05} +{"train_loss": 0.01722041890025139, "grad_norm": 0.11567272990942001, "global_step": 16036, "epoch": 151, "lr": 7.905980712677057e-05} +{"train_loss": 0.01286318525671959, "grad_norm": 0.06909835338592529, "global_step": 16037, "epoch": 151, "lr": 7.905739527057625e-05} +{"train_loss": 0.01561309490352869, "grad_norm": 0.15990318357944489, "global_step": 16038, "epoch": 151, "lr": 7.90549833122869e-05} +{"train_loss": 0.016471341252326965, "grad_norm": 0.08349008113145828, "global_step": 16039, "epoch": 151, "lr": 7.905257125191094e-05} +{"train_loss": 0.020561710000038147, "grad_norm": 0.06519912928342819, "global_step": 16040, "epoch": 151, "lr": 7.90501590894569e-05} +{"train_loss": 0.015189860947430134, "grad_norm": 0.0739639401435852, "global_step": 16041, "epoch": 151, "lr": 7.904774682493324e-05} +{"train_loss": 0.01448944304138422, "grad_norm": 0.0814436599612236, "global_step": 16042, "epoch": 151, "lr": 7.904533445834841e-05} +{"train_loss": 0.016886785626411438, "grad_norm": 0.05189257860183716, "global_step": 16043, "epoch": 151, "lr": 7.904292198971093e-05} +{"train_loss": 0.015463228337466717, "grad_norm": 0.09088948369026184, "global_step": 16044, "epoch": 151, "lr": 7.904050941902923e-05} +{"train_loss": 0.014767816290259361, "grad_norm": 0.11900152266025543, "global_step": 16045, "epoch": 151, "lr": 7.903809674631182e-05} +{"train_loss": 0.014318972826004028, "grad_norm": 0.07548347115516663, "global_step": 16046, "epoch": 151, "lr": 7.903568397156717e-05} +{"train_loss": 0.014974497258663177, "grad_norm": 0.0888148844242096, "global_step": 16047, "epoch": 151, "lr": 7.903327109480375e-05} +{"train_loss": 0.01618264988064766, "grad_norm": 0.08516992628574371, "global_step": 16048, "epoch": 151, "lr": 7.903085811603004e-05} +{"train_loss": 0.01618056371808052, "grad_norm": 0.09307794272899628, "global_step": 16049, "epoch": 151, "lr": 7.902844503525453e-05} +{"train_loss": 0.014409814961254597, "grad_norm": 0.11012329906225204, "global_step": 16050, "epoch": 151, "lr": 7.902603185248567e-05} +{"train_loss": 0.014949530363082886, "grad_norm": 0.050953902304172516, "global_step": 16051, "epoch": 151, "lr": 7.902361856773197e-05} +{"train_loss": 0.01938105747103691, "grad_norm": 0.07150938361883163, "global_step": 16052, "epoch": 151, "lr": 7.902120518100188e-05} +{"train_loss": 0.017118748277425766, "grad_norm": 0.05618544667959213, "global_step": 16053, "epoch": 151, "lr": 7.901879169230393e-05} +{"train_loss": 0.018288327381014824, "grad_norm": 0.08875259757041931, "global_step": 16054, "epoch": 151, "lr": 7.901637810164654e-05} +{"train_loss": 0.015167837962508202, "grad_norm": 0.10314834862947464, "global_step": 16055, "epoch": 151, "lr": 7.901396440903822e-05} +{"train_loss": 0.019518917426466942, "grad_norm": 0.05458387732505798, "global_step": 16056, "epoch": 151, "lr": 7.901155061448746e-05} +{"train_loss": 0.01705838181078434, "grad_norm": 0.1111990436911583, "global_step": 16057, "epoch": 151, "lr": 7.900913671800272e-05} +{"train_loss": 0.01589960604906082, "grad_norm": 0.06789423525333405, "global_step": 16058, "epoch": 151, "lr": 7.900672271959247e-05} +{"train_loss": 0.017021294683218002, "grad_norm": 0.057757847011089325, "global_step": 16059, "epoch": 151, "lr": 7.900430861926523e-05} +{"train_loss": 0.0188455730676651, "grad_norm": 0.11652487516403198, "global_step": 16060, "epoch": 151, "lr": 7.900189441702947e-05} +{"train_loss": 0.017367525026202202, "grad_norm": 0.08784352242946625, "global_step": 16061, "epoch": 151, "lr": 7.899948011289365e-05} +{"train_loss": 0.01645578071475029, "grad_norm": 0.09108506888151169, "global_step": 16062, "epoch": 151, "lr": 7.899706570686627e-05} +{"train_loss": 0.017301859334111214, "grad_norm": 0.058417029678821564, "global_step": 16063, "epoch": 151, "lr": 7.899465119895581e-05} +{"train_loss": 0.01542816124856472, "grad_norm": 0.0964847132563591, "global_step": 16064, "epoch": 151, "lr": 7.899223658917076e-05} +{"train_loss": 0.016511401161551476, "grad_norm": 0.08362915366888046, "global_step": 16065, "epoch": 151, "lr": 7.89898218775196e-05} +{"train_loss": 0.015647079795598984, "grad_norm": 0.07851240038871765, "global_step": 16066, "epoch": 151, "lr": 7.898740706401082e-05} +{"train_loss": 0.019343309104442596, "grad_norm": 0.16749002039432526, "global_step": 16067, "epoch": 151, "lr": 7.89849921486529e-05} +{"train_loss": 0.017023351043462753, "grad_norm": 0.0858861580491066, "global_step": 16068, "epoch": 151, "lr": 7.89825771314543e-05} +{"train_loss": 0.01585981994867325, "grad_norm": 0.05018841475248337, "global_step": 16069, "epoch": 151, "lr": 7.898016201242356e-05} +{"train_loss": 0.01632256619632244, "grad_norm": 0.07825854420661926, "global_step": 16070, "epoch": 151, "lr": 7.897774679156911e-05} +{"train_loss": 0.014254982583224773, "grad_norm": 0.09918320178985596, "global_step": 16071, "epoch": 151, "lr": 7.897533146889946e-05} +{"train_loss": 0.016606498509645462, "grad_norm": 0.13778135180473328, "global_step": 16072, "epoch": 151, "lr": 7.89729160444231e-05} +{"train_loss": 0.018104296177625656, "grad_norm": 0.08012199401855469, "global_step": 16073, "epoch": 151, "lr": 7.897050051814852e-05} +{"train_loss": 0.017693571746349335, "grad_norm": 0.057307351380586624, "global_step": 16074, "epoch": 151, "lr": 7.896808489008419e-05} +{"train_loss": 0.015964403748512268, "grad_norm": 0.07098998129367828, "global_step": 16075, "epoch": 151, "lr": 7.896566916023861e-05} +{"train_loss": 0.017938226461410522, "grad_norm": 0.09474223107099533, "global_step": 16076, "epoch": 151, "lr": 7.896325332862027e-05} +{"train_loss": 0.015301348641514778, "grad_norm": 0.09644721448421478, "global_step": 16077, "epoch": 151, "lr": 7.896083739523765e-05} +{"train_loss": 0.017603665590286255, "grad_norm": 0.14805108308792114, "global_step": 16078, "epoch": 151, "lr": 7.895842136009925e-05} +{"train_loss": 0.017998237162828445, "grad_norm": 0.05857568234205246, "global_step": 16079, "epoch": 151, "lr": 7.895600522321354e-05} +{"train_loss": 0.01600622572004795, "grad_norm": 0.08205600827932358, "global_step": 16080, "epoch": 151, "lr": 7.895358898458902e-05} +{"train_loss": 0.017963318154215813, "grad_norm": 0.114109106361866, "global_step": 16081, "epoch": 151, "lr": 7.895117264423419e-05} +{"train_loss": 0.016338204964995384, "grad_norm": 0.06760245561599731, "global_step": 16082, "epoch": 151, "lr": 7.894875620215752e-05} +{"train_loss": 0.013246607966721058, "grad_norm": 0.10747473686933517, "global_step": 16083, "epoch": 151, "lr": 7.894633965836753e-05} +{"train_loss": 0.016420919448137283, "grad_norm": 0.14831702411174774, "global_step": 16084, "epoch": 151, "lr": 7.894392301287267e-05} +{"train_loss": 0.015029101632535458, "grad_norm": 0.05658896639943123, "global_step": 16085, "epoch": 151, "lr": 7.894150626568146e-05} +{"train_loss": 0.015322632156312466, "grad_norm": 0.08043922483921051, "global_step": 16086, "epoch": 151, "lr": 7.893908941680237e-05} +{"train_loss": 0.01820691116154194, "grad_norm": 0.07390925288200378, "global_step": 16087, "epoch": 151, "lr": 7.893667246624392e-05} +{"train_loss": 0.01547708734869957, "grad_norm": 0.0729353204369545, "global_step": 16088, "epoch": 151, "lr": 7.893425541401458e-05} +{"train_loss": 0.01900586113333702, "grad_norm": 0.10159585624933243, "global_step": 16089, "epoch": 151, "lr": 7.893183826012286e-05} +{"train_loss": 0.01724730059504509, "grad_norm": 0.059884052723646164, "global_step": 16090, "epoch": 151, "lr": 7.892942100457723e-05} +{"train_loss": 0.01780165545642376, "grad_norm": 0.09719670563936234, "global_step": 16091, "epoch": 151, "lr": 7.892700364738621e-05} +{"train_loss": 0.015653615817427635, "grad_norm": 0.09317365288734436, "global_step": 16092, "epoch": 151, "lr": 7.892458618855827e-05} +{"train_loss": 0.018113385885953903, "grad_norm": 0.10490590333938599, "global_step": 16093, "epoch": 151, "lr": 7.892216862810191e-05} +{"train_loss": 0.01661526970565319, "grad_norm": 0.07843761146068573, "global_step": 16094, "epoch": 151, "lr": 7.891975096602563e-05} +{"train_loss": 0.014626819640398026, "grad_norm": 0.06788240373134613, "global_step": 16095, "epoch": 151, "lr": 7.891733320233792e-05} +{"train_loss": 0.01674043759703636, "grad_norm": 0.07856542617082596, "global_step": 16096, "epoch": 151, "lr": 7.891491533704728e-05} +{"train_loss": 0.017550107091665268, "grad_norm": 0.0672609806060791, "global_step": 16097, "epoch": 151, "lr": 7.89124973701622e-05} +{"train_loss": 0.016443921253085136, "grad_norm": 0.05109558254480362, "global_step": 16098, "epoch": 151, "lr": 7.891007930169119e-05} +{"train_loss": 0.017784472554922104, "grad_norm": 0.1022024005651474, "global_step": 16099, "epoch": 151, "lr": 7.890766113164271e-05} +{"train_loss": 0.017720703035593033, "grad_norm": 0.08005185425281525, "global_step": 16100, "epoch": 151, "lr": 7.89052428600253e-05} +{"train_loss": 0.01710684224963188, "grad_norm": 0.06532962620258331, "global_step": 16101, "epoch": 151, "lr": 7.890282448684743e-05} +{"train_loss": 0.01675828918814659, "grad_norm": 0.05231403186917305, "global_step": 16102, "epoch": 151, "lr": 7.890040601211761e-05} +{"train_loss": 0.01435093767940998, "grad_norm": 0.05436224117875099, "global_step": 16103, "epoch": 151, "lr": 7.889798743584434e-05} +{"train_loss": 0.017860090360045433, "grad_norm": 0.08859942853450775, "global_step": 16104, "epoch": 151, "lr": 7.889556875803608e-05} +{"train_loss": 0.017421264201402664, "grad_norm": 0.10750434547662735, "global_step": 16105, "epoch": 151, "lr": 7.889314997870139e-05} +{"train_loss": 0.01975308358669281, "grad_norm": 0.10105845332145691, "global_step": 16106, "epoch": 151, "lr": 7.889073109784871e-05} +{"train_loss": 0.018234487622976303, "grad_norm": 0.12016262859106064, "global_step": 16107, "epoch": 151, "lr": 7.888831211548657e-05} +{"train_loss": 0.01723436824977398, "grad_norm": 0.07120708376169205, "global_step": 16108, "epoch": 151, "lr": 7.888589303162348e-05} +{"train_loss": 0.018088599666953087, "grad_norm": 0.10571218281984329, "global_step": 16109, "epoch": 151, "lr": 7.888347384626791e-05} +{"train_loss": 0.015141613781452179, "grad_norm": 0.11037875711917877, "global_step": 16110, "epoch": 151, "lr": 7.888105455942838e-05} +{"train_loss": 0.016581072656543187, "grad_norm": 0.19043448567390442, "global_step": 16111, "epoch": 151, "lr": 7.887863517111338e-05, "val_loss": 0.01927286386489868} +{"train_loss": 0.018670113757252693, "grad_norm": 0.09614893049001694, "global_step": 16112, "epoch": 152, "lr": 7.88762156813314e-05} +{"train_loss": 0.020913224667310715, "grad_norm": 0.06123846769332886, "global_step": 16113, "epoch": 152, "lr": 7.8873796090091e-05} +{"train_loss": 0.0174575075507164, "grad_norm": 0.07009746134281158, "global_step": 16114, "epoch": 152, "lr": 7.88713763974006e-05} +{"train_loss": 0.014857439324259758, "grad_norm": 0.07344570010900497, "global_step": 16115, "epoch": 152, "lr": 7.886895660326876e-05} +{"train_loss": 0.014134084805846214, "grad_norm": 0.06798722594976425, "global_step": 16116, "epoch": 152, "lr": 7.886653670770396e-05} +{"train_loss": 0.018419422209262848, "grad_norm": 0.06205129623413086, "global_step": 16117, "epoch": 152, "lr": 7.886411671071468e-05} +{"train_loss": 0.013275312259793282, "grad_norm": 0.07248982042074203, "global_step": 16118, "epoch": 152, "lr": 7.886169661230947e-05} +{"train_loss": 0.016703277826309204, "grad_norm": 0.09288588166236877, "global_step": 16119, "epoch": 152, "lr": 7.88592764124968e-05} +{"train_loss": 0.01741204410791397, "grad_norm": 0.0879514068365097, "global_step": 16120, "epoch": 152, "lr": 7.88568561112852e-05} +{"train_loss": 0.01620279625058174, "grad_norm": 0.12495121359825134, "global_step": 16121, "epoch": 152, "lr": 7.885443570868312e-05} +{"train_loss": 0.016386769711971283, "grad_norm": 0.12695692479610443, "global_step": 16122, "epoch": 152, "lr": 7.885201520469914e-05} +{"train_loss": 0.016371870413422585, "grad_norm": 0.09617724269628525, "global_step": 16123, "epoch": 152, "lr": 7.88495945993417e-05} +{"train_loss": 0.019123520702123642, "grad_norm": 0.07398655265569687, "global_step": 16124, "epoch": 152, "lr": 7.884717389261934e-05} +{"train_loss": 0.013952219858765602, "grad_norm": 0.10410044342279434, "global_step": 16125, "epoch": 152, "lr": 7.884475308454056e-05} +{"train_loss": 0.01475189533084631, "grad_norm": 0.1407647579908371, "global_step": 16126, "epoch": 152, "lr": 7.884233217511386e-05} +{"train_loss": 0.01545537170022726, "grad_norm": 0.07216376066207886, "global_step": 16127, "epoch": 152, "lr": 7.883991116434774e-05} +{"train_loss": 0.016310764476656914, "grad_norm": 0.06198253855109215, "global_step": 16128, "epoch": 152, "lr": 7.883749005225072e-05} +{"train_loss": 0.015526474453508854, "grad_norm": 0.060876358300447464, "global_step": 16129, "epoch": 152, "lr": 7.883506883883131e-05} +{"train_loss": 0.01555332075804472, "grad_norm": 0.05318645015358925, "global_step": 16130, "epoch": 152, "lr": 7.8832647524098e-05} +{"train_loss": 0.016917651519179344, "grad_norm": 0.06261078268289566, "global_step": 16131, "epoch": 152, "lr": 7.883022610805931e-05} +{"train_loss": 0.016981318593025208, "grad_norm": 0.04657699167728424, "global_step": 16132, "epoch": 152, "lr": 7.882780459072373e-05} +{"train_loss": 0.017286034300923347, "grad_norm": 0.07123509049415588, "global_step": 16133, "epoch": 152, "lr": 7.882538297209978e-05} +{"train_loss": 0.017813213169574738, "grad_norm": 0.051437754184007645, "global_step": 16134, "epoch": 152, "lr": 7.882296125219599e-05} +{"train_loss": 0.015055060386657715, "grad_norm": 0.057745881378650665, "global_step": 16135, "epoch": 152, "lr": 7.882053943102083e-05} +{"train_loss": 0.01908917725086212, "grad_norm": 0.1218089610338211, "global_step": 16136, "epoch": 152, "lr": 7.881811750858285e-05} +{"train_loss": 0.015748482197523117, "grad_norm": 0.09417014569044113, "global_step": 16137, "epoch": 152, "lr": 7.881569548489053e-05} +{"train_loss": 0.015947559848427773, "grad_norm": 0.13018445670604706, "global_step": 16138, "epoch": 152, "lr": 7.881327335995238e-05} +{"train_loss": 0.01609145477414131, "grad_norm": 0.06257826089859009, "global_step": 16139, "epoch": 152, "lr": 7.881085113377691e-05} +{"train_loss": 0.014521196484565735, "grad_norm": 0.04627697542309761, "global_step": 16140, "epoch": 152, "lr": 7.880842880637267e-05} +{"train_loss": 0.01862039789557457, "grad_norm": 0.08587773144245148, "global_step": 16141, "epoch": 152, "lr": 7.88060063777481e-05} +{"train_loss": 0.014900085516273975, "grad_norm": 0.17444291710853577, "global_step": 16142, "epoch": 152, "lr": 7.880358384791177e-05} +{"train_loss": 0.0166411604732275, "grad_norm": 0.12034782022237778, "global_step": 16143, "epoch": 152, "lr": 7.880116121687219e-05} +{"train_loss": 0.015966063365340233, "grad_norm": 0.07692671567201614, "global_step": 16144, "epoch": 152, "lr": 7.879873848463783e-05} +{"train_loss": 0.015190603211522102, "grad_norm": 0.07351644337177277, "global_step": 16145, "epoch": 152, "lr": 7.879631565121724e-05} +{"train_loss": 0.017725739628076553, "grad_norm": 0.04711572453379631, "global_step": 16146, "epoch": 152, "lr": 7.87938927166189e-05} +{"train_loss": 0.014171645045280457, "grad_norm": 0.11405017971992493, "global_step": 16147, "epoch": 152, "lr": 7.879146968085137e-05} +{"train_loss": 0.016705568879842758, "grad_norm": 0.15435892343521118, "global_step": 16148, "epoch": 152, "lr": 7.878904654392311e-05} +{"train_loss": 0.017156245186924934, "grad_norm": 0.07626673579216003, "global_step": 16149, "epoch": 152, "lr": 7.87866233058427e-05} +{"train_loss": 0.017774425446987152, "grad_norm": 0.12023130059242249, "global_step": 16150, "epoch": 152, "lr": 7.878419996661859e-05} +{"train_loss": 0.015747524797916412, "grad_norm": 0.0687621608376503, "global_step": 16151, "epoch": 152, "lr": 7.878177652625931e-05} +{"train_loss": 0.015826428309082985, "grad_norm": 0.06463376432657242, "global_step": 16152, "epoch": 152, "lr": 7.87793529847734e-05} +{"train_loss": 0.0158158577978611, "grad_norm": 0.08527253568172455, "global_step": 16153, "epoch": 152, "lr": 7.877692934216935e-05} +{"train_loss": 0.017205964773893356, "grad_norm": 0.07284586876630783, "global_step": 16154, "epoch": 152, "lr": 7.877450559845568e-05} +{"train_loss": 0.015026381239295006, "grad_norm": 0.10843116044998169, "global_step": 16155, "epoch": 152, "lr": 7.877208175364092e-05} +{"train_loss": 0.014573685824871063, "grad_norm": 0.133173868060112, "global_step": 16156, "epoch": 152, "lr": 7.876965780773358e-05} +{"train_loss": 0.018500465899705887, "grad_norm": 0.14538808166980743, "global_step": 16157, "epoch": 152, "lr": 7.876723376074218e-05} +{"train_loss": 0.017823342233896255, "grad_norm": 0.05122412368655205, "global_step": 16158, "epoch": 152, "lr": 7.876480961267521e-05} +{"train_loss": 0.01725315861403942, "grad_norm": 0.06878591328859329, "global_step": 16159, "epoch": 152, "lr": 7.876238536354122e-05} +{"train_loss": 0.01789509691298008, "grad_norm": 0.055202316492795944, "global_step": 16160, "epoch": 152, "lr": 7.875996101334873e-05} +{"train_loss": 0.015243468806147575, "grad_norm": 0.08568396419286728, "global_step": 16161, "epoch": 152, "lr": 7.875753656210622e-05} +{"train_loss": 0.01668129675090313, "grad_norm": 0.06228028982877731, "global_step": 16162, "epoch": 152, "lr": 7.875511200982224e-05} +{"train_loss": 0.015050144866108894, "grad_norm": 0.10354641079902649, "global_step": 16163, "epoch": 152, "lr": 7.875268735650531e-05} +{"train_loss": 0.014059413224458694, "grad_norm": 0.07196225970983505, "global_step": 16164, "epoch": 152, "lr": 7.875026260216393e-05} +{"train_loss": 0.013744814321398735, "grad_norm": 0.11622730642557144, "global_step": 16165, "epoch": 152, "lr": 7.874783774680664e-05} +{"train_loss": 0.017073415219783783, "grad_norm": 0.1112489402294159, "global_step": 16166, "epoch": 152, "lr": 7.874541279044196e-05} +{"train_loss": 0.01719525083899498, "grad_norm": 0.07962502539157867, "global_step": 16167, "epoch": 152, "lr": 7.874298773307838e-05} +{"train_loss": 0.018040155991911888, "grad_norm": 0.09014514833688736, "global_step": 16168, "epoch": 152, "lr": 7.874056257472444e-05} +{"train_loss": 0.01554007176309824, "grad_norm": 0.06187088042497635, "global_step": 16169, "epoch": 152, "lr": 7.873813731538868e-05} +{"train_loss": 0.01451652217656374, "grad_norm": 0.06866742670536041, "global_step": 16170, "epoch": 152, "lr": 7.873571195507958e-05} +{"train_loss": 0.015123363584280014, "grad_norm": 0.08981140702962875, "global_step": 16171, "epoch": 152, "lr": 7.87332864938057e-05} +{"train_loss": 0.014688922092318535, "grad_norm": 0.0982314869761467, "global_step": 16172, "epoch": 152, "lr": 7.873086093157556e-05} +{"train_loss": 0.015460356138646603, "grad_norm": 0.09231691062450409, "global_step": 16173, "epoch": 152, "lr": 7.872843526839764e-05} +{"train_loss": 0.018793467432260513, "grad_norm": 0.13034020364284515, "global_step": 16174, "epoch": 152, "lr": 7.87260095042805e-05} +{"train_loss": 0.018985308706760406, "grad_norm": 0.2056588977575302, "global_step": 16175, "epoch": 152, "lr": 7.872358363923269e-05} +{"train_loss": 0.0179757047444582, "grad_norm": 0.11486051231622696, "global_step": 16176, "epoch": 152, "lr": 7.872115767326266e-05} +{"train_loss": 0.017363175749778748, "grad_norm": 0.0679311454296112, "global_step": 16177, "epoch": 152, "lr": 7.871873160637897e-05} +{"train_loss": 0.01778840832412243, "grad_norm": 0.049555156379938126, "global_step": 16178, "epoch": 152, "lr": 7.871630543859016e-05} +{"train_loss": 0.018691280856728554, "grad_norm": 0.09904862940311432, "global_step": 16179, "epoch": 152, "lr": 7.871387916990475e-05} +{"train_loss": 0.01653401367366314, "grad_norm": 0.06180192530155182, "global_step": 16180, "epoch": 152, "lr": 7.871145280033125e-05} +{"train_loss": 0.015299173071980476, "grad_norm": 0.04805586859583855, "global_step": 16181, "epoch": 152, "lr": 7.870902632987819e-05} +{"train_loss": 0.014829980209469795, "grad_norm": 0.08353474736213684, "global_step": 16182, "epoch": 152, "lr": 7.87065997585541e-05} +{"train_loss": 0.01702173613011837, "grad_norm": 0.11998283863067627, "global_step": 16183, "epoch": 152, "lr": 7.870417308636749e-05} +{"train_loss": 0.014238467440009117, "grad_norm": 0.08949672430753708, "global_step": 16184, "epoch": 152, "lr": 7.870174631332691e-05} +{"train_loss": 0.017303194850683212, "grad_norm": 0.05708421766757965, "global_step": 16185, "epoch": 152, "lr": 7.869931943944087e-05} +{"train_loss": 0.016214029863476753, "grad_norm": 0.0858733132481575, "global_step": 16186, "epoch": 152, "lr": 7.86968924647179e-05} +{"train_loss": 0.016212530434131622, "grad_norm": 0.052374113351106644, "global_step": 16187, "epoch": 152, "lr": 7.869446538916654e-05} +{"train_loss": 0.01816316694021225, "grad_norm": 0.08973336964845657, "global_step": 16188, "epoch": 152, "lr": 7.86920382127953e-05} +{"train_loss": 0.01687510684132576, "grad_norm": 0.07648905366659164, "global_step": 16189, "epoch": 152, "lr": 7.86896109356127e-05} +{"train_loss": 0.02055925689637661, "grad_norm": 0.0789458304643631, "global_step": 16190, "epoch": 152, "lr": 7.868718355762732e-05} +{"train_loss": 0.01652602106332779, "grad_norm": 0.07311418652534485, "global_step": 16191, "epoch": 152, "lr": 7.868475607884762e-05} +{"train_loss": 0.0175100676715374, "grad_norm": 0.09529376775026321, "global_step": 16192, "epoch": 152, "lr": 7.868232849928218e-05} +{"train_loss": 0.015851004049181938, "grad_norm": 0.050355080515146255, "global_step": 16193, "epoch": 152, "lr": 7.867990081893951e-05} +{"train_loss": 0.017718330025672913, "grad_norm": 0.060646090656518936, "global_step": 16194, "epoch": 152, "lr": 7.867747303782814e-05} +{"train_loss": 0.0154680535197258, "grad_norm": 0.052714455872774124, "global_step": 16195, "epoch": 152, "lr": 7.867504515595661e-05} +{"train_loss": 0.014854555018246174, "grad_norm": 0.0652415081858635, "global_step": 16196, "epoch": 152, "lr": 7.867261717333343e-05} +{"train_loss": 0.01948555000126362, "grad_norm": 0.06718461960554123, "global_step": 16197, "epoch": 152, "lr": 7.867018908996713e-05} +{"train_loss": 0.015853019431233406, "grad_norm": 0.09897205233573914, "global_step": 16198, "epoch": 152, "lr": 7.866776090586627e-05} +{"train_loss": 0.01635592058300972, "grad_norm": 0.0754166916012764, "global_step": 16199, "epoch": 152, "lr": 7.866533262103936e-05} +{"train_loss": 0.017892781645059586, "grad_norm": 0.09466316550970078, "global_step": 16200, "epoch": 152, "lr": 7.866290423549495e-05} +{"train_loss": 0.018068384379148483, "grad_norm": 0.052517492324113846, "global_step": 16201, "epoch": 152, "lr": 7.866047574924155e-05} +{"train_loss": 0.019304459914565086, "grad_norm": 0.10301981121301651, "global_step": 16202, "epoch": 152, "lr": 7.865804716228771e-05} +{"train_loss": 0.015115288086235523, "grad_norm": 0.059941839426755905, "global_step": 16203, "epoch": 152, "lr": 7.865561847464194e-05} +{"train_loss": 0.016460422426462173, "grad_norm": 0.06733648478984833, "global_step": 16204, "epoch": 152, "lr": 7.86531896863128e-05} +{"train_loss": 0.016682259738445282, "grad_norm": 0.04727660119533539, "global_step": 16205, "epoch": 152, "lr": 7.865076079730879e-05} +{"train_loss": 0.016707507893443108, "grad_norm": 0.06960318982601166, "global_step": 16206, "epoch": 152, "lr": 7.86483318076385e-05} +{"train_loss": 0.016291998326778412, "grad_norm": 0.09498941898345947, "global_step": 16207, "epoch": 152, "lr": 7.864590271731041e-05} +{"train_loss": 0.018168039619922638, "grad_norm": 0.0796579122543335, "global_step": 16208, "epoch": 152, "lr": 7.864347352633309e-05} +{"train_loss": 0.016862567514181137, "grad_norm": 0.06820371001958847, "global_step": 16209, "epoch": 152, "lr": 7.864104423471506e-05} +{"train_loss": 0.015699446201324463, "grad_norm": 0.0934322252869606, "global_step": 16210, "epoch": 152, "lr": 7.863861484246484e-05} +{"train_loss": 0.015564416535198689, "grad_norm": 0.055611565709114075, "global_step": 16211, "epoch": 152, "lr": 7.863618534959098e-05} +{"train_loss": 0.014889858663082123, "grad_norm": 0.06224023923277855, "global_step": 16212, "epoch": 152, "lr": 7.863375575610203e-05} +{"train_loss": 0.01790616102516651, "grad_norm": 0.1065293475985527, "global_step": 16213, "epoch": 152, "lr": 7.863132606200652e-05} +{"train_loss": 0.01578204706311226, "grad_norm": 0.09541567414999008, "global_step": 16214, "epoch": 152, "lr": 7.862889626731298e-05} +{"train_loss": 0.01706026680767536, "grad_norm": 0.0876951664686203, "global_step": 16215, "epoch": 152, "lr": 7.862646637202993e-05} +{"train_loss": 0.01757059618830681, "grad_norm": 0.06386757642030716, "global_step": 16216, "epoch": 152, "lr": 7.862403637616595e-05} +{"train_loss": 0.016575996561924804, "grad_norm": 0.08455079793930054, "global_step": 16217, "epoch": 152, "lr": 7.862160627972955e-05, "val_loss": 0.019139746204018593} +{"train_loss": 0.01586150750517845, "grad_norm": 0.04781487211585045, "global_step": 16218, "epoch": 153, "lr": 7.861917608272926e-05} +{"train_loss": 0.015140303410589695, "grad_norm": 0.10296764969825745, "global_step": 16219, "epoch": 153, "lr": 7.861674578517366e-05} +{"train_loss": 0.020164065062999725, "grad_norm": 0.061479244381189346, "global_step": 16220, "epoch": 153, "lr": 7.861431538707123e-05} +{"train_loss": 0.02023392915725708, "grad_norm": 0.06129677593708038, "global_step": 16221, "epoch": 153, "lr": 7.861188488843054e-05} +{"train_loss": 0.017065580934286118, "grad_norm": 0.09596886485815048, "global_step": 16222, "epoch": 153, "lr": 7.860945428926015e-05} +{"train_loss": 0.015210609883069992, "grad_norm": 0.08080939948558807, "global_step": 16223, "epoch": 153, "lr": 7.860702358956857e-05} +{"train_loss": 0.01673184148967266, "grad_norm": 0.1233241856098175, "global_step": 16224, "epoch": 153, "lr": 7.860459278936434e-05} +{"train_loss": 0.01810465008020401, "grad_norm": 0.15715231001377106, "global_step": 16225, "epoch": 153, "lr": 7.860216188865604e-05} +{"train_loss": 0.016077499836683273, "grad_norm": 0.13012515008449554, "global_step": 16226, "epoch": 153, "lr": 7.859973088745213e-05} +{"train_loss": 0.017428508028388023, "grad_norm": 0.10318278521299362, "global_step": 16227, "epoch": 153, "lr": 7.859729978576124e-05} +{"train_loss": 0.014144753105938435, "grad_norm": 0.08146960288286209, "global_step": 16228, "epoch": 153, "lr": 7.859486858359189e-05} +{"train_loss": 0.01566941663622856, "grad_norm": 0.04346666485071182, "global_step": 16229, "epoch": 153, "lr": 7.859243728095259e-05} +{"train_loss": 0.017179150134325027, "grad_norm": 0.0719953328371048, "global_step": 16230, "epoch": 153, "lr": 7.859000587785188e-05} +{"train_loss": 0.018140625208616257, "grad_norm": 0.060420118272304535, "global_step": 16231, "epoch": 153, "lr": 7.858757437429834e-05} +{"train_loss": 0.016032693907618523, "grad_norm": 0.10948531329631805, "global_step": 16232, "epoch": 153, "lr": 7.858514277030049e-05} +{"train_loss": 0.015346945263445377, "grad_norm": 0.06902489811182022, "global_step": 16233, "epoch": 153, "lr": 7.85827110658669e-05} +{"train_loss": 0.015062075108289719, "grad_norm": 0.06912069767713547, "global_step": 16234, "epoch": 153, "lr": 7.858027926100607e-05} +{"train_loss": 0.016845108941197395, "grad_norm": 0.06324514001607895, "global_step": 16235, "epoch": 153, "lr": 7.857784735572658e-05} +{"train_loss": 0.01846596971154213, "grad_norm": 0.10214792191982269, "global_step": 16236, "epoch": 153, "lr": 7.857541535003697e-05} +{"train_loss": 0.017180636525154114, "grad_norm": 0.11773890256881714, "global_step": 16237, "epoch": 153, "lr": 7.857298324394575e-05} +{"train_loss": 0.015434294939041138, "grad_norm": 0.14082440733909607, "global_step": 16238, "epoch": 153, "lr": 7.857055103746151e-05} +{"train_loss": 0.018305208534002304, "grad_norm": 0.15000170469284058, "global_step": 16239, "epoch": 153, "lr": 7.856811873059279e-05} +{"train_loss": 0.015511193312704563, "grad_norm": 0.047257255762815475, "global_step": 16240, "epoch": 153, "lr": 7.85656863233481e-05} +{"train_loss": 0.01648598723113537, "grad_norm": 0.10211534798145294, "global_step": 16241, "epoch": 153, "lr": 7.856325381573603e-05} +{"train_loss": 0.016848646104335785, "grad_norm": 0.06133751571178436, "global_step": 16242, "epoch": 153, "lr": 7.856082120776511e-05} +{"train_loss": 0.016840822994709015, "grad_norm": 0.09960299730300903, "global_step": 16243, "epoch": 153, "lr": 7.855838849944388e-05} +{"train_loss": 0.016100937500596046, "grad_norm": 0.11025764793157578, "global_step": 16244, "epoch": 153, "lr": 7.855595569078089e-05} +{"train_loss": 0.01659376546740532, "grad_norm": 0.0698494091629982, "global_step": 16245, "epoch": 153, "lr": 7.855352278178469e-05} +{"train_loss": 0.01976742595434189, "grad_norm": 0.08561109751462936, "global_step": 16246, "epoch": 153, "lr": 7.855108977246384e-05} +{"train_loss": 0.01655346155166626, "grad_norm": 0.059344105422496796, "global_step": 16247, "epoch": 153, "lr": 7.854865666282685e-05} +{"train_loss": 0.01591401919722557, "grad_norm": 0.06492538005113602, "global_step": 16248, "epoch": 153, "lr": 7.854622345288231e-05} +{"train_loss": 0.017294352874159813, "grad_norm": 0.06601648032665253, "global_step": 16249, "epoch": 153, "lr": 7.854379014263876e-05} +{"train_loss": 0.016248909756541252, "grad_norm": 0.07068575173616409, "global_step": 16250, "epoch": 153, "lr": 7.854135673210474e-05} +{"train_loss": 0.01728139817714691, "grad_norm": 0.09785855561494827, "global_step": 16251, "epoch": 153, "lr": 7.853892322128882e-05} +{"train_loss": 0.017653122544288635, "grad_norm": 0.1036660224199295, "global_step": 16252, "epoch": 153, "lr": 7.853648961019952e-05} +{"train_loss": 0.018153918907046318, "grad_norm": 0.0864323228597641, "global_step": 16253, "epoch": 153, "lr": 7.85340558988454e-05} +{"train_loss": 0.01596241258084774, "grad_norm": 0.06197286397218704, "global_step": 16254, "epoch": 153, "lr": 7.853162208723502e-05} +{"train_loss": 0.015730710700154305, "grad_norm": 0.12377695739269257, "global_step": 16255, "epoch": 153, "lr": 7.852918817537694e-05} +{"train_loss": 0.01733342930674553, "grad_norm": 0.13148333132266998, "global_step": 16256, "epoch": 153, "lr": 7.852675416327968e-05} +{"train_loss": 0.015387636609375477, "grad_norm": 0.08458185940980911, "global_step": 16257, "epoch": 153, "lr": 7.852432005095183e-05} +{"train_loss": 0.0171505119651556, "grad_norm": 0.07186643779277802, "global_step": 16258, "epoch": 153, "lr": 7.852188583840192e-05} +{"train_loss": 0.016131602227687836, "grad_norm": 0.07760056853294373, "global_step": 16259, "epoch": 153, "lr": 7.85194515256385e-05} +{"train_loss": 0.018873916938900948, "grad_norm": 0.08989277482032776, "global_step": 16260, "epoch": 153, "lr": 7.851701711267015e-05} +{"train_loss": 0.014723126776516438, "grad_norm": 0.09902406483888626, "global_step": 16261, "epoch": 153, "lr": 7.851458259950538e-05} +{"train_loss": 0.01693306304514408, "grad_norm": 0.056152183562517166, "global_step": 16262, "epoch": 153, "lr": 7.851214798615279e-05} +{"train_loss": 0.01734957844018936, "grad_norm": 0.10458804666996002, "global_step": 16263, "epoch": 153, "lr": 7.850971327262089e-05} +{"train_loss": 0.015507848002016544, "grad_norm": 0.05506380647420883, "global_step": 16264, "epoch": 153, "lr": 7.850727845891826e-05} +{"train_loss": 0.01605229452252388, "grad_norm": 0.05695263296365738, "global_step": 16265, "epoch": 153, "lr": 7.850484354505345e-05} +{"train_loss": 0.016895249485969543, "grad_norm": 0.07490215450525284, "global_step": 16266, "epoch": 153, "lr": 7.850240853103504e-05} +{"train_loss": 0.014750098809599876, "grad_norm": 0.0530664324760437, "global_step": 16267, "epoch": 153, "lr": 7.849997341687153e-05} +{"train_loss": 0.015681598335504532, "grad_norm": 0.06768584251403809, "global_step": 16268, "epoch": 153, "lr": 7.849753820257153e-05} +{"train_loss": 0.016723118722438812, "grad_norm": 0.11920249462127686, "global_step": 16269, "epoch": 153, "lr": 7.849510288814356e-05} +{"train_loss": 0.017525052651762962, "grad_norm": 0.05115969106554985, "global_step": 16270, "epoch": 153, "lr": 7.849266747359619e-05} +{"train_loss": 0.01721136085689068, "grad_norm": 0.08536206185817719, "global_step": 16271, "epoch": 153, "lr": 7.849023195893798e-05} +{"train_loss": 0.015463312156498432, "grad_norm": 0.12152300029993057, "global_step": 16272, "epoch": 153, "lr": 7.84877963441775e-05} +{"train_loss": 0.017761189490556717, "grad_norm": 0.07034119218587875, "global_step": 16273, "epoch": 153, "lr": 7.848536062932328e-05} +{"train_loss": 0.01846218854188919, "grad_norm": 0.07901826500892639, "global_step": 16274, "epoch": 153, "lr": 7.848292481438389e-05} +{"train_loss": 0.01576359197497368, "grad_norm": 0.07681477814912796, "global_step": 16275, "epoch": 153, "lr": 7.84804888993679e-05} +{"train_loss": 0.016472432762384415, "grad_norm": 0.056531891226768494, "global_step": 16276, "epoch": 153, "lr": 7.847805288428384e-05} +{"train_loss": 0.015167735517024994, "grad_norm": 0.09916769713163376, "global_step": 16277, "epoch": 153, "lr": 7.847561676914029e-05} +{"train_loss": 0.015582630410790443, "grad_norm": 0.10694393515586853, "global_step": 16278, "epoch": 153, "lr": 7.847318055394581e-05} +{"train_loss": 0.015040763653814793, "grad_norm": 0.060143645852804184, "global_step": 16279, "epoch": 153, "lr": 7.847074423870896e-05} +{"train_loss": 0.013887149281799793, "grad_norm": 0.06581293046474457, "global_step": 16280, "epoch": 153, "lr": 7.84683078234383e-05} +{"train_loss": 0.01569407619535923, "grad_norm": 0.05402044951915741, "global_step": 16281, "epoch": 153, "lr": 7.846587130814238e-05} +{"train_loss": 0.01839369349181652, "grad_norm": 0.12265180796384811, "global_step": 16282, "epoch": 153, "lr": 7.846343469282977e-05} +{"train_loss": 0.014334069564938545, "grad_norm": 0.08880352973937988, "global_step": 16283, "epoch": 153, "lr": 7.846099797750903e-05} +{"train_loss": 0.01734769530594349, "grad_norm": 0.06014612689614296, "global_step": 16284, "epoch": 153, "lr": 7.845856116218871e-05} +{"train_loss": 0.015048549510538578, "grad_norm": 0.06969718635082245, "global_step": 16285, "epoch": 153, "lr": 7.845612424687739e-05} +{"train_loss": 0.017674652859568596, "grad_norm": 0.09852053970098495, "global_step": 16286, "epoch": 153, "lr": 7.845368723158363e-05} +{"train_loss": 0.016504887491464615, "grad_norm": 0.0736142247915268, "global_step": 16287, "epoch": 153, "lr": 7.845125011631598e-05} +{"train_loss": 0.015481872484087944, "grad_norm": 0.0491146445274353, "global_step": 16288, "epoch": 153, "lr": 7.844881290108301e-05} +{"train_loss": 0.015607304871082306, "grad_norm": 0.10549911111593246, "global_step": 16289, "epoch": 153, "lr": 7.844637558589329e-05} +{"train_loss": 0.016907984390854836, "grad_norm": 0.05996283143758774, "global_step": 16290, "epoch": 153, "lr": 7.844393817075538e-05} +{"train_loss": 0.016723591834306717, "grad_norm": 0.09059809148311615, "global_step": 16291, "epoch": 153, "lr": 7.844150065567783e-05} +{"train_loss": 0.017408620566129684, "grad_norm": 0.07925789803266525, "global_step": 16292, "epoch": 153, "lr": 7.843906304066921e-05} +{"train_loss": 0.017060892656445503, "grad_norm": 0.07262954115867615, "global_step": 16293, "epoch": 153, "lr": 7.843662532573811e-05} +{"train_loss": 0.014691994525492191, "grad_norm": 0.06292247772216797, "global_step": 16294, "epoch": 153, "lr": 7.843418751089306e-05} +{"train_loss": 0.016818562522530556, "grad_norm": 0.064037024974823, "global_step": 16295, "epoch": 153, "lr": 7.843174959614264e-05} +{"train_loss": 0.01532069407403469, "grad_norm": 0.052539002150297165, "global_step": 16296, "epoch": 153, "lr": 7.842931158149543e-05} +{"train_loss": 0.015284501947462559, "grad_norm": 0.08362314105033875, "global_step": 16297, "epoch": 153, "lr": 7.842687346695999e-05} +{"train_loss": 0.016265399754047394, "grad_norm": 0.07812722027301788, "global_step": 16298, "epoch": 153, "lr": 7.842443525254486e-05} +{"train_loss": 0.015242503024637699, "grad_norm": 0.045506905764341354, "global_step": 16299, "epoch": 153, "lr": 7.842199693825862e-05} +{"train_loss": 0.01533467136323452, "grad_norm": 0.055390045046806335, "global_step": 16300, "epoch": 153, "lr": 7.841955852410986e-05} +{"train_loss": 0.015939921140670776, "grad_norm": 0.07910756021738052, "global_step": 16301, "epoch": 153, "lr": 7.841712001010712e-05} +{"train_loss": 0.018151719123125076, "grad_norm": 0.1565558910369873, "global_step": 16302, "epoch": 153, "lr": 7.841468139625898e-05} +{"train_loss": 0.014434108510613441, "grad_norm": 0.05025235936045647, "global_step": 16303, "epoch": 153, "lr": 7.841224268257402e-05} +{"train_loss": 0.017106033861637115, "grad_norm": 0.05235766991972923, "global_step": 16304, "epoch": 153, "lr": 7.840980386906078e-05} +{"train_loss": 0.015329521149396896, "grad_norm": 0.061937060207128525, "global_step": 16305, "epoch": 153, "lr": 7.840736495572785e-05} +{"train_loss": 0.014564032666385174, "grad_norm": 0.09172889590263367, "global_step": 16306, "epoch": 153, "lr": 7.84049259425838e-05} +{"train_loss": 0.017728067934513092, "grad_norm": 0.06780162453651428, "global_step": 16307, "epoch": 153, "lr": 7.840248682963718e-05} +{"train_loss": 0.015442777425050735, "grad_norm": 0.05770047754049301, "global_step": 16308, "epoch": 153, "lr": 7.840004761689659e-05} +{"train_loss": 0.015300121158361435, "grad_norm": 0.09895483404397964, "global_step": 16309, "epoch": 153, "lr": 7.839760830437057e-05} +{"train_loss": 0.017935670912265778, "grad_norm": 0.0769362598657608, "global_step": 16310, "epoch": 153, "lr": 7.83951688920677e-05} +{"train_loss": 0.015125916339457035, "grad_norm": 0.08207497745752335, "global_step": 16311, "epoch": 153, "lr": 7.839272937999658e-05} +{"train_loss": 0.017006313428282738, "grad_norm": 0.1263037621974945, "global_step": 16312, "epoch": 153, "lr": 7.839028976816572e-05} +{"train_loss": 0.019588623195886612, "grad_norm": 0.06152444705367088, "global_step": 16313, "epoch": 153, "lr": 7.838785005658375e-05} +{"train_loss": 0.013256823644042015, "grad_norm": 0.07067272812128067, "global_step": 16314, "epoch": 153, "lr": 7.838541024525923e-05} +{"train_loss": 0.01727205514907837, "grad_norm": 0.06517612189054489, "global_step": 16315, "epoch": 153, "lr": 7.83829703342007e-05} +{"train_loss": 0.02005608379840851, "grad_norm": 0.07440128922462463, "global_step": 16316, "epoch": 153, "lr": 7.838053032341678e-05} +{"train_loss": 0.019865408539772034, "grad_norm": 0.058229487389326096, "global_step": 16317, "epoch": 153, "lr": 7.8378090212916e-05} +{"train_loss": 0.01676030457019806, "grad_norm": 0.07320040464401245, "global_step": 16318, "epoch": 153, "lr": 7.837565000270695e-05} +{"train_loss": 0.017710912972688675, "grad_norm": 0.08661411702632904, "global_step": 16319, "epoch": 153, "lr": 7.837320969279822e-05} +{"train_loss": 0.016471801325678825, "grad_norm": 0.06604107469320297, "global_step": 16320, "epoch": 153, "lr": 7.837076928319838e-05} +{"train_loss": 0.01855294406414032, "grad_norm": 0.09719384461641312, "global_step": 16321, "epoch": 153, "lr": 7.836832877391596e-05} +{"train_loss": 0.014428568072617054, "grad_norm": 0.0539524219930172, "global_step": 16322, "epoch": 153, "lr": 7.83658881649596e-05} +{"train_loss": 0.016542891558822034, "grad_norm": 0.08186177164316177, "global_step": 16323, "epoch": 153, "lr": 7.836344745633783e-05, "val_loss": 0.019102197140455246} +{"train_loss": 0.018224047496914864, "grad_norm": 0.0639219582080841, "global_step": 16324, "epoch": 154, "lr": 7.836100664805924e-05} +{"train_loss": 0.018410922959446907, "grad_norm": 0.1336960792541504, "global_step": 16325, "epoch": 154, "lr": 7.835856574013242e-05} +{"train_loss": 0.014975406229496002, "grad_norm": 0.12766553461551666, "global_step": 16326, "epoch": 154, "lr": 7.835612473256593e-05} +{"train_loss": 0.018555261194705963, "grad_norm": 0.07454002648591995, "global_step": 16327, "epoch": 154, "lr": 7.835368362536832e-05} +{"train_loss": 0.014499791897833347, "grad_norm": 0.06385470181703568, "global_step": 16328, "epoch": 154, "lr": 7.835124241854823e-05} +{"train_loss": 0.01727822609245777, "grad_norm": 0.06832482665777206, "global_step": 16329, "epoch": 154, "lr": 7.83488011121142e-05} +{"train_loss": 0.018535492941737175, "grad_norm": 0.10030176490545273, "global_step": 16330, "epoch": 154, "lr": 7.83463597060748e-05} +{"train_loss": 0.0167439766228199, "grad_norm": 0.05130910873413086, "global_step": 16331, "epoch": 154, "lr": 7.834391820043862e-05} +{"train_loss": 0.0161091648042202, "grad_norm": 0.05467824265360832, "global_step": 16332, "epoch": 154, "lr": 7.834147659521425e-05} +{"train_loss": 0.01471485011279583, "grad_norm": 0.05503873527050018, "global_step": 16333, "epoch": 154, "lr": 7.833903489041025e-05} +{"train_loss": 0.015886543318629265, "grad_norm": 0.08166465908288956, "global_step": 16334, "epoch": 154, "lr": 7.83365930860352e-05} +{"train_loss": 0.016829025000333786, "grad_norm": 0.05521402135491371, "global_step": 16335, "epoch": 154, "lr": 7.83341511820977e-05} +{"train_loss": 0.015238120220601559, "grad_norm": 0.1075616180896759, "global_step": 16336, "epoch": 154, "lr": 7.83317091786063e-05} +{"train_loss": 0.017350606620311737, "grad_norm": 0.11272100359201431, "global_step": 16337, "epoch": 154, "lr": 7.83292670755696e-05} +{"train_loss": 0.015555412508547306, "grad_norm": 0.06000693514943123, "global_step": 16338, "epoch": 154, "lr": 7.832682487299617e-05} +{"train_loss": 0.015423394739627838, "grad_norm": 0.11295244097709656, "global_step": 16339, "epoch": 154, "lr": 7.83243825708946e-05} +{"train_loss": 0.018587781116366386, "grad_norm": 0.0741489827632904, "global_step": 16340, "epoch": 154, "lr": 7.832194016927347e-05} +{"train_loss": 0.017850615084171295, "grad_norm": 0.07478760927915573, "global_step": 16341, "epoch": 154, "lr": 7.831949766814136e-05} +{"train_loss": 0.01547914557158947, "grad_norm": 0.16394254565238953, "global_step": 16342, "epoch": 154, "lr": 7.831705506750684e-05} +{"train_loss": 0.015386920422315598, "grad_norm": 0.044101860374212265, "global_step": 16343, "epoch": 154, "lr": 7.831461236737852e-05} +{"train_loss": 0.017451096326112747, "grad_norm": 0.10497143864631653, "global_step": 16344, "epoch": 154, "lr": 7.831216956776497e-05} +{"train_loss": 0.015700004994869232, "grad_norm": 0.055224254727363586, "global_step": 16345, "epoch": 154, "lr": 7.830972666867475e-05} +{"train_loss": 0.01836920529603958, "grad_norm": 0.05859008803963661, "global_step": 16346, "epoch": 154, "lr": 7.830728367011648e-05} +{"train_loss": 0.015082159079611301, "grad_norm": 0.05704256892204285, "global_step": 16347, "epoch": 154, "lr": 7.83048405720987e-05} +{"train_loss": 0.01808345690369606, "grad_norm": 0.10864625871181488, "global_step": 16348, "epoch": 154, "lr": 7.830239737463004e-05} +{"train_loss": 0.017608631402254105, "grad_norm": 0.058971699327230453, "global_step": 16349, "epoch": 154, "lr": 7.829995407771907e-05} +{"train_loss": 0.015976913273334503, "grad_norm": 0.05777604132890701, "global_step": 16350, "epoch": 154, "lr": 7.829751068137434e-05} +{"train_loss": 0.015238904394209385, "grad_norm": 0.11042383313179016, "global_step": 16351, "epoch": 154, "lr": 7.829506718560449e-05} +{"train_loss": 0.019290229305624962, "grad_norm": 0.055033374577760696, "global_step": 16352, "epoch": 154, "lr": 7.829262359041807e-05} +{"train_loss": 0.015850495547056198, "grad_norm": 0.04591494798660278, "global_step": 16353, "epoch": 154, "lr": 7.829017989582369e-05} +{"train_loss": 0.01670493558049202, "grad_norm": 0.16642343997955322, "global_step": 16354, "epoch": 154, "lr": 7.828773610182989e-05} +{"train_loss": 0.017680345103144646, "grad_norm": 0.06766390800476074, "global_step": 16355, "epoch": 154, "lr": 7.828529220844532e-05} +{"train_loss": 0.013839183375239372, "grad_norm": 0.10472825169563293, "global_step": 16356, "epoch": 154, "lr": 7.82828482156785e-05} +{"train_loss": 0.01612546667456627, "grad_norm": 0.058020997792482376, "global_step": 16357, "epoch": 154, "lr": 7.828040412353808e-05} +{"train_loss": 0.015792524442076683, "grad_norm": 0.10277558118104935, "global_step": 16358, "epoch": 154, "lr": 7.82779599320326e-05} +{"train_loss": 0.016337551176548004, "grad_norm": 0.08110053837299347, "global_step": 16359, "epoch": 154, "lr": 7.827551564117068e-05} +{"train_loss": 0.01761830970644951, "grad_norm": 0.09291433542966843, "global_step": 16360, "epoch": 154, "lr": 7.827307125096089e-05} +{"train_loss": 0.016022365540266037, "grad_norm": 0.07217149436473846, "global_step": 16361, "epoch": 154, "lr": 7.827062676141181e-05} +{"train_loss": 0.017801430076360703, "grad_norm": 0.07248256355524063, "global_step": 16362, "epoch": 154, "lr": 7.826818217253207e-05} +{"train_loss": 0.014444101601839066, "grad_norm": 0.05520239844918251, "global_step": 16363, "epoch": 154, "lr": 7.826573748433022e-05} +{"train_loss": 0.017972523346543312, "grad_norm": 0.051711760461330414, "global_step": 16364, "epoch": 154, "lr": 7.826329269681484e-05} +{"train_loss": 0.015099935233592987, "grad_norm": 0.1813584566116333, "global_step": 16365, "epoch": 154, "lr": 7.826084780999456e-05} +{"train_loss": 0.01674770377576351, "grad_norm": 0.06181400641798973, "global_step": 16366, "epoch": 154, "lr": 7.825840282387795e-05} +{"train_loss": 0.013270758092403412, "grad_norm": 0.04971475154161453, "global_step": 16367, "epoch": 154, "lr": 7.825595773847361e-05} +{"train_loss": 0.014292307198047638, "grad_norm": 0.08480380475521088, "global_step": 16368, "epoch": 154, "lr": 7.82535125537901e-05} +{"train_loss": 0.01672763004899025, "grad_norm": 0.05530053749680519, "global_step": 16369, "epoch": 154, "lr": 7.825106726983606e-05} +{"train_loss": 0.0157938152551651, "grad_norm": 0.07776214927434921, "global_step": 16370, "epoch": 154, "lr": 7.824862188662003e-05} +{"train_loss": 0.016227416694164276, "grad_norm": 0.060376014560461044, "global_step": 16371, "epoch": 154, "lr": 7.824617640415064e-05} +{"train_loss": 0.013543752953410149, "grad_norm": 0.0442391112446785, "global_step": 16372, "epoch": 154, "lr": 7.824373082243648e-05} +{"train_loss": 0.016958344727754593, "grad_norm": 0.09739039093255997, "global_step": 16373, "epoch": 154, "lr": 7.824128514148612e-05} +{"train_loss": 0.016723567619919777, "grad_norm": 0.06543377041816711, "global_step": 16374, "epoch": 154, "lr": 7.823883936130817e-05} +{"train_loss": 0.01528090052306652, "grad_norm": 0.0788540169596672, "global_step": 16375, "epoch": 154, "lr": 7.823639348191121e-05} +{"train_loss": 0.016274254769086838, "grad_norm": 0.09273525327444077, "global_step": 16376, "epoch": 154, "lr": 7.823394750330387e-05} +{"train_loss": 0.016701921820640564, "grad_norm": 0.07264802604913712, "global_step": 16377, "epoch": 154, "lr": 7.823150142549469e-05} +{"train_loss": 0.015845153480768204, "grad_norm": 0.11111766844987869, "global_step": 16378, "epoch": 154, "lr": 7.82290552484923e-05} +{"train_loss": 0.015603741630911827, "grad_norm": 0.07396785914897919, "global_step": 16379, "epoch": 154, "lr": 7.822660897230529e-05} +{"train_loss": 0.018606437370181084, "grad_norm": 0.05437244474887848, "global_step": 16380, "epoch": 154, "lr": 7.822416259694225e-05} +{"train_loss": 0.016453836113214493, "grad_norm": 0.06663935631513596, "global_step": 16381, "epoch": 154, "lr": 7.822171612241177e-05} +{"train_loss": 0.015324566513299942, "grad_norm": 0.08841203153133392, "global_step": 16382, "epoch": 154, "lr": 7.821926954872244e-05} +{"train_loss": 0.015231408178806305, "grad_norm": 0.08387113362550735, "global_step": 16383, "epoch": 154, "lr": 7.82168228758829e-05} +{"train_loss": 0.01619376614689827, "grad_norm": 0.0948064997792244, "global_step": 16384, "epoch": 154, "lr": 7.82143761039017e-05} +{"train_loss": 0.01844424568116665, "grad_norm": 0.06601981073617935, "global_step": 16385, "epoch": 154, "lr": 7.821192923278746e-05} +{"train_loss": 0.015495484694838524, "grad_norm": 0.09225602447986603, "global_step": 16386, "epoch": 154, "lr": 7.820948226254877e-05} +{"train_loss": 0.016560181975364685, "grad_norm": 0.08717449754476547, "global_step": 16387, "epoch": 154, "lr": 7.820703519319422e-05} +{"train_loss": 0.01637837290763855, "grad_norm": 0.06238924339413643, "global_step": 16388, "epoch": 154, "lr": 7.820458802473241e-05} +{"train_loss": 0.014207050204277039, "grad_norm": 0.12195595353841782, "global_step": 16389, "epoch": 154, "lr": 7.820214075717194e-05} +{"train_loss": 0.01738901063799858, "grad_norm": 0.15869592130184174, "global_step": 16390, "epoch": 154, "lr": 7.819969339052145e-05} +{"train_loss": 0.020103607326745987, "grad_norm": 0.06038569286465645, "global_step": 16391, "epoch": 154, "lr": 7.819724592478945e-05} +{"train_loss": 0.016309916973114014, "grad_norm": 0.12064987421035767, "global_step": 16392, "epoch": 154, "lr": 7.819479835998463e-05} +{"train_loss": 0.01887815073132515, "grad_norm": 0.0854085236787796, "global_step": 16393, "epoch": 154, "lr": 7.819235069611553e-05} +{"train_loss": 0.016960661858320236, "grad_norm": 0.05872495099902153, "global_step": 16394, "epoch": 154, "lr": 7.818990293319077e-05} +{"train_loss": 0.02239980548620224, "grad_norm": 0.12097242474555969, "global_step": 16395, "epoch": 154, "lr": 7.818745507121896e-05} +{"train_loss": 0.0168064646422863, "grad_norm": 0.11719357222318649, "global_step": 16396, "epoch": 154, "lr": 7.818500711020869e-05} +{"train_loss": 0.01634027063846588, "grad_norm": 0.09127798676490784, "global_step": 16397, "epoch": 154, "lr": 7.818255905016856e-05} +{"train_loss": 0.015072035603225231, "grad_norm": 0.0747528150677681, "global_step": 16398, "epoch": 154, "lr": 7.818011089110716e-05} +{"train_loss": 0.015385888516902924, "grad_norm": 0.08657777309417725, "global_step": 16399, "epoch": 154, "lr": 7.817766263303313e-05} +{"train_loss": 0.015636781230568886, "grad_norm": 0.05546913295984268, "global_step": 16400, "epoch": 154, "lr": 7.817521427595503e-05} +{"train_loss": 0.017990071326494217, "grad_norm": 0.09322430193424225, "global_step": 16401, "epoch": 154, "lr": 7.817276581988149e-05} +{"train_loss": 0.014189690351486206, "grad_norm": 0.15409888327121735, "global_step": 16402, "epoch": 154, "lr": 7.81703172648211e-05} +{"train_loss": 0.017397303134202957, "grad_norm": 0.09047873318195343, "global_step": 16403, "epoch": 154, "lr": 7.816786861078246e-05} +{"train_loss": 0.016073741018772125, "grad_norm": 0.07282979041337967, "global_step": 16404, "epoch": 154, "lr": 7.816541985777418e-05} +{"train_loss": 0.01885109767317772, "grad_norm": 0.09899991005659103, "global_step": 16405, "epoch": 154, "lr": 7.816297100580487e-05} +{"train_loss": 0.01974741369485855, "grad_norm": 0.08104700595140457, "global_step": 16406, "epoch": 154, "lr": 7.816052205488314e-05} +{"train_loss": 0.01727362535893917, "grad_norm": 0.06731662899255753, "global_step": 16407, "epoch": 154, "lr": 7.815807300501755e-05} +{"train_loss": 0.016374090686440468, "grad_norm": 0.060343656688928604, "global_step": 16408, "epoch": 154, "lr": 7.815562385621677e-05} +{"train_loss": 0.01650744304060936, "grad_norm": 0.07883909344673157, "global_step": 16409, "epoch": 154, "lr": 7.815317460848934e-05} +{"train_loss": 0.014326058328151703, "grad_norm": 0.06377067416906357, "global_step": 16410, "epoch": 154, "lr": 7.815072526184392e-05} +{"train_loss": 0.015363669022917747, "grad_norm": 0.06341592222452164, "global_step": 16411, "epoch": 154, "lr": 7.814827581628909e-05} +{"train_loss": 0.017080705612897873, "grad_norm": 0.05775747448205948, "global_step": 16412, "epoch": 154, "lr": 7.814582627183347e-05} +{"train_loss": 0.017293216660618782, "grad_norm": 0.07154940813779831, "global_step": 16413, "epoch": 154, "lr": 7.814337662848563e-05} +{"train_loss": 0.017338406294584274, "grad_norm": 0.07555205374956131, "global_step": 16414, "epoch": 154, "lr": 7.814092688625423e-05} +{"train_loss": 0.015695977956056595, "grad_norm": 0.06933211535215378, "global_step": 16415, "epoch": 154, "lr": 7.813847704514783e-05} +{"train_loss": 0.018804185092449188, "grad_norm": 0.08613213151693344, "global_step": 16416, "epoch": 154, "lr": 7.813602710517508e-05} +{"train_loss": 0.016325034201145172, "grad_norm": 0.07997168600559235, "global_step": 16417, "epoch": 154, "lr": 7.813357706634456e-05} +{"train_loss": 0.013783617876470089, "grad_norm": 0.058675430715084076, "global_step": 16418, "epoch": 154, "lr": 7.813112692866487e-05} +{"train_loss": 0.01836775243282318, "grad_norm": 0.05329589918255806, "global_step": 16419, "epoch": 154, "lr": 7.812867669214466e-05} +{"train_loss": 0.015119802206754684, "grad_norm": 0.056278958916664124, "global_step": 16420, "epoch": 154, "lr": 7.81262263567925e-05} +{"train_loss": 0.01803784817457199, "grad_norm": 0.058056507259607315, "global_step": 16421, "epoch": 154, "lr": 7.8123775922617e-05} +{"train_loss": 0.018279213458299637, "grad_norm": 0.07111446559429169, "global_step": 16422, "epoch": 154, "lr": 7.81213253896268e-05} +{"train_loss": 0.01674582250416279, "grad_norm": 0.13022322952747345, "global_step": 16423, "epoch": 154, "lr": 7.811887475783049e-05} +{"train_loss": 0.014401743188500404, "grad_norm": 0.14796045422554016, "global_step": 16424, "epoch": 154, "lr": 7.811642402723667e-05} +{"train_loss": 0.015964793041348457, "grad_norm": 0.07146327942609787, "global_step": 16425, "epoch": 154, "lr": 7.811397319785397e-05} +{"train_loss": 0.016576038673520088, "grad_norm": 0.08023186028003693, "global_step": 16426, "epoch": 154, "lr": 7.8111522269691e-05} +{"train_loss": 0.015493609011173248, "grad_norm": 0.0960596576333046, "global_step": 16427, "epoch": 154, "lr": 7.810907124275634e-05} +{"train_loss": 0.015899833291769028, "grad_norm": 0.06048830598592758, "global_step": 16428, "epoch": 154, "lr": 7.810662011705867e-05} +{"train_loss": 0.016531177611916133, "grad_norm": 0.09067585319280624, "global_step": 16429, "epoch": 154, "lr": 7.810416889260653e-05, "val_loss": 0.018989114090800285} +{"train_loss": 0.015839379280805588, "grad_norm": 0.05430709943175316, "global_step": 16430, "epoch": 155, "lr": 7.810171756940859e-05} +{"train_loss": 0.016693267971277237, "grad_norm": 0.07714955508708954, "global_step": 16431, "epoch": 155, "lr": 7.80992661474734e-05} +{"train_loss": 0.019707519561052322, "grad_norm": 0.15322190523147583, "global_step": 16432, "epoch": 155, "lr": 7.809681462680963e-05} +{"train_loss": 0.016726233065128326, "grad_norm": 0.08396568894386292, "global_step": 16433, "epoch": 155, "lr": 7.809436300742588e-05} +{"train_loss": 0.016633184626698494, "grad_norm": 0.11036770790815353, "global_step": 16434, "epoch": 155, "lr": 7.809191128933073e-05} +{"train_loss": 0.015648456290364265, "grad_norm": 0.10773719847202301, "global_step": 16435, "epoch": 155, "lr": 7.808945947253286e-05} +{"train_loss": 0.015738477930426598, "grad_norm": 0.1449725478887558, "global_step": 16436, "epoch": 155, "lr": 7.808700755704081e-05} +{"train_loss": 0.01594744808971882, "grad_norm": 0.05668416619300842, "global_step": 16437, "epoch": 155, "lr": 7.808455554286325e-05} +{"train_loss": 0.018317393958568573, "grad_norm": 0.15357671678066254, "global_step": 16438, "epoch": 155, "lr": 7.808210343000875e-05} +{"train_loss": 0.015432512387633324, "grad_norm": 0.07365588843822479, "global_step": 16439, "epoch": 155, "lr": 7.807965121848597e-05} +{"train_loss": 0.01621781662106514, "grad_norm": 0.06925023347139359, "global_step": 16440, "epoch": 155, "lr": 7.807719890830352e-05} +{"train_loss": 0.017584679648280144, "grad_norm": 0.16597393155097961, "global_step": 16441, "epoch": 155, "lr": 7.807474649946997e-05} +{"train_loss": 0.015738680958747864, "grad_norm": 0.08000588417053223, "global_step": 16442, "epoch": 155, "lr": 7.807229399199399e-05} +{"train_loss": 0.017117470502853394, "grad_norm": 0.07921939343214035, "global_step": 16443, "epoch": 155, "lr": 7.806984138588418e-05} +{"train_loss": 0.01618868112564087, "grad_norm": 0.06166933849453926, "global_step": 16444, "epoch": 155, "lr": 7.806738868114915e-05} +{"train_loss": 0.017882227897644043, "grad_norm": 0.12434632331132889, "global_step": 16445, "epoch": 155, "lr": 7.806493587779753e-05} +{"train_loss": 0.015898820012807846, "grad_norm": 0.11941179633140564, "global_step": 16446, "epoch": 155, "lr": 7.806248297583792e-05} +{"train_loss": 0.015229182317852974, "grad_norm": 0.06574534624814987, "global_step": 16447, "epoch": 155, "lr": 7.806002997527895e-05} +{"train_loss": 0.018216649070382118, "grad_norm": 0.0684833973646164, "global_step": 16448, "epoch": 155, "lr": 7.805757687612925e-05} +{"train_loss": 0.015484523959457874, "grad_norm": 0.05606460943818092, "global_step": 16449, "epoch": 155, "lr": 7.805512367839741e-05} +{"train_loss": 0.017739567905664444, "grad_norm": 0.06101968139410019, "global_step": 16450, "epoch": 155, "lr": 7.805267038209207e-05} +{"train_loss": 0.015928905457258224, "grad_norm": 0.0844915360212326, "global_step": 16451, "epoch": 155, "lr": 7.805021698722186e-05} +{"train_loss": 0.01662478968501091, "grad_norm": 0.07012877613306046, "global_step": 16452, "epoch": 155, "lr": 7.804776349379537e-05} +{"train_loss": 0.013274986296892166, "grad_norm": 0.10584936290979385, "global_step": 16453, "epoch": 155, "lr": 7.804530990182123e-05} +{"train_loss": 0.016512185335159302, "grad_norm": 0.07538271695375443, "global_step": 16454, "epoch": 155, "lr": 7.80428562113081e-05} +{"train_loss": 0.017207719385623932, "grad_norm": 0.053757067769765854, "global_step": 16455, "epoch": 155, "lr": 7.804040242226453e-05} +{"train_loss": 0.01620900258421898, "grad_norm": 0.16726526618003845, "global_step": 16456, "epoch": 155, "lr": 7.803794853469922e-05} +{"train_loss": 0.015911052003502846, "grad_norm": 0.05047477409243584, "global_step": 16457, "epoch": 155, "lr": 7.803549454862072e-05} +{"train_loss": 0.015533183701336384, "grad_norm": 0.07267080247402191, "global_step": 16458, "epoch": 155, "lr": 7.80330404640377e-05} +{"train_loss": 0.015158033929765224, "grad_norm": 0.05532323569059372, "global_step": 16459, "epoch": 155, "lr": 7.803058628095877e-05} +{"train_loss": 0.014951190911233425, "grad_norm": 0.06147822365164757, "global_step": 16460, "epoch": 155, "lr": 7.802813199939253e-05} +{"train_loss": 0.01488927099853754, "grad_norm": 0.07514916360378265, "global_step": 16461, "epoch": 155, "lr": 7.802567761934765e-05} +{"train_loss": 0.01623353734612465, "grad_norm": 0.08633553981781006, "global_step": 16462, "epoch": 155, "lr": 7.80232231408327e-05} +{"train_loss": 0.01699085161089897, "grad_norm": 0.09461452066898346, "global_step": 16463, "epoch": 155, "lr": 7.802076856385634e-05} +{"train_loss": 0.01666911691427231, "grad_norm": 0.06602849811315536, "global_step": 16464, "epoch": 155, "lr": 7.80183138884272e-05} +{"train_loss": 0.013441130518913269, "grad_norm": 0.05364808440208435, "global_step": 16465, "epoch": 155, "lr": 7.801585911455387e-05} +{"train_loss": 0.017355382442474365, "grad_norm": 0.10904406011104584, "global_step": 16466, "epoch": 155, "lr": 7.801340424224501e-05} +{"train_loss": 0.013333753682672977, "grad_norm": 0.05358786880970001, "global_step": 16467, "epoch": 155, "lr": 7.801094927150922e-05} +{"train_loss": 0.018820486962795258, "grad_norm": 0.1474631130695343, "global_step": 16468, "epoch": 155, "lr": 7.800849420235515e-05} +{"train_loss": 0.015501603484153748, "grad_norm": 0.06899230927228928, "global_step": 16469, "epoch": 155, "lr": 7.80060390347914e-05} +{"train_loss": 0.018849879503250122, "grad_norm": 0.14693060517311096, "global_step": 16470, "epoch": 155, "lr": 7.80035837688266e-05} +{"train_loss": 0.01918609067797661, "grad_norm": 0.09181402623653412, "global_step": 16471, "epoch": 155, "lr": 7.80011284044694e-05} +{"train_loss": 0.01695145107805729, "grad_norm": 0.13095618784427643, "global_step": 16472, "epoch": 155, "lr": 7.799867294172842e-05} +{"train_loss": 0.016957156360149384, "grad_norm": 0.0894048810005188, "global_step": 16473, "epoch": 155, "lr": 7.799621738061225e-05} +{"train_loss": 0.016564827412366867, "grad_norm": 0.11325298994779587, "global_step": 16474, "epoch": 155, "lr": 7.799376172112957e-05} +{"train_loss": 0.017682675272226334, "grad_norm": 0.06245372071862221, "global_step": 16475, "epoch": 155, "lr": 7.799130596328899e-05} +{"train_loss": 0.016408482566475868, "grad_norm": 0.07786943018436432, "global_step": 16476, "epoch": 155, "lr": 7.798885010709911e-05} +{"train_loss": 0.017323778942227364, "grad_norm": 0.11474871635437012, "global_step": 16477, "epoch": 155, "lr": 7.79863941525686e-05} +{"train_loss": 0.016581445932388306, "grad_norm": 0.07899041473865509, "global_step": 16478, "epoch": 155, "lr": 7.798393809970607e-05} +{"train_loss": 0.01914709433913231, "grad_norm": 0.06685496866703033, "global_step": 16479, "epoch": 155, "lr": 7.798148194852015e-05} +{"train_loss": 0.014445758424699306, "grad_norm": 0.06485815346240997, "global_step": 16480, "epoch": 155, "lr": 7.797902569901947e-05} +{"train_loss": 0.01581745222210884, "grad_norm": 0.1600753515958786, "global_step": 16481, "epoch": 155, "lr": 7.797656935121266e-05} +{"train_loss": 0.015836719423532486, "grad_norm": 0.04538951814174652, "global_step": 16482, "epoch": 155, "lr": 7.797411290510835e-05} +{"train_loss": 0.015889972448349, "grad_norm": 0.06504907459020615, "global_step": 16483, "epoch": 155, "lr": 7.797165636071517e-05} +{"train_loss": 0.013351261615753174, "grad_norm": 0.09054933488368988, "global_step": 16484, "epoch": 155, "lr": 7.796919971804177e-05} +{"train_loss": 0.01765523850917816, "grad_norm": 0.0666302964091301, "global_step": 16485, "epoch": 155, "lr": 7.796674297709676e-05} +{"train_loss": 0.016475291922688484, "grad_norm": 0.05511735752224922, "global_step": 16486, "epoch": 155, "lr": 7.796428613788877e-05} +{"train_loss": 0.020948149263858795, "grad_norm": 0.07575405389070511, "global_step": 16487, "epoch": 155, "lr": 7.796182920042646e-05} +{"train_loss": 0.016949988901615143, "grad_norm": 0.054929181933403015, "global_step": 16488, "epoch": 155, "lr": 7.795937216471843e-05} +{"train_loss": 0.014439735561609268, "grad_norm": 0.06383064389228821, "global_step": 16489, "epoch": 155, "lr": 7.79569150307733e-05} +{"train_loss": 0.015767818316817284, "grad_norm": 0.0794300064444542, "global_step": 16490, "epoch": 155, "lr": 7.795445779859976e-05} +{"train_loss": 0.014985358342528343, "grad_norm": 0.0786142572760582, "global_step": 16491, "epoch": 155, "lr": 7.795200046820641e-05} +{"train_loss": 0.018471701070666313, "grad_norm": 0.07247431576251984, "global_step": 16492, "epoch": 155, "lr": 7.794954303960187e-05} +{"train_loss": 0.01741168648004532, "grad_norm": 0.053390175104141235, "global_step": 16493, "epoch": 155, "lr": 7.794708551279481e-05} +{"train_loss": 0.01541145984083414, "grad_norm": 0.04653292894363403, "global_step": 16494, "epoch": 155, "lr": 7.794462788779383e-05} +{"train_loss": 0.01636122725903988, "grad_norm": 0.07565927505493164, "global_step": 16495, "epoch": 155, "lr": 7.794217016460759e-05} +{"train_loss": 0.018416374921798706, "grad_norm": 0.08480500429868698, "global_step": 16496, "epoch": 155, "lr": 7.793971234324471e-05} +{"train_loss": 0.015157431364059448, "grad_norm": 0.10482785105705261, "global_step": 16497, "epoch": 155, "lr": 7.793725442371382e-05} +{"train_loss": 0.016340414062142372, "grad_norm": 0.11683913320302963, "global_step": 16498, "epoch": 155, "lr": 7.79347964060236e-05} +{"train_loss": 0.019183028489351273, "grad_norm": 0.07593610882759094, "global_step": 16499, "epoch": 155, "lr": 7.793233829018262e-05} +{"train_loss": 0.0141685139387846, "grad_norm": 0.05340683460235596, "global_step": 16500, "epoch": 155, "lr": 7.792988007619957e-05} +{"train_loss": 0.018345940858125687, "grad_norm": 0.07596855610609055, "global_step": 16501, "epoch": 155, "lr": 7.792742176408306e-05} +{"train_loss": 0.020122544839978218, "grad_norm": 0.16088156402111053, "global_step": 16502, "epoch": 155, "lr": 7.792496335384173e-05} +{"train_loss": 0.016377484425902367, "grad_norm": 0.06891389936208725, "global_step": 16503, "epoch": 155, "lr": 7.792250484548424e-05} +{"train_loss": 0.016383938491344452, "grad_norm": 0.07818552851676941, "global_step": 16504, "epoch": 155, "lr": 7.79200462390192e-05} +{"train_loss": 0.01728975400328636, "grad_norm": 0.06417401880025864, "global_step": 16505, "epoch": 155, "lr": 7.791758753445528e-05} +{"train_loss": 0.019646193832159042, "grad_norm": 0.08747981488704681, "global_step": 16506, "epoch": 155, "lr": 7.791512873180107e-05} +{"train_loss": 0.018578294664621353, "grad_norm": 0.10241106897592545, "global_step": 16507, "epoch": 155, "lr": 7.791266983106524e-05} +{"train_loss": 0.014786770567297935, "grad_norm": 0.05935933068394661, "global_step": 16508, "epoch": 155, "lr": 7.791021083225644e-05} +{"train_loss": 0.015097413212060928, "grad_norm": 0.07460986077785492, "global_step": 16509, "epoch": 155, "lr": 7.790775173538329e-05} +{"train_loss": 0.016580630093812943, "grad_norm": 0.09075094759464264, "global_step": 16510, "epoch": 155, "lr": 7.790529254045445e-05} +{"train_loss": 0.015485983341932297, "grad_norm": 0.051691412925720215, "global_step": 16511, "epoch": 155, "lr": 7.790283324747853e-05} +{"train_loss": 0.014241712167859077, "grad_norm": 0.056541331112384796, "global_step": 16512, "epoch": 155, "lr": 7.79003738564642e-05} +{"train_loss": 0.01778407022356987, "grad_norm": 0.06648404151201248, "global_step": 16513, "epoch": 155, "lr": 7.78979143674201e-05} +{"train_loss": 0.015744488686323166, "grad_norm": 0.06992802768945694, "global_step": 16514, "epoch": 155, "lr": 7.789545478035484e-05} +{"train_loss": 0.0165714044123888, "grad_norm": 0.1505649983882904, "global_step": 16515, "epoch": 155, "lr": 7.78929950952771e-05} +{"train_loss": 0.018800290301442146, "grad_norm": 0.07315529137849808, "global_step": 16516, "epoch": 155, "lr": 7.789053531219549e-05} +{"train_loss": 0.014514348469674587, "grad_norm": 0.04869204759597778, "global_step": 16517, "epoch": 155, "lr": 7.788807543111866e-05} +{"train_loss": 0.017033174633979797, "grad_norm": 0.14224231243133545, "global_step": 16518, "epoch": 155, "lr": 7.788561545205528e-05} +{"train_loss": 0.016137678176164627, "grad_norm": 0.07788530737161636, "global_step": 16519, "epoch": 155, "lr": 7.788315537501399e-05} +{"train_loss": 0.015531491488218307, "grad_norm": 0.058457449078559875, "global_step": 16520, "epoch": 155, "lr": 7.788069520000338e-05} +{"train_loss": 0.017050856724381447, "grad_norm": 0.07570943981409073, "global_step": 16521, "epoch": 155, "lr": 7.787823492703216e-05} +{"train_loss": 0.01590866595506668, "grad_norm": 0.07859206199645996, "global_step": 16522, "epoch": 155, "lr": 7.787577455610893e-05} +{"train_loss": 0.016315758228302002, "grad_norm": 0.12173677235841751, "global_step": 16523, "epoch": 155, "lr": 7.787331408724237e-05} +{"train_loss": 0.01725040003657341, "grad_norm": 0.10708409547805786, "global_step": 16524, "epoch": 155, "lr": 7.787085352044108e-05} +{"train_loss": 0.016028299927711487, "grad_norm": 0.06270268559455872, "global_step": 16525, "epoch": 155, "lr": 7.786839285571375e-05} +{"train_loss": 0.015284955501556396, "grad_norm": 0.06868349015712738, "global_step": 16526, "epoch": 155, "lr": 7.7865932093069e-05} +{"train_loss": 0.01573478803038597, "grad_norm": 0.1080661490559578, "global_step": 16527, "epoch": 155, "lr": 7.786347123251549e-05} +{"train_loss": 0.015101495198905468, "grad_norm": 0.04671895131468773, "global_step": 16528, "epoch": 155, "lr": 7.786101027406185e-05} +{"train_loss": 0.017680421471595764, "grad_norm": 0.057259514927864075, "global_step": 16529, "epoch": 155, "lr": 7.785854921771674e-05} +{"train_loss": 0.017522716894745827, "grad_norm": 0.10234397649765015, "global_step": 16530, "epoch": 155, "lr": 7.785608806348881e-05} +{"train_loss": 0.016550015658140182, "grad_norm": 0.06845614314079285, "global_step": 16531, "epoch": 155, "lr": 7.78536268113867e-05} +{"train_loss": 0.015430420637130737, "grad_norm": 0.07137191295623779, "global_step": 16532, "epoch": 155, "lr": 7.785116546141905e-05} +{"train_loss": 0.018177611753344536, "grad_norm": 0.11026351153850555, "global_step": 16533, "epoch": 155, "lr": 7.784870401359452e-05} +{"train_loss": 0.017810598015785217, "grad_norm": 0.09444023668766022, "global_step": 16534, "epoch": 155, "lr": 7.784624246792176e-05} +{"train_loss": 0.016524691732143455, "grad_norm": 0.07940012216567993, "global_step": 16535, "epoch": 155, "lr": 7.784378082440941e-05, "val_loss": 0.018923867493867874} +{"train_loss": 0.016715046018362045, "grad_norm": 0.0599551796913147, "global_step": 16536, "epoch": 156, "lr": 7.784131908306611e-05} +{"train_loss": 0.016624420881271362, "grad_norm": 0.057362984865903854, "global_step": 16537, "epoch": 156, "lr": 7.783885724390055e-05} +{"train_loss": 0.017155352979898453, "grad_norm": 0.1595297008752823, "global_step": 16538, "epoch": 156, "lr": 7.783639530692133e-05} +{"train_loss": 0.015616336837410927, "grad_norm": 0.07116656005382538, "global_step": 16539, "epoch": 156, "lr": 7.783393327213713e-05} +{"train_loss": 0.017752304673194885, "grad_norm": 0.061775121837854385, "global_step": 16540, "epoch": 156, "lr": 7.78314711395566e-05} +{"train_loss": 0.01442296989262104, "grad_norm": 0.10962101072072983, "global_step": 16541, "epoch": 156, "lr": 7.782900890918839e-05} +{"train_loss": 0.016959311440587044, "grad_norm": 0.06429335474967957, "global_step": 16542, "epoch": 156, "lr": 7.782654658104111e-05} +{"train_loss": 0.014804154634475708, "grad_norm": 0.08720134943723679, "global_step": 16543, "epoch": 156, "lr": 7.782408415512348e-05} +{"train_loss": 0.01947498694062233, "grad_norm": 0.05001262575387955, "global_step": 16544, "epoch": 156, "lr": 7.782162163144411e-05} +{"train_loss": 0.016811713576316833, "grad_norm": 0.04929021745920181, "global_step": 16545, "epoch": 156, "lr": 7.781915901001165e-05} +{"train_loss": 0.01806582324206829, "grad_norm": 0.17433872818946838, "global_step": 16546, "epoch": 156, "lr": 7.781669629083479e-05} +{"train_loss": 0.017705654725432396, "grad_norm": 0.05888622626662254, "global_step": 16547, "epoch": 156, "lr": 7.781423347392213e-05} +{"train_loss": 0.016429925337433815, "grad_norm": 0.09829378128051758, "global_step": 16548, "epoch": 156, "lr": 7.781177055928235e-05} +{"train_loss": 0.01443831529468298, "grad_norm": 0.0685439482331276, "global_step": 16549, "epoch": 156, "lr": 7.780930754692413e-05} +{"train_loss": 0.014042952097952366, "grad_norm": 0.06562751531600952, "global_step": 16550, "epoch": 156, "lr": 7.780684443685607e-05} +{"train_loss": 0.016541998833417892, "grad_norm": 0.07904151827096939, "global_step": 16551, "epoch": 156, "lr": 7.780438122908686e-05} +{"train_loss": 0.016376953572034836, "grad_norm": 0.0885089710354805, "global_step": 16552, "epoch": 156, "lr": 7.780191792362515e-05} +{"train_loss": 0.017258811742067337, "grad_norm": 0.06394676119089127, "global_step": 16553, "epoch": 156, "lr": 7.779945452047958e-05} +{"train_loss": 0.017310278490185738, "grad_norm": 0.13628284633159637, "global_step": 16554, "epoch": 156, "lr": 7.779699101965883e-05} +{"train_loss": 0.017484549432992935, "grad_norm": 0.09098999202251434, "global_step": 16555, "epoch": 156, "lr": 7.779452742117155e-05} +{"train_loss": 0.017492083832621574, "grad_norm": 0.09485159069299698, "global_step": 16556, "epoch": 156, "lr": 7.779206372502637e-05} +{"train_loss": 0.018187854439020157, "grad_norm": 0.0691891685128212, "global_step": 16557, "epoch": 156, "lr": 7.778959993123195e-05} +{"train_loss": 0.016868146136403084, "grad_norm": 0.06954730302095413, "global_step": 16558, "epoch": 156, "lr": 7.778713603979699e-05} +{"train_loss": 0.018822800368070602, "grad_norm": 0.1292111575603485, "global_step": 16559, "epoch": 156, "lr": 7.778467205073013e-05} +{"train_loss": 0.01763429306447506, "grad_norm": 0.13376958668231964, "global_step": 16560, "epoch": 156, "lr": 7.778220796403997e-05} +{"train_loss": 0.016302116215229034, "grad_norm": 0.05564875900745392, "global_step": 16561, "epoch": 156, "lr": 7.777974377973525e-05} +{"train_loss": 0.020696144551038742, "grad_norm": 0.09626412391662598, "global_step": 16562, "epoch": 156, "lr": 7.777727949782458e-05} +{"train_loss": 0.014254494570195675, "grad_norm": 0.06715654581785202, "global_step": 16563, "epoch": 156, "lr": 7.777481511831664e-05} +{"train_loss": 0.01472516730427742, "grad_norm": 0.07227525115013123, "global_step": 16564, "epoch": 156, "lr": 7.777235064122007e-05} +{"train_loss": 0.016957931220531464, "grad_norm": 0.09808256477117538, "global_step": 16565, "epoch": 156, "lr": 7.776988606654352e-05} +{"train_loss": 0.013501705601811409, "grad_norm": 0.049414437264204025, "global_step": 16566, "epoch": 156, "lr": 7.77674213942957e-05} +{"train_loss": 0.016907453536987305, "grad_norm": 0.07853830605745316, "global_step": 16567, "epoch": 156, "lr": 7.77649566244852e-05} +{"train_loss": 0.01642788201570511, "grad_norm": 0.12346579134464264, "global_step": 16568, "epoch": 156, "lr": 7.776249175712073e-05} +{"train_loss": 0.015954341739416122, "grad_norm": 0.12426995486021042, "global_step": 16569, "epoch": 156, "lr": 7.776002679221097e-05} +{"train_loss": 0.016254514455795288, "grad_norm": 0.09298672527074814, "global_step": 16570, "epoch": 156, "lr": 7.775756172976452e-05} +{"train_loss": 0.015534867532551289, "grad_norm": 0.07764849811792374, "global_step": 16571, "epoch": 156, "lr": 7.775509656979006e-05} +{"train_loss": 0.016683051362633705, "grad_norm": 0.10757061094045639, "global_step": 16572, "epoch": 156, "lr": 7.775263131229629e-05} +{"train_loss": 0.016546064987778664, "grad_norm": 0.08437192440032959, "global_step": 16573, "epoch": 156, "lr": 7.77501659572918e-05} +{"train_loss": 0.01617138460278511, "grad_norm": 0.13028226792812347, "global_step": 16574, "epoch": 156, "lr": 7.774770050478534e-05} +{"train_loss": 0.015830909833312035, "grad_norm": 0.0979337990283966, "global_step": 16575, "epoch": 156, "lr": 7.774523495478552e-05} +{"train_loss": 0.01895652338862419, "grad_norm": 0.09653473645448685, "global_step": 16576, "epoch": 156, "lr": 7.774276930730098e-05} +{"train_loss": 0.020149260759353638, "grad_norm": 0.09176543354988098, "global_step": 16577, "epoch": 156, "lr": 7.774030356234044e-05} +{"train_loss": 0.017155520617961884, "grad_norm": 0.17581091821193695, "global_step": 16578, "epoch": 156, "lr": 7.773783771991253e-05} +{"train_loss": 0.01976652815937996, "grad_norm": 0.12226743996143341, "global_step": 16579, "epoch": 156, "lr": 7.773537178002593e-05} +{"train_loss": 0.014514806680381298, "grad_norm": 0.06635064631700516, "global_step": 16580, "epoch": 156, "lr": 7.773290574268927e-05} +{"train_loss": 0.01597265526652336, "grad_norm": 0.05682876706123352, "global_step": 16581, "epoch": 156, "lr": 7.773043960791126e-05} +{"train_loss": 0.01609991118311882, "grad_norm": 0.056399937719106674, "global_step": 16582, "epoch": 156, "lr": 7.772797337570054e-05} +{"train_loss": 0.01584792137145996, "grad_norm": 0.10232427716255188, "global_step": 16583, "epoch": 156, "lr": 7.772550704606578e-05} +{"train_loss": 0.016985295340418816, "grad_norm": 0.072697214782238, "global_step": 16584, "epoch": 156, "lr": 7.772304061901565e-05} +{"train_loss": 0.014635933563113213, "grad_norm": 0.05339365452528, "global_step": 16585, "epoch": 156, "lr": 7.772057409455882e-05} +{"train_loss": 0.014791854657232761, "grad_norm": 0.06410782039165497, "global_step": 16586, "epoch": 156, "lr": 7.771810747270394e-05} +{"train_loss": 0.017803415656089783, "grad_norm": 0.053788457065820694, "global_step": 16587, "epoch": 156, "lr": 7.771564075345967e-05} +{"train_loss": 0.017416158691048622, "grad_norm": 0.04842124879360199, "global_step": 16588, "epoch": 156, "lr": 7.771317393683471e-05} +{"train_loss": 0.014625778421759605, "grad_norm": 0.08364487439393997, "global_step": 16589, "epoch": 156, "lr": 7.77107070228377e-05} +{"train_loss": 0.016804058104753494, "grad_norm": 0.05811478942632675, "global_step": 16590, "epoch": 156, "lr": 7.770824001147732e-05} +{"train_loss": 0.017691845074295998, "grad_norm": 0.09679017215967178, "global_step": 16591, "epoch": 156, "lr": 7.770577290276225e-05} +{"train_loss": 0.01617126725614071, "grad_norm": 0.083589568734169, "global_step": 16592, "epoch": 156, "lr": 7.77033056967011e-05} +{"train_loss": 0.01688091829419136, "grad_norm": 0.07390958070755005, "global_step": 16593, "epoch": 156, "lr": 7.770083839330262e-05} +{"train_loss": 0.01936301961541176, "grad_norm": 0.11916512995958328, "global_step": 16594, "epoch": 156, "lr": 7.769837099257542e-05} +{"train_loss": 0.015434741042554379, "grad_norm": 0.1598595827817917, "global_step": 16595, "epoch": 156, "lr": 7.76959034945282e-05} +{"train_loss": 0.013688869774341583, "grad_norm": 0.05386875197291374, "global_step": 16596, "epoch": 156, "lr": 7.769343589916961e-05} +{"train_loss": 0.018749551847577095, "grad_norm": 0.08874478191137314, "global_step": 16597, "epoch": 156, "lr": 7.769096820650836e-05} +{"train_loss": 0.015812072902917862, "grad_norm": 0.15312153100967407, "global_step": 16598, "epoch": 156, "lr": 7.768850041655306e-05} +{"train_loss": 0.017071716487407684, "grad_norm": 0.06567858904600143, "global_step": 16599, "epoch": 156, "lr": 7.768603252931243e-05} +{"train_loss": 0.016383133828639984, "grad_norm": 0.11707835644483566, "global_step": 16600, "epoch": 156, "lr": 7.76835645447951e-05} +{"train_loss": 0.017581814900040627, "grad_norm": 0.12320347875356674, "global_step": 16601, "epoch": 156, "lr": 7.768109646300978e-05} +{"train_loss": 0.016024962067604065, "grad_norm": 0.11849816888570786, "global_step": 16602, "epoch": 156, "lr": 7.767862828396512e-05} +{"train_loss": 0.014724694192409515, "grad_norm": 0.08118749409914017, "global_step": 16603, "epoch": 156, "lr": 7.76761600076698e-05} +{"train_loss": 0.020162925124168396, "grad_norm": 0.08048942685127258, "global_step": 16604, "epoch": 156, "lr": 7.767369163413248e-05} +{"train_loss": 0.016408298164606094, "grad_norm": 0.05488108843564987, "global_step": 16605, "epoch": 156, "lr": 7.767122316336186e-05} +{"train_loss": 0.01674548164010048, "grad_norm": 0.07217881828546524, "global_step": 16606, "epoch": 156, "lr": 7.766875459536658e-05} +{"train_loss": 0.016436096280813217, "grad_norm": 0.07480203360319138, "global_step": 16607, "epoch": 156, "lr": 7.766628593015534e-05} +{"train_loss": 0.01695876568555832, "grad_norm": 0.061419084668159485, "global_step": 16608, "epoch": 156, "lr": 7.766381716773679e-05} +{"train_loss": 0.01672542281448841, "grad_norm": 0.07225492596626282, "global_step": 16609, "epoch": 156, "lr": 7.766134830811962e-05} +{"train_loss": 0.01619839295744896, "grad_norm": 0.07088025659322739, "global_step": 16610, "epoch": 156, "lr": 7.76588793513125e-05} +{"train_loss": 0.01440689992159605, "grad_norm": 0.11246234178543091, "global_step": 16611, "epoch": 156, "lr": 7.765641029732413e-05} +{"train_loss": 0.016154741868376732, "grad_norm": 0.07871216535568237, "global_step": 16612, "epoch": 156, "lr": 7.765394114616313e-05} +{"train_loss": 0.014800542034208775, "grad_norm": 0.0909244492650032, "global_step": 16613, "epoch": 156, "lr": 7.765147189783823e-05} +{"train_loss": 0.017473282292485237, "grad_norm": 0.10601646453142166, "global_step": 16614, "epoch": 156, "lr": 7.764900255235808e-05} +{"train_loss": 0.01618688553571701, "grad_norm": 0.08974675089120865, "global_step": 16615, "epoch": 156, "lr": 7.764653310973136e-05} +{"train_loss": 0.017068367451429367, "grad_norm": 0.10876256227493286, "global_step": 16616, "epoch": 156, "lr": 7.764406356996675e-05} +{"train_loss": 0.015751443803310394, "grad_norm": 0.0626518726348877, "global_step": 16617, "epoch": 156, "lr": 7.76415939330729e-05} +{"train_loss": 0.013506382703781128, "grad_norm": 0.15578900277614594, "global_step": 16618, "epoch": 156, "lr": 7.763912419905852e-05} +{"train_loss": 0.01592717133462429, "grad_norm": 0.08152294903993607, "global_step": 16619, "epoch": 156, "lr": 7.763665436793228e-05} +{"train_loss": 0.014798944815993309, "grad_norm": 0.07555702328681946, "global_step": 16620, "epoch": 156, "lr": 7.763418443970286e-05} +{"train_loss": 0.018440844491124153, "grad_norm": 0.07191561907529831, "global_step": 16621, "epoch": 156, "lr": 7.763171441437894e-05} +{"train_loss": 0.018437165766954422, "grad_norm": 0.09166269749403, "global_step": 16622, "epoch": 156, "lr": 7.762924429196917e-05} +{"train_loss": 0.01580648496747017, "grad_norm": 0.09070006012916565, "global_step": 16623, "epoch": 156, "lr": 7.762677407248227e-05} +{"train_loss": 0.015947189182043076, "grad_norm": 0.047493670135736465, "global_step": 16624, "epoch": 156, "lr": 7.762430375592688e-05} +{"train_loss": 0.01509038731455803, "grad_norm": 0.05265912041068077, "global_step": 16625, "epoch": 156, "lr": 7.762183334231172e-05} +{"train_loss": 0.01598210074007511, "grad_norm": 0.05100281164050102, "global_step": 16626, "epoch": 156, "lr": 7.761936283164544e-05} +{"train_loss": 0.015928152948617935, "grad_norm": 0.09758557379245758, "global_step": 16627, "epoch": 156, "lr": 7.761689222393674e-05} +{"train_loss": 0.01880071684718132, "grad_norm": 0.13160023093223572, "global_step": 16628, "epoch": 156, "lr": 7.76144215191943e-05} +{"train_loss": 0.01451211329549551, "grad_norm": 0.046488743275403976, "global_step": 16629, "epoch": 156, "lr": 7.761195071742675e-05} +{"train_loss": 0.01631660759449005, "grad_norm": 0.04834868758916855, "global_step": 16630, "epoch": 156, "lr": 7.760947981864285e-05} +{"train_loss": 0.017104608938097954, "grad_norm": 0.07043763995170593, "global_step": 16631, "epoch": 156, "lr": 7.760700882285124e-05} +{"train_loss": 0.015500232577323914, "grad_norm": 0.05722678080201149, "global_step": 16632, "epoch": 156, "lr": 7.76045377300606e-05} +{"train_loss": 0.016184084117412567, "grad_norm": 0.06733647733926773, "global_step": 16633, "epoch": 156, "lr": 7.760206654027961e-05} +{"train_loss": 0.014803491532802582, "grad_norm": 0.0819176733493805, "global_step": 16634, "epoch": 156, "lr": 7.759959525351698e-05} +{"train_loss": 0.015550365671515465, "grad_norm": 0.08849659562110901, "global_step": 16635, "epoch": 156, "lr": 7.759712386978137e-05} +{"train_loss": 0.019655099138617516, "grad_norm": 0.07326085865497589, "global_step": 16636, "epoch": 156, "lr": 7.759465238908145e-05} +{"train_loss": 0.016218500211834908, "grad_norm": 0.09221283346414566, "global_step": 16637, "epoch": 156, "lr": 7.759218081142596e-05} +{"train_loss": 0.016339663416147232, "grad_norm": 0.0829700380563736, "global_step": 16638, "epoch": 156, "lr": 7.758970913682351e-05} +{"train_loss": 0.0157717764377594, "grad_norm": 0.05073500797152519, "global_step": 16639, "epoch": 156, "lr": 7.758723736528284e-05} +{"train_loss": 0.018398528918623924, "grad_norm": 0.08385580778121948, "global_step": 16640, "epoch": 156, "lr": 7.758476549681262e-05} +{"train_loss": 0.016520095874889278, "grad_norm": 0.08227008581161499, "global_step": 16641, "epoch": 156, "lr": 7.758229353142152e-05, "val_loss": 0.018782781437039375} +{"train_loss": 0.017312731593847275, "grad_norm": 0.08710628747940063, "global_step": 16642, "epoch": 157, "lr": 7.757982146911824e-05} +{"train_loss": 0.016927331686019897, "grad_norm": 0.07281142473220825, "global_step": 16643, "epoch": 157, "lr": 7.757734930991145e-05} +{"train_loss": 0.015605488792061806, "grad_norm": 0.09338519722223282, "global_step": 16644, "epoch": 157, "lr": 7.757487705380985e-05} +{"train_loss": 0.015006412751972675, "grad_norm": 0.10381820797920227, "global_step": 16645, "epoch": 157, "lr": 7.757240470082214e-05} +{"train_loss": 0.018363093957304955, "grad_norm": 0.06932274252176285, "global_step": 16646, "epoch": 157, "lr": 7.756993225095699e-05} +{"train_loss": 0.01692131720483303, "grad_norm": 0.05335591733455658, "global_step": 16647, "epoch": 157, "lr": 7.756745970422306e-05} +{"train_loss": 0.01848415657877922, "grad_norm": 0.1466921865940094, "global_step": 16648, "epoch": 157, "lr": 7.756498706062909e-05} +{"train_loss": 0.017913468182086945, "grad_norm": 0.22726917266845703, "global_step": 16649, "epoch": 157, "lr": 7.756251432018375e-05} +{"train_loss": 0.015317639335989952, "grad_norm": 0.07923774421215057, "global_step": 16650, "epoch": 157, "lr": 7.75600414828957e-05} +{"train_loss": 0.01870805211365223, "grad_norm": 0.10815320909023285, "global_step": 16651, "epoch": 157, "lr": 7.755756854877366e-05} +{"train_loss": 0.01498553529381752, "grad_norm": 0.06298542022705078, "global_step": 16652, "epoch": 157, "lr": 7.75550955178263e-05} +{"train_loss": 0.017140112817287445, "grad_norm": 0.178767591714859, "global_step": 16653, "epoch": 157, "lr": 7.755262239006232e-05} +{"train_loss": 0.014588434249162674, "grad_norm": 0.085031658411026, "global_step": 16654, "epoch": 157, "lr": 7.755014916549041e-05} +{"train_loss": 0.017827430739998817, "grad_norm": 0.1141417846083641, "global_step": 16655, "epoch": 157, "lr": 7.754767584411927e-05} +{"train_loss": 0.01722778007388115, "grad_norm": 0.08825987577438354, "global_step": 16656, "epoch": 157, "lr": 7.754520242595756e-05} +{"train_loss": 0.018571874126791954, "grad_norm": 0.0743182972073555, "global_step": 16657, "epoch": 157, "lr": 7.7542728911014e-05} +{"train_loss": 0.01547561027109623, "grad_norm": 0.12300998717546463, "global_step": 16658, "epoch": 157, "lr": 7.754025529929727e-05} +{"train_loss": 0.01820651814341545, "grad_norm": 0.06189408525824547, "global_step": 16659, "epoch": 157, "lr": 7.753778159081603e-05} +{"train_loss": 0.01625503972172737, "grad_norm": 0.1493532806634903, "global_step": 16660, "epoch": 157, "lr": 7.753530778557904e-05} +{"train_loss": 0.018083356320858, "grad_norm": 0.21031871438026428, "global_step": 16661, "epoch": 157, "lr": 7.753283388359494e-05} +{"train_loss": 0.012370915152132511, "grad_norm": 0.04930544272065163, "global_step": 16662, "epoch": 157, "lr": 7.753035988487243e-05} +{"train_loss": 0.014021039940416813, "grad_norm": 0.14477288722991943, "global_step": 16663, "epoch": 157, "lr": 7.752788578942021e-05} +{"train_loss": 0.014288361184298992, "grad_norm": 0.14924848079681396, "global_step": 16664, "epoch": 157, "lr": 7.752541159724698e-05} +{"train_loss": 0.018097583204507828, "grad_norm": 0.13316598534584045, "global_step": 16665, "epoch": 157, "lr": 7.752293730836141e-05} +{"train_loss": 0.015206446871161461, "grad_norm": 0.05645846948027611, "global_step": 16666, "epoch": 157, "lr": 7.752046292277222e-05} +{"train_loss": 0.017477940768003464, "grad_norm": 0.21678376197814941, "global_step": 16667, "epoch": 157, "lr": 7.751798844048808e-05} +{"train_loss": 0.013573436066508293, "grad_norm": 0.03884104639291763, "global_step": 16668, "epoch": 157, "lr": 7.75155138615177e-05} +{"train_loss": 0.015567801892757416, "grad_norm": 0.055184733122587204, "global_step": 16669, "epoch": 157, "lr": 7.751303918586976e-05} +{"train_loss": 0.016074050217866898, "grad_norm": 0.07109038531780243, "global_step": 16670, "epoch": 157, "lr": 7.751056441355297e-05} +{"train_loss": 0.016805635765194893, "grad_norm": 0.14484632015228271, "global_step": 16671, "epoch": 157, "lr": 7.750808954457605e-05} +{"train_loss": 0.015898320823907852, "grad_norm": 0.087053082883358, "global_step": 16672, "epoch": 157, "lr": 7.750561457894764e-05} +{"train_loss": 0.016121625900268555, "grad_norm": 0.06590700149536133, "global_step": 16673, "epoch": 157, "lr": 7.750313951667646e-05} +{"train_loss": 0.01639411598443985, "grad_norm": 0.11183208227157593, "global_step": 16674, "epoch": 157, "lr": 7.750066435777122e-05} +{"train_loss": 0.01703605428338051, "grad_norm": 0.0729459747672081, "global_step": 16675, "epoch": 157, "lr": 7.74981891022406e-05} +{"train_loss": 0.015705348923802376, "grad_norm": 0.1588049829006195, "global_step": 16676, "epoch": 157, "lr": 7.74957137500933e-05} +{"train_loss": 0.015383584424853325, "grad_norm": 0.10271337628364563, "global_step": 16677, "epoch": 157, "lr": 7.749323830133801e-05} +{"train_loss": 0.016535744071006775, "grad_norm": 0.22128921747207642, "global_step": 16678, "epoch": 157, "lr": 7.749076275598346e-05} +{"train_loss": 0.01514347456395626, "grad_norm": 0.12998057901859283, "global_step": 16679, "epoch": 157, "lr": 7.74882871140383e-05} +{"train_loss": 0.01638994924724102, "grad_norm": 0.14528897404670715, "global_step": 16680, "epoch": 157, "lr": 7.748581137551127e-05} +{"train_loss": 0.01691269874572754, "grad_norm": 0.07513415068387985, "global_step": 16681, "epoch": 157, "lr": 7.748333554041105e-05} +{"train_loss": 0.018077978864312172, "grad_norm": 0.1048295721411705, "global_step": 16682, "epoch": 157, "lr": 7.748085960874634e-05} +{"train_loss": 0.015656081959605217, "grad_norm": 0.11613134294748306, "global_step": 16683, "epoch": 157, "lr": 7.747838358052584e-05} +{"train_loss": 0.01591850444674492, "grad_norm": 0.0462447851896286, "global_step": 16684, "epoch": 157, "lr": 7.747590745575825e-05} +{"train_loss": 0.016787629574537277, "grad_norm": 0.05270972102880478, "global_step": 16685, "epoch": 157, "lr": 7.747343123445228e-05} +{"train_loss": 0.015465530566871166, "grad_norm": 0.07378225028514862, "global_step": 16686, "epoch": 157, "lr": 7.74709549166166e-05} +{"train_loss": 0.01623978279531002, "grad_norm": 0.08143425732851028, "global_step": 16687, "epoch": 157, "lr": 7.746847850225996e-05} +{"train_loss": 0.017872771248221397, "grad_norm": 0.13167159259319305, "global_step": 16688, "epoch": 157, "lr": 7.7466001991391e-05} +{"train_loss": 0.015487346798181534, "grad_norm": 0.06381972879171371, "global_step": 16689, "epoch": 157, "lr": 7.746352538401846e-05} +{"train_loss": 0.018069103360176086, "grad_norm": 0.17782138288021088, "global_step": 16690, "epoch": 157, "lr": 7.746104868015107e-05} +{"train_loss": 0.014978058636188507, "grad_norm": 0.10125172883272171, "global_step": 16691, "epoch": 157, "lr": 7.745857187979745e-05} +{"train_loss": 0.017043787986040115, "grad_norm": 0.0546160452067852, "global_step": 16692, "epoch": 157, "lr": 7.745609498296638e-05} +{"train_loss": 0.016990941017866135, "grad_norm": 0.09754692763090134, "global_step": 16693, "epoch": 157, "lr": 7.745361798966653e-05} +{"train_loss": 0.015085466206073761, "grad_norm": 0.1675204187631607, "global_step": 16694, "epoch": 157, "lr": 7.74511408999066e-05} +{"train_loss": 0.015887470915913582, "grad_norm": 0.0739717036485672, "global_step": 16695, "epoch": 157, "lr": 7.744866371369529e-05} +{"train_loss": 0.015762750059366226, "grad_norm": 0.06690430641174316, "global_step": 16696, "epoch": 157, "lr": 7.744618643104132e-05} +{"train_loss": 0.015908977016806602, "grad_norm": 0.15015427768230438, "global_step": 16697, "epoch": 157, "lr": 7.744370905195339e-05} +{"train_loss": 0.016717886552214622, "grad_norm": 0.13292980194091797, "global_step": 16698, "epoch": 157, "lr": 7.744123157644019e-05} +{"train_loss": 0.01825406216084957, "grad_norm": 0.05983138829469681, "global_step": 16699, "epoch": 157, "lr": 7.743875400451047e-05} +{"train_loss": 0.01552608609199524, "grad_norm": 0.12526781857013702, "global_step": 16700, "epoch": 157, "lr": 7.743627633617287e-05} +{"train_loss": 0.018256202340126038, "grad_norm": 0.11194369941949844, "global_step": 16701, "epoch": 157, "lr": 7.743379857143613e-05} +{"train_loss": 0.017156528308987617, "grad_norm": 0.14712944626808167, "global_step": 16702, "epoch": 157, "lr": 7.743132071030895e-05} +{"train_loss": 0.014907702803611755, "grad_norm": 0.07944370061159134, "global_step": 16703, "epoch": 157, "lr": 7.742884275280005e-05} +{"train_loss": 0.015580043196678162, "grad_norm": 0.057180795818567276, "global_step": 16704, "epoch": 157, "lr": 7.74263646989181e-05} +{"train_loss": 0.014578256756067276, "grad_norm": 0.09039764851331711, "global_step": 16705, "epoch": 157, "lr": 7.742388654867187e-05} +{"train_loss": 0.02005530521273613, "grad_norm": 0.1581229418516159, "global_step": 16706, "epoch": 157, "lr": 7.742140830207001e-05} +{"train_loss": 0.015285130590200424, "grad_norm": 0.09074673801660538, "global_step": 16707, "epoch": 157, "lr": 7.741892995912123e-05} +{"train_loss": 0.01793753169476986, "grad_norm": 0.11381660401821136, "global_step": 16708, "epoch": 157, "lr": 7.741645151983429e-05} +{"train_loss": 0.01600385084748268, "grad_norm": 0.05381712689995766, "global_step": 16709, "epoch": 157, "lr": 7.741397298421784e-05} +{"train_loss": 0.017252974212169647, "grad_norm": 0.07298114150762558, "global_step": 16710, "epoch": 157, "lr": 7.741149435228061e-05} +{"train_loss": 0.014599225483834743, "grad_norm": 0.09109818935394287, "global_step": 16711, "epoch": 157, "lr": 7.740901562403132e-05} +{"train_loss": 0.01759733259677887, "grad_norm": 0.06545591354370117, "global_step": 16712, "epoch": 157, "lr": 7.740653679947867e-05} +{"train_loss": 0.016009630635380745, "grad_norm": 0.07155317068099976, "global_step": 16713, "epoch": 157, "lr": 7.740405787863134e-05} +{"train_loss": 0.01835278794169426, "grad_norm": 0.05934864655137062, "global_step": 16714, "epoch": 157, "lr": 7.740157886149811e-05} +{"train_loss": 0.019507242366671562, "grad_norm": 0.11425304412841797, "global_step": 16715, "epoch": 157, "lr": 7.73990997480876e-05} +{"train_loss": 0.019103899598121643, "grad_norm": 0.06468339264392853, "global_step": 16716, "epoch": 157, "lr": 7.739662053840861e-05} +{"train_loss": 0.018413597717881203, "grad_norm": 0.05664927884936333, "global_step": 16717, "epoch": 157, "lr": 7.739414123246981e-05} +{"train_loss": 0.015526466071605682, "grad_norm": 0.04919053986668587, "global_step": 16718, "epoch": 157, "lr": 7.739166183027989e-05} +{"train_loss": 0.01767619326710701, "grad_norm": 0.10052013397216797, "global_step": 16719, "epoch": 157, "lr": 7.738918233184759e-05} +{"train_loss": 0.016936030238866806, "grad_norm": 0.11022308468818665, "global_step": 16720, "epoch": 157, "lr": 7.738670273718163e-05} +{"train_loss": 0.01944660395383835, "grad_norm": 0.10053467750549316, "global_step": 16721, "epoch": 157, "lr": 7.738422304629068e-05} +{"train_loss": 0.018376359716057777, "grad_norm": 0.20803526043891907, "global_step": 16722, "epoch": 157, "lr": 7.73817432591835e-05} +{"train_loss": 0.017666134983301163, "grad_norm": 0.11132723093032837, "global_step": 16723, "epoch": 157, "lr": 7.737926337586877e-05} +{"train_loss": 0.015986204147338867, "grad_norm": 0.08368849009275436, "global_step": 16724, "epoch": 157, "lr": 7.737678339635522e-05} +{"train_loss": 0.018983300775289536, "grad_norm": 0.07542780786752701, "global_step": 16725, "epoch": 157, "lr": 7.737430332065156e-05} +{"train_loss": 0.015868697315454483, "grad_norm": 0.07034873217344284, "global_step": 16726, "epoch": 157, "lr": 7.73718231487665e-05} +{"train_loss": 0.016776803880929947, "grad_norm": 0.062263648957014084, "global_step": 16727, "epoch": 157, "lr": 7.736934288070877e-05} +{"train_loss": 0.015211566351354122, "grad_norm": 0.0981661006808281, "global_step": 16728, "epoch": 157, "lr": 7.736686251648704e-05} +{"train_loss": 0.016547322273254395, "grad_norm": 0.058403387665748596, "global_step": 16729, "epoch": 157, "lr": 7.73643820561101e-05} +{"train_loss": 0.014850407838821411, "grad_norm": 0.06906774640083313, "global_step": 16730, "epoch": 157, "lr": 7.736190149958658e-05} +{"train_loss": 0.019087299704551697, "grad_norm": 0.12191828340291977, "global_step": 16731, "epoch": 157, "lr": 7.735942084692526e-05} +{"train_loss": 0.01531812734901905, "grad_norm": 0.08380360901355743, "global_step": 16732, "epoch": 157, "lr": 7.735694009813483e-05} +{"train_loss": 0.01731184497475624, "grad_norm": 0.11738578230142593, "global_step": 16733, "epoch": 157, "lr": 7.7354459253224e-05} +{"train_loss": 0.014750992879271507, "grad_norm": 0.10621483623981476, "global_step": 16734, "epoch": 157, "lr": 7.735197831220151e-05} +{"train_loss": 0.015918780118227005, "grad_norm": 0.12234146893024445, "global_step": 16735, "epoch": 157, "lr": 7.734949727507606e-05} +{"train_loss": 0.017913764342665672, "grad_norm": 0.09195522218942642, "global_step": 16736, "epoch": 157, "lr": 7.734701614185638e-05} +{"train_loss": 0.016107849776744843, "grad_norm": 0.08105767518281937, "global_step": 16737, "epoch": 157, "lr": 7.734453491255115e-05} +{"train_loss": 0.01672980561852455, "grad_norm": 0.10847173631191254, "global_step": 16738, "epoch": 157, "lr": 7.734205358716915e-05} +{"train_loss": 0.01832696981728077, "grad_norm": 0.16109120845794678, "global_step": 16739, "epoch": 157, "lr": 7.733957216571905e-05} +{"train_loss": 0.015489794313907623, "grad_norm": 0.06391629576683044, "global_step": 16740, "epoch": 157, "lr": 7.733709064820958e-05} +{"train_loss": 0.016097847372293472, "grad_norm": 0.06698612868785858, "global_step": 16741, "epoch": 157, "lr": 7.733460903464947e-05} +{"train_loss": 0.017196033149957657, "grad_norm": 0.07775815576314926, "global_step": 16742, "epoch": 157, "lr": 7.733212732504741e-05} +{"train_loss": 0.0160876102745533, "grad_norm": 0.07088912278413773, "global_step": 16743, "epoch": 157, "lr": 7.732964551941217e-05} +{"train_loss": 0.016351716592907906, "grad_norm": 0.04891616106033325, "global_step": 16744, "epoch": 157, "lr": 7.732716361775242e-05} +{"train_loss": 0.014976201578974724, "grad_norm": 0.0594186894595623, "global_step": 16745, "epoch": 157, "lr": 7.732468162007692e-05} +{"train_loss": 0.014913015067577362, "grad_norm": 0.08416306227445602, "global_step": 16746, "epoch": 157, "lr": 7.732219952639435e-05} +{"train_loss": 0.016563961115437297, "grad_norm": 0.07566406577825546, "global_step": 16747, "epoch": 157, "lr": 7.731971733671346e-05, "val_loss": 0.019382920116186142} +{"train_loss": 0.01750265620648861, "grad_norm": 0.07309070229530334, "global_step": 16748, "epoch": 158, "lr": 7.731723505104295e-05} +{"train_loss": 0.013266068883240223, "grad_norm": 0.10339947044849396, "global_step": 16749, "epoch": 158, "lr": 7.731475266939159e-05} +{"train_loss": 0.017647605389356613, "grad_norm": 0.14654967188835144, "global_step": 16750, "epoch": 158, "lr": 7.731227019176806e-05} +{"train_loss": 0.01672455482184887, "grad_norm": 0.15888845920562744, "global_step": 16751, "epoch": 158, "lr": 7.730978761818107e-05} +{"train_loss": 0.016666661947965622, "grad_norm": 0.06810978800058365, "global_step": 16752, "epoch": 158, "lr": 7.730730494863938e-05} +{"train_loss": 0.01671457104384899, "grad_norm": 0.22481752932071686, "global_step": 16753, "epoch": 158, "lr": 7.73048221831517e-05} +{"train_loss": 0.01998058520257473, "grad_norm": 0.07622046023607254, "global_step": 16754, "epoch": 158, "lr": 7.730233932172672e-05} +{"train_loss": 0.014064240269362926, "grad_norm": 0.16771413385868073, "global_step": 16755, "epoch": 158, "lr": 7.729985636437322e-05} +{"train_loss": 0.015835216268897057, "grad_norm": 0.06863424926996231, "global_step": 16756, "epoch": 158, "lr": 7.72973733110999e-05} +{"train_loss": 0.01473948359489441, "grad_norm": 0.05163348838686943, "global_step": 16757, "epoch": 158, "lr": 7.729489016191546e-05} +{"train_loss": 0.01779705472290516, "grad_norm": 0.08552654832601547, "global_step": 16758, "epoch": 158, "lr": 7.729240691682866e-05} +{"train_loss": 0.015600579790771008, "grad_norm": 0.05795343965291977, "global_step": 16759, "epoch": 158, "lr": 7.728992357584822e-05} +{"train_loss": 0.018747661262750626, "grad_norm": 0.17105481028556824, "global_step": 16760, "epoch": 158, "lr": 7.728744013898283e-05} +{"train_loss": 0.01569432020187378, "grad_norm": 0.08404164016246796, "global_step": 16761, "epoch": 158, "lr": 7.728495660624127e-05} +{"train_loss": 0.016057312488555908, "grad_norm": 0.12469396740198135, "global_step": 16762, "epoch": 158, "lr": 7.728247297763222e-05} +{"train_loss": 0.014876054599881172, "grad_norm": 0.13266703486442566, "global_step": 16763, "epoch": 158, "lr": 7.727998925316444e-05} +{"train_loss": 0.01699504815042019, "grad_norm": 0.08360137790441513, "global_step": 16764, "epoch": 158, "lr": 7.727750543284663e-05} +{"train_loss": 0.016792859882116318, "grad_norm": 0.11518945544958115, "global_step": 16765, "epoch": 158, "lr": 7.727502151668753e-05} +{"train_loss": 0.014118209481239319, "grad_norm": 0.09485239535570145, "global_step": 16766, "epoch": 158, "lr": 7.727253750469588e-05} +{"train_loss": 0.01877179928123951, "grad_norm": 0.05898645892739296, "global_step": 16767, "epoch": 158, "lr": 7.727005339688037e-05} +{"train_loss": 0.019239064306020737, "grad_norm": 0.12442246079444885, "global_step": 16768, "epoch": 158, "lr": 7.726756919324976e-05} +{"train_loss": 0.015980159863829613, "grad_norm": 0.05880589783191681, "global_step": 16769, "epoch": 158, "lr": 7.726508489381278e-05} +{"train_loss": 0.016225147992372513, "grad_norm": 0.06068945303559303, "global_step": 16770, "epoch": 158, "lr": 7.726260049857814e-05} +{"train_loss": 0.01380912959575653, "grad_norm": 0.07540369033813477, "global_step": 16771, "epoch": 158, "lr": 7.726011600755458e-05} +{"train_loss": 0.01648406870663166, "grad_norm": 0.05559805408120155, "global_step": 16772, "epoch": 158, "lr": 7.725763142075083e-05} +{"train_loss": 0.014592000283300877, "grad_norm": 0.08293800801038742, "global_step": 16773, "epoch": 158, "lr": 7.725514673817561e-05} +{"train_loss": 0.01605323515832424, "grad_norm": 0.07021674513816833, "global_step": 16774, "epoch": 158, "lr": 7.725266195983766e-05} +{"train_loss": 0.017219306901097298, "grad_norm": 0.10652478784322739, "global_step": 16775, "epoch": 158, "lr": 7.72501770857457e-05} +{"train_loss": 0.01594630628824234, "grad_norm": 0.08068746328353882, "global_step": 16776, "epoch": 158, "lr": 7.72476921159085e-05} +{"train_loss": 0.017456064000725746, "grad_norm": 0.06880416721105576, "global_step": 16777, "epoch": 158, "lr": 7.724520705033473e-05} +{"train_loss": 0.01998627558350563, "grad_norm": 0.10467396676540375, "global_step": 16778, "epoch": 158, "lr": 7.724272188903316e-05} +{"train_loss": 0.01765218749642372, "grad_norm": 0.1811106950044632, "global_step": 16779, "epoch": 158, "lr": 7.724023663201251e-05} +{"train_loss": 0.017286986112594604, "grad_norm": 0.1208028569817543, "global_step": 16780, "epoch": 158, "lr": 7.723775127928152e-05} +{"train_loss": 0.015215233899652958, "grad_norm": 0.07489262521266937, "global_step": 16781, "epoch": 158, "lr": 7.723526583084891e-05} +{"train_loss": 0.016688302159309387, "grad_norm": 0.13676245510578156, "global_step": 16782, "epoch": 158, "lr": 7.723278028672343e-05} +{"train_loss": 0.01696206070482731, "grad_norm": 0.07969005405902863, "global_step": 16783, "epoch": 158, "lr": 7.723029464691379e-05} +{"train_loss": 0.019001446664333344, "grad_norm": 0.08554897457361221, "global_step": 16784, "epoch": 158, "lr": 7.722780891142875e-05} +{"train_loss": 0.016213154420256615, "grad_norm": 0.050717875361442566, "global_step": 16785, "epoch": 158, "lr": 7.722532308027703e-05} +{"train_loss": 0.0162276029586792, "grad_norm": 0.0785892978310585, "global_step": 16786, "epoch": 158, "lr": 7.722283715346737e-05} +{"train_loss": 0.016417039558291435, "grad_norm": 0.07110006362199783, "global_step": 16787, "epoch": 158, "lr": 7.722035113100849e-05} +{"train_loss": 0.01706686243414879, "grad_norm": 0.09249052405357361, "global_step": 16788, "epoch": 158, "lr": 7.721786501290912e-05} +{"train_loss": 0.016099780797958374, "grad_norm": 0.12699918448925018, "global_step": 16789, "epoch": 158, "lr": 7.721537879917802e-05} +{"train_loss": 0.016145065426826477, "grad_norm": 0.0946129709482193, "global_step": 16790, "epoch": 158, "lr": 7.721289248982391e-05} +{"train_loss": 0.01767687499523163, "grad_norm": 0.13171228766441345, "global_step": 16791, "epoch": 158, "lr": 7.721040608485555e-05} +{"train_loss": 0.013594824820756912, "grad_norm": 0.05834289267659187, "global_step": 16792, "epoch": 158, "lr": 7.720791958428164e-05} +{"train_loss": 0.013602020218968391, "grad_norm": 0.09145577251911163, "global_step": 16793, "epoch": 158, "lr": 7.720543298811093e-05} +{"train_loss": 0.018337076529860497, "grad_norm": 0.061022862792015076, "global_step": 16794, "epoch": 158, "lr": 7.720294629635217e-05} +{"train_loss": 0.019465025514364243, "grad_norm": 0.13570739328861237, "global_step": 16795, "epoch": 158, "lr": 7.720045950901408e-05} +{"train_loss": 0.016236335039138794, "grad_norm": 0.1009458526968956, "global_step": 16796, "epoch": 158, "lr": 7.719797262610541e-05} +{"train_loss": 0.017445378005504608, "grad_norm": 0.06761082261800766, "global_step": 16797, "epoch": 158, "lr": 7.71954856476349e-05} +{"train_loss": 0.01651313155889511, "grad_norm": 0.08133523166179657, "global_step": 16798, "epoch": 158, "lr": 7.719299857361127e-05} +{"train_loss": 0.016714001074433327, "grad_norm": 0.06857401877641678, "global_step": 16799, "epoch": 158, "lr": 7.719051140404327e-05} +{"train_loss": 0.019460570067167282, "grad_norm": 0.10531237721443176, "global_step": 16800, "epoch": 158, "lr": 7.718802413893963e-05} +{"train_loss": 0.017510611563920975, "grad_norm": 0.08497582376003265, "global_step": 16801, "epoch": 158, "lr": 7.71855367783091e-05} +{"train_loss": 0.017659783363342285, "grad_norm": 0.05281788855791092, "global_step": 16802, "epoch": 158, "lr": 7.718304932216043e-05} +{"train_loss": 0.016701355576515198, "grad_norm": 0.08694794028997421, "global_step": 16803, "epoch": 158, "lr": 7.718056177050233e-05} +{"train_loss": 0.014522504061460495, "grad_norm": 0.08492754399776459, "global_step": 16804, "epoch": 158, "lr": 7.717807412334357e-05} +{"train_loss": 0.016997136175632477, "grad_norm": 0.06097264215350151, "global_step": 16805, "epoch": 158, "lr": 7.717558638069287e-05} +{"train_loss": 0.016713030636310577, "grad_norm": 0.10680493712425232, "global_step": 16806, "epoch": 158, "lr": 7.717309854255896e-05} +{"train_loss": 0.01676136627793312, "grad_norm": 0.13666459918022156, "global_step": 16807, "epoch": 158, "lr": 7.717061060895062e-05} +{"train_loss": 0.020307345315814018, "grad_norm": 0.05866008996963501, "global_step": 16808, "epoch": 158, "lr": 7.716812257987658e-05} +{"train_loss": 0.016599450260400772, "grad_norm": 0.08929470181465149, "global_step": 16809, "epoch": 158, "lr": 7.716563445534556e-05} +{"train_loss": 0.017201339825987816, "grad_norm": 0.07511591911315918, "global_step": 16810, "epoch": 158, "lr": 7.716314623536631e-05} +{"train_loss": 0.01653071865439415, "grad_norm": 0.061630427837371826, "global_step": 16811, "epoch": 158, "lr": 7.71606579199476e-05} +{"train_loss": 0.017296768724918365, "grad_norm": 0.11263523250818253, "global_step": 16812, "epoch": 158, "lr": 7.71581695090981e-05} +{"train_loss": 0.015756001695990562, "grad_norm": 0.05077522248029709, "global_step": 16813, "epoch": 158, "lr": 7.715568100282664e-05} +{"train_loss": 0.019544251263141632, "grad_norm": 0.05682169273495674, "global_step": 16814, "epoch": 158, "lr": 7.715319240114193e-05} +{"train_loss": 0.013567580841481686, "grad_norm": 0.09057500958442688, "global_step": 16815, "epoch": 158, "lr": 7.71507037040527e-05} +{"train_loss": 0.016053976491093636, "grad_norm": 0.11906079947948456, "global_step": 16816, "epoch": 158, "lr": 7.714821491156772e-05} +{"train_loss": 0.015617569908499718, "grad_norm": 0.11548137664794922, "global_step": 16817, "epoch": 158, "lr": 7.714572602369569e-05} +{"train_loss": 0.013783225789666176, "grad_norm": 0.0762588307261467, "global_step": 16818, "epoch": 158, "lr": 7.714323704044541e-05} +{"train_loss": 0.016609031707048416, "grad_norm": 0.08359956741333008, "global_step": 16819, "epoch": 158, "lr": 7.714074796182558e-05} +{"train_loss": 0.01676642708480358, "grad_norm": 0.11426820605993271, "global_step": 16820, "epoch": 158, "lr": 7.713825878784497e-05} +{"train_loss": 0.01850309781730175, "grad_norm": 0.06744576245546341, "global_step": 16821, "epoch": 158, "lr": 7.713576951851233e-05} +{"train_loss": 0.017049893736839294, "grad_norm": 0.08571697771549225, "global_step": 16822, "epoch": 158, "lr": 7.713328015383639e-05} +{"train_loss": 0.016056938096880913, "grad_norm": 0.07810155302286148, "global_step": 16823, "epoch": 158, "lr": 7.713079069382591e-05} +{"train_loss": 0.01595383882522583, "grad_norm": 0.08488724380731583, "global_step": 16824, "epoch": 158, "lr": 7.712830113848962e-05} +{"train_loss": 0.015635179355740547, "grad_norm": 0.08194772154092789, "global_step": 16825, "epoch": 158, "lr": 7.712581148783628e-05} +{"train_loss": 0.018121987581253052, "grad_norm": 0.08489657938480377, "global_step": 16826, "epoch": 158, "lr": 7.712332174187464e-05} +{"train_loss": 0.014763032086193562, "grad_norm": 0.06928875297307968, "global_step": 16827, "epoch": 158, "lr": 7.712083190061344e-05} +{"train_loss": 0.014629469253122807, "grad_norm": 0.07398916780948639, "global_step": 16828, "epoch": 158, "lr": 7.711834196406144e-05} +{"train_loss": 0.014119831845164299, "grad_norm": 0.15732038021087646, "global_step": 16829, "epoch": 158, "lr": 7.711585193222737e-05} +{"train_loss": 0.014842044562101364, "grad_norm": 0.05851209908723831, "global_step": 16830, "epoch": 158, "lr": 7.711336180512e-05} +{"train_loss": 0.016431158408522606, "grad_norm": 0.09834342449903488, "global_step": 16831, "epoch": 158, "lr": 7.711087158274806e-05} +{"train_loss": 0.017776213586330414, "grad_norm": 0.08385053277015686, "global_step": 16832, "epoch": 158, "lr": 7.71083812651203e-05} +{"train_loss": 0.017311055213212967, "grad_norm": 0.05875099077820778, "global_step": 16833, "epoch": 158, "lr": 7.710589085224548e-05} +{"train_loss": 0.015419306233525276, "grad_norm": 0.06933867931365967, "global_step": 16834, "epoch": 158, "lr": 7.710340034413235e-05} +{"train_loss": 0.015185361728072166, "grad_norm": 0.11848689615726471, "global_step": 16835, "epoch": 158, "lr": 7.710090974078966e-05} +{"train_loss": 0.01624222844839096, "grad_norm": 0.062463268637657166, "global_step": 16836, "epoch": 158, "lr": 7.709841904222616e-05} +{"train_loss": 0.016073303297162056, "grad_norm": 0.07631082832813263, "global_step": 16837, "epoch": 158, "lr": 7.70959282484506e-05} +{"train_loss": 0.01682424359023571, "grad_norm": 0.06417495012283325, "global_step": 16838, "epoch": 158, "lr": 7.709343735947173e-05} +{"train_loss": 0.017728060483932495, "grad_norm": 0.05959315225481987, "global_step": 16839, "epoch": 158, "lr": 7.70909463752983e-05} +{"train_loss": 0.015512659214437008, "grad_norm": 0.05374446138739586, "global_step": 16840, "epoch": 158, "lr": 7.708845529593907e-05} +{"train_loss": 0.017703915014863014, "grad_norm": 0.12392506003379822, "global_step": 16841, "epoch": 158, "lr": 7.708596412140279e-05} +{"train_loss": 0.014828486368060112, "grad_norm": 0.058941375464200974, "global_step": 16842, "epoch": 158, "lr": 7.70834728516982e-05} +{"train_loss": 0.014652775600552559, "grad_norm": 0.08768120408058167, "global_step": 16843, "epoch": 158, "lr": 7.70809814868341e-05} +{"train_loss": 0.015753958374261856, "grad_norm": 0.05433278530836105, "global_step": 16844, "epoch": 158, "lr": 7.707849002681917e-05} +{"train_loss": 0.01665608398616314, "grad_norm": 0.06639625132083893, "global_step": 16845, "epoch": 158, "lr": 7.707599847166222e-05} +{"train_loss": 0.017704810947179794, "grad_norm": 0.096624456346035, "global_step": 16846, "epoch": 158, "lr": 7.707350682137198e-05} +{"train_loss": 0.015865596011281013, "grad_norm": 0.08886754512786865, "global_step": 16847, "epoch": 158, "lr": 7.707101507595721e-05} +{"train_loss": 0.018282171338796616, "grad_norm": 0.10756021738052368, "global_step": 16848, "epoch": 158, "lr": 7.706852323542665e-05} +{"train_loss": 0.014264355413615704, "grad_norm": 0.05826522037386894, "global_step": 16849, "epoch": 158, "lr": 7.70660312997891e-05} +{"train_loss": 0.016331475228071213, "grad_norm": 0.049791254103183746, "global_step": 16850, "epoch": 158, "lr": 7.706353926905329e-05} +{"train_loss": 0.013852317817509174, "grad_norm": 0.05771268531680107, "global_step": 16851, "epoch": 158, "lr": 7.706104714322795e-05} +{"train_loss": 0.01793508231639862, "grad_norm": 0.05777997896075249, "global_step": 16852, "epoch": 158, "lr": 7.705855492232189e-05} +{"train_loss": 0.016496592649381678, "grad_norm": 0.10475674271583557, "global_step": 16853, "epoch": 158, "lr": 7.705606260634379e-05, "val_loss": 0.01912252977490425} +{"train_loss": 0.01438339427113533, "grad_norm": 0.10972756892442703, "global_step": 16854, "epoch": 159, "lr": 7.705357019530248e-05} +{"train_loss": 0.01760086603462696, "grad_norm": 0.0733913779258728, "global_step": 16855, "epoch": 159, "lr": 7.70510776892067e-05} +{"train_loss": 0.016705788671970367, "grad_norm": 0.052195243537425995, "global_step": 16856, "epoch": 159, "lr": 7.704858508806517e-05} +{"train_loss": 0.014725083485245705, "grad_norm": 0.07408105581998825, "global_step": 16857, "epoch": 159, "lr": 7.70460923918867e-05} +{"train_loss": 0.015790946781635284, "grad_norm": 0.056894369423389435, "global_step": 16858, "epoch": 159, "lr": 7.704359960068001e-05} +{"train_loss": 0.018162492662668228, "grad_norm": 0.055985111743211746, "global_step": 16859, "epoch": 159, "lr": 7.704110671445385e-05} +{"train_loss": 0.01414848305284977, "grad_norm": 0.10056398808956146, "global_step": 16860, "epoch": 159, "lr": 7.703861373321703e-05} +{"train_loss": 0.018373236060142517, "grad_norm": 0.11409889161586761, "global_step": 16861, "epoch": 159, "lr": 7.703612065697828e-05} +{"train_loss": 0.015470551326870918, "grad_norm": 0.054594434797763824, "global_step": 16862, "epoch": 159, "lr": 7.703362748574637e-05} +{"train_loss": 0.017277013510465622, "grad_norm": 0.09704572707414627, "global_step": 16863, "epoch": 159, "lr": 7.703113421953001e-05} +{"train_loss": 0.016320642083883286, "grad_norm": 0.04118404537439346, "global_step": 16864, "epoch": 159, "lr": 7.702864085833803e-05} +{"train_loss": 0.021816164255142212, "grad_norm": 0.08630330115556717, "global_step": 16865, "epoch": 159, "lr": 7.702614740217914e-05} +{"train_loss": 0.016459038481116295, "grad_norm": 0.07793382555246353, "global_step": 16866, "epoch": 159, "lr": 7.702365385106216e-05} +{"train_loss": 0.017109587788581848, "grad_norm": 0.06686940044164658, "global_step": 16867, "epoch": 159, "lr": 7.702116020499579e-05} +{"train_loss": 0.015390199609100819, "grad_norm": 0.06895819306373596, "global_step": 16868, "epoch": 159, "lr": 7.701866646398881e-05} +{"train_loss": 0.016713019460439682, "grad_norm": 0.09600435197353363, "global_step": 16869, "epoch": 159, "lr": 7.701617262805e-05} +{"train_loss": 0.015421690419316292, "grad_norm": 0.05675851181149483, "global_step": 16870, "epoch": 159, "lr": 7.701367869718808e-05} +{"train_loss": 0.016481002792716026, "grad_norm": 0.0733802542090416, "global_step": 16871, "epoch": 159, "lr": 7.701118467141187e-05} +{"train_loss": 0.016201265156269073, "grad_norm": 0.09448444843292236, "global_step": 16872, "epoch": 159, "lr": 7.700869055073007e-05} +{"train_loss": 0.015102449804544449, "grad_norm": 0.07858886569738388, "global_step": 16873, "epoch": 159, "lr": 7.700619633515152e-05} +{"train_loss": 0.017411502078175545, "grad_norm": 0.07591665536165237, "global_step": 16874, "epoch": 159, "lr": 7.700370202468489e-05} +{"train_loss": 0.0160101018846035, "grad_norm": 0.07956868410110474, "global_step": 16875, "epoch": 159, "lr": 7.700120761933904e-05} +{"train_loss": 0.01702427864074707, "grad_norm": 0.056641895323991776, "global_step": 16876, "epoch": 159, "lr": 7.699871311912268e-05} +{"train_loss": 0.014733141288161278, "grad_norm": 0.07026250660419464, "global_step": 16877, "epoch": 159, "lr": 7.699621852404457e-05} +{"train_loss": 0.01605253294110298, "grad_norm": 0.07959650456905365, "global_step": 16878, "epoch": 159, "lr": 7.69937238341135e-05} +{"train_loss": 0.017018545418977737, "grad_norm": 0.17676982283592224, "global_step": 16879, "epoch": 159, "lr": 7.699122904933821e-05} +{"train_loss": 0.015998581424355507, "grad_norm": 0.05012091249227524, "global_step": 16880, "epoch": 159, "lr": 7.69887341697275e-05} +{"train_loss": 0.015976663678884506, "grad_norm": 0.07894092798233032, "global_step": 16881, "epoch": 159, "lr": 7.698623919529009e-05} +{"train_loss": 0.016428060829639435, "grad_norm": 0.13926342129707336, "global_step": 16882, "epoch": 159, "lr": 7.698374412603479e-05} +{"train_loss": 0.014027245342731476, "grad_norm": 0.07117030024528503, "global_step": 16883, "epoch": 159, "lr": 7.698124896197034e-05} +{"train_loss": 0.01675155758857727, "grad_norm": 0.08791676163673401, "global_step": 16884, "epoch": 159, "lr": 7.697875370310552e-05} +{"train_loss": 0.01454196311533451, "grad_norm": 0.07010802626609802, "global_step": 16885, "epoch": 159, "lr": 7.69762583494491e-05} +{"train_loss": 0.017295587807893753, "grad_norm": 0.08781412988901138, "global_step": 16886, "epoch": 159, "lr": 7.697376290100982e-05} +{"train_loss": 0.017596114426851273, "grad_norm": 0.054074518382549286, "global_step": 16887, "epoch": 159, "lr": 7.697126735779647e-05} +{"train_loss": 0.017458295449614525, "grad_norm": 0.07753826677799225, "global_step": 16888, "epoch": 159, "lr": 7.696877171981784e-05} +{"train_loss": 0.014069435186684132, "grad_norm": 0.04869566112756729, "global_step": 16889, "epoch": 159, "lr": 7.696627598708264e-05} +{"train_loss": 0.016801506280899048, "grad_norm": 0.06088269129395485, "global_step": 16890, "epoch": 159, "lr": 7.696378015959971e-05} +{"train_loss": 0.016462942585349083, "grad_norm": 0.04904190078377724, "global_step": 16891, "epoch": 159, "lr": 7.696128423737775e-05} +{"train_loss": 0.015883252024650574, "grad_norm": 0.05374883860349655, "global_step": 16892, "epoch": 159, "lr": 7.695878822042558e-05} +{"train_loss": 0.018957093358039856, "grad_norm": 0.11680395156145096, "global_step": 16893, "epoch": 159, "lr": 7.695629210875195e-05} +{"train_loss": 0.017510322853922844, "grad_norm": 0.06370937079191208, "global_step": 16894, "epoch": 159, "lr": 7.695379590236563e-05} +{"train_loss": 0.013520179316401482, "grad_norm": 0.04332303628325462, "global_step": 16895, "epoch": 159, "lr": 7.695129960127539e-05} +{"train_loss": 0.01854414865374565, "grad_norm": 0.07231970131397247, "global_step": 16896, "epoch": 159, "lr": 7.694880320549002e-05} +{"train_loss": 0.014623085036873817, "grad_norm": 0.08355854451656342, "global_step": 16897, "epoch": 159, "lr": 7.694630671501826e-05} +{"train_loss": 0.0170159712433815, "grad_norm": 0.09486830979585648, "global_step": 16898, "epoch": 159, "lr": 7.694381012986891e-05} +{"train_loss": 0.016521356999874115, "grad_norm": 0.1437557339668274, "global_step": 16899, "epoch": 159, "lr": 7.694131345005071e-05} +{"train_loss": 0.016429295763373375, "grad_norm": 0.08662506937980652, "global_step": 16900, "epoch": 159, "lr": 7.693881667557247e-05} +{"train_loss": 0.01932726800441742, "grad_norm": 0.06397676467895508, "global_step": 16901, "epoch": 159, "lr": 7.693631980644291e-05} +{"train_loss": 0.017230702564120293, "grad_norm": 0.057150959968566895, "global_step": 16902, "epoch": 159, "lr": 7.693382284267088e-05} +{"train_loss": 0.015435758978128433, "grad_norm": 0.05148434638977051, "global_step": 16903, "epoch": 159, "lr": 7.693132578426508e-05} +{"train_loss": 0.016223154962062836, "grad_norm": 0.06374792009592056, "global_step": 16904, "epoch": 159, "lr": 7.692882863123433e-05} +{"train_loss": 0.01615910604596138, "grad_norm": 0.09597420692443848, "global_step": 16905, "epoch": 159, "lr": 7.692633138358739e-05} +{"train_loss": 0.013798274099826813, "grad_norm": 0.09875676035881042, "global_step": 16906, "epoch": 159, "lr": 7.692383404133301e-05} +{"train_loss": 0.015547842718660831, "grad_norm": 0.09533905237913132, "global_step": 16907, "epoch": 159, "lr": 7.692133660448e-05} +{"train_loss": 0.017897188663482666, "grad_norm": 0.06973385810852051, "global_step": 16908, "epoch": 159, "lr": 7.691883907303712e-05} +{"train_loss": 0.01712319441139698, "grad_norm": 0.05494208633899689, "global_step": 16909, "epoch": 159, "lr": 7.691634144701314e-05} +{"train_loss": 0.014825785532593727, "grad_norm": 0.07209347188472748, "global_step": 16910, "epoch": 159, "lr": 7.691384372641686e-05} +{"train_loss": 0.014627421274781227, "grad_norm": 0.11379925906658173, "global_step": 16911, "epoch": 159, "lr": 7.691134591125702e-05} +{"train_loss": 0.017161868512630463, "grad_norm": 0.0696694478392601, "global_step": 16912, "epoch": 159, "lr": 7.690884800154242e-05} +{"train_loss": 0.015319343656301498, "grad_norm": 0.09807322919368744, "global_step": 16913, "epoch": 159, "lr": 7.690634999728182e-05} +{"train_loss": 0.018186388537287712, "grad_norm": 0.07060938328504562, "global_step": 16914, "epoch": 159, "lr": 7.690385189848404e-05} +{"train_loss": 0.017343945801258087, "grad_norm": 0.09227798879146576, "global_step": 16915, "epoch": 159, "lr": 7.690135370515779e-05} +{"train_loss": 0.01716594025492668, "grad_norm": 0.10439830273389816, "global_step": 16916, "epoch": 159, "lr": 7.68988554173119e-05} +{"train_loss": 0.01655639335513115, "grad_norm": 0.0545896552503109, "global_step": 16917, "epoch": 159, "lr": 7.689635703495513e-05} +{"train_loss": 0.015130378305912018, "grad_norm": 0.04608491435647011, "global_step": 16918, "epoch": 159, "lr": 7.689385855809624e-05} +{"train_loss": 0.01582808792591095, "grad_norm": 0.09246613085269928, "global_step": 16919, "epoch": 159, "lr": 7.689135998674403e-05} +{"train_loss": 0.01715264283120632, "grad_norm": 0.05570041388273239, "global_step": 16920, "epoch": 159, "lr": 7.68888613209073e-05} +{"train_loss": 0.015149771235883236, "grad_norm": 0.07357918471097946, "global_step": 16921, "epoch": 159, "lr": 7.688636256059478e-05} +{"train_loss": 0.0179926585406065, "grad_norm": 0.12496023625135422, "global_step": 16922, "epoch": 159, "lr": 7.688386370581529e-05} +{"train_loss": 0.020715586841106415, "grad_norm": 0.05689515918493271, "global_step": 16923, "epoch": 159, "lr": 7.688136475657759e-05} +{"train_loss": 0.01696343719959259, "grad_norm": 0.1579652577638626, "global_step": 16924, "epoch": 159, "lr": 7.687886571289046e-05} +{"train_loss": 0.015508128330111504, "grad_norm": 0.08752183616161346, "global_step": 16925, "epoch": 159, "lr": 7.687636657476269e-05} +{"train_loss": 0.017777297645807266, "grad_norm": 0.055089954286813736, "global_step": 16926, "epoch": 159, "lr": 7.687386734220306e-05} +{"train_loss": 0.017572589218616486, "grad_norm": 0.10488224029541016, "global_step": 16927, "epoch": 159, "lr": 7.687136801522034e-05} +{"train_loss": 0.01539550069719553, "grad_norm": 0.05286628007888794, "global_step": 16928, "epoch": 159, "lr": 7.686886859382331e-05} +{"train_loss": 0.01824563555419445, "grad_norm": 0.08825907111167908, "global_step": 16929, "epoch": 159, "lr": 7.686636907802078e-05} +{"train_loss": 0.014807282015681267, "grad_norm": 0.08414987474679947, "global_step": 16930, "epoch": 159, "lr": 7.68638694678215e-05} +{"train_loss": 0.01558323297649622, "grad_norm": 0.09706920385360718, "global_step": 16931, "epoch": 159, "lr": 7.686136976323427e-05} +{"train_loss": 0.018492799252271652, "grad_norm": 0.09537816047668457, "global_step": 16932, "epoch": 159, "lr": 7.685886996426785e-05} +{"train_loss": 0.01568373292684555, "grad_norm": 0.07031610608100891, "global_step": 16933, "epoch": 159, "lr": 7.685637007093108e-05} +{"train_loss": 0.019440921023488045, "grad_norm": 0.11877679079771042, "global_step": 16934, "epoch": 159, "lr": 7.685387008323268e-05} +{"train_loss": 0.013779550790786743, "grad_norm": 0.05276176705956459, "global_step": 16935, "epoch": 159, "lr": 7.685137000118145e-05} +{"train_loss": 0.01774432510137558, "grad_norm": 0.12980715930461884, "global_step": 16936, "epoch": 159, "lr": 7.684886982478621e-05} +{"train_loss": 0.016661709174513817, "grad_norm": 0.08280890434980392, "global_step": 16937, "epoch": 159, "lr": 7.684636955405571e-05} +{"train_loss": 0.015167620964348316, "grad_norm": 0.0691356211900711, "global_step": 16938, "epoch": 159, "lr": 7.684386918899873e-05} +{"train_loss": 0.015187468379735947, "grad_norm": 0.05949738249182701, "global_step": 16939, "epoch": 159, "lr": 7.684136872962407e-05} +{"train_loss": 0.016773615032434464, "grad_norm": 0.0855766162276268, "global_step": 16940, "epoch": 159, "lr": 7.683886817594052e-05} +{"train_loss": 0.016567571088671684, "grad_norm": 0.08140648901462555, "global_step": 16941, "epoch": 159, "lr": 7.683636752795686e-05} +{"train_loss": 0.013873359188437462, "grad_norm": 0.1376623809337616, "global_step": 16942, "epoch": 159, "lr": 7.683386678568187e-05} +{"train_loss": 0.017598064616322517, "grad_norm": 0.14569278061389923, "global_step": 16943, "epoch": 159, "lr": 7.683136594912434e-05} +{"train_loss": 0.016247503459453583, "grad_norm": 0.06736411899328232, "global_step": 16944, "epoch": 159, "lr": 7.682886501829307e-05} +{"train_loss": 0.017634954303503036, "grad_norm": 0.098465695977211, "global_step": 16945, "epoch": 159, "lr": 7.682636399319683e-05} +{"train_loss": 0.01554420031607151, "grad_norm": 0.05980054661631584, "global_step": 16946, "epoch": 159, "lr": 7.682386287384442e-05} +{"train_loss": 0.016306694597005844, "grad_norm": 0.06583091616630554, "global_step": 16947, "epoch": 159, "lr": 7.68213616602446e-05} +{"train_loss": 0.015071065165102482, "grad_norm": 0.06952419131994247, "global_step": 16948, "epoch": 159, "lr": 7.68188603524062e-05} +{"train_loss": 0.019635478034615517, "grad_norm": 0.1518777310848236, "global_step": 16949, "epoch": 159, "lr": 7.681635895033799e-05} +{"train_loss": 0.01858169585466385, "grad_norm": 0.11330047994852066, "global_step": 16950, "epoch": 159, "lr": 7.681385745404875e-05} +{"train_loss": 0.019658580422401428, "grad_norm": 0.10057581216096878, "global_step": 16951, "epoch": 159, "lr": 7.681135586354727e-05} +{"train_loss": 0.01459956169128418, "grad_norm": 0.07732075452804565, "global_step": 16952, "epoch": 159, "lr": 7.680885417884234e-05} +{"train_loss": 0.017056215554475784, "grad_norm": 0.05996926873922348, "global_step": 16953, "epoch": 159, "lr": 7.680635239994276e-05} +{"train_loss": 0.014487143605947495, "grad_norm": 0.055899396538734436, "global_step": 16954, "epoch": 159, "lr": 7.680385052685732e-05} +{"train_loss": 0.01537800021469593, "grad_norm": 0.06942583620548248, "global_step": 16955, "epoch": 159, "lr": 7.680134855959483e-05} +{"train_loss": 0.01445748284459114, "grad_norm": 0.06694250553846359, "global_step": 16956, "epoch": 159, "lr": 7.679884649816402e-05} +{"train_loss": 0.016849419102072716, "grad_norm": 0.09641154855489731, "global_step": 16957, "epoch": 159, "lr": 7.679634434257372e-05} +{"train_loss": 0.016349975019693375, "grad_norm": 0.09345051646232605, "global_step": 16958, "epoch": 159, "lr": 7.679384209283273e-05} +{"train_loss": 0.016472098417580128, "grad_norm": 0.08289613574743271, "global_step": 16959, "epoch": 159, "lr": 7.679133974894983e-05, "val_loss": 0.018931468948721886} +{"train_loss": 0.013926140032708645, "grad_norm": 0.044860366731882095, "global_step": 16960, "epoch": 160, "lr": 7.678883731093383e-05} +{"train_loss": 0.017745036631822586, "grad_norm": 0.07424524426460266, "global_step": 16961, "epoch": 160, "lr": 7.678633477879348e-05} +{"train_loss": 0.013048578053712845, "grad_norm": 0.05010386183857918, "global_step": 16962, "epoch": 160, "lr": 7.678383215253762e-05} +{"train_loss": 0.018597694113850594, "grad_norm": 0.062038592994213104, "global_step": 16963, "epoch": 160, "lr": 7.6781329432175e-05} +{"train_loss": 0.017065204679965973, "grad_norm": 0.08694618195295334, "global_step": 16964, "epoch": 160, "lr": 7.677882661771445e-05} +{"train_loss": 0.01684744283556938, "grad_norm": 0.05325810983777046, "global_step": 16965, "epoch": 160, "lr": 7.677632370916474e-05} +{"train_loss": 0.019263114780187607, "grad_norm": 0.09921012073755264, "global_step": 16966, "epoch": 160, "lr": 7.677382070653467e-05} +{"train_loss": 0.015408394858241081, "grad_norm": 0.07012929022312164, "global_step": 16967, "epoch": 160, "lr": 7.677131760983306e-05} +{"train_loss": 0.01643660105764866, "grad_norm": 0.04679909348487854, "global_step": 16968, "epoch": 160, "lr": 7.676881441906866e-05} +{"train_loss": 0.018888331949710846, "grad_norm": 0.1271604299545288, "global_step": 16969, "epoch": 160, "lr": 7.67663111342503e-05} +{"train_loss": 0.015120768919587135, "grad_norm": 0.08801925182342529, "global_step": 16970, "epoch": 160, "lr": 7.676380775538676e-05} +{"train_loss": 0.018579445779323578, "grad_norm": 0.05757155641913414, "global_step": 16971, "epoch": 160, "lr": 7.676130428248684e-05} +{"train_loss": 0.0175497867166996, "grad_norm": 0.0754735991358757, "global_step": 16972, "epoch": 160, "lr": 7.675880071555934e-05} +{"train_loss": 0.017144406214356422, "grad_norm": 0.0745139941573143, "global_step": 16973, "epoch": 160, "lr": 7.675629705461303e-05} +{"train_loss": 0.016605684533715248, "grad_norm": 0.0775049701333046, "global_step": 16974, "epoch": 160, "lr": 7.675379329965675e-05} +{"train_loss": 0.014656629413366318, "grad_norm": 0.13100340962409973, "global_step": 16975, "epoch": 160, "lr": 7.675128945069927e-05} +{"train_loss": 0.018936023116111755, "grad_norm": 0.09856877475976944, "global_step": 16976, "epoch": 160, "lr": 7.674878550774939e-05} +{"train_loss": 0.01676446758210659, "grad_norm": 0.10917244851589203, "global_step": 16977, "epoch": 160, "lr": 7.674628147081591e-05} +{"train_loss": 0.0159995686262846, "grad_norm": 0.08113961666822433, "global_step": 16978, "epoch": 160, "lr": 7.674377733990763e-05} +{"train_loss": 0.015013821423053741, "grad_norm": 0.07621704787015915, "global_step": 16979, "epoch": 160, "lr": 7.674127311503335e-05} +{"train_loss": 0.018296800553798676, "grad_norm": 0.13167044520378113, "global_step": 16980, "epoch": 160, "lr": 7.673876879620185e-05} +{"train_loss": 0.015070207417011261, "grad_norm": 0.1005980595946312, "global_step": 16981, "epoch": 160, "lr": 7.673626438342197e-05} +{"train_loss": 0.017553048208355904, "grad_norm": 0.17926543951034546, "global_step": 16982, "epoch": 160, "lr": 7.673375987670247e-05} +{"train_loss": 0.016335105523467064, "grad_norm": 0.05490699037909508, "global_step": 16983, "epoch": 160, "lr": 7.673125527605216e-05} +{"train_loss": 0.016933832317590714, "grad_norm": 0.06496945768594742, "global_step": 16984, "epoch": 160, "lr": 7.672875058147986e-05} +{"train_loss": 0.016887186095118523, "grad_norm": 0.11108364164829254, "global_step": 16985, "epoch": 160, "lr": 7.672624579299434e-05} +{"train_loss": 0.017655685544013977, "grad_norm": 0.0489824041724205, "global_step": 16986, "epoch": 160, "lr": 7.672374091060441e-05} +{"train_loss": 0.0181559007614851, "grad_norm": 0.09366606920957565, "global_step": 16987, "epoch": 160, "lr": 7.672123593431889e-05} +{"train_loss": 0.018150677904486656, "grad_norm": 0.07612712681293488, "global_step": 16988, "epoch": 160, "lr": 7.671873086414655e-05} +{"train_loss": 0.01652567833662033, "grad_norm": 0.1075764074921608, "global_step": 16989, "epoch": 160, "lr": 7.671622570009623e-05} +{"train_loss": 0.016218237578868866, "grad_norm": 0.09549324959516525, "global_step": 16990, "epoch": 160, "lr": 7.671372044217671e-05} +{"train_loss": 0.014274552464485168, "grad_norm": 0.06427457183599472, "global_step": 16991, "epoch": 160, "lr": 7.671121509039678e-05} +{"train_loss": 0.015372623689472675, "grad_norm": 0.07018148899078369, "global_step": 16992, "epoch": 160, "lr": 7.670870964476526e-05} +{"train_loss": 0.017730526626110077, "grad_norm": 0.060303304344415665, "global_step": 16993, "epoch": 160, "lr": 7.670620410529097e-05} +{"train_loss": 0.01587802544236183, "grad_norm": 0.046136945486068726, "global_step": 16994, "epoch": 160, "lr": 7.670369847198267e-05} +{"train_loss": 0.017000459134578705, "grad_norm": 0.06148964539170265, "global_step": 16995, "epoch": 160, "lr": 7.67011927448492e-05} +{"train_loss": 0.015380822122097015, "grad_norm": 0.06143878027796745, "global_step": 16996, "epoch": 160, "lr": 7.669868692389934e-05} +{"train_loss": 0.016659919172525406, "grad_norm": 0.06169930472970009, "global_step": 16997, "epoch": 160, "lr": 7.66961810091419e-05} +{"train_loss": 0.01813790574669838, "grad_norm": 0.06856651604175568, "global_step": 16998, "epoch": 160, "lr": 7.66936750005857e-05} +{"train_loss": 0.01699579507112503, "grad_norm": 0.08105606585741043, "global_step": 16999, "epoch": 160, "lr": 7.669116889823955e-05} +{"train_loss": 0.016626913100481033, "grad_norm": 0.12306013703346252, "global_step": 17000, "epoch": 160, "lr": 7.668866270211222e-05} +{"train_loss": 0.017291922122240067, "grad_norm": 0.09040305018424988, "global_step": 17001, "epoch": 160, "lr": 7.668615641221253e-05} +{"train_loss": 0.016447782516479492, "grad_norm": 0.05884033069014549, "global_step": 17002, "epoch": 160, "lr": 7.668365002854931e-05} +{"train_loss": 0.01944783888757229, "grad_norm": 0.08023329824209213, "global_step": 17003, "epoch": 160, "lr": 7.668114355113134e-05} +{"train_loss": 0.015239641070365906, "grad_norm": 0.06841615587472916, "global_step": 17004, "epoch": 160, "lr": 7.667863697996742e-05} +{"train_loss": 0.0156855508685112, "grad_norm": 0.09036166220903397, "global_step": 17005, "epoch": 160, "lr": 7.667613031506639e-05} +{"train_loss": 0.01684420183300972, "grad_norm": 0.08841612190008163, "global_step": 17006, "epoch": 160, "lr": 7.667362355643705e-05} +{"train_loss": 0.016397137194871902, "grad_norm": 0.07316911965608597, "global_step": 17007, "epoch": 160, "lr": 7.667111670408818e-05} +{"train_loss": 0.015988096594810486, "grad_norm": 0.0846271812915802, "global_step": 17008, "epoch": 160, "lr": 7.66686097580286e-05} +{"train_loss": 0.017603646963834763, "grad_norm": 0.08782289922237396, "global_step": 17009, "epoch": 160, "lr": 7.666610271826715e-05} +{"train_loss": 0.018309619277715683, "grad_norm": 0.06682725995779037, "global_step": 17010, "epoch": 160, "lr": 7.666359558481257e-05} +{"train_loss": 0.014465701766312122, "grad_norm": 0.10138984024524689, "global_step": 17011, "epoch": 160, "lr": 7.666108835767375e-05} +{"train_loss": 0.015378424897789955, "grad_norm": 0.06635411828756332, "global_step": 17012, "epoch": 160, "lr": 7.665858103685944e-05} +{"train_loss": 0.011769701726734638, "grad_norm": 0.057008981704711914, "global_step": 17013, "epoch": 160, "lr": 7.665607362237846e-05} +{"train_loss": 0.016383729875087738, "grad_norm": 0.0787212923169136, "global_step": 17014, "epoch": 160, "lr": 7.665356611423966e-05} +{"train_loss": 0.01683509349822998, "grad_norm": 0.06505931913852692, "global_step": 17015, "epoch": 160, "lr": 7.66510585124518e-05} +{"train_loss": 0.014252038672566414, "grad_norm": 0.13244421780109406, "global_step": 17016, "epoch": 160, "lr": 7.664855081702371e-05} +{"train_loss": 0.014322235248982906, "grad_norm": 0.07234131544828415, "global_step": 17017, "epoch": 160, "lr": 7.66460430279642e-05} +{"train_loss": 0.01662607491016388, "grad_norm": 0.07239360362291336, "global_step": 17018, "epoch": 160, "lr": 7.664353514528208e-05} +{"train_loss": 0.016458917409181595, "grad_norm": 0.07109290361404419, "global_step": 17019, "epoch": 160, "lr": 7.664102716898619e-05} +{"train_loss": 0.018637778237462044, "grad_norm": 0.05508050695061684, "global_step": 17020, "epoch": 160, "lr": 7.66385190990853e-05} +{"train_loss": 0.01826256513595581, "grad_norm": 0.1870468109846115, "global_step": 17021, "epoch": 160, "lr": 7.663601093558823e-05} +{"train_loss": 0.016423705965280533, "grad_norm": 0.12464529275894165, "global_step": 17022, "epoch": 160, "lr": 7.663350267850383e-05} +{"train_loss": 0.015169644728302956, "grad_norm": 0.06982786953449249, "global_step": 17023, "epoch": 160, "lr": 7.663099432784084e-05} +{"train_loss": 0.018555477261543274, "grad_norm": 0.07636991888284683, "global_step": 17024, "epoch": 160, "lr": 7.662848588360814e-05} +{"train_loss": 0.015573247335851192, "grad_norm": 0.051588863134384155, "global_step": 17025, "epoch": 160, "lr": 7.662597734581454e-05} +{"train_loss": 0.012772923335433006, "grad_norm": 0.11452452093362808, "global_step": 17026, "epoch": 160, "lr": 7.662346871446881e-05} +{"train_loss": 0.015686877071857452, "grad_norm": 0.13609546422958374, "global_step": 17027, "epoch": 160, "lr": 7.662095998957978e-05} +{"train_loss": 0.013932844623923302, "grad_norm": 0.07226776331663132, "global_step": 17028, "epoch": 160, "lr": 7.661845117115632e-05} +{"train_loss": 0.017728812992572784, "grad_norm": 0.08157337456941605, "global_step": 17029, "epoch": 160, "lr": 7.661594225920714e-05} +{"train_loss": 0.013413562439382076, "grad_norm": 0.07948985695838928, "global_step": 17030, "epoch": 160, "lr": 7.661343325374116e-05} +{"train_loss": 0.017717570066452026, "grad_norm": 0.13590162992477417, "global_step": 17031, "epoch": 160, "lr": 7.661092415476712e-05} +{"train_loss": 0.021089795976877213, "grad_norm": 0.08433297276496887, "global_step": 17032, "epoch": 160, "lr": 7.660841496229388e-05} +{"train_loss": 0.013209892436861992, "grad_norm": 0.06182707101106644, "global_step": 17033, "epoch": 160, "lr": 7.660590567633024e-05} +{"train_loss": 0.01622086763381958, "grad_norm": 0.07257705926895142, "global_step": 17034, "epoch": 160, "lr": 7.660339629688502e-05} +{"train_loss": 0.0172232948243618, "grad_norm": 0.09824188798666, "global_step": 17035, "epoch": 160, "lr": 7.660088682396702e-05} +{"train_loss": 0.015681292861700058, "grad_norm": 0.04505743831396103, "global_step": 17036, "epoch": 160, "lr": 7.65983772575851e-05} +{"train_loss": 0.017612101510167122, "grad_norm": 0.10925383865833282, "global_step": 17037, "epoch": 160, "lr": 7.659586759774803e-05} +{"train_loss": 0.01535114087164402, "grad_norm": 0.09081120789051056, "global_step": 17038, "epoch": 160, "lr": 7.659335784446463e-05} +{"train_loss": 0.017397355288267136, "grad_norm": 0.06049519032239914, "global_step": 17039, "epoch": 160, "lr": 7.659084799774376e-05} +{"train_loss": 0.01834668591618538, "grad_norm": 0.04850706458091736, "global_step": 17040, "epoch": 160, "lr": 7.658833805759423e-05} +{"train_loss": 0.015165355056524277, "grad_norm": 0.1381121277809143, "global_step": 17041, "epoch": 160, "lr": 7.65858280240248e-05} +{"train_loss": 0.01547835674136877, "grad_norm": 0.08284926414489746, "global_step": 17042, "epoch": 160, "lr": 7.658331789704438e-05} +{"train_loss": 0.017057662829756737, "grad_norm": 0.07213804125785828, "global_step": 17043, "epoch": 160, "lr": 7.658080767666172e-05} +{"train_loss": 0.01420564204454422, "grad_norm": 0.07761417329311371, "global_step": 17044, "epoch": 160, "lr": 7.657829736288564e-05} +{"train_loss": 0.018591418862342834, "grad_norm": 0.05164128914475441, "global_step": 17045, "epoch": 160, "lr": 7.657578695572501e-05} +{"train_loss": 0.01260373368859291, "grad_norm": 0.10601046681404114, "global_step": 17046, "epoch": 160, "lr": 7.657327645518862e-05} +{"train_loss": 0.017338700592517853, "grad_norm": 0.05446638539433479, "global_step": 17047, "epoch": 160, "lr": 7.657076586128527e-05} +{"train_loss": 0.014305436983704567, "grad_norm": 0.05302467569708824, "global_step": 17048, "epoch": 160, "lr": 7.656825517402383e-05} +{"train_loss": 0.015231652185320854, "grad_norm": 0.068498894572258, "global_step": 17049, "epoch": 160, "lr": 7.656574439341309e-05} +{"train_loss": 0.017651613801717758, "grad_norm": 0.05736912786960602, "global_step": 17050, "epoch": 160, "lr": 7.656323351946186e-05} +{"train_loss": 0.015725435689091682, "grad_norm": 0.08037205040454865, "global_step": 17051, "epoch": 160, "lr": 7.656072255217899e-05} +{"train_loss": 0.020182248204946518, "grad_norm": 0.1492345780134201, "global_step": 17052, "epoch": 160, "lr": 7.655821149157331e-05} +{"train_loss": 0.016757141798734665, "grad_norm": 0.10475466400384903, "global_step": 17053, "epoch": 160, "lr": 7.655570033765361e-05} +{"train_loss": 0.016840994358062744, "grad_norm": 0.16319426894187927, "global_step": 17054, "epoch": 160, "lr": 7.655318909042871e-05} +{"train_loss": 0.01838500425219536, "grad_norm": 0.11144229769706726, "global_step": 17055, "epoch": 160, "lr": 7.655067774990747e-05} +{"train_loss": 0.01674514263868332, "grad_norm": 0.09714847058057785, "global_step": 17056, "epoch": 160, "lr": 7.65481663160987e-05} +{"train_loss": 0.01913559064269066, "grad_norm": 0.12060146778821945, "global_step": 17057, "epoch": 160, "lr": 7.654565478901121e-05} +{"train_loss": 0.01768488995730877, "grad_norm": 0.08623437583446503, "global_step": 17058, "epoch": 160, "lr": 7.654314316865383e-05} +{"train_loss": 0.013933905400335789, "grad_norm": 0.06304014474153519, "global_step": 17059, "epoch": 160, "lr": 7.65406314550354e-05} +{"train_loss": 0.017196305096149445, "grad_norm": 0.09930817782878876, "global_step": 17060, "epoch": 160, "lr": 7.653811964816473e-05} +{"train_loss": 0.015105510130524635, "grad_norm": 0.05612803250551224, "global_step": 17061, "epoch": 160, "lr": 7.653560774805064e-05} +{"train_loss": 0.014938393607735634, "grad_norm": 0.060210514813661575, "global_step": 17062, "epoch": 160, "lr": 7.653309575470197e-05} +{"train_loss": 0.014536410570144653, "grad_norm": 0.05579468980431557, "global_step": 17063, "epoch": 160, "lr": 7.653058366812753e-05} +{"train_loss": 0.015110103413462639, "grad_norm": 0.14871416985988617, "global_step": 17064, "epoch": 160, "lr": 7.652807148833619e-05} +{"train_loss": 0.016450867592796403, "grad_norm": 0.062556192278862, "global_step": 17065, "epoch": 160, "lr": 7.65255592153367e-05, "train_batch_mse_error": 0.014696627855300903, "val_loss": 0.018946684896945953} +{"train_loss": 0.017172638326883316, "grad_norm": 0.07107359915971756, "global_step": 17066, "epoch": 161, "lr": 7.652304684913796e-05} +{"train_loss": 0.017477570101618767, "grad_norm": 0.061027321964502335, "global_step": 17067, "epoch": 161, "lr": 7.652053438974877e-05} +{"train_loss": 0.014931892976164818, "grad_norm": 0.1704912781715393, "global_step": 17068, "epoch": 161, "lr": 7.651802183717793e-05} +{"train_loss": 0.016570791602134705, "grad_norm": 0.07770279049873352, "global_step": 17069, "epoch": 161, "lr": 7.651550919143432e-05} +{"train_loss": 0.015564549714326859, "grad_norm": 0.06372241675853729, "global_step": 17070, "epoch": 161, "lr": 7.651299645252672e-05} +{"train_loss": 0.014492291957139969, "grad_norm": 0.07802852243185043, "global_step": 17071, "epoch": 161, "lr": 7.651048362046399e-05} +{"train_loss": 0.013486357405781746, "grad_norm": 0.08139556646347046, "global_step": 17072, "epoch": 161, "lr": 7.650797069525495e-05} +{"train_loss": 0.016803879290819168, "grad_norm": 0.07249969244003296, "global_step": 17073, "epoch": 161, "lr": 7.650545767690843e-05} +{"train_loss": 0.0170593224465847, "grad_norm": 0.09606540948152542, "global_step": 17074, "epoch": 161, "lr": 7.650294456543326e-05} +{"train_loss": 0.01738143153488636, "grad_norm": 0.1120714321732521, "global_step": 17075, "epoch": 161, "lr": 7.650043136083826e-05} +{"train_loss": 0.015252118930220604, "grad_norm": 0.08609259873628616, "global_step": 17076, "epoch": 161, "lr": 7.64979180631323e-05} +{"train_loss": 0.015027233399450779, "grad_norm": 0.05316295102238655, "global_step": 17077, "epoch": 161, "lr": 7.649540467232414e-05} +{"train_loss": 0.016381140798330307, "grad_norm": 0.0889013260602951, "global_step": 17078, "epoch": 161, "lr": 7.649289118842267e-05} +{"train_loss": 0.016255123540759087, "grad_norm": 0.0643870159983635, "global_step": 17079, "epoch": 161, "lr": 7.649037761143669e-05} +{"train_loss": 0.016504615545272827, "grad_norm": 0.08700006455183029, "global_step": 17080, "epoch": 161, "lr": 7.648786394137505e-05} +{"train_loss": 0.017727605998516083, "grad_norm": 0.048194751143455505, "global_step": 17081, "epoch": 161, "lr": 7.648535017824659e-05} +{"train_loss": 0.01483856700360775, "grad_norm": 0.0524241179227829, "global_step": 17082, "epoch": 161, "lr": 7.64828363220601e-05} +{"train_loss": 0.018623296171426773, "grad_norm": 0.07516439259052277, "global_step": 17083, "epoch": 161, "lr": 7.648032237282445e-05} +{"train_loss": 0.016674943268299103, "grad_norm": 0.05588299408555031, "global_step": 17084, "epoch": 161, "lr": 7.647780833054849e-05} +{"train_loss": 0.015343474224209785, "grad_norm": 0.06651915609836578, "global_step": 17085, "epoch": 161, "lr": 7.647529419524099e-05} +{"train_loss": 0.019264021888375282, "grad_norm": 0.11464487761259079, "global_step": 17086, "epoch": 161, "lr": 7.647277996691084e-05} +{"train_loss": 0.013518938794732094, "grad_norm": 0.057191152125597, "global_step": 17087, "epoch": 161, "lr": 7.647026564556685e-05} +{"train_loss": 0.01631721295416355, "grad_norm": 0.08970154821872711, "global_step": 17088, "epoch": 161, "lr": 7.646775123121786e-05} +{"train_loss": 0.013712141662836075, "grad_norm": 0.08097171038389206, "global_step": 17089, "epoch": 161, "lr": 7.646523672387269e-05} +{"train_loss": 0.014566630125045776, "grad_norm": 0.06189638003706932, "global_step": 17090, "epoch": 161, "lr": 7.646272212354019e-05} +{"train_loss": 0.018324408680200577, "grad_norm": 0.09386265277862549, "global_step": 17091, "epoch": 161, "lr": 7.646020743022921e-05} +{"train_loss": 0.017619626596570015, "grad_norm": 0.08245860785245895, "global_step": 17092, "epoch": 161, "lr": 7.645769264394855e-05} +{"train_loss": 0.014427079819142818, "grad_norm": 0.10062900185585022, "global_step": 17093, "epoch": 161, "lr": 7.645517776470707e-05} +{"train_loss": 0.016821298748254776, "grad_norm": 0.05083681270480156, "global_step": 17094, "epoch": 161, "lr": 7.645266279251361e-05} +{"train_loss": 0.01622205600142479, "grad_norm": 0.07390854507684708, "global_step": 17095, "epoch": 161, "lr": 7.645014772737698e-05} +{"train_loss": 0.016512587666511536, "grad_norm": 0.06505375355482101, "global_step": 17096, "epoch": 161, "lr": 7.644763256930605e-05} +{"train_loss": 0.019570019096136093, "grad_norm": 0.05506083369255066, "global_step": 17097, "epoch": 161, "lr": 7.644511731830964e-05} +{"train_loss": 0.020039085298776627, "grad_norm": 0.1244930624961853, "global_step": 17098, "epoch": 161, "lr": 7.644260197439657e-05} +{"train_loss": 0.01738208904862404, "grad_norm": 0.059963326901197433, "global_step": 17099, "epoch": 161, "lr": 7.644008653757571e-05} +{"train_loss": 0.016429532319307327, "grad_norm": 0.14290745556354523, "global_step": 17100, "epoch": 161, "lr": 7.64375710078559e-05} +{"train_loss": 0.0173370111733675, "grad_norm": 0.07021724432706833, "global_step": 17101, "epoch": 161, "lr": 7.643505538524593e-05} +{"train_loss": 0.017017517238855362, "grad_norm": 0.07535601407289505, "global_step": 17102, "epoch": 161, "lr": 7.64325396697547e-05} +{"train_loss": 0.01632557064294815, "grad_norm": 0.07586872577667236, "global_step": 17103, "epoch": 161, "lr": 7.6430023861391e-05} +{"train_loss": 0.01710601896047592, "grad_norm": 0.08939430117607117, "global_step": 17104, "epoch": 161, "lr": 7.64275079601637e-05} +{"train_loss": 0.017076857388019562, "grad_norm": 0.0784582644701004, "global_step": 17105, "epoch": 161, "lr": 7.642499196608162e-05} +{"train_loss": 0.017204752191901207, "grad_norm": 0.08988060802221298, "global_step": 17106, "epoch": 161, "lr": 7.642247587915364e-05} +{"train_loss": 0.017865583300590515, "grad_norm": 0.07013488560914993, "global_step": 17107, "epoch": 161, "lr": 7.641995969938854e-05} +{"train_loss": 0.015375564806163311, "grad_norm": 0.0529572032392025, "global_step": 17108, "epoch": 161, "lr": 7.64174434267952e-05} +{"train_loss": 0.018138837069272995, "grad_norm": 0.07406363636255264, "global_step": 17109, "epoch": 161, "lr": 7.641492706138244e-05} +{"train_loss": 0.014709174633026123, "grad_norm": 0.09169144183397293, "global_step": 17110, "epoch": 161, "lr": 7.641241060315914e-05} +{"train_loss": 0.014017338864505291, "grad_norm": 0.07776273041963577, "global_step": 17111, "epoch": 161, "lr": 7.64098940521341e-05} +{"train_loss": 0.014803219586610794, "grad_norm": 0.12927404046058655, "global_step": 17112, "epoch": 161, "lr": 7.640737740831619e-05} +{"train_loss": 0.014675663784146309, "grad_norm": 0.12009667605161667, "global_step": 17113, "epoch": 161, "lr": 7.640486067171422e-05} +{"train_loss": 0.015209663659334183, "grad_norm": 0.06525324285030365, "global_step": 17114, "epoch": 161, "lr": 7.640234384233707e-05} +{"train_loss": 0.015850182622671127, "grad_norm": 0.06085161864757538, "global_step": 17115, "epoch": 161, "lr": 7.639982692019355e-05} +{"train_loss": 0.017456497997045517, "grad_norm": 0.0681353360414505, "global_step": 17116, "epoch": 161, "lr": 7.639730990529253e-05} +{"train_loss": 0.0143938809633255, "grad_norm": 0.07772050052881241, "global_step": 17117, "epoch": 161, "lr": 7.639479279764283e-05} +{"train_loss": 0.016885317862033844, "grad_norm": 0.0589483268558979, "global_step": 17118, "epoch": 161, "lr": 7.639227559725332e-05} +{"train_loss": 0.015264536254107952, "grad_norm": 0.060015592724084854, "global_step": 17119, "epoch": 161, "lr": 7.638975830413284e-05} +{"train_loss": 0.016841154545545578, "grad_norm": 0.05504573881626129, "global_step": 17120, "epoch": 161, "lr": 7.638724091829021e-05} +{"train_loss": 0.019166916608810425, "grad_norm": 0.08018984645605087, "global_step": 17121, "epoch": 161, "lr": 7.638472343973429e-05} +{"train_loss": 0.016226518899202347, "grad_norm": 0.06118568032979965, "global_step": 17122, "epoch": 161, "lr": 7.638220586847395e-05} +{"train_loss": 0.015752753242850304, "grad_norm": 0.1265784204006195, "global_step": 17123, "epoch": 161, "lr": 7.637968820451797e-05} +{"train_loss": 0.0155983567237854, "grad_norm": 0.04725303500890732, "global_step": 17124, "epoch": 161, "lr": 7.637717044787527e-05} +{"train_loss": 0.014619177207350731, "grad_norm": 0.05923209711909294, "global_step": 17125, "epoch": 161, "lr": 7.637465259855466e-05} +{"train_loss": 0.01755116879940033, "grad_norm": 0.06601086258888245, "global_step": 17126, "epoch": 161, "lr": 7.637213465656499e-05} +{"train_loss": 0.015263590030372143, "grad_norm": 0.046924442052841187, "global_step": 17127, "epoch": 161, "lr": 7.636961662191511e-05} +{"train_loss": 0.015325121581554413, "grad_norm": 0.08690431714057922, "global_step": 17128, "epoch": 161, "lr": 7.636709849461385e-05} +{"train_loss": 0.018666306510567665, "grad_norm": 0.10109741240739822, "global_step": 17129, "epoch": 161, "lr": 7.636458027467009e-05} +{"train_loss": 0.015676848590373993, "grad_norm": 0.05788679048418999, "global_step": 17130, "epoch": 161, "lr": 7.636206196209264e-05} +{"train_loss": 0.015971938148140907, "grad_norm": 0.06410983204841614, "global_step": 17131, "epoch": 161, "lr": 7.63595435568904e-05} +{"train_loss": 0.017493734136223793, "grad_norm": 0.05098143592476845, "global_step": 17132, "epoch": 161, "lr": 7.635702505907217e-05} +{"train_loss": 0.01674920879304409, "grad_norm": 0.09213748574256897, "global_step": 17133, "epoch": 161, "lr": 7.635450646864679e-05} +{"train_loss": 0.015699880197644234, "grad_norm": 0.06123480200767517, "global_step": 17134, "epoch": 161, "lr": 7.635198778562318e-05} +{"train_loss": 0.01654316484928131, "grad_norm": 0.07376441359519958, "global_step": 17135, "epoch": 161, "lr": 7.634946901001012e-05} +{"train_loss": 0.01761392503976822, "grad_norm": 0.08349652588367462, "global_step": 17136, "epoch": 161, "lr": 7.634695014181649e-05} +{"train_loss": 0.01603485271334648, "grad_norm": 0.07318582385778427, "global_step": 17137, "epoch": 161, "lr": 7.634443118105115e-05} +{"train_loss": 0.017063455656170845, "grad_norm": 0.05619324371218681, "global_step": 17138, "epoch": 161, "lr": 7.63419121277229e-05} +{"train_loss": 0.015322966501116753, "grad_norm": 0.10837779194116592, "global_step": 17139, "epoch": 161, "lr": 7.633939298184065e-05} +{"train_loss": 0.01711614988744259, "grad_norm": 0.06965567916631699, "global_step": 17140, "epoch": 161, "lr": 7.633687374341323e-05} +{"train_loss": 0.014848221093416214, "grad_norm": 0.07529766112565994, "global_step": 17141, "epoch": 161, "lr": 7.633435441244949e-05} +{"train_loss": 0.015638403594493866, "grad_norm": 0.041206687688827515, "global_step": 17142, "epoch": 161, "lr": 7.633183498895828e-05} +{"train_loss": 0.015335580334067345, "grad_norm": 0.06623527407646179, "global_step": 17143, "epoch": 161, "lr": 7.632931547294843e-05} +{"train_loss": 0.018045853823423386, "grad_norm": 0.08379923552274704, "global_step": 17144, "epoch": 161, "lr": 7.632679586442886e-05} +{"train_loss": 0.016804765909910202, "grad_norm": 0.07619378715753555, "global_step": 17145, "epoch": 161, "lr": 7.632427616340835e-05} +{"train_loss": 0.01402520202100277, "grad_norm": 0.07432097941637039, "global_step": 17146, "epoch": 161, "lr": 7.632175636989578e-05} +{"train_loss": 0.017898783087730408, "grad_norm": 0.05792972445487976, "global_step": 17147, "epoch": 161, "lr": 7.631923648390002e-05} +{"train_loss": 0.01858493685722351, "grad_norm": 0.1061505675315857, "global_step": 17148, "epoch": 161, "lr": 7.631671650542989e-05} +{"train_loss": 0.017197228968143463, "grad_norm": 0.07977823913097382, "global_step": 17149, "epoch": 161, "lr": 7.631419643449428e-05} +{"train_loss": 0.016885951161384583, "grad_norm": 0.06127943471074104, "global_step": 17150, "epoch": 161, "lr": 7.6311676271102e-05} +{"train_loss": 0.01634838618338108, "grad_norm": 0.05011805519461632, "global_step": 17151, "epoch": 161, "lr": 7.630915601526197e-05} +{"train_loss": 0.01664053648710251, "grad_norm": 0.05361788719892502, "global_step": 17152, "epoch": 161, "lr": 7.630663566698298e-05} +{"train_loss": 0.015549637377262115, "grad_norm": 0.05451244115829468, "global_step": 17153, "epoch": 161, "lr": 7.630411522627392e-05} +{"train_loss": 0.018984705209732056, "grad_norm": 0.0861780121922493, "global_step": 17154, "epoch": 161, "lr": 7.630159469314365e-05} +{"train_loss": 0.014589768834412098, "grad_norm": 0.09244825690984726, "global_step": 17155, "epoch": 161, "lr": 7.6299074067601e-05} +{"train_loss": 0.016335587948560715, "grad_norm": 0.05399777367711067, "global_step": 17156, "epoch": 161, "lr": 7.629655334965485e-05} +{"train_loss": 0.01685170643031597, "grad_norm": 0.07610608637332916, "global_step": 17157, "epoch": 161, "lr": 7.629403253931404e-05} +{"train_loss": 0.01737772300839424, "grad_norm": 0.05801551416516304, "global_step": 17158, "epoch": 161, "lr": 7.629151163658746e-05} +{"train_loss": 0.015741124749183655, "grad_norm": 0.05829638987779617, "global_step": 17159, "epoch": 161, "lr": 7.628899064148391e-05} +{"train_loss": 0.01672479137778282, "grad_norm": 0.07576444000005722, "global_step": 17160, "epoch": 161, "lr": 7.628646955401228e-05} +{"train_loss": 0.01906978338956833, "grad_norm": 0.05541375279426575, "global_step": 17161, "epoch": 161, "lr": 7.628394837418145e-05} +{"train_loss": 0.01770809292793274, "grad_norm": 0.0860438346862793, "global_step": 17162, "epoch": 161, "lr": 7.628142710200023e-05} +{"train_loss": 0.01645416021347046, "grad_norm": 0.08819621056318283, "global_step": 17163, "epoch": 161, "lr": 7.627890573747752e-05} +{"train_loss": 0.017885535955429077, "grad_norm": 0.09603273123502731, "global_step": 17164, "epoch": 161, "lr": 7.627638428062219e-05} +{"train_loss": 0.017282800748944283, "grad_norm": 0.08751237392425537, "global_step": 17165, "epoch": 161, "lr": 7.627386273144303e-05} +{"train_loss": 0.01208533626049757, "grad_norm": 0.05062710493803024, "global_step": 17166, "epoch": 161, "lr": 7.627134108994896e-05} +{"train_loss": 0.018664525821805, "grad_norm": 0.06258664280176163, "global_step": 17167, "epoch": 161, "lr": 7.626881935614882e-05} +{"train_loss": 0.01695370301604271, "grad_norm": 0.0537654384970665, "global_step": 17168, "epoch": 161, "lr": 7.626629753005146e-05} +{"train_loss": 0.017370952293276787, "grad_norm": 0.055342189967632294, "global_step": 17169, "epoch": 161, "lr": 7.626377561166578e-05} +{"train_loss": 0.015248899348080158, "grad_norm": 0.10063527524471283, "global_step": 17170, "epoch": 161, "lr": 7.62612536010006e-05} +{"train_loss": 0.016421282233424345, "grad_norm": 0.09827682375907898, "global_step": 17171, "epoch": 161, "lr": 7.62587314980648e-05, "val_loss": 0.01927938684821129} +{"train_loss": 0.015944406390190125, "grad_norm": 0.05257101356983185, "global_step": 17172, "epoch": 162, "lr": 7.625620930286723e-05} +{"train_loss": 0.016749627888202667, "grad_norm": 0.057385265827178955, "global_step": 17173, "epoch": 162, "lr": 7.625368701541675e-05} +{"train_loss": 0.018229424953460693, "grad_norm": 0.10491510480642319, "global_step": 17174, "epoch": 162, "lr": 7.625116463572226e-05} +{"train_loss": 0.018170813098549843, "grad_norm": 0.07498368620872498, "global_step": 17175, "epoch": 162, "lr": 7.624864216379257e-05} +{"train_loss": 0.015746282413601875, "grad_norm": 0.06736114621162415, "global_step": 17176, "epoch": 162, "lr": 7.624611959963658e-05} +{"train_loss": 0.015551882795989513, "grad_norm": 0.05953213945031166, "global_step": 17177, "epoch": 162, "lr": 7.624359694326313e-05} +{"train_loss": 0.013171585276722908, "grad_norm": 0.05824385955929756, "global_step": 17178, "epoch": 162, "lr": 7.62410741946811e-05} +{"train_loss": 0.016606297343969345, "grad_norm": 0.11076485365629196, "global_step": 17179, "epoch": 162, "lr": 7.623855135389933e-05} +{"train_loss": 0.016400417312979698, "grad_norm": 0.22538290917873383, "global_step": 17180, "epoch": 162, "lr": 7.623602842092672e-05} +{"train_loss": 0.015891579911112785, "grad_norm": 0.05623553693294525, "global_step": 17181, "epoch": 162, "lr": 7.623350539577212e-05} +{"train_loss": 0.01617574132978916, "grad_norm": 0.06332867592573166, "global_step": 17182, "epoch": 162, "lr": 7.623098227844438e-05} +{"train_loss": 0.01472475565969944, "grad_norm": 0.052955012768507004, "global_step": 17183, "epoch": 162, "lr": 7.622845906895237e-05} +{"train_loss": 0.01396938320249319, "grad_norm": 0.06232517212629318, "global_step": 17184, "epoch": 162, "lr": 7.622593576730498e-05} +{"train_loss": 0.019800059497356415, "grad_norm": 0.08768901228904724, "global_step": 17185, "epoch": 162, "lr": 7.622341237351105e-05} +{"train_loss": 0.014114925637841225, "grad_norm": 0.06679033488035202, "global_step": 17186, "epoch": 162, "lr": 7.622088888757945e-05} +{"train_loss": 0.01693081296980381, "grad_norm": 0.05289885029196739, "global_step": 17187, "epoch": 162, "lr": 7.621836530951905e-05} +{"train_loss": 0.013163216412067413, "grad_norm": 0.048394735902547836, "global_step": 17188, "epoch": 162, "lr": 7.621584163933871e-05} +{"train_loss": 0.015878213569521904, "grad_norm": 0.04737045615911484, "global_step": 17189, "epoch": 162, "lr": 7.621331787704731e-05} +{"train_loss": 0.017055846750736237, "grad_norm": 0.10020797699689865, "global_step": 17190, "epoch": 162, "lr": 7.621079402265372e-05} +{"train_loss": 0.020050764083862305, "grad_norm": 0.07199933379888535, "global_step": 17191, "epoch": 162, "lr": 7.620827007616678e-05} +{"train_loss": 0.017463117837905884, "grad_norm": 0.12345310300588608, "global_step": 17192, "epoch": 162, "lr": 7.620574603759538e-05} +{"train_loss": 0.015844037756323814, "grad_norm": 0.060748789459466934, "global_step": 17193, "epoch": 162, "lr": 7.62032219069484e-05} +{"train_loss": 0.016448238864541054, "grad_norm": 0.10100569576025009, "global_step": 17194, "epoch": 162, "lr": 7.620069768423468e-05} +{"train_loss": 0.016568046063184738, "grad_norm": 0.10541284829378128, "global_step": 17195, "epoch": 162, "lr": 7.61981733694631e-05} +{"train_loss": 0.01768016256392002, "grad_norm": 0.07061110436916351, "global_step": 17196, "epoch": 162, "lr": 7.619564896264255e-05} +{"train_loss": 0.018896089866757393, "grad_norm": 0.07359176129102707, "global_step": 17197, "epoch": 162, "lr": 7.619312446378186e-05} +{"train_loss": 0.015798060223460197, "grad_norm": 0.07745258510112762, "global_step": 17198, "epoch": 162, "lr": 7.619059987288994e-05} +{"train_loss": 0.018609430640935898, "grad_norm": 0.10092101246118546, "global_step": 17199, "epoch": 162, "lr": 7.618807518997563e-05} +{"train_loss": 0.017735905945301056, "grad_norm": 0.060451801866292953, "global_step": 17200, "epoch": 162, "lr": 7.618555041504783e-05} +{"train_loss": 0.018891897052526474, "grad_norm": 0.1514795869588852, "global_step": 17201, "epoch": 162, "lr": 7.618302554811537e-05} +{"train_loss": 0.016460483893752098, "grad_norm": 0.08781996369361877, "global_step": 17202, "epoch": 162, "lr": 7.618050058918716e-05} +{"train_loss": 0.015815723687410355, "grad_norm": 0.07074614614248276, "global_step": 17203, "epoch": 162, "lr": 7.617797553827204e-05} +{"train_loss": 0.01760457456111908, "grad_norm": 0.0774676576256752, "global_step": 17204, "epoch": 162, "lr": 7.617545039537892e-05} +{"train_loss": 0.017106270417571068, "grad_norm": 0.06637787073850632, "global_step": 17205, "epoch": 162, "lr": 7.617292516051665e-05} +{"train_loss": 0.01583331637084484, "grad_norm": 0.07025285810232162, "global_step": 17206, "epoch": 162, "lr": 7.617039983369408e-05} +{"train_loss": 0.01484043337404728, "grad_norm": 0.09794646501541138, "global_step": 17207, "epoch": 162, "lr": 7.616787441492013e-05} +{"train_loss": 0.015493892133235931, "grad_norm": 0.09134110808372498, "global_step": 17208, "epoch": 162, "lr": 7.616534890420364e-05} +{"train_loss": 0.01686706766486168, "grad_norm": 0.07081518322229385, "global_step": 17209, "epoch": 162, "lr": 7.616282330155349e-05} +{"train_loss": 0.016762416809797287, "grad_norm": 0.0836278423666954, "global_step": 17210, "epoch": 162, "lr": 7.616029760697856e-05} +{"train_loss": 0.016754481941461563, "grad_norm": 0.0880400612950325, "global_step": 17211, "epoch": 162, "lr": 7.615777182048774e-05} +{"train_loss": 0.014149900525808334, "grad_norm": 0.06515873968601227, "global_step": 17212, "epoch": 162, "lr": 7.615524594208985e-05} +{"train_loss": 0.017474962398409843, "grad_norm": 0.05883295089006424, "global_step": 17213, "epoch": 162, "lr": 7.615271997179381e-05} +{"train_loss": 0.017840677872300148, "grad_norm": 0.12476321309804916, "global_step": 17214, "epoch": 162, "lr": 7.61501939096085e-05} +{"train_loss": 0.017039060592651367, "grad_norm": 0.06566646695137024, "global_step": 17215, "epoch": 162, "lr": 7.614766775554278e-05} +{"train_loss": 0.015555836260318756, "grad_norm": 0.06313154846429825, "global_step": 17216, "epoch": 162, "lr": 7.614514150960552e-05} +{"train_loss": 0.020831674337387085, "grad_norm": 0.16252419352531433, "global_step": 17217, "epoch": 162, "lr": 7.61426151718056e-05} +{"train_loss": 0.017392870038747787, "grad_norm": 0.05954151228070259, "global_step": 17218, "epoch": 162, "lr": 7.61400887421519e-05} +{"train_loss": 0.017262350767850876, "grad_norm": 0.05195525661110878, "global_step": 17219, "epoch": 162, "lr": 7.613756222065331e-05} +{"train_loss": 0.0158829502761364, "grad_norm": 0.1084916889667511, "global_step": 17220, "epoch": 162, "lr": 7.613503560731868e-05} +{"train_loss": 0.016611546277999878, "grad_norm": 0.09654314070940018, "global_step": 17221, "epoch": 162, "lr": 7.61325089021569e-05} +{"train_loss": 0.016036272048950195, "grad_norm": 0.06030209735035896, "global_step": 17222, "epoch": 162, "lr": 7.612998210517686e-05} +{"train_loss": 0.017835382372140884, "grad_norm": 0.14644433557987213, "global_step": 17223, "epoch": 162, "lr": 7.61274552163874e-05} +{"train_loss": 0.014053208753466606, "grad_norm": 0.05797302722930908, "global_step": 17224, "epoch": 162, "lr": 7.612492823579745e-05} +{"train_loss": 0.01694527268409729, "grad_norm": 0.059240084141492844, "global_step": 17225, "epoch": 162, "lr": 7.612240116341585e-05} +{"train_loss": 0.01621183380484581, "grad_norm": 0.09752881526947021, "global_step": 17226, "epoch": 162, "lr": 7.61198739992515e-05} +{"train_loss": 0.0166607778519392, "grad_norm": 0.06851925700902939, "global_step": 17227, "epoch": 162, "lr": 7.611734674331328e-05} +{"train_loss": 0.017720825970172882, "grad_norm": 0.06369470059871674, "global_step": 17228, "epoch": 162, "lr": 7.611481939561005e-05} +{"train_loss": 0.01725940965116024, "grad_norm": 0.07608457654714584, "global_step": 17229, "epoch": 162, "lr": 7.611229195615071e-05} +{"train_loss": 0.014506608247756958, "grad_norm": 0.05757923051714897, "global_step": 17230, "epoch": 162, "lr": 7.61097644249441e-05} +{"train_loss": 0.01587085984647274, "grad_norm": 0.10127698630094528, "global_step": 17231, "epoch": 162, "lr": 7.610723680199919e-05} +{"train_loss": 0.013534330762922764, "grad_norm": 0.12352241575717926, "global_step": 17232, "epoch": 162, "lr": 7.610470908732476e-05} +{"train_loss": 0.016562893986701965, "grad_norm": 0.08002454787492752, "global_step": 17233, "epoch": 162, "lr": 7.610218128092975e-05} +{"train_loss": 0.0171941090375185, "grad_norm": 0.11912520229816437, "global_step": 17234, "epoch": 162, "lr": 7.609965338282303e-05} +{"train_loss": 0.016029421240091324, "grad_norm": 0.054988279938697815, "global_step": 17235, "epoch": 162, "lr": 7.609712539301346e-05} +{"train_loss": 0.016928158700466156, "grad_norm": 0.07854562997817993, "global_step": 17236, "epoch": 162, "lr": 7.609459731150997e-05} +{"train_loss": 0.01756984367966652, "grad_norm": 0.10114947706460953, "global_step": 17237, "epoch": 162, "lr": 7.609206913832139e-05} +{"train_loss": 0.016798608005046844, "grad_norm": 0.1214367151260376, "global_step": 17238, "epoch": 162, "lr": 7.608954087345663e-05} +{"train_loss": 0.01790352538228035, "grad_norm": 0.061568669974803925, "global_step": 17239, "epoch": 162, "lr": 7.608701251692457e-05} +{"train_loss": 0.015510035678744316, "grad_norm": 0.11468912661075592, "global_step": 17240, "epoch": 162, "lr": 7.60844840687341e-05} +{"train_loss": 0.017769336700439453, "grad_norm": 0.061456747353076935, "global_step": 17241, "epoch": 162, "lr": 7.608195552889409e-05} +{"train_loss": 0.015203813090920448, "grad_norm": 0.08381953835487366, "global_step": 17242, "epoch": 162, "lr": 7.607942689741343e-05} +{"train_loss": 0.015206197276711464, "grad_norm": 0.046675559133291245, "global_step": 17243, "epoch": 162, "lr": 7.607689817430101e-05} +{"train_loss": 0.017900407314300537, "grad_norm": 0.10460394620895386, "global_step": 17244, "epoch": 162, "lr": 7.607436935956571e-05} +{"train_loss": 0.01791098713874817, "grad_norm": 0.07063847035169601, "global_step": 17245, "epoch": 162, "lr": 7.60718404532164e-05} +{"train_loss": 0.016678906977176666, "grad_norm": 0.09464599937200546, "global_step": 17246, "epoch": 162, "lr": 7.606931145526202e-05} +{"train_loss": 0.01947515830397606, "grad_norm": 0.09140849113464355, "global_step": 17247, "epoch": 162, "lr": 7.606678236571137e-05} +{"train_loss": 0.012339435517787933, "grad_norm": 0.06772521138191223, "global_step": 17248, "epoch": 162, "lr": 7.606425318457341e-05} +{"train_loss": 0.015029516071081161, "grad_norm": 0.087513267993927, "global_step": 17249, "epoch": 162, "lr": 7.6061723911857e-05} +{"train_loss": 0.016957901418209076, "grad_norm": 0.08028281480073929, "global_step": 17250, "epoch": 162, "lr": 7.605919454757102e-05} +{"train_loss": 0.015988284721970558, "grad_norm": 0.05944051593542099, "global_step": 17251, "epoch": 162, "lr": 7.605666509172436e-05} +{"train_loss": 0.017378468066453934, "grad_norm": 0.10553738474845886, "global_step": 17252, "epoch": 162, "lr": 7.605413554432593e-05} +{"train_loss": 0.014956608414649963, "grad_norm": 0.06279396265745163, "global_step": 17253, "epoch": 162, "lr": 7.605160590538458e-05} +{"train_loss": 0.01599186286330223, "grad_norm": 0.06607764959335327, "global_step": 17254, "epoch": 162, "lr": 7.604907617490921e-05} +{"train_loss": 0.013967008329927921, "grad_norm": 0.05242322012782097, "global_step": 17255, "epoch": 162, "lr": 7.604654635290873e-05} +{"train_loss": 0.01402832381427288, "grad_norm": 0.08544974774122238, "global_step": 17256, "epoch": 162, "lr": 7.604401643939201e-05} +{"train_loss": 0.01718878000974655, "grad_norm": 0.07179352641105652, "global_step": 17257, "epoch": 162, "lr": 7.604148643436792e-05} +{"train_loss": 0.01748616248369217, "grad_norm": 0.05709857493638992, "global_step": 17258, "epoch": 162, "lr": 7.603895633784541e-05} +{"train_loss": 0.016744563356041908, "grad_norm": 0.08085039258003235, "global_step": 17259, "epoch": 162, "lr": 7.60364261498333e-05} +{"train_loss": 0.015018254518508911, "grad_norm": 0.10325612872838974, "global_step": 17260, "epoch": 162, "lr": 7.603389587034053e-05} +{"train_loss": 0.018711183220148087, "grad_norm": 0.12439648807048798, "global_step": 17261, "epoch": 162, "lr": 7.603136549937597e-05} +{"train_loss": 0.01479608565568924, "grad_norm": 0.09652842581272125, "global_step": 17262, "epoch": 162, "lr": 7.602883503694851e-05} +{"train_loss": 0.016374260187149048, "grad_norm": 0.11503572016954422, "global_step": 17263, "epoch": 162, "lr": 7.602630448306705e-05} +{"train_loss": 0.01539512351155281, "grad_norm": 0.06987351924180984, "global_step": 17264, "epoch": 162, "lr": 7.602377383774046e-05} +{"train_loss": 0.016154389828443527, "grad_norm": 0.04648295417428017, "global_step": 17265, "epoch": 162, "lr": 7.602124310097764e-05} +{"train_loss": 0.01667991653084755, "grad_norm": 0.05662120506167412, "global_step": 17266, "epoch": 162, "lr": 7.601871227278752e-05} +{"train_loss": 0.015816614031791687, "grad_norm": 0.06944148987531662, "global_step": 17267, "epoch": 162, "lr": 7.601618135317893e-05} +{"train_loss": 0.015640558674931526, "grad_norm": 0.08702206611633301, "global_step": 17268, "epoch": 162, "lr": 7.601365034216081e-05} +{"train_loss": 0.01601947657763958, "grad_norm": 0.10055369883775711, "global_step": 17269, "epoch": 162, "lr": 7.601111923974202e-05} +{"train_loss": 0.01596793718636036, "grad_norm": 0.12319665402173996, "global_step": 17270, "epoch": 162, "lr": 7.600858804593148e-05} +{"train_loss": 0.016625171527266502, "grad_norm": 0.06611271947622299, "global_step": 17271, "epoch": 162, "lr": 7.600605676073807e-05} +{"train_loss": 0.017318610101938248, "grad_norm": 0.07687732577323914, "global_step": 17272, "epoch": 162, "lr": 7.600352538417071e-05} +{"train_loss": 0.014615103602409363, "grad_norm": 0.12645363807678223, "global_step": 17273, "epoch": 162, "lr": 7.600099391623823e-05} +{"train_loss": 0.018635405227541924, "grad_norm": 0.07184655219316483, "global_step": 17274, "epoch": 162, "lr": 7.599846235694959e-05} +{"train_loss": 0.016651514917612076, "grad_norm": 0.09232358634471893, "global_step": 17275, "epoch": 162, "lr": 7.599593070631366e-05} +{"train_loss": 0.016368024051189423, "grad_norm": 0.05388503149151802, "global_step": 17276, "epoch": 162, "lr": 7.59933989643393e-05} +{"train_loss": 0.016415229998528957, "grad_norm": 0.08619125187397003, "global_step": 17277, "epoch": 162, "lr": 7.599086713103547e-05, "val_loss": 0.018649734556674957} +{"train_loss": 0.015101145952939987, "grad_norm": 0.058063142001628876, "global_step": 17278, "epoch": 163, "lr": 7.598833520641104e-05} +{"train_loss": 0.01657581701874733, "grad_norm": 0.07116007804870605, "global_step": 17279, "epoch": 163, "lr": 7.59858031904749e-05} +{"train_loss": 0.015113215893507004, "grad_norm": 0.05558868125081062, "global_step": 17280, "epoch": 163, "lr": 7.598327108323596e-05} +{"train_loss": 0.01549984235316515, "grad_norm": 0.0661105215549469, "global_step": 17281, "epoch": 163, "lr": 7.598073888470307e-05} +{"train_loss": 0.018905751407146454, "grad_norm": 0.08177034556865692, "global_step": 17282, "epoch": 163, "lr": 7.59782065948852e-05} +{"train_loss": 0.015486367978155613, "grad_norm": 0.0709204450249672, "global_step": 17283, "epoch": 163, "lr": 7.597567421379116e-05} +{"train_loss": 0.01726675033569336, "grad_norm": 0.0621064230799675, "global_step": 17284, "epoch": 163, "lr": 7.597314174142995e-05} +{"train_loss": 0.0162477046251297, "grad_norm": 0.08307506144046783, "global_step": 17285, "epoch": 163, "lr": 7.597060917781039e-05} +{"train_loss": 0.014566441997885704, "grad_norm": 0.0782352089881897, "global_step": 17286, "epoch": 163, "lr": 7.59680765229414e-05} +{"train_loss": 0.014059267938137054, "grad_norm": 0.05175749957561493, "global_step": 17287, "epoch": 163, "lr": 7.596554377683191e-05} +{"train_loss": 0.015246963128447533, "grad_norm": 0.06129327043890953, "global_step": 17288, "epoch": 163, "lr": 7.596301093949076e-05} +{"train_loss": 0.018031220883131027, "grad_norm": 0.07923734188079834, "global_step": 17289, "epoch": 163, "lr": 7.59604780109269e-05} +{"train_loss": 0.013440556824207306, "grad_norm": 0.05328401178121567, "global_step": 17290, "epoch": 163, "lr": 7.595794499114919e-05} +{"train_loss": 0.018852384760975838, "grad_norm": 0.13616962730884552, "global_step": 17291, "epoch": 163, "lr": 7.595541188016657e-05} +{"train_loss": 0.015538070350885391, "grad_norm": 0.0623052679002285, "global_step": 17292, "epoch": 163, "lr": 7.595287867798791e-05} +{"train_loss": 0.01704377308487892, "grad_norm": 0.06757239252328873, "global_step": 17293, "epoch": 163, "lr": 7.595034538462212e-05} +{"train_loss": 0.01534847728908062, "grad_norm": 0.06685849279165268, "global_step": 17294, "epoch": 163, "lr": 7.59478120000781e-05} +{"train_loss": 0.01491358783096075, "grad_norm": 0.09148439019918442, "global_step": 17295, "epoch": 163, "lr": 7.594527852436478e-05} +{"train_loss": 0.018032453954219818, "grad_norm": 0.06976255774497986, "global_step": 17296, "epoch": 163, "lr": 7.594274495749101e-05} +{"train_loss": 0.013158884830772877, "grad_norm": 0.04965578019618988, "global_step": 17297, "epoch": 163, "lr": 7.594021129946571e-05} +{"train_loss": 0.017387013882398605, "grad_norm": 0.05587119981646538, "global_step": 17298, "epoch": 163, "lr": 7.59376775502978e-05} +{"train_loss": 0.01936314068734646, "grad_norm": 0.09577468037605286, "global_step": 17299, "epoch": 163, "lr": 7.593514370999617e-05} +{"train_loss": 0.015807028859853745, "grad_norm": 0.12137621641159058, "global_step": 17300, "epoch": 163, "lr": 7.593260977856971e-05} +{"train_loss": 0.01846972480416298, "grad_norm": 0.09589225053787231, "global_step": 17301, "epoch": 163, "lr": 7.593007575602735e-05} +{"train_loss": 0.016004163771867752, "grad_norm": 0.060093045234680176, "global_step": 17302, "epoch": 163, "lr": 7.592754164237798e-05} +{"train_loss": 0.016556503251194954, "grad_norm": 0.06988713145256042, "global_step": 17303, "epoch": 163, "lr": 7.592500743763051e-05} +{"train_loss": 0.017569538205862045, "grad_norm": 0.07892832159996033, "global_step": 17304, "epoch": 163, "lr": 7.592247314179383e-05} +{"train_loss": 0.016329340636730194, "grad_norm": 0.051031120121479034, "global_step": 17305, "epoch": 163, "lr": 7.591993875487687e-05} +{"train_loss": 0.01541954092681408, "grad_norm": 0.09962908923625946, "global_step": 17306, "epoch": 163, "lr": 7.59174042768885e-05} +{"train_loss": 0.019572339951992035, "grad_norm": 0.09615641832351685, "global_step": 17307, "epoch": 163, "lr": 7.591486970783763e-05} +{"train_loss": 0.016184698790311813, "grad_norm": 0.06219852343201637, "global_step": 17308, "epoch": 163, "lr": 7.591233504773321e-05} +{"train_loss": 0.018710847944021225, "grad_norm": 0.06001013517379761, "global_step": 17309, "epoch": 163, "lr": 7.59098002965841e-05} +{"train_loss": 0.01733338087797165, "grad_norm": 0.07766344398260117, "global_step": 17310, "epoch": 163, "lr": 7.59072654543992e-05} +{"train_loss": 0.01900925487279892, "grad_norm": 0.0799349993467331, "global_step": 17311, "epoch": 163, "lr": 7.590473052118747e-05} +{"train_loss": 0.015898723155260086, "grad_norm": 0.06642807275056839, "global_step": 17312, "epoch": 163, "lr": 7.590219549695775e-05} +{"train_loss": 0.020310167223215103, "grad_norm": 0.06434231251478195, "global_step": 17313, "epoch": 163, "lr": 7.5899660381719e-05} +{"train_loss": 0.016871966421604156, "grad_norm": 0.08535657078027725, "global_step": 17314, "epoch": 163, "lr": 7.589712517548011e-05} +{"train_loss": 0.017098870128393173, "grad_norm": 0.1916702389717102, "global_step": 17315, "epoch": 163, "lr": 7.589458987824997e-05} +{"train_loss": 0.015811236575245857, "grad_norm": 0.06518005579710007, "global_step": 17316, "epoch": 163, "lr": 7.589205449003751e-05} +{"train_loss": 0.019797753542661667, "grad_norm": 0.07896491885185242, "global_step": 17317, "epoch": 163, "lr": 7.588951901085163e-05} +{"train_loss": 0.014638523571193218, "grad_norm": 0.10372689366340637, "global_step": 17318, "epoch": 163, "lr": 7.588698344070124e-05} +{"train_loss": 0.01560872234404087, "grad_norm": 0.0728064626455307, "global_step": 17319, "epoch": 163, "lr": 7.588444777959524e-05} +{"train_loss": 0.013544495217502117, "grad_norm": 0.07742198556661606, "global_step": 17320, "epoch": 163, "lr": 7.588191202754256e-05} +{"train_loss": 0.017949407920241356, "grad_norm": 0.11660978198051453, "global_step": 17321, "epoch": 163, "lr": 7.587937618455209e-05} +{"train_loss": 0.01432738546282053, "grad_norm": 0.07776381820440292, "global_step": 17322, "epoch": 163, "lr": 7.587684025063275e-05} +{"train_loss": 0.016039693728089333, "grad_norm": 0.0812002494931221, "global_step": 17323, "epoch": 163, "lr": 7.587430422579344e-05} +{"train_loss": 0.0175136998295784, "grad_norm": 0.05323266610503197, "global_step": 17324, "epoch": 163, "lr": 7.587176811004308e-05} +{"train_loss": 0.01838727667927742, "grad_norm": 0.14240404963493347, "global_step": 17325, "epoch": 163, "lr": 7.586923190339059e-05} +{"train_loss": 0.016813762485980988, "grad_norm": 0.08971378952264786, "global_step": 17326, "epoch": 163, "lr": 7.586669560584487e-05} +{"train_loss": 0.01622619852423668, "grad_norm": 0.07218893617391586, "global_step": 17327, "epoch": 163, "lr": 7.586415921741481e-05} +{"train_loss": 0.01955082081258297, "grad_norm": 0.2129344940185547, "global_step": 17328, "epoch": 163, "lr": 7.586162273810935e-05} +{"train_loss": 0.014989856630563736, "grad_norm": 0.11336860060691833, "global_step": 17329, "epoch": 163, "lr": 7.58590861679374e-05} +{"train_loss": 0.017319265753030777, "grad_norm": 0.09199995547533035, "global_step": 17330, "epoch": 163, "lr": 7.585654950690786e-05} +{"train_loss": 0.018770486116409302, "grad_norm": 0.09333483874797821, "global_step": 17331, "epoch": 163, "lr": 7.585401275502967e-05} +{"train_loss": 0.014546016231179237, "grad_norm": 0.1168627217411995, "global_step": 17332, "epoch": 163, "lr": 7.585147591231169e-05} +{"train_loss": 0.01431154366582632, "grad_norm": 0.08327782154083252, "global_step": 17333, "epoch": 163, "lr": 7.584893897876289e-05} +{"train_loss": 0.016908787190914154, "grad_norm": 0.18176402151584625, "global_step": 17334, "epoch": 163, "lr": 7.584640195439215e-05} +{"train_loss": 0.016231274232268333, "grad_norm": 0.09137267619371414, "global_step": 17335, "epoch": 163, "lr": 7.58438648392084e-05} +{"train_loss": 0.016298159956932068, "grad_norm": 0.1590486466884613, "global_step": 17336, "epoch": 163, "lr": 7.584132763322054e-05} +{"train_loss": 0.01690133661031723, "grad_norm": 0.059942908585071564, "global_step": 17337, "epoch": 163, "lr": 7.58387903364375e-05} +{"train_loss": 0.015733301639556885, "grad_norm": 0.08676763623952866, "global_step": 17338, "epoch": 163, "lr": 7.583625294886819e-05} +{"train_loss": 0.016270089894533157, "grad_norm": 0.08142821490764618, "global_step": 17339, "epoch": 163, "lr": 7.583371547052151e-05} +{"train_loss": 0.014959825202822685, "grad_norm": 0.0736822709441185, "global_step": 17340, "epoch": 163, "lr": 7.583117790140642e-05} +{"train_loss": 0.01554432138800621, "grad_norm": 0.11733902990818024, "global_step": 17341, "epoch": 163, "lr": 7.582864024153177e-05} +{"train_loss": 0.016652626916766167, "grad_norm": 0.13119001686573029, "global_step": 17342, "epoch": 163, "lr": 7.582610249090651e-05} +{"train_loss": 0.01725139282643795, "grad_norm": 0.08912377804517746, "global_step": 17343, "epoch": 163, "lr": 7.582356464953959e-05} +{"train_loss": 0.017047181725502014, "grad_norm": 0.11138271540403366, "global_step": 17344, "epoch": 163, "lr": 7.582102671743987e-05} +{"train_loss": 0.016825295984745026, "grad_norm": 0.046202320605516434, "global_step": 17345, "epoch": 163, "lr": 7.58184886946163e-05} +{"train_loss": 0.01825963892042637, "grad_norm": 0.07460615783929825, "global_step": 17346, "epoch": 163, "lr": 7.581595058107778e-05} +{"train_loss": 0.014886720106005669, "grad_norm": 0.06978054344654083, "global_step": 17347, "epoch": 163, "lr": 7.581341237683324e-05} +{"train_loss": 0.015613283030688763, "grad_norm": 0.06183362752199173, "global_step": 17348, "epoch": 163, "lr": 7.58108740818916e-05} +{"train_loss": 0.017230896279215813, "grad_norm": 0.1331496238708496, "global_step": 17349, "epoch": 163, "lr": 7.580833569626177e-05} +{"train_loss": 0.0193830206990242, "grad_norm": 0.05868206545710564, "global_step": 17350, "epoch": 163, "lr": 7.580579721995267e-05} +{"train_loss": 0.017348382622003555, "grad_norm": 0.14118796586990356, "global_step": 17351, "epoch": 163, "lr": 7.580325865297322e-05} +{"train_loss": 0.018835440278053284, "grad_norm": 0.12010659277439117, "global_step": 17352, "epoch": 163, "lr": 7.580071999533234e-05} +{"train_loss": 0.01740400865674019, "grad_norm": 0.05503435432910919, "global_step": 17353, "epoch": 163, "lr": 7.579818124703897e-05} +{"train_loss": 0.014780597761273384, "grad_norm": 0.0962085872888565, "global_step": 17354, "epoch": 163, "lr": 7.579564240810198e-05} +{"train_loss": 0.016602635383605957, "grad_norm": 0.04722711816430092, "global_step": 17355, "epoch": 163, "lr": 7.579310347853035e-05} +{"train_loss": 0.018459195271134377, "grad_norm": 0.06450387090444565, "global_step": 17356, "epoch": 163, "lr": 7.579056445833294e-05} +{"train_loss": 0.014409391209483147, "grad_norm": 0.12369991838932037, "global_step": 17357, "epoch": 163, "lr": 7.578802534751872e-05} +{"train_loss": 0.01630105823278427, "grad_norm": 0.05639450624585152, "global_step": 17358, "epoch": 163, "lr": 7.57854861460966e-05} +{"train_loss": 0.01674932800233364, "grad_norm": 0.06432532519102097, "global_step": 17359, "epoch": 163, "lr": 7.578294685407548e-05} +{"train_loss": 0.014880175702273846, "grad_norm": 0.06647700071334839, "global_step": 17360, "epoch": 163, "lr": 7.578040747146431e-05} +{"train_loss": 0.015432490967214108, "grad_norm": 0.12731751799583435, "global_step": 17361, "epoch": 163, "lr": 7.5777867998272e-05} +{"train_loss": 0.016737764701247215, "grad_norm": 0.07891174405813217, "global_step": 17362, "epoch": 163, "lr": 7.577532843450747e-05} +{"train_loss": 0.01944817416369915, "grad_norm": 0.06541158258914948, "global_step": 17363, "epoch": 163, "lr": 7.577278878017963e-05} +{"train_loss": 0.013765616342425346, "grad_norm": 0.047067612409591675, "global_step": 17364, "epoch": 163, "lr": 7.577024903529743e-05} +{"train_loss": 0.017440874129533768, "grad_norm": 0.059297237545251846, "global_step": 17365, "epoch": 163, "lr": 7.576770919986979e-05} +{"train_loss": 0.013202900998294353, "grad_norm": 0.07084129750728607, "global_step": 17366, "epoch": 163, "lr": 7.576516927390562e-05} +{"train_loss": 0.013745061121881008, "grad_norm": 0.08310546725988388, "global_step": 17367, "epoch": 163, "lr": 7.576262925741386e-05} +{"train_loss": 0.01486937701702118, "grad_norm": 0.060674622654914856, "global_step": 17368, "epoch": 163, "lr": 7.57600891504034e-05} +{"train_loss": 0.018010329455137253, "grad_norm": 0.06405849754810333, "global_step": 17369, "epoch": 163, "lr": 7.57575489528832e-05} +{"train_loss": 0.014400511980056763, "grad_norm": 0.09423726052045822, "global_step": 17370, "epoch": 163, "lr": 7.575500866486219e-05} +{"train_loss": 0.014860663563013077, "grad_norm": 0.04627582058310509, "global_step": 17371, "epoch": 163, "lr": 7.575246828634926e-05} +{"train_loss": 0.014111275784671307, "grad_norm": 0.07039492577314377, "global_step": 17372, "epoch": 163, "lr": 7.574992781735338e-05} +{"train_loss": 0.014531239867210388, "grad_norm": 0.05121895670890808, "global_step": 17373, "epoch": 163, "lr": 7.574738725788343e-05} +{"train_loss": 0.018774228170514107, "grad_norm": 0.11905943602323532, "global_step": 17374, "epoch": 163, "lr": 7.574484660794836e-05} +{"train_loss": 0.017249923199415207, "grad_norm": 0.08477447181940079, "global_step": 17375, "epoch": 163, "lr": 7.57423058675571e-05} +{"train_loss": 0.01649683713912964, "grad_norm": 0.0805089920759201, "global_step": 17376, "epoch": 163, "lr": 7.573976503671857e-05} +{"train_loss": 0.01707107573747635, "grad_norm": 0.06329793483018875, "global_step": 17377, "epoch": 163, "lr": 7.573722411544171e-05} +{"train_loss": 0.01616101898252964, "grad_norm": 0.1079985499382019, "global_step": 17378, "epoch": 163, "lr": 7.573468310373543e-05} +{"train_loss": 0.0171494260430336, "grad_norm": 0.12704184651374817, "global_step": 17379, "epoch": 163, "lr": 7.573214200160867e-05} +{"train_loss": 0.014612453058362007, "grad_norm": 0.08663767576217651, "global_step": 17380, "epoch": 163, "lr": 7.572960080907035e-05} +{"train_loss": 0.014864235185086727, "grad_norm": 0.07122282683849335, "global_step": 17381, "epoch": 163, "lr": 7.57270595261294e-05} +{"train_loss": 0.015846025198698044, "grad_norm": 0.11927660554647446, "global_step": 17382, "epoch": 163, "lr": 7.572451815279475e-05} +{"train_loss": 0.016423666032628632, "grad_norm": 0.0901157557964325, "global_step": 17383, "epoch": 163, "lr": 7.572197668907532e-05, "val_loss": 0.019453687593340874} +{"train_loss": 0.01598626933991909, "grad_norm": 0.07498014718294144, "global_step": 17384, "epoch": 164, "lr": 7.571943513498007e-05} +{"train_loss": 0.017598524689674377, "grad_norm": 0.11964844167232513, "global_step": 17385, "epoch": 164, "lr": 7.571689349051789e-05} +{"train_loss": 0.01465400867164135, "grad_norm": 0.07156448811292648, "global_step": 17386, "epoch": 164, "lr": 7.571435175569776e-05} +{"train_loss": 0.017290674149990082, "grad_norm": 0.11098194867372513, "global_step": 17387, "epoch": 164, "lr": 7.571180993052855e-05} +{"train_loss": 0.01737075299024582, "grad_norm": 0.09018095582723618, "global_step": 17388, "epoch": 164, "lr": 7.570926801501925e-05} +{"train_loss": 0.017668195068836212, "grad_norm": 0.09610375761985779, "global_step": 17389, "epoch": 164, "lr": 7.570672600917874e-05} +{"train_loss": 0.016733843833208084, "grad_norm": 0.0837068259716034, "global_step": 17390, "epoch": 164, "lr": 7.5704183913016e-05} +{"train_loss": 0.014964655041694641, "grad_norm": 0.05218822509050369, "global_step": 17391, "epoch": 164, "lr": 7.570164172653991e-05} +{"train_loss": 0.014275208115577698, "grad_norm": 0.06905365735292435, "global_step": 17392, "epoch": 164, "lr": 7.569909944975943e-05} +{"train_loss": 0.0169031023979187, "grad_norm": 0.08488357067108154, "global_step": 17393, "epoch": 164, "lr": 7.56965570826835e-05} +{"train_loss": 0.017646506428718567, "grad_norm": 0.08893544971942902, "global_step": 17394, "epoch": 164, "lr": 7.569401462532104e-05} +{"train_loss": 0.0177963525056839, "grad_norm": 0.057243525981903076, "global_step": 17395, "epoch": 164, "lr": 7.569147207768099e-05} +{"train_loss": 0.015583242289721966, "grad_norm": 0.1315053254365921, "global_step": 17396, "epoch": 164, "lr": 7.568892943977228e-05} +{"train_loss": 0.01664612628519535, "grad_norm": 0.13650290668010712, "global_step": 17397, "epoch": 164, "lr": 7.568638671160384e-05} +{"train_loss": 0.01529885083436966, "grad_norm": 0.052279211580753326, "global_step": 17398, "epoch": 164, "lr": 7.568384389318461e-05} +{"train_loss": 0.016496825963258743, "grad_norm": 0.0535234659910202, "global_step": 17399, "epoch": 164, "lr": 7.568130098452351e-05} +{"train_loss": 0.01572537049651146, "grad_norm": 0.07505931705236435, "global_step": 17400, "epoch": 164, "lr": 7.56787579856295e-05} +{"train_loss": 0.018251225352287292, "grad_norm": 0.10970044881105423, "global_step": 17401, "epoch": 164, "lr": 7.567621489651151e-05} +{"train_loss": 0.016316860914230347, "grad_norm": 0.06244690716266632, "global_step": 17402, "epoch": 164, "lr": 7.567367171717845e-05} +{"train_loss": 0.01805046759545803, "grad_norm": 0.07459188997745514, "global_step": 17403, "epoch": 164, "lr": 7.567112844763929e-05} +{"train_loss": 0.016689276322722435, "grad_norm": 0.05357915908098221, "global_step": 17404, "epoch": 164, "lr": 7.566858508790293e-05} +{"train_loss": 0.016124356538057327, "grad_norm": 0.08871380984783173, "global_step": 17405, "epoch": 164, "lr": 7.566604163797835e-05} +{"train_loss": 0.017128322273492813, "grad_norm": 0.14022773504257202, "global_step": 17406, "epoch": 164, "lr": 7.566349809787443e-05} +{"train_loss": 0.017089717090129852, "grad_norm": 0.04905856400728226, "global_step": 17407, "epoch": 164, "lr": 7.566095446760015e-05} +{"train_loss": 0.016889385879039764, "grad_norm": 0.051893699914216995, "global_step": 17408, "epoch": 164, "lr": 7.565841074716446e-05} +{"train_loss": 0.01604551635682583, "grad_norm": 0.07613218575716019, "global_step": 17409, "epoch": 164, "lr": 7.565586693657623e-05} +{"train_loss": 0.01824459619820118, "grad_norm": 0.0641292929649353, "global_step": 17410, "epoch": 164, "lr": 7.565332303584447e-05} +{"train_loss": 0.01797306537628174, "grad_norm": 0.06406553834676743, "global_step": 17411, "epoch": 164, "lr": 7.565077904497809e-05} +{"train_loss": 0.01581016555428505, "grad_norm": 0.11552782356739044, "global_step": 17412, "epoch": 164, "lr": 7.5648234963986e-05} +{"train_loss": 0.01812821999192238, "grad_norm": 0.0603516586124897, "global_step": 17413, "epoch": 164, "lr": 7.56456907928772e-05} +{"train_loss": 0.019091635942459106, "grad_norm": 0.16356182098388672, "global_step": 17414, "epoch": 164, "lr": 7.564314653166058e-05} +{"train_loss": 0.01564370095729828, "grad_norm": 0.15077950060367584, "global_step": 17415, "epoch": 164, "lr": 7.56406021803451e-05} +{"train_loss": 0.016389120370149612, "grad_norm": 0.05996242165565491, "global_step": 17416, "epoch": 164, "lr": 7.563805773893971e-05} +{"train_loss": 0.01595519483089447, "grad_norm": 0.06454804539680481, "global_step": 17417, "epoch": 164, "lr": 7.56355132074533e-05} +{"train_loss": 0.018511898815631866, "grad_norm": 0.07980853319168091, "global_step": 17418, "epoch": 164, "lr": 7.563296858589488e-05} +{"train_loss": 0.01715298742055893, "grad_norm": 0.06383078545331955, "global_step": 17419, "epoch": 164, "lr": 7.563042387427332e-05} +{"train_loss": 0.01733241230249405, "grad_norm": 0.15155929327011108, "global_step": 17420, "epoch": 164, "lr": 7.562787907259763e-05} +{"train_loss": 0.01757616549730301, "grad_norm": 0.05432967469096184, "global_step": 17421, "epoch": 164, "lr": 7.562533418087669e-05} +{"train_loss": 0.01733548566699028, "grad_norm": 0.12279970943927765, "global_step": 17422, "epoch": 164, "lr": 7.56227891991195e-05} +{"train_loss": 0.015878554433584213, "grad_norm": 0.05607225373387337, "global_step": 17423, "epoch": 164, "lr": 7.562024412733495e-05} +{"train_loss": 0.017102550715208054, "grad_norm": 0.08247368037700653, "global_step": 17424, "epoch": 164, "lr": 7.5617698965532e-05} +{"train_loss": 0.014709790237247944, "grad_norm": 0.044164810329675674, "global_step": 17425, "epoch": 164, "lr": 7.561515371371961e-05} +{"train_loss": 0.016065549105405807, "grad_norm": 0.17166246473789215, "global_step": 17426, "epoch": 164, "lr": 7.56126083719067e-05} +{"train_loss": 0.01700480654835701, "grad_norm": 0.060103464871644974, "global_step": 17427, "epoch": 164, "lr": 7.561006294010223e-05} +{"train_loss": 0.017676187679171562, "grad_norm": 0.0803321823477745, "global_step": 17428, "epoch": 164, "lr": 7.560751741831513e-05} +{"train_loss": 0.015254481695592403, "grad_norm": 0.0979037806391716, "global_step": 17429, "epoch": 164, "lr": 7.560497180655437e-05} +{"train_loss": 0.016915902495384216, "grad_norm": 0.08251183480024338, "global_step": 17430, "epoch": 164, "lr": 7.560242610482886e-05} +{"train_loss": 0.013467039912939072, "grad_norm": 0.058090902864933014, "global_step": 17431, "epoch": 164, "lr": 7.559988031314755e-05} +{"train_loss": 0.016054224222898483, "grad_norm": 0.0939495861530304, "global_step": 17432, "epoch": 164, "lr": 7.55973344315194e-05} +{"train_loss": 0.016095487400889397, "grad_norm": 0.07985429465770721, "global_step": 17433, "epoch": 164, "lr": 7.559478845995335e-05} +{"train_loss": 0.014836291782557964, "grad_norm": 0.07530621439218521, "global_step": 17434, "epoch": 164, "lr": 7.559224239845835e-05} +{"train_loss": 0.015937969088554382, "grad_norm": 0.062338687479496, "global_step": 17435, "epoch": 164, "lr": 7.558969624704333e-05} +{"train_loss": 0.015446212142705917, "grad_norm": 0.05232979357242584, "global_step": 17436, "epoch": 164, "lr": 7.558715000571726e-05} +{"train_loss": 0.016043560579419136, "grad_norm": 0.109130859375, "global_step": 17437, "epoch": 164, "lr": 7.558460367448906e-05} +{"train_loss": 0.017064427956938744, "grad_norm": 0.1049060970544815, "global_step": 17438, "epoch": 164, "lr": 7.558205725336769e-05} +{"train_loss": 0.017744280397892, "grad_norm": 0.10080790519714355, "global_step": 17439, "epoch": 164, "lr": 7.557951074236209e-05} +{"train_loss": 0.01685820147395134, "grad_norm": 0.05301402136683464, "global_step": 17440, "epoch": 164, "lr": 7.557696414148123e-05} +{"train_loss": 0.017068341374397278, "grad_norm": 0.14802344143390656, "global_step": 17441, "epoch": 164, "lr": 7.557441745073403e-05} +{"train_loss": 0.016200080513954163, "grad_norm": 0.05029083043336868, "global_step": 17442, "epoch": 164, "lr": 7.557187067012946e-05} +{"train_loss": 0.013604113832116127, "grad_norm": 0.06511116027832031, "global_step": 17443, "epoch": 164, "lr": 7.556932379967645e-05} +{"train_loss": 0.0167007464915514, "grad_norm": 0.07180943340063095, "global_step": 17444, "epoch": 164, "lr": 7.556677683938396e-05} +{"train_loss": 0.01638037897646427, "grad_norm": 0.058934446424245834, "global_step": 17445, "epoch": 164, "lr": 7.556422978926093e-05} +{"train_loss": 0.016599372029304504, "grad_norm": 0.06684087961912155, "global_step": 17446, "epoch": 164, "lr": 7.556168264931631e-05} +{"train_loss": 0.015176148153841496, "grad_norm": 0.06223152205348015, "global_step": 17447, "epoch": 164, "lr": 7.555913541955906e-05} +{"train_loss": 0.014821273274719715, "grad_norm": 0.11898905783891678, "global_step": 17448, "epoch": 164, "lr": 7.555658809999811e-05} +{"train_loss": 0.014428691938519478, "grad_norm": 0.04907217621803284, "global_step": 17449, "epoch": 164, "lr": 7.555404069064245e-05} +{"train_loss": 0.01681920886039734, "grad_norm": 0.053239792585372925, "global_step": 17450, "epoch": 164, "lr": 7.555149319150098e-05} +{"train_loss": 0.015308843925595284, "grad_norm": 0.09911061823368073, "global_step": 17451, "epoch": 164, "lr": 7.554894560258268e-05} +{"train_loss": 0.018035877496004105, "grad_norm": 0.056683044880628586, "global_step": 17452, "epoch": 164, "lr": 7.554639792389651e-05} +{"train_loss": 0.017923206090927124, "grad_norm": 0.09700452536344528, "global_step": 17453, "epoch": 164, "lr": 7.554385015545139e-05} +{"train_loss": 0.016699079424142838, "grad_norm": 0.09594886749982834, "global_step": 17454, "epoch": 164, "lr": 7.55413022972563e-05} +{"train_loss": 0.01745673269033432, "grad_norm": 0.06713014096021652, "global_step": 17455, "epoch": 164, "lr": 7.553875434932019e-05} +{"train_loss": 0.01620682142674923, "grad_norm": 0.1025993674993515, "global_step": 17456, "epoch": 164, "lr": 7.553620631165198e-05} +{"train_loss": 0.015437737107276917, "grad_norm": 0.05011937394738197, "global_step": 17457, "epoch": 164, "lr": 7.553365818426066e-05} +{"train_loss": 0.016095461323857307, "grad_norm": 0.18246416747570038, "global_step": 17458, "epoch": 164, "lr": 7.553110996715518e-05} +{"train_loss": 0.014751401729881763, "grad_norm": 0.05726141482591629, "global_step": 17459, "epoch": 164, "lr": 7.552856166034445e-05} +{"train_loss": 0.017146605998277664, "grad_norm": 0.05989678576588631, "global_step": 17460, "epoch": 164, "lr": 7.552601326383748e-05} +{"train_loss": 0.016120363026857376, "grad_norm": 0.07443948090076447, "global_step": 17461, "epoch": 164, "lr": 7.55234647776432e-05} +{"train_loss": 0.01577616110444069, "grad_norm": 0.06092974543571472, "global_step": 17462, "epoch": 164, "lr": 7.552091620177056e-05} +{"train_loss": 0.015727732330560684, "grad_norm": 0.07949318736791611, "global_step": 17463, "epoch": 164, "lr": 7.551836753622852e-05} +{"train_loss": 0.017783842980861664, "grad_norm": 0.06279056519269943, "global_step": 17464, "epoch": 164, "lr": 7.551581878102605e-05} +{"train_loss": 0.014953836798667908, "grad_norm": 0.10022337734699249, "global_step": 17465, "epoch": 164, "lr": 7.551326993617207e-05} +{"train_loss": 0.016527676954865456, "grad_norm": 0.06876810640096664, "global_step": 17466, "epoch": 164, "lr": 7.551072100167556e-05} +{"train_loss": 0.016095882281661034, "grad_norm": 0.13725592195987701, "global_step": 17467, "epoch": 164, "lr": 7.550817197754547e-05} +{"train_loss": 0.016297325491905212, "grad_norm": 0.09012053161859512, "global_step": 17468, "epoch": 164, "lr": 7.550562286379075e-05} +{"train_loss": 0.017808526754379272, "grad_norm": 0.06446875631809235, "global_step": 17469, "epoch": 164, "lr": 7.550307366042037e-05} +{"train_loss": 0.015955224633216858, "grad_norm": 0.06408561021089554, "global_step": 17470, "epoch": 164, "lr": 7.55005243674433e-05} +{"train_loss": 0.015679366886615753, "grad_norm": 0.07534487545490265, "global_step": 17471, "epoch": 164, "lr": 7.549797498486846e-05} +{"train_loss": 0.017486844211816788, "grad_norm": 0.07697629928588867, "global_step": 17472, "epoch": 164, "lr": 7.54954255127048e-05} +{"train_loss": 0.015658224001526833, "grad_norm": 0.0678897351026535, "global_step": 17473, "epoch": 164, "lr": 7.549287595096134e-05} +{"train_loss": 0.016277533024549484, "grad_norm": 0.11768817901611328, "global_step": 17474, "epoch": 164, "lr": 7.549032629964698e-05} +{"train_loss": 0.01644909381866455, "grad_norm": 0.06015736609697342, "global_step": 17475, "epoch": 164, "lr": 7.54877765587707e-05} +{"train_loss": 0.013924002647399902, "grad_norm": 0.15505611896514893, "global_step": 17476, "epoch": 164, "lr": 7.548522672834149e-05} +{"train_loss": 0.014482022263109684, "grad_norm": 0.06131541356444359, "global_step": 17477, "epoch": 164, "lr": 7.548267680836823e-05} +{"train_loss": 0.015975376591086388, "grad_norm": 0.07894691079854965, "global_step": 17478, "epoch": 164, "lr": 7.548012679885995e-05} +{"train_loss": 0.015531636774539948, "grad_norm": 0.06132965907454491, "global_step": 17479, "epoch": 164, "lr": 7.547757669982559e-05} +{"train_loss": 0.01881030574440956, "grad_norm": 0.11672148108482361, "global_step": 17480, "epoch": 164, "lr": 7.547502651127407e-05} +{"train_loss": 0.014104228466749191, "grad_norm": 0.10323943197727203, "global_step": 17481, "epoch": 164, "lr": 7.547247623321443e-05} +{"train_loss": 0.014899156987667084, "grad_norm": 0.07324232906103134, "global_step": 17482, "epoch": 164, "lr": 7.546992586565555e-05} +{"train_loss": 0.015478979796171188, "grad_norm": 0.09940162301063538, "global_step": 17483, "epoch": 164, "lr": 7.546737540860645e-05} +{"train_loss": 0.01679958403110504, "grad_norm": 0.1317136436700821, "global_step": 17484, "epoch": 164, "lr": 7.546482486207607e-05} +{"train_loss": 0.016508303582668304, "grad_norm": 0.13813777267932892, "global_step": 17485, "epoch": 164, "lr": 7.546227422607336e-05} +{"train_loss": 0.014550899155437946, "grad_norm": 0.08489435166120529, "global_step": 17486, "epoch": 164, "lr": 7.545972350060731e-05} +{"train_loss": 0.017210932448506355, "grad_norm": 0.10930787026882172, "global_step": 17487, "epoch": 164, "lr": 7.545717268568684e-05} +{"train_loss": 0.01970335841178894, "grad_norm": 0.07551367580890656, "global_step": 17488, "epoch": 164, "lr": 7.545462178132094e-05} +{"train_loss": 0.016442630642076145, "grad_norm": 0.08856408298015594, "global_step": 17489, "epoch": 164, "lr": 7.545207078751857e-05, "val_loss": 0.01883189007639885} +{"train_loss": 0.018102969974279404, "grad_norm": 0.0753418430685997, "global_step": 17490, "epoch": 165, "lr": 7.544951970428871e-05} +{"train_loss": 0.017598837614059448, "grad_norm": 0.07962290197610855, "global_step": 17491, "epoch": 165, "lr": 7.544696853164027e-05} +{"train_loss": 0.017101259902119637, "grad_norm": 0.05800877511501312, "global_step": 17492, "epoch": 165, "lr": 7.544441726958227e-05} +{"train_loss": 0.015726543962955475, "grad_norm": 0.05580757185816765, "global_step": 17493, "epoch": 165, "lr": 7.544186591812366e-05} +{"train_loss": 0.017482422292232513, "grad_norm": 0.08239003270864487, "global_step": 17494, "epoch": 165, "lr": 7.54393144772734e-05} +{"train_loss": 0.01585877314209938, "grad_norm": 0.19660422205924988, "global_step": 17495, "epoch": 165, "lr": 7.543676294704043e-05} +{"train_loss": 0.01860535703599453, "grad_norm": 0.06027780473232269, "global_step": 17496, "epoch": 165, "lr": 7.543421132743375e-05} +{"train_loss": 0.020133955404162407, "grad_norm": 0.09977560490369797, "global_step": 17497, "epoch": 165, "lr": 7.543165961846231e-05} +{"train_loss": 0.015576206147670746, "grad_norm": 0.1353483498096466, "global_step": 17498, "epoch": 165, "lr": 7.542910782013508e-05} +{"train_loss": 0.016559718176722527, "grad_norm": 0.13676629960536957, "global_step": 17499, "epoch": 165, "lr": 7.542655593246103e-05} +{"train_loss": 0.015379432588815689, "grad_norm": 0.07078739255666733, "global_step": 17500, "epoch": 165, "lr": 7.542400395544912e-05} +{"train_loss": 0.014630543999373913, "grad_norm": 0.08564520627260208, "global_step": 17501, "epoch": 165, "lr": 7.54214518891083e-05} +{"train_loss": 0.016517169773578644, "grad_norm": 0.09563732147216797, "global_step": 17502, "epoch": 165, "lr": 7.541889973344758e-05} +{"train_loss": 0.014006372541189194, "grad_norm": 0.07426133751869202, "global_step": 17503, "epoch": 165, "lr": 7.541634748847588e-05} +{"train_loss": 0.017292849719524384, "grad_norm": 0.062040071934461594, "global_step": 17504, "epoch": 165, "lr": 7.541379515420218e-05} +{"train_loss": 0.015221362933516502, "grad_norm": 0.04572124779224396, "global_step": 17505, "epoch": 165, "lr": 7.541124273063549e-05} +{"train_loss": 0.016260767355561256, "grad_norm": 0.07906606793403625, "global_step": 17506, "epoch": 165, "lr": 7.540869021778472e-05} +{"train_loss": 0.014576459303498268, "grad_norm": 0.07912221550941467, "global_step": 17507, "epoch": 165, "lr": 7.540613761565886e-05} +{"train_loss": 0.016719702631235123, "grad_norm": 0.10924986749887466, "global_step": 17508, "epoch": 165, "lr": 7.54035849242669e-05} +{"train_loss": 0.01628028228878975, "grad_norm": 0.06878918409347534, "global_step": 17509, "epoch": 165, "lr": 7.540103214361776e-05} +{"train_loss": 0.01715703308582306, "grad_norm": 0.09650194644927979, "global_step": 17510, "epoch": 165, "lr": 7.539847927372045e-05} +{"train_loss": 0.014869765378534794, "grad_norm": 0.1373167634010315, "global_step": 17511, "epoch": 165, "lr": 7.539592631458394e-05} +{"train_loss": 0.01645508222281933, "grad_norm": 0.09172850102186203, "global_step": 17512, "epoch": 165, "lr": 7.53933732662172e-05} +{"train_loss": 0.015305970795452595, "grad_norm": 0.05448652803897858, "global_step": 17513, "epoch": 165, "lr": 7.539082012862917e-05} +{"train_loss": 0.016955722123384476, "grad_norm": 0.06819384545087814, "global_step": 17514, "epoch": 165, "lr": 7.538826690182885e-05} +{"train_loss": 0.01853090524673462, "grad_norm": 0.12391413003206253, "global_step": 17515, "epoch": 165, "lr": 7.53857135858252e-05} +{"train_loss": 0.014352524653077126, "grad_norm": 0.05549461022019386, "global_step": 17516, "epoch": 165, "lr": 7.538316018062718e-05} +{"train_loss": 0.016666295006871223, "grad_norm": 0.0638534426689148, "global_step": 17517, "epoch": 165, "lr": 7.538060668624379e-05} +{"train_loss": 0.01457471027970314, "grad_norm": 0.08184541016817093, "global_step": 17518, "epoch": 165, "lr": 7.537805310268398e-05} +{"train_loss": 0.016303647309541702, "grad_norm": 0.07362022250890732, "global_step": 17519, "epoch": 165, "lr": 7.537549942995672e-05} +{"train_loss": 0.016398422420024872, "grad_norm": 0.05503705143928528, "global_step": 17520, "epoch": 165, "lr": 7.5372945668071e-05} +{"train_loss": 0.014745866879820824, "grad_norm": 0.05733417347073555, "global_step": 17521, "epoch": 165, "lr": 7.537039181703578e-05} +{"train_loss": 0.01776197925209999, "grad_norm": 0.06945421546697617, "global_step": 17522, "epoch": 165, "lr": 7.536783787686004e-05} +{"train_loss": 0.015862297266721725, "grad_norm": 0.07342083007097244, "global_step": 17523, "epoch": 165, "lr": 7.536528384755274e-05} +{"train_loss": 0.015688516199588776, "grad_norm": 0.09151774644851685, "global_step": 17524, "epoch": 165, "lr": 7.536272972912289e-05} +{"train_loss": 0.01709425076842308, "grad_norm": 0.0779668539762497, "global_step": 17525, "epoch": 165, "lr": 7.536017552157942e-05} +{"train_loss": 0.018893782049417496, "grad_norm": 0.12089414149522781, "global_step": 17526, "epoch": 165, "lr": 7.535762122493131e-05} +{"train_loss": 0.015039711259305477, "grad_norm": 0.08775986731052399, "global_step": 17527, "epoch": 165, "lr": 7.535506683918756e-05} +{"train_loss": 0.017690617591142654, "grad_norm": 0.17380374670028687, "global_step": 17528, "epoch": 165, "lr": 7.535251236435712e-05} +{"train_loss": 0.014266366139054298, "grad_norm": 0.11551293730735779, "global_step": 17529, "epoch": 165, "lr": 7.5349957800449e-05} +{"train_loss": 0.01564035937190056, "grad_norm": 0.07439670711755753, "global_step": 17530, "epoch": 165, "lr": 7.534740314747213e-05} +{"train_loss": 0.012727269902825356, "grad_norm": 0.06124551221728325, "global_step": 17531, "epoch": 165, "lr": 7.534484840543552e-05} +{"train_loss": 0.019058246165513992, "grad_norm": 0.07488173991441727, "global_step": 17532, "epoch": 165, "lr": 7.534229357434814e-05} +{"train_loss": 0.01623588241636753, "grad_norm": 0.15589112043380737, "global_step": 17533, "epoch": 165, "lr": 7.533973865421894e-05} +{"train_loss": 0.017945734784007072, "grad_norm": 0.06049981340765953, "global_step": 17534, "epoch": 165, "lr": 7.533718364505695e-05} +{"train_loss": 0.018363600596785545, "grad_norm": 0.08675514161586761, "global_step": 17535, "epoch": 165, "lr": 7.53346285468711e-05} +{"train_loss": 0.016876228153705597, "grad_norm": 0.09446138143539429, "global_step": 17536, "epoch": 165, "lr": 7.533207335967037e-05} +{"train_loss": 0.01589253917336464, "grad_norm": 0.1046978086233139, "global_step": 17537, "epoch": 165, "lr": 7.532951808346375e-05} +{"train_loss": 0.01554579846560955, "grad_norm": 0.0759560763835907, "global_step": 17538, "epoch": 165, "lr": 7.532696271826026e-05} +{"train_loss": 0.015467539429664612, "grad_norm": 0.10511928796768188, "global_step": 17539, "epoch": 165, "lr": 7.53244072640688e-05} +{"train_loss": 0.01662585511803627, "grad_norm": 0.07149843871593475, "global_step": 17540, "epoch": 165, "lr": 7.532185172089838e-05} +{"train_loss": 0.017276540398597717, "grad_norm": 0.1275663822889328, "global_step": 17541, "epoch": 165, "lr": 7.5319296088758e-05} +{"train_loss": 0.016530435532331467, "grad_norm": 0.06320462375879288, "global_step": 17542, "epoch": 165, "lr": 7.531674036765662e-05} +{"train_loss": 0.01875270903110504, "grad_norm": 0.068548783659935, "global_step": 17543, "epoch": 165, "lr": 7.531418455760324e-05} +{"train_loss": 0.01627189666032791, "grad_norm": 0.0859253779053688, "global_step": 17544, "epoch": 165, "lr": 7.531162865860682e-05} +{"train_loss": 0.014286977238953114, "grad_norm": 0.04026983305811882, "global_step": 17545, "epoch": 165, "lr": 7.530907267067633e-05} +{"train_loss": 0.015904784202575684, "grad_norm": 0.05289466306567192, "global_step": 17546, "epoch": 165, "lr": 7.530651659382078e-05} +{"train_loss": 0.017171625047922134, "grad_norm": 0.1033354103565216, "global_step": 17547, "epoch": 165, "lr": 7.530396042804912e-05} +{"train_loss": 0.016248811036348343, "grad_norm": 0.07068001478910446, "global_step": 17548, "epoch": 165, "lr": 7.530140417337037e-05} +{"train_loss": 0.017350103706121445, "grad_norm": 0.10654439777135849, "global_step": 17549, "epoch": 165, "lr": 7.529884782979346e-05} +{"train_loss": 0.01769108697772026, "grad_norm": 0.10870544612407684, "global_step": 17550, "epoch": 165, "lr": 7.529629139732743e-05} +{"train_loss": 0.01911904662847519, "grad_norm": 0.13323481380939484, "global_step": 17551, "epoch": 165, "lr": 7.529373487598122e-05} +{"train_loss": 0.016268575564026833, "grad_norm": 0.15994076430797577, "global_step": 17552, "epoch": 165, "lr": 7.529117826576381e-05} +{"train_loss": 0.018616244196891785, "grad_norm": 0.09853824228048325, "global_step": 17553, "epoch": 165, "lr": 7.528862156668422e-05} +{"train_loss": 0.014556562528014183, "grad_norm": 0.04950655624270439, "global_step": 17554, "epoch": 165, "lr": 7.528606477875141e-05} +{"train_loss": 0.018025003373622894, "grad_norm": 0.0833330899477005, "global_step": 17555, "epoch": 165, "lr": 7.528350790197436e-05} +{"train_loss": 0.018049143254756927, "grad_norm": 0.09949424117803574, "global_step": 17556, "epoch": 165, "lr": 7.528095093636205e-05} +{"train_loss": 0.016098521649837494, "grad_norm": 0.1029515340924263, "global_step": 17557, "epoch": 165, "lr": 7.527839388192348e-05} +{"train_loss": 0.01496086735278368, "grad_norm": 0.07252884656190872, "global_step": 17558, "epoch": 165, "lr": 7.527583673866764e-05} +{"train_loss": 0.015972986817359924, "grad_norm": 0.1438283771276474, "global_step": 17559, "epoch": 165, "lr": 7.527327950660347e-05} +{"train_loss": 0.016078297048807144, "grad_norm": 0.10481756925582886, "global_step": 17560, "epoch": 165, "lr": 7.527072218574003e-05} +{"train_loss": 0.016756845638155937, "grad_norm": 0.10557880997657776, "global_step": 17561, "epoch": 165, "lr": 7.526816477608625e-05} +{"train_loss": 0.015216364525258541, "grad_norm": 0.057818856090307236, "global_step": 17562, "epoch": 165, "lr": 7.52656072776511e-05} +{"train_loss": 0.01663130894303322, "grad_norm": 0.068821981549263, "global_step": 17563, "epoch": 165, "lr": 7.526304969044361e-05} +{"train_loss": 0.016585394740104675, "grad_norm": 0.06258615851402283, "global_step": 17564, "epoch": 165, "lr": 7.526049201447276e-05} +{"train_loss": 0.014744872227311134, "grad_norm": 0.061251137405633926, "global_step": 17565, "epoch": 165, "lr": 7.525793424974753e-05} +{"train_loss": 0.017061861231923103, "grad_norm": 0.08569250255823135, "global_step": 17566, "epoch": 165, "lr": 7.525537639627687e-05} +{"train_loss": 0.01537359319627285, "grad_norm": 0.10598961263895035, "global_step": 17567, "epoch": 165, "lr": 7.525281845406983e-05} +{"train_loss": 0.013795854523777962, "grad_norm": 0.07163291424512863, "global_step": 17568, "epoch": 165, "lr": 7.525026042313537e-05} +{"train_loss": 0.015040801838040352, "grad_norm": 0.07498642057180405, "global_step": 17569, "epoch": 165, "lr": 7.524770230348245e-05} +{"train_loss": 0.016901833936572075, "grad_norm": 0.06443734467029572, "global_step": 17570, "epoch": 165, "lr": 7.524514409512012e-05} +{"train_loss": 0.017521411180496216, "grad_norm": 0.09952973574399948, "global_step": 17571, "epoch": 165, "lr": 7.52425857980573e-05} +{"train_loss": 0.01832103542983532, "grad_norm": 0.1051158756017685, "global_step": 17572, "epoch": 165, "lr": 7.524002741230301e-05} +{"train_loss": 0.017285702750086784, "grad_norm": 0.06009551137685776, "global_step": 17573, "epoch": 165, "lr": 7.523746893786628e-05} +{"train_loss": 0.01726897805929184, "grad_norm": 0.07715771347284317, "global_step": 17574, "epoch": 165, "lr": 7.523491037475601e-05} +{"train_loss": 0.015562424436211586, "grad_norm": 0.06511194258928299, "global_step": 17575, "epoch": 165, "lr": 7.523235172298128e-05} +{"train_loss": 0.01581096462905407, "grad_norm": 0.06147991865873337, "global_step": 17576, "epoch": 165, "lr": 7.522979298255102e-05} +{"train_loss": 0.01759851910173893, "grad_norm": 0.0834047943353653, "global_step": 17577, "epoch": 165, "lr": 7.522723415347424e-05} +{"train_loss": 0.0185384564101696, "grad_norm": 0.07430636137723923, "global_step": 17578, "epoch": 165, "lr": 7.522467523575992e-05} +{"train_loss": 0.019480805844068527, "grad_norm": 0.05751081183552742, "global_step": 17579, "epoch": 165, "lr": 7.522211622941709e-05} +{"train_loss": 0.015521805733442307, "grad_norm": 0.05559675768017769, "global_step": 17580, "epoch": 165, "lr": 7.521955713445469e-05} +{"train_loss": 0.018923049792647362, "grad_norm": 0.07264871895313263, "global_step": 17581, "epoch": 165, "lr": 7.521699795088173e-05} +{"train_loss": 0.016982970759272575, "grad_norm": 0.06893820315599442, "global_step": 17582, "epoch": 165, "lr": 7.521443867870723e-05} +{"train_loss": 0.01609782874584198, "grad_norm": 0.08126577734947205, "global_step": 17583, "epoch": 165, "lr": 7.521187931794014e-05} +{"train_loss": 0.016110872849822044, "grad_norm": 0.07475396245718002, "global_step": 17584, "epoch": 165, "lr": 7.520931986858947e-05} +{"train_loss": 0.016591183841228485, "grad_norm": 0.10014039278030396, "global_step": 17585, "epoch": 165, "lr": 7.520676033066422e-05} +{"train_loss": 0.013304557651281357, "grad_norm": 0.1032792404294014, "global_step": 17586, "epoch": 165, "lr": 7.520420070417335e-05} +{"train_loss": 0.014903604984283447, "grad_norm": 0.1284964382648468, "global_step": 17587, "epoch": 165, "lr": 7.520164098912593e-05} +{"train_loss": 0.013999132439494133, "grad_norm": 0.1424623280763626, "global_step": 17588, "epoch": 165, "lr": 7.519908118553086e-05} +{"train_loss": 0.01649828627705574, "grad_norm": 0.10643559694290161, "global_step": 17589, "epoch": 165, "lr": 7.51965212933972e-05} +{"train_loss": 0.014784705825150013, "grad_norm": 0.07935098558664322, "global_step": 17590, "epoch": 165, "lr": 7.519396131273393e-05} +{"train_loss": 0.0156928189098835, "grad_norm": 0.05469084531068802, "global_step": 17591, "epoch": 165, "lr": 7.519140124355e-05} +{"train_loss": 0.014969808049499989, "grad_norm": 0.0540236160159111, "global_step": 17592, "epoch": 165, "lr": 7.518884108585447e-05} +{"train_loss": 0.01777978613972664, "grad_norm": 0.06353891640901566, "global_step": 17593, "epoch": 165, "lr": 7.518628083965631e-05} +{"train_loss": 0.01592787727713585, "grad_norm": 0.10110148787498474, "global_step": 17594, "epoch": 165, "lr": 7.518372050496449e-05} +{"train_loss": 0.016403728990622285, "grad_norm": 0.08180152624845505, "global_step": 17595, "epoch": 165, "lr": 7.518116008178805e-05, "val_loss": 0.019107719883322716} +{"train_loss": 0.01725434511899948, "grad_norm": 0.05002943426370621, "global_step": 17596, "epoch": 166, "lr": 7.517859957013595e-05} +{"train_loss": 0.014312772080302238, "grad_norm": 0.07861027121543884, "global_step": 17597, "epoch": 166, "lr": 7.517603897001721e-05} +{"train_loss": 0.019108012318611145, "grad_norm": 0.07897857576608658, "global_step": 17598, "epoch": 166, "lr": 7.517347828144081e-05} +{"train_loss": 0.01659977063536644, "grad_norm": 0.09799584746360779, "global_step": 17599, "epoch": 166, "lr": 7.517091750441576e-05} +{"train_loss": 0.017414426431059837, "grad_norm": 0.10667186230421066, "global_step": 17600, "epoch": 166, "lr": 7.516835663895106e-05} +{"train_loss": 0.016645271331071854, "grad_norm": 0.0893632248044014, "global_step": 17601, "epoch": 166, "lr": 7.516579568505569e-05} +{"train_loss": 0.014628520235419273, "grad_norm": 0.056279804557561874, "global_step": 17602, "epoch": 166, "lr": 7.516323464273867e-05} +{"train_loss": 0.015410281717777252, "grad_norm": 0.1045784205198288, "global_step": 17603, "epoch": 166, "lr": 7.516067351200896e-05} +{"train_loss": 0.01788075640797615, "grad_norm": 0.10162533074617386, "global_step": 17604, "epoch": 166, "lr": 7.51581122928756e-05} +{"train_loss": 0.015517344698309898, "grad_norm": 0.06241333857178688, "global_step": 17605, "epoch": 166, "lr": 7.515555098534759e-05} +{"train_loss": 0.016822634264826775, "grad_norm": 0.09096033126115799, "global_step": 17606, "epoch": 166, "lr": 7.515298958943389e-05} +{"train_loss": 0.012963550165295601, "grad_norm": 0.11718598008155823, "global_step": 17607, "epoch": 166, "lr": 7.515042810514355e-05} +{"train_loss": 0.017882389947772026, "grad_norm": 0.061304282397031784, "global_step": 17608, "epoch": 166, "lr": 7.514786653248551e-05} +{"train_loss": 0.013597354292869568, "grad_norm": 0.04883306473493576, "global_step": 17609, "epoch": 166, "lr": 7.514530487146882e-05} +{"train_loss": 0.016899237409234047, "grad_norm": 0.06912025064229965, "global_step": 17610, "epoch": 166, "lr": 7.514274312210246e-05} +{"train_loss": 0.016696549952030182, "grad_norm": 0.10050952434539795, "global_step": 17611, "epoch": 166, "lr": 7.514018128439546e-05} +{"train_loss": 0.01534486934542656, "grad_norm": 0.06802013516426086, "global_step": 17612, "epoch": 166, "lr": 7.513761935835677e-05} +{"train_loss": 0.018925843760371208, "grad_norm": 0.07356858253479004, "global_step": 17613, "epoch": 166, "lr": 7.513505734399542e-05} +{"train_loss": 0.015192972496151924, "grad_norm": 0.13539311289787292, "global_step": 17614, "epoch": 166, "lr": 7.513249524132042e-05} +{"train_loss": 0.01713360846042633, "grad_norm": 0.06698717176914215, "global_step": 17615, "epoch": 166, "lr": 7.512993305034074e-05} +{"train_loss": 0.016152527183294296, "grad_norm": 0.044968388974666595, "global_step": 17616, "epoch": 166, "lr": 7.512737077106542e-05} +{"train_loss": 0.014426512643694878, "grad_norm": 0.1549062728881836, "global_step": 17617, "epoch": 166, "lr": 7.512480840350345e-05} +{"train_loss": 0.018212880939245224, "grad_norm": 0.12582363188266754, "global_step": 17618, "epoch": 166, "lr": 7.512224594766382e-05} +{"train_loss": 0.013823263347148895, "grad_norm": 0.054734326899051666, "global_step": 17619, "epoch": 166, "lr": 7.511968340355556e-05} +{"train_loss": 0.01826806738972664, "grad_norm": 0.08240871131420135, "global_step": 17620, "epoch": 166, "lr": 7.511712077118764e-05} +{"train_loss": 0.018426885828375816, "grad_norm": 0.11128167062997818, "global_step": 17621, "epoch": 166, "lr": 7.511455805056908e-05} +{"train_loss": 0.017148567363619804, "grad_norm": 0.0792635902762413, "global_step": 17622, "epoch": 166, "lr": 7.511199524170889e-05} +{"train_loss": 0.016205348074436188, "grad_norm": 0.06362685561180115, "global_step": 17623, "epoch": 166, "lr": 7.510943234461607e-05} +{"train_loss": 0.018519315868616104, "grad_norm": 0.1107560321688652, "global_step": 17624, "epoch": 166, "lr": 7.510686935929962e-05} +{"train_loss": 0.014004794880747795, "grad_norm": 0.07723170518875122, "global_step": 17625, "epoch": 166, "lr": 7.510430628576856e-05} +{"train_loss": 0.015394405461847782, "grad_norm": 0.07060215622186661, "global_step": 17626, "epoch": 166, "lr": 7.510174312403189e-05} +{"train_loss": 0.019031507894396782, "grad_norm": 0.17776241898536682, "global_step": 17627, "epoch": 166, "lr": 7.509917987409859e-05} +{"train_loss": 0.016974065452814102, "grad_norm": 0.12366729229688644, "global_step": 17628, "epoch": 166, "lr": 7.509661653597771e-05} +{"train_loss": 0.01706846058368683, "grad_norm": 0.06734087318181992, "global_step": 17629, "epoch": 166, "lr": 7.50940531096782e-05} +{"train_loss": 0.017614465206861496, "grad_norm": 0.08729253709316254, "global_step": 17630, "epoch": 166, "lr": 7.509148959520913e-05} +{"train_loss": 0.013905475847423077, "grad_norm": 0.06887869536876678, "global_step": 17631, "epoch": 166, "lr": 7.508892599257949e-05} +{"train_loss": 0.01547195203602314, "grad_norm": 0.05618147924542427, "global_step": 17632, "epoch": 166, "lr": 7.508636230179824e-05} +{"train_loss": 0.01699424348771572, "grad_norm": 0.052432265132665634, "global_step": 17633, "epoch": 166, "lr": 7.508379852287444e-05} +{"train_loss": 0.01765001192688942, "grad_norm": 0.11291375756263733, "global_step": 17634, "epoch": 166, "lr": 7.508123465581708e-05} +{"train_loss": 0.01643737033009529, "grad_norm": 0.08553626388311386, "global_step": 17635, "epoch": 166, "lr": 7.507867070063517e-05} +{"train_loss": 0.01728694885969162, "grad_norm": 0.09519807994365692, "global_step": 17636, "epoch": 166, "lr": 7.50761066573377e-05} +{"train_loss": 0.014487648382782936, "grad_norm": 0.05363494157791138, "global_step": 17637, "epoch": 166, "lr": 7.507354252593373e-05} +{"train_loss": 0.016672927886247635, "grad_norm": 0.05490026995539665, "global_step": 17638, "epoch": 166, "lr": 7.507097830643221e-05} +{"train_loss": 0.016597025096416473, "grad_norm": 0.06131264567375183, "global_step": 17639, "epoch": 166, "lr": 7.506841399884217e-05} +{"train_loss": 0.018891721963882446, "grad_norm": 0.09607148170471191, "global_step": 17640, "epoch": 166, "lr": 7.506584960317266e-05} +{"train_loss": 0.016113080084323883, "grad_norm": 0.08646932244300842, "global_step": 17641, "epoch": 166, "lr": 7.506328511943263e-05} +{"train_loss": 0.016513407230377197, "grad_norm": 0.15064890682697296, "global_step": 17642, "epoch": 166, "lr": 7.506072054763111e-05} +{"train_loss": 0.019358351826667786, "grad_norm": 0.15616394579410553, "global_step": 17643, "epoch": 166, "lr": 7.505815588777713e-05} +{"train_loss": 0.016666654497385025, "grad_norm": 0.07404809445142746, "global_step": 17644, "epoch": 166, "lr": 7.505559113987967e-05} +{"train_loss": 0.014374787919223309, "grad_norm": 0.07943888008594513, "global_step": 17645, "epoch": 166, "lr": 7.505302630394778e-05} +{"train_loss": 0.018764294683933258, "grad_norm": 0.0990172028541565, "global_step": 17646, "epoch": 166, "lr": 7.505046137999044e-05} +{"train_loss": 0.013253746554255486, "grad_norm": 0.06240566819906235, "global_step": 17647, "epoch": 166, "lr": 7.504789636801668e-05} +{"train_loss": 0.01949435845017433, "grad_norm": 0.08796641230583191, "global_step": 17648, "epoch": 166, "lr": 7.50453312680355e-05} +{"train_loss": 0.015970535576343536, "grad_norm": 0.08403340727090836, "global_step": 17649, "epoch": 166, "lr": 7.504276608005591e-05} +{"train_loss": 0.018936995416879654, "grad_norm": 0.0925733745098114, "global_step": 17650, "epoch": 166, "lr": 7.504020080408695e-05} +{"train_loss": 0.017086246982216835, "grad_norm": 0.06274165958166122, "global_step": 17651, "epoch": 166, "lr": 7.50376354401376e-05} +{"train_loss": 0.016301598399877548, "grad_norm": 0.07358761876821518, "global_step": 17652, "epoch": 166, "lr": 7.50350699882169e-05} +{"train_loss": 0.015232665464282036, "grad_norm": 0.061942193657159805, "global_step": 17653, "epoch": 166, "lr": 7.503250444833384e-05} +{"train_loss": 0.01983005739748478, "grad_norm": 0.07477854937314987, "global_step": 17654, "epoch": 166, "lr": 7.502993882049745e-05} +{"train_loss": 0.015948910266160965, "grad_norm": 0.1566680371761322, "global_step": 17655, "epoch": 166, "lr": 7.502737310471674e-05} +{"train_loss": 0.01571446657180786, "grad_norm": 0.09219003468751907, "global_step": 17656, "epoch": 166, "lr": 7.502480730100073e-05} +{"train_loss": 0.01918783411383629, "grad_norm": 0.11261871457099915, "global_step": 17657, "epoch": 166, "lr": 7.50222414093584e-05} +{"train_loss": 0.01479201391339302, "grad_norm": 0.056101009249687195, "global_step": 17658, "epoch": 166, "lr": 7.501967542979884e-05} +{"train_loss": 0.01461498811841011, "grad_norm": 0.0661710649728775, "global_step": 17659, "epoch": 166, "lr": 7.501710936233099e-05} +{"train_loss": 0.018358023837208748, "grad_norm": 0.18050061166286469, "global_step": 17660, "epoch": 166, "lr": 7.50145432069639e-05} +{"train_loss": 0.016599522903561592, "grad_norm": 0.07565430551767349, "global_step": 17661, "epoch": 166, "lr": 7.50119769637066e-05} +{"train_loss": 0.015543588437139988, "grad_norm": 0.07427577674388885, "global_step": 17662, "epoch": 166, "lr": 7.500941063256808e-05} +{"train_loss": 0.018592137843370438, "grad_norm": 0.08750015497207642, "global_step": 17663, "epoch": 166, "lr": 7.500684421355737e-05} +{"train_loss": 0.01796654984354973, "grad_norm": 0.18525750935077667, "global_step": 17664, "epoch": 166, "lr": 7.500427770668348e-05} +{"train_loss": 0.015178333967924118, "grad_norm": 0.09738436341285706, "global_step": 17665, "epoch": 166, "lr": 7.500171111195544e-05} +{"train_loss": 0.014088591560721397, "grad_norm": 0.07272892445325851, "global_step": 17666, "epoch": 166, "lr": 7.499914442938225e-05} +{"train_loss": 0.01550964917987585, "grad_norm": 0.05952739715576172, "global_step": 17667, "epoch": 166, "lr": 7.499657765897297e-05} +{"train_loss": 0.01717568188905716, "grad_norm": 0.09593690186738968, "global_step": 17668, "epoch": 166, "lr": 7.499401080073655e-05} +{"train_loss": 0.017201097682118416, "grad_norm": 0.06696923822164536, "global_step": 17669, "epoch": 166, "lr": 7.499144385468205e-05} +{"train_loss": 0.017763176932930946, "grad_norm": 0.08937664330005646, "global_step": 17670, "epoch": 166, "lr": 7.498887682081851e-05} +{"train_loss": 0.01612555794417858, "grad_norm": 0.12635529041290283, "global_step": 17671, "epoch": 166, "lr": 7.49863096991549e-05} +{"train_loss": 0.015987731516361237, "grad_norm": 0.08458953350782394, "global_step": 17672, "epoch": 166, "lr": 7.498374248970025e-05} +{"train_loss": 0.015366281382739544, "grad_norm": 0.15957443416118622, "global_step": 17673, "epoch": 166, "lr": 7.498117519246362e-05} +{"train_loss": 0.014427050948143005, "grad_norm": 0.059496402740478516, "global_step": 17674, "epoch": 166, "lr": 7.4978607807454e-05} +{"train_loss": 0.01710820011794567, "grad_norm": 0.11151669919490814, "global_step": 17675, "epoch": 166, "lr": 7.497604033468041e-05} +{"train_loss": 0.016429241746664047, "grad_norm": 0.13221971690654755, "global_step": 17676, "epoch": 166, "lr": 7.49734727741519e-05} +{"train_loss": 0.016489047557115555, "grad_norm": 0.06232978403568268, "global_step": 17677, "epoch": 166, "lr": 7.497090512587744e-05} +{"train_loss": 0.01578369550406933, "grad_norm": 0.10395646095275879, "global_step": 17678, "epoch": 166, "lr": 7.496833738986608e-05} +{"train_loss": 0.016818488016724586, "grad_norm": 0.061984121799468994, "global_step": 17679, "epoch": 166, "lr": 7.496576956612686e-05} +{"train_loss": 0.015316500328481197, "grad_norm": 0.10437276214361191, "global_step": 17680, "epoch": 166, "lr": 7.496320165466876e-05} +{"train_loss": 0.014708434231579304, "grad_norm": 0.16676610708236694, "global_step": 17681, "epoch": 166, "lr": 7.496063365550085e-05} +{"train_loss": 0.01568695716559887, "grad_norm": 0.09490852057933807, "global_step": 17682, "epoch": 166, "lr": 7.49580655686321e-05} +{"train_loss": 0.016854839399456978, "grad_norm": 0.0734461098909378, "global_step": 17683, "epoch": 166, "lr": 7.495549739407157e-05} +{"train_loss": 0.014768220484256744, "grad_norm": 0.11929964274168015, "global_step": 17684, "epoch": 166, "lr": 7.49529291318283e-05} +{"train_loss": 0.016663623973727226, "grad_norm": 0.15201975405216217, "global_step": 17685, "epoch": 166, "lr": 7.495036078191126e-05} +{"train_loss": 0.019021179527044296, "grad_norm": 0.13267697393894196, "global_step": 17686, "epoch": 166, "lr": 7.494779234432951e-05} +{"train_loss": 0.015011431649327278, "grad_norm": 0.08069521933794022, "global_step": 17687, "epoch": 166, "lr": 7.494522381909207e-05} +{"train_loss": 0.01393977366387844, "grad_norm": 0.04104398563504219, "global_step": 17688, "epoch": 166, "lr": 7.494265520620796e-05} +{"train_loss": 0.015856388956308365, "grad_norm": 0.07512945681810379, "global_step": 17689, "epoch": 166, "lr": 7.494008650568621e-05} +{"train_loss": 0.016448257490992546, "grad_norm": 0.12330182641744614, "global_step": 17690, "epoch": 166, "lr": 7.493751771753584e-05} +{"train_loss": 0.01419215090572834, "grad_norm": 0.07452715188264847, "global_step": 17691, "epoch": 166, "lr": 7.493494884176588e-05} +{"train_loss": 0.015340132638812065, "grad_norm": 0.08747071772813797, "global_step": 17692, "epoch": 166, "lr": 7.493237987838535e-05} +{"train_loss": 0.017299626022577286, "grad_norm": 0.05467025563120842, "global_step": 17693, "epoch": 166, "lr": 7.49298108274033e-05} +{"train_loss": 0.0163821242749691, "grad_norm": 0.10832887142896652, "global_step": 17694, "epoch": 166, "lr": 7.492724168882871e-05} +{"train_loss": 0.015954939648509026, "grad_norm": 0.08260241150856018, "global_step": 17695, "epoch": 166, "lr": 7.492467246267063e-05} +{"train_loss": 0.015430019237101078, "grad_norm": 0.05574267357587814, "global_step": 17696, "epoch": 166, "lr": 7.492210314893812e-05} +{"train_loss": 0.014953463338315487, "grad_norm": 0.061800915747880936, "global_step": 17697, "epoch": 166, "lr": 7.491953374764015e-05} +{"train_loss": 0.015478197485208511, "grad_norm": 0.09446857124567032, "global_step": 17698, "epoch": 166, "lr": 7.491696425878579e-05} +{"train_loss": 0.01682646945118904, "grad_norm": 0.09441006183624268, "global_step": 17699, "epoch": 166, "lr": 7.491439468238403e-05} +{"train_loss": 0.016721757128834724, "grad_norm": 0.05407644808292389, "global_step": 17700, "epoch": 166, "lr": 7.491182501844396e-05} +{"train_loss": 0.016390496724337903, "grad_norm": 0.12115304917097092, "global_step": 17701, "epoch": 166, "lr": 7.490925526697455e-05, "val_loss": 0.01917063072323799} +{"train_loss": 0.016223663464188576, "grad_norm": 0.15046028792858124, "global_step": 17702, "epoch": 167, "lr": 7.490668542798484e-05} +{"train_loss": 0.01775018684566021, "grad_norm": 0.05531026050448418, "global_step": 17703, "epoch": 167, "lr": 7.490411550148388e-05} +{"train_loss": 0.01668483205139637, "grad_norm": 0.11628025025129318, "global_step": 17704, "epoch": 167, "lr": 7.49015454874807e-05} +{"train_loss": 0.01711735688149929, "grad_norm": 0.06773393601179123, "global_step": 17705, "epoch": 167, "lr": 7.489897538598432e-05} +{"train_loss": 0.014713254757225513, "grad_norm": 0.05134247988462448, "global_step": 17706, "epoch": 167, "lr": 7.489640519700375e-05} +{"train_loss": 0.015494105406105518, "grad_norm": 0.12509720027446747, "global_step": 17707, "epoch": 167, "lr": 7.489383492054804e-05} +{"train_loss": 0.016351571306586266, "grad_norm": 0.06137126311659813, "global_step": 17708, "epoch": 167, "lr": 7.489126455662622e-05} +{"train_loss": 0.015184522606432438, "grad_norm": 0.0537443570792675, "global_step": 17709, "epoch": 167, "lr": 7.488869410524732e-05} +{"train_loss": 0.016027262434363365, "grad_norm": 0.05619236081838608, "global_step": 17710, "epoch": 167, "lr": 7.48861235664204e-05} +{"train_loss": 0.016781779006123543, "grad_norm": 0.09375306963920593, "global_step": 17711, "epoch": 167, "lr": 7.488355294015445e-05} +{"train_loss": 0.017969433218240738, "grad_norm": 0.11756335198879242, "global_step": 17712, "epoch": 167, "lr": 7.48809822264585e-05} +{"train_loss": 0.014575313776731491, "grad_norm": 0.11647327244281769, "global_step": 17713, "epoch": 167, "lr": 7.487841142534161e-05} +{"train_loss": 0.015964314341545105, "grad_norm": 0.11105640977621078, "global_step": 17714, "epoch": 167, "lr": 7.48758405368128e-05} +{"train_loss": 0.016132494434714317, "grad_norm": 0.054964978247880936, "global_step": 17715, "epoch": 167, "lr": 7.48732695608811e-05} +{"train_loss": 0.015637464821338654, "grad_norm": 0.09436411410570145, "global_step": 17716, "epoch": 167, "lr": 7.487069849755557e-05} +{"train_loss": 0.018478618934750557, "grad_norm": 0.062260109931230545, "global_step": 17717, "epoch": 167, "lr": 7.48681273468452e-05} +{"train_loss": 0.014771101996302605, "grad_norm": 0.0931699201464653, "global_step": 17718, "epoch": 167, "lr": 7.486555610875905e-05} +{"train_loss": 0.01714317686855793, "grad_norm": 0.07600676268339157, "global_step": 17719, "epoch": 167, "lr": 7.486298478330615e-05} +{"train_loss": 0.016435984522104263, "grad_norm": 0.15928906202316284, "global_step": 17720, "epoch": 167, "lr": 7.486041337049555e-05} +{"train_loss": 0.015293143689632416, "grad_norm": 0.05674395337700844, "global_step": 17721, "epoch": 167, "lr": 7.485784187033624e-05} +{"train_loss": 0.015334287658333778, "grad_norm": 0.05502691864967346, "global_step": 17722, "epoch": 167, "lr": 7.48552702828373e-05} +{"train_loss": 0.014701060019433498, "grad_norm": 0.13396047055721283, "global_step": 17723, "epoch": 167, "lr": 7.485269860800774e-05} +{"train_loss": 0.015559348277747631, "grad_norm": 0.07205626368522644, "global_step": 17724, "epoch": 167, "lr": 7.485012684585662e-05} +{"train_loss": 0.014870360493659973, "grad_norm": 0.10265028476715088, "global_step": 17725, "epoch": 167, "lr": 7.484755499639296e-05} +{"train_loss": 0.015418985858559608, "grad_norm": 0.1303691864013672, "global_step": 17726, "epoch": 167, "lr": 7.484498305962579e-05} +{"train_loss": 0.013360452838242054, "grad_norm": 0.05144720897078514, "global_step": 17727, "epoch": 167, "lr": 7.484241103556416e-05} +{"train_loss": 0.014890937134623528, "grad_norm": 0.06732425093650818, "global_step": 17728, "epoch": 167, "lr": 7.483983892421709e-05} +{"train_loss": 0.016965527087450027, "grad_norm": 0.057853203266859055, "global_step": 17729, "epoch": 167, "lr": 7.483726672559364e-05} +{"train_loss": 0.015000253915786743, "grad_norm": 0.0397210568189621, "global_step": 17730, "epoch": 167, "lr": 7.483469443970283e-05} +{"train_loss": 0.014465885236859322, "grad_norm": 0.0856906846165657, "global_step": 17731, "epoch": 167, "lr": 7.48321220665537e-05} +{"train_loss": 0.014806954190135002, "grad_norm": 0.11765032261610031, "global_step": 17732, "epoch": 167, "lr": 7.48295496061553e-05} +{"train_loss": 0.015474837273359299, "grad_norm": 0.13049328327178955, "global_step": 17733, "epoch": 167, "lr": 7.482697705851666e-05} +{"train_loss": 0.013566478155553341, "grad_norm": 0.05213760957121849, "global_step": 17734, "epoch": 167, "lr": 7.482440442364682e-05} +{"train_loss": 0.01482604630291462, "grad_norm": 0.051141638308763504, "global_step": 17735, "epoch": 167, "lr": 7.482183170155482e-05} +{"train_loss": 0.01665717363357544, "grad_norm": 0.04855543375015259, "global_step": 17736, "epoch": 167, "lr": 7.481925889224968e-05} +{"train_loss": 0.017333602532744408, "grad_norm": 0.11923136562108994, "global_step": 17737, "epoch": 167, "lr": 7.481668599574049e-05} +{"train_loss": 0.01588987186551094, "grad_norm": 0.05640874058008194, "global_step": 17738, "epoch": 167, "lr": 7.481411301203622e-05} +{"train_loss": 0.01673915982246399, "grad_norm": 0.06003378704190254, "global_step": 17739, "epoch": 167, "lr": 7.481153994114598e-05} +{"train_loss": 0.01730302721261978, "grad_norm": 0.12049978971481323, "global_step": 17740, "epoch": 167, "lr": 7.480896678307878e-05} +{"train_loss": 0.018008068203926086, "grad_norm": 0.11687802523374557, "global_step": 17741, "epoch": 167, "lr": 7.480639353784363e-05} +{"train_loss": 0.015792062506079674, "grad_norm": 0.08630826324224472, "global_step": 17742, "epoch": 167, "lr": 7.480382020544963e-05} +{"train_loss": 0.015947308391332626, "grad_norm": 0.13554096221923828, "global_step": 17743, "epoch": 167, "lr": 7.480124678590577e-05} +{"train_loss": 0.01666146144270897, "grad_norm": 0.06451673060655594, "global_step": 17744, "epoch": 167, "lr": 7.479867327922114e-05} +{"train_loss": 0.016662105917930603, "grad_norm": 0.044591642916202545, "global_step": 17745, "epoch": 167, "lr": 7.479609968540474e-05} +{"train_loss": 0.0159144327044487, "grad_norm": 0.09579800069332123, "global_step": 17746, "epoch": 167, "lr": 7.479352600446562e-05} +{"train_loss": 0.017409497871994972, "grad_norm": 0.086044080555439, "global_step": 17747, "epoch": 167, "lr": 7.479095223641286e-05} +{"train_loss": 0.017002450302243233, "grad_norm": 0.07700822502374649, "global_step": 17748, "epoch": 167, "lr": 7.478837838125545e-05} +{"train_loss": 0.01819760352373123, "grad_norm": 0.11588239669799805, "global_step": 17749, "epoch": 167, "lr": 7.478580443900247e-05} +{"train_loss": 0.015389818698167801, "grad_norm": 0.11806012690067291, "global_step": 17750, "epoch": 167, "lr": 7.478323040966295e-05} +{"train_loss": 0.018405955284833908, "grad_norm": 0.06980954855680466, "global_step": 17751, "epoch": 167, "lr": 7.478065629324593e-05} +{"train_loss": 0.017560161650180817, "grad_norm": 0.06925726681947708, "global_step": 17752, "epoch": 167, "lr": 7.477808208976048e-05} +{"train_loss": 0.017627160996198654, "grad_norm": 0.06794685125350952, "global_step": 17753, "epoch": 167, "lr": 7.47755077992156e-05} +{"train_loss": 0.016119539737701416, "grad_norm": 0.06587459146976471, "global_step": 17754, "epoch": 167, "lr": 7.477293342162039e-05} +{"train_loss": 0.017868410795927048, "grad_norm": 0.10724014043807983, "global_step": 17755, "epoch": 167, "lr": 7.477035895698384e-05} +{"train_loss": 0.016231413930654526, "grad_norm": 0.05960933864116669, "global_step": 17756, "epoch": 167, "lr": 7.476778440531502e-05} +{"train_loss": 0.01736322045326233, "grad_norm": 0.06798441708087921, "global_step": 17757, "epoch": 167, "lr": 7.4765209766623e-05} +{"train_loss": 0.017962319776415825, "grad_norm": 0.05432489886879921, "global_step": 17758, "epoch": 167, "lr": 7.476263504091678e-05} +{"train_loss": 0.015047578141093254, "grad_norm": 0.08782994747161865, "global_step": 17759, "epoch": 167, "lr": 7.476006022820545e-05} +{"train_loss": 0.01766614243388176, "grad_norm": 0.12509432435035706, "global_step": 17760, "epoch": 167, "lr": 7.475748532849801e-05} +{"train_loss": 0.017606094479560852, "grad_norm": 0.09094678610563278, "global_step": 17761, "epoch": 167, "lr": 7.475491034180355e-05} +{"train_loss": 0.01701187714934349, "grad_norm": 0.08826960623264313, "global_step": 17762, "epoch": 167, "lr": 7.47523352681311e-05} +{"train_loss": 0.014435714110732079, "grad_norm": 0.07414978742599487, "global_step": 17763, "epoch": 167, "lr": 7.474976010748971e-05} +{"train_loss": 0.017165597528219223, "grad_norm": 0.06812794506549835, "global_step": 17764, "epoch": 167, "lr": 7.474718485988843e-05} +{"train_loss": 0.016849689185619354, "grad_norm": 0.08360555022954941, "global_step": 17765, "epoch": 167, "lr": 7.474460952533628e-05} +{"train_loss": 0.01764792948961258, "grad_norm": 0.10301326215267181, "global_step": 17766, "epoch": 167, "lr": 7.474203410384236e-05} +{"train_loss": 0.01731288805603981, "grad_norm": 0.10902177542448044, "global_step": 17767, "epoch": 167, "lr": 7.473945859541569e-05} +{"train_loss": 0.01517766248434782, "grad_norm": 0.06511856615543365, "global_step": 17768, "epoch": 167, "lr": 7.473688300006532e-05} +{"train_loss": 0.015718529000878334, "grad_norm": 0.07313834130764008, "global_step": 17769, "epoch": 167, "lr": 7.473430731780028e-05} +{"train_loss": 0.016275890171527863, "grad_norm": 0.062288131564855576, "global_step": 17770, "epoch": 167, "lr": 7.473173154862967e-05} +{"train_loss": 0.016508635133504868, "grad_norm": 0.0877312570810318, "global_step": 17771, "epoch": 167, "lr": 7.472915569256249e-05} +{"train_loss": 0.014659767039120197, "grad_norm": 0.08367768675088882, "global_step": 17772, "epoch": 167, "lr": 7.472657974960782e-05} +{"train_loss": 0.019644740968942642, "grad_norm": 0.13241638243198395, "global_step": 17773, "epoch": 167, "lr": 7.47240037197747e-05} +{"train_loss": 0.015757421031594276, "grad_norm": 0.1225508525967598, "global_step": 17774, "epoch": 167, "lr": 7.472142760307219e-05} +{"train_loss": 0.015705835074186325, "grad_norm": 0.07483657449483871, "global_step": 17775, "epoch": 167, "lr": 7.471885139950931e-05} +{"train_loss": 0.016066111624240875, "grad_norm": 0.056690558791160583, "global_step": 17776, "epoch": 167, "lr": 7.471627510909518e-05} +{"train_loss": 0.0177173912525177, "grad_norm": 0.06682711094617844, "global_step": 17777, "epoch": 167, "lr": 7.471369873183876e-05} +{"train_loss": 0.017779864370822906, "grad_norm": 0.06099327653646469, "global_step": 17778, "epoch": 167, "lr": 7.471112226774919e-05} +{"train_loss": 0.018536455929279327, "grad_norm": 0.054843202233314514, "global_step": 17779, "epoch": 167, "lr": 7.470854571683545e-05} +{"train_loss": 0.017698660492897034, "grad_norm": 0.10678239166736603, "global_step": 17780, "epoch": 167, "lr": 7.470596907910666e-05} +{"train_loss": 0.01627405732870102, "grad_norm": 0.15125377476215363, "global_step": 17781, "epoch": 167, "lr": 7.470339235457183e-05} +{"train_loss": 0.01692856289446354, "grad_norm": 0.07144197076559067, "global_step": 17782, "epoch": 167, "lr": 7.470081554324e-05} +{"train_loss": 0.016041196882724762, "grad_norm": 0.07376031577587128, "global_step": 17783, "epoch": 167, "lr": 7.469823864512026e-05} +{"train_loss": 0.018353041261434555, "grad_norm": 0.04865410551428795, "global_step": 17784, "epoch": 167, "lr": 7.469566166022165e-05} +{"train_loss": 0.014326350763440132, "grad_norm": 0.1134382113814354, "global_step": 17785, "epoch": 167, "lr": 7.469308458855321e-05} +{"train_loss": 0.019482586532831192, "grad_norm": 0.11630912125110626, "global_step": 17786, "epoch": 167, "lr": 7.469050743012403e-05} +{"train_loss": 0.016336161643266678, "grad_norm": 0.052084822207689285, "global_step": 17787, "epoch": 167, "lr": 7.468793018494315e-05} +{"train_loss": 0.01457154005765915, "grad_norm": 0.11477256566286087, "global_step": 17788, "epoch": 167, "lr": 7.468535285301961e-05} +{"train_loss": 0.014225952327251434, "grad_norm": 0.08821432292461395, "global_step": 17789, "epoch": 167, "lr": 7.468277543436246e-05} +{"train_loss": 0.01726159267127514, "grad_norm": 0.09333650767803192, "global_step": 17790, "epoch": 167, "lr": 7.46801979289808e-05} +{"train_loss": 0.01498099323362112, "grad_norm": 0.04437263309955597, "global_step": 17791, "epoch": 167, "lr": 7.467762033688363e-05} +{"train_loss": 0.0188882015645504, "grad_norm": 0.058597009629011154, "global_step": 17792, "epoch": 167, "lr": 7.467504265808006e-05} +{"train_loss": 0.017968924716114998, "grad_norm": 0.08434301614761353, "global_step": 17793, "epoch": 167, "lr": 7.467246489257911e-05} +{"train_loss": 0.01678878627717495, "grad_norm": 0.06845913082361221, "global_step": 17794, "epoch": 167, "lr": 7.466988704038983e-05} +{"train_loss": 0.014236350543797016, "grad_norm": 0.0571778379380703, "global_step": 17795, "epoch": 167, "lr": 7.466730910152131e-05} +{"train_loss": 0.01566522940993309, "grad_norm": 0.0767509937286377, "global_step": 17796, "epoch": 167, "lr": 7.466473107598259e-05} +{"train_loss": 0.014941344037652016, "grad_norm": 0.14282022416591644, "global_step": 17797, "epoch": 167, "lr": 7.466215296378275e-05} +{"train_loss": 0.013523263856768608, "grad_norm": 0.060764189809560776, "global_step": 17798, "epoch": 167, "lr": 7.465957476493082e-05} +{"train_loss": 0.015544656664133072, "grad_norm": 0.05417846888303757, "global_step": 17799, "epoch": 167, "lr": 7.465699647943586e-05} +{"train_loss": 0.018308935686945915, "grad_norm": 0.13674291968345642, "global_step": 17800, "epoch": 167, "lr": 7.465441810730695e-05} +{"train_loss": 0.01580849662423134, "grad_norm": 0.08332331478595734, "global_step": 17801, "epoch": 167, "lr": 7.46518396485531e-05} +{"train_loss": 0.01730332523584366, "grad_norm": 0.10009995102882385, "global_step": 17802, "epoch": 167, "lr": 7.464926110318346e-05} +{"train_loss": 0.013627320528030396, "grad_norm": 0.0678110271692276, "global_step": 17803, "epoch": 167, "lr": 7.4646682471207e-05} +{"train_loss": 0.015964563935995102, "grad_norm": 0.07744147628545761, "global_step": 17804, "epoch": 167, "lr": 7.464410375263283e-05} +{"train_loss": 0.017646227031946182, "grad_norm": 0.0822150781750679, "global_step": 17805, "epoch": 167, "lr": 7.464152494747001e-05} +{"train_loss": 0.01795867644250393, "grad_norm": 0.05760815367102623, "global_step": 17806, "epoch": 167, "lr": 7.463894605572757e-05} +{"train_loss": 0.0163768093488267, "grad_norm": 0.09133931994438171, "global_step": 17807, "epoch": 167, "lr": 7.463636707741458e-05, "val_loss": 0.019589990377426147} +{"train_loss": 0.015712041407823563, "grad_norm": 0.10054510086774826, "global_step": 17808, "epoch": 168, "lr": 7.463378801254012e-05} +{"train_loss": 0.01711483672261238, "grad_norm": 0.06827091425657272, "global_step": 17809, "epoch": 168, "lr": 7.463120886111324e-05} +{"train_loss": 0.016784923151135445, "grad_norm": 0.06447385251522064, "global_step": 17810, "epoch": 168, "lr": 7.462862962314301e-05} +{"train_loss": 0.016906458884477615, "grad_norm": 0.11529164016246796, "global_step": 17811, "epoch": 168, "lr": 7.462605029863848e-05} +{"train_loss": 0.016779223456978798, "grad_norm": 0.08949092775583267, "global_step": 17812, "epoch": 168, "lr": 7.462347088760872e-05} +{"train_loss": 0.0138692706823349, "grad_norm": 0.10322090983390808, "global_step": 17813, "epoch": 168, "lr": 7.462089139006281e-05} +{"train_loss": 0.014333045110106468, "grad_norm": 0.054799024015665054, "global_step": 17814, "epoch": 168, "lr": 7.461831180600977e-05} +{"train_loss": 0.017709074541926384, "grad_norm": 0.09345414489507675, "global_step": 17815, "epoch": 168, "lr": 7.461573213545871e-05} +{"train_loss": 0.016195781528949738, "grad_norm": 0.05472252890467644, "global_step": 17816, "epoch": 168, "lr": 7.461315237841864e-05} +{"train_loss": 0.019503828138113022, "grad_norm": 0.07572971284389496, "global_step": 17817, "epoch": 168, "lr": 7.461057253489869e-05} +{"train_loss": 0.015272406861186028, "grad_norm": 0.07221336662769318, "global_step": 17818, "epoch": 168, "lr": 7.460799260490787e-05} +{"train_loss": 0.016687167808413506, "grad_norm": 0.08278181403875351, "global_step": 17819, "epoch": 168, "lr": 7.460541258845527e-05} +{"train_loss": 0.015954749658703804, "grad_norm": 0.06621301919221878, "global_step": 17820, "epoch": 168, "lr": 7.460283248554997e-05} +{"train_loss": 0.01470958162099123, "grad_norm": 0.11818543076515198, "global_step": 17821, "epoch": 168, "lr": 7.460025229620099e-05} +{"train_loss": 0.016781799495220184, "grad_norm": 0.11372719705104828, "global_step": 17822, "epoch": 168, "lr": 7.459767202041743e-05} +{"train_loss": 0.015578980557620525, "grad_norm": 0.05311983823776245, "global_step": 17823, "epoch": 168, "lr": 7.459509165820835e-05} +{"train_loss": 0.01794549636542797, "grad_norm": 0.06444462388753891, "global_step": 17824, "epoch": 168, "lr": 7.459251120958281e-05} +{"train_loss": 0.014999193139374256, "grad_norm": 0.04513052850961685, "global_step": 17825, "epoch": 168, "lr": 7.458993067454989e-05} +{"train_loss": 0.015148479491472244, "grad_norm": 0.06802065670490265, "global_step": 17826, "epoch": 168, "lr": 7.458735005311864e-05} +{"train_loss": 0.013780752196907997, "grad_norm": 0.06578003615140915, "global_step": 17827, "epoch": 168, "lr": 7.458476934529814e-05} +{"train_loss": 0.018470164388418198, "grad_norm": 0.05177069827914238, "global_step": 17828, "epoch": 168, "lr": 7.458218855109744e-05} +{"train_loss": 0.015711728483438492, "grad_norm": 0.04482915997505188, "global_step": 17829, "epoch": 168, "lr": 7.457960767052563e-05} +{"train_loss": 0.016111906617879868, "grad_norm": 0.055683016777038574, "global_step": 17830, "epoch": 168, "lr": 7.457702670359176e-05} +{"train_loss": 0.017518620938062668, "grad_norm": 0.06933958828449249, "global_step": 17831, "epoch": 168, "lr": 7.457444565030493e-05} +{"train_loss": 0.016825031489133835, "grad_norm": 0.06084984168410301, "global_step": 17832, "epoch": 168, "lr": 7.457186451067416e-05} +{"train_loss": 0.01631942391395569, "grad_norm": 0.0594598762691021, "global_step": 17833, "epoch": 168, "lr": 7.456928328470854e-05} +{"train_loss": 0.015517272055149078, "grad_norm": 0.1164337620139122, "global_step": 17834, "epoch": 168, "lr": 7.456670197241717e-05} +{"train_loss": 0.016883939504623413, "grad_norm": 0.0576428584754467, "global_step": 17835, "epoch": 168, "lr": 7.456412057380907e-05} +{"train_loss": 0.017443101853132248, "grad_norm": 0.10329225659370422, "global_step": 17836, "epoch": 168, "lr": 7.456153908889334e-05} +{"train_loss": 0.016871610656380653, "grad_norm": 0.05281962454319, "global_step": 17837, "epoch": 168, "lr": 7.455895751767905e-05} +{"train_loss": 0.017024416476488113, "grad_norm": 0.0624818317592144, "global_step": 17838, "epoch": 168, "lr": 7.455637586017525e-05} +{"train_loss": 0.018092865124344826, "grad_norm": 0.06191340833902359, "global_step": 17839, "epoch": 168, "lr": 7.455379411639104e-05} +{"train_loss": 0.017203398048877716, "grad_norm": 0.054011035710573196, "global_step": 17840, "epoch": 168, "lr": 7.455121228633547e-05} +{"train_loss": 0.018151409924030304, "grad_norm": 0.09950528293848038, "global_step": 17841, "epoch": 168, "lr": 7.454863037001761e-05} +{"train_loss": 0.013974851928651333, "grad_norm": 0.07097631692886353, "global_step": 17842, "epoch": 168, "lr": 7.454604836744655e-05} +{"train_loss": 0.016898851841688156, "grad_norm": 0.04988330602645874, "global_step": 17843, "epoch": 168, "lr": 7.454346627863134e-05} +{"train_loss": 0.015449494123458862, "grad_norm": 0.06491855531930923, "global_step": 17844, "epoch": 168, "lr": 7.454088410358108e-05} +{"train_loss": 0.015855275094509125, "grad_norm": 0.09144392609596252, "global_step": 17845, "epoch": 168, "lr": 7.453830184230481e-05} +{"train_loss": 0.01758120208978653, "grad_norm": 0.06014002859592438, "global_step": 17846, "epoch": 168, "lr": 7.453571949481162e-05} +{"train_loss": 0.015616821125149727, "grad_norm": 0.13630014657974243, "global_step": 17847, "epoch": 168, "lr": 7.453313706111058e-05} +{"train_loss": 0.016157448291778564, "grad_norm": 0.06444909423589706, "global_step": 17848, "epoch": 168, "lr": 7.453055454121077e-05} +{"train_loss": 0.015382131561636925, "grad_norm": 0.11939076334238052, "global_step": 17849, "epoch": 168, "lr": 7.452797193512127e-05} +{"train_loss": 0.015177580527961254, "grad_norm": 0.06599822640419006, "global_step": 17850, "epoch": 168, "lr": 7.452538924285114e-05} +{"train_loss": 0.017949486151337624, "grad_norm": 0.13665032386779785, "global_step": 17851, "epoch": 168, "lr": 7.452280646440943e-05} +{"train_loss": 0.016279075294733047, "grad_norm": 0.12596158683300018, "global_step": 17852, "epoch": 168, "lr": 7.452022359980527e-05} +{"train_loss": 0.015545854344964027, "grad_norm": 0.0848514586687088, "global_step": 17853, "epoch": 168, "lr": 7.45176406490477e-05} +{"train_loss": 0.015256645157933235, "grad_norm": 0.09052671492099762, "global_step": 17854, "epoch": 168, "lr": 7.451505761214581e-05} +{"train_loss": 0.01588793843984604, "grad_norm": 0.07562758773565292, "global_step": 17855, "epoch": 168, "lr": 7.451247448910867e-05} +{"train_loss": 0.01750008389353752, "grad_norm": 0.04735115170478821, "global_step": 17856, "epoch": 168, "lr": 7.450989127994534e-05} +{"train_loss": 0.015344848856329918, "grad_norm": 0.07579077780246735, "global_step": 17857, "epoch": 168, "lr": 7.450730798466492e-05} +{"train_loss": 0.017818059772253036, "grad_norm": 0.07193679362535477, "global_step": 17858, "epoch": 168, "lr": 7.450472460327649e-05} +{"train_loss": 0.01437508687376976, "grad_norm": 0.08091617375612259, "global_step": 17859, "epoch": 168, "lr": 7.450214113578908e-05} +{"train_loss": 0.016226675361394882, "grad_norm": 0.07396508753299713, "global_step": 17860, "epoch": 168, "lr": 7.449955758221183e-05} +{"train_loss": 0.01574423536658287, "grad_norm": 0.058683376759290695, "global_step": 17861, "epoch": 168, "lr": 7.449697394255379e-05} +{"train_loss": 0.017133599147200584, "grad_norm": 0.05491071194410324, "global_step": 17862, "epoch": 168, "lr": 7.449439021682403e-05} +{"train_loss": 0.018122777342796326, "grad_norm": 0.08085533231496811, "global_step": 17863, "epoch": 168, "lr": 7.449180640503163e-05} +{"train_loss": 0.016103388741612434, "grad_norm": 0.06293300539255142, "global_step": 17864, "epoch": 168, "lr": 7.448922250718569e-05} +{"train_loss": 0.01865726336836815, "grad_norm": 0.05174696817994118, "global_step": 17865, "epoch": 168, "lr": 7.448663852329525e-05} +{"train_loss": 0.015226798132061958, "grad_norm": 0.06412272155284882, "global_step": 17866, "epoch": 168, "lr": 7.448405445336943e-05} +{"train_loss": 0.019970908761024475, "grad_norm": 0.13668939471244812, "global_step": 17867, "epoch": 168, "lr": 7.448147029741728e-05} +{"train_loss": 0.015533328987658024, "grad_norm": 0.06537610292434692, "global_step": 17868, "epoch": 168, "lr": 7.44788860554479e-05} +{"train_loss": 0.018977005034685135, "grad_norm": 0.22291648387908936, "global_step": 17869, "epoch": 168, "lr": 7.447630172747033e-05} +{"train_loss": 0.015983719378709793, "grad_norm": 0.06454876810312271, "global_step": 17870, "epoch": 168, "lr": 7.447371731349372e-05} +{"train_loss": 0.013895219191908836, "grad_norm": 0.05285225808620453, "global_step": 17871, "epoch": 168, "lr": 7.447113281352709e-05} +{"train_loss": 0.01754045858979225, "grad_norm": 0.08851472288370132, "global_step": 17872, "epoch": 168, "lr": 7.446854822757954e-05} +{"train_loss": 0.015343077480793, "grad_norm": 0.05583282932639122, "global_step": 17873, "epoch": 168, "lr": 7.446596355566017e-05} +{"train_loss": 0.017268497496843338, "grad_norm": 0.08736441284418106, "global_step": 17874, "epoch": 168, "lr": 7.446337879777802e-05} +{"train_loss": 0.01435344759374857, "grad_norm": 0.10488231480121613, "global_step": 17875, "epoch": 168, "lr": 7.446079395394221e-05} +{"train_loss": 0.01427392102777958, "grad_norm": 0.10374638438224792, "global_step": 17876, "epoch": 168, "lr": 7.44582090241618e-05} +{"train_loss": 0.018547819927334785, "grad_norm": 0.1044638380408287, "global_step": 17877, "epoch": 168, "lr": 7.445562400844589e-05} +{"train_loss": 0.017191458493471146, "grad_norm": 0.05587329715490341, "global_step": 17878, "epoch": 168, "lr": 7.445303890680356e-05} +{"train_loss": 0.01625152677297592, "grad_norm": 0.09680161625146866, "global_step": 17879, "epoch": 168, "lr": 7.445045371924386e-05} +{"train_loss": 0.01867086999118328, "grad_norm": 0.11339915543794632, "global_step": 17880, "epoch": 168, "lr": 7.444786844577592e-05} +{"train_loss": 0.013961631804704666, "grad_norm": 0.10766779631376266, "global_step": 17881, "epoch": 168, "lr": 7.444528308640879e-05} +{"train_loss": 0.016656847670674324, "grad_norm": 0.08613252639770508, "global_step": 17882, "epoch": 168, "lr": 7.444269764115157e-05} +{"train_loss": 0.022493332624435425, "grad_norm": 0.18969610333442688, "global_step": 17883, "epoch": 168, "lr": 7.444011211001334e-05} +{"train_loss": 0.017771504819393158, "grad_norm": 0.1445469707250595, "global_step": 17884, "epoch": 168, "lr": 7.44375264930032e-05} +{"train_loss": 0.01549205556511879, "grad_norm": 0.06372091174125671, "global_step": 17885, "epoch": 168, "lr": 7.44349407901302e-05} +{"train_loss": 0.013821804895997047, "grad_norm": 0.052609629929065704, "global_step": 17886, "epoch": 168, "lr": 7.443235500140344e-05} +{"train_loss": 0.013625474646687508, "grad_norm": 0.04905082285404205, "global_step": 17887, "epoch": 168, "lr": 7.442976912683204e-05} +{"train_loss": 0.015726368874311447, "grad_norm": 0.05792760103940964, "global_step": 17888, "epoch": 168, "lr": 7.442718316642502e-05} +{"train_loss": 0.016945408657193184, "grad_norm": 0.09380726516246796, "global_step": 17889, "epoch": 168, "lr": 7.442459712019151e-05} +{"train_loss": 0.015276370570063591, "grad_norm": 0.1351502239704132, "global_step": 17890, "epoch": 168, "lr": 7.442201098814061e-05} +{"train_loss": 0.019879436120390892, "grad_norm": 0.12957103550434113, "global_step": 17891, "epoch": 168, "lr": 7.441942477028136e-05} +{"train_loss": 0.01903066411614418, "grad_norm": 0.11260921508073807, "global_step": 17892, "epoch": 168, "lr": 7.441683846662289e-05} +{"train_loss": 0.016176974400877953, "grad_norm": 0.12157019972801208, "global_step": 17893, "epoch": 168, "lr": 7.441425207717425e-05} +{"train_loss": 0.013567547313869, "grad_norm": 0.10923068970441818, "global_step": 17894, "epoch": 168, "lr": 7.441166560194456e-05} +{"train_loss": 0.01805276796221733, "grad_norm": 0.05654657259583473, "global_step": 17895, "epoch": 168, "lr": 7.440907904094288e-05} +{"train_loss": 0.017348164692521095, "grad_norm": 0.0809839591383934, "global_step": 17896, "epoch": 168, "lr": 7.440649239417832e-05} +{"train_loss": 0.0178617462515831, "grad_norm": 0.10344274342060089, "global_step": 17897, "epoch": 168, "lr": 7.440390566165996e-05} +{"train_loss": 0.014867144636809826, "grad_norm": 0.06953412294387817, "global_step": 17898, "epoch": 168, "lr": 7.440131884339688e-05} +{"train_loss": 0.014012448489665985, "grad_norm": 0.0715479925274849, "global_step": 17899, "epoch": 168, "lr": 7.43987319393982e-05} +{"train_loss": 0.016433458775281906, "grad_norm": 0.06139540299773216, "global_step": 17900, "epoch": 168, "lr": 7.439614494967298e-05} +{"train_loss": 0.017379362136125565, "grad_norm": 0.07728417962789536, "global_step": 17901, "epoch": 168, "lr": 7.43935578742303e-05} +{"train_loss": 0.0192390289157629, "grad_norm": 0.1976296752691269, "global_step": 17902, "epoch": 168, "lr": 7.439097071307928e-05} +{"train_loss": 0.014098228886723518, "grad_norm": 0.06509636342525482, "global_step": 17903, "epoch": 168, "lr": 7.438838346622898e-05} +{"train_loss": 0.014689594507217407, "grad_norm": 0.0508853904902935, "global_step": 17904, "epoch": 168, "lr": 7.438579613368852e-05} +{"train_loss": 0.015092908404767513, "grad_norm": 0.13206811249256134, "global_step": 17905, "epoch": 168, "lr": 7.438320871546697e-05} +{"train_loss": 0.015642503276467323, "grad_norm": 0.07839494943618774, "global_step": 17906, "epoch": 168, "lr": 7.438062121157344e-05} +{"train_loss": 0.013867856003344059, "grad_norm": 0.07986024767160416, "global_step": 17907, "epoch": 168, "lr": 7.4378033622017e-05} +{"train_loss": 0.01370216254144907, "grad_norm": 0.07820650190114975, "global_step": 17908, "epoch": 168, "lr": 7.437544594680675e-05} +{"train_loss": 0.016072504222393036, "grad_norm": 0.06065057963132858, "global_step": 17909, "epoch": 168, "lr": 7.437285818595178e-05} +{"train_loss": 0.014285129494965076, "grad_norm": 0.05502074956893921, "global_step": 17910, "epoch": 168, "lr": 7.437027033946119e-05} +{"train_loss": 0.015839606523513794, "grad_norm": 0.08913208544254303, "global_step": 17911, "epoch": 168, "lr": 7.436768240734408e-05} +{"train_loss": 0.019085586071014404, "grad_norm": 0.10655654221773148, "global_step": 17912, "epoch": 168, "lr": 7.436509438960953e-05} +{"train_loss": 0.016350005410681916, "grad_norm": 0.06801232695579529, "global_step": 17913, "epoch": 168, "lr": 7.436250628626662e-05, "val_loss": 0.019198857247829437} +{"train_loss": 0.016455302014946938, "grad_norm": 0.08245311677455902, "global_step": 17914, "epoch": 169, "lr": 7.435991809732448e-05} +{"train_loss": 0.01480431854724884, "grad_norm": 0.09210636466741562, "global_step": 17915, "epoch": 169, "lr": 7.435732982279216e-05} +{"train_loss": 0.013776620849967003, "grad_norm": 0.07903244346380234, "global_step": 17916, "epoch": 169, "lr": 7.435474146267878e-05} +{"train_loss": 0.015407493337988853, "grad_norm": 0.06629600375890732, "global_step": 17917, "epoch": 169, "lr": 7.435215301699343e-05} +{"train_loss": 0.01795031502842903, "grad_norm": 0.05690672621130943, "global_step": 17918, "epoch": 169, "lr": 7.43495644857452e-05} +{"train_loss": 0.018279731273651123, "grad_norm": 0.11984718590974808, "global_step": 17919, "epoch": 169, "lr": 7.43469758689432e-05} +{"train_loss": 0.014634286053478718, "grad_norm": 0.09128722548484802, "global_step": 17920, "epoch": 169, "lr": 7.434438716659652e-05} +{"train_loss": 0.01719207689166069, "grad_norm": 0.07139415293931961, "global_step": 17921, "epoch": 169, "lr": 7.434179837871424e-05} +{"train_loss": 0.016324922442436218, "grad_norm": 0.07470247149467468, "global_step": 17922, "epoch": 169, "lr": 7.433920950530547e-05} +{"train_loss": 0.015660051256418228, "grad_norm": 0.05228177830576897, "global_step": 17923, "epoch": 169, "lr": 7.433662054637931e-05} +{"train_loss": 0.01529269851744175, "grad_norm": 0.09627723693847656, "global_step": 17924, "epoch": 169, "lr": 7.433403150194483e-05} +{"train_loss": 0.015245163813233376, "grad_norm": 0.077059306204319, "global_step": 17925, "epoch": 169, "lr": 7.433144237201115e-05} +{"train_loss": 0.015879429876804352, "grad_norm": 0.08404992520809174, "global_step": 17926, "epoch": 169, "lr": 7.432885315658738e-05} +{"train_loss": 0.01394236832857132, "grad_norm": 0.09075745940208435, "global_step": 17927, "epoch": 169, "lr": 7.432626385568259e-05} +{"train_loss": 0.017125245183706284, "grad_norm": 0.11898007243871689, "global_step": 17928, "epoch": 169, "lr": 7.43236744693059e-05} +{"train_loss": 0.01605529710650444, "grad_norm": 0.06736240535974503, "global_step": 17929, "epoch": 169, "lr": 7.432108499746639e-05} +{"train_loss": 0.014963638037443161, "grad_norm": 0.0998174175620079, "global_step": 17930, "epoch": 169, "lr": 7.431849544017314e-05} +{"train_loss": 0.015703266486525536, "grad_norm": 0.06463181972503662, "global_step": 17931, "epoch": 169, "lr": 7.43159057974353e-05} +{"train_loss": 0.018984679132699966, "grad_norm": 0.10346121340990067, "global_step": 17932, "epoch": 169, "lr": 7.431331606926193e-05} +{"train_loss": 0.013970059342682362, "grad_norm": 0.07841470092535019, "global_step": 17933, "epoch": 169, "lr": 7.431072625566215e-05} +{"train_loss": 0.016485042870044708, "grad_norm": 0.09159883111715317, "global_step": 17934, "epoch": 169, "lr": 7.430813635664504e-05} +{"train_loss": 0.01621612161397934, "grad_norm": 0.08949573338031769, "global_step": 17935, "epoch": 169, "lr": 7.430554637221971e-05} +{"train_loss": 0.01640729233622551, "grad_norm": 0.08512772619724274, "global_step": 17936, "epoch": 169, "lr": 7.430295630239527e-05} +{"train_loss": 0.015584897249937057, "grad_norm": 0.0907873660326004, "global_step": 17937, "epoch": 169, "lr": 7.430036614718082e-05} +{"train_loss": 0.015545221045613289, "grad_norm": 0.12271376699209213, "global_step": 17938, "epoch": 169, "lr": 7.429777590658543e-05} +{"train_loss": 0.015577539801597595, "grad_norm": 0.11154258996248245, "global_step": 17939, "epoch": 169, "lr": 7.429518558061821e-05} +{"train_loss": 0.016581803560256958, "grad_norm": 0.08789842575788498, "global_step": 17940, "epoch": 169, "lr": 7.429259516928831e-05} +{"train_loss": 0.017406269907951355, "grad_norm": 0.1301969289779663, "global_step": 17941, "epoch": 169, "lr": 7.429000467260477e-05} +{"train_loss": 0.01691325008869171, "grad_norm": 0.05695464089512825, "global_step": 17942, "epoch": 169, "lr": 7.428741409057672e-05} +{"train_loss": 0.016523540019989014, "grad_norm": 0.07685570418834686, "global_step": 17943, "epoch": 169, "lr": 7.428482342321328e-05} +{"train_loss": 0.0155725022777915, "grad_norm": 0.08134245127439499, "global_step": 17944, "epoch": 169, "lr": 7.42822326705235e-05} +{"train_loss": 0.015446520410478115, "grad_norm": 0.048549868166446686, "global_step": 17945, "epoch": 169, "lr": 7.427964183251654e-05} +{"train_loss": 0.015686918050050735, "grad_norm": 0.06060410290956497, "global_step": 17946, "epoch": 169, "lr": 7.427705090920146e-05} +{"train_loss": 0.015835681930184364, "grad_norm": 0.056924931704998016, "global_step": 17947, "epoch": 169, "lr": 7.42744599005874e-05} +{"train_loss": 0.01721024699509144, "grad_norm": 0.0491584837436676, "global_step": 17948, "epoch": 169, "lr": 7.427186880668344e-05} +{"train_loss": 0.01601785235106945, "grad_norm": 0.11317792534828186, "global_step": 17949, "epoch": 169, "lr": 7.426927762749867e-05} +{"train_loss": 0.015873992815613747, "grad_norm": 0.09262232482433319, "global_step": 17950, "epoch": 169, "lr": 7.426668636304223e-05} +{"train_loss": 0.014212924987077713, "grad_norm": 0.0558304525911808, "global_step": 17951, "epoch": 169, "lr": 7.42640950133232e-05} +{"train_loss": 0.017388828098773956, "grad_norm": 0.09641379117965698, "global_step": 17952, "epoch": 169, "lr": 7.42615035783507e-05} +{"train_loss": 0.015541637316346169, "grad_norm": 0.07701234519481659, "global_step": 17953, "epoch": 169, "lr": 7.425891205813382e-05} +{"train_loss": 0.018263380974531174, "grad_norm": 0.1593819111585617, "global_step": 17954, "epoch": 169, "lr": 7.425632045268167e-05} +{"train_loss": 0.015898767858743668, "grad_norm": 0.12409733235836029, "global_step": 17955, "epoch": 169, "lr": 7.425372876200338e-05} +{"train_loss": 0.014617444947361946, "grad_norm": 0.10073507577180862, "global_step": 17956, "epoch": 169, "lr": 7.4251136986108e-05} +{"train_loss": 0.016166958957910538, "grad_norm": 0.09213466942310333, "global_step": 17957, "epoch": 169, "lr": 7.424854512500469e-05} +{"train_loss": 0.014878279529511929, "grad_norm": 0.06609246879816055, "global_step": 17958, "epoch": 169, "lr": 7.424595317870253e-05} +{"train_loss": 0.016322441399097443, "grad_norm": 0.09604594111442566, "global_step": 17959, "epoch": 169, "lr": 7.424336114721066e-05} +{"train_loss": 0.017024116590619087, "grad_norm": 0.07656273990869522, "global_step": 17960, "epoch": 169, "lr": 7.424076903053814e-05} +{"train_loss": 0.016519799828529358, "grad_norm": 0.04906943812966347, "global_step": 17961, "epoch": 169, "lr": 7.423817682869409e-05} +{"train_loss": 0.013837823644280434, "grad_norm": 0.082915760576725, "global_step": 17962, "epoch": 169, "lr": 7.423558454168765e-05} +{"train_loss": 0.01643611118197441, "grad_norm": 0.08890574425458908, "global_step": 17963, "epoch": 169, "lr": 7.423299216952791e-05} +{"train_loss": 0.015334838069975376, "grad_norm": 0.07324935495853424, "global_step": 17964, "epoch": 169, "lr": 7.423039971222396e-05} +{"train_loss": 0.017597809433937073, "grad_norm": 0.07680170238018036, "global_step": 17965, "epoch": 169, "lr": 7.422780716978493e-05} +{"train_loss": 0.013626256957650185, "grad_norm": 0.0760163813829422, "global_step": 17966, "epoch": 169, "lr": 7.42252145422199e-05} +{"train_loss": 0.018206551671028137, "grad_norm": 0.07724042981863022, "global_step": 17967, "epoch": 169, "lr": 7.422262182953802e-05} +{"train_loss": 0.017352323979139328, "grad_norm": 0.07978086173534393, "global_step": 17968, "epoch": 169, "lr": 7.422002903174838e-05} +{"train_loss": 0.017070986330509186, "grad_norm": 0.07086734473705292, "global_step": 17969, "epoch": 169, "lr": 7.421743614886008e-05} +{"train_loss": 0.015118955634534359, "grad_norm": 0.060283780097961426, "global_step": 17970, "epoch": 169, "lr": 7.421484318088227e-05} +{"train_loss": 0.01462484709918499, "grad_norm": 0.05132295936346054, "global_step": 17971, "epoch": 169, "lr": 7.4212250127824e-05} +{"train_loss": 0.014230633154511452, "grad_norm": 0.08400724083185196, "global_step": 17972, "epoch": 169, "lr": 7.420965698969443e-05} +{"train_loss": 0.017704008147120476, "grad_norm": 0.09641249477863312, "global_step": 17973, "epoch": 169, "lr": 7.420706376650265e-05} +{"train_loss": 0.018809255212545395, "grad_norm": 0.05449500307440758, "global_step": 17974, "epoch": 169, "lr": 7.420447045825778e-05} +{"train_loss": 0.0183904767036438, "grad_norm": 0.05599457025527954, "global_step": 17975, "epoch": 169, "lr": 7.420187706496891e-05} +{"train_loss": 0.016281872987747192, "grad_norm": 0.06989698857069016, "global_step": 17976, "epoch": 169, "lr": 7.419928358664519e-05} +{"train_loss": 0.014914201572537422, "grad_norm": 0.04916849359869957, "global_step": 17977, "epoch": 169, "lr": 7.41966900232957e-05} +{"train_loss": 0.014833401888608932, "grad_norm": 0.13564351201057434, "global_step": 17978, "epoch": 169, "lr": 7.419409637492956e-05} +{"train_loss": 0.016262520104646683, "grad_norm": 0.08373338729143143, "global_step": 17979, "epoch": 169, "lr": 7.41915026415559e-05} +{"train_loss": 0.01654350757598877, "grad_norm": 0.052495915442705154, "global_step": 17980, "epoch": 169, "lr": 7.418890882318381e-05} +{"train_loss": 0.015239396132528782, "grad_norm": 0.07313378155231476, "global_step": 17981, "epoch": 169, "lr": 7.418631491982242e-05} +{"train_loss": 0.016067080199718475, "grad_norm": 0.14413821697235107, "global_step": 17982, "epoch": 169, "lr": 7.418372093148085e-05} +{"train_loss": 0.01870596408843994, "grad_norm": 0.07805848866701126, "global_step": 17983, "epoch": 169, "lr": 7.418112685816818e-05} +{"train_loss": 0.012609570287168026, "grad_norm": 0.05875926464796066, "global_step": 17984, "epoch": 169, "lr": 7.417853269989355e-05} +{"train_loss": 0.018352890387177467, "grad_norm": 0.05817331373691559, "global_step": 17985, "epoch": 169, "lr": 7.417593845666607e-05} +{"train_loss": 0.016876859590411186, "grad_norm": 0.14151740074157715, "global_step": 17986, "epoch": 169, "lr": 7.417334412849486e-05} +{"train_loss": 0.015719031915068626, "grad_norm": 0.0664214938879013, "global_step": 17987, "epoch": 169, "lr": 7.417074971538903e-05} +{"train_loss": 0.017451373860239983, "grad_norm": 0.07062214612960815, "global_step": 17988, "epoch": 169, "lr": 7.416815521735771e-05} +{"train_loss": 0.015594898723065853, "grad_norm": 0.07293522357940674, "global_step": 17989, "epoch": 169, "lr": 7.416556063441e-05} +{"train_loss": 0.017604298889636993, "grad_norm": 0.06794943660497665, "global_step": 17990, "epoch": 169, "lr": 7.4162965966555e-05} +{"train_loss": 0.01761074736714363, "grad_norm": 0.07727498561143875, "global_step": 17991, "epoch": 169, "lr": 7.416037121380187e-05} +{"train_loss": 0.017693523317575455, "grad_norm": 0.1544433981180191, "global_step": 17992, "epoch": 169, "lr": 7.415777637615968e-05} +{"train_loss": 0.017256587743759155, "grad_norm": 0.1018846407532692, "global_step": 17993, "epoch": 169, "lr": 7.415518145363758e-05} +{"train_loss": 0.019663769751787186, "grad_norm": 0.05787411332130432, "global_step": 17994, "epoch": 169, "lr": 7.415258644624468e-05} +{"train_loss": 0.014833501540124416, "grad_norm": 0.059684768319129944, "global_step": 17995, "epoch": 169, "lr": 7.414999135399008e-05} +{"train_loss": 0.01686403714120388, "grad_norm": 0.06188201531767845, "global_step": 17996, "epoch": 169, "lr": 7.414739617688293e-05} +{"train_loss": 0.017831895500421524, "grad_norm": 0.09557931125164032, "global_step": 17997, "epoch": 169, "lr": 7.414480091493232e-05} +{"train_loss": 0.018518691882491112, "grad_norm": 0.06528694182634354, "global_step": 17998, "epoch": 169, "lr": 7.41422055681474e-05} +{"train_loss": 0.015625834465026855, "grad_norm": 0.09606022387742996, "global_step": 17999, "epoch": 169, "lr": 7.413961013653726e-05} +{"train_loss": 0.016953522339463234, "grad_norm": 0.08232740312814713, "global_step": 18000, "epoch": 169, "lr": 7.413701462011101e-05} +{"train_loss": 0.017907723784446716, "grad_norm": 0.18745024502277374, "global_step": 18001, "epoch": 169, "lr": 7.41344190188778e-05} +{"train_loss": 0.015849586576223373, "grad_norm": 0.052866820245981216, "global_step": 18002, "epoch": 169, "lr": 7.413182333284674e-05} +{"train_loss": 0.017923779785633087, "grad_norm": 0.11728382110595703, "global_step": 18003, "epoch": 169, "lr": 7.412922756202694e-05} +{"train_loss": 0.015730813145637512, "grad_norm": 0.10769321769475937, "global_step": 18004, "epoch": 169, "lr": 7.412663170642754e-05} +{"train_loss": 0.017547262832522392, "grad_norm": 0.05614352226257324, "global_step": 18005, "epoch": 169, "lr": 7.412403576605765e-05} +{"train_loss": 0.01670612022280693, "grad_norm": 0.11540210992097855, "global_step": 18006, "epoch": 169, "lr": 7.412143974092637e-05} +{"train_loss": 0.016005560755729675, "grad_norm": 0.1131635382771492, "global_step": 18007, "epoch": 169, "lr": 7.411884363104284e-05} +{"train_loss": 0.016777832061052322, "grad_norm": 0.08754023909568787, "global_step": 18008, "epoch": 169, "lr": 7.411624743641621e-05} +{"train_loss": 0.01739739067852497, "grad_norm": 0.14226016402244568, "global_step": 18009, "epoch": 169, "lr": 7.411365115705556e-05} +{"train_loss": 0.019377462565898895, "grad_norm": 0.06622321903705597, "global_step": 18010, "epoch": 169, "lr": 7.411105479297003e-05} +{"train_loss": 0.015273925848305225, "grad_norm": 0.05395326390862465, "global_step": 18011, "epoch": 169, "lr": 7.410845834416873e-05} +{"train_loss": 0.017009584233164787, "grad_norm": 0.05123721808195114, "global_step": 18012, "epoch": 169, "lr": 7.41058618106608e-05} +{"train_loss": 0.01907620206475258, "grad_norm": 0.0660707876086235, "global_step": 18013, "epoch": 169, "lr": 7.410326519245534e-05} +{"train_loss": 0.015652507543563843, "grad_norm": 0.05461793392896652, "global_step": 18014, "epoch": 169, "lr": 7.41006684895615e-05} +{"train_loss": 0.015615249052643776, "grad_norm": 0.11025948822498322, "global_step": 18015, "epoch": 169, "lr": 7.409807170198839e-05} +{"train_loss": 0.018358664587140083, "grad_norm": 0.12566739320755005, "global_step": 18016, "epoch": 169, "lr": 7.409547482974514e-05} +{"train_loss": 0.01638297364115715, "grad_norm": 0.06956823170185089, "global_step": 18017, "epoch": 169, "lr": 7.409287787284088e-05} +{"train_loss": 0.014706836082041264, "grad_norm": 0.0740470439195633, "global_step": 18018, "epoch": 169, "lr": 7.40902808312847e-05} +{"train_loss": 0.01634695392348013, "grad_norm": 0.10596293956041336, "global_step": 18019, "epoch": 169, "lr": 7.408768370508576e-05, "val_loss": 0.019163595512509346} +{"train_loss": 0.01647481508553028, "grad_norm": 0.09177670627832413, "global_step": 18020, "epoch": 170, "lr": 7.408508649425319e-05} +{"train_loss": 0.015140727162361145, "grad_norm": 0.0885259360074997, "global_step": 18021, "epoch": 170, "lr": 7.40824891987961e-05} +{"train_loss": 0.01859760656952858, "grad_norm": 0.06556679308414459, "global_step": 18022, "epoch": 170, "lr": 7.407989181872359e-05} +{"train_loss": 0.017448510974645615, "grad_norm": 0.09554775804281235, "global_step": 18023, "epoch": 170, "lr": 7.407729435404483e-05} +{"train_loss": 0.017816372215747833, "grad_norm": 0.12444054335355759, "global_step": 18024, "epoch": 170, "lr": 7.407469680476892e-05} +{"train_loss": 0.016222313046455383, "grad_norm": 0.09255456924438477, "global_step": 18025, "epoch": 170, "lr": 7.4072099170905e-05} +{"train_loss": 0.019201163202524185, "grad_norm": 0.09686516970396042, "global_step": 18026, "epoch": 170, "lr": 7.406950145246218e-05} +{"train_loss": 0.013987393118441105, "grad_norm": 0.11591432988643646, "global_step": 18027, "epoch": 170, "lr": 7.406690364944963e-05} +{"train_loss": 0.01640523225069046, "grad_norm": 0.0998554676771164, "global_step": 18028, "epoch": 170, "lr": 7.406430576187642e-05} +{"train_loss": 0.014840533025562763, "grad_norm": 0.07300571352243423, "global_step": 18029, "epoch": 170, "lr": 7.406170778975172e-05} +{"train_loss": 0.015322688035666943, "grad_norm": 0.06529301404953003, "global_step": 18030, "epoch": 170, "lr": 7.405910973308463e-05} +{"train_loss": 0.015263786539435387, "grad_norm": 0.05968625843524933, "global_step": 18031, "epoch": 170, "lr": 7.40565115918843e-05} +{"train_loss": 0.015816234052181244, "grad_norm": 0.05182256922125816, "global_step": 18032, "epoch": 170, "lr": 7.405391336615985e-05} +{"train_loss": 0.01582001894712448, "grad_norm": 0.08066345751285553, "global_step": 18033, "epoch": 170, "lr": 7.405131505592041e-05} +{"train_loss": 0.01482902280986309, "grad_norm": 0.15456031262874603, "global_step": 18034, "epoch": 170, "lr": 7.404871666117511e-05} +{"train_loss": 0.016844147816300392, "grad_norm": 0.06606779247522354, "global_step": 18035, "epoch": 170, "lr": 7.404611818193307e-05} +{"train_loss": 0.017374690622091293, "grad_norm": 0.15444685518741608, "global_step": 18036, "epoch": 170, "lr": 7.404351961820343e-05} +{"train_loss": 0.016534045338630676, "grad_norm": 0.0838298574090004, "global_step": 18037, "epoch": 170, "lr": 7.404092096999534e-05} +{"train_loss": 0.013273550197482109, "grad_norm": 0.04817153885960579, "global_step": 18038, "epoch": 170, "lr": 7.403832223731787e-05} +{"train_loss": 0.015633560717105865, "grad_norm": 0.15683412551879883, "global_step": 18039, "epoch": 170, "lr": 7.403572342018021e-05} +{"train_loss": 0.015915967524051666, "grad_norm": 0.10313797742128372, "global_step": 18040, "epoch": 170, "lr": 7.403312451859148e-05} +{"train_loss": 0.016121359542012215, "grad_norm": 0.15508392453193665, "global_step": 18041, "epoch": 170, "lr": 7.403052553256078e-05} +{"train_loss": 0.017212266102433205, "grad_norm": 0.05376740172505379, "global_step": 18042, "epoch": 170, "lr": 7.402792646209728e-05} +{"train_loss": 0.01650002971291542, "grad_norm": 0.0860297754406929, "global_step": 18043, "epoch": 170, "lr": 7.402532730721009e-05} +{"train_loss": 0.016566960141062737, "grad_norm": 0.06356900185346603, "global_step": 18044, "epoch": 170, "lr": 7.402272806790835e-05} +{"train_loss": 0.014629168435931206, "grad_norm": 0.09148939698934555, "global_step": 18045, "epoch": 170, "lr": 7.402012874420119e-05} +{"train_loss": 0.015097909606993198, "grad_norm": 0.055156078189611435, "global_step": 18046, "epoch": 170, "lr": 7.401752933609775e-05} +{"train_loss": 0.015063820406794548, "grad_norm": 0.1434897929430008, "global_step": 18047, "epoch": 170, "lr": 7.401492984360715e-05} +{"train_loss": 0.018684377893805504, "grad_norm": 0.08750167489051819, "global_step": 18048, "epoch": 170, "lr": 7.401233026673853e-05} +{"train_loss": 0.014622745104134083, "grad_norm": 0.16663043200969696, "global_step": 18049, "epoch": 170, "lr": 7.400973060550103e-05} +{"train_loss": 0.015264056622982025, "grad_norm": 0.07737243175506592, "global_step": 18050, "epoch": 170, "lr": 7.400713085990378e-05} +{"train_loss": 0.015366693027317524, "grad_norm": 0.044721927493810654, "global_step": 18051, "epoch": 170, "lr": 7.40045310299559e-05} +{"train_loss": 0.014452429488301277, "grad_norm": 0.08581320941448212, "global_step": 18052, "epoch": 170, "lr": 7.400193111566654e-05} +{"train_loss": 0.013702137395739555, "grad_norm": 0.10267221182584763, "global_step": 18053, "epoch": 170, "lr": 7.399933111704483e-05} +{"train_loss": 0.017602764070034027, "grad_norm": 0.11517535150051117, "global_step": 18054, "epoch": 170, "lr": 7.399673103409991e-05} +{"train_loss": 0.015486164018511772, "grad_norm": 0.10423150658607483, "global_step": 18055, "epoch": 170, "lr": 7.399413086684093e-05} +{"train_loss": 0.015884697437286377, "grad_norm": 0.07035855948925018, "global_step": 18056, "epoch": 170, "lr": 7.3991530615277e-05} +{"train_loss": 0.01596120372414589, "grad_norm": 0.046387311071157455, "global_step": 18057, "epoch": 170, "lr": 7.398893027941724e-05} +{"train_loss": 0.014996528625488281, "grad_norm": 0.11387888342142105, "global_step": 18058, "epoch": 170, "lr": 7.398632985927084e-05} +{"train_loss": 0.015965573489665985, "grad_norm": 0.06137767806649208, "global_step": 18059, "epoch": 170, "lr": 7.39837293548469e-05} +{"train_loss": 0.014897597953677177, "grad_norm": 0.10959691554307938, "global_step": 18060, "epoch": 170, "lr": 7.398112876615457e-05} +{"train_loss": 0.015583906322717667, "grad_norm": 0.1254223883152008, "global_step": 18061, "epoch": 170, "lr": 7.397852809320299e-05} +{"train_loss": 0.013623912818729877, "grad_norm": 0.08136002719402313, "global_step": 18062, "epoch": 170, "lr": 7.397592733600128e-05} +{"train_loss": 0.016566261649131775, "grad_norm": 0.05089816823601723, "global_step": 18063, "epoch": 170, "lr": 7.397332649455857e-05} +{"train_loss": 0.016705740243196487, "grad_norm": 0.06409107148647308, "global_step": 18064, "epoch": 170, "lr": 7.397072556888405e-05} +{"train_loss": 0.017833031713962555, "grad_norm": 0.05343252420425415, "global_step": 18065, "epoch": 170, "lr": 7.396812455898679e-05} +{"train_loss": 0.015318496152758598, "grad_norm": 0.07292269915342331, "global_step": 18066, "epoch": 170, "lr": 7.396552346487598e-05} +{"train_loss": 0.01740087941288948, "grad_norm": 0.07911333441734314, "global_step": 18067, "epoch": 170, "lr": 7.396292228656074e-05} +{"train_loss": 0.015436173416674137, "grad_norm": 0.05280006676912308, "global_step": 18068, "epoch": 170, "lr": 7.396032102405023e-05} +{"train_loss": 0.013191563077270985, "grad_norm": 0.05580463632941246, "global_step": 18069, "epoch": 170, "lr": 7.395771967735356e-05} +{"train_loss": 0.01674247533082962, "grad_norm": 0.04438292607665062, "global_step": 18070, "epoch": 170, "lr": 7.395511824647987e-05} +{"train_loss": 0.017111120745539665, "grad_norm": 0.13861078023910522, "global_step": 18071, "epoch": 170, "lr": 7.395251673143833e-05} +{"train_loss": 0.01764574833214283, "grad_norm": 0.0752171128988266, "global_step": 18072, "epoch": 170, "lr": 7.394991513223806e-05} +{"train_loss": 0.018994368612766266, "grad_norm": 0.08097389340400696, "global_step": 18073, "epoch": 170, "lr": 7.39473134488882e-05} +{"train_loss": 0.01711384207010269, "grad_norm": 0.06505036354064941, "global_step": 18074, "epoch": 170, "lr": 7.394471168139789e-05} +{"train_loss": 0.02009328454732895, "grad_norm": 0.09791193902492523, "global_step": 18075, "epoch": 170, "lr": 7.394210982977627e-05} +{"train_loss": 0.013410953804850578, "grad_norm": 0.06093186140060425, "global_step": 18076, "epoch": 170, "lr": 7.393950789403252e-05} +{"train_loss": 0.014380937442183495, "grad_norm": 0.09871974587440491, "global_step": 18077, "epoch": 170, "lr": 7.393690587417573e-05} +{"train_loss": 0.017443621531128883, "grad_norm": 0.0625331848859787, "global_step": 18078, "epoch": 170, "lr": 7.393430377021505e-05} +{"train_loss": 0.01674574986100197, "grad_norm": 0.09768127650022507, "global_step": 18079, "epoch": 170, "lr": 7.393170158215966e-05} +{"train_loss": 0.015294915065169334, "grad_norm": 0.05711659416556358, "global_step": 18080, "epoch": 170, "lr": 7.392909931001865e-05} +{"train_loss": 0.016524123027920723, "grad_norm": 0.0699821263551712, "global_step": 18081, "epoch": 170, "lr": 7.392649695380123e-05} +{"train_loss": 0.01951880007982254, "grad_norm": 0.0863136574625969, "global_step": 18082, "epoch": 170, "lr": 7.392389451351647e-05} +{"train_loss": 0.015031160786747932, "grad_norm": 0.10385800898075104, "global_step": 18083, "epoch": 170, "lr": 7.392129198917356e-05} +{"train_loss": 0.016114506870508194, "grad_norm": 0.16094499826431274, "global_step": 18084, "epoch": 170, "lr": 7.391868938078163e-05} +{"train_loss": 0.0177223552018404, "grad_norm": 0.09240596741437912, "global_step": 18085, "epoch": 170, "lr": 7.391608668834983e-05} +{"train_loss": 0.017817210406064987, "grad_norm": 0.07204978168010712, "global_step": 18086, "epoch": 170, "lr": 7.391348391188731e-05} +{"train_loss": 0.01531421858817339, "grad_norm": 0.06612453609704971, "global_step": 18087, "epoch": 170, "lr": 7.39108810514032e-05} +{"train_loss": 0.018061917275190353, "grad_norm": 0.09965203702449799, "global_step": 18088, "epoch": 170, "lr": 7.390827810690666e-05} +{"train_loss": 0.016451280564069748, "grad_norm": 0.1076091080904007, "global_step": 18089, "epoch": 170, "lr": 7.390567507840681e-05} +{"train_loss": 0.017912521958351135, "grad_norm": 0.053768858313560486, "global_step": 18090, "epoch": 170, "lr": 7.390307196591284e-05} +{"train_loss": 0.01693311706185341, "grad_norm": 0.1301172524690628, "global_step": 18091, "epoch": 170, "lr": 7.390046876943387e-05} +{"train_loss": 0.015815120190382004, "grad_norm": 0.10574227571487427, "global_step": 18092, "epoch": 170, "lr": 7.389786548897902e-05} +{"train_loss": 0.015890231356024742, "grad_norm": 0.08020978420972824, "global_step": 18093, "epoch": 170, "lr": 7.389526212455748e-05} +{"train_loss": 0.019342508167028427, "grad_norm": 0.08659608662128448, "global_step": 18094, "epoch": 170, "lr": 7.389265867617835e-05} +{"train_loss": 0.015957091003656387, "grad_norm": 0.053458891808986664, "global_step": 18095, "epoch": 170, "lr": 7.389005514385086e-05} +{"train_loss": 0.015606052242219448, "grad_norm": 0.06281135231256485, "global_step": 18096, "epoch": 170, "lr": 7.388745152758407e-05} +{"train_loss": 0.013506527058780193, "grad_norm": 0.05973358824849129, "global_step": 18097, "epoch": 170, "lr": 7.388484782738718e-05} +{"train_loss": 0.016232280060648918, "grad_norm": 0.07062845677137375, "global_step": 18098, "epoch": 170, "lr": 7.388224404326931e-05} +{"train_loss": 0.0166314784437418, "grad_norm": 0.12396503239870071, "global_step": 18099, "epoch": 170, "lr": 7.387964017523964e-05} +{"train_loss": 0.014799494296312332, "grad_norm": 0.07771538943052292, "global_step": 18100, "epoch": 170, "lr": 7.387703622330728e-05} +{"train_loss": 0.018607815727591515, "grad_norm": 0.14847618341445923, "global_step": 18101, "epoch": 170, "lr": 7.387443218748142e-05} +{"train_loss": 0.016403548419475555, "grad_norm": 0.19155946373939514, "global_step": 18102, "epoch": 170, "lr": 7.387182806777117e-05} +{"train_loss": 0.01661452278494835, "grad_norm": 0.07517998665571213, "global_step": 18103, "epoch": 170, "lr": 7.38692238641857e-05} +{"train_loss": 0.017014995217323303, "grad_norm": 0.15514670312404633, "global_step": 18104, "epoch": 170, "lr": 7.386661957673414e-05} +{"train_loss": 0.014895637519657612, "grad_norm": 0.06231395900249481, "global_step": 18105, "epoch": 170, "lr": 7.38640152054257e-05} +{"train_loss": 0.01648823916912079, "grad_norm": 0.05272648483514786, "global_step": 18106, "epoch": 170, "lr": 7.386141075026944e-05} +{"train_loss": 0.016553469002246857, "grad_norm": 0.066008560359478, "global_step": 18107, "epoch": 170, "lr": 7.38588062112746e-05} +{"train_loss": 0.015308253467082977, "grad_norm": 0.09516116976737976, "global_step": 18108, "epoch": 170, "lr": 7.385620158845028e-05} +{"train_loss": 0.018637236207723618, "grad_norm": 0.12383227050304413, "global_step": 18109, "epoch": 170, "lr": 7.385359688180563e-05} +{"train_loss": 0.015293150208890438, "grad_norm": 0.07221118360757828, "global_step": 18110, "epoch": 170, "lr": 7.385099209134982e-05} +{"train_loss": 0.016475459560751915, "grad_norm": 0.07392049580812454, "global_step": 18111, "epoch": 170, "lr": 7.384838721709201e-05} +{"train_loss": 0.018372833728790283, "grad_norm": 0.08913876116275787, "global_step": 18112, "epoch": 170, "lr": 7.384578225904132e-05} +{"train_loss": 0.018630826845765114, "grad_norm": 0.09463031589984894, "global_step": 18113, "epoch": 170, "lr": 7.384317721720694e-05} +{"train_loss": 0.014218080788850784, "grad_norm": 0.10594061762094498, "global_step": 18114, "epoch": 170, "lr": 7.3840572091598e-05} +{"train_loss": 0.018442224711179733, "grad_norm": 0.07969643175601959, "global_step": 18115, "epoch": 170, "lr": 7.383796688222364e-05} +{"train_loss": 0.018400415778160095, "grad_norm": 0.1009916439652443, "global_step": 18116, "epoch": 170, "lr": 7.383536158909303e-05} +{"train_loss": 0.017261460423469543, "grad_norm": 0.06620003283023834, "global_step": 18117, "epoch": 170, "lr": 7.383275621221536e-05} +{"train_loss": 0.01556440070271492, "grad_norm": 0.06516175717115402, "global_step": 18118, "epoch": 170, "lr": 7.383015075159971e-05} +{"train_loss": 0.015527637675404549, "grad_norm": 0.08446647226810455, "global_step": 18119, "epoch": 170, "lr": 7.38275452072553e-05} +{"train_loss": 0.015127480030059814, "grad_norm": 0.098509781062603, "global_step": 18120, "epoch": 170, "lr": 7.382493957919125e-05} +{"train_loss": 0.018341204151511192, "grad_norm": 0.07093361020088196, "global_step": 18121, "epoch": 170, "lr": 7.382233386741673e-05} +{"train_loss": 0.018496308475732803, "grad_norm": 0.13900181651115417, "global_step": 18122, "epoch": 170, "lr": 7.381972807194088e-05} +{"train_loss": 0.017457852140069008, "grad_norm": 0.12768420577049255, "global_step": 18123, "epoch": 170, "lr": 7.381712219277286e-05} +{"train_loss": 0.016388269141316414, "grad_norm": 0.09430566430091858, "global_step": 18124, "epoch": 170, "lr": 7.381451622992184e-05} +{"train_loss": 0.01632686763263817, "grad_norm": 0.08879798650741577, "global_step": 18125, "epoch": 170, "lr": 7.381191018339696e-05, "train_batch_mse_error": 0.014818989671766758, "val_loss": 0.019168905913829803} +{"train_loss": 0.017042864114046097, "grad_norm": 0.06874406337738037, "global_step": 18126, "epoch": 171, "lr": 7.380930405320739e-05} +{"train_loss": 0.019444188103079796, "grad_norm": 0.1796048879623413, "global_step": 18127, "epoch": 171, "lr": 7.380669783936227e-05} +{"train_loss": 0.015178989619016647, "grad_norm": 0.15076152980327606, "global_step": 18128, "epoch": 171, "lr": 7.380409154187076e-05} +{"train_loss": 0.015903467312455177, "grad_norm": 0.05788008123636246, "global_step": 18129, "epoch": 171, "lr": 7.380148516074206e-05} +{"train_loss": 0.014561932533979416, "grad_norm": 0.08956040441989899, "global_step": 18130, "epoch": 171, "lr": 7.379887869598528e-05} +{"train_loss": 0.014696464873850346, "grad_norm": 0.050078704953193665, "global_step": 18131, "epoch": 171, "lr": 7.379627214760957e-05} +{"train_loss": 0.015116719529032707, "grad_norm": 0.06838522851467133, "global_step": 18132, "epoch": 171, "lr": 7.379366551562413e-05} +{"train_loss": 0.01576206460595131, "grad_norm": 0.05396084487438202, "global_step": 18133, "epoch": 171, "lr": 7.379105880003807e-05} +{"train_loss": 0.017809243872761726, "grad_norm": 0.14846831560134888, "global_step": 18134, "epoch": 171, "lr": 7.37884520008606e-05} +{"train_loss": 0.015221862122416496, "grad_norm": 0.05259007215499878, "global_step": 18135, "epoch": 171, "lr": 7.378584511810084e-05} +{"train_loss": 0.014991803094744682, "grad_norm": 0.04353373870253563, "global_step": 18136, "epoch": 171, "lr": 7.378323815176797e-05} +{"train_loss": 0.01757025718688965, "grad_norm": 0.0931229442358017, "global_step": 18137, "epoch": 171, "lr": 7.378063110187115e-05} +{"train_loss": 0.013669267296791077, "grad_norm": 0.0782000944018364, "global_step": 18138, "epoch": 171, "lr": 7.377802396841952e-05} +{"train_loss": 0.018101535737514496, "grad_norm": 0.08309734612703323, "global_step": 18139, "epoch": 171, "lr": 7.377541675142226e-05} +{"train_loss": 0.015706539154052734, "grad_norm": 0.08598383516073227, "global_step": 18140, "epoch": 171, "lr": 7.377280945088853e-05} +{"train_loss": 0.016643721610307693, "grad_norm": 0.060359444469213486, "global_step": 18141, "epoch": 171, "lr": 7.377020206682748e-05} +{"train_loss": 0.015587538480758667, "grad_norm": 0.0614611841738224, "global_step": 18142, "epoch": 171, "lr": 7.37675945992483e-05} +{"train_loss": 0.017878375947475433, "grad_norm": 0.07646296918392181, "global_step": 18143, "epoch": 171, "lr": 7.37649870481601e-05} +{"train_loss": 0.015323599800467491, "grad_norm": 0.0564577579498291, "global_step": 18144, "epoch": 171, "lr": 7.37623794135721e-05} +{"train_loss": 0.01761952042579651, "grad_norm": 0.09451878070831299, "global_step": 18145, "epoch": 171, "lr": 7.375977169549339e-05} +{"train_loss": 0.014495128765702248, "grad_norm": 0.0914553627371788, "global_step": 18146, "epoch": 171, "lr": 7.375716389393321e-05} +{"train_loss": 0.018661564216017723, "grad_norm": 0.06966786831617355, "global_step": 18147, "epoch": 171, "lr": 7.375455600890066e-05} +{"train_loss": 0.017474999651312828, "grad_norm": 0.10835977643728256, "global_step": 18148, "epoch": 171, "lr": 7.375194804040496e-05} +{"train_loss": 0.015074945986270905, "grad_norm": 0.08507249504327774, "global_step": 18149, "epoch": 171, "lr": 7.374933998845525e-05} +{"train_loss": 0.018120361492037773, "grad_norm": 0.058638401329517365, "global_step": 18150, "epoch": 171, "lr": 7.374673185306065e-05} +{"train_loss": 0.015126729384064674, "grad_norm": 0.059058014303445816, "global_step": 18151, "epoch": 171, "lr": 7.37441236342304e-05} +{"train_loss": 0.014209989458322525, "grad_norm": 0.0732535645365715, "global_step": 18152, "epoch": 171, "lr": 7.37415153319736e-05} +{"train_loss": 0.017958292737603188, "grad_norm": 0.08446234464645386, "global_step": 18153, "epoch": 171, "lr": 7.373890694629945e-05} +{"train_loss": 0.014626292511820793, "grad_norm": 0.053322963416576385, "global_step": 18154, "epoch": 171, "lr": 7.373629847721709e-05} +{"train_loss": 0.01859404146671295, "grad_norm": 0.0788792073726654, "global_step": 18155, "epoch": 171, "lr": 7.373368992473572e-05} +{"train_loss": 0.016612855717539787, "grad_norm": 0.15805283188819885, "global_step": 18156, "epoch": 171, "lr": 7.373108128886448e-05} +{"train_loss": 0.016710057854652405, "grad_norm": 0.17966531217098236, "global_step": 18157, "epoch": 171, "lr": 7.372847256961253e-05} +{"train_loss": 0.017134130001068115, "grad_norm": 0.07032974809408188, "global_step": 18158, "epoch": 171, "lr": 7.372586376698906e-05} +{"train_loss": 0.016502834856510162, "grad_norm": 0.06441202014684677, "global_step": 18159, "epoch": 171, "lr": 7.372325488100321e-05} +{"train_loss": 0.01702398806810379, "grad_norm": 0.1288236528635025, "global_step": 18160, "epoch": 171, "lr": 7.372064591166417e-05} +{"train_loss": 0.014989939518272877, "grad_norm": 0.14101316034793854, "global_step": 18161, "epoch": 171, "lr": 7.371803685898109e-05} +{"train_loss": 0.015142243355512619, "grad_norm": 0.09194817394018173, "global_step": 18162, "epoch": 171, "lr": 7.371542772296314e-05} +{"train_loss": 0.017043665051460266, "grad_norm": 0.09678780287504196, "global_step": 18163, "epoch": 171, "lr": 7.37128185036195e-05} +{"train_loss": 0.01563907228410244, "grad_norm": 0.23851512372493744, "global_step": 18164, "epoch": 171, "lr": 7.371020920095933e-05} +{"train_loss": 0.017540860921144485, "grad_norm": 0.19174616038799286, "global_step": 18165, "epoch": 171, "lr": 7.370759981499177e-05} +{"train_loss": 0.01739170216023922, "grad_norm": 0.11131598800420761, "global_step": 18166, "epoch": 171, "lr": 7.370499034572604e-05} +{"train_loss": 0.014838466420769691, "grad_norm": 0.07055386900901794, "global_step": 18167, "epoch": 171, "lr": 7.37023807931713e-05} +{"train_loss": 0.018091097474098206, "grad_norm": 0.12974625825881958, "global_step": 18168, "epoch": 171, "lr": 7.369977115733666e-05} +{"train_loss": 0.01725040003657341, "grad_norm": 0.057955533266067505, "global_step": 18169, "epoch": 171, "lr": 7.369716143823135e-05} +{"train_loss": 0.013571953400969505, "grad_norm": 0.07882249355316162, "global_step": 18170, "epoch": 171, "lr": 7.369455163586453e-05} +{"train_loss": 0.014808760955929756, "grad_norm": 0.08764121681451797, "global_step": 18171, "epoch": 171, "lr": 7.369194175024536e-05} +{"train_loss": 0.01741556078195572, "grad_norm": 0.109976626932621, "global_step": 18172, "epoch": 171, "lr": 7.368933178138299e-05} +{"train_loss": 0.018675263971090317, "grad_norm": 0.1455746293067932, "global_step": 18173, "epoch": 171, "lr": 7.368672172928662e-05} +{"train_loss": 0.02048051357269287, "grad_norm": 0.06517666578292847, "global_step": 18174, "epoch": 171, "lr": 7.368411159396541e-05} +{"train_loss": 0.015458822250366211, "grad_norm": 0.18663933873176575, "global_step": 18175, "epoch": 171, "lr": 7.368150137542853e-05} +{"train_loss": 0.017266057431697845, "grad_norm": 0.1368301510810852, "global_step": 18176, "epoch": 171, "lr": 7.367889107368516e-05} +{"train_loss": 0.01818317547440529, "grad_norm": 0.13643518090248108, "global_step": 18177, "epoch": 171, "lr": 7.367628068874445e-05} +{"train_loss": 0.015886690467596054, "grad_norm": 0.05068795755505562, "global_step": 18178, "epoch": 171, "lr": 7.36736702206156e-05} +{"train_loss": 0.015502415597438812, "grad_norm": 0.05932421237230301, "global_step": 18179, "epoch": 171, "lr": 7.367105966930778e-05} +{"train_loss": 0.01904848776757717, "grad_norm": 0.10885873436927795, "global_step": 18180, "epoch": 171, "lr": 7.366844903483014e-05} +{"train_loss": 0.01360192522406578, "grad_norm": 0.07078579068183899, "global_step": 18181, "epoch": 171, "lr": 7.366583831719185e-05} +{"train_loss": 0.014686685055494308, "grad_norm": 0.0658971294760704, "global_step": 18182, "epoch": 171, "lr": 7.366322751640211e-05} +{"train_loss": 0.015842288732528687, "grad_norm": 0.08301603049039841, "global_step": 18183, "epoch": 171, "lr": 7.366061663247008e-05} +{"train_loss": 0.01607201062142849, "grad_norm": 0.06122475489974022, "global_step": 18184, "epoch": 171, "lr": 7.365800566540493e-05} +{"train_loss": 0.014747634530067444, "grad_norm": 0.0615115612745285, "global_step": 18185, "epoch": 171, "lr": 7.365539461521583e-05} +{"train_loss": 0.01605318859219551, "grad_norm": 0.0571620836853981, "global_step": 18186, "epoch": 171, "lr": 7.365278348191196e-05} +{"train_loss": 0.017136789858341217, "grad_norm": 0.1035364493727684, "global_step": 18187, "epoch": 171, "lr": 7.365017226550251e-05} +{"train_loss": 0.016093630343675613, "grad_norm": 0.16769985854625702, "global_step": 18188, "epoch": 171, "lr": 7.364756096599663e-05} +{"train_loss": 0.016013408079743385, "grad_norm": 0.07665770500898361, "global_step": 18189, "epoch": 171, "lr": 7.36449495834035e-05} +{"train_loss": 0.01615295745432377, "grad_norm": 0.05875924229621887, "global_step": 18190, "epoch": 171, "lr": 7.364233811773231e-05} +{"train_loss": 0.01657710410654545, "grad_norm": 0.05922727659344673, "global_step": 18191, "epoch": 171, "lr": 7.363972656899222e-05} +{"train_loss": 0.016999125480651855, "grad_norm": 0.09924792498350143, "global_step": 18192, "epoch": 171, "lr": 7.363711493719241e-05} +{"train_loss": 0.01891031488776207, "grad_norm": 0.05769947171211243, "global_step": 18193, "epoch": 171, "lr": 7.363450322234206e-05} +{"train_loss": 0.016645219177007675, "grad_norm": 0.08715285360813141, "global_step": 18194, "epoch": 171, "lr": 7.363189142445034e-05} +{"train_loss": 0.02077306993305683, "grad_norm": 0.09338027983903885, "global_step": 18195, "epoch": 171, "lr": 7.362927954352644e-05} +{"train_loss": 0.015170292928814888, "grad_norm": 0.09497717022895813, "global_step": 18196, "epoch": 171, "lr": 7.362666757957952e-05} +{"train_loss": 0.0167770367115736, "grad_norm": 0.05378853902220726, "global_step": 18197, "epoch": 171, "lr": 7.362405553261877e-05} +{"train_loss": 0.015489812009036541, "grad_norm": 0.05862071365118027, "global_step": 18198, "epoch": 171, "lr": 7.362144340265336e-05} +{"train_loss": 0.017946533858776093, "grad_norm": 0.1140369400382042, "global_step": 18199, "epoch": 171, "lr": 7.361883118969247e-05} +{"train_loss": 0.012661611661314964, "grad_norm": 0.07014268636703491, "global_step": 18200, "epoch": 171, "lr": 7.361621889374529e-05} +{"train_loss": 0.016911007463932037, "grad_norm": 0.049279652535915375, "global_step": 18201, "epoch": 171, "lr": 7.361360651482097e-05} +{"train_loss": 0.015601333230733871, "grad_norm": 0.061453696340322495, "global_step": 18202, "epoch": 171, "lr": 7.361099405292871e-05} +{"train_loss": 0.01590372622013092, "grad_norm": 0.07286684960126877, "global_step": 18203, "epoch": 171, "lr": 7.360838150807768e-05} +{"train_loss": 0.016013771295547485, "grad_norm": 0.0677451640367508, "global_step": 18204, "epoch": 171, "lr": 7.360576888027708e-05} +{"train_loss": 0.016949055716395378, "grad_norm": 0.0591459758579731, "global_step": 18205, "epoch": 171, "lr": 7.360315616953606e-05} +{"train_loss": 0.01732615754008293, "grad_norm": 0.06268623471260071, "global_step": 18206, "epoch": 171, "lr": 7.360054337586382e-05} +{"train_loss": 0.017640255391597748, "grad_norm": 0.07834909856319427, "global_step": 18207, "epoch": 171, "lr": 7.359793049926953e-05} +{"train_loss": 0.014666961506009102, "grad_norm": 0.1025187224149704, "global_step": 18208, "epoch": 171, "lr": 7.359531753976239e-05} +{"train_loss": 0.01699843257665634, "grad_norm": 0.05143361911177635, "global_step": 18209, "epoch": 171, "lr": 7.359270449735157e-05} +{"train_loss": 0.019032763317227364, "grad_norm": 0.06884012371301651, "global_step": 18210, "epoch": 171, "lr": 7.359009137204622e-05} +{"train_loss": 0.0144282765686512, "grad_norm": 0.11541438847780228, "global_step": 18211, "epoch": 171, "lr": 7.358747816385556e-05} +{"train_loss": 0.01291346549987793, "grad_norm": 0.06419457495212555, "global_step": 18212, "epoch": 171, "lr": 7.358486487278876e-05} +{"train_loss": 0.01906679756939411, "grad_norm": 0.10362157970666885, "global_step": 18213, "epoch": 171, "lr": 7.358225149885499e-05} +{"train_loss": 0.018215805292129517, "grad_norm": 0.16827216744422913, "global_step": 18214, "epoch": 171, "lr": 7.357963804206347e-05} +{"train_loss": 0.016931068152189255, "grad_norm": 0.05929271876811981, "global_step": 18215, "epoch": 171, "lr": 7.357702450242333e-05} +{"train_loss": 0.019291570410132408, "grad_norm": 0.06823167204856873, "global_step": 18216, "epoch": 171, "lr": 7.35744108799438e-05} +{"train_loss": 0.016633059829473495, "grad_norm": 0.07648025453090668, "global_step": 18217, "epoch": 171, "lr": 7.357179717463403e-05} +{"train_loss": 0.016364257782697678, "grad_norm": 0.05844606086611748, "global_step": 18218, "epoch": 171, "lr": 7.356918338650322e-05} +{"train_loss": 0.015778986737132072, "grad_norm": 0.07356779277324677, "global_step": 18219, "epoch": 171, "lr": 7.356656951556053e-05} +{"train_loss": 0.013490142300724983, "grad_norm": 0.08188226073980331, "global_step": 18220, "epoch": 171, "lr": 7.35639555618152e-05} +{"train_loss": 0.017261117696762085, "grad_norm": 0.0753675326704979, "global_step": 18221, "epoch": 171, "lr": 7.356134152527634e-05} +{"train_loss": 0.01532595045864582, "grad_norm": 0.05688052624464035, "global_step": 18222, "epoch": 171, "lr": 7.355872740595321e-05} +{"train_loss": 0.014347397722303867, "grad_norm": 0.055226024240255356, "global_step": 18223, "epoch": 171, "lr": 7.355611320385494e-05} +{"train_loss": 0.015070239081978798, "grad_norm": 0.1198480874300003, "global_step": 18224, "epoch": 171, "lr": 7.355349891899072e-05} +{"train_loss": 0.013487514108419418, "grad_norm": 0.07839027047157288, "global_step": 18225, "epoch": 171, "lr": 7.355088455136976e-05} +{"train_loss": 0.016031891107559204, "grad_norm": 0.09342684596776962, "global_step": 18226, "epoch": 171, "lr": 7.354827010100125e-05} +{"train_loss": 0.016759535297751427, "grad_norm": 0.11624328047037125, "global_step": 18227, "epoch": 171, "lr": 7.354565556789432e-05} +{"train_loss": 0.014825168997049332, "grad_norm": 0.10052790492773056, "global_step": 18228, "epoch": 171, "lr": 7.354304095205823e-05} +{"train_loss": 0.015978962182998657, "grad_norm": 0.09829811751842499, "global_step": 18229, "epoch": 171, "lr": 7.354042625350213e-05} +{"train_loss": 0.01509919948875904, "grad_norm": 0.13162600994110107, "global_step": 18230, "epoch": 171, "lr": 7.35378114722352e-05} +{"train_loss": 0.01632366057063611, "grad_norm": 0.07943801581859589, "global_step": 18231, "epoch": 171, "lr": 7.353519660826665e-05, "val_loss": 0.019085217267274857} +{"train_loss": 0.01731368526816368, "grad_norm": 0.09382543712854385, "global_step": 18232, "epoch": 172, "lr": 7.353258166160563e-05} +{"train_loss": 0.0166181568056345, "grad_norm": 0.091678686439991, "global_step": 18233, "epoch": 172, "lr": 7.352996663226136e-05} +{"train_loss": 0.019172560423612595, "grad_norm": 0.06808403879404068, "global_step": 18234, "epoch": 172, "lr": 7.352735152024302e-05} +{"train_loss": 0.017534587532281876, "grad_norm": 0.047117412090301514, "global_step": 18235, "epoch": 172, "lr": 7.35247363255598e-05} +{"train_loss": 0.01852809637784958, "grad_norm": 0.06730162352323532, "global_step": 18236, "epoch": 172, "lr": 7.35221210482209e-05} +{"train_loss": 0.017348183318972588, "grad_norm": 0.06959903240203857, "global_step": 18237, "epoch": 172, "lr": 7.351950568823549e-05} +{"train_loss": 0.014707891270518303, "grad_norm": 0.0909530445933342, "global_step": 18238, "epoch": 172, "lr": 7.351689024561276e-05} +{"train_loss": 0.016558073461055756, "grad_norm": 0.10585121810436249, "global_step": 18239, "epoch": 172, "lr": 7.351427472036191e-05} +{"train_loss": 0.01859947293996811, "grad_norm": 0.13063403964042664, "global_step": 18240, "epoch": 172, "lr": 7.351165911249212e-05} +{"train_loss": 0.017355220392346382, "grad_norm": 0.2222055345773697, "global_step": 18241, "epoch": 172, "lr": 7.350904342201258e-05} +{"train_loss": 0.015256952494382858, "grad_norm": 0.09953993558883667, "global_step": 18242, "epoch": 172, "lr": 7.35064276489325e-05} +{"train_loss": 0.01654139906167984, "grad_norm": 0.12773443758487701, "global_step": 18243, "epoch": 172, "lr": 7.350381179326105e-05} +{"train_loss": 0.0167153961956501, "grad_norm": 0.07155897468328476, "global_step": 18244, "epoch": 172, "lr": 7.350119585500741e-05} +{"train_loss": 0.015499524772167206, "grad_norm": 0.08662581443786621, "global_step": 18245, "epoch": 172, "lr": 7.34985798341808e-05} +{"train_loss": 0.014260519295930862, "grad_norm": 0.07238735258579254, "global_step": 18246, "epoch": 172, "lr": 7.34959637307904e-05} +{"train_loss": 0.01581866294145584, "grad_norm": 0.04810621589422226, "global_step": 18247, "epoch": 172, "lr": 7.349334754484541e-05} +{"train_loss": 0.01680295541882515, "grad_norm": 0.07953635603189468, "global_step": 18248, "epoch": 172, "lr": 7.349073127635501e-05} +{"train_loss": 0.01658068783581257, "grad_norm": 0.060091208666563034, "global_step": 18249, "epoch": 172, "lr": 7.34881149253284e-05} +{"train_loss": 0.016331803053617477, "grad_norm": 0.05561324954032898, "global_step": 18250, "epoch": 172, "lr": 7.348549849177476e-05} +{"train_loss": 0.017380332574248314, "grad_norm": 0.19582560658454895, "global_step": 18251, "epoch": 172, "lr": 7.34828819757033e-05} +{"train_loss": 0.014336458407342434, "grad_norm": 0.1071043536067009, "global_step": 18252, "epoch": 172, "lr": 7.34802653771232e-05} +{"train_loss": 0.017539609223604202, "grad_norm": 0.09238873422145844, "global_step": 18253, "epoch": 172, "lr": 7.347764869604366e-05} +{"train_loss": 0.016612503677606583, "grad_norm": 0.1683579981327057, "global_step": 18254, "epoch": 172, "lr": 7.347503193247386e-05} +{"train_loss": 0.01494052167981863, "grad_norm": 0.06186068803071976, "global_step": 18255, "epoch": 172, "lr": 7.347241508642303e-05} +{"train_loss": 0.016777507960796356, "grad_norm": 0.0505499541759491, "global_step": 18256, "epoch": 172, "lr": 7.346979815790031e-05} +{"train_loss": 0.01675400882959366, "grad_norm": 0.07022182643413544, "global_step": 18257, "epoch": 172, "lr": 7.346718114691495e-05} +{"train_loss": 0.012940976768732071, "grad_norm": 0.04808833450078964, "global_step": 18258, "epoch": 172, "lr": 7.346456405347613e-05} +{"train_loss": 0.0169975683093071, "grad_norm": 0.07703568041324615, "global_step": 18259, "epoch": 172, "lr": 7.346194687759301e-05} +{"train_loss": 0.01649261638522148, "grad_norm": 0.09286561608314514, "global_step": 18260, "epoch": 172, "lr": 7.345932961927483e-05} +{"train_loss": 0.014477856457233429, "grad_norm": 0.06260628998279572, "global_step": 18261, "epoch": 172, "lr": 7.345671227853078e-05} +{"train_loss": 0.01828381046652794, "grad_norm": 0.07630634307861328, "global_step": 18262, "epoch": 172, "lr": 7.345409485537001e-05} +{"train_loss": 0.01617373153567314, "grad_norm": 0.05238383263349533, "global_step": 18263, "epoch": 172, "lr": 7.345147734980176e-05} +{"train_loss": 0.013508107513189316, "grad_norm": 0.08224928379058838, "global_step": 18264, "epoch": 172, "lr": 7.344885976183523e-05} +{"train_loss": 0.016703778877854347, "grad_norm": 0.06316094845533371, "global_step": 18265, "epoch": 172, "lr": 7.344624209147961e-05} +{"train_loss": 0.01582183502614498, "grad_norm": 0.0696854218840599, "global_step": 18266, "epoch": 172, "lr": 7.344362433874407e-05} +{"train_loss": 0.018635334447026253, "grad_norm": 0.06822196394205093, "global_step": 18267, "epoch": 172, "lr": 7.344100650363784e-05} +{"train_loss": 0.015377811156213284, "grad_norm": 0.07822569459676743, "global_step": 18268, "epoch": 172, "lr": 7.343838858617012e-05} +{"train_loss": 0.017820529639720917, "grad_norm": 0.09395374357700348, "global_step": 18269, "epoch": 172, "lr": 7.343577058635008e-05} +{"train_loss": 0.015081390738487244, "grad_norm": 0.12083189934492111, "global_step": 18270, "epoch": 172, "lr": 7.343315250418695e-05} +{"train_loss": 0.015972882509231567, "grad_norm": 0.09682013094425201, "global_step": 18271, "epoch": 172, "lr": 7.343053433968988e-05} +{"train_loss": 0.017887961119413376, "grad_norm": 0.12606890499591827, "global_step": 18272, "epoch": 172, "lr": 7.342791609286812e-05} +{"train_loss": 0.016390908509492874, "grad_norm": 0.05440367013216019, "global_step": 18273, "epoch": 172, "lr": 7.342529776373086e-05} +{"train_loss": 0.01400163397192955, "grad_norm": 0.08098835498094559, "global_step": 18274, "epoch": 172, "lr": 7.342267935228728e-05} +{"train_loss": 0.013741396367549896, "grad_norm": 0.11292406916618347, "global_step": 18275, "epoch": 172, "lr": 7.342006085854661e-05} +{"train_loss": 0.014303806237876415, "grad_norm": 0.05892001837491989, "global_step": 18276, "epoch": 172, "lr": 7.341744228251802e-05} +{"train_loss": 0.018969101831316948, "grad_norm": 0.07164687663316727, "global_step": 18277, "epoch": 172, "lr": 7.34148236242107e-05} +{"train_loss": 0.01889222115278244, "grad_norm": 0.13248494267463684, "global_step": 18278, "epoch": 172, "lr": 7.34122048836339e-05} +{"train_loss": 0.01768755167722702, "grad_norm": 0.061321862041950226, "global_step": 18279, "epoch": 172, "lr": 7.340958606079679e-05} +{"train_loss": 0.01695413887500763, "grad_norm": 0.05632888525724411, "global_step": 18280, "epoch": 172, "lr": 7.340696715570856e-05} +{"train_loss": 0.014211894944310188, "grad_norm": 0.0547969825565815, "global_step": 18281, "epoch": 172, "lr": 7.340434816837845e-05} +{"train_loss": 0.01528198178857565, "grad_norm": 0.10201559215784073, "global_step": 18282, "epoch": 172, "lr": 7.340172909881562e-05} +{"train_loss": 0.013421338982880116, "grad_norm": 0.08510071039199829, "global_step": 18283, "epoch": 172, "lr": 7.339910994702928e-05} +{"train_loss": 0.015152161940932274, "grad_norm": 0.08983000367879868, "global_step": 18284, "epoch": 172, "lr": 7.339649071302867e-05} +{"train_loss": 0.015958542004227638, "grad_norm": 0.09661099314689636, "global_step": 18285, "epoch": 172, "lr": 7.339387139682295e-05} +{"train_loss": 0.01866110786795616, "grad_norm": 0.07806805521249771, "global_step": 18286, "epoch": 172, "lr": 7.339125199842136e-05} +{"train_loss": 0.018947407603263855, "grad_norm": 0.08737854659557343, "global_step": 18287, "epoch": 172, "lr": 7.338863251783308e-05} +{"train_loss": 0.015933500602841377, "grad_norm": 0.09451691806316376, "global_step": 18288, "epoch": 172, "lr": 7.33860129550673e-05} +{"train_loss": 0.01583917625248432, "grad_norm": 0.06277189403772354, "global_step": 18289, "epoch": 172, "lr": 7.338339331013324e-05} +{"train_loss": 0.016208894550800323, "grad_norm": 0.10385045409202576, "global_step": 18290, "epoch": 172, "lr": 7.338077358304012e-05} +{"train_loss": 0.013883275911211967, "grad_norm": 0.04750684276223183, "global_step": 18291, "epoch": 172, "lr": 7.337815377379712e-05} +{"train_loss": 0.01631917804479599, "grad_norm": 0.05199733376502991, "global_step": 18292, "epoch": 172, "lr": 7.337553388241345e-05} +{"train_loss": 0.01546726655215025, "grad_norm": 0.0656183660030365, "global_step": 18293, "epoch": 172, "lr": 7.337291390889834e-05} +{"train_loss": 0.015040609985589981, "grad_norm": 0.08455170691013336, "global_step": 18294, "epoch": 172, "lr": 7.337029385326095e-05} +{"train_loss": 0.014644185081124306, "grad_norm": 0.07806442677974701, "global_step": 18295, "epoch": 172, "lr": 7.336767371551052e-05} +{"train_loss": 0.017191363498568535, "grad_norm": 0.07928033173084259, "global_step": 18296, "epoch": 172, "lr": 7.336505349565626e-05} +{"train_loss": 0.01666821911931038, "grad_norm": 0.05837379768490791, "global_step": 18297, "epoch": 172, "lr": 7.336243319370735e-05} +{"train_loss": 0.015587570145726204, "grad_norm": 0.07139269262552261, "global_step": 18298, "epoch": 172, "lr": 7.335981280967302e-05} +{"train_loss": 0.014580312184989452, "grad_norm": 0.0991414412856102, "global_step": 18299, "epoch": 172, "lr": 7.335719234356246e-05} +{"train_loss": 0.017363961786031723, "grad_norm": 0.09646715223789215, "global_step": 18300, "epoch": 172, "lr": 7.335457179538487e-05} +{"train_loss": 0.01720871403813362, "grad_norm": 0.05552738532423973, "global_step": 18301, "epoch": 172, "lr": 7.335195116514947e-05} +{"train_loss": 0.016396578401327133, "grad_norm": 0.056463997811079025, "global_step": 18302, "epoch": 172, "lr": 7.334933045286549e-05} +{"train_loss": 0.015282431617379189, "grad_norm": 0.08621768653392792, "global_step": 18303, "epoch": 172, "lr": 7.334670965854212e-05} +{"train_loss": 0.013026443310081959, "grad_norm": 0.056191060692071915, "global_step": 18304, "epoch": 172, "lr": 7.334408878218856e-05} +{"train_loss": 0.015453340485692024, "grad_norm": 0.06764781475067139, "global_step": 18305, "epoch": 172, "lr": 7.334146782381402e-05} +{"train_loss": 0.015806371346116066, "grad_norm": 0.06373046338558197, "global_step": 18306, "epoch": 172, "lr": 7.33388467834277e-05} +{"train_loss": 0.01778252050280571, "grad_norm": 0.08456244319677353, "global_step": 18307, "epoch": 172, "lr": 7.333622566103884e-05} +{"train_loss": 0.014072192832827568, "grad_norm": 0.11679070442914963, "global_step": 18308, "epoch": 172, "lr": 7.333360445665664e-05} +{"train_loss": 0.016058992594480515, "grad_norm": 0.10285793244838715, "global_step": 18309, "epoch": 172, "lr": 7.333098317029028e-05} +{"train_loss": 0.017721351236104965, "grad_norm": 0.07932906597852707, "global_step": 18310, "epoch": 172, "lr": 7.3328361801949e-05} +{"train_loss": 0.017354119569063187, "grad_norm": 0.0906699001789093, "global_step": 18311, "epoch": 172, "lr": 7.332574035164202e-05} +{"train_loss": 0.019634831696748734, "grad_norm": 0.0651642233133316, "global_step": 18312, "epoch": 172, "lr": 7.332311881937849e-05} +{"train_loss": 0.017837096005678177, "grad_norm": 0.0778549388051033, "global_step": 18313, "epoch": 172, "lr": 7.33204972051677e-05} +{"train_loss": 0.016644470393657684, "grad_norm": 0.08988878130912781, "global_step": 18314, "epoch": 172, "lr": 7.33178755090188e-05} +{"train_loss": 0.01496377494186163, "grad_norm": 0.051351193338632584, "global_step": 18315, "epoch": 172, "lr": 7.331525373094104e-05} +{"train_loss": 0.017922475934028625, "grad_norm": 0.08414600789546967, "global_step": 18316, "epoch": 172, "lr": 7.331263187094362e-05} +{"train_loss": 0.01862703636288643, "grad_norm": 0.1670786291360855, "global_step": 18317, "epoch": 172, "lr": 7.331000992903574e-05} +{"train_loss": 0.018360648304224014, "grad_norm": 0.061955999583005905, "global_step": 18318, "epoch": 172, "lr": 7.330738790522664e-05} +{"train_loss": 0.01617348939180374, "grad_norm": 0.11665238440036774, "global_step": 18319, "epoch": 172, "lr": 7.330476579952549e-05} +{"train_loss": 0.013242198154330254, "grad_norm": 0.10475541651248932, "global_step": 18320, "epoch": 172, "lr": 7.330214361194155e-05} +{"train_loss": 0.014127046801149845, "grad_norm": 0.08178450167179108, "global_step": 18321, "epoch": 172, "lr": 7.329952134248401e-05} +{"train_loss": 0.017948497086763382, "grad_norm": 0.08307617157697678, "global_step": 18322, "epoch": 172, "lr": 7.329689899116207e-05} +{"train_loss": 0.016258707270026207, "grad_norm": 0.09042010456323624, "global_step": 18323, "epoch": 172, "lr": 7.329427655798497e-05} +{"train_loss": 0.012852008454501629, "grad_norm": 0.16283895075321198, "global_step": 18324, "epoch": 172, "lr": 7.329165404296189e-05} +{"train_loss": 0.01799851655960083, "grad_norm": 0.08007900416851044, "global_step": 18325, "epoch": 172, "lr": 7.328903144610209e-05} +{"train_loss": 0.015297935344278812, "grad_norm": 0.06040005385875702, "global_step": 18326, "epoch": 172, "lr": 7.328640876741478e-05} +{"train_loss": 0.01369551196694374, "grad_norm": 0.09022417664527893, "global_step": 18327, "epoch": 172, "lr": 7.328378600690913e-05} +{"train_loss": 0.01630888506770134, "grad_norm": 0.05593738704919815, "global_step": 18328, "epoch": 172, "lr": 7.328116316459438e-05} +{"train_loss": 0.015455974265933037, "grad_norm": 0.061939287930727005, "global_step": 18329, "epoch": 172, "lr": 7.327854024047975e-05} +{"train_loss": 0.018047643825411797, "grad_norm": 0.10769469290971756, "global_step": 18330, "epoch": 172, "lr": 7.327591723457446e-05} +{"train_loss": 0.018441084772348404, "grad_norm": 0.0601179376244545, "global_step": 18331, "epoch": 172, "lr": 7.327329414688772e-05} +{"train_loss": 0.016554642468690872, "grad_norm": 0.08714743703603745, "global_step": 18332, "epoch": 172, "lr": 7.327067097742875e-05} +{"train_loss": 0.016916241496801376, "grad_norm": 0.05318838357925415, "global_step": 18333, "epoch": 172, "lr": 7.326804772620674e-05} +{"train_loss": 0.01619962975382805, "grad_norm": 0.06873800605535507, "global_step": 18334, "epoch": 172, "lr": 7.326542439323094e-05} +{"train_loss": 0.01849028840661049, "grad_norm": 0.08579268306493759, "global_step": 18335, "epoch": 172, "lr": 7.326280097851057e-05} +{"train_loss": 0.018193397670984268, "grad_norm": 0.08891243487596512, "global_step": 18336, "epoch": 172, "lr": 7.326017748205481e-05} +{"train_loss": 0.016302956865643554, "grad_norm": 0.06556317955255508, "global_step": 18337, "epoch": 172, "lr": 7.325755390387292e-05, "val_loss": 0.019162185490131378} +{"train_loss": 0.01613110676407814, "grad_norm": 0.06701085716485977, "global_step": 18338, "epoch": 173, "lr": 7.325493024397411e-05} +{"train_loss": 0.015873417258262634, "grad_norm": 0.10299823433160782, "global_step": 18339, "epoch": 173, "lr": 7.325230650236756e-05} +{"train_loss": 0.018244318664073944, "grad_norm": 0.11288284510374069, "global_step": 18340, "epoch": 173, "lr": 7.324968267906253e-05} +{"train_loss": 0.016053633764386177, "grad_norm": 0.12837651371955872, "global_step": 18341, "epoch": 173, "lr": 7.324705877406824e-05} +{"train_loss": 0.01473512127995491, "grad_norm": 0.08554008603096008, "global_step": 18342, "epoch": 173, "lr": 7.324443478739386e-05} +{"train_loss": 0.016046838834881783, "grad_norm": 0.057955630123615265, "global_step": 18343, "epoch": 173, "lr": 7.324181071904867e-05} +{"train_loss": 0.01631062477827072, "grad_norm": 0.15432938933372498, "global_step": 18344, "epoch": 173, "lr": 7.323918656904187e-05} +{"train_loss": 0.01668623462319374, "grad_norm": 0.09126869589090347, "global_step": 18345, "epoch": 173, "lr": 7.323656233738267e-05} +{"train_loss": 0.016891809180378914, "grad_norm": 0.11837142705917358, "global_step": 18346, "epoch": 173, "lr": 7.32339380240803e-05} +{"train_loss": 0.01587321050465107, "grad_norm": 0.0881541520357132, "global_step": 18347, "epoch": 173, "lr": 7.323131362914395e-05} +{"train_loss": 0.018174000084400177, "grad_norm": 0.051104336977005005, "global_step": 18348, "epoch": 173, "lr": 7.322868915258288e-05} +{"train_loss": 0.016630098223686218, "grad_norm": 0.059161800891160965, "global_step": 18349, "epoch": 173, "lr": 7.322606459440631e-05} +{"train_loss": 0.017326971516013145, "grad_norm": 0.09947504103183746, "global_step": 18350, "epoch": 173, "lr": 7.322343995462345e-05} +{"train_loss": 0.01348959468305111, "grad_norm": 0.1415696144104004, "global_step": 18351, "epoch": 173, "lr": 7.322081523324351e-05} +{"train_loss": 0.01705586537718773, "grad_norm": 0.11462660133838654, "global_step": 18352, "epoch": 173, "lr": 7.321819043027573e-05} +{"train_loss": 0.013599495403468609, "grad_norm": 0.07644017040729523, "global_step": 18353, "epoch": 173, "lr": 7.321556554572933e-05} +{"train_loss": 0.015651056542992592, "grad_norm": 0.0611642524600029, "global_step": 18354, "epoch": 173, "lr": 7.321294057961352e-05} +{"train_loss": 0.016500115394592285, "grad_norm": 0.11675399541854858, "global_step": 18355, "epoch": 173, "lr": 7.321031553193753e-05} +{"train_loss": 0.0172513909637928, "grad_norm": 0.07563062757253647, "global_step": 18356, "epoch": 173, "lr": 7.320769040271059e-05} +{"train_loss": 0.014276260510087013, "grad_norm": 0.08562470227479935, "global_step": 18357, "epoch": 173, "lr": 7.320506519194192e-05} +{"train_loss": 0.015667904168367386, "grad_norm": 0.09064920991659164, "global_step": 18358, "epoch": 173, "lr": 7.320243989964076e-05} +{"train_loss": 0.017896534875035286, "grad_norm": 0.1058931052684784, "global_step": 18359, "epoch": 173, "lr": 7.319981452581629e-05} +{"train_loss": 0.017410364001989365, "grad_norm": 0.07718769460916519, "global_step": 18360, "epoch": 173, "lr": 7.319718907047776e-05} +{"train_loss": 0.01990041695535183, "grad_norm": 0.07437458634376526, "global_step": 18361, "epoch": 173, "lr": 7.31945635336344e-05} +{"train_loss": 0.016929058358073235, "grad_norm": 0.0682314857840538, "global_step": 18362, "epoch": 173, "lr": 7.319193791529545e-05} +{"train_loss": 0.01672939769923687, "grad_norm": 0.10728128999471664, "global_step": 18363, "epoch": 173, "lr": 7.318931221547009e-05} +{"train_loss": 0.016366783529520035, "grad_norm": 0.048853471875190735, "global_step": 18364, "epoch": 173, "lr": 7.318668643416759e-05} +{"train_loss": 0.017100326716899872, "grad_norm": 0.1023840680718422, "global_step": 18365, "epoch": 173, "lr": 7.318406057139715e-05} +{"train_loss": 0.014858664013445377, "grad_norm": 0.10719063878059387, "global_step": 18366, "epoch": 173, "lr": 7.3181434627168e-05} +{"train_loss": 0.01612543687224388, "grad_norm": 0.08887644112110138, "global_step": 18367, "epoch": 173, "lr": 7.317880860148939e-05} +{"train_loss": 0.0175729189068079, "grad_norm": 0.06249357387423515, "global_step": 18368, "epoch": 173, "lr": 7.31761824943705e-05} +{"train_loss": 0.017922526225447655, "grad_norm": 0.11911357194185257, "global_step": 18369, "epoch": 173, "lr": 7.31735563058206e-05} +{"train_loss": 0.015442462638020515, "grad_norm": 0.07512497156858444, "global_step": 18370, "epoch": 173, "lr": 7.317093003584889e-05} +{"train_loss": 0.013811103999614716, "grad_norm": 0.08702303469181061, "global_step": 18371, "epoch": 173, "lr": 7.316830368446462e-05} +{"train_loss": 0.01574457623064518, "grad_norm": 0.05840303376317024, "global_step": 18372, "epoch": 173, "lr": 7.3165677251677e-05} +{"train_loss": 0.0168379507958889, "grad_norm": 0.058370355516672134, "global_step": 18373, "epoch": 173, "lr": 7.316305073749528e-05} +{"train_loss": 0.015279114246368408, "grad_norm": 0.11231312900781631, "global_step": 18374, "epoch": 173, "lr": 7.316042414192865e-05} +{"train_loss": 0.014926616102457047, "grad_norm": 0.05072246491909027, "global_step": 18375, "epoch": 173, "lr": 7.315779746498637e-05} +{"train_loss": 0.017212986946105957, "grad_norm": 0.05593303218483925, "global_step": 18376, "epoch": 173, "lr": 7.315517070667766e-05} +{"train_loss": 0.01588599383831024, "grad_norm": 0.09393300861120224, "global_step": 18377, "epoch": 173, "lr": 7.315254386701175e-05} +{"train_loss": 0.015565825626254082, "grad_norm": 0.07498308271169662, "global_step": 18378, "epoch": 173, "lr": 7.314991694599786e-05} +{"train_loss": 0.01333647221326828, "grad_norm": 0.08738694339990616, "global_step": 18379, "epoch": 173, "lr": 7.314728994364526e-05} +{"train_loss": 0.014252375811338425, "grad_norm": 0.056035663932561874, "global_step": 18380, "epoch": 173, "lr": 7.314466285996312e-05} +{"train_loss": 0.013613227754831314, "grad_norm": 0.09255608916282654, "global_step": 18381, "epoch": 173, "lr": 7.314203569496071e-05} +{"train_loss": 0.017491420730948448, "grad_norm": 0.08486992865800858, "global_step": 18382, "epoch": 173, "lr": 7.313940844864725e-05} +{"train_loss": 0.015031151473522186, "grad_norm": 0.10306117683649063, "global_step": 18383, "epoch": 173, "lr": 7.313678112103198e-05} +{"train_loss": 0.01688554510474205, "grad_norm": 0.06204553321003914, "global_step": 18384, "epoch": 173, "lr": 7.31341537121241e-05} +{"train_loss": 0.017062656581401825, "grad_norm": 0.09259549528360367, "global_step": 18385, "epoch": 173, "lr": 7.31315262219329e-05} +{"train_loss": 0.014199022203683853, "grad_norm": 0.06810259819030762, "global_step": 18386, "epoch": 173, "lr": 7.312889865046755e-05} +{"train_loss": 0.015486261807382107, "grad_norm": 0.09446506202220917, "global_step": 18387, "epoch": 173, "lr": 7.312627099773732e-05} +{"train_loss": 0.016194339841604233, "grad_norm": 0.06602500379085541, "global_step": 18388, "epoch": 173, "lr": 7.312364326375143e-05} +{"train_loss": 0.014776745811104774, "grad_norm": 0.1162911131978035, "global_step": 18389, "epoch": 173, "lr": 7.31210154485191e-05} +{"train_loss": 0.015574374236166477, "grad_norm": 0.06648849695920944, "global_step": 18390, "epoch": 173, "lr": 7.311838755204959e-05} +{"train_loss": 0.015254208818078041, "grad_norm": 0.1363660991191864, "global_step": 18391, "epoch": 173, "lr": 7.311575957435213e-05} +{"train_loss": 0.01638047769665718, "grad_norm": 0.06348586082458496, "global_step": 18392, "epoch": 173, "lr": 7.311313151543591e-05} +{"train_loss": 0.019573654979467392, "grad_norm": 0.13981492817401886, "global_step": 18393, "epoch": 173, "lr": 7.311050337531024e-05} +{"train_loss": 0.01856118068099022, "grad_norm": 0.06714068353176117, "global_step": 18394, "epoch": 173, "lr": 7.310787515398428e-05} +{"train_loss": 0.014338625594973564, "grad_norm": 0.06403357535600662, "global_step": 18395, "epoch": 173, "lr": 7.310524685146731e-05} +{"train_loss": 0.016805216670036316, "grad_norm": 0.0502096563577652, "global_step": 18396, "epoch": 173, "lr": 7.310261846776855e-05} +{"train_loss": 0.014810632914304733, "grad_norm": 0.12778319418430328, "global_step": 18397, "epoch": 173, "lr": 7.309999000289722e-05} +{"train_loss": 0.015231085941195488, "grad_norm": 0.11395304650068283, "global_step": 18398, "epoch": 173, "lr": 7.309736145686259e-05} +{"train_loss": 0.015369526110589504, "grad_norm": 0.06256242096424103, "global_step": 18399, "epoch": 173, "lr": 7.309473282967387e-05} +{"train_loss": 0.015370892360806465, "grad_norm": 0.04518330469727516, "global_step": 18400, "epoch": 173, "lr": 7.30921041213403e-05} +{"train_loss": 0.015014027245342731, "grad_norm": 0.06230318546295166, "global_step": 18401, "epoch": 173, "lr": 7.308947533187113e-05} +{"train_loss": 0.01605325937271118, "grad_norm": 0.05838192254304886, "global_step": 18402, "epoch": 173, "lr": 7.308684646127557e-05} +{"train_loss": 0.019679639488458633, "grad_norm": 0.06839525699615479, "global_step": 18403, "epoch": 173, "lr": 7.308421750956289e-05} +{"train_loss": 0.015747617930173874, "grad_norm": 0.060625240206718445, "global_step": 18404, "epoch": 173, "lr": 7.308158847674228e-05} +{"train_loss": 0.01628357172012329, "grad_norm": 0.07763556391000748, "global_step": 18405, "epoch": 173, "lr": 7.307895936282304e-05} +{"train_loss": 0.01511168759316206, "grad_norm": 0.08379319310188293, "global_step": 18406, "epoch": 173, "lr": 7.307633016781434e-05} +{"train_loss": 0.016183573752641678, "grad_norm": 0.08291555196046829, "global_step": 18407, "epoch": 173, "lr": 7.307370089172547e-05} +{"train_loss": 0.015725545585155487, "grad_norm": 0.08812811970710754, "global_step": 18408, "epoch": 173, "lr": 7.307107153456566e-05} +{"train_loss": 0.019277723506093025, "grad_norm": 0.08455710113048553, "global_step": 18409, "epoch": 173, "lr": 7.306844209634411e-05} +{"train_loss": 0.013820130378007889, "grad_norm": 0.07720206677913666, "global_step": 18410, "epoch": 173, "lr": 7.306581257707012e-05} +{"train_loss": 0.016572246327996254, "grad_norm": 0.053822483867406845, "global_step": 18411, "epoch": 173, "lr": 7.306318297675288e-05} +{"train_loss": 0.019152557477355003, "grad_norm": 0.08576826751232147, "global_step": 18412, "epoch": 173, "lr": 7.306055329540164e-05} +{"train_loss": 0.017105359584093094, "grad_norm": 0.06389492005109787, "global_step": 18413, "epoch": 173, "lr": 7.305792353302564e-05} +{"train_loss": 0.016772089526057243, "grad_norm": 0.08382300287485123, "global_step": 18414, "epoch": 173, "lr": 7.305529368963413e-05} +{"train_loss": 0.01634492538869381, "grad_norm": 0.09249993413686752, "global_step": 18415, "epoch": 173, "lr": 7.305266376523636e-05} +{"train_loss": 0.014744097366929054, "grad_norm": 0.09005115926265717, "global_step": 18416, "epoch": 173, "lr": 7.305003375984152e-05} +{"train_loss": 0.01749059557914734, "grad_norm": 0.0786871686577797, "global_step": 18417, "epoch": 173, "lr": 7.304740367345892e-05} +{"train_loss": 0.01891971193253994, "grad_norm": 0.11426806449890137, "global_step": 18418, "epoch": 173, "lr": 7.304477350609776e-05} +{"train_loss": 0.017149552702903748, "grad_norm": 0.13568714261054993, "global_step": 18419, "epoch": 173, "lr": 7.304214325776727e-05} +{"train_loss": 0.01760617084801197, "grad_norm": 0.18086594343185425, "global_step": 18420, "epoch": 173, "lr": 7.303951292847673e-05} +{"train_loss": 0.017032135277986526, "grad_norm": 0.0509825125336647, "global_step": 18421, "epoch": 173, "lr": 7.303688251823534e-05} +{"train_loss": 0.015134355053305626, "grad_norm": 0.08493020385503769, "global_step": 18422, "epoch": 173, "lr": 7.303425202705238e-05} +{"train_loss": 0.015400353819131851, "grad_norm": 0.08739855885505676, "global_step": 18423, "epoch": 173, "lr": 7.303162145493707e-05} +{"train_loss": 0.016726024448871613, "grad_norm": 0.1388469636440277, "global_step": 18424, "epoch": 173, "lr": 7.302899080189866e-05} +{"train_loss": 0.016546498984098434, "grad_norm": 0.10851989686489105, "global_step": 18425, "epoch": 173, "lr": 7.302636006794638e-05} +{"train_loss": 0.01590607687830925, "grad_norm": 0.06598234176635742, "global_step": 18426, "epoch": 173, "lr": 7.302372925308951e-05} +{"train_loss": 0.02005986124277115, "grad_norm": 0.09104184061288834, "global_step": 18427, "epoch": 173, "lr": 7.302109835733725e-05} +{"train_loss": 0.018061500042676926, "grad_norm": 0.07082691788673401, "global_step": 18428, "epoch": 173, "lr": 7.301846738069885e-05} +{"train_loss": 0.01734422706067562, "grad_norm": 0.07840048521757126, "global_step": 18429, "epoch": 173, "lr": 7.301583632318359e-05} +{"train_loss": 0.013715342618525028, "grad_norm": 0.06932748854160309, "global_step": 18430, "epoch": 173, "lr": 7.30132051848007e-05} +{"train_loss": 0.016717400401830673, "grad_norm": 0.1399083286523819, "global_step": 18431, "epoch": 173, "lr": 7.301057396555939e-05} +{"train_loss": 0.01829412207007408, "grad_norm": 0.07751869410276413, "global_step": 18432, "epoch": 173, "lr": 7.300794266546893e-05} +{"train_loss": 0.01791144721210003, "grad_norm": 0.10131148248910904, "global_step": 18433, "epoch": 173, "lr": 7.300531128453857e-05} +{"train_loss": 0.0164500679820776, "grad_norm": 0.07366158813238144, "global_step": 18434, "epoch": 173, "lr": 7.300267982277757e-05} +{"train_loss": 0.014037640765309334, "grad_norm": 0.04848073422908783, "global_step": 18435, "epoch": 173, "lr": 7.300004828019512e-05} +{"train_loss": 0.016788173466920853, "grad_norm": 0.11093547940254211, "global_step": 18436, "epoch": 173, "lr": 7.299741665680054e-05} +{"train_loss": 0.01544866431504488, "grad_norm": 0.0700744241476059, "global_step": 18437, "epoch": 173, "lr": 7.299478495260301e-05} +{"train_loss": 0.014797909185290337, "grad_norm": 0.15918241441249847, "global_step": 18438, "epoch": 173, "lr": 7.299215316761182e-05} +{"train_loss": 0.015183799900114536, "grad_norm": 0.12457624822854996, "global_step": 18439, "epoch": 173, "lr": 7.29895213018362e-05} +{"train_loss": 0.01733546331524849, "grad_norm": 0.06547139585018158, "global_step": 18440, "epoch": 173, "lr": 7.29868893552854e-05} +{"train_loss": 0.017049437388777733, "grad_norm": 0.05894242972135544, "global_step": 18441, "epoch": 173, "lr": 7.298425732796867e-05} +{"train_loss": 0.014990908093750477, "grad_norm": 0.13904190063476562, "global_step": 18442, "epoch": 173, "lr": 7.298162521989525e-05} +{"train_loss": 0.01628534335045601, "grad_norm": 0.08098051697015762, "global_step": 18443, "epoch": 173, "lr": 7.297899303107441e-05, "val_loss": 0.019153200089931488} +{"train_loss": 0.018223661929368973, "grad_norm": 0.06145209074020386, "global_step": 18444, "epoch": 174, "lr": 7.297636076151536e-05} +{"train_loss": 0.014896184206008911, "grad_norm": 0.0687694102525711, "global_step": 18445, "epoch": 174, "lr": 7.297372841122737e-05} +{"train_loss": 0.014703302644193172, "grad_norm": 0.11036968231201172, "global_step": 18446, "epoch": 174, "lr": 7.29710959802197e-05} +{"train_loss": 0.017201654613018036, "grad_norm": 0.10269507020711899, "global_step": 18447, "epoch": 174, "lr": 7.29684634685016e-05} +{"train_loss": 0.013540767133235931, "grad_norm": 0.053902849555015564, "global_step": 18448, "epoch": 174, "lr": 7.296583087608228e-05} +{"train_loss": 0.016635090112686157, "grad_norm": 0.12217509746551514, "global_step": 18449, "epoch": 174, "lr": 7.296319820297106e-05} +{"train_loss": 0.01806318759918213, "grad_norm": 0.08400868624448776, "global_step": 18450, "epoch": 174, "lr": 7.29605654491771e-05} +{"train_loss": 0.015459461137652397, "grad_norm": 0.11136182397603989, "global_step": 18451, "epoch": 174, "lr": 7.295793261470972e-05} +{"train_loss": 0.015728505328297615, "grad_norm": 0.078698068857193, "global_step": 18452, "epoch": 174, "lr": 7.295529969957815e-05} +{"train_loss": 0.013682423159480095, "grad_norm": 0.12837427854537964, "global_step": 18453, "epoch": 174, "lr": 7.295266670379164e-05} +{"train_loss": 0.014200204983353615, "grad_norm": 0.05201396718621254, "global_step": 18454, "epoch": 174, "lr": 7.295003362735944e-05} +{"train_loss": 0.017733369022607803, "grad_norm": 0.05892409756779671, "global_step": 18455, "epoch": 174, "lr": 7.294740047029079e-05} +{"train_loss": 0.013175315223634243, "grad_norm": 0.08091362565755844, "global_step": 18456, "epoch": 174, "lr": 7.294476723259499e-05} +{"train_loss": 0.013916756957769394, "grad_norm": 0.07419928163290024, "global_step": 18457, "epoch": 174, "lr": 7.29421339142812e-05} +{"train_loss": 0.014672965742647648, "grad_norm": 0.12827451527118683, "global_step": 18458, "epoch": 174, "lr": 7.293950051535879e-05} +{"train_loss": 0.016944892704486847, "grad_norm": 0.06296757608652115, "global_step": 18459, "epoch": 174, "lr": 7.29368670358369e-05} +{"train_loss": 0.01655573770403862, "grad_norm": 0.07553250342607498, "global_step": 18460, "epoch": 174, "lr": 7.293423347572485e-05} +{"train_loss": 0.015795955434441566, "grad_norm": 0.0547567680478096, "global_step": 18461, "epoch": 174, "lr": 7.29315998350319e-05} +{"train_loss": 0.014799796044826508, "grad_norm": 0.08758919686079025, "global_step": 18462, "epoch": 174, "lr": 7.292896611376726e-05} +{"train_loss": 0.013865772634744644, "grad_norm": 0.09055451303720474, "global_step": 18463, "epoch": 174, "lr": 7.292633231194021e-05} +{"train_loss": 0.01848534494638443, "grad_norm": 0.13859042525291443, "global_step": 18464, "epoch": 174, "lr": 7.292369842955999e-05} +{"train_loss": 0.0162532739341259, "grad_norm": 0.05328745022416115, "global_step": 18465, "epoch": 174, "lr": 7.292106446663588e-05} +{"train_loss": 0.016246136277914047, "grad_norm": 0.20452837646007538, "global_step": 18466, "epoch": 174, "lr": 7.29184304231771e-05} +{"train_loss": 0.018121732398867607, "grad_norm": 0.09668037295341492, "global_step": 18467, "epoch": 174, "lr": 7.291579629919294e-05} +{"train_loss": 0.016702787950634956, "grad_norm": 0.07408711314201355, "global_step": 18468, "epoch": 174, "lr": 7.291316209469263e-05} +{"train_loss": 0.01746954582631588, "grad_norm": 0.05073961615562439, "global_step": 18469, "epoch": 174, "lr": 7.291052780968543e-05} +{"train_loss": 0.01620243489742279, "grad_norm": 0.07007420808076859, "global_step": 18470, "epoch": 174, "lr": 7.29078934441806e-05} +{"train_loss": 0.01454123854637146, "grad_norm": 0.10542698949575424, "global_step": 18471, "epoch": 174, "lr": 7.290525899818739e-05} +{"train_loss": 0.016176752746105194, "grad_norm": 0.06453107297420502, "global_step": 18472, "epoch": 174, "lr": 7.290262447171508e-05} +{"train_loss": 0.016252771019935608, "grad_norm": 0.07216750085353851, "global_step": 18473, "epoch": 174, "lr": 7.28999898647729e-05} +{"train_loss": 0.013491244055330753, "grad_norm": 0.053470488637685776, "global_step": 18474, "epoch": 174, "lr": 7.28973551773701e-05} +{"train_loss": 0.016604509204626083, "grad_norm": 0.07044867426156998, "global_step": 18475, "epoch": 174, "lr": 7.289472040951595e-05} +{"train_loss": 0.019096052274107933, "grad_norm": 0.08274029940366745, "global_step": 18476, "epoch": 174, "lr": 7.289208556121974e-05} +{"train_loss": 0.016418278217315674, "grad_norm": 0.052448570728302, "global_step": 18477, "epoch": 174, "lr": 7.288945063249066e-05} +{"train_loss": 0.016408372670412064, "grad_norm": 0.06429766118526459, "global_step": 18478, "epoch": 174, "lr": 7.288681562333803e-05} +{"train_loss": 0.018796386197209358, "grad_norm": 0.058812521398067474, "global_step": 18479, "epoch": 174, "lr": 7.288418053377107e-05} +{"train_loss": 0.013605195097625256, "grad_norm": 0.11305839568376541, "global_step": 18480, "epoch": 174, "lr": 7.288154536379905e-05} +{"train_loss": 0.014347061514854431, "grad_norm": 0.10689090937376022, "global_step": 18481, "epoch": 174, "lr": 7.287891011343125e-05} +{"train_loss": 0.015202412381768227, "grad_norm": 0.10301913321018219, "global_step": 18482, "epoch": 174, "lr": 7.287627478267689e-05} +{"train_loss": 0.01679087057709694, "grad_norm": 0.06651270389556885, "global_step": 18483, "epoch": 174, "lr": 7.287363937154526e-05} +{"train_loss": 0.014108164235949516, "grad_norm": 0.08846521377563477, "global_step": 18484, "epoch": 174, "lr": 7.287100388004559e-05} +{"train_loss": 0.015134221874177456, "grad_norm": 0.10969746857881546, "global_step": 18485, "epoch": 174, "lr": 7.286836830818718e-05} +{"train_loss": 0.015988409519195557, "grad_norm": 0.06401176005601883, "global_step": 18486, "epoch": 174, "lr": 7.286573265597926e-05} +{"train_loss": 0.013858037069439888, "grad_norm": 0.1346447765827179, "global_step": 18487, "epoch": 174, "lr": 7.286309692343111e-05} +{"train_loss": 0.016298318281769753, "grad_norm": 0.06162365525960922, "global_step": 18488, "epoch": 174, "lr": 7.286046111055197e-05} +{"train_loss": 0.01520264707505703, "grad_norm": 0.07090231776237488, "global_step": 18489, "epoch": 174, "lr": 7.28578252173511e-05} +{"train_loss": 0.017561353743076324, "grad_norm": 0.04738591983914375, "global_step": 18490, "epoch": 174, "lr": 7.285518924383781e-05} +{"train_loss": 0.015536953695118427, "grad_norm": 0.07465435564517975, "global_step": 18491, "epoch": 174, "lr": 7.285255319002128e-05} +{"train_loss": 0.01416795700788498, "grad_norm": 0.07891886681318283, "global_step": 18492, "epoch": 174, "lr": 7.284991705591084e-05} +{"train_loss": 0.01564892567694187, "grad_norm": 0.048014748841524124, "global_step": 18493, "epoch": 174, "lr": 7.284728084151574e-05} +{"train_loss": 0.017815442755818367, "grad_norm": 0.11190541833639145, "global_step": 18494, "epoch": 174, "lr": 7.284464454684521e-05} +{"train_loss": 0.01660076156258583, "grad_norm": 0.07574672251939774, "global_step": 18495, "epoch": 174, "lr": 7.284200817190854e-05} +{"train_loss": 0.013968023471534252, "grad_norm": 0.08146000653505325, "global_step": 18496, "epoch": 174, "lr": 7.283937171671498e-05} +{"train_loss": 0.016357172280550003, "grad_norm": 0.1250649243593216, "global_step": 18497, "epoch": 174, "lr": 7.28367351812738e-05} +{"train_loss": 0.01736639440059662, "grad_norm": 0.10759621858596802, "global_step": 18498, "epoch": 174, "lr": 7.283409856559427e-05} +{"train_loss": 0.016685891896486282, "grad_norm": 0.05476857349276543, "global_step": 18499, "epoch": 174, "lr": 7.283146186968565e-05} +{"train_loss": 0.015983037650585175, "grad_norm": 0.07148446142673492, "global_step": 18500, "epoch": 174, "lr": 7.282882509355721e-05} +{"train_loss": 0.017482222989201546, "grad_norm": 0.07985250651836395, "global_step": 18501, "epoch": 174, "lr": 7.28261882372182e-05} +{"train_loss": 0.01553274691104889, "grad_norm": 0.05459815263748169, "global_step": 18502, "epoch": 174, "lr": 7.28235513006779e-05} +{"train_loss": 0.015310840681195259, "grad_norm": 0.08861102163791656, "global_step": 18503, "epoch": 174, "lr": 7.282091428394555e-05} +{"train_loss": 0.015264022164046764, "grad_norm": 0.10695645958185196, "global_step": 18504, "epoch": 174, "lr": 7.281827718703043e-05} +{"train_loss": 0.01758948341012001, "grad_norm": 0.07146143913269043, "global_step": 18505, "epoch": 174, "lr": 7.281564000994183e-05} +{"train_loss": 0.01642036996781826, "grad_norm": 0.052023664116859436, "global_step": 18506, "epoch": 174, "lr": 7.281300275268897e-05} +{"train_loss": 0.016998007893562317, "grad_norm": 0.12367598712444305, "global_step": 18507, "epoch": 174, "lr": 7.281036541528114e-05} +{"train_loss": 0.020043838769197464, "grad_norm": 0.09111791849136353, "global_step": 18508, "epoch": 174, "lr": 7.280772799772762e-05} +{"train_loss": 0.015538221225142479, "grad_norm": 0.06022235006093979, "global_step": 18509, "epoch": 174, "lr": 7.280509050003766e-05} +{"train_loss": 0.01626160554587841, "grad_norm": 0.056245118379592896, "global_step": 18510, "epoch": 174, "lr": 7.280245292222052e-05} +{"train_loss": 0.017951086163520813, "grad_norm": 0.04735235869884491, "global_step": 18511, "epoch": 174, "lr": 7.279981526428549e-05} +{"train_loss": 0.014688460156321526, "grad_norm": 0.06814566254615784, "global_step": 18512, "epoch": 174, "lr": 7.279717752624181e-05} +{"train_loss": 0.01766257919371128, "grad_norm": 0.06196705251932144, "global_step": 18513, "epoch": 174, "lr": 7.279453970809876e-05} +{"train_loss": 0.015549526549875736, "grad_norm": 0.04672616720199585, "global_step": 18514, "epoch": 174, "lr": 7.279190180986564e-05} +{"train_loss": 0.017470676451921463, "grad_norm": 0.17389191687107086, "global_step": 18515, "epoch": 174, "lr": 7.278926383155165e-05} +{"train_loss": 0.017396491020917892, "grad_norm": 0.08338190615177155, "global_step": 18516, "epoch": 174, "lr": 7.278662577316613e-05} +{"train_loss": 0.013422084972262383, "grad_norm": 0.10794680565595627, "global_step": 18517, "epoch": 174, "lr": 7.278398763471831e-05} +{"train_loss": 0.015593837015330791, "grad_norm": 0.09580373018980026, "global_step": 18518, "epoch": 174, "lr": 7.278134941621744e-05} +{"train_loss": 0.014873053878545761, "grad_norm": 0.06998050957918167, "global_step": 18519, "epoch": 174, "lr": 7.277871111767284e-05} +{"train_loss": 0.017269209027290344, "grad_norm": 0.1957724690437317, "global_step": 18520, "epoch": 174, "lr": 7.277607273909376e-05} +{"train_loss": 0.02065175399184227, "grad_norm": 0.07792770862579346, "global_step": 18521, "epoch": 174, "lr": 7.277343428048945e-05} +{"train_loss": 0.014758896082639694, "grad_norm": 0.1387569159269333, "global_step": 18522, "epoch": 174, "lr": 7.27707957418692e-05} +{"train_loss": 0.01570097729563713, "grad_norm": 0.11151820421218872, "global_step": 18523, "epoch": 174, "lr": 7.276815712324229e-05} +{"train_loss": 0.015335019677877426, "grad_norm": 0.07277459651231766, "global_step": 18524, "epoch": 174, "lr": 7.276551842461796e-05} +{"train_loss": 0.01707277074456215, "grad_norm": 0.0485420748591423, "global_step": 18525, "epoch": 174, "lr": 7.27628796460055e-05} +{"train_loss": 0.0171835757791996, "grad_norm": 0.1888631135225296, "global_step": 18526, "epoch": 174, "lr": 7.276024078741419e-05} +{"train_loss": 0.013997969217598438, "grad_norm": 0.06130818650126457, "global_step": 18527, "epoch": 174, "lr": 7.27576018488533e-05} +{"train_loss": 0.017885897308588028, "grad_norm": 0.11179971694946289, "global_step": 18528, "epoch": 174, "lr": 7.275496283033207e-05} +{"train_loss": 0.01674620807170868, "grad_norm": 0.04599260166287422, "global_step": 18529, "epoch": 174, "lr": 7.275232373185981e-05} +{"train_loss": 0.01638825051486492, "grad_norm": 0.0601484440267086, "global_step": 18530, "epoch": 174, "lr": 7.274968455344578e-05} +{"train_loss": 0.015742849558591843, "grad_norm": 0.075681172311306, "global_step": 18531, "epoch": 174, "lr": 7.274704529509924e-05} +{"train_loss": 0.019055543467402458, "grad_norm": 0.0793040543794632, "global_step": 18532, "epoch": 174, "lr": 7.274440595682949e-05} +{"train_loss": 0.016618236899375916, "grad_norm": 0.060079555958509445, "global_step": 18533, "epoch": 174, "lr": 7.274176653864578e-05} +{"train_loss": 0.016189273446798325, "grad_norm": 0.08659074455499649, "global_step": 18534, "epoch": 174, "lr": 7.273912704055739e-05} +{"train_loss": 0.017687812447547913, "grad_norm": 0.060933779925107956, "global_step": 18535, "epoch": 174, "lr": 7.27364874625736e-05} +{"train_loss": 0.015208559110760689, "grad_norm": 0.07441679388284683, "global_step": 18536, "epoch": 174, "lr": 7.273384780470369e-05} +{"train_loss": 0.018995940685272217, "grad_norm": 0.07384993880987167, "global_step": 18537, "epoch": 174, "lr": 7.273120806695693e-05} +{"train_loss": 0.016743676736950874, "grad_norm": 0.05604163929820061, "global_step": 18538, "epoch": 174, "lr": 7.272856824934257e-05} +{"train_loss": 0.020252004265785217, "grad_norm": 0.06732766330242157, "global_step": 18539, "epoch": 174, "lr": 7.272592835186991e-05} +{"train_loss": 0.018186315894126892, "grad_norm": 0.059217024594545364, "global_step": 18540, "epoch": 174, "lr": 7.272328837454823e-05} +{"train_loss": 0.015317080542445183, "grad_norm": 0.052163951098918915, "global_step": 18541, "epoch": 174, "lr": 7.272064831738679e-05} +{"train_loss": 0.017869969829916954, "grad_norm": 0.060108400881290436, "global_step": 18542, "epoch": 174, "lr": 7.271800818039487e-05} +{"train_loss": 0.01654599793255329, "grad_norm": 0.12519676983356476, "global_step": 18543, "epoch": 174, "lr": 7.271536796358176e-05} +{"train_loss": 0.02151569351553917, "grad_norm": 0.07138758897781372, "global_step": 18544, "epoch": 174, "lr": 7.271272766695671e-05} +{"train_loss": 0.016685333102941513, "grad_norm": 0.08811502903699875, "global_step": 18545, "epoch": 174, "lr": 7.271008729052901e-05} +{"train_loss": 0.016390517354011536, "grad_norm": 0.09020765125751495, "global_step": 18546, "epoch": 174, "lr": 7.270744683430796e-05} +{"train_loss": 0.016265729442238808, "grad_norm": 0.0703611969947815, "global_step": 18547, "epoch": 174, "lr": 7.270480629830281e-05} +{"train_loss": 0.015129434876143932, "grad_norm": 0.06384117901325226, "global_step": 18548, "epoch": 174, "lr": 7.270216568252285e-05} +{"train_loss": 0.016261235529662302, "grad_norm": 0.11034170538187027, "global_step": 18549, "epoch": 174, "lr": 7.269952498697734e-05, "val_loss": 0.019302459433674812} +{"train_loss": 0.017193973064422607, "grad_norm": 0.12051160633563995, "global_step": 18550, "epoch": 175, "lr": 7.269688421167558e-05} +{"train_loss": 0.016635695472359657, "grad_norm": 0.1525113433599472, "global_step": 18551, "epoch": 175, "lr": 7.269424335662685e-05} +{"train_loss": 0.01996823213994503, "grad_norm": 0.09083741158246994, "global_step": 18552, "epoch": 175, "lr": 7.26916024218404e-05} +{"train_loss": 0.01536624040454626, "grad_norm": 0.0717390924692154, "global_step": 18553, "epoch": 175, "lr": 7.268896140732553e-05} +{"train_loss": 0.014973508194088936, "grad_norm": 0.06556455045938492, "global_step": 18554, "epoch": 175, "lr": 7.268632031309151e-05} +{"train_loss": 0.01371987909078598, "grad_norm": 0.1673453003168106, "global_step": 18555, "epoch": 175, "lr": 7.268367913914764e-05} +{"train_loss": 0.01304683554917574, "grad_norm": 0.07296346873044968, "global_step": 18556, "epoch": 175, "lr": 7.268103788550318e-05} +{"train_loss": 0.0156672615557909, "grad_norm": 0.16420875489711761, "global_step": 18557, "epoch": 175, "lr": 7.267839655216742e-05} +{"train_loss": 0.014136788435280323, "grad_norm": 0.07111649215221405, "global_step": 18558, "epoch": 175, "lr": 7.267575513914963e-05} +{"train_loss": 0.018796904012560844, "grad_norm": 0.08537944406270981, "global_step": 18559, "epoch": 175, "lr": 7.26731136464591e-05} +{"train_loss": 0.01829850673675537, "grad_norm": 0.149770125746727, "global_step": 18560, "epoch": 175, "lr": 7.267047207410513e-05} +{"train_loss": 0.018053187057375908, "grad_norm": 0.10209164023399353, "global_step": 18561, "epoch": 175, "lr": 7.266783042209695e-05} +{"train_loss": 0.014448708854615688, "grad_norm": 0.07075591385364532, "global_step": 18562, "epoch": 175, "lr": 7.266518869044388e-05} +{"train_loss": 0.013193601742386818, "grad_norm": 0.10048487037420273, "global_step": 18563, "epoch": 175, "lr": 7.26625468791552e-05} +{"train_loss": 0.01672343537211418, "grad_norm": 0.09084806591272354, "global_step": 18564, "epoch": 175, "lr": 7.265990498824019e-05} +{"train_loss": 0.01658373884856701, "grad_norm": 0.09326938539743423, "global_step": 18565, "epoch": 175, "lr": 7.265726301770812e-05} +{"train_loss": 0.015715962275862694, "grad_norm": 0.13444142043590546, "global_step": 18566, "epoch": 175, "lr": 7.265462096756827e-05} +{"train_loss": 0.018316997215151787, "grad_norm": 0.1337110698223114, "global_step": 18567, "epoch": 175, "lr": 7.265197883782997e-05} +{"train_loss": 0.019682738929986954, "grad_norm": 0.10549133270978928, "global_step": 18568, "epoch": 175, "lr": 7.264933662850245e-05} +{"train_loss": 0.015354778617620468, "grad_norm": 0.07439526170492172, "global_step": 18569, "epoch": 175, "lr": 7.264669433959498e-05} +{"train_loss": 0.014009958133101463, "grad_norm": 0.05074336379766464, "global_step": 18570, "epoch": 175, "lr": 7.264405197111692e-05} +{"train_loss": 0.01591426134109497, "grad_norm": 0.07256931811571121, "global_step": 18571, "epoch": 175, "lr": 7.264140952307749e-05} +{"train_loss": 0.018554024398326874, "grad_norm": 0.11364999413490295, "global_step": 18572, "epoch": 175, "lr": 7.263876699548598e-05} +{"train_loss": 0.01563054881989956, "grad_norm": 0.06638867408037186, "global_step": 18573, "epoch": 175, "lr": 7.263612438835171e-05} +{"train_loss": 0.014895778149366379, "grad_norm": 0.06138085946440697, "global_step": 18574, "epoch": 175, "lr": 7.263348170168393e-05} +{"train_loss": 0.016198204830288887, "grad_norm": 0.06955757737159729, "global_step": 18575, "epoch": 175, "lr": 7.263083893549195e-05} +{"train_loss": 0.015886936336755753, "grad_norm": 0.23103728890419006, "global_step": 18576, "epoch": 175, "lr": 7.262819608978504e-05} +{"train_loss": 0.015692319720983505, "grad_norm": 0.06282523274421692, "global_step": 18577, "epoch": 175, "lr": 7.262555316457247e-05} +{"train_loss": 0.015903659164905548, "grad_norm": 0.06761510670185089, "global_step": 18578, "epoch": 175, "lr": 7.262291015986358e-05} +{"train_loss": 0.017238672822713852, "grad_norm": 0.07090338319540024, "global_step": 18579, "epoch": 175, "lr": 7.26202670756676e-05} +{"train_loss": 0.016371939331293106, "grad_norm": 0.1671333760023117, "global_step": 18580, "epoch": 175, "lr": 7.261762391199385e-05} +{"train_loss": 0.018316837027668953, "grad_norm": 0.05657024309039116, "global_step": 18581, "epoch": 175, "lr": 7.261498066885158e-05} +{"train_loss": 0.014002682641148567, "grad_norm": 0.057270508259534836, "global_step": 18582, "epoch": 175, "lr": 7.261233734625014e-05} +{"train_loss": 0.014912581071257591, "grad_norm": 0.046675000339746475, "global_step": 18583, "epoch": 175, "lr": 7.260969394419876e-05} +{"train_loss": 0.018518434837460518, "grad_norm": 0.06420528143644333, "global_step": 18584, "epoch": 175, "lr": 7.260705046270674e-05} +{"train_loss": 0.016165848821401596, "grad_norm": 0.06095227226614952, "global_step": 18585, "epoch": 175, "lr": 7.260440690178338e-05} +{"train_loss": 0.014868349768221378, "grad_norm": 0.06658191978931427, "global_step": 18586, "epoch": 175, "lr": 7.260176326143799e-05} +{"train_loss": 0.01537681370973587, "grad_norm": 0.0579698383808136, "global_step": 18587, "epoch": 175, "lr": 7.259911954167982e-05} +{"train_loss": 0.014760082587599754, "grad_norm": 0.06315580755472183, "global_step": 18588, "epoch": 175, "lr": 7.259647574251816e-05} +{"train_loss": 0.01828436367213726, "grad_norm": 0.0900505855679512, "global_step": 18589, "epoch": 175, "lr": 7.259383186396232e-05} +{"train_loss": 0.018443213775753975, "grad_norm": 0.10055259615182877, "global_step": 18590, "epoch": 175, "lr": 7.259118790602158e-05} +{"train_loss": 0.014644443988800049, "grad_norm": 0.14137382805347443, "global_step": 18591, "epoch": 175, "lr": 7.258854386870522e-05} +{"train_loss": 0.016074620187282562, "grad_norm": 0.0730765089392662, "global_step": 18592, "epoch": 175, "lr": 7.258589975202255e-05} +{"train_loss": 0.018463963642716408, "grad_norm": 0.1475139558315277, "global_step": 18593, "epoch": 175, "lr": 7.258325555598285e-05} +{"train_loss": 0.017350178211927414, "grad_norm": 0.09772112965583801, "global_step": 18594, "epoch": 175, "lr": 7.258061128059541e-05} +{"train_loss": 0.017976734787225723, "grad_norm": 0.08461809158325195, "global_step": 18595, "epoch": 175, "lr": 7.257796692586951e-05} +{"train_loss": 0.016691356897354126, "grad_norm": 0.09886153787374496, "global_step": 18596, "epoch": 175, "lr": 7.257532249181448e-05} +{"train_loss": 0.018431495875120163, "grad_norm": 0.098558709025383, "global_step": 18597, "epoch": 175, "lr": 7.257267797843958e-05} +{"train_loss": 0.015070242807269096, "grad_norm": 0.06918570399284363, "global_step": 18598, "epoch": 175, "lr": 7.257003338575407e-05} +{"train_loss": 0.014613919891417027, "grad_norm": 0.13197404146194458, "global_step": 18599, "epoch": 175, "lr": 7.256738871376732e-05} +{"train_loss": 0.012762550264596939, "grad_norm": 0.10165735334157944, "global_step": 18600, "epoch": 175, "lr": 7.256474396248856e-05} +{"train_loss": 0.016195332631468773, "grad_norm": 0.11914307624101639, "global_step": 18601, "epoch": 175, "lr": 7.256209913192711e-05} +{"train_loss": 0.016523677855730057, "grad_norm": 0.09804949909448624, "global_step": 18602, "epoch": 175, "lr": 7.255945422209227e-05} +{"train_loss": 0.015474186278879642, "grad_norm": 0.20153705775737762, "global_step": 18603, "epoch": 175, "lr": 7.255680923299328e-05} +{"train_loss": 0.01542920432984829, "grad_norm": 0.08264955133199692, "global_step": 18604, "epoch": 175, "lr": 7.255416416463951e-05} +{"train_loss": 0.017056673765182495, "grad_norm": 0.0602865032851696, "global_step": 18605, "epoch": 175, "lr": 7.255151901704019e-05} +{"train_loss": 0.018617939203977585, "grad_norm": 0.07913735508918762, "global_step": 18606, "epoch": 175, "lr": 7.254887379020466e-05} +{"train_loss": 0.01767498254776001, "grad_norm": 0.06565253436565399, "global_step": 18607, "epoch": 175, "lr": 7.254622848414218e-05} +{"train_loss": 0.0188094861805439, "grad_norm": 0.0577542744576931, "global_step": 18608, "epoch": 175, "lr": 7.254358309886207e-05} +{"train_loss": 0.019056372344493866, "grad_norm": 0.08099815249443054, "global_step": 18609, "epoch": 175, "lr": 7.254093763437359e-05} +{"train_loss": 0.014719056896865368, "grad_norm": 0.09327663481235504, "global_step": 18610, "epoch": 175, "lr": 7.253829209068607e-05} +{"train_loss": 0.014616238884627819, "grad_norm": 0.05592801794409752, "global_step": 18611, "epoch": 175, "lr": 7.25356464678088e-05} +{"train_loss": 0.018782129511237144, "grad_norm": 0.07364587485790253, "global_step": 18612, "epoch": 175, "lr": 7.253300076575107e-05} +{"train_loss": 0.013878440484404564, "grad_norm": 0.04762377217411995, "global_step": 18613, "epoch": 175, "lr": 7.253035498452216e-05} +{"train_loss": 0.014501459896564484, "grad_norm": 0.050535090267658234, "global_step": 18614, "epoch": 175, "lr": 7.25277091241314e-05} +{"train_loss": 0.01891901344060898, "grad_norm": 0.13806572556495667, "global_step": 18615, "epoch": 175, "lr": 7.252506318458804e-05} +{"train_loss": 0.014580994844436646, "grad_norm": 0.07480789721012115, "global_step": 18616, "epoch": 175, "lr": 7.252241716590144e-05} +{"train_loss": 0.016713976860046387, "grad_norm": 0.06526550650596619, "global_step": 18617, "epoch": 175, "lr": 7.251977106808083e-05} +{"train_loss": 0.014470821246504784, "grad_norm": 0.05685603246092796, "global_step": 18618, "epoch": 175, "lr": 7.251712489113556e-05} +{"train_loss": 0.01807810366153717, "grad_norm": 0.059335287660360336, "global_step": 18619, "epoch": 175, "lr": 7.251447863507488e-05} +{"train_loss": 0.015028007328510284, "grad_norm": 0.10533442348241806, "global_step": 18620, "epoch": 175, "lr": 7.251183229990815e-05} +{"train_loss": 0.015436660498380661, "grad_norm": 0.11270257830619812, "global_step": 18621, "epoch": 175, "lr": 7.25091858856446e-05} +{"train_loss": 0.014071607030928135, "grad_norm": 0.05317600816488266, "global_step": 18622, "epoch": 175, "lr": 7.250653939229357e-05} +{"train_loss": 0.01787605695426464, "grad_norm": 0.13502904772758484, "global_step": 18623, "epoch": 175, "lr": 7.250389281986435e-05} +{"train_loss": 0.015969527885317802, "grad_norm": 0.10064397007226944, "global_step": 18624, "epoch": 175, "lr": 7.250124616836624e-05} +{"train_loss": 0.01689278334379196, "grad_norm": 0.10136493295431137, "global_step": 18625, "epoch": 175, "lr": 7.249859943780852e-05} +{"train_loss": 0.015317810699343681, "grad_norm": 0.15889990329742432, "global_step": 18626, "epoch": 175, "lr": 7.249595262820053e-05} +{"train_loss": 0.01635606959462166, "grad_norm": 0.08752106130123138, "global_step": 18627, "epoch": 175, "lr": 7.249330573955153e-05} +{"train_loss": 0.018646692857146263, "grad_norm": 0.09112899005413055, "global_step": 18628, "epoch": 175, "lr": 7.249065877187084e-05} +{"train_loss": 0.017868002876639366, "grad_norm": 0.09912807494401932, "global_step": 18629, "epoch": 175, "lr": 7.248801172516775e-05} +{"train_loss": 0.016621587797999382, "grad_norm": 0.08371433615684509, "global_step": 18630, "epoch": 175, "lr": 7.248536459945158e-05} +{"train_loss": 0.017663173377513885, "grad_norm": 0.08405278623104095, "global_step": 18631, "epoch": 175, "lr": 7.248271739473161e-05} +{"train_loss": 0.014043359085917473, "grad_norm": 0.1647995412349701, "global_step": 18632, "epoch": 175, "lr": 7.248007011101714e-05} +{"train_loss": 0.01661285012960434, "grad_norm": 0.1171707734465599, "global_step": 18633, "epoch": 175, "lr": 7.247742274831749e-05} +{"train_loss": 0.01764495298266411, "grad_norm": 0.08843111991882324, "global_step": 18634, "epoch": 175, "lr": 7.247477530664194e-05} +{"train_loss": 0.0171674694865942, "grad_norm": 0.10881446301937103, "global_step": 18635, "epoch": 175, "lr": 7.247212778599983e-05} +{"train_loss": 0.01712896302342415, "grad_norm": 0.06589820981025696, "global_step": 18636, "epoch": 175, "lr": 7.246948018640041e-05} +{"train_loss": 0.014391396194696426, "grad_norm": 0.04351877048611641, "global_step": 18637, "epoch": 175, "lr": 7.246683250785303e-05} +{"train_loss": 0.016892047598958015, "grad_norm": 0.10761674493551254, "global_step": 18638, "epoch": 175, "lr": 7.246418475036696e-05} +{"train_loss": 0.01903941109776497, "grad_norm": 0.1036924421787262, "global_step": 18639, "epoch": 175, "lr": 7.24615369139515e-05} +{"train_loss": 0.01577426865696907, "grad_norm": 0.10499453544616699, "global_step": 18640, "epoch": 175, "lr": 7.245888899861599e-05} +{"train_loss": 0.019115224480628967, "grad_norm": 0.10631891340017319, "global_step": 18641, "epoch": 175, "lr": 7.24562410043697e-05} +{"train_loss": 0.015842298045754433, "grad_norm": 0.08167265355587006, "global_step": 18642, "epoch": 175, "lr": 7.245359293122194e-05} +{"train_loss": 0.014204330742359161, "grad_norm": 0.10760277509689331, "global_step": 18643, "epoch": 175, "lr": 7.245094477918203e-05} +{"train_loss": 0.01618017815053463, "grad_norm": 0.06862956285476685, "global_step": 18644, "epoch": 175, "lr": 7.244829654825925e-05} +{"train_loss": 0.01634030230343342, "grad_norm": 0.07747147977352142, "global_step": 18645, "epoch": 175, "lr": 7.244564823846294e-05} +{"train_loss": 0.013094591908156872, "grad_norm": 0.1081426739692688, "global_step": 18646, "epoch": 175, "lr": 7.244299984980237e-05} +{"train_loss": 0.015397503972053528, "grad_norm": 0.13004983961582184, "global_step": 18647, "epoch": 175, "lr": 7.244035138228686e-05} +{"train_loss": 0.01672535389661789, "grad_norm": 0.08706259727478027, "global_step": 18648, "epoch": 175, "lr": 7.243770283592571e-05} +{"train_loss": 0.015764687210321426, "grad_norm": 0.05181315541267395, "global_step": 18649, "epoch": 175, "lr": 7.243505421072824e-05} +{"train_loss": 0.01676955819129944, "grad_norm": 0.14113853871822357, "global_step": 18650, "epoch": 175, "lr": 7.243240550670374e-05} +{"train_loss": 0.015191501937806606, "grad_norm": 0.056047532707452774, "global_step": 18651, "epoch": 175, "lr": 7.242975672386152e-05} +{"train_loss": 0.018269315361976624, "grad_norm": 0.13360726833343506, "global_step": 18652, "epoch": 175, "lr": 7.242710786221089e-05} +{"train_loss": 0.015225129202008247, "grad_norm": 0.09064898639917374, "global_step": 18653, "epoch": 175, "lr": 7.242445892176116e-05} +{"train_loss": 0.015037468634545803, "grad_norm": 0.05629544332623482, "global_step": 18654, "epoch": 175, "lr": 7.242180990252162e-05} +{"train_loss": 0.016301319082179723, "grad_norm": 0.16295340657234192, "global_step": 18655, "epoch": 175, "lr": 7.241916080450163e-05, "val_loss": 0.019471295177936554} +{"train_loss": 0.01450920570641756, "grad_norm": 0.1176137700676918, "global_step": 18656, "epoch": 176, "lr": 7.241651162771042e-05} +{"train_loss": 0.017358096316456795, "grad_norm": 0.08380962908267975, "global_step": 18657, "epoch": 176, "lr": 7.241386237215735e-05} +{"train_loss": 0.016070649027824402, "grad_norm": 0.1815394014120102, "global_step": 18658, "epoch": 176, "lr": 7.24112130378517e-05} +{"train_loss": 0.0179342832416296, "grad_norm": 0.07429588586091995, "global_step": 18659, "epoch": 176, "lr": 7.240856362480281e-05} +{"train_loss": 0.017073864117264748, "grad_norm": 0.059566810727119446, "global_step": 18660, "epoch": 176, "lr": 7.240591413301998e-05} +{"train_loss": 0.016411039978265762, "grad_norm": 0.06375931203365326, "global_step": 18661, "epoch": 176, "lr": 7.240326456251251e-05} +{"train_loss": 0.015173880383372307, "grad_norm": 0.08025693148374557, "global_step": 18662, "epoch": 176, "lr": 7.24006149132897e-05} +{"train_loss": 0.01710551418364048, "grad_norm": 0.052830833941698074, "global_step": 18663, "epoch": 176, "lr": 7.239796518536087e-05} +{"train_loss": 0.015434637665748596, "grad_norm": 0.10199614614248276, "global_step": 18664, "epoch": 176, "lr": 7.239531537873533e-05} +{"train_loss": 0.016768530011177063, "grad_norm": 0.07339148968458176, "global_step": 18665, "epoch": 176, "lr": 7.239266549342241e-05} +{"train_loss": 0.016677126288414, "grad_norm": 0.048239704221487045, "global_step": 18666, "epoch": 176, "lr": 7.239001552943138e-05} +{"train_loss": 0.01680561527609825, "grad_norm": 0.055386826395988464, "global_step": 18667, "epoch": 176, "lr": 7.23873654867716e-05} +{"train_loss": 0.015075192786753178, "grad_norm": 0.05457114055752754, "global_step": 18668, "epoch": 176, "lr": 7.238471536545232e-05} +{"train_loss": 0.017264539375901222, "grad_norm": 0.10821101814508438, "global_step": 18669, "epoch": 176, "lr": 7.23820651654829e-05} +{"train_loss": 0.015650205314159393, "grad_norm": 0.08832619339227676, "global_step": 18670, "epoch": 176, "lr": 7.237941488687263e-05} +{"train_loss": 0.018275249749422073, "grad_norm": 0.0992516502737999, "global_step": 18671, "epoch": 176, "lr": 7.237676452963085e-05} +{"train_loss": 0.015088686719536781, "grad_norm": 0.08086522668600082, "global_step": 18672, "epoch": 176, "lr": 7.237411409376683e-05} +{"train_loss": 0.016949623823165894, "grad_norm": 0.1266503781080246, "global_step": 18673, "epoch": 176, "lr": 7.237146357928991e-05} +{"train_loss": 0.013887012377381325, "grad_norm": 0.10039370507001877, "global_step": 18674, "epoch": 176, "lr": 7.23688129862094e-05} +{"train_loss": 0.017832327634096146, "grad_norm": 0.0809464305639267, "global_step": 18675, "epoch": 176, "lr": 7.23661623145346e-05} +{"train_loss": 0.017185188829898834, "grad_norm": 0.07028880715370178, "global_step": 18676, "epoch": 176, "lr": 7.236351156427484e-05} +{"train_loss": 0.015705756843090057, "grad_norm": 0.04966641962528229, "global_step": 18677, "epoch": 176, "lr": 7.236086073543942e-05} +{"train_loss": 0.01602027378976345, "grad_norm": 0.06341423839330673, "global_step": 18678, "epoch": 176, "lr": 7.235820982803766e-05} +{"train_loss": 0.01653849333524704, "grad_norm": 0.08151846379041672, "global_step": 18679, "epoch": 176, "lr": 7.235555884207889e-05} +{"train_loss": 0.014709646813571453, "grad_norm": 0.0737832710146904, "global_step": 18680, "epoch": 176, "lr": 7.235290777757238e-05} +{"train_loss": 0.018298393115401268, "grad_norm": 0.050784382969141006, "global_step": 18681, "epoch": 176, "lr": 7.23502566345275e-05} +{"train_loss": 0.015647104009985924, "grad_norm": 0.1399199366569519, "global_step": 18682, "epoch": 176, "lr": 7.234760541295351e-05} +{"train_loss": 0.016735197976231575, "grad_norm": 0.21455839276313782, "global_step": 18683, "epoch": 176, "lr": 7.234495411285977e-05} +{"train_loss": 0.013751871883869171, "grad_norm": 0.04297490790486336, "global_step": 18684, "epoch": 176, "lr": 7.234230273425558e-05} +{"train_loss": 0.01623351126909256, "grad_norm": 0.05077889934182167, "global_step": 18685, "epoch": 176, "lr": 7.233965127715027e-05} +{"train_loss": 0.015959743410348892, "grad_norm": 0.07106872648000717, "global_step": 18686, "epoch": 176, "lr": 7.233699974155311e-05} +{"train_loss": 0.015818040817975998, "grad_norm": 0.08931416273117065, "global_step": 18687, "epoch": 176, "lr": 7.233434812747348e-05} +{"train_loss": 0.01798141747713089, "grad_norm": 0.09978264570236206, "global_step": 18688, "epoch": 176, "lr": 7.233169643492063e-05} +{"train_loss": 0.016717201098799706, "grad_norm": 0.048592254519462585, "global_step": 18689, "epoch": 176, "lr": 7.232904466390393e-05} +{"train_loss": 0.016270356252789497, "grad_norm": 0.06439266353845596, "global_step": 18690, "epoch": 176, "lr": 7.232639281443267e-05} +{"train_loss": 0.016865279525518417, "grad_norm": 0.049876898527145386, "global_step": 18691, "epoch": 176, "lr": 7.232374088651618e-05} +{"train_loss": 0.017808865755796432, "grad_norm": 0.08685148507356644, "global_step": 18692, "epoch": 176, "lr": 7.232108888016377e-05} +{"train_loss": 0.01674940064549446, "grad_norm": 0.1205131933093071, "global_step": 18693, "epoch": 176, "lr": 7.231843679538477e-05} +{"train_loss": 0.01779954880475998, "grad_norm": 0.09848673641681671, "global_step": 18694, "epoch": 176, "lr": 7.231578463218846e-05} +{"train_loss": 0.015340333804488182, "grad_norm": 0.08711118996143341, "global_step": 18695, "epoch": 176, "lr": 7.231313239058421e-05} +{"train_loss": 0.015553385019302368, "grad_norm": 0.07994382083415985, "global_step": 18696, "epoch": 176, "lr": 7.231048007058133e-05} +{"train_loss": 0.01351027749478817, "grad_norm": 0.10424245148897171, "global_step": 18697, "epoch": 176, "lr": 7.230782767218911e-05} +{"train_loss": 0.01743737980723381, "grad_norm": 0.10296308249235153, "global_step": 18698, "epoch": 176, "lr": 7.230517519541688e-05} +{"train_loss": 0.016377534717321396, "grad_norm": 0.11965806782245636, "global_step": 18699, "epoch": 176, "lr": 7.230252264027397e-05} +{"train_loss": 0.01590816304087639, "grad_norm": 0.07880628108978271, "global_step": 18700, "epoch": 176, "lr": 7.22998700067697e-05} +{"train_loss": 0.01732856221497059, "grad_norm": 0.06819142401218414, "global_step": 18701, "epoch": 176, "lr": 7.229721729491337e-05} +{"train_loss": 0.017172517254948616, "grad_norm": 0.08485452830791473, "global_step": 18702, "epoch": 176, "lr": 7.229456450471433e-05} +{"train_loss": 0.015827622264623642, "grad_norm": 0.04788447171449661, "global_step": 18703, "epoch": 176, "lr": 7.229191163618188e-05} +{"train_loss": 0.0169544480741024, "grad_norm": 0.05816394463181496, "global_step": 18704, "epoch": 176, "lr": 7.228925868932534e-05} +{"train_loss": 0.013645261526107788, "grad_norm": 0.07711812108755112, "global_step": 18705, "epoch": 176, "lr": 7.228660566415406e-05} +{"train_loss": 0.02060806006193161, "grad_norm": 0.07432174682617188, "global_step": 18706, "epoch": 176, "lr": 7.228395256067732e-05} +{"train_loss": 0.014099566265940666, "grad_norm": 0.05912581831216812, "global_step": 18707, "epoch": 176, "lr": 7.228129937890446e-05} +{"train_loss": 0.016835443675518036, "grad_norm": 0.09840301424264908, "global_step": 18708, "epoch": 176, "lr": 7.227864611884483e-05} +{"train_loss": 0.017096348106861115, "grad_norm": 0.07392902672290802, "global_step": 18709, "epoch": 176, "lr": 7.22759927805077e-05} +{"train_loss": 0.017010284587740898, "grad_norm": 0.09453248977661133, "global_step": 18710, "epoch": 176, "lr": 7.227333936390242e-05} +{"train_loss": 0.014631039462983608, "grad_norm": 0.1568458080291748, "global_step": 18711, "epoch": 176, "lr": 7.227068586903832e-05} +{"train_loss": 0.017600029706954956, "grad_norm": 0.04368261992931366, "global_step": 18712, "epoch": 176, "lr": 7.226803229592472e-05} +{"train_loss": 0.013953408226370811, "grad_norm": 0.11848455667495728, "global_step": 18713, "epoch": 176, "lr": 7.226537864457092e-05} +{"train_loss": 0.01562754064798355, "grad_norm": 0.14952734112739563, "global_step": 18714, "epoch": 176, "lr": 7.22627249149863e-05} +{"train_loss": 0.015150617808103561, "grad_norm": 0.0847732201218605, "global_step": 18715, "epoch": 176, "lr": 7.226007110718011e-05} +{"train_loss": 0.015430985018610954, "grad_norm": 0.08327406644821167, "global_step": 18716, "epoch": 176, "lr": 7.225741722116171e-05} +{"train_loss": 0.015016013756394386, "grad_norm": 0.08688166737556458, "global_step": 18717, "epoch": 176, "lr": 7.225476325694046e-05} +{"train_loss": 0.016702361404895782, "grad_norm": 0.12067940086126328, "global_step": 18718, "epoch": 176, "lr": 7.22521092145256e-05} +{"train_loss": 0.013220146298408508, "grad_norm": 0.15796788036823273, "global_step": 18719, "epoch": 176, "lr": 7.224945509392654e-05} +{"train_loss": 0.0176602341234684, "grad_norm": 0.15821358561515808, "global_step": 18720, "epoch": 176, "lr": 7.224680089515258e-05} +{"train_loss": 0.01804368570446968, "grad_norm": 0.19019922614097595, "global_step": 18721, "epoch": 176, "lr": 7.224414661821301e-05} +{"train_loss": 0.014677470549941063, "grad_norm": 0.14455388486385345, "global_step": 18722, "epoch": 176, "lr": 7.224149226311718e-05} +{"train_loss": 0.016103193163871765, "grad_norm": 0.07297620177268982, "global_step": 18723, "epoch": 176, "lr": 7.223883782987444e-05} +{"train_loss": 0.015591160394251347, "grad_norm": 0.07899966835975647, "global_step": 18724, "epoch": 176, "lr": 7.223618331849408e-05} +{"train_loss": 0.016917090862989426, "grad_norm": 0.063531294465065, "global_step": 18725, "epoch": 176, "lr": 7.223352872898545e-05} +{"train_loss": 0.01595614291727543, "grad_norm": 0.07069646567106247, "global_step": 18726, "epoch": 176, "lr": 7.223087406135788e-05} +{"train_loss": 0.014460785314440727, "grad_norm": 0.09797889739274979, "global_step": 18727, "epoch": 176, "lr": 7.222821931562068e-05} +{"train_loss": 0.014842311851680279, "grad_norm": 0.10410202294588089, "global_step": 18728, "epoch": 176, "lr": 7.222556449178317e-05} +{"train_loss": 0.014487597160041332, "grad_norm": 0.0616932176053524, "global_step": 18729, "epoch": 176, "lr": 7.22229095898547e-05} +{"train_loss": 0.019258446991443634, "grad_norm": 0.10834214091300964, "global_step": 18730, "epoch": 176, "lr": 7.22202546098446e-05} +{"train_loss": 0.015355273149907589, "grad_norm": 0.11127952486276627, "global_step": 18731, "epoch": 176, "lr": 7.221759955176218e-05} +{"train_loss": 0.01566185988485813, "grad_norm": 0.10569046437740326, "global_step": 18732, "epoch": 176, "lr": 7.221494441561677e-05} +{"train_loss": 0.016583586111664772, "grad_norm": 0.07386999577283859, "global_step": 18733, "epoch": 176, "lr": 7.22122892014177e-05} +{"train_loss": 0.01898498274385929, "grad_norm": 0.0735740140080452, "global_step": 18734, "epoch": 176, "lr": 7.220963390917432e-05} +{"train_loss": 0.016301490366458893, "grad_norm": 0.058285921812057495, "global_step": 18735, "epoch": 176, "lr": 7.220697853889597e-05} +{"train_loss": 0.015984131023287773, "grad_norm": 0.11182970553636551, "global_step": 18736, "epoch": 176, "lr": 7.220432309059191e-05} +{"train_loss": 0.017874829471111298, "grad_norm": 0.07101241499185562, "global_step": 18737, "epoch": 176, "lr": 7.220166756427154e-05} +{"train_loss": 0.017218925058841705, "grad_norm": 0.10470204055309296, "global_step": 18738, "epoch": 176, "lr": 7.219901195994416e-05} +{"train_loss": 0.015534834004938602, "grad_norm": 0.07253146916627884, "global_step": 18739, "epoch": 176, "lr": 7.21963562776191e-05} +{"train_loss": 0.01579531654715538, "grad_norm": 0.10417411476373672, "global_step": 18740, "epoch": 176, "lr": 7.219370051730571e-05} +{"train_loss": 0.018101418390870094, "grad_norm": 0.14789575338363647, "global_step": 18741, "epoch": 176, "lr": 7.21910446790133e-05} +{"train_loss": 0.01700868457555771, "grad_norm": 0.15175455808639526, "global_step": 18742, "epoch": 176, "lr": 7.218838876275122e-05} +{"train_loss": 0.01443982683122158, "grad_norm": 0.15475796163082123, "global_step": 18743, "epoch": 176, "lr": 7.218573276852877e-05} +{"train_loss": 0.015851397067308426, "grad_norm": 0.06823639571666718, "global_step": 18744, "epoch": 176, "lr": 7.218307669635532e-05} +{"train_loss": 0.016901716589927673, "grad_norm": 0.07812532037496567, "global_step": 18745, "epoch": 176, "lr": 7.218042054624017e-05} +{"train_loss": 0.015604279935359955, "grad_norm": 0.049480825662612915, "global_step": 18746, "epoch": 176, "lr": 7.217776431819269e-05} +{"train_loss": 0.015441169030964375, "grad_norm": 0.09367003291845322, "global_step": 18747, "epoch": 176, "lr": 7.217510801222218e-05} +{"train_loss": 0.017792075872421265, "grad_norm": 0.0757635161280632, "global_step": 18748, "epoch": 176, "lr": 7.217245162833798e-05} +{"train_loss": 0.015308520756661892, "grad_norm": 0.07755634933710098, "global_step": 18749, "epoch": 176, "lr": 7.216979516654943e-05} +{"train_loss": 0.016266029328107834, "grad_norm": 0.09689845889806747, "global_step": 18750, "epoch": 176, "lr": 7.216713862686586e-05} +{"train_loss": 0.01625840738415718, "grad_norm": 0.08493579924106598, "global_step": 18751, "epoch": 176, "lr": 7.216448200929661e-05} +{"train_loss": 0.015999820083379745, "grad_norm": 0.0992971807718277, "global_step": 18752, "epoch": 176, "lr": 7.216182531385101e-05} +{"train_loss": 0.014545470476150513, "grad_norm": 0.04553147032856941, "global_step": 18753, "epoch": 176, "lr": 7.215916854053839e-05} +{"train_loss": 0.01738162711262703, "grad_norm": 0.11982841789722443, "global_step": 18754, "epoch": 176, "lr": 7.215651168936809e-05} +{"train_loss": 0.017112448811531067, "grad_norm": 0.09931080788373947, "global_step": 18755, "epoch": 176, "lr": 7.215385476034944e-05} +{"train_loss": 0.01634884439408779, "grad_norm": 0.06193799525499344, "global_step": 18756, "epoch": 176, "lr": 7.215119775349179e-05} +{"train_loss": 0.016565540805459023, "grad_norm": 0.1452077031135559, "global_step": 18757, "epoch": 176, "lr": 7.214854066880446e-05} +{"train_loss": 0.017575319856405258, "grad_norm": 0.08515390008687973, "global_step": 18758, "epoch": 176, "lr": 7.214588350629679e-05} +{"train_loss": 0.01589212194085121, "grad_norm": 0.05772389471530914, "global_step": 18759, "epoch": 176, "lr": 7.21432262659781e-05} +{"train_loss": 0.016440201550722122, "grad_norm": 0.06476904451847076, "global_step": 18760, "epoch": 176, "lr": 7.214056894785775e-05} +{"train_loss": 0.016277926402904513, "grad_norm": 0.07644776999950409, "global_step": 18761, "epoch": 176, "lr": 7.21379115519451e-05, "val_loss": 0.019628316164016724} +{"train_loss": 0.015282428823411465, "grad_norm": 0.05391703173518181, "global_step": 18762, "epoch": 177, "lr": 7.213525407824942e-05} +{"train_loss": 0.014680782333016396, "grad_norm": 0.057384148240089417, "global_step": 18763, "epoch": 177, "lr": 7.213259652678009e-05} +{"train_loss": 0.015986453741788864, "grad_norm": 0.06889387965202332, "global_step": 18764, "epoch": 177, "lr": 7.212993889754645e-05} +{"train_loss": 0.01547242421656847, "grad_norm": 0.057715315371751785, "global_step": 18765, "epoch": 177, "lr": 7.212728119055781e-05} +{"train_loss": 0.015529867261648178, "grad_norm": 0.08738331496715546, "global_step": 18766, "epoch": 177, "lr": 7.212462340582356e-05} +{"train_loss": 0.016124939545989037, "grad_norm": 0.07398795336484909, "global_step": 18767, "epoch": 177, "lr": 7.212196554335298e-05} +{"train_loss": 0.015486907213926315, "grad_norm": 0.07624501734972, "global_step": 18768, "epoch": 177, "lr": 7.211930760315544e-05} +{"train_loss": 0.016225744038820267, "grad_norm": 0.06113467738032341, "global_step": 18769, "epoch": 177, "lr": 7.211664958524026e-05} +{"train_loss": 0.015672238543629646, "grad_norm": 0.058791909366846085, "global_step": 18770, "epoch": 177, "lr": 7.211399148961681e-05} +{"train_loss": 0.0193116944283247, "grad_norm": 0.1004161685705185, "global_step": 18771, "epoch": 177, "lr": 7.211133331629442e-05} +{"train_loss": 0.01882200315594673, "grad_norm": 0.07715707272291183, "global_step": 18772, "epoch": 177, "lr": 7.210867506528238e-05} +{"train_loss": 0.015927810221910477, "grad_norm": 0.09562600404024124, "global_step": 18773, "epoch": 177, "lr": 7.210601673659012e-05} +{"train_loss": 0.0170269962400198, "grad_norm": 0.06498812884092331, "global_step": 18774, "epoch": 177, "lr": 7.21033583302269e-05} +{"train_loss": 0.01722128689289093, "grad_norm": 0.06089336425065994, "global_step": 18775, "epoch": 177, "lr": 7.210069984620209e-05} +{"train_loss": 0.016803083941340446, "grad_norm": 0.0964447632431984, "global_step": 18776, "epoch": 177, "lr": 7.209804128452505e-05} +{"train_loss": 0.014698367565870285, "grad_norm": 0.14077335596084595, "global_step": 18777, "epoch": 177, "lr": 7.209538264520509e-05} +{"train_loss": 0.020087454468011856, "grad_norm": 0.08589358627796173, "global_step": 18778, "epoch": 177, "lr": 7.209272392825157e-05} +{"train_loss": 0.015203283168375492, "grad_norm": 0.0601675882935524, "global_step": 18779, "epoch": 177, "lr": 7.209006513367383e-05} +{"train_loss": 0.015755929052829742, "grad_norm": 0.08988464623689651, "global_step": 18780, "epoch": 177, "lr": 7.20874062614812e-05} +{"train_loss": 0.013630814850330353, "grad_norm": 0.08547337353229523, "global_step": 18781, "epoch": 177, "lr": 7.208474731168303e-05} +{"train_loss": 0.01776059716939926, "grad_norm": 0.05876963213086128, "global_step": 18782, "epoch": 177, "lr": 7.208208828428866e-05} +{"train_loss": 0.01688864827156067, "grad_norm": 0.04999759793281555, "global_step": 18783, "epoch": 177, "lr": 7.207942917930745e-05} +{"train_loss": 0.017233481630682945, "grad_norm": 0.10322919487953186, "global_step": 18784, "epoch": 177, "lr": 7.20767699967487e-05} +{"train_loss": 0.016344811767339706, "grad_norm": 0.11387596279382706, "global_step": 18785, "epoch": 177, "lr": 7.207411073662182e-05} +{"train_loss": 0.013025677762925625, "grad_norm": 0.12077383697032928, "global_step": 18786, "epoch": 177, "lr": 7.20714513989361e-05} +{"train_loss": 0.019153235480189323, "grad_norm": 0.10743390023708344, "global_step": 18787, "epoch": 177, "lr": 7.206879198370087e-05} +{"train_loss": 0.01888667792081833, "grad_norm": 0.06253093481063843, "global_step": 18788, "epoch": 177, "lr": 7.206613249092554e-05} +{"train_loss": 0.01741613820195198, "grad_norm": 0.08601366728544235, "global_step": 18789, "epoch": 177, "lr": 7.206347292061939e-05} +{"train_loss": 0.016420748084783554, "grad_norm": 0.0688512772321701, "global_step": 18790, "epoch": 177, "lr": 7.206081327279181e-05} +{"train_loss": 0.01773430034518242, "grad_norm": 0.07448091357946396, "global_step": 18791, "epoch": 177, "lr": 7.205815354745212e-05} +{"train_loss": 0.01817471534013748, "grad_norm": 0.07125494629144669, "global_step": 18792, "epoch": 177, "lr": 7.205549374460966e-05} +{"train_loss": 0.015223385766148567, "grad_norm": 0.1285279542207718, "global_step": 18793, "epoch": 177, "lr": 7.20528338642738e-05} +{"train_loss": 0.01655711606144905, "grad_norm": 0.11499086022377014, "global_step": 18794, "epoch": 177, "lr": 7.205017390645387e-05} +{"train_loss": 0.014610106125473976, "grad_norm": 0.0626099556684494, "global_step": 18795, "epoch": 177, "lr": 7.204751387115922e-05} +{"train_loss": 0.017246633768081665, "grad_norm": 0.11730341613292694, "global_step": 18796, "epoch": 177, "lr": 7.204485375839919e-05} +{"train_loss": 0.017885008826851845, "grad_norm": 0.08256564289331436, "global_step": 18797, "epoch": 177, "lr": 7.204219356818314e-05} +{"train_loss": 0.013743963092565536, "grad_norm": 0.05421401932835579, "global_step": 18798, "epoch": 177, "lr": 7.20395333005204e-05} +{"train_loss": 0.015590345486998558, "grad_norm": 0.07691794633865356, "global_step": 18799, "epoch": 177, "lr": 7.203687295542032e-05} +{"train_loss": 0.01737382635474205, "grad_norm": 0.12948669493198395, "global_step": 18800, "epoch": 177, "lr": 7.203421253289225e-05} +{"train_loss": 0.016856703907251358, "grad_norm": 0.05804494023323059, "global_step": 18801, "epoch": 177, "lr": 7.203155203294553e-05} +{"train_loss": 0.012876450084149837, "grad_norm": 0.08198986947536469, "global_step": 18802, "epoch": 177, "lr": 7.202889145558956e-05} +{"train_loss": 0.014427531510591507, "grad_norm": 0.060424838215112686, "global_step": 18803, "epoch": 177, "lr": 7.202623080083361e-05} +{"train_loss": 0.017018795013427734, "grad_norm": 0.07086679339408875, "global_step": 18804, "epoch": 177, "lr": 7.202357006868708e-05} +{"train_loss": 0.014966333284974098, "grad_norm": 0.06085817143321037, "global_step": 18805, "epoch": 177, "lr": 7.20209092591593e-05} +{"train_loss": 0.015508566983044147, "grad_norm": 0.06106641888618469, "global_step": 18806, "epoch": 177, "lr": 7.201824837225961e-05} +{"train_loss": 0.019897621124982834, "grad_norm": 0.08629675209522247, "global_step": 18807, "epoch": 177, "lr": 7.201558740799737e-05} +{"train_loss": 0.016223639249801636, "grad_norm": 0.08144543319940567, "global_step": 18808, "epoch": 177, "lr": 7.201292636638195e-05} +{"train_loss": 0.015018563717603683, "grad_norm": 0.04242207482457161, "global_step": 18809, "epoch": 177, "lr": 7.201026524742266e-05} +{"train_loss": 0.015502242371439934, "grad_norm": 0.061453677713871, "global_step": 18810, "epoch": 177, "lr": 7.200760405112889e-05} +{"train_loss": 0.01570066064596176, "grad_norm": 0.051996950060129166, "global_step": 18811, "epoch": 177, "lr": 7.200494277750995e-05} +{"train_loss": 0.01799270324409008, "grad_norm": 0.0999537855386734, "global_step": 18812, "epoch": 177, "lr": 7.200228142657522e-05} +{"train_loss": 0.014951132237911224, "grad_norm": 0.12045294791460037, "global_step": 18813, "epoch": 177, "lr": 7.199961999833403e-05} +{"train_loss": 0.016337823122739792, "grad_norm": 0.10526672750711441, "global_step": 18814, "epoch": 177, "lr": 7.199695849279576e-05} +{"train_loss": 0.014028964564204216, "grad_norm": 0.10204248875379562, "global_step": 18815, "epoch": 177, "lr": 7.199429690996974e-05} +{"train_loss": 0.0169316828250885, "grad_norm": 0.06356006860733032, "global_step": 18816, "epoch": 177, "lr": 7.199163524986532e-05} +{"train_loss": 0.016400139778852463, "grad_norm": 0.06682019680738449, "global_step": 18817, "epoch": 177, "lr": 7.198897351249188e-05} +{"train_loss": 0.017737381160259247, "grad_norm": 0.06508422642946243, "global_step": 18818, "epoch": 177, "lr": 7.198631169785871e-05} +{"train_loss": 0.013662631623446941, "grad_norm": 0.07760760933160782, "global_step": 18819, "epoch": 177, "lr": 7.198364980597523e-05} +{"train_loss": 0.01613711006939411, "grad_norm": 0.146782785654068, "global_step": 18820, "epoch": 177, "lr": 7.198098783685076e-05} +{"train_loss": 0.016028232872486115, "grad_norm": 0.05374530330300331, "global_step": 18821, "epoch": 177, "lr": 7.197832579049465e-05} +{"train_loss": 0.013556549325585365, "grad_norm": 0.07427959889173508, "global_step": 18822, "epoch": 177, "lr": 7.197566366691626e-05} +{"train_loss": 0.01745186187326908, "grad_norm": 0.06936220824718475, "global_step": 18823, "epoch": 177, "lr": 7.197300146612496e-05} +{"train_loss": 0.016130156815052032, "grad_norm": 0.08091186732053757, "global_step": 18824, "epoch": 177, "lr": 7.197033918813009e-05} +{"train_loss": 0.017657779157161713, "grad_norm": 0.07124114036560059, "global_step": 18825, "epoch": 177, "lr": 7.196767683294098e-05} +{"train_loss": 0.01831986755132675, "grad_norm": 0.06334177404642105, "global_step": 18826, "epoch": 177, "lr": 7.196501440056704e-05} +{"train_loss": 0.015241742134094238, "grad_norm": 0.07435416430234909, "global_step": 18827, "epoch": 177, "lr": 7.196235189101756e-05} +{"train_loss": 0.01469079963862896, "grad_norm": 0.07817361503839493, "global_step": 18828, "epoch": 177, "lr": 7.195968930430193e-05} +{"train_loss": 0.016713649034500122, "grad_norm": 0.061955228447914124, "global_step": 18829, "epoch": 177, "lr": 7.195702664042953e-05} +{"train_loss": 0.018769532442092896, "grad_norm": 0.05920257046818733, "global_step": 18830, "epoch": 177, "lr": 7.195436389940965e-05} +{"train_loss": 0.01736070215702057, "grad_norm": 0.11135722696781158, "global_step": 18831, "epoch": 177, "lr": 7.19517010812517e-05} +{"train_loss": 0.017288798466324806, "grad_norm": 0.0556146577000618, "global_step": 18832, "epoch": 177, "lr": 7.194903818596502e-05} +{"train_loss": 0.016953065991401672, "grad_norm": 0.07983887195587158, "global_step": 18833, "epoch": 177, "lr": 7.194637521355897e-05} +{"train_loss": 0.012797297909855843, "grad_norm": 0.1302722841501236, "global_step": 18834, "epoch": 177, "lr": 7.194371216404289e-05} +{"train_loss": 0.01543494500219822, "grad_norm": 0.12128068506717682, "global_step": 18835, "epoch": 177, "lr": 7.194104903742617e-05} +{"train_loss": 0.015389216132462025, "grad_norm": 0.07614383846521378, "global_step": 18836, "epoch": 177, "lr": 7.193838583371813e-05} +{"train_loss": 0.016636604443192482, "grad_norm": 0.13516508042812347, "global_step": 18837, "epoch": 177, "lr": 7.193572255292814e-05} +{"train_loss": 0.014491953887045383, "grad_norm": 0.0993485376238823, "global_step": 18838, "epoch": 177, "lr": 7.193305919506556e-05} +{"train_loss": 0.013498885557055473, "grad_norm": 0.08005673438310623, "global_step": 18839, "epoch": 177, "lr": 7.193039576013976e-05} +{"train_loss": 0.01866721361875534, "grad_norm": 0.08838843554258347, "global_step": 18840, "epoch": 177, "lr": 7.192773224816007e-05} +{"train_loss": 0.016700003296136856, "grad_norm": 0.07782598584890366, "global_step": 18841, "epoch": 177, "lr": 7.192506865913589e-05} +{"train_loss": 0.016013672575354576, "grad_norm": 0.07153964787721634, "global_step": 18842, "epoch": 177, "lr": 7.192240499307652e-05} +{"train_loss": 0.015280925668776035, "grad_norm": 0.09113515913486481, "global_step": 18843, "epoch": 177, "lr": 7.191974124999139e-05} +{"train_loss": 0.01551206037402153, "grad_norm": 0.09265542775392532, "global_step": 18844, "epoch": 177, "lr": 7.19170774298898e-05} +{"train_loss": 0.01723378524184227, "grad_norm": 0.07555790990591049, "global_step": 18845, "epoch": 177, "lr": 7.191441353278113e-05} +{"train_loss": 0.015227248892188072, "grad_norm": 0.08020582795143127, "global_step": 18846, "epoch": 177, "lr": 7.191174955867474e-05} +{"train_loss": 0.01908339001238346, "grad_norm": 0.07695061713457108, "global_step": 18847, "epoch": 177, "lr": 7.190908550758e-05} +{"train_loss": 0.014755110256373882, "grad_norm": 0.05651555210351944, "global_step": 18848, "epoch": 177, "lr": 7.190642137950625e-05} +{"train_loss": 0.0163309033960104, "grad_norm": 0.06248187646269798, "global_step": 18849, "epoch": 177, "lr": 7.190375717446288e-05} +{"train_loss": 0.015616510063409805, "grad_norm": 0.07640799880027771, "global_step": 18850, "epoch": 177, "lr": 7.190109289245922e-05} +{"train_loss": 0.018144112080335617, "grad_norm": 0.054937172681093216, "global_step": 18851, "epoch": 177, "lr": 7.189842853350464e-05} +{"train_loss": 0.017379026859998703, "grad_norm": 0.07161510735750198, "global_step": 18852, "epoch": 177, "lr": 7.189576409760852e-05} +{"train_loss": 0.016452709212899208, "grad_norm": 0.09825533628463745, "global_step": 18853, "epoch": 177, "lr": 7.18930995847802e-05} +{"train_loss": 0.016237974166870117, "grad_norm": 0.09175196290016174, "global_step": 18854, "epoch": 177, "lr": 7.189043499502904e-05} +{"train_loss": 0.014147168956696987, "grad_norm": 0.05934188887476921, "global_step": 18855, "epoch": 177, "lr": 7.188777032836444e-05} +{"train_loss": 0.0163930281996727, "grad_norm": 0.11263057589530945, "global_step": 18856, "epoch": 177, "lr": 7.188510558479572e-05} +{"train_loss": 0.017463166266679764, "grad_norm": 0.06627923995256424, "global_step": 18857, "epoch": 177, "lr": 7.188244076433223e-05} +{"train_loss": 0.014816707000136375, "grad_norm": 0.06234055757522583, "global_step": 18858, "epoch": 177, "lr": 7.187977586698339e-05} +{"train_loss": 0.017836399376392365, "grad_norm": 0.11984076350927353, "global_step": 18859, "epoch": 177, "lr": 7.187711089275852e-05} +{"train_loss": 0.014971021562814713, "grad_norm": 0.05154555290937424, "global_step": 18860, "epoch": 177, "lr": 7.1874445841667e-05} +{"train_loss": 0.015318326652050018, "grad_norm": 0.09257645905017853, "global_step": 18861, "epoch": 177, "lr": 7.187178071371818e-05} +{"train_loss": 0.015528158284723759, "grad_norm": 0.08389666676521301, "global_step": 18862, "epoch": 177, "lr": 7.186911550892145e-05} +{"train_loss": 0.019182655960321426, "grad_norm": 0.1320812702178955, "global_step": 18863, "epoch": 177, "lr": 7.186645022728615e-05} +{"train_loss": 0.013108570128679276, "grad_norm": 0.08182983845472336, "global_step": 18864, "epoch": 177, "lr": 7.186378486882165e-05} +{"train_loss": 0.01698426343500614, "grad_norm": 0.06996240466833115, "global_step": 18865, "epoch": 177, "lr": 7.186111943353734e-05} +{"train_loss": 0.014115951023995876, "grad_norm": 0.07082536816596985, "global_step": 18866, "epoch": 177, "lr": 7.185845392144255e-05} +{"train_loss": 0.016231997782048188, "grad_norm": 0.11745858192443848, "global_step": 18867, "epoch": 177, "lr": 7.185578833254664e-05, "val_loss": 0.019664563238620758} +{"train_loss": 0.016692664474248886, "grad_norm": 0.11163577437400818, "global_step": 18868, "epoch": 178, "lr": 7.185312266685903e-05} +{"train_loss": 0.017442677170038223, "grad_norm": 0.07142619043588638, "global_step": 18869, "epoch": 178, "lr": 7.185045692438902e-05} +{"train_loss": 0.014520573429763317, "grad_norm": 0.057071518152952194, "global_step": 18870, "epoch": 178, "lr": 7.184779110514604e-05} +{"train_loss": 0.0176188163459301, "grad_norm": 0.06473086029291153, "global_step": 18871, "epoch": 178, "lr": 7.184512520913939e-05} +{"train_loss": 0.014299432747066021, "grad_norm": 0.044176697731018066, "global_step": 18872, "epoch": 178, "lr": 7.184245923637848e-05} +{"train_loss": 0.01494821161031723, "grad_norm": 0.11089885234832764, "global_step": 18873, "epoch": 178, "lr": 7.183979318687266e-05} +{"train_loss": 0.01683684252202511, "grad_norm": 0.06751543283462524, "global_step": 18874, "epoch": 178, "lr": 7.183712706063133e-05} +{"train_loss": 0.015691032633185387, "grad_norm": 0.12784090638160706, "global_step": 18875, "epoch": 178, "lr": 7.183446085766382e-05} +{"train_loss": 0.014588313177227974, "grad_norm": 0.07739827036857605, "global_step": 18876, "epoch": 178, "lr": 7.18317945779795e-05} +{"train_loss": 0.015600945800542831, "grad_norm": 0.06673239916563034, "global_step": 18877, "epoch": 178, "lr": 7.182912822158775e-05} +{"train_loss": 0.018193257972598076, "grad_norm": 0.08237715810537338, "global_step": 18878, "epoch": 178, "lr": 7.182646178849794e-05} +{"train_loss": 0.017332004383206367, "grad_norm": 0.0883978083729744, "global_step": 18879, "epoch": 178, "lr": 7.182379527871945e-05} +{"train_loss": 0.01579352468252182, "grad_norm": 0.055806998163461685, "global_step": 18880, "epoch": 178, "lr": 7.182112869226162e-05} +{"train_loss": 0.016049066558480263, "grad_norm": 0.12872959673404694, "global_step": 18881, "epoch": 178, "lr": 7.181846202913382e-05} +{"train_loss": 0.01628180965781212, "grad_norm": 0.11980856209993362, "global_step": 18882, "epoch": 178, "lr": 7.181579528934545e-05} +{"train_loss": 0.016032138839364052, "grad_norm": 0.0746847614645958, "global_step": 18883, "epoch": 178, "lr": 7.181312847290587e-05} +{"train_loss": 0.01703043282032013, "grad_norm": 0.06721499562263489, "global_step": 18884, "epoch": 178, "lr": 7.181046157982443e-05} +{"train_loss": 0.01592235267162323, "grad_norm": 0.07615113258361816, "global_step": 18885, "epoch": 178, "lr": 7.180779461011052e-05} +{"train_loss": 0.016161950305104256, "grad_norm": 0.06589507311582565, "global_step": 18886, "epoch": 178, "lr": 7.180512756377349e-05} +{"train_loss": 0.017482254654169083, "grad_norm": 0.07326096296310425, "global_step": 18887, "epoch": 178, "lr": 7.180246044082273e-05} +{"train_loss": 0.01938675530254841, "grad_norm": 0.11105457693338394, "global_step": 18888, "epoch": 178, "lr": 7.179979324126762e-05} +{"train_loss": 0.015535663813352585, "grad_norm": 0.05570871755480766, "global_step": 18889, "epoch": 178, "lr": 7.17971259651175e-05} +{"train_loss": 0.016749318689107895, "grad_norm": 0.07056684046983719, "global_step": 18890, "epoch": 178, "lr": 7.179445861238178e-05} +{"train_loss": 0.01584043726325035, "grad_norm": 0.10610367357730865, "global_step": 18891, "epoch": 178, "lr": 7.17917911830698e-05} +{"train_loss": 0.0173837561160326, "grad_norm": 0.06528926640748978, "global_step": 18892, "epoch": 178, "lr": 7.178912367719094e-05} +{"train_loss": 0.01628594473004341, "grad_norm": 0.07493413239717484, "global_step": 18893, "epoch": 178, "lr": 7.178645609475457e-05} +{"train_loss": 0.013331364840269089, "grad_norm": 0.04712996259331703, "global_step": 18894, "epoch": 178, "lr": 7.178378843577007e-05} +{"train_loss": 0.01606965810060501, "grad_norm": 0.06119093298912048, "global_step": 18895, "epoch": 178, "lr": 7.178112070024682e-05} +{"train_loss": 0.01619994267821312, "grad_norm": 0.09281884878873825, "global_step": 18896, "epoch": 178, "lr": 7.177845288819418e-05} +{"train_loss": 0.017557598650455475, "grad_norm": 0.07914693653583527, "global_step": 18897, "epoch": 178, "lr": 7.177578499962152e-05} +{"train_loss": 0.0195297971367836, "grad_norm": 0.0712059736251831, "global_step": 18898, "epoch": 178, "lr": 7.177311703453822e-05} +{"train_loss": 0.015191584825515747, "grad_norm": 0.05503324046730995, "global_step": 18899, "epoch": 178, "lr": 7.177044899295368e-05} +{"train_loss": 0.014790250919759274, "grad_norm": 0.11049871891736984, "global_step": 18900, "epoch": 178, "lr": 7.176778087487722e-05} +{"train_loss": 0.018490780144929886, "grad_norm": 0.08694866299629211, "global_step": 18901, "epoch": 178, "lr": 7.176511268031826e-05} +{"train_loss": 0.016411975026130676, "grad_norm": 0.1411849856376648, "global_step": 18902, "epoch": 178, "lr": 7.176244440928615e-05} +{"train_loss": 0.014660708606243134, "grad_norm": 0.05504973605275154, "global_step": 18903, "epoch": 178, "lr": 7.17597760617903e-05} +{"train_loss": 0.015547960065305233, "grad_norm": 0.08445724099874496, "global_step": 18904, "epoch": 178, "lr": 7.175710763784003e-05} +{"train_loss": 0.018826864659786224, "grad_norm": 0.1475091129541397, "global_step": 18905, "epoch": 178, "lr": 7.175443913744477e-05} +{"train_loss": 0.018899070098996162, "grad_norm": 0.10231011360883713, "global_step": 18906, "epoch": 178, "lr": 7.175177056061385e-05} +{"train_loss": 0.017614461481571198, "grad_norm": 0.054400552064180374, "global_step": 18907, "epoch": 178, "lr": 7.174910190735668e-05} +{"train_loss": 0.016800420358777046, "grad_norm": 0.09230140596628189, "global_step": 18908, "epoch": 178, "lr": 7.174643317768262e-05} +{"train_loss": 0.017625026404857635, "grad_norm": 0.07095473259687424, "global_step": 18909, "epoch": 178, "lr": 7.174376437160106e-05} +{"train_loss": 0.01831584796309471, "grad_norm": 0.08475028723478317, "global_step": 18910, "epoch": 178, "lr": 7.174109548912135e-05} +{"train_loss": 0.01616034284234047, "grad_norm": 0.08051776140928268, "global_step": 18911, "epoch": 178, "lr": 7.17384265302529e-05} +{"train_loss": 0.017461121082305908, "grad_norm": 0.10378621518611908, "global_step": 18912, "epoch": 178, "lr": 7.173575749500507e-05} +{"train_loss": 0.015834344550967216, "grad_norm": 0.09014349430799484, "global_step": 18913, "epoch": 178, "lr": 7.173308838338722e-05} +{"train_loss": 0.015600036829710007, "grad_norm": 0.06947404891252518, "global_step": 18914, "epoch": 178, "lr": 7.173041919540879e-05} +{"train_loss": 0.014405259862542152, "grad_norm": 0.07248342782258987, "global_step": 18915, "epoch": 178, "lr": 7.17277499310791e-05} +{"train_loss": 0.018859945237636566, "grad_norm": 0.13509562611579895, "global_step": 18916, "epoch": 178, "lr": 7.172508059040753e-05} +{"train_loss": 0.014648962765932083, "grad_norm": 0.10083749145269394, "global_step": 18917, "epoch": 178, "lr": 7.172241117340349e-05} +{"train_loss": 0.015659932047128677, "grad_norm": 0.07551869004964828, "global_step": 18918, "epoch": 178, "lr": 7.171974168007633e-05} +{"train_loss": 0.015541617758572102, "grad_norm": 0.1105843111872673, "global_step": 18919, "epoch": 178, "lr": 7.171707211043545e-05} +{"train_loss": 0.014048390090465546, "grad_norm": 0.06469941884279251, "global_step": 18920, "epoch": 178, "lr": 7.171440246449024e-05} +{"train_loss": 0.014752361923456192, "grad_norm": 0.05196024477481842, "global_step": 18921, "epoch": 178, "lr": 7.171173274225005e-05} +{"train_loss": 0.016119450330734253, "grad_norm": 0.10223743319511414, "global_step": 18922, "epoch": 178, "lr": 7.170906294372426e-05} +{"train_loss": 0.014001693576574326, "grad_norm": 0.053785473108291626, "global_step": 18923, "epoch": 178, "lr": 7.170639306892228e-05} +{"train_loss": 0.014857364818453789, "grad_norm": 0.09905202686786652, "global_step": 18924, "epoch": 178, "lr": 7.170372311785347e-05} +{"train_loss": 0.01534336805343628, "grad_norm": 0.05663689225912094, "global_step": 18925, "epoch": 178, "lr": 7.170105309052721e-05} +{"train_loss": 0.01772576943039894, "grad_norm": 0.07061559706926346, "global_step": 18926, "epoch": 178, "lr": 7.169838298695292e-05} +{"train_loss": 0.015832096338272095, "grad_norm": 0.08704593777656555, "global_step": 18927, "epoch": 178, "lr": 7.169571280713992e-05} +{"train_loss": 0.015616016462445259, "grad_norm": 0.10835737735033035, "global_step": 18928, "epoch": 178, "lr": 7.169304255109761e-05} +{"train_loss": 0.01577822118997574, "grad_norm": 0.06563418358564377, "global_step": 18929, "epoch": 178, "lr": 7.16903722188354e-05} +{"train_loss": 0.015565158799290657, "grad_norm": 0.04807443544268608, "global_step": 18930, "epoch": 178, "lr": 7.168770181036265e-05} +{"train_loss": 0.014700963161885738, "grad_norm": 0.09086639434099197, "global_step": 18931, "epoch": 178, "lr": 7.168503132568875e-05} +{"train_loss": 0.015276436693966389, "grad_norm": 0.10355058312416077, "global_step": 18932, "epoch": 178, "lr": 7.168236076482308e-05} +{"train_loss": 0.015191011130809784, "grad_norm": 0.08495555073022842, "global_step": 18933, "epoch": 178, "lr": 7.167969012777503e-05} +{"train_loss": 0.01767430081963539, "grad_norm": 0.08558286726474762, "global_step": 18934, "epoch": 178, "lr": 7.167701941455396e-05} +{"train_loss": 0.016866866499185562, "grad_norm": 0.07056800276041031, "global_step": 18935, "epoch": 178, "lr": 7.167434862516928e-05} +{"train_loss": 0.01645304635167122, "grad_norm": 0.09495746344327927, "global_step": 18936, "epoch": 178, "lr": 7.167167775963037e-05} +{"train_loss": 0.018011033535003662, "grad_norm": 0.07473549991846085, "global_step": 18937, "epoch": 178, "lr": 7.166900681794659e-05} +{"train_loss": 0.017895936965942383, "grad_norm": 0.10428262501955032, "global_step": 18938, "epoch": 178, "lr": 7.166633580012737e-05} +{"train_loss": 0.016680452972650528, "grad_norm": 0.11744314432144165, "global_step": 18939, "epoch": 178, "lr": 7.166366470618206e-05} +{"train_loss": 0.01708761230111122, "grad_norm": 0.05613075569272041, "global_step": 18940, "epoch": 178, "lr": 7.166099353612006e-05} +{"train_loss": 0.017384137958288193, "grad_norm": 0.08194488286972046, "global_step": 18941, "epoch": 178, "lr": 7.165832228995072e-05} +{"train_loss": 0.01638847030699253, "grad_norm": 0.06974878162145615, "global_step": 18942, "epoch": 178, "lr": 7.165565096768348e-05} +{"train_loss": 0.015102429315447807, "grad_norm": 0.04662635922431946, "global_step": 18943, "epoch": 178, "lr": 7.165297956932771e-05} +{"train_loss": 0.014658981934189796, "grad_norm": 0.05307444930076599, "global_step": 18944, "epoch": 178, "lr": 7.165030809489274e-05} +{"train_loss": 0.01755756512284279, "grad_norm": 0.09991639852523804, "global_step": 18945, "epoch": 178, "lr": 7.164763654438804e-05} +{"train_loss": 0.016338447108864784, "grad_norm": 0.10846155136823654, "global_step": 18946, "epoch": 178, "lr": 7.164496491782295e-05} +{"train_loss": 0.014979716390371323, "grad_norm": 0.076153464615345, "global_step": 18947, "epoch": 178, "lr": 7.164229321520688e-05} +{"train_loss": 0.014742376282811165, "grad_norm": 0.0806950107216835, "global_step": 18948, "epoch": 178, "lr": 7.163962143654919e-05} +{"train_loss": 0.01575123891234398, "grad_norm": 0.056792572140693665, "global_step": 18949, "epoch": 178, "lr": 7.163694958185929e-05} +{"train_loss": 0.01473010890185833, "grad_norm": 0.07990724593400955, "global_step": 18950, "epoch": 178, "lr": 7.163427765114654e-05} +{"train_loss": 0.015056359581649303, "grad_norm": 0.11864573508501053, "global_step": 18951, "epoch": 178, "lr": 7.163160564442035e-05} +{"train_loss": 0.015386924147605896, "grad_norm": 0.051546577364206314, "global_step": 18952, "epoch": 178, "lr": 7.162893356169014e-05} +{"train_loss": 0.01631319895386696, "grad_norm": 0.05281418561935425, "global_step": 18953, "epoch": 178, "lr": 7.162626140296523e-05} +{"train_loss": 0.013311664573848248, "grad_norm": 0.07278405874967575, "global_step": 18954, "epoch": 178, "lr": 7.162358916825505e-05} +{"train_loss": 0.017228882759809494, "grad_norm": 0.14730824530124664, "global_step": 18955, "epoch": 178, "lr": 7.162091685756897e-05} +{"train_loss": 0.0179741308093071, "grad_norm": 0.06146959960460663, "global_step": 18956, "epoch": 178, "lr": 7.16182444709164e-05} +{"train_loss": 0.014708147384226322, "grad_norm": 0.0856548324227333, "global_step": 18957, "epoch": 178, "lr": 7.161557200830673e-05} +{"train_loss": 0.015988938510417938, "grad_norm": 0.056871917098760605, "global_step": 18958, "epoch": 178, "lr": 7.161289946974934e-05} +{"train_loss": 0.017795415595173836, "grad_norm": 0.05418357625603676, "global_step": 18959, "epoch": 178, "lr": 7.16102268552536e-05} +{"train_loss": 0.015770243480801582, "grad_norm": 0.11151750385761261, "global_step": 18960, "epoch": 178, "lr": 7.160755416482894e-05} +{"train_loss": 0.013985196128487587, "grad_norm": 0.06401875615119934, "global_step": 18961, "epoch": 178, "lr": 7.160488139848473e-05} +{"train_loss": 0.016250859946012497, "grad_norm": 0.06629817187786102, "global_step": 18962, "epoch": 178, "lr": 7.160220855623036e-05} +{"train_loss": 0.017413821071386337, "grad_norm": 0.09976908564567566, "global_step": 18963, "epoch": 178, "lr": 7.159953563807521e-05} +{"train_loss": 0.01313053723424673, "grad_norm": 0.05743509158492088, "global_step": 18964, "epoch": 178, "lr": 7.159686264402872e-05} +{"train_loss": 0.0163612961769104, "grad_norm": 0.11574781686067581, "global_step": 18965, "epoch": 178, "lr": 7.159418957410023e-05} +{"train_loss": 0.017086293548345566, "grad_norm": 0.0571138821542263, "global_step": 18966, "epoch": 178, "lr": 7.159151642829912e-05} +{"train_loss": 0.01671617478132248, "grad_norm": 0.10425407439470291, "global_step": 18967, "epoch": 178, "lr": 7.158884320663485e-05} +{"train_loss": 0.016685303300619125, "grad_norm": 0.08327899128198624, "global_step": 18968, "epoch": 178, "lr": 7.158616990911675e-05} +{"train_loss": 0.015027906745672226, "grad_norm": 0.04773470759391785, "global_step": 18969, "epoch": 178, "lr": 7.158349653575424e-05} +{"train_loss": 0.01743047684431076, "grad_norm": 0.11307920515537262, "global_step": 18970, "epoch": 178, "lr": 7.158082308655672e-05} +{"train_loss": 0.0166107676923275, "grad_norm": 0.07380907237529755, "global_step": 18971, "epoch": 178, "lr": 7.157814956153357e-05} +{"train_loss": 0.016995133832097054, "grad_norm": 0.08555611968040466, "global_step": 18972, "epoch": 178, "lr": 7.157547596069419e-05} +{"train_loss": 0.01621388607957172, "grad_norm": 0.14662542939186096, "global_step": 18973, "epoch": 178, "lr": 7.157280228404795e-05, "val_loss": 0.019431721419095993} +{"train_loss": 0.015190837904810905, "grad_norm": 0.049532126635313034, "global_step": 18974, "epoch": 179, "lr": 7.157012853160429e-05} +{"train_loss": 0.0155009925365448, "grad_norm": 0.059995703399181366, "global_step": 18975, "epoch": 179, "lr": 7.156745470337254e-05} +{"train_loss": 0.014661792665719986, "grad_norm": 0.05855049565434456, "global_step": 18976, "epoch": 179, "lr": 7.156478079936219e-05} +{"train_loss": 0.015178471803665161, "grad_norm": 0.06899131089448929, "global_step": 18977, "epoch": 179, "lr": 7.156210681958255e-05} +{"train_loss": 0.016817612573504448, "grad_norm": 0.08581149578094482, "global_step": 18978, "epoch": 179, "lr": 7.155943276404304e-05} +{"train_loss": 0.020052114501595497, "grad_norm": 0.18663622438907623, "global_step": 18979, "epoch": 179, "lr": 7.155675863275306e-05} +{"train_loss": 0.01946137845516205, "grad_norm": 0.08650992810726166, "global_step": 18980, "epoch": 179, "lr": 7.1554084425722e-05} +{"train_loss": 0.01694372110068798, "grad_norm": 0.11045939475297928, "global_step": 18981, "epoch": 179, "lr": 7.155141014295927e-05} +{"train_loss": 0.020427614450454712, "grad_norm": 0.0662938803434372, "global_step": 18982, "epoch": 179, "lr": 7.154873578447426e-05} +{"train_loss": 0.01578746736049652, "grad_norm": 0.06427653133869171, "global_step": 18983, "epoch": 179, "lr": 7.154606135027636e-05} +{"train_loss": 0.017352033406496048, "grad_norm": 0.08539792895317078, "global_step": 18984, "epoch": 179, "lr": 7.154338684037498e-05} +{"train_loss": 0.01670154742896557, "grad_norm": 0.08940979093313217, "global_step": 18985, "epoch": 179, "lr": 7.154071225477948e-05} +{"train_loss": 0.013872317969799042, "grad_norm": 0.07384371757507324, "global_step": 18986, "epoch": 179, "lr": 7.153803759349931e-05} +{"train_loss": 0.016305744647979736, "grad_norm": 0.06601084023714066, "global_step": 18987, "epoch": 179, "lr": 7.153536285654384e-05} +{"train_loss": 0.015421216376125813, "grad_norm": 0.06312690675258636, "global_step": 18988, "epoch": 179, "lr": 7.153268804392246e-05} +{"train_loss": 0.014696136116981506, "grad_norm": 0.12662062048912048, "global_step": 18989, "epoch": 179, "lr": 7.15300131556446e-05} +{"train_loss": 0.01720619946718216, "grad_norm": 0.16480569541454315, "global_step": 18990, "epoch": 179, "lr": 7.152733819171962e-05} +{"train_loss": 0.0165842454880476, "grad_norm": 0.10424897074699402, "global_step": 18991, "epoch": 179, "lr": 7.152466315215693e-05} +{"train_loss": 0.01605900563299656, "grad_norm": 0.10891887545585632, "global_step": 18992, "epoch": 179, "lr": 7.152198803696593e-05} +{"train_loss": 0.019593773409724236, "grad_norm": 0.08512328565120697, "global_step": 18993, "epoch": 179, "lr": 7.151931284615605e-05} +{"train_loss": 0.016586534678936005, "grad_norm": 0.08018910884857178, "global_step": 18994, "epoch": 179, "lr": 7.151663757973664e-05} +{"train_loss": 0.01719391718506813, "grad_norm": 0.0717882588505745, "global_step": 18995, "epoch": 179, "lr": 7.151396223771712e-05} +{"train_loss": 0.016837187111377716, "grad_norm": 0.08926547318696976, "global_step": 18996, "epoch": 179, "lr": 7.151128682010693e-05} +{"train_loss": 0.01869140937924385, "grad_norm": 0.11189354956150055, "global_step": 18997, "epoch": 179, "lr": 7.15086113269154e-05} +{"train_loss": 0.014632074162364006, "grad_norm": 0.06477437913417816, "global_step": 18998, "epoch": 179, "lr": 7.150593575815197e-05} +{"train_loss": 0.015708165243268013, "grad_norm": 0.13319066166877747, "global_step": 18999, "epoch": 179, "lr": 7.150326011382604e-05} +{"train_loss": 0.01611616648733616, "grad_norm": 0.07202883809804916, "global_step": 19000, "epoch": 179, "lr": 7.150058439394699e-05} +{"train_loss": 0.01446748711168766, "grad_norm": 0.047752950340509415, "global_step": 19001, "epoch": 179, "lr": 7.149790859852424e-05} +{"train_loss": 0.014756492339074612, "grad_norm": 0.09297160804271698, "global_step": 19002, "epoch": 179, "lr": 7.14952327275672e-05} +{"train_loss": 0.016299929469823837, "grad_norm": 0.057508595287799835, "global_step": 19003, "epoch": 179, "lr": 7.149255678108527e-05} +{"train_loss": 0.014998854137957096, "grad_norm": 0.08539947867393494, "global_step": 19004, "epoch": 179, "lr": 7.14898807590878e-05} +{"train_loss": 0.01623084954917431, "grad_norm": 0.11388752609491348, "global_step": 19005, "epoch": 179, "lr": 7.148720466158429e-05} +{"train_loss": 0.015478197485208511, "grad_norm": 0.07351431250572205, "global_step": 19006, "epoch": 179, "lr": 7.148452848858405e-05} +{"train_loss": 0.017767412588000298, "grad_norm": 0.07368744909763336, "global_step": 19007, "epoch": 179, "lr": 7.148185224009653e-05} +{"train_loss": 0.014964630827307701, "grad_norm": 0.07524997740983963, "global_step": 19008, "epoch": 179, "lr": 7.147917591613113e-05} +{"train_loss": 0.014266900718212128, "grad_norm": 0.07906051725149155, "global_step": 19009, "epoch": 179, "lr": 7.147649951669724e-05} +{"train_loss": 0.015911158174276352, "grad_norm": 0.06388695538043976, "global_step": 19010, "epoch": 179, "lr": 7.147382304180426e-05} +{"train_loss": 0.01615912839770317, "grad_norm": 0.07915978878736496, "global_step": 19011, "epoch": 179, "lr": 7.147114649146163e-05} +{"train_loss": 0.01559879258275032, "grad_norm": 0.07672049105167389, "global_step": 19012, "epoch": 179, "lr": 7.146846986567871e-05} +{"train_loss": 0.017192702740430832, "grad_norm": 0.07495120912790298, "global_step": 19013, "epoch": 179, "lr": 7.146579316446493e-05} +{"train_loss": 0.01401393860578537, "grad_norm": 0.05852208286523819, "global_step": 19014, "epoch": 179, "lr": 7.146311638782969e-05} +{"train_loss": 0.015324009582400322, "grad_norm": 0.04681601747870445, "global_step": 19015, "epoch": 179, "lr": 7.146043953578239e-05} +{"train_loss": 0.016808323562145233, "grad_norm": 0.08612901717424393, "global_step": 19016, "epoch": 179, "lr": 7.145776260833243e-05} +{"train_loss": 0.019079415127635002, "grad_norm": 0.07933183759450912, "global_step": 19017, "epoch": 179, "lr": 7.145508560548924e-05} +{"train_loss": 0.015898466110229492, "grad_norm": 0.054984092712402344, "global_step": 19018, "epoch": 179, "lr": 7.14524085272622e-05} +{"train_loss": 0.017300959676504135, "grad_norm": 0.162425234913826, "global_step": 19019, "epoch": 179, "lr": 7.144973137366072e-05} +{"train_loss": 0.01470857672393322, "grad_norm": 0.05744902417063713, "global_step": 19020, "epoch": 179, "lr": 7.144705414469423e-05} +{"train_loss": 0.01613452658057213, "grad_norm": 0.0749903991818428, "global_step": 19021, "epoch": 179, "lr": 7.14443768403721e-05} +{"train_loss": 0.017546867951750755, "grad_norm": 0.056943245232105255, "global_step": 19022, "epoch": 179, "lr": 7.144169946070376e-05} +{"train_loss": 0.016758227720856667, "grad_norm": 0.06983808428049088, "global_step": 19023, "epoch": 179, "lr": 7.143902200569863e-05} +{"train_loss": 0.013226830400526524, "grad_norm": 0.05405065789818764, "global_step": 19024, "epoch": 179, "lr": 7.143634447536606e-05} +{"train_loss": 0.015014990232884884, "grad_norm": 0.05812976136803627, "global_step": 19025, "epoch": 179, "lr": 7.143366686971554e-05} +{"train_loss": 0.015596807934343815, "grad_norm": 0.09824289381504059, "global_step": 19026, "epoch": 179, "lr": 7.143098918875643e-05} +{"train_loss": 0.015698885545134544, "grad_norm": 0.04920709505677223, "global_step": 19027, "epoch": 179, "lr": 7.142831143249812e-05} +{"train_loss": 0.01631433144211769, "grad_norm": 0.06961601972579956, "global_step": 19028, "epoch": 179, "lr": 7.142563360095007e-05} +{"train_loss": 0.01725246012210846, "grad_norm": 0.0938011184334755, "global_step": 19029, "epoch": 179, "lr": 7.142295569412165e-05} +{"train_loss": 0.015549078583717346, "grad_norm": 0.05246385559439659, "global_step": 19030, "epoch": 179, "lr": 7.142027771202227e-05} +{"train_loss": 0.014392338693141937, "grad_norm": 0.10243464261293411, "global_step": 19031, "epoch": 179, "lr": 7.141759965466135e-05} +{"train_loss": 0.01653420738875866, "grad_norm": 0.07769215106964111, "global_step": 19032, "epoch": 179, "lr": 7.141492152204832e-05} +{"train_loss": 0.017524544149637222, "grad_norm": 0.12287846207618713, "global_step": 19033, "epoch": 179, "lr": 7.141224331419255e-05} +{"train_loss": 0.018481310456991196, "grad_norm": 0.09118539094924927, "global_step": 19034, "epoch": 179, "lr": 7.140956503110348e-05} +{"train_loss": 0.017377950251102448, "grad_norm": 0.1333359181880951, "global_step": 19035, "epoch": 179, "lr": 7.140688667279052e-05} +{"train_loss": 0.01583525910973549, "grad_norm": 0.13765352964401245, "global_step": 19036, "epoch": 179, "lr": 7.140420823926304e-05} +{"train_loss": 0.018013257533311844, "grad_norm": 0.08280648291110992, "global_step": 19037, "epoch": 179, "lr": 7.14015297305305e-05} +{"train_loss": 0.013485714793205261, "grad_norm": 0.10920998454093933, "global_step": 19038, "epoch": 179, "lr": 7.139885114660228e-05} +{"train_loss": 0.017994321882724762, "grad_norm": 0.058836787939071655, "global_step": 19039, "epoch": 179, "lr": 7.139617248748781e-05} +{"train_loss": 0.014504339545965195, "grad_norm": 0.05016535893082619, "global_step": 19040, "epoch": 179, "lr": 7.139349375319649e-05} +{"train_loss": 0.017996778711676598, "grad_norm": 0.11565674096345901, "global_step": 19041, "epoch": 179, "lr": 7.139081494373774e-05} +{"train_loss": 0.015091324225068092, "grad_norm": 0.07027600705623627, "global_step": 19042, "epoch": 179, "lr": 7.138813605912098e-05} +{"train_loss": 0.02075984701514244, "grad_norm": 0.07914304733276367, "global_step": 19043, "epoch": 179, "lr": 7.138545709935561e-05} +{"train_loss": 0.016079269349575043, "grad_norm": 0.09872991591691971, "global_step": 19044, "epoch": 179, "lr": 7.138277806445103e-05} +{"train_loss": 0.015099014155566692, "grad_norm": 0.10714615136384964, "global_step": 19045, "epoch": 179, "lr": 7.138009895441668e-05} +{"train_loss": 0.015727147459983826, "grad_norm": 0.04587796330451965, "global_step": 19046, "epoch": 179, "lr": 7.137741976926196e-05} +{"train_loss": 0.014420376159250736, "grad_norm": 0.05440555512905121, "global_step": 19047, "epoch": 179, "lr": 7.137474050899627e-05} +{"train_loss": 0.015924308449029922, "grad_norm": 0.11009510606527328, "global_step": 19048, "epoch": 179, "lr": 7.137206117362904e-05} +{"train_loss": 0.015833796933293343, "grad_norm": 0.06238546594977379, "global_step": 19049, "epoch": 179, "lr": 7.13693817631697e-05} +{"train_loss": 0.01776539534330368, "grad_norm": 0.10133811086416245, "global_step": 19050, "epoch": 179, "lr": 7.13667022776276e-05} +{"train_loss": 0.019672231748700142, "grad_norm": 0.10401386767625809, "global_step": 19051, "epoch": 179, "lr": 7.136402271701225e-05} +{"train_loss": 0.014728379435837269, "grad_norm": 0.051900990307331085, "global_step": 19052, "epoch": 179, "lr": 7.136134308133299e-05} +{"train_loss": 0.01674412563443184, "grad_norm": 0.10729220509529114, "global_step": 19053, "epoch": 179, "lr": 7.135866337059926e-05} +{"train_loss": 0.015931496396660805, "grad_norm": 0.08738259971141815, "global_step": 19054, "epoch": 179, "lr": 7.13559835848205e-05} +{"train_loss": 0.016287650913000107, "grad_norm": 0.06615356355905533, "global_step": 19055, "epoch": 179, "lr": 7.135330372400608e-05} +{"train_loss": 0.019343234598636627, "grad_norm": 0.07302268594503403, "global_step": 19056, "epoch": 179, "lr": 7.135062378816544e-05} +{"train_loss": 0.018198929727077484, "grad_norm": 0.12319204956293106, "global_step": 19057, "epoch": 179, "lr": 7.1347943777308e-05} +{"train_loss": 0.01623638905584812, "grad_norm": 0.06564595550298691, "global_step": 19058, "epoch": 179, "lr": 7.134526369144317e-05} +{"train_loss": 0.015881434082984924, "grad_norm": 0.05041612312197685, "global_step": 19059, "epoch": 179, "lr": 7.134258353058035e-05} +{"train_loss": 0.01613743230700493, "grad_norm": 0.09768631309270859, "global_step": 19060, "epoch": 179, "lr": 7.133990329472898e-05} +{"train_loss": 0.015741106122732162, "grad_norm": 0.07357612997293472, "global_step": 19061, "epoch": 179, "lr": 7.133722298389849e-05} +{"train_loss": 0.01643640547990799, "grad_norm": 0.06004089117050171, "global_step": 19062, "epoch": 179, "lr": 7.133454259809824e-05} +{"train_loss": 0.015719421207904816, "grad_norm": 0.1483939290046692, "global_step": 19063, "epoch": 179, "lr": 7.133186213733772e-05} +{"train_loss": 0.012248525395989418, "grad_norm": 0.06784234195947647, "global_step": 19064, "epoch": 179, "lr": 7.132918160162631e-05} +{"train_loss": 0.01525653526186943, "grad_norm": 0.11850013583898544, "global_step": 19065, "epoch": 179, "lr": 7.13265009909734e-05} +{"train_loss": 0.01565680280327797, "grad_norm": 0.09771041572093964, "global_step": 19066, "epoch": 179, "lr": 7.132382030538847e-05} +{"train_loss": 0.015355334617197514, "grad_norm": 0.07458585500717163, "global_step": 19067, "epoch": 179, "lr": 7.13211395448809e-05} +{"train_loss": 0.01732611283659935, "grad_norm": 0.08475930988788605, "global_step": 19068, "epoch": 179, "lr": 7.131845870946013e-05} +{"train_loss": 0.01729932799935341, "grad_norm": 0.12465519458055496, "global_step": 19069, "epoch": 179, "lr": 7.131577779913554e-05} +{"train_loss": 0.016326459124684334, "grad_norm": 0.06972599774599075, "global_step": 19070, "epoch": 179, "lr": 7.13130968139166e-05} +{"train_loss": 0.014642490074038506, "grad_norm": 0.1406615823507309, "global_step": 19071, "epoch": 179, "lr": 7.13104157538127e-05} +{"train_loss": 0.015347328968346119, "grad_norm": 0.10080220550298691, "global_step": 19072, "epoch": 179, "lr": 7.130773461883325e-05} +{"train_loss": 0.014827604405581951, "grad_norm": 0.06368501484394073, "global_step": 19073, "epoch": 179, "lr": 7.130505340898772e-05} +{"train_loss": 0.015066917985677719, "grad_norm": 0.08610755950212479, "global_step": 19074, "epoch": 179, "lr": 7.130237212428548e-05} +{"train_loss": 0.01569148525595665, "grad_norm": 0.05645864084362984, "global_step": 19075, "epoch": 179, "lr": 7.129969076473596e-05} +{"train_loss": 0.014716390520334244, "grad_norm": 0.14497093856334686, "global_step": 19076, "epoch": 179, "lr": 7.129700933034862e-05} +{"train_loss": 0.01625078171491623, "grad_norm": 0.13331370055675507, "global_step": 19077, "epoch": 179, "lr": 7.129432782113281e-05} +{"train_loss": 0.01710038259625435, "grad_norm": 0.11991096287965775, "global_step": 19078, "epoch": 179, "lr": 7.129164623709802e-05} +{"train_loss": 0.01619458552625663, "grad_norm": 0.18216614425182343, "global_step": 19079, "epoch": 179, "lr": 7.128896457825364e-05, "val_loss": 0.018769681453704834} +{"train_loss": 0.02044401690363884, "grad_norm": 0.17879220843315125, "global_step": 19080, "epoch": 180, "lr": 7.128628284460908e-05} +{"train_loss": 0.015955880284309387, "grad_norm": 0.10219783335924149, "global_step": 19081, "epoch": 180, "lr": 7.128360103617381e-05} +{"train_loss": 0.014986351132392883, "grad_norm": 0.0660562589764595, "global_step": 19082, "epoch": 180, "lr": 7.12809191529572e-05} +{"train_loss": 0.014263717457652092, "grad_norm": 0.050943683832883835, "global_step": 19083, "epoch": 180, "lr": 7.12782371949687e-05} +{"train_loss": 0.016943057999014854, "grad_norm": 0.13020703196525574, "global_step": 19084, "epoch": 180, "lr": 7.127555516221772e-05} +{"train_loss": 0.01905660517513752, "grad_norm": 0.059733547270298004, "global_step": 19085, "epoch": 180, "lr": 7.12728730547137e-05} +{"train_loss": 0.014245780184864998, "grad_norm": 0.07121322304010391, "global_step": 19086, "epoch": 180, "lr": 7.127019087246607e-05} +{"train_loss": 0.016850877553224564, "grad_norm": 0.08761443942785263, "global_step": 19087, "epoch": 180, "lr": 7.126750861548421e-05} +{"train_loss": 0.015098460018634796, "grad_norm": 0.06169525533914566, "global_step": 19088, "epoch": 180, "lr": 7.12648262837776e-05} +{"train_loss": 0.017354004085063934, "grad_norm": 0.12656182050704956, "global_step": 19089, "epoch": 180, "lr": 7.126214387735562e-05} +{"train_loss": 0.016431104391813278, "grad_norm": 0.07764746993780136, "global_step": 19090, "epoch": 180, "lr": 7.125946139622772e-05} +{"train_loss": 0.017011530697345734, "grad_norm": 0.10687040537595749, "global_step": 19091, "epoch": 180, "lr": 7.125677884040331e-05} +{"train_loss": 0.01595689356327057, "grad_norm": 0.15895596146583557, "global_step": 19092, "epoch": 180, "lr": 7.125409620989184e-05} +{"train_loss": 0.016485724598169327, "grad_norm": 0.07341539859771729, "global_step": 19093, "epoch": 180, "lr": 7.125141350470273e-05} +{"train_loss": 0.014845544472336769, "grad_norm": 0.060829997062683105, "global_step": 19094, "epoch": 180, "lr": 7.124873072484537e-05} +{"train_loss": 0.013778665103018284, "grad_norm": 0.09358040243387222, "global_step": 19095, "epoch": 180, "lr": 7.124604787032922e-05} +{"train_loss": 0.014378410764038563, "grad_norm": 0.07426774501800537, "global_step": 19096, "epoch": 180, "lr": 7.124336494116368e-05} +{"train_loss": 0.0165808517485857, "grad_norm": 0.05159759148955345, "global_step": 19097, "epoch": 180, "lr": 7.124068193735821e-05} +{"train_loss": 0.01834043487906456, "grad_norm": 0.058718010783195496, "global_step": 19098, "epoch": 180, "lr": 7.123799885892223e-05} +{"train_loss": 0.015900548547506332, "grad_norm": 0.08775558322668076, "global_step": 19099, "epoch": 180, "lr": 7.123531570586514e-05} +{"train_loss": 0.01979237049818039, "grad_norm": 0.10908637940883636, "global_step": 19100, "epoch": 180, "lr": 7.12326324781964e-05} +{"train_loss": 0.0157766230404377, "grad_norm": 0.09587036073207855, "global_step": 19101, "epoch": 180, "lr": 7.12299491759254e-05} +{"train_loss": 0.016594357788562775, "grad_norm": 0.047482915222644806, "global_step": 19102, "epoch": 180, "lr": 7.122726579906162e-05} +{"train_loss": 0.01515316404402256, "grad_norm": 0.1053759977221489, "global_step": 19103, "epoch": 180, "lr": 7.122458234761446e-05} +{"train_loss": 0.015939127653837204, "grad_norm": 0.06267467141151428, "global_step": 19104, "epoch": 180, "lr": 7.122189882159331e-05} +{"train_loss": 0.016110582277178764, "grad_norm": 0.0642332136631012, "global_step": 19105, "epoch": 180, "lr": 7.121921522100768e-05} +{"train_loss": 0.015898341313004494, "grad_norm": 0.06347323209047318, "global_step": 19106, "epoch": 180, "lr": 7.121653154586692e-05} +{"train_loss": 0.0147376274690032, "grad_norm": 0.0714937299489975, "global_step": 19107, "epoch": 180, "lr": 7.121384779618052e-05} +{"train_loss": 0.016397763043642044, "grad_norm": 0.10432908684015274, "global_step": 19108, "epoch": 180, "lr": 7.121116397195789e-05} +{"train_loss": 0.014682806096971035, "grad_norm": 0.087321437895298, "global_step": 19109, "epoch": 180, "lr": 7.120848007320843e-05} +{"train_loss": 0.01686428114771843, "grad_norm": 0.05324516445398331, "global_step": 19110, "epoch": 180, "lr": 7.12057960999416e-05} +{"train_loss": 0.01590670645236969, "grad_norm": 0.05854417383670807, "global_step": 19111, "epoch": 180, "lr": 7.120311205216684e-05} +{"train_loss": 0.016605574637651443, "grad_norm": 0.06377104669809341, "global_step": 19112, "epoch": 180, "lr": 7.120042792989356e-05} +{"train_loss": 0.016952861100435257, "grad_norm": 0.1398850679397583, "global_step": 19113, "epoch": 180, "lr": 7.119774373313117e-05} +{"train_loss": 0.013654261827468872, "grad_norm": 0.11557933688163757, "global_step": 19114, "epoch": 180, "lr": 7.119505946188917e-05} +{"train_loss": 0.017992347478866577, "grad_norm": 0.08826445788145065, "global_step": 19115, "epoch": 180, "lr": 7.119237511617692e-05} +{"train_loss": 0.01771543174982071, "grad_norm": 0.07308859378099442, "global_step": 19116, "epoch": 180, "lr": 7.118969069600389e-05} +{"train_loss": 0.01650271564722061, "grad_norm": 0.07571225613355637, "global_step": 19117, "epoch": 180, "lr": 7.11870062013795e-05} +{"train_loss": 0.014970432966947556, "grad_norm": 0.062396179884672165, "global_step": 19118, "epoch": 180, "lr": 7.118432163231318e-05} +{"train_loss": 0.01765923574566841, "grad_norm": 0.10723896324634552, "global_step": 19119, "epoch": 180, "lr": 7.118163698881436e-05} +{"train_loss": 0.014599980786442757, "grad_norm": 0.10899478197097778, "global_step": 19120, "epoch": 180, "lr": 7.11789522708925e-05} +{"train_loss": 0.017126739025115967, "grad_norm": 0.09470653533935547, "global_step": 19121, "epoch": 180, "lr": 7.1176267478557e-05} +{"train_loss": 0.018351096659898758, "grad_norm": 0.10419037193059921, "global_step": 19122, "epoch": 180, "lr": 7.117358261181732e-05} +{"train_loss": 0.01849270984530449, "grad_norm": 0.10966286063194275, "global_step": 19123, "epoch": 180, "lr": 7.117089767068286e-05} +{"train_loss": 0.015408955514431, "grad_norm": 0.07773690670728683, "global_step": 19124, "epoch": 180, "lr": 7.116821265516307e-05} +{"train_loss": 0.015873435884714127, "grad_norm": 0.07681432366371155, "global_step": 19125, "epoch": 180, "lr": 7.116552756526739e-05} +{"train_loss": 0.015276730060577393, "grad_norm": 0.14639735221862793, "global_step": 19126, "epoch": 180, "lr": 7.116284240100527e-05} +{"train_loss": 0.015207584016025066, "grad_norm": 0.05395141988992691, "global_step": 19127, "epoch": 180, "lr": 7.116015716238611e-05} +{"train_loss": 0.01821684092283249, "grad_norm": 0.12074517458677292, "global_step": 19128, "epoch": 180, "lr": 7.115747184941936e-05} +{"train_loss": 0.015165623277425766, "grad_norm": 0.0668748989701271, "global_step": 19129, "epoch": 180, "lr": 7.115478646211447e-05} +{"train_loss": 0.01705893501639366, "grad_norm": 0.07178962975740433, "global_step": 19130, "epoch": 180, "lr": 7.115210100048083e-05} +{"train_loss": 0.016575124114751816, "grad_norm": 0.06742469221353531, "global_step": 19131, "epoch": 180, "lr": 7.114941546452793e-05} +{"train_loss": 0.016822736710309982, "grad_norm": 0.06995393335819244, "global_step": 19132, "epoch": 180, "lr": 7.114672985426516e-05} +{"train_loss": 0.015621153637766838, "grad_norm": 0.06739194691181183, "global_step": 19133, "epoch": 180, "lr": 7.114404416970199e-05} +{"train_loss": 0.017653794959187508, "grad_norm": 0.10731561481952667, "global_step": 19134, "epoch": 180, "lr": 7.114135841084786e-05} +{"train_loss": 0.016021614894270897, "grad_norm": 0.0427892729640007, "global_step": 19135, "epoch": 180, "lr": 7.113867257771217e-05} +{"train_loss": 0.017013249918818474, "grad_norm": 0.10776402801275253, "global_step": 19136, "epoch": 180, "lr": 7.11359866703044e-05} +{"train_loss": 0.014919018372893333, "grad_norm": 0.0642826035618782, "global_step": 19137, "epoch": 180, "lr": 7.113330068863394e-05} +{"train_loss": 0.016173556447029114, "grad_norm": 0.05501412972807884, "global_step": 19138, "epoch": 180, "lr": 7.113061463271025e-05} +{"train_loss": 0.017406029626727104, "grad_norm": 0.09773740917444229, "global_step": 19139, "epoch": 180, "lr": 7.11279285025428e-05} +{"train_loss": 0.01673031970858574, "grad_norm": 0.1388711780309677, "global_step": 19140, "epoch": 180, "lr": 7.112524229814096e-05} +{"train_loss": 0.013324206694960594, "grad_norm": 0.07264656573534012, "global_step": 19141, "epoch": 180, "lr": 7.112255601951423e-05} +{"train_loss": 0.016311097890138626, "grad_norm": 0.06459119915962219, "global_step": 19142, "epoch": 180, "lr": 7.111986966667202e-05} +{"train_loss": 0.014528103172779083, "grad_norm": 0.06602060794830322, "global_step": 19143, "epoch": 180, "lr": 7.111718323962377e-05} +{"train_loss": 0.015467804856598377, "grad_norm": 0.1133369728922844, "global_step": 19144, "epoch": 180, "lr": 7.111449673837893e-05} +{"train_loss": 0.018305685371160507, "grad_norm": 0.08689813315868378, "global_step": 19145, "epoch": 180, "lr": 7.11118101629469e-05} +{"train_loss": 0.013636372052133083, "grad_norm": 0.059045374393463135, "global_step": 19146, "epoch": 180, "lr": 7.11091235133372e-05} +{"train_loss": 0.01576266810297966, "grad_norm": 0.07534294575452805, "global_step": 19147, "epoch": 180, "lr": 7.110643678955919e-05} +{"train_loss": 0.01697760634124279, "grad_norm": 0.05385621264576912, "global_step": 19148, "epoch": 180, "lr": 7.110374999162234e-05} +{"train_loss": 0.017092974856495857, "grad_norm": 0.08076860010623932, "global_step": 19149, "epoch": 180, "lr": 7.11010631195361e-05} +{"train_loss": 0.016526440158486366, "grad_norm": 0.06057468429207802, "global_step": 19150, "epoch": 180, "lr": 7.109837617330988e-05} +{"train_loss": 0.016842463985085487, "grad_norm": 0.0850454643368721, "global_step": 19151, "epoch": 180, "lr": 7.109568915295317e-05} +{"train_loss": 0.014460070058703423, "grad_norm": 0.08267001807689667, "global_step": 19152, "epoch": 180, "lr": 7.109300205847537e-05} +{"train_loss": 0.017525305971503258, "grad_norm": 0.05428510904312134, "global_step": 19153, "epoch": 180, "lr": 7.109031488988593e-05} +{"train_loss": 0.013938519172370434, "grad_norm": 0.05533749610185623, "global_step": 19154, "epoch": 180, "lr": 7.10876276471943e-05} +{"train_loss": 0.01600027084350586, "grad_norm": 0.08832741528749466, "global_step": 19155, "epoch": 180, "lr": 7.10849403304099e-05} +{"train_loss": 0.015644129365682602, "grad_norm": 0.06667918711900711, "global_step": 19156, "epoch": 180, "lr": 7.108225293954222e-05} +{"train_loss": 0.012826932594180107, "grad_norm": 0.0639663115143776, "global_step": 19157, "epoch": 180, "lr": 7.107956547460064e-05} +{"train_loss": 0.015859706327319145, "grad_norm": 0.07210535556077957, "global_step": 19158, "epoch": 180, "lr": 7.107687793559466e-05} +{"train_loss": 0.016029540449380875, "grad_norm": 0.05299188569188118, "global_step": 19159, "epoch": 180, "lr": 7.107419032253368e-05} +{"train_loss": 0.014810733497142792, "grad_norm": 0.05691737309098244, "global_step": 19160, "epoch": 180, "lr": 7.107150263542717e-05} +{"train_loss": 0.017169784754514694, "grad_norm": 0.07622084766626358, "global_step": 19161, "epoch": 180, "lr": 7.106881487428456e-05} +{"train_loss": 0.01564185321331024, "grad_norm": 0.10793551802635193, "global_step": 19162, "epoch": 180, "lr": 7.10661270391153e-05} +{"train_loss": 0.01617690920829773, "grad_norm": 0.090348981320858, "global_step": 19163, "epoch": 180, "lr": 7.106343912992884e-05} +{"train_loss": 0.015295354649424553, "grad_norm": 0.07694821804761887, "global_step": 19164, "epoch": 180, "lr": 7.10607511467346e-05} +{"train_loss": 0.016488216817378998, "grad_norm": 0.04988277330994606, "global_step": 19165, "epoch": 180, "lr": 7.105806308954205e-05} +{"train_loss": 0.014901401475071907, "grad_norm": 0.05071108415722847, "global_step": 19166, "epoch": 180, "lr": 7.105537495836061e-05} +{"train_loss": 0.0175626240670681, "grad_norm": 0.083278127014637, "global_step": 19167, "epoch": 180, "lr": 7.105268675319976e-05} +{"train_loss": 0.016880664974451065, "grad_norm": 0.10215789079666138, "global_step": 19168, "epoch": 180, "lr": 7.10499984740689e-05} +{"train_loss": 0.014495471492409706, "grad_norm": 0.08137501776218414, "global_step": 19169, "epoch": 180, "lr": 7.104731012097751e-05} +{"train_loss": 0.014112644828855991, "grad_norm": 0.05883600562810898, "global_step": 19170, "epoch": 180, "lr": 7.104462169393504e-05} +{"train_loss": 0.016696227714419365, "grad_norm": 0.07554218173027039, "global_step": 19171, "epoch": 180, "lr": 7.104193319295089e-05} +{"train_loss": 0.016697056591510773, "grad_norm": 0.059462957084178925, "global_step": 19172, "epoch": 180, "lr": 7.103924461803456e-05} +{"train_loss": 0.01594788208603859, "grad_norm": 0.05864858999848366, "global_step": 19173, "epoch": 180, "lr": 7.103655596919549e-05} +{"train_loss": 0.01830659992992878, "grad_norm": 0.12864115834236145, "global_step": 19174, "epoch": 180, "lr": 7.103386724644307e-05} +{"train_loss": 0.016005899757146835, "grad_norm": 0.05808025598526001, "global_step": 19175, "epoch": 180, "lr": 7.103117844978681e-05} +{"train_loss": 0.01598518155515194, "grad_norm": 0.061918530613183975, "global_step": 19176, "epoch": 180, "lr": 7.102848957923613e-05} +{"train_loss": 0.015219045802950859, "grad_norm": 0.06315555423498154, "global_step": 19177, "epoch": 180, "lr": 7.10258006348005e-05} +{"train_loss": 0.01807733066380024, "grad_norm": 0.13775084912776947, "global_step": 19178, "epoch": 180, "lr": 7.102311161648934e-05} +{"train_loss": 0.018406642600893974, "grad_norm": 0.05925130844116211, "global_step": 19179, "epoch": 180, "lr": 7.102042252431211e-05} +{"train_loss": 0.01754618063569069, "grad_norm": 0.05668048560619354, "global_step": 19180, "epoch": 180, "lr": 7.101773335827825e-05} +{"train_loss": 0.016638286411762238, "grad_norm": 0.05248117446899414, "global_step": 19181, "epoch": 180, "lr": 7.10150441183972e-05} +{"train_loss": 0.018767623230814934, "grad_norm": 0.07490672171115875, "global_step": 19182, "epoch": 180, "lr": 7.101235480467847e-05} +{"train_loss": 0.017174050211906433, "grad_norm": 0.07818886637687683, "global_step": 19183, "epoch": 180, "lr": 7.100966541713141e-05} +{"train_loss": 0.01500758808106184, "grad_norm": 0.08639714121818542, "global_step": 19184, "epoch": 180, "lr": 7.100697595576555e-05} +{"train_loss": 0.016230019037875364, "grad_norm": 0.07327330112457275, "global_step": 19185, "epoch": 180, "lr": 7.100428642059033e-05, "train_batch_mse_error": 0.014591365121304989, "val_loss": 0.019131042063236237} +{"train_loss": 0.016257204115390778, "grad_norm": 0.16270458698272705, "global_step": 19186, "epoch": 181, "lr": 7.100159681161515e-05} +{"train_loss": 0.018769625574350357, "grad_norm": 0.108327217400074, "global_step": 19187, "epoch": 181, "lr": 7.09989071288495e-05} +{"train_loss": 0.01456044428050518, "grad_norm": 0.08931905776262283, "global_step": 19188, "epoch": 181, "lr": 7.099621737230282e-05} +{"train_loss": 0.01495848223567009, "grad_norm": 0.04797199368476868, "global_step": 19189, "epoch": 181, "lr": 7.099352754198459e-05} +{"train_loss": 0.019435759633779526, "grad_norm": 0.07646387070417404, "global_step": 19190, "epoch": 181, "lr": 7.099083763790421e-05} +{"train_loss": 0.015387728810310364, "grad_norm": 0.061010994017124176, "global_step": 19191, "epoch": 181, "lr": 7.098814766007116e-05} +{"train_loss": 0.01725328154861927, "grad_norm": 0.06186096742749214, "global_step": 19192, "epoch": 181, "lr": 7.098545760849489e-05} +{"train_loss": 0.015453371219336987, "grad_norm": 0.07454072684049606, "global_step": 19193, "epoch": 181, "lr": 7.098276748318484e-05} +{"train_loss": 0.01657993718981743, "grad_norm": 0.08261886239051819, "global_step": 19194, "epoch": 181, "lr": 7.098007728415048e-05} +{"train_loss": 0.017784854397177696, "grad_norm": 0.08922186493873596, "global_step": 19195, "epoch": 181, "lr": 7.097738701140125e-05} +{"train_loss": 0.017275985330343246, "grad_norm": 0.08392393589019775, "global_step": 19196, "epoch": 181, "lr": 7.09746966649466e-05} +{"train_loss": 0.015948869287967682, "grad_norm": 0.09471192210912704, "global_step": 19197, "epoch": 181, "lr": 7.097200624479599e-05} +{"train_loss": 0.013507556170225143, "grad_norm": 0.13792471587657928, "global_step": 19198, "epoch": 181, "lr": 7.096931575095886e-05} +{"train_loss": 0.016616003587841988, "grad_norm": 0.12150634080171585, "global_step": 19199, "epoch": 181, "lr": 7.096662518344468e-05} +{"train_loss": 0.015767935663461685, "grad_norm": 0.055274661630392075, "global_step": 19200, "epoch": 181, "lr": 7.09639345422629e-05} +{"train_loss": 0.01525081880390644, "grad_norm": 0.07028523087501526, "global_step": 19201, "epoch": 181, "lr": 7.096124382742297e-05} +{"train_loss": 0.017007092013955116, "grad_norm": 0.16990649700164795, "global_step": 19202, "epoch": 181, "lr": 7.095855303893436e-05} +{"train_loss": 0.015448322519659996, "grad_norm": 0.0557214617729187, "global_step": 19203, "epoch": 181, "lr": 7.095586217680648e-05} +{"train_loss": 0.01650114357471466, "grad_norm": 0.057330768555402756, "global_step": 19204, "epoch": 181, "lr": 7.095317124104883e-05} +{"train_loss": 0.016445180401206017, "grad_norm": 0.09330226480960846, "global_step": 19205, "epoch": 181, "lr": 7.095048023167085e-05} +{"train_loss": 0.013637926429510117, "grad_norm": 0.09018056094646454, "global_step": 19206, "epoch": 181, "lr": 7.0947789148682e-05} +{"train_loss": 0.016855942085385323, "grad_norm": 0.07225554436445236, "global_step": 19207, "epoch": 181, "lr": 7.094509799209173e-05} +{"train_loss": 0.015247207134962082, "grad_norm": 0.08714576810598373, "global_step": 19208, "epoch": 181, "lr": 7.094240676190947e-05} +{"train_loss": 0.017571432515978813, "grad_norm": 0.08307145535945892, "global_step": 19209, "epoch": 181, "lr": 7.093971545814473e-05} +{"train_loss": 0.017426980659365654, "grad_norm": 0.08481166511774063, "global_step": 19210, "epoch": 181, "lr": 7.09370240808069e-05} +{"train_loss": 0.01408066414296627, "grad_norm": 0.06023687124252319, "global_step": 19211, "epoch": 181, "lr": 7.093433262990551e-05} +{"train_loss": 0.01975969597697258, "grad_norm": 0.1995469480752945, "global_step": 19212, "epoch": 181, "lr": 7.093164110544995e-05} +{"train_loss": 0.016646889969706535, "grad_norm": 0.1424519121646881, "global_step": 19213, "epoch": 181, "lr": 7.092894950744973e-05} +{"train_loss": 0.018540117889642715, "grad_norm": 0.05460638925433159, "global_step": 19214, "epoch": 181, "lr": 7.092625783591427e-05} +{"train_loss": 0.014903387054800987, "grad_norm": 0.052737459540367126, "global_step": 19215, "epoch": 181, "lr": 7.092356609085304e-05} +{"train_loss": 0.016408195719122887, "grad_norm": 0.12770627439022064, "global_step": 19216, "epoch": 181, "lr": 7.092087427227551e-05} +{"train_loss": 0.018056508153676987, "grad_norm": 0.0905575081706047, "global_step": 19217, "epoch": 181, "lr": 7.091818238019111e-05} +{"train_loss": 0.016287092119455338, "grad_norm": 0.0866829976439476, "global_step": 19218, "epoch": 181, "lr": 7.091549041460933e-05} +{"train_loss": 0.016597401350736618, "grad_norm": 0.1605432629585266, "global_step": 19219, "epoch": 181, "lr": 7.09127983755396e-05} +{"train_loss": 0.016022153198719025, "grad_norm": 0.12384410947561264, "global_step": 19220, "epoch": 181, "lr": 7.091010626299139e-05} +{"train_loss": 0.015159987844526768, "grad_norm": 0.08874468505382538, "global_step": 19221, "epoch": 181, "lr": 7.090741407697418e-05} +{"train_loss": 0.01948212832212448, "grad_norm": 0.0731135830283165, "global_step": 19222, "epoch": 181, "lr": 7.090472181749739e-05} +{"train_loss": 0.0179017074406147, "grad_norm": 0.15455926954746246, "global_step": 19223, "epoch": 181, "lr": 7.090202948457051e-05} +{"train_loss": 0.016716009005904198, "grad_norm": 0.11243489384651184, "global_step": 19224, "epoch": 181, "lr": 7.089933707820298e-05} +{"train_loss": 0.016434045508503914, "grad_norm": 0.08703476935625076, "global_step": 19225, "epoch": 181, "lr": 7.089664459840426e-05} +{"train_loss": 0.01580415479838848, "grad_norm": 0.1193837821483612, "global_step": 19226, "epoch": 181, "lr": 7.089395204518385e-05} +{"train_loss": 0.018957527354359627, "grad_norm": 0.15565550327301025, "global_step": 19227, "epoch": 181, "lr": 7.089125941855114e-05} +{"train_loss": 0.014018593356013298, "grad_norm": 0.11985873430967331, "global_step": 19228, "epoch": 181, "lr": 7.088856671851565e-05} +{"train_loss": 0.013135162182152271, "grad_norm": 0.07014092057943344, "global_step": 19229, "epoch": 181, "lr": 7.088587394508683e-05} +{"train_loss": 0.014123608358204365, "grad_norm": 0.15238472819328308, "global_step": 19230, "epoch": 181, "lr": 7.088318109827412e-05} +{"train_loss": 0.015558378770947456, "grad_norm": 0.07130693644285202, "global_step": 19231, "epoch": 181, "lr": 7.088048817808698e-05} +{"train_loss": 0.013528218492865562, "grad_norm": 0.1059592142701149, "global_step": 19232, "epoch": 181, "lr": 7.087779518453491e-05} +{"train_loss": 0.015093321911990643, "grad_norm": 0.050437137484550476, "global_step": 19233, "epoch": 181, "lr": 7.087510211762733e-05} +{"train_loss": 0.01769966632127762, "grad_norm": 0.08750057965517044, "global_step": 19234, "epoch": 181, "lr": 7.087240897737374e-05} +{"train_loss": 0.01650497131049633, "grad_norm": 0.08210010081529617, "global_step": 19235, "epoch": 181, "lr": 7.086971576378356e-05} +{"train_loss": 0.01465383730828762, "grad_norm": 0.05594927445054054, "global_step": 19236, "epoch": 181, "lr": 7.086702247686629e-05} +{"train_loss": 0.016934111714363098, "grad_norm": 0.13518522679805756, "global_step": 19237, "epoch": 181, "lr": 7.086432911663137e-05} +{"train_loss": 0.014545101672410965, "grad_norm": 0.05711627006530762, "global_step": 19238, "epoch": 181, "lr": 7.086163568308828e-05} +{"train_loss": 0.017244994640350342, "grad_norm": 0.10493273288011551, "global_step": 19239, "epoch": 181, "lr": 7.085894217624645e-05} +{"train_loss": 0.01802113652229309, "grad_norm": 0.12857584655284882, "global_step": 19240, "epoch": 181, "lr": 7.085624859611539e-05} +{"train_loss": 0.018516356125473976, "grad_norm": 0.11402860283851624, "global_step": 19241, "epoch": 181, "lr": 7.085355494270454e-05} +{"train_loss": 0.017504937946796417, "grad_norm": 0.08142586797475815, "global_step": 19242, "epoch": 181, "lr": 7.085086121602336e-05} +{"train_loss": 0.015240218490362167, "grad_norm": 0.12008153647184372, "global_step": 19243, "epoch": 181, "lr": 7.084816741608133e-05} +{"train_loss": 0.014013159088790417, "grad_norm": 0.08984801173210144, "global_step": 19244, "epoch": 181, "lr": 7.08454735428879e-05} +{"train_loss": 0.015086114406585693, "grad_norm": 0.08003716170787811, "global_step": 19245, "epoch": 181, "lr": 7.084277959645255e-05} +{"train_loss": 0.01645144447684288, "grad_norm": 0.13327451050281525, "global_step": 19246, "epoch": 181, "lr": 7.084008557678473e-05} +{"train_loss": 0.015949364751577377, "grad_norm": 0.059912584722042084, "global_step": 19247, "epoch": 181, "lr": 7.083739148389391e-05} +{"train_loss": 0.01684115082025528, "grad_norm": 0.04768383130431175, "global_step": 19248, "epoch": 181, "lr": 7.083469731778956e-05} +{"train_loss": 0.016924317926168442, "grad_norm": 0.13771702349185944, "global_step": 19249, "epoch": 181, "lr": 7.083200307848116e-05} +{"train_loss": 0.016852684319019318, "grad_norm": 0.10903093963861465, "global_step": 19250, "epoch": 181, "lr": 7.082930876597814e-05} +{"train_loss": 0.015509220771491528, "grad_norm": 0.057773105800151825, "global_step": 19251, "epoch": 181, "lr": 7.082661438028999e-05} +{"train_loss": 0.01722291111946106, "grad_norm": 0.0771993100643158, "global_step": 19252, "epoch": 181, "lr": 7.082391992142618e-05} +{"train_loss": 0.019779302179813385, "grad_norm": 0.19422775506973267, "global_step": 19253, "epoch": 181, "lr": 7.082122538939617e-05} +{"train_loss": 0.015471571125090122, "grad_norm": 0.060972511768341064, "global_step": 19254, "epoch": 181, "lr": 7.081853078420943e-05} +{"train_loss": 0.01654386892914772, "grad_norm": 0.10403213649988174, "global_step": 19255, "epoch": 181, "lr": 7.081583610587543e-05} +{"train_loss": 0.015863534063100815, "grad_norm": 0.05959819629788399, "global_step": 19256, "epoch": 181, "lr": 7.081314135440362e-05} +{"train_loss": 0.016512081027030945, "grad_norm": 0.10759103298187256, "global_step": 19257, "epoch": 181, "lr": 7.08104465298035e-05} +{"train_loss": 0.016759086400270462, "grad_norm": 0.10149871557950974, "global_step": 19258, "epoch": 181, "lr": 7.080775163208451e-05} +{"train_loss": 0.018130429089069366, "grad_norm": 0.10281364619731903, "global_step": 19259, "epoch": 181, "lr": 7.080505666125614e-05} +{"train_loss": 0.014502840116620064, "grad_norm": 0.10946846008300781, "global_step": 19260, "epoch": 181, "lr": 7.080236161732784e-05} +{"train_loss": 0.01637793891131878, "grad_norm": 0.05456837639212608, "global_step": 19261, "epoch": 181, "lr": 7.07996665003091e-05} +{"train_loss": 0.015375955030322075, "grad_norm": 0.10848116874694824, "global_step": 19262, "epoch": 181, "lr": 7.079697131020936e-05} +{"train_loss": 0.015617584809660912, "grad_norm": 0.07349224388599396, "global_step": 19263, "epoch": 181, "lr": 7.079427604703809e-05} +{"train_loss": 0.01511764619499445, "grad_norm": 0.0818079486489296, "global_step": 19264, "epoch": 181, "lr": 7.079158071080482e-05} +{"train_loss": 0.017289824783802032, "grad_norm": 0.1847391128540039, "global_step": 19265, "epoch": 181, "lr": 7.078888530151897e-05} +{"train_loss": 0.015499094501137733, "grad_norm": 0.0641293078660965, "global_step": 19266, "epoch": 181, "lr": 7.078618981919e-05} +{"train_loss": 0.01554291881620884, "grad_norm": 0.10517621040344238, "global_step": 19267, "epoch": 181, "lr": 7.078349426382742e-05} +{"train_loss": 0.014420097693800926, "grad_norm": 0.09095081686973572, "global_step": 19268, "epoch": 181, "lr": 7.078079863544065e-05} +{"train_loss": 0.01569453813135624, "grad_norm": 0.08219046890735626, "global_step": 19269, "epoch": 181, "lr": 7.077810293403921e-05} +{"train_loss": 0.019138578325510025, "grad_norm": 0.08035609871149063, "global_step": 19270, "epoch": 181, "lr": 7.077540715963255e-05} +{"train_loss": 0.01441558264195919, "grad_norm": 0.06003569811582565, "global_step": 19271, "epoch": 181, "lr": 7.077271131223014e-05} +{"train_loss": 0.015070471912622452, "grad_norm": 0.08870960026979446, "global_step": 19272, "epoch": 181, "lr": 7.077001539184148e-05} +{"train_loss": 0.019671672955155373, "grad_norm": 0.09333296865224838, "global_step": 19273, "epoch": 181, "lr": 7.076731939847598e-05} +{"train_loss": 0.015201879665255547, "grad_norm": 0.08403648436069489, "global_step": 19274, "epoch": 181, "lr": 7.076462333214317e-05} +{"train_loss": 0.01668582484126091, "grad_norm": 0.09167566150426865, "global_step": 19275, "epoch": 181, "lr": 7.076192719285251e-05} +{"train_loss": 0.01494583860039711, "grad_norm": 0.07544207572937012, "global_step": 19276, "epoch": 181, "lr": 7.075923098061345e-05} +{"train_loss": 0.01555691845715046, "grad_norm": 0.10221470147371292, "global_step": 19277, "epoch": 181, "lr": 7.075653469543551e-05} +{"train_loss": 0.01706763543188572, "grad_norm": 0.05016252398490906, "global_step": 19278, "epoch": 181, "lr": 7.07538383373281e-05} +{"train_loss": 0.014382087625563145, "grad_norm": 0.12057246267795563, "global_step": 19279, "epoch": 181, "lr": 7.075114190630074e-05} +{"train_loss": 0.016504067927598953, "grad_norm": 0.18187236785888672, "global_step": 19280, "epoch": 181, "lr": 7.07484454023629e-05} +{"train_loss": 0.015575909055769444, "grad_norm": 0.08751565217971802, "global_step": 19281, "epoch": 181, "lr": 7.074574882552402e-05} +{"train_loss": 0.014329154044389725, "grad_norm": 0.07412662357091904, "global_step": 19282, "epoch": 181, "lr": 7.074305217579364e-05} +{"train_loss": 0.016230743378400803, "grad_norm": 0.09613265842199326, "global_step": 19283, "epoch": 181, "lr": 7.074035545318117e-05} +{"train_loss": 0.017817478626966476, "grad_norm": 0.07737307995557785, "global_step": 19284, "epoch": 181, "lr": 7.073765865769612e-05} +{"train_loss": 0.013461872935295105, "grad_norm": 0.1150800809264183, "global_step": 19285, "epoch": 181, "lr": 7.073496178934796e-05} +{"train_loss": 0.01712804101407528, "grad_norm": 0.0809461921453476, "global_step": 19286, "epoch": 181, "lr": 7.073226484814614e-05} +{"train_loss": 0.015166687779128551, "grad_norm": 0.10507775843143463, "global_step": 19287, "epoch": 181, "lr": 7.072956783410018e-05} +{"train_loss": 0.017691083252429962, "grad_norm": 0.06087486818432808, "global_step": 19288, "epoch": 181, "lr": 7.072687074721953e-05} +{"train_loss": 0.013434253633022308, "grad_norm": 0.09409730136394501, "global_step": 19289, "epoch": 181, "lr": 7.072417358751366e-05} +{"train_loss": 0.01561313308775425, "grad_norm": 0.08921504020690918, "global_step": 19290, "epoch": 181, "lr": 7.072147635499206e-05} +{"train_loss": 0.016198994556687912, "grad_norm": 0.07913854718208313, "global_step": 19291, "epoch": 181, "lr": 7.071877904966423e-05, "val_loss": 0.020008515566587448} +{"train_loss": 0.013713860884308815, "grad_norm": 0.11950859427452087, "global_step": 19292, "epoch": 182, "lr": 7.071608167153957e-05} +{"train_loss": 0.015851348638534546, "grad_norm": 0.07667955756187439, "global_step": 19293, "epoch": 182, "lr": 7.071338422062766e-05} +{"train_loss": 0.01727212220430374, "grad_norm": 0.19200517237186432, "global_step": 19294, "epoch": 182, "lr": 7.071068669693788e-05} +{"train_loss": 0.018471529707312584, "grad_norm": 0.06430045515298843, "global_step": 19295, "epoch": 182, "lr": 7.070798910047978e-05} +{"train_loss": 0.016236010938882828, "grad_norm": 0.0911673828959465, "global_step": 19296, "epoch": 182, "lr": 7.070529143126281e-05} +{"train_loss": 0.013949052430689335, "grad_norm": 0.1250390112400055, "global_step": 19297, "epoch": 182, "lr": 7.070259368929644e-05} +{"train_loss": 0.015745285898447037, "grad_norm": 0.18760327994823456, "global_step": 19298, "epoch": 182, "lr": 7.069989587459018e-05} +{"train_loss": 0.01360478438436985, "grad_norm": 0.04600067064166069, "global_step": 19299, "epoch": 182, "lr": 7.069719798715346e-05} +{"train_loss": 0.015103007666766644, "grad_norm": 0.06223613768815994, "global_step": 19300, "epoch": 182, "lr": 7.06945000269958e-05} +{"train_loss": 0.01601230911910534, "grad_norm": 0.127867192029953, "global_step": 19301, "epoch": 182, "lr": 7.069180199412669e-05} +{"train_loss": 0.015450999140739441, "grad_norm": 0.05640113353729248, "global_step": 19302, "epoch": 182, "lr": 7.068910388855555e-05} +{"train_loss": 0.013841408304870129, "grad_norm": 0.07574450224637985, "global_step": 19303, "epoch": 182, "lr": 7.068640571029192e-05} +{"train_loss": 0.015074849128723145, "grad_norm": 0.05896918475627899, "global_step": 19304, "epoch": 182, "lr": 7.068370745934524e-05} +{"train_loss": 0.017382102087140083, "grad_norm": 0.08579467982053757, "global_step": 19305, "epoch": 182, "lr": 7.068100913572502e-05} +{"train_loss": 0.015027811750769615, "grad_norm": 0.1430312842130661, "global_step": 19306, "epoch": 182, "lr": 7.067831073944072e-05} +{"train_loss": 0.016664961352944374, "grad_norm": 0.2169884890317917, "global_step": 19307, "epoch": 182, "lr": 7.067561227050182e-05} +{"train_loss": 0.017115427181124687, "grad_norm": 0.09926600009202957, "global_step": 19308, "epoch": 182, "lr": 7.067291372891784e-05} +{"train_loss": 0.01639391854405403, "grad_norm": 0.10247237235307693, "global_step": 19309, "epoch": 182, "lr": 7.06702151146982e-05} +{"train_loss": 0.01422996073961258, "grad_norm": 0.10978227108716965, "global_step": 19310, "epoch": 182, "lr": 7.066751642785243e-05} +{"train_loss": 0.014151024632155895, "grad_norm": 0.15099936723709106, "global_step": 19311, "epoch": 182, "lr": 7.066481766839e-05} +{"train_loss": 0.013752369210124016, "grad_norm": 0.14893242716789246, "global_step": 19312, "epoch": 182, "lr": 7.066211883632036e-05} +{"train_loss": 0.016201240941882133, "grad_norm": 0.06612485647201538, "global_step": 19313, "epoch": 182, "lr": 7.065941993165305e-05} +{"train_loss": 0.015277408994734287, "grad_norm": 0.0658135786652565, "global_step": 19314, "epoch": 182, "lr": 7.065672095439751e-05} +{"train_loss": 0.016962692141532898, "grad_norm": 0.11074024438858032, "global_step": 19315, "epoch": 182, "lr": 7.065402190456325e-05} +{"train_loss": 0.014381838962435722, "grad_norm": 0.07005827128887177, "global_step": 19316, "epoch": 182, "lr": 7.065132278215972e-05} +{"train_loss": 0.016621259972453117, "grad_norm": 0.1289687305688858, "global_step": 19317, "epoch": 182, "lr": 7.064862358719644e-05} +{"train_loss": 0.0160609669983387, "grad_norm": 0.06656360626220703, "global_step": 19318, "epoch": 182, "lr": 7.064592431968287e-05} +{"train_loss": 0.01517578400671482, "grad_norm": 0.0555645227432251, "global_step": 19319, "epoch": 182, "lr": 7.064322497962848e-05} +{"train_loss": 0.014792111702263355, "grad_norm": 0.13359959423542023, "global_step": 19320, "epoch": 182, "lr": 7.064052556704281e-05} +{"train_loss": 0.016122497618198395, "grad_norm": 0.09627906233072281, "global_step": 19321, "epoch": 182, "lr": 7.063782608193527e-05} +{"train_loss": 0.015477994456887245, "grad_norm": 0.12252500653266907, "global_step": 19322, "epoch": 182, "lr": 7.063512652431542e-05} +{"train_loss": 0.01503276638686657, "grad_norm": 0.13904590904712677, "global_step": 19323, "epoch": 182, "lr": 7.06324268941927e-05} +{"train_loss": 0.016771826893091202, "grad_norm": 0.11801528930664062, "global_step": 19324, "epoch": 182, "lr": 7.062972719157659e-05} +{"train_loss": 0.018766414374113083, "grad_norm": 0.13988789916038513, "global_step": 19325, "epoch": 182, "lr": 7.06270274164766e-05} +{"train_loss": 0.016134794801473618, "grad_norm": 0.10696014761924744, "global_step": 19326, "epoch": 182, "lr": 7.06243275689022e-05} +{"train_loss": 0.016174327582120895, "grad_norm": 0.0699736624956131, "global_step": 19327, "epoch": 182, "lr": 7.06216276488629e-05} +{"train_loss": 0.015412678942084312, "grad_norm": 0.12642529606819153, "global_step": 19328, "epoch": 182, "lr": 7.061892765636815e-05} +{"train_loss": 0.017050553113222122, "grad_norm": 0.13850504159927368, "global_step": 19329, "epoch": 182, "lr": 7.061622759142748e-05} +{"train_loss": 0.01605278067290783, "grad_norm": 0.05616449937224388, "global_step": 19330, "epoch": 182, "lr": 7.061352745405032e-05} +{"train_loss": 0.017535310238599777, "grad_norm": 0.12023478746414185, "global_step": 19331, "epoch": 182, "lr": 7.06108272442462e-05} +{"train_loss": 0.016268564388155937, "grad_norm": 0.08841021358966827, "global_step": 19332, "epoch": 182, "lr": 7.06081269620246e-05} +{"train_loss": 0.016968369483947754, "grad_norm": 0.05974116176366806, "global_step": 19333, "epoch": 182, "lr": 7.060542660739502e-05} +{"train_loss": 0.017516538500785828, "grad_norm": 0.07859713584184647, "global_step": 19334, "epoch": 182, "lr": 7.060272618036691e-05} +{"train_loss": 0.01712581142783165, "grad_norm": 0.052372537553310394, "global_step": 19335, "epoch": 182, "lr": 7.06000256809498e-05} +{"train_loss": 0.01895807683467865, "grad_norm": 0.07949735224246979, "global_step": 19336, "epoch": 182, "lr": 7.059732510915313e-05} +{"train_loss": 0.016207551583647728, "grad_norm": 0.07698630541563034, "global_step": 19337, "epoch": 182, "lr": 7.059462446498643e-05} +{"train_loss": 0.016679443418979645, "grad_norm": 0.16833657026290894, "global_step": 19338, "epoch": 182, "lr": 7.059192374845917e-05} +{"train_loss": 0.016092922538518906, "grad_norm": 0.05373474955558777, "global_step": 19339, "epoch": 182, "lr": 7.058922295958086e-05} +{"train_loss": 0.017344743013381958, "grad_norm": 0.08910930156707764, "global_step": 19340, "epoch": 182, "lr": 7.058652209836097e-05} +{"train_loss": 0.015537487342953682, "grad_norm": 0.11631426960229874, "global_step": 19341, "epoch": 182, "lr": 7.0583821164809e-05} +{"train_loss": 0.017977038398385048, "grad_norm": 0.056753285229206085, "global_step": 19342, "epoch": 182, "lr": 7.058112015893442e-05} +{"train_loss": 0.013483686372637749, "grad_norm": 0.11258582770824432, "global_step": 19343, "epoch": 182, "lr": 7.057841908074674e-05} +{"train_loss": 0.015488551929593086, "grad_norm": 0.14025144279003143, "global_step": 19344, "epoch": 182, "lr": 7.057571793025544e-05} +{"train_loss": 0.015260066837072372, "grad_norm": 0.08801960945129395, "global_step": 19345, "epoch": 182, "lr": 7.057301670747003e-05} +{"train_loss": 0.017190586775541306, "grad_norm": 0.0910712406039238, "global_step": 19346, "epoch": 182, "lr": 7.057031541239998e-05} +{"train_loss": 0.01999024674296379, "grad_norm": 0.087662473320961, "global_step": 19347, "epoch": 182, "lr": 7.056761404505477e-05} +{"train_loss": 0.01316166017204523, "grad_norm": 0.09714896231889725, "global_step": 19348, "epoch": 182, "lr": 7.056491260544392e-05} +{"train_loss": 0.019424350932240486, "grad_norm": 0.10077161341905594, "global_step": 19349, "epoch": 182, "lr": 7.056221109357692e-05} +{"train_loss": 0.018001440912485123, "grad_norm": 0.11964985728263855, "global_step": 19350, "epoch": 182, "lr": 7.055950950946324e-05} +{"train_loss": 0.016032209619879723, "grad_norm": 0.16887031495571136, "global_step": 19351, "epoch": 182, "lr": 7.055680785311239e-05} +{"train_loss": 0.017352350056171417, "grad_norm": 0.1307259351015091, "global_step": 19352, "epoch": 182, "lr": 7.055410612453387e-05} +{"train_loss": 0.01649153232574463, "grad_norm": 0.07179279625415802, "global_step": 19353, "epoch": 182, "lr": 7.055140432373715e-05} +{"train_loss": 0.015382862649857998, "grad_norm": 0.07484081387519836, "global_step": 19354, "epoch": 182, "lr": 7.054870245073171e-05} +{"train_loss": 0.016410861164331436, "grad_norm": 0.09419451653957367, "global_step": 19355, "epoch": 182, "lr": 7.05460005055271e-05} +{"train_loss": 0.01814192719757557, "grad_norm": 0.06218687817454338, "global_step": 19356, "epoch": 182, "lr": 7.054329848813278e-05} +{"train_loss": 0.015618322417140007, "grad_norm": 0.06738109141588211, "global_step": 19357, "epoch": 182, "lr": 7.054059639855821e-05} +{"train_loss": 0.017725519835948944, "grad_norm": 0.0755588635802269, "global_step": 19358, "epoch": 182, "lr": 7.053789423681294e-05} +{"train_loss": 0.01638716086745262, "grad_norm": 0.05720868706703186, "global_step": 19359, "epoch": 182, "lr": 7.053519200290644e-05} +{"train_loss": 0.01621907576918602, "grad_norm": 0.07853429019451141, "global_step": 19360, "epoch": 182, "lr": 7.05324896968482e-05} +{"train_loss": 0.014334348030388355, "grad_norm": 0.11400555074214935, "global_step": 19361, "epoch": 182, "lr": 7.052978731864772e-05} +{"train_loss": 0.016190584748983383, "grad_norm": 0.11413951963186264, "global_step": 19362, "epoch": 182, "lr": 7.052708486831451e-05} +{"train_loss": 0.014752219431102276, "grad_norm": 0.05623398721218109, "global_step": 19363, "epoch": 182, "lr": 7.052438234585804e-05} +{"train_loss": 0.01721632480621338, "grad_norm": 0.1321004331111908, "global_step": 19364, "epoch": 182, "lr": 7.052167975128782e-05} +{"train_loss": 0.01596938632428646, "grad_norm": 0.08158019185066223, "global_step": 19365, "epoch": 182, "lr": 7.051897708461332e-05} +{"train_loss": 0.016269125044345856, "grad_norm": 0.14184746146202087, "global_step": 19366, "epoch": 182, "lr": 7.051627434584408e-05} +{"train_loss": 0.016490258276462555, "grad_norm": 0.10722635686397552, "global_step": 19367, "epoch": 182, "lr": 7.051357153498958e-05} +{"train_loss": 0.015017276629805565, "grad_norm": 0.08278201520442963, "global_step": 19368, "epoch": 182, "lr": 7.051086865205929e-05} +{"train_loss": 0.01524447463452816, "grad_norm": 0.06152131408452988, "global_step": 19369, "epoch": 182, "lr": 7.050816569706274e-05} +{"train_loss": 0.01906982623040676, "grad_norm": 0.10200479626655579, "global_step": 19370, "epoch": 182, "lr": 7.050546267000942e-05} +{"train_loss": 0.016341546550393105, "grad_norm": 0.08295406401157379, "global_step": 19371, "epoch": 182, "lr": 7.050275957090881e-05} +{"train_loss": 0.015749236568808556, "grad_norm": 0.07184117287397385, "global_step": 19372, "epoch": 182, "lr": 7.050005639977043e-05} +{"train_loss": 0.01455643493682146, "grad_norm": 0.0734940841794014, "global_step": 19373, "epoch": 182, "lr": 7.049735315660376e-05} +{"train_loss": 0.017573941498994827, "grad_norm": 0.08823207765817642, "global_step": 19374, "epoch": 182, "lr": 7.04946498414183e-05} +{"train_loss": 0.018311481922864914, "grad_norm": 0.05271465703845024, "global_step": 19375, "epoch": 182, "lr": 7.049194645422356e-05} +{"train_loss": 0.014609203673899174, "grad_norm": 0.1480911374092102, "global_step": 19376, "epoch": 182, "lr": 7.048924299502904e-05} +{"train_loss": 0.017379704862833023, "grad_norm": 0.18926410377025604, "global_step": 19377, "epoch": 182, "lr": 7.048653946384422e-05} +{"train_loss": 0.013833935372531414, "grad_norm": 0.07024353742599487, "global_step": 19378, "epoch": 182, "lr": 7.04838358606786e-05} +{"train_loss": 0.01574305072426796, "grad_norm": 0.07495380192995071, "global_step": 19379, "epoch": 182, "lr": 7.04811321855417e-05} +{"train_loss": 0.015782661736011505, "grad_norm": 0.07969800382852554, "global_step": 19380, "epoch": 182, "lr": 7.0478428438443e-05} +{"train_loss": 0.016001321375370026, "grad_norm": 0.09485182166099548, "global_step": 19381, "epoch": 182, "lr": 7.047572461939204e-05} +{"train_loss": 0.016223309561610222, "grad_norm": 0.07629936933517456, "global_step": 19382, "epoch": 182, "lr": 7.047302072839826e-05} +{"train_loss": 0.01471559889614582, "grad_norm": 0.04791812598705292, "global_step": 19383, "epoch": 182, "lr": 7.04703167654712e-05} +{"train_loss": 0.018261021003127098, "grad_norm": 0.07313034683465958, "global_step": 19384, "epoch": 182, "lr": 7.046761273062035e-05} +{"train_loss": 0.01822749152779579, "grad_norm": 0.059586092829704285, "global_step": 19385, "epoch": 182, "lr": 7.04649086238552e-05} +{"train_loss": 0.016581092029809952, "grad_norm": 0.06459645926952362, "global_step": 19386, "epoch": 182, "lr": 7.046220444518526e-05} +{"train_loss": 0.01730608381330967, "grad_norm": 0.09974690526723862, "global_step": 19387, "epoch": 182, "lr": 7.045950019462003e-05} +{"train_loss": 0.014282586984336376, "grad_norm": 0.05752430856227875, "global_step": 19388, "epoch": 182, "lr": 7.045679587216904e-05} +{"train_loss": 0.017508285120129585, "grad_norm": 0.10169000923633575, "global_step": 19389, "epoch": 182, "lr": 7.045409147784173e-05} +{"train_loss": 0.018701691180467606, "grad_norm": 0.14609910547733307, "global_step": 19390, "epoch": 182, "lr": 7.045138701164766e-05} +{"train_loss": 0.01728644035756588, "grad_norm": 0.07120609283447266, "global_step": 19391, "epoch": 182, "lr": 7.04486824735963e-05} +{"train_loss": 0.015936195850372314, "grad_norm": 0.08213424682617188, "global_step": 19392, "epoch": 182, "lr": 7.044597786369717e-05} +{"train_loss": 0.016105279326438904, "grad_norm": 0.06991631537675858, "global_step": 19393, "epoch": 182, "lr": 7.044327318195976e-05} +{"train_loss": 0.01895272731781006, "grad_norm": 0.09859586507081985, "global_step": 19394, "epoch": 182, "lr": 7.044056842839358e-05} +{"train_loss": 0.015872135758399963, "grad_norm": 0.08825212717056274, "global_step": 19395, "epoch": 182, "lr": 7.043786360300813e-05} +{"train_loss": 0.014587687328457832, "grad_norm": 0.05776903033256531, "global_step": 19396, "epoch": 182, "lr": 7.043515870581294e-05} +{"train_loss": 0.016242787347368476, "grad_norm": 0.10021639615297318, "global_step": 19397, "epoch": 182, "lr": 7.043245373681747e-05, "val_loss": 0.01882544532418251} +{"train_loss": 0.014305021613836288, "grad_norm": 0.12160099297761917, "global_step": 19398, "epoch": 183, "lr": 7.042974869603122e-05} +{"train_loss": 0.016218001022934914, "grad_norm": 0.06490243971347809, "global_step": 19399, "epoch": 183, "lr": 7.042704358346375e-05} +{"train_loss": 0.015151643194258213, "grad_norm": 0.07135551422834396, "global_step": 19400, "epoch": 183, "lr": 7.042433839912452e-05} +{"train_loss": 0.017064157873392105, "grad_norm": 0.0745469406247139, "global_step": 19401, "epoch": 183, "lr": 7.042163314302304e-05} +{"train_loss": 0.018188267946243286, "grad_norm": 0.07917752116918564, "global_step": 19402, "epoch": 183, "lr": 7.041892781516884e-05} +{"train_loss": 0.017967589199543, "grad_norm": 0.061752837151288986, "global_step": 19403, "epoch": 183, "lr": 7.04162224155714e-05} +{"train_loss": 0.0184151791036129, "grad_norm": 0.06215363368391991, "global_step": 19404, "epoch": 183, "lr": 7.041351694424023e-05} +{"train_loss": 0.018018972128629684, "grad_norm": 0.08762207627296448, "global_step": 19405, "epoch": 183, "lr": 7.041081140118485e-05} +{"train_loss": 0.015927037224173546, "grad_norm": 0.09527963399887085, "global_step": 19406, "epoch": 183, "lr": 7.040810578641474e-05} +{"train_loss": 0.020118961110711098, "grad_norm": 0.09771968424320221, "global_step": 19407, "epoch": 183, "lr": 7.040540009993943e-05} +{"train_loss": 0.015059785917401314, "grad_norm": 0.047972653061151505, "global_step": 19408, "epoch": 183, "lr": 7.040269434176842e-05} +{"train_loss": 0.015628444030880928, "grad_norm": 0.05498719587922096, "global_step": 19409, "epoch": 183, "lr": 7.03999885119112e-05} +{"train_loss": 0.015433182939887047, "grad_norm": 0.057938266545534134, "global_step": 19410, "epoch": 183, "lr": 7.03972826103773e-05} +{"train_loss": 0.016892066225409508, "grad_norm": 0.07743038237094879, "global_step": 19411, "epoch": 183, "lr": 7.039457663717623e-05} +{"train_loss": 0.015809165313839912, "grad_norm": 0.11911968886852264, "global_step": 19412, "epoch": 183, "lr": 7.039187059231747e-05} +{"train_loss": 0.014865923672914505, "grad_norm": 0.07106057554483414, "global_step": 19413, "epoch": 183, "lr": 7.038916447581057e-05} +{"train_loss": 0.01850828155875206, "grad_norm": 0.07545018196105957, "global_step": 19414, "epoch": 183, "lr": 7.038645828766499e-05} +{"train_loss": 0.017643120139837265, "grad_norm": 0.11163712292909622, "global_step": 19415, "epoch": 183, "lr": 7.038375202789027e-05} +{"train_loss": 0.01530050951987505, "grad_norm": 0.06735217571258545, "global_step": 19416, "epoch": 183, "lr": 7.03810456964959e-05} +{"train_loss": 0.016422338783740997, "grad_norm": 0.1031644195318222, "global_step": 19417, "epoch": 183, "lr": 7.037833929349142e-05} +{"train_loss": 0.015236277133226395, "grad_norm": 0.13766992092132568, "global_step": 19418, "epoch": 183, "lr": 7.037563281888631e-05} +{"train_loss": 0.015343599952757359, "grad_norm": 0.05569297447800636, "global_step": 19419, "epoch": 183, "lr": 7.037292627269007e-05} +{"train_loss": 0.01581747829914093, "grad_norm": 0.049244850873947144, "global_step": 19420, "epoch": 183, "lr": 7.037021965491224e-05} +{"train_loss": 0.015188322402536869, "grad_norm": 0.07124121487140656, "global_step": 19421, "epoch": 183, "lr": 7.036751296556232e-05} +{"train_loss": 0.01601247303187847, "grad_norm": 0.0714133083820343, "global_step": 19422, "epoch": 183, "lr": 7.036480620464981e-05} +{"train_loss": 0.016238443553447723, "grad_norm": 0.05832355096936226, "global_step": 19423, "epoch": 183, "lr": 7.036209937218424e-05} +{"train_loss": 0.016870565712451935, "grad_norm": 0.05972594767808914, "global_step": 19424, "epoch": 183, "lr": 7.03593924681751e-05} +{"train_loss": 0.015564391389489174, "grad_norm": 0.05852927267551422, "global_step": 19425, "epoch": 183, "lr": 7.035668549263191e-05} +{"train_loss": 0.015815429389476776, "grad_norm": 0.07927476614713669, "global_step": 19426, "epoch": 183, "lr": 7.035397844556418e-05} +{"train_loss": 0.01887860894203186, "grad_norm": 0.08142589777708054, "global_step": 19427, "epoch": 183, "lr": 7.035127132698142e-05} +{"train_loss": 0.013599911704659462, "grad_norm": 0.058707110583782196, "global_step": 19428, "epoch": 183, "lr": 7.034856413689314e-05} +{"train_loss": 0.012632578611373901, "grad_norm": 0.08357702195644379, "global_step": 19429, "epoch": 183, "lr": 7.034585687530886e-05} +{"train_loss": 0.014618728309869766, "grad_norm": 0.07869770377874374, "global_step": 19430, "epoch": 183, "lr": 7.034314954223809e-05} +{"train_loss": 0.014230424538254738, "grad_norm": 0.10355820506811142, "global_step": 19431, "epoch": 183, "lr": 7.034044213769032e-05} +{"train_loss": 0.015512580052018166, "grad_norm": 0.1089949980378151, "global_step": 19432, "epoch": 183, "lr": 7.033773466167511e-05} +{"train_loss": 0.013956410810351372, "grad_norm": 0.05320058763027191, "global_step": 19433, "epoch": 183, "lr": 7.033502711420193e-05} +{"train_loss": 0.015375812537968159, "grad_norm": 0.09407305717468262, "global_step": 19434, "epoch": 183, "lr": 7.03323194952803e-05} +{"train_loss": 0.015809591859579086, "grad_norm": 0.10738880187273026, "global_step": 19435, "epoch": 183, "lr": 7.032961180491977e-05} +{"train_loss": 0.01336438674479723, "grad_norm": 0.060274649411439896, "global_step": 19436, "epoch": 183, "lr": 7.032690404312979e-05} +{"train_loss": 0.01437244564294815, "grad_norm": 0.12451223284006119, "global_step": 19437, "epoch": 183, "lr": 7.032419620991993e-05} +{"train_loss": 0.01611040160059929, "grad_norm": 0.05203504487872124, "global_step": 19438, "epoch": 183, "lr": 7.032148830529967e-05} +{"train_loss": 0.015870990231633186, "grad_norm": 0.06728050112724304, "global_step": 19439, "epoch": 183, "lr": 7.031878032927857e-05} +{"train_loss": 0.01502285711467266, "grad_norm": 0.13515183329582214, "global_step": 19440, "epoch": 183, "lr": 7.031607228186608e-05} +{"train_loss": 0.01812509447336197, "grad_norm": 0.13703590631484985, "global_step": 19441, "epoch": 183, "lr": 7.031336416307176e-05} +{"train_loss": 0.017452623695135117, "grad_norm": 0.10141590982675552, "global_step": 19442, "epoch": 183, "lr": 7.031065597290511e-05} +{"train_loss": 0.017821578308939934, "grad_norm": 0.12094894796609879, "global_step": 19443, "epoch": 183, "lr": 7.030794771137563e-05} +{"train_loss": 0.018418200314044952, "grad_norm": 0.09319452941417694, "global_step": 19444, "epoch": 183, "lr": 7.03052393784929e-05} +{"train_loss": 0.01710677146911621, "grad_norm": 0.054791130125522614, "global_step": 19445, "epoch": 183, "lr": 7.030253097426634e-05} +{"train_loss": 0.013808190822601318, "grad_norm": 0.11484147608280182, "global_step": 19446, "epoch": 183, "lr": 7.029982249870555e-05} +{"train_loss": 0.016880499199032784, "grad_norm": 0.07276731729507446, "global_step": 19447, "epoch": 183, "lr": 7.029711395181998e-05} +{"train_loss": 0.01604863442480564, "grad_norm": 0.09213249385356903, "global_step": 19448, "epoch": 183, "lr": 7.02944053336192e-05} +{"train_loss": 0.01737162098288536, "grad_norm": 0.11148445308208466, "global_step": 19449, "epoch": 183, "lr": 7.02916966441127e-05} +{"train_loss": 0.015645809471607208, "grad_norm": 0.0849560871720314, "global_step": 19450, "epoch": 183, "lr": 7.028898788331e-05} +{"train_loss": 0.017844902351498604, "grad_norm": 0.11361327767372131, "global_step": 19451, "epoch": 183, "lr": 7.028627905122063e-05} +{"train_loss": 0.01671673357486725, "grad_norm": 0.060239363461732864, "global_step": 19452, "epoch": 183, "lr": 7.028357014785409e-05} +{"train_loss": 0.014184008352458477, "grad_norm": 0.10856369137763977, "global_step": 19453, "epoch": 183, "lr": 7.02808611732199e-05} +{"train_loss": 0.01802704855799675, "grad_norm": 0.1314808428287506, "global_step": 19454, "epoch": 183, "lr": 7.02781521273276e-05} +{"train_loss": 0.018293602392077446, "grad_norm": 0.08157015591859818, "global_step": 19455, "epoch": 183, "lr": 7.027544301018668e-05} +{"train_loss": 0.016642233356833458, "grad_norm": 0.08073721826076508, "global_step": 19456, "epoch": 183, "lr": 7.027273382180669e-05} +{"train_loss": 0.015297330915927887, "grad_norm": 0.14326347410678864, "global_step": 19457, "epoch": 183, "lr": 7.027002456219709e-05} +{"train_loss": 0.01664331555366516, "grad_norm": 0.06093380227684975, "global_step": 19458, "epoch": 183, "lr": 7.026731523136747e-05} +{"train_loss": 0.015342649072408676, "grad_norm": 0.08630258589982986, "global_step": 19459, "epoch": 183, "lr": 7.02646058293273e-05} +{"train_loss": 0.017596129328012466, "grad_norm": 0.07907290756702423, "global_step": 19460, "epoch": 183, "lr": 7.026189635608613e-05} +{"train_loss": 0.016501106321811676, "grad_norm": 0.05638882517814636, "global_step": 19461, "epoch": 183, "lr": 7.025918681165349e-05} +{"train_loss": 0.015716008841991425, "grad_norm": 0.10360465943813324, "global_step": 19462, "epoch": 183, "lr": 7.025647719603884e-05} +{"train_loss": 0.015539051964879036, "grad_norm": 0.048084743320941925, "global_step": 19463, "epoch": 183, "lr": 7.025376750925176e-05} +{"train_loss": 0.014833668246865273, "grad_norm": 0.14407581090927124, "global_step": 19464, "epoch": 183, "lr": 7.025105775130174e-05} +{"train_loss": 0.013693667016923428, "grad_norm": 0.060547471046447754, "global_step": 19465, "epoch": 183, "lr": 7.02483479221983e-05} +{"train_loss": 0.017111968249082565, "grad_norm": 0.06394577026367188, "global_step": 19466, "epoch": 183, "lr": 7.024563802195099e-05} +{"train_loss": 0.017601752653717995, "grad_norm": 0.05661953240633011, "global_step": 19467, "epoch": 183, "lr": 7.024292805056932e-05} +{"train_loss": 0.017177360132336617, "grad_norm": 0.08559712022542953, "global_step": 19468, "epoch": 183, "lr": 7.024021800806278e-05} +{"train_loss": 0.01378550287336111, "grad_norm": 0.1039276123046875, "global_step": 19469, "epoch": 183, "lr": 7.023750789444091e-05} +{"train_loss": 0.014787047170102596, "grad_norm": 0.05644123628735542, "global_step": 19470, "epoch": 183, "lr": 7.023479770971327e-05} +{"train_loss": 0.016042791306972504, "grad_norm": 0.06534619629383087, "global_step": 19471, "epoch": 183, "lr": 7.023208745388933e-05} +{"train_loss": 0.018521733582019806, "grad_norm": 0.16673044860363007, "global_step": 19472, "epoch": 183, "lr": 7.022937712697863e-05} +{"train_loss": 0.01482375804334879, "grad_norm": 0.0961310863494873, "global_step": 19473, "epoch": 183, "lr": 7.022666672899072e-05} +{"train_loss": 0.01550346240401268, "grad_norm": 0.05991165712475777, "global_step": 19474, "epoch": 183, "lr": 7.022395625993507e-05} +{"train_loss": 0.016260197386145592, "grad_norm": 0.05468089506030083, "global_step": 19475, "epoch": 183, "lr": 7.022124571982124e-05} +{"train_loss": 0.016342516988515854, "grad_norm": 0.06261737644672394, "global_step": 19476, "epoch": 183, "lr": 7.021853510865877e-05} +{"train_loss": 0.016732659190893173, "grad_norm": 0.06779210269451141, "global_step": 19477, "epoch": 183, "lr": 7.021582442645715e-05} +{"train_loss": 0.015135560184717178, "grad_norm": 0.06761151552200317, "global_step": 19478, "epoch": 183, "lr": 7.02131136732259e-05} +{"train_loss": 0.015151134692132473, "grad_norm": 0.08903765678405762, "global_step": 19479, "epoch": 183, "lr": 7.021040284897458e-05} +{"train_loss": 0.019669942557811737, "grad_norm": 0.08952876180410385, "global_step": 19480, "epoch": 183, "lr": 7.020769195371267e-05} +{"train_loss": 0.013793271966278553, "grad_norm": 0.10240133851766586, "global_step": 19481, "epoch": 183, "lr": 7.020498098744971e-05} +{"train_loss": 0.01436556689441204, "grad_norm": 0.06240628659725189, "global_step": 19482, "epoch": 183, "lr": 7.020226995019527e-05} +{"train_loss": 0.014904920943081379, "grad_norm": 0.059638191014528275, "global_step": 19483, "epoch": 183, "lr": 7.019955884195883e-05} +{"train_loss": 0.01788792759180069, "grad_norm": 0.12578819692134857, "global_step": 19484, "epoch": 183, "lr": 7.01968476627499e-05} +{"train_loss": 0.017691239714622498, "grad_norm": 0.07546202838420868, "global_step": 19485, "epoch": 183, "lr": 7.019413641257808e-05} +{"train_loss": 0.016003912314772606, "grad_norm": 0.04733448103070259, "global_step": 19486, "epoch": 183, "lr": 7.01914250914528e-05} +{"train_loss": 0.01573726162314415, "grad_norm": 0.08467811346054077, "global_step": 19487, "epoch": 183, "lr": 7.018871369938365e-05} +{"train_loss": 0.014344850555062294, "grad_norm": 0.05339062213897705, "global_step": 19488, "epoch": 183, "lr": 7.018600223638013e-05} +{"train_loss": 0.016228530555963516, "grad_norm": 0.0705144852399826, "global_step": 19489, "epoch": 183, "lr": 7.01832907024518e-05} +{"train_loss": 0.019293174147605896, "grad_norm": 0.08983148634433746, "global_step": 19490, "epoch": 183, "lr": 7.018057909760815e-05} +{"train_loss": 0.016528550535440445, "grad_norm": 0.06119701638817787, "global_step": 19491, "epoch": 183, "lr": 7.017786742185871e-05} +{"train_loss": 0.01584446430206299, "grad_norm": 0.09414226561784744, "global_step": 19492, "epoch": 183, "lr": 7.017515567521303e-05} +{"train_loss": 0.01753910258412361, "grad_norm": 0.10136712342500687, "global_step": 19493, "epoch": 183, "lr": 7.017244385768062e-05} +{"train_loss": 0.015785764902830124, "grad_norm": 0.10696149617433548, "global_step": 19494, "epoch": 183, "lr": 7.016973196927104e-05} +{"train_loss": 0.016661016270518303, "grad_norm": 0.16439969837665558, "global_step": 19495, "epoch": 183, "lr": 7.016702000999376e-05} +{"train_loss": 0.015667177736759186, "grad_norm": 0.07927018404006958, "global_step": 19496, "epoch": 183, "lr": 7.016430797985837e-05} +{"train_loss": 0.0154709592461586, "grad_norm": 0.1153690293431282, "global_step": 19497, "epoch": 183, "lr": 7.016159587887435e-05} +{"train_loss": 0.014431467279791832, "grad_norm": 0.06908401846885681, "global_step": 19498, "epoch": 183, "lr": 7.015888370705126e-05} +{"train_loss": 0.015417101792991161, "grad_norm": 0.13479097187519073, "global_step": 19499, "epoch": 183, "lr": 7.015617146439863e-05} +{"train_loss": 0.014087335206568241, "grad_norm": 0.08872679620981216, "global_step": 19500, "epoch": 183, "lr": 7.015345915092594e-05} +{"train_loss": 0.017773551866412163, "grad_norm": 0.06287939101457596, "global_step": 19501, "epoch": 183, "lr": 7.01507467666428e-05} +{"train_loss": 0.01922442764043808, "grad_norm": 0.13862960040569305, "global_step": 19502, "epoch": 183, "lr": 7.01480343115587e-05} +{"train_loss": 0.016175415630469907, "grad_norm": 0.1087481752038002, "global_step": 19503, "epoch": 183, "lr": 7.014532178568314e-05, "val_loss": 0.019362252205610275} +{"train_loss": 0.015978824347257614, "grad_norm": 0.04434811696410179, "global_step": 19504, "epoch": 184, "lr": 7.014260918902569e-05} +{"train_loss": 0.017481394112110138, "grad_norm": 0.06103847920894623, "global_step": 19505, "epoch": 184, "lr": 7.013989652159588e-05} +{"train_loss": 0.01900133118033409, "grad_norm": 0.09807352721691132, "global_step": 19506, "epoch": 184, "lr": 7.013718378340323e-05} +{"train_loss": 0.01726285181939602, "grad_norm": 0.08827826380729675, "global_step": 19507, "epoch": 184, "lr": 7.013447097445727e-05} +{"train_loss": 0.017546528950333595, "grad_norm": 0.10735519230365753, "global_step": 19508, "epoch": 184, "lr": 7.013175809476754e-05} +{"train_loss": 0.017379725351929665, "grad_norm": 0.08064320683479309, "global_step": 19509, "epoch": 184, "lr": 7.012904514434356e-05} +{"train_loss": 0.015031819231808186, "grad_norm": 0.0667547956109047, "global_step": 19510, "epoch": 184, "lr": 7.012633212319486e-05} +{"train_loss": 0.017018835991621017, "grad_norm": 0.1076447144150734, "global_step": 19511, "epoch": 184, "lr": 7.012361903133102e-05} +{"train_loss": 0.01553487777709961, "grad_norm": 0.09454113990068436, "global_step": 19512, "epoch": 184, "lr": 7.01209058687615e-05} +{"train_loss": 0.015937380492687225, "grad_norm": 0.06010555103421211, "global_step": 19513, "epoch": 184, "lr": 7.011819263549587e-05} +{"train_loss": 0.016240566968917847, "grad_norm": 0.04819744452834129, "global_step": 19514, "epoch": 184, "lr": 7.011547933154367e-05} +{"train_loss": 0.018165845423936844, "grad_norm": 0.11241565644741058, "global_step": 19515, "epoch": 184, "lr": 7.011276595691441e-05} +{"train_loss": 0.01576741226017475, "grad_norm": 0.07531281560659409, "global_step": 19516, "epoch": 184, "lr": 7.011005251161766e-05} +{"train_loss": 0.015433678403496742, "grad_norm": 0.12319867312908173, "global_step": 19517, "epoch": 184, "lr": 7.010733899566292e-05} +{"train_loss": 0.01644635759294033, "grad_norm": 0.09323658794164658, "global_step": 19518, "epoch": 184, "lr": 7.010462540905974e-05} +{"train_loss": 0.017196647822856903, "grad_norm": 0.06694557517766953, "global_step": 19519, "epoch": 184, "lr": 7.010191175181764e-05} +{"train_loss": 0.014003882184624672, "grad_norm": 0.10199786722660065, "global_step": 19520, "epoch": 184, "lr": 7.009919802394617e-05} +{"train_loss": 0.014249036088585854, "grad_norm": 0.05101732909679413, "global_step": 19521, "epoch": 184, "lr": 7.009648422545485e-05} +{"train_loss": 0.014689870178699493, "grad_norm": 0.07615935057401657, "global_step": 19522, "epoch": 184, "lr": 7.009377035635323e-05} +{"train_loss": 0.015479274094104767, "grad_norm": 0.057883914560079575, "global_step": 19523, "epoch": 184, "lr": 7.009105641665084e-05} +{"train_loss": 0.014248112216591835, "grad_norm": 0.05636909231543541, "global_step": 19524, "epoch": 184, "lr": 7.008834240635723e-05} +{"train_loss": 0.016221655532717705, "grad_norm": 0.05124904587864876, "global_step": 19525, "epoch": 184, "lr": 7.008562832548189e-05} +{"train_loss": 0.014041868969798088, "grad_norm": 0.06458316743373871, "global_step": 19526, "epoch": 184, "lr": 7.00829141740344e-05} +{"train_loss": 0.014410234056413174, "grad_norm": 0.06143946573138237, "global_step": 19527, "epoch": 184, "lr": 7.008019995202428e-05} +{"train_loss": 0.015347929671406746, "grad_norm": 0.04132916405797005, "global_step": 19528, "epoch": 184, "lr": 7.007748565946108e-05} +{"train_loss": 0.019638681784272194, "grad_norm": 0.07510080188512802, "global_step": 19529, "epoch": 184, "lr": 7.007477129635433e-05} +{"train_loss": 0.014741750434041023, "grad_norm": 0.058036793023347855, "global_step": 19530, "epoch": 184, "lr": 7.007205686271357e-05} +{"train_loss": 0.016975436359643936, "grad_norm": 0.0895141139626503, "global_step": 19531, "epoch": 184, "lr": 7.006934235854831e-05} +{"train_loss": 0.015104344114661217, "grad_norm": 0.11192157864570618, "global_step": 19532, "epoch": 184, "lr": 7.006662778386812e-05} +{"train_loss": 0.01578228548169136, "grad_norm": 0.05871199071407318, "global_step": 19533, "epoch": 184, "lr": 7.006391313868253e-05} +{"train_loss": 0.016226032748818398, "grad_norm": 0.11523019522428513, "global_step": 19534, "epoch": 184, "lr": 7.006119842300106e-05} +{"train_loss": 0.015166140161454678, "grad_norm": 0.05222824588418007, "global_step": 19535, "epoch": 184, "lr": 7.005848363683327e-05} +{"train_loss": 0.015132025815546513, "grad_norm": 0.06040294095873833, "global_step": 19536, "epoch": 184, "lr": 7.00557687801887e-05} +{"train_loss": 0.01368277333676815, "grad_norm": 0.0765620544552803, "global_step": 19537, "epoch": 184, "lr": 7.005305385307687e-05} +{"train_loss": 0.016563232988119125, "grad_norm": 0.053287073969841, "global_step": 19538, "epoch": 184, "lr": 7.005033885550734e-05} +{"train_loss": 0.018083684146404266, "grad_norm": 0.0744931548833847, "global_step": 19539, "epoch": 184, "lr": 7.004762378748963e-05} +{"train_loss": 0.014774611219763756, "grad_norm": 0.157109797000885, "global_step": 19540, "epoch": 184, "lr": 7.004490864903329e-05} +{"train_loss": 0.01620742306113243, "grad_norm": 0.05806265398859978, "global_step": 19541, "epoch": 184, "lr": 7.004219344014786e-05} +{"train_loss": 0.016119766980409622, "grad_norm": 0.1078314334154129, "global_step": 19542, "epoch": 184, "lr": 7.003947816084287e-05} +{"train_loss": 0.015603524632751942, "grad_norm": 0.10284847021102905, "global_step": 19543, "epoch": 184, "lr": 7.003676281112788e-05} +{"train_loss": 0.016029100865125656, "grad_norm": 0.0669182613492012, "global_step": 19544, "epoch": 184, "lr": 7.003404739101242e-05} +{"train_loss": 0.015406697057187557, "grad_norm": 0.04696742072701454, "global_step": 19545, "epoch": 184, "lr": 7.003133190050604e-05} +{"train_loss": 0.017767932265996933, "grad_norm": 0.07740511000156403, "global_step": 19546, "epoch": 184, "lr": 7.002861633961825e-05} +{"train_loss": 0.01586531102657318, "grad_norm": 0.0575425960123539, "global_step": 19547, "epoch": 184, "lr": 7.002590070835861e-05} +{"train_loss": 0.01600080542266369, "grad_norm": 0.08960587531328201, "global_step": 19548, "epoch": 184, "lr": 7.002318500673669e-05} +{"train_loss": 0.01669432781636715, "grad_norm": 0.0608174130320549, "global_step": 19549, "epoch": 184, "lr": 7.002046923476199e-05} +{"train_loss": 0.017880816012620926, "grad_norm": 0.10682646930217743, "global_step": 19550, "epoch": 184, "lr": 7.001775339244407e-05} +{"train_loss": 0.015992460772395134, "grad_norm": 0.1553374081850052, "global_step": 19551, "epoch": 184, "lr": 7.001503747979245e-05} +{"train_loss": 0.017140088602900505, "grad_norm": 0.05052868649363518, "global_step": 19552, "epoch": 184, "lr": 7.001232149681672e-05} +{"train_loss": 0.01599051058292389, "grad_norm": 0.08846225589513779, "global_step": 19553, "epoch": 184, "lr": 7.00096054435264e-05} +{"train_loss": 0.01645004190504551, "grad_norm": 0.07175139337778091, "global_step": 19554, "epoch": 184, "lr": 7.000688931993099e-05} +{"train_loss": 0.014437021687626839, "grad_norm": 0.06891031563282013, "global_step": 19555, "epoch": 184, "lr": 7.00041731260401e-05} +{"train_loss": 0.0159553624689579, "grad_norm": 0.11564977467060089, "global_step": 19556, "epoch": 184, "lr": 7.000145686186324e-05} +{"train_loss": 0.016215672716498375, "grad_norm": 0.06249483674764633, "global_step": 19557, "epoch": 184, "lr": 6.999874052740996e-05} +{"train_loss": 0.015476546250283718, "grad_norm": 0.09563072770833969, "global_step": 19558, "epoch": 184, "lr": 6.99960241226898e-05} +{"train_loss": 0.015198849141597748, "grad_norm": 0.1287187933921814, "global_step": 19559, "epoch": 184, "lr": 6.999330764771231e-05} +{"train_loss": 0.01592210680246353, "grad_norm": 0.11178729683160782, "global_step": 19560, "epoch": 184, "lr": 6.999059110248704e-05} +{"train_loss": 0.015118667855858803, "grad_norm": 0.0708221048116684, "global_step": 19561, "epoch": 184, "lr": 6.998787448702352e-05} +{"train_loss": 0.015203981660306454, "grad_norm": 0.13791613280773163, "global_step": 19562, "epoch": 184, "lr": 6.99851578013313e-05} +{"train_loss": 0.016932537779211998, "grad_norm": 0.09757345914840698, "global_step": 19563, "epoch": 184, "lr": 6.998244104541992e-05} +{"train_loss": 0.015268851071596146, "grad_norm": 0.090646892786026, "global_step": 19564, "epoch": 184, "lr": 6.997972421929895e-05} +{"train_loss": 0.019253922626376152, "grad_norm": 0.08040981739759445, "global_step": 19565, "epoch": 184, "lr": 6.99770073229779e-05} +{"train_loss": 0.02016521245241165, "grad_norm": 0.07049061357975006, "global_step": 19566, "epoch": 184, "lr": 6.997429035646633e-05} +{"train_loss": 0.015625175088644028, "grad_norm": 0.1338876187801361, "global_step": 19567, "epoch": 184, "lr": 6.997157331977382e-05} +{"train_loss": 0.015910129994153976, "grad_norm": 0.08285584300756454, "global_step": 19568, "epoch": 184, "lr": 6.996885621290985e-05} +{"train_loss": 0.014897434040904045, "grad_norm": 0.10136034339666367, "global_step": 19569, "epoch": 184, "lr": 6.996613903588403e-05} +{"train_loss": 0.015422996133565903, "grad_norm": 0.055245548486709595, "global_step": 19570, "epoch": 184, "lr": 6.996342178870588e-05} +{"train_loss": 0.014313045889139175, "grad_norm": 0.04937981814146042, "global_step": 19571, "epoch": 184, "lr": 6.996070447138493e-05} +{"train_loss": 0.012932122685015202, "grad_norm": 0.08680450916290283, "global_step": 19572, "epoch": 184, "lr": 6.995798708393075e-05} +{"train_loss": 0.017525626346468925, "grad_norm": 0.09962138533592224, "global_step": 19573, "epoch": 184, "lr": 6.995526962635289e-05} +{"train_loss": 0.019160695374011993, "grad_norm": 0.09768126159906387, "global_step": 19574, "epoch": 184, "lr": 6.99525520986609e-05} +{"train_loss": 0.018500465899705887, "grad_norm": 0.08058004826307297, "global_step": 19575, "epoch": 184, "lr": 6.99498345008643e-05} +{"train_loss": 0.015278736129403114, "grad_norm": 0.10825501382350922, "global_step": 19576, "epoch": 184, "lr": 6.994711683297267e-05} +{"train_loss": 0.018141893669962883, "grad_norm": 0.06793158501386642, "global_step": 19577, "epoch": 184, "lr": 6.994439909499554e-05} +{"train_loss": 0.016666484996676445, "grad_norm": 0.11493689566850662, "global_step": 19578, "epoch": 184, "lr": 6.994168128694246e-05} +{"train_loss": 0.013377346098423004, "grad_norm": 0.05745943263173103, "global_step": 19579, "epoch": 184, "lr": 6.9938963408823e-05} +{"train_loss": 0.01720595173537731, "grad_norm": 0.13987275958061218, "global_step": 19580, "epoch": 184, "lr": 6.993624546064667e-05} +{"train_loss": 0.020201100036501884, "grad_norm": 0.07178701460361481, "global_step": 19581, "epoch": 184, "lr": 6.993352744242307e-05} +{"train_loss": 0.0151157071813941, "grad_norm": 0.1469395011663437, "global_step": 19582, "epoch": 184, "lr": 6.993080935416172e-05} +{"train_loss": 0.016907688230276108, "grad_norm": 0.20676375925540924, "global_step": 19583, "epoch": 184, "lr": 6.992809119587216e-05} +{"train_loss": 0.01706133410334587, "grad_norm": 0.06609804183244705, "global_step": 19584, "epoch": 184, "lr": 6.992537296756396e-05} +{"train_loss": 0.015311017632484436, "grad_norm": 0.06291870772838593, "global_step": 19585, "epoch": 184, "lr": 6.992265466924666e-05} +{"train_loss": 0.01646551303565502, "grad_norm": 0.06868167966604233, "global_step": 19586, "epoch": 184, "lr": 6.991993630092982e-05} +{"train_loss": 0.014925898984074593, "grad_norm": 0.08491859585046768, "global_step": 19587, "epoch": 184, "lr": 6.991721786262299e-05} +{"train_loss": 0.016109637916088104, "grad_norm": 0.0679563507437706, "global_step": 19588, "epoch": 184, "lr": 6.991449935433572e-05} +{"train_loss": 0.016931142657995224, "grad_norm": 0.0598091222345829, "global_step": 19589, "epoch": 184, "lr": 6.991178077607756e-05} +{"train_loss": 0.017338063567876816, "grad_norm": 0.05279885604977608, "global_step": 19590, "epoch": 184, "lr": 6.990906212785803e-05} +{"train_loss": 0.012876897118985653, "grad_norm": 0.06055154278874397, "global_step": 19591, "epoch": 184, "lr": 6.990634340968676e-05} +{"train_loss": 0.017363935708999634, "grad_norm": 0.07892268151044846, "global_step": 19592, "epoch": 184, "lr": 6.990362462157324e-05} +{"train_loss": 0.015153858810663223, "grad_norm": 0.09222376346588135, "global_step": 19593, "epoch": 184, "lr": 6.990090576352703e-05} +{"train_loss": 0.01591372862458229, "grad_norm": 0.13491050899028778, "global_step": 19594, "epoch": 184, "lr": 6.98981868355577e-05} +{"train_loss": 0.016621839255094528, "grad_norm": 0.13695219159126282, "global_step": 19595, "epoch": 184, "lr": 6.989546783767478e-05} +{"train_loss": 0.017540769651532173, "grad_norm": 0.07689894735813141, "global_step": 19596, "epoch": 184, "lr": 6.989274876988786e-05} +{"train_loss": 0.018242165446281433, "grad_norm": 0.08980947732925415, "global_step": 19597, "epoch": 184, "lr": 6.989002963220646e-05} +{"train_loss": 0.017746280878782272, "grad_norm": 0.062273282557725906, "global_step": 19598, "epoch": 184, "lr": 6.988731042464014e-05} +{"train_loss": 0.015430711209774017, "grad_norm": 0.08988196402788162, "global_step": 19599, "epoch": 184, "lr": 6.988459114719849e-05} +{"train_loss": 0.0162222720682621, "grad_norm": 0.10174383223056793, "global_step": 19600, "epoch": 184, "lr": 6.9881871799891e-05} +{"train_loss": 0.01393246278166771, "grad_norm": 0.06519675999879837, "global_step": 19601, "epoch": 184, "lr": 6.987915238272726e-05} +{"train_loss": 0.015311750583350658, "grad_norm": 0.0764462798833847, "global_step": 19602, "epoch": 184, "lr": 6.987643289571685e-05} +{"train_loss": 0.01837756857275963, "grad_norm": 0.11994964629411697, "global_step": 19603, "epoch": 184, "lr": 6.987371333886928e-05} +{"train_loss": 0.013867232948541641, "grad_norm": 0.07320252805948257, "global_step": 19604, "epoch": 184, "lr": 6.987099371219413e-05} +{"train_loss": 0.017851876094937325, "grad_norm": 0.17264701426029205, "global_step": 19605, "epoch": 184, "lr": 6.986827401570095e-05} +{"train_loss": 0.015049660578370094, "grad_norm": 0.05743706226348877, "global_step": 19606, "epoch": 184, "lr": 6.986555424939929e-05} +{"train_loss": 0.01631053537130356, "grad_norm": 0.08223137259483337, "global_step": 19607, "epoch": 184, "lr": 6.98628344132987e-05} +{"train_loss": 0.0167897529900074, "grad_norm": 0.06966038048267365, "global_step": 19608, "epoch": 184, "lr": 6.986011450740878e-05} +{"train_loss": 0.016162452252827725, "grad_norm": 0.07464909553527832, "global_step": 19609, "epoch": 184, "lr": 6.985739453173903e-05, "val_loss": 0.019077487289905548} +{"train_loss": 0.01947709545493126, "grad_norm": 0.09244270622730255, "global_step": 19610, "epoch": 185, "lr": 6.985467448629905e-05} +{"train_loss": 0.01423834078013897, "grad_norm": 0.051556121557950974, "global_step": 19611, "epoch": 185, "lr": 6.985195437109838e-05} +{"train_loss": 0.015233156271278858, "grad_norm": 0.06936296075582504, "global_step": 19612, "epoch": 185, "lr": 6.984923418614655e-05} +{"train_loss": 0.015328370966017246, "grad_norm": 0.08316421508789062, "global_step": 19613, "epoch": 185, "lr": 6.984651393145315e-05} +{"train_loss": 0.018392512574791908, "grad_norm": 0.0899483785033226, "global_step": 19614, "epoch": 185, "lr": 6.984379360702774e-05} +{"train_loss": 0.015173621475696564, "grad_norm": 0.06020955368876457, "global_step": 19615, "epoch": 185, "lr": 6.984107321287988e-05} +{"train_loss": 0.017489250749349594, "grad_norm": 0.09386066347360611, "global_step": 19616, "epoch": 185, "lr": 6.983835274901909e-05} +{"train_loss": 0.018708916381001472, "grad_norm": 0.10408459603786469, "global_step": 19617, "epoch": 185, "lr": 6.983563221545497e-05} +{"train_loss": 0.01391457486897707, "grad_norm": 0.04829808697104454, "global_step": 19618, "epoch": 185, "lr": 6.983291161219707e-05} +{"train_loss": 0.017967507243156433, "grad_norm": 0.09365382045507431, "global_step": 19619, "epoch": 185, "lr": 6.983019093925493e-05} +{"train_loss": 0.016283374279737473, "grad_norm": 0.06000503897666931, "global_step": 19620, "epoch": 185, "lr": 6.982747019663813e-05} +{"train_loss": 0.01854492351412773, "grad_norm": 0.09418933838605881, "global_step": 19621, "epoch": 185, "lr": 6.982474938435623e-05} +{"train_loss": 0.017630785703659058, "grad_norm": 0.05493147671222687, "global_step": 19622, "epoch": 185, "lr": 6.982202850241876e-05} +{"train_loss": 0.017404500395059586, "grad_norm": 0.14865824580192566, "global_step": 19623, "epoch": 185, "lr": 6.981930755083535e-05} +{"train_loss": 0.01783251389861107, "grad_norm": 0.053191330283880234, "global_step": 19624, "epoch": 185, "lr": 6.981658652961547e-05} +{"train_loss": 0.013464565388858318, "grad_norm": 0.07128264755010605, "global_step": 19625, "epoch": 185, "lr": 6.981386543876874e-05} +{"train_loss": 0.014821640215814114, "grad_norm": 0.05077338591217995, "global_step": 19626, "epoch": 185, "lr": 6.981114427830469e-05} +{"train_loss": 0.022229254245758057, "grad_norm": 0.06382593512535095, "global_step": 19627, "epoch": 185, "lr": 6.98084230482329e-05} +{"train_loss": 0.013722512871026993, "grad_norm": 0.12113835662603378, "global_step": 19628, "epoch": 185, "lr": 6.980570174856292e-05} +{"train_loss": 0.01398387923836708, "grad_norm": 0.04858161881566048, "global_step": 19629, "epoch": 185, "lr": 6.980298037930433e-05} +{"train_loss": 0.014405861496925354, "grad_norm": 0.06082725152373314, "global_step": 19630, "epoch": 185, "lr": 6.980025894046667e-05} +{"train_loss": 0.013929284177720547, "grad_norm": 0.0711049884557724, "global_step": 19631, "epoch": 185, "lr": 6.979753743205952e-05} +{"train_loss": 0.01897387020289898, "grad_norm": 0.08493251353502274, "global_step": 19632, "epoch": 185, "lr": 6.979481585409243e-05} +{"train_loss": 0.013610944151878357, "grad_norm": 0.04072684794664383, "global_step": 19633, "epoch": 185, "lr": 6.979209420657497e-05} +{"train_loss": 0.01506015844643116, "grad_norm": 0.13979612290859222, "global_step": 19634, "epoch": 185, "lr": 6.978937248951668e-05} +{"train_loss": 0.013668471947312355, "grad_norm": 0.04968777298927307, "global_step": 19635, "epoch": 185, "lr": 6.978665070292715e-05} +{"train_loss": 0.015595168806612492, "grad_norm": 0.06566713005304337, "global_step": 19636, "epoch": 185, "lr": 6.978392884681596e-05} +{"train_loss": 0.01627374440431595, "grad_norm": 0.06507029384374619, "global_step": 19637, "epoch": 185, "lr": 6.978120692119261e-05} +{"train_loss": 0.016432486474514008, "grad_norm": 0.08142787218093872, "global_step": 19638, "epoch": 185, "lr": 6.977848492606672e-05} +{"train_loss": 0.0182607714086771, "grad_norm": 0.0761771947145462, "global_step": 19639, "epoch": 185, "lr": 6.977576286144784e-05} +{"train_loss": 0.01650593988597393, "grad_norm": 0.09243384748697281, "global_step": 19640, "epoch": 185, "lr": 6.977304072734551e-05} +{"train_loss": 0.015104823745787144, "grad_norm": 0.09637568891048431, "global_step": 19641, "epoch": 185, "lr": 6.977031852376932e-05} +{"train_loss": 0.015546228736639023, "grad_norm": 0.1084364503622055, "global_step": 19642, "epoch": 185, "lr": 6.976759625072886e-05} +{"train_loss": 0.01633676514029503, "grad_norm": 0.060187697410583496, "global_step": 19643, "epoch": 185, "lr": 6.976487390823363e-05} +{"train_loss": 0.019383825361728668, "grad_norm": 0.06922852247953415, "global_step": 19644, "epoch": 185, "lr": 6.976215149629324e-05} +{"train_loss": 0.013508390635251999, "grad_norm": 0.09152846783399582, "global_step": 19645, "epoch": 185, "lr": 6.975942901491723e-05} +{"train_loss": 0.016172390431165695, "grad_norm": 0.06715220212936401, "global_step": 19646, "epoch": 185, "lr": 6.975670646411518e-05} +{"train_loss": 0.015653178095817566, "grad_norm": 0.10277801007032394, "global_step": 19647, "epoch": 185, "lr": 6.975398384389669e-05} +{"train_loss": 0.017851270735263824, "grad_norm": 0.08750848472118378, "global_step": 19648, "epoch": 185, "lr": 6.975126115427125e-05} +{"train_loss": 0.015373588539659977, "grad_norm": 0.1295953243970871, "global_step": 19649, "epoch": 185, "lr": 6.974853839524849e-05} +{"train_loss": 0.017416849732398987, "grad_norm": 0.045408498495817184, "global_step": 19650, "epoch": 185, "lr": 6.974581556683796e-05} +{"train_loss": 0.014169180765748024, "grad_norm": 0.05677523463964462, "global_step": 19651, "epoch": 185, "lr": 6.97430926690492e-05} +{"train_loss": 0.016239795833826065, "grad_norm": 0.07188834995031357, "global_step": 19652, "epoch": 185, "lr": 6.974036970189181e-05} +{"train_loss": 0.015213808976113796, "grad_norm": 0.08345651626586914, "global_step": 19653, "epoch": 185, "lr": 6.973764666537534e-05} +{"train_loss": 0.016079260036349297, "grad_norm": 0.08632184565067291, "global_step": 19654, "epoch": 185, "lr": 6.973492355950938e-05} +{"train_loss": 0.01604994758963585, "grad_norm": 0.07689925283193588, "global_step": 19655, "epoch": 185, "lr": 6.973220038430347e-05} +{"train_loss": 0.01675930619239807, "grad_norm": 0.15739639103412628, "global_step": 19656, "epoch": 185, "lr": 6.972947713976719e-05} +{"train_loss": 0.013380566611886024, "grad_norm": 0.06471871584653854, "global_step": 19657, "epoch": 185, "lr": 6.97267538259101e-05} +{"train_loss": 0.016133978962898254, "grad_norm": 0.06292903423309326, "global_step": 19658, "epoch": 185, "lr": 6.972403044274178e-05} +{"train_loss": 0.017907653003931046, "grad_norm": 0.12866975367069244, "global_step": 19659, "epoch": 185, "lr": 6.97213069902718e-05} +{"train_loss": 0.016614120453596115, "grad_norm": 0.06348021328449249, "global_step": 19660, "epoch": 185, "lr": 6.971858346850973e-05} +{"train_loss": 0.016662409529089928, "grad_norm": 0.08110078424215317, "global_step": 19661, "epoch": 185, "lr": 6.971585987746512e-05} +{"train_loss": 0.01577886939048767, "grad_norm": 0.06312251091003418, "global_step": 19662, "epoch": 185, "lr": 6.971313621714756e-05} +{"train_loss": 0.014462824910879135, "grad_norm": 0.05965469032526016, "global_step": 19663, "epoch": 185, "lr": 6.97104124875666e-05} +{"train_loss": 0.014352038502693176, "grad_norm": 0.07716476172208786, "global_step": 19664, "epoch": 185, "lr": 6.970768868873184e-05} +{"train_loss": 0.020630154758691788, "grad_norm": 0.08330216258764267, "global_step": 19665, "epoch": 185, "lr": 6.970496482065282e-05} +{"train_loss": 0.015332725830376148, "grad_norm": 0.0826607272028923, "global_step": 19666, "epoch": 185, "lr": 6.970224088333912e-05} +{"train_loss": 0.01566358096897602, "grad_norm": 0.051102764904499054, "global_step": 19667, "epoch": 185, "lr": 6.969951687680032e-05} +{"train_loss": 0.014939939603209496, "grad_norm": 0.08531420677900314, "global_step": 19668, "epoch": 185, "lr": 6.9696792801046e-05} +{"train_loss": 0.016481436789035797, "grad_norm": 0.050287116318941116, "global_step": 19669, "epoch": 185, "lr": 6.96940686560857e-05} +{"train_loss": 0.01576259732246399, "grad_norm": 0.052033714950084686, "global_step": 19670, "epoch": 185, "lr": 6.9691344441929e-05} +{"train_loss": 0.017239905893802643, "grad_norm": 0.1102413609623909, "global_step": 19671, "epoch": 185, "lr": 6.96886201585855e-05} +{"train_loss": 0.022504862397909164, "grad_norm": 0.06952173262834549, "global_step": 19672, "epoch": 185, "lr": 6.968589580606472e-05} +{"train_loss": 0.018364548683166504, "grad_norm": 0.06212928518652916, "global_step": 19673, "epoch": 185, "lr": 6.96831713843763e-05} +{"train_loss": 0.016616830602288246, "grad_norm": 0.0846366211771965, "global_step": 19674, "epoch": 185, "lr": 6.968044689352975e-05} +{"train_loss": 0.016139879822731018, "grad_norm": 0.05785083398222923, "global_step": 19675, "epoch": 185, "lr": 6.967772233353466e-05} +{"train_loss": 0.01702716574072838, "grad_norm": 0.08521907031536102, "global_step": 19676, "epoch": 185, "lr": 6.967499770440065e-05} +{"train_loss": 0.01655508205294609, "grad_norm": 0.08721954375505447, "global_step": 19677, "epoch": 185, "lr": 6.967227300613722e-05} +{"train_loss": 0.01641177013516426, "grad_norm": 0.10157045722007751, "global_step": 19678, "epoch": 185, "lr": 6.9669548238754e-05} +{"train_loss": 0.016946112737059593, "grad_norm": 0.0582953505218029, "global_step": 19679, "epoch": 185, "lr": 6.966682340226053e-05} +{"train_loss": 0.014691753312945366, "grad_norm": 0.07649726420640945, "global_step": 19680, "epoch": 185, "lr": 6.966409849666639e-05} +{"train_loss": 0.014923046343028545, "grad_norm": 0.08693370223045349, "global_step": 19681, "epoch": 185, "lr": 6.966137352198118e-05} +{"train_loss": 0.01650882512331009, "grad_norm": 0.06429551541805267, "global_step": 19682, "epoch": 185, "lr": 6.965864847821443e-05} +{"train_loss": 0.015283321961760521, "grad_norm": 0.11612887680530548, "global_step": 19683, "epoch": 185, "lr": 6.965592336537576e-05} +{"train_loss": 0.016890961676836014, "grad_norm": 0.14191196858882904, "global_step": 19684, "epoch": 185, "lr": 6.965319818347471e-05} +{"train_loss": 0.015065194107592106, "grad_norm": 0.08143480867147446, "global_step": 19685, "epoch": 185, "lr": 6.965047293252088e-05} +{"train_loss": 0.015496923588216305, "grad_norm": 0.07163768261671066, "global_step": 19686, "epoch": 185, "lr": 6.964774761252381e-05} +{"train_loss": 0.014692168682813644, "grad_norm": 0.051298752427101135, "global_step": 19687, "epoch": 185, "lr": 6.964502222349312e-05} +{"train_loss": 0.017368096858263016, "grad_norm": 0.07741349190473557, "global_step": 19688, "epoch": 185, "lr": 6.964229676543836e-05} +{"train_loss": 0.015125608071684837, "grad_norm": 0.07570477575063705, "global_step": 19689, "epoch": 185, "lr": 6.963957123836911e-05} +{"train_loss": 0.015261044725775719, "grad_norm": 0.13505858182907104, "global_step": 19690, "epoch": 185, "lr": 6.963684564229495e-05} +{"train_loss": 0.01689777337014675, "grad_norm": 0.12820640206336975, "global_step": 19691, "epoch": 185, "lr": 6.963411997722547e-05} +{"train_loss": 0.016554607078433037, "grad_norm": 0.07043357193470001, "global_step": 19692, "epoch": 185, "lr": 6.96313942431702e-05} +{"train_loss": 0.013240818865597248, "grad_norm": 0.05293857678771019, "global_step": 19693, "epoch": 185, "lr": 6.962866844013876e-05} +{"train_loss": 0.014194884337484837, "grad_norm": 0.05270552262663841, "global_step": 19694, "epoch": 185, "lr": 6.962594256814072e-05} +{"train_loss": 0.01510840468108654, "grad_norm": 0.09464924037456512, "global_step": 19695, "epoch": 185, "lr": 6.962321662718565e-05} +{"train_loss": 0.016119448468089104, "grad_norm": 0.10933278501033783, "global_step": 19696, "epoch": 185, "lr": 6.962049061728314e-05} +{"train_loss": 0.015551351010799408, "grad_norm": 0.15373757481575012, "global_step": 19697, "epoch": 185, "lr": 6.961776453844275e-05} +{"train_loss": 0.015126615762710571, "grad_norm": 0.07857881486415863, "global_step": 19698, "epoch": 185, "lr": 6.961503839067407e-05} +{"train_loss": 0.018476177006959915, "grad_norm": 0.12542766332626343, "global_step": 19699, "epoch": 185, "lr": 6.961231217398668e-05} +{"train_loss": 0.01705188676714897, "grad_norm": 0.14095717668533325, "global_step": 19700, "epoch": 185, "lr": 6.960958588839013e-05} +{"train_loss": 0.014041915535926819, "grad_norm": 0.11282780766487122, "global_step": 19701, "epoch": 185, "lr": 6.960685953389405e-05} +{"train_loss": 0.014151612296700478, "grad_norm": 0.04866831749677658, "global_step": 19702, "epoch": 185, "lr": 6.960413311050799e-05} +{"train_loss": 0.0180643443018198, "grad_norm": 0.08028282970190048, "global_step": 19703, "epoch": 185, "lr": 6.960140661824152e-05} +{"train_loss": 0.015104096382856369, "grad_norm": 0.08963628113269806, "global_step": 19704, "epoch": 185, "lr": 6.959868005710422e-05} +{"train_loss": 0.01563112996518612, "grad_norm": 0.07816334068775177, "global_step": 19705, "epoch": 185, "lr": 6.959595342710571e-05} +{"train_loss": 0.018406745046377182, "grad_norm": 0.07460926473140717, "global_step": 19706, "epoch": 185, "lr": 6.959322672825552e-05} +{"train_loss": 0.016160162165760994, "grad_norm": 0.08898323774337769, "global_step": 19707, "epoch": 185, "lr": 6.959049996056325e-05} +{"train_loss": 0.014608664438128471, "grad_norm": 0.08173906058073044, "global_step": 19708, "epoch": 185, "lr": 6.95877731240385e-05} +{"train_loss": 0.01657787710428238, "grad_norm": 0.1198798343539238, "global_step": 19709, "epoch": 185, "lr": 6.958504621869084e-05} +{"train_loss": 0.015077194198966026, "grad_norm": 0.053528279066085815, "global_step": 19710, "epoch": 185, "lr": 6.958231924452983e-05} +{"train_loss": 0.015206040814518929, "grad_norm": 0.059836484491825104, "global_step": 19711, "epoch": 185, "lr": 6.957959220156507e-05} +{"train_loss": 0.018127355724573135, "grad_norm": 0.10112208873033524, "global_step": 19712, "epoch": 185, "lr": 6.957686508980615e-05} +{"train_loss": 0.015385889448225498, "grad_norm": 0.10384335368871689, "global_step": 19713, "epoch": 185, "lr": 6.957413790926262e-05} +{"train_loss": 0.013958419673144817, "grad_norm": 0.057992275804281235, "global_step": 19714, "epoch": 185, "lr": 6.957141065994408e-05} +{"train_loss": 0.016137355692544073, "grad_norm": 0.07756262272596359, "global_step": 19715, "epoch": 185, "lr": 6.956868334186013e-05, "val_loss": 0.019986381754279137} +{"train_loss": 0.01585226319730282, "grad_norm": 0.06839148700237274, "global_step": 19716, "epoch": 186, "lr": 6.956595595502032e-05} +{"train_loss": 0.016035880893468857, "grad_norm": 0.06884024292230606, "global_step": 19717, "epoch": 186, "lr": 6.956322849943427e-05} +{"train_loss": 0.014611390419304371, "grad_norm": 0.053790099918842316, "global_step": 19718, "epoch": 186, "lr": 6.956050097511152e-05} +{"train_loss": 0.01536042895168066, "grad_norm": 0.18829061090946198, "global_step": 19719, "epoch": 186, "lr": 6.955777338206169e-05} +{"train_loss": 0.0152263343334198, "grad_norm": 0.11044315993785858, "global_step": 19720, "epoch": 186, "lr": 6.955504572029437e-05} +{"train_loss": 0.015665613114833832, "grad_norm": 0.10536449402570724, "global_step": 19721, "epoch": 186, "lr": 6.955231798981909e-05} +{"train_loss": 0.016920559108257294, "grad_norm": 0.08171537518501282, "global_step": 19722, "epoch": 186, "lr": 6.954959019064548e-05} +{"train_loss": 0.017655912786722183, "grad_norm": 0.07079502195119858, "global_step": 19723, "epoch": 186, "lr": 6.954686232278311e-05} +{"train_loss": 0.015315523371100426, "grad_norm": 0.061618782579898834, "global_step": 19724, "epoch": 186, "lr": 6.954413438624157e-05} +{"train_loss": 0.016462380066514015, "grad_norm": 0.05033478885889053, "global_step": 19725, "epoch": 186, "lr": 6.954140638103043e-05} +{"train_loss": 0.01719590649008751, "grad_norm": 0.06217411160469055, "global_step": 19726, "epoch": 186, "lr": 6.953867830715931e-05} +{"train_loss": 0.015034714713692665, "grad_norm": 0.07127174735069275, "global_step": 19727, "epoch": 186, "lr": 6.953595016463776e-05} +{"train_loss": 0.014007240533828735, "grad_norm": 0.06117250770330429, "global_step": 19728, "epoch": 186, "lr": 6.953322195347536e-05} +{"train_loss": 0.017920028418302536, "grad_norm": 0.0913638323545456, "global_step": 19729, "epoch": 186, "lr": 6.953049367368173e-05} +{"train_loss": 0.015446927398443222, "grad_norm": 0.06747057288885117, "global_step": 19730, "epoch": 186, "lr": 6.952776532526645e-05} +{"train_loss": 0.016942545771598816, "grad_norm": 0.1127859577536583, "global_step": 19731, "epoch": 186, "lr": 6.952503690823907e-05} +{"train_loss": 0.01471386943012476, "grad_norm": 0.05565556511282921, "global_step": 19732, "epoch": 186, "lr": 6.952230842260923e-05} +{"train_loss": 0.015757597982883453, "grad_norm": 0.06266812980175018, "global_step": 19733, "epoch": 186, "lr": 6.951957986838646e-05} +{"train_loss": 0.017253775149583817, "grad_norm": 0.07462223619222641, "global_step": 19734, "epoch": 186, "lr": 6.951685124558039e-05} +{"train_loss": 0.01707443594932556, "grad_norm": 0.06951602548360825, "global_step": 19735, "epoch": 186, "lr": 6.951412255420058e-05} +{"train_loss": 0.016815682873129845, "grad_norm": 0.05169214680790901, "global_step": 19736, "epoch": 186, "lr": 6.951139379425664e-05} +{"train_loss": 0.017307091504335403, "grad_norm": 0.0651216134428978, "global_step": 19737, "epoch": 186, "lr": 6.950866496575814e-05} +{"train_loss": 0.016190454363822937, "grad_norm": 0.07195398956537247, "global_step": 19738, "epoch": 186, "lr": 6.95059360687147e-05} +{"train_loss": 0.013749017380177975, "grad_norm": 0.06830429285764694, "global_step": 19739, "epoch": 186, "lr": 6.950320710313586e-05} +{"train_loss": 0.01786326989531517, "grad_norm": 0.10353925079107285, "global_step": 19740, "epoch": 186, "lr": 6.950047806903122e-05} +{"train_loss": 0.01907845214009285, "grad_norm": 0.07313872128725052, "global_step": 19741, "epoch": 186, "lr": 6.949774896641042e-05} +{"train_loss": 0.01683439314365387, "grad_norm": 0.1790987253189087, "global_step": 19742, "epoch": 186, "lr": 6.949501979528298e-05} +{"train_loss": 0.016556859016418457, "grad_norm": 0.09365720301866531, "global_step": 19743, "epoch": 186, "lr": 6.949229055565852e-05} +{"train_loss": 0.014594141393899918, "grad_norm": 0.093743234872818, "global_step": 19744, "epoch": 186, "lr": 6.948956124754665e-05} +{"train_loss": 0.01545364037156105, "grad_norm": 0.09542831778526306, "global_step": 19745, "epoch": 186, "lr": 6.94868318709569e-05} +{"train_loss": 0.015489369630813599, "grad_norm": 0.0739564448595047, "global_step": 19746, "epoch": 186, "lr": 6.948410242589893e-05} +{"train_loss": 0.01614540070295334, "grad_norm": 0.04302285239100456, "global_step": 19747, "epoch": 186, "lr": 6.948137291238228e-05} +{"train_loss": 0.01680513098835945, "grad_norm": 0.060705479234457016, "global_step": 19748, "epoch": 186, "lr": 6.947864333041655e-05} +{"train_loss": 0.017958952113986015, "grad_norm": 0.07070641964673996, "global_step": 19749, "epoch": 186, "lr": 6.947591368001138e-05} +{"train_loss": 0.01582535356283188, "grad_norm": 0.05535869300365448, "global_step": 19750, "epoch": 186, "lr": 6.947318396117628e-05} +{"train_loss": 0.01641177199780941, "grad_norm": 0.053501952439546585, "global_step": 19751, "epoch": 186, "lr": 6.947045417392088e-05} +{"train_loss": 0.015171973034739494, "grad_norm": 0.04880448430776596, "global_step": 19752, "epoch": 186, "lr": 6.946772431825479e-05} +{"train_loss": 0.014201520010828972, "grad_norm": 0.09240023046731949, "global_step": 19753, "epoch": 186, "lr": 6.946499439418758e-05} +{"train_loss": 0.014871213585138321, "grad_norm": 0.0796346440911293, "global_step": 19754, "epoch": 186, "lr": 6.946226440172884e-05} +{"train_loss": 0.012532556429505348, "grad_norm": 0.061986472457647324, "global_step": 19755, "epoch": 186, "lr": 6.945953434088816e-05} +{"train_loss": 0.015503356233239174, "grad_norm": 0.07286746054887772, "global_step": 19756, "epoch": 186, "lr": 6.945680421167515e-05} +{"train_loss": 0.016899101436138153, "grad_norm": 0.07543069124221802, "global_step": 19757, "epoch": 186, "lr": 6.945407401409939e-05} +{"train_loss": 0.014729676768183708, "grad_norm": 0.13703103363513947, "global_step": 19758, "epoch": 186, "lr": 6.945134374817048e-05} +{"train_loss": 0.017072102054953575, "grad_norm": 0.0968591570854187, "global_step": 19759, "epoch": 186, "lr": 6.9448613413898e-05} +{"train_loss": 0.01627396047115326, "grad_norm": 0.13768920302391052, "global_step": 19760, "epoch": 186, "lr": 6.944588301129155e-05} +{"train_loss": 0.01363280788064003, "grad_norm": 0.12876352667808533, "global_step": 19761, "epoch": 186, "lr": 6.944315254036073e-05} +{"train_loss": 0.01623060554265976, "grad_norm": 0.08767739683389664, "global_step": 19762, "epoch": 186, "lr": 6.944042200111511e-05} +{"train_loss": 0.01551416702568531, "grad_norm": 0.08786269277334213, "global_step": 19763, "epoch": 186, "lr": 6.943769139356431e-05} +{"train_loss": 0.015084335580468178, "grad_norm": 0.05561332777142525, "global_step": 19764, "epoch": 186, "lr": 6.943496071771792e-05} +{"train_loss": 0.015873204916715622, "grad_norm": 0.06372400373220444, "global_step": 19765, "epoch": 186, "lr": 6.943222997358554e-05} +{"train_loss": 0.02035048045217991, "grad_norm": 0.05854079872369766, "global_step": 19766, "epoch": 186, "lr": 6.942949916117674e-05} +{"train_loss": 0.015635166317224503, "grad_norm": 0.07123541086912155, "global_step": 19767, "epoch": 186, "lr": 6.942676828050113e-05} +{"train_loss": 0.01832611672580242, "grad_norm": 0.07327951490879059, "global_step": 19768, "epoch": 186, "lr": 6.942403733156832e-05} +{"train_loss": 0.015364399179816246, "grad_norm": 0.07540963590145111, "global_step": 19769, "epoch": 186, "lr": 6.942130631438786e-05} +{"train_loss": 0.01778346300125122, "grad_norm": 0.1118173599243164, "global_step": 19770, "epoch": 186, "lr": 6.94185752289694e-05} +{"train_loss": 0.016460632905364037, "grad_norm": 0.07154762744903564, "global_step": 19771, "epoch": 186, "lr": 6.941584407532251e-05} +{"train_loss": 0.017933249473571777, "grad_norm": 0.10655227303504944, "global_step": 19772, "epoch": 186, "lr": 6.941311285345677e-05} +{"train_loss": 0.01573445089161396, "grad_norm": 0.09360554814338684, "global_step": 19773, "epoch": 186, "lr": 6.941038156338182e-05} +{"train_loss": 0.015601173043251038, "grad_norm": 0.0627315491437912, "global_step": 19774, "epoch": 186, "lr": 6.940765020510721e-05} +{"train_loss": 0.01579046994447708, "grad_norm": 0.09217332303524017, "global_step": 19775, "epoch": 186, "lr": 6.940491877864256e-05} +{"train_loss": 0.013742946088314056, "grad_norm": 0.08460444211959839, "global_step": 19776, "epoch": 186, "lr": 6.940218728399747e-05} +{"train_loss": 0.017946770414710045, "grad_norm": 0.06338677555322647, "global_step": 19777, "epoch": 186, "lr": 6.939945572118154e-05} +{"train_loss": 0.015788724645972252, "grad_norm": 0.15275605022907257, "global_step": 19778, "epoch": 186, "lr": 6.939672409020435e-05} +{"train_loss": 0.015725482255220413, "grad_norm": 0.0841447114944458, "global_step": 19779, "epoch": 186, "lr": 6.939399239107552e-05} +{"train_loss": 0.018572144210338593, "grad_norm": 0.05206538364291191, "global_step": 19780, "epoch": 186, "lr": 6.939126062380462e-05} +{"train_loss": 0.013186869211494923, "grad_norm": 0.057169895619153976, "global_step": 19781, "epoch": 186, "lr": 6.938852878840126e-05} +{"train_loss": 0.018361348658800125, "grad_norm": 0.09158319234848022, "global_step": 19782, "epoch": 186, "lr": 6.938579688487504e-05} +{"train_loss": 0.018007470294833183, "grad_norm": 0.08910365402698517, "global_step": 19783, "epoch": 186, "lr": 6.938306491323558e-05} +{"train_loss": 0.017140325158834457, "grad_norm": 0.08710162341594696, "global_step": 19784, "epoch": 186, "lr": 6.938033287349243e-05} +{"train_loss": 0.015225788578391075, "grad_norm": 0.06060677766799927, "global_step": 19785, "epoch": 186, "lr": 6.937760076565525e-05} +{"train_loss": 0.01334584690630436, "grad_norm": 0.08188869059085846, "global_step": 19786, "epoch": 186, "lr": 6.937486858973359e-05} +{"train_loss": 0.016380738466978073, "grad_norm": 0.057942263782024384, "global_step": 19787, "epoch": 186, "lr": 6.937213634573707e-05} +{"train_loss": 0.015675878152251244, "grad_norm": 0.14890965819358826, "global_step": 19788, "epoch": 186, "lr": 6.936940403367528e-05} +{"train_loss": 0.016602452844381332, "grad_norm": 0.05157826095819473, "global_step": 19789, "epoch": 186, "lr": 6.936667165355783e-05} +{"train_loss": 0.012226911261677742, "grad_norm": 0.0966210886836052, "global_step": 19790, "epoch": 186, "lr": 6.936393920539433e-05} +{"train_loss": 0.01673600636422634, "grad_norm": 0.07335729897022247, "global_step": 19791, "epoch": 186, "lr": 6.936120668919437e-05} +{"train_loss": 0.01504237949848175, "grad_norm": 0.12781551480293274, "global_step": 19792, "epoch": 186, "lr": 6.935847410496755e-05} +{"train_loss": 0.018437134101986885, "grad_norm": 0.11509678512811661, "global_step": 19793, "epoch": 186, "lr": 6.935574145272345e-05} +{"train_loss": 0.017614832147955894, "grad_norm": 0.06036429852247238, "global_step": 19794, "epoch": 186, "lr": 6.935300873247171e-05} +{"train_loss": 0.01752585917711258, "grad_norm": 0.10191946476697922, "global_step": 19795, "epoch": 186, "lr": 6.935027594422192e-05} +{"train_loss": 0.01631603203713894, "grad_norm": 0.09690625965595245, "global_step": 19796, "epoch": 186, "lr": 6.934754308798365e-05} +{"train_loss": 0.017087429761886597, "grad_norm": 0.05892469361424446, "global_step": 19797, "epoch": 186, "lr": 6.934481016376656e-05} +{"train_loss": 0.01716928370296955, "grad_norm": 0.08595694601535797, "global_step": 19798, "epoch": 186, "lr": 6.934207717158019e-05} +{"train_loss": 0.016119129955768585, "grad_norm": 0.143662229180336, "global_step": 19799, "epoch": 186, "lr": 6.93393441114342e-05} +{"train_loss": 0.015566624701023102, "grad_norm": 0.07366931438446045, "global_step": 19800, "epoch": 186, "lr": 6.933661098333814e-05} +{"train_loss": 0.014516896568238735, "grad_norm": 0.061452317982912064, "global_step": 19801, "epoch": 186, "lr": 6.933387778730165e-05} +{"train_loss": 0.01412918884307146, "grad_norm": 0.07513534277677536, "global_step": 19802, "epoch": 186, "lr": 6.933114452333433e-05} +{"train_loss": 0.017605913802981377, "grad_norm": 0.14999711513519287, "global_step": 19803, "epoch": 186, "lr": 6.932841119144576e-05} +{"train_loss": 0.014985837042331696, "grad_norm": 0.044652096927165985, "global_step": 19804, "epoch": 186, "lr": 6.932567779164555e-05} +{"train_loss": 0.014498548582196236, "grad_norm": 0.0733148604631424, "global_step": 19805, "epoch": 186, "lr": 6.932294432394334e-05} +{"train_loss": 0.01779954880475998, "grad_norm": 0.06024729087948799, "global_step": 19806, "epoch": 186, "lr": 6.93202107883487e-05} +{"train_loss": 0.01804012805223465, "grad_norm": 0.06796102970838547, "global_step": 19807, "epoch": 186, "lr": 6.931747718487122e-05} +{"train_loss": 0.015148436650633812, "grad_norm": 0.07774668186903, "global_step": 19808, "epoch": 186, "lr": 6.931474351352055e-05} +{"train_loss": 0.01779867894947529, "grad_norm": 0.12478790432214737, "global_step": 19809, "epoch": 186, "lr": 6.931200977430626e-05} +{"train_loss": 0.014622770249843597, "grad_norm": 0.11767309904098511, "global_step": 19810, "epoch": 186, "lr": 6.930927596723796e-05} +{"train_loss": 0.01520280446857214, "grad_norm": 0.059591349214315414, "global_step": 19811, "epoch": 186, "lr": 6.930654209232528e-05} +{"train_loss": 0.015192165970802307, "grad_norm": 0.11391499638557434, "global_step": 19812, "epoch": 186, "lr": 6.93038081495778e-05} +{"train_loss": 0.014728359878063202, "grad_norm": 0.07114030420780182, "global_step": 19813, "epoch": 186, "lr": 6.930107413900514e-05} +{"train_loss": 0.016371950507164, "grad_norm": 0.123503178358078, "global_step": 19814, "epoch": 186, "lr": 6.929834006061689e-05} +{"train_loss": 0.01687488704919815, "grad_norm": 0.07488328218460083, "global_step": 19815, "epoch": 186, "lr": 6.929560591442266e-05} +{"train_loss": 0.016575884073972702, "grad_norm": 0.07385613024234772, "global_step": 19816, "epoch": 186, "lr": 6.929287170043207e-05} +{"train_loss": 0.015534870326519012, "grad_norm": 0.08245442807674408, "global_step": 19817, "epoch": 186, "lr": 6.929013741865471e-05} +{"train_loss": 0.017433583736419678, "grad_norm": 0.09366150200366974, "global_step": 19818, "epoch": 186, "lr": 6.92874030691002e-05} +{"train_loss": 0.019207099452614784, "grad_norm": 0.0647813156247139, "global_step": 19819, "epoch": 186, "lr": 6.928466865177815e-05} +{"train_loss": 0.01600079983472824, "grad_norm": 0.10894694924354553, "global_step": 19820, "epoch": 186, "lr": 6.928193416669816e-05} +{"train_loss": 0.016120705410908414, "grad_norm": 0.13065172731876373, "global_step": 19821, "epoch": 186, "lr": 6.927919961386984e-05, "val_loss": 0.01917235180735588} +{"train_loss": 0.0156018678098917, "grad_norm": 0.08483599126338959, "global_step": 19822, "epoch": 187, "lr": 6.927646499330279e-05} +{"train_loss": 0.01363968476653099, "grad_norm": 0.05319030582904816, "global_step": 19823, "epoch": 187, "lr": 6.927373030500663e-05} +{"train_loss": 0.014501695521175861, "grad_norm": 0.05145733430981636, "global_step": 19824, "epoch": 187, "lr": 6.927099554899096e-05} +{"train_loss": 0.014671813696622849, "grad_norm": 0.04962528496980667, "global_step": 19825, "epoch": 187, "lr": 6.92682607252654e-05} +{"train_loss": 0.014700084924697876, "grad_norm": 0.05915370583534241, "global_step": 19826, "epoch": 187, "lr": 6.926552583383955e-05} +{"train_loss": 0.016909725964069366, "grad_norm": 0.10124822705984116, "global_step": 19827, "epoch": 187, "lr": 6.9262790874723e-05} +{"train_loss": 0.015797406435012817, "grad_norm": 0.10232891887426376, "global_step": 19828, "epoch": 187, "lr": 6.92600558479254e-05} +{"train_loss": 0.017619313672184944, "grad_norm": 0.052383195608854294, "global_step": 19829, "epoch": 187, "lr": 6.925732075345634e-05} +{"train_loss": 0.01593206822872162, "grad_norm": 0.06691849231719971, "global_step": 19830, "epoch": 187, "lr": 6.925458559132543e-05} +{"train_loss": 0.014187818393111229, "grad_norm": 0.07645279169082642, "global_step": 19831, "epoch": 187, "lr": 6.925185036154227e-05} +{"train_loss": 0.014828141778707504, "grad_norm": 0.0790068730711937, "global_step": 19832, "epoch": 187, "lr": 6.924911506411649e-05} +{"train_loss": 0.014720841310918331, "grad_norm": 0.074264757335186, "global_step": 19833, "epoch": 187, "lr": 6.924637969905768e-05} +{"train_loss": 0.01610957272350788, "grad_norm": 0.08930182456970215, "global_step": 19834, "epoch": 187, "lr": 6.924364426637548e-05} +{"train_loss": 0.01697550341486931, "grad_norm": 0.04938315600156784, "global_step": 19835, "epoch": 187, "lr": 6.924090876607946e-05} +{"train_loss": 0.016933200880885124, "grad_norm": 0.05066047981381416, "global_step": 19836, "epoch": 187, "lr": 6.923817319817927e-05} +{"train_loss": 0.014824498444795609, "grad_norm": 0.08641932904720306, "global_step": 19837, "epoch": 187, "lr": 6.923543756268449e-05} +{"train_loss": 0.019925741478800774, "grad_norm": 0.08948403596878052, "global_step": 19838, "epoch": 187, "lr": 6.923270185960478e-05} +{"train_loss": 0.014492565765976906, "grad_norm": 0.08962316066026688, "global_step": 19839, "epoch": 187, "lr": 6.922996608894968e-05} +{"train_loss": 0.017042415216565132, "grad_norm": 0.14827947318553925, "global_step": 19840, "epoch": 187, "lr": 6.922723025072886e-05} +{"train_loss": 0.014238234609365463, "grad_norm": 0.07379326969385147, "global_step": 19841, "epoch": 187, "lr": 6.922449434495192e-05} +{"train_loss": 0.01628405600786209, "grad_norm": 0.1087275967001915, "global_step": 19842, "epoch": 187, "lr": 6.922175837162847e-05} +{"train_loss": 0.016637474298477173, "grad_norm": 0.09459803253412247, "global_step": 19843, "epoch": 187, "lr": 6.921902233076811e-05} +{"train_loss": 0.016587430611252785, "grad_norm": 0.05818074941635132, "global_step": 19844, "epoch": 187, "lr": 6.921628622238047e-05} +{"train_loss": 0.01557575911283493, "grad_norm": 0.06498367339372635, "global_step": 19845, "epoch": 187, "lr": 6.921355004647515e-05} +{"train_loss": 0.015301375649869442, "grad_norm": 0.07046645134687424, "global_step": 19846, "epoch": 187, "lr": 6.921081380306179e-05} +{"train_loss": 0.01783587783575058, "grad_norm": 0.10076767951250076, "global_step": 19847, "epoch": 187, "lr": 6.920807749214996e-05} +{"train_loss": 0.0161452516913414, "grad_norm": 0.06908009201288223, "global_step": 19848, "epoch": 187, "lr": 6.92053411137493e-05} +{"train_loss": 0.015444265678524971, "grad_norm": 0.10941749066114426, "global_step": 19849, "epoch": 187, "lr": 6.920260466786944e-05} +{"train_loss": 0.015222139656543732, "grad_norm": 0.07214994728565216, "global_step": 19850, "epoch": 187, "lr": 6.919986815451998e-05} +{"train_loss": 0.01693471148610115, "grad_norm": 0.07311838865280151, "global_step": 19851, "epoch": 187, "lr": 6.919713157371051e-05} +{"train_loss": 0.016517940908670425, "grad_norm": 0.07923068851232529, "global_step": 19852, "epoch": 187, "lr": 6.919439492545069e-05} +{"train_loss": 0.015073589980602264, "grad_norm": 0.09594476968050003, "global_step": 19853, "epoch": 187, "lr": 6.919165820975011e-05} +{"train_loss": 0.014890547841787338, "grad_norm": 0.13630971312522888, "global_step": 19854, "epoch": 187, "lr": 6.918892142661837e-05} +{"train_loss": 0.01777629926800728, "grad_norm": 0.06323091685771942, "global_step": 19855, "epoch": 187, "lr": 6.918618457606514e-05} +{"train_loss": 0.014322848990559578, "grad_norm": 0.11709334701299667, "global_step": 19856, "epoch": 187, "lr": 6.918344765809997e-05} +{"train_loss": 0.01932356506586075, "grad_norm": 0.13875222206115723, "global_step": 19857, "epoch": 187, "lr": 6.918071067273252e-05} +{"train_loss": 0.016673266887664795, "grad_norm": 0.056489136070013046, "global_step": 19858, "epoch": 187, "lr": 6.91779736199724e-05} +{"train_loss": 0.015711765736341476, "grad_norm": 0.10333171486854553, "global_step": 19859, "epoch": 187, "lr": 6.91752364998292e-05} +{"train_loss": 0.016631251201033592, "grad_norm": 0.050307054072618484, "global_step": 19860, "epoch": 187, "lr": 6.917249931231257e-05} +{"train_loss": 0.01257244311273098, "grad_norm": 0.09439770132303238, "global_step": 19861, "epoch": 187, "lr": 6.916976205743213e-05} +{"train_loss": 0.015605974942445755, "grad_norm": 0.0660996064543724, "global_step": 19862, "epoch": 187, "lr": 6.916702473519748e-05} +{"train_loss": 0.017780851572752, "grad_norm": 0.09952563047409058, "global_step": 19863, "epoch": 187, "lr": 6.916428734561822e-05} +{"train_loss": 0.014978030696511269, "grad_norm": 0.06362047046422958, "global_step": 19864, "epoch": 187, "lr": 6.9161549888704e-05} +{"train_loss": 0.01847459375858307, "grad_norm": 0.053860753774642944, "global_step": 19865, "epoch": 187, "lr": 6.915881236446443e-05} +{"train_loss": 0.017557945102453232, "grad_norm": 0.09813797473907471, "global_step": 19866, "epoch": 187, "lr": 6.915607477290911e-05} +{"train_loss": 0.020010145381093025, "grad_norm": 0.09916773438453674, "global_step": 19867, "epoch": 187, "lr": 6.91533371140477e-05} +{"train_loss": 0.012886236421763897, "grad_norm": 0.12083505839109421, "global_step": 19868, "epoch": 187, "lr": 6.915059938788977e-05} +{"train_loss": 0.01412961445748806, "grad_norm": 0.052637726068496704, "global_step": 19869, "epoch": 187, "lr": 6.914786159444497e-05} +{"train_loss": 0.01635228469967842, "grad_norm": 0.05972392484545708, "global_step": 19870, "epoch": 187, "lr": 6.914512373372291e-05} +{"train_loss": 0.016914930194616318, "grad_norm": 0.07998035103082657, "global_step": 19871, "epoch": 187, "lr": 6.914238580573321e-05} +{"train_loss": 0.014155061915516853, "grad_norm": 0.059503696858882904, "global_step": 19872, "epoch": 187, "lr": 6.91396478104855e-05} +{"train_loss": 0.015623983927071095, "grad_norm": 0.062113337218761444, "global_step": 19873, "epoch": 187, "lr": 6.913690974798938e-05} +{"train_loss": 0.01676490530371666, "grad_norm": 0.0871114507317543, "global_step": 19874, "epoch": 187, "lr": 6.91341716182545e-05} +{"train_loss": 0.016800962388515472, "grad_norm": 0.0789390280842781, "global_step": 19875, "epoch": 187, "lr": 6.913143342129045e-05} +{"train_loss": 0.014941050671041012, "grad_norm": 0.07397371530532837, "global_step": 19876, "epoch": 187, "lr": 6.912869515710687e-05} +{"train_loss": 0.015870943665504456, "grad_norm": 0.08891972154378891, "global_step": 19877, "epoch": 187, "lr": 6.912595682571336e-05} +{"train_loss": 0.01857640966773033, "grad_norm": 0.07208497822284698, "global_step": 19878, "epoch": 187, "lr": 6.912321842711955e-05} +{"train_loss": 0.015462086535990238, "grad_norm": 0.05082429200410843, "global_step": 19879, "epoch": 187, "lr": 6.912047996133508e-05} +{"train_loss": 0.017874330282211304, "grad_norm": 0.10009352117776871, "global_step": 19880, "epoch": 187, "lr": 6.911774142836957e-05} +{"train_loss": 0.014125705696642399, "grad_norm": 0.07219149172306061, "global_step": 19881, "epoch": 187, "lr": 6.91150028282326e-05} +{"train_loss": 0.019618945196270943, "grad_norm": 0.14869779348373413, "global_step": 19882, "epoch": 187, "lr": 6.911226416093386e-05} +{"train_loss": 0.017419148236513138, "grad_norm": 0.11682756245136261, "global_step": 19883, "epoch": 187, "lr": 6.91095254264829e-05} +{"train_loss": 0.013868238776922226, "grad_norm": 0.10191008448600769, "global_step": 19884, "epoch": 187, "lr": 6.91067866248894e-05} +{"train_loss": 0.01760144904255867, "grad_norm": 0.10264864563941956, "global_step": 19885, "epoch": 187, "lr": 6.910404775616296e-05} +{"train_loss": 0.016986427828669548, "grad_norm": 0.050593484193086624, "global_step": 19886, "epoch": 187, "lr": 6.910130882031321e-05} +{"train_loss": 0.015492857433855534, "grad_norm": 0.05774431303143501, "global_step": 19887, "epoch": 187, "lr": 6.909856981734974e-05} +{"train_loss": 0.016329392790794373, "grad_norm": 0.06519701331853867, "global_step": 19888, "epoch": 187, "lr": 6.909583074728223e-05} +{"train_loss": 0.014480292797088623, "grad_norm": 0.05064034461975098, "global_step": 19889, "epoch": 187, "lr": 6.909309161012026e-05} +{"train_loss": 0.0166312325745821, "grad_norm": 0.06136681139469147, "global_step": 19890, "epoch": 187, "lr": 6.909035240587348e-05} +{"train_loss": 0.017844749614596367, "grad_norm": 0.08043280243873596, "global_step": 19891, "epoch": 187, "lr": 6.90876131345515e-05} +{"train_loss": 0.014736760407686234, "grad_norm": 0.08966077864170074, "global_step": 19892, "epoch": 187, "lr": 6.908487379616394e-05} +{"train_loss": 0.018298331648111343, "grad_norm": 0.0696457028388977, "global_step": 19893, "epoch": 187, "lr": 6.908213439072044e-05} +{"train_loss": 0.014523020945489407, "grad_norm": 0.092725969851017, "global_step": 19894, "epoch": 187, "lr": 6.907939491823063e-05} +{"train_loss": 0.017490467056632042, "grad_norm": 0.07506576180458069, "global_step": 19895, "epoch": 187, "lr": 6.907665537870409e-05} +{"train_loss": 0.01607608050107956, "grad_norm": 0.09203118085861206, "global_step": 19896, "epoch": 187, "lr": 6.907391577215052e-05} +{"train_loss": 0.0160691998898983, "grad_norm": 0.08379799872636795, "global_step": 19897, "epoch": 187, "lr": 6.907117609857947e-05} +{"train_loss": 0.014301490038633347, "grad_norm": 0.10693300515413284, "global_step": 19898, "epoch": 187, "lr": 6.906843635800062e-05} +{"train_loss": 0.0182783380150795, "grad_norm": 0.07326987385749817, "global_step": 19899, "epoch": 187, "lr": 6.906569655042357e-05} +{"train_loss": 0.01706741936504841, "grad_norm": 0.09044539928436279, "global_step": 19900, "epoch": 187, "lr": 6.906295667585796e-05} +{"train_loss": 0.016620658338069916, "grad_norm": 0.06421218067407608, "global_step": 19901, "epoch": 187, "lr": 6.906021673431342e-05} +{"train_loss": 0.016368377953767776, "grad_norm": 0.08801553398370743, "global_step": 19902, "epoch": 187, "lr": 6.905747672579956e-05} +{"train_loss": 0.01604606769979, "grad_norm": 0.11386194825172424, "global_step": 19903, "epoch": 187, "lr": 6.9054736650326e-05} +{"train_loss": 0.01564284786581993, "grad_norm": 0.14581285417079926, "global_step": 19904, "epoch": 187, "lr": 6.90519965079024e-05} +{"train_loss": 0.018462825566530228, "grad_norm": 0.07960328459739685, "global_step": 19905, "epoch": 187, "lr": 6.904925629853835e-05} +{"train_loss": 0.018332798033952713, "grad_norm": 0.08859498798847198, "global_step": 19906, "epoch": 187, "lr": 6.904651602224352e-05} +{"train_loss": 0.01672041043639183, "grad_norm": 0.08622203767299652, "global_step": 19907, "epoch": 187, "lr": 6.90437756790275e-05} +{"train_loss": 0.015831759199500084, "grad_norm": 0.12366341799497604, "global_step": 19908, "epoch": 187, "lr": 6.904103526889996e-05} +{"train_loss": 0.01510033942759037, "grad_norm": 0.04757779836654663, "global_step": 19909, "epoch": 187, "lr": 6.903829479187048e-05} +{"train_loss": 0.013898796401917934, "grad_norm": 0.12384895980358124, "global_step": 19910, "epoch": 187, "lr": 6.90355542479487e-05} +{"train_loss": 0.015189534984529018, "grad_norm": 0.06926117837429047, "global_step": 19911, "epoch": 187, "lr": 6.903281363714428e-05} +{"train_loss": 0.017222613096237183, "grad_norm": 0.08215341717004776, "global_step": 19912, "epoch": 187, "lr": 6.903007295946682e-05} +{"train_loss": 0.01753653958439827, "grad_norm": 0.07466432452201843, "global_step": 19913, "epoch": 187, "lr": 6.902733221492597e-05} +{"train_loss": 0.015888068825006485, "grad_norm": 0.08915763348340988, "global_step": 19914, "epoch": 187, "lr": 6.902459140353135e-05} +{"train_loss": 0.01521589420735836, "grad_norm": 0.05039811134338379, "global_step": 19915, "epoch": 187, "lr": 6.902185052529257e-05} +{"train_loss": 0.016029495745897293, "grad_norm": 0.058287717401981354, "global_step": 19916, "epoch": 187, "lr": 6.90191095802193e-05} +{"train_loss": 0.016304224729537964, "grad_norm": 0.11136355251073837, "global_step": 19917, "epoch": 187, "lr": 6.901636856832114e-05} +{"train_loss": 0.0181307103484869, "grad_norm": 0.06757111102342606, "global_step": 19918, "epoch": 187, "lr": 6.901362748960773e-05} +{"train_loss": 0.01510570291429758, "grad_norm": 0.09624919295310974, "global_step": 19919, "epoch": 187, "lr": 6.90108863440887e-05} +{"train_loss": 0.015527419745922089, "grad_norm": 0.07393909245729446, "global_step": 19920, "epoch": 187, "lr": 6.90081451317737e-05} +{"train_loss": 0.01566949672996998, "grad_norm": 0.06511938571929932, "global_step": 19921, "epoch": 187, "lr": 6.900540385267234e-05} +{"train_loss": 0.014907288365066051, "grad_norm": 0.1645347625017166, "global_step": 19922, "epoch": 187, "lr": 6.900266250679423e-05} +{"train_loss": 0.015855882316827774, "grad_norm": 0.10312465578317642, "global_step": 19923, "epoch": 187, "lr": 6.899992109414907e-05} +{"train_loss": 0.014897515065968037, "grad_norm": 0.04821019992232323, "global_step": 19924, "epoch": 187, "lr": 6.899717961474642e-05} +{"train_loss": 0.016283437609672546, "grad_norm": 0.09581828117370605, "global_step": 19925, "epoch": 187, "lr": 6.899443806859595e-05} +{"train_loss": 0.01644754782319069, "grad_norm": 0.07451106607913971, "global_step": 19926, "epoch": 187, "lr": 6.899169645570729e-05} +{"train_loss": 0.016113108472090285, "grad_norm": 0.10547742247581482, "global_step": 19927, "epoch": 187, "lr": 6.898895477609007e-05, "val_loss": 0.019333533942699432} +{"train_loss": 0.014723850414156914, "grad_norm": 0.07870699465274811, "global_step": 19928, "epoch": 188, "lr": 6.89862130297539e-05} +{"train_loss": 0.016024213284254074, "grad_norm": 0.0856744647026062, "global_step": 19929, "epoch": 188, "lr": 6.898347121670845e-05} +{"train_loss": 0.013875528238713741, "grad_norm": 0.12031708657741547, "global_step": 19930, "epoch": 188, "lr": 6.898072933696333e-05} +{"train_loss": 0.01569204218685627, "grad_norm": 0.09976896643638611, "global_step": 19931, "epoch": 188, "lr": 6.89779873905282e-05} +{"train_loss": 0.01787177100777626, "grad_norm": 0.10996723920106888, "global_step": 19932, "epoch": 188, "lr": 6.897524537741267e-05} +{"train_loss": 0.015813862904906273, "grad_norm": 0.13738960027694702, "global_step": 19933, "epoch": 188, "lr": 6.897250329762637e-05} +{"train_loss": 0.014936714433133602, "grad_norm": 0.08435855805873871, "global_step": 19934, "epoch": 188, "lr": 6.896976115117893e-05} +{"train_loss": 0.01719577983021736, "grad_norm": 0.07454264163970947, "global_step": 19935, "epoch": 188, "lr": 6.896701893808005e-05} +{"train_loss": 0.019360806792974472, "grad_norm": 0.07774721086025238, "global_step": 19936, "epoch": 188, "lr": 6.896427665833925e-05} +{"train_loss": 0.014387386851012707, "grad_norm": 0.08325106650590897, "global_step": 19937, "epoch": 188, "lr": 6.896153431196627e-05} +{"train_loss": 0.015288390219211578, "grad_norm": 0.10561353713274002, "global_step": 19938, "epoch": 188, "lr": 6.89587918989707e-05} +{"train_loss": 0.02035200223326683, "grad_norm": 0.10778570175170898, "global_step": 19939, "epoch": 188, "lr": 6.895604941936217e-05} +{"train_loss": 0.018412698060274124, "grad_norm": 0.06053995341062546, "global_step": 19940, "epoch": 188, "lr": 6.895330687315033e-05} +{"train_loss": 0.01505054347217083, "grad_norm": 0.07977978140115738, "global_step": 19941, "epoch": 188, "lr": 6.895056426034481e-05} +{"train_loss": 0.017362693324685097, "grad_norm": 0.0803900733590126, "global_step": 19942, "epoch": 188, "lr": 6.894782158095525e-05} +{"train_loss": 0.017632845789194107, "grad_norm": 0.1571597456932068, "global_step": 19943, "epoch": 188, "lr": 6.89450788349913e-05} +{"train_loss": 0.01330984104424715, "grad_norm": 0.0716911256313324, "global_step": 19944, "epoch": 188, "lr": 6.894233602246256e-05} +{"train_loss": 0.01555292122066021, "grad_norm": 0.061700571328401566, "global_step": 19945, "epoch": 188, "lr": 6.89395931433787e-05} +{"train_loss": 0.014567328616976738, "grad_norm": 0.055573225021362305, "global_step": 19946, "epoch": 188, "lr": 6.893685019774934e-05} +{"train_loss": 0.014979508705437183, "grad_norm": 0.05092116445302963, "global_step": 19947, "epoch": 188, "lr": 6.893410718558414e-05} +{"train_loss": 0.015257459133863449, "grad_norm": 0.060586243867874146, "global_step": 19948, "epoch": 188, "lr": 6.893136410689269e-05} +{"train_loss": 0.01743549294769764, "grad_norm": 0.06645742803812027, "global_step": 19949, "epoch": 188, "lr": 6.892862096168469e-05} +{"train_loss": 0.01635443978011608, "grad_norm": 0.06863156706094742, "global_step": 19950, "epoch": 188, "lr": 6.892587774996975e-05} +{"train_loss": 0.016773372888565063, "grad_norm": 0.06323860585689545, "global_step": 19951, "epoch": 188, "lr": 6.892313447175748e-05} +{"train_loss": 0.01950565166771412, "grad_norm": 0.08510441333055496, "global_step": 19952, "epoch": 188, "lr": 6.892039112705756e-05} +{"train_loss": 0.016665423288941383, "grad_norm": 0.06317096203565598, "global_step": 19953, "epoch": 188, "lr": 6.89176477158796e-05} +{"train_loss": 0.01579020544886589, "grad_norm": 0.07027421146631241, "global_step": 19954, "epoch": 188, "lr": 6.891490423823328e-05} +{"train_loss": 0.017347779124975204, "grad_norm": 0.07560313493013382, "global_step": 19955, "epoch": 188, "lr": 6.891216069412819e-05} +{"train_loss": 0.016882292926311493, "grad_norm": 0.10150968283414841, "global_step": 19956, "epoch": 188, "lr": 6.8909417083574e-05} +{"train_loss": 0.01221832912415266, "grad_norm": 0.04944906011223793, "global_step": 19957, "epoch": 188, "lr": 6.890667340658034e-05} +{"train_loss": 0.016847355291247368, "grad_norm": 0.09310486167669296, "global_step": 19958, "epoch": 188, "lr": 6.890392966315687e-05} +{"train_loss": 0.017773831263184547, "grad_norm": 0.09403522312641144, "global_step": 19959, "epoch": 188, "lr": 6.890118585331319e-05} +{"train_loss": 0.018678372725844383, "grad_norm": 0.15987348556518555, "global_step": 19960, "epoch": 188, "lr": 6.889844197705897e-05} +{"train_loss": 0.013125553727149963, "grad_norm": 0.05988628417253494, "global_step": 19961, "epoch": 188, "lr": 6.889569803440384e-05} +{"train_loss": 0.018746189773082733, "grad_norm": 0.07944024354219437, "global_step": 19962, "epoch": 188, "lr": 6.889295402535744e-05} +{"train_loss": 0.01502478588372469, "grad_norm": 0.1304490864276886, "global_step": 19963, "epoch": 188, "lr": 6.889020994992942e-05} +{"train_loss": 0.01538750808686018, "grad_norm": 0.12724782526493073, "global_step": 19964, "epoch": 188, "lr": 6.888746580812942e-05} +{"train_loss": 0.01669294387102127, "grad_norm": 0.0943051353096962, "global_step": 19965, "epoch": 188, "lr": 6.888472159996709e-05} +{"train_loss": 0.01819433644413948, "grad_norm": 0.09291250258684158, "global_step": 19966, "epoch": 188, "lr": 6.888197732545205e-05} +{"train_loss": 0.01496371254324913, "grad_norm": 0.05762246251106262, "global_step": 19967, "epoch": 188, "lr": 6.887923298459394e-05} +{"train_loss": 0.018671074882149696, "grad_norm": 0.07716873288154602, "global_step": 19968, "epoch": 188, "lr": 6.887648857740244e-05} +{"train_loss": 0.02046911232173443, "grad_norm": 0.09413058310747147, "global_step": 19969, "epoch": 188, "lr": 6.887374410388715e-05} +{"train_loss": 0.01684289053082466, "grad_norm": 0.08501994609832764, "global_step": 19970, "epoch": 188, "lr": 6.887099956405774e-05} +{"train_loss": 0.01732061244547367, "grad_norm": 0.07388114184141159, "global_step": 19971, "epoch": 188, "lr": 6.886825495792384e-05} +{"train_loss": 0.014035562053322792, "grad_norm": 0.10339400172233582, "global_step": 19972, "epoch": 188, "lr": 6.88655102854951e-05} +{"train_loss": 0.014635867439210415, "grad_norm": 0.08540989458560944, "global_step": 19973, "epoch": 188, "lr": 6.886276554678117e-05} +{"train_loss": 0.015786919742822647, "grad_norm": 0.07653272897005081, "global_step": 19974, "epoch": 188, "lr": 6.886002074179168e-05} +{"train_loss": 0.015188752673566341, "grad_norm": 0.04452795907855034, "global_step": 19975, "epoch": 188, "lr": 6.885727587053625e-05} +{"train_loss": 0.016946978867053986, "grad_norm": 0.05544786900281906, "global_step": 19976, "epoch": 188, "lr": 6.88545309330246e-05} +{"train_loss": 0.01722714491188526, "grad_norm": 0.07947910577058792, "global_step": 19977, "epoch": 188, "lr": 6.88517859292663e-05} +{"train_loss": 0.013815822079777718, "grad_norm": 0.11003442853689194, "global_step": 19978, "epoch": 188, "lr": 6.884904085927104e-05} +{"train_loss": 0.016907494515180588, "grad_norm": 0.060223229229450226, "global_step": 19979, "epoch": 188, "lr": 6.884629572304844e-05} +{"train_loss": 0.01766910031437874, "grad_norm": 0.06853672862052917, "global_step": 19980, "epoch": 188, "lr": 6.884355052060814e-05} +{"train_loss": 0.015569977462291718, "grad_norm": 0.11780548095703125, "global_step": 19981, "epoch": 188, "lr": 6.884080525195981e-05} +{"train_loss": 0.015837213024497032, "grad_norm": 0.09093959629535675, "global_step": 19982, "epoch": 188, "lr": 6.883805991711307e-05} +{"train_loss": 0.01439947821199894, "grad_norm": 0.08337358385324478, "global_step": 19983, "epoch": 188, "lr": 6.88353145160776e-05} +{"train_loss": 0.015706457197666168, "grad_norm": 0.07579965889453888, "global_step": 19984, "epoch": 188, "lr": 6.883256904886301e-05} +{"train_loss": 0.013870642520487309, "grad_norm": 0.08006131649017334, "global_step": 19985, "epoch": 188, "lr": 6.882982351547898e-05} +{"train_loss": 0.014689583331346512, "grad_norm": 0.09049750864505768, "global_step": 19986, "epoch": 188, "lr": 6.882707791593513e-05} +{"train_loss": 0.017426924780011177, "grad_norm": 0.10491824150085449, "global_step": 19987, "epoch": 188, "lr": 6.88243322502411e-05} +{"train_loss": 0.015438465401530266, "grad_norm": 0.07935132831335068, "global_step": 19988, "epoch": 188, "lr": 6.88215865184066e-05} +{"train_loss": 0.015301911160349846, "grad_norm": 0.07769320160150528, "global_step": 19989, "epoch": 188, "lr": 6.881884072044119e-05} +{"train_loss": 0.016158469021320343, "grad_norm": 0.06998312473297119, "global_step": 19990, "epoch": 188, "lr": 6.881609485635457e-05} +{"train_loss": 0.01339237205684185, "grad_norm": 0.07003676891326904, "global_step": 19991, "epoch": 188, "lr": 6.881334892615637e-05} +{"train_loss": 0.014154745265841484, "grad_norm": 0.1288425326347351, "global_step": 19992, "epoch": 188, "lr": 6.881060292985624e-05} +{"train_loss": 0.017985673621296883, "grad_norm": 0.06237667053937912, "global_step": 19993, "epoch": 188, "lr": 6.880785686746384e-05} +{"train_loss": 0.017187099903821945, "grad_norm": 0.11633976548910141, "global_step": 19994, "epoch": 188, "lr": 6.880511073898881e-05} +{"train_loss": 0.013858327642083168, "grad_norm": 0.05166416987776756, "global_step": 19995, "epoch": 188, "lr": 6.880236454444079e-05} +{"train_loss": 0.015430372208356857, "grad_norm": 0.07853954285383224, "global_step": 19996, "epoch": 188, "lr": 6.879961828382945e-05} +{"train_loss": 0.016078496351838112, "grad_norm": 0.07734929770231247, "global_step": 19997, "epoch": 188, "lr": 6.879687195716443e-05} +{"train_loss": 0.017962953075766563, "grad_norm": 0.10224490612745285, "global_step": 19998, "epoch": 188, "lr": 6.879412556445536e-05} +{"train_loss": 0.012954000383615494, "grad_norm": 0.06028492748737335, "global_step": 19999, "epoch": 188, "lr": 6.879137910571191e-05} +{"train_loss": 0.016431866213679314, "grad_norm": 0.07835497707128525, "global_step": 20000, "epoch": 188, "lr": 6.878863258094373e-05} +{"train_loss": 0.013328703120350838, "grad_norm": 0.06455574929714203, "global_step": 20001, "epoch": 188, "lr": 6.878588599016047e-05} +{"train_loss": 0.016476429998874664, "grad_norm": 0.11180000752210617, "global_step": 20002, "epoch": 188, "lr": 6.878313933337178e-05} +{"train_loss": 0.014838186092674732, "grad_norm": 0.08723296970129013, "global_step": 20003, "epoch": 188, "lr": 6.87803926105873e-05} +{"train_loss": 0.01580282673239708, "grad_norm": 0.17964571714401245, "global_step": 20004, "epoch": 188, "lr": 6.877764582181667e-05} +{"train_loss": 0.014339808374643326, "grad_norm": 0.07666847854852676, "global_step": 20005, "epoch": 188, "lr": 6.877489896706958e-05} +{"train_loss": 0.019069543108344078, "grad_norm": 0.07477782666683197, "global_step": 20006, "epoch": 188, "lr": 6.877215204635566e-05} +{"train_loss": 0.012558710761368275, "grad_norm": 0.06248506158590317, "global_step": 20007, "epoch": 188, "lr": 6.876940505968457e-05} +{"train_loss": 0.015532176941633224, "grad_norm": 0.058216918259859085, "global_step": 20008, "epoch": 188, "lr": 6.876665800706595e-05} +{"train_loss": 0.017193980515003204, "grad_norm": 0.12732738256454468, "global_step": 20009, "epoch": 188, "lr": 6.876391088850943e-05} +{"train_loss": 0.015016591176390648, "grad_norm": 0.13023675978183746, "global_step": 20010, "epoch": 188, "lr": 6.87611637040247e-05} +{"train_loss": 0.013363957405090332, "grad_norm": 0.07570929080247879, "global_step": 20011, "epoch": 188, "lr": 6.87584164536214e-05} +{"train_loss": 0.01594390720129013, "grad_norm": 0.05701328068971634, "global_step": 20012, "epoch": 188, "lr": 6.87556691373092e-05} +{"train_loss": 0.0169178806245327, "grad_norm": 0.06367722898721695, "global_step": 20013, "epoch": 188, "lr": 6.875292175509773e-05} +{"train_loss": 0.01613708585500717, "grad_norm": 0.05515573173761368, "global_step": 20014, "epoch": 188, "lr": 6.875017430699663e-05} +{"train_loss": 0.01643703132867813, "grad_norm": 0.07282611727714539, "global_step": 20015, "epoch": 188, "lr": 6.874742679301559e-05} +{"train_loss": 0.019068732857704163, "grad_norm": 0.06128434091806412, "global_step": 20016, "epoch": 188, "lr": 6.874467921316422e-05} +{"train_loss": 0.019070260226726532, "grad_norm": 0.09949852526187897, "global_step": 20017, "epoch": 188, "lr": 6.874193156745222e-05} +{"train_loss": 0.015663128346204758, "grad_norm": 0.07293184846639633, "global_step": 20018, "epoch": 188, "lr": 6.87391838558892e-05} +{"train_loss": 0.015195332467556, "grad_norm": 0.08181995153427124, "global_step": 20019, "epoch": 188, "lr": 6.873643607848487e-05} +{"train_loss": 0.014867832884192467, "grad_norm": 0.08266498893499374, "global_step": 20020, "epoch": 188, "lr": 6.873368823524886e-05} +{"train_loss": 0.01519835740327835, "grad_norm": 0.05503992363810539, "global_step": 20021, "epoch": 188, "lr": 6.873094032619077e-05} +{"train_loss": 0.016303475946187973, "grad_norm": 0.08951500058174133, "global_step": 20022, "epoch": 188, "lr": 6.872819235132033e-05} +{"train_loss": 0.018351253122091293, "grad_norm": 0.123841792345047, "global_step": 20023, "epoch": 188, "lr": 6.872544431064717e-05} +{"train_loss": 0.014929323457181454, "grad_norm": 0.06909878551959991, "global_step": 20024, "epoch": 188, "lr": 6.872269620418093e-05} +{"train_loss": 0.012366442009806633, "grad_norm": 0.10068988054990768, "global_step": 20025, "epoch": 188, "lr": 6.87199480319313e-05} +{"train_loss": 0.01991310901939869, "grad_norm": 0.08188874274492264, "global_step": 20026, "epoch": 188, "lr": 6.871719979390789e-05} +{"train_loss": 0.01765703223645687, "grad_norm": 0.07023797929286957, "global_step": 20027, "epoch": 188, "lr": 6.871445149012038e-05} +{"train_loss": 0.01700541190803051, "grad_norm": 0.09106695652008057, "global_step": 20028, "epoch": 188, "lr": 6.871170312057842e-05} +{"train_loss": 0.016320396214723587, "grad_norm": 0.06115366145968437, "global_step": 20029, "epoch": 188, "lr": 6.870895468529171e-05} +{"train_loss": 0.017251988872885704, "grad_norm": 0.05675897002220154, "global_step": 20030, "epoch": 188, "lr": 6.870620618426984e-05} +{"train_loss": 0.017888491973280907, "grad_norm": 0.05859466269612312, "global_step": 20031, "epoch": 188, "lr": 6.87034576175225e-05} +{"train_loss": 0.013819766230881214, "grad_norm": 0.09732875972986221, "global_step": 20032, "epoch": 188, "lr": 6.870070898505934e-05} +{"train_loss": 0.01609119891522909, "grad_norm": 0.10319038480520248, "global_step": 20033, "epoch": 188, "lr": 6.869796028689001e-05, "val_loss": 0.01910584419965744} +{"train_loss": 0.016906049102544785, "grad_norm": 0.0783650130033493, "global_step": 20034, "epoch": 189, "lr": 6.86952115230242e-05} +{"train_loss": 0.015137308277189732, "grad_norm": 0.07704360038042068, "global_step": 20035, "epoch": 189, "lr": 6.869246269347156e-05} +{"train_loss": 0.01673140935599804, "grad_norm": 0.14936821162700653, "global_step": 20036, "epoch": 189, "lr": 6.868971379824171e-05} +{"train_loss": 0.017613230273127556, "grad_norm": 0.07803260535001755, "global_step": 20037, "epoch": 189, "lr": 6.868696483734435e-05} +{"train_loss": 0.01494770497083664, "grad_norm": 0.049356624484062195, "global_step": 20038, "epoch": 189, "lr": 6.868421581078912e-05} +{"train_loss": 0.016508135944604874, "grad_norm": 0.08237121254205704, "global_step": 20039, "epoch": 189, "lr": 6.868146671858567e-05} +{"train_loss": 0.018127718940377235, "grad_norm": 0.08557843416929245, "global_step": 20040, "epoch": 189, "lr": 6.867871756074366e-05} +{"train_loss": 0.015388182364404202, "grad_norm": 0.08312716335058212, "global_step": 20041, "epoch": 189, "lr": 6.86759683372728e-05} +{"train_loss": 0.013731062412261963, "grad_norm": 0.12496816366910934, "global_step": 20042, "epoch": 189, "lr": 6.867321904818267e-05} +{"train_loss": 0.01734497770667076, "grad_norm": 0.12600523233413696, "global_step": 20043, "epoch": 189, "lr": 6.867046969348298e-05} +{"train_loss": 0.01805056445300579, "grad_norm": 0.0653727650642395, "global_step": 20044, "epoch": 189, "lr": 6.86677202731834e-05} +{"train_loss": 0.018223075196146965, "grad_norm": 0.057716455310583115, "global_step": 20045, "epoch": 189, "lr": 6.866497078729353e-05} +{"train_loss": 0.014384515583515167, "grad_norm": 0.12000668793916702, "global_step": 20046, "epoch": 189, "lr": 6.866222123582309e-05} +{"train_loss": 0.015520562417805195, "grad_norm": 0.10473693162202835, "global_step": 20047, "epoch": 189, "lr": 6.865947161878173e-05} +{"train_loss": 0.017473315820097923, "grad_norm": 0.05459914728999138, "global_step": 20048, "epoch": 189, "lr": 6.865672193617909e-05} +{"train_loss": 0.014812297187745571, "grad_norm": 0.07349762320518494, "global_step": 20049, "epoch": 189, "lr": 6.865397218802485e-05} +{"train_loss": 0.019573867321014404, "grad_norm": 0.07500287890434265, "global_step": 20050, "epoch": 189, "lr": 6.865122237432866e-05} +{"train_loss": 0.0150513406842947, "grad_norm": 0.048282939940690994, "global_step": 20051, "epoch": 189, "lr": 6.864847249510019e-05} +{"train_loss": 0.016928602010011673, "grad_norm": 0.062122493982315063, "global_step": 20052, "epoch": 189, "lr": 6.864572255034909e-05} +{"train_loss": 0.014751658774912357, "grad_norm": 0.14200958609580994, "global_step": 20053, "epoch": 189, "lr": 6.864297254008504e-05} +{"train_loss": 0.017384907230734825, "grad_norm": 0.13239362835884094, "global_step": 20054, "epoch": 189, "lr": 6.86402224643177e-05} +{"train_loss": 0.014916369691491127, "grad_norm": 0.0897749587893486, "global_step": 20055, "epoch": 189, "lr": 6.86374723230567e-05} +{"train_loss": 0.018702253699302673, "grad_norm": 0.12000861763954163, "global_step": 20056, "epoch": 189, "lr": 6.863472211631174e-05} +{"train_loss": 0.016396425664424896, "grad_norm": 0.0819464772939682, "global_step": 20057, "epoch": 189, "lr": 6.863197184409245e-05} +{"train_loss": 0.014988197945058346, "grad_norm": 0.07131479680538177, "global_step": 20058, "epoch": 189, "lr": 6.862922150640855e-05} +{"train_loss": 0.016035882756114006, "grad_norm": 0.18298888206481934, "global_step": 20059, "epoch": 189, "lr": 6.862647110326965e-05} +{"train_loss": 0.01615728624165058, "grad_norm": 0.05955827608704567, "global_step": 20060, "epoch": 189, "lr": 6.862372063468543e-05} +{"train_loss": 0.014905795454978943, "grad_norm": 0.05780823528766632, "global_step": 20061, "epoch": 189, "lr": 6.862097010066556e-05} +{"train_loss": 0.016258377581834793, "grad_norm": 0.11856671422719955, "global_step": 20062, "epoch": 189, "lr": 6.861821950121968e-05} +{"train_loss": 0.01684756949543953, "grad_norm": 0.09834399819374084, "global_step": 20063, "epoch": 189, "lr": 6.861546883635749e-05} +{"train_loss": 0.012930233031511307, "grad_norm": 0.059476498514413834, "global_step": 20064, "epoch": 189, "lr": 6.861271810608864e-05} +{"train_loss": 0.015429090708494186, "grad_norm": 0.0911347046494484, "global_step": 20065, "epoch": 189, "lr": 6.86099673104228e-05} +{"train_loss": 0.014756123535335064, "grad_norm": 0.10565033555030823, "global_step": 20066, "epoch": 189, "lr": 6.86072164493696e-05} +{"train_loss": 0.015186992473900318, "grad_norm": 0.09622563421726227, "global_step": 20067, "epoch": 189, "lr": 6.860446552293876e-05} +{"train_loss": 0.017043791711330414, "grad_norm": 0.0756533071398735, "global_step": 20068, "epoch": 189, "lr": 6.860171453113992e-05} +{"train_loss": 0.017942562699317932, "grad_norm": 0.07081794738769531, "global_step": 20069, "epoch": 189, "lr": 6.859896347398272e-05} +{"train_loss": 0.01678476668894291, "grad_norm": 0.08085042983293533, "global_step": 20070, "epoch": 189, "lr": 6.859621235147688e-05} +{"train_loss": 0.016536150127649307, "grad_norm": 0.12566493451595306, "global_step": 20071, "epoch": 189, "lr": 6.8593461163632e-05} +{"train_loss": 0.01778058148920536, "grad_norm": 0.07694488018751144, "global_step": 20072, "epoch": 189, "lr": 6.859070991045782e-05} +{"train_loss": 0.016171125695109367, "grad_norm": 0.09592511504888535, "global_step": 20073, "epoch": 189, "lr": 6.858795859196396e-05} +{"train_loss": 0.016271663829684258, "grad_norm": 0.13934440910816193, "global_step": 20074, "epoch": 189, "lr": 6.858520720816009e-05} +{"train_loss": 0.01246949564665556, "grad_norm": 0.09504937380552292, "global_step": 20075, "epoch": 189, "lr": 6.858245575905589e-05} +{"train_loss": 0.01514216884970665, "grad_norm": 0.1188352033495903, "global_step": 20076, "epoch": 189, "lr": 6.857970424466101e-05} +{"train_loss": 0.014179809018969536, "grad_norm": 0.07350421696901321, "global_step": 20077, "epoch": 189, "lr": 6.857695266498515e-05} +{"train_loss": 0.016029585152864456, "grad_norm": 0.0916164293885231, "global_step": 20078, "epoch": 189, "lr": 6.857420102003795e-05} +{"train_loss": 0.014985046349465847, "grad_norm": 0.06302803754806519, "global_step": 20079, "epoch": 189, "lr": 6.857144930982908e-05} +{"train_loss": 0.01633761078119278, "grad_norm": 0.053842850029468536, "global_step": 20080, "epoch": 189, "lr": 6.856869753436821e-05} +{"train_loss": 0.018707897514104843, "grad_norm": 0.06491503864526749, "global_step": 20081, "epoch": 189, "lr": 6.856594569366502e-05} +{"train_loss": 0.015920056030154228, "grad_norm": 0.08598655462265015, "global_step": 20082, "epoch": 189, "lr": 6.856319378772918e-05} +{"train_loss": 0.013650203123688698, "grad_norm": 0.06369054317474365, "global_step": 20083, "epoch": 189, "lr": 6.856044181657033e-05} +{"train_loss": 0.014639750123023987, "grad_norm": 0.0852077379822731, "global_step": 20084, "epoch": 189, "lr": 6.855768978019817e-05} +{"train_loss": 0.014993995428085327, "grad_norm": 0.06690812855958939, "global_step": 20085, "epoch": 189, "lr": 6.855493767862235e-05} +{"train_loss": 0.016116246581077576, "grad_norm": 0.09747444093227386, "global_step": 20086, "epoch": 189, "lr": 6.855218551185255e-05} +{"train_loss": 0.016935430467128754, "grad_norm": 0.11282942444086075, "global_step": 20087, "epoch": 189, "lr": 6.854943327989845e-05} +{"train_loss": 0.017697470262646675, "grad_norm": 0.055887795984745026, "global_step": 20088, "epoch": 189, "lr": 6.85466809827697e-05} +{"train_loss": 0.014886163175106049, "grad_norm": 0.08476646989583969, "global_step": 20089, "epoch": 189, "lr": 6.854392862047598e-05} +{"train_loss": 0.018519587814807892, "grad_norm": 0.07524921000003815, "global_step": 20090, "epoch": 189, "lr": 6.854117619302696e-05} +{"train_loss": 0.01751583069562912, "grad_norm": 0.05723065882921219, "global_step": 20091, "epoch": 189, "lr": 6.853842370043231e-05} +{"train_loss": 0.01697346940636635, "grad_norm": 0.05150337144732475, "global_step": 20092, "epoch": 189, "lr": 6.85356711427017e-05} +{"train_loss": 0.01510000228881836, "grad_norm": 0.06706875562667847, "global_step": 20093, "epoch": 189, "lr": 6.85329185198448e-05} +{"train_loss": 0.014090807177126408, "grad_norm": 0.06894506514072418, "global_step": 20094, "epoch": 189, "lr": 6.85301658318713e-05} +{"train_loss": 0.018256084993481636, "grad_norm": 0.058594249188899994, "global_step": 20095, "epoch": 189, "lr": 6.852741307879083e-05} +{"train_loss": 0.015606649219989777, "grad_norm": 0.0553400032222271, "global_step": 20096, "epoch": 189, "lr": 6.852466026061311e-05} +{"train_loss": 0.015634801238775253, "grad_norm": 0.06430459022521973, "global_step": 20097, "epoch": 189, "lr": 6.852190737734778e-05} +{"train_loss": 0.014770634472370148, "grad_norm": 0.06632621586322784, "global_step": 20098, "epoch": 189, "lr": 6.851915442900452e-05} +{"train_loss": 0.017978528514504433, "grad_norm": 0.0905349925160408, "global_step": 20099, "epoch": 189, "lr": 6.8516401415593e-05} +{"train_loss": 0.016136743128299713, "grad_norm": 0.0762147605419159, "global_step": 20100, "epoch": 189, "lr": 6.851364833712292e-05} +{"train_loss": 0.014620444737374783, "grad_norm": 0.06899893283843994, "global_step": 20101, "epoch": 189, "lr": 6.85108951936039e-05} +{"train_loss": 0.015315205790102482, "grad_norm": 0.06839459389448166, "global_step": 20102, "epoch": 189, "lr": 6.850814198504568e-05} +{"train_loss": 0.013212256133556366, "grad_norm": 0.054321836680173874, "global_step": 20103, "epoch": 189, "lr": 6.850538871145787e-05} +{"train_loss": 0.014871894381940365, "grad_norm": 0.07059754431247711, "global_step": 20104, "epoch": 189, "lr": 6.850263537285019e-05} +{"train_loss": 0.018291786313056946, "grad_norm": 0.1252673864364624, "global_step": 20105, "epoch": 189, "lr": 6.849988196923229e-05} +{"train_loss": 0.015543910674750805, "grad_norm": 0.06707986444234848, "global_step": 20106, "epoch": 189, "lr": 6.849712850061385e-05} +{"train_loss": 0.020054759457707405, "grad_norm": 0.10620646178722382, "global_step": 20107, "epoch": 189, "lr": 6.849437496700453e-05} +{"train_loss": 0.018786881119012833, "grad_norm": 0.1283493936061859, "global_step": 20108, "epoch": 189, "lr": 6.849162136841404e-05} +{"train_loss": 0.014373735524713993, "grad_norm": 0.0695982351899147, "global_step": 20109, "epoch": 189, "lr": 6.848886770485203e-05} +{"train_loss": 0.014833880588412285, "grad_norm": 0.06722210347652435, "global_step": 20110, "epoch": 189, "lr": 6.848611397632818e-05} +{"train_loss": 0.016741380095481873, "grad_norm": 0.06486496329307556, "global_step": 20111, "epoch": 189, "lr": 6.848336018285216e-05} +{"train_loss": 0.014622297137975693, "grad_norm": 0.0683717206120491, "global_step": 20112, "epoch": 189, "lr": 6.848060632443365e-05} +{"train_loss": 0.018298353999853134, "grad_norm": 0.05775986239314079, "global_step": 20113, "epoch": 189, "lr": 6.847785240108232e-05} +{"train_loss": 0.015045976266264915, "grad_norm": 0.06513940542936325, "global_step": 20114, "epoch": 189, "lr": 6.847509841280788e-05} +{"train_loss": 0.014372250996530056, "grad_norm": 0.07882899045944214, "global_step": 20115, "epoch": 189, "lr": 6.847234435961994e-05} +{"train_loss": 0.01563204824924469, "grad_norm": 0.0550897978246212, "global_step": 20116, "epoch": 189, "lr": 6.846959024152823e-05} +{"train_loss": 0.015348276123404503, "grad_norm": 0.07324627786874771, "global_step": 20117, "epoch": 189, "lr": 6.846683605854243e-05} +{"train_loss": 0.016388140618801117, "grad_norm": 0.07864410430192947, "global_step": 20118, "epoch": 189, "lr": 6.846408181067217e-05} +{"train_loss": 0.016834180802106857, "grad_norm": 0.05567213147878647, "global_step": 20119, "epoch": 189, "lr": 6.846132749792718e-05} +{"train_loss": 0.01669219881296158, "grad_norm": 0.11586572229862213, "global_step": 20120, "epoch": 189, "lr": 6.845857312031712e-05} +{"train_loss": 0.012116074562072754, "grad_norm": 0.07675507664680481, "global_step": 20121, "epoch": 189, "lr": 6.845581867785163e-05} +{"train_loss": 0.01709260419011116, "grad_norm": 0.0611749067902565, "global_step": 20122, "epoch": 189, "lr": 6.845306417054045e-05} +{"train_loss": 0.01842491514980793, "grad_norm": 0.08029443025588989, "global_step": 20123, "epoch": 189, "lr": 6.84503095983932e-05} +{"train_loss": 0.018224041908979416, "grad_norm": 0.06431795656681061, "global_step": 20124, "epoch": 189, "lr": 6.844755496141961e-05} +{"train_loss": 0.015696680173277855, "grad_norm": 0.06497388333082199, "global_step": 20125, "epoch": 189, "lr": 6.844480025962931e-05} +{"train_loss": 0.014591092243790627, "grad_norm": 0.07176396250724792, "global_step": 20126, "epoch": 189, "lr": 6.844204549303204e-05} +{"train_loss": 0.016525153070688248, "grad_norm": 0.09388495981693268, "global_step": 20127, "epoch": 189, "lr": 6.84392906616374e-05} +{"train_loss": 0.017307288944721222, "grad_norm": 0.06896986067295074, "global_step": 20128, "epoch": 189, "lr": 6.843653576545513e-05} +{"train_loss": 0.015267833136022091, "grad_norm": 0.0525704063475132, "global_step": 20129, "epoch": 189, "lr": 6.843378080449491e-05} +{"train_loss": 0.016573341563344002, "grad_norm": 0.05288240313529968, "global_step": 20130, "epoch": 189, "lr": 6.843102577876638e-05} +{"train_loss": 0.016251254826784134, "grad_norm": 0.11706726253032684, "global_step": 20131, "epoch": 189, "lr": 6.842827068827924e-05} +{"train_loss": 0.01539682038128376, "grad_norm": 0.08119738101959229, "global_step": 20132, "epoch": 189, "lr": 6.842551553304317e-05} +{"train_loss": 0.017405567690730095, "grad_norm": 0.07791149616241455, "global_step": 20133, "epoch": 189, "lr": 6.842276031306787e-05} +{"train_loss": 0.016328977420926094, "grad_norm": 0.08133210241794586, "global_step": 20134, "epoch": 189, "lr": 6.842000502836299e-05} +{"train_loss": 0.014607222750782967, "grad_norm": 0.09738238900899887, "global_step": 20135, "epoch": 189, "lr": 6.841724967893822e-05} +{"train_loss": 0.01660376414656639, "grad_norm": 0.09241413325071335, "global_step": 20136, "epoch": 189, "lr": 6.841449426480324e-05} +{"train_loss": 0.017259977757930756, "grad_norm": 0.07523222267627716, "global_step": 20137, "epoch": 189, "lr": 6.841173878596774e-05} +{"train_loss": 0.01601102389395237, "grad_norm": 0.05987994000315666, "global_step": 20138, "epoch": 189, "lr": 6.84089832424414e-05} +{"train_loss": 0.016092205095052155, "grad_norm": 0.1479187160730362, "global_step": 20139, "epoch": 189, "lr": 6.840622763423391e-05, "val_loss": 0.019465913996100426} +{"train_loss": 0.015376475639641285, "grad_norm": 0.06175846606492996, "global_step": 20140, "epoch": 190, "lr": 6.840347196135491e-05} +{"train_loss": 0.016315946355462074, "grad_norm": 0.06044004485011101, "global_step": 20141, "epoch": 190, "lr": 6.840071622381414e-05} +{"train_loss": 0.013067254796624184, "grad_norm": 0.08113151043653488, "global_step": 20142, "epoch": 190, "lr": 6.839796042162125e-05} +{"train_loss": 0.013041017577052116, "grad_norm": 0.054251331835985184, "global_step": 20143, "epoch": 190, "lr": 6.839520455478592e-05} +{"train_loss": 0.015930697321891785, "grad_norm": 0.09857705980539322, "global_step": 20144, "epoch": 190, "lr": 6.839244862331785e-05} +{"train_loss": 0.014628183096647263, "grad_norm": 0.11451303213834763, "global_step": 20145, "epoch": 190, "lr": 6.83896926272267e-05} +{"train_loss": 0.015387746505439281, "grad_norm": 0.12741439044475555, "global_step": 20146, "epoch": 190, "lr": 6.838693656652219e-05} +{"train_loss": 0.01913345232605934, "grad_norm": 0.10071401298046112, "global_step": 20147, "epoch": 190, "lr": 6.838418044121397e-05} +{"train_loss": 0.014806609600782394, "grad_norm": 0.09181995689868927, "global_step": 20148, "epoch": 190, "lr": 6.838142425131173e-05} +{"train_loss": 0.014957401901483536, "grad_norm": 0.06174156442284584, "global_step": 20149, "epoch": 190, "lr": 6.837866799682515e-05} +{"train_loss": 0.016655514016747475, "grad_norm": 0.15102535486221313, "global_step": 20150, "epoch": 190, "lr": 6.837591167776396e-05} +{"train_loss": 0.016127049922943115, "grad_norm": 0.11065182834863663, "global_step": 20151, "epoch": 190, "lr": 6.837315529413778e-05} +{"train_loss": 0.016736559569835663, "grad_norm": 0.0616818405687809, "global_step": 20152, "epoch": 190, "lr": 6.837039884595631e-05} +{"train_loss": 0.015770435333251953, "grad_norm": 0.06283034384250641, "global_step": 20153, "epoch": 190, "lr": 6.836764233322928e-05} +{"train_loss": 0.016129691153764725, "grad_norm": 0.06745070219039917, "global_step": 20154, "epoch": 190, "lr": 6.83648857559663e-05} +{"train_loss": 0.018888413906097412, "grad_norm": 0.05824146047234535, "global_step": 20155, "epoch": 190, "lr": 6.836212911417715e-05} +{"train_loss": 0.01748310960829258, "grad_norm": 0.07942182570695877, "global_step": 20156, "epoch": 190, "lr": 6.835937240787144e-05} +{"train_loss": 0.017385561019182205, "grad_norm": 0.15610621869564056, "global_step": 20157, "epoch": 190, "lr": 6.835661563705888e-05} +{"train_loss": 0.01649368181824684, "grad_norm": 0.06631537526845932, "global_step": 20158, "epoch": 190, "lr": 6.835385880174916e-05} +{"train_loss": 0.016118627041578293, "grad_norm": 0.07962317019701004, "global_step": 20159, "epoch": 190, "lr": 6.835110190195195e-05} +{"train_loss": 0.016684640198946, "grad_norm": 0.1218232661485672, "global_step": 20160, "epoch": 190, "lr": 6.834834493767695e-05} +{"train_loss": 0.01692378893494606, "grad_norm": 0.07465766370296478, "global_step": 20161, "epoch": 190, "lr": 6.834558790893385e-05} +{"train_loss": 0.01310691423714161, "grad_norm": 0.07270505279302597, "global_step": 20162, "epoch": 190, "lr": 6.834283081573234e-05} +{"train_loss": 0.016929470002651215, "grad_norm": 0.0717877522110939, "global_step": 20163, "epoch": 190, "lr": 6.834007365808209e-05} +{"train_loss": 0.014420400373637676, "grad_norm": 0.09916933625936508, "global_step": 20164, "epoch": 190, "lr": 6.83373164359928e-05} +{"train_loss": 0.017664533108472824, "grad_norm": 0.05698566883802414, "global_step": 20165, "epoch": 190, "lr": 6.833455914947416e-05} +{"train_loss": 0.0167299322783947, "grad_norm": 0.07731932401657104, "global_step": 20166, "epoch": 190, "lr": 6.833180179853583e-05} +{"train_loss": 0.01607278361916542, "grad_norm": 0.11520765721797943, "global_step": 20167, "epoch": 190, "lr": 6.832904438318754e-05} +{"train_loss": 0.014792895875871181, "grad_norm": 0.05616018548607826, "global_step": 20168, "epoch": 190, "lr": 6.832628690343896e-05} +{"train_loss": 0.01472647674381733, "grad_norm": 0.11199519783258438, "global_step": 20169, "epoch": 190, "lr": 6.832352935929978e-05} +{"train_loss": 0.017564846202731133, "grad_norm": 0.13496965169906616, "global_step": 20170, "epoch": 190, "lr": 6.832077175077969e-05} +{"train_loss": 0.015485480427742004, "grad_norm": 0.14573447406291962, "global_step": 20171, "epoch": 190, "lr": 6.831801407788834e-05} +{"train_loss": 0.01544552855193615, "grad_norm": 0.08015801757574081, "global_step": 20172, "epoch": 190, "lr": 6.83152563406355e-05} +{"train_loss": 0.014382656663656235, "grad_norm": 0.07624971866607666, "global_step": 20173, "epoch": 190, "lr": 6.831249853903078e-05} +{"train_loss": 0.016667652875185013, "grad_norm": 0.10262637585401535, "global_step": 20174, "epoch": 190, "lr": 6.830974067308393e-05} +{"train_loss": 0.015534445643424988, "grad_norm": 0.06812921166419983, "global_step": 20175, "epoch": 190, "lr": 6.830698274280461e-05} +{"train_loss": 0.015955794602632523, "grad_norm": 0.05659254640340805, "global_step": 20176, "epoch": 190, "lr": 6.83042247482025e-05} +{"train_loss": 0.014822011813521385, "grad_norm": 0.0838339775800705, "global_step": 20177, "epoch": 190, "lr": 6.830146668928732e-05} +{"train_loss": 0.015101603232324123, "grad_norm": 0.07749788463115692, "global_step": 20178, "epoch": 190, "lr": 6.829870856606872e-05} +{"train_loss": 0.01968710869550705, "grad_norm": 0.06470643728971481, "global_step": 20179, "epoch": 190, "lr": 6.829595037855645e-05} +{"train_loss": 0.015658829361200333, "grad_norm": 0.07058008760213852, "global_step": 20180, "epoch": 190, "lr": 6.829319212676014e-05} +{"train_loss": 0.013369238004088402, "grad_norm": 0.05476847290992737, "global_step": 20181, "epoch": 190, "lr": 6.829043381068952e-05} +{"train_loss": 0.01767856813967228, "grad_norm": 0.1177809089422226, "global_step": 20182, "epoch": 190, "lr": 6.828767543035428e-05} +{"train_loss": 0.01825932413339615, "grad_norm": 0.12055113911628723, "global_step": 20183, "epoch": 190, "lr": 6.828491698576407e-05} +{"train_loss": 0.016141092404723167, "grad_norm": 0.110247902572155, "global_step": 20184, "epoch": 190, "lr": 6.828215847692864e-05} +{"train_loss": 0.014266684651374817, "grad_norm": 0.0715947225689888, "global_step": 20185, "epoch": 190, "lr": 6.827939990385766e-05} +{"train_loss": 0.017251459881663322, "grad_norm": 0.05785534158349037, "global_step": 20186, "epoch": 190, "lr": 6.827664126656079e-05} +{"train_loss": 0.017496515065431595, "grad_norm": 0.09302695840597153, "global_step": 20187, "epoch": 190, "lr": 6.827388256504778e-05} +{"train_loss": 0.016448698937892914, "grad_norm": 0.07120176404714584, "global_step": 20188, "epoch": 190, "lr": 6.827112379932829e-05} +{"train_loss": 0.018717151135206223, "grad_norm": 0.14237342774868011, "global_step": 20189, "epoch": 190, "lr": 6.8268364969412e-05} +{"train_loss": 0.014203803613781929, "grad_norm": 0.04764846712350845, "global_step": 20190, "epoch": 190, "lr": 6.826560607530863e-05} +{"train_loss": 0.01799386739730835, "grad_norm": 0.057311590760946274, "global_step": 20191, "epoch": 190, "lr": 6.826284711702786e-05} +{"train_loss": 0.01606018841266632, "grad_norm": 0.07474160194396973, "global_step": 20192, "epoch": 190, "lr": 6.82600880945794e-05} +{"train_loss": 0.016180209815502167, "grad_norm": 0.06110837683081627, "global_step": 20193, "epoch": 190, "lr": 6.825732900797292e-05} +{"train_loss": 0.013889987021684647, "grad_norm": 0.1415320336818695, "global_step": 20194, "epoch": 190, "lr": 6.825456985721814e-05} +{"train_loss": 0.017003372311592102, "grad_norm": 0.09817428886890411, "global_step": 20195, "epoch": 190, "lr": 6.82518106423247e-05} +{"train_loss": 0.01590234972536564, "grad_norm": 0.15981830656528473, "global_step": 20196, "epoch": 190, "lr": 6.824905136330237e-05} +{"train_loss": 0.015939921140670776, "grad_norm": 0.1292046755552292, "global_step": 20197, "epoch": 190, "lr": 6.824629202016082e-05} +{"train_loss": 0.017977256327867508, "grad_norm": 0.08553974330425262, "global_step": 20198, "epoch": 190, "lr": 6.82435326129097e-05} +{"train_loss": 0.016773678362369537, "grad_norm": 0.12784934043884277, "global_step": 20199, "epoch": 190, "lr": 6.824077314155877e-05} +{"train_loss": 0.01878567412495613, "grad_norm": 0.06388729065656662, "global_step": 20200, "epoch": 190, "lr": 6.823801360611768e-05} +{"train_loss": 0.01648576185107231, "grad_norm": 0.09138859808444977, "global_step": 20201, "epoch": 190, "lr": 6.823525400659615e-05} +{"train_loss": 0.01230858825147152, "grad_norm": 0.05995249003171921, "global_step": 20202, "epoch": 190, "lr": 6.823249434300386e-05} +{"train_loss": 0.013627966865897179, "grad_norm": 0.057778652757406235, "global_step": 20203, "epoch": 190, "lr": 6.822973461535051e-05} +{"train_loss": 0.017166871577501297, "grad_norm": 0.10015825182199478, "global_step": 20204, "epoch": 190, "lr": 6.822697482364582e-05} +{"train_loss": 0.015371985733509064, "grad_norm": 0.08476891368627548, "global_step": 20205, "epoch": 190, "lr": 6.822421496789945e-05} +{"train_loss": 0.014521717093884945, "grad_norm": 0.04941399395465851, "global_step": 20206, "epoch": 190, "lr": 6.822145504812111e-05} +{"train_loss": 0.01663220487535, "grad_norm": 0.09778106957674026, "global_step": 20207, "epoch": 190, "lr": 6.821869506432052e-05} +{"train_loss": 0.01527821458876133, "grad_norm": 0.0708746686577797, "global_step": 20208, "epoch": 190, "lr": 6.821593501650736e-05} +{"train_loss": 0.019115421921014786, "grad_norm": 0.07923179119825363, "global_step": 20209, "epoch": 190, "lr": 6.821317490469131e-05} +{"train_loss": 0.017254330217838287, "grad_norm": 0.1983504593372345, "global_step": 20210, "epoch": 190, "lr": 6.821041472888208e-05} +{"train_loss": 0.014685171656310558, "grad_norm": 0.07511918246746063, "global_step": 20211, "epoch": 190, "lr": 6.82076544890894e-05} +{"train_loss": 0.01645616814494133, "grad_norm": 0.06902629882097244, "global_step": 20212, "epoch": 190, "lr": 6.820489418532291e-05} +{"train_loss": 0.0169723778963089, "grad_norm": 0.052297137677669525, "global_step": 20213, "epoch": 190, "lr": 6.820213381759235e-05} +{"train_loss": 0.018376853317022324, "grad_norm": 0.1944241225719452, "global_step": 20214, "epoch": 190, "lr": 6.81993733859074e-05} +{"train_loss": 0.018929915502667427, "grad_norm": 0.05755582079291344, "global_step": 20215, "epoch": 190, "lr": 6.819661289027779e-05} +{"train_loss": 0.017810940742492676, "grad_norm": 0.08232974261045456, "global_step": 20216, "epoch": 190, "lr": 6.819385233071319e-05} +{"train_loss": 0.015428634360432625, "grad_norm": 0.0917598158121109, "global_step": 20217, "epoch": 190, "lr": 6.819109170722329e-05} +{"train_loss": 0.013390269130468369, "grad_norm": 0.09767266362905502, "global_step": 20218, "epoch": 190, "lr": 6.81883310198178e-05} +{"train_loss": 0.01848726160824299, "grad_norm": 0.08417169749736786, "global_step": 20219, "epoch": 190, "lr": 6.818557026850645e-05} +{"train_loss": 0.014681731350719929, "grad_norm": 0.07776837795972824, "global_step": 20220, "epoch": 190, "lr": 6.81828094532989e-05} +{"train_loss": 0.017973750829696655, "grad_norm": 0.11482386291027069, "global_step": 20221, "epoch": 190, "lr": 6.818004857420485e-05} +{"train_loss": 0.017316974699497223, "grad_norm": 0.09139683842658997, "global_step": 20222, "epoch": 190, "lr": 6.817728763123404e-05} +{"train_loss": 0.015766551718115807, "grad_norm": 0.06678652763366699, "global_step": 20223, "epoch": 190, "lr": 6.817452662439614e-05} +{"train_loss": 0.01494440995156765, "grad_norm": 0.0700310543179512, "global_step": 20224, "epoch": 190, "lr": 6.817176555370084e-05} +{"train_loss": 0.014807011932134628, "grad_norm": 0.08222010731697083, "global_step": 20225, "epoch": 190, "lr": 6.816900441915787e-05} +{"train_loss": 0.015869546681642532, "grad_norm": 0.07588684558868408, "global_step": 20226, "epoch": 190, "lr": 6.816624322077694e-05} +{"train_loss": 0.014806198887526989, "grad_norm": 0.06342966109514236, "global_step": 20227, "epoch": 190, "lr": 6.816348195856772e-05} +{"train_loss": 0.015546334907412529, "grad_norm": 0.06082786247134209, "global_step": 20228, "epoch": 190, "lr": 6.816072063253991e-05} +{"train_loss": 0.013142010197043419, "grad_norm": 0.04937122017145157, "global_step": 20229, "epoch": 190, "lr": 6.815795924270325e-05} +{"train_loss": 0.017347555607557297, "grad_norm": 0.09207291156053543, "global_step": 20230, "epoch": 190, "lr": 6.815519778906739e-05} +{"train_loss": 0.0162477008998394, "grad_norm": 0.07372944802045822, "global_step": 20231, "epoch": 190, "lr": 6.815243627164207e-05} +{"train_loss": 0.01595308445394039, "grad_norm": 0.058106403797864914, "global_step": 20232, "epoch": 190, "lr": 6.8149674690437e-05} +{"train_loss": 0.016250457614660263, "grad_norm": 0.07582610100507736, "global_step": 20233, "epoch": 190, "lr": 6.814691304546185e-05} +{"train_loss": 0.015457307919859886, "grad_norm": 0.09904926270246506, "global_step": 20234, "epoch": 190, "lr": 6.814415133672633e-05} +{"train_loss": 0.016995832324028015, "grad_norm": 0.060888610780239105, "global_step": 20235, "epoch": 190, "lr": 6.814138956424017e-05} +{"train_loss": 0.016718409955501556, "grad_norm": 0.09664671868085861, "global_step": 20236, "epoch": 190, "lr": 6.813862772801305e-05} +{"train_loss": 0.016848234459757805, "grad_norm": 0.10235672444105148, "global_step": 20237, "epoch": 190, "lr": 6.813586582805468e-05} +{"train_loss": 0.017978433519601822, "grad_norm": 0.13362635672092438, "global_step": 20238, "epoch": 190, "lr": 6.813310386437478e-05} +{"train_loss": 0.015546529553830624, "grad_norm": 0.052105218172073364, "global_step": 20239, "epoch": 190, "lr": 6.813034183698301e-05} +{"train_loss": 0.013765893876552582, "grad_norm": 0.05401398241519928, "global_step": 20240, "epoch": 190, "lr": 6.812757974588912e-05} +{"train_loss": 0.01542544737458229, "grad_norm": 0.06983031332492828, "global_step": 20241, "epoch": 190, "lr": 6.81248175911028e-05} +{"train_loss": 0.014118988066911697, "grad_norm": 0.08158712834119797, "global_step": 20242, "epoch": 190, "lr": 6.812205537263375e-05} +{"train_loss": 0.016057902947068214, "grad_norm": 0.07213649153709412, "global_step": 20243, "epoch": 190, "lr": 6.811929309049167e-05} +{"train_loss": 0.018978018313646317, "grad_norm": 0.0892384946346283, "global_step": 20244, "epoch": 190, "lr": 6.811653074468629e-05} +{"train_loss": 0.016098207433901307, "grad_norm": 0.07590225338935852, "global_step": 20245, "epoch": 190, "lr": 6.811376833522729e-05, "train_batch_mse_error": 0.014285900630056858, "val_loss": 0.020045818760991096} +{"train_loss": 0.01447917241603136, "grad_norm": 0.05201185494661331, "global_step": 20246, "epoch": 191, "lr": 6.811100586212438e-05} +{"train_loss": 0.016969207674264908, "grad_norm": 0.0809054747223854, "global_step": 20247, "epoch": 191, "lr": 6.810824332538728e-05} +{"train_loss": 0.01720776967704296, "grad_norm": 0.07187933474779129, "global_step": 20248, "epoch": 191, "lr": 6.810548072502569e-05} +{"train_loss": 0.018694713711738586, "grad_norm": 0.1398807018995285, "global_step": 20249, "epoch": 191, "lr": 6.810271806104931e-05} +{"train_loss": 0.015586365945637226, "grad_norm": 0.1161041110754013, "global_step": 20250, "epoch": 191, "lr": 6.809995533346786e-05} +{"train_loss": 0.016561420634388924, "grad_norm": 0.08159936964511871, "global_step": 20251, "epoch": 191, "lr": 6.809719254229104e-05} +{"train_loss": 0.01687433198094368, "grad_norm": 0.10439258813858032, "global_step": 20252, "epoch": 191, "lr": 6.809442968752853e-05} +{"train_loss": 0.015247484669089317, "grad_norm": 0.13373297452926636, "global_step": 20253, "epoch": 191, "lr": 6.809166676919008e-05} +{"train_loss": 0.01663154736161232, "grad_norm": 0.0733165517449379, "global_step": 20254, "epoch": 191, "lr": 6.808890378728538e-05} +{"train_loss": 0.016085121780633926, "grad_norm": 0.07060954719781876, "global_step": 20255, "epoch": 191, "lr": 6.808614074182414e-05} +{"train_loss": 0.017501991242170334, "grad_norm": 0.09404154121875763, "global_step": 20256, "epoch": 191, "lr": 6.808337763281607e-05} +{"train_loss": 0.013131966814398766, "grad_norm": 0.06685720384120941, "global_step": 20257, "epoch": 191, "lr": 6.808061446027087e-05} +{"train_loss": 0.01525474339723587, "grad_norm": 0.11032722890377045, "global_step": 20258, "epoch": 191, "lr": 6.807785122419826e-05} +{"train_loss": 0.01673676073551178, "grad_norm": 0.06947795301675797, "global_step": 20259, "epoch": 191, "lr": 6.807508792460794e-05} +{"train_loss": 0.017767705023288727, "grad_norm": 0.05875759199261665, "global_step": 20260, "epoch": 191, "lr": 6.807232456150962e-05} +{"train_loss": 0.013975519686937332, "grad_norm": 0.05281014367938042, "global_step": 20261, "epoch": 191, "lr": 6.806956113491302e-05} +{"train_loss": 0.01621224917471409, "grad_norm": 0.10259240865707397, "global_step": 20262, "epoch": 191, "lr": 6.806679764482783e-05} +{"train_loss": 0.013350254856050014, "grad_norm": 0.08280880749225616, "global_step": 20263, "epoch": 191, "lr": 6.806403409126377e-05} +{"train_loss": 0.014760054647922516, "grad_norm": 0.06567764282226562, "global_step": 20264, "epoch": 191, "lr": 6.806127047423055e-05} +{"train_loss": 0.016847655177116394, "grad_norm": 0.07177789509296417, "global_step": 20265, "epoch": 191, "lr": 6.805850679373788e-05} +{"train_loss": 0.014579052105545998, "grad_norm": 0.08183585107326508, "global_step": 20266, "epoch": 191, "lr": 6.805574304979548e-05} +{"train_loss": 0.01671295240521431, "grad_norm": 0.14630572497844696, "global_step": 20267, "epoch": 191, "lr": 6.805297924241305e-05} +{"train_loss": 0.014315001666545868, "grad_norm": 0.10290291905403137, "global_step": 20268, "epoch": 191, "lr": 6.805021537160027e-05} +{"train_loss": 0.01565227471292019, "grad_norm": 0.1333632469177246, "global_step": 20269, "epoch": 191, "lr": 6.804745143736692e-05} +{"train_loss": 0.017675243318080902, "grad_norm": 0.14729227125644684, "global_step": 20270, "epoch": 191, "lr": 6.804468743972266e-05} +{"train_loss": 0.016092417761683464, "grad_norm": 0.0800212100148201, "global_step": 20271, "epoch": 191, "lr": 6.804192337867723e-05} +{"train_loss": 0.018978245556354523, "grad_norm": 0.06557062268257141, "global_step": 20272, "epoch": 191, "lr": 6.803915925424032e-05} +{"train_loss": 0.01860198751091957, "grad_norm": 0.15320082008838654, "global_step": 20273, "epoch": 191, "lr": 6.803639506642163e-05} +{"train_loss": 0.013425180688500404, "grad_norm": 0.07457021623849869, "global_step": 20274, "epoch": 191, "lr": 6.803363081523092e-05} +{"train_loss": 0.015573178417980671, "grad_norm": 0.06037720665335655, "global_step": 20275, "epoch": 191, "lr": 6.803086650067785e-05} +{"train_loss": 0.016505509614944458, "grad_norm": 0.12064246088266373, "global_step": 20276, "epoch": 191, "lr": 6.802810212277217e-05} +{"train_loss": 0.013955752365291119, "grad_norm": 0.11871317774057388, "global_step": 20277, "epoch": 191, "lr": 6.802533768152359e-05} +{"train_loss": 0.01538938283920288, "grad_norm": 0.0596480593085289, "global_step": 20278, "epoch": 191, "lr": 6.802257317694179e-05} +{"train_loss": 0.014287006109952927, "grad_norm": 0.10052420198917389, "global_step": 20279, "epoch": 191, "lr": 6.801980860903651e-05} +{"train_loss": 0.017366576939821243, "grad_norm": 0.14747096598148346, "global_step": 20280, "epoch": 191, "lr": 6.801704397781746e-05} +{"train_loss": 0.01580468751490116, "grad_norm": 0.09115054458379745, "global_step": 20281, "epoch": 191, "lr": 6.801427928329436e-05} +{"train_loss": 0.015300285071134567, "grad_norm": 0.11711592972278595, "global_step": 20282, "epoch": 191, "lr": 6.801151452547692e-05} +{"train_loss": 0.01921377331018448, "grad_norm": 0.10789227485656738, "global_step": 20283, "epoch": 191, "lr": 6.800874970437484e-05} +{"train_loss": 0.01656993478536606, "grad_norm": 0.10654141753911972, "global_step": 20284, "epoch": 191, "lr": 6.800598481999786e-05} +{"train_loss": 0.0152959693223238, "grad_norm": 0.1282954066991806, "global_step": 20285, "epoch": 191, "lr": 6.800321987235567e-05} +{"train_loss": 0.014293866232037544, "grad_norm": 0.061337754130363464, "global_step": 20286, "epoch": 191, "lr": 6.800045486145799e-05} +{"train_loss": 0.01658577285706997, "grad_norm": 0.08789549022912979, "global_step": 20287, "epoch": 191, "lr": 6.799768978731453e-05} +{"train_loss": 0.017926372587680817, "grad_norm": 0.057719726115465164, "global_step": 20288, "epoch": 191, "lr": 6.799492464993504e-05} +{"train_loss": 0.01641150936484337, "grad_norm": 0.08913791179656982, "global_step": 20289, "epoch": 191, "lr": 6.799215944932921e-05} +{"train_loss": 0.018934644758701324, "grad_norm": 0.0741257295012474, "global_step": 20290, "epoch": 191, "lr": 6.798939418550673e-05} +{"train_loss": 0.014323962852358818, "grad_norm": 0.0975627452135086, "global_step": 20291, "epoch": 191, "lr": 6.798662885847737e-05} +{"train_loss": 0.01764802820980549, "grad_norm": 0.07432875782251358, "global_step": 20292, "epoch": 191, "lr": 6.798386346825081e-05} +{"train_loss": 0.01878402568399906, "grad_norm": 0.12029676884412766, "global_step": 20293, "epoch": 191, "lr": 6.798109801483676e-05} +{"train_loss": 0.015294952318072319, "grad_norm": 0.054411642253398895, "global_step": 20294, "epoch": 191, "lr": 6.797833249824497e-05} +{"train_loss": 0.017441514879465103, "grad_norm": 0.1717439889907837, "global_step": 20295, "epoch": 191, "lr": 6.797556691848514e-05} +{"train_loss": 0.015754595398902893, "grad_norm": 0.13039755821228027, "global_step": 20296, "epoch": 191, "lr": 6.797280127556697e-05} +{"train_loss": 0.017610587179660797, "grad_norm": 0.1052040085196495, "global_step": 20297, "epoch": 191, "lr": 6.79700355695002e-05} +{"train_loss": 0.01637936756014824, "grad_norm": 0.11811348050832748, "global_step": 20298, "epoch": 191, "lr": 6.796726980029454e-05} +{"train_loss": 0.01589813083410263, "grad_norm": 0.0635126531124115, "global_step": 20299, "epoch": 191, "lr": 6.79645039679597e-05} +{"train_loss": 0.016998745501041412, "grad_norm": 0.09646868705749512, "global_step": 20300, "epoch": 191, "lr": 6.796173807250542e-05} +{"train_loss": 0.014983514323830605, "grad_norm": 0.05663931742310524, "global_step": 20301, "epoch": 191, "lr": 6.79589721139414e-05} +{"train_loss": 0.017773114144802094, "grad_norm": 0.08688186854124069, "global_step": 20302, "epoch": 191, "lr": 6.795620609227736e-05} +{"train_loss": 0.015107683837413788, "grad_norm": 0.10596172511577606, "global_step": 20303, "epoch": 191, "lr": 6.795344000752302e-05} +{"train_loss": 0.014640573412179947, "grad_norm": 0.11825074255466461, "global_step": 20304, "epoch": 191, "lr": 6.795067385968809e-05} +{"train_loss": 0.01575443148612976, "grad_norm": 0.051371946930885315, "global_step": 20305, "epoch": 191, "lr": 6.794790764878231e-05} +{"train_loss": 0.016569770872592926, "grad_norm": 0.10191413760185242, "global_step": 20306, "epoch": 191, "lr": 6.794514137481538e-05} +{"train_loss": 0.016810303553938866, "grad_norm": 0.05869712308049202, "global_step": 20307, "epoch": 191, "lr": 6.794237503779704e-05} +{"train_loss": 0.01587229035794735, "grad_norm": 0.06703885644674301, "global_step": 20308, "epoch": 191, "lr": 6.7939608637737e-05} +{"train_loss": 0.016805291175842285, "grad_norm": 0.08144862204790115, "global_step": 20309, "epoch": 191, "lr": 6.793684217464496e-05} +{"train_loss": 0.014043373987078667, "grad_norm": 0.07512792944908142, "global_step": 20310, "epoch": 191, "lr": 6.793407564853064e-05} +{"train_loss": 0.017986610531806946, "grad_norm": 0.07876162230968475, "global_step": 20311, "epoch": 191, "lr": 6.793130905940381e-05} +{"train_loss": 0.015803515911102295, "grad_norm": 0.05852856487035751, "global_step": 20312, "epoch": 191, "lr": 6.792854240727416e-05} +{"train_loss": 0.013910925947129726, "grad_norm": 0.06305111944675446, "global_step": 20313, "epoch": 191, "lr": 6.792577569215139e-05} +{"train_loss": 0.01522595714777708, "grad_norm": 0.05562930181622505, "global_step": 20314, "epoch": 191, "lr": 6.792300891404526e-05} +{"train_loss": 0.015414313413202763, "grad_norm": 0.05675666406750679, "global_step": 20315, "epoch": 191, "lr": 6.792024207296546e-05} +{"train_loss": 0.017025060951709747, "grad_norm": 0.07041074335575104, "global_step": 20316, "epoch": 191, "lr": 6.791747516892171e-05} +{"train_loss": 0.014525053091347218, "grad_norm": 0.1087627187371254, "global_step": 20317, "epoch": 191, "lr": 6.791470820192375e-05} +{"train_loss": 0.01577826216816902, "grad_norm": 0.1279895305633545, "global_step": 20318, "epoch": 191, "lr": 6.791194117198131e-05} +{"train_loss": 0.014371245168149471, "grad_norm": 0.04668959975242615, "global_step": 20319, "epoch": 191, "lr": 6.790917407910408e-05} +{"train_loss": 0.015192625112831593, "grad_norm": 0.13475774228572845, "global_step": 20320, "epoch": 191, "lr": 6.790640692330184e-05} +{"train_loss": 0.014577271416783333, "grad_norm": 0.06377686560153961, "global_step": 20321, "epoch": 191, "lr": 6.790363970458422e-05} +{"train_loss": 0.015748634934425354, "grad_norm": 0.07847660779953003, "global_step": 20322, "epoch": 191, "lr": 6.790087242296103e-05} +{"train_loss": 0.01509622111916542, "grad_norm": 0.1233188658952713, "global_step": 20323, "epoch": 191, "lr": 6.789810507844195e-05} +{"train_loss": 0.017443407326936722, "grad_norm": 0.07137778401374817, "global_step": 20324, "epoch": 191, "lr": 6.789533767103671e-05} +{"train_loss": 0.020153295248746872, "grad_norm": 0.15365520119667053, "global_step": 20325, "epoch": 191, "lr": 6.789257020075504e-05} +{"train_loss": 0.01595938950777054, "grad_norm": 0.0793236792087555, "global_step": 20326, "epoch": 191, "lr": 6.788980266760666e-05} +{"train_loss": 0.01806265488266945, "grad_norm": 0.15535226464271545, "global_step": 20327, "epoch": 191, "lr": 6.78870350716013e-05} +{"train_loss": 0.015977192670106888, "grad_norm": 0.07443591952323914, "global_step": 20328, "epoch": 191, "lr": 6.788426741274867e-05} +{"train_loss": 0.014019114896655083, "grad_norm": 0.04718790948390961, "global_step": 20329, "epoch": 191, "lr": 6.78814996910585e-05} +{"train_loss": 0.014146529138088226, "grad_norm": 0.060629952698946, "global_step": 20330, "epoch": 191, "lr": 6.787873190654052e-05} +{"train_loss": 0.015642931684851646, "grad_norm": 0.052800871431827545, "global_step": 20331, "epoch": 191, "lr": 6.787596405920444e-05} +{"train_loss": 0.017182324081659317, "grad_norm": 0.05484134331345558, "global_step": 20332, "epoch": 191, "lr": 6.787319614906001e-05} +{"train_loss": 0.019309746101498604, "grad_norm": 0.12233902513980865, "global_step": 20333, "epoch": 191, "lr": 6.787042817611693e-05} +{"train_loss": 0.018027808517217636, "grad_norm": 0.08793966472148895, "global_step": 20334, "epoch": 191, "lr": 6.786766014038495e-05} +{"train_loss": 0.013926309533417225, "grad_norm": 0.1326882392168045, "global_step": 20335, "epoch": 191, "lr": 6.786489204187377e-05} +{"train_loss": 0.01697639934718609, "grad_norm": 0.15382881462574005, "global_step": 20336, "epoch": 191, "lr": 6.786212388059314e-05} +{"train_loss": 0.017636612057685852, "grad_norm": 0.11882263422012329, "global_step": 20337, "epoch": 191, "lr": 6.785935565655277e-05} +{"train_loss": 0.014712655916810036, "grad_norm": 0.05535959452390671, "global_step": 20338, "epoch": 191, "lr": 6.785658736976239e-05} +{"train_loss": 0.01671268604695797, "grad_norm": 0.10769816488027573, "global_step": 20339, "epoch": 191, "lr": 6.785381902023174e-05} +{"train_loss": 0.016168830916285515, "grad_norm": 0.09758675843477249, "global_step": 20340, "epoch": 191, "lr": 6.785105060797052e-05} +{"train_loss": 0.016912145540118217, "grad_norm": 0.0885717049241066, "global_step": 20341, "epoch": 191, "lr": 6.784828213298847e-05} +{"train_loss": 0.015575291588902473, "grad_norm": 0.0710063949227333, "global_step": 20342, "epoch": 191, "lr": 6.784551359529533e-05} +{"train_loss": 0.014623848721385002, "grad_norm": 0.05680268630385399, "global_step": 20343, "epoch": 191, "lr": 6.78427449949008e-05} +{"train_loss": 0.01700843870639801, "grad_norm": 0.09792659431695938, "global_step": 20344, "epoch": 191, "lr": 6.783997633181464e-05} +{"train_loss": 0.018384337425231934, "grad_norm": 0.0760565847158432, "global_step": 20345, "epoch": 191, "lr": 6.783720760604655e-05} +{"train_loss": 0.0140617024153471, "grad_norm": 0.058137644082307816, "global_step": 20346, "epoch": 191, "lr": 6.783443881760628e-05} +{"train_loss": 0.016224388033151627, "grad_norm": 0.19552509486675262, "global_step": 20347, "epoch": 191, "lr": 6.783166996650355e-05} +{"train_loss": 0.01714157871901989, "grad_norm": 0.1826876997947693, "global_step": 20348, "epoch": 191, "lr": 6.782890105274805e-05} +{"train_loss": 0.017207643017172813, "grad_norm": 0.11007676273584366, "global_step": 20349, "epoch": 191, "lr": 6.78261320763496e-05} +{"train_loss": 0.014210027642548084, "grad_norm": 0.14474791288375854, "global_step": 20350, "epoch": 191, "lr": 6.782336303731785e-05} +{"train_loss": 0.01610733435998829, "grad_norm": 0.11453620344400406, "global_step": 20351, "epoch": 191, "lr": 6.782059393566253e-05, "val_loss": 0.019684340804815292} +{"train_loss": 0.014716229401528835, "grad_norm": 0.06917930394411087, "global_step": 20352, "epoch": 192, "lr": 6.781782477139343e-05} +{"train_loss": 0.019690144807100296, "grad_norm": 0.06542833894491196, "global_step": 20353, "epoch": 192, "lr": 6.781505554452022e-05} +{"train_loss": 0.01619075983762741, "grad_norm": 0.09167168289422989, "global_step": 20354, "epoch": 192, "lr": 6.781228625505267e-05} +{"train_loss": 0.017090026289224625, "grad_norm": 0.0940096452832222, "global_step": 20355, "epoch": 192, "lr": 6.780951690300048e-05} +{"train_loss": 0.01600988768041134, "grad_norm": 0.10791292786598206, "global_step": 20356, "epoch": 192, "lr": 6.780674748837338e-05} +{"train_loss": 0.015374008566141129, "grad_norm": 0.086519755423069, "global_step": 20357, "epoch": 192, "lr": 6.780397801118113e-05} +{"train_loss": 0.018778890371322632, "grad_norm": 0.06305214017629623, "global_step": 20358, "epoch": 192, "lr": 6.780120847143346e-05} +{"train_loss": 0.018825802952051163, "grad_norm": 0.07736486196517944, "global_step": 20359, "epoch": 192, "lr": 6.779843886914006e-05} +{"train_loss": 0.016149817034602165, "grad_norm": 0.06955497711896896, "global_step": 20360, "epoch": 192, "lr": 6.779566920431068e-05} +{"train_loss": 0.015785429626703262, "grad_norm": 0.11871301382780075, "global_step": 20361, "epoch": 192, "lr": 6.779289947695508e-05} +{"train_loss": 0.016541961580514908, "grad_norm": 0.060686759650707245, "global_step": 20362, "epoch": 192, "lr": 6.779012968708296e-05} +{"train_loss": 0.015763631090521812, "grad_norm": 0.12235192954540253, "global_step": 20363, "epoch": 192, "lr": 6.778735983470406e-05} +{"train_loss": 0.01630694419145584, "grad_norm": 0.06056307256221771, "global_step": 20364, "epoch": 192, "lr": 6.778458991982812e-05} +{"train_loss": 0.014238141477108002, "grad_norm": 0.06066422164440155, "global_step": 20365, "epoch": 192, "lr": 6.778181994246486e-05} +{"train_loss": 0.01652735285460949, "grad_norm": 0.080287866294384, "global_step": 20366, "epoch": 192, "lr": 6.777904990262401e-05} +{"train_loss": 0.01622900925576687, "grad_norm": 0.1481754630804062, "global_step": 20367, "epoch": 192, "lr": 6.777627980031532e-05} +{"train_loss": 0.014627596363425255, "grad_norm": 0.04732082411646843, "global_step": 20368, "epoch": 192, "lr": 6.777350963554851e-05} +{"train_loss": 0.01661291904747486, "grad_norm": 0.11250591278076172, "global_step": 20369, "epoch": 192, "lr": 6.777073940833332e-05} +{"train_loss": 0.016220571473240852, "grad_norm": 0.13153497874736786, "global_step": 20370, "epoch": 192, "lr": 6.776796911867949e-05} +{"train_loss": 0.01750798709690571, "grad_norm": 0.17117202281951904, "global_step": 20371, "epoch": 192, "lr": 6.776519876659673e-05} +{"train_loss": 0.015822846442461014, "grad_norm": 0.10526077449321747, "global_step": 20372, "epoch": 192, "lr": 6.776242835209477e-05} +{"train_loss": 0.014285476878285408, "grad_norm": 0.06750769913196564, "global_step": 20373, "epoch": 192, "lr": 6.775965787518339e-05} +{"train_loss": 0.015426548197865486, "grad_norm": 0.12013252079486847, "global_step": 20374, "epoch": 192, "lr": 6.775688733587228e-05} +{"train_loss": 0.014071833342313766, "grad_norm": 0.13312558829784393, "global_step": 20375, "epoch": 192, "lr": 6.77541167341712e-05} +{"train_loss": 0.013334132730960846, "grad_norm": 0.06487354636192322, "global_step": 20376, "epoch": 192, "lr": 6.775134607008988e-05} +{"train_loss": 0.015191575512290001, "grad_norm": 0.07274580001831055, "global_step": 20377, "epoch": 192, "lr": 6.774857534363804e-05} +{"train_loss": 0.01602357253432274, "grad_norm": 0.06520958989858627, "global_step": 20378, "epoch": 192, "lr": 6.774580455482543e-05} +{"train_loss": 0.01636306755244732, "grad_norm": 0.05043511837720871, "global_step": 20379, "epoch": 192, "lr": 6.774303370366176e-05} +{"train_loss": 0.01873895712196827, "grad_norm": 0.06482305377721786, "global_step": 20380, "epoch": 192, "lr": 6.774026279015681e-05} +{"train_loss": 0.015691619366407394, "grad_norm": 0.053123198449611664, "global_step": 20381, "epoch": 192, "lr": 6.773749181432027e-05} +{"train_loss": 0.01672383025288582, "grad_norm": 0.1512644737958908, "global_step": 20382, "epoch": 192, "lr": 6.773472077616193e-05} +{"train_loss": 0.015742937102913857, "grad_norm": 0.1531081646680832, "global_step": 20383, "epoch": 192, "lr": 6.773194967569146e-05} +{"train_loss": 0.016039730980992317, "grad_norm": 0.09216512739658356, "global_step": 20384, "epoch": 192, "lr": 6.772917851291864e-05} +{"train_loss": 0.019023148342967033, "grad_norm": 0.1003810241818428, "global_step": 20385, "epoch": 192, "lr": 6.772640728785322e-05} +{"train_loss": 0.01502265501767397, "grad_norm": 0.13306953012943268, "global_step": 20386, "epoch": 192, "lr": 6.772363600050488e-05} +{"train_loss": 0.01591142825782299, "grad_norm": 0.04673243686556816, "global_step": 20387, "epoch": 192, "lr": 6.772086465088341e-05} +{"train_loss": 0.015628159046173096, "grad_norm": 0.14750583469867706, "global_step": 20388, "epoch": 192, "lr": 6.771809323899853e-05} +{"train_loss": 0.015914926305413246, "grad_norm": 0.05493243411183357, "global_step": 20389, "epoch": 192, "lr": 6.771532176485997e-05} +{"train_loss": 0.015485798940062523, "grad_norm": 0.060116007924079895, "global_step": 20390, "epoch": 192, "lr": 6.771255022847746e-05} +{"train_loss": 0.014872103929519653, "grad_norm": 0.05876067653298378, "global_step": 20391, "epoch": 192, "lr": 6.770977862986078e-05} +{"train_loss": 0.01858210563659668, "grad_norm": 0.06494089215993881, "global_step": 20392, "epoch": 192, "lr": 6.770700696901962e-05} +{"train_loss": 0.015068158507347107, "grad_norm": 0.11189830303192139, "global_step": 20393, "epoch": 192, "lr": 6.770423524596374e-05} +{"train_loss": 0.014959130436182022, "grad_norm": 0.051513414829969406, "global_step": 20394, "epoch": 192, "lr": 6.770146346070289e-05} +{"train_loss": 0.01917215622961521, "grad_norm": 0.09115200489759445, "global_step": 20395, "epoch": 192, "lr": 6.769869161324678e-05} +{"train_loss": 0.015196076594293118, "grad_norm": 0.13371522724628448, "global_step": 20396, "epoch": 192, "lr": 6.769591970360517e-05} +{"train_loss": 0.01419043354690075, "grad_norm": 0.12263672053813934, "global_step": 20397, "epoch": 192, "lr": 6.76931477317878e-05} +{"train_loss": 0.018105261027812958, "grad_norm": 0.17010070383548737, "global_step": 20398, "epoch": 192, "lr": 6.76903756978044e-05} +{"train_loss": 0.013891026377677917, "grad_norm": 0.10378789156675339, "global_step": 20399, "epoch": 192, "lr": 6.768760360166471e-05} +{"train_loss": 0.01581871137022972, "grad_norm": 0.06620737165212631, "global_step": 20400, "epoch": 192, "lr": 6.768483144337847e-05} +{"train_loss": 0.01599104329943657, "grad_norm": 0.1020989939570427, "global_step": 20401, "epoch": 192, "lr": 6.768205922295542e-05} +{"train_loss": 0.015059959143400192, "grad_norm": 0.06947902590036392, "global_step": 20402, "epoch": 192, "lr": 6.76792869404053e-05} +{"train_loss": 0.019521353766322136, "grad_norm": 0.07057084143161774, "global_step": 20403, "epoch": 192, "lr": 6.767651459573787e-05} +{"train_loss": 0.01541117299348116, "grad_norm": 0.0971861407160759, "global_step": 20404, "epoch": 192, "lr": 6.767374218896286e-05} +{"train_loss": 0.015738695859909058, "grad_norm": 0.07452128082513809, "global_step": 20405, "epoch": 192, "lr": 6.767096972008999e-05} +{"train_loss": 0.015681207180023193, "grad_norm": 0.0678488090634346, "global_step": 20406, "epoch": 192, "lr": 6.766819718912903e-05} +{"train_loss": 0.017495855689048767, "grad_norm": 0.10436723381280899, "global_step": 20407, "epoch": 192, "lr": 6.76654245960897e-05} +{"train_loss": 0.013868006877601147, "grad_norm": 0.05178077518939972, "global_step": 20408, "epoch": 192, "lr": 6.766265194098175e-05} +{"train_loss": 0.014126937836408615, "grad_norm": 0.07600659132003784, "global_step": 20409, "epoch": 192, "lr": 6.765987922381494e-05} +{"train_loss": 0.015413731336593628, "grad_norm": 0.062156446278095245, "global_step": 20410, "epoch": 192, "lr": 6.765710644459898e-05} +{"train_loss": 0.015274579636752605, "grad_norm": 0.06612560153007507, "global_step": 20411, "epoch": 192, "lr": 6.765433360334363e-05} +{"train_loss": 0.014632279053330421, "grad_norm": 0.09159047901630402, "global_step": 20412, "epoch": 192, "lr": 6.765156070005861e-05} +{"train_loss": 0.01806400716304779, "grad_norm": 0.1400267332792282, "global_step": 20413, "epoch": 192, "lr": 6.764878773475369e-05} +{"train_loss": 0.016042789444327354, "grad_norm": 0.0884241983294487, "global_step": 20414, "epoch": 192, "lr": 6.764601470743863e-05} +{"train_loss": 0.014676855877041817, "grad_norm": 0.08615270256996155, "global_step": 20415, "epoch": 192, "lr": 6.764324161812312e-05} +{"train_loss": 0.01467417273670435, "grad_norm": 0.08224025368690491, "global_step": 20416, "epoch": 192, "lr": 6.764046846681694e-05} +{"train_loss": 0.014593333005905151, "grad_norm": 0.09780406951904297, "global_step": 20417, "epoch": 192, "lr": 6.763769525352984e-05} +{"train_loss": 0.016736295074224472, "grad_norm": 0.07013507932424545, "global_step": 20418, "epoch": 192, "lr": 6.763492197827153e-05} +{"train_loss": 0.01609746366739273, "grad_norm": 0.05514901876449585, "global_step": 20419, "epoch": 192, "lr": 6.763214864105178e-05} +{"train_loss": 0.01781940460205078, "grad_norm": 0.05652279406785965, "global_step": 20420, "epoch": 192, "lr": 6.762937524188033e-05} +{"train_loss": 0.017371460795402527, "grad_norm": 0.07199415564537048, "global_step": 20421, "epoch": 192, "lr": 6.762660178076692e-05} +{"train_loss": 0.016622234135866165, "grad_norm": 0.0591302290558815, "global_step": 20422, "epoch": 192, "lr": 6.76238282577213e-05} +{"train_loss": 0.013626718893647194, "grad_norm": 0.050745148211717606, "global_step": 20423, "epoch": 192, "lr": 6.762105467275322e-05} +{"train_loss": 0.015768319368362427, "grad_norm": 0.10397592186927795, "global_step": 20424, "epoch": 192, "lr": 6.76182810258724e-05} +{"train_loss": 0.014766549691557884, "grad_norm": 0.04771400988101959, "global_step": 20425, "epoch": 192, "lr": 6.76155073170886e-05} +{"train_loss": 0.01726323738694191, "grad_norm": 0.07391531765460968, "global_step": 20426, "epoch": 192, "lr": 6.761273354641159e-05} +{"train_loss": 0.014146705158054829, "grad_norm": 0.06487789005041122, "global_step": 20427, "epoch": 192, "lr": 6.760995971385108e-05} +{"train_loss": 0.0166691355407238, "grad_norm": 0.08899789303541183, "global_step": 20428, "epoch": 192, "lr": 6.760718581941682e-05} +{"train_loss": 0.014480712823569775, "grad_norm": 0.12231937050819397, "global_step": 20429, "epoch": 192, "lr": 6.760441186311859e-05} +{"train_loss": 0.017499621957540512, "grad_norm": 0.10038967430591583, "global_step": 20430, "epoch": 192, "lr": 6.760163784496609e-05} +{"train_loss": 0.015683960169553757, "grad_norm": 0.09248004853725433, "global_step": 20431, "epoch": 192, "lr": 6.75988637649691e-05} +{"train_loss": 0.017641182988882065, "grad_norm": 0.1076987162232399, "global_step": 20432, "epoch": 192, "lr": 6.759608962313736e-05} +{"train_loss": 0.01475953496992588, "grad_norm": 0.08291595429182053, "global_step": 20433, "epoch": 192, "lr": 6.75933154194806e-05} +{"train_loss": 0.016081981360912323, "grad_norm": 0.08677928894758224, "global_step": 20434, "epoch": 192, "lr": 6.75905411540086e-05} +{"train_loss": 0.015627872198820114, "grad_norm": 0.09598715603351593, "global_step": 20435, "epoch": 192, "lr": 6.758776682673108e-05} +{"train_loss": 0.016304422169923782, "grad_norm": 0.07279616594314575, "global_step": 20436, "epoch": 192, "lr": 6.758499243765779e-05} +{"train_loss": 0.014632810838520527, "grad_norm": 0.054526183754205704, "global_step": 20437, "epoch": 192, "lr": 6.758221798679848e-05} +{"train_loss": 0.016748245805501938, "grad_norm": 0.08896622806787491, "global_step": 20438, "epoch": 192, "lr": 6.757944347416293e-05} +{"train_loss": 0.018442846834659576, "grad_norm": 0.09071136265993118, "global_step": 20439, "epoch": 192, "lr": 6.757666889976085e-05} +{"train_loss": 0.01694396324455738, "grad_norm": 0.06398650258779526, "global_step": 20440, "epoch": 192, "lr": 6.757389426360197e-05} +{"train_loss": 0.016415737569332123, "grad_norm": 0.06709631532430649, "global_step": 20441, "epoch": 192, "lr": 6.757111956569611e-05} +{"train_loss": 0.015831520780920982, "grad_norm": 0.07093136012554169, "global_step": 20442, "epoch": 192, "lr": 6.756834480605295e-05} +{"train_loss": 0.016192730516195297, "grad_norm": 0.060864076018333435, "global_step": 20443, "epoch": 192, "lr": 6.756556998468226e-05} +{"train_loss": 0.017081091180443764, "grad_norm": 0.13338805735111237, "global_step": 20444, "epoch": 192, "lr": 6.756279510159382e-05} +{"train_loss": 0.015011349692940712, "grad_norm": 0.09450946003198624, "global_step": 20445, "epoch": 192, "lr": 6.756002015679733e-05} +{"train_loss": 0.014619100838899612, "grad_norm": 0.05416858196258545, "global_step": 20446, "epoch": 192, "lr": 6.75572451503026e-05} +{"train_loss": 0.013914095237851143, "grad_norm": 0.160471573472023, "global_step": 20447, "epoch": 192, "lr": 6.755447008211932e-05} +{"train_loss": 0.018163932487368584, "grad_norm": 0.05611386150121689, "global_step": 20448, "epoch": 192, "lr": 6.755169495225727e-05} +{"train_loss": 0.017393533140420914, "grad_norm": 0.07826049625873566, "global_step": 20449, "epoch": 192, "lr": 6.75489197607262e-05} +{"train_loss": 0.015359938144683838, "grad_norm": 0.08862345665693283, "global_step": 20450, "epoch": 192, "lr": 6.754614450753585e-05} +{"train_loss": 0.015758298337459564, "grad_norm": 0.048817507922649384, "global_step": 20451, "epoch": 192, "lr": 6.754336919269599e-05} +{"train_loss": 0.017007403075695038, "grad_norm": 0.14123333990573883, "global_step": 20452, "epoch": 192, "lr": 6.754059381621635e-05} +{"train_loss": 0.015303989872336388, "grad_norm": 0.08881498128175735, "global_step": 20453, "epoch": 192, "lr": 6.753781837810671e-05} +{"train_loss": 0.019411595538258553, "grad_norm": 0.12481246143579483, "global_step": 20454, "epoch": 192, "lr": 6.753504287837679e-05} +{"train_loss": 0.014985565096139908, "grad_norm": 0.07062278687953949, "global_step": 20455, "epoch": 192, "lr": 6.753226731703636e-05} +{"train_loss": 0.01615395024418831, "grad_norm": 0.05898265913128853, "global_step": 20456, "epoch": 192, "lr": 6.752949169409517e-05} +{"train_loss": 0.0160788995266523, "grad_norm": 0.1071171760559082, "global_step": 20457, "epoch": 192, "lr": 6.752671600956295e-05, "val_loss": 0.020519979298114777} +{"train_loss": 0.014336385764181614, "grad_norm": 0.08875154703855515, "global_step": 20458, "epoch": 193, "lr": 6.752394026344951e-05} +{"train_loss": 0.01867416687309742, "grad_norm": 0.07327119261026382, "global_step": 20459, "epoch": 193, "lr": 6.752116445576452e-05} +{"train_loss": 0.015296634286642075, "grad_norm": 0.06646626442670822, "global_step": 20460, "epoch": 193, "lr": 6.75183885865178e-05} +{"train_loss": 0.01570476032793522, "grad_norm": 0.09340527653694153, "global_step": 20461, "epoch": 193, "lr": 6.751561265571909e-05} +{"train_loss": 0.014773747883737087, "grad_norm": 0.050843290984630585, "global_step": 20462, "epoch": 193, "lr": 6.75128366633781e-05} +{"train_loss": 0.01734755001962185, "grad_norm": 0.07160162180662155, "global_step": 20463, "epoch": 193, "lr": 6.751006060950466e-05} +{"train_loss": 0.015527847222983837, "grad_norm": 0.0763372853398323, "global_step": 20464, "epoch": 193, "lr": 6.750728449410846e-05} +{"train_loss": 0.01392338890582323, "grad_norm": 0.08570840209722519, "global_step": 20465, "epoch": 193, "lr": 6.750450831719928e-05} +{"train_loss": 0.015073948539793491, "grad_norm": 0.05235312506556511, "global_step": 20466, "epoch": 193, "lr": 6.750173207878686e-05} +{"train_loss": 0.01481518056243658, "grad_norm": 0.07081067562103271, "global_step": 20467, "epoch": 193, "lr": 6.749895577888099e-05} +{"train_loss": 0.016373109072446823, "grad_norm": 0.08856171369552612, "global_step": 20468, "epoch": 193, "lr": 6.749617941749138e-05} +{"train_loss": 0.015526091679930687, "grad_norm": 0.05474379286170006, "global_step": 20469, "epoch": 193, "lr": 6.749340299462781e-05} +{"train_loss": 0.014672832563519478, "grad_norm": 0.06725301593542099, "global_step": 20470, "epoch": 193, "lr": 6.749062651030004e-05} +{"train_loss": 0.014903508126735687, "grad_norm": 0.11972790956497192, "global_step": 20471, "epoch": 193, "lr": 6.74878499645178e-05} +{"train_loss": 0.014776546508073807, "grad_norm": 0.05419294536113739, "global_step": 20472, "epoch": 193, "lr": 6.748507335729085e-05} +{"train_loss": 0.015149847604334354, "grad_norm": 0.07199320197105408, "global_step": 20473, "epoch": 193, "lr": 6.748229668862898e-05} +{"train_loss": 0.01626696065068245, "grad_norm": 0.13911040127277374, "global_step": 20474, "epoch": 193, "lr": 6.74795199585419e-05} +{"train_loss": 0.016185157001018524, "grad_norm": 0.07260572910308838, "global_step": 20475, "epoch": 193, "lr": 6.74767431670394e-05} +{"train_loss": 0.014895424246788025, "grad_norm": 0.05547460541129112, "global_step": 20476, "epoch": 193, "lr": 6.747396631413123e-05} +{"train_loss": 0.01733846217393875, "grad_norm": 0.09631259739398956, "global_step": 20477, "epoch": 193, "lr": 6.747118939982714e-05} +{"train_loss": 0.01659601368010044, "grad_norm": 0.07897883653640747, "global_step": 20478, "epoch": 193, "lr": 6.746841242413689e-05} +{"train_loss": 0.01835193671286106, "grad_norm": 0.09590340405702591, "global_step": 20479, "epoch": 193, "lr": 6.746563538707022e-05} +{"train_loss": 0.017651032656431198, "grad_norm": 0.09154652804136276, "global_step": 20480, "epoch": 193, "lr": 6.746285828863693e-05} +{"train_loss": 0.013689852319657803, "grad_norm": 0.06279975175857544, "global_step": 20481, "epoch": 193, "lr": 6.746008112884672e-05} +{"train_loss": 0.01543857716023922, "grad_norm": 0.12086346745491028, "global_step": 20482, "epoch": 193, "lr": 6.745730390770941e-05} +{"train_loss": 0.017229609191417694, "grad_norm": 0.0772877037525177, "global_step": 20483, "epoch": 193, "lr": 6.74545266252347e-05} +{"train_loss": 0.014072473160922527, "grad_norm": 0.048593636602163315, "global_step": 20484, "epoch": 193, "lr": 6.745174928143238e-05} +{"train_loss": 0.015784267336130142, "grad_norm": 0.07750427722930908, "global_step": 20485, "epoch": 193, "lr": 6.744897187631221e-05} +{"train_loss": 0.01764099672436714, "grad_norm": 0.07164104282855988, "global_step": 20486, "epoch": 193, "lr": 6.744619440988393e-05} +{"train_loss": 0.017344677820801735, "grad_norm": 0.05909447371959686, "global_step": 20487, "epoch": 193, "lr": 6.744341688215732e-05} +{"train_loss": 0.012895063497126102, "grad_norm": 0.07383120805025101, "global_step": 20488, "epoch": 193, "lr": 6.744063929314212e-05} +{"train_loss": 0.016835220158100128, "grad_norm": 0.08727879077196121, "global_step": 20489, "epoch": 193, "lr": 6.74378616428481e-05} +{"train_loss": 0.01630564033985138, "grad_norm": 0.08006767928600311, "global_step": 20490, "epoch": 193, "lr": 6.743508393128504e-05} +{"train_loss": 0.016451813280582428, "grad_norm": 0.05121724680066109, "global_step": 20491, "epoch": 193, "lr": 6.743230615846265e-05} +{"train_loss": 0.013358326628804207, "grad_norm": 0.09913074225187302, "global_step": 20492, "epoch": 193, "lr": 6.742952832439073e-05} +{"train_loss": 0.015750400722026825, "grad_norm": 0.13226133584976196, "global_step": 20493, "epoch": 193, "lr": 6.742675042907903e-05} +{"train_loss": 0.016334768384695053, "grad_norm": 0.09553802013397217, "global_step": 20494, "epoch": 193, "lr": 6.74239724725373e-05} +{"train_loss": 0.013353869318962097, "grad_norm": 0.07125535607337952, "global_step": 20495, "epoch": 193, "lr": 6.742119445477531e-05} +{"train_loss": 0.015946824103593826, "grad_norm": 0.13478592038154602, "global_step": 20496, "epoch": 193, "lr": 6.741841637580281e-05} +{"train_loss": 0.014704596251249313, "grad_norm": 0.13621138036251068, "global_step": 20497, "epoch": 193, "lr": 6.741563823562958e-05} +{"train_loss": 0.015338148921728134, "grad_norm": 0.09679339826107025, "global_step": 20498, "epoch": 193, "lr": 6.741286003426537e-05} +{"train_loss": 0.015500565990805626, "grad_norm": 0.11414901167154312, "global_step": 20499, "epoch": 193, "lr": 6.741008177171995e-05} +{"train_loss": 0.015136039815843105, "grad_norm": 0.047230858355760574, "global_step": 20500, "epoch": 193, "lr": 6.740730344800305e-05} +{"train_loss": 0.01432967558503151, "grad_norm": 0.12015258520841599, "global_step": 20501, "epoch": 193, "lr": 6.740452506312448e-05} +{"train_loss": 0.01304299384355545, "grad_norm": 0.06853408366441727, "global_step": 20502, "epoch": 193, "lr": 6.740174661709397e-05} +{"train_loss": 0.013438438065350056, "grad_norm": 0.06262601166963577, "global_step": 20503, "epoch": 193, "lr": 6.73989681099213e-05} +{"train_loss": 0.017841599881649017, "grad_norm": 0.08089854568243027, "global_step": 20504, "epoch": 193, "lr": 6.739618954161621e-05} +{"train_loss": 0.017406079918146133, "grad_norm": 0.07861776649951935, "global_step": 20505, "epoch": 193, "lr": 6.739341091218847e-05} +{"train_loss": 0.017682021483778954, "grad_norm": 0.10434865206480026, "global_step": 20506, "epoch": 193, "lr": 6.739063222164787e-05} +{"train_loss": 0.016392376273870468, "grad_norm": 0.10871836543083191, "global_step": 20507, "epoch": 193, "lr": 6.738785347000414e-05} +{"train_loss": 0.017013534903526306, "grad_norm": 0.05175836756825447, "global_step": 20508, "epoch": 193, "lr": 6.738507465726704e-05} +{"train_loss": 0.017435966059565544, "grad_norm": 0.18402349948883057, "global_step": 20509, "epoch": 193, "lr": 6.738229578344637e-05} +{"train_loss": 0.017932694405317307, "grad_norm": 0.16985085606575012, "global_step": 20510, "epoch": 193, "lr": 6.737951684855185e-05} +{"train_loss": 0.01768999919295311, "grad_norm": 0.1975586861371994, "global_step": 20511, "epoch": 193, "lr": 6.73767378525933e-05} +{"train_loss": 0.017524458467960358, "grad_norm": 0.12510091066360474, "global_step": 20512, "epoch": 193, "lr": 6.737395879558041e-05} +{"train_loss": 0.016502391546964645, "grad_norm": 0.07896684855222702, "global_step": 20513, "epoch": 193, "lr": 6.7371179677523e-05} +{"train_loss": 0.01609698310494423, "grad_norm": 0.08838225901126862, "global_step": 20514, "epoch": 193, "lr": 6.736840049843083e-05} +{"train_loss": 0.014220194891095161, "grad_norm": 0.06855789572000504, "global_step": 20515, "epoch": 193, "lr": 6.736562125831365e-05} +{"train_loss": 0.020974811166524887, "grad_norm": 0.09680873900651932, "global_step": 20516, "epoch": 193, "lr": 6.736284195718122e-05} +{"train_loss": 0.015369554981589317, "grad_norm": 0.09199593961238861, "global_step": 20517, "epoch": 193, "lr": 6.736006259504333e-05} +{"train_loss": 0.01568252593278885, "grad_norm": 0.09144474565982819, "global_step": 20518, "epoch": 193, "lr": 6.735728317190971e-05} +{"train_loss": 0.016346734017133713, "grad_norm": 0.055613674223423004, "global_step": 20519, "epoch": 193, "lr": 6.735450368779016e-05} +{"train_loss": 0.017384111881256104, "grad_norm": 0.1195744201540947, "global_step": 20520, "epoch": 193, "lr": 6.735172414269443e-05} +{"train_loss": 0.017368048429489136, "grad_norm": 0.07160007953643799, "global_step": 20521, "epoch": 193, "lr": 6.734894453663229e-05} +{"train_loss": 0.01636451855301857, "grad_norm": 0.12312807142734528, "global_step": 20522, "epoch": 193, "lr": 6.734616486961349e-05} +{"train_loss": 0.017528105527162552, "grad_norm": 0.1081608310341835, "global_step": 20523, "epoch": 193, "lr": 6.734338514164782e-05} +{"train_loss": 0.017399324104189873, "grad_norm": 0.08370940387248993, "global_step": 20524, "epoch": 193, "lr": 6.734060535274503e-05} +{"train_loss": 0.016554422676563263, "grad_norm": 0.052466705441474915, "global_step": 20525, "epoch": 193, "lr": 6.73378255029149e-05} +{"train_loss": 0.015736542642116547, "grad_norm": 0.15576216578483582, "global_step": 20526, "epoch": 193, "lr": 6.733504559216719e-05} +{"train_loss": 0.017097122967243195, "grad_norm": 0.1285078376531601, "global_step": 20527, "epoch": 193, "lr": 6.733226562051167e-05} +{"train_loss": 0.015975307673215866, "grad_norm": 0.11590997874736786, "global_step": 20528, "epoch": 193, "lr": 6.732948558795811e-05} +{"train_loss": 0.015598117373883724, "grad_norm": 0.07802644371986389, "global_step": 20529, "epoch": 193, "lr": 6.732670549451627e-05} +{"train_loss": 0.014600737951695919, "grad_norm": 0.05957318842411041, "global_step": 20530, "epoch": 193, "lr": 6.732392534019591e-05} +{"train_loss": 0.016490891575813293, "grad_norm": 0.07679864764213562, "global_step": 20531, "epoch": 193, "lr": 6.732114512500682e-05} +{"train_loss": 0.017070556059479713, "grad_norm": 0.0587899312376976, "global_step": 20532, "epoch": 193, "lr": 6.731836484895877e-05} +{"train_loss": 0.015891535207629204, "grad_norm": 0.05749310553073883, "global_step": 20533, "epoch": 193, "lr": 6.73155845120615e-05} +{"train_loss": 0.02089429646730423, "grad_norm": 0.10215212404727936, "global_step": 20534, "epoch": 193, "lr": 6.73128041143248e-05} +{"train_loss": 0.015824073925614357, "grad_norm": 0.05648885667324066, "global_step": 20535, "epoch": 193, "lr": 6.731002365575844e-05} +{"train_loss": 0.016254961490631104, "grad_norm": 0.21250462532043457, "global_step": 20536, "epoch": 193, "lr": 6.73072431363722e-05} +{"train_loss": 0.014987721107900143, "grad_norm": 0.08655884116888046, "global_step": 20537, "epoch": 193, "lr": 6.730446255617581e-05} +{"train_loss": 0.016289934515953064, "grad_norm": 0.050033409148454666, "global_step": 20538, "epoch": 193, "lr": 6.730168191517907e-05} +{"train_loss": 0.013897924683988094, "grad_norm": 0.09104623645544052, "global_step": 20539, "epoch": 193, "lr": 6.729890121339175e-05} +{"train_loss": 0.018055716529488564, "grad_norm": 0.16527432203292847, "global_step": 20540, "epoch": 193, "lr": 6.72961204508236e-05} +{"train_loss": 0.018619149923324585, "grad_norm": 0.14022119343280792, "global_step": 20541, "epoch": 193, "lr": 6.729333962748441e-05} +{"train_loss": 0.01550852321088314, "grad_norm": 0.05749589204788208, "global_step": 20542, "epoch": 193, "lr": 6.729055874338397e-05} +{"train_loss": 0.014642472378909588, "grad_norm": 0.11533574759960175, "global_step": 20543, "epoch": 193, "lr": 6.7287777798532e-05} +{"train_loss": 0.014102645218372345, "grad_norm": 0.1538245528936386, "global_step": 20544, "epoch": 193, "lr": 6.72849967929383e-05} +{"train_loss": 0.015430949628353119, "grad_norm": 0.13575629889965057, "global_step": 20545, "epoch": 193, "lr": 6.728221572661265e-05} +{"train_loss": 0.016903026029467583, "grad_norm": 0.06978285312652588, "global_step": 20546, "epoch": 193, "lr": 6.72794345995648e-05} +{"train_loss": 0.016323544085025787, "grad_norm": 0.06142335757613182, "global_step": 20547, "epoch": 193, "lr": 6.727665341180453e-05} +{"train_loss": 0.015453977510333061, "grad_norm": 0.06947090476751328, "global_step": 20548, "epoch": 193, "lr": 6.727387216334162e-05} +{"train_loss": 0.01427440345287323, "grad_norm": 0.06277341395616531, "global_step": 20549, "epoch": 193, "lr": 6.727109085418583e-05} +{"train_loss": 0.01573638617992401, "grad_norm": 0.08260504901409149, "global_step": 20550, "epoch": 193, "lr": 6.726830948434694e-05} +{"train_loss": 0.014739589765667915, "grad_norm": 0.11154069006443024, "global_step": 20551, "epoch": 193, "lr": 6.726552805383472e-05} +{"train_loss": 0.01672174222767353, "grad_norm": 0.07872626185417175, "global_step": 20552, "epoch": 193, "lr": 6.726274656265894e-05} +{"train_loss": 0.018124405294656754, "grad_norm": 0.07885835319757462, "global_step": 20553, "epoch": 193, "lr": 6.725996501082938e-05} +{"train_loss": 0.01780402846634388, "grad_norm": 0.09131311625242233, "global_step": 20554, "epoch": 193, "lr": 6.725718339835581e-05} +{"train_loss": 0.01665041223168373, "grad_norm": 0.06170278042554855, "global_step": 20555, "epoch": 193, "lr": 6.7254401725248e-05} +{"train_loss": 0.018458962440490723, "grad_norm": 0.092259980738163, "global_step": 20556, "epoch": 193, "lr": 6.725161999151574e-05} +{"train_loss": 0.015709534287452698, "grad_norm": 0.05446651950478554, "global_step": 20557, "epoch": 193, "lr": 6.724883819716877e-05} +{"train_loss": 0.01589011773467064, "grad_norm": 0.06517937779426575, "global_step": 20558, "epoch": 193, "lr": 6.724605634221689e-05} +{"train_loss": 0.017637677490711212, "grad_norm": 0.07058238238096237, "global_step": 20559, "epoch": 193, "lr": 6.724327442666987e-05} +{"train_loss": 0.016819650307297707, "grad_norm": 0.1596049666404724, "global_step": 20560, "epoch": 193, "lr": 6.724049245053749e-05} +{"train_loss": 0.014448175206780434, "grad_norm": 0.06123719736933708, "global_step": 20561, "epoch": 193, "lr": 6.72377104138295e-05} +{"train_loss": 0.015520617365837097, "grad_norm": 0.06663532555103302, "global_step": 20562, "epoch": 193, "lr": 6.72349283165557e-05} +{"train_loss": 0.01607947537394346, "grad_norm": 0.12048640102148056, "global_step": 20563, "epoch": 193, "lr": 6.723214615872585e-05, "val_loss": 0.019795237109065056} +{"train_loss": 0.016877517104148865, "grad_norm": 0.09417565166950226, "global_step": 20564, "epoch": 194, "lr": 6.722936394034976e-05} +{"train_loss": 0.01568678393959999, "grad_norm": 0.05451922491192818, "global_step": 20565, "epoch": 194, "lr": 6.722658166143715e-05} +{"train_loss": 0.014086378738284111, "grad_norm": 0.054125022143125534, "global_step": 20566, "epoch": 194, "lr": 6.722379932199782e-05} +{"train_loss": 0.01360725425183773, "grad_norm": 0.06634704768657684, "global_step": 20567, "epoch": 194, "lr": 6.722101692204158e-05} +{"train_loss": 0.01722649484872818, "grad_norm": 0.06324507296085358, "global_step": 20568, "epoch": 194, "lr": 6.721823446157814e-05} +{"train_loss": 0.019805502146482468, "grad_norm": 0.05808921903371811, "global_step": 20569, "epoch": 194, "lr": 6.721545194061733e-05} +{"train_loss": 0.018723992630839348, "grad_norm": 0.07515238970518112, "global_step": 20570, "epoch": 194, "lr": 6.721266935916892e-05} +{"train_loss": 0.01589713618159294, "grad_norm": 0.062052227556705475, "global_step": 20571, "epoch": 194, "lr": 6.720988671724267e-05} +{"train_loss": 0.01442517526447773, "grad_norm": 0.16917479038238525, "global_step": 20572, "epoch": 194, "lr": 6.720710401484834e-05} +{"train_loss": 0.016902964562177658, "grad_norm": 0.06818723678588867, "global_step": 20573, "epoch": 194, "lr": 6.720432125199575e-05} +{"train_loss": 0.01559870783239603, "grad_norm": 0.1162947416305542, "global_step": 20574, "epoch": 194, "lr": 6.720153842869465e-05} +{"train_loss": 0.014895521104335785, "grad_norm": 0.11940912157297134, "global_step": 20575, "epoch": 194, "lr": 6.719875554495482e-05} +{"train_loss": 0.015554975718259811, "grad_norm": 0.06598075479269028, "global_step": 20576, "epoch": 194, "lr": 6.719597260078608e-05} +{"train_loss": 0.016537532210350037, "grad_norm": 0.061479050666093826, "global_step": 20577, "epoch": 194, "lr": 6.719318959619812e-05} +{"train_loss": 0.014864286407828331, "grad_norm": 0.06706493347883224, "global_step": 20578, "epoch": 194, "lr": 6.719040653120079e-05} +{"train_loss": 0.019371870905160904, "grad_norm": 0.07390616834163666, "global_step": 20579, "epoch": 194, "lr": 6.718762340580386e-05} +{"train_loss": 0.015592051669955254, "grad_norm": 0.07803995162248611, "global_step": 20580, "epoch": 194, "lr": 6.718484022001707e-05} +{"train_loss": 0.0168367438018322, "grad_norm": 0.06579761207103729, "global_step": 20581, "epoch": 194, "lr": 6.718205697385023e-05} +{"train_loss": 0.016765262931585312, "grad_norm": 0.07445737719535828, "global_step": 20582, "epoch": 194, "lr": 6.717927366731314e-05} +{"train_loss": 0.016698207706212997, "grad_norm": 0.0820925310254097, "global_step": 20583, "epoch": 194, "lr": 6.717649030041552e-05} +{"train_loss": 0.01585027202963829, "grad_norm": 0.05983508750796318, "global_step": 20584, "epoch": 194, "lr": 6.717370687316721e-05} +{"train_loss": 0.017433952540159225, "grad_norm": 0.05967006832361221, "global_step": 20585, "epoch": 194, "lr": 6.717092338557794e-05} +{"train_loss": 0.017003212124109268, "grad_norm": 0.09262320399284363, "global_step": 20586, "epoch": 194, "lr": 6.716813983765753e-05} +{"train_loss": 0.016376139596104622, "grad_norm": 0.11174791306257248, "global_step": 20587, "epoch": 194, "lr": 6.716535622941573e-05} +{"train_loss": 0.014814003370702267, "grad_norm": 0.06607231497764587, "global_step": 20588, "epoch": 194, "lr": 6.716257256086235e-05} +{"train_loss": 0.016305560246109962, "grad_norm": 0.09706393629312515, "global_step": 20589, "epoch": 194, "lr": 6.715978883200715e-05} +{"train_loss": 0.018208304420113564, "grad_norm": 0.06744960695505142, "global_step": 20590, "epoch": 194, "lr": 6.715700504285988e-05} +{"train_loss": 0.013675063848495483, "grad_norm": 0.06971942633390427, "global_step": 20591, "epoch": 194, "lr": 6.715422119343041e-05} +{"train_loss": 0.015199746936559677, "grad_norm": 0.08368705958127975, "global_step": 20592, "epoch": 194, "lr": 6.715143728372844e-05} +{"train_loss": 0.01594109833240509, "grad_norm": 0.07149416953325272, "global_step": 20593, "epoch": 194, "lr": 6.714865331376377e-05} +{"train_loss": 0.0154076237231493, "grad_norm": 0.06014375388622284, "global_step": 20594, "epoch": 194, "lr": 6.714586928354621e-05} +{"train_loss": 0.015197808854281902, "grad_norm": 0.1077808290719986, "global_step": 20595, "epoch": 194, "lr": 6.71430851930855e-05} +{"train_loss": 0.01727212406694889, "grad_norm": 0.053161416202783585, "global_step": 20596, "epoch": 194, "lr": 6.714030104239147e-05} +{"train_loss": 0.018917283043265343, "grad_norm": 0.1242104321718216, "global_step": 20597, "epoch": 194, "lr": 6.713751683147385e-05} +{"train_loss": 0.015678193420171738, "grad_norm": 0.057576000690460205, "global_step": 20598, "epoch": 194, "lr": 6.713473256034246e-05} +{"train_loss": 0.016956262290477753, "grad_norm": 0.08265666663646698, "global_step": 20599, "epoch": 194, "lr": 6.713194822900706e-05} +{"train_loss": 0.016557777300477028, "grad_norm": 0.09136410802602768, "global_step": 20600, "epoch": 194, "lr": 6.712916383747747e-05} +{"train_loss": 0.016379663720726967, "grad_norm": 0.059867702424526215, "global_step": 20601, "epoch": 194, "lr": 6.712637938576343e-05} +{"train_loss": 0.01638193055987358, "grad_norm": 0.11326576769351959, "global_step": 20602, "epoch": 194, "lr": 6.712359487387474e-05} +{"train_loss": 0.017116200178861618, "grad_norm": 0.10416177660226822, "global_step": 20603, "epoch": 194, "lr": 6.71208103018212e-05} +{"train_loss": 0.01488710381090641, "grad_norm": 0.06228620931506157, "global_step": 20604, "epoch": 194, "lr": 6.711802566961256e-05} +{"train_loss": 0.017166730016469955, "grad_norm": 0.08908311277627945, "global_step": 20605, "epoch": 194, "lr": 6.711524097725863e-05} +{"train_loss": 0.014715082943439484, "grad_norm": 0.05708640441298485, "global_step": 20606, "epoch": 194, "lr": 6.711245622476919e-05} +{"train_loss": 0.015523727983236313, "grad_norm": 0.08684361726045609, "global_step": 20607, "epoch": 194, "lr": 6.710967141215401e-05} +{"train_loss": 0.017327073961496353, "grad_norm": 0.05240601301193237, "global_step": 20608, "epoch": 194, "lr": 6.710688653942289e-05} +{"train_loss": 0.015279274433851242, "grad_norm": 0.07326704263687134, "global_step": 20609, "epoch": 194, "lr": 6.710410160658561e-05} +{"train_loss": 0.014643508940935135, "grad_norm": 0.054897550493478775, "global_step": 20610, "epoch": 194, "lr": 6.710131661365197e-05} +{"train_loss": 0.01591741107404232, "grad_norm": 0.07949820160865784, "global_step": 20611, "epoch": 194, "lr": 6.709853156063173e-05} +{"train_loss": 0.016023172065615654, "grad_norm": 0.13243074715137482, "global_step": 20612, "epoch": 194, "lr": 6.709574644753468e-05} +{"train_loss": 0.015572626143693924, "grad_norm": 0.0701199397444725, "global_step": 20613, "epoch": 194, "lr": 6.709296127437061e-05} +{"train_loss": 0.01604740507900715, "grad_norm": 0.0804671049118042, "global_step": 20614, "epoch": 194, "lr": 6.709017604114932e-05} +{"train_loss": 0.017250053584575653, "grad_norm": 0.17194950580596924, "global_step": 20615, "epoch": 194, "lr": 6.708739074788057e-05} +{"train_loss": 0.014174375683069229, "grad_norm": 0.06456228345632553, "global_step": 20616, "epoch": 194, "lr": 6.708460539457418e-05} +{"train_loss": 0.014311601407825947, "grad_norm": 0.08621355146169662, "global_step": 20617, "epoch": 194, "lr": 6.708181998123991e-05} +{"train_loss": 0.01605832204222679, "grad_norm": 0.05417374148964882, "global_step": 20618, "epoch": 194, "lr": 6.707903450788754e-05} +{"train_loss": 0.018301045522093773, "grad_norm": 0.0713580995798111, "global_step": 20619, "epoch": 194, "lr": 6.707624897452687e-05} +{"train_loss": 0.014178968966007233, "grad_norm": 0.06366270780563354, "global_step": 20620, "epoch": 194, "lr": 6.70734633811677e-05} +{"train_loss": 0.015978295356035233, "grad_norm": 0.13245044648647308, "global_step": 20621, "epoch": 194, "lr": 6.707067772781978e-05} +{"train_loss": 0.01683734729886055, "grad_norm": 0.11735183745622635, "global_step": 20622, "epoch": 194, "lr": 6.706789201449295e-05} +{"train_loss": 0.0138436583802104, "grad_norm": 0.08061753958463669, "global_step": 20623, "epoch": 194, "lr": 6.706510624119697e-05} +{"train_loss": 0.014294403605163097, "grad_norm": 0.07183291763067245, "global_step": 20624, "epoch": 194, "lr": 6.706232040794161e-05} +{"train_loss": 0.017138652503490448, "grad_norm": 0.08588941395282745, "global_step": 20625, "epoch": 194, "lr": 6.70595345147367e-05} +{"train_loss": 0.016503429040312767, "grad_norm": 0.12783662974834442, "global_step": 20626, "epoch": 194, "lr": 6.705674856159199e-05} +{"train_loss": 0.01640985906124115, "grad_norm": 0.06268487870693207, "global_step": 20627, "epoch": 194, "lr": 6.705396254851727e-05} +{"train_loss": 0.01690712571144104, "grad_norm": 0.19305570423603058, "global_step": 20628, "epoch": 194, "lr": 6.705117647552237e-05} +{"train_loss": 0.016459548845887184, "grad_norm": 0.06511060893535614, "global_step": 20629, "epoch": 194, "lr": 6.704839034261703e-05} +{"train_loss": 0.01254476048052311, "grad_norm": 0.07969283312559128, "global_step": 20630, "epoch": 194, "lr": 6.704560414981106e-05} +{"train_loss": 0.015339619480073452, "grad_norm": 0.12175125628709793, "global_step": 20631, "epoch": 194, "lr": 6.704281789711427e-05} +{"train_loss": 0.016197316348552704, "grad_norm": 0.08026934415102005, "global_step": 20632, "epoch": 194, "lr": 6.704003158453643e-05} +{"train_loss": 0.015430204570293427, "grad_norm": 0.10000510513782501, "global_step": 20633, "epoch": 194, "lr": 6.703724521208729e-05} +{"train_loss": 0.01726686581969261, "grad_norm": 0.08846414089202881, "global_step": 20634, "epoch": 194, "lr": 6.703445877977672e-05} +{"train_loss": 0.014515603892505169, "grad_norm": 0.05827120319008827, "global_step": 20635, "epoch": 194, "lr": 6.703167228761447e-05} +{"train_loss": 0.016474485397338867, "grad_norm": 0.05644087493419647, "global_step": 20636, "epoch": 194, "lr": 6.70288857356103e-05} +{"train_loss": 0.014761170372366905, "grad_norm": 0.14300718903541565, "global_step": 20637, "epoch": 194, "lr": 6.702609912377405e-05} +{"train_loss": 0.014798139221966267, "grad_norm": 0.11415623128414154, "global_step": 20638, "epoch": 194, "lr": 6.702331245211549e-05} +{"train_loss": 0.016648489981889725, "grad_norm": 0.07920345664024353, "global_step": 20639, "epoch": 194, "lr": 6.702052572064442e-05} +{"train_loss": 0.018958047032356262, "grad_norm": 0.07272698730230331, "global_step": 20640, "epoch": 194, "lr": 6.701773892937063e-05} +{"train_loss": 0.01621711254119873, "grad_norm": 0.07310021668672562, "global_step": 20641, "epoch": 194, "lr": 6.701495207830388e-05} +{"train_loss": 0.01749478653073311, "grad_norm": 0.06604554504156113, "global_step": 20642, "epoch": 194, "lr": 6.701216516745402e-05} +{"train_loss": 0.016398735344409943, "grad_norm": 0.08794109523296356, "global_step": 20643, "epoch": 194, "lr": 6.700937819683078e-05} +{"train_loss": 0.01649577170610428, "grad_norm": 0.06642377376556396, "global_step": 20644, "epoch": 194, "lr": 6.7006591166444e-05} +{"train_loss": 0.017948688939213753, "grad_norm": 0.06864650547504425, "global_step": 20645, "epoch": 194, "lr": 6.700380407630345e-05} +{"train_loss": 0.014512691646814346, "grad_norm": 0.09379845857620239, "global_step": 20646, "epoch": 194, "lr": 6.700101692641892e-05} +{"train_loss": 0.012936171144247055, "grad_norm": 0.06106732040643692, "global_step": 20647, "epoch": 194, "lr": 6.699822971680023e-05} +{"train_loss": 0.014443090185523033, "grad_norm": 0.06618540734052658, "global_step": 20648, "epoch": 194, "lr": 6.699544244745712e-05} +{"train_loss": 0.014036723412573338, "grad_norm": 0.08090423792600632, "global_step": 20649, "epoch": 194, "lr": 6.699265511839944e-05} +{"train_loss": 0.015718698501586914, "grad_norm": 0.07087281346321106, "global_step": 20650, "epoch": 194, "lr": 6.698986772963696e-05} +{"train_loss": 0.016191799193620682, "grad_norm": 0.07880721241235733, "global_step": 20651, "epoch": 194, "lr": 6.698708028117946e-05} +{"train_loss": 0.017561323940753937, "grad_norm": 0.05758490785956383, "global_step": 20652, "epoch": 194, "lr": 6.698429277303677e-05} +{"train_loss": 0.016063041985034943, "grad_norm": 0.12983636558055878, "global_step": 20653, "epoch": 194, "lr": 6.698150520521864e-05} +{"train_loss": 0.016555149108171463, "grad_norm": 0.06769882887601852, "global_step": 20654, "epoch": 194, "lr": 6.697871757773491e-05} +{"train_loss": 0.01599142700433731, "grad_norm": 0.06335501372814178, "global_step": 20655, "epoch": 194, "lr": 6.697592989059531e-05} +{"train_loss": 0.017518918961286545, "grad_norm": 0.20929200947284698, "global_step": 20656, "epoch": 194, "lr": 6.697314214380971e-05} +{"train_loss": 0.013290701434016228, "grad_norm": 0.05142287537455559, "global_step": 20657, "epoch": 194, "lr": 6.697035433738786e-05} +{"train_loss": 0.01328151486814022, "grad_norm": 0.17541606724262238, "global_step": 20658, "epoch": 194, "lr": 6.696756647133957e-05} +{"train_loss": 0.01662359945476055, "grad_norm": 0.09357704222202301, "global_step": 20659, "epoch": 194, "lr": 6.696477854567462e-05} +{"train_loss": 0.014079299755394459, "grad_norm": 0.0513654462993145, "global_step": 20660, "epoch": 194, "lr": 6.696199056040283e-05} +{"train_loss": 0.017451290041208267, "grad_norm": 0.09831640124320984, "global_step": 20661, "epoch": 194, "lr": 6.695920251553399e-05} +{"train_loss": 0.017821408808231354, "grad_norm": 0.05395011976361275, "global_step": 20662, "epoch": 194, "lr": 6.695641441107786e-05} +{"train_loss": 0.017316434532403946, "grad_norm": 0.07915431261062622, "global_step": 20663, "epoch": 194, "lr": 6.695362624704426e-05} +{"train_loss": 0.016258439049124718, "grad_norm": 0.05804353952407837, "global_step": 20664, "epoch": 194, "lr": 6.695083802344301e-05} +{"train_loss": 0.01719476282596588, "grad_norm": 0.09182300418615341, "global_step": 20665, "epoch": 194, "lr": 6.694804974028388e-05} +{"train_loss": 0.0167111624032259, "grad_norm": 0.06476632505655289, "global_step": 20666, "epoch": 194, "lr": 6.694526139757667e-05} +{"train_loss": 0.015815647318959236, "grad_norm": 0.07622049748897552, "global_step": 20667, "epoch": 194, "lr": 6.694247299533119e-05} +{"train_loss": 0.015841970220208168, "grad_norm": 0.118852898478508, "global_step": 20668, "epoch": 194, "lr": 6.693968453355723e-05} +{"train_loss": 0.0160362444695015, "grad_norm": 0.0687745213508606, "global_step": 20669, "epoch": 194, "lr": 6.693689601226458e-05, "val_loss": 0.019548077136278152} +{"train_loss": 0.017142683267593384, "grad_norm": 0.16978690028190613, "global_step": 20670, "epoch": 195, "lr": 6.693410743146304e-05} +{"train_loss": 0.01582888700067997, "grad_norm": 0.15139524638652802, "global_step": 20671, "epoch": 195, "lr": 6.693131879116241e-05} +{"train_loss": 0.01637406274676323, "grad_norm": 0.1116630882024765, "global_step": 20672, "epoch": 195, "lr": 6.692853009137248e-05} +{"train_loss": 0.01506071537733078, "grad_norm": 0.0769069492816925, "global_step": 20673, "epoch": 195, "lr": 6.692574133210309e-05} +{"train_loss": 0.013708766549825668, "grad_norm": 0.08488696068525314, "global_step": 20674, "epoch": 195, "lr": 6.692295251336398e-05} +{"train_loss": 0.01871970109641552, "grad_norm": 0.10498607903718948, "global_step": 20675, "epoch": 195, "lr": 6.692016363516498e-05} +{"train_loss": 0.01558596920222044, "grad_norm": 0.09953334927558899, "global_step": 20676, "epoch": 195, "lr": 6.691737469751589e-05} +{"train_loss": 0.017507772892713547, "grad_norm": 0.09440379589796066, "global_step": 20677, "epoch": 195, "lr": 6.691458570042648e-05} +{"train_loss": 0.016307462006807327, "grad_norm": 0.05456031486392021, "global_step": 20678, "epoch": 195, "lr": 6.691179664390659e-05} +{"train_loss": 0.014862545765936375, "grad_norm": 0.05053120106458664, "global_step": 20679, "epoch": 195, "lr": 6.690900752796601e-05} +{"train_loss": 0.01396570261567831, "grad_norm": 0.13282041251659393, "global_step": 20680, "epoch": 195, "lr": 6.690621835261451e-05} +{"train_loss": 0.014704007655382156, "grad_norm": 0.0781433954834938, "global_step": 20681, "epoch": 195, "lr": 6.690342911786192e-05} +{"train_loss": 0.015458762645721436, "grad_norm": 0.18341165781021118, "global_step": 20682, "epoch": 195, "lr": 6.690063982371804e-05} +{"train_loss": 0.012928856536746025, "grad_norm": 0.08649476617574692, "global_step": 20683, "epoch": 195, "lr": 6.689785047019266e-05} +{"train_loss": 0.015642158687114716, "grad_norm": 0.1417422592639923, "global_step": 20684, "epoch": 195, "lr": 6.689506105729556e-05} +{"train_loss": 0.01579214259982109, "grad_norm": 0.07194344699382782, "global_step": 20685, "epoch": 195, "lr": 6.689227158503659e-05} +{"train_loss": 0.013458121567964554, "grad_norm": 0.08030594140291214, "global_step": 20686, "epoch": 195, "lr": 6.688948205342551e-05} +{"train_loss": 0.016780579462647438, "grad_norm": 0.07107453793287277, "global_step": 20687, "epoch": 195, "lr": 6.688669246247214e-05} +{"train_loss": 0.015318870544433594, "grad_norm": 0.08579955995082855, "global_step": 20688, "epoch": 195, "lr": 6.688390281218629e-05} +{"train_loss": 0.017812589183449745, "grad_norm": 0.06819132715463638, "global_step": 20689, "epoch": 195, "lr": 6.68811131025777e-05} +{"train_loss": 0.016868956387043, "grad_norm": 0.06359848380088806, "global_step": 20690, "epoch": 195, "lr": 6.687832333365627e-05} +{"train_loss": 0.014468757435679436, "grad_norm": 0.10920458287000656, "global_step": 20691, "epoch": 195, "lr": 6.687553350543173e-05} +{"train_loss": 0.015298726968467236, "grad_norm": 0.06692865490913391, "global_step": 20692, "epoch": 195, "lr": 6.687274361791391e-05} +{"train_loss": 0.016911182552576065, "grad_norm": 0.08367744088172913, "global_step": 20693, "epoch": 195, "lr": 6.68699536711126e-05} +{"train_loss": 0.016228150576353073, "grad_norm": 0.11805057525634766, "global_step": 20694, "epoch": 195, "lr": 6.686716366503763e-05} +{"train_loss": 0.01426880992949009, "grad_norm": 0.10972616821527481, "global_step": 20695, "epoch": 195, "lr": 6.686437359969876e-05} +{"train_loss": 0.015384451486170292, "grad_norm": 0.04995176941156387, "global_step": 20696, "epoch": 195, "lr": 6.686158347510582e-05} +{"train_loss": 0.015691684558987617, "grad_norm": 0.1726275533437729, "global_step": 20697, "epoch": 195, "lr": 6.685879329126861e-05} +{"train_loss": 0.01231057196855545, "grad_norm": 0.05415583401918411, "global_step": 20698, "epoch": 195, "lr": 6.685600304819693e-05} +{"train_loss": 0.014914195984601974, "grad_norm": 0.08143548667430878, "global_step": 20699, "epoch": 195, "lr": 6.685321274590059e-05} +{"train_loss": 0.01413386408239603, "grad_norm": 0.05591834336519241, "global_step": 20700, "epoch": 195, "lr": 6.68504223843894e-05} +{"train_loss": 0.016906563192605972, "grad_norm": 0.05476698279380798, "global_step": 20701, "epoch": 195, "lr": 6.684763196367313e-05} +{"train_loss": 0.01740531250834465, "grad_norm": 0.1002078726887703, "global_step": 20702, "epoch": 195, "lr": 6.684484148376162e-05} +{"train_loss": 0.015703478828072548, "grad_norm": 0.0664219930768013, "global_step": 20703, "epoch": 195, "lr": 6.684205094466467e-05} +{"train_loss": 0.014448530040681362, "grad_norm": 0.09989647567272186, "global_step": 20704, "epoch": 195, "lr": 6.683926034639206e-05} +{"train_loss": 0.01667330414056778, "grad_norm": 0.06992867588996887, "global_step": 20705, "epoch": 195, "lr": 6.683646968895363e-05} +{"train_loss": 0.01660093665122986, "grad_norm": 0.05789885297417641, "global_step": 20706, "epoch": 195, "lr": 6.683367897235916e-05} +{"train_loss": 0.015479180961847305, "grad_norm": 0.058508820831775665, "global_step": 20707, "epoch": 195, "lr": 6.683088819661846e-05} +{"train_loss": 0.017050597816705704, "grad_norm": 0.13576464354991913, "global_step": 20708, "epoch": 195, "lr": 6.682809736174135e-05} +{"train_loss": 0.01630595326423645, "grad_norm": 0.1142331138253212, "global_step": 20709, "epoch": 195, "lr": 6.682530646773762e-05} +{"train_loss": 0.015665249899029732, "grad_norm": 0.09742129594087601, "global_step": 20710, "epoch": 195, "lr": 6.682251551461707e-05} +{"train_loss": 0.014454702846705914, "grad_norm": 0.07702821493148804, "global_step": 20711, "epoch": 195, "lr": 6.681972450238953e-05} +{"train_loss": 0.018914226442575455, "grad_norm": 0.10359830409288406, "global_step": 20712, "epoch": 195, "lr": 6.681693343106477e-05} +{"train_loss": 0.018732964992523193, "grad_norm": 0.08917275071144104, "global_step": 20713, "epoch": 195, "lr": 6.681414230065264e-05} +{"train_loss": 0.014972289092838764, "grad_norm": 0.0513245053589344, "global_step": 20714, "epoch": 195, "lr": 6.681135111116292e-05} +{"train_loss": 0.014355640858411789, "grad_norm": 0.050741881132125854, "global_step": 20715, "epoch": 195, "lr": 6.680855986260544e-05} +{"train_loss": 0.016519831493496895, "grad_norm": 0.1431603580713272, "global_step": 20716, "epoch": 195, "lr": 6.680576855498996e-05} +{"train_loss": 0.017799116671085358, "grad_norm": 0.08026804774999619, "global_step": 20717, "epoch": 195, "lr": 6.680297718832634e-05} +{"train_loss": 0.014768430963158607, "grad_norm": 0.0754459947347641, "global_step": 20718, "epoch": 195, "lr": 6.680018576262434e-05} +{"train_loss": 0.01630086451768875, "grad_norm": 0.05287376418709755, "global_step": 20719, "epoch": 195, "lr": 6.679739427789383e-05} +{"train_loss": 0.016376392915844917, "grad_norm": 0.08190371096134186, "global_step": 20720, "epoch": 195, "lr": 6.679460273414455e-05} +{"train_loss": 0.01634792983531952, "grad_norm": 0.10489574819803238, "global_step": 20721, "epoch": 195, "lr": 6.679181113138635e-05} +{"train_loss": 0.018247101455926895, "grad_norm": 0.06958014518022537, "global_step": 20722, "epoch": 195, "lr": 6.678901946962903e-05} +{"train_loss": 0.016884520649909973, "grad_norm": 0.07100752741098404, "global_step": 20723, "epoch": 195, "lr": 6.678622774888241e-05} +{"train_loss": 0.017290815711021423, "grad_norm": 0.061533279716968536, "global_step": 20724, "epoch": 195, "lr": 6.678343596915627e-05} +{"train_loss": 0.01365235261619091, "grad_norm": 0.07515501976013184, "global_step": 20725, "epoch": 195, "lr": 6.678064413046043e-05} +{"train_loss": 0.016649749130010605, "grad_norm": 0.08407701551914215, "global_step": 20726, "epoch": 195, "lr": 6.677785223280473e-05} +{"train_loss": 0.01572948880493641, "grad_norm": 0.09268585592508316, "global_step": 20727, "epoch": 195, "lr": 6.677506027619893e-05} +{"train_loss": 0.014966887421905994, "grad_norm": 0.08720120042562485, "global_step": 20728, "epoch": 195, "lr": 6.677226826065286e-05} +{"train_loss": 0.016839340329170227, "grad_norm": 0.08831176161766052, "global_step": 20729, "epoch": 195, "lr": 6.676947618617636e-05} +{"train_loss": 0.018168054521083832, "grad_norm": 0.07568345218896866, "global_step": 20730, "epoch": 195, "lr": 6.676668405277919e-05} +{"train_loss": 0.0162874273955822, "grad_norm": 0.0929374024271965, "global_step": 20731, "epoch": 195, "lr": 6.676389186047119e-05} +{"train_loss": 0.017634164541959763, "grad_norm": 0.1355287879705429, "global_step": 20732, "epoch": 195, "lr": 6.676109960926217e-05} +{"train_loss": 0.015745211392641068, "grad_norm": 0.08332559466362, "global_step": 20733, "epoch": 195, "lr": 6.675830729916192e-05} +{"train_loss": 0.017249658703804016, "grad_norm": 0.06406576186418533, "global_step": 20734, "epoch": 195, "lr": 6.675551493018029e-05} +{"train_loss": 0.014756896533071995, "grad_norm": 0.06964913010597229, "global_step": 20735, "epoch": 195, "lr": 6.675272250232705e-05} +{"train_loss": 0.016187984496355057, "grad_norm": 0.0562400221824646, "global_step": 20736, "epoch": 195, "lr": 6.674993001561203e-05} +{"train_loss": 0.015371372923254967, "grad_norm": 0.07294279336929321, "global_step": 20737, "epoch": 195, "lr": 6.674713747004503e-05} +{"train_loss": 0.014502329751849174, "grad_norm": 0.10573569685220718, "global_step": 20738, "epoch": 195, "lr": 6.67443448656359e-05} +{"train_loss": 0.01860564574599266, "grad_norm": 0.07159052789211273, "global_step": 20739, "epoch": 195, "lr": 6.674155220239441e-05} +{"train_loss": 0.014298937283456326, "grad_norm": 0.12415273487567902, "global_step": 20740, "epoch": 195, "lr": 6.673875948033038e-05} +{"train_loss": 0.01690981537103653, "grad_norm": 0.06446350365877151, "global_step": 20741, "epoch": 195, "lr": 6.673596669945364e-05} +{"train_loss": 0.01680753380060196, "grad_norm": 0.07526630163192749, "global_step": 20742, "epoch": 195, "lr": 6.673317385977398e-05} +{"train_loss": 0.017368443310260773, "grad_norm": 0.08583925664424896, "global_step": 20743, "epoch": 195, "lr": 6.673038096130124e-05} +{"train_loss": 0.0185982845723629, "grad_norm": 0.15309679508209229, "global_step": 20744, "epoch": 195, "lr": 6.672758800404521e-05} +{"train_loss": 0.020190495997667313, "grad_norm": 0.07851628214120865, "global_step": 20745, "epoch": 195, "lr": 6.672479498801569e-05} +{"train_loss": 0.013256710022687912, "grad_norm": 0.0639178454875946, "global_step": 20746, "epoch": 195, "lr": 6.672200191322253e-05} +{"train_loss": 0.016143623739480972, "grad_norm": 0.05530250817537308, "global_step": 20747, "epoch": 195, "lr": 6.671920877967554e-05} +{"train_loss": 0.017752118408679962, "grad_norm": 0.08995069563388824, "global_step": 20748, "epoch": 195, "lr": 6.671641558738451e-05} +{"train_loss": 0.016297392547130585, "grad_norm": 0.0587581992149353, "global_step": 20749, "epoch": 195, "lr": 6.671362233635925e-05} +{"train_loss": 0.015389221720397472, "grad_norm": 0.10200778394937515, "global_step": 20750, "epoch": 195, "lr": 6.671082902660961e-05} +{"train_loss": 0.01638169214129448, "grad_norm": 0.16251391172409058, "global_step": 20751, "epoch": 195, "lr": 6.670803565814538e-05} +{"train_loss": 0.014382468536496162, "grad_norm": 0.09800483286380768, "global_step": 20752, "epoch": 195, "lr": 6.670524223097637e-05} +{"train_loss": 0.017182182520627975, "grad_norm": 0.16760312020778656, "global_step": 20753, "epoch": 195, "lr": 6.670244874511243e-05} +{"train_loss": 0.01778339222073555, "grad_norm": 0.05638888478279114, "global_step": 20754, "epoch": 195, "lr": 6.669965520056331e-05} +{"train_loss": 0.016849596053361893, "grad_norm": 0.14105503261089325, "global_step": 20755, "epoch": 195, "lr": 6.669686159733888e-05} +{"train_loss": 0.016191992908716202, "grad_norm": 0.0841042771935463, "global_step": 20756, "epoch": 195, "lr": 6.669406793544897e-05} +{"train_loss": 0.020350899547338486, "grad_norm": 0.07653150707483292, "global_step": 20757, "epoch": 195, "lr": 6.669127421490334e-05} +{"train_loss": 0.015226437710225582, "grad_norm": 0.0708027184009552, "global_step": 20758, "epoch": 195, "lr": 6.668848043571183e-05} +{"train_loss": 0.014778544194996357, "grad_norm": 0.0779673233628273, "global_step": 20759, "epoch": 195, "lr": 6.668568659788425e-05} +{"train_loss": 0.016911417245864868, "grad_norm": 0.06446856260299683, "global_step": 20760, "epoch": 195, "lr": 6.668289270143043e-05} +{"train_loss": 0.014454749412834644, "grad_norm": 0.09364775568246841, "global_step": 20761, "epoch": 195, "lr": 6.668009874636019e-05} +{"train_loss": 0.01635325327515602, "grad_norm": 0.05779220163822174, "global_step": 20762, "epoch": 195, "lr": 6.667730473268333e-05} +{"train_loss": 0.0167763140052557, "grad_norm": 0.10640240460634232, "global_step": 20763, "epoch": 195, "lr": 6.667451066040966e-05} +{"train_loss": 0.014290060847997665, "grad_norm": 0.07947995513677597, "global_step": 20764, "epoch": 195, "lr": 6.667171652954903e-05} +{"train_loss": 0.0172104574739933, "grad_norm": 0.09415263682603836, "global_step": 20765, "epoch": 195, "lr": 6.666892234011123e-05} +{"train_loss": 0.017513949424028397, "grad_norm": 0.10078898817300797, "global_step": 20766, "epoch": 195, "lr": 6.66661280921061e-05} +{"train_loss": 0.01615491881966591, "grad_norm": 0.08162328600883484, "global_step": 20767, "epoch": 195, "lr": 6.666333378554343e-05} +{"train_loss": 0.0132821686565876, "grad_norm": 0.048233337700366974, "global_step": 20768, "epoch": 195, "lr": 6.666053942043306e-05} +{"train_loss": 0.015406640246510506, "grad_norm": 0.06565477699041367, "global_step": 20769, "epoch": 195, "lr": 6.665774499678477e-05} +{"train_loss": 0.01556266937404871, "grad_norm": 0.062098730355501175, "global_step": 20770, "epoch": 195, "lr": 6.665495051460844e-05} +{"train_loss": 0.015955673530697823, "grad_norm": 0.06503020226955414, "global_step": 20771, "epoch": 195, "lr": 6.665215597391385e-05} +{"train_loss": 0.018561940640211105, "grad_norm": 0.08033916354179382, "global_step": 20772, "epoch": 195, "lr": 6.664936137471083e-05} +{"train_loss": 0.016918115317821503, "grad_norm": 0.06938810646533966, "global_step": 20773, "epoch": 195, "lr": 6.66465667170092e-05} +{"train_loss": 0.014011609368026257, "grad_norm": 0.12902650237083435, "global_step": 20774, "epoch": 195, "lr": 6.664377200081875e-05} +{"train_loss": 0.016055110068816058, "grad_norm": 0.06436820328235626, "global_step": 20775, "epoch": 195, "lr": 6.664097722614934e-05, "val_loss": 0.0191080030053854} +{"train_loss": 0.016798196360468864, "grad_norm": 0.08692991733551025, "global_step": 20776, "epoch": 196, "lr": 6.663818239301076e-05} +{"train_loss": 0.015645071864128113, "grad_norm": 0.06746401637792587, "global_step": 20777, "epoch": 196, "lr": 6.663538750141286e-05} +{"train_loss": 0.016259610652923584, "grad_norm": 0.09927313029766083, "global_step": 20778, "epoch": 196, "lr": 6.663259255136543e-05} +{"train_loss": 0.013696624897420406, "grad_norm": 0.050710778683423996, "global_step": 20779, "epoch": 196, "lr": 6.662979754287831e-05} +{"train_loss": 0.013710797764360905, "grad_norm": 0.1472272425889969, "global_step": 20780, "epoch": 196, "lr": 6.662700247596131e-05} +{"train_loss": 0.014765889383852482, "grad_norm": 0.07383769005537033, "global_step": 20781, "epoch": 196, "lr": 6.662420735062425e-05} +{"train_loss": 0.015133639797568321, "grad_norm": 0.07246451824903488, "global_step": 20782, "epoch": 196, "lr": 6.662141216687699e-05} +{"train_loss": 0.016533635556697845, "grad_norm": 0.05267493426799774, "global_step": 20783, "epoch": 196, "lr": 6.661861692472927e-05} +{"train_loss": 0.01651209220290184, "grad_norm": 0.06944035738706589, "global_step": 20784, "epoch": 196, "lr": 6.661582162419098e-05} +{"train_loss": 0.013295261189341545, "grad_norm": 0.06254343688488007, "global_step": 20785, "epoch": 196, "lr": 6.661302626527192e-05} +{"train_loss": 0.016966113820672035, "grad_norm": 0.09936144202947617, "global_step": 20786, "epoch": 196, "lr": 6.661023084798189e-05} +{"train_loss": 0.01598844677209854, "grad_norm": 0.0890655592083931, "global_step": 20787, "epoch": 196, "lr": 6.660743537233075e-05} +{"train_loss": 0.015505681745707989, "grad_norm": 0.13182024657726288, "global_step": 20788, "epoch": 196, "lr": 6.66046398383283e-05} +{"train_loss": 0.0173058919608593, "grad_norm": 0.11736856400966644, "global_step": 20789, "epoch": 196, "lr": 6.660184424598437e-05} +{"train_loss": 0.015717627480626106, "grad_norm": 0.12188403308391571, "global_step": 20790, "epoch": 196, "lr": 6.659904859530878e-05} +{"train_loss": 0.014539161697030067, "grad_norm": 0.1170007660984993, "global_step": 20791, "epoch": 196, "lr": 6.659625288631135e-05} +{"train_loss": 0.015370991080999374, "grad_norm": 0.08631604164838791, "global_step": 20792, "epoch": 196, "lr": 6.659345711900192e-05} +{"train_loss": 0.016309645026922226, "grad_norm": 0.07457482069730759, "global_step": 20793, "epoch": 196, "lr": 6.659066129339026e-05} +{"train_loss": 0.016906501725316048, "grad_norm": 0.08733667433261871, "global_step": 20794, "epoch": 196, "lr": 6.658786540948626e-05} +{"train_loss": 0.013773631304502487, "grad_norm": 0.09217801690101624, "global_step": 20795, "epoch": 196, "lr": 6.658506946729971e-05} +{"train_loss": 0.015201158821582794, "grad_norm": 0.08011303097009659, "global_step": 20796, "epoch": 196, "lr": 6.658227346684043e-05} +{"train_loss": 0.017911642789840698, "grad_norm": 0.06128756329417229, "global_step": 20797, "epoch": 196, "lr": 6.657947740811827e-05} +{"train_loss": 0.015114616602659225, "grad_norm": 0.1047673150897026, "global_step": 20798, "epoch": 196, "lr": 6.657668129114302e-05} +{"train_loss": 0.01728120446205139, "grad_norm": 0.054614730179309845, "global_step": 20799, "epoch": 196, "lr": 6.657388511592452e-05} +{"train_loss": 0.01583884283900261, "grad_norm": 0.07359426468610764, "global_step": 20800, "epoch": 196, "lr": 6.657108888247261e-05} +{"train_loss": 0.016177263110876083, "grad_norm": 0.08795816451311111, "global_step": 20801, "epoch": 196, "lr": 6.65682925907971e-05} +{"train_loss": 0.017131119966506958, "grad_norm": 0.1491500288248062, "global_step": 20802, "epoch": 196, "lr": 6.656549624090779e-05} +{"train_loss": 0.017095297574996948, "grad_norm": 0.05563884600996971, "global_step": 20803, "epoch": 196, "lr": 6.656269983281456e-05} +{"train_loss": 0.01670040935277939, "grad_norm": 0.082398422062397, "global_step": 20804, "epoch": 196, "lr": 6.65599033665272e-05} +{"train_loss": 0.015263136476278305, "grad_norm": 0.0833800658583641, "global_step": 20805, "epoch": 196, "lr": 6.655710684205551e-05} +{"train_loss": 0.019543152302503586, "grad_norm": 0.1177842915058136, "global_step": 20806, "epoch": 196, "lr": 6.655431025940939e-05} +{"train_loss": 0.0172496996819973, "grad_norm": 0.09842613339424133, "global_step": 20807, "epoch": 196, "lr": 6.655151361859861e-05} +{"train_loss": 0.016864405944943428, "grad_norm": 0.08664785325527191, "global_step": 20808, "epoch": 196, "lr": 6.6548716919633e-05} +{"train_loss": 0.01637984812259674, "grad_norm": 0.09489990025758743, "global_step": 20809, "epoch": 196, "lr": 6.65459201625224e-05} +{"train_loss": 0.019909676164388657, "grad_norm": 0.11903820186853409, "global_step": 20810, "epoch": 196, "lr": 6.654312334727662e-05} +{"train_loss": 0.013911280781030655, "grad_norm": 0.0676935613155365, "global_step": 20811, "epoch": 196, "lr": 6.654032647390554e-05} +{"train_loss": 0.016599636524915695, "grad_norm": 0.07246695458889008, "global_step": 20812, "epoch": 196, "lr": 6.65375295424189e-05} +{"train_loss": 0.016436981037259102, "grad_norm": 0.06445387750864029, "global_step": 20813, "epoch": 196, "lr": 6.653473255282658e-05} +{"train_loss": 0.014938602223992348, "grad_norm": 0.0576297864317894, "global_step": 20814, "epoch": 196, "lr": 6.653193550513843e-05} +{"train_loss": 0.016739491373300552, "grad_norm": 0.11626686155796051, "global_step": 20815, "epoch": 196, "lr": 6.652913839936422e-05} +{"train_loss": 0.013349376618862152, "grad_norm": 0.0711154118180275, "global_step": 20816, "epoch": 196, "lr": 6.652634123551382e-05} +{"train_loss": 0.01442350447177887, "grad_norm": 0.10587240755558014, "global_step": 20817, "epoch": 196, "lr": 6.652354401359704e-05} +{"train_loss": 0.01737929880619049, "grad_norm": 0.09865991026163101, "global_step": 20818, "epoch": 196, "lr": 6.65207467336237e-05} +{"train_loss": 0.014084051363170147, "grad_norm": 0.07470224052667618, "global_step": 20819, "epoch": 196, "lr": 6.651794939560366e-05} +{"train_loss": 0.017467478290200233, "grad_norm": 0.0573626309633255, "global_step": 20820, "epoch": 196, "lr": 6.651515199954672e-05} +{"train_loss": 0.014985837042331696, "grad_norm": 0.057732608169317245, "global_step": 20821, "epoch": 196, "lr": 6.651235454546272e-05} +{"train_loss": 0.013987861573696136, "grad_norm": 0.07671311497688293, "global_step": 20822, "epoch": 196, "lr": 6.650955703336148e-05} +{"train_loss": 0.017485754564404488, "grad_norm": 0.07238456606864929, "global_step": 20823, "epoch": 196, "lr": 6.650675946325286e-05} +{"train_loss": 0.018717709928750992, "grad_norm": 0.1560886800289154, "global_step": 20824, "epoch": 196, "lr": 6.650396183514664e-05} +{"train_loss": 0.014867515303194523, "grad_norm": 0.11088203638792038, "global_step": 20825, "epoch": 196, "lr": 6.650116414905267e-05} +{"train_loss": 0.0152560630813241, "grad_norm": 0.1027662605047226, "global_step": 20826, "epoch": 196, "lr": 6.649836640498082e-05} +{"train_loss": 0.015328618697822094, "grad_norm": 0.10162774473428726, "global_step": 20827, "epoch": 196, "lr": 6.649556860294086e-05} +{"train_loss": 0.01648709550499916, "grad_norm": 0.05515016242861748, "global_step": 20828, "epoch": 196, "lr": 6.649277074294264e-05} +{"train_loss": 0.01587229035794735, "grad_norm": 0.06486000865697861, "global_step": 20829, "epoch": 196, "lr": 6.648997282499602e-05} +{"train_loss": 0.017771273851394653, "grad_norm": 0.10179295390844345, "global_step": 20830, "epoch": 196, "lr": 6.648717484911079e-05} +{"train_loss": 0.014102162793278694, "grad_norm": 0.07679900527000427, "global_step": 20831, "epoch": 196, "lr": 6.64843768152968e-05} +{"train_loss": 0.014849156141281128, "grad_norm": 0.16220442950725555, "global_step": 20832, "epoch": 196, "lr": 6.648157872356389e-05} +{"train_loss": 0.017401963472366333, "grad_norm": 0.0865938663482666, "global_step": 20833, "epoch": 196, "lr": 6.647878057392186e-05} +{"train_loss": 0.01708149164915085, "grad_norm": 0.12979912757873535, "global_step": 20834, "epoch": 196, "lr": 6.647598236638056e-05} +{"train_loss": 0.017912687733769417, "grad_norm": 0.0769977942109108, "global_step": 20835, "epoch": 196, "lr": 6.647318410094984e-05} +{"train_loss": 0.015628881752490997, "grad_norm": 0.08365622907876968, "global_step": 20836, "epoch": 196, "lr": 6.647038577763952e-05} +{"train_loss": 0.016590937972068787, "grad_norm": 0.07932699471712112, "global_step": 20837, "epoch": 196, "lr": 6.646758739645941e-05} +{"train_loss": 0.016870548948645592, "grad_norm": 0.12261662632226944, "global_step": 20838, "epoch": 196, "lr": 6.646478895741938e-05} +{"train_loss": 0.017692623659968376, "grad_norm": 0.11521265655755997, "global_step": 20839, "epoch": 196, "lr": 6.646199046052921e-05} +{"train_loss": 0.015688389539718628, "grad_norm": 0.05830013006925583, "global_step": 20840, "epoch": 196, "lr": 6.64591919057988e-05} +{"train_loss": 0.01586972177028656, "grad_norm": 0.10170595347881317, "global_step": 20841, "epoch": 196, "lr": 6.645639329323792e-05} +{"train_loss": 0.014793284237384796, "grad_norm": 0.08585122227668762, "global_step": 20842, "epoch": 196, "lr": 6.645359462285644e-05} +{"train_loss": 0.014911109581589699, "grad_norm": 0.1058616116642952, "global_step": 20843, "epoch": 196, "lr": 6.645079589466418e-05} +{"train_loss": 0.01582416705787182, "grad_norm": 0.10424390435218811, "global_step": 20844, "epoch": 196, "lr": 6.644799710867099e-05} +{"train_loss": 0.018435785546898842, "grad_norm": 0.0880209282040596, "global_step": 20845, "epoch": 196, "lr": 6.644519826488669e-05} +{"train_loss": 0.015762675553560257, "grad_norm": 0.06269232928752899, "global_step": 20846, "epoch": 196, "lr": 6.644239936332111e-05} +{"train_loss": 0.015870023518800735, "grad_norm": 0.09662098437547684, "global_step": 20847, "epoch": 196, "lr": 6.643960040398409e-05} +{"train_loss": 0.01358258631080389, "grad_norm": 0.07407810539007187, "global_step": 20848, "epoch": 196, "lr": 6.643680138688547e-05} +{"train_loss": 0.01644429937005043, "grad_norm": 0.07476560771465302, "global_step": 20849, "epoch": 196, "lr": 6.643400231203505e-05} +{"train_loss": 0.01901291310787201, "grad_norm": 0.0812612920999527, "global_step": 20850, "epoch": 196, "lr": 6.643120317944273e-05} +{"train_loss": 0.017544543370604515, "grad_norm": 0.08298009634017944, "global_step": 20851, "epoch": 196, "lr": 6.642840398911828e-05} +{"train_loss": 0.018695104867219925, "grad_norm": 0.09409671276807785, "global_step": 20852, "epoch": 196, "lr": 6.642560474107157e-05} +{"train_loss": 0.0166071318089962, "grad_norm": 0.1391746550798416, "global_step": 20853, "epoch": 196, "lr": 6.642280543531243e-05} +{"train_loss": 0.017308149486780167, "grad_norm": 0.10677672177553177, "global_step": 20854, "epoch": 196, "lr": 6.642000607185071e-05} +{"train_loss": 0.014345905743539333, "grad_norm": 0.10468702763319016, "global_step": 20855, "epoch": 196, "lr": 6.64172066506962e-05} +{"train_loss": 0.012502167373895645, "grad_norm": 0.0655047744512558, "global_step": 20856, "epoch": 196, "lr": 6.641440717185879e-05} +{"train_loss": 0.01779337227344513, "grad_norm": 0.15343596041202545, "global_step": 20857, "epoch": 196, "lr": 6.641160763534829e-05} +{"train_loss": 0.017006482928991318, "grad_norm": 0.07882913202047348, "global_step": 20858, "epoch": 196, "lr": 6.640880804117452e-05} +{"train_loss": 0.014499938115477562, "grad_norm": 0.05854278802871704, "global_step": 20859, "epoch": 196, "lr": 6.640600838934734e-05} +{"train_loss": 0.015025502070784569, "grad_norm": 0.09173537790775299, "global_step": 20860, "epoch": 196, "lr": 6.64032086798766e-05} +{"train_loss": 0.013866065070033073, "grad_norm": 0.06667964160442352, "global_step": 20861, "epoch": 196, "lr": 6.640040891277209e-05} +{"train_loss": 0.015864741057157516, "grad_norm": 0.09191873669624329, "global_step": 20862, "epoch": 196, "lr": 6.63976090880437e-05} +{"train_loss": 0.015739765018224716, "grad_norm": 0.06693750619888306, "global_step": 20863, "epoch": 196, "lr": 6.639480920570123e-05} +{"train_loss": 0.016159474849700928, "grad_norm": 0.08029001206159592, "global_step": 20864, "epoch": 196, "lr": 6.639200926575453e-05} +{"train_loss": 0.013857684098184109, "grad_norm": 0.06197284907102585, "global_step": 20865, "epoch": 196, "lr": 6.638920926821345e-05} +{"train_loss": 0.016508352011442184, "grad_norm": 0.08831518143415451, "global_step": 20866, "epoch": 196, "lr": 6.63864092130878e-05} +{"train_loss": 0.015376745723187923, "grad_norm": 0.08182887732982635, "global_step": 20867, "epoch": 196, "lr": 6.638360910038744e-05} +{"train_loss": 0.017951838672161102, "grad_norm": 0.05313487723469734, "global_step": 20868, "epoch": 196, "lr": 6.63808089301222e-05} +{"train_loss": 0.01359370443969965, "grad_norm": 0.06839549541473389, "global_step": 20869, "epoch": 196, "lr": 6.637800870230193e-05} +{"train_loss": 0.01372880395501852, "grad_norm": 0.07174067944288254, "global_step": 20870, "epoch": 196, "lr": 6.637520841693645e-05} +{"train_loss": 0.01588357985019684, "grad_norm": 0.08479734510183334, "global_step": 20871, "epoch": 196, "lr": 6.637240807403562e-05} +{"train_loss": 0.016437925398349762, "grad_norm": 0.08402108401060104, "global_step": 20872, "epoch": 196, "lr": 6.636960767360925e-05} +{"train_loss": 0.015626458451151848, "grad_norm": 0.052034102380275726, "global_step": 20873, "epoch": 196, "lr": 6.636680721566723e-05} +{"train_loss": 0.016505418345332146, "grad_norm": 0.07785630971193314, "global_step": 20874, "epoch": 196, "lr": 6.636400670021934e-05} +{"train_loss": 0.01920153573155403, "grad_norm": 0.07419466227293015, "global_step": 20875, "epoch": 196, "lr": 6.636120612727546e-05} +{"train_loss": 0.015332169830799103, "grad_norm": 0.06327776610851288, "global_step": 20876, "epoch": 196, "lr": 6.635840549684542e-05} +{"train_loss": 0.017953643575310707, "grad_norm": 0.062358058989048004, "global_step": 20877, "epoch": 196, "lr": 6.635560480893903e-05} +{"train_loss": 0.014729734510183334, "grad_norm": 0.06935225427150726, "global_step": 20878, "epoch": 196, "lr": 6.635280406356618e-05} +{"train_loss": 0.01631186157464981, "grad_norm": 0.05056016147136688, "global_step": 20879, "epoch": 196, "lr": 6.635000326073669e-05} +{"train_loss": 0.014807745814323425, "grad_norm": 0.08942669630050659, "global_step": 20880, "epoch": 196, "lr": 6.634720240046039e-05} +{"train_loss": 0.0160319715527431, "grad_norm": 0.09071563929319382, "global_step": 20881, "epoch": 196, "lr": 6.634440148274713e-05, "val_loss": 0.019379569217562675} +{"train_loss": 0.016612183302640915, "grad_norm": 0.06471020728349686, "global_step": 20882, "epoch": 197, "lr": 6.634160050760676e-05} +{"train_loss": 0.014392032288014889, "grad_norm": 0.0659182146191597, "global_step": 20883, "epoch": 197, "lr": 6.63387994750491e-05} +{"train_loss": 0.017266862094402313, "grad_norm": 0.06814926117658615, "global_step": 20884, "epoch": 197, "lr": 6.6335998385084e-05} +{"train_loss": 0.014891436323523521, "grad_norm": 0.08829507976770401, "global_step": 20885, "epoch": 197, "lr": 6.633319723772133e-05} +{"train_loss": 0.01629202626645565, "grad_norm": 0.07750885933637619, "global_step": 20886, "epoch": 197, "lr": 6.63303960329709e-05} +{"train_loss": 0.01738070882856846, "grad_norm": 0.06345014274120331, "global_step": 20887, "epoch": 197, "lr": 6.632759477084255e-05} +{"train_loss": 0.012748372741043568, "grad_norm": 0.04587143659591675, "global_step": 20888, "epoch": 197, "lr": 6.632479345134614e-05} +{"train_loss": 0.016134239733219147, "grad_norm": 0.07630528509616852, "global_step": 20889, "epoch": 197, "lr": 6.63219920744915e-05} +{"train_loss": 0.015087906271219254, "grad_norm": 0.05942599102854729, "global_step": 20890, "epoch": 197, "lr": 6.631919064028848e-05} +{"train_loss": 0.01690659672021866, "grad_norm": 0.056374792009592056, "global_step": 20891, "epoch": 197, "lr": 6.631638914874692e-05} +{"train_loss": 0.015332041308283806, "grad_norm": 0.059448547661304474, "global_step": 20892, "epoch": 197, "lr": 6.631358759987667e-05} +{"train_loss": 0.014170216396450996, "grad_norm": 0.05513264238834381, "global_step": 20893, "epoch": 197, "lr": 6.631078599368757e-05} +{"train_loss": 0.015220467001199722, "grad_norm": 0.06758752465248108, "global_step": 20894, "epoch": 197, "lr": 6.630798433018946e-05} +{"train_loss": 0.01631353795528412, "grad_norm": 0.05905097723007202, "global_step": 20895, "epoch": 197, "lr": 6.630518260939216e-05} +{"train_loss": 0.017873505130410194, "grad_norm": 0.06355689465999603, "global_step": 20896, "epoch": 197, "lr": 6.630238083130558e-05} +{"train_loss": 0.013807257637381554, "grad_norm": 0.05450735613703728, "global_step": 20897, "epoch": 197, "lr": 6.629957899593949e-05} +{"train_loss": 0.015724189579486847, "grad_norm": 0.07543981820344925, "global_step": 20898, "epoch": 197, "lr": 6.629677710330379e-05} +{"train_loss": 0.01524495892226696, "grad_norm": 0.06223800405859947, "global_step": 20899, "epoch": 197, "lr": 6.62939751534083e-05} +{"train_loss": 0.014755096286535263, "grad_norm": 0.12446633726358414, "global_step": 20900, "epoch": 197, "lr": 6.629117314626285e-05} +{"train_loss": 0.016285255551338196, "grad_norm": 0.058854084461927414, "global_step": 20901, "epoch": 197, "lr": 6.628837108187732e-05} +{"train_loss": 0.014715351164340973, "grad_norm": 0.0550888292491436, "global_step": 20902, "epoch": 197, "lr": 6.628556896026152e-05} +{"train_loss": 0.016515977680683136, "grad_norm": 0.12687039375305176, "global_step": 20903, "epoch": 197, "lr": 6.628276678142534e-05} +{"train_loss": 0.015184701420366764, "grad_norm": 0.08515510708093643, "global_step": 20904, "epoch": 197, "lr": 6.627996454537859e-05} +{"train_loss": 0.01696532964706421, "grad_norm": 0.08174976706504822, "global_step": 20905, "epoch": 197, "lr": 6.62771622521311e-05} +{"train_loss": 0.014797752723097801, "grad_norm": 0.10594161599874496, "global_step": 20906, "epoch": 197, "lr": 6.627435990169277e-05} +{"train_loss": 0.01737816259264946, "grad_norm": 0.0685046911239624, "global_step": 20907, "epoch": 197, "lr": 6.62715574940734e-05} +{"train_loss": 0.01901606097817421, "grad_norm": 0.12309093028306961, "global_step": 20908, "epoch": 197, "lr": 6.626875502928285e-05} +{"train_loss": 0.015504873357713223, "grad_norm": 0.043270546942949295, "global_step": 20909, "epoch": 197, "lr": 6.626595250733099e-05} +{"train_loss": 0.016219742596149445, "grad_norm": 0.10200153291225433, "global_step": 20910, "epoch": 197, "lr": 6.626314992822764e-05} +{"train_loss": 0.014435674995183945, "grad_norm": 0.14632108807563782, "global_step": 20911, "epoch": 197, "lr": 6.626034729198265e-05} +{"train_loss": 0.017647454515099525, "grad_norm": 0.06956859678030014, "global_step": 20912, "epoch": 197, "lr": 6.625754459860588e-05} +{"train_loss": 0.015434669330716133, "grad_norm": 0.1438082903623581, "global_step": 20913, "epoch": 197, "lr": 6.625474184810715e-05} +{"train_loss": 0.014508544467389584, "grad_norm": 0.08366988599300385, "global_step": 20914, "epoch": 197, "lr": 6.625193904049634e-05} +{"train_loss": 0.017461638897657394, "grad_norm": 0.06319918483495712, "global_step": 20915, "epoch": 197, "lr": 6.624913617578328e-05} +{"train_loss": 0.01627087965607643, "grad_norm": 0.06431204080581665, "global_step": 20916, "epoch": 197, "lr": 6.624633325397781e-05} +{"train_loss": 0.016111116856336594, "grad_norm": 0.061310868710279465, "global_step": 20917, "epoch": 197, "lr": 6.62435302750898e-05} +{"train_loss": 0.017305217683315277, "grad_norm": 0.05555944889783859, "global_step": 20918, "epoch": 197, "lr": 6.62407272391291e-05} +{"train_loss": 0.016900185495615005, "grad_norm": 0.09684955328702927, "global_step": 20919, "epoch": 197, "lr": 6.623792414610552e-05} +{"train_loss": 0.015611965209245682, "grad_norm": 0.1416427195072174, "global_step": 20920, "epoch": 197, "lr": 6.623512099602897e-05} +{"train_loss": 0.015829341486096382, "grad_norm": 0.19290420413017273, "global_step": 20921, "epoch": 197, "lr": 6.623231778890924e-05} +{"train_loss": 0.014697966165840626, "grad_norm": 0.09579221159219742, "global_step": 20922, "epoch": 197, "lr": 6.622951452475621e-05} +{"train_loss": 0.013525400310754776, "grad_norm": 0.12497318536043167, "global_step": 20923, "epoch": 197, "lr": 6.622671120357975e-05} +{"train_loss": 0.015786603093147278, "grad_norm": 0.1375366598367691, "global_step": 20924, "epoch": 197, "lr": 6.622390782538964e-05} +{"train_loss": 0.016551103442907333, "grad_norm": 0.1507832109928131, "global_step": 20925, "epoch": 197, "lr": 6.62211043901958e-05} +{"train_loss": 0.017812326550483704, "grad_norm": 0.06889768689870834, "global_step": 20926, "epoch": 197, "lr": 6.621830089800804e-05} +{"train_loss": 0.014497516676783562, "grad_norm": 0.06222685053944588, "global_step": 20927, "epoch": 197, "lr": 6.621549734883624e-05} +{"train_loss": 0.01881038025021553, "grad_norm": 0.08195746690034866, "global_step": 20928, "epoch": 197, "lr": 6.621269374269023e-05} +{"train_loss": 0.01501296553760767, "grad_norm": 0.09893128275871277, "global_step": 20929, "epoch": 197, "lr": 6.620989007957987e-05} +{"train_loss": 0.01740426942706108, "grad_norm": 0.08353684842586517, "global_step": 20930, "epoch": 197, "lr": 6.6207086359515e-05} +{"train_loss": 0.016834210604429245, "grad_norm": 0.08731986582279205, "global_step": 20931, "epoch": 197, "lr": 6.620428258250546e-05} +{"train_loss": 0.017269819974899292, "grad_norm": 0.062477707862854004, "global_step": 20932, "epoch": 197, "lr": 6.620147874856114e-05} +{"train_loss": 0.019593169912695885, "grad_norm": 0.1303967982530594, "global_step": 20933, "epoch": 197, "lr": 6.619867485769184e-05} +{"train_loss": 0.0161882434040308, "grad_norm": 0.0977821797132492, "global_step": 20934, "epoch": 197, "lr": 6.619587090990748e-05} +{"train_loss": 0.017009390518069267, "grad_norm": 0.10283684730529785, "global_step": 20935, "epoch": 197, "lr": 6.619306690521786e-05} +{"train_loss": 0.01634277030825615, "grad_norm": 0.08398342877626419, "global_step": 20936, "epoch": 197, "lr": 6.619026284363283e-05} +{"train_loss": 0.01492040604352951, "grad_norm": 0.056382980197668076, "global_step": 20937, "epoch": 197, "lr": 6.618745872516226e-05} +{"train_loss": 0.01970171555876732, "grad_norm": 0.0737595409154892, "global_step": 20938, "epoch": 197, "lr": 6.618465454981602e-05} +{"train_loss": 0.013922126032412052, "grad_norm": 0.06378095597028732, "global_step": 20939, "epoch": 197, "lr": 6.618185031760394e-05} +{"train_loss": 0.013546532019972801, "grad_norm": 0.062266070395708084, "global_step": 20940, "epoch": 197, "lr": 6.617904602853585e-05} +{"train_loss": 0.01687491312623024, "grad_norm": 0.09730079025030136, "global_step": 20941, "epoch": 197, "lr": 6.617624168262164e-05} +{"train_loss": 0.013603913597762585, "grad_norm": 0.055983006954193115, "global_step": 20942, "epoch": 197, "lr": 6.617343727987116e-05} +{"train_loss": 0.013177613727748394, "grad_norm": 0.07376119494438171, "global_step": 20943, "epoch": 197, "lr": 6.617063282029424e-05} +{"train_loss": 0.01599523052573204, "grad_norm": 0.08458247035741806, "global_step": 20944, "epoch": 197, "lr": 6.616782830390077e-05} +{"train_loss": 0.01474354974925518, "grad_norm": 0.056068748235702515, "global_step": 20945, "epoch": 197, "lr": 6.616502373070058e-05} +{"train_loss": 0.015473627485334873, "grad_norm": 0.10431777685880661, "global_step": 20946, "epoch": 197, "lr": 6.61622191007035e-05} +{"train_loss": 0.015141415409743786, "grad_norm": 0.10478939861059189, "global_step": 20947, "epoch": 197, "lr": 6.615941441391946e-05} +{"train_loss": 0.016352035105228424, "grad_norm": 0.04942898452281952, "global_step": 20948, "epoch": 197, "lr": 6.615660967035822e-05} +{"train_loss": 0.017381498590111732, "grad_norm": 0.08699819445610046, "global_step": 20949, "epoch": 197, "lr": 6.615380487002969e-05} +{"train_loss": 0.017204858362674713, "grad_norm": 0.06189354136586189, "global_step": 20950, "epoch": 197, "lr": 6.615100001294372e-05} +{"train_loss": 0.014915845356881618, "grad_norm": 0.09250633418560028, "global_step": 20951, "epoch": 197, "lr": 6.614819509911016e-05} +{"train_loss": 0.014412693679332733, "grad_norm": 0.10414185374975204, "global_step": 20952, "epoch": 197, "lr": 6.614539012853886e-05} +{"train_loss": 0.017820753157138824, "grad_norm": 0.15569831430912018, "global_step": 20953, "epoch": 197, "lr": 6.614258510123968e-05} +{"train_loss": 0.01580480858683586, "grad_norm": 0.07186871767044067, "global_step": 20954, "epoch": 197, "lr": 6.613978001722248e-05} +{"train_loss": 0.013049732893705368, "grad_norm": 0.11502253264188766, "global_step": 20955, "epoch": 197, "lr": 6.613697487649712e-05} +{"train_loss": 0.015060382895171642, "grad_norm": 0.06828362494707108, "global_step": 20956, "epoch": 197, "lr": 6.613416967907345e-05} +{"train_loss": 0.017001545056700706, "grad_norm": 0.0780630111694336, "global_step": 20957, "epoch": 197, "lr": 6.613136442496131e-05} +{"train_loss": 0.01671856828033924, "grad_norm": 0.1199423223733902, "global_step": 20958, "epoch": 197, "lr": 6.612855911417056e-05} +{"train_loss": 0.01650238037109375, "grad_norm": 0.10262947529554367, "global_step": 20959, "epoch": 197, "lr": 6.61257537467111e-05} +{"train_loss": 0.015187419950962067, "grad_norm": 0.13071958720684052, "global_step": 20960, "epoch": 197, "lr": 6.612294832259272e-05} +{"train_loss": 0.015468837693333626, "grad_norm": 0.12123832106590271, "global_step": 20961, "epoch": 197, "lr": 6.612014284182533e-05} +{"train_loss": 0.019520841538906097, "grad_norm": 0.08439338207244873, "global_step": 20962, "epoch": 197, "lr": 6.611733730441878e-05} +{"train_loss": 0.014502317644655704, "grad_norm": 0.0499948114156723, "global_step": 20963, "epoch": 197, "lr": 6.61145317103829e-05} +{"train_loss": 0.016520008444786072, "grad_norm": 0.12495120614767075, "global_step": 20964, "epoch": 197, "lr": 6.611172605972758e-05} +{"train_loss": 0.017350777983665466, "grad_norm": 0.08361589163541794, "global_step": 20965, "epoch": 197, "lr": 6.610892035246263e-05} +{"train_loss": 0.018336985260248184, "grad_norm": 0.09387120604515076, "global_step": 20966, "epoch": 197, "lr": 6.610611458859796e-05} +{"train_loss": 0.014948593452572823, "grad_norm": 0.11504095047712326, "global_step": 20967, "epoch": 197, "lr": 6.610330876814342e-05} +{"train_loss": 0.01904519833624363, "grad_norm": 0.09240544587373734, "global_step": 20968, "epoch": 197, "lr": 6.610050289110882e-05} +{"train_loss": 0.015121052041649818, "grad_norm": 0.12324775755405426, "global_step": 20969, "epoch": 197, "lr": 6.60976969575041e-05} +{"train_loss": 0.018754754215478897, "grad_norm": 0.12369013577699661, "global_step": 20970, "epoch": 197, "lr": 6.609489096733904e-05} +{"train_loss": 0.015623597428202629, "grad_norm": 0.049129970371723175, "global_step": 20971, "epoch": 197, "lr": 6.609208492062354e-05} +{"train_loss": 0.01909484714269638, "grad_norm": 0.10803695023059845, "global_step": 20972, "epoch": 197, "lr": 6.608927881736745e-05} +{"train_loss": 0.014501717872917652, "grad_norm": 0.06892549246549606, "global_step": 20973, "epoch": 197, "lr": 6.608647265758064e-05} +{"train_loss": 0.01592850498855114, "grad_norm": 0.058943234384059906, "global_step": 20974, "epoch": 197, "lr": 6.608366644127296e-05} +{"train_loss": 0.015724390745162964, "grad_norm": 0.0672084391117096, "global_step": 20975, "epoch": 197, "lr": 6.608086016845426e-05} +{"train_loss": 0.01354159601032734, "grad_norm": 0.10511207580566406, "global_step": 20976, "epoch": 197, "lr": 6.607805383913443e-05} +{"train_loss": 0.012963425368070602, "grad_norm": 0.0695221945643425, "global_step": 20977, "epoch": 197, "lr": 6.60752474533233e-05} +{"train_loss": 0.015313619747757912, "grad_norm": 0.10845393687486649, "global_step": 20978, "epoch": 197, "lr": 6.607244101103074e-05} +{"train_loss": 0.01738857291638851, "grad_norm": 0.11312732845544815, "global_step": 20979, "epoch": 197, "lr": 6.606963451226662e-05} +{"train_loss": 0.018788419663906097, "grad_norm": 0.06516267359256744, "global_step": 20980, "epoch": 197, "lr": 6.606682795704078e-05} +{"train_loss": 0.013989354483783245, "grad_norm": 0.06625001132488251, "global_step": 20981, "epoch": 197, "lr": 6.60640213453631e-05} +{"train_loss": 0.015996992588043213, "grad_norm": 0.06726408004760742, "global_step": 20982, "epoch": 197, "lr": 6.606121467724342e-05} +{"train_loss": 0.01668909192085266, "grad_norm": 0.06977444887161255, "global_step": 20983, "epoch": 197, "lr": 6.605840795269163e-05} +{"train_loss": 0.016577251255512238, "grad_norm": 0.06679703295230865, "global_step": 20984, "epoch": 197, "lr": 6.605560117171758e-05} +{"train_loss": 0.01516798511147499, "grad_norm": 0.06597291678190231, "global_step": 20985, "epoch": 197, "lr": 6.605279433433114e-05} +{"train_loss": 0.015623295679688454, "grad_norm": 0.11320822685956955, "global_step": 20986, "epoch": 197, "lr": 6.604998744054215e-05} +{"train_loss": 0.016033405991587438, "grad_norm": 0.08250285685062408, "global_step": 20987, "epoch": 197, "lr": 6.604718049036048e-05, "val_loss": 0.019526146352291107} +{"train_loss": 0.015261641703546047, "grad_norm": 0.05678169056773186, "global_step": 20988, "epoch": 198, "lr": 6.604437348379601e-05} +{"train_loss": 0.020175648853182793, "grad_norm": 0.0675521045923233, "global_step": 20989, "epoch": 198, "lr": 6.604156642085857e-05} +{"train_loss": 0.01698336936533451, "grad_norm": 0.07237906008958817, "global_step": 20990, "epoch": 198, "lr": 6.603875930155805e-05} +{"train_loss": 0.015481029637157917, "grad_norm": 0.09091062843799591, "global_step": 20991, "epoch": 198, "lr": 6.603595212590431e-05} +{"train_loss": 0.014706889167428017, "grad_norm": 0.079965740442276, "global_step": 20992, "epoch": 198, "lr": 6.603314489390721e-05} +{"train_loss": 0.01790987327694893, "grad_norm": 0.06576857715845108, "global_step": 20993, "epoch": 198, "lr": 6.60303376055766e-05} +{"train_loss": 0.01761205866932869, "grad_norm": 0.07947739213705063, "global_step": 20994, "epoch": 198, "lr": 6.602753026092237e-05} +{"train_loss": 0.015589557588100433, "grad_norm": 0.12851811945438385, "global_step": 20995, "epoch": 198, "lr": 6.602472285995435e-05} +{"train_loss": 0.015119696967303753, "grad_norm": 0.06850320845842361, "global_step": 20996, "epoch": 198, "lr": 6.602191540268243e-05} +{"train_loss": 0.0173975620418787, "grad_norm": 0.09720759838819504, "global_step": 20997, "epoch": 198, "lr": 6.601910788911648e-05} +{"train_loss": 0.016663499176502228, "grad_norm": 0.05834507197141647, "global_step": 20998, "epoch": 198, "lr": 6.601630031926635e-05} +{"train_loss": 0.016397684812545776, "grad_norm": 0.07571884989738464, "global_step": 20999, "epoch": 198, "lr": 6.601349269314188e-05} +{"train_loss": 0.01527286134660244, "grad_norm": 0.11463499069213867, "global_step": 21000, "epoch": 198, "lr": 6.6010685010753e-05} +{"train_loss": 0.016331039369106293, "grad_norm": 0.05780528858304024, "global_step": 21001, "epoch": 198, "lr": 6.60078772721095e-05} +{"train_loss": 0.016175979748368263, "grad_norm": 0.07515032589435577, "global_step": 21002, "epoch": 198, "lr": 6.600506947722129e-05} +{"train_loss": 0.01409976091235876, "grad_norm": 0.050761885941028595, "global_step": 21003, "epoch": 198, "lr": 6.600226162609824e-05} +{"train_loss": 0.014344464987516403, "grad_norm": 0.10754922777414322, "global_step": 21004, "epoch": 198, "lr": 6.599945371875018e-05} +{"train_loss": 0.015238182619214058, "grad_norm": 0.07702372223138809, "global_step": 21005, "epoch": 198, "lr": 6.599664575518702e-05} +{"train_loss": 0.015474759042263031, "grad_norm": 0.06213277578353882, "global_step": 21006, "epoch": 198, "lr": 6.59938377354186e-05} +{"train_loss": 0.014531904831528664, "grad_norm": 0.09170116484165192, "global_step": 21007, "epoch": 198, "lr": 6.599102965945478e-05} +{"train_loss": 0.015702510252594948, "grad_norm": 0.07262861728668213, "global_step": 21008, "epoch": 198, "lr": 6.598822152730544e-05} +{"train_loss": 0.018822461366653442, "grad_norm": 0.0655079185962677, "global_step": 21009, "epoch": 198, "lr": 6.598541333898043e-05} +{"train_loss": 0.019024450331926346, "grad_norm": 0.08386816084384918, "global_step": 21010, "epoch": 198, "lr": 6.598260509448966e-05} +{"train_loss": 0.017038587480783463, "grad_norm": 0.0926794558763504, "global_step": 21011, "epoch": 198, "lr": 6.597979679384294e-05} +{"train_loss": 0.014421243220567703, "grad_norm": 0.12498670816421509, "global_step": 21012, "epoch": 198, "lr": 6.597698843705018e-05} +{"train_loss": 0.01707274466753006, "grad_norm": 0.09664691239595413, "global_step": 21013, "epoch": 198, "lr": 6.597418002412121e-05} +{"train_loss": 0.016061929985880852, "grad_norm": 0.05179087072610855, "global_step": 21014, "epoch": 198, "lr": 6.597137155506594e-05} +{"train_loss": 0.01681765913963318, "grad_norm": 0.06255970895290375, "global_step": 21015, "epoch": 198, "lr": 6.596856302989422e-05} +{"train_loss": 0.015378594398498535, "grad_norm": 0.05687198042869568, "global_step": 21016, "epoch": 198, "lr": 6.596575444861588e-05} +{"train_loss": 0.019148005172610283, "grad_norm": 0.10845281183719635, "global_step": 21017, "epoch": 198, "lr": 6.596294581124087e-05} +{"train_loss": 0.014688430353999138, "grad_norm": 0.07558596134185791, "global_step": 21018, "epoch": 198, "lr": 6.596013711777896e-05} +{"train_loss": 0.017135825008153915, "grad_norm": 0.08142692595720291, "global_step": 21019, "epoch": 198, "lr": 6.59573283682401e-05} +{"train_loss": 0.017683720216155052, "grad_norm": 0.04764103889465332, "global_step": 21020, "epoch": 198, "lr": 6.595451956263413e-05} +{"train_loss": 0.01598406955599785, "grad_norm": 0.08997898548841476, "global_step": 21021, "epoch": 198, "lr": 6.59517107009709e-05} +{"train_loss": 0.01657146029174328, "grad_norm": 0.07982181012630463, "global_step": 21022, "epoch": 198, "lr": 6.59489017832603e-05} +{"train_loss": 0.013100077398121357, "grad_norm": 0.1611068993806839, "global_step": 21023, "epoch": 198, "lr": 6.59460928095122e-05} +{"train_loss": 0.014931468293070793, "grad_norm": 0.10700815171003342, "global_step": 21024, "epoch": 198, "lr": 6.594328377973646e-05} +{"train_loss": 0.014066172763705254, "grad_norm": 0.13824748992919922, "global_step": 21025, "epoch": 198, "lr": 6.594047469394296e-05} +{"train_loss": 0.016515430063009262, "grad_norm": 0.1506127417087555, "global_step": 21026, "epoch": 198, "lr": 6.593766555214157e-05} +{"train_loss": 0.014279961585998535, "grad_norm": 0.05207766592502594, "global_step": 21027, "epoch": 198, "lr": 6.593485635434214e-05} +{"train_loss": 0.015988733619451523, "grad_norm": 0.1438044309616089, "global_step": 21028, "epoch": 198, "lr": 6.593204710055457e-05} +{"train_loss": 0.014655040577054024, "grad_norm": 0.08844542503356934, "global_step": 21029, "epoch": 198, "lr": 6.59292377907887e-05} +{"train_loss": 0.015262664295732975, "grad_norm": 0.06450895220041275, "global_step": 21030, "epoch": 198, "lr": 6.59264284250544e-05} +{"train_loss": 0.015518750995397568, "grad_norm": 0.160141259431839, "global_step": 21031, "epoch": 198, "lr": 6.592361900336159e-05} +{"train_loss": 0.01497095450758934, "grad_norm": 0.07058323919773102, "global_step": 21032, "epoch": 198, "lr": 6.59208095257201e-05} +{"train_loss": 0.012488827109336853, "grad_norm": 0.05706531181931496, "global_step": 21033, "epoch": 198, "lr": 6.59179999921398e-05} +{"train_loss": 0.017026297748088837, "grad_norm": 0.06351568549871445, "global_step": 21034, "epoch": 198, "lr": 6.591519040263059e-05} +{"train_loss": 0.01730160042643547, "grad_norm": 0.11652021110057831, "global_step": 21035, "epoch": 198, "lr": 6.591238075720231e-05} +{"train_loss": 0.016362078487873077, "grad_norm": 0.15486544370651245, "global_step": 21036, "epoch": 198, "lr": 6.590957105586484e-05} +{"train_loss": 0.014673975296318531, "grad_norm": 0.1065719872713089, "global_step": 21037, "epoch": 198, "lr": 6.590676129862805e-05} +{"train_loss": 0.017177242785692215, "grad_norm": 0.0717516839504242, "global_step": 21038, "epoch": 198, "lr": 6.590395148550183e-05} +{"train_loss": 0.014378505758941174, "grad_norm": 0.07032857835292816, "global_step": 21039, "epoch": 198, "lr": 6.590114161649604e-05} +{"train_loss": 0.017140643671154976, "grad_norm": 0.153941810131073, "global_step": 21040, "epoch": 198, "lr": 6.589833169162054e-05} +{"train_loss": 0.013906973414123058, "grad_norm": 0.11353722959756851, "global_step": 21041, "epoch": 198, "lr": 6.589552171088525e-05} +{"train_loss": 0.017431728541851044, "grad_norm": 0.1299198865890503, "global_step": 21042, "epoch": 198, "lr": 6.589271167429997e-05} +{"train_loss": 0.017504841089248657, "grad_norm": 0.20426994562149048, "global_step": 21043, "epoch": 198, "lr": 6.588990158187463e-05} +{"train_loss": 0.015691174194216728, "grad_norm": 0.08000669628381729, "global_step": 21044, "epoch": 198, "lr": 6.588709143361911e-05} +{"train_loss": 0.016628071665763855, "grad_norm": 0.09742563962936401, "global_step": 21045, "epoch": 198, "lr": 6.588428122954321e-05} +{"train_loss": 0.017646469175815582, "grad_norm": 0.07038760930299759, "global_step": 21046, "epoch": 198, "lr": 6.588147096965688e-05} +{"train_loss": 0.014371618628501892, "grad_norm": 0.10912897437810898, "global_step": 21047, "epoch": 198, "lr": 6.587866065396997e-05} +{"train_loss": 0.017204931005835533, "grad_norm": 0.09290916472673416, "global_step": 21048, "epoch": 198, "lr": 6.587585028249235e-05} +{"train_loss": 0.01492521446198225, "grad_norm": 0.06477206945419312, "global_step": 21049, "epoch": 198, "lr": 6.587303985523391e-05} +{"train_loss": 0.015558654442429543, "grad_norm": 0.1647190898656845, "global_step": 21050, "epoch": 198, "lr": 6.587022937220449e-05} +{"train_loss": 0.01826411485671997, "grad_norm": 0.14414182305335999, "global_step": 21051, "epoch": 198, "lr": 6.5867418833414e-05} +{"train_loss": 0.016820766031742096, "grad_norm": 0.17969095706939697, "global_step": 21052, "epoch": 198, "lr": 6.586460823887228e-05} +{"train_loss": 0.019432663917541504, "grad_norm": 0.11971370130777359, "global_step": 21053, "epoch": 198, "lr": 6.586179758858926e-05} +{"train_loss": 0.014931367710232735, "grad_norm": 0.21291927993297577, "global_step": 21054, "epoch": 198, "lr": 6.585898688257474e-05} +{"train_loss": 0.015959380194544792, "grad_norm": 0.051890626549720764, "global_step": 21055, "epoch": 198, "lr": 6.585617612083866e-05} +{"train_loss": 0.015377703122794628, "grad_norm": 0.06078333780169487, "global_step": 21056, "epoch": 198, "lr": 6.585336530339088e-05} +{"train_loss": 0.01379852183163166, "grad_norm": 0.07262390106916428, "global_step": 21057, "epoch": 198, "lr": 6.585055443024125e-05} +{"train_loss": 0.015563873574137688, "grad_norm": 0.09358981996774673, "global_step": 21058, "epoch": 198, "lr": 6.584774350139968e-05} +{"train_loss": 0.014122984372079372, "grad_norm": 0.10685884952545166, "global_step": 21059, "epoch": 198, "lr": 6.584493251687603e-05} +{"train_loss": 0.01586667262017727, "grad_norm": 0.059983912855386734, "global_step": 21060, "epoch": 198, "lr": 6.584212147668017e-05} +{"train_loss": 0.014227291569113731, "grad_norm": 0.15158304572105408, "global_step": 21061, "epoch": 198, "lr": 6.583931038082199e-05} +{"train_loss": 0.01910923607647419, "grad_norm": 0.05572525039315224, "global_step": 21062, "epoch": 198, "lr": 6.583649922931136e-05} +{"train_loss": 0.014204186387360096, "grad_norm": 0.04553089663386345, "global_step": 21063, "epoch": 198, "lr": 6.583368802215816e-05} +{"train_loss": 0.015956133604049683, "grad_norm": 0.07585891336202621, "global_step": 21064, "epoch": 198, "lr": 6.583087675937227e-05} +{"train_loss": 0.014844803139567375, "grad_norm": 0.10554740577936172, "global_step": 21065, "epoch": 198, "lr": 6.582806544096356e-05} +{"train_loss": 0.01570664346218109, "grad_norm": 0.07834812998771667, "global_step": 21066, "epoch": 198, "lr": 6.582525406694191e-05} +{"train_loss": 0.018534507602453232, "grad_norm": 0.1551251858472824, "global_step": 21067, "epoch": 198, "lr": 6.582244263731719e-05} +{"train_loss": 0.013952432200312614, "grad_norm": 0.12668867409229279, "global_step": 21068, "epoch": 198, "lr": 6.581963115209931e-05} +{"train_loss": 0.020436029881238937, "grad_norm": 0.1222536489367485, "global_step": 21069, "epoch": 198, "lr": 6.58168196112981e-05} +{"train_loss": 0.01589268073439598, "grad_norm": 0.061611928045749664, "global_step": 21070, "epoch": 198, "lr": 6.581400801492347e-05} +{"train_loss": 0.01603858917951584, "grad_norm": 0.14759030938148499, "global_step": 21071, "epoch": 198, "lr": 6.58111963629853e-05} +{"train_loss": 0.015448559075593948, "grad_norm": 0.09565532207489014, "global_step": 21072, "epoch": 198, "lr": 6.580838465549347e-05} +{"train_loss": 0.0167899988591671, "grad_norm": 0.09476994723081589, "global_step": 21073, "epoch": 198, "lr": 6.580557289245786e-05} +{"train_loss": 0.01860222965478897, "grad_norm": 0.10110805183649063, "global_step": 21074, "epoch": 198, "lr": 6.580276107388831e-05} +{"train_loss": 0.01726667582988739, "grad_norm": 0.07636366784572601, "global_step": 21075, "epoch": 198, "lr": 6.579994919979474e-05} +{"train_loss": 0.016892731189727783, "grad_norm": 0.09111575037240982, "global_step": 21076, "epoch": 198, "lr": 6.579713727018703e-05} +{"train_loss": 0.012745745480060577, "grad_norm": 0.05471574887633324, "global_step": 21077, "epoch": 198, "lr": 6.579432528507504e-05} +{"train_loss": 0.015912087634205818, "grad_norm": 0.14860089123249054, "global_step": 21078, "epoch": 198, "lr": 6.579151324446866e-05} +{"train_loss": 0.014894062653183937, "grad_norm": 0.12837281823158264, "global_step": 21079, "epoch": 198, "lr": 6.578870114837779e-05} +{"train_loss": 0.01585560292005539, "grad_norm": 0.07446205615997314, "global_step": 21080, "epoch": 198, "lr": 6.578588899681226e-05} +{"train_loss": 0.015382322482764721, "grad_norm": 0.14645403623580933, "global_step": 21081, "epoch": 198, "lr": 6.578307678978199e-05} +{"train_loss": 0.015431507490575314, "grad_norm": 0.14548082649707794, "global_step": 21082, "epoch": 198, "lr": 6.578026452729686e-05} +{"train_loss": 0.014518510550260544, "grad_norm": 0.05930795148015022, "global_step": 21083, "epoch": 198, "lr": 6.577745220936676e-05} +{"train_loss": 0.014861923642456532, "grad_norm": 0.12668386101722717, "global_step": 21084, "epoch": 198, "lr": 6.577463983600152e-05} +{"train_loss": 0.016645949333906174, "grad_norm": 0.21312358975410461, "global_step": 21085, "epoch": 198, "lr": 6.577182740721108e-05} +{"train_loss": 0.016327567398548126, "grad_norm": 0.10695593059062958, "global_step": 21086, "epoch": 198, "lr": 6.576901492300528e-05} +{"train_loss": 0.016764327883720398, "grad_norm": 0.07226622104644775, "global_step": 21087, "epoch": 198, "lr": 6.576620238339403e-05} +{"train_loss": 0.0162262711673975, "grad_norm": 0.1031906008720398, "global_step": 21088, "epoch": 198, "lr": 6.57633897883872e-05} +{"train_loss": 0.01581648178398609, "grad_norm": 0.11259531229734421, "global_step": 21089, "epoch": 198, "lr": 6.576057713799466e-05} +{"train_loss": 0.017330994829535484, "grad_norm": 0.11008541285991669, "global_step": 21090, "epoch": 198, "lr": 6.575776443222631e-05} +{"train_loss": 0.016719365492463112, "grad_norm": 0.1556287556886673, "global_step": 21091, "epoch": 198, "lr": 6.575495167109206e-05} +{"train_loss": 0.017086492851376534, "grad_norm": 0.07279566675424576, "global_step": 21092, "epoch": 198, "lr": 6.575213885460173e-05} +{"train_loss": 0.01607178575095703, "grad_norm": 0.15932288765907288, "global_step": 21093, "epoch": 198, "lr": 6.574932598276525e-05, "val_loss": 0.01984304189682007} +{"train_loss": 0.015907438471913338, "grad_norm": 0.13966448605060577, "global_step": 21094, "epoch": 199, "lr": 6.574651305559248e-05} +{"train_loss": 0.01562977209687233, "grad_norm": 0.13704551756381989, "global_step": 21095, "epoch": 199, "lr": 6.574370007309331e-05} +{"train_loss": 0.014368586242198944, "grad_norm": 0.10317036509513855, "global_step": 21096, "epoch": 199, "lr": 6.574088703527762e-05} +{"train_loss": 0.014553690329194069, "grad_norm": 0.13282345235347748, "global_step": 21097, "epoch": 199, "lr": 6.573807394215532e-05} +{"train_loss": 0.017979655414819717, "grad_norm": 0.08181961625814438, "global_step": 21098, "epoch": 199, "lr": 6.573526079373624e-05} +{"train_loss": 0.01480306126177311, "grad_norm": 0.06037026643753052, "global_step": 21099, "epoch": 199, "lr": 6.573244759003032e-05} +{"train_loss": 0.015949685126543045, "grad_norm": 0.08325924724340439, "global_step": 21100, "epoch": 199, "lr": 6.572963433104741e-05} +{"train_loss": 0.016277093440294266, "grad_norm": 0.14107654988765717, "global_step": 21101, "epoch": 199, "lr": 6.572682101679741e-05} +{"train_loss": 0.016268450766801834, "grad_norm": 0.07419080287218094, "global_step": 21102, "epoch": 199, "lr": 6.572400764729021e-05} +{"train_loss": 0.014911746606230736, "grad_norm": 0.05517919361591339, "global_step": 21103, "epoch": 199, "lr": 6.572119422253568e-05} +{"train_loss": 0.01661427691578865, "grad_norm": 0.14019951224327087, "global_step": 21104, "epoch": 199, "lr": 6.57183807425437e-05} +{"train_loss": 0.015463704243302345, "grad_norm": 0.04956068843603134, "global_step": 21105, "epoch": 199, "lr": 6.571556720732416e-05} +{"train_loss": 0.015768498182296753, "grad_norm": 0.15342797338962555, "global_step": 21106, "epoch": 199, "lr": 6.571275361688698e-05} +{"train_loss": 0.018226884305477142, "grad_norm": 0.1747709959745407, "global_step": 21107, "epoch": 199, "lr": 6.5709939971242e-05} +{"train_loss": 0.016062118113040924, "grad_norm": 0.1493813544511795, "global_step": 21108, "epoch": 199, "lr": 6.570712627039912e-05} +{"train_loss": 0.016262933611869812, "grad_norm": 0.09052644670009613, "global_step": 21109, "epoch": 199, "lr": 6.570431251436823e-05} +{"train_loss": 0.013527360744774342, "grad_norm": 0.07596086710691452, "global_step": 21110, "epoch": 199, "lr": 6.570149870315923e-05} +{"train_loss": 0.015756087377667427, "grad_norm": 0.15166506171226501, "global_step": 21111, "epoch": 199, "lr": 6.569868483678196e-05} +{"train_loss": 0.015450662933290005, "grad_norm": 0.05873342975974083, "global_step": 21112, "epoch": 199, "lr": 6.569587091524639e-05} +{"train_loss": 0.017533903941512108, "grad_norm": 0.08971696346998215, "global_step": 21113, "epoch": 199, "lr": 6.56930569385623e-05} +{"train_loss": 0.018337585031986237, "grad_norm": 0.06717560440301895, "global_step": 21114, "epoch": 199, "lr": 6.569024290673967e-05} +{"train_loss": 0.015552831813693047, "grad_norm": 0.061661552637815475, "global_step": 21115, "epoch": 199, "lr": 6.568742881978834e-05} +{"train_loss": 0.01553464587777853, "grad_norm": 0.11719288676977158, "global_step": 21116, "epoch": 199, "lr": 6.56846146777182e-05} +{"train_loss": 0.018328532576560974, "grad_norm": 0.07381481677293777, "global_step": 21117, "epoch": 199, "lr": 6.568180048053915e-05} +{"train_loss": 0.017164036631584167, "grad_norm": 0.06340494751930237, "global_step": 21118, "epoch": 199, "lr": 6.567898622826108e-05} +{"train_loss": 0.015996064990758896, "grad_norm": 0.10088811069726944, "global_step": 21119, "epoch": 199, "lr": 6.567617192089386e-05} +{"train_loss": 0.014904836192727089, "grad_norm": 0.08559863269329071, "global_step": 21120, "epoch": 199, "lr": 6.567335755844739e-05} +{"train_loss": 0.01459156908094883, "grad_norm": 0.11725407838821411, "global_step": 21121, "epoch": 199, "lr": 6.567054314093156e-05} +{"train_loss": 0.013231809251010418, "grad_norm": 0.13180015981197357, "global_step": 21122, "epoch": 199, "lr": 6.566772866835625e-05} +{"train_loss": 0.016563165932893753, "grad_norm": 0.08250138908624649, "global_step": 21123, "epoch": 199, "lr": 6.566491414073137e-05} +{"train_loss": 0.015382477082312107, "grad_norm": 0.06981062144041061, "global_step": 21124, "epoch": 199, "lr": 6.56620995580668e-05} +{"train_loss": 0.01811639405786991, "grad_norm": 0.18622222542762756, "global_step": 21125, "epoch": 199, "lr": 6.565928492037239e-05} +{"train_loss": 0.014674859121441841, "grad_norm": 0.11304934322834015, "global_step": 21126, "epoch": 199, "lr": 6.565647022765809e-05} +{"train_loss": 0.015402618795633316, "grad_norm": 0.05703834071755409, "global_step": 21127, "epoch": 199, "lr": 6.565365547993375e-05} +{"train_loss": 0.015679996460676193, "grad_norm": 0.054002463817596436, "global_step": 21128, "epoch": 199, "lr": 6.565084067720928e-05} +{"train_loss": 0.016261843964457512, "grad_norm": 0.07144582271575928, "global_step": 21129, "epoch": 199, "lr": 6.564802581949457e-05} +{"train_loss": 0.014523045159876347, "grad_norm": 0.13096356391906738, "global_step": 21130, "epoch": 199, "lr": 6.564521090679948e-05} +{"train_loss": 0.015780238434672356, "grad_norm": 0.10161618888378143, "global_step": 21131, "epoch": 199, "lr": 6.564239593913394e-05} +{"train_loss": 0.014226621948182583, "grad_norm": 0.09394466876983643, "global_step": 21132, "epoch": 199, "lr": 6.563958091650781e-05} +{"train_loss": 0.018491966649889946, "grad_norm": 0.11111018806695938, "global_step": 21133, "epoch": 199, "lr": 6.563676583893102e-05} +{"train_loss": 0.016062993556261063, "grad_norm": 0.057645902037620544, "global_step": 21134, "epoch": 199, "lr": 6.56339507064134e-05} +{"train_loss": 0.015474929474294186, "grad_norm": 0.07305127382278442, "global_step": 21135, "epoch": 199, "lr": 6.56311355189649e-05} +{"train_loss": 0.014068655669689178, "grad_norm": 0.15355241298675537, "global_step": 21136, "epoch": 199, "lr": 6.56283202765954e-05} +{"train_loss": 0.0163717158138752, "grad_norm": 0.1231064721941948, "global_step": 21137, "epoch": 199, "lr": 6.562550497931475e-05} +{"train_loss": 0.01636994071304798, "grad_norm": 0.07036930322647095, "global_step": 21138, "epoch": 199, "lr": 6.562268962713288e-05} +{"train_loss": 0.01713278889656067, "grad_norm": 0.14919179677963257, "global_step": 21139, "epoch": 199, "lr": 6.561987422005966e-05} +{"train_loss": 0.015211024321615696, "grad_norm": 0.1963898092508316, "global_step": 21140, "epoch": 199, "lr": 6.561705875810502e-05} +{"train_loss": 0.01630188152194023, "grad_norm": 0.07527428865432739, "global_step": 21141, "epoch": 199, "lr": 6.56142432412788e-05} +{"train_loss": 0.016678903251886368, "grad_norm": 0.10132062435150146, "global_step": 21142, "epoch": 199, "lr": 6.561142766959095e-05} +{"train_loss": 0.013881637714803219, "grad_norm": 0.048209935426712036, "global_step": 21143, "epoch": 199, "lr": 6.560861204305132e-05} +{"train_loss": 0.01630261540412903, "grad_norm": 0.06508183479309082, "global_step": 21144, "epoch": 199, "lr": 6.560579636166981e-05} +{"train_loss": 0.01487358845770359, "grad_norm": 0.1236909031867981, "global_step": 21145, "epoch": 199, "lr": 6.560298062545633e-05} +{"train_loss": 0.014999983832240105, "grad_norm": 0.07138754427433014, "global_step": 21146, "epoch": 199, "lr": 6.560016483442075e-05} +{"train_loss": 0.017494266852736473, "grad_norm": 0.09089723229408264, "global_step": 21147, "epoch": 199, "lr": 6.559734898857299e-05} +{"train_loss": 0.014829089865088463, "grad_norm": 0.06410158425569534, "global_step": 21148, "epoch": 199, "lr": 6.559453308792291e-05} +{"train_loss": 0.016604572534561157, "grad_norm": 0.09926889836788177, "global_step": 21149, "epoch": 199, "lr": 6.559171713248042e-05} +{"train_loss": 0.016408242285251617, "grad_norm": 0.12758517265319824, "global_step": 21150, "epoch": 199, "lr": 6.558890112225544e-05} +{"train_loss": 0.014048365876078606, "grad_norm": 0.07259323447942734, "global_step": 21151, "epoch": 199, "lr": 6.55860850572578e-05} +{"train_loss": 0.01642865687608719, "grad_norm": 0.06418589502573013, "global_step": 21152, "epoch": 199, "lr": 6.558326893749747e-05} +{"train_loss": 0.01640588790178299, "grad_norm": 0.058147747069597244, "global_step": 21153, "epoch": 199, "lr": 6.558045276298432e-05} +{"train_loss": 0.019240228459239006, "grad_norm": 0.07391288876533508, "global_step": 21154, "epoch": 199, "lr": 6.557763653372822e-05} +{"train_loss": 0.018000585958361626, "grad_norm": 0.08634722977876663, "global_step": 21155, "epoch": 199, "lr": 6.557482024973907e-05} +{"train_loss": 0.015910811722278595, "grad_norm": 0.05814376100897789, "global_step": 21156, "epoch": 199, "lr": 6.557200391102678e-05} +{"train_loss": 0.015982847660779953, "grad_norm": 0.15081767737865448, "global_step": 21157, "epoch": 199, "lr": 6.556918751760124e-05} +{"train_loss": 0.015269639901816845, "grad_norm": 0.10250008851289749, "global_step": 21158, "epoch": 199, "lr": 6.556637106947235e-05} +{"train_loss": 0.015239626169204712, "grad_norm": 0.12624938786029816, "global_step": 21159, "epoch": 199, "lr": 6.556355456665e-05} +{"train_loss": 0.01865539699792862, "grad_norm": 0.06406892836093903, "global_step": 21160, "epoch": 199, "lr": 6.55607380091441e-05} +{"train_loss": 0.01407907996326685, "grad_norm": 0.09504808485507965, "global_step": 21161, "epoch": 199, "lr": 6.55579213969645e-05} +{"train_loss": 0.017285577952861786, "grad_norm": 0.11056866496801376, "global_step": 21162, "epoch": 199, "lr": 6.555510473012116e-05} +{"train_loss": 0.015251566655933857, "grad_norm": 0.11382214725017548, "global_step": 21163, "epoch": 199, "lr": 6.555228800862395e-05} +{"train_loss": 0.018694184720516205, "grad_norm": 0.06763602048158646, "global_step": 21164, "epoch": 199, "lr": 6.554947123248276e-05} +{"train_loss": 0.016521655023097992, "grad_norm": 0.08405863493680954, "global_step": 21165, "epoch": 199, "lr": 6.554665440170748e-05} +{"train_loss": 0.017705898731946945, "grad_norm": 0.05609182268381119, "global_step": 21166, "epoch": 199, "lr": 6.5543837516308e-05} +{"train_loss": 0.01585972309112549, "grad_norm": 0.06122775748372078, "global_step": 21167, "epoch": 199, "lr": 6.554102057629428e-05} +{"train_loss": 0.01502909604460001, "grad_norm": 0.06472467631101608, "global_step": 21168, "epoch": 199, "lr": 6.553820358167613e-05} +{"train_loss": 0.018957719206809998, "grad_norm": 0.09570223838090897, "global_step": 21169, "epoch": 199, "lr": 6.553538653246351e-05} +{"train_loss": 0.015442718751728535, "grad_norm": 0.12324782460927963, "global_step": 21170, "epoch": 199, "lr": 6.55325694286663e-05} +{"train_loss": 0.017962392419576645, "grad_norm": 0.055011454969644547, "global_step": 21171, "epoch": 199, "lr": 6.552975227029439e-05} +{"train_loss": 0.016967184841632843, "grad_norm": 0.10351298749446869, "global_step": 21172, "epoch": 199, "lr": 6.552693505735767e-05} +{"train_loss": 0.015350358560681343, "grad_norm": 0.06543825566768646, "global_step": 21173, "epoch": 199, "lr": 6.552411778986608e-05} +{"train_loss": 0.013387016020715237, "grad_norm": 0.07477276027202606, "global_step": 21174, "epoch": 199, "lr": 6.552130046782949e-05} +{"train_loss": 0.016886252909898758, "grad_norm": 0.09033855050802231, "global_step": 21175, "epoch": 199, "lr": 6.551848309125777e-05} +{"train_loss": 0.015358654782176018, "grad_norm": 0.053761761635541916, "global_step": 21176, "epoch": 199, "lr": 6.551566566016088e-05} +{"train_loss": 0.014699936844408512, "grad_norm": 0.09970765560865402, "global_step": 21177, "epoch": 199, "lr": 6.551284817454868e-05} +{"train_loss": 0.018332701176404953, "grad_norm": 0.11008315533399582, "global_step": 21178, "epoch": 199, "lr": 6.551003063443107e-05} +{"train_loss": 0.017138786613941193, "grad_norm": 0.06674224883317947, "global_step": 21179, "epoch": 199, "lr": 6.550721303981797e-05} +{"train_loss": 0.014409090392291546, "grad_norm": 0.06329146027565002, "global_step": 21180, "epoch": 199, "lr": 6.550439539071924e-05} +{"train_loss": 0.016698617488145828, "grad_norm": 0.10755740106105804, "global_step": 21181, "epoch": 199, "lr": 6.550157768714483e-05} +{"train_loss": 0.012052970938384533, "grad_norm": 0.05396788567304611, "global_step": 21182, "epoch": 199, "lr": 6.549875992910463e-05} +{"train_loss": 0.01435274537652731, "grad_norm": 0.12700627744197845, "global_step": 21183, "epoch": 199, "lr": 6.54959421166085e-05} +{"train_loss": 0.017020735889673233, "grad_norm": 0.07226254045963287, "global_step": 21184, "epoch": 199, "lr": 6.549312424966639e-05} +{"train_loss": 0.016040615737438202, "grad_norm": 0.08459586650133133, "global_step": 21185, "epoch": 199, "lr": 6.549030632828816e-05} +{"train_loss": 0.016230858862400055, "grad_norm": 0.09873185306787491, "global_step": 21186, "epoch": 199, "lr": 6.548748835248376e-05} +{"train_loss": 0.01641063019633293, "grad_norm": 0.13894523680210114, "global_step": 21187, "epoch": 199, "lr": 6.548467032226303e-05} +{"train_loss": 0.017674751579761505, "grad_norm": 0.1317022293806076, "global_step": 21188, "epoch": 199, "lr": 6.548185223763591e-05} +{"train_loss": 0.014372874051332474, "grad_norm": 0.06612447649240494, "global_step": 21189, "epoch": 199, "lr": 6.547903409861231e-05} +{"train_loss": 0.014724121429026127, "grad_norm": 0.06484293937683105, "global_step": 21190, "epoch": 199, "lr": 6.547621590520209e-05} +{"train_loss": 0.015883486717939377, "grad_norm": 0.09179046005010605, "global_step": 21191, "epoch": 199, "lr": 6.54733976574152e-05} +{"train_loss": 0.01588783785700798, "grad_norm": 0.06884537637233734, "global_step": 21192, "epoch": 199, "lr": 6.54705793552615e-05} +{"train_loss": 0.01672571338713169, "grad_norm": 0.11380942910909653, "global_step": 21193, "epoch": 199, "lr": 6.546776099875093e-05} +{"train_loss": 0.0149193424731493, "grad_norm": 0.08588182926177979, "global_step": 21194, "epoch": 199, "lr": 6.546494258789338e-05} +{"train_loss": 0.015104459598660469, "grad_norm": 0.06958756595849991, "global_step": 21195, "epoch": 199, "lr": 6.546212412269872e-05} +{"train_loss": 0.018575111404061317, "grad_norm": 0.06605023890733719, "global_step": 21196, "epoch": 199, "lr": 6.54593056031769e-05} +{"train_loss": 0.016344673931598663, "grad_norm": 0.06626801937818527, "global_step": 21197, "epoch": 199, "lr": 6.54564870293378e-05} +{"train_loss": 0.01842968910932541, "grad_norm": 0.1144520714879036, "global_step": 21198, "epoch": 199, "lr": 6.545366840119131e-05} +{"train_loss": 0.01602265704423189, "grad_norm": 0.06144961342215538, "global_step": 21199, "epoch": 199, "lr": 6.545084971874738e-05, "val_loss": 0.02062038518488407} +{"train_loss": 0.015149029903113842, "grad_norm": 0.05842997133731842, "global_step": 21200, "epoch": 200, "lr": 6.544803098201586e-05} +{"train_loss": 0.016260556876659393, "grad_norm": 0.07235254347324371, "global_step": 21201, "epoch": 200, "lr": 6.544521219100669e-05} +{"train_loss": 0.016241632401943207, "grad_norm": 0.06396517902612686, "global_step": 21202, "epoch": 200, "lr": 6.544239334572974e-05} +{"train_loss": 0.0120774544775486, "grad_norm": 0.06981725245714188, "global_step": 21203, "epoch": 200, "lr": 6.543957444619498e-05} +{"train_loss": 0.01666717603802681, "grad_norm": 0.0880008339881897, "global_step": 21204, "epoch": 200, "lr": 6.543675549241224e-05} +{"train_loss": 0.012428175657987595, "grad_norm": 0.09555987268686295, "global_step": 21205, "epoch": 200, "lr": 6.543393648439144e-05} +{"train_loss": 0.016570886597037315, "grad_norm": 0.11980032920837402, "global_step": 21206, "epoch": 200, "lr": 6.543111742214253e-05} +{"train_loss": 0.019864926114678383, "grad_norm": 0.0963447168469429, "global_step": 21207, "epoch": 200, "lr": 6.542829830567537e-05} +{"train_loss": 0.016195908188819885, "grad_norm": 0.201241597533226, "global_step": 21208, "epoch": 200, "lr": 6.542547913499988e-05} +{"train_loss": 0.014975321479141712, "grad_norm": 0.06415987759828568, "global_step": 21209, "epoch": 200, "lr": 6.542265991012596e-05} +{"train_loss": 0.015631675720214844, "grad_norm": 0.06747116893529892, "global_step": 21210, "epoch": 200, "lr": 6.541984063106352e-05} +{"train_loss": 0.015069546177983284, "grad_norm": 0.05917944386601448, "global_step": 21211, "epoch": 200, "lr": 6.541702129782247e-05} +{"train_loss": 0.01545337401330471, "grad_norm": 0.09337276965379715, "global_step": 21212, "epoch": 200, "lr": 6.541420191041273e-05} +{"train_loss": 0.01612124964594841, "grad_norm": 0.06705808639526367, "global_step": 21213, "epoch": 200, "lr": 6.541138246884417e-05} +{"train_loss": 0.015709465369582176, "grad_norm": 0.08277980983257294, "global_step": 21214, "epoch": 200, "lr": 6.54085629731267e-05} +{"train_loss": 0.015721822157502174, "grad_norm": 0.06831374764442444, "global_step": 21215, "epoch": 200, "lr": 6.540574342327027e-05} +{"train_loss": 0.016290564090013504, "grad_norm": 0.07076849043369293, "global_step": 21216, "epoch": 200, "lr": 6.540292381928474e-05} +{"train_loss": 0.014264773577451706, "grad_norm": 0.132229745388031, "global_step": 21217, "epoch": 200, "lr": 6.540010416118004e-05} +{"train_loss": 0.014639385044574738, "grad_norm": 0.07703359425067902, "global_step": 21218, "epoch": 200, "lr": 6.539728444896606e-05} +{"train_loss": 0.016222868114709854, "grad_norm": 0.09830373525619507, "global_step": 21219, "epoch": 200, "lr": 6.539446468265271e-05} +{"train_loss": 0.018557429313659668, "grad_norm": 0.09818582981824875, "global_step": 21220, "epoch": 200, "lr": 6.539164486224994e-05} +{"train_loss": 0.014909074641764164, "grad_norm": 0.17360147833824158, "global_step": 21221, "epoch": 200, "lr": 6.53888249877676e-05} +{"train_loss": 0.0180918350815773, "grad_norm": 0.18774215877056122, "global_step": 21222, "epoch": 200, "lr": 6.538600505921563e-05} +{"train_loss": 0.014532649889588356, "grad_norm": 0.10907744616270065, "global_step": 21223, "epoch": 200, "lr": 6.538318507660394e-05} +{"train_loss": 0.014020116999745369, "grad_norm": 0.11325138062238693, "global_step": 21224, "epoch": 200, "lr": 6.538036503994241e-05} +{"train_loss": 0.01650811731815338, "grad_norm": 0.0732986330986023, "global_step": 21225, "epoch": 200, "lr": 6.537754494924096e-05} +{"train_loss": 0.015697523951530457, "grad_norm": 0.09227647632360458, "global_step": 21226, "epoch": 200, "lr": 6.537472480450953e-05} +{"train_loss": 0.015976767987012863, "grad_norm": 0.05264011397957802, "global_step": 21227, "epoch": 200, "lr": 6.537190460575798e-05} +{"train_loss": 0.016116928309202194, "grad_norm": 0.0747796967625618, "global_step": 21228, "epoch": 200, "lr": 6.536908435299624e-05} +{"train_loss": 0.013657873496413231, "grad_norm": 0.08806414902210236, "global_step": 21229, "epoch": 200, "lr": 6.536626404623424e-05} +{"train_loss": 0.016318775713443756, "grad_norm": 0.09684278815984726, "global_step": 21230, "epoch": 200, "lr": 6.536344368548186e-05} +{"train_loss": 0.017168262973427773, "grad_norm": 0.05246901512145996, "global_step": 21231, "epoch": 200, "lr": 6.536062327074903e-05} +{"train_loss": 0.017551833763718605, "grad_norm": 0.0857740193605423, "global_step": 21232, "epoch": 200, "lr": 6.535780280204564e-05} +{"train_loss": 0.014976974576711655, "grad_norm": 0.07448935508728027, "global_step": 21233, "epoch": 200, "lr": 6.535498227938161e-05} +{"train_loss": 0.01570483110845089, "grad_norm": 0.07235913723707199, "global_step": 21234, "epoch": 200, "lr": 6.535216170276685e-05} +{"train_loss": 0.01907048001885414, "grad_norm": 0.06747983396053314, "global_step": 21235, "epoch": 200, "lr": 6.534934107221129e-05} +{"train_loss": 0.013998894020915031, "grad_norm": 0.12221033126115799, "global_step": 21236, "epoch": 200, "lr": 6.53465203877248e-05} +{"train_loss": 0.015537556260824203, "grad_norm": 0.08889424055814743, "global_step": 21237, "epoch": 200, "lr": 6.534369964931731e-05} +{"train_loss": 0.017803087830543518, "grad_norm": 0.06912294030189514, "global_step": 21238, "epoch": 200, "lr": 6.534087885699873e-05} +{"train_loss": 0.016398262232542038, "grad_norm": 0.07124222069978714, "global_step": 21239, "epoch": 200, "lr": 6.533805801077899e-05} +{"train_loss": 0.017468329519033432, "grad_norm": 0.08897528797388077, "global_step": 21240, "epoch": 200, "lr": 6.533523711066798e-05} +{"train_loss": 0.017492394894361496, "grad_norm": 0.13957509398460388, "global_step": 21241, "epoch": 200, "lr": 6.533241615667561e-05} +{"train_loss": 0.014451153576374054, "grad_norm": 0.058487143367528915, "global_step": 21242, "epoch": 200, "lr": 6.532959514881181e-05} +{"train_loss": 0.017753681167960167, "grad_norm": 0.08306212723255157, "global_step": 21243, "epoch": 200, "lr": 6.532677408708645e-05} +{"train_loss": 0.014324722811579704, "grad_norm": 0.04767855629324913, "global_step": 21244, "epoch": 200, "lr": 6.532395297150951e-05} +{"train_loss": 0.014647718518972397, "grad_norm": 0.06489930301904678, "global_step": 21245, "epoch": 200, "lr": 6.532113180209083e-05} +{"train_loss": 0.013514496386051178, "grad_norm": 0.04368439316749573, "global_step": 21246, "epoch": 200, "lr": 6.531831057884037e-05} +{"train_loss": 0.015530447475612164, "grad_norm": 0.09290633350610733, "global_step": 21247, "epoch": 200, "lr": 6.531548930176803e-05} +{"train_loss": 0.01779995858669281, "grad_norm": 0.11256594955921173, "global_step": 21248, "epoch": 200, "lr": 6.531266797088371e-05} +{"train_loss": 0.016323624178767204, "grad_norm": 0.10565757006406784, "global_step": 21249, "epoch": 200, "lr": 6.530984658619734e-05} +{"train_loss": 0.015184981748461723, "grad_norm": 0.07958800345659256, "global_step": 21250, "epoch": 200, "lr": 6.530702514771883e-05} +{"train_loss": 0.01579272374510765, "grad_norm": 0.16233982145786285, "global_step": 21251, "epoch": 200, "lr": 6.53042036554581e-05} +{"train_loss": 0.015319719910621643, "grad_norm": 0.08938726037740707, "global_step": 21252, "epoch": 200, "lr": 6.530138210942505e-05} +{"train_loss": 0.014414352364838123, "grad_norm": 0.06742933392524719, "global_step": 21253, "epoch": 200, "lr": 6.529856050962958e-05} +{"train_loss": 0.015815233811736107, "grad_norm": 0.06334054470062256, "global_step": 21254, "epoch": 200, "lr": 6.529573885608164e-05} +{"train_loss": 0.01476312056183815, "grad_norm": 0.060137633234262466, "global_step": 21255, "epoch": 200, "lr": 6.529291714879111e-05} +{"train_loss": 0.01655663177371025, "grad_norm": 0.05432787537574768, "global_step": 21256, "epoch": 200, "lr": 6.529009538776794e-05} +{"train_loss": 0.017538700252771378, "grad_norm": 0.12263422459363937, "global_step": 21257, "epoch": 200, "lr": 6.528727357302201e-05} +{"train_loss": 0.012773832306265831, "grad_norm": 0.13726146519184113, "global_step": 21258, "epoch": 200, "lr": 6.528445170456325e-05} +{"train_loss": 0.018067630007863045, "grad_norm": 0.07843396812677383, "global_step": 21259, "epoch": 200, "lr": 6.528162978240158e-05} +{"train_loss": 0.016572196036577225, "grad_norm": 0.079398512840271, "global_step": 21260, "epoch": 200, "lr": 6.52788078065469e-05} +{"train_loss": 0.017020422965288162, "grad_norm": 0.14039859175682068, "global_step": 21261, "epoch": 200, "lr": 6.527598577700914e-05} +{"train_loss": 0.01565564051270485, "grad_norm": 0.09607988595962524, "global_step": 21262, "epoch": 200, "lr": 6.527316369379822e-05} +{"train_loss": 0.01555723138153553, "grad_norm": 0.07483799755573273, "global_step": 21263, "epoch": 200, "lr": 6.527034155692402e-05} +{"train_loss": 0.01619366742670536, "grad_norm": 0.12537288665771484, "global_step": 21264, "epoch": 200, "lr": 6.526751936639649e-05} +{"train_loss": 0.014539480209350586, "grad_norm": 0.1008429005742073, "global_step": 21265, "epoch": 200, "lr": 6.526469712222554e-05} +{"train_loss": 0.016702283173799515, "grad_norm": 0.130098357796669, "global_step": 21266, "epoch": 200, "lr": 6.526187482442109e-05} +{"train_loss": 0.01659133844077587, "grad_norm": 0.0915154442191124, "global_step": 21267, "epoch": 200, "lr": 6.525905247299304e-05} +{"train_loss": 0.01818693056702614, "grad_norm": 0.09919179975986481, "global_step": 21268, "epoch": 200, "lr": 6.525623006795132e-05} +{"train_loss": 0.016584977507591248, "grad_norm": 0.06960610300302505, "global_step": 21269, "epoch": 200, "lr": 6.525340760930584e-05} +{"train_loss": 0.01721745729446411, "grad_norm": 0.07098750025033951, "global_step": 21270, "epoch": 200, "lr": 6.525058509706651e-05} +{"train_loss": 0.01723570190370083, "grad_norm": 0.147694393992424, "global_step": 21271, "epoch": 200, "lr": 6.524776253124327e-05} +{"train_loss": 0.014270339161157608, "grad_norm": 0.13110321760177612, "global_step": 21272, "epoch": 200, "lr": 6.524493991184601e-05} +{"train_loss": 0.01885666325688362, "grad_norm": 0.08344707638025284, "global_step": 21273, "epoch": 200, "lr": 6.524211723888468e-05} +{"train_loss": 0.016878437250852585, "grad_norm": 0.06833361834287643, "global_step": 21274, "epoch": 200, "lr": 6.523929451236917e-05} +{"train_loss": 0.01905744895339012, "grad_norm": 0.09439950436353683, "global_step": 21275, "epoch": 200, "lr": 6.523647173230941e-05} +{"train_loss": 0.014442481100559235, "grad_norm": 0.0633915513753891, "global_step": 21276, "epoch": 200, "lr": 6.523364889871531e-05} +{"train_loss": 0.018724452704191208, "grad_norm": 0.0717410072684288, "global_step": 21277, "epoch": 200, "lr": 6.523082601159678e-05} +{"train_loss": 0.014040878042578697, "grad_norm": 0.07652761787176132, "global_step": 21278, "epoch": 200, "lr": 6.522800307096376e-05} +{"train_loss": 0.01543312519788742, "grad_norm": 0.0881245955824852, "global_step": 21279, "epoch": 200, "lr": 6.522518007682616e-05} +{"train_loss": 0.015017664059996605, "grad_norm": 0.06526985764503479, "global_step": 21280, "epoch": 200, "lr": 6.522235702919391e-05} +{"train_loss": 0.016842756420373917, "grad_norm": 0.23809416592121124, "global_step": 21281, "epoch": 200, "lr": 6.521953392807691e-05} +{"train_loss": 0.017121266573667526, "grad_norm": 0.159355029463768, "global_step": 21282, "epoch": 200, "lr": 6.521671077348508e-05} +{"train_loss": 0.014201784506440163, "grad_norm": 0.1665779948234558, "global_step": 21283, "epoch": 200, "lr": 6.521388756542835e-05} +{"train_loss": 0.01649099960923195, "grad_norm": 0.1100899875164032, "global_step": 21284, "epoch": 200, "lr": 6.521106430391665e-05} +{"train_loss": 0.01839396357536316, "grad_norm": 0.14259427785873413, "global_step": 21285, "epoch": 200, "lr": 6.520824098895987e-05} +{"train_loss": 0.014136704616248608, "grad_norm": 0.07858214527368546, "global_step": 21286, "epoch": 200, "lr": 6.520541762056794e-05} +{"train_loss": 0.017886681482195854, "grad_norm": 0.06066802144050598, "global_step": 21287, "epoch": 200, "lr": 6.52025941987508e-05} +{"train_loss": 0.015078327618539333, "grad_norm": 0.06968405842781067, "global_step": 21288, "epoch": 200, "lr": 6.519977072351836e-05} +{"train_loss": 0.015880057588219643, "grad_norm": 0.14034539461135864, "global_step": 21289, "epoch": 200, "lr": 6.51969471948805e-05} +{"train_loss": 0.017990462481975555, "grad_norm": 0.10970611125230789, "global_step": 21290, "epoch": 200, "lr": 6.51941236128472e-05} +{"train_loss": 0.014428623020648956, "grad_norm": 0.1060534194111824, "global_step": 21291, "epoch": 200, "lr": 6.519129997742836e-05} +{"train_loss": 0.016830481588840485, "grad_norm": 0.1885872781276703, "global_step": 21292, "epoch": 200, "lr": 6.51884762886339e-05} +{"train_loss": 0.017717529088258743, "grad_norm": 0.11456809192895889, "global_step": 21293, "epoch": 200, "lr": 6.518565254647373e-05} +{"train_loss": 0.017975833266973495, "grad_norm": 0.18101193010807037, "global_step": 21294, "epoch": 200, "lr": 6.51828287509578e-05} +{"train_loss": 0.01737062633037567, "grad_norm": 0.0896020159125328, "global_step": 21295, "epoch": 200, "lr": 6.518000490209599e-05} +{"train_loss": 0.013682771474123001, "grad_norm": 0.06503336876630783, "global_step": 21296, "epoch": 200, "lr": 6.517718099989825e-05} +{"train_loss": 0.013894721865653992, "grad_norm": 0.06397821754217148, "global_step": 21297, "epoch": 200, "lr": 6.51743570443745e-05} +{"train_loss": 0.0154600590467453, "grad_norm": 0.05237184837460518, "global_step": 21298, "epoch": 200, "lr": 6.517153303553466e-05} +{"train_loss": 0.01601886749267578, "grad_norm": 0.10847246646881104, "global_step": 21299, "epoch": 200, "lr": 6.516870897338863e-05} +{"train_loss": 0.013810672797262669, "grad_norm": 0.10791022330522537, "global_step": 21300, "epoch": 200, "lr": 6.516588485794639e-05} +{"train_loss": 0.016332784667611122, "grad_norm": 0.06426512449979782, "global_step": 21301, "epoch": 200, "lr": 6.51630606892178e-05} +{"train_loss": 0.017564907670021057, "grad_norm": 0.077165387570858, "global_step": 21302, "epoch": 200, "lr": 6.516023646721281e-05} +{"train_loss": 0.01784047856926918, "grad_norm": 0.07501532137393951, "global_step": 21303, "epoch": 200, "lr": 6.515741219194137e-05} +{"train_loss": 0.015201494097709656, "grad_norm": 0.07176163047552109, "global_step": 21304, "epoch": 200, "lr": 6.515458786341335e-05} +{"train_loss": 0.016023931392240075, "grad_norm": 0.1876533031463623, "global_step": 21305, "epoch": 200, "lr": 6.515176348163871e-05, "train_batch_mse_error": 0.014227085746824741, "val_loss": 0.01938086748123169} +{"train_loss": 0.015297042205929756, "grad_norm": 0.11402695626020432, "global_step": 21306, "epoch": 201, "lr": 6.514893904662736e-05} +{"train_loss": 0.01647583395242691, "grad_norm": 0.09265843033790588, "global_step": 21307, "epoch": 201, "lr": 6.514611455838921e-05} +{"train_loss": 0.015772320330142975, "grad_norm": 0.1699845939874649, "global_step": 21308, "epoch": 201, "lr": 6.514329001693422e-05} +{"train_loss": 0.016914188861846924, "grad_norm": 0.07753081619739532, "global_step": 21309, "epoch": 201, "lr": 6.514046542227227e-05} +{"train_loss": 0.013394029811024666, "grad_norm": 0.10595091432332993, "global_step": 21310, "epoch": 201, "lr": 6.513764077441333e-05} +{"train_loss": 0.013962335884571075, "grad_norm": 0.11119158565998077, "global_step": 21311, "epoch": 201, "lr": 6.51348160733673e-05} +{"train_loss": 0.018057025969028473, "grad_norm": 0.07624967396259308, "global_step": 21312, "epoch": 201, "lr": 6.513199131914412e-05} +{"train_loss": 0.017273474484682083, "grad_norm": 0.06725598126649857, "global_step": 21313, "epoch": 201, "lr": 6.512916651175369e-05} +{"train_loss": 0.015060419216752052, "grad_norm": 0.06133638694882393, "global_step": 21314, "epoch": 201, "lr": 6.512634165120594e-05} +{"train_loss": 0.014744330197572708, "grad_norm": 0.07133574783802032, "global_step": 21315, "epoch": 201, "lr": 6.512351673751082e-05} +{"train_loss": 0.01695253886282444, "grad_norm": 0.11533050239086151, "global_step": 21316, "epoch": 201, "lr": 6.512069177067823e-05} +{"train_loss": 0.017093610018491745, "grad_norm": 0.1114993542432785, "global_step": 21317, "epoch": 201, "lr": 6.511786675071812e-05} +{"train_loss": 0.01288771815598011, "grad_norm": 0.06924577057361603, "global_step": 21318, "epoch": 201, "lr": 6.511504167764037e-05} +{"train_loss": 0.014140739105641842, "grad_norm": 0.11381103843450546, "global_step": 21319, "epoch": 201, "lr": 6.511221655145496e-05} +{"train_loss": 0.014208069071173668, "grad_norm": 0.054674528539180756, "global_step": 21320, "epoch": 201, "lr": 6.51093913721718e-05} +{"train_loss": 0.015487393364310265, "grad_norm": 0.06549041718244553, "global_step": 21321, "epoch": 201, "lr": 6.51065661398008e-05} +{"train_loss": 0.01349144522100687, "grad_norm": 0.14161819219589233, "global_step": 21322, "epoch": 201, "lr": 6.510374085435188e-05} +{"train_loss": 0.015994900837540627, "grad_norm": 0.14767827093601227, "global_step": 21323, "epoch": 201, "lr": 6.5100915515835e-05} +{"train_loss": 0.014681072905659676, "grad_norm": 0.08809882402420044, "global_step": 21324, "epoch": 201, "lr": 6.509809012426007e-05} +{"train_loss": 0.015028695575892925, "grad_norm": 0.07535557448863983, "global_step": 21325, "epoch": 201, "lr": 6.509526467963701e-05} +{"train_loss": 0.015993982553482056, "grad_norm": 0.05797708034515381, "global_step": 21326, "epoch": 201, "lr": 6.509243918197576e-05} +{"train_loss": 0.016110515221953392, "grad_norm": 0.19076010584831238, "global_step": 21327, "epoch": 201, "lr": 6.508961363128625e-05} +{"train_loss": 0.0174831822514534, "grad_norm": 0.07511738687753677, "global_step": 21328, "epoch": 201, "lr": 6.508678802757838e-05} +{"train_loss": 0.01699165627360344, "grad_norm": 0.09687338024377823, "global_step": 21329, "epoch": 201, "lr": 6.508396237086212e-05} +{"train_loss": 0.016438405960798264, "grad_norm": 0.08905792981386185, "global_step": 21330, "epoch": 201, "lr": 6.508113666114736e-05} +{"train_loss": 0.013711677864193916, "grad_norm": 0.06813809275627136, "global_step": 21331, "epoch": 201, "lr": 6.507831089844405e-05} +{"train_loss": 0.018622376024723053, "grad_norm": 0.08377137780189514, "global_step": 21332, "epoch": 201, "lr": 6.507548508276212e-05} +{"train_loss": 0.014462826773524284, "grad_norm": 0.07005809247493744, "global_step": 21333, "epoch": 201, "lr": 6.507265921411147e-05} +{"train_loss": 0.018935125321149826, "grad_norm": 0.13924263417720795, "global_step": 21334, "epoch": 201, "lr": 6.506983329250207e-05} +{"train_loss": 0.016189053654670715, "grad_norm": 0.07379860430955887, "global_step": 21335, "epoch": 201, "lr": 6.506700731794382e-05} +{"train_loss": 0.014689884148538113, "grad_norm": 0.10970525443553925, "global_step": 21336, "epoch": 201, "lr": 6.506418129044667e-05} +{"train_loss": 0.016433073207736015, "grad_norm": 0.07509703189134598, "global_step": 21337, "epoch": 201, "lr": 6.506135521002054e-05} +{"train_loss": 0.01909499801695347, "grad_norm": 0.11951065063476562, "global_step": 21338, "epoch": 201, "lr": 6.505852907667535e-05} +{"train_loss": 0.018979281187057495, "grad_norm": 0.1943691521883011, "global_step": 21339, "epoch": 201, "lr": 6.505570289042103e-05} +{"train_loss": 0.01712045818567276, "grad_norm": 0.06436923146247864, "global_step": 21340, "epoch": 201, "lr": 6.505287665126752e-05} +{"train_loss": 0.014969965443015099, "grad_norm": 0.06091941148042679, "global_step": 21341, "epoch": 201, "lr": 6.505005035922476e-05} +{"train_loss": 0.01854287087917328, "grad_norm": 0.04873059689998627, "global_step": 21342, "epoch": 201, "lr": 6.504722401430265e-05} +{"train_loss": 0.015205927193164825, "grad_norm": 0.09099682420492172, "global_step": 21343, "epoch": 201, "lr": 6.504439761651115e-05} +{"train_loss": 0.018301203846931458, "grad_norm": 0.1256212443113327, "global_step": 21344, "epoch": 201, "lr": 6.504157116586018e-05} +{"train_loss": 0.015044467523694038, "grad_norm": 0.0975731611251831, "global_step": 21345, "epoch": 201, "lr": 6.503874466235966e-05} +{"train_loss": 0.016428053379058838, "grad_norm": 0.09877954423427582, "global_step": 21346, "epoch": 201, "lr": 6.503591810601954e-05} +{"train_loss": 0.015583233907818794, "grad_norm": 0.10972363501787186, "global_step": 21347, "epoch": 201, "lr": 6.503309149684975e-05} +{"train_loss": 0.01500033400952816, "grad_norm": 0.06568720936775208, "global_step": 21348, "epoch": 201, "lr": 6.503026483486021e-05} +{"train_loss": 0.0201038159430027, "grad_norm": 0.059341445565223694, "global_step": 21349, "epoch": 201, "lr": 6.502743812006085e-05} +{"train_loss": 0.015687260776758194, "grad_norm": 0.0723738744854927, "global_step": 21350, "epoch": 201, "lr": 6.502461135246162e-05} +{"train_loss": 0.016334280371665955, "grad_norm": 0.061413686722517014, "global_step": 21351, "epoch": 201, "lr": 6.502178453207243e-05} +{"train_loss": 0.01783125102519989, "grad_norm": 0.059692054986953735, "global_step": 21352, "epoch": 201, "lr": 6.501895765890321e-05} +{"train_loss": 0.017075583338737488, "grad_norm": 0.059746235609054565, "global_step": 21353, "epoch": 201, "lr": 6.501613073296393e-05} +{"train_loss": 0.014862321317195892, "grad_norm": 0.09367616474628448, "global_step": 21354, "epoch": 201, "lr": 6.501330375426448e-05} +{"train_loss": 0.015490494668483734, "grad_norm": 0.04830572009086609, "global_step": 21355, "epoch": 201, "lr": 6.50104767228148e-05} +{"train_loss": 0.015900898724794388, "grad_norm": 0.18576385080814362, "global_step": 21356, "epoch": 201, "lr": 6.500764963862486e-05} +{"train_loss": 0.01362655684351921, "grad_norm": 0.08566837012767792, "global_step": 21357, "epoch": 201, "lr": 6.500482250170454e-05} +{"train_loss": 0.014480875805020332, "grad_norm": 0.0697820782661438, "global_step": 21358, "epoch": 201, "lr": 6.500199531206382e-05} +{"train_loss": 0.015899736434221268, "grad_norm": 0.06521345674991608, "global_step": 21359, "epoch": 201, "lr": 6.49991680697126e-05} +{"train_loss": 0.01534479483962059, "grad_norm": 0.07404521852731705, "global_step": 21360, "epoch": 201, "lr": 6.499634077466084e-05} +{"train_loss": 0.015294138342142105, "grad_norm": 0.05689575895667076, "global_step": 21361, "epoch": 201, "lr": 6.499351342691845e-05} +{"train_loss": 0.015913449227809906, "grad_norm": 0.10545463114976883, "global_step": 21362, "epoch": 201, "lr": 6.499068602649536e-05} +{"train_loss": 0.017732087522745132, "grad_norm": 0.058396562933921814, "global_step": 21363, "epoch": 201, "lr": 6.498785857340154e-05} +{"train_loss": 0.017424248158931732, "grad_norm": 0.09279318153858185, "global_step": 21364, "epoch": 201, "lr": 6.49850310676469e-05} +{"train_loss": 0.013283831998705864, "grad_norm": 0.06972178816795349, "global_step": 21365, "epoch": 201, "lr": 6.498220350924137e-05} +{"train_loss": 0.015675749629735947, "grad_norm": 0.07241201400756836, "global_step": 21366, "epoch": 201, "lr": 6.49793758981949e-05} +{"train_loss": 0.016007622703909874, "grad_norm": 0.09149014204740524, "global_step": 21367, "epoch": 201, "lr": 6.49765482345174e-05} +{"train_loss": 0.015788834542036057, "grad_norm": 0.0543319471180439, "global_step": 21368, "epoch": 201, "lr": 6.497372051821885e-05} +{"train_loss": 0.015443440526723862, "grad_norm": 0.0746123418211937, "global_step": 21369, "epoch": 201, "lr": 6.497089274930913e-05} +{"train_loss": 0.01579403690993786, "grad_norm": 0.058673735707998276, "global_step": 21370, "epoch": 201, "lr": 6.496806492779824e-05} +{"train_loss": 0.017884332686662674, "grad_norm": 0.13111786544322968, "global_step": 21371, "epoch": 201, "lr": 6.496523705369605e-05} +{"train_loss": 0.015667103230953217, "grad_norm": 0.0785430520772934, "global_step": 21372, "epoch": 201, "lr": 6.496240912701252e-05} +{"train_loss": 0.014238440431654453, "grad_norm": 0.06245988607406616, "global_step": 21373, "epoch": 201, "lr": 6.495958114775762e-05} +{"train_loss": 0.014525156468153, "grad_norm": 0.04670649394392967, "global_step": 21374, "epoch": 201, "lr": 6.495675311594123e-05} +{"train_loss": 0.017323816195130348, "grad_norm": 0.07560522109270096, "global_step": 21375, "epoch": 201, "lr": 6.495392503157333e-05} +{"train_loss": 0.015805140137672424, "grad_norm": 0.10411902517080307, "global_step": 21376, "epoch": 201, "lr": 6.495109689466384e-05} +{"train_loss": 0.014655635692179203, "grad_norm": 0.09196492284536362, "global_step": 21377, "epoch": 201, "lr": 6.494826870522268e-05} +{"train_loss": 0.015132194384932518, "grad_norm": 0.08899381756782532, "global_step": 21378, "epoch": 201, "lr": 6.494544046325983e-05} +{"train_loss": 0.016536805778741837, "grad_norm": 0.11210591346025467, "global_step": 21379, "epoch": 201, "lr": 6.494261216878519e-05} +{"train_loss": 0.017584092915058136, "grad_norm": 0.09976574033498764, "global_step": 21380, "epoch": 201, "lr": 6.493978382180871e-05} +{"train_loss": 0.015390102751553059, "grad_norm": 0.07654006779193878, "global_step": 21381, "epoch": 201, "lr": 6.493695542234031e-05} +{"train_loss": 0.013580761849880219, "grad_norm": 0.07519929111003876, "global_step": 21382, "epoch": 201, "lr": 6.493412697038997e-05} +{"train_loss": 0.015727246180176735, "grad_norm": 0.0877961814403534, "global_step": 21383, "epoch": 201, "lr": 6.493129846596759e-05} +{"train_loss": 0.01834283024072647, "grad_norm": 0.06424031406641006, "global_step": 21384, "epoch": 201, "lr": 6.492846990908312e-05} +{"train_loss": 0.01502871885895729, "grad_norm": 0.0933530181646347, "global_step": 21385, "epoch": 201, "lr": 6.492564129974652e-05} +{"train_loss": 0.01665792427957058, "grad_norm": 0.08121700584888458, "global_step": 21386, "epoch": 201, "lr": 6.492281263796767e-05} +{"train_loss": 0.016383634880185127, "grad_norm": 0.05314330756664276, "global_step": 21387, "epoch": 201, "lr": 6.491998392375658e-05} +{"train_loss": 0.015504607930779457, "grad_norm": 0.09212009608745575, "global_step": 21388, "epoch": 201, "lr": 6.491715515712313e-05} +{"train_loss": 0.01496264711022377, "grad_norm": 0.06736426800489426, "global_step": 21389, "epoch": 201, "lr": 6.49143263380773e-05} +{"train_loss": 0.01406531035900116, "grad_norm": 0.07907094061374664, "global_step": 21390, "epoch": 201, "lr": 6.491149746662901e-05} +{"train_loss": 0.01765201985836029, "grad_norm": 0.08505652844905853, "global_step": 21391, "epoch": 201, "lr": 6.490866854278821e-05} +{"train_loss": 0.014490293338894844, "grad_norm": 0.048949290066957474, "global_step": 21392, "epoch": 201, "lr": 6.490583956656482e-05} +{"train_loss": 0.017015554010868073, "grad_norm": 0.09388077259063721, "global_step": 21393, "epoch": 201, "lr": 6.490301053796878e-05} +{"train_loss": 0.014052314683794975, "grad_norm": 0.08188992738723755, "global_step": 21394, "epoch": 201, "lr": 6.490018145701008e-05} +{"train_loss": 0.01640496402978897, "grad_norm": 0.07696185261011124, "global_step": 21395, "epoch": 201, "lr": 6.48973523236986e-05} +{"train_loss": 0.017194397747516632, "grad_norm": 0.13627266883850098, "global_step": 21396, "epoch": 201, "lr": 6.48945231380443e-05} +{"train_loss": 0.01849103718996048, "grad_norm": 0.06377949565649033, "global_step": 21397, "epoch": 201, "lr": 6.489169390005712e-05} +{"train_loss": 0.01738852635025978, "grad_norm": 0.09487320482730865, "global_step": 21398, "epoch": 201, "lr": 6.488886460974702e-05} +{"train_loss": 0.0166871789842844, "grad_norm": 0.07212460786104202, "global_step": 21399, "epoch": 201, "lr": 6.48860352671239e-05} +{"train_loss": 0.015879154205322266, "grad_norm": 0.08523281663656235, "global_step": 21400, "epoch": 201, "lr": 6.488320587219776e-05} +{"train_loss": 0.017078805714845657, "grad_norm": 0.09732378274202347, "global_step": 21401, "epoch": 201, "lr": 6.488037642497847e-05} +{"train_loss": 0.014022724702954292, "grad_norm": 0.07698561251163483, "global_step": 21402, "epoch": 201, "lr": 6.487754692547603e-05} +{"train_loss": 0.016063407063484192, "grad_norm": 0.05964614823460579, "global_step": 21403, "epoch": 201, "lr": 6.487471737370035e-05} +{"train_loss": 0.015005938708782196, "grad_norm": 0.07391974329948425, "global_step": 21404, "epoch": 201, "lr": 6.48718877696614e-05} +{"train_loss": 0.01618230529129505, "grad_norm": 0.07108127325773239, "global_step": 21405, "epoch": 201, "lr": 6.486905811336909e-05} +{"train_loss": 0.01638980768620968, "grad_norm": 0.06903880089521408, "global_step": 21406, "epoch": 201, "lr": 6.486622840483339e-05} +{"train_loss": 0.01834964007139206, "grad_norm": 0.09905563294887543, "global_step": 21407, "epoch": 201, "lr": 6.486339864406422e-05} +{"train_loss": 0.01533487904816866, "grad_norm": 0.08865121752023697, "global_step": 21408, "epoch": 201, "lr": 6.48605688310715e-05} +{"train_loss": 0.015413048677146435, "grad_norm": 0.059429146349430084, "global_step": 21409, "epoch": 201, "lr": 6.485773896586524e-05} +{"train_loss": 0.016244009137153625, "grad_norm": 0.07369918376207352, "global_step": 21410, "epoch": 201, "lr": 6.485490904845534e-05} +{"train_loss": 0.016001580133682716, "grad_norm": 0.10460370779037476, "global_step": 21411, "epoch": 201, "lr": 6.485207907885175e-05, "val_loss": 0.018954282626509666} +{"train_loss": 0.015462467446923256, "grad_norm": 0.057500723749399185, "global_step": 21412, "epoch": 202, "lr": 6.484924905706442e-05} +{"train_loss": 0.01638638973236084, "grad_norm": 0.06585859507322311, "global_step": 21413, "epoch": 202, "lr": 6.484641898310327e-05} +{"train_loss": 0.015959925949573517, "grad_norm": 0.05442315712571144, "global_step": 21414, "epoch": 202, "lr": 6.484358885697826e-05} +{"train_loss": 0.0156920924782753, "grad_norm": 0.04761599004268646, "global_step": 21415, "epoch": 202, "lr": 6.484075867869935e-05} +{"train_loss": 0.015436754561960697, "grad_norm": 0.05535956472158432, "global_step": 21416, "epoch": 202, "lr": 6.483792844827646e-05} +{"train_loss": 0.016381733119487762, "grad_norm": 0.06090140342712402, "global_step": 21417, "epoch": 202, "lr": 6.483509816571954e-05} +{"train_loss": 0.018886318430304527, "grad_norm": 0.1027078777551651, "global_step": 21418, "epoch": 202, "lr": 6.483226783103855e-05} +{"train_loss": 0.015480884350836277, "grad_norm": 0.08369351178407669, "global_step": 21419, "epoch": 202, "lr": 6.482943744424341e-05} +{"train_loss": 0.01556581910699606, "grad_norm": 0.05926113948225975, "global_step": 21420, "epoch": 202, "lr": 6.482660700534408e-05} +{"train_loss": 0.017247464507818222, "grad_norm": 0.0677654892206192, "global_step": 21421, "epoch": 202, "lr": 6.48237765143505e-05} +{"train_loss": 0.015009480528533459, "grad_norm": 0.04628843441605568, "global_step": 21422, "epoch": 202, "lr": 6.482094597127261e-05} +{"train_loss": 0.015590103343129158, "grad_norm": 0.09519051760435104, "global_step": 21423, "epoch": 202, "lr": 6.481811537612038e-05} +{"train_loss": 0.017407182604074478, "grad_norm": 0.07184399664402008, "global_step": 21424, "epoch": 202, "lr": 6.48152847289037e-05} +{"train_loss": 0.01586657017469406, "grad_norm": 0.06583739072084427, "global_step": 21425, "epoch": 202, "lr": 6.481245402963258e-05} +{"train_loss": 0.015912890434265137, "grad_norm": 0.06974449008703232, "global_step": 21426, "epoch": 202, "lr": 6.480962327831693e-05} +{"train_loss": 0.015830133110284805, "grad_norm": 0.1041688546538353, "global_step": 21427, "epoch": 202, "lr": 6.48067924749667e-05} +{"train_loss": 0.01823403686285019, "grad_norm": 0.067860908806324, "global_step": 21428, "epoch": 202, "lr": 6.480396161959185e-05} +{"train_loss": 0.016812553629279137, "grad_norm": 0.05889124050736427, "global_step": 21429, "epoch": 202, "lr": 6.480113071220232e-05} +{"train_loss": 0.015032891184091568, "grad_norm": 0.14034238457679749, "global_step": 21430, "epoch": 202, "lr": 6.479829975280805e-05} +{"train_loss": 0.01549453753978014, "grad_norm": 0.13638654351234436, "global_step": 21431, "epoch": 202, "lr": 6.479546874141899e-05} +{"train_loss": 0.01619141921401024, "grad_norm": 0.07192537188529968, "global_step": 21432, "epoch": 202, "lr": 6.479263767804508e-05} +{"train_loss": 0.015558039769530296, "grad_norm": 0.062103211879730225, "global_step": 21433, "epoch": 202, "lr": 6.478980656269628e-05} +{"train_loss": 0.014131896197795868, "grad_norm": 0.05774262547492981, "global_step": 21434, "epoch": 202, "lr": 6.478697539538253e-05} +{"train_loss": 0.015421322546899319, "grad_norm": 0.1236925721168518, "global_step": 21435, "epoch": 202, "lr": 6.478414417611378e-05} +{"train_loss": 0.015622572973370552, "grad_norm": 0.08309481292963028, "global_step": 21436, "epoch": 202, "lr": 6.478131290489998e-05} +{"train_loss": 0.014339148998260498, "grad_norm": 0.05638117715716362, "global_step": 21437, "epoch": 202, "lr": 6.477848158175106e-05} +{"train_loss": 0.0159109216183424, "grad_norm": 0.1187235563993454, "global_step": 21438, "epoch": 202, "lr": 6.477565020667701e-05} +{"train_loss": 0.018037088215351105, "grad_norm": 0.057824280112981796, "global_step": 21439, "epoch": 202, "lr": 6.477281877968772e-05} +{"train_loss": 0.014871444553136826, "grad_norm": 0.05539052188396454, "global_step": 21440, "epoch": 202, "lr": 6.476998730079319e-05} +{"train_loss": 0.01442137360572815, "grad_norm": 0.0633869618177414, "global_step": 21441, "epoch": 202, "lr": 6.476715577000335e-05} +{"train_loss": 0.017292272299528122, "grad_norm": 0.09053564816713333, "global_step": 21442, "epoch": 202, "lr": 6.476432418732814e-05} +{"train_loss": 0.01515979040414095, "grad_norm": 0.06047860532999039, "global_step": 21443, "epoch": 202, "lr": 6.476149255277751e-05} +{"train_loss": 0.01750163733959198, "grad_norm": 0.11973240226507187, "global_step": 21444, "epoch": 202, "lr": 6.475866086636142e-05} +{"train_loss": 0.01449779886752367, "grad_norm": 0.06434234231710434, "global_step": 21445, "epoch": 202, "lr": 6.475582912808982e-05} +{"train_loss": 0.0170974712818861, "grad_norm": 0.1115056499838829, "global_step": 21446, "epoch": 202, "lr": 6.475299733797264e-05} +{"train_loss": 0.014894017949700356, "grad_norm": 0.05733944848179817, "global_step": 21447, "epoch": 202, "lr": 6.475016549601986e-05} +{"train_loss": 0.013842528685927391, "grad_norm": 0.05426478758454323, "global_step": 21448, "epoch": 202, "lr": 6.47473336022414e-05} +{"train_loss": 0.01665215939283371, "grad_norm": 0.06484545767307281, "global_step": 21449, "epoch": 202, "lr": 6.474450165664721e-05} +{"train_loss": 0.01415847335010767, "grad_norm": 0.0910826176404953, "global_step": 21450, "epoch": 202, "lr": 6.47416696592473e-05} +{"train_loss": 0.01319923996925354, "grad_norm": 0.06723545491695404, "global_step": 21451, "epoch": 202, "lr": 6.473883761005154e-05} +{"train_loss": 0.01709941029548645, "grad_norm": 0.08531476557254791, "global_step": 21452, "epoch": 202, "lr": 6.473600550906991e-05} +{"train_loss": 0.014960873872041702, "grad_norm": 0.08349650353193283, "global_step": 21453, "epoch": 202, "lr": 6.473317335631238e-05} +{"train_loss": 0.017034996300935745, "grad_norm": 0.05751029774546623, "global_step": 21454, "epoch": 202, "lr": 6.473034115178888e-05} +{"train_loss": 0.015215697698295116, "grad_norm": 0.08036304265260696, "global_step": 21455, "epoch": 202, "lr": 6.472750889550936e-05} +{"train_loss": 0.012407617643475533, "grad_norm": 0.07856106013059616, "global_step": 21456, "epoch": 202, "lr": 6.472467658748379e-05} +{"train_loss": 0.017037034034729004, "grad_norm": 0.06747186183929443, "global_step": 21457, "epoch": 202, "lr": 6.47218442277221e-05} +{"train_loss": 0.01545846275985241, "grad_norm": 0.055238742381334305, "global_step": 21458, "epoch": 202, "lr": 6.471901181623426e-05} +{"train_loss": 0.013721290975809097, "grad_norm": 0.05033999681472778, "global_step": 21459, "epoch": 202, "lr": 6.471617935303021e-05} +{"train_loss": 0.01621266081929207, "grad_norm": 0.078460693359375, "global_step": 21460, "epoch": 202, "lr": 6.471334683811991e-05} +{"train_loss": 0.01902097277343273, "grad_norm": 0.12144363671541214, "global_step": 21461, "epoch": 202, "lr": 6.47105142715133e-05} +{"train_loss": 0.01407137606292963, "grad_norm": 0.05304180085659027, "global_step": 21462, "epoch": 202, "lr": 6.470768165322035e-05} +{"train_loss": 0.01663801819086075, "grad_norm": 0.06846848130226135, "global_step": 21463, "epoch": 202, "lr": 6.470484898325099e-05} +{"train_loss": 0.01854550652205944, "grad_norm": 0.06746425479650497, "global_step": 21464, "epoch": 202, "lr": 6.47020162616152e-05} +{"train_loss": 0.014314243569970131, "grad_norm": 0.06909484416246414, "global_step": 21465, "epoch": 202, "lr": 6.469918348832292e-05} +{"train_loss": 0.01547315064817667, "grad_norm": 0.12938667833805084, "global_step": 21466, "epoch": 202, "lr": 6.469635066338409e-05} +{"train_loss": 0.01629922166466713, "grad_norm": 0.16759127378463745, "global_step": 21467, "epoch": 202, "lr": 6.469351778680868e-05} +{"train_loss": 0.018073461949825287, "grad_norm": 0.12042158097028732, "global_step": 21468, "epoch": 202, "lr": 6.469068485860663e-05} +{"train_loss": 0.016814913600683212, "grad_norm": 0.061034925282001495, "global_step": 21469, "epoch": 202, "lr": 6.468785187878791e-05} +{"train_loss": 0.01635107398033142, "grad_norm": 0.09805139154195786, "global_step": 21470, "epoch": 202, "lr": 6.468501884736248e-05} +{"train_loss": 0.01754595898091793, "grad_norm": 0.06567899137735367, "global_step": 21471, "epoch": 202, "lr": 6.468218576434026e-05} +{"train_loss": 0.0172981359064579, "grad_norm": 0.1024932786822319, "global_step": 21472, "epoch": 202, "lr": 6.467935262973122e-05} +{"train_loss": 0.017259076237678528, "grad_norm": 0.07267281413078308, "global_step": 21473, "epoch": 202, "lr": 6.467651944354534e-05} +{"train_loss": 0.015130904503166676, "grad_norm": 0.07941336184740067, "global_step": 21474, "epoch": 202, "lr": 6.467368620579254e-05} +{"train_loss": 0.014946908690035343, "grad_norm": 0.0612790510058403, "global_step": 21475, "epoch": 202, "lr": 6.467085291648279e-05} +{"train_loss": 0.01418511476367712, "grad_norm": 0.047058768570423126, "global_step": 21476, "epoch": 202, "lr": 6.466801957562604e-05} +{"train_loss": 0.017458217218518257, "grad_norm": 0.07305118441581726, "global_step": 21477, "epoch": 202, "lr": 6.466518618323226e-05} +{"train_loss": 0.015339663252234459, "grad_norm": 0.1454271674156189, "global_step": 21478, "epoch": 202, "lr": 6.466235273931138e-05} +{"train_loss": 0.01624375395476818, "grad_norm": 0.042629875242710114, "global_step": 21479, "epoch": 202, "lr": 6.465951924387338e-05} +{"train_loss": 0.015336902812123299, "grad_norm": 0.1195022240281105, "global_step": 21480, "epoch": 202, "lr": 6.465668569692819e-05} +{"train_loss": 0.016808999702334404, "grad_norm": 0.07858670502901077, "global_step": 21481, "epoch": 202, "lr": 6.465385209848578e-05} +{"train_loss": 0.01747378520667553, "grad_norm": 0.07200855016708374, "global_step": 21482, "epoch": 202, "lr": 6.465101844855613e-05} +{"train_loss": 0.015160888433456421, "grad_norm": 0.06730309873819351, "global_step": 21483, "epoch": 202, "lr": 6.464818474714916e-05} +{"train_loss": 0.01718999072909355, "grad_norm": 0.11596277356147766, "global_step": 21484, "epoch": 202, "lr": 6.464535099427484e-05} +{"train_loss": 0.01646510884165764, "grad_norm": 0.09188637882471085, "global_step": 21485, "epoch": 202, "lr": 6.464251718994312e-05} +{"train_loss": 0.016089582815766335, "grad_norm": 0.08312776684761047, "global_step": 21486, "epoch": 202, "lr": 6.463968333416397e-05} +{"train_loss": 0.015939349308609962, "grad_norm": 0.0648253783583641, "global_step": 21487, "epoch": 202, "lr": 6.463684942694733e-05} +{"train_loss": 0.01642794907093048, "grad_norm": 0.10309121012687683, "global_step": 21488, "epoch": 202, "lr": 6.463401546830317e-05} +{"train_loss": 0.017422256991267204, "grad_norm": 0.0643516555428505, "global_step": 21489, "epoch": 202, "lr": 6.463118145824146e-05} +{"train_loss": 0.017443954944610596, "grad_norm": 0.14602304995059967, "global_step": 21490, "epoch": 202, "lr": 6.462834739677211e-05} +{"train_loss": 0.016439387574791908, "grad_norm": 0.1173785999417305, "global_step": 21491, "epoch": 202, "lr": 6.462551328390514e-05} +{"train_loss": 0.014133498072624207, "grad_norm": 0.056702833622694016, "global_step": 21492, "epoch": 202, "lr": 6.462267911965045e-05} +{"train_loss": 0.014352118596434593, "grad_norm": 0.0967480018734932, "global_step": 21493, "epoch": 202, "lr": 6.461984490401803e-05} +{"train_loss": 0.016837934032082558, "grad_norm": 0.06126921996474266, "global_step": 21494, "epoch": 202, "lr": 6.461701063701786e-05} +{"train_loss": 0.015906982123851776, "grad_norm": 0.09833311289548874, "global_step": 21495, "epoch": 202, "lr": 6.461417631865984e-05} +{"train_loss": 0.015685204416513443, "grad_norm": 0.0838807001709938, "global_step": 21496, "epoch": 202, "lr": 6.461134194895397e-05} +{"train_loss": 0.01359393261373043, "grad_norm": 0.11328837275505066, "global_step": 21497, "epoch": 202, "lr": 6.46085075279102e-05} +{"train_loss": 0.014868560247123241, "grad_norm": 0.08149579912424088, "global_step": 21498, "epoch": 202, "lr": 6.460567305553849e-05} +{"train_loss": 0.014943255111575127, "grad_norm": 0.06988420337438583, "global_step": 21499, "epoch": 202, "lr": 6.460283853184879e-05} +{"train_loss": 0.017411008477211, "grad_norm": 0.06594877690076828, "global_step": 21500, "epoch": 202, "lr": 6.460000395685108e-05} +{"train_loss": 0.017130345106124878, "grad_norm": 0.07695065438747406, "global_step": 21501, "epoch": 202, "lr": 6.459716933055529e-05} +{"train_loss": 0.017542235553264618, "grad_norm": 0.1231493204832077, "global_step": 21502, "epoch": 202, "lr": 6.45943346529714e-05} +{"train_loss": 0.016190283000469208, "grad_norm": 0.1572873592376709, "global_step": 21503, "epoch": 202, "lr": 6.459149992410937e-05} +{"train_loss": 0.014556790702044964, "grad_norm": 0.08814068138599396, "global_step": 21504, "epoch": 202, "lr": 6.458866514397916e-05} +{"train_loss": 0.016724027693271637, "grad_norm": 0.12806902825832367, "global_step": 21505, "epoch": 202, "lr": 6.45858303125907e-05} +{"train_loss": 0.016071762889623642, "grad_norm": 0.1077137216925621, "global_step": 21506, "epoch": 202, "lr": 6.458299542995401e-05} +{"train_loss": 0.015560774132609367, "grad_norm": 0.051547884941101074, "global_step": 21507, "epoch": 202, "lr": 6.458016049607897e-05} +{"train_loss": 0.015542866662144661, "grad_norm": 0.10609371960163116, "global_step": 21508, "epoch": 202, "lr": 6.457732551097562e-05} +{"train_loss": 0.01691776141524315, "grad_norm": 0.10442269593477249, "global_step": 21509, "epoch": 202, "lr": 6.457449047465389e-05} +{"train_loss": 0.01720288023352623, "grad_norm": 0.07277772575616837, "global_step": 21510, "epoch": 202, "lr": 6.457165538712372e-05} +{"train_loss": 0.01743936538696289, "grad_norm": 0.0884835496544838, "global_step": 21511, "epoch": 202, "lr": 6.456882024839511e-05} +{"train_loss": 0.014659060165286064, "grad_norm": 0.07829990983009338, "global_step": 21512, "epoch": 202, "lr": 6.4565985058478e-05} +{"train_loss": 0.016481339931488037, "grad_norm": 0.061434097588062286, "global_step": 21513, "epoch": 202, "lr": 6.456314981738233e-05} +{"train_loss": 0.016122328117489815, "grad_norm": 0.10450029373168945, "global_step": 21514, "epoch": 202, "lr": 6.456031452511811e-05} +{"train_loss": 0.013766300864517689, "grad_norm": 0.08454945683479309, "global_step": 21515, "epoch": 202, "lr": 6.455747918169528e-05} +{"train_loss": 0.01630299910902977, "grad_norm": 0.08918001502752304, "global_step": 21516, "epoch": 202, "lr": 6.455464378712378e-05} +{"train_loss": 0.015953752115579707, "grad_norm": 0.1330401599407196, "global_step": 21517, "epoch": 202, "lr": 6.455180834141359e-05, "val_loss": 0.019598357379436493} +{"train_loss": 0.01584514044225216, "grad_norm": 0.07923362404108047, "global_step": 21518, "epoch": 203, "lr": 6.454897284457469e-05} +{"train_loss": 0.014595653861761093, "grad_norm": 0.0629870668053627, "global_step": 21519, "epoch": 203, "lr": 6.454613729661702e-05} +{"train_loss": 0.01494402252137661, "grad_norm": 0.05662769079208374, "global_step": 21520, "epoch": 203, "lr": 6.454330169755055e-05} +{"train_loss": 0.014764553867280483, "grad_norm": 0.09297079592943192, "global_step": 21521, "epoch": 203, "lr": 6.454046604738524e-05} +{"train_loss": 0.015156070701777935, "grad_norm": 0.057022254914045334, "global_step": 21522, "epoch": 203, "lr": 6.453763034613105e-05} +{"train_loss": 0.014925435185432434, "grad_norm": 0.09604286402463913, "global_step": 21523, "epoch": 203, "lr": 6.453479459379796e-05} +{"train_loss": 0.014141710475087166, "grad_norm": 0.05812866985797882, "global_step": 21524, "epoch": 203, "lr": 6.453195879039591e-05} +{"train_loss": 0.016452418640255928, "grad_norm": 0.07108379155397415, "global_step": 21525, "epoch": 203, "lr": 6.452912293593489e-05} +{"train_loss": 0.017292268574237823, "grad_norm": 0.0697195902466774, "global_step": 21526, "epoch": 203, "lr": 6.452628703042484e-05} +{"train_loss": 0.017639435827732086, "grad_norm": 0.11865561455488205, "global_step": 21527, "epoch": 203, "lr": 6.452345107387574e-05} +{"train_loss": 0.014883143827319145, "grad_norm": 0.09998670220375061, "global_step": 21528, "epoch": 203, "lr": 6.452061506629755e-05} +{"train_loss": 0.013800316490232944, "grad_norm": 0.05312628671526909, "global_step": 21529, "epoch": 203, "lr": 6.451777900770024e-05} +{"train_loss": 0.017336508259177208, "grad_norm": 0.08039401471614838, "global_step": 21530, "epoch": 203, "lr": 6.451494289809376e-05} +{"train_loss": 0.016880076378583908, "grad_norm": 0.09038424491882324, "global_step": 21531, "epoch": 203, "lr": 6.451210673748808e-05} +{"train_loss": 0.017326869070529938, "grad_norm": 0.0772567167878151, "global_step": 21532, "epoch": 203, "lr": 6.450927052589318e-05} +{"train_loss": 0.017340335994958878, "grad_norm": 0.07367836683988571, "global_step": 21533, "epoch": 203, "lr": 6.4506434263319e-05} +{"train_loss": 0.017574720084667206, "grad_norm": 0.05662863329052925, "global_step": 21534, "epoch": 203, "lr": 6.450359794977553e-05} +{"train_loss": 0.016992121934890747, "grad_norm": 0.09283188730478287, "global_step": 21535, "epoch": 203, "lr": 6.450076158527272e-05} +{"train_loss": 0.015290139243006706, "grad_norm": 0.06987079232931137, "global_step": 21536, "epoch": 203, "lr": 6.449792516982053e-05} +{"train_loss": 0.01439162902534008, "grad_norm": 0.06570541858673096, "global_step": 21537, "epoch": 203, "lr": 6.449508870342895e-05} +{"train_loss": 0.0159502811729908, "grad_norm": 0.1101289913058281, "global_step": 21538, "epoch": 203, "lr": 6.449225218610791e-05} +{"train_loss": 0.015376213937997818, "grad_norm": 0.0607445202767849, "global_step": 21539, "epoch": 203, "lr": 6.448941561786742e-05} +{"train_loss": 0.01845155656337738, "grad_norm": 0.11190222203731537, "global_step": 21540, "epoch": 203, "lr": 6.448657899871741e-05} +{"train_loss": 0.014112921431660652, "grad_norm": 0.056179456412792206, "global_step": 21541, "epoch": 203, "lr": 6.448374232866788e-05} +{"train_loss": 0.016110168769955635, "grad_norm": 0.11137726902961731, "global_step": 21542, "epoch": 203, "lr": 6.448090560772877e-05} +{"train_loss": 0.0149493757635355, "grad_norm": 0.06117734685540199, "global_step": 21543, "epoch": 203, "lr": 6.447806883591005e-05} +{"train_loss": 0.016016550362110138, "grad_norm": 0.10114637762308121, "global_step": 21544, "epoch": 203, "lr": 6.44752320132217e-05} +{"train_loss": 0.015621882863342762, "grad_norm": 0.051747631281614304, "global_step": 21545, "epoch": 203, "lr": 6.447239513967368e-05} +{"train_loss": 0.01929028518497944, "grad_norm": 0.07922844588756561, "global_step": 21546, "epoch": 203, "lr": 6.446955821527593e-05} +{"train_loss": 0.015752729028463364, "grad_norm": 0.07124051451683044, "global_step": 21547, "epoch": 203, "lr": 6.446672124003849e-05} +{"train_loss": 0.019533423706889153, "grad_norm": 0.06507008522748947, "global_step": 21548, "epoch": 203, "lr": 6.446388421397126e-05} +{"train_loss": 0.01701178401708603, "grad_norm": 0.07381647825241089, "global_step": 21549, "epoch": 203, "lr": 6.446104713708423e-05} +{"train_loss": 0.016658691689372063, "grad_norm": 0.06887658685445786, "global_step": 21550, "epoch": 203, "lr": 6.445821000938739e-05} +{"train_loss": 0.015862256288528442, "grad_norm": 0.05373270809650421, "global_step": 21551, "epoch": 203, "lr": 6.445537283089065e-05} +{"train_loss": 0.01803719811141491, "grad_norm": 0.062467072159051895, "global_step": 21552, "epoch": 203, "lr": 6.445253560160404e-05} +{"train_loss": 0.016206517815589905, "grad_norm": 0.10393025726079941, "global_step": 21553, "epoch": 203, "lr": 6.44496983215375e-05} +{"train_loss": 0.016257859766483307, "grad_norm": 0.057207200676202774, "global_step": 21554, "epoch": 203, "lr": 6.4446860990701e-05} +{"train_loss": 0.014364526607096195, "grad_norm": 0.06659843027591705, "global_step": 21555, "epoch": 203, "lr": 6.444402360910453e-05} +{"train_loss": 0.01567230373620987, "grad_norm": 0.049792539328336716, "global_step": 21556, "epoch": 203, "lr": 6.444118617675803e-05} +{"train_loss": 0.015939148142933846, "grad_norm": 0.10516174882650375, "global_step": 21557, "epoch": 203, "lr": 6.44383486936715e-05} +{"train_loss": 0.01581992208957672, "grad_norm": 0.05567120388150215, "global_step": 21558, "epoch": 203, "lr": 6.443551115985484e-05} +{"train_loss": 0.01772739179432392, "grad_norm": 0.09565255790948868, "global_step": 21559, "epoch": 203, "lr": 6.443267357531813e-05} +{"train_loss": 0.01585179753601551, "grad_norm": 0.08295264840126038, "global_step": 21560, "epoch": 203, "lr": 6.442983594007125e-05} +{"train_loss": 0.017226526513695717, "grad_norm": 0.06030775606632233, "global_step": 21561, "epoch": 203, "lr": 6.44269982541242e-05} +{"train_loss": 0.014457335695624352, "grad_norm": 0.060588233172893524, "global_step": 21562, "epoch": 203, "lr": 6.442416051748699e-05} +{"train_loss": 0.01565692573785782, "grad_norm": 0.07555330544710159, "global_step": 21563, "epoch": 203, "lr": 6.44213227301695e-05} +{"train_loss": 0.01650051958858967, "grad_norm": 0.08843399584293365, "global_step": 21564, "epoch": 203, "lr": 6.441848489218179e-05} +{"train_loss": 0.014225096441805363, "grad_norm": 0.046179868280887604, "global_step": 21565, "epoch": 203, "lr": 6.441564700353378e-05} +{"train_loss": 0.015873026102781296, "grad_norm": 0.051494281738996506, "global_step": 21566, "epoch": 203, "lr": 6.441280906423545e-05} +{"train_loss": 0.016130078583955765, "grad_norm": 0.06396668404340744, "global_step": 21567, "epoch": 203, "lr": 6.44099710742968e-05} +{"train_loss": 0.014397567138075829, "grad_norm": 0.07056460529565811, "global_step": 21568, "epoch": 203, "lr": 6.440713303372776e-05} +{"train_loss": 0.015121841803193092, "grad_norm": 0.1425853967666626, "global_step": 21569, "epoch": 203, "lr": 6.440429494253832e-05} +{"train_loss": 0.016273221001029015, "grad_norm": 0.11412382870912552, "global_step": 21570, "epoch": 203, "lr": 6.440145680073847e-05} +{"train_loss": 0.01499250065535307, "grad_norm": 0.1222250908613205, "global_step": 21571, "epoch": 203, "lr": 6.439861860833814e-05} +{"train_loss": 0.013988202437758446, "grad_norm": 0.11148333549499512, "global_step": 21572, "epoch": 203, "lr": 6.439578036534734e-05} +{"train_loss": 0.017582692205905914, "grad_norm": 0.07792297005653381, "global_step": 21573, "epoch": 203, "lr": 6.439294207177603e-05} +{"train_loss": 0.016509156674146652, "grad_norm": 0.10137059539556503, "global_step": 21574, "epoch": 203, "lr": 6.439010372763417e-05} +{"train_loss": 0.016214940696954727, "grad_norm": 0.05990932136774063, "global_step": 21575, "epoch": 203, "lr": 6.438726533293175e-05} +{"train_loss": 0.015154873952269554, "grad_norm": 0.05014864727854729, "global_step": 21576, "epoch": 203, "lr": 6.438442688767874e-05} +{"train_loss": 0.01612897962331772, "grad_norm": 0.06285735964775085, "global_step": 21577, "epoch": 203, "lr": 6.438158839188509e-05} +{"train_loss": 0.016991160809993744, "grad_norm": 0.10269410908222198, "global_step": 21578, "epoch": 203, "lr": 6.437874984556082e-05} +{"train_loss": 0.015082264319062233, "grad_norm": 0.06343261897563934, "global_step": 21579, "epoch": 203, "lr": 6.437591124871587e-05} +{"train_loss": 0.01587173528969288, "grad_norm": 0.10063444077968597, "global_step": 21580, "epoch": 203, "lr": 6.43730726013602e-05} +{"train_loss": 0.017175614833831787, "grad_norm": 0.05653528496623039, "global_step": 21581, "epoch": 203, "lr": 6.437023390350383e-05} +{"train_loss": 0.015192466787993908, "grad_norm": 0.10226460546255112, "global_step": 21582, "epoch": 203, "lr": 6.436739515515669e-05} +{"train_loss": 0.015841282904148102, "grad_norm": 0.13267336785793304, "global_step": 21583, "epoch": 203, "lr": 6.436455635632877e-05} +{"train_loss": 0.016218744218349457, "grad_norm": 0.19258631765842438, "global_step": 21584, "epoch": 203, "lr": 6.436171750703006e-05} +{"train_loss": 0.015566810965538025, "grad_norm": 0.14371581375598907, "global_step": 21585, "epoch": 203, "lr": 6.435887860727051e-05} +{"train_loss": 0.015325730666518211, "grad_norm": 0.12039179354906082, "global_step": 21586, "epoch": 203, "lr": 6.435603965706011e-05} +{"train_loss": 0.017311979085206985, "grad_norm": 0.11402911692857742, "global_step": 21587, "epoch": 203, "lr": 6.435320065640882e-05} +{"train_loss": 0.015477830544114113, "grad_norm": 0.1094517782330513, "global_step": 21588, "epoch": 203, "lr": 6.435036160532663e-05} +{"train_loss": 0.014132007956504822, "grad_norm": 0.08761093020439148, "global_step": 21589, "epoch": 203, "lr": 6.434752250382352e-05} +{"train_loss": 0.014416486024856567, "grad_norm": 0.0776248499751091, "global_step": 21590, "epoch": 203, "lr": 6.434468335190943e-05} +{"train_loss": 0.01685851253569126, "grad_norm": 0.07832586020231247, "global_step": 21591, "epoch": 203, "lr": 6.434184414959441e-05} +{"train_loss": 0.01579596847295761, "grad_norm": 0.09300152957439423, "global_step": 21592, "epoch": 203, "lr": 6.433900489688835e-05} +{"train_loss": 0.018490757793188095, "grad_norm": 0.18218566477298737, "global_step": 21593, "epoch": 203, "lr": 6.433616559380127e-05} +{"train_loss": 0.01614108681678772, "grad_norm": 0.16777069866657257, "global_step": 21594, "epoch": 203, "lr": 6.433332624034313e-05} +{"train_loss": 0.0193138737231493, "grad_norm": 0.21075640618801117, "global_step": 21595, "epoch": 203, "lr": 6.433048683652392e-05} +{"train_loss": 0.016202347353100777, "grad_norm": 0.1979319453239441, "global_step": 21596, "epoch": 203, "lr": 6.432764738235361e-05} +{"train_loss": 0.017002275213599205, "grad_norm": 0.06994806975126266, "global_step": 21597, "epoch": 203, "lr": 6.432480787784218e-05} +{"train_loss": 0.016552245244383812, "grad_norm": 0.060895033180713654, "global_step": 21598, "epoch": 203, "lr": 6.432196832299962e-05} +{"train_loss": 0.016428545117378235, "grad_norm": 0.08020670711994171, "global_step": 21599, "epoch": 203, "lr": 6.431912871783586e-05} +{"train_loss": 0.016323141753673553, "grad_norm": 0.10902684181928635, "global_step": 21600, "epoch": 203, "lr": 6.431628906236095e-05} +{"train_loss": 0.016341492533683777, "grad_norm": 0.10374227911233902, "global_step": 21601, "epoch": 203, "lr": 6.43134493565848e-05} +{"train_loss": 0.014257357455790043, "grad_norm": 0.09888190776109695, "global_step": 21602, "epoch": 203, "lr": 6.431060960051742e-05} +{"train_loss": 0.015333970077335835, "grad_norm": 0.08006808161735535, "global_step": 21603, "epoch": 203, "lr": 6.430776979416878e-05} +{"train_loss": 0.016074413433670998, "grad_norm": 0.10816717147827148, "global_step": 21604, "epoch": 203, "lr": 6.430492993754884e-05} +{"train_loss": 0.016484733670949936, "grad_norm": 0.10384082049131393, "global_step": 21605, "epoch": 203, "lr": 6.430209003066762e-05} +{"train_loss": 0.015027709305286407, "grad_norm": 0.07823915779590607, "global_step": 21606, "epoch": 203, "lr": 6.429925007353508e-05} +{"train_loss": 0.014971070922911167, "grad_norm": 0.07799401879310608, "global_step": 21607, "epoch": 203, "lr": 6.429641006616119e-05} +{"train_loss": 0.015291777439415455, "grad_norm": 0.08304505050182343, "global_step": 21608, "epoch": 203, "lr": 6.429357000855592e-05} +{"train_loss": 0.016430683434009552, "grad_norm": 0.06371433287858963, "global_step": 21609, "epoch": 203, "lr": 6.429072990072926e-05} +{"train_loss": 0.015687737613916397, "grad_norm": 0.10913296788930893, "global_step": 21610, "epoch": 203, "lr": 6.428788974269122e-05} +{"train_loss": 0.0165800079703331, "grad_norm": 0.13337470591068268, "global_step": 21611, "epoch": 203, "lr": 6.428504953445171e-05} +{"train_loss": 0.013761820271611214, "grad_norm": 0.12033949792385101, "global_step": 21612, "epoch": 203, "lr": 6.428220927602077e-05} +{"train_loss": 0.016437672078609467, "grad_norm": 0.11405539512634277, "global_step": 21613, "epoch": 203, "lr": 6.427936896740836e-05} +{"train_loss": 0.016291998326778412, "grad_norm": 0.09340351074934006, "global_step": 21614, "epoch": 203, "lr": 6.427652860862444e-05} +{"train_loss": 0.015859173610806465, "grad_norm": 0.04866351932287216, "global_step": 21615, "epoch": 203, "lr": 6.427368819967903e-05} +{"train_loss": 0.016555296257138252, "grad_norm": 0.13456420600414276, "global_step": 21616, "epoch": 203, "lr": 6.427084774058206e-05} +{"train_loss": 0.01647975854575634, "grad_norm": 0.11274365335702896, "global_step": 21617, "epoch": 203, "lr": 6.426800723134355e-05} +{"train_loss": 0.015238659456372261, "grad_norm": 0.0591629333794117, "global_step": 21618, "epoch": 203, "lr": 6.426516667197347e-05} +{"train_loss": 0.016382724046707153, "grad_norm": 0.12427366524934769, "global_step": 21619, "epoch": 203, "lr": 6.42623260624818e-05} +{"train_loss": 0.014517181552946568, "grad_norm": 0.05912750959396362, "global_step": 21620, "epoch": 203, "lr": 6.425948540287853e-05} +{"train_loss": 0.015320545062422752, "grad_norm": 0.141187384724617, "global_step": 21621, "epoch": 203, "lr": 6.42566446931736e-05} +{"train_loss": 0.013897999189794064, "grad_norm": 0.07991015911102295, "global_step": 21622, "epoch": 203, "lr": 6.425380393337705e-05} +{"train_loss": 0.015984439958800684, "grad_norm": 0.1849372386932373, "global_step": 21623, "epoch": 203, "lr": 6.42509631234988e-05, "val_loss": 0.0195656456053257} +{"train_loss": 0.0162928719073534, "grad_norm": 0.09287852793931961, "global_step": 21624, "epoch": 204, "lr": 6.42481222635489e-05} +{"train_loss": 0.017453672364354134, "grad_norm": 0.07864650338888168, "global_step": 21625, "epoch": 204, "lr": 6.424528135353726e-05} +{"train_loss": 0.01683371514081955, "grad_norm": 0.18198764324188232, "global_step": 21626, "epoch": 204, "lr": 6.424244039347393e-05} +{"train_loss": 0.016020793467760086, "grad_norm": 0.11761803925037384, "global_step": 21627, "epoch": 204, "lr": 6.423959938336883e-05} +{"train_loss": 0.015056516975164413, "grad_norm": 0.10778395086526871, "global_step": 21628, "epoch": 204, "lr": 6.423675832323198e-05} +{"train_loss": 0.017283771187067032, "grad_norm": 0.06405016034841537, "global_step": 21629, "epoch": 204, "lr": 6.423391721307338e-05} +{"train_loss": 0.01563318260014057, "grad_norm": 0.10846063494682312, "global_step": 21630, "epoch": 204, "lr": 6.423107605290295e-05} +{"train_loss": 0.015466487035155296, "grad_norm": 0.0663798525929451, "global_step": 21631, "epoch": 204, "lr": 6.42282348427307e-05} +{"train_loss": 0.01602446846663952, "grad_norm": 0.07037924230098724, "global_step": 21632, "epoch": 204, "lr": 6.422539358256665e-05} +{"train_loss": 0.015018348582088947, "grad_norm": 0.11084694415330887, "global_step": 21633, "epoch": 204, "lr": 6.422255227242073e-05} +{"train_loss": 0.01568196713924408, "grad_norm": 0.05636114627122879, "global_step": 21634, "epoch": 204, "lr": 6.421971091230296e-05} +{"train_loss": 0.01652277447283268, "grad_norm": 0.16087278723716736, "global_step": 21635, "epoch": 204, "lr": 6.421686950222331e-05} +{"train_loss": 0.015223442576825619, "grad_norm": 0.08624347299337387, "global_step": 21636, "epoch": 204, "lr": 6.421402804219177e-05} +{"train_loss": 0.013922750018537045, "grad_norm": 0.09973011910915375, "global_step": 21637, "epoch": 204, "lr": 6.421118653221829e-05} +{"train_loss": 0.016650982201099396, "grad_norm": 0.11002197116613388, "global_step": 21638, "epoch": 204, "lr": 6.42083449723129e-05} +{"train_loss": 0.01580437645316124, "grad_norm": 0.14685474336147308, "global_step": 21639, "epoch": 204, "lr": 6.420550336248558e-05} +{"train_loss": 0.015547425486147404, "grad_norm": 0.12370102107524872, "global_step": 21640, "epoch": 204, "lr": 6.420266170274627e-05} +{"train_loss": 0.016660481691360474, "grad_norm": 0.11146190017461777, "global_step": 21641, "epoch": 204, "lr": 6.4199819993105e-05} +{"train_loss": 0.016477137804031372, "grad_norm": 0.07415655255317688, "global_step": 21642, "epoch": 204, "lr": 6.419697823357174e-05} +{"train_loss": 0.014482041820883751, "grad_norm": 0.0581401027739048, "global_step": 21643, "epoch": 204, "lr": 6.419413642415645e-05} +{"train_loss": 0.01758926920592785, "grad_norm": 0.06329084932804108, "global_step": 21644, "epoch": 204, "lr": 6.419129456486917e-05} +{"train_loss": 0.01615765690803528, "grad_norm": 0.10095631331205368, "global_step": 21645, "epoch": 204, "lr": 6.418845265571982e-05} +{"train_loss": 0.014911716803908348, "grad_norm": 0.08800960332155228, "global_step": 21646, "epoch": 204, "lr": 6.418561069671844e-05} +{"train_loss": 0.016830850392580032, "grad_norm": 0.06938700377941132, "global_step": 21647, "epoch": 204, "lr": 6.418276868787498e-05} +{"train_loss": 0.013610264286398888, "grad_norm": 0.06094842031598091, "global_step": 21648, "epoch": 204, "lr": 6.417992662919945e-05} +{"train_loss": 0.015682784840464592, "grad_norm": 0.09134435653686523, "global_step": 21649, "epoch": 204, "lr": 6.417708452070183e-05} +{"train_loss": 0.018337875604629517, "grad_norm": 0.0732802078127861, "global_step": 21650, "epoch": 204, "lr": 6.417424236239209e-05} +{"train_loss": 0.01261722669005394, "grad_norm": 0.06118921935558319, "global_step": 21651, "epoch": 204, "lr": 6.417140015428023e-05} +{"train_loss": 0.01617804542183876, "grad_norm": 0.1501893252134323, "global_step": 21652, "epoch": 204, "lr": 6.416855789637622e-05} +{"train_loss": 0.015921112149953842, "grad_norm": 0.10151845961809158, "global_step": 21653, "epoch": 204, "lr": 6.416571558869006e-05} +{"train_loss": 0.01817498356103897, "grad_norm": 0.12437233328819275, "global_step": 21654, "epoch": 204, "lr": 6.416287323123176e-05} +{"train_loss": 0.014570996165275574, "grad_norm": 0.059183087199926376, "global_step": 21655, "epoch": 204, "lr": 6.416003082401125e-05} +{"train_loss": 0.017139744013547897, "grad_norm": 0.06059086695313454, "global_step": 21656, "epoch": 204, "lr": 6.415718836703859e-05} +{"train_loss": 0.016587378457188606, "grad_norm": 0.09019181877374649, "global_step": 21657, "epoch": 204, "lr": 6.41543458603237e-05} +{"train_loss": 0.01566636934876442, "grad_norm": 0.06536693125963211, "global_step": 21658, "epoch": 204, "lr": 6.415150330387659e-05} +{"train_loss": 0.014909091405570507, "grad_norm": 0.05764808505773544, "global_step": 21659, "epoch": 204, "lr": 6.414866069770727e-05} +{"train_loss": 0.0146027822047472, "grad_norm": 0.0653577372431755, "global_step": 21660, "epoch": 204, "lr": 6.414581804182569e-05} +{"train_loss": 0.014387346804141998, "grad_norm": 0.08204664289951324, "global_step": 21661, "epoch": 204, "lr": 6.414297533624188e-05} +{"train_loss": 0.01180284470319748, "grad_norm": 0.057295747101306915, "global_step": 21662, "epoch": 204, "lr": 6.414013258096578e-05} +{"train_loss": 0.017001952975988388, "grad_norm": 0.07263273745775223, "global_step": 21663, "epoch": 204, "lr": 6.413728977600742e-05} +{"train_loss": 0.016857096925377846, "grad_norm": 0.12527737021446228, "global_step": 21664, "epoch": 204, "lr": 6.413444692137677e-05} +{"train_loss": 0.015288444235920906, "grad_norm": 0.07354512065649033, "global_step": 21665, "epoch": 204, "lr": 6.413160401708383e-05} +{"train_loss": 0.01719105802476406, "grad_norm": 0.13922247290611267, "global_step": 21666, "epoch": 204, "lr": 6.412876106313856e-05} +{"train_loss": 0.015688933432102203, "grad_norm": 0.07851283252239227, "global_step": 21667, "epoch": 204, "lr": 6.412591805955096e-05} +{"train_loss": 0.016619473695755005, "grad_norm": 0.06300275772809982, "global_step": 21668, "epoch": 204, "lr": 6.412307500633106e-05} +{"train_loss": 0.017295440658926964, "grad_norm": 0.08316396176815033, "global_step": 21669, "epoch": 204, "lr": 6.412023190348878e-05} +{"train_loss": 0.017704203724861145, "grad_norm": 0.12929747998714447, "global_step": 21670, "epoch": 204, "lr": 6.411738875103418e-05} +{"train_loss": 0.019741803407669067, "grad_norm": 0.10179318487644196, "global_step": 21671, "epoch": 204, "lr": 6.41145455489772e-05} +{"train_loss": 0.014064997434616089, "grad_norm": 0.0508878119289875, "global_step": 21672, "epoch": 204, "lr": 6.411170229732783e-05} +{"train_loss": 0.015852343291044235, "grad_norm": 0.09209619462490082, "global_step": 21673, "epoch": 204, "lr": 6.410885899609611e-05} +{"train_loss": 0.017685767263174057, "grad_norm": 0.08318644016981125, "global_step": 21674, "epoch": 204, "lr": 6.410601564529197e-05} +{"train_loss": 0.017590943723917007, "grad_norm": 0.07881186902523041, "global_step": 21675, "epoch": 204, "lr": 6.410317224492543e-05} +{"train_loss": 0.01627987250685692, "grad_norm": 0.09851321578025818, "global_step": 21676, "epoch": 204, "lr": 6.410032879500647e-05} +{"train_loss": 0.015267735347151756, "grad_norm": 0.11091582477092743, "global_step": 21677, "epoch": 204, "lr": 6.40974852955451e-05} +{"train_loss": 0.01609540358185768, "grad_norm": 0.10363224893808365, "global_step": 21678, "epoch": 204, "lr": 6.409464174655129e-05} +{"train_loss": 0.01447458565235138, "grad_norm": 0.06882409006357193, "global_step": 21679, "epoch": 204, "lr": 6.409179814803504e-05} +{"train_loss": 0.01675400882959366, "grad_norm": 0.05855070799589157, "global_step": 21680, "epoch": 204, "lr": 6.408895450000635e-05} +{"train_loss": 0.016316797584295273, "grad_norm": 0.08192905783653259, "global_step": 21681, "epoch": 204, "lr": 6.408611080247518e-05} +{"train_loss": 0.014045949093997478, "grad_norm": 0.1356065422296524, "global_step": 21682, "epoch": 204, "lr": 6.408326705545155e-05} +{"train_loss": 0.015425833873450756, "grad_norm": 0.06008793041110039, "global_step": 21683, "epoch": 204, "lr": 6.408042325894545e-05} +{"train_loss": 0.014221340417861938, "grad_norm": 0.0598207525908947, "global_step": 21684, "epoch": 204, "lr": 6.407757941296685e-05} +{"train_loss": 0.016637101769447327, "grad_norm": 0.08202499896287918, "global_step": 21685, "epoch": 204, "lr": 6.40747355175258e-05} +{"train_loss": 0.014717829413712025, "grad_norm": 0.06813865154981613, "global_step": 21686, "epoch": 204, "lr": 6.40718915726322e-05} +{"train_loss": 0.016643166542053223, "grad_norm": 0.1335180252790451, "global_step": 21687, "epoch": 204, "lr": 6.406904757829612e-05} +{"train_loss": 0.013668366707861423, "grad_norm": 0.05960729345679283, "global_step": 21688, "epoch": 204, "lr": 6.406620353452752e-05} +{"train_loss": 0.017942456528544426, "grad_norm": 0.05105162411928177, "global_step": 21689, "epoch": 204, "lr": 6.40633594413364e-05} +{"train_loss": 0.01781701296567917, "grad_norm": 0.13817958533763885, "global_step": 21690, "epoch": 204, "lr": 6.406051529873275e-05} +{"train_loss": 0.01615748554468155, "grad_norm": 0.12218718975782394, "global_step": 21691, "epoch": 204, "lr": 6.405767110672657e-05} +{"train_loss": 0.015920527279376984, "grad_norm": 0.14311715960502625, "global_step": 21692, "epoch": 204, "lr": 6.405482686532784e-05} +{"train_loss": 0.017422938719391823, "grad_norm": 0.08270945399999619, "global_step": 21693, "epoch": 204, "lr": 6.405198257454656e-05} +{"train_loss": 0.015140200965106487, "grad_norm": 0.07506046444177628, "global_step": 21694, "epoch": 204, "lr": 6.404913823439272e-05} +{"train_loss": 0.015554985031485558, "grad_norm": 0.08352483063936234, "global_step": 21695, "epoch": 204, "lr": 6.404629384487633e-05} +{"train_loss": 0.016856549307703972, "grad_norm": 0.11977139860391617, "global_step": 21696, "epoch": 204, "lr": 6.404344940600735e-05} +{"train_loss": 0.015927720814943314, "grad_norm": 0.06284865736961365, "global_step": 21697, "epoch": 204, "lr": 6.404060491779583e-05} +{"train_loss": 0.016032861545681953, "grad_norm": 0.0681903287768364, "global_step": 21698, "epoch": 204, "lr": 6.40377603802517e-05} +{"train_loss": 0.01665942743420601, "grad_norm": 0.07891065627336502, "global_step": 21699, "epoch": 204, "lr": 6.403491579338499e-05} +{"train_loss": 0.014801369979977608, "grad_norm": 0.05658002570271492, "global_step": 21700, "epoch": 204, "lr": 6.403207115720571e-05} +{"train_loss": 0.017465628683567047, "grad_norm": 0.08725845068693161, "global_step": 21701, "epoch": 204, "lr": 6.402922647172381e-05} +{"train_loss": 0.018429376184940338, "grad_norm": 0.06223282963037491, "global_step": 21702, "epoch": 204, "lr": 6.402638173694933e-05} +{"train_loss": 0.01596413180232048, "grad_norm": 0.060470376163721085, "global_step": 21703, "epoch": 204, "lr": 6.402353695289223e-05} +{"train_loss": 0.014429621398448944, "grad_norm": 0.09391628205776215, "global_step": 21704, "epoch": 204, "lr": 6.402069211956254e-05} +{"train_loss": 0.014828357845544815, "grad_norm": 0.11425546556711197, "global_step": 21705, "epoch": 204, "lr": 6.40178472369702e-05} +{"train_loss": 0.016310211271047592, "grad_norm": 0.07336299866437912, "global_step": 21706, "epoch": 204, "lr": 6.401500230512528e-05} +{"train_loss": 0.015075193718075752, "grad_norm": 0.07018011063337326, "global_step": 21707, "epoch": 204, "lr": 6.401215732403772e-05} +{"train_loss": 0.015613069757819176, "grad_norm": 0.08462390303611755, "global_step": 21708, "epoch": 204, "lr": 6.40093122937175e-05} +{"train_loss": 0.015099592506885529, "grad_norm": 0.07306435704231262, "global_step": 21709, "epoch": 204, "lr": 6.40064672141747e-05} +{"train_loss": 0.01834152825176716, "grad_norm": 0.07473232597112656, "global_step": 21710, "epoch": 204, "lr": 6.400362208541925e-05} +{"train_loss": 0.018289435654878616, "grad_norm": 0.0936347171664238, "global_step": 21711, "epoch": 204, "lr": 6.400077690746113e-05} +{"train_loss": 0.014031649567186832, "grad_norm": 0.07434546202421188, "global_step": 21712, "epoch": 204, "lr": 6.399793168031042e-05} +{"train_loss": 0.015595820732414722, "grad_norm": 0.18656960129737854, "global_step": 21713, "epoch": 204, "lr": 6.399508640397702e-05} +{"train_loss": 0.015318293124437332, "grad_norm": 0.1189509928226471, "global_step": 21714, "epoch": 204, "lr": 6.399224107847101e-05} +{"train_loss": 0.014998182654380798, "grad_norm": 0.07737938314676285, "global_step": 21715, "epoch": 204, "lr": 6.398939570380234e-05} +{"train_loss": 0.01651090383529663, "grad_norm": 0.0869385153055191, "global_step": 21716, "epoch": 204, "lr": 6.3986550279981e-05} +{"train_loss": 0.016876401379704475, "grad_norm": 0.14125564694404602, "global_step": 21717, "epoch": 204, "lr": 6.398370480701702e-05} +{"train_loss": 0.01906869001686573, "grad_norm": 0.08349723368883133, "global_step": 21718, "epoch": 204, "lr": 6.398085928492038e-05} +{"train_loss": 0.015131614170968533, "grad_norm": 0.06777140498161316, "global_step": 21719, "epoch": 204, "lr": 6.397801371370108e-05} +{"train_loss": 0.012798160314559937, "grad_norm": 0.04831920936703682, "global_step": 21720, "epoch": 204, "lr": 6.397516809336912e-05} +{"train_loss": 0.015521956607699394, "grad_norm": 0.13290831446647644, "global_step": 21721, "epoch": 204, "lr": 6.397232242393449e-05} +{"train_loss": 0.01742362603545189, "grad_norm": 0.06487634032964706, "global_step": 21722, "epoch": 204, "lr": 6.39694767054072e-05} +{"train_loss": 0.014223570935428143, "grad_norm": 0.0481862872838974, "global_step": 21723, "epoch": 204, "lr": 6.396663093779725e-05} +{"train_loss": 0.019086651504039764, "grad_norm": 0.12833048403263092, "global_step": 21724, "epoch": 204, "lr": 6.396378512111463e-05} +{"train_loss": 0.013880363665521145, "grad_norm": 0.08037858456373215, "global_step": 21725, "epoch": 204, "lr": 6.396093925536933e-05} +{"train_loss": 0.01698439195752144, "grad_norm": 0.09284967929124832, "global_step": 21726, "epoch": 204, "lr": 6.395809334057138e-05} +{"train_loss": 0.01554846204817295, "grad_norm": 0.06876107305288315, "global_step": 21727, "epoch": 204, "lr": 6.395524737673073e-05} +{"train_loss": 0.01493108831346035, "grad_norm": 0.06350433826446533, "global_step": 21728, "epoch": 204, "lr": 6.395240136385743e-05} +{"train_loss": 0.01598424305436465, "grad_norm": 0.09257853776216507, "global_step": 21729, "epoch": 204, "lr": 6.394955530196147e-05, "val_loss": 0.018980994820594788} +{"train_loss": 0.014345097355544567, "grad_norm": 0.07691168785095215, "global_step": 21730, "epoch": 205, "lr": 6.394670919105282e-05} +{"train_loss": 0.016770102083683014, "grad_norm": 0.06521140784025192, "global_step": 21731, "epoch": 205, "lr": 6.394386303114151e-05} +{"train_loss": 0.017090510576963425, "grad_norm": 0.0664728656411171, "global_step": 21732, "epoch": 205, "lr": 6.394101682223754e-05} +{"train_loss": 0.014012766070663929, "grad_norm": 0.09344440698623657, "global_step": 21733, "epoch": 205, "lr": 6.393817056435087e-05} +{"train_loss": 0.0170600488781929, "grad_norm": 0.05677347630262375, "global_step": 21734, "epoch": 205, "lr": 6.393532425749156e-05} +{"train_loss": 0.01527680829167366, "grad_norm": 0.0907934382557869, "global_step": 21735, "epoch": 205, "lr": 6.393247790166957e-05} +{"train_loss": 0.015883060172200203, "grad_norm": 0.09482689946889877, "global_step": 21736, "epoch": 205, "lr": 6.39296314968949e-05} +{"train_loss": 0.015892762690782547, "grad_norm": 0.10796505957841873, "global_step": 21737, "epoch": 205, "lr": 6.392678504317756e-05} +{"train_loss": 0.01445543859153986, "grad_norm": 0.059726595878601074, "global_step": 21738, "epoch": 205, "lr": 6.392393854052758e-05} +{"train_loss": 0.013102048076689243, "grad_norm": 0.07016900181770325, "global_step": 21739, "epoch": 205, "lr": 6.392109198895492e-05} +{"train_loss": 0.017733212560415268, "grad_norm": 0.08416958898305893, "global_step": 21740, "epoch": 205, "lr": 6.39182453884696e-05} +{"train_loss": 0.015820985659956932, "grad_norm": 0.06501578539609909, "global_step": 21741, "epoch": 205, "lr": 6.391539873908162e-05} +{"train_loss": 0.01572488434612751, "grad_norm": 0.07175463438034058, "global_step": 21742, "epoch": 205, "lr": 6.391255204080096e-05} +{"train_loss": 0.017215903848409653, "grad_norm": 0.08377203345298767, "global_step": 21743, "epoch": 205, "lr": 6.390970529363766e-05} +{"train_loss": 0.0158828217536211, "grad_norm": 0.08799539506435394, "global_step": 21744, "epoch": 205, "lr": 6.390685849760172e-05} +{"train_loss": 0.013676866888999939, "grad_norm": 0.09115789085626602, "global_step": 21745, "epoch": 205, "lr": 6.390401165270311e-05} +{"train_loss": 0.016075383871793747, "grad_norm": 0.0745827704668045, "global_step": 21746, "epoch": 205, "lr": 6.390116475895185e-05} +{"train_loss": 0.017332199960947037, "grad_norm": 0.11188586056232452, "global_step": 21747, "epoch": 205, "lr": 6.389831781635794e-05} +{"train_loss": 0.017257457599043846, "grad_norm": 0.07093868404626846, "global_step": 21748, "epoch": 205, "lr": 6.38954708249314e-05} +{"train_loss": 0.01684289239346981, "grad_norm": 0.06482785940170288, "global_step": 21749, "epoch": 205, "lr": 6.38926237846822e-05} +{"train_loss": 0.015965556725859642, "grad_norm": 0.07951407134532928, "global_step": 21750, "epoch": 205, "lr": 6.388977669562037e-05} +{"train_loss": 0.016071835532784462, "grad_norm": 0.08757605403661728, "global_step": 21751, "epoch": 205, "lr": 6.38869295577559e-05} +{"train_loss": 0.015030380338430405, "grad_norm": 0.11053568124771118, "global_step": 21752, "epoch": 205, "lr": 6.388408237109879e-05} +{"train_loss": 0.015686865895986557, "grad_norm": 0.05371216684579849, "global_step": 21753, "epoch": 205, "lr": 6.388123513565909e-05} +{"train_loss": 0.018179312348365784, "grad_norm": 0.09259311109781265, "global_step": 21754, "epoch": 205, "lr": 6.387838785144673e-05} +{"train_loss": 0.015755670145154, "grad_norm": 0.07868403941392899, "global_step": 21755, "epoch": 205, "lr": 6.387554051847177e-05} +{"train_loss": 0.013504468835890293, "grad_norm": 0.1341843605041504, "global_step": 21756, "epoch": 205, "lr": 6.387269313674418e-05} +{"train_loss": 0.012844309210777283, "grad_norm": 0.07575584203004837, "global_step": 21757, "epoch": 205, "lr": 6.3869845706274e-05} +{"train_loss": 0.01608482375741005, "grad_norm": 0.14622803032398224, "global_step": 21758, "epoch": 205, "lr": 6.38669982270712e-05} +{"train_loss": 0.016429457813501358, "grad_norm": 0.06366726011037827, "global_step": 21759, "epoch": 205, "lr": 6.38641506991458e-05} +{"train_loss": 0.015428031794726849, "grad_norm": 0.05122965946793556, "global_step": 21760, "epoch": 205, "lr": 6.38613031225078e-05} +{"train_loss": 0.013660017400979996, "grad_norm": 0.12038819491863251, "global_step": 21761, "epoch": 205, "lr": 6.385845549716723e-05} +{"train_loss": 0.016297150403261185, "grad_norm": 0.0570378415286541, "global_step": 21762, "epoch": 205, "lr": 6.385560782313405e-05} +{"train_loss": 0.016078820452094078, "grad_norm": 0.08270244300365448, "global_step": 21763, "epoch": 205, "lr": 6.38527601004183e-05} +{"train_loss": 0.020034629851579666, "grad_norm": 0.05627332627773285, "global_step": 21764, "epoch": 205, "lr": 6.384991232902995e-05} +{"train_loss": 0.014577732421457767, "grad_norm": 0.0842195525765419, "global_step": 21765, "epoch": 205, "lr": 6.384706450897906e-05} +{"train_loss": 0.013105337508022785, "grad_norm": 0.08744052797555923, "global_step": 21766, "epoch": 205, "lr": 6.38442166402756e-05} +{"train_loss": 0.01728561520576477, "grad_norm": 0.09663479030132294, "global_step": 21767, "epoch": 205, "lr": 6.384136872292959e-05} +{"train_loss": 0.015514936298131943, "grad_norm": 0.12155278027057648, "global_step": 21768, "epoch": 205, "lr": 6.383852075695102e-05} +{"train_loss": 0.015764683485031128, "grad_norm": 0.06619208306074142, "global_step": 21769, "epoch": 205, "lr": 6.383567274234992e-05} +{"train_loss": 0.01827380247414112, "grad_norm": 0.06333501636981964, "global_step": 21770, "epoch": 205, "lr": 6.383282467913627e-05} +{"train_loss": 0.015304172411561012, "grad_norm": 0.06343510746955872, "global_step": 21771, "epoch": 205, "lr": 6.38299765673201e-05} +{"train_loss": 0.017435509711503983, "grad_norm": 0.10358484089374542, "global_step": 21772, "epoch": 205, "lr": 6.382712840691139e-05} +{"train_loss": 0.014943975023925304, "grad_norm": 0.06153208017349243, "global_step": 21773, "epoch": 205, "lr": 6.382428019792018e-05} +{"train_loss": 0.015200808644294739, "grad_norm": 0.07430420815944672, "global_step": 21774, "epoch": 205, "lr": 6.382143194035646e-05} +{"train_loss": 0.01601403020322323, "grad_norm": 0.06344261765480042, "global_step": 21775, "epoch": 205, "lr": 6.381858363423024e-05} +{"train_loss": 0.013914013281464577, "grad_norm": 0.07745009660720825, "global_step": 21776, "epoch": 205, "lr": 6.381573527955152e-05} +{"train_loss": 0.017803579568862915, "grad_norm": 0.06022312119603157, "global_step": 21777, "epoch": 205, "lr": 6.381288687633031e-05} +{"train_loss": 0.017845870926976204, "grad_norm": 0.1170203685760498, "global_step": 21778, "epoch": 205, "lr": 6.381003842457662e-05} +{"train_loss": 0.015488184988498688, "grad_norm": 0.056981392204761505, "global_step": 21779, "epoch": 205, "lr": 6.380718992430048e-05} +{"train_loss": 0.014624844305217266, "grad_norm": 0.05594807118177414, "global_step": 21780, "epoch": 205, "lr": 6.380434137551187e-05} +{"train_loss": 0.014730607159435749, "grad_norm": 0.04560224711894989, "global_step": 21781, "epoch": 205, "lr": 6.38014927782208e-05} +{"train_loss": 0.015663713216781616, "grad_norm": 0.07233058661222458, "global_step": 21782, "epoch": 205, "lr": 6.37986441324373e-05} +{"train_loss": 0.01584860309958458, "grad_norm": 0.08921979367733002, "global_step": 21783, "epoch": 205, "lr": 6.379579543817135e-05} +{"train_loss": 0.01595061644911766, "grad_norm": 0.17365120351314545, "global_step": 21784, "epoch": 205, "lr": 6.379294669543299e-05} +{"train_loss": 0.015495194122195244, "grad_norm": 0.08595428615808487, "global_step": 21785, "epoch": 205, "lr": 6.37900979042322e-05} +{"train_loss": 0.016821861267089844, "grad_norm": 0.09204169362783432, "global_step": 21786, "epoch": 205, "lr": 6.3787249064579e-05} +{"train_loss": 0.01319917943328619, "grad_norm": 0.04671445116400719, "global_step": 21787, "epoch": 205, "lr": 6.378440017648342e-05} +{"train_loss": 0.01629308983683586, "grad_norm": 0.08909501135349274, "global_step": 21788, "epoch": 205, "lr": 6.378155123995543e-05} +{"train_loss": 0.016196519136428833, "grad_norm": 0.12191341817378998, "global_step": 21789, "epoch": 205, "lr": 6.37787022550051e-05} +{"train_loss": 0.01732177659869194, "grad_norm": 0.08575504273176193, "global_step": 21790, "epoch": 205, "lr": 6.377585322164236e-05} +{"train_loss": 0.014582382515072823, "grad_norm": 0.06461527943611145, "global_step": 21791, "epoch": 205, "lr": 6.37730041398773e-05} +{"train_loss": 0.014813674613833427, "grad_norm": 0.06130707636475563, "global_step": 21792, "epoch": 205, "lr": 6.377015500971988e-05} +{"train_loss": 0.0172311719506979, "grad_norm": 0.06060812249779701, "global_step": 21793, "epoch": 205, "lr": 6.37673058311801e-05} +{"train_loss": 0.01670592650771141, "grad_norm": 0.07249744236469269, "global_step": 21794, "epoch": 205, "lr": 6.376445660426803e-05} +{"train_loss": 0.014445985667407513, "grad_norm": 0.06684152036905289, "global_step": 21795, "epoch": 205, "lr": 6.376160732899362e-05} +{"train_loss": 0.015039607882499695, "grad_norm": 0.061755869537591934, "global_step": 21796, "epoch": 205, "lr": 6.375875800536691e-05} +{"train_loss": 0.017617039382457733, "grad_norm": 0.10753736644983292, "global_step": 21797, "epoch": 205, "lr": 6.375590863339792e-05} +{"train_loss": 0.01630113460123539, "grad_norm": 0.054121412336826324, "global_step": 21798, "epoch": 205, "lr": 6.375305921309665e-05} +{"train_loss": 0.016827203333377838, "grad_norm": 0.06693485379219055, "global_step": 21799, "epoch": 205, "lr": 6.37502097444731e-05} +{"train_loss": 0.013617593795061111, "grad_norm": 0.08236673474311829, "global_step": 21800, "epoch": 205, "lr": 6.374736022753729e-05} +{"train_loss": 0.01493590697646141, "grad_norm": 0.1027202382683754, "global_step": 21801, "epoch": 205, "lr": 6.374451066229925e-05} +{"train_loss": 0.01667487621307373, "grad_norm": 0.06305442750453949, "global_step": 21802, "epoch": 205, "lr": 6.374166104876896e-05} +{"train_loss": 0.017102312296628952, "grad_norm": 0.0843110978603363, "global_step": 21803, "epoch": 205, "lr": 6.373881138695646e-05} +{"train_loss": 0.015588657930493355, "grad_norm": 0.10787127912044525, "global_step": 21804, "epoch": 205, "lr": 6.373596167687175e-05} +{"train_loss": 0.015185834839940071, "grad_norm": 0.05110612139105797, "global_step": 21805, "epoch": 205, "lr": 6.373311191852483e-05} +{"train_loss": 0.015398643910884857, "grad_norm": 0.1017678827047348, "global_step": 21806, "epoch": 205, "lr": 6.373026211192575e-05} +{"train_loss": 0.015935838222503662, "grad_norm": 0.04929617419838905, "global_step": 21807, "epoch": 205, "lr": 6.372741225708449e-05} +{"train_loss": 0.018117370083928108, "grad_norm": 0.09363003075122833, "global_step": 21808, "epoch": 205, "lr": 6.372456235401107e-05} +{"train_loss": 0.016574952751398087, "grad_norm": 0.11157915741205215, "global_step": 21809, "epoch": 205, "lr": 6.372171240271552e-05} +{"train_loss": 0.015660593286156654, "grad_norm": 0.07112262398004532, "global_step": 21810, "epoch": 205, "lr": 6.371886240320782e-05} +{"train_loss": 0.014412753283977509, "grad_norm": 0.06131505221128464, "global_step": 21811, "epoch": 205, "lr": 6.371601235549802e-05} +{"train_loss": 0.017888836562633514, "grad_norm": 0.08764289319515228, "global_step": 21812, "epoch": 205, "lr": 6.371316225959612e-05} +{"train_loss": 0.015862490981817245, "grad_norm": 0.11331059783697128, "global_step": 21813, "epoch": 205, "lr": 6.371031211551212e-05} +{"train_loss": 0.016151970252394676, "grad_norm": 0.05370144173502922, "global_step": 21814, "epoch": 205, "lr": 6.370746192325605e-05} +{"train_loss": 0.01615876890718937, "grad_norm": 0.07952672988176346, "global_step": 21815, "epoch": 205, "lr": 6.370461168283791e-05} +{"train_loss": 0.01607225276529789, "grad_norm": 0.0726214200258255, "global_step": 21816, "epoch": 205, "lr": 6.370176139426774e-05} +{"train_loss": 0.01794511266052723, "grad_norm": 0.06941307336091995, "global_step": 21817, "epoch": 205, "lr": 6.369891105755553e-05} +{"train_loss": 0.01569126546382904, "grad_norm": 0.06412015110254288, "global_step": 21818, "epoch": 205, "lr": 6.369606067271132e-05} +{"train_loss": 0.017635062336921692, "grad_norm": 0.0985688716173172, "global_step": 21819, "epoch": 205, "lr": 6.369321023974508e-05} +{"train_loss": 0.01819288358092308, "grad_norm": 0.07667482644319534, "global_step": 21820, "epoch": 205, "lr": 6.369035975866686e-05} +{"train_loss": 0.017170514911413193, "grad_norm": 0.06063094362616539, "global_step": 21821, "epoch": 205, "lr": 6.368750922948669e-05} +{"train_loss": 0.014780200086534023, "grad_norm": 0.056906089186668396, "global_step": 21822, "epoch": 205, "lr": 6.368465865221455e-05} +{"train_loss": 0.015685278922319412, "grad_norm": 0.09975515305995941, "global_step": 21823, "epoch": 205, "lr": 6.368180802686048e-05} +{"train_loss": 0.019574670121073723, "grad_norm": 0.07506859302520752, "global_step": 21824, "epoch": 205, "lr": 6.367895735343448e-05} +{"train_loss": 0.014067986980080605, "grad_norm": 0.07908809930086136, "global_step": 21825, "epoch": 205, "lr": 6.367610663194658e-05} +{"train_loss": 0.015440298244357109, "grad_norm": 0.10290918499231339, "global_step": 21826, "epoch": 205, "lr": 6.367325586240678e-05} +{"train_loss": 0.015538967214524746, "grad_norm": 0.08674860000610352, "global_step": 21827, "epoch": 205, "lr": 6.36704050448251e-05} +{"train_loss": 0.014733964577317238, "grad_norm": 0.11560185998678207, "global_step": 21828, "epoch": 205, "lr": 6.366755417921158e-05} +{"train_loss": 0.01643393561244011, "grad_norm": 0.10966053605079651, "global_step": 21829, "epoch": 205, "lr": 6.36647032655762e-05} +{"train_loss": 0.0138429906219244, "grad_norm": 0.09214101731777191, "global_step": 21830, "epoch": 205, "lr": 6.366185230392901e-05} +{"train_loss": 0.015625279396772385, "grad_norm": 0.06723421066999435, "global_step": 21831, "epoch": 205, "lr": 6.365900129428001e-05} +{"train_loss": 0.014883032068610191, "grad_norm": 0.06993565708398819, "global_step": 21832, "epoch": 205, "lr": 6.365615023663923e-05} +{"train_loss": 0.015908075496554375, "grad_norm": 0.07765748351812363, "global_step": 21833, "epoch": 205, "lr": 6.365329913101665e-05} +{"train_loss": 0.01904306747019291, "grad_norm": 0.07860075682401657, "global_step": 21834, "epoch": 205, "lr": 6.365044797742232e-05} +{"train_loss": 0.01592628374028037, "grad_norm": 0.1304624080657959, "global_step": 21835, "epoch": 205, "lr": 6.364759677586627e-05, "val_loss": 0.019342809915542603} +{"train_loss": 0.013586707413196564, "grad_norm": 0.07639317214488983, "global_step": 21836, "epoch": 206, "lr": 6.364474552635848e-05} +{"train_loss": 0.013599490746855736, "grad_norm": 0.12564820051193237, "global_step": 21837, "epoch": 206, "lr": 6.3641894228909e-05} +{"train_loss": 0.018051208928227425, "grad_norm": 0.07965650409460068, "global_step": 21838, "epoch": 206, "lr": 6.363904288352784e-05} +{"train_loss": 0.015011198818683624, "grad_norm": 0.055115774273872375, "global_step": 21839, "epoch": 206, "lr": 6.3636191490225e-05} +{"train_loss": 0.015526773408055305, "grad_norm": 0.08079566806554794, "global_step": 21840, "epoch": 206, "lr": 6.363334004901053e-05} +{"train_loss": 0.016577836126089096, "grad_norm": 0.06769149005413055, "global_step": 21841, "epoch": 206, "lr": 6.363048855989443e-05} +{"train_loss": 0.017377609387040138, "grad_norm": 0.15409575402736664, "global_step": 21842, "epoch": 206, "lr": 6.362763702288672e-05} +{"train_loss": 0.015224491246044636, "grad_norm": 0.08701207488775253, "global_step": 21843, "epoch": 206, "lr": 6.362478543799741e-05} +{"train_loss": 0.01778293214738369, "grad_norm": 0.17813517153263092, "global_step": 21844, "epoch": 206, "lr": 6.362193380523654e-05} +{"train_loss": 0.015342051163315773, "grad_norm": 0.06943036615848541, "global_step": 21845, "epoch": 206, "lr": 6.361908212461412e-05} +{"train_loss": 0.016794200986623764, "grad_norm": 0.06615705788135529, "global_step": 21846, "epoch": 206, "lr": 6.361623039614014e-05} +{"train_loss": 0.01380713377147913, "grad_norm": 0.14268596470355988, "global_step": 21847, "epoch": 206, "lr": 6.361337861982468e-05} +{"train_loss": 0.015309940092265606, "grad_norm": 0.10071995109319687, "global_step": 21848, "epoch": 206, "lr": 6.361052679567771e-05} +{"train_loss": 0.01721186563372612, "grad_norm": 0.08351824432611465, "global_step": 21849, "epoch": 206, "lr": 6.360767492370928e-05} +{"train_loss": 0.015987291932106018, "grad_norm": 0.12958234548568726, "global_step": 21850, "epoch": 206, "lr": 6.360482300392941e-05} +{"train_loss": 0.016380440443754196, "grad_norm": 0.059722039848566055, "global_step": 21851, "epoch": 206, "lr": 6.360197103634808e-05} +{"train_loss": 0.016862772405147552, "grad_norm": 0.10134056210517883, "global_step": 21852, "epoch": 206, "lr": 6.359911902097535e-05} +{"train_loss": 0.014145989902317524, "grad_norm": 0.050753649324178696, "global_step": 21853, "epoch": 206, "lr": 6.359626695782122e-05} +{"train_loss": 0.01447281427681446, "grad_norm": 0.09939240664243698, "global_step": 21854, "epoch": 206, "lr": 6.359341484689574e-05} +{"train_loss": 0.01777336746454239, "grad_norm": 0.09750627726316452, "global_step": 21855, "epoch": 206, "lr": 6.359056268820891e-05} +{"train_loss": 0.01602812670171261, "grad_norm": 0.05620637536048889, "global_step": 21856, "epoch": 206, "lr": 6.358771048177075e-05} +{"train_loss": 0.016094397753477097, "grad_norm": 0.09525686502456665, "global_step": 21857, "epoch": 206, "lr": 6.358485822759128e-05} +{"train_loss": 0.015990039333701134, "grad_norm": 0.1540132313966751, "global_step": 21858, "epoch": 206, "lr": 6.358200592568051e-05} +{"train_loss": 0.014685286208987236, "grad_norm": 0.0739462599158287, "global_step": 21859, "epoch": 206, "lr": 6.357915357604851e-05} +{"train_loss": 0.014996560290455818, "grad_norm": 0.06278233230113983, "global_step": 21860, "epoch": 206, "lr": 6.357630117870526e-05} +{"train_loss": 0.0190014336258173, "grad_norm": 0.09430308640003204, "global_step": 21861, "epoch": 206, "lr": 6.357344873366078e-05} +{"train_loss": 0.016992397606372833, "grad_norm": 0.15008409321308136, "global_step": 21862, "epoch": 206, "lr": 6.357059624092511e-05} +{"train_loss": 0.013708068057894707, "grad_norm": 0.06048910319805145, "global_step": 21863, "epoch": 206, "lr": 6.356774370050826e-05} +{"train_loss": 0.014413194730877876, "grad_norm": 0.09059823304414749, "global_step": 21864, "epoch": 206, "lr": 6.356489111242028e-05} +{"train_loss": 0.016269516199827194, "grad_norm": 0.1498962938785553, "global_step": 21865, "epoch": 206, "lr": 6.356203847667115e-05} +{"train_loss": 0.01748787984251976, "grad_norm": 0.11600185930728912, "global_step": 21866, "epoch": 206, "lr": 6.355918579327092e-05} +{"train_loss": 0.014480273239314556, "grad_norm": 0.10297378152608871, "global_step": 21867, "epoch": 206, "lr": 6.355633306222961e-05} +{"train_loss": 0.01693853735923767, "grad_norm": 0.1624254584312439, "global_step": 21868, "epoch": 206, "lr": 6.355348028355725e-05} +{"train_loss": 0.017590636387467384, "grad_norm": 0.09629254788160324, "global_step": 21869, "epoch": 206, "lr": 6.355062745726384e-05} +{"train_loss": 0.01755841076374054, "grad_norm": 0.1098286435008049, "global_step": 21870, "epoch": 206, "lr": 6.354777458335941e-05} +{"train_loss": 0.01357848010957241, "grad_norm": 0.11885029077529907, "global_step": 21871, "epoch": 206, "lr": 6.354492166185403e-05} +{"train_loss": 0.015849195420742035, "grad_norm": 0.10672228038311005, "global_step": 21872, "epoch": 206, "lr": 6.354206869275767e-05} +{"train_loss": 0.016383331269025803, "grad_norm": 0.09242191910743713, "global_step": 21873, "epoch": 206, "lr": 6.353921567608035e-05} +{"train_loss": 0.015802007168531418, "grad_norm": 0.08651930093765259, "global_step": 21874, "epoch": 206, "lr": 6.353636261183213e-05} +{"train_loss": 0.016908762976527214, "grad_norm": 0.06850980222225189, "global_step": 21875, "epoch": 206, "lr": 6.353350950002301e-05} +{"train_loss": 0.017897160723805428, "grad_norm": 0.10068058967590332, "global_step": 21876, "epoch": 206, "lr": 6.353065634066304e-05} +{"train_loss": 0.016570182517170906, "grad_norm": 0.07667852938175201, "global_step": 21877, "epoch": 206, "lr": 6.35278031337622e-05} +{"train_loss": 0.015242448076605797, "grad_norm": 0.07274771481752396, "global_step": 21878, "epoch": 206, "lr": 6.352494987933057e-05} +{"train_loss": 0.0140299741178751, "grad_norm": 0.058778904378414154, "global_step": 21879, "epoch": 206, "lr": 6.352209657737814e-05} +{"train_loss": 0.016987046226859093, "grad_norm": 0.07924820482730865, "global_step": 21880, "epoch": 206, "lr": 6.351924322791494e-05} +{"train_loss": 0.014982507564127445, "grad_norm": 0.10749230533838272, "global_step": 21881, "epoch": 206, "lr": 6.3516389830951e-05} +{"train_loss": 0.01686323992908001, "grad_norm": 0.09787087142467499, "global_step": 21882, "epoch": 206, "lr": 6.351353638649634e-05} +{"train_loss": 0.018201088532805443, "grad_norm": 0.0588560551404953, "global_step": 21883, "epoch": 206, "lr": 6.351068289456099e-05} +{"train_loss": 0.01526571810245514, "grad_norm": 0.06969673186540604, "global_step": 21884, "epoch": 206, "lr": 6.350782935515497e-05} +{"train_loss": 0.017420165240764618, "grad_norm": 0.09310811758041382, "global_step": 21885, "epoch": 206, "lr": 6.350497576828833e-05} +{"train_loss": 0.014948029071092606, "grad_norm": 0.07208896428346634, "global_step": 21886, "epoch": 206, "lr": 6.350212213397107e-05} +{"train_loss": 0.015086852014064789, "grad_norm": 0.06660645455121994, "global_step": 21887, "epoch": 206, "lr": 6.349926845221323e-05} +{"train_loss": 0.018282221630215645, "grad_norm": 0.10631541162729263, "global_step": 21888, "epoch": 206, "lr": 6.349641472302483e-05} +{"train_loss": 0.017643850296735764, "grad_norm": 0.05798842012882233, "global_step": 21889, "epoch": 206, "lr": 6.34935609464159e-05} +{"train_loss": 0.015478406101465225, "grad_norm": 0.05242794379591942, "global_step": 21890, "epoch": 206, "lr": 6.349070712239645e-05} +{"train_loss": 0.015504159964621067, "grad_norm": 0.10602406412363052, "global_step": 21891, "epoch": 206, "lr": 6.348785325097653e-05} +{"train_loss": 0.014921972528100014, "grad_norm": 0.08561990410089493, "global_step": 21892, "epoch": 206, "lr": 6.348499933216615e-05} +{"train_loss": 0.016639068722724915, "grad_norm": 0.11414624005556107, "global_step": 21893, "epoch": 206, "lr": 6.348214536597536e-05} +{"train_loss": 0.01544971950352192, "grad_norm": 0.07560910284519196, "global_step": 21894, "epoch": 206, "lr": 6.347929135241417e-05} +{"train_loss": 0.016535472124814987, "grad_norm": 0.06100911647081375, "global_step": 21895, "epoch": 206, "lr": 6.347643729149262e-05} +{"train_loss": 0.014435704797506332, "grad_norm": 0.07687325775623322, "global_step": 21896, "epoch": 206, "lr": 6.347358318322072e-05} +{"train_loss": 0.017191607505083084, "grad_norm": 0.051381487399339676, "global_step": 21897, "epoch": 206, "lr": 6.34707290276085e-05} +{"train_loss": 0.01624796912074089, "grad_norm": 0.08386018872261047, "global_step": 21898, "epoch": 206, "lr": 6.346787482466603e-05} +{"train_loss": 0.01497262716293335, "grad_norm": 0.09836698323488235, "global_step": 21899, "epoch": 206, "lr": 6.346502057440327e-05} +{"train_loss": 0.015511490404605865, "grad_norm": 0.05833970010280609, "global_step": 21900, "epoch": 206, "lr": 6.34621662768303e-05} +{"train_loss": 0.01374170370399952, "grad_norm": 0.08790483325719833, "global_step": 21901, "epoch": 206, "lr": 6.345931193195713e-05} +{"train_loss": 0.014959919266402721, "grad_norm": 0.05495370924472809, "global_step": 21902, "epoch": 206, "lr": 6.34564575397938e-05} +{"train_loss": 0.014055067673325539, "grad_norm": 0.0590798556804657, "global_step": 21903, "epoch": 206, "lr": 6.345360310035032e-05} +{"train_loss": 0.01441016886383295, "grad_norm": 0.06686470657587051, "global_step": 21904, "epoch": 206, "lr": 6.345074861363671e-05} +{"train_loss": 0.015733899548649788, "grad_norm": 0.06935305148363113, "global_step": 21905, "epoch": 206, "lr": 6.344789407966305e-05} +{"train_loss": 0.01530023105442524, "grad_norm": 0.09120973199605942, "global_step": 21906, "epoch": 206, "lr": 6.344503949843932e-05} +{"train_loss": 0.015621769241988659, "grad_norm": 0.06742064654827118, "global_step": 21907, "epoch": 206, "lr": 6.344218486997559e-05} +{"train_loss": 0.01560797356069088, "grad_norm": 0.14715349674224854, "global_step": 21908, "epoch": 206, "lr": 6.343933019428186e-05} +{"train_loss": 0.017134342342615128, "grad_norm": 0.08735810965299606, "global_step": 21909, "epoch": 206, "lr": 6.343647547136815e-05} +{"train_loss": 0.019382940605282784, "grad_norm": 0.10997778177261353, "global_step": 21910, "epoch": 206, "lr": 6.343362070124453e-05} +{"train_loss": 0.017300505191087723, "grad_norm": 0.10827268660068512, "global_step": 21911, "epoch": 206, "lr": 6.343076588392099e-05} +{"train_loss": 0.015153510496020317, "grad_norm": 0.10261169075965881, "global_step": 21912, "epoch": 206, "lr": 6.342791101940759e-05} +{"train_loss": 0.016722317785024643, "grad_norm": 0.10784358531236649, "global_step": 21913, "epoch": 206, "lr": 6.342505610771435e-05} +{"train_loss": 0.016566773876547813, "grad_norm": 0.11445080488920212, "global_step": 21914, "epoch": 206, "lr": 6.34222011488513e-05} +{"train_loss": 0.016520865261554718, "grad_norm": 0.12072630226612091, "global_step": 21915, "epoch": 206, "lr": 6.341934614282848e-05} +{"train_loss": 0.014877555891871452, "grad_norm": 0.07463917136192322, "global_step": 21916, "epoch": 206, "lr": 6.341649108965591e-05} +{"train_loss": 0.01739933341741562, "grad_norm": 0.0904310792684555, "global_step": 21917, "epoch": 206, "lr": 6.341363598934361e-05} +{"train_loss": 0.014766616746783257, "grad_norm": 0.10669448226690292, "global_step": 21918, "epoch": 206, "lr": 6.341078084190165e-05} +{"train_loss": 0.015994563698768616, "grad_norm": 0.13413049280643463, "global_step": 21919, "epoch": 206, "lr": 6.340792564734001e-05} +{"train_loss": 0.01576736941933632, "grad_norm": 0.09911277145147324, "global_step": 21920, "epoch": 206, "lr": 6.340507040566877e-05} +{"train_loss": 0.01302279345691204, "grad_norm": 0.08018626272678375, "global_step": 21921, "epoch": 206, "lr": 6.340221511689794e-05} +{"train_loss": 0.014576666057109833, "grad_norm": 0.09484464675188065, "global_step": 21922, "epoch": 206, "lr": 6.339935978103756e-05} +{"train_loss": 0.014277337118983269, "grad_norm": 0.10652245581150055, "global_step": 21923, "epoch": 206, "lr": 6.339650439809766e-05} +{"train_loss": 0.014502582140266895, "grad_norm": 0.06382295489311218, "global_step": 21924, "epoch": 206, "lr": 6.339364896808826e-05} +{"train_loss": 0.016244299709796906, "grad_norm": 0.08306457847356796, "global_step": 21925, "epoch": 206, "lr": 6.339079349101939e-05} +{"train_loss": 0.016400190070271492, "grad_norm": 0.0853559598326683, "global_step": 21926, "epoch": 206, "lr": 6.338793796690112e-05} +{"train_loss": 0.015679815784096718, "grad_norm": 0.10509125888347626, "global_step": 21927, "epoch": 206, "lr": 6.338508239574344e-05} +{"train_loss": 0.01882406324148178, "grad_norm": 0.17210671305656433, "global_step": 21928, "epoch": 206, "lr": 6.33822267775564e-05} +{"train_loss": 0.015849413350224495, "grad_norm": 0.0690104141831398, "global_step": 21929, "epoch": 206, "lr": 6.337937111235006e-05} +{"train_loss": 0.019316475838422775, "grad_norm": 0.15982156991958618, "global_step": 21930, "epoch": 206, "lr": 6.33765154001344e-05} +{"train_loss": 0.016301659867167473, "grad_norm": 0.054864197969436646, "global_step": 21931, "epoch": 206, "lr": 6.337365964091949e-05} +{"train_loss": 0.01764051616191864, "grad_norm": 0.08686880767345428, "global_step": 21932, "epoch": 206, "lr": 6.337080383471537e-05} +{"train_loss": 0.01716814935207367, "grad_norm": 0.15346942842006683, "global_step": 21933, "epoch": 206, "lr": 6.336794798153204e-05} +{"train_loss": 0.013454040512442589, "grad_norm": 0.05093365162611008, "global_step": 21934, "epoch": 206, "lr": 6.336509208137956e-05} +{"train_loss": 0.014584152027964592, "grad_norm": 0.06782552599906921, "global_step": 21935, "epoch": 206, "lr": 6.336223613426798e-05} +{"train_loss": 0.014711987227201462, "grad_norm": 0.06521442532539368, "global_step": 21936, "epoch": 206, "lr": 6.335938014020728e-05} +{"train_loss": 0.01538627315312624, "grad_norm": 0.06885314732789993, "global_step": 21937, "epoch": 206, "lr": 6.335652409920755e-05} +{"train_loss": 0.01487863902002573, "grad_norm": 0.08447171747684479, "global_step": 21938, "epoch": 206, "lr": 6.33536680112788e-05} +{"train_loss": 0.016917672008275986, "grad_norm": 0.0667870044708252, "global_step": 21939, "epoch": 206, "lr": 6.335081187643106e-05} +{"train_loss": 0.01687118597328663, "grad_norm": 0.08908744156360626, "global_step": 21940, "epoch": 206, "lr": 6.334795569467437e-05} +{"train_loss": 0.01594537138974048, "grad_norm": 0.186200812458992, "global_step": 21941, "epoch": 206, "lr": 6.334509946601879e-05, "val_loss": 0.01965395361185074} +{"train_loss": 0.016253378242254257, "grad_norm": 0.10513779520988464, "global_step": 21942, "epoch": 207, "lr": 6.334224319047431e-05} +{"train_loss": 0.01440175250172615, "grad_norm": 0.0912427231669426, "global_step": 21943, "epoch": 207, "lr": 6.333938686805099e-05} +{"train_loss": 0.015420900657773018, "grad_norm": 0.05826359987258911, "global_step": 21944, "epoch": 207, "lr": 6.33365304987589e-05} +{"train_loss": 0.01709870621562004, "grad_norm": 0.08820713311433792, "global_step": 21945, "epoch": 207, "lr": 6.333367408260801e-05} +{"train_loss": 0.015929898247122765, "grad_norm": 0.058983176946640015, "global_step": 21946, "epoch": 207, "lr": 6.33308176196084e-05} +{"train_loss": 0.015174933709204197, "grad_norm": 0.06820591539144516, "global_step": 21947, "epoch": 207, "lr": 6.33279611097701e-05} +{"train_loss": 0.015121834352612495, "grad_norm": 0.14467012882232666, "global_step": 21948, "epoch": 207, "lr": 6.332510455310311e-05} +{"train_loss": 0.014924589544534683, "grad_norm": 0.07444927841424942, "global_step": 21949, "epoch": 207, "lr": 6.332224794961752e-05} +{"train_loss": 0.014404364861547947, "grad_norm": 0.08183915168046951, "global_step": 21950, "epoch": 207, "lr": 6.331939129932334e-05} +{"train_loss": 0.01866707019507885, "grad_norm": 0.09622782468795776, "global_step": 21951, "epoch": 207, "lr": 6.331653460223061e-05} +{"train_loss": 0.017870623618364334, "grad_norm": 0.06053335592150688, "global_step": 21952, "epoch": 207, "lr": 6.331367785834939e-05} +{"train_loss": 0.016403820365667343, "grad_norm": 0.07815409451723099, "global_step": 21953, "epoch": 207, "lr": 6.331082106768967e-05} +{"train_loss": 0.015052447095513344, "grad_norm": 0.06496585160493851, "global_step": 21954, "epoch": 207, "lr": 6.330796423026154e-05} +{"train_loss": 0.013397375121712685, "grad_norm": 0.07398369908332825, "global_step": 21955, "epoch": 207, "lr": 6.330510734607497e-05} +{"train_loss": 0.014561902731657028, "grad_norm": 0.13849571347236633, "global_step": 21956, "epoch": 207, "lr": 6.330225041514008e-05} +{"train_loss": 0.01582152768969536, "grad_norm": 0.07585245370864868, "global_step": 21957, "epoch": 207, "lr": 6.329939343746684e-05} +{"train_loss": 0.01695990189909935, "grad_norm": 0.10002555698156357, "global_step": 21958, "epoch": 207, "lr": 6.329653641306532e-05} +{"train_loss": 0.016744673252105713, "grad_norm": 0.1732659786939621, "global_step": 21959, "epoch": 207, "lr": 6.329367934194556e-05} +{"train_loss": 0.013917902484536171, "grad_norm": 0.15804249048233032, "global_step": 21960, "epoch": 207, "lr": 6.329082222411758e-05} +{"train_loss": 0.016552140936255455, "grad_norm": 0.08926963806152344, "global_step": 21961, "epoch": 207, "lr": 6.328796505959145e-05} +{"train_loss": 0.014922944828867912, "grad_norm": 0.06481630355119705, "global_step": 21962, "epoch": 207, "lr": 6.328510784837718e-05} +{"train_loss": 0.015387170948088169, "grad_norm": 0.05833122879266739, "global_step": 21963, "epoch": 207, "lr": 6.328225059048483e-05} +{"train_loss": 0.013023053295910358, "grad_norm": 0.07038791477680206, "global_step": 21964, "epoch": 207, "lr": 6.32793932859244e-05} +{"train_loss": 0.014845740050077438, "grad_norm": 0.09483982622623444, "global_step": 21965, "epoch": 207, "lr": 6.327653593470598e-05} +{"train_loss": 0.015241620130836964, "grad_norm": 0.0606885589659214, "global_step": 21966, "epoch": 207, "lr": 6.327367853683959e-05} +{"train_loss": 0.017326518893241882, "grad_norm": 0.06991172581911087, "global_step": 21967, "epoch": 207, "lr": 6.327082109233525e-05} +{"train_loss": 0.01453801803290844, "grad_norm": 0.07493996620178223, "global_step": 21968, "epoch": 207, "lr": 6.326796360120302e-05} +{"train_loss": 0.015812449157238007, "grad_norm": 0.05144374445080757, "global_step": 21969, "epoch": 207, "lr": 6.326510606345295e-05} +{"train_loss": 0.01714516058564186, "grad_norm": 0.09539763629436493, "global_step": 21970, "epoch": 207, "lr": 6.326224847909504e-05} +{"train_loss": 0.01780487596988678, "grad_norm": 0.08310835808515549, "global_step": 21971, "epoch": 207, "lr": 6.325939084813939e-05} +{"train_loss": 0.014960557222366333, "grad_norm": 0.08801090717315674, "global_step": 21972, "epoch": 207, "lr": 6.325653317059596e-05} +{"train_loss": 0.015507888048887253, "grad_norm": 0.06677146255970001, "global_step": 21973, "epoch": 207, "lr": 6.325367544647489e-05} +{"train_loss": 0.014558576047420502, "grad_norm": 0.05933206155896187, "global_step": 21974, "epoch": 207, "lr": 6.325081767578614e-05} +{"train_loss": 0.01671888306736946, "grad_norm": 0.0765325054526329, "global_step": 21975, "epoch": 207, "lr": 6.324795985853978e-05} +{"train_loss": 0.013982690870761871, "grad_norm": 0.08094708621501923, "global_step": 21976, "epoch": 207, "lr": 6.324510199474584e-05} +{"train_loss": 0.01573142781853676, "grad_norm": 0.09626217186450958, "global_step": 21977, "epoch": 207, "lr": 6.32422440844144e-05} +{"train_loss": 0.014706969261169434, "grad_norm": 0.04500771313905716, "global_step": 21978, "epoch": 207, "lr": 6.323938612755546e-05} +{"train_loss": 0.016368871554732323, "grad_norm": 0.12338811159133911, "global_step": 21979, "epoch": 207, "lr": 6.323652812417908e-05} +{"train_loss": 0.01472289115190506, "grad_norm": 0.15188509225845337, "global_step": 21980, "epoch": 207, "lr": 6.323367007429528e-05} +{"train_loss": 0.0150988744571805, "grad_norm": 0.05451624467968941, "global_step": 21981, "epoch": 207, "lr": 6.323081197791413e-05} +{"train_loss": 0.013983801007270813, "grad_norm": 0.09050831943750381, "global_step": 21982, "epoch": 207, "lr": 6.322795383504567e-05} +{"train_loss": 0.015911489725112915, "grad_norm": 0.08905088901519775, "global_step": 21983, "epoch": 207, "lr": 6.322509564569992e-05} +{"train_loss": 0.016271468251943588, "grad_norm": 0.07448888570070267, "global_step": 21984, "epoch": 207, "lr": 6.322223740988691e-05} +{"train_loss": 0.0185173861682415, "grad_norm": 0.07751867920160294, "global_step": 21985, "epoch": 207, "lr": 6.321937912761676e-05} +{"train_loss": 0.019065028056502342, "grad_norm": 0.08944400399923325, "global_step": 21986, "epoch": 207, "lr": 6.321652079889942e-05} +{"train_loss": 0.014303108677268028, "grad_norm": 0.1265360563993454, "global_step": 21987, "epoch": 207, "lr": 6.3213662423745e-05} +{"train_loss": 0.01613418199121952, "grad_norm": 0.10219793021678925, "global_step": 21988, "epoch": 207, "lr": 6.321080400216351e-05} +{"train_loss": 0.015606096014380455, "grad_norm": 0.06636390089988708, "global_step": 21989, "epoch": 207, "lr": 6.3207945534165e-05} +{"train_loss": 0.016002219170331955, "grad_norm": 0.13603287935256958, "global_step": 21990, "epoch": 207, "lr": 6.32050870197595e-05} +{"train_loss": 0.018712162971496582, "grad_norm": 0.09112303704023361, "global_step": 21991, "epoch": 207, "lr": 6.320222845895708e-05} +{"train_loss": 0.016184277832508087, "grad_norm": 0.08277129381895065, "global_step": 21992, "epoch": 207, "lr": 6.319936985176776e-05} +{"train_loss": 0.01531447097659111, "grad_norm": 0.08566031605005264, "global_step": 21993, "epoch": 207, "lr": 6.319651119820162e-05} +{"train_loss": 0.013850889168679714, "grad_norm": 0.07975301891565323, "global_step": 21994, "epoch": 207, "lr": 6.319365249826865e-05} +{"train_loss": 0.015836400911211967, "grad_norm": 0.07013983279466629, "global_step": 21995, "epoch": 207, "lr": 6.319079375197894e-05} +{"train_loss": 0.01425158977508545, "grad_norm": 0.047367800027132034, "global_step": 21996, "epoch": 207, "lr": 6.31879349593425e-05} +{"train_loss": 0.018219616264104843, "grad_norm": 0.1213223934173584, "global_step": 21997, "epoch": 207, "lr": 6.318507612036942e-05} +{"train_loss": 0.01659250631928444, "grad_norm": 0.1322294920682907, "global_step": 21998, "epoch": 207, "lr": 6.31822172350697e-05} +{"train_loss": 0.016321025788784027, "grad_norm": 0.1360144019126892, "global_step": 21999, "epoch": 207, "lr": 6.317935830345338e-05} +{"train_loss": 0.015271938405930996, "grad_norm": 0.07950279861688614, "global_step": 22000, "epoch": 207, "lr": 6.317649932553055e-05} +{"train_loss": 0.01621263287961483, "grad_norm": 0.0847177803516388, "global_step": 22001, "epoch": 207, "lr": 6.317364030131123e-05} +{"train_loss": 0.019059740006923676, "grad_norm": 0.07218761742115021, "global_step": 22002, "epoch": 207, "lr": 6.317078123080546e-05} +{"train_loss": 0.0181355569511652, "grad_norm": 0.11755744367837906, "global_step": 22003, "epoch": 207, "lr": 6.316792211402331e-05} +{"train_loss": 0.016979530453681946, "grad_norm": 0.06951164454221725, "global_step": 22004, "epoch": 207, "lr": 6.316506295097479e-05} +{"train_loss": 0.016015324741601944, "grad_norm": 0.05734115093946457, "global_step": 22005, "epoch": 207, "lr": 6.316220374166998e-05} +{"train_loss": 0.012643922120332718, "grad_norm": 0.09045600891113281, "global_step": 22006, "epoch": 207, "lr": 6.31593444861189e-05} +{"train_loss": 0.015001527965068817, "grad_norm": 0.15203794836997986, "global_step": 22007, "epoch": 207, "lr": 6.31564851843316e-05} +{"train_loss": 0.016825104132294655, "grad_norm": 0.06404966115951538, "global_step": 22008, "epoch": 207, "lr": 6.315362583631814e-05} +{"train_loss": 0.015901297330856323, "grad_norm": 0.05378432944417, "global_step": 22009, "epoch": 207, "lr": 6.315076644208858e-05} +{"train_loss": 0.01626300998032093, "grad_norm": 0.07536845654249191, "global_step": 22010, "epoch": 207, "lr": 6.314790700165291e-05} +{"train_loss": 0.016277652233839035, "grad_norm": 0.07431603968143463, "global_step": 22011, "epoch": 207, "lr": 6.314504751502122e-05} +{"train_loss": 0.017668116837739944, "grad_norm": 0.06490553170442581, "global_step": 22012, "epoch": 207, "lr": 6.314218798220357e-05} +{"train_loss": 0.017265107482671738, "grad_norm": 0.11314995586872101, "global_step": 22013, "epoch": 207, "lr": 6.313932840320996e-05} +{"train_loss": 0.016319550573825836, "grad_norm": 0.14607985317707062, "global_step": 22014, "epoch": 207, "lr": 6.313646877805049e-05} +{"train_loss": 0.015920842066407204, "grad_norm": 0.12924537062644958, "global_step": 22015, "epoch": 207, "lr": 6.313360910673516e-05} +{"train_loss": 0.014433987438678741, "grad_norm": 0.11110381782054901, "global_step": 22016, "epoch": 207, "lr": 6.313074938927405e-05} +{"train_loss": 0.01612689718604088, "grad_norm": 0.11844292283058167, "global_step": 22017, "epoch": 207, "lr": 6.31278896256772e-05} +{"train_loss": 0.02096184343099594, "grad_norm": 0.09246623516082764, "global_step": 22018, "epoch": 207, "lr": 6.312502981595465e-05} +{"train_loss": 0.013600455597043037, "grad_norm": 0.08248206973075867, "global_step": 22019, "epoch": 207, "lr": 6.312216996011647e-05} +{"train_loss": 0.014203419908881187, "grad_norm": 0.08563138544559479, "global_step": 22020, "epoch": 207, "lr": 6.311931005817266e-05} +{"train_loss": 0.01563207060098648, "grad_norm": 0.12479773908853531, "global_step": 22021, "epoch": 207, "lr": 6.311645011013331e-05} +{"train_loss": 0.017761006951332092, "grad_norm": 0.09223609417676926, "global_step": 22022, "epoch": 207, "lr": 6.311359011600848e-05} +{"train_loss": 0.018245643004775047, "grad_norm": 0.11977439373731613, "global_step": 22023, "epoch": 207, "lr": 6.311073007580817e-05} +{"train_loss": 0.01833871752023697, "grad_norm": 0.110125832259655, "global_step": 22024, "epoch": 207, "lr": 6.310786998954247e-05} +{"train_loss": 0.015601439401507378, "grad_norm": 0.07160742580890656, "global_step": 22025, "epoch": 207, "lr": 6.31050098572214e-05} +{"train_loss": 0.016755027696490288, "grad_norm": 0.09935586899518967, "global_step": 22026, "epoch": 207, "lr": 6.310214967885506e-05} +{"train_loss": 0.015514413826167583, "grad_norm": 0.08215860277414322, "global_step": 22027, "epoch": 207, "lr": 6.309928945445342e-05} +{"train_loss": 0.015175841748714447, "grad_norm": 0.04921814426779747, "global_step": 22028, "epoch": 207, "lr": 6.309642918402661e-05} +{"train_loss": 0.01554443221539259, "grad_norm": 0.06827785819768906, "global_step": 22029, "epoch": 207, "lr": 6.309356886758464e-05} +{"train_loss": 0.015551792457699776, "grad_norm": 0.07173021882772446, "global_step": 22030, "epoch": 207, "lr": 6.309070850513754e-05} +{"train_loss": 0.012420153245329857, "grad_norm": 0.04369265213608742, "global_step": 22031, "epoch": 207, "lr": 6.308784809669538e-05} +{"train_loss": 0.01873774267733097, "grad_norm": 0.09934913367033005, "global_step": 22032, "epoch": 207, "lr": 6.308498764226824e-05} +{"train_loss": 0.015284476801753044, "grad_norm": 0.09818822145462036, "global_step": 22033, "epoch": 207, "lr": 6.308212714186614e-05} +{"train_loss": 0.01728270947933197, "grad_norm": 0.08551785349845886, "global_step": 22034, "epoch": 207, "lr": 6.307926659549912e-05} +{"train_loss": 0.016309436410665512, "grad_norm": 0.15366214513778687, "global_step": 22035, "epoch": 207, "lr": 6.307640600317726e-05} +{"train_loss": 0.01579943299293518, "grad_norm": 0.07004063576459885, "global_step": 22036, "epoch": 207, "lr": 6.307354536491059e-05} +{"train_loss": 0.0170252937823534, "grad_norm": 0.06124597042798996, "global_step": 22037, "epoch": 207, "lr": 6.307068468070916e-05} +{"train_loss": 0.015826566144824028, "grad_norm": 0.08114569634199142, "global_step": 22038, "epoch": 207, "lr": 6.306782395058304e-05} +{"train_loss": 0.018640335649251938, "grad_norm": 0.05946021154522896, "global_step": 22039, "epoch": 207, "lr": 6.306496317454227e-05} +{"train_loss": 0.014235020615160465, "grad_norm": 0.12238413095474243, "global_step": 22040, "epoch": 207, "lr": 6.306210235259687e-05} +{"train_loss": 0.012930301949381828, "grad_norm": 0.06393264979124069, "global_step": 22041, "epoch": 207, "lr": 6.305924148475698e-05} +{"train_loss": 0.01626618392765522, "grad_norm": 0.101500004529953, "global_step": 22042, "epoch": 207, "lr": 6.305638057103253e-05} +{"train_loss": 0.015204302966594696, "grad_norm": 0.0592053085565567, "global_step": 22043, "epoch": 207, "lr": 6.305351961143368e-05} +{"train_loss": 0.016516387462615967, "grad_norm": 0.05965864285826683, "global_step": 22044, "epoch": 207, "lr": 6.305065860597042e-05} +{"train_loss": 0.016789758577942848, "grad_norm": 0.18167440593242645, "global_step": 22045, "epoch": 207, "lr": 6.304779755465285e-05} +{"train_loss": 0.015708547085523605, "grad_norm": 0.09221319854259491, "global_step": 22046, "epoch": 207, "lr": 6.304493645749097e-05} +{"train_loss": 0.015923315597662947, "grad_norm": 0.09394650161266327, "global_step": 22047, "epoch": 207, "lr": 6.304207531449486e-05, "val_loss": 0.018957288935780525} +{"train_loss": 0.01749356836080551, "grad_norm": 0.20045426487922668, "global_step": 22048, "epoch": 208, "lr": 6.303921412567457e-05} +{"train_loss": 0.01599755883216858, "grad_norm": 0.07071459293365479, "global_step": 22049, "epoch": 208, "lr": 6.303635289104014e-05} +{"train_loss": 0.016063835471868515, "grad_norm": 0.0844649076461792, "global_step": 22050, "epoch": 208, "lr": 6.303349161060165e-05} +{"train_loss": 0.017009474337100983, "grad_norm": 0.06477964669466019, "global_step": 22051, "epoch": 208, "lr": 6.303063028436913e-05} +{"train_loss": 0.016810031607747078, "grad_norm": 0.08201690018177032, "global_step": 22052, "epoch": 208, "lr": 6.302776891235264e-05} +{"train_loss": 0.01617090404033661, "grad_norm": 0.06900449842214584, "global_step": 22053, "epoch": 208, "lr": 6.302490749456226e-05} +{"train_loss": 0.01472448743879795, "grad_norm": 0.1354961097240448, "global_step": 22054, "epoch": 208, "lr": 6.302204603100799e-05} +{"train_loss": 0.016110874712467194, "grad_norm": 0.16081537306308746, "global_step": 22055, "epoch": 208, "lr": 6.301918452169992e-05} +{"train_loss": 0.017004186287522316, "grad_norm": 0.09531269967556, "global_step": 22056, "epoch": 208, "lr": 6.301632296664812e-05} +{"train_loss": 0.016074372455477715, "grad_norm": 0.1405303180217743, "global_step": 22057, "epoch": 208, "lr": 6.301346136586259e-05} +{"train_loss": 0.01564345695078373, "grad_norm": 0.10226229578256607, "global_step": 22058, "epoch": 208, "lr": 6.301059971935344e-05} +{"train_loss": 0.017592009156942368, "grad_norm": 0.09925533831119537, "global_step": 22059, "epoch": 208, "lr": 6.300773802713069e-05} +{"train_loss": 0.015287825837731361, "grad_norm": 0.08409744501113892, "global_step": 22060, "epoch": 208, "lr": 6.30048762892044e-05} +{"train_loss": 0.015040983445942402, "grad_norm": 0.07192770391702652, "global_step": 22061, "epoch": 208, "lr": 6.300201450558464e-05} +{"train_loss": 0.016101641580462456, "grad_norm": 0.1811036616563797, "global_step": 22062, "epoch": 208, "lr": 6.299915267628146e-05} +{"train_loss": 0.013111468404531479, "grad_norm": 0.06069868057966232, "global_step": 22063, "epoch": 208, "lr": 6.29962908013049e-05} +{"train_loss": 0.01645713299512863, "grad_norm": 0.11879180371761322, "global_step": 22064, "epoch": 208, "lr": 6.299342888066502e-05} +{"train_loss": 0.01675121858716011, "grad_norm": 0.13995647430419922, "global_step": 22065, "epoch": 208, "lr": 6.299056691437191e-05} +{"train_loss": 0.01607922650873661, "grad_norm": 0.12252381443977356, "global_step": 22066, "epoch": 208, "lr": 6.298770490243556e-05} +{"train_loss": 0.017790265381336212, "grad_norm": 0.10425348579883575, "global_step": 22067, "epoch": 208, "lr": 6.298484284486608e-05} +{"train_loss": 0.017595916986465454, "grad_norm": 0.09276139736175537, "global_step": 22068, "epoch": 208, "lr": 6.298198074167351e-05} +{"train_loss": 0.012296080589294434, "grad_norm": 0.05875410512089729, "global_step": 22069, "epoch": 208, "lr": 6.297911859286792e-05} +{"train_loss": 0.016176823526620865, "grad_norm": 0.12874634563922882, "global_step": 22070, "epoch": 208, "lr": 6.297625639845934e-05} +{"train_loss": 0.014129556715488434, "grad_norm": 0.06810235977172852, "global_step": 22071, "epoch": 208, "lr": 6.297339415845782e-05} +{"train_loss": 0.016253765672445297, "grad_norm": 0.05390848591923714, "global_step": 22072, "epoch": 208, "lr": 6.297053187287346e-05} +{"train_loss": 0.01847922056913376, "grad_norm": 0.1402573585510254, "global_step": 22073, "epoch": 208, "lr": 6.296766954171628e-05} +{"train_loss": 0.01539432629942894, "grad_norm": 0.05582919344305992, "global_step": 22074, "epoch": 208, "lr": 6.296480716499636e-05} +{"train_loss": 0.014905773103237152, "grad_norm": 0.07853275537490845, "global_step": 22075, "epoch": 208, "lr": 6.296194474272373e-05} +{"train_loss": 0.017894750460982323, "grad_norm": 0.06563897430896759, "global_step": 22076, "epoch": 208, "lr": 6.295908227490848e-05} +{"train_loss": 0.016095338389277458, "grad_norm": 0.10013724863529205, "global_step": 22077, "epoch": 208, "lr": 6.295621976156063e-05} +{"train_loss": 0.01687769964337349, "grad_norm": 0.09068334847688675, "global_step": 22078, "epoch": 208, "lr": 6.295335720269027e-05} +{"train_loss": 0.018430635333061218, "grad_norm": 0.07826784998178482, "global_step": 22079, "epoch": 208, "lr": 6.295049459830746e-05} +{"train_loss": 0.016634609550237656, "grad_norm": 0.06149410828948021, "global_step": 22080, "epoch": 208, "lr": 6.294763194842223e-05} +{"train_loss": 0.015285500325262547, "grad_norm": 0.09660937637090683, "global_step": 22081, "epoch": 208, "lr": 6.294476925304463e-05} +{"train_loss": 0.013836905360221863, "grad_norm": 0.0906410813331604, "global_step": 22082, "epoch": 208, "lr": 6.294190651218478e-05} +{"train_loss": 0.0159841850399971, "grad_norm": 0.07220091670751572, "global_step": 22083, "epoch": 208, "lr": 6.293904372585267e-05} +{"train_loss": 0.01655968837440014, "grad_norm": 0.07418446987867355, "global_step": 22084, "epoch": 208, "lr": 6.29361808940584e-05} +{"train_loss": 0.015896901488304138, "grad_norm": 0.07052387297153473, "global_step": 22085, "epoch": 208, "lr": 6.293331801681201e-05} +{"train_loss": 0.014700232073664665, "grad_norm": 0.0771370381116867, "global_step": 22086, "epoch": 208, "lr": 6.293045509412357e-05} +{"train_loss": 0.015307262539863586, "grad_norm": 0.06715691834688187, "global_step": 22087, "epoch": 208, "lr": 6.292759212600313e-05} +{"train_loss": 0.013977322727441788, "grad_norm": 0.06105664372444153, "global_step": 22088, "epoch": 208, "lr": 6.292472911246076e-05} +{"train_loss": 0.016031380742788315, "grad_norm": 0.11264406889677048, "global_step": 22089, "epoch": 208, "lr": 6.292186605350651e-05} +{"train_loss": 0.016089368611574173, "grad_norm": 0.051719747483730316, "global_step": 22090, "epoch": 208, "lr": 6.291900294915046e-05} +{"train_loss": 0.015131904743611813, "grad_norm": 0.14677050709724426, "global_step": 22091, "epoch": 208, "lr": 6.291613979940262e-05} +{"train_loss": 0.015202630311250687, "grad_norm": 0.14999322593212128, "global_step": 22092, "epoch": 208, "lr": 6.29132766042731e-05} +{"train_loss": 0.015566864982247353, "grad_norm": 0.1307961791753769, "global_step": 22093, "epoch": 208, "lr": 6.291041336377193e-05} +{"train_loss": 0.012910483404994011, "grad_norm": 0.11129117757081985, "global_step": 22094, "epoch": 208, "lr": 6.290755007790919e-05} +{"train_loss": 0.013330556452274323, "grad_norm": 0.10693199187517166, "global_step": 22095, "epoch": 208, "lr": 6.290468674669493e-05} +{"train_loss": 0.016706326976418495, "grad_norm": 0.07513386011123657, "global_step": 22096, "epoch": 208, "lr": 6.290182337013921e-05} +{"train_loss": 0.01697273552417755, "grad_norm": 0.10658592730760574, "global_step": 22097, "epoch": 208, "lr": 6.289895994825211e-05} +{"train_loss": 0.017970632761716843, "grad_norm": 0.05846360698342323, "global_step": 22098, "epoch": 208, "lr": 6.289609648104366e-05} +{"train_loss": 0.01412421278655529, "grad_norm": 0.06812511384487152, "global_step": 22099, "epoch": 208, "lr": 6.289323296852393e-05} +{"train_loss": 0.016676580533385277, "grad_norm": 0.12354016304016113, "global_step": 22100, "epoch": 208, "lr": 6.2890369410703e-05} +{"train_loss": 0.017065631225705147, "grad_norm": 0.1254308670759201, "global_step": 22101, "epoch": 208, "lr": 6.28875058075909e-05} +{"train_loss": 0.01489417627453804, "grad_norm": 0.09102915972471237, "global_step": 22102, "epoch": 208, "lr": 6.288464215919774e-05} +{"train_loss": 0.014947920106351376, "grad_norm": 0.05884338542819023, "global_step": 22103, "epoch": 208, "lr": 6.288177846553353e-05} +{"train_loss": 0.014471222646534443, "grad_norm": 0.05991145223379135, "global_step": 22104, "epoch": 208, "lr": 6.287891472660835e-05} +{"train_loss": 0.01573445275425911, "grad_norm": 0.15281973779201508, "global_step": 22105, "epoch": 208, "lr": 6.287605094243226e-05} +{"train_loss": 0.015274971723556519, "grad_norm": 0.18319092690944672, "global_step": 22106, "epoch": 208, "lr": 6.287318711301535e-05} +{"train_loss": 0.016634216532111168, "grad_norm": 0.09282511472702026, "global_step": 22107, "epoch": 208, "lr": 6.287032323836763e-05} +{"train_loss": 0.019045475870370865, "grad_norm": 0.11518394201993942, "global_step": 22108, "epoch": 208, "lr": 6.286745931849922e-05} +{"train_loss": 0.01940513774752617, "grad_norm": 0.12940475344657898, "global_step": 22109, "epoch": 208, "lr": 6.286459535342014e-05} +{"train_loss": 0.01702798530459404, "grad_norm": 0.14749832451343536, "global_step": 22110, "epoch": 208, "lr": 6.286173134314046e-05} +{"train_loss": 0.01612863689661026, "grad_norm": 0.07444845139980316, "global_step": 22111, "epoch": 208, "lr": 6.285886728767026e-05} +{"train_loss": 0.019344432279467583, "grad_norm": 0.11948349326848984, "global_step": 22112, "epoch": 208, "lr": 6.285600318701959e-05} +{"train_loss": 0.017166636884212494, "grad_norm": 0.07292699068784714, "global_step": 22113, "epoch": 208, "lr": 6.28531390411985e-05} +{"train_loss": 0.014400122687220573, "grad_norm": 0.154367595911026, "global_step": 22114, "epoch": 208, "lr": 6.28502748502171e-05} +{"train_loss": 0.01647433452308178, "grad_norm": 0.15563848614692688, "global_step": 22115, "epoch": 208, "lr": 6.284741061408541e-05} +{"train_loss": 0.016332514584064484, "grad_norm": 0.13455940783023834, "global_step": 22116, "epoch": 208, "lr": 6.284454633281349e-05} +{"train_loss": 0.016845032572746277, "grad_norm": 0.1496414989233017, "global_step": 22117, "epoch": 208, "lr": 6.284168200641143e-05} +{"train_loss": 0.015659168362617493, "grad_norm": 0.10030996799468994, "global_step": 22118, "epoch": 208, "lr": 6.283881763488929e-05} +{"train_loss": 0.012752356007695198, "grad_norm": 0.05677887797355652, "global_step": 22119, "epoch": 208, "lr": 6.283595321825713e-05} +{"train_loss": 0.013132991269230843, "grad_norm": 0.13065916299819946, "global_step": 22120, "epoch": 208, "lr": 6.2833088756525e-05} +{"train_loss": 0.01739928126335144, "grad_norm": 0.06972116231918335, "global_step": 22121, "epoch": 208, "lr": 6.2830224249703e-05} +{"train_loss": 0.020089613273739815, "grad_norm": 0.08222822844982147, "global_step": 22122, "epoch": 208, "lr": 6.282735969780114e-05} +{"train_loss": 0.01357826218008995, "grad_norm": 0.06577584892511368, "global_step": 22123, "epoch": 208, "lr": 6.282449510082952e-05} +{"train_loss": 0.01919931173324585, "grad_norm": 0.10047587007284164, "global_step": 22124, "epoch": 208, "lr": 6.282163045879823e-05} +{"train_loss": 0.014662089757621288, "grad_norm": 0.07584488391876221, "global_step": 22125, "epoch": 208, "lr": 6.28187657717173e-05} +{"train_loss": 0.014670148491859436, "grad_norm": 0.05547865480184555, "global_step": 22126, "epoch": 208, "lr": 6.281590103959679e-05} +{"train_loss": 0.015448689460754395, "grad_norm": 0.18163840472698212, "global_step": 22127, "epoch": 208, "lr": 6.281303626244676e-05} +{"train_loss": 0.0178135484457016, "grad_norm": 0.0718865841627121, "global_step": 22128, "epoch": 208, "lr": 6.281017144027732e-05} +{"train_loss": 0.016356024891138077, "grad_norm": 0.07067427784204483, "global_step": 22129, "epoch": 208, "lr": 6.28073065730985e-05} +{"train_loss": 0.013352281413972378, "grad_norm": 0.1078362762928009, "global_step": 22130, "epoch": 208, "lr": 6.280444166092039e-05} +{"train_loss": 0.014641864225268364, "grad_norm": 0.07676048576831818, "global_step": 22131, "epoch": 208, "lr": 6.280157670375301e-05} +{"train_loss": 0.015161757357418537, "grad_norm": 0.10539352148771286, "global_step": 22132, "epoch": 208, "lr": 6.279871170160648e-05} +{"train_loss": 0.017038628458976746, "grad_norm": 0.11979108303785324, "global_step": 22133, "epoch": 208, "lr": 6.279584665449084e-05} +{"train_loss": 0.01549489889293909, "grad_norm": 0.06927001476287842, "global_step": 22134, "epoch": 208, "lr": 6.279298156241615e-05} +{"train_loss": 0.01507275179028511, "grad_norm": 0.07335930317640305, "global_step": 22135, "epoch": 208, "lr": 6.27901164253925e-05} +{"train_loss": 0.015445886179804802, "grad_norm": 0.09591080993413925, "global_step": 22136, "epoch": 208, "lr": 6.278725124342992e-05} +{"train_loss": 0.013500344008207321, "grad_norm": 0.09644000977277756, "global_step": 22137, "epoch": 208, "lr": 6.278438601653852e-05} +{"train_loss": 0.015948239713907242, "grad_norm": 0.06563997268676758, "global_step": 22138, "epoch": 208, "lr": 6.278152074472837e-05} +{"train_loss": 0.017970846965909004, "grad_norm": 0.06962087750434875, "global_step": 22139, "epoch": 208, "lr": 6.277865542800946e-05} +{"train_loss": 0.015736933797597885, "grad_norm": 0.07930740714073181, "global_step": 22140, "epoch": 208, "lr": 6.277579006639194e-05} +{"train_loss": 0.015937354415655136, "grad_norm": 0.06967473030090332, "global_step": 22141, "epoch": 208, "lr": 6.277292465988585e-05} +{"train_loss": 0.012818785384297371, "grad_norm": 0.06395111232995987, "global_step": 22142, "epoch": 208, "lr": 6.277005920850126e-05} +{"train_loss": 0.01708328351378441, "grad_norm": 0.06609511375427246, "global_step": 22143, "epoch": 208, "lr": 6.276719371224823e-05} +{"train_loss": 0.015266239643096924, "grad_norm": 0.06918609142303467, "global_step": 22144, "epoch": 208, "lr": 6.276432817113684e-05} +{"train_loss": 0.018605703487992287, "grad_norm": 0.10792641341686249, "global_step": 22145, "epoch": 208, "lr": 6.276146258517714e-05} +{"train_loss": 0.015615671873092651, "grad_norm": 0.07766368240118027, "global_step": 22146, "epoch": 208, "lr": 6.275859695437921e-05} +{"train_loss": 0.01839689537882805, "grad_norm": 0.08737577497959137, "global_step": 22147, "epoch": 208, "lr": 6.275573127875313e-05} +{"train_loss": 0.013576269149780273, "grad_norm": 0.07443741708993912, "global_step": 22148, "epoch": 208, "lr": 6.275286555830897e-05} +{"train_loss": 0.0189824178814888, "grad_norm": 0.1228606328368187, "global_step": 22149, "epoch": 208, "lr": 6.274999979305675e-05} +{"train_loss": 0.014520796947181225, "grad_norm": 0.07698127627372742, "global_step": 22150, "epoch": 208, "lr": 6.274713398300662e-05} +{"train_loss": 0.015627775341272354, "grad_norm": 0.0942479595541954, "global_step": 22151, "epoch": 208, "lr": 6.274426812816857e-05} +{"train_loss": 0.01587371528148651, "grad_norm": 0.10625436902046204, "global_step": 22152, "epoch": 208, "lr": 6.274140222855271e-05} +{"train_loss": 0.015943618605510808, "grad_norm": 0.07444816082715988, "global_step": 22153, "epoch": 208, "lr": 6.273853628416911e-05, "val_loss": 0.01906854659318924} +{"train_loss": 0.014522754587233067, "grad_norm": 0.04864024743437767, "global_step": 22154, "epoch": 209, "lr": 6.273567029502783e-05} +{"train_loss": 0.01475323922932148, "grad_norm": 0.05967302992939949, "global_step": 22155, "epoch": 209, "lr": 6.273280426113895e-05} +{"train_loss": 0.017994174733757973, "grad_norm": 0.1011260598897934, "global_step": 22156, "epoch": 209, "lr": 6.272993818251254e-05} +{"train_loss": 0.015762515366077423, "grad_norm": 0.07683460414409637, "global_step": 22157, "epoch": 209, "lr": 6.272707205915865e-05} +{"train_loss": 0.016827311366796494, "grad_norm": 0.09495527297258377, "global_step": 22158, "epoch": 209, "lr": 6.272420589108735e-05} +{"train_loss": 0.015870941802859306, "grad_norm": 0.08230182528495789, "global_step": 22159, "epoch": 209, "lr": 6.272133967830875e-05} +{"train_loss": 0.0163734033703804, "grad_norm": 0.11654860526323318, "global_step": 22160, "epoch": 209, "lr": 6.271847342083288e-05} +{"train_loss": 0.01727224886417389, "grad_norm": 0.07867121696472168, "global_step": 22161, "epoch": 209, "lr": 6.271560711866983e-05} +{"train_loss": 0.014915351755917072, "grad_norm": 0.06882168352603912, "global_step": 22162, "epoch": 209, "lr": 6.271274077182967e-05} +{"train_loss": 0.014214766211807728, "grad_norm": 0.08567960560321808, "global_step": 22163, "epoch": 209, "lr": 6.270987438032246e-05} +{"train_loss": 0.017032761126756668, "grad_norm": 0.06491614133119583, "global_step": 22164, "epoch": 209, "lr": 6.270700794415828e-05} +{"train_loss": 0.015759209170937538, "grad_norm": 0.0866013690829277, "global_step": 22165, "epoch": 209, "lr": 6.27041414633472e-05} +{"train_loss": 0.016002655029296875, "grad_norm": 0.09325561672449112, "global_step": 22166, "epoch": 209, "lr": 6.27012749378993e-05} +{"train_loss": 0.01421617902815342, "grad_norm": 0.0686461329460144, "global_step": 22167, "epoch": 209, "lr": 6.269840836782464e-05} +{"train_loss": 0.01403297483921051, "grad_norm": 0.06980384886264801, "global_step": 22168, "epoch": 209, "lr": 6.269554175313329e-05} +{"train_loss": 0.015725547447800636, "grad_norm": 0.06284435838460922, "global_step": 22169, "epoch": 209, "lr": 6.269267509383534e-05} +{"train_loss": 0.014050271362066269, "grad_norm": 0.1529197245836258, "global_step": 22170, "epoch": 209, "lr": 6.268980838994083e-05} +{"train_loss": 0.013234341517090797, "grad_norm": 0.0714406743645668, "global_step": 22171, "epoch": 209, "lr": 6.268694164145988e-05} +{"train_loss": 0.014720192179083824, "grad_norm": 0.11870044469833374, "global_step": 22172, "epoch": 209, "lr": 6.268407484840251e-05} +{"train_loss": 0.01999887451529503, "grad_norm": 0.08064909279346466, "global_step": 22173, "epoch": 209, "lr": 6.268120801077881e-05} +{"train_loss": 0.015574874356389046, "grad_norm": 0.057035550475120544, "global_step": 22174, "epoch": 209, "lr": 6.267834112859888e-05} +{"train_loss": 0.016354326158761978, "grad_norm": 0.059428632259368896, "global_step": 22175, "epoch": 209, "lr": 6.267547420187276e-05} +{"train_loss": 0.018145499750971794, "grad_norm": 0.07110060006380081, "global_step": 22176, "epoch": 209, "lr": 6.267260723061056e-05} +{"train_loss": 0.015090596862137318, "grad_norm": 0.09154713153839111, "global_step": 22177, "epoch": 209, "lr": 6.266974021482232e-05} +{"train_loss": 0.014132575131952763, "grad_norm": 0.09402184188365936, "global_step": 22178, "epoch": 209, "lr": 6.266687315451809e-05} +{"train_loss": 0.01764860935509205, "grad_norm": 0.07777629792690277, "global_step": 22179, "epoch": 209, "lr": 6.266400604970802e-05} +{"train_loss": 0.01519980002194643, "grad_norm": 0.06559764593839645, "global_step": 22180, "epoch": 209, "lr": 6.266113890040211e-05} +{"train_loss": 0.016381364315748215, "grad_norm": 0.0722036212682724, "global_step": 22181, "epoch": 209, "lr": 6.265827170661048e-05} +{"train_loss": 0.015625905245542526, "grad_norm": 0.06822850555181503, "global_step": 22182, "epoch": 209, "lr": 6.265540446834318e-05} +{"train_loss": 0.014669675379991531, "grad_norm": 0.06602784246206284, "global_step": 22183, "epoch": 209, "lr": 6.265253718561031e-05} +{"train_loss": 0.01406717486679554, "grad_norm": 0.07510772347450256, "global_step": 22184, "epoch": 209, "lr": 6.264966985842191e-05} +{"train_loss": 0.014951037243008614, "grad_norm": 0.17208708822727203, "global_step": 22185, "epoch": 209, "lr": 6.264680248678808e-05} +{"train_loss": 0.015553346835076809, "grad_norm": 0.07193507254123688, "global_step": 22186, "epoch": 209, "lr": 6.264393507071887e-05} +{"train_loss": 0.01582208275794983, "grad_norm": 0.06552013754844666, "global_step": 22187, "epoch": 209, "lr": 6.264106761022437e-05} +{"train_loss": 0.014099819585680962, "grad_norm": 0.05522608011960983, "global_step": 22188, "epoch": 209, "lr": 6.263820010531468e-05} +{"train_loss": 0.016627158969640732, "grad_norm": 0.1017174944281578, "global_step": 22189, "epoch": 209, "lr": 6.263533255599985e-05} +{"train_loss": 0.016311844810843468, "grad_norm": 0.09324970096349716, "global_step": 22190, "epoch": 209, "lr": 6.263246496228992e-05} +{"train_loss": 0.017870964482426643, "grad_norm": 0.06520089507102966, "global_step": 22191, "epoch": 209, "lr": 6.262959732419504e-05} +{"train_loss": 0.016147026792168617, "grad_norm": 0.11344914883375168, "global_step": 22192, "epoch": 209, "lr": 6.262672964172522e-05} +{"train_loss": 0.012990236282348633, "grad_norm": 0.052174050360918045, "global_step": 22193, "epoch": 209, "lr": 6.262386191489059e-05} +{"train_loss": 0.015377519652247429, "grad_norm": 0.10774970799684525, "global_step": 22194, "epoch": 209, "lr": 6.262099414370118e-05} +{"train_loss": 0.01672421582043171, "grad_norm": 0.08244992792606354, "global_step": 22195, "epoch": 209, "lr": 6.26181263281671e-05} +{"train_loss": 0.01648217812180519, "grad_norm": 0.0650024265050888, "global_step": 22196, "epoch": 209, "lr": 6.26152584682984e-05} +{"train_loss": 0.01785242184996605, "grad_norm": 0.16474081575870514, "global_step": 22197, "epoch": 209, "lr": 6.261239056410518e-05} +{"train_loss": 0.0146012082695961, "grad_norm": 0.06996162235736847, "global_step": 22198, "epoch": 209, "lr": 6.26095226155975e-05} +{"train_loss": 0.0171107966452837, "grad_norm": 0.06276430934667587, "global_step": 22199, "epoch": 209, "lr": 6.260665462278544e-05} +{"train_loss": 0.014574929140508175, "grad_norm": 0.09864567965269089, "global_step": 22200, "epoch": 209, "lr": 6.260378658567907e-05} +{"train_loss": 0.015555393882095814, "grad_norm": 0.06792142242193222, "global_step": 22201, "epoch": 209, "lr": 6.260091850428847e-05} +{"train_loss": 0.019354531541466713, "grad_norm": 0.1206243708729744, "global_step": 22202, "epoch": 209, "lr": 6.259805037862373e-05} +{"train_loss": 0.016914237290620804, "grad_norm": 0.0731128677725792, "global_step": 22203, "epoch": 209, "lr": 6.259518220869493e-05} +{"train_loss": 0.01779812015593052, "grad_norm": 0.08685001730918884, "global_step": 22204, "epoch": 209, "lr": 6.259231399451211e-05} +{"train_loss": 0.017415393143892288, "grad_norm": 0.1186940148472786, "global_step": 22205, "epoch": 209, "lr": 6.25894457360854e-05} +{"train_loss": 0.01574925146996975, "grad_norm": 0.07149659097194672, "global_step": 22206, "epoch": 209, "lr": 6.258657743342486e-05} +{"train_loss": 0.016824673861265182, "grad_norm": 0.11675009876489639, "global_step": 22207, "epoch": 209, "lr": 6.258370908654055e-05} +{"train_loss": 0.015114828944206238, "grad_norm": 0.05582493171095848, "global_step": 22208, "epoch": 209, "lr": 6.258084069544256e-05} +{"train_loss": 0.016872577369213104, "grad_norm": 0.10956836491823196, "global_step": 22209, "epoch": 209, "lr": 6.257797226014096e-05} +{"train_loss": 0.015338508412241936, "grad_norm": 0.12543362379074097, "global_step": 22210, "epoch": 209, "lr": 6.257510378064584e-05} +{"train_loss": 0.016141239553689957, "grad_norm": 0.08865508437156677, "global_step": 22211, "epoch": 209, "lr": 6.257223525696726e-05} +{"train_loss": 0.014729894697666168, "grad_norm": 0.11051714420318604, "global_step": 22212, "epoch": 209, "lr": 6.256936668911534e-05} +{"train_loss": 0.0166034959256649, "grad_norm": 0.05105689913034439, "global_step": 22213, "epoch": 209, "lr": 6.256649807710012e-05} +{"train_loss": 0.01586647517979145, "grad_norm": 0.09054414927959442, "global_step": 22214, "epoch": 209, "lr": 6.256362942093168e-05} +{"train_loss": 0.014902394264936447, "grad_norm": 0.058245111256837845, "global_step": 22215, "epoch": 209, "lr": 6.256076072062012e-05} +{"train_loss": 0.015978192910552025, "grad_norm": 0.05315937101840973, "global_step": 22216, "epoch": 209, "lr": 6.255789197617552e-05} +{"train_loss": 0.015555125661194324, "grad_norm": 0.08128491789102554, "global_step": 22217, "epoch": 209, "lr": 6.255502318760792e-05} +{"train_loss": 0.0140545554459095, "grad_norm": 0.047430574893951416, "global_step": 22218, "epoch": 209, "lr": 6.255215435492746e-05} +{"train_loss": 0.01688859798014164, "grad_norm": 0.09143611043691635, "global_step": 22219, "epoch": 209, "lr": 6.254928547814416e-05} +{"train_loss": 0.0151487085968256, "grad_norm": 0.07423879951238632, "global_step": 22220, "epoch": 209, "lr": 6.254641655726815e-05} +{"train_loss": 0.015862300992012024, "grad_norm": 0.08316123485565186, "global_step": 22221, "epoch": 209, "lr": 6.254354759230948e-05} +{"train_loss": 0.016320331022143364, "grad_norm": 0.0810021162033081, "global_step": 22222, "epoch": 209, "lr": 6.254067858327824e-05} +{"train_loss": 0.016242485493421555, "grad_norm": 0.07132884114980698, "global_step": 22223, "epoch": 209, "lr": 6.25378095301845e-05} +{"train_loss": 0.01559268869459629, "grad_norm": 0.07901477813720703, "global_step": 22224, "epoch": 209, "lr": 6.253494043303836e-05} +{"train_loss": 0.01739344373345375, "grad_norm": 0.07228703796863556, "global_step": 22225, "epoch": 209, "lr": 6.253207129184991e-05} +{"train_loss": 0.015270061790943146, "grad_norm": 0.05984459072351456, "global_step": 22226, "epoch": 209, "lr": 6.252920210662919e-05} +{"train_loss": 0.01868542470037937, "grad_norm": 0.11328183859586716, "global_step": 22227, "epoch": 209, "lr": 6.252633287738629e-05} +{"train_loss": 0.016931526362895966, "grad_norm": 0.09098894149065018, "global_step": 22228, "epoch": 209, "lr": 6.252346360413133e-05} +{"train_loss": 0.014140117913484573, "grad_norm": 0.09193792194128036, "global_step": 22229, "epoch": 209, "lr": 6.252059428687434e-05} +{"train_loss": 0.016274333000183105, "grad_norm": 0.14902399480342865, "global_step": 22230, "epoch": 209, "lr": 6.251772492562544e-05} +{"train_loss": 0.01731613092124462, "grad_norm": 0.06731373071670532, "global_step": 22231, "epoch": 209, "lr": 6.251485552039468e-05} +{"train_loss": 0.01414186879992485, "grad_norm": 0.06535898894071579, "global_step": 22232, "epoch": 209, "lr": 6.251198607119221e-05} +{"train_loss": 0.017287015914916992, "grad_norm": 0.08579995483160019, "global_step": 22233, "epoch": 209, "lr": 6.250911657802801e-05} +{"train_loss": 0.01858443394303322, "grad_norm": 0.08985315263271332, "global_step": 22234, "epoch": 209, "lr": 6.250624704091224e-05} +{"train_loss": 0.016550030559301376, "grad_norm": 0.10375042259693146, "global_step": 22235, "epoch": 209, "lr": 6.250337745985497e-05} +{"train_loss": 0.014381559565663338, "grad_norm": 0.06383075565099716, "global_step": 22236, "epoch": 209, "lr": 6.250050783486623e-05} +{"train_loss": 0.01643061265349388, "grad_norm": 0.13766638934612274, "global_step": 22237, "epoch": 209, "lr": 6.249763816595616e-05} +{"train_loss": 0.0180064644664526, "grad_norm": 0.0672001987695694, "global_step": 22238, "epoch": 209, "lr": 6.249476845313484e-05} +{"train_loss": 0.012934043072164059, "grad_norm": 0.06872405111789703, "global_step": 22239, "epoch": 209, "lr": 6.249189869641233e-05} +{"train_loss": 0.016407156363129616, "grad_norm": 0.1174374595284462, "global_step": 22240, "epoch": 209, "lr": 6.248902889579871e-05} +{"train_loss": 0.014026167802512646, "grad_norm": 0.0774659663438797, "global_step": 22241, "epoch": 209, "lr": 6.248615905130407e-05} +{"train_loss": 0.014995639212429523, "grad_norm": 0.1556662619113922, "global_step": 22242, "epoch": 209, "lr": 6.248328916293851e-05} +{"train_loss": 0.0174742192029953, "grad_norm": 0.08347020298242569, "global_step": 22243, "epoch": 209, "lr": 6.248041923071209e-05} +{"train_loss": 0.017970845103263855, "grad_norm": 0.08434256166219711, "global_step": 22244, "epoch": 209, "lr": 6.247754925463492e-05} +{"train_loss": 0.016895567998290062, "grad_norm": 0.059046898037195206, "global_step": 22245, "epoch": 209, "lr": 6.247467923471705e-05} +{"train_loss": 0.01766144670546055, "grad_norm": 0.12042922526597977, "global_step": 22246, "epoch": 209, "lr": 6.24718091709686e-05} +{"train_loss": 0.017323046922683716, "grad_norm": 0.14442332088947296, "global_step": 22247, "epoch": 209, "lr": 6.246893906339963e-05} +{"train_loss": 0.014757477678358555, "grad_norm": 0.16387100517749786, "global_step": 22248, "epoch": 209, "lr": 6.246606891202022e-05} +{"train_loss": 0.014803841710090637, "grad_norm": 0.08230804651975632, "global_step": 22249, "epoch": 209, "lr": 6.246319871684048e-05} +{"train_loss": 0.014931490644812584, "grad_norm": 0.07211115211248398, "global_step": 22250, "epoch": 209, "lr": 6.246032847787047e-05} +{"train_loss": 0.012712173163890839, "grad_norm": 0.07525090873241425, "global_step": 22251, "epoch": 209, "lr": 6.245745819512029e-05} +{"train_loss": 0.014687019400298595, "grad_norm": 0.0643862932920456, "global_step": 22252, "epoch": 209, "lr": 6.245458786860003e-05} +{"train_loss": 0.015060844831168652, "grad_norm": 0.07290265709161758, "global_step": 22253, "epoch": 209, "lr": 6.245171749831974e-05} +{"train_loss": 0.015150769613683224, "grad_norm": 0.11017557233572006, "global_step": 22254, "epoch": 209, "lr": 6.244884708428954e-05} +{"train_loss": 0.014276832342147827, "grad_norm": 0.0510827973484993, "global_step": 22255, "epoch": 209, "lr": 6.24459766265195e-05} +{"train_loss": 0.014385299757122993, "grad_norm": 0.08676110953092575, "global_step": 22256, "epoch": 209, "lr": 6.244310612501974e-05} +{"train_loss": 0.017534475773572922, "grad_norm": 0.08157628774642944, "global_step": 22257, "epoch": 209, "lr": 6.244023557980029e-05} +{"train_loss": 0.01696273311972618, "grad_norm": 0.1383904665708542, "global_step": 22258, "epoch": 209, "lr": 6.243736499087125e-05} +{"train_loss": 0.015894714365098275, "grad_norm": 0.06643647700548172, "global_step": 22259, "epoch": 209, "lr": 6.243449435824276e-05, "val_loss": 0.0192640982568264} +{"train_loss": 0.016063613817095757, "grad_norm": 0.130844846367836, "global_step": 22260, "epoch": 210, "lr": 6.243162368192482e-05} +{"train_loss": 0.015670742839574814, "grad_norm": 0.09087876230478287, "global_step": 22261, "epoch": 210, "lr": 6.242875296192758e-05} +{"train_loss": 0.015232774429023266, "grad_norm": 0.06288009881973267, "global_step": 22262, "epoch": 210, "lr": 6.242588219826111e-05} +{"train_loss": 0.013899097219109535, "grad_norm": 0.07202231138944626, "global_step": 22263, "epoch": 210, "lr": 6.24230113909355e-05} +{"train_loss": 0.015682537108659744, "grad_norm": 0.11704086512327194, "global_step": 22264, "epoch": 210, "lr": 6.242014053996082e-05} +{"train_loss": 0.014805009588599205, "grad_norm": 0.06530243903398514, "global_step": 22265, "epoch": 210, "lr": 6.241726964534717e-05} +{"train_loss": 0.014070217497646809, "grad_norm": 0.088016577064991, "global_step": 22266, "epoch": 210, "lr": 6.241439870710462e-05} +{"train_loss": 0.016225798055529594, "grad_norm": 0.06994492560625076, "global_step": 22267, "epoch": 210, "lr": 6.241152772524329e-05} +{"train_loss": 0.016544118523597717, "grad_norm": 0.1081104725599289, "global_step": 22268, "epoch": 210, "lr": 6.240865669977324e-05} +{"train_loss": 0.016999034211039543, "grad_norm": 0.07782840728759766, "global_step": 22269, "epoch": 210, "lr": 6.240578563070459e-05} +{"train_loss": 0.018783755600452423, "grad_norm": 0.05971760302782059, "global_step": 22270, "epoch": 210, "lr": 6.240291451804737e-05} +{"train_loss": 0.015758642926812172, "grad_norm": 0.08032383769750595, "global_step": 22271, "epoch": 210, "lr": 6.240004336181174e-05} +{"train_loss": 0.0149723831564188, "grad_norm": 0.11201485246419907, "global_step": 22272, "epoch": 210, "lr": 6.23971721620077e-05} +{"train_loss": 0.01586558297276497, "grad_norm": 0.07822033762931824, "global_step": 22273, "epoch": 210, "lr": 6.239430091864543e-05} +{"train_loss": 0.01801934838294983, "grad_norm": 0.05591136962175369, "global_step": 22274, "epoch": 210, "lr": 6.239142963173497e-05} +{"train_loss": 0.01388456393033266, "grad_norm": 0.054587431252002716, "global_step": 22275, "epoch": 210, "lr": 6.23885583012864e-05} +{"train_loss": 0.015510962344706059, "grad_norm": 0.06572332233190536, "global_step": 22276, "epoch": 210, "lr": 6.238568692730984e-05} +{"train_loss": 0.01690264791250229, "grad_norm": 0.10619863122701645, "global_step": 22277, "epoch": 210, "lr": 6.238281550981536e-05} +{"train_loss": 0.015742279589176178, "grad_norm": 0.05809535086154938, "global_step": 22278, "epoch": 210, "lr": 6.237994404881305e-05} +{"train_loss": 0.01663312315940857, "grad_norm": 0.056922320276498795, "global_step": 22279, "epoch": 210, "lr": 6.237707254431298e-05} +{"train_loss": 0.01576128415763378, "grad_norm": 0.06465401500463486, "global_step": 22280, "epoch": 210, "lr": 6.237420099632529e-05} +{"train_loss": 0.017361972481012344, "grad_norm": 0.1159091666340828, "global_step": 22281, "epoch": 210, "lr": 6.237132940486003e-05} +{"train_loss": 0.01634722575545311, "grad_norm": 0.09715783596038818, "global_step": 22282, "epoch": 210, "lr": 6.23684577699273e-05} +{"train_loss": 0.013357177376747131, "grad_norm": 0.054496120661497116, "global_step": 22283, "epoch": 210, "lr": 6.236558609153718e-05} +{"train_loss": 0.01682971604168415, "grad_norm": 0.20149318873882294, "global_step": 22284, "epoch": 210, "lr": 6.236271436969976e-05} +{"train_loss": 0.016072895377874374, "grad_norm": 0.06953061372041702, "global_step": 22285, "epoch": 210, "lr": 6.235984260442516e-05} +{"train_loss": 0.015100350603461266, "grad_norm": 0.07092943787574768, "global_step": 22286, "epoch": 210, "lr": 6.235697079572343e-05} +{"train_loss": 0.015836201608181, "grad_norm": 0.06494711339473724, "global_step": 22287, "epoch": 210, "lr": 6.235409894360469e-05} +{"train_loss": 0.013959027826786041, "grad_norm": 0.08180435001850128, "global_step": 22288, "epoch": 210, "lr": 6.235122704807901e-05} +{"train_loss": 0.016969231888651848, "grad_norm": 0.11602859199047089, "global_step": 22289, "epoch": 210, "lr": 6.23483551091565e-05} +{"train_loss": 0.014005021192133427, "grad_norm": 0.12628920376300812, "global_step": 22290, "epoch": 210, "lr": 6.234548312684723e-05} +{"train_loss": 0.015492177568376064, "grad_norm": 0.06136773154139519, "global_step": 22291, "epoch": 210, "lr": 6.234261110116131e-05} +{"train_loss": 0.01642942614853382, "grad_norm": 0.1385558396577835, "global_step": 22292, "epoch": 210, "lr": 6.233973903210881e-05} +{"train_loss": 0.015245168469846249, "grad_norm": 0.08907204121351242, "global_step": 22293, "epoch": 210, "lr": 6.233686691969984e-05} +{"train_loss": 0.017188183963298798, "grad_norm": 0.060973457992076874, "global_step": 22294, "epoch": 210, "lr": 6.233399476394449e-05} +{"train_loss": 0.015573074109852314, "grad_norm": 0.06304018199443817, "global_step": 22295, "epoch": 210, "lr": 6.233112256485285e-05} +{"train_loss": 0.016401754692196846, "grad_norm": 0.06889478117227554, "global_step": 22296, "epoch": 210, "lr": 6.232825032243498e-05} +{"train_loss": 0.013986831530928612, "grad_norm": 0.0908733531832695, "global_step": 22297, "epoch": 210, "lr": 6.232537803670105e-05} +{"train_loss": 0.014284428209066391, "grad_norm": 0.06374262273311615, "global_step": 22298, "epoch": 210, "lr": 6.232250570766107e-05} +{"train_loss": 0.014555367641150951, "grad_norm": 0.06244324892759323, "global_step": 22299, "epoch": 210, "lr": 6.231963333532515e-05} +{"train_loss": 0.014559727162122726, "grad_norm": 0.04843738302588463, "global_step": 22300, "epoch": 210, "lr": 6.231676091970343e-05} +{"train_loss": 0.015827637165784836, "grad_norm": 0.07045713812112808, "global_step": 22301, "epoch": 210, "lr": 6.231388846080594e-05} +{"train_loss": 0.014951014891266823, "grad_norm": 0.0478721484541893, "global_step": 22302, "epoch": 210, "lr": 6.231101595864282e-05} +{"train_loss": 0.013219403102993965, "grad_norm": 0.10907457768917084, "global_step": 22303, "epoch": 210, "lr": 6.230814341322413e-05} +{"train_loss": 0.017962628975510597, "grad_norm": 0.11224687099456787, "global_step": 22304, "epoch": 210, "lr": 6.230527082455998e-05} +{"train_loss": 0.017882488667964935, "grad_norm": 0.09517774730920792, "global_step": 22305, "epoch": 210, "lr": 6.230239819266046e-05} +{"train_loss": 0.014419076964259148, "grad_norm": 0.0669223889708519, "global_step": 22306, "epoch": 210, "lr": 6.229952551753568e-05} +{"train_loss": 0.018144715577363968, "grad_norm": 0.08172161132097244, "global_step": 22307, "epoch": 210, "lr": 6.22966527991957e-05} +{"train_loss": 0.01600990816950798, "grad_norm": 0.08706723153591156, "global_step": 22308, "epoch": 210, "lr": 6.229378003765062e-05} +{"train_loss": 0.013005747459828854, "grad_norm": 0.08379200845956802, "global_step": 22309, "epoch": 210, "lr": 6.229090723291056e-05} +{"train_loss": 0.015458381734788418, "grad_norm": 0.09760672599077225, "global_step": 22310, "epoch": 210, "lr": 6.228803438498559e-05} +{"train_loss": 0.014043285511434078, "grad_norm": 0.05412629619240761, "global_step": 22311, "epoch": 210, "lr": 6.228516149388581e-05} +{"train_loss": 0.015051282942295074, "grad_norm": 0.11433810740709305, "global_step": 22312, "epoch": 210, "lr": 6.228228855962133e-05} +{"train_loss": 0.01684195175766945, "grad_norm": 0.07474393397569656, "global_step": 22313, "epoch": 210, "lr": 6.22794155822022e-05} +{"train_loss": 0.0169539712369442, "grad_norm": 0.12517677247524261, "global_step": 22314, "epoch": 210, "lr": 6.227654256163857e-05} +{"train_loss": 0.01761382445693016, "grad_norm": 0.115018330514431, "global_step": 22315, "epoch": 210, "lr": 6.22736694979405e-05} +{"train_loss": 0.01587359793484211, "grad_norm": 0.0601213239133358, "global_step": 22316, "epoch": 210, "lr": 6.227079639111809e-05} +{"train_loss": 0.013984791934490204, "grad_norm": 0.040767353028059006, "global_step": 22317, "epoch": 210, "lr": 6.226792324118144e-05} +{"train_loss": 0.01681092381477356, "grad_norm": 0.08491755276918411, "global_step": 22318, "epoch": 210, "lr": 6.226505004814065e-05} +{"train_loss": 0.015778828412294388, "grad_norm": 0.05791290104389191, "global_step": 22319, "epoch": 210, "lr": 6.22621768120058e-05} +{"train_loss": 0.014104954898357391, "grad_norm": 0.09345867484807968, "global_step": 22320, "epoch": 210, "lr": 6.2259303532787e-05} +{"train_loss": 0.015576489269733429, "grad_norm": 0.07000155746936798, "global_step": 22321, "epoch": 210, "lr": 6.225643021049433e-05} +{"train_loss": 0.017384003847837448, "grad_norm": 0.046383630484342575, "global_step": 22322, "epoch": 210, "lr": 6.22535568451379e-05} +{"train_loss": 0.017253905534744263, "grad_norm": 0.08140142261981964, "global_step": 22323, "epoch": 210, "lr": 6.225068343672779e-05} +{"train_loss": 0.014071190729737282, "grad_norm": 0.06551752984523773, "global_step": 22324, "epoch": 210, "lr": 6.224780998527413e-05} +{"train_loss": 0.013650793582201004, "grad_norm": 0.08957107365131378, "global_step": 22325, "epoch": 210, "lr": 6.224493649078696e-05} +{"train_loss": 0.014079926535487175, "grad_norm": 0.07486355304718018, "global_step": 22326, "epoch": 210, "lr": 6.224206295327643e-05} +{"train_loss": 0.013458004221320152, "grad_norm": 0.05491610988974571, "global_step": 22327, "epoch": 210, "lr": 6.223918937275261e-05} +{"train_loss": 0.018747029826045036, "grad_norm": 0.06008021906018257, "global_step": 22328, "epoch": 210, "lr": 6.22363157492256e-05} +{"train_loss": 0.014819720759987831, "grad_norm": 0.07341880351305008, "global_step": 22329, "epoch": 210, "lr": 6.223344208270551e-05} +{"train_loss": 0.016579341143369675, "grad_norm": 0.09956467151641846, "global_step": 22330, "epoch": 210, "lr": 6.22305683732024e-05} +{"train_loss": 0.015750352293252945, "grad_norm": 0.09937072545289993, "global_step": 22331, "epoch": 210, "lr": 6.22276946207264e-05} +{"train_loss": 0.015811201184988022, "grad_norm": 0.08428239822387695, "global_step": 22332, "epoch": 210, "lr": 6.222482082528761e-05} +{"train_loss": 0.016644801944494247, "grad_norm": 0.08267975598573685, "global_step": 22333, "epoch": 210, "lr": 6.222194698689611e-05} +{"train_loss": 0.01694858819246292, "grad_norm": 0.05403175204992294, "global_step": 22334, "epoch": 210, "lr": 6.2219073105562e-05} +{"train_loss": 0.017830871045589447, "grad_norm": 0.06902915984392166, "global_step": 22335, "epoch": 210, "lr": 6.22161991812954e-05} +{"train_loss": 0.014121487736701965, "grad_norm": 0.060974203050136566, "global_step": 22336, "epoch": 210, "lr": 6.221332521410637e-05} +{"train_loss": 0.01647503674030304, "grad_norm": 0.06451989710330963, "global_step": 22337, "epoch": 210, "lr": 6.221045120400503e-05} +{"train_loss": 0.01842091791331768, "grad_norm": 0.1440388709306717, "global_step": 22338, "epoch": 210, "lr": 6.220757715100148e-05} +{"train_loss": 0.016581544652581215, "grad_norm": 0.0636003389954567, "global_step": 22339, "epoch": 210, "lr": 6.220470305510581e-05} +{"train_loss": 0.014068545773625374, "grad_norm": 0.11263008415699005, "global_step": 22340, "epoch": 210, "lr": 6.220182891632811e-05} +{"train_loss": 0.01576165109872818, "grad_norm": 0.07423367351293564, "global_step": 22341, "epoch": 210, "lr": 6.21989547346785e-05} +{"train_loss": 0.015330296009778976, "grad_norm": 0.05823412910103798, "global_step": 22342, "epoch": 210, "lr": 6.219608051016707e-05} +{"train_loss": 0.017452411353588104, "grad_norm": 0.09811824560165405, "global_step": 22343, "epoch": 210, "lr": 6.21932062428039e-05} +{"train_loss": 0.016134843230247498, "grad_norm": 0.05439307540655136, "global_step": 22344, "epoch": 210, "lr": 6.219033193259912e-05} +{"train_loss": 0.015561701729893684, "grad_norm": 0.06907713413238525, "global_step": 22345, "epoch": 210, "lr": 6.218745757956283e-05} +{"train_loss": 0.017940709367394447, "grad_norm": 0.0687878355383873, "global_step": 22346, "epoch": 210, "lr": 6.218458318370509e-05} +{"train_loss": 0.01523534394800663, "grad_norm": 0.09048033505678177, "global_step": 22347, "epoch": 210, "lr": 6.218170874503604e-05} +{"train_loss": 0.014994053170084953, "grad_norm": 0.06607743352651596, "global_step": 22348, "epoch": 210, "lr": 6.217883426356574e-05} +{"train_loss": 0.017310524359345436, "grad_norm": 0.08301644027233124, "global_step": 22349, "epoch": 210, "lr": 6.217595973930434e-05} +{"train_loss": 0.016721010208129883, "grad_norm": 0.0903281569480896, "global_step": 22350, "epoch": 210, "lr": 6.21730851722619e-05} +{"train_loss": 0.015070488676428795, "grad_norm": 0.06591979414224625, "global_step": 22351, "epoch": 210, "lr": 6.217021056244854e-05} +{"train_loss": 0.015113599598407745, "grad_norm": 0.06657161563634872, "global_step": 22352, "epoch": 210, "lr": 6.216733590987434e-05} +{"train_loss": 0.014696530066430569, "grad_norm": 0.0922798290848732, "global_step": 22353, "epoch": 210, "lr": 6.216446121454942e-05} +{"train_loss": 0.015792246907949448, "grad_norm": 0.11740175634622574, "global_step": 22354, "epoch": 210, "lr": 6.216158647648386e-05} +{"train_loss": 0.01669873110949993, "grad_norm": 0.05399152263998985, "global_step": 22355, "epoch": 210, "lr": 6.215871169568779e-05} +{"train_loss": 0.016944576054811478, "grad_norm": 0.058414291590452194, "global_step": 22356, "epoch": 210, "lr": 6.21558368721713e-05} +{"train_loss": 0.016658566892147064, "grad_norm": 0.07176047563552856, "global_step": 22357, "epoch": 210, "lr": 6.215296200594447e-05} +{"train_loss": 0.017559941858053207, "grad_norm": 0.07228312641382217, "global_step": 22358, "epoch": 210, "lr": 6.215008709701743e-05} +{"train_loss": 0.015598271042108536, "grad_norm": 0.08481790870428085, "global_step": 22359, "epoch": 210, "lr": 6.214721214540027e-05} +{"train_loss": 0.01520019955933094, "grad_norm": 0.09250380843877792, "global_step": 22360, "epoch": 210, "lr": 6.214433715110308e-05} +{"train_loss": 0.01928050071001053, "grad_norm": 0.0639931708574295, "global_step": 22361, "epoch": 210, "lr": 6.214146211413597e-05} +{"train_loss": 0.01830350235104561, "grad_norm": 0.0631767064332962, "global_step": 22362, "epoch": 210, "lr": 6.213858703450905e-05} +{"train_loss": 0.016643723472952843, "grad_norm": 0.1097634881734848, "global_step": 22363, "epoch": 210, "lr": 6.21357119122324e-05} +{"train_loss": 0.017679795622825623, "grad_norm": 0.05959505960345268, "global_step": 22364, "epoch": 210, "lr": 6.213283674731614e-05} +{"train_loss": 0.015871278010308743, "grad_norm": 0.06702709943056107, "global_step": 22365, "epoch": 210, "lr": 6.212996153977037e-05, "train_batch_mse_error": 0.014092660509049892, "val_loss": 0.01900581829249859} +{"train_loss": 0.014111088588833809, "grad_norm": 0.09200765192508698, "global_step": 22366, "epoch": 211, "lr": 6.212708628960519e-05} +{"train_loss": 0.016926653683185577, "grad_norm": 0.15072783827781677, "global_step": 22367, "epoch": 211, "lr": 6.212421099683072e-05} +{"train_loss": 0.01825612038373947, "grad_norm": 0.06005450710654259, "global_step": 22368, "epoch": 211, "lr": 6.212133566145704e-05} +{"train_loss": 0.01323095615953207, "grad_norm": 0.050786830484867096, "global_step": 22369, "epoch": 211, "lr": 6.211846028349423e-05} +{"train_loss": 0.013562459498643875, "grad_norm": 0.07758779078722, "global_step": 22370, "epoch": 211, "lr": 6.211558486295243e-05} +{"train_loss": 0.01620321348309517, "grad_norm": 0.059228237718343735, "global_step": 22371, "epoch": 211, "lr": 6.211270939984176e-05} +{"train_loss": 0.01202914584428072, "grad_norm": 0.06227526813745499, "global_step": 22372, "epoch": 211, "lr": 6.210983389417227e-05} +{"train_loss": 0.01516399160027504, "grad_norm": 0.04731680452823639, "global_step": 22373, "epoch": 211, "lr": 6.21069583459541e-05} +{"train_loss": 0.013779470697045326, "grad_norm": 0.06022581085562706, "global_step": 22374, "epoch": 211, "lr": 6.210408275519735e-05} +{"train_loss": 0.015426097437739372, "grad_norm": 0.08283282071352005, "global_step": 22375, "epoch": 211, "lr": 6.210120712191212e-05} +{"train_loss": 0.017149467021226883, "grad_norm": 0.05983554199337959, "global_step": 22376, "epoch": 211, "lr": 6.209833144610849e-05} +{"train_loss": 0.015892215073108673, "grad_norm": 0.07376354932785034, "global_step": 22377, "epoch": 211, "lr": 6.20954557277966e-05} +{"train_loss": 0.016277749091386795, "grad_norm": 0.06789767742156982, "global_step": 22378, "epoch": 211, "lr": 6.209257996698655e-05} +{"train_loss": 0.017358114942908287, "grad_norm": 0.06509197503328323, "global_step": 22379, "epoch": 211, "lr": 6.208970416368841e-05} +{"train_loss": 0.01581917144358158, "grad_norm": 0.09967581182718277, "global_step": 22380, "epoch": 211, "lr": 6.208682831791232e-05} +{"train_loss": 0.01742991805076599, "grad_norm": 0.14760300517082214, "global_step": 22381, "epoch": 211, "lr": 6.208395242966834e-05} +{"train_loss": 0.013761833310127258, "grad_norm": 0.06410973519086838, "global_step": 22382, "epoch": 211, "lr": 6.208107649896666e-05} +{"train_loss": 0.016645634546875954, "grad_norm": 0.0767689198255539, "global_step": 22383, "epoch": 211, "lr": 6.207820052581729e-05} +{"train_loss": 0.01847803220152855, "grad_norm": 0.06045506149530411, "global_step": 22384, "epoch": 211, "lr": 6.207532451023039e-05} +{"train_loss": 0.017331402748823166, "grad_norm": 0.07935234904289246, "global_step": 22385, "epoch": 211, "lr": 6.207244845221607e-05} +{"train_loss": 0.014780228957533836, "grad_norm": 0.0672781765460968, "global_step": 22386, "epoch": 211, "lr": 6.20695723517844e-05} +{"train_loss": 0.015566862188279629, "grad_norm": 0.08491644263267517, "global_step": 22387, "epoch": 211, "lr": 6.206669620894549e-05} +{"train_loss": 0.014303820207715034, "grad_norm": 0.08753463625907898, "global_step": 22388, "epoch": 211, "lr": 6.206382002370946e-05} +{"train_loss": 0.016306785866618156, "grad_norm": 0.06611645221710205, "global_step": 22389, "epoch": 211, "lr": 6.206094379608642e-05} +{"train_loss": 0.01674473285675049, "grad_norm": 0.06781173497438431, "global_step": 22390, "epoch": 211, "lr": 6.205806752608647e-05} +{"train_loss": 0.017072025686502457, "grad_norm": 0.09460485726594925, "global_step": 22391, "epoch": 211, "lr": 6.20551912137197e-05} +{"train_loss": 0.01628117449581623, "grad_norm": 0.055113423615694046, "global_step": 22392, "epoch": 211, "lr": 6.205231485899625e-05} +{"train_loss": 0.016638807952404022, "grad_norm": 0.07214689254760742, "global_step": 22393, "epoch": 211, "lr": 6.20494384619262e-05} +{"train_loss": 0.016753165051341057, "grad_norm": 0.06497582048177719, "global_step": 22394, "epoch": 211, "lr": 6.204656202251968e-05} +{"train_loss": 0.015109913423657417, "grad_norm": 0.08004211634397507, "global_step": 22395, "epoch": 211, "lr": 6.204368554078675e-05} +{"train_loss": 0.0157794076949358, "grad_norm": 0.06973731517791748, "global_step": 22396, "epoch": 211, "lr": 6.204080901673755e-05} +{"train_loss": 0.016867496073246002, "grad_norm": 0.10560378432273865, "global_step": 22397, "epoch": 211, "lr": 6.203793245038222e-05} +{"train_loss": 0.013827495276927948, "grad_norm": 0.0699675977230072, "global_step": 22398, "epoch": 211, "lr": 6.203505584173079e-05} +{"train_loss": 0.014443780295550823, "grad_norm": 0.06393861770629883, "global_step": 22399, "epoch": 211, "lr": 6.203217919079342e-05} +{"train_loss": 0.015456008724868298, "grad_norm": 0.06800699979066849, "global_step": 22400, "epoch": 211, "lr": 6.20293024975802e-05} +{"train_loss": 0.0176625344902277, "grad_norm": 0.09344136714935303, "global_step": 22401, "epoch": 211, "lr": 6.202642576210125e-05} +{"train_loss": 0.01629277877509594, "grad_norm": 0.1097768247127533, "global_step": 22402, "epoch": 211, "lr": 6.202354898436669e-05} +{"train_loss": 0.013651303015649319, "grad_norm": 0.051595915108919144, "global_step": 22403, "epoch": 211, "lr": 6.202067216438659e-05} +{"train_loss": 0.016794541850686073, "grad_norm": 0.061868760734796524, "global_step": 22404, "epoch": 211, "lr": 6.201779530217107e-05} +{"train_loss": 0.01428306009620428, "grad_norm": 0.09544157981872559, "global_step": 22405, "epoch": 211, "lr": 6.201491839773024e-05} +{"train_loss": 0.01491004228591919, "grad_norm": 0.0738249197602272, "global_step": 22406, "epoch": 211, "lr": 6.201204145107422e-05} +{"train_loss": 0.017519226297736168, "grad_norm": 0.08585767447948456, "global_step": 22407, "epoch": 211, "lr": 6.20091644622131e-05} +{"train_loss": 0.014576442539691925, "grad_norm": 0.07353504747152328, "global_step": 22408, "epoch": 211, "lr": 6.2006287431157e-05} +{"train_loss": 0.01691630110144615, "grad_norm": 0.07359138131141663, "global_step": 22409, "epoch": 211, "lr": 6.200341035791606e-05} +{"train_loss": 0.014081768691539764, "grad_norm": 0.057029202580451965, "global_step": 22410, "epoch": 211, "lr": 6.200053324250032e-05} +{"train_loss": 0.01786000467836857, "grad_norm": 0.0528358519077301, "global_step": 22411, "epoch": 211, "lr": 6.199765608491994e-05} +{"train_loss": 0.01476207934319973, "grad_norm": 0.061520930379629135, "global_step": 22412, "epoch": 211, "lr": 6.199477888518502e-05} +{"train_loss": 0.016769450157880783, "grad_norm": 0.06892799586057663, "global_step": 22413, "epoch": 211, "lr": 6.199190164330564e-05} +{"train_loss": 0.014668470248579979, "grad_norm": 0.057898085564374924, "global_step": 22414, "epoch": 211, "lr": 6.198902435929195e-05} +{"train_loss": 0.013610780239105225, "grad_norm": 0.06655235588550568, "global_step": 22415, "epoch": 211, "lr": 6.198614703315404e-05} +{"train_loss": 0.01875532791018486, "grad_norm": 0.1426781266927719, "global_step": 22416, "epoch": 211, "lr": 6.198326966490203e-05} +{"train_loss": 0.01654662936925888, "grad_norm": 0.07111211866140366, "global_step": 22417, "epoch": 211, "lr": 6.1980392254546e-05} +{"train_loss": 0.016614265739917755, "grad_norm": 0.10771135985851288, "global_step": 22418, "epoch": 211, "lr": 6.19775148020961e-05} +{"train_loss": 0.01567765511572361, "grad_norm": 0.11031649261713028, "global_step": 22419, "epoch": 211, "lr": 6.197463730756242e-05} +{"train_loss": 0.015624395571649075, "grad_norm": 0.07361184805631638, "global_step": 22420, "epoch": 211, "lr": 6.197175977095505e-05} +{"train_loss": 0.01542059425264597, "grad_norm": 0.11954380571842194, "global_step": 22421, "epoch": 211, "lr": 6.196888219228416e-05} +{"train_loss": 0.016383351758122444, "grad_norm": 0.08330542594194412, "global_step": 22422, "epoch": 211, "lr": 6.196600457155979e-05} +{"train_loss": 0.015615217387676239, "grad_norm": 0.06244652718305588, "global_step": 22423, "epoch": 211, "lr": 6.19631269087921e-05} +{"train_loss": 0.013239750638604164, "grad_norm": 0.05874069035053253, "global_step": 22424, "epoch": 211, "lr": 6.196024920399118e-05} +{"train_loss": 0.017270978540182114, "grad_norm": 0.1022082045674324, "global_step": 22425, "epoch": 211, "lr": 6.195737145716715e-05} +{"train_loss": 0.015431907959282398, "grad_norm": 0.05938044562935829, "global_step": 22426, "epoch": 211, "lr": 6.195449366833012e-05} +{"train_loss": 0.016151996329426765, "grad_norm": 0.06942524015903473, "global_step": 22427, "epoch": 211, "lr": 6.195161583749019e-05} +{"train_loss": 0.015383058227598667, "grad_norm": 0.0847998857498169, "global_step": 22428, "epoch": 211, "lr": 6.194873796465749e-05} +{"train_loss": 0.01635671965777874, "grad_norm": 0.10226823389530182, "global_step": 22429, "epoch": 211, "lr": 6.194586004984211e-05} +{"train_loss": 0.015102864243090153, "grad_norm": 0.08870228379964828, "global_step": 22430, "epoch": 211, "lr": 6.194298209305418e-05} +{"train_loss": 0.01408647932112217, "grad_norm": 0.12850235402584076, "global_step": 22431, "epoch": 211, "lr": 6.19401040943038e-05} +{"train_loss": 0.014133048243820667, "grad_norm": 0.06165674328804016, "global_step": 22432, "epoch": 211, "lr": 6.193722605360109e-05} +{"train_loss": 0.016358407214283943, "grad_norm": 0.12286238372325897, "global_step": 22433, "epoch": 211, "lr": 6.193434797095616e-05} +{"train_loss": 0.018461767584085464, "grad_norm": 0.05366251617670059, "global_step": 22434, "epoch": 211, "lr": 6.19314698463791e-05} +{"train_loss": 0.015768809244036674, "grad_norm": 0.13825945556163788, "global_step": 22435, "epoch": 211, "lr": 6.192859167988009e-05} +{"train_loss": 0.013202685862779617, "grad_norm": 0.08285067975521088, "global_step": 22436, "epoch": 211, "lr": 6.192571347146916e-05} +{"train_loss": 0.017208755016326904, "grad_norm": 0.07697033882141113, "global_step": 22437, "epoch": 211, "lr": 6.192283522115646e-05} +{"train_loss": 0.01662397012114525, "grad_norm": 0.11718926578760147, "global_step": 22438, "epoch": 211, "lr": 6.191995692895212e-05} +{"train_loss": 0.01747504249215126, "grad_norm": 0.06697969883680344, "global_step": 22439, "epoch": 211, "lr": 6.191707859486622e-05} +{"train_loss": 0.015456320717930794, "grad_norm": 0.06990907341241837, "global_step": 22440, "epoch": 211, "lr": 6.19142002189089e-05} +{"train_loss": 0.012861561961472034, "grad_norm": 0.04906648024916649, "global_step": 22441, "epoch": 211, "lr": 6.191132180109025e-05} +{"train_loss": 0.015932597219944, "grad_norm": 0.1502632051706314, "global_step": 22442, "epoch": 211, "lr": 6.190844334142041e-05} +{"train_loss": 0.015779156237840652, "grad_norm": 0.1361982524394989, "global_step": 22443, "epoch": 211, "lr": 6.190556483990948e-05} +{"train_loss": 0.016127724200487137, "grad_norm": 0.07648969441652298, "global_step": 22444, "epoch": 211, "lr": 6.190268629656754e-05} +{"train_loss": 0.01745481602847576, "grad_norm": 0.10680285841226578, "global_step": 22445, "epoch": 211, "lr": 6.189980771140477e-05} +{"train_loss": 0.015232356265187263, "grad_norm": 0.07792644202709198, "global_step": 22446, "epoch": 211, "lr": 6.189692908443123e-05} +{"train_loss": 0.017709393054246902, "grad_norm": 0.1159210130572319, "global_step": 22447, "epoch": 211, "lr": 6.189405041565708e-05} +{"train_loss": 0.016056040301918983, "grad_norm": 0.09504406154155731, "global_step": 22448, "epoch": 211, "lr": 6.189117170509239e-05} +{"train_loss": 0.014759562909603119, "grad_norm": 0.10371110588312149, "global_step": 22449, "epoch": 211, "lr": 6.18882929527473e-05} +{"train_loss": 0.0179213248193264, "grad_norm": 0.10296563059091568, "global_step": 22450, "epoch": 211, "lr": 6.188541415863192e-05} +{"train_loss": 0.01757560297846794, "grad_norm": 0.11653335392475128, "global_step": 22451, "epoch": 211, "lr": 6.188253532275635e-05} +{"train_loss": 0.014626001007854939, "grad_norm": 0.05526035651564598, "global_step": 22452, "epoch": 211, "lr": 6.187965644513072e-05} +{"train_loss": 0.015430881641805172, "grad_norm": 0.10152526944875717, "global_step": 22453, "epoch": 211, "lr": 6.187677752576516e-05} +{"train_loss": 0.016598427668213844, "grad_norm": 0.0603756420314312, "global_step": 22454, "epoch": 211, "lr": 6.187389856466977e-05} +{"train_loss": 0.015866719186306, "grad_norm": 0.08618717640638351, "global_step": 22455, "epoch": 211, "lr": 6.187101956185466e-05} +{"train_loss": 0.016877301037311554, "grad_norm": 0.09485508501529694, "global_step": 22456, "epoch": 211, "lr": 6.186814051732995e-05} +{"train_loss": 0.015681510791182518, "grad_norm": 0.04889734461903572, "global_step": 22457, "epoch": 211, "lr": 6.186526143110573e-05} +{"train_loss": 0.013384297490119934, "grad_norm": 0.13388414680957794, "global_step": 22458, "epoch": 211, "lr": 6.186238230319218e-05} +{"train_loss": 0.017018668353557587, "grad_norm": 0.1056327372789383, "global_step": 22459, "epoch": 211, "lr": 6.185950313359936e-05} +{"train_loss": 0.014927083626389503, "grad_norm": 0.1547766774892807, "global_step": 22460, "epoch": 211, "lr": 6.185662392233739e-05} +{"train_loss": 0.016737457364797592, "grad_norm": 0.0919857919216156, "global_step": 22461, "epoch": 211, "lr": 6.185374466941641e-05} +{"train_loss": 0.016232788562774658, "grad_norm": 0.118894062936306, "global_step": 22462, "epoch": 211, "lr": 6.185086537484654e-05} +{"train_loss": 0.017324257642030716, "grad_norm": 0.07548046112060547, "global_step": 22463, "epoch": 211, "lr": 6.184798603863787e-05} +{"train_loss": 0.017710264772176743, "grad_norm": 0.0678795650601387, "global_step": 22464, "epoch": 211, "lr": 6.184510666080053e-05} +{"train_loss": 0.01602175459265709, "grad_norm": 0.0975673496723175, "global_step": 22465, "epoch": 211, "lr": 6.184222724134465e-05} +{"train_loss": 0.012865006923675537, "grad_norm": 0.10563909262418747, "global_step": 22466, "epoch": 211, "lr": 6.183934778028031e-05} +{"train_loss": 0.013990772888064384, "grad_norm": 0.10641949623823166, "global_step": 22467, "epoch": 211, "lr": 6.183646827761766e-05} +{"train_loss": 0.018182283267378807, "grad_norm": 0.14624448120594025, "global_step": 22468, "epoch": 211, "lr": 6.183358873336682e-05} +{"train_loss": 0.015455423854291439, "grad_norm": 0.06799839437007904, "global_step": 22469, "epoch": 211, "lr": 6.183070914753789e-05} +{"train_loss": 0.02075163647532463, "grad_norm": 0.070172518491745, "global_step": 22470, "epoch": 211, "lr": 6.182782952014099e-05} +{"train_loss": 0.01586357000688056, "grad_norm": 0.11154089123010635, "global_step": 22471, "epoch": 211, "lr": 6.182494985118624e-05, "val_loss": 0.01898675039410591} +{"train_loss": 0.014297783374786377, "grad_norm": 0.04697401449084282, "global_step": 22472, "epoch": 212, "lr": 6.182207014068377e-05} +{"train_loss": 0.013674244284629822, "grad_norm": 0.10536494851112366, "global_step": 22473, "epoch": 212, "lr": 6.181919038864365e-05} +{"train_loss": 0.01654670760035515, "grad_norm": 0.11239839345216751, "global_step": 22474, "epoch": 212, "lr": 6.181631059507607e-05} +{"train_loss": 0.018895260989665985, "grad_norm": 0.11218101531267166, "global_step": 22475, "epoch": 212, "lr": 6.181343075999111e-05} +{"train_loss": 0.018738117069005966, "grad_norm": 0.11336833238601685, "global_step": 22476, "epoch": 212, "lr": 6.181055088339888e-05} +{"train_loss": 0.016464699059724808, "grad_norm": 0.10902849584817886, "global_step": 22477, "epoch": 212, "lr": 6.180767096530953e-05} +{"train_loss": 0.017499633133411407, "grad_norm": 0.06940726190805435, "global_step": 22478, "epoch": 212, "lr": 6.180479100573313e-05} +{"train_loss": 0.017748398706316948, "grad_norm": 0.061227839440107346, "global_step": 22479, "epoch": 212, "lr": 6.180191100467986e-05} +{"train_loss": 0.01631947234272957, "grad_norm": 0.11385099589824677, "global_step": 22480, "epoch": 212, "lr": 6.179903096215979e-05} +{"train_loss": 0.014061352238059044, "grad_norm": 0.08705426007509232, "global_step": 22481, "epoch": 212, "lr": 6.179615087818305e-05} +{"train_loss": 0.01737680472433567, "grad_norm": 0.1306925117969513, "global_step": 22482, "epoch": 212, "lr": 6.179327075275978e-05} +{"train_loss": 0.015310123562812805, "grad_norm": 0.06087815389037132, "global_step": 22483, "epoch": 212, "lr": 6.179039058590008e-05} +{"train_loss": 0.01475433073937893, "grad_norm": 0.07582809776067734, "global_step": 22484, "epoch": 212, "lr": 6.178751037761408e-05} +{"train_loss": 0.015027920715510845, "grad_norm": 0.0704062357544899, "global_step": 22485, "epoch": 212, "lr": 6.178463012791188e-05} +{"train_loss": 0.013700615614652634, "grad_norm": 0.16512714326381683, "global_step": 22486, "epoch": 212, "lr": 6.178174983680362e-05} +{"train_loss": 0.017336497083306313, "grad_norm": 0.0707419142127037, "global_step": 22487, "epoch": 212, "lr": 6.177886950429943e-05} +{"train_loss": 0.015419432893395424, "grad_norm": 0.12084060907363892, "global_step": 22488, "epoch": 212, "lr": 6.17759891304094e-05} +{"train_loss": 0.015652846544981003, "grad_norm": 0.0987849161028862, "global_step": 22489, "epoch": 212, "lr": 6.177310871514366e-05} +{"train_loss": 0.016670644283294678, "grad_norm": 0.10489901900291443, "global_step": 22490, "epoch": 212, "lr": 6.177022825851234e-05} +{"train_loss": 0.01615912653505802, "grad_norm": 0.1331661343574524, "global_step": 22491, "epoch": 212, "lr": 6.176734776052557e-05} +{"train_loss": 0.01544075459241867, "grad_norm": 0.1020403578877449, "global_step": 22492, "epoch": 212, "lr": 6.176446722119344e-05} +{"train_loss": 0.016448449343442917, "grad_norm": 0.08839414268732071, "global_step": 22493, "epoch": 212, "lr": 6.176158664052611e-05} +{"train_loss": 0.015982026234269142, "grad_norm": 0.06909456849098206, "global_step": 22494, "epoch": 212, "lr": 6.175870601853368e-05} +{"train_loss": 0.016313042491674423, "grad_norm": 0.07273279130458832, "global_step": 22495, "epoch": 212, "lr": 6.175582535522624e-05} +{"train_loss": 0.01581837609410286, "grad_norm": 0.07153692841529846, "global_step": 22496, "epoch": 212, "lr": 6.175294465061396e-05} +{"train_loss": 0.01438475027680397, "grad_norm": 0.07349298894405365, "global_step": 22497, "epoch": 212, "lr": 6.175006390470695e-05} +{"train_loss": 0.01414121501147747, "grad_norm": 0.06099668890237808, "global_step": 22498, "epoch": 212, "lr": 6.174718311751532e-05} +{"train_loss": 0.014750516042113304, "grad_norm": 0.1139870211482048, "global_step": 22499, "epoch": 212, "lr": 6.174430228904919e-05} +{"train_loss": 0.01558553148061037, "grad_norm": 0.14657527208328247, "global_step": 22500, "epoch": 212, "lr": 6.17414214193187e-05} +{"train_loss": 0.017073268070816994, "grad_norm": 0.15243172645568848, "global_step": 22501, "epoch": 212, "lr": 6.173854050833396e-05} +{"train_loss": 0.014326752163469791, "grad_norm": 0.05650436505675316, "global_step": 22502, "epoch": 212, "lr": 6.173565955610508e-05} +{"train_loss": 0.014848601073026657, "grad_norm": 0.06797781586647034, "global_step": 22503, "epoch": 212, "lr": 6.173277856264221e-05} +{"train_loss": 0.01522558182477951, "grad_norm": 0.0696222111582756, "global_step": 22504, "epoch": 212, "lr": 6.172989752795544e-05} +{"train_loss": 0.01429501362144947, "grad_norm": 0.07284040004014969, "global_step": 22505, "epoch": 212, "lr": 6.172701645205493e-05} +{"train_loss": 0.015229864045977592, "grad_norm": 0.10651414841413498, "global_step": 22506, "epoch": 212, "lr": 6.17241353349508e-05} +{"train_loss": 0.018116671591997147, "grad_norm": 0.07783056050539017, "global_step": 22507, "epoch": 212, "lr": 6.172125417665311e-05} +{"train_loss": 0.01665164716541767, "grad_norm": 0.051450006663799286, "global_step": 22508, "epoch": 212, "lr": 6.171837297717206e-05} +{"train_loss": 0.017109226435422897, "grad_norm": 0.061478324234485626, "global_step": 22509, "epoch": 212, "lr": 6.171549173651776e-05} +{"train_loss": 0.016688846051692963, "grad_norm": 0.07450953125953674, "global_step": 22510, "epoch": 212, "lr": 6.171261045470029e-05} +{"train_loss": 0.014597134664654732, "grad_norm": 0.060504987835884094, "global_step": 22511, "epoch": 212, "lr": 6.170972913172981e-05} +{"train_loss": 0.018303964287042618, "grad_norm": 0.07376646995544434, "global_step": 22512, "epoch": 212, "lr": 6.170684776761644e-05} +{"train_loss": 0.013358679600059986, "grad_norm": 0.07645292580127716, "global_step": 22513, "epoch": 212, "lr": 6.17039663623703e-05} +{"train_loss": 0.015239924192428589, "grad_norm": 0.11935076117515564, "global_step": 22514, "epoch": 212, "lr": 6.170108491600149e-05} +{"train_loss": 0.014494944363832474, "grad_norm": 0.05722307413816452, "global_step": 22515, "epoch": 212, "lr": 6.169820342852017e-05} +{"train_loss": 0.018205909058451653, "grad_norm": 0.1455964297056198, "global_step": 22516, "epoch": 212, "lr": 6.169532189993646e-05} +{"train_loss": 0.01499009970575571, "grad_norm": 0.14223237335681915, "global_step": 22517, "epoch": 212, "lr": 6.169244033026045e-05} +{"train_loss": 0.01598227582871914, "grad_norm": 0.1160767674446106, "global_step": 22518, "epoch": 212, "lr": 6.168955871950232e-05} +{"train_loss": 0.016483768820762634, "grad_norm": 0.05285852774977684, "global_step": 22519, "epoch": 212, "lr": 6.168667706767214e-05} +{"train_loss": 0.014618834480643272, "grad_norm": 0.08565697818994522, "global_step": 22520, "epoch": 212, "lr": 6.168379537478006e-05} +{"train_loss": 0.013401830568909645, "grad_norm": 0.056278251111507416, "global_step": 22521, "epoch": 212, "lr": 6.168091364083621e-05} +{"train_loss": 0.014318956062197685, "grad_norm": 0.05838475748896599, "global_step": 22522, "epoch": 212, "lr": 6.167803186585073e-05} +{"train_loss": 0.01569931022822857, "grad_norm": 0.08369634300470352, "global_step": 22523, "epoch": 212, "lr": 6.167515004983369e-05} +{"train_loss": 0.01548743061721325, "grad_norm": 0.08437631279230118, "global_step": 22524, "epoch": 212, "lr": 6.167226819279528e-05} +{"train_loss": 0.01697644032537937, "grad_norm": 0.09101230651140213, "global_step": 22525, "epoch": 212, "lr": 6.166938629474558e-05} +{"train_loss": 0.01804378256201744, "grad_norm": 0.08955918997526169, "global_step": 22526, "epoch": 212, "lr": 6.166650435569472e-05} +{"train_loss": 0.015200422145426273, "grad_norm": 0.04977742210030556, "global_step": 22527, "epoch": 212, "lr": 6.166362237565287e-05} +{"train_loss": 0.014885948039591312, "grad_norm": 0.09334983676671982, "global_step": 22528, "epoch": 212, "lr": 6.16607403546301e-05} +{"train_loss": 0.016252301633358, "grad_norm": 0.20450137555599213, "global_step": 22529, "epoch": 212, "lr": 6.165785829263654e-05} +{"train_loss": 0.01667005941271782, "grad_norm": 0.12378035485744476, "global_step": 22530, "epoch": 212, "lr": 6.16549761896824e-05} +{"train_loss": 0.014780926518142223, "grad_norm": 0.09162460267543793, "global_step": 22531, "epoch": 212, "lr": 6.165209404577767e-05} +{"train_loss": 0.015000151470303535, "grad_norm": 0.06455585360527039, "global_step": 22532, "epoch": 212, "lr": 6.16492118609326e-05} +{"train_loss": 0.01620493456721306, "grad_norm": 0.06867750734090805, "global_step": 22533, "epoch": 212, "lr": 6.164632963515724e-05} +{"train_loss": 0.017358053475618362, "grad_norm": 0.07367256283760071, "global_step": 22534, "epoch": 212, "lr": 6.164344736846175e-05} +{"train_loss": 0.016992410644888878, "grad_norm": 0.08121750503778458, "global_step": 22535, "epoch": 212, "lr": 6.164056506085626e-05} +{"train_loss": 0.01718531921505928, "grad_norm": 0.06659160554409027, "global_step": 22536, "epoch": 212, "lr": 6.163768271235086e-05} +{"train_loss": 0.015420741401612759, "grad_norm": 0.08589769899845123, "global_step": 22537, "epoch": 212, "lr": 6.163480032295572e-05} +{"train_loss": 0.01643887162208557, "grad_norm": 0.0956200659275055, "global_step": 22538, "epoch": 212, "lr": 6.163191789268095e-05} +{"train_loss": 0.017804071307182312, "grad_norm": 0.10502830147743225, "global_step": 22539, "epoch": 212, "lr": 6.162903542153669e-05} +{"train_loss": 0.016405262053012848, "grad_norm": 0.09420424699783325, "global_step": 22540, "epoch": 212, "lr": 6.162615290953304e-05} +{"train_loss": 0.018462788313627243, "grad_norm": 0.07821367681026459, "global_step": 22541, "epoch": 212, "lr": 6.162327035668015e-05} +{"train_loss": 0.019152965396642685, "grad_norm": 0.04885091632604599, "global_step": 22542, "epoch": 212, "lr": 6.162038776298815e-05} +{"train_loss": 0.0193626768887043, "grad_norm": 0.055467646569013596, "global_step": 22543, "epoch": 212, "lr": 6.161750512846715e-05} +{"train_loss": 0.015569611452519894, "grad_norm": 0.05759948492050171, "global_step": 22544, "epoch": 212, "lr": 6.16146224531273e-05} +{"train_loss": 0.014672151766717434, "grad_norm": 0.06355119496583939, "global_step": 22545, "epoch": 212, "lr": 6.161173973697871e-05} +{"train_loss": 0.014378965832293034, "grad_norm": 0.0765034630894661, "global_step": 22546, "epoch": 212, "lr": 6.160885698003152e-05} +{"train_loss": 0.013549430295825005, "grad_norm": 0.10776685923337936, "global_step": 22547, "epoch": 212, "lr": 6.160597418229586e-05} +{"train_loss": 0.01447560079395771, "grad_norm": 0.054384004324674606, "global_step": 22548, "epoch": 212, "lr": 6.160309134378184e-05} +{"train_loss": 0.01624755561351776, "grad_norm": 0.12147213518619537, "global_step": 22549, "epoch": 212, "lr": 6.16002084644996e-05} +{"train_loss": 0.017811503261327744, "grad_norm": 0.08501540869474411, "global_step": 22550, "epoch": 212, "lr": 6.15973255444593e-05} +{"train_loss": 0.01519777625799179, "grad_norm": 0.07268347591161728, "global_step": 22551, "epoch": 212, "lr": 6.159444258367101e-05} +{"train_loss": 0.01345033198595047, "grad_norm": 0.12378540635108948, "global_step": 22552, "epoch": 212, "lr": 6.159155958214492e-05} +{"train_loss": 0.015231097117066383, "grad_norm": 0.09283207356929779, "global_step": 22553, "epoch": 212, "lr": 6.15886765398911e-05} +{"train_loss": 0.015073629096150398, "grad_norm": 0.06036701425909996, "global_step": 22554, "epoch": 212, "lr": 6.158579345691974e-05} +{"train_loss": 0.01763695292174816, "grad_norm": 0.08873064070940018, "global_step": 22555, "epoch": 212, "lr": 6.158291033324091e-05} +{"train_loss": 0.014884434640407562, "grad_norm": 0.08090047538280487, "global_step": 22556, "epoch": 212, "lr": 6.15800271688648e-05} +{"train_loss": 0.0161735936999321, "grad_norm": 0.12269841879606247, "global_step": 22557, "epoch": 212, "lr": 6.157714396380152e-05} +{"train_loss": 0.01877775602042675, "grad_norm": 0.07252109050750732, "global_step": 22558, "epoch": 212, "lr": 6.157426071806114e-05} +{"train_loss": 0.015364417806267738, "grad_norm": 0.0842629000544548, "global_step": 22559, "epoch": 212, "lr": 6.15713774316539e-05} +{"train_loss": 0.014914598315954208, "grad_norm": 0.06593278795480728, "global_step": 22560, "epoch": 212, "lr": 6.156849410458982e-05} +{"train_loss": 0.016473006457090378, "grad_norm": 0.10325901210308075, "global_step": 22561, "epoch": 212, "lr": 6.15656107368791e-05} +{"train_loss": 0.015016507357358932, "grad_norm": 0.11097811162471771, "global_step": 22562, "epoch": 212, "lr": 6.156272732853187e-05} +{"train_loss": 0.016121506690979004, "grad_norm": 0.05470168590545654, "global_step": 22563, "epoch": 212, "lr": 6.155984387955823e-05} +{"train_loss": 0.017161410301923752, "grad_norm": 0.07082784920930862, "global_step": 22564, "epoch": 212, "lr": 6.155696038996834e-05} +{"train_loss": 0.015087476931512356, "grad_norm": 0.06457824259996414, "global_step": 22565, "epoch": 212, "lr": 6.15540768597723e-05} +{"train_loss": 0.014824881218373775, "grad_norm": 0.06444527953863144, "global_step": 22566, "epoch": 212, "lr": 6.155119328898027e-05} +{"train_loss": 0.015352470800280571, "grad_norm": 0.09241227060556412, "global_step": 22567, "epoch": 212, "lr": 6.154830967760236e-05} +{"train_loss": 0.014887060970067978, "grad_norm": 0.11465774476528168, "global_step": 22568, "epoch": 212, "lr": 6.154542602564873e-05} +{"train_loss": 0.014498701319098473, "grad_norm": 0.0510164238512516, "global_step": 22569, "epoch": 212, "lr": 6.154254233312947e-05} +{"train_loss": 0.015672123059630394, "grad_norm": 0.13267755508422852, "global_step": 22570, "epoch": 212, "lr": 6.153965860005474e-05} +{"train_loss": 0.016174159944057465, "grad_norm": 0.10405973345041275, "global_step": 22571, "epoch": 212, "lr": 6.153677482643469e-05} +{"train_loss": 0.017214734107255936, "grad_norm": 0.06371335685253143, "global_step": 22572, "epoch": 212, "lr": 6.15338910122794e-05} +{"train_loss": 0.01625913754105568, "grad_norm": 0.13710927963256836, "global_step": 22573, "epoch": 212, "lr": 6.153100715759905e-05} +{"train_loss": 0.015865463763475418, "grad_norm": 0.08271978795528412, "global_step": 22574, "epoch": 212, "lr": 6.152812326240375e-05} +{"train_loss": 0.013745717704296112, "grad_norm": 0.07123526930809021, "global_step": 22575, "epoch": 212, "lr": 6.152523932670366e-05} +{"train_loss": 0.01408373098820448, "grad_norm": 0.07939404994249344, "global_step": 22576, "epoch": 212, "lr": 6.152235535050886e-05} +{"train_loss": 0.01586631544039778, "grad_norm": 0.10138387233018875, "global_step": 22577, "epoch": 212, "lr": 6.151947133382954e-05, "val_loss": 0.019022272899746895} +{"train_loss": 0.013390030711889267, "grad_norm": 0.08836940675973892, "global_step": 22578, "epoch": 213, "lr": 6.151658727667579e-05} +{"train_loss": 0.016237959265708923, "grad_norm": 0.11424694955348969, "global_step": 22579, "epoch": 213, "lr": 6.151370317905777e-05} +{"train_loss": 0.01365329697728157, "grad_norm": 0.058787859976291656, "global_step": 22580, "epoch": 213, "lr": 6.15108190409856e-05} +{"train_loss": 0.017234232276678085, "grad_norm": 0.05965558439493179, "global_step": 22581, "epoch": 213, "lr": 6.150793486246941e-05} +{"train_loss": 0.015289516188204288, "grad_norm": 0.059477243572473526, "global_step": 22582, "epoch": 213, "lr": 6.150505064351935e-05} +{"train_loss": 0.016684694215655327, "grad_norm": 0.12145861983299255, "global_step": 22583, "epoch": 213, "lr": 6.150216638414555e-05} +{"train_loss": 0.016408156603574753, "grad_norm": 0.08134899288415909, "global_step": 22584, "epoch": 213, "lr": 6.149928208435814e-05} +{"train_loss": 0.01746009849011898, "grad_norm": 0.08339138329029083, "global_step": 22585, "epoch": 213, "lr": 6.149639774416724e-05} +{"train_loss": 0.016698196530342102, "grad_norm": 0.13226526975631714, "global_step": 22586, "epoch": 213, "lr": 6.149351336358302e-05} +{"train_loss": 0.015967434272170067, "grad_norm": 0.07365628331899643, "global_step": 22587, "epoch": 213, "lr": 6.149062894261557e-05} +{"train_loss": 0.01547587662935257, "grad_norm": 0.05884310603141785, "global_step": 22588, "epoch": 213, "lr": 6.148774448127507e-05} +{"train_loss": 0.015494780614972115, "grad_norm": 0.06945919245481491, "global_step": 22589, "epoch": 213, "lr": 6.14848599795716e-05} +{"train_loss": 0.014192819595336914, "grad_norm": 0.08208396285772324, "global_step": 22590, "epoch": 213, "lr": 6.148197543751535e-05} +{"train_loss": 0.016429539769887924, "grad_norm": 0.14039301872253418, "global_step": 22591, "epoch": 213, "lr": 6.147909085511642e-05} +{"train_loss": 0.018784351646900177, "grad_norm": 0.05764765664935112, "global_step": 22592, "epoch": 213, "lr": 6.147620623238498e-05} +{"train_loss": 0.017016705125570297, "grad_norm": 0.1860041320323944, "global_step": 22593, "epoch": 213, "lr": 6.147332156933111e-05} +{"train_loss": 0.016893398016691208, "grad_norm": 0.06426798552274704, "global_step": 22594, "epoch": 213, "lr": 6.1470436865965e-05} +{"train_loss": 0.013162946328520775, "grad_norm": 0.13384807109832764, "global_step": 22595, "epoch": 213, "lr": 6.146755212229676e-05} +{"train_loss": 0.014484401792287827, "grad_norm": 0.07627754658460617, "global_step": 22596, "epoch": 213, "lr": 6.146466733833651e-05} +{"train_loss": 0.016382433474063873, "grad_norm": 0.09072411805391312, "global_step": 22597, "epoch": 213, "lr": 6.146178251409443e-05} +{"train_loss": 0.018549997359514236, "grad_norm": 0.10324574261903763, "global_step": 22598, "epoch": 213, "lr": 6.145889764958061e-05} +{"train_loss": 0.01614735648036003, "grad_norm": 0.0740964263677597, "global_step": 22599, "epoch": 213, "lr": 6.145601274480521e-05} +{"train_loss": 0.01667928695678711, "grad_norm": 0.09995201975107193, "global_step": 22600, "epoch": 213, "lr": 6.145312779977839e-05} +{"train_loss": 0.014761483296751976, "grad_norm": 0.045392654836177826, "global_step": 22601, "epoch": 213, "lr": 6.145024281451023e-05} +{"train_loss": 0.015256790444254875, "grad_norm": 0.09992317855358124, "global_step": 22602, "epoch": 213, "lr": 6.144735778901091e-05} +{"train_loss": 0.01507707592099905, "grad_norm": 0.08501929044723511, "global_step": 22603, "epoch": 213, "lr": 6.144447272329054e-05} +{"train_loss": 0.017291875556111336, "grad_norm": 0.10313167423009872, "global_step": 22604, "epoch": 213, "lr": 6.144158761735928e-05} +{"train_loss": 0.014557748101651669, "grad_norm": 0.057609789073467255, "global_step": 22605, "epoch": 213, "lr": 6.143870247122727e-05} +{"train_loss": 0.015373112633824348, "grad_norm": 0.058919213712215424, "global_step": 22606, "epoch": 213, "lr": 6.143581728490461e-05} +{"train_loss": 0.016706474125385284, "grad_norm": 0.12261290848255157, "global_step": 22607, "epoch": 213, "lr": 6.143293205840148e-05} +{"train_loss": 0.01769944280385971, "grad_norm": 0.15071456134319305, "global_step": 22608, "epoch": 213, "lr": 6.143004679172799e-05} +{"train_loss": 0.015901722013950348, "grad_norm": 0.07495511323213577, "global_step": 22609, "epoch": 213, "lr": 6.14271614848943e-05} +{"train_loss": 0.017871107906103134, "grad_norm": 0.05211123451590538, "global_step": 22610, "epoch": 213, "lr": 6.142427613791052e-05} +{"train_loss": 0.015856608748435974, "grad_norm": 0.06934152543544769, "global_step": 22611, "epoch": 213, "lr": 6.142139075078678e-05} +{"train_loss": 0.01594134233891964, "grad_norm": 0.11811748147010803, "global_step": 22612, "epoch": 213, "lr": 6.141850532353328e-05} +{"train_loss": 0.014768369495868683, "grad_norm": 0.07294169813394547, "global_step": 22613, "epoch": 213, "lr": 6.141561985616009e-05} +{"train_loss": 0.013154616579413414, "grad_norm": 0.06914298236370087, "global_step": 22614, "epoch": 213, "lr": 6.14127343486774e-05} +{"train_loss": 0.017571667209267616, "grad_norm": 0.09630921483039856, "global_step": 22615, "epoch": 213, "lr": 6.140984880109531e-05} +{"train_loss": 0.01655549183487892, "grad_norm": 0.07145149260759354, "global_step": 22616, "epoch": 213, "lr": 6.140696321342398e-05} +{"train_loss": 0.014352511614561081, "grad_norm": 0.09145695716142654, "global_step": 22617, "epoch": 213, "lr": 6.140407758567353e-05} +{"train_loss": 0.018419934436678886, "grad_norm": 0.11589807271957397, "global_step": 22618, "epoch": 213, "lr": 6.140119191785412e-05} +{"train_loss": 0.014840645715594292, "grad_norm": 0.08934598416090012, "global_step": 22619, "epoch": 213, "lr": 6.139830620997588e-05} +{"train_loss": 0.015172039158642292, "grad_norm": 0.11183957010507584, "global_step": 22620, "epoch": 213, "lr": 6.139542046204894e-05} +{"train_loss": 0.016656730324029922, "grad_norm": 0.10590954124927521, "global_step": 22621, "epoch": 213, "lr": 6.139253467408347e-05} +{"train_loss": 0.014476967975497246, "grad_norm": 0.10495628416538239, "global_step": 22622, "epoch": 213, "lr": 6.138964884608957e-05} +{"train_loss": 0.015111107379198074, "grad_norm": 0.1275629699230194, "global_step": 22623, "epoch": 213, "lr": 6.138676297807739e-05} +{"train_loss": 0.015786191448569298, "grad_norm": 0.07609722763299942, "global_step": 22624, "epoch": 213, "lr": 6.13838770700571e-05} +{"train_loss": 0.017673591151833534, "grad_norm": 0.08529842644929886, "global_step": 22625, "epoch": 213, "lr": 6.138099112203881e-05} +{"train_loss": 0.015218226239085197, "grad_norm": 0.08758148550987244, "global_step": 22626, "epoch": 213, "lr": 6.137810513403265e-05} +{"train_loss": 0.016195137053728104, "grad_norm": 0.08980940282344818, "global_step": 22627, "epoch": 213, "lr": 6.13752191060488e-05} +{"train_loss": 0.014775849878787994, "grad_norm": 0.06460276246070862, "global_step": 22628, "epoch": 213, "lr": 6.137233303809735e-05} +{"train_loss": 0.014026190154254436, "grad_norm": 0.08030392974615097, "global_step": 22629, "epoch": 213, "lr": 6.136944693018848e-05} +{"train_loss": 0.016197560355067253, "grad_norm": 0.06299524009227753, "global_step": 22630, "epoch": 213, "lr": 6.136656078233232e-05} +{"train_loss": 0.01564864069223404, "grad_norm": 0.07661500573158264, "global_step": 22631, "epoch": 213, "lr": 6.136367459453902e-05} +{"train_loss": 0.012518337927758694, "grad_norm": 0.1302487850189209, "global_step": 22632, "epoch": 213, "lr": 6.13607883668187e-05} +{"train_loss": 0.014898453839123249, "grad_norm": 0.12787973880767822, "global_step": 22633, "epoch": 213, "lr": 6.13579020991815e-05} +{"train_loss": 0.017900168895721436, "grad_norm": 0.08160622417926788, "global_step": 22634, "epoch": 213, "lr": 6.135501579163758e-05} +{"train_loss": 0.016349181532859802, "grad_norm": 0.1328376680612564, "global_step": 22635, "epoch": 213, "lr": 6.135212944419708e-05} +{"train_loss": 0.015374850481748581, "grad_norm": 0.11455775052309036, "global_step": 22636, "epoch": 213, "lr": 6.134924305687012e-05} +{"train_loss": 0.01723705418407917, "grad_norm": 0.08316077291965485, "global_step": 22637, "epoch": 213, "lr": 6.134635662966687e-05} +{"train_loss": 0.016616661101579666, "grad_norm": 0.09490106254816055, "global_step": 22638, "epoch": 213, "lr": 6.134347016259745e-05} +{"train_loss": 0.017070718109607697, "grad_norm": 0.11673960834741592, "global_step": 22639, "epoch": 213, "lr": 6.1340583655672e-05} +{"train_loss": 0.016673162579536438, "grad_norm": 0.08353164792060852, "global_step": 22640, "epoch": 213, "lr": 6.133769710890067e-05} +{"train_loss": 0.013435354456305504, "grad_norm": 0.090370312333107, "global_step": 22641, "epoch": 213, "lr": 6.133481052229363e-05} +{"train_loss": 0.016391338780522346, "grad_norm": 0.06503285467624664, "global_step": 22642, "epoch": 213, "lr": 6.133192389586097e-05} +{"train_loss": 0.017941914498806, "grad_norm": 0.06120501086115837, "global_step": 22643, "epoch": 213, "lr": 6.132903722961286e-05} +{"train_loss": 0.015604047104716301, "grad_norm": 0.15000304579734802, "global_step": 22644, "epoch": 213, "lr": 6.132615052355945e-05} +{"train_loss": 0.01608380861580372, "grad_norm": 0.06092950329184532, "global_step": 22645, "epoch": 213, "lr": 6.132326377771086e-05} +{"train_loss": 0.015096801333129406, "grad_norm": 0.09456679224967957, "global_step": 22646, "epoch": 213, "lr": 6.132037699207726e-05} +{"train_loss": 0.012858772650361061, "grad_norm": 0.06596601754426956, "global_step": 22647, "epoch": 213, "lr": 6.131749016666877e-05} +{"train_loss": 0.01782362163066864, "grad_norm": 0.12485501915216446, "global_step": 22648, "epoch": 213, "lr": 6.131460330149554e-05} +{"train_loss": 0.01609318144619465, "grad_norm": 0.054895587265491486, "global_step": 22649, "epoch": 213, "lr": 6.131171639656772e-05} +{"train_loss": 0.014853824861347675, "grad_norm": 0.09042062610387802, "global_step": 22650, "epoch": 213, "lr": 6.130882945189545e-05} +{"train_loss": 0.017142467200756073, "grad_norm": 0.08201266080141068, "global_step": 22651, "epoch": 213, "lr": 6.130594246748887e-05} +{"train_loss": 0.01616733893752098, "grad_norm": 0.05445320904254913, "global_step": 22652, "epoch": 213, "lr": 6.130305544335811e-05} +{"train_loss": 0.017266687005758286, "grad_norm": 0.058679524809122086, "global_step": 22653, "epoch": 213, "lr": 6.130016837951336e-05} +{"train_loss": 0.016006212681531906, "grad_norm": 0.09150351583957672, "global_step": 22654, "epoch": 213, "lr": 6.129728127596468e-05} +{"train_loss": 0.016619035974144936, "grad_norm": 0.07181491702795029, "global_step": 22655, "epoch": 213, "lr": 6.129439413272231e-05} +{"train_loss": 0.016526665538549423, "grad_norm": 0.07408144325017929, "global_step": 22656, "epoch": 213, "lr": 6.129150694979636e-05} +{"train_loss": 0.017778834328055382, "grad_norm": 0.07358867675065994, "global_step": 22657, "epoch": 213, "lr": 6.128861972719693e-05} +{"train_loss": 0.015095008537173271, "grad_norm": 0.07095479965209961, "global_step": 22658, "epoch": 213, "lr": 6.128573246493423e-05} +{"train_loss": 0.014846350997686386, "grad_norm": 0.09173612296581268, "global_step": 22659, "epoch": 213, "lr": 6.128284516301836e-05} +{"train_loss": 0.01569528505206108, "grad_norm": 0.09469900280237198, "global_step": 22660, "epoch": 213, "lr": 6.127995782145947e-05} +{"train_loss": 0.01833125576376915, "grad_norm": 0.05234703794121742, "global_step": 22661, "epoch": 213, "lr": 6.127707044026773e-05} +{"train_loss": 0.016399744898080826, "grad_norm": 0.11920054256916046, "global_step": 22662, "epoch": 213, "lr": 6.127418301945325e-05} +{"train_loss": 0.015283729881048203, "grad_norm": 0.07648025453090668, "global_step": 22663, "epoch": 213, "lr": 6.12712955590262e-05} +{"train_loss": 0.016353491693735123, "grad_norm": 0.08582989871501923, "global_step": 22664, "epoch": 213, "lr": 6.126840805899672e-05} +{"train_loss": 0.013651402667164803, "grad_norm": 0.06563981622457504, "global_step": 22665, "epoch": 213, "lr": 6.126552051937497e-05} +{"train_loss": 0.015228768810629845, "grad_norm": 0.10061417520046234, "global_step": 22666, "epoch": 213, "lr": 6.126263294017107e-05} +{"train_loss": 0.017052508890628815, "grad_norm": 0.08240921795368195, "global_step": 22667, "epoch": 213, "lr": 6.125974532139517e-05} +{"train_loss": 0.016379961743950844, "grad_norm": 0.05859340354800224, "global_step": 22668, "epoch": 213, "lr": 6.125685766305743e-05} +{"train_loss": 0.013432543724775314, "grad_norm": 0.09198905527591705, "global_step": 22669, "epoch": 213, "lr": 6.125396996516798e-05} +{"train_loss": 0.015266859903931618, "grad_norm": 0.08533906936645508, "global_step": 22670, "epoch": 213, "lr": 6.125108222773698e-05} +{"train_loss": 0.015636496245861053, "grad_norm": 0.1510918140411377, "global_step": 22671, "epoch": 213, "lr": 6.124819445077457e-05} +{"train_loss": 0.01487375982105732, "grad_norm": 0.06174599379301071, "global_step": 22672, "epoch": 213, "lr": 6.124530663429089e-05} +{"train_loss": 0.014314249157905579, "grad_norm": 0.1141582801938057, "global_step": 22673, "epoch": 213, "lr": 6.124241877829609e-05} +{"train_loss": 0.0152910640463233, "grad_norm": 0.08956696838140488, "global_step": 22674, "epoch": 213, "lr": 6.123953088280033e-05} +{"train_loss": 0.0153204882517457, "grad_norm": 0.08816356211900711, "global_step": 22675, "epoch": 213, "lr": 6.123664294781374e-05} +{"train_loss": 0.016094444319605827, "grad_norm": 0.11618027836084366, "global_step": 22676, "epoch": 213, "lr": 6.12337549733465e-05} +{"train_loss": 0.013663495890796185, "grad_norm": 0.06924653798341751, "global_step": 22677, "epoch": 213, "lr": 6.123086695940869e-05} +{"train_loss": 0.014664238318800926, "grad_norm": 0.06133144721388817, "global_step": 22678, "epoch": 213, "lr": 6.122797890601052e-05} +{"train_loss": 0.016543041914701462, "grad_norm": 0.07793606072664261, "global_step": 22679, "epoch": 213, "lr": 6.12250908131621e-05} +{"train_loss": 0.01379571296274662, "grad_norm": 0.07865504175424576, "global_step": 22680, "epoch": 213, "lr": 6.122220268087361e-05} +{"train_loss": 0.016368389129638672, "grad_norm": 0.131895512342453, "global_step": 22681, "epoch": 213, "lr": 6.121931450915517e-05} +{"train_loss": 0.01958247646689415, "grad_norm": 0.11818206310272217, "global_step": 22682, "epoch": 213, "lr": 6.121642629801695e-05} +{"train_loss": 0.015868314875746675, "grad_norm": 0.12665581703186035, "global_step": 22683, "epoch": 213, "lr": 6.121353804746907e-05, "val_loss": 0.019070979207754135} +{"train_loss": 0.015053878538310528, "grad_norm": 0.053966689854860306, "global_step": 22684, "epoch": 214, "lr": 6.121064975752169e-05} +{"train_loss": 0.012625018134713173, "grad_norm": 0.13885867595672607, "global_step": 22685, "epoch": 214, "lr": 6.120776142818499e-05} +{"train_loss": 0.017483564093708992, "grad_norm": 0.06812161207199097, "global_step": 22686, "epoch": 214, "lr": 6.120487305946906e-05} +{"train_loss": 0.0151685681194067, "grad_norm": 0.11336040496826172, "global_step": 22687, "epoch": 214, "lr": 6.12019846513841e-05} +{"train_loss": 0.01704508066177368, "grad_norm": 0.053361617028713226, "global_step": 22688, "epoch": 214, "lr": 6.119909620394022e-05} +{"train_loss": 0.020051468163728714, "grad_norm": 0.06618383526802063, "global_step": 22689, "epoch": 214, "lr": 6.11962077171476e-05} +{"train_loss": 0.017296429723501205, "grad_norm": 0.06120287999510765, "global_step": 22690, "epoch": 214, "lr": 6.119331919101636e-05} +{"train_loss": 0.017144344747066498, "grad_norm": 0.08025362342596054, "global_step": 22691, "epoch": 214, "lr": 6.119043062555669e-05} +{"train_loss": 0.016147710382938385, "grad_norm": 0.07593422383069992, "global_step": 22692, "epoch": 214, "lr": 6.118754202077869e-05} +{"train_loss": 0.015208526514470577, "grad_norm": 0.07591373473405838, "global_step": 22693, "epoch": 214, "lr": 6.118465337669254e-05} +{"train_loss": 0.016279403120279312, "grad_norm": 0.09924490749835968, "global_step": 22694, "epoch": 214, "lr": 6.118176469330839e-05} +{"train_loss": 0.01621946692466736, "grad_norm": 0.20991788804531097, "global_step": 22695, "epoch": 214, "lr": 6.117887597063638e-05} +{"train_loss": 0.016742896288633347, "grad_norm": 0.0699024349451065, "global_step": 22696, "epoch": 214, "lr": 6.117598720868663e-05} +{"train_loss": 0.015918130055069923, "grad_norm": 0.07889315485954285, "global_step": 22697, "epoch": 214, "lr": 6.117309840746937e-05} +{"train_loss": 0.014305838383734226, "grad_norm": 0.08713648468255997, "global_step": 22698, "epoch": 214, "lr": 6.117020956699466e-05} +{"train_loss": 0.01644175499677658, "grad_norm": 0.05915002152323723, "global_step": 22699, "epoch": 214, "lr": 6.116732068727271e-05} +{"train_loss": 0.013723544776439667, "grad_norm": 0.08967415988445282, "global_step": 22700, "epoch": 214, "lr": 6.116443176831366e-05} +{"train_loss": 0.01595166325569153, "grad_norm": 0.077779121696949, "global_step": 22701, "epoch": 214, "lr": 6.116154281012765e-05} +{"train_loss": 0.014746307395398617, "grad_norm": 0.055386416614055634, "global_step": 22702, "epoch": 214, "lr": 6.115865381272483e-05} +{"train_loss": 0.01562124490737915, "grad_norm": 0.17193135619163513, "global_step": 22703, "epoch": 214, "lr": 6.115576477611534e-05} +{"train_loss": 0.013389571569859982, "grad_norm": 0.15880325436592102, "global_step": 22704, "epoch": 214, "lr": 6.115287570030936e-05} +{"train_loss": 0.01829157955944538, "grad_norm": 0.050358086824417114, "global_step": 22705, "epoch": 214, "lr": 6.114998658531702e-05} +{"train_loss": 0.016892801970243454, "grad_norm": 0.10540381073951721, "global_step": 22706, "epoch": 214, "lr": 6.114709743114847e-05} +{"train_loss": 0.016516035422682762, "grad_norm": 0.07842482626438141, "global_step": 22707, "epoch": 214, "lr": 6.114420823781387e-05} +{"train_loss": 0.015009667724370956, "grad_norm": 0.06362024694681168, "global_step": 22708, "epoch": 214, "lr": 6.114131900532337e-05} +{"train_loss": 0.015422360971570015, "grad_norm": 0.10817652195692062, "global_step": 22709, "epoch": 214, "lr": 6.113842973368714e-05} +{"train_loss": 0.015068440698087215, "grad_norm": 0.06230948120355606, "global_step": 22710, "epoch": 214, "lr": 6.113554042291527e-05} +{"train_loss": 0.015493527986109257, "grad_norm": 0.07678206264972687, "global_step": 22711, "epoch": 214, "lr": 6.113265107301798e-05} +{"train_loss": 0.015102334320545197, "grad_norm": 0.09842176735401154, "global_step": 22712, "epoch": 214, "lr": 6.11297616840054e-05} +{"train_loss": 0.016641996800899506, "grad_norm": 0.0810776799917221, "global_step": 22713, "epoch": 214, "lr": 6.112687225588766e-05} +{"train_loss": 0.015262826345860958, "grad_norm": 0.1340099275112152, "global_step": 22714, "epoch": 214, "lr": 6.112398278867494e-05} +{"train_loss": 0.016152359545230865, "grad_norm": 0.0639750212430954, "global_step": 22715, "epoch": 214, "lr": 6.112109328237739e-05} +{"train_loss": 0.016785219311714172, "grad_norm": 0.06518292427062988, "global_step": 22716, "epoch": 214, "lr": 6.111820373700514e-05} +{"train_loss": 0.016271494328975677, "grad_norm": 0.0886882022023201, "global_step": 22717, "epoch": 214, "lr": 6.111531415256837e-05} +{"train_loss": 0.013986078090965748, "grad_norm": 0.058672402054071426, "global_step": 22718, "epoch": 214, "lr": 6.111242452907721e-05} +{"train_loss": 0.013946223072707653, "grad_norm": 0.09122082591056824, "global_step": 22719, "epoch": 214, "lr": 6.110953486654183e-05} +{"train_loss": 0.01561756618320942, "grad_norm": 0.12035620212554932, "global_step": 22720, "epoch": 214, "lr": 6.110664516497237e-05} +{"train_loss": 0.017367083579301834, "grad_norm": 0.09097637236118317, "global_step": 22721, "epoch": 214, "lr": 6.110375542437899e-05} +{"train_loss": 0.015712562948465347, "grad_norm": 0.16077545285224915, "global_step": 22722, "epoch": 214, "lr": 6.110086564477184e-05} +{"train_loss": 0.01616576686501503, "grad_norm": 0.05186936631798744, "global_step": 22723, "epoch": 214, "lr": 6.109797582616107e-05} +{"train_loss": 0.01717105694115162, "grad_norm": 0.08297381550073624, "global_step": 22724, "epoch": 214, "lr": 6.109508596855687e-05} +{"train_loss": 0.013605398125946522, "grad_norm": 0.05979324132204056, "global_step": 22725, "epoch": 214, "lr": 6.109219607196933e-05} +{"train_loss": 0.015223206952214241, "grad_norm": 0.08412157744169235, "global_step": 22726, "epoch": 214, "lr": 6.108930613640865e-05} +{"train_loss": 0.015347693115472794, "grad_norm": 0.0754770040512085, "global_step": 22727, "epoch": 214, "lr": 6.108641616188497e-05} +{"train_loss": 0.016310200095176697, "grad_norm": 0.11894889175891876, "global_step": 22728, "epoch": 214, "lr": 6.108352614840845e-05} +{"train_loss": 0.016194259747862816, "grad_norm": 0.07486891001462936, "global_step": 22729, "epoch": 214, "lr": 6.108063609598923e-05} +{"train_loss": 0.015555862337350845, "grad_norm": 0.07905598729848862, "global_step": 22730, "epoch": 214, "lr": 6.107774600463748e-05} +{"train_loss": 0.016920261085033417, "grad_norm": 0.05207059904932976, "global_step": 22731, "epoch": 214, "lr": 6.107485587436334e-05} +{"train_loss": 0.0176627766340971, "grad_norm": 0.06029297783970833, "global_step": 22732, "epoch": 214, "lr": 6.107196570517698e-05} +{"train_loss": 0.01591060869395733, "grad_norm": 0.06147344782948494, "global_step": 22733, "epoch": 214, "lr": 6.106907549708856e-05} +{"train_loss": 0.01495431736111641, "grad_norm": 0.08757176995277405, "global_step": 22734, "epoch": 214, "lr": 6.106618525010818e-05} +{"train_loss": 0.014056443236768246, "grad_norm": 0.054609041661024094, "global_step": 22735, "epoch": 214, "lr": 6.106329496424608e-05} +{"train_loss": 0.01271725632250309, "grad_norm": 0.08099626749753952, "global_step": 22736, "epoch": 214, "lr": 6.106040463951237e-05} +{"train_loss": 0.012695129960775375, "grad_norm": 0.08213739097118378, "global_step": 22737, "epoch": 214, "lr": 6.105751427591719e-05} +{"train_loss": 0.016840524971485138, "grad_norm": 0.06661432236433029, "global_step": 22738, "epoch": 214, "lr": 6.105462387347072e-05} +{"train_loss": 0.015155920758843422, "grad_norm": 0.0744260773062706, "global_step": 22739, "epoch": 214, "lr": 6.10517334321831e-05} +{"train_loss": 0.01604672521352768, "grad_norm": 0.12375032901763916, "global_step": 22740, "epoch": 214, "lr": 6.104884295206452e-05} +{"train_loss": 0.018271686509251595, "grad_norm": 0.06717780977487564, "global_step": 22741, "epoch": 214, "lr": 6.104595243312508e-05} +{"train_loss": 0.01601787842810154, "grad_norm": 0.05834762752056122, "global_step": 22742, "epoch": 214, "lr": 6.1043061875375e-05} +{"train_loss": 0.01624472066760063, "grad_norm": 0.08048538118600845, "global_step": 22743, "epoch": 214, "lr": 6.104017127882437e-05} +{"train_loss": 0.014583888463675976, "grad_norm": 0.08310897648334503, "global_step": 22744, "epoch": 214, "lr": 6.10372806434834e-05} +{"train_loss": 0.01547008752822876, "grad_norm": 0.04706120491027832, "global_step": 22745, "epoch": 214, "lr": 6.103438996936223e-05} +{"train_loss": 0.017506510019302368, "grad_norm": 0.07381176948547363, "global_step": 22746, "epoch": 214, "lr": 6.1031499256471e-05} +{"train_loss": 0.01599765196442604, "grad_norm": 0.09556696563959122, "global_step": 22747, "epoch": 214, "lr": 6.102860850481988e-05} +{"train_loss": 0.015069812536239624, "grad_norm": 0.06305291503667831, "global_step": 22748, "epoch": 214, "lr": 6.102571771441903e-05} +{"train_loss": 0.015422162599861622, "grad_norm": 0.05814586579799652, "global_step": 22749, "epoch": 214, "lr": 6.102282688527859e-05} +{"train_loss": 0.014480339363217354, "grad_norm": 0.07066275924444199, "global_step": 22750, "epoch": 214, "lr": 6.101993601740876e-05} +{"train_loss": 0.017544077709317207, "grad_norm": 0.09894011169672012, "global_step": 22751, "epoch": 214, "lr": 6.1017045110819636e-05} +{"train_loss": 0.014360876753926277, "grad_norm": 0.07622494548559189, "global_step": 22752, "epoch": 214, "lr": 6.101415416552142e-05} +{"train_loss": 0.018351394683122635, "grad_norm": 0.06686196476221085, "global_step": 22753, "epoch": 214, "lr": 6.101126318152427e-05} +{"train_loss": 0.013834211975336075, "grad_norm": 0.0718332827091217, "global_step": 22754, "epoch": 214, "lr": 6.10083721588383e-05} +{"train_loss": 0.015778731554746628, "grad_norm": 0.057239778339862823, "global_step": 22755, "epoch": 214, "lr": 6.100548109747372e-05} +{"train_loss": 0.014975909143686295, "grad_norm": 0.062112100422382355, "global_step": 22756, "epoch": 214, "lr": 6.100258999744065e-05} +{"train_loss": 0.0148171316832304, "grad_norm": 0.09863913059234619, "global_step": 22757, "epoch": 214, "lr": 6.0999698858749276e-05} +{"train_loss": 0.016629740595817566, "grad_norm": 0.07311734557151794, "global_step": 22758, "epoch": 214, "lr": 6.099680768140974e-05} +{"train_loss": 0.013778295367956161, "grad_norm": 0.060301944613456726, "global_step": 22759, "epoch": 214, "lr": 6.0993916465432213e-05} +{"train_loss": 0.015857022255659103, "grad_norm": 0.0656035766005516, "global_step": 22760, "epoch": 214, "lr": 6.0991025210826834e-05} +{"train_loss": 0.015059992671012878, "grad_norm": 0.07898586988449097, "global_step": 22761, "epoch": 214, "lr": 6.0988133917603765e-05} +{"train_loss": 0.019432954490184784, "grad_norm": 0.06509273499250412, "global_step": 22762, "epoch": 214, "lr": 6.0985242585773185e-05} +{"train_loss": 0.015938596799969673, "grad_norm": 0.0722893550992012, "global_step": 22763, "epoch": 214, "lr": 6.098235121534523e-05} +{"train_loss": 0.014818531461060047, "grad_norm": 0.100294329226017, "global_step": 22764, "epoch": 214, "lr": 6.0979459806330076e-05} +{"train_loss": 0.017659759148955345, "grad_norm": 0.0766601711511612, "global_step": 22765, "epoch": 214, "lr": 6.097656835873788e-05} +{"train_loss": 0.016741231083869934, "grad_norm": 0.059289056807756424, "global_step": 22766, "epoch": 214, "lr": 6.0973676872578786e-05} +{"train_loss": 0.01562132965773344, "grad_norm": 0.07081300020217896, "global_step": 22767, "epoch": 214, "lr": 6.097078534786297e-05} +{"train_loss": 0.01619519479572773, "grad_norm": 0.06653357297182083, "global_step": 22768, "epoch": 214, "lr": 6.0967893784600585e-05} +{"train_loss": 0.018175940960645676, "grad_norm": 0.14115437865257263, "global_step": 22769, "epoch": 214, "lr": 6.096500218280179e-05} +{"train_loss": 0.015583178959786892, "grad_norm": 0.054832421243190765, "global_step": 22770, "epoch": 214, "lr": 6.0962110542476746e-05} +{"train_loss": 0.019192852079868317, "grad_norm": 0.20092004537582397, "global_step": 22771, "epoch": 214, "lr": 6.095921886363561e-05} +{"train_loss": 0.016293583437800407, "grad_norm": 0.1665869802236557, "global_step": 22772, "epoch": 214, "lr": 6.095632714628855e-05} +{"train_loss": 0.016887271776795387, "grad_norm": 0.1656091809272766, "global_step": 22773, "epoch": 214, "lr": 6.095343539044571e-05} +{"train_loss": 0.016948245465755463, "grad_norm": 0.0781615823507309, "global_step": 22774, "epoch": 214, "lr": 6.095054359611728e-05} +{"train_loss": 0.01576012559235096, "grad_norm": 0.06792444735765457, "global_step": 22775, "epoch": 214, "lr": 6.0947651763313395e-05} +{"train_loss": 0.016658294945955276, "grad_norm": 0.057102005928754807, "global_step": 22776, "epoch": 214, "lr": 6.094475989204421e-05} +{"train_loss": 0.01704208180308342, "grad_norm": 0.06203453242778778, "global_step": 22777, "epoch": 214, "lr": 6.094186798231992e-05} +{"train_loss": 0.013993609696626663, "grad_norm": 0.073360376060009, "global_step": 22778, "epoch": 214, "lr": 6.093897603415064e-05} +{"train_loss": 0.017137203365564346, "grad_norm": 0.16712595522403717, "global_step": 22779, "epoch": 214, "lr": 6.093608404754657e-05} +{"train_loss": 0.015606564469635487, "grad_norm": 0.13404855132102966, "global_step": 22780, "epoch": 214, "lr": 6.093319202251786e-05} +{"train_loss": 0.013095357455313206, "grad_norm": 0.1305471956729889, "global_step": 22781, "epoch": 214, "lr": 6.093029995907466e-05} +{"train_loss": 0.015252591110765934, "grad_norm": 0.07134976983070374, "global_step": 22782, "epoch": 214, "lr": 6.0927407857227146e-05} +{"train_loss": 0.01572927087545395, "grad_norm": 0.09899213910102844, "global_step": 22783, "epoch": 214, "lr": 6.092451571698546e-05} +{"train_loss": 0.014468317851424217, "grad_norm": 0.16596202552318573, "global_step": 22784, "epoch": 214, "lr": 6.0921623538359795e-05} +{"train_loss": 0.016804657876491547, "grad_norm": 0.08114958554506302, "global_step": 22785, "epoch": 214, "lr": 6.091873132136029e-05} +{"train_loss": 0.016789725050330162, "grad_norm": 0.08200520277023315, "global_step": 22786, "epoch": 214, "lr": 6.0915839065997104e-05} +{"train_loss": 0.015790138393640518, "grad_norm": 0.05185891315340996, "global_step": 22787, "epoch": 214, "lr": 6.091294677228041e-05} +{"train_loss": 0.016585808247327805, "grad_norm": 0.11986991763114929, "global_step": 22788, "epoch": 214, "lr": 6.091005444022036e-05} +{"train_loss": 0.015852631957112055, "grad_norm": 0.16308479011058807, "global_step": 22789, "epoch": 214, "lr": 6.090716206982714e-05, "val_loss": 0.019853699952363968} +{"train_loss": 0.018404919654130936, "grad_norm": 0.0600145161151886, "global_step": 22790, "epoch": 215, "lr": 6.090426966111087e-05} +{"train_loss": 0.015675013884902, "grad_norm": 0.052927274256944656, "global_step": 22791, "epoch": 215, "lr": 6.0901377214081767e-05} +{"train_loss": 0.014328311197459698, "grad_norm": 0.10485221445560455, "global_step": 22792, "epoch": 215, "lr": 6.089848472874995e-05} +{"train_loss": 0.015518260188400745, "grad_norm": 0.05735195428133011, "global_step": 22793, "epoch": 215, "lr": 6.0895592205125594e-05} +{"train_loss": 0.017586031928658485, "grad_norm": 0.0818725973367691, "global_step": 22794, "epoch": 215, "lr": 6.089269964321889e-05} +{"train_loss": 0.015359150245785713, "grad_norm": 0.07180462777614594, "global_step": 22795, "epoch": 215, "lr": 6.0889807043039947e-05} +{"train_loss": 0.015135172754526138, "grad_norm": 0.05348873510956764, "global_step": 22796, "epoch": 215, "lr": 6.088691440459897e-05} +{"train_loss": 0.014798576943576336, "grad_norm": 0.06964205950498581, "global_step": 22797, "epoch": 215, "lr": 6.0884021727906104e-05} +{"train_loss": 0.013464384712278843, "grad_norm": 0.1045183464884758, "global_step": 22798, "epoch": 215, "lr": 6.0881129012971525e-05} +{"train_loss": 0.01717308536171913, "grad_norm": 0.09508753567934036, "global_step": 22799, "epoch": 215, "lr": 6.0878236259805396e-05} +{"train_loss": 0.016197018325328827, "grad_norm": 0.06400890648365021, "global_step": 22800, "epoch": 215, "lr": 6.087534346841787e-05} +{"train_loss": 0.01464875228703022, "grad_norm": 0.08718618005514145, "global_step": 22801, "epoch": 215, "lr": 6.087245063881911e-05} +{"train_loss": 0.013694558292627335, "grad_norm": 0.06280338764190674, "global_step": 22802, "epoch": 215, "lr": 6.086955777101929e-05} +{"train_loss": 0.017649156972765923, "grad_norm": 0.10645045340061188, "global_step": 22803, "epoch": 215, "lr": 6.086666486502859e-05} +{"train_loss": 0.01598861813545227, "grad_norm": 0.0740148276090622, "global_step": 22804, "epoch": 215, "lr": 6.086377192085714e-05} +{"train_loss": 0.014667862094938755, "grad_norm": 0.10604981333017349, "global_step": 22805, "epoch": 215, "lr": 6.0860878938515106e-05} +{"train_loss": 0.012995894998311996, "grad_norm": 0.0625510886311531, "global_step": 22806, "epoch": 215, "lr": 6.085798591801268e-05} +{"train_loss": 0.015213776379823685, "grad_norm": 0.08632899820804596, "global_step": 22807, "epoch": 215, "lr": 6.085509285936002e-05} +{"train_loss": 0.014767168089747429, "grad_norm": 0.06763399392366409, "global_step": 22808, "epoch": 215, "lr": 6.085219976256727e-05} +{"train_loss": 0.016761265695095062, "grad_norm": 0.13749174773693085, "global_step": 22809, "epoch": 215, "lr": 6.084930662764462e-05} +{"train_loss": 0.017367731779813766, "grad_norm": 0.06214131414890289, "global_step": 22810, "epoch": 215, "lr": 6.084641345460223e-05} +{"train_loss": 0.01646392047405243, "grad_norm": 0.04755532369017601, "global_step": 22811, "epoch": 215, "lr": 6.084352024345025e-05} +{"train_loss": 0.014468315988779068, "grad_norm": 0.13866935670375824, "global_step": 22812, "epoch": 215, "lr": 6.0840626994198856e-05} +{"train_loss": 0.01596299558877945, "grad_norm": 0.06335099041461945, "global_step": 22813, "epoch": 215, "lr": 6.0837733706858216e-05} +{"train_loss": 0.015026864595711231, "grad_norm": 0.05750586465001106, "global_step": 22814, "epoch": 215, "lr": 6.0834840381438484e-05} +{"train_loss": 0.018495038151741028, "grad_norm": 0.06111885979771614, "global_step": 22815, "epoch": 215, "lr": 6.083194701794985e-05} +{"train_loss": 0.01728069595992565, "grad_norm": 0.060378383845090866, "global_step": 22816, "epoch": 215, "lr": 6.0829053616402466e-05} +{"train_loss": 0.015266695991158485, "grad_norm": 0.06527194380760193, "global_step": 22817, "epoch": 215, "lr": 6.082616017680647e-05} +{"train_loss": 0.017384923994541168, "grad_norm": 0.07241448760032654, "global_step": 22818, "epoch": 215, "lr": 6.082326669917209e-05} +{"train_loss": 0.015674423426389694, "grad_norm": 0.0770181193947792, "global_step": 22819, "epoch": 215, "lr": 6.082037318350944e-05} +{"train_loss": 0.015801649540662766, "grad_norm": 0.07594572007656097, "global_step": 22820, "epoch": 215, "lr": 6.08174796298287e-05} +{"train_loss": 0.017356589436531067, "grad_norm": 0.06510189920663834, "global_step": 22821, "epoch": 215, "lr": 6.081458603814005e-05} +{"train_loss": 0.01660497672855854, "grad_norm": 0.09245426952838898, "global_step": 22822, "epoch": 215, "lr": 6.081169240845364e-05} +{"train_loss": 0.018826384097337723, "grad_norm": 0.06586574018001556, "global_step": 22823, "epoch": 215, "lr": 6.080879874077965e-05} +{"train_loss": 0.015659242868423462, "grad_norm": 0.06440502405166626, "global_step": 22824, "epoch": 215, "lr": 6.080590503512824e-05} +{"train_loss": 0.016028180718421936, "grad_norm": 0.11488845944404602, "global_step": 22825, "epoch": 215, "lr": 6.080301129150959e-05} +{"train_loss": 0.01464911364018917, "grad_norm": 0.09130346029996872, "global_step": 22826, "epoch": 215, "lr": 6.080011750993384e-05} +{"train_loss": 0.0191442109644413, "grad_norm": 0.08906198292970657, "global_step": 22827, "epoch": 215, "lr": 6.0797223690411186e-05} +{"train_loss": 0.013677338138222694, "grad_norm": 0.07614383101463318, "global_step": 22828, "epoch": 215, "lr": 6.079432983295178e-05} +{"train_loss": 0.0178261399269104, "grad_norm": 0.08667250722646713, "global_step": 22829, "epoch": 215, "lr": 6.079143593756579e-05} +{"train_loss": 0.018438268452882767, "grad_norm": 0.14807148277759552, "global_step": 22830, "epoch": 215, "lr": 6.078854200426339e-05} +{"train_loss": 0.0138178002089262, "grad_norm": 0.07662547379732132, "global_step": 22831, "epoch": 215, "lr": 6.078564803305474e-05} +{"train_loss": 0.018142662942409515, "grad_norm": 0.19456622004508972, "global_step": 22832, "epoch": 215, "lr": 6.078275402395002e-05} +{"train_loss": 0.018027517944574356, "grad_norm": 0.09287342429161072, "global_step": 22833, "epoch": 215, "lr": 6.07798599769594e-05} +{"train_loss": 0.018165895715355873, "grad_norm": 0.07482253760099411, "global_step": 22834, "epoch": 215, "lr": 6.077696589209302e-05} +{"train_loss": 0.01818188652396202, "grad_norm": 0.06671115010976791, "global_step": 22835, "epoch": 215, "lr": 6.077407176936108e-05} +{"train_loss": 0.01654568687081337, "grad_norm": 0.09229031205177307, "global_step": 22836, "epoch": 215, "lr": 6.0771177608773735e-05} +{"train_loss": 0.013556011952459812, "grad_norm": 0.09886956959962845, "global_step": 22837, "epoch": 215, "lr": 6.076828341034115e-05} +{"train_loss": 0.016082702204585075, "grad_norm": 0.0903453528881073, "global_step": 22838, "epoch": 215, "lr": 6.076538917407352e-05} +{"train_loss": 0.015692327171564102, "grad_norm": 0.09336564689874649, "global_step": 22839, "epoch": 215, "lr": 6.076249489998097e-05} +{"train_loss": 0.018062636256217957, "grad_norm": 0.07186765968799591, "global_step": 22840, "epoch": 215, "lr": 6.0759600588073705e-05} +{"train_loss": 0.014005091041326523, "grad_norm": 0.09207842499017715, "global_step": 22841, "epoch": 215, "lr": 6.075670623836189e-05} +{"train_loss": 0.01636933535337448, "grad_norm": 0.08105505257844925, "global_step": 22842, "epoch": 215, "lr": 6.075381185085568e-05} +{"train_loss": 0.0147381741553545, "grad_norm": 0.06797001510858536, "global_step": 22843, "epoch": 215, "lr": 6.075091742556524e-05} +{"train_loss": 0.015082045458257198, "grad_norm": 0.15575538575649261, "global_step": 22844, "epoch": 215, "lr": 6.074802296250077e-05} +{"train_loss": 0.01691165566444397, "grad_norm": 0.08464013040065765, "global_step": 22845, "epoch": 215, "lr": 6.074512846167242e-05} +{"train_loss": 0.014584271237254143, "grad_norm": 0.07538646459579468, "global_step": 22846, "epoch": 215, "lr": 6.074223392309034e-05} +{"train_loss": 0.01580153778195381, "grad_norm": 0.06146892532706261, "global_step": 22847, "epoch": 215, "lr": 6.0739339346764755e-05} +{"train_loss": 0.015154100954532623, "grad_norm": 0.04880475997924805, "global_step": 22848, "epoch": 215, "lr": 6.073644473270576e-05} +{"train_loss": 0.01666388474404812, "grad_norm": 0.07191973179578781, "global_step": 22849, "epoch": 215, "lr": 6.0733550080923604e-05} +{"train_loss": 0.01646208018064499, "grad_norm": 0.06589432805776596, "global_step": 22850, "epoch": 215, "lr": 6.0730655391428414e-05} +{"train_loss": 0.013751879334449768, "grad_norm": 0.04771618917584419, "global_step": 22851, "epoch": 215, "lr": 6.072776066423036e-05} +{"train_loss": 0.016662001609802246, "grad_norm": 0.12234044075012207, "global_step": 22852, "epoch": 215, "lr": 6.072486589933962e-05} +{"train_loss": 0.015454625710844994, "grad_norm": 0.07632824033498764, "global_step": 22853, "epoch": 215, "lr": 6.072197109676637e-05} +{"train_loss": 0.016428770497441292, "grad_norm": 0.060317475348711014, "global_step": 22854, "epoch": 215, "lr": 6.071907625652078e-05} +{"train_loss": 0.014997370541095734, "grad_norm": 0.051789600402116776, "global_step": 22855, "epoch": 215, "lr": 6.0716181378613015e-05} +{"train_loss": 0.015469225123524666, "grad_norm": 0.0895785465836525, "global_step": 22856, "epoch": 215, "lr": 6.071328646305324e-05} +{"train_loss": 0.015133045613765717, "grad_norm": 0.057697005569934845, "global_step": 22857, "epoch": 215, "lr": 6.071039150985165e-05} +{"train_loss": 0.01563795655965805, "grad_norm": 0.10256022214889526, "global_step": 22858, "epoch": 215, "lr": 6.070749651901838e-05} +{"train_loss": 0.015973690897226334, "grad_norm": 0.07020442932844162, "global_step": 22859, "epoch": 215, "lr": 6.070460149056365e-05} +{"train_loss": 0.01754453405737877, "grad_norm": 0.08977489173412323, "global_step": 22860, "epoch": 215, "lr": 6.070170642449759e-05} +{"train_loss": 0.018256695941090584, "grad_norm": 0.2248886674642563, "global_step": 22861, "epoch": 215, "lr": 6.0698811320830384e-05} +{"train_loss": 0.017129845917224884, "grad_norm": 0.08316760510206223, "global_step": 22862, "epoch": 215, "lr": 6.069591617957221e-05} +{"train_loss": 0.01546589843928814, "grad_norm": 0.0688801109790802, "global_step": 22863, "epoch": 215, "lr": 6.069302100073324e-05} +{"train_loss": 0.017647339031100273, "grad_norm": 0.1222374364733696, "global_step": 22864, "epoch": 215, "lr": 6.069012578432365e-05} +{"train_loss": 0.013670568354427814, "grad_norm": 0.09813722223043442, "global_step": 22865, "epoch": 215, "lr": 6.068723053035361e-05} +{"train_loss": 0.016595173627138138, "grad_norm": 0.075954370200634, "global_step": 22866, "epoch": 215, "lr": 6.068433523883328e-05} +{"train_loss": 0.019174426794052124, "grad_norm": 0.08292697370052338, "global_step": 22867, "epoch": 215, "lr": 6.068143990977285e-05} +{"train_loss": 0.014489496126770973, "grad_norm": 0.07633484899997711, "global_step": 22868, "epoch": 215, "lr": 6.067854454318248e-05} +{"train_loss": 0.017087474465370178, "grad_norm": 0.056317128241062164, "global_step": 22869, "epoch": 215, "lr": 6.0675649139072354e-05} +{"train_loss": 0.016177164390683174, "grad_norm": 0.054356567561626434, "global_step": 22870, "epoch": 215, "lr": 6.0672753697452625e-05} +{"train_loss": 0.015469647943973541, "grad_norm": 0.07378744333982468, "global_step": 22871, "epoch": 215, "lr": 6.066985821833351e-05} +{"train_loss": 0.014593660831451416, "grad_norm": 0.08640041202306747, "global_step": 22872, "epoch": 215, "lr": 6.066696270172512e-05} +{"train_loss": 0.015129541978240013, "grad_norm": 0.09342389553785324, "global_step": 22873, "epoch": 215, "lr": 6.0664067147637684e-05} +{"train_loss": 0.014987667091190815, "grad_norm": 0.11603853106498718, "global_step": 22874, "epoch": 215, "lr": 6.066117155608135e-05} +{"train_loss": 0.01640295423567295, "grad_norm": 0.15720149874687195, "global_step": 22875, "epoch": 215, "lr": 6.0658275927066285e-05} +{"train_loss": 0.01543535478413105, "grad_norm": 0.05936023220419884, "global_step": 22876, "epoch": 215, "lr": 6.065538026060268e-05} +{"train_loss": 0.016550572589039803, "grad_norm": 0.06455874443054199, "global_step": 22877, "epoch": 215, "lr": 6.065248455670071e-05} +{"train_loss": 0.015260269865393639, "grad_norm": 0.09266481548547745, "global_step": 22878, "epoch": 215, "lr": 6.064958881537053e-05} +{"train_loss": 0.014265899546444416, "grad_norm": 0.057808905839920044, "global_step": 22879, "epoch": 215, "lr": 6.0646693036622334e-05} +{"train_loss": 0.012386775575578213, "grad_norm": 0.05513865873217583, "global_step": 22880, "epoch": 215, "lr": 6.064379722046629e-05} +{"train_loss": 0.015752246603369713, "grad_norm": 0.08954264223575592, "global_step": 22881, "epoch": 215, "lr": 6.064090136691256e-05} +{"train_loss": 0.014101406559348106, "grad_norm": 0.1284465193748474, "global_step": 22882, "epoch": 215, "lr": 6.063800547597134e-05} +{"train_loss": 0.013169514015316963, "grad_norm": 0.053480975329875946, "global_step": 22883, "epoch": 215, "lr": 6.06351095476528e-05} +{"train_loss": 0.016998283565044403, "grad_norm": 0.05596550181508064, "global_step": 22884, "epoch": 215, "lr": 6.06322135819671e-05} +{"train_loss": 0.014753006398677826, "grad_norm": 0.14696915447711945, "global_step": 22885, "epoch": 215, "lr": 6.0629317578924426e-05} +{"train_loss": 0.014224065467715263, "grad_norm": 0.08855127543210983, "global_step": 22886, "epoch": 215, "lr": 6.062642153853496e-05} +{"train_loss": 0.016501732170581818, "grad_norm": 0.07166864722967148, "global_step": 22887, "epoch": 215, "lr": 6.062352546080886e-05} +{"train_loss": 0.015439711511135101, "grad_norm": 0.05957630276679993, "global_step": 22888, "epoch": 215, "lr": 6.062062934575633e-05} +{"train_loss": 0.014806397259235382, "grad_norm": 0.18923871219158173, "global_step": 22889, "epoch": 215, "lr": 6.0617733193387506e-05} +{"train_loss": 0.01669974997639656, "grad_norm": 0.11434072256088257, "global_step": 22890, "epoch": 215, "lr": 6.061483700371259e-05} +{"train_loss": 0.014149677939713001, "grad_norm": 0.06846004724502563, "global_step": 22891, "epoch": 215, "lr": 6.061194077674176e-05} +{"train_loss": 0.016804855316877365, "grad_norm": 0.06253034621477127, "global_step": 22892, "epoch": 215, "lr": 6.060904451248518e-05} +{"train_loss": 0.01411089301109314, "grad_norm": 0.09854968637228012, "global_step": 22893, "epoch": 215, "lr": 6.060614821095303e-05} +{"train_loss": 0.015388602390885353, "grad_norm": 0.09469936043024063, "global_step": 22894, "epoch": 215, "lr": 6.06032518721555e-05} +{"train_loss": 0.015834736708059627, "grad_norm": 0.09281116724014282, "global_step": 22895, "epoch": 215, "lr": 6.0600355496102745e-05, "val_loss": 0.018733210861682892} +{"train_loss": 0.016945909708738327, "grad_norm": 0.10539905726909637, "global_step": 22896, "epoch": 216, "lr": 6.059745908280494e-05} +{"train_loss": 0.01664259098470211, "grad_norm": 0.09739135205745697, "global_step": 22897, "epoch": 216, "lr": 6.059456263227229e-05} +{"train_loss": 0.015536055900156498, "grad_norm": 0.09265489131212234, "global_step": 22898, "epoch": 216, "lr": 6.0591666144514946e-05} +{"train_loss": 0.017901644110679626, "grad_norm": 0.13951103389263153, "global_step": 22899, "epoch": 216, "lr": 6.0588769619543086e-05} +{"train_loss": 0.01635802909731865, "grad_norm": 0.08742374181747437, "global_step": 22900, "epoch": 216, "lr": 6.058587305736692e-05} +{"train_loss": 0.018155019730329514, "grad_norm": 0.08045801520347595, "global_step": 22901, "epoch": 216, "lr": 6.058297645799656e-05} +{"train_loss": 0.015450960025191307, "grad_norm": 0.05166394263505936, "global_step": 22902, "epoch": 216, "lr": 6.058007982144225e-05} +{"train_loss": 0.016776632517576218, "grad_norm": 0.14215655624866486, "global_step": 22903, "epoch": 216, "lr": 6.057718314771416e-05} +{"train_loss": 0.01768610253930092, "grad_norm": 0.19725832343101501, "global_step": 22904, "epoch": 216, "lr": 6.0574286436822415e-05} +{"train_loss": 0.01662653684616089, "grad_norm": 0.06799231469631195, "global_step": 22905, "epoch": 216, "lr": 6.0571389688777225e-05} +{"train_loss": 0.015133113600313663, "grad_norm": 0.08491774648427963, "global_step": 22906, "epoch": 216, "lr": 6.0568492903588794e-05} +{"train_loss": 0.013272427022457123, "grad_norm": 0.06068149581551552, "global_step": 22907, "epoch": 216, "lr": 6.056559608126725e-05} +{"train_loss": 0.015143169090151787, "grad_norm": 0.08874060958623886, "global_step": 22908, "epoch": 216, "lr": 6.056269922182282e-05} +{"train_loss": 0.015921201556921005, "grad_norm": 0.10893966257572174, "global_step": 22909, "epoch": 216, "lr": 6.0559802325265634e-05} +{"train_loss": 0.017202815040946007, "grad_norm": 0.06849487125873566, "global_step": 22910, "epoch": 216, "lr": 6.055690539160591e-05} +{"train_loss": 0.01724756509065628, "grad_norm": 0.09383576363325119, "global_step": 22911, "epoch": 216, "lr": 6.05540084208538e-05} +{"train_loss": 0.018166426569223404, "grad_norm": 0.14178761839866638, "global_step": 22912, "epoch": 216, "lr": 6.055111141301951e-05} +{"train_loss": 0.014031443744897842, "grad_norm": 0.14120514690876007, "global_step": 22913, "epoch": 216, "lr": 6.054821436811319e-05} +{"train_loss": 0.01579994708299637, "grad_norm": 0.06366591155529022, "global_step": 22914, "epoch": 216, "lr": 6.054531728614504e-05} +{"train_loss": 0.016459479928016663, "grad_norm": 0.12026240676641464, "global_step": 22915, "epoch": 216, "lr": 6.054242016712524e-05} +{"train_loss": 0.013892712071537971, "grad_norm": 0.08953040093183517, "global_step": 22916, "epoch": 216, "lr": 6.053952301106394e-05} +{"train_loss": 0.012148953974246979, "grad_norm": 0.055084750056266785, "global_step": 22917, "epoch": 216, "lr": 6.053662581797135e-05} +{"train_loss": 0.017252933233976364, "grad_norm": 0.08835466206073761, "global_step": 22918, "epoch": 216, "lr": 6.053372858785764e-05} +{"train_loss": 0.01672782376408577, "grad_norm": 0.07286075502634048, "global_step": 22919, "epoch": 216, "lr": 6.053083132073298e-05} +{"train_loss": 0.01489790715277195, "grad_norm": 0.06128103286027908, "global_step": 22920, "epoch": 216, "lr": 6.0527934016607565e-05} +{"train_loss": 0.01791365258395672, "grad_norm": 0.0945943295955658, "global_step": 22921, "epoch": 216, "lr": 6.0525036675491576e-05} +{"train_loss": 0.015673473477363586, "grad_norm": 0.08823640644550323, "global_step": 22922, "epoch": 216, "lr": 6.052213929739517e-05} +{"train_loss": 0.01658092625439167, "grad_norm": 0.08405942469835281, "global_step": 22923, "epoch": 216, "lr": 6.0519241882328546e-05} +{"train_loss": 0.015536144375801086, "grad_norm": 0.08982524275779724, "global_step": 22924, "epoch": 216, "lr": 6.0516344430301894e-05} +{"train_loss": 0.013459396548569202, "grad_norm": 0.09850028157234192, "global_step": 22925, "epoch": 216, "lr": 6.051344694132538e-05} +{"train_loss": 0.01901136338710785, "grad_norm": 0.0897068902850151, "global_step": 22926, "epoch": 216, "lr": 6.0510549415409166e-05} +{"train_loss": 0.015169505029916763, "grad_norm": 0.12934426963329315, "global_step": 22927, "epoch": 216, "lr": 6.050765185256347e-05} +{"train_loss": 0.018274351954460144, "grad_norm": 0.05823444575071335, "global_step": 22928, "epoch": 216, "lr": 6.0504754252798445e-05} +{"train_loss": 0.01551077887415886, "grad_norm": 0.06701314449310303, "global_step": 22929, "epoch": 216, "lr": 6.050185661612429e-05} +{"train_loss": 0.01572384685277939, "grad_norm": 0.07600223273038864, "global_step": 22930, "epoch": 216, "lr": 6.049895894255116e-05} +{"train_loss": 0.014317604713141918, "grad_norm": 0.0502617210149765, "global_step": 22931, "epoch": 216, "lr": 6.0496061232089276e-05} +{"train_loss": 0.01576233096420765, "grad_norm": 0.11391398310661316, "global_step": 22932, "epoch": 216, "lr": 6.0493163484748795e-05} +{"train_loss": 0.016595399007201195, "grad_norm": 0.16821372509002686, "global_step": 22933, "epoch": 216, "lr": 6.04902657005399e-05} +{"train_loss": 0.013699987903237343, "grad_norm": 0.07491831481456757, "global_step": 22934, "epoch": 216, "lr": 6.048736787947277e-05} +{"train_loss": 0.013019870035350323, "grad_norm": 0.1470419317483902, "global_step": 22935, "epoch": 216, "lr": 6.048447002155758e-05} +{"train_loss": 0.014591678977012634, "grad_norm": 0.12219761312007904, "global_step": 22936, "epoch": 216, "lr": 6.048157212680453e-05} +{"train_loss": 0.014412197284400463, "grad_norm": 0.044937193393707275, "global_step": 22937, "epoch": 216, "lr": 6.0478674195223804e-05} +{"train_loss": 0.01597324199974537, "grad_norm": 0.062216490507125854, "global_step": 22938, "epoch": 216, "lr": 6.0475776226825566e-05} +{"train_loss": 0.016350507736206055, "grad_norm": 0.10965686291456223, "global_step": 22939, "epoch": 216, "lr": 6.047287822162e-05} +{"train_loss": 0.01693866401910782, "grad_norm": 0.06223473325371742, "global_step": 22940, "epoch": 216, "lr": 6.0469980179617295e-05} +{"train_loss": 0.01674441248178482, "grad_norm": 0.05821703001856804, "global_step": 22941, "epoch": 216, "lr": 6.0467082100827644e-05} +{"train_loss": 0.017506618052721024, "grad_norm": 0.06319406628608704, "global_step": 22942, "epoch": 216, "lr": 6.0464183985261206e-05} +{"train_loss": 0.014932218007743359, "grad_norm": 0.05667176842689514, "global_step": 22943, "epoch": 216, "lr": 6.0461285832928184e-05} +{"train_loss": 0.014994446188211441, "grad_norm": 0.059886716306209564, "global_step": 22944, "epoch": 216, "lr": 6.045838764383876e-05} +{"train_loss": 0.012626579031348228, "grad_norm": 0.13285672664642334, "global_step": 22945, "epoch": 216, "lr": 6.045548941800309e-05} +{"train_loss": 0.017071908339858055, "grad_norm": 0.18858879804611206, "global_step": 22946, "epoch": 216, "lr": 6.045259115543139e-05} +{"train_loss": 0.016916055232286453, "grad_norm": 0.08090727776288986, "global_step": 22947, "epoch": 216, "lr": 6.0449692856133825e-05} +{"train_loss": 0.015913689509034157, "grad_norm": 0.08170405775308609, "global_step": 22948, "epoch": 216, "lr": 6.0446794520120584e-05} +{"train_loss": 0.01464966218918562, "grad_norm": 0.04974154010415077, "global_step": 22949, "epoch": 216, "lr": 6.0443896147401854e-05} +{"train_loss": 0.014739170670509338, "grad_norm": 0.06460472196340561, "global_step": 22950, "epoch": 216, "lr": 6.044099773798783e-05} +{"train_loss": 0.018367066979408264, "grad_norm": 0.17145074903964996, "global_step": 22951, "epoch": 216, "lr": 6.043809929188866e-05} +{"train_loss": 0.013685103505849838, "grad_norm": 0.08268528431653976, "global_step": 22952, "epoch": 216, "lr": 6.043520080911454e-05} +{"train_loss": 0.01449641864746809, "grad_norm": 0.09357724338769913, "global_step": 22953, "epoch": 216, "lr": 6.043230228967569e-05} +{"train_loss": 0.018674390390515327, "grad_norm": 0.09405836462974548, "global_step": 22954, "epoch": 216, "lr": 6.042940373358226e-05} +{"train_loss": 0.015483764000236988, "grad_norm": 0.08130957186222076, "global_step": 22955, "epoch": 216, "lr": 6.042650514084442e-05} +{"train_loss": 0.01753973215818405, "grad_norm": 0.07438212633132935, "global_step": 22956, "epoch": 216, "lr": 6.04236065114724e-05} +{"train_loss": 0.016115382313728333, "grad_norm": 0.1245887503027916, "global_step": 22957, "epoch": 216, "lr": 6.042070784547633e-05} +{"train_loss": 0.0175637137144804, "grad_norm": 0.06650401651859283, "global_step": 22958, "epoch": 216, "lr": 6.041780914286645e-05} +{"train_loss": 0.017330307513475418, "grad_norm": 0.08930088579654694, "global_step": 22959, "epoch": 216, "lr": 6.041491040365291e-05} +{"train_loss": 0.01546928659081459, "grad_norm": 0.08250775933265686, "global_step": 22960, "epoch": 216, "lr": 6.041201162784591e-05} +{"train_loss": 0.017141632735729218, "grad_norm": 0.17855240404605865, "global_step": 22961, "epoch": 216, "lr": 6.0409112815455624e-05} +{"train_loss": 0.01656714454293251, "grad_norm": 0.0871891900897026, "global_step": 22962, "epoch": 216, "lr": 6.0406213966492254e-05} +{"train_loss": 0.013404536992311478, "grad_norm": 0.08273150026798248, "global_step": 22963, "epoch": 216, "lr": 6.040331508096596e-05} +{"train_loss": 0.016428835690021515, "grad_norm": 0.08933258801698685, "global_step": 22964, "epoch": 216, "lr": 6.0400416158886944e-05} +{"train_loss": 0.015439228154718876, "grad_norm": 0.13689808547496796, "global_step": 22965, "epoch": 216, "lr": 6.0397517200265394e-05} +{"train_loss": 0.014069200493395329, "grad_norm": 0.10217175632715225, "global_step": 22966, "epoch": 216, "lr": 6.039461820511148e-05} +{"train_loss": 0.014259140007197857, "grad_norm": 0.08845917135477066, "global_step": 22967, "epoch": 216, "lr": 6.03917191734354e-05} +{"train_loss": 0.014371274039149284, "grad_norm": 0.09021341800689697, "global_step": 22968, "epoch": 216, "lr": 6.0388820105247355e-05} +{"train_loss": 0.016992658376693726, "grad_norm": 0.07390060275793076, "global_step": 22969, "epoch": 216, "lr": 6.038592100055749e-05} +{"train_loss": 0.017080187797546387, "grad_norm": 0.12685567140579224, "global_step": 22970, "epoch": 216, "lr": 6.038302185937603e-05} +{"train_loss": 0.01635642722249031, "grad_norm": 0.05925677716732025, "global_step": 22971, "epoch": 216, "lr": 6.038012268171314e-05} +{"train_loss": 0.01620824635028839, "grad_norm": 0.13219259679317474, "global_step": 22972, "epoch": 216, "lr": 6.037722346757901e-05} +{"train_loss": 0.012930642813444138, "grad_norm": 0.055665742605924606, "global_step": 22973, "epoch": 216, "lr": 6.037432421698385e-05} +{"train_loss": 0.012343806214630604, "grad_norm": 0.07635754346847534, "global_step": 22974, "epoch": 216, "lr": 6.0371424929937817e-05} +{"train_loss": 0.01567205786705017, "grad_norm": 0.06498623639345169, "global_step": 22975, "epoch": 216, "lr": 6.03685256064511e-05} +{"train_loss": 0.01811286248266697, "grad_norm": 0.1372373104095459, "global_step": 22976, "epoch": 216, "lr": 6.036562624653389e-05} +{"train_loss": 0.0185835137963295, "grad_norm": 0.11606502532958984, "global_step": 22977, "epoch": 216, "lr": 6.036272685019638e-05} +{"train_loss": 0.017477713525295258, "grad_norm": 0.08248896151781082, "global_step": 22978, "epoch": 216, "lr": 6.035982741744877e-05} +{"train_loss": 0.015552922151982784, "grad_norm": 0.060830648988485336, "global_step": 22979, "epoch": 216, "lr": 6.03569279483012e-05} +{"train_loss": 0.017101243138313293, "grad_norm": 0.07371368259191513, "global_step": 22980, "epoch": 216, "lr": 6.0354028442763924e-05} +{"train_loss": 0.017142849043011665, "grad_norm": 0.06688327342271805, "global_step": 22981, "epoch": 216, "lr": 6.035112890084707e-05} +{"train_loss": 0.014644021168351173, "grad_norm": 0.11024785786867142, "global_step": 22982, "epoch": 216, "lr": 6.034822932256086e-05} +{"train_loss": 0.014351408928632736, "grad_norm": 0.10536915808916092, "global_step": 22983, "epoch": 216, "lr": 6.034532970791549e-05} +{"train_loss": 0.014349877834320068, "grad_norm": 0.06168512627482414, "global_step": 22984, "epoch": 216, "lr": 6.03424300569211e-05} +{"train_loss": 0.01824416033923626, "grad_norm": 0.09034063667058945, "global_step": 22985, "epoch": 216, "lr": 6.033953036958793e-05} +{"train_loss": 0.018838994204998016, "grad_norm": 0.08729290217161179, "global_step": 22986, "epoch": 216, "lr": 6.0336630645926126e-05} +{"train_loss": 0.01635177992284298, "grad_norm": 0.0982510894536972, "global_step": 22987, "epoch": 216, "lr": 6.0333730885945915e-05} +{"train_loss": 0.016962610185146332, "grad_norm": 0.13670872151851654, "global_step": 22988, "epoch": 216, "lr": 6.0330831089657455e-05} +{"train_loss": 0.012437770143151283, "grad_norm": 0.10959111154079437, "global_step": 22989, "epoch": 216, "lr": 6.0327931257070956e-05} +{"train_loss": 0.013889187015593052, "grad_norm": 0.04549911990761757, "global_step": 22990, "epoch": 216, "lr": 6.032503138819661e-05} +{"train_loss": 0.013082362711429596, "grad_norm": 0.09460975229740143, "global_step": 22991, "epoch": 216, "lr": 6.0322131483044575e-05} +{"train_loss": 0.014242719858884811, "grad_norm": 0.13273486495018005, "global_step": 22992, "epoch": 216, "lr": 6.031923154162506e-05} +{"train_loss": 0.014720780774950981, "grad_norm": 0.1102442815899849, "global_step": 22993, "epoch": 216, "lr": 6.031633156394825e-05} +{"train_loss": 0.016692381352186203, "grad_norm": 0.1979355663061142, "global_step": 22994, "epoch": 216, "lr": 6.031343155002436e-05} +{"train_loss": 0.015782758593559265, "grad_norm": 0.06845494359731674, "global_step": 22995, "epoch": 216, "lr": 6.031053149986354e-05} +{"train_loss": 0.017904117703437805, "grad_norm": 0.11551762372255325, "global_step": 22996, "epoch": 216, "lr": 6.0307631413476e-05} +{"train_loss": 0.015088794752955437, "grad_norm": 0.059188369661569595, "global_step": 22997, "epoch": 216, "lr": 6.030473129087193e-05} +{"train_loss": 0.0163421630859375, "grad_norm": 0.11746638268232346, "global_step": 22998, "epoch": 216, "lr": 6.0301831132061494e-05} +{"train_loss": 0.016360826790332794, "grad_norm": 0.06761063635349274, "global_step": 22999, "epoch": 216, "lr": 6.029893093705492e-05} +{"train_loss": 0.014603940770030022, "grad_norm": 0.05965970829129219, "global_step": 23000, "epoch": 216, "lr": 6.0296030705862386e-05} +{"train_loss": 0.01585600909970279, "grad_norm": 0.07650642842054367, "global_step": 23001, "epoch": 216, "lr": 6.029313043849407e-05, "val_loss": 0.01959189958870411} +{"train_loss": 0.01749582402408123, "grad_norm": 0.07128646969795227, "global_step": 23002, "epoch": 217, "lr": 6.029023013496018e-05} +{"train_loss": 0.016643578186631203, "grad_norm": 0.05387938767671585, "global_step": 23003, "epoch": 217, "lr": 6.0287329795270886e-05} +{"train_loss": 0.015842441469430923, "grad_norm": 0.06697103381156921, "global_step": 23004, "epoch": 217, "lr": 6.028442941943639e-05} +{"train_loss": 0.015020505525171757, "grad_norm": 0.0895896703004837, "global_step": 23005, "epoch": 217, "lr": 6.028152900746689e-05} +{"train_loss": 0.015657037496566772, "grad_norm": 0.10282573848962784, "global_step": 23006, "epoch": 217, "lr": 6.027862855937256e-05} +{"train_loss": 0.014752434566617012, "grad_norm": 0.0769796371459961, "global_step": 23007, "epoch": 217, "lr": 6.0275728075163596e-05} +{"train_loss": 0.01374349370598793, "grad_norm": 0.06932953000068665, "global_step": 23008, "epoch": 217, "lr": 6.027282755485019e-05} +{"train_loss": 0.015709860250353813, "grad_norm": 0.06890720129013062, "global_step": 23009, "epoch": 217, "lr": 6.026992699844256e-05} +{"train_loss": 0.014446517452597618, "grad_norm": 0.0980161726474762, "global_step": 23010, "epoch": 217, "lr": 6.026702640595084e-05} +{"train_loss": 0.0162215419113636, "grad_norm": 0.1276974081993103, "global_step": 23011, "epoch": 217, "lr": 6.026412577738527e-05} +{"train_loss": 0.015061937272548676, "grad_norm": 0.05341026932001114, "global_step": 23012, "epoch": 217, "lr": 6.026122511275604e-05} +{"train_loss": 0.015035782940685749, "grad_norm": 0.05883289873600006, "global_step": 23013, "epoch": 217, "lr": 6.025832441207331e-05} +{"train_loss": 0.015127111226320267, "grad_norm": 0.07651907205581665, "global_step": 23014, "epoch": 217, "lr": 6.0255423675347286e-05} +{"train_loss": 0.014090200886130333, "grad_norm": 0.13354326784610748, "global_step": 23015, "epoch": 217, "lr": 6.025252290258817e-05} +{"train_loss": 0.015260230749845505, "grad_norm": 0.04760459065437317, "global_step": 23016, "epoch": 217, "lr": 6.024962209380615e-05} +{"train_loss": 0.017041975632309914, "grad_norm": 0.10218080133199692, "global_step": 23017, "epoch": 217, "lr": 6.024672124901141e-05} +{"train_loss": 0.016540568321943283, "grad_norm": 0.06527291983366013, "global_step": 23018, "epoch": 217, "lr": 6.0243820368214155e-05} +{"train_loss": 0.014474228024482727, "grad_norm": 0.08093325048685074, "global_step": 23019, "epoch": 217, "lr": 6.024091945142456e-05} +{"train_loss": 0.013453938066959381, "grad_norm": 0.0909554660320282, "global_step": 23020, "epoch": 217, "lr": 6.023801849865283e-05} +{"train_loss": 0.01485179178416729, "grad_norm": 0.0590764582157135, "global_step": 23021, "epoch": 217, "lr": 6.023511750990917e-05} +{"train_loss": 0.01710427738726139, "grad_norm": 0.06150476634502411, "global_step": 23022, "epoch": 217, "lr": 6.023221648520373e-05} +{"train_loss": 0.0166263859719038, "grad_norm": 0.08446324616670609, "global_step": 23023, "epoch": 217, "lr": 6.022931542454675e-05} +{"train_loss": 0.016206277534365654, "grad_norm": 0.08602142333984375, "global_step": 23024, "epoch": 217, "lr": 6.022641432794841e-05} +{"train_loss": 0.018392274156212807, "grad_norm": 0.15244784951210022, "global_step": 23025, "epoch": 217, "lr": 6.022351319541888e-05} +{"train_loss": 0.01502402313053608, "grad_norm": 0.05200482904911041, "global_step": 23026, "epoch": 217, "lr": 6.022061202696838e-05} +{"train_loss": 0.014645129442214966, "grad_norm": 0.09299635142087936, "global_step": 23027, "epoch": 217, "lr": 6.02177108226071e-05} +{"train_loss": 0.01382466871291399, "grad_norm": 0.08533396571874619, "global_step": 23028, "epoch": 217, "lr": 6.0214809582345224e-05} +{"train_loss": 0.012561121955513954, "grad_norm": 0.08894811570644379, "global_step": 23029, "epoch": 217, "lr": 6.0211908306192945e-05} +{"train_loss": 0.017528286203742027, "grad_norm": 0.06436283141374588, "global_step": 23030, "epoch": 217, "lr": 6.0209006994160465e-05} +{"train_loss": 0.01703590340912342, "grad_norm": 0.08929053694009781, "global_step": 23031, "epoch": 217, "lr": 6.020610564625797e-05} +{"train_loss": 0.01626010611653328, "grad_norm": 0.05772397667169571, "global_step": 23032, "epoch": 217, "lr": 6.020320426249565e-05} +{"train_loss": 0.015911301597952843, "grad_norm": 0.06481673568487167, "global_step": 23033, "epoch": 217, "lr": 6.020030284288374e-05} +{"train_loss": 0.014450574293732643, "grad_norm": 0.06440910696983337, "global_step": 23034, "epoch": 217, "lr": 6.019740138743237e-05} +{"train_loss": 0.015682632103562355, "grad_norm": 0.08522647619247437, "global_step": 23035, "epoch": 217, "lr": 6.0194499896151776e-05} +{"train_loss": 0.015577740035951138, "grad_norm": 0.07146057486534119, "global_step": 23036, "epoch": 217, "lr": 6.019159836905215e-05} +{"train_loss": 0.015274973586201668, "grad_norm": 0.06452832370996475, "global_step": 23037, "epoch": 217, "lr": 6.018869680614366e-05} +{"train_loss": 0.014868424274027348, "grad_norm": 0.08790639787912369, "global_step": 23038, "epoch": 217, "lr": 6.018579520743655e-05} +{"train_loss": 0.014295145869255066, "grad_norm": 0.11208731681108475, "global_step": 23039, "epoch": 217, "lr": 6.0182893572940956e-05} +{"train_loss": 0.014715177938342094, "grad_norm": 0.07098067551851273, "global_step": 23040, "epoch": 217, "lr": 6.017999190266711e-05} +{"train_loss": 0.016167081892490387, "grad_norm": 0.05400553345680237, "global_step": 23041, "epoch": 217, "lr": 6.017709019662522e-05} +{"train_loss": 0.015413734130561352, "grad_norm": 0.07352443039417267, "global_step": 23042, "epoch": 217, "lr": 6.017418845482543e-05} +{"train_loss": 0.015022581443190575, "grad_norm": 0.10201042890548706, "global_step": 23043, "epoch": 217, "lr": 6.017128667727798e-05} +{"train_loss": 0.01655019260942936, "grad_norm": 0.08187651634216309, "global_step": 23044, "epoch": 217, "lr": 6.016838486399307e-05} +{"train_loss": 0.015437820926308632, "grad_norm": 0.11985871940851212, "global_step": 23045, "epoch": 217, "lr": 6.016548301498086e-05} +{"train_loss": 0.02036529779434204, "grad_norm": 0.18417201936244965, "global_step": 23046, "epoch": 217, "lr": 6.0162581130251563e-05} +{"train_loss": 0.01711188815534115, "grad_norm": 0.14734883606433868, "global_step": 23047, "epoch": 217, "lr": 6.0159679209815376e-05} +{"train_loss": 0.015824316069483757, "grad_norm": 0.06754953414201736, "global_step": 23048, "epoch": 217, "lr": 6.0156777253682495e-05} +{"train_loss": 0.015398524701595306, "grad_norm": 0.07616391777992249, "global_step": 23049, "epoch": 217, "lr": 6.0153875261863114e-05} +{"train_loss": 0.016191639006137848, "grad_norm": 0.11399365216493607, "global_step": 23050, "epoch": 217, "lr": 6.015097323436745e-05} +{"train_loss": 0.014725632965564728, "grad_norm": 0.10144631564617157, "global_step": 23051, "epoch": 217, "lr": 6.014807117120564e-05} +{"train_loss": 0.016120223328471184, "grad_norm": 0.07561779022216797, "global_step": 23052, "epoch": 217, "lr": 6.0145169072387954e-05} +{"train_loss": 0.018551915884017944, "grad_norm": 0.12409640848636627, "global_step": 23053, "epoch": 217, "lr": 6.0142266937924565e-05} +{"train_loss": 0.015930965542793274, "grad_norm": 0.0795898362994194, "global_step": 23054, "epoch": 217, "lr": 6.0139364767825626e-05} +{"train_loss": 0.017648743465542793, "grad_norm": 0.06397701799869537, "global_step": 23055, "epoch": 217, "lr": 6.013646256210139e-05} +{"train_loss": 0.014721368439495564, "grad_norm": 0.07741601765155792, "global_step": 23056, "epoch": 217, "lr": 6.0133560320762026e-05} +{"train_loss": 0.01443365029990673, "grad_norm": 0.10586750507354736, "global_step": 23057, "epoch": 217, "lr": 6.013065804381774e-05} +{"train_loss": 0.016006607562303543, "grad_norm": 0.055587269365787506, "global_step": 23058, "epoch": 217, "lr": 6.012775573127873e-05} +{"train_loss": 0.017629653215408325, "grad_norm": 0.05753711983561516, "global_step": 23059, "epoch": 217, "lr": 6.012485338315519e-05} +{"train_loss": 0.014465875923633575, "grad_norm": 0.10558056831359863, "global_step": 23060, "epoch": 217, "lr": 6.012195099945732e-05} +{"train_loss": 0.015321414917707443, "grad_norm": 0.07844436168670654, "global_step": 23061, "epoch": 217, "lr": 6.0119048580195305e-05} +{"train_loss": 0.015769174322485924, "grad_norm": 0.08053530752658844, "global_step": 23062, "epoch": 217, "lr": 6.0116146125379366e-05} +{"train_loss": 0.01600196585059166, "grad_norm": 0.08365144580602646, "global_step": 23063, "epoch": 217, "lr": 6.011324363501969e-05} +{"train_loss": 0.01761973462998867, "grad_norm": 0.13218361139297485, "global_step": 23064, "epoch": 217, "lr": 6.0110341109126465e-05} +{"train_loss": 0.015592368319630623, "grad_norm": 0.07600822299718857, "global_step": 23065, "epoch": 217, "lr": 6.010743854770992e-05} +{"train_loss": 0.01655070297420025, "grad_norm": 0.06811904907226562, "global_step": 23066, "epoch": 217, "lr": 6.0104535950780195e-05} +{"train_loss": 0.0130621287971735, "grad_norm": 0.1458464413881302, "global_step": 23067, "epoch": 217, "lr": 6.010163331834755e-05} +{"train_loss": 0.015358742326498032, "grad_norm": 0.07537338882684708, "global_step": 23068, "epoch": 217, "lr": 6.009873065042215e-05} +{"train_loss": 0.01518076378852129, "grad_norm": 0.10760114341974258, "global_step": 23069, "epoch": 217, "lr": 6.0095827947014204e-05} +{"train_loss": 0.017426399514079094, "grad_norm": 0.08287624269723892, "global_step": 23070, "epoch": 217, "lr": 6.009292520813392e-05} +{"train_loss": 0.01528589241206646, "grad_norm": 0.07144138216972351, "global_step": 23071, "epoch": 217, "lr": 6.0090022433791474e-05} +{"train_loss": 0.014554079622030258, "grad_norm": 0.0950353741645813, "global_step": 23072, "epoch": 217, "lr": 6.008711962399708e-05} +{"train_loss": 0.013805626891553402, "grad_norm": 0.07335788756608963, "global_step": 23073, "epoch": 217, "lr": 6.0084216778760926e-05} +{"train_loss": 0.017725318670272827, "grad_norm": 0.10307872295379639, "global_step": 23074, "epoch": 217, "lr": 6.008131389809324e-05} +{"train_loss": 0.015784673392772675, "grad_norm": 0.18835946917533875, "global_step": 23075, "epoch": 217, "lr": 6.007841098200418e-05} +{"train_loss": 0.015178748406469822, "grad_norm": 0.050682950764894485, "global_step": 23076, "epoch": 217, "lr": 6.007550803050397e-05} +{"train_loss": 0.016778185963630676, "grad_norm": 0.10922826826572418, "global_step": 23077, "epoch": 217, "lr": 6.007260504360283e-05} +{"train_loss": 0.015603317879140377, "grad_norm": 0.09338221698999405, "global_step": 23078, "epoch": 217, "lr": 6.0069702021310906e-05} +{"train_loss": 0.014677152037620544, "grad_norm": 0.08229538053274155, "global_step": 23079, "epoch": 217, "lr": 6.006679896363844e-05} +{"train_loss": 0.018072279170155525, "grad_norm": 0.07345863431692123, "global_step": 23080, "epoch": 217, "lr": 6.006389587059562e-05} +{"train_loss": 0.014995165169239044, "grad_norm": 0.09401150792837143, "global_step": 23081, "epoch": 217, "lr": 6.0060992742192655e-05} +{"train_loss": 0.018147123977541924, "grad_norm": 0.06325140595436096, "global_step": 23082, "epoch": 217, "lr": 6.0058089578439725e-05} +{"train_loss": 0.01647745817899704, "grad_norm": 0.08635060489177704, "global_step": 23083, "epoch": 217, "lr": 6.005518637934705e-05} +{"train_loss": 0.017868854105472565, "grad_norm": 0.06489122658967972, "global_step": 23084, "epoch": 217, "lr": 6.0052283144924824e-05} +{"train_loss": 0.016787748783826828, "grad_norm": 0.1430659145116806, "global_step": 23085, "epoch": 217, "lr": 6.0049379875183244e-05} +{"train_loss": 0.016323670744895935, "grad_norm": 0.08483406156301498, "global_step": 23086, "epoch": 217, "lr": 6.004647657013251e-05} +{"train_loss": 0.016265185549855232, "grad_norm": 0.0747479647397995, "global_step": 23087, "epoch": 217, "lr": 6.0043573229782835e-05} +{"train_loss": 0.0181281678378582, "grad_norm": 0.07522618770599365, "global_step": 23088, "epoch": 217, "lr": 6.00406698541444e-05} +{"train_loss": 0.014862732961773872, "grad_norm": 0.07684924453496933, "global_step": 23089, "epoch": 217, "lr": 6.003776644322743e-05} +{"train_loss": 0.01830637827515602, "grad_norm": 0.08558741956949234, "global_step": 23090, "epoch": 217, "lr": 6.0034862997042095e-05} +{"train_loss": 0.01895606704056263, "grad_norm": 0.080691397190094, "global_step": 23091, "epoch": 217, "lr": 6.003195951559864e-05} +{"train_loss": 0.01769394986331463, "grad_norm": 0.11264591664075851, "global_step": 23092, "epoch": 217, "lr": 6.002905599890722e-05} +{"train_loss": 0.016312820836901665, "grad_norm": 0.09928002953529358, "global_step": 23093, "epoch": 217, "lr": 6.0026152446978067e-05} +{"train_loss": 0.013748150318861008, "grad_norm": 0.08787280321121216, "global_step": 23094, "epoch": 217, "lr": 6.002324885982139e-05} +{"train_loss": 0.01570160873234272, "grad_norm": 0.10707782953977585, "global_step": 23095, "epoch": 217, "lr": 6.002034523744735e-05} +{"train_loss": 0.013686694204807281, "grad_norm": 0.05386923626065254, "global_step": 23096, "epoch": 217, "lr": 6.0017441579866184e-05} +{"train_loss": 0.016539039090275764, "grad_norm": 0.07957685738801956, "global_step": 23097, "epoch": 217, "lr": 6.0014537887088086e-05} +{"train_loss": 0.016056500375270844, "grad_norm": 0.08770597726106644, "global_step": 23098, "epoch": 217, "lr": 6.0011634159123263e-05} +{"train_loss": 0.016802627593278885, "grad_norm": 0.08412443846464157, "global_step": 23099, "epoch": 217, "lr": 6.00087303959819e-05} +{"train_loss": 0.014391189441084862, "grad_norm": 0.07895954698324203, "global_step": 23100, "epoch": 217, "lr": 6.000582659767422e-05} +{"train_loss": 0.012555706314742565, "grad_norm": 0.054937005043029785, "global_step": 23101, "epoch": 217, "lr": 6.000292276421041e-05} +{"train_loss": 0.015313628129661083, "grad_norm": 0.07501857727766037, "global_step": 23102, "epoch": 217, "lr": 6.000001889560068e-05} +{"train_loss": 0.017229700461030006, "grad_norm": 0.15285494923591614, "global_step": 23103, "epoch": 217, "lr": 5.999711499185525e-05} +{"train_loss": 0.015022083185613155, "grad_norm": 0.07594608515501022, "global_step": 23104, "epoch": 217, "lr": 5.9994211052984295e-05} +{"train_loss": 0.015416976064443588, "grad_norm": 0.07835983484983444, "global_step": 23105, "epoch": 217, "lr": 5.999130707899801e-05} +{"train_loss": 0.01653713546693325, "grad_norm": 0.07260588556528091, "global_step": 23106, "epoch": 217, "lr": 5.998840306990665e-05} +{"train_loss": 0.01583827025134046, "grad_norm": 0.1300419420003891, "global_step": 23107, "epoch": 217, "lr": 5.9985499025720346e-05, "val_loss": 0.01967255026102066} +{"train_loss": 0.016991659998893738, "grad_norm": 0.10814594477415085, "global_step": 23108, "epoch": 218, "lr": 5.998259494644937e-05} +{"train_loss": 0.017654728144407272, "grad_norm": 0.10377027839422226, "global_step": 23109, "epoch": 218, "lr": 5.997969083210389e-05} +{"train_loss": 0.016785988584160805, "grad_norm": 0.10411734133958817, "global_step": 23110, "epoch": 218, "lr": 5.997678668269412e-05} +{"train_loss": 0.016112126410007477, "grad_norm": 0.09151162952184677, "global_step": 23111, "epoch": 218, "lr": 5.997388249823026e-05} +{"train_loss": 0.018270257860422134, "grad_norm": 0.09972614049911499, "global_step": 23112, "epoch": 218, "lr": 5.997097827872251e-05} +{"train_loss": 0.01668677292764187, "grad_norm": 0.07494430989027023, "global_step": 23113, "epoch": 218, "lr": 5.996807402418108e-05} +{"train_loss": 0.015743393450975418, "grad_norm": 0.09364204108715057, "global_step": 23114, "epoch": 218, "lr": 5.9965169734616166e-05} +{"train_loss": 0.015147922560572624, "grad_norm": 0.05422314628958702, "global_step": 23115, "epoch": 218, "lr": 5.996226541003799e-05} +{"train_loss": 0.018707700073719025, "grad_norm": 0.08963147550821304, "global_step": 23116, "epoch": 218, "lr": 5.9959361050456744e-05} +{"train_loss": 0.01734434999525547, "grad_norm": 0.14593388140201569, "global_step": 23117, "epoch": 218, "lr": 5.995645665588262e-05} +{"train_loss": 0.01575440540909767, "grad_norm": 0.08385607600212097, "global_step": 23118, "epoch": 218, "lr": 5.995355222632586e-05} +{"train_loss": 0.013445756398141384, "grad_norm": 0.05309002846479416, "global_step": 23119, "epoch": 218, "lr": 5.995064776179663e-05} +{"train_loss": 0.015098247677087784, "grad_norm": 0.11674542725086212, "global_step": 23120, "epoch": 218, "lr": 5.994774326230516e-05} +{"train_loss": 0.017543761059641838, "grad_norm": 0.09858250617980957, "global_step": 23121, "epoch": 218, "lr": 5.994483872786165e-05} +{"train_loss": 0.015301126055419445, "grad_norm": 0.09610287845134735, "global_step": 23122, "epoch": 218, "lr": 5.994193415847629e-05} +{"train_loss": 0.01658153533935547, "grad_norm": 0.18502821028232574, "global_step": 23123, "epoch": 218, "lr": 5.99390295541593e-05} +{"train_loss": 0.015720853582024574, "grad_norm": 0.11525873839855194, "global_step": 23124, "epoch": 218, "lr": 5.993612491492088e-05} +{"train_loss": 0.016665250062942505, "grad_norm": 0.07011733204126358, "global_step": 23125, "epoch": 218, "lr": 5.9933220240771235e-05} +{"train_loss": 0.018919430673122406, "grad_norm": 0.11925407499074936, "global_step": 23126, "epoch": 218, "lr": 5.9930315531720595e-05} +{"train_loss": 0.013876767829060555, "grad_norm": 0.07596466690301895, "global_step": 23127, "epoch": 218, "lr": 5.992741078777913e-05} +{"train_loss": 0.016822446137666702, "grad_norm": 0.16125723719596863, "global_step": 23128, "epoch": 218, "lr": 5.992450600895706e-05} +{"train_loss": 0.01487260963767767, "grad_norm": 0.07463328540325165, "global_step": 23129, "epoch": 218, "lr": 5.992160119526459e-05} +{"train_loss": 0.015504308976233006, "grad_norm": 0.093343086540699, "global_step": 23130, "epoch": 218, "lr": 5.991869634671193e-05} +{"train_loss": 0.016151707619428635, "grad_norm": 0.07236691564321518, "global_step": 23131, "epoch": 218, "lr": 5.991579146330928e-05} +{"train_loss": 0.019992534071207047, "grad_norm": 0.08121390640735626, "global_step": 23132, "epoch": 218, "lr": 5.991288654506686e-05} +{"train_loss": 0.015951551496982574, "grad_norm": 0.09793706238269806, "global_step": 23133, "epoch": 218, "lr": 5.990998159199488e-05} +{"train_loss": 0.013629947789013386, "grad_norm": 0.06797707825899124, "global_step": 23134, "epoch": 218, "lr": 5.990707660410351e-05} +{"train_loss": 0.016510963439941406, "grad_norm": 0.07430986315011978, "global_step": 23135, "epoch": 218, "lr": 5.990417158140299e-05} +{"train_loss": 0.014764189720153809, "grad_norm": 0.08686722069978714, "global_step": 23136, "epoch": 218, "lr": 5.990126652390352e-05} +{"train_loss": 0.012624964118003845, "grad_norm": 0.07813575118780136, "global_step": 23137, "epoch": 218, "lr": 5.9898361431615304e-05} +{"train_loss": 0.01632286235690117, "grad_norm": 0.09300758689641953, "global_step": 23138, "epoch": 218, "lr": 5.989545630454855e-05} +{"train_loss": 0.01856805570423603, "grad_norm": 0.1394391655921936, "global_step": 23139, "epoch": 218, "lr": 5.9892551142713473e-05} +{"train_loss": 0.01624235138297081, "grad_norm": 0.06336914002895355, "global_step": 23140, "epoch": 218, "lr": 5.9889645946120264e-05} +{"train_loss": 0.014704408124089241, "grad_norm": 0.07865376025438309, "global_step": 23141, "epoch": 218, "lr": 5.988674071477914e-05} +{"train_loss": 0.014195379801094532, "grad_norm": 0.08267197012901306, "global_step": 23142, "epoch": 218, "lr": 5.988383544870032e-05} +{"train_loss": 0.015907123684883118, "grad_norm": 0.05908115580677986, "global_step": 23143, "epoch": 218, "lr": 5.9880930147893985e-05} +{"train_loss": 0.016628341749310493, "grad_norm": 0.06552096456289291, "global_step": 23144, "epoch": 218, "lr": 5.987802481237037e-05} +{"train_loss": 0.01599164307117462, "grad_norm": 0.07345213741064072, "global_step": 23145, "epoch": 218, "lr": 5.987511944213967e-05} +{"train_loss": 0.014657403342425823, "grad_norm": 0.07534939050674438, "global_step": 23146, "epoch": 218, "lr": 5.987221403721208e-05} +{"train_loss": 0.01611914113163948, "grad_norm": 0.06246444210410118, "global_step": 23147, "epoch": 218, "lr": 5.986930859759785e-05} +{"train_loss": 0.01780419424176216, "grad_norm": 0.10532552003860474, "global_step": 23148, "epoch": 218, "lr": 5.9866403123307135e-05} +{"train_loss": 0.015325027517974377, "grad_norm": 0.13453486561775208, "global_step": 23149, "epoch": 218, "lr": 5.98634976143502e-05} +{"train_loss": 0.016331223770976067, "grad_norm": 0.17745272815227509, "global_step": 23150, "epoch": 218, "lr": 5.98605920707372e-05} +{"train_loss": 0.014509061351418495, "grad_norm": 0.1481417566537857, "global_step": 23151, "epoch": 218, "lr": 5.985768649247837e-05} +{"train_loss": 0.014916179701685905, "grad_norm": 0.056030530482530594, "global_step": 23152, "epoch": 218, "lr": 5.985478087958393e-05} +{"train_loss": 0.017216427251696587, "grad_norm": 0.055282000452280045, "global_step": 23153, "epoch": 218, "lr": 5.9851875232064056e-05} +{"train_loss": 0.01480235904455185, "grad_norm": 0.07274219393730164, "global_step": 23154, "epoch": 218, "lr": 5.9848969549929e-05} +{"train_loss": 0.016963448375463486, "grad_norm": 0.08121850341558456, "global_step": 23155, "epoch": 218, "lr": 5.9846063833188934e-05} +{"train_loss": 0.016212312504649162, "grad_norm": 0.11420435458421707, "global_step": 23156, "epoch": 218, "lr": 5.984315808185408e-05} +{"train_loss": 0.01577489823102951, "grad_norm": 0.07663532346487045, "global_step": 23157, "epoch": 218, "lr": 5.984025229593466e-05} +{"train_loss": 0.017495250329375267, "grad_norm": 0.08429213613271713, "global_step": 23158, "epoch": 218, "lr": 5.983734647544086e-05} +{"train_loss": 0.01747213490307331, "grad_norm": 0.07691638916730881, "global_step": 23159, "epoch": 218, "lr": 5.9834440620382916e-05} +{"train_loss": 0.01305006816983223, "grad_norm": 0.06916801631450653, "global_step": 23160, "epoch": 218, "lr": 5.9831534730771e-05} +{"train_loss": 0.016612786799669266, "grad_norm": 0.12177388370037079, "global_step": 23161, "epoch": 218, "lr": 5.982862880661537e-05} +{"train_loss": 0.014770936220884323, "grad_norm": 0.09279268980026245, "global_step": 23162, "epoch": 218, "lr": 5.9825722847926214e-05} +{"train_loss": 0.016020793467760086, "grad_norm": 0.16038846969604492, "global_step": 23163, "epoch": 218, "lr": 5.982281685471372e-05} +{"train_loss": 0.017076464369893074, "grad_norm": 0.05589614808559418, "global_step": 23164, "epoch": 218, "lr": 5.9819910826988137e-05} +{"train_loss": 0.01837954670190811, "grad_norm": 0.1145278736948967, "global_step": 23165, "epoch": 218, "lr": 5.981700476475965e-05} +{"train_loss": 0.013529328629374504, "grad_norm": 0.07749522477388382, "global_step": 23166, "epoch": 218, "lr": 5.981409866803849e-05} +{"train_loss": 0.016017362475395203, "grad_norm": 0.1957501322031021, "global_step": 23167, "epoch": 218, "lr": 5.981119253683485e-05} +{"train_loss": 0.014808512292802334, "grad_norm": 0.07103514671325684, "global_step": 23168, "epoch": 218, "lr": 5.980828637115894e-05} +{"train_loss": 0.01380198635160923, "grad_norm": 0.13720203936100006, "global_step": 23169, "epoch": 218, "lr": 5.9805380171020974e-05} +{"train_loss": 0.017370130866765976, "grad_norm": 0.10113701969385147, "global_step": 23170, "epoch": 218, "lr": 5.980247393643117e-05} +{"train_loss": 0.0159225482493639, "grad_norm": 0.06784343719482422, "global_step": 23171, "epoch": 218, "lr": 5.979956766739975e-05} +{"train_loss": 0.013947109691798687, "grad_norm": 0.08879305422306061, "global_step": 23172, "epoch": 218, "lr": 5.979666136393689e-05} +{"train_loss": 0.016857128590345383, "grad_norm": 0.07198568433523178, "global_step": 23173, "epoch": 218, "lr": 5.979375502605282e-05} +{"train_loss": 0.013990659266710281, "grad_norm": 0.08139645308256149, "global_step": 23174, "epoch": 218, "lr": 5.979084865375778e-05} +{"train_loss": 0.014911911450326443, "grad_norm": 0.11426737904548645, "global_step": 23175, "epoch": 218, "lr": 5.978794224706193e-05} +{"train_loss": 0.01512438990175724, "grad_norm": 0.08653786778450012, "global_step": 23176, "epoch": 218, "lr": 5.978503580597552e-05} +{"train_loss": 0.0199313722550869, "grad_norm": 0.0837448388338089, "global_step": 23177, "epoch": 218, "lr": 5.978212933050875e-05} +{"train_loss": 0.014061317779123783, "grad_norm": 0.05783243849873543, "global_step": 23178, "epoch": 218, "lr": 5.977922282067183e-05} +{"train_loss": 0.014811071567237377, "grad_norm": 0.05609763786196709, "global_step": 23179, "epoch": 218, "lr": 5.977631627647497e-05} +{"train_loss": 0.013643775135278702, "grad_norm": 0.0548282191157341, "global_step": 23180, "epoch": 218, "lr": 5.977340969792839e-05} +{"train_loss": 0.01683131791651249, "grad_norm": 0.0931706503033638, "global_step": 23181, "epoch": 218, "lr": 5.9770503085042306e-05} +{"train_loss": 0.014585231430828571, "grad_norm": 0.05977192893624306, "global_step": 23182, "epoch": 218, "lr": 5.9767596437826915e-05} +{"train_loss": 0.015406208112835884, "grad_norm": 0.11831158399581909, "global_step": 23183, "epoch": 218, "lr": 5.9764689756292446e-05} +{"train_loss": 0.015368498861789703, "grad_norm": 0.06268355995416641, "global_step": 23184, "epoch": 218, "lr": 5.976178304044909e-05} +{"train_loss": 0.014684014022350311, "grad_norm": 0.06548784673213959, "global_step": 23185, "epoch": 218, "lr": 5.9758876290307074e-05} +{"train_loss": 0.018152475357055664, "grad_norm": 0.06284922361373901, "global_step": 23186, "epoch": 218, "lr": 5.975596950587663e-05} +{"train_loss": 0.013494440354406834, "grad_norm": 0.05784580856561661, "global_step": 23187, "epoch": 218, "lr": 5.9753062687167925e-05} +{"train_loss": 0.012494754046201706, "grad_norm": 0.0896509438753128, "global_step": 23188, "epoch": 218, "lr": 5.975015583419122e-05} +{"train_loss": 0.013852539472281933, "grad_norm": 0.09283234924077988, "global_step": 23189, "epoch": 218, "lr": 5.97472489469567e-05} +{"train_loss": 0.014284878969192505, "grad_norm": 0.09723867475986481, "global_step": 23190, "epoch": 218, "lr": 5.974434202547459e-05} +{"train_loss": 0.014569034799933434, "grad_norm": 0.07647769898176193, "global_step": 23191, "epoch": 218, "lr": 5.97414350697551e-05} +{"train_loss": 0.016088198870420456, "grad_norm": 0.08811500668525696, "global_step": 23192, "epoch": 218, "lr": 5.9738528079808434e-05} +{"train_loss": 0.014948359690606594, "grad_norm": 0.08936340361833572, "global_step": 23193, "epoch": 218, "lr": 5.973562105564482e-05} +{"train_loss": 0.015415726229548454, "grad_norm": 0.1185150146484375, "global_step": 23194, "epoch": 218, "lr": 5.973271399727447e-05} +{"train_loss": 0.01710132509469986, "grad_norm": 0.10547370463609695, "global_step": 23195, "epoch": 218, "lr": 5.97298069047076e-05} +{"train_loss": 0.015847964212298393, "grad_norm": 0.09067650139331818, "global_step": 23196, "epoch": 218, "lr": 5.972689977795441e-05} +{"train_loss": 0.016857445240020752, "grad_norm": 0.13142217695713043, "global_step": 23197, "epoch": 218, "lr": 5.972399261702514e-05} +{"train_loss": 0.01866902783513069, "grad_norm": 0.08849938958883286, "global_step": 23198, "epoch": 218, "lr": 5.9721085421929966e-05} +{"train_loss": 0.01320379227399826, "grad_norm": 0.07363645732402802, "global_step": 23199, "epoch": 218, "lr": 5.971817819267913e-05} +{"train_loss": 0.017048310488462448, "grad_norm": 0.17671938240528107, "global_step": 23200, "epoch": 218, "lr": 5.971527092928286e-05} +{"train_loss": 0.016901619732379913, "grad_norm": 0.09000556170940399, "global_step": 23201, "epoch": 218, "lr": 5.971236363175133e-05} +{"train_loss": 0.014263525605201721, "grad_norm": 0.07306893169879913, "global_step": 23202, "epoch": 218, "lr": 5.97094563000948e-05} +{"train_loss": 0.01816370338201523, "grad_norm": 0.06495556980371475, "global_step": 23203, "epoch": 218, "lr": 5.970654893432346e-05} +{"train_loss": 0.015655748546123505, "grad_norm": 0.058810919523239136, "global_step": 23204, "epoch": 218, "lr": 5.970364153444752e-05} +{"train_loss": 0.013422507792711258, "grad_norm": 0.05066102370619774, "global_step": 23205, "epoch": 218, "lr": 5.9700734100477205e-05} +{"train_loss": 0.017216075211763382, "grad_norm": 0.1246342882514, "global_step": 23206, "epoch": 218, "lr": 5.9697826632422724e-05} +{"train_loss": 0.015247195959091187, "grad_norm": 0.07555962353944778, "global_step": 23207, "epoch": 218, "lr": 5.96949191302943e-05} +{"train_loss": 0.015818146988749504, "grad_norm": 0.06022302061319351, "global_step": 23208, "epoch": 218, "lr": 5.969201159410216e-05} +{"train_loss": 0.014660279266536236, "grad_norm": 0.08368310332298279, "global_step": 23209, "epoch": 218, "lr": 5.968910402385649e-05} +{"train_loss": 0.01485847681760788, "grad_norm": 0.07829807698726654, "global_step": 23210, "epoch": 218, "lr": 5.968619641956753e-05} +{"train_loss": 0.016578728333115578, "grad_norm": 0.10097953677177429, "global_step": 23211, "epoch": 218, "lr": 5.968328878124548e-05} +{"train_loss": 0.014924447983503342, "grad_norm": 0.05889544636011124, "global_step": 23212, "epoch": 218, "lr": 5.968038110890059e-05} +{"train_loss": 0.0158209326624308, "grad_norm": 0.06646173447370529, "global_step": 23213, "epoch": 218, "lr": 5.967747340254303e-05, "val_loss": 0.019773025065660477} +{"train_loss": 0.014283474534749985, "grad_norm": 0.12865784764289856, "global_step": 23214, "epoch": 219, "lr": 5.9674565662183027e-05} +{"train_loss": 0.015081670135259628, "grad_norm": 0.13211333751678467, "global_step": 23215, "epoch": 219, "lr": 5.967165788783082e-05} +{"train_loss": 0.01637277379631996, "grad_norm": 0.06216401606798172, "global_step": 23216, "epoch": 219, "lr": 5.966875007949661e-05} +{"train_loss": 0.017708254978060722, "grad_norm": 0.05750800669193268, "global_step": 23217, "epoch": 219, "lr": 5.966584223719062e-05} +{"train_loss": 0.014451985247433186, "grad_norm": 0.10853613168001175, "global_step": 23218, "epoch": 219, "lr": 5.9662934360923064e-05} +{"train_loss": 0.012652253732085228, "grad_norm": 0.1091822162270546, "global_step": 23219, "epoch": 219, "lr": 5.966002645070416e-05} +{"train_loss": 0.017566075548529625, "grad_norm": 0.11424542218446732, "global_step": 23220, "epoch": 219, "lr": 5.965711850654412e-05} +{"train_loss": 0.01444261521100998, "grad_norm": 0.07018058747053146, "global_step": 23221, "epoch": 219, "lr": 5.965421052845317e-05} +{"train_loss": 0.015318714082241058, "grad_norm": 0.06962354481220245, "global_step": 23222, "epoch": 219, "lr": 5.965130251644152e-05} +{"train_loss": 0.017260290682315826, "grad_norm": 0.11963605880737305, "global_step": 23223, "epoch": 219, "lr": 5.964839447051939e-05} +{"train_loss": 0.013587510213255882, "grad_norm": 0.09776996076107025, "global_step": 23224, "epoch": 219, "lr": 5.9645486390697e-05} +{"train_loss": 0.015087776817381382, "grad_norm": 0.07689030468463898, "global_step": 23225, "epoch": 219, "lr": 5.964257827698456e-05} +{"train_loss": 0.01377357542514801, "grad_norm": 0.08249539881944656, "global_step": 23226, "epoch": 219, "lr": 5.963967012939229e-05} +{"train_loss": 0.01713627576828003, "grad_norm": 0.06958632916212082, "global_step": 23227, "epoch": 219, "lr": 5.963676194793043e-05} +{"train_loss": 0.01598740555346012, "grad_norm": 0.09406625479459763, "global_step": 23228, "epoch": 219, "lr": 5.963385373260916e-05} +{"train_loss": 0.017438340932130814, "grad_norm": 0.06037105992436409, "global_step": 23229, "epoch": 219, "lr": 5.963094548343873e-05} +{"train_loss": 0.01618339493870735, "grad_norm": 0.07979096472263336, "global_step": 23230, "epoch": 219, "lr": 5.962803720042934e-05} +{"train_loss": 0.01637757569551468, "grad_norm": 0.07945352047681808, "global_step": 23231, "epoch": 219, "lr": 5.962512888359122e-05} +{"train_loss": 0.01697150431573391, "grad_norm": 0.10974536091089249, "global_step": 23232, "epoch": 219, "lr": 5.962222053293458e-05} +{"train_loss": 0.016388840973377228, "grad_norm": 0.08076480776071548, "global_step": 23233, "epoch": 219, "lr": 5.961931214846964e-05} +{"train_loss": 0.014537614770233631, "grad_norm": 0.08205827325582504, "global_step": 23234, "epoch": 219, "lr": 5.9616403730206615e-05} +{"train_loss": 0.013156631030142307, "grad_norm": 0.11566387861967087, "global_step": 23235, "epoch": 219, "lr": 5.961349527815574e-05} +{"train_loss": 0.015774324536323547, "grad_norm": 0.14180512726306915, "global_step": 23236, "epoch": 219, "lr": 5.961058679232722e-05} +{"train_loss": 0.013981284573674202, "grad_norm": 0.088582843542099, "global_step": 23237, "epoch": 219, "lr": 5.960767827273128e-05} +{"train_loss": 0.01541857235133648, "grad_norm": 0.06768403947353363, "global_step": 23238, "epoch": 219, "lr": 5.960476971937813e-05} +{"train_loss": 0.018365971744060516, "grad_norm": 0.1437310129404068, "global_step": 23239, "epoch": 219, "lr": 5.9601861132278e-05} +{"train_loss": 0.016327179968357086, "grad_norm": 0.09788306802511215, "global_step": 23240, "epoch": 219, "lr": 5.9598952511441094e-05} +{"train_loss": 0.016694722697138786, "grad_norm": 0.10947110503911972, "global_step": 23241, "epoch": 219, "lr": 5.959604385687767e-05} +{"train_loss": 0.01779622584581375, "grad_norm": 0.09467671811580658, "global_step": 23242, "epoch": 219, "lr": 5.95931351685979e-05} +{"train_loss": 0.014857442118227482, "grad_norm": 0.09749879688024521, "global_step": 23243, "epoch": 219, "lr": 5.959022644661203e-05} +{"train_loss": 0.016902830451726913, "grad_norm": 0.1272462010383606, "global_step": 23244, "epoch": 219, "lr": 5.9587317690930275e-05} +{"train_loss": 0.017937079071998596, "grad_norm": 0.1029186025261879, "global_step": 23245, "epoch": 219, "lr": 5.958440890156285e-05} +{"train_loss": 0.017359629273414612, "grad_norm": 0.152426615357399, "global_step": 23246, "epoch": 219, "lr": 5.9581500078519994e-05} +{"train_loss": 0.01593780517578125, "grad_norm": 0.054072435945272446, "global_step": 23247, "epoch": 219, "lr": 5.9578591221811896e-05} +{"train_loss": 0.01766214519739151, "grad_norm": 0.13612620532512665, "global_step": 23248, "epoch": 219, "lr": 5.9575682331448814e-05} +{"train_loss": 0.013191581703722477, "grad_norm": 0.13696564733982086, "global_step": 23249, "epoch": 219, "lr": 5.957277340744094e-05} +{"train_loss": 0.01533629558980465, "grad_norm": 0.07539558410644531, "global_step": 23250, "epoch": 219, "lr": 5.95698644497985e-05} +{"train_loss": 0.015465985052287579, "grad_norm": 0.06777019053697586, "global_step": 23251, "epoch": 219, "lr": 5.9566955458531716e-05} +{"train_loss": 0.015608994290232658, "grad_norm": 0.09086579829454422, "global_step": 23252, "epoch": 219, "lr": 5.956404643365081e-05} +{"train_loss": 0.01658310554921627, "grad_norm": 0.08172911405563354, "global_step": 23253, "epoch": 219, "lr": 5.956113737516602e-05} +{"train_loss": 0.013902872800827026, "grad_norm": 0.12232622504234314, "global_step": 23254, "epoch": 219, "lr": 5.955822828308754e-05} +{"train_loss": 0.015253630466759205, "grad_norm": 0.09630575031042099, "global_step": 23255, "epoch": 219, "lr": 5.9555319157425594e-05} +{"train_loss": 0.016614455729722977, "grad_norm": 0.05522792413830757, "global_step": 23256, "epoch": 219, "lr": 5.955240999819043e-05} +{"train_loss": 0.013992106541991234, "grad_norm": 0.08822514861822128, "global_step": 23257, "epoch": 219, "lr": 5.9549500805392235e-05} +{"train_loss": 0.014869175851345062, "grad_norm": 0.09451171010732651, "global_step": 23258, "epoch": 219, "lr": 5.9546591579041256e-05} +{"train_loss": 0.014769194647669792, "grad_norm": 0.12035724520683289, "global_step": 23259, "epoch": 219, "lr": 5.95436823191477e-05} +{"train_loss": 0.01783042401075363, "grad_norm": 0.16066686809062958, "global_step": 23260, "epoch": 219, "lr": 5.95407730257218e-05} +{"train_loss": 0.015035421587526798, "grad_norm": 0.05967264994978905, "global_step": 23261, "epoch": 219, "lr": 5.9537863698773765e-05} +{"train_loss": 0.012238595634698868, "grad_norm": 0.06722687929868698, "global_step": 23262, "epoch": 219, "lr": 5.953495433831384e-05} +{"train_loss": 0.016399912536144257, "grad_norm": 0.059347186237573624, "global_step": 23263, "epoch": 219, "lr": 5.953204494435223e-05} +{"train_loss": 0.01354664284735918, "grad_norm": 0.0789601057767868, "global_step": 23264, "epoch": 219, "lr": 5.952913551689915e-05} +{"train_loss": 0.014495913870632648, "grad_norm": 0.05795305222272873, "global_step": 23265, "epoch": 219, "lr": 5.9526226055964826e-05} +{"train_loss": 0.01730206608772278, "grad_norm": 0.060616523027420044, "global_step": 23266, "epoch": 219, "lr": 5.9523316561559503e-05} +{"train_loss": 0.01644996367394924, "grad_norm": 0.10048482567071915, "global_step": 23267, "epoch": 219, "lr": 5.952040703369337e-05} +{"train_loss": 0.015085854567587376, "grad_norm": 0.05123687535524368, "global_step": 23268, "epoch": 219, "lr": 5.9517497472376694e-05} +{"train_loss": 0.01775534451007843, "grad_norm": 0.08284371346235275, "global_step": 23269, "epoch": 219, "lr": 5.951458787761963e-05} +{"train_loss": 0.016761863604187965, "grad_norm": 0.06649919599294662, "global_step": 23270, "epoch": 219, "lr": 5.951167824943248e-05} +{"train_loss": 0.015731342136859894, "grad_norm": 0.10010667145252228, "global_step": 23271, "epoch": 219, "lr": 5.950876858782542e-05} +{"train_loss": 0.01598387211561203, "grad_norm": 0.15834251046180725, "global_step": 23272, "epoch": 219, "lr": 5.950585889280866e-05} +{"train_loss": 0.017430096864700317, "grad_norm": 0.06318134814500809, "global_step": 23273, "epoch": 219, "lr": 5.9502949164392476e-05} +{"train_loss": 0.015949536114931107, "grad_norm": 0.06310857832431793, "global_step": 23274, "epoch": 219, "lr": 5.950003940258705e-05} +{"train_loss": 0.016169048845767975, "grad_norm": 0.1043735072016716, "global_step": 23275, "epoch": 219, "lr": 5.949712960740261e-05} +{"train_loss": 0.016366083174943924, "grad_norm": 0.07302268594503403, "global_step": 23276, "epoch": 219, "lr": 5.949421977884939e-05} +{"train_loss": 0.014663377776741982, "grad_norm": 0.09689809381961823, "global_step": 23277, "epoch": 219, "lr": 5.949130991693762e-05} +{"train_loss": 0.015508806332945824, "grad_norm": 0.06200190633535385, "global_step": 23278, "epoch": 219, "lr": 5.94884000216775e-05} +{"train_loss": 0.015153961256146431, "grad_norm": 0.08223932981491089, "global_step": 23279, "epoch": 219, "lr": 5.948549009307927e-05} +{"train_loss": 0.014846554026007652, "grad_norm": 0.06608986109495163, "global_step": 23280, "epoch": 219, "lr": 5.9482580131153176e-05} +{"train_loss": 0.01416489202529192, "grad_norm": 0.05813351273536682, "global_step": 23281, "epoch": 219, "lr": 5.9479670135909395e-05} +{"train_loss": 0.012129864655435085, "grad_norm": 0.1272084265947342, "global_step": 23282, "epoch": 219, "lr": 5.947676010735818e-05} +{"train_loss": 0.018143966794013977, "grad_norm": 0.0974234864115715, "global_step": 23283, "epoch": 219, "lr": 5.947385004550976e-05} +{"train_loss": 0.01590377278625965, "grad_norm": 0.0780012235045433, "global_step": 23284, "epoch": 219, "lr": 5.9470939950374335e-05} +{"train_loss": 0.016633816063404083, "grad_norm": 0.0890420600771904, "global_step": 23285, "epoch": 219, "lr": 5.9468029821962154e-05} +{"train_loss": 0.015624920837581158, "grad_norm": 0.0743003711104393, "global_step": 23286, "epoch": 219, "lr": 5.946511966028344e-05} +{"train_loss": 0.016767974942922592, "grad_norm": 0.14055997133255005, "global_step": 23287, "epoch": 219, "lr": 5.9462209465348415e-05} +{"train_loss": 0.015847861766815186, "grad_norm": 0.09238044172525406, "global_step": 23288, "epoch": 219, "lr": 5.9459299237167286e-05} +{"train_loss": 0.018276376649737358, "grad_norm": 0.08047722280025482, "global_step": 23289, "epoch": 219, "lr": 5.945638897575031e-05} +{"train_loss": 0.015846237540245056, "grad_norm": 0.07150080800056458, "global_step": 23290, "epoch": 219, "lr": 5.9453478681107686e-05} +{"train_loss": 0.01495822612196207, "grad_norm": 0.08862314373254776, "global_step": 23291, "epoch": 219, "lr": 5.9450568353249645e-05} +{"train_loss": 0.01626562513411045, "grad_norm": 0.07688214629888535, "global_step": 23292, "epoch": 219, "lr": 5.944765799218642e-05} +{"train_loss": 0.015137862414121628, "grad_norm": 0.06104513630270958, "global_step": 23293, "epoch": 219, "lr": 5.944474759792824e-05} +{"train_loss": 0.013403227552771568, "grad_norm": 0.05459609255194664, "global_step": 23294, "epoch": 219, "lr": 5.944183717048533e-05} +{"train_loss": 0.01653917133808136, "grad_norm": 0.1027897447347641, "global_step": 23295, "epoch": 219, "lr": 5.943892670986789e-05} +{"train_loss": 0.016049381345510483, "grad_norm": 0.08819586783647537, "global_step": 23296, "epoch": 219, "lr": 5.9436016216086164e-05} +{"train_loss": 0.015413926914334297, "grad_norm": 0.0725993663072586, "global_step": 23297, "epoch": 219, "lr": 5.9433105689150416e-05} +{"train_loss": 0.01682082563638687, "grad_norm": 0.05365943908691406, "global_step": 23298, "epoch": 219, "lr": 5.94301951290708e-05} +{"train_loss": 0.01614629104733467, "grad_norm": 0.09822043776512146, "global_step": 23299, "epoch": 219, "lr": 5.942728453585759e-05} +{"train_loss": 0.017013832926750183, "grad_norm": 0.13407352566719055, "global_step": 23300, "epoch": 219, "lr": 5.942437390952101e-05} +{"train_loss": 0.01990356296300888, "grad_norm": 0.13159166276454926, "global_step": 23301, "epoch": 219, "lr": 5.942146325007128e-05} +{"train_loss": 0.014540230855345726, "grad_norm": 0.1045660749077797, "global_step": 23302, "epoch": 219, "lr": 5.941855255751861e-05} +{"train_loss": 0.016097217798233032, "grad_norm": 0.0793725997209549, "global_step": 23303, "epoch": 219, "lr": 5.9415641831873246e-05} +{"train_loss": 0.017628369852900505, "grad_norm": 0.08278705179691315, "global_step": 23304, "epoch": 219, "lr": 5.941273107314542e-05} +{"train_loss": 0.01592247188091278, "grad_norm": 0.08510610461235046, "global_step": 23305, "epoch": 219, "lr": 5.940982028134535e-05} +{"train_loss": 0.017774831503629684, "grad_norm": 0.11024481058120728, "global_step": 23306, "epoch": 219, "lr": 5.940690945648326e-05} +{"train_loss": 0.015774428844451904, "grad_norm": 0.09614885598421097, "global_step": 23307, "epoch": 219, "lr": 5.940399859856939e-05} +{"train_loss": 0.015569246374070644, "grad_norm": 0.08394858241081238, "global_step": 23308, "epoch": 219, "lr": 5.9401087707613934e-05} +{"train_loss": 0.017347555607557297, "grad_norm": 0.06156883016228676, "global_step": 23309, "epoch": 219, "lr": 5.939817678362718e-05} +{"train_loss": 0.01802973635494709, "grad_norm": 0.07611829042434692, "global_step": 23310, "epoch": 219, "lr": 5.9395265826619285e-05} +{"train_loss": 0.014008046127855778, "grad_norm": 0.09537801146507263, "global_step": 23311, "epoch": 219, "lr": 5.939235483660053e-05} +{"train_loss": 0.014309991151094437, "grad_norm": 0.17034602165222168, "global_step": 23312, "epoch": 219, "lr": 5.938944381358114e-05} +{"train_loss": 0.013543435372412205, "grad_norm": 0.1684098243713379, "global_step": 23313, "epoch": 219, "lr": 5.938653275757129e-05} +{"train_loss": 0.015408260747790337, "grad_norm": 0.08705462515354156, "global_step": 23314, "epoch": 219, "lr": 5.938362166858127e-05} +{"train_loss": 0.018579265102744102, "grad_norm": 0.09118127822875977, "global_step": 23315, "epoch": 219, "lr": 5.938071054662129e-05} +{"train_loss": 0.013591269962489605, "grad_norm": 0.06218785420060158, "global_step": 23316, "epoch": 219, "lr": 5.937779939170156e-05} +{"train_loss": 0.01839391328394413, "grad_norm": 0.08637836575508118, "global_step": 23317, "epoch": 219, "lr": 5.9374888203832326e-05} +{"train_loss": 0.014612806960940361, "grad_norm": 0.14144283533096313, "global_step": 23318, "epoch": 219, "lr": 5.9371976983023803e-05} +{"train_loss": 0.015833594590284914, "grad_norm": 0.1463993489742279, "global_step": 23319, "epoch": 219, "lr": 5.9369065729286245e-05, "val_loss": 0.01910969987511635} +{"train_loss": 0.014118414372205734, "grad_norm": 0.07544376701116562, "global_step": 23320, "epoch": 220, "lr": 5.936615444262984e-05} +{"train_loss": 0.01536751352250576, "grad_norm": 0.09340684115886688, "global_step": 23321, "epoch": 220, "lr": 5.9363243123064874e-05} +{"train_loss": 0.015622889623045921, "grad_norm": 0.06448132544755936, "global_step": 23322, "epoch": 220, "lr": 5.936033177060152e-05} +{"train_loss": 0.016502268612384796, "grad_norm": 0.06577585637569427, "global_step": 23323, "epoch": 220, "lr": 5.935742038525003e-05} +{"train_loss": 0.015591500326991081, "grad_norm": 0.0733269453048706, "global_step": 23324, "epoch": 220, "lr": 5.935450896702065e-05} +{"train_loss": 0.01412157528102398, "grad_norm": 0.10337356477975845, "global_step": 23325, "epoch": 220, "lr": 5.9351597515923575e-05} +{"train_loss": 0.015471382066607475, "grad_norm": 0.11336258053779602, "global_step": 23326, "epoch": 220, "lr": 5.934868603196907e-05} +{"train_loss": 0.017859909683465958, "grad_norm": 0.1226758137345314, "global_step": 23327, "epoch": 220, "lr": 5.934577451516734e-05} +{"train_loss": 0.014818001538515091, "grad_norm": 0.2068326324224472, "global_step": 23328, "epoch": 220, "lr": 5.934286296552862e-05} +{"train_loss": 0.017321355640888214, "grad_norm": 0.09931077063083649, "global_step": 23329, "epoch": 220, "lr": 5.933995138306315e-05} +{"train_loss": 0.01631934940814972, "grad_norm": 0.060937587171792984, "global_step": 23330, "epoch": 220, "lr": 5.9337039767781154e-05} +{"train_loss": 0.015142105519771576, "grad_norm": 0.13583219051361084, "global_step": 23331, "epoch": 220, "lr": 5.933412811969286e-05} +{"train_loss": 0.015999171882867813, "grad_norm": 0.09340597689151764, "global_step": 23332, "epoch": 220, "lr": 5.933121643880848e-05} +{"train_loss": 0.015172825194895267, "grad_norm": 0.16271410882472992, "global_step": 23333, "epoch": 220, "lr": 5.9328304725138296e-05} +{"train_loss": 0.017722047865390778, "grad_norm": 0.06726782768964767, "global_step": 23334, "epoch": 220, "lr": 5.932539297869249e-05} +{"train_loss": 0.01643824391067028, "grad_norm": 0.1310960352420807, "global_step": 23335, "epoch": 220, "lr": 5.9322481199481295e-05} +{"train_loss": 0.01560964249074459, "grad_norm": 0.08649836480617523, "global_step": 23336, "epoch": 220, "lr": 5.9319569387514974e-05} +{"train_loss": 0.017290949821472168, "grad_norm": 0.12479974329471588, "global_step": 23337, "epoch": 220, "lr": 5.931665754280372e-05} +{"train_loss": 0.016823163256049156, "grad_norm": 0.1840762495994568, "global_step": 23338, "epoch": 220, "lr": 5.93137456653578e-05} +{"train_loss": 0.01519082859158516, "grad_norm": 0.12314589321613312, "global_step": 23339, "epoch": 220, "lr": 5.931083375518742e-05} +{"train_loss": 0.013927724212408066, "grad_norm": 0.12276537716388702, "global_step": 23340, "epoch": 220, "lr": 5.930792181230283e-05} +{"train_loss": 0.01772911846637726, "grad_norm": 0.11196660995483398, "global_step": 23341, "epoch": 220, "lr": 5.9305009836714257e-05} +{"train_loss": 0.015635013580322266, "grad_norm": 0.07075200229883194, "global_step": 23342, "epoch": 220, "lr": 5.93020978284319e-05} +{"train_loss": 0.015391897410154343, "grad_norm": 0.10458245128393173, "global_step": 23343, "epoch": 220, "lr": 5.929918578746603e-05} +{"train_loss": 0.019567664712667465, "grad_norm": 0.11198333650827408, "global_step": 23344, "epoch": 220, "lr": 5.929627371382687e-05} +{"train_loss": 0.016113631427288055, "grad_norm": 0.08619556576013565, "global_step": 23345, "epoch": 220, "lr": 5.929336160752465e-05} +{"train_loss": 0.013655276037752628, "grad_norm": 0.11930813640356064, "global_step": 23346, "epoch": 220, "lr": 5.9290449468569585e-05} +{"train_loss": 0.015124307945370674, "grad_norm": 0.051462918519973755, "global_step": 23347, "epoch": 220, "lr": 5.928753729697193e-05} +{"train_loss": 0.01688065566122532, "grad_norm": 0.09970209747552872, "global_step": 23348, "epoch": 220, "lr": 5.928462509274191e-05} +{"train_loss": 0.013805280439555645, "grad_norm": 0.0560837984085083, "global_step": 23349, "epoch": 220, "lr": 5.9281712855889725e-05} +{"train_loss": 0.0188567154109478, "grad_norm": 0.0642981082201004, "global_step": 23350, "epoch": 220, "lr": 5.9278800586425676e-05} +{"train_loss": 0.014155440032482147, "grad_norm": 0.057544682174921036, "global_step": 23351, "epoch": 220, "lr": 5.927588828435995e-05} +{"train_loss": 0.018776364624500275, "grad_norm": 0.19112490117549896, "global_step": 23352, "epoch": 220, "lr": 5.927297594970277e-05} +{"train_loss": 0.015840768814086914, "grad_norm": 0.05586344003677368, "global_step": 23353, "epoch": 220, "lr": 5.927006358246441e-05} +{"train_loss": 0.016615822911262512, "grad_norm": 0.09210440516471863, "global_step": 23354, "epoch": 220, "lr": 5.9267151182655045e-05} +{"train_loss": 0.015886906534433365, "grad_norm": 0.07640177756547928, "global_step": 23355, "epoch": 220, "lr": 5.9264238750284964e-05} +{"train_loss": 0.0168411023914814, "grad_norm": 0.07508932799100876, "global_step": 23356, "epoch": 220, "lr": 5.926132628536437e-05} +{"train_loss": 0.013102997094392776, "grad_norm": 0.06186237558722496, "global_step": 23357, "epoch": 220, "lr": 5.925841378790351e-05} +{"train_loss": 0.01584789901971817, "grad_norm": 0.10459694266319275, "global_step": 23358, "epoch": 220, "lr": 5.9255501257912606e-05} +{"train_loss": 0.01901813969016075, "grad_norm": 0.0988146960735321, "global_step": 23359, "epoch": 220, "lr": 5.92525886954019e-05} +{"train_loss": 0.015701893717050552, "grad_norm": 0.09717283397912979, "global_step": 23360, "epoch": 220, "lr": 5.9249676100381614e-05} +{"train_loss": 0.014965154230594635, "grad_norm": 0.07040685415267944, "global_step": 23361, "epoch": 220, "lr": 5.924676347286198e-05} +{"train_loss": 0.014628268778324127, "grad_norm": 0.0825047492980957, "global_step": 23362, "epoch": 220, "lr": 5.9243850812853264e-05} +{"train_loss": 0.013734543696045876, "grad_norm": 0.11662013828754425, "global_step": 23363, "epoch": 220, "lr": 5.924093812036567e-05} +{"train_loss": 0.016386747360229492, "grad_norm": 0.05020393431186676, "global_step": 23364, "epoch": 220, "lr": 5.923802539540941e-05} +{"train_loss": 0.01705281250178814, "grad_norm": 0.06091763451695442, "global_step": 23365, "epoch": 220, "lr": 5.9235112637994795e-05} +{"train_loss": 0.01582728698849678, "grad_norm": 0.11949550360441208, "global_step": 23366, "epoch": 220, "lr": 5.9232199848131965e-05} +{"train_loss": 0.01598343439400196, "grad_norm": 0.06931930780410767, "global_step": 23367, "epoch": 220, "lr": 5.922928702583123e-05} +{"train_loss": 0.01668347418308258, "grad_norm": 0.07132957130670547, "global_step": 23368, "epoch": 220, "lr": 5.922637417110277e-05} +{"train_loss": 0.015086912550032139, "grad_norm": 0.06888791173696518, "global_step": 23369, "epoch": 220, "lr": 5.922346128395686e-05} +{"train_loss": 0.01715867780148983, "grad_norm": 0.11575713008642197, "global_step": 23370, "epoch": 220, "lr": 5.9220548364403716e-05} +{"train_loss": 0.01814942993223667, "grad_norm": 0.08393552899360657, "global_step": 23371, "epoch": 220, "lr": 5.9217635412453576e-05} +{"train_loss": 0.01483171433210373, "grad_norm": 0.07909422367811203, "global_step": 23372, "epoch": 220, "lr": 5.921472242811668e-05} +{"train_loss": 0.01661921665072441, "grad_norm": 0.07925423234701157, "global_step": 23373, "epoch": 220, "lr": 5.921180941140324e-05} +{"train_loss": 0.015899933874607086, "grad_norm": 0.1678633987903595, "global_step": 23374, "epoch": 220, "lr": 5.920889636232352e-05} +{"train_loss": 0.017192617058753967, "grad_norm": 0.09956706315279007, "global_step": 23375, "epoch": 220, "lr": 5.9205983280887735e-05} +{"train_loss": 0.013363346457481384, "grad_norm": 0.05480686575174332, "global_step": 23376, "epoch": 220, "lr": 5.920307016710612e-05} +{"train_loss": 0.01508614607155323, "grad_norm": 0.08614643663167953, "global_step": 23377, "epoch": 220, "lr": 5.9200157020988946e-05} +{"train_loss": 0.014852474443614483, "grad_norm": 0.1682243049144745, "global_step": 23378, "epoch": 220, "lr": 5.9197243842546404e-05} +{"train_loss": 0.016716646030545235, "grad_norm": 0.07652239501476288, "global_step": 23379, "epoch": 220, "lr": 5.919433063178874e-05} +{"train_loss": 0.013004379346966743, "grad_norm": 0.092976875603199, "global_step": 23380, "epoch": 220, "lr": 5.9191417388726225e-05} +{"train_loss": 0.01532416045665741, "grad_norm": 0.0803752914071083, "global_step": 23381, "epoch": 220, "lr": 5.918850411336904e-05} +{"train_loss": 0.015690743923187256, "grad_norm": 0.1269846111536026, "global_step": 23382, "epoch": 220, "lr": 5.918559080572745e-05} +{"train_loss": 0.015697307884693146, "grad_norm": 0.05360075831413269, "global_step": 23383, "epoch": 220, "lr": 5.91826774658117e-05} +{"train_loss": 0.015600025653839111, "grad_norm": 0.06147860363125801, "global_step": 23384, "epoch": 220, "lr": 5.9179764093632005e-05} +{"train_loss": 0.0174274705350399, "grad_norm": 0.0712641254067421, "global_step": 23385, "epoch": 220, "lr": 5.917685068919861e-05} +{"train_loss": 0.017876390367746353, "grad_norm": 0.07875707745552063, "global_step": 23386, "epoch": 220, "lr": 5.917393725252176e-05} +{"train_loss": 0.014339854940772057, "grad_norm": 0.0772336795926094, "global_step": 23387, "epoch": 220, "lr": 5.9171023783611677e-05} +{"train_loss": 0.01585926115512848, "grad_norm": 0.08049017935991287, "global_step": 23388, "epoch": 220, "lr": 5.9168110282478604e-05} +{"train_loss": 0.016843192279338837, "grad_norm": 0.23266248404979706, "global_step": 23389, "epoch": 220, "lr": 5.91651967491328e-05} +{"train_loss": 0.013027427718043327, "grad_norm": 0.0736910030245781, "global_step": 23390, "epoch": 220, "lr": 5.916228318358445e-05} +{"train_loss": 0.016016390174627304, "grad_norm": 0.1276603788137436, "global_step": 23391, "epoch": 220, "lr": 5.915936958584385e-05} +{"train_loss": 0.014647776260972023, "grad_norm": 0.05391780659556389, "global_step": 23392, "epoch": 220, "lr": 5.9156455955921186e-05} +{"train_loss": 0.016433287411928177, "grad_norm": 0.09907449036836624, "global_step": 23393, "epoch": 220, "lr": 5.915354229382673e-05} +{"train_loss": 0.018107721582055092, "grad_norm": 0.07856852561235428, "global_step": 23394, "epoch": 220, "lr": 5.91506285995707e-05} +{"train_loss": 0.014539601281285286, "grad_norm": 0.05122426152229309, "global_step": 23395, "epoch": 220, "lr": 5.914771487316334e-05} +{"train_loss": 0.016041601076722145, "grad_norm": 0.057370152324438095, "global_step": 23396, "epoch": 220, "lr": 5.914480111461489e-05} +{"train_loss": 0.015365397557616234, "grad_norm": 0.08341636508703232, "global_step": 23397, "epoch": 220, "lr": 5.9141887323935596e-05} +{"train_loss": 0.01539568044245243, "grad_norm": 0.04826810210943222, "global_step": 23398, "epoch": 220, "lr": 5.9138973501135676e-05} +{"train_loss": 0.018807969987392426, "grad_norm": 0.057326365262269974, "global_step": 23399, "epoch": 220, "lr": 5.9136059646225375e-05} +{"train_loss": 0.014605858363211155, "grad_norm": 0.11932793259620667, "global_step": 23400, "epoch": 220, "lr": 5.913314575921494e-05} +{"train_loss": 0.01578233391046524, "grad_norm": 0.08539266139268875, "global_step": 23401, "epoch": 220, "lr": 5.91302318401146e-05} +{"train_loss": 0.012398388236761093, "grad_norm": 0.06930920481681824, "global_step": 23402, "epoch": 220, "lr": 5.9127317888934595e-05} +{"train_loss": 0.013819701969623566, "grad_norm": 0.10192341357469559, "global_step": 23403, "epoch": 220, "lr": 5.912440390568517e-05} +{"train_loss": 0.016067007556557655, "grad_norm": 0.1271803081035614, "global_step": 23404, "epoch": 220, "lr": 5.912148989037656e-05} +{"train_loss": 0.013768400065600872, "grad_norm": 0.08971705287694931, "global_step": 23405, "epoch": 220, "lr": 5.911857584301898e-05} +{"train_loss": 0.014659236185252666, "grad_norm": 0.08476833999156952, "global_step": 23406, "epoch": 220, "lr": 5.9115661763622706e-05} +{"train_loss": 0.019543761387467384, "grad_norm": 0.09507419168949127, "global_step": 23407, "epoch": 220, "lr": 5.9112747652197964e-05} +{"train_loss": 0.01432037353515625, "grad_norm": 0.10261056572198868, "global_step": 23408, "epoch": 220, "lr": 5.910983350875498e-05} +{"train_loss": 0.01496954821050167, "grad_norm": 0.08825533837080002, "global_step": 23409, "epoch": 220, "lr": 5.910691933330401e-05} +{"train_loss": 0.01576535403728485, "grad_norm": 0.10901574045419693, "global_step": 23410, "epoch": 220, "lr": 5.910400512585529e-05} +{"train_loss": 0.01954098418354988, "grad_norm": 0.06936545670032501, "global_step": 23411, "epoch": 220, "lr": 5.910109088641904e-05} +{"train_loss": 0.01808381825685501, "grad_norm": 0.10197970271110535, "global_step": 23412, "epoch": 220, "lr": 5.909817661500553e-05} +{"train_loss": 0.014793353155255318, "grad_norm": 0.06674806028604507, "global_step": 23413, "epoch": 220, "lr": 5.909526231162498e-05} +{"train_loss": 0.016103651374578476, "grad_norm": 0.054457277059555054, "global_step": 23414, "epoch": 220, "lr": 5.909234797628763e-05} +{"train_loss": 0.014003155753016472, "grad_norm": 0.09337523579597473, "global_step": 23415, "epoch": 220, "lr": 5.908943360900372e-05} +{"train_loss": 0.015883173793554306, "grad_norm": 0.05975873023271561, "global_step": 23416, "epoch": 220, "lr": 5.908651920978351e-05} +{"train_loss": 0.016377266496419907, "grad_norm": 0.07001911103725433, "global_step": 23417, "epoch": 220, "lr": 5.9083604778637194e-05} +{"train_loss": 0.014868042431771755, "grad_norm": 0.07329859584569931, "global_step": 23418, "epoch": 220, "lr": 5.908069031557507e-05} +{"train_loss": 0.015616882592439651, "grad_norm": 0.08956946432590485, "global_step": 23419, "epoch": 220, "lr": 5.9077775820607326e-05} +{"train_loss": 0.014507909305393696, "grad_norm": 0.05439429730176926, "global_step": 23420, "epoch": 220, "lr": 5.907486129374424e-05} +{"train_loss": 0.015564775094389915, "grad_norm": 0.08009200543165207, "global_step": 23421, "epoch": 220, "lr": 5.9071946734996054e-05} +{"train_loss": 0.018651962280273438, "grad_norm": 0.06607768684625626, "global_step": 23422, "epoch": 220, "lr": 5.906903214437296e-05} +{"train_loss": 0.015340238809585571, "grad_norm": 0.056713320314884186, "global_step": 23423, "epoch": 220, "lr": 5.906611752188524e-05} +{"train_loss": 0.013543449342250824, "grad_norm": 0.07280632108449936, "global_step": 23424, "epoch": 220, "lr": 5.906320286754313e-05} +{"train_loss": 0.01580144968410989, "grad_norm": 0.11108912527561188, "global_step": 23425, "epoch": 220, "lr": 5.906028818135687e-05, "train_batch_mse_error": 0.013995746150612831, "val_loss": 0.019285591319203377} +{"train_loss": 0.016023706644773483, "grad_norm": 0.06251387298107147, "global_step": 23426, "epoch": 221, "lr": 5.905737346333669e-05} +{"train_loss": 0.015247974544763565, "grad_norm": 0.08647878468036652, "global_step": 23427, "epoch": 221, "lr": 5.905445871349286e-05} +{"train_loss": 0.01558306347578764, "grad_norm": 0.052324648946523666, "global_step": 23428, "epoch": 221, "lr": 5.9051543931835575e-05} +{"train_loss": 0.014301576651632786, "grad_norm": 0.07107242941856384, "global_step": 23429, "epoch": 221, "lr": 5.90486291183751e-05} +{"train_loss": 0.016124848276376724, "grad_norm": 0.04606110230088234, "global_step": 23430, "epoch": 221, "lr": 5.90457142731217e-05} +{"train_loss": 0.017536301165819168, "grad_norm": 0.07894758135080338, "global_step": 23431, "epoch": 221, "lr": 5.9042799396085576e-05} +{"train_loss": 0.014837943948805332, "grad_norm": 0.07794592529535294, "global_step": 23432, "epoch": 221, "lr": 5.903988448727699e-05} +{"train_loss": 0.016342071816325188, "grad_norm": 0.08035045117139816, "global_step": 23433, "epoch": 221, "lr": 5.903696954670619e-05} +{"train_loss": 0.017088264226913452, "grad_norm": 0.07014977186918259, "global_step": 23434, "epoch": 221, "lr": 5.903405457438339e-05} +{"train_loss": 0.0153890959918499, "grad_norm": 0.12768003344535828, "global_step": 23435, "epoch": 221, "lr": 5.903113957031887e-05} +{"train_loss": 0.014344628900289536, "grad_norm": 0.08056159317493439, "global_step": 23436, "epoch": 221, "lr": 5.902822453452285e-05} +{"train_loss": 0.014695393852889538, "grad_norm": 0.05398640036582947, "global_step": 23437, "epoch": 221, "lr": 5.902530946700556e-05} +{"train_loss": 0.017186589539051056, "grad_norm": 0.12337492406368256, "global_step": 23438, "epoch": 221, "lr": 5.902239436777728e-05} +{"train_loss": 0.015922900289297104, "grad_norm": 0.06551169604063034, "global_step": 23439, "epoch": 221, "lr": 5.9019479236848216e-05} +{"train_loss": 0.017382726073265076, "grad_norm": 0.05923660099506378, "global_step": 23440, "epoch": 221, "lr": 5.901656407422863e-05} +{"train_loss": 0.017502151429653168, "grad_norm": 0.07151515781879425, "global_step": 23441, "epoch": 221, "lr": 5.901364887992875e-05} +{"train_loss": 0.013145070523023605, "grad_norm": 0.08161623030900955, "global_step": 23442, "epoch": 221, "lr": 5.901073365395884e-05} +{"train_loss": 0.01644659787416458, "grad_norm": 0.1103212833404541, "global_step": 23443, "epoch": 221, "lr": 5.9007818396329125e-05} +{"train_loss": 0.016647737473249435, "grad_norm": 0.07182623445987701, "global_step": 23444, "epoch": 221, "lr": 5.9004903107049857e-05} +{"train_loss": 0.01488537434488535, "grad_norm": 0.05816693231463432, "global_step": 23445, "epoch": 221, "lr": 5.900198778613127e-05} +{"train_loss": 0.01545853540301323, "grad_norm": 0.09181475639343262, "global_step": 23446, "epoch": 221, "lr": 5.89990724335836e-05} +{"train_loss": 0.015945330262184143, "grad_norm": 0.12220998853445053, "global_step": 23447, "epoch": 221, "lr": 5.8996157049417134e-05} +{"train_loss": 0.015639375895261765, "grad_norm": 0.0730312392115593, "global_step": 23448, "epoch": 221, "lr": 5.899324163364206e-05} +{"train_loss": 0.01594013161957264, "grad_norm": 0.08206367492675781, "global_step": 23449, "epoch": 221, "lr": 5.899032618626865e-05} +{"train_loss": 0.016834575682878494, "grad_norm": 0.13854283094406128, "global_step": 23450, "epoch": 221, "lr": 5.898741070730717e-05} +{"train_loss": 0.01962261274456978, "grad_norm": 0.07561897486448288, "global_step": 23451, "epoch": 221, "lr": 5.898449519676781e-05} +{"train_loss": 0.01578112132847309, "grad_norm": 0.0940614640712738, "global_step": 23452, "epoch": 221, "lr": 5.898157965466085e-05} +{"train_loss": 0.020388757809996605, "grad_norm": 0.1249506026506424, "global_step": 23453, "epoch": 221, "lr": 5.8978664080996516e-05} +{"train_loss": 0.016228774562478065, "grad_norm": 0.05751967802643776, "global_step": 23454, "epoch": 221, "lr": 5.8975748475785064e-05} +{"train_loss": 0.015189539641141891, "grad_norm": 0.06947022676467896, "global_step": 23455, "epoch": 221, "lr": 5.897283283903674e-05} +{"train_loss": 0.014351686462759972, "grad_norm": 0.05200623348355293, "global_step": 23456, "epoch": 221, "lr": 5.896991717076178e-05} +{"train_loss": 0.0150102274492383, "grad_norm": 0.07227031886577606, "global_step": 23457, "epoch": 221, "lr": 5.8967001470970426e-05} +{"train_loss": 0.019783709198236465, "grad_norm": 0.0869322419166565, "global_step": 23458, "epoch": 221, "lr": 5.8964085739672937e-05} +{"train_loss": 0.015225173905491829, "grad_norm": 0.13926956057548523, "global_step": 23459, "epoch": 221, "lr": 5.896116997687956e-05} +{"train_loss": 0.016834955662488937, "grad_norm": 0.11538653075695038, "global_step": 23460, "epoch": 221, "lr": 5.895825418260051e-05} +{"train_loss": 0.017445005476474762, "grad_norm": 0.0661938264966011, "global_step": 23461, "epoch": 221, "lr": 5.8955338356846055e-05} +{"train_loss": 0.015905119478702545, "grad_norm": 0.06189025938510895, "global_step": 23462, "epoch": 221, "lr": 5.8952422499626445e-05} +{"train_loss": 0.014168365858495235, "grad_norm": 0.06758729368448257, "global_step": 23463, "epoch": 221, "lr": 5.89495066109519e-05} +{"train_loss": 0.015528212301433086, "grad_norm": 0.08392702788114548, "global_step": 23464, "epoch": 221, "lr": 5.8946590690832694e-05} +{"train_loss": 0.014991558156907558, "grad_norm": 0.11699268966913223, "global_step": 23465, "epoch": 221, "lr": 5.894367473927905e-05} +{"train_loss": 0.014163696207106113, "grad_norm": 0.058957941830158234, "global_step": 23466, "epoch": 221, "lr": 5.8940758756301226e-05} +{"train_loss": 0.011523926630616188, "grad_norm": 0.058330364525318146, "global_step": 23467, "epoch": 221, "lr": 5.893784274190947e-05} +{"train_loss": 0.01563597470521927, "grad_norm": 0.060488179326057434, "global_step": 23468, "epoch": 221, "lr": 5.8934926696114026e-05} +{"train_loss": 0.014800574630498886, "grad_norm": 0.06421375274658203, "global_step": 23469, "epoch": 221, "lr": 5.8932010618925116e-05} +{"train_loss": 0.014416907913982868, "grad_norm": 0.16518397629261017, "global_step": 23470, "epoch": 221, "lr": 5.892909451035301e-05} +{"train_loss": 0.014035020023584366, "grad_norm": 0.09978179633617401, "global_step": 23471, "epoch": 221, "lr": 5.892617837040797e-05} +{"train_loss": 0.013314304873347282, "grad_norm": 0.0685553252696991, "global_step": 23472, "epoch": 221, "lr": 5.8923262199100205e-05} +{"train_loss": 0.016622930765151978, "grad_norm": 0.06586313992738724, "global_step": 23473, "epoch": 221, "lr": 5.8920345996439966e-05} +{"train_loss": 0.015611769631505013, "grad_norm": 0.11238884180784225, "global_step": 23474, "epoch": 221, "lr": 5.891742976243753e-05} +{"train_loss": 0.016661223024129868, "grad_norm": 0.08061637729406357, "global_step": 23475, "epoch": 221, "lr": 5.891451349710311e-05} +{"train_loss": 0.01622830331325531, "grad_norm": 0.09912282973527908, "global_step": 23476, "epoch": 221, "lr": 5.8911597200446964e-05} +{"train_loss": 0.013319098390638828, "grad_norm": 0.10823361575603485, "global_step": 23477, "epoch": 221, "lr": 5.890868087247936e-05} +{"train_loss": 0.013980351388454437, "grad_norm": 0.09690544009208679, "global_step": 23478, "epoch": 221, "lr": 5.89057645132105e-05} +{"train_loss": 0.016225077211856842, "grad_norm": 0.10198795795440674, "global_step": 23479, "epoch": 221, "lr": 5.8902848122650675e-05} +{"train_loss": 0.020186059176921844, "grad_norm": 0.18269027769565582, "global_step": 23480, "epoch": 221, "lr": 5.889993170081012e-05} +{"train_loss": 0.016136623919010162, "grad_norm": 0.05767425149679184, "global_step": 23481, "epoch": 221, "lr": 5.889701524769906e-05} +{"train_loss": 0.01570429839193821, "grad_norm": 0.06560461223125458, "global_step": 23482, "epoch": 221, "lr": 5.8894098763327766e-05} +{"train_loss": 0.016886305063962936, "grad_norm": 0.0687393769621849, "global_step": 23483, "epoch": 221, "lr": 5.889118224770648e-05} +{"train_loss": 0.01751590520143509, "grad_norm": 0.0652039498090744, "global_step": 23484, "epoch": 221, "lr": 5.8888265700845436e-05} +{"train_loss": 0.015548419207334518, "grad_norm": 0.08334024250507355, "global_step": 23485, "epoch": 221, "lr": 5.8885349122754885e-05} +{"train_loss": 0.016391992568969727, "grad_norm": 0.0759173035621643, "global_step": 23486, "epoch": 221, "lr": 5.8882432513445106e-05} +{"train_loss": 0.014531634747982025, "grad_norm": 0.1495712250471115, "global_step": 23487, "epoch": 221, "lr": 5.8879515872926296e-05} +{"train_loss": 0.01516999676823616, "grad_norm": 0.20668242871761322, "global_step": 23488, "epoch": 221, "lr": 5.887659920120875e-05} +{"train_loss": 0.016433443874120712, "grad_norm": 0.15507303178310394, "global_step": 23489, "epoch": 221, "lr": 5.887368249830268e-05} +{"train_loss": 0.01583327166736126, "grad_norm": 0.13546201586723328, "global_step": 23490, "epoch": 221, "lr": 5.887076576421836e-05} +{"train_loss": 0.016828324645757675, "grad_norm": 0.12203044444322586, "global_step": 23491, "epoch": 221, "lr": 5.886784899896602e-05} +{"train_loss": 0.015193916857242584, "grad_norm": 0.07930758595466614, "global_step": 23492, "epoch": 221, "lr": 5.886493220255591e-05} +{"train_loss": 0.013246545568108559, "grad_norm": 0.06617022305727005, "global_step": 23493, "epoch": 221, "lr": 5.886201537499829e-05} +{"train_loss": 0.017787296324968338, "grad_norm": 0.0704529732465744, "global_step": 23494, "epoch": 221, "lr": 5.885909851630341e-05} +{"train_loss": 0.015551363117992878, "grad_norm": 0.1249404102563858, "global_step": 23495, "epoch": 221, "lr": 5.88561816264815e-05} +{"train_loss": 0.015634430572390556, "grad_norm": 0.1337207406759262, "global_step": 23496, "epoch": 221, "lr": 5.885326470554282e-05} +{"train_loss": 0.014606527984142303, "grad_norm": 0.10114353895187378, "global_step": 23497, "epoch": 221, "lr": 5.885034775349762e-05} +{"train_loss": 0.012629996985197067, "grad_norm": 0.09169211238622665, "global_step": 23498, "epoch": 221, "lr": 5.884743077035615e-05} +{"train_loss": 0.015830347314476967, "grad_norm": 0.07639127969741821, "global_step": 23499, "epoch": 221, "lr": 5.884451375612865e-05} +{"train_loss": 0.01646466925740242, "grad_norm": 0.0643322765827179, "global_step": 23500, "epoch": 221, "lr": 5.8841596710825386e-05} +{"train_loss": 0.015720557421445847, "grad_norm": 0.20384293794631958, "global_step": 23501, "epoch": 221, "lr": 5.88386796344566e-05} +{"train_loss": 0.016531197354197502, "grad_norm": 0.07134788483381271, "global_step": 23502, "epoch": 221, "lr": 5.88357625270325e-05} +{"train_loss": 0.015202285721898079, "grad_norm": 0.08963260054588318, "global_step": 23503, "epoch": 221, "lr": 5.8832845388563415e-05} +{"train_loss": 0.017521629109978676, "grad_norm": 0.05284943804144859, "global_step": 23504, "epoch": 221, "lr": 5.8829928219059526e-05} +{"train_loss": 0.01642979308962822, "grad_norm": 0.14090237021446228, "global_step": 23505, "epoch": 221, "lr": 5.882701101853112e-05} +{"train_loss": 0.014982979744672775, "grad_norm": 0.10035574436187744, "global_step": 23506, "epoch": 221, "lr": 5.8824093786988434e-05} +{"train_loss": 0.016082001850008965, "grad_norm": 0.10127793252468109, "global_step": 23507, "epoch": 221, "lr": 5.882117652444172e-05} +{"train_loss": 0.016812993213534355, "grad_norm": 0.05520198494195938, "global_step": 23508, "epoch": 221, "lr": 5.881825923090123e-05} +{"train_loss": 0.016498949378728867, "grad_norm": 0.06393568962812424, "global_step": 23509, "epoch": 221, "lr": 5.881534190637722e-05} +{"train_loss": 0.018398679792881012, "grad_norm": 0.06529742479324341, "global_step": 23510, "epoch": 221, "lr": 5.881242455087992e-05} +{"train_loss": 0.015648089349269867, "grad_norm": 0.06296087801456451, "global_step": 23511, "epoch": 221, "lr": 5.88095071644196e-05} +{"train_loss": 0.01643921062350273, "grad_norm": 0.0682777464389801, "global_step": 23512, "epoch": 221, "lr": 5.8806589747006505e-05} +{"train_loss": 0.016017504036426544, "grad_norm": 0.08043394237756729, "global_step": 23513, "epoch": 221, "lr": 5.880367229865087e-05} +{"train_loss": 0.01756991259753704, "grad_norm": 0.0968056321144104, "global_step": 23514, "epoch": 221, "lr": 5.880075481936296e-05} +{"train_loss": 0.01601811870932579, "grad_norm": 0.12485696375370026, "global_step": 23515, "epoch": 221, "lr": 5.879783730915305e-05} +{"train_loss": 0.014839204028248787, "grad_norm": 0.07672029733657837, "global_step": 23516, "epoch": 221, "lr": 5.879491976803133e-05} +{"train_loss": 0.016268290579319, "grad_norm": 0.11671639233827591, "global_step": 23517, "epoch": 221, "lr": 5.87920021960081e-05} +{"train_loss": 0.01607527583837509, "grad_norm": 0.12341026961803436, "global_step": 23518, "epoch": 221, "lr": 5.87890845930936e-05} +{"train_loss": 0.013991945423185825, "grad_norm": 0.1317870318889618, "global_step": 23519, "epoch": 221, "lr": 5.878616695929808e-05} +{"train_loss": 0.013917004689574242, "grad_norm": 0.08543431013822556, "global_step": 23520, "epoch": 221, "lr": 5.87832492946318e-05} +{"train_loss": 0.014554580673575401, "grad_norm": 0.09717928618192673, "global_step": 23521, "epoch": 221, "lr": 5.8780331599104996e-05} +{"train_loss": 0.015104781836271286, "grad_norm": 0.05945736542344093, "global_step": 23522, "epoch": 221, "lr": 5.877741387272793e-05} +{"train_loss": 0.014136547222733498, "grad_norm": 0.07065319269895554, "global_step": 23523, "epoch": 221, "lr": 5.877449611551084e-05} +{"train_loss": 0.015607102774083614, "grad_norm": 0.11555381119251251, "global_step": 23524, "epoch": 221, "lr": 5.8771578327463986e-05} +{"train_loss": 0.018643882125616074, "grad_norm": 0.07424964755773544, "global_step": 23525, "epoch": 221, "lr": 5.8768660508597615e-05} +{"train_loss": 0.017423804849386215, "grad_norm": 0.07643967121839523, "global_step": 23526, "epoch": 221, "lr": 5.876574265892199e-05} +{"train_loss": 0.013863923028111458, "grad_norm": 0.07513070851564407, "global_step": 23527, "epoch": 221, "lr": 5.8762824778447366e-05} +{"train_loss": 0.014382686465978622, "grad_norm": 0.10381913930177689, "global_step": 23528, "epoch": 221, "lr": 5.8759906867183965e-05} +{"train_loss": 0.01642083376646042, "grad_norm": 0.05752858892083168, "global_step": 23529, "epoch": 221, "lr": 5.8756988925142084e-05} +{"train_loss": 0.014927806332707405, "grad_norm": 0.10454080253839493, "global_step": 23530, "epoch": 221, "lr": 5.8754070952331955e-05} +{"train_loss": 0.015818035695701838, "grad_norm": 0.15132048726081848, "global_step": 23531, "epoch": 221, "lr": 5.875115294876381e-05, "val_loss": 0.018801430240273476} +{"train_loss": 0.016802465543150902, "grad_norm": 0.0577550008893013, "global_step": 23532, "epoch": 222, "lr": 5.874823491444792e-05} +{"train_loss": 0.01261361874639988, "grad_norm": 0.09415534883737564, "global_step": 23533, "epoch": 222, "lr": 5.874531684939455e-05} +{"train_loss": 0.017001628875732422, "grad_norm": 0.059987641870975494, "global_step": 23534, "epoch": 222, "lr": 5.8742398753613924e-05} +{"train_loss": 0.01261292677372694, "grad_norm": 0.08344515413045883, "global_step": 23535, "epoch": 222, "lr": 5.873948062711633e-05} +{"train_loss": 0.013727319426834583, "grad_norm": 0.05697598680853844, "global_step": 23536, "epoch": 222, "lr": 5.8736562469911984e-05} +{"train_loss": 0.013058113865554333, "grad_norm": 0.13896465301513672, "global_step": 23537, "epoch": 222, "lr": 5.8733644282011166e-05} +{"train_loss": 0.014054574072360992, "grad_norm": 0.11124493181705475, "global_step": 23538, "epoch": 222, "lr": 5.8730726063424114e-05} +{"train_loss": 0.015244566835463047, "grad_norm": 0.058648835867643356, "global_step": 23539, "epoch": 222, "lr": 5.872780781416109e-05} +{"train_loss": 0.015172429382801056, "grad_norm": 0.06273049116134644, "global_step": 23540, "epoch": 222, "lr": 5.872488953423234e-05} +{"train_loss": 0.015541684813797474, "grad_norm": 0.08186265081167221, "global_step": 23541, "epoch": 222, "lr": 5.8721971223648145e-05} +{"train_loss": 0.01591717079281807, "grad_norm": 0.0987263172864914, "global_step": 23542, "epoch": 222, "lr": 5.8719052882418725e-05} +{"train_loss": 0.01690492033958435, "grad_norm": 0.13727571070194244, "global_step": 23543, "epoch": 222, "lr": 5.871613451055433e-05} +{"train_loss": 0.01602262631058693, "grad_norm": 0.10408034175634384, "global_step": 23544, "epoch": 222, "lr": 5.871321610806525e-05} +{"train_loss": 0.01691434532403946, "grad_norm": 0.08613449335098267, "global_step": 23545, "epoch": 222, "lr": 5.87102976749617e-05} +{"train_loss": 0.0158012006431818, "grad_norm": 0.08317092806100845, "global_step": 23546, "epoch": 222, "lr": 5.8707379211253956e-05} +{"train_loss": 0.014800745993852615, "grad_norm": 0.07835224270820618, "global_step": 23547, "epoch": 222, "lr": 5.870446071695227e-05} +{"train_loss": 0.014521672390401363, "grad_norm": 0.07956063747406006, "global_step": 23548, "epoch": 222, "lr": 5.870154219206689e-05} +{"train_loss": 0.015934176743030548, "grad_norm": 0.08411932736635208, "global_step": 23549, "epoch": 222, "lr": 5.8698623636608086e-05} +{"train_loss": 0.015576453879475594, "grad_norm": 0.06356032937765121, "global_step": 23550, "epoch": 222, "lr": 5.86957050505861e-05} +{"train_loss": 0.015551757998764515, "grad_norm": 0.06815628707408905, "global_step": 23551, "epoch": 222, "lr": 5.869278643401118e-05} +{"train_loss": 0.016057392582297325, "grad_norm": 0.055535897612571716, "global_step": 23552, "epoch": 222, "lr": 5.8689867786893606e-05} +{"train_loss": 0.017309512943029404, "grad_norm": 0.08821971714496613, "global_step": 23553, "epoch": 222, "lr": 5.86869491092436e-05} +{"train_loss": 0.016113553196191788, "grad_norm": 0.08750560134649277, "global_step": 23554, "epoch": 222, "lr": 5.868403040107143e-05} +{"train_loss": 0.012682611122727394, "grad_norm": 0.04754219949245453, "global_step": 23555, "epoch": 222, "lr": 5.8681111662387354e-05} +{"train_loss": 0.01936100795865059, "grad_norm": 0.13053026795387268, "global_step": 23556, "epoch": 222, "lr": 5.8678192893201645e-05} +{"train_loss": 0.020095620304346085, "grad_norm": 0.09839367866516113, "global_step": 23557, "epoch": 222, "lr": 5.867527409352451e-05} +{"train_loss": 0.014930395409464836, "grad_norm": 0.09418200701475143, "global_step": 23558, "epoch": 222, "lr": 5.8672355263366255e-05} +{"train_loss": 0.01659422367811203, "grad_norm": 0.07931885123252869, "global_step": 23559, "epoch": 222, "lr": 5.866943640273712e-05} +{"train_loss": 0.015972072258591652, "grad_norm": 0.050237324088811874, "global_step": 23560, "epoch": 222, "lr": 5.866651751164734e-05} +{"train_loss": 0.01579134911298752, "grad_norm": 0.054411739110946655, "global_step": 23561, "epoch": 222, "lr": 5.866359859010719e-05} +{"train_loss": 0.016096273437142372, "grad_norm": 0.06281239539384842, "global_step": 23562, "epoch": 222, "lr": 5.8660679638126936e-05} +{"train_loss": 0.018196092918515205, "grad_norm": 0.14305061101913452, "global_step": 23563, "epoch": 222, "lr": 5.865776065571681e-05} +{"train_loss": 0.015556835569441319, "grad_norm": 0.06984895467758179, "global_step": 23564, "epoch": 222, "lr": 5.865484164288708e-05} +{"train_loss": 0.015186846256256104, "grad_norm": 0.06719434261322021, "global_step": 23565, "epoch": 222, "lr": 5.8651922599648003e-05} +{"train_loss": 0.017124582082033157, "grad_norm": 0.09978237003087997, "global_step": 23566, "epoch": 222, "lr": 5.864900352600983e-05} +{"train_loss": 0.016796842217445374, "grad_norm": 0.0682898461818695, "global_step": 23567, "epoch": 222, "lr": 5.8646084421982806e-05} +{"train_loss": 0.015731684863567352, "grad_norm": 0.11207001656293869, "global_step": 23568, "epoch": 222, "lr": 5.8643165287577225e-05} +{"train_loss": 0.0166738610714674, "grad_norm": 0.07868632674217224, "global_step": 23569, "epoch": 222, "lr": 5.86402461228033e-05} +{"train_loss": 0.014647230505943298, "grad_norm": 0.06341895461082458, "global_step": 23570, "epoch": 222, "lr": 5.8637326927671324e-05} +{"train_loss": 0.015834560617804527, "grad_norm": 0.08036196231842041, "global_step": 23571, "epoch": 222, "lr": 5.863440770219154e-05} +{"train_loss": 0.015403102152049541, "grad_norm": 0.09724003076553345, "global_step": 23572, "epoch": 222, "lr": 5.863148844637418e-05} +{"train_loss": 0.015229059383273125, "grad_norm": 0.07939014583826065, "global_step": 23573, "epoch": 222, "lr": 5.862856916022954e-05} +{"train_loss": 0.017283663153648376, "grad_norm": 0.06311754137277603, "global_step": 23574, "epoch": 222, "lr": 5.862564984376785e-05} +{"train_loss": 0.016169384121894836, "grad_norm": 0.1107548177242279, "global_step": 23575, "epoch": 222, "lr": 5.8622730496999386e-05} +{"train_loss": 0.014688674360513687, "grad_norm": 0.05306967720389366, "global_step": 23576, "epoch": 222, "lr": 5.86198111199344e-05} +{"train_loss": 0.0181666761636734, "grad_norm": 0.09684717655181885, "global_step": 23577, "epoch": 222, "lr": 5.861689171258314e-05} +{"train_loss": 0.015022533014416695, "grad_norm": 0.10383988916873932, "global_step": 23578, "epoch": 222, "lr": 5.8613972274955875e-05} +{"train_loss": 0.014630332589149475, "grad_norm": 0.07934226095676422, "global_step": 23579, "epoch": 222, "lr": 5.861105280706284e-05} +{"train_loss": 0.0150712039321661, "grad_norm": 0.10139206051826477, "global_step": 23580, "epoch": 222, "lr": 5.8608133308914335e-05} +{"train_loss": 0.014422539621591568, "grad_norm": 0.09522093087434769, "global_step": 23581, "epoch": 222, "lr": 5.860521378052059e-05} +{"train_loss": 0.012994447723031044, "grad_norm": 0.05499538779258728, "global_step": 23582, "epoch": 222, "lr": 5.860229422189184e-05} +{"train_loss": 0.015812896192073822, "grad_norm": 0.09588054567575455, "global_step": 23583, "epoch": 222, "lr": 5.8599374633038403e-05} +{"train_loss": 0.01708805561065674, "grad_norm": 0.13896451890468597, "global_step": 23584, "epoch": 222, "lr": 5.859645501397048e-05} +{"train_loss": 0.016725702211260796, "grad_norm": 0.08815816789865494, "global_step": 23585, "epoch": 222, "lr": 5.859353536469836e-05} +{"train_loss": 0.01425998006016016, "grad_norm": 0.07453617453575134, "global_step": 23586, "epoch": 222, "lr": 5.8590615685232296e-05} +{"train_loss": 0.01623515412211418, "grad_norm": 0.12712644040584564, "global_step": 23587, "epoch": 222, "lr": 5.8587695975582545e-05} +{"train_loss": 0.017999805510044098, "grad_norm": 0.0995744839310646, "global_step": 23588, "epoch": 222, "lr": 5.8584776235759366e-05} +{"train_loss": 0.018278004601597786, "grad_norm": 0.08514904230833054, "global_step": 23589, "epoch": 222, "lr": 5.858185646577301e-05} +{"train_loss": 0.01491545420140028, "grad_norm": 0.06485096365213394, "global_step": 23590, "epoch": 222, "lr": 5.857893666563374e-05} +{"train_loss": 0.015048578381538391, "grad_norm": 0.07072173804044724, "global_step": 23591, "epoch": 222, "lr": 5.857601683535182e-05} +{"train_loss": 0.015081290155649185, "grad_norm": 0.0677427276968956, "global_step": 23592, "epoch": 222, "lr": 5.8573096974937523e-05} +{"train_loss": 0.015311524271965027, "grad_norm": 0.11074910312891006, "global_step": 23593, "epoch": 222, "lr": 5.857017708440108e-05} +{"train_loss": 0.012970361858606339, "grad_norm": 0.13225345313549042, "global_step": 23594, "epoch": 222, "lr": 5.8567257163752766e-05} +{"train_loss": 0.017564881592988968, "grad_norm": 0.09415580332279205, "global_step": 23595, "epoch": 222, "lr": 5.856433721300283e-05} +{"train_loss": 0.016107888892292976, "grad_norm": 0.08835813403129578, "global_step": 23596, "epoch": 222, "lr": 5.856141723216152e-05} +{"train_loss": 0.016135312616825104, "grad_norm": 0.059651803225278854, "global_step": 23597, "epoch": 222, "lr": 5.855849722123915e-05} +{"train_loss": 0.014355393126606941, "grad_norm": 0.07921043783426285, "global_step": 23598, "epoch": 222, "lr": 5.855557718024591e-05} +{"train_loss": 0.01276993565261364, "grad_norm": 0.055732399225234985, "global_step": 23599, "epoch": 222, "lr": 5.855265710919211e-05} +{"train_loss": 0.015265095047652721, "grad_norm": 0.05603528395295143, "global_step": 23600, "epoch": 222, "lr": 5.8549737008088e-05} +{"train_loss": 0.016387134790420532, "grad_norm": 0.12495886534452438, "global_step": 23601, "epoch": 222, "lr": 5.854681687694381e-05} +{"train_loss": 0.015395313501358032, "grad_norm": 0.05776235833764076, "global_step": 23602, "epoch": 222, "lr": 5.854389671576983e-05} +{"train_loss": 0.01477055624127388, "grad_norm": 0.09409205615520477, "global_step": 23603, "epoch": 222, "lr": 5.854097652457632e-05} +{"train_loss": 0.014019976370036602, "grad_norm": 0.16371995210647583, "global_step": 23604, "epoch": 222, "lr": 5.853805630337353e-05} +{"train_loss": 0.015207529999315739, "grad_norm": 0.05768972635269165, "global_step": 23605, "epoch": 222, "lr": 5.853513605217173e-05} +{"train_loss": 0.016193699091672897, "grad_norm": 0.0563703291118145, "global_step": 23606, "epoch": 222, "lr": 5.8532215770981155e-05} +{"train_loss": 0.017698511481285095, "grad_norm": 0.07128755003213882, "global_step": 23607, "epoch": 222, "lr": 5.852929545981211e-05} +{"train_loss": 0.015958216041326523, "grad_norm": 0.06891926378011703, "global_step": 23608, "epoch": 222, "lr": 5.8526375118674806e-05} +{"train_loss": 0.01722106710076332, "grad_norm": 0.062150463461875916, "global_step": 23609, "epoch": 222, "lr": 5.852345474757954e-05} +{"train_loss": 0.015606323257088661, "grad_norm": 0.07901118695735931, "global_step": 23610, "epoch": 222, "lr": 5.852053434653656e-05} +{"train_loss": 0.01626501977443695, "grad_norm": 0.10445744544267654, "global_step": 23611, "epoch": 222, "lr": 5.8517613915556116e-05} +{"train_loss": 0.015152432955801487, "grad_norm": 0.06761153042316437, "global_step": 23612, "epoch": 222, "lr": 5.8514693454648504e-05} +{"train_loss": 0.01683362014591694, "grad_norm": 0.06638087332248688, "global_step": 23613, "epoch": 222, "lr": 5.851177296382394e-05} +{"train_loss": 0.018709588795900345, "grad_norm": 0.07126125693321228, "global_step": 23614, "epoch": 222, "lr": 5.850885244309271e-05} +{"train_loss": 0.013741249218583107, "grad_norm": 0.05097503215074539, "global_step": 23615, "epoch": 222, "lr": 5.850593189246508e-05} +{"train_loss": 0.015213506296277046, "grad_norm": 0.07081654667854309, "global_step": 23616, "epoch": 222, "lr": 5.8503011311951307e-05} +{"train_loss": 0.013206062838435173, "grad_norm": 0.051709406077861786, "global_step": 23617, "epoch": 222, "lr": 5.850009070156165e-05} +{"train_loss": 0.01704603061079979, "grad_norm": 0.05457436665892601, "global_step": 23618, "epoch": 222, "lr": 5.849717006130637e-05} +{"train_loss": 0.01575450412929058, "grad_norm": 0.06267713755369186, "global_step": 23619, "epoch": 222, "lr": 5.849424939119572e-05} +{"train_loss": 0.017856314778327942, "grad_norm": 0.10664316266775131, "global_step": 23620, "epoch": 222, "lr": 5.8491328691239976e-05} +{"train_loss": 0.01714983582496643, "grad_norm": 0.060367003083229065, "global_step": 23621, "epoch": 222, "lr": 5.848840796144941e-05} +{"train_loss": 0.016760386526584625, "grad_norm": 0.09102980047464371, "global_step": 23622, "epoch": 222, "lr": 5.8485487201834255e-05} +{"train_loss": 0.016451280564069748, "grad_norm": 0.08508352935314178, "global_step": 23623, "epoch": 222, "lr": 5.8482566412404785e-05} +{"train_loss": 0.016848541796207428, "grad_norm": 0.11382682621479034, "global_step": 23624, "epoch": 222, "lr": 5.847964559317128e-05} +{"train_loss": 0.014102322980761528, "grad_norm": 0.08523990213871002, "global_step": 23625, "epoch": 222, "lr": 5.847672474414397e-05} +{"train_loss": 0.015165598131716251, "grad_norm": 0.13666725158691406, "global_step": 23626, "epoch": 222, "lr": 5.847380386533315e-05} +{"train_loss": 0.015735723078250885, "grad_norm": 0.07914135605096817, "global_step": 23627, "epoch": 222, "lr": 5.847088295674907e-05} +{"train_loss": 0.01620848849415779, "grad_norm": 0.07962347567081451, "global_step": 23628, "epoch": 222, "lr": 5.8467962018401987e-05} +{"train_loss": 0.0168559942394495, "grad_norm": 0.08251664787530899, "global_step": 23629, "epoch": 222, "lr": 5.846504105030217e-05} +{"train_loss": 0.015901360660791397, "grad_norm": 0.11387446522712708, "global_step": 23630, "epoch": 222, "lr": 5.846212005245988e-05} +{"train_loss": 0.014326936565339565, "grad_norm": 0.07453421503305435, "global_step": 23631, "epoch": 222, "lr": 5.845919902488537e-05} +{"train_loss": 0.01552506908774376, "grad_norm": 0.1424170732498169, "global_step": 23632, "epoch": 222, "lr": 5.8456277967588926e-05} +{"train_loss": 0.01605764776468277, "grad_norm": 0.062274571508169174, "global_step": 23633, "epoch": 222, "lr": 5.845335688058079e-05} +{"train_loss": 0.016872504726052284, "grad_norm": 0.07332976907491684, "global_step": 23634, "epoch": 222, "lr": 5.845043576387123e-05} +{"train_loss": 0.015250210650265217, "grad_norm": 0.07306558638811111, "global_step": 23635, "epoch": 222, "lr": 5.844751461747052e-05} +{"train_loss": 0.01822797954082489, "grad_norm": 0.07404482364654541, "global_step": 23636, "epoch": 222, "lr": 5.844459344138893e-05} +{"train_loss": 0.015768254105493706, "grad_norm": 0.11089994013309479, "global_step": 23637, "epoch": 222, "lr": 5.844167223563669e-05, "val_loss": 0.019640764221549034} +{"train_loss": 0.016977554187178612, "grad_norm": 0.1572345346212387, "global_step": 23638, "epoch": 223, "lr": 5.8438751000224104e-05} +{"train_loss": 0.01647716388106346, "grad_norm": 0.12087187170982361, "global_step": 23639, "epoch": 223, "lr": 5.8435829735161416e-05} +{"train_loss": 0.017157932743430138, "grad_norm": 0.0706942230463028, "global_step": 23640, "epoch": 223, "lr": 5.843290844045886e-05} +{"train_loss": 0.017147382721304893, "grad_norm": 0.08345558494329453, "global_step": 23641, "epoch": 223, "lr": 5.842998711612677e-05} +{"train_loss": 0.015410376712679863, "grad_norm": 0.06355809420347214, "global_step": 23642, "epoch": 223, "lr": 5.842706576217535e-05} +{"train_loss": 0.019478417932987213, "grad_norm": 0.06101405993103981, "global_step": 23643, "epoch": 223, "lr": 5.84241443786149e-05} +{"train_loss": 0.016570765525102615, "grad_norm": 0.06178887188434601, "global_step": 23644, "epoch": 223, "lr": 5.842122296545566e-05} +{"train_loss": 0.01802373304963112, "grad_norm": 0.07938551902770996, "global_step": 23645, "epoch": 223, "lr": 5.8418301522707907e-05} +{"train_loss": 0.017744656652212143, "grad_norm": 0.10164864361286163, "global_step": 23646, "epoch": 223, "lr": 5.8415380050381894e-05} +{"train_loss": 0.015993379056453705, "grad_norm": 0.11814112216234207, "global_step": 23647, "epoch": 223, "lr": 5.841245854848791e-05} +{"train_loss": 0.015145080164074898, "grad_norm": 0.06908135116100311, "global_step": 23648, "epoch": 223, "lr": 5.8409537017036196e-05} +{"train_loss": 0.01273654866963625, "grad_norm": 0.07792531698942184, "global_step": 23649, "epoch": 223, "lr": 5.840661545603703e-05} +{"train_loss": 0.016214072704315186, "grad_norm": 0.07042643427848816, "global_step": 23650, "epoch": 223, "lr": 5.8403693865500677e-05} +{"train_loss": 0.01550205983221531, "grad_norm": 0.09391041100025177, "global_step": 23651, "epoch": 223, "lr": 5.840077224543739e-05} +{"train_loss": 0.01664520800113678, "grad_norm": 0.09864435344934464, "global_step": 23652, "epoch": 223, "lr": 5.839785059585744e-05} +{"train_loss": 0.0137417521327734, "grad_norm": 0.10094129294157028, "global_step": 23653, "epoch": 223, "lr": 5.839492891677112e-05} +{"train_loss": 0.014595911838114262, "grad_norm": 0.07672082632780075, "global_step": 23654, "epoch": 223, "lr": 5.8392007208188635e-05} +{"train_loss": 0.01384340412914753, "grad_norm": 0.09046576917171478, "global_step": 23655, "epoch": 223, "lr": 5.8389085470120306e-05} +{"train_loss": 0.015258613042533398, "grad_norm": 0.06756529211997986, "global_step": 23656, "epoch": 223, "lr": 5.838616370257638e-05} +{"train_loss": 0.01720162108540535, "grad_norm": 0.0988551452755928, "global_step": 23657, "epoch": 223, "lr": 5.838324190556712e-05} +{"train_loss": 0.01739947311580181, "grad_norm": 0.07885131984949112, "global_step": 23658, "epoch": 223, "lr": 5.838032007910279e-05} +{"train_loss": 0.014547747559845448, "grad_norm": 0.08138462156057358, "global_step": 23659, "epoch": 223, "lr": 5.837739822319366e-05} +{"train_loss": 0.014941984787583351, "grad_norm": 0.0935707539319992, "global_step": 23660, "epoch": 223, "lr": 5.8374476337850003e-05} +{"train_loss": 0.016005557030439377, "grad_norm": 0.06218868866562843, "global_step": 23661, "epoch": 223, "lr": 5.8371554423082064e-05} +{"train_loss": 0.014645302668213844, "grad_norm": 0.07643045485019684, "global_step": 23662, "epoch": 223, "lr": 5.836863247890014e-05} +{"train_loss": 0.014995764009654522, "grad_norm": 0.058300718665122986, "global_step": 23663, "epoch": 223, "lr": 5.8365710505314476e-05} +{"train_loss": 0.01706361025571823, "grad_norm": 0.0672922283411026, "global_step": 23664, "epoch": 223, "lr": 5.836278850233533e-05} +{"train_loss": 0.017158064991235733, "grad_norm": 0.08549091219902039, "global_step": 23665, "epoch": 223, "lr": 5.8359866469973e-05} +{"train_loss": 0.016114529222249985, "grad_norm": 0.08460143953561783, "global_step": 23666, "epoch": 223, "lr": 5.835694440823771e-05} +{"train_loss": 0.015363993123173714, "grad_norm": 0.059405211359262466, "global_step": 23667, "epoch": 223, "lr": 5.835402231713978e-05} +{"train_loss": 0.01552066020667553, "grad_norm": 0.07137585431337357, "global_step": 23668, "epoch": 223, "lr": 5.8351100196689444e-05} +{"train_loss": 0.013116885907948017, "grad_norm": 0.12852737307548523, "global_step": 23669, "epoch": 223, "lr": 5.834817804689696e-05} +{"train_loss": 0.017168909311294556, "grad_norm": 0.14611662924289703, "global_step": 23670, "epoch": 223, "lr": 5.834525586777262e-05} +{"train_loss": 0.015727343037724495, "grad_norm": 0.06435041129589081, "global_step": 23671, "epoch": 223, "lr": 5.8342333659326674e-05} +{"train_loss": 0.015485625714063644, "grad_norm": 0.07214899361133575, "global_step": 23672, "epoch": 223, "lr": 5.83394114215694e-05} +{"train_loss": 0.017829861491918564, "grad_norm": 0.07575248181819916, "global_step": 23673, "epoch": 223, "lr": 5.8336489154511064e-05} +{"train_loss": 0.016643188893795013, "grad_norm": 0.12042871862649918, "global_step": 23674, "epoch": 223, "lr": 5.8333566858161924e-05} +{"train_loss": 0.0167703740298748, "grad_norm": 0.07472610473632812, "global_step": 23675, "epoch": 223, "lr": 5.833064453253225e-05} +{"train_loss": 0.01421976089477539, "grad_norm": 0.10269611328840256, "global_step": 23676, "epoch": 223, "lr": 5.832772217763232e-05} +{"train_loss": 0.01576862297952175, "grad_norm": 0.060620155185461044, "global_step": 23677, "epoch": 223, "lr": 5.832479979347241e-05} +{"train_loss": 0.015885207802057266, "grad_norm": 0.08485394716262817, "global_step": 23678, "epoch": 223, "lr": 5.832187738006274e-05} +{"train_loss": 0.01689949259161949, "grad_norm": 0.08235900849103928, "global_step": 23679, "epoch": 223, "lr": 5.831895493741364e-05} +{"train_loss": 0.015597960911691189, "grad_norm": 0.07325784862041473, "global_step": 23680, "epoch": 223, "lr": 5.831603246553534e-05} +{"train_loss": 0.014576844871044159, "grad_norm": 0.14824311435222626, "global_step": 23681, "epoch": 223, "lr": 5.831310996443812e-05} +{"train_loss": 0.015307755209505558, "grad_norm": 0.10584014654159546, "global_step": 23682, "epoch": 223, "lr": 5.831018743413224e-05} +{"train_loss": 0.012299267575144768, "grad_norm": 0.08352874964475632, "global_step": 23683, "epoch": 223, "lr": 5.8307264874627984e-05} +{"train_loss": 0.014621200039982796, "grad_norm": 0.10835009068250656, "global_step": 23684, "epoch": 223, "lr": 5.830434228593562e-05} +{"train_loss": 0.017167745158076286, "grad_norm": 0.10608124732971191, "global_step": 23685, "epoch": 223, "lr": 5.830141966806539e-05} +{"train_loss": 0.014765794388949871, "grad_norm": 0.09801804274320602, "global_step": 23686, "epoch": 223, "lr": 5.8298497021027596e-05} +{"train_loss": 0.014517001807689667, "grad_norm": 0.08356918394565582, "global_step": 23687, "epoch": 223, "lr": 5.829557434483248e-05} +{"train_loss": 0.01783009245991707, "grad_norm": 0.06922710686922073, "global_step": 23688, "epoch": 223, "lr": 5.829265163949032e-05} +{"train_loss": 0.01581132970750332, "grad_norm": 0.08824290335178375, "global_step": 23689, "epoch": 223, "lr": 5.828972890501142e-05} +{"train_loss": 0.01751658320426941, "grad_norm": 0.06581234931945801, "global_step": 23690, "epoch": 223, "lr": 5.828680614140599e-05} +{"train_loss": 0.014016842469573021, "grad_norm": 0.06488309055566788, "global_step": 23691, "epoch": 223, "lr": 5.828388334868433e-05} +{"train_loss": 0.017512701451778412, "grad_norm": 0.0638142079114914, "global_step": 23692, "epoch": 223, "lr": 5.828096052685672e-05} +{"train_loss": 0.016343500465154648, "grad_norm": 0.06458062678575516, "global_step": 23693, "epoch": 223, "lr": 5.82780376759334e-05} +{"train_loss": 0.012632942758500576, "grad_norm": 0.07326045632362366, "global_step": 23694, "epoch": 223, "lr": 5.8275114795924665e-05} +{"train_loss": 0.016648385673761368, "grad_norm": 0.05824951454997063, "global_step": 23695, "epoch": 223, "lr": 5.827219188684077e-05} +{"train_loss": 0.01618175581097603, "grad_norm": 0.05920727550983429, "global_step": 23696, "epoch": 223, "lr": 5.826926894869199e-05} +{"train_loss": 0.013261744752526283, "grad_norm": 0.08212126791477203, "global_step": 23697, "epoch": 223, "lr": 5.8266345981488604e-05} +{"train_loss": 0.0144696906208992, "grad_norm": 0.05865243822336197, "global_step": 23698, "epoch": 223, "lr": 5.8263422985240856e-05} +{"train_loss": 0.013733774423599243, "grad_norm": 0.06796275079250336, "global_step": 23699, "epoch": 223, "lr": 5.826049995995905e-05} +{"train_loss": 0.01507798582315445, "grad_norm": 0.05980096384882927, "global_step": 23700, "epoch": 223, "lr": 5.825757690565343e-05} +{"train_loss": 0.014712532050907612, "grad_norm": 0.0848148912191391, "global_step": 23701, "epoch": 223, "lr": 5.825465382233428e-05} +{"train_loss": 0.014719866216182709, "grad_norm": 0.0671469047665596, "global_step": 23702, "epoch": 223, "lr": 5.825173071001186e-05} +{"train_loss": 0.0155029296875, "grad_norm": 0.05874180048704147, "global_step": 23703, "epoch": 223, "lr": 5.824880756869646e-05} +{"train_loss": 0.017089154571294785, "grad_norm": 0.09782818704843521, "global_step": 23704, "epoch": 223, "lr": 5.824588439839833e-05} +{"train_loss": 0.014181924983859062, "grad_norm": 0.09913099557161331, "global_step": 23705, "epoch": 223, "lr": 5.824296119912775e-05} +{"train_loss": 0.015904022380709648, "grad_norm": 0.11755447834730148, "global_step": 23706, "epoch": 223, "lr": 5.824003797089499e-05} +{"train_loss": 0.017029881477355957, "grad_norm": 0.073454350233078, "global_step": 23707, "epoch": 223, "lr": 5.8237114713710304e-05} +{"train_loss": 0.016716746613383293, "grad_norm": 0.112062968313694, "global_step": 23708, "epoch": 223, "lr": 5.8234191427583995e-05} +{"train_loss": 0.015328310430049896, "grad_norm": 0.09463812410831451, "global_step": 23709, "epoch": 223, "lr": 5.8231268112526326e-05} +{"train_loss": 0.014141712337732315, "grad_norm": 0.07486895471811295, "global_step": 23710, "epoch": 223, "lr": 5.822834476854755e-05} +{"train_loss": 0.015386251732707024, "grad_norm": 0.07129359990358353, "global_step": 23711, "epoch": 223, "lr": 5.8225421395657944e-05} +{"train_loss": 0.015490061603486538, "grad_norm": 0.05458788573741913, "global_step": 23712, "epoch": 223, "lr": 5.822249799386779e-05} +{"train_loss": 0.01437021791934967, "grad_norm": 0.08479182422161102, "global_step": 23713, "epoch": 223, "lr": 5.821957456318735e-05} +{"train_loss": 0.01795317605137825, "grad_norm": 0.13491584360599518, "global_step": 23714, "epoch": 223, "lr": 5.8216651103626906e-05} +{"train_loss": 0.016838539391756058, "grad_norm": 0.06989605724811554, "global_step": 23715, "epoch": 223, "lr": 5.8213727615196725e-05} +{"train_loss": 0.018102284520864487, "grad_norm": 0.0653291791677475, "global_step": 23716, "epoch": 223, "lr": 5.821080409790707e-05} +{"train_loss": 0.014660377986729145, "grad_norm": 0.1558806151151657, "global_step": 23717, "epoch": 223, "lr": 5.820788055176821e-05} +{"train_loss": 0.014231624081730843, "grad_norm": 0.06562559306621552, "global_step": 23718, "epoch": 223, "lr": 5.820495697679046e-05} +{"train_loss": 0.013504499569535255, "grad_norm": 0.052855029702186584, "global_step": 23719, "epoch": 223, "lr": 5.820203337298403e-05} +{"train_loss": 0.013385685160756111, "grad_norm": 0.06541462242603302, "global_step": 23720, "epoch": 223, "lr": 5.8199109740359225e-05} +{"train_loss": 0.013643654994666576, "grad_norm": 0.08678042888641357, "global_step": 23721, "epoch": 223, "lr": 5.819618607892634e-05} +{"train_loss": 0.01603836938738823, "grad_norm": 0.08305364102125168, "global_step": 23722, "epoch": 223, "lr": 5.819326238869559e-05} +{"train_loss": 0.014212409034371376, "grad_norm": 0.07975053787231445, "global_step": 23723, "epoch": 223, "lr": 5.8190338669677294e-05} +{"train_loss": 0.013460551388561726, "grad_norm": 0.0662568211555481, "global_step": 23724, "epoch": 223, "lr": 5.81874149218817e-05} +{"train_loss": 0.016206342726945877, "grad_norm": 0.05826586112380028, "global_step": 23725, "epoch": 223, "lr": 5.81844911453191e-05} +{"train_loss": 0.015605950728058815, "grad_norm": 0.06326355040073395, "global_step": 23726, "epoch": 223, "lr": 5.818156733999975e-05} +{"train_loss": 0.015824204310774803, "grad_norm": 0.09499285370111465, "global_step": 23727, "epoch": 223, "lr": 5.817864350593394e-05} +{"train_loss": 0.018548931926488876, "grad_norm": 0.14879459142684937, "global_step": 23728, "epoch": 223, "lr": 5.8175719643131935e-05} +{"train_loss": 0.016679182648658752, "grad_norm": 0.06865821778774261, "global_step": 23729, "epoch": 223, "lr": 5.817279575160398e-05} +{"train_loss": 0.01641881838440895, "grad_norm": 0.06788735836744308, "global_step": 23730, "epoch": 223, "lr": 5.816987183136041e-05} +{"train_loss": 0.018193762749433517, "grad_norm": 0.08737003058195114, "global_step": 23731, "epoch": 223, "lr": 5.816694788241145e-05} +{"train_loss": 0.014700603671371937, "grad_norm": 0.05177370086312294, "global_step": 23732, "epoch": 223, "lr": 5.816402390476738e-05} +{"train_loss": 0.01857246644794941, "grad_norm": 0.09176532179117203, "global_step": 23733, "epoch": 223, "lr": 5.81610998984385e-05} +{"train_loss": 0.018185600638389587, "grad_norm": 0.077640600502491, "global_step": 23734, "epoch": 223, "lr": 5.8158175863435035e-05} +{"train_loss": 0.013554573059082031, "grad_norm": 0.0980391576886177, "global_step": 23735, "epoch": 223, "lr": 5.815525179976731e-05} +{"train_loss": 0.015705611556768417, "grad_norm": 0.11790197342634201, "global_step": 23736, "epoch": 223, "lr": 5.8152327707445566e-05} +{"train_loss": 0.017542509362101555, "grad_norm": 0.08521036803722382, "global_step": 23737, "epoch": 223, "lr": 5.814940358648009e-05} +{"train_loss": 0.017397958785295486, "grad_norm": 0.07629265636205673, "global_step": 23738, "epoch": 223, "lr": 5.814647943688116e-05} +{"train_loss": 0.015019508078694344, "grad_norm": 0.0856679156422615, "global_step": 23739, "epoch": 223, "lr": 5.814355525865904e-05} +{"train_loss": 0.01672343537211418, "grad_norm": 0.06154319643974304, "global_step": 23740, "epoch": 223, "lr": 5.814063105182401e-05} +{"train_loss": 0.01726989448070526, "grad_norm": 0.07363653928041458, "global_step": 23741, "epoch": 223, "lr": 5.813770681638634e-05} +{"train_loss": 0.015375673770904541, "grad_norm": 0.0724586620926857, "global_step": 23742, "epoch": 223, "lr": 5.8134782552356306e-05} +{"train_loss": 0.01577398913719182, "grad_norm": 0.09627464413642883, "global_step": 23743, "epoch": 223, "lr": 5.813185825974419e-05, "val_loss": 0.019254228100180626} +{"train_loss": 0.01469099149107933, "grad_norm": 0.05309363454580307, "global_step": 23744, "epoch": 224, "lr": 5.812893393856026e-05} +{"train_loss": 0.01668887585401535, "grad_norm": 0.08156008273363113, "global_step": 23745, "epoch": 224, "lr": 5.812600958881478e-05} +{"train_loss": 0.018788328394293785, "grad_norm": 0.09687155485153198, "global_step": 23746, "epoch": 224, "lr": 5.8123085210518036e-05} +{"train_loss": 0.013269070535898209, "grad_norm": 0.0695481151342392, "global_step": 23747, "epoch": 224, "lr": 5.812016080368032e-05} +{"train_loss": 0.015789691358804703, "grad_norm": 0.08167697489261627, "global_step": 23748, "epoch": 224, "lr": 5.811723636831188e-05} +{"train_loss": 0.01379468571394682, "grad_norm": 0.06286647915840149, "global_step": 23749, "epoch": 224, "lr": 5.8114311904423004e-05} +{"train_loss": 0.01371740736067295, "grad_norm": 0.09701699763536453, "global_step": 23750, "epoch": 224, "lr": 5.8111387412023976e-05} +{"train_loss": 0.01519034430384636, "grad_norm": 0.06781645864248276, "global_step": 23751, "epoch": 224, "lr": 5.8108462891125034e-05} +{"train_loss": 0.016038862988352776, "grad_norm": 0.06725603342056274, "global_step": 23752, "epoch": 224, "lr": 5.8105538341736485e-05} +{"train_loss": 0.017641790211200714, "grad_norm": 0.11728107184171677, "global_step": 23753, "epoch": 224, "lr": 5.810261376386862e-05} +{"train_loss": 0.015484873205423355, "grad_norm": 0.0553005151450634, "global_step": 23754, "epoch": 224, "lr": 5.809968915753168e-05} +{"train_loss": 0.014746329747140408, "grad_norm": 0.10041022300720215, "global_step": 23755, "epoch": 224, "lr": 5.809676452273596e-05} +{"train_loss": 0.015207993797957897, "grad_norm": 0.09113379567861557, "global_step": 23756, "epoch": 224, "lr": 5.809383985949173e-05} +{"train_loss": 0.016864683479070663, "grad_norm": 0.06284619867801666, "global_step": 23757, "epoch": 224, "lr": 5.809091516780926e-05} +{"train_loss": 0.012729332782328129, "grad_norm": 0.12588901817798615, "global_step": 23758, "epoch": 224, "lr": 5.808799044769883e-05} +{"train_loss": 0.016143877059221268, "grad_norm": 0.08365723490715027, "global_step": 23759, "epoch": 224, "lr": 5.808506569917074e-05} +{"train_loss": 0.0162217877805233, "grad_norm": 0.10599202662706375, "global_step": 23760, "epoch": 224, "lr": 5.808214092223523e-05} +{"train_loss": 0.016309823840856552, "grad_norm": 0.08220252394676208, "global_step": 23761, "epoch": 224, "lr": 5.8079216116902576e-05} +{"train_loss": 0.015935655683279037, "grad_norm": 0.09807039052248001, "global_step": 23762, "epoch": 224, "lr": 5.8076291283183105e-05} +{"train_loss": 0.01653420552611351, "grad_norm": 0.11764761060476303, "global_step": 23763, "epoch": 224, "lr": 5.8073366421087026e-05} +{"train_loss": 0.016077857464551926, "grad_norm": 0.04980666562914848, "global_step": 23764, "epoch": 224, "lr": 5.807044153062466e-05} +{"train_loss": 0.019444413483142853, "grad_norm": 0.14032961428165436, "global_step": 23765, "epoch": 224, "lr": 5.806751661180627e-05} +{"train_loss": 0.016853969544172287, "grad_norm": 0.05515296757221222, "global_step": 23766, "epoch": 224, "lr": 5.8064591664642144e-05} +{"train_loss": 0.012140586040914059, "grad_norm": 0.05997854843735695, "global_step": 23767, "epoch": 224, "lr": 5.8061666689142544e-05} +{"train_loss": 0.01637420989573002, "grad_norm": 0.07064563035964966, "global_step": 23768, "epoch": 224, "lr": 5.805874168531776e-05} +{"train_loss": 0.01769111678004265, "grad_norm": 0.09870839864015579, "global_step": 23769, "epoch": 224, "lr": 5.8055816653178064e-05} +{"train_loss": 0.015509134158492088, "grad_norm": 0.05344617739319801, "global_step": 23770, "epoch": 224, "lr": 5.8052891592733724e-05} +{"train_loss": 0.01673867553472519, "grad_norm": 0.08929353952407837, "global_step": 23771, "epoch": 224, "lr": 5.804996650399502e-05} +{"train_loss": 0.014109673909842968, "grad_norm": 0.06502310186624527, "global_step": 23772, "epoch": 224, "lr": 5.804704138697225e-05} +{"train_loss": 0.016890281811356544, "grad_norm": 0.054357681423425674, "global_step": 23773, "epoch": 224, "lr": 5.804411624167565e-05} +{"train_loss": 0.014461306855082512, "grad_norm": 0.1019696518778801, "global_step": 23774, "epoch": 224, "lr": 5.8041191068115555e-05} +{"train_loss": 0.016491789370775223, "grad_norm": 0.07126767933368683, "global_step": 23775, "epoch": 224, "lr": 5.803826586630219e-05} +{"train_loss": 0.01516305934637785, "grad_norm": 0.06343577802181244, "global_step": 23776, "epoch": 224, "lr": 5.8035340636245863e-05} +{"train_loss": 0.015407312661409378, "grad_norm": 0.07630612701177597, "global_step": 23777, "epoch": 224, "lr": 5.803241537795684e-05} +{"train_loss": 0.01625490002334118, "grad_norm": 0.09920114278793335, "global_step": 23778, "epoch": 224, "lr": 5.80294900914454e-05} +{"train_loss": 0.015946246683597565, "grad_norm": 0.09092973172664642, "global_step": 23779, "epoch": 224, "lr": 5.802656477672184e-05} +{"train_loss": 0.01595890335738659, "grad_norm": 0.09201785922050476, "global_step": 23780, "epoch": 224, "lr": 5.802363943379641e-05} +{"train_loss": 0.013335063122212887, "grad_norm": 0.09549643099308014, "global_step": 23781, "epoch": 224, "lr": 5.8020714062679405e-05} +{"train_loss": 0.015606213361024857, "grad_norm": 0.053645357489585876, "global_step": 23782, "epoch": 224, "lr": 5.801778866338109e-05} +{"train_loss": 0.016859082505106926, "grad_norm": 0.09931798279285431, "global_step": 23783, "epoch": 224, "lr": 5.8014863235911765e-05} +{"train_loss": 0.016530726104974747, "grad_norm": 0.08154395967721939, "global_step": 23784, "epoch": 224, "lr": 5.801193778028169e-05} +{"train_loss": 0.016451381146907806, "grad_norm": 0.05657649412751198, "global_step": 23785, "epoch": 224, "lr": 5.800901229650113e-05} +{"train_loss": 0.012781811878085136, "grad_norm": 0.07843425124883652, "global_step": 23786, "epoch": 224, "lr": 5.800608678458042e-05} +{"train_loss": 0.015214436687529087, "grad_norm": 0.07951373606920242, "global_step": 23787, "epoch": 224, "lr": 5.8003161244529766e-05} +{"train_loss": 0.0164362583309412, "grad_norm": 0.07947102934122086, "global_step": 23788, "epoch": 224, "lr": 5.8000235676359505e-05} +{"train_loss": 0.01266029104590416, "grad_norm": 0.059825602918863297, "global_step": 23789, "epoch": 224, "lr": 5.799731008007991e-05} +{"train_loss": 0.017871331423521042, "grad_norm": 0.12832999229431152, "global_step": 23790, "epoch": 224, "lr": 5.799438445570122e-05} +{"train_loss": 0.016268063336610794, "grad_norm": 0.11954671144485474, "global_step": 23791, "epoch": 224, "lr": 5.799145880323374e-05} +{"train_loss": 0.013325569219887257, "grad_norm": 0.07422857731580734, "global_step": 23792, "epoch": 224, "lr": 5.798853312268776e-05} +{"train_loss": 0.017371462658047676, "grad_norm": 0.0656687393784523, "global_step": 23793, "epoch": 224, "lr": 5.798560741407355e-05} +{"train_loss": 0.014583941549062729, "grad_norm": 0.05300338566303253, "global_step": 23794, "epoch": 224, "lr": 5.7982681677401384e-05} +{"train_loss": 0.01800435408949852, "grad_norm": 0.06971060484647751, "global_step": 23795, "epoch": 224, "lr": 5.797975591268155e-05} +{"train_loss": 0.018267758190631866, "grad_norm": 0.05320161581039429, "global_step": 23796, "epoch": 224, "lr": 5.797683011992432e-05} +{"train_loss": 0.017424602061510086, "grad_norm": 0.0771561861038208, "global_step": 23797, "epoch": 224, "lr": 5.797390429913998e-05} +{"train_loss": 0.013644086197018623, "grad_norm": 0.05334946885704994, "global_step": 23798, "epoch": 224, "lr": 5.7970978450338806e-05} +{"train_loss": 0.015340401791036129, "grad_norm": 0.0525130033493042, "global_step": 23799, "epoch": 224, "lr": 5.7968052573531084e-05} +{"train_loss": 0.01784159056842327, "grad_norm": 0.0704018622636795, "global_step": 23800, "epoch": 224, "lr": 5.796512666872709e-05} +{"train_loss": 0.01611243188381195, "grad_norm": 0.1287936419248581, "global_step": 23801, "epoch": 224, "lr": 5.79622007359371e-05} +{"train_loss": 0.013270479626953602, "grad_norm": 0.07745173573493958, "global_step": 23802, "epoch": 224, "lr": 5.795927477517139e-05} +{"train_loss": 0.01716557890176773, "grad_norm": 0.08092548698186874, "global_step": 23803, "epoch": 224, "lr": 5.795634878644027e-05} +{"train_loss": 0.015922071412205696, "grad_norm": 0.08408236503601074, "global_step": 23804, "epoch": 224, "lr": 5.7953422769753984e-05} +{"train_loss": 0.016162268817424774, "grad_norm": 0.08352901041507721, "global_step": 23805, "epoch": 224, "lr": 5.795049672512284e-05} +{"train_loss": 0.014268247410655022, "grad_norm": 0.15234674513339996, "global_step": 23806, "epoch": 224, "lr": 5.79475706525571e-05} +{"train_loss": 0.014616936445236206, "grad_norm": 0.054668549448251724, "global_step": 23807, "epoch": 224, "lr": 5.794464455206706e-05} +{"train_loss": 0.015120483934879303, "grad_norm": 0.06476598978042603, "global_step": 23808, "epoch": 224, "lr": 5.794171842366301e-05} +{"train_loss": 0.013189507648348808, "grad_norm": 0.06201820448040962, "global_step": 23809, "epoch": 224, "lr": 5.793879226735519e-05} +{"train_loss": 0.018000373616814613, "grad_norm": 0.060183048248291016, "global_step": 23810, "epoch": 224, "lr": 5.793586608315391e-05} +{"train_loss": 0.014984449371695518, "grad_norm": 0.10183580219745636, "global_step": 23811, "epoch": 224, "lr": 5.793293987106946e-05} +{"train_loss": 0.01572815701365471, "grad_norm": 0.09576442092657089, "global_step": 23812, "epoch": 224, "lr": 5.7930013631112104e-05} +{"train_loss": 0.014123070985078812, "grad_norm": 0.09337266534566879, "global_step": 23813, "epoch": 224, "lr": 5.792708736329212e-05} +{"train_loss": 0.014373017475008965, "grad_norm": 0.05115852877497673, "global_step": 23814, "epoch": 224, "lr": 5.7924161067619796e-05} +{"train_loss": 0.015167199075222015, "grad_norm": 0.1104893609881401, "global_step": 23815, "epoch": 224, "lr": 5.792123474410544e-05} +{"train_loss": 0.016682006418704987, "grad_norm": 0.08156517148017883, "global_step": 23816, "epoch": 224, "lr": 5.791830839275928e-05} +{"train_loss": 0.015587709844112396, "grad_norm": 0.10714569687843323, "global_step": 23817, "epoch": 224, "lr": 5.7915382013591635e-05} +{"train_loss": 0.015418123453855515, "grad_norm": 0.07995390146970749, "global_step": 23818, "epoch": 224, "lr": 5.7912455606612805e-05} +{"train_loss": 0.01607457548379898, "grad_norm": 0.0675269141793251, "global_step": 23819, "epoch": 224, "lr": 5.790952917183302e-05} +{"train_loss": 0.019732780754566193, "grad_norm": 0.09810709208250046, "global_step": 23820, "epoch": 224, "lr": 5.79066027092626e-05} +{"train_loss": 0.01626347377896309, "grad_norm": 0.09328339248895645, "global_step": 23821, "epoch": 224, "lr": 5.7903676218911817e-05} +{"train_loss": 0.013893812894821167, "grad_norm": 0.05418448895215988, "global_step": 23822, "epoch": 224, "lr": 5.790074970079095e-05} +{"train_loss": 0.0157175213098526, "grad_norm": 0.08640122413635254, "global_step": 23823, "epoch": 224, "lr": 5.789782315491028e-05} +{"train_loss": 0.014653405174612999, "grad_norm": 0.10509663075208664, "global_step": 23824, "epoch": 224, "lr": 5.78948965812801e-05} +{"train_loss": 0.016249990090727806, "grad_norm": 0.06754902005195618, "global_step": 23825, "epoch": 224, "lr": 5.789196997991069e-05} +{"train_loss": 0.016057007014751434, "grad_norm": 0.13585436344146729, "global_step": 23826, "epoch": 224, "lr": 5.7889043350812314e-05} +{"train_loss": 0.01534693781286478, "grad_norm": 0.06385637819766998, "global_step": 23827, "epoch": 224, "lr": 5.7886116693995295e-05} +{"train_loss": 0.017134107649326324, "grad_norm": 0.11636575311422348, "global_step": 23828, "epoch": 224, "lr": 5.788319000946987e-05} +{"train_loss": 0.016558581963181496, "grad_norm": 0.0757041871547699, "global_step": 23829, "epoch": 224, "lr": 5.788026329724635e-05} +{"train_loss": 0.013583973050117493, "grad_norm": 0.08222400397062302, "global_step": 23830, "epoch": 224, "lr": 5.787733655733503e-05} +{"train_loss": 0.01673598401248455, "grad_norm": 0.08760453760623932, "global_step": 23831, "epoch": 224, "lr": 5.787440978974615e-05} +{"train_loss": 0.017796499654650688, "grad_norm": 0.05944807827472687, "global_step": 23832, "epoch": 224, "lr": 5.7871482994490034e-05} +{"train_loss": 0.013764956034719944, "grad_norm": 0.09905920922756195, "global_step": 23833, "epoch": 224, "lr": 5.7868556171576935e-05} +{"train_loss": 0.019331082701683044, "grad_norm": 0.06110917776823044, "global_step": 23834, "epoch": 224, "lr": 5.786562932101717e-05} +{"train_loss": 0.01629457250237465, "grad_norm": 0.06197904422879219, "global_step": 23835, "epoch": 224, "lr": 5.7862702442820994e-05} +{"train_loss": 0.014098165556788445, "grad_norm": 0.05819772183895111, "global_step": 23836, "epoch": 224, "lr": 5.78597755369987e-05} +{"train_loss": 0.016161076724529266, "grad_norm": 0.0938912034034729, "global_step": 23837, "epoch": 224, "lr": 5.785684860356057e-05} +{"train_loss": 0.014428095892071724, "grad_norm": 0.11488551646471024, "global_step": 23838, "epoch": 224, "lr": 5.7853921642516885e-05} +{"train_loss": 0.015952659770846367, "grad_norm": 0.07654184848070145, "global_step": 23839, "epoch": 224, "lr": 5.785099465387797e-05} +{"train_loss": 0.014986097812652588, "grad_norm": 0.07190265506505966, "global_step": 23840, "epoch": 224, "lr": 5.784806763765405e-05} +{"train_loss": 0.01781104877591133, "grad_norm": 0.09887873381376266, "global_step": 23841, "epoch": 224, "lr": 5.784514059385542e-05} +{"train_loss": 0.017314329743385315, "grad_norm": 0.1447940617799759, "global_step": 23842, "epoch": 224, "lr": 5.78422135224924e-05} +{"train_loss": 0.01498924195766449, "grad_norm": 0.07939957082271576, "global_step": 23843, "epoch": 224, "lr": 5.783928642357523e-05} +{"train_loss": 0.01604440249502659, "grad_norm": 0.08121614903211594, "global_step": 23844, "epoch": 224, "lr": 5.783635929711424e-05} +{"train_loss": 0.015013515949249268, "grad_norm": 0.09485858678817749, "global_step": 23845, "epoch": 224, "lr": 5.783343214311967e-05} +{"train_loss": 0.016177158802747726, "grad_norm": 0.06061017885804176, "global_step": 23846, "epoch": 224, "lr": 5.7830504961601837e-05} +{"train_loss": 0.01367099303752184, "grad_norm": 0.05318930745124817, "global_step": 23847, "epoch": 224, "lr": 5.7827577752571025e-05} +{"train_loss": 0.016861196607351303, "grad_norm": 0.056398387998342514, "global_step": 23848, "epoch": 224, "lr": 5.7824650516037495e-05} +{"train_loss": 0.015760924693478166, "grad_norm": 0.0773148313164711, "global_step": 23849, "epoch": 224, "lr": 5.782172325201155e-05, "val_loss": 0.01871657557785511} +{"train_loss": 0.016230862587690353, "grad_norm": 0.17787431180477142, "global_step": 23850, "epoch": 225, "lr": 5.781879596050347e-05} +{"train_loss": 0.01847606524825096, "grad_norm": 0.07813715934753418, "global_step": 23851, "epoch": 225, "lr": 5.781586864152354e-05} +{"train_loss": 0.015058944001793861, "grad_norm": 0.06104031577706337, "global_step": 23852, "epoch": 225, "lr": 5.7812941295082056e-05} +{"train_loss": 0.015173608437180519, "grad_norm": 0.0652000829577446, "global_step": 23853, "epoch": 225, "lr": 5.7810013921189285e-05} +{"train_loss": 0.014936503022909164, "grad_norm": 0.06461230665445328, "global_step": 23854, "epoch": 225, "lr": 5.780708651985553e-05} +{"train_loss": 0.014749756082892418, "grad_norm": 0.0645160898566246, "global_step": 23855, "epoch": 225, "lr": 5.780415909109105e-05} +{"train_loss": 0.01718316227197647, "grad_norm": 0.0832105427980423, "global_step": 23856, "epoch": 225, "lr": 5.780123163490617e-05} +{"train_loss": 0.015396280214190483, "grad_norm": 0.08743540942668915, "global_step": 23857, "epoch": 225, "lr": 5.7798304151311136e-05} +{"train_loss": 0.01477527804672718, "grad_norm": 0.08808553963899612, "global_step": 23858, "epoch": 225, "lr": 5.779537664031626e-05} +{"train_loss": 0.014498207718133926, "grad_norm": 0.13935142755508423, "global_step": 23859, "epoch": 225, "lr": 5.7792449101931836e-05} +{"train_loss": 0.014413970522582531, "grad_norm": 0.06828417629003525, "global_step": 23860, "epoch": 225, "lr": 5.778952153616811e-05} +{"train_loss": 0.015286085195839405, "grad_norm": 0.12798745930194855, "global_step": 23861, "epoch": 225, "lr": 5.7786593943035405e-05} +{"train_loss": 0.015119528397917747, "grad_norm": 0.052767932415008545, "global_step": 23862, "epoch": 225, "lr": 5.7783666322543994e-05} +{"train_loss": 0.013349689543247223, "grad_norm": 0.1439286321401596, "global_step": 23863, "epoch": 225, "lr": 5.778073867470416e-05} +{"train_loss": 0.016041092574596405, "grad_norm": 0.0799546167254448, "global_step": 23864, "epoch": 225, "lr": 5.777781099952621e-05} +{"train_loss": 0.015578383579850197, "grad_norm": 0.10008978098630905, "global_step": 23865, "epoch": 225, "lr": 5.77748832970204e-05} +{"train_loss": 0.01694326102733612, "grad_norm": 0.12521786987781525, "global_step": 23866, "epoch": 225, "lr": 5.777195556719703e-05} +{"train_loss": 0.01636185497045517, "grad_norm": 0.06828297674655914, "global_step": 23867, "epoch": 225, "lr": 5.776902781006639e-05} +{"train_loss": 0.016382453963160515, "grad_norm": 0.04968775808811188, "global_step": 23868, "epoch": 225, "lr": 5.776610002563878e-05} +{"train_loss": 0.012798848561942577, "grad_norm": 0.055372558534145355, "global_step": 23869, "epoch": 225, "lr": 5.776317221392445e-05} +{"train_loss": 0.01664995402097702, "grad_norm": 0.07391456514596939, "global_step": 23870, "epoch": 225, "lr": 5.7760244374933717e-05} +{"train_loss": 0.01706872135400772, "grad_norm": 0.0636802539229393, "global_step": 23871, "epoch": 225, "lr": 5.7757316508676875e-05} +{"train_loss": 0.017482707276940346, "grad_norm": 0.07954297959804535, "global_step": 23872, "epoch": 225, "lr": 5.775438861516417e-05} +{"train_loss": 0.014090381562709808, "grad_norm": 0.08208087831735611, "global_step": 23873, "epoch": 225, "lr": 5.775146069440593e-05} +{"train_loss": 0.01600588671863079, "grad_norm": 0.07153093814849854, "global_step": 23874, "epoch": 225, "lr": 5.774853274641243e-05} +{"train_loss": 0.01369630265980959, "grad_norm": 0.06401733309030533, "global_step": 23875, "epoch": 225, "lr": 5.774560477119395e-05} +{"train_loss": 0.019354768097400665, "grad_norm": 0.08793262392282486, "global_step": 23876, "epoch": 225, "lr": 5.774267676876079e-05} +{"train_loss": 0.016544083133339882, "grad_norm": 0.1609826683998108, "global_step": 23877, "epoch": 225, "lr": 5.773974873912322e-05} +{"train_loss": 0.017263885587453842, "grad_norm": 0.10071050375699997, "global_step": 23878, "epoch": 225, "lr": 5.7736820682291545e-05} +{"train_loss": 0.014433469623327255, "grad_norm": 0.09155695885419846, "global_step": 23879, "epoch": 225, "lr": 5.773389259827604e-05} +{"train_loss": 0.014417020604014397, "grad_norm": 0.08190181106328964, "global_step": 23880, "epoch": 225, "lr": 5.773096448708702e-05} +{"train_loss": 0.018746063113212585, "grad_norm": 0.11132021993398666, "global_step": 23881, "epoch": 225, "lr": 5.772803634873474e-05} +{"train_loss": 0.014972271397709846, "grad_norm": 0.08687742799520493, "global_step": 23882, "epoch": 225, "lr": 5.772510818322949e-05} +{"train_loss": 0.017920199781656265, "grad_norm": 0.11974186450242996, "global_step": 23883, "epoch": 225, "lr": 5.772217999058159e-05} +{"train_loss": 0.013925239443778992, "grad_norm": 0.07701143622398376, "global_step": 23884, "epoch": 225, "lr": 5.77192517708013e-05} +{"train_loss": 0.01450041402131319, "grad_norm": 0.06286855787038803, "global_step": 23885, "epoch": 225, "lr": 5.771632352389891e-05} +{"train_loss": 0.014164713211357594, "grad_norm": 0.07816126942634583, "global_step": 23886, "epoch": 225, "lr": 5.7713395249884716e-05} +{"train_loss": 0.01614038273692131, "grad_norm": 0.07398059219121933, "global_step": 23887, "epoch": 225, "lr": 5.7710466948769014e-05} +{"train_loss": 0.014422244392335415, "grad_norm": 0.0748295709490776, "global_step": 23888, "epoch": 225, "lr": 5.770753862056209e-05} +{"train_loss": 0.013823676854372025, "grad_norm": 0.05878489464521408, "global_step": 23889, "epoch": 225, "lr": 5.770461026527421e-05} +{"train_loss": 0.014087710529565811, "grad_norm": 0.08535976707935333, "global_step": 23890, "epoch": 225, "lr": 5.7701681882915694e-05} +{"train_loss": 0.015766365453600883, "grad_norm": 0.06965283304452896, "global_step": 23891, "epoch": 225, "lr": 5.76987534734968e-05} +{"train_loss": 0.015269081108272076, "grad_norm": 0.13138364255428314, "global_step": 23892, "epoch": 225, "lr": 5.769582503702785e-05} +{"train_loss": 0.014447207562625408, "grad_norm": 0.08208583295345306, "global_step": 23893, "epoch": 225, "lr": 5.7692896573519104e-05} +{"train_loss": 0.01790861040353775, "grad_norm": 0.09690213948488235, "global_step": 23894, "epoch": 225, "lr": 5.7689968082980874e-05} +{"train_loss": 0.0149283641949296, "grad_norm": 0.05020574480295181, "global_step": 23895, "epoch": 225, "lr": 5.768703956542345e-05} +{"train_loss": 0.018442906439304352, "grad_norm": 0.09470648318529129, "global_step": 23896, "epoch": 225, "lr": 5.768411102085709e-05} +{"train_loss": 0.016016094014048576, "grad_norm": 0.055102936923503876, "global_step": 23897, "epoch": 225, "lr": 5.768118244929212e-05} +{"train_loss": 0.01577366515994072, "grad_norm": 0.07600676268339157, "global_step": 23898, "epoch": 225, "lr": 5.7678253850738816e-05} +{"train_loss": 0.013433839194476604, "grad_norm": 0.08041119575500488, "global_step": 23899, "epoch": 225, "lr": 5.7675325225207455e-05} +{"train_loss": 0.013574883341789246, "grad_norm": 0.08458106219768524, "global_step": 23900, "epoch": 225, "lr": 5.767239657270837e-05} +{"train_loss": 0.01610175147652626, "grad_norm": 0.08192694187164307, "global_step": 23901, "epoch": 225, "lr": 5.7669467893251785e-05} +{"train_loss": 0.01736076921224594, "grad_norm": 0.10376357287168503, "global_step": 23902, "epoch": 225, "lr": 5.7666539186848036e-05} +{"train_loss": 0.013269484974443913, "grad_norm": 0.06914038956165314, "global_step": 23903, "epoch": 225, "lr": 5.766361045350741e-05} +{"train_loss": 0.018406644463539124, "grad_norm": 0.07048266381025314, "global_step": 23904, "epoch": 225, "lr": 5.766068169324018e-05} +{"train_loss": 0.016636855900287628, "grad_norm": 0.1013534739613533, "global_step": 23905, "epoch": 225, "lr": 5.7657752906056653e-05} +{"train_loss": 0.01505117304623127, "grad_norm": 0.1681012213230133, "global_step": 23906, "epoch": 225, "lr": 5.7654824091967105e-05} +{"train_loss": 0.015362238511443138, "grad_norm": 0.06459256261587143, "global_step": 23907, "epoch": 225, "lr": 5.765189525098185e-05} +{"train_loss": 0.016446180641651154, "grad_norm": 0.0875392034649849, "global_step": 23908, "epoch": 225, "lr": 5.7648966383111135e-05} +{"train_loss": 0.013207407668232918, "grad_norm": 0.057558950036764145, "global_step": 23909, "epoch": 225, "lr": 5.76460374883653e-05} +{"train_loss": 0.013563822954893112, "grad_norm": 0.05057445541024208, "global_step": 23910, "epoch": 225, "lr": 5.764310856675461e-05} +{"train_loss": 0.018597116693854332, "grad_norm": 0.06342824548482895, "global_step": 23911, "epoch": 225, "lr": 5.7640179618289355e-05} +{"train_loss": 0.015855520963668823, "grad_norm": 0.10697009414434433, "global_step": 23912, "epoch": 225, "lr": 5.7637250642979845e-05} +{"train_loss": 0.01807507686316967, "grad_norm": 0.07644376158714294, "global_step": 23913, "epoch": 225, "lr": 5.7634321640836345e-05} +{"train_loss": 0.016208089888095856, "grad_norm": 0.06384764611721039, "global_step": 23914, "epoch": 225, "lr": 5.763139261186916e-05} +{"train_loss": 0.015922628343105316, "grad_norm": 0.08820055425167084, "global_step": 23915, "epoch": 225, "lr": 5.7628463556088586e-05} +{"train_loss": 0.01426267996430397, "grad_norm": 0.06450007855892181, "global_step": 23916, "epoch": 225, "lr": 5.762553447350491e-05} +{"train_loss": 0.013111688196659088, "grad_norm": 0.08872624486684799, "global_step": 23917, "epoch": 225, "lr": 5.762260536412841e-05} +{"train_loss": 0.016876723617315292, "grad_norm": 0.10989866405725479, "global_step": 23918, "epoch": 225, "lr": 5.76196762279694e-05} +{"train_loss": 0.015464844182133675, "grad_norm": 0.09325538575649261, "global_step": 23919, "epoch": 225, "lr": 5.7616747065038167e-05} +{"train_loss": 0.015562404878437519, "grad_norm": 0.10990780591964722, "global_step": 23920, "epoch": 225, "lr": 5.761381787534499e-05} +{"train_loss": 0.015841811895370483, "grad_norm": 0.12758052349090576, "global_step": 23921, "epoch": 225, "lr": 5.761088865890016e-05} +{"train_loss": 0.014567491598427296, "grad_norm": 0.08604541420936584, "global_step": 23922, "epoch": 225, "lr": 5.7607959415713995e-05} +{"train_loss": 0.01584377884864807, "grad_norm": 0.07786999642848969, "global_step": 23923, "epoch": 225, "lr": 5.760503014579674e-05} +{"train_loss": 0.013758406974375248, "grad_norm": 0.050125040113925934, "global_step": 23924, "epoch": 225, "lr": 5.760210084915876e-05} +{"train_loss": 0.016387056559324265, "grad_norm": 0.11609857529401779, "global_step": 23925, "epoch": 225, "lr": 5.759917152581026e-05} +{"train_loss": 0.01508658193051815, "grad_norm": 0.10574457794427872, "global_step": 23926, "epoch": 225, "lr": 5.75962421757616e-05} +{"train_loss": 0.016700880602002144, "grad_norm": 0.06346813589334488, "global_step": 23927, "epoch": 225, "lr": 5.759331279902305e-05} +{"train_loss": 0.01432991772890091, "grad_norm": 0.04803788289427757, "global_step": 23928, "epoch": 225, "lr": 5.7590383395604885e-05} +{"train_loss": 0.01680256426334381, "grad_norm": 0.13165636360645294, "global_step": 23929, "epoch": 225, "lr": 5.758745396551742e-05} +{"train_loss": 0.016615647822618484, "grad_norm": 0.055204253643751144, "global_step": 23930, "epoch": 225, "lr": 5.758452450877095e-05} +{"train_loss": 0.017546396702528, "grad_norm": 0.09203335642814636, "global_step": 23931, "epoch": 225, "lr": 5.7581595025375756e-05} +{"train_loss": 0.017525065690279007, "grad_norm": 0.13045017421245575, "global_step": 23932, "epoch": 225, "lr": 5.757866551534213e-05} +{"train_loss": 0.014052973128855228, "grad_norm": 0.10662336647510529, "global_step": 23933, "epoch": 225, "lr": 5.757573597868038e-05} +{"train_loss": 0.017406482249498367, "grad_norm": 0.1411609947681427, "global_step": 23934, "epoch": 225, "lr": 5.757280641540077e-05} +{"train_loss": 0.017490025609731674, "grad_norm": 0.06260263919830322, "global_step": 23935, "epoch": 225, "lr": 5.7569876825513615e-05} +{"train_loss": 0.0143740214407444, "grad_norm": 0.0663275495171547, "global_step": 23936, "epoch": 225, "lr": 5.756694720902922e-05} +{"train_loss": 0.016459954902529716, "grad_norm": 0.109640933573246, "global_step": 23937, "epoch": 225, "lr": 5.756401756595785e-05} +{"train_loss": 0.015765808522701263, "grad_norm": 0.06723566353321075, "global_step": 23938, "epoch": 225, "lr": 5.756108789630982e-05} +{"train_loss": 0.016833949834108353, "grad_norm": 0.06793941557407379, "global_step": 23939, "epoch": 225, "lr": 5.7558158200095426e-05} +{"train_loss": 0.016601726412773132, "grad_norm": 0.07253391295671463, "global_step": 23940, "epoch": 225, "lr": 5.7555228477324917e-05} +{"train_loss": 0.017530664801597595, "grad_norm": 0.1049254834651947, "global_step": 23941, "epoch": 225, "lr": 5.7552298728008646e-05} +{"train_loss": 0.017357278615236282, "grad_norm": 0.1000915914773941, "global_step": 23942, "epoch": 225, "lr": 5.754936895215688e-05} +{"train_loss": 0.015624098479747772, "grad_norm": 0.06495574861764908, "global_step": 23943, "epoch": 225, "lr": 5.754643914977991e-05} +{"train_loss": 0.015790145844221115, "grad_norm": 0.08449327945709229, "global_step": 23944, "epoch": 225, "lr": 5.754350932088803e-05} +{"train_loss": 0.014342176727950573, "grad_norm": 0.06697221845388412, "global_step": 23945, "epoch": 225, "lr": 5.754057946549155e-05} +{"train_loss": 0.017482895404100418, "grad_norm": 0.07331187278032303, "global_step": 23946, "epoch": 225, "lr": 5.753764958360075e-05} +{"train_loss": 0.01394356694072485, "grad_norm": 0.13896198570728302, "global_step": 23947, "epoch": 225, "lr": 5.753471967522592e-05} +{"train_loss": 0.015394496731460094, "grad_norm": 0.12453000992536545, "global_step": 23948, "epoch": 225, "lr": 5.753178974037737e-05} +{"train_loss": 0.013800378888845444, "grad_norm": 0.0906771793961525, "global_step": 23949, "epoch": 225, "lr": 5.752885977906539e-05} +{"train_loss": 0.016022764146327972, "grad_norm": 0.07245342433452606, "global_step": 23950, "epoch": 225, "lr": 5.7525929791300256e-05} +{"train_loss": 0.017933132126927376, "grad_norm": 0.07747688889503479, "global_step": 23951, "epoch": 225, "lr": 5.752299977709229e-05} +{"train_loss": 0.017332764342427254, "grad_norm": 0.06785530596971512, "global_step": 23952, "epoch": 225, "lr": 5.7520069736451766e-05} +{"train_loss": 0.016450200229883194, "grad_norm": 0.09999838471412659, "global_step": 23953, "epoch": 225, "lr": 5.7517139669388995e-05} +{"train_loss": 0.01670827344059944, "grad_norm": 0.06655997037887573, "global_step": 23954, "epoch": 225, "lr": 5.751420957591426e-05} +{"train_loss": 0.015767808379781135, "grad_norm": 0.1082150861620903, "global_step": 23955, "epoch": 225, "lr": 5.751127945603786e-05, "val_loss": 0.019948294386267662} +{"train_loss": 0.01748657040297985, "grad_norm": 0.1221042200922966, "global_step": 23956, "epoch": 226, "lr": 5.750834930977009e-05} +{"train_loss": 0.016219530254602432, "grad_norm": 0.06691151857376099, "global_step": 23957, "epoch": 226, "lr": 5.750541913712124e-05} +{"train_loss": 0.016352910548448563, "grad_norm": 0.07517894357442856, "global_step": 23958, "epoch": 226, "lr": 5.750248893810163e-05} +{"train_loss": 0.015270311385393143, "grad_norm": 0.14857091009616852, "global_step": 23959, "epoch": 226, "lr": 5.749955871272153e-05} +{"train_loss": 0.015665467828512192, "grad_norm": 0.09574244916439056, "global_step": 23960, "epoch": 226, "lr": 5.749662846099123e-05} +{"train_loss": 0.012893302366137505, "grad_norm": 0.05078722909092903, "global_step": 23961, "epoch": 226, "lr": 5.749369818292105e-05} +{"train_loss": 0.016914259642362595, "grad_norm": 0.05959901213645935, "global_step": 23962, "epoch": 226, "lr": 5.7490767878521267e-05} +{"train_loss": 0.013448084704577923, "grad_norm": 0.07231323421001434, "global_step": 23963, "epoch": 226, "lr": 5.7487837547802184e-05} +{"train_loss": 0.016970409080386162, "grad_norm": 0.14916221797466278, "global_step": 23964, "epoch": 226, "lr": 5.748490719077408e-05} +{"train_loss": 0.015046404674649239, "grad_norm": 0.10797195136547089, "global_step": 23965, "epoch": 226, "lr": 5.74819768074473e-05} +{"train_loss": 0.013928942382335663, "grad_norm": 0.09880262613296509, "global_step": 23966, "epoch": 226, "lr": 5.747904639783208e-05} +{"train_loss": 0.014853579923510551, "grad_norm": 0.06316748261451721, "global_step": 23967, "epoch": 226, "lr": 5.747611596193876e-05} +{"train_loss": 0.01806776225566864, "grad_norm": 0.08480411767959595, "global_step": 23968, "epoch": 226, "lr": 5.7473185499777624e-05} +{"train_loss": 0.017472393810749054, "grad_norm": 0.06501903384923935, "global_step": 23969, "epoch": 226, "lr": 5.747025501135895e-05} +{"train_loss": 0.017470989376306534, "grad_norm": 0.09417525678873062, "global_step": 23970, "epoch": 226, "lr": 5.7467324496693056e-05} +{"train_loss": 0.013850964605808258, "grad_norm": 0.08392487466335297, "global_step": 23971, "epoch": 226, "lr": 5.746439395579023e-05} +{"train_loss": 0.014464100822806358, "grad_norm": 0.09133314341306686, "global_step": 23972, "epoch": 226, "lr": 5.746146338866076e-05} +{"train_loss": 0.015253188088536263, "grad_norm": 0.05819419026374817, "global_step": 23973, "epoch": 226, "lr": 5.7458532795314965e-05} +{"train_loss": 0.014642361551523209, "grad_norm": 0.14540515840053558, "global_step": 23974, "epoch": 226, "lr": 5.7455602175763136e-05} +{"train_loss": 0.01605496183037758, "grad_norm": 0.11247506737709045, "global_step": 23975, "epoch": 226, "lr": 5.745267153001555e-05} +{"train_loss": 0.015033816918730736, "grad_norm": 0.05900344252586365, "global_step": 23976, "epoch": 226, "lr": 5.7449740858082515e-05} +{"train_loss": 0.01578756608068943, "grad_norm": 0.07732827216386795, "global_step": 23977, "epoch": 226, "lr": 5.744681015997434e-05} +{"train_loss": 0.016210880130529404, "grad_norm": 0.07508116215467453, "global_step": 23978, "epoch": 226, "lr": 5.744387943570131e-05} +{"train_loss": 0.01692153885960579, "grad_norm": 0.06798487901687622, "global_step": 23979, "epoch": 226, "lr": 5.744094868527372e-05} +{"train_loss": 0.01491248607635498, "grad_norm": 0.1066148579120636, "global_step": 23980, "epoch": 226, "lr": 5.7438017908701894e-05} +{"train_loss": 0.016342908143997192, "grad_norm": 0.11005849391222, "global_step": 23981, "epoch": 226, "lr": 5.743508710599606e-05} +{"train_loss": 0.0161272119730711, "grad_norm": 0.05701565369963646, "global_step": 23982, "epoch": 226, "lr": 5.7432156277166604e-05} +{"train_loss": 0.01577020063996315, "grad_norm": 0.07575549930334091, "global_step": 23983, "epoch": 226, "lr": 5.742922542222378e-05} +{"train_loss": 0.014260376803576946, "grad_norm": 0.051643919199705124, "global_step": 23984, "epoch": 226, "lr": 5.7426294541177874e-05} +{"train_loss": 0.014456402510404587, "grad_norm": 0.06280190497636795, "global_step": 23985, "epoch": 226, "lr": 5.742336363403921e-05} +{"train_loss": 0.016968373209238052, "grad_norm": 0.08776769787073135, "global_step": 23986, "epoch": 226, "lr": 5.742043270081807e-05} +{"train_loss": 0.017597917467355728, "grad_norm": 0.10007082670927048, "global_step": 23987, "epoch": 226, "lr": 5.741750174152476e-05} +{"train_loss": 0.015139486640691757, "grad_norm": 0.11104290187358856, "global_step": 23988, "epoch": 226, "lr": 5.7414570756169563e-05} +{"train_loss": 0.014257974922657013, "grad_norm": 0.06023918837308884, "global_step": 23989, "epoch": 226, "lr": 5.7411639744762803e-05} +{"train_loss": 0.016338691115379333, "grad_norm": 0.08244354277849197, "global_step": 23990, "epoch": 226, "lr": 5.740870870731476e-05} +{"train_loss": 0.01821770891547203, "grad_norm": 0.08312587440013885, "global_step": 23991, "epoch": 226, "lr": 5.740577764383572e-05} +{"train_loss": 0.02028396725654602, "grad_norm": 0.11002767086029053, "global_step": 23992, "epoch": 226, "lr": 5.7402846554336035e-05} +{"train_loss": 0.018700260668992996, "grad_norm": 0.06744362413883209, "global_step": 23993, "epoch": 226, "lr": 5.7399915438825923e-05} +{"train_loss": 0.016789834946393967, "grad_norm": 0.07164998352527618, "global_step": 23994, "epoch": 226, "lr": 5.739698429731575e-05} +{"train_loss": 0.017032209783792496, "grad_norm": 0.1361248940229416, "global_step": 23995, "epoch": 226, "lr": 5.739405312981579e-05} +{"train_loss": 0.01516849733889103, "grad_norm": 0.12455900758504868, "global_step": 23996, "epoch": 226, "lr": 5.739112193633635e-05} +{"train_loss": 0.014258438721299171, "grad_norm": 0.06199558451771736, "global_step": 23997, "epoch": 226, "lr": 5.738819071688771e-05} +{"train_loss": 0.016626447439193726, "grad_norm": 0.06054901331663132, "global_step": 23998, "epoch": 226, "lr": 5.738525947148018e-05} +{"train_loss": 0.018832948058843613, "grad_norm": 0.08731196075677872, "global_step": 23999, "epoch": 226, "lr": 5.738232820012407e-05} +{"train_loss": 0.01582249253988266, "grad_norm": 0.09258832037448883, "global_step": 24000, "epoch": 226, "lr": 5.737939690282966e-05} +{"train_loss": 0.013815573416650295, "grad_norm": 0.10050380229949951, "global_step": 24001, "epoch": 226, "lr": 5.7376465579607264e-05} +{"train_loss": 0.016597015783190727, "grad_norm": 0.09693633019924164, "global_step": 24002, "epoch": 226, "lr": 5.737353423046718e-05} +{"train_loss": 0.015432650223374367, "grad_norm": 0.0811600610613823, "global_step": 24003, "epoch": 226, "lr": 5.73706028554197e-05} +{"train_loss": 0.017931031063199043, "grad_norm": 0.0822368934750557, "global_step": 24004, "epoch": 226, "lr": 5.736767145447511e-05} +{"train_loss": 0.01502138189971447, "grad_norm": 0.06278267502784729, "global_step": 24005, "epoch": 226, "lr": 5.736474002764374e-05} +{"train_loss": 0.014238524250686169, "grad_norm": 0.09743599593639374, "global_step": 24006, "epoch": 226, "lr": 5.7361808574935884e-05} +{"train_loss": 0.01671833172440529, "grad_norm": 0.18532675504684448, "global_step": 24007, "epoch": 226, "lr": 5.735887709636183e-05} +{"train_loss": 0.015050881542265415, "grad_norm": 0.05348263308405876, "global_step": 24008, "epoch": 226, "lr": 5.735594559193187e-05} +{"train_loss": 0.015771305188536644, "grad_norm": 0.08550211787223816, "global_step": 24009, "epoch": 226, "lr": 5.735301406165634e-05} +{"train_loss": 0.0139816515147686, "grad_norm": 0.09454859048128128, "global_step": 24010, "epoch": 226, "lr": 5.735008250554549e-05} +{"train_loss": 0.01510757114738226, "grad_norm": 0.05826922506093979, "global_step": 24011, "epoch": 226, "lr": 5.734715092360966e-05} +{"train_loss": 0.01712595298886299, "grad_norm": 0.09867395460605621, "global_step": 24012, "epoch": 226, "lr": 5.734421931585914e-05} +{"train_loss": 0.014543595723807812, "grad_norm": 0.05713147297501564, "global_step": 24013, "epoch": 226, "lr": 5.7341287682304226e-05} +{"train_loss": 0.01562708243727684, "grad_norm": 0.054800838232040405, "global_step": 24014, "epoch": 226, "lr": 5.733835602295521e-05} +{"train_loss": 0.01690155267715454, "grad_norm": 0.08695980906486511, "global_step": 24015, "epoch": 226, "lr": 5.733542433782241e-05} +{"train_loss": 0.014643442817032337, "grad_norm": 0.07827750593423843, "global_step": 24016, "epoch": 226, "lr": 5.733249262691612e-05} +{"train_loss": 0.016589824110269547, "grad_norm": 0.053664904087781906, "global_step": 24017, "epoch": 226, "lr": 5.732956089024663e-05} +{"train_loss": 0.014674798585474491, "grad_norm": 0.13545143604278564, "global_step": 24018, "epoch": 226, "lr": 5.732662912782427e-05} +{"train_loss": 0.012670010328292847, "grad_norm": 0.06187533214688301, "global_step": 24019, "epoch": 226, "lr": 5.732369733965931e-05} +{"train_loss": 0.0181218683719635, "grad_norm": 0.06723851710557938, "global_step": 24020, "epoch": 226, "lr": 5.7320765525762044e-05} +{"train_loss": 0.01564055308699608, "grad_norm": 0.10933048278093338, "global_step": 24021, "epoch": 226, "lr": 5.731783368614283e-05} +{"train_loss": 0.01387837715446949, "grad_norm": 0.08987804502248764, "global_step": 24022, "epoch": 226, "lr": 5.7314901820811896e-05} +{"train_loss": 0.015320736914873123, "grad_norm": 0.08734174072742462, "global_step": 24023, "epoch": 226, "lr": 5.7311969929779594e-05} +{"train_loss": 0.01733268052339554, "grad_norm": 0.08913692086935043, "global_step": 24024, "epoch": 226, "lr": 5.730903801305621e-05} +{"train_loss": 0.015294959768652916, "grad_norm": 0.07664801180362701, "global_step": 24025, "epoch": 226, "lr": 5.730610607065204e-05} +{"train_loss": 0.017868362367153168, "grad_norm": 0.09457021951675415, "global_step": 24026, "epoch": 226, "lr": 5.73031741025774e-05} +{"train_loss": 0.016034450381994247, "grad_norm": 0.05878100544214249, "global_step": 24027, "epoch": 226, "lr": 5.7300242108842574e-05} +{"train_loss": 0.015039963647723198, "grad_norm": 0.06163269281387329, "global_step": 24028, "epoch": 226, "lr": 5.729731008945788e-05} +{"train_loss": 0.01506047323346138, "grad_norm": 0.07369837909936905, "global_step": 24029, "epoch": 226, "lr": 5.7294378044433605e-05} +{"train_loss": 0.017243437469005585, "grad_norm": 0.07911091297864914, "global_step": 24030, "epoch": 226, "lr": 5.729144597378006e-05} +{"train_loss": 0.015624300576746464, "grad_norm": 0.06891331821680069, "global_step": 24031, "epoch": 226, "lr": 5.728851387750754e-05} +{"train_loss": 0.013186704367399216, "grad_norm": 0.1159646287560463, "global_step": 24032, "epoch": 226, "lr": 5.728558175562635e-05} +{"train_loss": 0.015925906598567963, "grad_norm": 0.08391639590263367, "global_step": 24033, "epoch": 226, "lr": 5.728264960814681e-05} +{"train_loss": 0.015413784421980381, "grad_norm": 0.09283844381570816, "global_step": 24034, "epoch": 226, "lr": 5.7279717435079184e-05} +{"train_loss": 0.015090086497366428, "grad_norm": 0.07643632590770721, "global_step": 24035, "epoch": 226, "lr": 5.72767852364338e-05} +{"train_loss": 0.01598755642771721, "grad_norm": 0.05474923551082611, "global_step": 24036, "epoch": 226, "lr": 5.7273853012220977e-05} +{"train_loss": 0.015958797186613083, "grad_norm": 0.06680357456207275, "global_step": 24037, "epoch": 226, "lr": 5.7270920762450984e-05} +{"train_loss": 0.015469747595489025, "grad_norm": 0.06859848648309708, "global_step": 24038, "epoch": 226, "lr": 5.726798848713415e-05} +{"train_loss": 0.017389867454767227, "grad_norm": 0.07222037762403488, "global_step": 24039, "epoch": 226, "lr": 5.7265056186280754e-05} +{"train_loss": 0.01437629945576191, "grad_norm": 0.11780471354722977, "global_step": 24040, "epoch": 226, "lr": 5.7262123859901104e-05} +{"train_loss": 0.014334185048937798, "grad_norm": 0.06413930654525757, "global_step": 24041, "epoch": 226, "lr": 5.725919150800553e-05} +{"train_loss": 0.014475895091891289, "grad_norm": 0.07589324563741684, "global_step": 24042, "epoch": 226, "lr": 5.725625913060429e-05} +{"train_loss": 0.016856061294674873, "grad_norm": 0.07102182507514954, "global_step": 24043, "epoch": 226, "lr": 5.725332672770774e-05} +{"train_loss": 0.016058407723903656, "grad_norm": 0.07717488706111908, "global_step": 24044, "epoch": 226, "lr": 5.7250394299326116e-05} +{"train_loss": 0.01580239273607731, "grad_norm": 0.10789709538221359, "global_step": 24045, "epoch": 226, "lr": 5.724746184546978e-05} +{"train_loss": 0.014142027124762535, "grad_norm": 0.06532123684883118, "global_step": 24046, "epoch": 226, "lr": 5.7244529366149016e-05} +{"train_loss": 0.015068091452121735, "grad_norm": 0.09988152980804443, "global_step": 24047, "epoch": 226, "lr": 5.724159686137414e-05} +{"train_loss": 0.01623978093266487, "grad_norm": 0.10062652826309204, "global_step": 24048, "epoch": 226, "lr": 5.7238664331155425e-05} +{"train_loss": 0.01458461582660675, "grad_norm": 0.17229421436786652, "global_step": 24049, "epoch": 226, "lr": 5.7235731775503186e-05} +{"train_loss": 0.015005199238657951, "grad_norm": 0.058653492480516434, "global_step": 24050, "epoch": 226, "lr": 5.723279919442774e-05} +{"train_loss": 0.015349277295172215, "grad_norm": 0.056677430868148804, "global_step": 24051, "epoch": 226, "lr": 5.7229866587939385e-05} +{"train_loss": 0.015561763197183609, "grad_norm": 0.06483541429042816, "global_step": 24052, "epoch": 226, "lr": 5.7226933956048415e-05} +{"train_loss": 0.014658141881227493, "grad_norm": 0.06934233754873276, "global_step": 24053, "epoch": 226, "lr": 5.722400129876515e-05} +{"train_loss": 0.01621916890144348, "grad_norm": 0.173355832695961, "global_step": 24054, "epoch": 226, "lr": 5.722106861609988e-05} +{"train_loss": 0.014840683899819851, "grad_norm": 0.06792210787534714, "global_step": 24055, "epoch": 226, "lr": 5.7218135908062934e-05} +{"train_loss": 0.013491693884134293, "grad_norm": 0.05042143166065216, "global_step": 24056, "epoch": 226, "lr": 5.721520317466458e-05} +{"train_loss": 0.017017437145113945, "grad_norm": 0.12318599224090576, "global_step": 24057, "epoch": 226, "lr": 5.721227041591514e-05} +{"train_loss": 0.0156550332903862, "grad_norm": 0.09545484930276871, "global_step": 24058, "epoch": 226, "lr": 5.7209337631824924e-05} +{"train_loss": 0.01738724671304226, "grad_norm": 0.14265789091587067, "global_step": 24059, "epoch": 226, "lr": 5.7206404822404224e-05} +{"train_loss": 0.0167527012526989, "grad_norm": 0.12198261171579361, "global_step": 24060, "epoch": 226, "lr": 5.720347198766336e-05} +{"train_loss": 0.015745263142546394, "grad_norm": 0.09498534351587296, "global_step": 24061, "epoch": 226, "lr": 5.7200539127612604e-05, "val_loss": 0.019307099282741547} +{"train_loss": 0.017249615862965584, "grad_norm": 0.07089756429195404, "global_step": 24062, "epoch": 227, "lr": 5.7197606242262325e-05} +{"train_loss": 0.018243279308080673, "grad_norm": 0.11185919493436813, "global_step": 24063, "epoch": 227, "lr": 5.719467333162275e-05} +{"train_loss": 0.01883968710899353, "grad_norm": 0.07245081663131714, "global_step": 24064, "epoch": 227, "lr": 5.719174039570424e-05} +{"train_loss": 0.016692839562892914, "grad_norm": 0.151051327586174, "global_step": 24065, "epoch": 227, "lr": 5.718880743451709e-05} +{"train_loss": 0.01943059451878071, "grad_norm": 0.09084323048591614, "global_step": 24066, "epoch": 227, "lr": 5.7185874448071586e-05} +{"train_loss": 0.015862468630075455, "grad_norm": 0.06465145945549011, "global_step": 24067, "epoch": 227, "lr": 5.718294143637805e-05} +{"train_loss": 0.013203523121774197, "grad_norm": 0.0732077956199646, "global_step": 24068, "epoch": 227, "lr": 5.718000839944677e-05} +{"train_loss": 0.016003865748643875, "grad_norm": 0.1343669295310974, "global_step": 24069, "epoch": 227, "lr": 5.717707533728807e-05} +{"train_loss": 0.01425805501639843, "grad_norm": 0.062189992517232895, "global_step": 24070, "epoch": 227, "lr": 5.7174142249912253e-05} +{"train_loss": 0.013381846249103546, "grad_norm": 0.07552272081375122, "global_step": 24071, "epoch": 227, "lr": 5.717120913732962e-05} +{"train_loss": 0.016615770757198334, "grad_norm": 0.08239627629518509, "global_step": 24072, "epoch": 227, "lr": 5.716827599955047e-05} +{"train_loss": 0.01461958047002554, "grad_norm": 0.08862177282571793, "global_step": 24073, "epoch": 227, "lr": 5.71653428365851e-05} +{"train_loss": 0.01424838975071907, "grad_norm": 0.10735643655061722, "global_step": 24074, "epoch": 227, "lr": 5.716240964844387e-05} +{"train_loss": 0.017156992107629776, "grad_norm": 0.0993771106004715, "global_step": 24075, "epoch": 227, "lr": 5.7159476435137015e-05} +{"train_loss": 0.016925351694226265, "grad_norm": 0.05573251098394394, "global_step": 24076, "epoch": 227, "lr": 5.715654319667488e-05} +{"train_loss": 0.014681767672300339, "grad_norm": 0.05968521907925606, "global_step": 24077, "epoch": 227, "lr": 5.7153609933067784e-05} +{"train_loss": 0.01666121557354927, "grad_norm": 0.07857554405927658, "global_step": 24078, "epoch": 227, "lr": 5.7150676644326e-05} +{"train_loss": 0.015194112434983253, "grad_norm": 0.06471412628889084, "global_step": 24079, "epoch": 227, "lr": 5.714774333045985e-05} +{"train_loss": 0.017114980146288872, "grad_norm": 0.10829131305217743, "global_step": 24080, "epoch": 227, "lr": 5.714480999147963e-05} +{"train_loss": 0.014988607726991177, "grad_norm": 0.07857916504144669, "global_step": 24081, "epoch": 227, "lr": 5.714187662739566e-05} +{"train_loss": 0.01291240006685257, "grad_norm": 0.05443219467997551, "global_step": 24082, "epoch": 227, "lr": 5.7138943238218245e-05} +{"train_loss": 0.016405196860432625, "grad_norm": 0.10789947211742401, "global_step": 24083, "epoch": 227, "lr": 5.7136009823957684e-05} +{"train_loss": 0.01780417189002037, "grad_norm": 0.06793291121721268, "global_step": 24084, "epoch": 227, "lr": 5.713307638462428e-05} +{"train_loss": 0.01491096243262291, "grad_norm": 0.08078140020370483, "global_step": 24085, "epoch": 227, "lr": 5.7130142920228356e-05} +{"train_loss": 0.013655299320816994, "grad_norm": 0.050496939569711685, "global_step": 24086, "epoch": 227, "lr": 5.7127209430780224e-05} +{"train_loss": 0.015362955629825592, "grad_norm": 0.08121751993894577, "global_step": 24087, "epoch": 227, "lr": 5.712427591629016e-05} +{"train_loss": 0.01658770814538002, "grad_norm": 0.06359812617301941, "global_step": 24088, "epoch": 227, "lr": 5.7121342376768495e-05} +{"train_loss": 0.015615681186318398, "grad_norm": 0.07518124580383301, "global_step": 24089, "epoch": 227, "lr": 5.711840881222553e-05} +{"train_loss": 0.01686786860227585, "grad_norm": 0.0719117522239685, "global_step": 24090, "epoch": 227, "lr": 5.711547522267157e-05} +{"train_loss": 0.015530072152614594, "grad_norm": 0.08487050235271454, "global_step": 24091, "epoch": 227, "lr": 5.711254160811692e-05} +{"train_loss": 0.0163480993360281, "grad_norm": 0.078940749168396, "global_step": 24092, "epoch": 227, "lr": 5.71096079685719e-05} +{"train_loss": 0.013334089890122414, "grad_norm": 0.0972006544470787, "global_step": 24093, "epoch": 227, "lr": 5.7106674304046815e-05} +{"train_loss": 0.015478374436497688, "grad_norm": 0.06789242476224899, "global_step": 24094, "epoch": 227, "lr": 5.710374061455195e-05} +{"train_loss": 0.01770791783928871, "grad_norm": 0.10843947529792786, "global_step": 24095, "epoch": 227, "lr": 5.710080690009765e-05} +{"train_loss": 0.015551947057247162, "grad_norm": 0.12514737248420715, "global_step": 24096, "epoch": 227, "lr": 5.7097873160694185e-05} +{"train_loss": 0.0140012726187706, "grad_norm": 0.06931417435407639, "global_step": 24097, "epoch": 227, "lr": 5.7094939396351884e-05} +{"train_loss": 0.01611477881669998, "grad_norm": 0.060688383877277374, "global_step": 24098, "epoch": 227, "lr": 5.709200560708106e-05} +{"train_loss": 0.01473180390894413, "grad_norm": 0.095732681453228, "global_step": 24099, "epoch": 227, "lr": 5.7089071792892e-05} +{"train_loss": 0.01544356718659401, "grad_norm": 0.10587681084871292, "global_step": 24100, "epoch": 227, "lr": 5.7086137953795037e-05} +{"train_loss": 0.01470424234867096, "grad_norm": 0.09508733451366425, "global_step": 24101, "epoch": 227, "lr": 5.708320408980047e-05} +{"train_loss": 0.01941613107919693, "grad_norm": 0.09969396144151688, "global_step": 24102, "epoch": 227, "lr": 5.708027020091858e-05} +{"train_loss": 0.01457245647907257, "grad_norm": 0.08904023468494415, "global_step": 24103, "epoch": 227, "lr": 5.707733628715972e-05} +{"train_loss": 0.01566062867641449, "grad_norm": 0.20377178490161896, "global_step": 24104, "epoch": 227, "lr": 5.707440234853417e-05} +{"train_loss": 0.016601169481873512, "grad_norm": 0.1216619461774826, "global_step": 24105, "epoch": 227, "lr": 5.707146838505225e-05} +{"train_loss": 0.013864963315427303, "grad_norm": 0.08199003338813782, "global_step": 24106, "epoch": 227, "lr": 5.706853439672428e-05} +{"train_loss": 0.01654302328824997, "grad_norm": 0.05723690241575241, "global_step": 24107, "epoch": 227, "lr": 5.7065600383560525e-05} +{"train_loss": 0.01620546542108059, "grad_norm": 0.15860231220722198, "global_step": 24108, "epoch": 227, "lr": 5.7062666345571335e-05} +{"train_loss": 0.015859652310609818, "grad_norm": 0.08337444812059402, "global_step": 24109, "epoch": 227, "lr": 5.705973228276702e-05} +{"train_loss": 0.018350552767515182, "grad_norm": 0.06272081285715103, "global_step": 24110, "epoch": 227, "lr": 5.705679819515787e-05} +{"train_loss": 0.016736721619963646, "grad_norm": 0.08876371383666992, "global_step": 24111, "epoch": 227, "lr": 5.705386408275419e-05} +{"train_loss": 0.015603308565914631, "grad_norm": 0.06465838104486465, "global_step": 24112, "epoch": 227, "lr": 5.705092994556631e-05} +{"train_loss": 0.016700757667422295, "grad_norm": 0.12146192789077759, "global_step": 24113, "epoch": 227, "lr": 5.7047995783604524e-05} +{"train_loss": 0.016698747873306274, "grad_norm": 0.06196731701493263, "global_step": 24114, "epoch": 227, "lr": 5.7045061596879134e-05} +{"train_loss": 0.017629917711019516, "grad_norm": 0.07275894284248352, "global_step": 24115, "epoch": 227, "lr": 5.704212738540048e-05} +{"train_loss": 0.017577670514583588, "grad_norm": 0.08413907140493393, "global_step": 24116, "epoch": 227, "lr": 5.7039193149178835e-05} +{"train_loss": 0.0157524012029171, "grad_norm": 0.06975645571947098, "global_step": 24117, "epoch": 227, "lr": 5.703625888822454e-05} +{"train_loss": 0.014243623241782188, "grad_norm": 0.06450498849153519, "global_step": 24118, "epoch": 227, "lr": 5.70333246025479e-05} +{"train_loss": 0.01590029150247574, "grad_norm": 0.06202591583132744, "global_step": 24119, "epoch": 227, "lr": 5.7030390292159195e-05} +{"train_loss": 0.017106331884860992, "grad_norm": 0.12119890004396439, "global_step": 24120, "epoch": 227, "lr": 5.702745595706877e-05} +{"train_loss": 0.01459591556340456, "grad_norm": 0.0744481161236763, "global_step": 24121, "epoch": 227, "lr": 5.7024521597286916e-05} +{"train_loss": 0.01367681659758091, "grad_norm": 0.08640751987695694, "global_step": 24122, "epoch": 227, "lr": 5.702158721282395e-05} +{"train_loss": 0.016143420711159706, "grad_norm": 0.07893094420433044, "global_step": 24123, "epoch": 227, "lr": 5.7018652803690175e-05} +{"train_loss": 0.01623000204563141, "grad_norm": 0.0686720758676529, "global_step": 24124, "epoch": 227, "lr": 5.701571836989591e-05} +{"train_loss": 0.013739002868533134, "grad_norm": 0.11981615424156189, "global_step": 24125, "epoch": 227, "lr": 5.701278391145146e-05} +{"train_loss": 0.018184930086135864, "grad_norm": 0.06116136536002159, "global_step": 24126, "epoch": 227, "lr": 5.700984942836713e-05} +{"train_loss": 0.0152805270627141, "grad_norm": 0.05189374461770058, "global_step": 24127, "epoch": 227, "lr": 5.7006914920653266e-05} +{"train_loss": 0.017758697271347046, "grad_norm": 0.10568360984325409, "global_step": 24128, "epoch": 227, "lr": 5.700398038832013e-05} +{"train_loss": 0.0166090726852417, "grad_norm": 0.0814097598195076, "global_step": 24129, "epoch": 227, "lr": 5.700104583137804e-05} +{"train_loss": 0.013972422108054161, "grad_norm": 0.09223313629627228, "global_step": 24130, "epoch": 227, "lr": 5.699811124983735e-05} +{"train_loss": 0.017443707212805748, "grad_norm": 0.08967462927103043, "global_step": 24131, "epoch": 227, "lr": 5.6995176643708306e-05} +{"train_loss": 0.014204890467226505, "grad_norm": 0.0679994598031044, "global_step": 24132, "epoch": 227, "lr": 5.699224201300127e-05} +{"train_loss": 0.013314621523022652, "grad_norm": 0.06749362498521805, "global_step": 24133, "epoch": 227, "lr": 5.698930735772654e-05} +{"train_loss": 0.016649998724460602, "grad_norm": 0.1230861097574234, "global_step": 24134, "epoch": 227, "lr": 5.698637267789443e-05} +{"train_loss": 0.014814947731792927, "grad_norm": 0.07011910527944565, "global_step": 24135, "epoch": 227, "lr": 5.698343797351523e-05} +{"train_loss": 0.014742283150553703, "grad_norm": 0.09717167168855667, "global_step": 24136, "epoch": 227, "lr": 5.698050324459927e-05} +{"train_loss": 0.0185033418238163, "grad_norm": 0.06547096371650696, "global_step": 24137, "epoch": 227, "lr": 5.6977568491156866e-05} +{"train_loss": 0.014459621161222458, "grad_norm": 0.08146071434020996, "global_step": 24138, "epoch": 227, "lr": 5.69746337131983e-05} +{"train_loss": 0.015919599682092667, "grad_norm": 0.07638449966907501, "global_step": 24139, "epoch": 227, "lr": 5.697169891073393e-05} +{"train_loss": 0.016372589394450188, "grad_norm": 0.12678103148937225, "global_step": 24140, "epoch": 227, "lr": 5.696876408377403e-05} +{"train_loss": 0.01708368770778179, "grad_norm": 0.14100077748298645, "global_step": 24141, "epoch": 227, "lr": 5.6965829232328915e-05} +{"train_loss": 0.017493562772870064, "grad_norm": 0.051526397466659546, "global_step": 24142, "epoch": 227, "lr": 5.6962894356408926e-05} +{"train_loss": 0.014231499284505844, "grad_norm": 0.09696805477142334, "global_step": 24143, "epoch": 227, "lr": 5.695995945602432e-05} +{"train_loss": 0.014006385579705238, "grad_norm": 0.0831560492515564, "global_step": 24144, "epoch": 227, "lr": 5.695702453118548e-05} +{"train_loss": 0.014697715640068054, "grad_norm": 0.06371760368347168, "global_step": 24145, "epoch": 227, "lr": 5.695408958190267e-05} +{"train_loss": 0.015287276357412338, "grad_norm": 0.07783376425504684, "global_step": 24146, "epoch": 227, "lr": 5.695115460818621e-05} +{"train_loss": 0.016327427700161934, "grad_norm": 0.14504793286323547, "global_step": 24147, "epoch": 227, "lr": 5.6948219610046413e-05} +{"train_loss": 0.015710756182670593, "grad_norm": 0.08340292423963547, "global_step": 24148, "epoch": 227, "lr": 5.69452845874936e-05} +{"train_loss": 0.014324398711323738, "grad_norm": 0.14723117649555206, "global_step": 24149, "epoch": 227, "lr": 5.694234954053807e-05} +{"train_loss": 0.014917714521288872, "grad_norm": 0.052918896079063416, "global_step": 24150, "epoch": 227, "lr": 5.6939414469190154e-05} +{"train_loss": 0.014078237116336823, "grad_norm": 0.06484859436750412, "global_step": 24151, "epoch": 227, "lr": 5.693647937346014e-05} +{"train_loss": 0.01678577996790409, "grad_norm": 0.15461400151252747, "global_step": 24152, "epoch": 227, "lr": 5.693354425335836e-05} +{"train_loss": 0.016167916357517242, "grad_norm": 0.10621029883623123, "global_step": 24153, "epoch": 227, "lr": 5.693060910889513e-05} +{"train_loss": 0.015864133834838867, "grad_norm": 0.07845046371221542, "global_step": 24154, "epoch": 227, "lr": 5.6927673940080736e-05} +{"train_loss": 0.016272127628326416, "grad_norm": 0.05468861013650894, "global_step": 24155, "epoch": 227, "lr": 5.692473874692552e-05} +{"train_loss": 0.016468290239572525, "grad_norm": 0.06454361230134964, "global_step": 24156, "epoch": 227, "lr": 5.692180352943979e-05} +{"train_loss": 0.014994551427662373, "grad_norm": 0.08615436404943466, "global_step": 24157, "epoch": 227, "lr": 5.691886828763384e-05} +{"train_loss": 0.016642732545733452, "grad_norm": 0.08348476141691208, "global_step": 24158, "epoch": 227, "lr": 5.6915933021517996e-05} +{"train_loss": 0.012170080095529556, "grad_norm": 0.05420951172709465, "global_step": 24159, "epoch": 227, "lr": 5.691299773110258e-05} +{"train_loss": 0.01702071726322174, "grad_norm": 0.0932312160730362, "global_step": 24160, "epoch": 227, "lr": 5.691006241639788e-05} +{"train_loss": 0.015161260962486267, "grad_norm": 0.08472535759210587, "global_step": 24161, "epoch": 227, "lr": 5.690712707741425e-05} +{"train_loss": 0.014393123798072338, "grad_norm": 0.10912743210792542, "global_step": 24162, "epoch": 227, "lr": 5.690419171416196e-05} +{"train_loss": 0.014818379655480385, "grad_norm": 0.06126757711172104, "global_step": 24163, "epoch": 227, "lr": 5.6901256326651354e-05} +{"train_loss": 0.01529882661998272, "grad_norm": 0.07648153603076935, "global_step": 24164, "epoch": 227, "lr": 5.689832091489272e-05} +{"train_loss": 0.012748748064041138, "grad_norm": 0.0490918904542923, "global_step": 24165, "epoch": 227, "lr": 5.68953854788964e-05} +{"train_loss": 0.015459917485713959, "grad_norm": 0.12492979317903519, "global_step": 24166, "epoch": 227, "lr": 5.689245001867269e-05} +{"train_loss": 0.015725823892456182, "grad_norm": 0.09391198307275772, "global_step": 24167, "epoch": 227, "lr": 5.68895145342319e-05, "val_loss": 0.019595643505454063} +{"train_loss": 0.018928224220871925, "grad_norm": 0.16267316043376923, "global_step": 24168, "epoch": 228, "lr": 5.6886579025584374e-05} +{"train_loss": 0.016842814162373543, "grad_norm": 0.06115914508700371, "global_step": 24169, "epoch": 228, "lr": 5.688364349274039e-05} +{"train_loss": 0.016345664858818054, "grad_norm": 0.06571590900421143, "global_step": 24170, "epoch": 228, "lr": 5.688070793571026e-05} +{"train_loss": 0.016368674114346504, "grad_norm": 0.08577221632003784, "global_step": 24171, "epoch": 228, "lr": 5.687777235450435e-05} +{"train_loss": 0.01659480109810829, "grad_norm": 0.11404722929000854, "global_step": 24172, "epoch": 228, "lr": 5.687483674913291e-05} +{"train_loss": 0.01573290303349495, "grad_norm": 0.10091004520654678, "global_step": 24173, "epoch": 228, "lr": 5.68719011196063e-05} +{"train_loss": 0.014688687399029732, "grad_norm": 0.05587597191333771, "global_step": 24174, "epoch": 228, "lr": 5.686896546593481e-05} +{"train_loss": 0.01710052415728569, "grad_norm": 0.06404215097427368, "global_step": 24175, "epoch": 228, "lr": 5.686602978812876e-05} +{"train_loss": 0.013499281369149685, "grad_norm": 0.07609640061855316, "global_step": 24176, "epoch": 228, "lr": 5.6863094086198475e-05} +{"train_loss": 0.015440412797033787, "grad_norm": 0.07125308364629745, "global_step": 24177, "epoch": 228, "lr": 5.686015836015426e-05} +{"train_loss": 0.015284463763237, "grad_norm": 0.09159689396619797, "global_step": 24178, "epoch": 228, "lr": 5.685722261000643e-05} +{"train_loss": 0.0191606804728508, "grad_norm": 0.10510339587926865, "global_step": 24179, "epoch": 228, "lr": 5.6854286835765306e-05} +{"train_loss": 0.013663235120475292, "grad_norm": 0.06947863847017288, "global_step": 24180, "epoch": 228, "lr": 5.685135103744119e-05} +{"train_loss": 0.014705805107951164, "grad_norm": 0.06945072114467621, "global_step": 24181, "epoch": 228, "lr": 5.684841521504442e-05} +{"train_loss": 0.01404687948524952, "grad_norm": 0.05664119869470596, "global_step": 24182, "epoch": 228, "lr": 5.6845479368585275e-05} +{"train_loss": 0.017667261883616447, "grad_norm": 0.07180334627628326, "global_step": 24183, "epoch": 228, "lr": 5.684254349807412e-05} +{"train_loss": 0.01487177424132824, "grad_norm": 0.07792436331510544, "global_step": 24184, "epoch": 228, "lr": 5.683960760352122e-05} +{"train_loss": 0.01631525717675686, "grad_norm": 0.06778662651777267, "global_step": 24185, "epoch": 228, "lr": 5.6836671684936914e-05} +{"train_loss": 0.01631152629852295, "grad_norm": 0.052163172513246536, "global_step": 24186, "epoch": 228, "lr": 5.683373574233154e-05} +{"train_loss": 0.016805579885840416, "grad_norm": 0.08420392125844955, "global_step": 24187, "epoch": 228, "lr": 5.683079977571537e-05} +{"train_loss": 0.015942247584462166, "grad_norm": 0.06083909049630165, "global_step": 24188, "epoch": 228, "lr": 5.682786378509874e-05} +{"train_loss": 0.014892598614096642, "grad_norm": 0.07496768981218338, "global_step": 24189, "epoch": 228, "lr": 5.682492777049197e-05} +{"train_loss": 0.013720565475523472, "grad_norm": 0.06025275960564613, "global_step": 24190, "epoch": 228, "lr": 5.682199173190539e-05} +{"train_loss": 0.01641862466931343, "grad_norm": 0.06734707951545715, "global_step": 24191, "epoch": 228, "lr": 5.681905566934928e-05} +{"train_loss": 0.01263665035367012, "grad_norm": 0.060360729694366455, "global_step": 24192, "epoch": 228, "lr": 5.681611958283397e-05} +{"train_loss": 0.015831101685762405, "grad_norm": 0.1284349113702774, "global_step": 24193, "epoch": 228, "lr": 5.681318347236979e-05} +{"train_loss": 0.015203903429210186, "grad_norm": 0.05389149859547615, "global_step": 24194, "epoch": 228, "lr": 5.6810247337967036e-05} +{"train_loss": 0.013518866151571274, "grad_norm": 0.08137726783752441, "global_step": 24195, "epoch": 228, "lr": 5.680731117963605e-05} +{"train_loss": 0.016479764133691788, "grad_norm": 0.07684203237295151, "global_step": 24196, "epoch": 228, "lr": 5.680437499738711e-05} +{"train_loss": 0.015674851834774017, "grad_norm": 0.05977926775813103, "global_step": 24197, "epoch": 228, "lr": 5.680143879123058e-05} +{"train_loss": 0.015017759054899216, "grad_norm": 0.08209255337715149, "global_step": 24198, "epoch": 228, "lr": 5.679850256117675e-05} +{"train_loss": 0.016797451302409172, "grad_norm": 0.07673901319503784, "global_step": 24199, "epoch": 228, "lr": 5.6795566307235915e-05} +{"train_loss": 0.016328729689121246, "grad_norm": 0.059082988649606705, "global_step": 24200, "epoch": 228, "lr": 5.6792630029418435e-05} +{"train_loss": 0.015330573543906212, "grad_norm": 0.07913114130496979, "global_step": 24201, "epoch": 228, "lr": 5.6789693727734594e-05} +{"train_loss": 0.016865324229002, "grad_norm": 0.07452443242073059, "global_step": 24202, "epoch": 228, "lr": 5.6786757402194736e-05} +{"train_loss": 0.01872955821454525, "grad_norm": 0.1725718230009079, "global_step": 24203, "epoch": 228, "lr": 5.678382105280916e-05} +{"train_loss": 0.01563280262053013, "grad_norm": 0.07084998488426208, "global_step": 24204, "epoch": 228, "lr": 5.678088467958819e-05} +{"train_loss": 0.017080392688512802, "grad_norm": 0.0673845037817955, "global_step": 24205, "epoch": 228, "lr": 5.6777948282542126e-05} +{"train_loss": 0.01804891973733902, "grad_norm": 0.09271572530269623, "global_step": 24206, "epoch": 228, "lr": 5.677501186168131e-05} +{"train_loss": 0.016044262796640396, "grad_norm": 0.050989776849746704, "global_step": 24207, "epoch": 228, "lr": 5.677207541701605e-05} +{"train_loss": 0.016184784471988678, "grad_norm": 0.0636858120560646, "global_step": 24208, "epoch": 228, "lr": 5.6769138948556665e-05} +{"train_loss": 0.013610901311039925, "grad_norm": 0.11715782433748245, "global_step": 24209, "epoch": 228, "lr": 5.676620245631346e-05} +{"train_loss": 0.015086710453033447, "grad_norm": 0.06125929579138756, "global_step": 24210, "epoch": 228, "lr": 5.6763265940296774e-05} +{"train_loss": 0.014911022037267685, "grad_norm": 0.04968219995498657, "global_step": 24211, "epoch": 228, "lr": 5.6760329400516885e-05} +{"train_loss": 0.018834754824638367, "grad_norm": 0.13213318586349487, "global_step": 24212, "epoch": 228, "lr": 5.6757392836984166e-05} +{"train_loss": 0.015145735815167427, "grad_norm": 0.07596337050199509, "global_step": 24213, "epoch": 228, "lr": 5.67544562497089e-05} +{"train_loss": 0.013234181329607964, "grad_norm": 0.0935242548584938, "global_step": 24214, "epoch": 228, "lr": 5.675151963870141e-05} +{"train_loss": 0.016086075454950333, "grad_norm": 0.07665915042161942, "global_step": 24215, "epoch": 228, "lr": 5.6748583003972037e-05} +{"train_loss": 0.015416821464896202, "grad_norm": 0.11145305633544922, "global_step": 24216, "epoch": 228, "lr": 5.674564634553103e-05} +{"train_loss": 0.017252985388040543, "grad_norm": 0.053960639983415604, "global_step": 24217, "epoch": 228, "lr": 5.674270966338879e-05} +{"train_loss": 0.015284473076462746, "grad_norm": 0.06643344461917877, "global_step": 24218, "epoch": 228, "lr": 5.67397729575556e-05} +{"train_loss": 0.01614041067659855, "grad_norm": 0.13780397176742554, "global_step": 24219, "epoch": 228, "lr": 5.673683622804178e-05} +{"train_loss": 0.015244739130139351, "grad_norm": 0.05801526457071304, "global_step": 24220, "epoch": 228, "lr": 5.6733899474857634e-05} +{"train_loss": 0.017621491104364395, "grad_norm": 0.07731181383132935, "global_step": 24221, "epoch": 228, "lr": 5.6730962698013505e-05} +{"train_loss": 0.015183823183178902, "grad_norm": 0.06564309448003769, "global_step": 24222, "epoch": 228, "lr": 5.67280258975197e-05} +{"train_loss": 0.014311591163277626, "grad_norm": 0.06368400156497955, "global_step": 24223, "epoch": 228, "lr": 5.6725089073386514e-05} +{"train_loss": 0.014876941218972206, "grad_norm": 0.07302811741828918, "global_step": 24224, "epoch": 228, "lr": 5.672215222562433e-05} +{"train_loss": 0.017266083508729935, "grad_norm": 0.06859399378299713, "global_step": 24225, "epoch": 228, "lr": 5.6719215354243393e-05} +{"train_loss": 0.014398760162293911, "grad_norm": 0.04547909274697304, "global_step": 24226, "epoch": 228, "lr": 5.671627845925407e-05} +{"train_loss": 0.01590947061777115, "grad_norm": 0.10172734409570694, "global_step": 24227, "epoch": 228, "lr": 5.671334154066668e-05} +{"train_loss": 0.014921417459845543, "grad_norm": 0.06047436594963074, "global_step": 24228, "epoch": 228, "lr": 5.6710404598491494e-05} +{"train_loss": 0.014365101233124733, "grad_norm": 0.0858275517821312, "global_step": 24229, "epoch": 228, "lr": 5.6707467632738885e-05} +{"train_loss": 0.015356438234448433, "grad_norm": 0.092383474111557, "global_step": 24230, "epoch": 228, "lr": 5.670453064341915e-05} +{"train_loss": 0.015311290509998798, "grad_norm": 0.1916760802268982, "global_step": 24231, "epoch": 228, "lr": 5.6701593630542615e-05} +{"train_loss": 0.01503310538828373, "grad_norm": 0.08747278898954391, "global_step": 24232, "epoch": 228, "lr": 5.669865659411958e-05} +{"train_loss": 0.015585438348352909, "grad_norm": 0.05663623660802841, "global_step": 24233, "epoch": 228, "lr": 5.6695719534160395e-05} +{"train_loss": 0.01718020625412464, "grad_norm": 0.081253282725811, "global_step": 24234, "epoch": 228, "lr": 5.669278245067535e-05} +{"train_loss": 0.01500497292727232, "grad_norm": 0.09283652156591415, "global_step": 24235, "epoch": 228, "lr": 5.668984534367477e-05} +{"train_loss": 0.015661131590604782, "grad_norm": 0.06102128326892853, "global_step": 24236, "epoch": 228, "lr": 5.668690821316901e-05} +{"train_loss": 0.016435950994491577, "grad_norm": 0.09464914351701736, "global_step": 24237, "epoch": 228, "lr": 5.668397105916835e-05} +{"train_loss": 0.016630901023745537, "grad_norm": 0.07883801311254501, "global_step": 24238, "epoch": 228, "lr": 5.668103388168311e-05} +{"train_loss": 0.016092387959361076, "grad_norm": 0.10676135122776031, "global_step": 24239, "epoch": 228, "lr": 5.667809668072365e-05} +{"train_loss": 0.016949737444519997, "grad_norm": 0.052627082914114, "global_step": 24240, "epoch": 228, "lr": 5.667515945630023e-05} +{"train_loss": 0.01508444081991911, "grad_norm": 0.10582076758146286, "global_step": 24241, "epoch": 228, "lr": 5.667222220842322e-05} +{"train_loss": 0.014013243839144707, "grad_norm": 0.0746879056096077, "global_step": 24242, "epoch": 228, "lr": 5.666928493710293e-05} +{"train_loss": 0.01643989607691765, "grad_norm": 0.0706089437007904, "global_step": 24243, "epoch": 228, "lr": 5.6666347642349657e-05} +{"train_loss": 0.0168132521212101, "grad_norm": 0.0669689029455185, "global_step": 24244, "epoch": 228, "lr": 5.666341032417375e-05} +{"train_loss": 0.017058227211236954, "grad_norm": 0.10015343874692917, "global_step": 24245, "epoch": 228, "lr": 5.666047298258551e-05} +{"train_loss": 0.01567229814827442, "grad_norm": 0.06535972654819489, "global_step": 24246, "epoch": 228, "lr": 5.665753561759527e-05} +{"train_loss": 0.01581859588623047, "grad_norm": 0.06672853231430054, "global_step": 24247, "epoch": 228, "lr": 5.665459822921333e-05} +{"train_loss": 0.017090823501348495, "grad_norm": 0.1539086550474167, "global_step": 24248, "epoch": 228, "lr": 5.665166081745006e-05} +{"train_loss": 0.014709357172250748, "grad_norm": 0.07052146643400192, "global_step": 24249, "epoch": 228, "lr": 5.6648723382315715e-05} +{"train_loss": 0.014936208724975586, "grad_norm": 0.10452849417924881, "global_step": 24250, "epoch": 228, "lr": 5.664578592382066e-05} +{"train_loss": 0.015654321759939194, "grad_norm": 0.06901899725198746, "global_step": 24251, "epoch": 228, "lr": 5.66428484419752e-05} +{"train_loss": 0.01588868722319603, "grad_norm": 0.11872482299804688, "global_step": 24252, "epoch": 228, "lr": 5.6639910936789654e-05} +{"train_loss": 0.01592380367219448, "grad_norm": 0.08987695723772049, "global_step": 24253, "epoch": 228, "lr": 5.6636973408274366e-05} +{"train_loss": 0.0175173357129097, "grad_norm": 0.05039278417825699, "global_step": 24254, "epoch": 228, "lr": 5.663403585643962e-05} +{"train_loss": 0.013716025277972221, "grad_norm": 0.06821783632040024, "global_step": 24255, "epoch": 228, "lr": 5.663109828129576e-05} +{"train_loss": 0.014127489179372787, "grad_norm": 0.09684760868549347, "global_step": 24256, "epoch": 228, "lr": 5.6628160682853125e-05} +{"train_loss": 0.015037713572382927, "grad_norm": 0.06211910396814346, "global_step": 24257, "epoch": 228, "lr": 5.662522306112199e-05} +{"train_loss": 0.012266721576452255, "grad_norm": 0.09608496725559235, "global_step": 24258, "epoch": 228, "lr": 5.662228541611271e-05} +{"train_loss": 0.016348619014024734, "grad_norm": 0.07713185250759125, "global_step": 24259, "epoch": 228, "lr": 5.661934774783561e-05} +{"train_loss": 0.015429620631039143, "grad_norm": 0.06421364098787308, "global_step": 24260, "epoch": 228, "lr": 5.661641005630098e-05} +{"train_loss": 0.01643540896475315, "grad_norm": 0.07488812506198883, "global_step": 24261, "epoch": 228, "lr": 5.661347234151918e-05} +{"train_loss": 0.015437016263604164, "grad_norm": 0.0881853848695755, "global_step": 24262, "epoch": 228, "lr": 5.661053460350052e-05} +{"train_loss": 0.016983021050691605, "grad_norm": 0.08833356201648712, "global_step": 24263, "epoch": 228, "lr": 5.6607596842255294e-05} +{"train_loss": 0.016162186861038208, "grad_norm": 0.15330073237419128, "global_step": 24264, "epoch": 228, "lr": 5.660465905779385e-05} +{"train_loss": 0.015339318662881851, "grad_norm": 0.09547054022550583, "global_step": 24265, "epoch": 228, "lr": 5.660172125012653e-05} +{"train_loss": 0.014968041330575943, "grad_norm": 0.0610269159078598, "global_step": 24266, "epoch": 228, "lr": 5.6598783419263614e-05} +{"train_loss": 0.01450091041624546, "grad_norm": 0.0742325484752655, "global_step": 24267, "epoch": 228, "lr": 5.6595845565215444e-05} +{"train_loss": 0.015665428712964058, "grad_norm": 0.06962946802377701, "global_step": 24268, "epoch": 228, "lr": 5.6592907687992344e-05} +{"train_loss": 0.01352682150900364, "grad_norm": 0.06675846874713898, "global_step": 24269, "epoch": 228, "lr": 5.658996978760462e-05} +{"train_loss": 0.01717280223965645, "grad_norm": 0.13797704875469208, "global_step": 24270, "epoch": 228, "lr": 5.658703186406262e-05} +{"train_loss": 0.017357122153043747, "grad_norm": 0.12265873700380325, "global_step": 24271, "epoch": 228, "lr": 5.658409391737667e-05} +{"train_loss": 0.016584832221269608, "grad_norm": 0.1256847083568573, "global_step": 24272, "epoch": 228, "lr": 5.6581155947557066e-05} +{"train_loss": 0.01570355191053647, "grad_norm": 0.07629803568124771, "global_step": 24273, "epoch": 228, "lr": 5.6578217954614134e-05, "val_loss": 0.01905321329832077} +{"train_loss": 0.01612158492207527, "grad_norm": 0.05692785233259201, "global_step": 24274, "epoch": 229, "lr": 5.65752799385582e-05} +{"train_loss": 0.016981258988380432, "grad_norm": 0.054961904883384705, "global_step": 24275, "epoch": 229, "lr": 5.6572341899399606e-05} +{"train_loss": 0.015587856993079185, "grad_norm": 0.07598694413900375, "global_step": 24276, "epoch": 229, "lr": 5.656940383714865e-05} +{"train_loss": 0.016393594443798065, "grad_norm": 0.07393114268779755, "global_step": 24277, "epoch": 229, "lr": 5.6566465751815675e-05} +{"train_loss": 0.017994508147239685, "grad_norm": 0.0690297782421112, "global_step": 24278, "epoch": 229, "lr": 5.6563527643411e-05} +{"train_loss": 0.014237788505852222, "grad_norm": 0.16189706325531006, "global_step": 24279, "epoch": 229, "lr": 5.656058951194493e-05} +{"train_loss": 0.015861134976148605, "grad_norm": 0.08215107023715973, "global_step": 24280, "epoch": 229, "lr": 5.655765135742781e-05} +{"train_loss": 0.01738952472805977, "grad_norm": 0.11914549767971039, "global_step": 24281, "epoch": 229, "lr": 5.6554713179869945e-05} +{"train_loss": 0.016512302681803703, "grad_norm": 0.11610062420368195, "global_step": 24282, "epoch": 229, "lr": 5.655177497928168e-05} +{"train_loss": 0.016827503219246864, "grad_norm": 0.07846724987030029, "global_step": 24283, "epoch": 229, "lr": 5.654883675567332e-05} +{"train_loss": 0.01573164202272892, "grad_norm": 0.06267593055963516, "global_step": 24284, "epoch": 229, "lr": 5.65458985090552e-05} +{"train_loss": 0.01732623763382435, "grad_norm": 0.15547198057174683, "global_step": 24285, "epoch": 229, "lr": 5.6542960239437635e-05} +{"train_loss": 0.015393981710076332, "grad_norm": 0.05900339037179947, "global_step": 24286, "epoch": 229, "lr": 5.6540021946830953e-05} +{"train_loss": 0.01227467879652977, "grad_norm": 0.08613725006580353, "global_step": 24287, "epoch": 229, "lr": 5.653708363124548e-05} +{"train_loss": 0.018684808164834976, "grad_norm": 0.11216239631175995, "global_step": 24288, "epoch": 229, "lr": 5.6534145292691545e-05} +{"train_loss": 0.014946148730814457, "grad_norm": 0.05048000440001488, "global_step": 24289, "epoch": 229, "lr": 5.653120693117946e-05} +{"train_loss": 0.01645437441766262, "grad_norm": 0.080658458173275, "global_step": 24290, "epoch": 229, "lr": 5.6528268546719555e-05} +{"train_loss": 0.018903780728578568, "grad_norm": 0.12243261188268661, "global_step": 24291, "epoch": 229, "lr": 5.652533013932214e-05} +{"train_loss": 0.017811380326747894, "grad_norm": 0.08144727349281311, "global_step": 24292, "epoch": 229, "lr": 5.6522391708997566e-05} +{"train_loss": 0.013928212225437164, "grad_norm": 0.06807393580675125, "global_step": 24293, "epoch": 229, "lr": 5.651945325575614e-05} +{"train_loss": 0.015884246677160263, "grad_norm": 0.14555516839027405, "global_step": 24294, "epoch": 229, "lr": 5.651651477960819e-05} +{"train_loss": 0.012943804264068604, "grad_norm": 0.05906856432557106, "global_step": 24295, "epoch": 229, "lr": 5.651357628056405e-05} +{"train_loss": 0.017343346029520035, "grad_norm": 0.06660997122526169, "global_step": 24296, "epoch": 229, "lr": 5.651063775863402e-05} +{"train_loss": 0.014217112213373184, "grad_norm": 0.071440689265728, "global_step": 24297, "epoch": 229, "lr": 5.650769921382846e-05} +{"train_loss": 0.013618545606732368, "grad_norm": 0.10637351125478745, "global_step": 24298, "epoch": 229, "lr": 5.650476064615766e-05} +{"train_loss": 0.01707407645881176, "grad_norm": 0.081947460770607, "global_step": 24299, "epoch": 229, "lr": 5.650182205563197e-05} +{"train_loss": 0.015569393523037434, "grad_norm": 0.06126946955919266, "global_step": 24300, "epoch": 229, "lr": 5.6498883442261695e-05} +{"train_loss": 0.013912108726799488, "grad_norm": 0.06470692902803421, "global_step": 24301, "epoch": 229, "lr": 5.6495944806057175e-05} +{"train_loss": 0.01220429502427578, "grad_norm": 0.08483187854290009, "global_step": 24302, "epoch": 229, "lr": 5.6493006147028723e-05} +{"train_loss": 0.016459260135889053, "grad_norm": 0.06486830860376358, "global_step": 24303, "epoch": 229, "lr": 5.649006746518668e-05} +{"train_loss": 0.0176386721432209, "grad_norm": 0.08489709347486496, "global_step": 24304, "epoch": 229, "lr": 5.648712876054136e-05} +{"train_loss": 0.015407171100378036, "grad_norm": 0.046326298266649246, "global_step": 24305, "epoch": 229, "lr": 5.648419003310308e-05} +{"train_loss": 0.016289211809635162, "grad_norm": 0.11783294379711151, "global_step": 24306, "epoch": 229, "lr": 5.6481251282882194e-05} +{"train_loss": 0.017358556389808655, "grad_norm": 0.07580389082431793, "global_step": 24307, "epoch": 229, "lr": 5.647831250988901e-05} +{"train_loss": 0.01463501900434494, "grad_norm": 0.08736163377761841, "global_step": 24308, "epoch": 229, "lr": 5.647537371413384e-05} +{"train_loss": 0.014944920316338539, "grad_norm": 0.09605914354324341, "global_step": 24309, "epoch": 229, "lr": 5.6472434895627036e-05} +{"train_loss": 0.01376961637288332, "grad_norm": 0.09521032869815826, "global_step": 24310, "epoch": 229, "lr": 5.646949605437889e-05} +{"train_loss": 0.014586656354367733, "grad_norm": 0.11334042996168137, "global_step": 24311, "epoch": 229, "lr": 5.646655719039976e-05} +{"train_loss": 0.016132915392518044, "grad_norm": 0.07098987698554993, "global_step": 24312, "epoch": 229, "lr": 5.646361830369996e-05} +{"train_loss": 0.017064280807971954, "grad_norm": 0.08582428842782974, "global_step": 24313, "epoch": 229, "lr": 5.646067939428983e-05} +{"train_loss": 0.015349524095654488, "grad_norm": 0.12201662361621857, "global_step": 24314, "epoch": 229, "lr": 5.645774046217967e-05} +{"train_loss": 0.013435639441013336, "grad_norm": 0.05218212679028511, "global_step": 24315, "epoch": 229, "lr": 5.6454801507379806e-05} +{"train_loss": 0.014713884331285954, "grad_norm": 0.11310609430074692, "global_step": 24316, "epoch": 229, "lr": 5.64518625299006e-05} +{"train_loss": 0.015497811138629913, "grad_norm": 0.09056118875741959, "global_step": 24317, "epoch": 229, "lr": 5.644892352975234e-05} +{"train_loss": 0.014613795094192028, "grad_norm": 0.09590451419353485, "global_step": 24318, "epoch": 229, "lr": 5.644598450694537e-05} +{"train_loss": 0.016295816749334335, "grad_norm": 0.07482486218214035, "global_step": 24319, "epoch": 229, "lr": 5.6443045461490016e-05} +{"train_loss": 0.01766999438405037, "grad_norm": 0.056008633226156235, "global_step": 24320, "epoch": 229, "lr": 5.64401063933966e-05} +{"train_loss": 0.015287930145859718, "grad_norm": 0.07853423058986664, "global_step": 24321, "epoch": 229, "lr": 5.643716730267545e-05} +{"train_loss": 0.01752035692334175, "grad_norm": 0.05950506404042244, "global_step": 24322, "epoch": 229, "lr": 5.643422818933689e-05} +{"train_loss": 0.01637117564678192, "grad_norm": 0.08255747705698013, "global_step": 24323, "epoch": 229, "lr": 5.643128905339126e-05} +{"train_loss": 0.014210689812898636, "grad_norm": 0.05895749852061272, "global_step": 24324, "epoch": 229, "lr": 5.6428349894848876e-05} +{"train_loss": 0.015731683000922203, "grad_norm": 0.07461091130971909, "global_step": 24325, "epoch": 229, "lr": 5.642541071372006e-05} +{"train_loss": 0.016103068366646767, "grad_norm": 0.13632558286190033, "global_step": 24326, "epoch": 229, "lr": 5.642247151001515e-05} +{"train_loss": 0.014595652930438519, "grad_norm": 0.08659520000219345, "global_step": 24327, "epoch": 229, "lr": 5.641953228374446e-05} +{"train_loss": 0.01598823070526123, "grad_norm": 0.0559062696993351, "global_step": 24328, "epoch": 229, "lr": 5.641659303491833e-05} +{"train_loss": 0.01793050952255726, "grad_norm": 0.07416117191314697, "global_step": 24329, "epoch": 229, "lr": 5.641365376354709e-05} +{"train_loss": 0.016389477998018265, "grad_norm": 0.07554947584867477, "global_step": 24330, "epoch": 229, "lr": 5.641071446964106e-05} +{"train_loss": 0.013247901573777199, "grad_norm": 0.07084835320711136, "global_step": 24331, "epoch": 229, "lr": 5.6407775153210564e-05} +{"train_loss": 0.013154184445738792, "grad_norm": 0.059747371822595596, "global_step": 24332, "epoch": 229, "lr": 5.640483581426591e-05} +{"train_loss": 0.016718465834856033, "grad_norm": 0.08323008567094803, "global_step": 24333, "epoch": 229, "lr": 5.640189645281748e-05} +{"train_loss": 0.015775788575410843, "grad_norm": 0.07766791433095932, "global_step": 24334, "epoch": 229, "lr": 5.6398957068875545e-05} +{"train_loss": 0.014833490364253521, "grad_norm": 0.07700232416391373, "global_step": 24335, "epoch": 229, "lr": 5.639601766245048e-05} +{"train_loss": 0.014862796291708946, "grad_norm": 0.059081871062517166, "global_step": 24336, "epoch": 229, "lr": 5.6393078233552585e-05} +{"train_loss": 0.014221124351024628, "grad_norm": 0.15584169328212738, "global_step": 24337, "epoch": 229, "lr": 5.6390138782192194e-05} +{"train_loss": 0.01647375151515007, "grad_norm": 0.077059805393219, "global_step": 24338, "epoch": 229, "lr": 5.638719930837962e-05} +{"train_loss": 0.018129266798496246, "grad_norm": 0.10670394450426102, "global_step": 24339, "epoch": 229, "lr": 5.638425981212522e-05} +{"train_loss": 0.0159919373691082, "grad_norm": 0.062214016914367676, "global_step": 24340, "epoch": 229, "lr": 5.63813202934393e-05} +{"train_loss": 0.014608262106776237, "grad_norm": 0.1098688617348671, "global_step": 24341, "epoch": 229, "lr": 5.6378380752332205e-05} +{"train_loss": 0.01615985855460167, "grad_norm": 0.06312716007232666, "global_step": 24342, "epoch": 229, "lr": 5.637544118881425e-05} +{"train_loss": 0.0146493474021554, "grad_norm": 0.0685943141579628, "global_step": 24343, "epoch": 229, "lr": 5.637250160289576e-05} +{"train_loss": 0.018467191606760025, "grad_norm": 0.08631492406129837, "global_step": 24344, "epoch": 229, "lr": 5.636956199458707e-05} +{"train_loss": 0.01841379702091217, "grad_norm": 0.08971443772315979, "global_step": 24345, "epoch": 229, "lr": 5.6366622363898524e-05} +{"train_loss": 0.014974335208535194, "grad_norm": 0.05908212438225746, "global_step": 24346, "epoch": 229, "lr": 5.6363682710840425e-05} +{"train_loss": 0.015553204342722893, "grad_norm": 0.08740001171827316, "global_step": 24347, "epoch": 229, "lr": 5.63607430354231e-05} +{"train_loss": 0.015254426747560501, "grad_norm": 0.06262015551328659, "global_step": 24348, "epoch": 229, "lr": 5.6357803337656924e-05} +{"train_loss": 0.014526508748531342, "grad_norm": 0.08639254420995712, "global_step": 24349, "epoch": 229, "lr": 5.6354863617552156e-05} +{"train_loss": 0.015907451510429382, "grad_norm": 0.08351625502109528, "global_step": 24350, "epoch": 229, "lr": 5.635192387511919e-05} +{"train_loss": 0.019512906670570374, "grad_norm": 0.14080902934074402, "global_step": 24351, "epoch": 229, "lr": 5.6348984110368307e-05} +{"train_loss": 0.01370551623404026, "grad_norm": 0.05799974128603935, "global_step": 24352, "epoch": 229, "lr": 5.634604432330985e-05} +{"train_loss": 0.0157442856580019, "grad_norm": 0.12144413590431213, "global_step": 24353, "epoch": 229, "lr": 5.634310451395417e-05} +{"train_loss": 0.016363946720957756, "grad_norm": 0.06541028618812561, "global_step": 24354, "epoch": 229, "lr": 5.634016468231158e-05} +{"train_loss": 0.01562723144888878, "grad_norm": 0.08303461968898773, "global_step": 24355, "epoch": 229, "lr": 5.63372248283924e-05} +{"train_loss": 0.01411503553390503, "grad_norm": 0.1061309352517128, "global_step": 24356, "epoch": 229, "lr": 5.633428495220697e-05} +{"train_loss": 0.015424638986587524, "grad_norm": 0.0526629239320755, "global_step": 24357, "epoch": 229, "lr": 5.6331345053765617e-05} +{"train_loss": 0.014946438372135162, "grad_norm": 0.05671350657939911, "global_step": 24358, "epoch": 229, "lr": 5.6328405133078676e-05} +{"train_loss": 0.01670657843351364, "grad_norm": 0.08810275048017502, "global_step": 24359, "epoch": 229, "lr": 5.632546519015647e-05} +{"train_loss": 0.015836138278245926, "grad_norm": 0.08224452286958694, "global_step": 24360, "epoch": 229, "lr": 5.6322525225009325e-05} +{"train_loss": 0.017982125282287598, "grad_norm": 0.12342378497123718, "global_step": 24361, "epoch": 229, "lr": 5.631958523764756e-05} +{"train_loss": 0.018398310989141464, "grad_norm": 0.07489565759897232, "global_step": 24362, "epoch": 229, "lr": 5.6316645228081556e-05} +{"train_loss": 0.014991610310971737, "grad_norm": 0.1060427650809288, "global_step": 24363, "epoch": 229, "lr": 5.631370519632157e-05} +{"train_loss": 0.01720389910042286, "grad_norm": 0.0992174968123436, "global_step": 24364, "epoch": 229, "lr": 5.6310765142378e-05} +{"train_loss": 0.014124289155006409, "grad_norm": 0.06949111819267273, "global_step": 24365, "epoch": 229, "lr": 5.630782506626114e-05} +{"train_loss": 0.012732482515275478, "grad_norm": 0.06590016931295395, "global_step": 24366, "epoch": 229, "lr": 5.630488496798132e-05} +{"train_loss": 0.015015355311334133, "grad_norm": 0.10518518835306168, "global_step": 24367, "epoch": 229, "lr": 5.630194484754887e-05} +{"train_loss": 0.017730748280882835, "grad_norm": 0.1736106425523758, "global_step": 24368, "epoch": 229, "lr": 5.6299004704974146e-05} +{"train_loss": 0.013986640609800816, "grad_norm": 0.06489469110965729, "global_step": 24369, "epoch": 229, "lr": 5.629606454026745e-05} +{"train_loss": 0.014513673260807991, "grad_norm": 0.06956357508897781, "global_step": 24370, "epoch": 229, "lr": 5.629312435343911e-05} +{"train_loss": 0.01452292874455452, "grad_norm": 0.06946933269500732, "global_step": 24371, "epoch": 229, "lr": 5.629018414449948e-05} +{"train_loss": 0.016138024628162384, "grad_norm": 0.058979813009500504, "global_step": 24372, "epoch": 229, "lr": 5.628724391345887e-05} +{"train_loss": 0.01670290157198906, "grad_norm": 0.07567928731441498, "global_step": 24373, "epoch": 229, "lr": 5.628430366032762e-05} +{"train_loss": 0.013978477567434311, "grad_norm": 0.05774906650185585, "global_step": 24374, "epoch": 229, "lr": 5.628136338511607e-05} +{"train_loss": 0.015193410217761993, "grad_norm": 0.06530709564685822, "global_step": 24375, "epoch": 229, "lr": 5.627842308783453e-05} +{"train_loss": 0.017455296590924263, "grad_norm": 0.07489001750946045, "global_step": 24376, "epoch": 229, "lr": 5.6275482768493346e-05} +{"train_loss": 0.015167798846960068, "grad_norm": 0.0804462879896164, "global_step": 24377, "epoch": 229, "lr": 5.627254242710286e-05} +{"train_loss": 0.013739069923758507, "grad_norm": 0.08104975521564484, "global_step": 24378, "epoch": 229, "lr": 5.626960206367336e-05} +{"train_loss": 0.015725313589187724, "grad_norm": 0.10593549907207489, "global_step": 24379, "epoch": 229, "lr": 5.6266661678215216e-05, "val_loss": 0.018797222524881363} +{"train_loss": 0.017991166561841965, "grad_norm": 0.058374010026454926, "global_step": 24380, "epoch": 230, "lr": 5.6263721270738746e-05} +{"train_loss": 0.013698292896151543, "grad_norm": 0.055959559977054596, "global_step": 24381, "epoch": 230, "lr": 5.626078084125429e-05} +{"train_loss": 0.014762422069907188, "grad_norm": 0.0872507244348526, "global_step": 24382, "epoch": 230, "lr": 5.6257840389772177e-05} +{"train_loss": 0.01398785412311554, "grad_norm": 0.1252344250679016, "global_step": 24383, "epoch": 230, "lr": 5.6254899916302725e-05} +{"train_loss": 0.014615990221500397, "grad_norm": 0.06532800197601318, "global_step": 24384, "epoch": 230, "lr": 5.625195942085628e-05} +{"train_loss": 0.016404740512371063, "grad_norm": 0.07505743950605392, "global_step": 24385, "epoch": 230, "lr": 5.624901890344315e-05} +{"train_loss": 0.012742397375404835, "grad_norm": 0.11466670781373978, "global_step": 24386, "epoch": 230, "lr": 5.624607836407372e-05} +{"train_loss": 0.01854172721505165, "grad_norm": 0.08300948143005371, "global_step": 24387, "epoch": 230, "lr": 5.624313780275826e-05} +{"train_loss": 0.018333055078983307, "grad_norm": 0.10864506661891937, "global_step": 24388, "epoch": 230, "lr": 5.624019721950713e-05} +{"train_loss": 0.01399632915854454, "grad_norm": 0.06448693573474884, "global_step": 24389, "epoch": 230, "lr": 5.623725661433068e-05} +{"train_loss": 0.016402535140514374, "grad_norm": 0.10616303235292435, "global_step": 24390, "epoch": 230, "lr": 5.623431598723919e-05} +{"train_loss": 0.014718309044837952, "grad_norm": 0.11354493349790573, "global_step": 24391, "epoch": 230, "lr": 5.6231375338243055e-05} +{"train_loss": 0.014553281478583813, "grad_norm": 0.13157223165035248, "global_step": 24392, "epoch": 230, "lr": 5.622843466735257e-05} +{"train_loss": 0.014403912238776684, "grad_norm": 0.11995827406644821, "global_step": 24393, "epoch": 230, "lr": 5.622549397457807e-05} +{"train_loss": 0.01590728759765625, "grad_norm": 0.07166454195976257, "global_step": 24394, "epoch": 230, "lr": 5.6222553259929886e-05} +{"train_loss": 0.014133362099528313, "grad_norm": 0.1691320240497589, "global_step": 24395, "epoch": 230, "lr": 5.621961252341838e-05} +{"train_loss": 0.0149341756477952, "grad_norm": 0.05296597629785538, "global_step": 24396, "epoch": 230, "lr": 5.621667176505383e-05} +{"train_loss": 0.014363426715135574, "grad_norm": 0.07477577030658722, "global_step": 24397, "epoch": 230, "lr": 5.621373098484661e-05} +{"train_loss": 0.01636572927236557, "grad_norm": 0.1530318558216095, "global_step": 24398, "epoch": 230, "lr": 5.621079018280705e-05} +{"train_loss": 0.014669498428702354, "grad_norm": 0.09106464684009552, "global_step": 24399, "epoch": 230, "lr": 5.620784935894547e-05} +{"train_loss": 0.017755862325429916, "grad_norm": 0.09567330777645111, "global_step": 24400, "epoch": 230, "lr": 5.6204908513272194e-05} +{"train_loss": 0.014198324643075466, "grad_norm": 0.10653755813837051, "global_step": 24401, "epoch": 230, "lr": 5.6201967645797595e-05} +{"train_loss": 0.014155313372612, "grad_norm": 0.07862454652786255, "global_step": 24402, "epoch": 230, "lr": 5.619902675653195e-05} +{"train_loss": 0.018363507464528084, "grad_norm": 0.11773822456598282, "global_step": 24403, "epoch": 230, "lr": 5.619608584548565e-05} +{"train_loss": 0.01567796990275383, "grad_norm": 0.12859876453876495, "global_step": 24404, "epoch": 230, "lr": 5.619314491266897e-05} +{"train_loss": 0.017129771411418915, "grad_norm": 0.14484651386737823, "global_step": 24405, "epoch": 230, "lr": 5.619020395809228e-05} +{"train_loss": 0.016348328441381454, "grad_norm": 0.079419806599617, "global_step": 24406, "epoch": 230, "lr": 5.618726298176593e-05} +{"train_loss": 0.015412398613989353, "grad_norm": 0.06683651357889175, "global_step": 24407, "epoch": 230, "lr": 5.6184321983700196e-05} +{"train_loss": 0.014177734963595867, "grad_norm": 0.0611412450671196, "global_step": 24408, "epoch": 230, "lr": 5.618138096390546e-05} +{"train_loss": 0.014319262467324734, "grad_norm": 0.07077926397323608, "global_step": 24409, "epoch": 230, "lr": 5.6178439922392036e-05} +{"train_loss": 0.016622040420770645, "grad_norm": 0.06812625378370285, "global_step": 24410, "epoch": 230, "lr": 5.6175498859170264e-05} +{"train_loss": 0.016395261511206627, "grad_norm": 0.07550213485956192, "global_step": 24411, "epoch": 230, "lr": 5.617255777425047e-05} +{"train_loss": 0.014931214973330498, "grad_norm": 0.07439503073692322, "global_step": 24412, "epoch": 230, "lr": 5.6169616667643e-05} +{"train_loss": 0.017492089420557022, "grad_norm": 0.12190170586109161, "global_step": 24413, "epoch": 230, "lr": 5.6166675539358174e-05} +{"train_loss": 0.01355660893023014, "grad_norm": 0.051225319504737854, "global_step": 24414, "epoch": 230, "lr": 5.616373438940632e-05} +{"train_loss": 0.017277495935559273, "grad_norm": 0.06470341980457306, "global_step": 24415, "epoch": 230, "lr": 5.6160793217797806e-05} +{"train_loss": 0.01606704853475094, "grad_norm": 0.14399276673793793, "global_step": 24416, "epoch": 230, "lr": 5.615785202454294e-05} +{"train_loss": 0.0162773746997118, "grad_norm": 0.14332327246665955, "global_step": 24417, "epoch": 230, "lr": 5.615491080965204e-05} +{"train_loss": 0.017127368599176407, "grad_norm": 0.08629626035690308, "global_step": 24418, "epoch": 230, "lr": 5.615196957313549e-05} +{"train_loss": 0.015251509845256805, "grad_norm": 0.05269043520092964, "global_step": 24419, "epoch": 230, "lr": 5.614902831500356e-05} +{"train_loss": 0.014091282151639462, "grad_norm": 0.1922912448644638, "global_step": 24420, "epoch": 230, "lr": 5.614608703526665e-05} +{"train_loss": 0.01557167712599039, "grad_norm": 0.12293822318315506, "global_step": 24421, "epoch": 230, "lr": 5.6143145733935046e-05} +{"train_loss": 0.01418595016002655, "grad_norm": 0.0660133808851242, "global_step": 24422, "epoch": 230, "lr": 5.6140204411019105e-05} +{"train_loss": 0.01559862494468689, "grad_norm": 0.06958970427513123, "global_step": 24423, "epoch": 230, "lr": 5.613726306652915e-05} +{"train_loss": 0.01808883249759674, "grad_norm": 0.10334974527359009, "global_step": 24424, "epoch": 230, "lr": 5.613432170047553e-05} +{"train_loss": 0.014536661095917225, "grad_norm": 0.07627446949481964, "global_step": 24425, "epoch": 230, "lr": 5.613138031286857e-05} +{"train_loss": 0.017429184168577194, "grad_norm": 0.08695012331008911, "global_step": 24426, "epoch": 230, "lr": 5.6128438903718594e-05} +{"train_loss": 0.014187220484018326, "grad_norm": 0.05160696059465408, "global_step": 24427, "epoch": 230, "lr": 5.612549747303597e-05} +{"train_loss": 0.01592784747481346, "grad_norm": 0.05688219517469406, "global_step": 24428, "epoch": 230, "lr": 5.6122556020830997e-05} +{"train_loss": 0.014573478139936924, "grad_norm": 0.09828431904315948, "global_step": 24429, "epoch": 230, "lr": 5.611961454711402e-05} +{"train_loss": 0.018374715000391006, "grad_norm": 0.088251493871212, "global_step": 24430, "epoch": 230, "lr": 5.6116673051895395e-05} +{"train_loss": 0.013149849139153957, "grad_norm": 0.07387787103652954, "global_step": 24431, "epoch": 230, "lr": 5.611373153518543e-05} +{"train_loss": 0.01504422351717949, "grad_norm": 0.054693225771188736, "global_step": 24432, "epoch": 230, "lr": 5.6110789996994474e-05} +{"train_loss": 0.016859322786331177, "grad_norm": 0.05680401995778084, "global_step": 24433, "epoch": 230, "lr": 5.610784843733286e-05} +{"train_loss": 0.015973463654518127, "grad_norm": 0.0777805745601654, "global_step": 24434, "epoch": 230, "lr": 5.6104906856210924e-05} +{"train_loss": 0.01936187781393528, "grad_norm": 0.08352038264274597, "global_step": 24435, "epoch": 230, "lr": 5.6101965253639e-05} +{"train_loss": 0.015301162376999855, "grad_norm": 0.05403298884630203, "global_step": 24436, "epoch": 230, "lr": 5.609902362962742e-05} +{"train_loss": 0.015201736241579056, "grad_norm": 0.10732530802488327, "global_step": 24437, "epoch": 230, "lr": 5.6096081984186536e-05} +{"train_loss": 0.014626851305365562, "grad_norm": 0.07018318772315979, "global_step": 24438, "epoch": 230, "lr": 5.609314031732665e-05} +{"train_loss": 0.016552528366446495, "grad_norm": 0.0652134120464325, "global_step": 24439, "epoch": 230, "lr": 5.6090198629058134e-05} +{"train_loss": 0.019059214740991592, "grad_norm": 0.1589212417602539, "global_step": 24440, "epoch": 230, "lr": 5.6087256919391306e-05} +{"train_loss": 0.015705831348896027, "grad_norm": 0.06506563723087311, "global_step": 24441, "epoch": 230, "lr": 5.608431518833649e-05} +{"train_loss": 0.015947682783007622, "grad_norm": 0.10411982983350754, "global_step": 24442, "epoch": 230, "lr": 5.608137343590406e-05} +{"train_loss": 0.01753871887922287, "grad_norm": 0.12785287201404572, "global_step": 24443, "epoch": 230, "lr": 5.6078431662104316e-05} +{"train_loss": 0.01614898443222046, "grad_norm": 0.07992104440927505, "global_step": 24444, "epoch": 230, "lr": 5.6075489866947605e-05} +{"train_loss": 0.016651105135679245, "grad_norm": 0.08070073276758194, "global_step": 24445, "epoch": 230, "lr": 5.607254805044428e-05} +{"train_loss": 0.01545169111341238, "grad_norm": 0.13618920743465424, "global_step": 24446, "epoch": 230, "lr": 5.606960621260464e-05} +{"train_loss": 0.016241658478975296, "grad_norm": 0.08341572433710098, "global_step": 24447, "epoch": 230, "lr": 5.606666435343906e-05} +{"train_loss": 0.01620555855333805, "grad_norm": 0.07779285311698914, "global_step": 24448, "epoch": 230, "lr": 5.6063722472957844e-05} +{"train_loss": 0.014776717871427536, "grad_norm": 0.08104748278856277, "global_step": 24449, "epoch": 230, "lr": 5.6060780571171354e-05} +{"train_loss": 0.017599981278181076, "grad_norm": 0.1124182790517807, "global_step": 24450, "epoch": 230, "lr": 5.605783864808992e-05} +{"train_loss": 0.01466911006718874, "grad_norm": 0.07005314528942108, "global_step": 24451, "epoch": 230, "lr": 5.605489670372387e-05} +{"train_loss": 0.01714770495891571, "grad_norm": 0.07790648192167282, "global_step": 24452, "epoch": 230, "lr": 5.605195473808356e-05} +{"train_loss": 0.013695128262043, "grad_norm": 0.056343235075473785, "global_step": 24453, "epoch": 230, "lr": 5.6049012751179295e-05} +{"train_loss": 0.015276163816452026, "grad_norm": 0.06324636936187744, "global_step": 24454, "epoch": 230, "lr": 5.604607074302144e-05} +{"train_loss": 0.015255330130457878, "grad_norm": 0.05139384791254997, "global_step": 24455, "epoch": 230, "lr": 5.604312871362031e-05} +{"train_loss": 0.014124391600489616, "grad_norm": 0.1273026019334793, "global_step": 24456, "epoch": 230, "lr": 5.604018666298627e-05} +{"train_loss": 0.019563322886824608, "grad_norm": 0.12466072291135788, "global_step": 24457, "epoch": 230, "lr": 5.6037244591129637e-05} +{"train_loss": 0.01618192344903946, "grad_norm": 0.07929325848817825, "global_step": 24458, "epoch": 230, "lr": 5.603430249806073e-05} +{"train_loss": 0.015559552237391472, "grad_norm": 0.15972736477851868, "global_step": 24459, "epoch": 230, "lr": 5.6031360383789935e-05} +{"train_loss": 0.015586732886731625, "grad_norm": 0.18356895446777344, "global_step": 24460, "epoch": 230, "lr": 5.602841824832754e-05} +{"train_loss": 0.016407795250415802, "grad_norm": 0.09062021225690842, "global_step": 24461, "epoch": 230, "lr": 5.6025476091683926e-05} +{"train_loss": 0.014867734163999557, "grad_norm": 0.1013147383928299, "global_step": 24462, "epoch": 230, "lr": 5.60225339138694e-05} +{"train_loss": 0.013759016990661621, "grad_norm": 0.08760762214660645, "global_step": 24463, "epoch": 230, "lr": 5.6019591714894313e-05} +{"train_loss": 0.017201563343405724, "grad_norm": 0.12744447588920593, "global_step": 24464, "epoch": 230, "lr": 5.601664949476899e-05} +{"train_loss": 0.01818513125181198, "grad_norm": 0.08034668117761612, "global_step": 24465, "epoch": 230, "lr": 5.601370725350379e-05} +{"train_loss": 0.013555851764976978, "grad_norm": 0.07593680918216705, "global_step": 24466, "epoch": 230, "lr": 5.6010764991109034e-05} +{"train_loss": 0.014800265431404114, "grad_norm": 0.08462294936180115, "global_step": 24467, "epoch": 230, "lr": 5.600782270759506e-05} +{"train_loss": 0.016846809536218643, "grad_norm": 0.07769928127527237, "global_step": 24468, "epoch": 230, "lr": 5.600488040297221e-05} +{"train_loss": 0.0170283280313015, "grad_norm": 0.12788864970207214, "global_step": 24469, "epoch": 230, "lr": 5.600193807725083e-05} +{"train_loss": 0.013418175280094147, "grad_norm": 0.1626013219356537, "global_step": 24470, "epoch": 230, "lr": 5.599899573044123e-05} +{"train_loss": 0.014012748375535011, "grad_norm": 0.08400387316942215, "global_step": 24471, "epoch": 230, "lr": 5.5996053362553793e-05} +{"train_loss": 0.01626330241560936, "grad_norm": 0.08229431509971619, "global_step": 24472, "epoch": 230, "lr": 5.5993110973598804e-05} +{"train_loss": 0.0136577058583498, "grad_norm": 0.06910218298435211, "global_step": 24473, "epoch": 230, "lr": 5.599016856358664e-05} +{"train_loss": 0.014447493478655815, "grad_norm": 0.07546856254339218, "global_step": 24474, "epoch": 230, "lr": 5.5987226132527645e-05} +{"train_loss": 0.014254092238843441, "grad_norm": 0.07392197847366333, "global_step": 24475, "epoch": 230, "lr": 5.598428368043212e-05} +{"train_loss": 0.016440724954009056, "grad_norm": 0.07797513157129288, "global_step": 24476, "epoch": 230, "lr": 5.5981341207310444e-05} +{"train_loss": 0.016049116849899292, "grad_norm": 0.14334546029567719, "global_step": 24477, "epoch": 230, "lr": 5.597839871317292e-05} +{"train_loss": 0.014790167100727558, "grad_norm": 0.1334647685289383, "global_step": 24478, "epoch": 230, "lr": 5.597545619802992e-05} +{"train_loss": 0.014494242146611214, "grad_norm": 0.09352901577949524, "global_step": 24479, "epoch": 230, "lr": 5.597251366189175e-05} +{"train_loss": 0.016383573412895203, "grad_norm": 0.09490363299846649, "global_step": 24480, "epoch": 230, "lr": 5.596957110476877e-05} +{"train_loss": 0.01668797805905342, "grad_norm": 0.05864353105425835, "global_step": 24481, "epoch": 230, "lr": 5.5966628526671314e-05} +{"train_loss": 0.014257851988077164, "grad_norm": 0.08385248482227325, "global_step": 24482, "epoch": 230, "lr": 5.596368592760971e-05} +{"train_loss": 0.019017459824681282, "grad_norm": 0.09194085747003555, "global_step": 24483, "epoch": 230, "lr": 5.596074330759432e-05} +{"train_loss": 0.018001791089773178, "grad_norm": 0.18820308148860931, "global_step": 24484, "epoch": 230, "lr": 5.595780066663546e-05} +{"train_loss": 0.015728616665275592, "grad_norm": 0.11763937771320343, "global_step": 24485, "epoch": 230, "lr": 5.595485800474349e-05, "train_batch_mse_error": 0.014234943315386772, "val_loss": 0.019544899463653564} +{"train_loss": 0.014303112402558327, "grad_norm": 0.05255540460348129, "global_step": 24486, "epoch": 231, "lr": 5.595191532192874e-05} +{"train_loss": 0.014952711760997772, "grad_norm": 0.13323718309402466, "global_step": 24487, "epoch": 231, "lr": 5.5948972618201534e-05} +{"train_loss": 0.01649368554353714, "grad_norm": 0.11969821900129318, "global_step": 24488, "epoch": 231, "lr": 5.594602989357224e-05} +{"train_loss": 0.017040487378835678, "grad_norm": 0.05937320739030838, "global_step": 24489, "epoch": 231, "lr": 5.5943087148051175e-05} +{"train_loss": 0.01604468561708927, "grad_norm": 0.08292064070701599, "global_step": 24490, "epoch": 231, "lr": 5.594014438164868e-05} +{"train_loss": 0.013886434957385063, "grad_norm": 0.0788184180855751, "global_step": 24491, "epoch": 231, "lr": 5.593720159437512e-05} +{"train_loss": 0.013502227142453194, "grad_norm": 0.07714121043682098, "global_step": 24492, "epoch": 231, "lr": 5.59342587862408e-05} +{"train_loss": 0.017154280096292496, "grad_norm": 0.07923654466867447, "global_step": 24493, "epoch": 231, "lr": 5.593131595725608e-05} +{"train_loss": 0.015887700021266937, "grad_norm": 0.17302873730659485, "global_step": 24494, "epoch": 231, "lr": 5.592837310743129e-05} +{"train_loss": 0.018200485035777092, "grad_norm": 0.11742474883794785, "global_step": 24495, "epoch": 231, "lr": 5.5925430236776785e-05} +{"train_loss": 0.014763489365577698, "grad_norm": 0.14001701772212982, "global_step": 24496, "epoch": 231, "lr": 5.59224873453029e-05} +{"train_loss": 0.019761495292186737, "grad_norm": 0.10924455523490906, "global_step": 24497, "epoch": 231, "lr": 5.5919544433019945e-05} +{"train_loss": 0.01545293815433979, "grad_norm": 0.0538940392434597, "global_step": 24498, "epoch": 231, "lr": 5.5916601499938316e-05} +{"train_loss": 0.01393350213766098, "grad_norm": 0.10754627734422684, "global_step": 24499, "epoch": 231, "lr": 5.5913658546068295e-05} +{"train_loss": 0.015201128087937832, "grad_norm": 0.11960075795650482, "global_step": 24500, "epoch": 231, "lr": 5.591071557142026e-05} +{"train_loss": 0.016999326646327972, "grad_norm": 0.09529732912778854, "global_step": 24501, "epoch": 231, "lr": 5.5907772576004546e-05} +{"train_loss": 0.01638122648000717, "grad_norm": 0.10180141776800156, "global_step": 24502, "epoch": 231, "lr": 5.5904829559831485e-05} +{"train_loss": 0.015868686139583588, "grad_norm": 0.13170821964740753, "global_step": 24503, "epoch": 231, "lr": 5.590188652291143e-05} +{"train_loss": 0.01329804677516222, "grad_norm": 0.10870913416147232, "global_step": 24504, "epoch": 231, "lr": 5.589894346525469e-05} +{"train_loss": 0.01369976345449686, "grad_norm": 0.04883098602294922, "global_step": 24505, "epoch": 231, "lr": 5.589600038687164e-05} +{"train_loss": 0.01960112154483795, "grad_norm": 0.134853333234787, "global_step": 24506, "epoch": 231, "lr": 5.589305728777261e-05} +{"train_loss": 0.015549023635685444, "grad_norm": 0.0541236437857151, "global_step": 24507, "epoch": 231, "lr": 5.589011416796794e-05} +{"train_loss": 0.01566752977669239, "grad_norm": 0.06708751618862152, "global_step": 24508, "epoch": 231, "lr": 5.5887171027467974e-05} +{"train_loss": 0.017183827236294746, "grad_norm": 0.0593377910554409, "global_step": 24509, "epoch": 231, "lr": 5.588422786628305e-05} +{"train_loss": 0.015244300477206707, "grad_norm": 0.045834820717573166, "global_step": 24510, "epoch": 231, "lr": 5.58812846844235e-05} +{"train_loss": 0.01433759555220604, "grad_norm": 0.07268347591161728, "global_step": 24511, "epoch": 231, "lr": 5.5878341481899665e-05} +{"train_loss": 0.016774896532297134, "grad_norm": 0.14420700073242188, "global_step": 24512, "epoch": 231, "lr": 5.587539825872192e-05} +{"train_loss": 0.014911586418747902, "grad_norm": 0.06984029710292816, "global_step": 24513, "epoch": 231, "lr": 5.587245501490056e-05} +{"train_loss": 0.013864833861589432, "grad_norm": 0.06340113282203674, "global_step": 24514, "epoch": 231, "lr": 5.5869511750445944e-05} +{"train_loss": 0.016533207148313522, "grad_norm": 0.07489439845085144, "global_step": 24515, "epoch": 231, "lr": 5.586656846536844e-05} +{"train_loss": 0.017207292839884758, "grad_norm": 0.1475619226694107, "global_step": 24516, "epoch": 231, "lr": 5.586362515967835e-05} +{"train_loss": 0.015748456120491028, "grad_norm": 0.07720690220594406, "global_step": 24517, "epoch": 231, "lr": 5.5860681833386034e-05} +{"train_loss": 0.01671181619167328, "grad_norm": 0.0843295305967331, "global_step": 24518, "epoch": 231, "lr": 5.585773848650182e-05} +{"train_loss": 0.014566989615559578, "grad_norm": 0.09327712655067444, "global_step": 24519, "epoch": 231, "lr": 5.585479511903607e-05} +{"train_loss": 0.013022175058722496, "grad_norm": 0.05282987654209137, "global_step": 24520, "epoch": 231, "lr": 5.5851851730999116e-05} +{"train_loss": 0.01549176499247551, "grad_norm": 0.07363691926002502, "global_step": 24521, "epoch": 231, "lr": 5.584890832240131e-05} +{"train_loss": 0.015014243312180042, "grad_norm": 0.10357969999313354, "global_step": 24522, "epoch": 231, "lr": 5.584596489325297e-05} +{"train_loss": 0.018080847337841988, "grad_norm": 0.14463776350021362, "global_step": 24523, "epoch": 231, "lr": 5.584302144356445e-05} +{"train_loss": 0.018920723348855972, "grad_norm": 0.0888637900352478, "global_step": 24524, "epoch": 231, "lr": 5.58400779733461e-05} +{"train_loss": 0.015603131614625454, "grad_norm": 0.1194423958659172, "global_step": 24525, "epoch": 231, "lr": 5.5837134482608264e-05} +{"train_loss": 0.016622204333543777, "grad_norm": 0.086890809237957, "global_step": 24526, "epoch": 231, "lr": 5.583419097136126e-05} +{"train_loss": 0.020281055942177773, "grad_norm": 0.10813620686531067, "global_step": 24527, "epoch": 231, "lr": 5.583124743961546e-05} +{"train_loss": 0.015147748403251171, "grad_norm": 0.09870882332324982, "global_step": 24528, "epoch": 231, "lr": 5.5828303887381176e-05} +{"train_loss": 0.016241978853940964, "grad_norm": 0.04861622303724289, "global_step": 24529, "epoch": 231, "lr": 5.5825360314668776e-05} +{"train_loss": 0.014854775741696358, "grad_norm": 0.11534436047077179, "global_step": 24530, "epoch": 231, "lr": 5.582241672148859e-05} +{"train_loss": 0.015026402659714222, "grad_norm": 0.07275497168302536, "global_step": 24531, "epoch": 231, "lr": 5.581947310785097e-05} +{"train_loss": 0.014246711507439613, "grad_norm": 0.0644616186618805, "global_step": 24532, "epoch": 231, "lr": 5.581652947376625e-05} +{"train_loss": 0.013615930452942848, "grad_norm": 0.10371849685907364, "global_step": 24533, "epoch": 231, "lr": 5.581358581924479e-05} +{"train_loss": 0.015524405054748058, "grad_norm": 0.0904485210776329, "global_step": 24534, "epoch": 231, "lr": 5.58106421442969e-05} +{"train_loss": 0.015443610027432442, "grad_norm": 0.05622000992298126, "global_step": 24535, "epoch": 231, "lr": 5.580769844893292e-05} +{"train_loss": 0.017302431166172028, "grad_norm": 0.1022908166050911, "global_step": 24536, "epoch": 231, "lr": 5.580475473316326e-05} +{"train_loss": 0.01522077340632677, "grad_norm": 0.08217497169971466, "global_step": 24537, "epoch": 231, "lr": 5.5801810996998185e-05} +{"train_loss": 0.013797643594443798, "grad_norm": 0.05662687122821808, "global_step": 24538, "epoch": 231, "lr": 5.579886724044807e-05} +{"train_loss": 0.01438475027680397, "grad_norm": 0.06222042813897133, "global_step": 24539, "epoch": 231, "lr": 5.579592346352327e-05} +{"train_loss": 0.01719808019697666, "grad_norm": 0.0831044539809227, "global_step": 24540, "epoch": 231, "lr": 5.5792979666234105e-05} +{"train_loss": 0.016823282465338707, "grad_norm": 0.10948386043310165, "global_step": 24541, "epoch": 231, "lr": 5.579003584859093e-05} +{"train_loss": 0.016862092539668083, "grad_norm": 0.10405219346284866, "global_step": 24542, "epoch": 231, "lr": 5.578709201060409e-05} +{"train_loss": 0.017562445253133774, "grad_norm": 0.09192352741956711, "global_step": 24543, "epoch": 231, "lr": 5.578414815228393e-05} +{"train_loss": 0.014796679839491844, "grad_norm": 0.08811265975236893, "global_step": 24544, "epoch": 231, "lr": 5.578120427364079e-05} +{"train_loss": 0.013854618184268475, "grad_norm": 0.08303549885749817, "global_step": 24545, "epoch": 231, "lr": 5.5778260374684995e-05} +{"train_loss": 0.015316970646381378, "grad_norm": 0.11941851675510406, "global_step": 24546, "epoch": 231, "lr": 5.577531645542693e-05} +{"train_loss": 0.016339126974344254, "grad_norm": 0.11683104187250137, "global_step": 24547, "epoch": 231, "lr": 5.5772372515876904e-05} +{"train_loss": 0.01567033864557743, "grad_norm": 0.0794186219573021, "global_step": 24548, "epoch": 231, "lr": 5.5769428556045266e-05} +{"train_loss": 0.01598312333226204, "grad_norm": 0.06413298100233078, "global_step": 24549, "epoch": 231, "lr": 5.576648457594237e-05} +{"train_loss": 0.012526347301900387, "grad_norm": 0.06469608843326569, "global_step": 24550, "epoch": 231, "lr": 5.5763540575578544e-05} +{"train_loss": 0.016500158235430717, "grad_norm": 0.07782185077667236, "global_step": 24551, "epoch": 231, "lr": 5.576059655496416e-05} +{"train_loss": 0.01335539948195219, "grad_norm": 0.13697105646133423, "global_step": 24552, "epoch": 231, "lr": 5.575765251410954e-05} +{"train_loss": 0.015320984646677971, "grad_norm": 0.05303184688091278, "global_step": 24553, "epoch": 231, "lr": 5.575470845302504e-05} +{"train_loss": 0.016998963430523872, "grad_norm": 0.1424005776643753, "global_step": 24554, "epoch": 231, "lr": 5.5751764371720983e-05} +{"train_loss": 0.014496690593659878, "grad_norm": 0.101600781083107, "global_step": 24555, "epoch": 231, "lr": 5.5748820270207724e-05} +{"train_loss": 0.014146246947348118, "grad_norm": 0.11795108020305634, "global_step": 24556, "epoch": 231, "lr": 5.5745876148495635e-05} +{"train_loss": 0.01480051688849926, "grad_norm": 0.2213788479566574, "global_step": 24557, "epoch": 231, "lr": 5.5742932006595004e-05} +{"train_loss": 0.015666812658309937, "grad_norm": 0.09372315555810928, "global_step": 24558, "epoch": 231, "lr": 5.5739987844516225e-05} +{"train_loss": 0.015238381922245026, "grad_norm": 0.1092192679643631, "global_step": 24559, "epoch": 231, "lr": 5.5737043662269626e-05} +{"train_loss": 0.015878744423389435, "grad_norm": 0.06590569019317627, "global_step": 24560, "epoch": 231, "lr": 5.5734099459865554e-05} +{"train_loss": 0.017420675605535507, "grad_norm": 0.10109584033489227, "global_step": 24561, "epoch": 231, "lr": 5.5731155237314335e-05} +{"train_loss": 0.017274722456932068, "grad_norm": 0.07281128317117691, "global_step": 24562, "epoch": 231, "lr": 5.572821099462634e-05} +{"train_loss": 0.01580895483493805, "grad_norm": 0.08262613415718079, "global_step": 24563, "epoch": 231, "lr": 5.572526673181191e-05} +{"train_loss": 0.014588972553610802, "grad_norm": 0.07358469814062119, "global_step": 24564, "epoch": 231, "lr": 5.572232244888136e-05} +{"train_loss": 0.014868631027638912, "grad_norm": 0.09411944448947906, "global_step": 24565, "epoch": 231, "lr": 5.571937814584508e-05} +{"train_loss": 0.016937149688601494, "grad_norm": 0.15593977272510529, "global_step": 24566, "epoch": 231, "lr": 5.571643382271339e-05} +{"train_loss": 0.016241490840911865, "grad_norm": 0.10811632126569748, "global_step": 24567, "epoch": 231, "lr": 5.5713489479496614e-05} +{"train_loss": 0.015803884714841843, "grad_norm": 0.07957707345485687, "global_step": 24568, "epoch": 231, "lr": 5.5710545116205146e-05} +{"train_loss": 0.016341570764780045, "grad_norm": 0.1565689742565155, "global_step": 24569, "epoch": 231, "lr": 5.5707600732849296e-05} +{"train_loss": 0.014252914115786552, "grad_norm": 0.09859894961118698, "global_step": 24570, "epoch": 231, "lr": 5.570465632943942e-05} +{"train_loss": 0.015941129997372627, "grad_norm": 0.08673341572284698, "global_step": 24571, "epoch": 231, "lr": 5.5701711905985863e-05} +{"train_loss": 0.019501173868775368, "grad_norm": 0.10490158200263977, "global_step": 24572, "epoch": 231, "lr": 5.569876746249897e-05} +{"train_loss": 0.01454702764749527, "grad_norm": 0.08110351115465164, "global_step": 24573, "epoch": 231, "lr": 5.569582299898909e-05} +{"train_loss": 0.015109702944755554, "grad_norm": 0.10734279453754425, "global_step": 24574, "epoch": 231, "lr": 5.5692878515466565e-05} +{"train_loss": 0.014872826635837555, "grad_norm": 0.11015690863132477, "global_step": 24575, "epoch": 231, "lr": 5.568993401194174e-05} +{"train_loss": 0.016392232850193977, "grad_norm": 0.08037099242210388, "global_step": 24576, "epoch": 231, "lr": 5.5686989488424955e-05} +{"train_loss": 0.015144074335694313, "grad_norm": 0.08345004916191101, "global_step": 24577, "epoch": 231, "lr": 5.568404494492656e-05} +{"train_loss": 0.016811372712254524, "grad_norm": 0.09649480879306793, "global_step": 24578, "epoch": 231, "lr": 5.568110038145692e-05} +{"train_loss": 0.014288464561104774, "grad_norm": 0.07232814282178879, "global_step": 24579, "epoch": 231, "lr": 5.567815579802633e-05} +{"train_loss": 0.015550455078482628, "grad_norm": 0.11331447958946228, "global_step": 24580, "epoch": 231, "lr": 5.567521119464522e-05} +{"train_loss": 0.014259204268455505, "grad_norm": 0.09745226800441742, "global_step": 24581, "epoch": 231, "lr": 5.567226657132384e-05} +{"train_loss": 0.014620911329984665, "grad_norm": 0.13011138141155243, "global_step": 24582, "epoch": 231, "lr": 5.5669321928072604e-05} +{"train_loss": 0.017296474426984787, "grad_norm": 0.16193228960037231, "global_step": 24583, "epoch": 231, "lr": 5.5666377264901826e-05} +{"train_loss": 0.015633028000593185, "grad_norm": 0.05982610955834389, "global_step": 24584, "epoch": 231, "lr": 5.566343258182187e-05} +{"train_loss": 0.013285109773278236, "grad_norm": 0.12623660266399384, "global_step": 24585, "epoch": 231, "lr": 5.5660487878843084e-05} +{"train_loss": 0.018014658242464066, "grad_norm": 0.07193739712238312, "global_step": 24586, "epoch": 231, "lr": 5.565754315597579e-05} +{"train_loss": 0.016827065497636795, "grad_norm": 0.08692209422588348, "global_step": 24587, "epoch": 231, "lr": 5.5654598413230365e-05} +{"train_loss": 0.014093277975916862, "grad_norm": 0.11636753380298615, "global_step": 24588, "epoch": 231, "lr": 5.565165365061713e-05} +{"train_loss": 0.016825566068291664, "grad_norm": 0.10220589488744736, "global_step": 24589, "epoch": 231, "lr": 5.564870886814644e-05} +{"train_loss": 0.01668233796954155, "grad_norm": 0.08456601202487946, "global_step": 24590, "epoch": 231, "lr": 5.5645764065828646e-05} +{"train_loss": 0.015730768705735792, "grad_norm": 0.08885445445775986, "global_step": 24591, "epoch": 231, "lr": 5.564281924367408e-05, "val_loss": 0.01947629451751709} +{"train_loss": 0.012304709292948246, "grad_norm": 0.07494348287582397, "global_step": 24592, "epoch": 232, "lr": 5.563987440169313e-05} +{"train_loss": 0.01936580426990986, "grad_norm": 0.05902280658483505, "global_step": 24593, "epoch": 232, "lr": 5.5636929539896085e-05} +{"train_loss": 0.015829458832740784, "grad_norm": 0.0978493019938469, "global_step": 24594, "epoch": 232, "lr": 5.563398465829334e-05} +{"train_loss": 0.01569417491555214, "grad_norm": 0.07401056587696075, "global_step": 24595, "epoch": 232, "lr": 5.563103975689522e-05} +{"train_loss": 0.017260778695344925, "grad_norm": 0.05782538652420044, "global_step": 24596, "epoch": 232, "lr": 5.562809483571206e-05} +{"train_loss": 0.01280301809310913, "grad_norm": 0.18557238578796387, "global_step": 24597, "epoch": 232, "lr": 5.562514989475424e-05} +{"train_loss": 0.015567908994853497, "grad_norm": 0.07930974662303925, "global_step": 24598, "epoch": 232, "lr": 5.562220493403208e-05} +{"train_loss": 0.014387676492333412, "grad_norm": 0.1064373031258583, "global_step": 24599, "epoch": 232, "lr": 5.5619259953555945e-05} +{"train_loss": 0.014757469296455383, "grad_norm": 0.18822930753231049, "global_step": 24600, "epoch": 232, "lr": 5.561631495333617e-05} +{"train_loss": 0.018338140100240707, "grad_norm": 0.12208042293787003, "global_step": 24601, "epoch": 232, "lr": 5.5613369933383105e-05} +{"train_loss": 0.01787080615758896, "grad_norm": 0.05725414678454399, "global_step": 24602, "epoch": 232, "lr": 5.5610424893707117e-05} +{"train_loss": 0.01557043008506298, "grad_norm": 0.10364457964897156, "global_step": 24603, "epoch": 232, "lr": 5.560747983431851e-05} +{"train_loss": 0.014928989112377167, "grad_norm": 0.06868423521518707, "global_step": 24604, "epoch": 232, "lr": 5.5604534755227666e-05} +{"train_loss": 0.016794320195913315, "grad_norm": 0.09706521779298782, "global_step": 24605, "epoch": 232, "lr": 5.560158965644493e-05} +{"train_loss": 0.01857694983482361, "grad_norm": 0.16076070070266724, "global_step": 24606, "epoch": 232, "lr": 5.559864453798064e-05} +{"train_loss": 0.016395706683397293, "grad_norm": 0.10079335421323776, "global_step": 24607, "epoch": 232, "lr": 5.5595699399845146e-05} +{"train_loss": 0.015305833891034126, "grad_norm": 0.06757069379091263, "global_step": 24608, "epoch": 232, "lr": 5.5592754242048794e-05} +{"train_loss": 0.015167078003287315, "grad_norm": 0.11046790331602097, "global_step": 24609, "epoch": 232, "lr": 5.558980906460195e-05} +{"train_loss": 0.01599617302417755, "grad_norm": 0.21241183578968048, "global_step": 24610, "epoch": 232, "lr": 5.558686386751493e-05} +{"train_loss": 0.0156460702419281, "grad_norm": 0.06337539851665497, "global_step": 24611, "epoch": 232, "lr": 5.5583918650798104e-05} +{"train_loss": 0.013053863309323788, "grad_norm": 0.06059330329298973, "global_step": 24612, "epoch": 232, "lr": 5.558097341446182e-05} +{"train_loss": 0.015337331220507622, "grad_norm": 0.09903053194284439, "global_step": 24613, "epoch": 232, "lr": 5.5578028158516435e-05} +{"train_loss": 0.01355949230492115, "grad_norm": 0.05912334844470024, "global_step": 24614, "epoch": 232, "lr": 5.557508288297227e-05} +{"train_loss": 0.01433173380792141, "grad_norm": 0.049671582877635956, "global_step": 24615, "epoch": 232, "lr": 5.5572137587839704e-05} +{"train_loss": 0.016685694456100464, "grad_norm": 0.08931700885295868, "global_step": 24616, "epoch": 232, "lr": 5.556919227312905e-05} +{"train_loss": 0.015139563009142876, "grad_norm": 0.11120434105396271, "global_step": 24617, "epoch": 232, "lr": 5.5566246938850684e-05} +{"train_loss": 0.015702906996011734, "grad_norm": 0.07012562453746796, "global_step": 24618, "epoch": 232, "lr": 5.5563301585014946e-05} +{"train_loss": 0.015478407964110374, "grad_norm": 0.09101297706365585, "global_step": 24619, "epoch": 232, "lr": 5.556035621163219e-05} +{"train_loss": 0.014815649017691612, "grad_norm": 0.11165626347064972, "global_step": 24620, "epoch": 232, "lr": 5.555741081871275e-05} +{"train_loss": 0.014279215596616268, "grad_norm": 0.10225644707679749, "global_step": 24621, "epoch": 232, "lr": 5.5554465406266996e-05} +{"train_loss": 0.015013317577540874, "grad_norm": 0.07031800597906113, "global_step": 24622, "epoch": 232, "lr": 5.5551519974305255e-05} +{"train_loss": 0.017247002571821213, "grad_norm": 0.07653434574604034, "global_step": 24623, "epoch": 232, "lr": 5.5548574522837905e-05} +{"train_loss": 0.014968916773796082, "grad_norm": 0.11156819015741348, "global_step": 24624, "epoch": 232, "lr": 5.554562905187527e-05} +{"train_loss": 0.013515196740627289, "grad_norm": 0.06261537969112396, "global_step": 24625, "epoch": 232, "lr": 5.55426835614277e-05} +{"train_loss": 0.017107848078012466, "grad_norm": 0.07527927309274673, "global_step": 24626, "epoch": 232, "lr": 5.553973805150555e-05} +{"train_loss": 0.015521122142672539, "grad_norm": 0.14653737843036652, "global_step": 24627, "epoch": 232, "lr": 5.553679252211918e-05} +{"train_loss": 0.014602217823266983, "grad_norm": 0.048051100224256516, "global_step": 24628, "epoch": 232, "lr": 5.553384697327891e-05} +{"train_loss": 0.01655052974820137, "grad_norm": 0.07417440414428711, "global_step": 24629, "epoch": 232, "lr": 5.5530901404995126e-05} +{"train_loss": 0.014260570518672466, "grad_norm": 0.10159892588853836, "global_step": 24630, "epoch": 232, "lr": 5.5527955817278155e-05} +{"train_loss": 0.015698105096817017, "grad_norm": 0.11329448223114014, "global_step": 24631, "epoch": 232, "lr": 5.552501021013835e-05} +{"train_loss": 0.01651402935385704, "grad_norm": 0.058496855199337006, "global_step": 24632, "epoch": 232, "lr": 5.552206458358605e-05} +{"train_loss": 0.015895845368504524, "grad_norm": 0.12464585155248642, "global_step": 24633, "epoch": 232, "lr": 5.551911893763164e-05} +{"train_loss": 0.013706454075872898, "grad_norm": 0.085686594247818, "global_step": 24634, "epoch": 232, "lr": 5.5516173272285444e-05} +{"train_loss": 0.01577070727944374, "grad_norm": 0.10743188858032227, "global_step": 24635, "epoch": 232, "lr": 5.551322758755779e-05} +{"train_loss": 0.017033621668815613, "grad_norm": 0.14048060774803162, "global_step": 24636, "epoch": 232, "lr": 5.551028188345908e-05} +{"train_loss": 0.012202916666865349, "grad_norm": 0.06943970918655396, "global_step": 24637, "epoch": 232, "lr": 5.550733615999961e-05} +{"train_loss": 0.017011117190122604, "grad_norm": 0.11863101273775101, "global_step": 24638, "epoch": 232, "lr": 5.550439041718978e-05} +{"train_loss": 0.015518906526267529, "grad_norm": 0.07488681375980377, "global_step": 24639, "epoch": 232, "lr": 5.55014446550399e-05} +{"train_loss": 0.016103386878967285, "grad_norm": 0.09409025311470032, "global_step": 24640, "epoch": 232, "lr": 5.5498498873560344e-05} +{"train_loss": 0.015583069063723087, "grad_norm": 0.06221427768468857, "global_step": 24641, "epoch": 232, "lr": 5.5495553072761454e-05} +{"train_loss": 0.014472532086074352, "grad_norm": 0.07323632389307022, "global_step": 24642, "epoch": 232, "lr": 5.5492607252653574e-05} +{"train_loss": 0.015184925869107246, "grad_norm": 0.08865859359502792, "global_step": 24643, "epoch": 232, "lr": 5.5489661413247065e-05} +{"train_loss": 0.015122287906706333, "grad_norm": 0.126302108168602, "global_step": 24644, "epoch": 232, "lr": 5.548671555455226e-05} +{"train_loss": 0.01910720393061638, "grad_norm": 0.139618918299675, "global_step": 24645, "epoch": 232, "lr": 5.548376967657955e-05} +{"train_loss": 0.014933068305253983, "grad_norm": 0.05823707953095436, "global_step": 24646, "epoch": 232, "lr": 5.548082377933924e-05} +{"train_loss": 0.015337013639509678, "grad_norm": 0.06975466012954712, "global_step": 24647, "epoch": 232, "lr": 5.547787786284169e-05} +{"train_loss": 0.016561487689614296, "grad_norm": 0.07991921901702881, "global_step": 24648, "epoch": 232, "lr": 5.5474931927097284e-05} +{"train_loss": 0.014078167267143726, "grad_norm": 0.11293799430131912, "global_step": 24649, "epoch": 232, "lr": 5.547198597211633e-05} +{"train_loss": 0.016764530912041664, "grad_norm": 0.07030737400054932, "global_step": 24650, "epoch": 232, "lr": 5.54690399979092e-05} +{"train_loss": 0.017151322215795517, "grad_norm": 0.05290296673774719, "global_step": 24651, "epoch": 232, "lr": 5.546609400448624e-05} +{"train_loss": 0.016209663823246956, "grad_norm": 0.1042841449379921, "global_step": 24652, "epoch": 232, "lr": 5.546314799185781e-05} +{"train_loss": 0.013393471017479897, "grad_norm": 0.06284341216087341, "global_step": 24653, "epoch": 232, "lr": 5.5460201960034255e-05} +{"train_loss": 0.017093561589717865, "grad_norm": 0.07096980512142181, "global_step": 24654, "epoch": 232, "lr": 5.545725590902592e-05} +{"train_loss": 0.01441102009266615, "grad_norm": 0.15807895362377167, "global_step": 24655, "epoch": 232, "lr": 5.545430983884317e-05} +{"train_loss": 0.016944583505392075, "grad_norm": 0.07247784733772278, "global_step": 24656, "epoch": 232, "lr": 5.545136374949633e-05} +{"train_loss": 0.015967324376106262, "grad_norm": 0.056732986122369766, "global_step": 24657, "epoch": 232, "lr": 5.5448417640995783e-05} +{"train_loss": 0.01601715013384819, "grad_norm": 0.0969778522849083, "global_step": 24658, "epoch": 232, "lr": 5.544547151335186e-05} +{"train_loss": 0.01629810407757759, "grad_norm": 0.07988203316926956, "global_step": 24659, "epoch": 232, "lr": 5.544252536657491e-05} +{"train_loss": 0.01673639565706253, "grad_norm": 0.12561820447444916, "global_step": 24660, "epoch": 232, "lr": 5.543957920067532e-05} +{"train_loss": 0.01597757823765278, "grad_norm": 0.11450036615133286, "global_step": 24661, "epoch": 232, "lr": 5.5436633015663375e-05} +{"train_loss": 0.015409031882882118, "grad_norm": 0.06113402917981148, "global_step": 24662, "epoch": 232, "lr": 5.54336868115495e-05} +{"train_loss": 0.014272020198404789, "grad_norm": 0.05441627278923988, "global_step": 24663, "epoch": 232, "lr": 5.543074058834399e-05} +{"train_loss": 0.015833178535103798, "grad_norm": 0.0676870048046112, "global_step": 24664, "epoch": 232, "lr": 5.542779434605723e-05} +{"train_loss": 0.01369963027536869, "grad_norm": 0.1129714697599411, "global_step": 24665, "epoch": 232, "lr": 5.542484808469958e-05} +{"train_loss": 0.016750432550907135, "grad_norm": 0.06133797764778137, "global_step": 24666, "epoch": 232, "lr": 5.542190180428134e-05} +{"train_loss": 0.014367073774337769, "grad_norm": 0.08347451686859131, "global_step": 24667, "epoch": 232, "lr": 5.541895550481291e-05} +{"train_loss": 0.01761004887521267, "grad_norm": 0.05733903869986534, "global_step": 24668, "epoch": 232, "lr": 5.5416009186304616e-05} +{"train_loss": 0.015240354463458061, "grad_norm": 0.060103658586740494, "global_step": 24669, "epoch": 232, "lr": 5.541306284876684e-05} +{"train_loss": 0.018192779272794724, "grad_norm": 0.0513550266623497, "global_step": 24670, "epoch": 232, "lr": 5.5410116492209894e-05} +{"train_loss": 0.015418866649270058, "grad_norm": 0.0784909650683403, "global_step": 24671, "epoch": 232, "lr": 5.540717011664417e-05} +{"train_loss": 0.014221969991922379, "grad_norm": 0.0560491643846035, "global_step": 24672, "epoch": 232, "lr": 5.540422372207999e-05} +{"train_loss": 0.015365585684776306, "grad_norm": 0.07427532970905304, "global_step": 24673, "epoch": 232, "lr": 5.5401277308527724e-05} +{"train_loss": 0.01607312634587288, "grad_norm": 0.06872973591089249, "global_step": 24674, "epoch": 232, "lr": 5.5398330875997716e-05} +{"train_loss": 0.018788572400808334, "grad_norm": 0.06420666724443436, "global_step": 24675, "epoch": 232, "lr": 5.539538442450033e-05} +{"train_loss": 0.015477275475859642, "grad_norm": 0.07438287883996964, "global_step": 24676, "epoch": 232, "lr": 5.5392437954045893e-05} +{"train_loss": 0.018040068447589874, "grad_norm": 0.09019174426794052, "global_step": 24677, "epoch": 232, "lr": 5.538949146464479e-05} +{"train_loss": 0.01640363782644272, "grad_norm": 0.0945986807346344, "global_step": 24678, "epoch": 232, "lr": 5.538654495630734e-05} +{"train_loss": 0.015380431897938251, "grad_norm": 0.09829791635274887, "global_step": 24679, "epoch": 232, "lr": 5.538359842904392e-05} +{"train_loss": 0.015551907010376453, "grad_norm": 0.08173496276140213, "global_step": 24680, "epoch": 232, "lr": 5.538065188286489e-05} +{"train_loss": 0.012592580169439316, "grad_norm": 0.09197079390287399, "global_step": 24681, "epoch": 232, "lr": 5.537770531778058e-05} +{"train_loss": 0.017953049391508102, "grad_norm": 0.07563810795545578, "global_step": 24682, "epoch": 232, "lr": 5.537475873380135e-05} +{"train_loss": 0.015163715928792953, "grad_norm": 0.066531702876091, "global_step": 24683, "epoch": 232, "lr": 5.537181213093756e-05} +{"train_loss": 0.015544744208455086, "grad_norm": 0.07474568486213684, "global_step": 24684, "epoch": 232, "lr": 5.536886550919955e-05} +{"train_loss": 0.016215980052947998, "grad_norm": 0.08994908630847931, "global_step": 24685, "epoch": 232, "lr": 5.536591886859769e-05} +{"train_loss": 0.015514122322201729, "grad_norm": 0.09833815693855286, "global_step": 24686, "epoch": 232, "lr": 5.536297220914233e-05} +{"train_loss": 0.019007325172424316, "grad_norm": 0.14067292213439941, "global_step": 24687, "epoch": 232, "lr": 5.536002553084381e-05} +{"train_loss": 0.01733068935573101, "grad_norm": 0.06831197440624237, "global_step": 24688, "epoch": 232, "lr": 5.535707883371247e-05} +{"train_loss": 0.013027532026171684, "grad_norm": 0.0574667789041996, "global_step": 24689, "epoch": 232, "lr": 5.535413211775873e-05} +{"train_loss": 0.014097094535827637, "grad_norm": 0.10604648292064667, "global_step": 24690, "epoch": 232, "lr": 5.535118538299285e-05} +{"train_loss": 0.014825398102402687, "grad_norm": 0.07336753606796265, "global_step": 24691, "epoch": 232, "lr": 5.534823862942527e-05} +{"train_loss": 0.016380300745368004, "grad_norm": 0.11752161383628845, "global_step": 24692, "epoch": 232, "lr": 5.5345291857066285e-05} +{"train_loss": 0.015886101871728897, "grad_norm": 0.06541737914085388, "global_step": 24693, "epoch": 232, "lr": 5.534234506592627e-05} +{"train_loss": 0.013940408825874329, "grad_norm": 0.08537112176418304, "global_step": 24694, "epoch": 232, "lr": 5.533939825601558e-05} +{"train_loss": 0.016735803335905075, "grad_norm": 0.11950427293777466, "global_step": 24695, "epoch": 232, "lr": 5.533645142734458e-05} +{"train_loss": 0.017212269827723503, "grad_norm": 0.06671090424060822, "global_step": 24696, "epoch": 232, "lr": 5.53335045799236e-05} +{"train_loss": 0.015715633088476816, "grad_norm": 0.08274126052856445, "global_step": 24697, "epoch": 232, "lr": 5.5330557713763e-05, "val_loss": 0.0199736375361681} +{"train_loss": 0.015919379889965057, "grad_norm": 0.1371936798095703, "global_step": 24698, "epoch": 233, "lr": 5.5327610828873146e-05} +{"train_loss": 0.016714900732040405, "grad_norm": 0.06577103585004807, "global_step": 24699, "epoch": 233, "lr": 5.532466392526438e-05} +{"train_loss": 0.01595805212855339, "grad_norm": 0.06480088084936142, "global_step": 24700, "epoch": 233, "lr": 5.532171700294705e-05} +{"train_loss": 0.01606700010597706, "grad_norm": 0.13870413601398468, "global_step": 24701, "epoch": 233, "lr": 5.531877006193155e-05} +{"train_loss": 0.01639527827501297, "grad_norm": 0.08169861882925034, "global_step": 24702, "epoch": 233, "lr": 5.531582310222817e-05} +{"train_loss": 0.014684312976896763, "grad_norm": 0.12146032601594925, "global_step": 24703, "epoch": 233, "lr": 5.531287612384732e-05} +{"train_loss": 0.014220770448446274, "grad_norm": 0.05997654050588608, "global_step": 24704, "epoch": 233, "lr": 5.5309929126799345e-05} +{"train_loss": 0.014226334169507027, "grad_norm": 0.11583703756332397, "global_step": 24705, "epoch": 233, "lr": 5.5306982111094566e-05} +{"train_loss": 0.017604675143957138, "grad_norm": 0.05124472826719284, "global_step": 24706, "epoch": 233, "lr": 5.5304035076743374e-05} +{"train_loss": 0.016418581828475, "grad_norm": 0.08371815830469131, "global_step": 24707, "epoch": 233, "lr": 5.5301088023756095e-05} +{"train_loss": 0.015800833702087402, "grad_norm": 0.06622610986232758, "global_step": 24708, "epoch": 233, "lr": 5.529814095214311e-05} +{"train_loss": 0.014995493926107883, "grad_norm": 0.05972912535071373, "global_step": 24709, "epoch": 233, "lr": 5.529519386191475e-05} +{"train_loss": 0.019555676728487015, "grad_norm": 0.08198987692594528, "global_step": 24710, "epoch": 233, "lr": 5.529224675308139e-05} +{"train_loss": 0.017838772386312485, "grad_norm": 0.07686346769332886, "global_step": 24711, "epoch": 233, "lr": 5.528929962565338e-05} +{"train_loss": 0.014267010614275932, "grad_norm": 0.0790591686964035, "global_step": 24712, "epoch": 233, "lr": 5.528635247964107e-05} +{"train_loss": 0.017727382481098175, "grad_norm": 0.06696897000074387, "global_step": 24713, "epoch": 233, "lr": 5.528340531505481e-05} +{"train_loss": 0.01408221572637558, "grad_norm": 0.07778510451316833, "global_step": 24714, "epoch": 233, "lr": 5.528045813190496e-05} +{"train_loss": 0.013297113589942455, "grad_norm": 0.10994283854961395, "global_step": 24715, "epoch": 233, "lr": 5.5277510930201894e-05} +{"train_loss": 0.01701623573899269, "grad_norm": 0.1392745077610016, "global_step": 24716, "epoch": 233, "lr": 5.527456370995594e-05} +{"train_loss": 0.014638207852840424, "grad_norm": 0.11321587860584259, "global_step": 24717, "epoch": 233, "lr": 5.5271616471177444e-05} +{"train_loss": 0.016413137316703796, "grad_norm": 0.1763170063495636, "global_step": 24718, "epoch": 233, "lr": 5.526866921387681e-05} +{"train_loss": 0.015095772221684456, "grad_norm": 0.06159273907542229, "global_step": 24719, "epoch": 233, "lr": 5.526572193806434e-05} +{"train_loss": 0.011507811024785042, "grad_norm": 0.06576775759458542, "global_step": 24720, "epoch": 233, "lr": 5.5262774643750425e-05} +{"train_loss": 0.017106637358665466, "grad_norm": 0.07135096192359924, "global_step": 24721, "epoch": 233, "lr": 5.5259827330945405e-05} +{"train_loss": 0.013788855634629726, "grad_norm": 0.10262567549943924, "global_step": 24722, "epoch": 233, "lr": 5.5256879999659636e-05} +{"train_loss": 0.017594680190086365, "grad_norm": 0.06950998306274414, "global_step": 24723, "epoch": 233, "lr": 5.5253932649903495e-05} +{"train_loss": 0.013047785498201847, "grad_norm": 0.049031030386686325, "global_step": 24724, "epoch": 233, "lr": 5.525098528168731e-05} +{"train_loss": 0.015891067683696747, "grad_norm": 0.073647141456604, "global_step": 24725, "epoch": 233, "lr": 5.5248037895021434e-05} +{"train_loss": 0.015528107061982155, "grad_norm": 0.08510079234838486, "global_step": 24726, "epoch": 233, "lr": 5.524509048991625e-05} +{"train_loss": 0.01637854054570198, "grad_norm": 0.07805082201957703, "global_step": 24727, "epoch": 233, "lr": 5.524214306638209e-05} +{"train_loss": 0.0156282726675272, "grad_norm": 0.08353309333324432, "global_step": 24728, "epoch": 233, "lr": 5.523919562442932e-05} +{"train_loss": 0.018452569842338562, "grad_norm": 0.173679381608963, "global_step": 24729, "epoch": 233, "lr": 5.523624816406829e-05} +{"train_loss": 0.017112836241722107, "grad_norm": 0.13849472999572754, "global_step": 24730, "epoch": 233, "lr": 5.523330068530937e-05} +{"train_loss": 0.01765180192887783, "grad_norm": 0.09345485270023346, "global_step": 24731, "epoch": 233, "lr": 5.523035318816291e-05} +{"train_loss": 0.01510243397206068, "grad_norm": 0.11091292649507523, "global_step": 24732, "epoch": 233, "lr": 5.522740567263925e-05} +{"train_loss": 0.014739811420440674, "grad_norm": 0.0665905773639679, "global_step": 24733, "epoch": 233, "lr": 5.522445813874878e-05} +{"train_loss": 0.015170767903327942, "grad_norm": 0.08561711013317108, "global_step": 24734, "epoch": 233, "lr": 5.522151058650182e-05} +{"train_loss": 0.01426190696656704, "grad_norm": 0.07505837827920914, "global_step": 24735, "epoch": 233, "lr": 5.5218563015908755e-05} +{"train_loss": 0.015995565801858902, "grad_norm": 0.1051427349448204, "global_step": 24736, "epoch": 233, "lr": 5.521561542697992e-05} +{"train_loss": 0.015493620187044144, "grad_norm": 0.06277536600828171, "global_step": 24737, "epoch": 233, "lr": 5.5212667819725686e-05} +{"train_loss": 0.014816224575042725, "grad_norm": 0.07569114118814468, "global_step": 24738, "epoch": 233, "lr": 5.5209720194156386e-05} +{"train_loss": 0.01502145640552044, "grad_norm": 0.07009759545326233, "global_step": 24739, "epoch": 233, "lr": 5.520677255028241e-05} +{"train_loss": 0.014185931533575058, "grad_norm": 0.05758323147892952, "global_step": 24740, "epoch": 233, "lr": 5.520382488811411e-05} +{"train_loss": 0.016553323715925217, "grad_norm": 0.08735726028680801, "global_step": 24741, "epoch": 233, "lr": 5.520087720766181e-05} +{"train_loss": 0.013579608872532845, "grad_norm": 0.1337190717458725, "global_step": 24742, "epoch": 233, "lr": 5.5197929508935906e-05} +{"train_loss": 0.017274044454097748, "grad_norm": 0.05937682092189789, "global_step": 24743, "epoch": 233, "lr": 5.519498179194673e-05} +{"train_loss": 0.015392833389341831, "grad_norm": 0.12095291167497635, "global_step": 24744, "epoch": 233, "lr": 5.5192034056704643e-05} +{"train_loss": 0.01621934585273266, "grad_norm": 0.093251071870327, "global_step": 24745, "epoch": 233, "lr": 5.5189086303220026e-05} +{"train_loss": 0.016176018863916397, "grad_norm": 0.05281109735369682, "global_step": 24746, "epoch": 233, "lr": 5.5186138531503194e-05} +{"train_loss": 0.014686567708849907, "grad_norm": 0.13588106632232666, "global_step": 24747, "epoch": 233, "lr": 5.518319074156454e-05} +{"train_loss": 0.01512753963470459, "grad_norm": 0.06432969123125076, "global_step": 24748, "epoch": 233, "lr": 5.51802429334144e-05} +{"train_loss": 0.015680644661188126, "grad_norm": 0.05438949912786484, "global_step": 24749, "epoch": 233, "lr": 5.517729510706315e-05} +{"train_loss": 0.01564149372279644, "grad_norm": 0.07145922631025314, "global_step": 24750, "epoch": 233, "lr": 5.517434726252113e-05} +{"train_loss": 0.014940080232918262, "grad_norm": 0.07387284189462662, "global_step": 24751, "epoch": 233, "lr": 5.5171399399798706e-05} +{"train_loss": 0.016320014372467995, "grad_norm": 0.12310859560966492, "global_step": 24752, "epoch": 233, "lr": 5.5168451518906225e-05} +{"train_loss": 0.015314595773816109, "grad_norm": 0.058013346046209335, "global_step": 24753, "epoch": 233, "lr": 5.5165503619854056e-05} +{"train_loss": 0.0135404197499156, "grad_norm": 0.06316167116165161, "global_step": 24754, "epoch": 233, "lr": 5.516255570265257e-05} +{"train_loss": 0.016370385885238647, "grad_norm": 0.09765703976154327, "global_step": 24755, "epoch": 233, "lr": 5.5159607767312096e-05} +{"train_loss": 0.015956945717334747, "grad_norm": 0.061828576028347015, "global_step": 24756, "epoch": 233, "lr": 5.5156659813843006e-05} +{"train_loss": 0.016386525705456734, "grad_norm": 0.06997206807136536, "global_step": 24757, "epoch": 233, "lr": 5.515371184225565e-05} +{"train_loss": 0.01806749776005745, "grad_norm": 0.07622089982032776, "global_step": 24758, "epoch": 233, "lr": 5.5150763852560385e-05} +{"train_loss": 0.018826141953468323, "grad_norm": 0.06511920690536499, "global_step": 24759, "epoch": 233, "lr": 5.5147815844767604e-05} +{"train_loss": 0.015439681708812714, "grad_norm": 0.06749504804611206, "global_step": 24760, "epoch": 233, "lr": 5.5144867818887605e-05} +{"train_loss": 0.015585213899612427, "grad_norm": 0.06739593297243118, "global_step": 24761, "epoch": 233, "lr": 5.514191977493079e-05} +{"train_loss": 0.016141802072525024, "grad_norm": 0.1338914930820465, "global_step": 24762, "epoch": 233, "lr": 5.5138971712907525e-05} +{"train_loss": 0.014920034445822239, "grad_norm": 0.10821373760700226, "global_step": 24763, "epoch": 233, "lr": 5.513602363282812e-05} +{"train_loss": 0.013601992279291153, "grad_norm": 0.14896152913570404, "global_step": 24764, "epoch": 233, "lr": 5.513307553470297e-05} +{"train_loss": 0.018811505287885666, "grad_norm": 0.10778233408927917, "global_step": 24765, "epoch": 233, "lr": 5.5130127418542435e-05} +{"train_loss": 0.014778118580579758, "grad_norm": 0.07188506424427032, "global_step": 24766, "epoch": 233, "lr": 5.512717928435685e-05} +{"train_loss": 0.015363836660981178, "grad_norm": 0.05698423460125923, "global_step": 24767, "epoch": 233, "lr": 5.512423113215659e-05} +{"train_loss": 0.014676503837108612, "grad_norm": 0.11242977529764175, "global_step": 24768, "epoch": 233, "lr": 5.5121282961952016e-05} +{"train_loss": 0.01349790021777153, "grad_norm": 0.0772961899638176, "global_step": 24769, "epoch": 233, "lr": 5.5118334773753474e-05} +{"train_loss": 0.014815621078014374, "grad_norm": 0.1013699471950531, "global_step": 24770, "epoch": 233, "lr": 5.5115386567571326e-05} +{"train_loss": 0.017329808324575424, "grad_norm": 0.08671937137842178, "global_step": 24771, "epoch": 233, "lr": 5.511243834341595e-05} +{"train_loss": 0.016853876411914825, "grad_norm": 0.18133795261383057, "global_step": 24772, "epoch": 233, "lr": 5.510949010129767e-05} +{"train_loss": 0.014877310954034328, "grad_norm": 0.11124799400568008, "global_step": 24773, "epoch": 233, "lr": 5.510654184122688e-05} +{"train_loss": 0.015495656058192253, "grad_norm": 0.1271393597126007, "global_step": 24774, "epoch": 233, "lr": 5.510359356321393e-05} +{"train_loss": 0.016564171761274338, "grad_norm": 0.0648597702383995, "global_step": 24775, "epoch": 233, "lr": 5.5100645267269146e-05} +{"train_loss": 0.01548765879124403, "grad_norm": 0.08138012886047363, "global_step": 24776, "epoch": 233, "lr": 5.509769695340292e-05} +{"train_loss": 0.012652818113565445, "grad_norm": 0.09880959242582321, "global_step": 24777, "epoch": 233, "lr": 5.509474862162561e-05} +{"train_loss": 0.01742955669760704, "grad_norm": 0.05550302192568779, "global_step": 24778, "epoch": 233, "lr": 5.509180027194757e-05} +{"train_loss": 0.01686388999223709, "grad_norm": 0.12323526293039322, "global_step": 24779, "epoch": 233, "lr": 5.508885190437916e-05} +{"train_loss": 0.01660700887441635, "grad_norm": 0.08251014351844788, "global_step": 24780, "epoch": 233, "lr": 5.5085903518930726e-05} +{"train_loss": 0.015507902018725872, "grad_norm": 0.11733755469322205, "global_step": 24781, "epoch": 233, "lr": 5.5082955115612656e-05} +{"train_loss": 0.015738099813461304, "grad_norm": 0.08130268007516861, "global_step": 24782, "epoch": 233, "lr": 5.508000669443527e-05} +{"train_loss": 0.015441768802702427, "grad_norm": 0.06851830333471298, "global_step": 24783, "epoch": 233, "lr": 5.5077058255408975e-05} +{"train_loss": 0.014437097124755383, "grad_norm": 0.059124626219272614, "global_step": 24784, "epoch": 233, "lr": 5.507410979854409e-05} +{"train_loss": 0.014826719649136066, "grad_norm": 0.14263461530208588, "global_step": 24785, "epoch": 233, "lr": 5.507116132385098e-05} +{"train_loss": 0.01542935986071825, "grad_norm": 0.07131768018007278, "global_step": 24786, "epoch": 233, "lr": 5.506821283134004e-05} +{"train_loss": 0.0173686183989048, "grad_norm": 0.08335709571838379, "global_step": 24787, "epoch": 233, "lr": 5.506526432102158e-05} +{"train_loss": 0.014020942151546478, "grad_norm": 0.09511632472276688, "global_step": 24788, "epoch": 233, "lr": 5.5062315792905994e-05} +{"train_loss": 0.018273476511240005, "grad_norm": 0.06757592409849167, "global_step": 24789, "epoch": 233, "lr": 5.505936724700364e-05} +{"train_loss": 0.016247816383838654, "grad_norm": 0.10703771561384201, "global_step": 24790, "epoch": 233, "lr": 5.505641868332486e-05} +{"train_loss": 0.016969621181488037, "grad_norm": 0.07845433801412582, "global_step": 24791, "epoch": 233, "lr": 5.5053470101880034e-05} +{"train_loss": 0.014717672020196915, "grad_norm": 0.06551291793584824, "global_step": 24792, "epoch": 233, "lr": 5.50505215026795e-05} +{"train_loss": 0.014226228930056095, "grad_norm": 0.08542169630527496, "global_step": 24793, "epoch": 233, "lr": 5.504757288573363e-05} +{"train_loss": 0.017168106511235237, "grad_norm": 0.09192050248384476, "global_step": 24794, "epoch": 233, "lr": 5.504462425105278e-05} +{"train_loss": 0.014723189175128937, "grad_norm": 0.09105847775936127, "global_step": 24795, "epoch": 233, "lr": 5.5041675598647336e-05} +{"train_loss": 0.016957353800535202, "grad_norm": 0.09094943851232529, "global_step": 24796, "epoch": 233, "lr": 5.503872692852763e-05} +{"train_loss": 0.016638074070215225, "grad_norm": 0.08581811934709549, "global_step": 24797, "epoch": 233, "lr": 5.503577824070401e-05} +{"train_loss": 0.016365546733140945, "grad_norm": 0.0955720767378807, "global_step": 24798, "epoch": 233, "lr": 5.503282953518687e-05} +{"train_loss": 0.0153316305950284, "grad_norm": 0.07286076247692108, "global_step": 24799, "epoch": 233, "lr": 5.5029880811986544e-05} +{"train_loss": 0.013854059390723705, "grad_norm": 0.13578200340270996, "global_step": 24800, "epoch": 233, "lr": 5.502693207111341e-05} +{"train_loss": 0.01654450222849846, "grad_norm": 0.06270000338554382, "global_step": 24801, "epoch": 233, "lr": 5.502398331257783e-05} +{"train_loss": 0.015197411179542542, "grad_norm": 0.14389249682426453, "global_step": 24802, "epoch": 233, "lr": 5.502103453639016e-05} +{"train_loss": 0.015695633208554872, "grad_norm": 0.08178450912237167, "global_step": 24803, "epoch": 233, "lr": 5.5018085742560744e-05, "val_loss": 0.019586710259318352} +{"train_loss": 0.020685728639364243, "grad_norm": 0.10484112054109573, "global_step": 24804, "epoch": 234, "lr": 5.5015136931099964e-05} +{"train_loss": 0.015396768227219582, "grad_norm": 0.0699346587061882, "global_step": 24805, "epoch": 234, "lr": 5.501218810201818e-05} +{"train_loss": 0.019617535173892975, "grad_norm": 0.08863263577222824, "global_step": 24806, "epoch": 234, "lr": 5.500923925532574e-05} +{"train_loss": 0.01624210737645626, "grad_norm": 0.1376296430826187, "global_step": 24807, "epoch": 234, "lr": 5.500629039103301e-05} +{"train_loss": 0.01658562943339348, "grad_norm": 0.05529920756816864, "global_step": 24808, "epoch": 234, "lr": 5.500334150915036e-05} +{"train_loss": 0.014407520182430744, "grad_norm": 0.0692824050784111, "global_step": 24809, "epoch": 234, "lr": 5.500039260968814e-05} +{"train_loss": 0.013470010831952095, "grad_norm": 0.05998528003692627, "global_step": 24810, "epoch": 234, "lr": 5.4997443692656714e-05} +{"train_loss": 0.017446991056203842, "grad_norm": 0.06549690663814545, "global_step": 24811, "epoch": 234, "lr": 5.499449475806644e-05} +{"train_loss": 0.017059728503227234, "grad_norm": 0.07386383414268494, "global_step": 24812, "epoch": 234, "lr": 5.4991545805927706e-05} +{"train_loss": 0.016661379486322403, "grad_norm": 0.10747276991605759, "global_step": 24813, "epoch": 234, "lr": 5.498859683625083e-05} +{"train_loss": 0.015383971855044365, "grad_norm": 0.05486901476979256, "global_step": 24814, "epoch": 234, "lr": 5.498564784904618e-05} +{"train_loss": 0.014995074830949306, "grad_norm": 0.09170038998126984, "global_step": 24815, "epoch": 234, "lr": 5.498269884432416e-05} +{"train_loss": 0.014854311011731625, "grad_norm": 0.071549192070961, "global_step": 24816, "epoch": 234, "lr": 5.497974982209508e-05} +{"train_loss": 0.015035519376397133, "grad_norm": 0.12346377968788147, "global_step": 24817, "epoch": 234, "lr": 5.497680078236934e-05} +{"train_loss": 0.014783995226025581, "grad_norm": 0.06983442604541779, "global_step": 24818, "epoch": 234, "lr": 5.497385172515729e-05} +{"train_loss": 0.016658343374729156, "grad_norm": 0.06025661900639534, "global_step": 24819, "epoch": 234, "lr": 5.497090265046928e-05} +{"train_loss": 0.015481268055737019, "grad_norm": 0.09488026797771454, "global_step": 24820, "epoch": 234, "lr": 5.496795355831568e-05} +{"train_loss": 0.015187421813607216, "grad_norm": 0.056668851524591446, "global_step": 24821, "epoch": 234, "lr": 5.4965004448706845e-05} +{"train_loss": 0.017698945477604866, "grad_norm": 0.08538076281547546, "global_step": 24822, "epoch": 234, "lr": 5.496205532165316e-05} +{"train_loss": 0.015597783029079437, "grad_norm": 0.0477023608982563, "global_step": 24823, "epoch": 234, "lr": 5.4959106177164954e-05} +{"train_loss": 0.016520384699106216, "grad_norm": 0.1694912314414978, "global_step": 24824, "epoch": 234, "lr": 5.495615701525262e-05} +{"train_loss": 0.01387297548353672, "grad_norm": 0.10166227072477341, "global_step": 24825, "epoch": 234, "lr": 5.49532078359265e-05} +{"train_loss": 0.017418302595615387, "grad_norm": 0.08399415016174316, "global_step": 24826, "epoch": 234, "lr": 5.495025863919695e-05} +{"train_loss": 0.01432070042937994, "grad_norm": 0.06379733979701996, "global_step": 24827, "epoch": 234, "lr": 5.494730942507437e-05} +{"train_loss": 0.01584041118621826, "grad_norm": 0.07651031762361526, "global_step": 24828, "epoch": 234, "lr": 5.4944360193569064e-05} +{"train_loss": 0.013084395788609982, "grad_norm": 0.05866040289402008, "global_step": 24829, "epoch": 234, "lr": 5.494141094469145e-05} +{"train_loss": 0.017722398042678833, "grad_norm": 0.1351979523897171, "global_step": 24830, "epoch": 234, "lr": 5.493846167845186e-05} +{"train_loss": 0.01742679812014103, "grad_norm": 0.13311868906021118, "global_step": 24831, "epoch": 234, "lr": 5.493551239486067e-05} +{"train_loss": 0.015305285342037678, "grad_norm": 0.07152532786130905, "global_step": 24832, "epoch": 234, "lr": 5.4932563093928224e-05} +{"train_loss": 0.015627244487404823, "grad_norm": 0.06891431659460068, "global_step": 24833, "epoch": 234, "lr": 5.49296137756649e-05} +{"train_loss": 0.013225030153989792, "grad_norm": 0.08984903991222382, "global_step": 24834, "epoch": 234, "lr": 5.492666444008107e-05} +{"train_loss": 0.015028286725282669, "grad_norm": 0.06331285089254379, "global_step": 24835, "epoch": 234, "lr": 5.4923715087187065e-05} +{"train_loss": 0.013614453375339508, "grad_norm": 0.0903095006942749, "global_step": 24836, "epoch": 234, "lr": 5.492076571699327e-05} +{"train_loss": 0.014431624673306942, "grad_norm": 0.06629641354084015, "global_step": 24837, "epoch": 234, "lr": 5.4917816329510055e-05} +{"train_loss": 0.015679441392421722, "grad_norm": 0.1263950616121292, "global_step": 24838, "epoch": 234, "lr": 5.491486692474775e-05} +{"train_loss": 0.014835457317531109, "grad_norm": 0.06153594329953194, "global_step": 24839, "epoch": 234, "lr": 5.491191750271677e-05} +{"train_loss": 0.015302751213312149, "grad_norm": 0.07861758023500443, "global_step": 24840, "epoch": 234, "lr": 5.4908968063427424e-05} +{"train_loss": 0.016489332541823387, "grad_norm": 0.20088328421115875, "global_step": 24841, "epoch": 234, "lr": 5.490601860689011e-05} +{"train_loss": 0.014259601011872292, "grad_norm": 0.08055611699819565, "global_step": 24842, "epoch": 234, "lr": 5.490306913311519e-05} +{"train_loss": 0.012644868344068527, "grad_norm": 0.059881385415792465, "global_step": 24843, "epoch": 234, "lr": 5.4900119642113e-05} +{"train_loss": 0.014847107231616974, "grad_norm": 0.08129757642745972, "global_step": 24844, "epoch": 234, "lr": 5.4897170133893925e-05} +{"train_loss": 0.016199614852666855, "grad_norm": 0.057862911373376846, "global_step": 24845, "epoch": 234, "lr": 5.489422060846833e-05} +{"train_loss": 0.01548106037080288, "grad_norm": 0.13743138313293457, "global_step": 24846, "epoch": 234, "lr": 5.4891271065846574e-05} +{"train_loss": 0.01653873920440674, "grad_norm": 0.08214771747589111, "global_step": 24847, "epoch": 234, "lr": 5.488832150603901e-05} +{"train_loss": 0.013585546053946018, "grad_norm": 0.06746629625558853, "global_step": 24848, "epoch": 234, "lr": 5.488537192905601e-05} +{"train_loss": 0.01326661091297865, "grad_norm": 0.10748142749071121, "global_step": 24849, "epoch": 234, "lr": 5.4882422334907945e-05} +{"train_loss": 0.01603354886174202, "grad_norm": 0.12277315557003021, "global_step": 24850, "epoch": 234, "lr": 5.4879472723605155e-05} +{"train_loss": 0.01653505116701126, "grad_norm": 0.10755155235528946, "global_step": 24851, "epoch": 234, "lr": 5.487652309515804e-05} +{"train_loss": 0.017226941883563995, "grad_norm": 0.15174242854118347, "global_step": 24852, "epoch": 234, "lr": 5.487357344957692e-05} +{"train_loss": 0.016342345625162125, "grad_norm": 0.09727489203214645, "global_step": 24853, "epoch": 234, "lr": 5.48706237868722e-05} +{"train_loss": 0.017004869878292084, "grad_norm": 0.14227375388145447, "global_step": 24854, "epoch": 234, "lr": 5.4867674107054236e-05} +{"train_loss": 0.015839070081710815, "grad_norm": 0.14748302102088928, "global_step": 24855, "epoch": 234, "lr": 5.486472441013336e-05} +{"train_loss": 0.015147756785154343, "grad_norm": 0.16741004586219788, "global_step": 24856, "epoch": 234, "lr": 5.486177469611998e-05} +{"train_loss": 0.01703568920493126, "grad_norm": 0.07319459319114685, "global_step": 24857, "epoch": 234, "lr": 5.4858824965024416e-05} +{"train_loss": 0.014427083544433117, "grad_norm": 0.12006242573261261, "global_step": 24858, "epoch": 234, "lr": 5.4855875216857066e-05} +{"train_loss": 0.014845965430140495, "grad_norm": 0.15152499079704285, "global_step": 24859, "epoch": 234, "lr": 5.485292545162828e-05} +{"train_loss": 0.014421576634049416, "grad_norm": 0.062183305621147156, "global_step": 24860, "epoch": 234, "lr": 5.4849975669348417e-05} +{"train_loss": 0.015354767441749573, "grad_norm": 0.1159818023443222, "global_step": 24861, "epoch": 234, "lr": 5.4847025870027866e-05} +{"train_loss": 0.015840783715248108, "grad_norm": 0.13895680010318756, "global_step": 24862, "epoch": 234, "lr": 5.4844076053676966e-05} +{"train_loss": 0.016350157558918, "grad_norm": 0.09156337380409241, "global_step": 24863, "epoch": 234, "lr": 5.4841126220306084e-05} +{"train_loss": 0.015196545049548149, "grad_norm": 0.11040928959846497, "global_step": 24864, "epoch": 234, "lr": 5.48381763699256e-05} +{"train_loss": 0.015731070190668106, "grad_norm": 0.10260818153619766, "global_step": 24865, "epoch": 234, "lr": 5.483522650254586e-05} +{"train_loss": 0.016288578510284424, "grad_norm": 0.053154923021793365, "global_step": 24866, "epoch": 234, "lr": 5.4832276618177236e-05} +{"train_loss": 0.0158630833029747, "grad_norm": 0.0686689019203186, "global_step": 24867, "epoch": 234, "lr": 5.482932671683009e-05} +{"train_loss": 0.017899300903081894, "grad_norm": 0.07747925817966461, "global_step": 24868, "epoch": 234, "lr": 5.48263767985148e-05} +{"train_loss": 0.016671352088451385, "grad_norm": 0.1207164004445076, "global_step": 24869, "epoch": 234, "lr": 5.4823426863241714e-05} +{"train_loss": 0.015138721093535423, "grad_norm": 0.07931690663099289, "global_step": 24870, "epoch": 234, "lr": 5.48204769110212e-05} +{"train_loss": 0.014911990612745285, "grad_norm": 0.11370761692523956, "global_step": 24871, "epoch": 234, "lr": 5.481752694186364e-05} +{"train_loss": 0.015617018565535545, "grad_norm": 0.07620910555124283, "global_step": 24872, "epoch": 234, "lr": 5.481457695577937e-05} +{"train_loss": 0.015699027106165886, "grad_norm": 0.06614870578050613, "global_step": 24873, "epoch": 234, "lr": 5.481162695277878e-05} +{"train_loss": 0.017711548134684563, "grad_norm": 0.11032868176698685, "global_step": 24874, "epoch": 234, "lr": 5.480867693287224e-05} +{"train_loss": 0.01662634313106537, "grad_norm": 0.09052717685699463, "global_step": 24875, "epoch": 234, "lr": 5.480572689607008e-05} +{"train_loss": 0.020445503294467926, "grad_norm": 0.12228794395923615, "global_step": 24876, "epoch": 234, "lr": 5.4802776842382686e-05} +{"train_loss": 0.012296855449676514, "grad_norm": 0.0638510137796402, "global_step": 24877, "epoch": 234, "lr": 5.479982677182043e-05} +{"train_loss": 0.01842973381280899, "grad_norm": 0.07850512117147446, "global_step": 24878, "epoch": 234, "lr": 5.4796876684393675e-05} +{"train_loss": 0.016684554517269135, "grad_norm": 0.12681546807289124, "global_step": 24879, "epoch": 234, "lr": 5.4793926580112755e-05} +{"train_loss": 0.014038272202014923, "grad_norm": 0.06676293164491653, "global_step": 24880, "epoch": 234, "lr": 5.47909764589881e-05} +{"train_loss": 0.020852725952863693, "grad_norm": 0.0797717422246933, "global_step": 24881, "epoch": 234, "lr": 5.478802632103001e-05} +{"train_loss": 0.016221147030591965, "grad_norm": 0.10049350559711456, "global_step": 24882, "epoch": 234, "lr": 5.4785076166248884e-05} +{"train_loss": 0.014430264942348003, "grad_norm": 0.08220536261796951, "global_step": 24883, "epoch": 234, "lr": 5.478212599465509e-05} +{"train_loss": 0.013898167759180069, "grad_norm": 0.12544383108615875, "global_step": 24884, "epoch": 234, "lr": 5.477917580625897e-05} +{"train_loss": 0.015371821820735931, "grad_norm": 0.1154586598277092, "global_step": 24885, "epoch": 234, "lr": 5.4776225601070904e-05} +{"train_loss": 0.015523258596658707, "grad_norm": 0.13818243145942688, "global_step": 24886, "epoch": 234, "lr": 5.4773275379101276e-05} +{"train_loss": 0.016631128266453743, "grad_norm": 0.0716092437505722, "global_step": 24887, "epoch": 234, "lr": 5.477032514036042e-05} +{"train_loss": 0.013118775561451912, "grad_norm": 0.07166992872953415, "global_step": 24888, "epoch": 234, "lr": 5.4767374884858724e-05} +{"train_loss": 0.0194920115172863, "grad_norm": 0.0793256089091301, "global_step": 24889, "epoch": 234, "lr": 5.476442461260655e-05} +{"train_loss": 0.013758320361375809, "grad_norm": 0.07393505424261093, "global_step": 24890, "epoch": 234, "lr": 5.4761474323614246e-05} +{"train_loss": 0.016021035611629486, "grad_norm": 0.08675696700811386, "global_step": 24891, "epoch": 234, "lr": 5.47585240178922e-05} +{"train_loss": 0.015736788511276245, "grad_norm": 0.06961364299058914, "global_step": 24892, "epoch": 234, "lr": 5.475557369545078e-05} +{"train_loss": 0.013194084167480469, "grad_norm": 0.08048780262470245, "global_step": 24893, "epoch": 234, "lr": 5.475262335630033e-05} +{"train_loss": 0.016909202560782433, "grad_norm": 0.07383467257022858, "global_step": 24894, "epoch": 234, "lr": 5.474967300045122e-05} +{"train_loss": 0.016552014276385307, "grad_norm": 0.06282174587249756, "global_step": 24895, "epoch": 234, "lr": 5.474672262791384e-05} +{"train_loss": 0.016036244109272957, "grad_norm": 0.09694795310497284, "global_step": 24896, "epoch": 234, "lr": 5.474377223869853e-05} +{"train_loss": 0.014972757548093796, "grad_norm": 0.09765642881393433, "global_step": 24897, "epoch": 234, "lr": 5.474082183281567e-05} +{"train_loss": 0.012311476282775402, "grad_norm": 0.06025359407067299, "global_step": 24898, "epoch": 234, "lr": 5.473787141027563e-05} +{"train_loss": 0.015767723321914673, "grad_norm": 0.06317342817783356, "global_step": 24899, "epoch": 234, "lr": 5.473492097108877e-05} +{"train_loss": 0.0158996544778347, "grad_norm": 0.08051037788391113, "global_step": 24900, "epoch": 234, "lr": 5.473197051526545e-05} +{"train_loss": 0.015648841857910156, "grad_norm": 0.0815102830529213, "global_step": 24901, "epoch": 234, "lr": 5.472902004281605e-05} +{"train_loss": 0.01444682851433754, "grad_norm": 0.1313340812921524, "global_step": 24902, "epoch": 234, "lr": 5.4726069553750926e-05} +{"train_loss": 0.01598130725324154, "grad_norm": 0.08903270959854126, "global_step": 24903, "epoch": 234, "lr": 5.472311904808044e-05} +{"train_loss": 0.015329679474234581, "grad_norm": 0.059035055339336395, "global_step": 24904, "epoch": 234, "lr": 5.4720168525814985e-05} +{"train_loss": 0.014933032914996147, "grad_norm": 0.09305793046951294, "global_step": 24905, "epoch": 234, "lr": 5.471721798696491e-05} +{"train_loss": 0.016228165477514267, "grad_norm": 0.055725302547216415, "global_step": 24906, "epoch": 234, "lr": 5.4714267431540575e-05} +{"train_loss": 0.015870321542024612, "grad_norm": 0.06425096839666367, "global_step": 24907, "epoch": 234, "lr": 5.4711316859552355e-05} +{"train_loss": 0.014712046831846237, "grad_norm": 0.059590838849544525, "global_step": 24908, "epoch": 234, "lr": 5.4708366271010604e-05} +{"train_loss": 0.015696001388484013, "grad_norm": 0.11005467921495438, "global_step": 24909, "epoch": 234, "lr": 5.470541566592573e-05, "val_loss": 0.019678007811307907} +{"train_loss": 0.014029387384653091, "grad_norm": 0.06030266359448433, "global_step": 24910, "epoch": 235, "lr": 5.470246504430805e-05} +{"train_loss": 0.014768010005354881, "grad_norm": 0.09071695804595947, "global_step": 24911, "epoch": 235, "lr": 5.469951440616795e-05} +{"train_loss": 0.015498301014304161, "grad_norm": 0.07142110168933868, "global_step": 24912, "epoch": 235, "lr": 5.4696563751515826e-05} +{"train_loss": 0.01565754972398281, "grad_norm": 0.08103258162736893, "global_step": 24913, "epoch": 235, "lr": 5.4693613080361997e-05} +{"train_loss": 0.01414508931338787, "grad_norm": 0.0887034460902214, "global_step": 24914, "epoch": 235, "lr": 5.469066239271685e-05} +{"train_loss": 0.015297588892281055, "grad_norm": 0.11588742583990097, "global_step": 24915, "epoch": 235, "lr": 5.468771168859077e-05} +{"train_loss": 0.017245739698410034, "grad_norm": 0.07480263710021973, "global_step": 24916, "epoch": 235, "lr": 5.4684760967994096e-05} +{"train_loss": 0.0169566348195076, "grad_norm": 0.06994858384132385, "global_step": 24917, "epoch": 235, "lr": 5.468181023093723e-05} +{"train_loss": 0.0153274554759264, "grad_norm": 0.08948315680027008, "global_step": 24918, "epoch": 235, "lr": 5.46788594774305e-05} +{"train_loss": 0.016053926199674606, "grad_norm": 0.07920871675014496, "global_step": 24919, "epoch": 235, "lr": 5.4675908707484294e-05} +{"train_loss": 0.017422867938876152, "grad_norm": 0.08980097621679306, "global_step": 24920, "epoch": 235, "lr": 5.467295792110897e-05} +{"train_loss": 0.013349118642508984, "grad_norm": 0.09844417124986649, "global_step": 24921, "epoch": 235, "lr": 5.4670007118314933e-05} +{"train_loss": 0.014903327450156212, "grad_norm": 0.0803418755531311, "global_step": 24922, "epoch": 235, "lr": 5.46670562991125e-05} +{"train_loss": 0.01715845614671707, "grad_norm": 0.07489161193370819, "global_step": 24923, "epoch": 235, "lr": 5.466410546351206e-05} +{"train_loss": 0.015615816228091717, "grad_norm": 0.11759690940380096, "global_step": 24924, "epoch": 235, "lr": 5.4661154611524e-05} +{"train_loss": 0.013714387081563473, "grad_norm": 0.08154455572366714, "global_step": 24925, "epoch": 235, "lr": 5.4658203743158654e-05} +{"train_loss": 0.01652006432414055, "grad_norm": 0.07833194732666016, "global_step": 24926, "epoch": 235, "lr": 5.465525285842641e-05} +{"train_loss": 0.016454532742500305, "grad_norm": 0.055524539202451706, "global_step": 24927, "epoch": 235, "lr": 5.4652301957337624e-05} +{"train_loss": 0.015530705451965332, "grad_norm": 0.12396927922964096, "global_step": 24928, "epoch": 235, "lr": 5.4649351039902684e-05} +{"train_loss": 0.01513246912509203, "grad_norm": 0.05736807361245155, "global_step": 24929, "epoch": 235, "lr": 5.464640010613195e-05} +{"train_loss": 0.015316713601350784, "grad_norm": 0.05073250085115433, "global_step": 24930, "epoch": 235, "lr": 5.4643449156035775e-05} +{"train_loss": 0.015582486987113953, "grad_norm": 0.0846409797668457, "global_step": 24931, "epoch": 235, "lr": 5.464049818962454e-05} +{"train_loss": 0.017518021166324615, "grad_norm": 0.15314052999019623, "global_step": 24932, "epoch": 235, "lr": 5.46375472069086e-05} +{"train_loss": 0.013679636642336845, "grad_norm": 0.13567163050174713, "global_step": 24933, "epoch": 235, "lr": 5.463459620789837e-05} +{"train_loss": 0.015796350315213203, "grad_norm": 0.06447223573923111, "global_step": 24934, "epoch": 235, "lr": 5.463164519260416e-05} +{"train_loss": 0.014519801363348961, "grad_norm": 0.06397759914398193, "global_step": 24935, "epoch": 235, "lr": 5.4628694161036364e-05} +{"train_loss": 0.015115110203623772, "grad_norm": 0.04875732213258743, "global_step": 24936, "epoch": 235, "lr": 5.462574311320536e-05} +{"train_loss": 0.015619019977748394, "grad_norm": 0.11606283485889435, "global_step": 24937, "epoch": 235, "lr": 5.4622792049121493e-05} +{"train_loss": 0.013698909431695938, "grad_norm": 0.08441942185163498, "global_step": 24938, "epoch": 235, "lr": 5.461984096879516e-05} +{"train_loss": 0.01799199730157852, "grad_norm": 0.1075572744011879, "global_step": 24939, "epoch": 235, "lr": 5.461688987223671e-05} +{"train_loss": 0.014524608850479126, "grad_norm": 0.07438942790031433, "global_step": 24940, "epoch": 235, "lr": 5.461393875945651e-05} +{"train_loss": 0.012663740664720535, "grad_norm": 0.054475992918014526, "global_step": 24941, "epoch": 235, "lr": 5.461098763046494e-05} +{"train_loss": 0.017135750502347946, "grad_norm": 0.11601395905017853, "global_step": 24942, "epoch": 235, "lr": 5.460803648527237e-05} +{"train_loss": 0.01595376618206501, "grad_norm": 0.06877898424863815, "global_step": 24943, "epoch": 235, "lr": 5.460508532388916e-05} +{"train_loss": 0.01742761954665184, "grad_norm": 0.05624747648835182, "global_step": 24944, "epoch": 235, "lr": 5.460213414632568e-05} +{"train_loss": 0.013581410050392151, "grad_norm": 0.06339351087808609, "global_step": 24945, "epoch": 235, "lr": 5.459918295259231e-05} +{"train_loss": 0.015126562677323818, "grad_norm": 0.0874592587351799, "global_step": 24946, "epoch": 235, "lr": 5.45962317426994e-05} +{"train_loss": 0.016746513545513153, "grad_norm": 0.10763376206159592, "global_step": 24947, "epoch": 235, "lr": 5.459328051665733e-05} +{"train_loss": 0.01628408581018448, "grad_norm": 0.08874499052762985, "global_step": 24948, "epoch": 235, "lr": 5.4590329274476484e-05} +{"train_loss": 0.01531109120696783, "grad_norm": 0.09284248948097229, "global_step": 24949, "epoch": 235, "lr": 5.45873780161672e-05} +{"train_loss": 0.01569806970655918, "grad_norm": 0.08229201287031174, "global_step": 24950, "epoch": 235, "lr": 5.458442674173987e-05} +{"train_loss": 0.016086269170045853, "grad_norm": 0.05606485530734062, "global_step": 24951, "epoch": 235, "lr": 5.458147545120487e-05} +{"train_loss": 0.016731232404708862, "grad_norm": 0.06553088128566742, "global_step": 24952, "epoch": 235, "lr": 5.457852414457254e-05} +{"train_loss": 0.01461346261203289, "grad_norm": 0.07792175561189651, "global_step": 24953, "epoch": 235, "lr": 5.457557282185328e-05} +{"train_loss": 0.015137420035898685, "grad_norm": 0.0726156160235405, "global_step": 24954, "epoch": 235, "lr": 5.457262148305744e-05} +{"train_loss": 0.014230269938707352, "grad_norm": 0.08542655408382416, "global_step": 24955, "epoch": 235, "lr": 5.4569670128195406e-05} +{"train_loss": 0.017117928713560104, "grad_norm": 0.10747991502285004, "global_step": 24956, "epoch": 235, "lr": 5.456671875727754e-05} +{"train_loss": 0.01447097584605217, "grad_norm": 0.06909888237714767, "global_step": 24957, "epoch": 235, "lr": 5.45637673703142e-05} +{"train_loss": 0.012539168819785118, "grad_norm": 0.07671031355857849, "global_step": 24958, "epoch": 235, "lr": 5.456081596731577e-05} +{"train_loss": 0.017795592546463013, "grad_norm": 0.0924597829580307, "global_step": 24959, "epoch": 235, "lr": 5.4557864548292616e-05} +{"train_loss": 0.015307619236409664, "grad_norm": 0.05838959664106369, "global_step": 24960, "epoch": 235, "lr": 5.45549131132551e-05} +{"train_loss": 0.015524161979556084, "grad_norm": 0.1058051660656929, "global_step": 24961, "epoch": 235, "lr": 5.45519616622136e-05} +{"train_loss": 0.014952065423130989, "grad_norm": 0.05609089508652687, "global_step": 24962, "epoch": 235, "lr": 5.4549010195178505e-05} +{"train_loss": 0.015839722007513046, "grad_norm": 0.07587237656116486, "global_step": 24963, "epoch": 235, "lr": 5.454605871216015e-05} +{"train_loss": 0.013267950154840946, "grad_norm": 0.05628424510359764, "global_step": 24964, "epoch": 235, "lr": 5.454310721316892e-05} +{"train_loss": 0.01699594408273697, "grad_norm": 0.07401081919670105, "global_step": 24965, "epoch": 235, "lr": 5.4540155698215215e-05} +{"train_loss": 0.015246224589645863, "grad_norm": 0.07203833013772964, "global_step": 24966, "epoch": 235, "lr": 5.4537204167309344e-05} +{"train_loss": 0.015690024942159653, "grad_norm": 0.09060923010110855, "global_step": 24967, "epoch": 235, "lr": 5.453425262046172e-05} +{"train_loss": 0.014446569606661797, "grad_norm": 0.0629040002822876, "global_step": 24968, "epoch": 235, "lr": 5.453130105768271e-05} +{"train_loss": 0.014492545276880264, "grad_norm": 0.06474221497774124, "global_step": 24969, "epoch": 235, "lr": 5.4528349478982676e-05} +{"train_loss": 0.01423032395541668, "grad_norm": 0.11774206906557083, "global_step": 24970, "epoch": 235, "lr": 5.4525397884371986e-05} +{"train_loss": 0.01838330738246441, "grad_norm": 0.08418676257133484, "global_step": 24971, "epoch": 235, "lr": 5.452244627386103e-05} +{"train_loss": 0.017380615696310997, "grad_norm": 0.05902167037129402, "global_step": 24972, "epoch": 235, "lr": 5.4519494647460156e-05} +{"train_loss": 0.015517100691795349, "grad_norm": 0.06918242573738098, "global_step": 24973, "epoch": 235, "lr": 5.451654300517973e-05} +{"train_loss": 0.01478457823395729, "grad_norm": 0.08153531700372696, "global_step": 24974, "epoch": 235, "lr": 5.451359134703016e-05} +{"train_loss": 0.01731569692492485, "grad_norm": 0.10339874029159546, "global_step": 24975, "epoch": 235, "lr": 5.451063967302178e-05} +{"train_loss": 0.016177862882614136, "grad_norm": 0.08675321191549301, "global_step": 24976, "epoch": 235, "lr": 5.450768798316497e-05} +{"train_loss": 0.01703650876879692, "grad_norm": 0.08599849790334702, "global_step": 24977, "epoch": 235, "lr": 5.450473627747012e-05} +{"train_loss": 0.013441940769553185, "grad_norm": 0.07170110940933228, "global_step": 24978, "epoch": 235, "lr": 5.4501784555947564e-05} +{"train_loss": 0.01571837067604065, "grad_norm": 0.08362060785293579, "global_step": 24979, "epoch": 235, "lr": 5.4498832818607716e-05} +{"train_loss": 0.014522275887429714, "grad_norm": 0.07395458966493607, "global_step": 24980, "epoch": 235, "lr": 5.4495881065460904e-05} +{"train_loss": 0.013051401823759079, "grad_norm": 0.09135811030864716, "global_step": 24981, "epoch": 235, "lr": 5.4492929296517545e-05} +{"train_loss": 0.013598012737929821, "grad_norm": 0.06855058670043945, "global_step": 24982, "epoch": 235, "lr": 5.448997751178797e-05} +{"train_loss": 0.01596793159842491, "grad_norm": 0.12071088701486588, "global_step": 24983, "epoch": 235, "lr": 5.448702571128258e-05} +{"train_loss": 0.016603736206889153, "grad_norm": 0.14385823905467987, "global_step": 24984, "epoch": 235, "lr": 5.448407389501172e-05} +{"train_loss": 0.014847107231616974, "grad_norm": 0.10858629643917084, "global_step": 24985, "epoch": 235, "lr": 5.448112206298579e-05} +{"train_loss": 0.015732252970337868, "grad_norm": 0.12830445170402527, "global_step": 24986, "epoch": 235, "lr": 5.447817021521513e-05} +{"train_loss": 0.013316456228494644, "grad_norm": 0.08035020530223846, "global_step": 24987, "epoch": 235, "lr": 5.4475218351710135e-05} +{"train_loss": 0.01729718968272209, "grad_norm": 0.12251584231853485, "global_step": 24988, "epoch": 235, "lr": 5.447226647248116e-05} +{"train_loss": 0.01396622508764267, "grad_norm": 0.0908965915441513, "global_step": 24989, "epoch": 235, "lr": 5.44693145775386e-05} +{"train_loss": 0.016868414357304573, "grad_norm": 0.07692781090736389, "global_step": 24990, "epoch": 235, "lr": 5.4466362666892804e-05} +{"train_loss": 0.019708385691046715, "grad_norm": 0.11620406806468964, "global_step": 24991, "epoch": 235, "lr": 5.446341074055414e-05} +{"train_loss": 0.013668897561728954, "grad_norm": 0.10359501093626022, "global_step": 24992, "epoch": 235, "lr": 5.446045879853302e-05} +{"train_loss": 0.017733804881572723, "grad_norm": 0.06447545439004898, "global_step": 24993, "epoch": 235, "lr": 5.445750684083977e-05} +{"train_loss": 0.013634772971272469, "grad_norm": 0.05929621681571007, "global_step": 24994, "epoch": 235, "lr": 5.4454554867484774e-05} +{"train_loss": 0.015301248989999294, "grad_norm": 0.08166791498661041, "global_step": 24995, "epoch": 235, "lr": 5.445160287847841e-05} +{"train_loss": 0.01734144799411297, "grad_norm": 0.06714365631341934, "global_step": 24996, "epoch": 235, "lr": 5.4448650873831067e-05} +{"train_loss": 0.017299052327871323, "grad_norm": 0.06758546084165573, "global_step": 24997, "epoch": 235, "lr": 5.444569885355308e-05} +{"train_loss": 0.014428731054067612, "grad_norm": 0.059073805809020996, "global_step": 24998, "epoch": 235, "lr": 5.4442746817654855e-05} +{"train_loss": 0.014673980884253979, "grad_norm": 0.058326102793216705, "global_step": 24999, "epoch": 235, "lr": 5.4439794766146746e-05} +{"train_loss": 0.017178241163492203, "grad_norm": 0.14716491103172302, "global_step": 25000, "epoch": 235, "lr": 5.4436842699039115e-05} +{"train_loss": 0.015272394753992558, "grad_norm": 0.08339071273803711, "global_step": 25001, "epoch": 235, "lr": 5.443389061634238e-05} +{"train_loss": 0.016041859984397888, "grad_norm": 0.05625681206583977, "global_step": 25002, "epoch": 235, "lr": 5.443093851806684e-05} +{"train_loss": 0.01719561219215393, "grad_norm": 0.08859352022409439, "global_step": 25003, "epoch": 235, "lr": 5.442798640422294e-05} +{"train_loss": 0.020160280168056488, "grad_norm": 0.07932545989751816, "global_step": 25004, "epoch": 235, "lr": 5.442503427482102e-05} +{"train_loss": 0.016220755875110626, "grad_norm": 0.06088491529226303, "global_step": 25005, "epoch": 235, "lr": 5.442208212987144e-05} +{"train_loss": 0.01599704846739769, "grad_norm": 0.12084708362817764, "global_step": 25006, "epoch": 235, "lr": 5.4419129969384595e-05} +{"train_loss": 0.01494397409260273, "grad_norm": 0.09561973065137863, "global_step": 25007, "epoch": 235, "lr": 5.441617779337085e-05} +{"train_loss": 0.015574341639876366, "grad_norm": 0.07205232232809067, "global_step": 25008, "epoch": 235, "lr": 5.441322560184057e-05} +{"train_loss": 0.01795630156993866, "grad_norm": 0.09633769094944, "global_step": 25009, "epoch": 235, "lr": 5.441027339480413e-05} +{"train_loss": 0.016317158937454224, "grad_norm": 0.07193585485219955, "global_step": 25010, "epoch": 235, "lr": 5.440732117227193e-05} +{"train_loss": 0.01904689520597458, "grad_norm": 0.0958481952548027, "global_step": 25011, "epoch": 235, "lr": 5.4404368934254304e-05} +{"train_loss": 0.015580184757709503, "grad_norm": 0.13033604621887207, "global_step": 25012, "epoch": 235, "lr": 5.440141668076164e-05} +{"train_loss": 0.015413791872560978, "grad_norm": 0.06406713277101517, "global_step": 25013, "epoch": 235, "lr": 5.439846441180432e-05} +{"train_loss": 0.015998851507902145, "grad_norm": 0.15492257475852966, "global_step": 25014, "epoch": 235, "lr": 5.43955121273927e-05} +{"train_loss": 0.01566710116622864, "grad_norm": 0.08343837410211563, "global_step": 25015, "epoch": 235, "lr": 5.439255982753717e-05, "val_loss": 0.019159767776727676} +{"train_loss": 0.01574822887778282, "grad_norm": 0.15437081456184387, "global_step": 25016, "epoch": 236, "lr": 5.4389607512248085e-05} +{"train_loss": 0.016584297642111778, "grad_norm": 0.12729060649871826, "global_step": 25017, "epoch": 236, "lr": 5.4386655181535826e-05} +{"train_loss": 0.01541665755212307, "grad_norm": 0.0684821754693985, "global_step": 25018, "epoch": 236, "lr": 5.4383702835410786e-05} +{"train_loss": 0.016971804201602936, "grad_norm": 0.08977828174829483, "global_step": 25019, "epoch": 236, "lr": 5.4380750473883305e-05} +{"train_loss": 0.018486667424440384, "grad_norm": 0.08963246643543243, "global_step": 25020, "epoch": 236, "lr": 5.437779809696377e-05} +{"train_loss": 0.015382439829409122, "grad_norm": 0.11596155911684036, "global_step": 25021, "epoch": 236, "lr": 5.437484570466258e-05} +{"train_loss": 0.0146256722509861, "grad_norm": 0.09755636751651764, "global_step": 25022, "epoch": 236, "lr": 5.437189329699006e-05} +{"train_loss": 0.017642660066485405, "grad_norm": 0.09769722074270248, "global_step": 25023, "epoch": 236, "lr": 5.436894087395662e-05} +{"train_loss": 0.017001260071992874, "grad_norm": 0.07287468761205673, "global_step": 25024, "epoch": 236, "lr": 5.436598843557262e-05} +{"train_loss": 0.01616756245493889, "grad_norm": 0.06484466791152954, "global_step": 25025, "epoch": 236, "lr": 5.436303598184843e-05} +{"train_loss": 0.014330476522445679, "grad_norm": 0.06017974019050598, "global_step": 25026, "epoch": 236, "lr": 5.436008351279443e-05} +{"train_loss": 0.01555272750556469, "grad_norm": 0.09430480748414993, "global_step": 25027, "epoch": 236, "lr": 5.4357131028421005e-05} +{"train_loss": 0.015882708132267, "grad_norm": 0.0715029388666153, "global_step": 25028, "epoch": 236, "lr": 5.4354178528738506e-05} +{"train_loss": 0.013024525716900826, "grad_norm": 0.09405140578746796, "global_step": 25029, "epoch": 236, "lr": 5.4351226013757306e-05} +{"train_loss": 0.0169207900762558, "grad_norm": 0.11327336728572845, "global_step": 25030, "epoch": 236, "lr": 5.4348273483487825e-05} +{"train_loss": 0.015831386670470238, "grad_norm": 0.15805450081825256, "global_step": 25031, "epoch": 236, "lr": 5.4345320937940356e-05} +{"train_loss": 0.015341443940997124, "grad_norm": 0.06940074265003204, "global_step": 25032, "epoch": 236, "lr": 5.434236837712535e-05} +{"train_loss": 0.015273054130375385, "grad_norm": 0.08081063628196716, "global_step": 25033, "epoch": 236, "lr": 5.433941580105315e-05} +{"train_loss": 0.0181315615773201, "grad_norm": 0.06812320649623871, "global_step": 25034, "epoch": 236, "lr": 5.433646320973411e-05} +{"train_loss": 0.018305832520127296, "grad_norm": 0.0722605437040329, "global_step": 25035, "epoch": 236, "lr": 5.433351060317864e-05} +{"train_loss": 0.015495953150093555, "grad_norm": 0.09036920964717865, "global_step": 25036, "epoch": 236, "lr": 5.43305579813971e-05} +{"train_loss": 0.015560660511255264, "grad_norm": 0.06557589769363403, "global_step": 25037, "epoch": 236, "lr": 5.4327605344399854e-05} +{"train_loss": 0.016865499317646027, "grad_norm": 0.1030639186501503, "global_step": 25038, "epoch": 236, "lr": 5.43246526921973e-05} +{"train_loss": 0.013858936727046967, "grad_norm": 0.07449331134557724, "global_step": 25039, "epoch": 236, "lr": 5.432170002479978e-05} +{"train_loss": 0.012919245287775993, "grad_norm": 0.0942942202091217, "global_step": 25040, "epoch": 236, "lr": 5.431874734221769e-05} +{"train_loss": 0.016352064907550812, "grad_norm": 0.12431198358535767, "global_step": 25041, "epoch": 236, "lr": 5.4315794644461396e-05} +{"train_loss": 0.015023469924926758, "grad_norm": 0.07262533158063889, "global_step": 25042, "epoch": 236, "lr": 5.4312841931541294e-05} +{"train_loss": 0.013971409760415554, "grad_norm": 0.0657319575548172, "global_step": 25043, "epoch": 236, "lr": 5.4309889203467725e-05} +{"train_loss": 0.01462380588054657, "grad_norm": 0.104672871530056, "global_step": 25044, "epoch": 236, "lr": 5.430693646025108e-05} +{"train_loss": 0.015399893745779991, "grad_norm": 0.0996813252568245, "global_step": 25045, "epoch": 236, "lr": 5.430398370190174e-05} +{"train_loss": 0.01608838513493538, "grad_norm": 0.08305816352367401, "global_step": 25046, "epoch": 236, "lr": 5.4301030928430065e-05} +{"train_loss": 0.0187092125415802, "grad_norm": 0.06107950955629349, "global_step": 25047, "epoch": 236, "lr": 5.429807813984644e-05} +{"train_loss": 0.017905548214912415, "grad_norm": 0.07727614790201187, "global_step": 25048, "epoch": 236, "lr": 5.4295125336161244e-05} +{"train_loss": 0.014900727197527885, "grad_norm": 0.0840948149561882, "global_step": 25049, "epoch": 236, "lr": 5.429217251738484e-05} +{"train_loss": 0.014165177009999752, "grad_norm": 0.07769498229026794, "global_step": 25050, "epoch": 236, "lr": 5.4289219683527617e-05} +{"train_loss": 0.016356609761714935, "grad_norm": 0.07331153750419617, "global_step": 25051, "epoch": 236, "lr": 5.428626683459993e-05} +{"train_loss": 0.018768707290291786, "grad_norm": 0.07231040298938751, "global_step": 25052, "epoch": 236, "lr": 5.428331397061217e-05} +{"train_loss": 0.014940169639885426, "grad_norm": 0.07973747700452805, "global_step": 25053, "epoch": 236, "lr": 5.428036109157469e-05} +{"train_loss": 0.013871327042579651, "grad_norm": 0.05555589124560356, "global_step": 25054, "epoch": 236, "lr": 5.4277408197497914e-05} +{"train_loss": 0.017181461676955223, "grad_norm": 0.12180780619382858, "global_step": 25055, "epoch": 236, "lr": 5.427445528839218e-05} +{"train_loss": 0.013982008211314678, "grad_norm": 0.05669577419757843, "global_step": 25056, "epoch": 236, "lr": 5.427150236426784e-05} +{"train_loss": 0.0165114626288414, "grad_norm": 0.06652196496725082, "global_step": 25057, "epoch": 236, "lr": 5.426854942513534e-05} +{"train_loss": 0.01744229905307293, "grad_norm": 0.08664311468601227, "global_step": 25058, "epoch": 236, "lr": 5.4265596471004976e-05} +{"train_loss": 0.01717028021812439, "grad_norm": 0.06607342511415482, "global_step": 25059, "epoch": 236, "lr": 5.426264350188719e-05} +{"train_loss": 0.01917869597673416, "grad_norm": 0.08193189650774002, "global_step": 25060, "epoch": 236, "lr": 5.4259690517792316e-05} +{"train_loss": 0.015232952311635017, "grad_norm": 0.12407933175563812, "global_step": 25061, "epoch": 236, "lr": 5.4256737518730735e-05} +{"train_loss": 0.01687578670680523, "grad_norm": 0.07055707275867462, "global_step": 25062, "epoch": 236, "lr": 5.425378450471286e-05} +{"train_loss": 0.01734272390604019, "grad_norm": 0.12152565270662308, "global_step": 25063, "epoch": 236, "lr": 5.4250831475748995e-05} +{"train_loss": 0.015473468229174614, "grad_norm": 0.06718472391366959, "global_step": 25064, "epoch": 236, "lr": 5.4247878431849576e-05} +{"train_loss": 0.01641807332634926, "grad_norm": 0.06074398010969162, "global_step": 25065, "epoch": 236, "lr": 5.4244925373024956e-05} +{"train_loss": 0.0166153684258461, "grad_norm": 0.07581967115402222, "global_step": 25066, "epoch": 236, "lr": 5.4241972299285516e-05} +{"train_loss": 0.016724228858947754, "grad_norm": 0.070909284055233, "global_step": 25067, "epoch": 236, "lr": 5.4239019210641636e-05} +{"train_loss": 0.01490966696292162, "grad_norm": 0.111148901283741, "global_step": 25068, "epoch": 236, "lr": 5.423606610710368e-05} +{"train_loss": 0.016100846230983734, "grad_norm": 0.08703317493200302, "global_step": 25069, "epoch": 236, "lr": 5.423311298868203e-05} +{"train_loss": 0.015345839783549309, "grad_norm": 0.06631585955619812, "global_step": 25070, "epoch": 236, "lr": 5.423015985538705e-05} +{"train_loss": 0.015678483992815018, "grad_norm": 0.06840493530035019, "global_step": 25071, "epoch": 236, "lr": 5.422720670722916e-05} +{"train_loss": 0.01818954572081566, "grad_norm": 0.10972172766923904, "global_step": 25072, "epoch": 236, "lr": 5.422425354421868e-05} +{"train_loss": 0.015363918617367744, "grad_norm": 0.10013175755739212, "global_step": 25073, "epoch": 236, "lr": 5.4221300366366004e-05} +{"train_loss": 0.016688302159309387, "grad_norm": 0.0719740018248558, "global_step": 25074, "epoch": 236, "lr": 5.421834717368154e-05} +{"train_loss": 0.014969389885663986, "grad_norm": 0.0833970308303833, "global_step": 25075, "epoch": 236, "lr": 5.421539396617561e-05} +{"train_loss": 0.014804059639573097, "grad_norm": 0.08846698701381683, "global_step": 25076, "epoch": 236, "lr": 5.421244074385864e-05} +{"train_loss": 0.013728069141507149, "grad_norm": 0.08005394041538239, "global_step": 25077, "epoch": 236, "lr": 5.420948750674097e-05} +{"train_loss": 0.015121204778552055, "grad_norm": 0.1616552174091339, "global_step": 25078, "epoch": 236, "lr": 5.420653425483301e-05} +{"train_loss": 0.014749421738088131, "grad_norm": 0.15431320667266846, "global_step": 25079, "epoch": 236, "lr": 5.4203580988145095e-05} +{"train_loss": 0.014728979207575321, "grad_norm": 0.07081722468137741, "global_step": 25080, "epoch": 236, "lr": 5.420062770668764e-05} +{"train_loss": 0.015074039809405804, "grad_norm": 0.08864481747150421, "global_step": 25081, "epoch": 236, "lr": 5.4197674410471014e-05} +{"train_loss": 0.014888792298734188, "grad_norm": 0.10970379412174225, "global_step": 25082, "epoch": 236, "lr": 5.419472109950555e-05} +{"train_loss": 0.015261899679899216, "grad_norm": 0.10302072763442993, "global_step": 25083, "epoch": 236, "lr": 5.4191767773801706e-05} +{"train_loss": 0.014456442557275295, "grad_norm": 0.060905326157808304, "global_step": 25084, "epoch": 236, "lr": 5.4188814433369795e-05} +{"train_loss": 0.015545373782515526, "grad_norm": 0.065448097884655, "global_step": 25085, "epoch": 236, "lr": 5.418586107822019e-05} +{"train_loss": 0.013987944461405277, "grad_norm": 0.10699462890625, "global_step": 25086, "epoch": 236, "lr": 5.418290770836332e-05} +{"train_loss": 0.016443558037281036, "grad_norm": 0.07149070501327515, "global_step": 25087, "epoch": 236, "lr": 5.417995432380951e-05} +{"train_loss": 0.015203225426375866, "grad_norm": 0.07247688621282578, "global_step": 25088, "epoch": 236, "lr": 5.4177000924569164e-05} +{"train_loss": 0.016639776527881622, "grad_norm": 0.2155803143978119, "global_step": 25089, "epoch": 236, "lr": 5.417404751065266e-05} +{"train_loss": 0.014551319181919098, "grad_norm": 0.19119590520858765, "global_step": 25090, "epoch": 236, "lr": 5.4171094082070375e-05} +{"train_loss": 0.0164461862295866, "grad_norm": 0.12960802018642426, "global_step": 25091, "epoch": 236, "lr": 5.4168140638832666e-05} +{"train_loss": 0.01600806973874569, "grad_norm": 0.06411595642566681, "global_step": 25092, "epoch": 236, "lr": 5.4165187180949926e-05} +{"train_loss": 0.01709989830851555, "grad_norm": 0.2283211201429367, "global_step": 25093, "epoch": 236, "lr": 5.416223370843253e-05} +{"train_loss": 0.016064833849668503, "grad_norm": 0.08089860528707504, "global_step": 25094, "epoch": 236, "lr": 5.4159280221290854e-05} +{"train_loss": 0.01628100499510765, "grad_norm": 0.17158958315849304, "global_step": 25095, "epoch": 236, "lr": 5.415632671953528e-05} +{"train_loss": 0.015939578413963318, "grad_norm": 0.12013030052185059, "global_step": 25096, "epoch": 236, "lr": 5.415337320317618e-05} +{"train_loss": 0.017452968284487724, "grad_norm": 0.07684348523616791, "global_step": 25097, "epoch": 236, "lr": 5.4150419672223915e-05} +{"train_loss": 0.013853725977241993, "grad_norm": 0.1107056587934494, "global_step": 25098, "epoch": 236, "lr": 5.414746612668892e-05} +{"train_loss": 0.016223270446062088, "grad_norm": 0.12355770170688629, "global_step": 25099, "epoch": 236, "lr": 5.414451256658149e-05} +{"train_loss": 0.014795331284403801, "grad_norm": 0.04907817766070366, "global_step": 25100, "epoch": 236, "lr": 5.4141558991912067e-05} +{"train_loss": 0.013800887390971184, "grad_norm": 0.06008811295032501, "global_step": 25101, "epoch": 236, "lr": 5.413860540269102e-05} +{"train_loss": 0.01559691783040762, "grad_norm": 0.09637555480003357, "global_step": 25102, "epoch": 236, "lr": 5.413565179892869e-05} +{"train_loss": 0.013942931778728962, "grad_norm": 0.07260735332965851, "global_step": 25103, "epoch": 236, "lr": 5.413269818063549e-05} +{"train_loss": 0.014132167212665081, "grad_norm": 0.06828490644693375, "global_step": 25104, "epoch": 236, "lr": 5.412974454782178e-05} +{"train_loss": 0.015153865329921246, "grad_norm": 0.1170244812965393, "global_step": 25105, "epoch": 236, "lr": 5.412679090049795e-05} +{"train_loss": 0.014430433511734009, "grad_norm": 0.11534876376390457, "global_step": 25106, "epoch": 236, "lr": 5.412383723867437e-05} +{"train_loss": 0.017209433019161224, "grad_norm": 0.06413159519433975, "global_step": 25107, "epoch": 236, "lr": 5.4120883562361415e-05} +{"train_loss": 0.012879163026809692, "grad_norm": 0.14620789885520935, "global_step": 25108, "epoch": 236, "lr": 5.411792987156947e-05} +{"train_loss": 0.01441238448023796, "grad_norm": 0.127753347158432, "global_step": 25109, "epoch": 236, "lr": 5.411497616630891e-05} +{"train_loss": 0.016853727400302887, "grad_norm": 0.13858970999717712, "global_step": 25110, "epoch": 236, "lr": 5.411202244659012e-05} +{"train_loss": 0.01562923938035965, "grad_norm": 0.09308816492557526, "global_step": 25111, "epoch": 236, "lr": 5.410906871242345e-05} +{"train_loss": 0.014722866006195545, "grad_norm": 0.05854783579707146, "global_step": 25112, "epoch": 236, "lr": 5.410611496381932e-05} +{"train_loss": 0.01430183369666338, "grad_norm": 0.10939595103263855, "global_step": 25113, "epoch": 236, "lr": 5.4103161200788077e-05} +{"train_loss": 0.013393253087997437, "grad_norm": 0.06757881492376328, "global_step": 25114, "epoch": 236, "lr": 5.410020742334011e-05} +{"train_loss": 0.016413019970059395, "grad_norm": 0.07982642203569412, "global_step": 25115, "epoch": 236, "lr": 5.409725363148581e-05} +{"train_loss": 0.015707172453403473, "grad_norm": 0.06056388467550278, "global_step": 25116, "epoch": 236, "lr": 5.409429982523553e-05} +{"train_loss": 0.01513631921261549, "grad_norm": 0.10403968393802643, "global_step": 25117, "epoch": 236, "lr": 5.409134600459966e-05} +{"train_loss": 0.017694421112537384, "grad_norm": 0.1996488720178604, "global_step": 25118, "epoch": 236, "lr": 5.4088392169588587e-05} +{"train_loss": 0.013685113750398159, "grad_norm": 0.1430988758802414, "global_step": 25119, "epoch": 236, "lr": 5.408543832021268e-05} +{"train_loss": 0.016985667869448662, "grad_norm": 0.14684626460075378, "global_step": 25120, "epoch": 236, "lr": 5.408248445648232e-05} +{"train_loss": 0.01569884920598201, "grad_norm": 0.10871955007314682, "global_step": 25121, "epoch": 236, "lr": 5.407953057840789e-05, "val_loss": 0.019522253423929214} +{"train_loss": 0.017424950376152992, "grad_norm": 0.14876693487167358, "global_step": 25122, "epoch": 237, "lr": 5.4076576685999744e-05} +{"train_loss": 0.017857752740383148, "grad_norm": 0.18566574156284332, "global_step": 25123, "epoch": 237, "lr": 5.407362277926831e-05} +{"train_loss": 0.014636875130236149, "grad_norm": 0.06961997598409653, "global_step": 25124, "epoch": 237, "lr": 5.4070668858223916e-05} +{"train_loss": 0.01841682940721512, "grad_norm": 0.08997999131679535, "global_step": 25125, "epoch": 237, "lr": 5.4067714922876964e-05} +{"train_loss": 0.01445687748491764, "grad_norm": 0.057974833995103836, "global_step": 25126, "epoch": 237, "lr": 5.406476097323783e-05} +{"train_loss": 0.015265114605426788, "grad_norm": 0.09875068068504333, "global_step": 25127, "epoch": 237, "lr": 5.406180700931691e-05} +{"train_loss": 0.018299177289009094, "grad_norm": 0.09291883558034897, "global_step": 25128, "epoch": 237, "lr": 5.4058853031124534e-05} +{"train_loss": 0.015788353979587555, "grad_norm": 0.05226879566907883, "global_step": 25129, "epoch": 237, "lr": 5.405589903867114e-05} +{"train_loss": 0.015291348099708557, "grad_norm": 0.07497580349445343, "global_step": 25130, "epoch": 237, "lr": 5.4052945031967086e-05} +{"train_loss": 0.014532862231135368, "grad_norm": 0.06491352617740631, "global_step": 25131, "epoch": 237, "lr": 5.4049991011022726e-05} +{"train_loss": 0.019517067819833755, "grad_norm": 0.14755375683307648, "global_step": 25132, "epoch": 237, "lr": 5.404703697584846e-05} +{"train_loss": 0.018175557255744934, "grad_norm": 0.09885863959789276, "global_step": 25133, "epoch": 237, "lr": 5.404408292645468e-05} +{"train_loss": 0.017061466351151466, "grad_norm": 0.22004525363445282, "global_step": 25134, "epoch": 237, "lr": 5.4041128862851745e-05} +{"train_loss": 0.012330884113907814, "grad_norm": 0.11683092266321182, "global_step": 25135, "epoch": 237, "lr": 5.403817478505005e-05} +{"train_loss": 0.014180577360093594, "grad_norm": 0.10454009473323822, "global_step": 25136, "epoch": 237, "lr": 5.403522069305994e-05} +{"train_loss": 0.017017796635627747, "grad_norm": 0.18538278341293335, "global_step": 25137, "epoch": 237, "lr": 5.4032266586891844e-05} +{"train_loss": 0.014051105827093124, "grad_norm": 0.05719161406159401, "global_step": 25138, "epoch": 237, "lr": 5.4029312466556095e-05} +{"train_loss": 0.014653687365353107, "grad_norm": 0.11536287516355515, "global_step": 25139, "epoch": 237, "lr": 5.402635833206312e-05} +{"train_loss": 0.015259956009685993, "grad_norm": 0.06328816711902618, "global_step": 25140, "epoch": 237, "lr": 5.402340418342324e-05} +{"train_loss": 0.015490330755710602, "grad_norm": 0.07766924053430557, "global_step": 25141, "epoch": 237, "lr": 5.4020450020646894e-05} +{"train_loss": 0.01480996236205101, "grad_norm": 0.08251360803842545, "global_step": 25142, "epoch": 237, "lr": 5.4017495843744435e-05} +{"train_loss": 0.015102964825928211, "grad_norm": 0.07828304171562195, "global_step": 25143, "epoch": 237, "lr": 5.4014541652726227e-05} +{"train_loss": 0.017691481858491898, "grad_norm": 0.0752781555056572, "global_step": 25144, "epoch": 237, "lr": 5.4011587447602676e-05} +{"train_loss": 0.015620993450284004, "grad_norm": 0.05968403443694115, "global_step": 25145, "epoch": 237, "lr": 5.400863322838415e-05} +{"train_loss": 0.015211448073387146, "grad_norm": 0.13215859234333038, "global_step": 25146, "epoch": 237, "lr": 5.4005678995081035e-05} +{"train_loss": 0.020354535430669785, "grad_norm": 0.10637548565864563, "global_step": 25147, "epoch": 237, "lr": 5.4002724747703694e-05} +{"train_loss": 0.01620173640549183, "grad_norm": 0.17416667938232422, "global_step": 25148, "epoch": 237, "lr": 5.399977048626252e-05} +{"train_loss": 0.018247565254569054, "grad_norm": 0.1610606461763382, "global_step": 25149, "epoch": 237, "lr": 5.399681621076791e-05} +{"train_loss": 0.014723102562129498, "grad_norm": 0.14917004108428955, "global_step": 25150, "epoch": 237, "lr": 5.399386192123019e-05} +{"train_loss": 0.014156230725347996, "grad_norm": 0.06156245619058609, "global_step": 25151, "epoch": 237, "lr": 5.3990907617659824e-05} +{"train_loss": 0.014600349590182304, "grad_norm": 0.11643099039793015, "global_step": 25152, "epoch": 237, "lr": 5.398795330006712e-05} +{"train_loss": 0.015280937775969505, "grad_norm": 0.117449089884758, "global_step": 25153, "epoch": 237, "lr": 5.3984998968462466e-05} +{"train_loss": 0.01683618500828743, "grad_norm": 0.05508704110980034, "global_step": 25154, "epoch": 237, "lr": 5.3982044622856285e-05} +{"train_loss": 0.015213534235954285, "grad_norm": 0.056445036083459854, "global_step": 25155, "epoch": 237, "lr": 5.3979090263258916e-05} +{"train_loss": 0.017935872077941895, "grad_norm": 0.09376409649848938, "global_step": 25156, "epoch": 237, "lr": 5.3976135889680753e-05} +{"train_loss": 0.01718241721391678, "grad_norm": 0.07396199554204941, "global_step": 25157, "epoch": 237, "lr": 5.3973181502132185e-05} +{"train_loss": 0.017236454412341118, "grad_norm": 0.06971392035484314, "global_step": 25158, "epoch": 237, "lr": 5.397022710062359e-05} +{"train_loss": 0.01308067049831152, "grad_norm": 0.09895934909582138, "global_step": 25159, "epoch": 237, "lr": 5.3967272685165335e-05} +{"train_loss": 0.016391068696975708, "grad_norm": 0.14292961359024048, "global_step": 25160, "epoch": 237, "lr": 5.396431825576781e-05} +{"train_loss": 0.013856901787221432, "grad_norm": 0.08169490098953247, "global_step": 25161, "epoch": 237, "lr": 5.39613638124414e-05} +{"train_loss": 0.015019135549664497, "grad_norm": 0.08485500514507294, "global_step": 25162, "epoch": 237, "lr": 5.395840935519648e-05} +{"train_loss": 0.014841096475720406, "grad_norm": 0.12827898561954498, "global_step": 25163, "epoch": 237, "lr": 5.3955454884043425e-05} +{"train_loss": 0.014556000009179115, "grad_norm": 0.20472203195095062, "global_step": 25164, "epoch": 237, "lr": 5.395250039899262e-05} +{"train_loss": 0.014279786497354507, "grad_norm": 0.11948343366384506, "global_step": 25165, "epoch": 237, "lr": 5.394954590005446e-05} +{"train_loss": 0.01510613039135933, "grad_norm": 0.06863804161548615, "global_step": 25166, "epoch": 237, "lr": 5.39465913872393e-05} +{"train_loss": 0.015155713073909283, "grad_norm": 0.10337604582309723, "global_step": 25167, "epoch": 237, "lr": 5.394363686055753e-05} +{"train_loss": 0.014890587888658047, "grad_norm": 0.06751686334609985, "global_step": 25168, "epoch": 237, "lr": 5.3940682320019565e-05} +{"train_loss": 0.015944700688123703, "grad_norm": 0.08418887108564377, "global_step": 25169, "epoch": 237, "lr": 5.3937727765635714e-05} +{"train_loss": 0.01605679839849472, "grad_norm": 0.06772330403327942, "global_step": 25170, "epoch": 237, "lr": 5.393477319741642e-05} +{"train_loss": 0.016120731830596924, "grad_norm": 0.09021598845720291, "global_step": 25171, "epoch": 237, "lr": 5.3931818615372065e-05} +{"train_loss": 0.0157932061702013, "grad_norm": 0.07349664717912674, "global_step": 25172, "epoch": 237, "lr": 5.392886401951298e-05} +{"train_loss": 0.013834434561431408, "grad_norm": 0.060941535979509354, "global_step": 25173, "epoch": 237, "lr": 5.3925909409849586e-05} +{"train_loss": 0.014861875213682652, "grad_norm": 0.08903414011001587, "global_step": 25174, "epoch": 237, "lr": 5.392295478639225e-05} +{"train_loss": 0.015221419744193554, "grad_norm": 0.07985341548919678, "global_step": 25175, "epoch": 237, "lr": 5.392000014915136e-05} +{"train_loss": 0.016402550041675568, "grad_norm": 0.09525183588266373, "global_step": 25176, "epoch": 237, "lr": 5.39170454981373e-05} +{"train_loss": 0.015247496776282787, "grad_norm": 0.05757768079638481, "global_step": 25177, "epoch": 237, "lr": 5.3914090833360445e-05} +{"train_loss": 0.015367644838988781, "grad_norm": 0.09343861788511276, "global_step": 25178, "epoch": 237, "lr": 5.391113615483116e-05} +{"train_loss": 0.015357567928731441, "grad_norm": 0.13397946953773499, "global_step": 25179, "epoch": 237, "lr": 5.390818146255985e-05} +{"train_loss": 0.015935657545924187, "grad_norm": 0.08375504612922668, "global_step": 25180, "epoch": 237, "lr": 5.3905226756556904e-05} +{"train_loss": 0.015287570655345917, "grad_norm": 0.09476130455732346, "global_step": 25181, "epoch": 237, "lr": 5.390227203683269e-05} +{"train_loss": 0.014332910999655724, "grad_norm": 0.09304238855838776, "global_step": 25182, "epoch": 237, "lr": 5.3899317303397566e-05} +{"train_loss": 0.01333312876522541, "grad_norm": 0.09135527908802032, "global_step": 25183, "epoch": 237, "lr": 5.3896362556261956e-05} +{"train_loss": 0.015665767714381218, "grad_norm": 0.08319953829050064, "global_step": 25184, "epoch": 237, "lr": 5.3893407795436214e-05} +{"train_loss": 0.01645112782716751, "grad_norm": 0.080349400639534, "global_step": 25185, "epoch": 237, "lr": 5.389045302093073e-05} +{"train_loss": 0.015694590285420418, "grad_norm": 0.06225232407450676, "global_step": 25186, "epoch": 237, "lr": 5.3887498232755895e-05} +{"train_loss": 0.013829994946718216, "grad_norm": 0.08717259019613266, "global_step": 25187, "epoch": 237, "lr": 5.3884543430922076e-05} +{"train_loss": 0.017610033974051476, "grad_norm": 0.10499149560928345, "global_step": 25188, "epoch": 237, "lr": 5.388158861543966e-05} +{"train_loss": 0.016765784472227097, "grad_norm": 0.09420439600944519, "global_step": 25189, "epoch": 237, "lr": 5.3878633786319035e-05} +{"train_loss": 0.014801356010138988, "grad_norm": 0.06900058686733246, "global_step": 25190, "epoch": 237, "lr": 5.3875678943570576e-05} +{"train_loss": 0.01332906074821949, "grad_norm": 0.05967109277844429, "global_step": 25191, "epoch": 237, "lr": 5.387272408720465e-05} +{"train_loss": 0.016300570219755173, "grad_norm": 0.07155004888772964, "global_step": 25192, "epoch": 237, "lr": 5.386976921723168e-05} +{"train_loss": 0.01470816321671009, "grad_norm": 0.07464559376239777, "global_step": 25193, "epoch": 237, "lr": 5.386681433366201e-05} +{"train_loss": 0.01468263566493988, "grad_norm": 0.14945709705352783, "global_step": 25194, "epoch": 237, "lr": 5.3863859436506034e-05} +{"train_loss": 0.017521029338240623, "grad_norm": 0.07297633588314056, "global_step": 25195, "epoch": 237, "lr": 5.386090452577415e-05} +{"train_loss": 0.01343696191906929, "grad_norm": 0.12499811500310898, "global_step": 25196, "epoch": 237, "lr": 5.385794960147671e-05} +{"train_loss": 0.014483396895229816, "grad_norm": 0.06790278851985931, "global_step": 25197, "epoch": 237, "lr": 5.3854994663624126e-05} +{"train_loss": 0.01657465659081936, "grad_norm": 0.054431524127721786, "global_step": 25198, "epoch": 237, "lr": 5.3852039712226765e-05} +{"train_loss": 0.01429060660302639, "grad_norm": 0.10206763446331024, "global_step": 25199, "epoch": 237, "lr": 5.384908474729501e-05} +{"train_loss": 0.017929406836628914, "grad_norm": 0.1111350804567337, "global_step": 25200, "epoch": 237, "lr": 5.3846129768839235e-05} +{"train_loss": 0.01478668674826622, "grad_norm": 0.08129511773586273, "global_step": 25201, "epoch": 237, "lr": 5.384317477686984e-05} +{"train_loss": 0.015494988299906254, "grad_norm": 0.05651369318366051, "global_step": 25202, "epoch": 237, "lr": 5.3840219771397207e-05} +{"train_loss": 0.01708703488111496, "grad_norm": 0.07988790422677994, "global_step": 25203, "epoch": 237, "lr": 5.3837264752431715e-05} +{"train_loss": 0.01572810858488083, "grad_norm": 0.06464061886072159, "global_step": 25204, "epoch": 237, "lr": 5.3834309719983736e-05} +{"train_loss": 0.016803529113531113, "grad_norm": 0.09030020236968994, "global_step": 25205, "epoch": 237, "lr": 5.383135467406366e-05} +{"train_loss": 0.013706320896744728, "grad_norm": 0.08165638148784637, "global_step": 25206, "epoch": 237, "lr": 5.3828399614681854e-05} +{"train_loss": 0.01899852231144905, "grad_norm": 0.08206748217344284, "global_step": 25207, "epoch": 237, "lr": 5.382544454184875e-05} +{"train_loss": 0.016706261783838272, "grad_norm": 0.16555075347423553, "global_step": 25208, "epoch": 237, "lr": 5.382248945557466e-05} +{"train_loss": 0.016809716820716858, "grad_norm": 0.0671880915760994, "global_step": 25209, "epoch": 237, "lr": 5.3819534355870025e-05} +{"train_loss": 0.014870080165565014, "grad_norm": 0.1263323724269867, "global_step": 25210, "epoch": 237, "lr": 5.3816579242745214e-05} +{"train_loss": 0.01449205819517374, "grad_norm": 0.12208329886198044, "global_step": 25211, "epoch": 237, "lr": 5.381362411621059e-05} +{"train_loss": 0.015742622315883636, "grad_norm": 0.11180528998374939, "global_step": 25212, "epoch": 237, "lr": 5.381066897627657e-05} +{"train_loss": 0.018125630915164948, "grad_norm": 0.06362244486808777, "global_step": 25213, "epoch": 237, "lr": 5.3807713822953487e-05} +{"train_loss": 0.018277931958436966, "grad_norm": 0.09157296270132065, "global_step": 25214, "epoch": 237, "lr": 5.3804758656251775e-05} +{"train_loss": 0.01455993764102459, "grad_norm": 0.054804086685180664, "global_step": 25215, "epoch": 237, "lr": 5.380180347618179e-05} +{"train_loss": 0.013819479383528233, "grad_norm": 0.06860218197107315, "global_step": 25216, "epoch": 237, "lr": 5.3798848282753925e-05} +{"train_loss": 0.015653828158974648, "grad_norm": 0.10797330737113953, "global_step": 25217, "epoch": 237, "lr": 5.379589307597854e-05} +{"train_loss": 0.015398059040307999, "grad_norm": 0.0871758759021759, "global_step": 25218, "epoch": 237, "lr": 5.379293785586606e-05} +{"train_loss": 0.015781652182340622, "grad_norm": 0.05661555752158165, "global_step": 25219, "epoch": 237, "lr": 5.3789982622426836e-05} +{"train_loss": 0.017331067472696304, "grad_norm": 0.07831897586584091, "global_step": 25220, "epoch": 237, "lr": 5.378702737567125e-05} +{"train_loss": 0.014413917437195778, "grad_norm": 0.07619587332010269, "global_step": 25221, "epoch": 237, "lr": 5.378407211560973e-05} +{"train_loss": 0.01468418538570404, "grad_norm": 0.12302497774362564, "global_step": 25222, "epoch": 237, "lr": 5.37811168422526e-05} +{"train_loss": 0.01631559431552887, "grad_norm": 0.05988095700740814, "global_step": 25223, "epoch": 237, "lr": 5.3778161555610265e-05} +{"train_loss": 0.014170186594128609, "grad_norm": 0.06034615635871887, "global_step": 25224, "epoch": 237, "lr": 5.3775206255693134e-05} +{"train_loss": 0.016231417655944824, "grad_norm": 0.08252927660942078, "global_step": 25225, "epoch": 237, "lr": 5.377225094251155e-05} +{"train_loss": 0.015072640031576157, "grad_norm": 0.04958001896739006, "global_step": 25226, "epoch": 237, "lr": 5.376929561607593e-05} +{"train_loss": 0.015694756473783614, "grad_norm": 0.07613502442836761, "global_step": 25227, "epoch": 237, "lr": 5.3766340276396646e-05, "val_loss": 0.019233819097280502} +{"train_loss": 0.014378765597939491, "grad_norm": 0.06263439357280731, "global_step": 25228, "epoch": 238, "lr": 5.376338492348407e-05} +{"train_loss": 0.01631723716855049, "grad_norm": 0.08270128071308136, "global_step": 25229, "epoch": 238, "lr": 5.37604295573486e-05} +{"train_loss": 0.014895268715918064, "grad_norm": 0.0736287385225296, "global_step": 25230, "epoch": 238, "lr": 5.3757474178000623e-05} +{"train_loss": 0.017795220017433167, "grad_norm": 0.09555274248123169, "global_step": 25231, "epoch": 238, "lr": 5.3754518785450505e-05} +{"train_loss": 0.016687750816345215, "grad_norm": 0.06313120573759079, "global_step": 25232, "epoch": 238, "lr": 5.375156337970865e-05} +{"train_loss": 0.014025854878127575, "grad_norm": 0.12022696435451508, "global_step": 25233, "epoch": 238, "lr": 5.374860796078542e-05} +{"train_loss": 0.016348203644156456, "grad_norm": 0.0956331342458725, "global_step": 25234, "epoch": 238, "lr": 5.3745652528691225e-05} +{"train_loss": 0.014455657452344894, "grad_norm": 0.06011439114809036, "global_step": 25235, "epoch": 238, "lr": 5.374269708343642e-05} +{"train_loss": 0.016843456774950027, "grad_norm": 0.06163947284221649, "global_step": 25236, "epoch": 238, "lr": 5.3739741625031435e-05} +{"train_loss": 0.014099175110459328, "grad_norm": 0.04916617274284363, "global_step": 25237, "epoch": 238, "lr": 5.37367861534866e-05} +{"train_loss": 0.015981897711753845, "grad_norm": 0.1270580142736435, "global_step": 25238, "epoch": 238, "lr": 5.373383066881232e-05} +{"train_loss": 0.015682339668273926, "grad_norm": 0.061339493840932846, "global_step": 25239, "epoch": 238, "lr": 5.373087517101899e-05} +{"train_loss": 0.014476814307272434, "grad_norm": 0.06814314424991608, "global_step": 25240, "epoch": 238, "lr": 5.3727919660117e-05} +{"train_loss": 0.015781991183757782, "grad_norm": 0.05511023476719856, "global_step": 25241, "epoch": 238, "lr": 5.3724964136116703e-05} +{"train_loss": 0.016506031155586243, "grad_norm": 0.11041633784770966, "global_step": 25242, "epoch": 238, "lr": 5.372200859902852e-05} +{"train_loss": 0.01349632441997528, "grad_norm": 0.06418117880821228, "global_step": 25243, "epoch": 238, "lr": 5.37190530488628e-05} +{"train_loss": 0.01657385379076004, "grad_norm": 0.05946116894483566, "global_step": 25244, "epoch": 238, "lr": 5.371609748562996e-05} +{"train_loss": 0.015226636081933975, "grad_norm": 0.14803658425807953, "global_step": 25245, "epoch": 238, "lr": 5.371314190934036e-05} +{"train_loss": 0.01563408225774765, "grad_norm": 0.08468759804964066, "global_step": 25246, "epoch": 238, "lr": 5.3710186320004393e-05} +{"train_loss": 0.019083861261606216, "grad_norm": 0.05864644795656204, "global_step": 25247, "epoch": 238, "lr": 5.370723071763244e-05} +{"train_loss": 0.015216147527098656, "grad_norm": 0.0923013836145401, "global_step": 25248, "epoch": 238, "lr": 5.37042751022349e-05} +{"train_loss": 0.014756988734006882, "grad_norm": 0.08351493626832962, "global_step": 25249, "epoch": 238, "lr": 5.3701319473822146e-05} +{"train_loss": 0.01617627590894699, "grad_norm": 0.07564585655927658, "global_step": 25250, "epoch": 238, "lr": 5.3698363832404566e-05} +{"train_loss": 0.016735874116420746, "grad_norm": 0.11633318662643433, "global_step": 25251, "epoch": 238, "lr": 5.369540817799256e-05} +{"train_loss": 0.016360731795430183, "grad_norm": 0.15404005348682404, "global_step": 25252, "epoch": 238, "lr": 5.369245251059646e-05} +{"train_loss": 0.013979262672364712, "grad_norm": 0.09471624344587326, "global_step": 25253, "epoch": 238, "lr": 5.3689496830226706e-05} +{"train_loss": 0.014572864398360252, "grad_norm": 0.05444871634244919, "global_step": 25254, "epoch": 238, "lr": 5.368654113689368e-05} +{"train_loss": 0.01667500101029873, "grad_norm": 0.17580726742744446, "global_step": 25255, "epoch": 238, "lr": 5.368358543060773e-05} +{"train_loss": 0.017368778586387634, "grad_norm": 0.09024208039045334, "global_step": 25256, "epoch": 238, "lr": 5.3680629711379274e-05} +{"train_loss": 0.015652090311050415, "grad_norm": 0.08317223936319351, "global_step": 25257, "epoch": 238, "lr": 5.367767397921868e-05} +{"train_loss": 0.014518478885293007, "grad_norm": 0.09379688650369644, "global_step": 25258, "epoch": 238, "lr": 5.3674718234136345e-05} +{"train_loss": 0.014792936854064465, "grad_norm": 0.08155845105648041, "global_step": 25259, "epoch": 238, "lr": 5.367176247614263e-05} +{"train_loss": 0.015046700835227966, "grad_norm": 0.061754949390888214, "global_step": 25260, "epoch": 238, "lr": 5.366880670524797e-05} +{"train_loss": 0.016160592436790466, "grad_norm": 0.06672953814268112, "global_step": 25261, "epoch": 238, "lr": 5.366585092146269e-05} +{"train_loss": 0.015951508656144142, "grad_norm": 0.1633169949054718, "global_step": 25262, "epoch": 238, "lr": 5.366289512479723e-05} +{"train_loss": 0.013884272426366806, "grad_norm": 0.13188622891902924, "global_step": 25263, "epoch": 238, "lr": 5.3659939315261934e-05} +{"train_loss": 0.014799533411860466, "grad_norm": 0.07525257021188736, "global_step": 25264, "epoch": 238, "lr": 5.36569834928672e-05} +{"train_loss": 0.01597270742058754, "grad_norm": 0.1217484325170517, "global_step": 25265, "epoch": 238, "lr": 5.365402765762343e-05} +{"train_loss": 0.016135316342115402, "grad_norm": 0.09287328273057938, "global_step": 25266, "epoch": 238, "lr": 5.365107180954098e-05} +{"train_loss": 0.01563628762960434, "grad_norm": 0.14213122427463531, "global_step": 25267, "epoch": 238, "lr": 5.364811594863025e-05} +{"train_loss": 0.016592467203736305, "grad_norm": 0.059828020632267, "global_step": 25268, "epoch": 238, "lr": 5.3645160074901636e-05} +{"train_loss": 0.013664813712239265, "grad_norm": 0.05652882903814316, "global_step": 25269, "epoch": 238, "lr": 5.364220418836552e-05} +{"train_loss": 0.01673416793346405, "grad_norm": 0.10000153630971909, "global_step": 25270, "epoch": 238, "lr": 5.363924828903227e-05} +{"train_loss": 0.017111964523792267, "grad_norm": 0.08503206819295883, "global_step": 25271, "epoch": 238, "lr": 5.363629237691229e-05} +{"train_loss": 0.01659569889307022, "grad_norm": 0.06919599324464798, "global_step": 25272, "epoch": 238, "lr": 5.3633336452015967e-05} +{"train_loss": 0.01743996888399124, "grad_norm": 0.0659404918551445, "global_step": 25273, "epoch": 238, "lr": 5.3630380514353675e-05} +{"train_loss": 0.015633201226592064, "grad_norm": 0.08807412534952164, "global_step": 25274, "epoch": 238, "lr": 5.36274245639358e-05} +{"train_loss": 0.01786598190665245, "grad_norm": 0.0652291476726532, "global_step": 25275, "epoch": 238, "lr": 5.3624468600772735e-05} +{"train_loss": 0.016576193273067474, "grad_norm": 0.07854143530130386, "global_step": 25276, "epoch": 238, "lr": 5.362151262487486e-05} +{"train_loss": 0.015487591736018658, "grad_norm": 0.08317255973815918, "global_step": 25277, "epoch": 238, "lr": 5.361855663625258e-05} +{"train_loss": 0.014334773644804955, "grad_norm": 0.08119288086891174, "global_step": 25278, "epoch": 238, "lr": 5.361560063491624e-05} +{"train_loss": 0.016940053552389145, "grad_norm": 0.0895688608288765, "global_step": 25279, "epoch": 238, "lr": 5.361264462087627e-05} +{"train_loss": 0.015051502734422684, "grad_norm": 0.11796198040246964, "global_step": 25280, "epoch": 238, "lr": 5.360968859414305e-05} +{"train_loss": 0.015207081101834774, "grad_norm": 0.19809621572494507, "global_step": 25281, "epoch": 238, "lr": 5.360673255472692e-05} +{"train_loss": 0.01759793981909752, "grad_norm": 0.10578976571559906, "global_step": 25282, "epoch": 238, "lr": 5.360377650263833e-05} +{"train_loss": 0.01819555275142193, "grad_norm": 0.12422042340040207, "global_step": 25283, "epoch": 238, "lr": 5.360082043788762e-05} +{"train_loss": 0.014675239101052284, "grad_norm": 0.07087268680334091, "global_step": 25284, "epoch": 238, "lr": 5.359786436048519e-05} +{"train_loss": 0.014781080186367035, "grad_norm": 0.06651928275823593, "global_step": 25285, "epoch": 238, "lr": 5.3594908270441444e-05} +{"train_loss": 0.01483086310327053, "grad_norm": 0.08770644664764404, "global_step": 25286, "epoch": 238, "lr": 5.359195216776675e-05} +{"train_loss": 0.01739661768078804, "grad_norm": 0.09915974736213684, "global_step": 25287, "epoch": 238, "lr": 5.3588996052471496e-05} +{"train_loss": 0.0160091370344162, "grad_norm": 0.08083923161029816, "global_step": 25288, "epoch": 238, "lr": 5.358603992456606e-05} +{"train_loss": 0.01696551777422428, "grad_norm": 0.15612377226352692, "global_step": 25289, "epoch": 238, "lr": 5.3583083784060864e-05} +{"train_loss": 0.014553086832165718, "grad_norm": 0.086527518928051, "global_step": 25290, "epoch": 238, "lr": 5.358012763096624e-05} +{"train_loss": 0.012249762192368507, "grad_norm": 0.13511306047439575, "global_step": 25291, "epoch": 238, "lr": 5.357717146529262e-05} +{"train_loss": 0.017422553151845932, "grad_norm": 0.09378225356340408, "global_step": 25292, "epoch": 238, "lr": 5.3574215287050386e-05} +{"train_loss": 0.017987513914704323, "grad_norm": 0.08429889380931854, "global_step": 25293, "epoch": 238, "lr": 5.357125909624989e-05} +{"train_loss": 0.01635531708598137, "grad_norm": 0.10786400735378265, "global_step": 25294, "epoch": 238, "lr": 5.3568302892901556e-05} +{"train_loss": 0.013736343011260033, "grad_norm": 0.07255319505929947, "global_step": 25295, "epoch": 238, "lr": 5.356534667701576e-05} +{"train_loss": 0.017555953934788704, "grad_norm": 0.08019925653934479, "global_step": 25296, "epoch": 238, "lr": 5.3562390448602886e-05} +{"train_loss": 0.015433233231306076, "grad_norm": 0.05863672122359276, "global_step": 25297, "epoch": 238, "lr": 5.355943420767332e-05} +{"train_loss": 0.014307338744401932, "grad_norm": 0.08457516878843307, "global_step": 25298, "epoch": 238, "lr": 5.355647795423745e-05} +{"train_loss": 0.017763514071702957, "grad_norm": 0.10851845145225525, "global_step": 25299, "epoch": 238, "lr": 5.355352168830565e-05} +{"train_loss": 0.012755265459418297, "grad_norm": 0.06285303831100464, "global_step": 25300, "epoch": 238, "lr": 5.355056540988833e-05} +{"train_loss": 0.015477489680051804, "grad_norm": 0.0922471433877945, "global_step": 25301, "epoch": 238, "lr": 5.354760911899588e-05} +{"train_loss": 0.01677703484892845, "grad_norm": 0.12206096947193146, "global_step": 25302, "epoch": 238, "lr": 5.3544652815638665e-05} +{"train_loss": 0.016054116189479828, "grad_norm": 0.04639160633087158, "global_step": 25303, "epoch": 238, "lr": 5.3541696499827066e-05} +{"train_loss": 0.013742774724960327, "grad_norm": 0.08529473096132278, "global_step": 25304, "epoch": 238, "lr": 5.353874017157151e-05} +{"train_loss": 0.01357576809823513, "grad_norm": 0.05920341983437538, "global_step": 25305, "epoch": 238, "lr": 5.353578383088235e-05} +{"train_loss": 0.014477222226560116, "grad_norm": 0.08462854474782944, "global_step": 25306, "epoch": 238, "lr": 5.3532827477769975e-05} +{"train_loss": 0.014674857258796692, "grad_norm": 0.09997956454753876, "global_step": 25307, "epoch": 238, "lr": 5.352987111224479e-05} +{"train_loss": 0.01577853597700596, "grad_norm": 0.2048804759979248, "global_step": 25308, "epoch": 238, "lr": 5.352691473431717e-05} +{"train_loss": 0.014511422254145145, "grad_norm": 0.15706923604011536, "global_step": 25309, "epoch": 238, "lr": 5.3523958343997505e-05} +{"train_loss": 0.017561763525009155, "grad_norm": 0.06868434697389603, "global_step": 25310, "epoch": 238, "lr": 5.352100194129619e-05} +{"train_loss": 0.013314180076122284, "grad_norm": 0.057152800261974335, "global_step": 25311, "epoch": 238, "lr": 5.3518045526223604e-05} +{"train_loss": 0.015299658291041851, "grad_norm": 0.065645232796669, "global_step": 25312, "epoch": 238, "lr": 5.3515089098790125e-05} +{"train_loss": 0.01699129119515419, "grad_norm": 0.15605492889881134, "global_step": 25313, "epoch": 238, "lr": 5.351213265900617e-05} +{"train_loss": 0.014814024791121483, "grad_norm": 0.06336545944213867, "global_step": 25314, "epoch": 238, "lr": 5.350917620688209e-05} +{"train_loss": 0.016481878235936165, "grad_norm": 0.17883948981761932, "global_step": 25315, "epoch": 238, "lr": 5.35062197424283e-05} +{"train_loss": 0.014395024627447128, "grad_norm": 0.1471058428287506, "global_step": 25316, "epoch": 238, "lr": 5.350326326565518e-05} +{"train_loss": 0.017704587429761887, "grad_norm": 0.07139752805233002, "global_step": 25317, "epoch": 238, "lr": 5.3500306776573104e-05} +{"train_loss": 0.017316671088337898, "grad_norm": 0.08190994709730148, "global_step": 25318, "epoch": 238, "lr": 5.349735027519249e-05} +{"train_loss": 0.013322667218744755, "grad_norm": 0.05705781280994415, "global_step": 25319, "epoch": 238, "lr": 5.3494393761523685e-05} +{"train_loss": 0.013654506765305996, "grad_norm": 0.11680469661951065, "global_step": 25320, "epoch": 238, "lr": 5.3491437235577124e-05} +{"train_loss": 0.01618529111146927, "grad_norm": 0.18641044199466705, "global_step": 25321, "epoch": 238, "lr": 5.348848069736316e-05} +{"train_loss": 0.016547124832868576, "grad_norm": 0.07734829932451248, "global_step": 25322, "epoch": 238, "lr": 5.348552414689218e-05} +{"train_loss": 0.014979586936533451, "grad_norm": 0.06679294258356094, "global_step": 25323, "epoch": 238, "lr": 5.34825675841746e-05} +{"train_loss": 0.0161746796220541, "grad_norm": 0.14696408808231354, "global_step": 25324, "epoch": 238, "lr": 5.34796110092208e-05} +{"train_loss": 0.015385475941002369, "grad_norm": 0.10643605142831802, "global_step": 25325, "epoch": 238, "lr": 5.347665442204115e-05} +{"train_loss": 0.015447656624019146, "grad_norm": 0.07784000039100647, "global_step": 25326, "epoch": 238, "lr": 5.347369782264605e-05} +{"train_loss": 0.014905577525496483, "grad_norm": 0.08061294257640839, "global_step": 25327, "epoch": 238, "lr": 5.3470741211045884e-05} +{"train_loss": 0.0170893594622612, "grad_norm": 0.07686441391706467, "global_step": 25328, "epoch": 238, "lr": 5.3467784587251045e-05} +{"train_loss": 0.015515941195189953, "grad_norm": 0.06868129968643188, "global_step": 25329, "epoch": 238, "lr": 5.34648279512719e-05} +{"train_loss": 0.016775771975517273, "grad_norm": 0.059806060045957565, "global_step": 25330, "epoch": 238, "lr": 5.346187130311888e-05} +{"train_loss": 0.016282623633742332, "grad_norm": 0.06784054636955261, "global_step": 25331, "epoch": 238, "lr": 5.345891464280235e-05} +{"train_loss": 0.014711840078234673, "grad_norm": 0.06804288178682327, "global_step": 25332, "epoch": 238, "lr": 5.345595797033268e-05} +{"train_loss": 0.015660028676238825, "grad_norm": 0.07325094938278198, "global_step": 25333, "epoch": 238, "lr": 5.345300128572031e-05, "val_loss": 0.019223839044570923} +{"train_loss": 0.016968851909041405, "grad_norm": 0.08322575688362122, "global_step": 25334, "epoch": 239, "lr": 5.3450044588975556e-05} +{"train_loss": 0.015082432888448238, "grad_norm": 0.10712337493896484, "global_step": 25335, "epoch": 239, "lr": 5.344708788010886e-05} +{"train_loss": 0.015485192649066448, "grad_norm": 0.07755093276500702, "global_step": 25336, "epoch": 239, "lr": 5.3444131159130606e-05} +{"train_loss": 0.014148430898785591, "grad_norm": 0.06031225249171257, "global_step": 25337, "epoch": 239, "lr": 5.3441174426051164e-05} +{"train_loss": 0.012527496553957462, "grad_norm": 0.08793346583843231, "global_step": 25338, "epoch": 239, "lr": 5.343821768088093e-05} +{"train_loss": 0.013026420958340168, "grad_norm": 0.07485390454530716, "global_step": 25339, "epoch": 239, "lr": 5.34352609236303e-05} +{"train_loss": 0.012968303635716438, "grad_norm": 0.08502355962991714, "global_step": 25340, "epoch": 239, "lr": 5.3432304154309655e-05} +{"train_loss": 0.01891564391553402, "grad_norm": 0.06640677154064178, "global_step": 25341, "epoch": 239, "lr": 5.3429347372929385e-05} +{"train_loss": 0.01406127866357565, "grad_norm": 0.06100067123770714, "global_step": 25342, "epoch": 239, "lr": 5.342639057949989e-05} +{"train_loss": 0.017456643283367157, "grad_norm": 0.07941629737615585, "global_step": 25343, "epoch": 239, "lr": 5.342343377403154e-05} +{"train_loss": 0.016428600996732712, "grad_norm": 0.14831840991973877, "global_step": 25344, "epoch": 239, "lr": 5.342047695653471e-05} +{"train_loss": 0.014663479290902615, "grad_norm": 0.14205706119537354, "global_step": 25345, "epoch": 239, "lr": 5.341752012701985e-05} +{"train_loss": 0.014126244932413101, "grad_norm": 0.0480838306248188, "global_step": 25346, "epoch": 239, "lr": 5.341456328549729e-05} +{"train_loss": 0.014280186966061592, "grad_norm": 0.05404123291373253, "global_step": 25347, "epoch": 239, "lr": 5.341160643197744e-05} +{"train_loss": 0.015994787216186523, "grad_norm": 0.08383593708276749, "global_step": 25348, "epoch": 239, "lr": 5.3408649566470695e-05} +{"train_loss": 0.01441151462495327, "grad_norm": 0.09817802160978317, "global_step": 25349, "epoch": 239, "lr": 5.340569268898743e-05} +{"train_loss": 0.01442628912627697, "grad_norm": 0.14498069882392883, "global_step": 25350, "epoch": 239, "lr": 5.3402735799538053e-05} +{"train_loss": 0.014443406835198402, "grad_norm": 0.07644272595643997, "global_step": 25351, "epoch": 239, "lr": 5.339977889813294e-05} +{"train_loss": 0.01473926194012165, "grad_norm": 0.10061989724636078, "global_step": 25352, "epoch": 239, "lr": 5.3396821984782474e-05} +{"train_loss": 0.01782451942563057, "grad_norm": 0.11022539436817169, "global_step": 25353, "epoch": 239, "lr": 5.339386505949706e-05} +{"train_loss": 0.014140845276415348, "grad_norm": 0.05698394402861595, "global_step": 25354, "epoch": 239, "lr": 5.3390908122287074e-05} +{"train_loss": 0.015132686123251915, "grad_norm": 0.06872676312923431, "global_step": 25355, "epoch": 239, "lr": 5.338795117316292e-05} +{"train_loss": 0.015246976166963577, "grad_norm": 0.07348663359880447, "global_step": 25356, "epoch": 239, "lr": 5.3384994212134965e-05} +{"train_loss": 0.016172898933291435, "grad_norm": 0.10859203338623047, "global_step": 25357, "epoch": 239, "lr": 5.338203723921364e-05} +{"train_loss": 0.01587752252817154, "grad_norm": 0.11751396954059601, "global_step": 25358, "epoch": 239, "lr": 5.3379080254409263e-05} +{"train_loss": 0.014197206124663353, "grad_norm": 0.13233068585395813, "global_step": 25359, "epoch": 239, "lr": 5.3376123257732293e-05} +{"train_loss": 0.015086355619132519, "grad_norm": 0.054325398057699203, "global_step": 25360, "epoch": 239, "lr": 5.337316624919311e-05} +{"train_loss": 0.01427009142935276, "grad_norm": 0.08297012746334076, "global_step": 25361, "epoch": 239, "lr": 5.337020922880206e-05} +{"train_loss": 0.015053210780024529, "grad_norm": 0.07732072472572327, "global_step": 25362, "epoch": 239, "lr": 5.336725219656958e-05} +{"train_loss": 0.018314406275749207, "grad_norm": 0.1176048144698143, "global_step": 25363, "epoch": 239, "lr": 5.3364295152506025e-05} +{"train_loss": 0.011878829449415207, "grad_norm": 0.07335899770259857, "global_step": 25364, "epoch": 239, "lr": 5.3361338096621805e-05} +{"train_loss": 0.014422863721847534, "grad_norm": 0.05519339069724083, "global_step": 25365, "epoch": 239, "lr": 5.335838102892731e-05} +{"train_loss": 0.01703265868127346, "grad_norm": 0.07357073575258255, "global_step": 25366, "epoch": 239, "lr": 5.3355423949432926e-05} +{"train_loss": 0.019007287919521332, "grad_norm": 0.09781895577907562, "global_step": 25367, "epoch": 239, "lr": 5.335246685814903e-05} +{"train_loss": 0.013904593884944916, "grad_norm": 0.09623205661773682, "global_step": 25368, "epoch": 239, "lr": 5.3349509755086036e-05} +{"train_loss": 0.013541718944907188, "grad_norm": 0.08357712626457214, "global_step": 25369, "epoch": 239, "lr": 5.334655264025432e-05} +{"train_loss": 0.0145967286080122, "grad_norm": 0.16546505689620972, "global_step": 25370, "epoch": 239, "lr": 5.334359551366426e-05} +{"train_loss": 0.016399256885051727, "grad_norm": 0.10343209654092789, "global_step": 25371, "epoch": 239, "lr": 5.334063837532628e-05} +{"train_loss": 0.018851645290851593, "grad_norm": 0.11525237560272217, "global_step": 25372, "epoch": 239, "lr": 5.333768122525074e-05} +{"train_loss": 0.017706137150526047, "grad_norm": 0.06119031459093094, "global_step": 25373, "epoch": 239, "lr": 5.333472406344803e-05} +{"train_loss": 0.015841200947761536, "grad_norm": 0.08994505554437637, "global_step": 25374, "epoch": 239, "lr": 5.3331766889928556e-05} +{"train_loss": 0.01662491075694561, "grad_norm": 0.06396379321813583, "global_step": 25375, "epoch": 239, "lr": 5.3328809704702696e-05} +{"train_loss": 0.01564224809408188, "grad_norm": 0.06444724649190903, "global_step": 25376, "epoch": 239, "lr": 5.332585250778086e-05} +{"train_loss": 0.015210434794425964, "grad_norm": 0.1396731585264206, "global_step": 25377, "epoch": 239, "lr": 5.3322895299173416e-05} +{"train_loss": 0.01759687438607216, "grad_norm": 0.11852949112653732, "global_step": 25378, "epoch": 239, "lr": 5.331993807889077e-05} +{"train_loss": 0.01459585502743721, "grad_norm": 0.07236400991678238, "global_step": 25379, "epoch": 239, "lr": 5.33169808469433e-05} +{"train_loss": 0.0157088041305542, "grad_norm": 0.06672690063714981, "global_step": 25380, "epoch": 239, "lr": 5.33140236033414e-05} +{"train_loss": 0.018244866281747818, "grad_norm": 0.1297210454940796, "global_step": 25381, "epoch": 239, "lr": 5.331106634809547e-05} +{"train_loss": 0.016811709851026535, "grad_norm": 0.12807567417621613, "global_step": 25382, "epoch": 239, "lr": 5.330810908121589e-05} +{"train_loss": 0.01632319577038288, "grad_norm": 0.1389886438846588, "global_step": 25383, "epoch": 239, "lr": 5.330515180271305e-05} +{"train_loss": 0.017010875046253204, "grad_norm": 0.06744474172592163, "global_step": 25384, "epoch": 239, "lr": 5.3302194512597346e-05} +{"train_loss": 0.015902621671557426, "grad_norm": 0.07049352675676346, "global_step": 25385, "epoch": 239, "lr": 5.329923721087916e-05} +{"train_loss": 0.01533049438148737, "grad_norm": 0.116742342710495, "global_step": 25386, "epoch": 239, "lr": 5.32962798975689e-05} +{"train_loss": 0.016871381551027298, "grad_norm": 0.08190127462148666, "global_step": 25387, "epoch": 239, "lr": 5.329332257267694e-05} +{"train_loss": 0.017140284180641174, "grad_norm": 0.10612378269433975, "global_step": 25388, "epoch": 239, "lr": 5.329036523621367e-05} +{"train_loss": 0.014458868652582169, "grad_norm": 0.08422604203224182, "global_step": 25389, "epoch": 239, "lr": 5.328740788818951e-05} +{"train_loss": 0.01428607665002346, "grad_norm": 0.058397095650434494, "global_step": 25390, "epoch": 239, "lr": 5.32844505286148e-05} +{"train_loss": 0.015800263732671738, "grad_norm": 0.0593939833343029, "global_step": 25391, "epoch": 239, "lr": 5.328149315749997e-05} +{"train_loss": 0.014536350965499878, "grad_norm": 0.08992195129394531, "global_step": 25392, "epoch": 239, "lr": 5.327853577485541e-05} +{"train_loss": 0.012443814426660538, "grad_norm": 0.08017229288816452, "global_step": 25393, "epoch": 239, "lr": 5.327557838069149e-05} +{"train_loss": 0.015228403732180595, "grad_norm": 0.0762568935751915, "global_step": 25394, "epoch": 239, "lr": 5.3272620975018626e-05} +{"train_loss": 0.017937177792191505, "grad_norm": 0.18350227177143097, "global_step": 25395, "epoch": 239, "lr": 5.326966355784718e-05} +{"train_loss": 0.014854210428893566, "grad_norm": 0.10247328877449036, "global_step": 25396, "epoch": 239, "lr": 5.3266706129187574e-05} +{"train_loss": 0.015262714587152004, "grad_norm": 0.06543554365634918, "global_step": 25397, "epoch": 239, "lr": 5.326374868905015e-05} +{"train_loss": 0.019047655165195465, "grad_norm": 0.09723979234695435, "global_step": 25398, "epoch": 239, "lr": 5.326079123744537e-05} +{"train_loss": 0.015887126326560974, "grad_norm": 0.09276290982961655, "global_step": 25399, "epoch": 239, "lr": 5.325783377438357e-05} +{"train_loss": 0.014994433149695396, "grad_norm": 0.06472491472959518, "global_step": 25400, "epoch": 239, "lr": 5.3254876299875154e-05} +{"train_loss": 0.014935890212655067, "grad_norm": 0.12035739421844482, "global_step": 25401, "epoch": 239, "lr": 5.325191881393055e-05} +{"train_loss": 0.014745140448212624, "grad_norm": 0.07605472952127457, "global_step": 25402, "epoch": 239, "lr": 5.3248961316560076e-05} +{"train_loss": 0.015427024103701115, "grad_norm": 0.13372987508773804, "global_step": 25403, "epoch": 239, "lr": 5.324600380777418e-05} +{"train_loss": 0.017114246264100075, "grad_norm": 0.104634128510952, "global_step": 25404, "epoch": 239, "lr": 5.324304628758324e-05} +{"train_loss": 0.015663743019104004, "grad_norm": 0.17009498178958893, "global_step": 25405, "epoch": 239, "lr": 5.3240088755997654e-05} +{"train_loss": 0.01603071019053459, "grad_norm": 0.07080288976430893, "global_step": 25406, "epoch": 239, "lr": 5.3237131213027794e-05} +{"train_loss": 0.013379447162151337, "grad_norm": 0.17280755937099457, "global_step": 25407, "epoch": 239, "lr": 5.323417365868407e-05} +{"train_loss": 0.015825949609279633, "grad_norm": 0.0707051157951355, "global_step": 25408, "epoch": 239, "lr": 5.3231216092976865e-05} +{"train_loss": 0.01571783609688282, "grad_norm": 0.07544814050197601, "global_step": 25409, "epoch": 239, "lr": 5.322825851591655e-05} +{"train_loss": 0.015638817101716995, "grad_norm": 0.06203893944621086, "global_step": 25410, "epoch": 239, "lr": 5.322530092751358e-05} +{"train_loss": 0.01567775569856167, "grad_norm": 0.09368347376585007, "global_step": 25411, "epoch": 239, "lr": 5.322234332777828e-05} +{"train_loss": 0.014231845736503601, "grad_norm": 0.053805552423000336, "global_step": 25412, "epoch": 239, "lr": 5.321938571672107e-05} +{"train_loss": 0.01704428531229496, "grad_norm": 0.0712401494383812, "global_step": 25413, "epoch": 239, "lr": 5.3216428094352346e-05} +{"train_loss": 0.01611006073653698, "grad_norm": 0.16025477647781372, "global_step": 25414, "epoch": 239, "lr": 5.321347046068247e-05} +{"train_loss": 0.017680056393146515, "grad_norm": 0.06297673285007477, "global_step": 25415, "epoch": 239, "lr": 5.321051281572188e-05} +{"train_loss": 0.0176087636500597, "grad_norm": 0.09348335862159729, "global_step": 25416, "epoch": 239, "lr": 5.320755515948093e-05} +{"train_loss": 0.017698824405670166, "grad_norm": 0.06545668840408325, "global_step": 25417, "epoch": 239, "lr": 5.320459749197003e-05} +{"train_loss": 0.014258909970521927, "grad_norm": 0.06025082990527153, "global_step": 25418, "epoch": 239, "lr": 5.320163981319957e-05} +{"train_loss": 0.014475082978606224, "grad_norm": 0.08303101360797882, "global_step": 25419, "epoch": 239, "lr": 5.319868212317994e-05} +{"train_loss": 0.016117315739393234, "grad_norm": 0.08963549137115479, "global_step": 25420, "epoch": 239, "lr": 5.319572442192153e-05} +{"train_loss": 0.014925380237400532, "grad_norm": 0.09500337392091751, "global_step": 25421, "epoch": 239, "lr": 5.319276670943474e-05} +{"train_loss": 0.017085226252675056, "grad_norm": 0.1704062670469284, "global_step": 25422, "epoch": 239, "lr": 5.318980898572995e-05} +{"train_loss": 0.014790482819080353, "grad_norm": 0.13599048554897308, "global_step": 25423, "epoch": 239, "lr": 5.318685125081757e-05} +{"train_loss": 0.016930224373936653, "grad_norm": 0.08630219846963882, "global_step": 25424, "epoch": 239, "lr": 5.318389350470797e-05} +{"train_loss": 0.015098245814442635, "grad_norm": 0.054568517953157425, "global_step": 25425, "epoch": 239, "lr": 5.318093574741155e-05} +{"train_loss": 0.01519689243286848, "grad_norm": 0.1414492130279541, "global_step": 25426, "epoch": 239, "lr": 5.31779779789387e-05} +{"train_loss": 0.014934324659407139, "grad_norm": 0.10952093452215195, "global_step": 25427, "epoch": 239, "lr": 5.317502019929984e-05} +{"train_loss": 0.01845366135239601, "grad_norm": 0.16456614434719086, "global_step": 25428, "epoch": 239, "lr": 5.3172062408505317e-05} +{"train_loss": 0.017222851514816284, "grad_norm": 0.07036449015140533, "global_step": 25429, "epoch": 239, "lr": 5.3169104606565555e-05} +{"train_loss": 0.018139678984880447, "grad_norm": 0.13691909611225128, "global_step": 25430, "epoch": 239, "lr": 5.3166146793490956e-05} +{"train_loss": 0.013954734429717064, "grad_norm": 0.049513667821884155, "global_step": 25431, "epoch": 239, "lr": 5.316318896929187e-05} +{"train_loss": 0.016231616958975792, "grad_norm": 0.10530465841293335, "global_step": 25432, "epoch": 239, "lr": 5.316023113397871e-05} +{"train_loss": 0.017537053674459457, "grad_norm": 0.10377463698387146, "global_step": 25433, "epoch": 239, "lr": 5.3157273287561885e-05} +{"train_loss": 0.011900760233402252, "grad_norm": 0.08302545547485352, "global_step": 25434, "epoch": 239, "lr": 5.315431543005177e-05} +{"train_loss": 0.015927942469716072, "grad_norm": 0.09577129781246185, "global_step": 25435, "epoch": 239, "lr": 5.315135756145877e-05} +{"train_loss": 0.017514023929834366, "grad_norm": 0.13733400404453278, "global_step": 25436, "epoch": 239, "lr": 5.314839968179327e-05} +{"train_loss": 0.015272201038897038, "grad_norm": 0.2763756811618805, "global_step": 25437, "epoch": 239, "lr": 5.314544179106565e-05} +{"train_loss": 0.018632300198078156, "grad_norm": 0.23478220403194427, "global_step": 25438, "epoch": 239, "lr": 5.314248388928632e-05} +{"train_loss": 0.01570298665045005, "grad_norm": 0.0959850549697876, "global_step": 25439, "epoch": 239, "lr": 5.313952597646568e-05, "val_loss": 0.01988735981285572} +{"train_loss": 0.013728784397244453, "grad_norm": 0.15591450035572052, "global_step": 25440, "epoch": 240, "lr": 5.3136568052614097e-05} +{"train_loss": 0.014799746684730053, "grad_norm": 0.07184896618127823, "global_step": 25441, "epoch": 240, "lr": 5.3133610117741974e-05} +{"train_loss": 0.017110353335738182, "grad_norm": 0.09036922454833984, "global_step": 25442, "epoch": 240, "lr": 5.313065217185973e-05} +{"train_loss": 0.014482662081718445, "grad_norm": 0.10551134496927261, "global_step": 25443, "epoch": 240, "lr": 5.312769421497772e-05} +{"train_loss": 0.01502249762415886, "grad_norm": 0.12202378362417221, "global_step": 25444, "epoch": 240, "lr": 5.312473624710635e-05} +{"train_loss": 0.014065960422158241, "grad_norm": 0.08933380246162415, "global_step": 25445, "epoch": 240, "lr": 5.3121778268256015e-05} +{"train_loss": 0.015361180528998375, "grad_norm": 0.08558626472949982, "global_step": 25446, "epoch": 240, "lr": 5.311882027843712e-05} +{"train_loss": 0.015649043023586273, "grad_norm": 0.08566922694444656, "global_step": 25447, "epoch": 240, "lr": 5.311586227766004e-05} +{"train_loss": 0.017491336911916733, "grad_norm": 0.07677825540304184, "global_step": 25448, "epoch": 240, "lr": 5.3112904265935184e-05} +{"train_loss": 0.015490083955228329, "grad_norm": 0.09507188946008682, "global_step": 25449, "epoch": 240, "lr": 5.310994624327292e-05} +{"train_loss": 0.015897899866104126, "grad_norm": 0.13964134454727173, "global_step": 25450, "epoch": 240, "lr": 5.310698820968366e-05} +{"train_loss": 0.018586397171020508, "grad_norm": 0.2221655249595642, "global_step": 25451, "epoch": 240, "lr": 5.310403016517782e-05} +{"train_loss": 0.014806699007749557, "grad_norm": 0.07367150485515594, "global_step": 25452, "epoch": 240, "lr": 5.310107210976575e-05} +{"train_loss": 0.01645713299512863, "grad_norm": 0.08674976974725723, "global_step": 25453, "epoch": 240, "lr": 5.309811404345785e-05} +{"train_loss": 0.016937069594860077, "grad_norm": 0.09892615675926208, "global_step": 25454, "epoch": 240, "lr": 5.309515596626454e-05} +{"train_loss": 0.01789885200560093, "grad_norm": 0.10892413556575775, "global_step": 25455, "epoch": 240, "lr": 5.309219787819618e-05} +{"train_loss": 0.015632735565304756, "grad_norm": 0.12899525463581085, "global_step": 25456, "epoch": 240, "lr": 5.30892397792632e-05} +{"train_loss": 0.014861281961202621, "grad_norm": 0.09688123315572739, "global_step": 25457, "epoch": 240, "lr": 5.308628166947597e-05} +{"train_loss": 0.015022091567516327, "grad_norm": 0.07029866427183151, "global_step": 25458, "epoch": 240, "lr": 5.308332354884489e-05} +{"train_loss": 0.015298392623662949, "grad_norm": 0.0827673003077507, "global_step": 25459, "epoch": 240, "lr": 5.308036541738034e-05} +{"train_loss": 0.0157832819968462, "grad_norm": 0.08431757986545563, "global_step": 25460, "epoch": 240, "lr": 5.307740727509274e-05} +{"train_loss": 0.017235660925507545, "grad_norm": 0.07231701165437698, "global_step": 25461, "epoch": 240, "lr": 5.3074449121992465e-05} +{"train_loss": 0.015926312655210495, "grad_norm": 0.08042221516370773, "global_step": 25462, "epoch": 240, "lr": 5.307149095808991e-05} +{"train_loss": 0.015471098944544792, "grad_norm": 0.10198777914047241, "global_step": 25463, "epoch": 240, "lr": 5.306853278339547e-05} +{"train_loss": 0.015486624091863632, "grad_norm": 0.09130294620990753, "global_step": 25464, "epoch": 240, "lr": 5.306557459791955e-05} +{"train_loss": 0.016163287684321404, "grad_norm": 0.09400835633277893, "global_step": 25465, "epoch": 240, "lr": 5.306261640167253e-05} +{"train_loss": 0.016612432897090912, "grad_norm": 0.1171262264251709, "global_step": 25466, "epoch": 240, "lr": 5.305965819466481e-05} +{"train_loss": 0.015683859586715698, "grad_norm": 0.054133519530296326, "global_step": 25467, "epoch": 240, "lr": 5.305669997690677e-05} +{"train_loss": 0.014722629450261593, "grad_norm": 0.12989477813243866, "global_step": 25468, "epoch": 240, "lr": 5.3053741748408824e-05} +{"train_loss": 0.01487276703119278, "grad_norm": 0.0863448902964592, "global_step": 25469, "epoch": 240, "lr": 5.3050783509181344e-05} +{"train_loss": 0.015694715082645416, "grad_norm": 0.07747256755828857, "global_step": 25470, "epoch": 240, "lr": 5.304782525923475e-05} +{"train_loss": 0.014792554080486298, "grad_norm": 0.098921038210392, "global_step": 25471, "epoch": 240, "lr": 5.304486699857943e-05} +{"train_loss": 0.012403508648276329, "grad_norm": 0.08424151688814163, "global_step": 25472, "epoch": 240, "lr": 5.304190872722576e-05} +{"train_loss": 0.016234630718827248, "grad_norm": 0.09740160405635834, "global_step": 25473, "epoch": 240, "lr": 5.3038950445184154e-05} +{"train_loss": 0.014537995681166649, "grad_norm": 0.16379143297672272, "global_step": 25474, "epoch": 240, "lr": 5.3035992152465e-05} +{"train_loss": 0.014226273633539677, "grad_norm": 0.10270722210407257, "global_step": 25475, "epoch": 240, "lr": 5.303303384907868e-05} +{"train_loss": 0.015758533030748367, "grad_norm": 0.07724060863256454, "global_step": 25476, "epoch": 240, "lr": 5.303007553503561e-05} +{"train_loss": 0.015496990643441677, "grad_norm": 0.10811885446310043, "global_step": 25477, "epoch": 240, "lr": 5.302711721034616e-05} +{"train_loss": 0.016726544126868248, "grad_norm": 0.10234297811985016, "global_step": 25478, "epoch": 240, "lr": 5.3024158875020743e-05} +{"train_loss": 0.015895264223217964, "grad_norm": 0.08077729493379593, "global_step": 25479, "epoch": 240, "lr": 5.3021200529069735e-05} +{"train_loss": 0.016506252810359, "grad_norm": 0.07734628021717072, "global_step": 25480, "epoch": 240, "lr": 5.301824217250356e-05} +{"train_loss": 0.01889132708311081, "grad_norm": 0.07436002790927887, "global_step": 25481, "epoch": 240, "lr": 5.3015283805332594e-05} +{"train_loss": 0.015589450486004353, "grad_norm": 0.09631703794002533, "global_step": 25482, "epoch": 240, "lr": 5.301232542756721e-05} +{"train_loss": 0.01667475327849388, "grad_norm": 0.08448736369609833, "global_step": 25483, "epoch": 240, "lr": 5.3009367039217846e-05} +{"train_loss": 0.014929953962564468, "grad_norm": 0.07189032435417175, "global_step": 25484, "epoch": 240, "lr": 5.300640864029486e-05} +{"train_loss": 0.01583784446120262, "grad_norm": 0.0883055329322815, "global_step": 25485, "epoch": 240, "lr": 5.300345023080867e-05} +{"train_loss": 0.01793580874800682, "grad_norm": 0.06943512707948685, "global_step": 25486, "epoch": 240, "lr": 5.300049181076967e-05} +{"train_loss": 0.01698962040245533, "grad_norm": 0.1053534597158432, "global_step": 25487, "epoch": 240, "lr": 5.299753338018824e-05} +{"train_loss": 0.013917500153183937, "grad_norm": 0.08541947603225708, "global_step": 25488, "epoch": 240, "lr": 5.299457493907477e-05} +{"train_loss": 0.01728440448641777, "grad_norm": 0.16267706453800201, "global_step": 25489, "epoch": 240, "lr": 5.299161648743968e-05} +{"train_loss": 0.01589350588619709, "grad_norm": 0.1321432739496231, "global_step": 25490, "epoch": 240, "lr": 5.298865802529335e-05} +{"train_loss": 0.01738106459379196, "grad_norm": 0.1123790591955185, "global_step": 25491, "epoch": 240, "lr": 5.298569955264617e-05} +{"train_loss": 0.011445934884250164, "grad_norm": 0.05898229777812958, "global_step": 25492, "epoch": 240, "lr": 5.298274106950854e-05} +{"train_loss": 0.016871202737092972, "grad_norm": 0.061034031212329865, "global_step": 25493, "epoch": 240, "lr": 5.297978257589086e-05} +{"train_loss": 0.01328937616199255, "grad_norm": 0.055320896208286285, "global_step": 25494, "epoch": 240, "lr": 5.297682407180351e-05} +{"train_loss": 0.01533922366797924, "grad_norm": 0.1049460768699646, "global_step": 25495, "epoch": 240, "lr": 5.2973865557256907e-05} +{"train_loss": 0.015438152477145195, "grad_norm": 0.07763606309890747, "global_step": 25496, "epoch": 240, "lr": 5.297090703226142e-05} +{"train_loss": 0.01561853289604187, "grad_norm": 0.0766831561923027, "global_step": 25497, "epoch": 240, "lr": 5.296794849682747e-05} +{"train_loss": 0.01700703427195549, "grad_norm": 0.07170093059539795, "global_step": 25498, "epoch": 240, "lr": 5.296498995096544e-05} +{"train_loss": 0.014179878868162632, "grad_norm": 0.06413944810628891, "global_step": 25499, "epoch": 240, "lr": 5.296203139468572e-05} +{"train_loss": 0.013867197558283806, "grad_norm": 0.08494674414396286, "global_step": 25500, "epoch": 240, "lr": 5.295907282799871e-05} +{"train_loss": 0.016207393258810043, "grad_norm": 0.05539007857441902, "global_step": 25501, "epoch": 240, "lr": 5.29561142509148e-05} +{"train_loss": 0.015212692320346832, "grad_norm": 0.051259011030197144, "global_step": 25502, "epoch": 240, "lr": 5.29531556634444e-05} +{"train_loss": 0.014164678752422333, "grad_norm": 0.09412695467472076, "global_step": 25503, "epoch": 240, "lr": 5.2950197065597896e-05} +{"train_loss": 0.014561808668076992, "grad_norm": 0.05740401893854141, "global_step": 25504, "epoch": 240, "lr": 5.294723845738567e-05} +{"train_loss": 0.015261917375028133, "grad_norm": 0.07893650233745575, "global_step": 25505, "epoch": 240, "lr": 5.294427983881815e-05} +{"train_loss": 0.015188094228506088, "grad_norm": 0.08181730657815933, "global_step": 25506, "epoch": 240, "lr": 5.294132120990568e-05} +{"train_loss": 0.01723070815205574, "grad_norm": 0.16734135150909424, "global_step": 25507, "epoch": 240, "lr": 5.293836257065872e-05} +{"train_loss": 0.014958636835217476, "grad_norm": 0.11782942712306976, "global_step": 25508, "epoch": 240, "lr": 5.29354039210876e-05} +{"train_loss": 0.018692508339881897, "grad_norm": 0.21675314009189606, "global_step": 25509, "epoch": 240, "lr": 5.293244526120277e-05} +{"train_loss": 0.014385546557605267, "grad_norm": 0.08150757104158401, "global_step": 25510, "epoch": 240, "lr": 5.292948659101461e-05} +{"train_loss": 0.014678971841931343, "grad_norm": 0.07623536884784698, "global_step": 25511, "epoch": 240, "lr": 5.292652791053347e-05} +{"train_loss": 0.018772918730974197, "grad_norm": 0.08796267956495285, "global_step": 25512, "epoch": 240, "lr": 5.292356921976982e-05} +{"train_loss": 0.014174976386129856, "grad_norm": 0.05369216948747635, "global_step": 25513, "epoch": 240, "lr": 5.2920610518734005e-05} +{"train_loss": 0.014352383092045784, "grad_norm": 0.0693700835108757, "global_step": 25514, "epoch": 240, "lr": 5.291765180743644e-05} +{"train_loss": 0.014451557770371437, "grad_norm": 0.05459294095635414, "global_step": 25515, "epoch": 240, "lr": 5.291469308588751e-05} +{"train_loss": 0.015184732154011726, "grad_norm": 0.07370409369468689, "global_step": 25516, "epoch": 240, "lr": 5.291173435409762e-05} +{"train_loss": 0.016295097768306732, "grad_norm": 0.12358757108449936, "global_step": 25517, "epoch": 240, "lr": 5.2908775612077164e-05} +{"train_loss": 0.01640951633453369, "grad_norm": 0.08658195286989212, "global_step": 25518, "epoch": 240, "lr": 5.290581685983653e-05} +{"train_loss": 0.015470964834094048, "grad_norm": 0.11808601021766663, "global_step": 25519, "epoch": 240, "lr": 5.290285809738612e-05} +{"train_loss": 0.014109764248132706, "grad_norm": 0.08632597327232361, "global_step": 25520, "epoch": 240, "lr": 5.2899899324736334e-05} +{"train_loss": 0.01678461581468582, "grad_norm": 0.13888800144195557, "global_step": 25521, "epoch": 240, "lr": 5.2896940541897556e-05} +{"train_loss": 0.017198581248521805, "grad_norm": 0.17515332996845245, "global_step": 25522, "epoch": 240, "lr": 5.289398174888019e-05} +{"train_loss": 0.016771188005805016, "grad_norm": 0.12615112960338593, "global_step": 25523, "epoch": 240, "lr": 5.2891022945694626e-05} +{"train_loss": 0.015209982171654701, "grad_norm": 0.05919010192155838, "global_step": 25524, "epoch": 240, "lr": 5.288806413235128e-05} +{"train_loss": 0.01680087298154831, "grad_norm": 0.1120060607790947, "global_step": 25525, "epoch": 240, "lr": 5.2885105308860515e-05} +{"train_loss": 0.018625222146511078, "grad_norm": 0.08030383288860321, "global_step": 25526, "epoch": 240, "lr": 5.2882146475232754e-05} +{"train_loss": 0.017932618036866188, "grad_norm": 0.08508540689945221, "global_step": 25527, "epoch": 240, "lr": 5.287918763147839e-05} +{"train_loss": 0.01570076122879982, "grad_norm": 0.07907377928495407, "global_step": 25528, "epoch": 240, "lr": 5.2876228777607806e-05} +{"train_loss": 0.01626158319413662, "grad_norm": 0.0524798147380352, "global_step": 25529, "epoch": 240, "lr": 5.287326991363141e-05} +{"train_loss": 0.017182419076561928, "grad_norm": 0.06228945031762123, "global_step": 25530, "epoch": 240, "lr": 5.287031103955958e-05} +{"train_loss": 0.013396739959716797, "grad_norm": 0.05635051429271698, "global_step": 25531, "epoch": 240, "lr": 5.286735215540274e-05} +{"train_loss": 0.013574045151472092, "grad_norm": 0.08521870523691177, "global_step": 25532, "epoch": 240, "lr": 5.286439326117126e-05} +{"train_loss": 0.01545268390327692, "grad_norm": 0.10220135003328323, "global_step": 25533, "epoch": 240, "lr": 5.286143435687555e-05} +{"train_loss": 0.015507975593209267, "grad_norm": 0.08084000647068024, "global_step": 25534, "epoch": 240, "lr": 5.2858475442526014e-05} +{"train_loss": 0.016654133796691895, "grad_norm": 0.09924208372831345, "global_step": 25535, "epoch": 240, "lr": 5.285551651813302e-05} +{"train_loss": 0.013406503945589066, "grad_norm": 0.059052467346191406, "global_step": 25536, "epoch": 240, "lr": 5.2852557583707e-05} +{"train_loss": 0.013823479413986206, "grad_norm": 0.06823112070560455, "global_step": 25537, "epoch": 240, "lr": 5.284959863925831e-05} +{"train_loss": 0.016015879809856415, "grad_norm": 0.1364496648311615, "global_step": 25538, "epoch": 240, "lr": 5.2846639684797396e-05} +{"train_loss": 0.01597791165113449, "grad_norm": 0.07110951095819473, "global_step": 25539, "epoch": 240, "lr": 5.284368072033463e-05} +{"train_loss": 0.017538167536258698, "grad_norm": 0.12999095022678375, "global_step": 25540, "epoch": 240, "lr": 5.284072174588038e-05} +{"train_loss": 0.014520712196826935, "grad_norm": 0.07476259022951126, "global_step": 25541, "epoch": 240, "lr": 5.2837762761445086e-05} +{"train_loss": 0.015510272234678268, "grad_norm": 0.08462607860565186, "global_step": 25542, "epoch": 240, "lr": 5.2834803767039135e-05} +{"train_loss": 0.013987571001052856, "grad_norm": 0.09520208090543747, "global_step": 25543, "epoch": 240, "lr": 5.28318447626729e-05} +{"train_loss": 0.016662495210766792, "grad_norm": 0.06158110126852989, "global_step": 25544, "epoch": 240, "lr": 5.2828885748356805e-05} +{"train_loss": 0.01568047131218438, "grad_norm": 0.18254618346691132, "global_step": 25545, "epoch": 240, "lr": 5.2825926724101236e-05, "train_batch_mse_error": 0.0140411127358675, "val_loss": 0.019300339743494987} +{"train_loss": 0.016424886882305145, "grad_norm": 0.07084614038467407, "global_step": 25546, "epoch": 241, "lr": 5.282296768991658e-05} +{"train_loss": 0.0171499140560627, "grad_norm": 0.10838650166988373, "global_step": 25547, "epoch": 241, "lr": 5.2820008645813245e-05} +{"train_loss": 0.016097698360681534, "grad_norm": 0.1411072462797165, "global_step": 25548, "epoch": 241, "lr": 5.281704959180164e-05} +{"train_loss": 0.016303775832057, "grad_norm": 0.08208154886960983, "global_step": 25549, "epoch": 241, "lr": 5.281409052789212e-05} +{"train_loss": 0.018109815195202827, "grad_norm": 0.08333863317966461, "global_step": 25550, "epoch": 241, "lr": 5.2811131454095134e-05} +{"train_loss": 0.015023458749055862, "grad_norm": 0.09898381680250168, "global_step": 25551, "epoch": 241, "lr": 5.280817237042106e-05} +{"train_loss": 0.012724881991744041, "grad_norm": 0.06017206609249115, "global_step": 25552, "epoch": 241, "lr": 5.2805213276880274e-05} +{"train_loss": 0.01452246680855751, "grad_norm": 0.10774501413106918, "global_step": 25553, "epoch": 241, "lr": 5.280225417348319e-05} +{"train_loss": 0.017433715984225273, "grad_norm": 0.1783779114484787, "global_step": 25554, "epoch": 241, "lr": 5.2799295060240205e-05} +{"train_loss": 0.01609937846660614, "grad_norm": 0.07212509959936142, "global_step": 25555, "epoch": 241, "lr": 5.279633593716171e-05} +{"train_loss": 0.01773891970515251, "grad_norm": 0.12027228623628616, "global_step": 25556, "epoch": 241, "lr": 5.279337680425812e-05} +{"train_loss": 0.015319768339395523, "grad_norm": 0.1469055861234665, "global_step": 25557, "epoch": 241, "lr": 5.27904176615398e-05} +{"train_loss": 0.014043762348592281, "grad_norm": 0.12130016088485718, "global_step": 25558, "epoch": 241, "lr": 5.2787458509017186e-05} +{"train_loss": 0.016848254948854446, "grad_norm": 0.08677351474761963, "global_step": 25559, "epoch": 241, "lr": 5.2784499346700634e-05} +{"train_loss": 0.01457127183675766, "grad_norm": 0.08858277648687363, "global_step": 25560, "epoch": 241, "lr": 5.278154017460059e-05} +{"train_loss": 0.014393354766070843, "grad_norm": 0.06630853563547134, "global_step": 25561, "epoch": 241, "lr": 5.27785809927274e-05} +{"train_loss": 0.013676001690328121, "grad_norm": 0.17029666900634766, "global_step": 25562, "epoch": 241, "lr": 5.2775621801091494e-05} +{"train_loss": 0.018060121685266495, "grad_norm": 0.11567401140928268, "global_step": 25563, "epoch": 241, "lr": 5.277266259970326e-05} +{"train_loss": 0.01625354215502739, "grad_norm": 0.10473699122667313, "global_step": 25564, "epoch": 241, "lr": 5.276970338857309e-05} +{"train_loss": 0.0116205383092165, "grad_norm": 0.053340211510658264, "global_step": 25565, "epoch": 241, "lr": 5.276674416771139e-05} +{"train_loss": 0.015348294749855995, "grad_norm": 0.08547333627939224, "global_step": 25566, "epoch": 241, "lr": 5.276378493712855e-05} +{"train_loss": 0.013916900381445885, "grad_norm": 0.14130091667175293, "global_step": 25567, "epoch": 241, "lr": 5.2760825696834985e-05} +{"train_loss": 0.018143020570278168, "grad_norm": 0.14796793460845947, "global_step": 25568, "epoch": 241, "lr": 5.2757866446841076e-05} +{"train_loss": 0.016522644087672234, "grad_norm": 0.06495649367570877, "global_step": 25569, "epoch": 241, "lr": 5.275490718715721e-05} +{"train_loss": 0.012361228466033936, "grad_norm": 0.0642973855137825, "global_step": 25570, "epoch": 241, "lr": 5.2751947917793796e-05} +{"train_loss": 0.018228448927402496, "grad_norm": 0.09065132588148117, "global_step": 25571, "epoch": 241, "lr": 5.2748988638761254e-05} +{"train_loss": 0.018403537571430206, "grad_norm": 0.06723101437091827, "global_step": 25572, "epoch": 241, "lr": 5.274602935006995e-05} +{"train_loss": 0.016864623874425888, "grad_norm": 0.05462091416120529, "global_step": 25573, "epoch": 241, "lr": 5.274307005173029e-05} +{"train_loss": 0.01370193436741829, "grad_norm": 0.07551993429660797, "global_step": 25574, "epoch": 241, "lr": 5.274011074375268e-05} +{"train_loss": 0.014322913251817226, "grad_norm": 0.0663532018661499, "global_step": 25575, "epoch": 241, "lr": 5.2737151426147514e-05} +{"train_loss": 0.017298106104135513, "grad_norm": 0.07722382247447968, "global_step": 25576, "epoch": 241, "lr": 5.273419209892519e-05} +{"train_loss": 0.01668434590101242, "grad_norm": 0.08401291072368622, "global_step": 25577, "epoch": 241, "lr": 5.2731232762096104e-05} +{"train_loss": 0.01462618913501501, "grad_norm": 0.08191487938165665, "global_step": 25578, "epoch": 241, "lr": 5.272827341567064e-05} +{"train_loss": 0.015335471369326115, "grad_norm": 0.09606770426034927, "global_step": 25579, "epoch": 241, "lr": 5.2725314059659214e-05} +{"train_loss": 0.01642000302672386, "grad_norm": 0.06190621852874756, "global_step": 25580, "epoch": 241, "lr": 5.272235469407224e-05} +{"train_loss": 0.013439847156405449, "grad_norm": 0.10059665143489838, "global_step": 25581, "epoch": 241, "lr": 5.271939531892007e-05} +{"train_loss": 0.015911651775240898, "grad_norm": 0.10701430588960648, "global_step": 25582, "epoch": 241, "lr": 5.271643593421314e-05} +{"train_loss": 0.01573687419295311, "grad_norm": 0.061507437378168106, "global_step": 25583, "epoch": 241, "lr": 5.271347653996184e-05} +{"train_loss": 0.01645960472524166, "grad_norm": 0.0615520216524601, "global_step": 25584, "epoch": 241, "lr": 5.2710517136176555e-05} +{"train_loss": 0.015903057530522346, "grad_norm": 0.07371523976325989, "global_step": 25585, "epoch": 241, "lr": 5.270755772286769e-05} +{"train_loss": 0.015041723847389221, "grad_norm": 0.06955540180206299, "global_step": 25586, "epoch": 241, "lr": 5.270459830004565e-05} +{"train_loss": 0.01474725641310215, "grad_norm": 0.0675121396780014, "global_step": 25587, "epoch": 241, "lr": 5.270163886772083e-05} +{"train_loss": 0.015837134793400764, "grad_norm": 0.08266821503639221, "global_step": 25588, "epoch": 241, "lr": 5.269867942590361e-05} +{"train_loss": 0.0171087384223938, "grad_norm": 0.09553468972444534, "global_step": 25589, "epoch": 241, "lr": 5.2695719974604427e-05} +{"train_loss": 0.016272984445095062, "grad_norm": 0.10689225047826767, "global_step": 25590, "epoch": 241, "lr": 5.269276051383365e-05} +{"train_loss": 0.01615925133228302, "grad_norm": 0.0680173933506012, "global_step": 25591, "epoch": 241, "lr": 5.268980104360166e-05} +{"train_loss": 0.01595606654882431, "grad_norm": 0.0872679278254509, "global_step": 25592, "epoch": 241, "lr": 5.268684156391891e-05} +{"train_loss": 0.015719175338745117, "grad_norm": 0.10322600603103638, "global_step": 25593, "epoch": 241, "lr": 5.268388207479574e-05} +{"train_loss": 0.015465657226741314, "grad_norm": 0.0922946110367775, "global_step": 25594, "epoch": 241, "lr": 5.26809225762426e-05} +{"train_loss": 0.014578571543097496, "grad_norm": 0.07216960936784744, "global_step": 25595, "epoch": 241, "lr": 5.267796306826985e-05} +{"train_loss": 0.015219694003462791, "grad_norm": 0.06476747244596481, "global_step": 25596, "epoch": 241, "lr": 5.2675003550887916e-05} +{"train_loss": 0.017116950824856758, "grad_norm": 0.1261564940214157, "global_step": 25597, "epoch": 241, "lr": 5.267204402410717e-05} +{"train_loss": 0.017543673515319824, "grad_norm": 0.09094831347465515, "global_step": 25598, "epoch": 241, "lr": 5.266908448793803e-05} +{"train_loss": 0.015112856402993202, "grad_norm": 0.07417212426662445, "global_step": 25599, "epoch": 241, "lr": 5.266612494239088e-05} +{"train_loss": 0.01961374282836914, "grad_norm": 0.08539197593927383, "global_step": 25600, "epoch": 241, "lr": 5.2663165387476135e-05} +{"train_loss": 0.014821380376815796, "grad_norm": 0.09212726354598999, "global_step": 25601, "epoch": 241, "lr": 5.266020582320419e-05} +{"train_loss": 0.014410952106118202, "grad_norm": 0.10100575536489487, "global_step": 25602, "epoch": 241, "lr": 5.265724624958542e-05} +{"train_loss": 0.017254777252674103, "grad_norm": 0.0655331015586853, "global_step": 25603, "epoch": 241, "lr": 5.265428666663025e-05} +{"train_loss": 0.018264712765812874, "grad_norm": 0.13773001730442047, "global_step": 25604, "epoch": 241, "lr": 5.2651327074349085e-05} +{"train_loss": 0.012859523296356201, "grad_norm": 0.07083318382501602, "global_step": 25605, "epoch": 241, "lr": 5.264836747275229e-05} +{"train_loss": 0.014122871682047844, "grad_norm": 0.07100584357976913, "global_step": 25606, "epoch": 241, "lr": 5.264540786185029e-05} +{"train_loss": 0.015032253228127956, "grad_norm": 0.06823374330997467, "global_step": 25607, "epoch": 241, "lr": 5.264244824165347e-05} +{"train_loss": 0.016100861132144928, "grad_norm": 0.08244577050209045, "global_step": 25608, "epoch": 241, "lr": 5.263948861217225e-05} +{"train_loss": 0.013577617704868317, "grad_norm": 0.09193841367959976, "global_step": 25609, "epoch": 241, "lr": 5.263652897341701e-05} +{"train_loss": 0.015622631646692753, "grad_norm": 0.049433231353759766, "global_step": 25610, "epoch": 241, "lr": 5.2633569325398155e-05} +{"train_loss": 0.015241465531289577, "grad_norm": 0.12218185514211655, "global_step": 25611, "epoch": 241, "lr": 5.2630609668126086e-05} +{"train_loss": 0.015788696706295013, "grad_norm": 0.06899707019329071, "global_step": 25612, "epoch": 241, "lr": 5.262765000161119e-05} +{"train_loss": 0.015409993007779121, "grad_norm": 0.07458800077438354, "global_step": 25613, "epoch": 241, "lr": 5.262469032586388e-05} +{"train_loss": 0.014334583654999733, "grad_norm": 0.14580239355564117, "global_step": 25614, "epoch": 241, "lr": 5.262173064089455e-05} +{"train_loss": 0.016536526381969452, "grad_norm": 0.0608329139649868, "global_step": 25615, "epoch": 241, "lr": 5.2618770946713604e-05} +{"train_loss": 0.015322358347475529, "grad_norm": 0.1030288115143776, "global_step": 25616, "epoch": 241, "lr": 5.2615811243331426e-05} +{"train_loss": 0.013600439764559269, "grad_norm": 0.08364738523960114, "global_step": 25617, "epoch": 241, "lr": 5.2612851530758426e-05} +{"train_loss": 0.01657928340137005, "grad_norm": 0.062153104692697525, "global_step": 25618, "epoch": 241, "lr": 5.2609891809005005e-05} +{"train_loss": 0.01531553827226162, "grad_norm": 0.07927282154560089, "global_step": 25619, "epoch": 241, "lr": 5.2606932078081564e-05} +{"train_loss": 0.015337605029344559, "grad_norm": 0.054829102009534836, "global_step": 25620, "epoch": 241, "lr": 5.260397233799847e-05} +{"train_loss": 0.015855927020311356, "grad_norm": 0.0931285172700882, "global_step": 25621, "epoch": 241, "lr": 5.2601012588766184e-05} +{"train_loss": 0.015706613659858704, "grad_norm": 0.0792309120297432, "global_step": 25622, "epoch": 241, "lr": 5.259805283039505e-05} +{"train_loss": 0.014552555978298187, "grad_norm": 0.08706053346395493, "global_step": 25623, "epoch": 241, "lr": 5.25950930628955e-05} +{"train_loss": 0.014968120492994785, "grad_norm": 0.14607927203178406, "global_step": 25624, "epoch": 241, "lr": 5.259213328627792e-05} +{"train_loss": 0.0147794708609581, "grad_norm": 0.10211881250143051, "global_step": 25625, "epoch": 241, "lr": 5.258917350055271e-05} +{"train_loss": 0.017142459750175476, "grad_norm": 0.07600001990795135, "global_step": 25626, "epoch": 241, "lr": 5.258621370573027e-05} +{"train_loss": 0.014867733232676983, "grad_norm": 0.1213143989443779, "global_step": 25627, "epoch": 241, "lr": 5.2583253901820997e-05} +{"train_loss": 0.01681218296289444, "grad_norm": 0.07407982647418976, "global_step": 25628, "epoch": 241, "lr": 5.258029408883529e-05} +{"train_loss": 0.014933358877897263, "grad_norm": 0.05045786499977112, "global_step": 25629, "epoch": 241, "lr": 5.257733426678355e-05} +{"train_loss": 0.01340404711663723, "grad_norm": 0.06036248430609703, "global_step": 25630, "epoch": 241, "lr": 5.25743744356762e-05} +{"train_loss": 0.01727372407913208, "grad_norm": 0.07253903895616531, "global_step": 25631, "epoch": 241, "lr": 5.2571414595523604e-05} +{"train_loss": 0.014477442018687725, "grad_norm": 0.1451762616634369, "global_step": 25632, "epoch": 241, "lr": 5.256845474633616e-05} +{"train_loss": 0.01794269308447838, "grad_norm": 0.08509267121553421, "global_step": 25633, "epoch": 241, "lr": 5.256549488812431e-05} +{"train_loss": 0.014419058337807655, "grad_norm": 0.08579488098621368, "global_step": 25634, "epoch": 241, "lr": 5.256253502089841e-05} +{"train_loss": 0.014160290360450745, "grad_norm": 0.10483244061470032, "global_step": 25635, "epoch": 241, "lr": 5.255957514466887e-05} +{"train_loss": 0.015176782384514809, "grad_norm": 0.05973799526691437, "global_step": 25636, "epoch": 241, "lr": 5.255661525944611e-05} +{"train_loss": 0.013561734929680824, "grad_norm": 0.10472295433282852, "global_step": 25637, "epoch": 241, "lr": 5.255365536524052e-05} +{"train_loss": 0.016629450023174286, "grad_norm": 0.10500399768352509, "global_step": 25638, "epoch": 241, "lr": 5.255069546206248e-05} +{"train_loss": 0.019423115998506546, "grad_norm": 0.1505368947982788, "global_step": 25639, "epoch": 241, "lr": 5.254773554992242e-05} +{"train_loss": 0.014516409486532211, "grad_norm": 0.07160244137048721, "global_step": 25640, "epoch": 241, "lr": 5.254477562883071e-05} +{"train_loss": 0.015034377574920654, "grad_norm": 0.0734189972281456, "global_step": 25641, "epoch": 241, "lr": 5.2541815698797784e-05} +{"train_loss": 0.014869749546051025, "grad_norm": 0.07275104522705078, "global_step": 25642, "epoch": 241, "lr": 5.2538855759834005e-05} +{"train_loss": 0.015514304861426353, "grad_norm": 0.08697044104337692, "global_step": 25643, "epoch": 241, "lr": 5.25358958119498e-05} +{"train_loss": 0.01610758900642395, "grad_norm": 0.0764162465929985, "global_step": 25644, "epoch": 241, "lr": 5.2532935855155555e-05} +{"train_loss": 0.01725814864039421, "grad_norm": 0.08513373136520386, "global_step": 25645, "epoch": 241, "lr": 5.252997588946168e-05} +{"train_loss": 0.01349277701228857, "grad_norm": 0.09313491731882095, "global_step": 25646, "epoch": 241, "lr": 5.2527015914878567e-05} +{"train_loss": 0.016003839671611786, "grad_norm": 0.07403009384870529, "global_step": 25647, "epoch": 241, "lr": 5.252405593141662e-05} +{"train_loss": 0.015322688966989517, "grad_norm": 0.0669599175453186, "global_step": 25648, "epoch": 241, "lr": 5.252109593908624e-05} +{"train_loss": 0.017896078526973724, "grad_norm": 0.08847663551568985, "global_step": 25649, "epoch": 241, "lr": 5.2518135937897814e-05} +{"train_loss": 0.013607382774353027, "grad_norm": 0.07029126584529877, "global_step": 25650, "epoch": 241, "lr": 5.251517592786177e-05} +{"train_loss": 0.015627991109383555, "grad_norm": 0.16071847081184387, "global_step": 25651, "epoch": 241, "lr": 5.2512215908988484e-05, "val_loss": 0.01997688040137291} +{"train_loss": 0.014856168068945408, "grad_norm": 0.053273577243089676, "global_step": 25652, "epoch": 242, "lr": 5.250925588128837e-05} +{"train_loss": 0.015406511723995209, "grad_norm": 0.059272680431604385, "global_step": 25653, "epoch": 242, "lr": 5.250629584477181e-05} +{"train_loss": 0.014516860246658325, "grad_norm": 0.05086040869355202, "global_step": 25654, "epoch": 242, "lr": 5.250333579944923e-05} +{"train_loss": 0.015815945342183113, "grad_norm": 0.13701696693897247, "global_step": 25655, "epoch": 242, "lr": 5.250037574533101e-05} +{"train_loss": 0.014820762909948826, "grad_norm": 0.12541554868221283, "global_step": 25656, "epoch": 242, "lr": 5.249741568242754e-05} +{"train_loss": 0.014581350609660149, "grad_norm": 0.15726596117019653, "global_step": 25657, "epoch": 242, "lr": 5.249445561074927e-05} +{"train_loss": 0.015077026560902596, "grad_norm": 0.1303694099187851, "global_step": 25658, "epoch": 242, "lr": 5.2491495530306535e-05} +{"train_loss": 0.01702195033431053, "grad_norm": 0.05635468289256096, "global_step": 25659, "epoch": 242, "lr": 5.248853544110979e-05} +{"train_loss": 0.016329560428857803, "grad_norm": 0.06426255404949188, "global_step": 25660, "epoch": 242, "lr": 5.2485575343169425e-05} +{"train_loss": 0.01543545350432396, "grad_norm": 0.14745008945465088, "global_step": 25661, "epoch": 242, "lr": 5.24826152364958e-05} +{"train_loss": 0.01903800293803215, "grad_norm": 0.06194726377725601, "global_step": 25662, "epoch": 242, "lr": 5.247965512109936e-05} +{"train_loss": 0.016598280519247055, "grad_norm": 0.10260899364948273, "global_step": 25663, "epoch": 242, "lr": 5.247669499699048e-05} +{"train_loss": 0.01891256496310234, "grad_norm": 0.13612376153469086, "global_step": 25664, "epoch": 242, "lr": 5.247373486417958e-05} +{"train_loss": 0.015451193787157536, "grad_norm": 0.09845252335071564, "global_step": 25665, "epoch": 242, "lr": 5.247077472267705e-05} +{"train_loss": 0.015235574916005135, "grad_norm": 0.06238473206758499, "global_step": 25666, "epoch": 242, "lr": 5.2467814572493304e-05} +{"train_loss": 0.014091286808252335, "grad_norm": 0.13617601990699768, "global_step": 25667, "epoch": 242, "lr": 5.246485441363871e-05} +{"train_loss": 0.01744842529296875, "grad_norm": 0.09352531284093857, "global_step": 25668, "epoch": 242, "lr": 5.24618942461237e-05} +{"train_loss": 0.017240585759282112, "grad_norm": 0.06497342884540558, "global_step": 25669, "epoch": 242, "lr": 5.245893406995867e-05} +{"train_loss": 0.014134548604488373, "grad_norm": 0.145962655544281, "global_step": 25670, "epoch": 242, "lr": 5.245597388515401e-05} +{"train_loss": 0.015331422910094261, "grad_norm": 0.1405712366104126, "global_step": 25671, "epoch": 242, "lr": 5.245301369172012e-05} +{"train_loss": 0.016352113336324692, "grad_norm": 0.07525318115949631, "global_step": 25672, "epoch": 242, "lr": 5.245005348966741e-05} +{"train_loss": 0.014980928041040897, "grad_norm": 0.07062871754169464, "global_step": 25673, "epoch": 242, "lr": 5.2447093279006265e-05} +{"train_loss": 0.016554828733205795, "grad_norm": 0.06393235176801682, "global_step": 25674, "epoch": 242, "lr": 5.244413305974713e-05} +{"train_loss": 0.01615859568119049, "grad_norm": 0.08408579975366592, "global_step": 25675, "epoch": 242, "lr": 5.244117283190034e-05} +{"train_loss": 0.016691124066710472, "grad_norm": 0.11831535398960114, "global_step": 25676, "epoch": 242, "lr": 5.243821259547634e-05} +{"train_loss": 0.019053202122449875, "grad_norm": 0.12462961673736572, "global_step": 25677, "epoch": 242, "lr": 5.2435252350485555e-05} +{"train_loss": 0.014832409098744392, "grad_norm": 0.056935232132673264, "global_step": 25678, "epoch": 242, "lr": 5.243229209693831e-05} +{"train_loss": 0.015684835612773895, "grad_norm": 0.08781783282756805, "global_step": 25679, "epoch": 242, "lr": 5.2429331834845066e-05} +{"train_loss": 0.014782157726585865, "grad_norm": 0.05949367582798004, "global_step": 25680, "epoch": 242, "lr": 5.2426371564216206e-05} +{"train_loss": 0.016954699531197548, "grad_norm": 0.12269795686006546, "global_step": 25681, "epoch": 242, "lr": 5.2423411285062115e-05} +{"train_loss": 0.014872034080326557, "grad_norm": 0.10089893639087677, "global_step": 25682, "epoch": 242, "lr": 5.2420450997393236e-05} +{"train_loss": 0.015179840847849846, "grad_norm": 0.10658490657806396, "global_step": 25683, "epoch": 242, "lr": 5.241749070121993e-05} +{"train_loss": 0.01661890558898449, "grad_norm": 0.11537434905767441, "global_step": 25684, "epoch": 242, "lr": 5.2414530396552616e-05} +{"train_loss": 0.01700899563729763, "grad_norm": 0.15069954097270966, "global_step": 25685, "epoch": 242, "lr": 5.2411570083401686e-05} +{"train_loss": 0.016717346385121346, "grad_norm": 0.18838651478290558, "global_step": 25686, "epoch": 242, "lr": 5.240860976177756e-05} +{"train_loss": 0.015514916740357876, "grad_norm": 0.15704452991485596, "global_step": 25687, "epoch": 242, "lr": 5.240564943169062e-05} +{"train_loss": 0.016537155956029892, "grad_norm": 0.1093209981918335, "global_step": 25688, "epoch": 242, "lr": 5.240268909315127e-05} +{"train_loss": 0.01492872554808855, "grad_norm": 0.13941603899002075, "global_step": 25689, "epoch": 242, "lr": 5.239972874616994e-05} +{"train_loss": 0.014971176162362099, "grad_norm": 0.08097933977842331, "global_step": 25690, "epoch": 242, "lr": 5.2396768390756965e-05} +{"train_loss": 0.015211603604257107, "grad_norm": 0.15549813210964203, "global_step": 25691, "epoch": 242, "lr": 5.239380802692282e-05} +{"train_loss": 0.016240842640399933, "grad_norm": 0.0667460709810257, "global_step": 25692, "epoch": 242, "lr": 5.2390847654677865e-05} +{"train_loss": 0.014606026001274586, "grad_norm": 0.08616966009140015, "global_step": 25693, "epoch": 242, "lr": 5.238788727403251e-05} +{"train_loss": 0.01579974964261055, "grad_norm": 0.07738563418388367, "global_step": 25694, "epoch": 242, "lr": 5.2384926884997166e-05} +{"train_loss": 0.015500403009355068, "grad_norm": 0.06390642374753952, "global_step": 25695, "epoch": 242, "lr": 5.238196648758221e-05} +{"train_loss": 0.01724928617477417, "grad_norm": 0.1291557252407074, "global_step": 25696, "epoch": 242, "lr": 5.237900608179808e-05} +{"train_loss": 0.01617373526096344, "grad_norm": 0.06866530328989029, "global_step": 25697, "epoch": 242, "lr": 5.2376045667655124e-05} +{"train_loss": 0.013442902825772762, "grad_norm": 0.0671846866607666, "global_step": 25698, "epoch": 242, "lr": 5.2373085245163814e-05} +{"train_loss": 0.015551868826150894, "grad_norm": 0.067807137966156, "global_step": 25699, "epoch": 242, "lr": 5.2370124814334495e-05} +{"train_loss": 0.014980577863752842, "grad_norm": 0.08264750987291336, "global_step": 25700, "epoch": 242, "lr": 5.236716437517759e-05} +{"train_loss": 0.015725523233413696, "grad_norm": 0.0949167013168335, "global_step": 25701, "epoch": 242, "lr": 5.2364203927703515e-05} +{"train_loss": 0.013690554536879063, "grad_norm": 0.07113663852214813, "global_step": 25702, "epoch": 242, "lr": 5.2361243471922625e-05} +{"train_loss": 0.014650560915470123, "grad_norm": 0.06603531539440155, "global_step": 25703, "epoch": 242, "lr": 5.235828300784538e-05} +{"train_loss": 0.016031673178076744, "grad_norm": 0.11687663197517395, "global_step": 25704, "epoch": 242, "lr": 5.235532253548213e-05} +{"train_loss": 0.01333933137357235, "grad_norm": 0.05774097889661789, "global_step": 25705, "epoch": 242, "lr": 5.2352362054843316e-05} +{"train_loss": 0.013975128531455994, "grad_norm": 0.05610669031739235, "global_step": 25706, "epoch": 242, "lr": 5.2349401565939324e-05} +{"train_loss": 0.015705252066254616, "grad_norm": 0.06916432827711105, "global_step": 25707, "epoch": 242, "lr": 5.234644106878056e-05} +{"train_loss": 0.016812250018119812, "grad_norm": 0.09481999278068542, "global_step": 25708, "epoch": 242, "lr": 5.2343480563377425e-05} +{"train_loss": 0.013446684926748276, "grad_norm": 0.053580787032842636, "global_step": 25709, "epoch": 242, "lr": 5.2340520049740295e-05} +{"train_loss": 0.015070827677845955, "grad_norm": 0.06034071370959282, "global_step": 25710, "epoch": 242, "lr": 5.233755952787962e-05} +{"train_loss": 0.015380583703517914, "grad_norm": 0.07199772447347641, "global_step": 25711, "epoch": 242, "lr": 5.2334598997805774e-05} +{"train_loss": 0.01362697035074234, "grad_norm": 0.089541956782341, "global_step": 25712, "epoch": 242, "lr": 5.233163845952914e-05} +{"train_loss": 0.013318896293640137, "grad_norm": 0.09571754932403564, "global_step": 25713, "epoch": 242, "lr": 5.232867791306018e-05} +{"train_loss": 0.014250386506319046, "grad_norm": 0.05097179114818573, "global_step": 25714, "epoch": 242, "lr": 5.232571735840923e-05} +{"train_loss": 0.013944721780717373, "grad_norm": 0.08088909834623337, "global_step": 25715, "epoch": 242, "lr": 5.232275679558673e-05} +{"train_loss": 0.015239613130688667, "grad_norm": 0.06192591041326523, "global_step": 25716, "epoch": 242, "lr": 5.231979622460307e-05} +{"train_loss": 0.01775745116174221, "grad_norm": 0.10478121787309647, "global_step": 25717, "epoch": 242, "lr": 5.231683564546865e-05} +{"train_loss": 0.013658432289958, "grad_norm": 0.07546394318342209, "global_step": 25718, "epoch": 242, "lr": 5.231387505819391e-05} +{"train_loss": 0.013864134438335896, "grad_norm": 0.07711856067180634, "global_step": 25719, "epoch": 242, "lr": 5.2310914462789176e-05} +{"train_loss": 0.015058828517794609, "grad_norm": 0.10308864712715149, "global_step": 25720, "epoch": 242, "lr": 5.2307953859264916e-05} +{"train_loss": 0.014396274462342262, "grad_norm": 0.056909117847681046, "global_step": 25721, "epoch": 242, "lr": 5.230499324763151e-05} +{"train_loss": 0.01361842267215252, "grad_norm": 0.05692135915160179, "global_step": 25722, "epoch": 242, "lr": 5.2302032627899346e-05} +{"train_loss": 0.014964226633310318, "grad_norm": 0.05797944590449333, "global_step": 25723, "epoch": 242, "lr": 5.2299072000078854e-05} +{"train_loss": 0.014734689146280289, "grad_norm": 0.09942925721406937, "global_step": 25724, "epoch": 242, "lr": 5.229611136418041e-05} +{"train_loss": 0.016349781304597855, "grad_norm": 0.05505308881402016, "global_step": 25725, "epoch": 242, "lr": 5.229315072021445e-05} +{"train_loss": 0.016450755298137665, "grad_norm": 0.09583234041929245, "global_step": 25726, "epoch": 242, "lr": 5.2290190068191325e-05} +{"train_loss": 0.01631028577685356, "grad_norm": 0.08289099484682083, "global_step": 25727, "epoch": 242, "lr": 5.2287229408121496e-05} +{"train_loss": 0.015413484536111355, "grad_norm": 0.09861603379249573, "global_step": 25728, "epoch": 242, "lr": 5.228426874001533e-05} +{"train_loss": 0.015554392710328102, "grad_norm": 0.10605248808860779, "global_step": 25729, "epoch": 242, "lr": 5.228130806388323e-05} +{"train_loss": 0.016759736463427544, "grad_norm": 0.07594107836484909, "global_step": 25730, "epoch": 242, "lr": 5.2278347379735617e-05} +{"train_loss": 0.01606752723455429, "grad_norm": 0.12082476913928986, "global_step": 25731, "epoch": 242, "lr": 5.227538668758287e-05} +{"train_loss": 0.015993114560842514, "grad_norm": 0.0678931400179863, "global_step": 25732, "epoch": 242, "lr": 5.22724259874354e-05} +{"train_loss": 0.013745209202170372, "grad_norm": 0.09778252989053726, "global_step": 25733, "epoch": 242, "lr": 5.226946527930363e-05} +{"train_loss": 0.016463445499539375, "grad_norm": 0.09133502840995789, "global_step": 25734, "epoch": 242, "lr": 5.2266504563197935e-05} +{"train_loss": 0.015220281668007374, "grad_norm": 0.07490182667970657, "global_step": 25735, "epoch": 242, "lr": 5.226354383912874e-05} +{"train_loss": 0.016817355528473854, "grad_norm": 0.17749452590942383, "global_step": 25736, "epoch": 242, "lr": 5.2260583107106434e-05} +{"train_loss": 0.01585439406335354, "grad_norm": 0.12071049958467484, "global_step": 25737, "epoch": 242, "lr": 5.225762236714141e-05} +{"train_loss": 0.015137184411287308, "grad_norm": 0.11511452496051788, "global_step": 25738, "epoch": 242, "lr": 5.225466161924408e-05} +{"train_loss": 0.014651091769337654, "grad_norm": 0.09530948847532272, "global_step": 25739, "epoch": 242, "lr": 5.225170086342487e-05} +{"train_loss": 0.01826053112745285, "grad_norm": 0.06681656092405319, "global_step": 25740, "epoch": 242, "lr": 5.224874009969415e-05} +{"train_loss": 0.015563086606562138, "grad_norm": 0.06910424679517746, "global_step": 25741, "epoch": 242, "lr": 5.224577932806233e-05} +{"train_loss": 0.013473916798830032, "grad_norm": 0.05914558842778206, "global_step": 25742, "epoch": 242, "lr": 5.224281854853983e-05} +{"train_loss": 0.01859823241829872, "grad_norm": 0.08552997559309006, "global_step": 25743, "epoch": 242, "lr": 5.2239857761137034e-05} +{"train_loss": 0.015051409602165222, "grad_norm": 0.08807777613401413, "global_step": 25744, "epoch": 242, "lr": 5.223689696586435e-05} +{"train_loss": 0.01663297414779663, "grad_norm": 0.07732544839382172, "global_step": 25745, "epoch": 242, "lr": 5.2233936162732187e-05} +{"train_loss": 0.01608177460730076, "grad_norm": 0.11665203422307968, "global_step": 25746, "epoch": 242, "lr": 5.223097535175094e-05} +{"train_loss": 0.016036398708820343, "grad_norm": 0.06970459222793579, "global_step": 25747, "epoch": 242, "lr": 5.2228014532931027e-05} +{"train_loss": 0.017004644498229027, "grad_norm": 0.05791104584932327, "global_step": 25748, "epoch": 242, "lr": 5.222505370628283e-05} +{"train_loss": 0.016615716740489006, "grad_norm": 0.24286885559558868, "global_step": 25749, "epoch": 242, "lr": 5.2222092871816766e-05} +{"train_loss": 0.016848888248205185, "grad_norm": 0.06354067474603653, "global_step": 25750, "epoch": 242, "lr": 5.2219132029543225e-05} +{"train_loss": 0.017363641411066055, "grad_norm": 0.07119455933570862, "global_step": 25751, "epoch": 242, "lr": 5.221617117947263e-05} +{"train_loss": 0.017381146550178528, "grad_norm": 0.08860907703638077, "global_step": 25752, "epoch": 242, "lr": 5.221321032161537e-05} +{"train_loss": 0.017826087772846222, "grad_norm": 0.07218007743358612, "global_step": 25753, "epoch": 242, "lr": 5.221024945598184e-05} +{"train_loss": 0.015705537050962448, "grad_norm": 0.08976621925830841, "global_step": 25754, "epoch": 242, "lr": 5.220728858258248e-05} +{"train_loss": 0.014311217702925205, "grad_norm": 0.07074794173240662, "global_step": 25755, "epoch": 242, "lr": 5.220432770142765e-05} +{"train_loss": 0.015395080670714378, "grad_norm": 0.09850920736789703, "global_step": 25756, "epoch": 242, "lr": 5.2201366812527764e-05} +{"train_loss": 0.015638790330586006, "grad_norm": 0.11270152032375336, "global_step": 25757, "epoch": 242, "lr": 5.219840591589325e-05, "val_loss": 0.019400982186198235} +{"train_loss": 0.014928964897990227, "grad_norm": 0.09188257157802582, "global_step": 25758, "epoch": 243, "lr": 5.219544501153448e-05} +{"train_loss": 0.016129085794091225, "grad_norm": 0.10699324309825897, "global_step": 25759, "epoch": 243, "lr": 5.219248409946188e-05} +{"train_loss": 0.013255982659757137, "grad_norm": 0.1422048956155777, "global_step": 25760, "epoch": 243, "lr": 5.218952317968584e-05} +{"train_loss": 0.013262726366519928, "grad_norm": 0.08059675991535187, "global_step": 25761, "epoch": 243, "lr": 5.2186562252216774e-05} +{"train_loss": 0.016724172979593277, "grad_norm": 0.06699463725090027, "global_step": 25762, "epoch": 243, "lr": 5.2183601317065075e-05} +{"train_loss": 0.015167804434895515, "grad_norm": 0.12667997181415558, "global_step": 25763, "epoch": 243, "lr": 5.2180640374241154e-05} +{"train_loss": 0.014717932790517807, "grad_norm": 0.09809134900569916, "global_step": 25764, "epoch": 243, "lr": 5.217767942375541e-05} +{"train_loss": 0.016807232052087784, "grad_norm": 0.053641993552446365, "global_step": 25765, "epoch": 243, "lr": 5.2174718465618236e-05} +{"train_loss": 0.016790946945548058, "grad_norm": 0.13439597189426422, "global_step": 25766, "epoch": 243, "lr": 5.2171757499840054e-05} +{"train_loss": 0.016454525291919708, "grad_norm": 0.08617392182350159, "global_step": 25767, "epoch": 243, "lr": 5.2168796526431265e-05} +{"train_loss": 0.017424065619707108, "grad_norm": 0.10295908153057098, "global_step": 25768, "epoch": 243, "lr": 5.2165835545402275e-05} +{"train_loss": 0.01195558998733759, "grad_norm": 0.05723830685019493, "global_step": 25769, "epoch": 243, "lr": 5.2162874556763474e-05} +{"train_loss": 0.015425454825162888, "grad_norm": 0.06931956857442856, "global_step": 25770, "epoch": 243, "lr": 5.2159913560525256e-05} +{"train_loss": 0.017207790166139603, "grad_norm": 0.0985039621591568, "global_step": 25771, "epoch": 243, "lr": 5.215695255669807e-05} +{"train_loss": 0.013982187956571579, "grad_norm": 0.0654652863740921, "global_step": 25772, "epoch": 243, "lr": 5.215399154529226e-05} +{"train_loss": 0.016778061166405678, "grad_norm": 0.07002198696136475, "global_step": 25773, "epoch": 243, "lr": 5.215103052631828e-05} +{"train_loss": 0.014551639556884766, "grad_norm": 0.10787772387266159, "global_step": 25774, "epoch": 243, "lr": 5.214806949978651e-05} +{"train_loss": 0.012087334878742695, "grad_norm": 0.08392014354467392, "global_step": 25775, "epoch": 243, "lr": 5.214510846570736e-05} +{"train_loss": 0.01421571895480156, "grad_norm": 0.06254182755947113, "global_step": 25776, "epoch": 243, "lr": 5.214214742409123e-05} +{"train_loss": 0.012857090681791306, "grad_norm": 0.10615869611501694, "global_step": 25777, "epoch": 243, "lr": 5.2139186374948525e-05} +{"train_loss": 0.013668999075889587, "grad_norm": 0.05497147887945175, "global_step": 25778, "epoch": 243, "lr": 5.213622531828966e-05} +{"train_loss": 0.01566535234451294, "grad_norm": 0.11775858700275421, "global_step": 25779, "epoch": 243, "lr": 5.2133264254125015e-05} +{"train_loss": 0.01636650413274765, "grad_norm": 0.10268232971429825, "global_step": 25780, "epoch": 243, "lr": 5.213030318246501e-05} +{"train_loss": 0.015510546043515205, "grad_norm": 0.13006141781806946, "global_step": 25781, "epoch": 243, "lr": 5.212734210332004e-05} +{"train_loss": 0.016691656783223152, "grad_norm": 0.10669595003128052, "global_step": 25782, "epoch": 243, "lr": 5.212438101670052e-05} +{"train_loss": 0.015095449984073639, "grad_norm": 0.053744006901979446, "global_step": 25783, "epoch": 243, "lr": 5.212141992261685e-05} +{"train_loss": 0.018044058233499527, "grad_norm": 0.10487275570631027, "global_step": 25784, "epoch": 243, "lr": 5.211845882107943e-05} +{"train_loss": 0.015983816236257553, "grad_norm": 0.10257094353437424, "global_step": 25785, "epoch": 243, "lr": 5.211549771209867e-05} +{"train_loss": 0.015137601643800735, "grad_norm": 0.09721456468105316, "global_step": 25786, "epoch": 243, "lr": 5.211253659568497e-05} +{"train_loss": 0.0166394654661417, "grad_norm": 0.14684244990348816, "global_step": 25787, "epoch": 243, "lr": 5.2109575471848734e-05} +{"train_loss": 0.016664795577526093, "grad_norm": 0.0741022601723671, "global_step": 25788, "epoch": 243, "lr": 5.2106614340600376e-05} +{"train_loss": 0.015220953151583672, "grad_norm": 0.09349251538515091, "global_step": 25789, "epoch": 243, "lr": 5.2103653201950285e-05} +{"train_loss": 0.015660708770155907, "grad_norm": 0.05019226297736168, "global_step": 25790, "epoch": 243, "lr": 5.2100692055908876e-05} +{"train_loss": 0.016922973096370697, "grad_norm": 0.07301098108291626, "global_step": 25791, "epoch": 243, "lr": 5.2097730902486544e-05} +{"train_loss": 0.018409304320812225, "grad_norm": 0.11740466207265854, "global_step": 25792, "epoch": 243, "lr": 5.20947697416937e-05} +{"train_loss": 0.015372585505247116, "grad_norm": 0.0793672502040863, "global_step": 25793, "epoch": 243, "lr": 5.2091808573540744e-05} +{"train_loss": 0.015738124027848244, "grad_norm": 0.11486057937145233, "global_step": 25794, "epoch": 243, "lr": 5.208884739803808e-05} +{"train_loss": 0.015833009034395218, "grad_norm": 0.10109136253595352, "global_step": 25795, "epoch": 243, "lr": 5.2085886215196124e-05} +{"train_loss": 0.01325751282274723, "grad_norm": 0.06395837664604187, "global_step": 25796, "epoch": 243, "lr": 5.2082925025025254e-05} +{"train_loss": 0.013930068351328373, "grad_norm": 0.06215275079011917, "global_step": 25797, "epoch": 243, "lr": 5.2079963827535904e-05} +{"train_loss": 0.01396425161510706, "grad_norm": 0.08542846143245697, "global_step": 25798, "epoch": 243, "lr": 5.207700262273847e-05} +{"train_loss": 0.016568923369050026, "grad_norm": 0.08683162927627563, "global_step": 25799, "epoch": 243, "lr": 5.207404141064334e-05} +{"train_loss": 0.01535709947347641, "grad_norm": 0.1188051775097847, "global_step": 25800, "epoch": 243, "lr": 5.2071080191260936e-05} +{"train_loss": 0.013924108818173409, "grad_norm": 0.09322596341371536, "global_step": 25801, "epoch": 243, "lr": 5.2068118964601655e-05} +{"train_loss": 0.017523135989904404, "grad_norm": 0.09799005836248398, "global_step": 25802, "epoch": 243, "lr": 5.206515773067591e-05} +{"train_loss": 0.01333431527018547, "grad_norm": 0.0626535713672638, "global_step": 25803, "epoch": 243, "lr": 5.2062196489494086e-05} +{"train_loss": 0.015901820734143257, "grad_norm": 0.05319809168577194, "global_step": 25804, "epoch": 243, "lr": 5.205923524106662e-05} +{"train_loss": 0.014895319938659668, "grad_norm": 0.06455640494823456, "global_step": 25805, "epoch": 243, "lr": 5.205627398540388e-05} +{"train_loss": 0.015075568109750748, "grad_norm": 0.1122332215309143, "global_step": 25806, "epoch": 243, "lr": 5.2053312722516276e-05} +{"train_loss": 0.015685411170125008, "grad_norm": 0.07077991217374802, "global_step": 25807, "epoch": 243, "lr": 5.205035145241426e-05} +{"train_loss": 0.015639308840036392, "grad_norm": 0.09400693327188492, "global_step": 25808, "epoch": 243, "lr": 5.204739017510817e-05} +{"train_loss": 0.014598686248064041, "grad_norm": 0.09795037657022476, "global_step": 25809, "epoch": 243, "lr": 5.204442889060845e-05} +{"train_loss": 0.015074579045176506, "grad_norm": 0.09957095235586166, "global_step": 25810, "epoch": 243, "lr": 5.204146759892551e-05} +{"train_loss": 0.01578153669834137, "grad_norm": 0.07418469339609146, "global_step": 25811, "epoch": 243, "lr": 5.2038506300069726e-05} +{"train_loss": 0.015522422268986702, "grad_norm": 0.05434596538543701, "global_step": 25812, "epoch": 243, "lr": 5.203554499405152e-05} +{"train_loss": 0.017513958737254143, "grad_norm": 0.06277477741241455, "global_step": 25813, "epoch": 243, "lr": 5.203258368088129e-05} +{"train_loss": 0.015141413547098637, "grad_norm": 0.09004689007997513, "global_step": 25814, "epoch": 243, "lr": 5.202962236056945e-05} +{"train_loss": 0.015554219484329224, "grad_norm": 0.053007710725069046, "global_step": 25815, "epoch": 243, "lr": 5.20266610331264e-05} +{"train_loss": 0.018253356218338013, "grad_norm": 0.08931880444288254, "global_step": 25816, "epoch": 243, "lr": 5.2023699698562544e-05} +{"train_loss": 0.017815809696912766, "grad_norm": 0.09885890781879425, "global_step": 25817, "epoch": 243, "lr": 5.2020738356888285e-05} +{"train_loss": 0.014298910275101662, "grad_norm": 0.05720297619700432, "global_step": 25818, "epoch": 243, "lr": 5.2017777008114046e-05} +{"train_loss": 0.018208298832178116, "grad_norm": 0.07107096165418625, "global_step": 25819, "epoch": 243, "lr": 5.2014815652250196e-05} +{"train_loss": 0.015089673921465874, "grad_norm": 0.06810152530670166, "global_step": 25820, "epoch": 243, "lr": 5.201185428930717e-05} +{"train_loss": 0.015872538089752197, "grad_norm": 0.14785142242908478, "global_step": 25821, "epoch": 243, "lr": 5.200889291929536e-05} +{"train_loss": 0.014682058244943619, "grad_norm": 0.06060314550995827, "global_step": 25822, "epoch": 243, "lr": 5.2005931542225174e-05} +{"train_loss": 0.015333974733948708, "grad_norm": 0.09106787294149399, "global_step": 25823, "epoch": 243, "lr": 5.2002970158107e-05} +{"train_loss": 0.016271650791168213, "grad_norm": 0.07527491450309753, "global_step": 25824, "epoch": 243, "lr": 5.2000008766951293e-05} +{"train_loss": 0.017717985436320305, "grad_norm": 0.14080214500427246, "global_step": 25825, "epoch": 243, "lr": 5.199704736876839e-05} +{"train_loss": 0.018312280997633934, "grad_norm": 0.1102215126156807, "global_step": 25826, "epoch": 243, "lr": 5.199408596356876e-05} +{"train_loss": 0.016429288312792778, "grad_norm": 0.07614078372716904, "global_step": 25827, "epoch": 243, "lr": 5.199112455136278e-05} +{"train_loss": 0.016328420490026474, "grad_norm": 0.1082921251654625, "global_step": 25828, "epoch": 243, "lr": 5.1988163132160826e-05} +{"train_loss": 0.017049530521035194, "grad_norm": 0.09805165231227875, "global_step": 25829, "epoch": 243, "lr": 5.198520170597335e-05} +{"train_loss": 0.01545555517077446, "grad_norm": 0.09322842955589294, "global_step": 25830, "epoch": 243, "lr": 5.1982240272810747e-05} +{"train_loss": 0.018774019554257393, "grad_norm": 0.13158845901489258, "global_step": 25831, "epoch": 243, "lr": 5.1979278832683395e-05} +{"train_loss": 0.014770107343792915, "grad_norm": 0.06767912954092026, "global_step": 25832, "epoch": 243, "lr": 5.1976317385601726e-05} +{"train_loss": 0.015767216682434082, "grad_norm": 0.07582547515630722, "global_step": 25833, "epoch": 243, "lr": 5.197335593157614e-05} +{"train_loss": 0.018275044858455658, "grad_norm": 0.09619489312171936, "global_step": 25834, "epoch": 243, "lr": 5.197039447061703e-05} +{"train_loss": 0.013703197240829468, "grad_norm": 0.08963077515363693, "global_step": 25835, "epoch": 243, "lr": 5.196743300273481e-05} +{"train_loss": 0.015325394459068775, "grad_norm": 0.05743243545293808, "global_step": 25836, "epoch": 243, "lr": 5.196447152793991e-05} +{"train_loss": 0.014345807023346424, "grad_norm": 0.06711487472057343, "global_step": 25837, "epoch": 243, "lr": 5.196151004624269e-05} +{"train_loss": 0.013721609488129616, "grad_norm": 0.08025467395782471, "global_step": 25838, "epoch": 243, "lr": 5.195854855765357e-05} +{"train_loss": 0.015208836644887924, "grad_norm": 0.05945946276187897, "global_step": 25839, "epoch": 243, "lr": 5.1955587062183e-05} +{"train_loss": 0.016083452850580215, "grad_norm": 0.059477683156728745, "global_step": 25840, "epoch": 243, "lr": 5.195262555984131e-05} +{"train_loss": 0.015157036483287811, "grad_norm": 0.07596208900213242, "global_step": 25841, "epoch": 243, "lr": 5.1949664050638946e-05} +{"train_loss": 0.016362521797418594, "grad_norm": 0.08638621121644974, "global_step": 25842, "epoch": 243, "lr": 5.194670253458632e-05} +{"train_loss": 0.017006564885377884, "grad_norm": 0.07480500638484955, "global_step": 25843, "epoch": 243, "lr": 5.194374101169383e-05} +{"train_loss": 0.012609729543328285, "grad_norm": 0.08189389109611511, "global_step": 25844, "epoch": 243, "lr": 5.194077948197188e-05} +{"train_loss": 0.016562240198254585, "grad_norm": 0.058164555579423904, "global_step": 25845, "epoch": 243, "lr": 5.193781794543087e-05} +{"train_loss": 0.014950892888009548, "grad_norm": 0.06396420300006866, "global_step": 25846, "epoch": 243, "lr": 5.193485640208121e-05} +{"train_loss": 0.0169156976044178, "grad_norm": 0.14706969261169434, "global_step": 25847, "epoch": 243, "lr": 5.1931894851933304e-05} +{"train_loss": 0.015446295961737633, "grad_norm": 0.06122240424156189, "global_step": 25848, "epoch": 243, "lr": 5.1928933294997574e-05} +{"train_loss": 0.015612524002790451, "grad_norm": 0.054000284522771835, "global_step": 25849, "epoch": 243, "lr": 5.1925971731284395e-05} +{"train_loss": 0.014885958284139633, "grad_norm": 0.12649023532867432, "global_step": 25850, "epoch": 243, "lr": 5.192301016080419e-05} +{"train_loss": 0.017016232013702393, "grad_norm": 0.09304507821798325, "global_step": 25851, "epoch": 243, "lr": 5.192004858356738e-05} +{"train_loss": 0.015587685629725456, "grad_norm": 0.07199712097644806, "global_step": 25852, "epoch": 243, "lr": 5.191708699958433e-05} +{"train_loss": 0.017897646874189377, "grad_norm": 0.11492499709129333, "global_step": 25853, "epoch": 243, "lr": 5.191412540886548e-05} +{"train_loss": 0.014604588970541954, "grad_norm": 0.11852068454027176, "global_step": 25854, "epoch": 243, "lr": 5.1911163811421227e-05} +{"train_loss": 0.016067663207650185, "grad_norm": 0.07495363056659698, "global_step": 25855, "epoch": 243, "lr": 5.190820220726198e-05} +{"train_loss": 0.01569180376827717, "grad_norm": 0.09956982731819153, "global_step": 25856, "epoch": 243, "lr": 5.190524059639814e-05} +{"train_loss": 0.014434010721743107, "grad_norm": 0.08521702885627747, "global_step": 25857, "epoch": 243, "lr": 5.1902278978840116e-05} +{"train_loss": 0.015589114278554916, "grad_norm": 0.06710653752088547, "global_step": 25858, "epoch": 243, "lr": 5.18993173545983e-05} +{"train_loss": 0.014399362727999687, "grad_norm": 0.055589765310287476, "global_step": 25859, "epoch": 243, "lr": 5.189635572368311e-05} +{"train_loss": 0.016203993931412697, "grad_norm": 0.059712693095207214, "global_step": 25860, "epoch": 243, "lr": 5.189339408610496e-05} +{"train_loss": 0.01603279635310173, "grad_norm": 0.07099246233701706, "global_step": 25861, "epoch": 243, "lr": 5.189043244187424e-05} +{"train_loss": 0.01339947059750557, "grad_norm": 0.05604681372642517, "global_step": 25862, "epoch": 243, "lr": 5.188747079100136e-05} +{"train_loss": 0.015597885975368181, "grad_norm": 0.0707770437002182, "global_step": 25863, "epoch": 243, "lr": 5.188450913349674e-05, "val_loss": 0.01901603862643242} +{"train_loss": 0.015264800749719143, "grad_norm": 0.13387389481067657, "global_step": 25864, "epoch": 244, "lr": 5.188154746937075e-05} +{"train_loss": 0.01517781987786293, "grad_norm": 0.10050241649150848, "global_step": 25865, "epoch": 244, "lr": 5.187858579863384e-05} +{"train_loss": 0.013561537489295006, "grad_norm": 0.06547768414020538, "global_step": 25866, "epoch": 244, "lr": 5.187562412129639e-05} +{"train_loss": 0.018709201365709305, "grad_norm": 0.12297260016202927, "global_step": 25867, "epoch": 244, "lr": 5.187266243736882e-05} +{"train_loss": 0.014042277820408344, "grad_norm": 0.08494792878627777, "global_step": 25868, "epoch": 244, "lr": 5.186970074686154e-05} +{"train_loss": 0.015866262838244438, "grad_norm": 0.09592799097299576, "global_step": 25869, "epoch": 244, "lr": 5.186673904978492e-05} +{"train_loss": 0.01596113294363022, "grad_norm": 0.13704510033130646, "global_step": 25870, "epoch": 244, "lr": 5.18637773461494e-05} +{"train_loss": 0.013853989541530609, "grad_norm": 0.09275896847248077, "global_step": 25871, "epoch": 244, "lr": 5.186081563596538e-05} +{"train_loss": 0.015123776160180569, "grad_norm": 0.0880596712231636, "global_step": 25872, "epoch": 244, "lr": 5.185785391924326e-05} +{"train_loss": 0.0169198177754879, "grad_norm": 0.08335043489933014, "global_step": 25873, "epoch": 244, "lr": 5.1854892195993445e-05} +{"train_loss": 0.015082797035574913, "grad_norm": 0.05752218887209892, "global_step": 25874, "epoch": 244, "lr": 5.185193046622635e-05} +{"train_loss": 0.015375250950455666, "grad_norm": 0.07490423321723938, "global_step": 25875, "epoch": 244, "lr": 5.184896872995238e-05} +{"train_loss": 0.017108283936977386, "grad_norm": 0.13195271790027618, "global_step": 25876, "epoch": 244, "lr": 5.184600698718193e-05} +{"train_loss": 0.014488518238067627, "grad_norm": 0.11781960725784302, "global_step": 25877, "epoch": 244, "lr": 5.1843045237925424e-05} +{"train_loss": 0.0175591092556715, "grad_norm": 0.08166911453008652, "global_step": 25878, "epoch": 244, "lr": 5.184008348219326e-05} +{"train_loss": 0.01857803203165531, "grad_norm": 0.09435441344976425, "global_step": 25879, "epoch": 244, "lr": 5.1837121719995816e-05} +{"train_loss": 0.01718415878713131, "grad_norm": 0.07993324846029282, "global_step": 25880, "epoch": 244, "lr": 5.1834159951343564e-05} +{"train_loss": 0.01604701578617096, "grad_norm": 0.0798938050866127, "global_step": 25881, "epoch": 244, "lr": 5.183119817624684e-05} +{"train_loss": 0.016987856477499008, "grad_norm": 0.12175668776035309, "global_step": 25882, "epoch": 244, "lr": 5.18282363947161e-05} +{"train_loss": 0.014456437900662422, "grad_norm": 0.05834601819515228, "global_step": 25883, "epoch": 244, "lr": 5.182527460676172e-05} +{"train_loss": 0.015308782458305359, "grad_norm": 0.05977797508239746, "global_step": 25884, "epoch": 244, "lr": 5.182231281239414e-05} +{"train_loss": 0.016525574028491974, "grad_norm": 0.06329299509525299, "global_step": 25885, "epoch": 244, "lr": 5.181935101162372e-05} +{"train_loss": 0.016508586704730988, "grad_norm": 0.05583031848073006, "global_step": 25886, "epoch": 244, "lr": 5.18163892044609e-05} +{"train_loss": 0.015400135889649391, "grad_norm": 0.05819296836853027, "global_step": 25887, "epoch": 244, "lr": 5.1813427390916095e-05} +{"train_loss": 0.015367181971669197, "grad_norm": 0.1371866911649704, "global_step": 25888, "epoch": 244, "lr": 5.181046557099967e-05} +{"train_loss": 0.015817519277334213, "grad_norm": 0.07290874421596527, "global_step": 25889, "epoch": 244, "lr": 5.180750374472208e-05} +{"train_loss": 0.014217261224985123, "grad_norm": 0.07997894287109375, "global_step": 25890, "epoch": 244, "lr": 5.18045419120937e-05} +{"train_loss": 0.016689492389559746, "grad_norm": 0.10334058851003647, "global_step": 25891, "epoch": 244, "lr": 5.1801580073124924e-05} +{"train_loss": 0.015024891123175621, "grad_norm": 0.08999497443437576, "global_step": 25892, "epoch": 244, "lr": 5.17986182278262e-05} +{"train_loss": 0.016736190766096115, "grad_norm": 0.0834040492773056, "global_step": 25893, "epoch": 244, "lr": 5.1795656376207904e-05} +{"train_loss": 0.018984897062182426, "grad_norm": 0.06608475744724274, "global_step": 25894, "epoch": 244, "lr": 5.179269451828046e-05} +{"train_loss": 0.014515379443764687, "grad_norm": 0.09918522089719772, "global_step": 25895, "epoch": 244, "lr": 5.1789732654054255e-05} +{"train_loss": 0.015681084245443344, "grad_norm": 0.07516974210739136, "global_step": 25896, "epoch": 244, "lr": 5.178677078353973e-05} +{"train_loss": 0.013039822690188885, "grad_norm": 0.11611910909414291, "global_step": 25897, "epoch": 244, "lr": 5.178380890674726e-05} +{"train_loss": 0.015562830492854118, "grad_norm": 0.06733622401952744, "global_step": 25898, "epoch": 244, "lr": 5.178084702368725e-05} +{"train_loss": 0.01599632203578949, "grad_norm": 0.16556574404239655, "global_step": 25899, "epoch": 244, "lr": 5.177788513437013e-05} +{"train_loss": 0.018721550703048706, "grad_norm": 0.06507240980863571, "global_step": 25900, "epoch": 244, "lr": 5.177492323880628e-05} +{"train_loss": 0.018736444413661957, "grad_norm": 0.07285095751285553, "global_step": 25901, "epoch": 244, "lr": 5.177196133700614e-05} +{"train_loss": 0.017752084881067276, "grad_norm": 0.05577744171023369, "global_step": 25902, "epoch": 244, "lr": 5.176899942898009e-05} +{"train_loss": 0.014996057376265526, "grad_norm": 0.06950978189706802, "global_step": 25903, "epoch": 244, "lr": 5.176603751473854e-05} +{"train_loss": 0.01398308202624321, "grad_norm": 0.07224980741739273, "global_step": 25904, "epoch": 244, "lr": 5.176307559429192e-05} +{"train_loss": 0.01461085770279169, "grad_norm": 0.05812124162912369, "global_step": 25905, "epoch": 244, "lr": 5.17601136676506e-05} +{"train_loss": 0.015007875859737396, "grad_norm": 0.06303998082876205, "global_step": 25906, "epoch": 244, "lr": 5.175715173482501e-05} +{"train_loss": 0.014621260575950146, "grad_norm": 0.05995042249560356, "global_step": 25907, "epoch": 244, "lr": 5.175418979582557e-05} +{"train_loss": 0.01420009694993496, "grad_norm": 0.10550595074892044, "global_step": 25908, "epoch": 244, "lr": 5.175122785066264e-05} +{"train_loss": 0.014798698015511036, "grad_norm": 0.06571204215288162, "global_step": 25909, "epoch": 244, "lr": 5.174826589934668e-05} +{"train_loss": 0.01811833679676056, "grad_norm": 0.07617288827896118, "global_step": 25910, "epoch": 244, "lr": 5.174530394188807e-05} +{"train_loss": 0.015055084601044655, "grad_norm": 0.08627678453922272, "global_step": 25911, "epoch": 244, "lr": 5.1742341978297216e-05} +{"train_loss": 0.01496911607682705, "grad_norm": 0.16021932661533356, "global_step": 25912, "epoch": 244, "lr": 5.1739380008584537e-05} +{"train_loss": 0.016748519614338875, "grad_norm": 0.13295356929302216, "global_step": 25913, "epoch": 244, "lr": 5.1736418032760426e-05} +{"train_loss": 0.01406182162463665, "grad_norm": 0.07848623394966125, "global_step": 25914, "epoch": 244, "lr": 5.1733456050835294e-05} +{"train_loss": 0.015528284013271332, "grad_norm": 0.06935334205627441, "global_step": 25915, "epoch": 244, "lr": 5.1730494062819554e-05} +{"train_loss": 0.014457957819104195, "grad_norm": 0.1301928013563156, "global_step": 25916, "epoch": 244, "lr": 5.172753206872363e-05} +{"train_loss": 0.017781130969524384, "grad_norm": 0.11109257489442825, "global_step": 25917, "epoch": 244, "lr": 5.172457006855788e-05} +{"train_loss": 0.015272222459316254, "grad_norm": 0.08446869254112244, "global_step": 25918, "epoch": 244, "lr": 5.172160806233276e-05} +{"train_loss": 0.015128294937312603, "grad_norm": 0.11376448720693588, "global_step": 25919, "epoch": 244, "lr": 5.1718646050058653e-05} +{"train_loss": 0.016540396958589554, "grad_norm": 0.08547939360141754, "global_step": 25920, "epoch": 244, "lr": 5.171568403174596e-05} +{"train_loss": 0.014105078764259815, "grad_norm": 0.10589198023080826, "global_step": 25921, "epoch": 244, "lr": 5.1712722007405125e-05} +{"train_loss": 0.014848338440060616, "grad_norm": 0.11475980281829834, "global_step": 25922, "epoch": 244, "lr": 5.17097599770465e-05} +{"train_loss": 0.015015652403235435, "grad_norm": 0.07262491434812546, "global_step": 25923, "epoch": 244, "lr": 5.170679794068054e-05} +{"train_loss": 0.01469143945723772, "grad_norm": 0.06470390409231186, "global_step": 25924, "epoch": 244, "lr": 5.1703835898317635e-05} +{"train_loss": 0.015190218575298786, "grad_norm": 0.060481492429971695, "global_step": 25925, "epoch": 244, "lr": 5.170087384996818e-05} +{"train_loss": 0.015330134890973568, "grad_norm": 0.061219897121191025, "global_step": 25926, "epoch": 244, "lr": 5.1697911795642606e-05} +{"train_loss": 0.016293611377477646, "grad_norm": 0.11975213885307312, "global_step": 25927, "epoch": 244, "lr": 5.169494973535129e-05} +{"train_loss": 0.015335393138229847, "grad_norm": 0.12579523026943207, "global_step": 25928, "epoch": 244, "lr": 5.169198766910468e-05} +{"train_loss": 0.014263940043747425, "grad_norm": 0.05785627290606499, "global_step": 25929, "epoch": 244, "lr": 5.168902559691315e-05} +{"train_loss": 0.01652839221060276, "grad_norm": 0.05656745657324791, "global_step": 25930, "epoch": 244, "lr": 5.168606351878712e-05} +{"train_loss": 0.015484275296330452, "grad_norm": 0.19604578614234924, "global_step": 25931, "epoch": 244, "lr": 5.1683101434737e-05} +{"train_loss": 0.013614149764180183, "grad_norm": 0.10999502241611481, "global_step": 25932, "epoch": 244, "lr": 5.168013934477318e-05} +{"train_loss": 0.016096610575914383, "grad_norm": 0.09847231209278107, "global_step": 25933, "epoch": 244, "lr": 5.1677177248906104e-05} +{"train_loss": 0.015322001650929451, "grad_norm": 0.10786700248718262, "global_step": 25934, "epoch": 244, "lr": 5.167421514714613e-05} +{"train_loss": 0.016982898116111755, "grad_norm": 0.1785230189561844, "global_step": 25935, "epoch": 244, "lr": 5.167125303950371e-05} +{"train_loss": 0.015241017565131187, "grad_norm": 0.046686284244060516, "global_step": 25936, "epoch": 244, "lr": 5.166829092598924e-05} +{"train_loss": 0.015804678201675415, "grad_norm": 0.10327710956335068, "global_step": 25937, "epoch": 244, "lr": 5.16653288066131e-05} +{"train_loss": 0.01400013454258442, "grad_norm": 0.10498788207769394, "global_step": 25938, "epoch": 244, "lr": 5.166236668138573e-05} +{"train_loss": 0.015343111008405685, "grad_norm": 0.12704549729824066, "global_step": 25939, "epoch": 244, "lr": 5.165940455031752e-05} +{"train_loss": 0.016784965991973877, "grad_norm": 0.06889644265174866, "global_step": 25940, "epoch": 244, "lr": 5.16564424134189e-05} +{"train_loss": 0.01599852740764618, "grad_norm": 0.06971518695354462, "global_step": 25941, "epoch": 244, "lr": 5.165348027070025e-05} +{"train_loss": 0.014933031052350998, "grad_norm": 0.05705498531460762, "global_step": 25942, "epoch": 244, "lr": 5.1650518122172e-05} +{"train_loss": 0.016957465559244156, "grad_norm": 0.0990041196346283, "global_step": 25943, "epoch": 244, "lr": 5.164755596784453e-05} +{"train_loss": 0.014691921882331371, "grad_norm": 0.07158693671226501, "global_step": 25944, "epoch": 244, "lr": 5.164459380772827e-05} +{"train_loss": 0.012233661487698555, "grad_norm": 0.11688517779111862, "global_step": 25945, "epoch": 244, "lr": 5.1641631641833634e-05} +{"train_loss": 0.015823476016521454, "grad_norm": 0.06930642575025558, "global_step": 25946, "epoch": 244, "lr": 5.1638669470171e-05} +{"train_loss": 0.013305654749274254, "grad_norm": 0.08411958068609238, "global_step": 25947, "epoch": 244, "lr": 5.16357072927508e-05} +{"train_loss": 0.016621213406324387, "grad_norm": 0.10940581560134888, "global_step": 25948, "epoch": 244, "lr": 5.163274510958346e-05} +{"train_loss": 0.015605820342898369, "grad_norm": 0.05805882066488266, "global_step": 25949, "epoch": 244, "lr": 5.162978292067933e-05} +{"train_loss": 0.014161113649606705, "grad_norm": 0.05840284004807472, "global_step": 25950, "epoch": 244, "lr": 5.1626820726048866e-05} +{"train_loss": 0.01726692169904709, "grad_norm": 0.07322060316801071, "global_step": 25951, "epoch": 244, "lr": 5.1623858525702465e-05} +{"train_loss": 0.015081660822033882, "grad_norm": 0.06737755984067917, "global_step": 25952, "epoch": 244, "lr": 5.1620896319650524e-05} +{"train_loss": 0.015321975573897362, "grad_norm": 0.07616747915744781, "global_step": 25953, "epoch": 244, "lr": 5.1617934107903465e-05} +{"train_loss": 0.01442315150052309, "grad_norm": 0.05849093943834305, "global_step": 25954, "epoch": 244, "lr": 5.1614971890471684e-05} +{"train_loss": 0.014161713421344757, "grad_norm": 0.07973384112119675, "global_step": 25955, "epoch": 244, "lr": 5.16120096673656e-05} +{"train_loss": 0.017758673056960106, "grad_norm": 0.08537402749061584, "global_step": 25956, "epoch": 244, "lr": 5.160904743859559e-05} +{"train_loss": 0.01650751382112503, "grad_norm": 0.07621174305677414, "global_step": 25957, "epoch": 244, "lr": 5.160608520417213e-05} +{"train_loss": 0.013119556941092014, "grad_norm": 0.0677933543920517, "global_step": 25958, "epoch": 244, "lr": 5.1603122964105566e-05} +{"train_loss": 0.01434175856411457, "grad_norm": 0.08865824341773987, "global_step": 25959, "epoch": 244, "lr": 5.160016071840631e-05} +{"train_loss": 0.01832367666065693, "grad_norm": 0.06816413998603821, "global_step": 25960, "epoch": 244, "lr": 5.159719846708481e-05} +{"train_loss": 0.014641258865594864, "grad_norm": 0.0511084645986557, "global_step": 25961, "epoch": 244, "lr": 5.159423621015142e-05} +{"train_loss": 0.01526285894215107, "grad_norm": 0.10246460884809494, "global_step": 25962, "epoch": 244, "lr": 5.159127394761659e-05} +{"train_loss": 0.018594030290842056, "grad_norm": 0.08317197859287262, "global_step": 25963, "epoch": 244, "lr": 5.158831167949072e-05} +{"train_loss": 0.01602604240179062, "grad_norm": 0.05863790959119797, "global_step": 25964, "epoch": 244, "lr": 5.158534940578422e-05} +{"train_loss": 0.015879549086093903, "grad_norm": 0.0899277850985527, "global_step": 25965, "epoch": 244, "lr": 5.158238712650748e-05} +{"train_loss": 0.014408101327717304, "grad_norm": 0.07494094967842102, "global_step": 25966, "epoch": 244, "lr": 5.157942484167093e-05} +{"train_loss": 0.016876593232154846, "grad_norm": 0.06285639107227325, "global_step": 25967, "epoch": 244, "lr": 5.157646255128496e-05} +{"train_loss": 0.013668776489794254, "grad_norm": 0.09491309523582458, "global_step": 25968, "epoch": 244, "lr": 5.1573500255359986e-05} +{"train_loss": 0.015617568757525593, "grad_norm": 0.1113269180059433, "global_step": 25969, "epoch": 244, "lr": 5.157053795390642e-05, "val_loss": 0.019459469243884087} +{"train_loss": 0.015321608632802963, "grad_norm": 0.09835889935493469, "global_step": 25970, "epoch": 245, "lr": 5.1567575646934663e-05} +{"train_loss": 0.013982512056827545, "grad_norm": 0.0857643112540245, "global_step": 25971, "epoch": 245, "lr": 5.156461333445513e-05} +{"train_loss": 0.014117308892309666, "grad_norm": 0.10371529310941696, "global_step": 25972, "epoch": 245, "lr": 5.156165101647823e-05} +{"train_loss": 0.016222335398197174, "grad_norm": 0.08156286180019379, "global_step": 25973, "epoch": 245, "lr": 5.1558688693014357e-05} +{"train_loss": 0.01265835203230381, "grad_norm": 0.06441352516412735, "global_step": 25974, "epoch": 245, "lr": 5.155572636407394e-05} +{"train_loss": 0.015915624797344208, "grad_norm": 0.08712355047464371, "global_step": 25975, "epoch": 245, "lr": 5.155276402966736e-05} +{"train_loss": 0.016191687434911728, "grad_norm": 0.0489237941801548, "global_step": 25976, "epoch": 245, "lr": 5.154980168980507e-05} +{"train_loss": 0.017439285293221474, "grad_norm": 0.12006228417158127, "global_step": 25977, "epoch": 245, "lr": 5.154683934449745e-05} +{"train_loss": 0.014512652531266212, "grad_norm": 0.11455360054969788, "global_step": 25978, "epoch": 245, "lr": 5.154387699375487e-05} +{"train_loss": 0.014622125774621964, "grad_norm": 0.07181132584810257, "global_step": 25979, "epoch": 245, "lr": 5.154091463758781e-05} +{"train_loss": 0.016789384186267853, "grad_norm": 0.1296800673007965, "global_step": 25980, "epoch": 245, "lr": 5.1537952276006644e-05} +{"train_loss": 0.01691536232829094, "grad_norm": 0.08912822604179382, "global_step": 25981, "epoch": 245, "lr": 5.153498990902177e-05} +{"train_loss": 0.0163232684135437, "grad_norm": 0.06374416500329971, "global_step": 25982, "epoch": 245, "lr": 5.153202753664362e-05} +{"train_loss": 0.01583736576139927, "grad_norm": 0.07693987339735031, "global_step": 25983, "epoch": 245, "lr": 5.152906515888258e-05} +{"train_loss": 0.017110874876379967, "grad_norm": 0.10108160227537155, "global_step": 25984, "epoch": 245, "lr": 5.152610277574909e-05} +{"train_loss": 0.014483126811683178, "grad_norm": 0.09182053059339523, "global_step": 25985, "epoch": 245, "lr": 5.152314038725351e-05} +{"train_loss": 0.012422074563801289, "grad_norm": 0.057702045887708664, "global_step": 25986, "epoch": 245, "lr": 5.1520177993406304e-05} +{"train_loss": 0.01562325470149517, "grad_norm": 0.071882463991642, "global_step": 25987, "epoch": 245, "lr": 5.151721559421785e-05} +{"train_loss": 0.013686088845133781, "grad_norm": 0.0862627774477005, "global_step": 25988, "epoch": 245, "lr": 5.151425318969854e-05} +{"train_loss": 0.01491324882954359, "grad_norm": 0.06268979609012604, "global_step": 25989, "epoch": 245, "lr": 5.1511290779858825e-05} +{"train_loss": 0.01405344344675541, "grad_norm": 0.06643475592136383, "global_step": 25990, "epoch": 245, "lr": 5.150832836470908e-05} +{"train_loss": 0.0162862129509449, "grad_norm": 0.0634458065032959, "global_step": 25991, "epoch": 245, "lr": 5.150536594425972e-05} +{"train_loss": 0.013445623219013214, "grad_norm": 0.09138168394565582, "global_step": 25992, "epoch": 245, "lr": 5.150240351852117e-05} +{"train_loss": 0.01698731817305088, "grad_norm": 0.060254015028476715, "global_step": 25993, "epoch": 245, "lr": 5.149944108750383e-05} +{"train_loss": 0.015119615942239761, "grad_norm": 0.061447445303201675, "global_step": 25994, "epoch": 245, "lr": 5.14964786512181e-05} +{"train_loss": 0.0140297319740057, "grad_norm": 0.06706878542900085, "global_step": 25995, "epoch": 245, "lr": 5.149351620967439e-05} +{"train_loss": 0.01885467767715454, "grad_norm": 0.07638711482286453, "global_step": 25996, "epoch": 245, "lr": 5.149055376288312e-05} +{"train_loss": 0.01611141674220562, "grad_norm": 0.08838546276092529, "global_step": 25997, "epoch": 245, "lr": 5.1487591310854675e-05} +{"train_loss": 0.01586063951253891, "grad_norm": 0.06651704013347626, "global_step": 25998, "epoch": 245, "lr": 5.1484628853599514e-05} +{"train_loss": 0.012801719829440117, "grad_norm": 0.06302487850189209, "global_step": 25999, "epoch": 245, "lr": 5.148166639112799e-05} +{"train_loss": 0.013640817254781723, "grad_norm": 0.05583278462290764, "global_step": 26000, "epoch": 245, "lr": 5.147870392345053e-05} +{"train_loss": 0.018354248255491257, "grad_norm": 0.06513050943613052, "global_step": 26001, "epoch": 245, "lr": 5.1475741450577575e-05} +{"train_loss": 0.01368024107068777, "grad_norm": 0.077143095433712, "global_step": 26002, "epoch": 245, "lr": 5.1472778972519475e-05} +{"train_loss": 0.018107648938894272, "grad_norm": 0.06975533813238144, "global_step": 26003, "epoch": 245, "lr": 5.1469816489286695e-05} +{"train_loss": 0.014091265387833118, "grad_norm": 0.06684695929288864, "global_step": 26004, "epoch": 245, "lr": 5.146685400088961e-05} +{"train_loss": 0.013767199590802193, "grad_norm": 0.10160938650369644, "global_step": 26005, "epoch": 245, "lr": 5.146389150733863e-05} +{"train_loss": 0.01708057150244713, "grad_norm": 0.05307856202125549, "global_step": 26006, "epoch": 245, "lr": 5.1460929008644174e-05} +{"train_loss": 0.015929076820611954, "grad_norm": 0.06489923596382141, "global_step": 26007, "epoch": 245, "lr": 5.1457966504816656e-05} +{"train_loss": 0.015172578394412994, "grad_norm": 0.1063961461186409, "global_step": 26008, "epoch": 245, "lr": 5.145500399586648e-05} +{"train_loss": 0.013687756843864918, "grad_norm": 0.04926665499806404, "global_step": 26009, "epoch": 245, "lr": 5.145204148180405e-05} +{"train_loss": 0.013544511049985886, "grad_norm": 0.09430702030658722, "global_step": 26010, "epoch": 245, "lr": 5.144907896263977e-05} +{"train_loss": 0.01495266705751419, "grad_norm": 0.13618700206279755, "global_step": 26011, "epoch": 245, "lr": 5.144611643838406e-05} +{"train_loss": 0.016096871346235275, "grad_norm": 0.04917115718126297, "global_step": 26012, "epoch": 245, "lr": 5.1443153909047327e-05} +{"train_loss": 0.013688648119568825, "grad_norm": 0.053986694663763046, "global_step": 26013, "epoch": 245, "lr": 5.144019137463999e-05} +{"train_loss": 0.015248266980051994, "grad_norm": 0.10067078471183777, "global_step": 26014, "epoch": 245, "lr": 5.1437228835172425e-05} +{"train_loss": 0.0160248801112175, "grad_norm": 0.08335158973932266, "global_step": 26015, "epoch": 245, "lr": 5.143426629065508e-05} +{"train_loss": 0.021470610052347183, "grad_norm": 0.1277247667312622, "global_step": 26016, "epoch": 245, "lr": 5.143130374109835e-05} +{"train_loss": 0.016584571450948715, "grad_norm": 0.09867247939109802, "global_step": 26017, "epoch": 245, "lr": 5.142834118651262e-05} +{"train_loss": 0.014906351454555988, "grad_norm": 0.0938088595867157, "global_step": 26018, "epoch": 245, "lr": 5.1425378626908326e-05} +{"train_loss": 0.01642981916666031, "grad_norm": 0.08814051002264023, "global_step": 26019, "epoch": 245, "lr": 5.1422416062295886e-05} +{"train_loss": 0.015611112117767334, "grad_norm": 0.05965160205960274, "global_step": 26020, "epoch": 245, "lr": 5.141945349268569e-05} +{"train_loss": 0.01671500876545906, "grad_norm": 0.06474733352661133, "global_step": 26021, "epoch": 245, "lr": 5.1416490918088147e-05} +{"train_loss": 0.015149177052080631, "grad_norm": 0.08081137388944626, "global_step": 26022, "epoch": 245, "lr": 5.141352833851367e-05} +{"train_loss": 0.018665174022316933, "grad_norm": 0.10903045535087585, "global_step": 26023, "epoch": 245, "lr": 5.141056575397267e-05} +{"train_loss": 0.014067616313695908, "grad_norm": 0.07952277362346649, "global_step": 26024, "epoch": 245, "lr": 5.140760316447556e-05} +{"train_loss": 0.01781991496682167, "grad_norm": 0.13157373666763306, "global_step": 26025, "epoch": 245, "lr": 5.140464057003274e-05} +{"train_loss": 0.014232784509658813, "grad_norm": 0.09601282328367233, "global_step": 26026, "epoch": 245, "lr": 5.140167797065462e-05} +{"train_loss": 0.015167278237640858, "grad_norm": 0.08927828073501587, "global_step": 26027, "epoch": 245, "lr": 5.1398715366351623e-05} +{"train_loss": 0.014310121536254883, "grad_norm": 0.07279497385025024, "global_step": 26028, "epoch": 245, "lr": 5.1395752757134144e-05} +{"train_loss": 0.016773996874690056, "grad_norm": 0.06695765256881714, "global_step": 26029, "epoch": 245, "lr": 5.1392790143012595e-05} +{"train_loss": 0.014980059117078781, "grad_norm": 0.08423298597335815, "global_step": 26030, "epoch": 245, "lr": 5.138982752399739e-05} +{"train_loss": 0.01810036227107048, "grad_norm": 0.15321359038352966, "global_step": 26031, "epoch": 245, "lr": 5.138686490009893e-05} +{"train_loss": 0.015352565795183182, "grad_norm": 0.06173066049814224, "global_step": 26032, "epoch": 245, "lr": 5.138390227132763e-05} +{"train_loss": 0.01592719554901123, "grad_norm": 0.10689082741737366, "global_step": 26033, "epoch": 245, "lr": 5.1380939637693906e-05} +{"train_loss": 0.014775989577174187, "grad_norm": 0.06736967712640762, "global_step": 26034, "epoch": 245, "lr": 5.137797699920815e-05} +{"train_loss": 0.015056867152452469, "grad_norm": 0.0702318549156189, "global_step": 26035, "epoch": 245, "lr": 5.1375014355880805e-05} +{"train_loss": 0.016969334334135056, "grad_norm": 0.07668045163154602, "global_step": 26036, "epoch": 245, "lr": 5.137205170772223e-05} +{"train_loss": 0.01650749333202839, "grad_norm": 0.08491842448711395, "global_step": 26037, "epoch": 245, "lr": 5.136908905474288e-05} +{"train_loss": 0.016137847676873207, "grad_norm": 0.13776680827140808, "global_step": 26038, "epoch": 245, "lr": 5.136612639695315e-05} +{"train_loss": 0.01777285896241665, "grad_norm": 0.06042136996984482, "global_step": 26039, "epoch": 245, "lr": 5.136316373436343e-05} +{"train_loss": 0.01751807890832424, "grad_norm": 0.12035729736089706, "global_step": 26040, "epoch": 245, "lr": 5.136020106698416e-05} +{"train_loss": 0.013712747022509575, "grad_norm": 0.08076745271682739, "global_step": 26041, "epoch": 245, "lr": 5.1357238394825713e-05} +{"train_loss": 0.015367310494184494, "grad_norm": 0.0719127207994461, "global_step": 26042, "epoch": 245, "lr": 5.135427571789855e-05} +{"train_loss": 0.016606122255325317, "grad_norm": 0.14108076691627502, "global_step": 26043, "epoch": 245, "lr": 5.135131303621302e-05} +{"train_loss": 0.017043771222233772, "grad_norm": 0.1189698576927185, "global_step": 26044, "epoch": 245, "lr": 5.134835034977959e-05} +{"train_loss": 0.01593485288321972, "grad_norm": 0.06754665076732635, "global_step": 26045, "epoch": 245, "lr": 5.134538765860862e-05} +{"train_loss": 0.01584940403699875, "grad_norm": 0.07921300828456879, "global_step": 26046, "epoch": 245, "lr": 5.134242496271057e-05} +{"train_loss": 0.015314731746912003, "grad_norm": 0.13428694009780884, "global_step": 26047, "epoch": 245, "lr": 5.133946226209581e-05} +{"train_loss": 0.016911868005990982, "grad_norm": 0.08636637777090073, "global_step": 26048, "epoch": 245, "lr": 5.1336499556774756e-05} +{"train_loss": 0.014619098976254463, "grad_norm": 0.10605490952730179, "global_step": 26049, "epoch": 245, "lr": 5.133353684675782e-05} +{"train_loss": 0.01616535149514675, "grad_norm": 0.06543918699026108, "global_step": 26050, "epoch": 245, "lr": 5.133057413205543e-05} +{"train_loss": 0.013916194438934326, "grad_norm": 0.07762128114700317, "global_step": 26051, "epoch": 245, "lr": 5.132761141267798e-05} +{"train_loss": 0.015391763299703598, "grad_norm": 0.11565124243497849, "global_step": 26052, "epoch": 245, "lr": 5.132464868863588e-05} +{"train_loss": 0.015256844460964203, "grad_norm": 0.06883081793785095, "global_step": 26053, "epoch": 245, "lr": 5.1321685959939525e-05} +{"train_loss": 0.014497329480946064, "grad_norm": 0.07070939242839813, "global_step": 26054, "epoch": 245, "lr": 5.131872322659936e-05} +{"train_loss": 0.014303820207715034, "grad_norm": 0.07482675462961197, "global_step": 26055, "epoch": 245, "lr": 5.131576048862575e-05} +{"train_loss": 0.015177195891737938, "grad_norm": 0.06328827887773514, "global_step": 26056, "epoch": 245, "lr": 5.1312797746029153e-05} +{"train_loss": 0.015816401690244675, "grad_norm": 0.07105052471160889, "global_step": 26057, "epoch": 245, "lr": 5.130983499881996e-05} +{"train_loss": 0.014842027798295021, "grad_norm": 0.06562888622283936, "global_step": 26058, "epoch": 245, "lr": 5.1306872247008555e-05} +{"train_loss": 0.017542598769068718, "grad_norm": 0.07341421395540237, "global_step": 26059, "epoch": 245, "lr": 5.1303909490605386e-05} +{"train_loss": 0.013990199193358421, "grad_norm": 0.07137653976678848, "global_step": 26060, "epoch": 245, "lr": 5.130094672962083e-05} +{"train_loss": 0.013655589893460274, "grad_norm": 0.08207305520772934, "global_step": 26061, "epoch": 245, "lr": 5.129798396406533e-05} +{"train_loss": 0.017644114792346954, "grad_norm": 0.072968028485775, "global_step": 26062, "epoch": 245, "lr": 5.1295021193949274e-05} +{"train_loss": 0.015821080654859543, "grad_norm": 0.08785893768072128, "global_step": 26063, "epoch": 245, "lr": 5.129205841928307e-05} +{"train_loss": 0.015390312299132347, "grad_norm": 0.05835491791367531, "global_step": 26064, "epoch": 245, "lr": 5.1289095640077146e-05} +{"train_loss": 0.018809104338288307, "grad_norm": 0.19809621572494507, "global_step": 26065, "epoch": 245, "lr": 5.1286132856341884e-05} +{"train_loss": 0.01705661229789257, "grad_norm": 0.12362218648195267, "global_step": 26066, "epoch": 245, "lr": 5.128317006808773e-05} +{"train_loss": 0.013120342046022415, "grad_norm": 0.07324797660112381, "global_step": 26067, "epoch": 245, "lr": 5.128020727532507e-05} +{"train_loss": 0.01381717436015606, "grad_norm": 0.07963007688522339, "global_step": 26068, "epoch": 245, "lr": 5.12772444780643e-05} +{"train_loss": 0.01827181503176689, "grad_norm": 0.09267140924930573, "global_step": 26069, "epoch": 245, "lr": 5.127428167631587e-05} +{"train_loss": 0.01399319153279066, "grad_norm": 0.09867088496685028, "global_step": 26070, "epoch": 245, "lr": 5.127131887009015e-05} +{"train_loss": 0.015366737730801105, "grad_norm": 0.05853705480694771, "global_step": 26071, "epoch": 245, "lr": 5.1268356059397594e-05} +{"train_loss": 0.015491504222154617, "grad_norm": 0.06338609755039215, "global_step": 26072, "epoch": 245, "lr": 5.1265393244248564e-05} +{"train_loss": 0.015163562260568142, "grad_norm": 0.09774801880121231, "global_step": 26073, "epoch": 245, "lr": 5.12624304246535e-05} +{"train_loss": 0.016627507284283638, "grad_norm": 0.07493624836206436, "global_step": 26074, "epoch": 245, "lr": 5.12594676006228e-05} +{"train_loss": 0.015598763606317763, "grad_norm": 0.12378718703985214, "global_step": 26075, "epoch": 245, "lr": 5.125650477216688e-05, "val_loss": 0.019317176192998886} +{"train_loss": 0.013240167871117592, "grad_norm": 0.05838853865861893, "global_step": 26076, "epoch": 246, "lr": 5.1253541939296145e-05} +{"train_loss": 0.014114459976553917, "grad_norm": 0.11856363713741302, "global_step": 26077, "epoch": 246, "lr": 5.125057910202102e-05} +{"train_loss": 0.016014646738767624, "grad_norm": 0.0818706527352333, "global_step": 26078, "epoch": 246, "lr": 5.1247616260351905e-05} +{"train_loss": 0.016552835702896118, "grad_norm": 0.05997484549880028, "global_step": 26079, "epoch": 246, "lr": 5.1244653414299196e-05} +{"train_loss": 0.014266693964600563, "grad_norm": 0.05975482612848282, "global_step": 26080, "epoch": 246, "lr": 5.124169056387332e-05} +{"train_loss": 0.015840116888284683, "grad_norm": 0.0889262780547142, "global_step": 26081, "epoch": 246, "lr": 5.123872770908469e-05} +{"train_loss": 0.013444913551211357, "grad_norm": 0.06040441244840622, "global_step": 26082, "epoch": 246, "lr": 5.12357648499437e-05} +{"train_loss": 0.01628229208290577, "grad_norm": 0.059334222227334976, "global_step": 26083, "epoch": 246, "lr": 5.123280198646079e-05} +{"train_loss": 0.016930177807807922, "grad_norm": 0.12154310196638107, "global_step": 26084, "epoch": 246, "lr": 5.122983911864632e-05} +{"train_loss": 0.01697433367371559, "grad_norm": 0.07525894045829773, "global_step": 26085, "epoch": 246, "lr": 5.122687624651075e-05} +{"train_loss": 0.01616620272397995, "grad_norm": 0.06532983481884003, "global_step": 26086, "epoch": 246, "lr": 5.122391337006447e-05} +{"train_loss": 0.015237582847476006, "grad_norm": 0.07623601704835892, "global_step": 26087, "epoch": 246, "lr": 5.1220950489317875e-05} +{"train_loss": 0.01344628632068634, "grad_norm": 0.0863436684012413, "global_step": 26088, "epoch": 246, "lr": 5.12179876042814e-05} +{"train_loss": 0.01731521263718605, "grad_norm": 0.1325850486755371, "global_step": 26089, "epoch": 246, "lr": 5.1215024714965456e-05} +{"train_loss": 0.013069835491478443, "grad_norm": 0.10967385023832321, "global_step": 26090, "epoch": 246, "lr": 5.121206182138044e-05} +{"train_loss": 0.01630120351910591, "grad_norm": 0.07891993224620819, "global_step": 26091, "epoch": 246, "lr": 5.120909892353676e-05} +{"train_loss": 0.017268920317292213, "grad_norm": 0.07118865102529526, "global_step": 26092, "epoch": 246, "lr": 5.120613602144484e-05} +{"train_loss": 0.014339329674839973, "grad_norm": 0.06301508098840714, "global_step": 26093, "epoch": 246, "lr": 5.1203173115115075e-05} +{"train_loss": 0.01736295409500599, "grad_norm": 0.07671615481376648, "global_step": 26094, "epoch": 246, "lr": 5.120021020455787e-05} +{"train_loss": 0.016103502362966537, "grad_norm": 0.06039823219180107, "global_step": 26095, "epoch": 246, "lr": 5.119724728978368e-05} +{"train_loss": 0.015408024191856384, "grad_norm": 0.0711599737405777, "global_step": 26096, "epoch": 246, "lr": 5.119428437080285e-05} +{"train_loss": 0.015692470595240593, "grad_norm": 0.059127502143383026, "global_step": 26097, "epoch": 246, "lr": 5.119132144762584e-05} +{"train_loss": 0.016380494460463524, "grad_norm": 0.07056339830160141, "global_step": 26098, "epoch": 246, "lr": 5.118835852026306e-05} +{"train_loss": 0.015814954414963722, "grad_norm": 0.09466597437858582, "global_step": 26099, "epoch": 246, "lr": 5.118539558872488e-05} +{"train_loss": 0.015378741547465324, "grad_norm": 0.05112893879413605, "global_step": 26100, "epoch": 246, "lr": 5.118243265302175e-05} +{"train_loss": 0.0164092555642128, "grad_norm": 0.11448222398757935, "global_step": 26101, "epoch": 246, "lr": 5.117946971316406e-05} +{"train_loss": 0.0158355925232172, "grad_norm": 0.08740618824958801, "global_step": 26102, "epoch": 246, "lr": 5.1176506769162224e-05} +{"train_loss": 0.01656392216682434, "grad_norm": 0.145475834608078, "global_step": 26103, "epoch": 246, "lr": 5.117354382102666e-05} +{"train_loss": 0.014600634574890137, "grad_norm": 0.10628459602594376, "global_step": 26104, "epoch": 246, "lr": 5.117058086876778e-05} +{"train_loss": 0.0117585938423872, "grad_norm": 0.06905145943164825, "global_step": 26105, "epoch": 246, "lr": 5.116761791239598e-05} +{"train_loss": 0.014092628844082355, "grad_norm": 0.0938204899430275, "global_step": 26106, "epoch": 246, "lr": 5.116465495192166e-05} +{"train_loss": 0.01375825610011816, "grad_norm": 0.0745413526892662, "global_step": 26107, "epoch": 246, "lr": 5.116169198735529e-05} +{"train_loss": 0.014949736185371876, "grad_norm": 0.12938250601291656, "global_step": 26108, "epoch": 246, "lr": 5.115872901870722e-05} +{"train_loss": 0.013229265809059143, "grad_norm": 0.10096187889575958, "global_step": 26109, "epoch": 246, "lr": 5.115576604598786e-05} +{"train_loss": 0.012645965442061424, "grad_norm": 0.06371624767780304, "global_step": 26110, "epoch": 246, "lr": 5.1152803069207665e-05} +{"train_loss": 0.015553402714431286, "grad_norm": 0.13103701174259186, "global_step": 26111, "epoch": 246, "lr": 5.114984008837701e-05} +{"train_loss": 0.015520033426582813, "grad_norm": 0.08714213967323303, "global_step": 26112, "epoch": 246, "lr": 5.114687710350632e-05} +{"train_loss": 0.01629956066608429, "grad_norm": 0.08182907849550247, "global_step": 26113, "epoch": 246, "lr": 5.1143914114606006e-05} +{"train_loss": 0.017188746482133865, "grad_norm": 0.11292433738708496, "global_step": 26114, "epoch": 246, "lr": 5.114095112168648e-05} +{"train_loss": 0.0150143476203084, "grad_norm": 0.06793415546417236, "global_step": 26115, "epoch": 246, "lr": 5.113798812475814e-05} +{"train_loss": 0.01556370873004198, "grad_norm": 0.09148689359426498, "global_step": 26116, "epoch": 246, "lr": 5.1135025123831405e-05} +{"train_loss": 0.016632158309221268, "grad_norm": 0.10631536692380905, "global_step": 26117, "epoch": 246, "lr": 5.113206211891668e-05} +{"train_loss": 0.01624308153986931, "grad_norm": 0.0650547444820404, "global_step": 26118, "epoch": 246, "lr": 5.112909911002439e-05} +{"train_loss": 0.015918657183647156, "grad_norm": 0.09347577393054962, "global_step": 26119, "epoch": 246, "lr": 5.1126136097164935e-05} +{"train_loss": 0.018620498478412628, "grad_norm": 0.12807683646678925, "global_step": 26120, "epoch": 246, "lr": 5.112317308034873e-05} +{"train_loss": 0.01607927680015564, "grad_norm": 0.15691553056240082, "global_step": 26121, "epoch": 246, "lr": 5.112021005958618e-05} +{"train_loss": 0.014492863789200783, "grad_norm": 0.10721702128648758, "global_step": 26122, "epoch": 246, "lr": 5.1117247034887696e-05} +{"train_loss": 0.019694995135068893, "grad_norm": 0.10250025242567062, "global_step": 26123, "epoch": 246, "lr": 5.111428400626369e-05} +{"train_loss": 0.016362791880965233, "grad_norm": 0.13478675484657288, "global_step": 26124, "epoch": 246, "lr": 5.111132097372459e-05} +{"train_loss": 0.017229534685611725, "grad_norm": 0.08393586426973343, "global_step": 26125, "epoch": 246, "lr": 5.110835793728077e-05} +{"train_loss": 0.015245882794260979, "grad_norm": 0.13173280656337738, "global_step": 26126, "epoch": 246, "lr": 5.1105394896942674e-05} +{"train_loss": 0.016053639352321625, "grad_norm": 0.1443362832069397, "global_step": 26127, "epoch": 246, "lr": 5.110243185272071e-05} +{"train_loss": 0.014929596334695816, "grad_norm": 0.12229280173778534, "global_step": 26128, "epoch": 246, "lr": 5.109946880462526e-05} +{"train_loss": 0.016225654631853104, "grad_norm": 0.13930851221084595, "global_step": 26129, "epoch": 246, "lr": 5.109650575266677e-05} +{"train_loss": 0.015965532511472702, "grad_norm": 0.12928977608680725, "global_step": 26130, "epoch": 246, "lr": 5.109354269685562e-05} +{"train_loss": 0.015719547867774963, "grad_norm": 0.086899533867836, "global_step": 26131, "epoch": 246, "lr": 5.109057963720225e-05} +{"train_loss": 0.016407344490289688, "grad_norm": 0.0738476887345314, "global_step": 26132, "epoch": 246, "lr": 5.108761657371706e-05} +{"train_loss": 0.014487029984593391, "grad_norm": 0.12594296038150787, "global_step": 26133, "epoch": 246, "lr": 5.108465350641045e-05} +{"train_loss": 0.017098261043429375, "grad_norm": 0.11860737204551697, "global_step": 26134, "epoch": 246, "lr": 5.108169043529284e-05} +{"train_loss": 0.016408436000347137, "grad_norm": 0.15385861694812775, "global_step": 26135, "epoch": 246, "lr": 5.1078727360374636e-05} +{"train_loss": 0.015027499757707119, "grad_norm": 0.10965567827224731, "global_step": 26136, "epoch": 246, "lr": 5.107576428166627e-05} +{"train_loss": 0.013762311078608036, "grad_norm": 0.05732040852308273, "global_step": 26137, "epoch": 246, "lr": 5.107280119917812e-05} +{"train_loss": 0.01423619594424963, "grad_norm": 0.09316134452819824, "global_step": 26138, "epoch": 246, "lr": 5.106983811292061e-05} +{"train_loss": 0.017078924924135208, "grad_norm": 0.1547517478466034, "global_step": 26139, "epoch": 246, "lr": 5.106687502290418e-05} +{"train_loss": 0.013416500762104988, "grad_norm": 0.07404924184083939, "global_step": 26140, "epoch": 246, "lr": 5.1063911929139186e-05} +{"train_loss": 0.015755463391542435, "grad_norm": 0.14654242992401123, "global_step": 26141, "epoch": 246, "lr": 5.106094883163608e-05} +{"train_loss": 0.01733662188053131, "grad_norm": 0.13979248702526093, "global_step": 26142, "epoch": 246, "lr": 5.105798573040527e-05} +{"train_loss": 0.015586132183670998, "grad_norm": 0.07355143129825592, "global_step": 26143, "epoch": 246, "lr": 5.1055022625457136e-05} +{"train_loss": 0.015878304839134216, "grad_norm": 0.058994997292757034, "global_step": 26144, "epoch": 246, "lr": 5.105205951680213e-05} +{"train_loss": 0.014606384560465813, "grad_norm": 0.09262882173061371, "global_step": 26145, "epoch": 246, "lr": 5.104909640445064e-05} +{"train_loss": 0.015104606747627258, "grad_norm": 0.1253623217344284, "global_step": 26146, "epoch": 246, "lr": 5.104613328841308e-05} +{"train_loss": 0.018542177975177765, "grad_norm": 0.0751931220293045, "global_step": 26147, "epoch": 246, "lr": 5.104317016869985e-05} +{"train_loss": 0.014995964244008064, "grad_norm": 0.07642759382724762, "global_step": 26148, "epoch": 246, "lr": 5.10402070453214e-05} +{"train_loss": 0.015961196273565292, "grad_norm": 0.1347586214542389, "global_step": 26149, "epoch": 246, "lr": 5.10372439182881e-05} +{"train_loss": 0.01775388792157173, "grad_norm": 0.07091248035430908, "global_step": 26150, "epoch": 246, "lr": 5.1034280787610366e-05} +{"train_loss": 0.014467127621173859, "grad_norm": 0.06303033232688904, "global_step": 26151, "epoch": 246, "lr": 5.103131765329865e-05} +{"train_loss": 0.016483094543218613, "grad_norm": 0.07622997462749481, "global_step": 26152, "epoch": 246, "lr": 5.1028354515363294e-05} +{"train_loss": 0.014803964644670486, "grad_norm": 0.07281231880187988, "global_step": 26153, "epoch": 246, "lr": 5.102539137381478e-05} +{"train_loss": 0.02085386961698532, "grad_norm": 0.09520228952169418, "global_step": 26154, "epoch": 246, "lr": 5.102242822866347e-05} +{"train_loss": 0.015616367571055889, "grad_norm": 0.19989584386348724, "global_step": 26155, "epoch": 246, "lr": 5.1019465079919794e-05} +{"train_loss": 0.01484999991953373, "grad_norm": 0.08055958896875381, "global_step": 26156, "epoch": 246, "lr": 5.101650192759416e-05} +{"train_loss": 0.014615718275308609, "grad_norm": 0.06144820153713226, "global_step": 26157, "epoch": 246, "lr": 5.101353877169699e-05} +{"train_loss": 0.014720488339662552, "grad_norm": 0.09915363788604736, "global_step": 26158, "epoch": 246, "lr": 5.1010575612238674e-05} +{"train_loss": 0.01644570752978325, "grad_norm": 0.17315329611301422, "global_step": 26159, "epoch": 246, "lr": 5.100761244922964e-05} +{"train_loss": 0.017099758610129356, "grad_norm": 0.07058950513601303, "global_step": 26160, "epoch": 246, "lr": 5.1004649282680296e-05} +{"train_loss": 0.016159478574991226, "grad_norm": 0.12103955447673798, "global_step": 26161, "epoch": 246, "lr": 5.100168611260104e-05} +{"train_loss": 0.01615806668996811, "grad_norm": 0.11104346811771393, "global_step": 26162, "epoch": 246, "lr": 5.09987229390023e-05} +{"train_loss": 0.018368428573012352, "grad_norm": 0.08969438821077347, "global_step": 26163, "epoch": 246, "lr": 5.099575976189449e-05} +{"train_loss": 0.013997742906212807, "grad_norm": 0.060198258608579636, "global_step": 26164, "epoch": 246, "lr": 5.0992796581288e-05} +{"train_loss": 0.014807740226387978, "grad_norm": 0.06487943977117538, "global_step": 26165, "epoch": 246, "lr": 5.0989833397193266e-05} +{"train_loss": 0.015903010964393616, "grad_norm": 0.09745623916387558, "global_step": 26166, "epoch": 246, "lr": 5.0986870209620694e-05} +{"train_loss": 0.014602325856685638, "grad_norm": 0.06285613030195236, "global_step": 26167, "epoch": 246, "lr": 5.098390701858067e-05} +{"train_loss": 0.0153232142329216, "grad_norm": 0.0613420233130455, "global_step": 26168, "epoch": 246, "lr": 5.0980943824083636e-05} +{"train_loss": 0.012609679251909256, "grad_norm": 0.06919606029987335, "global_step": 26169, "epoch": 246, "lr": 5.0977980626139986e-05} +{"train_loss": 0.01531766727566719, "grad_norm": 0.06280852854251862, "global_step": 26170, "epoch": 246, "lr": 5.097501742476014e-05} +{"train_loss": 0.014820759184658527, "grad_norm": 0.05320243537425995, "global_step": 26171, "epoch": 246, "lr": 5.0972054219954514e-05} +{"train_loss": 0.015080992132425308, "grad_norm": 0.08879145979881287, "global_step": 26172, "epoch": 246, "lr": 5.09690910117335e-05} +{"train_loss": 0.015414485707879066, "grad_norm": 0.06770657002925873, "global_step": 26173, "epoch": 246, "lr": 5.096612780010753e-05} +{"train_loss": 0.014066983014345169, "grad_norm": 0.07024774700403214, "global_step": 26174, "epoch": 246, "lr": 5.096316458508701e-05} +{"train_loss": 0.014473006129264832, "grad_norm": 0.11621100455522537, "global_step": 26175, "epoch": 246, "lr": 5.096020136668235e-05} +{"train_loss": 0.017117809504270554, "grad_norm": 0.13688044250011444, "global_step": 26176, "epoch": 246, "lr": 5.0957238144903943e-05} +{"train_loss": 0.012645525857806206, "grad_norm": 0.10076295584440231, "global_step": 26177, "epoch": 246, "lr": 5.0954274919762235e-05} +{"train_loss": 0.01631172187626362, "grad_norm": 0.14225232601165771, "global_step": 26178, "epoch": 246, "lr": 5.095131169126761e-05} +{"train_loss": 0.016448430716991425, "grad_norm": 0.10030841082334518, "global_step": 26179, "epoch": 246, "lr": 5.094834845943049e-05} +{"train_loss": 0.017598967999219894, "grad_norm": 0.09523827582597733, "global_step": 26180, "epoch": 246, "lr": 5.09453852242613e-05} +{"train_loss": 0.015618733016176606, "grad_norm": 0.1156000867486, "global_step": 26181, "epoch": 246, "lr": 5.094242198577042e-05, "val_loss": 0.01952778361737728} +{"train_loss": 0.01430467888712883, "grad_norm": 0.0732257068157196, "global_step": 26182, "epoch": 247, "lr": 5.093945874396827e-05} +{"train_loss": 0.017148196697235107, "grad_norm": 0.1349712610244751, "global_step": 26183, "epoch": 247, "lr": 5.093649549886529e-05} +{"train_loss": 0.013166945427656174, "grad_norm": 0.07240191847085953, "global_step": 26184, "epoch": 247, "lr": 5.0933532250471875e-05} +{"train_loss": 0.014187333174049854, "grad_norm": 0.1491444706916809, "global_step": 26185, "epoch": 247, "lr": 5.0930568998798425e-05} +{"train_loss": 0.015553418546915054, "grad_norm": 0.10312175005674362, "global_step": 26186, "epoch": 247, "lr": 5.092760574385535e-05} +{"train_loss": 0.016609249636530876, "grad_norm": 0.06475084275007248, "global_step": 26187, "epoch": 247, "lr": 5.0924642485653095e-05} +{"train_loss": 0.01713465340435505, "grad_norm": 0.06941218674182892, "global_step": 26188, "epoch": 247, "lr": 5.092167922420203e-05} +{"train_loss": 0.018823161721229553, "grad_norm": 0.06015859916806221, "global_step": 26189, "epoch": 247, "lr": 5.09187159595126e-05} +{"train_loss": 0.014581263065338135, "grad_norm": 0.09270542114973068, "global_step": 26190, "epoch": 247, "lr": 5.09157526915952e-05} +{"train_loss": 0.01609659567475319, "grad_norm": 0.21139740943908691, "global_step": 26191, "epoch": 247, "lr": 5.091278942046022e-05} +{"train_loss": 0.014386530965566635, "grad_norm": 0.0663372352719307, "global_step": 26192, "epoch": 247, "lr": 5.090982614611812e-05} +{"train_loss": 0.013251245953142643, "grad_norm": 0.055811576545238495, "global_step": 26193, "epoch": 247, "lr": 5.090686286857927e-05} +{"train_loss": 0.012663298286497593, "grad_norm": 0.17533446848392487, "global_step": 26194, "epoch": 247, "lr": 5.090389958785411e-05} +{"train_loss": 0.017708752304315567, "grad_norm": 0.08120795339345932, "global_step": 26195, "epoch": 247, "lr": 5.090093630395304e-05} +{"train_loss": 0.016520852223038673, "grad_norm": 0.07209377735853195, "global_step": 26196, "epoch": 247, "lr": 5.0897973016886455e-05} +{"train_loss": 0.013857146725058556, "grad_norm": 0.06767407059669495, "global_step": 26197, "epoch": 247, "lr": 5.08950097266648e-05} +{"train_loss": 0.015371355228126049, "grad_norm": 0.06083713099360466, "global_step": 26198, "epoch": 247, "lr": 5.089204643329847e-05} +{"train_loss": 0.01652674749493599, "grad_norm": 0.12041877210140228, "global_step": 26199, "epoch": 247, "lr": 5.0889083136797875e-05} +{"train_loss": 0.01906406134366989, "grad_norm": 0.07944535464048386, "global_step": 26200, "epoch": 247, "lr": 5.088611983717343e-05} +{"train_loss": 0.014636832289397717, "grad_norm": 0.05115479230880737, "global_step": 26201, "epoch": 247, "lr": 5.088315653443554e-05} +{"train_loss": 0.012280400842428207, "grad_norm": 0.08780358731746674, "global_step": 26202, "epoch": 247, "lr": 5.088019322859463e-05} +{"train_loss": 0.01675630733370781, "grad_norm": 0.1197672113776207, "global_step": 26203, "epoch": 247, "lr": 5.087722991966108e-05} +{"train_loss": 0.012963377870619297, "grad_norm": 0.08195440471172333, "global_step": 26204, "epoch": 247, "lr": 5.087426660764536e-05} +{"train_loss": 0.015617960132658482, "grad_norm": 0.057245269417762756, "global_step": 26205, "epoch": 247, "lr": 5.087130329255782e-05} +{"train_loss": 0.017277607694268227, "grad_norm": 0.07968708127737045, "global_step": 26206, "epoch": 247, "lr": 5.086833997440891e-05} +{"train_loss": 0.01500934362411499, "grad_norm": 0.060506172478199005, "global_step": 26207, "epoch": 247, "lr": 5.086537665320904e-05} +{"train_loss": 0.013590976595878601, "grad_norm": 0.11699379235506058, "global_step": 26208, "epoch": 247, "lr": 5.0862413328968594e-05} +{"train_loss": 0.015261001884937286, "grad_norm": 0.08538370579481125, "global_step": 26209, "epoch": 247, "lr": 5.0859450001698016e-05} +{"train_loss": 0.014579365961253643, "grad_norm": 0.07019560784101486, "global_step": 26210, "epoch": 247, "lr": 5.08564866714077e-05} +{"train_loss": 0.015603943727910519, "grad_norm": 0.05794377997517586, "global_step": 26211, "epoch": 247, "lr": 5.0853523338108054e-05} +{"train_loss": 0.018374165520071983, "grad_norm": 0.08563631772994995, "global_step": 26212, "epoch": 247, "lr": 5.085056000180951e-05} +{"train_loss": 0.014822511933743954, "grad_norm": 0.054084692150354385, "global_step": 26213, "epoch": 247, "lr": 5.0847596662522466e-05} +{"train_loss": 0.015272709541022778, "grad_norm": 0.059727251529693604, "global_step": 26214, "epoch": 247, "lr": 5.084463332025734e-05} +{"train_loss": 0.017517561092972755, "grad_norm": 0.06844289600849152, "global_step": 26215, "epoch": 247, "lr": 5.084166997502452e-05} +{"train_loss": 0.016606029123067856, "grad_norm": 0.15045616030693054, "global_step": 26216, "epoch": 247, "lr": 5.083870662683446e-05} +{"train_loss": 0.015321705490350723, "grad_norm": 0.05526377260684967, "global_step": 26217, "epoch": 247, "lr": 5.083574327569756e-05} +{"train_loss": 0.016066845506429672, "grad_norm": 0.06837356835603714, "global_step": 26218, "epoch": 247, "lr": 5.083277992162419e-05} +{"train_loss": 0.013374668546020985, "grad_norm": 0.04673825204372406, "global_step": 26219, "epoch": 247, "lr": 5.082981656462481e-05} +{"train_loss": 0.014537282288074493, "grad_norm": 0.08679737895727158, "global_step": 26220, "epoch": 247, "lr": 5.082685320470981e-05} +{"train_loss": 0.016697533428668976, "grad_norm": 0.1361725628376007, "global_step": 26221, "epoch": 247, "lr": 5.082388984188962e-05} +{"train_loss": 0.014992138370871544, "grad_norm": 0.1281212866306305, "global_step": 26222, "epoch": 247, "lr": 5.0820926476174626e-05} +{"train_loss": 0.015507321804761887, "grad_norm": 0.0565948560833931, "global_step": 26223, "epoch": 247, "lr": 5.081796310757525e-05} +{"train_loss": 0.013261985965073109, "grad_norm": 0.07658813893795013, "global_step": 26224, "epoch": 247, "lr": 5.081499973610193e-05} +{"train_loss": 0.014729579910635948, "grad_norm": 0.057685501873493195, "global_step": 26225, "epoch": 247, "lr": 5.0812036361765027e-05} +{"train_loss": 0.01571130007505417, "grad_norm": 0.08853229880332947, "global_step": 26226, "epoch": 247, "lr": 5.080907298457499e-05} +{"train_loss": 0.018876276910305023, "grad_norm": 0.13521088659763336, "global_step": 26227, "epoch": 247, "lr": 5.080610960454223e-05} +{"train_loss": 0.013494882732629776, "grad_norm": 0.06314419955015182, "global_step": 26228, "epoch": 247, "lr": 5.080314622167714e-05} +{"train_loss": 0.015513394959270954, "grad_norm": 0.10507135838270187, "global_step": 26229, "epoch": 247, "lr": 5.0800182835990165e-05} +{"train_loss": 0.01474179420620203, "grad_norm": 0.07886174321174622, "global_step": 26230, "epoch": 247, "lr": 5.079721944749168e-05} +{"train_loss": 0.015988674014806747, "grad_norm": 0.10126139223575592, "global_step": 26231, "epoch": 247, "lr": 5.0794256056192114e-05} +{"train_loss": 0.013558004051446915, "grad_norm": 0.10629529505968094, "global_step": 26232, "epoch": 247, "lr": 5.0791292662101864e-05} +{"train_loss": 0.015717318281531334, "grad_norm": 0.1508687138557434, "global_step": 26233, "epoch": 247, "lr": 5.078832926523138e-05} +{"train_loss": 0.016274504363536835, "grad_norm": 0.07713348418474197, "global_step": 26234, "epoch": 247, "lr": 5.078536586559104e-05} +{"train_loss": 0.015639591962099075, "grad_norm": 0.0697270929813385, "global_step": 26235, "epoch": 247, "lr": 5.078240246319126e-05} +{"train_loss": 0.013575381599366665, "grad_norm": 0.06438083946704865, "global_step": 26236, "epoch": 247, "lr": 5.0779439058042476e-05} +{"train_loss": 0.013618573546409607, "grad_norm": 0.061825498938560486, "global_step": 26237, "epoch": 247, "lr": 5.077647565015506e-05} +{"train_loss": 0.01632542535662651, "grad_norm": 0.11312565952539444, "global_step": 26238, "epoch": 247, "lr": 5.077351223953946e-05} +{"train_loss": 0.014578958041965961, "grad_norm": 0.0655512586236, "global_step": 26239, "epoch": 247, "lr": 5.077054882620606e-05} +{"train_loss": 0.017103997990489006, "grad_norm": 0.08679782599210739, "global_step": 26240, "epoch": 247, "lr": 5.076758541016531e-05} +{"train_loss": 0.021148908883333206, "grad_norm": 0.13622508943080902, "global_step": 26241, "epoch": 247, "lr": 5.0764621991427585e-05} +{"train_loss": 0.019577963277697563, "grad_norm": 0.1343744546175003, "global_step": 26242, "epoch": 247, "lr": 5.076165857000331e-05} +{"train_loss": 0.01735284924507141, "grad_norm": 0.08158565312623978, "global_step": 26243, "epoch": 247, "lr": 5.07586951459029e-05} +{"train_loss": 0.015628580003976822, "grad_norm": 0.13755081593990326, "global_step": 26244, "epoch": 247, "lr": 5.0755731719136756e-05} +{"train_loss": 0.013709395192563534, "grad_norm": 0.12281889468431473, "global_step": 26245, "epoch": 247, "lr": 5.075276828971531e-05} +{"train_loss": 0.01634974032640457, "grad_norm": 0.07817269116640091, "global_step": 26246, "epoch": 247, "lr": 5.0749804857648964e-05} +{"train_loss": 0.01468614675104618, "grad_norm": 0.12094184011220932, "global_step": 26247, "epoch": 247, "lr": 5.074684142294812e-05} +{"train_loss": 0.015547447837889194, "grad_norm": 0.05646174028515816, "global_step": 26248, "epoch": 247, "lr": 5.074387798562322e-05} +{"train_loss": 0.015102101489901543, "grad_norm": 0.07790161669254303, "global_step": 26249, "epoch": 247, "lr": 5.074091454568464e-05} +{"train_loss": 0.015842357650399208, "grad_norm": 0.06783425807952881, "global_step": 26250, "epoch": 247, "lr": 5.073795110314281e-05} +{"train_loss": 0.015441518276929855, "grad_norm": 0.08188223838806152, "global_step": 26251, "epoch": 247, "lr": 5.073498765800814e-05} +{"train_loss": 0.015187632292509079, "grad_norm": 0.17722095549106598, "global_step": 26252, "epoch": 247, "lr": 5.0732024210291054e-05} +{"train_loss": 0.017867106944322586, "grad_norm": 0.08078020811080933, "global_step": 26253, "epoch": 247, "lr": 5.072906076000194e-05} +{"train_loss": 0.016115181148052216, "grad_norm": 0.09617321193218231, "global_step": 26254, "epoch": 247, "lr": 5.072609730715123e-05} +{"train_loss": 0.012811329215765, "grad_norm": 0.060750190168619156, "global_step": 26255, "epoch": 247, "lr": 5.072313385174933e-05} +{"train_loss": 0.01605241186916828, "grad_norm": 0.05900431051850319, "global_step": 26256, "epoch": 247, "lr": 5.072017039380663e-05} +{"train_loss": 0.01747908629477024, "grad_norm": 0.07505884021520615, "global_step": 26257, "epoch": 247, "lr": 5.07172069333336e-05} +{"train_loss": 0.013760577887296677, "grad_norm": 0.07255176454782486, "global_step": 26258, "epoch": 247, "lr": 5.0714243470340605e-05} +{"train_loss": 0.01884811744093895, "grad_norm": 0.06384256482124329, "global_step": 26259, "epoch": 247, "lr": 5.071128000483806e-05} +{"train_loss": 0.014759370125830173, "grad_norm": 0.09712207317352295, "global_step": 26260, "epoch": 247, "lr": 5.07083165368364e-05} +{"train_loss": 0.0164859090000391, "grad_norm": 0.0723397433757782, "global_step": 26261, "epoch": 247, "lr": 5.0705353066346006e-05} +{"train_loss": 0.014003693126142025, "grad_norm": 0.10748021304607391, "global_step": 26262, "epoch": 247, "lr": 5.070238959337732e-05} +{"train_loss": 0.015107598155736923, "grad_norm": 0.06670370697975159, "global_step": 26263, "epoch": 247, "lr": 5.069942611794074e-05} +{"train_loss": 0.015495449304580688, "grad_norm": 0.057824525982141495, "global_step": 26264, "epoch": 247, "lr": 5.0696462640046685e-05} +{"train_loss": 0.014977185986936092, "grad_norm": 0.19512739777565002, "global_step": 26265, "epoch": 247, "lr": 5.069349915970556e-05} +{"train_loss": 0.01687423512339592, "grad_norm": 0.09061513096094131, "global_step": 26266, "epoch": 247, "lr": 5.0690535676927773e-05} +{"train_loss": 0.012514847330749035, "grad_norm": 0.07927603274583817, "global_step": 26267, "epoch": 247, "lr": 5.0687572191723755e-05} +{"train_loss": 0.015024644322693348, "grad_norm": 0.09575961530208588, "global_step": 26268, "epoch": 247, "lr": 5.068460870410391e-05} +{"train_loss": 0.017217764630913734, "grad_norm": 0.07473451644182205, "global_step": 26269, "epoch": 247, "lr": 5.068164521407864e-05} +{"train_loss": 0.016152624040842056, "grad_norm": 0.11763516068458557, "global_step": 26270, "epoch": 247, "lr": 5.0678681721658364e-05} +{"train_loss": 0.01332327164709568, "grad_norm": 0.07260815054178238, "global_step": 26271, "epoch": 247, "lr": 5.0675718226853485e-05} +{"train_loss": 0.014182154089212418, "grad_norm": 0.13369375467300415, "global_step": 26272, "epoch": 247, "lr": 5.067275472967445e-05} +{"train_loss": 0.015436409041285515, "grad_norm": 0.0703812688589096, "global_step": 26273, "epoch": 247, "lr": 5.0669791230131624e-05} +{"train_loss": 0.013711404986679554, "grad_norm": 0.07302822172641754, "global_step": 26274, "epoch": 247, "lr": 5.066682772823548e-05} +{"train_loss": 0.01725030317902565, "grad_norm": 0.1090417355298996, "global_step": 26275, "epoch": 247, "lr": 5.066386422399636e-05} +{"train_loss": 0.016394149512052536, "grad_norm": 0.051123205572366714, "global_step": 26276, "epoch": 247, "lr": 5.066090071742471e-05} +{"train_loss": 0.01711871102452278, "grad_norm": 0.1563718318939209, "global_step": 26277, "epoch": 247, "lr": 5.0657937208530956e-05} +{"train_loss": 0.017936818301677704, "grad_norm": 0.19020847976207733, "global_step": 26278, "epoch": 247, "lr": 5.065497369732549e-05} +{"train_loss": 0.01684824377298355, "grad_norm": 0.09052400290966034, "global_step": 26279, "epoch": 247, "lr": 5.065201018381873e-05} +{"train_loss": 0.016727736219763756, "grad_norm": 0.11512335389852524, "global_step": 26280, "epoch": 247, "lr": 5.0649046668021094e-05} +{"train_loss": 0.013769673183560371, "grad_norm": 0.0903083086013794, "global_step": 26281, "epoch": 247, "lr": 5.064608314994299e-05} +{"train_loss": 0.018219266086816788, "grad_norm": 0.056029971688985825, "global_step": 26282, "epoch": 247, "lr": 5.064311962959483e-05} +{"train_loss": 0.01677786558866501, "grad_norm": 0.08285494893789291, "global_step": 26283, "epoch": 247, "lr": 5.064015610698703e-05} +{"train_loss": 0.013688642531633377, "grad_norm": 0.07271703332662582, "global_step": 26284, "epoch": 247, "lr": 5.063719258213e-05} +{"train_loss": 0.013847382739186287, "grad_norm": 0.13971169292926788, "global_step": 26285, "epoch": 247, "lr": 5.063422905503413e-05} +{"train_loss": 0.014608906581997871, "grad_norm": 0.055684372782707214, "global_step": 26286, "epoch": 247, "lr": 5.063126552570989e-05} +{"train_loss": 0.015619237353708947, "grad_norm": 0.08118250966072083, "global_step": 26287, "epoch": 247, "lr": 5.062830199416764e-05, "val_loss": 0.018997374922037125} +{"train_loss": 0.01422164123505354, "grad_norm": 0.08509894460439682, "global_step": 26288, "epoch": 248, "lr": 5.0625338460417805e-05} +{"train_loss": 0.018381591886281967, "grad_norm": 0.0985139012336731, "global_step": 26289, "epoch": 248, "lr": 5.0622374924470814e-05} +{"train_loss": 0.01680673658847809, "grad_norm": 0.06612999737262726, "global_step": 26290, "epoch": 248, "lr": 5.0619411386337055e-05} +{"train_loss": 0.01723099686205387, "grad_norm": 0.11125237494707108, "global_step": 26291, "epoch": 248, "lr": 5.061644784602696e-05} +{"train_loss": 0.01774683967232704, "grad_norm": 0.10110421478748322, "global_step": 26292, "epoch": 248, "lr": 5.061348430355094e-05} +{"train_loss": 0.013797794468700886, "grad_norm": 0.0862487256526947, "global_step": 26293, "epoch": 248, "lr": 5.06105207589194e-05} +{"train_loss": 0.01443635392934084, "grad_norm": 0.09468182921409607, "global_step": 26294, "epoch": 248, "lr": 5.0607557212142765e-05} +{"train_loss": 0.016239464282989502, "grad_norm": 0.0904146358370781, "global_step": 26295, "epoch": 248, "lr": 5.060459366323143e-05} +{"train_loss": 0.0122663713991642, "grad_norm": 0.08619300276041031, "global_step": 26296, "epoch": 248, "lr": 5.060163011219581e-05} +{"train_loss": 0.016298707574605942, "grad_norm": 0.10123294591903687, "global_step": 26297, "epoch": 248, "lr": 5.059866655904633e-05} +{"train_loss": 0.01368292048573494, "grad_norm": 0.07118801772594452, "global_step": 26298, "epoch": 248, "lr": 5.0595703003793395e-05} +{"train_loss": 0.020406799390912056, "grad_norm": 0.10067152231931686, "global_step": 26299, "epoch": 248, "lr": 5.0592739446447425e-05} +{"train_loss": 0.017538046464323997, "grad_norm": 0.07698585838079453, "global_step": 26300, "epoch": 248, "lr": 5.05897758870188e-05} +{"train_loss": 0.014015201479196548, "grad_norm": 0.08651898056268692, "global_step": 26301, "epoch": 248, "lr": 5.0586812325517995e-05} +{"train_loss": 0.014653490856289864, "grad_norm": 0.06297852843999863, "global_step": 26302, "epoch": 248, "lr": 5.058384876195535e-05} +{"train_loss": 0.016724558547139168, "grad_norm": 0.04915525019168854, "global_step": 26303, "epoch": 248, "lr": 5.0580885196341344e-05} +{"train_loss": 0.013628948479890823, "grad_norm": 0.1063367947936058, "global_step": 26304, "epoch": 248, "lr": 5.057792162868635e-05} +{"train_loss": 0.016649747267365456, "grad_norm": 0.1283130794763565, "global_step": 26305, "epoch": 248, "lr": 5.0574958059000786e-05} +{"train_loss": 0.014414122328162193, "grad_norm": 0.0747404545545578, "global_step": 26306, "epoch": 248, "lr": 5.057199448729508e-05} +{"train_loss": 0.014597244560718536, "grad_norm": 0.1165444552898407, "global_step": 26307, "epoch": 248, "lr": 5.056903091357963e-05} +{"train_loss": 0.014027606695890427, "grad_norm": 0.1028573289513588, "global_step": 26308, "epoch": 248, "lr": 5.0566067337864854e-05} +{"train_loss": 0.017310233786702156, "grad_norm": 0.055122923105955124, "global_step": 26309, "epoch": 248, "lr": 5.056310376016116e-05} +{"train_loss": 0.013238056562840939, "grad_norm": 0.0807059183716774, "global_step": 26310, "epoch": 248, "lr": 5.0560140180478964e-05} +{"train_loss": 0.01566278748214245, "grad_norm": 0.07515102624893188, "global_step": 26311, "epoch": 248, "lr": 5.0557176598828684e-05} +{"train_loss": 0.014977161772549152, "grad_norm": 0.06895024329423904, "global_step": 26312, "epoch": 248, "lr": 5.055421301522072e-05} +{"train_loss": 0.014561645686626434, "grad_norm": 0.08822675049304962, "global_step": 26313, "epoch": 248, "lr": 5.0551249429665505e-05} +{"train_loss": 0.01744324155151844, "grad_norm": 0.10576077550649643, "global_step": 26314, "epoch": 248, "lr": 5.0548285842173415e-05} +{"train_loss": 0.017836907878518105, "grad_norm": 0.11163023114204407, "global_step": 26315, "epoch": 248, "lr": 5.0545322252754915e-05} +{"train_loss": 0.016460556536912918, "grad_norm": 0.1603328138589859, "global_step": 26316, "epoch": 248, "lr": 5.0542358661420386e-05} +{"train_loss": 0.017338059842586517, "grad_norm": 0.06520581245422363, "global_step": 26317, "epoch": 248, "lr": 5.0539395068180215e-05} +{"train_loss": 0.014524197205901146, "grad_norm": 0.07238636165857315, "global_step": 26318, "epoch": 248, "lr": 5.053643147304487e-05} +{"train_loss": 0.017330579459667206, "grad_norm": 0.06822309643030167, "global_step": 26319, "epoch": 248, "lr": 5.053346787602473e-05} +{"train_loss": 0.01372196339070797, "grad_norm": 0.0637284442782402, "global_step": 26320, "epoch": 248, "lr": 5.053050427713023e-05} +{"train_loss": 0.01674325205385685, "grad_norm": 0.09053182601928711, "global_step": 26321, "epoch": 248, "lr": 5.0527540676371756e-05} +{"train_loss": 0.013587769120931625, "grad_norm": 0.10826021432876587, "global_step": 26322, "epoch": 248, "lr": 5.0524577073759735e-05} +{"train_loss": 0.017316311597824097, "grad_norm": 0.06296470016241074, "global_step": 26323, "epoch": 248, "lr": 5.0521613469304584e-05} +{"train_loss": 0.014284754171967506, "grad_norm": 0.07552313804626465, "global_step": 26324, "epoch": 248, "lr": 5.05186498630167e-05} +{"train_loss": 0.01592295616865158, "grad_norm": 0.07300710678100586, "global_step": 26325, "epoch": 248, "lr": 5.051568625490651e-05} +{"train_loss": 0.01526311133056879, "grad_norm": 0.07880520075559616, "global_step": 26326, "epoch": 248, "lr": 5.0512722644984434e-05} +{"train_loss": 0.01457065436989069, "grad_norm": 0.0903935581445694, "global_step": 26327, "epoch": 248, "lr": 5.050975903326086e-05} +{"train_loss": 0.015564286150038242, "grad_norm": 0.07901360839605331, "global_step": 26328, "epoch": 248, "lr": 5.0506795419746224e-05} +{"train_loss": 0.013999562710523605, "grad_norm": 0.14745540916919708, "global_step": 26329, "epoch": 248, "lr": 5.050383180445091e-05} +{"train_loss": 0.01572228968143463, "grad_norm": 0.09863252192735672, "global_step": 26330, "epoch": 248, "lr": 5.0500868187385374e-05} +{"train_loss": 0.015772737562656403, "grad_norm": 0.07126528024673462, "global_step": 26331, "epoch": 248, "lr": 5.0497904568559984e-05} +{"train_loss": 0.01576242409646511, "grad_norm": 0.09379930794239044, "global_step": 26332, "epoch": 248, "lr": 5.049494094798518e-05} +{"train_loss": 0.014561889693140984, "grad_norm": 0.06635984778404236, "global_step": 26333, "epoch": 248, "lr": 5.049197732567138e-05} +{"train_loss": 0.01565883308649063, "grad_norm": 0.12367615103721619, "global_step": 26334, "epoch": 248, "lr": 5.0489013701628974e-05} +{"train_loss": 0.016463547945022583, "grad_norm": 0.07413572072982788, "global_step": 26335, "epoch": 248, "lr": 5.0486050075868396e-05} +{"train_loss": 0.016613278537988663, "grad_norm": 0.06768643110990524, "global_step": 26336, "epoch": 248, "lr": 5.048308644840004e-05} +{"train_loss": 0.016059182584285736, "grad_norm": 0.15501753985881805, "global_step": 26337, "epoch": 248, "lr": 5.0480122819234334e-05} +{"train_loss": 0.016673289239406586, "grad_norm": 0.06159656494855881, "global_step": 26338, "epoch": 248, "lr": 5.0477159188381686e-05} +{"train_loss": 0.01507692039012909, "grad_norm": 0.06339003890752792, "global_step": 26339, "epoch": 248, "lr": 5.047419555585251e-05} +{"train_loss": 0.014530216343700886, "grad_norm": 0.05519767105579376, "global_step": 26340, "epoch": 248, "lr": 5.047123192165721e-05} +{"train_loss": 0.01584324985742569, "grad_norm": 0.15638810396194458, "global_step": 26341, "epoch": 248, "lr": 5.0468268285806194e-05} +{"train_loss": 0.015092987567186356, "grad_norm": 0.12310511618852615, "global_step": 26342, "epoch": 248, "lr": 5.046530464830992e-05} +{"train_loss": 0.015477822162210941, "grad_norm": 0.11122315376996994, "global_step": 26343, "epoch": 248, "lr": 5.046234100917874e-05} +{"train_loss": 0.015419588424265385, "grad_norm": 0.19327782094478607, "global_step": 26344, "epoch": 248, "lr": 5.0459377368423096e-05} +{"train_loss": 0.015686234459280968, "grad_norm": 0.07175862789154053, "global_step": 26345, "epoch": 248, "lr": 5.045641372605343e-05} +{"train_loss": 0.013276062905788422, "grad_norm": 0.10915056616067886, "global_step": 26346, "epoch": 248, "lr": 5.045345008208009e-05} +{"train_loss": 0.018150584772229195, "grad_norm": 0.05847331881523132, "global_step": 26347, "epoch": 248, "lr": 5.045048643651353e-05} +{"train_loss": 0.012629497796297073, "grad_norm": 0.07525064051151276, "global_step": 26348, "epoch": 248, "lr": 5.044752278936417e-05} +{"train_loss": 0.013605592772364616, "grad_norm": 0.1252431720495224, "global_step": 26349, "epoch": 248, "lr": 5.04445591406424e-05} +{"train_loss": 0.013346143066883087, "grad_norm": 0.10633272677659988, "global_step": 26350, "epoch": 248, "lr": 5.044159549035865e-05} +{"train_loss": 0.01674102246761322, "grad_norm": 0.09793001413345337, "global_step": 26351, "epoch": 248, "lr": 5.043863183852332e-05} +{"train_loss": 0.015621719881892204, "grad_norm": 0.10658065974712372, "global_step": 26352, "epoch": 248, "lr": 5.0435668185146835e-05} +{"train_loss": 0.017818251624703407, "grad_norm": 0.08465307950973511, "global_step": 26353, "epoch": 248, "lr": 5.043270453023958e-05} +{"train_loss": 0.015033317729830742, "grad_norm": 0.06914734095335007, "global_step": 26354, "epoch": 248, "lr": 5.0429740873812015e-05} +{"train_loss": 0.016489088535308838, "grad_norm": 0.12684598565101624, "global_step": 26355, "epoch": 248, "lr": 5.042677721587452e-05} +{"train_loss": 0.015594140626490116, "grad_norm": 0.09958824515342712, "global_step": 26356, "epoch": 248, "lr": 5.0423813556437504e-05} +{"train_loss": 0.0128003666177392, "grad_norm": 0.09173853695392609, "global_step": 26357, "epoch": 248, "lr": 5.042084989551141e-05} +{"train_loss": 0.014420436695218086, "grad_norm": 0.10182053595781326, "global_step": 26358, "epoch": 248, "lr": 5.041788623310662e-05} +{"train_loss": 0.018374893814325333, "grad_norm": 0.07320109009742737, "global_step": 26359, "epoch": 248, "lr": 5.0414922569233555e-05} +{"train_loss": 0.01589461974799633, "grad_norm": 0.12967582046985626, "global_step": 26360, "epoch": 248, "lr": 5.041195890390265e-05} +{"train_loss": 0.014715764671564102, "grad_norm": 0.050910938531160355, "global_step": 26361, "epoch": 248, "lr": 5.040899523712429e-05} +{"train_loss": 0.01671868935227394, "grad_norm": 0.06865539401769638, "global_step": 26362, "epoch": 248, "lr": 5.0406031568908896e-05} +{"train_loss": 0.015309474430978298, "grad_norm": 0.06635241955518723, "global_step": 26363, "epoch": 248, "lr": 5.040306789926689e-05} +{"train_loss": 0.01677120476961136, "grad_norm": 0.06762445718050003, "global_step": 26364, "epoch": 248, "lr": 5.040010422820868e-05} +{"train_loss": 0.015558121725916862, "grad_norm": 0.07934700697660446, "global_step": 26365, "epoch": 248, "lr": 5.039714055574466e-05} +{"train_loss": 0.015265138819813728, "grad_norm": 0.0794905498623848, "global_step": 26366, "epoch": 248, "lr": 5.039417688188528e-05} +{"train_loss": 0.014944193884730339, "grad_norm": 0.08561873435974121, "global_step": 26367, "epoch": 248, "lr": 5.039121320664093e-05} +{"train_loss": 0.016593467444181442, "grad_norm": 0.11480777710676193, "global_step": 26368, "epoch": 248, "lr": 5.038824953002201e-05} +{"train_loss": 0.013652430847287178, "grad_norm": 0.10178609937429428, "global_step": 26369, "epoch": 248, "lr": 5.038528585203898e-05} +{"train_loss": 0.01750427484512329, "grad_norm": 0.06410206109285355, "global_step": 26370, "epoch": 248, "lr": 5.03823221727022e-05} +{"train_loss": 0.01542702130973339, "grad_norm": 0.13902834057807922, "global_step": 26371, "epoch": 248, "lr": 5.037935849202211e-05} +{"train_loss": 0.012883704155683517, "grad_norm": 0.055898454040288925, "global_step": 26372, "epoch": 248, "lr": 5.037639481000913e-05} +{"train_loss": 0.014589102938771248, "grad_norm": 0.06514007598161697, "global_step": 26373, "epoch": 248, "lr": 5.0373431126673655e-05} +{"train_loss": 0.017557872459292412, "grad_norm": 0.13913670182228088, "global_step": 26374, "epoch": 248, "lr": 5.0370467442026113e-05} +{"train_loss": 0.015290151350200176, "grad_norm": 0.09121206402778625, "global_step": 26375, "epoch": 248, "lr": 5.036750375607689e-05} +{"train_loss": 0.01655820943415165, "grad_norm": 0.06392048299312592, "global_step": 26376, "epoch": 248, "lr": 5.036454006883644e-05} +{"train_loss": 0.01442531868815422, "grad_norm": 0.055037833750247955, "global_step": 26377, "epoch": 248, "lr": 5.036157638031515e-05} +{"train_loss": 0.01905209757387638, "grad_norm": 0.07498465478420258, "global_step": 26378, "epoch": 248, "lr": 5.035861269052343e-05} +{"train_loss": 0.01500711590051651, "grad_norm": 0.15019522607326508, "global_step": 26379, "epoch": 248, "lr": 5.0355648999471706e-05} +{"train_loss": 0.015152065083384514, "grad_norm": 0.09356672316789627, "global_step": 26380, "epoch": 248, "lr": 5.035268530717039e-05} +{"train_loss": 0.015167108736932278, "grad_norm": 0.15660043060779572, "global_step": 26381, "epoch": 248, "lr": 5.034972161362989e-05} +{"train_loss": 0.01413436233997345, "grad_norm": 0.17759495973587036, "global_step": 26382, "epoch": 248, "lr": 5.0346757918860595e-05} +{"train_loss": 0.014261351898312569, "grad_norm": 0.15996487438678741, "global_step": 26383, "epoch": 248, "lr": 5.034379422287298e-05} +{"train_loss": 0.018948080018162727, "grad_norm": 0.09693977981805801, "global_step": 26384, "epoch": 248, "lr": 5.0340830525677395e-05} +{"train_loss": 0.014386218041181564, "grad_norm": 0.10025312006473541, "global_step": 26385, "epoch": 248, "lr": 5.0337866827284284e-05} +{"train_loss": 0.012997251003980637, "grad_norm": 0.06730449944734573, "global_step": 26386, "epoch": 248, "lr": 5.033490312770408e-05} +{"train_loss": 0.017308108508586884, "grad_norm": 0.07190846651792526, "global_step": 26387, "epoch": 248, "lr": 5.033193942694715e-05} +{"train_loss": 0.017152851447463036, "grad_norm": 0.07628253847360611, "global_step": 26388, "epoch": 248, "lr": 5.032897572502393e-05} +{"train_loss": 0.017114385962486267, "grad_norm": 0.13453976809978485, "global_step": 26389, "epoch": 248, "lr": 5.032601202194483e-05} +{"train_loss": 0.015771348029375076, "grad_norm": 0.1104254499077797, "global_step": 26390, "epoch": 248, "lr": 5.032304831772028e-05} +{"train_loss": 0.016028329730033875, "grad_norm": 0.11569427698850632, "global_step": 26391, "epoch": 248, "lr": 5.0320084612360676e-05} +{"train_loss": 0.015472878701984882, "grad_norm": 0.10474969446659088, "global_step": 26392, "epoch": 248, "lr": 5.031712090587642e-05} +{"train_loss": 0.015619452331074566, "grad_norm": 0.15220198035240173, "global_step": 26393, "epoch": 248, "lr": 5.0314157198277954e-05, "val_loss": 0.019034964963793755} +{"train_loss": 0.013648751191794872, "grad_norm": 0.1119682565331459, "global_step": 26394, "epoch": 249, "lr": 5.031119348957566e-05} +{"train_loss": 0.015341615304350853, "grad_norm": 0.08462482690811157, "global_step": 26395, "epoch": 249, "lr": 5.0308229779779995e-05} +{"train_loss": 0.013616098091006279, "grad_norm": 0.07849660515785217, "global_step": 26396, "epoch": 249, "lr": 5.030526606890133e-05} +{"train_loss": 0.014908039011061192, "grad_norm": 0.09268754720687866, "global_step": 26397, "epoch": 249, "lr": 5.030230235695007e-05} +{"train_loss": 0.01699395850300789, "grad_norm": 0.0772811621427536, "global_step": 26398, "epoch": 249, "lr": 5.029933864393669e-05} +{"train_loss": 0.015816301107406616, "grad_norm": 0.09408700466156006, "global_step": 26399, "epoch": 249, "lr": 5.029637492987153e-05} +{"train_loss": 0.014755120500922203, "grad_norm": 0.05595673620700836, "global_step": 26400, "epoch": 249, "lr": 5.0293411214765053e-05} +{"train_loss": 0.015497921034693718, "grad_norm": 0.13278403878211975, "global_step": 26401, "epoch": 249, "lr": 5.029044749862766e-05} +{"train_loss": 0.015577701851725578, "grad_norm": 0.047787077724933624, "global_step": 26402, "epoch": 249, "lr": 5.028748378146976e-05} +{"train_loss": 0.015660308301448822, "grad_norm": 0.07323931157588959, "global_step": 26403, "epoch": 249, "lr": 5.028452006330175e-05} +{"train_loss": 0.017624661326408386, "grad_norm": 0.07486504316329956, "global_step": 26404, "epoch": 249, "lr": 5.028155634413409e-05} +{"train_loss": 0.01557900570333004, "grad_norm": 0.1124865785241127, "global_step": 26405, "epoch": 249, "lr": 5.0278592623977136e-05} +{"train_loss": 0.01629958674311638, "grad_norm": 0.06036081910133362, "global_step": 26406, "epoch": 249, "lr": 5.0275628902841346e-05} +{"train_loss": 0.015395764261484146, "grad_norm": 0.08417605608701706, "global_step": 26407, "epoch": 249, "lr": 5.0272665180737114e-05} +{"train_loss": 0.016824334859848022, "grad_norm": 0.10251498967409134, "global_step": 26408, "epoch": 249, "lr": 5.026970145767485e-05} +{"train_loss": 0.01597464829683304, "grad_norm": 0.13947793841362, "global_step": 26409, "epoch": 249, "lr": 5.026673773366496e-05} +{"train_loss": 0.014944438822567463, "grad_norm": 0.07863091677427292, "global_step": 26410, "epoch": 249, "lr": 5.02637740087179e-05} +{"train_loss": 0.01534629613161087, "grad_norm": 0.14822912216186523, "global_step": 26411, "epoch": 249, "lr": 5.0260810282844026e-05} +{"train_loss": 0.01805480197072029, "grad_norm": 0.07694675773382187, "global_step": 26412, "epoch": 249, "lr": 5.025784655605379e-05} +{"train_loss": 0.014282293617725372, "grad_norm": 0.06477918475866318, "global_step": 26413, "epoch": 249, "lr": 5.025488282835759e-05} +{"train_loss": 0.016342446208000183, "grad_norm": 0.059507403522729874, "global_step": 26414, "epoch": 249, "lr": 5.025191909976584e-05} +{"train_loss": 0.01516435295343399, "grad_norm": 0.07909421622753143, "global_step": 26415, "epoch": 249, "lr": 5.024895537028897e-05} +{"train_loss": 0.013632950372993946, "grad_norm": 0.11055386811494827, "global_step": 26416, "epoch": 249, "lr": 5.0245991639937364e-05} +{"train_loss": 0.014632821083068848, "grad_norm": 0.07921476662158966, "global_step": 26417, "epoch": 249, "lr": 5.024302790872145e-05} +{"train_loss": 0.013437552377581596, "grad_norm": 0.06365278363227844, "global_step": 26418, "epoch": 249, "lr": 5.024006417665166e-05} +{"train_loss": 0.013351593166589737, "grad_norm": 0.09924353659152985, "global_step": 26419, "epoch": 249, "lr": 5.023710044373837e-05} +{"train_loss": 0.013549456372857094, "grad_norm": 0.058226946741342545, "global_step": 26420, "epoch": 249, "lr": 5.023413670999202e-05} +{"train_loss": 0.015676476061344147, "grad_norm": 0.056662943214178085, "global_step": 26421, "epoch": 249, "lr": 5.0231172975423e-05} +{"train_loss": 0.016783196479082108, "grad_norm": 0.11078380048274994, "global_step": 26422, "epoch": 249, "lr": 5.022820924004177e-05} +{"train_loss": 0.014406145550310612, "grad_norm": 0.07460644841194153, "global_step": 26423, "epoch": 249, "lr": 5.0225245503858685e-05} +{"train_loss": 0.01579214632511139, "grad_norm": 0.05556658282876015, "global_step": 26424, "epoch": 249, "lr": 5.0222281766884206e-05} +{"train_loss": 0.014983458444476128, "grad_norm": 0.08687266707420349, "global_step": 26425, "epoch": 249, "lr": 5.0219318029128716e-05} +{"train_loss": 0.01572146825492382, "grad_norm": 0.07152833044528961, "global_step": 26426, "epoch": 249, "lr": 5.021635429060262e-05} +{"train_loss": 0.0133073590695858, "grad_norm": 0.06321435421705246, "global_step": 26427, "epoch": 249, "lr": 5.021339055131637e-05} +{"train_loss": 0.013549786061048508, "grad_norm": 0.06197807937860489, "global_step": 26428, "epoch": 249, "lr": 5.021042681128035e-05} +{"train_loss": 0.016228165477514267, "grad_norm": 0.06417525559663773, "global_step": 26429, "epoch": 249, "lr": 5.020746307050499e-05} +{"train_loss": 0.015160409733653069, "grad_norm": 0.07067984342575073, "global_step": 26430, "epoch": 249, "lr": 5.020449932900069e-05} +{"train_loss": 0.017806271091103554, "grad_norm": 0.0713888630270958, "global_step": 26431, "epoch": 249, "lr": 5.020153558677787e-05} +{"train_loss": 0.015471704304218292, "grad_norm": 0.05638571456074715, "global_step": 26432, "epoch": 249, "lr": 5.019857184384693e-05} +{"train_loss": 0.015880899503827095, "grad_norm": 0.06364455074071884, "global_step": 26433, "epoch": 249, "lr": 5.0195608100218315e-05} +{"train_loss": 0.016506249085068703, "grad_norm": 0.07310302555561066, "global_step": 26434, "epoch": 249, "lr": 5.0192644355902406e-05} +{"train_loss": 0.016103895381093025, "grad_norm": 0.06704879552125931, "global_step": 26435, "epoch": 249, "lr": 5.0189680610909615e-05} +{"train_loss": 0.015256560407578945, "grad_norm": 0.09002842009067535, "global_step": 26436, "epoch": 249, "lr": 5.0186716865250406e-05} +{"train_loss": 0.01678512431681156, "grad_norm": 0.059968795627355576, "global_step": 26437, "epoch": 249, "lr": 5.0183753118935125e-05} +{"train_loss": 0.014326505362987518, "grad_norm": 0.06736309826374054, "global_step": 26438, "epoch": 249, "lr": 5.0180789371974216e-05} +{"train_loss": 0.016115516424179077, "grad_norm": 0.0817752331495285, "global_step": 26439, "epoch": 249, "lr": 5.0177825624378114e-05} +{"train_loss": 0.0163560900837183, "grad_norm": 0.0949256494641304, "global_step": 26440, "epoch": 249, "lr": 5.0174861876157185e-05} +{"train_loss": 0.016543537378311157, "grad_norm": 0.07075732946395874, "global_step": 26441, "epoch": 249, "lr": 5.017189812732188e-05} +{"train_loss": 0.014202000573277473, "grad_norm": 0.07685967534780502, "global_step": 26442, "epoch": 249, "lr": 5.0168934377882594e-05} +{"train_loss": 0.015271392650902271, "grad_norm": 0.10732387006282806, "global_step": 26443, "epoch": 249, "lr": 5.0165970627849754e-05} +{"train_loss": 0.01610788144171238, "grad_norm": 0.06499549746513367, "global_step": 26444, "epoch": 249, "lr": 5.016300687723376e-05} +{"train_loss": 0.013869151473045349, "grad_norm": 0.06446380168199539, "global_step": 26445, "epoch": 249, "lr": 5.016004312604503e-05} +{"train_loss": 0.017273426055908203, "grad_norm": 0.08060739934444427, "global_step": 26446, "epoch": 249, "lr": 5.0157079374293983e-05} +{"train_loss": 0.017548002302646637, "grad_norm": 0.08507315814495087, "global_step": 26447, "epoch": 249, "lr": 5.015411562199103e-05} +{"train_loss": 0.01910768449306488, "grad_norm": 0.07909742742776871, "global_step": 26448, "epoch": 249, "lr": 5.015115186914657e-05} +{"train_loss": 0.015578366816043854, "grad_norm": 0.07140316069126129, "global_step": 26449, "epoch": 249, "lr": 5.014818811577104e-05} +{"train_loss": 0.016460735350847244, "grad_norm": 0.0691438689827919, "global_step": 26450, "epoch": 249, "lr": 5.014522436187482e-05} +{"train_loss": 0.015379674732685089, "grad_norm": 0.08795184642076492, "global_step": 26451, "epoch": 249, "lr": 5.0142260607468374e-05} +{"train_loss": 0.01630716770887375, "grad_norm": 0.0963984802365303, "global_step": 26452, "epoch": 249, "lr": 5.0139296852562065e-05} +{"train_loss": 0.01628880947828293, "grad_norm": 0.06983022391796112, "global_step": 26453, "epoch": 249, "lr": 5.0136333097166334e-05} +{"train_loss": 0.013697361573576927, "grad_norm": 0.06207584962248802, "global_step": 26454, "epoch": 249, "lr": 5.01333693412916e-05} +{"train_loss": 0.015409106388688087, "grad_norm": 0.1511480212211609, "global_step": 26455, "epoch": 249, "lr": 5.0130405584948245e-05} +{"train_loss": 0.01732906699180603, "grad_norm": 0.07503838837146759, "global_step": 26456, "epoch": 249, "lr": 5.012744182814672e-05} +{"train_loss": 0.0141674242913723, "grad_norm": 0.055265702307224274, "global_step": 26457, "epoch": 249, "lr": 5.0124478070897394e-05} +{"train_loss": 0.01590431109070778, "grad_norm": 0.05582623556256294, "global_step": 26458, "epoch": 249, "lr": 5.012151431321073e-05} +{"train_loss": 0.01785709708929062, "grad_norm": 0.06757041066884995, "global_step": 26459, "epoch": 249, "lr": 5.011855055509711e-05} +{"train_loss": 0.015041758306324482, "grad_norm": 0.04869258031249046, "global_step": 26460, "epoch": 249, "lr": 5.011558679656695e-05} +{"train_loss": 0.014196962118148804, "grad_norm": 0.13226716220378876, "global_step": 26461, "epoch": 249, "lr": 5.011262303763068e-05} +{"train_loss": 0.018043819814920425, "grad_norm": 0.08489210903644562, "global_step": 26462, "epoch": 249, "lr": 5.010965927829868e-05} +{"train_loss": 0.01642928272485733, "grad_norm": 0.07569915801286697, "global_step": 26463, "epoch": 249, "lr": 5.010669551858141e-05} +{"train_loss": 0.017686771228909492, "grad_norm": 0.06932160258293152, "global_step": 26464, "epoch": 249, "lr": 5.0103731758489237e-05} +{"train_loss": 0.015263380482792854, "grad_norm": 0.0919276773929596, "global_step": 26465, "epoch": 249, "lr": 5.0100767998032606e-05} +{"train_loss": 0.013973334804177284, "grad_norm": 0.08710181713104248, "global_step": 26466, "epoch": 249, "lr": 5.0097804237221935e-05} +{"train_loss": 0.012629098258912563, "grad_norm": 0.06593415141105652, "global_step": 26467, "epoch": 249, "lr": 5.009484047606759e-05} +{"train_loss": 0.013906233012676239, "grad_norm": 0.2367919236421585, "global_step": 26468, "epoch": 249, "lr": 5.009187671458004e-05} +{"train_loss": 0.015401543118059635, "grad_norm": 0.07693003118038177, "global_step": 26469, "epoch": 249, "lr": 5.0088912952769676e-05} +{"train_loss": 0.016630740836262703, "grad_norm": 0.19773805141448975, "global_step": 26470, "epoch": 249, "lr": 5.00859491906469e-05} +{"train_loss": 0.01748747192323208, "grad_norm": 0.11611030250787735, "global_step": 26471, "epoch": 249, "lr": 5.008298542822214e-05} +{"train_loss": 0.017101339995861053, "grad_norm": 0.1356743425130844, "global_step": 26472, "epoch": 249, "lr": 5.00800216655058e-05} +{"train_loss": 0.014583831652998924, "grad_norm": 0.07442060112953186, "global_step": 26473, "epoch": 249, "lr": 5.0077057902508304e-05} +{"train_loss": 0.01445270050317049, "grad_norm": 0.050627753138542175, "global_step": 26474, "epoch": 249, "lr": 5.0074094139240066e-05} +{"train_loss": 0.013642804697155952, "grad_norm": 0.07460930943489075, "global_step": 26475, "epoch": 249, "lr": 5.0071130375711484e-05} +{"train_loss": 0.01553766056895256, "grad_norm": 0.07688920944929123, "global_step": 26476, "epoch": 249, "lr": 5.0068166611932986e-05} +{"train_loss": 0.014721402898430824, "grad_norm": 0.10844773054122925, "global_step": 26477, "epoch": 249, "lr": 5.006520284791498e-05} +{"train_loss": 0.0171385258436203, "grad_norm": 0.07675160467624664, "global_step": 26478, "epoch": 249, "lr": 5.0062239083667885e-05} +{"train_loss": 0.01697302982211113, "grad_norm": 0.07881268113851547, "global_step": 26479, "epoch": 249, "lr": 5.0059275319202084e-05} +{"train_loss": 0.01608545146882534, "grad_norm": 0.11293242871761322, "global_step": 26480, "epoch": 249, "lr": 5.005631155452804e-05} +{"train_loss": 0.01476921234279871, "grad_norm": 0.11229152232408524, "global_step": 26481, "epoch": 249, "lr": 5.0053347789656134e-05} +{"train_loss": 0.015534075908362865, "grad_norm": 0.05829848721623421, "global_step": 26482, "epoch": 249, "lr": 5.005038402459679e-05} +{"train_loss": 0.01383983250707388, "grad_norm": 0.07533440738916397, "global_step": 26483, "epoch": 249, "lr": 5.0047420259360424e-05} +{"train_loss": 0.015359793789684772, "grad_norm": 0.08901263773441315, "global_step": 26484, "epoch": 249, "lr": 5.0044456493957426e-05} +{"train_loss": 0.015137123875319958, "grad_norm": 0.08329447358846664, "global_step": 26485, "epoch": 249, "lr": 5.0041492728398234e-05} +{"train_loss": 0.014921455644071102, "grad_norm": 0.06540796905755997, "global_step": 26486, "epoch": 249, "lr": 5.0038528962693265e-05} +{"train_loss": 0.019479287788271904, "grad_norm": 0.07649742066860199, "global_step": 26487, "epoch": 249, "lr": 5.003556519685291e-05} +{"train_loss": 0.017676308751106262, "grad_norm": 0.08893275260925293, "global_step": 26488, "epoch": 249, "lr": 5.003260143088761e-05} +{"train_loss": 0.01682453788816929, "grad_norm": 0.07598995417356491, "global_step": 26489, "epoch": 249, "lr": 5.002963766480775e-05} +{"train_loss": 0.013131883926689625, "grad_norm": 0.10428853332996368, "global_step": 26490, "epoch": 249, "lr": 5.002667389862375e-05} +{"train_loss": 0.01712999865412712, "grad_norm": 0.06254027038812637, "global_step": 26491, "epoch": 249, "lr": 5.0023710132346026e-05} +{"train_loss": 0.014511686749756336, "grad_norm": 0.13245725631713867, "global_step": 26492, "epoch": 249, "lr": 5.0020746365985026e-05} +{"train_loss": 0.014161765575408936, "grad_norm": 0.09402293711900711, "global_step": 26493, "epoch": 249, "lr": 5.00177825995511e-05} +{"train_loss": 0.014431258663535118, "grad_norm": 0.12533344328403473, "global_step": 26494, "epoch": 249, "lr": 5.001481883305471e-05} +{"train_loss": 0.018643904477357864, "grad_norm": 0.0804532915353775, "global_step": 26495, "epoch": 249, "lr": 5.0011855066506254e-05} +{"train_loss": 0.012346445582807064, "grad_norm": 0.08819403499364853, "global_step": 26496, "epoch": 249, "lr": 5.000889129991613e-05} +{"train_loss": 0.015725452452898026, "grad_norm": 0.12042383849620819, "global_step": 26497, "epoch": 249, "lr": 5.0005927533294785e-05} +{"train_loss": 0.014632763341069221, "grad_norm": 0.06292734295129776, "global_step": 26498, "epoch": 249, "lr": 5.00029637666526e-05} +{"train_loss": 0.015579545413548092, "grad_norm": 0.09340249747037888, "global_step": 26499, "epoch": 249, "lr": 5e-05, "val_loss": 0.019026141613721848} +{"train_loss": 0.017061179503798485, "grad_norm": 0.09472901374101639, "global_step": 26500, "epoch": 250, "lr": 4.9997036233347414e-05} +{"train_loss": 0.014225933700799942, "grad_norm": 0.07946814596652985, "global_step": 26501, "epoch": 250, "lr": 4.999407246670522e-05} +{"train_loss": 0.013881885446608067, "grad_norm": 0.06242925301194191, "global_step": 26502, "epoch": 250, "lr": 4.999110870008389e-05} +{"train_loss": 0.015854638069868088, "grad_norm": 0.07943099737167358, "global_step": 26503, "epoch": 250, "lr": 4.9988144933493765e-05} +{"train_loss": 0.016840897500514984, "grad_norm": 0.07413942366838455, "global_step": 26504, "epoch": 250, "lr": 4.998518116694531e-05} +{"train_loss": 0.01307513378560543, "grad_norm": 0.10327368229627609, "global_step": 26505, "epoch": 250, "lr": 4.998221740044891e-05} +{"train_loss": 0.015782779082655907, "grad_norm": 0.06500153988599777, "global_step": 26506, "epoch": 250, "lr": 4.997925363401499e-05} +{"train_loss": 0.01765640825033188, "grad_norm": 0.05814603716135025, "global_step": 26507, "epoch": 250, "lr": 4.997628986765397e-05} +{"train_loss": 0.013311208225786686, "grad_norm": 0.05792751908302307, "global_step": 26508, "epoch": 250, "lr": 4.997332610137627e-05} +{"train_loss": 0.016397910192608833, "grad_norm": 0.09325214475393295, "global_step": 26509, "epoch": 250, "lr": 4.997036233519228e-05} +{"train_loss": 0.01300613209605217, "grad_norm": 0.1392531543970108, "global_step": 26510, "epoch": 250, "lr": 4.996739856911241e-05} +{"train_loss": 0.016785304993391037, "grad_norm": 0.11499277502298355, "global_step": 26511, "epoch": 250, "lr": 4.99644348031471e-05} +{"train_loss": 0.014721320942044258, "grad_norm": 0.09222981333732605, "global_step": 26512, "epoch": 250, "lr": 4.996147103730675e-05} +{"train_loss": 0.018040381371974945, "grad_norm": 0.06811954081058502, "global_step": 26513, "epoch": 250, "lr": 4.9958507271601764e-05} +{"train_loss": 0.014438167214393616, "grad_norm": 0.059111952781677246, "global_step": 26514, "epoch": 250, "lr": 4.995554350604259e-05} +{"train_loss": 0.014924449846148491, "grad_norm": 0.05134730041027069, "global_step": 26515, "epoch": 250, "lr": 4.99525797406396e-05} +{"train_loss": 0.014818050898611546, "grad_norm": 0.05786249786615372, "global_step": 26516, "epoch": 250, "lr": 4.9949615975403224e-05} +{"train_loss": 0.013838007114827633, "grad_norm": 0.06748278439044952, "global_step": 26517, "epoch": 250, "lr": 4.994665221034388e-05} +{"train_loss": 0.015290193259716034, "grad_norm": 0.09256864339113235, "global_step": 26518, "epoch": 250, "lr": 4.994368844547196e-05} +{"train_loss": 0.014973970130085945, "grad_norm": 0.06958530843257904, "global_step": 26519, "epoch": 250, "lr": 4.9940724680797914e-05} +{"train_loss": 0.015743926167488098, "grad_norm": 0.08555825054645538, "global_step": 26520, "epoch": 250, "lr": 4.993776091633215e-05} +{"train_loss": 0.018928393721580505, "grad_norm": 0.08547219634056091, "global_step": 26521, "epoch": 250, "lr": 4.9934797152085044e-05} +{"train_loss": 0.015249921008944511, "grad_norm": 0.05143160745501518, "global_step": 26522, "epoch": 250, "lr": 4.9931833388067026e-05} +{"train_loss": 0.016576629132032394, "grad_norm": 0.09892240166664124, "global_step": 26523, "epoch": 250, "lr": 4.992886962428853e-05} +{"train_loss": 0.01562977209687233, "grad_norm": 0.08247721195220947, "global_step": 26524, "epoch": 250, "lr": 4.9925905860759946e-05} +{"train_loss": 0.014760829508304596, "grad_norm": 0.05186295136809349, "global_step": 26525, "epoch": 250, "lr": 4.9922942097491694e-05} +{"train_loss": 0.01598617434501648, "grad_norm": 0.09057820588350296, "global_step": 26526, "epoch": 250, "lr": 4.9919978334494215e-05} +{"train_loss": 0.015161871910095215, "grad_norm": 0.08176370710134506, "global_step": 26527, "epoch": 250, "lr": 4.9917014571777877e-05} +{"train_loss": 0.013905683532357216, "grad_norm": 0.05284770578145981, "global_step": 26528, "epoch": 250, "lr": 4.9914050809353114e-05} +{"train_loss": 0.017399564385414124, "grad_norm": 0.07478949427604675, "global_step": 26529, "epoch": 250, "lr": 4.991108704723034e-05} +{"train_loss": 0.015614712610840797, "grad_norm": 0.0691574290394783, "global_step": 26530, "epoch": 250, "lr": 4.990812328541996e-05} +{"train_loss": 0.01650777831673622, "grad_norm": 0.0642521008849144, "global_step": 26531, "epoch": 250, "lr": 4.9905159523932405e-05} +{"train_loss": 0.01662137545645237, "grad_norm": 0.08811617642641068, "global_step": 26532, "epoch": 250, "lr": 4.990219576277809e-05} +{"train_loss": 0.01536789070814848, "grad_norm": 0.06173926964402199, "global_step": 26533, "epoch": 250, "lr": 4.9899232001967406e-05} +{"train_loss": 0.017329588532447815, "grad_norm": 0.16254493594169617, "global_step": 26534, "epoch": 250, "lr": 4.9896268241510775e-05} +{"train_loss": 0.016504690051078796, "grad_norm": 0.052685223519802094, "global_step": 26535, "epoch": 250, "lr": 4.98933044814186e-05} +{"train_loss": 0.015631165355443954, "grad_norm": 0.06690119206905365, "global_step": 26536, "epoch": 250, "lr": 4.9890340721701326e-05} +{"train_loss": 0.013208371587097645, "grad_norm": 0.07461608946323395, "global_step": 26537, "epoch": 250, "lr": 4.9887376962369325e-05} +{"train_loss": 0.016895700246095657, "grad_norm": 0.10011476278305054, "global_step": 26538, "epoch": 250, "lr": 4.9884413203433064e-05} +{"train_loss": 0.01600456051528454, "grad_norm": 0.10877549648284912, "global_step": 26539, "epoch": 250, "lr": 4.9881449444902904e-05} +{"train_loss": 0.014753159135580063, "grad_norm": 0.05385522171854973, "global_step": 26540, "epoch": 250, "lr": 4.987848568678928e-05} +{"train_loss": 0.013460008427500725, "grad_norm": 0.09593182802200317, "global_step": 26541, "epoch": 250, "lr": 4.987552192910261e-05} +{"train_loss": 0.016306478530168533, "grad_norm": 0.10020642727613449, "global_step": 26542, "epoch": 250, "lr": 4.987255817185329e-05} +{"train_loss": 0.013947071507573128, "grad_norm": 0.06684119999408722, "global_step": 26543, "epoch": 250, "lr": 4.986959441505176e-05} +{"train_loss": 0.014594506472349167, "grad_norm": 0.08852999657392502, "global_step": 26544, "epoch": 250, "lr": 4.9866630658708416e-05} +{"train_loss": 0.015790734440088272, "grad_norm": 0.10513253509998322, "global_step": 26545, "epoch": 250, "lr": 4.986366690283368e-05} +{"train_loss": 0.01567484624683857, "grad_norm": 0.09701821953058243, "global_step": 26546, "epoch": 250, "lr": 4.9860703147437946e-05} +{"train_loss": 0.013714080676436424, "grad_norm": 0.10796481370925903, "global_step": 26547, "epoch": 250, "lr": 4.985773939253163e-05} +{"train_loss": 0.01608884707093239, "grad_norm": 0.05754756182432175, "global_step": 26548, "epoch": 250, "lr": 4.9854775638125174e-05} +{"train_loss": 0.016153903678059578, "grad_norm": 0.057073239237070084, "global_step": 26549, "epoch": 250, "lr": 4.985181188422897e-05} +{"train_loss": 0.01455841027200222, "grad_norm": 0.08438888192176819, "global_step": 26550, "epoch": 250, "lr": 4.9848848130853445e-05} +{"train_loss": 0.017191525548696518, "grad_norm": 0.06228925287723541, "global_step": 26551, "epoch": 250, "lr": 4.9845884378008984e-05} +{"train_loss": 0.01369721069931984, "grad_norm": 0.05353275313973427, "global_step": 26552, "epoch": 250, "lr": 4.984292062570602e-05} +{"train_loss": 0.01682310551404953, "grad_norm": 0.12762555480003357, "global_step": 26553, "epoch": 250, "lr": 4.983995687395498e-05} +{"train_loss": 0.015278004109859467, "grad_norm": 0.08348296582698822, "global_step": 26554, "epoch": 250, "lr": 4.983699312276624e-05} +{"train_loss": 0.015264008194208145, "grad_norm": 0.08909564465284348, "global_step": 26555, "epoch": 250, "lr": 4.983402937215027e-05} +{"train_loss": 0.019286032766103745, "grad_norm": 0.07805123180150986, "global_step": 26556, "epoch": 250, "lr": 4.983106562211741e-05} +{"train_loss": 0.01724223792552948, "grad_norm": 0.08750657737255096, "global_step": 26557, "epoch": 250, "lr": 4.982810187267814e-05} +{"train_loss": 0.013385867699980736, "grad_norm": 0.06916651129722595, "global_step": 26558, "epoch": 250, "lr": 4.9825138123842826e-05} +{"train_loss": 0.01592225581407547, "grad_norm": 0.0777912586927414, "global_step": 26559, "epoch": 250, "lr": 4.98221743756219e-05} +{"train_loss": 0.01730453595519066, "grad_norm": 0.08599868416786194, "global_step": 26560, "epoch": 250, "lr": 4.981921062802578e-05} +{"train_loss": 0.013762632384896278, "grad_norm": 0.05648017302155495, "global_step": 26561, "epoch": 250, "lr": 4.981624688106489e-05} +{"train_loss": 0.01443299651145935, "grad_norm": 0.06371527910232544, "global_step": 26562, "epoch": 250, "lr": 4.981328313474962e-05} +{"train_loss": 0.01586509868502617, "grad_norm": 0.0685407817363739, "global_step": 26563, "epoch": 250, "lr": 4.981031938909039e-05} +{"train_loss": 0.015482200309634209, "grad_norm": 0.08409886807203293, "global_step": 26564, "epoch": 250, "lr": 4.9807355644097606e-05} +{"train_loss": 0.017998185008764267, "grad_norm": 0.08032235503196716, "global_step": 26565, "epoch": 250, "lr": 4.9804391899781696e-05} +{"train_loss": 0.013947863131761551, "grad_norm": 0.0943124070763588, "global_step": 26566, "epoch": 250, "lr": 4.980142815615306e-05} +{"train_loss": 0.016761675477027893, "grad_norm": 0.05143970996141434, "global_step": 26567, "epoch": 250, "lr": 4.979846441322216e-05} +{"train_loss": 0.013389412313699722, "grad_norm": 0.057882990688085556, "global_step": 26568, "epoch": 250, "lr": 4.979550067099932e-05} +{"train_loss": 0.013320770114660263, "grad_norm": 0.06343324482440948, "global_step": 26569, "epoch": 250, "lr": 4.9792536929495027e-05} +{"train_loss": 0.016370084136724472, "grad_norm": 0.09494437277317047, "global_step": 26570, "epoch": 250, "lr": 4.978957318871966e-05} +{"train_loss": 0.013971712440252304, "grad_norm": 0.07845652103424072, "global_step": 26571, "epoch": 250, "lr": 4.978660944868364e-05} +{"train_loss": 0.015736740082502365, "grad_norm": 0.0885753259062767, "global_step": 26572, "epoch": 250, "lr": 4.978364570939738e-05} +{"train_loss": 0.013487854972481728, "grad_norm": 0.060441210865974426, "global_step": 26573, "epoch": 250, "lr": 4.978068197087131e-05} +{"train_loss": 0.017122961580753326, "grad_norm": 0.14244812726974487, "global_step": 26574, "epoch": 250, "lr": 4.977771823311582e-05} +{"train_loss": 0.01606534793972969, "grad_norm": 0.0635756105184555, "global_step": 26575, "epoch": 250, "lr": 4.9774754496141327e-05} +{"train_loss": 0.01661047898232937, "grad_norm": 0.07266970723867416, "global_step": 26576, "epoch": 250, "lr": 4.977179075995824e-05} +{"train_loss": 0.014915253967046738, "grad_norm": 0.10317661613225937, "global_step": 26577, "epoch": 250, "lr": 4.9768827024577e-05} +{"train_loss": 0.016397831961512566, "grad_norm": 0.09115664660930634, "global_step": 26578, "epoch": 250, "lr": 4.976586329000799e-05} +{"train_loss": 0.015748223289847374, "grad_norm": 0.0961068868637085, "global_step": 26579, "epoch": 250, "lr": 4.976289955626165e-05} +{"train_loss": 0.014890531077980995, "grad_norm": 0.06982168555259705, "global_step": 26580, "epoch": 250, "lr": 4.975993582334836e-05} +{"train_loss": 0.014971025288105011, "grad_norm": 0.10808324813842773, "global_step": 26581, "epoch": 250, "lr": 4.9756972091278556e-05} +{"train_loss": 0.015030916780233383, "grad_norm": 0.05318360775709152, "global_step": 26582, "epoch": 250, "lr": 4.975400836006265e-05} +{"train_loss": 0.017766498029232025, "grad_norm": 0.1297651082277298, "global_step": 26583, "epoch": 250, "lr": 4.975104462971104e-05} +{"train_loss": 0.01755215972661972, "grad_norm": 0.06484668701887131, "global_step": 26584, "epoch": 250, "lr": 4.9748080900234156e-05} +{"train_loss": 0.013025090098381042, "grad_norm": 0.049646858125925064, "global_step": 26585, "epoch": 250, "lr": 4.9745117171642417e-05} +{"train_loss": 0.016858547925949097, "grad_norm": 0.12738710641860962, "global_step": 26586, "epoch": 250, "lr": 4.974215344394622e-05} +{"train_loss": 0.016085278242826462, "grad_norm": 0.07096655666828156, "global_step": 26587, "epoch": 250, "lr": 4.9739189717155986e-05} +{"train_loss": 0.014846151694655418, "grad_norm": 0.06214570999145508, "global_step": 26588, "epoch": 250, "lr": 4.973622599128211e-05} +{"train_loss": 0.015343911945819855, "grad_norm": 0.058570098131895065, "global_step": 26589, "epoch": 250, "lr": 4.973326226633503e-05} +{"train_loss": 0.01693829707801342, "grad_norm": 0.08424944430589676, "global_step": 26590, "epoch": 250, "lr": 4.973029854232516e-05} +{"train_loss": 0.015416061505675316, "grad_norm": 0.0892808809876442, "global_step": 26591, "epoch": 250, "lr": 4.972733481926291e-05} +{"train_loss": 0.015617342665791512, "grad_norm": 0.075938880443573, "global_step": 26592, "epoch": 250, "lr": 4.972437109715866e-05} +{"train_loss": 0.01467910222709179, "grad_norm": 0.057040221989154816, "global_step": 26593, "epoch": 250, "lr": 4.972140737602287e-05} +{"train_loss": 0.01496871653944254, "grad_norm": 0.08105143904685974, "global_step": 26594, "epoch": 250, "lr": 4.9718443655865925e-05} +{"train_loss": 0.014532501809298992, "grad_norm": 0.07241593301296234, "global_step": 26595, "epoch": 250, "lr": 4.971547993669824e-05} +{"train_loss": 0.014699523337185383, "grad_norm": 0.09114746749401093, "global_step": 26596, "epoch": 250, "lr": 4.9712516218530244e-05} +{"train_loss": 0.01750226318836212, "grad_norm": 0.06778500229120255, "global_step": 26597, "epoch": 250, "lr": 4.970955250137235e-05} +{"train_loss": 0.017052853479981422, "grad_norm": 0.08230709284543991, "global_step": 26598, "epoch": 250, "lr": 4.970658878523496e-05} +{"train_loss": 0.01558825746178627, "grad_norm": 0.11527016013860703, "global_step": 26599, "epoch": 250, "lr": 4.970362507012848e-05} +{"train_loss": 0.017126191407442093, "grad_norm": 0.12079214304685593, "global_step": 26600, "epoch": 250, "lr": 4.970066135606333e-05} +{"train_loss": 0.01484967302531004, "grad_norm": 0.13300476968288422, "global_step": 26601, "epoch": 250, "lr": 4.9697697643049926e-05} +{"train_loss": 0.016316622495651245, "grad_norm": 0.07778725773096085, "global_step": 26602, "epoch": 250, "lr": 4.96947339310987e-05} +{"train_loss": 0.014936452731490135, "grad_norm": 0.09149415791034698, "global_step": 26603, "epoch": 250, "lr": 4.969177022022003e-05} +{"train_loss": 0.014760310761630535, "grad_norm": 0.08232836425304413, "global_step": 26604, "epoch": 250, "lr": 4.968880651042435e-05} +{"train_loss": 0.015560568283962191, "grad_norm": 0.09644865244626999, "global_step": 26605, "epoch": 250, "lr": 4.968584280172206e-05, "train_batch_mse_error": 0.014017603360116482, "val_loss": 0.01954106241464615} +{"train_loss": 0.013680249452590942, "grad_norm": 0.0723101869225502, "global_step": 26606, "epoch": 251, "lr": 4.968287909412358e-05} +{"train_loss": 0.01855139061808586, "grad_norm": 0.11476875841617584, "global_step": 26607, "epoch": 251, "lr": 4.9679915387639336e-05} +{"train_loss": 0.01865643449127674, "grad_norm": 0.09911464154720306, "global_step": 26608, "epoch": 251, "lr": 4.967695168227974e-05} +{"train_loss": 0.01751076430082321, "grad_norm": 0.07818412780761719, "global_step": 26609, "epoch": 251, "lr": 4.967398797805517e-05} +{"train_loss": 0.016392666846513748, "grad_norm": 0.10698314011096954, "global_step": 26610, "epoch": 251, "lr": 4.967102427497608e-05} +{"train_loss": 0.016159899532794952, "grad_norm": 0.1216774433851242, "global_step": 26611, "epoch": 251, "lr": 4.9668060573052866e-05} +{"train_loss": 0.0163553636521101, "grad_norm": 0.09480496495962143, "global_step": 26612, "epoch": 251, "lr": 4.966509687229593e-05} +{"train_loss": 0.014973009005188942, "grad_norm": 0.07162047177553177, "global_step": 26613, "epoch": 251, "lr": 4.966213317271571e-05} +{"train_loss": 0.016397159546613693, "grad_norm": 0.07713507115840912, "global_step": 26614, "epoch": 251, "lr": 4.965916947432262e-05} +{"train_loss": 0.020079853013157845, "grad_norm": 0.19605101644992828, "global_step": 26615, "epoch": 251, "lr": 4.965620577712705e-05} +{"train_loss": 0.01683058962225914, "grad_norm": 0.07894551753997803, "global_step": 26616, "epoch": 251, "lr": 4.9653242081139416e-05} +{"train_loss": 0.016600701957941055, "grad_norm": 0.10211687535047531, "global_step": 26617, "epoch": 251, "lr": 4.965027838637013e-05} +{"train_loss": 0.014193983748555183, "grad_norm": 0.07481793314218521, "global_step": 26618, "epoch": 251, "lr": 4.964731469282963e-05} +{"train_loss": 0.014243422076106071, "grad_norm": 0.0769565999507904, "global_step": 26619, "epoch": 251, "lr": 4.9644351000528306e-05} +{"train_loss": 0.014755766838788986, "grad_norm": 0.0679745301604271, "global_step": 26620, "epoch": 251, "lr": 4.964138730947659e-05} +{"train_loss": 0.015451186336576939, "grad_norm": 0.049605514854192734, "global_step": 26621, "epoch": 251, "lr": 4.9638423619684864e-05} +{"train_loss": 0.017010370269417763, "grad_norm": 0.17257025837898254, "global_step": 26622, "epoch": 251, "lr": 4.963545993116358e-05} +{"train_loss": 0.015520773828029633, "grad_norm": 0.0771869644522667, "global_step": 26623, "epoch": 251, "lr": 4.963249624392312e-05} +{"train_loss": 0.016955284401774406, "grad_norm": 0.09468168020248413, "global_step": 26624, "epoch": 251, "lr": 4.96295325579739e-05} +{"train_loss": 0.015334800817072392, "grad_norm": 0.08079081028699875, "global_step": 26625, "epoch": 251, "lr": 4.9626568873326343e-05} +{"train_loss": 0.01623963750898838, "grad_norm": 0.06748689711093903, "global_step": 26626, "epoch": 251, "lr": 4.962360518999088e-05} +{"train_loss": 0.0143981222063303, "grad_norm": 0.06957948952913284, "global_step": 26627, "epoch": 251, "lr": 4.96206415079779e-05} +{"train_loss": 0.01391647569835186, "grad_norm": 0.06446510553359985, "global_step": 26628, "epoch": 251, "lr": 4.961767782729781e-05} +{"train_loss": 0.015267583541572094, "grad_norm": 0.15123315155506134, "global_step": 26629, "epoch": 251, "lr": 4.9614714147961026e-05} +{"train_loss": 0.013224031776189804, "grad_norm": 0.06892885267734528, "global_step": 26630, "epoch": 251, "lr": 4.961175046997799e-05} +{"train_loss": 0.018377698957920074, "grad_norm": 0.11421442031860352, "global_step": 26631, "epoch": 251, "lr": 4.960878679335908e-05} +{"train_loss": 0.015433081425726414, "grad_norm": 0.06943773478269577, "global_step": 26632, "epoch": 251, "lr": 4.960582311811474e-05} +{"train_loss": 0.015249962918460369, "grad_norm": 0.0635485127568245, "global_step": 26633, "epoch": 251, "lr": 4.960285944425536e-05} +{"train_loss": 0.016175683587789536, "grad_norm": 0.0896025225520134, "global_step": 26634, "epoch": 251, "lr": 4.959989577179134e-05} +{"train_loss": 0.015395630151033401, "grad_norm": 0.1251378357410431, "global_step": 26635, "epoch": 251, "lr": 4.959693210073312e-05} +{"train_loss": 0.018194224685430527, "grad_norm": 0.0833367183804512, "global_step": 26636, "epoch": 251, "lr": 4.9593968431091116e-05} +{"train_loss": 0.01372431218624115, "grad_norm": 0.09298522770404816, "global_step": 26637, "epoch": 251, "lr": 4.9591004762875715e-05} +{"train_loss": 0.013245219364762306, "grad_norm": 0.05964965745806694, "global_step": 26638, "epoch": 251, "lr": 4.9588041096097364e-05} +{"train_loss": 0.013478809967637062, "grad_norm": 0.1037377342581749, "global_step": 26639, "epoch": 251, "lr": 4.958507743076645e-05} +{"train_loss": 0.016766130924224854, "grad_norm": 0.15096457302570343, "global_step": 26640, "epoch": 251, "lr": 4.9582113766893395e-05} +{"train_loss": 0.015077047049999237, "grad_norm": 0.08602481335401535, "global_step": 26641, "epoch": 251, "lr": 4.95791501044886e-05} +{"train_loss": 0.013225157745182514, "grad_norm": 0.06952427327632904, "global_step": 26642, "epoch": 251, "lr": 4.9576186443562494e-05} +{"train_loss": 0.014759477227926254, "grad_norm": 0.0762273445725441, "global_step": 26643, "epoch": 251, "lr": 4.957322278412549e-05} +{"train_loss": 0.013548236340284348, "grad_norm": 0.06333008408546448, "global_step": 26644, "epoch": 251, "lr": 4.9570259126188004e-05} +{"train_loss": 0.015530215576291084, "grad_norm": 0.08645160496234894, "global_step": 26645, "epoch": 251, "lr": 4.956729546976043e-05} +{"train_loss": 0.012544894590973854, "grad_norm": 0.05526641011238098, "global_step": 26646, "epoch": 251, "lr": 4.9564331814853184e-05} +{"train_loss": 0.016879852861166, "grad_norm": 0.0802977979183197, "global_step": 26647, "epoch": 251, "lr": 4.956136816147669e-05} +{"train_loss": 0.01646638847887516, "grad_norm": 0.16365964710712433, "global_step": 26648, "epoch": 251, "lr": 4.955840450964136e-05} +{"train_loss": 0.014433165080845356, "grad_norm": 0.07424459606409073, "global_step": 26649, "epoch": 251, "lr": 4.9555440859357594e-05} +{"train_loss": 0.016450311988592148, "grad_norm": 0.07599227875471115, "global_step": 26650, "epoch": 251, "lr": 4.955247721063584e-05} +{"train_loss": 0.01731831766664982, "grad_norm": 0.10130546241998672, "global_step": 26651, "epoch": 251, "lr": 4.954951356348647e-05} +{"train_loss": 0.014162261039018631, "grad_norm": 0.09108918160200119, "global_step": 26652, "epoch": 251, "lr": 4.954654991791992e-05} +{"train_loss": 0.01604367606341839, "grad_norm": 0.08588676154613495, "global_step": 26653, "epoch": 251, "lr": 4.954358627394658e-05} +{"train_loss": 0.01691693253815174, "grad_norm": 0.1714579313993454, "global_step": 26654, "epoch": 251, "lr": 4.95406226315769e-05} +{"train_loss": 0.017400935292243958, "grad_norm": 0.16263949871063232, "global_step": 26655, "epoch": 251, "lr": 4.953765899082127e-05} +{"train_loss": 0.014229378663003445, "grad_norm": 0.07185041904449463, "global_step": 26656, "epoch": 251, "lr": 4.9534695351690104e-05} +{"train_loss": 0.01750657521188259, "grad_norm": 0.07820543646812439, "global_step": 26657, "epoch": 251, "lr": 4.953173171419381e-05} +{"train_loss": 0.014104317873716354, "grad_norm": 0.06531640887260437, "global_step": 26658, "epoch": 251, "lr": 4.95287680783428e-05} +{"train_loss": 0.014951465651392937, "grad_norm": 0.08889325708150864, "global_step": 26659, "epoch": 251, "lr": 4.95258044441475e-05} +{"train_loss": 0.014934155158698559, "grad_norm": 0.1051652655005455, "global_step": 26660, "epoch": 251, "lr": 4.9522840811618326e-05} +{"train_loss": 0.01438536774367094, "grad_norm": 0.05103279650211334, "global_step": 26661, "epoch": 251, "lr": 4.951987718076569e-05} +{"train_loss": 0.016737231984734535, "grad_norm": 0.07589316368103027, "global_step": 26662, "epoch": 251, "lr": 4.951691355159998e-05} +{"train_loss": 0.018146388232707977, "grad_norm": 0.06970321387052536, "global_step": 26663, "epoch": 251, "lr": 4.951394992413162e-05} +{"train_loss": 0.015175271779298782, "grad_norm": 0.05705196410417557, "global_step": 26664, "epoch": 251, "lr": 4.951098629837104e-05} +{"train_loss": 0.017406560480594635, "grad_norm": 0.07989040017127991, "global_step": 26665, "epoch": 251, "lr": 4.9508022674328636e-05} +{"train_loss": 0.01794358342885971, "grad_norm": 0.058151401579380035, "global_step": 26666, "epoch": 251, "lr": 4.950505905201482e-05} +{"train_loss": 0.02040667086839676, "grad_norm": 0.06818913668394089, "global_step": 26667, "epoch": 251, "lr": 4.9502095431440034e-05} +{"train_loss": 0.014990758150815964, "grad_norm": 0.06448972225189209, "global_step": 26668, "epoch": 251, "lr": 4.949913181261465e-05} +{"train_loss": 0.015062755905091763, "grad_norm": 0.07603759318590164, "global_step": 26669, "epoch": 251, "lr": 4.9496168195549104e-05} +{"train_loss": 0.013748623430728912, "grad_norm": 0.0566764660179615, "global_step": 26670, "epoch": 251, "lr": 4.9493204580253795e-05} +{"train_loss": 0.015833698213100433, "grad_norm": 0.14285115897655487, "global_step": 26671, "epoch": 251, "lr": 4.949024096673915e-05} +{"train_loss": 0.013154007494449615, "grad_norm": 0.10005886852741241, "global_step": 26672, "epoch": 251, "lr": 4.9487277355015584e-05} +{"train_loss": 0.014855217188596725, "grad_norm": 0.08991583436727524, "global_step": 26673, "epoch": 251, "lr": 4.9484313745093507e-05} +{"train_loss": 0.015333782881498337, "grad_norm": 0.07843217998743057, "global_step": 26674, "epoch": 251, "lr": 4.948135013698332e-05} +{"train_loss": 0.014033416286110878, "grad_norm": 0.0994483083486557, "global_step": 26675, "epoch": 251, "lr": 4.9478386530695434e-05} +{"train_loss": 0.014836716465651989, "grad_norm": 0.09854789823293686, "global_step": 26676, "epoch": 251, "lr": 4.9475422926240276e-05} +{"train_loss": 0.014352709986269474, "grad_norm": 0.048626434057950974, "global_step": 26677, "epoch": 251, "lr": 4.9472459323628256e-05} +{"train_loss": 0.014656738378107548, "grad_norm": 0.07663619518280029, "global_step": 26678, "epoch": 251, "lr": 4.946949572286977e-05} +{"train_loss": 0.013891543261706829, "grad_norm": 0.10975369811058044, "global_step": 26679, "epoch": 251, "lr": 4.946653212397527e-05} +{"train_loss": 0.015437541529536247, "grad_norm": 0.08746492862701416, "global_step": 26680, "epoch": 251, "lr": 4.946356852695514e-05} +{"train_loss": 0.016366327181458473, "grad_norm": 0.05680212005972862, "global_step": 26681, "epoch": 251, "lr": 4.946060493181979e-05} +{"train_loss": 0.015803160145878792, "grad_norm": 0.09769291430711746, "global_step": 26682, "epoch": 251, "lr": 4.945764133857963e-05} +{"train_loss": 0.016571972519159317, "grad_norm": 0.14639924466609955, "global_step": 26683, "epoch": 251, "lr": 4.945467774724509e-05} +{"train_loss": 0.01478653959929943, "grad_norm": 0.09070976078510284, "global_step": 26684, "epoch": 251, "lr": 4.945171415782658e-05} +{"train_loss": 0.014502257108688354, "grad_norm": 0.06105200946331024, "global_step": 26685, "epoch": 251, "lr": 4.944875057033452e-05} +{"train_loss": 0.0152326300740242, "grad_norm": 0.052275825291872025, "global_step": 26686, "epoch": 251, "lr": 4.94457869847793e-05} +{"train_loss": 0.014978616498410702, "grad_norm": 0.0473693311214447, "global_step": 26687, "epoch": 251, "lr": 4.944282340117133e-05} +{"train_loss": 0.013191154226660728, "grad_norm": 0.07677722722291946, "global_step": 26688, "epoch": 251, "lr": 4.943985981952105e-05} +{"train_loss": 0.015904342755675316, "grad_norm": 0.07387811690568924, "global_step": 26689, "epoch": 251, "lr": 4.9436896239838846e-05} +{"train_loss": 0.014451533555984497, "grad_norm": 0.17797492444515228, "global_step": 26690, "epoch": 251, "lr": 4.9433932662135144e-05} +{"train_loss": 0.015160602517426014, "grad_norm": 0.07796827703714371, "global_step": 26691, "epoch": 251, "lr": 4.943096908642039e-05} +{"train_loss": 0.01697836071252823, "grad_norm": 0.06261613219976425, "global_step": 26692, "epoch": 251, "lr": 4.9428005512704925e-05} +{"train_loss": 0.017764948308467865, "grad_norm": 0.08213337510824203, "global_step": 26693, "epoch": 251, "lr": 4.942504194099922e-05} +{"train_loss": 0.012367360293865204, "grad_norm": 0.05698632821440697, "global_step": 26694, "epoch": 251, "lr": 4.9422078371313666e-05} +{"train_loss": 0.01598464325070381, "grad_norm": 0.09237523376941681, "global_step": 26695, "epoch": 251, "lr": 4.9419114803658654e-05} +{"train_loss": 0.01822250708937645, "grad_norm": 0.07197877764701843, "global_step": 26696, "epoch": 251, "lr": 4.9416151238044646e-05} +{"train_loss": 0.015262151136994362, "grad_norm": 0.08480307459831238, "global_step": 26697, "epoch": 251, "lr": 4.941318767448203e-05} +{"train_loss": 0.0142756262794137, "grad_norm": 0.06359361857175827, "global_step": 26698, "epoch": 251, "lr": 4.941022411298121e-05} +{"train_loss": 0.01353343203663826, "grad_norm": 0.09013012051582336, "global_step": 26699, "epoch": 251, "lr": 4.940726055355259e-05} +{"train_loss": 0.013348894193768501, "grad_norm": 0.09691040217876434, "global_step": 26700, "epoch": 251, "lr": 4.9404296996206616e-05} +{"train_loss": 0.017382899299263954, "grad_norm": 0.1295894980430603, "global_step": 26701, "epoch": 251, "lr": 4.940133344095368e-05} +{"train_loss": 0.01714429445564747, "grad_norm": 0.08323200792074203, "global_step": 26702, "epoch": 251, "lr": 4.9398369887804214e-05} +{"train_loss": 0.017126817256212234, "grad_norm": 0.11120431125164032, "global_step": 26703, "epoch": 251, "lr": 4.93954063367686e-05} +{"train_loss": 0.015093576163053513, "grad_norm": 0.06509759277105331, "global_step": 26704, "epoch": 251, "lr": 4.9392442787857254e-05} +{"train_loss": 0.017005812376737595, "grad_norm": 0.09600286930799484, "global_step": 26705, "epoch": 251, "lr": 4.938947924108061e-05} +{"train_loss": 0.015239092521369457, "grad_norm": 0.06584744155406952, "global_step": 26706, "epoch": 251, "lr": 4.938651569644907e-05} +{"train_loss": 0.015084534883499146, "grad_norm": 0.09578823298215866, "global_step": 26707, "epoch": 251, "lr": 4.9383552153973034e-05} +{"train_loss": 0.01403452455997467, "grad_norm": 0.1052199974656105, "global_step": 26708, "epoch": 251, "lr": 4.9380588613662956e-05} +{"train_loss": 0.013083897531032562, "grad_norm": 0.058140095323324203, "global_step": 26709, "epoch": 251, "lr": 4.9377625075529204e-05} +{"train_loss": 0.017166530713438988, "grad_norm": 0.08589695394039154, "global_step": 26710, "epoch": 251, "lr": 4.937466153958221e-05} +{"train_loss": 0.015590800965239978, "grad_norm": 0.13398416340351105, "global_step": 26711, "epoch": 251, "lr": 4.9371698005832365e-05, "val_loss": 0.0189947672188282} +{"train_loss": 0.014589831233024597, "grad_norm": 0.1115676611661911, "global_step": 26712, "epoch": 252, "lr": 4.936873447429012e-05} +{"train_loss": 0.015096066519618034, "grad_norm": 0.06067771464586258, "global_step": 26713, "epoch": 252, "lr": 4.9365770944965865e-05} +{"train_loss": 0.015538986772298813, "grad_norm": 0.06377781927585602, "global_step": 26714, "epoch": 252, "lr": 4.936280741787003e-05} +{"train_loss": 0.014814293012022972, "grad_norm": 0.06531964242458344, "global_step": 26715, "epoch": 252, "lr": 4.9359843893013e-05} +{"train_loss": 0.015717187896370888, "grad_norm": 0.06917809695005417, "global_step": 26716, "epoch": 252, "lr": 4.935688037040518e-05} +{"train_loss": 0.013235126622021198, "grad_norm": 0.0857773944735527, "global_step": 26717, "epoch": 252, "lr": 4.9353916850057016e-05} +{"train_loss": 0.016342075541615486, "grad_norm": 0.10911070555448532, "global_step": 26718, "epoch": 252, "lr": 4.935095333197892e-05} +{"train_loss": 0.016356637701392174, "grad_norm": 0.08301994204521179, "global_step": 26719, "epoch": 252, "lr": 4.934798981618127e-05} +{"train_loss": 0.013121861033141613, "grad_norm": 0.08797284960746765, "global_step": 26720, "epoch": 252, "lr": 4.934502630267453e-05} +{"train_loss": 0.014959121122956276, "grad_norm": 0.07303521782159805, "global_step": 26721, "epoch": 252, "lr": 4.9342062791469056e-05} +{"train_loss": 0.017812296748161316, "grad_norm": 0.07083917409181595, "global_step": 26722, "epoch": 252, "lr": 4.93390992825753e-05} +{"train_loss": 0.01588619127869606, "grad_norm": 0.0634671077132225, "global_step": 26723, "epoch": 252, "lr": 4.933613577600365e-05} +{"train_loss": 0.014583238400518894, "grad_norm": 0.11276832222938538, "global_step": 26724, "epoch": 252, "lr": 4.933317227176454e-05} +{"train_loss": 0.015679126605391502, "grad_norm": 0.0711459293961525, "global_step": 26725, "epoch": 252, "lr": 4.933020876986837e-05} +{"train_loss": 0.01494244858622551, "grad_norm": 0.08070755004882812, "global_step": 26726, "epoch": 252, "lr": 4.9327245270325565e-05} +{"train_loss": 0.015554512850940228, "grad_norm": 0.10116207599639893, "global_step": 26727, "epoch": 252, "lr": 4.932428177314653e-05} +{"train_loss": 0.015632545575499535, "grad_norm": 0.11468277871608734, "global_step": 26728, "epoch": 252, "lr": 4.932131827834164e-05} +{"train_loss": 0.017636695876717567, "grad_norm": 0.14901259541511536, "global_step": 26729, "epoch": 252, "lr": 4.931835478592137e-05} +{"train_loss": 0.015883754938840866, "grad_norm": 0.15039414167404175, "global_step": 26730, "epoch": 252, "lr": 4.931539129589611e-05} +{"train_loss": 0.012945275753736496, "grad_norm": 0.050979308784008026, "global_step": 26731, "epoch": 252, "lr": 4.931242780827624e-05} +{"train_loss": 0.014806274324655533, "grad_norm": 0.05623242259025574, "global_step": 26732, "epoch": 252, "lr": 4.930946432307224e-05} +{"train_loss": 0.013780426234006882, "grad_norm": 0.06373564898967743, "global_step": 26733, "epoch": 252, "lr": 4.930650084029445e-05} +{"train_loss": 0.015818966552615166, "grad_norm": 0.07637625187635422, "global_step": 26734, "epoch": 252, "lr": 4.930353735995333e-05} +{"train_loss": 0.01573624461889267, "grad_norm": 0.09319446235895157, "global_step": 26735, "epoch": 252, "lr": 4.930057388205927e-05} +{"train_loss": 0.015656962990760803, "grad_norm": 0.06134400516748428, "global_step": 26736, "epoch": 252, "lr": 4.929761040662268e-05} +{"train_loss": 0.01488696038722992, "grad_norm": 0.15402793884277344, "global_step": 26737, "epoch": 252, "lr": 4.929464693365399e-05} +{"train_loss": 0.01386553980410099, "grad_norm": 0.14159628748893738, "global_step": 26738, "epoch": 252, "lr": 4.929168346316362e-05} +{"train_loss": 0.015448255464434624, "grad_norm": 0.07755088061094284, "global_step": 26739, "epoch": 252, "lr": 4.9288719995161954e-05} +{"train_loss": 0.013038012199103832, "grad_norm": 0.08047236502170563, "global_step": 26740, "epoch": 252, "lr": 4.928575652965941e-05} +{"train_loss": 0.013949381187558174, "grad_norm": 0.1058482825756073, "global_step": 26741, "epoch": 252, "lr": 4.92827930666664e-05} +{"train_loss": 0.014811109751462936, "grad_norm": 0.15190304815769196, "global_step": 26742, "epoch": 252, "lr": 4.927982960619336e-05} +{"train_loss": 0.012687202543020248, "grad_norm": 0.08318600803613663, "global_step": 26743, "epoch": 252, "lr": 4.927686614825067e-05} +{"train_loss": 0.014920075424015522, "grad_norm": 0.07746760547161102, "global_step": 26744, "epoch": 252, "lr": 4.9273902692848795e-05} +{"train_loss": 0.014990179799497128, "grad_norm": 0.08063129335641861, "global_step": 26745, "epoch": 252, "lr": 4.927093923999807e-05} +{"train_loss": 0.016828462481498718, "grad_norm": 0.0923447459936142, "global_step": 26746, "epoch": 252, "lr": 4.9267975789708964e-05} +{"train_loss": 0.016328416764736176, "grad_norm": 0.16003282368183136, "global_step": 26747, "epoch": 252, "lr": 4.926501234199187e-05} +{"train_loss": 0.014729490503668785, "grad_norm": 0.05795164778828621, "global_step": 26748, "epoch": 252, "lr": 4.9262048896857195e-05} +{"train_loss": 0.01521284133195877, "grad_norm": 0.1144447773694992, "global_step": 26749, "epoch": 252, "lr": 4.925908545431537e-05} +{"train_loss": 0.015707165002822876, "grad_norm": 0.06827998906373978, "global_step": 26750, "epoch": 252, "lr": 4.92561220143768e-05} +{"train_loss": 0.017804503440856934, "grad_norm": 0.09609293192625046, "global_step": 26751, "epoch": 252, "lr": 4.925315857705189e-05} +{"train_loss": 0.015230976045131683, "grad_norm": 0.12077213078737259, "global_step": 26752, "epoch": 252, "lr": 4.925019514235104e-05} +{"train_loss": 0.01718258485198021, "grad_norm": 0.16188594698905945, "global_step": 26753, "epoch": 252, "lr": 4.92472317102847e-05} +{"train_loss": 0.015706803649663925, "grad_norm": 0.14771512150764465, "global_step": 26754, "epoch": 252, "lr": 4.924426828086325e-05} +{"train_loss": 0.01556682214140892, "grad_norm": 0.17529034614562988, "global_step": 26755, "epoch": 252, "lr": 4.9241304854097124e-05} +{"train_loss": 0.015859249979257584, "grad_norm": 0.1530880182981491, "global_step": 26756, "epoch": 252, "lr": 4.9238341429996714e-05} +{"train_loss": 0.014885706827044487, "grad_norm": 0.13734732568264008, "global_step": 26757, "epoch": 252, "lr": 4.9235378008572434e-05} +{"train_loss": 0.015238077379763126, "grad_norm": 0.14601044356822968, "global_step": 26758, "epoch": 252, "lr": 4.9232414589834704e-05} +{"train_loss": 0.013105282559990883, "grad_norm": 0.07664696872234344, "global_step": 26759, "epoch": 252, "lr": 4.922945117379394e-05} +{"train_loss": 0.015391645021736622, "grad_norm": 0.05473652482032776, "global_step": 26760, "epoch": 252, "lr": 4.922648776046054e-05} +{"train_loss": 0.01669125072658062, "grad_norm": 0.07227745652198792, "global_step": 26761, "epoch": 252, "lr": 4.922352434984496e-05} +{"train_loss": 0.017505638301372528, "grad_norm": 0.10441620647907257, "global_step": 26762, "epoch": 252, "lr": 4.922056094195754e-05} +{"train_loss": 0.01456567831337452, "grad_norm": 0.07553204149007797, "global_step": 26763, "epoch": 252, "lr": 4.921759753680875e-05} +{"train_loss": 0.015743432566523552, "grad_norm": 0.11880514025688171, "global_step": 26764, "epoch": 252, "lr": 4.921463413440898e-05} +{"train_loss": 0.016126003116369247, "grad_norm": 0.10809165984392166, "global_step": 26765, "epoch": 252, "lr": 4.921167073476863e-05} +{"train_loss": 0.01790434867143631, "grad_norm": 0.06810292601585388, "global_step": 26766, "epoch": 252, "lr": 4.920870733789813e-05} +{"train_loss": 0.01419108547270298, "grad_norm": 0.10334090143442154, "global_step": 26767, "epoch": 252, "lr": 4.920574394380791e-05} +{"train_loss": 0.01590903103351593, "grad_norm": 0.07562986016273499, "global_step": 26768, "epoch": 252, "lr": 4.9202780552508344e-05} +{"train_loss": 0.018616721034049988, "grad_norm": 0.0825151801109314, "global_step": 26769, "epoch": 252, "lr": 4.919981716400985e-05} +{"train_loss": 0.01856023445725441, "grad_norm": 0.16120995581150055, "global_step": 26770, "epoch": 252, "lr": 4.919685377832286e-05} +{"train_loss": 0.014418978244066238, "grad_norm": 0.054063580930233, "global_step": 26771, "epoch": 252, "lr": 4.919389039545778e-05} +{"train_loss": 0.01637803018093109, "grad_norm": 0.059889115393161774, "global_step": 26772, "epoch": 252, "lr": 4.9190927015425e-05} +{"train_loss": 0.01617581583559513, "grad_norm": 0.1191745400428772, "global_step": 26773, "epoch": 252, "lr": 4.9187963638235e-05} +{"train_loss": 0.014674603007733822, "grad_norm": 0.05705966800451279, "global_step": 26774, "epoch": 252, "lr": 4.918500026389809e-05} +{"train_loss": 0.017069220542907715, "grad_norm": 0.08242328464984894, "global_step": 26775, "epoch": 252, "lr": 4.918203689242476e-05} +{"train_loss": 0.01777367666363716, "grad_norm": 0.1048569306731224, "global_step": 26776, "epoch": 252, "lr": 4.917907352382539e-05} +{"train_loss": 0.013177753426134586, "grad_norm": 0.0679858922958374, "global_step": 26777, "epoch": 252, "lr": 4.917611015811039e-05} +{"train_loss": 0.01469158474355936, "grad_norm": 0.04448067769408226, "global_step": 26778, "epoch": 252, "lr": 4.9173146795290196e-05} +{"train_loss": 0.018083879724144936, "grad_norm": 0.11464798450469971, "global_step": 26779, "epoch": 252, "lr": 4.9170183435375205e-05} +{"train_loss": 0.015337268821895123, "grad_norm": 0.059311866760253906, "global_step": 26780, "epoch": 252, "lr": 4.9167220078375824e-05} +{"train_loss": 0.01626252755522728, "grad_norm": 0.05852789431810379, "global_step": 26781, "epoch": 252, "lr": 4.916425672430246e-05} +{"train_loss": 0.016780804842710495, "grad_norm": 0.12638813257217407, "global_step": 26782, "epoch": 252, "lr": 4.916129337316554e-05} +{"train_loss": 0.015057483687996864, "grad_norm": 0.06578123569488525, "global_step": 26783, "epoch": 252, "lr": 4.915833002497547e-05} +{"train_loss": 0.016333002597093582, "grad_norm": 0.10616577416658401, "global_step": 26784, "epoch": 252, "lr": 4.915536667974266e-05} +{"train_loss": 0.01637178286910057, "grad_norm": 0.07214763015508652, "global_step": 26785, "epoch": 252, "lr": 4.915240333747755e-05} +{"train_loss": 0.016320865601301193, "grad_norm": 0.07818740606307983, "global_step": 26786, "epoch": 252, "lr": 4.91494399981905e-05} +{"train_loss": 0.015024163760244846, "grad_norm": 0.07213308662176132, "global_step": 26787, "epoch": 252, "lr": 4.914647666189195e-05} +{"train_loss": 0.016730928793549538, "grad_norm": 0.09734813868999481, "global_step": 26788, "epoch": 252, "lr": 4.9143513328592314e-05} +{"train_loss": 0.015792595222592354, "grad_norm": 0.075063556432724, "global_step": 26789, "epoch": 252, "lr": 4.914054999830199e-05} +{"train_loss": 0.015693077817559242, "grad_norm": 0.0696512833237648, "global_step": 26790, "epoch": 252, "lr": 4.9137586671031404e-05} +{"train_loss": 0.014616498723626137, "grad_norm": 0.05258084088563919, "global_step": 26791, "epoch": 252, "lr": 4.913462334679098e-05} +{"train_loss": 0.015526260249316692, "grad_norm": 0.05887426435947418, "global_step": 26792, "epoch": 252, "lr": 4.91316600255911e-05} +{"train_loss": 0.014404959976673126, "grad_norm": 0.059728968888521194, "global_step": 26793, "epoch": 252, "lr": 4.9128696707442194e-05} +{"train_loss": 0.014278649352490902, "grad_norm": 0.07755573838949203, "global_step": 26794, "epoch": 252, "lr": 4.912573339235465e-05} +{"train_loss": 0.014052190817892551, "grad_norm": 0.06489095091819763, "global_step": 26795, "epoch": 252, "lr": 4.912277008033892e-05} +{"train_loss": 0.014397932216525078, "grad_norm": 0.11644169688224792, "global_step": 26796, "epoch": 252, "lr": 4.911980677140539e-05} +{"train_loss": 0.017048466950654984, "grad_norm": 0.06450966000556946, "global_step": 26797, "epoch": 252, "lr": 4.9116843465564485e-05} +{"train_loss": 0.014372065663337708, "grad_norm": 0.14850422739982605, "global_step": 26798, "epoch": 252, "lr": 4.911388016282658e-05} +{"train_loss": 0.016783010214567184, "grad_norm": 0.1573619544506073, "global_step": 26799, "epoch": 252, "lr": 4.911091686320213e-05} +{"train_loss": 0.01591487228870392, "grad_norm": 0.04859171062707901, "global_step": 26800, "epoch": 252, "lr": 4.9107953566701535e-05} +{"train_loss": 0.017345014959573746, "grad_norm": 0.1338525116443634, "global_step": 26801, "epoch": 252, "lr": 4.910499027333519e-05} +{"train_loss": 0.014169040136039257, "grad_norm": 0.09223265200853348, "global_step": 26802, "epoch": 252, "lr": 4.9102026983113537e-05} +{"train_loss": 0.016062935814261436, "grad_norm": 0.07566728442907333, "global_step": 26803, "epoch": 252, "lr": 4.909906369604698e-05} +{"train_loss": 0.015949316322803497, "grad_norm": 0.08034088462591171, "global_step": 26804, "epoch": 252, "lr": 4.909610041214591e-05} +{"train_loss": 0.013563094660639763, "grad_norm": 0.05555583909153938, "global_step": 26805, "epoch": 252, "lr": 4.909313713142074e-05} +{"train_loss": 0.01785147190093994, "grad_norm": 0.06516445428133011, "global_step": 26806, "epoch": 252, "lr": 4.9090173853881885e-05} +{"train_loss": 0.013873133808374405, "grad_norm": 0.07750432938337326, "global_step": 26807, "epoch": 252, "lr": 4.908721057953978e-05} +{"train_loss": 0.015576016157865524, "grad_norm": 0.08022642135620117, "global_step": 26808, "epoch": 252, "lr": 4.908424730840483e-05} +{"train_loss": 0.014921274036169052, "grad_norm": 0.1551956981420517, "global_step": 26809, "epoch": 252, "lr": 4.9081284040487426e-05} +{"train_loss": 0.016096916049718857, "grad_norm": 0.11550319194793701, "global_step": 26810, "epoch": 252, "lr": 4.907832077579797e-05} +{"train_loss": 0.016728471964597702, "grad_norm": 0.10210337489843369, "global_step": 26811, "epoch": 252, "lr": 4.907535751434692e-05} +{"train_loss": 0.014802969992160797, "grad_norm": 0.09012618660926819, "global_step": 26812, "epoch": 252, "lr": 4.907239425614465e-05} +{"train_loss": 0.019926730543375015, "grad_norm": 0.08154293149709702, "global_step": 26813, "epoch": 252, "lr": 4.9069431001201573e-05} +{"train_loss": 0.014467371627688408, "grad_norm": 0.0592416450381279, "global_step": 26814, "epoch": 252, "lr": 4.906646774952815e-05} +{"train_loss": 0.016196172684431076, "grad_norm": 0.12092530727386475, "global_step": 26815, "epoch": 252, "lr": 4.906350450113472e-05} +{"train_loss": 0.016989830881357193, "grad_norm": 0.07110166549682617, "global_step": 26816, "epoch": 252, "lr": 4.9060541256031735e-05} +{"train_loss": 0.015568186731060158, "grad_norm": 0.16405776143074036, "global_step": 26817, "epoch": 252, "lr": 4.90575780142296e-05, "val_loss": 0.019170444458723068} +{"train_loss": 0.015486921183764935, "grad_norm": 0.07523512840270996, "global_step": 26818, "epoch": 253, "lr": 4.9054614775738714e-05} +{"train_loss": 0.015645988285541534, "grad_norm": 0.09373978525400162, "global_step": 26819, "epoch": 253, "lr": 4.905165154056952e-05} +{"train_loss": 0.015817567706108093, "grad_norm": 0.0640488937497139, "global_step": 26820, "epoch": 253, "lr": 4.904868830873241e-05} +{"train_loss": 0.017171885818243027, "grad_norm": 0.09359872341156006, "global_step": 26821, "epoch": 253, "lr": 4.904572508023778e-05} +{"train_loss": 0.013508801348507404, "grad_norm": 0.11585318297147751, "global_step": 26822, "epoch": 253, "lr": 4.9042761855096075e-05} +{"train_loss": 0.014916115440428257, "grad_norm": 0.04949492961168289, "global_step": 26823, "epoch": 253, "lr": 4.903979863331767e-05} +{"train_loss": 0.013850073330104351, "grad_norm": 0.06622122973203659, "global_step": 26824, "epoch": 253, "lr": 4.9036835414913004e-05} +{"train_loss": 0.016198139637708664, "grad_norm": 0.06044752895832062, "global_step": 26825, "epoch": 253, "lr": 4.9033872199892466e-05} +{"train_loss": 0.016648337244987488, "grad_norm": 0.06058742478489876, "global_step": 26826, "epoch": 253, "lr": 4.9030908988266516e-05} +{"train_loss": 0.017213229089975357, "grad_norm": 0.06814423203468323, "global_step": 26827, "epoch": 253, "lr": 4.90279457800455e-05} +{"train_loss": 0.014387814328074455, "grad_norm": 0.0770869255065918, "global_step": 26828, "epoch": 253, "lr": 4.902498257523987e-05} +{"train_loss": 0.015202192589640617, "grad_norm": 0.06870228052139282, "global_step": 26829, "epoch": 253, "lr": 4.9022019373860026e-05} +{"train_loss": 0.015612033195793629, "grad_norm": 0.06202499195933342, "global_step": 26830, "epoch": 253, "lr": 4.901905617591637e-05} +{"train_loss": 0.0146868284791708, "grad_norm": 0.10671684890985489, "global_step": 26831, "epoch": 253, "lr": 4.901609298141933e-05} +{"train_loss": 0.014547659084200859, "grad_norm": 0.08492612838745117, "global_step": 26832, "epoch": 253, "lr": 4.901312979037933e-05} +{"train_loss": 0.017533866688609123, "grad_norm": 0.07535364478826523, "global_step": 26833, "epoch": 253, "lr": 4.901016660280675e-05} +{"train_loss": 0.01572386920452118, "grad_norm": 0.08910840004682541, "global_step": 26834, "epoch": 253, "lr": 4.900720341871201e-05} +{"train_loss": 0.013732489198446274, "grad_norm": 0.1085842028260231, "global_step": 26835, "epoch": 253, "lr": 4.900424023810552e-05} +{"train_loss": 0.011922327801585197, "grad_norm": 0.05486302077770233, "global_step": 26836, "epoch": 253, "lr": 4.90012770609977e-05} +{"train_loss": 0.012959413230419159, "grad_norm": 0.08656761795282364, "global_step": 26837, "epoch": 253, "lr": 4.899831388739896e-05} +{"train_loss": 0.013449900783598423, "grad_norm": 0.10013628005981445, "global_step": 26838, "epoch": 253, "lr": 4.899535071731973e-05} +{"train_loss": 0.015493598766624928, "grad_norm": 0.07759121060371399, "global_step": 26839, "epoch": 253, "lr": 4.8992387550770373e-05} +{"train_loss": 0.014929949305951595, "grad_norm": 0.10200006514787674, "global_step": 26840, "epoch": 253, "lr": 4.898942438776134e-05} +{"train_loss": 0.015767928212881088, "grad_norm": 0.07749124616384506, "global_step": 26841, "epoch": 253, "lr": 4.898646122830303e-05} +{"train_loss": 0.015210801735520363, "grad_norm": 0.09032583236694336, "global_step": 26842, "epoch": 253, "lr": 4.898349807240584e-05} +{"train_loss": 0.013164588250219822, "grad_norm": 0.0870044082403183, "global_step": 26843, "epoch": 253, "lr": 4.8980534920080204e-05} +{"train_loss": 0.014206590130925179, "grad_norm": 0.08297275751829147, "global_step": 26844, "epoch": 253, "lr": 4.8977571771336544e-05} +{"train_loss": 0.013356605544686317, "grad_norm": 0.05352163687348366, "global_step": 26845, "epoch": 253, "lr": 4.897460862618524e-05} +{"train_loss": 0.01764872297644615, "grad_norm": 0.07848282158374786, "global_step": 26846, "epoch": 253, "lr": 4.897164548463671e-05} +{"train_loss": 0.016417894512414932, "grad_norm": 0.10629002749919891, "global_step": 26847, "epoch": 253, "lr": 4.896868234670136e-05} +{"train_loss": 0.015574321150779724, "grad_norm": 0.0998997911810875, "global_step": 26848, "epoch": 253, "lr": 4.8965719212389625e-05} +{"train_loss": 0.014546312391757965, "grad_norm": 0.05827284976840019, "global_step": 26849, "epoch": 253, "lr": 4.896275608171191e-05} +{"train_loss": 0.01599697768688202, "grad_norm": 0.054571766406297684, "global_step": 26850, "epoch": 253, "lr": 4.895979295467862e-05} +{"train_loss": 0.015709156170487404, "grad_norm": 0.05584414675831795, "global_step": 26851, "epoch": 253, "lr": 4.8956829831300155e-05} +{"train_loss": 0.016135117039084435, "grad_norm": 0.15542331337928772, "global_step": 26852, "epoch": 253, "lr": 4.895386671158693e-05} +{"train_loss": 0.017369654029607773, "grad_norm": 0.06678116321563721, "global_step": 26853, "epoch": 253, "lr": 4.895090359554937e-05} +{"train_loss": 0.015483174473047256, "grad_norm": 0.16165828704833984, "global_step": 26854, "epoch": 253, "lr": 4.894794048319787e-05} +{"train_loss": 0.017000699415802956, "grad_norm": 0.07112342864274979, "global_step": 26855, "epoch": 253, "lr": 4.894497737454288e-05} +{"train_loss": 0.01657182350754738, "grad_norm": 0.15490439534187317, "global_step": 26856, "epoch": 253, "lr": 4.894201426959475e-05} +{"train_loss": 0.01557350903749466, "grad_norm": 0.11593995988368988, "global_step": 26857, "epoch": 253, "lr": 4.893905116836393e-05} +{"train_loss": 0.014275526627898216, "grad_norm": 0.11234619468450546, "global_step": 26858, "epoch": 253, "lr": 4.8936088070860826e-05} +{"train_loss": 0.015576533041894436, "grad_norm": 0.08517096936702728, "global_step": 26859, "epoch": 253, "lr": 4.893312497709583e-05} +{"train_loss": 0.015628632158041, "grad_norm": 0.08627088367938995, "global_step": 26860, "epoch": 253, "lr": 4.8930161887079385e-05} +{"train_loss": 0.016593310981988907, "grad_norm": 0.08184105902910233, "global_step": 26861, "epoch": 253, "lr": 4.8927198800821895e-05} +{"train_loss": 0.015000002458691597, "grad_norm": 0.07732270658016205, "global_step": 26862, "epoch": 253, "lr": 4.8924235718333756e-05} +{"train_loss": 0.015051381662487984, "grad_norm": 0.07066328823566437, "global_step": 26863, "epoch": 253, "lr": 4.8921272639625375e-05} +{"train_loss": 0.01469920203089714, "grad_norm": 0.17247557640075684, "global_step": 26864, "epoch": 253, "lr": 4.891830956470716e-05} +{"train_loss": 0.018073327839374542, "grad_norm": 0.2174048274755478, "global_step": 26865, "epoch": 253, "lr": 4.891534649358956e-05} +{"train_loss": 0.018565626814961433, "grad_norm": 0.10377029329538345, "global_step": 26866, "epoch": 253, "lr": 4.8912383426282954e-05} +{"train_loss": 0.01355531346052885, "grad_norm": 0.1296093761920929, "global_step": 26867, "epoch": 253, "lr": 4.890942036279777e-05} +{"train_loss": 0.015943801030516624, "grad_norm": 0.16281086206436157, "global_step": 26868, "epoch": 253, "lr": 4.890645730314439e-05} +{"train_loss": 0.016240976750850677, "grad_norm": 0.0555393360555172, "global_step": 26869, "epoch": 253, "lr": 4.890349424733325e-05} +{"train_loss": 0.01749121956527233, "grad_norm": 0.0723521038889885, "global_step": 26870, "epoch": 253, "lr": 4.890053119537475e-05} +{"train_loss": 0.012994233518838882, "grad_norm": 0.1361348032951355, "global_step": 26871, "epoch": 253, "lr": 4.88975681472793e-05} +{"train_loss": 0.014505811966955662, "grad_norm": 0.049608126282691956, "global_step": 26872, "epoch": 253, "lr": 4.889460510305733e-05} +{"train_loss": 0.01761411502957344, "grad_norm": 0.06148013845086098, "global_step": 26873, "epoch": 253, "lr": 4.889164206271924e-05} +{"train_loss": 0.015979483723640442, "grad_norm": 0.07286856323480606, "global_step": 26874, "epoch": 253, "lr": 4.888867902627544e-05} +{"train_loss": 0.01700776070356369, "grad_norm": 0.11940963566303253, "global_step": 26875, "epoch": 253, "lr": 4.888571599373632e-05} +{"train_loss": 0.015814265236258507, "grad_norm": 0.07553862780332565, "global_step": 26876, "epoch": 253, "lr": 4.888275296511231e-05} +{"train_loss": 0.018511109054088593, "grad_norm": 0.10976450145244598, "global_step": 26877, "epoch": 253, "lr": 4.8879789940413826e-05} +{"train_loss": 0.016702108085155487, "grad_norm": 0.054928455501794815, "global_step": 26878, "epoch": 253, "lr": 4.887682691965128e-05} +{"train_loss": 0.015568702481687069, "grad_norm": 0.060114506632089615, "global_step": 26879, "epoch": 253, "lr": 4.887386390283508e-05} +{"train_loss": 0.014938464388251305, "grad_norm": 0.05070803686976433, "global_step": 26880, "epoch": 253, "lr": 4.887090088997562e-05} +{"train_loss": 0.014139288105070591, "grad_norm": 0.1582634598016739, "global_step": 26881, "epoch": 253, "lr": 4.886793788108333e-05} +{"train_loss": 0.016800886020064354, "grad_norm": 0.11784838885068893, "global_step": 26882, "epoch": 253, "lr": 4.886497487616861e-05} +{"train_loss": 0.01149473525583744, "grad_norm": 0.04684292525053024, "global_step": 26883, "epoch": 253, "lr": 4.8862011875241866e-05} +{"train_loss": 0.017126962542533875, "grad_norm": 0.077309750020504, "global_step": 26884, "epoch": 253, "lr": 4.885904887831352e-05} +{"train_loss": 0.015193099156022072, "grad_norm": 0.08420926332473755, "global_step": 26885, "epoch": 253, "lr": 4.8856085885394006e-05} +{"train_loss": 0.017381947487592697, "grad_norm": 0.12697486579418182, "global_step": 26886, "epoch": 253, "lr": 4.885312289649369e-05} +{"train_loss": 0.015184755437076092, "grad_norm": 0.10475842654705048, "global_step": 26887, "epoch": 253, "lr": 4.8850159911623e-05} +{"train_loss": 0.01496041938662529, "grad_norm": 0.11089296638965607, "global_step": 26888, "epoch": 253, "lr": 4.884719693079234e-05} +{"train_loss": 0.01392401847988367, "grad_norm": 0.08365306258201599, "global_step": 26889, "epoch": 253, "lr": 4.884423395401214e-05} +{"train_loss": 0.014728314243257046, "grad_norm": 0.11311953514814377, "global_step": 26890, "epoch": 253, "lr": 4.88412709812928e-05} +{"train_loss": 0.015439173206686974, "grad_norm": 0.06936432421207428, "global_step": 26891, "epoch": 253, "lr": 4.8838308012644736e-05} +{"train_loss": 0.014880162663757801, "grad_norm": 0.07849482446908951, "global_step": 26892, "epoch": 253, "lr": 4.883534504807835e-05} +{"train_loss": 0.014374744147062302, "grad_norm": 0.09980058670043945, "global_step": 26893, "epoch": 253, "lr": 4.883238208760403e-05} +{"train_loss": 0.015130230225622654, "grad_norm": 0.13195577263832092, "global_step": 26894, "epoch": 253, "lr": 4.882941913123223e-05} +{"train_loss": 0.01873486302793026, "grad_norm": 0.06734034419059753, "global_step": 26895, "epoch": 253, "lr": 4.882645617897334e-05} +{"train_loss": 0.018256139010190964, "grad_norm": 0.19327868521213531, "global_step": 26896, "epoch": 253, "lr": 4.882349323083777e-05} +{"train_loss": 0.016020797193050385, "grad_norm": 0.07517354190349579, "global_step": 26897, "epoch": 253, "lr": 4.882053028683595e-05} +{"train_loss": 0.015781259164214134, "grad_norm": 0.07614549994468689, "global_step": 26898, "epoch": 253, "lr": 4.8817567346978266e-05} +{"train_loss": 0.014474542811512947, "grad_norm": 0.08757394552230835, "global_step": 26899, "epoch": 253, "lr": 4.881460441127513e-05} +{"train_loss": 0.01408385206013918, "grad_norm": 0.06313320249319077, "global_step": 26900, "epoch": 253, "lr": 4.881164147973695e-05} +{"train_loss": 0.017803922295570374, "grad_norm": 0.08796116709709167, "global_step": 26901, "epoch": 253, "lr": 4.880867855237415e-05} +{"train_loss": 0.01677539572119713, "grad_norm": 0.05782880261540413, "global_step": 26902, "epoch": 253, "lr": 4.880571562919715e-05} +{"train_loss": 0.01642594486474991, "grad_norm": 0.09810248762369156, "global_step": 26903, "epoch": 253, "lr": 4.8802752710216345e-05} +{"train_loss": 0.017689526081085205, "grad_norm": 0.11313006281852722, "global_step": 26904, "epoch": 253, "lr": 4.879978979544214e-05} +{"train_loss": 0.0172116719186306, "grad_norm": 0.05398964136838913, "global_step": 26905, "epoch": 253, "lr": 4.8796826884884937e-05} +{"train_loss": 0.014266319572925568, "grad_norm": 0.07527825981378555, "global_step": 26906, "epoch": 253, "lr": 4.8793863978555174e-05} +{"train_loss": 0.016362041234970093, "grad_norm": 0.07111998647451401, "global_step": 26907, "epoch": 253, "lr": 4.879090107646325e-05} +{"train_loss": 0.013806549832224846, "grad_norm": 0.10233945399522781, "global_step": 26908, "epoch": 253, "lr": 4.878793817861959e-05} +{"train_loss": 0.014157399535179138, "grad_norm": 0.1139397844672203, "global_step": 26909, "epoch": 253, "lr": 4.8784975285034555e-05} +{"train_loss": 0.017023609951138496, "grad_norm": 0.054217200726270676, "global_step": 26910, "epoch": 253, "lr": 4.878201239571861e-05} +{"train_loss": 0.016259044408798218, "grad_norm": 0.11924509704113007, "global_step": 26911, "epoch": 253, "lr": 4.8779049510682136e-05} +{"train_loss": 0.015578832477331161, "grad_norm": 0.05747164413332939, "global_step": 26912, "epoch": 253, "lr": 4.877608662993554e-05} +{"train_loss": 0.016019335016608238, "grad_norm": 0.06308465451002121, "global_step": 26913, "epoch": 253, "lr": 4.877312375348925e-05} +{"train_loss": 0.012173539027571678, "grad_norm": 0.11059138178825378, "global_step": 26914, "epoch": 253, "lr": 4.877016088135369e-05} +{"train_loss": 0.016442710533738136, "grad_norm": 0.05641383305191994, "global_step": 26915, "epoch": 253, "lr": 4.8767198013539236e-05} +{"train_loss": 0.017480386421084404, "grad_norm": 0.11534440517425537, "global_step": 26916, "epoch": 253, "lr": 4.876423515005632e-05} +{"train_loss": 0.014458846300840378, "grad_norm": 0.07581394910812378, "global_step": 26917, "epoch": 253, "lr": 4.8761272290915316e-05} +{"train_loss": 0.016596727073192596, "grad_norm": 0.0579521544277668, "global_step": 26918, "epoch": 253, "lr": 4.875830943612669e-05} +{"train_loss": 0.012635264545679092, "grad_norm": 0.053071629256010056, "global_step": 26919, "epoch": 253, "lr": 4.875534658570081e-05} +{"train_loss": 0.015309952199459076, "grad_norm": 0.10665424913167953, "global_step": 26920, "epoch": 253, "lr": 4.875238373964812e-05} +{"train_loss": 0.012551475316286087, "grad_norm": 0.07177159190177917, "global_step": 26921, "epoch": 253, "lr": 4.8749420897979003e-05} +{"train_loss": 0.017386864870786667, "grad_norm": 0.11337423324584961, "global_step": 26922, "epoch": 253, "lr": 4.874645806070386e-05} +{"train_loss": 0.015564445212905138, "grad_norm": 0.16480882465839386, "global_step": 26923, "epoch": 253, "lr": 4.874349522783313e-05, "val_loss": 0.01951896958053112} +{"train_loss": 0.01591983437538147, "grad_norm": 0.07433261722326279, "global_step": 26924, "epoch": 254, "lr": 4.874053239937721e-05} +{"train_loss": 0.017038598656654358, "grad_norm": 0.06833735853433609, "global_step": 26925, "epoch": 254, "lr": 4.873756957534651e-05} +{"train_loss": 0.015511681325733662, "grad_norm": 0.10587684065103531, "global_step": 26926, "epoch": 254, "lr": 4.8734606755751455e-05} +{"train_loss": 0.014320248737931252, "grad_norm": 0.05772724375128746, "global_step": 26927, "epoch": 254, "lr": 4.873164394060243e-05} +{"train_loss": 0.017168210819363594, "grad_norm": 0.0965351015329361, "global_step": 26928, "epoch": 254, "lr": 4.872868112990986e-05} +{"train_loss": 0.013549097813665867, "grad_norm": 0.09001553803682327, "global_step": 26929, "epoch": 254, "lr": 4.8725718323684136e-05} +{"train_loss": 0.015986036509275436, "grad_norm": 0.054167237132787704, "global_step": 26930, "epoch": 254, "lr": 4.87227555219357e-05} +{"train_loss": 0.017296038568019867, "grad_norm": 0.12019867449998856, "global_step": 26931, "epoch": 254, "lr": 4.8719792724674944e-05} +{"train_loss": 0.017405763268470764, "grad_norm": 0.10393659770488739, "global_step": 26932, "epoch": 254, "lr": 4.8716829931912296e-05} +{"train_loss": 0.01445482112467289, "grad_norm": 0.08447016775608063, "global_step": 26933, "epoch": 254, "lr": 4.8713867143658134e-05} +{"train_loss": 0.01682616025209427, "grad_norm": 0.07054264098405838, "global_step": 26934, "epoch": 254, "lr": 4.8710904359922865e-05} +{"train_loss": 0.01712993159890175, "grad_norm": 0.10369491577148438, "global_step": 26935, "epoch": 254, "lr": 4.870794158071694e-05} +{"train_loss": 0.014761839061975479, "grad_norm": 0.07103222608566284, "global_step": 26936, "epoch": 254, "lr": 4.870497880605074e-05} +{"train_loss": 0.01625739224255085, "grad_norm": 0.066610187292099, "global_step": 26937, "epoch": 254, "lr": 4.870201603593467e-05} +{"train_loss": 0.016399085521697998, "grad_norm": 0.11845063418149948, "global_step": 26938, "epoch": 254, "lr": 4.869905327037917e-05} +{"train_loss": 0.013675933703780174, "grad_norm": 0.08833124488592148, "global_step": 26939, "epoch": 254, "lr": 4.8696090509394626e-05} +{"train_loss": 0.012715715914964676, "grad_norm": 0.041853949427604675, "global_step": 26940, "epoch": 254, "lr": 4.869312775299145e-05} +{"train_loss": 0.015372877940535545, "grad_norm": 0.18766061961650848, "global_step": 26941, "epoch": 254, "lr": 4.869016500118005e-05} +{"train_loss": 0.016897644847631454, "grad_norm": 0.0911860242486, "global_step": 26942, "epoch": 254, "lr": 4.868720225397085e-05} +{"train_loss": 0.015946965664625168, "grad_norm": 0.0982494130730629, "global_step": 26943, "epoch": 254, "lr": 4.8684239511374245e-05} +{"train_loss": 0.013044448569417, "grad_norm": 0.10465742647647858, "global_step": 26944, "epoch": 254, "lr": 4.868127677340066e-05} +{"train_loss": 0.01649024710059166, "grad_norm": 0.06458992511034012, "global_step": 26945, "epoch": 254, "lr": 4.8678314040060494e-05} +{"train_loss": 0.014344756491482258, "grad_norm": 0.12673655152320862, "global_step": 26946, "epoch": 254, "lr": 4.867535131136414e-05} +{"train_loss": 0.015592202544212341, "grad_norm": 0.1464034914970398, "global_step": 26947, "epoch": 254, "lr": 4.867238858732203e-05} +{"train_loss": 0.015330135822296143, "grad_norm": 0.07623691111803055, "global_step": 26948, "epoch": 254, "lr": 4.866942586794458e-05} +{"train_loss": 0.01345946453511715, "grad_norm": 0.08997666090726852, "global_step": 26949, "epoch": 254, "lr": 4.866646315324217e-05} +{"train_loss": 0.013244962319731712, "grad_norm": 0.051866039633750916, "global_step": 26950, "epoch": 254, "lr": 4.866350044322527e-05} +{"train_loss": 0.01675083488225937, "grad_norm": 0.061093222349882126, "global_step": 26951, "epoch": 254, "lr": 4.8660537737904205e-05} +{"train_loss": 0.016343262046575546, "grad_norm": 0.10656170547008514, "global_step": 26952, "epoch": 254, "lr": 4.8657575037289444e-05} +{"train_loss": 0.013235962018370628, "grad_norm": 0.11690998822450638, "global_step": 26953, "epoch": 254, "lr": 4.8654612341391384e-05} +{"train_loss": 0.017884276807308197, "grad_norm": 0.08139403164386749, "global_step": 26954, "epoch": 254, "lr": 4.865164965022042e-05} +{"train_loss": 0.01617058366537094, "grad_norm": 0.1003117561340332, "global_step": 26955, "epoch": 254, "lr": 4.864868696378699e-05} +{"train_loss": 0.017205599695444107, "grad_norm": 0.09238668531179428, "global_step": 26956, "epoch": 254, "lr": 4.864572428210148e-05} +{"train_loss": 0.016085095703601837, "grad_norm": 0.07131270319223404, "global_step": 26957, "epoch": 254, "lr": 4.86427616051743e-05} +{"train_loss": 0.01633549854159355, "grad_norm": 0.07574242353439331, "global_step": 26958, "epoch": 254, "lr": 4.8639798933015854e-05} +{"train_loss": 0.016809549182653427, "grad_norm": 0.09289611130952835, "global_step": 26959, "epoch": 254, "lr": 4.863683626563657e-05} +{"train_loss": 0.015130147337913513, "grad_norm": 0.06674382835626602, "global_step": 26960, "epoch": 254, "lr": 4.863387360304686e-05} +{"train_loss": 0.01670030690729618, "grad_norm": 0.0643770843744278, "global_step": 26961, "epoch": 254, "lr": 4.8630910945257134e-05} +{"train_loss": 0.015468527562916279, "grad_norm": 0.08495103567838669, "global_step": 26962, "epoch": 254, "lr": 4.862794829227778e-05} +{"train_loss": 0.01570984534919262, "grad_norm": 0.06652995944023132, "global_step": 26963, "epoch": 254, "lr": 4.862498564411921e-05} +{"train_loss": 0.014997338876128197, "grad_norm": 0.0985279530286789, "global_step": 26964, "epoch": 254, "lr": 4.862202300079185e-05} +{"train_loss": 0.0151559729129076, "grad_norm": 0.12336410582065582, "global_step": 26965, "epoch": 254, "lr": 4.86190603623061e-05} +{"train_loss": 0.017028670758008957, "grad_norm": 0.16750486195087433, "global_step": 26966, "epoch": 254, "lr": 4.861609772867237e-05} +{"train_loss": 0.014814094640314579, "grad_norm": 0.0984082818031311, "global_step": 26967, "epoch": 254, "lr": 4.861313509990108e-05} +{"train_loss": 0.013560157269239426, "grad_norm": 0.05057167261838913, "global_step": 26968, "epoch": 254, "lr": 4.861017247600263e-05} +{"train_loss": 0.016298837959766388, "grad_norm": 0.07883036881685257, "global_step": 26969, "epoch": 254, "lr": 4.860720985698742e-05} +{"train_loss": 0.015421012416481972, "grad_norm": 0.12534523010253906, "global_step": 26970, "epoch": 254, "lr": 4.860424724286586e-05} +{"train_loss": 0.017281552776694298, "grad_norm": 0.06669075042009354, "global_step": 26971, "epoch": 254, "lr": 4.860128463364839e-05} +{"train_loss": 0.015974927693605423, "grad_norm": 0.06648217141628265, "global_step": 26972, "epoch": 254, "lr": 4.859832202934538e-05} +{"train_loss": 0.012416305020451546, "grad_norm": 0.13019447028636932, "global_step": 26973, "epoch": 254, "lr": 4.859535942996728e-05} +{"train_loss": 0.013440021313726902, "grad_norm": 0.0785147100687027, "global_step": 26974, "epoch": 254, "lr": 4.859239683552446e-05} +{"train_loss": 0.013741834089159966, "grad_norm": 0.05978897213935852, "global_step": 26975, "epoch": 254, "lr": 4.858943424602734e-05} +{"train_loss": 0.01651288941502571, "grad_norm": 0.06701092422008514, "global_step": 26976, "epoch": 254, "lr": 4.858647166148634e-05} +{"train_loss": 0.015221480280160904, "grad_norm": 0.08571436256170273, "global_step": 26977, "epoch": 254, "lr": 4.8583509081911865e-05} +{"train_loss": 0.014757651835680008, "grad_norm": 0.0957278311252594, "global_step": 26978, "epoch": 254, "lr": 4.8580546507314317e-05} +{"train_loss": 0.01739484630525112, "grad_norm": 0.11439255625009537, "global_step": 26979, "epoch": 254, "lr": 4.857758393770414e-05} +{"train_loss": 0.01623361185193062, "grad_norm": 0.06900465488433838, "global_step": 26980, "epoch": 254, "lr": 4.857462137309168e-05} +{"train_loss": 0.015327665954828262, "grad_norm": 0.08016501367092133, "global_step": 26981, "epoch": 254, "lr": 4.857165881348739e-05} +{"train_loss": 0.014599261805415154, "grad_norm": 0.05919625610113144, "global_step": 26982, "epoch": 254, "lr": 4.856869625890166e-05} +{"train_loss": 0.016849931329488754, "grad_norm": 0.09883192926645279, "global_step": 26983, "epoch": 254, "lr": 4.856573370934493e-05} +{"train_loss": 0.013363633304834366, "grad_norm": 0.0841183215379715, "global_step": 26984, "epoch": 254, "lr": 4.856277116482757e-05} +{"train_loss": 0.014205874875187874, "grad_norm": 0.08309221267700195, "global_step": 26985, "epoch": 254, "lr": 4.855980862536004e-05} +{"train_loss": 0.013360046781599522, "grad_norm": 0.06826474517583847, "global_step": 26986, "epoch": 254, "lr": 4.855684609095269e-05} +{"train_loss": 0.015191365033388138, "grad_norm": 0.09890688955783844, "global_step": 26987, "epoch": 254, "lr": 4.855388356161594e-05} +{"train_loss": 0.013177447952330112, "grad_norm": 0.07078736275434494, "global_step": 26988, "epoch": 254, "lr": 4.855092103736024e-05} +{"train_loss": 0.015427248552441597, "grad_norm": 0.09233744442462921, "global_step": 26989, "epoch": 254, "lr": 4.854795851819596e-05} +{"train_loss": 0.01708531565964222, "grad_norm": 0.07371625304222107, "global_step": 26990, "epoch": 254, "lr": 4.854499600413352e-05} +{"train_loss": 0.015646878629922867, "grad_norm": 0.09658659249544144, "global_step": 26991, "epoch": 254, "lr": 4.854203349518336e-05} +{"train_loss": 0.015477888286113739, "grad_norm": 0.08580538630485535, "global_step": 26992, "epoch": 254, "lr": 4.853907099135583e-05} +{"train_loss": 0.01393899880349636, "grad_norm": 0.0985744297504425, "global_step": 26993, "epoch": 254, "lr": 4.853610849266138e-05} +{"train_loss": 0.01953405886888504, "grad_norm": 0.09091833978891373, "global_step": 26994, "epoch": 254, "lr": 4.853314599911041e-05} +{"train_loss": 0.015900760889053345, "grad_norm": 0.07123338431119919, "global_step": 26995, "epoch": 254, "lr": 4.853018351071331e-05} +{"train_loss": 0.013857632875442505, "grad_norm": 0.09470445662736893, "global_step": 26996, "epoch": 254, "lr": 4.8527221027480516e-05} +{"train_loss": 0.013105180114507675, "grad_norm": 0.06189107522368431, "global_step": 26997, "epoch": 254, "lr": 4.852425854942244e-05} +{"train_loss": 0.016120046377182007, "grad_norm": 0.1269347220659256, "global_step": 26998, "epoch": 254, "lr": 4.852129607654947e-05} +{"train_loss": 0.01656455546617508, "grad_norm": 0.08611476421356201, "global_step": 26999, "epoch": 254, "lr": 4.851833360887201e-05} +{"train_loss": 0.019315367564558983, "grad_norm": 0.15368948876857758, "global_step": 27000, "epoch": 254, "lr": 4.85153711464005e-05} +{"train_loss": 0.013531683012843132, "grad_norm": 0.07514435797929764, "global_step": 27001, "epoch": 254, "lr": 4.8512408689145317e-05} +{"train_loss": 0.016725774854421616, "grad_norm": 0.07295803725719452, "global_step": 27002, "epoch": 254, "lr": 4.850944623711688e-05} +{"train_loss": 0.015123707242310047, "grad_norm": 0.06435254961252213, "global_step": 27003, "epoch": 254, "lr": 4.850648379032563e-05} +{"train_loss": 0.0149232791736722, "grad_norm": 0.06236584857106209, "global_step": 27004, "epoch": 254, "lr": 4.8503521348781915e-05} +{"train_loss": 0.017896998673677444, "grad_norm": 0.0728648230433464, "global_step": 27005, "epoch": 254, "lr": 4.8500558912496186e-05} +{"train_loss": 0.015969591215252876, "grad_norm": 0.07539400458335876, "global_step": 27006, "epoch": 254, "lr": 4.849759648147884e-05} +{"train_loss": 0.014798330143094063, "grad_norm": 0.05834988132119179, "global_step": 27007, "epoch": 254, "lr": 4.849463405574028e-05} +{"train_loss": 0.01339356042444706, "grad_norm": 0.06606312096118927, "global_step": 27008, "epoch": 254, "lr": 4.849167163529095e-05} +{"train_loss": 0.017537295818328857, "grad_norm": 0.08589717745780945, "global_step": 27009, "epoch": 254, "lr": 4.848870922014119e-05} +{"train_loss": 0.015703799203038216, "grad_norm": 0.05757368728518486, "global_step": 27010, "epoch": 254, "lr": 4.848574681030147e-05} +{"train_loss": 0.016345303505659103, "grad_norm": 0.0985734611749649, "global_step": 27011, "epoch": 254, "lr": 4.8482784405782164e-05} +{"train_loss": 0.016737503930926323, "grad_norm": 0.09696188569068909, "global_step": 27012, "epoch": 254, "lr": 4.847982200659371e-05} +{"train_loss": 0.014441289938986301, "grad_norm": 0.11173982918262482, "global_step": 27013, "epoch": 254, "lr": 4.847685961274649e-05} +{"train_loss": 0.013540730811655521, "grad_norm": 0.07390611618757248, "global_step": 27014, "epoch": 254, "lr": 4.847389722425094e-05} +{"train_loss": 0.01564992032945156, "grad_norm": 0.1138182207942009, "global_step": 27015, "epoch": 254, "lr": 4.8470934841117435e-05} +{"train_loss": 0.014319772832095623, "grad_norm": 0.11311733722686768, "global_step": 27016, "epoch": 254, "lr": 4.846797246335639e-05} +{"train_loss": 0.0181935653090477, "grad_norm": 0.14532646536827087, "global_step": 27017, "epoch": 254, "lr": 4.846501009097824e-05} +{"train_loss": 0.014661774039268494, "grad_norm": 0.0925552099943161, "global_step": 27018, "epoch": 254, "lr": 4.8462047723993375e-05} +{"train_loss": 0.015297243371605873, "grad_norm": 0.07799819856882095, "global_step": 27019, "epoch": 254, "lr": 4.845908536241219e-05} +{"train_loss": 0.017521362751722336, "grad_norm": 0.08360964059829712, "global_step": 27020, "epoch": 254, "lr": 4.845612300624515e-05} +{"train_loss": 0.01438224222511053, "grad_norm": 0.09763603657484055, "global_step": 27021, "epoch": 254, "lr": 4.8453160655502585e-05} +{"train_loss": 0.01796356588602066, "grad_norm": 0.13162261247634888, "global_step": 27022, "epoch": 254, "lr": 4.845019831019495e-05} +{"train_loss": 0.015298932790756226, "grad_norm": 0.08898111432790756, "global_step": 27023, "epoch": 254, "lr": 4.844723597033264e-05} +{"train_loss": 0.015901509672403336, "grad_norm": 0.11687879264354706, "global_step": 27024, "epoch": 254, "lr": 4.844427363592607e-05} +{"train_loss": 0.015491600148379803, "grad_norm": 0.07402495294809341, "global_step": 27025, "epoch": 254, "lr": 4.844131130698565e-05} +{"train_loss": 0.01442023180425167, "grad_norm": 0.10678590834140778, "global_step": 27026, "epoch": 254, "lr": 4.8438348983521795e-05} +{"train_loss": 0.017050789669156075, "grad_norm": 0.07937275618314743, "global_step": 27027, "epoch": 254, "lr": 4.8435386665544885e-05} +{"train_loss": 0.01627613790333271, "grad_norm": 0.08767125755548477, "global_step": 27028, "epoch": 254, "lr": 4.843242435306534e-05} +{"train_loss": 0.015537054049518873, "grad_norm": 0.07073888927698135, "global_step": 27029, "epoch": 254, "lr": 4.8429462046093585e-05, "val_loss": 0.01922999508678913} +{"train_loss": 0.017178183421492577, "grad_norm": 0.09145879745483398, "global_step": 27030, "epoch": 255, "lr": 4.842649974464002e-05} +{"train_loss": 0.014510354027152061, "grad_norm": 0.112470842897892, "global_step": 27031, "epoch": 255, "lr": 4.842353744871504e-05} +{"train_loss": 0.015487932600080967, "grad_norm": 0.09748625010251999, "global_step": 27032, "epoch": 255, "lr": 4.842057515832909e-05} +{"train_loss": 0.013089370913803577, "grad_norm": 0.10080975294113159, "global_step": 27033, "epoch": 255, "lr": 4.841761287349253e-05} +{"train_loss": 0.01625821925699711, "grad_norm": 0.07587277889251709, "global_step": 27034, "epoch": 255, "lr": 4.841465059421579e-05} +{"train_loss": 0.01673820987343788, "grad_norm": 0.09266403317451477, "global_step": 27035, "epoch": 255, "lr": 4.841168832050929e-05} +{"train_loss": 0.014226325787603855, "grad_norm": 0.08499171584844589, "global_step": 27036, "epoch": 255, "lr": 4.8408726052383405e-05} +{"train_loss": 0.01636473834514618, "grad_norm": 0.06129612773656845, "global_step": 27037, "epoch": 255, "lr": 4.840576378984858e-05} +{"train_loss": 0.015794314444065094, "grad_norm": 0.06708990782499313, "global_step": 27038, "epoch": 255, "lr": 4.840280153291522e-05} +{"train_loss": 0.014365369454026222, "grad_norm": 0.09724634140729904, "global_step": 27039, "epoch": 255, "lr": 4.839983928159371e-05} +{"train_loss": 0.01648661121726036, "grad_norm": 0.06394580751657486, "global_step": 27040, "epoch": 255, "lr": 4.839687703589445e-05} +{"train_loss": 0.01755364239215851, "grad_norm": 0.06988738477230072, "global_step": 27041, "epoch": 255, "lr": 4.8393914795827884e-05} +{"train_loss": 0.018267616629600525, "grad_norm": 0.08208626508712769, "global_step": 27042, "epoch": 255, "lr": 4.83909525614044e-05} +{"train_loss": 0.016359008848667145, "grad_norm": 0.0965951457619667, "global_step": 27043, "epoch": 255, "lr": 4.83879903326344e-05} +{"train_loss": 0.015682421624660492, "grad_norm": 0.12686622142791748, "global_step": 27044, "epoch": 255, "lr": 4.8385028109528335e-05} +{"train_loss": 0.01448790542781353, "grad_norm": 0.05701703205704689, "global_step": 27045, "epoch": 255, "lr": 4.8382065892096547e-05} +{"train_loss": 0.015875566750764847, "grad_norm": 0.08946141600608826, "global_step": 27046, "epoch": 255, "lr": 4.837910368034949e-05} +{"train_loss": 0.016634654253721237, "grad_norm": 0.062184158712625504, "global_step": 27047, "epoch": 255, "lr": 4.837614147429755e-05} +{"train_loss": 0.013586949557065964, "grad_norm": 0.10836378484964371, "global_step": 27048, "epoch": 255, "lr": 4.837317927395114e-05} +{"train_loss": 0.014913234859704971, "grad_norm": 0.07133977115154266, "global_step": 27049, "epoch": 255, "lr": 4.837021707932067e-05} +{"train_loss": 0.014630962163209915, "grad_norm": 0.09574051201343536, "global_step": 27050, "epoch": 255, "lr": 4.8367254890416566e-05} +{"train_loss": 0.014964910224080086, "grad_norm": 0.05721886828541756, "global_step": 27051, "epoch": 255, "lr": 4.8364292707249206e-05} +{"train_loss": 0.014484953135251999, "grad_norm": 0.08168306201696396, "global_step": 27052, "epoch": 255, "lr": 4.836133052982901e-05} +{"train_loss": 0.013981672003865242, "grad_norm": 0.05479294806718826, "global_step": 27053, "epoch": 255, "lr": 4.835836835816638e-05} +{"train_loss": 0.01479949802160263, "grad_norm": 0.05933842435479164, "global_step": 27054, "epoch": 255, "lr": 4.835540619227173e-05} +{"train_loss": 0.014186505228281021, "grad_norm": 0.09648239612579346, "global_step": 27055, "epoch": 255, "lr": 4.835244403215547e-05} +{"train_loss": 0.016303803771734238, "grad_norm": 0.06040021404623985, "global_step": 27056, "epoch": 255, "lr": 4.8349481877828026e-05} +{"train_loss": 0.014261464588344097, "grad_norm": 0.060963716357946396, "global_step": 27057, "epoch": 255, "lr": 4.834651972929975e-05} +{"train_loss": 0.014690767973661423, "grad_norm": 0.05422201380133629, "global_step": 27058, "epoch": 255, "lr": 4.834355758658111e-05} +{"train_loss": 0.016833454370498657, "grad_norm": 0.16186906397342682, "global_step": 27059, "epoch": 255, "lr": 4.834059544968248e-05} +{"train_loss": 0.01498730480670929, "grad_norm": 0.05957312881946564, "global_step": 27060, "epoch": 255, "lr": 4.833763331861427e-05} +{"train_loss": 0.014424684457480907, "grad_norm": 0.10747090727090836, "global_step": 27061, "epoch": 255, "lr": 4.833467119338692e-05} +{"train_loss": 0.015948979184031487, "grad_norm": 0.08821334689855576, "global_step": 27062, "epoch": 255, "lr": 4.833170907401078e-05} +{"train_loss": 0.015507757663726807, "grad_norm": 0.09594450145959854, "global_step": 27063, "epoch": 255, "lr": 4.83287469604963e-05} +{"train_loss": 0.01657899282872677, "grad_norm": 0.10451284795999527, "global_step": 27064, "epoch": 255, "lr": 4.832578485285388e-05} +{"train_loss": 0.01577325165271759, "grad_norm": 0.09921726584434509, "global_step": 27065, "epoch": 255, "lr": 4.832282275109391e-05} +{"train_loss": 0.016692638397216797, "grad_norm": 0.08857370167970657, "global_step": 27066, "epoch": 255, "lr": 4.831986065522682e-05} +{"train_loss": 0.016246002167463303, "grad_norm": 0.11887871474027634, "global_step": 27067, "epoch": 255, "lr": 4.831689856526302e-05} +{"train_loss": 0.015911925584077835, "grad_norm": 0.15107153356075287, "global_step": 27068, "epoch": 255, "lr": 4.83139364812129e-05} +{"train_loss": 0.01730470173060894, "grad_norm": 0.0851912871003151, "global_step": 27069, "epoch": 255, "lr": 4.831097440308685e-05} +{"train_loss": 0.016479970887303352, "grad_norm": 0.05791442096233368, "global_step": 27070, "epoch": 255, "lr": 4.830801233089533e-05} +{"train_loss": 0.01547444798052311, "grad_norm": 0.13708209991455078, "global_step": 27071, "epoch": 255, "lr": 4.830505026464871e-05} +{"train_loss": 0.015216870233416557, "grad_norm": 0.1203676387667656, "global_step": 27072, "epoch": 255, "lr": 4.83020882043574e-05} +{"train_loss": 0.01468040980398655, "grad_norm": 0.057956431061029434, "global_step": 27073, "epoch": 255, "lr": 4.829912615003184e-05} +{"train_loss": 0.015319923870265484, "grad_norm": 0.06700523942708969, "global_step": 27074, "epoch": 255, "lr": 4.8296164101682384e-05} +{"train_loss": 0.019174251705408096, "grad_norm": 0.10256398469209671, "global_step": 27075, "epoch": 255, "lr": 4.829320205931948e-05} +{"train_loss": 0.017918091267347336, "grad_norm": 0.06523875147104263, "global_step": 27076, "epoch": 255, "lr": 4.829024002295351e-05} +{"train_loss": 0.015965597704052925, "grad_norm": 0.08081074059009552, "global_step": 27077, "epoch": 255, "lr": 4.828727799259489e-05} +{"train_loss": 0.015949036926031113, "grad_norm": 0.08107151836156845, "global_step": 27078, "epoch": 255, "lr": 4.828431596825404e-05} +{"train_loss": 0.013764215633273125, "grad_norm": 0.08885249495506287, "global_step": 27079, "epoch": 255, "lr": 4.828135394994137e-05} +{"train_loss": 0.01508687436580658, "grad_norm": 0.1412552446126938, "global_step": 27080, "epoch": 255, "lr": 4.827839193766726e-05} +{"train_loss": 0.01596381887793541, "grad_norm": 0.11521361768245697, "global_step": 27081, "epoch": 255, "lr": 4.827542993144213e-05} +{"train_loss": 0.014210053719580173, "grad_norm": 0.06898178905248642, "global_step": 27082, "epoch": 255, "lr": 4.827246793127639e-05} +{"train_loss": 0.015310624614357948, "grad_norm": 0.09352125972509384, "global_step": 27083, "epoch": 255, "lr": 4.826950593718045e-05} +{"train_loss": 0.015509488061070442, "grad_norm": 0.09479362517595291, "global_step": 27084, "epoch": 255, "lr": 4.826654394916471e-05} +{"train_loss": 0.017111727967858315, "grad_norm": 0.13120752573013306, "global_step": 27085, "epoch": 255, "lr": 4.82635819672396e-05} +{"train_loss": 0.012086864560842514, "grad_norm": 0.07190384715795517, "global_step": 27086, "epoch": 255, "lr": 4.826061999141548e-05} +{"train_loss": 0.014555150642991066, "grad_norm": 0.0537632517516613, "global_step": 27087, "epoch": 255, "lr": 4.8257658021702796e-05} +{"train_loss": 0.016097083687782288, "grad_norm": 0.08156678080558777, "global_step": 27088, "epoch": 255, "lr": 4.825469605811194e-05} +{"train_loss": 0.01637152209877968, "grad_norm": 0.10651831328868866, "global_step": 27089, "epoch": 255, "lr": 4.825173410065332e-05} +{"train_loss": 0.01621396839618683, "grad_norm": 0.07273255288600922, "global_step": 27090, "epoch": 255, "lr": 4.824877214933736e-05} +{"train_loss": 0.01078038476407528, "grad_norm": 0.06141015142202377, "global_step": 27091, "epoch": 255, "lr": 4.824581020417445e-05} +{"train_loss": 0.01343509554862976, "grad_norm": 0.08225639909505844, "global_step": 27092, "epoch": 255, "lr": 4.8242848265175e-05} +{"train_loss": 0.015155859291553497, "grad_norm": 0.0656796470284462, "global_step": 27093, "epoch": 255, "lr": 4.823988633234942e-05} +{"train_loss": 0.01577860675752163, "grad_norm": 0.07614439725875854, "global_step": 27094, "epoch": 255, "lr": 4.823692440570809e-05} +{"train_loss": 0.014721282757818699, "grad_norm": 0.11577203869819641, "global_step": 27095, "epoch": 255, "lr": 4.8233962485261464e-05} +{"train_loss": 0.015229938551783562, "grad_norm": 0.13603848218917847, "global_step": 27096, "epoch": 255, "lr": 4.8231000571019915e-05} +{"train_loss": 0.017237920314073563, "grad_norm": 0.07383681833744049, "global_step": 27097, "epoch": 255, "lr": 4.822803866299388e-05} +{"train_loss": 0.016634086146950722, "grad_norm": 0.10740028321743011, "global_step": 27098, "epoch": 255, "lr": 4.822507676119372e-05} +{"train_loss": 0.012912801466882229, "grad_norm": 0.05576285719871521, "global_step": 27099, "epoch": 255, "lr": 4.8222114865629886e-05} +{"train_loss": 0.013839470222592354, "grad_norm": 0.053742699325084686, "global_step": 27100, "epoch": 255, "lr": 4.8219152976312756e-05} +{"train_loss": 0.01621933840215206, "grad_norm": 0.07235608249902725, "global_step": 27101, "epoch": 255, "lr": 4.821619109325275e-05} +{"train_loss": 0.015042012557387352, "grad_norm": 0.10828270018100739, "global_step": 27102, "epoch": 255, "lr": 4.821322921646027e-05} +{"train_loss": 0.015212241560220718, "grad_norm": 0.08533358573913574, "global_step": 27103, "epoch": 255, "lr": 4.821026734594575e-05} +{"train_loss": 0.017628319561481476, "grad_norm": 0.096245177090168, "global_step": 27104, "epoch": 255, "lr": 4.820730548171956e-05} +{"train_loss": 0.015994291752576828, "grad_norm": 0.09484022110700607, "global_step": 27105, "epoch": 255, "lr": 4.820434362379211e-05} +{"train_loss": 0.013293924741446972, "grad_norm": 0.0796395093202591, "global_step": 27106, "epoch": 255, "lr": 4.820138177217381e-05} +{"train_loss": 0.014817677438259125, "grad_norm": 0.07399557530879974, "global_step": 27107, "epoch": 255, "lr": 4.819841992687508e-05} +{"train_loss": 0.015999075025320053, "grad_norm": 0.06124669685959816, "global_step": 27108, "epoch": 255, "lr": 4.8195458087906334e-05} +{"train_loss": 0.015504956245422363, "grad_norm": 0.06897790729999542, "global_step": 27109, "epoch": 255, "lr": 4.819249625527795e-05} +{"train_loss": 0.016700344160199165, "grad_norm": 0.06815126538276672, "global_step": 27110, "epoch": 255, "lr": 4.8189534429000344e-05} +{"train_loss": 0.019572049379348755, "grad_norm": 0.07860194891691208, "global_step": 27111, "epoch": 255, "lr": 4.8186572609083924e-05} +{"train_loss": 0.014074587263166904, "grad_norm": 0.0987456813454628, "global_step": 27112, "epoch": 255, "lr": 4.81836107955391e-05} +{"train_loss": 0.017585186287760735, "grad_norm": 0.0928025022149086, "global_step": 27113, "epoch": 255, "lr": 4.8180648988376273e-05} +{"train_loss": 0.013868114911019802, "grad_norm": 0.09849300980567932, "global_step": 27114, "epoch": 255, "lr": 4.8177687187605894e-05} +{"train_loss": 0.016214651986956596, "grad_norm": 0.0706956759095192, "global_step": 27115, "epoch": 255, "lr": 4.8174725393238285e-05} +{"train_loss": 0.016587989404797554, "grad_norm": 0.07692666351795197, "global_step": 27116, "epoch": 255, "lr": 4.8171763605283914e-05} +{"train_loss": 0.016535375267267227, "grad_norm": 0.14961294829845428, "global_step": 27117, "epoch": 255, "lr": 4.8168801823753176e-05} +{"train_loss": 0.015569484792649746, "grad_norm": 0.07085084915161133, "global_step": 27118, "epoch": 255, "lr": 4.8165840048656454e-05} +{"train_loss": 0.01269165612757206, "grad_norm": 0.05566996708512306, "global_step": 27119, "epoch": 255, "lr": 4.816287828000418e-05} +{"train_loss": 0.015979861840605736, "grad_norm": 0.06293214857578278, "global_step": 27120, "epoch": 255, "lr": 4.815991651780677e-05} +{"train_loss": 0.016857873648405075, "grad_norm": 0.06956339627504349, "global_step": 27121, "epoch": 255, "lr": 4.8156954762074595e-05} +{"train_loss": 0.016019701957702637, "grad_norm": 0.1126973107457161, "global_step": 27122, "epoch": 255, "lr": 4.815399301281809e-05} +{"train_loss": 0.014695116318762302, "grad_norm": 0.07032357901334763, "global_step": 27123, "epoch": 255, "lr": 4.815103127004763e-05} +{"train_loss": 0.014746937900781631, "grad_norm": 0.11827671527862549, "global_step": 27124, "epoch": 255, "lr": 4.814806953377366e-05} +{"train_loss": 0.016040517017245293, "grad_norm": 0.09719901531934738, "global_step": 27125, "epoch": 255, "lr": 4.814510780400657e-05} +{"train_loss": 0.014707603491842747, "grad_norm": 0.053498949855566025, "global_step": 27126, "epoch": 255, "lr": 4.8142146080756765e-05} +{"train_loss": 0.014255248010158539, "grad_norm": 0.12453830987215042, "global_step": 27127, "epoch": 255, "lr": 4.813918436403463e-05} +{"train_loss": 0.0180574469268322, "grad_norm": 0.11952224373817444, "global_step": 27128, "epoch": 255, "lr": 4.813622265385061e-05} +{"train_loss": 0.01564428023993969, "grad_norm": 0.08847897499799728, "global_step": 27129, "epoch": 255, "lr": 4.813326095021509e-05} +{"train_loss": 0.01600000262260437, "grad_norm": 0.10448214411735535, "global_step": 27130, "epoch": 255, "lr": 4.813029925313847e-05} +{"train_loss": 0.015723831951618195, "grad_norm": 0.07510218769311905, "global_step": 27131, "epoch": 255, "lr": 4.812733756263118e-05} +{"train_loss": 0.016042431816458702, "grad_norm": 0.11542956531047821, "global_step": 27132, "epoch": 255, "lr": 4.812437587870361e-05} +{"train_loss": 0.014195320196449757, "grad_norm": 0.1422494500875473, "global_step": 27133, "epoch": 255, "lr": 4.8121414201366174e-05} +{"train_loss": 0.015362711623311043, "grad_norm": 0.07379265874624252, "global_step": 27134, "epoch": 255, "lr": 4.811845253062926e-05} +{"train_loss": 0.015536037660291735, "grad_norm": 0.08220957219600677, "global_step": 27135, "epoch": 255, "lr": 4.811549086650327e-05, "val_loss": 0.01919291540980339} +{"train_loss": 0.014443357475101948, "grad_norm": 0.11164244264364243, "global_step": 27136, "epoch": 256, "lr": 4.811252920899865e-05} +{"train_loss": 0.01318889670073986, "grad_norm": 0.07324323058128357, "global_step": 27137, "epoch": 256, "lr": 4.810956755812577e-05} +{"train_loss": 0.015087920241057873, "grad_norm": 0.07281924039125443, "global_step": 27138, "epoch": 256, "lr": 4.810660591389506e-05} +{"train_loss": 0.01637842319905758, "grad_norm": 0.08218259364366531, "global_step": 27139, "epoch": 256, "lr": 4.81036442763169e-05} +{"train_loss": 0.014349629171192646, "grad_norm": 0.115211121737957, "global_step": 27140, "epoch": 256, "lr": 4.810068264540171e-05} +{"train_loss": 0.01641380414366722, "grad_norm": 0.05626394972205162, "global_step": 27141, "epoch": 256, "lr": 4.80977210211599e-05} +{"train_loss": 0.018193725496530533, "grad_norm": 0.05771976336836815, "global_step": 27142, "epoch": 256, "lr": 4.809475940360186e-05} +{"train_loss": 0.01480780728161335, "grad_norm": 0.062399089336395264, "global_step": 27143, "epoch": 256, "lr": 4.809179779273802e-05} +{"train_loss": 0.01557351928204298, "grad_norm": 0.13757726550102234, "global_step": 27144, "epoch": 256, "lr": 4.808883618857878e-05} +{"train_loss": 0.015080582350492477, "grad_norm": 0.07142791152000427, "global_step": 27145, "epoch": 256, "lr": 4.8085874591134525e-05} +{"train_loss": 0.015420809388160706, "grad_norm": 0.0808967873454094, "global_step": 27146, "epoch": 256, "lr": 4.808291300041568e-05} +{"train_loss": 0.01779491826891899, "grad_norm": 0.10243230313062668, "global_step": 27147, "epoch": 256, "lr": 4.8079951416432635e-05} +{"train_loss": 0.01592651754617691, "grad_norm": 0.07133098691701889, "global_step": 27148, "epoch": 256, "lr": 4.8076989839195816e-05} +{"train_loss": 0.014949706383049488, "grad_norm": 0.05255034938454628, "global_step": 27149, "epoch": 256, "lr": 4.8074028268715616e-05} +{"train_loss": 0.016353439539670944, "grad_norm": 0.07294964790344238, "global_step": 27150, "epoch": 256, "lr": 4.807106670500245e-05} +{"train_loss": 0.01233509462326765, "grad_norm": 0.05503194034099579, "global_step": 27151, "epoch": 256, "lr": 4.8068105148066714e-05} +{"train_loss": 0.01525789499282837, "grad_norm": 0.09821762889623642, "global_step": 27152, "epoch": 256, "lr": 4.80651435979188e-05} +{"train_loss": 0.014569179154932499, "grad_norm": 0.07782482355833054, "global_step": 27153, "epoch": 256, "lr": 4.8062182054569135e-05} +{"train_loss": 0.01496786531060934, "grad_norm": 0.07359810173511505, "global_step": 27154, "epoch": 256, "lr": 4.805922051802813e-05} +{"train_loss": 0.015464246273040771, "grad_norm": 0.18019238114356995, "global_step": 27155, "epoch": 256, "lr": 4.805625898830617e-05} +{"train_loss": 0.015690281987190247, "grad_norm": 0.05759849026799202, "global_step": 27156, "epoch": 256, "lr": 4.8053297465413686e-05} +{"train_loss": 0.015511829406023026, "grad_norm": 0.11864272505044937, "global_step": 27157, "epoch": 256, "lr": 4.805033594936106e-05} +{"train_loss": 0.011905239894986153, "grad_norm": 0.1119086742401123, "global_step": 27158, "epoch": 256, "lr": 4.8047374440158705e-05} +{"train_loss": 0.01462823711335659, "grad_norm": 0.11163122951984406, "global_step": 27159, "epoch": 256, "lr": 4.804441293781702e-05} +{"train_loss": 0.015517800115048885, "grad_norm": 0.08503799885511398, "global_step": 27160, "epoch": 256, "lr": 4.804145144234642e-05} +{"train_loss": 0.016493719071149826, "grad_norm": 0.0799136608839035, "global_step": 27161, "epoch": 256, "lr": 4.8038489953757324e-05} +{"train_loss": 0.014977769926190376, "grad_norm": 0.07038149982690811, "global_step": 27162, "epoch": 256, "lr": 4.803552847206011e-05} +{"train_loss": 0.015815719962120056, "grad_norm": 0.11252664774656296, "global_step": 27163, "epoch": 256, "lr": 4.8032566997265196e-05} +{"train_loss": 0.016571417450904846, "grad_norm": 0.12464745342731476, "global_step": 27164, "epoch": 256, "lr": 4.802960552938297e-05} +{"train_loss": 0.013308961875736713, "grad_norm": 0.0531308650970459, "global_step": 27165, "epoch": 256, "lr": 4.802664406842387e-05} +{"train_loss": 0.017462056130170822, "grad_norm": 0.08031465113162994, "global_step": 27166, "epoch": 256, "lr": 4.802368261439828e-05} +{"train_loss": 0.015537803061306477, "grad_norm": 0.06521885842084885, "global_step": 27167, "epoch": 256, "lr": 4.8020721167316624e-05} +{"train_loss": 0.015332783572375774, "grad_norm": 0.06538698822259903, "global_step": 27168, "epoch": 256, "lr": 4.801775972718927e-05} +{"train_loss": 0.016978958621621132, "grad_norm": 0.07201644033193588, "global_step": 27169, "epoch": 256, "lr": 4.801479829402666e-05} +{"train_loss": 0.01460050605237484, "grad_norm": 0.15219762921333313, "global_step": 27170, "epoch": 256, "lr": 4.8011836867839185e-05} +{"train_loss": 0.015056555159389973, "grad_norm": 0.07356135547161102, "global_step": 27171, "epoch": 256, "lr": 4.800887544863723e-05} +{"train_loss": 0.01788492128252983, "grad_norm": 0.07367122173309326, "global_step": 27172, "epoch": 256, "lr": 4.8005914036431245e-05} +{"train_loss": 0.015065121464431286, "grad_norm": 0.05469347536563873, "global_step": 27173, "epoch": 256, "lr": 4.8002952631231615e-05} +{"train_loss": 0.013542501255869865, "grad_norm": 0.056157276034355164, "global_step": 27174, "epoch": 256, "lr": 4.799999123304873e-05} +{"train_loss": 0.01703239046037197, "grad_norm": 0.05231979489326477, "global_step": 27175, "epoch": 256, "lr": 4.7997029841893e-05} +{"train_loss": 0.01418243907392025, "grad_norm": 0.09868797659873962, "global_step": 27176, "epoch": 256, "lr": 4.799406845777484e-05} +{"train_loss": 0.014402616769075394, "grad_norm": 0.09513132274150848, "global_step": 27177, "epoch": 256, "lr": 4.799110708070466e-05} +{"train_loss": 0.015416277572512627, "grad_norm": 0.077757328748703, "global_step": 27178, "epoch": 256, "lr": 4.798814571069284e-05} +{"train_loss": 0.012312538921833038, "grad_norm": 0.0627814531326294, "global_step": 27179, "epoch": 256, "lr": 4.798518434774982e-05} +{"train_loss": 0.017360201105475426, "grad_norm": 0.07067642360925674, "global_step": 27180, "epoch": 256, "lr": 4.7982222991885986e-05} +{"train_loss": 0.014274952933192253, "grad_norm": 0.05952206254005432, "global_step": 27181, "epoch": 256, "lr": 4.797926164311172e-05} +{"train_loss": 0.0183984637260437, "grad_norm": 0.0658101737499237, "global_step": 27182, "epoch": 256, "lr": 4.797630030143747e-05} +{"train_loss": 0.01709935814142227, "grad_norm": 0.07025006413459778, "global_step": 27183, "epoch": 256, "lr": 4.797333896687361e-05} +{"train_loss": 0.014790253713726997, "grad_norm": 0.10849294066429138, "global_step": 27184, "epoch": 256, "lr": 4.7970377639430547e-05} +{"train_loss": 0.013820979744195938, "grad_norm": 0.08708370476961136, "global_step": 27185, "epoch": 256, "lr": 4.7967416319118716e-05} +{"train_loss": 0.016323642805218697, "grad_norm": 0.059841644018888474, "global_step": 27186, "epoch": 256, "lr": 4.79644550059485e-05} +{"train_loss": 0.016314759850502014, "grad_norm": 0.0650930106639862, "global_step": 27187, "epoch": 256, "lr": 4.796149369993029e-05} +{"train_loss": 0.01502471137791872, "grad_norm": 0.08249985426664352, "global_step": 27188, "epoch": 256, "lr": 4.7958532401074504e-05} +{"train_loss": 0.017049433663487434, "grad_norm": 0.09519532322883606, "global_step": 27189, "epoch": 256, "lr": 4.795557110939155e-05} +{"train_loss": 0.016490522772073746, "grad_norm": 0.07355707883834839, "global_step": 27190, "epoch": 256, "lr": 4.795260982489183e-05} +{"train_loss": 0.016079960390925407, "grad_norm": 0.05043838173151016, "global_step": 27191, "epoch": 256, "lr": 4.794964854758577e-05} +{"train_loss": 0.01597624458372593, "grad_norm": 0.07865741848945618, "global_step": 27192, "epoch": 256, "lr": 4.794668727748373e-05} +{"train_loss": 0.013781045563519001, "grad_norm": 0.04831182584166527, "global_step": 27193, "epoch": 256, "lr": 4.794372601459614e-05} +{"train_loss": 0.014714447781443596, "grad_norm": 0.10651496797800064, "global_step": 27194, "epoch": 256, "lr": 4.79407647589334e-05} +{"train_loss": 0.017898529767990112, "grad_norm": 0.14743547141551971, "global_step": 27195, "epoch": 256, "lr": 4.7937803510505926e-05} +{"train_loss": 0.016321327537298203, "grad_norm": 0.24778597056865692, "global_step": 27196, "epoch": 256, "lr": 4.7934842269324096e-05} +{"train_loss": 0.015955369919538498, "grad_norm": 0.07578878104686737, "global_step": 27197, "epoch": 256, "lr": 4.7931881035398356e-05} +{"train_loss": 0.013357330113649368, "grad_norm": 0.051651325076818466, "global_step": 27198, "epoch": 256, "lr": 4.792891980873908e-05} +{"train_loss": 0.014888869598507881, "grad_norm": 0.0557645745575428, "global_step": 27199, "epoch": 256, "lr": 4.7925958589356675e-05} +{"train_loss": 0.015475661493837833, "grad_norm": 0.07483727484941483, "global_step": 27200, "epoch": 256, "lr": 4.792299737726154e-05} +{"train_loss": 0.018113072961568832, "grad_norm": 0.06886593252420425, "global_step": 27201, "epoch": 256, "lr": 4.79200361724641e-05} +{"train_loss": 0.014860768802464008, "grad_norm": 0.07867275923490524, "global_step": 27202, "epoch": 256, "lr": 4.7917074974974744e-05} +{"train_loss": 0.014353587292134762, "grad_norm": 0.0747763141989708, "global_step": 27203, "epoch": 256, "lr": 4.79141137848039e-05} +{"train_loss": 0.014478206634521484, "grad_norm": 0.10057763755321503, "global_step": 27204, "epoch": 256, "lr": 4.791115260196194e-05} +{"train_loss": 0.017942149192094803, "grad_norm": 0.07526592910289764, "global_step": 27205, "epoch": 256, "lr": 4.790819142645927e-05} +{"train_loss": 0.016430281102657318, "grad_norm": 0.09797398746013641, "global_step": 27206, "epoch": 256, "lr": 4.790523025830631e-05} +{"train_loss": 0.014850165694952011, "grad_norm": 0.154933899641037, "global_step": 27207, "epoch": 256, "lr": 4.790226909751347e-05} +{"train_loss": 0.013956088572740555, "grad_norm": 0.05398786440491676, "global_step": 27208, "epoch": 256, "lr": 4.789930794409115e-05} +{"train_loss": 0.015702415257692337, "grad_norm": 0.13927143812179565, "global_step": 27209, "epoch": 256, "lr": 4.789634679804973e-05} +{"train_loss": 0.015311058610677719, "grad_norm": 0.11248670518398285, "global_step": 27210, "epoch": 256, "lr": 4.789338565939963e-05} +{"train_loss": 0.015133881941437721, "grad_norm": 0.08579976856708527, "global_step": 27211, "epoch": 256, "lr": 4.789042452815127e-05} +{"train_loss": 0.016383904963731766, "grad_norm": 0.05513418838381767, "global_step": 27212, "epoch": 256, "lr": 4.7887463404315034e-05} +{"train_loss": 0.016019344329833984, "grad_norm": 0.12269948422908783, "global_step": 27213, "epoch": 256, "lr": 4.7884502287901326e-05} +{"train_loss": 0.015134770423173904, "grad_norm": 0.053344402462244034, "global_step": 27214, "epoch": 256, "lr": 4.788154117892058e-05} +{"train_loss": 0.014780016615986824, "grad_norm": 0.13856033980846405, "global_step": 27215, "epoch": 256, "lr": 4.7878580077383166e-05} +{"train_loss": 0.01840377226471901, "grad_norm": 0.10984466969966888, "global_step": 27216, "epoch": 256, "lr": 4.78756189832995e-05} +{"train_loss": 0.016699964180588722, "grad_norm": 0.0876447930932045, "global_step": 27217, "epoch": 256, "lr": 4.787265789667997e-05} +{"train_loss": 0.014871563762426376, "grad_norm": 0.07509278506040573, "global_step": 27218, "epoch": 256, "lr": 4.7869696817535e-05} +{"train_loss": 0.015539905056357384, "grad_norm": 0.10445340722799301, "global_step": 27219, "epoch": 256, "lr": 4.7866735745875004e-05} +{"train_loss": 0.016867050901055336, "grad_norm": 0.09050735086202621, "global_step": 27220, "epoch": 256, "lr": 4.7863774681710374e-05} +{"train_loss": 0.014763490296900272, "grad_norm": 0.1297142505645752, "global_step": 27221, "epoch": 256, "lr": 4.78608136250515e-05} +{"train_loss": 0.015081003308296204, "grad_norm": 0.08357680588960648, "global_step": 27222, "epoch": 256, "lr": 4.785785257590878e-05} +{"train_loss": 0.014295175671577454, "grad_norm": 0.09326208382844925, "global_step": 27223, "epoch": 256, "lr": 4.785489153429265e-05} +{"train_loss": 0.014462298713624477, "grad_norm": 0.09870059788227081, "global_step": 27224, "epoch": 256, "lr": 4.78519305002135e-05} +{"train_loss": 0.01418071798980236, "grad_norm": 0.1595362275838852, "global_step": 27225, "epoch": 256, "lr": 4.784896947368172e-05} +{"train_loss": 0.016320424154400826, "grad_norm": 0.07458250224590302, "global_step": 27226, "epoch": 256, "lr": 4.784600845470775e-05} +{"train_loss": 0.01608026772737503, "grad_norm": 0.057631831616163254, "global_step": 27227, "epoch": 256, "lr": 4.784304744330196e-05} +{"train_loss": 0.013409433886408806, "grad_norm": 0.08066391944885254, "global_step": 27228, "epoch": 256, "lr": 4.7840086439474755e-05} +{"train_loss": 0.017238255590200424, "grad_norm": 0.09628763794898987, "global_step": 27229, "epoch": 256, "lr": 4.7837125443236544e-05} +{"train_loss": 0.015634991228580475, "grad_norm": 0.11130910366773605, "global_step": 27230, "epoch": 256, "lr": 4.783416445459774e-05} +{"train_loss": 0.016675960272550583, "grad_norm": 0.06740988790988922, "global_step": 27231, "epoch": 256, "lr": 4.783120347356874e-05} +{"train_loss": 0.014744254760444164, "grad_norm": 0.073246531188488, "global_step": 27232, "epoch": 256, "lr": 4.782824250015996e-05} +{"train_loss": 0.01628001779317856, "grad_norm": 0.06613511592149734, "global_step": 27233, "epoch": 256, "lr": 4.782528153438178e-05} +{"train_loss": 0.01667032390832901, "grad_norm": 0.06939082592725754, "global_step": 27234, "epoch": 256, "lr": 4.7822320576244604e-05} +{"train_loss": 0.013783716596662998, "grad_norm": 0.11032672971487045, "global_step": 27235, "epoch": 256, "lr": 4.781935962575886e-05} +{"train_loss": 0.01441770140081644, "grad_norm": 0.08714119344949722, "global_step": 27236, "epoch": 256, "lr": 4.7816398682934937e-05} +{"train_loss": 0.017093965783715248, "grad_norm": 0.10951925814151764, "global_step": 27237, "epoch": 256, "lr": 4.7813437747783224e-05} +{"train_loss": 0.01555576641112566, "grad_norm": 0.09107640385627747, "global_step": 27238, "epoch": 256, "lr": 4.7810476820314174e-05} +{"train_loss": 0.017526816576719284, "grad_norm": 0.09502002596855164, "global_step": 27239, "epoch": 256, "lr": 4.780751590053813e-05} +{"train_loss": 0.017859570682048798, "grad_norm": 0.11711487919092178, "global_step": 27240, "epoch": 256, "lr": 4.780455498846553e-05} +{"train_loss": 0.015538690796227387, "grad_norm": 0.13307783007621765, "global_step": 27241, "epoch": 256, "lr": 4.7801594084106763e-05, "val_loss": 0.01950847916305065} +{"train_loss": 0.015596050769090652, "grad_norm": 0.1176757737994194, "global_step": 27242, "epoch": 257, "lr": 4.779863318747223e-05} +{"train_loss": 0.01444997452199459, "grad_norm": 0.14548437297344208, "global_step": 27243, "epoch": 257, "lr": 4.7795672298572355e-05} +{"train_loss": 0.014370481483638287, "grad_norm": 0.07437202334403992, "global_step": 27244, "epoch": 257, "lr": 4.779271141741754e-05} +{"train_loss": 0.014506797306239605, "grad_norm": 0.10969436168670654, "global_step": 27245, "epoch": 257, "lr": 4.778975054401817e-05} +{"train_loss": 0.015587844885885715, "grad_norm": 0.08036737889051437, "global_step": 27246, "epoch": 257, "lr": 4.778678967838463e-05} +{"train_loss": 0.01474457886070013, "grad_norm": 0.0773569643497467, "global_step": 27247, "epoch": 257, "lr": 4.7783828820527373e-05} +{"train_loss": 0.015012037009000778, "grad_norm": 0.061546728014945984, "global_step": 27248, "epoch": 257, "lr": 4.7780867970456786e-05} +{"train_loss": 0.014530787244439125, "grad_norm": 0.061698127537965775, "global_step": 27249, "epoch": 257, "lr": 4.777790712818324e-05} +{"train_loss": 0.01496361754834652, "grad_norm": 0.06757310032844543, "global_step": 27250, "epoch": 257, "lr": 4.7774946293717194e-05} +{"train_loss": 0.01583496294915676, "grad_norm": 0.17151996493339539, "global_step": 27251, "epoch": 257, "lr": 4.7771985467068985e-05} +{"train_loss": 0.014408577233552933, "grad_norm": 0.1173228845000267, "global_step": 27252, "epoch": 257, "lr": 4.776902464824907e-05} +{"train_loss": 0.015386318787932396, "grad_norm": 0.17265595495700836, "global_step": 27253, "epoch": 257, "lr": 4.7766063837267825e-05} +{"train_loss": 0.0185776986181736, "grad_norm": 0.15757760405540466, "global_step": 27254, "epoch": 257, "lr": 4.776310303413565e-05} +{"train_loss": 0.015415722504258156, "grad_norm": 0.09566076099872589, "global_step": 27255, "epoch": 257, "lr": 4.776014223886297e-05} +{"train_loss": 0.016084175556898117, "grad_norm": 0.14420144259929657, "global_step": 27256, "epoch": 257, "lr": 4.7757181451460184e-05} +{"train_loss": 0.01367906853556633, "grad_norm": 0.07320311665534973, "global_step": 27257, "epoch": 257, "lr": 4.7754220671937684e-05} +{"train_loss": 0.016990268602967262, "grad_norm": 0.09702218323945999, "global_step": 27258, "epoch": 257, "lr": 4.7751259900305864e-05} +{"train_loss": 0.012966133654117584, "grad_norm": 0.06095248460769653, "global_step": 27259, "epoch": 257, "lr": 4.774829913657514e-05} +{"train_loss": 0.016375362873077393, "grad_norm": 0.07737415283918381, "global_step": 27260, "epoch": 257, "lr": 4.774533838075592e-05} +{"train_loss": 0.020025594159960747, "grad_norm": 0.07046809792518616, "global_step": 27261, "epoch": 257, "lr": 4.7742377632858616e-05} +{"train_loss": 0.013814638368785381, "grad_norm": 0.05535393953323364, "global_step": 27262, "epoch": 257, "lr": 4.77394168928936e-05} +{"train_loss": 0.014453805983066559, "grad_norm": 0.07628555595874786, "global_step": 27263, "epoch": 257, "lr": 4.7736456160871275e-05} +{"train_loss": 0.012809052132070065, "grad_norm": 0.054407328367233276, "global_step": 27264, "epoch": 257, "lr": 4.773349543680207e-05} +{"train_loss": 0.01695634052157402, "grad_norm": 0.05824476480484009, "global_step": 27265, "epoch": 257, "lr": 4.773053472069638e-05} +{"train_loss": 0.014849819242954254, "grad_norm": 0.11737697571516037, "global_step": 27266, "epoch": 257, "lr": 4.772757401256459e-05} +{"train_loss": 0.016617922112345695, "grad_norm": 0.11098139733076096, "global_step": 27267, "epoch": 257, "lr": 4.7724613312417155e-05} +{"train_loss": 0.015101109631359577, "grad_norm": 0.1870907098054886, "global_step": 27268, "epoch": 257, "lr": 4.77216526202644e-05} +{"train_loss": 0.015191360376775265, "grad_norm": 0.07710718363523483, "global_step": 27269, "epoch": 257, "lr": 4.771869193611678e-05} +{"train_loss": 0.01417018473148346, "grad_norm": 0.15343578159809113, "global_step": 27270, "epoch": 257, "lr": 4.771573125998468e-05} +{"train_loss": 0.0184329766780138, "grad_norm": 0.07539411634206772, "global_step": 27271, "epoch": 257, "lr": 4.771277059187851e-05} +{"train_loss": 0.016445541754364967, "grad_norm": 0.07502000033855438, "global_step": 27272, "epoch": 257, "lr": 4.770980993180867e-05} +{"train_loss": 0.018703782930970192, "grad_norm": 0.07560452818870544, "global_step": 27273, "epoch": 257, "lr": 4.7706849279785577e-05} +{"train_loss": 0.018453124910593033, "grad_norm": 0.0902031883597374, "global_step": 27274, "epoch": 257, "lr": 4.7703888635819606e-05} +{"train_loss": 0.018506161868572235, "grad_norm": 0.08618109673261642, "global_step": 27275, "epoch": 257, "lr": 4.770092799992116e-05} +{"train_loss": 0.015374580398201942, "grad_norm": 0.08270756155252457, "global_step": 27276, "epoch": 257, "lr": 4.7697967372100665e-05} +{"train_loss": 0.014875564724206924, "grad_norm": 0.06014026328921318, "global_step": 27277, "epoch": 257, "lr": 4.769500675236851e-05} +{"train_loss": 0.01601407676935196, "grad_norm": 0.08284439891576767, "global_step": 27278, "epoch": 257, "lr": 4.769204614073509e-05} +{"train_loss": 0.016748100519180298, "grad_norm": 0.05263080075383186, "global_step": 27279, "epoch": 257, "lr": 4.768908553721085e-05} +{"train_loss": 0.016120117157697678, "grad_norm": 0.06356018036603928, "global_step": 27280, "epoch": 257, "lr": 4.768612494180612e-05} +{"train_loss": 0.016214055940508842, "grad_norm": 0.06005898863077164, "global_step": 27281, "epoch": 257, "lr": 4.768316435453135e-05} +{"train_loss": 0.01718638464808464, "grad_norm": 0.09908659011125565, "global_step": 27282, "epoch": 257, "lr": 4.7680203775396945e-05} +{"train_loss": 0.01427658274769783, "grad_norm": 0.08033977448940277, "global_step": 27283, "epoch": 257, "lr": 4.767724320441327e-05} +{"train_loss": 0.01492829155176878, "grad_norm": 0.09278520941734314, "global_step": 27284, "epoch": 257, "lr": 4.7674282641590774e-05} +{"train_loss": 0.014627265743911266, "grad_norm": 0.057429540902376175, "global_step": 27285, "epoch": 257, "lr": 4.767132208693984e-05} +{"train_loss": 0.016633179038763046, "grad_norm": 0.0877266377210617, "global_step": 27286, "epoch": 257, "lr": 4.7668361540470865e-05} +{"train_loss": 0.019224535673856735, "grad_norm": 0.11322962492704391, "global_step": 27287, "epoch": 257, "lr": 4.766540100219424e-05} +{"train_loss": 0.013114817440509796, "grad_norm": 0.06423217803239822, "global_step": 27288, "epoch": 257, "lr": 4.7662440472120385e-05} +{"train_loss": 0.01619713008403778, "grad_norm": 0.0983363464474678, "global_step": 27289, "epoch": 257, "lr": 4.76594799502597e-05} +{"train_loss": 0.013308985158801079, "grad_norm": 0.09181281179189682, "global_step": 27290, "epoch": 257, "lr": 4.765651943662258e-05} +{"train_loss": 0.016422031447291374, "grad_norm": 0.06969121843576431, "global_step": 27291, "epoch": 257, "lr": 4.765355893121946e-05} +{"train_loss": 0.01752086728811264, "grad_norm": 0.10784067213535309, "global_step": 27292, "epoch": 257, "lr": 4.765059843406068e-05} +{"train_loss": 0.013059169985353947, "grad_norm": 0.0744423121213913, "global_step": 27293, "epoch": 257, "lr": 4.7647637945156695e-05} +{"train_loss": 0.016917873173952103, "grad_norm": 0.09594251960515976, "global_step": 27294, "epoch": 257, "lr": 4.7644677464517874e-05} +{"train_loss": 0.01496857963502407, "grad_norm": 0.05581444501876831, "global_step": 27295, "epoch": 257, "lr": 4.7641716992154626e-05} +{"train_loss": 0.013054070994257927, "grad_norm": 0.0869864970445633, "global_step": 27296, "epoch": 257, "lr": 4.763875652807737e-05} +{"train_loss": 0.014618908986449242, "grad_norm": 0.09173531085252762, "global_step": 27297, "epoch": 257, "lr": 4.763579607229651e-05} +{"train_loss": 0.016790736466646194, "grad_norm": 0.12061993777751923, "global_step": 27298, "epoch": 257, "lr": 4.7632835624822424e-05} +{"train_loss": 0.015640081837773323, "grad_norm": 0.0796779915690422, "global_step": 27299, "epoch": 257, "lr": 4.762987518566551e-05} +{"train_loss": 0.01537314523011446, "grad_norm": 0.05654338374733925, "global_step": 27300, "epoch": 257, "lr": 4.762691475483619e-05} +{"train_loss": 0.014576482586562634, "grad_norm": 0.050485238432884216, "global_step": 27301, "epoch": 257, "lr": 4.7623954332344874e-05} +{"train_loss": 0.014984790235757828, "grad_norm": 0.09573233127593994, "global_step": 27302, "epoch": 257, "lr": 4.762099391820193e-05} +{"train_loss": 0.013920358382165432, "grad_norm": 0.062122978270053864, "global_step": 27303, "epoch": 257, "lr": 4.761803351241781e-05} +{"train_loss": 0.015507224947214127, "grad_norm": 0.05483656004071236, "global_step": 27304, "epoch": 257, "lr": 4.761507311500285e-05} +{"train_loss": 0.014112930744886398, "grad_norm": 0.05483631417155266, "global_step": 27305, "epoch": 257, "lr": 4.76121127259675e-05} +{"train_loss": 0.017327498644590378, "grad_norm": 0.08294110745191574, "global_step": 27306, "epoch": 257, "lr": 4.7609152345322146e-05} +{"train_loss": 0.01597301848232746, "grad_norm": 0.07623185962438583, "global_step": 27307, "epoch": 257, "lr": 4.760619197307718e-05} +{"train_loss": 0.014032341539859772, "grad_norm": 0.0756271481513977, "global_step": 27308, "epoch": 257, "lr": 4.7603231609243026e-05} +{"train_loss": 0.01598535291850567, "grad_norm": 0.057864584028720856, "global_step": 27309, "epoch": 257, "lr": 4.7600271253830085e-05} +{"train_loss": 0.015103533864021301, "grad_norm": 0.07018385827541351, "global_step": 27310, "epoch": 257, "lr": 4.759731090684874e-05} +{"train_loss": 0.015585263259708881, "grad_norm": 0.06485629081726074, "global_step": 27311, "epoch": 257, "lr": 4.759435056830939e-05} +{"train_loss": 0.015799911692738533, "grad_norm": 0.1062554195523262, "global_step": 27312, "epoch": 257, "lr": 4.7591390238222445e-05} +{"train_loss": 0.013394663110375404, "grad_norm": 0.05237482115626335, "global_step": 27313, "epoch": 257, "lr": 4.758842991659831e-05} +{"train_loss": 0.015509242191910744, "grad_norm": 0.04959177225828171, "global_step": 27314, "epoch": 257, "lr": 4.75854696034474e-05} +{"train_loss": 0.01869272254407406, "grad_norm": 0.06798722594976425, "global_step": 27315, "epoch": 257, "lr": 4.758250929878009e-05} +{"train_loss": 0.014333890751004219, "grad_norm": 0.14671467244625092, "global_step": 27316, "epoch": 257, "lr": 4.7579549002606776e-05} +{"train_loss": 0.014152130112051964, "grad_norm": 0.14210984110832214, "global_step": 27317, "epoch": 257, "lr": 4.757658871493789e-05} +{"train_loss": 0.015141350217163563, "grad_norm": 0.06916411966085434, "global_step": 27318, "epoch": 257, "lr": 4.757362843578381e-05} +{"train_loss": 0.018593285232782364, "grad_norm": 0.12304867804050446, "global_step": 27319, "epoch": 257, "lr": 4.757066816515494e-05} +{"train_loss": 0.015067989937961102, "grad_norm": 0.13451607525348663, "global_step": 27320, "epoch": 257, "lr": 4.756770790306171e-05} +{"train_loss": 0.015948249027132988, "grad_norm": 0.08329768478870392, "global_step": 27321, "epoch": 257, "lr": 4.756474764951448e-05} +{"train_loss": 0.01431027241051197, "grad_norm": 0.13024355471134186, "global_step": 27322, "epoch": 257, "lr": 4.7561787404523664e-05} +{"train_loss": 0.015287136659026146, "grad_norm": 0.05406413599848747, "global_step": 27323, "epoch": 257, "lr": 4.755882716809967e-05} +{"train_loss": 0.01615801639854908, "grad_norm": 0.1670663058757782, "global_step": 27324, "epoch": 257, "lr": 4.7555866940252884e-05} +{"train_loss": 0.012837553396821022, "grad_norm": 0.09665976464748383, "global_step": 27325, "epoch": 257, "lr": 4.755290672099373e-05} +{"train_loss": 0.01598363369703293, "grad_norm": 0.08979135751724243, "global_step": 27326, "epoch": 257, "lr": 4.754994651033261e-05} +{"train_loss": 0.01423923671245575, "grad_norm": 0.06656324863433838, "global_step": 27327, "epoch": 257, "lr": 4.7546986308279903e-05} +{"train_loss": 0.01306843850761652, "grad_norm": 0.0689968690276146, "global_step": 27328, "epoch": 257, "lr": 4.754402611484601e-05} +{"train_loss": 0.016967426985502243, "grad_norm": 0.11669765412807465, "global_step": 27329, "epoch": 257, "lr": 4.7541065930041345e-05} +{"train_loss": 0.014859789982438087, "grad_norm": 0.06204671412706375, "global_step": 27330, "epoch": 257, "lr": 4.753810575387631e-05} +{"train_loss": 0.014861016534268856, "grad_norm": 0.12003457546234131, "global_step": 27331, "epoch": 257, "lr": 4.753514558636128e-05} +{"train_loss": 0.014024005271494389, "grad_norm": 0.0721222385764122, "global_step": 27332, "epoch": 257, "lr": 4.753218542750673e-05} +{"train_loss": 0.01670980453491211, "grad_norm": 0.0725896954536438, "global_step": 27333, "epoch": 257, "lr": 4.752922527732296e-05} +{"train_loss": 0.01213136501610279, "grad_norm": 0.10145766288042068, "global_step": 27334, "epoch": 257, "lr": 4.752626513582043e-05} +{"train_loss": 0.01853397861123085, "grad_norm": 0.06784190982580185, "global_step": 27335, "epoch": 257, "lr": 4.752330500300953e-05} +{"train_loss": 0.015320960432291031, "grad_norm": 0.13156673312187195, "global_step": 27336, "epoch": 257, "lr": 4.7520344878900644e-05} +{"train_loss": 0.016269173473119736, "grad_norm": 0.15490496158599854, "global_step": 27337, "epoch": 257, "lr": 4.7517384763504206e-05} +{"train_loss": 0.014341216534376144, "grad_norm": 0.0928618535399437, "global_step": 27338, "epoch": 257, "lr": 4.75144246568306e-05} +{"train_loss": 0.015473616309463978, "grad_norm": 0.10845282673835754, "global_step": 27339, "epoch": 257, "lr": 4.7511464558890223e-05} +{"train_loss": 0.014103984460234642, "grad_norm": 0.13528606295585632, "global_step": 27340, "epoch": 257, "lr": 4.750850446969347e-05} +{"train_loss": 0.01870468072593212, "grad_norm": 0.15424631536006927, "global_step": 27341, "epoch": 257, "lr": 4.7505544389250747e-05} +{"train_loss": 0.0144585482776165, "grad_norm": 0.07112596929073334, "global_step": 27342, "epoch": 257, "lr": 4.750258431757246e-05} +{"train_loss": 0.018057966604828835, "grad_norm": 0.0840681791305542, "global_step": 27343, "epoch": 257, "lr": 4.749962425466901e-05} +{"train_loss": 0.016445210203528404, "grad_norm": 0.06430317461490631, "global_step": 27344, "epoch": 257, "lr": 4.74966642005508e-05} +{"train_loss": 0.014497745782136917, "grad_norm": 0.057973477989435196, "global_step": 27345, "epoch": 257, "lr": 4.74937041552282e-05} +{"train_loss": 0.016602367162704468, "grad_norm": 0.0662020668387413, "global_step": 27346, "epoch": 257, "lr": 4.749074411871165e-05} +{"train_loss": 0.015526386812540158, "grad_norm": 0.10228434950113297, "global_step": 27347, "epoch": 257, "lr": 4.748778409101153e-05, "val_loss": 0.01946510747075081} +{"train_loss": 0.015602117404341698, "grad_norm": 0.09410811960697174, "global_step": 27348, "epoch": 258, "lr": 4.7484824072138234e-05} +{"train_loss": 0.018547646701335907, "grad_norm": 0.10079379379749298, "global_step": 27349, "epoch": 258, "lr": 4.7481864062102184e-05} +{"train_loss": 0.014984939247369766, "grad_norm": 0.08823216706514359, "global_step": 27350, "epoch": 258, "lr": 4.747890406091377e-05} +{"train_loss": 0.01521449163556099, "grad_norm": 0.07326964288949966, "global_step": 27351, "epoch": 258, "lr": 4.7475944068583395e-05} +{"train_loss": 0.01296248473227024, "grad_norm": 0.11043375730514526, "global_step": 27352, "epoch": 258, "lr": 4.747298408512145e-05} +{"train_loss": 0.015162971802055836, "grad_norm": 0.0692199319601059, "global_step": 27353, "epoch": 258, "lr": 4.747002411053833e-05} +{"train_loss": 0.01558606419712305, "grad_norm": 0.0637301355600357, "global_step": 27354, "epoch": 258, "lr": 4.746706414484445e-05} +{"train_loss": 0.015515772625803947, "grad_norm": 0.09873977303504944, "global_step": 27355, "epoch": 258, "lr": 4.7464104188050214e-05} +{"train_loss": 0.01297544501721859, "grad_norm": 0.0825556144118309, "global_step": 27356, "epoch": 258, "lr": 4.746114424016602e-05} +{"train_loss": 0.014278425835072994, "grad_norm": 0.15756383538246155, "global_step": 27357, "epoch": 258, "lr": 4.7458184301202235e-05} +{"train_loss": 0.015709947794675827, "grad_norm": 0.12319222837686539, "global_step": 27358, "epoch": 258, "lr": 4.74552243711693e-05} +{"train_loss": 0.016917206346988678, "grad_norm": 0.1031629666686058, "global_step": 27359, "epoch": 258, "lr": 4.74522644500776e-05} +{"train_loss": 0.017152119427919388, "grad_norm": 0.07751347124576569, "global_step": 27360, "epoch": 258, "lr": 4.7449304537937524e-05} +{"train_loss": 0.017815304920077324, "grad_norm": 0.10947040468454361, "global_step": 27361, "epoch": 258, "lr": 4.744634463475951e-05} +{"train_loss": 0.01251286081969738, "grad_norm": 0.0684458464384079, "global_step": 27362, "epoch": 258, "lr": 4.74433847405539e-05} +{"train_loss": 0.014791663736104965, "grad_norm": 0.07353179156780243, "global_step": 27363, "epoch": 258, "lr": 4.744042485533114e-05} +{"train_loss": 0.015809576958417892, "grad_norm": 0.05219363793730736, "global_step": 27364, "epoch": 258, "lr": 4.743746497910161e-05} +{"train_loss": 0.014991730451583862, "grad_norm": 0.08139792829751968, "global_step": 27365, "epoch": 258, "lr": 4.743450511187571e-05} +{"train_loss": 0.01696006767451763, "grad_norm": 0.06635311990976334, "global_step": 27366, "epoch": 258, "lr": 4.743154525366384e-05} +{"train_loss": 0.014636117964982986, "grad_norm": 0.07301940768957138, "global_step": 27367, "epoch": 258, "lr": 4.742858540447642e-05} +{"train_loss": 0.014358035288751125, "grad_norm": 0.10297663509845734, "global_step": 27368, "epoch": 258, "lr": 4.742562556432382e-05} +{"train_loss": 0.01478649117052555, "grad_norm": 0.08000275492668152, "global_step": 27369, "epoch": 258, "lr": 4.742266573321646e-05} +{"train_loss": 0.01390901580452919, "grad_norm": 0.07431997358798981, "global_step": 27370, "epoch": 258, "lr": 4.7419705911164715e-05} +{"train_loss": 0.014419494196772575, "grad_norm": 0.05379346385598183, "global_step": 27371, "epoch": 258, "lr": 4.7416746098179015e-05} +{"train_loss": 0.015567674301564693, "grad_norm": 0.12171190232038498, "global_step": 27372, "epoch": 258, "lr": 4.7413786294269734e-05} +{"train_loss": 0.016038350760936737, "grad_norm": 0.05844999849796295, "global_step": 27373, "epoch": 258, "lr": 4.7410826499447314e-05} +{"train_loss": 0.01531872060149908, "grad_norm": 0.07189171016216278, "global_step": 27374, "epoch": 258, "lr": 4.7407866713722095e-05} +{"train_loss": 0.013215521350502968, "grad_norm": 0.08240152895450592, "global_step": 27375, "epoch": 258, "lr": 4.740490693710451e-05} +{"train_loss": 0.015114536508917809, "grad_norm": 0.06281284242868423, "global_step": 27376, "epoch": 258, "lr": 4.740194716960496e-05} +{"train_loss": 0.017106402665376663, "grad_norm": 0.07359471917152405, "global_step": 27377, "epoch": 258, "lr": 4.739898741123382e-05} +{"train_loss": 0.013773111626505852, "grad_norm": 0.06659791618585587, "global_step": 27378, "epoch": 258, "lr": 4.7396027662001526e-05} +{"train_loss": 0.01524051371961832, "grad_norm": 0.11872658878564835, "global_step": 27379, "epoch": 258, "lr": 4.739306792191846e-05} +{"train_loss": 0.014549126848578453, "grad_norm": 0.08802451193332672, "global_step": 27380, "epoch": 258, "lr": 4.7390108190995014e-05} +{"train_loss": 0.012214560061693192, "grad_norm": 0.0628180131316185, "global_step": 27381, "epoch": 258, "lr": 4.738714846924159e-05} +{"train_loss": 0.013448754325509071, "grad_norm": 0.06244463846087456, "global_step": 27382, "epoch": 258, "lr": 4.7384188756668586e-05} +{"train_loss": 0.017732743173837662, "grad_norm": 0.08053220063447952, "global_step": 27383, "epoch": 258, "lr": 4.738122905328641e-05} +{"train_loss": 0.01467967126518488, "grad_norm": 0.1451992690563202, "global_step": 27384, "epoch": 258, "lr": 4.737826935910546e-05} +{"train_loss": 0.016026655212044716, "grad_norm": 0.06816601753234863, "global_step": 27385, "epoch": 258, "lr": 4.737530967413614e-05} +{"train_loss": 0.018743839114904404, "grad_norm": 0.19503812491893768, "global_step": 27386, "epoch": 258, "lr": 4.737234999838882e-05} +{"train_loss": 0.015333415940403938, "grad_norm": 0.11117490381002426, "global_step": 27387, "epoch": 258, "lr": 4.7369390331873926e-05} +{"train_loss": 0.014221149496734142, "grad_norm": 0.120201475918293, "global_step": 27388, "epoch": 258, "lr": 4.736643067460185e-05} +{"train_loss": 0.017649389803409576, "grad_norm": 0.086610808968544, "global_step": 27389, "epoch": 258, "lr": 4.7363471026582986e-05} +{"train_loss": 0.01715730130672455, "grad_norm": 0.11254855990409851, "global_step": 27390, "epoch": 258, "lr": 4.736051138782774e-05} +{"train_loss": 0.015368949621915817, "grad_norm": 0.11073161661624908, "global_step": 27391, "epoch": 258, "lr": 4.735755175834653e-05} +{"train_loss": 0.01641571894288063, "grad_norm": 0.07862737029790878, "global_step": 27392, "epoch": 258, "lr": 4.7354592138149726e-05} +{"train_loss": 0.01668032817542553, "grad_norm": 0.057915907353162766, "global_step": 27393, "epoch": 258, "lr": 4.735163252724772e-05} +{"train_loss": 0.015635155141353607, "grad_norm": 0.10374761372804642, "global_step": 27394, "epoch": 258, "lr": 4.7348672925650926e-05} +{"train_loss": 0.016057375818490982, "grad_norm": 0.09054160863161087, "global_step": 27395, "epoch": 258, "lr": 4.734571333336975e-05} +{"train_loss": 0.014489457942545414, "grad_norm": 0.05243081972002983, "global_step": 27396, "epoch": 258, "lr": 4.7342753750414584e-05} +{"train_loss": 0.016705654561519623, "grad_norm": 0.09729719161987305, "global_step": 27397, "epoch": 258, "lr": 4.7339794176795834e-05} +{"train_loss": 0.014848945662379265, "grad_norm": 0.09505927562713623, "global_step": 27398, "epoch": 258, "lr": 4.7336834612523883e-05} +{"train_loss": 0.01741463877260685, "grad_norm": 0.11354436725378036, "global_step": 27399, "epoch": 258, "lr": 4.7333875057609126e-05} +{"train_loss": 0.015364773571491241, "grad_norm": 0.22926758229732513, "global_step": 27400, "epoch": 258, "lr": 4.7330915512061976e-05} +{"train_loss": 0.016593851149082184, "grad_norm": 0.09212099015712738, "global_step": 27401, "epoch": 258, "lr": 4.732795597589283e-05} +{"train_loss": 0.01624283939599991, "grad_norm": 0.10953358560800552, "global_step": 27402, "epoch": 258, "lr": 4.732499644911209e-05} +{"train_loss": 0.014522606506943703, "grad_norm": 0.11431203782558441, "global_step": 27403, "epoch": 258, "lr": 4.7322036931730154e-05} +{"train_loss": 0.017766237258911133, "grad_norm": 0.1169983521103859, "global_step": 27404, "epoch": 258, "lr": 4.731907742375741e-05} +{"train_loss": 0.01685384288430214, "grad_norm": 0.06522554904222488, "global_step": 27405, "epoch": 258, "lr": 4.731611792520426e-05} +{"train_loss": 0.01556468941271305, "grad_norm": 0.0828142911195755, "global_step": 27406, "epoch": 258, "lr": 4.7313158436081094e-05} +{"train_loss": 0.014167760498821735, "grad_norm": 0.08072060346603394, "global_step": 27407, "epoch": 258, "lr": 4.731019895639834e-05} +{"train_loss": 0.017265798524022102, "grad_norm": 0.07168927788734436, "global_step": 27408, "epoch": 258, "lr": 4.7307239486166364e-05} +{"train_loss": 0.01293235458433628, "grad_norm": 0.07346990704536438, "global_step": 27409, "epoch": 258, "lr": 4.73042800253956e-05} +{"train_loss": 0.015032034367322922, "grad_norm": 0.12736821174621582, "global_step": 27410, "epoch": 258, "lr": 4.73013205740964e-05} +{"train_loss": 0.015460248105227947, "grad_norm": 0.050572559237480164, "global_step": 27411, "epoch": 258, "lr": 4.7298361132279185e-05} +{"train_loss": 0.01564529910683632, "grad_norm": 0.09877723455429077, "global_step": 27412, "epoch": 258, "lr": 4.729540169995436e-05} +{"train_loss": 0.018275616690516472, "grad_norm": 0.10055150091648102, "global_step": 27413, "epoch": 258, "lr": 4.729244227713232e-05} +{"train_loss": 0.01430630125105381, "grad_norm": 0.10106974095106125, "global_step": 27414, "epoch": 258, "lr": 4.728948286382347e-05} +{"train_loss": 0.015506263822317123, "grad_norm": 0.06970565021038055, "global_step": 27415, "epoch": 258, "lr": 4.728652346003818e-05} +{"train_loss": 0.015812650322914124, "grad_norm": 0.07527848333120346, "global_step": 27416, "epoch": 258, "lr": 4.728356406578687e-05} +{"train_loss": 0.014503010548651218, "grad_norm": 0.05699701979756355, "global_step": 27417, "epoch": 258, "lr": 4.728060468107994e-05} +{"train_loss": 0.018367905169725418, "grad_norm": 0.09134242683649063, "global_step": 27418, "epoch": 258, "lr": 4.727764530592777e-05} +{"train_loss": 0.015685327351093292, "grad_norm": 0.06737541407346725, "global_step": 27419, "epoch": 258, "lr": 4.7274685940340784e-05} +{"train_loss": 0.015605050139129162, "grad_norm": 0.08924392610788345, "global_step": 27420, "epoch": 258, "lr": 4.727172658432937e-05} +{"train_loss": 0.01597747579216957, "grad_norm": 0.08172836154699326, "global_step": 27421, "epoch": 258, "lr": 4.726876723790392e-05} +{"train_loss": 0.01503307931125164, "grad_norm": 0.07413801550865173, "global_step": 27422, "epoch": 258, "lr": 4.726580790107483e-05} +{"train_loss": 0.016541972756385803, "grad_norm": 0.0651329755783081, "global_step": 27423, "epoch": 258, "lr": 4.726284857385249e-05} +{"train_loss": 0.014454416930675507, "grad_norm": 0.08469852060079575, "global_step": 27424, "epoch": 258, "lr": 4.725988925624732e-05} +{"train_loss": 0.01568015292286873, "grad_norm": 0.09246112406253815, "global_step": 27425, "epoch": 258, "lr": 4.725692994826972e-05} +{"train_loss": 0.01861771196126938, "grad_norm": 0.06188911199569702, "global_step": 27426, "epoch": 258, "lr": 4.725397064993007e-05} +{"train_loss": 0.01571650803089142, "grad_norm": 0.07766852527856827, "global_step": 27427, "epoch": 258, "lr": 4.7251011361238765e-05} +{"train_loss": 0.01685110665857792, "grad_norm": 0.09912139177322388, "global_step": 27428, "epoch": 258, "lr": 4.724805208220621e-05} +{"train_loss": 0.015851512551307678, "grad_norm": 0.08564972132444382, "global_step": 27429, "epoch": 258, "lr": 4.7245092812842805e-05} +{"train_loss": 0.014317642897367477, "grad_norm": 0.08430201560258865, "global_step": 27430, "epoch": 258, "lr": 4.7242133553158935e-05} +{"train_loss": 0.017817001789808273, "grad_norm": 0.10323944687843323, "global_step": 27431, "epoch": 258, "lr": 4.723917430316502e-05} +{"train_loss": 0.017635153606534004, "grad_norm": 0.13201192021369934, "global_step": 27432, "epoch": 258, "lr": 4.7236215062871455e-05} +{"train_loss": 0.013692370615899563, "grad_norm": 0.1590036153793335, "global_step": 27433, "epoch": 258, "lr": 4.723325583228863e-05} +{"train_loss": 0.016202447935938835, "grad_norm": 0.07474292069673538, "global_step": 27434, "epoch": 258, "lr": 4.723029661142693e-05} +{"train_loss": 0.013330947607755661, "grad_norm": 0.05772808939218521, "global_step": 27435, "epoch": 258, "lr": 4.722733740029675e-05} +{"train_loss": 0.014633025974035263, "grad_norm": 0.11823505908250809, "global_step": 27436, "epoch": 258, "lr": 4.722437819890851e-05} +{"train_loss": 0.015691373497247696, "grad_norm": 0.13750356435775757, "global_step": 27437, "epoch": 258, "lr": 4.7221419007272605e-05} +{"train_loss": 0.01667167991399765, "grad_norm": 0.1286531686782837, "global_step": 27438, "epoch": 258, "lr": 4.721845982539944e-05} +{"train_loss": 0.015037862583994865, "grad_norm": 0.0705302283167839, "global_step": 27439, "epoch": 258, "lr": 4.721550065329938e-05} +{"train_loss": 0.014093786478042603, "grad_norm": 0.05909619852900505, "global_step": 27440, "epoch": 258, "lr": 4.7212541490982825e-05} +{"train_loss": 0.014321343041956425, "grad_norm": 0.09229255467653275, "global_step": 27441, "epoch": 258, "lr": 4.7209582338460204e-05} +{"train_loss": 0.01641547493636608, "grad_norm": 0.14787138998508453, "global_step": 27442, "epoch": 258, "lr": 4.720662319574189e-05} +{"train_loss": 0.014435198158025742, "grad_norm": 0.08696933090686798, "global_step": 27443, "epoch": 258, "lr": 4.720366406283829e-05} +{"train_loss": 0.013577573001384735, "grad_norm": 0.11775019764900208, "global_step": 27444, "epoch": 258, "lr": 4.720070493975981e-05} +{"train_loss": 0.017306413501501083, "grad_norm": 0.0696638971567154, "global_step": 27445, "epoch": 258, "lr": 4.719774582651683e-05} +{"train_loss": 0.014290953055024147, "grad_norm": 0.07184088975191116, "global_step": 27446, "epoch": 258, "lr": 4.7194786723119744e-05} +{"train_loss": 0.016586652025580406, "grad_norm": 0.08090310543775558, "global_step": 27447, "epoch": 258, "lr": 4.719182762957895e-05} +{"train_loss": 0.01768585294485092, "grad_norm": 0.11301136016845703, "global_step": 27448, "epoch": 258, "lr": 4.7188868545904864e-05} +{"train_loss": 0.014136210083961487, "grad_norm": 0.10215646773576736, "global_step": 27449, "epoch": 258, "lr": 4.718590947210788e-05} +{"train_loss": 0.01636962965130806, "grad_norm": 0.05475997179746628, "global_step": 27450, "epoch": 258, "lr": 4.7182950408198386e-05} +{"train_loss": 0.012349987402558327, "grad_norm": 0.06911809742450714, "global_step": 27451, "epoch": 258, "lr": 4.717999135418677e-05} +{"train_loss": 0.01477684173732996, "grad_norm": 0.06319551914930344, "global_step": 27452, "epoch": 258, "lr": 4.717703231008343e-05} +{"train_loss": 0.015526483511938801, "grad_norm": 0.1731353998184204, "global_step": 27453, "epoch": 258, "lr": 4.7174073275898776e-05, "val_loss": 0.019612502306699753} +{"train_loss": 0.018062585964798927, "grad_norm": 0.092633455991745, "global_step": 27454, "epoch": 259, "lr": 4.717111425164321e-05} +{"train_loss": 0.0152227021753788, "grad_norm": 0.07809816300868988, "global_step": 27455, "epoch": 259, "lr": 4.7168155237327096e-05} +{"train_loss": 0.013899678364396095, "grad_norm": 0.14001816511154175, "global_step": 27456, "epoch": 259, "lr": 4.716519623296088e-05} +{"train_loss": 0.014917293563485146, "grad_norm": 0.10760480165481567, "global_step": 27457, "epoch": 259, "lr": 4.7162237238554925e-05} +{"train_loss": 0.014256228692829609, "grad_norm": 0.09610935300588608, "global_step": 27458, "epoch": 259, "lr": 4.715927825411963e-05} +{"train_loss": 0.015625569969415665, "grad_norm": 0.061508454382419586, "global_step": 27459, "epoch": 259, "lr": 4.715631927966538e-05} +{"train_loss": 0.014851011335849762, "grad_norm": 0.13318204879760742, "global_step": 27460, "epoch": 259, "lr": 4.71533603152026e-05} +{"train_loss": 0.013945600017905235, "grad_norm": 0.13514438271522522, "global_step": 27461, "epoch": 259, "lr": 4.7150401360741694e-05} +{"train_loss": 0.01558276079595089, "grad_norm": 0.07667370140552521, "global_step": 27462, "epoch": 259, "lr": 4.714744241629302e-05} +{"train_loss": 0.015757087618112564, "grad_norm": 0.08404546976089478, "global_step": 27463, "epoch": 259, "lr": 4.7144483481867e-05} +{"train_loss": 0.01474953442811966, "grad_norm": 0.07481441646814346, "global_step": 27464, "epoch": 259, "lr": 4.7141524557474004e-05} +{"train_loss": 0.015399554744362831, "grad_norm": 0.08853898197412491, "global_step": 27465, "epoch": 259, "lr": 4.7138565643124463e-05} +{"train_loss": 0.015701398253440857, "grad_norm": 0.05320865660905838, "global_step": 27466, "epoch": 259, "lr": 4.713560673882875e-05} +{"train_loss": 0.01427605003118515, "grad_norm": 0.06330661475658417, "global_step": 27467, "epoch": 259, "lr": 4.7132647844597286e-05} +{"train_loss": 0.016864579170942307, "grad_norm": 0.0768568366765976, "global_step": 27468, "epoch": 259, "lr": 4.7129688960440445e-05} +{"train_loss": 0.01418154127895832, "grad_norm": 0.10248079150915146, "global_step": 27469, "epoch": 259, "lr": 4.712673008636861e-05} +{"train_loss": 0.016326334327459335, "grad_norm": 0.0931960865855217, "global_step": 27470, "epoch": 259, "lr": 4.712377122239221e-05} +{"train_loss": 0.015604234300553799, "grad_norm": 0.08379421383142471, "global_step": 27471, "epoch": 259, "lr": 4.712081236852163e-05} +{"train_loss": 0.01900136098265648, "grad_norm": 0.10992845892906189, "global_step": 27472, "epoch": 259, "lr": 4.7117853524767244e-05} +{"train_loss": 0.014753292314708233, "grad_norm": 0.07221990823745728, "global_step": 27473, "epoch": 259, "lr": 4.71148946911395e-05} +{"train_loss": 0.016715455800294876, "grad_norm": 0.11092375218868256, "global_step": 27474, "epoch": 259, "lr": 4.711193586764874e-05} +{"train_loss": 0.015596005134284496, "grad_norm": 0.06514491140842438, "global_step": 27475, "epoch": 259, "lr": 4.7108977054305386e-05} +{"train_loss": 0.013273217715322971, "grad_norm": 0.06540881842374802, "global_step": 27476, "epoch": 259, "lr": 4.710601825111981e-05} +{"train_loss": 0.015923505648970604, "grad_norm": 0.10310038179159164, "global_step": 27477, "epoch": 259, "lr": 4.710305945810245e-05} +{"train_loss": 0.01399417407810688, "grad_norm": 0.05859098955988884, "global_step": 27478, "epoch": 259, "lr": 4.710010067526368e-05} +{"train_loss": 0.018907800316810608, "grad_norm": 0.1638174057006836, "global_step": 27479, "epoch": 259, "lr": 4.70971419026139e-05} +{"train_loss": 0.015902740880846977, "grad_norm": 0.07325268536806107, "global_step": 27480, "epoch": 259, "lr": 4.709418314016349e-05} +{"train_loss": 0.01782221533358097, "grad_norm": 0.08498970419168472, "global_step": 27481, "epoch": 259, "lr": 4.709122438792285e-05} +{"train_loss": 0.015771616250276566, "grad_norm": 0.12407591193914413, "global_step": 27482, "epoch": 259, "lr": 4.708826564590239e-05} +{"train_loss": 0.013308349996805191, "grad_norm": 0.07762154936790466, "global_step": 27483, "epoch": 259, "lr": 4.70853069141125e-05} +{"train_loss": 0.013369081541895866, "grad_norm": 0.059721242636442184, "global_step": 27484, "epoch": 259, "lr": 4.708234819256356e-05} +{"train_loss": 0.014948518946766853, "grad_norm": 0.05743288993835449, "global_step": 27485, "epoch": 259, "lr": 4.7079389481266006e-05} +{"train_loss": 0.015975642949342728, "grad_norm": 0.09289638698101044, "global_step": 27486, "epoch": 259, "lr": 4.7076430780230194e-05} +{"train_loss": 0.014264404773712158, "grad_norm": 0.09196540713310242, "global_step": 27487, "epoch": 259, "lr": 4.707347208946653e-05} +{"train_loss": 0.01664123311638832, "grad_norm": 0.07415588945150375, "global_step": 27488, "epoch": 259, "lr": 4.7070513408985405e-05} +{"train_loss": 0.01706942915916443, "grad_norm": 0.07989279925823212, "global_step": 27489, "epoch": 259, "lr": 4.706755473879723e-05} +{"train_loss": 0.015176631510257721, "grad_norm": 0.08125822246074677, "global_step": 27490, "epoch": 259, "lr": 4.70645960789124e-05} +{"train_loss": 0.014119277708232403, "grad_norm": 0.08396466076374054, "global_step": 27491, "epoch": 259, "lr": 4.706163742934131e-05} +{"train_loss": 0.014618455432355404, "grad_norm": 0.07006116211414337, "global_step": 27492, "epoch": 259, "lr": 4.705867879009433e-05} +{"train_loss": 0.014314503408968449, "grad_norm": 0.07314234226942062, "global_step": 27493, "epoch": 259, "lr": 4.705572016118187e-05} +{"train_loss": 0.013866965658962727, "grad_norm": 0.08861749619245529, "global_step": 27494, "epoch": 259, "lr": 4.705276154261434e-05} +{"train_loss": 0.01693299040198326, "grad_norm": 0.06589876115322113, "global_step": 27495, "epoch": 259, "lr": 4.704980293440212e-05} +{"train_loss": 0.013389375060796738, "grad_norm": 0.14325492084026337, "global_step": 27496, "epoch": 259, "lr": 4.7046844336555604e-05} +{"train_loss": 0.016244711354374886, "grad_norm": 0.14116141200065613, "global_step": 27497, "epoch": 259, "lr": 4.704388574908521e-05} +{"train_loss": 0.01786450482904911, "grad_norm": 0.06793157011270523, "global_step": 27498, "epoch": 259, "lr": 4.70409271720013e-05} +{"train_loss": 0.01532833743840456, "grad_norm": 0.05427934229373932, "global_step": 27499, "epoch": 259, "lr": 4.703796860531429e-05} +{"train_loss": 0.015091096051037312, "grad_norm": 0.06568821519613266, "global_step": 27500, "epoch": 259, "lr": 4.7035010049034575e-05} +{"train_loss": 0.012393125332891941, "grad_norm": 0.05713662505149841, "global_step": 27501, "epoch": 259, "lr": 4.703205150317253e-05} +{"train_loss": 0.017604444175958633, "grad_norm": 0.13723759353160858, "global_step": 27502, "epoch": 259, "lr": 4.7029092967738576e-05} +{"train_loss": 0.01685193181037903, "grad_norm": 0.10353314876556396, "global_step": 27503, "epoch": 259, "lr": 4.7026134442743105e-05} +{"train_loss": 0.01608385518193245, "grad_norm": 0.058257583528757095, "global_step": 27504, "epoch": 259, "lr": 4.7023175928196504e-05} +{"train_loss": 0.015821512788534164, "grad_norm": 0.059224314987659454, "global_step": 27505, "epoch": 259, "lr": 4.702021742410915e-05} +{"train_loss": 0.01841723918914795, "grad_norm": 0.07199589163064957, "global_step": 27506, "epoch": 259, "lr": 4.701725893049147e-05} +{"train_loss": 0.016754519194364548, "grad_norm": 0.09379763156175613, "global_step": 27507, "epoch": 259, "lr": 4.7014300447353846e-05} +{"train_loss": 0.01599552482366562, "grad_norm": 0.05931377410888672, "global_step": 27508, "epoch": 259, "lr": 4.701134197470665e-05} +{"train_loss": 0.016911394894123077, "grad_norm": 0.11314176023006439, "global_step": 27509, "epoch": 259, "lr": 4.700838351256034e-05} +{"train_loss": 0.013189954683184624, "grad_norm": 0.06373920291662216, "global_step": 27510, "epoch": 259, "lr": 4.7005425060925234e-05} +{"train_loss": 0.015546373091638088, "grad_norm": 0.1017918586730957, "global_step": 27511, "epoch": 259, "lr": 4.7002466619811775e-05} +{"train_loss": 0.014613382518291473, "grad_norm": 0.06533566862344742, "global_step": 27512, "epoch": 259, "lr": 4.699950818923034e-05} +{"train_loss": 0.014388414099812508, "grad_norm": 0.062238480895757675, "global_step": 27513, "epoch": 259, "lr": 4.699654976919133e-05} +{"train_loss": 0.015323488041758537, "grad_norm": 0.06698929518461227, "global_step": 27514, "epoch": 259, "lr": 4.6993591359705144e-05} +{"train_loss": 0.01617426425218582, "grad_norm": 0.0852786973118782, "global_step": 27515, "epoch": 259, "lr": 4.699063296078217e-05} +{"train_loss": 0.01611970365047455, "grad_norm": 0.05962172895669937, "global_step": 27516, "epoch": 259, "lr": 4.69876745724328e-05} +{"train_loss": 0.013860350474715233, "grad_norm": 0.10607434809207916, "global_step": 27517, "epoch": 259, "lr": 4.6984716194667424e-05} +{"train_loss": 0.01550840213894844, "grad_norm": 0.0872739627957344, "global_step": 27518, "epoch": 259, "lr": 4.6981757827496444e-05} +{"train_loss": 0.016656944528222084, "grad_norm": 0.06775306165218353, "global_step": 27519, "epoch": 259, "lr": 4.697879947093027e-05} +{"train_loss": 0.017308969050645828, "grad_norm": 0.0814647302031517, "global_step": 27520, "epoch": 259, "lr": 4.697584112497928e-05} +{"train_loss": 0.016597982496023178, "grad_norm": 0.09909597784280777, "global_step": 27521, "epoch": 259, "lr": 4.697288278965386e-05} +{"train_loss": 0.01675158366560936, "grad_norm": 0.05917965620756149, "global_step": 27522, "epoch": 259, "lr": 4.696992446496441e-05} +{"train_loss": 0.01590621843934059, "grad_norm": 0.08827895671129227, "global_step": 27523, "epoch": 259, "lr": 4.6966966150921335e-05} +{"train_loss": 0.016198942437767982, "grad_norm": 0.07846946269273758, "global_step": 27524, "epoch": 259, "lr": 4.696400784753501e-05} +{"train_loss": 0.015827879309654236, "grad_norm": 0.08903108537197113, "global_step": 27525, "epoch": 259, "lr": 4.6961049554815844e-05} +{"train_loss": 0.01535504125058651, "grad_norm": 0.06537842005491257, "global_step": 27526, "epoch": 259, "lr": 4.695809127277426e-05} +{"train_loss": 0.017070060595870018, "grad_norm": 0.06861746311187744, "global_step": 27527, "epoch": 259, "lr": 4.695513300142058e-05} +{"train_loss": 0.015231159515678883, "grad_norm": 0.11471215635538101, "global_step": 27528, "epoch": 259, "lr": 4.6952174740765254e-05} +{"train_loss": 0.01493272464722395, "grad_norm": 0.055074308067560196, "global_step": 27529, "epoch": 259, "lr": 4.694921649081867e-05} +{"train_loss": 0.013961022719740868, "grad_norm": 0.08439292013645172, "global_step": 27530, "epoch": 259, "lr": 4.694625825159119e-05} +{"train_loss": 0.014785814099013805, "grad_norm": 0.05010173097252846, "global_step": 27531, "epoch": 259, "lr": 4.694330002309324e-05} +{"train_loss": 0.017020925879478455, "grad_norm": 0.0770227387547493, "global_step": 27532, "epoch": 259, "lr": 4.694034180533522e-05} +{"train_loss": 0.017443981021642685, "grad_norm": 0.06378915905952454, "global_step": 27533, "epoch": 259, "lr": 4.6937383598327495e-05} +{"train_loss": 0.016550110653042793, "grad_norm": 0.08175934851169586, "global_step": 27534, "epoch": 259, "lr": 4.693442540208046e-05} +{"train_loss": 0.016776010394096375, "grad_norm": 0.05959600582718849, "global_step": 27535, "epoch": 259, "lr": 4.693146721660454e-05} +{"train_loss": 0.016260065138339996, "grad_norm": 0.07578536868095398, "global_step": 27536, "epoch": 259, "lr": 4.6928509041910093e-05} +{"train_loss": 0.01423265878111124, "grad_norm": 0.09120424836874008, "global_step": 27537, "epoch": 259, "lr": 4.692555087800753e-05} +{"train_loss": 0.013311877846717834, "grad_norm": 0.06479141116142273, "global_step": 27538, "epoch": 259, "lr": 4.6922592724907285e-05} +{"train_loss": 0.017138393595814705, "grad_norm": 0.08888863772153854, "global_step": 27539, "epoch": 259, "lr": 4.691963458261966e-05} +{"train_loss": 0.016406290233135223, "grad_norm": 0.10903007537126541, "global_step": 27540, "epoch": 259, "lr": 4.691667645115513e-05} +{"train_loss": 0.01459228340536356, "grad_norm": 0.05234399065375328, "global_step": 27541, "epoch": 259, "lr": 4.691371833052404e-05} +{"train_loss": 0.014803177677094936, "grad_norm": 0.0545293428003788, "global_step": 27542, "epoch": 259, "lr": 4.69107602207368e-05} +{"train_loss": 0.014427288435399532, "grad_norm": 0.052542444318532944, "global_step": 27543, "epoch": 259, "lr": 4.6907802121803816e-05} +{"train_loss": 0.016310375183820724, "grad_norm": 0.10060185194015503, "global_step": 27544, "epoch": 259, "lr": 4.690484403373548e-05} +{"train_loss": 0.013694066554307938, "grad_norm": 0.07133334130048752, "global_step": 27545, "epoch": 259, "lr": 4.690188595654217e-05} +{"train_loss": 0.013498062267899513, "grad_norm": 0.1172456443309784, "global_step": 27546, "epoch": 259, "lr": 4.689892789023427e-05} +{"train_loss": 0.012061430141329765, "grad_norm": 0.10955411940813065, "global_step": 27547, "epoch": 259, "lr": 4.6895969834822196e-05} +{"train_loss": 0.0167187862098217, "grad_norm": 0.06856496632099152, "global_step": 27548, "epoch": 259, "lr": 4.6893011790316336e-05} +{"train_loss": 0.013603465631604195, "grad_norm": 0.07591453194618225, "global_step": 27549, "epoch": 259, "lr": 4.689005375672707e-05} +{"train_loss": 0.01176038570702076, "grad_norm": 0.08709581941366196, "global_step": 27550, "epoch": 259, "lr": 4.688709573406484e-05} +{"train_loss": 0.01734018884599209, "grad_norm": 0.11221333593130112, "global_step": 27551, "epoch": 259, "lr": 4.688413772233997e-05} +{"train_loss": 0.014414266683161259, "grad_norm": 0.059050239622592926, "global_step": 27552, "epoch": 259, "lr": 4.688117972156289e-05} +{"train_loss": 0.016420988366007805, "grad_norm": 0.07331971824169159, "global_step": 27553, "epoch": 259, "lr": 4.687822173174399e-05} +{"train_loss": 0.018317919224500656, "grad_norm": 0.09180363267660141, "global_step": 27554, "epoch": 259, "lr": 4.687526375289365e-05} +{"train_loss": 0.01581392250955105, "grad_norm": 0.14213547110557556, "global_step": 27555, "epoch": 259, "lr": 4.6872305785022284e-05} +{"train_loss": 0.012403911910951138, "grad_norm": 0.07293781638145447, "global_step": 27556, "epoch": 259, "lr": 4.6869347828140295e-05} +{"train_loss": 0.0170162133872509, "grad_norm": 0.06941153109073639, "global_step": 27557, "epoch": 259, "lr": 4.686638988225803e-05} +{"train_loss": 0.015994615852832794, "grad_norm": 0.08048999309539795, "global_step": 27558, "epoch": 259, "lr": 4.686343194738591e-05} +{"train_loss": 0.015497620287790613, "grad_norm": 0.10718203336000443, "global_step": 27559, "epoch": 259, "lr": 4.6860474023534335e-05, "val_loss": 0.018834920600056648} +{"train_loss": 0.013471543788909912, "grad_norm": 0.09752929955720901, "global_step": 27560, "epoch": 260, "lr": 4.6857516110713685e-05} +{"train_loss": 0.015345156192779541, "grad_norm": 0.08101554214954376, "global_step": 27561, "epoch": 260, "lr": 4.6854558208934346e-05} +{"train_loss": 0.01753240078687668, "grad_norm": 0.10220110416412354, "global_step": 27562, "epoch": 260, "lr": 4.6851600318206755e-05} +{"train_loss": 0.013752501457929611, "grad_norm": 0.060229796916246414, "global_step": 27563, "epoch": 260, "lr": 4.684864243854124e-05} +{"train_loss": 0.01632777787744999, "grad_norm": 0.11123151332139969, "global_step": 27564, "epoch": 260, "lr": 4.684568456994824e-05} +{"train_loss": 0.016440309584140778, "grad_norm": 0.10685975849628448, "global_step": 27565, "epoch": 260, "lr": 4.684272671243812e-05} +{"train_loss": 0.01454581692814827, "grad_norm": 0.05524398013949394, "global_step": 27566, "epoch": 260, "lr": 4.683976886602128e-05} +{"train_loss": 0.018245786428451538, "grad_norm": 0.12840110063552856, "global_step": 27567, "epoch": 260, "lr": 4.683681103070816e-05} +{"train_loss": 0.014360702596604824, "grad_norm": 0.11569087207317352, "global_step": 27568, "epoch": 260, "lr": 4.683385320650907e-05} +{"train_loss": 0.016685528680682182, "grad_norm": 0.11494965851306915, "global_step": 27569, "epoch": 260, "lr": 4.6830895393434456e-05} +{"train_loss": 0.014460884965956211, "grad_norm": 0.05943940952420235, "global_step": 27570, "epoch": 260, "lr": 4.682793759149469e-05} +{"train_loss": 0.017999079078435898, "grad_norm": 0.09980308264493942, "global_step": 27571, "epoch": 260, "lr": 4.6824979800700166e-05} +{"train_loss": 0.01577993668615818, "grad_norm": 0.11275983601808548, "global_step": 27572, "epoch": 260, "lr": 4.68220220210613e-05} +{"train_loss": 0.014960633590817451, "grad_norm": 0.060550034046173096, "global_step": 27573, "epoch": 260, "lr": 4.6819064252588466e-05} +{"train_loss": 0.0171387679874897, "grad_norm": 0.1671769618988037, "global_step": 27574, "epoch": 260, "lr": 4.681610649529206e-05} +{"train_loss": 0.013734055683016777, "grad_norm": 0.14366036653518677, "global_step": 27575, "epoch": 260, "lr": 4.6813148749182446e-05} +{"train_loss": 0.015235871076583862, "grad_norm": 0.18392620980739594, "global_step": 27576, "epoch": 260, "lr": 4.6810191014270055e-05} +{"train_loss": 0.01603248529136181, "grad_norm": 0.06386715173721313, "global_step": 27577, "epoch": 260, "lr": 4.6807233290565266e-05} +{"train_loss": 0.014414737932384014, "grad_norm": 0.07434795796871185, "global_step": 27578, "epoch": 260, "lr": 4.6804275578078465e-05} +{"train_loss": 0.0160280279815197, "grad_norm": 0.06908171623945236, "global_step": 27579, "epoch": 260, "lr": 4.6801317876820076e-05} +{"train_loss": 0.0161549411714077, "grad_norm": 0.08934220671653748, "global_step": 27580, "epoch": 260, "lr": 4.679836018680044e-05} +{"train_loss": 0.013962408527731895, "grad_norm": 0.052704911679029465, "global_step": 27581, "epoch": 260, "lr": 4.679540250802998e-05} +{"train_loss": 0.01619240641593933, "grad_norm": 0.09096942842006683, "global_step": 27582, "epoch": 260, "lr": 4.6792444840519084e-05} +{"train_loss": 0.014172116294503212, "grad_norm": 0.10162340104579926, "global_step": 27583, "epoch": 260, "lr": 4.678948718427813e-05} +{"train_loss": 0.016651447862386703, "grad_norm": 0.07105565816164017, "global_step": 27584, "epoch": 260, "lr": 4.678652953931753e-05} +{"train_loss": 0.013853127136826515, "grad_norm": 0.0843600183725357, "global_step": 27585, "epoch": 260, "lr": 4.6783571905647686e-05} +{"train_loss": 0.015003075823187828, "grad_norm": 0.1272994577884674, "global_step": 27586, "epoch": 260, "lr": 4.678061428327895e-05} +{"train_loss": 0.0153767429292202, "grad_norm": 0.06592391431331635, "global_step": 27587, "epoch": 260, "lr": 4.677765667222173e-05} +{"train_loss": 0.016483481973409653, "grad_norm": 0.10409150272607803, "global_step": 27588, "epoch": 260, "lr": 4.677469907248643e-05} +{"train_loss": 0.016006316989660263, "grad_norm": 0.08215712010860443, "global_step": 27589, "epoch": 260, "lr": 4.677174148408344e-05} +{"train_loss": 0.014529142528772354, "grad_norm": 0.0829014778137207, "global_step": 27590, "epoch": 260, "lr": 4.676878390702314e-05} +{"train_loss": 0.015859508886933327, "grad_norm": 0.06307441741228104, "global_step": 27591, "epoch": 260, "lr": 4.6765826341315954e-05} +{"train_loss": 0.015665486454963684, "grad_norm": 0.0773182362318039, "global_step": 27592, "epoch": 260, "lr": 4.676286878697221e-05} +{"train_loss": 0.013743292540311813, "grad_norm": 0.070346899330616, "global_step": 27593, "epoch": 260, "lr": 4.675991124400236e-05} +{"train_loss": 0.014988045208156109, "grad_norm": 0.09574327617883682, "global_step": 27594, "epoch": 260, "lr": 4.675695371241677e-05} +{"train_loss": 0.018252700567245483, "grad_norm": 0.08877518028020859, "global_step": 27595, "epoch": 260, "lr": 4.6753996192225815e-05} +{"train_loss": 0.013431498780846596, "grad_norm": 0.049570806324481964, "global_step": 27596, "epoch": 260, "lr": 4.675103868343992e-05} +{"train_loss": 0.017638763412833214, "grad_norm": 0.08314394950866699, "global_step": 27597, "epoch": 260, "lr": 4.6748081186069485e-05} +{"train_loss": 0.0178779736161232, "grad_norm": 0.13378280401229858, "global_step": 27598, "epoch": 260, "lr": 4.674512370012486e-05} +{"train_loss": 0.015665877610445023, "grad_norm": 0.060966312885284424, "global_step": 27599, "epoch": 260, "lr": 4.674216622561644e-05} +{"train_loss": 0.013696005567908287, "grad_norm": 0.10197695344686508, "global_step": 27600, "epoch": 260, "lr": 4.6739208762554635e-05} +{"train_loss": 0.016699504107236862, "grad_norm": 0.06416244804859161, "global_step": 27601, "epoch": 260, "lr": 4.673625131094984e-05} +{"train_loss": 0.01601177453994751, "grad_norm": 0.11801286041736603, "global_step": 27602, "epoch": 260, "lr": 4.6733293870812445e-05} +{"train_loss": 0.016903411597013474, "grad_norm": 0.06609342247247696, "global_step": 27603, "epoch": 260, "lr": 4.673033644215284e-05} +{"train_loss": 0.014532536268234253, "grad_norm": 0.0906885489821434, "global_step": 27604, "epoch": 260, "lr": 4.6727379024981385e-05} +{"train_loss": 0.013934526592493057, "grad_norm": 0.0596466027200222, "global_step": 27605, "epoch": 260, "lr": 4.672442161930851e-05} +{"train_loss": 0.015543556772172451, "grad_norm": 0.12396065890789032, "global_step": 27606, "epoch": 260, "lr": 4.67214642251446e-05} +{"train_loss": 0.013283906504511833, "grad_norm": 0.08988875150680542, "global_step": 27607, "epoch": 260, "lr": 4.671850684250002e-05} +{"train_loss": 0.014045562595129013, "grad_norm": 0.06454051285982132, "global_step": 27608, "epoch": 260, "lr": 4.6715549471385196e-05} +{"train_loss": 0.013241746462881565, "grad_norm": 0.08924004435539246, "global_step": 27609, "epoch": 260, "lr": 4.671259211181051e-05} +{"train_loss": 0.014380577951669693, "grad_norm": 0.08327623456716537, "global_step": 27610, "epoch": 260, "lr": 4.6709634763786336e-05} +{"train_loss": 0.01823541149497032, "grad_norm": 0.09280531853437424, "global_step": 27611, "epoch": 260, "lr": 4.6706677427323075e-05} +{"train_loss": 0.016234278678894043, "grad_norm": 0.07219980657100677, "global_step": 27612, "epoch": 260, "lr": 4.670372010243111e-05} +{"train_loss": 0.01526571437716484, "grad_norm": 0.08409937471151352, "global_step": 27613, "epoch": 260, "lr": 4.6700762789120845e-05} +{"train_loss": 0.014600708149373531, "grad_norm": 0.07966621220111847, "global_step": 27614, "epoch": 260, "lr": 4.669780548740267e-05} +{"train_loss": 0.017784317955374718, "grad_norm": 0.08041975647211075, "global_step": 27615, "epoch": 260, "lr": 4.6694848197286974e-05} +{"train_loss": 0.01428885105997324, "grad_norm": 0.06885921955108643, "global_step": 27616, "epoch": 260, "lr": 4.669189091878412e-05} +{"train_loss": 0.01633507013320923, "grad_norm": 0.09037936478853226, "global_step": 27617, "epoch": 260, "lr": 4.6688933651904545e-05} +{"train_loss": 0.014470399357378483, "grad_norm": 0.06848260760307312, "global_step": 27618, "epoch": 260, "lr": 4.6685976396658605e-05} +{"train_loss": 0.016977857798337936, "grad_norm": 0.09417830407619476, "global_step": 27619, "epoch": 260, "lr": 4.66830191530567e-05} +{"train_loss": 0.01521826721727848, "grad_norm": 0.05745280161499977, "global_step": 27620, "epoch": 260, "lr": 4.668006192110925e-05} +{"train_loss": 0.01730506122112274, "grad_norm": 0.07683119922876358, "global_step": 27621, "epoch": 260, "lr": 4.667710470082659e-05} +{"train_loss": 0.017091523855924606, "grad_norm": 0.08384272456169128, "global_step": 27622, "epoch": 260, "lr": 4.667414749221915e-05} +{"train_loss": 0.01689356379210949, "grad_norm": 0.06999511271715164, "global_step": 27623, "epoch": 260, "lr": 4.667119029529731e-05} +{"train_loss": 0.014456093311309814, "grad_norm": 0.07191202044487, "global_step": 27624, "epoch": 260, "lr": 4.666823311007145e-05} +{"train_loss": 0.014533501118421555, "grad_norm": 0.08954572677612305, "global_step": 27625, "epoch": 260, "lr": 4.666527593655198e-05} +{"train_loss": 0.013249007984995842, "grad_norm": 0.08518026024103165, "global_step": 27626, "epoch": 260, "lr": 4.666231877474929e-05} +{"train_loss": 0.01728379726409912, "grad_norm": 0.05307656154036522, "global_step": 27627, "epoch": 260, "lr": 4.665936162467375e-05} +{"train_loss": 0.017533503472805023, "grad_norm": 0.18061573803424835, "global_step": 27628, "epoch": 260, "lr": 4.665640448633575e-05} +{"train_loss": 0.01598268374800682, "grad_norm": 0.06696441769599915, "global_step": 27629, "epoch": 260, "lr": 4.665344735974569e-05} +{"train_loss": 0.014412302523851395, "grad_norm": 0.17541663348674774, "global_step": 27630, "epoch": 260, "lr": 4.6650490244913975e-05} +{"train_loss": 0.01537284068763256, "grad_norm": 0.0718657523393631, "global_step": 27631, "epoch": 260, "lr": 4.6647533141850975e-05} +{"train_loss": 0.017036205157637596, "grad_norm": 0.08788207918405533, "global_step": 27632, "epoch": 260, "lr": 4.664457605056709e-05} +{"train_loss": 0.013900421559810638, "grad_norm": 0.061548586934804916, "global_step": 27633, "epoch": 260, "lr": 4.6641618971072696e-05} +{"train_loss": 0.015530094504356384, "grad_norm": 0.06266488134860992, "global_step": 27634, "epoch": 260, "lr": 4.66386619033782e-05} +{"train_loss": 0.014019660651683807, "grad_norm": 0.05711551383137703, "global_step": 27635, "epoch": 260, "lr": 4.663570484749398e-05} +{"train_loss": 0.015476967208087444, "grad_norm": 0.09975330531597137, "global_step": 27636, "epoch": 260, "lr": 4.663274780343043e-05} +{"train_loss": 0.013611620292067528, "grad_norm": 0.08546876162290573, "global_step": 27637, "epoch": 260, "lr": 4.662979077119794e-05} +{"train_loss": 0.015198417007923126, "grad_norm": 0.09488919377326965, "global_step": 27638, "epoch": 260, "lr": 4.662683375080692e-05} +{"train_loss": 0.015294413082301617, "grad_norm": 0.06454545259475708, "global_step": 27639, "epoch": 260, "lr": 4.662387674226771e-05} +{"train_loss": 0.018708806484937668, "grad_norm": 0.08855339139699936, "global_step": 27640, "epoch": 260, "lr": 4.662091974559075e-05} +{"train_loss": 0.015811173245310783, "grad_norm": 0.08080253005027771, "global_step": 27641, "epoch": 260, "lr": 4.661796276078638e-05} +{"train_loss": 0.01619209349155426, "grad_norm": 0.08470537513494492, "global_step": 27642, "epoch": 260, "lr": 4.661500578786504e-05} +{"train_loss": 0.013550783507525921, "grad_norm": 0.06470342725515366, "global_step": 27643, "epoch": 260, "lr": 4.66120488268371e-05} +{"train_loss": 0.014162678271532059, "grad_norm": 0.07912260293960571, "global_step": 27644, "epoch": 260, "lr": 4.6609091877712944e-05} +{"train_loss": 0.015978163108229637, "grad_norm": 0.06172347068786621, "global_step": 27645, "epoch": 260, "lr": 4.660613494050295e-05} +{"train_loss": 0.01869741827249527, "grad_norm": 0.07928600162267685, "global_step": 27646, "epoch": 260, "lr": 4.660317801521754e-05} +{"train_loss": 0.013962964527308941, "grad_norm": 0.05951833724975586, "global_step": 27647, "epoch": 260, "lr": 4.6600221101867075e-05} +{"train_loss": 0.015484895557165146, "grad_norm": 0.05514828860759735, "global_step": 27648, "epoch": 260, "lr": 4.659726420046195e-05} +{"train_loss": 0.01563972234725952, "grad_norm": 0.0687190368771553, "global_step": 27649, "epoch": 260, "lr": 4.659430731101256e-05} +{"train_loss": 0.015548640862107277, "grad_norm": 0.05299476906657219, "global_step": 27650, "epoch": 260, "lr": 4.659135043352932e-05} +{"train_loss": 0.015313507989048958, "grad_norm": 0.0898602232336998, "global_step": 27651, "epoch": 260, "lr": 4.658839356802257e-05} +{"train_loss": 0.015082525089383125, "grad_norm": 0.1606220155954361, "global_step": 27652, "epoch": 260, "lr": 4.658543671450273e-05} +{"train_loss": 0.01632615551352501, "grad_norm": 0.08511221408843994, "global_step": 27653, "epoch": 260, "lr": 4.6582479872980154e-05} +{"train_loss": 0.015115050598978996, "grad_norm": 0.10022645443677902, "global_step": 27654, "epoch": 260, "lr": 4.6579523043465285e-05} +{"train_loss": 0.014358008280396461, "grad_norm": 0.055391792207956314, "global_step": 27655, "epoch": 260, "lr": 4.657656622596848e-05} +{"train_loss": 0.01625685766339302, "grad_norm": 0.06774303317070007, "global_step": 27656, "epoch": 260, "lr": 4.657360942050014e-05} +{"train_loss": 0.01676075905561447, "grad_norm": 0.07809793204069138, "global_step": 27657, "epoch": 260, "lr": 4.6570652627070634e-05} +{"train_loss": 0.013079404830932617, "grad_norm": 0.13942070305347443, "global_step": 27658, "epoch": 260, "lr": 4.656769584569035e-05} +{"train_loss": 0.014121567830443382, "grad_norm": 0.05552024021744728, "global_step": 27659, "epoch": 260, "lr": 4.656473907636971e-05} +{"train_loss": 0.016276486217975616, "grad_norm": 0.06743738055229187, "global_step": 27660, "epoch": 260, "lr": 4.656178231911907e-05} +{"train_loss": 0.017542805522680283, "grad_norm": 0.07338879257440567, "global_step": 27661, "epoch": 260, "lr": 4.655882557394884e-05} +{"train_loss": 0.014183567836880684, "grad_norm": 0.06290583312511444, "global_step": 27662, "epoch": 260, "lr": 4.655586884086941e-05} +{"train_loss": 0.015052462927997112, "grad_norm": 0.06989289820194244, "global_step": 27663, "epoch": 260, "lr": 4.655291211989114e-05} +{"train_loss": 0.014727055095136166, "grad_norm": 0.09517835825681686, "global_step": 27664, "epoch": 260, "lr": 4.6549955411024456e-05} +{"train_loss": 0.015498722780425593, "grad_norm": 0.09398680180311203, "global_step": 27665, "epoch": 260, "lr": 4.654699871427971e-05, "train_batch_mse_error": 0.013933533802628517, "val_loss": 0.01902000419795513} +{"train_loss": 0.01390991359949112, "grad_norm": 0.06219290941953659, "global_step": 27666, "epoch": 261, "lr": 4.654404202966732e-05} +{"train_loss": 0.01443948782980442, "grad_norm": 0.06711309403181076, "global_step": 27667, "epoch": 261, "lr": 4.654108535719767e-05} +{"train_loss": 0.01444102544337511, "grad_norm": 0.07339321076869965, "global_step": 27668, "epoch": 261, "lr": 4.6538128696881135e-05} +{"train_loss": 0.016858620569109917, "grad_norm": 0.10346321761608124, "global_step": 27669, "epoch": 261, "lr": 4.653517204872811e-05} +{"train_loss": 0.016790494322776794, "grad_norm": 0.07500023394823074, "global_step": 27670, "epoch": 261, "lr": 4.6532215412748966e-05} +{"train_loss": 0.01739034429192543, "grad_norm": 0.07856184244155884, "global_step": 27671, "epoch": 261, "lr": 4.652925878895413e-05} +{"train_loss": 0.017432862892746925, "grad_norm": 0.10056919604539871, "global_step": 27672, "epoch": 261, "lr": 4.652630217735396e-05} +{"train_loss": 0.01567593216896057, "grad_norm": 0.09819930046796799, "global_step": 27673, "epoch": 261, "lr": 4.6523345577958876e-05} +{"train_loss": 0.016761712729930878, "grad_norm": 0.13049760460853577, "global_step": 27674, "epoch": 261, "lr": 4.652038899077921e-05} +{"train_loss": 0.016394702717661858, "grad_norm": 0.08903462439775467, "global_step": 27675, "epoch": 261, "lr": 4.6517432415825405e-05} +{"train_loss": 0.015692617744207382, "grad_norm": 0.1450011432170868, "global_step": 27676, "epoch": 261, "lr": 4.651447585310782e-05} +{"train_loss": 0.014967462047934532, "grad_norm": 0.07967881113290787, "global_step": 27677, "epoch": 261, "lr": 4.6511519302636844e-05} +{"train_loss": 0.01602848619222641, "grad_norm": 0.0874265730381012, "global_step": 27678, "epoch": 261, "lr": 4.650856276442288e-05} +{"train_loss": 0.013824083842337132, "grad_norm": 0.14378301799297333, "global_step": 27679, "epoch": 261, "lr": 4.6505606238476326e-05} +{"train_loss": 0.0150205809623003, "grad_norm": 0.11875386536121368, "global_step": 27680, "epoch": 261, "lr": 4.6502649724807535e-05} +{"train_loss": 0.013658074662089348, "grad_norm": 0.0695972740650177, "global_step": 27681, "epoch": 261, "lr": 4.649969322342691e-05} +{"train_loss": 0.017718464136123657, "grad_norm": 0.08651243895292282, "global_step": 27682, "epoch": 261, "lr": 4.649673673434483e-05} +{"train_loss": 0.015979869291186333, "grad_norm": 0.12617924809455872, "global_step": 27683, "epoch": 261, "lr": 4.649378025757171e-05} +{"train_loss": 0.015756960958242416, "grad_norm": 0.11745218932628632, "global_step": 27684, "epoch": 261, "lr": 4.649082379311792e-05} +{"train_loss": 0.015117939561605453, "grad_norm": 0.06082650274038315, "global_step": 27685, "epoch": 261, "lr": 4.648786734099386e-05} +{"train_loss": 0.016432229429483414, "grad_norm": 0.09953808039426804, "global_step": 27686, "epoch": 261, "lr": 4.648491090120989e-05} +{"train_loss": 0.015563245862722397, "grad_norm": 0.08636962622404099, "global_step": 27687, "epoch": 261, "lr": 4.6481954473776414e-05} +{"train_loss": 0.017475910484790802, "grad_norm": 0.06862594187259674, "global_step": 27688, "epoch": 261, "lr": 4.647899805870382e-05} +{"train_loss": 0.016518983989953995, "grad_norm": 0.09213639795780182, "global_step": 27689, "epoch": 261, "lr": 4.6476041656002487e-05} +{"train_loss": 0.017317509278655052, "grad_norm": 0.06866325438022614, "global_step": 27690, "epoch": 261, "lr": 4.647308526568283e-05} +{"train_loss": 0.015297740697860718, "grad_norm": 0.09997490793466568, "global_step": 27691, "epoch": 261, "lr": 4.647012888775522e-05} +{"train_loss": 0.01797698251903057, "grad_norm": 0.09111219644546509, "global_step": 27692, "epoch": 261, "lr": 4.646717252223004e-05} +{"train_loss": 0.014837962575256824, "grad_norm": 0.09122911095619202, "global_step": 27693, "epoch": 261, "lr": 4.646421616911767e-05} +{"train_loss": 0.015377188101410866, "grad_norm": 0.06241230294108391, "global_step": 27694, "epoch": 261, "lr": 4.64612598284285e-05} +{"train_loss": 0.014410057105123997, "grad_norm": 0.061784107238054276, "global_step": 27695, "epoch": 261, "lr": 4.645830350017293e-05} +{"train_loss": 0.016512073576450348, "grad_norm": 0.1243661567568779, "global_step": 27696, "epoch": 261, "lr": 4.6455347184361347e-05} +{"train_loss": 0.01700681820511818, "grad_norm": 0.09034125506877899, "global_step": 27697, "epoch": 261, "lr": 4.645239088100414e-05} +{"train_loss": 0.016405100002884865, "grad_norm": 0.13614454865455627, "global_step": 27698, "epoch": 261, "lr": 4.644943459011168e-05} +{"train_loss": 0.015461644157767296, "grad_norm": 0.12167663127183914, "global_step": 27699, "epoch": 261, "lr": 4.644647831169435e-05} +{"train_loss": 0.01598811149597168, "grad_norm": 0.1428021639585495, "global_step": 27700, "epoch": 261, "lr": 4.6443522045762565e-05} +{"train_loss": 0.015721525996923447, "grad_norm": 0.0744866281747818, "global_step": 27701, "epoch": 261, "lr": 4.6440565792326696e-05} +{"train_loss": 0.017389792948961258, "grad_norm": 0.09200850874185562, "global_step": 27702, "epoch": 261, "lr": 4.643760955139711e-05} +{"train_loss": 0.01561648864299059, "grad_norm": 0.09761283546686172, "global_step": 27703, "epoch": 261, "lr": 4.643465332298425e-05} +{"train_loss": 0.014330592006444931, "grad_norm": 0.06108025088906288, "global_step": 27704, "epoch": 261, "lr": 4.6431697107098456e-05} +{"train_loss": 0.015818770974874496, "grad_norm": 0.12413434684276581, "global_step": 27705, "epoch": 261, "lr": 4.642874090375012e-05} +{"train_loss": 0.014908287674188614, "grad_norm": 0.0693294107913971, "global_step": 27706, "epoch": 261, "lr": 4.6425784712949625e-05} +{"train_loss": 0.015731077641248703, "grad_norm": 0.05761509761214256, "global_step": 27707, "epoch": 261, "lr": 4.642282853470738e-05} +{"train_loss": 0.017199641093611717, "grad_norm": 0.11533311009407043, "global_step": 27708, "epoch": 261, "lr": 4.6419872369033765e-05} +{"train_loss": 0.013741958886384964, "grad_norm": 0.05662951245903969, "global_step": 27709, "epoch": 261, "lr": 4.641691621593917e-05} +{"train_loss": 0.01602061279118061, "grad_norm": 0.09824365377426147, "global_step": 27710, "epoch": 261, "lr": 4.641396007543395e-05} +{"train_loss": 0.014067322947084904, "grad_norm": 0.11720350384712219, "global_step": 27711, "epoch": 261, "lr": 4.641100394752852e-05} +{"train_loss": 0.015582231804728508, "grad_norm": 0.11078555136919022, "global_step": 27712, "epoch": 261, "lr": 4.6408047832233264e-05} +{"train_loss": 0.01507275179028511, "grad_norm": 0.07290301471948624, "global_step": 27713, "epoch": 261, "lr": 4.640509172955857e-05} +{"train_loss": 0.011785874143242836, "grad_norm": 0.0993601456284523, "global_step": 27714, "epoch": 261, "lr": 4.6402135639514825e-05} +{"train_loss": 0.013202053494751453, "grad_norm": 0.12404408305883408, "global_step": 27715, "epoch": 261, "lr": 4.639917956211239e-05} +{"train_loss": 0.017648007720708847, "grad_norm": 0.06986997276544571, "global_step": 27716, "epoch": 261, "lr": 4.639622349736169e-05} +{"train_loss": 0.01637333631515503, "grad_norm": 0.05574982985854149, "global_step": 27717, "epoch": 261, "lr": 4.639326744527309e-05} +{"train_loss": 0.014496874064207077, "grad_norm": 0.14370566606521606, "global_step": 27718, "epoch": 261, "lr": 4.639031140585697e-05} +{"train_loss": 0.013877295888960361, "grad_norm": 0.05611346662044525, "global_step": 27719, "epoch": 261, "lr": 4.6387355379123734e-05} +{"train_loss": 0.013683659955859184, "grad_norm": 0.07940513640642166, "global_step": 27720, "epoch": 261, "lr": 4.638439936508377e-05} +{"train_loss": 0.016111303120851517, "grad_norm": 0.09508007764816284, "global_step": 27721, "epoch": 261, "lr": 4.6381443363747444e-05} +{"train_loss": 0.015899114310741425, "grad_norm": 0.07696502655744553, "global_step": 27722, "epoch": 261, "lr": 4.637848737512516e-05} +{"train_loss": 0.014792785979807377, "grad_norm": 0.08744677901268005, "global_step": 27723, "epoch": 261, "lr": 4.637553139922727e-05} +{"train_loss": 0.014840789139270782, "grad_norm": 0.06353195011615753, "global_step": 27724, "epoch": 261, "lr": 4.637257543606421e-05} +{"train_loss": 0.016259804368019104, "grad_norm": 0.09644034504890442, "global_step": 27725, "epoch": 261, "lr": 4.6369619485646336e-05} +{"train_loss": 0.015422559343278408, "grad_norm": 0.06165697053074837, "global_step": 27726, "epoch": 261, "lr": 4.636666354798406e-05} +{"train_loss": 0.014574745669960976, "grad_norm": 0.0573214590549469, "global_step": 27727, "epoch": 261, "lr": 4.6363707623087726e-05} +{"train_loss": 0.015171057544648647, "grad_norm": 0.06181688979268074, "global_step": 27728, "epoch": 261, "lr": 4.6360751710967734e-05} +{"train_loss": 0.01526817586272955, "grad_norm": 0.0581013523042202, "global_step": 27729, "epoch": 261, "lr": 4.63577958116345e-05} +{"train_loss": 0.014710313640534878, "grad_norm": 0.0679471492767334, "global_step": 27730, "epoch": 261, "lr": 4.6354839925098376e-05} +{"train_loss": 0.015304012224078178, "grad_norm": 0.08639699220657349, "global_step": 27731, "epoch": 261, "lr": 4.6351884051369745e-05} +{"train_loss": 0.015431847423315048, "grad_norm": 0.08126374334096909, "global_step": 27732, "epoch": 261, "lr": 4.634892819045904e-05} +{"train_loss": 0.01600152999162674, "grad_norm": 0.12217210233211517, "global_step": 27733, "epoch": 261, "lr": 4.6345972342376596e-05} +{"train_loss": 0.014479228295385838, "grad_norm": 0.0732904002070427, "global_step": 27734, "epoch": 261, "lr": 4.634301650713282e-05} +{"train_loss": 0.015124978497624397, "grad_norm": 0.1015411838889122, "global_step": 27735, "epoch": 261, "lr": 4.634006068473808e-05} +{"train_loss": 0.015484409406781197, "grad_norm": 0.07644370943307877, "global_step": 27736, "epoch": 261, "lr": 4.633710487520279e-05} +{"train_loss": 0.014886507764458656, "grad_norm": 0.09218259900808334, "global_step": 27737, "epoch": 261, "lr": 4.6334149078537316e-05} +{"train_loss": 0.013982427306473255, "grad_norm": 0.11125747859477997, "global_step": 27738, "epoch": 261, "lr": 4.633119329475206e-05} +{"train_loss": 0.016701579093933105, "grad_norm": 0.08875289559364319, "global_step": 27739, "epoch": 261, "lr": 4.632823752385738e-05} +{"train_loss": 0.019400304183363914, "grad_norm": 0.08558500558137894, "global_step": 27740, "epoch": 261, "lr": 4.6325281765863674e-05} +{"train_loss": 0.014438060112297535, "grad_norm": 0.14812049269676208, "global_step": 27741, "epoch": 261, "lr": 4.632232602078133e-05} +{"train_loss": 0.01364518329501152, "grad_norm": 0.09138277918100357, "global_step": 27742, "epoch": 261, "lr": 4.631937028862074e-05} +{"train_loss": 0.01556469313800335, "grad_norm": 0.08094631880521774, "global_step": 27743, "epoch": 261, "lr": 4.631641456939227e-05} +{"train_loss": 0.015199003741145134, "grad_norm": 0.09644473344087601, "global_step": 27744, "epoch": 261, "lr": 4.631345886310634e-05} +{"train_loss": 0.01442944910377264, "grad_norm": 0.0864238291978836, "global_step": 27745, "epoch": 261, "lr": 4.63105031697733e-05} +{"train_loss": 0.013605844229459763, "grad_norm": 0.11746584624052048, "global_step": 27746, "epoch": 261, "lr": 4.630754748940355e-05} +{"train_loss": 0.016447944566607475, "grad_norm": 0.16021916270256042, "global_step": 27747, "epoch": 261, "lr": 4.630459182200746e-05} +{"train_loss": 0.01529225055128336, "grad_norm": 0.07418682426214218, "global_step": 27748, "epoch": 261, "lr": 4.630163616759544e-05} +{"train_loss": 0.018722761422395706, "grad_norm": 0.14472270011901855, "global_step": 27749, "epoch": 261, "lr": 4.629868052617786e-05} +{"train_loss": 0.012840572744607925, "grad_norm": 0.07484566420316696, "global_step": 27750, "epoch": 261, "lr": 4.6295724897765115e-05} +{"train_loss": 0.01510526705533266, "grad_norm": 0.08351563662290573, "global_step": 27751, "epoch": 261, "lr": 4.6292769282367574e-05} +{"train_loss": 0.015650443732738495, "grad_norm": 0.057586558163166046, "global_step": 27752, "epoch": 261, "lr": 4.628981367999562e-05} +{"train_loss": 0.015904217958450317, "grad_norm": 0.059750862419605255, "global_step": 27753, "epoch": 261, "lr": 4.6286858090659655e-05} +{"train_loss": 0.015535388141870499, "grad_norm": 0.09286236763000488, "global_step": 27754, "epoch": 261, "lr": 4.628390251437006e-05} +{"train_loss": 0.015807613730430603, "grad_norm": 0.06586197018623352, "global_step": 27755, "epoch": 261, "lr": 4.62809469511372e-05} +{"train_loss": 0.016327911987900734, "grad_norm": 0.0857769250869751, "global_step": 27756, "epoch": 261, "lr": 4.627799140097151e-05} +{"train_loss": 0.01603255234658718, "grad_norm": 0.10289350152015686, "global_step": 27757, "epoch": 261, "lr": 4.62750358638833e-05} +{"train_loss": 0.013346259482204914, "grad_norm": 0.08289454132318497, "global_step": 27758, "epoch": 261, "lr": 4.6272080339883014e-05} +{"train_loss": 0.014533642679452896, "grad_norm": 0.07493064552545547, "global_step": 27759, "epoch": 261, "lr": 4.6269124828981014e-05} +{"train_loss": 0.01452835462987423, "grad_norm": 0.07181296497583389, "global_step": 27760, "epoch": 261, "lr": 4.6266169331187676e-05} +{"train_loss": 0.013216368854045868, "grad_norm": 0.06076422333717346, "global_step": 27761, "epoch": 261, "lr": 4.626321384651341e-05} +{"train_loss": 0.01496692281216383, "grad_norm": 0.06615234911441803, "global_step": 27762, "epoch": 261, "lr": 4.626025837496859e-05} +{"train_loss": 0.016646940261125565, "grad_norm": 0.06246495246887207, "global_step": 27763, "epoch": 261, "lr": 4.6257302916563585e-05} +{"train_loss": 0.01688074879348278, "grad_norm": 0.13374099135398865, "global_step": 27764, "epoch": 261, "lr": 4.625434747130879e-05} +{"train_loss": 0.01358333695679903, "grad_norm": 0.06983213126659393, "global_step": 27765, "epoch": 261, "lr": 4.6251392039214583e-05} +{"train_loss": 0.014686908572912216, "grad_norm": 0.07275588065385818, "global_step": 27766, "epoch": 261, "lr": 4.624843662029136e-05} +{"train_loss": 0.016948789358139038, "grad_norm": 0.09381676465272903, "global_step": 27767, "epoch": 261, "lr": 4.624548121454951e-05} +{"train_loss": 0.016632722690701485, "grad_norm": 0.11407369375228882, "global_step": 27768, "epoch": 261, "lr": 4.624252582199941e-05} +{"train_loss": 0.016965746879577637, "grad_norm": 0.08616533875465393, "global_step": 27769, "epoch": 261, "lr": 4.623957044265141e-05} +{"train_loss": 0.014210041612386703, "grad_norm": 0.1269732415676117, "global_step": 27770, "epoch": 261, "lr": 4.623661507651594e-05} +{"train_loss": 0.015494839604592548, "grad_norm": 0.11062797904014587, "global_step": 27771, "epoch": 261, "lr": 4.623365972360337e-05, "val_loss": 0.019690997898578644} +{"train_loss": 0.01659976691007614, "grad_norm": 0.06873546540737152, "global_step": 27772, "epoch": 262, "lr": 4.6230704383924075e-05} +{"train_loss": 0.011910594999790192, "grad_norm": 0.04755705967545509, "global_step": 27773, "epoch": 262, "lr": 4.622774905748846e-05} +{"train_loss": 0.014831178821623325, "grad_norm": 0.060193512588739395, "global_step": 27774, "epoch": 262, "lr": 4.622479374430689e-05} +{"train_loss": 0.015694040805101395, "grad_norm": 0.0638665035367012, "global_step": 27775, "epoch": 262, "lr": 4.622183844438974e-05} +{"train_loss": 0.014099961146712303, "grad_norm": 0.05183309689164162, "global_step": 27776, "epoch": 262, "lr": 4.621888315774741e-05} +{"train_loss": 0.01684590056538582, "grad_norm": 0.10231749713420868, "global_step": 27777, "epoch": 262, "lr": 4.6215927884390285e-05} +{"train_loss": 0.01703217625617981, "grad_norm": 0.16014252603054047, "global_step": 27778, "epoch": 262, "lr": 4.621297262432875e-05} +{"train_loss": 0.01678849756717682, "grad_norm": 0.0733378455042839, "global_step": 27779, "epoch": 262, "lr": 4.621001737757318e-05} +{"train_loss": 0.014516589231789112, "grad_norm": 0.07386260479688644, "global_step": 27780, "epoch": 262, "lr": 4.620706214413396e-05} +{"train_loss": 0.01605388894677162, "grad_norm": 0.0652376338839531, "global_step": 27781, "epoch": 262, "lr": 4.620410692402146e-05} +{"train_loss": 0.014464380219578743, "grad_norm": 0.05284612253308296, "global_step": 27782, "epoch": 262, "lr": 4.6201151717246094e-05} +{"train_loss": 0.0167839378118515, "grad_norm": 0.11713355779647827, "global_step": 27783, "epoch": 262, "lr": 4.619819652381823e-05} +{"train_loss": 0.015728550031781197, "grad_norm": 0.07642385363578796, "global_step": 27784, "epoch": 262, "lr": 4.619524134374823e-05} +{"train_loss": 0.015648430213332176, "grad_norm": 0.07468385994434357, "global_step": 27785, "epoch": 262, "lr": 4.619228617704653e-05} +{"train_loss": 0.016672125086188316, "grad_norm": 0.0681244507431984, "global_step": 27786, "epoch": 262, "lr": 4.6189331023723455e-05} +{"train_loss": 0.015041236765682697, "grad_norm": 0.1342405080795288, "global_step": 27787, "epoch": 262, "lr": 4.618637588378942e-05} +{"train_loss": 0.015456820838153362, "grad_norm": 0.11480148881673813, "global_step": 27788, "epoch": 262, "lr": 4.6183420757254804e-05} +{"train_loss": 0.01540270447731018, "grad_norm": 0.07050396502017975, "global_step": 27789, "epoch": 262, "lr": 4.618046564412997e-05} +{"train_loss": 0.01564844138920307, "grad_norm": 0.05217339098453522, "global_step": 27790, "epoch": 262, "lr": 4.6177510544425335e-05} +{"train_loss": 0.014677256345748901, "grad_norm": 0.06625501811504364, "global_step": 27791, "epoch": 262, "lr": 4.617455545815128e-05} +{"train_loss": 0.015019074082374573, "grad_norm": 0.060721978545188904, "global_step": 27792, "epoch": 262, "lr": 4.617160038531816e-05} +{"train_loss": 0.01639753207564354, "grad_norm": 0.07184704393148422, "global_step": 27793, "epoch": 262, "lr": 4.6168645325936354e-05} +{"train_loss": 0.01573130302131176, "grad_norm": 0.06059534102678299, "global_step": 27794, "epoch": 262, "lr": 4.616569028001628e-05} +{"train_loss": 0.013514164835214615, "grad_norm": 0.06253107637166977, "global_step": 27795, "epoch": 262, "lr": 4.61627352475683e-05} +{"train_loss": 0.016289521008729935, "grad_norm": 0.07043479382991791, "global_step": 27796, "epoch": 262, "lr": 4.615978022860279e-05} +{"train_loss": 0.015544748865067959, "grad_norm": 0.05895364657044411, "global_step": 27797, "epoch": 262, "lr": 4.6156825223130176e-05} +{"train_loss": 0.01660153642296791, "grad_norm": 0.06342453509569168, "global_step": 27798, "epoch": 262, "lr": 4.615387023116077e-05} +{"train_loss": 0.015289269387722015, "grad_norm": 0.08627792447805405, "global_step": 27799, "epoch": 262, "lr": 4.615091525270501e-05} +{"train_loss": 0.01633034273982048, "grad_norm": 0.14487218856811523, "global_step": 27800, "epoch": 262, "lr": 4.6147960287773254e-05} +{"train_loss": 0.015526304952800274, "grad_norm": 0.060265034437179565, "global_step": 27801, "epoch": 262, "lr": 4.614500533637588e-05} +{"train_loss": 0.01495376043021679, "grad_norm": 0.1447821706533432, "global_step": 27802, "epoch": 262, "lr": 4.614205039852329e-05} +{"train_loss": 0.015509950928390026, "grad_norm": 0.08381486684083939, "global_step": 27803, "epoch": 262, "lr": 4.613909547422587e-05} +{"train_loss": 0.014737699180841446, "grad_norm": 0.06614376604557037, "global_step": 27804, "epoch": 262, "lr": 4.613614056349398e-05} +{"train_loss": 0.015910133719444275, "grad_norm": 0.08368422091007233, "global_step": 27805, "epoch": 262, "lr": 4.6133185666338e-05} +{"train_loss": 0.015474982559680939, "grad_norm": 0.05308925360441208, "global_step": 27806, "epoch": 262, "lr": 4.613023078276833e-05} +{"train_loss": 0.01452350988984108, "grad_norm": 0.1354081630706787, "global_step": 27807, "epoch": 262, "lr": 4.612727591279535e-05} +{"train_loss": 0.016947053372859955, "grad_norm": 0.09715183079242706, "global_step": 27808, "epoch": 262, "lr": 4.612432105642943e-05} +{"train_loss": 0.016666661947965622, "grad_norm": 0.06960756331682205, "global_step": 27809, "epoch": 262, "lr": 4.612136621368099e-05} +{"train_loss": 0.017038725316524506, "grad_norm": 0.059843145310878754, "global_step": 27810, "epoch": 262, "lr": 4.611841138456035e-05} +{"train_loss": 0.016553640365600586, "grad_norm": 0.11266475170850754, "global_step": 27811, "epoch": 262, "lr": 4.6115456569077936e-05} +{"train_loss": 0.01468693744391203, "grad_norm": 0.07231706380844116, "global_step": 27812, "epoch": 262, "lr": 4.611250176724412e-05} +{"train_loss": 0.017046688124537468, "grad_norm": 0.05651417002081871, "global_step": 27813, "epoch": 262, "lr": 4.6109546979069265e-05} +{"train_loss": 0.013604886829853058, "grad_norm": 0.06275834143161774, "global_step": 27814, "epoch": 262, "lr": 4.610659220456381e-05} +{"train_loss": 0.014177034609019756, "grad_norm": 0.06825776398181915, "global_step": 27815, "epoch": 262, "lr": 4.610363744373806e-05} +{"train_loss": 0.013054355047643185, "grad_norm": 0.05714523047208786, "global_step": 27816, "epoch": 262, "lr": 4.610068269660245e-05} +{"train_loss": 0.014912394806742668, "grad_norm": 0.13340699672698975, "global_step": 27817, "epoch": 262, "lr": 4.609772796316733e-05} +{"train_loss": 0.015073142945766449, "grad_norm": 0.1249169260263443, "global_step": 27818, "epoch": 262, "lr": 4.60947732434431e-05} +{"train_loss": 0.014271344989538193, "grad_norm": 0.09449906647205353, "global_step": 27819, "epoch": 262, "lr": 4.609181853744014e-05} +{"train_loss": 0.014962786808609962, "grad_norm": 0.07021905481815338, "global_step": 27820, "epoch": 262, "lr": 4.6088863845168855e-05} +{"train_loss": 0.016707487404346466, "grad_norm": 0.07898515462875366, "global_step": 27821, "epoch": 262, "lr": 4.608590916663959e-05} +{"train_loss": 0.014526464976370335, "grad_norm": 0.04972619190812111, "global_step": 27822, "epoch": 262, "lr": 4.608295450186271e-05} +{"train_loss": 0.016483275219798088, "grad_norm": 0.07155811786651611, "global_step": 27823, "epoch": 262, "lr": 4.607999985084865e-05} +{"train_loss": 0.01686159521341324, "grad_norm": 0.05840126425027847, "global_step": 27824, "epoch": 262, "lr": 4.607704521360776e-05} +{"train_loss": 0.015675880014896393, "grad_norm": 0.15032973885536194, "global_step": 27825, "epoch": 262, "lr": 4.607409059015041e-05} +{"train_loss": 0.01831173151731491, "grad_norm": 0.12102219462394714, "global_step": 27826, "epoch": 262, "lr": 4.607113598048704e-05} +{"train_loss": 0.01620868593454361, "grad_norm": 0.06119832023978233, "global_step": 27827, "epoch": 262, "lr": 4.606818138462796e-05} +{"train_loss": 0.011586316861212254, "grad_norm": 0.09707926213741302, "global_step": 27828, "epoch": 262, "lr": 4.6065226802583584e-05} +{"train_loss": 0.01787232607603073, "grad_norm": 0.09022266417741776, "global_step": 27829, "epoch": 262, "lr": 4.606227223436429e-05} +{"train_loss": 0.013491918332874775, "grad_norm": 0.10959045588970184, "global_step": 27830, "epoch": 262, "lr": 4.605931767998045e-05} +{"train_loss": 0.017313513904809952, "grad_norm": 0.11716040968894958, "global_step": 27831, "epoch": 262, "lr": 4.605636313944247e-05} +{"train_loss": 0.017298411577939987, "grad_norm": 0.1254490464925766, "global_step": 27832, "epoch": 262, "lr": 4.605340861276072e-05} +{"train_loss": 0.015433110296726227, "grad_norm": 0.08182433992624283, "global_step": 27833, "epoch": 262, "lr": 4.605045409994556e-05} +{"train_loss": 0.015750113874673843, "grad_norm": 0.07784204185009003, "global_step": 27834, "epoch": 262, "lr": 4.604749960100738e-05} +{"train_loss": 0.014379093423485756, "grad_norm": 0.07284973561763763, "global_step": 27835, "epoch": 262, "lr": 4.604454511595658e-05} +{"train_loss": 0.013691294007003307, "grad_norm": 0.04793848097324371, "global_step": 27836, "epoch": 262, "lr": 4.6041590644803535e-05} +{"train_loss": 0.01448679156601429, "grad_norm": 0.10254820436239243, "global_step": 27837, "epoch": 262, "lr": 4.6038636187558595e-05} +{"train_loss": 0.017724424600601196, "grad_norm": 0.15063263475894928, "global_step": 27838, "epoch": 262, "lr": 4.603568174423221e-05} +{"train_loss": 0.017291970551013947, "grad_norm": 0.08320051431655884, "global_step": 27839, "epoch": 262, "lr": 4.603272731483467e-05} +{"train_loss": 0.015283049084246159, "grad_norm": 0.07582476735115051, "global_step": 27840, "epoch": 262, "lr": 4.602977289937642e-05} +{"train_loss": 0.016630608588457108, "grad_norm": 0.08993843197822571, "global_step": 27841, "epoch": 262, "lr": 4.602681849786783e-05} +{"train_loss": 0.01377624087035656, "grad_norm": 0.08671657741069794, "global_step": 27842, "epoch": 262, "lr": 4.6023864110319245e-05} +{"train_loss": 0.014558963477611542, "grad_norm": 0.10444945842027664, "global_step": 27843, "epoch": 262, "lr": 4.602090973674109e-05} +{"train_loss": 0.01477036066353321, "grad_norm": 0.08361678570508957, "global_step": 27844, "epoch": 262, "lr": 4.601795537714373e-05} +{"train_loss": 0.01491217315196991, "grad_norm": 0.07147137075662613, "global_step": 27845, "epoch": 262, "lr": 4.601500103153754e-05} +{"train_loss": 0.014890449121594429, "grad_norm": 0.056573536247015, "global_step": 27846, "epoch": 262, "lr": 4.6012046699932894e-05} +{"train_loss": 0.01389295794069767, "grad_norm": 0.08749551326036453, "global_step": 27847, "epoch": 262, "lr": 4.6009092382340194e-05} +{"train_loss": 0.016342168673872948, "grad_norm": 0.06251626461744308, "global_step": 27848, "epoch": 262, "lr": 4.60061380787698e-05} +{"train_loss": 0.01436383742839098, "grad_norm": 0.10003171861171722, "global_step": 27849, "epoch": 262, "lr": 4.60031837892321e-05} +{"train_loss": 0.015627535060048103, "grad_norm": 0.09309343248605728, "global_step": 27850, "epoch": 262, "lr": 4.600022951373749e-05} +{"train_loss": 0.013903059996664524, "grad_norm": 0.057503387331962585, "global_step": 27851, "epoch": 262, "lr": 4.599727525229631e-05} +{"train_loss": 0.014353952370584011, "grad_norm": 0.09423239529132843, "global_step": 27852, "epoch": 262, "lr": 4.599432100491898e-05} +{"train_loss": 0.01467689499258995, "grad_norm": 0.08954177796840668, "global_step": 27853, "epoch": 262, "lr": 4.599136677161586e-05} +{"train_loss": 0.016932204365730286, "grad_norm": 0.10377011448144913, "global_step": 27854, "epoch": 262, "lr": 4.598841255239733e-05} +{"train_loss": 0.014094125479459763, "grad_norm": 0.057247478514909744, "global_step": 27855, "epoch": 262, "lr": 4.598545834727377e-05} +{"train_loss": 0.019328976050019264, "grad_norm": 0.08367355912923813, "global_step": 27856, "epoch": 262, "lr": 4.598250415625558e-05} +{"train_loss": 0.01607438549399376, "grad_norm": 0.05942697077989578, "global_step": 27857, "epoch": 262, "lr": 4.597954997935312e-05} +{"train_loss": 0.015528939664363861, "grad_norm": 0.08660992234945297, "global_step": 27858, "epoch": 262, "lr": 4.597659581657677e-05} +{"train_loss": 0.015378919430077076, "grad_norm": 0.07124044746160507, "global_step": 27859, "epoch": 262, "lr": 4.597364166793689e-05} +{"train_loss": 0.013243146240711212, "grad_norm": 0.06546945869922638, "global_step": 27860, "epoch": 262, "lr": 4.597068753344391e-05} +{"train_loss": 0.0182008258998394, "grad_norm": 0.09525210410356522, "global_step": 27861, "epoch": 262, "lr": 4.5967733413108174e-05} +{"train_loss": 0.016959264874458313, "grad_norm": 0.0934247374534607, "global_step": 27862, "epoch": 262, "lr": 4.5964779306940074e-05} +{"train_loss": 0.016522090882062912, "grad_norm": 0.10193703323602676, "global_step": 27863, "epoch": 262, "lr": 4.596182521494997e-05} +{"train_loss": 0.014129932038486004, "grad_norm": 0.07234909385442734, "global_step": 27864, "epoch": 262, "lr": 4.595887113714827e-05} +{"train_loss": 0.01414097473025322, "grad_norm": 0.08724084496498108, "global_step": 27865, "epoch": 262, "lr": 4.595591707354533e-05} +{"train_loss": 0.016137156635522842, "grad_norm": 0.08079493790864944, "global_step": 27866, "epoch": 262, "lr": 4.5952963024151536e-05} +{"train_loss": 0.017296355217695236, "grad_norm": 0.08191283047199249, "global_step": 27867, "epoch": 262, "lr": 4.59500089889773e-05} +{"train_loss": 0.015263608656823635, "grad_norm": 0.07285638153553009, "global_step": 27868, "epoch": 262, "lr": 4.594705496803294e-05} +{"train_loss": 0.014129048213362694, "grad_norm": 0.06289328634738922, "global_step": 27869, "epoch": 262, "lr": 4.5944100961328876e-05} +{"train_loss": 0.018927454948425293, "grad_norm": 0.17466090619564056, "global_step": 27870, "epoch": 262, "lr": 4.594114696887547e-05} +{"train_loss": 0.013136763125658035, "grad_norm": 0.0991448238492012, "global_step": 27871, "epoch": 262, "lr": 4.5938192990683106e-05} +{"train_loss": 0.015619440004229546, "grad_norm": 0.06914794445037842, "global_step": 27872, "epoch": 262, "lr": 4.5935239026762176e-05} +{"train_loss": 0.015116922557353973, "grad_norm": 0.09668111801147461, "global_step": 27873, "epoch": 262, "lr": 4.5932285077123054e-05} +{"train_loss": 0.014711379073560238, "grad_norm": 0.07842118293046951, "global_step": 27874, "epoch": 262, "lr": 4.59293311417761e-05} +{"train_loss": 0.014409679919481277, "grad_norm": 0.07508010417222977, "global_step": 27875, "epoch": 262, "lr": 4.592637722073171e-05} +{"train_loss": 0.013619998469948769, "grad_norm": 0.09695201367139816, "global_step": 27876, "epoch": 262, "lr": 4.592342331400026e-05} +{"train_loss": 0.015495373797163649, "grad_norm": 0.1290998011827469, "global_step": 27877, "epoch": 262, "lr": 4.592046942159213e-05, "val_loss": 0.018941570073366165} +{"train_loss": 0.014716895297169685, "grad_norm": 0.10475698858499527, "global_step": 27878, "epoch": 263, "lr": 4.591751554351768e-05} +{"train_loss": 0.016703424975275993, "grad_norm": 0.08082917332649231, "global_step": 27879, "epoch": 263, "lr": 4.5914561679787345e-05} +{"train_loss": 0.015855489298701286, "grad_norm": 0.11105143278837204, "global_step": 27880, "epoch": 263, "lr": 4.5911607830411425e-05} +{"train_loss": 0.014685065485537052, "grad_norm": 0.07373388856649399, "global_step": 27881, "epoch": 263, "lr": 4.5908653995400344e-05} +{"train_loss": 0.013574624434113503, "grad_norm": 0.08514147996902466, "global_step": 27882, "epoch": 263, "lr": 4.590570017476449e-05} +{"train_loss": 0.013390911743044853, "grad_norm": 0.05893382057547569, "global_step": 27883, "epoch": 263, "lr": 4.59027463685142e-05} +{"train_loss": 0.015284046530723572, "grad_norm": 0.1005391851067543, "global_step": 27884, "epoch": 263, "lr": 4.589979257665989e-05} +{"train_loss": 0.014944910071790218, "grad_norm": 0.09264475107192993, "global_step": 27885, "epoch": 263, "lr": 4.589683879921194e-05} +{"train_loss": 0.014896806329488754, "grad_norm": 0.14111414551734924, "global_step": 27886, "epoch": 263, "lr": 4.58938850361807e-05} +{"train_loss": 0.0155815863981843, "grad_norm": 0.10727662593126297, "global_step": 27887, "epoch": 263, "lr": 4.5890931287576565e-05} +{"train_loss": 0.01467402745038271, "grad_norm": 0.12650536000728607, "global_step": 27888, "epoch": 263, "lr": 4.58879775534099e-05} +{"train_loss": 0.016775308176875114, "grad_norm": 0.07345494627952576, "global_step": 27889, "epoch": 263, "lr": 4.58850238336911e-05} +{"train_loss": 0.01646437682211399, "grad_norm": 0.07391058653593063, "global_step": 27890, "epoch": 263, "lr": 4.588207012843054e-05} +{"train_loss": 0.016783276572823524, "grad_norm": 0.10991103202104568, "global_step": 27891, "epoch": 263, "lr": 4.587911643763861e-05} +{"train_loss": 0.01678585633635521, "grad_norm": 0.0922311320900917, "global_step": 27892, "epoch": 263, "lr": 4.587616276132565e-05} +{"train_loss": 0.015720508992671967, "grad_norm": 0.09466003626585007, "global_step": 27893, "epoch": 263, "lr": 4.587320909950206e-05} +{"train_loss": 0.018757492303848267, "grad_norm": 0.05855084955692291, "global_step": 27894, "epoch": 263, "lr": 4.587025545217823e-05} +{"train_loss": 0.01447016466408968, "grad_norm": 0.07268023490905762, "global_step": 27895, "epoch": 263, "lr": 4.586730181936451e-05} +{"train_loss": 0.015438973903656006, "grad_norm": 0.17990493774414062, "global_step": 27896, "epoch": 263, "lr": 4.586434820107131e-05} +{"train_loss": 0.016131769865751266, "grad_norm": 0.09166108071804047, "global_step": 27897, "epoch": 263, "lr": 4.5861394597309e-05} +{"train_loss": 0.01478938851505518, "grad_norm": 0.0693153664469719, "global_step": 27898, "epoch": 263, "lr": 4.585844100808794e-05} +{"train_loss": 0.016434788703918457, "grad_norm": 0.04855313152074814, "global_step": 27899, "epoch": 263, "lr": 4.585548743341851e-05} +{"train_loss": 0.012899590656161308, "grad_norm": 0.08599600195884705, "global_step": 27900, "epoch": 263, "lr": 4.585253387331109e-05} +{"train_loss": 0.0161733441054821, "grad_norm": 0.08249716460704803, "global_step": 27901, "epoch": 263, "lr": 4.5849580327776076e-05} +{"train_loss": 0.016554098576307297, "grad_norm": 0.08822637796401978, "global_step": 27902, "epoch": 263, "lr": 4.584662679682383e-05} +{"train_loss": 0.015402777120471, "grad_norm": 0.10219129920005798, "global_step": 27903, "epoch": 263, "lr": 4.584367328046474e-05} +{"train_loss": 0.01665736734867096, "grad_norm": 0.055551677942276, "global_step": 27904, "epoch": 263, "lr": 4.584071977870915e-05} +{"train_loss": 0.01630839891731739, "grad_norm": 0.09488903731107712, "global_step": 27905, "epoch": 263, "lr": 4.583776629156748e-05} +{"train_loss": 0.014083953574299812, "grad_norm": 0.11497291922569275, "global_step": 27906, "epoch": 263, "lr": 4.5834812819050085e-05} +{"train_loss": 0.015098320320248604, "grad_norm": 0.07087408751249313, "global_step": 27907, "epoch": 263, "lr": 4.583185936116733e-05} +{"train_loss": 0.014529511332511902, "grad_norm": 0.06659524142742157, "global_step": 27908, "epoch": 263, "lr": 4.582890591792963e-05} +{"train_loss": 0.014279941096901894, "grad_norm": 0.10906781256198883, "global_step": 27909, "epoch": 263, "lr": 4.5825952489347346e-05} +{"train_loss": 0.016864100471138954, "grad_norm": 0.12456190586090088, "global_step": 27910, "epoch": 263, "lr": 4.582299907543084e-05} +{"train_loss": 0.013679870404303074, "grad_norm": 0.06343681365251541, "global_step": 27911, "epoch": 263, "lr": 4.5820045676190496e-05} +{"train_loss": 0.015268661081790924, "grad_norm": 0.07744181901216507, "global_step": 27912, "epoch": 263, "lr": 4.5817092291636686e-05} +{"train_loss": 0.01694473624229431, "grad_norm": 0.11398985981941223, "global_step": 27913, "epoch": 263, "lr": 4.5814138921779806e-05} +{"train_loss": 0.015622071921825409, "grad_norm": 0.07983022928237915, "global_step": 27914, "epoch": 263, "lr": 4.5811185566630224e-05} +{"train_loss": 0.014382634311914444, "grad_norm": 0.06741008907556534, "global_step": 27915, "epoch": 263, "lr": 4.580823222619832e-05} +{"train_loss": 0.015491990372538567, "grad_norm": 0.07882080227136612, "global_step": 27916, "epoch": 263, "lr": 4.5805278900494455e-05} +{"train_loss": 0.015793651342391968, "grad_norm": 0.07993197441101074, "global_step": 27917, "epoch": 263, "lr": 4.5802325589529004e-05} +{"train_loss": 0.01652788557112217, "grad_norm": 0.07219294458627701, "global_step": 27918, "epoch": 263, "lr": 4.579937229331237e-05} +{"train_loss": 0.0167926587164402, "grad_norm": 0.07525388151407242, "global_step": 27919, "epoch": 263, "lr": 4.579641901185491e-05} +{"train_loss": 0.017688695341348648, "grad_norm": 0.07264430075883865, "global_step": 27920, "epoch": 263, "lr": 4.579346574516702e-05} +{"train_loss": 0.018703270703554153, "grad_norm": 0.14217732846736908, "global_step": 27921, "epoch": 263, "lr": 4.579051249325903e-05} +{"train_loss": 0.0195387564599514, "grad_norm": 0.06683482229709625, "global_step": 27922, "epoch": 263, "lr": 4.578755925614138e-05} +{"train_loss": 0.0145559785887599, "grad_norm": 0.06335653364658356, "global_step": 27923, "epoch": 263, "lr": 4.57846060338244e-05} +{"train_loss": 0.015058547258377075, "grad_norm": 0.1370135247707367, "global_step": 27924, "epoch": 263, "lr": 4.578165282631847e-05} +{"train_loss": 0.015400995500385761, "grad_norm": 0.09173844754695892, "global_step": 27925, "epoch": 263, "lr": 4.5778699633633994e-05} +{"train_loss": 0.014239270240068436, "grad_norm": 0.08949591964483261, "global_step": 27926, "epoch": 263, "lr": 4.577574645578134e-05} +{"train_loss": 0.0162542462348938, "grad_norm": 0.07315421104431152, "global_step": 27927, "epoch": 263, "lr": 4.5772793292770864e-05} +{"train_loss": 0.014493763446807861, "grad_norm": 0.07043354958295822, "global_step": 27928, "epoch": 263, "lr": 4.576984014461296e-05} +{"train_loss": 0.014311875216662884, "grad_norm": 0.09155687689781189, "global_step": 27929, "epoch": 263, "lr": 4.5766887011317975e-05} +{"train_loss": 0.015520323067903519, "grad_norm": 0.07930301129817963, "global_step": 27930, "epoch": 263, "lr": 4.576393389289633e-05} +{"train_loss": 0.015575382858514786, "grad_norm": 0.07269859313964844, "global_step": 27931, "epoch": 263, "lr": 4.5760980789358375e-05} +{"train_loss": 0.01401800662279129, "grad_norm": 0.13413666188716888, "global_step": 27932, "epoch": 263, "lr": 4.57580277007145e-05} +{"train_loss": 0.01654387265443802, "grad_norm": 0.07069505751132965, "global_step": 27933, "epoch": 263, "lr": 4.5755074626975055e-05} +{"train_loss": 0.01504763588309288, "grad_norm": 0.05401114001870155, "global_step": 27934, "epoch": 263, "lr": 4.5752121568150435e-05} +{"train_loss": 0.01575077511370182, "grad_norm": 0.146937295794487, "global_step": 27935, "epoch": 263, "lr": 4.574916852425102e-05} +{"train_loss": 0.015158995985984802, "grad_norm": 0.061332181096076965, "global_step": 27936, "epoch": 263, "lr": 4.574621549528716e-05} +{"train_loss": 0.012481320649385452, "grad_norm": 0.051633741706609726, "global_step": 27937, "epoch": 263, "lr": 4.574326248126926e-05} +{"train_loss": 0.014091181568801403, "grad_norm": 0.06234476715326309, "global_step": 27938, "epoch": 263, "lr": 4.57403094822077e-05} +{"train_loss": 0.012512891553342342, "grad_norm": 0.08328710496425629, "global_step": 27939, "epoch": 263, "lr": 4.573735649811283e-05} +{"train_loss": 0.015304146334528923, "grad_norm": 0.0584491491317749, "global_step": 27940, "epoch": 263, "lr": 4.573440352899503e-05} +{"train_loss": 0.015632782131433487, "grad_norm": 0.0778658539056778, "global_step": 27941, "epoch": 263, "lr": 4.573145057486467e-05} +{"train_loss": 0.014845035038888454, "grad_norm": 0.07743106037378311, "global_step": 27942, "epoch": 263, "lr": 4.5728497635732155e-05} +{"train_loss": 0.014044500887393951, "grad_norm": 0.05379488319158554, "global_step": 27943, "epoch": 263, "lr": 4.572554471160783e-05} +{"train_loss": 0.016073741018772125, "grad_norm": 0.07084233313798904, "global_step": 27944, "epoch": 263, "lr": 4.5722591802502105e-05} +{"train_loss": 0.013636564835906029, "grad_norm": 0.0965164303779602, "global_step": 27945, "epoch": 263, "lr": 4.571963890842531e-05} +{"train_loss": 0.015364324674010277, "grad_norm": 0.08383236825466156, "global_step": 27946, "epoch": 263, "lr": 4.571668602938784e-05} +{"train_loss": 0.016237858682870865, "grad_norm": 0.06588871777057648, "global_step": 27947, "epoch": 263, "lr": 4.571373316540008e-05} +{"train_loss": 0.01689205691218376, "grad_norm": 0.11928470432758331, "global_step": 27948, "epoch": 263, "lr": 4.571078031647239e-05} +{"train_loss": 0.0158253014087677, "grad_norm": 0.10227900743484497, "global_step": 27949, "epoch": 263, "lr": 4.5707827482615165e-05} +{"train_loss": 0.017554476857185364, "grad_norm": 0.07629323750734329, "global_step": 27950, "epoch": 263, "lr": 4.570487466383877e-05} +{"train_loss": 0.01750263012945652, "grad_norm": 0.13214147090911865, "global_step": 27951, "epoch": 263, "lr": 4.570192186015357e-05} +{"train_loss": 0.015058944001793861, "grad_norm": 0.1073225811123848, "global_step": 27952, "epoch": 263, "lr": 4.5698969071569946e-05} +{"train_loss": 0.01547286193817854, "grad_norm": 0.09396035969257355, "global_step": 27953, "epoch": 263, "lr": 4.5696016298098265e-05} +{"train_loss": 0.01518059242516756, "grad_norm": 0.06400403380393982, "global_step": 27954, "epoch": 263, "lr": 4.569306353974893e-05} +{"train_loss": 0.01719590090215206, "grad_norm": 0.07619789242744446, "global_step": 27955, "epoch": 263, "lr": 4.569011079653228e-05} +{"train_loss": 0.01635020226240158, "grad_norm": 0.1020130142569542, "global_step": 27956, "epoch": 263, "lr": 4.568715806845873e-05} +{"train_loss": 0.01352774165570736, "grad_norm": 0.0764724537730217, "global_step": 27957, "epoch": 263, "lr": 4.5684205355538616e-05} +{"train_loss": 0.01488417387008667, "grad_norm": 0.127783864736557, "global_step": 27958, "epoch": 263, "lr": 4.568125265778232e-05} +{"train_loss": 0.015329371206462383, "grad_norm": 0.109792560338974, "global_step": 27959, "epoch": 263, "lr": 4.567829997520023e-05} +{"train_loss": 0.012734773568809032, "grad_norm": 0.05230342596769333, "global_step": 27960, "epoch": 263, "lr": 4.567534730780272e-05} +{"train_loss": 0.016344083473086357, "grad_norm": 0.09218916296958923, "global_step": 27961, "epoch": 263, "lr": 4.5672394655600144e-05} +{"train_loss": 0.013769902288913727, "grad_norm": 0.08265678584575653, "global_step": 27962, "epoch": 263, "lr": 4.566944201860291e-05} +{"train_loss": 0.01436894852668047, "grad_norm": 0.07436974346637726, "global_step": 27963, "epoch": 263, "lr": 4.566648939682137e-05} +{"train_loss": 0.013613738119602203, "grad_norm": 0.05060431361198425, "global_step": 27964, "epoch": 263, "lr": 4.5663536790265896e-05} +{"train_loss": 0.01660173572599888, "grad_norm": 0.06659732013940811, "global_step": 27965, "epoch": 263, "lr": 4.5660584198946854e-05} +{"train_loss": 0.01624738797545433, "grad_norm": 0.095375657081604, "global_step": 27966, "epoch": 263, "lr": 4.565763162287465e-05} +{"train_loss": 0.015119043178856373, "grad_norm": 0.05835779383778572, "global_step": 27967, "epoch": 263, "lr": 4.565467906205965e-05} +{"train_loss": 0.012391532771289349, "grad_norm": 0.10376501083374023, "global_step": 27968, "epoch": 263, "lr": 4.565172651651221e-05} +{"train_loss": 0.018059801310300827, "grad_norm": 0.06396123766899109, "global_step": 27969, "epoch": 263, "lr": 4.56487739862427e-05} +{"train_loss": 0.016012098640203476, "grad_norm": 0.06501786410808563, "global_step": 27970, "epoch": 263, "lr": 4.5645821471261505e-05} +{"train_loss": 0.015564107336103916, "grad_norm": 0.07540281862020493, "global_step": 27971, "epoch": 263, "lr": 4.564286897157901e-05} +{"train_loss": 0.014736597426235676, "grad_norm": 0.09167865663766861, "global_step": 27972, "epoch": 263, "lr": 4.563991648720557e-05} +{"train_loss": 0.015289283357560635, "grad_norm": 0.08658964931964874, "global_step": 27973, "epoch": 263, "lr": 4.563696401815158e-05} +{"train_loss": 0.015242256224155426, "grad_norm": 0.16092799603939056, "global_step": 27974, "epoch": 263, "lr": 4.563401156442739e-05} +{"train_loss": 0.016424359753727913, "grad_norm": 0.11038784682750702, "global_step": 27975, "epoch": 263, "lr": 4.5631059126043394e-05} +{"train_loss": 0.018390705808997154, "grad_norm": 0.09203445911407471, "global_step": 27976, "epoch": 263, "lr": 4.562810670300995e-05} +{"train_loss": 0.01784585602581501, "grad_norm": 0.05776996910572052, "global_step": 27977, "epoch": 263, "lr": 4.562515429533742e-05} +{"train_loss": 0.017901185899972916, "grad_norm": 0.12179336696863174, "global_step": 27978, "epoch": 263, "lr": 4.562220190303622e-05} +{"train_loss": 0.01617690920829773, "grad_norm": 0.0725424513220787, "global_step": 27979, "epoch": 263, "lr": 4.5619249526116706e-05} +{"train_loss": 0.014091286808252335, "grad_norm": 0.0765627920627594, "global_step": 27980, "epoch": 263, "lr": 4.561629716458923e-05} +{"train_loss": 0.014695042744278908, "grad_norm": 0.053586024791002274, "global_step": 27981, "epoch": 263, "lr": 4.561334481846418e-05} +{"train_loss": 0.014881055802106857, "grad_norm": 0.09989448636770248, "global_step": 27982, "epoch": 263, "lr": 4.561039248775192e-05} +{"train_loss": 0.01547864579402332, "grad_norm": 0.07261334359645844, "global_step": 27983, "epoch": 263, "lr": 4.560744017246284e-05, "val_loss": 0.019416967406868935} +{"train_loss": 0.0164871197193861, "grad_norm": 0.057195648550987244, "global_step": 27984, "epoch": 264, "lr": 4.560448787260731e-05} +{"train_loss": 0.014050271362066269, "grad_norm": 0.0956452414393425, "global_step": 27985, "epoch": 264, "lr": 4.560153558819571e-05} +{"train_loss": 0.01632729172706604, "grad_norm": 0.0883314460515976, "global_step": 27986, "epoch": 264, "lr": 4.559858331923838e-05} +{"train_loss": 0.015840988606214523, "grad_norm": 0.08867482841014862, "global_step": 27987, "epoch": 264, "lr": 4.559563106574571e-05} +{"train_loss": 0.013232178054749966, "grad_norm": 0.07053685933351517, "global_step": 27988, "epoch": 264, "lr": 4.5592678827728085e-05} +{"train_loss": 0.01513365376740694, "grad_norm": 0.08270246535539627, "global_step": 27989, "epoch": 264, "lr": 4.558972660519587e-05} +{"train_loss": 0.016068585216999054, "grad_norm": 0.10251979529857635, "global_step": 27990, "epoch": 264, "lr": 4.558677439815943e-05} +{"train_loss": 0.017177250236272812, "grad_norm": 0.06202757731080055, "global_step": 27991, "epoch": 264, "lr": 4.558382220662917e-05} +{"train_loss": 0.016359999775886536, "grad_norm": 0.09883853793144226, "global_step": 27992, "epoch": 264, "lr": 4.558087003061542e-05} +{"train_loss": 0.014712350443005562, "grad_norm": 0.05885936692357063, "global_step": 27993, "epoch": 264, "lr": 4.557791787012857e-05} +{"train_loss": 0.016603998839855194, "grad_norm": 0.05266955494880676, "global_step": 27994, "epoch": 264, "lr": 4.5574965725178985e-05} +{"train_loss": 0.01432529091835022, "grad_norm": 0.07344881445169449, "global_step": 27995, "epoch": 264, "lr": 4.557201359577706e-05} +{"train_loss": 0.01559259183704853, "grad_norm": 0.06226569786667824, "global_step": 27996, "epoch": 264, "lr": 4.556906148193315e-05} +{"train_loss": 0.016831835731863976, "grad_norm": 0.11509264260530472, "global_step": 27997, "epoch": 264, "lr": 4.5566109383657646e-05} +{"train_loss": 0.015792187303304672, "grad_norm": 0.06451644003391266, "global_step": 27998, "epoch": 264, "lr": 4.556315730096089e-05} +{"train_loss": 0.016002092510461807, "grad_norm": 0.06198321655392647, "global_step": 27999, "epoch": 264, "lr": 4.5560205233853266e-05} +{"train_loss": 0.01675383560359478, "grad_norm": 0.08474620431661606, "global_step": 28000, "epoch": 264, "lr": 4.5557253182345156e-05} +{"train_loss": 0.01739485375583172, "grad_norm": 0.07218937575817108, "global_step": 28001, "epoch": 264, "lr": 4.555430114644692e-05} +{"train_loss": 0.01721423864364624, "grad_norm": 0.06269384175539017, "global_step": 28002, "epoch": 264, "lr": 4.555134912616894e-05} +{"train_loss": 0.014741554856300354, "grad_norm": 0.11482679843902588, "global_step": 28003, "epoch": 264, "lr": 4.554839712152159e-05} +{"train_loss": 0.015419090166687965, "grad_norm": 0.06959269195795059, "global_step": 28004, "epoch": 264, "lr": 4.554544513251524e-05} +{"train_loss": 0.015201445668935776, "grad_norm": 0.13098612427711487, "global_step": 28005, "epoch": 264, "lr": 4.554249315916025e-05} +{"train_loss": 0.014708403497934341, "grad_norm": 0.06612633913755417, "global_step": 28006, "epoch": 264, "lr": 4.553954120146699e-05} +{"train_loss": 0.015399754047393799, "grad_norm": 0.08625061064958572, "global_step": 28007, "epoch": 264, "lr": 4.5536589259445857e-05} +{"train_loss": 0.016403447836637497, "grad_norm": 0.13416625559329987, "global_step": 28008, "epoch": 264, "lr": 4.55336373331072e-05} +{"train_loss": 0.016421057283878326, "grad_norm": 0.11662411689758301, "global_step": 28009, "epoch": 264, "lr": 4.553068542246142e-05} +{"train_loss": 0.016759445890784264, "grad_norm": 0.11253885924816132, "global_step": 28010, "epoch": 264, "lr": 4.5527733527518856e-05} +{"train_loss": 0.014064199291169643, "grad_norm": 0.07510460168123245, "global_step": 28011, "epoch": 264, "lr": 4.5524781648289876e-05} +{"train_loss": 0.015429414808750153, "grad_norm": 0.08152227848768234, "global_step": 28012, "epoch": 264, "lr": 4.552182978478488e-05} +{"train_loss": 0.014697040431201458, "grad_norm": 0.07346025109291077, "global_step": 28013, "epoch": 264, "lr": 4.5518877937014226e-05} +{"train_loss": 0.017107343301177025, "grad_norm": 0.08338330686092377, "global_step": 28014, "epoch": 264, "lr": 4.551592610498828e-05} +{"train_loss": 0.013720293529331684, "grad_norm": 0.06142083927989006, "global_step": 28015, "epoch": 264, "lr": 4.551297428871744e-05} +{"train_loss": 0.015399699099361897, "grad_norm": 0.05468602105975151, "global_step": 28016, "epoch": 264, "lr": 4.5510022488212033e-05} +{"train_loss": 0.014090090990066528, "grad_norm": 0.0771699920296669, "global_step": 28017, "epoch": 264, "lr": 4.550707070348247e-05} +{"train_loss": 0.018864203244447708, "grad_norm": 0.07943136990070343, "global_step": 28018, "epoch": 264, "lr": 4.55041189345391e-05} +{"train_loss": 0.015209762379527092, "grad_norm": 0.05107297748327255, "global_step": 28019, "epoch": 264, "lr": 4.550116718139229e-05} +{"train_loss": 0.015113516710698605, "grad_norm": 0.0892801284790039, "global_step": 28020, "epoch": 264, "lr": 4.549821544405245e-05} +{"train_loss": 0.017185138538479805, "grad_norm": 0.076870858669281, "global_step": 28021, "epoch": 264, "lr": 4.54952637225299e-05} +{"train_loss": 0.015030506066977978, "grad_norm": 0.13269908726215363, "global_step": 28022, "epoch": 264, "lr": 4.549231201683504e-05} +{"train_loss": 0.015321075916290283, "grad_norm": 0.11661943048238754, "global_step": 28023, "epoch": 264, "lr": 4.548936032697823e-05} +{"train_loss": 0.01527538150548935, "grad_norm": 0.09702155739068985, "global_step": 28024, "epoch": 264, "lr": 4.5486408652969846e-05} +{"train_loss": 0.017453093081712723, "grad_norm": 0.09398170560598373, "global_step": 28025, "epoch": 264, "lr": 4.5483456994820266e-05} +{"train_loss": 0.015612914226949215, "grad_norm": 0.07988196611404419, "global_step": 28026, "epoch": 264, "lr": 4.548050535253987e-05} +{"train_loss": 0.015869490802288055, "grad_norm": 0.0686248242855072, "global_step": 28027, "epoch": 264, "lr": 4.5477553726138995e-05} +{"train_loss": 0.01641787774860859, "grad_norm": 0.08025467395782471, "global_step": 28028, "epoch": 264, "lr": 4.547460211562802e-05} +{"train_loss": 0.015098247677087784, "grad_norm": 0.0733846053481102, "global_step": 28029, "epoch": 264, "lr": 4.5471650521017335e-05} +{"train_loss": 0.014340321533381939, "grad_norm": 0.09488754719495773, "global_step": 28030, "epoch": 264, "lr": 4.54686989423173e-05} +{"train_loss": 0.01471838727593422, "grad_norm": 0.06327470391988754, "global_step": 28031, "epoch": 264, "lr": 4.546574737953828e-05} +{"train_loss": 0.01531832106411457, "grad_norm": 0.12146075814962387, "global_step": 28032, "epoch": 264, "lr": 4.546279583269067e-05} +{"train_loss": 0.014234865084290504, "grad_norm": 0.09616222232580185, "global_step": 28033, "epoch": 264, "lr": 4.545984430178481e-05} +{"train_loss": 0.013893971219658852, "grad_norm": 0.10447274148464203, "global_step": 28034, "epoch": 264, "lr": 4.545689278683108e-05} +{"train_loss": 0.013714428059756756, "grad_norm": 0.0732535719871521, "global_step": 28035, "epoch": 264, "lr": 4.5453941287839855e-05} +{"train_loss": 0.014916555024683475, "grad_norm": 0.1258813440799713, "global_step": 28036, "epoch": 264, "lr": 4.5450989804821506e-05} +{"train_loss": 0.015853509306907654, "grad_norm": 0.059752363711595535, "global_step": 28037, "epoch": 264, "lr": 4.54480383377864e-05} +{"train_loss": 0.017177198082208633, "grad_norm": 0.08580102771520615, "global_step": 28038, "epoch": 264, "lr": 4.544508688674492e-05} +{"train_loss": 0.015693271532654762, "grad_norm": 0.13254444301128387, "global_step": 28039, "epoch": 264, "lr": 4.544213545170741e-05} +{"train_loss": 0.01617567427456379, "grad_norm": 0.10892295837402344, "global_step": 28040, "epoch": 264, "lr": 4.543918403268424e-05} +{"train_loss": 0.015725117176771164, "grad_norm": 0.1254301369190216, "global_step": 28041, "epoch": 264, "lr": 4.5436232629685816e-05} +{"train_loss": 0.012692499905824661, "grad_norm": 0.062001150101423264, "global_step": 28042, "epoch": 264, "lr": 4.543328124272248e-05} +{"train_loss": 0.014127660542726517, "grad_norm": 0.07404117286205292, "global_step": 28043, "epoch": 264, "lr": 4.543032987180459e-05} +{"train_loss": 0.015392133966088295, "grad_norm": 0.09623989462852478, "global_step": 28044, "epoch": 264, "lr": 4.542737851694257e-05} +{"train_loss": 0.01489191222935915, "grad_norm": 0.10618366301059723, "global_step": 28045, "epoch": 264, "lr": 4.5424427178146735e-05} +{"train_loss": 0.0152592109516263, "grad_norm": 0.13138194382190704, "global_step": 28046, "epoch": 264, "lr": 4.542147585542747e-05} +{"train_loss": 0.016858981922268867, "grad_norm": 0.09398338198661804, "global_step": 28047, "epoch": 264, "lr": 4.541852454879515e-05} +{"train_loss": 0.015660272911190987, "grad_norm": 0.06920400261878967, "global_step": 28048, "epoch": 264, "lr": 4.541557325826013e-05} +{"train_loss": 0.015392913483083248, "grad_norm": 0.1101192831993103, "global_step": 28049, "epoch": 264, "lr": 4.5412621983832804e-05} +{"train_loss": 0.014099281281232834, "grad_norm": 0.07102766633033752, "global_step": 28050, "epoch": 264, "lr": 4.540967072552354e-05} +{"train_loss": 0.018287789076566696, "grad_norm": 0.0956493392586708, "global_step": 28051, "epoch": 264, "lr": 4.540671948334269e-05} +{"train_loss": 0.013918912038207054, "grad_norm": 0.09196069836616516, "global_step": 28052, "epoch": 264, "lr": 4.540376825730061e-05} +{"train_loss": 0.015482032671570778, "grad_norm": 0.06301019340753555, "global_step": 28053, "epoch": 264, "lr": 4.540081704740771e-05} +{"train_loss": 0.017477039247751236, "grad_norm": 0.11066659539937973, "global_step": 28054, "epoch": 264, "lr": 4.5397865853674335e-05} +{"train_loss": 0.014830180443823338, "grad_norm": 0.10311318933963776, "global_step": 28055, "epoch": 264, "lr": 4.5394914676110846e-05} +{"train_loss": 0.016384286805987358, "grad_norm": 0.14324091374874115, "global_step": 28056, "epoch": 264, "lr": 4.539196351472765e-05} +{"train_loss": 0.017525669187307358, "grad_norm": 0.08240954577922821, "global_step": 28057, "epoch": 264, "lr": 4.5389012369535065e-05} +{"train_loss": 0.014024067670106888, "grad_norm": 0.06554743647575378, "global_step": 28058, "epoch": 264, "lr": 4.5386061240543495e-05} +{"train_loss": 0.017233561724424362, "grad_norm": 0.09879231452941895, "global_step": 28059, "epoch": 264, "lr": 4.538311012776331e-05} +{"train_loss": 0.014847625978291035, "grad_norm": 0.057744819670915604, "global_step": 28060, "epoch": 264, "lr": 4.538015903120485e-05} +{"train_loss": 0.014316187240183353, "grad_norm": 0.08938916772603989, "global_step": 28061, "epoch": 264, "lr": 4.5377207950878505e-05} +{"train_loss": 0.013180104084312916, "grad_norm": 0.1079905554652214, "global_step": 28062, "epoch": 264, "lr": 4.5374256886794656e-05} +{"train_loss": 0.015378860756754875, "grad_norm": 0.0851631909608841, "global_step": 28063, "epoch": 264, "lr": 4.537130583896365e-05} +{"train_loss": 0.015193932689726353, "grad_norm": 0.1412680596113205, "global_step": 28064, "epoch": 264, "lr": 4.536835480739585e-05} +{"train_loss": 0.014180876314640045, "grad_norm": 0.06610744446516037, "global_step": 28065, "epoch": 264, "lr": 4.536540379210164e-05} +{"train_loss": 0.014516046270728111, "grad_norm": 0.09502194821834564, "global_step": 28066, "epoch": 264, "lr": 4.5362452793091396e-05} +{"train_loss": 0.013953622430562973, "grad_norm": 0.0695594772696495, "global_step": 28067, "epoch": 264, "lr": 4.535950181037548e-05} +{"train_loss": 0.016473248600959778, "grad_norm": 0.07206729799509048, "global_step": 28068, "epoch": 264, "lr": 4.535655084396425e-05} +{"train_loss": 0.013513205572962761, "grad_norm": 0.07236175984144211, "global_step": 28069, "epoch": 264, "lr": 4.535359989386807e-05} +{"train_loss": 0.015414368361234665, "grad_norm": 0.11157184839248657, "global_step": 28070, "epoch": 264, "lr": 4.535064896009733e-05} +{"train_loss": 0.014977315440773964, "grad_norm": 0.06033441051840782, "global_step": 28071, "epoch": 264, "lr": 4.534769804266239e-05} +{"train_loss": 0.018632620573043823, "grad_norm": 0.09995436668395996, "global_step": 28072, "epoch": 264, "lr": 4.53447471415736e-05} +{"train_loss": 0.015786513686180115, "grad_norm": 0.09080265462398529, "global_step": 28073, "epoch": 264, "lr": 4.534179625684137e-05} +{"train_loss": 0.015575550496578217, "grad_norm": 0.06184704229235649, "global_step": 28074, "epoch": 264, "lr": 4.533884538847602e-05} +{"train_loss": 0.01531482394784689, "grad_norm": 0.16428980231285095, "global_step": 28075, "epoch": 264, "lr": 4.533589453648795e-05} +{"train_loss": 0.016666334122419357, "grad_norm": 0.06313350051641464, "global_step": 28076, "epoch": 264, "lr": 4.5332943700887514e-05} +{"train_loss": 0.014683539047837257, "grad_norm": 0.06017985939979553, "global_step": 28077, "epoch": 264, "lr": 4.532999288168508e-05} +{"train_loss": 0.015006613917648792, "grad_norm": 0.08233364671468735, "global_step": 28078, "epoch": 264, "lr": 4.5327042078891026e-05} +{"train_loss": 0.015498845838010311, "grad_norm": 0.07461003959178925, "global_step": 28079, "epoch": 264, "lr": 4.5324091292515724e-05} +{"train_loss": 0.013643978163599968, "grad_norm": 0.07307589054107666, "global_step": 28080, "epoch": 264, "lr": 4.532114052256953e-05} +{"train_loss": 0.01462609227746725, "grad_norm": 0.09801962971687317, "global_step": 28081, "epoch": 264, "lr": 4.531818976906279e-05} +{"train_loss": 0.014933929778635502, "grad_norm": 0.06344431638717651, "global_step": 28082, "epoch": 264, "lr": 4.5315239032005915e-05} +{"train_loss": 0.01488066278398037, "grad_norm": 0.06869126856327057, "global_step": 28083, "epoch": 264, "lr": 4.531228831140925e-05} +{"train_loss": 0.014235305599868298, "grad_norm": 0.09174655377864838, "global_step": 28084, "epoch": 264, "lr": 4.530933760728315e-05} +{"train_loss": 0.013719378039240837, "grad_norm": 0.10747160762548447, "global_step": 28085, "epoch": 264, "lr": 4.5306386919638035e-05} +{"train_loss": 0.01659044250845909, "grad_norm": 0.08368295431137085, "global_step": 28086, "epoch": 264, "lr": 4.53034362484842e-05} +{"train_loss": 0.015050733461976051, "grad_norm": 0.07001921534538269, "global_step": 28087, "epoch": 264, "lr": 4.530048559383206e-05} +{"train_loss": 0.01702200248837471, "grad_norm": 0.07579416036605835, "global_step": 28088, "epoch": 264, "lr": 4.5297534955691964e-05} +{"train_loss": 0.015482868015203835, "grad_norm": 0.09134124964475632, "global_step": 28089, "epoch": 264, "lr": 4.529458433407429e-05, "val_loss": 0.019769106060266495} +{"train_loss": 0.015612262301146984, "grad_norm": 0.10540466010570526, "global_step": 28090, "epoch": 265, "lr": 4.5291633728989394e-05} +{"train_loss": 0.01681242510676384, "grad_norm": 0.07956637442111969, "global_step": 28091, "epoch": 265, "lr": 4.528868314044767e-05} +{"train_loss": 0.014252059161663055, "grad_norm": 0.07919548451900482, "global_step": 28092, "epoch": 265, "lr": 4.528573256845945e-05} +{"train_loss": 0.014855079352855682, "grad_norm": 0.11188533902168274, "global_step": 28093, "epoch": 265, "lr": 4.52827820130351e-05} +{"train_loss": 0.014775583520531654, "grad_norm": 0.09456335008144379, "global_step": 28094, "epoch": 265, "lr": 4.527983147418503e-05} +{"train_loss": 0.015157968737185001, "grad_norm": 0.15141402184963226, "global_step": 28095, "epoch": 265, "lr": 4.527688095191956e-05} +{"train_loss": 0.015603475272655487, "grad_norm": 0.07345812022686005, "global_step": 28096, "epoch": 265, "lr": 4.527393044624907e-05} +{"train_loss": 0.018392901867628098, "grad_norm": 0.14794793725013733, "global_step": 28097, "epoch": 265, "lr": 4.5270979957183975e-05} +{"train_loss": 0.015930674970149994, "grad_norm": 0.06151358038187027, "global_step": 28098, "epoch": 265, "lr": 4.526802948473457e-05} +{"train_loss": 0.019348978996276855, "grad_norm": 0.09205304086208344, "global_step": 28099, "epoch": 265, "lr": 4.5265079028911244e-05} +{"train_loss": 0.014359132386744022, "grad_norm": 0.07706337422132492, "global_step": 28100, "epoch": 265, "lr": 4.526212858972439e-05} +{"train_loss": 0.01741880178451538, "grad_norm": 0.05986829474568367, "global_step": 28101, "epoch": 265, "lr": 4.525917816718433e-05} +{"train_loss": 0.017043378204107285, "grad_norm": 0.12940561771392822, "global_step": 28102, "epoch": 265, "lr": 4.525622776130147e-05} +{"train_loss": 0.01397663913667202, "grad_norm": 0.16663379967212677, "global_step": 28103, "epoch": 265, "lr": 4.525327737208618e-05} +{"train_loss": 0.014673195779323578, "grad_norm": 0.05925609916448593, "global_step": 28104, "epoch": 265, "lr": 4.52503269995488e-05} +{"train_loss": 0.014567846432328224, "grad_norm": 0.11374933272600174, "global_step": 28105, "epoch": 265, "lr": 4.5247376643699684e-05} +{"train_loss": 0.016011469066143036, "grad_norm": 0.13791880011558533, "global_step": 28106, "epoch": 265, "lr": 4.5244426304549234e-05} +{"train_loss": 0.015691153705120087, "grad_norm": 0.12251519411802292, "global_step": 28107, "epoch": 265, "lr": 4.52414759821078e-05} +{"train_loss": 0.0171198733150959, "grad_norm": 0.08968175202608109, "global_step": 28108, "epoch": 265, "lr": 4.5238525676385745e-05} +{"train_loss": 0.01657014898955822, "grad_norm": 0.1397448480129242, "global_step": 28109, "epoch": 265, "lr": 4.5235575387393476e-05} +{"train_loss": 0.016741178929805756, "grad_norm": 0.062016021460294724, "global_step": 28110, "epoch": 265, "lr": 4.523262511514128e-05} +{"train_loss": 0.015653803944587708, "grad_norm": 0.12543261051177979, "global_step": 28111, "epoch": 265, "lr": 4.522967485963959e-05} +{"train_loss": 0.01625688374042511, "grad_norm": 0.06412749737501144, "global_step": 28112, "epoch": 265, "lr": 4.5226724620898736e-05} +{"train_loss": 0.013713199645280838, "grad_norm": 0.06273171305656433, "global_step": 28113, "epoch": 265, "lr": 4.522377439892909e-05} +{"train_loss": 0.014735384844243526, "grad_norm": 0.07740776985883713, "global_step": 28114, "epoch": 265, "lr": 4.522082419374103e-05} +{"train_loss": 0.01400228962302208, "grad_norm": 0.07510001957416534, "global_step": 28115, "epoch": 265, "lr": 4.521787400534493e-05} +{"train_loss": 0.014707657508552074, "grad_norm": 0.06649399548768997, "global_step": 28116, "epoch": 265, "lr": 4.5214923833751135e-05} +{"train_loss": 0.016504302620887756, "grad_norm": 0.06633973121643066, "global_step": 28117, "epoch": 265, "lr": 4.521197367897001e-05} +{"train_loss": 0.013673683628439903, "grad_norm": 0.08561601489782333, "global_step": 28118, "epoch": 265, "lr": 4.5209023541011914e-05} +{"train_loss": 0.016388189047574997, "grad_norm": 0.08472107350826263, "global_step": 28119, "epoch": 265, "lr": 4.520607341988724e-05} +{"train_loss": 0.014685840345919132, "grad_norm": 0.0681210309267044, "global_step": 28120, "epoch": 265, "lr": 4.520312331560636e-05} +{"train_loss": 0.01934994012117386, "grad_norm": 0.1211354210972786, "global_step": 28121, "epoch": 265, "lr": 4.520017322817959e-05} +{"train_loss": 0.015247507020831108, "grad_norm": 0.0523109957575798, "global_step": 28122, "epoch": 265, "lr": 4.519722315761732e-05} +{"train_loss": 0.01598307117819786, "grad_norm": 0.09488724172115326, "global_step": 28123, "epoch": 265, "lr": 4.5194273103929934e-05} +{"train_loss": 0.017871104180812836, "grad_norm": 0.15584450960159302, "global_step": 28124, "epoch": 265, "lr": 4.519132306712778e-05} +{"train_loss": 0.014276016503572464, "grad_norm": 0.07219307869672775, "global_step": 28125, "epoch": 265, "lr": 4.518837304722121e-05} +{"train_loss": 0.017537660896778107, "grad_norm": 0.0677671954035759, "global_step": 28126, "epoch": 265, "lr": 4.518542304422064e-05} +{"train_loss": 0.016513269394636154, "grad_norm": 0.07268631458282471, "global_step": 28127, "epoch": 265, "lr": 4.518247305813637e-05} +{"train_loss": 0.012311919592320919, "grad_norm": 0.09232702851295471, "global_step": 28128, "epoch": 265, "lr": 4.5179523088978805e-05} +{"train_loss": 0.014969266951084137, "grad_norm": 0.07095231115818024, "global_step": 28129, "epoch": 265, "lr": 4.51765731367583e-05} +{"train_loss": 0.015786906704306602, "grad_norm": 0.056750111281871796, "global_step": 28130, "epoch": 265, "lr": 4.5173623201485205e-05} +{"train_loss": 0.014393860474228859, "grad_norm": 0.08681497722864151, "global_step": 28131, "epoch": 265, "lr": 4.5170673283169915e-05} +{"train_loss": 0.014964355155825615, "grad_norm": 0.10296648740768433, "global_step": 28132, "epoch": 265, "lr": 4.516772338182278e-05} +{"train_loss": 0.014686526730656624, "grad_norm": 0.08002357929944992, "global_step": 28133, "epoch": 265, "lr": 4.516477349745417e-05} +{"train_loss": 0.016357548534870148, "grad_norm": 0.08058489859104156, "global_step": 28134, "epoch": 265, "lr": 4.516182363007442e-05} +{"train_loss": 0.01430498342961073, "grad_norm": 0.0794747844338417, "global_step": 28135, "epoch": 265, "lr": 4.515887377969393e-05} +{"train_loss": 0.01776876300573349, "grad_norm": 0.06851644814014435, "global_step": 28136, "epoch": 265, "lr": 4.5155923946323046e-05} +{"train_loss": 0.015432070009410381, "grad_norm": 0.06553179025650024, "global_step": 28137, "epoch": 265, "lr": 4.515297412997214e-05} +{"train_loss": 0.01501300185918808, "grad_norm": 0.060973551124334335, "global_step": 28138, "epoch": 265, "lr": 4.5150024330651595e-05} +{"train_loss": 0.014688679948449135, "grad_norm": 0.08120465278625488, "global_step": 28139, "epoch": 265, "lr": 4.5147074548371735e-05} +{"train_loss": 0.015109138563275337, "grad_norm": 0.08375521749258041, "global_step": 28140, "epoch": 265, "lr": 4.514412478314295e-05} +{"train_loss": 0.014330752193927765, "grad_norm": 0.05370510369539261, "global_step": 28141, "epoch": 265, "lr": 4.5141175034975595e-05} +{"train_loss": 0.013508202508091927, "grad_norm": 0.06883262097835541, "global_step": 28142, "epoch": 265, "lr": 4.513822530388003e-05} +{"train_loss": 0.01565372198820114, "grad_norm": 0.06900683045387268, "global_step": 28143, "epoch": 265, "lr": 4.5135275589866636e-05} +{"train_loss": 0.017608830705285072, "grad_norm": 0.0577491857111454, "global_step": 28144, "epoch": 265, "lr": 4.513232589294578e-05} +{"train_loss": 0.016508691012859344, "grad_norm": 0.09746897220611572, "global_step": 28145, "epoch": 265, "lr": 4.512937621312781e-05} +{"train_loss": 0.013171656057238579, "grad_norm": 0.07729437202215195, "global_step": 28146, "epoch": 265, "lr": 4.512642655042308e-05} +{"train_loss": 0.014101915061473846, "grad_norm": 0.10415159910917282, "global_step": 28147, "epoch": 265, "lr": 4.5123476904841974e-05} +{"train_loss": 0.01609005033969879, "grad_norm": 0.12361811846494675, "global_step": 28148, "epoch": 265, "lr": 4.512052727639485e-05} +{"train_loss": 0.015411019325256348, "grad_norm": 0.15520261228084564, "global_step": 28149, "epoch": 265, "lr": 4.511757766509207e-05} +{"train_loss": 0.01494970079511404, "grad_norm": 0.15259680151939392, "global_step": 28150, "epoch": 265, "lr": 4.511462807094402e-05} +{"train_loss": 0.01421712338924408, "grad_norm": 0.12071486562490463, "global_step": 28151, "epoch": 265, "lr": 4.511167849396101e-05} +{"train_loss": 0.0138634592294693, "grad_norm": 0.08337505906820297, "global_step": 28152, "epoch": 265, "lr": 4.5108728934153444e-05} +{"train_loss": 0.01746945083141327, "grad_norm": 0.0665067583322525, "global_step": 28153, "epoch": 265, "lr": 4.510577939153168e-05} +{"train_loss": 0.01581171154975891, "grad_norm": 0.15491145849227905, "global_step": 28154, "epoch": 265, "lr": 4.510282986610607e-05} +{"train_loss": 0.016207944601774216, "grad_norm": 0.0916457548737526, "global_step": 28155, "epoch": 265, "lr": 4.5099880357887e-05} +{"train_loss": 0.016804318875074387, "grad_norm": 0.09389253705739975, "global_step": 28156, "epoch": 265, "lr": 4.509693086688483e-05} +{"train_loss": 0.013739073649048805, "grad_norm": 0.11229845136404037, "global_step": 28157, "epoch": 265, "lr": 4.5093981393109895e-05} +{"train_loss": 0.016806740313768387, "grad_norm": 0.14562514424324036, "global_step": 28158, "epoch": 265, "lr": 4.509103193657259e-05} +{"train_loss": 0.015380282886326313, "grad_norm": 0.0780845358967781, "global_step": 28159, "epoch": 265, "lr": 4.5088082497283235e-05} +{"train_loss": 0.016318727284669876, "grad_norm": 0.07484686374664307, "global_step": 28160, "epoch": 265, "lr": 4.508513307525225e-05} +{"train_loss": 0.015942972153425217, "grad_norm": 0.14711640775203705, "global_step": 28161, "epoch": 265, "lr": 4.508218367048995e-05} +{"train_loss": 0.013297284953296185, "grad_norm": 0.0606653168797493, "global_step": 28162, "epoch": 265, "lr": 4.5079234283006745e-05} +{"train_loss": 0.015736816450953484, "grad_norm": 0.08865936845541, "global_step": 28163, "epoch": 265, "lr": 4.507628491281295e-05} +{"train_loss": 0.015240127220749855, "grad_norm": 0.07931050658226013, "global_step": 28164, "epoch": 265, "lr": 4.507333555991895e-05} +{"train_loss": 0.013568347319960594, "grad_norm": 0.06805858016014099, "global_step": 28165, "epoch": 265, "lr": 4.507038622433511e-05} +{"train_loss": 0.015538375824689865, "grad_norm": 0.07487878948450089, "global_step": 28166, "epoch": 265, "lr": 4.5067436906071774e-05} +{"train_loss": 0.015626979991793633, "grad_norm": 0.06776686757802963, "global_step": 28167, "epoch": 265, "lr": 4.506448760513933e-05} +{"train_loss": 0.012962760403752327, "grad_norm": 0.12228602170944214, "global_step": 28168, "epoch": 265, "lr": 4.506153832154815e-05} +{"train_loss": 0.015742411836981773, "grad_norm": 0.07240018993616104, "global_step": 28169, "epoch": 265, "lr": 4.505858905530857e-05} +{"train_loss": 0.017124999314546585, "grad_norm": 0.08433626592159271, "global_step": 28170, "epoch": 265, "lr": 4.505563980643094e-05} +{"train_loss": 0.01705649122595787, "grad_norm": 0.1225099191069603, "global_step": 28171, "epoch": 265, "lr": 4.505269057492564e-05} +{"train_loss": 0.013959186151623726, "grad_norm": 0.07431799173355103, "global_step": 28172, "epoch": 265, "lr": 4.504974136080305e-05} +{"train_loss": 0.016041111201047897, "grad_norm": 0.10000484436750412, "global_step": 28173, "epoch": 265, "lr": 4.504679216407352e-05} +{"train_loss": 0.01757221668958664, "grad_norm": 0.07484287023544312, "global_step": 28174, "epoch": 265, "lr": 4.50438429847474e-05} +{"train_loss": 0.013556313700973988, "grad_norm": 0.1512155532836914, "global_step": 28175, "epoch": 265, "lr": 4.5040893822835064e-05} +{"train_loss": 0.015393292531371117, "grad_norm": 0.08007940649986267, "global_step": 28176, "epoch": 265, "lr": 4.503794467834685e-05} +{"train_loss": 0.013984577730298042, "grad_norm": 0.1154414564371109, "global_step": 28177, "epoch": 265, "lr": 4.503499555129316e-05} +{"train_loss": 0.01600910723209381, "grad_norm": 0.06683448702096939, "global_step": 28178, "epoch": 265, "lr": 4.503204644168433e-05} +{"train_loss": 0.016080209985375404, "grad_norm": 0.06510180979967117, "global_step": 28179, "epoch": 265, "lr": 4.5029097349530745e-05} +{"train_loss": 0.016431298106908798, "grad_norm": 0.0851062685251236, "global_step": 28180, "epoch": 265, "lr": 4.502614827484273e-05} +{"train_loss": 0.015506254509091377, "grad_norm": 0.14519567787647247, "global_step": 28181, "epoch": 265, "lr": 4.5023199217630666e-05} +{"train_loss": 0.015163077041506767, "grad_norm": 0.13073699176311493, "global_step": 28182, "epoch": 265, "lr": 4.502025017790493e-05} +{"train_loss": 0.014479885809123516, "grad_norm": 0.11149861663579941, "global_step": 28183, "epoch": 265, "lr": 4.501730115567585e-05} +{"train_loss": 0.013775749132037163, "grad_norm": 0.07303782552480698, "global_step": 28184, "epoch": 265, "lr": 4.5014352150953823e-05} +{"train_loss": 0.019679557532072067, "grad_norm": 0.08134250342845917, "global_step": 28185, "epoch": 265, "lr": 4.50114031637492e-05} +{"train_loss": 0.015318484976887703, "grad_norm": 0.06330157071352005, "global_step": 28186, "epoch": 265, "lr": 4.5008454194072326e-05} +{"train_loss": 0.01563219353556633, "grad_norm": 0.07631642371416092, "global_step": 28187, "epoch": 265, "lr": 4.500550524193357e-05} +{"train_loss": 0.015548440627753735, "grad_norm": 0.09234649688005447, "global_step": 28188, "epoch": 265, "lr": 4.500255630734329e-05} +{"train_loss": 0.015877939760684967, "grad_norm": 0.10476658493280411, "global_step": 28189, "epoch": 265, "lr": 4.499960739031187e-05} +{"train_loss": 0.015845119953155518, "grad_norm": 0.09724965691566467, "global_step": 28190, "epoch": 265, "lr": 4.4996658490849644e-05} +{"train_loss": 0.013882039114832878, "grad_norm": 0.08228226751089096, "global_step": 28191, "epoch": 265, "lr": 4.4993709608967006e-05} +{"train_loss": 0.013098346069455147, "grad_norm": 0.15275472402572632, "global_step": 28192, "epoch": 265, "lr": 4.499076074467427e-05} +{"train_loss": 0.016410870477557182, "grad_norm": 0.1053369864821434, "global_step": 28193, "epoch": 265, "lr": 4.498781189798183e-05} +{"train_loss": 0.012668222188949585, "grad_norm": 0.09521663188934326, "global_step": 28194, "epoch": 265, "lr": 4.498486306890004e-05} +{"train_loss": 0.015492768070818681, "grad_norm": 0.07144469767808914, "global_step": 28195, "epoch": 265, "lr": 4.498191425743925e-05, "val_loss": 0.01954566314816475} +{"train_loss": 0.017347371205687523, "grad_norm": 0.06438728421926498, "global_step": 28196, "epoch": 266, "lr": 4.4978965463609845e-05} +{"train_loss": 0.013331957161426544, "grad_norm": 0.12111533433198929, "global_step": 28197, "epoch": 266, "lr": 4.497601668742218e-05} +{"train_loss": 0.014347686432301998, "grad_norm": 0.07259324938058853, "global_step": 28198, "epoch": 266, "lr": 4.49730679288866e-05} +{"train_loss": 0.013673045672476292, "grad_norm": 0.07056672126054764, "global_step": 28199, "epoch": 266, "lr": 4.497011918801347e-05} +{"train_loss": 0.016859013587236404, "grad_norm": 0.08215384185314178, "global_step": 28200, "epoch": 266, "lr": 4.4967170464813135e-05} +{"train_loss": 0.014874991960823536, "grad_norm": 0.07434935867786407, "global_step": 28201, "epoch": 266, "lr": 4.4964221759296e-05} +{"train_loss": 0.016889559105038643, "grad_norm": 0.07295850664377213, "global_step": 28202, "epoch": 266, "lr": 4.496127307147239e-05} +{"train_loss": 0.01615581475198269, "grad_norm": 0.0711652934551239, "global_step": 28203, "epoch": 266, "lr": 4.495832440135268e-05} +{"train_loss": 0.014050193130970001, "grad_norm": 0.07289097458124161, "global_step": 28204, "epoch": 266, "lr": 4.495537574894723e-05} +{"train_loss": 0.017985772341489792, "grad_norm": 0.07283981144428253, "global_step": 28205, "epoch": 266, "lr": 4.495242711426638e-05} +{"train_loss": 0.01576080173254013, "grad_norm": 0.06372298300266266, "global_step": 28206, "epoch": 266, "lr": 4.4949478497320515e-05} +{"train_loss": 0.016595451161265373, "grad_norm": 0.15756404399871826, "global_step": 28207, "epoch": 266, "lr": 4.494652989811997e-05} +{"train_loss": 0.015636030584573746, "grad_norm": 0.11797741055488586, "global_step": 28208, "epoch": 266, "lr": 4.4943581316675135e-05} +{"train_loss": 0.013532291166484356, "grad_norm": 0.06275651603937149, "global_step": 28209, "epoch": 266, "lr": 4.494063275299637e-05} +{"train_loss": 0.01576634868979454, "grad_norm": 0.09774945676326752, "global_step": 28210, "epoch": 266, "lr": 4.493768420709401e-05} +{"train_loss": 0.01734185963869095, "grad_norm": 0.12531796097755432, "global_step": 28211, "epoch": 266, "lr": 4.493473567897843e-05} +{"train_loss": 0.016657082363963127, "grad_norm": 0.0957568883895874, "global_step": 28212, "epoch": 266, "lr": 4.493178716865997e-05} +{"train_loss": 0.015117522329092026, "grad_norm": 0.0796465203166008, "global_step": 28213, "epoch": 266, "lr": 4.492883867614902e-05} +{"train_loss": 0.014988252893090248, "grad_norm": 0.08884904533624649, "global_step": 28214, "epoch": 266, "lr": 4.492589020145592e-05} +{"train_loss": 0.014159595593810081, "grad_norm": 0.06748697906732559, "global_step": 28215, "epoch": 266, "lr": 4.492294174459105e-05} +{"train_loss": 0.015973364934325218, "grad_norm": 0.06538478285074234, "global_step": 28216, "epoch": 266, "lr": 4.491999330556475e-05} +{"train_loss": 0.016296207904815674, "grad_norm": 0.06290334463119507, "global_step": 28217, "epoch": 266, "lr": 4.491704488438736e-05} +{"train_loss": 0.015009690076112747, "grad_norm": 0.05437954515218735, "global_step": 28218, "epoch": 266, "lr": 4.491409648106928e-05} +{"train_loss": 0.014878090471029282, "grad_norm": 0.0899929627776146, "global_step": 28219, "epoch": 266, "lr": 4.4911148095620854e-05} +{"train_loss": 0.015111605636775494, "grad_norm": 0.06296233087778091, "global_step": 28220, "epoch": 266, "lr": 4.4908199728052454e-05} +{"train_loss": 0.015092089772224426, "grad_norm": 0.14070555567741394, "global_step": 28221, "epoch": 266, "lr": 4.49052513783744e-05} +{"train_loss": 0.01641324907541275, "grad_norm": 0.05619725584983826, "global_step": 28222, "epoch": 266, "lr": 4.4902303046597085e-05} +{"train_loss": 0.01179422251880169, "grad_norm": 0.08276049047708511, "global_step": 28223, "epoch": 266, "lr": 4.489935473273087e-05} +{"train_loss": 0.013735760934650898, "grad_norm": 0.12252435088157654, "global_step": 28224, "epoch": 266, "lr": 4.489640643678609e-05} +{"train_loss": 0.017967993393540382, "grad_norm": 0.0918658971786499, "global_step": 28225, "epoch": 266, "lr": 4.489345815877312e-05} +{"train_loss": 0.015887927263975143, "grad_norm": 0.08578674495220184, "global_step": 28226, "epoch": 266, "lr": 4.4890509898702334e-05} +{"train_loss": 0.016359323635697365, "grad_norm": 0.09856196492910385, "global_step": 28227, "epoch": 266, "lr": 4.488756165658407e-05} +{"train_loss": 0.014220613986253738, "grad_norm": 0.06788921356201172, "global_step": 28228, "epoch": 266, "lr": 4.4884613432428686e-05} +{"train_loss": 0.014100153930485249, "grad_norm": 0.0641094222664833, "global_step": 28229, "epoch": 266, "lr": 4.488166522624653e-05} +{"train_loss": 0.01682981476187706, "grad_norm": 0.09057337790727615, "global_step": 28230, "epoch": 266, "lr": 4.4878717038047996e-05} +{"train_loss": 0.013696474954485893, "grad_norm": 0.10591244697570801, "global_step": 28231, "epoch": 266, "lr": 4.4875768867843416e-05} +{"train_loss": 0.01587880775332451, "grad_norm": 0.11753536760807037, "global_step": 28232, "epoch": 266, "lr": 4.487282071564317e-05} +{"train_loss": 0.013854147866368294, "grad_norm": 0.12597504258155823, "global_step": 28233, "epoch": 266, "lr": 4.48698725814576e-05} +{"train_loss": 0.016869019716978073, "grad_norm": 0.057426828891038895, "global_step": 28234, "epoch": 266, "lr": 4.486692446529704e-05} +{"train_loss": 0.013268261216580868, "grad_norm": 0.05660372972488403, "global_step": 28235, "epoch": 266, "lr": 4.4863976367171886e-05} +{"train_loss": 0.01496114768087864, "grad_norm": 0.057498179376125336, "global_step": 28236, "epoch": 266, "lr": 4.4861028287092486e-05} +{"train_loss": 0.01617780514061451, "grad_norm": 0.0639643743634224, "global_step": 28237, "epoch": 266, "lr": 4.4858080225069206e-05} +{"train_loss": 0.01592855155467987, "grad_norm": 0.054707806557416916, "global_step": 28238, "epoch": 266, "lr": 4.48551321811124e-05} +{"train_loss": 0.015536597929894924, "grad_norm": 0.07689452916383743, "global_step": 28239, "epoch": 266, "lr": 4.485218415523242e-05} +{"train_loss": 0.01580001786351204, "grad_norm": 0.11793532222509384, "global_step": 28240, "epoch": 266, "lr": 4.484923614743962e-05} +{"train_loss": 0.014060061424970627, "grad_norm": 0.12095087021589279, "global_step": 28241, "epoch": 266, "lr": 4.484628815774436e-05} +{"train_loss": 0.01634855754673481, "grad_norm": 0.11777323484420776, "global_step": 28242, "epoch": 266, "lr": 4.4843340186157006e-05} +{"train_loss": 0.015132229775190353, "grad_norm": 0.1080075055360794, "global_step": 28243, "epoch": 266, "lr": 4.4840392232687916e-05} +{"train_loss": 0.014042158611118793, "grad_norm": 0.05376375839114189, "global_step": 28244, "epoch": 266, "lr": 4.483744429734745e-05} +{"train_loss": 0.015126386657357216, "grad_norm": 0.07625412940979004, "global_step": 28245, "epoch": 266, "lr": 4.483449638014595e-05} +{"train_loss": 0.018307968974113464, "grad_norm": 0.12217259407043457, "global_step": 28246, "epoch": 266, "lr": 4.483154848109378e-05} +{"train_loss": 0.01476212963461876, "grad_norm": 0.07728638499975204, "global_step": 28247, "epoch": 266, "lr": 4.4828600600201305e-05} +{"train_loss": 0.013272834941744804, "grad_norm": 0.06936250627040863, "global_step": 28248, "epoch": 266, "lr": 4.482565273747888e-05} +{"train_loss": 0.0164981409907341, "grad_norm": 0.1470872312784195, "global_step": 28249, "epoch": 266, "lr": 4.482270489293685e-05} +{"train_loss": 0.014876443892717361, "grad_norm": 0.12392349541187286, "global_step": 28250, "epoch": 266, "lr": 4.48197570665856e-05} +{"train_loss": 0.015733763575553894, "grad_norm": 0.06855107843875885, "global_step": 28251, "epoch": 266, "lr": 4.481680925843547e-05} +{"train_loss": 0.017522234469652176, "grad_norm": 0.09270576387643814, "global_step": 28252, "epoch": 266, "lr": 4.481386146849681e-05} +{"train_loss": 0.015878232195973396, "grad_norm": 0.09007063508033752, "global_step": 28253, "epoch": 266, "lr": 4.481091369677998e-05} +{"train_loss": 0.017349623143672943, "grad_norm": 0.08178729563951492, "global_step": 28254, "epoch": 266, "lr": 4.480796594329535e-05} +{"train_loss": 0.01404192578047514, "grad_norm": 0.0630013719201088, "global_step": 28255, "epoch": 266, "lr": 4.480501820805327e-05} +{"train_loss": 0.016553735360503197, "grad_norm": 0.18095077574253082, "global_step": 28256, "epoch": 266, "lr": 4.4802070491064106e-05} +{"train_loss": 0.01708933338522911, "grad_norm": 0.09826966375112534, "global_step": 28257, "epoch": 266, "lr": 4.4799122792338196e-05} +{"train_loss": 0.01965894177556038, "grad_norm": 0.1251436322927475, "global_step": 28258, "epoch": 266, "lr": 4.4796175111885905e-05} +{"train_loss": 0.012693457305431366, "grad_norm": 0.0983845666050911, "global_step": 28259, "epoch": 266, "lr": 4.479322744971759e-05} +{"train_loss": 0.01621513068675995, "grad_norm": 0.07466795295476913, "global_step": 28260, "epoch": 266, "lr": 4.479027980584362e-05} +{"train_loss": 0.018627703189849854, "grad_norm": 0.06544651836156845, "global_step": 28261, "epoch": 266, "lr": 4.478733218027432e-05} +{"train_loss": 0.01451040618121624, "grad_norm": 0.10416660457849503, "global_step": 28262, "epoch": 266, "lr": 4.478438457302009e-05} +{"train_loss": 0.012881268747150898, "grad_norm": 0.057229023426771164, "global_step": 28263, "epoch": 266, "lr": 4.4781436984091263e-05} +{"train_loss": 0.016411634162068367, "grad_norm": 0.10190615057945251, "global_step": 28264, "epoch": 266, "lr": 4.477848941349819e-05} +{"train_loss": 0.016058098524808884, "grad_norm": 0.04897778108716011, "global_step": 28265, "epoch": 266, "lr": 4.4775541861251226e-05} +{"train_loss": 0.013719933107495308, "grad_norm": 0.06252043694257736, "global_step": 28266, "epoch": 266, "lr": 4.477259432736075e-05} +{"train_loss": 0.016571205109357834, "grad_norm": 0.06519898027181625, "global_step": 28267, "epoch": 266, "lr": 4.4769646811837093e-05} +{"train_loss": 0.01767270639538765, "grad_norm": 0.17661207914352417, "global_step": 28268, "epoch": 266, "lr": 4.476669931469064e-05} +{"train_loss": 0.014959970489144325, "grad_norm": 0.10507222265005112, "global_step": 28269, "epoch": 266, "lr": 4.476375183593172e-05} +{"train_loss": 0.013981945812702179, "grad_norm": 0.09261909127235413, "global_step": 28270, "epoch": 266, "lr": 4.4760804375570686e-05} +{"train_loss": 0.014684703201055527, "grad_norm": 0.0885407030582428, "global_step": 28271, "epoch": 266, "lr": 4.475785693361792e-05} +{"train_loss": 0.01388732623308897, "grad_norm": 0.07072605937719345, "global_step": 28272, "epoch": 266, "lr": 4.475490951008376e-05} +{"train_loss": 0.01692640781402588, "grad_norm": 0.059520166367292404, "global_step": 28273, "epoch": 266, "lr": 4.4751962104978584e-05} +{"train_loss": 0.01620231382548809, "grad_norm": 0.08027102798223495, "global_step": 28274, "epoch": 266, "lr": 4.474901471831272e-05} +{"train_loss": 0.013760892674326897, "grad_norm": 0.07240394502878189, "global_step": 28275, "epoch": 266, "lr": 4.474606735009652e-05} +{"train_loss": 0.014044368639588356, "grad_norm": 0.09517867863178253, "global_step": 28276, "epoch": 266, "lr": 4.474312000034037e-05} +{"train_loss": 0.014355341903865337, "grad_norm": 0.06552550941705704, "global_step": 28277, "epoch": 266, "lr": 4.47401726690546e-05} +{"train_loss": 0.014757655560970306, "grad_norm": 0.10277207940816879, "global_step": 28278, "epoch": 266, "lr": 4.473722535624958e-05} +{"train_loss": 0.014468028210103512, "grad_norm": 0.13282683491706848, "global_step": 28279, "epoch": 266, "lr": 4.473427806193567e-05} +{"train_loss": 0.015541797503829002, "grad_norm": 0.06262603402137756, "global_step": 28280, "epoch": 266, "lr": 4.473133078612321e-05} +{"train_loss": 0.015996426343917847, "grad_norm": 0.11760276556015015, "global_step": 28281, "epoch": 266, "lr": 4.472838352882257e-05} +{"train_loss": 0.015526385977864265, "grad_norm": 0.12593454122543335, "global_step": 28282, "epoch": 266, "lr": 4.472543629004407e-05} +{"train_loss": 0.014866793528199196, "grad_norm": 0.13793160021305084, "global_step": 28283, "epoch": 266, "lr": 4.472248906979812e-05} +{"train_loss": 0.015652548521757126, "grad_norm": 0.08594000339508057, "global_step": 28284, "epoch": 266, "lr": 4.471954186809504e-05} +{"train_loss": 0.01467984914779663, "grad_norm": 0.05813455581665039, "global_step": 28285, "epoch": 266, "lr": 4.471659468494521e-05} +{"train_loss": 0.01833554357290268, "grad_norm": 0.10469448566436768, "global_step": 28286, "epoch": 266, "lr": 4.471364752035895e-05} +{"train_loss": 0.015203086659312248, "grad_norm": 0.058215126395225525, "global_step": 28287, "epoch": 266, "lr": 4.471070037434663e-05} +{"train_loss": 0.01589667610824108, "grad_norm": 0.08477196097373962, "global_step": 28288, "epoch": 266, "lr": 4.470775324691861e-05} +{"train_loss": 0.014709364622831345, "grad_norm": 0.05440893396735191, "global_step": 28289, "epoch": 266, "lr": 4.470480613808526e-05} +{"train_loss": 0.014233248308300972, "grad_norm": 0.06157815828919411, "global_step": 28290, "epoch": 266, "lr": 4.4701859047856895e-05} +{"train_loss": 0.016818197444081306, "grad_norm": 0.10190579295158386, "global_step": 28291, "epoch": 266, "lr": 4.469891197624392e-05} +{"train_loss": 0.01599578931927681, "grad_norm": 0.06661026924848557, "global_step": 28292, "epoch": 266, "lr": 4.469596492325665e-05} +{"train_loss": 0.013799807988107204, "grad_norm": 0.10343397408723831, "global_step": 28293, "epoch": 266, "lr": 4.4693017888905445e-05} +{"train_loss": 0.015560904517769814, "grad_norm": 0.061952315270900726, "global_step": 28294, "epoch": 266, "lr": 4.469007087320067e-05} +{"train_loss": 0.01810063049197197, "grad_norm": 0.06399679183959961, "global_step": 28295, "epoch": 266, "lr": 4.4687123876152684e-05} +{"train_loss": 0.017114391550421715, "grad_norm": 0.11270082741975784, "global_step": 28296, "epoch": 266, "lr": 4.468417689777183e-05} +{"train_loss": 0.015931248664855957, "grad_norm": 0.16521945595741272, "global_step": 28297, "epoch": 266, "lr": 4.468122993806848e-05} +{"train_loss": 0.015755990520119667, "grad_norm": 0.1035136729478836, "global_step": 28298, "epoch": 266, "lr": 4.467828299705296e-05} +{"train_loss": 0.014282699674367905, "grad_norm": 0.11401496082544327, "global_step": 28299, "epoch": 266, "lr": 4.467533607473563e-05} +{"train_loss": 0.015477614477276802, "grad_norm": 0.07394903898239136, "global_step": 28300, "epoch": 266, "lr": 4.4672389171126865e-05} +{"train_loss": 0.015454736283435574, "grad_norm": 0.07472541183233261, "global_step": 28301, "epoch": 266, "lr": 4.466944228623701e-05, "val_loss": 0.01930285431444645} +{"train_loss": 0.013885591179132462, "grad_norm": 0.09915804862976074, "global_step": 28302, "epoch": 267, "lr": 4.4666495420076405e-05} +{"train_loss": 0.012804843485355377, "grad_norm": 0.0796368420124054, "global_step": 28303, "epoch": 267, "lr": 4.466354857265544e-05} +{"train_loss": 0.014955997467041016, "grad_norm": 0.047095198184251785, "global_step": 28304, "epoch": 267, "lr": 4.466060174398442e-05} +{"train_loss": 0.015060242265462875, "grad_norm": 0.07321594655513763, "global_step": 28305, "epoch": 267, "lr": 4.465765493407374e-05} +{"train_loss": 0.016358794644474983, "grad_norm": 0.11568333208560944, "global_step": 28306, "epoch": 267, "lr": 4.4654708142933726e-05} +{"train_loss": 0.017110511660575867, "grad_norm": 0.0849088579416275, "global_step": 28307, "epoch": 267, "lr": 4.465176137057474e-05} +{"train_loss": 0.013929672539234161, "grad_norm": 0.05726700648665428, "global_step": 28308, "epoch": 267, "lr": 4.464881461700715e-05} +{"train_loss": 0.012993166223168373, "grad_norm": 0.058389898389577866, "global_step": 28309, "epoch": 267, "lr": 4.46458678822413e-05} +{"train_loss": 0.018044978380203247, "grad_norm": 0.09860721230506897, "global_step": 28310, "epoch": 267, "lr": 4.464292116628754e-05} +{"train_loss": 0.01680520549416542, "grad_norm": 0.07173418253660202, "global_step": 28311, "epoch": 267, "lr": 4.463997446915621e-05} +{"train_loss": 0.01628078520298004, "grad_norm": 0.0616387240588665, "global_step": 28312, "epoch": 267, "lr": 4.463702779085769e-05} +{"train_loss": 0.01421978510916233, "grad_norm": 0.06029301881790161, "global_step": 28313, "epoch": 267, "lr": 4.4634081131402316e-05} +{"train_loss": 0.01670263707637787, "grad_norm": 0.06960390508174896, "global_step": 28314, "epoch": 267, "lr": 4.463113449080045e-05} +{"train_loss": 0.014546968042850494, "grad_norm": 0.11990205198526382, "global_step": 28315, "epoch": 267, "lr": 4.4628187869062463e-05} +{"train_loss": 0.0136313047260046, "grad_norm": 0.10755201429128647, "global_step": 28316, "epoch": 267, "lr": 4.462524126619866e-05} +{"train_loss": 0.014498339965939522, "grad_norm": 0.07551157474517822, "global_step": 28317, "epoch": 267, "lr": 4.4622294682219436e-05} +{"train_loss": 0.016630591824650764, "grad_norm": 0.11032135784626007, "global_step": 28318, "epoch": 267, "lr": 4.461934811713512e-05} +{"train_loss": 0.017110371962189674, "grad_norm": 0.07316609472036362, "global_step": 28319, "epoch": 267, "lr": 4.4616401570956075e-05} +{"train_loss": 0.01501311082392931, "grad_norm": 0.08342886716127396, "global_step": 28320, "epoch": 267, "lr": 4.461345504369267e-05} +{"train_loss": 0.014899974688887596, "grad_norm": 0.13468952476978302, "global_step": 28321, "epoch": 267, "lr": 4.461050853535523e-05} +{"train_loss": 0.015043901279568672, "grad_norm": 0.07847434282302856, "global_step": 28322, "epoch": 267, "lr": 4.460756204595412e-05} +{"train_loss": 0.014525175094604492, "grad_norm": 0.06431082636117935, "global_step": 28323, "epoch": 267, "lr": 4.460461557549968e-05} +{"train_loss": 0.014094633981585503, "grad_norm": 0.07057454437017441, "global_step": 28324, "epoch": 267, "lr": 4.460166912400229e-05} +{"train_loss": 0.016243668273091316, "grad_norm": 0.06460206210613251, "global_step": 28325, "epoch": 267, "lr": 4.459872269147228e-05} +{"train_loss": 0.01452336274087429, "grad_norm": 0.1697603017091751, "global_step": 28326, "epoch": 267, "lr": 4.4595776277920026e-05} +{"train_loss": 0.014484548941254616, "grad_norm": 0.05470254644751549, "global_step": 28327, "epoch": 267, "lr": 4.459282988335585e-05} +{"train_loss": 0.015670882537961006, "grad_norm": 0.06252120435237885, "global_step": 28328, "epoch": 267, "lr": 4.458988350779011e-05} +{"train_loss": 0.01569315791130066, "grad_norm": 0.0816805437207222, "global_step": 28329, "epoch": 267, "lr": 4.458693715123318e-05} +{"train_loss": 0.013455374166369438, "grad_norm": 0.09523226320743561, "global_step": 28330, "epoch": 267, "lr": 4.4583990813695395e-05} +{"train_loss": 0.016721472144126892, "grad_norm": 0.11776384711265564, "global_step": 28331, "epoch": 267, "lr": 4.4581044495187095e-05} +{"train_loss": 0.01602128893136978, "grad_norm": 0.07472991943359375, "global_step": 28332, "epoch": 267, "lr": 4.4578098195718686e-05} +{"train_loss": 0.0159623846411705, "grad_norm": 0.12961934506893158, "global_step": 28333, "epoch": 267, "lr": 4.457515191530045e-05} +{"train_loss": 0.01643683761358261, "grad_norm": 0.062335532158613205, "global_step": 28334, "epoch": 267, "lr": 4.4572205653942775e-05} +{"train_loss": 0.0134342722594738, "grad_norm": 0.08132406324148178, "global_step": 28335, "epoch": 267, "lr": 4.4569259411656015e-05} +{"train_loss": 0.01600043661892414, "grad_norm": 0.06940510123968124, "global_step": 28336, "epoch": 267, "lr": 4.456631318845051e-05} +{"train_loss": 0.01456766203045845, "grad_norm": 0.10599429160356522, "global_step": 28337, "epoch": 267, "lr": 4.4563366984336617e-05} +{"train_loss": 0.017142802476882935, "grad_norm": 0.0812763124704361, "global_step": 28338, "epoch": 267, "lr": 4.456042079932471e-05} +{"train_loss": 0.014989442192018032, "grad_norm": 0.08156085759401321, "global_step": 28339, "epoch": 267, "lr": 4.45574746334251e-05} +{"train_loss": 0.014131907373666763, "grad_norm": 0.06717371940612793, "global_step": 28340, "epoch": 267, "lr": 4.455452848664815e-05} +{"train_loss": 0.015437447465956211, "grad_norm": 0.07832439988851547, "global_step": 28341, "epoch": 267, "lr": 4.455158235900423e-05} +{"train_loss": 0.014784907922148705, "grad_norm": 0.11681731045246124, "global_step": 28342, "epoch": 267, "lr": 4.454863625050368e-05} +{"train_loss": 0.015030036680400372, "grad_norm": 0.07091207802295685, "global_step": 28343, "epoch": 267, "lr": 4.4545690161156836e-05} +{"train_loss": 0.014977598562836647, "grad_norm": 0.06454429775476456, "global_step": 28344, "epoch": 267, "lr": 4.45427440909741e-05} +{"train_loss": 0.017815817147493362, "grad_norm": 0.08165426552295685, "global_step": 28345, "epoch": 267, "lr": 4.453979803996576e-05} +{"train_loss": 0.01920866221189499, "grad_norm": 0.06724045425653458, "global_step": 28346, "epoch": 267, "lr": 4.45368520081422e-05} +{"train_loss": 0.014502977021038532, "grad_norm": 0.0687137320637703, "global_step": 28347, "epoch": 267, "lr": 4.453390599551376e-05} +{"train_loss": 0.014778711833059788, "grad_norm": 0.0570816695690155, "global_step": 28348, "epoch": 267, "lr": 4.45309600020908e-05} +{"train_loss": 0.013129092752933502, "grad_norm": 0.05943099781870842, "global_step": 28349, "epoch": 267, "lr": 4.452801402788368e-05} +{"train_loss": 0.014100783504545689, "grad_norm": 0.07825350016355515, "global_step": 28350, "epoch": 267, "lr": 4.4525068072902734e-05} +{"train_loss": 0.01569451205432415, "grad_norm": 0.1710556000471115, "global_step": 28351, "epoch": 267, "lr": 4.4522122137158316e-05} +{"train_loss": 0.018534615635871887, "grad_norm": 0.06711166352033615, "global_step": 28352, "epoch": 267, "lr": 4.451917622066077e-05} +{"train_loss": 0.015999939292669296, "grad_norm": 0.0653710663318634, "global_step": 28353, "epoch": 267, "lr": 4.4516230323420464e-05} +{"train_loss": 0.014478554017841816, "grad_norm": 0.078939288854599, "global_step": 28354, "epoch": 267, "lr": 4.451328444544774e-05} +{"train_loss": 0.015693482011556625, "grad_norm": 0.06808263063430786, "global_step": 28355, "epoch": 267, "lr": 4.451033858675293e-05} +{"train_loss": 0.01348524447530508, "grad_norm": 0.07319290190935135, "global_step": 28356, "epoch": 267, "lr": 4.450739274734645e-05} +{"train_loss": 0.014222674071788788, "grad_norm": 0.11057192087173462, "global_step": 28357, "epoch": 267, "lr": 4.4504446927238564e-05} +{"train_loss": 0.015907632187008858, "grad_norm": 0.07220352441072464, "global_step": 28358, "epoch": 267, "lr": 4.450150112643967e-05} +{"train_loss": 0.01575312949717045, "grad_norm": 0.09523453563451767, "global_step": 28359, "epoch": 267, "lr": 4.4498555344960115e-05} +{"train_loss": 0.01889493316411972, "grad_norm": 0.06705476343631744, "global_step": 28360, "epoch": 267, "lr": 4.449560958281023e-05} +{"train_loss": 0.013555383309721947, "grad_norm": 0.07186535745859146, "global_step": 28361, "epoch": 267, "lr": 4.4492663840000386e-05} +{"train_loss": 0.015985656529664993, "grad_norm": 0.14622877538204193, "global_step": 28362, "epoch": 267, "lr": 4.4489718116540944e-05} +{"train_loss": 0.014840947464108467, "grad_norm": 0.07711530476808548, "global_step": 28363, "epoch": 267, "lr": 4.4486772412442215e-05} +{"train_loss": 0.017870469018816948, "grad_norm": 0.11254516243934631, "global_step": 28364, "epoch": 267, "lr": 4.448382672771457e-05} +{"train_loss": 0.014835989102721214, "grad_norm": 0.06469598412513733, "global_step": 28365, "epoch": 267, "lr": 4.448088106236836e-05} +{"train_loss": 0.014207419008016586, "grad_norm": 0.07181591540575027, "global_step": 28366, "epoch": 267, "lr": 4.4477935416413945e-05} +{"train_loss": 0.016333140432834625, "grad_norm": 0.10276100784540176, "global_step": 28367, "epoch": 267, "lr": 4.4474989789861645e-05} +{"train_loss": 0.01406482607126236, "grad_norm": 0.13233721256256104, "global_step": 28368, "epoch": 267, "lr": 4.447204418272187e-05} +{"train_loss": 0.015028939582407475, "grad_norm": 0.06496810913085938, "global_step": 28369, "epoch": 267, "lr": 4.446909859500488e-05} +{"train_loss": 0.014121071435511112, "grad_norm": 0.14290311932563782, "global_step": 28370, "epoch": 267, "lr": 4.4466153026721094e-05} +{"train_loss": 0.017107635736465454, "grad_norm": 0.13484883308410645, "global_step": 28371, "epoch": 267, "lr": 4.4463207477880834e-05} +{"train_loss": 0.01604742929339409, "grad_norm": 0.11118078231811523, "global_step": 28372, "epoch": 267, "lr": 4.4460261948494455e-05} +{"train_loss": 0.015285631641745567, "grad_norm": 0.08385685086250305, "global_step": 28373, "epoch": 267, "lr": 4.4457316438572324e-05} +{"train_loss": 0.015461643226444721, "grad_norm": 0.07786161452531815, "global_step": 28374, "epoch": 267, "lr": 4.4454370948124754e-05} +{"train_loss": 0.014127517119050026, "grad_norm": 0.18177534639835358, "global_step": 28375, "epoch": 267, "lr": 4.4451425477162114e-05} +{"train_loss": 0.01602352410554886, "grad_norm": 0.06289760768413544, "global_step": 28376, "epoch": 267, "lr": 4.444848002569475e-05} +{"train_loss": 0.01311334129422903, "grad_norm": 0.05916134640574455, "global_step": 28377, "epoch": 267, "lr": 4.444553459373301e-05} +{"train_loss": 0.01775638572871685, "grad_norm": 0.06493186950683594, "global_step": 28378, "epoch": 267, "lr": 4.4442589181287253e-05} +{"train_loss": 0.01435010600835085, "grad_norm": 0.07885731756687164, "global_step": 28379, "epoch": 267, "lr": 4.443964378836783e-05} +{"train_loss": 0.015625877305865288, "grad_norm": 0.11070924997329712, "global_step": 28380, "epoch": 267, "lr": 4.443669841498507e-05} +{"train_loss": 0.0177347119897604, "grad_norm": 0.09510467946529388, "global_step": 28381, "epoch": 267, "lr": 4.443375306114933e-05} +{"train_loss": 0.01629185676574707, "grad_norm": 0.07399151474237442, "global_step": 28382, "epoch": 267, "lr": 4.443080772687096e-05} +{"train_loss": 0.013128806836903095, "grad_norm": 0.0798690989613533, "global_step": 28383, "epoch": 267, "lr": 4.4427862412160314e-05} +{"train_loss": 0.013978218659758568, "grad_norm": 0.0769357681274414, "global_step": 28384, "epoch": 267, "lr": 4.442491711702773e-05} +{"train_loss": 0.015016959980130196, "grad_norm": 0.06126983463764191, "global_step": 28385, "epoch": 267, "lr": 4.442197184148359e-05} +{"train_loss": 0.014968465082347393, "grad_norm": 0.14039888978004456, "global_step": 28386, "epoch": 267, "lr": 4.441902658553818e-05} +{"train_loss": 0.01642472855746746, "grad_norm": 0.07313910126686096, "global_step": 28387, "epoch": 267, "lr": 4.44160813492019e-05} +{"train_loss": 0.016968395560979843, "grad_norm": 0.0872957780957222, "global_step": 28388, "epoch": 267, "lr": 4.4413136132485076e-05} +{"train_loss": 0.015036260709166527, "grad_norm": 0.0928301215171814, "global_step": 28389, "epoch": 267, "lr": 4.4410190935398063e-05} +{"train_loss": 0.013031069189310074, "grad_norm": 0.056313250213861465, "global_step": 28390, "epoch": 267, "lr": 4.440724575795121e-05} +{"train_loss": 0.015688367187976837, "grad_norm": 0.08015602827072144, "global_step": 28391, "epoch": 267, "lr": 4.440430060015487e-05} +{"train_loss": 0.017006073147058487, "grad_norm": 0.0691286250948906, "global_step": 28392, "epoch": 267, "lr": 4.440135546201938e-05} +{"train_loss": 0.018617652356624603, "grad_norm": 0.10134217888116837, "global_step": 28393, "epoch": 267, "lr": 4.439841034355508e-05} +{"train_loss": 0.017035001888871193, "grad_norm": 0.09715494513511658, "global_step": 28394, "epoch": 267, "lr": 4.439546524477234e-05} +{"train_loss": 0.015568575821816921, "grad_norm": 0.10261786729097366, "global_step": 28395, "epoch": 267, "lr": 4.43925201656815e-05} +{"train_loss": 0.015972666442394257, "grad_norm": 0.05408347398042679, "global_step": 28396, "epoch": 267, "lr": 4.438957510629289e-05} +{"train_loss": 0.013547249138355255, "grad_norm": 0.09629428386688232, "global_step": 28397, "epoch": 267, "lr": 4.438663006661691e-05} +{"train_loss": 0.015135470777750015, "grad_norm": 0.1054898053407669, "global_step": 28398, "epoch": 267, "lr": 4.438368504666384e-05} +{"train_loss": 0.01691352389752865, "grad_norm": 0.10826858133077621, "global_step": 28399, "epoch": 267, "lr": 4.4380740046444066e-05} +{"train_loss": 0.013110746629536152, "grad_norm": 0.059806711971759796, "global_step": 28400, "epoch": 267, "lr": 4.437779506596792e-05} +{"train_loss": 0.014522222802042961, "grad_norm": 0.07636551558971405, "global_step": 28401, "epoch": 267, "lr": 4.437485010524576e-05} +{"train_loss": 0.01619207113981247, "grad_norm": 0.056995995342731476, "global_step": 28402, "epoch": 267, "lr": 4.437190516428794e-05} +{"train_loss": 0.015160981565713882, "grad_norm": 0.1264537274837494, "global_step": 28403, "epoch": 267, "lr": 4.43689602431048e-05} +{"train_loss": 0.017073825001716614, "grad_norm": 0.0798979103565216, "global_step": 28404, "epoch": 267, "lr": 4.4366015341706673e-05} +{"train_loss": 0.01757807284593582, "grad_norm": 0.07766000926494598, "global_step": 28405, "epoch": 267, "lr": 4.4363070460103926e-05} +{"train_loss": 0.016456864774227142, "grad_norm": 0.125684916973114, "global_step": 28406, "epoch": 267, "lr": 4.436012559830688e-05} +{"train_loss": 0.015452946079367737, "grad_norm": 0.10193521529436111, "global_step": 28407, "epoch": 267, "lr": 4.435718075632592e-05, "val_loss": 0.020053328946232796} +{"train_loss": 0.01939581148326397, "grad_norm": 0.09276756644248962, "global_step": 28408, "epoch": 268, "lr": 4.435423593417136e-05} +{"train_loss": 0.014711122028529644, "grad_norm": 0.13587243854999542, "global_step": 28409, "epoch": 268, "lr": 4.4351291131853586e-05} +{"train_loss": 0.013711451552808285, "grad_norm": 0.07975059002637863, "global_step": 28410, "epoch": 268, "lr": 4.4348346349382886e-05} +{"train_loss": 0.016205836087465286, "grad_norm": 0.0791175439953804, "global_step": 28411, "epoch": 268, "lr": 4.434540158676965e-05} +{"train_loss": 0.01676863618195057, "grad_norm": 0.08568083494901657, "global_step": 28412, "epoch": 268, "lr": 4.4342456844024216e-05} +{"train_loss": 0.01655585691332817, "grad_norm": 0.12766189873218536, "global_step": 28413, "epoch": 268, "lr": 4.433951212115692e-05} +{"train_loss": 0.015362901613116264, "grad_norm": 0.05803317949175835, "global_step": 28414, "epoch": 268, "lr": 4.433656741817813e-05} +{"train_loss": 0.015427783131599426, "grad_norm": 0.07005812972784042, "global_step": 28415, "epoch": 268, "lr": 4.433362273509818e-05} +{"train_loss": 0.013666789047420025, "grad_norm": 0.11928121745586395, "global_step": 28416, "epoch": 268, "lr": 4.4330678071927414e-05} +{"train_loss": 0.015922658145427704, "grad_norm": 0.058570198714733124, "global_step": 28417, "epoch": 268, "lr": 4.4327733428676174e-05} +{"train_loss": 0.016146011650562286, "grad_norm": 0.09041493386030197, "global_step": 28418, "epoch": 268, "lr": 4.43247888053548e-05} +{"train_loss": 0.012170204892754555, "grad_norm": 0.07908277213573456, "global_step": 28419, "epoch": 268, "lr": 4.432184420197366e-05} +{"train_loss": 0.014909781515598297, "grad_norm": 0.0831855833530426, "global_step": 28420, "epoch": 268, "lr": 4.4318899618543097e-05} +{"train_loss": 0.015667341649532318, "grad_norm": 0.059712208807468414, "global_step": 28421, "epoch": 268, "lr": 4.4315955055073456e-05} +{"train_loss": 0.015035409480333328, "grad_norm": 0.09467078745365143, "global_step": 28422, "epoch": 268, "lr": 4.431301051157506e-05} +{"train_loss": 0.015425803139805794, "grad_norm": 0.15776433050632477, "global_step": 28423, "epoch": 268, "lr": 4.431006598805828e-05} +{"train_loss": 0.01650175079703331, "grad_norm": 0.09007278084754944, "global_step": 28424, "epoch": 268, "lr": 4.430712148453345e-05} +{"train_loss": 0.01551454421132803, "grad_norm": 0.14337490499019623, "global_step": 28425, "epoch": 268, "lr": 4.430417700101091e-05} +{"train_loss": 0.014733390882611275, "grad_norm": 0.07709434628486633, "global_step": 28426, "epoch": 268, "lr": 4.430123253750105e-05} +{"train_loss": 0.015692558139562607, "grad_norm": 0.15670229494571686, "global_step": 28427, "epoch": 268, "lr": 4.429828809401415e-05} +{"train_loss": 0.01606074720621109, "grad_norm": 0.0740569531917572, "global_step": 28428, "epoch": 268, "lr": 4.42953436705606e-05} +{"train_loss": 0.01529995072633028, "grad_norm": 0.08985910564661026, "global_step": 28429, "epoch": 268, "lr": 4.429239926715072e-05} +{"train_loss": 0.012482758611440659, "grad_norm": 0.09702076762914658, "global_step": 28430, "epoch": 268, "lr": 4.428945488379486e-05} +{"train_loss": 0.01575489342212677, "grad_norm": 0.095635786652565, "global_step": 28431, "epoch": 268, "lr": 4.4286510520503384e-05} +{"train_loss": 0.015082430094480515, "grad_norm": 0.05927887186408043, "global_step": 28432, "epoch": 268, "lr": 4.4283566177286636e-05} +{"train_loss": 0.014750480651855469, "grad_norm": 0.07972610741853714, "global_step": 28433, "epoch": 268, "lr": 4.4280621854154944e-05} +{"train_loss": 0.016502268612384796, "grad_norm": 0.10272791236639023, "global_step": 28434, "epoch": 268, "lr": 4.427767755111865e-05} +{"train_loss": 0.017186835408210754, "grad_norm": 0.07176525890827179, "global_step": 28435, "epoch": 268, "lr": 4.427473326818811e-05} +{"train_loss": 0.016299812123179436, "grad_norm": 0.15569226443767548, "global_step": 28436, "epoch": 268, "lr": 4.427178900537367e-05} +{"train_loss": 0.01522749662399292, "grad_norm": 0.10927364975214005, "global_step": 28437, "epoch": 268, "lr": 4.4268844762685677e-05} +{"train_loss": 0.017843086272478104, "grad_norm": 0.11923885345458984, "global_step": 28438, "epoch": 268, "lr": 4.426590054013448e-05} +{"train_loss": 0.017570577561855316, "grad_norm": 0.10297606885433197, "global_step": 28439, "epoch": 268, "lr": 4.4262956337730385e-05} +{"train_loss": 0.015023807995021343, "grad_norm": 0.07054876536130905, "global_step": 28440, "epoch": 268, "lr": 4.4260012155483786e-05} +{"train_loss": 0.013902992941439152, "grad_norm": 0.067018523812294, "global_step": 28441, "epoch": 268, "lr": 4.425706799340501e-05} +{"train_loss": 0.014727838337421417, "grad_norm": 0.05092153698205948, "global_step": 28442, "epoch": 268, "lr": 4.425412385150438e-05} +{"train_loss": 0.015072331763803959, "grad_norm": 0.08216101676225662, "global_step": 28443, "epoch": 268, "lr": 4.4251179729792274e-05} +{"train_loss": 0.017493464052677155, "grad_norm": 0.11181216686964035, "global_step": 28444, "epoch": 268, "lr": 4.424823562827904e-05} +{"train_loss": 0.015476065687835217, "grad_norm": 0.11668824404478073, "global_step": 28445, "epoch": 268, "lr": 4.4245291546974985e-05} +{"train_loss": 0.014889413490891457, "grad_norm": 0.0648743212223053, "global_step": 28446, "epoch": 268, "lr": 4.424234748589048e-05} +{"train_loss": 0.01431818213313818, "grad_norm": 0.07854458689689636, "global_step": 28447, "epoch": 268, "lr": 4.4239403445035844e-05} +{"train_loss": 0.012349092401564121, "grad_norm": 0.05000175163149834, "global_step": 28448, "epoch": 268, "lr": 4.4236459424421454e-05} +{"train_loss": 0.01763453707098961, "grad_norm": 0.07995907962322235, "global_step": 28449, "epoch": 268, "lr": 4.4233515424057636e-05} +{"train_loss": 0.013624539598822594, "grad_norm": 0.12136060744524002, "global_step": 28450, "epoch": 268, "lr": 4.423057144395475e-05} +{"train_loss": 0.015820030122995377, "grad_norm": 0.11776338517665863, "global_step": 28451, "epoch": 268, "lr": 4.4227627484123115e-05} +{"train_loss": 0.014245414175093174, "grad_norm": 0.16274955868721008, "global_step": 28452, "epoch": 268, "lr": 4.422468354457308e-05} +{"train_loss": 0.01495268288999796, "grad_norm": 0.07366307079792023, "global_step": 28453, "epoch": 268, "lr": 4.422173962531501e-05} +{"train_loss": 0.01702750101685524, "grad_norm": 0.10393749922513962, "global_step": 28454, "epoch": 268, "lr": 4.4218795726359217e-05} +{"train_loss": 0.01507966872304678, "grad_norm": 0.08264654129743576, "global_step": 28455, "epoch": 268, "lr": 4.421585184771607e-05} +{"train_loss": 0.016081836074590683, "grad_norm": 0.10298094898462296, "global_step": 28456, "epoch": 268, "lr": 4.421290798939592e-05} +{"train_loss": 0.014978267252445221, "grad_norm": 0.15929624438285828, "global_step": 28457, "epoch": 268, "lr": 4.420996415140908e-05} +{"train_loss": 0.016790172085165977, "grad_norm": 0.06967739015817642, "global_step": 28458, "epoch": 268, "lr": 4.4207020333765906e-05} +{"train_loss": 0.01667425036430359, "grad_norm": 0.10023212432861328, "global_step": 28459, "epoch": 268, "lr": 4.4204076536476733e-05} +{"train_loss": 0.016245203092694283, "grad_norm": 0.10582254081964493, "global_step": 28460, "epoch": 268, "lr": 4.4201132759551934e-05} +{"train_loss": 0.014156347140669823, "grad_norm": 0.06523966044187546, "global_step": 28461, "epoch": 268, "lr": 4.419818900300182e-05} +{"train_loss": 0.01517079770565033, "grad_norm": 0.07179011404514313, "global_step": 28462, "epoch": 268, "lr": 4.4195245266836774e-05} +{"train_loss": 0.016520241275429726, "grad_norm": 0.060487426817417145, "global_step": 28463, "epoch": 268, "lr": 4.419230155106708e-05} +{"train_loss": 0.017678160220384598, "grad_norm": 0.11113854497671127, "global_step": 28464, "epoch": 268, "lr": 4.418935785570312e-05} +{"train_loss": 0.01411772333085537, "grad_norm": 0.06781446933746338, "global_step": 28465, "epoch": 268, "lr": 4.418641418075523e-05} +{"train_loss": 0.014297325164079666, "grad_norm": 0.10716581344604492, "global_step": 28466, "epoch": 268, "lr": 4.418347052623375e-05} +{"train_loss": 0.014291472733020782, "grad_norm": 0.06003936007618904, "global_step": 28467, "epoch": 268, "lr": 4.4180526892149025e-05} +{"train_loss": 0.01573437638580799, "grad_norm": 0.11587204784154892, "global_step": 28468, "epoch": 268, "lr": 4.4177583278511415e-05} +{"train_loss": 0.017574027180671692, "grad_norm": 0.07335434854030609, "global_step": 28469, "epoch": 268, "lr": 4.417463968533123e-05} +{"train_loss": 0.015485559590160847, "grad_norm": 0.08167880773544312, "global_step": 28470, "epoch": 268, "lr": 4.417169611261883e-05} +{"train_loss": 0.014462713152170181, "grad_norm": 0.08041156083345413, "global_step": 28471, "epoch": 268, "lr": 4.4168752560384555e-05} +{"train_loss": 0.017260586842894554, "grad_norm": 0.09625844657421112, "global_step": 28472, "epoch": 268, "lr": 4.416580902863874e-05} +{"train_loss": 0.01667618192732334, "grad_norm": 0.06374084949493408, "global_step": 28473, "epoch": 268, "lr": 4.416286551739176e-05} +{"train_loss": 0.018558964133262634, "grad_norm": 0.1025824323296547, "global_step": 28474, "epoch": 268, "lr": 4.4159922026653915e-05} +{"train_loss": 0.014392398297786713, "grad_norm": 0.07474003732204437, "global_step": 28475, "epoch": 268, "lr": 4.4156978556435564e-05} +{"train_loss": 0.017199117690324783, "grad_norm": 0.11032386124134064, "global_step": 28476, "epoch": 268, "lr": 4.4154035106747036e-05} +{"train_loss": 0.013589856214821339, "grad_norm": 0.07336647808551788, "global_step": 28477, "epoch": 268, "lr": 4.4151091677598704e-05} +{"train_loss": 0.015893345698714256, "grad_norm": 0.1401004195213318, "global_step": 28478, "epoch": 268, "lr": 4.414814826900089e-05} +{"train_loss": 0.015681609511375427, "grad_norm": 0.12888416647911072, "global_step": 28479, "epoch": 268, "lr": 4.4145204880963945e-05} +{"train_loss": 0.016725808382034302, "grad_norm": 0.12276611477136612, "global_step": 28480, "epoch": 268, "lr": 4.4142261513498185e-05} +{"train_loss": 0.014138646423816681, "grad_norm": 0.0648016482591629, "global_step": 28481, "epoch": 268, "lr": 4.4139318166613985e-05} +{"train_loss": 0.017102766782045364, "grad_norm": 0.08003733307123184, "global_step": 28482, "epoch": 268, "lr": 4.413637484032166e-05} +{"train_loss": 0.012401517480611801, "grad_norm": 0.06614610552787781, "global_step": 28483, "epoch": 268, "lr": 4.413343153463157e-05} +{"train_loss": 0.014133335091173649, "grad_norm": 0.08974120020866394, "global_step": 28484, "epoch": 268, "lr": 4.4130488249554054e-05} +{"train_loss": 0.016220955178141594, "grad_norm": 0.12886887788772583, "global_step": 28485, "epoch": 268, "lr": 4.4127544985099454e-05} +{"train_loss": 0.013794420287013054, "grad_norm": 0.06585362553596497, "global_step": 28486, "epoch": 268, "lr": 4.41246017412781e-05} +{"train_loss": 0.017844952642917633, "grad_norm": 0.08884294331073761, "global_step": 28487, "epoch": 268, "lr": 4.4121658518100347e-05} +{"train_loss": 0.014554074965417385, "grad_norm": 0.047221649438142776, "global_step": 28488, "epoch": 268, "lr": 4.4118715315576506e-05} +{"train_loss": 0.01734859123826027, "grad_norm": 0.07239959388971329, "global_step": 28489, "epoch": 268, "lr": 4.411577213371696e-05} +{"train_loss": 0.011735396459698677, "grad_norm": 0.08050671964883804, "global_step": 28490, "epoch": 268, "lr": 4.411282897253204e-05} +{"train_loss": 0.015735242515802383, "grad_norm": 0.08008153736591339, "global_step": 28491, "epoch": 268, "lr": 4.410988583203208e-05} +{"train_loss": 0.014310348778963089, "grad_norm": 0.06009720265865326, "global_step": 28492, "epoch": 268, "lr": 4.410694271222741e-05} +{"train_loss": 0.016001321375370026, "grad_norm": 0.06494665890932083, "global_step": 28493, "epoch": 268, "lr": 4.4103999613128364e-05} +{"train_loss": 0.013854365795850754, "grad_norm": 0.05534154549241066, "global_step": 28494, "epoch": 268, "lr": 4.410105653474532e-05} +{"train_loss": 0.015689048916101456, "grad_norm": 0.134418323636055, "global_step": 28495, "epoch": 268, "lr": 4.409811347708858e-05} +{"train_loss": 0.01761605031788349, "grad_norm": 0.07595348358154297, "global_step": 28496, "epoch": 268, "lr": 4.409517044016852e-05} +{"train_loss": 0.015546794980764389, "grad_norm": 0.07836618274450302, "global_step": 28497, "epoch": 268, "lr": 4.409222742399547e-05} +{"train_loss": 0.015410413034260273, "grad_norm": 0.08134573698043823, "global_step": 28498, "epoch": 268, "lr": 4.408928442857975e-05} +{"train_loss": 0.014765845611691475, "grad_norm": 0.0562509261071682, "global_step": 28499, "epoch": 268, "lr": 4.4086341453931716e-05} +{"train_loss": 0.013437381014227867, "grad_norm": 0.056546539068222046, "global_step": 28500, "epoch": 268, "lr": 4.40833985000617e-05} +{"train_loss": 0.01683872565627098, "grad_norm": 0.04971906915307045, "global_step": 28501, "epoch": 268, "lr": 4.408045556698005e-05} +{"train_loss": 0.015743521973490715, "grad_norm": 0.07835080474615097, "global_step": 28502, "epoch": 268, "lr": 4.407751265469712e-05} +{"train_loss": 0.014372988604009151, "grad_norm": 0.1288709193468094, "global_step": 28503, "epoch": 268, "lr": 4.407456976322323e-05} +{"train_loss": 0.016411125659942627, "grad_norm": 0.13750120997428894, "global_step": 28504, "epoch": 268, "lr": 4.4071626892568725e-05} +{"train_loss": 0.01516026258468628, "grad_norm": 0.17532438039779663, "global_step": 28505, "epoch": 268, "lr": 4.406868404274393e-05} +{"train_loss": 0.012503592297434807, "grad_norm": 0.07674624770879745, "global_step": 28506, "epoch": 268, "lr": 4.406574121375921e-05} +{"train_loss": 0.016879932954907417, "grad_norm": 0.06452370434999466, "global_step": 28507, "epoch": 268, "lr": 4.40627984056249e-05} +{"train_loss": 0.018436098471283913, "grad_norm": 0.08198876678943634, "global_step": 28508, "epoch": 268, "lr": 4.4059855618351316e-05} +{"train_loss": 0.01224667951464653, "grad_norm": 0.07319879531860352, "global_step": 28509, "epoch": 268, "lr": 4.4056912851948836e-05} +{"train_loss": 0.015720553696155548, "grad_norm": 0.12417370826005936, "global_step": 28510, "epoch": 268, "lr": 4.405397010642778e-05} +{"train_loss": 0.01335868425667286, "grad_norm": 0.0793064758181572, "global_step": 28511, "epoch": 268, "lr": 4.405102738179848e-05} +{"train_loss": 0.014992072246968746, "grad_norm": 0.07926027476787567, "global_step": 28512, "epoch": 268, "lr": 4.4048084678071266e-05} +{"train_loss": 0.015468048327163144, "grad_norm": 0.11617052555084229, "global_step": 28513, "epoch": 268, "lr": 4.404514199525651e-05, "val_loss": 0.01945105567574501} +{"train_loss": 0.017959855496883392, "grad_norm": 0.13047058880329132, "global_step": 28514, "epoch": 269, "lr": 4.4042199333364535e-05} +{"train_loss": 0.014991529285907745, "grad_norm": 0.08800195157527924, "global_step": 28515, "epoch": 269, "lr": 4.40392566924057e-05} +{"train_loss": 0.015165653079748154, "grad_norm": 0.0880211889743805, "global_step": 28516, "epoch": 269, "lr": 4.4036314072390304e-05} +{"train_loss": 0.017372382804751396, "grad_norm": 0.11496339738368988, "global_step": 28517, "epoch": 269, "lr": 4.40333714733287e-05} +{"train_loss": 0.015765024349093437, "grad_norm": 0.08351532369852066, "global_step": 28518, "epoch": 269, "lr": 4.403042889523124e-05} +{"train_loss": 0.014328496530652046, "grad_norm": 0.09021121263504028, "global_step": 28519, "epoch": 269, "lr": 4.402748633810826e-05} +{"train_loss": 0.014912762679159641, "grad_norm": 0.07985334098339081, "global_step": 28520, "epoch": 269, "lr": 4.4024543801970086e-05} +{"train_loss": 0.015039796009659767, "grad_norm": 0.05634916573762894, "global_step": 28521, "epoch": 269, "lr": 4.4021601286827096e-05} +{"train_loss": 0.014759378507733345, "grad_norm": 0.05494454875588417, "global_step": 28522, "epoch": 269, "lr": 4.4018658792689574e-05} +{"train_loss": 0.015703223645687103, "grad_norm": 0.07039700448513031, "global_step": 28523, "epoch": 269, "lr": 4.401571631956788e-05} +{"train_loss": 0.016258545219898224, "grad_norm": 0.04680049419403076, "global_step": 28524, "epoch": 269, "lr": 4.401277386747236e-05} +{"train_loss": 0.01561908982694149, "grad_norm": 0.1416051685810089, "global_step": 28525, "epoch": 269, "lr": 4.400983143641335e-05} +{"train_loss": 0.015914615243673325, "grad_norm": 0.11972667276859283, "global_step": 28526, "epoch": 269, "lr": 4.400688902640121e-05} +{"train_loss": 0.016419770196080208, "grad_norm": 0.13272513449192047, "global_step": 28527, "epoch": 269, "lr": 4.400394663744624e-05} +{"train_loss": 0.015642501413822174, "grad_norm": 0.07099589705467224, "global_step": 28528, "epoch": 269, "lr": 4.400100426955878e-05} +{"train_loss": 0.014114917255938053, "grad_norm": 0.08539378643035889, "global_step": 28529, "epoch": 269, "lr": 4.399806192274918e-05} +{"train_loss": 0.016065865755081177, "grad_norm": 0.08189869672060013, "global_step": 28530, "epoch": 269, "lr": 4.39951195970278e-05} +{"train_loss": 0.014860084280371666, "grad_norm": 0.056488487869501114, "global_step": 28531, "epoch": 269, "lr": 4.3992177292404954e-05} +{"train_loss": 0.017225058749318123, "grad_norm": 0.08047167211771011, "global_step": 28532, "epoch": 269, "lr": 4.398923500889099e-05} +{"train_loss": 0.014598667621612549, "grad_norm": 0.04613940045237541, "global_step": 28533, "epoch": 269, "lr": 4.3986292746496236e-05} +{"train_loss": 0.014111891388893127, "grad_norm": 0.062439557164907455, "global_step": 28534, "epoch": 269, "lr": 4.3983350505231016e-05} +{"train_loss": 0.014903521165251732, "grad_norm": 0.06772400438785553, "global_step": 28535, "epoch": 269, "lr": 4.39804082851057e-05} +{"train_loss": 0.01389063149690628, "grad_norm": 0.08523186296224594, "global_step": 28536, "epoch": 269, "lr": 4.397746608613061e-05} +{"train_loss": 0.015786994248628616, "grad_norm": 0.12857452034950256, "global_step": 28537, "epoch": 269, "lr": 4.397452390831608e-05} +{"train_loss": 0.014084305614233017, "grad_norm": 0.05633847042918205, "global_step": 28538, "epoch": 269, "lr": 4.3971581751672466e-05} +{"train_loss": 0.016129145398736, "grad_norm": 0.138031467795372, "global_step": 28539, "epoch": 269, "lr": 4.3968639616210084e-05} +{"train_loss": 0.01881491020321846, "grad_norm": 0.08439414948225021, "global_step": 28540, "epoch": 269, "lr": 4.396569750193928e-05} +{"train_loss": 0.01811281219124794, "grad_norm": 0.08323654532432556, "global_step": 28541, "epoch": 269, "lr": 4.396275540887038e-05} +{"train_loss": 0.015459680929780006, "grad_norm": 0.0921194851398468, "global_step": 28542, "epoch": 269, "lr": 4.395981333701374e-05} +{"train_loss": 0.014641718938946724, "grad_norm": 0.07926227897405624, "global_step": 28543, "epoch": 269, "lr": 4.39568712863797e-05} +{"train_loss": 0.014510630629956722, "grad_norm": 0.10924577713012695, "global_step": 28544, "epoch": 269, "lr": 4.395392925697859e-05} +{"train_loss": 0.015840696170926094, "grad_norm": 0.09080497920513153, "global_step": 28545, "epoch": 269, "lr": 4.395098724882073e-05} +{"train_loss": 0.01679207757115364, "grad_norm": 0.2491435557603836, "global_step": 28546, "epoch": 269, "lr": 4.394804526191646e-05} +{"train_loss": 0.015399293974041939, "grad_norm": 0.10491068661212921, "global_step": 28547, "epoch": 269, "lr": 4.3945103296276135e-05} +{"train_loss": 0.014461269602179527, "grad_norm": 0.08317916840314865, "global_step": 28548, "epoch": 269, "lr": 4.394216135191009e-05} +{"train_loss": 0.013948578387498856, "grad_norm": 0.11653327941894531, "global_step": 28549, "epoch": 269, "lr": 4.3939219428828644e-05} +{"train_loss": 0.015861820429563522, "grad_norm": 0.09457520395517349, "global_step": 28550, "epoch": 269, "lr": 4.393627752704216e-05} +{"train_loss": 0.014120586216449738, "grad_norm": 0.062382638454437256, "global_step": 28551, "epoch": 269, "lr": 4.393333564656096e-05} +{"train_loss": 0.014100219123065472, "grad_norm": 0.06656235456466675, "global_step": 28552, "epoch": 269, "lr": 4.393039378739537e-05} +{"train_loss": 0.01625562086701393, "grad_norm": 0.05606238543987274, "global_step": 28553, "epoch": 269, "lr": 4.392745194955573e-05} +{"train_loss": 0.013819519430398941, "grad_norm": 0.07203999161720276, "global_step": 28554, "epoch": 269, "lr": 4.392451013305239e-05} +{"train_loss": 0.01574547402560711, "grad_norm": 0.07003159821033478, "global_step": 28555, "epoch": 269, "lr": 4.392156833789569e-05} +{"train_loss": 0.016073687002062798, "grad_norm": 0.08251528441905975, "global_step": 28556, "epoch": 269, "lr": 4.391862656409596e-05} +{"train_loss": 0.017315909266471863, "grad_norm": 0.1677524894475937, "global_step": 28557, "epoch": 269, "lr": 4.3915684811663514e-05} +{"train_loss": 0.01415807381272316, "grad_norm": 0.06386575847864151, "global_step": 28558, "epoch": 269, "lr": 4.3912743080608705e-05} +{"train_loss": 0.01409151405096054, "grad_norm": 0.07827294617891312, "global_step": 28559, "epoch": 269, "lr": 4.390980137094187e-05} +{"train_loss": 0.016252120956778526, "grad_norm": 0.0823802649974823, "global_step": 28560, "epoch": 269, "lr": 4.390685968267336e-05} +{"train_loss": 0.015764782205224037, "grad_norm": 0.08306296914815903, "global_step": 28561, "epoch": 269, "lr": 4.390391801581347e-05} +{"train_loss": 0.015251385048031807, "grad_norm": 0.1336965262889862, "global_step": 28562, "epoch": 269, "lr": 4.3900976370372596e-05} +{"train_loss": 0.015959983691573143, "grad_norm": 0.08357059955596924, "global_step": 28563, "epoch": 269, "lr": 4.389803474636101e-05} +{"train_loss": 0.016077546402812004, "grad_norm": 0.11364033818244934, "global_step": 28564, "epoch": 269, "lr": 4.389509314378909e-05} +{"train_loss": 0.01583884283900261, "grad_norm": 0.05601575970649719, "global_step": 28565, "epoch": 269, "lr": 4.389215156266715e-05} +{"train_loss": 0.018132727593183517, "grad_norm": 0.06688734889030457, "global_step": 28566, "epoch": 269, "lr": 4.3889210003005524e-05} +{"train_loss": 0.014327597804367542, "grad_norm": 0.05240827798843384, "global_step": 28567, "epoch": 269, "lr": 4.3886268464814574e-05} +{"train_loss": 0.013846728019416332, "grad_norm": 0.1142161637544632, "global_step": 28568, "epoch": 269, "lr": 4.3883326948104617e-05} +{"train_loss": 0.014067746698856354, "grad_norm": 0.11881738901138306, "global_step": 28569, "epoch": 269, "lr": 4.3880385452885986e-05} +{"train_loss": 0.01592269353568554, "grad_norm": 0.05742734298110008, "global_step": 28570, "epoch": 269, "lr": 4.387744397916901e-05} +{"train_loss": 0.01651019975543022, "grad_norm": 0.054189879447221756, "global_step": 28571, "epoch": 269, "lr": 4.387450252696404e-05} +{"train_loss": 0.01668955571949482, "grad_norm": 0.06709766387939453, "global_step": 28572, "epoch": 269, "lr": 4.3871561096281404e-05} +{"train_loss": 0.01434309221804142, "grad_norm": 0.07666775584220886, "global_step": 28573, "epoch": 269, "lr": 4.386861968713145e-05} +{"train_loss": 0.016274159774184227, "grad_norm": 0.06976965069770813, "global_step": 28574, "epoch": 269, "lr": 4.386567829952449e-05} +{"train_loss": 0.015154438093304634, "grad_norm": 0.05750885605812073, "global_step": 28575, "epoch": 269, "lr": 4.386273693347086e-05} +{"train_loss": 0.014441275037825108, "grad_norm": 0.08770408481359482, "global_step": 28576, "epoch": 269, "lr": 4.385979558898091e-05} +{"train_loss": 0.0163310244679451, "grad_norm": 0.06967377662658691, "global_step": 28577, "epoch": 269, "lr": 4.3856854266064966e-05} +{"train_loss": 0.012685868889093399, "grad_norm": 0.07092933356761932, "global_step": 28578, "epoch": 269, "lr": 4.3853912964733356e-05} +{"train_loss": 0.013728720135986805, "grad_norm": 0.08493136614561081, "global_step": 28579, "epoch": 269, "lr": 4.3850971684996444e-05} +{"train_loss": 0.014353681355714798, "grad_norm": 0.05589921027421951, "global_step": 28580, "epoch": 269, "lr": 4.384803042686453e-05} +{"train_loss": 0.015595960430800915, "grad_norm": 0.15427689254283905, "global_step": 28581, "epoch": 269, "lr": 4.384508919034797e-05} +{"train_loss": 0.013405222445726395, "grad_norm": 0.058530669659376144, "global_step": 28582, "epoch": 269, "lr": 4.3842147975457074e-05} +{"train_loss": 0.01659863442182541, "grad_norm": 0.07222110778093338, "global_step": 28583, "epoch": 269, "lr": 4.38392067822022e-05} +{"train_loss": 0.01600884459912777, "grad_norm": 0.13838841021060944, "global_step": 28584, "epoch": 269, "lr": 4.383626561059368e-05} +{"train_loss": 0.012736205011606216, "grad_norm": 0.06814168393611908, "global_step": 28585, "epoch": 269, "lr": 4.383332446064185e-05} +{"train_loss": 0.01443270593881607, "grad_norm": 0.07077997177839279, "global_step": 28586, "epoch": 269, "lr": 4.3830383332357026e-05} +{"train_loss": 0.015967879444360733, "grad_norm": 0.05365065485239029, "global_step": 28587, "epoch": 269, "lr": 4.382744222574954e-05} +{"train_loss": 0.016631998121738434, "grad_norm": 0.08777333050966263, "global_step": 28588, "epoch": 269, "lr": 4.382450114082975e-05} +{"train_loss": 0.014841138385236263, "grad_norm": 0.08599138259887695, "global_step": 28589, "epoch": 269, "lr": 4.3821560077607975e-05} +{"train_loss": 0.01577821932733059, "grad_norm": 0.08972825855016708, "global_step": 28590, "epoch": 269, "lr": 4.381861903609455e-05} +{"train_loss": 0.017689133062958717, "grad_norm": 0.08944724500179291, "global_step": 28591, "epoch": 269, "lr": 4.381567801629983e-05} +{"train_loss": 0.01587151363492012, "grad_norm": 0.12176403403282166, "global_step": 28592, "epoch": 269, "lr": 4.3812737018234095e-05} +{"train_loss": 0.01587468758225441, "grad_norm": 0.06351503729820251, "global_step": 28593, "epoch": 269, "lr": 4.3809796041907725e-05} +{"train_loss": 0.015267537906765938, "grad_norm": 0.07614632695913315, "global_step": 28594, "epoch": 269, "lr": 4.380685508733105e-05} +{"train_loss": 0.015195388346910477, "grad_norm": 0.12247823178768158, "global_step": 28595, "epoch": 269, "lr": 4.380391415451436e-05} +{"train_loss": 0.013462472707033157, "grad_norm": 0.06681158393621445, "global_step": 28596, "epoch": 269, "lr": 4.380097324346805e-05} +{"train_loss": 0.01859620213508606, "grad_norm": 0.10364490747451782, "global_step": 28597, "epoch": 269, "lr": 4.379803235420243e-05} +{"train_loss": 0.016702722758054733, "grad_norm": 0.1488652229309082, "global_step": 28598, "epoch": 269, "lr": 4.379509148672781e-05} +{"train_loss": 0.016132038086652756, "grad_norm": 0.07743873447179794, "global_step": 28599, "epoch": 269, "lr": 4.379215064105454e-05} +{"train_loss": 0.017130577936768532, "grad_norm": 0.14391393959522247, "global_step": 28600, "epoch": 269, "lr": 4.378920981719296e-05} +{"train_loss": 0.01402821484953165, "grad_norm": 0.10368470847606659, "global_step": 28601, "epoch": 269, "lr": 4.378626901515339e-05} +{"train_loss": 0.015054996125400066, "grad_norm": 0.060877926647663116, "global_step": 28602, "epoch": 269, "lr": 4.3783328234946165e-05} +{"train_loss": 0.014889651909470558, "grad_norm": 0.13883358240127563, "global_step": 28603, "epoch": 269, "lr": 4.3780387476581654e-05} +{"train_loss": 0.015539679676294327, "grad_norm": 0.056740064173936844, "global_step": 28604, "epoch": 269, "lr": 4.377744674007012e-05} +{"train_loss": 0.015322509221732616, "grad_norm": 0.079439178109169, "global_step": 28605, "epoch": 269, "lr": 4.377450602542194e-05} +{"train_loss": 0.01574244536459446, "grad_norm": 0.07867991179227829, "global_step": 28606, "epoch": 269, "lr": 4.377156533264745e-05} +{"train_loss": 0.014628708362579346, "grad_norm": 0.12147387117147446, "global_step": 28607, "epoch": 269, "lr": 4.3768624661756943e-05} +{"train_loss": 0.015989549458026886, "grad_norm": 0.17268632352352142, "global_step": 28608, "epoch": 269, "lr": 4.37656840127608e-05} +{"train_loss": 0.014608476310968399, "grad_norm": 0.09494777023792267, "global_step": 28609, "epoch": 269, "lr": 4.376274338566934e-05} +{"train_loss": 0.014723301865160465, "grad_norm": 0.05294683203101158, "global_step": 28610, "epoch": 269, "lr": 4.3759802780492884e-05} +{"train_loss": 0.018944712355732918, "grad_norm": 0.15503355860710144, "global_step": 28611, "epoch": 269, "lr": 4.375686219724175e-05} +{"train_loss": 0.017275620251893997, "grad_norm": 0.06344863027334213, "global_step": 28612, "epoch": 269, "lr": 4.37539216359263e-05} +{"train_loss": 0.01668636128306389, "grad_norm": 0.11512042582035065, "global_step": 28613, "epoch": 269, "lr": 4.3750981096556846e-05} +{"train_loss": 0.014045793563127518, "grad_norm": 0.08589348942041397, "global_step": 28614, "epoch": 269, "lr": 4.3748040579143726e-05} +{"train_loss": 0.014192020520567894, "grad_norm": 0.09230896830558777, "global_step": 28615, "epoch": 269, "lr": 4.37451000836973e-05} +{"train_loss": 0.014861024916172028, "grad_norm": 0.09893038123846054, "global_step": 28616, "epoch": 269, "lr": 4.374215961022784e-05} +{"train_loss": 0.012989100068807602, "grad_norm": 0.09780740737915039, "global_step": 28617, "epoch": 269, "lr": 4.373921915874572e-05} +{"train_loss": 0.015366623178124428, "grad_norm": 0.13898789882659912, "global_step": 28618, "epoch": 269, "lr": 4.373627872926126e-05} +{"train_loss": 0.01546096274594091, "grad_norm": 0.11088893562555313, "global_step": 28619, "epoch": 269, "lr": 4.373333832178478e-05, "val_loss": 0.019331935793161392} +{"train_loss": 0.016662096604704857, "grad_norm": 0.07076999545097351, "global_step": 28620, "epoch": 270, "lr": 4.373039793632664e-05} +{"train_loss": 0.018055032938718796, "grad_norm": 0.08842083811759949, "global_step": 28621, "epoch": 270, "lr": 4.372745757289717e-05} +{"train_loss": 0.018033161759376526, "grad_norm": 0.06585000455379486, "global_step": 28622, "epoch": 270, "lr": 4.3724517231506665e-05} +{"train_loss": 0.017605993896722794, "grad_norm": 0.104530468583107, "global_step": 28623, "epoch": 270, "lr": 4.3721576912165484e-05} +{"train_loss": 0.014998573809862137, "grad_norm": 0.14050915837287903, "global_step": 28624, "epoch": 270, "lr": 4.3718636614883934e-05} +{"train_loss": 0.011629670858383179, "grad_norm": 0.10687234252691269, "global_step": 28625, "epoch": 270, "lr": 4.3715696339672377e-05} +{"train_loss": 0.015809401869773865, "grad_norm": 0.07859908044338226, "global_step": 28626, "epoch": 270, "lr": 4.3712756086541145e-05} +{"train_loss": 0.016127970069646835, "grad_norm": 0.07991190254688263, "global_step": 28627, "epoch": 270, "lr": 4.370981585550054e-05} +{"train_loss": 0.016322078183293343, "grad_norm": 0.1227673813700676, "global_step": 28628, "epoch": 270, "lr": 4.3706875646560894e-05} +{"train_loss": 0.013716880232095718, "grad_norm": 0.09671778976917267, "global_step": 28629, "epoch": 270, "lr": 4.370393545973257e-05} +{"train_loss": 0.017539506778120995, "grad_norm": 0.15493090450763702, "global_step": 28630, "epoch": 270, "lr": 4.370099529502587e-05} +{"train_loss": 0.015795107930898666, "grad_norm": 0.07951091229915619, "global_step": 28631, "epoch": 270, "lr": 4.369805515245112e-05} +{"train_loss": 0.01525383722037077, "grad_norm": 0.07625004649162292, "global_step": 28632, "epoch": 270, "lr": 4.36951150320187e-05} +{"train_loss": 0.01528206653892994, "grad_norm": 0.06129101663827896, "global_step": 28633, "epoch": 270, "lr": 4.369217493373887e-05} +{"train_loss": 0.014876590110361576, "grad_norm": 0.13578806817531586, "global_step": 28634, "epoch": 270, "lr": 4.3689234857622006e-05} +{"train_loss": 0.014789106324315071, "grad_norm": 0.12066632509231567, "global_step": 28635, "epoch": 270, "lr": 4.368629480367843e-05} +{"train_loss": 0.013061266392469406, "grad_norm": 0.07508038729429245, "global_step": 28636, "epoch": 270, "lr": 4.3683354771918455e-05} +{"train_loss": 0.012753423303365707, "grad_norm": 0.07736799865961075, "global_step": 28637, "epoch": 270, "lr": 4.3680414762352435e-05} +{"train_loss": 0.01721106469631195, "grad_norm": 0.09560491144657135, "global_step": 28638, "epoch": 270, "lr": 4.36774747749907e-05} +{"train_loss": 0.016938336193561554, "grad_norm": 0.08501037955284119, "global_step": 28639, "epoch": 270, "lr": 4.367453480984356e-05} +{"train_loss": 0.01452468242496252, "grad_norm": 0.058074139058589935, "global_step": 28640, "epoch": 270, "lr": 4.367159486692134e-05} +{"train_loss": 0.015130002051591873, "grad_norm": 0.08776014298200607, "global_step": 28641, "epoch": 270, "lr": 4.3668654946234395e-05} +{"train_loss": 0.016566498205065727, "grad_norm": 0.08568110316991806, "global_step": 28642, "epoch": 270, "lr": 4.366571504779304e-05} +{"train_loss": 0.014537416398525238, "grad_norm": 0.0780845507979393, "global_step": 28643, "epoch": 270, "lr": 4.36627751716076e-05} +{"train_loss": 0.015695735812187195, "grad_norm": 0.14542244374752045, "global_step": 28644, "epoch": 270, "lr": 4.365983531768845e-05} +{"train_loss": 0.01700824499130249, "grad_norm": 0.07391951233148575, "global_step": 28645, "epoch": 270, "lr": 4.365689548604583e-05} +{"train_loss": 0.015163643285632133, "grad_norm": 0.08223827928304672, "global_step": 28646, "epoch": 270, "lr": 4.365395567669015e-05} +{"train_loss": 0.015953758731484413, "grad_norm": 0.10973711311817169, "global_step": 28647, "epoch": 270, "lr": 4.3651015889631705e-05} +{"train_loss": 0.015298187732696533, "grad_norm": 0.15385767817497253, "global_step": 28648, "epoch": 270, "lr": 4.364807612488082e-05} +{"train_loss": 0.01663646288216114, "grad_norm": 0.13789191842079163, "global_step": 28649, "epoch": 270, "lr": 4.3645136382447836e-05} +{"train_loss": 0.016730070114135742, "grad_norm": 0.06746716797351837, "global_step": 28650, "epoch": 270, "lr": 4.36421966623431e-05} +{"train_loss": 0.014240652322769165, "grad_norm": 0.1312391757965088, "global_step": 28651, "epoch": 270, "lr": 4.36392569645769e-05} +{"train_loss": 0.013607341796159744, "grad_norm": 0.13339881598949432, "global_step": 28652, "epoch": 270, "lr": 4.3636317289159587e-05} +{"train_loss": 0.018477678298950195, "grad_norm": 0.17154358327388763, "global_step": 28653, "epoch": 270, "lr": 4.363337763610149e-05} +{"train_loss": 0.01855715736746788, "grad_norm": 0.12104656547307968, "global_step": 28654, "epoch": 270, "lr": 4.363043800541293e-05} +{"train_loss": 0.014793400652706623, "grad_norm": 0.08878601342439651, "global_step": 28655, "epoch": 270, "lr": 4.362749839710424e-05} +{"train_loss": 0.015681665390729904, "grad_norm": 0.05665847659111023, "global_step": 28656, "epoch": 270, "lr": 4.3624558811185775e-05} +{"train_loss": 0.01446765661239624, "grad_norm": 0.10724978893995285, "global_step": 28657, "epoch": 270, "lr": 4.3621619247667814e-05} +{"train_loss": 0.016163688153028488, "grad_norm": 0.1235087662935257, "global_step": 28658, "epoch": 270, "lr": 4.3618679706560704e-05} +{"train_loss": 0.013936484232544899, "grad_norm": 0.07103781402111053, "global_step": 28659, "epoch": 270, "lr": 4.3615740187874794e-05} +{"train_loss": 0.014017031528055668, "grad_norm": 0.06469572335481644, "global_step": 28660, "epoch": 270, "lr": 4.3612800691620376e-05} +{"train_loss": 0.016369905322790146, "grad_norm": 0.06505842506885529, "global_step": 28661, "epoch": 270, "lr": 4.360986121780781e-05} +{"train_loss": 0.014116507023572922, "grad_norm": 0.17232255637645721, "global_step": 28662, "epoch": 270, "lr": 4.360692176644743e-05} +{"train_loss": 0.015852373093366623, "grad_norm": 0.0704650804400444, "global_step": 28663, "epoch": 270, "lr": 4.360398233754953e-05} +{"train_loss": 0.013564212247729301, "grad_norm": 0.09689764678478241, "global_step": 28664, "epoch": 270, "lr": 4.360104293112446e-05} +{"train_loss": 0.015111910179257393, "grad_norm": 0.12207960337400436, "global_step": 28665, "epoch": 270, "lr": 4.3598103547182525e-05} +{"train_loss": 0.015039484947919846, "grad_norm": 0.0973627045750618, "global_step": 28666, "epoch": 270, "lr": 4.359516418573409e-05} +{"train_loss": 0.01434529572725296, "grad_norm": 0.08189330250024796, "global_step": 28667, "epoch": 270, "lr": 4.3592224846789454e-05} +{"train_loss": 0.016557006165385246, "grad_norm": 0.11940321326255798, "global_step": 28668, "epoch": 270, "lr": 4.358928553035897e-05} +{"train_loss": 0.017407836392521858, "grad_norm": 0.08529611676931381, "global_step": 28669, "epoch": 270, "lr": 4.358634623645292e-05} +{"train_loss": 0.015393375419080257, "grad_norm": 0.14768363535404205, "global_step": 28670, "epoch": 270, "lr": 4.3583406965081676e-05} +{"train_loss": 0.015072648413479328, "grad_norm": 0.09784866124391556, "global_step": 28671, "epoch": 270, "lr": 4.358046771625554e-05} +{"train_loss": 0.014951517805457115, "grad_norm": 0.07140572369098663, "global_step": 28672, "epoch": 270, "lr": 4.3577528489984854e-05} +{"train_loss": 0.014488324522972107, "grad_norm": 0.06614010035991669, "global_step": 28673, "epoch": 270, "lr": 4.357458928627994e-05} +{"train_loss": 0.01619977504014969, "grad_norm": 0.05660534277558327, "global_step": 28674, "epoch": 270, "lr": 4.357165010515114e-05} +{"train_loss": 0.015904167667031288, "grad_norm": 0.10757626593112946, "global_step": 28675, "epoch": 270, "lr": 4.3568710946608754e-05} +{"train_loss": 0.016461890190839767, "grad_norm": 0.060998037457466125, "global_step": 28676, "epoch": 270, "lr": 4.356577181066312e-05} +{"train_loss": 0.014888458885252476, "grad_norm": 0.08903935551643372, "global_step": 28677, "epoch": 270, "lr": 4.356283269732456e-05} +{"train_loss": 0.017617296427488327, "grad_norm": 0.07887780666351318, "global_step": 28678, "epoch": 270, "lr": 4.355989360660341e-05} +{"train_loss": 0.013888117857277393, "grad_norm": 0.1032223328948021, "global_step": 28679, "epoch": 270, "lr": 4.355695453851e-05} +{"train_loss": 0.015170255675911903, "grad_norm": 0.11488890647888184, "global_step": 28680, "epoch": 270, "lr": 4.355401549305465e-05} +{"train_loss": 0.013533435761928558, "grad_norm": 0.09369491040706635, "global_step": 28681, "epoch": 270, "lr": 4.355107647024767e-05} +{"train_loss": 0.014420539140701294, "grad_norm": 0.13058991730213165, "global_step": 28682, "epoch": 270, "lr": 4.3548137470099415e-05} +{"train_loss": 0.01632724516093731, "grad_norm": 0.07667070627212524, "global_step": 28683, "epoch": 270, "lr": 4.35451984926202e-05} +{"train_loss": 0.017445199191570282, "grad_norm": 0.13010747730731964, "global_step": 28684, "epoch": 270, "lr": 4.354225953782034e-05} +{"train_loss": 0.018524769693613052, "grad_norm": 0.28551754355430603, "global_step": 28685, "epoch": 270, "lr": 4.35393206057102e-05} +{"train_loss": 0.014142051339149475, "grad_norm": 0.07580056041479111, "global_step": 28686, "epoch": 270, "lr": 4.3536381696300046e-05} +{"train_loss": 0.017273256555199623, "grad_norm": 0.08936280757188797, "global_step": 28687, "epoch": 270, "lr": 4.3533442809600244e-05} +{"train_loss": 0.012274830602109432, "grad_norm": 0.06482792645692825, "global_step": 28688, "epoch": 270, "lr": 4.353050394562112e-05} +{"train_loss": 0.014824675396084785, "grad_norm": 0.09383077174425125, "global_step": 28689, "epoch": 270, "lr": 4.3527565104372976e-05} +{"train_loss": 0.01501157321035862, "grad_norm": 0.10086795687675476, "global_step": 28690, "epoch": 270, "lr": 4.3524626285866164e-05} +{"train_loss": 0.015169442631304264, "grad_norm": 0.12633998692035675, "global_step": 28691, "epoch": 270, "lr": 4.3521687490111015e-05} +{"train_loss": 0.01697610318660736, "grad_norm": 0.10295940190553665, "global_step": 28692, "epoch": 270, "lr": 4.351874871711782e-05} +{"train_loss": 0.014675173908472061, "grad_norm": 0.07808393239974976, "global_step": 28693, "epoch": 270, "lr": 4.3515809966896924e-05} +{"train_loss": 0.015176918357610703, "grad_norm": 0.10270686447620392, "global_step": 28694, "epoch": 270, "lr": 4.351287123945865e-05} +{"train_loss": 0.01481285784393549, "grad_norm": 0.05199640244245529, "global_step": 28695, "epoch": 270, "lr": 4.3509932534813326e-05} +{"train_loss": 0.014052929356694221, "grad_norm": 0.06388210505247116, "global_step": 28696, "epoch": 270, "lr": 4.350699385297128e-05} +{"train_loss": 0.017336592078208923, "grad_norm": 0.22848787903785706, "global_step": 28697, "epoch": 270, "lr": 4.350405519394285e-05} +{"train_loss": 0.015593403950333595, "grad_norm": 0.09790224581956863, "global_step": 28698, "epoch": 270, "lr": 4.350111655773832e-05} +{"train_loss": 0.014616058208048344, "grad_norm": 0.15802712738513947, "global_step": 28699, "epoch": 270, "lr": 4.3498177944368044e-05} +{"train_loss": 0.013191241770982742, "grad_norm": 0.08748796582221985, "global_step": 28700, "epoch": 270, "lr": 4.349523935384235e-05} +{"train_loss": 0.014548690989613533, "grad_norm": 0.07834520190954208, "global_step": 28701, "epoch": 270, "lr": 4.3492300786171547e-05} +{"train_loss": 0.01542741246521473, "grad_norm": 0.064112089574337, "global_step": 28702, "epoch": 270, "lr": 4.3489362241365975e-05} +{"train_loss": 0.014196740463376045, "grad_norm": 0.0550478920340538, "global_step": 28703, "epoch": 270, "lr": 4.3486423719435966e-05} +{"train_loss": 0.015445372089743614, "grad_norm": 0.09301640838384628, "global_step": 28704, "epoch": 270, "lr": 4.348348522039182e-05} +{"train_loss": 0.014765128493309021, "grad_norm": 0.06397873908281326, "global_step": 28705, "epoch": 270, "lr": 4.348054674424387e-05} +{"train_loss": 0.01651560701429844, "grad_norm": 0.05791627615690231, "global_step": 28706, "epoch": 270, "lr": 4.347760829100244e-05} +{"train_loss": 0.015245147980749607, "grad_norm": 0.08076923340559006, "global_step": 28707, "epoch": 270, "lr": 4.347466986067786e-05} +{"train_loss": 0.01374167762696743, "grad_norm": 0.06375986337661743, "global_step": 28708, "epoch": 270, "lr": 4.347173145328046e-05} +{"train_loss": 0.013887999579310417, "grad_norm": 0.06600922346115112, "global_step": 28709, "epoch": 270, "lr": 4.346879306882056e-05} +{"train_loss": 0.015432722866535187, "grad_norm": 0.08790378272533417, "global_step": 28710, "epoch": 270, "lr": 4.346585470730847e-05} +{"train_loss": 0.016607556492090225, "grad_norm": 0.0782712772488594, "global_step": 28711, "epoch": 270, "lr": 4.3462916368754524e-05} +{"train_loss": 0.014074031263589859, "grad_norm": 0.10091346502304077, "global_step": 28712, "epoch": 270, "lr": 4.345997805316905e-05} +{"train_loss": 0.016717832535505295, "grad_norm": 0.11729162931442261, "global_step": 28713, "epoch": 270, "lr": 4.345703976056237e-05} +{"train_loss": 0.017019469290971756, "grad_norm": 0.0872674360871315, "global_step": 28714, "epoch": 270, "lr": 4.3454101490944806e-05} +{"train_loss": 0.016374439001083374, "grad_norm": 0.07457534223794937, "global_step": 28715, "epoch": 270, "lr": 4.3451163244326695e-05} +{"train_loss": 0.015999101102352142, "grad_norm": 0.22219008207321167, "global_step": 28716, "epoch": 270, "lr": 4.3448225020718335e-05} +{"train_loss": 0.013259951025247574, "grad_norm": 0.22257772088050842, "global_step": 28717, "epoch": 270, "lr": 4.344528682013007e-05} +{"train_loss": 0.015277458354830742, "grad_norm": 0.06185974180698395, "global_step": 28718, "epoch": 270, "lr": 4.34423486425722e-05} +{"train_loss": 0.015999358147382736, "grad_norm": 0.11428412050008774, "global_step": 28719, "epoch": 270, "lr": 4.3439410488055075e-05} +{"train_loss": 0.01567232981324196, "grad_norm": 0.05974419787526131, "global_step": 28720, "epoch": 270, "lr": 4.3436472356589006e-05} +{"train_loss": 0.01779954880475998, "grad_norm": 0.08956971764564514, "global_step": 28721, "epoch": 270, "lr": 4.3433534248184336e-05} +{"train_loss": 0.015370827168226242, "grad_norm": 0.08302546292543411, "global_step": 28722, "epoch": 270, "lr": 4.343059616285136e-05} +{"train_loss": 0.015916874632239342, "grad_norm": 0.06439607590436935, "global_step": 28723, "epoch": 270, "lr": 4.34276581006004e-05} +{"train_loss": 0.01730836182832718, "grad_norm": 0.14353083074092865, "global_step": 28724, "epoch": 270, "lr": 4.3424720061441804e-05} +{"train_loss": 0.015486031674059495, "grad_norm": 0.0685146376490593, "global_step": 28725, "epoch": 270, "lr": 4.342178204538588e-05, "train_batch_mse_error": 0.013791442848742008, "val_loss": 0.01937871240079403} +{"train_loss": 0.018789734691381454, "grad_norm": 0.1150437593460083, "global_step": 28726, "epoch": 271, "lr": 4.3418844052442966e-05} +{"train_loss": 0.017384642735123634, "grad_norm": 0.06572361290454865, "global_step": 28727, "epoch": 271, "lr": 4.341590608262335e-05} +{"train_loss": 0.015562674030661583, "grad_norm": 0.07125550508499146, "global_step": 28728, "epoch": 271, "lr": 4.341296813593738e-05} +{"train_loss": 0.01323073823004961, "grad_norm": 0.06123499944806099, "global_step": 28729, "epoch": 271, "lr": 4.341003021239538e-05} +{"train_loss": 0.01748976856470108, "grad_norm": 0.10927674174308777, "global_step": 28730, "epoch": 271, "lr": 4.340709231200766e-05} +{"train_loss": 0.015687445178627968, "grad_norm": 0.09241097420454025, "global_step": 28731, "epoch": 271, "lr": 4.340415443478456e-05} +{"train_loss": 0.01632658764719963, "grad_norm": 0.08901528269052505, "global_step": 28732, "epoch": 271, "lr": 4.3401216580736405e-05} +{"train_loss": 0.016478156670928, "grad_norm": 0.07239983975887299, "global_step": 28733, "epoch": 271, "lr": 4.339827874987349e-05} +{"train_loss": 0.01793638989329338, "grad_norm": 0.06453298777341843, "global_step": 28734, "epoch": 271, "lr": 4.3395340942206156e-05} +{"train_loss": 0.016198553144931793, "grad_norm": 0.07914620637893677, "global_step": 28735, "epoch": 271, "lr": 4.339240315774471e-05} +{"train_loss": 0.012966667301952839, "grad_norm": 0.10910467058420181, "global_step": 28736, "epoch": 271, "lr": 4.33894653964995e-05} +{"train_loss": 0.016133982688188553, "grad_norm": 0.05543491244316101, "global_step": 28737, "epoch": 271, "lr": 4.338652765848083e-05} +{"train_loss": 0.014973624609410763, "grad_norm": 0.07306285202503204, "global_step": 28738, "epoch": 271, "lr": 4.338358994369903e-05} +{"train_loss": 0.016138698905706406, "grad_norm": 0.18896742165088654, "global_step": 28739, "epoch": 271, "lr": 4.3380652252164404e-05} +{"train_loss": 0.014309396967291832, "grad_norm": 0.08507610857486725, "global_step": 28740, "epoch": 271, "lr": 4.3377714583887296e-05} +{"train_loss": 0.014666660688817501, "grad_norm": 0.0955846756696701, "global_step": 28741, "epoch": 271, "lr": 4.337477693887802e-05} +{"train_loss": 0.014950605109333992, "grad_norm": 0.05798432603478432, "global_step": 28742, "epoch": 271, "lr": 4.3371839317146886e-05} +{"train_loss": 0.014249999076128006, "grad_norm": 0.06396898627281189, "global_step": 28743, "epoch": 271, "lr": 4.336890171870424e-05} +{"train_loss": 0.01490296795964241, "grad_norm": 0.10123047232627869, "global_step": 28744, "epoch": 271, "lr": 4.3365964143560395e-05} +{"train_loss": 0.015396995469927788, "grad_norm": 0.09844996780157089, "global_step": 28745, "epoch": 271, "lr": 4.336302659172566e-05} +{"train_loss": 0.015216031111776829, "grad_norm": 0.10126292705535889, "global_step": 28746, "epoch": 271, "lr": 4.336008906321036e-05} +{"train_loss": 0.015436607412993908, "grad_norm": 0.05811261013150215, "global_step": 28747, "epoch": 271, "lr": 4.335715155802481e-05} +{"train_loss": 0.01449512131512165, "grad_norm": 0.08872276544570923, "global_step": 28748, "epoch": 271, "lr": 4.335421407617935e-05} +{"train_loss": 0.013217592611908913, "grad_norm": 0.06527211517095566, "global_step": 28749, "epoch": 271, "lr": 4.335127661768429e-05} +{"train_loss": 0.016009200364351273, "grad_norm": 0.08539866656064987, "global_step": 28750, "epoch": 271, "lr": 4.3348339182549965e-05} +{"train_loss": 0.012527935206890106, "grad_norm": 0.07099296897649765, "global_step": 28751, "epoch": 271, "lr": 4.3345401770786676e-05} +{"train_loss": 0.01464321929961443, "grad_norm": 0.07161767035722733, "global_step": 28752, "epoch": 271, "lr": 4.3342464382404735e-05} +{"train_loss": 0.014614379033446312, "grad_norm": 0.09303927421569824, "global_step": 28753, "epoch": 271, "lr": 4.33395270174145e-05} +{"train_loss": 0.013560038059949875, "grad_norm": 0.052688032388687134, "global_step": 28754, "epoch": 271, "lr": 4.3336589675826255e-05} +{"train_loss": 0.015008251182734966, "grad_norm": 0.08078455179929733, "global_step": 28755, "epoch": 271, "lr": 4.3333652357650335e-05} +{"train_loss": 0.016614563763141632, "grad_norm": 0.1298048496246338, "global_step": 28756, "epoch": 271, "lr": 4.3330715062897085e-05} +{"train_loss": 0.013970836065709591, "grad_norm": 0.11843349784612656, "global_step": 28757, "epoch": 271, "lr": 4.3327777791576785e-05} +{"train_loss": 0.01690191775560379, "grad_norm": 0.05758257955312729, "global_step": 28758, "epoch": 271, "lr": 4.3324840543699775e-05} +{"train_loss": 0.014183485880494118, "grad_norm": 0.13310964405536652, "global_step": 28759, "epoch": 271, "lr": 4.332190331927636e-05} +{"train_loss": 0.016473708674311638, "grad_norm": 0.09048304706811905, "global_step": 28760, "epoch": 271, "lr": 4.3318966118316886e-05} +{"train_loss": 0.01614067330956459, "grad_norm": 0.08556155115365982, "global_step": 28761, "epoch": 271, "lr": 4.331602894083166e-05} +{"train_loss": 0.017739325761795044, "grad_norm": 0.09554001688957214, "global_step": 28762, "epoch": 271, "lr": 4.331309178683101e-05} +{"train_loss": 0.01701374538242817, "grad_norm": 0.11619161069393158, "global_step": 28763, "epoch": 271, "lr": 4.331015465632523e-05} +{"train_loss": 0.015099856071174145, "grad_norm": 0.06884048879146576, "global_step": 28764, "epoch": 271, "lr": 4.3307217549324655e-05} +{"train_loss": 0.016463058069348335, "grad_norm": 0.07769046723842621, "global_step": 28765, "epoch": 271, "lr": 4.330428046583962e-05} +{"train_loss": 0.01365622691810131, "grad_norm": 0.08772812783718109, "global_step": 28766, "epoch": 271, "lr": 4.330134340588043e-05} +{"train_loss": 0.014359287917613983, "grad_norm": 0.06249620020389557, "global_step": 28767, "epoch": 271, "lr": 4.329840636945739e-05} +{"train_loss": 0.018058203160762787, "grad_norm": 0.09642738848924637, "global_step": 28768, "epoch": 271, "lr": 4.329546935658086e-05} +{"train_loss": 0.015897491946816444, "grad_norm": 0.06166728585958481, "global_step": 28769, "epoch": 271, "lr": 4.3292532367261126e-05} +{"train_loss": 0.015584561973810196, "grad_norm": 0.08122760057449341, "global_step": 28770, "epoch": 271, "lr": 4.328959540150852e-05} +{"train_loss": 0.016332847997546196, "grad_norm": 0.12769442796707153, "global_step": 28771, "epoch": 271, "lr": 4.3286658459333335e-05} +{"train_loss": 0.01845986768603325, "grad_norm": 0.09673572331666946, "global_step": 28772, "epoch": 271, "lr": 4.328372154074593e-05} +{"train_loss": 0.017883099615573883, "grad_norm": 0.06472346186637878, "global_step": 28773, "epoch": 271, "lr": 4.3280784645756605e-05} +{"train_loss": 0.014584851451218128, "grad_norm": 0.04438105225563049, "global_step": 28774, "epoch": 271, "lr": 4.327784777437569e-05} +{"train_loss": 0.015402695164084435, "grad_norm": 0.10701344907283783, "global_step": 28775, "epoch": 271, "lr": 4.327491092661349e-05} +{"train_loss": 0.014997852966189384, "grad_norm": 0.1420903205871582, "global_step": 28776, "epoch": 271, "lr": 4.327197410248032e-05} +{"train_loss": 0.015268473885953426, "grad_norm": 0.09945929795503616, "global_step": 28777, "epoch": 271, "lr": 4.32690373019865e-05} +{"train_loss": 0.01627962850034237, "grad_norm": 0.08490937203168869, "global_step": 28778, "epoch": 271, "lr": 4.326610052514237e-05} +{"train_loss": 0.016308829188346863, "grad_norm": 0.06907633692026138, "global_step": 28779, "epoch": 271, "lr": 4.326316377195825e-05} +{"train_loss": 0.016199812293052673, "grad_norm": 0.06417082995176315, "global_step": 28780, "epoch": 271, "lr": 4.3260227042444425e-05} +{"train_loss": 0.01540575735270977, "grad_norm": 0.07949229329824448, "global_step": 28781, "epoch": 271, "lr": 4.3257290336611214e-05} +{"train_loss": 0.015066217631101608, "grad_norm": 0.06286453455686569, "global_step": 28782, "epoch": 271, "lr": 4.3254353654468974e-05} +{"train_loss": 0.014173520728945732, "grad_norm": 0.09345313161611557, "global_step": 28783, "epoch": 271, "lr": 4.325141699602798e-05} +{"train_loss": 0.01703052595257759, "grad_norm": 0.06719651073217392, "global_step": 28784, "epoch": 271, "lr": 4.324848036129859e-05} +{"train_loss": 0.01632564142346382, "grad_norm": 0.08786125481128693, "global_step": 28785, "epoch": 271, "lr": 4.3245543750291114e-05} +{"train_loss": 0.017142444849014282, "grad_norm": 0.08930903673171997, "global_step": 28786, "epoch": 271, "lr": 4.3242607163015845e-05} +{"train_loss": 0.01393360085785389, "grad_norm": 0.05628194287419319, "global_step": 28787, "epoch": 271, "lr": 4.323967059948312e-05} +{"train_loss": 0.01618249900639057, "grad_norm": 0.07106485962867737, "global_step": 28788, "epoch": 271, "lr": 4.3236734059703245e-05} +{"train_loss": 0.014456013217568398, "grad_norm": 0.192782923579216, "global_step": 28789, "epoch": 271, "lr": 4.323379754368655e-05} +{"train_loss": 0.012924105860292912, "grad_norm": 0.06971436738967896, "global_step": 28790, "epoch": 271, "lr": 4.323086105144335e-05} +{"train_loss": 0.01664106175303459, "grad_norm": 0.0869406908750534, "global_step": 28791, "epoch": 271, "lr": 4.322792458298397e-05} +{"train_loss": 0.014842192642390728, "grad_norm": 0.06666021794080734, "global_step": 28792, "epoch": 271, "lr": 4.322498813831871e-05} +{"train_loss": 0.014503465965390205, "grad_norm": 0.12899638712406158, "global_step": 28793, "epoch": 271, "lr": 4.322205171745788e-05} +{"train_loss": 0.014863536693155766, "grad_norm": 0.1119607537984848, "global_step": 28794, "epoch": 271, "lr": 4.321911532041183e-05} +{"train_loss": 0.016935020685195923, "grad_norm": 0.1296122819185257, "global_step": 28795, "epoch": 271, "lr": 4.3216178947190854e-05} +{"train_loss": 0.015404614619910717, "grad_norm": 0.09006442874670029, "global_step": 28796, "epoch": 271, "lr": 4.321324259780526e-05} +{"train_loss": 0.01522277295589447, "grad_norm": 0.06521360576152802, "global_step": 28797, "epoch": 271, "lr": 4.321030627226541e-05} +{"train_loss": 0.014568082988262177, "grad_norm": 0.05754534900188446, "global_step": 28798, "epoch": 271, "lr": 4.320736997058158e-05} +{"train_loss": 0.013290421105921268, "grad_norm": 0.09770732372999191, "global_step": 28799, "epoch": 271, "lr": 4.3204433692764096e-05} +{"train_loss": 0.01637822762131691, "grad_norm": 0.07521360367536545, "global_step": 28800, "epoch": 271, "lr": 4.320149743882327e-05} +{"train_loss": 0.015433333814144135, "grad_norm": 0.10142914205789566, "global_step": 28801, "epoch": 271, "lr": 4.319856120876943e-05} +{"train_loss": 0.016918115317821503, "grad_norm": 0.08742471039295197, "global_step": 28802, "epoch": 271, "lr": 4.3195625002612885e-05} +{"train_loss": 0.014339236542582512, "grad_norm": 0.06373370438814163, "global_step": 28803, "epoch": 271, "lr": 4.319268882036398e-05} +{"train_loss": 0.014835561625659466, "grad_norm": 0.06253691762685776, "global_step": 28804, "epoch": 271, "lr": 4.3189752662032976e-05} +{"train_loss": 0.015680331736803055, "grad_norm": 0.1046176552772522, "global_step": 28805, "epoch": 271, "lr": 4.318681652763022e-05} +{"train_loss": 0.016504941508173943, "grad_norm": 0.05881650373339653, "global_step": 28806, "epoch": 271, "lr": 4.3183880417166034e-05} +{"train_loss": 0.01605413295328617, "grad_norm": 0.13694748282432556, "global_step": 28807, "epoch": 271, "lr": 4.318094433065073e-05} +{"train_loss": 0.013403009623289108, "grad_norm": 0.07897152006626129, "global_step": 28808, "epoch": 271, "lr": 4.3178008268094616e-05} +{"train_loss": 0.013980070129036903, "grad_norm": 0.07691141217947006, "global_step": 28809, "epoch": 271, "lr": 4.317507222950803e-05} +{"train_loss": 0.013557256199419498, "grad_norm": 0.07707623392343521, "global_step": 28810, "epoch": 271, "lr": 4.317213621490126e-05} +{"train_loss": 0.015983810648322105, "grad_norm": 0.1017366573214531, "global_step": 28811, "epoch": 271, "lr": 4.316920022428464e-05} +{"train_loss": 0.014949694275856018, "grad_norm": 0.07246974110603333, "global_step": 28812, "epoch": 271, "lr": 4.316626425766847e-05} +{"train_loss": 0.014314482919871807, "grad_norm": 0.13166607916355133, "global_step": 28813, "epoch": 271, "lr": 4.316332831506308e-05} +{"train_loss": 0.015461872331798077, "grad_norm": 0.0790814533829689, "global_step": 28814, "epoch": 271, "lr": 4.316039239647878e-05} +{"train_loss": 0.012869341298937798, "grad_norm": 0.06625989079475403, "global_step": 28815, "epoch": 271, "lr": 4.31574565019259e-05} +{"train_loss": 0.014292770065367222, "grad_norm": 0.06926260888576508, "global_step": 28816, "epoch": 271, "lr": 4.315452063141474e-05} +{"train_loss": 0.017179667949676514, "grad_norm": 0.07750408351421356, "global_step": 28817, "epoch": 271, "lr": 4.3151584784955595e-05} +{"train_loss": 0.014595643617212772, "grad_norm": 0.10031700134277344, "global_step": 28818, "epoch": 271, "lr": 4.3148648962558816e-05} +{"train_loss": 0.016905376687645912, "grad_norm": 0.06815527379512787, "global_step": 28819, "epoch": 271, "lr": 4.3145713164234706e-05} +{"train_loss": 0.014753242023289204, "grad_norm": 0.09680857509374619, "global_step": 28820, "epoch": 271, "lr": 4.314277738999357e-05} +{"train_loss": 0.014778289012610912, "grad_norm": 0.0840778797864914, "global_step": 28821, "epoch": 271, "lr": 4.313984163984576e-05} +{"train_loss": 0.014315323904156685, "grad_norm": 0.06310855597257614, "global_step": 28822, "epoch": 271, "lr": 4.3136905913801536e-05} +{"train_loss": 0.018343307077884674, "grad_norm": 0.06308916211128235, "global_step": 28823, "epoch": 271, "lr": 4.3133970211871247e-05} +{"train_loss": 0.01311587356030941, "grad_norm": 0.05785522237420082, "global_step": 28824, "epoch": 271, "lr": 4.313103453406521e-05} +{"train_loss": 0.015504159033298492, "grad_norm": 0.08607766777276993, "global_step": 28825, "epoch": 271, "lr": 4.3128098880393705e-05} +{"train_loss": 0.014314571395516396, "grad_norm": 0.11742792278528214, "global_step": 28826, "epoch": 271, "lr": 4.31251632508671e-05} +{"train_loss": 0.016849935054779053, "grad_norm": 0.0618540495634079, "global_step": 28827, "epoch": 271, "lr": 4.312222764549567e-05} +{"train_loss": 0.015767822042107582, "grad_norm": 0.05756499618291855, "global_step": 28828, "epoch": 271, "lr": 4.3119292064289744e-05} +{"train_loss": 0.018034810200333595, "grad_norm": 0.1166144534945488, "global_step": 28829, "epoch": 271, "lr": 4.311635650725962e-05} +{"train_loss": 0.01381506584584713, "grad_norm": 0.10054589062929153, "global_step": 28830, "epoch": 271, "lr": 4.3113420974415645e-05} +{"train_loss": 0.015436833750737726, "grad_norm": 0.07481833547353745, "global_step": 28831, "epoch": 271, "lr": 4.31104854657681e-05, "val_loss": 0.01946205645799637} +{"train_loss": 0.012107893824577332, "grad_norm": 0.05786966159939766, "global_step": 28832, "epoch": 272, "lr": 4.3107549981327325e-05} +{"train_loss": 0.01719817891716957, "grad_norm": 0.0793478712439537, "global_step": 28833, "epoch": 272, "lr": 4.310461452110362e-05} +{"train_loss": 0.01453230157494545, "grad_norm": 0.08979542553424835, "global_step": 28834, "epoch": 272, "lr": 4.3101679085107286e-05} +{"train_loss": 0.015294499695301056, "grad_norm": 0.08968529850244522, "global_step": 28835, "epoch": 272, "lr": 4.3098743673348664e-05} +{"train_loss": 0.01668800786137581, "grad_norm": 0.08746981620788574, "global_step": 28836, "epoch": 272, "lr": 4.3095808285838055e-05} +{"train_loss": 0.014323342591524124, "grad_norm": 0.11064601689577103, "global_step": 28837, "epoch": 272, "lr": 4.309287292258576e-05} +{"train_loss": 0.018884912133216858, "grad_norm": 0.06520971655845642, "global_step": 28838, "epoch": 272, "lr": 4.3089937583602126e-05} +{"train_loss": 0.01704230345785618, "grad_norm": 0.07340946793556213, "global_step": 28839, "epoch": 272, "lr": 4.3087002268897434e-05} +{"train_loss": 0.014439810998737812, "grad_norm": 0.08528357744216919, "global_step": 28840, "epoch": 272, "lr": 4.3084066978482015e-05} +{"train_loss": 0.013080568984150887, "grad_norm": 0.06499454379081726, "global_step": 28841, "epoch": 272, "lr": 4.308113171236617e-05} +{"train_loss": 0.014551641419529915, "grad_norm": 0.06729467213153839, "global_step": 28842, "epoch": 272, "lr": 4.307819647056022e-05} +{"train_loss": 0.014753386378288269, "grad_norm": 0.053772807121276855, "global_step": 28843, "epoch": 272, "lr": 4.307526125307448e-05} +{"train_loss": 0.01629640907049179, "grad_norm": 0.05952461063861847, "global_step": 28844, "epoch": 272, "lr": 4.3072326059919275e-05} +{"train_loss": 0.01722479611635208, "grad_norm": 0.07430070638656616, "global_step": 28845, "epoch": 272, "lr": 4.3069390891104895e-05} +{"train_loss": 0.018965288996696472, "grad_norm": 0.09045466035604477, "global_step": 28846, "epoch": 272, "lr": 4.306645574664164e-05} +{"train_loss": 0.01525375247001648, "grad_norm": 0.05878862366080284, "global_step": 28847, "epoch": 272, "lr": 4.306352062653986e-05} +{"train_loss": 0.01483725756406784, "grad_norm": 0.15195971727371216, "global_step": 28848, "epoch": 272, "lr": 4.306058553080986e-05} +{"train_loss": 0.016884148120880127, "grad_norm": 0.08100114017724991, "global_step": 28849, "epoch": 272, "lr": 4.305765045946193e-05} +{"train_loss": 0.01925577223300934, "grad_norm": 0.14990168809890747, "global_step": 28850, "epoch": 272, "lr": 4.305471541250642e-05} +{"train_loss": 0.014584597200155258, "grad_norm": 0.046198975294828415, "global_step": 28851, "epoch": 272, "lr": 4.30517803899536e-05} +{"train_loss": 0.015428273007273674, "grad_norm": 0.09056417644023895, "global_step": 28852, "epoch": 272, "lr": 4.304884539181381e-05} +{"train_loss": 0.015385255217552185, "grad_norm": 0.10662931203842163, "global_step": 28853, "epoch": 272, "lr": 4.304591041809735e-05} +{"train_loss": 0.014021489769220352, "grad_norm": 0.08167009055614471, "global_step": 28854, "epoch": 272, "lr": 4.304297546881452e-05} +{"train_loss": 0.015587466768920422, "grad_norm": 0.09937093406915665, "global_step": 28855, "epoch": 272, "lr": 4.304004054397567e-05} +{"train_loss": 0.01598074473440647, "grad_norm": 0.1287238895893097, "global_step": 28856, "epoch": 272, "lr": 4.30371056435911e-05} +{"train_loss": 0.016330065205693245, "grad_norm": 0.07171451300382614, "global_step": 28857, "epoch": 272, "lr": 4.30341707676711e-05} +{"train_loss": 0.014522629790008068, "grad_norm": 0.06953803449869156, "global_step": 28858, "epoch": 272, "lr": 4.303123591622598e-05} +{"train_loss": 0.017819155007600784, "grad_norm": 0.14065219461917877, "global_step": 28859, "epoch": 272, "lr": 4.3028301089266084e-05} +{"train_loss": 0.015313752926886082, "grad_norm": 0.06436395645141602, "global_step": 28860, "epoch": 272, "lr": 4.30253662868017e-05} +{"train_loss": 0.01652245596051216, "grad_norm": 0.07394417375326157, "global_step": 28861, "epoch": 272, "lr": 4.302243150884314e-05} +{"train_loss": 0.015155212953686714, "grad_norm": 0.06079167127609253, "global_step": 28862, "epoch": 272, "lr": 4.301949675540075e-05} +{"train_loss": 0.015742303803563118, "grad_norm": 0.06896871328353882, "global_step": 28863, "epoch": 272, "lr": 4.301656202648478e-05} +{"train_loss": 0.016634082421660423, "grad_norm": 0.11212588101625443, "global_step": 28864, "epoch": 272, "lr": 4.301362732210559e-05} +{"train_loss": 0.015012478455901146, "grad_norm": 0.07209134846925735, "global_step": 28865, "epoch": 272, "lr": 4.301069264227347e-05} +{"train_loss": 0.0168621726334095, "grad_norm": 0.11773180961608887, "global_step": 28866, "epoch": 272, "lr": 4.3007757986998726e-05} +{"train_loss": 0.013401871547102928, "grad_norm": 0.11338002234697342, "global_step": 28867, "epoch": 272, "lr": 4.300482335629169e-05} +{"train_loss": 0.01588883437216282, "grad_norm": 0.06956617534160614, "global_step": 28868, "epoch": 272, "lr": 4.3001888750162674e-05} +{"train_loss": 0.014885448850691319, "grad_norm": 0.11914218962192535, "global_step": 28869, "epoch": 272, "lr": 4.2998954168621966e-05} +{"train_loss": 0.014954264275729656, "grad_norm": 0.06989309936761856, "global_step": 28870, "epoch": 272, "lr": 4.299601961167988e-05} +{"train_loss": 0.01557623315602541, "grad_norm": 0.0702008455991745, "global_step": 28871, "epoch": 272, "lr": 4.299308507934675e-05} +{"train_loss": 0.01624275930225849, "grad_norm": 0.08286242932081223, "global_step": 28872, "epoch": 272, "lr": 4.299015057163287e-05} +{"train_loss": 0.016127925366163254, "grad_norm": 0.10981210321187973, "global_step": 28873, "epoch": 272, "lr": 4.298721608854853e-05} +{"train_loss": 0.015769455581903458, "grad_norm": 0.09278108924627304, "global_step": 28874, "epoch": 272, "lr": 4.2984281630104105e-05} +{"train_loss": 0.01426019798964262, "grad_norm": 0.07815331220626831, "global_step": 28875, "epoch": 272, "lr": 4.2981347196309836e-05} +{"train_loss": 0.014360652305185795, "grad_norm": 0.1255805790424347, "global_step": 28876, "epoch": 272, "lr": 4.2978412787176064e-05} +{"train_loss": 0.014057433232665062, "grad_norm": 0.09738747775554657, "global_step": 28877, "epoch": 272, "lr": 4.2975478402713096e-05} +{"train_loss": 0.015701184049248695, "grad_norm": 0.06968901306390762, "global_step": 28878, "epoch": 272, "lr": 4.297254404293124e-05} +{"train_loss": 0.01662568189203739, "grad_norm": 0.07859436422586441, "global_step": 28879, "epoch": 272, "lr": 4.296960970784082e-05} +{"train_loss": 0.015244530513882637, "grad_norm": 0.0687723457813263, "global_step": 28880, "epoch": 272, "lr": 4.296667539745212e-05} +{"train_loss": 0.014750069938600063, "grad_norm": 0.08622734248638153, "global_step": 28881, "epoch": 272, "lr": 4.2963741111775466e-05} +{"train_loss": 0.013334674760699272, "grad_norm": 0.06378357857465744, "global_step": 28882, "epoch": 272, "lr": 4.296080685082117e-05} +{"train_loss": 0.01846127212047577, "grad_norm": 0.1713755577802658, "global_step": 28883, "epoch": 272, "lr": 4.2957872614599525e-05} +{"train_loss": 0.014157796278595924, "grad_norm": 0.08250418305397034, "global_step": 28884, "epoch": 272, "lr": 4.295493840312087e-05} +{"train_loss": 0.014636335894465446, "grad_norm": 0.115647092461586, "global_step": 28885, "epoch": 272, "lr": 4.29520042163955e-05} +{"train_loss": 0.012516599148511887, "grad_norm": 0.087702177464962, "global_step": 28886, "epoch": 272, "lr": 4.2949070054433715e-05} +{"train_loss": 0.016453444957733154, "grad_norm": 0.0774250403046608, "global_step": 28887, "epoch": 272, "lr": 4.294613591724582e-05} +{"train_loss": 0.015023577958345413, "grad_norm": 0.06568612903356552, "global_step": 28888, "epoch": 272, "lr": 4.294320180484215e-05} +{"train_loss": 0.013678371906280518, "grad_norm": 0.06919417530298233, "global_step": 28889, "epoch": 272, "lr": 4.2940267717233e-05} +{"train_loss": 0.017312131822109222, "grad_norm": 0.09011959284543991, "global_step": 28890, "epoch": 272, "lr": 4.2937333654428656e-05} +{"train_loss": 0.01546940766274929, "grad_norm": 0.08466338366270065, "global_step": 28891, "epoch": 272, "lr": 4.2934399616439494e-05} +{"train_loss": 0.01574144884943962, "grad_norm": 0.11008687317371368, "global_step": 28892, "epoch": 272, "lr": 4.293146560327574e-05} +{"train_loss": 0.01703595370054245, "grad_norm": 0.17656177282333374, "global_step": 28893, "epoch": 272, "lr": 4.292853161494776e-05} +{"train_loss": 0.015680409967899323, "grad_norm": 0.12396032363176346, "global_step": 28894, "epoch": 272, "lr": 4.2925597651465846e-05} +{"train_loss": 0.014627823606133461, "grad_norm": 0.08151854574680328, "global_step": 28895, "epoch": 272, "lr": 4.292266371284028e-05} +{"train_loss": 0.01430545374751091, "grad_norm": 0.11875554919242859, "global_step": 28896, "epoch": 272, "lr": 4.2919729799081424e-05} +{"train_loss": 0.01375892385840416, "grad_norm": 0.08587995916604996, "global_step": 28897, "epoch": 272, "lr": 4.291679591019956e-05} +{"train_loss": 0.01373650785535574, "grad_norm": 0.07464727759361267, "global_step": 28898, "epoch": 272, "lr": 4.291386204620498e-05} +{"train_loss": 0.01514488272368908, "grad_norm": 0.11714004725217819, "global_step": 28899, "epoch": 272, "lr": 4.2910928207108005e-05} +{"train_loss": 0.015586532652378082, "grad_norm": 0.06571022421121597, "global_step": 28900, "epoch": 272, "lr": 4.290799439291895e-05} +{"train_loss": 0.018021326512098312, "grad_norm": 0.07746629416942596, "global_step": 28901, "epoch": 272, "lr": 4.290506060364812e-05} +{"train_loss": 0.015235882252454758, "grad_norm": 0.09047360718250275, "global_step": 28902, "epoch": 272, "lr": 4.290212683930581e-05} +{"train_loss": 0.01616630144417286, "grad_norm": 0.08581464737653732, "global_step": 28903, "epoch": 272, "lr": 4.289919309990238e-05} +{"train_loss": 0.014225395396351814, "grad_norm": 0.07620417326688766, "global_step": 28904, "epoch": 272, "lr": 4.2896259385448054e-05} +{"train_loss": 0.01523161306977272, "grad_norm": 0.08025587350130081, "global_step": 28905, "epoch": 272, "lr": 4.28933256959532e-05} +{"train_loss": 0.013590551912784576, "grad_norm": 0.0934344157576561, "global_step": 28906, "epoch": 272, "lr": 4.289039203142811e-05} +{"train_loss": 0.015907274559140205, "grad_norm": 0.08145445585250854, "global_step": 28907, "epoch": 272, "lr": 4.288745839188308e-05} +{"train_loss": 0.016221964731812477, "grad_norm": 0.07558316737413406, "global_step": 28908, "epoch": 272, "lr": 4.2884524777328436e-05} +{"train_loss": 0.014393179677426815, "grad_norm": 0.07007817178964615, "global_step": 28909, "epoch": 272, "lr": 4.288159118777449e-05} +{"train_loss": 0.015409514307975769, "grad_norm": 0.09310285747051239, "global_step": 28910, "epoch": 272, "lr": 4.2878657623231524e-05} +{"train_loss": 0.012972204945981503, "grad_norm": 0.07769281417131424, "global_step": 28911, "epoch": 272, "lr": 4.287572408370985e-05} +{"train_loss": 0.016928423196077347, "grad_norm": 0.07706429064273834, "global_step": 28912, "epoch": 272, "lr": 4.287279056921979e-05} +{"train_loss": 0.014893152751028538, "grad_norm": 0.09462451189756393, "global_step": 28913, "epoch": 272, "lr": 4.286985707977164e-05} +{"train_loss": 0.016664335504174232, "grad_norm": 0.08442527055740356, "global_step": 28914, "epoch": 272, "lr": 4.286692361537571e-05} +{"train_loss": 0.015846336260437965, "grad_norm": 0.07775966823101044, "global_step": 28915, "epoch": 272, "lr": 4.2863990176042335e-05} +{"train_loss": 0.017246700823307037, "grad_norm": 0.0796603411436081, "global_step": 28916, "epoch": 272, "lr": 4.286105676178177e-05} +{"train_loss": 0.015054884366691113, "grad_norm": 0.07377790659666061, "global_step": 28917, "epoch": 272, "lr": 4.285812337260435e-05} +{"train_loss": 0.014144963584840298, "grad_norm": 0.10393215715885162, "global_step": 28918, "epoch": 272, "lr": 4.285519000852039e-05} +{"train_loss": 0.01720546744763851, "grad_norm": 0.08137425035238266, "global_step": 28919, "epoch": 272, "lr": 4.285225666954016e-05} +{"train_loss": 0.015538722276687622, "grad_norm": 0.09241917729377747, "global_step": 28920, "epoch": 272, "lr": 4.2849323355674006e-05} +{"train_loss": 0.016853485256433487, "grad_norm": 0.09952262043952942, "global_step": 28921, "epoch": 272, "lr": 4.2846390066932235e-05} +{"train_loss": 0.015808939933776855, "grad_norm": 0.12207285314798355, "global_step": 28922, "epoch": 272, "lr": 4.284345680332513e-05} +{"train_loss": 0.0184287428855896, "grad_norm": 0.08137913048267365, "global_step": 28923, "epoch": 272, "lr": 4.2840523564862997e-05} +{"train_loss": 0.01580568589270115, "grad_norm": 0.08386595547199249, "global_step": 28924, "epoch": 272, "lr": 4.283759035155614e-05} +{"train_loss": 0.017099570482969284, "grad_norm": 0.06835035234689713, "global_step": 28925, "epoch": 272, "lr": 4.2834657163414896e-05} +{"train_loss": 0.014588648453354836, "grad_norm": 0.0931345671415329, "global_step": 28926, "epoch": 272, "lr": 4.283172400044954e-05} +{"train_loss": 0.01441293116658926, "grad_norm": 0.07183562219142914, "global_step": 28927, "epoch": 272, "lr": 4.282879086267041e-05} +{"train_loss": 0.014817901886999607, "grad_norm": 0.05738784000277519, "global_step": 28928, "epoch": 272, "lr": 4.282585775008776e-05} +{"train_loss": 0.0143012385815382, "grad_norm": 0.06787025183439255, "global_step": 28929, "epoch": 272, "lr": 4.2822924662711936e-05} +{"train_loss": 0.014873938634991646, "grad_norm": 0.1095690056681633, "global_step": 28930, "epoch": 272, "lr": 4.2819991600553246e-05} +{"train_loss": 0.015760352835059166, "grad_norm": 0.08185385912656784, "global_step": 28931, "epoch": 272, "lr": 4.2817058563621956e-05} +{"train_loss": 0.013517094776034355, "grad_norm": 0.09407417476177216, "global_step": 28932, "epoch": 272, "lr": 4.281412555192844e-05} +{"train_loss": 0.016152502968907356, "grad_norm": 0.08697231858968735, "global_step": 28933, "epoch": 272, "lr": 4.281119256548293e-05} +{"train_loss": 0.013652661815285683, "grad_norm": 0.10332173109054565, "global_step": 28934, "epoch": 272, "lr": 4.280825960429577e-05} +{"train_loss": 0.014242850244045258, "grad_norm": 0.06670967489480972, "global_step": 28935, "epoch": 272, "lr": 4.280532666837726e-05} +{"train_loss": 0.01474094670265913, "grad_norm": 0.08048458397388458, "global_step": 28936, "epoch": 272, "lr": 4.280239375773769e-05} +{"train_loss": 0.015427942370187561, "grad_norm": 0.13020125031471252, "global_step": 28937, "epoch": 272, "lr": 4.2799460872387394e-05, "val_loss": 0.020327599719166756} +{"train_loss": 0.015928642824292183, "grad_norm": 0.05865379422903061, "global_step": 28938, "epoch": 273, "lr": 4.279652801233667e-05} +{"train_loss": 0.013684130273759365, "grad_norm": 0.1552124321460724, "global_step": 28939, "epoch": 273, "lr": 4.2793595177595795e-05} +{"train_loss": 0.013620799407362938, "grad_norm": 0.060152918100357056, "global_step": 28940, "epoch": 273, "lr": 4.2790662368175094e-05} +{"train_loss": 0.016208846122026443, "grad_norm": 0.05824818089604378, "global_step": 28941, "epoch": 273, "lr": 4.2787729584084874e-05} +{"train_loss": 0.01624884642660618, "grad_norm": 0.06944581866264343, "global_step": 28942, "epoch": 273, "lr": 4.278479682533544e-05} +{"train_loss": 0.013820813968777657, "grad_norm": 0.06552563607692719, "global_step": 28943, "epoch": 273, "lr": 4.278186409193707e-05} +{"train_loss": 0.015368646010756493, "grad_norm": 0.06987527012825012, "global_step": 28944, "epoch": 273, "lr": 4.2778931383900136e-05} +{"train_loss": 0.01761305332183838, "grad_norm": 0.07967661321163177, "global_step": 28945, "epoch": 273, "lr": 4.277599870123486e-05} +{"train_loss": 0.013779712840914726, "grad_norm": 0.06821957230567932, "global_step": 28946, "epoch": 273, "lr": 4.277306604395159e-05} +{"train_loss": 0.014456378296017647, "grad_norm": 0.05587854981422424, "global_step": 28947, "epoch": 273, "lr": 4.2770133412060633e-05} +{"train_loss": 0.018441984429955482, "grad_norm": 0.06750322133302689, "global_step": 28948, "epoch": 273, "lr": 4.276720080557226e-05} +{"train_loss": 0.014071608893573284, "grad_norm": 0.08292052894830704, "global_step": 28949, "epoch": 273, "lr": 4.276426822449682e-05} +{"train_loss": 0.013841638341546059, "grad_norm": 0.09292131662368774, "global_step": 28950, "epoch": 273, "lr": 4.27613356688446e-05} +{"train_loss": 0.016034670174121857, "grad_norm": 0.09217531979084015, "global_step": 28951, "epoch": 273, "lr": 4.275840313862589e-05} +{"train_loss": 0.017668385058641434, "grad_norm": 0.08696215599775314, "global_step": 28952, "epoch": 273, "lr": 4.2755470633850996e-05} +{"train_loss": 0.017558474093675613, "grad_norm": 0.09154540300369263, "global_step": 28953, "epoch": 273, "lr": 4.275253815453022e-05} +{"train_loss": 0.0148942731320858, "grad_norm": 0.07385676354169846, "global_step": 28954, "epoch": 273, "lr": 4.274960570067388e-05} +{"train_loss": 0.016843808814883232, "grad_norm": 0.13185887038707733, "global_step": 28955, "epoch": 273, "lr": 4.274667327229228e-05} +{"train_loss": 0.016877511516213417, "grad_norm": 0.12575054168701172, "global_step": 28956, "epoch": 273, "lr": 4.2743740869395726e-05} +{"train_loss": 0.017418254166841507, "grad_norm": 0.08441346138715744, "global_step": 28957, "epoch": 273, "lr": 4.274080849199449e-05} +{"train_loss": 0.014221245422959328, "grad_norm": 0.06177089735865593, "global_step": 28958, "epoch": 273, "lr": 4.27378761400989e-05} +{"train_loss": 0.014391997829079628, "grad_norm": 0.05731669440865517, "global_step": 28959, "epoch": 273, "lr": 4.273494381371926e-05} +{"train_loss": 0.01586028002202511, "grad_norm": 0.08811679482460022, "global_step": 28960, "epoch": 273, "lr": 4.273201151286585e-05} +{"train_loss": 0.014989483170211315, "grad_norm": 0.05845467373728752, "global_step": 28961, "epoch": 273, "lr": 4.272907923754901e-05} +{"train_loss": 0.015891920775175095, "grad_norm": 0.07490700483322144, "global_step": 28962, "epoch": 273, "lr": 4.272614698777903e-05} +{"train_loss": 0.01596725359559059, "grad_norm": 0.06676233559846878, "global_step": 28963, "epoch": 273, "lr": 4.27232147635662e-05} +{"train_loss": 0.016290143132209778, "grad_norm": 0.08666010200977325, "global_step": 28964, "epoch": 273, "lr": 4.272028256492083e-05} +{"train_loss": 0.014695008285343647, "grad_norm": 0.17284032702445984, "global_step": 28965, "epoch": 273, "lr": 4.27173503918532e-05} +{"train_loss": 0.013193827122449875, "grad_norm": 0.06960812211036682, "global_step": 28966, "epoch": 273, "lr": 4.2714418244373655e-05} +{"train_loss": 0.01513394620269537, "grad_norm": 0.1612679362297058, "global_step": 28967, "epoch": 273, "lr": 4.271148612249247e-05} +{"train_loss": 0.014927682466804981, "grad_norm": 0.06942537426948547, "global_step": 28968, "epoch": 273, "lr": 4.2708554026219966e-05} +{"train_loss": 0.012814776971936226, "grad_norm": 0.07091069966554642, "global_step": 28969, "epoch": 273, "lr": 4.2705621955566413e-05} +{"train_loss": 0.01542736031115055, "grad_norm": 0.09194076806306839, "global_step": 28970, "epoch": 273, "lr": 4.270268991054214e-05} +{"train_loss": 0.016011852771043777, "grad_norm": 0.07546248286962509, "global_step": 28971, "epoch": 273, "lr": 4.269975789115744e-05} +{"train_loss": 0.01313849538564682, "grad_norm": 0.058573734015226364, "global_step": 28972, "epoch": 273, "lr": 4.26968258974226e-05} +{"train_loss": 0.013685910031199455, "grad_norm": 0.09542335569858551, "global_step": 28973, "epoch": 273, "lr": 4.269389392934796e-05} +{"train_loss": 0.016147274523973465, "grad_norm": 0.08334290981292725, "global_step": 28974, "epoch": 273, "lr": 4.26909619869438e-05} +{"train_loss": 0.014828350394964218, "grad_norm": 0.08286501467227936, "global_step": 28975, "epoch": 273, "lr": 4.268803007022042e-05} +{"train_loss": 0.014147433452308178, "grad_norm": 0.051197752356529236, "global_step": 28976, "epoch": 273, "lr": 4.2685098179188116e-05} +{"train_loss": 0.01233813725411892, "grad_norm": 0.08791352808475494, "global_step": 28977, "epoch": 273, "lr": 4.2682166313857186e-05} +{"train_loss": 0.017699232324957848, "grad_norm": 0.07604411244392395, "global_step": 28978, "epoch": 273, "lr": 4.267923447423795e-05} +{"train_loss": 0.014709031209349632, "grad_norm": 0.08404066413640976, "global_step": 28979, "epoch": 273, "lr": 4.2676302660340715e-05} +{"train_loss": 0.015370731242001057, "grad_norm": 0.06737720221281052, "global_step": 28980, "epoch": 273, "lr": 4.267337087217575e-05} +{"train_loss": 0.01669168286025524, "grad_norm": 0.09542062133550644, "global_step": 28981, "epoch": 273, "lr": 4.267043910975338e-05} +{"train_loss": 0.015349974855780602, "grad_norm": 0.18561460077762604, "global_step": 28982, "epoch": 273, "lr": 4.266750737308389e-05} +{"train_loss": 0.015591617673635483, "grad_norm": 0.16105671226978302, "global_step": 28983, "epoch": 273, "lr": 4.2664575662177605e-05} +{"train_loss": 0.015463674440979958, "grad_norm": 0.1281389594078064, "global_step": 28984, "epoch": 273, "lr": 4.26616439770448e-05} +{"train_loss": 0.01573564112186432, "grad_norm": 0.09219034761190414, "global_step": 28985, "epoch": 273, "lr": 4.2658712317695805e-05} +{"train_loss": 0.013488495722413063, "grad_norm": 0.07219754159450531, "global_step": 28986, "epoch": 273, "lr": 4.265578068414088e-05} +{"train_loss": 0.016030415892601013, "grad_norm": 0.05219288542866707, "global_step": 28987, "epoch": 273, "lr": 4.265284907639035e-05} +{"train_loss": 0.015025248751044273, "grad_norm": 0.0967194065451622, "global_step": 28988, "epoch": 273, "lr": 4.264991749445452e-05} +{"train_loss": 0.01685408689081669, "grad_norm": 0.06301340460777283, "global_step": 28989, "epoch": 273, "lr": 4.2646985938343674e-05} +{"train_loss": 0.013768441043794155, "grad_norm": 0.05630674213171005, "global_step": 28990, "epoch": 273, "lr": 4.264405440806813e-05} +{"train_loss": 0.01610521785914898, "grad_norm": 0.08628056198358536, "global_step": 28991, "epoch": 273, "lr": 4.2641122903638195e-05} +{"train_loss": 0.015823446214199066, "grad_norm": 0.06805291026830673, "global_step": 28992, "epoch": 273, "lr": 4.2638191425064135e-05} +{"train_loss": 0.016514606773853302, "grad_norm": 0.11120936274528503, "global_step": 28993, "epoch": 273, "lr": 4.263525997235627e-05} +{"train_loss": 0.015437079593539238, "grad_norm": 0.09972231090068817, "global_step": 28994, "epoch": 273, "lr": 4.263232854552489e-05} +{"train_loss": 0.0172329843044281, "grad_norm": 0.14739379286766052, "global_step": 28995, "epoch": 273, "lr": 4.262939714458032e-05} +{"train_loss": 0.015016607940196991, "grad_norm": 0.07121525704860687, "global_step": 28996, "epoch": 273, "lr": 4.2626465769532834e-05} +{"train_loss": 0.013319653458893299, "grad_norm": 0.06240503862500191, "global_step": 28997, "epoch": 273, "lr": 4.262353442039276e-05} +{"train_loss": 0.014886610209941864, "grad_norm": 0.08390186727046967, "global_step": 28998, "epoch": 273, "lr": 4.262060309717035e-05} +{"train_loss": 0.0162361953407526, "grad_norm": 0.09168051183223724, "global_step": 28999, "epoch": 273, "lr": 4.2617671799875944e-05} +{"train_loss": 0.017436757683753967, "grad_norm": 0.1273173689842224, "global_step": 29000, "epoch": 273, "lr": 4.261474052851983e-05} +{"train_loss": 0.01610065996646881, "grad_norm": 0.16280299425125122, "global_step": 29001, "epoch": 273, "lr": 4.2611809283112294e-05} +{"train_loss": 0.013263925909996033, "grad_norm": 0.050441812723875046, "global_step": 29002, "epoch": 273, "lr": 4.260887806366366e-05} +{"train_loss": 0.015985503792762756, "grad_norm": 0.09001802653074265, "global_step": 29003, "epoch": 273, "lr": 4.260594687018422e-05} +{"train_loss": 0.015008402988314629, "grad_norm": 0.13191625475883484, "global_step": 29004, "epoch": 273, "lr": 4.2603015702684255e-05} +{"train_loss": 0.014328476041555405, "grad_norm": 0.12752577662467957, "global_step": 29005, "epoch": 273, "lr": 4.260008456117409e-05} +{"train_loss": 0.012781593948602676, "grad_norm": 0.08604859560728073, "global_step": 29006, "epoch": 273, "lr": 4.2597153445663984e-05} +{"train_loss": 0.01647760346531868, "grad_norm": 0.06425318121910095, "global_step": 29007, "epoch": 273, "lr": 4.2594222356164276e-05} +{"train_loss": 0.016307324171066284, "grad_norm": 0.07522588968276978, "global_step": 29008, "epoch": 273, "lr": 4.259129129268525e-05} +{"train_loss": 0.013228639028966427, "grad_norm": 0.08490745723247528, "global_step": 29009, "epoch": 273, "lr": 4.258836025523722e-05} +{"train_loss": 0.017162669450044632, "grad_norm": 0.1349300742149353, "global_step": 29010, "epoch": 273, "lr": 4.2585429243830455e-05} +{"train_loss": 0.015381284058094025, "grad_norm": 0.10187694430351257, "global_step": 29011, "epoch": 273, "lr": 4.258249825847526e-05} +{"train_loss": 0.013169670477509499, "grad_norm": 0.10607163608074188, "global_step": 29012, "epoch": 273, "lr": 4.257956729918194e-05} +{"train_loss": 0.01858917437493801, "grad_norm": 0.10740862786769867, "global_step": 29013, "epoch": 273, "lr": 4.25766363659608e-05} +{"train_loss": 0.0171361044049263, "grad_norm": 0.09513232856988907, "global_step": 29014, "epoch": 273, "lr": 4.2573705458822124e-05} +{"train_loss": 0.017224973067641258, "grad_norm": 0.15164700150489807, "global_step": 29015, "epoch": 273, "lr": 4.257077457777624e-05} +{"train_loss": 0.016662636771798134, "grad_norm": 0.08624199032783508, "global_step": 29016, "epoch": 273, "lr": 4.256784372283341e-05} +{"train_loss": 0.020652128383517265, "grad_norm": 0.1007721871137619, "global_step": 29017, "epoch": 273, "lr": 4.256491289400394e-05} +{"train_loss": 0.014432378113269806, "grad_norm": 0.08392877131700516, "global_step": 29018, "epoch": 273, "lr": 4.2561982091298125e-05} +{"train_loss": 0.016726067289710045, "grad_norm": 0.12099017947912216, "global_step": 29019, "epoch": 273, "lr": 4.255905131472628e-05} +{"train_loss": 0.014753356575965881, "grad_norm": 0.1537877917289734, "global_step": 29020, "epoch": 273, "lr": 4.2556120564298695e-05} +{"train_loss": 0.01441221870481968, "grad_norm": 0.07580993324518204, "global_step": 29021, "epoch": 273, "lr": 4.2553189840025675e-05} +{"train_loss": 0.019810441881418228, "grad_norm": 0.06783491373062134, "global_step": 29022, "epoch": 273, "lr": 4.2550259141917497e-05} +{"train_loss": 0.014192303642630577, "grad_norm": 0.08187086135149002, "global_step": 29023, "epoch": 273, "lr": 4.254732846998446e-05} +{"train_loss": 0.016894403845071793, "grad_norm": 0.09621985256671906, "global_step": 29024, "epoch": 273, "lr": 4.2544397824236875e-05} +{"train_loss": 0.014084924943745136, "grad_norm": 0.06484324485063553, "global_step": 29025, "epoch": 273, "lr": 4.254146720468504e-05} +{"train_loss": 0.015817422419786453, "grad_norm": 0.10105996578931808, "global_step": 29026, "epoch": 273, "lr": 4.253853661133923e-05} +{"train_loss": 0.01706075295805931, "grad_norm": 0.05384274199604988, "global_step": 29027, "epoch": 273, "lr": 4.253560604420978e-05} +{"train_loss": 0.017502235248684883, "grad_norm": 0.08747703582048416, "global_step": 29028, "epoch": 273, "lr": 4.2532675503306955e-05} +{"train_loss": 0.015439502894878387, "grad_norm": 0.08151740580797195, "global_step": 29029, "epoch": 273, "lr": 4.252974498864106e-05} +{"train_loss": 0.013702332973480225, "grad_norm": 0.15633432567119598, "global_step": 29030, "epoch": 273, "lr": 4.252681450022238e-05} +{"train_loss": 0.013113253749907017, "grad_norm": 0.11183279752731323, "global_step": 29031, "epoch": 273, "lr": 4.252388403806123e-05} +{"train_loss": 0.0157767403870821, "grad_norm": 0.09572528302669525, "global_step": 29032, "epoch": 273, "lr": 4.2520953602167926e-05} +{"train_loss": 0.01479268167167902, "grad_norm": 0.06946151703596115, "global_step": 29033, "epoch": 273, "lr": 4.251802319255271e-05} +{"train_loss": 0.016295332461595535, "grad_norm": 0.1534060388803482, "global_step": 29034, "epoch": 273, "lr": 4.2515092809225923e-05} +{"train_loss": 0.014373620972037315, "grad_norm": 0.08684448897838593, "global_step": 29035, "epoch": 273, "lr": 4.251216245219782e-05} +{"train_loss": 0.016001097857952118, "grad_norm": 0.07781321555376053, "global_step": 29036, "epoch": 273, "lr": 4.2509232121478745e-05} +{"train_loss": 0.012709841132164001, "grad_norm": 0.0536470040678978, "global_step": 29037, "epoch": 273, "lr": 4.250630181707896e-05} +{"train_loss": 0.015126489102840424, "grad_norm": 0.057245347648859024, "global_step": 29038, "epoch": 273, "lr": 4.250337153900879e-05} +{"train_loss": 0.01630576327443123, "grad_norm": 0.1191820278763771, "global_step": 29039, "epoch": 273, "lr": 4.2500441287278505e-05} +{"train_loss": 0.016703270375728607, "grad_norm": 0.08514168858528137, "global_step": 29040, "epoch": 273, "lr": 4.249751106189839e-05} +{"train_loss": 0.014150861650705338, "grad_norm": 0.07455436885356903, "global_step": 29041, "epoch": 273, "lr": 4.249458086287876e-05} +{"train_loss": 0.013901480473577976, "grad_norm": 0.08291790634393692, "global_step": 29042, "epoch": 273, "lr": 4.249165069022991e-05} +{"train_loss": 0.0154236396686789, "grad_norm": 0.0843878909945488, "global_step": 29043, "epoch": 273, "lr": 4.2488720543962146e-05, "val_loss": 0.019575027748942375} +{"train_loss": 0.01517101563513279, "grad_norm": 0.09542392939329147, "global_step": 29044, "epoch": 274, "lr": 4.248579042408576e-05} +{"train_loss": 0.015061499550938606, "grad_norm": 0.09593912959098816, "global_step": 29045, "epoch": 274, "lr": 4.2482860330611023e-05} +{"train_loss": 0.015177519991993904, "grad_norm": 0.13601063191890717, "global_step": 29046, "epoch": 274, "lr": 4.247993026354825e-05} +{"train_loss": 0.01437393482774496, "grad_norm": 0.08020474761724472, "global_step": 29047, "epoch": 274, "lr": 4.247700022290772e-05} +{"train_loss": 0.017198804765939713, "grad_norm": 0.10879165679216385, "global_step": 29048, "epoch": 274, "lr": 4.247407020869975e-05} +{"train_loss": 0.015003768727183342, "grad_norm": 0.158150315284729, "global_step": 29049, "epoch": 274, "lr": 4.247114022093462e-05} +{"train_loss": 0.013758635148406029, "grad_norm": 0.09181197732686996, "global_step": 29050, "epoch": 274, "lr": 4.246821025962265e-05} +{"train_loss": 0.013283595442771912, "grad_norm": 0.08720622956752777, "global_step": 29051, "epoch": 274, "lr": 4.2465280324774096e-05} +{"train_loss": 0.015218909829854965, "grad_norm": 0.09463995695114136, "global_step": 29052, "epoch": 274, "lr": 4.246235041639927e-05} +{"train_loss": 0.01259088609367609, "grad_norm": 0.07313650101423264, "global_step": 29053, "epoch": 274, "lr": 4.245942053450846e-05} +{"train_loss": 0.012577269226312637, "grad_norm": 0.09750209003686905, "global_step": 29054, "epoch": 274, "lr": 4.2456490679111974e-05} +{"train_loss": 0.017198238521814346, "grad_norm": 0.09585599601268768, "global_step": 29055, "epoch": 274, "lr": 4.2453560850220095e-05} +{"train_loss": 0.017793193459510803, "grad_norm": 0.0716155394911766, "global_step": 29056, "epoch": 274, "lr": 4.245063104784314e-05} +{"train_loss": 0.014057258144021034, "grad_norm": 0.09655101597309113, "global_step": 29057, "epoch": 274, "lr": 4.2447701271991366e-05} +{"train_loss": 0.015224861912429333, "grad_norm": 0.10555152595043182, "global_step": 29058, "epoch": 274, "lr": 4.244477152267509e-05} +{"train_loss": 0.013423018157482147, "grad_norm": 0.05520884320139885, "global_step": 29059, "epoch": 274, "lr": 4.244184179990459e-05} +{"train_loss": 0.016178416088223457, "grad_norm": 0.17623378336429596, "global_step": 29060, "epoch": 274, "lr": 4.243891210369018e-05} +{"train_loss": 0.015093682333827019, "grad_norm": 0.0847790464758873, "global_step": 29061, "epoch": 274, "lr": 4.243598243404215e-05} +{"train_loss": 0.01685422472655773, "grad_norm": 0.1176287904381752, "global_step": 29062, "epoch": 274, "lr": 4.24330527909708e-05} +{"train_loss": 0.017720863223075867, "grad_norm": 0.07108443230390549, "global_step": 29063, "epoch": 274, "lr": 4.243012317448639e-05} +{"train_loss": 0.016566447913646698, "grad_norm": 0.06950747221708298, "global_step": 29064, "epoch": 274, "lr": 4.2427193584599237e-05} +{"train_loss": 0.01582983508706093, "grad_norm": 0.10380759835243225, "global_step": 29065, "epoch": 274, "lr": 4.242426402131964e-05} +{"train_loss": 0.017083972692489624, "grad_norm": 0.13115574419498444, "global_step": 29066, "epoch": 274, "lr": 4.242133448465788e-05} +{"train_loss": 0.01621434837579727, "grad_norm": 0.0770440548658371, "global_step": 29067, "epoch": 274, "lr": 4.241840497462424e-05} +{"train_loss": 0.016957547515630722, "grad_norm": 0.10308957099914551, "global_step": 29068, "epoch": 274, "lr": 4.241547549122907e-05} +{"train_loss": 0.013175496831536293, "grad_norm": 0.09851223230361938, "global_step": 29069, "epoch": 274, "lr": 4.241254603448259e-05} +{"train_loss": 0.015094089321792126, "grad_norm": 0.13586336374282837, "global_step": 29070, "epoch": 274, "lr": 4.240961660439513e-05} +{"train_loss": 0.017374247312545776, "grad_norm": 0.15783196687698364, "global_step": 29071, "epoch": 274, "lr": 4.240668720097696e-05} +{"train_loss": 0.013870435766875744, "grad_norm": 0.08816345781087875, "global_step": 29072, "epoch": 274, "lr": 4.240375782423841e-05} +{"train_loss": 0.0129860183224082, "grad_norm": 0.061994317919015884, "global_step": 29073, "epoch": 274, "lr": 4.240082847418974e-05} +{"train_loss": 0.019023558124899864, "grad_norm": 0.12340597808361053, "global_step": 29074, "epoch": 274, "lr": 4.239789915084127e-05} +{"train_loss": 0.014646323397755623, "grad_norm": 0.06156755983829498, "global_step": 29075, "epoch": 274, "lr": 4.239496985420326e-05} +{"train_loss": 0.01616506464779377, "grad_norm": 0.10289890319108963, "global_step": 29076, "epoch": 274, "lr": 4.239204058428602e-05} +{"train_loss": 0.015969425439834595, "grad_norm": 0.09599918127059937, "global_step": 29077, "epoch": 274, "lr": 4.238911134109984e-05} +{"train_loss": 0.017108289524912834, "grad_norm": 0.08560250699520111, "global_step": 29078, "epoch": 274, "lr": 4.238618212465502e-05} +{"train_loss": 0.016328759491443634, "grad_norm": 0.0795338973402977, "global_step": 29079, "epoch": 274, "lr": 4.238325293496186e-05} +{"train_loss": 0.014008995145559311, "grad_norm": 0.06127798929810524, "global_step": 29080, "epoch": 274, "lr": 4.2380323772030614e-05} +{"train_loss": 0.014191818423569202, "grad_norm": 0.0676162838935852, "global_step": 29081, "epoch": 274, "lr": 4.237739463587159e-05} +{"train_loss": 0.013215264305472374, "grad_norm": 0.21436136960983276, "global_step": 29082, "epoch": 274, "lr": 4.23744655264951e-05} +{"train_loss": 0.013265167362987995, "grad_norm": 0.06301908940076828, "global_step": 29083, "epoch": 274, "lr": 4.2371536443911425e-05} +{"train_loss": 0.011968339793384075, "grad_norm": 0.11288465559482574, "global_step": 29084, "epoch": 274, "lr": 4.236860738813084e-05} +{"train_loss": 0.015255836769938469, "grad_norm": 0.11303803324699402, "global_step": 29085, "epoch": 274, "lr": 4.236567835916366e-05} +{"train_loss": 0.01627987064421177, "grad_norm": 0.09345447272062302, "global_step": 29086, "epoch": 274, "lr": 4.236274935702017e-05} +{"train_loss": 0.013753268867731094, "grad_norm": 0.07272511720657349, "global_step": 29087, "epoch": 274, "lr": 4.235982038171065e-05} +{"train_loss": 0.016697239130735397, "grad_norm": 0.10523463040590286, "global_step": 29088, "epoch": 274, "lr": 4.2356891433245395e-05} +{"train_loss": 0.016875997185707092, "grad_norm": 0.1269947588443756, "global_step": 29089, "epoch": 274, "lr": 4.2353962511634704e-05} +{"train_loss": 0.016032561659812927, "grad_norm": 0.12408751994371414, "global_step": 29090, "epoch": 274, "lr": 4.235103361688886e-05} +{"train_loss": 0.01606207713484764, "grad_norm": 0.08228456228971481, "global_step": 29091, "epoch": 274, "lr": 4.234810474901818e-05} +{"train_loss": 0.017032651230692863, "grad_norm": 0.10793229192495346, "global_step": 29092, "epoch": 274, "lr": 4.234517590803291e-05} +{"train_loss": 0.013540472835302353, "grad_norm": 0.06511960178613663, "global_step": 29093, "epoch": 274, "lr": 4.234224709394336e-05} +{"train_loss": 0.012316273525357246, "grad_norm": 0.05423564463853836, "global_step": 29094, "epoch": 274, "lr": 4.233931830675983e-05} +{"train_loss": 0.015167335979640484, "grad_norm": 0.06332050263881683, "global_step": 29095, "epoch": 274, "lr": 4.2336389546492606e-05} +{"train_loss": 0.01700436696410179, "grad_norm": 0.09330284595489502, "global_step": 29096, "epoch": 274, "lr": 4.233346081315196e-05} +{"train_loss": 0.01587037183344364, "grad_norm": 0.10306402295827866, "global_step": 29097, "epoch": 274, "lr": 4.2330532106748227e-05} +{"train_loss": 0.016322718933224678, "grad_norm": 0.07251071184873581, "global_step": 29098, "epoch": 274, "lr": 4.232760342729166e-05} +{"train_loss": 0.01644607074558735, "grad_norm": 0.08479426801204681, "global_step": 29099, "epoch": 274, "lr": 4.232467477479255e-05} +{"train_loss": 0.017028216272592545, "grad_norm": 0.08472151309251785, "global_step": 29100, "epoch": 274, "lr": 4.232174614926119e-05} +{"train_loss": 0.012824198231101036, "grad_norm": 0.08673352003097534, "global_step": 29101, "epoch": 274, "lr": 4.231881755070788e-05} +{"train_loss": 0.01389443501830101, "grad_norm": 0.08107903599739075, "global_step": 29102, "epoch": 274, "lr": 4.231588897914291e-05} +{"train_loss": 0.013619001023471355, "grad_norm": 0.1144237145781517, "global_step": 29103, "epoch": 274, "lr": 4.231296043457658e-05} +{"train_loss": 0.016991060227155685, "grad_norm": 0.09826664626598358, "global_step": 29104, "epoch": 274, "lr": 4.231003191701914e-05} +{"train_loss": 0.016601519659161568, "grad_norm": 0.1240740492939949, "global_step": 29105, "epoch": 274, "lr": 4.23071034264809e-05} +{"train_loss": 0.015858108177781105, "grad_norm": 0.1554030030965805, "global_step": 29106, "epoch": 274, "lr": 4.2304174962972164e-05} +{"train_loss": 0.01697036251425743, "grad_norm": 0.2081712782382965, "global_step": 29107, "epoch": 274, "lr": 4.2301246526503205e-05} +{"train_loss": 0.015069299377501011, "grad_norm": 0.09454609453678131, "global_step": 29108, "epoch": 274, "lr": 4.229831811708431e-05} +{"train_loss": 0.016405615955591202, "grad_norm": 0.07853465527296066, "global_step": 29109, "epoch": 274, "lr": 4.229538973472581e-05} +{"train_loss": 0.014785418286919594, "grad_norm": 0.075750432908535, "global_step": 29110, "epoch": 274, "lr": 4.229246137943793e-05} +{"train_loss": 0.014617894776165485, "grad_norm": 0.05708698183298111, "global_step": 29111, "epoch": 274, "lr": 4.2289533051231e-05} +{"train_loss": 0.01503774058073759, "grad_norm": 0.06251616775989532, "global_step": 29112, "epoch": 274, "lr": 4.228660475011529e-05} +{"train_loss": 0.015870539471507072, "grad_norm": 0.09138116985559464, "global_step": 29113, "epoch": 274, "lr": 4.228367647610109e-05} +{"train_loss": 0.0161482784897089, "grad_norm": 0.08510932326316833, "global_step": 29114, "epoch": 274, "lr": 4.228074822919871e-05} +{"train_loss": 0.014646276831626892, "grad_norm": 0.15080967545509338, "global_step": 29115, "epoch": 274, "lr": 4.227782000941842e-05} +{"train_loss": 0.014111768454313278, "grad_norm": 0.06315100193023682, "global_step": 29116, "epoch": 274, "lr": 4.227489181677052e-05} +{"train_loss": 0.014297340996563435, "grad_norm": 0.10909181088209152, "global_step": 29117, "epoch": 274, "lr": 4.2271963651265266e-05} +{"train_loss": 0.017739389091730118, "grad_norm": 0.1307784616947174, "global_step": 29118, "epoch": 274, "lr": 4.226903551291299e-05} +{"train_loss": 0.014044409617781639, "grad_norm": 0.07435973733663559, "global_step": 29119, "epoch": 274, "lr": 4.226610740172396e-05} +{"train_loss": 0.017934352159500122, "grad_norm": 0.054483577609062195, "global_step": 29120, "epoch": 274, "lr": 4.226317931770845e-05} +{"train_loss": 0.017405612394213676, "grad_norm": 0.08695616573095322, "global_step": 29121, "epoch": 274, "lr": 4.22602512608768e-05} +{"train_loss": 0.01718129776418209, "grad_norm": 0.12474953383207321, "global_step": 29122, "epoch": 274, "lr": 4.2257323231239225e-05} +{"train_loss": 0.016128117218613625, "grad_norm": 0.1136712059378624, "global_step": 29123, "epoch": 274, "lr": 4.225439522880606e-05} +{"train_loss": 0.01656920462846756, "grad_norm": 0.07332849502563477, "global_step": 29124, "epoch": 274, "lr": 4.225146725358759e-05} +{"train_loss": 0.014796745032072067, "grad_norm": 0.0604020357131958, "global_step": 29125, "epoch": 274, "lr": 4.224853930559407e-05} +{"train_loss": 0.014653106220066547, "grad_norm": 0.17849290370941162, "global_step": 29126, "epoch": 274, "lr": 4.224561138483583e-05} +{"train_loss": 0.014488275162875652, "grad_norm": 0.06559272855520248, "global_step": 29127, "epoch": 274, "lr": 4.2242683491323157e-05} +{"train_loss": 0.01379341259598732, "grad_norm": 0.06572260707616806, "global_step": 29128, "epoch": 274, "lr": 4.2239755625066295e-05} +{"train_loss": 0.015021100640296936, "grad_norm": 0.07381989806890488, "global_step": 29129, "epoch": 274, "lr": 4.223682778607555e-05} +{"train_loss": 0.018731333315372467, "grad_norm": 0.12276949733495712, "global_step": 29130, "epoch": 274, "lr": 4.2233899974361237e-05} +{"train_loss": 0.015192187391221523, "grad_norm": 0.08649533987045288, "global_step": 29131, "epoch": 274, "lr": 4.223097218993361e-05} +{"train_loss": 0.018528155982494354, "grad_norm": 0.06109686940908432, "global_step": 29132, "epoch": 274, "lr": 4.222804443280299e-05} +{"train_loss": 0.015623560175299644, "grad_norm": 0.11351198703050613, "global_step": 29133, "epoch": 274, "lr": 4.222511670297962e-05} +{"train_loss": 0.01913543976843357, "grad_norm": 0.058033425360918045, "global_step": 29134, "epoch": 274, "lr": 4.2222189000473803e-05} +{"train_loss": 0.014910725876688957, "grad_norm": 0.0665932223200798, "global_step": 29135, "epoch": 274, "lr": 4.2219261325295845e-05} +{"train_loss": 0.013140913099050522, "grad_norm": 0.10640604048967361, "global_step": 29136, "epoch": 274, "lr": 4.2216333677456024e-05} +{"train_loss": 0.013493754900991917, "grad_norm": 0.0917159914970398, "global_step": 29137, "epoch": 274, "lr": 4.22134060569646e-05} +{"train_loss": 0.01667354255914688, "grad_norm": 0.11348924040794373, "global_step": 29138, "epoch": 274, "lr": 4.221047846383191e-05} +{"train_loss": 0.017275771126151085, "grad_norm": 0.07824922353029251, "global_step": 29139, "epoch": 274, "lr": 4.220755089806819e-05} +{"train_loss": 0.013477972708642483, "grad_norm": 0.05638829991221428, "global_step": 29140, "epoch": 274, "lr": 4.2204623359683754e-05} +{"train_loss": 0.016624636948108673, "grad_norm": 0.09132009744644165, "global_step": 29141, "epoch": 274, "lr": 4.2201695848688875e-05} +{"train_loss": 0.01876775175333023, "grad_norm": 0.09440567344427109, "global_step": 29142, "epoch": 274, "lr": 4.2198768365093846e-05} +{"train_loss": 0.015410509891808033, "grad_norm": 0.09237529337406158, "global_step": 29143, "epoch": 274, "lr": 4.219584090890896e-05} +{"train_loss": 0.01215268298983574, "grad_norm": 0.06673303991556168, "global_step": 29144, "epoch": 274, "lr": 4.21929134801445e-05} +{"train_loss": 0.016085153445601463, "grad_norm": 0.10045374929904938, "global_step": 29145, "epoch": 274, "lr": 4.218998607881074e-05} +{"train_loss": 0.014499098062515259, "grad_norm": 0.07546558976173401, "global_step": 29146, "epoch": 274, "lr": 4.2187058704917956e-05} +{"train_loss": 0.013883806765079498, "grad_norm": 0.08722126483917236, "global_step": 29147, "epoch": 274, "lr": 4.2184131358476466e-05} +{"train_loss": 0.015963779762387276, "grad_norm": 0.08900865912437439, "global_step": 29148, "epoch": 274, "lr": 4.218120403949654e-05} +{"train_loss": 0.015436451319815978, "grad_norm": 0.09510314464569092, "global_step": 29149, "epoch": 274, "lr": 4.2178276747988446e-05, "val_loss": 0.019461261108517647} +{"train_loss": 0.01544621679931879, "grad_norm": 0.07438641041517258, "global_step": 29150, "epoch": 275, "lr": 4.217534948396253e-05} +{"train_loss": 0.014429251663386822, "grad_norm": 0.09502341598272324, "global_step": 29151, "epoch": 275, "lr": 4.2172422247428986e-05} +{"train_loss": 0.014461357146501541, "grad_norm": 0.05243165045976639, "global_step": 29152, "epoch": 275, "lr": 4.216949503839817e-05} +{"train_loss": 0.015857793390750885, "grad_norm": 0.06267325580120087, "global_step": 29153, "epoch": 275, "lr": 4.216656785688034e-05} +{"train_loss": 0.01585395447909832, "grad_norm": 0.14804530143737793, "global_step": 29154, "epoch": 275, "lr": 4.216364070288577e-05} +{"train_loss": 0.01706540212035179, "grad_norm": 0.06904072314500809, "global_step": 29155, "epoch": 275, "lr": 4.216071357642477e-05} +{"train_loss": 0.014229918830096722, "grad_norm": 0.06263057887554169, "global_step": 29156, "epoch": 275, "lr": 4.215778647750762e-05} +{"train_loss": 0.016713161021471024, "grad_norm": 0.05481409281492233, "global_step": 29157, "epoch": 275, "lr": 4.21548594061446e-05} +{"train_loss": 0.012930132448673248, "grad_norm": 0.09043888002634048, "global_step": 29158, "epoch": 275, "lr": 4.215193236234597e-05} +{"train_loss": 0.014696056954562664, "grad_norm": 0.1083165854215622, "global_step": 29159, "epoch": 275, "lr": 4.214900534612205e-05} +{"train_loss": 0.01567414402961731, "grad_norm": 0.12851282954216003, "global_step": 29160, "epoch": 275, "lr": 4.214607835748311e-05} +{"train_loss": 0.01595817692577839, "grad_norm": 0.10463451594114304, "global_step": 29161, "epoch": 275, "lr": 4.2143151396439425e-05} +{"train_loss": 0.015120865777134895, "grad_norm": 0.08573631942272186, "global_step": 29162, "epoch": 275, "lr": 4.214022446300132e-05} +{"train_loss": 0.017730409279465675, "grad_norm": 0.09333706647157669, "global_step": 29163, "epoch": 275, "lr": 4.2137297557179024e-05} +{"train_loss": 0.0137464739382267, "grad_norm": 0.07708395272493362, "global_step": 29164, "epoch": 275, "lr": 4.213437067898285e-05} +{"train_loss": 0.01678633503615856, "grad_norm": 0.08187355846166611, "global_step": 29165, "epoch": 275, "lr": 4.213144382842307e-05} +{"train_loss": 0.01902320794761181, "grad_norm": 0.10598893463611603, "global_step": 29166, "epoch": 275, "lr": 4.212851700550997e-05} +{"train_loss": 0.01718776300549507, "grad_norm": 0.08136714994907379, "global_step": 29167, "epoch": 275, "lr": 4.2125590210253853e-05} +{"train_loss": 0.016212081536650658, "grad_norm": 0.07785273343324661, "global_step": 29168, "epoch": 275, "lr": 4.212266344266499e-05} +{"train_loss": 0.016592849045991898, "grad_norm": 0.08388692140579224, "global_step": 29169, "epoch": 275, "lr": 4.211973670275365e-05} +{"train_loss": 0.015451726503670216, "grad_norm": 0.06137784942984581, "global_step": 29170, "epoch": 275, "lr": 4.2116809990530135e-05} +{"train_loss": 0.013923274353146553, "grad_norm": 0.06696122139692307, "global_step": 29171, "epoch": 275, "lr": 4.2113883306004717e-05} +{"train_loss": 0.016692718490958214, "grad_norm": 0.0693054348230362, "global_step": 29172, "epoch": 275, "lr": 4.2110956649187684e-05} +{"train_loss": 0.015097007155418396, "grad_norm": 0.09095071256160736, "global_step": 29173, "epoch": 275, "lr": 4.2108030020089303e-05} +{"train_loss": 0.015818826854228973, "grad_norm": 0.06724540144205093, "global_step": 29174, "epoch": 275, "lr": 4.2105103418719915e-05} +{"train_loss": 0.014620412141084671, "grad_norm": 0.07997658103704453, "global_step": 29175, "epoch": 275, "lr": 4.2102176845089724e-05} +{"train_loss": 0.015196292661130428, "grad_norm": 0.07917221635580063, "global_step": 29176, "epoch": 275, "lr": 4.2099250299209064e-05} +{"train_loss": 0.01935063675045967, "grad_norm": 0.19451452791690826, "global_step": 29177, "epoch": 275, "lr": 4.2096323781088195e-05} +{"train_loss": 0.014170730486512184, "grad_norm": 0.1295395940542221, "global_step": 29178, "epoch": 275, "lr": 4.20933972907374e-05} +{"train_loss": 0.01324152946472168, "grad_norm": 0.09452158957719803, "global_step": 29179, "epoch": 275, "lr": 4.209047082816699e-05} +{"train_loss": 0.01471647247672081, "grad_norm": 0.0891806036233902, "global_step": 29180, "epoch": 275, "lr": 4.208754439338722e-05} +{"train_loss": 0.016845185309648514, "grad_norm": 0.0833572968840599, "global_step": 29181, "epoch": 275, "lr": 4.208461798640837e-05} +{"train_loss": 0.017458384856581688, "grad_norm": 0.057155828922986984, "global_step": 29182, "epoch": 275, "lr": 4.2081691607240735e-05} +{"train_loss": 0.014731625095009804, "grad_norm": 0.10452067106962204, "global_step": 29183, "epoch": 275, "lr": 4.2078765255894574e-05} +{"train_loss": 0.014942601323127747, "grad_norm": 0.056269410997629166, "global_step": 29184, "epoch": 275, "lr": 4.207583893238021e-05} +{"train_loss": 0.01725960522890091, "grad_norm": 0.11176096647977829, "global_step": 29185, "epoch": 275, "lr": 4.20729126367079e-05} +{"train_loss": 0.014993364922702312, "grad_norm": 0.08399399369955063, "global_step": 29186, "epoch": 275, "lr": 4.206998636888793e-05} +{"train_loss": 0.016805890947580338, "grad_norm": 0.09499887377023697, "global_step": 29187, "epoch": 275, "lr": 4.206706012893056e-05} +{"train_loss": 0.01313125342130661, "grad_norm": 0.07652140408754349, "global_step": 29188, "epoch": 275, "lr": 4.20641339168461e-05} +{"train_loss": 0.015887105837464333, "grad_norm": 0.11183665692806244, "global_step": 29189, "epoch": 275, "lr": 4.206120773264482e-05} +{"train_loss": 0.015015000477433205, "grad_norm": 0.1340070217847824, "global_step": 29190, "epoch": 275, "lr": 4.2058281576337e-05} +{"train_loss": 0.01570645533502102, "grad_norm": 0.12899932265281677, "global_step": 29191, "epoch": 275, "lr": 4.205535544793295e-05} +{"train_loss": 0.014965632930397987, "grad_norm": 0.07609520107507706, "global_step": 29192, "epoch": 275, "lr": 4.20524293474429e-05} +{"train_loss": 0.013389014638960361, "grad_norm": 0.11212348937988281, "global_step": 29193, "epoch": 275, "lr": 4.2049503274877166e-05} +{"train_loss": 0.012365235947072506, "grad_norm": 0.09044864028692245, "global_step": 29194, "epoch": 275, "lr": 4.204657723024602e-05} +{"train_loss": 0.01522837020456791, "grad_norm": 0.09469648450613022, "global_step": 29195, "epoch": 275, "lr": 4.204365121355973e-05} +{"train_loss": 0.014807539992034435, "grad_norm": 0.07489396631717682, "global_step": 29196, "epoch": 275, "lr": 4.2040725224828606e-05} +{"train_loss": 0.014865281991660595, "grad_norm": 0.08166851103305817, "global_step": 29197, "epoch": 275, "lr": 4.2037799264062924e-05} +{"train_loss": 0.01319311372935772, "grad_norm": 0.07899013161659241, "global_step": 29198, "epoch": 275, "lr": 4.203487333127293e-05} +{"train_loss": 0.01696641743183136, "grad_norm": 0.1087363138794899, "global_step": 29199, "epoch": 275, "lr": 4.203194742646893e-05} +{"train_loss": 0.014294328168034554, "grad_norm": 0.07160230726003647, "global_step": 29200, "epoch": 275, "lr": 4.20290215496612e-05} +{"train_loss": 0.01657027006149292, "grad_norm": 0.10098594427108765, "global_step": 29201, "epoch": 275, "lr": 4.202609570086003e-05} +{"train_loss": 0.017468079924583435, "grad_norm": 0.11133364588022232, "global_step": 29202, "epoch": 275, "lr": 4.202316988007567e-05} +{"train_loss": 0.016233637928962708, "grad_norm": 0.18178284168243408, "global_step": 29203, "epoch": 275, "lr": 4.202024408731847e-05} +{"train_loss": 0.01343938522040844, "grad_norm": 0.07635259628295898, "global_step": 29204, "epoch": 275, "lr": 4.201731832259863e-05} +{"train_loss": 0.013349184766411781, "grad_norm": 0.06955300271511078, "global_step": 29205, "epoch": 275, "lr": 4.201439258592646e-05} +{"train_loss": 0.013756411150097847, "grad_norm": 0.09257682412862778, "global_step": 29206, "epoch": 275, "lr": 4.2011466877312243e-05} +{"train_loss": 0.014216626062989235, "grad_norm": 0.07180868089199066, "global_step": 29207, "epoch": 275, "lr": 4.200854119676626e-05} +{"train_loss": 0.013777215965092182, "grad_norm": 0.05416443571448326, "global_step": 29208, "epoch": 275, "lr": 4.200561554429878e-05} +{"train_loss": 0.017069566994905472, "grad_norm": 0.07071294635534286, "global_step": 29209, "epoch": 275, "lr": 4.200268991992011e-05} +{"train_loss": 0.016164761036634445, "grad_norm": 0.06362681090831757, "global_step": 29210, "epoch": 275, "lr": 4.19997643236405e-05} +{"train_loss": 0.016866974532604218, "grad_norm": 0.06406354904174805, "global_step": 29211, "epoch": 275, "lr": 4.199683875547024e-05} +{"train_loss": 0.01401899941265583, "grad_norm": 0.11160014569759369, "global_step": 29212, "epoch": 275, "lr": 4.1993913215419594e-05} +{"train_loss": 0.014806948602199554, "grad_norm": 0.10515757650136948, "global_step": 29213, "epoch": 275, "lr": 4.199098770349887e-05} +{"train_loss": 0.01580875553190708, "grad_norm": 0.08084625750780106, "global_step": 29214, "epoch": 275, "lr": 4.198806221971833e-05} +{"train_loss": 0.01581384800374508, "grad_norm": 0.163066104054451, "global_step": 29215, "epoch": 275, "lr": 4.198513676408826e-05} +{"train_loss": 0.014832677319645882, "grad_norm": 0.13948963582515717, "global_step": 29216, "epoch": 275, "lr": 4.198221133661892e-05} +{"train_loss": 0.013771135360002518, "grad_norm": 0.08221707493066788, "global_step": 29217, "epoch": 275, "lr": 4.1979285937320614e-05} +{"train_loss": 0.01642349734902382, "grad_norm": 0.11143527179956436, "global_step": 29218, "epoch": 275, "lr": 4.19763605662036e-05} +{"train_loss": 0.01938321441411972, "grad_norm": 0.08795685321092606, "global_step": 29219, "epoch": 275, "lr": 4.1973435223278166e-05} +{"train_loss": 0.012016495689749718, "grad_norm": 0.0466141514480114, "global_step": 29220, "epoch": 275, "lr": 4.197050990855459e-05} +{"train_loss": 0.012913379818201065, "grad_norm": 0.05343818664550781, "global_step": 29221, "epoch": 275, "lr": 4.1967584622043165e-05} +{"train_loss": 0.01491180807352066, "grad_norm": 0.06368301808834076, "global_step": 29222, "epoch": 275, "lr": 4.196465936375415e-05} +{"train_loss": 0.01683729514479637, "grad_norm": 0.05504599213600159, "global_step": 29223, "epoch": 275, "lr": 4.196173413369782e-05} +{"train_loss": 0.016389118507504463, "grad_norm": 0.15869949758052826, "global_step": 29224, "epoch": 275, "lr": 4.195880893188445e-05} +{"train_loss": 0.015733011066913605, "grad_norm": 0.06304576992988586, "global_step": 29225, "epoch": 275, "lr": 4.195588375832434e-05} +{"train_loss": 0.016183042898774147, "grad_norm": 0.060089074075222015, "global_step": 29226, "epoch": 275, "lr": 4.1952958613027764e-05} +{"train_loss": 0.015853166580200195, "grad_norm": 0.06513327360153198, "global_step": 29227, "epoch": 275, "lr": 4.1950033496005e-05} +{"train_loss": 0.013461463153362274, "grad_norm": 0.05926833301782608, "global_step": 29228, "epoch": 275, "lr": 4.194710840726629e-05} +{"train_loss": 0.01429867185652256, "grad_norm": 0.07356845587491989, "global_step": 29229, "epoch": 275, "lr": 4.1944183346821954e-05} +{"train_loss": 0.014770649373531342, "grad_norm": 0.07995186746120453, "global_step": 29230, "epoch": 275, "lr": 4.194125831468225e-05} +{"train_loss": 0.014507549814879894, "grad_norm": 0.08893211930990219, "global_step": 29231, "epoch": 275, "lr": 4.193833331085745e-05} +{"train_loss": 0.015125837177038193, "grad_norm": 0.08852805942296982, "global_step": 29232, "epoch": 275, "lr": 4.1935408335357874e-05} +{"train_loss": 0.017005035653710365, "grad_norm": 0.11809355020523071, "global_step": 29233, "epoch": 275, "lr": 4.1932483388193735e-05} +{"train_loss": 0.016016844660043716, "grad_norm": 0.08989783376455307, "global_step": 29234, "epoch": 275, "lr": 4.192955846937535e-05} +{"train_loss": 0.01403086632490158, "grad_norm": 0.06930866837501526, "global_step": 29235, "epoch": 275, "lr": 4.1926633578912985e-05} +{"train_loss": 0.016095075756311417, "grad_norm": 0.0716019794344902, "global_step": 29236, "epoch": 275, "lr": 4.1923708716816914e-05} +{"train_loss": 0.015218173153698444, "grad_norm": 0.1363941878080368, "global_step": 29237, "epoch": 275, "lr": 4.192078388309743e-05} +{"train_loss": 0.014845007099211216, "grad_norm": 0.07089709490537643, "global_step": 29238, "epoch": 275, "lr": 4.1917859077764796e-05} +{"train_loss": 0.015435739420354366, "grad_norm": 0.07122445851564407, "global_step": 29239, "epoch": 275, "lr": 4.191493430082929e-05} +{"train_loss": 0.0155642656609416, "grad_norm": 0.09401379525661469, "global_step": 29240, "epoch": 275, "lr": 4.1912009552301185e-05} +{"train_loss": 0.01488959789276123, "grad_norm": 0.1079958826303482, "global_step": 29241, "epoch": 275, "lr": 4.190908483219075e-05} +{"train_loss": 0.014379632659256458, "grad_norm": 0.08789148926734924, "global_step": 29242, "epoch": 275, "lr": 4.190616014050829e-05} +{"train_loss": 0.016324803233146667, "grad_norm": 0.06949430704116821, "global_step": 29243, "epoch": 275, "lr": 4.190323547726406e-05} +{"train_loss": 0.017501041293144226, "grad_norm": 0.11757589876651764, "global_step": 29244, "epoch": 275, "lr": 4.1900310842468344e-05} +{"train_loss": 0.01742219552397728, "grad_norm": 0.07833985239267349, "global_step": 29245, "epoch": 275, "lr": 4.18973862361314e-05} +{"train_loss": 0.016517404466867447, "grad_norm": 0.10111799836158752, "global_step": 29246, "epoch": 275, "lr": 4.189446165826352e-05} +{"train_loss": 0.015320559963583946, "grad_norm": 0.09478054940700531, "global_step": 29247, "epoch": 275, "lr": 4.189153710887498e-05} +{"train_loss": 0.014882889576256275, "grad_norm": 0.068726547062397, "global_step": 29248, "epoch": 275, "lr": 4.188861258797604e-05} +{"train_loss": 0.015056194737553596, "grad_norm": 0.05526404827833176, "global_step": 29249, "epoch": 275, "lr": 4.1885688095577e-05} +{"train_loss": 0.014900344423949718, "grad_norm": 0.09476882964372635, "global_step": 29250, "epoch": 275, "lr": 4.188276363168814e-05} +{"train_loss": 0.014490671455860138, "grad_norm": 0.05922741815447807, "global_step": 29251, "epoch": 275, "lr": 4.18798391963197e-05} +{"train_loss": 0.017816975712776184, "grad_norm": 0.09507297724485397, "global_step": 29252, "epoch": 275, "lr": 4.1876914789481975e-05} +{"train_loss": 0.016636481508612633, "grad_norm": 0.07061716914176941, "global_step": 29253, "epoch": 275, "lr": 4.1873990411185224e-05} +{"train_loss": 0.015571869909763336, "grad_norm": 0.08116088807582855, "global_step": 29254, "epoch": 275, "lr": 4.187106606143976e-05} +{"train_loss": 0.015419892378081128, "grad_norm": 0.09500442445278168, "global_step": 29255, "epoch": 275, "lr": 4.1868141740255823e-05, "val_loss": 0.019374269992113113} +{"train_loss": 0.014495685696601868, "grad_norm": 0.06497623771429062, "global_step": 29256, "epoch": 276, "lr": 4.186521744764371e-05} +{"train_loss": 0.014879106543958187, "grad_norm": 0.07050599157810211, "global_step": 29257, "epoch": 276, "lr": 4.186229318361367e-05} +{"train_loss": 0.018943609669804573, "grad_norm": 0.0894370898604393, "global_step": 29258, "epoch": 276, "lr": 4.1859368948176007e-05} +{"train_loss": 0.016799796372652054, "grad_norm": 0.09924580901861191, "global_step": 29259, "epoch": 276, "lr": 4.185644474134097e-05} +{"train_loss": 0.015708953142166138, "grad_norm": 0.06100882217288017, "global_step": 29260, "epoch": 276, "lr": 4.185352056311884e-05} +{"train_loss": 0.017326626926660538, "grad_norm": 0.11475172638893127, "global_step": 29261, "epoch": 276, "lr": 4.1850596413519904e-05} +{"train_loss": 0.014713282696902752, "grad_norm": 0.14115794003009796, "global_step": 29262, "epoch": 276, "lr": 4.1847672292554446e-05} +{"train_loss": 0.014700817875564098, "grad_norm": 0.06857987493276596, "global_step": 29263, "epoch": 276, "lr": 4.184474820023271e-05} +{"train_loss": 0.014842947013676167, "grad_norm": 0.07211155444383621, "global_step": 29264, "epoch": 276, "lr": 4.1841824136564976e-05} +{"train_loss": 0.015658635646104813, "grad_norm": 0.07251809537410736, "global_step": 29265, "epoch": 276, "lr": 4.183890010156152e-05} +{"train_loss": 0.012729746289551258, "grad_norm": 0.06411857903003693, "global_step": 29266, "epoch": 276, "lr": 4.1835976095232624e-05} +{"train_loss": 0.01616794243454933, "grad_norm": 0.07108867168426514, "global_step": 29267, "epoch": 276, "lr": 4.1833052117588556e-05} +{"train_loss": 0.013962308876216412, "grad_norm": 0.06509076803922653, "global_step": 29268, "epoch": 276, "lr": 4.183012816863961e-05} +{"train_loss": 0.013350240886211395, "grad_norm": 0.05334869399666786, "global_step": 29269, "epoch": 276, "lr": 4.1827204248396026e-05} +{"train_loss": 0.014827455393970013, "grad_norm": 0.06581069529056549, "global_step": 29270, "epoch": 276, "lr": 4.1824280356868076e-05} +{"train_loss": 0.017070386558771133, "grad_norm": 0.15015016496181488, "global_step": 29271, "epoch": 276, "lr": 4.182135649406607e-05} +{"train_loss": 0.016415925696492195, "grad_norm": 0.07872607558965683, "global_step": 29272, "epoch": 276, "lr": 4.181843266000026e-05} +{"train_loss": 0.014470784924924374, "grad_norm": 0.07916931062936783, "global_step": 29273, "epoch": 276, "lr": 4.1815508854680904e-05} +{"train_loss": 0.014834100380539894, "grad_norm": 0.09717284142971039, "global_step": 29274, "epoch": 276, "lr": 4.181258507811831e-05} +{"train_loss": 0.014328470453619957, "grad_norm": 0.08643323928117752, "global_step": 29275, "epoch": 276, "lr": 4.1809661330322724e-05} +{"train_loss": 0.01830979622900486, "grad_norm": 0.09649964421987534, "global_step": 29276, "epoch": 276, "lr": 4.1806737611304425e-05} +{"train_loss": 0.012406865134835243, "grad_norm": 0.08958892524242401, "global_step": 29277, "epoch": 276, "lr": 4.180381392107367e-05} +{"train_loss": 0.013856956735253334, "grad_norm": 0.09129436314105988, "global_step": 29278, "epoch": 276, "lr": 4.180089025964077e-05} +{"train_loss": 0.01676652580499649, "grad_norm": 0.10524119436740875, "global_step": 29279, "epoch": 276, "lr": 4.179796662701597e-05} +{"train_loss": 0.01828448474407196, "grad_norm": 0.09499669820070267, "global_step": 29280, "epoch": 276, "lr": 4.179504302320957e-05} +{"train_loss": 0.01650039479136467, "grad_norm": 0.08327220380306244, "global_step": 29281, "epoch": 276, "lr": 4.17921194482318e-05} +{"train_loss": 0.014437930658459663, "grad_norm": 0.09377875179052353, "global_step": 29282, "epoch": 276, "lr": 4.178919590209294e-05} +{"train_loss": 0.016526788473129272, "grad_norm": 0.07243180274963379, "global_step": 29283, "epoch": 276, "lr": 4.178627238480329e-05} +{"train_loss": 0.01690136268734932, "grad_norm": 0.09238000214099884, "global_step": 29284, "epoch": 276, "lr": 4.1783348896373106e-05} +{"train_loss": 0.016643330454826355, "grad_norm": 0.08527465164661407, "global_step": 29285, "epoch": 276, "lr": 4.178042543681267e-05} +{"train_loss": 0.014414241537451744, "grad_norm": 0.050434410572052, "global_step": 29286, "epoch": 276, "lr": 4.177750200613222e-05} +{"train_loss": 0.013339480385184288, "grad_norm": 0.06452347338199615, "global_step": 29287, "epoch": 276, "lr": 4.177457860434207e-05} +{"train_loss": 0.017457827925682068, "grad_norm": 0.07364623993635178, "global_step": 29288, "epoch": 276, "lr": 4.1771655231452466e-05} +{"train_loss": 0.015889160335063934, "grad_norm": 0.049445684999227524, "global_step": 29289, "epoch": 276, "lr": 4.1768731887473685e-05} +{"train_loss": 0.01710222288966179, "grad_norm": 0.1280701607465744, "global_step": 29290, "epoch": 276, "lr": 4.1765808572416003e-05} +{"train_loss": 0.01473325490951538, "grad_norm": 0.06055566295981407, "global_step": 29291, "epoch": 276, "lr": 4.176288528628971e-05} +{"train_loss": 0.013937650248408318, "grad_norm": 0.04804982617497444, "global_step": 29292, "epoch": 276, "lr": 4.175996202910503e-05} +{"train_loss": 0.015653841197490692, "grad_norm": 0.17334680259227753, "global_step": 29293, "epoch": 276, "lr": 4.175703880087227e-05} +{"train_loss": 0.013945730403065681, "grad_norm": 0.08655806630849838, "global_step": 29294, "epoch": 276, "lr": 4.1754115601601676e-05} +{"train_loss": 0.016539206728339195, "grad_norm": 0.11848637461662292, "global_step": 29295, "epoch": 276, "lr": 4.175119243130355e-05} +{"train_loss": 0.01544700562953949, "grad_norm": 0.09926426410675049, "global_step": 29296, "epoch": 276, "lr": 4.174826928998814e-05} +{"train_loss": 0.01601715385913849, "grad_norm": 0.06566742807626724, "global_step": 29297, "epoch": 276, "lr": 4.174534617766574e-05} +{"train_loss": 0.013829429633915424, "grad_norm": 0.060147590935230255, "global_step": 29298, "epoch": 276, "lr": 4.174242309434659e-05} +{"train_loss": 0.014022285118699074, "grad_norm": 0.06823486089706421, "global_step": 29299, "epoch": 276, "lr": 4.1739500040040964e-05} +{"train_loss": 0.01667439006268978, "grad_norm": 0.06289792060852051, "global_step": 29300, "epoch": 276, "lr": 4.173657701475915e-05} +{"train_loss": 0.015588011592626572, "grad_norm": 0.10802502185106277, "global_step": 29301, "epoch": 276, "lr": 4.1733654018511415e-05} +{"train_loss": 0.012807922437787056, "grad_norm": 0.06573213636875153, "global_step": 29302, "epoch": 276, "lr": 4.173073105130801e-05} +{"train_loss": 0.015765530988574028, "grad_norm": 0.10671571642160416, "global_step": 29303, "epoch": 276, "lr": 4.1727808113159244e-05} +{"train_loss": 0.01661754958331585, "grad_norm": 0.05393664911389351, "global_step": 29304, "epoch": 276, "lr": 4.172488520407535e-05} +{"train_loss": 0.01659497618675232, "grad_norm": 0.05811240151524544, "global_step": 29305, "epoch": 276, "lr": 4.1721962324066614e-05} +{"train_loss": 0.016187485307455063, "grad_norm": 0.11747564375400543, "global_step": 29306, "epoch": 276, "lr": 4.171903947314329e-05} +{"train_loss": 0.016165070235729218, "grad_norm": 0.10847378522157669, "global_step": 29307, "epoch": 276, "lr": 4.1716116651315674e-05} +{"train_loss": 0.015203040093183517, "grad_norm": 0.07636341452598572, "global_step": 29308, "epoch": 276, "lr": 4.171319385859401e-05} +{"train_loss": 0.01411210186779499, "grad_norm": 0.08608683198690414, "global_step": 29309, "epoch": 276, "lr": 4.1710271094988606e-05} +{"train_loss": 0.014251397922635078, "grad_norm": 0.10731761157512665, "global_step": 29310, "epoch": 276, "lr": 4.170734836050969e-05} +{"train_loss": 0.01454894058406353, "grad_norm": 0.09161384403705597, "global_step": 29311, "epoch": 276, "lr": 4.1704425655167525e-05} +{"train_loss": 0.017100002616643906, "grad_norm": 0.07333535701036453, "global_step": 29312, "epoch": 276, "lr": 4.1701502978972416e-05} +{"train_loss": 0.016922157257795334, "grad_norm": 0.08391191810369492, "global_step": 29313, "epoch": 276, "lr": 4.1698580331934614e-05} +{"train_loss": 0.01788283884525299, "grad_norm": 0.11729021370410919, "global_step": 29314, "epoch": 276, "lr": 4.1695657714064384e-05} +{"train_loss": 0.014641806483268738, "grad_norm": 0.07860952615737915, "global_step": 29315, "epoch": 276, "lr": 4.169273512537202e-05} +{"train_loss": 0.013961004093289375, "grad_norm": 0.05472024902701378, "global_step": 29316, "epoch": 276, "lr": 4.168981256586777e-05} +{"train_loss": 0.016054511070251465, "grad_norm": 0.08219826221466064, "global_step": 29317, "epoch": 276, "lr": 4.1686890035561894e-05} +{"train_loss": 0.014538035728037357, "grad_norm": 0.0606716088950634, "global_step": 29318, "epoch": 276, "lr": 4.168396753446466e-05} +{"train_loss": 0.013593770563602448, "grad_norm": 0.08300649374723434, "global_step": 29319, "epoch": 276, "lr": 4.168104506258636e-05} +{"train_loss": 0.016270453110337257, "grad_norm": 0.0746501162648201, "global_step": 29320, "epoch": 276, "lr": 4.167812261993726e-05} +{"train_loss": 0.016583848744630814, "grad_norm": 0.11763472855091095, "global_step": 29321, "epoch": 276, "lr": 4.167520020652762e-05} +{"train_loss": 0.015859046950936317, "grad_norm": 0.06998229771852493, "global_step": 29322, "epoch": 276, "lr": 4.16722778223677e-05} +{"train_loss": 0.015446760691702366, "grad_norm": 0.07431820034980774, "global_step": 29323, "epoch": 276, "lr": 4.166935546746776e-05} +{"train_loss": 0.013435011729598045, "grad_norm": 0.12243487685918808, "global_step": 29324, "epoch": 276, "lr": 4.166643314183809e-05} +{"train_loss": 0.013989489525556564, "grad_norm": 0.06322822719812393, "global_step": 29325, "epoch": 276, "lr": 4.1663510845488954e-05} +{"train_loss": 0.01669580303132534, "grad_norm": 0.1302734613418579, "global_step": 29326, "epoch": 276, "lr": 4.16605885784306e-05} +{"train_loss": 0.016196653246879578, "grad_norm": 0.07303079962730408, "global_step": 29327, "epoch": 276, "lr": 4.1657666340673344e-05} +{"train_loss": 0.014384875074028969, "grad_norm": 0.07819291204214096, "global_step": 29328, "epoch": 276, "lr": 4.165474413222739e-05} +{"train_loss": 0.015371797606348991, "grad_norm": 0.1023167297244072, "global_step": 29329, "epoch": 276, "lr": 4.1651821953103045e-05} +{"train_loss": 0.01960267499089241, "grad_norm": 0.12748901546001434, "global_step": 29330, "epoch": 276, "lr": 4.164889980331057e-05} +{"train_loss": 0.01480822078883648, "grad_norm": 0.05163757875561714, "global_step": 29331, "epoch": 276, "lr": 4.164597768286022e-05} +{"train_loss": 0.017295636236667633, "grad_norm": 0.08338703215122223, "global_step": 29332, "epoch": 276, "lr": 4.1643055591762294e-05} +{"train_loss": 0.013864364475011826, "grad_norm": 0.059162687510252, "global_step": 29333, "epoch": 276, "lr": 4.164013353002702e-05} +{"train_loss": 0.014960975386202335, "grad_norm": 0.10065808892250061, "global_step": 29334, "epoch": 276, "lr": 4.1637211497664686e-05} +{"train_loss": 0.01611703634262085, "grad_norm": 0.05567801743745804, "global_step": 29335, "epoch": 276, "lr": 4.163428949468554e-05} +{"train_loss": 0.01523449644446373, "grad_norm": 0.08578077703714371, "global_step": 29336, "epoch": 276, "lr": 4.163136752109987e-05} +{"train_loss": 0.015101268887519836, "grad_norm": 0.05194183066487312, "global_step": 29337, "epoch": 276, "lr": 4.162844557691794e-05} +{"train_loss": 0.014601876959204674, "grad_norm": 0.12103313952684402, "global_step": 29338, "epoch": 276, "lr": 4.162552366215002e-05} +{"train_loss": 0.014382830820977688, "grad_norm": 0.061743609607219696, "global_step": 29339, "epoch": 276, "lr": 4.162260177680636e-05} +{"train_loss": 0.016119137406349182, "grad_norm": 0.07483182847499847, "global_step": 29340, "epoch": 276, "lr": 4.161967992089723e-05} +{"train_loss": 0.016682550311088562, "grad_norm": 0.06183410435914993, "global_step": 29341, "epoch": 276, "lr": 4.1616758094432894e-05} +{"train_loss": 0.012835947796702385, "grad_norm": 0.15197397768497467, "global_step": 29342, "epoch": 276, "lr": 4.1613836297423635e-05} +{"train_loss": 0.015197736211121082, "grad_norm": 0.08896157890558243, "global_step": 29343, "epoch": 276, "lr": 4.16109145298797e-05} +{"train_loss": 0.018221933394670486, "grad_norm": 0.06879651546478271, "global_step": 29344, "epoch": 276, "lr": 4.1607992791811376e-05} +{"train_loss": 0.014821164309978485, "grad_norm": 0.07083355635404587, "global_step": 29345, "epoch": 276, "lr": 4.160507108322891e-05} +{"train_loss": 0.015800733119249344, "grad_norm": 0.08204060792922974, "global_step": 29346, "epoch": 276, "lr": 4.1602149404142574e-05} +{"train_loss": 0.014898117631673813, "grad_norm": 0.06504299491643906, "global_step": 29347, "epoch": 276, "lr": 4.159922775456262e-05} +{"train_loss": 0.015733491629362106, "grad_norm": 0.15419374406337738, "global_step": 29348, "epoch": 276, "lr": 4.1596306134499335e-05} +{"train_loss": 0.013725262135267258, "grad_norm": 0.0524246022105217, "global_step": 29349, "epoch": 276, "lr": 4.159338454396298e-05} +{"train_loss": 0.0152592733502388, "grad_norm": 0.08549880981445312, "global_step": 29350, "epoch": 276, "lr": 4.159046298296382e-05} +{"train_loss": 0.017312955111265182, "grad_norm": 0.08137859404087067, "global_step": 29351, "epoch": 276, "lr": 4.158754145151211e-05} +{"train_loss": 0.015164054930210114, "grad_norm": 0.07635831087827682, "global_step": 29352, "epoch": 276, "lr": 4.158461994961811e-05} +{"train_loss": 0.014929273165762424, "grad_norm": 0.06810646504163742, "global_step": 29353, "epoch": 276, "lr": 4.1581698477292105e-05} +{"train_loss": 0.014143677428364754, "grad_norm": 0.07109243422746658, "global_step": 29354, "epoch": 276, "lr": 4.1578777034544354e-05} +{"train_loss": 0.016750670969486237, "grad_norm": 0.1018824502825737, "global_step": 29355, "epoch": 276, "lr": 4.1575855621385105e-05} +{"train_loss": 0.015157395973801613, "grad_norm": 0.10823821276426315, "global_step": 29356, "epoch": 276, "lr": 4.157293423782467e-05} +{"train_loss": 0.012803371995687485, "grad_norm": 0.06328557431697845, "global_step": 29357, "epoch": 276, "lr": 4.1570012883873246e-05} +{"train_loss": 0.014665110036730766, "grad_norm": 0.06759359687566757, "global_step": 29358, "epoch": 276, "lr": 4.156709155954114e-05} +{"train_loss": 0.015462308190762997, "grad_norm": 0.08877011388540268, "global_step": 29359, "epoch": 276, "lr": 4.1564170264838595e-05} +{"train_loss": 0.01469416357576847, "grad_norm": 0.09743420779705048, "global_step": 29360, "epoch": 276, "lr": 4.15612489997759e-05} +{"train_loss": 0.015398841693927095, "grad_norm": 0.10350363701581955, "global_step": 29361, "epoch": 276, "lr": 4.155832776436331e-05, "val_loss": 0.019566364586353302} +{"train_loss": 0.014042339287698269, "grad_norm": 0.07687479257583618, "global_step": 29362, "epoch": 277, "lr": 4.155540655861109e-05} +{"train_loss": 0.014300215989351273, "grad_norm": 0.07131650298833847, "global_step": 29363, "epoch": 277, "lr": 4.155248538252949e-05} +{"train_loss": 0.015073530375957489, "grad_norm": 0.0726105272769928, "global_step": 29364, "epoch": 277, "lr": 4.154956423612877e-05} +{"train_loss": 0.01810719072818756, "grad_norm": 0.15012425184249878, "global_step": 29365, "epoch": 277, "lr": 4.1546643119419215e-05} +{"train_loss": 0.014821968972682953, "grad_norm": 0.07399605959653854, "global_step": 29366, "epoch": 277, "lr": 4.1543722032411086e-05} +{"train_loss": 0.01744479313492775, "grad_norm": 0.10381004214286804, "global_step": 29367, "epoch": 277, "lr": 4.154080097511462e-05} +{"train_loss": 0.015479683876037598, "grad_norm": 0.0837995633482933, "global_step": 29368, "epoch": 277, "lr": 4.1537879947540146e-05} +{"train_loss": 0.017615152522921562, "grad_norm": 0.10668696463108063, "global_step": 29369, "epoch": 277, "lr": 4.1534958949697846e-05} +{"train_loss": 0.01564636640250683, "grad_norm": 0.11409398913383484, "global_step": 29370, "epoch": 277, "lr": 4.1532037981598025e-05} +{"train_loss": 0.015402913093566895, "grad_norm": 0.0636730045080185, "global_step": 29371, "epoch": 277, "lr": 4.152911704325094e-05} +{"train_loss": 0.016443414613604546, "grad_norm": 0.09696248918771744, "global_step": 29372, "epoch": 277, "lr": 4.1526196134666846e-05} +{"train_loss": 0.0144881010055542, "grad_norm": 0.06706306338310242, "global_step": 29373, "epoch": 277, "lr": 4.152327525585603e-05} +{"train_loss": 0.015291367657482624, "grad_norm": 0.06588523089885712, "global_step": 29374, "epoch": 277, "lr": 4.1520354406828736e-05} +{"train_loss": 0.01396390050649643, "grad_norm": 0.11822879314422607, "global_step": 29375, "epoch": 277, "lr": 4.151743358759522e-05} +{"train_loss": 0.015112437307834625, "grad_norm": 0.09330092370510101, "global_step": 29376, "epoch": 277, "lr": 4.151451279816575e-05} +{"train_loss": 0.016176847741007805, "grad_norm": 0.10258565098047256, "global_step": 29377, "epoch": 277, "lr": 4.15115920385506e-05} +{"train_loss": 0.016472676768898964, "grad_norm": 0.08116023242473602, "global_step": 29378, "epoch": 277, "lr": 4.150867130876003e-05} +{"train_loss": 0.015270115807652473, "grad_norm": 0.058829352259635925, "global_step": 29379, "epoch": 277, "lr": 4.1505750608804275e-05} +{"train_loss": 0.015677768737077713, "grad_norm": 0.08024225383996964, "global_step": 29380, "epoch": 277, "lr": 4.150282993869365e-05} +{"train_loss": 0.015281898900866508, "grad_norm": 0.1072511300444603, "global_step": 29381, "epoch": 277, "lr": 4.149990929843836e-05} +{"train_loss": 0.01660827174782753, "grad_norm": 0.06632537394762039, "global_step": 29382, "epoch": 277, "lr": 4.1496988688048705e-05} +{"train_loss": 0.01542673259973526, "grad_norm": 0.15860411524772644, "global_step": 29383, "epoch": 277, "lr": 4.149406810753493e-05} +{"train_loss": 0.014902896247804165, "grad_norm": 0.06659884750843048, "global_step": 29384, "epoch": 277, "lr": 4.1491147556907285e-05} +{"train_loss": 0.015393191948533058, "grad_norm": 0.12238667905330658, "global_step": 29385, "epoch": 277, "lr": 4.148822703617607e-05} +{"train_loss": 0.015393625944852829, "grad_norm": 0.09627742320299149, "global_step": 29386, "epoch": 277, "lr": 4.148530654535152e-05} +{"train_loss": 0.014933373779058456, "grad_norm": 0.07629276812076569, "global_step": 29387, "epoch": 277, "lr": 4.1482386084443895e-05} +{"train_loss": 0.014957441948354244, "grad_norm": 0.07093742489814758, "global_step": 29388, "epoch": 277, "lr": 4.147946565346345e-05} +{"train_loss": 0.01572585105895996, "grad_norm": 0.06619825959205627, "global_step": 29389, "epoch": 277, "lr": 4.147654525242047e-05} +{"train_loss": 0.01615973375737667, "grad_norm": 0.12412129342556, "global_step": 29390, "epoch": 277, "lr": 4.14736248813252e-05} +{"train_loss": 0.013312357477843761, "grad_norm": 0.09848357737064362, "global_step": 29391, "epoch": 277, "lr": 4.1470704540187924e-05} +{"train_loss": 0.016287116333842278, "grad_norm": 0.07203865796327591, "global_step": 29392, "epoch": 277, "lr": 4.1467784229018857e-05} +{"train_loss": 0.01801438257098198, "grad_norm": 0.1487332135438919, "global_step": 29393, "epoch": 277, "lr": 4.1464863947828285e-05} +{"train_loss": 0.01480364240705967, "grad_norm": 0.07411666214466095, "global_step": 29394, "epoch": 277, "lr": 4.146194369662648e-05} +{"train_loss": 0.014780020341277122, "grad_norm": 0.06437785923480988, "global_step": 29395, "epoch": 277, "lr": 4.145902347542369e-05} +{"train_loss": 0.014203434810042381, "grad_norm": 0.08128464967012405, "global_step": 29396, "epoch": 277, "lr": 4.1456103284230165e-05} +{"train_loss": 0.017286663874983788, "grad_norm": 0.15912804007530212, "global_step": 29397, "epoch": 277, "lr": 4.145318312305621e-05} +{"train_loss": 0.014855814166367054, "grad_norm": 0.05676219239830971, "global_step": 29398, "epoch": 277, "lr": 4.1450262991912025e-05} +{"train_loss": 0.015579590573906898, "grad_norm": 0.10189638286828995, "global_step": 29399, "epoch": 277, "lr": 4.144734289080791e-05} +{"train_loss": 0.015145405195653439, "grad_norm": 0.13316850364208221, "global_step": 29400, "epoch": 277, "lr": 4.14444228197541e-05} +{"train_loss": 0.01499072927981615, "grad_norm": 0.16763699054718018, "global_step": 29401, "epoch": 277, "lr": 4.1441502778760864e-05} +{"train_loss": 0.01422897633165121, "grad_norm": 0.07189618796110153, "global_step": 29402, "epoch": 277, "lr": 4.1438582767838474e-05} +{"train_loss": 0.013404646888375282, "grad_norm": 0.08394069224596024, "global_step": 29403, "epoch": 277, "lr": 4.14356627869972e-05} +{"train_loss": 0.015360381454229355, "grad_norm": 0.06199534237384796, "global_step": 29404, "epoch": 277, "lr": 4.1432742836247266e-05} +{"train_loss": 0.01664820872247219, "grad_norm": 0.08240555971860886, "global_step": 29405, "epoch": 277, "lr": 4.142982291559893e-05} +{"train_loss": 0.016933167353272438, "grad_norm": 0.058564286679029465, "global_step": 29406, "epoch": 277, "lr": 4.142690302506249e-05} +{"train_loss": 0.015318721532821655, "grad_norm": 0.09057144075632095, "global_step": 29407, "epoch": 277, "lr": 4.142398316464818e-05} +{"train_loss": 0.016781151294708252, "grad_norm": 0.14921224117279053, "global_step": 29408, "epoch": 277, "lr": 4.142106333436625e-05} +{"train_loss": 0.015513627789914608, "grad_norm": 0.20687410235404968, "global_step": 29409, "epoch": 277, "lr": 4.1418143534227014e-05} +{"train_loss": 0.01308611873537302, "grad_norm": 0.07084112614393234, "global_step": 29410, "epoch": 277, "lr": 4.141522376424065e-05} +{"train_loss": 0.013473110273480415, "grad_norm": 0.13203661143779755, "global_step": 29411, "epoch": 277, "lr": 4.141230402441747e-05} +{"train_loss": 0.014535084366798401, "grad_norm": 0.1433347910642624, "global_step": 29412, "epoch": 277, "lr": 4.1409384314767716e-05} +{"train_loss": 0.016211621463298798, "grad_norm": 0.07102730870246887, "global_step": 29413, "epoch": 277, "lr": 4.140646463530164e-05} +{"train_loss": 0.015411708503961563, "grad_norm": 0.1488940417766571, "global_step": 29414, "epoch": 277, "lr": 4.140354498602952e-05} +{"train_loss": 0.01847897469997406, "grad_norm": 0.08518915623426437, "global_step": 29415, "epoch": 277, "lr": 4.140062536696162e-05} +{"train_loss": 0.01575089618563652, "grad_norm": 0.1375827044248581, "global_step": 29416, "epoch": 277, "lr": 4.139770577810817e-05} +{"train_loss": 0.014016196131706238, "grad_norm": 0.14431189000606537, "global_step": 29417, "epoch": 277, "lr": 4.1394786219479424e-05} +{"train_loss": 0.016310855746269226, "grad_norm": 0.18142205476760864, "global_step": 29418, "epoch": 277, "lr": 4.139186669108567e-05} +{"train_loss": 0.016318440437316895, "grad_norm": 0.06748933345079422, "global_step": 29419, "epoch": 277, "lr": 4.138894719293715e-05} +{"train_loss": 0.013541748747229576, "grad_norm": 0.06210844963788986, "global_step": 29420, "epoch": 277, "lr": 4.138602772504413e-05} +{"train_loss": 0.015906695276498795, "grad_norm": 0.05796832591295242, "global_step": 29421, "epoch": 277, "lr": 4.138310828741688e-05} +{"train_loss": 0.01518197264522314, "grad_norm": 0.08277677744626999, "global_step": 29422, "epoch": 277, "lr": 4.1380188880065614e-05} +{"train_loss": 0.015942126512527466, "grad_norm": 0.12972502410411835, "global_step": 29423, "epoch": 277, "lr": 4.137726950300062e-05} +{"train_loss": 0.014346042647957802, "grad_norm": 0.07862348109483719, "global_step": 29424, "epoch": 277, "lr": 4.1374350156232155e-05} +{"train_loss": 0.01581542007625103, "grad_norm": 0.09170033037662506, "global_step": 29425, "epoch": 277, "lr": 4.1371430839770463e-05} +{"train_loss": 0.017047898843884468, "grad_norm": 0.07339763641357422, "global_step": 29426, "epoch": 277, "lr": 4.136851155362582e-05} +{"train_loss": 0.012126311659812927, "grad_norm": 0.10272137820720673, "global_step": 29427, "epoch": 277, "lr": 4.1365592297808484e-05} +{"train_loss": 0.015405509620904922, "grad_norm": 0.15789827704429626, "global_step": 29428, "epoch": 277, "lr": 4.1362673072328695e-05} +{"train_loss": 0.013880771584808826, "grad_norm": 0.0656486451625824, "global_step": 29429, "epoch": 277, "lr": 4.1359753877196714e-05} +{"train_loss": 0.015152381733059883, "grad_norm": 0.06493937969207764, "global_step": 29430, "epoch": 277, "lr": 4.1356834712422786e-05} +{"train_loss": 0.0159929022192955, "grad_norm": 0.0914129912853241, "global_step": 29431, "epoch": 277, "lr": 4.135391557801719e-05} +{"train_loss": 0.014614081010222435, "grad_norm": 0.08350619673728943, "global_step": 29432, "epoch": 277, "lr": 4.135099647399017e-05} +{"train_loss": 0.016945350915193558, "grad_norm": 0.11104924231767654, "global_step": 29433, "epoch": 277, "lr": 4.134807740035202e-05} +{"train_loss": 0.013447003439068794, "grad_norm": 0.1276710480451584, "global_step": 29434, "epoch": 277, "lr": 4.134515835711293e-05} +{"train_loss": 0.01464110892266035, "grad_norm": 0.084751196205616, "global_step": 29435, "epoch": 277, "lr": 4.13422393442832e-05} +{"train_loss": 0.01551552303135395, "grad_norm": 0.12759189307689667, "global_step": 29436, "epoch": 277, "lr": 4.1339320361873075e-05} +{"train_loss": 0.017217552289366722, "grad_norm": 0.06066959351301193, "global_step": 29437, "epoch": 277, "lr": 4.13364014098928e-05} +{"train_loss": 0.015322507359087467, "grad_norm": 0.1035100594162941, "global_step": 29438, "epoch": 277, "lr": 4.133348248835268e-05} +{"train_loss": 0.015988092869520187, "grad_norm": 0.1084229052066803, "global_step": 29439, "epoch": 277, "lr": 4.133056359726289e-05} +{"train_loss": 0.015195518732070923, "grad_norm": 0.12870077788829803, "global_step": 29440, "epoch": 277, "lr": 4.132764473663375e-05} +{"train_loss": 0.015689264982938766, "grad_norm": 0.10761572420597076, "global_step": 29441, "epoch": 277, "lr": 4.13247259064755e-05} +{"train_loss": 0.014229637570679188, "grad_norm": 0.05071501433849335, "global_step": 29442, "epoch": 277, "lr": 4.132180710679837e-05} +{"train_loss": 0.015343422070145607, "grad_norm": 0.05212453380227089, "global_step": 29443, "epoch": 277, "lr": 4.131888833761265e-05} +{"train_loss": 0.014887729659676552, "grad_norm": 0.11193735152482986, "global_step": 29444, "epoch": 277, "lr": 4.131596959892859e-05} +{"train_loss": 0.015571287833154202, "grad_norm": 0.07015200704336166, "global_step": 29445, "epoch": 277, "lr": 4.131305089075643e-05} +{"train_loss": 0.012796875089406967, "grad_norm": 0.08636446297168732, "global_step": 29446, "epoch": 277, "lr": 4.131013221310641e-05} +{"train_loss": 0.019407449290156364, "grad_norm": 0.09064168483018875, "global_step": 29447, "epoch": 277, "lr": 4.130721356598883e-05} +{"train_loss": 0.01638445258140564, "grad_norm": 0.06914439052343369, "global_step": 29448, "epoch": 277, "lr": 4.1304294949413915e-05} +{"train_loss": 0.01599220000207424, "grad_norm": 0.14919087290763855, "global_step": 29449, "epoch": 277, "lr": 4.1301376363391905e-05} +{"train_loss": 0.014251366257667542, "grad_norm": 0.08002641052007675, "global_step": 29450, "epoch": 277, "lr": 4.129845780793312e-05} +{"train_loss": 0.014144552871584892, "grad_norm": 0.07031925022602081, "global_step": 29451, "epoch": 277, "lr": 4.129553928304774e-05} +{"train_loss": 0.014697219245135784, "grad_norm": 0.1094224750995636, "global_step": 29452, "epoch": 277, "lr": 4.1292620788746055e-05} +{"train_loss": 0.014616532251238823, "grad_norm": 0.0728483572602272, "global_step": 29453, "epoch": 277, "lr": 4.128970232503831e-05} +{"train_loss": 0.016490666195750237, "grad_norm": 0.16095837950706482, "global_step": 29454, "epoch": 277, "lr": 4.1286783891934766e-05} +{"train_loss": 0.014736787416040897, "grad_norm": 0.0641225203871727, "global_step": 29455, "epoch": 277, "lr": 4.128386548944567e-05} +{"train_loss": 0.014974081888794899, "grad_norm": 0.07885605096817017, "global_step": 29456, "epoch": 277, "lr": 4.12809471175813e-05} +{"train_loss": 0.017527824267745018, "grad_norm": 0.09053628146648407, "global_step": 29457, "epoch": 277, "lr": 4.127802877635187e-05} +{"train_loss": 0.014463940635323524, "grad_norm": 0.07864905893802643, "global_step": 29458, "epoch": 277, "lr": 4.1275110465767664e-05} +{"train_loss": 0.015710972249507904, "grad_norm": 0.15254928171634674, "global_step": 29459, "epoch": 277, "lr": 4.127219218583891e-05} +{"train_loss": 0.016341235488653183, "grad_norm": 0.12334917485713959, "global_step": 29460, "epoch": 277, "lr": 4.126927393657589e-05} +{"train_loss": 0.015741655603051186, "grad_norm": 0.0716090127825737, "global_step": 29461, "epoch": 277, "lr": 4.126635571798884e-05} +{"train_loss": 0.014867546036839485, "grad_norm": 0.10633275657892227, "global_step": 29462, "epoch": 277, "lr": 4.1263437530088035e-05} +{"train_loss": 0.017183564603328705, "grad_norm": 0.08779723197221756, "global_step": 29463, "epoch": 277, "lr": 4.126051937288369e-05} +{"train_loss": 0.016400841996073723, "grad_norm": 0.08272354304790497, "global_step": 29464, "epoch": 277, "lr": 4.125760124638608e-05} +{"train_loss": 0.01818358525633812, "grad_norm": 0.05603805556893349, "global_step": 29465, "epoch": 277, "lr": 4.125468315060546e-05} +{"train_loss": 0.014337508007884026, "grad_norm": 0.1720200926065445, "global_step": 29466, "epoch": 277, "lr": 4.125176508555208e-05} +{"train_loss": 0.015431382482484827, "grad_norm": 0.10690641403198242, "global_step": 29467, "epoch": 277, "lr": 4.1248847051236195e-05, "val_loss": 0.020316971465945244} +{"train_loss": 0.014879293739795685, "grad_norm": 0.08932913094758987, "global_step": 29468, "epoch": 278, "lr": 4.124592904766807e-05} +{"train_loss": 0.015690527856349945, "grad_norm": 0.07223252207040787, "global_step": 29469, "epoch": 278, "lr": 4.124301107485793e-05} +{"train_loss": 0.014838133938610554, "grad_norm": 0.12341979891061783, "global_step": 29470, "epoch": 278, "lr": 4.124009313281604e-05} +{"train_loss": 0.016039783135056496, "grad_norm": 0.16806620359420776, "global_step": 29471, "epoch": 278, "lr": 4.1237175221552645e-05} +{"train_loss": 0.015268145129084587, "grad_norm": 0.1713080257177353, "global_step": 29472, "epoch": 278, "lr": 4.123425734107801e-05} +{"train_loss": 0.01545325294137001, "grad_norm": 0.21621400117874146, "global_step": 29473, "epoch": 278, "lr": 4.123133949140239e-05} +{"train_loss": 0.017075765877962112, "grad_norm": 0.09659790992736816, "global_step": 29474, "epoch": 278, "lr": 4.122842167253604e-05} +{"train_loss": 0.01387416385114193, "grad_norm": 0.09110711514949799, "global_step": 29475, "epoch": 278, "lr": 4.122550388448917e-05} +{"train_loss": 0.016304034739732742, "grad_norm": 0.07404238730669022, "global_step": 29476, "epoch": 278, "lr": 4.122258612727209e-05} +{"train_loss": 0.01433481089770794, "grad_norm": 0.13758040964603424, "global_step": 29477, "epoch": 278, "lr": 4.1219668400895016e-05} +{"train_loss": 0.014185287058353424, "grad_norm": 0.08896695077419281, "global_step": 29478, "epoch": 278, "lr": 4.12167507053682e-05} +{"train_loss": 0.017807234078645706, "grad_norm": 0.07823046296834946, "global_step": 29479, "epoch": 278, "lr": 4.121383304070191e-05} +{"train_loss": 0.018078239634633064, "grad_norm": 0.0679866373538971, "global_step": 29480, "epoch": 278, "lr": 4.1210915406906405e-05} +{"train_loss": 0.017885703593492508, "grad_norm": 0.1779993772506714, "global_step": 29481, "epoch": 278, "lr": 4.1207997803991904e-05} +{"train_loss": 0.01382632739841938, "grad_norm": 0.07724010199308395, "global_step": 29482, "epoch": 278, "lr": 4.120508023196868e-05} +{"train_loss": 0.015367879532277584, "grad_norm": 0.05086567997932434, "global_step": 29483, "epoch": 278, "lr": 4.120216269084697e-05} +{"train_loss": 0.015685653313994408, "grad_norm": 0.08143341541290283, "global_step": 29484, "epoch": 278, "lr": 4.119924518063704e-05} +{"train_loss": 0.016569707542657852, "grad_norm": 0.12915079295635223, "global_step": 29485, "epoch": 278, "lr": 4.1196327701349146e-05} +{"train_loss": 0.016937412321567535, "grad_norm": 0.05440415441989899, "global_step": 29486, "epoch": 278, "lr": 4.119341025299353e-05} +{"train_loss": 0.01206957921385765, "grad_norm": 0.0823194608092308, "global_step": 29487, "epoch": 278, "lr": 4.1190492835580414e-05} +{"train_loss": 0.017096932977437973, "grad_norm": 0.13718226552009583, "global_step": 29488, "epoch": 278, "lr": 4.118757544912009e-05} +{"train_loss": 0.017676623538136482, "grad_norm": 0.1316756159067154, "global_step": 29489, "epoch": 278, "lr": 4.11846580936228e-05} +{"train_loss": 0.015554310753941536, "grad_norm": 0.11230847984552383, "global_step": 29490, "epoch": 278, "lr": 4.118174076909876e-05} +{"train_loss": 0.014158849604427814, "grad_norm": 0.062481410801410675, "global_step": 29491, "epoch": 278, "lr": 4.1178823475558296e-05} +{"train_loss": 0.016828738152980804, "grad_norm": 0.19117574393749237, "global_step": 29492, "epoch": 278, "lr": 4.117590621301157e-05} +{"train_loss": 0.014271987602114677, "grad_norm": 0.07694713771343231, "global_step": 29493, "epoch": 278, "lr": 4.117298898146889e-05} +{"train_loss": 0.01425749622285366, "grad_norm": 0.0865975171327591, "global_step": 29494, "epoch": 278, "lr": 4.1170071780940486e-05} +{"train_loss": 0.0138156209141016, "grad_norm": 0.13519315421581268, "global_step": 29495, "epoch": 278, "lr": 4.1167154611436596e-05} +{"train_loss": 0.014016878791153431, "grad_norm": 0.05618367716670036, "global_step": 29496, "epoch": 278, "lr": 4.116423747296749e-05} +{"train_loss": 0.012720640748739243, "grad_norm": 0.06098568066954613, "global_step": 29497, "epoch": 278, "lr": 4.1161320365543434e-05} +{"train_loss": 0.012783054262399673, "grad_norm": 0.08611273765563965, "global_step": 29498, "epoch": 278, "lr": 4.115840328917463e-05} +{"train_loss": 0.016238044947385788, "grad_norm": 0.09731418639421463, "global_step": 29499, "epoch": 278, "lr": 4.115548624387137e-05} +{"train_loss": 0.015275850892066956, "grad_norm": 0.06437636911869049, "global_step": 29500, "epoch": 278, "lr": 4.115256922964386e-05} +{"train_loss": 0.013809267431497574, "grad_norm": 0.07286286354064941, "global_step": 29501, "epoch": 278, "lr": 4.114965224650238e-05} +{"train_loss": 0.014715357683598995, "grad_norm": 0.1301223337650299, "global_step": 29502, "epoch": 278, "lr": 4.114673529445719e-05} +{"train_loss": 0.014247996732592583, "grad_norm": 0.14626134932041168, "global_step": 29503, "epoch": 278, "lr": 4.114381837351852e-05} +{"train_loss": 0.01651168242096901, "grad_norm": 0.08344188332557678, "global_step": 29504, "epoch": 278, "lr": 4.1140901483696606e-05} +{"train_loss": 0.013298097997903824, "grad_norm": 0.0565393827855587, "global_step": 29505, "epoch": 278, "lr": 4.113798462500171e-05} +{"train_loss": 0.016008440405130386, "grad_norm": 0.06370586901903152, "global_step": 29506, "epoch": 278, "lr": 4.1135067797444096e-05} +{"train_loss": 0.016335515305399895, "grad_norm": 0.11139339208602905, "global_step": 29507, "epoch": 278, "lr": 4.113215100103398e-05} +{"train_loss": 0.017421120777726173, "grad_norm": 0.11770834773778915, "global_step": 29508, "epoch": 278, "lr": 4.112923423578164e-05} +{"train_loss": 0.014110613614320755, "grad_norm": 0.15605328977108002, "global_step": 29509, "epoch": 278, "lr": 4.112631750169733e-05} +{"train_loss": 0.014163161627948284, "grad_norm": 0.06816352158784866, "global_step": 29510, "epoch": 278, "lr": 4.112340079879127e-05} +{"train_loss": 0.014919420704245567, "grad_norm": 0.053974222391843796, "global_step": 29511, "epoch": 278, "lr": 4.112048412707371e-05} +{"train_loss": 0.01524925697594881, "grad_norm": 0.08141951262950897, "global_step": 29512, "epoch": 278, "lr": 4.1117567486554906e-05} +{"train_loss": 0.016517026349902153, "grad_norm": 0.14002181589603424, "global_step": 29513, "epoch": 278, "lr": 4.1114650877245106e-05} +{"train_loss": 0.01362977921962738, "grad_norm": 0.07573530822992325, "global_step": 29514, "epoch": 278, "lr": 4.111173429915457e-05} +{"train_loss": 0.016718991100788116, "grad_norm": 0.0683487132191658, "global_step": 29515, "epoch": 278, "lr": 4.110881775229354e-05} +{"train_loss": 0.015973255038261414, "grad_norm": 0.059861574321985245, "global_step": 29516, "epoch": 278, "lr": 4.110590123667224e-05} +{"train_loss": 0.014063897542655468, "grad_norm": 0.11974133551120758, "global_step": 29517, "epoch": 278, "lr": 4.1102984752300944e-05} +{"train_loss": 0.017826426774263382, "grad_norm": 0.06609297543764114, "global_step": 29518, "epoch": 278, "lr": 4.1100068299189895e-05} +{"train_loss": 0.015793126076459885, "grad_norm": 0.1011669859290123, "global_step": 29519, "epoch": 278, "lr": 4.1097151877349316e-05} +{"train_loss": 0.01424395851790905, "grad_norm": 0.09241931885480881, "global_step": 29520, "epoch": 278, "lr": 4.109423548678949e-05} +{"train_loss": 0.01668611168861389, "grad_norm": 0.12102046608924866, "global_step": 29521, "epoch": 278, "lr": 4.109131912752066e-05} +{"train_loss": 0.015264367684721947, "grad_norm": 0.1464819759130478, "global_step": 29522, "epoch": 278, "lr": 4.108840279955305e-05} +{"train_loss": 0.01413784921169281, "grad_norm": 0.08775245398283005, "global_step": 29523, "epoch": 278, "lr": 4.108548650289691e-05} +{"train_loss": 0.017082396894693375, "grad_norm": 0.10674022883176804, "global_step": 29524, "epoch": 278, "lr": 4.1082570237562475e-05} +{"train_loss": 0.015020778402686119, "grad_norm": 0.09400612115859985, "global_step": 29525, "epoch": 278, "lr": 4.107965400356004e-05} +{"train_loss": 0.013660351745784283, "grad_norm": 0.07740423828363419, "global_step": 29526, "epoch": 278, "lr": 4.107673780089981e-05} +{"train_loss": 0.017284134402871132, "grad_norm": 0.12744377553462982, "global_step": 29527, "epoch": 278, "lr": 4.107382162959206e-05} +{"train_loss": 0.016340864822268486, "grad_norm": 0.06250962615013123, "global_step": 29528, "epoch": 278, "lr": 4.1070905489647006e-05} +{"train_loss": 0.01544509083032608, "grad_norm": 0.14356811344623566, "global_step": 29529, "epoch": 278, "lr": 4.106798938107489e-05} +{"train_loss": 0.014525110833346844, "grad_norm": 0.0778239443898201, "global_step": 29530, "epoch": 278, "lr": 4.106507330388599e-05} +{"train_loss": 0.017266903072595596, "grad_norm": 0.10005787760019302, "global_step": 29531, "epoch": 278, "lr": 4.1062157258090536e-05} +{"train_loss": 0.01710851490497589, "grad_norm": 0.12907230854034424, "global_step": 29532, "epoch": 278, "lr": 4.105924124369877e-05} +{"train_loss": 0.018377942964434624, "grad_norm": 0.06287463754415512, "global_step": 29533, "epoch": 278, "lr": 4.105632526072096e-05} +{"train_loss": 0.016492491587996483, "grad_norm": 0.08969639241695404, "global_step": 29534, "epoch": 278, "lr": 4.105340930916732e-05} +{"train_loss": 0.014118390157818794, "grad_norm": 0.12729650735855103, "global_step": 29535, "epoch": 278, "lr": 4.105049338904811e-05} +{"train_loss": 0.013450095430016518, "grad_norm": 0.10506390035152435, "global_step": 29536, "epoch": 278, "lr": 4.104757750037356e-05} +{"train_loss": 0.015403324738144875, "grad_norm": 0.05560022592544556, "global_step": 29537, "epoch": 278, "lr": 4.104466164315395e-05} +{"train_loss": 0.017217621207237244, "grad_norm": 0.07338914275169373, "global_step": 29538, "epoch": 278, "lr": 4.1041745817399507e-05} +{"train_loss": 0.014234296046197414, "grad_norm": 0.16048020124435425, "global_step": 29539, "epoch": 278, "lr": 4.103883002312047e-05} +{"train_loss": 0.012563636526465416, "grad_norm": 0.05877382680773735, "global_step": 29540, "epoch": 278, "lr": 4.103591426032708e-05} +{"train_loss": 0.015704315155744553, "grad_norm": 0.08929730206727982, "global_step": 29541, "epoch": 278, "lr": 4.103299852902958e-05} +{"train_loss": 0.013534251600503922, "grad_norm": 0.1264205127954483, "global_step": 29542, "epoch": 278, "lr": 4.103008282923823e-05} +{"train_loss": 0.01698511466383934, "grad_norm": 0.07854002714157104, "global_step": 29543, "epoch": 278, "lr": 4.102716716096327e-05} +{"train_loss": 0.014759998768568039, "grad_norm": 0.08443088829517365, "global_step": 29544, "epoch": 278, "lr": 4.1024251524214954e-05} +{"train_loss": 0.01484636589884758, "grad_norm": 0.0850796326994896, "global_step": 29545, "epoch": 278, "lr": 4.10213359190035e-05} +{"train_loss": 0.013929231092333794, "grad_norm": 0.050678614526987076, "global_step": 29546, "epoch": 278, "lr": 4.101842034533917e-05} +{"train_loss": 0.015415813773870468, "grad_norm": 0.07619301974773407, "global_step": 29547, "epoch": 278, "lr": 4.101550480323221e-05} +{"train_loss": 0.015800707042217255, "grad_norm": 0.13699238002300262, "global_step": 29548, "epoch": 278, "lr": 4.101258929269284e-05} +{"train_loss": 0.01665518805384636, "grad_norm": 0.0635232925415039, "global_step": 29549, "epoch": 278, "lr": 4.100967381373134e-05} +{"train_loss": 0.01453142985701561, "grad_norm": 0.08588311821222305, "global_step": 29550, "epoch": 278, "lr": 4.100675836635795e-05} +{"train_loss": 0.012805834412574768, "grad_norm": 0.08153356611728668, "global_step": 29551, "epoch": 278, "lr": 4.1003842950582885e-05} +{"train_loss": 0.014085883274674416, "grad_norm": 0.10526023060083389, "global_step": 29552, "epoch": 278, "lr": 4.10009275664164e-05} +{"train_loss": 0.016627447679638863, "grad_norm": 0.06278491020202637, "global_step": 29553, "epoch": 278, "lr": 4.0998012213868734e-05} +{"train_loss": 0.014009863138198853, "grad_norm": 0.06049450859427452, "global_step": 29554, "epoch": 278, "lr": 4.0995096892950155e-05} +{"train_loss": 0.018042072653770447, "grad_norm": 0.07438760995864868, "global_step": 29555, "epoch": 278, "lr": 4.099218160367088e-05} +{"train_loss": 0.016297779977321625, "grad_norm": 0.1908247321844101, "global_step": 29556, "epoch": 278, "lr": 4.098926634604118e-05} +{"train_loss": 0.017791952937841415, "grad_norm": 0.06784393638372421, "global_step": 29557, "epoch": 278, "lr": 4.098635112007127e-05} +{"train_loss": 0.017038390040397644, "grad_norm": 0.09693509340286255, "global_step": 29558, "epoch": 278, "lr": 4.098343592577138e-05} +{"train_loss": 0.015057644806802273, "grad_norm": 0.09404078871011734, "global_step": 29559, "epoch": 278, "lr": 4.0980520763151796e-05} +{"train_loss": 0.01618429832160473, "grad_norm": 0.0875513032078743, "global_step": 29560, "epoch": 278, "lr": 4.0977605632222735e-05} +{"train_loss": 0.013832496479153633, "grad_norm": 0.07752838730812073, "global_step": 29561, "epoch": 278, "lr": 4.097469053299443e-05} +{"train_loss": 0.016169123351573944, "grad_norm": 0.12066558748483658, "global_step": 29562, "epoch": 278, "lr": 4.097177546547717e-05} +{"train_loss": 0.017143864184617996, "grad_norm": 0.054459985345602036, "global_step": 29563, "epoch": 278, "lr": 4.096886042968114e-05} +{"train_loss": 0.017661873251199722, "grad_norm": 0.08159485459327698, "global_step": 29564, "epoch": 278, "lr": 4.096594542561662e-05} +{"train_loss": 0.012910104356706142, "grad_norm": 0.10259821265935898, "global_step": 29565, "epoch": 278, "lr": 4.096303045329382e-05} +{"train_loss": 0.015866953879594803, "grad_norm": 0.14282944798469543, "global_step": 29566, "epoch": 278, "lr": 4.096011551272301e-05} +{"train_loss": 0.015163550153374672, "grad_norm": 0.07013406604528427, "global_step": 29567, "epoch": 278, "lr": 4.095720060391443e-05} +{"train_loss": 0.015214371494948864, "grad_norm": 0.08035800606012344, "global_step": 29568, "epoch": 278, "lr": 4.095428572687832e-05} +{"train_loss": 0.013063372112810612, "grad_norm": 0.0495508573949337, "global_step": 29569, "epoch": 278, "lr": 4.095137088162491e-05} +{"train_loss": 0.01469743624329567, "grad_norm": 0.089617520570755, "global_step": 29570, "epoch": 278, "lr": 4.094845606816443e-05} +{"train_loss": 0.015482894144952297, "grad_norm": 0.08176674693822861, "global_step": 29571, "epoch": 278, "lr": 4.094554128650716e-05} +{"train_loss": 0.015187202021479607, "grad_norm": 0.16025491058826447, "global_step": 29572, "epoch": 278, "lr": 4.094262653666331e-05} +{"train_loss": 0.015404176494141794, "grad_norm": 0.0816504955291748, "global_step": 29573, "epoch": 278, "lr": 4.093971181864313e-05, "val_loss": 0.019128425046801567} +{"train_loss": 0.017989810556173325, "grad_norm": 0.16940444707870483, "global_step": 29574, "epoch": 279, "lr": 4.093679713245688e-05} +{"train_loss": 0.014183364808559418, "grad_norm": 0.05719219520688057, "global_step": 29575, "epoch": 279, "lr": 4.093388247811476e-05} +{"train_loss": 0.014731942676007748, "grad_norm": 0.07843434065580368, "global_step": 29576, "epoch": 279, "lr": 4.093096785562705e-05} +{"train_loss": 0.014694164507091045, "grad_norm": 0.08082351088523865, "global_step": 29577, "epoch": 279, "lr": 4.0928053265003965e-05} +{"train_loss": 0.015767939388751984, "grad_norm": 0.05107276886701584, "global_step": 29578, "epoch": 279, "lr": 4.092513870625576e-05} +{"train_loss": 0.017007935792207718, "grad_norm": 0.09990403801202774, "global_step": 29579, "epoch": 279, "lr": 4.092222417939267e-05} +{"train_loss": 0.01682281494140625, "grad_norm": 0.11941898614168167, "global_step": 29580, "epoch": 279, "lr": 4.091930968442495e-05} +{"train_loss": 0.019698020070791245, "grad_norm": 0.10290884226560593, "global_step": 29581, "epoch": 279, "lr": 4.091639522136281e-05} +{"train_loss": 0.01636064052581787, "grad_norm": 0.08134761452674866, "global_step": 29582, "epoch": 279, "lr": 4.091348079021651e-05} +{"train_loss": 0.01864585466682911, "grad_norm": 0.08634336292743683, "global_step": 29583, "epoch": 279, "lr": 4.0910566390996285e-05} +{"train_loss": 0.016206271946430206, "grad_norm": 0.0675666332244873, "global_step": 29584, "epoch": 279, "lr": 4.0907652023712386e-05} +{"train_loss": 0.017710240557789803, "grad_norm": 0.0845705196261406, "global_step": 29585, "epoch": 279, "lr": 4.0904737688375046e-05} +{"train_loss": 0.014868607744574547, "grad_norm": 0.06924370676279068, "global_step": 29586, "epoch": 279, "lr": 4.090182338499449e-05} +{"train_loss": 0.014771826565265656, "grad_norm": 0.07339876890182495, "global_step": 29587, "epoch": 279, "lr": 4.0898909113580966e-05} +{"train_loss": 0.013790301978588104, "grad_norm": 0.07749916613101959, "global_step": 29588, "epoch": 279, "lr": 4.089599487414473e-05} +{"train_loss": 0.015715165063738823, "grad_norm": 0.0701560229063034, "global_step": 29589, "epoch": 279, "lr": 4.0893080666696e-05} +{"train_loss": 0.016514355316758156, "grad_norm": 0.08054365962743759, "global_step": 29590, "epoch": 279, "lr": 4.089016649124502e-05} +{"train_loss": 0.01568056084215641, "grad_norm": 0.08036987483501434, "global_step": 29591, "epoch": 279, "lr": 4.0887252347802055e-05} +{"train_loss": 0.014742547646164894, "grad_norm": 0.07199035584926605, "global_step": 29592, "epoch": 279, "lr": 4.0884338236377305e-05} +{"train_loss": 0.016779229044914246, "grad_norm": 0.06654129922389984, "global_step": 29593, "epoch": 279, "lr": 4.0881424156981026e-05} +{"train_loss": 0.015339922159910202, "grad_norm": 0.05501265078783035, "global_step": 29594, "epoch": 279, "lr": 4.087851010962346e-05} +{"train_loss": 0.016438627615571022, "grad_norm": 0.11651443690061569, "global_step": 29595, "epoch": 279, "lr": 4.0875596094314836e-05} +{"train_loss": 0.013557132333517075, "grad_norm": 0.19539803266525269, "global_step": 29596, "epoch": 279, "lr": 4.0872682111065416e-05} +{"train_loss": 0.016802143305540085, "grad_norm": 0.10701177269220352, "global_step": 29597, "epoch": 279, "lr": 4.0869768159885416e-05} +{"train_loss": 0.015067875385284424, "grad_norm": 0.06565709412097931, "global_step": 29598, "epoch": 279, "lr": 4.086685424078508e-05} +{"train_loss": 0.017098357900977135, "grad_norm": 0.07965188473463058, "global_step": 29599, "epoch": 279, "lr": 4.086394035377463e-05} +{"train_loss": 0.014129092916846275, "grad_norm": 0.07640189677476883, "global_step": 29600, "epoch": 279, "lr": 4.0861026498864335e-05} +{"train_loss": 0.015678107738494873, "grad_norm": 0.07134170830249786, "global_step": 29601, "epoch": 279, "lr": 4.085811267606442e-05} +{"train_loss": 0.017498023808002472, "grad_norm": 0.08097881078720093, "global_step": 29602, "epoch": 279, "lr": 4.0855198885385104e-05} +{"train_loss": 0.014522974379360676, "grad_norm": 0.11231489479541779, "global_step": 29603, "epoch": 279, "lr": 4.085228512683667e-05} +{"train_loss": 0.016415296122431755, "grad_norm": 0.0708068236708641, "global_step": 29604, "epoch": 279, "lr": 4.084937140042931e-05} +{"train_loss": 0.013469865545630455, "grad_norm": 0.11040510982275009, "global_step": 29605, "epoch": 279, "lr": 4.084645770617329e-05} +{"train_loss": 0.015143328346312046, "grad_norm": 0.10173742473125458, "global_step": 29606, "epoch": 279, "lr": 4.084354404407882e-05} +{"train_loss": 0.013724533841013908, "grad_norm": 0.11601147800683975, "global_step": 29607, "epoch": 279, "lr": 4.084063041415617e-05} +{"train_loss": 0.01476946659386158, "grad_norm": 0.06915229558944702, "global_step": 29608, "epoch": 279, "lr": 4.083771681641555e-05} +{"train_loss": 0.019459573552012444, "grad_norm": 0.1046476662158966, "global_step": 29609, "epoch": 279, "lr": 4.0834803250867224e-05} +{"train_loss": 0.011609572917222977, "grad_norm": 0.06936569511890411, "global_step": 29610, "epoch": 279, "lr": 4.083188971752141e-05} +{"train_loss": 0.015356676653027534, "grad_norm": 0.06046285480260849, "global_step": 29611, "epoch": 279, "lr": 4.082897621638833e-05} +{"train_loss": 0.014922497794032097, "grad_norm": 0.10175013542175293, "global_step": 29612, "epoch": 279, "lr": 4.082606274747825e-05} +{"train_loss": 0.014531301334500313, "grad_norm": 0.10382281988859177, "global_step": 29613, "epoch": 279, "lr": 4.08231493108014e-05} +{"train_loss": 0.015309898182749748, "grad_norm": 0.05842450261116028, "global_step": 29614, "epoch": 279, "lr": 4.0820235906368e-05} +{"train_loss": 0.015277578495442867, "grad_norm": 0.07659156620502472, "global_step": 29615, "epoch": 279, "lr": 4.081732253418833e-05} +{"train_loss": 0.018877863883972168, "grad_norm": 0.06388925015926361, "global_step": 29616, "epoch": 279, "lr": 4.081440919427256e-05} +{"train_loss": 0.014291038736701012, "grad_norm": 0.059421710669994354, "global_step": 29617, "epoch": 279, "lr": 4.081149588663098e-05} +{"train_loss": 0.012633267790079117, "grad_norm": 0.07319259643554688, "global_step": 29618, "epoch": 279, "lr": 4.0808582611273786e-05} +{"train_loss": 0.013254331424832344, "grad_norm": 0.06815825402736664, "global_step": 29619, "epoch": 279, "lr": 4.0805669368211256e-05} +{"train_loss": 0.013784024864435196, "grad_norm": 0.10940028727054596, "global_step": 29620, "epoch": 279, "lr": 4.08027561574536e-05} +{"train_loss": 0.016909752041101456, "grad_norm": 0.058068543672561646, "global_step": 29621, "epoch": 279, "lr": 4.079984297901107e-05} +{"train_loss": 0.014967898838222027, "grad_norm": 0.0772809237241745, "global_step": 29622, "epoch": 279, "lr": 4.079692983289389e-05} +{"train_loss": 0.014496320858597755, "grad_norm": 0.05062062293291092, "global_step": 29623, "epoch": 279, "lr": 4.079401671911227e-05} +{"train_loss": 0.01462559960782528, "grad_norm": 0.08978183567523956, "global_step": 29624, "epoch": 279, "lr": 4.0791103637676494e-05} +{"train_loss": 0.015262640081346035, "grad_norm": 0.08332601934671402, "global_step": 29625, "epoch": 279, "lr": 4.078819058859677e-05} +{"train_loss": 0.015365134924650192, "grad_norm": 0.09917417168617249, "global_step": 29626, "epoch": 279, "lr": 4.078527757188333e-05} +{"train_loss": 0.014943422749638557, "grad_norm": 0.09999217838048935, "global_step": 29627, "epoch": 279, "lr": 4.078236458754644e-05} +{"train_loss": 0.012288892641663551, "grad_norm": 0.07474522292613983, "global_step": 29628, "epoch": 279, "lr": 4.0779451635596296e-05} +{"train_loss": 0.01603909581899643, "grad_norm": 0.09529680758714676, "global_step": 29629, "epoch": 279, "lr": 4.077653871604315e-05} +{"train_loss": 0.015365641564130783, "grad_norm": 0.06456822156906128, "global_step": 29630, "epoch": 279, "lr": 4.0773625828897234e-05} +{"train_loss": 0.014318505302071571, "grad_norm": 0.10623734444379807, "global_step": 29631, "epoch": 279, "lr": 4.0770712974168776e-05} +{"train_loss": 0.015915699303150177, "grad_norm": 0.08205311745405197, "global_step": 29632, "epoch": 279, "lr": 4.0767800151868026e-05} +{"train_loss": 0.01592586189508438, "grad_norm": 0.1259484589099884, "global_step": 29633, "epoch": 279, "lr": 4.076488736200523e-05} +{"train_loss": 0.014356772415339947, "grad_norm": 0.11148879677057266, "global_step": 29634, "epoch": 279, "lr": 4.076197460459059e-05} +{"train_loss": 0.014212055131793022, "grad_norm": 0.08361551910638809, "global_step": 29635, "epoch": 279, "lr": 4.0759061879634345e-05} +{"train_loss": 0.013793072663247585, "grad_norm": 0.0749175027012825, "global_step": 29636, "epoch": 279, "lr": 4.075614918714675e-05} +{"train_loss": 0.01482241041958332, "grad_norm": 0.07351550459861755, "global_step": 29637, "epoch": 279, "lr": 4.075323652713802e-05} +{"train_loss": 0.01599995233118534, "grad_norm": 0.08419254422187805, "global_step": 29638, "epoch": 279, "lr": 4.075032389961841e-05} +{"train_loss": 0.01735256239771843, "grad_norm": 0.10491392016410828, "global_step": 29639, "epoch": 279, "lr": 4.074741130459813e-05} +{"train_loss": 0.014545345678925514, "grad_norm": 0.06995130330324173, "global_step": 29640, "epoch": 279, "lr": 4.0744498742087405e-05} +{"train_loss": 0.013875668868422508, "grad_norm": 0.1432180255651474, "global_step": 29641, "epoch": 279, "lr": 4.07415862120965e-05} +{"train_loss": 0.01591898687183857, "grad_norm": 0.1399700939655304, "global_step": 29642, "epoch": 279, "lr": 4.0738673714635636e-05} +{"train_loss": 0.015287086367607117, "grad_norm": 0.0593363381922245, "global_step": 29643, "epoch": 279, "lr": 4.0735761249715034e-05} +{"train_loss": 0.015194021165370941, "grad_norm": 0.10353033244609833, "global_step": 29644, "epoch": 279, "lr": 4.073284881734496e-05} +{"train_loss": 0.013795956037938595, "grad_norm": 0.09013509005308151, "global_step": 29645, "epoch": 279, "lr": 4.072993641753561e-05} +{"train_loss": 0.016013719141483307, "grad_norm": 0.10814033448696136, "global_step": 29646, "epoch": 279, "lr": 4.072702405029724e-05} +{"train_loss": 0.015676235780119896, "grad_norm": 0.10807789862155914, "global_step": 29647, "epoch": 279, "lr": 4.072411171564006e-05} +{"train_loss": 0.014358915388584137, "grad_norm": 0.10123542696237564, "global_step": 29648, "epoch": 279, "lr": 4.072119941357433e-05} +{"train_loss": 0.013016406446695328, "grad_norm": 0.07780171185731888, "global_step": 29649, "epoch": 279, "lr": 4.0718287144110266e-05} +{"train_loss": 0.01675010845065117, "grad_norm": 0.09116441011428833, "global_step": 29650, "epoch": 279, "lr": 4.071537490725812e-05} +{"train_loss": 0.015439135953783989, "grad_norm": 0.07804267853498459, "global_step": 29651, "epoch": 279, "lr": 4.07124627030281e-05} +{"train_loss": 0.015270598232746124, "grad_norm": 0.09907136857509613, "global_step": 29652, "epoch": 279, "lr": 4.070955053143043e-05} +{"train_loss": 0.01734738051891327, "grad_norm": 0.08243060857057571, "global_step": 29653, "epoch": 279, "lr": 4.070663839247537e-05} +{"train_loss": 0.01537823025137186, "grad_norm": 0.06485806405544281, "global_step": 29654, "epoch": 279, "lr": 4.0703726286173136e-05} +{"train_loss": 0.013963762670755386, "grad_norm": 0.10365960747003555, "global_step": 29655, "epoch": 279, "lr": 4.070081421253396e-05} +{"train_loss": 0.015551784075796604, "grad_norm": 0.16284477710723877, "global_step": 29656, "epoch": 279, "lr": 4.069790217156812e-05} +{"train_loss": 0.01853695511817932, "grad_norm": 0.08508677035570145, "global_step": 29657, "epoch": 279, "lr": 4.069499016328576e-05} +{"train_loss": 0.014595039188861847, "grad_norm": 0.09873338043689728, "global_step": 29658, "epoch": 279, "lr": 4.069207818769718e-05} +{"train_loss": 0.012926062569022179, "grad_norm": 0.058791160583496094, "global_step": 29659, "epoch": 279, "lr": 4.068916624481258e-05} +{"train_loss": 0.016949402168393135, "grad_norm": 0.07266980409622192, "global_step": 29660, "epoch": 279, "lr": 4.06862543346422e-05} +{"train_loss": 0.014099279418587685, "grad_norm": 0.11632086336612701, "global_step": 29661, "epoch": 279, "lr": 4.068334245719627e-05} +{"train_loss": 0.014064018614590168, "grad_norm": 0.10379280894994736, "global_step": 29662, "epoch": 279, "lr": 4.0680430612485045e-05} +{"train_loss": 0.01544724777340889, "grad_norm": 0.0872945785522461, "global_step": 29663, "epoch": 279, "lr": 4.067751880051872e-05} +{"train_loss": 0.013421524316072464, "grad_norm": 0.055609554052352905, "global_step": 29664, "epoch": 279, "lr": 4.0674607021307525e-05} +{"train_loss": 0.014547780156135559, "grad_norm": 0.057472456246614456, "global_step": 29665, "epoch": 279, "lr": 4.067169527486172e-05} +{"train_loss": 0.015607139095664024, "grad_norm": 0.0920054242014885, "global_step": 29666, "epoch": 279, "lr": 4.066878356119152e-05} +{"train_loss": 0.013281227089464664, "grad_norm": 0.060400333255529404, "global_step": 29667, "epoch": 279, "lr": 4.0665871880307146e-05} +{"train_loss": 0.01467328704893589, "grad_norm": 0.0803227350115776, "global_step": 29668, "epoch": 279, "lr": 4.066296023221887e-05} +{"train_loss": 0.0148312421515584, "grad_norm": 0.08999823778867722, "global_step": 29669, "epoch": 279, "lr": 4.066004861693686e-05} +{"train_loss": 0.016371652483940125, "grad_norm": 0.09698393195867538, "global_step": 29670, "epoch": 279, "lr": 4.065713703447139e-05} +{"train_loss": 0.013736272230744362, "grad_norm": 0.1061948612332344, "global_step": 29671, "epoch": 279, "lr": 4.0654225484832677e-05} +{"train_loss": 0.015453455038368702, "grad_norm": 0.08561278879642487, "global_step": 29672, "epoch": 279, "lr": 4.065131396803094e-05} +{"train_loss": 0.018367242068052292, "grad_norm": 0.11742269992828369, "global_step": 29673, "epoch": 279, "lr": 4.064840248407642e-05} +{"train_loss": 0.018530676141381264, "grad_norm": 0.06704572588205338, "global_step": 29674, "epoch": 279, "lr": 4.0645491032979375e-05} +{"train_loss": 0.015007361769676208, "grad_norm": 0.056224044412374496, "global_step": 29675, "epoch": 279, "lr": 4.064257961474998e-05} +{"train_loss": 0.01870485581457615, "grad_norm": 0.06894560903310776, "global_step": 29676, "epoch": 279, "lr": 4.0639668229398486e-05} +{"train_loss": 0.015039818361401558, "grad_norm": 0.0648188665509224, "global_step": 29677, "epoch": 279, "lr": 4.0636756876935145e-05} +{"train_loss": 0.014489241875708103, "grad_norm": 0.12959451973438263, "global_step": 29678, "epoch": 279, "lr": 4.0633845557370156e-05} +{"train_loss": 0.015402103255871893, "grad_norm": 0.10353564471006393, "global_step": 29679, "epoch": 279, "lr": 4.063093427071376e-05, "val_loss": 0.01878144033253193} +{"train_loss": 0.017398934811353683, "grad_norm": 0.0952475517988205, "global_step": 29680, "epoch": 280, "lr": 4.0628023016976215e-05} +{"train_loss": 0.016594793647527695, "grad_norm": 0.06425143033266068, "global_step": 29681, "epoch": 280, "lr": 4.062511179616769e-05} +{"train_loss": 0.016008736565709114, "grad_norm": 0.08932135999202728, "global_step": 29682, "epoch": 280, "lr": 4.062220060829846e-05} +{"train_loss": 0.01688496768474579, "grad_norm": 0.07035776227712631, "global_step": 29683, "epoch": 280, "lr": 4.061928945337873e-05} +{"train_loss": 0.011729749850928783, "grad_norm": 0.06679034978151321, "global_step": 29684, "epoch": 280, "lr": 4.061637833141873e-05} +{"train_loss": 0.015918370336294174, "grad_norm": 0.06340747326612473, "global_step": 29685, "epoch": 280, "lr": 4.061346724242871e-05} +{"train_loss": 0.014678355306386948, "grad_norm": 0.06850495934486389, "global_step": 29686, "epoch": 280, "lr": 4.0610556186418894e-05} +{"train_loss": 0.017455678433179855, "grad_norm": 0.11195707321166992, "global_step": 29687, "epoch": 280, "lr": 4.060764516339948e-05} +{"train_loss": 0.013332676142454147, "grad_norm": 0.06526962667703629, "global_step": 29688, "epoch": 280, "lr": 4.0604734173380726e-05} +{"train_loss": 0.015600715763866901, "grad_norm": 0.08372124284505844, "global_step": 29689, "epoch": 280, "lr": 4.0601823216372836e-05} +{"train_loss": 0.0161898210644722, "grad_norm": 0.10894669592380524, "global_step": 29690, "epoch": 280, "lr": 4.0598912292386064e-05} +{"train_loss": 0.015749815851449966, "grad_norm": 0.13515225052833557, "global_step": 29691, "epoch": 280, "lr": 4.059600140143064e-05} +{"train_loss": 0.015675771981477737, "grad_norm": 0.08006196469068527, "global_step": 29692, "epoch": 280, "lr": 4.059309054351676e-05} +{"train_loss": 0.012421930208802223, "grad_norm": 0.05610749498009682, "global_step": 29693, "epoch": 280, "lr": 4.059017971865466e-05} +{"train_loss": 0.015564065426588058, "grad_norm": 0.09183825552463531, "global_step": 29694, "epoch": 280, "lr": 4.058726892685459e-05} +{"train_loss": 0.01623249426484108, "grad_norm": 0.06437645852565765, "global_step": 29695, "epoch": 280, "lr": 4.058435816812676e-05} +{"train_loss": 0.0166555717587471, "grad_norm": 0.07045499980449677, "global_step": 29696, "epoch": 280, "lr": 4.058144744248139e-05} +{"train_loss": 0.01362302154302597, "grad_norm": 0.10552800446748734, "global_step": 29697, "epoch": 280, "lr": 4.0578536749928754e-05} +{"train_loss": 0.01341081690043211, "grad_norm": 0.051956672221422195, "global_step": 29698, "epoch": 280, "lr": 4.0575626090479005e-05} +{"train_loss": 0.01571701094508171, "grad_norm": 0.09911597520112991, "global_step": 29699, "epoch": 280, "lr": 4.0572715464142416e-05} +{"train_loss": 0.01566966250538826, "grad_norm": 0.088900625705719, "global_step": 29700, "epoch": 280, "lr": 4.056980487092922e-05} +{"train_loss": 0.015563261695206165, "grad_norm": 0.09402809292078018, "global_step": 29701, "epoch": 280, "lr": 4.05668943108496e-05} +{"train_loss": 0.015572922304272652, "grad_norm": 0.07485317438840866, "global_step": 29702, "epoch": 280, "lr": 4.0563983783913834e-05} +{"train_loss": 0.014186129905283451, "grad_norm": 0.08659406006336212, "global_step": 29703, "epoch": 280, "lr": 4.056107329013213e-05} +{"train_loss": 0.01312580518424511, "grad_norm": 0.07049264758825302, "global_step": 29704, "epoch": 280, "lr": 4.0558162829514703e-05} +{"train_loss": 0.016620058566331863, "grad_norm": 0.09467746317386627, "global_step": 29705, "epoch": 280, "lr": 4.055525240207177e-05} +{"train_loss": 0.013156196102499962, "grad_norm": 0.0749656930565834, "global_step": 29706, "epoch": 280, "lr": 4.0552342007813584e-05} +{"train_loss": 0.017471451312303543, "grad_norm": 0.09428790211677551, "global_step": 29707, "epoch": 280, "lr": 4.054943164675036e-05} +{"train_loss": 0.013652155175805092, "grad_norm": 0.06849324703216553, "global_step": 29708, "epoch": 280, "lr": 4.054652131889232e-05} +{"train_loss": 0.015920255333185196, "grad_norm": 0.06557799130678177, "global_step": 29709, "epoch": 280, "lr": 4.054361102424972e-05} +{"train_loss": 0.016154695302248, "grad_norm": 0.07692921906709671, "global_step": 29710, "epoch": 280, "lr": 4.054070076283272e-05} +{"train_loss": 0.01635664328932762, "grad_norm": 0.10096082091331482, "global_step": 29711, "epoch": 280, "lr": 4.05377905346516e-05} +{"train_loss": 0.014877095818519592, "grad_norm": 0.07318662106990814, "global_step": 29712, "epoch": 280, "lr": 4.0534880339716566e-05} +{"train_loss": 0.015030089765787125, "grad_norm": 0.06624364107847214, "global_step": 29713, "epoch": 280, "lr": 4.053197017803784e-05} +{"train_loss": 0.013632595539093018, "grad_norm": 0.13180720806121826, "global_step": 29714, "epoch": 280, "lr": 4.052906004962566e-05} +{"train_loss": 0.018471764400601387, "grad_norm": 0.06414078176021576, "global_step": 29715, "epoch": 280, "lr": 4.0526149954490256e-05} +{"train_loss": 0.016207223758101463, "grad_norm": 0.10004827380180359, "global_step": 29716, "epoch": 280, "lr": 4.0523239892641835e-05} +{"train_loss": 0.01574457809329033, "grad_norm": 0.09981569647789001, "global_step": 29717, "epoch": 280, "lr": 4.0520329864090624e-05} +{"train_loss": 0.013669254258275032, "grad_norm": 0.06082158163189888, "global_step": 29718, "epoch": 280, "lr": 4.051741986884684e-05} +{"train_loss": 0.01353275403380394, "grad_norm": 0.13013555109500885, "global_step": 29719, "epoch": 280, "lr": 4.051450990692073e-05} +{"train_loss": 0.015035760588943958, "grad_norm": 0.05583425238728523, "global_step": 29720, "epoch": 280, "lr": 4.0511599978322496e-05} +{"train_loss": 0.016516845673322678, "grad_norm": 0.08664865791797638, "global_step": 29721, "epoch": 280, "lr": 4.050869008306241e-05} +{"train_loss": 0.01311410591006279, "grad_norm": 0.07221771776676178, "global_step": 29722, "epoch": 280, "lr": 4.050578022115062e-05} +{"train_loss": 0.01716797798871994, "grad_norm": 0.07833869010210037, "global_step": 29723, "epoch": 280, "lr": 4.0502870392597405e-05} +{"train_loss": 0.013713287189602852, "grad_norm": 0.09318812936544418, "global_step": 29724, "epoch": 280, "lr": 4.049996059741297e-05} +{"train_loss": 0.01552807167172432, "grad_norm": 0.06886059045791626, "global_step": 29725, "epoch": 280, "lr": 4.049705083560753e-05} +{"train_loss": 0.015397662296891212, "grad_norm": 0.13904264569282532, "global_step": 29726, "epoch": 280, "lr": 4.049414110719133e-05} +{"train_loss": 0.014593749307096004, "grad_norm": 0.07959523797035217, "global_step": 29727, "epoch": 280, "lr": 4.04912314121746e-05} +{"train_loss": 0.01353751216083765, "grad_norm": 0.06637827306985855, "global_step": 29728, "epoch": 280, "lr": 4.048832175056754e-05} +{"train_loss": 0.016108863055706024, "grad_norm": 0.10791119188070297, "global_step": 29729, "epoch": 280, "lr": 4.0485412122380374e-05} +{"train_loss": 0.013620292767882347, "grad_norm": 0.0885891392827034, "global_step": 29730, "epoch": 280, "lr": 4.0482502527623324e-05} +{"train_loss": 0.016475412994623184, "grad_norm": 0.0636383444070816, "global_step": 29731, "epoch": 280, "lr": 4.047959296630663e-05} +{"train_loss": 0.017554758116602898, "grad_norm": 0.07891888171434402, "global_step": 29732, "epoch": 280, "lr": 4.047668343844051e-05} +{"train_loss": 0.015284164808690548, "grad_norm": 0.07337524741888046, "global_step": 29733, "epoch": 280, "lr": 4.0473773944035186e-05} +{"train_loss": 0.014087008312344551, "grad_norm": 0.08258993923664093, "global_step": 29734, "epoch": 280, "lr": 4.0470864483100865e-05} +{"train_loss": 0.016215410083532333, "grad_norm": 0.10223587602376938, "global_step": 29735, "epoch": 280, "lr": 4.046795505564779e-05} +{"train_loss": 0.015523944981396198, "grad_norm": 0.06067387014627457, "global_step": 29736, "epoch": 280, "lr": 4.0465045661686176e-05} +{"train_loss": 0.014003974385559559, "grad_norm": 0.07789948582649231, "global_step": 29737, "epoch": 280, "lr": 4.0462136301226226e-05} +{"train_loss": 0.015481359325349331, "grad_norm": 0.05501081049442291, "global_step": 29738, "epoch": 280, "lr": 4.045922697427822e-05} +{"train_loss": 0.01539112813770771, "grad_norm": 0.09145242720842361, "global_step": 29739, "epoch": 280, "lr": 4.0456317680852314e-05} +{"train_loss": 0.01623529940843582, "grad_norm": 0.1297225058078766, "global_step": 29740, "epoch": 280, "lr": 4.045340842095876e-05} +{"train_loss": 0.014486938714981079, "grad_norm": 0.06151343137025833, "global_step": 29741, "epoch": 280, "lr": 4.0450499194607784e-05} +{"train_loss": 0.016423506662249565, "grad_norm": 0.13175031542778015, "global_step": 29742, "epoch": 280, "lr": 4.044759000180958e-05} +{"train_loss": 0.014620029367506504, "grad_norm": 0.06395341455936432, "global_step": 29743, "epoch": 280, "lr": 4.044468084257441e-05} +{"train_loss": 0.017600758001208305, "grad_norm": 0.07501306384801865, "global_step": 29744, "epoch": 280, "lr": 4.044177171691249e-05} +{"train_loss": 0.019055690616369247, "grad_norm": 0.10059612989425659, "global_step": 29745, "epoch": 280, "lr": 4.0438862624834006e-05} +{"train_loss": 0.014864569529891014, "grad_norm": 0.11667098104953766, "global_step": 29746, "epoch": 280, "lr": 4.04359535663492e-05} +{"train_loss": 0.014029424637556076, "grad_norm": 0.10868161916732788, "global_step": 29747, "epoch": 280, "lr": 4.043304454146829e-05} +{"train_loss": 0.014142833650112152, "grad_norm": 0.07921639084815979, "global_step": 29748, "epoch": 280, "lr": 4.043013555020151e-05} +{"train_loss": 0.015618538483977318, "grad_norm": 0.13517874479293823, "global_step": 29749, "epoch": 280, "lr": 4.0427226592559065e-05} +{"train_loss": 0.01599125564098358, "grad_norm": 0.08554074913263321, "global_step": 29750, "epoch": 280, "lr": 4.042431766855122e-05} +{"train_loss": 0.014704981818795204, "grad_norm": 0.08067584782838821, "global_step": 29751, "epoch": 280, "lr": 4.042140877818811e-05} +{"train_loss": 0.012101255357265472, "grad_norm": 0.07543127983808517, "global_step": 29752, "epoch": 280, "lr": 4.0418499921480025e-05} +{"train_loss": 0.017200006172060966, "grad_norm": 0.08733359724283218, "global_step": 29753, "epoch": 280, "lr": 4.0415591098437157e-05} +{"train_loss": 0.015445864759385586, "grad_norm": 0.1016322523355484, "global_step": 29754, "epoch": 280, "lr": 4.041268230906973e-05} +{"train_loss": 0.013455061241984367, "grad_norm": 0.07520417869091034, "global_step": 29755, "epoch": 280, "lr": 4.0409773553387977e-05} +{"train_loss": 0.0142157431691885, "grad_norm": 0.06423259526491165, "global_step": 29756, "epoch": 280, "lr": 4.0406864831402116e-05} +{"train_loss": 0.016650903970003128, "grad_norm": 0.16684851050376892, "global_step": 29757, "epoch": 280, "lr": 4.040395614312236e-05} +{"train_loss": 0.013514420948922634, "grad_norm": 0.1222839206457138, "global_step": 29758, "epoch": 280, "lr": 4.040104748855891e-05} +{"train_loss": 0.01459460984915495, "grad_norm": 0.10849642753601074, "global_step": 29759, "epoch": 280, "lr": 4.039813886772201e-05} +{"train_loss": 0.015847597271203995, "grad_norm": 0.078017957508564, "global_step": 29760, "epoch": 280, "lr": 4.0395230280621884e-05} +{"train_loss": 0.017923792824149132, "grad_norm": 0.179929718375206, "global_step": 29761, "epoch": 280, "lr": 4.0392321727268737e-05} +{"train_loss": 0.017906514927744865, "grad_norm": 0.12415698170661926, "global_step": 29762, "epoch": 280, "lr": 4.03894132076728e-05} +{"train_loss": 0.015700513496994972, "grad_norm": 0.08665144443511963, "global_step": 29763, "epoch": 280, "lr": 4.038650472184427e-05} +{"train_loss": 0.014714241027832031, "grad_norm": 0.06448553502559662, "global_step": 29764, "epoch": 280, "lr": 4.038359626979339e-05} +{"train_loss": 0.01605554297566414, "grad_norm": 0.06466027349233627, "global_step": 29765, "epoch": 280, "lr": 4.0380687851530376e-05} +{"train_loss": 0.015581213869154453, "grad_norm": 0.0839073657989502, "global_step": 29766, "epoch": 280, "lr": 4.0377779467065427e-05} +{"train_loss": 0.015393448062241077, "grad_norm": 0.07162447273731232, "global_step": 29767, "epoch": 280, "lr": 4.037487111640879e-05} +{"train_loss": 0.01887301355600357, "grad_norm": 0.09638268500566483, "global_step": 29768, "epoch": 280, "lr": 4.037196279957067e-05} +{"train_loss": 0.015037008561193943, "grad_norm": 0.07995239645242691, "global_step": 29769, "epoch": 280, "lr": 4.036905451656128e-05} +{"train_loss": 0.01587974652647972, "grad_norm": 0.1215425431728363, "global_step": 29770, "epoch": 280, "lr": 4.0366146267390845e-05} +{"train_loss": 0.01710369996726513, "grad_norm": 0.16967397928237915, "global_step": 29771, "epoch": 280, "lr": 4.036323805206958e-05} +{"train_loss": 0.016097115352749825, "grad_norm": 0.15996821224689484, "global_step": 29772, "epoch": 280, "lr": 4.0360329870607704e-05} +{"train_loss": 0.01463058777153492, "grad_norm": 0.09839537739753723, "global_step": 29773, "epoch": 280, "lr": 4.035742172301544e-05} +{"train_loss": 0.017068766057491302, "grad_norm": 0.22497904300689697, "global_step": 29774, "epoch": 280, "lr": 4.0354513609303024e-05} +{"train_loss": 0.01656666211783886, "grad_norm": 0.1295560598373413, "global_step": 29775, "epoch": 280, "lr": 4.0351605529480616e-05} +{"train_loss": 0.01610700413584709, "grad_norm": 0.06996387988328934, "global_step": 29776, "epoch": 280, "lr": 4.034869748355849e-05} +{"train_loss": 0.012982061132788658, "grad_norm": 0.09678523987531662, "global_step": 29777, "epoch": 280, "lr": 4.034578947154684e-05} +{"train_loss": 0.015951499342918396, "grad_norm": 0.10587340593338013, "global_step": 29778, "epoch": 280, "lr": 4.034288149345588e-05} +{"train_loss": 0.015456823632121086, "grad_norm": 0.06418726593255997, "global_step": 29779, "epoch": 280, "lr": 4.033997354929584e-05} +{"train_loss": 0.01664547622203827, "grad_norm": 0.10528377443552017, "global_step": 29780, "epoch": 280, "lr": 4.033706563907695e-05} +{"train_loss": 0.012778883799910545, "grad_norm": 0.05483781918883324, "global_step": 29781, "epoch": 280, "lr": 4.033415776280939e-05} +{"train_loss": 0.013831615447998047, "grad_norm": 0.053348515182733536, "global_step": 29782, "epoch": 280, "lr": 4.03312499205034e-05} +{"train_loss": 0.014047699049115181, "grad_norm": 0.053499337285757065, "global_step": 29783, "epoch": 280, "lr": 4.0328342112169185e-05} +{"train_loss": 0.015634655952453613, "grad_norm": 0.22566556930541992, "global_step": 29784, "epoch": 280, "lr": 4.032543433781698e-05} +{"train_loss": 0.015424324471925228, "grad_norm": 0.13831152021884918, "global_step": 29785, "epoch": 280, "lr": 4.032252659745699e-05, "train_batch_mse_error": 0.013852579519152641, "val_loss": 0.019218025729060173} +{"train_loss": 0.016258813440799713, "grad_norm": 0.07433492690324783, "global_step": 29786, "epoch": 281, "lr": 4.0319618891099445e-05} +{"train_loss": 0.016192980110645294, "grad_norm": 0.14718583226203918, "global_step": 29787, "epoch": 281, "lr": 4.0316711218754536e-05} +{"train_loss": 0.015360163524746895, "grad_norm": 0.14003999531269073, "global_step": 29788, "epoch": 281, "lr": 4.031380358043248e-05} +{"train_loss": 0.01581227034330368, "grad_norm": 0.1357421875, "global_step": 29789, "epoch": 281, "lr": 4.0310895976143517e-05} +{"train_loss": 0.01639864221215248, "grad_norm": 0.05763864517211914, "global_step": 29790, "epoch": 281, "lr": 4.030798840589786e-05} +{"train_loss": 0.015115944668650627, "grad_norm": 0.12670636177062988, "global_step": 29791, "epoch": 281, "lr": 4.0305080869705715e-05} +{"train_loss": 0.014715568162500858, "grad_norm": 0.10740336775779724, "global_step": 29792, "epoch": 281, "lr": 4.030217336757728e-05} +{"train_loss": 0.01453973725438118, "grad_norm": 0.0932590588927269, "global_step": 29793, "epoch": 281, "lr": 4.0299265899522806e-05} +{"train_loss": 0.014622123911976814, "grad_norm": 0.1123754158616066, "global_step": 29794, "epoch": 281, "lr": 4.02963584655525e-05} +{"train_loss": 0.014145564287900925, "grad_norm": 0.08013828098773956, "global_step": 29795, "epoch": 281, "lr": 4.0293451065676545e-05} +{"train_loss": 0.01529469434171915, "grad_norm": 0.09108706563711166, "global_step": 29796, "epoch": 281, "lr": 4.02905436999052e-05} +{"train_loss": 0.01394985243678093, "grad_norm": 0.08938659727573395, "global_step": 29797, "epoch": 281, "lr": 4.0287636368248674e-05} +{"train_loss": 0.015222582966089249, "grad_norm": 0.10835165530443192, "global_step": 29798, "epoch": 281, "lr": 4.028472907071716e-05} +{"train_loss": 0.015221530571579933, "grad_norm": 0.10267005115747452, "global_step": 29799, "epoch": 281, "lr": 4.028182180732088e-05} +{"train_loss": 0.016922231763601303, "grad_norm": 0.11157328635454178, "global_step": 29800, "epoch": 281, "lr": 4.027891457807004e-05} +{"train_loss": 0.014327493496239185, "grad_norm": 0.08155220001935959, "global_step": 29801, "epoch": 281, "lr": 4.027600738297488e-05} +{"train_loss": 0.013904335908591747, "grad_norm": 0.14436060190200806, "global_step": 29802, "epoch": 281, "lr": 4.027310022204559e-05} +{"train_loss": 0.015127969905734062, "grad_norm": 0.0750047042965889, "global_step": 29803, "epoch": 281, "lr": 4.0270193095292427e-05} +{"train_loss": 0.016135836020112038, "grad_norm": 0.08606347441673279, "global_step": 29804, "epoch": 281, "lr": 4.0267286002725544e-05} +{"train_loss": 0.014240579679608345, "grad_norm": 0.11270267516374588, "global_step": 29805, "epoch": 281, "lr": 4.026437894435518e-05} +{"train_loss": 0.01666741445660591, "grad_norm": 0.049102190881967545, "global_step": 29806, "epoch": 281, "lr": 4.026147192019158e-05} +{"train_loss": 0.014480560086667538, "grad_norm": 0.08038902282714844, "global_step": 29807, "epoch": 281, "lr": 4.0258564930244904e-05} +{"train_loss": 0.015774058178067207, "grad_norm": 0.0912134051322937, "global_step": 29808, "epoch": 281, "lr": 4.0255657974525416e-05} +{"train_loss": 0.01791631057858467, "grad_norm": 0.12424705922603607, "global_step": 29809, "epoch": 281, "lr": 4.025275105304331e-05} +{"train_loss": 0.014779018238186836, "grad_norm": 0.0800357535481453, "global_step": 29810, "epoch": 281, "lr": 4.024984416580879e-05} +{"train_loss": 0.01487310416996479, "grad_norm": 0.10333342850208282, "global_step": 29811, "epoch": 281, "lr": 4.024693731283209e-05} +{"train_loss": 0.015011216513812542, "grad_norm": 0.061885979026556015, "global_step": 29812, "epoch": 281, "lr": 4.024403049412338e-05} +{"train_loss": 0.014525867998600006, "grad_norm": 0.08707500249147415, "global_step": 29813, "epoch": 281, "lr": 4.0241123709692924e-05} +{"train_loss": 0.01529559213668108, "grad_norm": 0.06496673822402954, "global_step": 29814, "epoch": 281, "lr": 4.0238216959550914e-05} +{"train_loss": 0.01606527343392372, "grad_norm": 0.08415743708610535, "global_step": 29815, "epoch": 281, "lr": 4.023531024370758e-05} +{"train_loss": 0.015709737315773964, "grad_norm": 0.0961444303393364, "global_step": 29816, "epoch": 281, "lr": 4.0232403562173103e-05} +{"train_loss": 0.014795922674238682, "grad_norm": 0.058983538299798965, "global_step": 29817, "epoch": 281, "lr": 4.0229496914957706e-05} +{"train_loss": 0.014601669274270535, "grad_norm": 0.09896251559257507, "global_step": 29818, "epoch": 281, "lr": 4.022659030207161e-05} +{"train_loss": 0.015790261328220367, "grad_norm": 0.10396619886159897, "global_step": 29819, "epoch": 281, "lr": 4.022368372352503e-05} +{"train_loss": 0.015450049191713333, "grad_norm": 0.09142641723155975, "global_step": 29820, "epoch": 281, "lr": 4.022077717932817e-05} +{"train_loss": 0.017926082015037537, "grad_norm": 0.07407435774803162, "global_step": 29821, "epoch": 281, "lr": 4.021787066949126e-05} +{"train_loss": 0.016227025538682938, "grad_norm": 0.06405235826969147, "global_step": 29822, "epoch": 281, "lr": 4.0214964194024486e-05} +{"train_loss": 0.013426161371171474, "grad_norm": 0.11649326980113983, "global_step": 29823, "epoch": 281, "lr": 4.021205775293807e-05} +{"train_loss": 0.015139210037887096, "grad_norm": 0.1916818618774414, "global_step": 29824, "epoch": 281, "lr": 4.020915134624223e-05} +{"train_loss": 0.016293806955218315, "grad_norm": 0.1454629749059677, "global_step": 29825, "epoch": 281, "lr": 4.020624497394717e-05} +{"train_loss": 0.01580960303544998, "grad_norm": 0.07481163740158081, "global_step": 29826, "epoch": 281, "lr": 4.0203338636063114e-05} +{"train_loss": 0.015113530680537224, "grad_norm": 0.11904505640268326, "global_step": 29827, "epoch": 281, "lr": 4.020043233260028e-05} +{"train_loss": 0.015147496946156025, "grad_norm": 0.09141053259372711, "global_step": 29828, "epoch": 281, "lr": 4.019752606356885e-05} +{"train_loss": 0.014736086130142212, "grad_norm": 0.07701902091503143, "global_step": 29829, "epoch": 281, "lr": 4.019461982897903e-05} +{"train_loss": 0.015185471624135971, "grad_norm": 0.13938844203948975, "global_step": 29830, "epoch": 281, "lr": 4.0191713628841074e-05} +{"train_loss": 0.014489117078483105, "grad_norm": 0.057765983045101166, "global_step": 29831, "epoch": 281, "lr": 4.018880746316517e-05} +{"train_loss": 0.015806686133146286, "grad_norm": 0.062042780220508575, "global_step": 29832, "epoch": 281, "lr": 4.018590133196151e-05} +{"train_loss": 0.01610027812421322, "grad_norm": 0.07333832234144211, "global_step": 29833, "epoch": 281, "lr": 4.018299523524036e-05} +{"train_loss": 0.016424335539340973, "grad_norm": 0.11290613561868668, "global_step": 29834, "epoch": 281, "lr": 4.018008917301187e-05} +{"train_loss": 0.014656584709882736, "grad_norm": 0.08069978654384613, "global_step": 29835, "epoch": 281, "lr": 4.017718314528628e-05} +{"train_loss": 0.013706730678677559, "grad_norm": 0.09089525789022446, "global_step": 29836, "epoch": 281, "lr": 4.01742771520738e-05} +{"train_loss": 0.015829039737582207, "grad_norm": 0.08754734694957733, "global_step": 29837, "epoch": 281, "lr": 4.0171371193384636e-05} +{"train_loss": 0.01246909610927105, "grad_norm": 0.09119552373886108, "global_step": 29838, "epoch": 281, "lr": 4.0168465269229007e-05} +{"train_loss": 0.01579543948173523, "grad_norm": 0.11298487335443497, "global_step": 29839, "epoch": 281, "lr": 4.016555937961711e-05} +{"train_loss": 0.013831594958901405, "grad_norm": 0.15858334302902222, "global_step": 29840, "epoch": 281, "lr": 4.0162653524559155e-05} +{"train_loss": 0.016284063458442688, "grad_norm": 0.09310844540596008, "global_step": 29841, "epoch": 281, "lr": 4.0159747704065356e-05} +{"train_loss": 0.017529183998703957, "grad_norm": 0.06763456761837006, "global_step": 29842, "epoch": 281, "lr": 4.0156841918145924e-05} +{"train_loss": 0.016922079026699066, "grad_norm": 0.13500908017158508, "global_step": 29843, "epoch": 281, "lr": 4.015393616681108e-05} +{"train_loss": 0.015987606719136238, "grad_norm": 0.07327516376972198, "global_step": 29844, "epoch": 281, "lr": 4.015103045007103e-05} +{"train_loss": 0.017311889678239822, "grad_norm": 0.07406497001647949, "global_step": 29845, "epoch": 281, "lr": 4.0148124767935955e-05} +{"train_loss": 0.01470071543008089, "grad_norm": 0.10086169838905334, "global_step": 29846, "epoch": 281, "lr": 4.014521912041609e-05} +{"train_loss": 0.016692105680704117, "grad_norm": 0.1009160727262497, "global_step": 29847, "epoch": 281, "lr": 4.0142313507521636e-05} +{"train_loss": 0.015936581417918205, "grad_norm": 0.06480618566274643, "global_step": 29848, "epoch": 281, "lr": 4.0139407929262814e-05} +{"train_loss": 0.016185497865080833, "grad_norm": 0.08015122264623642, "global_step": 29849, "epoch": 281, "lr": 4.013650238564981e-05} +{"train_loss": 0.014694176614284515, "grad_norm": 0.07249771803617477, "global_step": 29850, "epoch": 281, "lr": 4.013359687669287e-05} +{"train_loss": 0.014433194883167744, "grad_norm": 0.07696995884180069, "global_step": 29851, "epoch": 281, "lr": 4.0130691402402165e-05} +{"train_loss": 0.012959640473127365, "grad_norm": 0.08209279924631119, "global_step": 29852, "epoch": 281, "lr": 4.0127785962787926e-05} +{"train_loss": 0.01589396595954895, "grad_norm": 0.06817827373743057, "global_step": 29853, "epoch": 281, "lr": 4.012488055786034e-05} +{"train_loss": 0.015808822587132454, "grad_norm": 0.10461927205324173, "global_step": 29854, "epoch": 281, "lr": 4.0121975187629634e-05} +{"train_loss": 0.015077630057930946, "grad_norm": 0.07715585082769394, "global_step": 29855, "epoch": 281, "lr": 4.011906985210602e-05} +{"train_loss": 0.015378416515886784, "grad_norm": 0.06559944152832031, "global_step": 29856, "epoch": 281, "lr": 4.0116164551299704e-05} +{"train_loss": 0.01505021657794714, "grad_norm": 0.0921771228313446, "global_step": 29857, "epoch": 281, "lr": 4.011325928522088e-05} +{"train_loss": 0.01494581438601017, "grad_norm": 0.07034538686275482, "global_step": 29858, "epoch": 281, "lr": 4.011035405387975e-05} +{"train_loss": 0.014874096028506756, "grad_norm": 0.07015722244977951, "global_step": 29859, "epoch": 281, "lr": 4.0107448857286545e-05} +{"train_loss": 0.016582271084189415, "grad_norm": 0.06752418726682663, "global_step": 29860, "epoch": 281, "lr": 4.0104543695451454e-05} +{"train_loss": 0.017599375918507576, "grad_norm": 0.11976142972707748, "global_step": 29861, "epoch": 281, "lr": 4.0101638568384694e-05} +{"train_loss": 0.01448705792427063, "grad_norm": 0.06704989820718765, "global_step": 29862, "epoch": 281, "lr": 4.00987334760965e-05} +{"train_loss": 0.016870152205228806, "grad_norm": 0.09295535832643509, "global_step": 29863, "epoch": 281, "lr": 4.009582841859703e-05} +{"train_loss": 0.014928951859474182, "grad_norm": 0.07523646205663681, "global_step": 29864, "epoch": 281, "lr": 4.0092923395896504e-05} +{"train_loss": 0.015918398275971413, "grad_norm": 0.05408979952335358, "global_step": 29865, "epoch": 281, "lr": 4.009001840800514e-05} +{"train_loss": 0.014025688171386719, "grad_norm": 0.09438318014144897, "global_step": 29866, "epoch": 281, "lr": 4.008711345493314e-05} +{"train_loss": 0.0142221599817276, "grad_norm": 0.09508363902568817, "global_step": 29867, "epoch": 281, "lr": 4.0084208536690714e-05} +{"train_loss": 0.014434502460062504, "grad_norm": 0.061212968081235886, "global_step": 29868, "epoch": 281, "lr": 4.0081303653288086e-05} +{"train_loss": 0.015171857550740242, "grad_norm": 0.07710292935371399, "global_step": 29869, "epoch": 281, "lr": 4.007839880473543e-05} +{"train_loss": 0.014424271881580353, "grad_norm": 0.07811453938484192, "global_step": 29870, "epoch": 281, "lr": 4.007549399104295e-05} +{"train_loss": 0.018801860511302948, "grad_norm": 0.11805461347103119, "global_step": 29871, "epoch": 281, "lr": 4.007258921222088e-05} +{"train_loss": 0.014954810030758381, "grad_norm": 0.08609165996313095, "global_step": 29872, "epoch": 281, "lr": 4.0069684468279424e-05} +{"train_loss": 0.013336519710719585, "grad_norm": 0.1164848580956459, "global_step": 29873, "epoch": 281, "lr": 4.0066779759228756e-05} +{"train_loss": 0.013440944254398346, "grad_norm": 0.08907683938741684, "global_step": 29874, "epoch": 281, "lr": 4.006387508507914e-05} +{"train_loss": 0.013813307508826256, "grad_norm": 0.05520330369472504, "global_step": 29875, "epoch": 281, "lr": 4.006097044584072e-05} +{"train_loss": 0.016985461115837097, "grad_norm": 0.08334307372570038, "global_step": 29876, "epoch": 281, "lr": 4.005806584152373e-05} +{"train_loss": 0.016779480502009392, "grad_norm": 0.05545276030898094, "global_step": 29877, "epoch": 281, "lr": 4.0055161272138366e-05} +{"train_loss": 0.014005362056195736, "grad_norm": 0.1600244641304016, "global_step": 29878, "epoch": 281, "lr": 4.005225673769485e-05} +{"train_loss": 0.017149444669485092, "grad_norm": 0.08079303801059723, "global_step": 29879, "epoch": 281, "lr": 4.0049352238203376e-05} +{"train_loss": 0.016185510903596878, "grad_norm": 0.08200245350599289, "global_step": 29880, "epoch": 281, "lr": 4.004644777367416e-05} +{"train_loss": 0.01672523468732834, "grad_norm": 0.06014280766248703, "global_step": 29881, "epoch": 281, "lr": 4.004354334411739e-05} +{"train_loss": 0.015243550762534142, "grad_norm": 0.048371437937021255, "global_step": 29882, "epoch": 281, "lr": 4.004063894954327e-05} +{"train_loss": 0.014671796932816505, "grad_norm": 0.05773431435227394, "global_step": 29883, "epoch": 281, "lr": 4.0037734589962016e-05} +{"train_loss": 0.01687200367450714, "grad_norm": 0.06538427621126175, "global_step": 29884, "epoch": 281, "lr": 4.003483026538384e-05} +{"train_loss": 0.017322756350040436, "grad_norm": 0.09407337009906769, "global_step": 29885, "epoch": 281, "lr": 4.003192597581892e-05} +{"train_loss": 0.014474459923803806, "grad_norm": 0.0706738755106926, "global_step": 29886, "epoch": 281, "lr": 4.002902172127751e-05} +{"train_loss": 0.0128503842279315, "grad_norm": 0.07368865609169006, "global_step": 29887, "epoch": 281, "lr": 4.0026117501769754e-05} +{"train_loss": 0.015447646379470825, "grad_norm": 0.11324260383844376, "global_step": 29888, "epoch": 281, "lr": 4.002321331730589e-05} +{"train_loss": 0.014720620587468147, "grad_norm": 0.11324828863143921, "global_step": 29889, "epoch": 281, "lr": 4.002030916789612e-05} +{"train_loss": 0.015799514949321747, "grad_norm": 0.08366817235946655, "global_step": 29890, "epoch": 281, "lr": 4.001740505355063e-05} +{"train_loss": 0.015386793217709605, "grad_norm": 0.12238389253616333, "global_step": 29891, "epoch": 281, "lr": 4.001450097427966e-05, "val_loss": 0.01951020024716854} +{"train_loss": 0.01567712239921093, "grad_norm": 0.1133229061961174, "global_step": 29892, "epoch": 282, "lr": 4.0011596930093375e-05} +{"train_loss": 0.014397552236914635, "grad_norm": 0.08472545444965363, "global_step": 29893, "epoch": 282, "lr": 4.0008692921002e-05} +{"train_loss": 0.014290458522737026, "grad_norm": 0.0845695286989212, "global_step": 29894, "epoch": 282, "lr": 4.000578894701572e-05} +{"train_loss": 0.014729375019669533, "grad_norm": 0.10681792348623276, "global_step": 29895, "epoch": 282, "lr": 4.0002885008144766e-05} +{"train_loss": 0.017480552196502686, "grad_norm": 0.07523535937070847, "global_step": 29896, "epoch": 282, "lr": 3.999998110439932e-05} +{"train_loss": 0.016742998734116554, "grad_norm": 0.15349148213863373, "global_step": 29897, "epoch": 282, "lr": 3.9997077235789606e-05} +{"train_loss": 0.016192439943552017, "grad_norm": 0.12504643201828003, "global_step": 29898, "epoch": 282, "lr": 3.999417340232579e-05} +{"train_loss": 0.015612713992595673, "grad_norm": 0.1027119979262352, "global_step": 29899, "epoch": 282, "lr": 3.99912696040181e-05} +{"train_loss": 0.015876013785600662, "grad_norm": 0.08158618956804276, "global_step": 29900, "epoch": 282, "lr": 3.998836584087675e-05} +{"train_loss": 0.014171337708830833, "grad_norm": 0.06625012308359146, "global_step": 29901, "epoch": 282, "lr": 3.9985462112911926e-05} +{"train_loss": 0.016960281878709793, "grad_norm": 0.14205466210842133, "global_step": 29902, "epoch": 282, "lr": 3.9982558420133814e-05} +{"train_loss": 0.014159169048070908, "grad_norm": 0.05953915789723396, "global_step": 29903, "epoch": 282, "lr": 3.997965476255267e-05} +{"train_loss": 0.01387291494756937, "grad_norm": 0.05386253446340561, "global_step": 29904, "epoch": 282, "lr": 3.997675114017864e-05} +{"train_loss": 0.017046509310603142, "grad_norm": 0.13134045898914337, "global_step": 29905, "epoch": 282, "lr": 3.9973847553021945e-05} +{"train_loss": 0.016541052609682083, "grad_norm": 0.051121365278959274, "global_step": 29906, "epoch": 282, "lr": 3.997094400109278e-05} +{"train_loss": 0.01723024994134903, "grad_norm": 0.06739833950996399, "global_step": 29907, "epoch": 282, "lr": 3.9968040484401375e-05} +{"train_loss": 0.014696065336465836, "grad_norm": 0.05470532178878784, "global_step": 29908, "epoch": 282, "lr": 3.996513700295791e-05} +{"train_loss": 0.013513743877410889, "grad_norm": 0.10036435723304749, "global_step": 29909, "epoch": 282, "lr": 3.99622335567726e-05} +{"train_loss": 0.015353255905210972, "grad_norm": 0.06857803463935852, "global_step": 29910, "epoch": 282, "lr": 3.995933014585562e-05} +{"train_loss": 0.015308413654565811, "grad_norm": 0.06650937348604202, "global_step": 29911, "epoch": 282, "lr": 3.995642677021718e-05} +{"train_loss": 0.014687791466712952, "grad_norm": 0.07694265991449356, "global_step": 29912, "epoch": 282, "lr": 3.99535234298675e-05} +{"train_loss": 0.013793276622891426, "grad_norm": 0.08889661729335785, "global_step": 29913, "epoch": 282, "lr": 3.995062012481677e-05} +{"train_loss": 0.014747243374586105, "grad_norm": 0.1405337005853653, "global_step": 29914, "epoch": 282, "lr": 3.994771685507518e-05} +{"train_loss": 0.014169272035360336, "grad_norm": 0.07732043415307999, "global_step": 29915, "epoch": 282, "lr": 3.994481362065297e-05} +{"train_loss": 0.014632963575422764, "grad_norm": 0.0625431016087532, "global_step": 29916, "epoch": 282, "lr": 3.994191042156028e-05} +{"train_loss": 0.016461875289678574, "grad_norm": 0.0720742866396904, "global_step": 29917, "epoch": 282, "lr": 3.9939007257807357e-05} +{"train_loss": 0.015703849494457245, "grad_norm": 0.08745894581079483, "global_step": 29918, "epoch": 282, "lr": 3.993610412940439e-05} +{"train_loss": 0.013867436908185482, "grad_norm": 0.1013903096318245, "global_step": 29919, "epoch": 282, "lr": 3.9933201036361554e-05} +{"train_loss": 0.014910636469721794, "grad_norm": 0.12372034043073654, "global_step": 29920, "epoch": 282, "lr": 3.99302979786891e-05} +{"train_loss": 0.01581783965229988, "grad_norm": 0.06244837865233421, "global_step": 29921, "epoch": 282, "lr": 3.9927394956397196e-05} +{"train_loss": 0.014907062985002995, "grad_norm": 0.0780249759554863, "global_step": 29922, "epoch": 282, "lr": 3.992449196949604e-05} +{"train_loss": 0.01549357920885086, "grad_norm": 0.1785062849521637, "global_step": 29923, "epoch": 282, "lr": 3.992158901799582e-05} +{"train_loss": 0.01388093363493681, "grad_norm": 0.060799334198236465, "global_step": 29924, "epoch": 282, "lr": 3.9918686101906774e-05} +{"train_loss": 0.01609976217150688, "grad_norm": 0.11911050975322723, "global_step": 29925, "epoch": 282, "lr": 3.991578322123908e-05} +{"train_loss": 0.014489318244159222, "grad_norm": 0.125274196267128, "global_step": 29926, "epoch": 282, "lr": 3.991288037600292e-05} +{"train_loss": 0.015273296274244785, "grad_norm": 0.11741528660058975, "global_step": 29927, "epoch": 282, "lr": 3.990997756620855e-05} +{"train_loss": 0.017178025096654892, "grad_norm": 0.10827770829200745, "global_step": 29928, "epoch": 282, "lr": 3.9907074791866096e-05} +{"train_loss": 0.013341193087399006, "grad_norm": 0.11190946400165558, "global_step": 29929, "epoch": 282, "lr": 3.99041720529858e-05} +{"train_loss": 0.01274595595896244, "grad_norm": 0.06860141456127167, "global_step": 29930, "epoch": 282, "lr": 3.9901269349577855e-05} +{"train_loss": 0.013941410928964615, "grad_norm": 0.07839372009038925, "global_step": 29931, "epoch": 282, "lr": 3.989836668165245e-05} +{"train_loss": 0.017121771350502968, "grad_norm": 0.13556551933288574, "global_step": 29932, "epoch": 282, "lr": 3.9895464049219796e-05} +{"train_loss": 0.015783045440912247, "grad_norm": 0.058886539191007614, "global_step": 29933, "epoch": 282, "lr": 3.9892561452290106e-05} +{"train_loss": 0.016303349286317825, "grad_norm": 0.09267594665288925, "global_step": 29934, "epoch": 282, "lr": 3.988965889087355e-05} +{"train_loss": 0.015117300674319267, "grad_norm": 0.09077513962984085, "global_step": 29935, "epoch": 282, "lr": 3.9886756364980316e-05} +{"train_loss": 0.014739790931344032, "grad_norm": 0.10267874598503113, "global_step": 29936, "epoch": 282, "lr": 3.988385387462064e-05} +{"train_loss": 0.015487183816730976, "grad_norm": 0.06161818653345108, "global_step": 29937, "epoch": 282, "lr": 3.988095141980469e-05} +{"train_loss": 0.017958011478185654, "grad_norm": 0.11374340206384659, "global_step": 29938, "epoch": 282, "lr": 3.98780490005427e-05} +{"train_loss": 0.019559413194656372, "grad_norm": 0.09265576303005219, "global_step": 29939, "epoch": 282, "lr": 3.9875146616844825e-05} +{"train_loss": 0.017043907195329666, "grad_norm": 0.08272260427474976, "global_step": 29940, "epoch": 282, "lr": 3.987224426872128e-05} +{"train_loss": 0.014816811308264732, "grad_norm": 0.06445331871509552, "global_step": 29941, "epoch": 282, "lr": 3.9869341956182263e-05} +{"train_loss": 0.01673867180943489, "grad_norm": 0.08031754195690155, "global_step": 29942, "epoch": 282, "lr": 3.9866439679237985e-05} +{"train_loss": 0.014020996168255806, "grad_norm": 0.06262647360563278, "global_step": 29943, "epoch": 282, "lr": 3.9863537437898615e-05} +{"train_loss": 0.015594061464071274, "grad_norm": 0.10503742098808289, "global_step": 29944, "epoch": 282, "lr": 3.986063523217439e-05} +{"train_loss": 0.01297377236187458, "grad_norm": 0.11571663618087769, "global_step": 29945, "epoch": 282, "lr": 3.985773306207547e-05} +{"train_loss": 0.016595399007201195, "grad_norm": 0.10119113326072693, "global_step": 29946, "epoch": 282, "lr": 3.985483092761206e-05} +{"train_loss": 0.016726983711123466, "grad_norm": 0.12043026834726334, "global_step": 29947, "epoch": 282, "lr": 3.985192882879436e-05} +{"train_loss": 0.016342666000127792, "grad_norm": 0.10129199922084808, "global_step": 29948, "epoch": 282, "lr": 3.984902676563257e-05} +{"train_loss": 0.016614561900496483, "grad_norm": 0.11432153731584549, "global_step": 29949, "epoch": 282, "lr": 3.984612473813689e-05} +{"train_loss": 0.013962590135633945, "grad_norm": 0.10636874288320541, "global_step": 29950, "epoch": 282, "lr": 3.984322274631752e-05} +{"train_loss": 0.01296955719590187, "grad_norm": 0.08177277445793152, "global_step": 29951, "epoch": 282, "lr": 3.984032079018464e-05} +{"train_loss": 0.014835440553724766, "grad_norm": 0.11505062878131866, "global_step": 29952, "epoch": 282, "lr": 3.983741886974845e-05} +{"train_loss": 0.014050627127289772, "grad_norm": 0.08196557313203812, "global_step": 29953, "epoch": 282, "lr": 3.983451698501916e-05} +{"train_loss": 0.015822172164916992, "grad_norm": 0.10071564465761185, "global_step": 29954, "epoch": 282, "lr": 3.983161513600695e-05} +{"train_loss": 0.012984133325517178, "grad_norm": 0.07581306993961334, "global_step": 29955, "epoch": 282, "lr": 3.982871332272201e-05} +{"train_loss": 0.014539437368512154, "grad_norm": 0.10652455687522888, "global_step": 29956, "epoch": 282, "lr": 3.982581154517459e-05} +{"train_loss": 0.01809021085500717, "grad_norm": 0.07498767971992493, "global_step": 29957, "epoch": 282, "lr": 3.98229098033748e-05} +{"train_loss": 0.016800448298454285, "grad_norm": 0.0772048756480217, "global_step": 29958, "epoch": 282, "lr": 3.982000809733289e-05} +{"train_loss": 0.013917427510023117, "grad_norm": 0.10015641152858734, "global_step": 29959, "epoch": 282, "lr": 3.981710642705906e-05} +{"train_loss": 0.01443257462233305, "grad_norm": 0.10895968228578568, "global_step": 29960, "epoch": 282, "lr": 3.9814204792563464e-05} +{"train_loss": 0.0159728042781353, "grad_norm": 0.10035080462694168, "global_step": 29961, "epoch": 282, "lr": 3.981130319385634e-05} +{"train_loss": 0.014808109030127525, "grad_norm": 0.12801611423492432, "global_step": 29962, "epoch": 282, "lr": 3.9808401630947876e-05} +{"train_loss": 0.01907459646463394, "grad_norm": 0.17699678242206573, "global_step": 29963, "epoch": 282, "lr": 3.980550010384824e-05} +{"train_loss": 0.015216049738228321, "grad_norm": 0.0736667737364769, "global_step": 29964, "epoch": 282, "lr": 3.980259861256764e-05} +{"train_loss": 0.012819558382034302, "grad_norm": 0.10441450029611588, "global_step": 29965, "epoch": 282, "lr": 3.979969715711628e-05} +{"train_loss": 0.017540432512760162, "grad_norm": 0.10077661275863647, "global_step": 29966, "epoch": 282, "lr": 3.979679573750434e-05} +{"train_loss": 0.01508511696010828, "grad_norm": 0.10187029093503952, "global_step": 29967, "epoch": 282, "lr": 3.979389435374203e-05} +{"train_loss": 0.014953571371734142, "grad_norm": 0.22046828269958496, "global_step": 29968, "epoch": 282, "lr": 3.979099300583956e-05} +{"train_loss": 0.01454211212694645, "grad_norm": 0.12830182909965515, "global_step": 29969, "epoch": 282, "lr": 3.978809169380707e-05} +{"train_loss": 0.014577032998204231, "grad_norm": 0.07464036345481873, "global_step": 29970, "epoch": 282, "lr": 3.9785190417654795e-05} +{"train_loss": 0.0164039209485054, "grad_norm": 0.07231815159320831, "global_step": 29971, "epoch": 282, "lr": 3.9782289177392914e-05} +{"train_loss": 0.01742415875196457, "grad_norm": 0.08018889278173447, "global_step": 29972, "epoch": 282, "lr": 3.977938797303162e-05} +{"train_loss": 0.012713136151432991, "grad_norm": 0.054194483906030655, "global_step": 29973, "epoch": 282, "lr": 3.977648680458112e-05} +{"train_loss": 0.01512038242071867, "grad_norm": 0.1298406571149826, "global_step": 29974, "epoch": 282, "lr": 3.977358567205161e-05} +{"train_loss": 0.016712380573153496, "grad_norm": 0.05976805463433266, "global_step": 29975, "epoch": 282, "lr": 3.977068457545326e-05} +{"train_loss": 0.014410941861569881, "grad_norm": 0.08125896006822586, "global_step": 29976, "epoch": 282, "lr": 3.9767783514796284e-05} +{"train_loss": 0.017997732385993004, "grad_norm": 0.06747180223464966, "global_step": 29977, "epoch": 282, "lr": 3.976488249009085e-05} +{"train_loss": 0.015178915113210678, "grad_norm": 0.16317807137966156, "global_step": 29978, "epoch": 282, "lr": 3.976198150134718e-05} +{"train_loss": 0.014302311465144157, "grad_norm": 0.08526436984539032, "global_step": 29979, "epoch": 282, "lr": 3.975908054857545e-05} +{"train_loss": 0.015948912128806114, "grad_norm": 0.05376473069190979, "global_step": 29980, "epoch": 282, "lr": 3.975617963178587e-05} +{"train_loss": 0.015877440571784973, "grad_norm": 0.17218755185604095, "global_step": 29981, "epoch": 282, "lr": 3.97532787509886e-05} +{"train_loss": 0.014633862301707268, "grad_norm": 0.16392071545124054, "global_step": 29982, "epoch": 282, "lr": 3.9750377906193864e-05} +{"train_loss": 0.01784638874232769, "grad_norm": 0.11701297760009766, "global_step": 29983, "epoch": 282, "lr": 3.974747709741184e-05} +{"train_loss": 0.014876360073685646, "grad_norm": 0.04632133990526199, "global_step": 29984, "epoch": 282, "lr": 3.974457632465272e-05} +{"train_loss": 0.013043390586972237, "grad_norm": 0.12426674365997314, "global_step": 29985, "epoch": 282, "lr": 3.97416755879267e-05} +{"train_loss": 0.014718759804964066, "grad_norm": 0.12992075085639954, "global_step": 29986, "epoch": 282, "lr": 3.973877488724398e-05} +{"train_loss": 0.012049724347889423, "grad_norm": 0.06441562622785568, "global_step": 29987, "epoch": 282, "lr": 3.973587422261474e-05} +{"train_loss": 0.016864223405718803, "grad_norm": 0.11730305105447769, "global_step": 29988, "epoch": 282, "lr": 3.973297359404917e-05} +{"train_loss": 0.015313489362597466, "grad_norm": 0.08058914542198181, "global_step": 29989, "epoch": 282, "lr": 3.973007300155746e-05} +{"train_loss": 0.01803295686841011, "grad_norm": 0.08769150823354721, "global_step": 29990, "epoch": 282, "lr": 3.972717244514981e-05} +{"train_loss": 0.01532529853284359, "grad_norm": 0.10549117624759674, "global_step": 29991, "epoch": 282, "lr": 3.972427192483642e-05} +{"train_loss": 0.014192086644470692, "grad_norm": 0.08335579186677933, "global_step": 29992, "epoch": 282, "lr": 3.9721371440627464e-05} +{"train_loss": 0.014860481023788452, "grad_norm": 0.08911164104938507, "global_step": 29993, "epoch": 282, "lr": 3.971847099253312e-05} +{"train_loss": 0.016250312328338623, "grad_norm": 0.09681147336959839, "global_step": 29994, "epoch": 282, "lr": 3.971557058056362e-05} +{"train_loss": 0.016153518110513687, "grad_norm": 0.13814595341682434, "global_step": 29995, "epoch": 282, "lr": 3.971267020472913e-05} +{"train_loss": 0.014579100534319878, "grad_norm": 0.05851109325885773, "global_step": 29996, "epoch": 282, "lr": 3.970976986503982e-05} +{"train_loss": 0.015415242451401252, "grad_norm": 0.13380898535251617, "global_step": 29997, "epoch": 282, "lr": 3.9706869561505946e-05, "val_loss": 0.020078139379620552} +{"train_loss": 0.015646684914827347, "grad_norm": 0.12908799946308136, "global_step": 29998, "epoch": 283, "lr": 3.9703969294137625e-05} +{"train_loss": 0.01367955282330513, "grad_norm": 0.08148963004350662, "global_step": 29999, "epoch": 283, "lr": 3.970106906294509e-05} +{"train_loss": 0.01662852242588997, "grad_norm": 0.16000883281230927, "global_step": 30000, "epoch": 283, "lr": 3.969816886793851e-05} +{"train_loss": 0.01624414324760437, "grad_norm": 0.07807441800832748, "global_step": 30001, "epoch": 283, "lr": 3.969526870912809e-05} +{"train_loss": 0.013933797366917133, "grad_norm": 0.11716998368501663, "global_step": 30002, "epoch": 283, "lr": 3.9692368586524006e-05} +{"train_loss": 0.018799960613250732, "grad_norm": 0.13229018449783325, "global_step": 30003, "epoch": 283, "lr": 3.968946850013648e-05} +{"train_loss": 0.0150454081594944, "grad_norm": 0.13357649743556976, "global_step": 30004, "epoch": 283, "lr": 3.968656844997566e-05} +{"train_loss": 0.017038876190781593, "grad_norm": 0.23771925270557404, "global_step": 30005, "epoch": 283, "lr": 3.9683668436051755e-05} +{"train_loss": 0.016415193676948547, "grad_norm": 0.08429424464702606, "global_step": 30006, "epoch": 283, "lr": 3.968076845837494e-05} +{"train_loss": 0.015270276926457882, "grad_norm": 0.07068862020969391, "global_step": 30007, "epoch": 283, "lr": 3.967786851695544e-05} +{"train_loss": 0.013898489996790886, "grad_norm": 0.06917911767959595, "global_step": 30008, "epoch": 283, "lr": 3.9674968611803395e-05} +{"train_loss": 0.013493120670318604, "grad_norm": 0.08749283105134964, "global_step": 30009, "epoch": 283, "lr": 3.9672068742929056e-05} +{"train_loss": 0.014577420428395271, "grad_norm": 0.15527762472629547, "global_step": 30010, "epoch": 283, "lr": 3.966916891034255e-05} +{"train_loss": 0.012771216221153736, "grad_norm": 0.153866246342659, "global_step": 30011, "epoch": 283, "lr": 3.96662691140541e-05} +{"train_loss": 0.013987850397825241, "grad_norm": 0.14333230257034302, "global_step": 30012, "epoch": 283, "lr": 3.966336935407388e-05} +{"train_loss": 0.01613270863890648, "grad_norm": 0.0860758051276207, "global_step": 30013, "epoch": 283, "lr": 3.9660469630412077e-05} +{"train_loss": 0.01597176119685173, "grad_norm": 0.12549501657485962, "global_step": 30014, "epoch": 283, "lr": 3.96575699430789e-05} +{"train_loss": 0.01589304395020008, "grad_norm": 0.17424264550209045, "global_step": 30015, "epoch": 283, "lr": 3.965467029208454e-05} +{"train_loss": 0.013446642085909843, "grad_norm": 0.05211734399199486, "global_step": 30016, "epoch": 283, "lr": 3.9651770677439147e-05} +{"train_loss": 0.017148954793810844, "grad_norm": 0.11778000742197037, "global_step": 30017, "epoch": 283, "lr": 3.964887109915294e-05} +{"train_loss": 0.017463987693190575, "grad_norm": 0.0969262570142746, "global_step": 30018, "epoch": 283, "lr": 3.964597155723609e-05} +{"train_loss": 0.014449009671807289, "grad_norm": 0.09332267194986343, "global_step": 30019, "epoch": 283, "lr": 3.96430720516988e-05} +{"train_loss": 0.01590760424733162, "grad_norm": 0.05908504128456116, "global_step": 30020, "epoch": 283, "lr": 3.964017258255125e-05} +{"train_loss": 0.014742567203938961, "grad_norm": 0.08961743861436844, "global_step": 30021, "epoch": 283, "lr": 3.963727314980363e-05} +{"train_loss": 0.014322654344141483, "grad_norm": 0.08682585507631302, "global_step": 30022, "epoch": 283, "lr": 3.963437375346612e-05} +{"train_loss": 0.01553921028971672, "grad_norm": 0.07940391451120377, "global_step": 30023, "epoch": 283, "lr": 3.963147439354892e-05} +{"train_loss": 0.015726909041404724, "grad_norm": 0.12197455018758774, "global_step": 30024, "epoch": 283, "lr": 3.96285750700622e-05} +{"train_loss": 0.014995262026786804, "grad_norm": 0.07087688148021698, "global_step": 30025, "epoch": 283, "lr": 3.9625675783016156e-05} +{"train_loss": 0.015670986846089363, "grad_norm": 0.056259676814079285, "global_step": 30026, "epoch": 283, "lr": 3.962277653242098e-05} +{"train_loss": 0.015030520036816597, "grad_norm": 0.07588516175746918, "global_step": 30027, "epoch": 283, "lr": 3.961987731828687e-05} +{"train_loss": 0.01377813145518303, "grad_norm": 0.06778910756111145, "global_step": 30028, "epoch": 283, "lr": 3.9616978140623987e-05} +{"train_loss": 0.015220487490296364, "grad_norm": 0.10091546922922134, "global_step": 30029, "epoch": 283, "lr": 3.9614078999442524e-05} +{"train_loss": 0.01417003944516182, "grad_norm": 0.08061165362596512, "global_step": 30030, "epoch": 283, "lr": 3.9611179894752656e-05} +{"train_loss": 0.014101969078183174, "grad_norm": 0.06238490715622902, "global_step": 30031, "epoch": 283, "lr": 3.96082808265646e-05} +{"train_loss": 0.018024839460849762, "grad_norm": 0.07700727134943008, "global_step": 30032, "epoch": 283, "lr": 3.960538179488852e-05} +{"train_loss": 0.016284987330436707, "grad_norm": 0.11706088483333588, "global_step": 30033, "epoch": 283, "lr": 3.960248279973463e-05} +{"train_loss": 0.015639979392290115, "grad_norm": 0.12156867235898972, "global_step": 30034, "epoch": 283, "lr": 3.959958384111306e-05} +{"train_loss": 0.016403593122959137, "grad_norm": 0.06871386617422104, "global_step": 30035, "epoch": 283, "lr": 3.959668491903405e-05} +{"train_loss": 0.015748269855976105, "grad_norm": 0.07690861821174622, "global_step": 30036, "epoch": 283, "lr": 3.9593786033507765e-05} +{"train_loss": 0.01562567800283432, "grad_norm": 0.1627187281847, "global_step": 30037, "epoch": 283, "lr": 3.9590887184544374e-05} +{"train_loss": 0.016209334135055542, "grad_norm": 0.11416277289390564, "global_step": 30038, "epoch": 283, "lr": 3.958798837215409e-05} +{"train_loss": 0.01596762239933014, "grad_norm": 0.062258388847112656, "global_step": 30039, "epoch": 283, "lr": 3.9585089596347094e-05} +{"train_loss": 0.01603894680738449, "grad_norm": 0.08353819698095322, "global_step": 30040, "epoch": 283, "lr": 3.958219085713356e-05} +{"train_loss": 0.015286462381482124, "grad_norm": 0.0627666786313057, "global_step": 30041, "epoch": 283, "lr": 3.957929215452367e-05} +{"train_loss": 0.013554541394114494, "grad_norm": 0.08135700970888138, "global_step": 30042, "epoch": 283, "lr": 3.957639348852761e-05} +{"train_loss": 0.016996409744024277, "grad_norm": 0.08868072181940079, "global_step": 30043, "epoch": 283, "lr": 3.957349485915558e-05} +{"train_loss": 0.014879653230309486, "grad_norm": 0.062422607094049454, "global_step": 30044, "epoch": 283, "lr": 3.957059626641777e-05} +{"train_loss": 0.01600370556116104, "grad_norm": 0.07315311580896378, "global_step": 30045, "epoch": 283, "lr": 3.956769771032433e-05} +{"train_loss": 0.01417627464979887, "grad_norm": 0.07801031321287155, "global_step": 30046, "epoch": 283, "lr": 3.956479919088547e-05} +{"train_loss": 0.017006002366542816, "grad_norm": 0.06302697956562042, "global_step": 30047, "epoch": 283, "lr": 3.9561900708111356e-05} +{"train_loss": 0.01588376611471176, "grad_norm": 0.10096662491559982, "global_step": 30048, "epoch": 283, "lr": 3.955900226201219e-05} +{"train_loss": 0.011065024882555008, "grad_norm": 0.14418190717697144, "global_step": 30049, "epoch": 283, "lr": 3.955610385259815e-05} +{"train_loss": 0.014572327956557274, "grad_norm": 0.10195645689964294, "global_step": 30050, "epoch": 283, "lr": 3.955320547987943e-05} +{"train_loss": 0.013978658244013786, "grad_norm": 0.07767025381326675, "global_step": 30051, "epoch": 283, "lr": 3.955030714386618e-05} +{"train_loss": 0.014942670240998268, "grad_norm": 0.06280319392681122, "global_step": 30052, "epoch": 283, "lr": 3.9547408844568626e-05} +{"train_loss": 0.015080436132848263, "grad_norm": 0.07443488389253616, "global_step": 30053, "epoch": 283, "lr": 3.9544510581996915e-05} +{"train_loss": 0.014727814123034477, "grad_norm": 0.0917980819940567, "global_step": 30054, "epoch": 283, "lr": 3.954161235616125e-05} +{"train_loss": 0.01239117980003357, "grad_norm": 0.062245480716228485, "global_step": 30055, "epoch": 283, "lr": 3.953871416707182e-05} +{"train_loss": 0.014630483463406563, "grad_norm": 0.1045956164598465, "global_step": 30056, "epoch": 283, "lr": 3.9535816014738806e-05} +{"train_loss": 0.012973377481102943, "grad_norm": 0.09046360105276108, "global_step": 30057, "epoch": 283, "lr": 3.953291789917238e-05} +{"train_loss": 0.017859112471342087, "grad_norm": 0.06022585928440094, "global_step": 30058, "epoch": 283, "lr": 3.953001982038272e-05} +{"train_loss": 0.015481367707252502, "grad_norm": 0.05992472544312477, "global_step": 30059, "epoch": 283, "lr": 3.952712177838001e-05} +{"train_loss": 0.015453112311661243, "grad_norm": 0.08685551583766937, "global_step": 30060, "epoch": 283, "lr": 3.952422377317445e-05} +{"train_loss": 0.015796735882759094, "grad_norm": 0.13352222740650177, "global_step": 30061, "epoch": 283, "lr": 3.952132580477621e-05} +{"train_loss": 0.013073723763227463, "grad_norm": 0.11860347539186478, "global_step": 30062, "epoch": 283, "lr": 3.951842787319548e-05} +{"train_loss": 0.013371016830205917, "grad_norm": 0.05337043106555939, "global_step": 30063, "epoch": 283, "lr": 3.951552997844243e-05} +{"train_loss": 0.015221786685287952, "grad_norm": 0.07244899123907089, "global_step": 30064, "epoch": 283, "lr": 3.951263212052725e-05} +{"train_loss": 0.013386384584009647, "grad_norm": 0.07257552444934845, "global_step": 30065, "epoch": 283, "lr": 3.9509734299460115e-05} +{"train_loss": 0.014140836894512177, "grad_norm": 0.11163096129894257, "global_step": 30066, "epoch": 283, "lr": 3.9506836515251203e-05} +{"train_loss": 0.01465700101107359, "grad_norm": 0.07915212959051132, "global_step": 30067, "epoch": 283, "lr": 3.950393876791072e-05} +{"train_loss": 0.01617170125246048, "grad_norm": 0.10089713335037231, "global_step": 30068, "epoch": 283, "lr": 3.950104105744884e-05} +{"train_loss": 0.015366645529866219, "grad_norm": 0.09820713847875595, "global_step": 30069, "epoch": 283, "lr": 3.949814338387572e-05} +{"train_loss": 0.015166176483035088, "grad_norm": 0.06257019937038422, "global_step": 30070, "epoch": 283, "lr": 3.949524574720157e-05} +{"train_loss": 0.012785010039806366, "grad_norm": 0.07547071576118469, "global_step": 30071, "epoch": 283, "lr": 3.949234814743654e-05} +{"train_loss": 0.016767676919698715, "grad_norm": 0.0736137256026268, "global_step": 30072, "epoch": 283, "lr": 3.948945058459084e-05} +{"train_loss": 0.015608766116201878, "grad_norm": 0.09424219280481339, "global_step": 30073, "epoch": 283, "lr": 3.948655305867464e-05} +{"train_loss": 0.01673204079270363, "grad_norm": 0.10148404538631439, "global_step": 30074, "epoch": 283, "lr": 3.9483655569698124e-05} +{"train_loss": 0.01769854500889778, "grad_norm": 0.0970749706029892, "global_step": 30075, "epoch": 283, "lr": 3.9480758117671465e-05} +{"train_loss": 0.014861940406262875, "grad_norm": 0.07393843680620193, "global_step": 30076, "epoch": 283, "lr": 3.947786070260484e-05} +{"train_loss": 0.013672692701220512, "grad_norm": 0.09841851145029068, "global_step": 30077, "epoch": 283, "lr": 3.9474963324508435e-05} +{"train_loss": 0.015606777742505074, "grad_norm": 0.08648623526096344, "global_step": 30078, "epoch": 283, "lr": 3.947206598339244e-05} +{"train_loss": 0.016174865886569023, "grad_norm": 0.10599920153617859, "global_step": 30079, "epoch": 283, "lr": 3.946916867926702e-05} +{"train_loss": 0.015080615878105164, "grad_norm": 0.07309933006763458, "global_step": 30080, "epoch": 283, "lr": 3.946627141214237e-05} +{"train_loss": 0.01570838689804077, "grad_norm": 0.06064635515213013, "global_step": 30081, "epoch": 283, "lr": 3.946337418202866e-05} +{"train_loss": 0.018232561647892, "grad_norm": 0.08596591651439667, "global_step": 30082, "epoch": 283, "lr": 3.946047698893607e-05} +{"train_loss": 0.015327135100960732, "grad_norm": 0.12796202301979065, "global_step": 30083, "epoch": 283, "lr": 3.945757983287477e-05} +{"train_loss": 0.018455887213349342, "grad_norm": 0.08888740092515945, "global_step": 30084, "epoch": 283, "lr": 3.9454682713854964e-05} +{"train_loss": 0.014981931075453758, "grad_norm": 0.06896381080150604, "global_step": 30085, "epoch": 283, "lr": 3.945178563188681e-05} +{"train_loss": 0.014942619018256664, "grad_norm": 0.12240505963563919, "global_step": 30086, "epoch": 283, "lr": 3.9448888586980505e-05} +{"train_loss": 0.016513757407665253, "grad_norm": 0.05942267179489136, "global_step": 30087, "epoch": 283, "lr": 3.944599157914621e-05} +{"train_loss": 0.017925512045621872, "grad_norm": 0.07288889586925507, "global_step": 30088, "epoch": 283, "lr": 3.94430946083941e-05} +{"train_loss": 0.015040402300655842, "grad_norm": 0.0797029659152031, "global_step": 30089, "epoch": 283, "lr": 3.944019767473437e-05} +{"train_loss": 0.01657566800713539, "grad_norm": 0.06833896040916443, "global_step": 30090, "epoch": 283, "lr": 3.94373007781772e-05} +{"train_loss": 0.015593800693750381, "grad_norm": 0.06238815188407898, "global_step": 30091, "epoch": 283, "lr": 3.9434403918732766e-05} +{"train_loss": 0.013823745772242546, "grad_norm": 0.09373702853918076, "global_step": 30092, "epoch": 283, "lr": 3.9431507096411224e-05} +{"train_loss": 0.015015942975878716, "grad_norm": 0.08301049470901489, "global_step": 30093, "epoch": 283, "lr": 3.942861031122278e-05} +{"train_loss": 0.013115105219185352, "grad_norm": 0.07399795949459076, "global_step": 30094, "epoch": 283, "lr": 3.9425713563177604e-05} +{"train_loss": 0.015563730150461197, "grad_norm": 0.1333066076040268, "global_step": 30095, "epoch": 283, "lr": 3.942281685228586e-05} +{"train_loss": 0.01820404641330242, "grad_norm": 0.0691198855638504, "global_step": 30096, "epoch": 283, "lr": 3.9419920178557744e-05} +{"train_loss": 0.014099071733653545, "grad_norm": 0.10088140517473221, "global_step": 30097, "epoch": 283, "lr": 3.941702354200344e-05} +{"train_loss": 0.018357660621404648, "grad_norm": 0.13083875179290771, "global_step": 30098, "epoch": 283, "lr": 3.941412694263311e-05} +{"train_loss": 0.016438711434602737, "grad_norm": 0.11048922687768936, "global_step": 30099, "epoch": 283, "lr": 3.9411230380456925e-05} +{"train_loss": 0.016737161204218864, "grad_norm": 0.06487569957971573, "global_step": 30100, "epoch": 283, "lr": 3.9408333855485066e-05} +{"train_loss": 0.017019886523485184, "grad_norm": 0.08603296428918839, "global_step": 30101, "epoch": 283, "lr": 3.940543736772772e-05} +{"train_loss": 0.015987208113074303, "grad_norm": 0.10114595293998718, "global_step": 30102, "epoch": 283, "lr": 3.940254091719506e-05} +{"train_loss": 0.015378638267306224, "grad_norm": 0.14461596310138702, "global_step": 30103, "epoch": 283, "lr": 3.939964450389728e-05, "val_loss": 0.019857514649629593} +{"train_loss": 0.014407874085009098, "grad_norm": 0.06714391708374023, "global_step": 30104, "epoch": 284, "lr": 3.939674812784453e-05} +{"train_loss": 0.014743495732545853, "grad_norm": 0.062134820967912674, "global_step": 30105, "epoch": 284, "lr": 3.939385178904698e-05} +{"train_loss": 0.016238778829574585, "grad_norm": 0.17000098526477814, "global_step": 30106, "epoch": 284, "lr": 3.939095548751483e-05} +{"train_loss": 0.015459113754332066, "grad_norm": 0.07182831317186356, "global_step": 30107, "epoch": 284, "lr": 3.938805922325826e-05} +{"train_loss": 0.016439838334918022, "grad_norm": 0.07240131497383118, "global_step": 30108, "epoch": 284, "lr": 3.938516299628741e-05} +{"train_loss": 0.01605270802974701, "grad_norm": 0.15306736528873444, "global_step": 30109, "epoch": 284, "lr": 3.938226680661251e-05} +{"train_loss": 0.01766297221183777, "grad_norm": 0.05859505385160446, "global_step": 30110, "epoch": 284, "lr": 3.93793706542437e-05} +{"train_loss": 0.013605325482785702, "grad_norm": 0.08768723905086517, "global_step": 30111, "epoch": 284, "lr": 3.9376474539191156e-05} +{"train_loss": 0.014438124373555183, "grad_norm": 0.09891083091497421, "global_step": 30112, "epoch": 284, "lr": 3.937357846146505e-05} +{"train_loss": 0.01876937597990036, "grad_norm": 0.09468518197536469, "global_step": 30113, "epoch": 284, "lr": 3.937068242107558e-05} +{"train_loss": 0.014511536806821823, "grad_norm": 0.0792255774140358, "global_step": 30114, "epoch": 284, "lr": 3.936778641803291e-05} +{"train_loss": 0.014323579147458076, "grad_norm": 0.07114468514919281, "global_step": 30115, "epoch": 284, "lr": 3.9364890452347225e-05} +{"train_loss": 0.016516316682100296, "grad_norm": 0.06934025883674622, "global_step": 30116, "epoch": 284, "lr": 3.936199452402868e-05} +{"train_loss": 0.016499415040016174, "grad_norm": 0.06992267072200775, "global_step": 30117, "epoch": 284, "lr": 3.935909863308744e-05} +{"train_loss": 0.01424202136695385, "grad_norm": 0.07724721729755402, "global_step": 30118, "epoch": 284, "lr": 3.9356202779533725e-05} +{"train_loss": 0.01669350638985634, "grad_norm": 0.05853491649031639, "global_step": 30119, "epoch": 284, "lr": 3.935330696337768e-05} +{"train_loss": 0.014983205124735832, "grad_norm": 0.0780056044459343, "global_step": 30120, "epoch": 284, "lr": 3.935041118462947e-05} +{"train_loss": 0.019802764058113098, "grad_norm": 0.08554745465517044, "global_step": 30121, "epoch": 284, "lr": 3.9347515443299304e-05} +{"train_loss": 0.01747310534119606, "grad_norm": 0.06829908490180969, "global_step": 30122, "epoch": 284, "lr": 3.9344619739397335e-05} +{"train_loss": 0.013537333346903324, "grad_norm": 0.08255276083946228, "global_step": 30123, "epoch": 284, "lr": 3.934172407293373e-05} +{"train_loss": 0.015114227309823036, "grad_norm": 0.08859244734048843, "global_step": 30124, "epoch": 284, "lr": 3.9338828443918665e-05} +{"train_loss": 0.01429800782352686, "grad_norm": 0.13333848118782043, "global_step": 30125, "epoch": 284, "lr": 3.933593285236232e-05} +{"train_loss": 0.014888640493154526, "grad_norm": 0.12216895073652267, "global_step": 30126, "epoch": 284, "lr": 3.9333037298274886e-05} +{"train_loss": 0.01395466085523367, "grad_norm": 0.09068391472101212, "global_step": 30127, "epoch": 284, "lr": 3.933014178166652e-05} +{"train_loss": 0.015114806592464447, "grad_norm": 0.06359349936246872, "global_step": 30128, "epoch": 284, "lr": 3.932724630254738e-05} +{"train_loss": 0.016156306490302086, "grad_norm": 0.08774525672197342, "global_step": 30129, "epoch": 284, "lr": 3.932435086092766e-05} +{"train_loss": 0.014059791341423988, "grad_norm": 0.08893056958913803, "global_step": 30130, "epoch": 284, "lr": 3.932145545681753e-05} +{"train_loss": 0.014512212947010994, "grad_norm": 0.06911090016365051, "global_step": 30131, "epoch": 284, "lr": 3.9318560090227166e-05} +{"train_loss": 0.015913359820842743, "grad_norm": 0.06899970024824142, "global_step": 30132, "epoch": 284, "lr": 3.931566476116671e-05} +{"train_loss": 0.014841911382973194, "grad_norm": 0.07086402177810669, "global_step": 30133, "epoch": 284, "lr": 3.931276946964641e-05} +{"train_loss": 0.01548114512115717, "grad_norm": 0.10395383089780807, "global_step": 30134, "epoch": 284, "lr": 3.930987421567636e-05} +{"train_loss": 0.014485441148281097, "grad_norm": 0.05973048508167267, "global_step": 30135, "epoch": 284, "lr": 3.9306978999266765e-05} +{"train_loss": 0.014484090730547905, "grad_norm": 0.08131840825080872, "global_step": 30136, "epoch": 284, "lr": 3.9304083820427794e-05} +{"train_loss": 0.014435797929763794, "grad_norm": 0.05270450562238693, "global_step": 30137, "epoch": 284, "lr": 3.930118867916962e-05} +{"train_loss": 0.013082893565297127, "grad_norm": 0.08619125187397003, "global_step": 30138, "epoch": 284, "lr": 3.929829357550242e-05} +{"train_loss": 0.013687019236385822, "grad_norm": 0.08458583056926727, "global_step": 30139, "epoch": 284, "lr": 3.9295398509436376e-05} +{"train_loss": 0.015179604291915894, "grad_norm": 0.06304536759853363, "global_step": 30140, "epoch": 284, "lr": 3.9292503480981633e-05} +{"train_loss": 0.015633588656783104, "grad_norm": 0.07756993174552917, "global_step": 30141, "epoch": 284, "lr": 3.928960849014837e-05} +{"train_loss": 0.015769384801387787, "grad_norm": 0.174698606133461, "global_step": 30142, "epoch": 284, "lr": 3.928671353694677e-05} +{"train_loss": 0.015311181545257568, "grad_norm": 0.11117216944694519, "global_step": 30143, "epoch": 284, "lr": 3.9283818621387e-05} +{"train_loss": 0.013228904455900192, "grad_norm": 0.1022569015622139, "global_step": 30144, "epoch": 284, "lr": 3.928092374347925e-05} +{"train_loss": 0.01472887210547924, "grad_norm": 0.07315260171890259, "global_step": 30145, "epoch": 284, "lr": 3.9278028903233647e-05} +{"train_loss": 0.014205686748027802, "grad_norm": 0.06366334110498428, "global_step": 30146, "epoch": 284, "lr": 3.9275134100660385e-05} +{"train_loss": 0.016095085069537163, "grad_norm": 0.08972060680389404, "global_step": 30147, "epoch": 284, "lr": 3.9272239335769654e-05} +{"train_loss": 0.0191645547747612, "grad_norm": 0.09160061180591583, "global_step": 30148, "epoch": 284, "lr": 3.9269344608571604e-05} +{"train_loss": 0.012530840933322906, "grad_norm": 0.06584420055150986, "global_step": 30149, "epoch": 284, "lr": 3.92664499190764e-05} +{"train_loss": 0.01701105572283268, "grad_norm": 0.0857376977801323, "global_step": 30150, "epoch": 284, "lr": 3.926355526729424e-05} +{"train_loss": 0.016529656946659088, "grad_norm": 0.10987404733896255, "global_step": 30151, "epoch": 284, "lr": 3.926066065323527e-05} +{"train_loss": 0.01526218093931675, "grad_norm": 0.1436326503753662, "global_step": 30152, "epoch": 284, "lr": 3.9257766076909666e-05} +{"train_loss": 0.016262086108326912, "grad_norm": 0.0690106749534607, "global_step": 30153, "epoch": 284, "lr": 3.925487153832759e-05} +{"train_loss": 0.016450226306915283, "grad_norm": 0.13124753534793854, "global_step": 30154, "epoch": 284, "lr": 3.925197703749924e-05} +{"train_loss": 0.01464143767952919, "grad_norm": 0.13163669407367706, "global_step": 30155, "epoch": 284, "lr": 3.924908257443476e-05} +{"train_loss": 0.014964423142373562, "grad_norm": 0.08606700599193573, "global_step": 30156, "epoch": 284, "lr": 3.9246188149144346e-05} +{"train_loss": 0.01634415239095688, "grad_norm": 0.09022621065378189, "global_step": 30157, "epoch": 284, "lr": 3.924329376163813e-05} +{"train_loss": 0.014539789408445358, "grad_norm": 0.07542528957128525, "global_step": 30158, "epoch": 284, "lr": 3.92403994119263e-05} +{"train_loss": 0.015840623527765274, "grad_norm": 0.15564240515232086, "global_step": 30159, "epoch": 284, "lr": 3.923750510001903e-05} +{"train_loss": 0.015333116054534912, "grad_norm": 0.07371345907449722, "global_step": 30160, "epoch": 284, "lr": 3.92346108259265e-05} +{"train_loss": 0.015937337651848793, "grad_norm": 0.058167386800050735, "global_step": 30161, "epoch": 284, "lr": 3.9231716589658845e-05} +{"train_loss": 0.013741480186581612, "grad_norm": 0.07693076133728027, "global_step": 30162, "epoch": 284, "lr": 3.922882239122629e-05} +{"train_loss": 0.015481565147638321, "grad_norm": 0.10224001854658127, "global_step": 30163, "epoch": 284, "lr": 3.922592823063893e-05} +{"train_loss": 0.012891815975308418, "grad_norm": 0.07934239506721497, "global_step": 30164, "epoch": 284, "lr": 3.922303410790699e-05} +{"train_loss": 0.015079963020980358, "grad_norm": 0.07672494649887085, "global_step": 30165, "epoch": 284, "lr": 3.9220140023040614e-05} +{"train_loss": 0.014686365611851215, "grad_norm": 0.14421620965003967, "global_step": 30166, "epoch": 284, "lr": 3.921724597604998e-05} +{"train_loss": 0.015724780037999153, "grad_norm": 0.09068632870912552, "global_step": 30167, "epoch": 284, "lr": 3.921435196694526e-05} +{"train_loss": 0.01668514311313629, "grad_norm": 0.07353593409061432, "global_step": 30168, "epoch": 284, "lr": 3.9211457995736626e-05} +{"train_loss": 0.014085281640291214, "grad_norm": 0.07575300335884094, "global_step": 30169, "epoch": 284, "lr": 3.920856406243423e-05} +{"train_loss": 0.01597435027360916, "grad_norm": 0.0717843547463417, "global_step": 30170, "epoch": 284, "lr": 3.9205670167048234e-05} +{"train_loss": 0.01581377163529396, "grad_norm": 0.08761665225028992, "global_step": 30171, "epoch": 284, "lr": 3.9202776309588826e-05} +{"train_loss": 0.015848271548748016, "grad_norm": 0.11909737437963486, "global_step": 30172, "epoch": 284, "lr": 3.9199882490066166e-05} +{"train_loss": 0.01637890562415123, "grad_norm": 0.0620027594268322, "global_step": 30173, "epoch": 284, "lr": 3.9196988708490414e-05} +{"train_loss": 0.01443033292889595, "grad_norm": 0.08104278892278671, "global_step": 30174, "epoch": 284, "lr": 3.919409496487177e-05} +{"train_loss": 0.013430400751531124, "grad_norm": 0.06188960000872612, "global_step": 30175, "epoch": 284, "lr": 3.9191201259220354e-05} +{"train_loss": 0.015524568036198616, "grad_norm": 0.06645114719867706, "global_step": 30176, "epoch": 284, "lr": 3.918830759154637e-05} +{"train_loss": 0.014579134993255138, "grad_norm": 0.11596021801233292, "global_step": 30177, "epoch": 284, "lr": 3.918541396185996e-05} +{"train_loss": 0.016961492598056793, "grad_norm": 0.07600999623537064, "global_step": 30178, "epoch": 284, "lr": 3.91825203701713e-05} +{"train_loss": 0.01613813266158104, "grad_norm": 0.12568701803684235, "global_step": 30179, "epoch": 284, "lr": 3.917962681649056e-05} +{"train_loss": 0.01566382497549057, "grad_norm": 0.0660502016544342, "global_step": 30180, "epoch": 284, "lr": 3.917673330082793e-05} +{"train_loss": 0.01507152896374464, "grad_norm": 0.06719421595335007, "global_step": 30181, "epoch": 284, "lr": 3.917383982319354e-05} +{"train_loss": 0.015383351594209671, "grad_norm": 0.09365534782409668, "global_step": 30182, "epoch": 284, "lr": 3.917094638359755e-05} +{"train_loss": 0.014986719004809856, "grad_norm": 0.07479545474052429, "global_step": 30183, "epoch": 284, "lr": 3.916805298205016e-05} +{"train_loss": 0.016177933663129807, "grad_norm": 0.07861143350601196, "global_step": 30184, "epoch": 284, "lr": 3.916515961856152e-05} +{"train_loss": 0.015032189898192883, "grad_norm": 0.055615466088056564, "global_step": 30185, "epoch": 284, "lr": 3.916226629314179e-05} +{"train_loss": 0.015207439661026001, "grad_norm": 0.07551437616348267, "global_step": 30186, "epoch": 284, "lr": 3.915937300580116e-05} +{"train_loss": 0.016059521585702896, "grad_norm": 0.06772112846374512, "global_step": 30187, "epoch": 284, "lr": 3.9156479756549763e-05} +{"train_loss": 0.01703484170138836, "grad_norm": 0.0809733122587204, "global_step": 30188, "epoch": 284, "lr": 3.9153586545397785e-05} +{"train_loss": 0.01209673285484314, "grad_norm": 0.06731706857681274, "global_step": 30189, "epoch": 284, "lr": 3.915069337235539e-05} +{"train_loss": 0.014417644590139389, "grad_norm": 0.06266781687736511, "global_step": 30190, "epoch": 284, "lr": 3.914780023743273e-05} +{"train_loss": 0.016727160662412643, "grad_norm": 0.098085917532444, "global_step": 30191, "epoch": 284, "lr": 3.9144907140640007e-05} +{"train_loss": 0.01591983437538147, "grad_norm": 0.09580071270465851, "global_step": 30192, "epoch": 284, "lr": 3.914201408198733e-05} +{"train_loss": 0.016186824068427086, "grad_norm": 0.08379720896482468, "global_step": 30193, "epoch": 284, "lr": 3.9139121061484905e-05} +{"train_loss": 0.013643385842442513, "grad_norm": 0.064163938164711, "global_step": 30194, "epoch": 284, "lr": 3.913622807914288e-05} +{"train_loss": 0.016996093094348907, "grad_norm": 0.15308824181556702, "global_step": 30195, "epoch": 284, "lr": 3.913333513497143e-05} +{"train_loss": 0.012935595586895943, "grad_norm": 0.053033675998449326, "global_step": 30196, "epoch": 284, "lr": 3.9130442228980716e-05} +{"train_loss": 0.014652641490101814, "grad_norm": 0.09634672850370407, "global_step": 30197, "epoch": 284, "lr": 3.91275493611809e-05} +{"train_loss": 0.016403421759605408, "grad_norm": 0.09275663644075394, "global_step": 30198, "epoch": 284, "lr": 3.912465653158215e-05} +{"train_loss": 0.01503694336861372, "grad_norm": 0.0591893196105957, "global_step": 30199, "epoch": 284, "lr": 3.9121763740194616e-05} +{"train_loss": 0.01573500782251358, "grad_norm": 0.07723264396190643, "global_step": 30200, "epoch": 284, "lr": 3.9118870987028486e-05} +{"train_loss": 0.014406029134988785, "grad_norm": 0.05086302012205124, "global_step": 30201, "epoch": 284, "lr": 3.911597827209391e-05} +{"train_loss": 0.016808051615953445, "grad_norm": 0.12536649405956268, "global_step": 30202, "epoch": 284, "lr": 3.9113085595401037e-05} +{"train_loss": 0.013330329209566116, "grad_norm": 0.13608784973621368, "global_step": 30203, "epoch": 284, "lr": 3.911019295696008e-05} +{"train_loss": 0.016054291278123856, "grad_norm": 0.09669376909732819, "global_step": 30204, "epoch": 284, "lr": 3.910730035678114e-05} +{"train_loss": 0.017087072134017944, "grad_norm": 0.060767218470573425, "global_step": 30205, "epoch": 284, "lr": 3.910440779487442e-05} +{"train_loss": 0.018292326480150223, "grad_norm": 0.08109444379806519, "global_step": 30206, "epoch": 284, "lr": 3.910151527125007e-05} +{"train_loss": 0.01707669161260128, "grad_norm": 0.0665464773774147, "global_step": 30207, "epoch": 284, "lr": 3.9098622785918245e-05} +{"train_loss": 0.014994313940405846, "grad_norm": 0.06427446752786636, "global_step": 30208, "epoch": 284, "lr": 3.909573033888912e-05} +{"train_loss": 0.015353879592609856, "grad_norm": 0.1032523661851883, "global_step": 30209, "epoch": 284, "lr": 3.9092837930172884e-05, "val_loss": 0.01891227811574936} +{"train_loss": 0.013611562550067902, "grad_norm": 0.07934018969535828, "global_step": 30210, "epoch": 285, "lr": 3.9089945559779656e-05} +{"train_loss": 0.0146854342892766, "grad_norm": 0.12148784846067429, "global_step": 30211, "epoch": 285, "lr": 3.90870532277196e-05} +{"train_loss": 0.015358728356659412, "grad_norm": 0.06550950556993484, "global_step": 30212, "epoch": 285, "lr": 3.9084160934002914e-05} +{"train_loss": 0.015088984742760658, "grad_norm": 0.06653185188770294, "global_step": 30213, "epoch": 285, "lr": 3.908126867863973e-05} +{"train_loss": 0.014786826446652412, "grad_norm": 0.11521249264478683, "global_step": 30214, "epoch": 285, "lr": 3.907837646164021e-05} +{"train_loss": 0.014472909271717072, "grad_norm": 0.09840833395719528, "global_step": 30215, "epoch": 285, "lr": 3.907548428301455e-05} +{"train_loss": 0.013114100322127342, "grad_norm": 0.06398005038499832, "global_step": 30216, "epoch": 285, "lr": 3.9072592142772866e-05} +{"train_loss": 0.01593923754990101, "grad_norm": 0.09550777822732925, "global_step": 30217, "epoch": 285, "lr": 3.906970004092535e-05} +{"train_loss": 0.015197530388832092, "grad_norm": 0.06439173966646194, "global_step": 30218, "epoch": 285, "lr": 3.9066807977482156e-05} +{"train_loss": 0.01614542491734028, "grad_norm": 0.07495865225791931, "global_step": 30219, "epoch": 285, "lr": 3.9063915952453426e-05} +{"train_loss": 0.01264965906739235, "grad_norm": 0.06184166297316551, "global_step": 30220, "epoch": 285, "lr": 3.9061023965849356e-05} +{"train_loss": 0.016086162999272346, "grad_norm": 0.16261078417301178, "global_step": 30221, "epoch": 285, "lr": 3.9058132017680104e-05} +{"train_loss": 0.015199583023786545, "grad_norm": 0.06816131621599197, "global_step": 30222, "epoch": 285, "lr": 3.905524010795581e-05} +{"train_loss": 0.013237569481134415, "grad_norm": 0.0699605867266655, "global_step": 30223, "epoch": 285, "lr": 3.905234823668662e-05} +{"train_loss": 0.014400852844119072, "grad_norm": 0.07044940441846848, "global_step": 30224, "epoch": 285, "lr": 3.904945640388273e-05} +{"train_loss": 0.0178545992821455, "grad_norm": 0.12332470715045929, "global_step": 30225, "epoch": 285, "lr": 3.904656460955429e-05} +{"train_loss": 0.016253061592578888, "grad_norm": 0.08773540705442429, "global_step": 30226, "epoch": 285, "lr": 3.9043672853711447e-05} +{"train_loss": 0.015157054178416729, "grad_norm": 0.1375553011894226, "global_step": 30227, "epoch": 285, "lr": 3.9040781136364405e-05} +{"train_loss": 0.015271024778485298, "grad_norm": 0.0638374462723732, "global_step": 30228, "epoch": 285, "lr": 3.9037889457523265e-05} +{"train_loss": 0.017304467037320137, "grad_norm": 0.07850633561611176, "global_step": 30229, "epoch": 285, "lr": 3.9034997817198224e-05} +{"train_loss": 0.01517850998789072, "grad_norm": 0.08652666211128235, "global_step": 30230, "epoch": 285, "lr": 3.9032106215399426e-05} +{"train_loss": 0.015937577933073044, "grad_norm": 0.0778869017958641, "global_step": 30231, "epoch": 285, "lr": 3.902921465213703e-05} +{"train_loss": 0.013892849907279015, "grad_norm": 0.06259387731552124, "global_step": 30232, "epoch": 285, "lr": 3.902632312742122e-05} +{"train_loss": 0.0144124710932374, "grad_norm": 0.0923343300819397, "global_step": 30233, "epoch": 285, "lr": 3.902343164126214e-05} +{"train_loss": 0.012833540327847004, "grad_norm": 0.0685824528336525, "global_step": 30234, "epoch": 285, "lr": 3.9020540193669936e-05} +{"train_loss": 0.016557900235056877, "grad_norm": 0.07639201730489731, "global_step": 30235, "epoch": 285, "lr": 3.901764878465478e-05} +{"train_loss": 0.01176836621016264, "grad_norm": 0.14720866084098816, "global_step": 30236, "epoch": 285, "lr": 3.901475741422682e-05} +{"train_loss": 0.013061371631920338, "grad_norm": 0.060294944792985916, "global_step": 30237, "epoch": 285, "lr": 3.901186608239624e-05} +{"train_loss": 0.014955462887883186, "grad_norm": 0.05965026095509529, "global_step": 30238, "epoch": 285, "lr": 3.900897478917318e-05} +{"train_loss": 0.014666338451206684, "grad_norm": 0.07659003883600235, "global_step": 30239, "epoch": 285, "lr": 3.900608353456782e-05} +{"train_loss": 0.01622377708554268, "grad_norm": 0.12451522052288055, "global_step": 30240, "epoch": 285, "lr": 3.9003192318590265e-05} +{"train_loss": 0.017841458320617676, "grad_norm": 0.07196450233459473, "global_step": 30241, "epoch": 285, "lr": 3.9000301141250736e-05} +{"train_loss": 0.018010597676038742, "grad_norm": 0.1608969271183014, "global_step": 30242, "epoch": 285, "lr": 3.8997410002559355e-05} +{"train_loss": 0.015736309811472893, "grad_norm": 0.08244980871677399, "global_step": 30243, "epoch": 285, "lr": 3.899451890252628e-05} +{"train_loss": 0.01390928030014038, "grad_norm": 0.05678693205118179, "global_step": 30244, "epoch": 285, "lr": 3.899162784116172e-05} +{"train_loss": 0.014842214062809944, "grad_norm": 0.09261275827884674, "global_step": 30245, "epoch": 285, "lr": 3.898873681847576e-05} +{"train_loss": 0.01626826450228691, "grad_norm": 0.11767978221178055, "global_step": 30246, "epoch": 285, "lr": 3.898584583447859e-05} +{"train_loss": 0.01682927832007408, "grad_norm": 0.14156600832939148, "global_step": 30247, "epoch": 285, "lr": 3.8982954889180375e-05} +{"train_loss": 0.014293023385107517, "grad_norm": 0.062161218374967575, "global_step": 30248, "epoch": 285, "lr": 3.8980063982591255e-05} +{"train_loss": 0.015412856824696064, "grad_norm": 0.06673628836870193, "global_step": 30249, "epoch": 285, "lr": 3.897717311472141e-05} +{"train_loss": 0.01435659546405077, "grad_norm": 0.09610909223556519, "global_step": 30250, "epoch": 285, "lr": 3.897428228558099e-05} +{"train_loss": 0.016391700133681297, "grad_norm": 0.06960783153772354, "global_step": 30251, "epoch": 285, "lr": 3.8971391495180136e-05} +{"train_loss": 0.016383077949285507, "grad_norm": 0.08717058598995209, "global_step": 30252, "epoch": 285, "lr": 3.8968500743529016e-05} +{"train_loss": 0.014202212914824486, "grad_norm": 0.08832357823848724, "global_step": 30253, "epoch": 285, "lr": 3.8965610030637784e-05} +{"train_loss": 0.013848286122083664, "grad_norm": 0.08007144927978516, "global_step": 30254, "epoch": 285, "lr": 3.8962719356516605e-05} +{"train_loss": 0.015598829835653305, "grad_norm": 0.1056634858250618, "global_step": 30255, "epoch": 285, "lr": 3.895982872117562e-05} +{"train_loss": 0.015980597585439682, "grad_norm": 0.08421940356492996, "global_step": 30256, "epoch": 285, "lr": 3.895693812462503e-05} +{"train_loss": 0.015316237695515156, "grad_norm": 0.13234476745128632, "global_step": 30257, "epoch": 285, "lr": 3.895404756687492e-05} +{"train_loss": 0.014437592588365078, "grad_norm": 0.07808245718479156, "global_step": 30258, "epoch": 285, "lr": 3.89511570479355e-05} +{"train_loss": 0.014229790307581425, "grad_norm": 0.08128505200147629, "global_step": 30259, "epoch": 285, "lr": 3.8948266567816906e-05} +{"train_loss": 0.016235582530498505, "grad_norm": 0.0918973833322525, "global_step": 30260, "epoch": 285, "lr": 3.894537612652929e-05} +{"train_loss": 0.013379018753767014, "grad_norm": 0.06355854868888855, "global_step": 30261, "epoch": 285, "lr": 3.8942485724082816e-05} +{"train_loss": 0.016151707619428635, "grad_norm": 0.07298479229211807, "global_step": 30262, "epoch": 285, "lr": 3.8939595360487656e-05} +{"train_loss": 0.018009508028626442, "grad_norm": 0.10144422948360443, "global_step": 30263, "epoch": 285, "lr": 3.8936705035753935e-05} +{"train_loss": 0.012801723554730415, "grad_norm": 0.056721556931734085, "global_step": 30264, "epoch": 285, "lr": 3.893381474989182e-05} +{"train_loss": 0.013240628875792027, "grad_norm": 0.05822610855102539, "global_step": 30265, "epoch": 285, "lr": 3.8930924502911456e-05} +{"train_loss": 0.014578457921743393, "grad_norm": 0.06851553171873093, "global_step": 30266, "epoch": 285, "lr": 3.892803429482302e-05} +{"train_loss": 0.015573035925626755, "grad_norm": 0.06528772413730621, "global_step": 30267, "epoch": 285, "lr": 3.8925144125636655e-05} +{"train_loss": 0.016716908663511276, "grad_norm": 0.07250525802373886, "global_step": 30268, "epoch": 285, "lr": 3.892225399536255e-05} +{"train_loss": 0.018638096749782562, "grad_norm": 0.07553543895483017, "global_step": 30269, "epoch": 285, "lr": 3.891936390401078e-05} +{"train_loss": 0.016020676121115685, "grad_norm": 0.1575687974691391, "global_step": 30270, "epoch": 285, "lr": 3.891647385159156e-05} +{"train_loss": 0.016508519649505615, "grad_norm": 0.14511539041996002, "global_step": 30271, "epoch": 285, "lr": 3.8913583838115045e-05} +{"train_loss": 0.013400539755821228, "grad_norm": 0.1785338968038559, "global_step": 30272, "epoch": 285, "lr": 3.891069386359135e-05} +{"train_loss": 0.015310580842196941, "grad_norm": 0.11069723218679428, "global_step": 30273, "epoch": 285, "lr": 3.890780392803067e-05} +{"train_loss": 0.018334921449422836, "grad_norm": 0.07902726531028748, "global_step": 30274, "epoch": 285, "lr": 3.890491403144315e-05} +{"train_loss": 0.014145612716674805, "grad_norm": 0.06102421134710312, "global_step": 30275, "epoch": 285, "lr": 3.890202417383894e-05} +{"train_loss": 0.01539740152657032, "grad_norm": 0.052569109946489334, "global_step": 30276, "epoch": 285, "lr": 3.889913435522817e-05} +{"train_loss": 0.016189686954021454, "grad_norm": 0.10896012932062149, "global_step": 30277, "epoch": 285, "lr": 3.8896244575621015e-05} +{"train_loss": 0.014143354259431362, "grad_norm": 0.08026502281427383, "global_step": 30278, "epoch": 285, "lr": 3.889335483502764e-05} +{"train_loss": 0.018123101443052292, "grad_norm": 0.06892024725675583, "global_step": 30279, "epoch": 285, "lr": 3.889046513345818e-05} +{"train_loss": 0.01522007305175066, "grad_norm": 0.08675073832273483, "global_step": 30280, "epoch": 285, "lr": 3.888757547092281e-05} +{"train_loss": 0.017709989100694656, "grad_norm": 0.06474678963422775, "global_step": 30281, "epoch": 285, "lr": 3.8884685847431645e-05} +{"train_loss": 0.014890210703015327, "grad_norm": 0.1066390872001648, "global_step": 30282, "epoch": 285, "lr": 3.888179626299487e-05} +{"train_loss": 0.017488688230514526, "grad_norm": 0.07528161257505417, "global_step": 30283, "epoch": 285, "lr": 3.8878906717622625e-05} +{"train_loss": 0.01670471951365471, "grad_norm": 0.10272973030805588, "global_step": 30284, "epoch": 285, "lr": 3.887601721132506e-05} +{"train_loss": 0.016927305608987808, "grad_norm": 0.13715702295303345, "global_step": 30285, "epoch": 285, "lr": 3.887312774411234e-05} +{"train_loss": 0.016555704176425934, "grad_norm": 0.058353450149297714, "global_step": 30286, "epoch": 285, "lr": 3.887023831599462e-05} +{"train_loss": 0.015084016136825085, "grad_norm": 0.12576663494110107, "global_step": 30287, "epoch": 285, "lr": 3.8867348926982033e-05} +{"train_loss": 0.01391751691699028, "grad_norm": 0.07812537252902985, "global_step": 30288, "epoch": 285, "lr": 3.886445957708474e-05} +{"train_loss": 0.01375608704984188, "grad_norm": 0.08679771423339844, "global_step": 30289, "epoch": 285, "lr": 3.886157026631288e-05} +{"train_loss": 0.01414173562079668, "grad_norm": 0.05701078847050667, "global_step": 30290, "epoch": 285, "lr": 3.885868099467664e-05} +{"train_loss": 0.015406174585223198, "grad_norm": 0.06695452332496643, "global_step": 30291, "epoch": 285, "lr": 3.885579176218614e-05} +{"train_loss": 0.0151373827829957, "grad_norm": 0.058891184628009796, "global_step": 30292, "epoch": 285, "lr": 3.885290256885155e-05} +{"train_loss": 0.017040979117155075, "grad_norm": 0.06701164692640305, "global_step": 30293, "epoch": 285, "lr": 3.885001341468301e-05} +{"train_loss": 0.017587371170520782, "grad_norm": 0.1268838346004486, "global_step": 30294, "epoch": 285, "lr": 3.884712429969066e-05} +{"train_loss": 0.014999940060079098, "grad_norm": 0.1316259354352951, "global_step": 30295, "epoch": 285, "lr": 3.884423522388467e-05} +{"train_loss": 0.014299841597676277, "grad_norm": 0.06295209378004074, "global_step": 30296, "epoch": 285, "lr": 3.884134618727518e-05} +{"train_loss": 0.015850290656089783, "grad_norm": 0.10302256047725677, "global_step": 30297, "epoch": 285, "lr": 3.883845718987238e-05} +{"train_loss": 0.01354283094406128, "grad_norm": 0.09381617605686188, "global_step": 30298, "epoch": 285, "lr": 3.883556823168635e-05} +{"train_loss": 0.01524246484041214, "grad_norm": 0.14401033520698547, "global_step": 30299, "epoch": 285, "lr": 3.8832679312727296e-05} +{"train_loss": 0.015700049698352814, "grad_norm": 0.0843791514635086, "global_step": 30300, "epoch": 285, "lr": 3.882979043300534e-05} +{"train_loss": 0.015613419003784657, "grad_norm": 0.10896077007055283, "global_step": 30301, "epoch": 285, "lr": 3.882690159253064e-05} +{"train_loss": 0.01699686422944069, "grad_norm": 0.08142931014299393, "global_step": 30302, "epoch": 285, "lr": 3.882401279131336e-05} +{"train_loss": 0.014554545283317566, "grad_norm": 0.08642749488353729, "global_step": 30303, "epoch": 285, "lr": 3.882112402936365e-05} +{"train_loss": 0.014295587316155434, "grad_norm": 0.058998800814151764, "global_step": 30304, "epoch": 285, "lr": 3.881823530669164e-05} +{"train_loss": 0.013359403237700462, "grad_norm": 0.06815040111541748, "global_step": 30305, "epoch": 285, "lr": 3.8815346623307483e-05} +{"train_loss": 0.01590551994740963, "grad_norm": 0.11382834613323212, "global_step": 30306, "epoch": 285, "lr": 3.881245797922132e-05} +{"train_loss": 0.01573830470442772, "grad_norm": 0.1018487960100174, "global_step": 30307, "epoch": 285, "lr": 3.8809569374443325e-05} +{"train_loss": 0.018057964742183685, "grad_norm": 0.19541916251182556, "global_step": 30308, "epoch": 285, "lr": 3.880668080898364e-05} +{"train_loss": 0.013587340712547302, "grad_norm": 0.09600879997015, "global_step": 30309, "epoch": 285, "lr": 3.8803792282852423e-05} +{"train_loss": 0.01727808639407158, "grad_norm": 0.10744782537221909, "global_step": 30310, "epoch": 285, "lr": 3.880090379605979e-05} +{"train_loss": 0.018605850636959076, "grad_norm": 0.15184253454208374, "global_step": 30311, "epoch": 285, "lr": 3.879801534861592e-05} +{"train_loss": 0.014592180028557777, "grad_norm": 0.10162750631570816, "global_step": 30312, "epoch": 285, "lr": 3.879512694053095e-05} +{"train_loss": 0.013323103077709675, "grad_norm": 0.12681128084659576, "global_step": 30313, "epoch": 285, "lr": 3.8792238571815024e-05} +{"train_loss": 0.01514316163957119, "grad_norm": 0.1510026603937149, "global_step": 30314, "epoch": 285, "lr": 3.878935024247831e-05} +{"train_loss": 0.015364743567089428, "grad_norm": 0.12311059981584549, "global_step": 30315, "epoch": 285, "lr": 3.878646195253095e-05, "val_loss": 0.019184596836566925} +{"train_loss": 0.016025826334953308, "grad_norm": 0.09391877055168152, "global_step": 30316, "epoch": 286, "lr": 3.878357370198308e-05} +{"train_loss": 0.01603233627974987, "grad_norm": 0.06850766390562057, "global_step": 30317, "epoch": 286, "lr": 3.878068549084485e-05} +{"train_loss": 0.01501577626913786, "grad_norm": 0.05851880460977554, "global_step": 30318, "epoch": 286, "lr": 3.87777973191264e-05} +{"train_loss": 0.014647669158875942, "grad_norm": 0.10192633420228958, "global_step": 30319, "epoch": 286, "lr": 3.87749091868379e-05} +{"train_loss": 0.013543648645281792, "grad_norm": 0.0687744989991188, "global_step": 30320, "epoch": 286, "lr": 3.877202109398949e-05} +{"train_loss": 0.016274945810437202, "grad_norm": 0.12164812535047531, "global_step": 30321, "epoch": 286, "lr": 3.8769133040591335e-05} +{"train_loss": 0.015196269378066063, "grad_norm": 0.06886057555675507, "global_step": 30322, "epoch": 286, "lr": 3.876624502665352e-05} +{"train_loss": 0.015720581635832787, "grad_norm": 0.14529719948768616, "global_step": 30323, "epoch": 286, "lr": 3.876335705218626e-05} +{"train_loss": 0.014148646034300327, "grad_norm": 0.1529051810503006, "global_step": 30324, "epoch": 286, "lr": 3.876046911719968e-05} +{"train_loss": 0.014003129675984383, "grad_norm": 0.14690060913562775, "global_step": 30325, "epoch": 286, "lr": 3.8757581221703906e-05} +{"train_loss": 0.015977168455719948, "grad_norm": 0.13574187457561493, "global_step": 30326, "epoch": 286, "lr": 3.875469336570911e-05} +{"train_loss": 0.016043439507484436, "grad_norm": 0.07774587720632553, "global_step": 30327, "epoch": 286, "lr": 3.8751805549225444e-05} +{"train_loss": 0.015384290367364883, "grad_norm": 0.04912261292338371, "global_step": 30328, "epoch": 286, "lr": 3.8748917772263036e-05} +{"train_loss": 0.014196626842021942, "grad_norm": 0.11197947710752487, "global_step": 30329, "epoch": 286, "lr": 3.8746030034832035e-05} +{"train_loss": 0.014100085943937302, "grad_norm": 0.1523129642009735, "global_step": 30330, "epoch": 286, "lr": 3.874314233694258e-05} +{"train_loss": 0.016620298847556114, "grad_norm": 0.06821873039007187, "global_step": 30331, "epoch": 286, "lr": 3.874025467860483e-05} +{"train_loss": 0.014069778844714165, "grad_norm": 0.08360102772712708, "global_step": 30332, "epoch": 286, "lr": 3.873736705982894e-05} +{"train_loss": 0.012950103729963303, "grad_norm": 0.07653363794088364, "global_step": 30333, "epoch": 286, "lr": 3.873447948062505e-05} +{"train_loss": 0.015051133930683136, "grad_norm": 0.09377642720937729, "global_step": 30334, "epoch": 286, "lr": 3.8731591941003286e-05} +{"train_loss": 0.013501943089067936, "grad_norm": 0.05722424015402794, "global_step": 30335, "epoch": 286, "lr": 3.87287044409738e-05} +{"train_loss": 0.014727646484971046, "grad_norm": 0.07927339524030685, "global_step": 30336, "epoch": 286, "lr": 3.872581698054676e-05} +{"train_loss": 0.015670444816350937, "grad_norm": 0.1091853529214859, "global_step": 30337, "epoch": 286, "lr": 3.872292955973229e-05} +{"train_loss": 0.014738025143742561, "grad_norm": 0.05628667771816254, "global_step": 30338, "epoch": 286, "lr": 3.872004217854053e-05} +{"train_loss": 0.015795867890119553, "grad_norm": 0.057951390743255615, "global_step": 30339, "epoch": 286, "lr": 3.8717154836981665e-05} +{"train_loss": 0.014401819556951523, "grad_norm": 0.07064151763916016, "global_step": 30340, "epoch": 286, "lr": 3.871426753506579e-05} +{"train_loss": 0.015699557960033417, "grad_norm": 0.10142135620117188, "global_step": 30341, "epoch": 286, "lr": 3.871138027280308e-05} +{"train_loss": 0.01519947499036789, "grad_norm": 0.1606736183166504, "global_step": 30342, "epoch": 286, "lr": 3.8708493050203654e-05} +{"train_loss": 0.015510795637965202, "grad_norm": 0.08658061176538467, "global_step": 30343, "epoch": 286, "lr": 3.870560586727768e-05} +{"train_loss": 0.01599450781941414, "grad_norm": 0.08892782777547836, "global_step": 30344, "epoch": 286, "lr": 3.8702718724035314e-05} +{"train_loss": 0.0124971317127347, "grad_norm": 0.21357426047325134, "global_step": 30345, "epoch": 286, "lr": 3.869983162048667e-05} +{"train_loss": 0.016111822798848152, "grad_norm": 0.06612510979175568, "global_step": 30346, "epoch": 286, "lr": 3.8696944556641903e-05} +{"train_loss": 0.017280422151088715, "grad_norm": 0.07843280583620071, "global_step": 30347, "epoch": 286, "lr": 3.8694057532511144e-05} +{"train_loss": 0.01168177928775549, "grad_norm": 0.08427519351243973, "global_step": 30348, "epoch": 286, "lr": 3.869117054810456e-05} +{"train_loss": 0.015517144463956356, "grad_norm": 0.14321288466453552, "global_step": 30349, "epoch": 286, "lr": 3.868828360343228e-05} +{"train_loss": 0.01787056028842926, "grad_norm": 0.07628722488880157, "global_step": 30350, "epoch": 286, "lr": 3.868539669850447e-05} +{"train_loss": 0.015413960441946983, "grad_norm": 0.06481686979532242, "global_step": 30351, "epoch": 286, "lr": 3.868250983333124e-05} +{"train_loss": 0.015095735900104046, "grad_norm": 0.07260521501302719, "global_step": 30352, "epoch": 286, "lr": 3.867962300792275e-05} +{"train_loss": 0.01583840884268284, "grad_norm": 0.07670240104198456, "global_step": 30353, "epoch": 286, "lr": 3.867673622228915e-05} +{"train_loss": 0.014993716031312943, "grad_norm": 0.06394411623477936, "global_step": 30354, "epoch": 286, "lr": 3.867384947644055e-05} +{"train_loss": 0.015175802633166313, "grad_norm": 0.06597702205181122, "global_step": 30355, "epoch": 286, "lr": 3.867096277038714e-05} +{"train_loss": 0.017013313248753548, "grad_norm": 0.08567392081022263, "global_step": 30356, "epoch": 286, "lr": 3.8668076104139046e-05} +{"train_loss": 0.01601121947169304, "grad_norm": 0.07369076460599899, "global_step": 30357, "epoch": 286, "lr": 3.8665189477706396e-05} +{"train_loss": 0.014556147158145905, "grad_norm": 0.09133881330490112, "global_step": 30358, "epoch": 286, "lr": 3.866230289109933e-05} +{"train_loss": 0.01443720143288374, "grad_norm": 0.16082914173603058, "global_step": 30359, "epoch": 286, "lr": 3.8659416344328e-05} +{"train_loss": 0.01681789755821228, "grad_norm": 0.08038689196109772, "global_step": 30360, "epoch": 286, "lr": 3.8656529837402566e-05} +{"train_loss": 0.015193179249763489, "grad_norm": 0.07117824256420135, "global_step": 30361, "epoch": 286, "lr": 3.865364337033314e-05} +{"train_loss": 0.014738375321030617, "grad_norm": 0.11787620931863785, "global_step": 30362, "epoch": 286, "lr": 3.86507569431299e-05} +{"train_loss": 0.015401845797896385, "grad_norm": 0.16166333854198456, "global_step": 30363, "epoch": 286, "lr": 3.8647870555802946e-05} +{"train_loss": 0.017572803422808647, "grad_norm": 0.21758943796157837, "global_step": 30364, "epoch": 286, "lr": 3.864498420836243e-05} +{"train_loss": 0.01665499433875084, "grad_norm": 0.0676460787653923, "global_step": 30365, "epoch": 286, "lr": 3.864209790081851e-05} +{"train_loss": 0.015462426468729973, "grad_norm": 0.06717830151319504, "global_step": 30366, "epoch": 286, "lr": 3.863921163318132e-05} +{"train_loss": 0.018196240067481995, "grad_norm": 0.07481806725263596, "global_step": 30367, "epoch": 286, "lr": 3.863632540546098e-05} +{"train_loss": 0.015534860081970692, "grad_norm": 0.06121942400932312, "global_step": 30368, "epoch": 286, "lr": 3.863343921766769e-05} +{"train_loss": 0.016741406172513962, "grad_norm": 0.08038779348134995, "global_step": 30369, "epoch": 286, "lr": 3.863055306981153e-05} +{"train_loss": 0.013637106865644455, "grad_norm": 0.06636054813861847, "global_step": 30370, "epoch": 286, "lr": 3.862766696190266e-05} +{"train_loss": 0.01412076037377119, "grad_norm": 0.05298543721437454, "global_step": 30371, "epoch": 286, "lr": 3.8624780893951215e-05} +{"train_loss": 0.017537668347358704, "grad_norm": 0.059645120054483414, "global_step": 30372, "epoch": 286, "lr": 3.862189486596735e-05} +{"train_loss": 0.013915843330323696, "grad_norm": 0.06510694324970245, "global_step": 30373, "epoch": 286, "lr": 3.86190088779612e-05} +{"train_loss": 0.01574859954416752, "grad_norm": 0.08248701691627502, "global_step": 30374, "epoch": 286, "lr": 3.861612292994292e-05} +{"train_loss": 0.014861679635941982, "grad_norm": 0.10832513123750687, "global_step": 30375, "epoch": 286, "lr": 3.861323702192262e-05} +{"train_loss": 0.013777869753539562, "grad_norm": 0.10439422726631165, "global_step": 30376, "epoch": 286, "lr": 3.861035115391044e-05} +{"train_loss": 0.0168914794921875, "grad_norm": 0.14938314259052277, "global_step": 30377, "epoch": 286, "lr": 3.8607465325916543e-05} +{"train_loss": 0.01705687679350376, "grad_norm": 0.10301972925662994, "global_step": 30378, "epoch": 286, "lr": 3.860457953795106e-05} +{"train_loss": 0.014768361113965511, "grad_norm": 0.1222304254770279, "global_step": 30379, "epoch": 286, "lr": 3.860169379002412e-05} +{"train_loss": 0.014541639015078545, "grad_norm": 0.09569110721349716, "global_step": 30380, "epoch": 286, "lr": 3.859880808214589e-05} +{"train_loss": 0.015571871772408485, "grad_norm": 0.11639642715454102, "global_step": 30381, "epoch": 286, "lr": 3.859592241432648e-05} +{"train_loss": 0.01569780521094799, "grad_norm": 0.1710570603609085, "global_step": 30382, "epoch": 286, "lr": 3.859303678657604e-05} +{"train_loss": 0.016619304195046425, "grad_norm": 0.07962533831596375, "global_step": 30383, "epoch": 286, "lr": 3.859015119890469e-05} +{"train_loss": 0.014951989986002445, "grad_norm": 0.06679916381835938, "global_step": 30384, "epoch": 286, "lr": 3.858726565132261e-05} +{"train_loss": 0.01673656702041626, "grad_norm": 0.05993252620100975, "global_step": 30385, "epoch": 286, "lr": 3.8584380143839904e-05} +{"train_loss": 0.015250420197844505, "grad_norm": 0.0691554993391037, "global_step": 30386, "epoch": 286, "lr": 3.858149467646674e-05} +{"train_loss": 0.01625565066933632, "grad_norm": 0.10077445954084396, "global_step": 30387, "epoch": 286, "lr": 3.857860924921322e-05} +{"train_loss": 0.015530012547969818, "grad_norm": 0.1517781764268875, "global_step": 30388, "epoch": 286, "lr": 3.8575723862089493e-05} +{"train_loss": 0.014083278365433216, "grad_norm": 0.09206794947385788, "global_step": 30389, "epoch": 286, "lr": 3.857283851510572e-05} +{"train_loss": 0.016385648399591446, "grad_norm": 0.10929359495639801, "global_step": 30390, "epoch": 286, "lr": 3.8569953208272015e-05} +{"train_loss": 0.01603015884757042, "grad_norm": 0.15043173730373383, "global_step": 30391, "epoch": 286, "lr": 3.856706794159852e-05} +{"train_loss": 0.01903805322945118, "grad_norm": 0.18798184394836426, "global_step": 30392, "epoch": 286, "lr": 3.85641827150954e-05} +{"train_loss": 0.014879588037729263, "grad_norm": 0.06838245689868927, "global_step": 30393, "epoch": 286, "lr": 3.856129752877274e-05} +{"train_loss": 0.015081954188644886, "grad_norm": 0.11246972531080246, "global_step": 30394, "epoch": 286, "lr": 3.8558412382640716e-05} +{"train_loss": 0.016352951526641846, "grad_norm": 0.0987105518579483, "global_step": 30395, "epoch": 286, "lr": 3.8555527276709456e-05} +{"train_loss": 0.01547392550855875, "grad_norm": 0.22299523651599884, "global_step": 30396, "epoch": 286, "lr": 3.855264221098909e-05} +{"train_loss": 0.014983659610152245, "grad_norm": 0.08275968581438065, "global_step": 30397, "epoch": 286, "lr": 3.854975718548978e-05} +{"train_loss": 0.014198236167430878, "grad_norm": 0.08413558453321457, "global_step": 30398, "epoch": 286, "lr": 3.8546872200221626e-05} +{"train_loss": 0.013847537338733673, "grad_norm": 0.05764134228229523, "global_step": 30399, "epoch": 286, "lr": 3.854398725519479e-05} +{"train_loss": 0.01359647698700428, "grad_norm": 0.11851207166910172, "global_step": 30400, "epoch": 286, "lr": 3.854110235041939e-05} +{"train_loss": 0.013251925818622112, "grad_norm": 0.04740937799215317, "global_step": 30401, "epoch": 286, "lr": 3.853821748590558e-05} +{"train_loss": 0.013465063646435738, "grad_norm": 0.08595229685306549, "global_step": 30402, "epoch": 286, "lr": 3.853533266166349e-05} +{"train_loss": 0.016796596348285675, "grad_norm": 0.12790167331695557, "global_step": 30403, "epoch": 286, "lr": 3.853244787770327e-05} +{"train_loss": 0.013896926306188107, "grad_norm": 0.05424116924405098, "global_step": 30404, "epoch": 286, "lr": 3.8529563134035025e-05} +{"train_loss": 0.016023997217416763, "grad_norm": 0.0948677659034729, "global_step": 30405, "epoch": 286, "lr": 3.85266784306689e-05} +{"train_loss": 0.015648748725652695, "grad_norm": 0.0752219706773758, "global_step": 30406, "epoch": 286, "lr": 3.852379376761505e-05} +{"train_loss": 0.017646275460720062, "grad_norm": 0.15259724855422974, "global_step": 30407, "epoch": 286, "lr": 3.852090914488359e-05} +{"train_loss": 0.01650826819241047, "grad_norm": 0.10552963614463806, "global_step": 30408, "epoch": 286, "lr": 3.8518024562484655e-05} +{"train_loss": 0.014919155277311802, "grad_norm": 0.05531283840537071, "global_step": 30409, "epoch": 286, "lr": 3.851514002042841e-05} +{"train_loss": 0.014466439373791218, "grad_norm": 0.08518410474061966, "global_step": 30410, "epoch": 286, "lr": 3.851225551872496e-05} +{"train_loss": 0.01725248619914055, "grad_norm": 0.1980505883693695, "global_step": 30411, "epoch": 286, "lr": 3.8509371057384444e-05} +{"train_loss": 0.01516261138021946, "grad_norm": 0.05192587524652481, "global_step": 30412, "epoch": 286, "lr": 3.8506486636416995e-05} +{"train_loss": 0.013586441054940224, "grad_norm": 0.10794702172279358, "global_step": 30413, "epoch": 286, "lr": 3.850360225583276e-05} +{"train_loss": 0.016204364597797394, "grad_norm": 0.06142156571149826, "global_step": 30414, "epoch": 286, "lr": 3.850071791564187e-05} +{"train_loss": 0.016383646056056023, "grad_norm": 0.10406555980443954, "global_step": 30415, "epoch": 286, "lr": 3.849783361585447e-05} +{"train_loss": 0.017364107072353363, "grad_norm": 0.17751669883728027, "global_step": 30416, "epoch": 286, "lr": 3.8494949356480664e-05} +{"train_loss": 0.013358019292354584, "grad_norm": 0.07716750353574753, "global_step": 30417, "epoch": 286, "lr": 3.849206513753059e-05} +{"train_loss": 0.016387835144996643, "grad_norm": 0.11575537174940109, "global_step": 30418, "epoch": 286, "lr": 3.848918095901441e-05} +{"train_loss": 0.015767689794301987, "grad_norm": 0.11240474879741669, "global_step": 30419, "epoch": 286, "lr": 3.8486296820942246e-05} +{"train_loss": 0.018115300685167313, "grad_norm": 0.1198076531291008, "global_step": 30420, "epoch": 286, "lr": 3.848341272332421e-05} +{"train_loss": 0.015404423197977385, "grad_norm": 0.09552592784166336, "global_step": 30421, "epoch": 286, "lr": 3.848052866617049e-05, "val_loss": 0.0190467219799757} +{"train_loss": 0.016147593036293983, "grad_norm": 0.06672896444797516, "global_step": 30422, "epoch": 287, "lr": 3.847764464949115e-05} +{"train_loss": 0.016357343643903732, "grad_norm": 0.18368199467658997, "global_step": 30423, "epoch": 287, "lr": 3.847476067329636e-05} +{"train_loss": 0.015622492879629135, "grad_norm": 0.09416083246469498, "global_step": 30424, "epoch": 287, "lr": 3.847187673759626e-05} +{"train_loss": 0.01664036139845848, "grad_norm": 0.07080314308404922, "global_step": 30425, "epoch": 287, "lr": 3.846899284240095e-05} +{"train_loss": 0.015856299549341202, "grad_norm": 0.057540733367204666, "global_step": 30426, "epoch": 287, "lr": 3.84661089877206e-05} +{"train_loss": 0.016051772981882095, "grad_norm": 0.07246620953083038, "global_step": 30427, "epoch": 287, "lr": 3.8463225173565335e-05} +{"train_loss": 0.013533584773540497, "grad_norm": 0.13257858157157898, "global_step": 30428, "epoch": 287, "lr": 3.846034139994527e-05} +{"train_loss": 0.016131192445755005, "grad_norm": 0.11662383377552032, "global_step": 30429, "epoch": 287, "lr": 3.845745766687054e-05} +{"train_loss": 0.01625051721930504, "grad_norm": 0.0586458183825016, "global_step": 30430, "epoch": 287, "lr": 3.845457397435129e-05} +{"train_loss": 0.016613449901342392, "grad_norm": 0.060312576591968536, "global_step": 30431, "epoch": 287, "lr": 3.845169032239765e-05} +{"train_loss": 0.01327848806977272, "grad_norm": 0.059087350964546204, "global_step": 30432, "epoch": 287, "lr": 3.844880671101974e-05} +{"train_loss": 0.01458451896905899, "grad_norm": 0.1388280689716339, "global_step": 30433, "epoch": 287, "lr": 3.844592314022772e-05} +{"train_loss": 0.01662779413163662, "grad_norm": 0.10466347634792328, "global_step": 30434, "epoch": 287, "lr": 3.8443039610031676e-05} +{"train_loss": 0.01696774736046791, "grad_norm": 0.06773121654987335, "global_step": 30435, "epoch": 287, "lr": 3.844015612044178e-05} +{"train_loss": 0.016912275925278664, "grad_norm": 0.10199661552906036, "global_step": 30436, "epoch": 287, "lr": 3.8437272671468146e-05} +{"train_loss": 0.016460781916975975, "grad_norm": 0.06073009595274925, "global_step": 30437, "epoch": 287, "lr": 3.8434389263120894e-05} +{"train_loss": 0.014021612703800201, "grad_norm": 0.10851982980966568, "global_step": 30438, "epoch": 287, "lr": 3.8431505895410184e-05} +{"train_loss": 0.012661127373576164, "grad_norm": 0.08050285279750824, "global_step": 30439, "epoch": 287, "lr": 3.842862256834613e-05} +{"train_loss": 0.01604180410504341, "grad_norm": 0.10656776279211044, "global_step": 30440, "epoch": 287, "lr": 3.842573928193887e-05} +{"train_loss": 0.015891917049884796, "grad_norm": 0.08032019436359406, "global_step": 30441, "epoch": 287, "lr": 3.84228560361985e-05} +{"train_loss": 0.015040304511785507, "grad_norm": 0.08968880772590637, "global_step": 30442, "epoch": 287, "lr": 3.84199728311352e-05} +{"train_loss": 0.013458301313221455, "grad_norm": 0.11438040435314178, "global_step": 30443, "epoch": 287, "lr": 3.841708966675908e-05} +{"train_loss": 0.01744232513010502, "grad_norm": 0.10357251018285751, "global_step": 30444, "epoch": 287, "lr": 3.8414206543080275e-05} +{"train_loss": 0.013015707954764366, "grad_norm": 0.05902593210339546, "global_step": 30445, "epoch": 287, "lr": 3.8411323460108906e-05} +{"train_loss": 0.014803659170866013, "grad_norm": 0.08510255813598633, "global_step": 30446, "epoch": 287, "lr": 3.8408440417855094e-05} +{"train_loss": 0.01604124903678894, "grad_norm": 0.07084964215755463, "global_step": 30447, "epoch": 287, "lr": 3.840555741632899e-05} +{"train_loss": 0.014017324894666672, "grad_norm": 0.08315904438495636, "global_step": 30448, "epoch": 287, "lr": 3.8402674455540715e-05} +{"train_loss": 0.013935187831521034, "grad_norm": 0.06285280734300613, "global_step": 30449, "epoch": 287, "lr": 3.839979153550039e-05} +{"train_loss": 0.012009713798761368, "grad_norm": 0.06391913443803787, "global_step": 30450, "epoch": 287, "lr": 3.839690865621818e-05} +{"train_loss": 0.014438916929066181, "grad_norm": 0.06616530567407608, "global_step": 30451, "epoch": 287, "lr": 3.839402581770416e-05} +{"train_loss": 0.017028389498591423, "grad_norm": 0.12438389658927917, "global_step": 30452, "epoch": 287, "lr": 3.8391143019968493e-05} +{"train_loss": 0.017035122960805893, "grad_norm": 0.057579200714826584, "global_step": 30453, "epoch": 287, "lr": 3.838826026302129e-05} +{"train_loss": 0.015234870836138725, "grad_norm": 0.07258496433496475, "global_step": 30454, "epoch": 287, "lr": 3.838537754687271e-05} +{"train_loss": 0.015489831566810608, "grad_norm": 0.06148986890912056, "global_step": 30455, "epoch": 287, "lr": 3.838249487153285e-05} +{"train_loss": 0.015251598320901394, "grad_norm": 0.07130742818117142, "global_step": 30456, "epoch": 287, "lr": 3.837961223701187e-05} +{"train_loss": 0.01982816495001316, "grad_norm": 0.08921931684017181, "global_step": 30457, "epoch": 287, "lr": 3.8376729643319864e-05} +{"train_loss": 0.013763239607214928, "grad_norm": 0.10181158035993576, "global_step": 30458, "epoch": 287, "lr": 3.837384709046697e-05} +{"train_loss": 0.013218594714999199, "grad_norm": 0.05983724817633629, "global_step": 30459, "epoch": 287, "lr": 3.837096457846333e-05} +{"train_loss": 0.014918576925992966, "grad_norm": 0.05345568433403969, "global_step": 30460, "epoch": 287, "lr": 3.836808210731906e-05} +{"train_loss": 0.016483530402183533, "grad_norm": 0.107376828789711, "global_step": 30461, "epoch": 287, "lr": 3.836519967704428e-05} +{"train_loss": 0.015473504550755024, "grad_norm": 0.06919838488101959, "global_step": 30462, "epoch": 287, "lr": 3.836231728764916e-05} +{"train_loss": 0.014969313517212868, "grad_norm": 0.0792374461889267, "global_step": 30463, "epoch": 287, "lr": 3.8359434939143766e-05} +{"train_loss": 0.016000274568796158, "grad_norm": 0.0748477429151535, "global_step": 30464, "epoch": 287, "lr": 3.835655263153826e-05} +{"train_loss": 0.015473057515919209, "grad_norm": 0.09149038791656494, "global_step": 30465, "epoch": 287, "lr": 3.8353670364842775e-05} +{"train_loss": 0.015738271176815033, "grad_norm": 0.05729985237121582, "global_step": 30466, "epoch": 287, "lr": 3.8350788139067414e-05} +{"train_loss": 0.015405142679810524, "grad_norm": 0.11968725919723511, "global_step": 30467, "epoch": 287, "lr": 3.834790595422232e-05} +{"train_loss": 0.01722959242761135, "grad_norm": 0.11954076588153839, "global_step": 30468, "epoch": 287, "lr": 3.8345023810317636e-05} +{"train_loss": 0.014360717497766018, "grad_norm": 0.09087279438972473, "global_step": 30469, "epoch": 287, "lr": 3.8342141707363454e-05} +{"train_loss": 0.01605270989239216, "grad_norm": 0.13966074585914612, "global_step": 30470, "epoch": 287, "lr": 3.833925964536991e-05} +{"train_loss": 0.017351871356368065, "grad_norm": 0.06752036511898041, "global_step": 30471, "epoch": 287, "lr": 3.833637762434714e-05} +{"train_loss": 0.014587962999939919, "grad_norm": 0.08399896323680878, "global_step": 30472, "epoch": 287, "lr": 3.8333495644305275e-05} +{"train_loss": 0.018155483528971672, "grad_norm": 0.130843847990036, "global_step": 30473, "epoch": 287, "lr": 3.833061370525442e-05} +{"train_loss": 0.01421772874891758, "grad_norm": 0.08577364683151245, "global_step": 30474, "epoch": 287, "lr": 3.832773180720475e-05} +{"train_loss": 0.016123948618769646, "grad_norm": 0.14347827434539795, "global_step": 30475, "epoch": 287, "lr": 3.8324849950166314e-05} +{"train_loss": 0.01587022840976715, "grad_norm": 0.07738497853279114, "global_step": 30476, "epoch": 287, "lr": 3.8321968134149286e-05} +{"train_loss": 0.013515131548047066, "grad_norm": 0.062449220567941666, "global_step": 30477, "epoch": 287, "lr": 3.8319086359163795e-05} +{"train_loss": 0.016702843829989433, "grad_norm": 0.07237207889556885, "global_step": 30478, "epoch": 287, "lr": 3.831620462521993e-05} +{"train_loss": 0.015228839591145515, "grad_norm": 0.09944729506969452, "global_step": 30479, "epoch": 287, "lr": 3.831332293232787e-05} +{"train_loss": 0.014020189642906189, "grad_norm": 0.07786960899829865, "global_step": 30480, "epoch": 287, "lr": 3.83104412804977e-05} +{"train_loss": 0.014675221405923367, "grad_norm": 0.059199873358011246, "global_step": 30481, "epoch": 287, "lr": 3.830755966973956e-05} +{"train_loss": 0.013479171320796013, "grad_norm": 0.06976983696222305, "global_step": 30482, "epoch": 287, "lr": 3.830467810006356e-05} +{"train_loss": 0.015475932508707047, "grad_norm": 0.09435218572616577, "global_step": 30483, "epoch": 287, "lr": 3.830179657147984e-05} +{"train_loss": 0.01532992534339428, "grad_norm": 0.07738891988992691, "global_step": 30484, "epoch": 287, "lr": 3.829891508399852e-05} +{"train_loss": 0.0156516432762146, "grad_norm": 0.08102691173553467, "global_step": 30485, "epoch": 287, "lr": 3.829603363762971e-05} +{"train_loss": 0.013256702572107315, "grad_norm": 0.06942308694124222, "global_step": 30486, "epoch": 287, "lr": 3.829315223238358e-05} +{"train_loss": 0.01571602188050747, "grad_norm": 0.07214635610580444, "global_step": 30487, "epoch": 287, "lr": 3.82902708682702e-05} +{"train_loss": 0.015314547345042229, "grad_norm": 0.07106699049472809, "global_step": 30488, "epoch": 287, "lr": 3.828738954529972e-05} +{"train_loss": 0.01630975864827633, "grad_norm": 0.07181244343519211, "global_step": 30489, "epoch": 287, "lr": 3.8284508263482254e-05} +{"train_loss": 0.017104264348745346, "grad_norm": 0.07839512079954147, "global_step": 30490, "epoch": 287, "lr": 3.828162702282793e-05} +{"train_loss": 0.013800124637782574, "grad_norm": 0.08573734015226364, "global_step": 30491, "epoch": 287, "lr": 3.827874582334687e-05} +{"train_loss": 0.01292729377746582, "grad_norm": 0.08143644779920578, "global_step": 30492, "epoch": 287, "lr": 3.827586466504922e-05} +{"train_loss": 0.014056912623345852, "grad_norm": 0.14802095293998718, "global_step": 30493, "epoch": 287, "lr": 3.8272983547945076e-05} +{"train_loss": 0.014107316732406616, "grad_norm": 0.06724021583795547, "global_step": 30494, "epoch": 287, "lr": 3.827010247204456e-05} +{"train_loss": 0.014245809987187386, "grad_norm": 0.09986749291419983, "global_step": 30495, "epoch": 287, "lr": 3.8267221437357794e-05} +{"train_loss": 0.01614822819828987, "grad_norm": 0.10149300843477249, "global_step": 30496, "epoch": 287, "lr": 3.826434044389492e-05} +{"train_loss": 0.014506390318274498, "grad_norm": 0.07677261531352997, "global_step": 30497, "epoch": 287, "lr": 3.826145949166607e-05} +{"train_loss": 0.017467116937041283, "grad_norm": 0.10786371678113937, "global_step": 30498, "epoch": 287, "lr": 3.825857858068132e-05} +{"train_loss": 0.015826387330889702, "grad_norm": 0.06541253626346588, "global_step": 30499, "epoch": 287, "lr": 3.825569771095082e-05} +{"train_loss": 0.01728648506104946, "grad_norm": 0.0830371156334877, "global_step": 30500, "epoch": 287, "lr": 3.8252816882484696e-05} +{"train_loss": 0.015371333807706833, "grad_norm": 0.0527934655547142, "global_step": 30501, "epoch": 287, "lr": 3.824993609529306e-05} +{"train_loss": 0.015582502819597721, "grad_norm": 0.05844603106379509, "global_step": 30502, "epoch": 287, "lr": 3.824705534938604e-05} +{"train_loss": 0.016886502504348755, "grad_norm": 0.07552470266819, "global_step": 30503, "epoch": 287, "lr": 3.824417464477378e-05} +{"train_loss": 0.014735398814082146, "grad_norm": 0.11361522972583771, "global_step": 30504, "epoch": 287, "lr": 3.8241293981466356e-05} +{"train_loss": 0.015946906059980392, "grad_norm": 0.07286041229963303, "global_step": 30505, "epoch": 287, "lr": 3.823841335947391e-05} +{"train_loss": 0.01582879200577736, "grad_norm": 0.090468630194664, "global_step": 30506, "epoch": 287, "lr": 3.823553277880657e-05} +{"train_loss": 0.012031580321490765, "grad_norm": 0.0761156901717186, "global_step": 30507, "epoch": 287, "lr": 3.823265223947443e-05} +{"train_loss": 0.01650996319949627, "grad_norm": 0.0948459655046463, "global_step": 30508, "epoch": 287, "lr": 3.822977174148766e-05} +{"train_loss": 0.014906663447618484, "grad_norm": 0.05548685044050217, "global_step": 30509, "epoch": 287, "lr": 3.8226891284856354e-05} +{"train_loss": 0.016326306387782097, "grad_norm": 0.1172402948141098, "global_step": 30510, "epoch": 287, "lr": 3.822401086959062e-05} +{"train_loss": 0.013342579826712608, "grad_norm": 0.06622736155986786, "global_step": 30511, "epoch": 287, "lr": 3.822113049570059e-05} +{"train_loss": 0.01655104011297226, "grad_norm": 0.06519490480422974, "global_step": 30512, "epoch": 287, "lr": 3.8218250163196384e-05} +{"train_loss": 0.01463704276829958, "grad_norm": 0.10627054423093796, "global_step": 30513, "epoch": 287, "lr": 3.821536987208813e-05} +{"train_loss": 0.012984119355678558, "grad_norm": 0.0738065168261528, "global_step": 30514, "epoch": 287, "lr": 3.821248962238593e-05} +{"train_loss": 0.017122585326433182, "grad_norm": 0.07993423193693161, "global_step": 30515, "epoch": 287, "lr": 3.8209609414099946e-05} +{"train_loss": 0.014594554901123047, "grad_norm": 0.11426720768213272, "global_step": 30516, "epoch": 287, "lr": 3.820672924724024e-05} +{"train_loss": 0.014506000094115734, "grad_norm": 0.0712675079703331, "global_step": 30517, "epoch": 287, "lr": 3.820384912181696e-05} +{"train_loss": 0.016458626836538315, "grad_norm": 0.06839601695537567, "global_step": 30518, "epoch": 287, "lr": 3.8200969037840225e-05} +{"train_loss": 0.01603209599852562, "grad_norm": 0.07345467805862427, "global_step": 30519, "epoch": 287, "lr": 3.819808899532015e-05} +{"train_loss": 0.016455676406621933, "grad_norm": 0.11020238697528839, "global_step": 30520, "epoch": 287, "lr": 3.819520899426686e-05} +{"train_loss": 0.014345688745379448, "grad_norm": 0.05730588734149933, "global_step": 30521, "epoch": 287, "lr": 3.819232903469049e-05} +{"train_loss": 0.01466354075819254, "grad_norm": 0.09421754628419876, "global_step": 30522, "epoch": 287, "lr": 3.818944911660113e-05} +{"train_loss": 0.01589386910200119, "grad_norm": 0.08870980888605118, "global_step": 30523, "epoch": 287, "lr": 3.8186569240008906e-05} +{"train_loss": 0.017086517065763474, "grad_norm": 0.07589206099510193, "global_step": 30524, "epoch": 287, "lr": 3.8183689404923926e-05} +{"train_loss": 0.013518350198864937, "grad_norm": 0.06863006949424744, "global_step": 30525, "epoch": 287, "lr": 3.8180809611356347e-05} +{"train_loss": 0.01509126927703619, "grad_norm": 0.0778374969959259, "global_step": 30526, "epoch": 287, "lr": 3.817792985931625e-05} +{"train_loss": 0.015349098569098508, "grad_norm": 0.07347920536994934, "global_step": 30527, "epoch": 287, "lr": 3.817505014881378e-05, "val_loss": 0.02015731856226921} +{"train_loss": 0.016103414818644524, "grad_norm": 0.07038009911775589, "global_step": 30528, "epoch": 288, "lr": 3.817217047985902e-05} +{"train_loss": 0.014703592285513878, "grad_norm": 0.08803915232419968, "global_step": 30529, "epoch": 288, "lr": 3.8169290852462123e-05} +{"train_loss": 0.015786876901984215, "grad_norm": 0.08622835576534271, "global_step": 30530, "epoch": 288, "lr": 3.816641126663319e-05} +{"train_loss": 0.01669280230998993, "grad_norm": 0.06225256994366646, "global_step": 30531, "epoch": 288, "lr": 3.816353172238233e-05} +{"train_loss": 0.014760175719857216, "grad_norm": 0.11373315006494522, "global_step": 30532, "epoch": 288, "lr": 3.8160652219719687e-05} +{"train_loss": 0.01465511228889227, "grad_norm": 0.09107762575149536, "global_step": 30533, "epoch": 288, "lr": 3.815777275865537e-05} +{"train_loss": 0.018084514886140823, "grad_norm": 0.07301794737577438, "global_step": 30534, "epoch": 288, "lr": 3.815489333919948e-05} +{"train_loss": 0.014200139790773392, "grad_norm": 0.1439700573682785, "global_step": 30535, "epoch": 288, "lr": 3.815201396136214e-05} +{"train_loss": 0.014221110381186008, "grad_norm": 0.09759778529405594, "global_step": 30536, "epoch": 288, "lr": 3.814913462515346e-05} +{"train_loss": 0.012916753999888897, "grad_norm": 0.08196980506181717, "global_step": 30537, "epoch": 288, "lr": 3.8146255330583584e-05} +{"train_loss": 0.017452657222747803, "grad_norm": 0.10808900743722916, "global_step": 30538, "epoch": 288, "lr": 3.814337607766261e-05} +{"train_loss": 0.01627323217689991, "grad_norm": 0.11777153611183167, "global_step": 30539, "epoch": 288, "lr": 3.814049686640067e-05} +{"train_loss": 0.015644943341612816, "grad_norm": 0.08558273315429688, "global_step": 30540, "epoch": 288, "lr": 3.813761769680784e-05} +{"train_loss": 0.017503919079899788, "grad_norm": 0.11129707098007202, "global_step": 30541, "epoch": 288, "lr": 3.8134738568894266e-05} +{"train_loss": 0.012447603978216648, "grad_norm": 0.05669001489877701, "global_step": 30542, "epoch": 288, "lr": 3.813185948267007e-05} +{"train_loss": 0.016064457595348358, "grad_norm": 0.10326319187879562, "global_step": 30543, "epoch": 288, "lr": 3.812898043814535e-05} +{"train_loss": 0.01645701378583908, "grad_norm": 0.0904163271188736, "global_step": 30544, "epoch": 288, "lr": 3.812610143533023e-05} +{"train_loss": 0.013547276146709919, "grad_norm": 0.059392914175987244, "global_step": 30545, "epoch": 288, "lr": 3.812322247423484e-05} +{"train_loss": 0.015572018921375275, "grad_norm": 0.0751071497797966, "global_step": 30546, "epoch": 288, "lr": 3.812034355486928e-05} +{"train_loss": 0.016814718022942543, "grad_norm": 0.06779459863901138, "global_step": 30547, "epoch": 288, "lr": 3.8117464677243665e-05} +{"train_loss": 0.013229572214186192, "grad_norm": 0.07231837511062622, "global_step": 30548, "epoch": 288, "lr": 3.811458584136809e-05} +{"train_loss": 0.014869590289890766, "grad_norm": 0.0533437579870224, "global_step": 30549, "epoch": 288, "lr": 3.811170704725271e-05} +{"train_loss": 0.016642753034830093, "grad_norm": 0.09982611238956451, "global_step": 30550, "epoch": 288, "lr": 3.8108828294907626e-05} +{"train_loss": 0.014925493858754635, "grad_norm": 0.07490606606006622, "global_step": 30551, "epoch": 288, "lr": 3.810594958434295e-05} +{"train_loss": 0.014481527730822563, "grad_norm": 0.0756709948182106, "global_step": 30552, "epoch": 288, "lr": 3.8103070915568784e-05} +{"train_loss": 0.01595439948141575, "grad_norm": 0.1878517121076584, "global_step": 30553, "epoch": 288, "lr": 3.810019228859524e-05} +{"train_loss": 0.013034537434577942, "grad_norm": 0.13727667927742004, "global_step": 30554, "epoch": 288, "lr": 3.8097313703432464e-05} +{"train_loss": 0.01592535525560379, "grad_norm": 0.12622863054275513, "global_step": 30555, "epoch": 288, "lr": 3.8094435160090536e-05} +{"train_loss": 0.015514894388616085, "grad_norm": 0.06745357066392899, "global_step": 30556, "epoch": 288, "lr": 3.809155665857962e-05} +{"train_loss": 0.015827983617782593, "grad_norm": 0.10124902427196503, "global_step": 30557, "epoch": 288, "lr": 3.808867819890976e-05} +{"train_loss": 0.012938899919390678, "grad_norm": 0.07847646623849869, "global_step": 30558, "epoch": 288, "lr": 3.808579978109111e-05} +{"train_loss": 0.015719257295131683, "grad_norm": 0.09966365247964859, "global_step": 30559, "epoch": 288, "lr": 3.808292140513379e-05} +{"train_loss": 0.014230706728994846, "grad_norm": 0.056285444647073746, "global_step": 30560, "epoch": 288, "lr": 3.808004307104789e-05} +{"train_loss": 0.016944367438554764, "grad_norm": 0.08996112644672394, "global_step": 30561, "epoch": 288, "lr": 3.807716477884354e-05} +{"train_loss": 0.014448588714003563, "grad_norm": 0.07875265926122665, "global_step": 30562, "epoch": 288, "lr": 3.807428652853086e-05} +{"train_loss": 0.0159447081387043, "grad_norm": 0.06229792535305023, "global_step": 30563, "epoch": 288, "lr": 3.807140832011994e-05} +{"train_loss": 0.014856313355267048, "grad_norm": 0.067717045545578, "global_step": 30564, "epoch": 288, "lr": 3.80685301536209e-05} +{"train_loss": 0.014379123225808144, "grad_norm": 0.13610726594924927, "global_step": 30565, "epoch": 288, "lr": 3.806565202904385e-05} +{"train_loss": 0.014402903616428375, "grad_norm": 0.0680188313126564, "global_step": 30566, "epoch": 288, "lr": 3.806277394639892e-05} +{"train_loss": 0.014995891600847244, "grad_norm": 0.17671486735343933, "global_step": 30567, "epoch": 288, "lr": 3.805989590569621e-05} +{"train_loss": 0.014348854310810566, "grad_norm": 0.06269507855176926, "global_step": 30568, "epoch": 288, "lr": 3.805701790694584e-05} +{"train_loss": 0.014983577653765678, "grad_norm": 0.09912371635437012, "global_step": 30569, "epoch": 288, "lr": 3.80541399501579e-05} +{"train_loss": 0.016710396856069565, "grad_norm": 0.08157583326101303, "global_step": 30570, "epoch": 288, "lr": 3.805126203534252e-05} +{"train_loss": 0.014154840260744095, "grad_norm": 0.11567391455173492, "global_step": 30571, "epoch": 288, "lr": 3.804838416250981e-05} +{"train_loss": 0.015285039320588112, "grad_norm": 0.10297884047031403, "global_step": 30572, "epoch": 288, "lr": 3.8045506331669876e-05} +{"train_loss": 0.014385515823960304, "grad_norm": 0.06889183819293976, "global_step": 30573, "epoch": 288, "lr": 3.8042628542832845e-05} +{"train_loss": 0.014886693097651005, "grad_norm": 0.06941761076450348, "global_step": 30574, "epoch": 288, "lr": 3.8039750796008825e-05} +{"train_loss": 0.014105543494224548, "grad_norm": 0.08349892497062683, "global_step": 30575, "epoch": 288, "lr": 3.8036873091207906e-05} +{"train_loss": 0.015534838661551476, "grad_norm": 0.07389751076698303, "global_step": 30576, "epoch": 288, "lr": 3.803399542844022e-05} +{"train_loss": 0.014404358342289925, "grad_norm": 0.1023470014333725, "global_step": 30577, "epoch": 288, "lr": 3.8031117807715855e-05} +{"train_loss": 0.015456940978765488, "grad_norm": 0.07467818260192871, "global_step": 30578, "epoch": 288, "lr": 3.8028240229044945e-05} +{"train_loss": 0.014953184872865677, "grad_norm": 0.12861303985118866, "global_step": 30579, "epoch": 288, "lr": 3.8025362692437596e-05} +{"train_loss": 0.014470787718892097, "grad_norm": 0.10598205775022507, "global_step": 30580, "epoch": 288, "lr": 3.8022485197903925e-05} +{"train_loss": 0.013327222317457199, "grad_norm": 0.058686044067144394, "global_step": 30581, "epoch": 288, "lr": 3.8019607745454015e-05} +{"train_loss": 0.017390381544828415, "grad_norm": 0.06907860934734344, "global_step": 30582, "epoch": 288, "lr": 3.801673033509799e-05} +{"train_loss": 0.01687607914209366, "grad_norm": 0.09343352913856506, "global_step": 30583, "epoch": 288, "lr": 3.801385296684597e-05} +{"train_loss": 0.01505945809185505, "grad_norm": 0.08514563739299774, "global_step": 30584, "epoch": 288, "lr": 3.801097564070805e-05} +{"train_loss": 0.015624032355844975, "grad_norm": 0.06301990151405334, "global_step": 30585, "epoch": 288, "lr": 3.800809835669436e-05} +{"train_loss": 0.015749717131257057, "grad_norm": 0.0864173099398613, "global_step": 30586, "epoch": 288, "lr": 3.8005221114815e-05} +{"train_loss": 0.014353795908391476, "grad_norm": 0.06872277706861496, "global_step": 30587, "epoch": 288, "lr": 3.8002343915080076e-05} +{"train_loss": 0.015221762470901012, "grad_norm": 0.07006188482046127, "global_step": 30588, "epoch": 288, "lr": 3.799946675749969e-05} +{"train_loss": 0.014188585802912712, "grad_norm": 0.051029182970523834, "global_step": 30589, "epoch": 288, "lr": 3.799658964208395e-05} +{"train_loss": 0.015110349282622337, "grad_norm": 0.0860385000705719, "global_step": 30590, "epoch": 288, "lr": 3.799371256884299e-05} +{"train_loss": 0.013517890125513077, "grad_norm": 0.10914207249879837, "global_step": 30591, "epoch": 288, "lr": 3.7990835537786895e-05} +{"train_loss": 0.016583267599344254, "grad_norm": 0.06646670401096344, "global_step": 30592, "epoch": 288, "lr": 3.79879585489258e-05} +{"train_loss": 0.016600351780653, "grad_norm": 0.059134341776371, "global_step": 30593, "epoch": 288, "lr": 3.798508160226978e-05} +{"train_loss": 0.01554375421255827, "grad_norm": 0.05811801552772522, "global_step": 30594, "epoch": 288, "lr": 3.798220469782894e-05} +{"train_loss": 0.0162454042583704, "grad_norm": 0.07262561470270157, "global_step": 30595, "epoch": 288, "lr": 3.797932783561343e-05} +{"train_loss": 0.014342354610562325, "grad_norm": 0.11527843028306961, "global_step": 30596, "epoch": 288, "lr": 3.797645101563333e-05} +{"train_loss": 0.015414584428071976, "grad_norm": 0.09848251938819885, "global_step": 30597, "epoch": 288, "lr": 3.797357423789876e-05} +{"train_loss": 0.01542794518172741, "grad_norm": 0.06823714822530746, "global_step": 30598, "epoch": 288, "lr": 3.79706975024198e-05} +{"train_loss": 0.015481299720704556, "grad_norm": 0.07776537537574768, "global_step": 30599, "epoch": 288, "lr": 3.796782080920659e-05} +{"train_loss": 0.014784066937863827, "grad_norm": 0.07043930143117905, "global_step": 30600, "epoch": 288, "lr": 3.7964944158269224e-05} +{"train_loss": 0.014713666401803493, "grad_norm": 0.08985758572816849, "global_step": 30601, "epoch": 288, "lr": 3.7962067549617796e-05} +{"train_loss": 0.016904719173908234, "grad_norm": 0.06843772530555725, "global_step": 30602, "epoch": 288, "lr": 3.7959190983262435e-05} +{"train_loss": 0.014824369922280312, "grad_norm": 0.12611141800880432, "global_step": 30603, "epoch": 288, "lr": 3.795631445921326e-05} +{"train_loss": 0.013097625225782394, "grad_norm": 0.08502832800149918, "global_step": 30604, "epoch": 288, "lr": 3.795343797748035e-05} +{"train_loss": 0.015982739627361298, "grad_norm": 0.10852091014385223, "global_step": 30605, "epoch": 288, "lr": 3.795056153807381e-05} +{"train_loss": 0.01563100703060627, "grad_norm": 0.07694236934185028, "global_step": 30606, "epoch": 288, "lr": 3.7947685141003753e-05} +{"train_loss": 0.01609390787780285, "grad_norm": 0.07936520129442215, "global_step": 30607, "epoch": 288, "lr": 3.794480878628029e-05} +{"train_loss": 0.015120895579457283, "grad_norm": 0.1132829412817955, "global_step": 30608, "epoch": 288, "lr": 3.794193247391354e-05} +{"train_loss": 0.0161313284188509, "grad_norm": 0.07513268291950226, "global_step": 30609, "epoch": 288, "lr": 3.7939056203913604e-05} +{"train_loss": 0.014280378818511963, "grad_norm": 0.06669998168945312, "global_step": 30610, "epoch": 288, "lr": 3.7936179976290555e-05} +{"train_loss": 0.018305176869034767, "grad_norm": 0.11343558877706528, "global_step": 30611, "epoch": 288, "lr": 3.793330379105453e-05} +{"train_loss": 0.015580959618091583, "grad_norm": 0.0987265482544899, "global_step": 30612, "epoch": 288, "lr": 3.7930427648215625e-05} +{"train_loss": 0.01714289002120495, "grad_norm": 0.08877497166395187, "global_step": 30613, "epoch": 288, "lr": 3.7927551547783944e-05} +{"train_loss": 0.0171404629945755, "grad_norm": 0.08504626899957657, "global_step": 30614, "epoch": 288, "lr": 3.7924675489769606e-05} +{"train_loss": 0.01476999744772911, "grad_norm": 0.06889922171831131, "global_step": 30615, "epoch": 288, "lr": 3.7921799474182716e-05} +{"train_loss": 0.015754710882902145, "grad_norm": 0.07653313130140305, "global_step": 30616, "epoch": 288, "lr": 3.791892350103337e-05} +{"train_loss": 0.015074378810822964, "grad_norm": 0.06713193655014038, "global_step": 30617, "epoch": 288, "lr": 3.791604757033166e-05} +{"train_loss": 0.01611807383596897, "grad_norm": 0.1220138669013977, "global_step": 30618, "epoch": 288, "lr": 3.79131716820877e-05} +{"train_loss": 0.013617133721709251, "grad_norm": 0.09506024420261383, "global_step": 30619, "epoch": 288, "lr": 3.7910295836311606e-05} +{"train_loss": 0.013794582337141037, "grad_norm": 0.11706946045160294, "global_step": 30620, "epoch": 288, "lr": 3.790742003301347e-05} +{"train_loss": 0.014153950847685337, "grad_norm": 0.1014678031206131, "global_step": 30621, "epoch": 288, "lr": 3.790454427220342e-05} +{"train_loss": 0.01662566140294075, "grad_norm": 0.1277819573879242, "global_step": 30622, "epoch": 288, "lr": 3.790166855389152e-05} +{"train_loss": 0.016672659665346146, "grad_norm": 0.08002007752656937, "global_step": 30623, "epoch": 288, "lr": 3.7898792878087894e-05} +{"train_loss": 0.01645701937377453, "grad_norm": 0.1027480885386467, "global_step": 30624, "epoch": 288, "lr": 3.7895917244802656e-05} +{"train_loss": 0.014418978244066238, "grad_norm": 0.0691094696521759, "global_step": 30625, "epoch": 288, "lr": 3.78930416540459e-05} +{"train_loss": 0.01624227687716484, "grad_norm": 0.06834811717271805, "global_step": 30626, "epoch": 288, "lr": 3.7890166105827726e-05} +{"train_loss": 0.0159851536154747, "grad_norm": 0.07697287946939468, "global_step": 30627, "epoch": 288, "lr": 3.7887290600158254e-05} +{"train_loss": 0.018990937620401382, "grad_norm": 0.10888834297657013, "global_step": 30628, "epoch": 288, "lr": 3.7884415137047566e-05} +{"train_loss": 0.014424784108996391, "grad_norm": 0.0769151970744133, "global_step": 30629, "epoch": 288, "lr": 3.788153971650577e-05} +{"train_loss": 0.01480143889784813, "grad_norm": 0.062303777784109116, "global_step": 30630, "epoch": 288, "lr": 3.787866433854298e-05} +{"train_loss": 0.016021832823753357, "grad_norm": 0.11285138875246048, "global_step": 30631, "epoch": 288, "lr": 3.787578900316929e-05} +{"train_loss": 0.0156738068908453, "grad_norm": 0.1116407960653305, "global_step": 30632, "epoch": 288, "lr": 3.787291371039481e-05} +{"train_loss": 0.015354607342127359, "grad_norm": 0.08204556256532669, "global_step": 30633, "epoch": 288, "lr": 3.787003846022964e-05, "val_loss": 0.019577743485569954} +{"train_loss": 0.0194342490285635, "grad_norm": 0.19391268491744995, "global_step": 30634, "epoch": 289, "lr": 3.7867163252683875e-05} +{"train_loss": 0.013811297714710236, "grad_norm": 0.13735011219978333, "global_step": 30635, "epoch": 289, "lr": 3.7864288087767606e-05} +{"train_loss": 0.016533399000763893, "grad_norm": 0.07401540130376816, "global_step": 30636, "epoch": 289, "lr": 3.786141296549097e-05} +{"train_loss": 0.015367616899311543, "grad_norm": 0.09387949109077454, "global_step": 30637, "epoch": 289, "lr": 3.7858537885864046e-05} +{"train_loss": 0.016258852556347847, "grad_norm": 0.16191956400871277, "global_step": 30638, "epoch": 289, "lr": 3.785566284889692e-05} +{"train_loss": 0.016135767102241516, "grad_norm": 0.05660521239042282, "global_step": 30639, "epoch": 289, "lr": 3.7852787854599746e-05} +{"train_loss": 0.016101384535431862, "grad_norm": 0.08484911173582077, "global_step": 30640, "epoch": 289, "lr": 3.784991290298258e-05} +{"train_loss": 0.013595735654234886, "grad_norm": 0.07653220742940903, "global_step": 30641, "epoch": 289, "lr": 3.784703799405553e-05} +{"train_loss": 0.017603106796741486, "grad_norm": 0.07548830658197403, "global_step": 30642, "epoch": 289, "lr": 3.78441631278287e-05} +{"train_loss": 0.013609912246465683, "grad_norm": 0.12944923341274261, "global_step": 30643, "epoch": 289, "lr": 3.7841288304312206e-05} +{"train_loss": 0.01596478372812271, "grad_norm": 0.051927391439676285, "global_step": 30644, "epoch": 289, "lr": 3.7838413523516135e-05} +{"train_loss": 0.014232408255338669, "grad_norm": 0.057846952229738235, "global_step": 30645, "epoch": 289, "lr": 3.78355387854506e-05} +{"train_loss": 0.01595677062869072, "grad_norm": 0.06367884576320648, "global_step": 30646, "epoch": 289, "lr": 3.783266409012568e-05} +{"train_loss": 0.01606954261660576, "grad_norm": 0.12189485877752304, "global_step": 30647, "epoch": 289, "lr": 3.7829789437551476e-05} +{"train_loss": 0.015324749052524567, "grad_norm": 0.10880216211080551, "global_step": 30648, "epoch": 289, "lr": 3.7826914827738106e-05} +{"train_loss": 0.0145468944683671, "grad_norm": 0.09258247911930084, "global_step": 30649, "epoch": 289, "lr": 3.782404026069567e-05} +{"train_loss": 0.014767544344067574, "grad_norm": 0.07952605932950974, "global_step": 30650, "epoch": 289, "lr": 3.782116573643427e-05} +{"train_loss": 0.01578586734831333, "grad_norm": 0.0613032728433609, "global_step": 30651, "epoch": 289, "lr": 3.781829125496399e-05} +{"train_loss": 0.014265401288866997, "grad_norm": 0.05081278085708618, "global_step": 30652, "epoch": 289, "lr": 3.7815416816294915e-05} +{"train_loss": 0.016245493665337563, "grad_norm": 0.07810468226671219, "global_step": 30653, "epoch": 289, "lr": 3.781254242043719e-05} +{"train_loss": 0.01469595916569233, "grad_norm": 0.11738462746143341, "global_step": 30654, "epoch": 289, "lr": 3.780966806740088e-05} +{"train_loss": 0.014606854878365993, "grad_norm": 0.06266926974058151, "global_step": 30655, "epoch": 289, "lr": 3.78067937571961e-05} +{"train_loss": 0.016749035567045212, "grad_norm": 0.062346845865249634, "global_step": 30656, "epoch": 289, "lr": 3.780391948983295e-05} +{"train_loss": 0.0167960487306118, "grad_norm": 0.1387333869934082, "global_step": 30657, "epoch": 289, "lr": 3.780104526532151e-05} +{"train_loss": 0.01648404821753502, "grad_norm": 0.07509605586528778, "global_step": 30658, "epoch": 289, "lr": 3.779817108367191e-05} +{"train_loss": 0.015759332105517387, "grad_norm": 0.0672476664185524, "global_step": 30659, "epoch": 289, "lr": 3.7795296944894204e-05} +{"train_loss": 0.01338128000497818, "grad_norm": 0.15675005316734314, "global_step": 30660, "epoch": 289, "lr": 3.779242284899853e-05} +{"train_loss": 0.014999381266534328, "grad_norm": 0.07452542334794998, "global_step": 30661, "epoch": 289, "lr": 3.7789548795994976e-05} +{"train_loss": 0.014939484186470509, "grad_norm": 0.06597364693880081, "global_step": 30662, "epoch": 289, "lr": 3.778667478589365e-05} +{"train_loss": 0.013696190901100636, "grad_norm": 0.06709959357976913, "global_step": 30663, "epoch": 289, "lr": 3.7783800818704624e-05} +{"train_loss": 0.012867191806435585, "grad_norm": 0.05985522270202637, "global_step": 30664, "epoch": 289, "lr": 3.7780926894438e-05} +{"train_loss": 0.017876189202070236, "grad_norm": 0.10112693160772324, "global_step": 30665, "epoch": 289, "lr": 3.777805301310389e-05} +{"train_loss": 0.011651268228888512, "grad_norm": 0.07568991184234619, "global_step": 30666, "epoch": 289, "lr": 3.77751791747124e-05} +{"train_loss": 0.017787747085094452, "grad_norm": 0.05951504781842232, "global_step": 30667, "epoch": 289, "lr": 3.777230537927359e-05} +{"train_loss": 0.017447687685489655, "grad_norm": 0.11033038794994354, "global_step": 30668, "epoch": 289, "lr": 3.776943162679761e-05} +{"train_loss": 0.013460297137498856, "grad_norm": 0.09072712808847427, "global_step": 30669, "epoch": 289, "lr": 3.776655791729451e-05} +{"train_loss": 0.01277926191687584, "grad_norm": 0.07521259039640427, "global_step": 30670, "epoch": 289, "lr": 3.776368425077441e-05} +{"train_loss": 0.01643385924398899, "grad_norm": 0.06280209869146347, "global_step": 30671, "epoch": 289, "lr": 3.7760810627247386e-05} +{"train_loss": 0.017014658078551292, "grad_norm": 0.15491053462028503, "global_step": 30672, "epoch": 289, "lr": 3.775793704672357e-05} +{"train_loss": 0.01647689752280712, "grad_norm": 0.19811207056045532, "global_step": 30673, "epoch": 289, "lr": 3.775506350921304e-05} +{"train_loss": 0.015925895422697067, "grad_norm": 0.08994852006435394, "global_step": 30674, "epoch": 289, "lr": 3.77521900147259e-05} +{"train_loss": 0.01611139625310898, "grad_norm": 0.07381758838891983, "global_step": 30675, "epoch": 289, "lr": 3.774931656327222e-05} +{"train_loss": 0.014772672206163406, "grad_norm": 0.08939643949270248, "global_step": 30676, "epoch": 289, "lr": 3.774644315486211e-05} +{"train_loss": 0.012730402871966362, "grad_norm": 0.11356762051582336, "global_step": 30677, "epoch": 289, "lr": 3.774356978950568e-05} +{"train_loss": 0.015226200222969055, "grad_norm": 0.13053488731384277, "global_step": 30678, "epoch": 289, "lr": 3.774069646721302e-05} +{"train_loss": 0.015198977664113045, "grad_norm": 0.06948510557413101, "global_step": 30679, "epoch": 289, "lr": 3.77378231879942e-05} +{"train_loss": 0.014357030391693115, "grad_norm": 0.0672576054930687, "global_step": 30680, "epoch": 289, "lr": 3.773494995185937e-05} +{"train_loss": 0.01737711764872074, "grad_norm": 0.10824138671159744, "global_step": 30681, "epoch": 289, "lr": 3.773207675881857e-05} +{"train_loss": 0.014445813372731209, "grad_norm": 0.10550323873758316, "global_step": 30682, "epoch": 289, "lr": 3.772920360888192e-05} +{"train_loss": 0.015193640254437923, "grad_norm": 0.07546082884073257, "global_step": 30683, "epoch": 289, "lr": 3.7726330502059515e-05} +{"train_loss": 0.01321918424218893, "grad_norm": 0.1450958251953125, "global_step": 30684, "epoch": 289, "lr": 3.772345743836143e-05} +{"train_loss": 0.01495132315903902, "grad_norm": 0.09465712308883667, "global_step": 30685, "epoch": 289, "lr": 3.77205844177978e-05} +{"train_loss": 0.01294559333473444, "grad_norm": 0.06512248516082764, "global_step": 30686, "epoch": 289, "lr": 3.7717711440378694e-05} +{"train_loss": 0.012769234366714954, "grad_norm": 0.051708657294511795, "global_step": 30687, "epoch": 289, "lr": 3.7714838506114204e-05} +{"train_loss": 0.015344489365816116, "grad_norm": 0.10738690197467804, "global_step": 30688, "epoch": 289, "lr": 3.771196561501442e-05} +{"train_loss": 0.017879506573081017, "grad_norm": 0.18069280683994293, "global_step": 30689, "epoch": 289, "lr": 3.770909276708945e-05} +{"train_loss": 0.015667598694562912, "grad_norm": 0.08202395588159561, "global_step": 30690, "epoch": 289, "lr": 3.770621996234939e-05} +{"train_loss": 0.01624641939997673, "grad_norm": 0.08746057003736496, "global_step": 30691, "epoch": 289, "lr": 3.770334720080431e-05} +{"train_loss": 0.0162237286567688, "grad_norm": 0.10184401273727417, "global_step": 30692, "epoch": 289, "lr": 3.770047448246435e-05} +{"train_loss": 0.016298729926347733, "grad_norm": 0.07074142247438431, "global_step": 30693, "epoch": 289, "lr": 3.7697601807339546e-05} +{"train_loss": 0.015028966590762138, "grad_norm": 0.1474902629852295, "global_step": 30694, "epoch": 289, "lr": 3.769472917544003e-05} +{"train_loss": 0.016264764592051506, "grad_norm": 0.11008653044700623, "global_step": 30695, "epoch": 289, "lr": 3.769185658677588e-05} +{"train_loss": 0.015814730897545815, "grad_norm": 0.07926152646541595, "global_step": 30696, "epoch": 289, "lr": 3.768898404135719e-05} +{"train_loss": 0.015623290091753006, "grad_norm": 0.079513780772686, "global_step": 30697, "epoch": 289, "lr": 3.768611153919407e-05} +{"train_loss": 0.014235001057386398, "grad_norm": 0.07134352624416351, "global_step": 30698, "epoch": 289, "lr": 3.7683239080296594e-05} +{"train_loss": 0.01716604083776474, "grad_norm": 0.06330233812332153, "global_step": 30699, "epoch": 289, "lr": 3.768036666467486e-05} +{"train_loss": 0.019264522939920425, "grad_norm": 0.09911483526229858, "global_step": 30700, "epoch": 289, "lr": 3.7677494292338944e-05} +{"train_loss": 0.018023991957306862, "grad_norm": 0.12969335913658142, "global_step": 30701, "epoch": 289, "lr": 3.7674621963298964e-05} +{"train_loss": 0.013971629552543163, "grad_norm": 0.062411725521087646, "global_step": 30702, "epoch": 289, "lr": 3.7671749677565014e-05} +{"train_loss": 0.01562627963721752, "grad_norm": 0.0973663255572319, "global_step": 30703, "epoch": 289, "lr": 3.766887743514717e-05} +{"train_loss": 0.01693752408027649, "grad_norm": 0.10898235440254211, "global_step": 30704, "epoch": 289, "lr": 3.766600523605552e-05} +{"train_loss": 0.014200661331415176, "grad_norm": 0.08053335547447205, "global_step": 30705, "epoch": 289, "lr": 3.766313308030016e-05} +{"train_loss": 0.018499594181776047, "grad_norm": 0.08629085123538971, "global_step": 30706, "epoch": 289, "lr": 3.766026096789119e-05} +{"train_loss": 0.01582149788737297, "grad_norm": 0.1464691013097763, "global_step": 30707, "epoch": 289, "lr": 3.7657388898838706e-05} +{"train_loss": 0.01745551824569702, "grad_norm": 0.11289829015731812, "global_step": 30708, "epoch": 289, "lr": 3.765451687315277e-05} +{"train_loss": 0.014414781704545021, "grad_norm": 0.08356571942567825, "global_step": 30709, "epoch": 289, "lr": 3.765164489084353e-05} +{"train_loss": 0.014906824566423893, "grad_norm": 0.05174483731389046, "global_step": 30710, "epoch": 289, "lr": 3.7648772951921004e-05} +{"train_loss": 0.014261915348470211, "grad_norm": 0.13870346546173096, "global_step": 30711, "epoch": 289, "lr": 3.7645901056395326e-05} +{"train_loss": 0.013979856856167316, "grad_norm": 0.08456799387931824, "global_step": 30712, "epoch": 289, "lr": 3.7643029204276576e-05} +{"train_loss": 0.01597924903035164, "grad_norm": 0.05356702208518982, "global_step": 30713, "epoch": 289, "lr": 3.764015739557486e-05} +{"train_loss": 0.01657918281853199, "grad_norm": 0.05655467510223389, "global_step": 30714, "epoch": 289, "lr": 3.763728563030025e-05} +{"train_loss": 0.016932103782892227, "grad_norm": 0.06315143406391144, "global_step": 30715, "epoch": 289, "lr": 3.763441390846285e-05} +{"train_loss": 0.01617567427456379, "grad_norm": 0.16465887427330017, "global_step": 30716, "epoch": 289, "lr": 3.763154223007273e-05} +{"train_loss": 0.015373519621789455, "grad_norm": 0.04977784305810928, "global_step": 30717, "epoch": 289, "lr": 3.762867059513998e-05} +{"train_loss": 0.012993857264518738, "grad_norm": 0.08328745514154434, "global_step": 30718, "epoch": 289, "lr": 3.762579900367472e-05} +{"train_loss": 0.015596969053149223, "grad_norm": 0.08083601295948029, "global_step": 30719, "epoch": 289, "lr": 3.762292745568702e-05} +{"train_loss": 0.014793947339057922, "grad_norm": 0.08039243519306183, "global_step": 30720, "epoch": 289, "lr": 3.762005595118696e-05} +{"train_loss": 0.014623226597905159, "grad_norm": 0.08072849363088608, "global_step": 30721, "epoch": 289, "lr": 3.761718449018467e-05} +{"train_loss": 0.012758498080074787, "grad_norm": 0.08515442162752151, "global_step": 30722, "epoch": 289, "lr": 3.761431307269017e-05} +{"train_loss": 0.012809672392904758, "grad_norm": 0.09130574017763138, "global_step": 30723, "epoch": 289, "lr": 3.7611441698713605e-05} +{"train_loss": 0.013986409641802311, "grad_norm": 0.09597300738096237, "global_step": 30724, "epoch": 289, "lr": 3.7608570368265045e-05} +{"train_loss": 0.014322311617434025, "grad_norm": 0.10272006690502167, "global_step": 30725, "epoch": 289, "lr": 3.7605699081354576e-05} +{"train_loss": 0.01672990620136261, "grad_norm": 0.08461250364780426, "global_step": 30726, "epoch": 289, "lr": 3.7602827837992286e-05} +{"train_loss": 0.012912913225591183, "grad_norm": 0.0838824063539505, "global_step": 30727, "epoch": 289, "lr": 3.7599956638188286e-05} +{"train_loss": 0.016474206000566483, "grad_norm": 0.07311540096998215, "global_step": 30728, "epoch": 289, "lr": 3.759708548195264e-05} +{"train_loss": 0.014791742898523808, "grad_norm": 0.1468837857246399, "global_step": 30729, "epoch": 289, "lr": 3.759421436929542e-05} +{"train_loss": 0.014618019573390484, "grad_norm": 0.18885912001132965, "global_step": 30730, "epoch": 289, "lr": 3.7591343300226754e-05} +{"train_loss": 0.015957003459334373, "grad_norm": 0.09235692769289017, "global_step": 30731, "epoch": 289, "lr": 3.7588472274756717e-05} +{"train_loss": 0.013553819619119167, "grad_norm": 0.1530807763338089, "global_step": 30732, "epoch": 289, "lr": 3.7585601292895364e-05} +{"train_loss": 0.016146957874298096, "grad_norm": 0.10632364451885223, "global_step": 30733, "epoch": 289, "lr": 3.7582730354652855e-05} +{"train_loss": 0.014698345214128494, "grad_norm": 0.06763458251953125, "global_step": 30734, "epoch": 289, "lr": 3.7579859460039196e-05} +{"train_loss": 0.013150429353117943, "grad_norm": 0.06173410639166832, "global_step": 30735, "epoch": 289, "lr": 3.757698860906452e-05} +{"train_loss": 0.01565711200237274, "grad_norm": 0.05723753571510315, "global_step": 30736, "epoch": 289, "lr": 3.75741178017389e-05} +{"train_loss": 0.015275951474905014, "grad_norm": 0.15037377178668976, "global_step": 30737, "epoch": 289, "lr": 3.7571247038072415e-05} +{"train_loss": 0.016803251579403877, "grad_norm": 0.07910627126693726, "global_step": 30738, "epoch": 289, "lr": 3.7568376318075175e-05} +{"train_loss": 0.015350051364808713, "grad_norm": 0.22091859579086304, "global_step": 30739, "epoch": 289, "lr": 3.756550564175727e-05, "val_loss": 0.01937817595899105} +{"train_loss": 0.01704672910273075, "grad_norm": 0.07080140709877014, "global_step": 30740, "epoch": 290, "lr": 3.756263500912875e-05} +{"train_loss": 0.015625137835741043, "grad_norm": 0.0832517221570015, "global_step": 30741, "epoch": 290, "lr": 3.7559764420199724e-05} +{"train_loss": 0.015407590195536613, "grad_norm": 0.10875972360372543, "global_step": 30742, "epoch": 290, "lr": 3.755689387498028e-05} +{"train_loss": 0.013200949877500534, "grad_norm": 0.13305586576461792, "global_step": 30743, "epoch": 290, "lr": 3.75540233734805e-05} +{"train_loss": 0.0197907742112875, "grad_norm": 0.15954606235027313, "global_step": 30744, "epoch": 290, "lr": 3.7551152915710456e-05} +{"train_loss": 0.015025397762656212, "grad_norm": 0.07752566784620285, "global_step": 30745, "epoch": 290, "lr": 3.754828250168028e-05} +{"train_loss": 0.015336922369897366, "grad_norm": 0.10548102855682373, "global_step": 30746, "epoch": 290, "lr": 3.754541213139999e-05} +{"train_loss": 0.01601332426071167, "grad_norm": 0.07979001104831696, "global_step": 30747, "epoch": 290, "lr": 3.754254180487972e-05} +{"train_loss": 0.01687948778271675, "grad_norm": 0.09147539734840393, "global_step": 30748, "epoch": 290, "lr": 3.7539671522129536e-05} +{"train_loss": 0.016158053651452065, "grad_norm": 0.08398675918579102, "global_step": 30749, "epoch": 290, "lr": 3.753680128315952e-05} +{"train_loss": 0.013754110783338547, "grad_norm": 0.08311989158391953, "global_step": 30750, "epoch": 290, "lr": 3.75339310879798e-05} +{"train_loss": 0.01448570191860199, "grad_norm": 0.07241616398096085, "global_step": 30751, "epoch": 290, "lr": 3.753106093660039e-05} +{"train_loss": 0.01430012471973896, "grad_norm": 0.057777442038059235, "global_step": 30752, "epoch": 290, "lr": 3.752819082903142e-05} +{"train_loss": 0.013906929641962051, "grad_norm": 0.06765292584896088, "global_step": 30753, "epoch": 290, "lr": 3.7525320765282964e-05} +{"train_loss": 0.014462693594396114, "grad_norm": 0.08801300078630447, "global_step": 30754, "epoch": 290, "lr": 3.752245074536509e-05} +{"train_loss": 0.016478944569826126, "grad_norm": 0.09374632686376572, "global_step": 30755, "epoch": 290, "lr": 3.751958076928791e-05} +{"train_loss": 0.015490042977035046, "grad_norm": 0.06712459772825241, "global_step": 30756, "epoch": 290, "lr": 3.7516710837061506e-05} +{"train_loss": 0.016593240201473236, "grad_norm": 0.07822316139936447, "global_step": 30757, "epoch": 290, "lr": 3.7513840948695945e-05} +{"train_loss": 0.016511429101228714, "grad_norm": 0.15338601171970367, "global_step": 30758, "epoch": 290, "lr": 3.75109711042013e-05} +{"train_loss": 0.015987172722816467, "grad_norm": 0.17254400253295898, "global_step": 30759, "epoch": 290, "lr": 3.750810130358769e-05} +{"train_loss": 0.01474788412451744, "grad_norm": 0.08470684289932251, "global_step": 30760, "epoch": 290, "lr": 3.750523154686518e-05} +{"train_loss": 0.014110717922449112, "grad_norm": 0.0893118605017662, "global_step": 30761, "epoch": 290, "lr": 3.750236183404383e-05} +{"train_loss": 0.016373973339796066, "grad_norm": 0.1110464334487915, "global_step": 30762, "epoch": 290, "lr": 3.749949216513379e-05} +{"train_loss": 0.01437730249017477, "grad_norm": 0.05976930633187294, "global_step": 30763, "epoch": 290, "lr": 3.749662254014505e-05} +{"train_loss": 0.013724472373723984, "grad_norm": 0.07913488149642944, "global_step": 30764, "epoch": 290, "lr": 3.749375295908777e-05} +{"train_loss": 0.017000311985611916, "grad_norm": 0.10609501600265503, "global_step": 30765, "epoch": 290, "lr": 3.7490883421972e-05} +{"train_loss": 0.01677507348358631, "grad_norm": 0.07121716439723969, "global_step": 30766, "epoch": 290, "lr": 3.7488013928807805e-05} +{"train_loss": 0.014605256728827953, "grad_norm": 0.12412336468696594, "global_step": 30767, "epoch": 290, "lr": 3.7485144479605306e-05} +{"train_loss": 0.01708918809890747, "grad_norm": 0.11758135259151459, "global_step": 30768, "epoch": 290, "lr": 3.748227507437457e-05} +{"train_loss": 0.014733269810676575, "grad_norm": 0.08359923958778381, "global_step": 30769, "epoch": 290, "lr": 3.747940571312567e-05} +{"train_loss": 0.016169637441635132, "grad_norm": 0.06420457363128662, "global_step": 30770, "epoch": 290, "lr": 3.7476536395868686e-05} +{"train_loss": 0.015168026089668274, "grad_norm": 0.0957827940583229, "global_step": 30771, "epoch": 290, "lr": 3.7473667122613714e-05} +{"train_loss": 0.014121163636446, "grad_norm": 0.07043880224227905, "global_step": 30772, "epoch": 290, "lr": 3.747079789337083e-05} +{"train_loss": 0.015497511252760887, "grad_norm": 0.05786805599927902, "global_step": 30773, "epoch": 290, "lr": 3.74679287081501e-05} +{"train_loss": 0.015792936086654663, "grad_norm": 0.07220129668712616, "global_step": 30774, "epoch": 290, "lr": 3.7465059566961655e-05} +{"train_loss": 0.01397356390953064, "grad_norm": 0.14742113649845123, "global_step": 30775, "epoch": 290, "lr": 3.74621904698155e-05} +{"train_loss": 0.017128966748714447, "grad_norm": 0.19340792298316956, "global_step": 30776, "epoch": 290, "lr": 3.745932141672177e-05} +{"train_loss": 0.014599497430026531, "grad_norm": 0.10931246727705002, "global_step": 30777, "epoch": 290, "lr": 3.7456452407690535e-05} +{"train_loss": 0.014602060429751873, "grad_norm": 0.08243770897388458, "global_step": 30778, "epoch": 290, "lr": 3.745358344273185e-05} +{"train_loss": 0.012933019548654556, "grad_norm": 0.10984766483306885, "global_step": 30779, "epoch": 290, "lr": 3.7450714521855835e-05} +{"train_loss": 0.015310186892747879, "grad_norm": 0.13440659642219543, "global_step": 30780, "epoch": 290, "lr": 3.744784564507256e-05} +{"train_loss": 0.015025840140879154, "grad_norm": 0.07533707469701767, "global_step": 30781, "epoch": 290, "lr": 3.7444976812392084e-05} +{"train_loss": 0.015062268823385239, "grad_norm": 0.07375263422727585, "global_step": 30782, "epoch": 290, "lr": 3.744210802382451e-05} +{"train_loss": 0.01667414791882038, "grad_norm": 0.09440402686595917, "global_step": 30783, "epoch": 290, "lr": 3.7439239279379885e-05} +{"train_loss": 0.016300734132528305, "grad_norm": 0.05355270579457283, "global_step": 30784, "epoch": 290, "lr": 3.7436370579068326e-05} +{"train_loss": 0.014186955988407135, "grad_norm": 0.052965398877859116, "global_step": 30785, "epoch": 290, "lr": 3.7433501922899895e-05} +{"train_loss": 0.014519779942929745, "grad_norm": 0.07244841009378433, "global_step": 30786, "epoch": 290, "lr": 3.7430633310884684e-05} +{"train_loss": 0.014896928332746029, "grad_norm": 0.05874544009566307, "global_step": 30787, "epoch": 290, "lr": 3.742776474303274e-05} +{"train_loss": 0.01712696999311447, "grad_norm": 0.1439201980829239, "global_step": 30788, "epoch": 290, "lr": 3.7424896219354175e-05} +{"train_loss": 0.018384311348199844, "grad_norm": 0.07978043705224991, "global_step": 30789, "epoch": 290, "lr": 3.742202773985906e-05} +{"train_loss": 0.01345946453511715, "grad_norm": 0.06318294256925583, "global_step": 30790, "epoch": 290, "lr": 3.741915930455745e-05} +{"train_loss": 0.01637430489063263, "grad_norm": 0.21129260957241058, "global_step": 30791, "epoch": 290, "lr": 3.741629091345945e-05} +{"train_loss": 0.018005065619945526, "grad_norm": 0.06703510880470276, "global_step": 30792, "epoch": 290, "lr": 3.741342256657515e-05} +{"train_loss": 0.016116399317979813, "grad_norm": 0.07083860784769058, "global_step": 30793, "epoch": 290, "lr": 3.74105542639146e-05} +{"train_loss": 0.015913916751742363, "grad_norm": 0.09100301563739777, "global_step": 30794, "epoch": 290, "lr": 3.7407686005487885e-05} +{"train_loss": 0.014903770759701729, "grad_norm": 0.0542755052447319, "global_step": 30795, "epoch": 290, "lr": 3.740481779130508e-05} +{"train_loss": 0.014937454834580421, "grad_norm": 0.06186676025390625, "global_step": 30796, "epoch": 290, "lr": 3.740194962137627e-05} +{"train_loss": 0.014929751865565777, "grad_norm": 0.08234784007072449, "global_step": 30797, "epoch": 290, "lr": 3.739908149571154e-05} +{"train_loss": 0.015717793256044388, "grad_norm": 0.07341238111257553, "global_step": 30798, "epoch": 290, "lr": 3.739621341432096e-05} +{"train_loss": 0.01674179546535015, "grad_norm": 0.052405085414648056, "global_step": 30799, "epoch": 290, "lr": 3.7393345377214586e-05} +{"train_loss": 0.0144197978079319, "grad_norm": 0.050544023513793945, "global_step": 30800, "epoch": 290, "lr": 3.739047738440252e-05} +{"train_loss": 0.016026359051465988, "grad_norm": 0.06214594468474388, "global_step": 30801, "epoch": 290, "lr": 3.7387609435894836e-05} +{"train_loss": 0.014326559379696846, "grad_norm": 0.08691409230232239, "global_step": 30802, "epoch": 290, "lr": 3.7384741531701595e-05} +{"train_loss": 0.01253008283674717, "grad_norm": 0.08286117762327194, "global_step": 30803, "epoch": 290, "lr": 3.7381873671832925e-05} +{"train_loss": 0.016449853777885437, "grad_norm": 0.05514468252658844, "global_step": 30804, "epoch": 290, "lr": 3.7379005856298826e-05} +{"train_loss": 0.013270258903503418, "grad_norm": 0.0661369040608406, "global_step": 30805, "epoch": 290, "lr": 3.7376138085109423e-05} +{"train_loss": 0.014326822943985462, "grad_norm": 0.1202007457613945, "global_step": 30806, "epoch": 290, "lr": 3.737327035827478e-05} +{"train_loss": 0.01606375351548195, "grad_norm": 0.11030758172273636, "global_step": 30807, "epoch": 290, "lr": 3.737040267580497e-05} +{"train_loss": 0.01606784388422966, "grad_norm": 0.13309943675994873, "global_step": 30808, "epoch": 290, "lr": 3.736753503771007e-05} +{"train_loss": 0.016232166439294815, "grad_norm": 0.0970396101474762, "global_step": 30809, "epoch": 290, "lr": 3.736466744400018e-05} +{"train_loss": 0.014224248006939888, "grad_norm": 0.05461222305893898, "global_step": 30810, "epoch": 290, "lr": 3.7361799894685336e-05} +{"train_loss": 0.014894211664795876, "grad_norm": 0.06980263441801071, "global_step": 30811, "epoch": 290, "lr": 3.7358932389775634e-05} +{"train_loss": 0.013886015862226486, "grad_norm": 0.09545323997735977, "global_step": 30812, "epoch": 290, "lr": 3.7356064929281135e-05} +{"train_loss": 0.013691699132323265, "grad_norm": 0.06524693965911865, "global_step": 30813, "epoch": 290, "lr": 3.735319751321194e-05} +{"train_loss": 0.015016920864582062, "grad_norm": 0.08964872360229492, "global_step": 30814, "epoch": 290, "lr": 3.7350330141578104e-05} +{"train_loss": 0.015516810119152069, "grad_norm": 0.0646577924489975, "global_step": 30815, "epoch": 290, "lr": 3.7347462814389724e-05} +{"train_loss": 0.015489507466554642, "grad_norm": 0.06236890330910683, "global_step": 30816, "epoch": 290, "lr": 3.7344595531656826e-05} +{"train_loss": 0.013966823928058147, "grad_norm": 0.07929373532533646, "global_step": 30817, "epoch": 290, "lr": 3.734172829338953e-05} +{"train_loss": 0.015175491571426392, "grad_norm": 0.0801941305398941, "global_step": 30818, "epoch": 290, "lr": 3.73388610995979e-05} +{"train_loss": 0.018953539431095123, "grad_norm": 0.11148229241371155, "global_step": 30819, "epoch": 290, "lr": 3.733599395029199e-05} +{"train_loss": 0.014894930645823479, "grad_norm": 0.1411440074443817, "global_step": 30820, "epoch": 290, "lr": 3.73331268454819e-05} +{"train_loss": 0.01520636584609747, "grad_norm": 0.10429004579782486, "global_step": 30821, "epoch": 290, "lr": 3.7330259785177714e-05} +{"train_loss": 0.014226438477635384, "grad_norm": 0.0971694067120552, "global_step": 30822, "epoch": 290, "lr": 3.7327392769389465e-05} +{"train_loss": 0.016176313161849976, "grad_norm": 0.11865464597940445, "global_step": 30823, "epoch": 290, "lr": 3.7324525798127243e-05} +{"train_loss": 0.013792285695672035, "grad_norm": 0.12230947613716125, "global_step": 30824, "epoch": 290, "lr": 3.732165887140112e-05} +{"train_loss": 0.015028806403279305, "grad_norm": 0.12794806063175201, "global_step": 30825, "epoch": 290, "lr": 3.731879198922119e-05} +{"train_loss": 0.01464131660759449, "grad_norm": 0.08472421765327454, "global_step": 30826, "epoch": 290, "lr": 3.73159251515975e-05} +{"train_loss": 0.013763810507953167, "grad_norm": 0.057842906564474106, "global_step": 30827, "epoch": 290, "lr": 3.7313058358540155e-05} +{"train_loss": 0.01353885605931282, "grad_norm": 0.047572050243616104, "global_step": 30828, "epoch": 290, "lr": 3.731019161005917e-05} +{"train_loss": 0.015418311581015587, "grad_norm": 0.08985146135091782, "global_step": 30829, "epoch": 290, "lr": 3.7307324906164676e-05} +{"train_loss": 0.015176694840192795, "grad_norm": 0.10526290535926819, "global_step": 30830, "epoch": 290, "lr": 3.7304458246866716e-05} +{"train_loss": 0.01802830398082733, "grad_norm": 0.07564181834459305, "global_step": 30831, "epoch": 290, "lr": 3.730159163217536e-05} +{"train_loss": 0.014606425538659096, "grad_norm": 0.08891265094280243, "global_step": 30832, "epoch": 290, "lr": 3.7298725062100706e-05} +{"train_loss": 0.01895744539797306, "grad_norm": 0.1004820168018341, "global_step": 30833, "epoch": 290, "lr": 3.72958585366528e-05} +{"train_loss": 0.01647178828716278, "grad_norm": 0.09537869691848755, "global_step": 30834, "epoch": 290, "lr": 3.729299205584173e-05} +{"train_loss": 0.015050822868943214, "grad_norm": 0.08570291101932526, "global_step": 30835, "epoch": 290, "lr": 3.729012561967755e-05} +{"train_loss": 0.015037324279546738, "grad_norm": 0.07938927412033081, "global_step": 30836, "epoch": 290, "lr": 3.728725922817033e-05} +{"train_loss": 0.01616022363305092, "grad_norm": 0.10763475298881531, "global_step": 30837, "epoch": 290, "lr": 3.7284392881330175e-05} +{"train_loss": 0.01536354050040245, "grad_norm": 0.08282812684774399, "global_step": 30838, "epoch": 290, "lr": 3.728152657916712e-05} +{"train_loss": 0.015257944352924824, "grad_norm": 0.11544790863990784, "global_step": 30839, "epoch": 290, "lr": 3.727866032169127e-05} +{"train_loss": 0.016562599688768387, "grad_norm": 0.08099313080310822, "global_step": 30840, "epoch": 290, "lr": 3.7275794108912665e-05} +{"train_loss": 0.014799179509282112, "grad_norm": 0.11146266758441925, "global_step": 30841, "epoch": 290, "lr": 3.7272927940841366e-05} +{"train_loss": 0.013476479798555374, "grad_norm": 0.08293202519416809, "global_step": 30842, "epoch": 290, "lr": 3.727006181748748e-05} +{"train_loss": 0.015361562371253967, "grad_norm": 0.059911906719207764, "global_step": 30843, "epoch": 290, "lr": 3.726719573886105e-05} +{"train_loss": 0.014620481058955193, "grad_norm": 0.06760440766811371, "global_step": 30844, "epoch": 290, "lr": 3.726432970497217e-05} +{"train_loss": 0.015344377405027736, "grad_norm": 0.07739457488059998, "global_step": 30845, "epoch": 290, "lr": 3.72614637158309e-05, "train_batch_mse_error": 0.0136458994820714, "val_loss": 0.01907583698630333} +{"train_loss": 0.016712533310055733, "grad_norm": 0.07966725528240204, "global_step": 30846, "epoch": 291, "lr": 3.72585977714473e-05} +{"train_loss": 0.017238814383745193, "grad_norm": 0.09292349219322205, "global_step": 30847, "epoch": 291, "lr": 3.7255731871831444e-05} +{"train_loss": 0.01355006918311119, "grad_norm": 0.09759806841611862, "global_step": 30848, "epoch": 291, "lr": 3.7252866016993396e-05} +{"train_loss": 0.015241069719195366, "grad_norm": 0.14397771656513214, "global_step": 30849, "epoch": 291, "lr": 3.7250000206943245e-05} +{"train_loss": 0.014602337963879108, "grad_norm": 0.17807459831237793, "global_step": 30850, "epoch": 291, "lr": 3.7247134441691054e-05} +{"train_loss": 0.015810195356607437, "grad_norm": 0.09010593593120575, "global_step": 30851, "epoch": 291, "lr": 3.724426872124688e-05} +{"train_loss": 0.013182701542973518, "grad_norm": 0.10360808670520782, "global_step": 30852, "epoch": 291, "lr": 3.7241403045620795e-05} +{"train_loss": 0.014437905512750149, "grad_norm": 0.0713430643081665, "global_step": 30853, "epoch": 291, "lr": 3.723853741482286e-05} +{"train_loss": 0.01654607243835926, "grad_norm": 0.10716353356838226, "global_step": 30854, "epoch": 291, "lr": 3.723567182886317e-05} +{"train_loss": 0.016234563663601875, "grad_norm": 0.06148093566298485, "global_step": 30855, "epoch": 291, "lr": 3.723280628775177e-05} +{"train_loss": 0.015609647147357464, "grad_norm": 0.13859446346759796, "global_step": 30856, "epoch": 291, "lr": 3.722994079149876e-05} +{"train_loss": 0.0175882950425148, "grad_norm": 0.09406888484954834, "global_step": 30857, "epoch": 291, "lr": 3.722707534011416e-05} +{"train_loss": 0.014918549917638302, "grad_norm": 0.05927456170320511, "global_step": 30858, "epoch": 291, "lr": 3.722420993360807e-05} +{"train_loss": 0.018669363111257553, "grad_norm": 0.07061483711004257, "global_step": 30859, "epoch": 291, "lr": 3.7221344571990544e-05} +{"train_loss": 0.014659011736512184, "grad_norm": 0.06089919060468674, "global_step": 30860, "epoch": 291, "lr": 3.721847925527165e-05} +{"train_loss": 0.012279007583856583, "grad_norm": 0.06789854168891907, "global_step": 30861, "epoch": 291, "lr": 3.721561398346148e-05} +{"train_loss": 0.013544410467147827, "grad_norm": 0.057984743267297745, "global_step": 30862, "epoch": 291, "lr": 3.721274875657008e-05} +{"train_loss": 0.017504140734672546, "grad_norm": 0.21567334234714508, "global_step": 30863, "epoch": 291, "lr": 3.720988357460752e-05} +{"train_loss": 0.013798200525343418, "grad_norm": 0.09207390248775482, "global_step": 30864, "epoch": 291, "lr": 3.720701843758386e-05} +{"train_loss": 0.011089242994785309, "grad_norm": 0.09243661910295486, "global_step": 30865, "epoch": 291, "lr": 3.720415334550917e-05} +{"train_loss": 0.012992607429623604, "grad_norm": 0.09052816033363342, "global_step": 30866, "epoch": 291, "lr": 3.7201288298393525e-05} +{"train_loss": 0.0130958566442132, "grad_norm": 0.1189667358994484, "global_step": 30867, "epoch": 291, "lr": 3.7198423296246984e-05} +{"train_loss": 0.012157846242189407, "grad_norm": 0.05233690142631531, "global_step": 30868, "epoch": 291, "lr": 3.719555833907964e-05} +{"train_loss": 0.015376482158899307, "grad_norm": 0.05710341036319733, "global_step": 30869, "epoch": 291, "lr": 3.71926934269015e-05} +{"train_loss": 0.013801638036966324, "grad_norm": 0.06671755015850067, "global_step": 30870, "epoch": 291, "lr": 3.718982855972268e-05} +{"train_loss": 0.014759980142116547, "grad_norm": 0.0686163455247879, "global_step": 30871, "epoch": 291, "lr": 3.7186963737553234e-05} +{"train_loss": 0.014994597062468529, "grad_norm": 0.15952786803245544, "global_step": 30872, "epoch": 291, "lr": 3.718409896040322e-05} +{"train_loss": 0.015053227543830872, "grad_norm": 0.073920339345932, "global_step": 30873, "epoch": 291, "lr": 3.718123422828271e-05} +{"train_loss": 0.015183836221694946, "grad_norm": 0.07127177715301514, "global_step": 30874, "epoch": 291, "lr": 3.7178369541201786e-05} +{"train_loss": 0.017609331756830215, "grad_norm": 0.11516369879245758, "global_step": 30875, "epoch": 291, "lr": 3.7175504899170476e-05} +{"train_loss": 0.013740625232458115, "grad_norm": 0.06655000895261765, "global_step": 30876, "epoch": 291, "lr": 3.717264030219887e-05} +{"train_loss": 0.017727574333548546, "grad_norm": 0.07056286931037903, "global_step": 30877, "epoch": 291, "lr": 3.716977575029702e-05} +{"train_loss": 0.016154373064637184, "grad_norm": 0.09398499876260757, "global_step": 30878, "epoch": 291, "lr": 3.7166911243475e-05} +{"train_loss": 0.013804754242300987, "grad_norm": 0.0709141194820404, "global_step": 30879, "epoch": 291, "lr": 3.716404678174288e-05} +{"train_loss": 0.01642385870218277, "grad_norm": 0.09073985368013382, "global_step": 30880, "epoch": 291, "lr": 3.716118236511072e-05} +{"train_loss": 0.015902772545814514, "grad_norm": 0.13675643503665924, "global_step": 30881, "epoch": 291, "lr": 3.7158317993588586e-05} +{"train_loss": 0.014633591286838055, "grad_norm": 0.06218444183468819, "global_step": 30882, "epoch": 291, "lr": 3.715545366718652e-05} +{"train_loss": 0.014804256148636341, "grad_norm": 0.12920033931732178, "global_step": 30883, "epoch": 291, "lr": 3.715258938591461e-05} +{"train_loss": 0.015354844741523266, "grad_norm": 0.1019260361790657, "global_step": 30884, "epoch": 291, "lr": 3.714972514978292e-05} +{"train_loss": 0.014548428356647491, "grad_norm": 0.140242338180542, "global_step": 30885, "epoch": 291, "lr": 3.714686095880149e-05} +{"train_loss": 0.016427520662546158, "grad_norm": 0.06920216977596283, "global_step": 30886, "epoch": 291, "lr": 3.714399681298043e-05} +{"train_loss": 0.014326723292469978, "grad_norm": 0.06109372526407242, "global_step": 30887, "epoch": 291, "lr": 3.7141132712329755e-05} +{"train_loss": 0.017869072034955025, "grad_norm": 0.07707774639129639, "global_step": 30888, "epoch": 291, "lr": 3.713826865685955e-05} +{"train_loss": 0.01354554109275341, "grad_norm": 0.06790819764137268, "global_step": 30889, "epoch": 291, "lr": 3.713540464657987e-05} +{"train_loss": 0.015623442828655243, "grad_norm": 0.11104229092597961, "global_step": 30890, "epoch": 291, "lr": 3.7132540681500795e-05} +{"train_loss": 0.014592812396585941, "grad_norm": 0.09235649555921555, "global_step": 30891, "epoch": 291, "lr": 3.712967676163237e-05} +{"train_loss": 0.014601478353142738, "grad_norm": 0.04688553512096405, "global_step": 30892, "epoch": 291, "lr": 3.712681288698468e-05} +{"train_loss": 0.012489814311265945, "grad_norm": 0.06089072674512863, "global_step": 30893, "epoch": 291, "lr": 3.7123949057567753e-05} +{"train_loss": 0.013697014190256596, "grad_norm": 0.10153691470623016, "global_step": 30894, "epoch": 291, "lr": 3.712108527339166e-05} +{"train_loss": 0.016383284702897072, "grad_norm": 0.056490086019039154, "global_step": 30895, "epoch": 291, "lr": 3.7118221534466484e-05} +{"train_loss": 0.016223886981606483, "grad_norm": 0.09968318045139313, "global_step": 30896, "epoch": 291, "lr": 3.711535784080228e-05} +{"train_loss": 0.013479864224791527, "grad_norm": 0.05457542464137077, "global_step": 30897, "epoch": 291, "lr": 3.711249419240909e-05} +{"train_loss": 0.01636466756463051, "grad_norm": 0.14390547573566437, "global_step": 30898, "epoch": 291, "lr": 3.710963058929701e-05} +{"train_loss": 0.015431814827024937, "grad_norm": 0.06288288533687592, "global_step": 30899, "epoch": 291, "lr": 3.7106767031476075e-05} +{"train_loss": 0.01583034172654152, "grad_norm": 0.052701935172080994, "global_step": 30900, "epoch": 291, "lr": 3.7103903518956355e-05} +{"train_loss": 0.015522553585469723, "grad_norm": 0.09980925917625427, "global_step": 30901, "epoch": 291, "lr": 3.71010400517479e-05} +{"train_loss": 0.01610862836241722, "grad_norm": 0.08058613538742065, "global_step": 30902, "epoch": 291, "lr": 3.7098176629860785e-05} +{"train_loss": 0.015981271862983704, "grad_norm": 0.06964249163866043, "global_step": 30903, "epoch": 291, "lr": 3.709531325330509e-05} +{"train_loss": 0.015226999297738075, "grad_norm": 0.060533516108989716, "global_step": 30904, "epoch": 291, "lr": 3.7092449922090825e-05} +{"train_loss": 0.015356197953224182, "grad_norm": 0.07063388079404831, "global_step": 30905, "epoch": 291, "lr": 3.708958663622808e-05} +{"train_loss": 0.014975817874073982, "grad_norm": 0.05892334505915642, "global_step": 30906, "epoch": 291, "lr": 3.7086723395726917e-05} +{"train_loss": 0.01320680696517229, "grad_norm": 0.07889900356531143, "global_step": 30907, "epoch": 291, "lr": 3.708386020059739e-05} +{"train_loss": 0.01710999757051468, "grad_norm": 0.10675305873155594, "global_step": 30908, "epoch": 291, "lr": 3.708099705084956e-05} +{"train_loss": 0.015357645228505135, "grad_norm": 0.09255351126194, "global_step": 30909, "epoch": 291, "lr": 3.707813394649351e-05} +{"train_loss": 0.015311850234866142, "grad_norm": 0.06955961138010025, "global_step": 30910, "epoch": 291, "lr": 3.707527088753926e-05} +{"train_loss": 0.01751193404197693, "grad_norm": 0.09037549048662186, "global_step": 30911, "epoch": 291, "lr": 3.707240787399687e-05} +{"train_loss": 0.015165550634264946, "grad_norm": 0.08216744661331177, "global_step": 30912, "epoch": 291, "lr": 3.706954490587644e-05} +{"train_loss": 0.014936480671167374, "grad_norm": 0.08213943988084793, "global_step": 30913, "epoch": 291, "lr": 3.7066681983188e-05} +{"train_loss": 0.016190936788916588, "grad_norm": 0.07875648140907288, "global_step": 30914, "epoch": 291, "lr": 3.70638191059416e-05} +{"train_loss": 0.01508653350174427, "grad_norm": 0.1300056278705597, "global_step": 30915, "epoch": 291, "lr": 3.706095627414734e-05} +{"train_loss": 0.01622290350496769, "grad_norm": 0.07956868410110474, "global_step": 30916, "epoch": 291, "lr": 3.705809348781525e-05} +{"train_loss": 0.014263050630688667, "grad_norm": 0.05684623867273331, "global_step": 30917, "epoch": 291, "lr": 3.705523074695538e-05} +{"train_loss": 0.015535159036517143, "grad_norm": 0.08108439296483994, "global_step": 30918, "epoch": 291, "lr": 3.705236805157779e-05} +{"train_loss": 0.01334176491945982, "grad_norm": 0.06959323585033417, "global_step": 30919, "epoch": 291, "lr": 3.704950540169256e-05} +{"train_loss": 0.01827467605471611, "grad_norm": 0.0711728185415268, "global_step": 30920, "epoch": 291, "lr": 3.704664279730974e-05} +{"train_loss": 0.01763048768043518, "grad_norm": 0.0950796902179718, "global_step": 30921, "epoch": 291, "lr": 3.7043780238439386e-05} +{"train_loss": 0.01570780761539936, "grad_norm": 0.10037851333618164, "global_step": 30922, "epoch": 291, "lr": 3.704091772509155e-05} +{"train_loss": 0.015252194367349148, "grad_norm": 0.10586128383874893, "global_step": 30923, "epoch": 291, "lr": 3.703805525727628e-05} +{"train_loss": 0.018555713817477226, "grad_norm": 0.13684016466140747, "global_step": 30924, "epoch": 291, "lr": 3.7035192835003655e-05} +{"train_loss": 0.015288409776985645, "grad_norm": 0.23647662997245789, "global_step": 30925, "epoch": 291, "lr": 3.7032330458283723e-05} +{"train_loss": 0.016492750495672226, "grad_norm": 0.084221750497818, "global_step": 30926, "epoch": 291, "lr": 3.702946812712654e-05} +{"train_loss": 0.013948355801403522, "grad_norm": 0.0758141353726387, "global_step": 30927, "epoch": 291, "lr": 3.7026605841542184e-05} +{"train_loss": 0.015048068016767502, "grad_norm": 0.1366114318370819, "global_step": 30928, "epoch": 291, "lr": 3.702374360154068e-05} +{"train_loss": 0.014368151314556599, "grad_norm": 0.09133480489253998, "global_step": 30929, "epoch": 291, "lr": 3.70208814071321e-05} +{"train_loss": 0.015597907826304436, "grad_norm": 0.05345352739095688, "global_step": 30930, "epoch": 291, "lr": 3.7018019258326484e-05} +{"train_loss": 0.014692207798361778, "grad_norm": 0.13107873499393463, "global_step": 30931, "epoch": 291, "lr": 3.7015157155133915e-05} +{"train_loss": 0.01700938679277897, "grad_norm": 0.09930622577667236, "global_step": 30932, "epoch": 291, "lr": 3.7012295097564444e-05} +{"train_loss": 0.01597282476723194, "grad_norm": 0.08310098201036453, "global_step": 30933, "epoch": 291, "lr": 3.700943308562812e-05} +{"train_loss": 0.015786362811923027, "grad_norm": 0.11590267717838287, "global_step": 30934, "epoch": 291, "lr": 3.700657111933499e-05} +{"train_loss": 0.013780277222394943, "grad_norm": 0.11363242566585541, "global_step": 30935, "epoch": 291, "lr": 3.7003709198695116e-05} +{"train_loss": 0.015348291955888271, "grad_norm": 0.07621560990810394, "global_step": 30936, "epoch": 291, "lr": 3.700084732371856e-05} +{"train_loss": 0.014795074239373207, "grad_norm": 0.09705396741628647, "global_step": 30937, "epoch": 291, "lr": 3.699798549441537e-05} +{"train_loss": 0.01550755649805069, "grad_norm": 0.08500891178846359, "global_step": 30938, "epoch": 291, "lr": 3.6995123710795596e-05} +{"train_loss": 0.014141296967864037, "grad_norm": 0.11919000744819641, "global_step": 30939, "epoch": 291, "lr": 3.6992261972869336e-05} +{"train_loss": 0.015623421408236027, "grad_norm": 0.07689248770475388, "global_step": 30940, "epoch": 291, "lr": 3.698940028064658e-05} +{"train_loss": 0.016431676223874092, "grad_norm": 0.09559708833694458, "global_step": 30941, "epoch": 291, "lr": 3.6986538634137416e-05} +{"train_loss": 0.017237262800335884, "grad_norm": 0.1484142392873764, "global_step": 30942, "epoch": 291, "lr": 3.69836770333519e-05} +{"train_loss": 0.01501513458788395, "grad_norm": 0.10091674327850342, "global_step": 30943, "epoch": 291, "lr": 3.698081547830008e-05} +{"train_loss": 0.015405353158712387, "grad_norm": 0.05614258348941803, "global_step": 30944, "epoch": 291, "lr": 3.6977953968992006e-05} +{"train_loss": 0.016683630645275116, "grad_norm": 0.1003977358341217, "global_step": 30945, "epoch": 291, "lr": 3.697509250543776e-05} +{"train_loss": 0.017499670386314392, "grad_norm": 0.060663238167762756, "global_step": 30946, "epoch": 291, "lr": 3.697223108764737e-05} +{"train_loss": 0.01715495064854622, "grad_norm": 0.08415812253952026, "global_step": 30947, "epoch": 291, "lr": 3.696936971563088e-05} +{"train_loss": 0.014852112159132957, "grad_norm": 0.07167047262191772, "global_step": 30948, "epoch": 291, "lr": 3.696650838939836e-05} +{"train_loss": 0.01349659450352192, "grad_norm": 0.06632620841264725, "global_step": 30949, "epoch": 291, "lr": 3.696364710895987e-05} +{"train_loss": 0.015528535470366478, "grad_norm": 0.0734134390950203, "global_step": 30950, "epoch": 291, "lr": 3.696078587432546e-05} +{"train_loss": 0.015332056569673543, "grad_norm": 0.15235033631324768, "global_step": 30951, "epoch": 291, "lr": 3.6957924685505167e-05, "val_loss": 0.01926155760884285} +{"train_loss": 0.015308992937207222, "grad_norm": 0.0890604630112648, "global_step": 30952, "epoch": 292, "lr": 3.695506354250905e-05} +{"train_loss": 0.014513990841805935, "grad_norm": 0.10545120388269424, "global_step": 30953, "epoch": 292, "lr": 3.695220244534717e-05} +{"train_loss": 0.01497574057430029, "grad_norm": 0.06606504321098328, "global_step": 30954, "epoch": 292, "lr": 3.694934139402958e-05} +{"train_loss": 0.01554867159575224, "grad_norm": 0.05803804099559784, "global_step": 30955, "epoch": 292, "lr": 3.694648038856632e-05} +{"train_loss": 0.017288971692323685, "grad_norm": 0.11028719693422318, "global_step": 30956, "epoch": 292, "lr": 3.694361942896747e-05} +{"train_loss": 0.014212679117918015, "grad_norm": 0.08355046808719635, "global_step": 30957, "epoch": 292, "lr": 3.6940758515243056e-05} +{"train_loss": 0.01606091298162937, "grad_norm": 0.08719319105148315, "global_step": 30958, "epoch": 292, "lr": 3.693789764740313e-05} +{"train_loss": 0.013543084263801575, "grad_norm": 0.07640721648931503, "global_step": 30959, "epoch": 292, "lr": 3.693503682545775e-05} +{"train_loss": 0.018392249941825867, "grad_norm": 0.08808426558971405, "global_step": 30960, "epoch": 292, "lr": 3.693217604941697e-05} +{"train_loss": 0.014682607725262642, "grad_norm": 0.10751159489154816, "global_step": 30961, "epoch": 292, "lr": 3.6929315319290845e-05} +{"train_loss": 0.015317317098379135, "grad_norm": 0.06763264536857605, "global_step": 30962, "epoch": 292, "lr": 3.6926454635089434e-05} +{"train_loss": 0.013395579531788826, "grad_norm": 0.06238575279712677, "global_step": 30963, "epoch": 292, "lr": 3.6923593996822756e-05} +{"train_loss": 0.015501504763960838, "grad_norm": 0.09011711925268173, "global_step": 30964, "epoch": 292, "lr": 3.6920733404500886e-05} +{"train_loss": 0.012035529129207134, "grad_norm": 0.0732434093952179, "global_step": 30965, "epoch": 292, "lr": 3.6917872858133874e-05} +{"train_loss": 0.013914582319557667, "grad_norm": 0.07280761748552322, "global_step": 30966, "epoch": 292, "lr": 3.691501235773176e-05} +{"train_loss": 0.015286815352737904, "grad_norm": 0.128679558634758, "global_step": 30967, "epoch": 292, "lr": 3.691215190330461e-05} +{"train_loss": 0.015254040248692036, "grad_norm": 0.08550634980201721, "global_step": 30968, "epoch": 292, "lr": 3.6909291494862484e-05} +{"train_loss": 0.01524701900780201, "grad_norm": 0.061048414558172226, "global_step": 30969, "epoch": 292, "lr": 3.690643113241539e-05} +{"train_loss": 0.014199713245034218, "grad_norm": 0.07417089492082596, "global_step": 30970, "epoch": 292, "lr": 3.69035708159734e-05} +{"train_loss": 0.01620381698012352, "grad_norm": 0.04976319149136543, "global_step": 30971, "epoch": 292, "lr": 3.690071054554658e-05} +{"train_loss": 0.015471464022994041, "grad_norm": 0.08468466252088547, "global_step": 30972, "epoch": 292, "lr": 3.689785032114495e-05} +{"train_loss": 0.017497317865490913, "grad_norm": 0.06314341723918915, "global_step": 30973, "epoch": 292, "lr": 3.689499014277859e-05} +{"train_loss": 0.016710659489035606, "grad_norm": 0.08674715459346771, "global_step": 30974, "epoch": 292, "lr": 3.6892130010457544e-05} +{"train_loss": 0.015199355781078339, "grad_norm": 0.11029642820358276, "global_step": 30975, "epoch": 292, "lr": 3.688926992419184e-05} +{"train_loss": 0.01587591879069805, "grad_norm": 0.06097002699971199, "global_step": 30976, "epoch": 292, "lr": 3.6886409883991534e-05} +{"train_loss": 0.014078238047659397, "grad_norm": 0.06191529706120491, "global_step": 30977, "epoch": 292, "lr": 3.6883549889866686e-05} +{"train_loss": 0.01791232079267502, "grad_norm": 0.12935012578964233, "global_step": 30978, "epoch": 292, "lr": 3.688068994182734e-05} +{"train_loss": 0.015671266242861748, "grad_norm": 0.06591534614562988, "global_step": 30979, "epoch": 292, "lr": 3.6877830039883546e-05} +{"train_loss": 0.014144089072942734, "grad_norm": 0.055688705295324326, "global_step": 30980, "epoch": 292, "lr": 3.6874970184045365e-05} +{"train_loss": 0.015598173253238201, "grad_norm": 0.07706043869256973, "global_step": 30981, "epoch": 292, "lr": 3.687211037432281e-05} +{"train_loss": 0.013632358983159065, "grad_norm": 0.062482643872499466, "global_step": 30982, "epoch": 292, "lr": 3.686925061072596e-05} +{"train_loss": 0.016134638339281082, "grad_norm": 0.09690994024276733, "global_step": 30983, "epoch": 292, "lr": 3.686639089326485e-05} +{"train_loss": 0.012746576219797134, "grad_norm": 0.060528431087732315, "global_step": 30984, "epoch": 292, "lr": 3.686353122194952e-05} +{"train_loss": 0.015570297837257385, "grad_norm": 0.06993266195058823, "global_step": 30985, "epoch": 292, "lr": 3.686067159679004e-05} +{"train_loss": 0.01437134388834238, "grad_norm": 0.16215786337852478, "global_step": 30986, "epoch": 292, "lr": 3.685781201779646e-05} +{"train_loss": 0.015919040888547897, "grad_norm": 0.17931751906871796, "global_step": 30987, "epoch": 292, "lr": 3.685495248497879e-05} +{"train_loss": 0.014867397025227547, "grad_norm": 0.05908755958080292, "global_step": 30988, "epoch": 292, "lr": 3.68520929983471e-05} +{"train_loss": 0.016764305531978607, "grad_norm": 0.08506914973258972, "global_step": 30989, "epoch": 292, "lr": 3.684923355791145e-05} +{"train_loss": 0.01691897213459015, "grad_norm": 0.07569332420825958, "global_step": 30990, "epoch": 292, "lr": 3.684637416368186e-05} +{"train_loss": 0.015213686972856522, "grad_norm": 0.13041932880878448, "global_step": 30991, "epoch": 292, "lr": 3.6843514815668393e-05} +{"train_loss": 0.01863887347280979, "grad_norm": 0.07756481319665909, "global_step": 30992, "epoch": 292, "lr": 3.6840655513881125e-05} +{"train_loss": 0.01631726510822773, "grad_norm": 0.08587056398391724, "global_step": 30993, "epoch": 292, "lr": 3.683779625833004e-05} +{"train_loss": 0.016303474083542824, "grad_norm": 0.08142607659101486, "global_step": 30994, "epoch": 292, "lr": 3.683493704902522e-05} +{"train_loss": 0.013828855007886887, "grad_norm": 0.07896892726421356, "global_step": 30995, "epoch": 292, "lr": 3.68320778859767e-05} +{"train_loss": 0.01760319620370865, "grad_norm": 0.10100659728050232, "global_step": 30996, "epoch": 292, "lr": 3.682921876919453e-05} +{"train_loss": 0.013609071262180805, "grad_norm": 0.07087025046348572, "global_step": 30997, "epoch": 292, "lr": 3.682635969868877e-05} +{"train_loss": 0.014564219862222672, "grad_norm": 0.06362144649028778, "global_step": 30998, "epoch": 292, "lr": 3.6823500674469457e-05} +{"train_loss": 0.01608896628022194, "grad_norm": 0.08236113935709, "global_step": 30999, "epoch": 292, "lr": 3.682064169654663e-05} +{"train_loss": 0.014388438314199448, "grad_norm": 0.12038721889257431, "global_step": 31000, "epoch": 292, "lr": 3.681778276493032e-05} +{"train_loss": 0.015112842433154583, "grad_norm": 0.06124702841043472, "global_step": 31001, "epoch": 292, "lr": 3.68149238796306e-05} +{"train_loss": 0.013785211369395256, "grad_norm": 0.0871519148349762, "global_step": 31002, "epoch": 292, "lr": 3.6812065040657504e-05} +{"train_loss": 0.013508819043636322, "grad_norm": 0.10857634246349335, "global_step": 31003, "epoch": 292, "lr": 3.680920624802109e-05} +{"train_loss": 0.017041921615600586, "grad_norm": 0.07753171026706696, "global_step": 31004, "epoch": 292, "lr": 3.680634750173137e-05} +{"train_loss": 0.018769696354866028, "grad_norm": 0.12060803174972534, "global_step": 31005, "epoch": 292, "lr": 3.68034888017984e-05} +{"train_loss": 0.01741911843419075, "grad_norm": 0.1111287996172905, "global_step": 31006, "epoch": 292, "lr": 3.680063014823224e-05} +{"train_loss": 0.014155840501189232, "grad_norm": 0.06253945827484131, "global_step": 31007, "epoch": 292, "lr": 3.679777154104293e-05} +{"train_loss": 0.01615113578736782, "grad_norm": 0.11520696431398392, "global_step": 31008, "epoch": 292, "lr": 3.6794912980240495e-05} +{"train_loss": 0.01535683125257492, "grad_norm": 0.13910813629627228, "global_step": 31009, "epoch": 292, "lr": 3.679205446583503e-05} +{"train_loss": 0.014769008383154869, "grad_norm": 0.0868898406624794, "global_step": 31010, "epoch": 292, "lr": 3.678919599783651e-05} +{"train_loss": 0.01412339136004448, "grad_norm": 0.09263177216053009, "global_step": 31011, "epoch": 292, "lr": 3.678633757625501e-05} +{"train_loss": 0.016300231218338013, "grad_norm": 0.09657587110996246, "global_step": 31012, "epoch": 292, "lr": 3.6783479201100584e-05} +{"train_loss": 0.013045519590377808, "grad_norm": 0.10253828018903732, "global_step": 31013, "epoch": 292, "lr": 3.6780620872383254e-05} +{"train_loss": 0.015110433101654053, "grad_norm": 0.07425237447023392, "global_step": 31014, "epoch": 292, "lr": 3.677776259011308e-05} +{"train_loss": 0.013601935468614101, "grad_norm": 0.04795326665043831, "global_step": 31015, "epoch": 292, "lr": 3.67749043543001e-05} +{"train_loss": 0.014697859063744545, "grad_norm": 0.10810834914445877, "global_step": 31016, "epoch": 292, "lr": 3.677204616495436e-05} +{"train_loss": 0.015018788166344166, "grad_norm": 0.06038471683859825, "global_step": 31017, "epoch": 292, "lr": 3.676918802208588e-05} +{"train_loss": 0.016342995688319206, "grad_norm": 0.09349660575389862, "global_step": 31018, "epoch": 292, "lr": 3.676632992570473e-05} +{"train_loss": 0.014874778687953949, "grad_norm": 0.13188348710536957, "global_step": 31019, "epoch": 292, "lr": 3.6763471875820936e-05} +{"train_loss": 0.015332777053117752, "grad_norm": 0.1470838040113449, "global_step": 31020, "epoch": 292, "lr": 3.6760613872444536e-05} +{"train_loss": 0.015323806554079056, "grad_norm": 0.0762786716222763, "global_step": 31021, "epoch": 292, "lr": 3.675775591558562e-05} +{"train_loss": 0.012837839312851429, "grad_norm": 0.058195002377033234, "global_step": 31022, "epoch": 292, "lr": 3.6754898005254156e-05} +{"train_loss": 0.016166843473911285, "grad_norm": 0.14069519937038422, "global_step": 31023, "epoch": 292, "lr": 3.6752040141460234e-05} +{"train_loss": 0.01619984768331051, "grad_norm": 0.08660370111465454, "global_step": 31024, "epoch": 292, "lr": 3.6749182324213876e-05} +{"train_loss": 0.015045996755361557, "grad_norm": 0.061240844428539276, "global_step": 31025, "epoch": 292, "lr": 3.6746324553525124e-05} +{"train_loss": 0.015294816344976425, "grad_norm": 0.07230597734451294, "global_step": 31026, "epoch": 292, "lr": 3.674346682940403e-05} +{"train_loss": 0.014111344702541828, "grad_norm": 0.06126699596643448, "global_step": 31027, "epoch": 292, "lr": 3.674060915186064e-05} +{"train_loss": 0.014647350646555424, "grad_norm": 0.0723581537604332, "global_step": 31028, "epoch": 292, "lr": 3.673775152090497e-05} +{"train_loss": 0.013563273474574089, "grad_norm": 0.0928550735116005, "global_step": 31029, "epoch": 292, "lr": 3.6734893936547064e-05} +{"train_loss": 0.017845073714852333, "grad_norm": 0.14279471337795258, "global_step": 31030, "epoch": 292, "lr": 3.673203639879699e-05} +{"train_loss": 0.013506617397069931, "grad_norm": 0.07350306957960129, "global_step": 31031, "epoch": 292, "lr": 3.672917890766476e-05} +{"train_loss": 0.01708969473838806, "grad_norm": 0.06339399516582489, "global_step": 31032, "epoch": 292, "lr": 3.672632146316042e-05} +{"train_loss": 0.014811556786298752, "grad_norm": 0.07457876950502396, "global_step": 31033, "epoch": 292, "lr": 3.672346406529403e-05} +{"train_loss": 0.012600588612258434, "grad_norm": 0.05465111508965492, "global_step": 31034, "epoch": 292, "lr": 3.67206067140756e-05} +{"train_loss": 0.014641134068369865, "grad_norm": 0.10105319321155548, "global_step": 31035, "epoch": 292, "lr": 3.671774940951519e-05} +{"train_loss": 0.01728535071015358, "grad_norm": 0.10132350772619247, "global_step": 31036, "epoch": 292, "lr": 3.671489215162283e-05} +{"train_loss": 0.014685675501823425, "grad_norm": 0.05673106014728546, "global_step": 31037, "epoch": 292, "lr": 3.671203494040855e-05} +{"train_loss": 0.013191252946853638, "grad_norm": 0.06795989722013474, "global_step": 31038, "epoch": 292, "lr": 3.670917777588241e-05} +{"train_loss": 0.011930357664823532, "grad_norm": 0.05419909954071045, "global_step": 31039, "epoch": 292, "lr": 3.670632065805445e-05} +{"train_loss": 0.015511816367506981, "grad_norm": 0.07510782033205032, "global_step": 31040, "epoch": 292, "lr": 3.670346358693469e-05} +{"train_loss": 0.01430490892380476, "grad_norm": 0.057066626846790314, "global_step": 31041, "epoch": 292, "lr": 3.6700606562533176e-05} +{"train_loss": 0.015636831521987915, "grad_norm": 0.09404542297124863, "global_step": 31042, "epoch": 292, "lr": 3.669774958485993e-05} +{"train_loss": 0.01843683421611786, "grad_norm": 0.08251264691352844, "global_step": 31043, "epoch": 292, "lr": 3.6694892653925026e-05} +{"train_loss": 0.017801392823457718, "grad_norm": 0.09350807964801788, "global_step": 31044, "epoch": 292, "lr": 3.669203576973848e-05} +{"train_loss": 0.013892803341150284, "grad_norm": 0.12508007884025574, "global_step": 31045, "epoch": 292, "lr": 3.668917893231035e-05} +{"train_loss": 0.015344621613621712, "grad_norm": 0.07529232650995255, "global_step": 31046, "epoch": 292, "lr": 3.668632214165063e-05} +{"train_loss": 0.016051849350333214, "grad_norm": 0.06408563256263733, "global_step": 31047, "epoch": 292, "lr": 3.668346539776939e-05} +{"train_loss": 0.016641605645418167, "grad_norm": 0.09569413959980011, "global_step": 31048, "epoch": 292, "lr": 3.668060870067666e-05} +{"train_loss": 0.015600181184709072, "grad_norm": 0.06313338875770569, "global_step": 31049, "epoch": 292, "lr": 3.6677752050382474e-05} +{"train_loss": 0.016292372718453407, "grad_norm": 0.10453897714614868, "global_step": 31050, "epoch": 292, "lr": 3.667489544689688e-05} +{"train_loss": 0.01665426790714264, "grad_norm": 0.05157201737165451, "global_step": 31051, "epoch": 292, "lr": 3.6672038890229935e-05} +{"train_loss": 0.016042958945035934, "grad_norm": 0.09108361601829529, "global_step": 31052, "epoch": 292, "lr": 3.666918238039162e-05} +{"train_loss": 0.015123434364795685, "grad_norm": 0.07894077897071838, "global_step": 31053, "epoch": 292, "lr": 3.666632591739201e-05} +{"train_loss": 0.01477720495313406, "grad_norm": 0.05640096589922905, "global_step": 31054, "epoch": 292, "lr": 3.666346950124112e-05} +{"train_loss": 0.014916799031198025, "grad_norm": 0.1106792464852333, "global_step": 31055, "epoch": 292, "lr": 3.6660613131949e-05} +{"train_loss": 0.016190212219953537, "grad_norm": 0.1273488998413086, "global_step": 31056, "epoch": 292, "lr": 3.66577568095257e-05} +{"train_loss": 0.015319273086651316, "grad_norm": 0.10741833597421646, "global_step": 31057, "epoch": 292, "lr": 3.665490053398123e-05, "val_loss": 0.019663609564304352} +{"train_loss": 0.01571081578731537, "grad_norm": 0.09618125110864639, "global_step": 31058, "epoch": 293, "lr": 3.665204430532563e-05} +{"train_loss": 0.017129678279161453, "grad_norm": 0.09826286882162094, "global_step": 31059, "epoch": 293, "lr": 3.664918812356895e-05} +{"train_loss": 0.014780530706048012, "grad_norm": 0.08490152657032013, "global_step": 31060, "epoch": 293, "lr": 3.664633198872122e-05} +{"train_loss": 0.01640385203063488, "grad_norm": 0.05889119580388069, "global_step": 31061, "epoch": 293, "lr": 3.664347590079245e-05} +{"train_loss": 0.017174966633319855, "grad_norm": 0.05658048763871193, "global_step": 31062, "epoch": 293, "lr": 3.6640619859792734e-05} +{"train_loss": 0.01542581059038639, "grad_norm": 0.10602045059204102, "global_step": 31063, "epoch": 293, "lr": 3.663776386573204e-05} +{"train_loss": 0.01860690489411354, "grad_norm": 0.09063775092363358, "global_step": 31064, "epoch": 293, "lr": 3.663490791862044e-05} +{"train_loss": 0.014284000732004642, "grad_norm": 0.07808676362037659, "global_step": 31065, "epoch": 293, "lr": 3.6632052018467974e-05} +{"train_loss": 0.015239646658301353, "grad_norm": 0.05498269945383072, "global_step": 31066, "epoch": 293, "lr": 3.662919616528464e-05} +{"train_loss": 0.014926053583621979, "grad_norm": 0.07735060900449753, "global_step": 31067, "epoch": 293, "lr": 3.662634035908051e-05} +{"train_loss": 0.013830868527293205, "grad_norm": 0.07781317830085754, "global_step": 31068, "epoch": 293, "lr": 3.662348459986562e-05} +{"train_loss": 0.014795009046792984, "grad_norm": 0.10455889999866486, "global_step": 31069, "epoch": 293, "lr": 3.662062888764997e-05} +{"train_loss": 0.017499927431344986, "grad_norm": 0.10463026911020279, "global_step": 31070, "epoch": 293, "lr": 3.6617773222443605e-05} +{"train_loss": 0.014103300869464874, "grad_norm": 0.10792683064937592, "global_step": 31071, "epoch": 293, "lr": 3.661491760425657e-05} +{"train_loss": 0.016437161713838577, "grad_norm": 0.05724746733903885, "global_step": 31072, "epoch": 293, "lr": 3.6612062033098894e-05} +{"train_loss": 0.018155548721551895, "grad_norm": 0.06494425982236862, "global_step": 31073, "epoch": 293, "lr": 3.660920650898061e-05} +{"train_loss": 0.016926845535635948, "grad_norm": 0.09173429757356644, "global_step": 31074, "epoch": 293, "lr": 3.6606351031911765e-05} +{"train_loss": 0.017096947878599167, "grad_norm": 0.07297459989786148, "global_step": 31075, "epoch": 293, "lr": 3.660349560190236e-05} +{"train_loss": 0.014189663343131542, "grad_norm": 0.15595167875289917, "global_step": 31076, "epoch": 293, "lr": 3.660064021896245e-05} +{"train_loss": 0.015753643587231636, "grad_norm": 0.10338699817657471, "global_step": 31077, "epoch": 293, "lr": 3.659778488310206e-05} +{"train_loss": 0.01828693598508835, "grad_norm": 0.0728905200958252, "global_step": 31078, "epoch": 293, "lr": 3.659492959433122e-05} +{"train_loss": 0.01517310831695795, "grad_norm": 0.07386867702007294, "global_step": 31079, "epoch": 293, "lr": 3.659207435265998e-05} +{"train_loss": 0.01885829120874405, "grad_norm": 0.06116727367043495, "global_step": 31080, "epoch": 293, "lr": 3.658921915809837e-05} +{"train_loss": 0.014727374538779259, "grad_norm": 0.11497941613197327, "global_step": 31081, "epoch": 293, "lr": 3.65863640106564e-05} +{"train_loss": 0.01380191557109356, "grad_norm": 0.09150439500808716, "global_step": 31082, "epoch": 293, "lr": 3.658350891034411e-05} +{"train_loss": 0.01426299475133419, "grad_norm": 0.07558506727218628, "global_step": 31083, "epoch": 293, "lr": 3.658065385717153e-05} +{"train_loss": 0.015314536169171333, "grad_norm": 0.09389808773994446, "global_step": 31084, "epoch": 293, "lr": 3.65777988511487e-05} +{"train_loss": 0.016344817355275154, "grad_norm": 0.1267179399728775, "global_step": 31085, "epoch": 293, "lr": 3.6574943892285654e-05} +{"train_loss": 0.014095915481448174, "grad_norm": 0.06457092612981796, "global_step": 31086, "epoch": 293, "lr": 3.657208898059242e-05} +{"train_loss": 0.014027664437890053, "grad_norm": 0.06526186317205429, "global_step": 31087, "epoch": 293, "lr": 3.6569234116079016e-05} +{"train_loss": 0.014611984603106976, "grad_norm": 0.08747623860836029, "global_step": 31088, "epoch": 293, "lr": 3.6566379298755485e-05} +{"train_loss": 0.015267709270119667, "grad_norm": 0.08140020072460175, "global_step": 31089, "epoch": 293, "lr": 3.656352452863186e-05} +{"train_loss": 0.01610506698489189, "grad_norm": 0.0913209617137909, "global_step": 31090, "epoch": 293, "lr": 3.656066980571815e-05} +{"train_loss": 0.016424421221017838, "grad_norm": 0.061617400497198105, "global_step": 31091, "epoch": 293, "lr": 3.655781513002442e-05} +{"train_loss": 0.016395092010498047, "grad_norm": 0.062394704669713974, "global_step": 31092, "epoch": 293, "lr": 3.655496050156069e-05} +{"train_loss": 0.01599481701850891, "grad_norm": 0.08232775330543518, "global_step": 31093, "epoch": 293, "lr": 3.6552105920336966e-05} +{"train_loss": 0.01375269703567028, "grad_norm": 0.09313569962978363, "global_step": 31094, "epoch": 293, "lr": 3.654925138636329e-05} +{"train_loss": 0.01502973772585392, "grad_norm": 0.09112679213285446, "global_step": 31095, "epoch": 293, "lr": 3.6546396899649694e-05} +{"train_loss": 0.015859398990869522, "grad_norm": 0.10676763206720352, "global_step": 31096, "epoch": 293, "lr": 3.6543542460206216e-05} +{"train_loss": 0.015555307269096375, "grad_norm": 0.10479871928691864, "global_step": 31097, "epoch": 293, "lr": 3.654068806804287e-05} +{"train_loss": 0.01365901529788971, "grad_norm": 0.09163474291563034, "global_step": 31098, "epoch": 293, "lr": 3.653783372316971e-05} +{"train_loss": 0.015253698453307152, "grad_norm": 0.06356773525476456, "global_step": 31099, "epoch": 293, "lr": 3.6534979425596743e-05} +{"train_loss": 0.01304347813129425, "grad_norm": 0.12829023599624634, "global_step": 31100, "epoch": 293, "lr": 3.653212517533399e-05} +{"train_loss": 0.01347125880420208, "grad_norm": 0.08033879846334457, "global_step": 31101, "epoch": 293, "lr": 3.652927097239149e-05} +{"train_loss": 0.015164462849497795, "grad_norm": 0.05887423828244209, "global_step": 31102, "epoch": 293, "lr": 3.6526416816779275e-05} +{"train_loss": 0.01600201055407524, "grad_norm": 0.07784178853034973, "global_step": 31103, "epoch": 293, "lr": 3.6523562708507403e-05} +{"train_loss": 0.014848084188997746, "grad_norm": 0.05749433487653732, "global_step": 31104, "epoch": 293, "lr": 3.652070864758584e-05} +{"train_loss": 0.01381911151111126, "grad_norm": 0.06122919172048569, "global_step": 31105, "epoch": 293, "lr": 3.651785463402465e-05} +{"train_loss": 0.018606560304760933, "grad_norm": 0.10013187676668167, "global_step": 31106, "epoch": 293, "lr": 3.6515000667833856e-05} +{"train_loss": 0.013079890981316566, "grad_norm": 0.09481581300497055, "global_step": 31107, "epoch": 293, "lr": 3.651214674902348e-05} +{"train_loss": 0.013702656142413616, "grad_norm": 0.11876002699136734, "global_step": 31108, "epoch": 293, "lr": 3.650929287760356e-05} +{"train_loss": 0.014841955155134201, "grad_norm": 0.07989948242902756, "global_step": 31109, "epoch": 293, "lr": 3.650643905358413e-05} +{"train_loss": 0.014467012137174606, "grad_norm": 0.10505102574825287, "global_step": 31110, "epoch": 293, "lr": 3.650358527697519e-05} +{"train_loss": 0.01749112270772457, "grad_norm": 0.1880575716495514, "global_step": 31111, "epoch": 293, "lr": 3.650073154778679e-05} +{"train_loss": 0.01653045415878296, "grad_norm": 0.06881967931985855, "global_step": 31112, "epoch": 293, "lr": 3.6497877866028934e-05} +{"train_loss": 0.01653868891298771, "grad_norm": 0.09868141263723373, "global_step": 31113, "epoch": 293, "lr": 3.649502423171167e-05} +{"train_loss": 0.016972750425338745, "grad_norm": 0.10537565499544144, "global_step": 31114, "epoch": 293, "lr": 3.6492170644845025e-05} +{"train_loss": 0.015702173113822937, "grad_norm": 0.1259537786245346, "global_step": 31115, "epoch": 293, "lr": 3.6489317105439023e-05} +{"train_loss": 0.0146239148452878, "grad_norm": 0.10853917896747589, "global_step": 31116, "epoch": 293, "lr": 3.6486463613503676e-05} +{"train_loss": 0.013864265754818916, "grad_norm": 0.07797709852457047, "global_step": 31117, "epoch": 293, "lr": 3.648361016904902e-05} +{"train_loss": 0.015695257112383842, "grad_norm": 0.07624892145395279, "global_step": 31118, "epoch": 293, "lr": 3.648075677208508e-05} +{"train_loss": 0.01623281091451645, "grad_norm": 0.1145811676979065, "global_step": 31119, "epoch": 293, "lr": 3.6477903422621865e-05} +{"train_loss": 0.01453495118767023, "grad_norm": 0.08201620727777481, "global_step": 31120, "epoch": 293, "lr": 3.6475050120669434e-05} +{"train_loss": 0.015371284447610378, "grad_norm": 0.06271952390670776, "global_step": 31121, "epoch": 293, "lr": 3.647219686623781e-05} +{"train_loss": 0.015747088938951492, "grad_norm": 0.14246056973934174, "global_step": 31122, "epoch": 293, "lr": 3.646934365933698e-05} +{"train_loss": 0.01370813325047493, "grad_norm": 0.09320364147424698, "global_step": 31123, "epoch": 293, "lr": 3.6466490499977e-05} +{"train_loss": 0.014304833486676216, "grad_norm": 0.06916438788175583, "global_step": 31124, "epoch": 293, "lr": 3.646363738816787e-05} +{"train_loss": 0.015303523279726505, "grad_norm": 0.053456999361515045, "global_step": 31125, "epoch": 293, "lr": 3.6460784323919654e-05} +{"train_loss": 0.017878200858831406, "grad_norm": 0.08159859478473663, "global_step": 31126, "epoch": 293, "lr": 3.6457931307242343e-05} +{"train_loss": 0.013740329071879387, "grad_norm": 0.083661288022995, "global_step": 31127, "epoch": 293, "lr": 3.645507833814599e-05} +{"train_loss": 0.01436246931552887, "grad_norm": 0.09795175492763519, "global_step": 31128, "epoch": 293, "lr": 3.645222541664059e-05} +{"train_loss": 0.014758807606995106, "grad_norm": 0.07156489789485931, "global_step": 31129, "epoch": 293, "lr": 3.644937254273616e-05} +{"train_loss": 0.015748005360364914, "grad_norm": 0.08045893907546997, "global_step": 31130, "epoch": 293, "lr": 3.6446519716442763e-05} +{"train_loss": 0.01589161530137062, "grad_norm": 0.06867103278636932, "global_step": 31131, "epoch": 293, "lr": 3.644366693777038e-05} +{"train_loss": 0.013002696447074413, "grad_norm": 0.09465562552213669, "global_step": 31132, "epoch": 293, "lr": 3.644081420672908e-05} +{"train_loss": 0.016684431582689285, "grad_norm": 0.09156246483325958, "global_step": 31133, "epoch": 293, "lr": 3.6437961523328857e-05} +{"train_loss": 0.016520420089364052, "grad_norm": 0.1227293610572815, "global_step": 31134, "epoch": 293, "lr": 3.643510888757974e-05} +{"train_loss": 0.014495874755084515, "grad_norm": 0.09096140414476395, "global_step": 31135, "epoch": 293, "lr": 3.643225629949175e-05} +{"train_loss": 0.012409133836627007, "grad_norm": 0.05539095029234886, "global_step": 31136, "epoch": 293, "lr": 3.64294037590749e-05} +{"train_loss": 0.017366599291563034, "grad_norm": 0.10185456275939941, "global_step": 31137, "epoch": 293, "lr": 3.6426551266339224e-05} +{"train_loss": 0.014979034662246704, "grad_norm": 0.07672199606895447, "global_step": 31138, "epoch": 293, "lr": 3.6423698821294756e-05} +{"train_loss": 0.0138889504596591, "grad_norm": 0.071385957300663, "global_step": 31139, "epoch": 293, "lr": 3.642084642395151e-05} +{"train_loss": 0.014894312247633934, "grad_norm": 0.08414362370967865, "global_step": 31140, "epoch": 293, "lr": 3.64179940743195e-05} +{"train_loss": 0.01262272335588932, "grad_norm": 0.05575164780020714, "global_step": 31141, "epoch": 293, "lr": 3.641514177240874e-05} +{"train_loss": 0.015083365142345428, "grad_norm": 0.11225275695323944, "global_step": 31142, "epoch": 293, "lr": 3.641228951822927e-05} +{"train_loss": 0.017512772232294083, "grad_norm": 0.08862771093845367, "global_step": 31143, "epoch": 293, "lr": 3.6409437311791105e-05} +{"train_loss": 0.015563436783850193, "grad_norm": 0.0776742696762085, "global_step": 31144, "epoch": 293, "lr": 3.640658515310426e-05} +{"train_loss": 0.01544187031686306, "grad_norm": 0.1281675100326538, "global_step": 31145, "epoch": 293, "lr": 3.640373304217878e-05} +{"train_loss": 0.0161641463637352, "grad_norm": 0.08633679151535034, "global_step": 31146, "epoch": 293, "lr": 3.6400880979024666e-05} +{"train_loss": 0.012637849897146225, "grad_norm": 0.07411130517721176, "global_step": 31147, "epoch": 293, "lr": 3.639802896365193e-05} +{"train_loss": 0.014184756204485893, "grad_norm": 0.1118691936135292, "global_step": 31148, "epoch": 293, "lr": 3.6395176996070605e-05} +{"train_loss": 0.015221287496387959, "grad_norm": 0.18556411564350128, "global_step": 31149, "epoch": 293, "lr": 3.639232507629072e-05} +{"train_loss": 0.013772834092378616, "grad_norm": 0.07946784794330597, "global_step": 31150, "epoch": 293, "lr": 3.638947320432229e-05} +{"train_loss": 0.013235240243375301, "grad_norm": 0.11428374797105789, "global_step": 31151, "epoch": 293, "lr": 3.6386621380175335e-05} +{"train_loss": 0.013563482090830803, "grad_norm": 0.1456376016139984, "global_step": 31152, "epoch": 293, "lr": 3.638376960385987e-05} +{"train_loss": 0.016264354810118675, "grad_norm": 0.049280159175395966, "global_step": 31153, "epoch": 293, "lr": 3.63809178753859e-05} +{"train_loss": 0.014380760490894318, "grad_norm": 0.05588043853640556, "global_step": 31154, "epoch": 293, "lr": 3.637806619476347e-05} +{"train_loss": 0.017321638762950897, "grad_norm": 0.11968239396810532, "global_step": 31155, "epoch": 293, "lr": 3.63752145620026e-05} +{"train_loss": 0.018866021186113358, "grad_norm": 0.08748351037502289, "global_step": 31156, "epoch": 293, "lr": 3.637236297711331e-05} +{"train_loss": 0.01424243114888668, "grad_norm": 0.07055618613958359, "global_step": 31157, "epoch": 293, "lr": 3.636951144010559e-05} +{"train_loss": 0.014173099771142006, "grad_norm": 0.12423616647720337, "global_step": 31158, "epoch": 293, "lr": 3.636665995098948e-05} +{"train_loss": 0.015020616352558136, "grad_norm": 0.04812520742416382, "global_step": 31159, "epoch": 293, "lr": 3.6363808509774996e-05} +{"train_loss": 0.01297839730978012, "grad_norm": 0.11300918459892273, "global_step": 31160, "epoch": 293, "lr": 3.636095711647216e-05} +{"train_loss": 0.014705007895827293, "grad_norm": 0.09199389070272446, "global_step": 31161, "epoch": 293, "lr": 3.6358105771091e-05} +{"train_loss": 0.01721026748418808, "grad_norm": 0.11801569163799286, "global_step": 31162, "epoch": 293, "lr": 3.6355254473641534e-05} +{"train_loss": 0.015310817828647932, "grad_norm": 0.11368794739246368, "global_step": 31163, "epoch": 293, "lr": 3.635240322413374e-05, "val_loss": 0.019684594124555588} +{"train_loss": 0.014970320276916027, "grad_norm": 0.13822583854198456, "global_step": 31164, "epoch": 294, "lr": 3.6349552022577694e-05} +{"train_loss": 0.014452156610786915, "grad_norm": 0.05389724671840668, "global_step": 31165, "epoch": 294, "lr": 3.634670086898336e-05} +{"train_loss": 0.016197141259908676, "grad_norm": 0.08218901604413986, "global_step": 31166, "epoch": 294, "lr": 3.634384976336079e-05} +{"train_loss": 0.0162283256649971, "grad_norm": 0.0653616189956665, "global_step": 31167, "epoch": 294, "lr": 3.634099870572e-05} +{"train_loss": 0.014646337367594242, "grad_norm": 0.07214796543121338, "global_step": 31168, "epoch": 294, "lr": 3.633814769607101e-05} +{"train_loss": 0.01608962193131447, "grad_norm": 0.08123692125082016, "global_step": 31169, "epoch": 294, "lr": 3.633529673442381e-05} +{"train_loss": 0.016253937035799026, "grad_norm": 0.06225213780999184, "global_step": 31170, "epoch": 294, "lr": 3.633244582078843e-05} +{"train_loss": 0.015020499937236309, "grad_norm": 0.10807059705257416, "global_step": 31171, "epoch": 294, "lr": 3.6329594955174906e-05} +{"train_loss": 0.019879328086972237, "grad_norm": 0.09571130573749542, "global_step": 31172, "epoch": 294, "lr": 3.632674413759323e-05} +{"train_loss": 0.015106331557035446, "grad_norm": 0.09111016988754272, "global_step": 31173, "epoch": 294, "lr": 3.632389336805343e-05} +{"train_loss": 0.013677440583705902, "grad_norm": 0.05368725582957268, "global_step": 31174, "epoch": 294, "lr": 3.6321042646565526e-05} +{"train_loss": 0.014806794933974743, "grad_norm": 0.07801894843578339, "global_step": 31175, "epoch": 294, "lr": 3.6318191973139534e-05} +{"train_loss": 0.01234695315361023, "grad_norm": 0.09477204084396362, "global_step": 31176, "epoch": 294, "lr": 3.631534134778546e-05} +{"train_loss": 0.014972256496548653, "grad_norm": 0.06963592022657394, "global_step": 31177, "epoch": 294, "lr": 3.631249077051331e-05} +{"train_loss": 0.011609536595642567, "grad_norm": 0.07186310738325119, "global_step": 31178, "epoch": 294, "lr": 3.630964024133313e-05} +{"train_loss": 0.015486432239413261, "grad_norm": 0.06717048585414886, "global_step": 31179, "epoch": 294, "lr": 3.630678976025492e-05} +{"train_loss": 0.015256579965353012, "grad_norm": 0.10613548755645752, "global_step": 31180, "epoch": 294, "lr": 3.6303939327288716e-05} +{"train_loss": 0.015369432047009468, "grad_norm": 0.058327991515398026, "global_step": 31181, "epoch": 294, "lr": 3.630108894244449e-05} +{"train_loss": 0.016356728971004486, "grad_norm": 0.06858434528112411, "global_step": 31182, "epoch": 294, "lr": 3.629823860573227e-05} +{"train_loss": 0.0161273255944252, "grad_norm": 0.09288130700588226, "global_step": 31183, "epoch": 294, "lr": 3.6295388317162094e-05} +{"train_loss": 0.016018077731132507, "grad_norm": 0.10635310411453247, "global_step": 31184, "epoch": 294, "lr": 3.6292538076743966e-05} +{"train_loss": 0.014107203111052513, "grad_norm": 0.12333373725414276, "global_step": 31185, "epoch": 294, "lr": 3.628968788448789e-05} +{"train_loss": 0.0162544846534729, "grad_norm": 0.11676157265901566, "global_step": 31186, "epoch": 294, "lr": 3.62868377404039e-05} +{"train_loss": 0.015752991661429405, "grad_norm": 0.08417651802301407, "global_step": 31187, "epoch": 294, "lr": 3.6283987644501994e-05} +{"train_loss": 0.013982024975121021, "grad_norm": 0.06932578235864639, "global_step": 31188, "epoch": 294, "lr": 3.628113759679219e-05} +{"train_loss": 0.01615506038069725, "grad_norm": 0.07335656881332397, "global_step": 31189, "epoch": 294, "lr": 3.627828759728449e-05} +{"train_loss": 0.016218453645706177, "grad_norm": 0.07709653675556183, "global_step": 31190, "epoch": 294, "lr": 3.627543764598893e-05} +{"train_loss": 0.013566562905907631, "grad_norm": 0.08573025465011597, "global_step": 31191, "epoch": 294, "lr": 3.627258774291551e-05} +{"train_loss": 0.014567902311682701, "grad_norm": 0.05360113084316254, "global_step": 31192, "epoch": 294, "lr": 3.626973788807426e-05} +{"train_loss": 0.01423671841621399, "grad_norm": 0.1128629520535469, "global_step": 31193, "epoch": 294, "lr": 3.6266888081475175e-05} +{"train_loss": 0.014971055090427399, "grad_norm": 0.07061339169740677, "global_step": 31194, "epoch": 294, "lr": 3.626403832312826e-05} +{"train_loss": 0.013384168036282063, "grad_norm": 0.10864878445863724, "global_step": 31195, "epoch": 294, "lr": 3.626118861304355e-05} +{"train_loss": 0.01378706470131874, "grad_norm": 0.07224355638027191, "global_step": 31196, "epoch": 294, "lr": 3.6258338951231046e-05} +{"train_loss": 0.014136068522930145, "grad_norm": 0.060637813061475754, "global_step": 31197, "epoch": 294, "lr": 3.6255489337700756e-05} +{"train_loss": 0.01501426100730896, "grad_norm": 0.09722166508436203, "global_step": 31198, "epoch": 294, "lr": 3.625263977246273e-05} +{"train_loss": 0.014207662083208561, "grad_norm": 0.07639764994382858, "global_step": 31199, "epoch": 294, "lr": 3.6249790255526915e-05} +{"train_loss": 0.01342700608074665, "grad_norm": 0.09560254961252213, "global_step": 31200, "epoch": 294, "lr": 3.624694078690337e-05} +{"train_loss": 0.015836680307984352, "grad_norm": 0.0773640125989914, "global_step": 31201, "epoch": 294, "lr": 3.62440913666021e-05} +{"train_loss": 0.014165811240673065, "grad_norm": 0.05296836793422699, "global_step": 31202, "epoch": 294, "lr": 3.6241241994633094e-05} +{"train_loss": 0.01723407581448555, "grad_norm": 0.09521189332008362, "global_step": 31203, "epoch": 294, "lr": 3.6238392671006395e-05} +{"train_loss": 0.01504512969404459, "grad_norm": 0.13631725311279297, "global_step": 31204, "epoch": 294, "lr": 3.6235543395732e-05} +{"train_loss": 0.01651996560394764, "grad_norm": 0.06922907382249832, "global_step": 31205, "epoch": 294, "lr": 3.623269416881991e-05} +{"train_loss": 0.01742524467408657, "grad_norm": 0.0818900465965271, "global_step": 31206, "epoch": 294, "lr": 3.6229844990280134e-05} +{"train_loss": 0.013297159224748611, "grad_norm": 0.07318101078271866, "global_step": 31207, "epoch": 294, "lr": 3.6226995860122713e-05} +{"train_loss": 0.013915786519646645, "grad_norm": 0.08179398626089096, "global_step": 31208, "epoch": 294, "lr": 3.622414677835764e-05} +{"train_loss": 0.01471927110105753, "grad_norm": 0.10395167022943497, "global_step": 31209, "epoch": 294, "lr": 3.622129774499493e-05} +{"train_loss": 0.013867801055312157, "grad_norm": 0.07880118489265442, "global_step": 31210, "epoch": 294, "lr": 3.621844876004458e-05} +{"train_loss": 0.017959285527467728, "grad_norm": 0.10719919204711914, "global_step": 31211, "epoch": 294, "lr": 3.6215599823516585e-05} +{"train_loss": 0.016282323747873306, "grad_norm": 0.0708015188574791, "global_step": 31212, "epoch": 294, "lr": 3.6212750935421e-05} +{"train_loss": 0.015261314809322357, "grad_norm": 0.06338805705308914, "global_step": 31213, "epoch": 294, "lr": 3.6209902095767814e-05} +{"train_loss": 0.01618635468184948, "grad_norm": 0.07196971029043198, "global_step": 31214, "epoch": 294, "lr": 3.620705330456702e-05} +{"train_loss": 0.01744356006383896, "grad_norm": 0.05956416204571724, "global_step": 31215, "epoch": 294, "lr": 3.6204204561828666e-05} +{"train_loss": 0.014099664986133575, "grad_norm": 0.1474798619747162, "global_step": 31216, "epoch": 294, "lr": 3.620135586756273e-05} +{"train_loss": 0.015042256563901901, "grad_norm": 0.10938140749931335, "global_step": 31217, "epoch": 294, "lr": 3.6198507221779214e-05} +{"train_loss": 0.01403263583779335, "grad_norm": 0.06775926053524017, "global_step": 31218, "epoch": 294, "lr": 3.619565862448814e-05} +{"train_loss": 0.01550012081861496, "grad_norm": 0.08110973984003067, "global_step": 31219, "epoch": 294, "lr": 3.6192810075699535e-05} +{"train_loss": 0.014799578115344048, "grad_norm": 0.08209886401891708, "global_step": 31220, "epoch": 294, "lr": 3.618996157542338e-05} +{"train_loss": 0.017287250608205795, "grad_norm": 0.10467501729726791, "global_step": 31221, "epoch": 294, "lr": 3.618711312366971e-05} +{"train_loss": 0.017079845070838928, "grad_norm": 0.060722559690475464, "global_step": 31222, "epoch": 294, "lr": 3.618426472044851e-05} +{"train_loss": 0.016162361949682236, "grad_norm": 0.06399800628423691, "global_step": 31223, "epoch": 294, "lr": 3.618141636576978e-05} +{"train_loss": 0.015084095299243927, "grad_norm": 0.05944984406232834, "global_step": 31224, "epoch": 294, "lr": 3.617856805964355e-05} +{"train_loss": 0.01565399020910263, "grad_norm": 0.08297323435544968, "global_step": 31225, "epoch": 294, "lr": 3.617571980207983e-05} +{"train_loss": 0.013742249459028244, "grad_norm": 0.05660012736916542, "global_step": 31226, "epoch": 294, "lr": 3.617287159308861e-05} +{"train_loss": 0.01418662816286087, "grad_norm": 0.08515956997871399, "global_step": 31227, "epoch": 294, "lr": 3.617002343267993e-05} +{"train_loss": 0.01711009070277214, "grad_norm": 0.12510351836681366, "global_step": 31228, "epoch": 294, "lr": 3.6167175320863744e-05} +{"train_loss": 0.015543429180979729, "grad_norm": 0.08124158531427383, "global_step": 31229, "epoch": 294, "lr": 3.61643272576501e-05} +{"train_loss": 0.015398099087178707, "grad_norm": 0.07413657009601593, "global_step": 31230, "epoch": 294, "lr": 3.616147924304899e-05} +{"train_loss": 0.014338338747620583, "grad_norm": 0.10432755202054977, "global_step": 31231, "epoch": 294, "lr": 3.615863127707041e-05} +{"train_loss": 0.01667887531220913, "grad_norm": 0.07856621593236923, "global_step": 31232, "epoch": 294, "lr": 3.615578335972439e-05} +{"train_loss": 0.014309681951999664, "grad_norm": 0.06556270271539688, "global_step": 31233, "epoch": 294, "lr": 3.615293549102095e-05} +{"train_loss": 0.01671963930130005, "grad_norm": 0.09478900581598282, "global_step": 31234, "epoch": 294, "lr": 3.615008767097006e-05} +{"train_loss": 0.015337567776441574, "grad_norm": 0.12231501936912537, "global_step": 31235, "epoch": 294, "lr": 3.614723989958172e-05} +{"train_loss": 0.01712915301322937, "grad_norm": 0.08204708248376846, "global_step": 31236, "epoch": 294, "lr": 3.6144392176865965e-05} +{"train_loss": 0.014902208931744099, "grad_norm": 0.07366674393415451, "global_step": 31237, "epoch": 294, "lr": 3.6141544502832795e-05} +{"train_loss": 0.017556659877300262, "grad_norm": 0.12113447487354279, "global_step": 31238, "epoch": 294, "lr": 3.6138696877492194e-05} +{"train_loss": 0.016363784670829773, "grad_norm": 0.1480700820684433, "global_step": 31239, "epoch": 294, "lr": 3.6135849300854225e-05} +{"train_loss": 0.016550470143556595, "grad_norm": 0.07507508993148804, "global_step": 31240, "epoch": 294, "lr": 3.613300177292881e-05} +{"train_loss": 0.014857616275548935, "grad_norm": 0.07357589155435562, "global_step": 31241, "epoch": 294, "lr": 3.613015429372602e-05} +{"train_loss": 0.01584295555949211, "grad_norm": 0.07145905494689941, "global_step": 31242, "epoch": 294, "lr": 3.612730686325583e-05} +{"train_loss": 0.015149933286011219, "grad_norm": 0.06778695434331894, "global_step": 31243, "epoch": 294, "lr": 3.612445948152823e-05} +{"train_loss": 0.013549108058214188, "grad_norm": 0.07855954021215439, "global_step": 31244, "epoch": 294, "lr": 3.6121612148553264e-05} +{"train_loss": 0.01497175544500351, "grad_norm": 0.06388075649738312, "global_step": 31245, "epoch": 294, "lr": 3.611876486434093e-05} +{"train_loss": 0.01438783761113882, "grad_norm": 0.08817816525697708, "global_step": 31246, "epoch": 294, "lr": 3.6115917628901205e-05} +{"train_loss": 0.016862941905856133, "grad_norm": 0.08356401324272156, "global_step": 31247, "epoch": 294, "lr": 3.6113070442244104e-05} +{"train_loss": 0.01551644317805767, "grad_norm": 0.07287590950727463, "global_step": 31248, "epoch": 294, "lr": 3.611022330437964e-05} +{"train_loss": 0.014413056895136833, "grad_norm": 0.07166354358196259, "global_step": 31249, "epoch": 294, "lr": 3.6107376215317815e-05} +{"train_loss": 0.013755051419138908, "grad_norm": 0.05972873792052269, "global_step": 31250, "epoch": 294, "lr": 3.6104529175068604e-05} +{"train_loss": 0.014960803091526031, "grad_norm": 0.1144426092505455, "global_step": 31251, "epoch": 294, "lr": 3.610168218364208e-05} +{"train_loss": 0.014993181452155113, "grad_norm": 0.17630921304225922, "global_step": 31252, "epoch": 294, "lr": 3.6098835241048165e-05} +{"train_loss": 0.015639007091522217, "grad_norm": 0.07362339645624161, "global_step": 31253, "epoch": 294, "lr": 3.609598834729691e-05} +{"train_loss": 0.014175986871123314, "grad_norm": 0.06829338520765305, "global_step": 31254, "epoch": 294, "lr": 3.6093141502398294e-05} +{"train_loss": 0.01807596907019615, "grad_norm": 0.06538716703653336, "global_step": 31255, "epoch": 294, "lr": 3.609029470636233e-05} +{"train_loss": 0.015999820083379745, "grad_norm": 0.06816470623016357, "global_step": 31256, "epoch": 294, "lr": 3.608744795919905e-05} +{"train_loss": 0.015456777065992355, "grad_norm": 0.06711295247077942, "global_step": 31257, "epoch": 294, "lr": 3.60846012609184e-05} +{"train_loss": 0.015085483901202679, "grad_norm": 0.07326959073543549, "global_step": 31258, "epoch": 294, "lr": 3.608175461153042e-05} +{"train_loss": 0.013285018503665924, "grad_norm": 0.11931417137384415, "global_step": 31259, "epoch": 294, "lr": 3.60789080110451e-05} +{"train_loss": 0.016818862408399582, "grad_norm": 0.07878085970878601, "global_step": 31260, "epoch": 294, "lr": 3.6076061459472424e-05} +{"train_loss": 0.014557771384716034, "grad_norm": 0.08147507905960083, "global_step": 31261, "epoch": 294, "lr": 3.6073214956822434e-05} +{"train_loss": 0.015252579003572464, "grad_norm": 0.09157969802618027, "global_step": 31262, "epoch": 294, "lr": 3.607036850310512e-05} +{"train_loss": 0.01592453569173813, "grad_norm": 0.0671328604221344, "global_step": 31263, "epoch": 294, "lr": 3.6067522098330456e-05} +{"train_loss": 0.013610854744911194, "grad_norm": 0.09165889769792557, "global_step": 31264, "epoch": 294, "lr": 3.606467574250846e-05} +{"train_loss": 0.017947625368833542, "grad_norm": 0.07322690635919571, "global_step": 31265, "epoch": 294, "lr": 3.606182943564913e-05} +{"train_loss": 0.016158096492290497, "grad_norm": 0.09323031455278397, "global_step": 31266, "epoch": 294, "lr": 3.6058983177762476e-05} +{"train_loss": 0.015608925372362137, "grad_norm": 0.11193472146987915, "global_step": 31267, "epoch": 294, "lr": 3.605613696885849e-05} +{"train_loss": 0.013667972758412361, "grad_norm": 0.07984122633934021, "global_step": 31268, "epoch": 294, "lr": 3.605329080894719e-05} +{"train_loss": 0.015294020608910974, "grad_norm": 0.08917395770549774, "global_step": 31269, "epoch": 294, "lr": 3.605044469803854e-05, "val_loss": 0.01910475827753544} +{"train_loss": 0.015008015558123589, "grad_norm": 0.05252016335725784, "global_step": 31270, "epoch": 295, "lr": 3.604759863614258e-05} +{"train_loss": 0.015448580496013165, "grad_norm": 0.06556005030870438, "global_step": 31271, "epoch": 295, "lr": 3.604475262326927e-05} +{"train_loss": 0.013494586572051048, "grad_norm": 0.1019243597984314, "global_step": 31272, "epoch": 295, "lr": 3.6041906659428635e-05} +{"train_loss": 0.015359289944171906, "grad_norm": 0.0862019956111908, "global_step": 31273, "epoch": 295, "lr": 3.603906074463068e-05} +{"train_loss": 0.015031895600259304, "grad_norm": 0.06080543249845505, "global_step": 31274, "epoch": 295, "lr": 3.60362148788854e-05} +{"train_loss": 0.01783963292837143, "grad_norm": 0.06440236419439316, "global_step": 31275, "epoch": 295, "lr": 3.6033369062202774e-05} +{"train_loss": 0.016054965555667877, "grad_norm": 0.06818746030330658, "global_step": 31276, "epoch": 295, "lr": 3.603052329459281e-05} +{"train_loss": 0.014615310356020927, "grad_norm": 0.05558806285262108, "global_step": 31277, "epoch": 295, "lr": 3.602767757606552e-05} +{"train_loss": 0.017648350447416306, "grad_norm": 0.0744401067495346, "global_step": 31278, "epoch": 295, "lr": 3.6024831906630896e-05} +{"train_loss": 0.017914369702339172, "grad_norm": 0.07684250175952911, "global_step": 31279, "epoch": 295, "lr": 3.602198628629893e-05} +{"train_loss": 0.0186406672000885, "grad_norm": 0.06529706716537476, "global_step": 31280, "epoch": 295, "lr": 3.6019140715079646e-05} +{"train_loss": 0.013427745550870895, "grad_norm": 0.1661325991153717, "global_step": 31281, "epoch": 295, "lr": 3.6016295192982995e-05} +{"train_loss": 0.015831410884857178, "grad_norm": 0.06340822577476501, "global_step": 31282, "epoch": 295, "lr": 3.601344972001901e-05} +{"train_loss": 0.014474639669060707, "grad_norm": 0.06023747846484184, "global_step": 31283, "epoch": 295, "lr": 3.601060429619768e-05} +{"train_loss": 0.01711306907236576, "grad_norm": 0.08860474824905396, "global_step": 31284, "epoch": 295, "lr": 3.6007758921529e-05} +{"train_loss": 0.01564418151974678, "grad_norm": 0.14546561241149902, "global_step": 31285, "epoch": 295, "lr": 3.6004913596022974e-05} +{"train_loss": 0.01540030725300312, "grad_norm": 0.061028871685266495, "global_step": 31286, "epoch": 295, "lr": 3.6002068319689605e-05} +{"train_loss": 0.01704167202115059, "grad_norm": 0.10822945833206177, "global_step": 31287, "epoch": 295, "lr": 3.5999223092538867e-05} +{"train_loss": 0.018025889992713928, "grad_norm": 0.08315058797597885, "global_step": 31288, "epoch": 295, "lr": 3.599637791458077e-05} +{"train_loss": 0.015232869423925877, "grad_norm": 0.08909562975168228, "global_step": 31289, "epoch": 295, "lr": 3.599353278582531e-05} +{"train_loss": 0.015794258564710617, "grad_norm": 0.08350195735692978, "global_step": 31290, "epoch": 295, "lr": 3.599068770628249e-05} +{"train_loss": 0.013414277695119381, "grad_norm": 0.14285361766815186, "global_step": 31291, "epoch": 295, "lr": 3.598784267596229e-05} +{"train_loss": 0.015373137779533863, "grad_norm": 0.07487520575523376, "global_step": 31292, "epoch": 295, "lr": 3.5984997694874754e-05} +{"train_loss": 0.014537900686264038, "grad_norm": 0.11179205030202866, "global_step": 31293, "epoch": 295, "lr": 3.59821527630298e-05} +{"train_loss": 0.015015043318271637, "grad_norm": 0.10278210043907166, "global_step": 31294, "epoch": 295, "lr": 3.597930788043748e-05} +{"train_loss": 0.014627491123974323, "grad_norm": 0.07091937214136124, "global_step": 31295, "epoch": 295, "lr": 3.597646304710778e-05} +{"train_loss": 0.014931876212358475, "grad_norm": 0.1339263617992401, "global_step": 31296, "epoch": 295, "lr": 3.5973618263050676e-05} +{"train_loss": 0.015231192111968994, "grad_norm": 0.06382561475038528, "global_step": 31297, "epoch": 295, "lr": 3.597077352827618e-05} +{"train_loss": 0.013022969476878643, "grad_norm": 0.08356885612010956, "global_step": 31298, "epoch": 295, "lr": 3.596792884279431e-05} +{"train_loss": 0.013319451361894608, "grad_norm": 0.12923665344715118, "global_step": 31299, "epoch": 295, "lr": 3.5965084206615014e-05} +{"train_loss": 0.014268746599555016, "grad_norm": 0.056215107440948486, "global_step": 31300, "epoch": 295, "lr": 3.596223961974831e-05} +{"train_loss": 0.016635019332170486, "grad_norm": 0.05607226490974426, "global_step": 31301, "epoch": 295, "lr": 3.595939508220418e-05} +{"train_loss": 0.016095591709017754, "grad_norm": 0.10628695785999298, "global_step": 31302, "epoch": 295, "lr": 3.5956550593992646e-05} +{"train_loss": 0.011446431279182434, "grad_norm": 0.05095295235514641, "global_step": 31303, "epoch": 295, "lr": 3.595370615512368e-05} +{"train_loss": 0.014989316463470459, "grad_norm": 0.10422342270612717, "global_step": 31304, "epoch": 295, "lr": 3.59508617656073e-05} +{"train_loss": 0.01426574494689703, "grad_norm": 0.07201571762561798, "global_step": 31305, "epoch": 295, "lr": 3.594801742545345e-05} +{"train_loss": 0.014528954401612282, "grad_norm": 0.07225276529788971, "global_step": 31306, "epoch": 295, "lr": 3.594517313467217e-05} +{"train_loss": 0.012224340811371803, "grad_norm": 0.07171090692281723, "global_step": 31307, "epoch": 295, "lr": 3.5942328893273445e-05} +{"train_loss": 0.01873205229640007, "grad_norm": 0.1257527619600296, "global_step": 31308, "epoch": 295, "lr": 3.5939484701267244e-05} +{"train_loss": 0.014807606115937233, "grad_norm": 0.1625450849533081, "global_step": 31309, "epoch": 295, "lr": 3.5936640558663616e-05} +{"train_loss": 0.014854692853987217, "grad_norm": 0.07094830274581909, "global_step": 31310, "epoch": 295, "lr": 3.593379646547248e-05} +{"train_loss": 0.014520584605634212, "grad_norm": 0.0843813493847847, "global_step": 31311, "epoch": 295, "lr": 3.593095242170389e-05} +{"train_loss": 0.015204383060336113, "grad_norm": 0.2350158542394638, "global_step": 31312, "epoch": 295, "lr": 3.5928108427367805e-05} +{"train_loss": 0.015983879566192627, "grad_norm": 0.10133771598339081, "global_step": 31313, "epoch": 295, "lr": 3.592526448247422e-05} +{"train_loss": 0.015946630388498306, "grad_norm": 0.10431212931871414, "global_step": 31314, "epoch": 295, "lr": 3.5922420587033136e-05} +{"train_loss": 0.015175970271229744, "grad_norm": 0.07043895125389099, "global_step": 31315, "epoch": 295, "lr": 3.591957674105456e-05} +{"train_loss": 0.015337938442826271, "grad_norm": 0.06148231774568558, "global_step": 31316, "epoch": 295, "lr": 3.591673294454846e-05} +{"train_loss": 0.012672457844018936, "grad_norm": 0.05211485177278519, "global_step": 31317, "epoch": 295, "lr": 3.591388919752482e-05} +{"train_loss": 0.012554856017231941, "grad_norm": 0.09312666207551956, "global_step": 31318, "epoch": 295, "lr": 3.591104549999367e-05} +{"train_loss": 0.017097383737564087, "grad_norm": 0.07433488965034485, "global_step": 31319, "epoch": 295, "lr": 3.590820185196497e-05} +{"train_loss": 0.01545781921595335, "grad_norm": 0.05949723720550537, "global_step": 31320, "epoch": 295, "lr": 3.5905358253448706e-05} +{"train_loss": 0.018972912803292274, "grad_norm": 0.20503510534763336, "global_step": 31321, "epoch": 295, "lr": 3.590251470445492e-05} +{"train_loss": 0.013211965560913086, "grad_norm": 0.056802764534950256, "global_step": 31322, "epoch": 295, "lr": 3.589967120499353e-05} +{"train_loss": 0.01601705327630043, "grad_norm": 0.08886738866567612, "global_step": 31323, "epoch": 295, "lr": 3.589682775507458e-05} +{"train_loss": 0.01566983386874199, "grad_norm": 0.10544843971729279, "global_step": 31324, "epoch": 295, "lr": 3.5893984354708045e-05} +{"train_loss": 0.013310843147337437, "grad_norm": 0.0942600667476654, "global_step": 31325, "epoch": 295, "lr": 3.58911410039039e-05} +{"train_loss": 0.013217109255492687, "grad_norm": 0.06141718477010727, "global_step": 31326, "epoch": 295, "lr": 3.5888297702672165e-05} +{"train_loss": 0.012490976601839066, "grad_norm": 0.06896350532770157, "global_step": 31327, "epoch": 295, "lr": 3.588545445102281e-05} +{"train_loss": 0.01427979301661253, "grad_norm": 0.05641532689332962, "global_step": 31328, "epoch": 295, "lr": 3.588261124896584e-05} +{"train_loss": 0.02061929553747177, "grad_norm": 0.10873951762914658, "global_step": 31329, "epoch": 295, "lr": 3.5879768096511226e-05} +{"train_loss": 0.017903588712215424, "grad_norm": 0.08504951745271683, "global_step": 31330, "epoch": 295, "lr": 3.587692499366896e-05} +{"train_loss": 0.01768176443874836, "grad_norm": 0.08310331404209137, "global_step": 31331, "epoch": 295, "lr": 3.587408194044904e-05} +{"train_loss": 0.01311499997973442, "grad_norm": 0.11083991825580597, "global_step": 31332, "epoch": 295, "lr": 3.587123893686145e-05} +{"train_loss": 0.015582796186208725, "grad_norm": 0.10909240692853928, "global_step": 31333, "epoch": 295, "lr": 3.5868395982916205e-05} +{"train_loss": 0.015198413282632828, "grad_norm": 0.08310171961784363, "global_step": 31334, "epoch": 295, "lr": 3.586555307862324e-05} +{"train_loss": 0.0159362331032753, "grad_norm": 0.07707911729812622, "global_step": 31335, "epoch": 295, "lr": 3.586271022399259e-05} +{"train_loss": 0.01382334902882576, "grad_norm": 0.08296672254800797, "global_step": 31336, "epoch": 295, "lr": 3.585986741903423e-05} +{"train_loss": 0.01384720392525196, "grad_norm": 0.0643564984202385, "global_step": 31337, "epoch": 295, "lr": 3.585702466375813e-05} +{"train_loss": 0.01326405256986618, "grad_norm": 0.15903350710868835, "global_step": 31338, "epoch": 295, "lr": 3.58541819581743e-05} +{"train_loss": 0.01522558182477951, "grad_norm": 0.114569753408432, "global_step": 31339, "epoch": 295, "lr": 3.585133930229275e-05} +{"train_loss": 0.012781171128153801, "grad_norm": 0.10680120438337326, "global_step": 31340, "epoch": 295, "lr": 3.5848496696123426e-05} +{"train_loss": 0.0138732073828578, "grad_norm": 0.08465930074453354, "global_step": 31341, "epoch": 295, "lr": 3.584565413967632e-05} +{"train_loss": 0.01830122247338295, "grad_norm": 0.10490085184574127, "global_step": 31342, "epoch": 295, "lr": 3.584281163296143e-05} +{"train_loss": 0.014742638915777206, "grad_norm": 0.0678393691778183, "global_step": 31343, "epoch": 295, "lr": 3.5839969175988745e-05} +{"train_loss": 0.017446059733629227, "grad_norm": 0.10277881473302841, "global_step": 31344, "epoch": 295, "lr": 3.5837126768768256e-05} +{"train_loss": 0.0151580311357975, "grad_norm": 0.13481198251247406, "global_step": 31345, "epoch": 295, "lr": 3.583428441130995e-05} +{"train_loss": 0.020117508247494698, "grad_norm": 0.15386617183685303, "global_step": 31346, "epoch": 295, "lr": 3.5831442103623794e-05} +{"train_loss": 0.014779873192310333, "grad_norm": 0.08716221898794174, "global_step": 31347, "epoch": 295, "lr": 3.582859984571979e-05} +{"train_loss": 0.014841757714748383, "grad_norm": 0.05954994633793831, "global_step": 31348, "epoch": 295, "lr": 3.582575763760793e-05} +{"train_loss": 0.016333678737282753, "grad_norm": 0.08319544047117233, "global_step": 31349, "epoch": 295, "lr": 3.582291547929818e-05} +{"train_loss": 0.016564827412366867, "grad_norm": 0.0970848947763443, "global_step": 31350, "epoch": 295, "lr": 3.582007337080055e-05} +{"train_loss": 0.014344600960612297, "grad_norm": 0.16339761018753052, "global_step": 31351, "epoch": 295, "lr": 3.5817231312125025e-05} +{"train_loss": 0.01642427034676075, "grad_norm": 0.11065950989723206, "global_step": 31352, "epoch": 295, "lr": 3.5814389303281574e-05} +{"train_loss": 0.01475664135068655, "grad_norm": 0.05934838578104973, "global_step": 31353, "epoch": 295, "lr": 3.581154734428019e-05} +{"train_loss": 0.01648261398077011, "grad_norm": 0.09405829757452011, "global_step": 31354, "epoch": 295, "lr": 3.580870543513084e-05} +{"train_loss": 0.01710129901766777, "grad_norm": 0.07159897685050964, "global_step": 31355, "epoch": 295, "lr": 3.5805863575843546e-05} +{"train_loss": 0.015828747302293777, "grad_norm": 0.0710427314043045, "global_step": 31356, "epoch": 295, "lr": 3.580302176642828e-05} +{"train_loss": 0.01524716429412365, "grad_norm": 0.08701307326555252, "global_step": 31357, "epoch": 295, "lr": 3.5800180006895026e-05} +{"train_loss": 0.017127161845564842, "grad_norm": 0.1753728985786438, "global_step": 31358, "epoch": 295, "lr": 3.5797338297253745e-05} +{"train_loss": 0.014722481369972229, "grad_norm": 0.06411628425121307, "global_step": 31359, "epoch": 295, "lr": 3.579449663751444e-05} +{"train_loss": 0.015800174325704575, "grad_norm": 0.08281698822975159, "global_step": 31360, "epoch": 295, "lr": 3.57916550276871e-05} +{"train_loss": 0.014405920170247555, "grad_norm": 0.05633961781859398, "global_step": 31361, "epoch": 295, "lr": 3.578881346778171e-05} +{"train_loss": 0.013272859156131744, "grad_norm": 0.11205407977104187, "global_step": 31362, "epoch": 295, "lr": 3.578597195780826e-05} +{"train_loss": 0.013431891798973083, "grad_norm": 0.05333373695611954, "global_step": 31363, "epoch": 295, "lr": 3.57831304977767e-05} +{"train_loss": 0.016336845234036446, "grad_norm": 0.06205934286117554, "global_step": 31364, "epoch": 295, "lr": 3.578028908769705e-05} +{"train_loss": 0.013615002855658531, "grad_norm": 0.05966082587838173, "global_step": 31365, "epoch": 295, "lr": 3.5777447727579275e-05} +{"train_loss": 0.015400066040456295, "grad_norm": 0.09007568657398224, "global_step": 31366, "epoch": 295, "lr": 3.577460641743336e-05} +{"train_loss": 0.01741020195186138, "grad_norm": 0.10333754867315292, "global_step": 31367, "epoch": 295, "lr": 3.57717651572693e-05} +{"train_loss": 0.014260290190577507, "grad_norm": 0.053318463265895844, "global_step": 31368, "epoch": 295, "lr": 3.5768923947097075e-05} +{"train_loss": 0.014777831733226776, "grad_norm": 0.07965926826000214, "global_step": 31369, "epoch": 295, "lr": 3.5766082786926656e-05} +{"train_loss": 0.015550453215837479, "grad_norm": 0.0524553507566452, "global_step": 31370, "epoch": 295, "lr": 3.576324167676803e-05} +{"train_loss": 0.01399858109652996, "grad_norm": 0.12674598395824432, "global_step": 31371, "epoch": 295, "lr": 3.5760400616631174e-05} +{"train_loss": 0.01563737541437149, "grad_norm": 0.08248141407966614, "global_step": 31372, "epoch": 295, "lr": 3.575755960652608e-05} +{"train_loss": 0.01581525430083275, "grad_norm": 0.1010831966996193, "global_step": 31373, "epoch": 295, "lr": 3.5754718646462736e-05} +{"train_loss": 0.013647053390741348, "grad_norm": 0.08291603624820709, "global_step": 31374, "epoch": 295, "lr": 3.5751877736451125e-05} +{"train_loss": 0.01531903161052263, "grad_norm": 0.09078270196914673, "global_step": 31375, "epoch": 295, "lr": 3.5749036876501194e-05, "val_loss": 0.01947309821844101} +{"train_loss": 0.016512326896190643, "grad_norm": 0.09550663828849792, "global_step": 31376, "epoch": 296, "lr": 3.5746196066622966e-05} +{"train_loss": 0.017487559467554092, "grad_norm": 0.07243788242340088, "global_step": 31377, "epoch": 296, "lr": 3.574335530682641e-05} +{"train_loss": 0.0172116719186306, "grad_norm": 0.09557165205478668, "global_step": 31378, "epoch": 296, "lr": 3.574051459712148e-05} +{"train_loss": 0.01300114206969738, "grad_norm": 0.13140738010406494, "global_step": 31379, "epoch": 296, "lr": 3.57376739375182e-05} +{"train_loss": 0.015513402409851551, "grad_norm": 0.08469167351722717, "global_step": 31380, "epoch": 296, "lr": 3.573483332802654e-05} +{"train_loss": 0.01513537485152483, "grad_norm": 0.10307104885578156, "global_step": 31381, "epoch": 296, "lr": 3.573199276865646e-05} +{"train_loss": 0.017555058002471924, "grad_norm": 0.06265593320131302, "global_step": 31382, "epoch": 296, "lr": 3.5729152259417954e-05} +{"train_loss": 0.016223574057221413, "grad_norm": 0.0660400539636612, "global_step": 31383, "epoch": 296, "lr": 3.572631180032099e-05} +{"train_loss": 0.015924308449029922, "grad_norm": 0.13019219040870667, "global_step": 31384, "epoch": 296, "lr": 3.572347139137556e-05} +{"train_loss": 0.015558814629912376, "grad_norm": 0.13485169410705566, "global_step": 31385, "epoch": 296, "lr": 3.5720631032591656e-05} +{"train_loss": 0.01726590096950531, "grad_norm": 0.0964304581284523, "global_step": 31386, "epoch": 296, "lr": 3.571779072397925e-05} +{"train_loss": 0.01831957697868347, "grad_norm": 0.10617959499359131, "global_step": 31387, "epoch": 296, "lr": 3.571495046554831e-05} +{"train_loss": 0.017078159376978874, "grad_norm": 0.05491090938448906, "global_step": 31388, "epoch": 296, "lr": 3.57121102573088e-05} +{"train_loss": 0.014385351911187172, "grad_norm": 0.06187622621655464, "global_step": 31389, "epoch": 296, "lr": 3.570927009927074e-05} +{"train_loss": 0.013225733302533627, "grad_norm": 0.1129649206995964, "global_step": 31390, "epoch": 296, "lr": 3.570642999144408e-05} +{"train_loss": 0.01611410081386566, "grad_norm": 0.07406102120876312, "global_step": 31391, "epoch": 296, "lr": 3.570358993383882e-05} +{"train_loss": 0.01452702097594738, "grad_norm": 0.06418658792972565, "global_step": 31392, "epoch": 296, "lr": 3.5700749926464935e-05} +{"train_loss": 0.014925455674529076, "grad_norm": 0.10606789588928223, "global_step": 31393, "epoch": 296, "lr": 3.569790996933239e-05} +{"train_loss": 0.013971935026347637, "grad_norm": 0.07442597299814224, "global_step": 31394, "epoch": 296, "lr": 3.5695070062451164e-05} +{"train_loss": 0.01565132662653923, "grad_norm": 0.10961355268955231, "global_step": 31395, "epoch": 296, "lr": 3.5692230205831235e-05} +{"train_loss": 0.01405324600636959, "grad_norm": 0.10538506507873535, "global_step": 31396, "epoch": 296, "lr": 3.5689390399482595e-05} +{"train_loss": 0.015532628633081913, "grad_norm": 0.18786481022834778, "global_step": 31397, "epoch": 296, "lr": 3.568655064341521e-05} +{"train_loss": 0.01284340862184763, "grad_norm": 0.0789230689406395, "global_step": 31398, "epoch": 296, "lr": 3.568371093763908e-05} +{"train_loss": 0.01673738844692707, "grad_norm": 0.08857312798500061, "global_step": 31399, "epoch": 296, "lr": 3.5680871282164144e-05} +{"train_loss": 0.014307960867881775, "grad_norm": 0.0903007760643959, "global_step": 31400, "epoch": 296, "lr": 3.567803167700039e-05} +{"train_loss": 0.015112973749637604, "grad_norm": 0.07629800587892532, "global_step": 31401, "epoch": 296, "lr": 3.5675192122157816e-05} +{"train_loss": 0.016082381829619408, "grad_norm": 0.07450113445520401, "global_step": 31402, "epoch": 296, "lr": 3.567235261764639e-05} +{"train_loss": 0.02015519328415394, "grad_norm": 0.15154729783535004, "global_step": 31403, "epoch": 296, "lr": 3.5669513163476075e-05} +{"train_loss": 0.01652003452181816, "grad_norm": 0.06701864302158356, "global_step": 31404, "epoch": 296, "lr": 3.566667375965688e-05} +{"train_loss": 0.01732044294476509, "grad_norm": 0.07801622897386551, "global_step": 31405, "epoch": 296, "lr": 3.566383440619875e-05} +{"train_loss": 0.016026873141527176, "grad_norm": 0.07525358349084854, "global_step": 31406, "epoch": 296, "lr": 3.5660995103111675e-05} +{"train_loss": 0.01626035012304783, "grad_norm": 0.09824446588754654, "global_step": 31407, "epoch": 296, "lr": 3.56581558504056e-05} +{"train_loss": 0.016582399606704712, "grad_norm": 0.07022063434123993, "global_step": 31408, "epoch": 296, "lr": 3.565531664809055e-05} +{"train_loss": 0.012771712616086006, "grad_norm": 0.16230618953704834, "global_step": 31409, "epoch": 296, "lr": 3.565247749617649e-05} +{"train_loss": 0.01456230878829956, "grad_norm": 0.06500129401683807, "global_step": 31410, "epoch": 296, "lr": 3.564963839467338e-05} +{"train_loss": 0.016102805733680725, "grad_norm": 0.11191106587648392, "global_step": 31411, "epoch": 296, "lr": 3.564679934359119e-05} +{"train_loss": 0.01397659257054329, "grad_norm": 0.11737117916345596, "global_step": 31412, "epoch": 296, "lr": 3.56439603429399e-05} +{"train_loss": 0.016232820227742195, "grad_norm": 0.12477476894855499, "global_step": 31413, "epoch": 296, "lr": 3.56411213927295e-05} +{"train_loss": 0.017645342275500298, "grad_norm": 0.06999124586582184, "global_step": 31414, "epoch": 296, "lr": 3.5638282492969955e-05} +{"train_loss": 0.01400272361934185, "grad_norm": 0.057686589658260345, "global_step": 31415, "epoch": 296, "lr": 3.5635443643671245e-05} +{"train_loss": 0.0129691231995821, "grad_norm": 0.06063501909375191, "global_step": 31416, "epoch": 296, "lr": 3.563260484484333e-05} +{"train_loss": 0.013903968967497349, "grad_norm": 0.08465971052646637, "global_step": 31417, "epoch": 296, "lr": 3.5629766096496185e-05} +{"train_loss": 0.01585238054394722, "grad_norm": 0.06531982123851776, "global_step": 31418, "epoch": 296, "lr": 3.562692739863981e-05} +{"train_loss": 0.014321675524115562, "grad_norm": 0.07956951856613159, "global_step": 31419, "epoch": 296, "lr": 3.5624088751284144e-05} +{"train_loss": 0.012862063013017178, "grad_norm": 0.11633152514696121, "global_step": 31420, "epoch": 296, "lr": 3.5621250154439186e-05} +{"train_loss": 0.01656418666243553, "grad_norm": 0.0929999127984047, "global_step": 31421, "epoch": 296, "lr": 3.561841160811491e-05} +{"train_loss": 0.01463787630200386, "grad_norm": 0.07370961457490921, "global_step": 31422, "epoch": 296, "lr": 3.5615573112321276e-05} +{"train_loss": 0.01734917238354683, "grad_norm": 0.08607282489538193, "global_step": 31423, "epoch": 296, "lr": 3.561273466706827e-05} +{"train_loss": 0.01407032459974289, "grad_norm": 0.1017489954829216, "global_step": 31424, "epoch": 296, "lr": 3.560989627236583e-05} +{"train_loss": 0.01817331463098526, "grad_norm": 0.1287631392478943, "global_step": 31425, "epoch": 296, "lr": 3.560705792822398e-05} +{"train_loss": 0.014678431674838066, "grad_norm": 0.12268538773059845, "global_step": 31426, "epoch": 296, "lr": 3.560421963465267e-05} +{"train_loss": 0.014608511701226234, "grad_norm": 0.09143971651792526, "global_step": 31427, "epoch": 296, "lr": 3.560138139166187e-05} +{"train_loss": 0.014031335711479187, "grad_norm": 0.06612291187047958, "global_step": 31428, "epoch": 296, "lr": 3.559854319926156e-05} +{"train_loss": 0.01617518626153469, "grad_norm": 0.0953393280506134, "global_step": 31429, "epoch": 296, "lr": 3.559570505746168e-05} +{"train_loss": 0.011989137157797813, "grad_norm": 0.06314723193645477, "global_step": 31430, "epoch": 296, "lr": 3.5592866966272245e-05} +{"train_loss": 0.013092244043946266, "grad_norm": 0.05741860717535019, "global_step": 31431, "epoch": 296, "lr": 3.559002892570321e-05} +{"train_loss": 0.01436598226428032, "grad_norm": 0.09838652610778809, "global_step": 31432, "epoch": 296, "lr": 3.558719093576454e-05} +{"train_loss": 0.017021752893924713, "grad_norm": 0.08231724798679352, "global_step": 31433, "epoch": 296, "lr": 3.558435299646623e-05} +{"train_loss": 0.01761864498257637, "grad_norm": 0.11430317163467407, "global_step": 31434, "epoch": 296, "lr": 3.558151510781822e-05} +{"train_loss": 0.015802020207047462, "grad_norm": 0.10511093586683273, "global_step": 31435, "epoch": 296, "lr": 3.55786772698305e-05} +{"train_loss": 0.013770999386906624, "grad_norm": 0.05707681179046631, "global_step": 31436, "epoch": 296, "lr": 3.557583948251303e-05} +{"train_loss": 0.014483179897069931, "grad_norm": 0.06685346364974976, "global_step": 31437, "epoch": 296, "lr": 3.557300174587579e-05} +{"train_loss": 0.012612244114279747, "grad_norm": 0.048690926283597946, "global_step": 31438, "epoch": 296, "lr": 3.5570164059928755e-05} +{"train_loss": 0.015118399634957314, "grad_norm": 0.06697622686624527, "global_step": 31439, "epoch": 296, "lr": 3.556732642468189e-05} +{"train_loss": 0.014634160324931145, "grad_norm": 0.0837692990899086, "global_step": 31440, "epoch": 296, "lr": 3.556448884014516e-05} +{"train_loss": 0.014758522622287273, "grad_norm": 0.06664738059043884, "global_step": 31441, "epoch": 296, "lr": 3.5561651306328525e-05} +{"train_loss": 0.015367265790700912, "grad_norm": 0.05444859713315964, "global_step": 31442, "epoch": 296, "lr": 3.555881382324198e-05} +{"train_loss": 0.015345023944973946, "grad_norm": 0.16309639811515808, "global_step": 31443, "epoch": 296, "lr": 3.5555976390895484e-05} +{"train_loss": 0.012557162903249264, "grad_norm": 0.07276185601949692, "global_step": 31444, "epoch": 296, "lr": 3.555313900929899e-05} +{"train_loss": 0.01632378250360489, "grad_norm": 0.09292978793382645, "global_step": 31445, "epoch": 296, "lr": 3.555030167846251e-05} +{"train_loss": 0.015271052718162537, "grad_norm": 0.08277526497840881, "global_step": 31446, "epoch": 296, "lr": 3.554746439839597e-05} +{"train_loss": 0.016148164868354797, "grad_norm": 0.08893530815839767, "global_step": 31447, "epoch": 296, "lr": 3.5544627169109354e-05} +{"train_loss": 0.016293618828058243, "grad_norm": 0.08851342648267746, "global_step": 31448, "epoch": 296, "lr": 3.554178999061263e-05} +{"train_loss": 0.014298871159553528, "grad_norm": 0.08520247042179108, "global_step": 31449, "epoch": 296, "lr": 3.553895286291577e-05} +{"train_loss": 0.014453299343585968, "grad_norm": 0.060545049607753754, "global_step": 31450, "epoch": 296, "lr": 3.553611578602875e-05} +{"train_loss": 0.015804864466190338, "grad_norm": 0.07457057386636734, "global_step": 31451, "epoch": 296, "lr": 3.5533278759961525e-05} +{"train_loss": 0.01615692675113678, "grad_norm": 0.10597069561481476, "global_step": 31452, "epoch": 296, "lr": 3.5530441784724065e-05} +{"train_loss": 0.014309436082839966, "grad_norm": 0.0755419135093689, "global_step": 31453, "epoch": 296, "lr": 3.552760486032633e-05} +{"train_loss": 0.015137141570448875, "grad_norm": 0.07925637066364288, "global_step": 31454, "epoch": 296, "lr": 3.5524767986778305e-05} +{"train_loss": 0.014028876088559628, "grad_norm": 0.09121433645486832, "global_step": 31455, "epoch": 296, "lr": 3.552193116408995e-05} +{"train_loss": 0.014013528823852539, "grad_norm": 0.0731259286403656, "global_step": 31456, "epoch": 296, "lr": 3.5519094392271245e-05} +{"train_loss": 0.015788167715072632, "grad_norm": 0.088486447930336, "global_step": 31457, "epoch": 296, "lr": 3.5516257671332144e-05} +{"train_loss": 0.014551326632499695, "grad_norm": 0.08105539530515671, "global_step": 31458, "epoch": 296, "lr": 3.5513421001282586e-05} +{"train_loss": 0.018703661859035492, "grad_norm": 0.07516792416572571, "global_step": 31459, "epoch": 296, "lr": 3.5510584382132587e-05} +{"train_loss": 0.016958437860012054, "grad_norm": 0.0737157016992569, "global_step": 31460, "epoch": 296, "lr": 3.550774781389209e-05} +{"train_loss": 0.012920886278152466, "grad_norm": 0.05896550789475441, "global_step": 31461, "epoch": 296, "lr": 3.5504911296571066e-05} +{"train_loss": 0.01708027720451355, "grad_norm": 0.15443451702594757, "global_step": 31462, "epoch": 296, "lr": 3.550207483017949e-05} +{"train_loss": 0.015904100611805916, "grad_norm": 0.0574209950864315, "global_step": 31463, "epoch": 296, "lr": 3.5499238414727305e-05} +{"train_loss": 0.01477176882326603, "grad_norm": 0.12234819680452347, "global_step": 31464, "epoch": 296, "lr": 3.54964020502245e-05} +{"train_loss": 0.014470629394054413, "grad_norm": 0.08510669320821762, "global_step": 31465, "epoch": 296, "lr": 3.549356573668101e-05} +{"train_loss": 0.015599008649587631, "grad_norm": 0.08619873970746994, "global_step": 31466, "epoch": 296, "lr": 3.549072947410683e-05} +{"train_loss": 0.015220200642943382, "grad_norm": 0.07699482142925262, "global_step": 31467, "epoch": 296, "lr": 3.5487893262511926e-05} +{"train_loss": 0.01396854780614376, "grad_norm": 0.06444716453552246, "global_step": 31468, "epoch": 296, "lr": 3.548505710190626e-05} +{"train_loss": 0.014662416651844978, "grad_norm": 0.10021225363016129, "global_step": 31469, "epoch": 296, "lr": 3.548222099229979e-05} +{"train_loss": 0.015542054548859596, "grad_norm": 0.07998890429735184, "global_step": 31470, "epoch": 296, "lr": 3.547938493370245e-05} +{"train_loss": 0.014752974733710289, "grad_norm": 0.08780062198638916, "global_step": 31471, "epoch": 296, "lr": 3.5476548926124265e-05} +{"train_loss": 0.01548117958009243, "grad_norm": 0.09204267710447311, "global_step": 31472, "epoch": 296, "lr": 3.547371296957517e-05} +{"train_loss": 0.014817027375102043, "grad_norm": 0.06765340268611908, "global_step": 31473, "epoch": 296, "lr": 3.547087706406511e-05} +{"train_loss": 0.014902903698384762, "grad_norm": 0.13969141244888306, "global_step": 31474, "epoch": 296, "lr": 3.5468041209604095e-05} +{"train_loss": 0.017569638788700104, "grad_norm": 0.08209621906280518, "global_step": 31475, "epoch": 296, "lr": 3.546520540620206e-05} +{"train_loss": 0.014962336048483849, "grad_norm": 0.09340521693229675, "global_step": 31476, "epoch": 296, "lr": 3.5462369653868964e-05} +{"train_loss": 0.017715424299240112, "grad_norm": 0.13298344612121582, "global_step": 31477, "epoch": 296, "lr": 3.545953395261476e-05} +{"train_loss": 0.014735493808984756, "grad_norm": 0.14375126361846924, "global_step": 31478, "epoch": 296, "lr": 3.545669830244946e-05} +{"train_loss": 0.014480342157185078, "grad_norm": 0.08857472240924835, "global_step": 31479, "epoch": 296, "lr": 3.545386270338299e-05} +{"train_loss": 0.013130348175764084, "grad_norm": 0.09345119446516037, "global_step": 31480, "epoch": 296, "lr": 3.545102715542533e-05} +{"train_loss": 0.015305034561469307, "grad_norm": 0.10148311406373978, "global_step": 31481, "epoch": 296, "lr": 3.544819165858642e-05, "val_loss": 0.019868802279233932} +{"train_loss": 0.01580982282757759, "grad_norm": 0.059162359684705734, "global_step": 31482, "epoch": 297, "lr": 3.544535621287623e-05} +{"train_loss": 0.015831442549824715, "grad_norm": 0.08617281168699265, "global_step": 31483, "epoch": 297, "lr": 3.5442520818304733e-05} +{"train_loss": 0.01232130266726017, "grad_norm": 0.05762385204434395, "global_step": 31484, "epoch": 297, "lr": 3.54396854748819e-05} +{"train_loss": 0.011420577764511108, "grad_norm": 0.06778816133737564, "global_step": 31485, "epoch": 297, "lr": 3.543685018261766e-05} +{"train_loss": 0.015492820180952549, "grad_norm": 0.12133621424436569, "global_step": 31486, "epoch": 297, "lr": 3.543401494152202e-05} +{"train_loss": 0.018166637048125267, "grad_norm": 0.09011431783437729, "global_step": 31487, "epoch": 297, "lr": 3.543117975160489e-05} +{"train_loss": 0.014621934853494167, "grad_norm": 0.08403845131397247, "global_step": 31488, "epoch": 297, "lr": 3.5428344612876274e-05} +{"train_loss": 0.015037763863801956, "grad_norm": 0.11158013343811035, "global_step": 31489, "epoch": 297, "lr": 3.542550952534612e-05} +{"train_loss": 0.015405004844069481, "grad_norm": 0.09380098432302475, "global_step": 31490, "epoch": 297, "lr": 3.542267448902437e-05} +{"train_loss": 0.016768570989370346, "grad_norm": 0.08856247365474701, "global_step": 31491, "epoch": 297, "lr": 3.541983950392102e-05} +{"train_loss": 0.013364169746637344, "grad_norm": 0.09542638063430786, "global_step": 31492, "epoch": 297, "lr": 3.541700457004602e-05} +{"train_loss": 0.014070102944970131, "grad_norm": 0.12976400554180145, "global_step": 31493, "epoch": 297, "lr": 3.54141696874093e-05} +{"train_loss": 0.014096998609602451, "grad_norm": 0.09601738303899765, "global_step": 31494, "epoch": 297, "lr": 3.541133485602086e-05} +{"train_loss": 0.014787642285227776, "grad_norm": 0.06703266501426697, "global_step": 31495, "epoch": 297, "lr": 3.540850007589064e-05} +{"train_loss": 0.016388099640607834, "grad_norm": 0.10252423584461212, "global_step": 31496, "epoch": 297, "lr": 3.54056653470286e-05} +{"train_loss": 0.015063818544149399, "grad_norm": 0.11366964131593704, "global_step": 31497, "epoch": 297, "lr": 3.540283066944471e-05} +{"train_loss": 0.01677444577217102, "grad_norm": 0.07170146703720093, "global_step": 31498, "epoch": 297, "lr": 3.5399996043148945e-05} +{"train_loss": 0.017009975388646126, "grad_norm": 0.09038608521223068, "global_step": 31499, "epoch": 297, "lr": 3.539716146815122e-05} +{"train_loss": 0.016757136210799217, "grad_norm": 0.08842708170413971, "global_step": 31500, "epoch": 297, "lr": 3.539432694446152e-05} +{"train_loss": 0.015292615629732609, "grad_norm": 0.10304185748100281, "global_step": 31501, "epoch": 297, "lr": 3.539149247208981e-05} +{"train_loss": 0.016689680516719818, "grad_norm": 0.0792197436094284, "global_step": 31502, "epoch": 297, "lr": 3.538865805104603e-05} +{"train_loss": 0.015372885391116142, "grad_norm": 0.09178687632083893, "global_step": 31503, "epoch": 297, "lr": 3.5385823681340156e-05} +{"train_loss": 0.014312174171209335, "grad_norm": 0.08551973104476929, "global_step": 31504, "epoch": 297, "lr": 3.5382989362982166e-05} +{"train_loss": 0.015409750863909721, "grad_norm": 0.12549716234207153, "global_step": 31505, "epoch": 297, "lr": 3.538015509598197e-05} +{"train_loss": 0.01727779023349285, "grad_norm": 0.0744692012667656, "global_step": 31506, "epoch": 297, "lr": 3.537732088034955e-05} +{"train_loss": 0.014210065826773643, "grad_norm": 0.06300849467515945, "global_step": 31507, "epoch": 297, "lr": 3.537448671609488e-05} +{"train_loss": 0.014448653906583786, "grad_norm": 0.06771985441446304, "global_step": 31508, "epoch": 297, "lr": 3.53716526032279e-05} +{"train_loss": 0.01707230508327484, "grad_norm": 0.09192728996276855, "global_step": 31509, "epoch": 297, "lr": 3.5368818541758575e-05} +{"train_loss": 0.014573088847100735, "grad_norm": 0.06286151707172394, "global_step": 31510, "epoch": 297, "lr": 3.536598453169685e-05} +{"train_loss": 0.017884671688079834, "grad_norm": 0.12098250538110733, "global_step": 31511, "epoch": 297, "lr": 3.5363150573052684e-05} +{"train_loss": 0.019528010860085487, "grad_norm": 0.1064780130982399, "global_step": 31512, "epoch": 297, "lr": 3.5360316665836043e-05} +{"train_loss": 0.014159264974296093, "grad_norm": 0.05619552731513977, "global_step": 31513, "epoch": 297, "lr": 3.5357482810056895e-05} +{"train_loss": 0.015734268352389336, "grad_norm": 0.09770924597978592, "global_step": 31514, "epoch": 297, "lr": 3.5354649005725166e-05} +{"train_loss": 0.015263580717146397, "grad_norm": 0.0858892947435379, "global_step": 31515, "epoch": 297, "lr": 3.5351815252850865e-05} +{"train_loss": 0.014461223036050797, "grad_norm": 0.1016758605837822, "global_step": 31516, "epoch": 297, "lr": 3.5348981551443885e-05} +{"train_loss": 0.0154017498716712, "grad_norm": 0.13869057595729828, "global_step": 31517, "epoch": 297, "lr": 3.534614790151422e-05} +{"train_loss": 0.016261979937553406, "grad_norm": 0.07016397267580032, "global_step": 31518, "epoch": 297, "lr": 3.534331430307182e-05} +{"train_loss": 0.015921883285045624, "grad_norm": 0.09413205832242966, "global_step": 31519, "epoch": 297, "lr": 3.5340480756126624e-05} +{"train_loss": 0.017213866114616394, "grad_norm": 0.0727173462510109, "global_step": 31520, "epoch": 297, "lr": 3.533764726068862e-05} +{"train_loss": 0.016916770488023758, "grad_norm": 0.05815599113702774, "global_step": 31521, "epoch": 297, "lr": 3.5334813816767764e-05} +{"train_loss": 0.014807427302002907, "grad_norm": 0.1778607964515686, "global_step": 31522, "epoch": 297, "lr": 3.5331980424373974e-05} +{"train_loss": 0.015068922191858292, "grad_norm": 0.09404764324426651, "global_step": 31523, "epoch": 297, "lr": 3.5329147083517214e-05} +{"train_loss": 0.015848152339458466, "grad_norm": 0.0978606566786766, "global_step": 31524, "epoch": 297, "lr": 3.532631379420746e-05} +{"train_loss": 0.01377144455909729, "grad_norm": 0.0557587631046772, "global_step": 31525, "epoch": 297, "lr": 3.532348055645467e-05} +{"train_loss": 0.013967493548989296, "grad_norm": 0.07132036238908768, "global_step": 31526, "epoch": 297, "lr": 3.5320647370268775e-05} +{"train_loss": 0.014523925259709358, "grad_norm": 0.06492532789707184, "global_step": 31527, "epoch": 297, "lr": 3.5317814235659766e-05} +{"train_loss": 0.013790598139166832, "grad_norm": 0.07864680886268616, "global_step": 31528, "epoch": 297, "lr": 3.531498115263754e-05} +{"train_loss": 0.016209721565246582, "grad_norm": 0.10664308071136475, "global_step": 31529, "epoch": 297, "lr": 3.53121481212121e-05} +{"train_loss": 0.014521939679980278, "grad_norm": 0.06661100685596466, "global_step": 31530, "epoch": 297, "lr": 3.530931514139338e-05} +{"train_loss": 0.01634056121110916, "grad_norm": 0.06517113745212555, "global_step": 31531, "epoch": 297, "lr": 3.530648221319133e-05} +{"train_loss": 0.012492980808019638, "grad_norm": 0.05581163242459297, "global_step": 31532, "epoch": 297, "lr": 3.530364933661592e-05} +{"train_loss": 0.014387187547981739, "grad_norm": 0.09569253027439117, "global_step": 31533, "epoch": 297, "lr": 3.53008165116771e-05} +{"train_loss": 0.01564219780266285, "grad_norm": 0.08213459700345993, "global_step": 31534, "epoch": 297, "lr": 3.529798373838481e-05} +{"train_loss": 0.012456159107387066, "grad_norm": 0.07752720266580582, "global_step": 31535, "epoch": 297, "lr": 3.5295151016749006e-05} +{"train_loss": 0.016759011894464493, "grad_norm": 0.12835121154785156, "global_step": 31536, "epoch": 297, "lr": 3.5292318346779656e-05} +{"train_loss": 0.01735779456794262, "grad_norm": 0.11978428065776825, "global_step": 31537, "epoch": 297, "lr": 3.52894857284867e-05} +{"train_loss": 0.014295072294771671, "grad_norm": 0.09519930928945541, "global_step": 31538, "epoch": 297, "lr": 3.528665316188009e-05} +{"train_loss": 0.014965826645493507, "grad_norm": 0.0618472695350647, "global_step": 31539, "epoch": 297, "lr": 3.528382064696981e-05} +{"train_loss": 0.015111373737454414, "grad_norm": 0.09612647444009781, "global_step": 31540, "epoch": 297, "lr": 3.528098818376575e-05} +{"train_loss": 0.014999532140791416, "grad_norm": 0.16018745303153992, "global_step": 31541, "epoch": 297, "lr": 3.527815577227791e-05} +{"train_loss": 0.014839475974440575, "grad_norm": 0.07518571615219116, "global_step": 31542, "epoch": 297, "lr": 3.5275323412516227e-05} +{"train_loss": 0.015708506107330322, "grad_norm": 0.08427818864583969, "global_step": 31543, "epoch": 297, "lr": 3.527249110449064e-05} +{"train_loss": 0.014337141998112202, "grad_norm": 0.1300572156906128, "global_step": 31544, "epoch": 297, "lr": 3.526965884821113e-05} +{"train_loss": 0.01388806477189064, "grad_norm": 0.07111120969057083, "global_step": 31545, "epoch": 297, "lr": 3.526682664368764e-05} +{"train_loss": 0.014352722093462944, "grad_norm": 0.07324336469173431, "global_step": 31546, "epoch": 297, "lr": 3.526399449093011e-05} +{"train_loss": 0.016675684601068497, "grad_norm": 0.06378886848688126, "global_step": 31547, "epoch": 297, "lr": 3.526116238994848e-05} +{"train_loss": 0.018336515873670578, "grad_norm": 0.0864873006939888, "global_step": 31548, "epoch": 297, "lr": 3.525833034075272e-05} +{"train_loss": 0.014277918264269829, "grad_norm": 0.05996587127447128, "global_step": 31549, "epoch": 297, "lr": 3.525549834335278e-05} +{"train_loss": 0.0180504210293293, "grad_norm": 0.07620507478713989, "global_step": 31550, "epoch": 297, "lr": 3.525266639775859e-05} +{"train_loss": 0.014974843710660934, "grad_norm": 0.1280822902917862, "global_step": 31551, "epoch": 297, "lr": 3.524983450398016e-05} +{"train_loss": 0.013595694676041603, "grad_norm": 0.13333332538604736, "global_step": 31552, "epoch": 297, "lr": 3.524700266202736e-05} +{"train_loss": 0.015427153557538986, "grad_norm": 0.21710321307182312, "global_step": 31553, "epoch": 297, "lr": 3.524417087191019e-05} +{"train_loss": 0.014687999151647091, "grad_norm": 0.07535284757614136, "global_step": 31554, "epoch": 297, "lr": 3.524133913363859e-05} +{"train_loss": 0.012131409719586372, "grad_norm": 0.19886842370033264, "global_step": 31555, "epoch": 297, "lr": 3.523850744722249e-05} +{"train_loss": 0.017650648951530457, "grad_norm": 0.12375246733427048, "global_step": 31556, "epoch": 297, "lr": 3.523567581267187e-05} +{"train_loss": 0.013773096725344658, "grad_norm": 0.1048145741224289, "global_step": 31557, "epoch": 297, "lr": 3.5232844229996665e-05} +{"train_loss": 0.014674571342766285, "grad_norm": 0.12611758708953857, "global_step": 31558, "epoch": 297, "lr": 3.523001269920682e-05} +{"train_loss": 0.017017940059304237, "grad_norm": 0.06561548262834549, "global_step": 31559, "epoch": 297, "lr": 3.522718122031229e-05} +{"train_loss": 0.012922088615596294, "grad_norm": 0.0685785710811615, "global_step": 31560, "epoch": 297, "lr": 3.5224349793322996e-05} +{"train_loss": 0.01473880372941494, "grad_norm": 0.0998942032456398, "global_step": 31561, "epoch": 297, "lr": 3.5221518418248936e-05} +{"train_loss": 0.01640324667096138, "grad_norm": 0.07145418971776962, "global_step": 31562, "epoch": 297, "lr": 3.521868709510004e-05} +{"train_loss": 0.014654609374701977, "grad_norm": 0.07527664303779602, "global_step": 31563, "epoch": 297, "lr": 3.5215855823886233e-05} +{"train_loss": 0.015575862489640713, "grad_norm": 0.06242331489920616, "global_step": 31564, "epoch": 297, "lr": 3.521302460461748e-05} +{"train_loss": 0.01553697232156992, "grad_norm": 0.06788700073957443, "global_step": 31565, "epoch": 297, "lr": 3.521019343730373e-05} +{"train_loss": 0.015861772000789642, "grad_norm": 0.09736190736293793, "global_step": 31566, "epoch": 297, "lr": 3.5207362321954927e-05} +{"train_loss": 0.013476693071424961, "grad_norm": 0.10201847553253174, "global_step": 31567, "epoch": 297, "lr": 3.520453125858102e-05} +{"train_loss": 0.016674119979143143, "grad_norm": 0.10408332943916321, "global_step": 31568, "epoch": 297, "lr": 3.5201700247191975e-05} +{"train_loss": 0.01587069407105446, "grad_norm": 0.07994431257247925, "global_step": 31569, "epoch": 297, "lr": 3.51988692877977e-05} +{"train_loss": 0.016047433018684387, "grad_norm": 0.10187485814094543, "global_step": 31570, "epoch": 297, "lr": 3.519603838040816e-05} +{"train_loss": 0.016823146492242813, "grad_norm": 0.15798979997634888, "global_step": 31571, "epoch": 297, "lr": 3.519320752503331e-05} +{"train_loss": 0.015011126175522804, "grad_norm": 0.15352928638458252, "global_step": 31572, "epoch": 297, "lr": 3.519037672168308e-05} +{"train_loss": 0.01514686644077301, "grad_norm": 0.10864505171775818, "global_step": 31573, "epoch": 297, "lr": 3.518754597036743e-05} +{"train_loss": 0.014325106516480446, "grad_norm": 0.09688634425401688, "global_step": 31574, "epoch": 297, "lr": 3.518471527109631e-05} +{"train_loss": 0.01598614826798439, "grad_norm": 0.07234685868024826, "global_step": 31575, "epoch": 297, "lr": 3.5181884623879656e-05} +{"train_loss": 0.01668509468436241, "grad_norm": 0.09366729855537415, "global_step": 31576, "epoch": 297, "lr": 3.5179054028727404e-05} +{"train_loss": 0.013967696577310562, "grad_norm": 0.11926832795143127, "global_step": 31577, "epoch": 297, "lr": 3.517622348564952e-05} +{"train_loss": 0.01607910357415676, "grad_norm": 0.1766643226146698, "global_step": 31578, "epoch": 297, "lr": 3.517339299465593e-05} +{"train_loss": 0.014539606869220734, "grad_norm": 0.09049981832504272, "global_step": 31579, "epoch": 297, "lr": 3.517056255575659e-05} +{"train_loss": 0.015764521434903145, "grad_norm": 0.08011813461780548, "global_step": 31580, "epoch": 297, "lr": 3.516773216896148e-05} +{"train_loss": 0.013984281569719315, "grad_norm": 0.08175372332334518, "global_step": 31581, "epoch": 297, "lr": 3.516490183428046e-05} +{"train_loss": 0.015519142150878906, "grad_norm": 0.10626599937677383, "global_step": 31582, "epoch": 297, "lr": 3.516207155172355e-05} +{"train_loss": 0.015940677374601364, "grad_norm": 0.09604257345199585, "global_step": 31583, "epoch": 297, "lr": 3.5159241321300664e-05} +{"train_loss": 0.015630580484867096, "grad_norm": 0.11339103430509567, "global_step": 31584, "epoch": 297, "lr": 3.5156411143021735e-05} +{"train_loss": 0.012756161391735077, "grad_norm": 0.08381592482328415, "global_step": 31585, "epoch": 297, "lr": 3.515358101689673e-05} +{"train_loss": 0.015830954536795616, "grad_norm": 0.10752711445093155, "global_step": 31586, "epoch": 297, "lr": 3.515075094293561e-05} +{"train_loss": 0.015313314519963175, "grad_norm": 0.10817353427410126, "global_step": 31587, "epoch": 297, "lr": 3.5147920921148267e-05, "val_loss": 0.018984707072377205} +{"train_loss": 0.01573154143989086, "grad_norm": 0.06168507784605026, "global_step": 31588, "epoch": 298, "lr": 3.514509095154467e-05} +{"train_loss": 0.015229922719299793, "grad_norm": 0.17086000740528107, "global_step": 31589, "epoch": 298, "lr": 3.514226103413476e-05} +{"train_loss": 0.018085528165102005, "grad_norm": 0.08945304155349731, "global_step": 31590, "epoch": 298, "lr": 3.513943116892849e-05} +{"train_loss": 0.014255429618060589, "grad_norm": 0.11232341825962067, "global_step": 31591, "epoch": 298, "lr": 3.513660135593581e-05} +{"train_loss": 0.015532242134213448, "grad_norm": 0.16605184972286224, "global_step": 31592, "epoch": 298, "lr": 3.513377159516664e-05} +{"train_loss": 0.015842380002141, "grad_norm": 0.1526072919368744, "global_step": 31593, "epoch": 298, "lr": 3.513094188663092e-05} +{"train_loss": 0.013512413017451763, "grad_norm": 0.177851140499115, "global_step": 31594, "epoch": 298, "lr": 3.5128112230338615e-05} +{"train_loss": 0.014863410964608192, "grad_norm": 0.15443575382232666, "global_step": 31595, "epoch": 298, "lr": 3.5125282626299656e-05} +{"train_loss": 0.012898927554488182, "grad_norm": 0.07003174722194672, "global_step": 31596, "epoch": 298, "lr": 3.512245307452397e-05} +{"train_loss": 0.017563022673130035, "grad_norm": 0.15290392935276031, "global_step": 31597, "epoch": 298, "lr": 3.5119623575021535e-05} +{"train_loss": 0.014472641050815582, "grad_norm": 0.08429745584726334, "global_step": 31598, "epoch": 298, "lr": 3.511679412780227e-05} +{"train_loss": 0.01480294018983841, "grad_norm": 0.13081152737140656, "global_step": 31599, "epoch": 298, "lr": 3.5113964732876106e-05} +{"train_loss": 0.012321775779128075, "grad_norm": 0.0856744572520256, "global_step": 31600, "epoch": 298, "lr": 3.5111135390253e-05} +{"train_loss": 0.015729228034615517, "grad_norm": 0.11252754181623459, "global_step": 31601, "epoch": 298, "lr": 3.510830609994288e-05} +{"train_loss": 0.012297652661800385, "grad_norm": 0.078766830265522, "global_step": 31602, "epoch": 298, "lr": 3.510547686195571e-05} +{"train_loss": 0.016846997663378716, "grad_norm": 0.07872792333364487, "global_step": 31603, "epoch": 298, "lr": 3.5102647676301416e-05} +{"train_loss": 0.014317006804049015, "grad_norm": 0.0754740759730339, "global_step": 31604, "epoch": 298, "lr": 3.509981854298995e-05} +{"train_loss": 0.017515547573566437, "grad_norm": 0.1662348359823227, "global_step": 31605, "epoch": 298, "lr": 3.5096989462031214e-05} +{"train_loss": 0.015570184215903282, "grad_norm": 0.07047143578529358, "global_step": 31606, "epoch": 298, "lr": 3.509416043343519e-05} +{"train_loss": 0.013478654436767101, "grad_norm": 0.0675385594367981, "global_step": 31607, "epoch": 298, "lr": 3.509133145721181e-05} +{"train_loss": 0.014765319414436817, "grad_norm": 0.07021298259496689, "global_step": 31608, "epoch": 298, "lr": 3.508850253337099e-05} +{"train_loss": 0.014958704821765423, "grad_norm": 0.12155430763959885, "global_step": 31609, "epoch": 298, "lr": 3.508567366192272e-05} +{"train_loss": 0.017114652320742607, "grad_norm": 0.12305569648742676, "global_step": 31610, "epoch": 298, "lr": 3.508284484287687e-05} +{"train_loss": 0.013519495725631714, "grad_norm": 0.1643909215927124, "global_step": 31611, "epoch": 298, "lr": 3.508001607624344e-05} +{"train_loss": 0.013053378090262413, "grad_norm": 0.08098427206277847, "global_step": 31612, "epoch": 298, "lr": 3.507718736203233e-05} +{"train_loss": 0.01578841730952263, "grad_norm": 0.07436853647232056, "global_step": 31613, "epoch": 298, "lr": 3.50743587002535e-05} +{"train_loss": 0.01530422456562519, "grad_norm": 0.1454007625579834, "global_step": 31614, "epoch": 298, "lr": 3.507153009091688e-05} +{"train_loss": 0.01355942152440548, "grad_norm": 0.10160142183303833, "global_step": 31615, "epoch": 298, "lr": 3.506870153403242e-05} +{"train_loss": 0.013814297504723072, "grad_norm": 0.07419871538877487, "global_step": 31616, "epoch": 298, "lr": 3.506587302961005e-05} +{"train_loss": 0.013702213764190674, "grad_norm": 0.08647437393665314, "global_step": 31617, "epoch": 298, "lr": 3.50630445776597e-05} +{"train_loss": 0.014649231918156147, "grad_norm": 0.05544643849134445, "global_step": 31618, "epoch": 298, "lr": 3.506021617819131e-05} +{"train_loss": 0.015874100849032402, "grad_norm": 0.07945756614208221, "global_step": 31619, "epoch": 298, "lr": 3.505738783121483e-05} +{"train_loss": 0.014276761561632156, "grad_norm": 0.09460902214050293, "global_step": 31620, "epoch": 298, "lr": 3.5054559536740183e-05} +{"train_loss": 0.014223297126591206, "grad_norm": 0.06379038095474243, "global_step": 31621, "epoch": 298, "lr": 3.505173129477733e-05} +{"train_loss": 0.017001356929540634, "grad_norm": 0.09227431565523148, "global_step": 31622, "epoch": 298, "lr": 3.504890310533618e-05} +{"train_loss": 0.01515962928533554, "grad_norm": 0.060691092163324356, "global_step": 31623, "epoch": 298, "lr": 3.5046074968426686e-05} +{"train_loss": 0.014917455613613129, "grad_norm": 0.1440214216709137, "global_step": 31624, "epoch": 298, "lr": 3.504324688405878e-05} +{"train_loss": 0.014077526517212391, "grad_norm": 0.07940001040697098, "global_step": 31625, "epoch": 298, "lr": 3.5040418852242396e-05} +{"train_loss": 0.015166993252933025, "grad_norm": 0.08919911831617355, "global_step": 31626, "epoch": 298, "lr": 3.503759087298748e-05} +{"train_loss": 0.013902435079216957, "grad_norm": 0.07311813533306122, "global_step": 31627, "epoch": 298, "lr": 3.503476294630397e-05} +{"train_loss": 0.016218863427639008, "grad_norm": 0.06535502523183823, "global_step": 31628, "epoch": 298, "lr": 3.503193507220179e-05} +{"train_loss": 0.01563623920083046, "grad_norm": 0.07049001008272171, "global_step": 31629, "epoch": 298, "lr": 3.502910725069087e-05} +{"train_loss": 0.014671583659946918, "grad_norm": 0.10307912528514862, "global_step": 31630, "epoch": 298, "lr": 3.502627948178116e-05} +{"train_loss": 0.013069536536931992, "grad_norm": 0.0898197814822197, "global_step": 31631, "epoch": 298, "lr": 3.5023451765482594e-05} +{"train_loss": 0.014544647186994553, "grad_norm": 0.10608141124248505, "global_step": 31632, "epoch": 298, "lr": 3.5020624101805114e-05} +{"train_loss": 0.017069362103939056, "grad_norm": 0.1795167475938797, "global_step": 31633, "epoch": 298, "lr": 3.501779649075865e-05} +{"train_loss": 0.017536822706460953, "grad_norm": 0.11997896432876587, "global_step": 31634, "epoch": 298, "lr": 3.5014968932353114e-05} +{"train_loss": 0.014977344311773777, "grad_norm": 0.17659978568553925, "global_step": 31635, "epoch": 298, "lr": 3.501214142659847e-05} +{"train_loss": 0.015944965183734894, "grad_norm": 0.09964116662740707, "global_step": 31636, "epoch": 298, "lr": 3.500931397350464e-05} +{"train_loss": 0.015758579596877098, "grad_norm": 0.07510120421648026, "global_step": 31637, "epoch": 298, "lr": 3.500648657308156e-05} +{"train_loss": 0.015009364113211632, "grad_norm": 0.07250110805034637, "global_step": 31638, "epoch": 298, "lr": 3.500365922533917e-05} +{"train_loss": 0.014622900635004044, "grad_norm": 0.08376763015985489, "global_step": 31639, "epoch": 298, "lr": 3.500083193028741e-05} +{"train_loss": 0.016266971826553345, "grad_norm": 0.10837633162736893, "global_step": 31640, "epoch": 298, "lr": 3.4998004687936196e-05} +{"train_loss": 0.01763584092259407, "grad_norm": 0.06262772530317307, "global_step": 31641, "epoch": 298, "lr": 3.499517749829547e-05} +{"train_loss": 0.015927786007523537, "grad_norm": 0.06627292931079865, "global_step": 31642, "epoch": 298, "lr": 3.4992350361375145e-05} +{"train_loss": 0.017248038202524185, "grad_norm": 0.0743560642004013, "global_step": 31643, "epoch": 298, "lr": 3.49895232771852e-05} +{"train_loss": 0.016823062673211098, "grad_norm": 0.08911766856908798, "global_step": 31644, "epoch": 298, "lr": 3.498669624573553e-05} +{"train_loss": 0.015990298241376877, "grad_norm": 0.0832752212882042, "global_step": 31645, "epoch": 298, "lr": 3.49838692670361e-05} +{"train_loss": 0.015909314155578613, "grad_norm": 0.07112868130207062, "global_step": 31646, "epoch": 298, "lr": 3.4981042341096804e-05} +{"train_loss": 0.015951335430145264, "grad_norm": 0.057329896837472916, "global_step": 31647, "epoch": 298, "lr": 3.497821546792759e-05} +{"train_loss": 0.015189635567367077, "grad_norm": 0.14161495864391327, "global_step": 31648, "epoch": 298, "lr": 3.49753886475384e-05} +{"train_loss": 0.014877515845000744, "grad_norm": 0.10669175535440445, "global_step": 31649, "epoch": 298, "lr": 3.497256187993916e-05} +{"train_loss": 0.017244352027773857, "grad_norm": 0.07874100655317307, "global_step": 31650, "epoch": 298, "lr": 3.4969735165139794e-05} +{"train_loss": 0.018059667199850082, "grad_norm": 0.07565506547689438, "global_step": 31651, "epoch": 298, "lr": 3.4966908503150263e-05} +{"train_loss": 0.0151866115629673, "grad_norm": 0.07335210591554642, "global_step": 31652, "epoch": 298, "lr": 3.496408189398046e-05} +{"train_loss": 0.018748849630355835, "grad_norm": 0.09272640198469162, "global_step": 31653, "epoch": 298, "lr": 3.496125533764034e-05} +{"train_loss": 0.013646486215293407, "grad_norm": 0.11069464683532715, "global_step": 31654, "epoch": 298, "lr": 3.495842883413982e-05} +{"train_loss": 0.015492448583245277, "grad_norm": 0.07148992270231247, "global_step": 31655, "epoch": 298, "lr": 3.495560238348885e-05} +{"train_loss": 0.014942670240998268, "grad_norm": 0.10107066482305527, "global_step": 31656, "epoch": 298, "lr": 3.495277598569735e-05} +{"train_loss": 0.014224663376808167, "grad_norm": 0.060505449771881104, "global_step": 31657, "epoch": 298, "lr": 3.4949949640775264e-05} +{"train_loss": 0.012246912345290184, "grad_norm": 0.05668627843260765, "global_step": 31658, "epoch": 298, "lr": 3.4947123348732496e-05} +{"train_loss": 0.012639589607715607, "grad_norm": 0.08364613354206085, "global_step": 31659, "epoch": 298, "lr": 3.494429710957898e-05} +{"train_loss": 0.01613426022231579, "grad_norm": 0.07749996334314346, "global_step": 31660, "epoch": 298, "lr": 3.494147092332467e-05} +{"train_loss": 0.013601630926132202, "grad_norm": 0.05317489057779312, "global_step": 31661, "epoch": 298, "lr": 3.493864478997948e-05} +{"train_loss": 0.01625690422952175, "grad_norm": 0.05649864301085472, "global_step": 31662, "epoch": 298, "lr": 3.493581870955335e-05} +{"train_loss": 0.015514776110649109, "grad_norm": 0.08284071087837219, "global_step": 31663, "epoch": 298, "lr": 3.493299268205619e-05} +{"train_loss": 0.01562298834323883, "grad_norm": 0.08297836780548096, "global_step": 31664, "epoch": 298, "lr": 3.493016670749794e-05} +{"train_loss": 0.014291202649474144, "grad_norm": 0.10137153416872025, "global_step": 31665, "epoch": 298, "lr": 3.4927340785888535e-05} +{"train_loss": 0.014124907553195953, "grad_norm": 0.07489131391048431, "global_step": 31666, "epoch": 298, "lr": 3.492451491723789e-05} +{"train_loss": 0.015776455402374268, "grad_norm": 0.06638013571500778, "global_step": 31667, "epoch": 298, "lr": 3.4921689101555955e-05} +{"train_loss": 0.01600211299955845, "grad_norm": 0.06675013154745102, "global_step": 31668, "epoch": 298, "lr": 3.491886333885265e-05} +{"train_loss": 0.016730064526200294, "grad_norm": 0.05230700597167015, "global_step": 31669, "epoch": 298, "lr": 3.49160376291379e-05} +{"train_loss": 0.015863753855228424, "grad_norm": 0.10369087755680084, "global_step": 31670, "epoch": 298, "lr": 3.491321197242163e-05} +{"train_loss": 0.017418133094906807, "grad_norm": 0.08765895664691925, "global_step": 31671, "epoch": 298, "lr": 3.4910386368713766e-05} +{"train_loss": 0.01603524759411812, "grad_norm": 0.10771311074495316, "global_step": 31672, "epoch": 298, "lr": 3.4907560818024246e-05} +{"train_loss": 0.01530715823173523, "grad_norm": 0.10135474801063538, "global_step": 31673, "epoch": 298, "lr": 3.490473532036299e-05} +{"train_loss": 0.013722579926252365, "grad_norm": 0.08583205938339233, "global_step": 31674, "epoch": 298, "lr": 3.4901909875739954e-05} +{"train_loss": 0.013192861340939999, "grad_norm": 0.08310636878013611, "global_step": 31675, "epoch": 298, "lr": 3.4899084484165024e-05} +{"train_loss": 0.015962697565555573, "grad_norm": 0.16266319155693054, "global_step": 31676, "epoch": 298, "lr": 3.489625914564812e-05} +{"train_loss": 0.015420932322740555, "grad_norm": 0.09476837515830994, "global_step": 31677, "epoch": 298, "lr": 3.489343386019922e-05} +{"train_loss": 0.01477896235883236, "grad_norm": 0.11234084516763687, "global_step": 31678, "epoch": 298, "lr": 3.489060862782821e-05} +{"train_loss": 0.015472638420760632, "grad_norm": 0.10955609381198883, "global_step": 31679, "epoch": 298, "lr": 3.488778344854504e-05} +{"train_loss": 0.018170062452554703, "grad_norm": 0.09466201066970825, "global_step": 31680, "epoch": 298, "lr": 3.488495832235963e-05} +{"train_loss": 0.014998476020991802, "grad_norm": 0.10501430928707123, "global_step": 31681, "epoch": 298, "lr": 3.48821332492819e-05} +{"train_loss": 0.017972277477383614, "grad_norm": 0.054085273295640945, "global_step": 31682, "epoch": 298, "lr": 3.487930822932178e-05} +{"train_loss": 0.01625341922044754, "grad_norm": 0.09236476570367813, "global_step": 31683, "epoch": 298, "lr": 3.4876483262489185e-05} +{"train_loss": 0.016285285353660583, "grad_norm": 0.07599151134490967, "global_step": 31684, "epoch": 298, "lr": 3.4873658348794056e-05} +{"train_loss": 0.014548830687999725, "grad_norm": 0.11936529725790024, "global_step": 31685, "epoch": 298, "lr": 3.487083348824632e-05} +{"train_loss": 0.015832483768463135, "grad_norm": 0.0964612290263176, "global_step": 31686, "epoch": 298, "lr": 3.48680086808559e-05} +{"train_loss": 0.015259530395269394, "grad_norm": 0.07315952330827713, "global_step": 31687, "epoch": 298, "lr": 3.486518392663271e-05} +{"train_loss": 0.014298714697360992, "grad_norm": 0.06644860655069351, "global_step": 31688, "epoch": 298, "lr": 3.4862359225586674e-05} +{"train_loss": 0.014943069778382778, "grad_norm": 0.07401132583618164, "global_step": 31689, "epoch": 298, "lr": 3.4859534577727726e-05} +{"train_loss": 0.017273448407649994, "grad_norm": 0.13561847805976868, "global_step": 31690, "epoch": 298, "lr": 3.4856709983065794e-05} +{"train_loss": 0.014688405208289623, "grad_norm": 0.07224708050489426, "global_step": 31691, "epoch": 298, "lr": 3.485388544161079e-05} +{"train_loss": 0.015391969121992588, "grad_norm": 0.09587457776069641, "global_step": 31692, "epoch": 298, "lr": 3.485106095337266e-05} +{"train_loss": 0.015318172904750649, "grad_norm": 0.0674748346209526, "global_step": 31693, "epoch": 298, "lr": 3.484823651836131e-05, "val_loss": 0.019568951800465584} +{"train_loss": 0.014420348219573498, "grad_norm": 0.08357253670692444, "global_step": 31694, "epoch": 299, "lr": 3.484541213658666e-05} +{"train_loss": 0.01622413471341133, "grad_norm": 0.13908278942108154, "global_step": 31695, "epoch": 299, "lr": 3.484258780805863e-05} +{"train_loss": 0.013002865016460419, "grad_norm": 0.08657559007406235, "global_step": 31696, "epoch": 299, "lr": 3.4839763532787177e-05} +{"train_loss": 0.014673419296741486, "grad_norm": 0.06436038017272949, "global_step": 31697, "epoch": 299, "lr": 3.4836939310782194e-05} +{"train_loss": 0.014043725095689297, "grad_norm": 0.07575978338718414, "global_step": 31698, "epoch": 299, "lr": 3.4834115142053626e-05} +{"train_loss": 0.013365482911467552, "grad_norm": 0.09854760766029358, "global_step": 31699, "epoch": 299, "lr": 3.4831291026611365e-05} +{"train_loss": 0.017145909368991852, "grad_norm": 0.06305201351642609, "global_step": 31700, "epoch": 299, "lr": 3.482846696446535e-05} +{"train_loss": 0.014249159023165703, "grad_norm": 0.09409652650356293, "global_step": 31701, "epoch": 299, "lr": 3.482564295562551e-05} +{"train_loss": 0.015262866392731667, "grad_norm": 0.08416597545146942, "global_step": 31702, "epoch": 299, "lr": 3.482281900010176e-05} +{"train_loss": 0.017184775322675705, "grad_norm": 0.07375723868608475, "global_step": 31703, "epoch": 299, "lr": 3.4819995097904014e-05} +{"train_loss": 0.012585334479808807, "grad_norm": 0.08061490207910538, "global_step": 31704, "epoch": 299, "lr": 3.4817171249042225e-05} +{"train_loss": 0.018810246139764786, "grad_norm": 0.07012970745563507, "global_step": 31705, "epoch": 299, "lr": 3.481434745352628e-05} +{"train_loss": 0.015189442783594131, "grad_norm": 0.08795595169067383, "global_step": 31706, "epoch": 299, "lr": 3.481152371136611e-05} +{"train_loss": 0.01542140357196331, "grad_norm": 0.09529457241296768, "global_step": 31707, "epoch": 299, "lr": 3.480870002257164e-05} +{"train_loss": 0.016481682658195496, "grad_norm": 0.06164869666099548, "global_step": 31708, "epoch": 299, "lr": 3.48058763871528e-05} +{"train_loss": 0.015169222839176655, "grad_norm": 0.08931286633014679, "global_step": 31709, "epoch": 299, "lr": 3.480305280511951e-05} +{"train_loss": 0.01784232072532177, "grad_norm": 0.07294349372386932, "global_step": 31710, "epoch": 299, "lr": 3.4800229276481676e-05} +{"train_loss": 0.016179120168089867, "grad_norm": 0.11395534873008728, "global_step": 31711, "epoch": 299, "lr": 3.4797405801249224e-05} +{"train_loss": 0.015591757372021675, "grad_norm": 0.07127970457077026, "global_step": 31712, "epoch": 299, "lr": 3.4794582379432066e-05} +{"train_loss": 0.015654077753424644, "grad_norm": 0.08307359367609024, "global_step": 31713, "epoch": 299, "lr": 3.479175901104014e-05} +{"train_loss": 0.015401948243379593, "grad_norm": 0.08142772316932678, "global_step": 31714, "epoch": 299, "lr": 3.4788935696083366e-05} +{"train_loss": 0.01593644544482231, "grad_norm": 0.08428216725587845, "global_step": 31715, "epoch": 299, "lr": 3.478611243457166e-05} +{"train_loss": 0.014180291444063187, "grad_norm": 0.07594528794288635, "global_step": 31716, "epoch": 299, "lr": 3.478328922651494e-05} +{"train_loss": 0.013844151049852371, "grad_norm": 0.10161947458982468, "global_step": 31717, "epoch": 299, "lr": 3.47804660719231e-05} +{"train_loss": 0.015766391530632973, "grad_norm": 0.11673842370510101, "global_step": 31718, "epoch": 299, "lr": 3.47776429708061e-05} +{"train_loss": 0.015424607321619987, "grad_norm": 0.07567629218101501, "global_step": 31719, "epoch": 299, "lr": 3.477481992317384e-05} +{"train_loss": 0.015145267359912395, "grad_norm": 0.07539226859807968, "global_step": 31720, "epoch": 299, "lr": 3.4771996929036234e-05} +{"train_loss": 0.014433810487389565, "grad_norm": 0.08543354272842407, "global_step": 31721, "epoch": 299, "lr": 3.476917398840323e-05} +{"train_loss": 0.013954998925328255, "grad_norm": 0.06285171210765839, "global_step": 31722, "epoch": 299, "lr": 3.476635110128471e-05} +{"train_loss": 0.017659496515989304, "grad_norm": 0.09839128702878952, "global_step": 31723, "epoch": 299, "lr": 3.476352826769062e-05} +{"train_loss": 0.015909582376480103, "grad_norm": 0.07912130653858185, "global_step": 31724, "epoch": 299, "lr": 3.476070548763084e-05} +{"train_loss": 0.01466938853263855, "grad_norm": 0.06654000282287598, "global_step": 31725, "epoch": 299, "lr": 3.475788276111533e-05} +{"train_loss": 0.017462771385908127, "grad_norm": 0.09517140686511993, "global_step": 31726, "epoch": 299, "lr": 3.475506008815399e-05} +{"train_loss": 0.014272416010499, "grad_norm": 0.0703023299574852, "global_step": 31727, "epoch": 299, "lr": 3.475223746875675e-05} +{"train_loss": 0.015688667073845863, "grad_norm": 0.07162769138813019, "global_step": 31728, "epoch": 299, "lr": 3.474941490293351e-05} +{"train_loss": 0.013900521211326122, "grad_norm": 0.18378852307796478, "global_step": 31729, "epoch": 299, "lr": 3.474659239069418e-05} +{"train_loss": 0.014465998858213425, "grad_norm": 0.08657281845808029, "global_step": 31730, "epoch": 299, "lr": 3.4743769932048695e-05} +{"train_loss": 0.015272601507604122, "grad_norm": 0.11349333822727203, "global_step": 31731, "epoch": 299, "lr": 3.474094752700697e-05} +{"train_loss": 0.012134047225117683, "grad_norm": 0.0768921822309494, "global_step": 31732, "epoch": 299, "lr": 3.473812517557892e-05} +{"train_loss": 0.013529561460018158, "grad_norm": 0.098051056265831, "global_step": 31733, "epoch": 299, "lr": 3.473530287777447e-05} +{"train_loss": 0.016612127423286438, "grad_norm": 0.07603808492422104, "global_step": 31734, "epoch": 299, "lr": 3.473248063360352e-05} +{"train_loss": 0.012843040749430656, "grad_norm": 0.06389998644590378, "global_step": 31735, "epoch": 299, "lr": 3.4729658443075995e-05} +{"train_loss": 0.012714551761746407, "grad_norm": 0.10087556391954422, "global_step": 31736, "epoch": 299, "lr": 3.47268363062018e-05} +{"train_loss": 0.015054207295179367, "grad_norm": 0.054100263863801956, "global_step": 31737, "epoch": 299, "lr": 3.472401422299086e-05} +{"train_loss": 0.015937775373458862, "grad_norm": 0.11691155284643173, "global_step": 31738, "epoch": 299, "lr": 3.47211921934531e-05} +{"train_loss": 0.015243740752339363, "grad_norm": 0.056366339325904846, "global_step": 31739, "epoch": 299, "lr": 3.471837021759844e-05} +{"train_loss": 0.015562831424176693, "grad_norm": 0.0560145266354084, "global_step": 31740, "epoch": 299, "lr": 3.4715548295436765e-05} +{"train_loss": 0.015854479745030403, "grad_norm": 0.08750268071889877, "global_step": 31741, "epoch": 299, "lr": 3.4712726426978e-05} +{"train_loss": 0.01673728972673416, "grad_norm": 0.1552957147359848, "global_step": 31742, "epoch": 299, "lr": 3.4709904612232076e-05} +{"train_loss": 0.017941802740097046, "grad_norm": 0.11040449887514114, "global_step": 31743, "epoch": 299, "lr": 3.47070828512089e-05} +{"train_loss": 0.016768567264080048, "grad_norm": 0.06758230924606323, "global_step": 31744, "epoch": 299, "lr": 3.470426114391836e-05} +{"train_loss": 0.014296193607151508, "grad_norm": 0.10339367389678955, "global_step": 31745, "epoch": 299, "lr": 3.470143949037043e-05} +{"train_loss": 0.016204794868826866, "grad_norm": 0.12780030071735382, "global_step": 31746, "epoch": 299, "lr": 3.469861789057497e-05} +{"train_loss": 0.01668773591518402, "grad_norm": 0.13646875321865082, "global_step": 31747, "epoch": 299, "lr": 3.469579634454192e-05} +{"train_loss": 0.017250964418053627, "grad_norm": 0.11125490814447403, "global_step": 31748, "epoch": 299, "lr": 3.469297485228118e-05} +{"train_loss": 0.01346514094620943, "grad_norm": 0.08293129503726959, "global_step": 31749, "epoch": 299, "lr": 3.469015341380266e-05} +{"train_loss": 0.016352292150259018, "grad_norm": 0.08011091500520706, "global_step": 31750, "epoch": 299, "lr": 3.468733202911629e-05} +{"train_loss": 0.014588048681616783, "grad_norm": 0.09196682274341583, "global_step": 31751, "epoch": 299, "lr": 3.468451069823199e-05} +{"train_loss": 0.01389764342457056, "grad_norm": 0.061474479734897614, "global_step": 31752, "epoch": 299, "lr": 3.468168942115965e-05} +{"train_loss": 0.015774866566061974, "grad_norm": 0.07533691823482513, "global_step": 31753, "epoch": 299, "lr": 3.467886819790918e-05} +{"train_loss": 0.013984410092234612, "grad_norm": 0.07849513739347458, "global_step": 31754, "epoch": 299, "lr": 3.467604702849051e-05} +{"train_loss": 0.013482972979545593, "grad_norm": 0.062126535922288895, "global_step": 31755, "epoch": 299, "lr": 3.467322591291355e-05} +{"train_loss": 0.016542088240385056, "grad_norm": 0.13300222158432007, "global_step": 31756, "epoch": 299, "lr": 3.4670404851188196e-05} +{"train_loss": 0.01773690991103649, "grad_norm": 0.10187585651874542, "global_step": 31757, "epoch": 299, "lr": 3.4667583843324413e-05} +{"train_loss": 0.01702289655804634, "grad_norm": 0.07816702127456665, "global_step": 31758, "epoch": 299, "lr": 3.4664762889332034e-05} +{"train_loss": 0.014651905745267868, "grad_norm": 0.11131440103054047, "global_step": 31759, "epoch": 299, "lr": 3.466194198922102e-05} +{"train_loss": 0.015854738652706146, "grad_norm": 0.06563296914100647, "global_step": 31760, "epoch": 299, "lr": 3.465912114300127e-05} +{"train_loss": 0.014624832198023796, "grad_norm": 0.06930740922689438, "global_step": 31761, "epoch": 299, "lr": 3.465630035068269e-05} +{"train_loss": 0.015177132561802864, "grad_norm": 0.05322878435254097, "global_step": 31762, "epoch": 299, "lr": 3.465347961227522e-05} +{"train_loss": 0.0142764151096344, "grad_norm": 0.06392662972211838, "global_step": 31763, "epoch": 299, "lr": 3.465065892778873e-05} +{"train_loss": 0.013499287888407707, "grad_norm": 0.06186080351471901, "global_step": 31764, "epoch": 299, "lr": 3.464783829723316e-05} +{"train_loss": 0.015277298167347908, "grad_norm": 0.1216139942407608, "global_step": 31765, "epoch": 299, "lr": 3.464501772061839e-05} +{"train_loss": 0.016556236892938614, "grad_norm": 0.057903777807950974, "global_step": 31766, "epoch": 299, "lr": 3.464219719795436e-05} +{"train_loss": 0.0139400539919734, "grad_norm": 0.05983249843120575, "global_step": 31767, "epoch": 299, "lr": 3.4639376729250976e-05} +{"train_loss": 0.01619105041027069, "grad_norm": 0.15380243957042694, "global_step": 31768, "epoch": 299, "lr": 3.463655631451815e-05} +{"train_loss": 0.013886488974094391, "grad_norm": 0.0842890664935112, "global_step": 31769, "epoch": 299, "lr": 3.4633735953765776e-05} +{"train_loss": 0.01432965137064457, "grad_norm": 0.10397075861692429, "global_step": 31770, "epoch": 299, "lr": 3.463091564700376e-05} +{"train_loss": 0.015055270865559578, "grad_norm": 0.0522116981446743, "global_step": 31771, "epoch": 299, "lr": 3.4628095394242033e-05} +{"train_loss": 0.015100881457328796, "grad_norm": 0.061403416097164154, "global_step": 31772, "epoch": 299, "lr": 3.4625275195490495e-05} +{"train_loss": 0.016275547444820404, "grad_norm": 0.06916072219610214, "global_step": 31773, "epoch": 299, "lr": 3.462245505075904e-05} +{"train_loss": 0.013177821412682533, "grad_norm": 0.06001753360033035, "global_step": 31774, "epoch": 299, "lr": 3.4619634960057625e-05} +{"train_loss": 0.01637910306453705, "grad_norm": 0.12224340438842773, "global_step": 31775, "epoch": 299, "lr": 3.4616814923396087e-05} +{"train_loss": 0.01742122694849968, "grad_norm": 0.08010371029376984, "global_step": 31776, "epoch": 299, "lr": 3.4613994940784385e-05} +{"train_loss": 0.016182061284780502, "grad_norm": 0.08570663630962372, "global_step": 31777, "epoch": 299, "lr": 3.461117501223241e-05} +{"train_loss": 0.015346406027674675, "grad_norm": 0.10533419251441956, "global_step": 31778, "epoch": 299, "lr": 3.460835513775007e-05} +{"train_loss": 0.016916397958993912, "grad_norm": 0.0756322517991066, "global_step": 31779, "epoch": 299, "lr": 3.460553531734728e-05} +{"train_loss": 0.01390254870057106, "grad_norm": 0.08680888265371323, "global_step": 31780, "epoch": 299, "lr": 3.460271555103396e-05} +{"train_loss": 0.014145833440124989, "grad_norm": 0.07433280348777771, "global_step": 31781, "epoch": 299, "lr": 3.459989583881999e-05} +{"train_loss": 0.014591129496693611, "grad_norm": 0.1049569845199585, "global_step": 31782, "epoch": 299, "lr": 3.4597076180715274e-05} +{"train_loss": 0.016264576464891434, "grad_norm": 0.057230331003665924, "global_step": 31783, "epoch": 299, "lr": 3.459425657672975e-05} +{"train_loss": 0.014434821903705597, "grad_norm": 0.12965847551822662, "global_step": 31784, "epoch": 299, "lr": 3.459143702687331e-05} +{"train_loss": 0.01778975874185562, "grad_norm": 0.0884193703532219, "global_step": 31785, "epoch": 299, "lr": 3.4588617531155844e-05} +{"train_loss": 0.015996497124433517, "grad_norm": 0.12488214671611786, "global_step": 31786, "epoch": 299, "lr": 3.4585798089587305e-05} +{"train_loss": 0.018540631979703903, "grad_norm": 0.2755005359649658, "global_step": 31787, "epoch": 299, "lr": 3.458297870217754e-05} +{"train_loss": 0.014521896839141846, "grad_norm": 0.14825798571109772, "global_step": 31788, "epoch": 299, "lr": 3.458015936893649e-05} +{"train_loss": 0.01381246279925108, "grad_norm": 0.19690822064876556, "global_step": 31789, "epoch": 299, "lr": 3.457734008987405e-05} +{"train_loss": 0.015395497903227806, "grad_norm": 0.09493719786405563, "global_step": 31790, "epoch": 299, "lr": 3.457452086500013e-05} +{"train_loss": 0.016276691108942032, "grad_norm": 0.061139509081840515, "global_step": 31791, "epoch": 299, "lr": 3.457170169432464e-05} +{"train_loss": 0.012628979049623013, "grad_norm": 0.1218334436416626, "global_step": 31792, "epoch": 299, "lr": 3.456888257785749e-05} +{"train_loss": 0.016376597806811333, "grad_norm": 0.11434423923492432, "global_step": 31793, "epoch": 299, "lr": 3.4566063515608565e-05} +{"train_loss": 0.014507526531815529, "grad_norm": 0.07574755698442459, "global_step": 31794, "epoch": 299, "lr": 3.4563244507587775e-05} +{"train_loss": 0.015504344366490841, "grad_norm": 0.06171521171927452, "global_step": 31795, "epoch": 299, "lr": 3.456042555380504e-05} +{"train_loss": 0.016353972256183624, "grad_norm": 0.18002033233642578, "global_step": 31796, "epoch": 299, "lr": 3.455760665427025e-05} +{"train_loss": 0.016132675111293793, "grad_norm": 0.0611998550593853, "global_step": 31797, "epoch": 299, "lr": 3.455478780899331e-05} +{"train_loss": 0.013369984924793243, "grad_norm": 0.09803590178489685, "global_step": 31798, "epoch": 299, "lr": 3.4551969017984155e-05} +{"train_loss": 0.015302651860522773, "grad_norm": 0.08547690510749817, "global_step": 31799, "epoch": 299, "lr": 3.4549150281252636e-05, "val_loss": 0.01921265572309494} +{"train_loss": 0.015855804085731506, "grad_norm": 0.1536698341369629, "global_step": 31800, "epoch": 300, "lr": 3.454633159880869e-05} +{"train_loss": 0.015983691439032555, "grad_norm": 0.08362852782011032, "global_step": 31801, "epoch": 300, "lr": 3.4543512970662215e-05} +{"train_loss": 0.013650277629494667, "grad_norm": 0.16938282549381256, "global_step": 31802, "epoch": 300, "lr": 3.45406943968231e-05} +{"train_loss": 0.013516830280423164, "grad_norm": 0.06334639340639114, "global_step": 31803, "epoch": 300, "lr": 3.453787587730128e-05} +{"train_loss": 0.01665571704506874, "grad_norm": 0.05429196357727051, "global_step": 31804, "epoch": 300, "lr": 3.4535057412106645e-05} +{"train_loss": 0.016405075788497925, "grad_norm": 0.0984673872590065, "global_step": 31805, "epoch": 300, "lr": 3.453223900124909e-05} +{"train_loss": 0.016658738255500793, "grad_norm": 0.07722064107656479, "global_step": 31806, "epoch": 300, "lr": 3.452942064473851e-05} +{"train_loss": 0.014696555212140083, "grad_norm": 0.06642891466617584, "global_step": 31807, "epoch": 300, "lr": 3.4526602342584815e-05} +{"train_loss": 0.014429580420255661, "grad_norm": 0.07518034428358078, "global_step": 31808, "epoch": 300, "lr": 3.4523784094797914e-05} +{"train_loss": 0.016248783096671104, "grad_norm": 0.09547816962003708, "global_step": 31809, "epoch": 300, "lr": 3.45209659013877e-05} +{"train_loss": 0.01719680242240429, "grad_norm": 0.1414981633424759, "global_step": 31810, "epoch": 300, "lr": 3.451814776236411e-05} +{"train_loss": 0.016007322818040848, "grad_norm": 0.08037551492452621, "global_step": 31811, "epoch": 300, "lr": 3.4515329677736984e-05} +{"train_loss": 0.015701720491051674, "grad_norm": 0.07922735065221786, "global_step": 31812, "epoch": 300, "lr": 3.451251164751626e-05} +{"train_loss": 0.013435987755656242, "grad_norm": 0.1118687093257904, "global_step": 31813, "epoch": 300, "lr": 3.4509693671711845e-05} +{"train_loss": 0.013547269627451897, "grad_norm": 0.0629807636141777, "global_step": 31814, "epoch": 300, "lr": 3.450687575033362e-05} +{"train_loss": 0.01411384902894497, "grad_norm": 0.05747457966208458, "global_step": 31815, "epoch": 300, "lr": 3.4504057883391514e-05} +{"train_loss": 0.015243534930050373, "grad_norm": 0.09762590378522873, "global_step": 31816, "epoch": 300, "lr": 3.450124007089539e-05} +{"train_loss": 0.014533596113324165, "grad_norm": 0.09737301617860794, "global_step": 31817, "epoch": 300, "lr": 3.449842231285518e-05} +{"train_loss": 0.013329248875379562, "grad_norm": 0.0460631437599659, "global_step": 31818, "epoch": 300, "lr": 3.449560460928076e-05} +{"train_loss": 0.015521468594670296, "grad_norm": 0.0782109871506691, "global_step": 31819, "epoch": 300, "lr": 3.4492786960182046e-05} +{"train_loss": 0.016469385474920273, "grad_norm": 0.08772487193346024, "global_step": 31820, "epoch": 300, "lr": 3.448996936556894e-05} +{"train_loss": 0.015607615001499653, "grad_norm": 0.07080353051424026, "global_step": 31821, "epoch": 300, "lr": 3.4487151825451344e-05} +{"train_loss": 0.017006121575832367, "grad_norm": 0.054063692688941956, "global_step": 31822, "epoch": 300, "lr": 3.448433433983914e-05} +{"train_loss": 0.015138426795601845, "grad_norm": 0.0736042708158493, "global_step": 31823, "epoch": 300, "lr": 3.448151690874223e-05} +{"train_loss": 0.014518997631967068, "grad_norm": 0.06844358146190643, "global_step": 31824, "epoch": 300, "lr": 3.4478699532170536e-05} +{"train_loss": 0.014139782637357712, "grad_norm": 0.0853656679391861, "global_step": 31825, "epoch": 300, "lr": 3.447588221013393e-05} +{"train_loss": 0.015152118168771267, "grad_norm": 0.10893867164850235, "global_step": 31826, "epoch": 300, "lr": 3.447306494264232e-05} +{"train_loss": 0.014583704061806202, "grad_norm": 0.24587468802928925, "global_step": 31827, "epoch": 300, "lr": 3.447024772970564e-05} +{"train_loss": 0.01615304872393608, "grad_norm": 0.062450848519802094, "global_step": 31828, "epoch": 300, "lr": 3.4467430571333715e-05} +{"train_loss": 0.0188553836196661, "grad_norm": 0.07893944531679153, "global_step": 31829, "epoch": 300, "lr": 3.44646134675365e-05} +{"train_loss": 0.017713062465190887, "grad_norm": 0.13077546656131744, "global_step": 31830, "epoch": 300, "lr": 3.446179641832388e-05} +{"train_loss": 0.017088623717427254, "grad_norm": 0.08430872857570648, "global_step": 31831, "epoch": 300, "lr": 3.445897942370574e-05} +{"train_loss": 0.014210495166480541, "grad_norm": 0.09112270921468735, "global_step": 31832, "epoch": 300, "lr": 3.4456162483692e-05} +{"train_loss": 0.013720544055104256, "grad_norm": 0.07200915366411209, "global_step": 31833, "epoch": 300, "lr": 3.445334559829255e-05} +{"train_loss": 0.015491878613829613, "grad_norm": 0.06923525780439377, "global_step": 31834, "epoch": 300, "lr": 3.445052876751727e-05} +{"train_loss": 0.01344643346965313, "grad_norm": 0.044189006090164185, "global_step": 31835, "epoch": 300, "lr": 3.4447711991376065e-05} +{"train_loss": 0.013022681698203087, "grad_norm": 0.1746624857187271, "global_step": 31836, "epoch": 300, "lr": 3.444489526987884e-05} +{"train_loss": 0.016243839636445045, "grad_norm": 0.08928442746400833, "global_step": 31837, "epoch": 300, "lr": 3.444207860303549e-05} +{"train_loss": 0.015224495902657509, "grad_norm": 0.15765218436717987, "global_step": 31838, "epoch": 300, "lr": 3.443926199085591e-05} +{"train_loss": 0.01605638861656189, "grad_norm": 0.1415041983127594, "global_step": 31839, "epoch": 300, "lr": 3.4436445433350015e-05} +{"train_loss": 0.01716184802353382, "grad_norm": 0.0712294802069664, "global_step": 31840, "epoch": 300, "lr": 3.443362893052765e-05} +{"train_loss": 0.015101802535355091, "grad_norm": 0.08638158440589905, "global_step": 31841, "epoch": 300, "lr": 3.4430812482398764e-05} +{"train_loss": 0.014712648466229439, "grad_norm": 0.06392432004213333, "global_step": 31842, "epoch": 300, "lr": 3.442799608897323e-05} +{"train_loss": 0.016994185745716095, "grad_norm": 0.06858090311288834, "global_step": 31843, "epoch": 300, "lr": 3.442517975026093e-05} +{"train_loss": 0.016952037811279297, "grad_norm": 0.07719594240188599, "global_step": 31844, "epoch": 300, "lr": 3.442236346627179e-05} +{"train_loss": 0.014233386144042015, "grad_norm": 0.06268008798360825, "global_step": 31845, "epoch": 300, "lr": 3.44195472370157e-05} +{"train_loss": 0.016899827867746353, "grad_norm": 0.0716463252902031, "global_step": 31846, "epoch": 300, "lr": 3.441673106250253e-05} +{"train_loss": 0.01393450703471899, "grad_norm": 0.07700613141059875, "global_step": 31847, "epoch": 300, "lr": 3.44139149427422e-05} +{"train_loss": 0.01697569154202938, "grad_norm": 0.11744220554828644, "global_step": 31848, "epoch": 300, "lr": 3.4411098877744574e-05} +{"train_loss": 0.015233899466693401, "grad_norm": 0.061244212090969086, "global_step": 31849, "epoch": 300, "lr": 3.440828286751958e-05} +{"train_loss": 0.01673385128378868, "grad_norm": 0.12884515523910522, "global_step": 31850, "epoch": 300, "lr": 3.44054669120771e-05} +{"train_loss": 0.018866270780563354, "grad_norm": 0.12173192203044891, "global_step": 31851, "epoch": 300, "lr": 3.4402651011427045e-05} +{"train_loss": 0.0138346366584301, "grad_norm": 0.1808246225118637, "global_step": 31852, "epoch": 300, "lr": 3.4399835165579266e-05} +{"train_loss": 0.014520535245537758, "grad_norm": 0.09329021722078323, "global_step": 31853, "epoch": 300, "lr": 3.439701937454368e-05} +{"train_loss": 0.013975991867482662, "grad_norm": 0.09958817809820175, "global_step": 31854, "epoch": 300, "lr": 3.43942036383302e-05} +{"train_loss": 0.01461615040898323, "grad_norm": 0.08847898244857788, "global_step": 31855, "epoch": 300, "lr": 3.439138795694868e-05} +{"train_loss": 0.016028478741645813, "grad_norm": 0.052514102309942245, "global_step": 31856, "epoch": 300, "lr": 3.4388572330409055e-05} +{"train_loss": 0.014372235164046288, "grad_norm": 0.07457128912210464, "global_step": 31857, "epoch": 300, "lr": 3.43857567587212e-05} +{"train_loss": 0.019581250846385956, "grad_norm": 0.12540468573570251, "global_step": 31858, "epoch": 300, "lr": 3.4382941241894996e-05} +{"train_loss": 0.01680586487054825, "grad_norm": 0.12224306911230087, "global_step": 31859, "epoch": 300, "lr": 3.438012577994034e-05} +{"train_loss": 0.015971913933753967, "grad_norm": 0.07578334212303162, "global_step": 31860, "epoch": 300, "lr": 3.437731037286713e-05} +{"train_loss": 0.015482237562537193, "grad_norm": 0.09853431582450867, "global_step": 31861, "epoch": 300, "lr": 3.437449502068526e-05} +{"train_loss": 0.017002906650304794, "grad_norm": 0.07570234686136246, "global_step": 31862, "epoch": 300, "lr": 3.437167972340464e-05} +{"train_loss": 0.015811480581760406, "grad_norm": 0.11093869060277939, "global_step": 31863, "epoch": 300, "lr": 3.436886448103511e-05} +{"train_loss": 0.013130361214280128, "grad_norm": 0.08526705950498581, "global_step": 31864, "epoch": 300, "lr": 3.43660492935866e-05} +{"train_loss": 0.016593188047409058, "grad_norm": 0.06652757525444031, "global_step": 31865, "epoch": 300, "lr": 3.4363234161069e-05} +{"train_loss": 0.013467421755194664, "grad_norm": 0.09981811046600342, "global_step": 31866, "epoch": 300, "lr": 3.436041908349219e-05} +{"train_loss": 0.012689445167779922, "grad_norm": 0.13150368630886078, "global_step": 31867, "epoch": 300, "lr": 3.435760406086606e-05} +{"train_loss": 0.015045073814690113, "grad_norm": 0.05782172456383705, "global_step": 31868, "epoch": 300, "lr": 3.4354789093200534e-05} +{"train_loss": 0.014417367056012154, "grad_norm": 0.10746131837368011, "global_step": 31869, "epoch": 300, "lr": 3.435197418050545e-05} +{"train_loss": 0.014959384687244892, "grad_norm": 0.06178983300924301, "global_step": 31870, "epoch": 300, "lr": 3.434915932279073e-05} +{"train_loss": 0.01570175215601921, "grad_norm": 0.07760998606681824, "global_step": 31871, "epoch": 300, "lr": 3.4346344520066256e-05} +{"train_loss": 0.016208328306674957, "grad_norm": 0.0674319639801979, "global_step": 31872, "epoch": 300, "lr": 3.4343529772341914e-05} +{"train_loss": 0.013568117283284664, "grad_norm": 0.08792106062173843, "global_step": 31873, "epoch": 300, "lr": 3.4340715079627605e-05} +{"train_loss": 0.01434149220585823, "grad_norm": 0.050287242978811264, "global_step": 31874, "epoch": 300, "lr": 3.433790044193323e-05} +{"train_loss": 0.015588768757879734, "grad_norm": 0.08283320814371109, "global_step": 31875, "epoch": 300, "lr": 3.4335085859268645e-05} +{"train_loss": 0.013318052515387535, "grad_norm": 0.08481132984161377, "global_step": 31876, "epoch": 300, "lr": 3.433227133164376e-05} +{"train_loss": 0.013613712042570114, "grad_norm": 0.09561292827129364, "global_step": 31877, "epoch": 300, "lr": 3.432945685906844e-05} +{"train_loss": 0.014346206560730934, "grad_norm": 0.10427084565162659, "global_step": 31878, "epoch": 300, "lr": 3.4326642441552616e-05} +{"train_loss": 0.014195811003446579, "grad_norm": 0.06665942817926407, "global_step": 31879, "epoch": 300, "lr": 3.432382807910615e-05} +{"train_loss": 0.015094775706529617, "grad_norm": 0.10329801589250565, "global_step": 31880, "epoch": 300, "lr": 3.432101377173895e-05} +{"train_loss": 0.01584572158753872, "grad_norm": 0.061044130474328995, "global_step": 31881, "epoch": 300, "lr": 3.431819951946086e-05} +{"train_loss": 0.015329168178141117, "grad_norm": 0.054039306938648224, "global_step": 31882, "epoch": 300, "lr": 3.431538532228181e-05} +{"train_loss": 0.014674559235572815, "grad_norm": 0.07454244047403336, "global_step": 31883, "epoch": 300, "lr": 3.4312571180211675e-05} +{"train_loss": 0.013181530870497227, "grad_norm": 0.09812761098146439, "global_step": 31884, "epoch": 300, "lr": 3.430975709326033e-05} +{"train_loss": 0.013237081468105316, "grad_norm": 0.05649905279278755, "global_step": 31885, "epoch": 300, "lr": 3.430694306143769e-05} +{"train_loss": 0.016401436179876328, "grad_norm": 0.06432874500751495, "global_step": 31886, "epoch": 300, "lr": 3.430412908475364e-05} +{"train_loss": 0.016193918883800507, "grad_norm": 0.13314837217330933, "global_step": 31887, "epoch": 300, "lr": 3.430131516321804e-05} +{"train_loss": 0.01475090067833662, "grad_norm": 0.10614664107561111, "global_step": 31888, "epoch": 300, "lr": 3.429850129684079e-05} +{"train_loss": 0.014701956883072853, "grad_norm": 0.10799393802881241, "global_step": 31889, "epoch": 300, "lr": 3.429568748563177e-05} +{"train_loss": 0.01492743194103241, "grad_norm": 0.09542937576770782, "global_step": 31890, "epoch": 300, "lr": 3.429287372960088e-05} +{"train_loss": 0.016342055052518845, "grad_norm": 0.11025862395763397, "global_step": 31891, "epoch": 300, "lr": 3.429006002875801e-05} +{"train_loss": 0.013804921880364418, "grad_norm": 0.07429753988981247, "global_step": 31892, "epoch": 300, "lr": 3.428724638311304e-05} +{"train_loss": 0.015883002430200577, "grad_norm": 0.06573814898729324, "global_step": 31893, "epoch": 300, "lr": 3.4284432792675835e-05} +{"train_loss": 0.017595145851373672, "grad_norm": 0.12220481038093567, "global_step": 31894, "epoch": 300, "lr": 3.428161925745631e-05} +{"train_loss": 0.015527859330177307, "grad_norm": 0.13919131457805634, "global_step": 31895, "epoch": 300, "lr": 3.427880577746434e-05} +{"train_loss": 0.014931906014680862, "grad_norm": 0.06265801936388016, "global_step": 31896, "epoch": 300, "lr": 3.42759923527098e-05} +{"train_loss": 0.01802806928753853, "grad_norm": 0.09308167546987534, "global_step": 31897, "epoch": 300, "lr": 3.427317898320259e-05} +{"train_loss": 0.015159938484430313, "grad_norm": 0.07198502123355865, "global_step": 31898, "epoch": 300, "lr": 3.42703656689526e-05} +{"train_loss": 0.013603691011667252, "grad_norm": 0.056933727115392685, "global_step": 31899, "epoch": 300, "lr": 3.426755240996969e-05} +{"train_loss": 0.013480732217431068, "grad_norm": 0.056804317981004715, "global_step": 31900, "epoch": 300, "lr": 3.426473920626377e-05} +{"train_loss": 0.013830152340233326, "grad_norm": 0.05904066190123558, "global_step": 31901, "epoch": 300, "lr": 3.4261926057844696e-05} +{"train_loss": 0.01511998102068901, "grad_norm": 0.07257190346717834, "global_step": 31902, "epoch": 300, "lr": 3.4259112964722386e-05} +{"train_loss": 0.017034240067005157, "grad_norm": 0.07273997366428375, "global_step": 31903, "epoch": 300, "lr": 3.42562999269067e-05} +{"train_loss": 0.017434192821383476, "grad_norm": 0.14220726490020752, "global_step": 31904, "epoch": 300, "lr": 3.4253486944407545e-05} +{"train_loss": 0.015284119525326873, "grad_norm": 0.07900657504796982, "global_step": 31905, "epoch": 300, "lr": 3.425067401723477e-05, "train_batch_mse_error": 0.013897435739636421, "val_loss": 0.01966799609363079} +{"train_loss": 0.0152894277125597, "grad_norm": 0.05571039766073227, "global_step": 31906, "epoch": 301, "lr": 3.4247861145398275e-05} +{"train_loss": 0.015222849324345589, "grad_norm": 0.07680514454841614, "global_step": 31907, "epoch": 301, "lr": 3.4245048328907955e-05} +{"train_loss": 0.014167040586471558, "grad_norm": 0.0817161276936531, "global_step": 31908, "epoch": 301, "lr": 3.4242235567773684e-05} +{"train_loss": 0.015490943565964699, "grad_norm": 0.055155497044324875, "global_step": 31909, "epoch": 301, "lr": 3.4239422862005335e-05} +{"train_loss": 0.014521146193146706, "grad_norm": 0.06636188924312592, "global_step": 31910, "epoch": 301, "lr": 3.4236610211612816e-05} +{"train_loss": 0.015570497140288353, "grad_norm": 0.10453999042510986, "global_step": 31911, "epoch": 301, "lr": 3.423379761660599e-05} +{"train_loss": 0.01607433706521988, "grad_norm": 0.11850442737340927, "global_step": 31912, "epoch": 301, "lr": 3.4230985076994736e-05} +{"train_loss": 0.015166985802352428, "grad_norm": 0.08651861548423767, "global_step": 31913, "epoch": 301, "lr": 3.422817259278893e-05} +{"train_loss": 0.01859090104699135, "grad_norm": 0.14901094138622284, "global_step": 31914, "epoch": 301, "lr": 3.4225360163998484e-05} +{"train_loss": 0.014443989843130112, "grad_norm": 0.13838429749011993, "global_step": 31915, "epoch": 301, "lr": 3.422254779063327e-05} +{"train_loss": 0.014702800661325455, "grad_norm": 0.07801881432533264, "global_step": 31916, "epoch": 301, "lr": 3.421973547270315e-05} +{"train_loss": 0.016734223812818527, "grad_norm": 0.07005485147237778, "global_step": 31917, "epoch": 301, "lr": 3.421692321021802e-05} +{"train_loss": 0.015115445479750633, "grad_norm": 0.07677523046731949, "global_step": 31918, "epoch": 301, "lr": 3.421411100318774e-05} +{"train_loss": 0.014727156609296799, "grad_norm": 0.05165054276585579, "global_step": 31919, "epoch": 301, "lr": 3.421129885162223e-05} +{"train_loss": 0.016657495871186256, "grad_norm": 0.07356393337249756, "global_step": 31920, "epoch": 301, "lr": 3.4208486755531344e-05} +{"train_loss": 0.017468998208642006, "grad_norm": 0.15789510309696198, "global_step": 31921, "epoch": 301, "lr": 3.420567471492498e-05} +{"train_loss": 0.016002003103494644, "grad_norm": 0.12700405716896057, "global_step": 31922, "epoch": 301, "lr": 3.420286272981298e-05} +{"train_loss": 0.01876063644886017, "grad_norm": 0.09209183603525162, "global_step": 31923, "epoch": 301, "lr": 3.420005080020526e-05} +{"train_loss": 0.016696367412805557, "grad_norm": 0.07626921683549881, "global_step": 31924, "epoch": 301, "lr": 3.41972389261117e-05} +{"train_loss": 0.014342195354402065, "grad_norm": 0.0872260257601738, "global_step": 31925, "epoch": 301, "lr": 3.419442710754216e-05} +{"train_loss": 0.016565313562750816, "grad_norm": 0.13408246636390686, "global_step": 31926, "epoch": 301, "lr": 3.4191615344506535e-05} +{"train_loss": 0.015052895992994308, "grad_norm": 0.12825818359851837, "global_step": 31927, "epoch": 301, "lr": 3.418880363701471e-05} +{"train_loss": 0.01551039144396782, "grad_norm": 0.07166895270347595, "global_step": 31928, "epoch": 301, "lr": 3.418599198507654e-05} +{"train_loss": 0.014201516285538673, "grad_norm": 0.0669729933142662, "global_step": 31929, "epoch": 301, "lr": 3.4183180388701915e-05} +{"train_loss": 0.013367712497711182, "grad_norm": 0.1172778308391571, "global_step": 31930, "epoch": 301, "lr": 3.418036884790071e-05} +{"train_loss": 0.015377060510218143, "grad_norm": 0.07814793288707733, "global_step": 31931, "epoch": 301, "lr": 3.417755736268282e-05} +{"train_loss": 0.013063717633485794, "grad_norm": 0.10718937963247299, "global_step": 31932, "epoch": 301, "lr": 3.417474593305811e-05} +{"train_loss": 0.01799583062529564, "grad_norm": 0.11376187950372696, "global_step": 31933, "epoch": 301, "lr": 3.417193455903647e-05} +{"train_loss": 0.01566193252801895, "grad_norm": 0.060013048350811005, "global_step": 31934, "epoch": 301, "lr": 3.4169123240627756e-05} +{"train_loss": 0.013802073895931244, "grad_norm": 0.07546462118625641, "global_step": 31935, "epoch": 301, "lr": 3.416631197784185e-05} +{"train_loss": 0.015938647091388702, "grad_norm": 0.0828702300786972, "global_step": 31936, "epoch": 301, "lr": 3.416350077068865e-05} +{"train_loss": 0.014522725716233253, "grad_norm": 0.06050528585910797, "global_step": 31937, "epoch": 301, "lr": 3.416068961917801e-05} +{"train_loss": 0.014463433995842934, "grad_norm": 0.057315316051244736, "global_step": 31938, "epoch": 301, "lr": 3.415787852331983e-05} +{"train_loss": 0.01681303046643734, "grad_norm": 0.057846952229738235, "global_step": 31939, "epoch": 301, "lr": 3.415506748312399e-05} +{"train_loss": 0.01726478338241577, "grad_norm": 0.056160859763622284, "global_step": 31940, "epoch": 301, "lr": 3.4152256498600335e-05} +{"train_loss": 0.013967305421829224, "grad_norm": 0.06338559091091156, "global_step": 31941, "epoch": 301, "lr": 3.414944556975876e-05} +{"train_loss": 0.014731111004948616, "grad_norm": 0.15343959629535675, "global_step": 31942, "epoch": 301, "lr": 3.414663469660913e-05} +{"train_loss": 0.013150272890925407, "grad_norm": 0.1038142591714859, "global_step": 31943, "epoch": 301, "lr": 3.414382387916134e-05} +{"train_loss": 0.01469193771481514, "grad_norm": 0.10298693925142288, "global_step": 31944, "epoch": 301, "lr": 3.414101311742526e-05} +{"train_loss": 0.014032095670700073, "grad_norm": 0.09752659499645233, "global_step": 31945, "epoch": 301, "lr": 3.413820241141077e-05} +{"train_loss": 0.016025768592953682, "grad_norm": 0.07522628456354141, "global_step": 31946, "epoch": 301, "lr": 3.4135391761127734e-05} +{"train_loss": 0.015776310116052628, "grad_norm": 0.10242005437612534, "global_step": 31947, "epoch": 301, "lr": 3.413258116658601e-05} +{"train_loss": 0.014543557539582253, "grad_norm": 0.10768167674541473, "global_step": 31948, "epoch": 301, "lr": 3.412977062779552e-05} +{"train_loss": 0.012694964185357094, "grad_norm": 0.09128671884536743, "global_step": 31949, "epoch": 301, "lr": 3.412696014476611e-05} +{"train_loss": 0.015589042566716671, "grad_norm": 0.07377742975950241, "global_step": 31950, "epoch": 301, "lr": 3.4124149717507645e-05} +{"train_loss": 0.014301292598247528, "grad_norm": 0.10812177509069443, "global_step": 31951, "epoch": 301, "lr": 3.4121339346030037e-05} +{"train_loss": 0.016302481293678284, "grad_norm": 0.08381785452365875, "global_step": 31952, "epoch": 301, "lr": 3.411852903034312e-05} +{"train_loss": 0.012956302613019943, "grad_norm": 0.11440807580947876, "global_step": 31953, "epoch": 301, "lr": 3.41157187704568e-05} +{"train_loss": 0.015641136094927788, "grad_norm": 0.08871814608573914, "global_step": 31954, "epoch": 301, "lr": 3.4112908566380905e-05} +{"train_loss": 0.015765104442834854, "grad_norm": 0.13653644919395447, "global_step": 31955, "epoch": 301, "lr": 3.411009841812537e-05} +{"train_loss": 0.015125283971428871, "grad_norm": 0.11738412082195282, "global_step": 31956, "epoch": 301, "lr": 3.410728832570002e-05} +{"train_loss": 0.016872694715857506, "grad_norm": 0.11726249009370804, "global_step": 31957, "epoch": 301, "lr": 3.410447828911477e-05} +{"train_loss": 0.017654376104474068, "grad_norm": 0.10781160742044449, "global_step": 31958, "epoch": 301, "lr": 3.4101668308379466e-05} +{"train_loss": 0.014271946623921394, "grad_norm": 0.07861822843551636, "global_step": 31959, "epoch": 301, "lr": 3.409885838350397e-05} +{"train_loss": 0.015617379918694496, "grad_norm": 0.08359599858522415, "global_step": 31960, "epoch": 301, "lr": 3.409604851449817e-05} +{"train_loss": 0.015917643904685974, "grad_norm": 0.07045785337686539, "global_step": 31961, "epoch": 301, "lr": 3.4093238701371956e-05} +{"train_loss": 0.014221934601664543, "grad_norm": 0.08430700749158859, "global_step": 31962, "epoch": 301, "lr": 3.409042894413518e-05} +{"train_loss": 0.015673335641622543, "grad_norm": 0.08704748749732971, "global_step": 31963, "epoch": 301, "lr": 3.408761924279772e-05} +{"train_loss": 0.016291063278913498, "grad_norm": 0.11700945347547531, "global_step": 31964, "epoch": 301, "lr": 3.4084809597369424e-05} +{"train_loss": 0.014384093694388866, "grad_norm": 0.12029308080673218, "global_step": 31965, "epoch": 301, "lr": 3.4082000007860205e-05} +{"train_loss": 0.013311772607266903, "grad_norm": 0.12507326900959015, "global_step": 31966, "epoch": 301, "lr": 3.40791904742799e-05} +{"train_loss": 0.014445880427956581, "grad_norm": 0.16507405042648315, "global_step": 31967, "epoch": 301, "lr": 3.40763809966384e-05} +{"train_loss": 0.016225747764110565, "grad_norm": 0.0611228346824646, "global_step": 31968, "epoch": 301, "lr": 3.4073571574945594e-05} +{"train_loss": 0.015557103790342808, "grad_norm": 0.08633020520210266, "global_step": 31969, "epoch": 301, "lr": 3.407076220921131e-05} +{"train_loss": 0.013664157129824162, "grad_norm": 0.1096448004245758, "global_step": 31970, "epoch": 301, "lr": 3.406795289944546e-05} +{"train_loss": 0.015228654257953167, "grad_norm": 0.07950242608785629, "global_step": 31971, "epoch": 301, "lr": 3.406514364565787e-05} +{"train_loss": 0.01666044443845749, "grad_norm": 0.11096253991127014, "global_step": 31972, "epoch": 301, "lr": 3.4062334447858446e-05} +{"train_loss": 0.01736881397664547, "grad_norm": 0.0817875787615776, "global_step": 31973, "epoch": 301, "lr": 3.405952530605705e-05} +{"train_loss": 0.01697111874818802, "grad_norm": 0.07259729504585266, "global_step": 31974, "epoch": 301, "lr": 3.405671622026355e-05} +{"train_loss": 0.013555007055401802, "grad_norm": 0.061090193688869476, "global_step": 31975, "epoch": 301, "lr": 3.405390719048782e-05} +{"train_loss": 0.015409741550683975, "grad_norm": 0.15670651197433472, "global_step": 31976, "epoch": 301, "lr": 3.405109821673971e-05} +{"train_loss": 0.013562167063355446, "grad_norm": 0.10432255268096924, "global_step": 31977, "epoch": 301, "lr": 3.404828929902911e-05} +{"train_loss": 0.017798589542508125, "grad_norm": 0.12345249205827713, "global_step": 31978, "epoch": 301, "lr": 3.404548043736589e-05} +{"train_loss": 0.014270685613155365, "grad_norm": 0.07211952656507492, "global_step": 31979, "epoch": 301, "lr": 3.40426716317599e-05} +{"train_loss": 0.015194700099527836, "grad_norm": 0.09231190383434296, "global_step": 31980, "epoch": 301, "lr": 3.403986288222105e-05} +{"train_loss": 0.01327090710401535, "grad_norm": 0.07584559917449951, "global_step": 31981, "epoch": 301, "lr": 3.4037054188759165e-05} +{"train_loss": 0.01531202532351017, "grad_norm": 0.11135606467723846, "global_step": 31982, "epoch": 301, "lr": 3.4034245551384123e-05} +{"train_loss": 0.015002084895968437, "grad_norm": 0.09554621577262878, "global_step": 31983, "epoch": 301, "lr": 3.4031436970105804e-05} +{"train_loss": 0.015594900585711002, "grad_norm": 0.11260183155536652, "global_step": 31984, "epoch": 301, "lr": 3.4028628444934066e-05} +{"train_loss": 0.015285974368453026, "grad_norm": 0.07536987215280533, "global_step": 31985, "epoch": 301, "lr": 3.402581997587879e-05} +{"train_loss": 0.01498504914343357, "grad_norm": 0.09492817521095276, "global_step": 31986, "epoch": 301, "lr": 3.4023011562949846e-05} +{"train_loss": 0.014052042737603188, "grad_norm": 0.10062411427497864, "global_step": 31987, "epoch": 301, "lr": 3.402020320615708e-05} +{"train_loss": 0.01424568984657526, "grad_norm": 0.08781249076128006, "global_step": 31988, "epoch": 301, "lr": 3.401739490551036e-05} +{"train_loss": 0.018598856404423714, "grad_norm": 0.1448853611946106, "global_step": 31989, "epoch": 301, "lr": 3.4014586661019565e-05} +{"train_loss": 0.01521628350019455, "grad_norm": 0.06580507755279541, "global_step": 31990, "epoch": 301, "lr": 3.4011778472694575e-05} +{"train_loss": 0.013460848480463028, "grad_norm": 0.11116081476211548, "global_step": 31991, "epoch": 301, "lr": 3.400897034054522e-05} +{"train_loss": 0.013305996544659138, "grad_norm": 0.10469537228345871, "global_step": 31992, "epoch": 301, "lr": 3.400616226458142e-05} +{"train_loss": 0.017514601349830627, "grad_norm": 0.0993722528219223, "global_step": 31993, "epoch": 301, "lr": 3.400335424481299e-05} +{"train_loss": 0.01363402046263218, "grad_norm": 0.07843003422021866, "global_step": 31994, "epoch": 301, "lr": 3.400054628124982e-05} +{"train_loss": 0.016287939622998238, "grad_norm": 0.10019250214099884, "global_step": 31995, "epoch": 301, "lr": 3.399773837390177e-05} +{"train_loss": 0.01563863269984722, "grad_norm": 0.09318975359201431, "global_step": 31996, "epoch": 301, "lr": 3.399493052277871e-05} +{"train_loss": 0.014841843396425247, "grad_norm": 0.07371237874031067, "global_step": 31997, "epoch": 301, "lr": 3.399212272789051e-05} +{"train_loss": 0.013212818652391434, "grad_norm": 0.07959289103746414, "global_step": 31998, "epoch": 301, "lr": 3.3989314989247036e-05} +{"train_loss": 0.014867393299937248, "grad_norm": 0.11809657514095306, "global_step": 31999, "epoch": 301, "lr": 3.3986507306858125e-05} +{"train_loss": 0.014174943789839745, "grad_norm": 0.09631205350160599, "global_step": 32000, "epoch": 301, "lr": 3.398369968073367e-05} +{"train_loss": 0.01574854739010334, "grad_norm": 0.076882503926754, "global_step": 32001, "epoch": 301, "lr": 3.3980892110883535e-05} +{"train_loss": 0.017839670181274414, "grad_norm": 0.07687272876501083, "global_step": 32002, "epoch": 301, "lr": 3.3978084597317574e-05} +{"train_loss": 0.0162753127515316, "grad_norm": 0.08140405267477036, "global_step": 32003, "epoch": 301, "lr": 3.3975277140045646e-05} +{"train_loss": 0.014041696675121784, "grad_norm": 0.09452550858259201, "global_step": 32004, "epoch": 301, "lr": 3.397246973907766e-05} +{"train_loss": 0.013472174294292927, "grad_norm": 0.09561508893966675, "global_step": 32005, "epoch": 301, "lr": 3.396966239442341e-05} +{"train_loss": 0.01670379936695099, "grad_norm": 0.07731081545352936, "global_step": 32006, "epoch": 301, "lr": 3.3966855106092806e-05} +{"train_loss": 0.014994846656918526, "grad_norm": 0.08999636769294739, "global_step": 32007, "epoch": 301, "lr": 3.396404787409571e-05} +{"train_loss": 0.015487229451537132, "grad_norm": 0.09625633805990219, "global_step": 32008, "epoch": 301, "lr": 3.3961240698441954e-05} +{"train_loss": 0.012820230796933174, "grad_norm": 0.08351191133260727, "global_step": 32009, "epoch": 301, "lr": 3.395843357914144e-05} +{"train_loss": 0.017353307455778122, "grad_norm": 0.07057182490825653, "global_step": 32010, "epoch": 301, "lr": 3.395562651620401e-05} +{"train_loss": 0.01529308625514496, "grad_norm": 0.09233162552118301, "global_step": 32011, "epoch": 301, "lr": 3.3952819509639534e-05, "val_loss": 0.019685490056872368} +{"train_loss": 0.013290749862790108, "grad_norm": 0.11196403205394745, "global_step": 32012, "epoch": 302, "lr": 3.395001255945786e-05} +{"train_loss": 0.016191866248846054, "grad_norm": 0.10795702785253525, "global_step": 32013, "epoch": 302, "lr": 3.394720566566887e-05} +{"train_loss": 0.013562476262450218, "grad_norm": 0.14061953127384186, "global_step": 32014, "epoch": 302, "lr": 3.394439882828242e-05} +{"train_loss": 0.014285563491284847, "grad_norm": 0.07628914713859558, "global_step": 32015, "epoch": 302, "lr": 3.394159204730838e-05} +{"train_loss": 0.016560180112719536, "grad_norm": 0.09723572432994843, "global_step": 32016, "epoch": 302, "lr": 3.393878532275659e-05} +{"train_loss": 0.015535172075033188, "grad_norm": 0.06461513042449951, "global_step": 32017, "epoch": 302, "lr": 3.3935978654636916e-05} +{"train_loss": 0.015116903930902481, "grad_norm": 0.1286451369524002, "global_step": 32018, "epoch": 302, "lr": 3.393317204295924e-05} +{"train_loss": 0.015323388390243053, "grad_norm": 0.11689965426921844, "global_step": 32019, "epoch": 302, "lr": 3.393036548773341e-05} +{"train_loss": 0.014302872121334076, "grad_norm": 0.05320051312446594, "global_step": 32020, "epoch": 302, "lr": 3.3927558988969264e-05} +{"train_loss": 0.015716571360826492, "grad_norm": 0.0769953802227974, "global_step": 32021, "epoch": 302, "lr": 3.392475254667672e-05} +{"train_loss": 0.016140084713697433, "grad_norm": 0.06648439168930054, "global_step": 32022, "epoch": 302, "lr": 3.3921946160865594e-05} +{"train_loss": 0.01697642169892788, "grad_norm": 0.05500148981809616, "global_step": 32023, "epoch": 302, "lr": 3.391913983154574e-05} +{"train_loss": 0.014426752924919128, "grad_norm": 0.09964733570814133, "global_step": 32024, "epoch": 302, "lr": 3.3916333558727045e-05} +{"train_loss": 0.01346760243177414, "grad_norm": 0.15511134266853333, "global_step": 32025, "epoch": 302, "lr": 3.391352734241936e-05} +{"train_loss": 0.01688542403280735, "grad_norm": 0.08638348430395126, "global_step": 32026, "epoch": 302, "lr": 3.3910721182632546e-05} +{"train_loss": 0.01519264467060566, "grad_norm": 0.11442066729068756, "global_step": 32027, "epoch": 302, "lr": 3.3907915079376476e-05} +{"train_loss": 0.01365046389400959, "grad_norm": 0.07873600721359253, "global_step": 32028, "epoch": 302, "lr": 3.3905109032660973e-05} +{"train_loss": 0.015223813243210316, "grad_norm": 0.0672302320599556, "global_step": 32029, "epoch": 302, "lr": 3.390230304249592e-05} +{"train_loss": 0.013091037981212139, "grad_norm": 0.05744652822613716, "global_step": 32030, "epoch": 302, "lr": 3.3899497108891176e-05} +{"train_loss": 0.014198062941432, "grad_norm": 0.07861826568841934, "global_step": 32031, "epoch": 302, "lr": 3.38966912318566e-05} +{"train_loss": 0.014719514176249504, "grad_norm": 0.057401519268751144, "global_step": 32032, "epoch": 302, "lr": 3.3893885411402035e-05} +{"train_loss": 0.01587778888642788, "grad_norm": 0.06751342862844467, "global_step": 32033, "epoch": 302, "lr": 3.389107964753738e-05} +{"train_loss": 0.013529239222407341, "grad_norm": 0.11127821356058121, "global_step": 32034, "epoch": 302, "lr": 3.388827394027244e-05} +{"train_loss": 0.014790037646889687, "grad_norm": 0.10598785430192947, "global_step": 32035, "epoch": 302, "lr": 3.3885468289617104e-05} +{"train_loss": 0.014464598149061203, "grad_norm": 0.08429554849863052, "global_step": 32036, "epoch": 302, "lr": 3.3882662695581236e-05} +{"train_loss": 0.01543474942445755, "grad_norm": 0.10929833352565765, "global_step": 32037, "epoch": 302, "lr": 3.387985715817466e-05} +{"train_loss": 0.019944004714488983, "grad_norm": 0.081547811627388, "global_step": 32038, "epoch": 302, "lr": 3.387705167740727e-05} +{"train_loss": 0.01597115397453308, "grad_norm": 0.12078128010034561, "global_step": 32039, "epoch": 302, "lr": 3.387424625328892e-05} +{"train_loss": 0.015085543505847454, "grad_norm": 0.1070263534784317, "global_step": 32040, "epoch": 302, "lr": 3.387144088582944e-05} +{"train_loss": 0.014885690994560719, "grad_norm": 0.06262239813804626, "global_step": 32041, "epoch": 302, "lr": 3.38686355750387e-05} +{"train_loss": 0.017562730237841606, "grad_norm": 0.09336145222187042, "global_step": 32042, "epoch": 302, "lr": 3.3865830320926566e-05} +{"train_loss": 0.014301853254437447, "grad_norm": 0.06696203351020813, "global_step": 32043, "epoch": 302, "lr": 3.386302512350289e-05} +{"train_loss": 0.01595771126449108, "grad_norm": 0.06149766594171524, "global_step": 32044, "epoch": 302, "lr": 3.386021998277751e-05} +{"train_loss": 0.015141630545258522, "grad_norm": 0.10303792357444763, "global_step": 32045, "epoch": 302, "lr": 3.385741489876033e-05} +{"train_loss": 0.016499625518918037, "grad_norm": 0.05690845847129822, "global_step": 32046, "epoch": 302, "lr": 3.385460987146115e-05} +{"train_loss": 0.013944216072559357, "grad_norm": 0.0862589180469513, "global_step": 32047, "epoch": 302, "lr": 3.385180490088985e-05} +{"train_loss": 0.015766773372888565, "grad_norm": 0.06751036643981934, "global_step": 32048, "epoch": 302, "lr": 3.3848999987056294e-05} +{"train_loss": 0.016172148287296295, "grad_norm": 0.06595408916473389, "global_step": 32049, "epoch": 302, "lr": 3.3846195129970315e-05} +{"train_loss": 0.016498977318406105, "grad_norm": 0.0964570939540863, "global_step": 32050, "epoch": 302, "lr": 3.384339032964179e-05} +{"train_loss": 0.012051100842654705, "grad_norm": 0.07178674638271332, "global_step": 32051, "epoch": 302, "lr": 3.3840585586080577e-05} +{"train_loss": 0.012440335005521774, "grad_norm": 0.0683087408542633, "global_step": 32052, "epoch": 302, "lr": 3.38377808992965e-05} +{"train_loss": 0.014758751727640629, "grad_norm": 0.06755030155181885, "global_step": 32053, "epoch": 302, "lr": 3.383497626929943e-05} +{"train_loss": 0.01655438542366028, "grad_norm": 0.08200185000896454, "global_step": 32054, "epoch": 302, "lr": 3.383217169609924e-05} +{"train_loss": 0.016658682376146317, "grad_norm": 0.07647895067930222, "global_step": 32055, "epoch": 302, "lr": 3.382936717970575e-05} +{"train_loss": 0.014036980457603931, "grad_norm": 0.06256401538848877, "global_step": 32056, "epoch": 302, "lr": 3.382656272012883e-05} +{"train_loss": 0.015651166439056396, "grad_norm": 0.07963147014379501, "global_step": 32057, "epoch": 302, "lr": 3.382375831737837e-05} +{"train_loss": 0.015036425553262234, "grad_norm": 0.15133382380008698, "global_step": 32058, "epoch": 302, "lr": 3.382095397146415e-05} +{"train_loss": 0.013752298429608345, "grad_norm": 0.08929823338985443, "global_step": 32059, "epoch": 302, "lr": 3.381814968239608e-05} +{"train_loss": 0.01465403288602829, "grad_norm": 0.08016111701726913, "global_step": 32060, "epoch": 302, "lr": 3.3815345450183996e-05} +{"train_loss": 0.018177494406700134, "grad_norm": 0.13596214354038239, "global_step": 32061, "epoch": 302, "lr": 3.381254127483773e-05} +{"train_loss": 0.015160493552684784, "grad_norm": 0.14778771996498108, "global_step": 32062, "epoch": 302, "lr": 3.380973715636719e-05} +{"train_loss": 0.015202658250927925, "grad_norm": 0.13749225437641144, "global_step": 32063, "epoch": 302, "lr": 3.380693309478216e-05} +{"train_loss": 0.015160651877522469, "grad_norm": 0.0923008918762207, "global_step": 32064, "epoch": 302, "lr": 3.380412909009254e-05} +{"train_loss": 0.015736529603600502, "grad_norm": 0.07659858465194702, "global_step": 32065, "epoch": 302, "lr": 3.380132514230816e-05} +{"train_loss": 0.015968402847647667, "grad_norm": 0.07287125289440155, "global_step": 32066, "epoch": 302, "lr": 3.379852125143887e-05} +{"train_loss": 0.015932125970721245, "grad_norm": 0.07084597647190094, "global_step": 32067, "epoch": 302, "lr": 3.379571741749454e-05} +{"train_loss": 0.01478195283561945, "grad_norm": 0.10311227291822433, "global_step": 32068, "epoch": 302, "lr": 3.3792913640485036e-05} +{"train_loss": 0.015592193230986595, "grad_norm": 0.06352411210536957, "global_step": 32069, "epoch": 302, "lr": 3.379010992042016e-05} +{"train_loss": 0.015518365427851677, "grad_norm": 0.08940484374761581, "global_step": 32070, "epoch": 302, "lr": 3.378730625730978e-05} +{"train_loss": 0.0154335405677557, "grad_norm": 0.10519900918006897, "global_step": 32071, "epoch": 302, "lr": 3.378450265116377e-05} +{"train_loss": 0.017024625092744827, "grad_norm": 0.060874052345752716, "global_step": 32072, "epoch": 302, "lr": 3.378169910199196e-05} +{"train_loss": 0.014387388713657856, "grad_norm": 0.07564997673034668, "global_step": 32073, "epoch": 302, "lr": 3.3778895609804195e-05} +{"train_loss": 0.016757110133767128, "grad_norm": 0.11368323117494583, "global_step": 32074, "epoch": 302, "lr": 3.3776092174610375e-05} +{"train_loss": 0.016878826543688774, "grad_norm": 0.0843566507101059, "global_step": 32075, "epoch": 302, "lr": 3.377328879642028e-05} +{"train_loss": 0.01460457593202591, "grad_norm": 0.053748030215501785, "global_step": 32076, "epoch": 302, "lr": 3.37704854752438e-05} +{"train_loss": 0.014709381386637688, "grad_norm": 0.08284750580787659, "global_step": 32077, "epoch": 302, "lr": 3.376768221109077e-05} +{"train_loss": 0.01734847202897072, "grad_norm": 0.05610492080450058, "global_step": 32078, "epoch": 302, "lr": 3.376487900397104e-05} +{"train_loss": 0.015246763825416565, "grad_norm": 0.08635225892066956, "global_step": 32079, "epoch": 302, "lr": 3.376207585389447e-05} +{"train_loss": 0.014854148030281067, "grad_norm": 0.06106133386492729, "global_step": 32080, "epoch": 302, "lr": 3.375927276087092e-05} +{"train_loss": 0.016018178313970566, "grad_norm": 0.09490502625703812, "global_step": 32081, "epoch": 302, "lr": 3.375646972491021e-05} +{"train_loss": 0.014306022785604, "grad_norm": 0.050286054611206055, "global_step": 32082, "epoch": 302, "lr": 3.37536667460222e-05} +{"train_loss": 0.014451757073402405, "grad_norm": 0.05778242647647858, "global_step": 32083, "epoch": 302, "lr": 3.375086382421674e-05} +{"train_loss": 0.016470562666654587, "grad_norm": 0.06502509862184525, "global_step": 32084, "epoch": 302, "lr": 3.374806095950368e-05} +{"train_loss": 0.014230649918317795, "grad_norm": 0.16442358493804932, "global_step": 32085, "epoch": 302, "lr": 3.374525815189286e-05} +{"train_loss": 0.014755290001630783, "grad_norm": 0.12043870240449905, "global_step": 32086, "epoch": 302, "lr": 3.3742455401394156e-05} +{"train_loss": 0.014807924628257751, "grad_norm": 0.06032520532608032, "global_step": 32087, "epoch": 302, "lr": 3.3739652708017365e-05} +{"train_loss": 0.01412068959325552, "grad_norm": 0.0960889384150505, "global_step": 32088, "epoch": 302, "lr": 3.373685007177238e-05} +{"train_loss": 0.01838533580303192, "grad_norm": 0.15955844521522522, "global_step": 32089, "epoch": 302, "lr": 3.373404749266902e-05} +{"train_loss": 0.016350513324141502, "grad_norm": 0.07062949985265732, "global_step": 32090, "epoch": 302, "lr": 3.373124497071714e-05} +{"train_loss": 0.017914574593305588, "grad_norm": 0.09114975482225418, "global_step": 32091, "epoch": 302, "lr": 3.37284425059266e-05} +{"train_loss": 0.014215151779353619, "grad_norm": 0.06805045157670975, "global_step": 32092, "epoch": 302, "lr": 3.3725640098307244e-05} +{"train_loss": 0.016759971156716347, "grad_norm": 0.05921371653676033, "global_step": 32093, "epoch": 302, "lr": 3.3722837747868906e-05} +{"train_loss": 0.013902124017477036, "grad_norm": 0.14173680543899536, "global_step": 32094, "epoch": 302, "lr": 3.3720035454621426e-05} +{"train_loss": 0.014497026801109314, "grad_norm": 0.12294351309537888, "global_step": 32095, "epoch": 302, "lr": 3.371723321857467e-05} +{"train_loss": 0.013340573757886887, "grad_norm": 0.05723404884338379, "global_step": 32096, "epoch": 302, "lr": 3.371443103973848e-05} +{"train_loss": 0.015011055395007133, "grad_norm": 0.18194644153118134, "global_step": 32097, "epoch": 302, "lr": 3.371162891812268e-05} +{"train_loss": 0.013585511595010757, "grad_norm": 0.0899248868227005, "global_step": 32098, "epoch": 302, "lr": 3.370882685373716e-05} +{"train_loss": 0.013579750433564186, "grad_norm": 0.06381408125162125, "global_step": 32099, "epoch": 302, "lr": 3.3706024846591715e-05} +{"train_loss": 0.01761132851243019, "grad_norm": 0.06640179455280304, "global_step": 32100, "epoch": 302, "lr": 3.3703222896696217e-05} +{"train_loss": 0.015481561422348022, "grad_norm": 0.11017037183046341, "global_step": 32101, "epoch": 302, "lr": 3.370042100406051e-05} +{"train_loss": 0.015049956738948822, "grad_norm": 0.09430393576622009, "global_step": 32102, "epoch": 302, "lr": 3.369761916869443e-05} +{"train_loss": 0.015488067641854286, "grad_norm": 0.07384152710437775, "global_step": 32103, "epoch": 302, "lr": 3.369481739060783e-05} +{"train_loss": 0.017266685143113136, "grad_norm": 0.07168170809745789, "global_step": 32104, "epoch": 302, "lr": 3.3692015669810564e-05} +{"train_loss": 0.013272636570036411, "grad_norm": 0.06738676130771637, "global_step": 32105, "epoch": 302, "lr": 3.368921400631245e-05} +{"train_loss": 0.01452798955142498, "grad_norm": 0.06032900512218475, "global_step": 32106, "epoch": 302, "lr": 3.3686412400123344e-05} +{"train_loss": 0.015806084498763084, "grad_norm": 0.06122005730867386, "global_step": 32107, "epoch": 302, "lr": 3.368361085125308e-05} +{"train_loss": 0.01257192250341177, "grad_norm": 0.06562384963035583, "global_step": 32108, "epoch": 302, "lr": 3.3680809359711524e-05} +{"train_loss": 0.017753615975379944, "grad_norm": 0.061319418251514435, "global_step": 32109, "epoch": 302, "lr": 3.36780079255085e-05} +{"train_loss": 0.012835502624511719, "grad_norm": 0.13881005346775055, "global_step": 32110, "epoch": 302, "lr": 3.3675206548653884e-05} +{"train_loss": 0.019390426576137543, "grad_norm": 0.07696311920881271, "global_step": 32111, "epoch": 302, "lr": 3.367240522915745e-05} +{"train_loss": 0.014012012630701065, "grad_norm": 0.1471341848373413, "global_step": 32112, "epoch": 302, "lr": 3.366960396702912e-05} +{"train_loss": 0.016179164871573448, "grad_norm": 0.09648822993040085, "global_step": 32113, "epoch": 302, "lr": 3.366680276227868e-05} +{"train_loss": 0.014288416132330894, "grad_norm": 0.050519973039627075, "global_step": 32114, "epoch": 302, "lr": 3.3664001614915987e-05} +{"train_loss": 0.014728829264640808, "grad_norm": 0.09602296352386475, "global_step": 32115, "epoch": 302, "lr": 3.366120052495092e-05} +{"train_loss": 0.016646169126033783, "grad_norm": 0.0619066096842289, "global_step": 32116, "epoch": 302, "lr": 3.3658399492393256e-05} +{"train_loss": 0.015262617263942957, "grad_norm": 0.1239420548081398, "global_step": 32117, "epoch": 302, "lr": 3.3655598517252885e-05, "val_loss": 0.019840827211737633} +{"train_loss": 0.01599736139178276, "grad_norm": 0.11507352441549301, "global_step": 32118, "epoch": 303, "lr": 3.365279759953963e-05} +{"train_loss": 0.015026044100522995, "grad_norm": 0.08770858496427536, "global_step": 32119, "epoch": 303, "lr": 3.3649996739263326e-05} +{"train_loss": 0.016802335157990456, "grad_norm": 0.0924220085144043, "global_step": 32120, "epoch": 303, "lr": 3.364719593643383e-05} +{"train_loss": 0.013802986592054367, "grad_norm": 0.07594610750675201, "global_step": 32121, "epoch": 303, "lr": 3.3644395191060986e-05} +{"train_loss": 0.017220348119735718, "grad_norm": 0.06322994083166122, "global_step": 32122, "epoch": 303, "lr": 3.3641594503154617e-05} +{"train_loss": 0.013969143852591515, "grad_norm": 0.07614809274673462, "global_step": 32123, "epoch": 303, "lr": 3.3638793872724556e-05} +{"train_loss": 0.016278449445962906, "grad_norm": 0.11776265501976013, "global_step": 32124, "epoch": 303, "lr": 3.363599329978066e-05} +{"train_loss": 0.0163835771381855, "grad_norm": 0.07288555800914764, "global_step": 32125, "epoch": 303, "lr": 3.363319278433279e-05} +{"train_loss": 0.01379456091672182, "grad_norm": 0.06984113156795502, "global_step": 32126, "epoch": 303, "lr": 3.3630392326390737e-05} +{"train_loss": 0.012883033603429794, "grad_norm": 0.08953548967838287, "global_step": 32127, "epoch": 303, "lr": 3.36275919259644e-05} +{"train_loss": 0.013413532637059689, "grad_norm": 0.06641176342964172, "global_step": 32128, "epoch": 303, "lr": 3.362479158306356e-05} +{"train_loss": 0.018691010773181915, "grad_norm": 0.1042971983551979, "global_step": 32129, "epoch": 303, "lr": 3.3621991297698084e-05} +{"train_loss": 0.014868858270347118, "grad_norm": 0.05342818424105644, "global_step": 32130, "epoch": 303, "lr": 3.3619191069877806e-05} +{"train_loss": 0.01643291860818863, "grad_norm": 0.0749429240822792, "global_step": 32131, "epoch": 303, "lr": 3.361639089961256e-05} +{"train_loss": 0.013773880898952484, "grad_norm": 0.054573606699705124, "global_step": 32132, "epoch": 303, "lr": 3.361359078691221e-05} +{"train_loss": 0.015101002529263496, "grad_norm": 0.11800309270620346, "global_step": 32133, "epoch": 303, "lr": 3.361079073178658e-05} +{"train_loss": 0.01686963438987732, "grad_norm": 0.10249659419059753, "global_step": 32134, "epoch": 303, "lr": 3.360799073424549e-05} +{"train_loss": 0.01737954095005989, "grad_norm": 0.14634773135185242, "global_step": 32135, "epoch": 303, "lr": 3.360519079429879e-05} +{"train_loss": 0.015367226675152779, "grad_norm": 0.13179728388786316, "global_step": 32136, "epoch": 303, "lr": 3.3602390911956316e-05} +{"train_loss": 0.015592633746564388, "grad_norm": 0.06497032195329666, "global_step": 32137, "epoch": 303, "lr": 3.359959108722791e-05} +{"train_loss": 0.015875494107604027, "grad_norm": 0.1073560044169426, "global_step": 32138, "epoch": 303, "lr": 3.359679132012342e-05} +{"train_loss": 0.016407325863838196, "grad_norm": 0.11140692979097366, "global_step": 32139, "epoch": 303, "lr": 3.3593991610652675e-05} +{"train_loss": 0.014379400759935379, "grad_norm": 0.0688202828168869, "global_step": 32140, "epoch": 303, "lr": 3.3591191958825494e-05} +{"train_loss": 0.015854328870773315, "grad_norm": 0.06407177448272705, "global_step": 32141, "epoch": 303, "lr": 3.358839236465173e-05} +{"train_loss": 0.0156851839274168, "grad_norm": 0.07698491215705872, "global_step": 32142, "epoch": 303, "lr": 3.358559282814122e-05} +{"train_loss": 0.01535599585622549, "grad_norm": 0.16786691546440125, "global_step": 32143, "epoch": 303, "lr": 3.35827933493038e-05} +{"train_loss": 0.014085294678807259, "grad_norm": 0.08189839124679565, "global_step": 32144, "epoch": 303, "lr": 3.35799939281493e-05} +{"train_loss": 0.016247060149908066, "grad_norm": 0.12501150369644165, "global_step": 32145, "epoch": 303, "lr": 3.357719456468758e-05} +{"train_loss": 0.016190189868211746, "grad_norm": 0.13047575950622559, "global_step": 32146, "epoch": 303, "lr": 3.3574395258928446e-05} +{"train_loss": 0.013369380496442318, "grad_norm": 0.0836142748594284, "global_step": 32147, "epoch": 303, "lr": 3.3571596010881735e-05} +{"train_loss": 0.018740994855761528, "grad_norm": 0.1156095564365387, "global_step": 32148, "epoch": 303, "lr": 3.3568796820557284e-05} +{"train_loss": 0.015224441885948181, "grad_norm": 0.11787385493516922, "global_step": 32149, "epoch": 303, "lr": 3.356599768796495e-05} +{"train_loss": 0.015693701803684235, "grad_norm": 0.1523115336894989, "global_step": 32150, "epoch": 303, "lr": 3.356319861311455e-05} +{"train_loss": 0.015871696174144745, "grad_norm": 0.09287459403276443, "global_step": 32151, "epoch": 303, "lr": 3.3560399596015935e-05} +{"train_loss": 0.012784412130713463, "grad_norm": 0.06929824501276016, "global_step": 32152, "epoch": 303, "lr": 3.355760063667891e-05} +{"train_loss": 0.014943177811801434, "grad_norm": 0.059563908725976944, "global_step": 32153, "epoch": 303, "lr": 3.355480173511332e-05} +{"train_loss": 0.016332492232322693, "grad_norm": 0.09007484465837479, "global_step": 32154, "epoch": 303, "lr": 3.3552002891329015e-05} +{"train_loss": 0.015178181231021881, "grad_norm": 0.06307646632194519, "global_step": 32155, "epoch": 303, "lr": 3.354920410533581e-05} +{"train_loss": 0.015084894374012947, "grad_norm": 0.07049515843391418, "global_step": 32156, "epoch": 303, "lr": 3.3546405377143556e-05} +{"train_loss": 0.014342648908495903, "grad_norm": 0.0660550519824028, "global_step": 32157, "epoch": 303, "lr": 3.354360670676209e-05} +{"train_loss": 0.014433132484555244, "grad_norm": 0.1052607074379921, "global_step": 32158, "epoch": 303, "lr": 3.354080809420122e-05} +{"train_loss": 0.01532072201371193, "grad_norm": 0.05717874690890312, "global_step": 32159, "epoch": 303, "lr": 3.353800953947079e-05} +{"train_loss": 0.012921958230435848, "grad_norm": 0.06973192095756531, "global_step": 32160, "epoch": 303, "lr": 3.353521104258063e-05} +{"train_loss": 0.01437896303832531, "grad_norm": 0.08385390788316727, "global_step": 32161, "epoch": 303, "lr": 3.353241260354059e-05} +{"train_loss": 0.016053853556513786, "grad_norm": 0.09906557947397232, "global_step": 32162, "epoch": 303, "lr": 3.352961422236049e-05} +{"train_loss": 0.015755435451865196, "grad_norm": 0.06878528743982315, "global_step": 32163, "epoch": 303, "lr": 3.352681589905017e-05} +{"train_loss": 0.01539527252316475, "grad_norm": 0.05513794720172882, "global_step": 32164, "epoch": 303, "lr": 3.3524017633619445e-05} +{"train_loss": 0.01517497655004263, "grad_norm": 0.18752004206180573, "global_step": 32165, "epoch": 303, "lr": 3.3521219426078144e-05} +{"train_loss": 0.014405898749828339, "grad_norm": 0.12366374582052231, "global_step": 32166, "epoch": 303, "lr": 3.351842127643613e-05} +{"train_loss": 0.014604130759835243, "grad_norm": 0.06173855811357498, "global_step": 32167, "epoch": 303, "lr": 3.351562318470321e-05} +{"train_loss": 0.01439015194773674, "grad_norm": 0.0685923770070076, "global_step": 32168, "epoch": 303, "lr": 3.351282515088923e-05} +{"train_loss": 0.01550537720322609, "grad_norm": 0.06974390894174576, "global_step": 32169, "epoch": 303, "lr": 3.3510027175003996e-05} +{"train_loss": 0.01406082697212696, "grad_norm": 0.07509218156337738, "global_step": 32170, "epoch": 303, "lr": 3.350722925705736e-05} +{"train_loss": 0.016147062182426453, "grad_norm": 0.12090960890054703, "global_step": 32171, "epoch": 303, "lr": 3.350443139705916e-05} +{"train_loss": 0.01596834883093834, "grad_norm": 0.10032682120800018, "global_step": 32172, "epoch": 303, "lr": 3.3501633595019187e-05} +{"train_loss": 0.013277895748615265, "grad_norm": 0.06254003196954727, "global_step": 32173, "epoch": 303, "lr": 3.349883585094732e-05} +{"train_loss": 0.016244927421212196, "grad_norm": 0.07905314862728119, "global_step": 32174, "epoch": 303, "lr": 3.349603816485338e-05} +{"train_loss": 0.01505146361887455, "grad_norm": 0.10148691385984421, "global_step": 32175, "epoch": 303, "lr": 3.349324053674717e-05} +{"train_loss": 0.014802064746618271, "grad_norm": 0.06421487778425217, "global_step": 32176, "epoch": 303, "lr": 3.349044296663853e-05} +{"train_loss": 0.01756410300731659, "grad_norm": 0.13433410227298737, "global_step": 32177, "epoch": 303, "lr": 3.3487645454537284e-05} +{"train_loss": 0.015722189098596573, "grad_norm": 0.08776362240314484, "global_step": 32178, "epoch": 303, "lr": 3.3484848000453284e-05} +{"train_loss": 0.015138832852244377, "grad_norm": 0.10046114772558212, "global_step": 32179, "epoch": 303, "lr": 3.348205060439635e-05} +{"train_loss": 0.017623938620090485, "grad_norm": 0.06074770912528038, "global_step": 32180, "epoch": 303, "lr": 3.347925326637631e-05} +{"train_loss": 0.01628013700246811, "grad_norm": 0.08719425648450851, "global_step": 32181, "epoch": 303, "lr": 3.3476455986402977e-05} +{"train_loss": 0.014254122972488403, "grad_norm": 0.08583378046751022, "global_step": 32182, "epoch": 303, "lr": 3.3473658764486195e-05} +{"train_loss": 0.016565503552556038, "grad_norm": 0.06333910673856735, "global_step": 32183, "epoch": 303, "lr": 3.347086160063579e-05} +{"train_loss": 0.015355086885392666, "grad_norm": 0.09805046021938324, "global_step": 32184, "epoch": 303, "lr": 3.346806449486158e-05} +{"train_loss": 0.01531315129250288, "grad_norm": 0.08542415499687195, "global_step": 32185, "epoch": 303, "lr": 3.346526744717341e-05} +{"train_loss": 0.013895710930228233, "grad_norm": 0.07921700179576874, "global_step": 32186, "epoch": 303, "lr": 3.3462470457581115e-05} +{"train_loss": 0.015422050841152668, "grad_norm": 0.07482069730758667, "global_step": 32187, "epoch": 303, "lr": 3.3459673526094495e-05} +{"train_loss": 0.01788335293531418, "grad_norm": 0.07491883635520935, "global_step": 32188, "epoch": 303, "lr": 3.345687665272338e-05} +{"train_loss": 0.014571067877113819, "grad_norm": 0.08278591930866241, "global_step": 32189, "epoch": 303, "lr": 3.34540798374776e-05} +{"train_loss": 0.014711705036461353, "grad_norm": 0.08963073790073395, "global_step": 32190, "epoch": 303, "lr": 3.345128308036701e-05} +{"train_loss": 0.01609237864613533, "grad_norm": 0.09197746217250824, "global_step": 32191, "epoch": 303, "lr": 3.34484863814014e-05} +{"train_loss": 0.014741111546754837, "grad_norm": 0.05253641679883003, "global_step": 32192, "epoch": 303, "lr": 3.344568974059063e-05} +{"train_loss": 0.01623690500855446, "grad_norm": 0.07467474788427353, "global_step": 32193, "epoch": 303, "lr": 3.344289315794449e-05} +{"train_loss": 0.014339996501803398, "grad_norm": 0.07225571572780609, "global_step": 32194, "epoch": 303, "lr": 3.3440096633472815e-05} +{"train_loss": 0.014134766533970833, "grad_norm": 0.08986783027648926, "global_step": 32195, "epoch": 303, "lr": 3.3437300167185445e-05} +{"train_loss": 0.015054166316986084, "grad_norm": 0.12611724436283112, "global_step": 32196, "epoch": 303, "lr": 3.3434503759092206e-05} +{"train_loss": 0.014660676941275597, "grad_norm": 0.113817498087883, "global_step": 32197, "epoch": 303, "lr": 3.3431707409202904e-05} +{"train_loss": 0.01572810858488083, "grad_norm": 0.11350835114717484, "global_step": 32198, "epoch": 303, "lr": 3.3428911117527405e-05} +{"train_loss": 0.01687958836555481, "grad_norm": 0.07510510087013245, "global_step": 32199, "epoch": 303, "lr": 3.342611488407549e-05} +{"train_loss": 0.014641143381595612, "grad_norm": 0.11844898015260696, "global_step": 32200, "epoch": 303, "lr": 3.3423318708856996e-05} +{"train_loss": 0.015666227787733078, "grad_norm": 0.0686047375202179, "global_step": 32201, "epoch": 303, "lr": 3.3420522591881744e-05} +{"train_loss": 0.013571678660809994, "grad_norm": 0.08578411489725113, "global_step": 32202, "epoch": 303, "lr": 3.3417726533159564e-05} +{"train_loss": 0.016752567142248154, "grad_norm": 0.05626799911260605, "global_step": 32203, "epoch": 303, "lr": 3.3414930532700294e-05} +{"train_loss": 0.016354337334632874, "grad_norm": 0.09603171795606613, "global_step": 32204, "epoch": 303, "lr": 3.3412134590513754e-05} +{"train_loss": 0.013269515708088875, "grad_norm": 0.05807791277766228, "global_step": 32205, "epoch": 303, "lr": 3.340933870660975e-05} +{"train_loss": 0.01365954801440239, "grad_norm": 0.08670909702777863, "global_step": 32206, "epoch": 303, "lr": 3.34065428809981e-05} +{"train_loss": 0.014293228276073933, "grad_norm": 0.07073263823986053, "global_step": 32207, "epoch": 303, "lr": 3.3403747113688656e-05} +{"train_loss": 0.016517795622348785, "grad_norm": 0.2176242172718048, "global_step": 32208, "epoch": 303, "lr": 3.340095140469123e-05} +{"train_loss": 0.014442750252783298, "grad_norm": 0.06906535476446152, "global_step": 32209, "epoch": 303, "lr": 3.339815575401562e-05} +{"train_loss": 0.015401250682771206, "grad_norm": 0.06756789982318878, "global_step": 32210, "epoch": 303, "lr": 3.339536016167171e-05} +{"train_loss": 0.015538147650659084, "grad_norm": 0.12415358424186707, "global_step": 32211, "epoch": 303, "lr": 3.339256462766926e-05} +{"train_loss": 0.01555521972477436, "grad_norm": 0.12403656542301178, "global_step": 32212, "epoch": 303, "lr": 3.338976915201811e-05} +{"train_loss": 0.016801584511995316, "grad_norm": 0.1421484798192978, "global_step": 32213, "epoch": 303, "lr": 3.338697373472809e-05} +{"train_loss": 0.01551077701151371, "grad_norm": 0.11081891506910324, "global_step": 32214, "epoch": 303, "lr": 3.338417837580903e-05} +{"train_loss": 0.01313549280166626, "grad_norm": 0.11299911141395569, "global_step": 32215, "epoch": 303, "lr": 3.338138307527074e-05} +{"train_loss": 0.013965966179966927, "grad_norm": 0.0672663226723671, "global_step": 32216, "epoch": 303, "lr": 3.3378587833123044e-05} +{"train_loss": 0.01557897962629795, "grad_norm": 0.05919318273663521, "global_step": 32217, "epoch": 303, "lr": 3.3375792649375754e-05} +{"train_loss": 0.016873948276042938, "grad_norm": 0.0719524547457695, "global_step": 32218, "epoch": 303, "lr": 3.337299752403869e-05} +{"train_loss": 0.011636101640760899, "grad_norm": 0.0544278658926487, "global_step": 32219, "epoch": 303, "lr": 3.337020245712169e-05} +{"train_loss": 0.016304973512887955, "grad_norm": 0.10028225183486938, "global_step": 32220, "epoch": 303, "lr": 3.3367407448634566e-05} +{"train_loss": 0.014668429270386696, "grad_norm": 0.12330174446105957, "global_step": 32221, "epoch": 303, "lr": 3.3364612498587156e-05} +{"train_loss": 0.014493001624941826, "grad_norm": 0.08473331481218338, "global_step": 32222, "epoch": 303, "lr": 3.3361817606989254e-05} +{"train_loss": 0.01526954593488349, "grad_norm": 0.09734170138835907, "global_step": 32223, "epoch": 303, "lr": 3.335902277385067e-05, "val_loss": 0.01965739205479622} +{"train_loss": 0.01303795538842678, "grad_norm": 0.07641559839248657, "global_step": 32224, "epoch": 304, "lr": 3.335622799918126e-05} +{"train_loss": 0.013944888487458229, "grad_norm": 0.06495299935340881, "global_step": 32225, "epoch": 304, "lr": 3.3353433282990816e-05} +{"train_loss": 0.016077928245067596, "grad_norm": 0.07574491202831268, "global_step": 32226, "epoch": 304, "lr": 3.335063862528918e-05} +{"train_loss": 0.014467140659689903, "grad_norm": 0.11231739819049835, "global_step": 32227, "epoch": 304, "lr": 3.334784402608616e-05} +{"train_loss": 0.014640863984823227, "grad_norm": 0.08691906183958054, "global_step": 32228, "epoch": 304, "lr": 3.3345049485391574e-05} +{"train_loss": 0.01257143821567297, "grad_norm": 0.0758906751871109, "global_step": 32229, "epoch": 304, "lr": 3.334225500321524e-05} +{"train_loss": 0.014622492715716362, "grad_norm": 0.0898144394159317, "global_step": 32230, "epoch": 304, "lr": 3.333946057956696e-05} +{"train_loss": 0.015064096078276634, "grad_norm": 0.09642442315816879, "global_step": 32231, "epoch": 304, "lr": 3.333666621445659e-05} +{"train_loss": 0.014635543338954449, "grad_norm": 0.08158499747514725, "global_step": 32232, "epoch": 304, "lr": 3.333387190789392e-05} +{"train_loss": 0.014714356511831284, "grad_norm": 0.07362090051174164, "global_step": 32233, "epoch": 304, "lr": 3.333107765988879e-05} +{"train_loss": 0.016184642910957336, "grad_norm": 0.1280646026134491, "global_step": 32234, "epoch": 304, "lr": 3.3328283470450994e-05} +{"train_loss": 0.017055531963706017, "grad_norm": 0.06949084997177124, "global_step": 32235, "epoch": 304, "lr": 3.332548933959034e-05} +{"train_loss": 0.015073252841830254, "grad_norm": 0.13800545036792755, "global_step": 32236, "epoch": 304, "lr": 3.332269526731669e-05} +{"train_loss": 0.017172671854496002, "grad_norm": 0.09678953140974045, "global_step": 32237, "epoch": 304, "lr": 3.331990125363983e-05} +{"train_loss": 0.014122888445854187, "grad_norm": 0.11057348549365997, "global_step": 32238, "epoch": 304, "lr": 3.331710729856957e-05} +{"train_loss": 0.014992139302194118, "grad_norm": 0.06802935898303986, "global_step": 32239, "epoch": 304, "lr": 3.331431340211576e-05} +{"train_loss": 0.015954522415995598, "grad_norm": 0.09003164619207382, "global_step": 32240, "epoch": 304, "lr": 3.331151956428819e-05} +{"train_loss": 0.013785767368972301, "grad_norm": 0.11657574772834778, "global_step": 32241, "epoch": 304, "lr": 3.330872578509668e-05} +{"train_loss": 0.015275136567652225, "grad_norm": 0.06869212538003922, "global_step": 32242, "epoch": 304, "lr": 3.330593206455104e-05} +{"train_loss": 0.017118927091360092, "grad_norm": 0.08030978590250015, "global_step": 32243, "epoch": 304, "lr": 3.3303138402661107e-05} +{"train_loss": 0.014399808831512928, "grad_norm": 0.06281294673681259, "global_step": 32244, "epoch": 304, "lr": 3.3300344799436686e-05} +{"train_loss": 0.015250023454427719, "grad_norm": 0.11141754686832428, "global_step": 32245, "epoch": 304, "lr": 3.329755125488759e-05} +{"train_loss": 0.014757795259356499, "grad_norm": 0.11030477285385132, "global_step": 32246, "epoch": 304, "lr": 3.329475776902364e-05} +{"train_loss": 0.015176557004451752, "grad_norm": 0.11794234067201614, "global_step": 32247, "epoch": 304, "lr": 3.3291964341854624e-05} +{"train_loss": 0.016291476786136627, "grad_norm": 0.09941504895687103, "global_step": 32248, "epoch": 304, "lr": 3.328917097339039e-05} +{"train_loss": 0.015944059938192368, "grad_norm": 0.08570542931556702, "global_step": 32249, "epoch": 304, "lr": 3.328637766364075e-05} +{"train_loss": 0.01747111976146698, "grad_norm": 0.10751368850469589, "global_step": 32250, "epoch": 304, "lr": 3.3283584412615496e-05} +{"train_loss": 0.013132009655237198, "grad_norm": 0.10410843789577484, "global_step": 32251, "epoch": 304, "lr": 3.328079122032449e-05} +{"train_loss": 0.015937281772494316, "grad_norm": 0.07139819860458374, "global_step": 32252, "epoch": 304, "lr": 3.327799808677747e-05} +{"train_loss": 0.01714077591896057, "grad_norm": 0.06281809508800507, "global_step": 32253, "epoch": 304, "lr": 3.327520501198432e-05} +{"train_loss": 0.01396813802421093, "grad_norm": 0.06060374155640602, "global_step": 32254, "epoch": 304, "lr": 3.3272411995954804e-05} +{"train_loss": 0.013422757387161255, "grad_norm": 0.121741883456707, "global_step": 32255, "epoch": 304, "lr": 3.3269619038698777e-05} +{"train_loss": 0.016335144639015198, "grad_norm": 0.09070092439651489, "global_step": 32256, "epoch": 304, "lr": 3.326682614022603e-05} +{"train_loss": 0.017619499936699867, "grad_norm": 0.07478884607553482, "global_step": 32257, "epoch": 304, "lr": 3.326403330054638e-05} +{"train_loss": 0.01505209319293499, "grad_norm": 0.14288505911827087, "global_step": 32258, "epoch": 304, "lr": 3.3261240519669637e-05} +{"train_loss": 0.012507718056440353, "grad_norm": 0.06764167547225952, "global_step": 32259, "epoch": 304, "lr": 3.3258447797605595e-05} +{"train_loss": 0.017414720728993416, "grad_norm": 0.06685039401054382, "global_step": 32260, "epoch": 304, "lr": 3.325565513436411e-05} +{"train_loss": 0.014368069358170033, "grad_norm": 0.1791514754295349, "global_step": 32261, "epoch": 304, "lr": 3.325286252995496e-05} +{"train_loss": 0.01370842382311821, "grad_norm": 0.06368203461170197, "global_step": 32262, "epoch": 304, "lr": 3.325006998438796e-05} +{"train_loss": 0.014578805305063725, "grad_norm": 0.06360533088445663, "global_step": 32263, "epoch": 304, "lr": 3.3247277497672975e-05} +{"train_loss": 0.01451193355023861, "grad_norm": 0.10600647330284119, "global_step": 32264, "epoch": 304, "lr": 3.324448506981973e-05} +{"train_loss": 0.01581159606575966, "grad_norm": 0.07055855542421341, "global_step": 32265, "epoch": 304, "lr": 3.324169270083809e-05} +{"train_loss": 0.016433190554380417, "grad_norm": 0.06658890843391418, "global_step": 32266, "epoch": 304, "lr": 3.323890039073785e-05} +{"train_loss": 0.015971794724464417, "grad_norm": 0.06580767035484314, "global_step": 32267, "epoch": 304, "lr": 3.323610813952881e-05} +{"train_loss": 0.014932923018932343, "grad_norm": 0.060935474932193756, "global_step": 32268, "epoch": 304, "lr": 3.3233315947220824e-05} +{"train_loss": 0.017771467566490173, "grad_norm": 0.07537975907325745, "global_step": 32269, "epoch": 304, "lr": 3.323052381382367e-05} +{"train_loss": 0.0150197334587574, "grad_norm": 0.09894587099552155, "global_step": 32270, "epoch": 304, "lr": 3.322773173934715e-05} +{"train_loss": 0.013849783688783646, "grad_norm": 0.07913800328969955, "global_step": 32271, "epoch": 304, "lr": 3.322493972380107e-05} +{"train_loss": 0.014260891824960709, "grad_norm": 0.0809478908777237, "global_step": 32272, "epoch": 304, "lr": 3.322214776719529e-05} +{"train_loss": 0.01606801524758339, "grad_norm": 0.07872264832258224, "global_step": 32273, "epoch": 304, "lr": 3.321935586953957e-05} +{"train_loss": 0.01494352426379919, "grad_norm": 0.11946652084589005, "global_step": 32274, "epoch": 304, "lr": 3.321656403084375e-05} +{"train_loss": 0.018945038318634033, "grad_norm": 0.10502852499485016, "global_step": 32275, "epoch": 304, "lr": 3.3213772251117615e-05} +{"train_loss": 0.015556962229311466, "grad_norm": 0.08905521035194397, "global_step": 32276, "epoch": 304, "lr": 3.321098053037097e-05} +{"train_loss": 0.014684812165796757, "grad_norm": 0.11201456934213638, "global_step": 32277, "epoch": 304, "lr": 3.320818886861365e-05} +{"train_loss": 0.01615484617650509, "grad_norm": 0.09291964769363403, "global_step": 32278, "epoch": 304, "lr": 3.320539726585545e-05} +{"train_loss": 0.016160590574145317, "grad_norm": 0.06833938509225845, "global_step": 32279, "epoch": 304, "lr": 3.3202605722106186e-05} +{"train_loss": 0.016624854877591133, "grad_norm": 0.08040252327919006, "global_step": 32280, "epoch": 304, "lr": 3.319981423737566e-05} +{"train_loss": 0.015261602587997913, "grad_norm": 0.075007863342762, "global_step": 32281, "epoch": 304, "lr": 3.319702281167369e-05} +{"train_loss": 0.0156838558614254, "grad_norm": 0.12841349840164185, "global_step": 32282, "epoch": 304, "lr": 3.3194231445010056e-05} +{"train_loss": 0.014369413256645203, "grad_norm": 0.08554645627737045, "global_step": 32283, "epoch": 304, "lr": 3.319144013739458e-05} +{"train_loss": 0.013680433854460716, "grad_norm": 0.072539784014225, "global_step": 32284, "epoch": 304, "lr": 3.318864888883708e-05} +{"train_loss": 0.014839636161923409, "grad_norm": 0.056162625551223755, "global_step": 32285, "epoch": 304, "lr": 3.3185857699347365e-05} +{"train_loss": 0.01949971541762352, "grad_norm": 0.10854046046733856, "global_step": 32286, "epoch": 304, "lr": 3.318306656893524e-05} +{"train_loss": 0.012288236990571022, "grad_norm": 0.1103755310177803, "global_step": 32287, "epoch": 304, "lr": 3.31802754976105e-05} +{"train_loss": 0.01253365259617567, "grad_norm": 0.059552498161792755, "global_step": 32288, "epoch": 304, "lr": 3.3177484485382935e-05} +{"train_loss": 0.016888760030269623, "grad_norm": 0.057981304824352264, "global_step": 32289, "epoch": 304, "lr": 3.31746935322624e-05} +{"train_loss": 0.013340462930500507, "grad_norm": 0.11671269685029984, "global_step": 32290, "epoch": 304, "lr": 3.3171902638258664e-05} +{"train_loss": 0.01618720032274723, "grad_norm": 0.09389884024858475, "global_step": 32291, "epoch": 304, "lr": 3.3169111803381535e-05} +{"train_loss": 0.016831597313284874, "grad_norm": 0.1255989372730255, "global_step": 32292, "epoch": 304, "lr": 3.316632102764086e-05} +{"train_loss": 0.014349746517837048, "grad_norm": 0.18516211211681366, "global_step": 32293, "epoch": 304, "lr": 3.316353031104638e-05} +{"train_loss": 0.013083324767649174, "grad_norm": 0.0891987532377243, "global_step": 32294, "epoch": 304, "lr": 3.316073965360794e-05} +{"train_loss": 0.016060415655374527, "grad_norm": 0.09805813431739807, "global_step": 32295, "epoch": 304, "lr": 3.3157949055335345e-05} +{"train_loss": 0.015532654710114002, "grad_norm": 0.06451055407524109, "global_step": 32296, "epoch": 304, "lr": 3.315515851623838e-05} +{"train_loss": 0.014762119390070438, "grad_norm": 0.08207471668720245, "global_step": 32297, "epoch": 304, "lr": 3.315236803632687e-05} +{"train_loss": 0.015331012196838856, "grad_norm": 0.07387778162956238, "global_step": 32298, "epoch": 304, "lr": 3.3149577615610626e-05} +{"train_loss": 0.016860565170645714, "grad_norm": 0.11396219581365585, "global_step": 32299, "epoch": 304, "lr": 3.314678725409942e-05} +{"train_loss": 0.017667148262262344, "grad_norm": 0.08885044604539871, "global_step": 32300, "epoch": 304, "lr": 3.314399695180307e-05} +{"train_loss": 0.016604311764240265, "grad_norm": 0.06214848905801773, "global_step": 32301, "epoch": 304, "lr": 3.31412067087314e-05} +{"train_loss": 0.015529442578554153, "grad_norm": 0.10312959551811218, "global_step": 32302, "epoch": 304, "lr": 3.3138416524894186e-05} +{"train_loss": 0.012308729812502861, "grad_norm": 0.07345784455537796, "global_step": 32303, "epoch": 304, "lr": 3.313562640030124e-05} +{"train_loss": 0.01463785395026207, "grad_norm": 0.06573005020618439, "global_step": 32304, "epoch": 304, "lr": 3.3132836334962405e-05} +{"train_loss": 0.015176007524132729, "grad_norm": 0.06548170745372772, "global_step": 32305, "epoch": 304, "lr": 3.3130046328887405e-05} +{"train_loss": 0.013987581245601177, "grad_norm": 0.07257638871669769, "global_step": 32306, "epoch": 304, "lr": 3.3127256382086103e-05} +{"train_loss": 0.012403886765241623, "grad_norm": 0.05309024453163147, "global_step": 32307, "epoch": 304, "lr": 3.312446649456828e-05} +{"train_loss": 0.012769155204296112, "grad_norm": 0.0783315971493721, "global_step": 32308, "epoch": 304, "lr": 3.3121676666343736e-05} +{"train_loss": 0.01394819188863039, "grad_norm": 0.051708370447158813, "global_step": 32309, "epoch": 304, "lr": 3.311888689742229e-05} +{"train_loss": 0.014277624897658825, "grad_norm": 0.06753595918416977, "global_step": 32310, "epoch": 304, "lr": 3.3116097187813745e-05} +{"train_loss": 0.014613157138228416, "grad_norm": 0.0703514814376831, "global_step": 32311, "epoch": 304, "lr": 3.3113307537527873e-05} +{"train_loss": 0.015345297753810883, "grad_norm": 0.170551598072052, "global_step": 32312, "epoch": 304, "lr": 3.311051794657449e-05} +{"train_loss": 0.014131803065538406, "grad_norm": 0.07180668413639069, "global_step": 32313, "epoch": 304, "lr": 3.310772841496342e-05} +{"train_loss": 0.01581558771431446, "grad_norm": 0.07498206198215485, "global_step": 32314, "epoch": 304, "lr": 3.310493894270444e-05} +{"train_loss": 0.018901415169239044, "grad_norm": 0.07194625586271286, "global_step": 32315, "epoch": 304, "lr": 3.310214952980737e-05} +{"train_loss": 0.015556694939732552, "grad_norm": 0.07729993760585785, "global_step": 32316, "epoch": 304, "lr": 3.309936017628198e-05} +{"train_loss": 0.01586413010954857, "grad_norm": 0.07190971076488495, "global_step": 32317, "epoch": 304, "lr": 3.309657088213808e-05} +{"train_loss": 0.019070979207754135, "grad_norm": 0.12489617615938187, "global_step": 32318, "epoch": 304, "lr": 3.30937816473855e-05} +{"train_loss": 0.015763115137815475, "grad_norm": 0.11057507991790771, "global_step": 32319, "epoch": 304, "lr": 3.309099247203401e-05} +{"train_loss": 0.015033738687634468, "grad_norm": 0.11991319060325623, "global_step": 32320, "epoch": 304, "lr": 3.3088203356093404e-05} +{"train_loss": 0.015761882066726685, "grad_norm": 0.06733139604330063, "global_step": 32321, "epoch": 304, "lr": 3.3085414299573536e-05} +{"train_loss": 0.015448886901140213, "grad_norm": 0.07836414128541946, "global_step": 32322, "epoch": 304, "lr": 3.3082625302484135e-05} +{"train_loss": 0.018236655741930008, "grad_norm": 0.08944923430681229, "global_step": 32323, "epoch": 304, "lr": 3.307983636483504e-05} +{"train_loss": 0.016878556460142136, "grad_norm": 0.1536586880683899, "global_step": 32324, "epoch": 304, "lr": 3.307704748663604e-05} +{"train_loss": 0.01349297072738409, "grad_norm": 0.05124272033572197, "global_step": 32325, "epoch": 304, "lr": 3.3074258667896925e-05} +{"train_loss": 0.015501584857702255, "grad_norm": 0.09970247000455856, "global_step": 32326, "epoch": 304, "lr": 3.3071469908627515e-05} +{"train_loss": 0.014055175706744194, "grad_norm": 0.1185511127114296, "global_step": 32327, "epoch": 304, "lr": 3.3068681208837605e-05} +{"train_loss": 0.013131547719240189, "grad_norm": 0.08694758266210556, "global_step": 32328, "epoch": 304, "lr": 3.306589256853698e-05} +{"train_loss": 0.015272684632536938, "grad_norm": 0.18196669220924377, "global_step": 32329, "epoch": 304, "lr": 3.3063103987735433e-05, "val_loss": 0.019961904734373093} +{"train_loss": 0.0164768286049366, "grad_norm": 0.06256529688835144, "global_step": 32330, "epoch": 305, "lr": 3.306031546644278e-05} +{"train_loss": 0.016722286120057106, "grad_norm": 0.056973643600940704, "global_step": 32331, "epoch": 305, "lr": 3.305752700466882e-05} +{"train_loss": 0.015814268961548805, "grad_norm": 0.07192496955394745, "global_step": 32332, "epoch": 305, "lr": 3.305473860242333e-05} +{"train_loss": 0.01450759544968605, "grad_norm": 0.11572393774986267, "global_step": 32333, "epoch": 305, "lr": 3.305195025971614e-05} +{"train_loss": 0.017067154869437218, "grad_norm": 0.1167244240641594, "global_step": 32334, "epoch": 305, "lr": 3.3049161976557006e-05} +{"train_loss": 0.01516677625477314, "grad_norm": 0.08491145819425583, "global_step": 32335, "epoch": 305, "lr": 3.304637375295575e-05} +{"train_loss": 0.01744496449828148, "grad_norm": 0.06751956045627594, "global_step": 32336, "epoch": 305, "lr": 3.304358558892216e-05} +{"train_loss": 0.018231898546218872, "grad_norm": 0.14016473293304443, "global_step": 32337, "epoch": 305, "lr": 3.304079748446603e-05} +{"train_loss": 0.013718023896217346, "grad_norm": 0.10885407775640488, "global_step": 32338, "epoch": 305, "lr": 3.303800943959718e-05} +{"train_loss": 0.015587632544338703, "grad_norm": 0.07728718966245651, "global_step": 32339, "epoch": 305, "lr": 3.3035221454325396e-05} +{"train_loss": 0.017125960439443588, "grad_norm": 0.08009462058544159, "global_step": 32340, "epoch": 305, "lr": 3.3032433528660445e-05} +{"train_loss": 0.01305202767252922, "grad_norm": 0.09958255290985107, "global_step": 32341, "epoch": 305, "lr": 3.302964566261214e-05} +{"train_loss": 0.012305326759815216, "grad_norm": 0.06680887192487717, "global_step": 32342, "epoch": 305, "lr": 3.30268578561903e-05} +{"train_loss": 0.015981057658791542, "grad_norm": 0.08470466732978821, "global_step": 32343, "epoch": 305, "lr": 3.302407010940468e-05} +{"train_loss": 0.016309265047311783, "grad_norm": 0.08650656044483185, "global_step": 32344, "epoch": 305, "lr": 3.30212824222651e-05} +{"train_loss": 0.012946061789989471, "grad_norm": 0.14638535678386688, "global_step": 32345, "epoch": 305, "lr": 3.3018494794781375e-05} +{"train_loss": 0.01795421727001667, "grad_norm": 0.1269211322069168, "global_step": 32346, "epoch": 305, "lr": 3.3015707226963246e-05} +{"train_loss": 0.01484011672437191, "grad_norm": 0.07856947928667068, "global_step": 32347, "epoch": 305, "lr": 3.301291971882055e-05} +{"train_loss": 0.014854634180665016, "grad_norm": 0.06345666944980621, "global_step": 32348, "epoch": 305, "lr": 3.301013227036305e-05} +{"train_loss": 0.01444096490740776, "grad_norm": 0.09330756217241287, "global_step": 32349, "epoch": 305, "lr": 3.300734488160056e-05} +{"train_loss": 0.013989794999361038, "grad_norm": 0.08209046721458435, "global_step": 32350, "epoch": 305, "lr": 3.3004557552542876e-05} +{"train_loss": 0.016865773126482964, "grad_norm": 0.08108662068843842, "global_step": 32351, "epoch": 305, "lr": 3.30017702831998e-05} +{"train_loss": 0.013997957110404968, "grad_norm": 0.056665219366550446, "global_step": 32352, "epoch": 305, "lr": 3.299898307358109e-05} +{"train_loss": 0.014468098059296608, "grad_norm": 0.05295948311686516, "global_step": 32353, "epoch": 305, "lr": 3.299619592369657e-05} +{"train_loss": 0.015972090885043144, "grad_norm": 0.11628449708223343, "global_step": 32354, "epoch": 305, "lr": 3.299340883355601e-05} +{"train_loss": 0.015989363193511963, "grad_norm": 0.07147860527038574, "global_step": 32355, "epoch": 305, "lr": 3.299062180316923e-05} +{"train_loss": 0.015419228933751583, "grad_norm": 0.06365826725959778, "global_step": 32356, "epoch": 305, "lr": 3.2987834832545996e-05} +{"train_loss": 0.014630688354372978, "grad_norm": 0.11391826719045639, "global_step": 32357, "epoch": 305, "lr": 3.2985047921696136e-05} +{"train_loss": 0.019304784014821053, "grad_norm": 0.09991130977869034, "global_step": 32358, "epoch": 305, "lr": 3.298226107062939e-05} +{"train_loss": 0.014965997077524662, "grad_norm": 0.07463239878416061, "global_step": 32359, "epoch": 305, "lr": 3.297947427935559e-05} +{"train_loss": 0.01579519733786583, "grad_norm": 0.11911933124065399, "global_step": 32360, "epoch": 305, "lr": 3.2976687547884515e-05} +{"train_loss": 0.017145084217190742, "grad_norm": 0.06767947971820831, "global_step": 32361, "epoch": 305, "lr": 3.297390087622595e-05} +{"train_loss": 0.015127784572541714, "grad_norm": 0.0733497366309166, "global_step": 32362, "epoch": 305, "lr": 3.29711142643897e-05} +{"train_loss": 0.014840767718851566, "grad_norm": 0.09900712966918945, "global_step": 32363, "epoch": 305, "lr": 3.2968327712385556e-05} +{"train_loss": 0.01382514275610447, "grad_norm": 0.1142122745513916, "global_step": 32364, "epoch": 305, "lr": 3.2965541220223295e-05} +{"train_loss": 0.014858867973089218, "grad_norm": 0.06717846542596817, "global_step": 32365, "epoch": 305, "lr": 3.296275478791271e-05} +{"train_loss": 0.015817856416106224, "grad_norm": 0.0706125870347023, "global_step": 32366, "epoch": 305, "lr": 3.295996841546359e-05} +{"train_loss": 0.013379653915762901, "grad_norm": 0.11011151969432831, "global_step": 32367, "epoch": 305, "lr": 3.295718210288574e-05} +{"train_loss": 0.01465971115976572, "grad_norm": 0.053132835775613785, "global_step": 32368, "epoch": 305, "lr": 3.295439585018895e-05} +{"train_loss": 0.012971587479114532, "grad_norm": 0.055798593908548355, "global_step": 32369, "epoch": 305, "lr": 3.2951609657382985e-05} +{"train_loss": 0.013158680871129036, "grad_norm": 0.17905940115451813, "global_step": 32370, "epoch": 305, "lr": 3.2948823524477646e-05} +{"train_loss": 0.014862073585391045, "grad_norm": 0.07354993373155594, "global_step": 32371, "epoch": 305, "lr": 3.294603745148273e-05} +{"train_loss": 0.017617030069231987, "grad_norm": 0.08785536140203476, "global_step": 32372, "epoch": 305, "lr": 3.2943251438408024e-05} +{"train_loss": 0.01151351910084486, "grad_norm": 0.12685611844062805, "global_step": 32373, "epoch": 305, "lr": 3.2940465485263304e-05} +{"train_loss": 0.014468220062553883, "grad_norm": 0.09911743551492691, "global_step": 32374, "epoch": 305, "lr": 3.29376795920584e-05} +{"train_loss": 0.01481361873447895, "grad_norm": 0.0751509815454483, "global_step": 32375, "epoch": 305, "lr": 3.2934893758803044e-05} +{"train_loss": 0.01656140387058258, "grad_norm": 0.07822650671005249, "global_step": 32376, "epoch": 305, "lr": 3.2932107985507054e-05} +{"train_loss": 0.013546045869588852, "grad_norm": 0.0734991803765297, "global_step": 32377, "epoch": 305, "lr": 3.292932227218022e-05} +{"train_loss": 0.013038041070103645, "grad_norm": 0.06919499486684799, "global_step": 32378, "epoch": 305, "lr": 3.292653661883231e-05} +{"train_loss": 0.015560328960418701, "grad_norm": 0.07729483395814896, "global_step": 32379, "epoch": 305, "lr": 3.292375102547314e-05} +{"train_loss": 0.01494187954813242, "grad_norm": 0.09410852938890457, "global_step": 32380, "epoch": 305, "lr": 3.2920965492112484e-05} +{"train_loss": 0.014887017197906971, "grad_norm": 0.10988906770944595, "global_step": 32381, "epoch": 305, "lr": 3.2918180018760124e-05} +{"train_loss": 0.013676708564162254, "grad_norm": 0.056035950779914856, "global_step": 32382, "epoch": 305, "lr": 3.2915394605425835e-05} +{"train_loss": 0.01614060439169407, "grad_norm": 0.06933876872062683, "global_step": 32383, "epoch": 305, "lr": 3.291260925211943e-05} +{"train_loss": 0.014121243730187416, "grad_norm": 0.07348062843084335, "global_step": 32384, "epoch": 305, "lr": 3.290982395885069e-05} +{"train_loss": 0.017318198457360268, "grad_norm": 0.07820666581392288, "global_step": 32385, "epoch": 305, "lr": 3.290703872562939e-05} +{"train_loss": 0.015933088958263397, "grad_norm": 0.11322500556707382, "global_step": 32386, "epoch": 305, "lr": 3.290425355246535e-05} +{"train_loss": 0.015475446358323097, "grad_norm": 0.06212056055665016, "global_step": 32387, "epoch": 305, "lr": 3.290146843936829e-05} +{"train_loss": 0.01776960678398609, "grad_norm": 0.11261296272277832, "global_step": 32388, "epoch": 305, "lr": 3.289868338634805e-05} +{"train_loss": 0.015579340048134327, "grad_norm": 0.07716884464025497, "global_step": 32389, "epoch": 305, "lr": 3.289589839341439e-05} +{"train_loss": 0.017187712714076042, "grad_norm": 0.0688491240143776, "global_step": 32390, "epoch": 305, "lr": 3.289311346057711e-05} +{"train_loss": 0.012980366125702858, "grad_norm": 0.05239873751997948, "global_step": 32391, "epoch": 305, "lr": 3.289032858784599e-05} +{"train_loss": 0.01316151861101389, "grad_norm": 0.058501891791820526, "global_step": 32392, "epoch": 305, "lr": 3.288754377523083e-05} +{"train_loss": 0.016079697757959366, "grad_norm": 0.06127137318253517, "global_step": 32393, "epoch": 305, "lr": 3.2884759022741386e-05} +{"train_loss": 0.014966299757361412, "grad_norm": 0.0638287365436554, "global_step": 32394, "epoch": 305, "lr": 3.2881974330387454e-05} +{"train_loss": 0.01271776482462883, "grad_norm": 0.071234792470932, "global_step": 32395, "epoch": 305, "lr": 3.287918969817881e-05} +{"train_loss": 0.01794583350419998, "grad_norm": 0.0684530958533287, "global_step": 32396, "epoch": 305, "lr": 3.287640512612526e-05} +{"train_loss": 0.013280600309371948, "grad_norm": 0.10500936955213547, "global_step": 32397, "epoch": 305, "lr": 3.287362061423658e-05} +{"train_loss": 0.017346521839499474, "grad_norm": 0.09371131658554077, "global_step": 32398, "epoch": 305, "lr": 3.287083616252256e-05} +{"train_loss": 0.014892911538481712, "grad_norm": 0.0974913239479065, "global_step": 32399, "epoch": 305, "lr": 3.286805177099293e-05} +{"train_loss": 0.013075603172183037, "grad_norm": 0.07535095512866974, "global_step": 32400, "epoch": 305, "lr": 3.2865267439657555e-05} +{"train_loss": 0.015286182053387165, "grad_norm": 0.08260506391525269, "global_step": 32401, "epoch": 305, "lr": 3.286248316852616e-05} +{"train_loss": 0.017631087452173233, "grad_norm": 0.08106601983308792, "global_step": 32402, "epoch": 305, "lr": 3.285969895760854e-05} +{"train_loss": 0.01945757307112217, "grad_norm": 0.1126921996474266, "global_step": 32403, "epoch": 305, "lr": 3.28569148069145e-05} +{"train_loss": 0.01529921218752861, "grad_norm": 0.0853162482380867, "global_step": 32404, "epoch": 305, "lr": 3.285413071645381e-05} +{"train_loss": 0.01707892306149006, "grad_norm": 0.11025966703891754, "global_step": 32405, "epoch": 305, "lr": 3.285134668623624e-05} +{"train_loss": 0.015260168351233006, "grad_norm": 0.1603931039571762, "global_step": 32406, "epoch": 305, "lr": 3.284856271627158e-05} +{"train_loss": 0.01595400460064411, "grad_norm": 0.08524734526872635, "global_step": 32407, "epoch": 305, "lr": 3.2845778806569604e-05} +{"train_loss": 0.01570849120616913, "grad_norm": 0.05224911496043205, "global_step": 32408, "epoch": 305, "lr": 3.28429949571401e-05} +{"train_loss": 0.013123482465744019, "grad_norm": 0.13377825915813446, "global_step": 32409, "epoch": 305, "lr": 3.2840211167992866e-05} +{"train_loss": 0.01569775678217411, "grad_norm": 0.09983460605144501, "global_step": 32410, "epoch": 305, "lr": 3.283742743913767e-05} +{"train_loss": 0.014447847381234169, "grad_norm": 0.09742370247840881, "global_step": 32411, "epoch": 305, "lr": 3.283464377058427e-05} +{"train_loss": 0.013829870149493217, "grad_norm": 0.1069311797618866, "global_step": 32412, "epoch": 305, "lr": 3.2831860162342474e-05} +{"train_loss": 0.01750783622264862, "grad_norm": 0.06502064317464828, "global_step": 32413, "epoch": 305, "lr": 3.282907661442206e-05} +{"train_loss": 0.014519309625029564, "grad_norm": 0.06594088673591614, "global_step": 32414, "epoch": 305, "lr": 3.28262931268328e-05} +{"train_loss": 0.014438867568969727, "grad_norm": 0.059574928134679794, "global_step": 32415, "epoch": 305, "lr": 3.2823509699584474e-05} +{"train_loss": 0.0159807987511158, "grad_norm": 0.08653969317674637, "global_step": 32416, "epoch": 305, "lr": 3.282072633268688e-05} +{"train_loss": 0.015937115997076035, "grad_norm": 0.07532792538404465, "global_step": 32417, "epoch": 305, "lr": 3.281794302614977e-05} +{"train_loss": 0.0144953029230237, "grad_norm": 0.10596828162670135, "global_step": 32418, "epoch": 305, "lr": 3.281515977998294e-05} +{"train_loss": 0.013740051537752151, "grad_norm": 0.09055279195308685, "global_step": 32419, "epoch": 305, "lr": 3.281237659419616e-05} +{"train_loss": 0.012273464351892471, "grad_norm": 0.09075769037008286, "global_step": 32420, "epoch": 305, "lr": 3.280959346879921e-05} +{"train_loss": 0.014985155314207077, "grad_norm": 0.18182502686977386, "global_step": 32421, "epoch": 305, "lr": 3.280681040380189e-05} +{"train_loss": 0.015054075047373772, "grad_norm": 0.127905935049057, "global_step": 32422, "epoch": 305, "lr": 3.280402739921395e-05} +{"train_loss": 0.014514757320284843, "grad_norm": 0.061840128153562546, "global_step": 32423, "epoch": 305, "lr": 3.280124445504519e-05} +{"train_loss": 0.014413060620427132, "grad_norm": 0.08770355582237244, "global_step": 32424, "epoch": 305, "lr": 3.279846157130535e-05} +{"train_loss": 0.015482492744922638, "grad_norm": 0.08452870696783066, "global_step": 32425, "epoch": 305, "lr": 3.2795678748004253e-05} +{"train_loss": 0.015266595408320427, "grad_norm": 0.12063273787498474, "global_step": 32426, "epoch": 305, "lr": 3.279289598515166e-05} +{"train_loss": 0.013921907171607018, "grad_norm": 0.06191801652312279, "global_step": 32427, "epoch": 305, "lr": 3.2790113282757364e-05} +{"train_loss": 0.01859811693429947, "grad_norm": 0.10072503983974457, "global_step": 32428, "epoch": 305, "lr": 3.2787330640831097e-05} +{"train_loss": 0.017021194100379944, "grad_norm": 0.07233145833015442, "global_step": 32429, "epoch": 305, "lr": 3.278454805938268e-05} +{"train_loss": 0.017491862177848816, "grad_norm": 0.12488329410552979, "global_step": 32430, "epoch": 305, "lr": 3.278176553842186e-05} +{"train_loss": 0.01583937555551529, "grad_norm": 0.06800978630781174, "global_step": 32431, "epoch": 305, "lr": 3.277898307795844e-05} +{"train_loss": 0.01662510260939598, "grad_norm": 0.07059376686811447, "global_step": 32432, "epoch": 305, "lr": 3.2776200678002176e-05} +{"train_loss": 0.014844954013824463, "grad_norm": 0.07537627965211868, "global_step": 32433, "epoch": 305, "lr": 3.277341833856287e-05} +{"train_loss": 0.012964582070708275, "grad_norm": 0.07591210305690765, "global_step": 32434, "epoch": 305, "lr": 3.277063605965027e-05} +{"train_loss": 0.015267202832718502, "grad_norm": 0.13037605583667755, "global_step": 32435, "epoch": 305, "lr": 3.276785384127415e-05, "val_loss": 0.019874516874551773} +{"train_loss": 0.01343056745827198, "grad_norm": 0.09136203676462173, "global_step": 32436, "epoch": 306, "lr": 3.2765071683444305e-05} +{"train_loss": 0.015054143965244293, "grad_norm": 0.06747337430715561, "global_step": 32437, "epoch": 306, "lr": 3.276228958617051e-05} +{"train_loss": 0.01449938677251339, "grad_norm": 0.16399991512298584, "global_step": 32438, "epoch": 306, "lr": 3.2759507549462524e-05} +{"train_loss": 0.015222368761897087, "grad_norm": 0.08159829676151276, "global_step": 32439, "epoch": 306, "lr": 3.275672557333015e-05} +{"train_loss": 0.014214163646101952, "grad_norm": 0.07233909517526627, "global_step": 32440, "epoch": 306, "lr": 3.2753943657783116e-05} +{"train_loss": 0.01284339651465416, "grad_norm": 0.059284452348947525, "global_step": 32441, "epoch": 306, "lr": 3.2751161802831243e-05} +{"train_loss": 0.01586206629872322, "grad_norm": 0.0943421795964241, "global_step": 32442, "epoch": 306, "lr": 3.274838000848428e-05} +{"train_loss": 0.015444306656718254, "grad_norm": 0.14565588533878326, "global_step": 32443, "epoch": 306, "lr": 3.2745598274752e-05} +{"train_loss": 0.015366148203611374, "grad_norm": 0.14683417975902557, "global_step": 32444, "epoch": 306, "lr": 3.274281660164419e-05} +{"train_loss": 0.015815112739801407, "grad_norm": 0.07459893822669983, "global_step": 32445, "epoch": 306, "lr": 3.274003498917063e-05} +{"train_loss": 0.015238882042467594, "grad_norm": 0.06413464993238449, "global_step": 32446, "epoch": 306, "lr": 3.273725343734106e-05} +{"train_loss": 0.017802350223064423, "grad_norm": 0.11206525564193726, "global_step": 32447, "epoch": 306, "lr": 3.273447194616529e-05} +{"train_loss": 0.014635591767728329, "grad_norm": 0.09717073291540146, "global_step": 32448, "epoch": 306, "lr": 3.2731690515653055e-05} +{"train_loss": 0.013960026204586029, "grad_norm": 0.06560521572828293, "global_step": 32449, "epoch": 306, "lr": 3.272890914581417e-05} +{"train_loss": 0.017971109598875046, "grad_norm": 0.07165653258562088, "global_step": 32450, "epoch": 306, "lr": 3.2726127836658385e-05} +{"train_loss": 0.015364476479589939, "grad_norm": 0.057909879833459854, "global_step": 32451, "epoch": 306, "lr": 3.272334658819548e-05} +{"train_loss": 0.017589746043086052, "grad_norm": 0.12507592141628265, "global_step": 32452, "epoch": 306, "lr": 3.272056540043522e-05} +{"train_loss": 0.013314667157828808, "grad_norm": 0.07503621280193329, "global_step": 32453, "epoch": 306, "lr": 3.271778427338736e-05} +{"train_loss": 0.015222105197608471, "grad_norm": 0.098517507314682, "global_step": 32454, "epoch": 306, "lr": 3.27150032070617e-05} +{"train_loss": 0.012733180075883865, "grad_norm": 0.08367323130369186, "global_step": 32455, "epoch": 306, "lr": 3.271222220146801e-05} +{"train_loss": 0.013509048148989677, "grad_norm": 0.129558265209198, "global_step": 32456, "epoch": 306, "lr": 3.270944125661604e-05} +{"train_loss": 0.014137862250208855, "grad_norm": 0.08412393182516098, "global_step": 32457, "epoch": 306, "lr": 3.2706660372515583e-05} +{"train_loss": 0.015983901917934418, "grad_norm": 0.0593208372592926, "global_step": 32458, "epoch": 306, "lr": 3.27038795491764e-05} +{"train_loss": 0.017200056463479996, "grad_norm": 0.07339105755090714, "global_step": 32459, "epoch": 306, "lr": 3.2701098786608265e-05} +{"train_loss": 0.0165396835654974, "grad_norm": 0.062179047614336014, "global_step": 32460, "epoch": 306, "lr": 3.269831808482093e-05} +{"train_loss": 0.01452985405921936, "grad_norm": 0.0671272724866867, "global_step": 32461, "epoch": 306, "lr": 3.26955374438242e-05} +{"train_loss": 0.015117999166250229, "grad_norm": 0.08150164783000946, "global_step": 32462, "epoch": 306, "lr": 3.269275686362781e-05} +{"train_loss": 0.015277540311217308, "grad_norm": 0.058382343500852585, "global_step": 32463, "epoch": 306, "lr": 3.268997634424157e-05} +{"train_loss": 0.015240302309393883, "grad_norm": 0.1981477588415146, "global_step": 32464, "epoch": 306, "lr": 3.2687195885675214e-05} +{"train_loss": 0.015702273696660995, "grad_norm": 0.11212250590324402, "global_step": 32465, "epoch": 306, "lr": 3.26844154879385e-05} +{"train_loss": 0.01598775014281273, "grad_norm": 0.09569628536701202, "global_step": 32466, "epoch": 306, "lr": 3.2681635151041246e-05} +{"train_loss": 0.013189870864152908, "grad_norm": 0.06542482227087021, "global_step": 32467, "epoch": 306, "lr": 3.2678854874993184e-05} +{"train_loss": 0.013959142379462719, "grad_norm": 0.06686674058437347, "global_step": 32468, "epoch": 306, "lr": 3.267607465980411e-05} +{"train_loss": 0.015866804867982864, "grad_norm": 0.10885749012231827, "global_step": 32469, "epoch": 306, "lr": 3.267329450548375e-05} +{"train_loss": 0.016396962106227875, "grad_norm": 0.09770749509334564, "global_step": 32470, "epoch": 306, "lr": 3.267051441204191e-05} +{"train_loss": 0.014722645282745361, "grad_norm": 0.07265034317970276, "global_step": 32471, "epoch": 306, "lr": 3.2667734379488343e-05} +{"train_loss": 0.015901312232017517, "grad_norm": 0.11381163448095322, "global_step": 32472, "epoch": 306, "lr": 3.266495440783281e-05} +{"train_loss": 0.016463588923215866, "grad_norm": 0.16257280111312866, "global_step": 32473, "epoch": 306, "lr": 3.266217449708511e-05} +{"train_loss": 0.014907985925674438, "grad_norm": 0.10694140195846558, "global_step": 32474, "epoch": 306, "lr": 3.265939464725498e-05} +{"train_loss": 0.019908586516976357, "grad_norm": 0.0752054899930954, "global_step": 32475, "epoch": 306, "lr": 3.26566148583522e-05} +{"train_loss": 0.015653207898139954, "grad_norm": 0.09611206501722336, "global_step": 32476, "epoch": 306, "lr": 3.265383513038653e-05} +{"train_loss": 0.014880040660500526, "grad_norm": 0.0666634812951088, "global_step": 32477, "epoch": 306, "lr": 3.265105546336773e-05} +{"train_loss": 0.013278698548674583, "grad_norm": 0.06412402540445328, "global_step": 32478, "epoch": 306, "lr": 3.264827585730558e-05} +{"train_loss": 0.014935581013560295, "grad_norm": 0.08555953949689865, "global_step": 32479, "epoch": 306, "lr": 3.264549631220985e-05} +{"train_loss": 0.015050562098622322, "grad_norm": 0.07904000580310822, "global_step": 32480, "epoch": 306, "lr": 3.2642716828090304e-05} +{"train_loss": 0.012563930824398994, "grad_norm": 0.05074755474925041, "global_step": 32481, "epoch": 306, "lr": 3.26399374049567e-05} +{"train_loss": 0.014973660930991173, "grad_norm": 0.16380718350410461, "global_step": 32482, "epoch": 306, "lr": 3.263715804281878e-05} +{"train_loss": 0.018105648458003998, "grad_norm": 0.10701541602611542, "global_step": 32483, "epoch": 306, "lr": 3.263437874168637e-05} +{"train_loss": 0.014530179090797901, "grad_norm": 0.06826914846897125, "global_step": 32484, "epoch": 306, "lr": 3.2631599501569176e-05} +{"train_loss": 0.017087485641241074, "grad_norm": 0.07819345593452454, "global_step": 32485, "epoch": 306, "lr": 3.262882032247699e-05} +{"train_loss": 0.01746492087841034, "grad_norm": 0.0626991018652916, "global_step": 32486, "epoch": 306, "lr": 3.26260412044196e-05} +{"train_loss": 0.016525179147720337, "grad_norm": 0.05973204970359802, "global_step": 32487, "epoch": 306, "lr": 3.262326214740673e-05} +{"train_loss": 0.014990628696978092, "grad_norm": 0.07907284796237946, "global_step": 32488, "epoch": 306, "lr": 3.262048315144815e-05} +{"train_loss": 0.012937165796756744, "grad_norm": 0.07049325853586197, "global_step": 32489, "epoch": 306, "lr": 3.261770421655364e-05} +{"train_loss": 0.016218086704611778, "grad_norm": 0.08466002345085144, "global_step": 32490, "epoch": 306, "lr": 3.261492534273296e-05} +{"train_loss": 0.013651751913130283, "grad_norm": 0.06399503350257874, "global_step": 32491, "epoch": 306, "lr": 3.261214652999588e-05} +{"train_loss": 0.014413781464099884, "grad_norm": 0.09621324390172958, "global_step": 32492, "epoch": 306, "lr": 3.260936777835216e-05} +{"train_loss": 0.014791772700846195, "grad_norm": 0.10598037391901016, "global_step": 32493, "epoch": 306, "lr": 3.2606589087811544e-05} +{"train_loss": 0.016680750995874405, "grad_norm": 0.07876762747764587, "global_step": 32494, "epoch": 306, "lr": 3.26038104583838e-05} +{"train_loss": 0.016287125647068024, "grad_norm": 0.10852960497140884, "global_step": 32495, "epoch": 306, "lr": 3.2601031890078714e-05} +{"train_loss": 0.013309521600604057, "grad_norm": 0.1246168464422226, "global_step": 32496, "epoch": 306, "lr": 3.259825338290604e-05} +{"train_loss": 0.01503700390458107, "grad_norm": 0.08196990191936493, "global_step": 32497, "epoch": 306, "lr": 3.2595474936875515e-05} +{"train_loss": 0.014839482493698597, "grad_norm": 0.0778796598315239, "global_step": 32498, "epoch": 306, "lr": 3.259269655199695e-05} +{"train_loss": 0.016027208417654037, "grad_norm": 0.0684964582324028, "global_step": 32499, "epoch": 306, "lr": 3.258991822828007e-05} +{"train_loss": 0.013629348948597908, "grad_norm": 0.06285896897315979, "global_step": 32500, "epoch": 306, "lr": 3.2587139965734645e-05} +{"train_loss": 0.015469597652554512, "grad_norm": 0.052789825946092606, "global_step": 32501, "epoch": 306, "lr": 3.258436176437042e-05} +{"train_loss": 0.01323799416422844, "grad_norm": 0.08920501917600632, "global_step": 32502, "epoch": 306, "lr": 3.2581583624197187e-05} +{"train_loss": 0.013832848519086838, "grad_norm": 0.06710635870695114, "global_step": 32503, "epoch": 306, "lr": 3.25788055452247e-05} +{"train_loss": 0.014221522957086563, "grad_norm": 0.05996577814221382, "global_step": 32504, "epoch": 306, "lr": 3.257602752746272e-05} +{"train_loss": 0.015825681388378143, "grad_norm": 0.061546921730041504, "global_step": 32505, "epoch": 306, "lr": 3.257324957092099e-05} +{"train_loss": 0.016128990799188614, "grad_norm": 0.045569758862257004, "global_step": 32506, "epoch": 306, "lr": 3.2570471675609284e-05} +{"train_loss": 0.014068545773625374, "grad_norm": 0.07379166781902313, "global_step": 32507, "epoch": 306, "lr": 3.2567693841537354e-05} +{"train_loss": 0.015789644792675972, "grad_norm": 0.13035845756530762, "global_step": 32508, "epoch": 306, "lr": 3.256491606871498e-05} +{"train_loss": 0.0162842757999897, "grad_norm": 0.06473157554864883, "global_step": 32509, "epoch": 306, "lr": 3.2562138357151884e-05} +{"train_loss": 0.016929704695940018, "grad_norm": 0.08019831031560898, "global_step": 32510, "epoch": 306, "lr": 3.2559360706857894e-05} +{"train_loss": 0.01765495538711548, "grad_norm": 0.09498666971921921, "global_step": 32511, "epoch": 306, "lr": 3.2556583117842686e-05} +{"train_loss": 0.015289047732949257, "grad_norm": 0.08940593898296356, "global_step": 32512, "epoch": 306, "lr": 3.255380559011608e-05} +{"train_loss": 0.013845494017004967, "grad_norm": 0.07796230912208557, "global_step": 32513, "epoch": 306, "lr": 3.2551028123687793e-05} +{"train_loss": 0.01467830315232277, "grad_norm": 0.08679250627756119, "global_step": 32514, "epoch": 306, "lr": 3.254825071856762e-05} +{"train_loss": 0.01875876821577549, "grad_norm": 0.1107528880238533, "global_step": 32515, "epoch": 306, "lr": 3.25454733747653e-05} +{"train_loss": 0.014268023893237114, "grad_norm": 0.08644185960292816, "global_step": 32516, "epoch": 306, "lr": 3.2542696092290614e-05} +{"train_loss": 0.017583180218935013, "grad_norm": 0.0905340388417244, "global_step": 32517, "epoch": 306, "lr": 3.253991887115329e-05} +{"train_loss": 0.013789528049528599, "grad_norm": 0.06966228038072586, "global_step": 32518, "epoch": 306, "lr": 3.2537141711363087e-05} +{"train_loss": 0.01678340882062912, "grad_norm": 0.13952964544296265, "global_step": 32519, "epoch": 306, "lr": 3.253436461292978e-05} +{"train_loss": 0.015602762810885906, "grad_norm": 0.07053884863853455, "global_step": 32520, "epoch": 306, "lr": 3.253158757586312e-05} +{"train_loss": 0.012950815260410309, "grad_norm": 0.061224315315485, "global_step": 32521, "epoch": 306, "lr": 3.252881060017288e-05} +{"train_loss": 0.01350292470306158, "grad_norm": 0.11063362658023834, "global_step": 32522, "epoch": 306, "lr": 3.2526033685868785e-05} +{"train_loss": 0.017109517008066177, "grad_norm": 0.09716300666332245, "global_step": 32523, "epoch": 306, "lr": 3.25232568329606e-05} +{"train_loss": 0.013872349634766579, "grad_norm": 0.08705822378396988, "global_step": 32524, "epoch": 306, "lr": 3.25204800414581e-05} +{"train_loss": 0.014900406822562218, "grad_norm": 0.08201837539672852, "global_step": 32525, "epoch": 306, "lr": 3.2517703311371036e-05} +{"train_loss": 0.017211783677339554, "grad_norm": 0.12092874944210052, "global_step": 32526, "epoch": 306, "lr": 3.2514926642709146e-05} +{"train_loss": 0.013239407911896706, "grad_norm": 0.08320542424917221, "global_step": 32527, "epoch": 306, "lr": 3.2512150035482224e-05} +{"train_loss": 0.01367692556232214, "grad_norm": 0.069513238966465, "global_step": 32528, "epoch": 306, "lr": 3.250937348969999e-05} +{"train_loss": 0.016601789742708206, "grad_norm": 0.0653335452079773, "global_step": 32529, "epoch": 306, "lr": 3.2506597005372205e-05} +{"train_loss": 0.01583203673362732, "grad_norm": 0.07721181213855743, "global_step": 32530, "epoch": 306, "lr": 3.2503820582508624e-05} +{"train_loss": 0.015150947496294975, "grad_norm": 0.07163980603218079, "global_step": 32531, "epoch": 306, "lr": 3.250104422111902e-05} +{"train_loss": 0.014975117519497871, "grad_norm": 0.08754263818264008, "global_step": 32532, "epoch": 306, "lr": 3.2498267921213135e-05} +{"train_loss": 0.015170923434197903, "grad_norm": 0.05867438018321991, "global_step": 32533, "epoch": 306, "lr": 3.2495491682800736e-05} +{"train_loss": 0.01762140542268753, "grad_norm": 0.07876574248075485, "global_step": 32534, "epoch": 306, "lr": 3.2492715505891555e-05} +{"train_loss": 0.014059152454137802, "grad_norm": 0.11541186273097992, "global_step": 32535, "epoch": 306, "lr": 3.248993939049535e-05} +{"train_loss": 0.014306866563856602, "grad_norm": 0.06725303828716278, "global_step": 32536, "epoch": 306, "lr": 3.248716333662189e-05} +{"train_loss": 0.01717546209692955, "grad_norm": 0.09261883050203323, "global_step": 32537, "epoch": 306, "lr": 3.2484387344280934e-05} +{"train_loss": 0.014150595292448997, "grad_norm": 0.15690065920352936, "global_step": 32538, "epoch": 306, "lr": 3.2481611413482204e-05} +{"train_loss": 0.015939658507704735, "grad_norm": 0.09124816209077835, "global_step": 32539, "epoch": 306, "lr": 3.247883554423549e-05} +{"train_loss": 0.015696045011281967, "grad_norm": 0.0925079733133316, "global_step": 32540, "epoch": 306, "lr": 3.2476059736550523e-05} +{"train_loss": 0.015253388050521881, "grad_norm": 0.13733862340450287, "global_step": 32541, "epoch": 306, "lr": 3.247328399043706e-05, "val_loss": 0.019328180700540543} +{"train_loss": 0.016213545575737953, "grad_norm": 0.08436677604913712, "global_step": 32542, "epoch": 307, "lr": 3.2470508305904847e-05} +{"train_loss": 0.013976888731122017, "grad_norm": 0.05645278841257095, "global_step": 32543, "epoch": 307, "lr": 3.246773268296365e-05} +{"train_loss": 0.017840346321463585, "grad_norm": 0.1265975385904312, "global_step": 32544, "epoch": 307, "lr": 3.2464957121623215e-05} +{"train_loss": 0.01351194642484188, "grad_norm": 0.06565920263528824, "global_step": 32545, "epoch": 307, "lr": 3.246218162189331e-05} +{"train_loss": 0.014704082161188126, "grad_norm": 0.07222362607717514, "global_step": 32546, "epoch": 307, "lr": 3.245940618378366e-05} +{"train_loss": 0.014973258599638939, "grad_norm": 0.0709298625588417, "global_step": 32547, "epoch": 307, "lr": 3.245663080730401e-05} +{"train_loss": 0.01643882319331169, "grad_norm": 0.059019941836595535, "global_step": 32548, "epoch": 307, "lr": 3.2453855492464144e-05} +{"train_loss": 0.013704933226108551, "grad_norm": 0.07124819606542587, "global_step": 32549, "epoch": 307, "lr": 3.2451080239273814e-05} +{"train_loss": 0.014242582023143768, "grad_norm": 0.07130333036184311, "global_step": 32550, "epoch": 307, "lr": 3.2448305047742734e-05} +{"train_loss": 0.017671357840299606, "grad_norm": 0.09288158267736435, "global_step": 32551, "epoch": 307, "lr": 3.24455299178807e-05} +{"train_loss": 0.014334003441035748, "grad_norm": 0.06350474059581757, "global_step": 32552, "epoch": 307, "lr": 3.244275484969742e-05} +{"train_loss": 0.01757223717868328, "grad_norm": 0.061106760054826736, "global_step": 32553, "epoch": 307, "lr": 3.2439979843202675e-05} +{"train_loss": 0.0155634805560112, "grad_norm": 0.06544573605060577, "global_step": 32554, "epoch": 307, "lr": 3.2437204898406206e-05} +{"train_loss": 0.01591116562485695, "grad_norm": 0.07351666688919067, "global_step": 32555, "epoch": 307, "lr": 3.243443001531774e-05} +{"train_loss": 0.01689700409770012, "grad_norm": 0.10551899671554565, "global_step": 32556, "epoch": 307, "lr": 3.2431655193947065e-05} +{"train_loss": 0.013033539056777954, "grad_norm": 0.11425241082906723, "global_step": 32557, "epoch": 307, "lr": 3.242888043430392e-05} +{"train_loss": 0.016395926475524902, "grad_norm": 0.11981305480003357, "global_step": 32558, "epoch": 307, "lr": 3.2426105736398035e-05} +{"train_loss": 0.014551149681210518, "grad_norm": 0.16068142652511597, "global_step": 32559, "epoch": 307, "lr": 3.2423331100239177e-05} +{"train_loss": 0.01599818468093872, "grad_norm": 0.055421434342861176, "global_step": 32560, "epoch": 307, "lr": 3.242055652583709e-05} +{"train_loss": 0.01729794219136238, "grad_norm": 0.07945308834314346, "global_step": 32561, "epoch": 307, "lr": 3.2417782013201514e-05} +{"train_loss": 0.013142779469490051, "grad_norm": 0.09782885760068893, "global_step": 32562, "epoch": 307, "lr": 3.2415007562342205e-05} +{"train_loss": 0.01597699522972107, "grad_norm": 0.15226615965366364, "global_step": 32563, "epoch": 307, "lr": 3.241223317326894e-05} +{"train_loss": 0.014716475270688534, "grad_norm": 0.08990422636270523, "global_step": 32564, "epoch": 307, "lr": 3.240945884599141e-05} +{"train_loss": 0.013581085950136185, "grad_norm": 0.05581739544868469, "global_step": 32565, "epoch": 307, "lr": 3.2406684580519396e-05} +{"train_loss": 0.013992764055728912, "grad_norm": 0.07226444035768509, "global_step": 32566, "epoch": 307, "lr": 3.240391037686265e-05} +{"train_loss": 0.013922718353569508, "grad_norm": 0.11048101633787155, "global_step": 32567, "epoch": 307, "lr": 3.2401136235030896e-05} +{"train_loss": 0.015408945269882679, "grad_norm": 0.1114685982465744, "global_step": 32568, "epoch": 307, "lr": 3.239836215503392e-05} +{"train_loss": 0.014177048578858376, "grad_norm": 0.14106403291225433, "global_step": 32569, "epoch": 307, "lr": 3.2395588136881425e-05} +{"train_loss": 0.015984663739800453, "grad_norm": 0.1793956309556961, "global_step": 32570, "epoch": 307, "lr": 3.239281418058319e-05} +{"train_loss": 0.017320770770311356, "grad_norm": 0.073935866355896, "global_step": 32571, "epoch": 307, "lr": 3.239004028614893e-05} +{"train_loss": 0.015323396772146225, "grad_norm": 0.07673170417547226, "global_step": 32572, "epoch": 307, "lr": 3.238726645358842e-05} +{"train_loss": 0.01753389835357666, "grad_norm": 0.10818927735090256, "global_step": 32573, "epoch": 307, "lr": 3.23844926829114e-05} +{"train_loss": 0.013437662273645401, "grad_norm": 0.11639394611120224, "global_step": 32574, "epoch": 307, "lr": 3.2381718974127616e-05} +{"train_loss": 0.013688893057405949, "grad_norm": 0.10464686900377274, "global_step": 32575, "epoch": 307, "lr": 3.237894532724681e-05} +{"train_loss": 0.014186532236635685, "grad_norm": 0.06326652318239212, "global_step": 32576, "epoch": 307, "lr": 3.2376171742278706e-05} +{"train_loss": 0.014761130325496197, "grad_norm": 0.10053309798240662, "global_step": 32577, "epoch": 307, "lr": 3.2373398219233084e-05} +{"train_loss": 0.014710500836372375, "grad_norm": 0.0771745815873146, "global_step": 32578, "epoch": 307, "lr": 3.237062475811968e-05} +{"train_loss": 0.014402243308722973, "grad_norm": 0.06816887855529785, "global_step": 32579, "epoch": 307, "lr": 3.236785135894822e-05} +{"train_loss": 0.015584085136651993, "grad_norm": 0.06528610736131668, "global_step": 32580, "epoch": 307, "lr": 3.236507802172849e-05} +{"train_loss": 0.01600981131196022, "grad_norm": 0.09680836647748947, "global_step": 32581, "epoch": 307, "lr": 3.236230474647018e-05} +{"train_loss": 0.01822165586054325, "grad_norm": 0.06917249411344528, "global_step": 32582, "epoch": 307, "lr": 3.235953153318307e-05} +{"train_loss": 0.01607784442603588, "grad_norm": 0.08005724847316742, "global_step": 32583, "epoch": 307, "lr": 3.2356758381876896e-05} +{"train_loss": 0.017431240528821945, "grad_norm": 0.16560335457324982, "global_step": 32584, "epoch": 307, "lr": 3.2353985292561387e-05} +{"train_loss": 0.01685677468776703, "grad_norm": 0.11752326041460037, "global_step": 32585, "epoch": 307, "lr": 3.23512122652463e-05} +{"train_loss": 0.015104977414011955, "grad_norm": 0.08461738377809525, "global_step": 32586, "epoch": 307, "lr": 3.2348439299941404e-05} +{"train_loss": 0.015879960730671883, "grad_norm": 0.05947275087237358, "global_step": 32587, "epoch": 307, "lr": 3.2345666396656406e-05} +{"train_loss": 0.013521014712750912, "grad_norm": 0.055156409740448, "global_step": 32588, "epoch": 307, "lr": 3.234289355540104e-05} +{"train_loss": 0.019261348992586136, "grad_norm": 0.07782530784606934, "global_step": 32589, "epoch": 307, "lr": 3.234012077618508e-05} +{"train_loss": 0.016559910029172897, "grad_norm": 0.0941450446844101, "global_step": 32590, "epoch": 307, "lr": 3.233734805901826e-05} +{"train_loss": 0.014061497524380684, "grad_norm": 0.07034283131361008, "global_step": 32591, "epoch": 307, "lr": 3.2334575403910296e-05} +{"train_loss": 0.016686774790287018, "grad_norm": 0.07907617092132568, "global_step": 32592, "epoch": 307, "lr": 3.233180281087099e-05} +{"train_loss": 0.013676386326551437, "grad_norm": 0.06276165693998337, "global_step": 32593, "epoch": 307, "lr": 3.232903027991001e-05} +{"train_loss": 0.013932363130152225, "grad_norm": 0.10186529159545898, "global_step": 32594, "epoch": 307, "lr": 3.2326257811037155e-05} +{"train_loss": 0.01501674298197031, "grad_norm": 0.068772092461586, "global_step": 32595, "epoch": 307, "lr": 3.232348540426213e-05} +{"train_loss": 0.015016511082649231, "grad_norm": 0.157806396484375, "global_step": 32596, "epoch": 307, "lr": 3.232071305959469e-05} +{"train_loss": 0.013507147319614887, "grad_norm": 0.08461380004882812, "global_step": 32597, "epoch": 307, "lr": 3.2317940777044574e-05} +{"train_loss": 0.014827236533164978, "grad_norm": 0.06533496826887131, "global_step": 32598, "epoch": 307, "lr": 3.2315168556621546e-05} +{"train_loss": 0.013514984399080276, "grad_norm": 0.07624675333499908, "global_step": 32599, "epoch": 307, "lr": 3.231239639833531e-05} +{"train_loss": 0.014304691925644875, "grad_norm": 0.08252748101949692, "global_step": 32600, "epoch": 307, "lr": 3.230962430219561e-05} +{"train_loss": 0.013218407519161701, "grad_norm": 0.07120852172374725, "global_step": 32601, "epoch": 307, "lr": 3.2306852268212206e-05} +{"train_loss": 0.016149641945958138, "grad_norm": 0.06671919673681259, "global_step": 32602, "epoch": 307, "lr": 3.230408029639483e-05} +{"train_loss": 0.014589646831154823, "grad_norm": 0.07467927038669586, "global_step": 32603, "epoch": 307, "lr": 3.2301308386753215e-05} +{"train_loss": 0.014641187153756618, "grad_norm": 0.1165449470281601, "global_step": 32604, "epoch": 307, "lr": 3.229853653929713e-05} +{"train_loss": 0.016292713582515717, "grad_norm": 0.07921648025512695, "global_step": 32605, "epoch": 307, "lr": 3.229576475403626e-05} +{"train_loss": 0.014564447104930878, "grad_norm": 0.12390996515750885, "global_step": 32606, "epoch": 307, "lr": 3.229299303098039e-05} +{"train_loss": 0.014564024284482002, "grad_norm": 0.07422786951065063, "global_step": 32607, "epoch": 307, "lr": 3.229022137013923e-05} +{"train_loss": 0.015029734000563622, "grad_norm": 0.10951457172632217, "global_step": 32608, "epoch": 307, "lr": 3.228744977152253e-05} +{"train_loss": 0.014866153709590435, "grad_norm": 0.0737207904458046, "global_step": 32609, "epoch": 307, "lr": 3.228467823514003e-05} +{"train_loss": 0.012692084535956383, "grad_norm": 0.07506996393203735, "global_step": 32610, "epoch": 307, "lr": 3.228190676100149e-05} +{"train_loss": 0.01491850521415472, "grad_norm": 0.12307199090719223, "global_step": 32611, "epoch": 307, "lr": 3.22791353491166e-05} +{"train_loss": 0.01531642023473978, "grad_norm": 0.10844908654689789, "global_step": 32612, "epoch": 307, "lr": 3.227636399949513e-05} +{"train_loss": 0.013338007032871246, "grad_norm": 0.06613650172948837, "global_step": 32613, "epoch": 307, "lr": 3.227359271214679e-05} +{"train_loss": 0.016297999769449234, "grad_norm": 0.11990547180175781, "global_step": 32614, "epoch": 307, "lr": 3.227082148708136e-05} +{"train_loss": 0.015636064112186432, "grad_norm": 0.05726200342178345, "global_step": 32615, "epoch": 307, "lr": 3.226805032430853e-05} +{"train_loss": 0.016459565609693527, "grad_norm": 0.12644122540950775, "global_step": 32616, "epoch": 307, "lr": 3.22652792238381e-05} +{"train_loss": 0.014833533205091953, "grad_norm": 0.0900920182466507, "global_step": 32617, "epoch": 307, "lr": 3.226250818567973e-05} +{"train_loss": 0.01602006144821644, "grad_norm": 0.09219470620155334, "global_step": 32618, "epoch": 307, "lr": 3.2259737209843204e-05} +{"train_loss": 0.01517418771982193, "grad_norm": 0.12402672320604324, "global_step": 32619, "epoch": 307, "lr": 3.225696629633825e-05} +{"train_loss": 0.012706002220511436, "grad_norm": 0.09666857868432999, "global_step": 32620, "epoch": 307, "lr": 3.2254195445174584e-05} +{"train_loss": 0.014639616012573242, "grad_norm": 0.07217501103878021, "global_step": 32621, "epoch": 307, "lr": 3.2251424656361995e-05} +{"train_loss": 0.016511976718902588, "grad_norm": 0.06730819493532181, "global_step": 32622, "epoch": 307, "lr": 3.224865392991015e-05} +{"train_loss": 0.01389859989285469, "grad_norm": 0.09261556714773178, "global_step": 32623, "epoch": 307, "lr": 3.224588326582881e-05} +{"train_loss": 0.016010526567697525, "grad_norm": 0.08793259412050247, "global_step": 32624, "epoch": 307, "lr": 3.2243112664127725e-05} +{"train_loss": 0.01602143980562687, "grad_norm": 0.06692442297935486, "global_step": 32625, "epoch": 307, "lr": 3.224034212481661e-05} +{"train_loss": 0.013836272060871124, "grad_norm": 0.11603953689336777, "global_step": 32626, "epoch": 307, "lr": 3.223757164790523e-05} +{"train_loss": 0.015900585800409317, "grad_norm": 0.09399596601724625, "global_step": 32627, "epoch": 307, "lr": 3.22348012334033e-05} +{"train_loss": 0.01795589178800583, "grad_norm": 0.10888449847698212, "global_step": 32628, "epoch": 307, "lr": 3.2232030881320544e-05} +{"train_loss": 0.014775971882045269, "grad_norm": 0.06643270701169968, "global_step": 32629, "epoch": 307, "lr": 3.222926059166669e-05} +{"train_loss": 0.015500686131417751, "grad_norm": 0.13277886807918549, "global_step": 32630, "epoch": 307, "lr": 3.22264903644515e-05} +{"train_loss": 0.015019126236438751, "grad_norm": 0.08616160601377487, "global_step": 32631, "epoch": 307, "lr": 3.222372019968469e-05} +{"train_loss": 0.016773249953985214, "grad_norm": 0.09238661080598831, "global_step": 32632, "epoch": 307, "lr": 3.222095009737599e-05} +{"train_loss": 0.017846226692199707, "grad_norm": 0.10046584904193878, "global_step": 32633, "epoch": 307, "lr": 3.221818005753516e-05} +{"train_loss": 0.013892069458961487, "grad_norm": 0.08170871436595917, "global_step": 32634, "epoch": 307, "lr": 3.2215410080171896e-05} +{"train_loss": 0.014530545100569725, "grad_norm": 0.1423361748456955, "global_step": 32635, "epoch": 307, "lr": 3.2212640165295945e-05} +{"train_loss": 0.015677087008953094, "grad_norm": 0.08008953928947449, "global_step": 32636, "epoch": 307, "lr": 3.220987031291705e-05} +{"train_loss": 0.014549199491739273, "grad_norm": 0.18063244223594666, "global_step": 32637, "epoch": 307, "lr": 3.220710052304492e-05} +{"train_loss": 0.015713926404714584, "grad_norm": 0.05592663586139679, "global_step": 32638, "epoch": 307, "lr": 3.2204330795689315e-05} +{"train_loss": 0.014600396156311035, "grad_norm": 0.08566679805517197, "global_step": 32639, "epoch": 307, "lr": 3.2201561130859957e-05} +{"train_loss": 0.015085444785654545, "grad_norm": 0.10750720649957657, "global_step": 32640, "epoch": 307, "lr": 3.2198791528566566e-05} +{"train_loss": 0.01637166179716587, "grad_norm": 0.09135565161705017, "global_step": 32641, "epoch": 307, "lr": 3.219602198881888e-05} +{"train_loss": 0.016431905329227448, "grad_norm": 0.11151878535747528, "global_step": 32642, "epoch": 307, "lr": 3.2193252511626614e-05} +{"train_loss": 0.01490606740117073, "grad_norm": 0.0824873223900795, "global_step": 32643, "epoch": 307, "lr": 3.219048309699953e-05} +{"train_loss": 0.014783583581447601, "grad_norm": 0.06872900575399399, "global_step": 32644, "epoch": 307, "lr": 3.2187713744947336e-05} +{"train_loss": 0.014877742156386375, "grad_norm": 0.06096155196428299, "global_step": 32645, "epoch": 307, "lr": 3.21849444554798e-05} +{"train_loss": 0.016349444165825844, "grad_norm": 0.08394777029752731, "global_step": 32646, "epoch": 307, "lr": 3.218217522860658e-05} +{"train_loss": 0.015256963810830746, "grad_norm": 0.10331056267023087, "global_step": 32647, "epoch": 307, "lr": 3.217940606433747e-05, "val_loss": 0.01920795440673828} +{"train_loss": 0.016953540965914726, "grad_norm": 0.05964737385511398, "global_step": 32648, "epoch": 308, "lr": 3.217663696268217e-05} +{"train_loss": 0.01709897816181183, "grad_norm": 0.08959456533193588, "global_step": 32649, "epoch": 308, "lr": 3.2173867923650415e-05} +{"train_loss": 0.014682379551231861, "grad_norm": 0.10718834400177002, "global_step": 32650, "epoch": 308, "lr": 3.2171098947251936e-05} +{"train_loss": 0.013567308895289898, "grad_norm": 0.10678490996360779, "global_step": 32651, "epoch": 308, "lr": 3.216833003349648e-05} +{"train_loss": 0.013697146438062191, "grad_norm": 0.06981935352087021, "global_step": 32652, "epoch": 308, "lr": 3.216556118239374e-05} +{"train_loss": 0.015612266026437283, "grad_norm": 0.10184766352176666, "global_step": 32653, "epoch": 308, "lr": 3.2162792393953466e-05} +{"train_loss": 0.01514207012951374, "grad_norm": 0.09528127312660217, "global_step": 32654, "epoch": 308, "lr": 3.216002366818537e-05} +{"train_loss": 0.01340017281472683, "grad_norm": 0.1453353315591812, "global_step": 32655, "epoch": 308, "lr": 3.215725500509921e-05} +{"train_loss": 0.017594661563634872, "grad_norm": 0.22231443226337433, "global_step": 32656, "epoch": 308, "lr": 3.2154486404704685e-05} +{"train_loss": 0.014890942722558975, "grad_norm": 0.08276847749948502, "global_step": 32657, "epoch": 308, "lr": 3.2151717867011546e-05} +{"train_loss": 0.016189411282539368, "grad_norm": 0.11555929481983185, "global_step": 32658, "epoch": 308, "lr": 3.2148949392029495e-05} +{"train_loss": 0.01708953082561493, "grad_norm": 0.061860229820013046, "global_step": 32659, "epoch": 308, "lr": 3.214618097976828e-05} +{"train_loss": 0.016569655388593674, "grad_norm": 0.1091516837477684, "global_step": 32660, "epoch": 308, "lr": 3.2143412630237616e-05} +{"train_loss": 0.015022503212094307, "grad_norm": 0.08984219282865524, "global_step": 32661, "epoch": 308, "lr": 3.214064434344723e-05} +{"train_loss": 0.014518726617097855, "grad_norm": 0.06235392019152641, "global_step": 32662, "epoch": 308, "lr": 3.213787611940686e-05} +{"train_loss": 0.015576569363474846, "grad_norm": 0.07768508046865463, "global_step": 32663, "epoch": 308, "lr": 3.213510795812624e-05} +{"train_loss": 0.01351260393857956, "grad_norm": 0.07237938046455383, "global_step": 32664, "epoch": 308, "lr": 3.213233985961507e-05} +{"train_loss": 0.013709723018109798, "grad_norm": 0.0726763978600502, "global_step": 32665, "epoch": 308, "lr": 3.2129571823883075e-05} +{"train_loss": 0.016088824719190598, "grad_norm": 0.12810903787612915, "global_step": 32666, "epoch": 308, "lr": 3.212680385094e-05} +{"train_loss": 0.013878025114536285, "grad_norm": 0.1481408327817917, "global_step": 32667, "epoch": 308, "lr": 3.212403594079556e-05} +{"train_loss": 0.01336514949798584, "grad_norm": 0.06179478392004967, "global_step": 32668, "epoch": 308, "lr": 3.2121268093459494e-05} +{"train_loss": 0.014563795179128647, "grad_norm": 0.0854497030377388, "global_step": 32669, "epoch": 308, "lr": 3.2118500308941526e-05} +{"train_loss": 0.01631566882133484, "grad_norm": 0.061199747025966644, "global_step": 32670, "epoch": 308, "lr": 3.2115732587251346e-05} +{"train_loss": 0.014091214165091515, "grad_norm": 0.09461960196495056, "global_step": 32671, "epoch": 308, "lr": 3.211296492839871e-05} +{"train_loss": 0.014072323217988014, "grad_norm": 0.11803384870290756, "global_step": 32672, "epoch": 308, "lr": 3.211019733239335e-05} +{"train_loss": 0.017563791945576668, "grad_norm": 0.09385885298252106, "global_step": 32673, "epoch": 308, "lr": 3.210742979924496e-05} +{"train_loss": 0.018258795142173767, "grad_norm": 0.09319807589054108, "global_step": 32674, "epoch": 308, "lr": 3.210466232896331e-05} +{"train_loss": 0.015012749470770359, "grad_norm": 0.08893075585365295, "global_step": 32675, "epoch": 308, "lr": 3.210189492155806e-05} +{"train_loss": 0.015292618423700333, "grad_norm": 0.06287860870361328, "global_step": 32676, "epoch": 308, "lr": 3.209912757703898e-05} +{"train_loss": 0.013661926612257957, "grad_norm": 0.1304529309272766, "global_step": 32677, "epoch": 308, "lr": 3.209636029541578e-05} +{"train_loss": 0.015010152012109756, "grad_norm": 0.06069906800985336, "global_step": 32678, "epoch": 308, "lr": 3.209359307669818e-05} +{"train_loss": 0.012056960724294186, "grad_norm": 0.06084524095058441, "global_step": 32679, "epoch": 308, "lr": 3.209082592089591e-05} +{"train_loss": 0.014611857011914253, "grad_norm": 0.07448511570692062, "global_step": 32680, "epoch": 308, "lr": 3.2088058828018705e-05} +{"train_loss": 0.018450487405061722, "grad_norm": 0.05841861292719841, "global_step": 32681, "epoch": 308, "lr": 3.208529179807626e-05} +{"train_loss": 0.01569327339529991, "grad_norm": 0.12943118810653687, "global_step": 32682, "epoch": 308, "lr": 3.2082524831078305e-05} +{"train_loss": 0.012071783654391766, "grad_norm": 0.06675808131694794, "global_step": 32683, "epoch": 308, "lr": 3.207975792703456e-05} +{"train_loss": 0.015542220324277878, "grad_norm": 0.05658099800348282, "global_step": 32684, "epoch": 308, "lr": 3.2076991085954755e-05} +{"train_loss": 0.015096401795744896, "grad_norm": 0.06504832208156586, "global_step": 32685, "epoch": 308, "lr": 3.207422430784861e-05} +{"train_loss": 0.017188020050525665, "grad_norm": 0.07776010036468506, "global_step": 32686, "epoch": 308, "lr": 3.207145759272586e-05} +{"train_loss": 0.014428351074457169, "grad_norm": 0.10281793773174286, "global_step": 32687, "epoch": 308, "lr": 3.206869094059619e-05} +{"train_loss": 0.013076861388981342, "grad_norm": 0.07429748773574829, "global_step": 32688, "epoch": 308, "lr": 3.2065924351469355e-05} +{"train_loss": 0.015161840245127678, "grad_norm": 0.10140176862478256, "global_step": 32689, "epoch": 308, "lr": 3.206315782535506e-05} +{"train_loss": 0.01839277893304825, "grad_norm": 0.08203579485416412, "global_step": 32690, "epoch": 308, "lr": 3.206039136226302e-05} +{"train_loss": 0.01463007740676403, "grad_norm": 0.10941288620233536, "global_step": 32691, "epoch": 308, "lr": 3.205762496220297e-05} +{"train_loss": 0.015873780474066734, "grad_norm": 0.1181550845503807, "global_step": 32692, "epoch": 308, "lr": 3.205485862518463e-05} +{"train_loss": 0.014736094512045383, "grad_norm": 0.15672411024570465, "global_step": 32693, "epoch": 308, "lr": 3.205209235121771e-05} +{"train_loss": 0.01367206871509552, "grad_norm": 0.07941509783267975, "global_step": 32694, "epoch": 308, "lr": 3.2049326140311925e-05} +{"train_loss": 0.017255788668990135, "grad_norm": 0.07463209331035614, "global_step": 32695, "epoch": 308, "lr": 3.2046559992476995e-05} +{"train_loss": 0.014827406965196133, "grad_norm": 0.09550176560878754, "global_step": 32696, "epoch": 308, "lr": 3.2043793907722655e-05} +{"train_loss": 0.01781386323273182, "grad_norm": 0.08090019226074219, "global_step": 32697, "epoch": 308, "lr": 3.2041027886058616e-05} +{"train_loss": 0.01322892028838396, "grad_norm": 0.08634702861309052, "global_step": 32698, "epoch": 308, "lr": 3.20382619274946e-05} +{"train_loss": 0.015841133892536163, "grad_norm": 0.15580663084983826, "global_step": 32699, "epoch": 308, "lr": 3.20354960320403e-05} +{"train_loss": 0.016402989625930786, "grad_norm": 0.07008025050163269, "global_step": 32700, "epoch": 308, "lr": 3.203273019970547e-05} +{"train_loss": 0.01600728929042816, "grad_norm": 0.09705021232366562, "global_step": 32701, "epoch": 308, "lr": 3.202996443049981e-05} +{"train_loss": 0.012495003640651703, "grad_norm": 0.11477762460708618, "global_step": 32702, "epoch": 308, "lr": 3.2027198724433024e-05} +{"train_loss": 0.0144457146525383, "grad_norm": 0.07279607653617859, "global_step": 32703, "epoch": 308, "lr": 3.202443308151487e-05} +{"train_loss": 0.01644599810242653, "grad_norm": 0.07450932264328003, "global_step": 32704, "epoch": 308, "lr": 3.202166750175504e-05} +{"train_loss": 0.010822666808962822, "grad_norm": 0.13712583482265472, "global_step": 32705, "epoch": 308, "lr": 3.201890198516324e-05} +{"train_loss": 0.01551584992557764, "grad_norm": 0.15898746252059937, "global_step": 32706, "epoch": 308, "lr": 3.201613653174921e-05} +{"train_loss": 0.014042001217603683, "grad_norm": 0.06307150423526764, "global_step": 32707, "epoch": 308, "lr": 3.201337114152263e-05} +{"train_loss": 0.015576522797346115, "grad_norm": 0.08899936079978943, "global_step": 32708, "epoch": 308, "lr": 3.2010605814493264e-05} +{"train_loss": 0.01586458459496498, "grad_norm": 0.1646491289138794, "global_step": 32709, "epoch": 308, "lr": 3.2007840550670795e-05} +{"train_loss": 0.0161262359470129, "grad_norm": 0.07110791653394699, "global_step": 32710, "epoch": 308, "lr": 3.200507535006497e-05} +{"train_loss": 0.013790380209684372, "grad_norm": 0.09938563406467438, "global_step": 32711, "epoch": 308, "lr": 3.200231021268547e-05} +{"train_loss": 0.016993392258882523, "grad_norm": 0.08873533457517624, "global_step": 32712, "epoch": 308, "lr": 3.199954513854202e-05} +{"train_loss": 0.014786912128329277, "grad_norm": 0.11831359565258026, "global_step": 32713, "epoch": 308, "lr": 3.1996780127644345e-05} +{"train_loss": 0.015656089410185814, "grad_norm": 0.09420118480920792, "global_step": 32714, "epoch": 308, "lr": 3.199401518000216e-05} +{"train_loss": 0.014181060716509819, "grad_norm": 0.10226567834615707, "global_step": 32715, "epoch": 308, "lr": 3.1991250295625154e-05} +{"train_loss": 0.01536291092634201, "grad_norm": 0.09022805094718933, "global_step": 32716, "epoch": 308, "lr": 3.198848547452309e-05} +{"train_loss": 0.014063796028494835, "grad_norm": 0.08730573207139969, "global_step": 32717, "epoch": 308, "lr": 3.198572071670565e-05} +{"train_loss": 0.015469258651137352, "grad_norm": 0.06038588657975197, "global_step": 32718, "epoch": 308, "lr": 3.198295602218254e-05} +{"train_loss": 0.013447361066937447, "grad_norm": 0.07428787648677826, "global_step": 32719, "epoch": 308, "lr": 3.1980191390963484e-05} +{"train_loss": 0.015508580021560192, "grad_norm": 0.07924509793519974, "global_step": 32720, "epoch": 308, "lr": 3.1977426823058216e-05} +{"train_loss": 0.014882052317261696, "grad_norm": 0.10514228790998459, "global_step": 32721, "epoch": 308, "lr": 3.197466231847643e-05} +{"train_loss": 0.015259785577654839, "grad_norm": 0.0936775878071785, "global_step": 32722, "epoch": 308, "lr": 3.197189787722784e-05} +{"train_loss": 0.014652920886874199, "grad_norm": 0.0871938019990921, "global_step": 32723, "epoch": 308, "lr": 3.1969133499322164e-05} +{"train_loss": 0.0163089819252491, "grad_norm": 0.08236553519964218, "global_step": 32724, "epoch": 308, "lr": 3.19663691847691e-05} +{"train_loss": 0.017766505479812622, "grad_norm": 0.11980973929166794, "global_step": 32725, "epoch": 308, "lr": 3.1963604933578374e-05} +{"train_loss": 0.014860428869724274, "grad_norm": 0.10805270075798035, "global_step": 32726, "epoch": 308, "lr": 3.19608407457597e-05} +{"train_loss": 0.014654774218797684, "grad_norm": 0.12441989034414291, "global_step": 32727, "epoch": 308, "lr": 3.19580766213228e-05} +{"train_loss": 0.012202451005578041, "grad_norm": 0.06967075914144516, "global_step": 32728, "epoch": 308, "lr": 3.1955312560277344e-05} +{"train_loss": 0.01435358077287674, "grad_norm": 0.12179794907569885, "global_step": 32729, "epoch": 308, "lr": 3.195254856263309e-05} +{"train_loss": 0.01494820136576891, "grad_norm": 0.12221761792898178, "global_step": 32730, "epoch": 308, "lr": 3.194978462839972e-05} +{"train_loss": 0.014879213646054268, "grad_norm": 0.06992090493440628, "global_step": 32731, "epoch": 308, "lr": 3.1947020757586965e-05} +{"train_loss": 0.01599746197462082, "grad_norm": 0.06785634160041809, "global_step": 32732, "epoch": 308, "lr": 3.194425695020453e-05} +{"train_loss": 0.014483978971838951, "grad_norm": 0.06528405845165253, "global_step": 32733, "epoch": 308, "lr": 3.194149320626213e-05} +{"train_loss": 0.015882251784205437, "grad_norm": 0.1051502451300621, "global_step": 32734, "epoch": 308, "lr": 3.193872952576946e-05} +{"train_loss": 0.016582190990447998, "grad_norm": 0.05519358441233635, "global_step": 32735, "epoch": 308, "lr": 3.1935965908736246e-05} +{"train_loss": 0.013881118036806583, "grad_norm": 0.08998721837997437, "global_step": 32736, "epoch": 308, "lr": 3.193320235517218e-05} +{"train_loss": 0.016954414546489716, "grad_norm": 0.06252137571573257, "global_step": 32737, "epoch": 308, "lr": 3.193043886508699e-05} +{"train_loss": 0.019143301993608475, "grad_norm": 0.08749761432409286, "global_step": 32738, "epoch": 308, "lr": 3.1927675438490384e-05} +{"train_loss": 0.014943523332476616, "grad_norm": 0.08529562503099442, "global_step": 32739, "epoch": 308, "lr": 3.1924912075392074e-05} +{"train_loss": 0.016362208873033524, "grad_norm": 0.16520985960960388, "global_step": 32740, "epoch": 308, "lr": 3.1922148775801754e-05} +{"train_loss": 0.013472329825162888, "grad_norm": 0.09529457241296768, "global_step": 32741, "epoch": 308, "lr": 3.191938553972913e-05} +{"train_loss": 0.017785347998142242, "grad_norm": 0.12400210648775101, "global_step": 32742, "epoch": 308, "lr": 3.1916622367183936e-05} +{"train_loss": 0.015107598155736923, "grad_norm": 0.08466239273548126, "global_step": 32743, "epoch": 308, "lr": 3.191385925817586e-05} +{"train_loss": 0.015703752636909485, "grad_norm": 0.12156815826892853, "global_step": 32744, "epoch": 308, "lr": 3.191109621271461e-05} +{"train_loss": 0.01576085388660431, "grad_norm": 0.0954715758562088, "global_step": 32745, "epoch": 308, "lr": 3.190833323080993e-05} +{"train_loss": 0.01498415693640709, "grad_norm": 0.10648767650127411, "global_step": 32746, "epoch": 308, "lr": 3.190557031247148e-05} +{"train_loss": 0.015755154192447662, "grad_norm": 0.06215953826904297, "global_step": 32747, "epoch": 308, "lr": 3.1902807457708986e-05} +{"train_loss": 0.01677963137626648, "grad_norm": 0.08892975002527237, "global_step": 32748, "epoch": 308, "lr": 3.190004466653215e-05} +{"train_loss": 0.014203978702425957, "grad_norm": 0.07942458242177963, "global_step": 32749, "epoch": 308, "lr": 3.189728193895069e-05} +{"train_loss": 0.015582048334181309, "grad_norm": 0.14101360738277435, "global_step": 32750, "epoch": 308, "lr": 3.189451927497431e-05} +{"train_loss": 0.01709236204624176, "grad_norm": 0.15380002558231354, "global_step": 32751, "epoch": 308, "lr": 3.189175667461273e-05} +{"train_loss": 0.014808392152190208, "grad_norm": 0.05985075980424881, "global_step": 32752, "epoch": 308, "lr": 3.188899413787563e-05} +{"train_loss": 0.015268616562814644, "grad_norm": 0.09010707587003708, "global_step": 32753, "epoch": 308, "lr": 3.188623166477272e-05, "val_loss": 0.019477497786283493} +{"train_loss": 0.015026389621198177, "grad_norm": 0.08197102695703506, "global_step": 32754, "epoch": 309, "lr": 3.1883469255313724e-05} +{"train_loss": 0.013198615051805973, "grad_norm": 0.06843086332082748, "global_step": 32755, "epoch": 309, "lr": 3.188070690950833e-05} +{"train_loss": 0.017708584666252136, "grad_norm": 0.08707060664892197, "global_step": 32756, "epoch": 309, "lr": 3.187794462736625e-05} +{"train_loss": 0.013763686642050743, "grad_norm": 0.0638330802321434, "global_step": 32757, "epoch": 309, "lr": 3.1875182408897214e-05} +{"train_loss": 0.012961437925696373, "grad_norm": 0.06431148946285248, "global_step": 32758, "epoch": 309, "lr": 3.187242025411089e-05} +{"train_loss": 0.014824684709310532, "grad_norm": 0.05686045438051224, "global_step": 32759, "epoch": 309, "lr": 3.1869658163017e-05} +{"train_loss": 0.014012256637215614, "grad_norm": 0.05308905616402626, "global_step": 32760, "epoch": 309, "lr": 3.1866896135625236e-05} +{"train_loss": 0.01244882307946682, "grad_norm": 0.05531042814254761, "global_step": 32761, "epoch": 309, "lr": 3.186413417194533e-05} +{"train_loss": 0.015922121703624725, "grad_norm": 0.06690610945224762, "global_step": 32762, "epoch": 309, "lr": 3.186137227198696e-05} +{"train_loss": 0.015092751011252403, "grad_norm": 0.05978957936167717, "global_step": 32763, "epoch": 309, "lr": 3.185861043575985e-05} +{"train_loss": 0.016174426302313805, "grad_norm": 0.10334643721580505, "global_step": 32764, "epoch": 309, "lr": 3.185584866327368e-05} +{"train_loss": 0.015564252622425556, "grad_norm": 0.08038253337144852, "global_step": 32765, "epoch": 309, "lr": 3.185308695453816e-05} +{"train_loss": 0.012906024232506752, "grad_norm": 0.09819662570953369, "global_step": 32766, "epoch": 309, "lr": 3.1850325309563014e-05} +{"train_loss": 0.018101971596479416, "grad_norm": 0.07089130580425262, "global_step": 32767, "epoch": 309, "lr": 3.1847563728357934e-05} +{"train_loss": 0.01556041557341814, "grad_norm": 0.12776845693588257, "global_step": 32768, "epoch": 309, "lr": 3.184480221093261e-05} +{"train_loss": 0.01638602651655674, "grad_norm": 0.113564133644104, "global_step": 32769, "epoch": 309, "lr": 3.184204075729678e-05} +{"train_loss": 0.0135699063539505, "grad_norm": 0.0792589858174324, "global_step": 32770, "epoch": 309, "lr": 3.18392793674601e-05} +{"train_loss": 0.016481807455420494, "grad_norm": 0.08752897381782532, "global_step": 32771, "epoch": 309, "lr": 3.183651804143229e-05} +{"train_loss": 0.01690816506743431, "grad_norm": 0.06341695040464401, "global_step": 32772, "epoch": 309, "lr": 3.183375677922306e-05} +{"train_loss": 0.015528889372944832, "grad_norm": 0.07316496968269348, "global_step": 32773, "epoch": 309, "lr": 3.183099558084212e-05} +{"train_loss": 0.017598304897546768, "grad_norm": 0.08681444823741913, "global_step": 32774, "epoch": 309, "lr": 3.182823444629916e-05} +{"train_loss": 0.016871396452188492, "grad_norm": 0.07524517923593521, "global_step": 32775, "epoch": 309, "lr": 3.1825473375603874e-05} +{"train_loss": 0.017887968569993973, "grad_norm": 0.08214753121137619, "global_step": 32776, "epoch": 309, "lr": 3.1822712368765974e-05} +{"train_loss": 0.015030000358819962, "grad_norm": 0.06460965424776077, "global_step": 32777, "epoch": 309, "lr": 3.1819951425795144e-05} +{"train_loss": 0.016359660774469376, "grad_norm": 0.0764426589012146, "global_step": 32778, "epoch": 309, "lr": 3.1817190546701116e-05} +{"train_loss": 0.01661060005426407, "grad_norm": 0.0699014663696289, "global_step": 32779, "epoch": 309, "lr": 3.1814429731493565e-05} +{"train_loss": 0.015081238001585007, "grad_norm": 0.09570774435997009, "global_step": 32780, "epoch": 309, "lr": 3.1811668980182215e-05} +{"train_loss": 0.015262901782989502, "grad_norm": 0.0779753029346466, "global_step": 32781, "epoch": 309, "lr": 3.1808908292776736e-05} +{"train_loss": 0.016910497099161148, "grad_norm": 0.15380960702896118, "global_step": 32782, "epoch": 309, "lr": 3.180614766928683e-05} +{"train_loss": 0.014616843312978745, "grad_norm": 0.06918594241142273, "global_step": 32783, "epoch": 309, "lr": 3.180338710972223e-05} +{"train_loss": 0.013681989163160324, "grad_norm": 0.1353774070739746, "global_step": 32784, "epoch": 309, "lr": 3.18006266140926e-05} +{"train_loss": 0.017684340476989746, "grad_norm": 0.09615530073642731, "global_step": 32785, "epoch": 309, "lr": 3.179786618240765e-05} +{"train_loss": 0.013262433931231499, "grad_norm": 0.09162808209657669, "global_step": 32786, "epoch": 309, "lr": 3.17951058146771e-05} +{"train_loss": 0.014232264831662178, "grad_norm": 0.09089967608451843, "global_step": 32787, "epoch": 309, "lr": 3.179234551091062e-05} +{"train_loss": 0.014252318069338799, "grad_norm": 0.05613201484084129, "global_step": 32788, "epoch": 309, "lr": 3.178958527111793e-05} +{"train_loss": 0.014974908903241158, "grad_norm": 0.06480491161346436, "global_step": 32789, "epoch": 309, "lr": 3.1786825095308696e-05} +{"train_loss": 0.015542829409241676, "grad_norm": 0.07240410894155502, "global_step": 32790, "epoch": 309, "lr": 3.178406498349265e-05} +{"train_loss": 0.013599388301372528, "grad_norm": 0.07810335606336594, "global_step": 32791, "epoch": 309, "lr": 3.178130493567948e-05} +{"train_loss": 0.01678408868610859, "grad_norm": 0.06428100913763046, "global_step": 32792, "epoch": 309, "lr": 3.17785449518789e-05} +{"train_loss": 0.014818927273154259, "grad_norm": 0.09556885808706284, "global_step": 32793, "epoch": 309, "lr": 3.177578503210057e-05} +{"train_loss": 0.01739589497447014, "grad_norm": 0.09157223999500275, "global_step": 32794, "epoch": 309, "lr": 3.177302517635419e-05} +{"train_loss": 0.01647552102804184, "grad_norm": 0.09621716290712357, "global_step": 32795, "epoch": 309, "lr": 3.1770265384649487e-05} +{"train_loss": 0.01589251309633255, "grad_norm": 0.0707775130867958, "global_step": 32796, "epoch": 309, "lr": 3.176750565699615e-05} +{"train_loss": 0.013956011272966862, "grad_norm": 0.08827739208936691, "global_step": 32797, "epoch": 309, "lr": 3.176474599340385e-05} +{"train_loss": 0.013478782959282398, "grad_norm": 0.0640551894903183, "global_step": 32798, "epoch": 309, "lr": 3.1761986393882335e-05} +{"train_loss": 0.015796517953276634, "grad_norm": 0.10550080239772797, "global_step": 32799, "epoch": 309, "lr": 3.175922685844125e-05} +{"train_loss": 0.014681834727525711, "grad_norm": 0.09695804864168167, "global_step": 32800, "epoch": 309, "lr": 3.1756467387090306e-05} +{"train_loss": 0.015272056683897972, "grad_norm": 0.07235261052846909, "global_step": 32801, "epoch": 309, "lr": 3.1753707979839195e-05} +{"train_loss": 0.015353258699178696, "grad_norm": 0.10950643569231033, "global_step": 32802, "epoch": 309, "lr": 3.175094863669763e-05} +{"train_loss": 0.012409594841301441, "grad_norm": 0.09387084096670151, "global_step": 32803, "epoch": 309, "lr": 3.174818935767529e-05} +{"train_loss": 0.015471654012799263, "grad_norm": 0.06144314631819725, "global_step": 32804, "epoch": 309, "lr": 3.174543014278189e-05} +{"train_loss": 0.015191884711384773, "grad_norm": 0.09561523795127869, "global_step": 32805, "epoch": 309, "lr": 3.17426709920271e-05} +{"train_loss": 0.01613384671509266, "grad_norm": 0.05307629704475403, "global_step": 32806, "epoch": 309, "lr": 3.1739911905420617e-05} +{"train_loss": 0.01566704362630844, "grad_norm": 0.0838203951716423, "global_step": 32807, "epoch": 309, "lr": 3.173715288297214e-05} +{"train_loss": 0.016045168042182922, "grad_norm": 0.10375383496284485, "global_step": 32808, "epoch": 309, "lr": 3.1734393924691374e-05} +{"train_loss": 0.015382744371891022, "grad_norm": 0.07544423639774323, "global_step": 32809, "epoch": 309, "lr": 3.1731635030587995e-05} +{"train_loss": 0.015419067814946175, "grad_norm": 0.05997597053647041, "global_step": 32810, "epoch": 309, "lr": 3.172887620067174e-05} +{"train_loss": 0.013589577749371529, "grad_norm": 0.08781062066555023, "global_step": 32811, "epoch": 309, "lr": 3.172611743495223e-05} +{"train_loss": 0.01402167696505785, "grad_norm": 0.17094363272190094, "global_step": 32812, "epoch": 309, "lr": 3.1723358733439216e-05} +{"train_loss": 0.014872532337903976, "grad_norm": 0.06410880386829376, "global_step": 32813, "epoch": 309, "lr": 3.172060009614236e-05} +{"train_loss": 0.014394339174032211, "grad_norm": 0.06555500626564026, "global_step": 32814, "epoch": 309, "lr": 3.1717841523071355e-05} +{"train_loss": 0.015044274739921093, "grad_norm": 0.0654781237244606, "global_step": 32815, "epoch": 309, "lr": 3.1715083014235924e-05} +{"train_loss": 0.014960480853915215, "grad_norm": 0.07150056213140488, "global_step": 32816, "epoch": 309, "lr": 3.171232456964575e-05} +{"train_loss": 0.018405064940452576, "grad_norm": 0.096544548869133, "global_step": 32817, "epoch": 309, "lr": 3.1709566189310494e-05} +{"train_loss": 0.01451928447932005, "grad_norm": 0.08048804849386215, "global_step": 32818, "epoch": 309, "lr": 3.170680787323986e-05} +{"train_loss": 0.013188442215323448, "grad_norm": 0.09327412396669388, "global_step": 32819, "epoch": 309, "lr": 3.1704049621443564e-05} +{"train_loss": 0.014817780815064907, "grad_norm": 0.1469946652650833, "global_step": 32820, "epoch": 309, "lr": 3.170129143393128e-05} +{"train_loss": 0.015440687537193298, "grad_norm": 0.11783525347709656, "global_step": 32821, "epoch": 309, "lr": 3.169853331071271e-05} +{"train_loss": 0.014969535171985626, "grad_norm": 0.06861893087625504, "global_step": 32822, "epoch": 309, "lr": 3.169577525179752e-05} +{"train_loss": 0.01471179910004139, "grad_norm": 0.14027218520641327, "global_step": 32823, "epoch": 309, "lr": 3.169301725719541e-05} +{"train_loss": 0.01711803488433361, "grad_norm": 0.17165015637874603, "global_step": 32824, "epoch": 309, "lr": 3.1690259326916084e-05} +{"train_loss": 0.018155422061681747, "grad_norm": 0.06532039493322372, "global_step": 32825, "epoch": 309, "lr": 3.168750146096923e-05} +{"train_loss": 0.013838639482855797, "grad_norm": 0.13484641909599304, "global_step": 32826, "epoch": 309, "lr": 3.168474365936451e-05} +{"train_loss": 0.01749902404844761, "grad_norm": 0.06935848295688629, "global_step": 32827, "epoch": 309, "lr": 3.168198592211166e-05} +{"train_loss": 0.015271386131644249, "grad_norm": 0.08167403936386108, "global_step": 32828, "epoch": 309, "lr": 3.167922824922034e-05} +{"train_loss": 0.014999521896243095, "grad_norm": 0.07036220282316208, "global_step": 32829, "epoch": 309, "lr": 3.167647064070024e-05} +{"train_loss": 0.015534882433712482, "grad_norm": 0.08433783054351807, "global_step": 32830, "epoch": 309, "lr": 3.167371309656104e-05} +{"train_loss": 0.015690594911575317, "grad_norm": 0.12781672179698944, "global_step": 32831, "epoch": 309, "lr": 3.167095561681246e-05} +{"train_loss": 0.014567868784070015, "grad_norm": 0.12262807041406631, "global_step": 32832, "epoch": 309, "lr": 3.1668198201464175e-05} +{"train_loss": 0.015908416360616684, "grad_norm": 0.12479275465011597, "global_step": 32833, "epoch": 309, "lr": 3.166544085052587e-05} +{"train_loss": 0.015373162925243378, "grad_norm": 0.11863779276609421, "global_step": 32834, "epoch": 309, "lr": 3.166268356400722e-05} +{"train_loss": 0.015993032604455948, "grad_norm": 0.07689306139945984, "global_step": 32835, "epoch": 309, "lr": 3.165992634191792e-05} +{"train_loss": 0.016151122748851776, "grad_norm": 0.08797850459814072, "global_step": 32836, "epoch": 309, "lr": 3.1657169184267675e-05} +{"train_loss": 0.0145435631275177, "grad_norm": 0.07880445569753647, "global_step": 32837, "epoch": 309, "lr": 3.165441209106616e-05} +{"train_loss": 0.012548718601465225, "grad_norm": 0.07406642287969589, "global_step": 32838, "epoch": 309, "lr": 3.165165506232305e-05} +{"train_loss": 0.014264519326388836, "grad_norm": 0.09674517065286636, "global_step": 32839, "epoch": 309, "lr": 3.164889809804808e-05} +{"train_loss": 0.016216792166233063, "grad_norm": 0.12205256521701813, "global_step": 32840, "epoch": 309, "lr": 3.164614119825087e-05} +{"train_loss": 0.015805691480636597, "grad_norm": 0.12114295363426208, "global_step": 32841, "epoch": 309, "lr": 3.164338436294114e-05} +{"train_loss": 0.013438057154417038, "grad_norm": 0.08989262580871582, "global_step": 32842, "epoch": 309, "lr": 3.164062759212858e-05} +{"train_loss": 0.015611274167895317, "grad_norm": 0.11068461835384369, "global_step": 32843, "epoch": 309, "lr": 3.163787088582286e-05} +{"train_loss": 0.016051815822720528, "grad_norm": 0.11551588773727417, "global_step": 32844, "epoch": 309, "lr": 3.163511424403368e-05} +{"train_loss": 0.01591968722641468, "grad_norm": 0.07627588510513306, "global_step": 32845, "epoch": 309, "lr": 3.163235766677074e-05} +{"train_loss": 0.012367631308734417, "grad_norm": 0.10496782511472702, "global_step": 32846, "epoch": 309, "lr": 3.16296011540437e-05} +{"train_loss": 0.013731859624385834, "grad_norm": 0.06251883506774902, "global_step": 32847, "epoch": 309, "lr": 3.1626844705862236e-05} +{"train_loss": 0.016071053221821785, "grad_norm": 0.06888087838888168, "global_step": 32848, "epoch": 309, "lr": 3.1624088322236054e-05} +{"train_loss": 0.017054181545972824, "grad_norm": 0.09728585928678513, "global_step": 32849, "epoch": 309, "lr": 3.1621332003174845e-05} +{"train_loss": 0.015831559896469116, "grad_norm": 0.06310975551605225, "global_step": 32850, "epoch": 309, "lr": 3.161857574868827e-05} +{"train_loss": 0.014058507978916168, "grad_norm": 0.053742457181215286, "global_step": 32851, "epoch": 309, "lr": 3.161581955878605e-05} +{"train_loss": 0.015913454815745354, "grad_norm": 0.07892832905054092, "global_step": 32852, "epoch": 309, "lr": 3.161306343347782e-05} +{"train_loss": 0.014625411480665207, "grad_norm": 0.10692183673381805, "global_step": 32853, "epoch": 309, "lr": 3.16103073727733e-05} +{"train_loss": 0.014873149804770947, "grad_norm": 0.0461660698056221, "global_step": 32854, "epoch": 309, "lr": 3.160755137668216e-05} +{"train_loss": 0.01634494960308075, "grad_norm": 0.0554293729364872, "global_step": 32855, "epoch": 309, "lr": 3.1604795445214075e-05} +{"train_loss": 0.014840701594948769, "grad_norm": 0.08378145098686218, "global_step": 32856, "epoch": 309, "lr": 3.160203957837875e-05} +{"train_loss": 0.012871475890278816, "grad_norm": 0.05147711932659149, "global_step": 32857, "epoch": 309, "lr": 3.159928377618587e-05} +{"train_loss": 0.016976626589894295, "grad_norm": 0.05859067663550377, "global_step": 32858, "epoch": 309, "lr": 3.159652803864509e-05} +{"train_loss": 0.015252449576092779, "grad_norm": 0.0871216282248497, "global_step": 32859, "epoch": 309, "lr": 3.1593772365766105e-05, "val_loss": 0.019236929714679718} +{"train_loss": 0.015761077404022217, "grad_norm": 0.1303800344467163, "global_step": 32860, "epoch": 310, "lr": 3.15910167575586e-05} +{"train_loss": 0.016236912459135056, "grad_norm": 0.07297186553478241, "global_step": 32861, "epoch": 310, "lr": 3.158826121403226e-05} +{"train_loss": 0.01475559826940298, "grad_norm": 0.09820235520601273, "global_step": 32862, "epoch": 310, "lr": 3.158550573519675e-05} +{"train_loss": 0.015589284710586071, "grad_norm": 0.055208753794431686, "global_step": 32863, "epoch": 310, "lr": 3.15827503210618e-05} +{"train_loss": 0.014327525161206722, "grad_norm": 0.1680326908826828, "global_step": 32864, "epoch": 310, "lr": 3.1579994971637026e-05} +{"train_loss": 0.016883110627532005, "grad_norm": 0.059809360653162, "global_step": 32865, "epoch": 310, "lr": 3.1577239686932146e-05} +{"train_loss": 0.014937033876776695, "grad_norm": 0.13917270302772522, "global_step": 32866, "epoch": 310, "lr": 3.157448446695683e-05} +{"train_loss": 0.015423191711306572, "grad_norm": 0.13445907831192017, "global_step": 32867, "epoch": 310, "lr": 3.1571729311720756e-05} +{"train_loss": 0.013525428250432014, "grad_norm": 0.06813456118106842, "global_step": 32868, "epoch": 310, "lr": 3.1568974221233626e-05} +{"train_loss": 0.015875037759542465, "grad_norm": 0.05861565098166466, "global_step": 32869, "epoch": 310, "lr": 3.1566219195505106e-05} +{"train_loss": 0.014355536550283432, "grad_norm": 0.06633356213569641, "global_step": 32870, "epoch": 310, "lr": 3.156346423454487e-05} +{"train_loss": 0.014478309080004692, "grad_norm": 0.08092594891786575, "global_step": 32871, "epoch": 310, "lr": 3.156070933836261e-05} +{"train_loss": 0.01658741757273674, "grad_norm": 0.1515478789806366, "global_step": 32872, "epoch": 310, "lr": 3.155795450696798e-05} +{"train_loss": 0.015123496763408184, "grad_norm": 0.06180568411946297, "global_step": 32873, "epoch": 310, "lr": 3.155519974037069e-05} +{"train_loss": 0.012627750635147095, "grad_norm": 0.12107347697019577, "global_step": 32874, "epoch": 310, "lr": 3.1552445038580416e-05} +{"train_loss": 0.013549983501434326, "grad_norm": 0.12227080017328262, "global_step": 32875, "epoch": 310, "lr": 3.1549690401606814e-05} +{"train_loss": 0.012558640912175179, "grad_norm": 0.08462247252464294, "global_step": 32876, "epoch": 310, "lr": 3.1546935829459565e-05} +{"train_loss": 0.01807696372270584, "grad_norm": 0.0817611888051033, "global_step": 32877, "epoch": 310, "lr": 3.154418132214837e-05} +{"train_loss": 0.014633497223258018, "grad_norm": 0.07921671122312546, "global_step": 32878, "epoch": 310, "lr": 3.15414268796829e-05} +{"train_loss": 0.01352771744132042, "grad_norm": 0.05700362101197243, "global_step": 32879, "epoch": 310, "lr": 3.1538672502072817e-05} +{"train_loss": 0.016244854778051376, "grad_norm": 0.0772949755191803, "global_step": 32880, "epoch": 310, "lr": 3.153591818932784e-05} +{"train_loss": 0.015142278745770454, "grad_norm": 0.0784962922334671, "global_step": 32881, "epoch": 310, "lr": 3.153316394145759e-05} +{"train_loss": 0.015576077625155449, "grad_norm": 0.057294394820928574, "global_step": 32882, "epoch": 310, "lr": 3.153040975847177e-05} +{"train_loss": 0.017938904464244843, "grad_norm": 0.1661912053823471, "global_step": 32883, "epoch": 310, "lr": 3.152765564038007e-05} +{"train_loss": 0.014970322139561176, "grad_norm": 0.1407698094844818, "global_step": 32884, "epoch": 310, "lr": 3.152490158719214e-05} +{"train_loss": 0.015193674713373184, "grad_norm": 0.07833051681518555, "global_step": 32885, "epoch": 310, "lr": 3.152214759891768e-05} +{"train_loss": 0.014196999371051788, "grad_norm": 0.06862270832061768, "global_step": 32886, "epoch": 310, "lr": 3.151939367556637e-05} +{"train_loss": 0.013160749338567257, "grad_norm": 0.05630992725491524, "global_step": 32887, "epoch": 310, "lr": 3.1516639817147866e-05} +{"train_loss": 0.014126630499958992, "grad_norm": 0.1613694429397583, "global_step": 32888, "epoch": 310, "lr": 3.151388602367183e-05} +{"train_loss": 0.014221258461475372, "grad_norm": 0.06550592929124832, "global_step": 32889, "epoch": 310, "lr": 3.1511132295147985e-05} +{"train_loss": 0.016472117975354195, "grad_norm": 0.08688368648290634, "global_step": 32890, "epoch": 310, "lr": 3.150837863158597e-05} +{"train_loss": 0.014333423227071762, "grad_norm": 0.05296105518937111, "global_step": 32891, "epoch": 310, "lr": 3.150562503299546e-05} +{"train_loss": 0.01505009550601244, "grad_norm": 0.10138433426618576, "global_step": 32892, "epoch": 310, "lr": 3.150287149938618e-05} +{"train_loss": 0.01656358502805233, "grad_norm": 0.0846528485417366, "global_step": 32893, "epoch": 310, "lr": 3.150011803076772e-05} +{"train_loss": 0.014752771705389023, "grad_norm": 0.14641021192073822, "global_step": 32894, "epoch": 310, "lr": 3.1497364627149824e-05} +{"train_loss": 0.014749222435057163, "grad_norm": 0.12620598077774048, "global_step": 32895, "epoch": 310, "lr": 3.149461128854213e-05} +{"train_loss": 0.013645884580910206, "grad_norm": 0.08928833901882172, "global_step": 32896, "epoch": 310, "lr": 3.149185801495432e-05} +{"train_loss": 0.01789795607328415, "grad_norm": 0.0626852959394455, "global_step": 32897, "epoch": 310, "lr": 3.148910480639609e-05} +{"train_loss": 0.01593957655131817, "grad_norm": 0.0684405267238617, "global_step": 32898, "epoch": 310, "lr": 3.14863516628771e-05} +{"train_loss": 0.015084706246852875, "grad_norm": 0.05380147323012352, "global_step": 32899, "epoch": 310, "lr": 3.1483598584407006e-05} +{"train_loss": 0.015967587009072304, "grad_norm": 0.08222374320030212, "global_step": 32900, "epoch": 310, "lr": 3.1480845570995496e-05} +{"train_loss": 0.013784145936369896, "grad_norm": 0.07457279413938522, "global_step": 32901, "epoch": 310, "lr": 3.147809262265223e-05} +{"train_loss": 0.01502479612827301, "grad_norm": 0.10148833692073822, "global_step": 32902, "epoch": 310, "lr": 3.14753397393869e-05} +{"train_loss": 0.014452444389462471, "grad_norm": 0.06893134862184525, "global_step": 32903, "epoch": 310, "lr": 3.147258692120916e-05} +{"train_loss": 0.016278881579637527, "grad_norm": 0.1328655332326889, "global_step": 32904, "epoch": 310, "lr": 3.146983416812873e-05} +{"train_loss": 0.016463415697216988, "grad_norm": 0.06528014689683914, "global_step": 32905, "epoch": 310, "lr": 3.1467081480155206e-05} +{"train_loss": 0.015345435589551926, "grad_norm": 0.08032024651765823, "global_step": 32906, "epoch": 310, "lr": 3.146432885729831e-05} +{"train_loss": 0.015570370480418205, "grad_norm": 0.13410066068172455, "global_step": 32907, "epoch": 310, "lr": 3.14615762995677e-05} +{"train_loss": 0.015376517549157143, "grad_norm": 0.09248727560043335, "global_step": 32908, "epoch": 310, "lr": 3.145882380697304e-05} +{"train_loss": 0.01454379316419363, "grad_norm": 0.06433273106813431, "global_step": 32909, "epoch": 310, "lr": 3.145607137952402e-05} +{"train_loss": 0.013972317799925804, "grad_norm": 0.08062854409217834, "global_step": 32910, "epoch": 310, "lr": 3.145331901723031e-05} +{"train_loss": 0.016093827784061432, "grad_norm": 0.0862375944852829, "global_step": 32911, "epoch": 310, "lr": 3.1450566720101564e-05} +{"train_loss": 0.01892077922821045, "grad_norm": 0.15459276735782623, "global_step": 32912, "epoch": 310, "lr": 3.144781448814746e-05} +{"train_loss": 0.01368441991508007, "grad_norm": 0.13984018564224243, "global_step": 32913, "epoch": 310, "lr": 3.1445062321377654e-05} +{"train_loss": 0.015131408348679543, "grad_norm": 0.1133291944861412, "global_step": 32914, "epoch": 310, "lr": 3.144231021980184e-05} +{"train_loss": 0.01601756177842617, "grad_norm": 0.09965850412845612, "global_step": 32915, "epoch": 310, "lr": 3.143955818342968e-05} +{"train_loss": 0.014482403174042702, "grad_norm": 0.06729641556739807, "global_step": 32916, "epoch": 310, "lr": 3.1436806212270854e-05} +{"train_loss": 0.016589602455496788, "grad_norm": 0.08067870885133743, "global_step": 32917, "epoch": 310, "lr": 3.1434054306334993e-05} +{"train_loss": 0.012712391093373299, "grad_norm": 0.14291085302829742, "global_step": 32918, "epoch": 310, "lr": 3.1431302465631796e-05} +{"train_loss": 0.012870529666543007, "grad_norm": 0.048319652676582336, "global_step": 32919, "epoch": 310, "lr": 3.142855069017093e-05} +{"train_loss": 0.014606660231947899, "grad_norm": 0.13309355080127716, "global_step": 32920, "epoch": 310, "lr": 3.142579897996206e-05} +{"train_loss": 0.016262751072645187, "grad_norm": 0.08566004037857056, "global_step": 32921, "epoch": 310, "lr": 3.1423047335014855e-05} +{"train_loss": 0.015048615634441376, "grad_norm": 0.06480053067207336, "global_step": 32922, "epoch": 310, "lr": 3.142029575533899e-05} +{"train_loss": 0.017503008246421814, "grad_norm": 0.09259495139122009, "global_step": 32923, "epoch": 310, "lr": 3.141754424094412e-05} +{"train_loss": 0.017438184469938278, "grad_norm": 0.07845774292945862, "global_step": 32924, "epoch": 310, "lr": 3.141479279183992e-05} +{"train_loss": 0.016141848638653755, "grad_norm": 0.15215060114860535, "global_step": 32925, "epoch": 310, "lr": 3.141204140803605e-05} +{"train_loss": 0.013751558028161526, "grad_norm": 0.12002307921648026, "global_step": 32926, "epoch": 310, "lr": 3.140929008954219e-05} +{"train_loss": 0.013597402721643448, "grad_norm": 0.14490003883838654, "global_step": 32927, "epoch": 310, "lr": 3.1406538836368005e-05} +{"train_loss": 0.014679732732474804, "grad_norm": 0.07745739072561264, "global_step": 32928, "epoch": 310, "lr": 3.140378764852315e-05} +{"train_loss": 0.01673399657011032, "grad_norm": 0.178809255361557, "global_step": 32929, "epoch": 310, "lr": 3.140103652601728e-05} +{"train_loss": 0.0166306272149086, "grad_norm": 0.0658189207315445, "global_step": 32930, "epoch": 310, "lr": 3.13982854688601e-05} +{"train_loss": 0.014867208898067474, "grad_norm": 0.10951819270849228, "global_step": 32931, "epoch": 310, "lr": 3.139553447706125e-05} +{"train_loss": 0.01628681644797325, "grad_norm": 0.05648832023143768, "global_step": 32932, "epoch": 310, "lr": 3.139278355063039e-05} +{"train_loss": 0.014213893562555313, "grad_norm": 0.06252717226743698, "global_step": 32933, "epoch": 310, "lr": 3.139003268957723e-05} +{"train_loss": 0.015350716188549995, "grad_norm": 0.09191127866506577, "global_step": 32934, "epoch": 310, "lr": 3.138728189391137e-05} +{"train_loss": 0.01518227905035019, "grad_norm": 0.07413414865732193, "global_step": 32935, "epoch": 310, "lr": 3.138453116364252e-05} +{"train_loss": 0.014988398179411888, "grad_norm": 0.14982600510120392, "global_step": 32936, "epoch": 310, "lr": 3.138178049878032e-05} +{"train_loss": 0.014635845087468624, "grad_norm": 0.12087174504995346, "global_step": 32937, "epoch": 310, "lr": 3.137902989933445e-05} +{"train_loss": 0.015491755679249763, "grad_norm": 0.12002820521593094, "global_step": 32938, "epoch": 310, "lr": 3.137627936531458e-05} +{"train_loss": 0.014212898910045624, "grad_norm": 0.06484437733888626, "global_step": 32939, "epoch": 310, "lr": 3.137352889673037e-05} +{"train_loss": 0.018422473222017288, "grad_norm": 0.07665753364562988, "global_step": 32940, "epoch": 310, "lr": 3.137077849359147e-05} +{"train_loss": 0.016701005399227142, "grad_norm": 0.08270623534917831, "global_step": 32941, "epoch": 310, "lr": 3.136802815590755e-05} +{"train_loss": 0.015132932923734188, "grad_norm": 0.10244104266166687, "global_step": 32942, "epoch": 310, "lr": 3.136527788368827e-05} +{"train_loss": 0.015578498132526875, "grad_norm": 0.09297177195549011, "global_step": 32943, "epoch": 310, "lr": 3.136252767694331e-05} +{"train_loss": 0.0146261565387249, "grad_norm": 0.08240170776844025, "global_step": 32944, "epoch": 310, "lr": 3.135977753568232e-05} +{"train_loss": 0.01399244274944067, "grad_norm": 0.08697867393493652, "global_step": 32945, "epoch": 310, "lr": 3.1357027459914976e-05} +{"train_loss": 0.016535751521587372, "grad_norm": 0.09952503442764282, "global_step": 32946, "epoch": 310, "lr": 3.135427744965091e-05} +{"train_loss": 0.016869543120265007, "grad_norm": 0.056652773171663284, "global_step": 32947, "epoch": 310, "lr": 3.135152750489982e-05} +{"train_loss": 0.015027125366032124, "grad_norm": 0.10246045142412186, "global_step": 32948, "epoch": 310, "lr": 3.1348777625671347e-05} +{"train_loss": 0.014671910554170609, "grad_norm": 0.11366371810436249, "global_step": 32949, "epoch": 310, "lr": 3.134602781197515e-05} +{"train_loss": 0.014091916382312775, "grad_norm": 0.080546073615551, "global_step": 32950, "epoch": 310, "lr": 3.1343278063820905e-05} +{"train_loss": 0.01659245416522026, "grad_norm": 0.07106447219848633, "global_step": 32951, "epoch": 310, "lr": 3.134052838121828e-05} +{"train_loss": 0.014082655310630798, "grad_norm": 0.06605508178472519, "global_step": 32952, "epoch": 310, "lr": 3.133777876417692e-05} +{"train_loss": 0.012576743960380554, "grad_norm": 0.07072247564792633, "global_step": 32953, "epoch": 310, "lr": 3.133502921270648e-05} +{"train_loss": 0.015300567261874676, "grad_norm": 0.06817414611577988, "global_step": 32954, "epoch": 310, "lr": 3.133227972681662e-05} +{"train_loss": 0.015570702962577343, "grad_norm": 0.09147234261035919, "global_step": 32955, "epoch": 310, "lr": 3.132953030651702e-05} +{"train_loss": 0.013662371784448624, "grad_norm": 0.07134772837162018, "global_step": 32956, "epoch": 310, "lr": 3.132678095181734e-05} +{"train_loss": 0.01571597158908844, "grad_norm": 0.07295035570859909, "global_step": 32957, "epoch": 310, "lr": 3.132403166272723e-05} +{"train_loss": 0.015606401488184929, "grad_norm": 0.07264166325330734, "global_step": 32958, "epoch": 310, "lr": 3.1321282439256337e-05} +{"train_loss": 0.015171356499195099, "grad_norm": 0.09641467034816742, "global_step": 32959, "epoch": 310, "lr": 3.131853328141434e-05} +{"train_loss": 0.016287222504615784, "grad_norm": 0.08539469540119171, "global_step": 32960, "epoch": 310, "lr": 3.13157841892109e-05} +{"train_loss": 0.017072882503271103, "grad_norm": 0.14893797039985657, "global_step": 32961, "epoch": 310, "lr": 3.1313035162655654e-05} +{"train_loss": 0.015557169914245605, "grad_norm": 0.10274288803339005, "global_step": 32962, "epoch": 310, "lr": 3.131028620175829e-05} +{"train_loss": 0.015176011249423027, "grad_norm": 0.09774211049079895, "global_step": 32963, "epoch": 310, "lr": 3.1307537306528456e-05} +{"train_loss": 0.016038361936807632, "grad_norm": 0.08012236654758453, "global_step": 32964, "epoch": 310, "lr": 3.13047884769758e-05} +{"train_loss": 0.015262990671579005, "grad_norm": 0.12017572671175003, "global_step": 32965, "epoch": 310, "lr": 3.130203971310999e-05, "train_batch_mse_error": 0.013781823217868805, "val_loss": 0.01926618628203869} +{"train_loss": 0.014693904668092728, "grad_norm": 0.10016592592000961, "global_step": 32966, "epoch": 311, "lr": 3.129929101494067e-05} +{"train_loss": 0.01582583785057068, "grad_norm": 0.0773499608039856, "global_step": 32967, "epoch": 311, "lr": 3.129654238247751e-05} +{"train_loss": 0.01714332401752472, "grad_norm": 0.13388323783874512, "global_step": 32968, "epoch": 311, "lr": 3.1293793815730177e-05} +{"train_loss": 0.014328500255942345, "grad_norm": 0.056892894208431244, "global_step": 32969, "epoch": 311, "lr": 3.129104531470832e-05} +{"train_loss": 0.01662736013531685, "grad_norm": 0.06970538198947906, "global_step": 32970, "epoch": 311, "lr": 3.128829687942159e-05} +{"train_loss": 0.0157708041369915, "grad_norm": 0.13918933272361755, "global_step": 32971, "epoch": 311, "lr": 3.128554850987962e-05} +{"train_loss": 0.016987016424536705, "grad_norm": 0.09506962448358536, "global_step": 32972, "epoch": 311, "lr": 3.128280020609212e-05} +{"train_loss": 0.015767471864819527, "grad_norm": 0.10745421051979065, "global_step": 32973, "epoch": 311, "lr": 3.128005196806872e-05} +{"train_loss": 0.01607564650475979, "grad_norm": 0.09721380472183228, "global_step": 32974, "epoch": 311, "lr": 3.1277303795819093e-05} +{"train_loss": 0.01425097044557333, "grad_norm": 0.07647181302309036, "global_step": 32975, "epoch": 311, "lr": 3.127455568935285e-05} +{"train_loss": 0.014769800938665867, "grad_norm": 0.08743997663259506, "global_step": 32976, "epoch": 311, "lr": 3.127180764867967e-05} +{"train_loss": 0.01551113836467266, "grad_norm": 0.114414282143116, "global_step": 32977, "epoch": 311, "lr": 3.126905967380923e-05} +{"train_loss": 0.015571667812764645, "grad_norm": 0.13183613121509552, "global_step": 32978, "epoch": 311, "lr": 3.126631176475116e-05} +{"train_loss": 0.015255767852067947, "grad_norm": 0.12268485873937607, "global_step": 32979, "epoch": 311, "lr": 3.1263563921515125e-05} +{"train_loss": 0.01674303412437439, "grad_norm": 0.08199069648981094, "global_step": 32980, "epoch": 311, "lr": 3.126081614411079e-05} +{"train_loss": 0.015333188697695732, "grad_norm": 0.08481737971305847, "global_step": 32981, "epoch": 311, "lr": 3.1258068432547794e-05} +{"train_loss": 0.014627872966229916, "grad_norm": 0.06343495100736618, "global_step": 32982, "epoch": 311, "lr": 3.125532078683579e-05} +{"train_loss": 0.01407222542911768, "grad_norm": 0.08859587460756302, "global_step": 32983, "epoch": 311, "lr": 3.1252573206984426e-05} +{"train_loss": 0.015414921566843987, "grad_norm": 0.0757843479514122, "global_step": 32984, "epoch": 311, "lr": 3.124982569300338e-05} +{"train_loss": 0.015556849539279938, "grad_norm": 0.09022361040115356, "global_step": 32985, "epoch": 311, "lr": 3.12470782449023e-05} +{"train_loss": 0.016403645277023315, "grad_norm": 0.07096721976995468, "global_step": 32986, "epoch": 311, "lr": 3.124433086269082e-05} +{"train_loss": 0.015513441525399685, "grad_norm": 0.10006485879421234, "global_step": 32987, "epoch": 311, "lr": 3.12415835463786e-05} +{"train_loss": 0.016080176457762718, "grad_norm": 0.11552536487579346, "global_step": 32988, "epoch": 311, "lr": 3.12388362959753e-05} +{"train_loss": 0.014583956450223923, "grad_norm": 0.05865348130464554, "global_step": 32989, "epoch": 311, "lr": 3.123608911149058e-05} +{"train_loss": 0.0143266711384058, "grad_norm": 0.10627637803554535, "global_step": 32990, "epoch": 311, "lr": 3.123334199293407e-05} +{"train_loss": 0.018201474100351334, "grad_norm": 0.13170383870601654, "global_step": 32991, "epoch": 311, "lr": 3.123059494031544e-05} +{"train_loss": 0.016706176102161407, "grad_norm": 0.09070494771003723, "global_step": 32992, "epoch": 311, "lr": 3.122784795364435e-05} +{"train_loss": 0.013013357296586037, "grad_norm": 0.059985410422086716, "global_step": 32993, "epoch": 311, "lr": 3.122510103293043e-05} +{"train_loss": 0.014491146430373192, "grad_norm": 0.14401282370090485, "global_step": 32994, "epoch": 311, "lr": 3.122235417818333e-05} +{"train_loss": 0.013164836913347244, "grad_norm": 0.0751277282834053, "global_step": 32995, "epoch": 311, "lr": 3.1219607389412705e-05} +{"train_loss": 0.014935300685465336, "grad_norm": 0.05322977155447006, "global_step": 32996, "epoch": 311, "lr": 3.121686066662823e-05} +{"train_loss": 0.01499616727232933, "grad_norm": 0.07869239151477814, "global_step": 32997, "epoch": 311, "lr": 3.1214114009839536e-05} +{"train_loss": 0.01596708409488201, "grad_norm": 0.09053623676300049, "global_step": 32998, "epoch": 311, "lr": 3.121136741905628e-05} +{"train_loss": 0.016334589570760727, "grad_norm": 0.06370741128921509, "global_step": 32999, "epoch": 311, "lr": 3.12086208942881e-05} +{"train_loss": 0.014144149608910084, "grad_norm": 0.1646163910627365, "global_step": 33000, "epoch": 311, "lr": 3.1205874435544654e-05} +{"train_loss": 0.01576520875096321, "grad_norm": 0.08399461954832077, "global_step": 33001, "epoch": 311, "lr": 3.120312804283559e-05} +{"train_loss": 0.014271564781665802, "grad_norm": 0.049166448414325714, "global_step": 33002, "epoch": 311, "lr": 3.120038171617056e-05} +{"train_loss": 0.01745091751217842, "grad_norm": 0.07899317890405655, "global_step": 33003, "epoch": 311, "lr": 3.1197635455559205e-05} +{"train_loss": 0.014720745384693146, "grad_norm": 0.08868823945522308, "global_step": 33004, "epoch": 311, "lr": 3.119488926101121e-05} +{"train_loss": 0.015030302107334137, "grad_norm": 0.17296119034290314, "global_step": 33005, "epoch": 311, "lr": 3.119214313253618e-05} +{"train_loss": 0.014773106202483177, "grad_norm": 0.09171207994222641, "global_step": 33006, "epoch": 311, "lr": 3.1189397070143776e-05} +{"train_loss": 0.01631757616996765, "grad_norm": 0.05811635032296181, "global_step": 33007, "epoch": 311, "lr": 3.1186651073843645e-05} +{"train_loss": 0.012982247397303581, "grad_norm": 0.07949553430080414, "global_step": 33008, "epoch": 311, "lr": 3.118390514364544e-05} +{"train_loss": 0.014524223282933235, "grad_norm": 0.08711192011833191, "global_step": 33009, "epoch": 311, "lr": 3.1181159279558827e-05} +{"train_loss": 0.015344955958425999, "grad_norm": 0.10682043433189392, "global_step": 33010, "epoch": 311, "lr": 3.117841348159343e-05} +{"train_loss": 0.015886183828115463, "grad_norm": 0.10341184586286545, "global_step": 33011, "epoch": 311, "lr": 3.1175667749758894e-05} +{"train_loss": 0.01400543563067913, "grad_norm": 0.11805196106433868, "global_step": 33012, "epoch": 311, "lr": 3.117292208406488e-05} +{"train_loss": 0.016518937423825264, "grad_norm": 0.10934779793024063, "global_step": 33013, "epoch": 311, "lr": 3.117017648452103e-05} +{"train_loss": 0.013692887499928474, "grad_norm": 0.11124707013368607, "global_step": 33014, "epoch": 311, "lr": 3.116743095113699e-05} +{"train_loss": 0.0132401492446661, "grad_norm": 0.07724946737289429, "global_step": 33015, "epoch": 311, "lr": 3.1164685483922393e-05} +{"train_loss": 0.01821506768465042, "grad_norm": 0.11422283947467804, "global_step": 33016, "epoch": 311, "lr": 3.116194008288693e-05} +{"train_loss": 0.015597136691212654, "grad_norm": 0.06965970993041992, "global_step": 33017, "epoch": 311, "lr": 3.11591947480402e-05} +{"train_loss": 0.016751863062381744, "grad_norm": 0.12649524211883545, "global_step": 33018, "epoch": 311, "lr": 3.1156449479391876e-05} +{"train_loss": 0.017674244940280914, "grad_norm": 0.08451595902442932, "global_step": 33019, "epoch": 311, "lr": 3.115370427695158e-05} +{"train_loss": 0.015200421214103699, "grad_norm": 0.06365160644054413, "global_step": 33020, "epoch": 311, "lr": 3.1150959140728976e-05} +{"train_loss": 0.014127700589597225, "grad_norm": 0.15573671460151672, "global_step": 33021, "epoch": 311, "lr": 3.114821407073371e-05} +{"train_loss": 0.016419824212789536, "grad_norm": 0.0619417168200016, "global_step": 33022, "epoch": 311, "lr": 3.114546906697542e-05} +{"train_loss": 0.01449551247060299, "grad_norm": 0.10088436305522919, "global_step": 33023, "epoch": 311, "lr": 3.114272412946375e-05} +{"train_loss": 0.017086710780858994, "grad_norm": 0.08351331204175949, "global_step": 33024, "epoch": 311, "lr": 3.1139979258208344e-05} +{"train_loss": 0.013745691627264023, "grad_norm": 0.057817354798316956, "global_step": 33025, "epoch": 311, "lr": 3.1137234453218846e-05} +{"train_loss": 0.013463318347930908, "grad_norm": 0.11246216297149658, "global_step": 33026, "epoch": 311, "lr": 3.113448971450491e-05} +{"train_loss": 0.0133736003190279, "grad_norm": 0.07066690921783447, "global_step": 33027, "epoch": 311, "lr": 3.1131745042076174e-05} +{"train_loss": 0.014934899285435677, "grad_norm": 0.06839709728956223, "global_step": 33028, "epoch": 311, "lr": 3.112900043594228e-05} +{"train_loss": 0.017103375867009163, "grad_norm": 0.0741162821650505, "global_step": 33029, "epoch": 311, "lr": 3.112625589611286e-05} +{"train_loss": 0.013953158631920815, "grad_norm": 0.06918099522590637, "global_step": 33030, "epoch": 311, "lr": 3.1123511422597576e-05} +{"train_loss": 0.01587209291756153, "grad_norm": 0.07776521891355515, "global_step": 33031, "epoch": 311, "lr": 3.112076701540607e-05} +{"train_loss": 0.015327091328799725, "grad_norm": 0.09438136965036392, "global_step": 33032, "epoch": 311, "lr": 3.1118022674547965e-05} +{"train_loss": 0.01774047128856182, "grad_norm": 0.08507992327213287, "global_step": 33033, "epoch": 311, "lr": 3.111527840003293e-05} +{"train_loss": 0.01340976171195507, "grad_norm": 0.11450425535440445, "global_step": 33034, "epoch": 311, "lr": 3.111253419187059e-05} +{"train_loss": 0.01205663662403822, "grad_norm": 0.06752921640872955, "global_step": 33035, "epoch": 311, "lr": 3.1109790050070595e-05} +{"train_loss": 0.01640264317393303, "grad_norm": 0.10522489994764328, "global_step": 33036, "epoch": 311, "lr": 3.110704597464257e-05} +{"train_loss": 0.015272548422217369, "grad_norm": 0.05856924504041672, "global_step": 33037, "epoch": 311, "lr": 3.110430196559617e-05} +{"train_loss": 0.015329904854297638, "grad_norm": 0.07402241230010986, "global_step": 33038, "epoch": 311, "lr": 3.1101558022941046e-05} +{"train_loss": 0.015160154551267624, "grad_norm": 0.07667675614356995, "global_step": 33039, "epoch": 311, "lr": 3.1098814146686834e-05} +{"train_loss": 0.013958610594272614, "grad_norm": 0.07904266566038132, "global_step": 33040, "epoch": 311, "lr": 3.109607033684316e-05} +{"train_loss": 0.015570368617773056, "grad_norm": 0.14419293403625488, "global_step": 33041, "epoch": 311, "lr": 3.109332659341966e-05} +{"train_loss": 0.01788846217095852, "grad_norm": 0.10911614447832108, "global_step": 33042, "epoch": 311, "lr": 3.1090582916426e-05} +{"train_loss": 0.015771647915244102, "grad_norm": 0.07706676423549652, "global_step": 33043, "epoch": 311, "lr": 3.1087839305871814e-05} +{"train_loss": 0.015185211785137653, "grad_norm": 0.06600473821163177, "global_step": 33044, "epoch": 311, "lr": 3.1085095761766726e-05} +{"train_loss": 0.014470558613538742, "grad_norm": 0.09074994176626205, "global_step": 33045, "epoch": 311, "lr": 3.10823522841204e-05} +{"train_loss": 0.014438388869166374, "grad_norm": 0.14073516428470612, "global_step": 33046, "epoch": 311, "lr": 3.1079608872942453e-05} +{"train_loss": 0.015114576555788517, "grad_norm": 0.11880695074796677, "global_step": 33047, "epoch": 311, "lr": 3.1076865528242534e-05} +{"train_loss": 0.015579462051391602, "grad_norm": 0.1057664230465889, "global_step": 33048, "epoch": 311, "lr": 3.107412225003027e-05} +{"train_loss": 0.015581110492348671, "grad_norm": 0.10702791064977646, "global_step": 33049, "epoch": 311, "lr": 3.107137903831531e-05} +{"train_loss": 0.01516962330788374, "grad_norm": 0.14464963972568512, "global_step": 33050, "epoch": 311, "lr": 3.10686358931073e-05} +{"train_loss": 0.014443157240748405, "grad_norm": 0.06634436547756195, "global_step": 33051, "epoch": 311, "lr": 3.1065892814415885e-05} +{"train_loss": 0.012325158342719078, "grad_norm": 0.18827587366104126, "global_step": 33052, "epoch": 311, "lr": 3.106314980225067e-05} +{"train_loss": 0.014903543516993523, "grad_norm": 0.09789150208234787, "global_step": 33053, "epoch": 311, "lr": 3.106040685662131e-05} +{"train_loss": 0.016253003850579262, "grad_norm": 0.09454437345266342, "global_step": 33054, "epoch": 311, "lr": 3.1057663977537446e-05} +{"train_loss": 0.013086160644888878, "grad_norm": 0.07584735006093979, "global_step": 33055, "epoch": 311, "lr": 3.105492116500872e-05} +{"train_loss": 0.016625192016363144, "grad_norm": 0.10378673672676086, "global_step": 33056, "epoch": 311, "lr": 3.105217841904474e-05} +{"train_loss": 0.014853404834866524, "grad_norm": 0.0658750906586647, "global_step": 33057, "epoch": 311, "lr": 3.104943573965521e-05} +{"train_loss": 0.015110976994037628, "grad_norm": 0.05534985288977623, "global_step": 33058, "epoch": 311, "lr": 3.104669312684968e-05} +{"train_loss": 0.017572084441781044, "grad_norm": 0.06720410287380219, "global_step": 33059, "epoch": 311, "lr": 3.1043950580637846e-05} +{"train_loss": 0.014538897201418877, "grad_norm": 0.05447359010577202, "global_step": 33060, "epoch": 311, "lr": 3.1041208101029305e-05} +{"train_loss": 0.016047146171331406, "grad_norm": 0.08750895410776138, "global_step": 33061, "epoch": 311, "lr": 3.103846568803374e-05} +{"train_loss": 0.013160855509340763, "grad_norm": 0.06504914164543152, "global_step": 33062, "epoch": 311, "lr": 3.103572334166074e-05} +{"train_loss": 0.015187449753284454, "grad_norm": 0.10729718953371048, "global_step": 33063, "epoch": 311, "lr": 3.1032981061919986e-05} +{"train_loss": 0.015177331864833832, "grad_norm": 0.06718473881483078, "global_step": 33064, "epoch": 311, "lr": 3.103023884882107e-05} +{"train_loss": 0.017123473808169365, "grad_norm": 0.10594584047794342, "global_step": 33065, "epoch": 311, "lr": 3.102749670237364e-05} +{"train_loss": 0.01442358922213316, "grad_norm": 0.08031964302062988, "global_step": 33066, "epoch": 311, "lr": 3.102475462258735e-05} +{"train_loss": 0.01430029608309269, "grad_norm": 0.09887395054101944, "global_step": 33067, "epoch": 311, "lr": 3.1022012609471805e-05} +{"train_loss": 0.01502305082976818, "grad_norm": 0.05516408011317253, "global_step": 33068, "epoch": 311, "lr": 3.101927066303666e-05} +{"train_loss": 0.013410758227109909, "grad_norm": 0.06151460111141205, "global_step": 33069, "epoch": 311, "lr": 3.101652878329156e-05} +{"train_loss": 0.015607570298016071, "grad_norm": 0.08244653046131134, "global_step": 33070, "epoch": 311, "lr": 3.1013786970246106e-05} +{"train_loss": 0.015235762313922058, "grad_norm": 0.11635182797908783, "global_step": 33071, "epoch": 311, "lr": 3.101104522390995e-05, "val_loss": 0.020027197897434235} +{"train_loss": 0.015822716057300568, "grad_norm": 0.07148279994726181, "global_step": 33072, "epoch": 312, "lr": 3.100830354429273e-05} +{"train_loss": 0.017948094755411148, "grad_norm": 0.11057163029909134, "global_step": 33073, "epoch": 312, "lr": 3.100556193140405e-05} +{"train_loss": 0.01247387658804655, "grad_norm": 0.06182442605495453, "global_step": 33074, "epoch": 312, "lr": 3.100282038525359e-05} +{"train_loss": 0.017011400312185287, "grad_norm": 0.08896640688180923, "global_step": 33075, "epoch": 312, "lr": 3.100007890585095e-05} +{"train_loss": 0.01732512004673481, "grad_norm": 0.07845842093229294, "global_step": 33076, "epoch": 312, "lr": 3.099733749320577e-05} +{"train_loss": 0.01302484143525362, "grad_norm": 0.05803638696670532, "global_step": 33077, "epoch": 312, "lr": 3.099459614732767e-05} +{"train_loss": 0.013914466835558414, "grad_norm": 0.05950765311717987, "global_step": 33078, "epoch": 312, "lr": 3.09918548682263e-05} +{"train_loss": 0.013388379476964474, "grad_norm": 0.05650518089532852, "global_step": 33079, "epoch": 312, "lr": 3.098911365591129e-05} +{"train_loss": 0.014638692140579224, "grad_norm": 0.06328543275594711, "global_step": 33080, "epoch": 312, "lr": 3.098637251039228e-05} +{"train_loss": 0.01347380317747593, "grad_norm": 0.08866571635007858, "global_step": 33081, "epoch": 312, "lr": 3.0983631431678874e-05} +{"train_loss": 0.01674819365143776, "grad_norm": 0.060335077345371246, "global_step": 33082, "epoch": 312, "lr": 3.098089041978071e-05} +{"train_loss": 0.014525368809700012, "grad_norm": 0.07744517922401428, "global_step": 33083, "epoch": 312, "lr": 3.097814947470743e-05} +{"train_loss": 0.013685306534171104, "grad_norm": 0.0726553276181221, "global_step": 33084, "epoch": 312, "lr": 3.097540859646867e-05} +{"train_loss": 0.01585330441594124, "grad_norm": 0.07613418251276016, "global_step": 33085, "epoch": 312, "lr": 3.097266778507403e-05} +{"train_loss": 0.01648392342031002, "grad_norm": 0.11666978895664215, "global_step": 33086, "epoch": 312, "lr": 3.09699270405332e-05} +{"train_loss": 0.01574867218732834, "grad_norm": 0.06793873012065887, "global_step": 33087, "epoch": 312, "lr": 3.096718636285574e-05} +{"train_loss": 0.016232244670391083, "grad_norm": 0.10238576680421829, "global_step": 33088, "epoch": 312, "lr": 3.0964445752051315e-05} +{"train_loss": 0.014548607170581818, "grad_norm": 0.08009688556194305, "global_step": 33089, "epoch": 312, "lr": 3.096170520812954e-05} +{"train_loss": 0.014422951266169548, "grad_norm": 0.08914534002542496, "global_step": 33090, "epoch": 312, "lr": 3.095896473110006e-05} +{"train_loss": 0.016502968966960907, "grad_norm": 0.08826722949743271, "global_step": 33091, "epoch": 312, "lr": 3.09562243209725e-05} +{"train_loss": 0.014500973746180534, "grad_norm": 0.11112145334482193, "global_step": 33092, "epoch": 312, "lr": 3.0953483977756505e-05} +{"train_loss": 0.017930403351783752, "grad_norm": 0.214854434132576, "global_step": 33093, "epoch": 312, "lr": 3.095074370146166e-05} +{"train_loss": 0.014961903914809227, "grad_norm": 0.1249406710267067, "global_step": 33094, "epoch": 312, "lr": 3.094800349209761e-05} +{"train_loss": 0.013972220942378044, "grad_norm": 0.06773985922336578, "global_step": 33095, "epoch": 312, "lr": 3.0945263349674e-05} +{"train_loss": 0.015847593545913696, "grad_norm": 0.13295379281044006, "global_step": 33096, "epoch": 312, "lr": 3.0942523274200456e-05} +{"train_loss": 0.012573454529047012, "grad_norm": 0.08062022179365158, "global_step": 33097, "epoch": 312, "lr": 3.093978326568659e-05} +{"train_loss": 0.01408761739730835, "grad_norm": 0.08061476051807404, "global_step": 33098, "epoch": 312, "lr": 3.0937043324142056e-05} +{"train_loss": 0.016034234315156937, "grad_norm": 0.07962436228990555, "global_step": 33099, "epoch": 312, "lr": 3.093430344957643e-05} +{"train_loss": 0.013768212869763374, "grad_norm": 0.11124461889266968, "global_step": 33100, "epoch": 312, "lr": 3.093156364199939e-05} +{"train_loss": 0.014980251900851727, "grad_norm": 0.07913127541542053, "global_step": 33101, "epoch": 312, "lr": 3.092882390142054e-05} +{"train_loss": 0.014587018638849258, "grad_norm": 0.14367860555648804, "global_step": 33102, "epoch": 312, "lr": 3.0926084227849496e-05} +{"train_loss": 0.015221867710351944, "grad_norm": 0.14229173958301544, "global_step": 33103, "epoch": 312, "lr": 3.0923344621295905e-05} +{"train_loss": 0.017299484461545944, "grad_norm": 0.06938313692808151, "global_step": 33104, "epoch": 312, "lr": 3.092060508176939e-05} +{"train_loss": 0.014946777373552322, "grad_norm": 0.05531296506524086, "global_step": 33105, "epoch": 312, "lr": 3.091786560927957e-05} +{"train_loss": 0.012171037495136261, "grad_norm": 0.0494597926735878, "global_step": 33106, "epoch": 312, "lr": 3.091512620383606e-05} +{"train_loss": 0.015848323702812195, "grad_norm": 0.1960172951221466, "global_step": 33107, "epoch": 312, "lr": 3.0912386865448515e-05} +{"train_loss": 0.015228117816150188, "grad_norm": 0.08893261849880219, "global_step": 33108, "epoch": 312, "lr": 3.0909647594126536e-05} +{"train_loss": 0.01476337667554617, "grad_norm": 0.05202791467308998, "global_step": 33109, "epoch": 312, "lr": 3.090690838987974e-05} +{"train_loss": 0.014988461509346962, "grad_norm": 0.09075017273426056, "global_step": 33110, "epoch": 312, "lr": 3.09041692527178e-05} +{"train_loss": 0.015918001532554626, "grad_norm": 0.09104370325803757, "global_step": 33111, "epoch": 312, "lr": 3.090143018265026e-05} +{"train_loss": 0.015417087823152542, "grad_norm": 0.061662450432777405, "global_step": 33112, "epoch": 312, "lr": 3.089869117968681e-05} +{"train_loss": 0.013987628743052483, "grad_norm": 0.06253213435411453, "global_step": 33113, "epoch": 312, "lr": 3.0895952243837054e-05} +{"train_loss": 0.015468481928110123, "grad_norm": 0.08861970901489258, "global_step": 33114, "epoch": 312, "lr": 3.0893213375110594e-05} +{"train_loss": 0.015301931649446487, "grad_norm": 0.10759254544973373, "global_step": 33115, "epoch": 312, "lr": 3.089047457351709e-05} +{"train_loss": 0.014867235906422138, "grad_norm": 0.05881636589765549, "global_step": 33116, "epoch": 312, "lr": 3.088773583906617e-05} +{"train_loss": 0.016568513587117195, "grad_norm": 0.08643986284732819, "global_step": 33117, "epoch": 312, "lr": 3.0884997171767404e-05} +{"train_loss": 0.01450599916279316, "grad_norm": 0.07499644160270691, "global_step": 33118, "epoch": 312, "lr": 3.088225857163044e-05} +{"train_loss": 0.016304021701216698, "grad_norm": 0.07100775837898254, "global_step": 33119, "epoch": 312, "lr": 3.087952003866492e-05} +{"train_loss": 0.01533007062971592, "grad_norm": 0.0687311589717865, "global_step": 33120, "epoch": 312, "lr": 3.087678157288045e-05} +{"train_loss": 0.014999985694885254, "grad_norm": 0.0826791301369667, "global_step": 33121, "epoch": 312, "lr": 3.087404317428665e-05} +{"train_loss": 0.01476916391402483, "grad_norm": 0.057475101202726364, "global_step": 33122, "epoch": 312, "lr": 3.087130484289316e-05} +{"train_loss": 0.01281049195677042, "grad_norm": 0.06648880988359451, "global_step": 33123, "epoch": 312, "lr": 3.086856657870957e-05} +{"train_loss": 0.015490957535803318, "grad_norm": 0.09155846387147903, "global_step": 33124, "epoch": 312, "lr": 3.086582838174551e-05} +{"train_loss": 0.014315934851765633, "grad_norm": 0.08813152462244034, "global_step": 33125, "epoch": 312, "lr": 3.0863090252010623e-05} +{"train_loss": 0.015426581725478172, "grad_norm": 0.0630951076745987, "global_step": 33126, "epoch": 312, "lr": 3.08603521895145e-05} +{"train_loss": 0.01538003608584404, "grad_norm": 0.09941408783197403, "global_step": 33127, "epoch": 312, "lr": 3.08576141942668e-05} +{"train_loss": 0.014702267944812775, "grad_norm": 0.08339928090572357, "global_step": 33128, "epoch": 312, "lr": 3.08548762662771e-05} +{"train_loss": 0.014062637463212013, "grad_norm": 0.07916191220283508, "global_step": 33129, "epoch": 312, "lr": 3.0852138405555045e-05} +{"train_loss": 0.017561374232172966, "grad_norm": 0.06977367401123047, "global_step": 33130, "epoch": 312, "lr": 3.0849400612110246e-05} +{"train_loss": 0.013713216409087181, "grad_norm": 0.08919315040111542, "global_step": 33131, "epoch": 312, "lr": 3.0846662885952315e-05} +{"train_loss": 0.015257716178894043, "grad_norm": 0.07874009013175964, "global_step": 33132, "epoch": 312, "lr": 3.0843925227090895e-05} +{"train_loss": 0.01869077980518341, "grad_norm": 0.10118099302053452, "global_step": 33133, "epoch": 312, "lr": 3.084118763553559e-05} +{"train_loss": 0.017545504495501518, "grad_norm": 0.07603172957897186, "global_step": 33134, "epoch": 312, "lr": 3.0838450111296026e-05} +{"train_loss": 0.016421522945165634, "grad_norm": 0.092757448554039, "global_step": 33135, "epoch": 312, "lr": 3.083571265438179e-05} +{"train_loss": 0.015957169234752655, "grad_norm": 0.07491238415241241, "global_step": 33136, "epoch": 312, "lr": 3.083297526480254e-05} +{"train_loss": 0.01562718115746975, "grad_norm": 0.06362355500459671, "global_step": 33137, "epoch": 312, "lr": 3.0830237942567885e-05} +{"train_loss": 0.015876851975917816, "grad_norm": 0.07082231342792511, "global_step": 33138, "epoch": 312, "lr": 3.0827500687687425e-05} +{"train_loss": 0.017916657030582428, "grad_norm": 0.08554348349571228, "global_step": 33139, "epoch": 312, "lr": 3.0824763500170816e-05} +{"train_loss": 0.014519112184643745, "grad_norm": 0.11813639849424362, "global_step": 33140, "epoch": 312, "lr": 3.082202638002762e-05} +{"train_loss": 0.013533275574445724, "grad_norm": 0.07738915085792542, "global_step": 33141, "epoch": 312, "lr": 3.08192893272675e-05} +{"train_loss": 0.014381885528564453, "grad_norm": 0.07282095402479172, "global_step": 33142, "epoch": 312, "lr": 3.0816552341900046e-05} +{"train_loss": 0.014681201428174973, "grad_norm": 0.10003329068422318, "global_step": 33143, "epoch": 312, "lr": 3.0813815423934875e-05} +{"train_loss": 0.014373810961842537, "grad_norm": 0.10592004656791687, "global_step": 33144, "epoch": 312, "lr": 3.081107857338163e-05} +{"train_loss": 0.015582334250211716, "grad_norm": 0.11706768721342087, "global_step": 33145, "epoch": 312, "lr": 3.080834179024992e-05} +{"train_loss": 0.013966697268188, "grad_norm": 0.07014504820108414, "global_step": 33146, "epoch": 312, "lr": 3.080560507454933e-05} +{"train_loss": 0.015627391636371613, "grad_norm": 0.09026424586772919, "global_step": 33147, "epoch": 312, "lr": 3.080286842628949e-05} +{"train_loss": 0.015310870483517647, "grad_norm": 0.08094482868909836, "global_step": 33148, "epoch": 312, "lr": 3.080013184548004e-05} +{"train_loss": 0.016060873866081238, "grad_norm": 0.06476397067308426, "global_step": 33149, "epoch": 312, "lr": 3.079739533213057e-05} +{"train_loss": 0.017207972705364227, "grad_norm": 0.059248704463243484, "global_step": 33150, "epoch": 312, "lr": 3.079465888625069e-05} +{"train_loss": 0.01567639596760273, "grad_norm": 0.07140260934829712, "global_step": 33151, "epoch": 312, "lr": 3.079192250785006e-05} +{"train_loss": 0.016064107418060303, "grad_norm": 0.10270190238952637, "global_step": 33152, "epoch": 312, "lr": 3.078918619693824e-05} +{"train_loss": 0.014428754337131977, "grad_norm": 0.09655993431806564, "global_step": 33153, "epoch": 312, "lr": 3.078644995352486e-05} +{"train_loss": 0.015845011919736862, "grad_norm": 0.08427423983812332, "global_step": 33154, "epoch": 312, "lr": 3.078371377761954e-05} +{"train_loss": 0.016408702358603477, "grad_norm": 0.04567229002714157, "global_step": 33155, "epoch": 312, "lr": 3.0780977669231895e-05} +{"train_loss": 0.012914227321743965, "grad_norm": 0.09510821104049683, "global_step": 33156, "epoch": 312, "lr": 3.077824162837154e-05} +{"train_loss": 0.015563005581498146, "grad_norm": 0.0684850811958313, "global_step": 33157, "epoch": 312, "lr": 3.077550565504809e-05} +{"train_loss": 0.013860109262168407, "grad_norm": 0.060956720262765884, "global_step": 33158, "epoch": 312, "lr": 3.077276974927115e-05} +{"train_loss": 0.014170065522193909, "grad_norm": 0.07250210642814636, "global_step": 33159, "epoch": 312, "lr": 3.077003391105033e-05} +{"train_loss": 0.016394643113017082, "grad_norm": 0.09611720591783524, "global_step": 33160, "epoch": 312, "lr": 3.076729814039524e-05} +{"train_loss": 0.015920182690024376, "grad_norm": 0.06732270121574402, "global_step": 33161, "epoch": 312, "lr": 3.076456243731551e-05} +{"train_loss": 0.01409191731363535, "grad_norm": 0.1386765092611313, "global_step": 33162, "epoch": 312, "lr": 3.076182680182073e-05} +{"train_loss": 0.01615586131811142, "grad_norm": 0.09606021642684937, "global_step": 33163, "epoch": 312, "lr": 3.075909123392056e-05} +{"train_loss": 0.01509978249669075, "grad_norm": 0.158463254570961, "global_step": 33164, "epoch": 312, "lr": 3.0756355733624546e-05} +{"train_loss": 0.014277160167694092, "grad_norm": 0.08399894833564758, "global_step": 33165, "epoch": 312, "lr": 3.0753620300942334e-05} +{"train_loss": 0.015558541752398014, "grad_norm": 0.07848063856363297, "global_step": 33166, "epoch": 312, "lr": 3.0750884935883526e-05} +{"train_loss": 0.014785872772336006, "grad_norm": 0.06981351226568222, "global_step": 33167, "epoch": 312, "lr": 3.074814963845774e-05} +{"train_loss": 0.015364598482847214, "grad_norm": 0.06897679716348648, "global_step": 33168, "epoch": 312, "lr": 3.0745414408674576e-05} +{"train_loss": 0.015158679336309433, "grad_norm": 0.0660921111702919, "global_step": 33169, "epoch": 312, "lr": 3.074267924654367e-05} +{"train_loss": 0.01806548982858658, "grad_norm": 0.14676934480667114, "global_step": 33170, "epoch": 312, "lr": 3.0739944152074615e-05} +{"train_loss": 0.015911582857370377, "grad_norm": 0.07591402530670166, "global_step": 33171, "epoch": 312, "lr": 3.073720912527701e-05} +{"train_loss": 0.014095474034547806, "grad_norm": 0.05734071135520935, "global_step": 33172, "epoch": 312, "lr": 3.0734474166160465e-05} +{"train_loss": 0.017446089535951614, "grad_norm": 0.12100774794816971, "global_step": 33173, "epoch": 312, "lr": 3.073173927473462e-05} +{"train_loss": 0.014086000621318817, "grad_norm": 0.057137392461299896, "global_step": 33174, "epoch": 312, "lr": 3.072900445100905e-05} +{"train_loss": 0.016171924769878387, "grad_norm": 0.09571437537670135, "global_step": 33175, "epoch": 312, "lr": 3.072626969499339e-05} +{"train_loss": 0.013812044635415077, "grad_norm": 0.07686339318752289, "global_step": 33176, "epoch": 312, "lr": 3.0723535006697226e-05} +{"train_loss": 0.015216488900752563, "grad_norm": 0.08654017746448517, "global_step": 33177, "epoch": 312, "lr": 3.072080038613018e-05, "val_loss": 0.01960039883852005} +{"train_loss": 0.01649286597967148, "grad_norm": 0.10032512247562408, "global_step": 33178, "epoch": 313, "lr": 3.0718065833301856e-05} +{"train_loss": 0.014312510378658772, "grad_norm": 0.08852733671665192, "global_step": 33179, "epoch": 313, "lr": 3.071533134822186e-05} +{"train_loss": 0.013892206363379955, "grad_norm": 0.05861404538154602, "global_step": 33180, "epoch": 313, "lr": 3.071259693089982e-05} +{"train_loss": 0.01862485148012638, "grad_norm": 0.12381633371114731, "global_step": 33181, "epoch": 313, "lr": 3.070986258134531e-05} +{"train_loss": 0.014410214498639107, "grad_norm": 0.06333653628826141, "global_step": 33182, "epoch": 313, "lr": 3.070712829956795e-05} +{"train_loss": 0.014366429299116135, "grad_norm": 0.12778744101524353, "global_step": 33183, "epoch": 313, "lr": 3.070439408557736e-05} +{"train_loss": 0.01251313928514719, "grad_norm": 0.0710734874010086, "global_step": 33184, "epoch": 313, "lr": 3.0701659939383126e-05} +{"train_loss": 0.01299634762108326, "grad_norm": 0.13861791789531708, "global_step": 33185, "epoch": 313, "lr": 3.0698925860994876e-05} +{"train_loss": 0.01499495841562748, "grad_norm": 0.05529477819800377, "global_step": 33186, "epoch": 313, "lr": 3.069619185042222e-05} +{"train_loss": 0.015527970157563686, "grad_norm": 0.0595431812107563, "global_step": 33187, "epoch": 313, "lr": 3.0693457907674735e-05} +{"train_loss": 0.01483161561191082, "grad_norm": 0.050737347453832626, "global_step": 33188, "epoch": 313, "lr": 3.0690724032762055e-05} +{"train_loss": 0.013434480875730515, "grad_norm": 0.06846101582050323, "global_step": 33189, "epoch": 313, "lr": 3.0687990225693746e-05} +{"train_loss": 0.014635631814599037, "grad_norm": 0.172554150223732, "global_step": 33190, "epoch": 313, "lr": 3.0685256486479464e-05} +{"train_loss": 0.013996763154864311, "grad_norm": 0.0658220574259758, "global_step": 33191, "epoch": 313, "lr": 3.068252281512877e-05} +{"train_loss": 0.01688181608915329, "grad_norm": 0.0954933613538742, "global_step": 33192, "epoch": 313, "lr": 3.0679789211651335e-05} +{"train_loss": 0.014633629471063614, "grad_norm": 0.08363725990056992, "global_step": 33193, "epoch": 313, "lr": 3.067705567605668e-05} +{"train_loss": 0.017314599826931953, "grad_norm": 0.13456284999847412, "global_step": 33194, "epoch": 313, "lr": 3.067432220835445e-05} +{"train_loss": 0.017344411462545395, "grad_norm": 0.05170860141515732, "global_step": 33195, "epoch": 313, "lr": 3.067158880855425e-05} +{"train_loss": 0.014422420412302017, "grad_norm": 0.0670749768614769, "global_step": 33196, "epoch": 313, "lr": 3.066885547666568e-05} +{"train_loss": 0.01474246010184288, "grad_norm": 0.09038372337818146, "global_step": 33197, "epoch": 313, "lr": 3.0666122212698357e-05} +{"train_loss": 0.016505151987075806, "grad_norm": 0.17564757168293, "global_step": 33198, "epoch": 313, "lr": 3.066338901666187e-05} +{"train_loss": 0.016266319900751114, "grad_norm": 0.10858581960201263, "global_step": 33199, "epoch": 313, "lr": 3.0660655888565825e-05} +{"train_loss": 0.013347888365387917, "grad_norm": 0.06519310921430588, "global_step": 33200, "epoch": 313, "lr": 3.065792282841982e-05} +{"train_loss": 0.016955919563770294, "grad_norm": 0.20952481031417847, "global_step": 33201, "epoch": 313, "lr": 3.0655189836233454e-05} +{"train_loss": 0.016326816752552986, "grad_norm": 0.07678969204425812, "global_step": 33202, "epoch": 313, "lr": 3.065245691201635e-05} +{"train_loss": 0.0156521275639534, "grad_norm": 0.14166474342346191, "global_step": 33203, "epoch": 313, "lr": 3.064972405577809e-05} +{"train_loss": 0.01648804359138012, "grad_norm": 0.06508252024650574, "global_step": 33204, "epoch": 313, "lr": 3.0646991267528304e-05} +{"train_loss": 0.01577468402683735, "grad_norm": 0.1970585286617279, "global_step": 33205, "epoch": 313, "lr": 3.064425854727655e-05} +{"train_loss": 0.01589464768767357, "grad_norm": 0.10256637632846832, "global_step": 33206, "epoch": 313, "lr": 3.0641525895032464e-05} +{"train_loss": 0.012629227712750435, "grad_norm": 0.08424834907054901, "global_step": 33207, "epoch": 313, "lr": 3.063879331080564e-05} +{"train_loss": 0.014335069805383682, "grad_norm": 0.07678165286779404, "global_step": 33208, "epoch": 313, "lr": 3.063606079460567e-05} +{"train_loss": 0.017625553533434868, "grad_norm": 0.08862198889255524, "global_step": 33209, "epoch": 313, "lr": 3.063332834644216e-05} +{"train_loss": 0.016498614102602005, "grad_norm": 0.06897298991680145, "global_step": 33210, "epoch": 313, "lr": 3.063059596632473e-05} +{"train_loss": 0.01514686830341816, "grad_norm": 0.18565431237220764, "global_step": 33211, "epoch": 313, "lr": 3.0627863654262944e-05} +{"train_loss": 0.01321520283818245, "grad_norm": 0.09855687618255615, "global_step": 33212, "epoch": 313, "lr": 3.062513141026643e-05} +{"train_loss": 0.01233438216149807, "grad_norm": 0.07077258080244064, "global_step": 33213, "epoch": 313, "lr": 3.062239923434476e-05} +{"train_loss": 0.015365403145551682, "grad_norm": 0.0651623085141182, "global_step": 33214, "epoch": 313, "lr": 3.0619667126507566e-05} +{"train_loss": 0.01279064454138279, "grad_norm": 0.12564067542552948, "global_step": 33215, "epoch": 313, "lr": 3.061693508676443e-05} +{"train_loss": 0.013592002913355827, "grad_norm": 0.08673177659511566, "global_step": 33216, "epoch": 313, "lr": 3.0614203115124975e-05} +{"train_loss": 0.019190587103366852, "grad_norm": 0.17957542836666107, "global_step": 33217, "epoch": 313, "lr": 3.0611471211598755e-05} +{"train_loss": 0.01666496880352497, "grad_norm": 0.09631568193435669, "global_step": 33218, "epoch": 313, "lr": 3.0608739376195396e-05} +{"train_loss": 0.018554694950580597, "grad_norm": 0.07646166533231735, "global_step": 33219, "epoch": 313, "lr": 3.06060076089245e-05} +{"train_loss": 0.017321903258562088, "grad_norm": 0.06866001337766647, "global_step": 33220, "epoch": 313, "lr": 3.0603275909795646e-05} +{"train_loss": 0.015266773290932178, "grad_norm": 0.06811078637838364, "global_step": 33221, "epoch": 313, "lr": 3.060054427881846e-05} +{"train_loss": 0.017447352409362793, "grad_norm": 0.08997078984975815, "global_step": 33222, "epoch": 313, "lr": 3.059781271600253e-05} +{"train_loss": 0.014366816729307175, "grad_norm": 0.05809260532259941, "global_step": 33223, "epoch": 313, "lr": 3.059508122135744e-05} +{"train_loss": 0.013532772660255432, "grad_norm": 0.06868259608745575, "global_step": 33224, "epoch": 313, "lr": 3.059234979489279e-05} +{"train_loss": 0.015492025762796402, "grad_norm": 0.061309926211833954, "global_step": 33225, "epoch": 313, "lr": 3.058961843661818e-05} +{"train_loss": 0.01445542462170124, "grad_norm": 0.09505193680524826, "global_step": 33226, "epoch": 313, "lr": 3.058688714654323e-05} +{"train_loss": 0.01898476853966713, "grad_norm": 0.06695294380187988, "global_step": 33227, "epoch": 313, "lr": 3.0584155924677504e-05} +{"train_loss": 0.013873206451535225, "grad_norm": 0.07279413193464279, "global_step": 33228, "epoch": 313, "lr": 3.0581424771030616e-05} +{"train_loss": 0.014313914813101292, "grad_norm": 0.08406130969524384, "global_step": 33229, "epoch": 313, "lr": 3.057869368561215e-05} +{"train_loss": 0.01802096888422966, "grad_norm": 0.07380163669586182, "global_step": 33230, "epoch": 313, "lr": 3.05759626684317e-05} +{"train_loss": 0.015565725974738598, "grad_norm": 0.050871044397354126, "global_step": 33231, "epoch": 313, "lr": 3.0573231719498875e-05} +{"train_loss": 0.01410096138715744, "grad_norm": 0.09311264753341675, "global_step": 33232, "epoch": 313, "lr": 3.057050083882327e-05} +{"train_loss": 0.014707296155393124, "grad_norm": 0.06978034228086472, "global_step": 33233, "epoch": 313, "lr": 3.0567770026414486e-05} +{"train_loss": 0.015308738686144352, "grad_norm": 0.08307290822267532, "global_step": 33234, "epoch": 313, "lr": 3.056503928228208e-05} +{"train_loss": 0.014730552211403847, "grad_norm": 0.06438178569078445, "global_step": 33235, "epoch": 313, "lr": 3.05623086064357e-05} +{"train_loss": 0.0126138711348176, "grad_norm": 0.05381127446889877, "global_step": 33236, "epoch": 313, "lr": 3.05595779988849e-05} +{"train_loss": 0.015234456397593021, "grad_norm": 0.10712979733943939, "global_step": 33237, "epoch": 313, "lr": 3.0556847459639285e-05} +{"train_loss": 0.017907045781612396, "grad_norm": 0.11204753816127777, "global_step": 33238, "epoch": 313, "lr": 3.055411698870846e-05} +{"train_loss": 0.014355534687638283, "grad_norm": 0.050590842962265015, "global_step": 33239, "epoch": 313, "lr": 3.055138658610202e-05} +{"train_loss": 0.013241669163107872, "grad_norm": 0.06357662379741669, "global_step": 33240, "epoch": 313, "lr": 3.054865625182954e-05} +{"train_loss": 0.01481076329946518, "grad_norm": 0.0998181477189064, "global_step": 33241, "epoch": 313, "lr": 3.054592598590063e-05} +{"train_loss": 0.017610736191272736, "grad_norm": 0.09744385629892349, "global_step": 33242, "epoch": 313, "lr": 3.054319578832485e-05} +{"train_loss": 0.014694491401314735, "grad_norm": 0.056555066257715225, "global_step": 33243, "epoch": 313, "lr": 3.054046565911184e-05} +{"train_loss": 0.015509710647165775, "grad_norm": 0.13115669786930084, "global_step": 33244, "epoch": 313, "lr": 3.053773559827117e-05} +{"train_loss": 0.013691408559679985, "grad_norm": 0.06117422506213188, "global_step": 33245, "epoch": 313, "lr": 3.0535005605812437e-05} +{"train_loss": 0.01727854460477829, "grad_norm": 0.13209138810634613, "global_step": 33246, "epoch": 313, "lr": 3.053227568174521e-05} +{"train_loss": 0.016225356608629227, "grad_norm": 0.08710476011037827, "global_step": 33247, "epoch": 313, "lr": 3.052954582607912e-05} +{"train_loss": 0.015083536505699158, "grad_norm": 0.0746326893568039, "global_step": 33248, "epoch": 313, "lr": 3.052681603882373e-05} +{"train_loss": 0.013048353604972363, "grad_norm": 0.08850692212581635, "global_step": 33249, "epoch": 313, "lr": 3.0524086319988634e-05} +{"train_loss": 0.016325490549206734, "grad_norm": 0.11392752081155777, "global_step": 33250, "epoch": 313, "lr": 3.0521356669583436e-05} +{"train_loss": 0.013033121824264526, "grad_norm": 0.06075949966907501, "global_step": 33251, "epoch": 313, "lr": 3.0518627087617735e-05} +{"train_loss": 0.016199298202991486, "grad_norm": 0.084686778485775, "global_step": 33252, "epoch": 313, "lr": 3.051589757410109e-05} +{"train_loss": 0.013084694743156433, "grad_norm": 0.09994646906852722, "global_step": 33253, "epoch": 313, "lr": 3.0513168129043108e-05} +{"train_loss": 0.018318191170692444, "grad_norm": 0.10462398082017899, "global_step": 33254, "epoch": 313, "lr": 3.0510438752453373e-05} +{"train_loss": 0.012655176222324371, "grad_norm": 0.05803120508790016, "global_step": 33255, "epoch": 313, "lr": 3.050770944434148e-05} +{"train_loss": 0.0155873978510499, "grad_norm": 0.12100912630558014, "global_step": 33256, "epoch": 313, "lr": 3.050498020471703e-05} +{"train_loss": 0.014238197356462479, "grad_norm": 0.0701175183057785, "global_step": 33257, "epoch": 313, "lr": 3.0502251033589614e-05} +{"train_loss": 0.017221111804246902, "grad_norm": 0.07964211702346802, "global_step": 33258, "epoch": 313, "lr": 3.049952193096879e-05} +{"train_loss": 0.01476273313164711, "grad_norm": 0.08440759032964706, "global_step": 33259, "epoch": 313, "lr": 3.0496792896864157e-05} +{"train_loss": 0.01551355142146349, "grad_norm": 0.08441288024187088, "global_step": 33260, "epoch": 313, "lr": 3.049406393128532e-05} +{"train_loss": 0.01650245115160942, "grad_norm": 0.0815618708729744, "global_step": 33261, "epoch": 313, "lr": 3.0491335034241862e-05} +{"train_loss": 0.014119340106844902, "grad_norm": 0.10920795798301697, "global_step": 33262, "epoch": 313, "lr": 3.048860620574336e-05} +{"train_loss": 0.016199547797441483, "grad_norm": 0.07020934671163559, "global_step": 33263, "epoch": 313, "lr": 3.0485877445799426e-05} +{"train_loss": 0.015973366796970367, "grad_norm": 0.06241936981678009, "global_step": 33264, "epoch": 313, "lr": 3.0483148754419633e-05} +{"train_loss": 0.013494694605469704, "grad_norm": 0.08953338861465454, "global_step": 33265, "epoch": 313, "lr": 3.0480420131613553e-05} +{"train_loss": 0.012907035648822784, "grad_norm": 0.07195830345153809, "global_step": 33266, "epoch": 313, "lr": 3.047769157739079e-05} +{"train_loss": 0.01299629919230938, "grad_norm": 0.054551392793655396, "global_step": 33267, "epoch": 313, "lr": 3.0474963091760933e-05} +{"train_loss": 0.015554845333099365, "grad_norm": 0.1744460165500641, "global_step": 33268, "epoch": 313, "lr": 3.0472234674733563e-05} +{"train_loss": 0.014346688985824585, "grad_norm": 0.05781034007668495, "global_step": 33269, "epoch": 313, "lr": 3.046950632631828e-05} +{"train_loss": 0.016055092215538025, "grad_norm": 0.09683553874492645, "global_step": 33270, "epoch": 313, "lr": 3.0466778046524648e-05} +{"train_loss": 0.015246730297803879, "grad_norm": 0.06941530108451843, "global_step": 33271, "epoch": 313, "lr": 3.0464049835362252e-05} +{"train_loss": 0.016690146178007126, "grad_norm": 0.08523572236299515, "global_step": 33272, "epoch": 313, "lr": 3.0461321692840706e-05} +{"train_loss": 0.014085923321545124, "grad_norm": 0.07763240486383438, "global_step": 33273, "epoch": 313, "lr": 3.0458593618969568e-05} +{"train_loss": 0.014560015872120857, "grad_norm": 0.07773212343454361, "global_step": 33274, "epoch": 313, "lr": 3.045586561375843e-05} +{"train_loss": 0.0151798902079463, "grad_norm": 0.0585334412753582, "global_step": 33275, "epoch": 313, "lr": 3.04531376772169e-05} +{"train_loss": 0.013925913721323013, "grad_norm": 0.11992071568965912, "global_step": 33276, "epoch": 313, "lr": 3.045040980935453e-05} +{"train_loss": 0.015667112544178963, "grad_norm": 0.10123707354068756, "global_step": 33277, "epoch": 313, "lr": 3.0447682010180923e-05} +{"train_loss": 0.01608114503324032, "grad_norm": 0.09754934161901474, "global_step": 33278, "epoch": 313, "lr": 3.0444954279705646e-05} +{"train_loss": 0.016630399972200394, "grad_norm": 0.15187953412532806, "global_step": 33279, "epoch": 313, "lr": 3.0442226617938307e-05} +{"train_loss": 0.015030216425657272, "grad_norm": 0.12140240520238876, "global_step": 33280, "epoch": 313, "lr": 3.0439499024888484e-05} +{"train_loss": 0.016285644844174385, "grad_norm": 0.06939470022916794, "global_step": 33281, "epoch": 313, "lr": 3.0436771500565752e-05} +{"train_loss": 0.016136737540364265, "grad_norm": 0.09599461406469345, "global_step": 33282, "epoch": 313, "lr": 3.0434044044979693e-05} +{"train_loss": 0.015262952575214067, "grad_norm": 0.15798434615135193, "global_step": 33283, "epoch": 313, "lr": 3.043131665813988e-05, "val_loss": 0.018869809806346893} +{"train_loss": 0.016346525400877, "grad_norm": 0.16428866982460022, "global_step": 33284, "epoch": 314, "lr": 3.0428589340055924e-05} +{"train_loss": 0.01783851906657219, "grad_norm": 0.0829542949795723, "global_step": 33285, "epoch": 314, "lr": 3.0425862090737394e-05} +{"train_loss": 0.01642073690891266, "grad_norm": 0.0982801541686058, "global_step": 33286, "epoch": 314, "lr": 3.0423134910193883e-05} +{"train_loss": 0.012585777789354324, "grad_norm": 0.07040286809206009, "global_step": 33287, "epoch": 314, "lr": 3.0420407798434947e-05} +{"train_loss": 0.013753462582826614, "grad_norm": 0.08090613782405853, "global_step": 33288, "epoch": 314, "lr": 3.041768075547018e-05} +{"train_loss": 0.013115493580698967, "grad_norm": 0.07378455251455307, "global_step": 33289, "epoch": 314, "lr": 3.0414953781309175e-05} +{"train_loss": 0.014289116486907005, "grad_norm": 0.1330394297838211, "global_step": 33290, "epoch": 314, "lr": 3.0412226875961502e-05} +{"train_loss": 0.016077477484941483, "grad_norm": 0.07352104038000107, "global_step": 33291, "epoch": 314, "lr": 3.0409500039436733e-05} +{"train_loss": 0.015093290247023106, "grad_norm": 0.07680940628051758, "global_step": 33292, "epoch": 314, "lr": 3.0406773271744488e-05} +{"train_loss": 0.015208989381790161, "grad_norm": 0.07667466998100281, "global_step": 33293, "epoch": 314, "lr": 3.0404046572894306e-05} +{"train_loss": 0.018402935937047005, "grad_norm": 0.07320345193147659, "global_step": 33294, "epoch": 314, "lr": 3.040131994289579e-05} +{"train_loss": 0.014637507498264313, "grad_norm": 0.06821432709693909, "global_step": 33295, "epoch": 314, "lr": 3.0398593381758493e-05} +{"train_loss": 0.014415302313864231, "grad_norm": 0.08069243282079697, "global_step": 33296, "epoch": 314, "lr": 3.039586688949203e-05} +{"train_loss": 0.019146066159009933, "grad_norm": 0.22271643579006195, "global_step": 33297, "epoch": 314, "lr": 3.0393140466105963e-05} +{"train_loss": 0.01613728702068329, "grad_norm": 0.13380718231201172, "global_step": 33298, "epoch": 314, "lr": 3.0390414111609884e-05} +{"train_loss": 0.01616688072681427, "grad_norm": 0.16778913140296936, "global_step": 33299, "epoch": 314, "lr": 3.038768782601335e-05} +{"train_loss": 0.014381920918822289, "grad_norm": 0.06491805613040924, "global_step": 33300, "epoch": 314, "lr": 3.0384961609325945e-05} +{"train_loss": 0.018259655684232712, "grad_norm": 0.16287066042423248, "global_step": 33301, "epoch": 314, "lr": 3.0382235461557263e-05} +{"train_loss": 0.015959065407514572, "grad_norm": 0.14903482794761658, "global_step": 33302, "epoch": 314, "lr": 3.0379509382716876e-05} +{"train_loss": 0.01380431279540062, "grad_norm": 0.06810624152421951, "global_step": 33303, "epoch": 314, "lr": 3.0376783372814342e-05} +{"train_loss": 0.01636650040745735, "grad_norm": 0.07664691656827927, "global_step": 33304, "epoch": 314, "lr": 3.0374057431859288e-05} +{"train_loss": 0.01569969579577446, "grad_norm": 0.08357331156730652, "global_step": 33305, "epoch": 314, "lr": 3.037133155986125e-05} +{"train_loss": 0.014738869853317738, "grad_norm": 0.07421524077653885, "global_step": 33306, "epoch": 314, "lr": 3.036860575682981e-05} +{"train_loss": 0.01421812642365694, "grad_norm": 0.08844751864671707, "global_step": 33307, "epoch": 314, "lr": 3.0365880022774546e-05} +{"train_loss": 0.019191015511751175, "grad_norm": 0.10491827130317688, "global_step": 33308, "epoch": 314, "lr": 3.0363154357705048e-05} +{"train_loss": 0.01367952674627304, "grad_norm": 0.06522247940301895, "global_step": 33309, "epoch": 314, "lr": 3.036042876163089e-05} +{"train_loss": 0.015787100419402122, "grad_norm": 0.07536174356937408, "global_step": 33310, "epoch": 314, "lr": 3.0357703234561652e-05} +{"train_loss": 0.015373770147562027, "grad_norm": 0.07238034904003143, "global_step": 33311, "epoch": 314, "lr": 3.0354977776506894e-05} +{"train_loss": 0.01387803815305233, "grad_norm": 0.06893174350261688, "global_step": 33312, "epoch": 314, "lr": 3.0352252387476187e-05} +{"train_loss": 0.012508789077401161, "grad_norm": 0.05586306378245354, "global_step": 33313, "epoch": 314, "lr": 3.0349527067479132e-05} +{"train_loss": 0.01652287319302559, "grad_norm": 0.13168756663799286, "global_step": 33314, "epoch": 314, "lr": 3.0346801816525295e-05} +{"train_loss": 0.014702297747135162, "grad_norm": 0.07102755457162857, "global_step": 33315, "epoch": 314, "lr": 3.034407663462424e-05} +{"train_loss": 0.014234292320907116, "grad_norm": 0.09446985274553299, "global_step": 33316, "epoch": 314, "lr": 3.0341351521785584e-05} +{"train_loss": 0.013916827738285065, "grad_norm": 0.05639719218015671, "global_step": 33317, "epoch": 314, "lr": 3.0338626478018835e-05} +{"train_loss": 0.01602301187813282, "grad_norm": 0.08939380198717117, "global_step": 33318, "epoch": 314, "lr": 3.033590150333361e-05} +{"train_loss": 0.016578156501054764, "grad_norm": 0.10224666446447372, "global_step": 33319, "epoch": 314, "lr": 3.0333176597739476e-05} +{"train_loss": 0.013149072416126728, "grad_norm": 0.06641875207424164, "global_step": 33320, "epoch": 314, "lr": 3.0330451761246002e-05} +{"train_loss": 0.014855382032692432, "grad_norm": 0.10265609622001648, "global_step": 33321, "epoch": 314, "lr": 3.032772699386277e-05} +{"train_loss": 0.014102935791015625, "grad_norm": 0.05935041606426239, "global_step": 33322, "epoch": 314, "lr": 3.032500229559937e-05} +{"train_loss": 0.013551251962780952, "grad_norm": 0.08233592659235, "global_step": 33323, "epoch": 314, "lr": 3.0322277666465336e-05} +{"train_loss": 0.014772389084100723, "grad_norm": 0.051087573170661926, "global_step": 33324, "epoch": 314, "lr": 3.0319553106470254e-05} +{"train_loss": 0.013065943494439125, "grad_norm": 0.09421395510435104, "global_step": 33325, "epoch": 314, "lr": 3.0316828615623716e-05} +{"train_loss": 0.015165816061198711, "grad_norm": 0.13639341294765472, "global_step": 33326, "epoch": 314, "lr": 3.0314104193935277e-05} +{"train_loss": 0.014340249821543694, "grad_norm": 0.06342867016792297, "global_step": 33327, "epoch": 314, "lr": 3.031137984141453e-05} +{"train_loss": 0.015314402990043163, "grad_norm": 0.06146029755473137, "global_step": 33328, "epoch": 314, "lr": 3.030865555807102e-05} +{"train_loss": 0.014231903478503227, "grad_norm": 0.07356183230876923, "global_step": 33329, "epoch": 314, "lr": 3.030593134391432e-05} +{"train_loss": 0.013257743790745735, "grad_norm": 0.1254851520061493, "global_step": 33330, "epoch": 314, "lr": 3.030320719895402e-05} +{"train_loss": 0.014211912639439106, "grad_norm": 0.06295780837535858, "global_step": 33331, "epoch": 314, "lr": 3.0300483123199684e-05} +{"train_loss": 0.014195689000189304, "grad_norm": 0.06265460699796677, "global_step": 33332, "epoch": 314, "lr": 3.0297759116660873e-05} +{"train_loss": 0.017663592472672462, "grad_norm": 0.1146862581372261, "global_step": 33333, "epoch": 314, "lr": 3.0295035179347198e-05} +{"train_loss": 0.01864337921142578, "grad_norm": 0.08631747961044312, "global_step": 33334, "epoch": 314, "lr": 3.029231131126818e-05} +{"train_loss": 0.016529837623238564, "grad_norm": 0.10134865343570709, "global_step": 33335, "epoch": 314, "lr": 3.0289587512433408e-05} +{"train_loss": 0.014899026602506638, "grad_norm": 0.09071790426969528, "global_step": 33336, "epoch": 314, "lr": 3.028686378285245e-05} +{"train_loss": 0.01399470865726471, "grad_norm": 0.06554295122623444, "global_step": 33337, "epoch": 314, "lr": 3.0284140122534886e-05} +{"train_loss": 0.01442909799516201, "grad_norm": 0.10993409901857376, "global_step": 33338, "epoch": 314, "lr": 3.0281416531490282e-05} +{"train_loss": 0.014726654626429081, "grad_norm": 0.07380712032318115, "global_step": 33339, "epoch": 314, "lr": 3.0278693009728214e-05} +{"train_loss": 0.01528872735798359, "grad_norm": 0.06701724976301193, "global_step": 33340, "epoch": 314, "lr": 3.0275969557258234e-05} +{"train_loss": 0.013963676989078522, "grad_norm": 0.07083345204591751, "global_step": 33341, "epoch": 314, "lr": 3.027324617408991e-05} +{"train_loss": 0.01748701184988022, "grad_norm": 0.08255118876695633, "global_step": 33342, "epoch": 314, "lr": 3.027052286023282e-05} +{"train_loss": 0.015328437089920044, "grad_norm": 0.11868125200271606, "global_step": 33343, "epoch": 314, "lr": 3.0267799615696546e-05} +{"train_loss": 0.01409545075148344, "grad_norm": 0.06397038698196411, "global_step": 33344, "epoch": 314, "lr": 3.0265076440490625e-05} +{"train_loss": 0.014538906514644623, "grad_norm": 0.16748353838920593, "global_step": 33345, "epoch": 314, "lr": 3.026235333462467e-05} +{"train_loss": 0.014771739020943642, "grad_norm": 0.12992249429225922, "global_step": 33346, "epoch": 314, "lr": 3.02596302981082e-05} +{"train_loss": 0.016611555591225624, "grad_norm": 0.11436320096254349, "global_step": 33347, "epoch": 314, "lr": 3.0256907330950813e-05} +{"train_loss": 0.013130756095051765, "grad_norm": 0.051984984427690506, "global_step": 33348, "epoch": 314, "lr": 3.0254184433162046e-05} +{"train_loss": 0.013955741189420223, "grad_norm": 0.10663740336894989, "global_step": 33349, "epoch": 314, "lr": 3.025146160475151e-05} +{"train_loss": 0.012943021953105927, "grad_norm": 0.061101581901311874, "global_step": 33350, "epoch": 314, "lr": 3.0248738845728748e-05} +{"train_loss": 0.015258681029081345, "grad_norm": 0.07525381445884705, "global_step": 33351, "epoch": 314, "lr": 3.0246016156103336e-05} +{"train_loss": 0.012661993503570557, "grad_norm": 0.06267230957746506, "global_step": 33352, "epoch": 314, "lr": 3.024329353588482e-05} +{"train_loss": 0.016174258664250374, "grad_norm": 0.08548238128423691, "global_step": 33353, "epoch": 314, "lr": 3.024057098508277e-05} +{"train_loss": 0.018112394958734512, "grad_norm": 0.0828811526298523, "global_step": 33354, "epoch": 314, "lr": 3.023784850370678e-05} +{"train_loss": 0.015391053631901741, "grad_norm": 0.07039335370063782, "global_step": 33355, "epoch": 314, "lr": 3.0235126091766386e-05} +{"train_loss": 0.014331764541566372, "grad_norm": 0.09640444815158844, "global_step": 33356, "epoch": 314, "lr": 3.0232403749271156e-05} +{"train_loss": 0.01666877791285515, "grad_norm": 0.06844811141490936, "global_step": 33357, "epoch": 314, "lr": 3.0229681476230686e-05} +{"train_loss": 0.01656930148601532, "grad_norm": 0.07210055738687515, "global_step": 33358, "epoch": 314, "lr": 3.0226959272654497e-05} +{"train_loss": 0.016774458810687065, "grad_norm": 0.06390690803527832, "global_step": 33359, "epoch": 314, "lr": 3.022423713855218e-05} +{"train_loss": 0.01737292855978012, "grad_norm": 0.06237782537937164, "global_step": 33360, "epoch": 314, "lr": 3.0221515073933294e-05} +{"train_loss": 0.015607360750436783, "grad_norm": 0.10742251574993134, "global_step": 33361, "epoch": 314, "lr": 3.021879307880739e-05} +{"train_loss": 0.015043509192764759, "grad_norm": 0.07484716922044754, "global_step": 33362, "epoch": 314, "lr": 3.0216071153184056e-05} +{"train_loss": 0.014733152464032173, "grad_norm": 0.10872015357017517, "global_step": 33363, "epoch": 314, "lr": 3.0213349297072857e-05} +{"train_loss": 0.015537221916019917, "grad_norm": 0.06722746789455414, "global_step": 33364, "epoch": 314, "lr": 3.021062751048333e-05} +{"train_loss": 0.014762045815587044, "grad_norm": 0.06886659562587738, "global_step": 33365, "epoch": 314, "lr": 3.020790579342504e-05} +{"train_loss": 0.015966644510626793, "grad_norm": 0.10090729594230652, "global_step": 33366, "epoch": 314, "lr": 3.0205184145907578e-05} +{"train_loss": 0.014258868992328644, "grad_norm": 0.07433152198791504, "global_step": 33367, "epoch": 314, "lr": 3.0202462567940487e-05} +{"train_loss": 0.015572292730212212, "grad_norm": 0.10224439203739166, "global_step": 33368, "epoch": 314, "lr": 3.0199741059533325e-05} +{"train_loss": 0.01642334833741188, "grad_norm": 0.10386994481086731, "global_step": 33369, "epoch": 314, "lr": 3.019701962069569e-05} +{"train_loss": 0.015665099024772644, "grad_norm": 0.07355256378650665, "global_step": 33370, "epoch": 314, "lr": 3.019429825143708e-05} +{"train_loss": 0.014209107495844364, "grad_norm": 0.07775034755468369, "global_step": 33371, "epoch": 314, "lr": 3.019157695176711e-05} +{"train_loss": 0.018785260617733, "grad_norm": 0.07169468700885773, "global_step": 33372, "epoch": 314, "lr": 3.0188855721695326e-05} +{"train_loss": 0.014529045671224594, "grad_norm": 0.05653240159153938, "global_step": 33373, "epoch": 314, "lr": 3.0186134561231273e-05} +{"train_loss": 0.013962533324956894, "grad_norm": 0.06416808068752289, "global_step": 33374, "epoch": 314, "lr": 3.0183413470384537e-05} +{"train_loss": 0.014658275991678238, "grad_norm": 0.11818471550941467, "global_step": 33375, "epoch": 314, "lr": 3.0180692449164684e-05} +{"train_loss": 0.013981511816382408, "grad_norm": 0.06314332783222198, "global_step": 33376, "epoch": 314, "lr": 3.0177971497581237e-05} +{"train_loss": 0.01618715561926365, "grad_norm": 0.12772810459136963, "global_step": 33377, "epoch": 314, "lr": 3.0175250615643775e-05} +{"train_loss": 0.015905942767858505, "grad_norm": 0.11014330387115479, "global_step": 33378, "epoch": 314, "lr": 3.0172529803361872e-05} +{"train_loss": 0.015745237469673157, "grad_norm": 0.11645624041557312, "global_step": 33379, "epoch": 314, "lr": 3.0169809060745068e-05} +{"train_loss": 0.013646159321069717, "grad_norm": 0.11558659374713898, "global_step": 33380, "epoch": 314, "lr": 3.0167088387802954e-05} +{"train_loss": 0.012628752738237381, "grad_norm": 0.05904276296496391, "global_step": 33381, "epoch": 314, "lr": 3.0164367784545045e-05} +{"train_loss": 0.014667957089841366, "grad_norm": 0.06666729599237442, "global_step": 33382, "epoch": 314, "lr": 3.016164725098092e-05} +{"train_loss": 0.013361457735300064, "grad_norm": 0.059349559247493744, "global_step": 33383, "epoch": 314, "lr": 3.0158926787120147e-05} +{"train_loss": 0.015260126441717148, "grad_norm": 0.06564806401729584, "global_step": 33384, "epoch": 314, "lr": 3.015620639297227e-05} +{"train_loss": 0.018141373991966248, "grad_norm": 0.095962293446064, "global_step": 33385, "epoch": 314, "lr": 3.015348606854685e-05} +{"train_loss": 0.01583835296332836, "grad_norm": 0.07413947582244873, "global_step": 33386, "epoch": 314, "lr": 3.0150765813853478e-05} +{"train_loss": 0.01308734156191349, "grad_norm": 0.10581979155540466, "global_step": 33387, "epoch": 314, "lr": 3.014804562890165e-05} +{"train_loss": 0.015363751910626888, "grad_norm": 0.10492023080587387, "global_step": 33388, "epoch": 314, "lr": 3.014532551370097e-05} +{"train_loss": 0.015229625767975484, "grad_norm": 0.11078017204999924, "global_step": 33389, "epoch": 314, "lr": 3.0142605468260978e-05, "val_loss": 0.01941358670592308} +{"train_loss": 0.013026423752307892, "grad_norm": 0.09797875583171844, "global_step": 33390, "epoch": 315, "lr": 3.0139885492591226e-05} +{"train_loss": 0.015364790335297585, "grad_norm": 0.053690824657678604, "global_step": 33391, "epoch": 315, "lr": 3.013716558670129e-05} +{"train_loss": 0.017274482175707817, "grad_norm": 0.0922258049249649, "global_step": 33392, "epoch": 315, "lr": 3.0134445750600727e-05} +{"train_loss": 0.013500581495463848, "grad_norm": 0.06967461109161377, "global_step": 33393, "epoch": 315, "lr": 3.0131725984299075e-05} +{"train_loss": 0.01493452861905098, "grad_norm": 0.07061479985713959, "global_step": 33394, "epoch": 315, "lr": 3.0129006287805884e-05} +{"train_loss": 0.01554781012237072, "grad_norm": 0.08248800784349442, "global_step": 33395, "epoch": 315, "lr": 3.012628666113073e-05} +{"train_loss": 0.020102741196751595, "grad_norm": 0.10118231922388077, "global_step": 33396, "epoch": 315, "lr": 3.0123567104283166e-05} +{"train_loss": 0.016033533960580826, "grad_norm": 0.06905637681484222, "global_step": 33397, "epoch": 315, "lr": 3.012084761727273e-05} +{"train_loss": 0.014954823069274426, "grad_norm": 0.06860294938087463, "global_step": 33398, "epoch": 315, "lr": 3.011812820010902e-05} +{"train_loss": 0.01796737313270569, "grad_norm": 0.09018759429454803, "global_step": 33399, "epoch": 315, "lr": 3.0115408852801535e-05} +{"train_loss": 0.013896642252802849, "grad_norm": 0.08254406601190567, "global_step": 33400, "epoch": 315, "lr": 3.0112689575359866e-05} +{"train_loss": 0.014837018214166164, "grad_norm": 0.07895736396312714, "global_step": 33401, "epoch": 315, "lr": 3.0109970367793557e-05} +{"train_loss": 0.014560151845216751, "grad_norm": 0.14439859986305237, "global_step": 33402, "epoch": 315, "lr": 3.0107251230112145e-05} +{"train_loss": 0.01590154878795147, "grad_norm": 0.05930926650762558, "global_step": 33403, "epoch": 315, "lr": 3.0104532162325215e-05} +{"train_loss": 0.014555240981280804, "grad_norm": 0.06993618607521057, "global_step": 33404, "epoch": 315, "lr": 3.0101813164442322e-05} +{"train_loss": 0.01550221350044012, "grad_norm": 0.12194427847862244, "global_step": 33405, "epoch": 315, "lr": 3.0099094236472993e-05} +{"train_loss": 0.013706183061003685, "grad_norm": 0.08505681157112122, "global_step": 33406, "epoch": 315, "lr": 3.009637537842678e-05} +{"train_loss": 0.016659589484333992, "grad_norm": 0.15808451175689697, "global_step": 33407, "epoch": 315, "lr": 3.009365659031326e-05} +{"train_loss": 0.01710893213748932, "grad_norm": 0.10080935060977936, "global_step": 33408, "epoch": 315, "lr": 3.0090937872141966e-05} +{"train_loss": 0.018627377226948738, "grad_norm": 0.10486789792776108, "global_step": 33409, "epoch": 315, "lr": 3.0088219223922453e-05} +{"train_loss": 0.015517791733145714, "grad_norm": 0.06086214631795883, "global_step": 33410, "epoch": 315, "lr": 3.008550064566431e-05} +{"train_loss": 0.014255066402256489, "grad_norm": 0.06437833607196808, "global_step": 33411, "epoch": 315, "lr": 3.0082782137377018e-05} +{"train_loss": 0.01573178544640541, "grad_norm": 0.06927741318941116, "global_step": 33412, "epoch": 315, "lr": 3.0080063699070192e-05} +{"train_loss": 0.016477622091770172, "grad_norm": 0.12129323929548264, "global_step": 33413, "epoch": 315, "lr": 3.007734533075335e-05} +{"train_loss": 0.01519007422029972, "grad_norm": 0.09377232939004898, "global_step": 33414, "epoch": 315, "lr": 3.007462703243604e-05} +{"train_loss": 0.017441660165786743, "grad_norm": 0.11785297095775604, "global_step": 33415, "epoch": 315, "lr": 3.0071908804127846e-05} +{"train_loss": 0.018217045813798904, "grad_norm": 0.07147329300642014, "global_step": 33416, "epoch": 315, "lr": 3.00691906458383e-05} +{"train_loss": 0.017290759831666946, "grad_norm": 0.1078360304236412, "global_step": 33417, "epoch": 315, "lr": 3.006647255757694e-05} +{"train_loss": 0.013500094413757324, "grad_norm": 0.15183024108409882, "global_step": 33418, "epoch": 315, "lr": 3.0063754539353334e-05} +{"train_loss": 0.015644853934645653, "grad_norm": 0.11862219125032425, "global_step": 33419, "epoch": 315, "lr": 3.0061036591177004e-05} +{"train_loss": 0.012947297655045986, "grad_norm": 0.09150082617998123, "global_step": 33420, "epoch": 315, "lr": 3.0058318713057542e-05} +{"train_loss": 0.014103317633271217, "grad_norm": 0.06988707929849625, "global_step": 33421, "epoch": 315, "lr": 3.0055600905004466e-05} +{"train_loss": 0.014835990034043789, "grad_norm": 0.0778597891330719, "global_step": 33422, "epoch": 315, "lr": 3.005288316702735e-05} +{"train_loss": 0.013463572598993778, "grad_norm": 0.06829792261123657, "global_step": 33423, "epoch": 315, "lr": 3.0050165499135706e-05} +{"train_loss": 0.016951467841863632, "grad_norm": 0.09707579016685486, "global_step": 33424, "epoch": 315, "lr": 3.0047447901339114e-05} +{"train_loss": 0.015601033344864845, "grad_norm": 0.06057083606719971, "global_step": 33425, "epoch": 315, "lr": 3.0044730373647113e-05} +{"train_loss": 0.01569291390478611, "grad_norm": 0.07939984649419785, "global_step": 33426, "epoch": 315, "lr": 3.004201291606924e-05} +{"train_loss": 0.015283117070794106, "grad_norm": 0.06209418922662735, "global_step": 33427, "epoch": 315, "lr": 3.003929552861507e-05} +{"train_loss": 0.015073107555508614, "grad_norm": 0.07983702421188354, "global_step": 33428, "epoch": 315, "lr": 3.003657821129414e-05} +{"train_loss": 0.016873350366950035, "grad_norm": 0.10658172518014908, "global_step": 33429, "epoch": 315, "lr": 3.0033860964115978e-05} +{"train_loss": 0.014368877746164799, "grad_norm": 0.06050272285938263, "global_step": 33430, "epoch": 315, "lr": 3.0031143787090156e-05} +{"train_loss": 0.014500539749860764, "grad_norm": 0.05056404694914818, "global_step": 33431, "epoch": 315, "lr": 3.0028426680226196e-05} +{"train_loss": 0.014586379751563072, "grad_norm": 0.05915190652012825, "global_step": 33432, "epoch": 315, "lr": 3.0025709643533662e-05} +{"train_loss": 0.012758947908878326, "grad_norm": 0.11087537556886673, "global_step": 33433, "epoch": 315, "lr": 3.002299267702212e-05} +{"train_loss": 0.017201397567987442, "grad_norm": 0.0733911395072937, "global_step": 33434, "epoch": 315, "lr": 3.002027578070108e-05} +{"train_loss": 0.01468854770064354, "grad_norm": 0.0920180156826973, "global_step": 33435, "epoch": 315, "lr": 3.0017558954580084e-05} +{"train_loss": 0.013763473369181156, "grad_norm": 0.08852507919073105, "global_step": 33436, "epoch": 315, "lr": 3.0014842198668712e-05} +{"train_loss": 0.013627314940094948, "grad_norm": 0.056309062987565994, "global_step": 33437, "epoch": 315, "lr": 3.0012125512976497e-05} +{"train_loss": 0.017923321574926376, "grad_norm": 0.11274344474077225, "global_step": 33438, "epoch": 315, "lr": 3.000940889751296e-05} +{"train_loss": 0.013523012399673462, "grad_norm": 0.09796646237373352, "global_step": 33439, "epoch": 315, "lr": 3.00066923522877e-05} +{"train_loss": 0.01338071096688509, "grad_norm": 0.07168182730674744, "global_step": 33440, "epoch": 315, "lr": 3.000397587731021e-05} +{"train_loss": 0.015379906632006168, "grad_norm": 0.07436566054821014, "global_step": 33441, "epoch": 315, "lr": 3.0001259472590047e-05} +{"train_loss": 0.016547903418540955, "grad_norm": 0.08825784921646118, "global_step": 33442, "epoch": 315, "lr": 2.999854313813677e-05} +{"train_loss": 0.015908265486359596, "grad_norm": 0.09797078371047974, "global_step": 33443, "epoch": 315, "lr": 2.9995826873959897e-05} +{"train_loss": 0.014907093718647957, "grad_norm": 0.07393795996904373, "global_step": 33444, "epoch": 315, "lr": 2.9993110680069008e-05} +{"train_loss": 0.016826607286930084, "grad_norm": 0.10311641544103622, "global_step": 33445, "epoch": 315, "lr": 2.9990394556473633e-05} +{"train_loss": 0.013302095234394073, "grad_norm": 0.05643144249916077, "global_step": 33446, "epoch": 315, "lr": 2.9987678503183304e-05} +{"train_loss": 0.01324534136801958, "grad_norm": 0.06229468062520027, "global_step": 33447, "epoch": 315, "lr": 2.998496252020756e-05} +{"train_loss": 0.013316179625689983, "grad_norm": 0.11091397702693939, "global_step": 33448, "epoch": 315, "lr": 2.998224660755595e-05} +{"train_loss": 0.012747804634273052, "grad_norm": 0.11032544821500778, "global_step": 33449, "epoch": 315, "lr": 2.997953076523803e-05} +{"train_loss": 0.014213519170880318, "grad_norm": 0.05610501393675804, "global_step": 33450, "epoch": 315, "lr": 2.9976814993263315e-05} +{"train_loss": 0.015085493214428425, "grad_norm": 0.1439729481935501, "global_step": 33451, "epoch": 315, "lr": 2.99740992916414e-05} +{"train_loss": 0.01359812542796135, "grad_norm": 0.14568178355693817, "global_step": 33452, "epoch": 315, "lr": 2.997138366038176e-05} +{"train_loss": 0.015158543363213539, "grad_norm": 0.11963233351707458, "global_step": 33453, "epoch": 315, "lr": 2.996866809949398e-05} +{"train_loss": 0.013234736397862434, "grad_norm": 0.09017926454544067, "global_step": 33454, "epoch": 315, "lr": 2.996595260898759e-05} +{"train_loss": 0.014208277687430382, "grad_norm": 0.09657511115074158, "global_step": 33455, "epoch": 315, "lr": 2.9963237188872117e-05} +{"train_loss": 0.01541119534522295, "grad_norm": 0.13228406012058258, "global_step": 33456, "epoch": 315, "lr": 2.996052183915712e-05} +{"train_loss": 0.014782055281102657, "grad_norm": 0.07482942193746567, "global_step": 33457, "epoch": 315, "lr": 2.9957806559852152e-05} +{"train_loss": 0.015622161328792572, "grad_norm": 0.07574136555194855, "global_step": 33458, "epoch": 315, "lr": 2.9955091350966725e-05} +{"train_loss": 0.013687816448509693, "grad_norm": 0.0655638575553894, "global_step": 33459, "epoch": 315, "lr": 2.9952376212510384e-05} +{"train_loss": 0.016394127160310745, "grad_norm": 0.1727205067873001, "global_step": 33460, "epoch": 315, "lr": 2.994966114449267e-05} +{"train_loss": 0.015629716217517853, "grad_norm": 0.0678042322397232, "global_step": 33461, "epoch": 315, "lr": 2.9946946146923132e-05} +{"train_loss": 0.015928160399198532, "grad_norm": 0.06251715123653412, "global_step": 33462, "epoch": 315, "lr": 2.994423121981131e-05} +{"train_loss": 0.01685524359345436, "grad_norm": 0.09594376385211945, "global_step": 33463, "epoch": 315, "lr": 2.9941516363166743e-05} +{"train_loss": 0.015355664305388927, "grad_norm": 0.09343121200799942, "global_step": 33464, "epoch": 315, "lr": 2.9938801576998947e-05} +{"train_loss": 0.012038614600896835, "grad_norm": 0.08581521362066269, "global_step": 33465, "epoch": 315, "lr": 2.9936086861317492e-05} +{"train_loss": 0.015038516372442245, "grad_norm": 0.10519744455814362, "global_step": 33466, "epoch": 315, "lr": 2.9933372216131895e-05} +{"train_loss": 0.0154033824801445, "grad_norm": 0.061626650393009186, "global_step": 33467, "epoch": 315, "lr": 2.9930657641451692e-05} +{"train_loss": 0.016399798914790154, "grad_norm": 0.16035939753055573, "global_step": 33468, "epoch": 315, "lr": 2.992794313728644e-05} +{"train_loss": 0.014829847030341625, "grad_norm": 0.06559371948242188, "global_step": 33469, "epoch": 315, "lr": 2.992522870364568e-05} +{"train_loss": 0.017084084451198578, "grad_norm": 0.06537987291812897, "global_step": 33470, "epoch": 315, "lr": 2.9922514340538922e-05} +{"train_loss": 0.01675983890891075, "grad_norm": 0.11449559777975082, "global_step": 33471, "epoch": 315, "lr": 2.991980004797572e-05} +{"train_loss": 0.015532080084085464, "grad_norm": 0.06478305160999298, "global_step": 33472, "epoch": 315, "lr": 2.9917085825965596e-05} +{"train_loss": 0.013000995852053165, "grad_norm": 0.10152030736207962, "global_step": 33473, "epoch": 315, "lr": 2.9914371674518116e-05} +{"train_loss": 0.016845718026161194, "grad_norm": 0.08081735670566559, "global_step": 33474, "epoch": 315, "lr": 2.9911657593642793e-05} +{"train_loss": 0.017039228230714798, "grad_norm": 0.09753042459487915, "global_step": 33475, "epoch": 315, "lr": 2.990894358334918e-05} +{"train_loss": 0.014118965715169907, "grad_norm": 0.07131467014551163, "global_step": 33476, "epoch": 315, "lr": 2.990622964364679e-05} +{"train_loss": 0.017071198672056198, "grad_norm": 0.16622065007686615, "global_step": 33477, "epoch": 315, "lr": 2.990351577454516e-05} +{"train_loss": 0.015812400728464127, "grad_norm": 0.05065707489848137, "global_step": 33478, "epoch": 315, "lr": 2.990080197605385e-05} +{"train_loss": 0.016603857278823853, "grad_norm": 0.07809498906135559, "global_step": 33479, "epoch": 315, "lr": 2.9898088248182366e-05} +{"train_loss": 0.013892068527638912, "grad_norm": 0.09982703626155853, "global_step": 33480, "epoch": 315, "lr": 2.989537459094029e-05} +{"train_loss": 0.01532801240682602, "grad_norm": 0.0812123492360115, "global_step": 33481, "epoch": 315, "lr": 2.9892661004337096e-05} +{"train_loss": 0.015723953023552895, "grad_norm": 0.09193894267082214, "global_step": 33482, "epoch": 315, "lr": 2.9889947488382353e-05} +{"train_loss": 0.016669578850269318, "grad_norm": 0.0709187313914299, "global_step": 33483, "epoch": 315, "lr": 2.9887234043085595e-05} +{"train_loss": 0.012438202276825905, "grad_norm": 0.08245003968477249, "global_step": 33484, "epoch": 315, "lr": 2.9884520668456335e-05} +{"train_loss": 0.014275469817221165, "grad_norm": 0.10728179663419724, "global_step": 33485, "epoch": 315, "lr": 2.9881807364504134e-05} +{"train_loss": 0.016474343836307526, "grad_norm": 0.13784503936767578, "global_step": 33486, "epoch": 315, "lr": 2.9879094131238516e-05} +{"train_loss": 0.014764492399990559, "grad_norm": 0.05607245862483978, "global_step": 33487, "epoch": 315, "lr": 2.9876380968669006e-05} +{"train_loss": 0.012094958685338497, "grad_norm": 0.08833103626966476, "global_step": 33488, "epoch": 315, "lr": 2.9873667876805144e-05} +{"train_loss": 0.015470709651708603, "grad_norm": 0.06261985003948212, "global_step": 33489, "epoch": 315, "lr": 2.9870954855656447e-05} +{"train_loss": 0.013893885537981987, "grad_norm": 0.081993468105793, "global_step": 33490, "epoch": 315, "lr": 2.9868241905232475e-05} +{"train_loss": 0.014050178229808807, "grad_norm": 0.09265491366386414, "global_step": 33491, "epoch": 315, "lr": 2.986552902554274e-05} +{"train_loss": 0.015334188006818295, "grad_norm": 0.12116684019565582, "global_step": 33492, "epoch": 315, "lr": 2.986281621659679e-05} +{"train_loss": 0.016692018136382103, "grad_norm": 0.09780571609735489, "global_step": 33493, "epoch": 315, "lr": 2.9860103478404135e-05} +{"train_loss": 0.015640119090676308, "grad_norm": 0.11124425381422043, "global_step": 33494, "epoch": 315, "lr": 2.9857390810974317e-05} +{"train_loss": 0.01521855027783873, "grad_norm": 0.14583854377269745, "global_step": 33495, "epoch": 315, "lr": 2.9854678214316873e-05, "val_loss": 0.019238490611314774} +{"train_loss": 0.01466099452227354, "grad_norm": 0.11255946010351181, "global_step": 33496, "epoch": 316, "lr": 2.9851965688441318e-05} +{"train_loss": 0.013769682496786118, "grad_norm": 0.07941626757383347, "global_step": 33497, "epoch": 316, "lr": 2.9849253233357205e-05} +{"train_loss": 0.018061283975839615, "grad_norm": 0.09638600051403046, "global_step": 33498, "epoch": 316, "lr": 2.9846540849074062e-05} +{"train_loss": 0.013665908016264439, "grad_norm": 0.06758378446102142, "global_step": 33499, "epoch": 316, "lr": 2.98438285356014e-05} +{"train_loss": 0.017070982605218887, "grad_norm": 0.07661163061857224, "global_step": 33500, "epoch": 316, "lr": 2.9841116292948756e-05} +{"train_loss": 0.013636916875839233, "grad_norm": 0.07538248598575592, "global_step": 33501, "epoch": 316, "lr": 2.9838404121125657e-05} +{"train_loss": 0.01385371945798397, "grad_norm": 0.10857138782739639, "global_step": 33502, "epoch": 316, "lr": 2.983569202014165e-05} +{"train_loss": 0.015758182853460312, "grad_norm": 0.1598394811153412, "global_step": 33503, "epoch": 316, "lr": 2.9832979990006244e-05} +{"train_loss": 0.015258672647178173, "grad_norm": 0.09181221574544907, "global_step": 33504, "epoch": 316, "lr": 2.983026803072899e-05} +{"train_loss": 0.014186922460794449, "grad_norm": 0.07557636499404907, "global_step": 33505, "epoch": 316, "lr": 2.982755614231938e-05} +{"train_loss": 0.016770198941230774, "grad_norm": 0.13350841403007507, "global_step": 33506, "epoch": 316, "lr": 2.9824844324786977e-05} +{"train_loss": 0.017320170998573303, "grad_norm": 0.13762202858924866, "global_step": 33507, "epoch": 316, "lr": 2.98221325781413e-05} +{"train_loss": 0.01684095710515976, "grad_norm": 0.06992700695991516, "global_step": 33508, "epoch": 316, "lr": 2.981942090239186e-05} +{"train_loss": 0.015824541449546814, "grad_norm": 0.11784794181585312, "global_step": 33509, "epoch": 316, "lr": 2.9816709297548207e-05} +{"train_loss": 0.015579108148813248, "grad_norm": 0.053796425461769104, "global_step": 33510, "epoch": 316, "lr": 2.981399776361987e-05} +{"train_loss": 0.014255097135901451, "grad_norm": 0.08799674361944199, "global_step": 33511, "epoch": 316, "lr": 2.9811286300616366e-05} +{"train_loss": 0.011280233040452003, "grad_norm": 0.07618376612663269, "global_step": 33512, "epoch": 316, "lr": 2.9808574908547214e-05} +{"train_loss": 0.01590210758149624, "grad_norm": 0.09287647902965546, "global_step": 33513, "epoch": 316, "lr": 2.980586358742194e-05} +{"train_loss": 0.012807518243789673, "grad_norm": 0.06351450085639954, "global_step": 33514, "epoch": 316, "lr": 2.9803152337250083e-05} +{"train_loss": 0.014031280763447285, "grad_norm": 0.07026548683643341, "global_step": 33515, "epoch": 316, "lr": 2.9800441158041175e-05} +{"train_loss": 0.013837888836860657, "grad_norm": 0.07883251458406448, "global_step": 33516, "epoch": 316, "lr": 2.979773004980474e-05} +{"train_loss": 0.015393145382404327, "grad_norm": 0.06343820691108704, "global_step": 33517, "epoch": 316, "lr": 2.9795019012550284e-05} +{"train_loss": 0.01558093260973692, "grad_norm": 0.09923160821199417, "global_step": 33518, "epoch": 316, "lr": 2.9792308046287333e-05} +{"train_loss": 0.01578601822257042, "grad_norm": 0.07791071385145187, "global_step": 33519, "epoch": 316, "lr": 2.9789597151025438e-05} +{"train_loss": 0.01372717134654522, "grad_norm": 0.05800815671682358, "global_step": 33520, "epoch": 316, "lr": 2.9786886326774106e-05} +{"train_loss": 0.013514697551727295, "grad_norm": 0.09699583053588867, "global_step": 33521, "epoch": 316, "lr": 2.9784175573542855e-05} +{"train_loss": 0.013802153058350086, "grad_norm": 0.05547833442687988, "global_step": 33522, "epoch": 316, "lr": 2.9781464891341238e-05} +{"train_loss": 0.013254490680992603, "grad_norm": 0.05481933057308197, "global_step": 33523, "epoch": 316, "lr": 2.9778754280178754e-05} +{"train_loss": 0.013901758939027786, "grad_norm": 0.0591549314558506, "global_step": 33524, "epoch": 316, "lr": 2.977604374006493e-05} +{"train_loss": 0.015341170132160187, "grad_norm": 0.059190817177295685, "global_step": 33525, "epoch": 316, "lr": 2.9773333271009284e-05} +{"train_loss": 0.016515690833330154, "grad_norm": 0.10878337919712067, "global_step": 33526, "epoch": 316, "lr": 2.9770622873021364e-05} +{"train_loss": 0.013191217556595802, "grad_norm": 0.06387189775705338, "global_step": 33527, "epoch": 316, "lr": 2.9767912546110675e-05} +{"train_loss": 0.014780115336179733, "grad_norm": 0.15558455884456635, "global_step": 33528, "epoch": 316, "lr": 2.9765202290286753e-05} +{"train_loss": 0.014817086048424244, "grad_norm": 0.1213054209947586, "global_step": 33529, "epoch": 316, "lr": 2.9762492105559092e-05} +{"train_loss": 0.015949875116348267, "grad_norm": 0.17882943153381348, "global_step": 33530, "epoch": 316, "lr": 2.975978199193723e-05} +{"train_loss": 0.016096796840429306, "grad_norm": 0.10808906704187393, "global_step": 33531, "epoch": 316, "lr": 2.9757071949430703e-05} +{"train_loss": 0.016676535829901695, "grad_norm": 0.11470413208007812, "global_step": 33532, "epoch": 316, "lr": 2.9754361978049016e-05} +{"train_loss": 0.015310680493712425, "grad_norm": 0.06012844666838646, "global_step": 33533, "epoch": 316, "lr": 2.9751652077801718e-05} +{"train_loss": 0.015331005677580833, "grad_norm": 0.08726021647453308, "global_step": 33534, "epoch": 316, "lr": 2.9748942248698276e-05} +{"train_loss": 0.016433339565992355, "grad_norm": 0.0726890116930008, "global_step": 33535, "epoch": 316, "lr": 2.974623249074826e-05} +{"train_loss": 0.017474275082349777, "grad_norm": 0.07550020515918732, "global_step": 33536, "epoch": 316, "lr": 2.974352280396117e-05} +{"train_loss": 0.017380617558956146, "grad_norm": 0.08818546682596207, "global_step": 33537, "epoch": 316, "lr": 2.9740813188346528e-05} +{"train_loss": 0.012138687074184418, "grad_norm": 0.0514964759349823, "global_step": 33538, "epoch": 316, "lr": 2.973810364391386e-05} +{"train_loss": 0.016565483063459396, "grad_norm": 0.10680598020553589, "global_step": 33539, "epoch": 316, "lr": 2.9735394170672704e-05} +{"train_loss": 0.016734236851334572, "grad_norm": 0.08330155164003372, "global_step": 33540, "epoch": 316, "lr": 2.9732684768632547e-05} +{"train_loss": 0.014867961406707764, "grad_norm": 0.056580618023872375, "global_step": 33541, "epoch": 316, "lr": 2.9729975437802914e-05} +{"train_loss": 0.01617540791630745, "grad_norm": 0.07190989702939987, "global_step": 33542, "epoch": 316, "lr": 2.9727266178193336e-05} +{"train_loss": 0.013762217015028, "grad_norm": 0.09783883392810822, "global_step": 33543, "epoch": 316, "lr": 2.9724556989813323e-05} +{"train_loss": 0.01666058413684368, "grad_norm": 0.0908561572432518, "global_step": 33544, "epoch": 316, "lr": 2.9721847872672413e-05} +{"train_loss": 0.014645067974925041, "grad_norm": 0.06161359325051308, "global_step": 33545, "epoch": 316, "lr": 2.9719138826780114e-05} +{"train_loss": 0.014265778474509716, "grad_norm": 0.13162875175476074, "global_step": 33546, "epoch": 316, "lr": 2.971642985214593e-05} +{"train_loss": 0.01468810997903347, "grad_norm": 0.09635350853204727, "global_step": 33547, "epoch": 316, "lr": 2.9713720948779383e-05} +{"train_loss": 0.015374485403299332, "grad_norm": 0.11324450373649597, "global_step": 33548, "epoch": 316, "lr": 2.9711012116690007e-05} +{"train_loss": 0.014228787273168564, "grad_norm": 0.11134658753871918, "global_step": 33549, "epoch": 316, "lr": 2.9708303355887313e-05} +{"train_loss": 0.01573583483695984, "grad_norm": 0.11267919093370438, "global_step": 33550, "epoch": 316, "lr": 2.9705594666380805e-05} +{"train_loss": 0.01407681219279766, "grad_norm": 0.08439331501722336, "global_step": 33551, "epoch": 316, "lr": 2.970288604818003e-05} +{"train_loss": 0.015844300389289856, "grad_norm": 0.06422968953847885, "global_step": 33552, "epoch": 316, "lr": 2.970017750129448e-05} +{"train_loss": 0.018001914024353027, "grad_norm": 0.07026628404855728, "global_step": 33553, "epoch": 316, "lr": 2.969746902573367e-05} +{"train_loss": 0.012881503440439701, "grad_norm": 0.06231115385890007, "global_step": 33554, "epoch": 316, "lr": 2.969476062150712e-05} +{"train_loss": 0.0178727637976408, "grad_norm": 0.13604222238063812, "global_step": 33555, "epoch": 316, "lr": 2.969205228862436e-05} +{"train_loss": 0.018994390964508057, "grad_norm": 0.07885026186704636, "global_step": 33556, "epoch": 316, "lr": 2.96893440270949e-05} +{"train_loss": 0.015716848894953728, "grad_norm": 0.06343555450439453, "global_step": 33557, "epoch": 316, "lr": 2.968663583692826e-05} +{"train_loss": 0.014695089310407639, "grad_norm": 0.09500113874673843, "global_step": 33558, "epoch": 316, "lr": 2.9683927718133943e-05} +{"train_loss": 0.01598936691880226, "grad_norm": 0.07250668108463287, "global_step": 33559, "epoch": 316, "lr": 2.968121967072145e-05} +{"train_loss": 0.014071499928832054, "grad_norm": 0.09971310198307037, "global_step": 33560, "epoch": 316, "lr": 2.9678511694700327e-05} +{"train_loss": 0.013144884258508682, "grad_norm": 0.06383812427520752, "global_step": 33561, "epoch": 316, "lr": 2.967580379008008e-05} +{"train_loss": 0.013799766078591347, "grad_norm": 0.061169471591711044, "global_step": 33562, "epoch": 316, "lr": 2.9673095956870202e-05} +{"train_loss": 0.013606649823486805, "grad_norm": 0.05129272863268852, "global_step": 33563, "epoch": 316, "lr": 2.967038819508025e-05} +{"train_loss": 0.014040399342775345, "grad_norm": 0.1180904358625412, "global_step": 33564, "epoch": 316, "lr": 2.966768050471971e-05} +{"train_loss": 0.014797613956034184, "grad_norm": 0.058304257690906525, "global_step": 33565, "epoch": 316, "lr": 2.9664972885798086e-05} +{"train_loss": 0.01411458570510149, "grad_norm": 0.10306287556886673, "global_step": 33566, "epoch": 316, "lr": 2.96622653383249e-05} +{"train_loss": 0.015615498647093773, "grad_norm": 0.08620038628578186, "global_step": 33567, "epoch": 316, "lr": 2.9659557862309673e-05} +{"train_loss": 0.014471346512436867, "grad_norm": 0.100193090736866, "global_step": 33568, "epoch": 316, "lr": 2.965685045776192e-05} +{"train_loss": 0.015386976301670074, "grad_norm": 0.16447028517723083, "global_step": 33569, "epoch": 316, "lr": 2.965414312469116e-05} +{"train_loss": 0.016921617090702057, "grad_norm": 0.07028591632843018, "global_step": 33570, "epoch": 316, "lr": 2.965143586310688e-05} +{"train_loss": 0.015184694901108742, "grad_norm": 0.09733334928750992, "global_step": 33571, "epoch": 316, "lr": 2.964872867301859e-05} +{"train_loss": 0.016665760427713394, "grad_norm": 0.14682790637016296, "global_step": 33572, "epoch": 316, "lr": 2.9646021554435832e-05} +{"train_loss": 0.015313174575567245, "grad_norm": 0.05860753357410431, "global_step": 33573, "epoch": 316, "lr": 2.9643314507368103e-05} +{"train_loss": 0.015607727691531181, "grad_norm": 0.06825031340122223, "global_step": 33574, "epoch": 316, "lr": 2.9640607531824904e-05} +{"train_loss": 0.013909104280173779, "grad_norm": 0.07116282731294632, "global_step": 33575, "epoch": 316, "lr": 2.9637900627815785e-05} +{"train_loss": 0.01250615157186985, "grad_norm": 0.06689057499170303, "global_step": 33576, "epoch": 316, "lr": 2.963519379535019e-05} +{"train_loss": 0.014108307659626007, "grad_norm": 0.067875936627388, "global_step": 33577, "epoch": 316, "lr": 2.963248703443769e-05} +{"train_loss": 0.01488993689417839, "grad_norm": 0.06203547120094299, "global_step": 33578, "epoch": 316, "lr": 2.9629780345087766e-05} +{"train_loss": 0.01360709872096777, "grad_norm": 0.10895020514726639, "global_step": 33579, "epoch": 316, "lr": 2.9627073727309928e-05} +{"train_loss": 0.01674412190914154, "grad_norm": 0.07233651727437973, "global_step": 33580, "epoch": 316, "lr": 2.962436718111371e-05} +{"train_loss": 0.014382224529981613, "grad_norm": 0.12090405821800232, "global_step": 33581, "epoch": 316, "lr": 2.9621660706508598e-05} +{"train_loss": 0.014106066897511482, "grad_norm": 0.09689558297395706, "global_step": 33582, "epoch": 316, "lr": 2.961895430350411e-05} +{"train_loss": 0.018758494406938553, "grad_norm": 0.12845027446746826, "global_step": 33583, "epoch": 316, "lr": 2.9616247972109733e-05} +{"train_loss": 0.015569908544421196, "grad_norm": 0.09649822115898132, "global_step": 33584, "epoch": 316, "lr": 2.9613541712335013e-05} +{"train_loss": 0.017669619992375374, "grad_norm": 0.07932775467634201, "global_step": 33585, "epoch": 316, "lr": 2.9610835524189447e-05} +{"train_loss": 0.017097270116209984, "grad_norm": 0.07108568400144577, "global_step": 33586, "epoch": 316, "lr": 2.960812940768254e-05} +{"train_loss": 0.015774643048644066, "grad_norm": 0.06001931428909302, "global_step": 33587, "epoch": 316, "lr": 2.960542336282379e-05} +{"train_loss": 0.01372070237994194, "grad_norm": 0.10585009306669235, "global_step": 33588, "epoch": 316, "lr": 2.96027173896227e-05} +{"train_loss": 0.013603451661765575, "grad_norm": 0.07060941308736801, "global_step": 33589, "epoch": 316, "lr": 2.960001148808881e-05} +{"train_loss": 0.018027445301413536, "grad_norm": 0.10185550153255463, "global_step": 33590, "epoch": 316, "lr": 2.95973056582316e-05} +{"train_loss": 0.01495925523340702, "grad_norm": 0.07681351155042648, "global_step": 33591, "epoch": 316, "lr": 2.959459990006057e-05} +{"train_loss": 0.013051614165306091, "grad_norm": 0.09128303825855255, "global_step": 33592, "epoch": 316, "lr": 2.959189421358527e-05} +{"train_loss": 0.014821234159171581, "grad_norm": 0.09232509881258011, "global_step": 33593, "epoch": 316, "lr": 2.9589188598815172e-05} +{"train_loss": 0.016475219279527664, "grad_norm": 0.059452131390571594, "global_step": 33594, "epoch": 316, "lr": 2.9586483055759785e-05} +{"train_loss": 0.015456940047442913, "grad_norm": 0.07931811362504959, "global_step": 33595, "epoch": 316, "lr": 2.9583777584428607e-05} +{"train_loss": 0.01828703097999096, "grad_norm": 0.06643518060445786, "global_step": 33596, "epoch": 316, "lr": 2.9581072184831165e-05} +{"train_loss": 0.01639515534043312, "grad_norm": 0.12854866683483124, "global_step": 33597, "epoch": 316, "lr": 2.9578366856976956e-05} +{"train_loss": 0.01380110438913107, "grad_norm": 0.08823250234127045, "global_step": 33598, "epoch": 316, "lr": 2.957566160087549e-05} +{"train_loss": 0.015198800712823868, "grad_norm": 0.062024615705013275, "global_step": 33599, "epoch": 316, "lr": 2.9572956416536267e-05} +{"train_loss": 0.016388334333896637, "grad_norm": 0.1546945720911026, "global_step": 33600, "epoch": 316, "lr": 2.9570251303968778e-05} +{"train_loss": 0.015219972161400431, "grad_norm": 0.10447367280721664, "global_step": 33601, "epoch": 316, "lr": 2.9567546263182556e-05, "val_loss": 0.019579323008656502} +{"train_loss": 0.015902245417237282, "grad_norm": 0.12701298296451569, "global_step": 33602, "epoch": 317, "lr": 2.9564841294187083e-05} +{"train_loss": 0.016665153205394745, "grad_norm": 0.11217308044433594, "global_step": 33603, "epoch": 317, "lr": 2.956213639699186e-05} +{"train_loss": 0.017867226153612137, "grad_norm": 0.09075865894556046, "global_step": 33604, "epoch": 317, "lr": 2.9559431571606434e-05} +{"train_loss": 0.01770954392850399, "grad_norm": 0.06996499001979828, "global_step": 33605, "epoch": 317, "lr": 2.955672681804025e-05} +{"train_loss": 0.016263246536254883, "grad_norm": 0.07555587589740753, "global_step": 33606, "epoch": 317, "lr": 2.955402213630284e-05} +{"train_loss": 0.01575225591659546, "grad_norm": 0.08287316560745239, "global_step": 33607, "epoch": 317, "lr": 2.9551317526403698e-05} +{"train_loss": 0.01468513160943985, "grad_norm": 0.08247414976358414, "global_step": 33608, "epoch": 317, "lr": 2.954861298835234e-05} +{"train_loss": 0.018297579139471054, "grad_norm": 0.07985290139913559, "global_step": 33609, "epoch": 317, "lr": 2.954590852215827e-05} +{"train_loss": 0.01757754385471344, "grad_norm": 0.10411430150270462, "global_step": 33610, "epoch": 317, "lr": 2.9543204127830986e-05} +{"train_loss": 0.01790766790509224, "grad_norm": 0.09968426823616028, "global_step": 33611, "epoch": 317, "lr": 2.954049980537998e-05} +{"train_loss": 0.016410188749432564, "grad_norm": 0.06888452917337418, "global_step": 33612, "epoch": 317, "lr": 2.9537795554814753e-05} +{"train_loss": 0.014288637787103653, "grad_norm": 0.11499246209859848, "global_step": 33613, "epoch": 317, "lr": 2.9535091376144814e-05} +{"train_loss": 0.01380676869302988, "grad_norm": 0.054478295147418976, "global_step": 33614, "epoch": 317, "lr": 2.953238726937967e-05} +{"train_loss": 0.01472490094602108, "grad_norm": 0.09199296683073044, "global_step": 33615, "epoch": 317, "lr": 2.9529683234528803e-05} +{"train_loss": 0.015666387975215912, "grad_norm": 0.07926710695028305, "global_step": 33616, "epoch": 317, "lr": 2.9526979271601758e-05} +{"train_loss": 0.013346537947654724, "grad_norm": 0.06588440388441086, "global_step": 33617, "epoch": 317, "lr": 2.9524275380607973e-05} +{"train_loss": 0.014521246775984764, "grad_norm": 0.09860767424106598, "global_step": 33618, "epoch": 317, "lr": 2.9521571561556992e-05} +{"train_loss": 0.014014454558491707, "grad_norm": 0.1607220470905304, "global_step": 33619, "epoch": 317, "lr": 2.9518867814458306e-05} +{"train_loss": 0.013778681866824627, "grad_norm": 0.15239186584949493, "global_step": 33620, "epoch": 317, "lr": 2.9516164139321394e-05} +{"train_loss": 0.013329937122762203, "grad_norm": 0.07201282680034637, "global_step": 33621, "epoch": 317, "lr": 2.951346053615579e-05} +{"train_loss": 0.016008161008358, "grad_norm": 0.0819154754281044, "global_step": 33622, "epoch": 317, "lr": 2.951075700497098e-05} +{"train_loss": 0.013339269906282425, "grad_norm": 0.15357589721679688, "global_step": 33623, "epoch": 317, "lr": 2.9508053545776455e-05} +{"train_loss": 0.014945680275559425, "grad_norm": 0.07945531606674194, "global_step": 33624, "epoch": 317, "lr": 2.9505350158581702e-05} +{"train_loss": 0.013994336128234863, "grad_norm": 0.06871544569730759, "global_step": 33625, "epoch": 317, "lr": 2.9502646843396252e-05} +{"train_loss": 0.01805941015481949, "grad_norm": 0.08973663300275803, "global_step": 33626, "epoch": 317, "lr": 2.9499943600229586e-05} +{"train_loss": 0.016421914100646973, "grad_norm": 0.06141816824674606, "global_step": 33627, "epoch": 317, "lr": 2.949724042909119e-05} +{"train_loss": 0.013631941750645638, "grad_norm": 0.052183058112859726, "global_step": 33628, "epoch": 317, "lr": 2.94945373299906e-05} +{"train_loss": 0.0166216641664505, "grad_norm": 0.1069527119398117, "global_step": 33629, "epoch": 317, "lr": 2.9491834302937265e-05} +{"train_loss": 0.013639099895954132, "grad_norm": 0.06478489935398102, "global_step": 33630, "epoch": 317, "lr": 2.948913134794071e-05} +{"train_loss": 0.0146593376994133, "grad_norm": 0.07170024514198303, "global_step": 33631, "epoch": 317, "lr": 2.9486428465010442e-05} +{"train_loss": 0.015730595216155052, "grad_norm": 0.06535442918539047, "global_step": 33632, "epoch": 317, "lr": 2.948372565415592e-05} +{"train_loss": 0.015622865408658981, "grad_norm": 0.11492446810007095, "global_step": 33633, "epoch": 317, "lr": 2.9481022915386698e-05} +{"train_loss": 0.01673906296491623, "grad_norm": 0.07398586720228195, "global_step": 33634, "epoch": 317, "lr": 2.9478320248712204e-05} +{"train_loss": 0.013312311843037605, "grad_norm": 0.08752808719873428, "global_step": 33635, "epoch": 317, "lr": 2.9475617654141986e-05} +{"train_loss": 0.01597645878791809, "grad_norm": 0.05482236295938492, "global_step": 33636, "epoch": 317, "lr": 2.9472915131685503e-05} +{"train_loss": 0.017370494082570076, "grad_norm": 0.10644172132015228, "global_step": 33637, "epoch": 317, "lr": 2.9470212681352284e-05} +{"train_loss": 0.01352810487151146, "grad_norm": 0.07634052634239197, "global_step": 33638, "epoch": 317, "lr": 2.9467510303151807e-05} +{"train_loss": 0.014016721397638321, "grad_norm": 0.07450360059738159, "global_step": 33639, "epoch": 317, "lr": 2.946480799709358e-05} +{"train_loss": 0.013269133865833282, "grad_norm": 0.06014426052570343, "global_step": 33640, "epoch": 317, "lr": 2.9462105763187074e-05} +{"train_loss": 0.016460515558719635, "grad_norm": 0.06804285198450089, "global_step": 33641, "epoch": 317, "lr": 2.945940360144179e-05} +{"train_loss": 0.014346458949148655, "grad_norm": 0.06398290395736694, "global_step": 33642, "epoch": 317, "lr": 2.9456701511867246e-05} +{"train_loss": 0.015416380949318409, "grad_norm": 0.0698070377111435, "global_step": 33643, "epoch": 317, "lr": 2.945399949447291e-05} +{"train_loss": 0.015177110210061073, "grad_norm": 0.07409746199846268, "global_step": 33644, "epoch": 317, "lr": 2.9451297549268275e-05} +{"train_loss": 0.014147637411952019, "grad_norm": 0.07063724845647812, "global_step": 33645, "epoch": 317, "lr": 2.944859567626288e-05} +{"train_loss": 0.013861192390322685, "grad_norm": 0.13776566088199615, "global_step": 33646, "epoch": 317, "lr": 2.9445893875466145e-05} +{"train_loss": 0.016470223665237427, "grad_norm": 0.06145366653800011, "global_step": 33647, "epoch": 317, "lr": 2.9443192146887617e-05} +{"train_loss": 0.014158807694911957, "grad_norm": 0.07434580475091934, "global_step": 33648, "epoch": 317, "lr": 2.9440490490536766e-05} +{"train_loss": 0.01447860524058342, "grad_norm": 0.09239226579666138, "global_step": 33649, "epoch": 317, "lr": 2.943778890642308e-05} +{"train_loss": 0.01484083104878664, "grad_norm": 0.0717797726392746, "global_step": 33650, "epoch": 317, "lr": 2.9435087394556078e-05} +{"train_loss": 0.01521545834839344, "grad_norm": 0.10963120311498642, "global_step": 33651, "epoch": 317, "lr": 2.943238595494524e-05} +{"train_loss": 0.012241515330970287, "grad_norm": 0.14660270512104034, "global_step": 33652, "epoch": 317, "lr": 2.9429684587600047e-05} +{"train_loss": 0.016322359442710876, "grad_norm": 0.10152526199817657, "global_step": 33653, "epoch": 317, "lr": 2.9426983292529986e-05} +{"train_loss": 0.014276294969022274, "grad_norm": 0.08387798070907593, "global_step": 33654, "epoch": 317, "lr": 2.9424282069744564e-05} +{"train_loss": 0.013577373698353767, "grad_norm": 0.08865022659301758, "global_step": 33655, "epoch": 317, "lr": 2.9421580919253265e-05} +{"train_loss": 0.015571773052215576, "grad_norm": 0.05949852988123894, "global_step": 33656, "epoch": 317, "lr": 2.9418879841065583e-05} +{"train_loss": 0.01591726765036583, "grad_norm": 0.08369998633861542, "global_step": 33657, "epoch": 317, "lr": 2.9416178835191028e-05} +{"train_loss": 0.014970027841627598, "grad_norm": 0.07957059890031815, "global_step": 33658, "epoch": 317, "lr": 2.9413477901639043e-05} +{"train_loss": 0.012892721220850945, "grad_norm": 0.11978455632925034, "global_step": 33659, "epoch": 317, "lr": 2.9410777040419145e-05} +{"train_loss": 0.014219372533261776, "grad_norm": 0.08893819153308868, "global_step": 33660, "epoch": 317, "lr": 2.9408076251540835e-05} +{"train_loss": 0.014796406030654907, "grad_norm": 0.07500488311052322, "global_step": 33661, "epoch": 317, "lr": 2.9405375535013568e-05} +{"train_loss": 0.012809954583644867, "grad_norm": 0.12704189121723175, "global_step": 33662, "epoch": 317, "lr": 2.940267489084687e-05} +{"train_loss": 0.013684302568435669, "grad_norm": 0.06918388605117798, "global_step": 33663, "epoch": 317, "lr": 2.939997431905023e-05} +{"train_loss": 0.01463092491030693, "grad_norm": 0.05650851130485535, "global_step": 33664, "epoch": 317, "lr": 2.939727381963311e-05} +{"train_loss": 0.015090620145201683, "grad_norm": 0.056216511875391006, "global_step": 33665, "epoch": 317, "lr": 2.9394573392604996e-05} +{"train_loss": 0.01435594167560339, "grad_norm": 0.06976586580276489, "global_step": 33666, "epoch": 317, "lr": 2.93918730379754e-05} +{"train_loss": 0.012032793834805489, "grad_norm": 0.07214352488517761, "global_step": 33667, "epoch": 317, "lr": 2.9389172755753804e-05} +{"train_loss": 0.011860286816954613, "grad_norm": 0.06647900491952896, "global_step": 33668, "epoch": 317, "lr": 2.9386472545949673e-05} +{"train_loss": 0.014127753674983978, "grad_norm": 0.06189865618944168, "global_step": 33669, "epoch": 317, "lr": 2.938377240857255e-05} +{"train_loss": 0.015116826631128788, "grad_norm": 0.06582627445459366, "global_step": 33670, "epoch": 317, "lr": 2.9381072343631853e-05} +{"train_loss": 0.015477996319532394, "grad_norm": 0.06831938028335571, "global_step": 33671, "epoch": 317, "lr": 2.9378372351137112e-05} +{"train_loss": 0.01661638915538788, "grad_norm": 0.087649405002594, "global_step": 33672, "epoch": 317, "lr": 2.9375672431097802e-05} +{"train_loss": 0.014586164616048336, "grad_norm": 0.1148483008146286, "global_step": 33673, "epoch": 317, "lr": 2.93729725835234e-05} +{"train_loss": 0.016318533569574356, "grad_norm": 0.073575459420681, "global_step": 33674, "epoch": 317, "lr": 2.937027280842341e-05} +{"train_loss": 0.017718229442834854, "grad_norm": 0.20360742509365082, "global_step": 33675, "epoch": 317, "lr": 2.9367573105807323e-05} +{"train_loss": 0.014397104270756245, "grad_norm": 0.10302425175905228, "global_step": 33676, "epoch": 317, "lr": 2.9364873475684596e-05} +{"train_loss": 0.01547081395983696, "grad_norm": 0.1405222862958908, "global_step": 33677, "epoch": 317, "lr": 2.9362173918064738e-05} +{"train_loss": 0.017833232879638672, "grad_norm": 0.09933997690677643, "global_step": 33678, "epoch": 317, "lr": 2.9359474432957206e-05} +{"train_loss": 0.01573699712753296, "grad_norm": 0.07395579665899277, "global_step": 33679, "epoch": 317, "lr": 2.935677502037152e-05} +{"train_loss": 0.014689480885863304, "grad_norm": 0.07807926833629608, "global_step": 33680, "epoch": 317, "lr": 2.935407568031715e-05} +{"train_loss": 0.014364399015903473, "grad_norm": 0.0648694857954979, "global_step": 33681, "epoch": 317, "lr": 2.935137641280359e-05} +{"train_loss": 0.014713071286678314, "grad_norm": 0.08441879600286484, "global_step": 33682, "epoch": 317, "lr": 2.9348677217840292e-05} +{"train_loss": 0.017322640866041183, "grad_norm": 0.06937238574028015, "global_step": 33683, "epoch": 317, "lr": 2.9345978095436767e-05} +{"train_loss": 0.015613728202879429, "grad_norm": 0.11476071178913116, "global_step": 33684, "epoch": 317, "lr": 2.9343279045602495e-05} +{"train_loss": 0.013498594984412193, "grad_norm": 0.06001697853207588, "global_step": 33685, "epoch": 317, "lr": 2.9340580068346946e-05} +{"train_loss": 0.013378542847931385, "grad_norm": 0.09029850363731384, "global_step": 33686, "epoch": 317, "lr": 2.9337881163679648e-05} +{"train_loss": 0.0113008301705122, "grad_norm": 0.06314588338136673, "global_step": 33687, "epoch": 317, "lr": 2.933518233161002e-05} +{"train_loss": 0.01783142425119877, "grad_norm": 0.06256205588579178, "global_step": 33688, "epoch": 317, "lr": 2.9332483572147585e-05} +{"train_loss": 0.014993151649832726, "grad_norm": 0.05857213959097862, "global_step": 33689, "epoch": 317, "lr": 2.932978488530181e-05} +{"train_loss": 0.013619967736303806, "grad_norm": 0.11593717336654663, "global_step": 33690, "epoch": 317, "lr": 2.9327086271082172e-05} +{"train_loss": 0.017269335687160492, "grad_norm": 0.06955503672361374, "global_step": 33691, "epoch": 317, "lr": 2.9324387729498175e-05} +{"train_loss": 0.016144685447216034, "grad_norm": 0.10874342173337936, "global_step": 33692, "epoch": 317, "lr": 2.93216892605593e-05} +{"train_loss": 0.01550050638616085, "grad_norm": 0.06102992594242096, "global_step": 33693, "epoch": 317, "lr": 2.9318990864275002e-05} +{"train_loss": 0.016151878982782364, "grad_norm": 0.08202846348285675, "global_step": 33694, "epoch": 317, "lr": 2.9316292540654765e-05} +{"train_loss": 0.015228810720145702, "grad_norm": 0.06224101781845093, "global_step": 33695, "epoch": 317, "lr": 2.9313594289708097e-05} +{"train_loss": 0.016860900446772575, "grad_norm": 0.10372914373874664, "global_step": 33696, "epoch": 317, "lr": 2.9310896111444457e-05} +{"train_loss": 0.016669806092977524, "grad_norm": 0.05557402968406677, "global_step": 33697, "epoch": 317, "lr": 2.9308198005873323e-05} +{"train_loss": 0.015887293964624405, "grad_norm": 0.06878157705068588, "global_step": 33698, "epoch": 317, "lr": 2.9305499973004212e-05} +{"train_loss": 0.015406159684062004, "grad_norm": 0.0629429891705513, "global_step": 33699, "epoch": 317, "lr": 2.930280201284654e-05} +{"train_loss": 0.014634929597377777, "grad_norm": 0.08048264682292938, "global_step": 33700, "epoch": 317, "lr": 2.9300104125409843e-05} +{"train_loss": 0.01505633257329464, "grad_norm": 0.1315208524465561, "global_step": 33701, "epoch": 317, "lr": 2.929740631070357e-05} +{"train_loss": 0.017625290900468826, "grad_norm": 0.17417611181735992, "global_step": 33702, "epoch": 317, "lr": 2.9294708568737194e-05} +{"train_loss": 0.018993202596902847, "grad_norm": 0.14348401129245758, "global_step": 33703, "epoch": 317, "lr": 2.9292010899520222e-05} +{"train_loss": 0.016495268791913986, "grad_norm": 0.059691883623600006, "global_step": 33704, "epoch": 317, "lr": 2.9289313303062127e-05} +{"train_loss": 0.0144206453114748, "grad_norm": 0.07450728118419647, "global_step": 33705, "epoch": 317, "lr": 2.928661577937237e-05} +{"train_loss": 0.015850737690925598, "grad_norm": 0.12080144882202148, "global_step": 33706, "epoch": 317, "lr": 2.9283918328460436e-05} +{"train_loss": 0.0152223491162624, "grad_norm": 0.09497914463281631, "global_step": 33707, "epoch": 317, "lr": 2.9281220950335796e-05, "val_loss": 0.019377393648028374} +{"train_loss": 0.013516167178750038, "grad_norm": 0.1255035698413849, "global_step": 33708, "epoch": 318, "lr": 2.927852364500794e-05} +{"train_loss": 0.014597034081816673, "grad_norm": 0.08721830695867538, "global_step": 33709, "epoch": 318, "lr": 2.9275826412486346e-05} +{"train_loss": 0.01525271125137806, "grad_norm": 0.1540905088186264, "global_step": 33710, "epoch": 318, "lr": 2.9273129252780495e-05} +{"train_loss": 0.014742955565452576, "grad_norm": 0.08631541579961777, "global_step": 33711, "epoch": 318, "lr": 2.927043216589983e-05} +{"train_loss": 0.013522831723093987, "grad_norm": 0.10177869349718094, "global_step": 33712, "epoch": 318, "lr": 2.9267735151853858e-05} +{"train_loss": 0.016150373965501785, "grad_norm": 0.05971662700176239, "global_step": 33713, "epoch": 318, "lr": 2.926503821065206e-05} +{"train_loss": 0.012406834401190281, "grad_norm": 0.06006448715925217, "global_step": 33714, "epoch": 318, "lr": 2.9262341342303878e-05} +{"train_loss": 0.015242332592606544, "grad_norm": 0.13694879412651062, "global_step": 33715, "epoch": 318, "lr": 2.9259644546818833e-05} +{"train_loss": 0.01343042403459549, "grad_norm": 0.06252002716064453, "global_step": 33716, "epoch": 318, "lr": 2.9256947824206378e-05} +{"train_loss": 0.016942981630563736, "grad_norm": 0.08326219022274017, "global_step": 33717, "epoch": 318, "lr": 2.9254251174475978e-05} +{"train_loss": 0.01332803349941969, "grad_norm": 0.09535514563322067, "global_step": 33718, "epoch": 318, "lr": 2.925155459763712e-05} +{"train_loss": 0.01592693291604519, "grad_norm": 0.056298334151506424, "global_step": 33719, "epoch": 318, "lr": 2.924885809369926e-05} +{"train_loss": 0.015600370243191719, "grad_norm": 0.1073579266667366, "global_step": 33720, "epoch": 318, "lr": 2.9246161662671904e-05} +{"train_loss": 0.01547105610370636, "grad_norm": 0.05585136637091637, "global_step": 33721, "epoch": 318, "lr": 2.924346530456451e-05} +{"train_loss": 0.01444606389850378, "grad_norm": 0.1420583724975586, "global_step": 33722, "epoch": 318, "lr": 2.924076901938656e-05} +{"train_loss": 0.013638908043503761, "grad_norm": 0.1168437972664833, "global_step": 33723, "epoch": 318, "lr": 2.9238072807147505e-05} +{"train_loss": 0.014009155333042145, "grad_norm": 0.08720552921295166, "global_step": 33724, "epoch": 318, "lr": 2.9235376667856838e-05} +{"train_loss": 0.013773753307759762, "grad_norm": 0.05947715789079666, "global_step": 33725, "epoch": 318, "lr": 2.9232680601524026e-05} +{"train_loss": 0.017081566154956818, "grad_norm": 0.12925103306770325, "global_step": 33726, "epoch": 318, "lr": 2.9229984608158532e-05} +{"train_loss": 0.013865340501070023, "grad_norm": 0.07159950584173203, "global_step": 33727, "epoch": 318, "lr": 2.922728868776986e-05} +{"train_loss": 0.014909802004694939, "grad_norm": 0.06889854371547699, "global_step": 33728, "epoch": 318, "lr": 2.9224592840367458e-05} +{"train_loss": 0.01580670475959778, "grad_norm": 0.06284992396831512, "global_step": 33729, "epoch": 318, "lr": 2.92218970659608e-05} +{"train_loss": 0.01665431633591652, "grad_norm": 0.1031990721821785, "global_step": 33730, "epoch": 318, "lr": 2.9219201364559358e-05} +{"train_loss": 0.01615051180124283, "grad_norm": 0.05289839208126068, "global_step": 33731, "epoch": 318, "lr": 2.9216505736172595e-05} +{"train_loss": 0.014648626558482647, "grad_norm": 0.060268547385931015, "global_step": 33732, "epoch": 318, "lr": 2.9213810180810003e-05} +{"train_loss": 0.015418488532304764, "grad_norm": 0.10199017822742462, "global_step": 33733, "epoch": 318, "lr": 2.921111469848105e-05} +{"train_loss": 0.013818200677633286, "grad_norm": 0.10750716924667358, "global_step": 33734, "epoch": 318, "lr": 2.9208419289195187e-05} +{"train_loss": 0.01624644361436367, "grad_norm": 0.13796059787273407, "global_step": 33735, "epoch": 318, "lr": 2.9205723952961905e-05} +{"train_loss": 0.01607048325240612, "grad_norm": 0.15420833230018616, "global_step": 33736, "epoch": 318, "lr": 2.9203028689790647e-05} +{"train_loss": 0.013884477317333221, "grad_norm": 0.059648118913173676, "global_step": 33737, "epoch": 318, "lr": 2.9200333499690918e-05} +{"train_loss": 0.014838455244898796, "grad_norm": 0.0719875916838646, "global_step": 33738, "epoch": 318, "lr": 2.919763838267216e-05} +{"train_loss": 0.015381868928670883, "grad_norm": 0.12354952096939087, "global_step": 33739, "epoch": 318, "lr": 2.9194943338743886e-05} +{"train_loss": 0.017524179071187973, "grad_norm": 0.07939194142818451, "global_step": 33740, "epoch": 318, "lr": 2.919224836791551e-05} +{"train_loss": 0.014456382021307945, "grad_norm": 0.10941452533006668, "global_step": 33741, "epoch": 318, "lr": 2.918955347019651e-05} +{"train_loss": 0.014210071414709091, "grad_norm": 0.0684194415807724, "global_step": 33742, "epoch": 318, "lr": 2.9186858645596378e-05} +{"train_loss": 0.013360200449824333, "grad_norm": 0.07907181978225708, "global_step": 33743, "epoch": 318, "lr": 2.9184163894124584e-05} +{"train_loss": 0.017369035631418228, "grad_norm": 0.07460018992424011, "global_step": 33744, "epoch": 318, "lr": 2.9181469215790568e-05} +{"train_loss": 0.016713667660951614, "grad_norm": 0.07281625270843506, "global_step": 33745, "epoch": 318, "lr": 2.9178774610603844e-05} +{"train_loss": 0.018344175070524216, "grad_norm": 0.0763808935880661, "global_step": 33746, "epoch": 318, "lr": 2.917608007857382e-05} +{"train_loss": 0.014601192437112331, "grad_norm": 0.11601374298334122, "global_step": 33747, "epoch": 318, "lr": 2.9173385619710026e-05} +{"train_loss": 0.015937019139528275, "grad_norm": 0.08424568921327591, "global_step": 33748, "epoch": 318, "lr": 2.9170691234021868e-05} +{"train_loss": 0.01408566813915968, "grad_norm": 0.14416050910949707, "global_step": 33749, "epoch": 318, "lr": 2.9167996921518848e-05} +{"train_loss": 0.014090567827224731, "grad_norm": 0.10821899771690369, "global_step": 33750, "epoch": 318, "lr": 2.9165302682210445e-05} +{"train_loss": 0.015031840652227402, "grad_norm": 0.08601967990398407, "global_step": 33751, "epoch": 318, "lr": 2.9162608516106105e-05} +{"train_loss": 0.017043163999915123, "grad_norm": 0.13876377046108246, "global_step": 33752, "epoch": 318, "lr": 2.915991442321528e-05} +{"train_loss": 0.01377861388027668, "grad_norm": 0.14040271937847137, "global_step": 33753, "epoch": 318, "lr": 2.915722040354747e-05} +{"train_loss": 0.017548291012644768, "grad_norm": 0.07446341961622238, "global_step": 33754, "epoch": 318, "lr": 2.915452645711211e-05} +{"train_loss": 0.016940750181674957, "grad_norm": 0.0837310329079628, "global_step": 33755, "epoch": 318, "lr": 2.915183258391867e-05} +{"train_loss": 0.01550772599875927, "grad_norm": 0.05839290842413902, "global_step": 33756, "epoch": 318, "lr": 2.914913878397665e-05} +{"train_loss": 0.01711111329495907, "grad_norm": 0.14180265367031097, "global_step": 33757, "epoch": 318, "lr": 2.914644505729548e-05} +{"train_loss": 0.017327746376395226, "grad_norm": 0.1251409649848938, "global_step": 33758, "epoch": 318, "lr": 2.914375140388462e-05} +{"train_loss": 0.01653178408741951, "grad_norm": 0.07157032191753387, "global_step": 33759, "epoch": 318, "lr": 2.9141057823753548e-05} +{"train_loss": 0.013778808526694775, "grad_norm": 0.17960090935230255, "global_step": 33760, "epoch": 318, "lr": 2.913836431691175e-05} +{"train_loss": 0.013398600742220879, "grad_norm": 0.0454150065779686, "global_step": 33761, "epoch": 318, "lr": 2.913567088336864e-05} +{"train_loss": 0.01734067127108574, "grad_norm": 0.0731535479426384, "global_step": 33762, "epoch": 318, "lr": 2.913297752313371e-05} +{"train_loss": 0.016611194238066673, "grad_norm": 0.10860767215490341, "global_step": 33763, "epoch": 318, "lr": 2.9130284236216443e-05} +{"train_loss": 0.014763634651899338, "grad_norm": 0.09067560732364655, "global_step": 33764, "epoch": 318, "lr": 2.9127591022626287e-05} +{"train_loss": 0.015419023111462593, "grad_norm": 0.10523824393749237, "global_step": 33765, "epoch": 318, "lr": 2.912489788237267e-05} +{"train_loss": 0.013144968077540398, "grad_norm": 0.12914176285266876, "global_step": 33766, "epoch": 318, "lr": 2.912220481546509e-05} +{"train_loss": 0.014215858653187752, "grad_norm": 0.08957512676715851, "global_step": 33767, "epoch": 318, "lr": 2.911951182191302e-05} +{"train_loss": 0.016522914171218872, "grad_norm": 0.08226198703050613, "global_step": 33768, "epoch": 318, "lr": 2.9116818901725883e-05} +{"train_loss": 0.014148790389299393, "grad_norm": 0.057361677289009094, "global_step": 33769, "epoch": 318, "lr": 2.9114126054913187e-05} +{"train_loss": 0.016080137342214584, "grad_norm": 0.10593491792678833, "global_step": 33770, "epoch": 318, "lr": 2.9111433281484367e-05} +{"train_loss": 0.015934575349092484, "grad_norm": 0.12342046946287155, "global_step": 33771, "epoch": 318, "lr": 2.9108740581448868e-05} +{"train_loss": 0.013343441300094128, "grad_norm": 0.056108471006155014, "global_step": 33772, "epoch": 318, "lr": 2.9106047954816163e-05} +{"train_loss": 0.014925897121429443, "grad_norm": 0.10026188939809799, "global_step": 33773, "epoch": 318, "lr": 2.9103355401595746e-05} +{"train_loss": 0.015241384506225586, "grad_norm": 0.09044428914785385, "global_step": 33774, "epoch": 318, "lr": 2.910066292179703e-05} +{"train_loss": 0.01439027301967144, "grad_norm": 0.08033677190542221, "global_step": 33775, "epoch": 318, "lr": 2.9097970515429512e-05} +{"train_loss": 0.016194790601730347, "grad_norm": 0.0912456288933754, "global_step": 33776, "epoch": 318, "lr": 2.9095278182502616e-05} +{"train_loss": 0.01647961139678955, "grad_norm": 0.10223118960857391, "global_step": 33777, "epoch": 318, "lr": 2.909258592302584e-05} +{"train_loss": 0.01552001666277647, "grad_norm": 0.15714959800243378, "global_step": 33778, "epoch": 318, "lr": 2.908989373700861e-05} +{"train_loss": 0.013781309127807617, "grad_norm": 0.13184943795204163, "global_step": 33779, "epoch": 318, "lr": 2.9087201624460398e-05} +{"train_loss": 0.014931073412299156, "grad_norm": 0.10835941880941391, "global_step": 33780, "epoch": 318, "lr": 2.9084509585390684e-05} +{"train_loss": 0.01704738475382328, "grad_norm": 0.08820988982915878, "global_step": 33781, "epoch": 318, "lr": 2.9081817619808905e-05} +{"train_loss": 0.014511387795209885, "grad_norm": 0.07139544934034348, "global_step": 33782, "epoch": 318, "lr": 2.9079125727724504e-05} +{"train_loss": 0.015505906194448471, "grad_norm": 0.13948458433151245, "global_step": 33783, "epoch": 318, "lr": 2.9076433909146956e-05} +{"train_loss": 0.014944205060601234, "grad_norm": 0.11772331595420837, "global_step": 33784, "epoch": 318, "lr": 2.907374216408574e-05} +{"train_loss": 0.015404464676976204, "grad_norm": 0.06806038320064545, "global_step": 33785, "epoch": 318, "lr": 2.907105049255028e-05} +{"train_loss": 0.01366511732339859, "grad_norm": 0.08185198158025742, "global_step": 33786, "epoch": 318, "lr": 2.9068358894550052e-05} +{"train_loss": 0.015551353804767132, "grad_norm": 0.09475799649953842, "global_step": 33787, "epoch": 318, "lr": 2.9065667370094497e-05} +{"train_loss": 0.012776949442923069, "grad_norm": 0.13287660479545593, "global_step": 33788, "epoch": 318, "lr": 2.90629759191931e-05} +{"train_loss": 0.01768600568175316, "grad_norm": 0.07205179333686829, "global_step": 33789, "epoch": 318, "lr": 2.9060284541855287e-05} +{"train_loss": 0.014233403839170933, "grad_norm": 0.10371389240026474, "global_step": 33790, "epoch": 318, "lr": 2.9057593238090525e-05} +{"train_loss": 0.013129785656929016, "grad_norm": 0.05187967047095299, "global_step": 33791, "epoch": 318, "lr": 2.905490200790829e-05} +{"train_loss": 0.01874285377562046, "grad_norm": 0.07170107215642929, "global_step": 33792, "epoch": 318, "lr": 2.905221085131802e-05} +{"train_loss": 0.014751903712749481, "grad_norm": 0.07440786063671112, "global_step": 33793, "epoch": 318, "lr": 2.9049519768329152e-05} +{"train_loss": 0.01501690223813057, "grad_norm": 0.12541693449020386, "global_step": 33794, "epoch": 318, "lr": 2.9046828758951185e-05} +{"train_loss": 0.014630753546953201, "grad_norm": 0.06885463744401932, "global_step": 33795, "epoch": 318, "lr": 2.9044137823193523e-05} +{"train_loss": 0.012701992876827717, "grad_norm": 0.0565013624727726, "global_step": 33796, "epoch": 318, "lr": 2.9041446961065654e-05} +{"train_loss": 0.014857977628707886, "grad_norm": 0.08299263566732407, "global_step": 33797, "epoch": 318, "lr": 2.903875617257704e-05} +{"train_loss": 0.016542669385671616, "grad_norm": 0.07814247161149979, "global_step": 33798, "epoch": 318, "lr": 2.9036065457737122e-05} +{"train_loss": 0.015558796934783459, "grad_norm": 0.04737284779548645, "global_step": 33799, "epoch": 318, "lr": 2.9033374816555338e-05} +{"train_loss": 0.014055624604225159, "grad_norm": 0.09438823163509369, "global_step": 33800, "epoch": 318, "lr": 2.903068424904115e-05} +{"train_loss": 0.016184192150831223, "grad_norm": 0.05984124168753624, "global_step": 33801, "epoch": 318, "lr": 2.902799375520404e-05} +{"train_loss": 0.016823848709464073, "grad_norm": 0.0647830069065094, "global_step": 33802, "epoch": 318, "lr": 2.9025303335053423e-05} +{"train_loss": 0.01674477383494377, "grad_norm": 0.11974137276411057, "global_step": 33803, "epoch": 318, "lr": 2.902261298859876e-05} +{"train_loss": 0.014565622434020042, "grad_norm": 0.1025555431842804, "global_step": 33804, "epoch": 318, "lr": 2.9019922715849536e-05} +{"train_loss": 0.013171891681849957, "grad_norm": 0.1280003935098648, "global_step": 33805, "epoch": 318, "lr": 2.9017232516815186e-05} +{"train_loss": 0.016948062926530838, "grad_norm": 0.1544678807258606, "global_step": 33806, "epoch": 318, "lr": 2.901454239150513e-05} +{"train_loss": 0.017953548580408096, "grad_norm": 0.0665324404835701, "global_step": 33807, "epoch": 318, "lr": 2.9011852339928848e-05} +{"train_loss": 0.015855465084314346, "grad_norm": 0.08005542308092117, "global_step": 33808, "epoch": 318, "lr": 2.9009162362095803e-05} +{"train_loss": 0.014239157550036907, "grad_norm": 0.07717667520046234, "global_step": 33809, "epoch": 318, "lr": 2.9006472458015422e-05} +{"train_loss": 0.014532726258039474, "grad_norm": 0.07712012529373169, "global_step": 33810, "epoch": 318, "lr": 2.9003782627697183e-05} +{"train_loss": 0.014892930164933205, "grad_norm": 0.1439049243927002, "global_step": 33811, "epoch": 318, "lr": 2.9001092871150514e-05} +{"train_loss": 0.014507509768009186, "grad_norm": 0.061814673244953156, "global_step": 33812, "epoch": 318, "lr": 2.8998403188384866e-05} +{"train_loss": 0.015215181938882143, "grad_norm": 0.13657987117767334, "global_step": 33813, "epoch": 318, "lr": 2.899571357940969e-05, "val_loss": 0.019475210458040237} +{"train_loss": 0.013062984682619572, "grad_norm": 0.07726309448480606, "global_step": 33814, "epoch": 319, "lr": 2.8993024044234457e-05} +{"train_loss": 0.016452839598059654, "grad_norm": 0.08449190109968185, "global_step": 33815, "epoch": 319, "lr": 2.8990334582868584e-05} +{"train_loss": 0.014833739958703518, "grad_norm": 0.09691468626260757, "global_step": 33816, "epoch": 319, "lr": 2.8987645195321566e-05} +{"train_loss": 0.01808813214302063, "grad_norm": 0.14721737802028656, "global_step": 33817, "epoch": 319, "lr": 2.8984955881602793e-05} +{"train_loss": 0.016185522079467773, "grad_norm": 0.09255204349756241, "global_step": 33818, "epoch": 319, "lr": 2.898226664172178e-05} +{"train_loss": 0.015095073729753494, "grad_norm": 0.07785381376743317, "global_step": 33819, "epoch": 319, "lr": 2.897957747568791e-05} +{"train_loss": 0.01253754273056984, "grad_norm": 0.06275546550750732, "global_step": 33820, "epoch": 319, "lr": 2.8976888383510665e-05} +{"train_loss": 0.019303657114505768, "grad_norm": 0.13725590705871582, "global_step": 33821, "epoch": 319, "lr": 2.897419936519951e-05} +{"train_loss": 0.016348857432603836, "grad_norm": 0.09018006920814514, "global_step": 33822, "epoch": 319, "lr": 2.897151042076388e-05} +{"train_loss": 0.01511449832469225, "grad_norm": 0.08681490272283554, "global_step": 33823, "epoch": 319, "lr": 2.89688215502132e-05} +{"train_loss": 0.015387879684567451, "grad_norm": 0.08426982164382935, "global_step": 33824, "epoch": 319, "lr": 2.8966132753556923e-05} +{"train_loss": 0.017028585076332092, "grad_norm": 0.06150156259536743, "global_step": 33825, "epoch": 319, "lr": 2.8963444030804537e-05} +{"train_loss": 0.014658021740615368, "grad_norm": 0.08269929885864258, "global_step": 33826, "epoch": 319, "lr": 2.8960755381965438e-05} +{"train_loss": 0.01584719493985176, "grad_norm": 0.0849432721734047, "global_step": 33827, "epoch": 319, "lr": 2.8958066807049096e-05} +{"train_loss": 0.012906977906823158, "grad_norm": 0.0811629593372345, "global_step": 33828, "epoch": 319, "lr": 2.8955378306064995e-05} +{"train_loss": 0.01422920636832714, "grad_norm": 0.06800738722085953, "global_step": 33829, "epoch": 319, "lr": 2.89526898790225e-05} +{"train_loss": 0.015367362648248672, "grad_norm": 0.08704254031181335, "global_step": 33830, "epoch": 319, "lr": 2.8950001525931104e-05} +{"train_loss": 0.014169087633490562, "grad_norm": 0.17292053997516632, "global_step": 33831, "epoch": 319, "lr": 2.8947313246800246e-05} +{"train_loss": 0.013501282781362534, "grad_norm": 0.064943827688694, "global_step": 33832, "epoch": 319, "lr": 2.8944625041639393e-05} +{"train_loss": 0.014865756034851074, "grad_norm": 0.09949790686368942, "global_step": 33833, "epoch": 319, "lr": 2.894193691045797e-05} +{"train_loss": 0.013536606915295124, "grad_norm": 0.10667228698730469, "global_step": 33834, "epoch": 319, "lr": 2.893924885326541e-05} +{"train_loss": 0.013672955334186554, "grad_norm": 0.059927210211753845, "global_step": 33835, "epoch": 319, "lr": 2.8936560870071184e-05} +{"train_loss": 0.015427237376570702, "grad_norm": 0.13239911198616028, "global_step": 33836, "epoch": 319, "lr": 2.8933872960884707e-05} +{"train_loss": 0.01548762060701847, "grad_norm": 0.1070876494050026, "global_step": 33837, "epoch": 319, "lr": 2.893118512571543e-05} +{"train_loss": 0.015760935842990875, "grad_norm": 0.064143106341362, "global_step": 33838, "epoch": 319, "lr": 2.8928497364572838e-05} +{"train_loss": 0.014732444658875465, "grad_norm": 0.10903871804475784, "global_step": 33839, "epoch": 319, "lr": 2.892580967746633e-05} +{"train_loss": 0.011894164606928825, "grad_norm": 0.04504638910293579, "global_step": 33840, "epoch": 319, "lr": 2.892312206440535e-05} +{"train_loss": 0.016326116397976875, "grad_norm": 0.0988645926117897, "global_step": 33841, "epoch": 319, "lr": 2.8920434525399353e-05} +{"train_loss": 0.013131742365658283, "grad_norm": 0.09505502879619598, "global_step": 33842, "epoch": 319, "lr": 2.89177470604578e-05} +{"train_loss": 0.01603800430893898, "grad_norm": 0.10895458608865738, "global_step": 33843, "epoch": 319, "lr": 2.8915059669590094e-05} +{"train_loss": 0.014306789264082909, "grad_norm": 0.09508350491523743, "global_step": 33844, "epoch": 319, "lr": 2.8912372352805707e-05} +{"train_loss": 0.015085531398653984, "grad_norm": 0.0633489117026329, "global_step": 33845, "epoch": 319, "lr": 2.890968511011408e-05} +{"train_loss": 0.017100267112255096, "grad_norm": 0.11446636915206909, "global_step": 33846, "epoch": 319, "lr": 2.890699794152466e-05} +{"train_loss": 0.015336329117417336, "grad_norm": 0.09678822010755539, "global_step": 33847, "epoch": 319, "lr": 2.8904310847046846e-05} +{"train_loss": 0.01587543450295925, "grad_norm": 0.08267673850059509, "global_step": 33848, "epoch": 319, "lr": 2.8901623826690117e-05} +{"train_loss": 0.014173794537782669, "grad_norm": 0.08636464923620224, "global_step": 33849, "epoch": 319, "lr": 2.8898936880463922e-05} +{"train_loss": 0.014737156219780445, "grad_norm": 0.07872116565704346, "global_step": 33850, "epoch": 319, "lr": 2.8896250008377667e-05} +{"train_loss": 0.013710624538362026, "grad_norm": 0.07435952872037888, "global_step": 33851, "epoch": 319, "lr": 2.8893563210440832e-05} +{"train_loss": 0.012299805879592896, "grad_norm": 0.06804698705673218, "global_step": 33852, "epoch": 319, "lr": 2.8890876486662833e-05} +{"train_loss": 0.016577323898673058, "grad_norm": 0.10619650781154633, "global_step": 33853, "epoch": 319, "lr": 2.88881898370531e-05} +{"train_loss": 0.014770790934562683, "grad_norm": 0.05742155387997627, "global_step": 33854, "epoch": 319, "lr": 2.8885503261621082e-05} +{"train_loss": 0.016351105645298958, "grad_norm": 0.1390051394701004, "global_step": 33855, "epoch": 319, "lr": 2.8882816760376247e-05} +{"train_loss": 0.012610487639904022, "grad_norm": 0.06271857768297195, "global_step": 33856, "epoch": 319, "lr": 2.8880130333327987e-05} +{"train_loss": 0.014991876669228077, "grad_norm": 0.08789452910423279, "global_step": 33857, "epoch": 319, "lr": 2.8877443980485787e-05} +{"train_loss": 0.017484216019511223, "grad_norm": 0.1295190304517746, "global_step": 33858, "epoch": 319, "lr": 2.887475770185904e-05} +{"train_loss": 0.016421779990196228, "grad_norm": 0.1002676859498024, "global_step": 33859, "epoch": 319, "lr": 2.887207149745723e-05} +{"train_loss": 0.015422276221215725, "grad_norm": 0.1258249580860138, "global_step": 33860, "epoch": 319, "lr": 2.8869385367289746e-05} +{"train_loss": 0.017354890704154968, "grad_norm": 0.07201837748289108, "global_step": 33861, "epoch": 319, "lr": 2.8866699311366062e-05} +{"train_loss": 0.01399395801126957, "grad_norm": 0.06739748269319534, "global_step": 33862, "epoch": 319, "lr": 2.886401332969562e-05} +{"train_loss": 0.01569889672100544, "grad_norm": 0.10719186812639236, "global_step": 33863, "epoch": 319, "lr": 2.8861327422287847e-05} +{"train_loss": 0.01662357896566391, "grad_norm": 0.06900671124458313, "global_step": 33864, "epoch": 319, "lr": 2.8858641589152158e-05} +{"train_loss": 0.013170403428375721, "grad_norm": 0.11374470591545105, "global_step": 33865, "epoch": 319, "lr": 2.8855955830298e-05} +{"train_loss": 0.013623920269310474, "grad_norm": 0.14890119433403015, "global_step": 33866, "epoch": 319, "lr": 2.8853270145734846e-05} +{"train_loss": 0.012287598103284836, "grad_norm": 0.11878199130296707, "global_step": 33867, "epoch": 319, "lr": 2.8850584535472082e-05} +{"train_loss": 0.01395510695874691, "grad_norm": 0.12688004970550537, "global_step": 33868, "epoch": 319, "lr": 2.8847898999519162e-05} +{"train_loss": 0.015658827498555183, "grad_norm": 0.08876528590917587, "global_step": 33869, "epoch": 319, "lr": 2.8845213537885574e-05} +{"train_loss": 0.016356028616428375, "grad_norm": 0.10062750428915024, "global_step": 33870, "epoch": 319, "lr": 2.8842528150580663e-05} +{"train_loss": 0.015408569946885109, "grad_norm": 0.08211371302604675, "global_step": 33871, "epoch": 319, "lr": 2.88398428376139e-05} +{"train_loss": 0.01587188057601452, "grad_norm": 0.10322772711515427, "global_step": 33872, "epoch": 319, "lr": 2.8837157598994752e-05} +{"train_loss": 0.014758305624127388, "grad_norm": 0.05367310345172882, "global_step": 33873, "epoch": 319, "lr": 2.8834472434732606e-05} +{"train_loss": 0.013066471554338932, "grad_norm": 0.10058112442493439, "global_step": 33874, "epoch": 319, "lr": 2.8831787344836923e-05} +{"train_loss": 0.014902925118803978, "grad_norm": 0.07784555852413177, "global_step": 33875, "epoch": 319, "lr": 2.8829102329317154e-05} +{"train_loss": 0.014985128305852413, "grad_norm": 0.11913259327411652, "global_step": 33876, "epoch": 319, "lr": 2.8826417388182713e-05} +{"train_loss": 0.015332557260990143, "grad_norm": 0.06294350326061249, "global_step": 33877, "epoch": 319, "lr": 2.8823732521443007e-05} +{"train_loss": 0.017208322882652283, "grad_norm": 0.07786642760038376, "global_step": 33878, "epoch": 319, "lr": 2.88210477291075e-05} +{"train_loss": 0.017078962177038193, "grad_norm": 0.06510552763938904, "global_step": 33879, "epoch": 319, "lr": 2.8818363011185647e-05} +{"train_loss": 0.017129819840192795, "grad_norm": 0.06393932551145554, "global_step": 33880, "epoch": 319, "lr": 2.8815678367686827e-05} +{"train_loss": 0.01351897232234478, "grad_norm": 0.07679557055234909, "global_step": 33881, "epoch": 319, "lr": 2.8812993798620512e-05} +{"train_loss": 0.014317646622657776, "grad_norm": 0.08379596471786499, "global_step": 33882, "epoch": 319, "lr": 2.881030930399612e-05} +{"train_loss": 0.016490794718265533, "grad_norm": 0.07666191458702087, "global_step": 33883, "epoch": 319, "lr": 2.8807624883823092e-05} +{"train_loss": 0.01563386432826519, "grad_norm": 0.06871876120567322, "global_step": 33884, "epoch": 319, "lr": 2.8804940538110848e-05} +{"train_loss": 0.013465704396367073, "grad_norm": 0.15525813400745392, "global_step": 33885, "epoch": 319, "lr": 2.880225626686881e-05} +{"train_loss": 0.013877132907509804, "grad_norm": 0.09157954156398773, "global_step": 33886, "epoch": 319, "lr": 2.8799572070106455e-05} +{"train_loss": 0.015236922539770603, "grad_norm": 0.0966096892952919, "global_step": 33887, "epoch": 319, "lr": 2.8796887947833185e-05} +{"train_loss": 0.017332132905721664, "grad_norm": 0.08423054218292236, "global_step": 33888, "epoch": 319, "lr": 2.8794203900058403e-05} +{"train_loss": 0.015451361425220966, "grad_norm": 0.10846945643424988, "global_step": 33889, "epoch": 319, "lr": 2.8791519926791564e-05} +{"train_loss": 0.017102468758821487, "grad_norm": 0.07506068795919418, "global_step": 33890, "epoch": 319, "lr": 2.8788836028042133e-05} +{"train_loss": 0.013368595391511917, "grad_norm": 0.08373203128576279, "global_step": 33891, "epoch": 319, "lr": 2.8786152203819482e-05} +{"train_loss": 0.014401332475244999, "grad_norm": 0.12702849507331848, "global_step": 33892, "epoch": 319, "lr": 2.8783468454133085e-05} +{"train_loss": 0.012643747963011265, "grad_norm": 0.07896319031715393, "global_step": 33893, "epoch": 319, "lr": 2.8780784778992353e-05} +{"train_loss": 0.014329015277326107, "grad_norm": 0.07051947712898254, "global_step": 33894, "epoch": 319, "lr": 2.877810117840669e-05} +{"train_loss": 0.016649279743433, "grad_norm": 0.06668378412723541, "global_step": 33895, "epoch": 319, "lr": 2.877541765238556e-05} +{"train_loss": 0.017435302957892418, "grad_norm": 0.10805987566709518, "global_step": 33896, "epoch": 319, "lr": 2.87727342009384e-05} +{"train_loss": 0.01496540755033493, "grad_norm": 0.07626928389072418, "global_step": 33897, "epoch": 319, "lr": 2.8770050824074594e-05} +{"train_loss": 0.016250284388661385, "grad_norm": 0.1049438863992691, "global_step": 33898, "epoch": 319, "lr": 2.8767367521803624e-05} +{"train_loss": 0.014538979157805443, "grad_norm": 0.1044292151927948, "global_step": 33899, "epoch": 319, "lr": 2.876468429413487e-05} +{"train_loss": 0.01742125302553177, "grad_norm": 0.0668497234582901, "global_step": 33900, "epoch": 319, "lr": 2.87620011410778e-05} +{"train_loss": 0.015241682529449463, "grad_norm": 0.06690964102745056, "global_step": 33901, "epoch": 319, "lr": 2.8759318062641793e-05} +{"train_loss": 0.018138539046049118, "grad_norm": 0.09532489627599716, "global_step": 33902, "epoch": 319, "lr": 2.8756635058836312e-05} +{"train_loss": 0.013415826484560966, "grad_norm": 0.07825880497694016, "global_step": 33903, "epoch": 319, "lr": 2.87539521296708e-05} +{"train_loss": 0.013108902610838413, "grad_norm": 0.05356425791978836, "global_step": 33904, "epoch": 319, "lr": 2.8751269275154657e-05} +{"train_loss": 0.01730107143521309, "grad_norm": 0.07309578359127045, "global_step": 33905, "epoch": 319, "lr": 2.874858649529729e-05} +{"train_loss": 0.014385158196091652, "grad_norm": 0.054920945316553116, "global_step": 33906, "epoch": 319, "lr": 2.8745903790108152e-05} +{"train_loss": 0.014859341084957123, "grad_norm": 0.10093231499195099, "global_step": 33907, "epoch": 319, "lr": 2.8743221159596688e-05} +{"train_loss": 0.015099737793207169, "grad_norm": 0.12265948951244354, "global_step": 33908, "epoch": 319, "lr": 2.8740538603772275e-05} +{"train_loss": 0.015415194444358349, "grad_norm": 0.13014782965183258, "global_step": 33909, "epoch": 319, "lr": 2.8737856122644368e-05} +{"train_loss": 0.01786145009100437, "grad_norm": 0.06689757108688354, "global_step": 33910, "epoch": 319, "lr": 2.873517371622243e-05} +{"train_loss": 0.013771571218967438, "grad_norm": 0.13343723118305206, "global_step": 33911, "epoch": 319, "lr": 2.8732491384515792e-05} +{"train_loss": 0.017225533723831177, "grad_norm": 0.08063244074583054, "global_step": 33912, "epoch": 319, "lr": 2.872980912753394e-05} +{"train_loss": 0.014828582294285297, "grad_norm": 0.07849116623401642, "global_step": 33913, "epoch": 319, "lr": 2.872712694528631e-05} +{"train_loss": 0.01616561785340309, "grad_norm": 0.0965508297085762, "global_step": 33914, "epoch": 319, "lr": 2.8724444837782276e-05} +{"train_loss": 0.013770808465778828, "grad_norm": 0.07807616144418716, "global_step": 33915, "epoch": 319, "lr": 2.8721762805031293e-05} +{"train_loss": 0.01736891269683838, "grad_norm": 0.08736489713191986, "global_step": 33916, "epoch": 319, "lr": 2.8719080847042813e-05} +{"train_loss": 0.015586130321025848, "grad_norm": 0.09186538308858871, "global_step": 33917, "epoch": 319, "lr": 2.871639896382622e-05} +{"train_loss": 0.015432214364409447, "grad_norm": 0.0700756385922432, "global_step": 33918, "epoch": 319, "lr": 2.8713717155390917e-05} +{"train_loss": 0.01520779914275374, "grad_norm": 0.1134834885597229, "global_step": 33919, "epoch": 319, "lr": 2.8711035421746367e-05, "val_loss": 0.019435495138168335} +{"train_loss": 0.01481652446091175, "grad_norm": 0.1343894898891449, "global_step": 33920, "epoch": 320, "lr": 2.8708353762901995e-05} +{"train_loss": 0.016837555915117264, "grad_norm": 0.0828360840678215, "global_step": 33921, "epoch": 320, "lr": 2.8705672178867184e-05} +{"train_loss": 0.01695612445473671, "grad_norm": 0.07577609270811081, "global_step": 33922, "epoch": 320, "lr": 2.870299066965141e-05} +{"train_loss": 0.01673426479101181, "grad_norm": 0.06876558065414429, "global_step": 33923, "epoch": 320, "lr": 2.8700309235264033e-05} +{"train_loss": 0.015515845268964767, "grad_norm": 0.08511392772197723, "global_step": 33924, "epoch": 320, "lr": 2.8697627875714538e-05} +{"train_loss": 0.016571789979934692, "grad_norm": 0.09097883105278015, "global_step": 33925, "epoch": 320, "lr": 2.869494659101229e-05} +{"train_loss": 0.01564783602952957, "grad_norm": 0.08677470684051514, "global_step": 33926, "epoch": 320, "lr": 2.8692265381166734e-05} +{"train_loss": 0.018522974103689194, "grad_norm": 0.10264692455530167, "global_step": 33927, "epoch": 320, "lr": 2.8689584246187307e-05} +{"train_loss": 0.01512141153216362, "grad_norm": 0.11606311798095703, "global_step": 33928, "epoch": 320, "lr": 2.8686903186083424e-05} +{"train_loss": 0.014760457910597324, "grad_norm": 0.07011550664901733, "global_step": 33929, "epoch": 320, "lr": 2.8684222200864464e-05} +{"train_loss": 0.012532094493508339, "grad_norm": 0.12913425266742706, "global_step": 33930, "epoch": 320, "lr": 2.8681541290539902e-05} +{"train_loss": 0.015059830620884895, "grad_norm": 0.06903993338346481, "global_step": 33931, "epoch": 320, "lr": 2.867886045511911e-05} +{"train_loss": 0.01875293254852295, "grad_norm": 0.08913233876228333, "global_step": 33932, "epoch": 320, "lr": 2.867617969461153e-05} +{"train_loss": 0.01595322974026203, "grad_norm": 0.1019599661231041, "global_step": 33933, "epoch": 320, "lr": 2.8673499009026583e-05} +{"train_loss": 0.015159687958657742, "grad_norm": 0.05443849414587021, "global_step": 33934, "epoch": 320, "lr": 2.8670818398373726e-05} +{"train_loss": 0.014260749332606792, "grad_norm": 0.06187254562973976, "global_step": 33935, "epoch": 320, "lr": 2.8668137862662303e-05} +{"train_loss": 0.015579490922391415, "grad_norm": 0.07409980148077011, "global_step": 33936, "epoch": 320, "lr": 2.866545740190176e-05} +{"train_loss": 0.014113621786236763, "grad_norm": 0.055215783417224884, "global_step": 33937, "epoch": 320, "lr": 2.8662777016101537e-05} +{"train_loss": 0.014298027381300926, "grad_norm": 0.06131797656416893, "global_step": 33938, "epoch": 320, "lr": 2.8660096705271022e-05} +{"train_loss": 0.015498567372560501, "grad_norm": 0.0713452473282814, "global_step": 33939, "epoch": 320, "lr": 2.8657416469419667e-05} +{"train_loss": 0.016906097531318665, "grad_norm": 0.1208358108997345, "global_step": 33940, "epoch": 320, "lr": 2.865473630855685e-05} +{"train_loss": 0.014368969947099686, "grad_norm": 0.07029528170824051, "global_step": 33941, "epoch": 320, "lr": 2.8652056222692025e-05} +{"train_loss": 0.015580005943775177, "grad_norm": 0.06399097293615341, "global_step": 33942, "epoch": 320, "lr": 2.8649376211834566e-05} +{"train_loss": 0.015284307301044464, "grad_norm": 0.08821991831064224, "global_step": 33943, "epoch": 320, "lr": 2.8646696275993923e-05} +{"train_loss": 0.014947878196835518, "grad_norm": 0.062554270029068, "global_step": 33944, "epoch": 320, "lr": 2.8644016415179514e-05} +{"train_loss": 0.016135049983859062, "grad_norm": 0.12633608281612396, "global_step": 33945, "epoch": 320, "lr": 2.864133662940075e-05} +{"train_loss": 0.013934871181845665, "grad_norm": 0.07570495456457138, "global_step": 33946, "epoch": 320, "lr": 2.863865691866702e-05} +{"train_loss": 0.014689748175442219, "grad_norm": 0.07812780141830444, "global_step": 33947, "epoch": 320, "lr": 2.8635977282987757e-05} +{"train_loss": 0.015257326886057854, "grad_norm": 0.07702655345201492, "global_step": 33948, "epoch": 320, "lr": 2.86332977223724e-05} +{"train_loss": 0.015754876658320427, "grad_norm": 0.08607576042413712, "global_step": 33949, "epoch": 320, "lr": 2.863061823683032e-05} +{"train_loss": 0.01433723047375679, "grad_norm": 0.06898089498281479, "global_step": 33950, "epoch": 320, "lr": 2.862793882637096e-05} +{"train_loss": 0.014884434640407562, "grad_norm": 0.11929915100336075, "global_step": 33951, "epoch": 320, "lr": 2.8625259491003766e-05} +{"train_loss": 0.011113937944173813, "grad_norm": 0.07418663054704666, "global_step": 33952, "epoch": 320, "lr": 2.862258023073807e-05} +{"train_loss": 0.01671326532959938, "grad_norm": 0.07883753627538681, "global_step": 33953, "epoch": 320, "lr": 2.8619901045583335e-05} +{"train_loss": 0.0158353503793478, "grad_norm": 0.12800025939941406, "global_step": 33954, "epoch": 320, "lr": 2.8617221935548986e-05} +{"train_loss": 0.012637508101761341, "grad_norm": 0.08176559954881668, "global_step": 33955, "epoch": 320, "lr": 2.8614542900644404e-05} +{"train_loss": 0.013046549633145332, "grad_norm": 0.11186870187520981, "global_step": 33956, "epoch": 320, "lr": 2.861186394087902e-05} +{"train_loss": 0.013752065598964691, "grad_norm": 0.0574185773730278, "global_step": 33957, "epoch": 320, "lr": 2.8609185056262262e-05} +{"train_loss": 0.014343425631523132, "grad_norm": 0.07029501348733902, "global_step": 33958, "epoch": 320, "lr": 2.860650624680352e-05} +{"train_loss": 0.013927764259278774, "grad_norm": 0.0856766328215599, "global_step": 33959, "epoch": 320, "lr": 2.86038275125122e-05} +{"train_loss": 0.01352862548083067, "grad_norm": 0.0886639878153801, "global_step": 33960, "epoch": 320, "lr": 2.860114885339772e-05} +{"train_loss": 0.01230650581419468, "grad_norm": 0.06561267375946045, "global_step": 33961, "epoch": 320, "lr": 2.8598470269469517e-05} +{"train_loss": 0.015999585390090942, "grad_norm": 0.06142749637365341, "global_step": 33962, "epoch": 320, "lr": 2.8595791760736963e-05} +{"train_loss": 0.015316754579544067, "grad_norm": 0.08867080509662628, "global_step": 33963, "epoch": 320, "lr": 2.859311332720951e-05} +{"train_loss": 0.012501295655965805, "grad_norm": 0.08548663556575775, "global_step": 33964, "epoch": 320, "lr": 2.8590434968896522e-05} +{"train_loss": 0.013901591300964355, "grad_norm": 0.08173169940710068, "global_step": 33965, "epoch": 320, "lr": 2.858775668580746e-05} +{"train_loss": 0.013667618855834007, "grad_norm": 0.0634414330124855, "global_step": 33966, "epoch": 320, "lr": 2.8585078477951688e-05} +{"train_loss": 0.016643818467855453, "grad_norm": 0.09625453501939774, "global_step": 33967, "epoch": 320, "lr": 2.8582400345338632e-05} +{"train_loss": 0.014383126981556416, "grad_norm": 0.08451397716999054, "global_step": 33968, "epoch": 320, "lr": 2.8579722287977738e-05} +{"train_loss": 0.014254575595259666, "grad_norm": 0.0693572387099266, "global_step": 33969, "epoch": 320, "lr": 2.8577044305878374e-05} +{"train_loss": 0.014202550053596497, "grad_norm": 0.102216437458992, "global_step": 33970, "epoch": 320, "lr": 2.8574366399049946e-05} +{"train_loss": 0.014428619295358658, "grad_norm": 0.07299862056970596, "global_step": 33971, "epoch": 320, "lr": 2.8571688567501893e-05} +{"train_loss": 0.015125640667974949, "grad_norm": 0.06114359200000763, "global_step": 33972, "epoch": 320, "lr": 2.856901081124359e-05} +{"train_loss": 0.01579221338033676, "grad_norm": 0.08809445798397064, "global_step": 33973, "epoch": 320, "lr": 2.856633313028446e-05} +{"train_loss": 0.016497567296028137, "grad_norm": 0.1419367790222168, "global_step": 33974, "epoch": 320, "lr": 2.8563655524633938e-05} +{"train_loss": 0.01688385382294655, "grad_norm": 0.08668526262044907, "global_step": 33975, "epoch": 320, "lr": 2.856097799430141e-05} +{"train_loss": 0.01618095487356186, "grad_norm": 0.11341656744480133, "global_step": 33976, "epoch": 320, "lr": 2.8558300539296252e-05} +{"train_loss": 0.015144981443881989, "grad_norm": 0.07908040285110474, "global_step": 33977, "epoch": 320, "lr": 2.855562315962791e-05} +{"train_loss": 0.0162937194108963, "grad_norm": 0.08605590462684631, "global_step": 33978, "epoch": 320, "lr": 2.8552945855305795e-05} +{"train_loss": 0.014202783815562725, "grad_norm": 0.10487062484025955, "global_step": 33979, "epoch": 320, "lr": 2.855026862633928e-05} +{"train_loss": 0.015483038499951363, "grad_norm": 0.06750810146331787, "global_step": 33980, "epoch": 320, "lr": 2.8547591472737802e-05} +{"train_loss": 0.014502748847007751, "grad_norm": 0.054406505078077316, "global_step": 33981, "epoch": 320, "lr": 2.8544914394510776e-05} +{"train_loss": 0.016231650486588478, "grad_norm": 0.11205954849720001, "global_step": 33982, "epoch": 320, "lr": 2.8542237391667588e-05} +{"train_loss": 0.01804422214627266, "grad_norm": 0.07588327676057816, "global_step": 33983, "epoch": 320, "lr": 2.853956046421763e-05} +{"train_loss": 0.01579337567090988, "grad_norm": 0.09294334053993225, "global_step": 33984, "epoch": 320, "lr": 2.8536883612170316e-05} +{"train_loss": 0.013879185542464256, "grad_norm": 0.0574180968105793, "global_step": 33985, "epoch": 320, "lr": 2.8534206835535083e-05} +{"train_loss": 0.014872186817228794, "grad_norm": 0.07498816400766373, "global_step": 33986, "epoch": 320, "lr": 2.8531530134321317e-05} +{"train_loss": 0.01804187335073948, "grad_norm": 0.0727737545967102, "global_step": 33987, "epoch": 320, "lr": 2.852885350853839e-05} +{"train_loss": 0.016283687204122543, "grad_norm": 0.09603538364171982, "global_step": 33988, "epoch": 320, "lr": 2.8526176958195738e-05} +{"train_loss": 0.013826572336256504, "grad_norm": 0.054177381098270416, "global_step": 33989, "epoch": 320, "lr": 2.852350048330278e-05} +{"train_loss": 0.013365967199206352, "grad_norm": 0.09785928577184677, "global_step": 33990, "epoch": 320, "lr": 2.852082408386888e-05} +{"train_loss": 0.014621911570429802, "grad_norm": 0.07229974865913391, "global_step": 33991, "epoch": 320, "lr": 2.851814775990346e-05} +{"train_loss": 0.014307286590337753, "grad_norm": 0.13010458648204803, "global_step": 33992, "epoch": 320, "lr": 2.8515471511415975e-05} +{"train_loss": 0.016975725069642067, "grad_norm": 0.09793467074632645, "global_step": 33993, "epoch": 320, "lr": 2.8512795338415736e-05} +{"train_loss": 0.014868451282382011, "grad_norm": 0.08355266600847244, "global_step": 33994, "epoch": 320, "lr": 2.851011924091219e-05} +{"train_loss": 0.018259014934301376, "grad_norm": 0.08086495101451874, "global_step": 33995, "epoch": 320, "lr": 2.850744321891476e-05} +{"train_loss": 0.014859084971249104, "grad_norm": 0.06428319960832596, "global_step": 33996, "epoch": 320, "lr": 2.8504767272432808e-05} +{"train_loss": 0.015393109992146492, "grad_norm": 0.1023629829287529, "global_step": 33997, "epoch": 320, "lr": 2.850209140147575e-05} +{"train_loss": 0.014901923015713692, "grad_norm": 0.1403742879629135, "global_step": 33998, "epoch": 320, "lr": 2.849941560605302e-05} +{"train_loss": 0.012227216735482216, "grad_norm": 0.05952265113592148, "global_step": 33999, "epoch": 320, "lr": 2.8496739886173995e-05} +{"train_loss": 0.01399952918291092, "grad_norm": 0.07479821145534515, "global_step": 34000, "epoch": 320, "lr": 2.849406424184805e-05} +{"train_loss": 0.014646993950009346, "grad_norm": 0.11042828857898712, "global_step": 34001, "epoch": 320, "lr": 2.8491388673084608e-05} +{"train_loss": 0.01799726113677025, "grad_norm": 0.10150448232889175, "global_step": 34002, "epoch": 320, "lr": 2.84887131798931e-05} +{"train_loss": 0.014963861554861069, "grad_norm": 0.14878042042255402, "global_step": 34003, "epoch": 320, "lr": 2.848603776228287e-05} +{"train_loss": 0.014969650655984879, "grad_norm": 0.08728639781475067, "global_step": 34004, "epoch": 320, "lr": 2.848336242026337e-05} +{"train_loss": 0.018777722492814064, "grad_norm": 0.1311585158109665, "global_step": 34005, "epoch": 320, "lr": 2.8480687153843966e-05} +{"train_loss": 0.0164153054356575, "grad_norm": 0.10998639464378357, "global_step": 34006, "epoch": 320, "lr": 2.847801196303408e-05} +{"train_loss": 0.015170015394687653, "grad_norm": 0.06352350115776062, "global_step": 34007, "epoch": 320, "lr": 2.847533684784308e-05} +{"train_loss": 0.01652638614177704, "grad_norm": 0.06583432853221893, "global_step": 34008, "epoch": 320, "lr": 2.8472661808280386e-05} +{"train_loss": 0.01828182116150856, "grad_norm": 0.07174843549728394, "global_step": 34009, "epoch": 320, "lr": 2.8469986844355424e-05} +{"train_loss": 0.016077952459454536, "grad_norm": 0.08218749612569809, "global_step": 34010, "epoch": 320, "lr": 2.8467311956077558e-05} +{"train_loss": 0.014995193108916283, "grad_norm": 0.07781101763248444, "global_step": 34011, "epoch": 320, "lr": 2.8464637143456173e-05} +{"train_loss": 0.014699508436024189, "grad_norm": 0.07128722965717316, "global_step": 34012, "epoch": 320, "lr": 2.846196240650071e-05} +{"train_loss": 0.01523665338754654, "grad_norm": 0.11005240678787231, "global_step": 34013, "epoch": 320, "lr": 2.845928774522052e-05} +{"train_loss": 0.016277890652418137, "grad_norm": 0.1097007542848587, "global_step": 34014, "epoch": 320, "lr": 2.8456613159625033e-05} +{"train_loss": 0.015876198187470436, "grad_norm": 0.1233079582452774, "global_step": 34015, "epoch": 320, "lr": 2.8453938649723655e-05} +{"train_loss": 0.016962910071015358, "grad_norm": 0.11152246594429016, "global_step": 34016, "epoch": 320, "lr": 2.8451264215525763e-05} +{"train_loss": 0.01516675390303135, "grad_norm": 0.08612894266843796, "global_step": 34017, "epoch": 320, "lr": 2.844858985704074e-05} +{"train_loss": 0.013254953548312187, "grad_norm": 0.06262017041444778, "global_step": 34018, "epoch": 320, "lr": 2.8445915574277997e-05} +{"train_loss": 0.014624172821640968, "grad_norm": 0.06292332708835602, "global_step": 34019, "epoch": 320, "lr": 2.8443241367246955e-05} +{"train_loss": 0.013115230947732925, "grad_norm": 0.08708181977272034, "global_step": 34020, "epoch": 320, "lr": 2.8440567235956972e-05} +{"train_loss": 0.016802744939923286, "grad_norm": 0.061524175107479095, "global_step": 34021, "epoch": 320, "lr": 2.843789318041745e-05} +{"train_loss": 0.011434114538133144, "grad_norm": 0.12323994934558868, "global_step": 34022, "epoch": 320, "lr": 2.8435219200637826e-05} +{"train_loss": 0.014325480908155441, "grad_norm": 0.09532592445611954, "global_step": 34023, "epoch": 320, "lr": 2.8432545296627454e-05} +{"train_loss": 0.014846909791231155, "grad_norm": 0.0518578365445137, "global_step": 34024, "epoch": 320, "lr": 2.842987146839573e-05} +{"train_loss": 0.01518242547007383, "grad_norm": 0.09065379947423935, "global_step": 34025, "epoch": 320, "lr": 2.8427197715952047e-05, "train_batch_mse_error": 0.013471435755491257, "val_loss": 0.01929120533168316} +{"train_loss": 0.016392480581998825, "grad_norm": 0.09461425989866257, "global_step": 34026, "epoch": 321, "lr": 2.8424524039305833e-05} +{"train_loss": 0.012733016163110733, "grad_norm": 0.08845475316047668, "global_step": 34027, "epoch": 321, "lr": 2.8421850438466435e-05} +{"train_loss": 0.014572909101843834, "grad_norm": 0.0576227605342865, "global_step": 34028, "epoch": 321, "lr": 2.841917691344329e-05} +{"train_loss": 0.015647809952497482, "grad_norm": 0.11139015853404999, "global_step": 34029, "epoch": 321, "lr": 2.841650346424578e-05} +{"train_loss": 0.014696658588945866, "grad_norm": 0.07437316328287125, "global_step": 34030, "epoch": 321, "lr": 2.8413830090883265e-05} +{"train_loss": 0.015329413115978241, "grad_norm": 0.07017015665769577, "global_step": 34031, "epoch": 321, "lr": 2.8411156793365163e-05} +{"train_loss": 0.013451507315039635, "grad_norm": 0.08286210894584656, "global_step": 34032, "epoch": 321, "lr": 2.8408483571700887e-05} +{"train_loss": 0.01580105349421501, "grad_norm": 0.14008617401123047, "global_step": 34033, "epoch": 321, "lr": 2.8405810425899793e-05} +{"train_loss": 0.012701828964054585, "grad_norm": 0.06447310000658035, "global_step": 34034, "epoch": 321, "lr": 2.840313735597131e-05} +{"train_loss": 0.014152787625789642, "grad_norm": 0.05545060709118843, "global_step": 34035, "epoch": 321, "lr": 2.8400464361924783e-05} +{"train_loss": 0.01508847065269947, "grad_norm": 0.08005297929048538, "global_step": 34036, "epoch": 321, "lr": 2.839779144376966e-05} +{"train_loss": 0.01626971922814846, "grad_norm": 0.057374902069568634, "global_step": 34037, "epoch": 321, "lr": 2.8395118601515282e-05} +{"train_loss": 0.01602310687303543, "grad_norm": 0.09770059585571289, "global_step": 34038, "epoch": 321, "lr": 2.8392445835171056e-05} +{"train_loss": 0.014957694336771965, "grad_norm": 0.0828540101647377, "global_step": 34039, "epoch": 321, "lr": 2.83897731447464e-05} +{"train_loss": 0.014132082462310791, "grad_norm": 0.055986158549785614, "global_step": 34040, "epoch": 321, "lr": 2.8387100530250687e-05} +{"train_loss": 0.01662338897585869, "grad_norm": 0.0686783418059349, "global_step": 34041, "epoch": 321, "lr": 2.838442799169328e-05} +{"train_loss": 0.013839426450431347, "grad_norm": 0.07013247162103653, "global_step": 34042, "epoch": 321, "lr": 2.83817555290836e-05} +{"train_loss": 0.019005868583917618, "grad_norm": 0.08359165489673615, "global_step": 34043, "epoch": 321, "lr": 2.837908314243104e-05} +{"train_loss": 0.015462762676179409, "grad_norm": 0.06882152706384659, "global_step": 34044, "epoch": 321, "lr": 2.837641083174496e-05} +{"train_loss": 0.016482342034578323, "grad_norm": 0.09275992214679718, "global_step": 34045, "epoch": 321, "lr": 2.8373738597034794e-05} +{"train_loss": 0.0190140288323164, "grad_norm": 0.054802410304546356, "global_step": 34046, "epoch": 321, "lr": 2.8371066438309877e-05} +{"train_loss": 0.01634664088487625, "grad_norm": 0.05631029233336449, "global_step": 34047, "epoch": 321, "lr": 2.8368394355579652e-05} +{"train_loss": 0.014238769188523293, "grad_norm": 0.10793628543615341, "global_step": 34048, "epoch": 321, "lr": 2.8365722348853462e-05} +{"train_loss": 0.015332306735217571, "grad_norm": 0.08395466953516006, "global_step": 34049, "epoch": 321, "lr": 2.8363050418140717e-05} +{"train_loss": 0.015733353793621063, "grad_norm": 0.07865191996097565, "global_step": 34050, "epoch": 321, "lr": 2.836037856345082e-05} +{"train_loss": 0.015664245933294296, "grad_norm": 0.11498081684112549, "global_step": 34051, "epoch": 321, "lr": 2.8357706784793147e-05} +{"train_loss": 0.015066903084516525, "grad_norm": 0.08275476098060608, "global_step": 34052, "epoch": 321, "lr": 2.835503508217705e-05} +{"train_loss": 0.011902667582035065, "grad_norm": 0.07377475500106812, "global_step": 34053, "epoch": 321, "lr": 2.8352363455611974e-05} +{"train_loss": 0.014121809974312782, "grad_norm": 0.06489446759223938, "global_step": 34054, "epoch": 321, "lr": 2.834969190510726e-05} +{"train_loss": 0.014523237943649292, "grad_norm": 0.07827476412057877, "global_step": 34055, "epoch": 321, "lr": 2.8347020430672313e-05} +{"train_loss": 0.01492511946707964, "grad_norm": 0.09031664580106735, "global_step": 34056, "epoch": 321, "lr": 2.8344349032316536e-05} +{"train_loss": 0.014544477686285973, "grad_norm": 0.048955950886011124, "global_step": 34057, "epoch": 321, "lr": 2.8341677710049295e-05} +{"train_loss": 0.011566491797566414, "grad_norm": 0.055934105068445206, "global_step": 34058, "epoch": 321, "lr": 2.833900646387997e-05} +{"train_loss": 0.017429523169994354, "grad_norm": 0.06102772057056427, "global_step": 34059, "epoch": 321, "lr": 2.833633529381795e-05} +{"train_loss": 0.015647439286112785, "grad_norm": 0.0663563460111618, "global_step": 34060, "epoch": 321, "lr": 2.8333664199872644e-05} +{"train_loss": 0.013821997679769993, "grad_norm": 0.10049322247505188, "global_step": 34061, "epoch": 321, "lr": 2.8330993182053406e-05} +{"train_loss": 0.014221768826246262, "grad_norm": 0.11508508026599884, "global_step": 34062, "epoch": 321, "lr": 2.8328322240369632e-05} +{"train_loss": 0.013562179170548916, "grad_norm": 0.06154429540038109, "global_step": 34063, "epoch": 321, "lr": 2.832565137483073e-05} +{"train_loss": 0.018587026745080948, "grad_norm": 0.08621933311223984, "global_step": 34064, "epoch": 321, "lr": 2.8322980585446058e-05} +{"train_loss": 0.01536194235086441, "grad_norm": 0.08853690326213837, "global_step": 34065, "epoch": 321, "lr": 2.8320309872224993e-05} +{"train_loss": 0.013818860054016113, "grad_norm": 0.13718731701374054, "global_step": 34066, "epoch": 321, "lr": 2.8317639235176917e-05} +{"train_loss": 0.012598400935530663, "grad_norm": 0.06857841461896896, "global_step": 34067, "epoch": 321, "lr": 2.831496867431126e-05} +{"train_loss": 0.014643491245806217, "grad_norm": 0.1284523457288742, "global_step": 34068, "epoch": 321, "lr": 2.8312298189637353e-05} +{"train_loss": 0.017511185258626938, "grad_norm": 0.08972270786762238, "global_step": 34069, "epoch": 321, "lr": 2.830962778116461e-05} +{"train_loss": 0.013130029663443565, "grad_norm": 0.06184014678001404, "global_step": 34070, "epoch": 321, "lr": 2.8306957448902404e-05} +{"train_loss": 0.014161976054310799, "grad_norm": 0.06520575284957886, "global_step": 34071, "epoch": 321, "lr": 2.8304287192860102e-05} +{"train_loss": 0.014201771467924118, "grad_norm": 0.07990498840808868, "global_step": 34072, "epoch": 321, "lr": 2.83016170130471e-05} +{"train_loss": 0.017269279807806015, "grad_norm": 0.09211920946836472, "global_step": 34073, "epoch": 321, "lr": 2.829894690947279e-05} +{"train_loss": 0.015737902373075485, "grad_norm": 0.08704400062561035, "global_step": 34074, "epoch": 321, "lr": 2.8296276882146532e-05} +{"train_loss": 0.015189291909337044, "grad_norm": 0.06579520553350449, "global_step": 34075, "epoch": 321, "lr": 2.829360693107773e-05} +{"train_loss": 0.018901072442531586, "grad_norm": 0.10884089022874832, "global_step": 34076, "epoch": 321, "lr": 2.829093705627574e-05} +{"train_loss": 0.012777849100530148, "grad_norm": 0.12950970232486725, "global_step": 34077, "epoch": 321, "lr": 2.8288267257749972e-05} +{"train_loss": 0.015297383069992065, "grad_norm": 0.07023101300001144, "global_step": 34078, "epoch": 321, "lr": 2.828559753550977e-05} +{"train_loss": 0.014443652704358101, "grad_norm": 0.07740286737680435, "global_step": 34079, "epoch": 321, "lr": 2.8282927889564542e-05} +{"train_loss": 0.01663333922624588, "grad_norm": 0.1477188616991043, "global_step": 34080, "epoch": 321, "lr": 2.8280258319923676e-05} +{"train_loss": 0.012929089367389679, "grad_norm": 0.1098618134856224, "global_step": 34081, "epoch": 321, "lr": 2.8277588826596535e-05} +{"train_loss": 0.015421167947351933, "grad_norm": 0.06379183381795883, "global_step": 34082, "epoch": 321, "lr": 2.8274919409592483e-05} +{"train_loss": 0.015249557793140411, "grad_norm": 0.05207139626145363, "global_step": 34083, "epoch": 321, "lr": 2.8272250068920913e-05} +{"train_loss": 0.016271959990262985, "grad_norm": 0.0775999203324318, "global_step": 34084, "epoch": 321, "lr": 2.8269580804591233e-05} +{"train_loss": 0.013437440618872643, "grad_norm": 0.08248165994882584, "global_step": 34085, "epoch": 321, "lr": 2.8266911616612767e-05} +{"train_loss": 0.014710510149598122, "grad_norm": 0.06881332397460938, "global_step": 34086, "epoch": 321, "lr": 2.8264242504994952e-05} +{"train_loss": 0.013685227371752262, "grad_norm": 0.14208750426769257, "global_step": 34087, "epoch": 321, "lr": 2.8261573469747125e-05} +{"train_loss": 0.013181453570723534, "grad_norm": 0.10018415749073029, "global_step": 34088, "epoch": 321, "lr": 2.8258904510878668e-05} +{"train_loss": 0.01354747824370861, "grad_norm": 0.11213566362857819, "global_step": 34089, "epoch": 321, "lr": 2.825623562839896e-05} +{"train_loss": 0.014558940194547176, "grad_norm": 0.07036694139242172, "global_step": 34090, "epoch": 321, "lr": 2.825356682231738e-05} +{"train_loss": 0.017460452392697334, "grad_norm": 0.08592374622821808, "global_step": 34091, "epoch": 321, "lr": 2.825089809264333e-05} +{"train_loss": 0.013917336240410805, "grad_norm": 0.07600405812263489, "global_step": 34092, "epoch": 321, "lr": 2.824822943938617e-05} +{"train_loss": 0.016839541494846344, "grad_norm": 0.058260247111320496, "global_step": 34093, "epoch": 321, "lr": 2.8245560862555253e-05} +{"train_loss": 0.017607465386390686, "grad_norm": 0.05665329098701477, "global_step": 34094, "epoch": 321, "lr": 2.824289236215999e-05} +{"train_loss": 0.016277052462100983, "grad_norm": 0.09181194007396698, "global_step": 34095, "epoch": 321, "lr": 2.824022393820972e-05} +{"train_loss": 0.014502260833978653, "grad_norm": 0.08578616380691528, "global_step": 34096, "epoch": 321, "lr": 2.8237555590713844e-05} +{"train_loss": 0.01288571022450924, "grad_norm": 0.06677304953336716, "global_step": 34097, "epoch": 321, "lr": 2.8234887319681748e-05} +{"train_loss": 0.013855446130037308, "grad_norm": 0.11243518441915512, "global_step": 34098, "epoch": 321, "lr": 2.8232219125122798e-05} +{"train_loss": 0.012906456366181374, "grad_norm": 0.07993864268064499, "global_step": 34099, "epoch": 321, "lr": 2.822955100704634e-05} +{"train_loss": 0.016730375587940216, "grad_norm": 0.20182761549949646, "global_step": 34100, "epoch": 321, "lr": 2.8226882965461777e-05} +{"train_loss": 0.013513360172510147, "grad_norm": 0.0752248764038086, "global_step": 34101, "epoch": 321, "lr": 2.8224215000378496e-05} +{"train_loss": 0.018396155908703804, "grad_norm": 0.08097759634256363, "global_step": 34102, "epoch": 321, "lr": 2.8221547111805834e-05} +{"train_loss": 0.014607511460781097, "grad_norm": 0.12090671807527542, "global_step": 34103, "epoch": 321, "lr": 2.821887929975318e-05} +{"train_loss": 0.016847245395183563, "grad_norm": 0.060657698661088943, "global_step": 34104, "epoch": 321, "lr": 2.8216211564229933e-05} +{"train_loss": 0.017237627878785133, "grad_norm": 0.059781573712825775, "global_step": 34105, "epoch": 321, "lr": 2.821354390524545e-05} +{"train_loss": 0.015942510217428207, "grad_norm": 0.06357094645500183, "global_step": 34106, "epoch": 321, "lr": 2.8210876322809076e-05} +{"train_loss": 0.014674575999379158, "grad_norm": 0.07606136798858643, "global_step": 34107, "epoch": 321, "lr": 2.8208208816930203e-05} +{"train_loss": 0.013229265809059143, "grad_norm": 0.09109841287136078, "global_step": 34108, "epoch": 321, "lr": 2.8205541387618233e-05} +{"train_loss": 0.013704844750463963, "grad_norm": 0.11548329889774323, "global_step": 34109, "epoch": 321, "lr": 2.8202874034882487e-05} +{"train_loss": 0.01726824976503849, "grad_norm": 0.0815959945321083, "global_step": 34110, "epoch": 321, "lr": 2.820020675873239e-05} +{"train_loss": 0.01642870530486107, "grad_norm": 0.06555888801813126, "global_step": 34111, "epoch": 321, "lr": 2.8197539559177276e-05} +{"train_loss": 0.016571711748838425, "grad_norm": 0.07407472282648087, "global_step": 34112, "epoch": 321, "lr": 2.8194872436226515e-05} +{"train_loss": 0.01836376078426838, "grad_norm": 0.0860709473490715, "global_step": 34113, "epoch": 321, "lr": 2.8192205389889487e-05} +{"train_loss": 0.016059577465057373, "grad_norm": 0.06212306767702103, "global_step": 34114, "epoch": 321, "lr": 2.818953842017559e-05} +{"train_loss": 0.013232773169875145, "grad_norm": 0.0894210934638977, "global_step": 34115, "epoch": 321, "lr": 2.818687152709414e-05} +{"train_loss": 0.016252757981419563, "grad_norm": 0.16442571580410004, "global_step": 34116, "epoch": 321, "lr": 2.818420471065456e-05} +{"train_loss": 0.015486708842217922, "grad_norm": 0.056700535118579865, "global_step": 34117, "epoch": 321, "lr": 2.8181537970866178e-05} +{"train_loss": 0.015452699735760689, "grad_norm": 0.07705473899841309, "global_step": 34118, "epoch": 321, "lr": 2.817887130773841e-05} +{"train_loss": 0.016865381971001625, "grad_norm": 0.06303562223911285, "global_step": 34119, "epoch": 321, "lr": 2.8176204721280563e-05} +{"train_loss": 0.0129639757797122, "grad_norm": 0.07886376976966858, "global_step": 34120, "epoch": 321, "lr": 2.8173538211502055e-05} +{"train_loss": 0.015951307490468025, "grad_norm": 0.08337034285068512, "global_step": 34121, "epoch": 321, "lr": 2.8170871778412256e-05} +{"train_loss": 0.01412893459200859, "grad_norm": 0.06562439352273941, "global_step": 34122, "epoch": 321, "lr": 2.816820542202052e-05} +{"train_loss": 0.013526750728487968, "grad_norm": 0.06046857312321663, "global_step": 34123, "epoch": 321, "lr": 2.81655391423362e-05} +{"train_loss": 0.015813250094652176, "grad_norm": 0.07739686220884323, "global_step": 34124, "epoch": 321, "lr": 2.8162872939368677e-05} +{"train_loss": 0.016311634331941605, "grad_norm": 0.14622318744659424, "global_step": 34125, "epoch": 321, "lr": 2.8160206813127345e-05} +{"train_loss": 0.015339741483330727, "grad_norm": 0.08870817720890045, "global_step": 34126, "epoch": 321, "lr": 2.8157540763621526e-05} +{"train_loss": 0.013631212525069714, "grad_norm": 0.10819292813539505, "global_step": 34127, "epoch": 321, "lr": 2.8154874790860608e-05} +{"train_loss": 0.01532047986984253, "grad_norm": 0.09396559745073318, "global_step": 34128, "epoch": 321, "lr": 2.8152208894854005e-05} +{"train_loss": 0.015794355422258377, "grad_norm": 0.06889933347702026, "global_step": 34129, "epoch": 321, "lr": 2.8149543075610995e-05} +{"train_loss": 0.017657002434134483, "grad_norm": 0.07838373631238937, "global_step": 34130, "epoch": 321, "lr": 2.8146877333140986e-05} +{"train_loss": 0.015195219347007433, "grad_norm": 0.08471982926130295, "global_step": 34131, "epoch": 321, "lr": 2.8144211667453368e-05, "val_loss": 0.019699152559041977} +{"train_loss": 0.016282638534903526, "grad_norm": 0.15704627335071564, "global_step": 34132, "epoch": 322, "lr": 2.8141546078557468e-05} +{"train_loss": 0.015387076884508133, "grad_norm": 0.07121980935335159, "global_step": 34133, "epoch": 322, "lr": 2.8138880566462666e-05} +{"train_loss": 0.01362529955804348, "grad_norm": 0.06092485040426254, "global_step": 34134, "epoch": 322, "lr": 2.8136215131178344e-05} +{"train_loss": 0.016037192195653915, "grad_norm": 0.10667695105075836, "global_step": 34135, "epoch": 322, "lr": 2.813354977271386e-05} +{"train_loss": 0.014410819858312607, "grad_norm": 0.06944363564252853, "global_step": 34136, "epoch": 322, "lr": 2.8130884491078557e-05} +{"train_loss": 0.013345390558242798, "grad_norm": 0.09538602083921432, "global_step": 34137, "epoch": 322, "lr": 2.8128219286281817e-05} +{"train_loss": 0.01764950528740883, "grad_norm": 0.10841909795999527, "global_step": 34138, "epoch": 322, "lr": 2.812555415833301e-05} +{"train_loss": 0.013414425775408745, "grad_norm": 0.0832764059305191, "global_step": 34139, "epoch": 322, "lr": 2.81228891072415e-05} +{"train_loss": 0.013880378566682339, "grad_norm": 0.058932289481163025, "global_step": 34140, "epoch": 322, "lr": 2.8120224133016625e-05} +{"train_loss": 0.015858447179198265, "grad_norm": 0.08237805962562561, "global_step": 34141, "epoch": 322, "lr": 2.811755923566777e-05} +{"train_loss": 0.016933206468820572, "grad_norm": 0.15283332765102386, "global_step": 34142, "epoch": 322, "lr": 2.8114894415204313e-05} +{"train_loss": 0.01587441749870777, "grad_norm": 0.11277730017900467, "global_step": 34143, "epoch": 322, "lr": 2.8112229671635576e-05} +{"train_loss": 0.01428152620792389, "grad_norm": 0.07134935259819031, "global_step": 34144, "epoch": 322, "lr": 2.810956500497095e-05} +{"train_loss": 0.016172491014003754, "grad_norm": 0.06737446039915085, "global_step": 34145, "epoch": 322, "lr": 2.8106900415219806e-05} +{"train_loss": 0.014453724958002567, "grad_norm": 0.09860553592443466, "global_step": 34146, "epoch": 322, "lr": 2.8104235902391497e-05} +{"train_loss": 0.018619803711771965, "grad_norm": 0.08153875172138214, "global_step": 34147, "epoch": 322, "lr": 2.810157146649536e-05} +{"train_loss": 0.016536768525838852, "grad_norm": 0.10230667144060135, "global_step": 34148, "epoch": 322, "lr": 2.809890710754078e-05} +{"train_loss": 0.01489163562655449, "grad_norm": 0.06401380896568298, "global_step": 34149, "epoch": 322, "lr": 2.8096242825537134e-05} +{"train_loss": 0.01594613865017891, "grad_norm": 0.07234834879636765, "global_step": 34150, "epoch": 322, "lr": 2.8093578620493743e-05} +{"train_loss": 0.01609630137681961, "grad_norm": 0.08241930603981018, "global_step": 34151, "epoch": 322, "lr": 2.8090914492420016e-05} +{"train_loss": 0.013063129037618637, "grad_norm": 0.06337897479534149, "global_step": 34152, "epoch": 322, "lr": 2.808825044132528e-05} +{"train_loss": 0.013922452926635742, "grad_norm": 0.06405377388000488, "global_step": 34153, "epoch": 322, "lr": 2.8085586467218884e-05} +{"train_loss": 0.013071272522211075, "grad_norm": 0.06472949683666229, "global_step": 34154, "epoch": 322, "lr": 2.8082922570110215e-05} +{"train_loss": 0.01615513488650322, "grad_norm": 0.06146308034658432, "global_step": 34155, "epoch": 322, "lr": 2.8080258750008638e-05} +{"train_loss": 0.01533736102283001, "grad_norm": 0.09111957252025604, "global_step": 34156, "epoch": 322, "lr": 2.8077595006923474e-05} +{"train_loss": 0.013499726541340351, "grad_norm": 0.06761318445205688, "global_step": 34157, "epoch": 322, "lr": 2.8074931340864136e-05} +{"train_loss": 0.011853886768221855, "grad_norm": 0.0596318356692791, "global_step": 34158, "epoch": 322, "lr": 2.807226775183993e-05} +{"train_loss": 0.01600765623152256, "grad_norm": 0.08210386335849762, "global_step": 34159, "epoch": 322, "lr": 2.806960423986026e-05} +{"train_loss": 0.015856541693210602, "grad_norm": 0.07183019071817398, "global_step": 34160, "epoch": 322, "lr": 2.8066940804934444e-05} +{"train_loss": 0.014520110562443733, "grad_norm": 0.055348701775074005, "global_step": 34161, "epoch": 322, "lr": 2.8064277447071862e-05} +{"train_loss": 0.015400096774101257, "grad_norm": 0.0652153417468071, "global_step": 34162, "epoch": 322, "lr": 2.8061614166281886e-05} +{"train_loss": 0.01607448235154152, "grad_norm": 0.08312585949897766, "global_step": 34163, "epoch": 322, "lr": 2.8058950962573855e-05} +{"train_loss": 0.012888812460005283, "grad_norm": 0.0788508728146553, "global_step": 34164, "epoch": 322, "lr": 2.8056287835957113e-05} +{"train_loss": 0.016338858753442764, "grad_norm": 0.09083951264619827, "global_step": 34165, "epoch": 322, "lr": 2.805362478644103e-05} +{"train_loss": 0.01258197333663702, "grad_norm": 0.08043932169675827, "global_step": 34166, "epoch": 322, "lr": 2.8050961814034983e-05} +{"train_loss": 0.017298495396971703, "grad_norm": 0.06434130668640137, "global_step": 34167, "epoch": 322, "lr": 2.8048298918748294e-05} +{"train_loss": 0.012427054345607758, "grad_norm": 0.14445295929908752, "global_step": 34168, "epoch": 322, "lr": 2.804563610059034e-05} +{"train_loss": 0.015470633283257484, "grad_norm": 0.1287238597869873, "global_step": 34169, "epoch": 322, "lr": 2.8042973359570506e-05} +{"train_loss": 0.015290936455130577, "grad_norm": 0.07540139555931091, "global_step": 34170, "epoch": 322, "lr": 2.804031069569808e-05} +{"train_loss": 0.014977030456066132, "grad_norm": 0.06103072315454483, "global_step": 34171, "epoch": 322, "lr": 2.803764810898245e-05} +{"train_loss": 0.014152051880955696, "grad_norm": 0.07438525557518005, "global_step": 34172, "epoch": 322, "lr": 2.8034985599432993e-05} +{"train_loss": 0.01488729752600193, "grad_norm": 0.09261063486337662, "global_step": 34173, "epoch": 322, "lr": 2.8032323167059023e-05} +{"train_loss": 0.015479027293622494, "grad_norm": 0.1153428927063942, "global_step": 34174, "epoch": 322, "lr": 2.8029660811869917e-05} +{"train_loss": 0.013458490371704102, "grad_norm": 0.09139307588338852, "global_step": 34175, "epoch": 322, "lr": 2.8026998533875047e-05} +{"train_loss": 0.016174234449863434, "grad_norm": 0.15255701541900635, "global_step": 34176, "epoch": 322, "lr": 2.802433633308375e-05} +{"train_loss": 0.016637034714221954, "grad_norm": 0.06152046471834183, "global_step": 34177, "epoch": 322, "lr": 2.8021674209505355e-05} +{"train_loss": 0.015376472845673561, "grad_norm": 0.10467244684696198, "global_step": 34178, "epoch": 322, "lr": 2.801901216314925e-05} +{"train_loss": 0.01658909022808075, "grad_norm": 0.12348370999097824, "global_step": 34179, "epoch": 322, "lr": 2.8016350194024787e-05} +{"train_loss": 0.013963375240564346, "grad_norm": 0.07135585695505142, "global_step": 34180, "epoch": 322, "lr": 2.801368830214129e-05} +{"train_loss": 0.014758256264030933, "grad_norm": 0.06728598475456238, "global_step": 34181, "epoch": 322, "lr": 2.801102648750815e-05} +{"train_loss": 0.01506393775343895, "grad_norm": 0.07816222310066223, "global_step": 34182, "epoch": 322, "lr": 2.8008364750134685e-05} +{"train_loss": 0.015584762208163738, "grad_norm": 0.12806324660778046, "global_step": 34183, "epoch": 322, "lr": 2.8005703090030278e-05} +{"train_loss": 0.014309735968708992, "grad_norm": 0.0866747498512268, "global_step": 34184, "epoch": 322, "lr": 2.8003041507204242e-05} +{"train_loss": 0.01740618795156479, "grad_norm": 0.1208847314119339, "global_step": 34185, "epoch": 322, "lr": 2.800038000166596e-05} +{"train_loss": 0.0166475847363472, "grad_norm": 0.09734224528074265, "global_step": 34186, "epoch": 322, "lr": 2.7997718573424792e-05} +{"train_loss": 0.01605876535177231, "grad_norm": 0.06211681663990021, "global_step": 34187, "epoch": 322, "lr": 2.7995057222490075e-05} +{"train_loss": 0.016139429062604904, "grad_norm": 0.17486493289470673, "global_step": 34188, "epoch": 322, "lr": 2.7992395948871132e-05} +{"train_loss": 0.016467519104480743, "grad_norm": 0.13247741758823395, "global_step": 34189, "epoch": 322, "lr": 2.7989734752577356e-05} +{"train_loss": 0.016088884323835373, "grad_norm": 0.09242246299982071, "global_step": 34190, "epoch": 322, "lr": 2.7987073633618068e-05} +{"train_loss": 0.014052346348762512, "grad_norm": 0.05784070864319801, "global_step": 34191, "epoch": 322, "lr": 2.798441259200263e-05} +{"train_loss": 0.014799638651311398, "grad_norm": 0.057339318096637726, "global_step": 34192, "epoch": 322, "lr": 2.7981751627740406e-05} +{"train_loss": 0.015053079463541508, "grad_norm": 0.10796259343624115, "global_step": 34193, "epoch": 322, "lr": 2.797909074084073e-05} +{"train_loss": 0.014401895925402641, "grad_norm": 0.10515404492616653, "global_step": 34194, "epoch": 322, "lr": 2.797642993131294e-05} +{"train_loss": 0.013746419921517372, "grad_norm": 0.0576908104121685, "global_step": 34195, "epoch": 322, "lr": 2.7973769199166396e-05} +{"train_loss": 0.01702917367219925, "grad_norm": 0.061504025012254715, "global_step": 34196, "epoch": 322, "lr": 2.7971108544410463e-05} +{"train_loss": 0.015839993953704834, "grad_norm": 0.0792335569858551, "global_step": 34197, "epoch": 322, "lr": 2.7968447967054456e-05} +{"train_loss": 0.013914256356656551, "grad_norm": 0.07004334777593613, "global_step": 34198, "epoch": 322, "lr": 2.796578746710776e-05} +{"train_loss": 0.018708664923906326, "grad_norm": 0.067428819835186, "global_step": 34199, "epoch": 322, "lr": 2.7963127044579697e-05} +{"train_loss": 0.014408077113330364, "grad_norm": 0.08368618041276932, "global_step": 34200, "epoch": 322, "lr": 2.7960466699479626e-05} +{"train_loss": 0.015401139855384827, "grad_norm": 0.07439396530389786, "global_step": 34201, "epoch": 322, "lr": 2.795780643181688e-05} +{"train_loss": 0.01495087705552578, "grad_norm": 0.12444275617599487, "global_step": 34202, "epoch": 322, "lr": 2.7955146241600815e-05} +{"train_loss": 0.014153016731142998, "grad_norm": 0.12698373198509216, "global_step": 34203, "epoch": 322, "lr": 2.7952486128840795e-05} +{"train_loss": 0.014600170776247978, "grad_norm": 0.05588918179273605, "global_step": 34204, "epoch": 322, "lr": 2.794982609354615e-05} +{"train_loss": 0.016209103167057037, "grad_norm": 0.1779795140028, "global_step": 34205, "epoch": 322, "lr": 2.794716613572621e-05} +{"train_loss": 0.014078600332140923, "grad_norm": 0.13171906769275665, "global_step": 34206, "epoch": 322, "lr": 2.7944506255390333e-05} +{"train_loss": 0.014555053785443306, "grad_norm": 0.07659878581762314, "global_step": 34207, "epoch": 322, "lr": 2.7941846452547898e-05} +{"train_loss": 0.015147874131798744, "grad_norm": 0.053910642862319946, "global_step": 34208, "epoch": 322, "lr": 2.79391867272082e-05} +{"train_loss": 0.013959289528429508, "grad_norm": 0.09486126154661179, "global_step": 34209, "epoch": 322, "lr": 2.7936527079380604e-05} +{"train_loss": 0.01746946945786476, "grad_norm": 0.08258408308029175, "global_step": 34210, "epoch": 322, "lr": 2.7933867509074496e-05} +{"train_loss": 0.013133583590388298, "grad_norm": 0.07370959967374802, "global_step": 34211, "epoch": 322, "lr": 2.7931208016299136e-05} +{"train_loss": 0.015129366889595985, "grad_norm": 0.05483679845929146, "global_step": 34212, "epoch": 322, "lr": 2.792854860106392e-05} +{"train_loss": 0.01334291696548462, "grad_norm": 0.06377952545881271, "global_step": 34213, "epoch": 322, "lr": 2.7925889263378203e-05} +{"train_loss": 0.016098275780677795, "grad_norm": 0.12077049165964127, "global_step": 34214, "epoch": 322, "lr": 2.7923230003251295e-05} +{"train_loss": 0.01352250948548317, "grad_norm": 0.09390760213136673, "global_step": 34215, "epoch": 322, "lr": 2.792057082069255e-05} +{"train_loss": 0.014969310723245144, "grad_norm": 0.06451401114463806, "global_step": 34216, "epoch": 322, "lr": 2.7917911715711344e-05} +{"train_loss": 0.015831928700208664, "grad_norm": 0.0848643109202385, "global_step": 34217, "epoch": 322, "lr": 2.791525268831699e-05} +{"train_loss": 0.014379134401679039, "grad_norm": 0.0933656245470047, "global_step": 34218, "epoch": 322, "lr": 2.791259373851881e-05} +{"train_loss": 0.015480514615774155, "grad_norm": 0.08840880542993546, "global_step": 34219, "epoch": 322, "lr": 2.790993486632617e-05} +{"train_loss": 0.014333495870232582, "grad_norm": 0.08321978151798248, "global_step": 34220, "epoch": 322, "lr": 2.7907276071748438e-05} +{"train_loss": 0.015094207599759102, "grad_norm": 0.14410939812660217, "global_step": 34221, "epoch": 322, "lr": 2.790461735479491e-05} +{"train_loss": 0.015495233237743378, "grad_norm": 0.10765740275382996, "global_step": 34222, "epoch": 322, "lr": 2.7901958715474965e-05} +{"train_loss": 0.014862450771033764, "grad_norm": 0.14202925562858582, "global_step": 34223, "epoch": 322, "lr": 2.7899300153797902e-05} +{"train_loss": 0.015429003164172173, "grad_norm": 0.08542248606681824, "global_step": 34224, "epoch": 322, "lr": 2.7896641669773114e-05} +{"train_loss": 0.016847355291247368, "grad_norm": 0.1653648018836975, "global_step": 34225, "epoch": 322, "lr": 2.7893983263409894e-05} +{"train_loss": 0.015700705349445343, "grad_norm": 0.05600477010011673, "global_step": 34226, "epoch": 322, "lr": 2.78913249347176e-05} +{"train_loss": 0.015638142824172974, "grad_norm": 0.06373470276594162, "global_step": 34227, "epoch": 322, "lr": 2.78886666837056e-05} +{"train_loss": 0.014679061248898506, "grad_norm": 0.10396329313516617, "global_step": 34228, "epoch": 322, "lr": 2.7886008510383205e-05} +{"train_loss": 0.01323889009654522, "grad_norm": 0.12859365344047546, "global_step": 34229, "epoch": 322, "lr": 2.788335041475974e-05} +{"train_loss": 0.014578256756067276, "grad_norm": 0.07217973470687866, "global_step": 34230, "epoch": 322, "lr": 2.7880692396844582e-05} +{"train_loss": 0.01822737604379654, "grad_norm": 0.1283717304468155, "global_step": 34231, "epoch": 322, "lr": 2.7878034456647035e-05} +{"train_loss": 0.01562967710196972, "grad_norm": 0.07350654155015945, "global_step": 34232, "epoch": 322, "lr": 2.787537659417645e-05} +{"train_loss": 0.014395028352737427, "grad_norm": 0.0560106597840786, "global_step": 34233, "epoch": 322, "lr": 2.7872718809442195e-05} +{"train_loss": 0.014864709228277206, "grad_norm": 0.08228041976690292, "global_step": 34234, "epoch": 322, "lr": 2.787006110245358e-05} +{"train_loss": 0.018174592405557632, "grad_norm": 0.06051289662718773, "global_step": 34235, "epoch": 322, "lr": 2.786740347321992e-05} +{"train_loss": 0.014955186285078526, "grad_norm": 0.06879031658172607, "global_step": 34236, "epoch": 322, "lr": 2.7864745921750592e-05} +{"train_loss": 0.01519824947529244, "grad_norm": 0.10230058431625366, "global_step": 34237, "epoch": 322, "lr": 2.7862088448054936e-05, "val_loss": 0.019287964329123497} +{"train_loss": 0.015653962269425392, "grad_norm": 0.05669616907835007, "global_step": 34238, "epoch": 323, "lr": 2.7859431052142248e-05} +{"train_loss": 0.014746537432074547, "grad_norm": 0.0744556114077568, "global_step": 34239, "epoch": 323, "lr": 2.7856773734021913e-05} +{"train_loss": 0.01543772965669632, "grad_norm": 0.09654837101697922, "global_step": 34240, "epoch": 323, "lr": 2.785411649370323e-05} +{"train_loss": 0.016358286142349243, "grad_norm": 0.06396947056055069, "global_step": 34241, "epoch": 323, "lr": 2.7851459331195568e-05} +{"train_loss": 0.01265505701303482, "grad_norm": 0.05651719495654106, "global_step": 34242, "epoch": 323, "lr": 2.784880224650822e-05} +{"train_loss": 0.01768704317510128, "grad_norm": 0.06609612703323364, "global_step": 34243, "epoch": 323, "lr": 2.7846145239650555e-05} +{"train_loss": 0.014755582436919212, "grad_norm": 0.08177240192890167, "global_step": 34244, "epoch": 323, "lr": 2.784348831063192e-05} +{"train_loss": 0.014737349934875965, "grad_norm": 0.07761470228433609, "global_step": 34245, "epoch": 323, "lr": 2.784083145946163e-05} +{"train_loss": 0.012865100055932999, "grad_norm": 0.09165672212839127, "global_step": 34246, "epoch": 323, "lr": 2.7838174686149e-05} +{"train_loss": 0.015521379187703133, "grad_norm": 0.06089697405695915, "global_step": 34247, "epoch": 323, "lr": 2.783551799070339e-05} +{"train_loss": 0.017441008239984512, "grad_norm": 0.11380951851606369, "global_step": 34248, "epoch": 323, "lr": 2.7832861373134143e-05} +{"train_loss": 0.01319785788655281, "grad_norm": 0.09426111727952957, "global_step": 34249, "epoch": 323, "lr": 2.7830204833450575e-05} +{"train_loss": 0.013482695445418358, "grad_norm": 0.057362865656614304, "global_step": 34250, "epoch": 323, "lr": 2.7827548371662017e-05} +{"train_loss": 0.017807630822062492, "grad_norm": 0.11170601099729538, "global_step": 34251, "epoch": 323, "lr": 2.7824891987777846e-05} +{"train_loss": 0.01583481766283512, "grad_norm": 0.09173500537872314, "global_step": 34252, "epoch": 323, "lr": 2.7822235681807328e-05} +{"train_loss": 0.016063842922449112, "grad_norm": 0.11719650030136108, "global_step": 34253, "epoch": 323, "lr": 2.7819579453759832e-05} +{"train_loss": 0.013837916776537895, "grad_norm": 0.04980695992708206, "global_step": 34254, "epoch": 323, "lr": 2.7816923303644704e-05} +{"train_loss": 0.015620691701769829, "grad_norm": 0.059937965124845505, "global_step": 34255, "epoch": 323, "lr": 2.7814267231471237e-05} +{"train_loss": 0.014754479750990868, "grad_norm": 0.08266167342662811, "global_step": 34256, "epoch": 323, "lr": 2.7811611237248792e-05} +{"train_loss": 0.01592584140598774, "grad_norm": 0.07049877196550369, "global_step": 34257, "epoch": 323, "lr": 2.780895532098671e-05} +{"train_loss": 0.013916047289967537, "grad_norm": 0.06480363011360168, "global_step": 34258, "epoch": 323, "lr": 2.7806299482694307e-05} +{"train_loss": 0.013062831945717335, "grad_norm": 0.06947341561317444, "global_step": 34259, "epoch": 323, "lr": 2.7803643722380906e-05} +{"train_loss": 0.018552696332335472, "grad_norm": 0.06442246586084366, "global_step": 34260, "epoch": 323, "lr": 2.7800988040055843e-05} +{"train_loss": 0.014670309610664845, "grad_norm": 0.06113254651427269, "global_step": 34261, "epoch": 323, "lr": 2.7798332435728468e-05} +{"train_loss": 0.015894977375864983, "grad_norm": 0.0678260326385498, "global_step": 34262, "epoch": 323, "lr": 2.7795676909408087e-05} +{"train_loss": 0.016697973012924194, "grad_norm": 0.10357767343521118, "global_step": 34263, "epoch": 323, "lr": 2.7793021461104056e-05} +{"train_loss": 0.015795473009347916, "grad_norm": 0.10498326271772385, "global_step": 34264, "epoch": 323, "lr": 2.779036609082567e-05} +{"train_loss": 0.013622120954096317, "grad_norm": 0.08810071647167206, "global_step": 34265, "epoch": 323, "lr": 2.77877107985823e-05} +{"train_loss": 0.01472976803779602, "grad_norm": 0.05892455205321312, "global_step": 34266, "epoch": 323, "lr": 2.7785055584383236e-05} +{"train_loss": 0.016498006880283356, "grad_norm": 0.08558612316846848, "global_step": 34267, "epoch": 323, "lr": 2.778240044823782e-05} +{"train_loss": 0.01649271510541439, "grad_norm": 0.06422746181488037, "global_step": 34268, "epoch": 323, "lr": 2.7779745390155414e-05} +{"train_loss": 0.014094133861362934, "grad_norm": 0.08207828551530838, "global_step": 34269, "epoch": 323, "lr": 2.777709041014531e-05} +{"train_loss": 0.014587441459298134, "grad_norm": 0.10034794360399246, "global_step": 34270, "epoch": 323, "lr": 2.7774435508216833e-05} +{"train_loss": 0.014632252976298332, "grad_norm": 0.10094735026359558, "global_step": 34271, "epoch": 323, "lr": 2.7771780684379344e-05} +{"train_loss": 0.01551838219165802, "grad_norm": 0.10192977637052536, "global_step": 34272, "epoch": 323, "lr": 2.7769125938642132e-05} +{"train_loss": 0.017592407763004303, "grad_norm": 0.06605851650238037, "global_step": 34273, "epoch": 323, "lr": 2.7766471271014537e-05} +{"train_loss": 0.012978320941329002, "grad_norm": 0.054016534239053726, "global_step": 34274, "epoch": 323, "lr": 2.776381668150592e-05} +{"train_loss": 0.013944428414106369, "grad_norm": 0.12126221507787704, "global_step": 34275, "epoch": 323, "lr": 2.7761162170125575e-05} +{"train_loss": 0.017078913748264313, "grad_norm": 0.09381278604269028, "global_step": 34276, "epoch": 323, "lr": 2.7758507736882822e-05} +{"train_loss": 0.014699644409120083, "grad_norm": 0.06092114374041557, "global_step": 34277, "epoch": 323, "lr": 2.7755853381786993e-05} +{"train_loss": 0.013381348922848701, "grad_norm": 0.06995820254087448, "global_step": 34278, "epoch": 323, "lr": 2.775319910484745e-05} +{"train_loss": 0.01627860590815544, "grad_norm": 0.11989352852106094, "global_step": 34279, "epoch": 323, "lr": 2.7750544906073463e-05} +{"train_loss": 0.014155438169836998, "grad_norm": 0.09777723252773285, "global_step": 34280, "epoch": 323, "lr": 2.7747890785474386e-05} +{"train_loss": 0.013895610347390175, "grad_norm": 0.08646547049283981, "global_step": 34281, "epoch": 323, "lr": 2.7745236743059567e-05} +{"train_loss": 0.013874373398721218, "grad_norm": 0.07398412376642227, "global_step": 34282, "epoch": 323, "lr": 2.77425827788383e-05} +{"train_loss": 0.014131385833024979, "grad_norm": 0.11363613605499268, "global_step": 34283, "epoch": 323, "lr": 2.77399288928199e-05} +{"train_loss": 0.016512766480445862, "grad_norm": 0.06881280243396759, "global_step": 34284, "epoch": 323, "lr": 2.7737275085013713e-05} +{"train_loss": 0.017561059445142746, "grad_norm": 0.12385014444589615, "global_step": 34285, "epoch": 323, "lr": 2.773462135542908e-05} +{"train_loss": 0.01982024870812893, "grad_norm": 0.08390249311923981, "global_step": 34286, "epoch": 323, "lr": 2.7731967704075285e-05} +{"train_loss": 0.014773257076740265, "grad_norm": 0.07330343872308731, "global_step": 34287, "epoch": 323, "lr": 2.7729314130961692e-05} +{"train_loss": 0.012613661587238312, "grad_norm": 0.07405252754688263, "global_step": 34288, "epoch": 323, "lr": 2.77266606360976e-05} +{"train_loss": 0.014207430183887482, "grad_norm": 0.10071369260549545, "global_step": 34289, "epoch": 323, "lr": 2.7724007219492315e-05} +{"train_loss": 0.013900693506002426, "grad_norm": 0.10964922606945038, "global_step": 34290, "epoch": 323, "lr": 2.772135388115519e-05} +{"train_loss": 0.015458240173757076, "grad_norm": 0.08863695710897446, "global_step": 34291, "epoch": 323, "lr": 2.771870062109555e-05} +{"train_loss": 0.014448045752942562, "grad_norm": 0.06368570774793625, "global_step": 34292, "epoch": 323, "lr": 2.77160474393227e-05} +{"train_loss": 0.01703636534512043, "grad_norm": 0.06502683460712433, "global_step": 34293, "epoch": 323, "lr": 2.771339433584596e-05} +{"train_loss": 0.014035126194357872, "grad_norm": 0.12446247041225433, "global_step": 34294, "epoch": 323, "lr": 2.771074131067466e-05} +{"train_loss": 0.015253460966050625, "grad_norm": 0.08767828345298767, "global_step": 34295, "epoch": 323, "lr": 2.7708088363818134e-05} +{"train_loss": 0.01412821002304554, "grad_norm": 0.0791066512465477, "global_step": 34296, "epoch": 323, "lr": 2.770543549528568e-05} +{"train_loss": 0.015304278582334518, "grad_norm": 0.10694080591201782, "global_step": 34297, "epoch": 323, "lr": 2.7702782705086626e-05} +{"train_loss": 0.016570741310715675, "grad_norm": 0.08966203778982162, "global_step": 34298, "epoch": 323, "lr": 2.7700129993230318e-05} +{"train_loss": 0.01620115526020527, "grad_norm": 0.07825917750597, "global_step": 34299, "epoch": 323, "lr": 2.7697477359726047e-05} +{"train_loss": 0.015991825610399246, "grad_norm": 0.0848800465464592, "global_step": 34300, "epoch": 323, "lr": 2.769482480458313e-05} +{"train_loss": 0.013750912621617317, "grad_norm": 0.07042993605136871, "global_step": 34301, "epoch": 323, "lr": 2.7692172327810894e-05} +{"train_loss": 0.015886450186371803, "grad_norm": 0.12491582334041595, "global_step": 34302, "epoch": 323, "lr": 2.7689519929418684e-05} +{"train_loss": 0.014053534716367722, "grad_norm": 0.06589893251657486, "global_step": 34303, "epoch": 323, "lr": 2.7686867609415784e-05} +{"train_loss": 0.01547283586114645, "grad_norm": 0.05711402744054794, "global_step": 34304, "epoch": 323, "lr": 2.7684215367811538e-05} +{"train_loss": 0.014049475081264973, "grad_norm": 0.08371956646442413, "global_step": 34305, "epoch": 323, "lr": 2.7681563204615247e-05} +{"train_loss": 0.018521735444664955, "grad_norm": 0.10009029507637024, "global_step": 34306, "epoch": 323, "lr": 2.767891111983625e-05} +{"train_loss": 0.016971968114376068, "grad_norm": 0.11031601577997208, "global_step": 34307, "epoch": 323, "lr": 2.7676259113483827e-05} +{"train_loss": 0.014215972274541855, "grad_norm": 0.06361750513315201, "global_step": 34308, "epoch": 323, "lr": 2.767360718556733e-05} +{"train_loss": 0.016317663714289665, "grad_norm": 0.10787203162908554, "global_step": 34309, "epoch": 323, "lr": 2.7670955336096084e-05} +{"train_loss": 0.015313589945435524, "grad_norm": 0.08305198699235916, "global_step": 34310, "epoch": 323, "lr": 2.7668303565079388e-05} +{"train_loss": 0.013478386215865612, "grad_norm": 0.14000798761844635, "global_step": 34311, "epoch": 323, "lr": 2.7665651872526543e-05} +{"train_loss": 0.014042486436665058, "grad_norm": 0.06316511332988739, "global_step": 34312, "epoch": 323, "lr": 2.7663000258446902e-05} +{"train_loss": 0.014039785601198673, "grad_norm": 0.08671104907989502, "global_step": 34313, "epoch": 323, "lr": 2.766034872284975e-05} +{"train_loss": 0.015259362757205963, "grad_norm": 0.14066250622272491, "global_step": 34314, "epoch": 323, "lr": 2.765769726574441e-05} +{"train_loss": 0.016683708876371384, "grad_norm": 0.06851048022508621, "global_step": 34315, "epoch": 323, "lr": 2.765504588714023e-05} +{"train_loss": 0.013960614800453186, "grad_norm": 0.057943958789110184, "global_step": 34316, "epoch": 323, "lr": 2.7652394587046504e-05} +{"train_loss": 0.015559526160359383, "grad_norm": 0.06547898799180984, "global_step": 34317, "epoch": 323, "lr": 2.764974336547252e-05} +{"train_loss": 0.01629716344177723, "grad_norm": 0.08505614846944809, "global_step": 34318, "epoch": 323, "lr": 2.7647092222427616e-05} +{"train_loss": 0.016635514795780182, "grad_norm": 0.1157422736287117, "global_step": 34319, "epoch": 323, "lr": 2.764444115792113e-05} +{"train_loss": 0.015764139592647552, "grad_norm": 0.058587659150362015, "global_step": 34320, "epoch": 323, "lr": 2.7641790171962345e-05} +{"train_loss": 0.013101506046950817, "grad_norm": 0.09625057131052017, "global_step": 34321, "epoch": 323, "lr": 2.7639139264560578e-05} +{"train_loss": 0.013701101765036583, "grad_norm": 0.15714047849178314, "global_step": 34322, "epoch": 323, "lr": 2.7636488435725172e-05} +{"train_loss": 0.017330581322312355, "grad_norm": 0.05947680398821831, "global_step": 34323, "epoch": 323, "lr": 2.7633837685465418e-05} +{"train_loss": 0.015322379767894745, "grad_norm": 0.09630092978477478, "global_step": 34324, "epoch": 323, "lr": 2.7631187013790616e-05} +{"train_loss": 0.014804304577410221, "grad_norm": 0.12703391909599304, "global_step": 34325, "epoch": 323, "lr": 2.7628536420710094e-05} +{"train_loss": 0.017298681661486626, "grad_norm": 0.09919028729200363, "global_step": 34326, "epoch": 323, "lr": 2.762588590623318e-05} +{"train_loss": 0.01524575799703598, "grad_norm": 0.09512761235237122, "global_step": 34327, "epoch": 323, "lr": 2.762323547036916e-05} +{"train_loss": 0.013629566878080368, "grad_norm": 0.04757342487573624, "global_step": 34328, "epoch": 323, "lr": 2.7620585113127373e-05} +{"train_loss": 0.01401321031153202, "grad_norm": 0.09360253065824509, "global_step": 34329, "epoch": 323, "lr": 2.7617934834517122e-05} +{"train_loss": 0.01292611937969923, "grad_norm": 0.09352635592222214, "global_step": 34330, "epoch": 323, "lr": 2.7615284634547688e-05} +{"train_loss": 0.015059988014400005, "grad_norm": 0.06248624622821808, "global_step": 34331, "epoch": 323, "lr": 2.761263451322842e-05} +{"train_loss": 0.015958836302161217, "grad_norm": 0.11928455531597137, "global_step": 34332, "epoch": 323, "lr": 2.760998447056863e-05} +{"train_loss": 0.014132044278085232, "grad_norm": 0.07121382653713226, "global_step": 34333, "epoch": 323, "lr": 2.7607334506577597e-05} +{"train_loss": 0.015108122490346432, "grad_norm": 0.08368460834026337, "global_step": 34334, "epoch": 323, "lr": 2.7604684621264677e-05} +{"train_loss": 0.01306786946952343, "grad_norm": 0.13483433425426483, "global_step": 34335, "epoch": 323, "lr": 2.7602034814639132e-05} +{"train_loss": 0.015406431630253792, "grad_norm": 0.10280890762805939, "global_step": 34336, "epoch": 323, "lr": 2.7599385086710317e-05} +{"train_loss": 0.01482113916426897, "grad_norm": 0.0807822197675705, "global_step": 34337, "epoch": 323, "lr": 2.75967354374875e-05} +{"train_loss": 0.01457981951534748, "grad_norm": 0.06293419003486633, "global_step": 34338, "epoch": 323, "lr": 2.7594085866980014e-05} +{"train_loss": 0.014459439553320408, "grad_norm": 0.09601133316755295, "global_step": 34339, "epoch": 323, "lr": 2.7591436375197188e-05} +{"train_loss": 0.015408600680530071, "grad_norm": 0.07307369261980057, "global_step": 34340, "epoch": 323, "lr": 2.7588786962148307e-05} +{"train_loss": 0.016041871160268784, "grad_norm": 0.09378533065319061, "global_step": 34341, "epoch": 323, "lr": 2.758613762784266e-05} +{"train_loss": 0.016159754246473312, "grad_norm": 0.07108434289693832, "global_step": 34342, "epoch": 323, "lr": 2.758348837228958e-05} +{"train_loss": 0.015194342875818037, "grad_norm": 0.08621884137392044, "global_step": 34343, "epoch": 323, "lr": 2.7580839195498398e-05, "val_loss": 0.02065584994852543} +{"train_loss": 0.016963090747594833, "grad_norm": 0.15269163250923157, "global_step": 34344, "epoch": 324, "lr": 2.7578190097478367e-05} +{"train_loss": 0.01415605004876852, "grad_norm": 0.08327504247426987, "global_step": 34345, "epoch": 324, "lr": 2.7575541078238854e-05} +{"train_loss": 0.014308653771877289, "grad_norm": 0.09426518529653549, "global_step": 34346, "epoch": 324, "lr": 2.757289213778913e-05} +{"train_loss": 0.014681999571621418, "grad_norm": 0.08417928218841553, "global_step": 34347, "epoch": 324, "lr": 2.7570243276138497e-05} +{"train_loss": 0.01729154959321022, "grad_norm": 0.14904135465621948, "global_step": 34348, "epoch": 324, "lr": 2.7567594493296268e-05} +{"train_loss": 0.01623496785759926, "grad_norm": 0.08101372420787811, "global_step": 34349, "epoch": 324, "lr": 2.756494578927178e-05} +{"train_loss": 0.011747242882847786, "grad_norm": 0.05481068044900894, "global_step": 34350, "epoch": 324, "lr": 2.7562297164074297e-05} +{"train_loss": 0.014057392254471779, "grad_norm": 0.0984446257352829, "global_step": 34351, "epoch": 324, "lr": 2.7559648617713164e-05} +{"train_loss": 0.014054856263101101, "grad_norm": 0.07358986139297485, "global_step": 34352, "epoch": 324, "lr": 2.755700015019764e-05} +{"train_loss": 0.015703603625297546, "grad_norm": 0.07713048905134201, "global_step": 34353, "epoch": 324, "lr": 2.7554351761537085e-05} +{"train_loss": 0.016168195754289627, "grad_norm": 0.0739465206861496, "global_step": 34354, "epoch": 324, "lr": 2.755170345174075e-05} +{"train_loss": 0.01436745747923851, "grad_norm": 0.05963362380862236, "global_step": 34355, "epoch": 324, "lr": 2.7549055220817976e-05} +{"train_loss": 0.016647156327962875, "grad_norm": 0.11278511583805084, "global_step": 34356, "epoch": 324, "lr": 2.754640706877807e-05} +{"train_loss": 0.013184035196900368, "grad_norm": 0.114765964448452, "global_step": 34357, "epoch": 324, "lr": 2.754375899563033e-05} +{"train_loss": 0.014346509240567684, "grad_norm": 0.0638832226395607, "global_step": 34358, "epoch": 324, "lr": 2.7541111001384035e-05} +{"train_loss": 0.012250298634171486, "grad_norm": 0.048104044049978256, "global_step": 34359, "epoch": 324, "lr": 2.7538463086048504e-05} +{"train_loss": 0.015731293708086014, "grad_norm": 0.114503875374794, "global_step": 34360, "epoch": 324, "lr": 2.7535815249633067e-05} +{"train_loss": 0.012494266033172607, "grad_norm": 0.11271942406892776, "global_step": 34361, "epoch": 324, "lr": 2.7533167492146982e-05} +{"train_loss": 0.01776759885251522, "grad_norm": 0.13001692295074463, "global_step": 34362, "epoch": 324, "lr": 2.7530519813599587e-05} +{"train_loss": 0.015323267318308353, "grad_norm": 0.12733034789562225, "global_step": 34363, "epoch": 324, "lr": 2.7527872214000182e-05} +{"train_loss": 0.014763206243515015, "grad_norm": 0.08372625708580017, "global_step": 34364, "epoch": 324, "lr": 2.7525224693358065e-05} +{"train_loss": 0.015516898594796658, "grad_norm": 0.08119523525238037, "global_step": 34365, "epoch": 324, "lr": 2.7522577251682518e-05} +{"train_loss": 0.015009433962404728, "grad_norm": 0.07860472798347473, "global_step": 34366, "epoch": 324, "lr": 2.7519929888982854e-05} +{"train_loss": 0.013685121200978756, "grad_norm": 0.07992903888225555, "global_step": 34367, "epoch": 324, "lr": 2.75172826052684e-05} +{"train_loss": 0.014437245205044746, "grad_norm": 0.07919321209192276, "global_step": 34368, "epoch": 324, "lr": 2.7514635400548428e-05} +{"train_loss": 0.01466456614434719, "grad_norm": 0.08232937008142471, "global_step": 34369, "epoch": 324, "lr": 2.7511988274832258e-05} +{"train_loss": 0.013309955596923828, "grad_norm": 0.0706169381737709, "global_step": 34370, "epoch": 324, "lr": 2.750934122812918e-05} +{"train_loss": 0.014408349990844727, "grad_norm": 0.05717794597148895, "global_step": 34371, "epoch": 324, "lr": 2.7506694260448485e-05} +{"train_loss": 0.015866398811340332, "grad_norm": 0.11352813988924026, "global_step": 34372, "epoch": 324, "lr": 2.750404737179948e-05} +{"train_loss": 0.016334086656570435, "grad_norm": 0.08886255323886871, "global_step": 34373, "epoch": 324, "lr": 2.750140056219149e-05} +{"train_loss": 0.01505508553236723, "grad_norm": 0.08879970014095306, "global_step": 34374, "epoch": 324, "lr": 2.7498753831633773e-05} +{"train_loss": 0.015273235738277435, "grad_norm": 0.09180247038602829, "global_step": 34375, "epoch": 324, "lr": 2.7496107180135666e-05} +{"train_loss": 0.015737121924757957, "grad_norm": 0.12940335273742676, "global_step": 34376, "epoch": 324, "lr": 2.7493460607706435e-05} +{"train_loss": 0.015101917088031769, "grad_norm": 0.09954255074262619, "global_step": 34377, "epoch": 324, "lr": 2.7490814114355417e-05} +{"train_loss": 0.01630396395921707, "grad_norm": 0.1418464481830597, "global_step": 34378, "epoch": 324, "lr": 2.7488167700091872e-05} +{"train_loss": 0.015095947310328484, "grad_norm": 0.05132199823856354, "global_step": 34379, "epoch": 324, "lr": 2.748552136492511e-05} +{"train_loss": 0.015545760281383991, "grad_norm": 0.06728993356227875, "global_step": 34380, "epoch": 324, "lr": 2.7482875108864458e-05} +{"train_loss": 0.014497380703687668, "grad_norm": 0.06666598469018936, "global_step": 34381, "epoch": 324, "lr": 2.7480228931919182e-05} +{"train_loss": 0.016843654215335846, "grad_norm": 0.07355468720197678, "global_step": 34382, "epoch": 324, "lr": 2.747758283409858e-05} +{"train_loss": 0.015587824396789074, "grad_norm": 0.08023045212030411, "global_step": 34383, "epoch": 324, "lr": 2.747493681541195e-05} +{"train_loss": 0.018406616523861885, "grad_norm": 0.14392274618148804, "global_step": 34384, "epoch": 324, "lr": 2.7472290875868616e-05} +{"train_loss": 0.016018599271774292, "grad_norm": 0.06735745817422867, "global_step": 34385, "epoch": 324, "lr": 2.746964501547784e-05} +{"train_loss": 0.015182003378868103, "grad_norm": 0.13877947628498077, "global_step": 34386, "epoch": 324, "lr": 2.7466999234248924e-05} +{"train_loss": 0.01354068424552679, "grad_norm": 0.1080036610364914, "global_step": 34387, "epoch": 324, "lr": 2.7464353532191223e-05} +{"train_loss": 0.015556718222796917, "grad_norm": 0.090095154941082, "global_step": 34388, "epoch": 324, "lr": 2.746170790931394e-05} +{"train_loss": 0.01799873262643814, "grad_norm": 0.10729126632213593, "global_step": 34389, "epoch": 324, "lr": 2.745906236562641e-05} +{"train_loss": 0.014602549374103546, "grad_norm": 0.06256213784217834, "global_step": 34390, "epoch": 324, "lr": 2.745641690113796e-05} +{"train_loss": 0.015230491757392883, "grad_norm": 0.07111617177724838, "global_step": 34391, "epoch": 324, "lr": 2.745377151585783e-05} +{"train_loss": 0.014617630280554295, "grad_norm": 0.09749279916286469, "global_step": 34392, "epoch": 324, "lr": 2.7451126209795363e-05} +{"train_loss": 0.013034021481871605, "grad_norm": 0.07167541980743408, "global_step": 34393, "epoch": 324, "lr": 2.7448480982959813e-05} +{"train_loss": 0.014481743797659874, "grad_norm": 0.07547558844089508, "global_step": 34394, "epoch": 324, "lr": 2.744583583536051e-05} +{"train_loss": 0.014816473238170147, "grad_norm": 0.09186170995235443, "global_step": 34395, "epoch": 324, "lr": 2.7443190767006722e-05} +{"train_loss": 0.014656870625913143, "grad_norm": 0.06753812730312347, "global_step": 34396, "epoch": 324, "lr": 2.7440545777907746e-05} +{"train_loss": 0.013867520727217197, "grad_norm": 0.11188460141420364, "global_step": 34397, "epoch": 324, "lr": 2.7437900868072897e-05} +{"train_loss": 0.017169443890452385, "grad_norm": 0.13927797973155975, "global_step": 34398, "epoch": 324, "lr": 2.7435256037511458e-05} +{"train_loss": 0.01814090833067894, "grad_norm": 0.060454778373241425, "global_step": 34399, "epoch": 324, "lr": 2.743261128623269e-05} +{"train_loss": 0.016434503719210625, "grad_norm": 0.08373651653528214, "global_step": 34400, "epoch": 324, "lr": 2.7429966614245915e-05} +{"train_loss": 0.013478163629770279, "grad_norm": 0.07286887615919113, "global_step": 34401, "epoch": 324, "lr": 2.7427322021560442e-05} +{"train_loss": 0.014151434414088726, "grad_norm": 0.11634767800569534, "global_step": 34402, "epoch": 324, "lr": 2.7424677508185527e-05} +{"train_loss": 0.015171731822192669, "grad_norm": 0.11918078362941742, "global_step": 34403, "epoch": 324, "lr": 2.742203307413047e-05} +{"train_loss": 0.015725040808320045, "grad_norm": 0.05304267257452011, "global_step": 34404, "epoch": 324, "lr": 2.74193887194046e-05} +{"train_loss": 0.01217992976307869, "grad_norm": 0.10833047330379486, "global_step": 34405, "epoch": 324, "lr": 2.7416744444017166e-05} +{"train_loss": 0.015136511996388435, "grad_norm": 0.1289835274219513, "global_step": 34406, "epoch": 324, "lr": 2.741410024797746e-05} +{"train_loss": 0.016345379874110222, "grad_norm": 0.06496129930019379, "global_step": 34407, "epoch": 324, "lr": 2.741145613129478e-05} +{"train_loss": 0.016630494967103004, "grad_norm": 0.10514447093009949, "global_step": 34408, "epoch": 324, "lr": 2.740881209397844e-05} +{"train_loss": 0.017161620780825615, "grad_norm": 0.06551665812730789, "global_step": 34409, "epoch": 324, "lr": 2.740616813603769e-05} +{"train_loss": 0.01401401124894619, "grad_norm": 0.06328555941581726, "global_step": 34410, "epoch": 324, "lr": 2.7403524257481856e-05} +{"train_loss": 0.013399045914411545, "grad_norm": 0.07223924994468689, "global_step": 34411, "epoch": 324, "lr": 2.740088045832021e-05} +{"train_loss": 0.014506415463984013, "grad_norm": 0.13427093625068665, "global_step": 34412, "epoch": 324, "lr": 2.7398236738562026e-05} +{"train_loss": 0.014535550028085709, "grad_norm": 0.09539506584405899, "global_step": 34413, "epoch": 324, "lr": 2.739559309821661e-05} +{"train_loss": 0.01541051547974348, "grad_norm": 0.11444131284952164, "global_step": 34414, "epoch": 324, "lr": 2.7392949537293267e-05} +{"train_loss": 0.012990661896765232, "grad_norm": 0.08268094807863235, "global_step": 34415, "epoch": 324, "lr": 2.739030605580125e-05} +{"train_loss": 0.018059205263853073, "grad_norm": 0.10794710367918015, "global_step": 34416, "epoch": 324, "lr": 2.738766265374988e-05} +{"train_loss": 0.01529952883720398, "grad_norm": 0.06333247572183609, "global_step": 34417, "epoch": 324, "lr": 2.7385019331148417e-05} +{"train_loss": 0.014932049438357353, "grad_norm": 0.07831043750047684, "global_step": 34418, "epoch": 324, "lr": 2.738237608800618e-05} +{"train_loss": 0.016301792114973068, "grad_norm": 0.1430332064628601, "global_step": 34419, "epoch": 324, "lr": 2.7379732924332414e-05} +{"train_loss": 0.016318678855895996, "grad_norm": 0.062487389892339706, "global_step": 34420, "epoch": 324, "lr": 2.7377089840136428e-05} +{"train_loss": 0.015194473788142204, "grad_norm": 0.06370587646961212, "global_step": 34421, "epoch": 324, "lr": 2.7374446835427524e-05} +{"train_loss": 0.0131561029702425, "grad_norm": 0.08454067260026932, "global_step": 34422, "epoch": 324, "lr": 2.7371803910214987e-05} +{"train_loss": 0.012950677424669266, "grad_norm": 0.10336731374263763, "global_step": 34423, "epoch": 324, "lr": 2.7369161064508063e-05} +{"train_loss": 0.015534574165940285, "grad_norm": 0.0633169636130333, "global_step": 34424, "epoch": 324, "lr": 2.736651829831607e-05} +{"train_loss": 0.015359820798039436, "grad_norm": 0.050302401185035706, "global_step": 34425, "epoch": 324, "lr": 2.7363875611648304e-05} +{"train_loss": 0.013055339455604553, "grad_norm": 0.09487423300743103, "global_step": 34426, "epoch": 324, "lr": 2.7361233004514013e-05} +{"train_loss": 0.015958677977323532, "grad_norm": 0.09198164939880371, "global_step": 34427, "epoch": 324, "lr": 2.7358590476922506e-05} +{"train_loss": 0.015538733452558517, "grad_norm": 0.08145955204963684, "global_step": 34428, "epoch": 324, "lr": 2.7355948028883103e-05} +{"train_loss": 0.01496485061943531, "grad_norm": 0.08461479842662811, "global_step": 34429, "epoch": 324, "lr": 2.735330566040502e-05} +{"train_loss": 0.012257494032382965, "grad_norm": 0.12530922889709473, "global_step": 34430, "epoch": 324, "lr": 2.7350663371497566e-05} +{"train_loss": 0.014577675610780716, "grad_norm": 0.09181882441043854, "global_step": 34431, "epoch": 324, "lr": 2.7348021162170055e-05} +{"train_loss": 0.01635724864900112, "grad_norm": 0.0975811555981636, "global_step": 34432, "epoch": 324, "lr": 2.7345379032431723e-05} +{"train_loss": 0.015347030945122242, "grad_norm": 0.10708591341972351, "global_step": 34433, "epoch": 324, "lr": 2.7342736982291878e-05} +{"train_loss": 0.017487958073616028, "grad_norm": 0.07185101509094238, "global_step": 34434, "epoch": 324, "lr": 2.7340095011759818e-05} +{"train_loss": 0.013748613186180592, "grad_norm": 0.057214654982089996, "global_step": 34435, "epoch": 324, "lr": 2.7337453120844813e-05} +{"train_loss": 0.017207743600010872, "grad_norm": 0.07262367755174637, "global_step": 34436, "epoch": 324, "lr": 2.7334811309556124e-05} +{"train_loss": 0.018246807157993317, "grad_norm": 0.17928852140903473, "global_step": 34437, "epoch": 324, "lr": 2.733216957790305e-05} +{"train_loss": 0.017296740785241127, "grad_norm": 0.16876854002475739, "global_step": 34438, "epoch": 324, "lr": 2.7329527925894893e-05} +{"train_loss": 0.01718120649456978, "grad_norm": 0.10243295133113861, "global_step": 34439, "epoch": 324, "lr": 2.7326886353540916e-05} +{"train_loss": 0.014679509215056896, "grad_norm": 0.09103616327047348, "global_step": 34440, "epoch": 324, "lr": 2.7324244860850378e-05} +{"train_loss": 0.013301584869623184, "grad_norm": 0.11018990725278854, "global_step": 34441, "epoch": 324, "lr": 2.7321603447832588e-05} +{"train_loss": 0.016995470970869064, "grad_norm": 0.08670633286237717, "global_step": 34442, "epoch": 324, "lr": 2.7318962114496838e-05} +{"train_loss": 0.01618427410721779, "grad_norm": 0.07239830493927002, "global_step": 34443, "epoch": 324, "lr": 2.7316320860852364e-05} +{"train_loss": 0.013534175232052803, "grad_norm": 0.06131694093346596, "global_step": 34444, "epoch": 324, "lr": 2.7313679686908482e-05} +{"train_loss": 0.01496492326259613, "grad_norm": 0.11557929217815399, "global_step": 34445, "epoch": 324, "lr": 2.7311038592674497e-05} +{"train_loss": 0.016486871987581253, "grad_norm": 0.06579525023698807, "global_step": 34446, "epoch": 324, "lr": 2.7308397578159627e-05} +{"train_loss": 0.01645306497812271, "grad_norm": 0.05913893133401871, "global_step": 34447, "epoch": 324, "lr": 2.730575664337317e-05} +{"train_loss": 0.015820227563381195, "grad_norm": 0.1176234781742096, "global_step": 34448, "epoch": 324, "lr": 2.730311578832444e-05} +{"train_loss": 0.015190846796305675, "grad_norm": 0.1323021799325943, "global_step": 34449, "epoch": 324, "lr": 2.7300475013022663e-05, "val_loss": 0.01964270882308483} +{"train_loss": 0.0161263570189476, "grad_norm": 0.09256311506032944, "global_step": 34450, "epoch": 325, "lr": 2.7297834317477155e-05} +{"train_loss": 0.01385507546365261, "grad_norm": 0.08052222430706024, "global_step": 34451, "epoch": 325, "lr": 2.7295193701697198e-05} +{"train_loss": 0.015481317415833473, "grad_norm": 0.06982877850532532, "global_step": 34452, "epoch": 325, "lr": 2.7292553165692057e-05} +{"train_loss": 0.012916154228150845, "grad_norm": 0.09645324945449829, "global_step": 34453, "epoch": 325, "lr": 2.728991270947099e-05} +{"train_loss": 0.016558794304728508, "grad_norm": 0.07409923523664474, "global_step": 34454, "epoch": 325, "lr": 2.728727233304329e-05} +{"train_loss": 0.01588749885559082, "grad_norm": 0.13633522391319275, "global_step": 34455, "epoch": 325, "lr": 2.7284632036418257e-05} +{"train_loss": 0.014694293960928917, "grad_norm": 0.09492358565330505, "global_step": 34456, "epoch": 325, "lr": 2.7281991819605136e-05} +{"train_loss": 0.015240823850035667, "grad_norm": 0.10687077045440674, "global_step": 34457, "epoch": 325, "lr": 2.727935168261323e-05} +{"train_loss": 0.017111526802182198, "grad_norm": 0.12773993611335754, "global_step": 34458, "epoch": 325, "lr": 2.7276711625451784e-05} +{"train_loss": 0.015391608700156212, "grad_norm": 0.12168664485216141, "global_step": 34459, "epoch": 325, "lr": 2.7274071648130107e-05} +{"train_loss": 0.018299806863069534, "grad_norm": 0.08773966878652573, "global_step": 34460, "epoch": 325, "lr": 2.7271431750657443e-05} +{"train_loss": 0.01709797978401184, "grad_norm": 0.10182024538516998, "global_step": 34461, "epoch": 325, "lr": 2.726879193304308e-05} +{"train_loss": 0.013201046735048294, "grad_norm": 0.06461741775274277, "global_step": 34462, "epoch": 325, "lr": 2.7266152195296323e-05} +{"train_loss": 0.015693239867687225, "grad_norm": 0.1495985984802246, "global_step": 34463, "epoch": 325, "lr": 2.7263512537426417e-05} +{"train_loss": 0.016148142516613007, "grad_norm": 0.06402361392974854, "global_step": 34464, "epoch": 325, "lr": 2.7260872959442617e-05} +{"train_loss": 0.015186343342065811, "grad_norm": 0.18465381860733032, "global_step": 34465, "epoch": 325, "lr": 2.725823346135422e-05} +{"train_loss": 0.013825749978423119, "grad_norm": 0.0711783841252327, "global_step": 34466, "epoch": 325, "lr": 2.725559404317053e-05} +{"train_loss": 0.015439597889780998, "grad_norm": 0.07442688196897507, "global_step": 34467, "epoch": 325, "lr": 2.7252954704900763e-05} +{"train_loss": 0.01551731675863266, "grad_norm": 0.08910517394542694, "global_step": 34468, "epoch": 325, "lr": 2.725031544655422e-05} +{"train_loss": 0.014724303968250751, "grad_norm": 0.06980767101049423, "global_step": 34469, "epoch": 325, "lr": 2.724767626814022e-05} +{"train_loss": 0.014492695219814777, "grad_norm": 0.07214657217264175, "global_step": 34470, "epoch": 325, "lr": 2.724503716966795e-05} +{"train_loss": 0.016236966475844383, "grad_norm": 0.13667164742946625, "global_step": 34471, "epoch": 325, "lr": 2.724239815114672e-05} +{"train_loss": 0.015949811786413193, "grad_norm": 0.11286725848913193, "global_step": 34472, "epoch": 325, "lr": 2.7239759212585826e-05} +{"train_loss": 0.017007362097501755, "grad_norm": 0.20564498007297516, "global_step": 34473, "epoch": 325, "lr": 2.7237120353994505e-05} +{"train_loss": 0.014653543941676617, "grad_norm": 0.076747365295887, "global_step": 34474, "epoch": 325, "lr": 2.7234481575382037e-05} +{"train_loss": 0.015394031070172787, "grad_norm": 0.08577226102352142, "global_step": 34475, "epoch": 325, "lr": 2.7231842876757723e-05} +{"train_loss": 0.014453485608100891, "grad_norm": 0.06056411564350128, "global_step": 34476, "epoch": 325, "lr": 2.7229204258130815e-05} +{"train_loss": 0.014751504175364971, "grad_norm": 0.08962411433458328, "global_step": 34477, "epoch": 325, "lr": 2.7226565719510554e-05} +{"train_loss": 0.012580696493387222, "grad_norm": 0.08138695359230042, "global_step": 34478, "epoch": 325, "lr": 2.7223927260906245e-05} +{"train_loss": 0.012804749421775341, "grad_norm": 0.09189854562282562, "global_step": 34479, "epoch": 325, "lr": 2.7221288882327167e-05} +{"train_loss": 0.015491544269025326, "grad_norm": 0.0706324651837349, "global_step": 34480, "epoch": 325, "lr": 2.721865058378255e-05} +{"train_loss": 0.01432595495134592, "grad_norm": 0.06261380761861801, "global_step": 34481, "epoch": 325, "lr": 2.7216012365281713e-05} +{"train_loss": 0.01410263404250145, "grad_norm": 0.09397372603416443, "global_step": 34482, "epoch": 325, "lr": 2.7213374226833876e-05} +{"train_loss": 0.014755653217434883, "grad_norm": 0.08089347183704376, "global_step": 34483, "epoch": 325, "lr": 2.7210736168448353e-05} +{"train_loss": 0.013906121253967285, "grad_norm": 0.08345927298069, "global_step": 34484, "epoch": 325, "lr": 2.7208098190134372e-05} +{"train_loss": 0.014997785910964012, "grad_norm": 0.07083707302808762, "global_step": 34485, "epoch": 325, "lr": 2.7205460291901224e-05} +{"train_loss": 0.015252267941832542, "grad_norm": 0.08199532330036163, "global_step": 34486, "epoch": 325, "lr": 2.7202822473758192e-05} +{"train_loss": 0.015524297952651978, "grad_norm": 0.07274369895458221, "global_step": 34487, "epoch": 325, "lr": 2.7200184735714536e-05} +{"train_loss": 0.014918037690222263, "grad_norm": 0.09103704243898392, "global_step": 34488, "epoch": 325, "lr": 2.719754707777949e-05} +{"train_loss": 0.015328955836594105, "grad_norm": 0.06228610500693321, "global_step": 34489, "epoch": 325, "lr": 2.7194909499962363e-05} +{"train_loss": 0.018086228519678116, "grad_norm": 0.07444008439779282, "global_step": 34490, "epoch": 325, "lr": 2.719227200227239e-05} +{"train_loss": 0.013921264559030533, "grad_norm": 0.05563310533761978, "global_step": 34491, "epoch": 325, "lr": 2.718963458471886e-05} +{"train_loss": 0.01300903782248497, "grad_norm": 0.10776915401220322, "global_step": 34492, "epoch": 325, "lr": 2.7186997247311046e-05} +{"train_loss": 0.015824919566512108, "grad_norm": 0.07067149877548218, "global_step": 34493, "epoch": 325, "lr": 2.7184359990058206e-05} +{"train_loss": 0.013853969052433968, "grad_norm": 0.057331036776304245, "global_step": 34494, "epoch": 325, "lr": 2.718172281296958e-05} +{"train_loss": 0.014965870417654514, "grad_norm": 0.08068937063217163, "global_step": 34495, "epoch": 325, "lr": 2.7179085716054458e-05} +{"train_loss": 0.013147955760359764, "grad_norm": 0.10121072828769684, "global_step": 34496, "epoch": 325, "lr": 2.717644869932213e-05} +{"train_loss": 0.015355450101196766, "grad_norm": 0.11182208359241486, "global_step": 34497, "epoch": 325, "lr": 2.717381176278181e-05} +{"train_loss": 0.013840528205037117, "grad_norm": 0.10157725214958191, "global_step": 34498, "epoch": 325, "lr": 2.717117490644281e-05} +{"train_loss": 0.013879689387977123, "grad_norm": 0.07880586385726929, "global_step": 34499, "epoch": 325, "lr": 2.716853813031435e-05} +{"train_loss": 0.016409169882535934, "grad_norm": 0.10092981159687042, "global_step": 34500, "epoch": 325, "lr": 2.716590143440574e-05} +{"train_loss": 0.01710812933743, "grad_norm": 0.10411639511585236, "global_step": 34501, "epoch": 325, "lr": 2.7163264818726197e-05} +{"train_loss": 0.014315871521830559, "grad_norm": 0.08571691066026688, "global_step": 34502, "epoch": 325, "lr": 2.7160628283285018e-05} +{"train_loss": 0.01382606104016304, "grad_norm": 0.09134085476398468, "global_step": 34503, "epoch": 325, "lr": 2.7157991828091476e-05} +{"train_loss": 0.01648683100938797, "grad_norm": 0.07894415408372879, "global_step": 34504, "epoch": 325, "lr": 2.7155355453154818e-05} +{"train_loss": 0.017598222941160202, "grad_norm": 0.07336448132991791, "global_step": 34505, "epoch": 325, "lr": 2.7152719158484286e-05} +{"train_loss": 0.016966454684734344, "grad_norm": 0.07305407524108887, "global_step": 34506, "epoch": 325, "lr": 2.7150082944089184e-05} +{"train_loss": 0.014066711068153381, "grad_norm": 0.06827395409345627, "global_step": 34507, "epoch": 325, "lr": 2.7147446809978728e-05} +{"train_loss": 0.015833575278520584, "grad_norm": 0.06073172390460968, "global_step": 34508, "epoch": 325, "lr": 2.7144810756162208e-05} +{"train_loss": 0.01615316793322563, "grad_norm": 0.0664573684334755, "global_step": 34509, "epoch": 325, "lr": 2.7142174782648887e-05} +{"train_loss": 0.015911847352981567, "grad_norm": 0.1481492966413498, "global_step": 34510, "epoch": 325, "lr": 2.7139538889448058e-05} +{"train_loss": 0.01429715845733881, "grad_norm": 0.06933896243572235, "global_step": 34511, "epoch": 325, "lr": 2.7136903076568908e-05} +{"train_loss": 0.01590207777917385, "grad_norm": 0.10824106633663177, "global_step": 34512, "epoch": 325, "lr": 2.7134267344020743e-05} +{"train_loss": 0.01751834712922573, "grad_norm": 0.0999588668346405, "global_step": 34513, "epoch": 325, "lr": 2.7131631691812836e-05} +{"train_loss": 0.016145776957273483, "grad_norm": 0.07918454706668854, "global_step": 34514, "epoch": 325, "lr": 2.7128996119954407e-05} +{"train_loss": 0.015454370528459549, "grad_norm": 0.06418463587760925, "global_step": 34515, "epoch": 325, "lr": 2.7126360628454746e-05} +{"train_loss": 0.017638778313994408, "grad_norm": 0.07263485342264175, "global_step": 34516, "epoch": 325, "lr": 2.712372521732312e-05} +{"train_loss": 0.015129514038562775, "grad_norm": 0.06199345737695694, "global_step": 34517, "epoch": 325, "lr": 2.7121089886568773e-05} +{"train_loss": 0.017719771713018417, "grad_norm": 0.08175584673881531, "global_step": 34518, "epoch": 325, "lr": 2.7118454636200953e-05} +{"train_loss": 0.01368161104619503, "grad_norm": 0.05955229699611664, "global_step": 34519, "epoch": 325, "lr": 2.711581946622893e-05} +{"train_loss": 0.015259205363690853, "grad_norm": 0.06227816268801689, "global_step": 34520, "epoch": 325, "lr": 2.711318437666198e-05} +{"train_loss": 0.01118466816842556, "grad_norm": 0.06372144818305969, "global_step": 34521, "epoch": 325, "lr": 2.7110549367509342e-05} +{"train_loss": 0.016730865463614464, "grad_norm": 0.0832132026553154, "global_step": 34522, "epoch": 325, "lr": 2.7107914438780285e-05} +{"train_loss": 0.016517117619514465, "grad_norm": 0.06427747011184692, "global_step": 34523, "epoch": 325, "lr": 2.7105279590484044e-05} +{"train_loss": 0.014616795815527439, "grad_norm": 0.10638003796339035, "global_step": 34524, "epoch": 325, "lr": 2.710264482262992e-05} +{"train_loss": 0.01800629124045372, "grad_norm": 0.1113903671503067, "global_step": 34525, "epoch": 325, "lr": 2.7100010135227115e-05} +{"train_loss": 0.01407812163233757, "grad_norm": 0.05405041575431824, "global_step": 34526, "epoch": 325, "lr": 2.709737552828492e-05} +{"train_loss": 0.01676136627793312, "grad_norm": 0.08908909559249878, "global_step": 34527, "epoch": 325, "lr": 2.7094741001812605e-05} +{"train_loss": 0.014280270785093307, "grad_norm": 0.10351724177598953, "global_step": 34528, "epoch": 325, "lr": 2.7092106555819412e-05} +{"train_loss": 0.016113169491291046, "grad_norm": 0.09061745554208755, "global_step": 34529, "epoch": 325, "lr": 2.708947219031458e-05} +{"train_loss": 0.014527148567140102, "grad_norm": 0.07106450200080872, "global_step": 34530, "epoch": 325, "lr": 2.708683790530739e-05} +{"train_loss": 0.01651149056851864, "grad_norm": 0.08726102113723755, "global_step": 34531, "epoch": 325, "lr": 2.7084203700807063e-05} +{"train_loss": 0.014181956648826599, "grad_norm": 0.05375375226140022, "global_step": 34532, "epoch": 325, "lr": 2.708156957682289e-05} +{"train_loss": 0.016239602118730545, "grad_norm": 0.08049110323190689, "global_step": 34533, "epoch": 325, "lr": 2.7078935533364126e-05} +{"train_loss": 0.015764273703098297, "grad_norm": 0.07136696577072144, "global_step": 34534, "epoch": 325, "lr": 2.707630157044002e-05} +{"train_loss": 0.015130302868783474, "grad_norm": 0.1119348555803299, "global_step": 34535, "epoch": 325, "lr": 2.70736676880598e-05} +{"train_loss": 0.01560062263160944, "grad_norm": 0.0800027847290039, "global_step": 34536, "epoch": 325, "lr": 2.7071033886232743e-05} +{"train_loss": 0.013324886560440063, "grad_norm": 0.09060393273830414, "global_step": 34537, "epoch": 325, "lr": 2.7068400164968116e-05} +{"train_loss": 0.014138351194560528, "grad_norm": 0.07658284902572632, "global_step": 34538, "epoch": 325, "lr": 2.7065766524275137e-05} +{"train_loss": 0.014201338402926922, "grad_norm": 0.09211943298578262, "global_step": 34539, "epoch": 325, "lr": 2.7063132964163085e-05} +{"train_loss": 0.015326458029448986, "grad_norm": 0.07084125280380249, "global_step": 34540, "epoch": 325, "lr": 2.7060499484641234e-05} +{"train_loss": 0.013508598320186138, "grad_norm": 0.09710191935300827, "global_step": 34541, "epoch": 325, "lr": 2.7057866085718803e-05} +{"train_loss": 0.014282379299402237, "grad_norm": 0.08683791011571884, "global_step": 34542, "epoch": 325, "lr": 2.7055232767405036e-05} +{"train_loss": 0.01407224778085947, "grad_norm": 0.050639715045690536, "global_step": 34543, "epoch": 325, "lr": 2.70525995297092e-05} +{"train_loss": 0.013780441135168076, "grad_norm": 0.08524280786514282, "global_step": 34544, "epoch": 325, "lr": 2.704996637264058e-05} +{"train_loss": 0.01353982649743557, "grad_norm": 0.0769241601228714, "global_step": 34545, "epoch": 325, "lr": 2.704733329620839e-05} +{"train_loss": 0.015130803920328617, "grad_norm": 0.0661519467830658, "global_step": 34546, "epoch": 325, "lr": 2.7044700300421867e-05} +{"train_loss": 0.015815671533346176, "grad_norm": 0.15692739188671112, "global_step": 34547, "epoch": 325, "lr": 2.7042067385290303e-05} +{"train_loss": 0.01667557656764984, "grad_norm": 0.09876860678195953, "global_step": 34548, "epoch": 325, "lr": 2.703943455082291e-05} +{"train_loss": 0.016801709309220314, "grad_norm": 0.09159047901630402, "global_step": 34549, "epoch": 325, "lr": 2.7036801797028965e-05} +{"train_loss": 0.01543596014380455, "grad_norm": 0.08652505278587341, "global_step": 34550, "epoch": 325, "lr": 2.703416912391773e-05} +{"train_loss": 0.0155864916741848, "grad_norm": 0.10847575962543488, "global_step": 34551, "epoch": 325, "lr": 2.7031536531498435e-05} +{"train_loss": 0.012754635885357857, "grad_norm": 0.07670033723115921, "global_step": 34552, "epoch": 325, "lr": 2.702890401978031e-05} +{"train_loss": 0.014788385480642319, "grad_norm": 0.11288068443536758, "global_step": 34553, "epoch": 325, "lr": 2.7026271588772635e-05} +{"train_loss": 0.015553014352917671, "grad_norm": 0.060324594378471375, "global_step": 34554, "epoch": 325, "lr": 2.702363923848466e-05} +{"train_loss": 0.015178456208405067, "grad_norm": 0.08233734965324402, "global_step": 34555, "epoch": 325, "lr": 2.702100696892561e-05, "val_loss": 0.019432861357927322} +{"train_loss": 0.013314476236701012, "grad_norm": 0.06670960038900375, "global_step": 34556, "epoch": 326, "lr": 2.7018374780104748e-05} +{"train_loss": 0.012542376294732094, "grad_norm": 0.09281881898641586, "global_step": 34557, "epoch": 326, "lr": 2.701574267203134e-05} +{"train_loss": 0.012500925920903683, "grad_norm": 0.0627337172627449, "global_step": 34558, "epoch": 326, "lr": 2.7013110644714623e-05} +{"train_loss": 0.013121678493916988, "grad_norm": 0.09608324617147446, "global_step": 34559, "epoch": 326, "lr": 2.701047869816381e-05} +{"train_loss": 0.015773873776197433, "grad_norm": 0.06769245117902756, "global_step": 34560, "epoch": 326, "lr": 2.7007846832388183e-05} +{"train_loss": 0.014968760311603546, "grad_norm": 0.0942748561501503, "global_step": 34561, "epoch": 326, "lr": 2.7005215047397004e-05} +{"train_loss": 0.015731552615761757, "grad_norm": 0.07167083024978638, "global_step": 34562, "epoch": 326, "lr": 2.7002583343199473e-05} +{"train_loss": 0.01694740727543831, "grad_norm": 0.08428753912448883, "global_step": 34563, "epoch": 326, "lr": 2.6999951719804883e-05} +{"train_loss": 0.016006730496883392, "grad_norm": 0.11024042218923569, "global_step": 34564, "epoch": 326, "lr": 2.699732017722245e-05} +{"train_loss": 0.014539925381541252, "grad_norm": 0.06273065507411957, "global_step": 34565, "epoch": 326, "lr": 2.699468871546144e-05} +{"train_loss": 0.016605546697974205, "grad_norm": 0.08021830022335052, "global_step": 34566, "epoch": 326, "lr": 2.6992057334531073e-05} +{"train_loss": 0.014745144173502922, "grad_norm": 0.10213792324066162, "global_step": 34567, "epoch": 326, "lr": 2.698942603444061e-05} +{"train_loss": 0.017235752195119858, "grad_norm": 0.08116758614778519, "global_step": 34568, "epoch": 326, "lr": 2.6986794815199322e-05} +{"train_loss": 0.01667291484773159, "grad_norm": 0.08536344766616821, "global_step": 34569, "epoch": 326, "lr": 2.6984163676816422e-05} +{"train_loss": 0.014267845079302788, "grad_norm": 0.05735240504145622, "global_step": 34570, "epoch": 326, "lr": 2.6981532619301143e-05} +{"train_loss": 0.015929974615573883, "grad_norm": 0.12043896317481995, "global_step": 34571, "epoch": 326, "lr": 2.697890164266277e-05} +{"train_loss": 0.01723613403737545, "grad_norm": 0.07647194713354111, "global_step": 34572, "epoch": 326, "lr": 2.6976270746910504e-05} +{"train_loss": 0.017548546195030212, "grad_norm": 0.07049006223678589, "global_step": 34573, "epoch": 326, "lr": 2.6973639932053607e-05} +{"train_loss": 0.01525111123919487, "grad_norm": 0.1237739846110344, "global_step": 34574, "epoch": 326, "lr": 2.697100919810135e-05} +{"train_loss": 0.014626417309045792, "grad_norm": 0.06262916326522827, "global_step": 34575, "epoch": 326, "lr": 2.696837854506295e-05} +{"train_loss": 0.01476677693426609, "grad_norm": 0.09177692234516144, "global_step": 34576, "epoch": 326, "lr": 2.696574797294763e-05} +{"train_loss": 0.013950757682323456, "grad_norm": 0.10010099411010742, "global_step": 34577, "epoch": 326, "lr": 2.6963117481764655e-05} +{"train_loss": 0.013820081017911434, "grad_norm": 0.04962959885597229, "global_step": 34578, "epoch": 326, "lr": 2.6960487071523287e-05} +{"train_loss": 0.014402247965335846, "grad_norm": 0.09061933308839798, "global_step": 34579, "epoch": 326, "lr": 2.6957856742232733e-05} +{"train_loss": 0.01613333262503147, "grad_norm": 0.059959929436445236, "global_step": 34580, "epoch": 326, "lr": 2.6955226493902242e-05} +{"train_loss": 0.01661141775548458, "grad_norm": 0.06754648685455322, "global_step": 34581, "epoch": 326, "lr": 2.695259632654109e-05} +{"train_loss": 0.015647727996110916, "grad_norm": 0.08939572423696518, "global_step": 34582, "epoch": 326, "lr": 2.6949966240158487e-05} +{"train_loss": 0.016112854704260826, "grad_norm": 0.05860156565904617, "global_step": 34583, "epoch": 326, "lr": 2.6947336234763652e-05} +{"train_loss": 0.015346059575676918, "grad_norm": 0.08624950796365738, "global_step": 34584, "epoch": 326, "lr": 2.6944706310365865e-05} +{"train_loss": 0.016748448833823204, "grad_norm": 0.10530861467123032, "global_step": 34585, "epoch": 326, "lr": 2.6942076466974365e-05} +{"train_loss": 0.01241428591310978, "grad_norm": 0.11049298197031021, "global_step": 34586, "epoch": 326, "lr": 2.6939446704598364e-05} +{"train_loss": 0.01509394682943821, "grad_norm": 0.071959488093853, "global_step": 34587, "epoch": 326, "lr": 2.693681702324714e-05} +{"train_loss": 0.013043687678873539, "grad_norm": 0.05765499919652939, "global_step": 34588, "epoch": 326, "lr": 2.6934187422929908e-05} +{"train_loss": 0.01720389537513256, "grad_norm": 0.09646228700876236, "global_step": 34589, "epoch": 326, "lr": 2.693155790365589e-05} +{"train_loss": 0.013493996113538742, "grad_norm": 0.15460534393787384, "global_step": 34590, "epoch": 326, "lr": 2.6928928465434344e-05} +{"train_loss": 0.01338177639991045, "grad_norm": 0.07564884424209595, "global_step": 34591, "epoch": 326, "lr": 2.692629910827453e-05} +{"train_loss": 0.015226146206259727, "grad_norm": 0.11170081794261932, "global_step": 34592, "epoch": 326, "lr": 2.6923669832185672e-05} +{"train_loss": 0.015109428204596043, "grad_norm": 0.0784926488995552, "global_step": 34593, "epoch": 326, "lr": 2.6921040637176982e-05} +{"train_loss": 0.01439222414046526, "grad_norm": 0.13541117310523987, "global_step": 34594, "epoch": 326, "lr": 2.6918411523257713e-05} +{"train_loss": 0.013059657998383045, "grad_norm": 0.1215999498963356, "global_step": 34595, "epoch": 326, "lr": 2.6915782490437136e-05} +{"train_loss": 0.015907038003206253, "grad_norm": 0.0832715556025505, "global_step": 34596, "epoch": 326, "lr": 2.6913153538724433e-05} +{"train_loss": 0.016596972942352295, "grad_norm": 0.08116818219423294, "global_step": 34597, "epoch": 326, "lr": 2.6910524668128867e-05} +{"train_loss": 0.014345066621899605, "grad_norm": 0.10291509330272675, "global_step": 34598, "epoch": 326, "lr": 2.6907895878659707e-05} +{"train_loss": 0.01352480798959732, "grad_norm": 0.06574776023626328, "global_step": 34599, "epoch": 326, "lr": 2.6905267170326143e-05} +{"train_loss": 0.01641671359539032, "grad_norm": 0.11786483973264694, "global_step": 34600, "epoch": 326, "lr": 2.6902638543137414e-05} +{"train_loss": 0.017463892698287964, "grad_norm": 0.10948650538921356, "global_step": 34601, "epoch": 326, "lr": 2.6900009997102772e-05} +{"train_loss": 0.015614197589457035, "grad_norm": 0.11351887881755829, "global_step": 34602, "epoch": 326, "lr": 2.689738153223147e-05} +{"train_loss": 0.014731429517269135, "grad_norm": 0.16638217866420746, "global_step": 34603, "epoch": 326, "lr": 2.6894753148532693e-05} +{"train_loss": 0.013653900474309921, "grad_norm": 0.0566512793302536, "global_step": 34604, "epoch": 326, "lr": 2.689212484601573e-05} +{"train_loss": 0.016407158225774765, "grad_norm": 0.07337186485528946, "global_step": 34605, "epoch": 326, "lr": 2.688949662468979e-05} +{"train_loss": 0.015081189572811127, "grad_norm": 0.08106349408626556, "global_step": 34606, "epoch": 326, "lr": 2.6886868484564094e-05} +{"train_loss": 0.016395270824432373, "grad_norm": 0.13192681968212128, "global_step": 34607, "epoch": 326, "lr": 2.688424042564788e-05} +{"train_loss": 0.013722194358706474, "grad_norm": 0.06553711742162704, "global_step": 34608, "epoch": 326, "lr": 2.6881612447950423e-05} +{"train_loss": 0.019206585362553596, "grad_norm": 0.0805431455373764, "global_step": 34609, "epoch": 326, "lr": 2.6878984551480903e-05} +{"train_loss": 0.014721252955496311, "grad_norm": 0.06426586210727692, "global_step": 34610, "epoch": 326, "lr": 2.6876356736248592e-05} +{"train_loss": 0.01596345752477646, "grad_norm": 0.09380584955215454, "global_step": 34611, "epoch": 326, "lr": 2.6873729002262692e-05} +{"train_loss": 0.017300348728895187, "grad_norm": 0.12201271206140518, "global_step": 34612, "epoch": 326, "lr": 2.687110134953247e-05} +{"train_loss": 0.01728546805679798, "grad_norm": 0.13344787061214447, "global_step": 34613, "epoch": 326, "lr": 2.6868473778067116e-05} +{"train_loss": 0.018471186980605125, "grad_norm": 0.07836572080850601, "global_step": 34614, "epoch": 326, "lr": 2.6865846287875886e-05} +{"train_loss": 0.015140687115490437, "grad_norm": 0.06394441425800323, "global_step": 34615, "epoch": 326, "lr": 2.6863218878968034e-05} +{"train_loss": 0.015109294094145298, "grad_norm": 0.0822560116648674, "global_step": 34616, "epoch": 326, "lr": 2.6860591551352767e-05} +{"train_loss": 0.015522073954343796, "grad_norm": 0.06399296224117279, "global_step": 34617, "epoch": 326, "lr": 2.6857964305039296e-05} +{"train_loss": 0.01563400775194168, "grad_norm": 0.1353275030851364, "global_step": 34618, "epoch": 326, "lr": 2.685533714003688e-05} +{"train_loss": 0.015062717720866203, "grad_norm": 0.0771125853061676, "global_step": 34619, "epoch": 326, "lr": 2.6852710056354757e-05} +{"train_loss": 0.01485161017626524, "grad_norm": 0.14357902109622955, "global_step": 34620, "epoch": 326, "lr": 2.6850083054002127e-05} +{"train_loss": 0.015942759811878204, "grad_norm": 0.11222779005765915, "global_step": 34621, "epoch": 326, "lr": 2.6847456132988248e-05} +{"train_loss": 0.01457321085035801, "grad_norm": 0.13763797283172607, "global_step": 34622, "epoch": 326, "lr": 2.684482929332235e-05} +{"train_loss": 0.014094160869717598, "grad_norm": 0.11165259778499603, "global_step": 34623, "epoch": 326, "lr": 2.684220253501365e-05} +{"train_loss": 0.01688612438738346, "grad_norm": 0.10409791767597198, "global_step": 34624, "epoch": 326, "lr": 2.6839575858071363e-05} +{"train_loss": 0.013560224324464798, "grad_norm": 0.05723238363862038, "global_step": 34625, "epoch": 326, "lr": 2.683694926250473e-05} +{"train_loss": 0.01436255406588316, "grad_norm": 0.06385556608438492, "global_step": 34626, "epoch": 326, "lr": 2.6834322748323006e-05} +{"train_loss": 0.014351148158311844, "grad_norm": 0.07441019266843796, "global_step": 34627, "epoch": 326, "lr": 2.683169631553538e-05} +{"train_loss": 0.014229067601263523, "grad_norm": 0.08191728591918945, "global_step": 34628, "epoch": 326, "lr": 2.6829069964151114e-05} +{"train_loss": 0.014900876209139824, "grad_norm": 0.10975710302591324, "global_step": 34629, "epoch": 326, "lr": 2.6826443694179427e-05} +{"train_loss": 0.013747608289122581, "grad_norm": 0.07369352877140045, "global_step": 34630, "epoch": 326, "lr": 2.6823817505629513e-05} +{"train_loss": 0.01459801010787487, "grad_norm": 0.09914571046829224, "global_step": 34631, "epoch": 326, "lr": 2.6821191398510626e-05} +{"train_loss": 0.015044566243886948, "grad_norm": 0.09585294872522354, "global_step": 34632, "epoch": 326, "lr": 2.6818565372832006e-05} +{"train_loss": 0.014117034152150154, "grad_norm": 0.07256841659545898, "global_step": 34633, "epoch": 326, "lr": 2.6815939428602854e-05} +{"train_loss": 0.015160025097429752, "grad_norm": 0.09386467188596725, "global_step": 34634, "epoch": 326, "lr": 2.681331356583242e-05} +{"train_loss": 0.013427033089101315, "grad_norm": 0.09673171490430832, "global_step": 34635, "epoch": 326, "lr": 2.681068778452991e-05} +{"train_loss": 0.014704529196023941, "grad_norm": 0.059505756944417953, "global_step": 34636, "epoch": 326, "lr": 2.6808062084704573e-05} +{"train_loss": 0.014536837115883827, "grad_norm": 0.060171954333782196, "global_step": 34637, "epoch": 326, "lr": 2.6805436466365598e-05} +{"train_loss": 0.016467541456222534, "grad_norm": 0.13212326169013977, "global_step": 34638, "epoch": 326, "lr": 2.6802810929522238e-05} +{"train_loss": 0.01584719493985176, "grad_norm": 0.1076926663517952, "global_step": 34639, "epoch": 326, "lr": 2.6800185474183725e-05} +{"train_loss": 0.013662582263350487, "grad_norm": 0.06278388202190399, "global_step": 34640, "epoch": 326, "lr": 2.6797560100359275e-05} +{"train_loss": 0.014303449541330338, "grad_norm": 0.11943162977695465, "global_step": 34641, "epoch": 326, "lr": 2.679493480805808e-05} +{"train_loss": 0.013912845402956009, "grad_norm": 0.13498692214488983, "global_step": 34642, "epoch": 326, "lr": 2.6792309597289405e-05} +{"train_loss": 0.01485825702548027, "grad_norm": 0.06247447058558464, "global_step": 34643, "epoch": 326, "lr": 2.6789684468062477e-05} +{"train_loss": 0.015495741739869118, "grad_norm": 0.08443178236484528, "global_step": 34644, "epoch": 326, "lr": 2.6787059420386485e-05} +{"train_loss": 0.013857554644346237, "grad_norm": 0.05308377742767334, "global_step": 34645, "epoch": 326, "lr": 2.6784434454270685e-05} +{"train_loss": 0.01482619158923626, "grad_norm": 0.08354423940181732, "global_step": 34646, "epoch": 326, "lr": 2.678180956972429e-05} +{"train_loss": 0.013300840742886066, "grad_norm": 0.07195762544870377, "global_step": 34647, "epoch": 326, "lr": 2.67791847667565e-05} +{"train_loss": 0.016206637024879456, "grad_norm": 0.16002577543258667, "global_step": 34648, "epoch": 326, "lr": 2.6776560045376552e-05} +{"train_loss": 0.013734566047787666, "grad_norm": 0.07160457968711853, "global_step": 34649, "epoch": 326, "lr": 2.6773935405593697e-05} +{"train_loss": 0.017430204898118973, "grad_norm": 0.06846269965171814, "global_step": 34650, "epoch": 326, "lr": 2.6771310847417114e-05} +{"train_loss": 0.013130204752087593, "grad_norm": 0.0993007943034172, "global_step": 34651, "epoch": 326, "lr": 2.6768686370856054e-05} +{"train_loss": 0.017300307750701904, "grad_norm": 0.06699804961681366, "global_step": 34652, "epoch": 326, "lr": 2.676606197591972e-05} +{"train_loss": 0.014528442174196243, "grad_norm": 0.09301982820034027, "global_step": 34653, "epoch": 326, "lr": 2.676343766261735e-05} +{"train_loss": 0.014293048530817032, "grad_norm": 0.07111528515815735, "global_step": 34654, "epoch": 326, "lr": 2.6760813430958138e-05} +{"train_loss": 0.01691560074687004, "grad_norm": 0.10440812259912491, "global_step": 34655, "epoch": 326, "lr": 2.675818928095132e-05} +{"train_loss": 0.01592233218252659, "grad_norm": 0.10067511349916458, "global_step": 34656, "epoch": 326, "lr": 2.6755565212606136e-05} +{"train_loss": 0.016179487109184265, "grad_norm": 0.12858282029628754, "global_step": 34657, "epoch": 326, "lr": 2.6752941225931793e-05} +{"train_loss": 0.016166167333722115, "grad_norm": 0.08079175651073456, "global_step": 34658, "epoch": 326, "lr": 2.675031732093748e-05} +{"train_loss": 0.017211422324180603, "grad_norm": 0.0950060784816742, "global_step": 34659, "epoch": 326, "lr": 2.674769349763244e-05} +{"train_loss": 0.013788009993731976, "grad_norm": 0.05122588574886322, "global_step": 34660, "epoch": 326, "lr": 2.674506975602592e-05} +{"train_loss": 0.01517903658051817, "grad_norm": 0.10906228423118591, "global_step": 34661, "epoch": 326, "lr": 2.674244609612708e-05, "val_loss": 0.019510777667164803} +{"train_loss": 0.014245372265577316, "grad_norm": 0.11139952391386032, "global_step": 34662, "epoch": 327, "lr": 2.6739822517945178e-05} +{"train_loss": 0.013553233817219734, "grad_norm": 0.07020650058984756, "global_step": 34663, "epoch": 327, "lr": 2.6737199021489462e-05} +{"train_loss": 0.014850275591015816, "grad_norm": 0.10618502646684647, "global_step": 34664, "epoch": 327, "lr": 2.6734575606769075e-05} +{"train_loss": 0.014258716255426407, "grad_norm": 0.1853157877922058, "global_step": 34665, "epoch": 327, "lr": 2.673195227379327e-05} +{"train_loss": 0.014986541122198105, "grad_norm": 0.07540576159954071, "global_step": 34666, "epoch": 327, "lr": 2.6729329022571257e-05} +{"train_loss": 0.014443555846810341, "grad_norm": 0.1270247995853424, "global_step": 34667, "epoch": 327, "lr": 2.6726705853112292e-05} +{"train_loss": 0.015132986009120941, "grad_norm": 0.07745731621980667, "global_step": 34668, "epoch": 327, "lr": 2.672408276542554e-05} +{"train_loss": 0.014168144203722477, "grad_norm": 0.16913457214832306, "global_step": 34669, "epoch": 327, "lr": 2.6721459759520257e-05} +{"train_loss": 0.014184129424393177, "grad_norm": 0.08475778251886368, "global_step": 34670, "epoch": 327, "lr": 2.6718836835405635e-05} +{"train_loss": 0.014165150001645088, "grad_norm": 0.08357308804988861, "global_step": 34671, "epoch": 327, "lr": 2.671621399309089e-05} +{"train_loss": 0.016380473971366882, "grad_norm": 0.07298901677131653, "global_step": 34672, "epoch": 327, "lr": 2.6713591232585232e-05} +{"train_loss": 0.015020176768302917, "grad_norm": 0.056475959718227386, "global_step": 34673, "epoch": 327, "lr": 2.671096855389791e-05} +{"train_loss": 0.01537301018834114, "grad_norm": 0.05817415192723274, "global_step": 34674, "epoch": 327, "lr": 2.6708345957038095e-05} +{"train_loss": 0.014817028306424618, "grad_norm": 0.1061990037560463, "global_step": 34675, "epoch": 327, "lr": 2.6705723442015047e-05} +{"train_loss": 0.015004156157374382, "grad_norm": 0.07914964109659195, "global_step": 34676, "epoch": 327, "lr": 2.6703101008837933e-05} +{"train_loss": 0.01382401678711176, "grad_norm": 0.08001682907342911, "global_step": 34677, "epoch": 327, "lr": 2.6700478657516016e-05} +{"train_loss": 0.015076483599841595, "grad_norm": 0.10252749174833298, "global_step": 34678, "epoch": 327, "lr": 2.669785638805845e-05} +{"train_loss": 0.012086553499102592, "grad_norm": 0.0635194182395935, "global_step": 34679, "epoch": 327, "lr": 2.6695234200474495e-05} +{"train_loss": 0.014394517056643963, "grad_norm": 0.09823836386203766, "global_step": 34680, "epoch": 327, "lr": 2.6692612094773373e-05} +{"train_loss": 0.015780121088027954, "grad_norm": 0.05570858344435692, "global_step": 34681, "epoch": 327, "lr": 2.668999007096427e-05} +{"train_loss": 0.016477089375257492, "grad_norm": 0.07492489367723465, "global_step": 34682, "epoch": 327, "lr": 2.668736812905639e-05} +{"train_loss": 0.01778934895992279, "grad_norm": 0.12307608872652054, "global_step": 34683, "epoch": 327, "lr": 2.6684746269058957e-05} +{"train_loss": 0.013824256137013435, "grad_norm": 0.13236384093761444, "global_step": 34684, "epoch": 327, "lr": 2.668212449098121e-05} +{"train_loss": 0.012388168834149837, "grad_norm": 0.07398799806833267, "global_step": 34685, "epoch": 327, "lr": 2.6679502794832312e-05} +{"train_loss": 0.012825822457671165, "grad_norm": 0.05633725970983505, "global_step": 34686, "epoch": 327, "lr": 2.6676881180621492e-05} +{"train_loss": 0.014194665476679802, "grad_norm": 0.0709223672747612, "global_step": 34687, "epoch": 327, "lr": 2.6674259648358023e-05} +{"train_loss": 0.01384247187525034, "grad_norm": 0.06495234370231628, "global_step": 34688, "epoch": 327, "lr": 2.667163819805102e-05} +{"train_loss": 0.016202345490455627, "grad_norm": 0.11275538802146912, "global_step": 34689, "epoch": 327, "lr": 2.6669016829709726e-05} +{"train_loss": 0.013164695352315903, "grad_norm": 0.17947231233119965, "global_step": 34690, "epoch": 327, "lr": 2.666639554334338e-05} +{"train_loss": 0.014712745323777199, "grad_norm": 0.13054519891738892, "global_step": 34691, "epoch": 327, "lr": 2.666377433896116e-05} +{"train_loss": 0.017168868333101273, "grad_norm": 0.06388545781373978, "global_step": 34692, "epoch": 327, "lr": 2.666115321657231e-05} +{"train_loss": 0.015757272019982338, "grad_norm": 0.06898020952939987, "global_step": 34693, "epoch": 327, "lr": 2.665853217618599e-05} +{"train_loss": 0.014158964157104492, "grad_norm": 0.09134907275438309, "global_step": 34694, "epoch": 327, "lr": 2.6655911217811463e-05} +{"train_loss": 0.014362534508109093, "grad_norm": 0.05629647150635719, "global_step": 34695, "epoch": 327, "lr": 2.6653290341457894e-05} +{"train_loss": 0.017837082967162132, "grad_norm": 0.086973637342453, "global_step": 34696, "epoch": 327, "lr": 2.66506695471345e-05} +{"train_loss": 0.01478310115635395, "grad_norm": 0.07704457640647888, "global_step": 34697, "epoch": 327, "lr": 2.6648048834850524e-05} +{"train_loss": 0.015401802957057953, "grad_norm": 0.14159823954105377, "global_step": 34698, "epoch": 327, "lr": 2.664542820461515e-05} +{"train_loss": 0.016207365319132805, "grad_norm": 0.07941363751888275, "global_step": 34699, "epoch": 327, "lr": 2.6642807656437562e-05} +{"train_loss": 0.015653438866138458, "grad_norm": 0.06256267428398132, "global_step": 34700, "epoch": 327, "lr": 2.6640187190326993e-05} +{"train_loss": 0.01552136242389679, "grad_norm": 0.0684133842587471, "global_step": 34701, "epoch": 327, "lr": 2.663756680629267e-05} +{"train_loss": 0.01629484072327614, "grad_norm": 0.058767739683389664, "global_step": 34702, "epoch": 327, "lr": 2.663494650434375e-05} +{"train_loss": 0.016705555841326714, "grad_norm": 0.07305694371461868, "global_step": 34703, "epoch": 327, "lr": 2.6632326284489473e-05} +{"train_loss": 0.015881003811955452, "grad_norm": 0.14398492872714996, "global_step": 34704, "epoch": 327, "lr": 2.6629706146739076e-05} +{"train_loss": 0.016392739489674568, "grad_norm": 0.1028173640370369, "global_step": 34705, "epoch": 327, "lr": 2.6627086091101683e-05} +{"train_loss": 0.014143296517431736, "grad_norm": 0.06740378588438034, "global_step": 34706, "epoch": 327, "lr": 2.6624466117586556e-05} +{"train_loss": 0.014774621464312077, "grad_norm": 0.056028060615062714, "global_step": 34707, "epoch": 327, "lr": 2.6621846226202902e-05} +{"train_loss": 0.01599244773387909, "grad_norm": 0.06078922748565674, "global_step": 34708, "epoch": 327, "lr": 2.66192264169599e-05} +{"train_loss": 0.015183630399405956, "grad_norm": 0.06072400137782097, "global_step": 34709, "epoch": 327, "lr": 2.6616606689866758e-05} +{"train_loss": 0.014556603506207466, "grad_norm": 0.1214391440153122, "global_step": 34710, "epoch": 327, "lr": 2.661398704493272e-05} +{"train_loss": 0.015287638641893864, "grad_norm": 0.08997039496898651, "global_step": 34711, "epoch": 327, "lr": 2.6611367482166953e-05} +{"train_loss": 0.01383628137409687, "grad_norm": 0.10967028886079788, "global_step": 34712, "epoch": 327, "lr": 2.6608748001578655e-05} +{"train_loss": 0.014341181144118309, "grad_norm": 0.09772668033838272, "global_step": 34713, "epoch": 327, "lr": 2.660612860317704e-05} +{"train_loss": 0.014369687996804714, "grad_norm": 0.07069061696529388, "global_step": 34714, "epoch": 327, "lr": 2.660350928697134e-05} +{"train_loss": 0.017298974096775055, "grad_norm": 0.12738002836704254, "global_step": 34715, "epoch": 327, "lr": 2.660089005297071e-05} +{"train_loss": 0.015162663534283638, "grad_norm": 0.15549880266189575, "global_step": 34716, "epoch": 327, "lr": 2.6598270901184398e-05} +{"train_loss": 0.014980917796492577, "grad_norm": 0.057976849377155304, "global_step": 34717, "epoch": 327, "lr": 2.6595651831621565e-05} +{"train_loss": 0.015284180641174316, "grad_norm": 0.060393232852220535, "global_step": 34718, "epoch": 327, "lr": 2.6593032844291456e-05} +{"train_loss": 0.015043716877698898, "grad_norm": 0.06640272587537766, "global_step": 34719, "epoch": 327, "lr": 2.6590413939203228e-05} +{"train_loss": 0.015051143243908882, "grad_norm": 0.09581837058067322, "global_step": 34720, "epoch": 327, "lr": 2.6587795116366097e-05} +{"train_loss": 0.0162484310567379, "grad_norm": 0.0633484274148941, "global_step": 34721, "epoch": 327, "lr": 2.65851763757893e-05} +{"train_loss": 0.01499677263200283, "grad_norm": 0.06399089097976685, "global_step": 34722, "epoch": 327, "lr": 2.6582557717482013e-05} +{"train_loss": 0.013718889094889164, "grad_norm": 0.12503108382225037, "global_step": 34723, "epoch": 327, "lr": 2.6579939141453414e-05} +{"train_loss": 0.016302794218063354, "grad_norm": 0.09116881340742111, "global_step": 34724, "epoch": 327, "lr": 2.6577320647712717e-05} +{"train_loss": 0.014941566623747349, "grad_norm": 0.12731443345546722, "global_step": 34725, "epoch": 327, "lr": 2.6574702236269155e-05} +{"train_loss": 0.017522726207971573, "grad_norm": 0.0814286470413208, "global_step": 34726, "epoch": 327, "lr": 2.657208390713188e-05} +{"train_loss": 0.01346617378294468, "grad_norm": 0.07018190622329712, "global_step": 34727, "epoch": 327, "lr": 2.6569465660310112e-05} +{"train_loss": 0.015133182518184185, "grad_norm": 0.06106986477971077, "global_step": 34728, "epoch": 327, "lr": 2.656684749581309e-05} +{"train_loss": 0.01349314209073782, "grad_norm": 0.049261268228292465, "global_step": 34729, "epoch": 327, "lr": 2.6564229413649944e-05} +{"train_loss": 0.016158726066350937, "grad_norm": 0.12882965803146362, "global_step": 34730, "epoch": 327, "lr": 2.656161141382989e-05} +{"train_loss": 0.014729531481862068, "grad_norm": 0.11105311661958694, "global_step": 34731, "epoch": 327, "lr": 2.6558993496362173e-05} +{"train_loss": 0.017648667097091675, "grad_norm": 0.07106862217187881, "global_step": 34732, "epoch": 327, "lr": 2.6556375661255932e-05} +{"train_loss": 0.013763152062892914, "grad_norm": 0.07425995171070099, "global_step": 34733, "epoch": 327, "lr": 2.6553757908520392e-05} +{"train_loss": 0.016212811693549156, "grad_norm": 0.1266046166419983, "global_step": 34734, "epoch": 327, "lr": 2.6551140238164774e-05} +{"train_loss": 0.014289293438196182, "grad_norm": 0.057150181382894516, "global_step": 34735, "epoch": 327, "lr": 2.6548522650198247e-05} +{"train_loss": 0.0140114426612854, "grad_norm": 0.07385212182998657, "global_step": 34736, "epoch": 327, "lr": 2.654590514463e-05} +{"train_loss": 0.01775321736931801, "grad_norm": 0.050783880054950714, "global_step": 34737, "epoch": 327, "lr": 2.6543287721469235e-05} +{"train_loss": 0.014453748241066933, "grad_norm": 0.11378937214612961, "global_step": 34738, "epoch": 327, "lr": 2.654067038072518e-05} +{"train_loss": 0.015330572612583637, "grad_norm": 0.0904349610209465, "global_step": 34739, "epoch": 327, "lr": 2.6538053122406985e-05} +{"train_loss": 0.013918913900852203, "grad_norm": 0.09447257220745087, "global_step": 34740, "epoch": 327, "lr": 2.653543594652389e-05} +{"train_loss": 0.015700148418545723, "grad_norm": 0.0781535878777504, "global_step": 34741, "epoch": 327, "lr": 2.6532818853085045e-05} +{"train_loss": 0.015048941597342491, "grad_norm": 0.08008640259504318, "global_step": 34742, "epoch": 327, "lr": 2.653020184209969e-05} +{"train_loss": 0.01501736044883728, "grad_norm": 0.10902084410190582, "global_step": 34743, "epoch": 327, "lr": 2.6527584913576987e-05} +{"train_loss": 0.016705136746168137, "grad_norm": 0.06498463451862335, "global_step": 34744, "epoch": 327, "lr": 2.6524968067526134e-05} +{"train_loss": 0.0154939666390419, "grad_norm": 0.12775814533233643, "global_step": 34745, "epoch": 327, "lr": 2.6522351303956372e-05} +{"train_loss": 0.01453922688961029, "grad_norm": 0.07462053000926971, "global_step": 34746, "epoch": 327, "lr": 2.6519734622876824e-05} +{"train_loss": 0.017245259135961533, "grad_norm": 0.07291746139526367, "global_step": 34747, "epoch": 327, "lr": 2.6517118024296716e-05} +{"train_loss": 0.01678038202226162, "grad_norm": 0.2212928980588913, "global_step": 34748, "epoch": 327, "lr": 2.6514501508225255e-05} +{"train_loss": 0.016165807843208313, "grad_norm": 0.0557042621076107, "global_step": 34749, "epoch": 327, "lr": 2.651188507467161e-05} +{"train_loss": 0.017542283982038498, "grad_norm": 0.0693347230553627, "global_step": 34750, "epoch": 327, "lr": 2.6509268723644987e-05} +{"train_loss": 0.01510581560432911, "grad_norm": 0.06908658146858215, "global_step": 34751, "epoch": 327, "lr": 2.650665245515459e-05} +{"train_loss": 0.01270194910466671, "grad_norm": 0.0828445702791214, "global_step": 34752, "epoch": 327, "lr": 2.650403626920961e-05} +{"train_loss": 0.01534862071275711, "grad_norm": 0.07903293520212173, "global_step": 34753, "epoch": 327, "lr": 2.6501420165819203e-05} +{"train_loss": 0.015349948778748512, "grad_norm": 0.08874500542879105, "global_step": 34754, "epoch": 327, "lr": 2.6498804144992585e-05} +{"train_loss": 0.012296457774937153, "grad_norm": 0.055287137627601624, "global_step": 34755, "epoch": 327, "lr": 2.649618820673897e-05} +{"train_loss": 0.01664774864912033, "grad_norm": 0.08772289007902145, "global_step": 34756, "epoch": 327, "lr": 2.649357235106751e-05} +{"train_loss": 0.01866341009736061, "grad_norm": 0.11475366353988647, "global_step": 34757, "epoch": 327, "lr": 2.6490956577987437e-05} +{"train_loss": 0.015583118423819542, "grad_norm": 0.08114921301603317, "global_step": 34758, "epoch": 327, "lr": 2.648834088750789e-05} +{"train_loss": 0.015395901165902615, "grad_norm": 0.07977302372455597, "global_step": 34759, "epoch": 327, "lr": 2.6485725279638113e-05} +{"train_loss": 0.013700989075005054, "grad_norm": 0.08728807419538498, "global_step": 34760, "epoch": 327, "lr": 2.6483109754387247e-05} +{"train_loss": 0.014921236783266068, "grad_norm": 0.09714026749134064, "global_step": 34761, "epoch": 327, "lr": 2.6480494311764515e-05} +{"train_loss": 0.014214802533388138, "grad_norm": 0.10729474574327469, "global_step": 34762, "epoch": 327, "lr": 2.647787895177911e-05} +{"train_loss": 0.016964970156550407, "grad_norm": 0.0556381456553936, "global_step": 34763, "epoch": 327, "lr": 2.647526367444021e-05} +{"train_loss": 0.015407530590891838, "grad_norm": 0.10604065656661987, "global_step": 34764, "epoch": 327, "lr": 2.6472648479756984e-05} +{"train_loss": 0.017822327092289925, "grad_norm": 0.0796356201171875, "global_step": 34765, "epoch": 327, "lr": 2.6470033367738662e-05} +{"train_loss": 0.01633262261748314, "grad_norm": 0.08862755447626114, "global_step": 34766, "epoch": 327, "lr": 2.6467418338394388e-05} +{"train_loss": 0.015173270753672663, "grad_norm": 0.08162163197994232, "global_step": 34767, "epoch": 327, "lr": 2.6464803391733374e-05, "val_loss": 0.019886035472154617} +{"train_loss": 0.015821464359760284, "grad_norm": 0.07627265155315399, "global_step": 34768, "epoch": 328, "lr": 2.64621885277648e-05} +{"train_loss": 0.01643085479736328, "grad_norm": 0.07084633409976959, "global_step": 34769, "epoch": 328, "lr": 2.64595737464979e-05} +{"train_loss": 0.01541367918252945, "grad_norm": 0.07200822979211807, "global_step": 34770, "epoch": 328, "lr": 2.6456959047941786e-05} +{"train_loss": 0.016026757657527924, "grad_norm": 0.06736855953931808, "global_step": 34771, "epoch": 328, "lr": 2.645434443210567e-05} +{"train_loss": 0.016995485872030258, "grad_norm": 0.08135418593883514, "global_step": 34772, "epoch": 328, "lr": 2.645172989899878e-05} +{"train_loss": 0.014480095356702805, "grad_norm": 0.1097385585308075, "global_step": 34773, "epoch": 328, "lr": 2.644911544863024e-05} +{"train_loss": 0.015681801363825798, "grad_norm": 0.1285945028066635, "global_step": 34774, "epoch": 328, "lr": 2.6446501081009273e-05} +{"train_loss": 0.013402990996837616, "grad_norm": 0.13351894915103912, "global_step": 34775, "epoch": 328, "lr": 2.6443886796145072e-05} +{"train_loss": 0.015184253454208374, "grad_norm": 0.10454200953245163, "global_step": 34776, "epoch": 328, "lr": 2.644127259404682e-05} +{"train_loss": 0.013830756768584251, "grad_norm": 0.06630079448223114, "global_step": 34777, "epoch": 328, "lr": 2.6438658474723654e-05} +{"train_loss": 0.015798324719071388, "grad_norm": 0.11447972804307938, "global_step": 34778, "epoch": 328, "lr": 2.6436044438184803e-05} +{"train_loss": 0.015980448573827744, "grad_norm": 0.08534553647041321, "global_step": 34779, "epoch": 328, "lr": 2.643343048443947e-05} +{"train_loss": 0.014459474012255669, "grad_norm": 0.1027996838092804, "global_step": 34780, "epoch": 328, "lr": 2.643081661349679e-05} +{"train_loss": 0.014667011797428131, "grad_norm": 0.14925819635391235, "global_step": 34781, "epoch": 328, "lr": 2.6428202825365988e-05} +{"train_loss": 0.01506496500223875, "grad_norm": 0.06038016080856323, "global_step": 34782, "epoch": 328, "lr": 2.642558912005621e-05} +{"train_loss": 0.014507908374071121, "grad_norm": 0.056946177035570145, "global_step": 34783, "epoch": 328, "lr": 2.6422975497576687e-05} +{"train_loss": 0.014911304228007793, "grad_norm": 0.07528028637170792, "global_step": 34784, "epoch": 328, "lr": 2.6420361957936545e-05} +{"train_loss": 0.012938662432134151, "grad_norm": 0.07491522282361984, "global_step": 34785, "epoch": 328, "lr": 2.6417748501145e-05} +{"train_loss": 0.015508879907429218, "grad_norm": 0.11477789282798767, "global_step": 34786, "epoch": 328, "lr": 2.6415135127211253e-05} +{"train_loss": 0.016071993857622147, "grad_norm": 0.17531898617744446, "global_step": 34787, "epoch": 328, "lr": 2.641252183614446e-05} +{"train_loss": 0.014454588294029236, "grad_norm": 0.09887045621871948, "global_step": 34788, "epoch": 328, "lr": 2.6409908627953796e-05} +{"train_loss": 0.013963429257273674, "grad_norm": 0.06448838859796524, "global_step": 34789, "epoch": 328, "lr": 2.6407295502648467e-05} +{"train_loss": 0.015520700253546238, "grad_norm": 0.04963543638586998, "global_step": 34790, "epoch": 328, "lr": 2.6404682460237617e-05} +{"train_loss": 0.014753887429833412, "grad_norm": 0.0646095797419548, "global_step": 34791, "epoch": 328, "lr": 2.640206950073046e-05} +{"train_loss": 0.014493104070425034, "grad_norm": 0.07425627112388611, "global_step": 34792, "epoch": 328, "lr": 2.6399456624136186e-05} +{"train_loss": 0.014963381923735142, "grad_norm": 0.05555059388279915, "global_step": 34793, "epoch": 328, "lr": 2.6396843830463958e-05} +{"train_loss": 0.01319880597293377, "grad_norm": 0.07183343172073364, "global_step": 34794, "epoch": 328, "lr": 2.6394231119722934e-05} +{"train_loss": 0.01567848026752472, "grad_norm": 0.08472861349582672, "global_step": 34795, "epoch": 328, "lr": 2.6391618491922315e-05} +{"train_loss": 0.016379181295633316, "grad_norm": 0.09109493345022202, "global_step": 34796, "epoch": 328, "lr": 2.6389005947071304e-05} +{"train_loss": 0.015135733410716057, "grad_norm": 0.15676386654376984, "global_step": 34797, "epoch": 328, "lr": 2.638639348517904e-05} +{"train_loss": 0.014923772774636745, "grad_norm": 0.14467781782150269, "global_step": 34798, "epoch": 328, "lr": 2.6383781106254735e-05} +{"train_loss": 0.014989820308983326, "grad_norm": 0.08777977526187897, "global_step": 34799, "epoch": 328, "lr": 2.6381168810307533e-05} +{"train_loss": 0.013900108635425568, "grad_norm": 0.08280140161514282, "global_step": 34800, "epoch": 328, "lr": 2.637855659734666e-05} +{"train_loss": 0.014076579362154007, "grad_norm": 0.08659213036298752, "global_step": 34801, "epoch": 328, "lr": 2.637594446738124e-05} +{"train_loss": 0.016261011362075806, "grad_norm": 0.09124212712049484, "global_step": 34802, "epoch": 328, "lr": 2.637333242042048e-05} +{"train_loss": 0.015851300209760666, "grad_norm": 0.06773319840431213, "global_step": 34803, "epoch": 328, "lr": 2.6370720456473573e-05} +{"train_loss": 0.015009751543402672, "grad_norm": 0.0658750981092453, "global_step": 34804, "epoch": 328, "lr": 2.6368108575549676e-05} +{"train_loss": 0.016650620847940445, "grad_norm": 0.07576509565114975, "global_step": 34805, "epoch": 328, "lr": 2.636549677765795e-05} +{"train_loss": 0.012554654851555824, "grad_norm": 0.0637413039803505, "global_step": 34806, "epoch": 328, "lr": 2.6362885062807612e-05} +{"train_loss": 0.015264606103301048, "grad_norm": 0.10596418380737305, "global_step": 34807, "epoch": 328, "lr": 2.6360273431007797e-05} +{"train_loss": 0.014246263541281223, "grad_norm": 0.07434583455324173, "global_step": 34808, "epoch": 328, "lr": 2.63576618822677e-05} +{"train_loss": 0.01634405180811882, "grad_norm": 0.07721755653619766, "global_step": 34809, "epoch": 328, "lr": 2.6355050416596517e-05} +{"train_loss": 0.01631258614361286, "grad_norm": 0.07189282029867172, "global_step": 34810, "epoch": 328, "lr": 2.63524390340034e-05} +{"train_loss": 0.01686534658074379, "grad_norm": 0.11817479133605957, "global_step": 34811, "epoch": 328, "lr": 2.6349827734497512e-05} +{"train_loss": 0.015158675611019135, "grad_norm": 0.07895170152187347, "global_step": 34812, "epoch": 328, "lr": 2.6347216518088046e-05} +{"train_loss": 0.015035610646009445, "grad_norm": 0.05890466645359993, "global_step": 34813, "epoch": 328, "lr": 2.6344605384784187e-05} +{"train_loss": 0.014458349905908108, "grad_norm": 0.0870930552482605, "global_step": 34814, "epoch": 328, "lr": 2.6341994334595087e-05} +{"train_loss": 0.016777006909251213, "grad_norm": 0.07513627409934998, "global_step": 34815, "epoch": 328, "lr": 2.6339383367529924e-05} +{"train_loss": 0.014189439825713634, "grad_norm": 0.0734284296631813, "global_step": 34816, "epoch": 328, "lr": 2.6336772483597898e-05} +{"train_loss": 0.016471505165100098, "grad_norm": 0.10672464966773987, "global_step": 34817, "epoch": 328, "lr": 2.6334161682808166e-05} +{"train_loss": 0.016368117183446884, "grad_norm": 0.11601173877716064, "global_step": 34818, "epoch": 328, "lr": 2.6331550965169876e-05} +{"train_loss": 0.014282836578786373, "grad_norm": 0.05046040564775467, "global_step": 34819, "epoch": 328, "lr": 2.632894033069222e-05} +{"train_loss": 0.016652796417474747, "grad_norm": 0.0677538588643074, "global_step": 34820, "epoch": 328, "lr": 2.6326329779384395e-05} +{"train_loss": 0.01352619007229805, "grad_norm": 0.08363558351993561, "global_step": 34821, "epoch": 328, "lr": 2.6323719311255533e-05} +{"train_loss": 0.015083683654665947, "grad_norm": 0.07648282498121262, "global_step": 34822, "epoch": 328, "lr": 2.632110892631485e-05} +{"train_loss": 0.017255660146474838, "grad_norm": 0.06673069298267365, "global_step": 34823, "epoch": 328, "lr": 2.6318498624571463e-05} +{"train_loss": 0.01433519460260868, "grad_norm": 0.10005093365907669, "global_step": 34824, "epoch": 328, "lr": 2.6315888406034605e-05} +{"train_loss": 0.014556060545146465, "grad_norm": 0.07017384469509125, "global_step": 34825, "epoch": 328, "lr": 2.631327827071338e-05} +{"train_loss": 0.014750669710338116, "grad_norm": 0.06715983152389526, "global_step": 34826, "epoch": 328, "lr": 2.6310668218617006e-05} +{"train_loss": 0.018375298008322716, "grad_norm": 0.1279134303331375, "global_step": 34827, "epoch": 328, "lr": 2.630805824975466e-05} +{"train_loss": 0.013110135681927204, "grad_norm": 0.0754280760884285, "global_step": 34828, "epoch": 328, "lr": 2.6305448364135488e-05} +{"train_loss": 0.015166863799095154, "grad_norm": 0.06781844794750214, "global_step": 34829, "epoch": 328, "lr": 2.630283856176865e-05} +{"train_loss": 0.013913292437791824, "grad_norm": 0.0849643275141716, "global_step": 34830, "epoch": 328, "lr": 2.6300228842663354e-05} +{"train_loss": 0.014656954444944859, "grad_norm": 0.052163492888212204, "global_step": 34831, "epoch": 328, "lr": 2.629761920682872e-05} +{"train_loss": 0.015046761371195316, "grad_norm": 0.07366997748613358, "global_step": 34832, "epoch": 328, "lr": 2.629500965427395e-05} +{"train_loss": 0.013648085296154022, "grad_norm": 0.05726170912384987, "global_step": 34833, "epoch": 328, "lr": 2.6292400185008227e-05} +{"train_loss": 0.016686519607901573, "grad_norm": 0.08103828877210617, "global_step": 34834, "epoch": 328, "lr": 2.6289790799040697e-05} +{"train_loss": 0.015658751130104065, "grad_norm": 0.11376102268695831, "global_step": 34835, "epoch": 328, "lr": 2.6287181496380507e-05} +{"train_loss": 0.016362231224775314, "grad_norm": 0.09046193957328796, "global_step": 34836, "epoch": 328, "lr": 2.6284572277036856e-05} +{"train_loss": 0.017323028296232224, "grad_norm": 0.06964079290628433, "global_step": 34837, "epoch": 328, "lr": 2.628196314101893e-05} +{"train_loss": 0.014749452471733093, "grad_norm": 0.08975189924240112, "global_step": 34838, "epoch": 328, "lr": 2.6279354088335838e-05} +{"train_loss": 0.0160396546125412, "grad_norm": 0.09965203702449799, "global_step": 34839, "epoch": 328, "lr": 2.627674511899678e-05} +{"train_loss": 0.014320239424705505, "grad_norm": 0.06352732330560684, "global_step": 34840, "epoch": 328, "lr": 2.627413623301095e-05} +{"train_loss": 0.015034109354019165, "grad_norm": 0.11552075296640396, "global_step": 34841, "epoch": 328, "lr": 2.6271527430387488e-05} +{"train_loss": 0.01538604125380516, "grad_norm": 0.06938125938177109, "global_step": 34842, "epoch": 328, "lr": 2.6268918711135538e-05} +{"train_loss": 0.01629427820444107, "grad_norm": 0.0665993019938469, "global_step": 34843, "epoch": 328, "lr": 2.626631007526428e-05} +{"train_loss": 0.015150055289268494, "grad_norm": 0.07376322150230408, "global_step": 34844, "epoch": 328, "lr": 2.6263701522782914e-05} +{"train_loss": 0.013533862307667732, "grad_norm": 0.10237675160169601, "global_step": 34845, "epoch": 328, "lr": 2.626109305370058e-05} +{"train_loss": 0.014057669788599014, "grad_norm": 0.09018851071596146, "global_step": 34846, "epoch": 328, "lr": 2.6258484668026417e-05} +{"train_loss": 0.013612795621156693, "grad_norm": 0.1594095081090927, "global_step": 34847, "epoch": 328, "lr": 2.625587636576963e-05} +{"train_loss": 0.017369845882058144, "grad_norm": 0.07752103358507156, "global_step": 34848, "epoch": 328, "lr": 2.625326814693936e-05} +{"train_loss": 0.014093281701207161, "grad_norm": 0.09398715943098068, "global_step": 34849, "epoch": 328, "lr": 2.625066001154477e-05} +{"train_loss": 0.01722143217921257, "grad_norm": 0.07875965535640717, "global_step": 34850, "epoch": 328, "lr": 2.624805195959505e-05} +{"train_loss": 0.01629297062754631, "grad_norm": 0.05405133217573166, "global_step": 34851, "epoch": 328, "lr": 2.624544399109935e-05} +{"train_loss": 0.015708526596426964, "grad_norm": 0.07033922523260117, "global_step": 34852, "epoch": 328, "lr": 2.6242836106066805e-05} +{"train_loss": 0.014585889875888824, "grad_norm": 0.06943653523921967, "global_step": 34853, "epoch": 328, "lr": 2.624022830450661e-05} +{"train_loss": 0.016263343393802643, "grad_norm": 0.0776669830083847, "global_step": 34854, "epoch": 328, "lr": 2.623762058642793e-05} +{"train_loss": 0.016504991799592972, "grad_norm": 0.09241282194852829, "global_step": 34855, "epoch": 328, "lr": 2.6235012951839903e-05} +{"train_loss": 0.0150326332077384, "grad_norm": 0.0989934578537941, "global_step": 34856, "epoch": 328, "lr": 2.6232405400751704e-05} +{"train_loss": 0.014705008827149868, "grad_norm": 0.12382283806800842, "global_step": 34857, "epoch": 328, "lr": 2.6229797933172518e-05} +{"train_loss": 0.01532807294279337, "grad_norm": 0.08838004618883133, "global_step": 34858, "epoch": 328, "lr": 2.6227190549111484e-05} +{"train_loss": 0.017011558637022972, "grad_norm": 0.07939722388982773, "global_step": 34859, "epoch": 328, "lr": 2.6224583248577746e-05} +{"train_loss": 0.015809370204806328, "grad_norm": 0.1165337935090065, "global_step": 34860, "epoch": 328, "lr": 2.6221976031580477e-05} +{"train_loss": 0.014776049181818962, "grad_norm": 0.05787750706076622, "global_step": 34861, "epoch": 328, "lr": 2.6219368898128872e-05} +{"train_loss": 0.01350405067205429, "grad_norm": 0.20160973072052002, "global_step": 34862, "epoch": 328, "lr": 2.6216761848232037e-05} +{"train_loss": 0.01621972769498825, "grad_norm": 0.05947059765458107, "global_step": 34863, "epoch": 328, "lr": 2.621415488189918e-05} +{"train_loss": 0.01210310123860836, "grad_norm": 0.0803825706243515, "global_step": 34864, "epoch": 328, "lr": 2.621154799913943e-05} +{"train_loss": 0.015911221504211426, "grad_norm": 0.07109101116657257, "global_step": 34865, "epoch": 328, "lr": 2.6208941199961946e-05} +{"train_loss": 0.014483623206615448, "grad_norm": 0.056435614824295044, "global_step": 34866, "epoch": 328, "lr": 2.6206334484375892e-05} +{"train_loss": 0.014053951017558575, "grad_norm": 0.06614623963832855, "global_step": 34867, "epoch": 328, "lr": 2.6203727852390446e-05} +{"train_loss": 0.014411772601306438, "grad_norm": 0.12050872296094894, "global_step": 34868, "epoch": 328, "lr": 2.620112130401474e-05} +{"train_loss": 0.014272595755755901, "grad_norm": 0.06404637545347214, "global_step": 34869, "epoch": 328, "lr": 2.619851483925796e-05} +{"train_loss": 0.013895941898226738, "grad_norm": 0.0459907129406929, "global_step": 34870, "epoch": 328, "lr": 2.619590845812923e-05} +{"train_loss": 0.015416843816637993, "grad_norm": 0.07532039284706116, "global_step": 34871, "epoch": 328, "lr": 2.6193302160637746e-05} +{"train_loss": 0.015352719463407993, "grad_norm": 0.09371642768383026, "global_step": 34872, "epoch": 328, "lr": 2.619069594679262e-05} +{"train_loss": 0.015159300071591476, "grad_norm": 0.09998602420091629, "global_step": 34873, "epoch": 328, "lr": 2.618808981660304e-05, "val_loss": 0.019474130123853683} +{"train_loss": 0.015144230797886848, "grad_norm": 0.09967939555644989, "global_step": 34874, "epoch": 329, "lr": 2.6185483770078173e-05} +{"train_loss": 0.016118574887514114, "grad_norm": 0.08902834355831146, "global_step": 34875, "epoch": 329, "lr": 2.618287780722716e-05} +{"train_loss": 0.01513420045375824, "grad_norm": 0.10534588247537613, "global_step": 34876, "epoch": 329, "lr": 2.618027192805914e-05} +{"train_loss": 0.01445558201521635, "grad_norm": 0.07743887603282928, "global_step": 34877, "epoch": 329, "lr": 2.6177666132583278e-05} +{"train_loss": 0.017610885202884674, "grad_norm": 0.07575958222150803, "global_step": 34878, "epoch": 329, "lr": 2.617506042080876e-05} +{"train_loss": 0.014982922933995724, "grad_norm": 0.05692901462316513, "global_step": 34879, "epoch": 329, "lr": 2.6172454792744706e-05} +{"train_loss": 0.02029753476381302, "grad_norm": 0.07747765630483627, "global_step": 34880, "epoch": 329, "lr": 2.616984924840028e-05} +{"train_loss": 0.014361962676048279, "grad_norm": 0.13358759880065918, "global_step": 34881, "epoch": 329, "lr": 2.6167243787784656e-05} +{"train_loss": 0.017679482698440552, "grad_norm": 0.05918737128376961, "global_step": 34882, "epoch": 329, "lr": 2.6164638410906973e-05} +{"train_loss": 0.014876142144203186, "grad_norm": 0.055484168231487274, "global_step": 34883, "epoch": 329, "lr": 2.6162033117776362e-05} +{"train_loss": 0.013101919554173946, "grad_norm": 0.07112209498882294, "global_step": 34884, "epoch": 329, "lr": 2.6159427908402012e-05} +{"train_loss": 0.014950823038816452, "grad_norm": 0.09050925821065903, "global_step": 34885, "epoch": 329, "lr": 2.615682278279307e-05} +{"train_loss": 0.015439055860042572, "grad_norm": 0.06222926080226898, "global_step": 34886, "epoch": 329, "lr": 2.6154217740958677e-05} +{"train_loss": 0.013174780644476414, "grad_norm": 0.1427595019340515, "global_step": 34887, "epoch": 329, "lr": 2.6151612782908e-05} +{"train_loss": 0.015860574319958687, "grad_norm": 0.07846523821353912, "global_step": 34888, "epoch": 329, "lr": 2.6149007908650192e-05} +{"train_loss": 0.01769436150789261, "grad_norm": 0.10433471202850342, "global_step": 34889, "epoch": 329, "lr": 2.6146403118194373e-05} +{"train_loss": 0.01791197806596756, "grad_norm": 0.15055276453495026, "global_step": 34890, "epoch": 329, "lr": 2.6143798411549724e-05} +{"train_loss": 0.01590358465909958, "grad_norm": 0.13155247271060944, "global_step": 34891, "epoch": 329, "lr": 2.6141193788725416e-05} +{"train_loss": 0.01347595825791359, "grad_norm": 0.06479694694280624, "global_step": 34892, "epoch": 329, "lr": 2.613858924973055e-05} +{"train_loss": 0.015821179375052452, "grad_norm": 0.09548301249742508, "global_step": 34893, "epoch": 329, "lr": 2.6135984794574325e-05} +{"train_loss": 0.012984318658709526, "grad_norm": 0.07762482762336731, "global_step": 34894, "epoch": 329, "lr": 2.613338042326585e-05} +{"train_loss": 0.01600473001599312, "grad_norm": 0.06533533334732056, "global_step": 34895, "epoch": 329, "lr": 2.6130776135814328e-05} +{"train_loss": 0.01387111097574234, "grad_norm": 0.10399176925420761, "global_step": 34896, "epoch": 329, "lr": 2.6128171932228845e-05} +{"train_loss": 0.016273140907287598, "grad_norm": 0.07971834391355515, "global_step": 34897, "epoch": 329, "lr": 2.6125567812518593e-05} +{"train_loss": 0.016342580318450928, "grad_norm": 0.07448242604732513, "global_step": 34898, "epoch": 329, "lr": 2.612296377669273e-05} +{"train_loss": 0.01399984396994114, "grad_norm": 0.06061488762497902, "global_step": 34899, "epoch": 329, "lr": 2.6120359824760388e-05} +{"train_loss": 0.014192109927535057, "grad_norm": 0.07234029471874237, "global_step": 34900, "epoch": 329, "lr": 2.6117755956730693e-05} +{"train_loss": 0.015793032944202423, "grad_norm": 0.0712786614894867, "global_step": 34901, "epoch": 329, "lr": 2.611515217261282e-05} +{"train_loss": 0.013772251084446907, "grad_norm": 0.08262979984283447, "global_step": 34902, "epoch": 329, "lr": 2.611254847241594e-05} +{"train_loss": 0.01657313108444214, "grad_norm": 0.09605865180492401, "global_step": 34903, "epoch": 329, "lr": 2.610994485614915e-05} +{"train_loss": 0.014549653977155685, "grad_norm": 0.08375087380409241, "global_step": 34904, "epoch": 329, "lr": 2.6107341323821643e-05} +{"train_loss": 0.013268424198031425, "grad_norm": 0.07158446311950684, "global_step": 34905, "epoch": 329, "lr": 2.6104737875442553e-05} +{"train_loss": 0.015002945438027382, "grad_norm": 0.05599100887775421, "global_step": 34906, "epoch": 329, "lr": 2.6102134511021003e-05} +{"train_loss": 0.012411341071128845, "grad_norm": 0.0915667787194252, "global_step": 34907, "epoch": 329, "lr": 2.6099531230566154e-05} +{"train_loss": 0.014603575691580772, "grad_norm": 0.08521510660648346, "global_step": 34908, "epoch": 329, "lr": 2.609692803408718e-05} +{"train_loss": 0.015396879985928535, "grad_norm": 0.11765812337398529, "global_step": 34909, "epoch": 329, "lr": 2.609432492159319e-05} +{"train_loss": 0.01673557236790657, "grad_norm": 0.06180732324719429, "global_step": 34910, "epoch": 329, "lr": 2.6091721893093363e-05} +{"train_loss": 0.015038250014185905, "grad_norm": 0.07053905725479126, "global_step": 34911, "epoch": 329, "lr": 2.6089118948596802e-05} +{"train_loss": 0.013692720793187618, "grad_norm": 0.07862876355648041, "global_step": 34912, "epoch": 329, "lr": 2.6086516088112713e-05} +{"train_loss": 0.014278995804488659, "grad_norm": 0.08581416308879852, "global_step": 34913, "epoch": 329, "lr": 2.6083913311650176e-05} +{"train_loss": 0.017001468688249588, "grad_norm": 0.10873737186193466, "global_step": 34914, "epoch": 329, "lr": 2.608131061921837e-05} +{"train_loss": 0.016503693535923958, "grad_norm": 0.13640102744102478, "global_step": 34915, "epoch": 329, "lr": 2.6078708010826453e-05} +{"train_loss": 0.013305912725627422, "grad_norm": 0.07740144431591034, "global_step": 34916, "epoch": 329, "lr": 2.6076105486483554e-05} +{"train_loss": 0.013428622856736183, "grad_norm": 0.138132706284523, "global_step": 34917, "epoch": 329, "lr": 2.60735030461988e-05} +{"train_loss": 0.015341350808739662, "grad_norm": 0.15500406920909882, "global_step": 34918, "epoch": 329, "lr": 2.6070900689981348e-05} +{"train_loss": 0.014829443767666817, "grad_norm": 0.08575008809566498, "global_step": 34919, "epoch": 329, "lr": 2.6068298417840364e-05} +{"train_loss": 0.016065407544374466, "grad_norm": 0.05183261260390282, "global_step": 34920, "epoch": 329, "lr": 2.6065696229784954e-05} +{"train_loss": 0.016718456521630287, "grad_norm": 0.1027197539806366, "global_step": 34921, "epoch": 329, "lr": 2.6063094125824272e-05} +{"train_loss": 0.014999203383922577, "grad_norm": 0.10617044568061829, "global_step": 34922, "epoch": 329, "lr": 2.6060492105967515e-05} +{"train_loss": 0.017699522897601128, "grad_norm": 0.11464152485132217, "global_step": 34923, "epoch": 329, "lr": 2.605789017022373e-05} +{"train_loss": 0.013849419541656971, "grad_norm": 0.07533153146505356, "global_step": 34924, "epoch": 329, "lr": 2.6055288318602117e-05} +{"train_loss": 0.01626792922616005, "grad_norm": 0.07910376787185669, "global_step": 34925, "epoch": 329, "lr": 2.6052686551111803e-05} +{"train_loss": 0.016049012541770935, "grad_norm": 0.06976396590471268, "global_step": 34926, "epoch": 329, "lr": 2.6050084867761954e-05} +{"train_loss": 0.01466987282037735, "grad_norm": 0.07555784285068512, "global_step": 34927, "epoch": 329, "lr": 2.6047483268561667e-05} +{"train_loss": 0.014063887298107147, "grad_norm": 0.07164645195007324, "global_step": 34928, "epoch": 329, "lr": 2.6044881753520134e-05} +{"train_loss": 0.014474518597126007, "grad_norm": 0.09932655096054077, "global_step": 34929, "epoch": 329, "lr": 2.6042280322646462e-05} +{"train_loss": 0.01536076795309782, "grad_norm": 0.11434989422559738, "global_step": 34930, "epoch": 329, "lr": 2.6039678975949787e-05} +{"train_loss": 0.014544984325766563, "grad_norm": 0.059818774461746216, "global_step": 34931, "epoch": 329, "lr": 2.6037077713439252e-05} +{"train_loss": 0.014377206563949585, "grad_norm": 0.17566139996051788, "global_step": 34932, "epoch": 329, "lr": 2.603447653512403e-05} +{"train_loss": 0.014615392312407494, "grad_norm": 0.1358993798494339, "global_step": 34933, "epoch": 329, "lr": 2.603187544101321e-05} +{"train_loss": 0.014489887282252312, "grad_norm": 0.15550675988197327, "global_step": 34934, "epoch": 329, "lr": 2.602927443111598e-05} +{"train_loss": 0.015292666852474213, "grad_norm": 0.11748546361923218, "global_step": 34935, "epoch": 329, "lr": 2.6026673505441435e-05} +{"train_loss": 0.015775630250573158, "grad_norm": 0.07604992389678955, "global_step": 34936, "epoch": 329, "lr": 2.602407266399875e-05} +{"train_loss": 0.01309497281908989, "grad_norm": 0.07564126700162888, "global_step": 34937, "epoch": 329, "lr": 2.602147190679703e-05} +{"train_loss": 0.016772549599409103, "grad_norm": 0.16980043053627014, "global_step": 34938, "epoch": 329, "lr": 2.601887123384543e-05} +{"train_loss": 0.01632516086101532, "grad_norm": 0.2598736584186554, "global_step": 34939, "epoch": 329, "lr": 2.601627064515311e-05} +{"train_loss": 0.016704687848687172, "grad_norm": 0.13959729671478271, "global_step": 34940, "epoch": 329, "lr": 2.6013670140729174e-05} +{"train_loss": 0.014968235045671463, "grad_norm": 0.07087933272123337, "global_step": 34941, "epoch": 329, "lr": 2.601106972058276e-05} +{"train_loss": 0.0155192781239748, "grad_norm": 0.12834258377552032, "global_step": 34942, "epoch": 329, "lr": 2.600846938472301e-05} +{"train_loss": 0.015366151928901672, "grad_norm": 0.10263620316982269, "global_step": 34943, "epoch": 329, "lr": 2.6005869133159088e-05} +{"train_loss": 0.015235207043588161, "grad_norm": 0.09275008738040924, "global_step": 34944, "epoch": 329, "lr": 2.6003268965900084e-05} +{"train_loss": 0.014903528615832329, "grad_norm": 0.07233748584985733, "global_step": 34945, "epoch": 329, "lr": 2.600066888295518e-05} +{"train_loss": 0.014821873977780342, "grad_norm": 0.0915505439043045, "global_step": 34946, "epoch": 329, "lr": 2.5998068884333483e-05} +{"train_loss": 0.015608429908752441, "grad_norm": 0.10735121369361877, "global_step": 34947, "epoch": 329, "lr": 2.5995468970044125e-05} +{"train_loss": 0.013332132250070572, "grad_norm": 0.09443511068820953, "global_step": 34948, "epoch": 329, "lr": 2.599286914009624e-05} +{"train_loss": 0.015505186282098293, "grad_norm": 0.12690316140651703, "global_step": 34949, "epoch": 329, "lr": 2.599026939449899e-05} +{"train_loss": 0.01497633196413517, "grad_norm": 0.06167853996157646, "global_step": 34950, "epoch": 329, "lr": 2.5987669733261478e-05} +{"train_loss": 0.013310502283275127, "grad_norm": 0.15085434913635254, "global_step": 34951, "epoch": 329, "lr": 2.598507015639287e-05} +{"train_loss": 0.015373741276562214, "grad_norm": 0.057188667356967926, "global_step": 34952, "epoch": 329, "lr": 2.5982470663902258e-05} +{"train_loss": 0.015015867538750172, "grad_norm": 0.059549957513809204, "global_step": 34953, "epoch": 329, "lr": 2.5979871255798826e-05} +{"train_loss": 0.015047929249703884, "grad_norm": 0.10008543729782104, "global_step": 34954, "epoch": 329, "lr": 2.5977271932091664e-05} +{"train_loss": 0.015757713466882706, "grad_norm": 0.10646508634090424, "global_step": 34955, "epoch": 329, "lr": 2.5974672692789915e-05} +{"train_loss": 0.014809143729507923, "grad_norm": 0.05714002624154091, "global_step": 34956, "epoch": 329, "lr": 2.597207353790273e-05} +{"train_loss": 0.017769932746887207, "grad_norm": 0.07288223505020142, "global_step": 34957, "epoch": 329, "lr": 2.5969474467439237e-05} +{"train_loss": 0.015755262225866318, "grad_norm": 0.10135837644338608, "global_step": 34958, "epoch": 329, "lr": 2.596687548140854e-05} +{"train_loss": 0.012810097075998783, "grad_norm": 0.0996926948428154, "global_step": 34959, "epoch": 329, "lr": 2.596427657981979e-05} +{"train_loss": 0.014716708101332188, "grad_norm": 0.1414356827735901, "global_step": 34960, "epoch": 329, "lr": 2.5961677762682147e-05} +{"train_loss": 0.013317231088876724, "grad_norm": 0.140142560005188, "global_step": 34961, "epoch": 329, "lr": 2.5959079030004685e-05} +{"train_loss": 0.012447543442249298, "grad_norm": 0.0647989958524704, "global_step": 34962, "epoch": 329, "lr": 2.5956480381796567e-05} +{"train_loss": 0.01601862907409668, "grad_norm": 0.05354791507124901, "global_step": 34963, "epoch": 329, "lr": 2.5953881818066962e-05} +{"train_loss": 0.0196041502058506, "grad_norm": 0.06599362939596176, "global_step": 34964, "epoch": 329, "lr": 2.5951283338824912e-05} +{"train_loss": 0.014906231313943863, "grad_norm": 0.10647754371166229, "global_step": 34965, "epoch": 329, "lr": 2.59486849440796e-05} +{"train_loss": 0.015421541407704353, "grad_norm": 0.06944722682237625, "global_step": 34966, "epoch": 329, "lr": 2.5946086633840166e-05} +{"train_loss": 0.014580458402633667, "grad_norm": 0.08420762419700623, "global_step": 34967, "epoch": 329, "lr": 2.5943488408115707e-05} +{"train_loss": 0.015092689543962479, "grad_norm": 0.06053929775953293, "global_step": 34968, "epoch": 329, "lr": 2.5940890266915362e-05} +{"train_loss": 0.01419855747371912, "grad_norm": 0.0884038507938385, "global_step": 34969, "epoch": 329, "lr": 2.5938292210248292e-05} +{"train_loss": 0.014905596151947975, "grad_norm": 0.09763640910387039, "global_step": 34970, "epoch": 329, "lr": 2.5935694238123598e-05} +{"train_loss": 0.014955028891563416, "grad_norm": 0.08411416411399841, "global_step": 34971, "epoch": 329, "lr": 2.593309635055039e-05} +{"train_loss": 0.01413523405790329, "grad_norm": 0.08070442825555801, "global_step": 34972, "epoch": 329, "lr": 2.5930498547537808e-05} +{"train_loss": 0.014253176748752594, "grad_norm": 0.10253364592790604, "global_step": 34973, "epoch": 329, "lr": 2.5927900829095008e-05} +{"train_loss": 0.014397788792848587, "grad_norm": 0.0789334699511528, "global_step": 34974, "epoch": 329, "lr": 2.5925303195231088e-05} +{"train_loss": 0.01753886044025421, "grad_norm": 0.08890190720558167, "global_step": 34975, "epoch": 329, "lr": 2.592270564595519e-05} +{"train_loss": 0.016413407400250435, "grad_norm": 0.11799191683530807, "global_step": 34976, "epoch": 329, "lr": 2.5920108181276416e-05} +{"train_loss": 0.01653248257935047, "grad_norm": 0.09241064637899399, "global_step": 34977, "epoch": 329, "lr": 2.5917510801203933e-05} +{"train_loss": 0.014934785664081573, "grad_norm": 0.059713706374168396, "global_step": 34978, "epoch": 329, "lr": 2.591491350574682e-05} +{"train_loss": 0.015166861980380315, "grad_norm": 0.09140583872795105, "global_step": 34979, "epoch": 329, "lr": 2.591231629491423e-05, "val_loss": 0.0205563735216856} +{"train_loss": 0.01655406877398491, "grad_norm": 0.1709422618150711, "global_step": 34980, "epoch": 330, "lr": 2.59097191687153e-05} +{"train_loss": 0.013087581843137741, "grad_norm": 0.21378950774669647, "global_step": 34981, "epoch": 330, "lr": 2.5907122127159146e-05} +{"train_loss": 0.016579173505306244, "grad_norm": 0.09740748256444931, "global_step": 34982, "epoch": 330, "lr": 2.5904525170254867e-05} +{"train_loss": 0.013937218114733696, "grad_norm": 0.12951438128948212, "global_step": 34983, "epoch": 330, "lr": 2.5901928298011602e-05} +{"train_loss": 0.01590079814195633, "grad_norm": 0.06913627684116364, "global_step": 34984, "epoch": 330, "lr": 2.5899331510438506e-05} +{"train_loss": 0.01819857582449913, "grad_norm": 0.0819242000579834, "global_step": 34985, "epoch": 330, "lr": 2.5896734807544653e-05} +{"train_loss": 0.010673671960830688, "grad_norm": 0.06863580644130707, "global_step": 34986, "epoch": 330, "lr": 2.58941381893392e-05} +{"train_loss": 0.01239902526140213, "grad_norm": 0.08192107826471329, "global_step": 34987, "epoch": 330, "lr": 2.58915416558313e-05} +{"train_loss": 0.013909182511270046, "grad_norm": 0.0801546648144722, "global_step": 34988, "epoch": 330, "lr": 2.5888945207029992e-05} +{"train_loss": 0.016701264306902885, "grad_norm": 0.0744735449552536, "global_step": 34989, "epoch": 330, "lr": 2.5886348842944452e-05} +{"train_loss": 0.015557906590402126, "grad_norm": 0.0916358232498169, "global_step": 34990, "epoch": 330, "lr": 2.5883752563583807e-05} +{"train_loss": 0.016537513583898544, "grad_norm": 0.07749925553798676, "global_step": 34991, "epoch": 330, "lr": 2.5881156368957155e-05} +{"train_loss": 0.013376191258430481, "grad_norm": 0.05585396662354469, "global_step": 34992, "epoch": 330, "lr": 2.5878560259073626e-05} +{"train_loss": 0.016811199486255646, "grad_norm": 0.08686313778162003, "global_step": 34993, "epoch": 330, "lr": 2.5875964233942368e-05} +{"train_loss": 0.012373443692922592, "grad_norm": 0.12820371985435486, "global_step": 34994, "epoch": 330, "lr": 2.587336829357248e-05} +{"train_loss": 0.017314612865447998, "grad_norm": 0.07569414377212524, "global_step": 34995, "epoch": 330, "lr": 2.5870772437973066e-05} +{"train_loss": 0.016141321510076523, "grad_norm": 0.08079490065574646, "global_step": 34996, "epoch": 330, "lr": 2.586817666715326e-05} +{"train_loss": 0.014126741327345371, "grad_norm": 0.0521608404815197, "global_step": 34997, "epoch": 330, "lr": 2.5865580981122207e-05} +{"train_loss": 0.013800815679132938, "grad_norm": 0.13437499105930328, "global_step": 34998, "epoch": 330, "lr": 2.586298537988901e-05} +{"train_loss": 0.014547718688845634, "grad_norm": 0.11863887310028076, "global_step": 34999, "epoch": 330, "lr": 2.5860389863462765e-05} +{"train_loss": 0.014637274667620659, "grad_norm": 0.09090901911258698, "global_step": 35000, "epoch": 330, "lr": 2.585779443185261e-05} +{"train_loss": 0.016885342076420784, "grad_norm": 0.1072206199169159, "global_step": 35001, "epoch": 330, "lr": 2.585519908506768e-05} +{"train_loss": 0.01584707945585251, "grad_norm": 0.06211837753653526, "global_step": 35002, "epoch": 330, "lr": 2.5852603823117072e-05} +{"train_loss": 0.014629151672124863, "grad_norm": 0.11538795381784439, "global_step": 35003, "epoch": 330, "lr": 2.585000864600991e-05} +{"train_loss": 0.013038702309131622, "grad_norm": 0.06574735790491104, "global_step": 35004, "epoch": 330, "lr": 2.584741355375535e-05} +{"train_loss": 0.016300242394208908, "grad_norm": 0.06423701345920563, "global_step": 35005, "epoch": 330, "lr": 2.5844818546362437e-05} +{"train_loss": 0.014366084709763527, "grad_norm": 0.09811826050281525, "global_step": 35006, "epoch": 330, "lr": 2.5842223623840332e-05} +{"train_loss": 0.01631002500653267, "grad_norm": 0.12020378559827805, "global_step": 35007, "epoch": 330, "lr": 2.5839628786198162e-05} +{"train_loss": 0.015436253510415554, "grad_norm": 0.06912145763635635, "global_step": 35008, "epoch": 330, "lr": 2.5837034033445008e-05} +{"train_loss": 0.014447826892137527, "grad_norm": 0.10605911165475845, "global_step": 35009, "epoch": 330, "lr": 2.5834439365590013e-05} +{"train_loss": 0.017285270616412163, "grad_norm": 0.16767431795597076, "global_step": 35010, "epoch": 330, "lr": 2.58318447826423e-05} +{"train_loss": 0.01649598591029644, "grad_norm": 0.11889395117759705, "global_step": 35011, "epoch": 330, "lr": 2.5829250284610985e-05} +{"train_loss": 0.018242809921503067, "grad_norm": 0.09655369818210602, "global_step": 35012, "epoch": 330, "lr": 2.5826655871505147e-05} +{"train_loss": 0.01477068942040205, "grad_norm": 0.06870654970407486, "global_step": 35013, "epoch": 330, "lr": 2.582406154333393e-05} +{"train_loss": 0.015126615762710571, "grad_norm": 0.05939750745892525, "global_step": 35014, "epoch": 330, "lr": 2.582146730010646e-05} +{"train_loss": 0.01405249536037445, "grad_norm": 0.1069200411438942, "global_step": 35015, "epoch": 330, "lr": 2.581887314183183e-05} +{"train_loss": 0.013502577319741249, "grad_norm": 0.09469062834978104, "global_step": 35016, "epoch": 330, "lr": 2.5816279068519176e-05} +{"train_loss": 0.017392758280038834, "grad_norm": 0.08071678131818771, "global_step": 35017, "epoch": 330, "lr": 2.5813685080177586e-05} +{"train_loss": 0.01617448963224888, "grad_norm": 0.10858514159917831, "global_step": 35018, "epoch": 330, "lr": 2.5811091176816203e-05} +{"train_loss": 0.01590900495648384, "grad_norm": 0.0747455433011055, "global_step": 35019, "epoch": 330, "lr": 2.5808497358444106e-05} +{"train_loss": 0.01701054722070694, "grad_norm": 0.07148580998182297, "global_step": 35020, "epoch": 330, "lr": 2.580590362507043e-05} +{"train_loss": 0.012970451265573502, "grad_norm": 0.06411862373352051, "global_step": 35021, "epoch": 330, "lr": 2.580330997670431e-05} +{"train_loss": 0.015990395098924637, "grad_norm": 0.17941616475582123, "global_step": 35022, "epoch": 330, "lr": 2.5800716413354832e-05} +{"train_loss": 0.015714535489678383, "grad_norm": 0.08971399068832397, "global_step": 35023, "epoch": 330, "lr": 2.5798122935031095e-05} +{"train_loss": 0.012527313083410263, "grad_norm": 0.06951452791690826, "global_step": 35024, "epoch": 330, "lr": 2.5795529541742246e-05} +{"train_loss": 0.01495880726724863, "grad_norm": 0.06932057440280914, "global_step": 35025, "epoch": 330, "lr": 2.579293623349736e-05} +{"train_loss": 0.014931158162653446, "grad_norm": 0.08192048966884613, "global_step": 35026, "epoch": 330, "lr": 2.5790343010305572e-05} +{"train_loss": 0.015329539775848389, "grad_norm": 0.12487252801656723, "global_step": 35027, "epoch": 330, "lr": 2.5787749872175993e-05} +{"train_loss": 0.01597805693745613, "grad_norm": 0.07174403220415115, "global_step": 35028, "epoch": 330, "lr": 2.578515681911776e-05} +{"train_loss": 0.015916448086500168, "grad_norm": 0.07842988520860672, "global_step": 35029, "epoch": 330, "lr": 2.5782563851139924e-05} +{"train_loss": 0.015793249011039734, "grad_norm": 0.09550482779741287, "global_step": 35030, "epoch": 330, "lr": 2.577997096825162e-05} +{"train_loss": 0.01468929834663868, "grad_norm": 0.07203710079193115, "global_step": 35031, "epoch": 330, "lr": 2.5777378170461997e-05} +{"train_loss": 0.015323498286306858, "grad_norm": 0.07326165586709976, "global_step": 35032, "epoch": 330, "lr": 2.5774785457780103e-05} +{"train_loss": 0.015016620047390461, "grad_norm": 0.06761019676923752, "global_step": 35033, "epoch": 330, "lr": 2.577219283021508e-05} +{"train_loss": 0.014007531106472015, "grad_norm": 0.09375175088644028, "global_step": 35034, "epoch": 330, "lr": 2.5769600287776054e-05} +{"train_loss": 0.015146255493164062, "grad_norm": 0.06903428584337234, "global_step": 35035, "epoch": 330, "lr": 2.5767007830472116e-05} +{"train_loss": 0.013606233522295952, "grad_norm": 0.0886458158493042, "global_step": 35036, "epoch": 330, "lr": 2.5764415458312358e-05} +{"train_loss": 0.016431596130132675, "grad_norm": 0.057844728231430054, "global_step": 35037, "epoch": 330, "lr": 2.57618231713059e-05} +{"train_loss": 0.015213247388601303, "grad_norm": 0.09003309905529022, "global_step": 35038, "epoch": 330, "lr": 2.5759230969461874e-05} +{"train_loss": 0.013756017200648785, "grad_norm": 0.06345781683921814, "global_step": 35039, "epoch": 330, "lr": 2.575663885278935e-05} +{"train_loss": 0.014870785176753998, "grad_norm": 0.08338513970375061, "global_step": 35040, "epoch": 330, "lr": 2.5754046821297474e-05} +{"train_loss": 0.01812148280441761, "grad_norm": 0.12543462216854095, "global_step": 35041, "epoch": 330, "lr": 2.575145487499531e-05} +{"train_loss": 0.01649477332830429, "grad_norm": 0.18024082481861115, "global_step": 35042, "epoch": 330, "lr": 2.574886301389201e-05} +{"train_loss": 0.014533694833517075, "grad_norm": 0.0991135984659195, "global_step": 35043, "epoch": 330, "lr": 2.5746271237996643e-05} +{"train_loss": 0.014250615611672401, "grad_norm": 0.08732368052005768, "global_step": 35044, "epoch": 330, "lr": 2.5743679547318323e-05} +{"train_loss": 0.016038943082094193, "grad_norm": 0.0728372260928154, "global_step": 35045, "epoch": 330, "lr": 2.5741087941866195e-05} +{"train_loss": 0.014661229215562344, "grad_norm": 0.08113379031419754, "global_step": 35046, "epoch": 330, "lr": 2.573849642164933e-05} +{"train_loss": 0.014863207936286926, "grad_norm": 0.06362058967351913, "global_step": 35047, "epoch": 330, "lr": 2.5735904986676808e-05} +{"train_loss": 0.014867288991808891, "grad_norm": 0.09346718341112137, "global_step": 35048, "epoch": 330, "lr": 2.573331363695779e-05} +{"train_loss": 0.017182473093271255, "grad_norm": 0.07719352841377258, "global_step": 35049, "epoch": 330, "lr": 2.5730722372501337e-05} +{"train_loss": 0.01477293111383915, "grad_norm": 0.057469986379146576, "global_step": 35050, "epoch": 330, "lr": 2.5728131193316575e-05} +{"train_loss": 0.01806122623383999, "grad_norm": 0.10089361667633057, "global_step": 35051, "epoch": 330, "lr": 2.5725540099412614e-05} +{"train_loss": 0.012754194438457489, "grad_norm": 0.06080589070916176, "global_step": 35052, "epoch": 330, "lr": 2.5722949090798555e-05} +{"train_loss": 0.015616180375218391, "grad_norm": 0.07602014392614365, "global_step": 35053, "epoch": 330, "lr": 2.5720358167483473e-05} +{"train_loss": 0.01426667720079422, "grad_norm": 0.08933648467063904, "global_step": 35054, "epoch": 330, "lr": 2.5717767329476493e-05} +{"train_loss": 0.01638077013194561, "grad_norm": 0.05962161719799042, "global_step": 35055, "epoch": 330, "lr": 2.571517657678674e-05} +{"train_loss": 0.016607951372861862, "grad_norm": 0.10686351358890533, "global_step": 35056, "epoch": 330, "lr": 2.571258590942328e-05} +{"train_loss": 0.01604642905294895, "grad_norm": 0.1578301191329956, "global_step": 35057, "epoch": 330, "lr": 2.570999532739524e-05} +{"train_loss": 0.01358419843018055, "grad_norm": 0.06437557190656662, "global_step": 35058, "epoch": 330, "lr": 2.57074048307117e-05} +{"train_loss": 0.016489826142787933, "grad_norm": 0.10400224477052689, "global_step": 35059, "epoch": 330, "lr": 2.5704814419381794e-05} +{"train_loss": 0.016010724008083344, "grad_norm": 0.06285044550895691, "global_step": 35060, "epoch": 330, "lr": 2.5702224093414583e-05} +{"train_loss": 0.014624754898250103, "grad_norm": 0.10066366195678711, "global_step": 35061, "epoch": 330, "lr": 2.5699633852819193e-05} +{"train_loss": 0.014096662402153015, "grad_norm": 0.06451692432165146, "global_step": 35062, "epoch": 330, "lr": 2.569704369760474e-05} +{"train_loss": 0.012691386044025421, "grad_norm": 0.13870085775852203, "global_step": 35063, "epoch": 330, "lr": 2.5694453627780303e-05} +{"train_loss": 0.01587729901075363, "grad_norm": 0.0667007714509964, "global_step": 35064, "epoch": 330, "lr": 2.5691863643354974e-05} +{"train_loss": 0.014684153720736504, "grad_norm": 0.06540943682193756, "global_step": 35065, "epoch": 330, "lr": 2.5689273744337878e-05} +{"train_loss": 0.015633128583431244, "grad_norm": 0.09429007023572922, "global_step": 35066, "epoch": 330, "lr": 2.568668393073808e-05} +{"train_loss": 0.014606280252337456, "grad_norm": 0.05916953459382057, "global_step": 35067, "epoch": 330, "lr": 2.5684094202564708e-05} +{"train_loss": 0.016335856169462204, "grad_norm": 0.05963955819606781, "global_step": 35068, "epoch": 330, "lr": 2.568150455982687e-05} +{"train_loss": 0.013589765876531601, "grad_norm": 0.0684085488319397, "global_step": 35069, "epoch": 330, "lr": 2.567891500253365e-05} +{"train_loss": 0.014141978695988655, "grad_norm": 0.08824513852596283, "global_step": 35070, "epoch": 330, "lr": 2.5676325530694123e-05} +{"train_loss": 0.014313094317913055, "grad_norm": 0.06295966356992722, "global_step": 35071, "epoch": 330, "lr": 2.567373614431742e-05} +{"train_loss": 0.015688739717006683, "grad_norm": 0.05739549919962883, "global_step": 35072, "epoch": 330, "lr": 2.5671146843412635e-05} +{"train_loss": 0.016544103622436523, "grad_norm": 0.07536476105451584, "global_step": 35073, "epoch": 330, "lr": 2.5668557627988844e-05} +{"train_loss": 0.01637943834066391, "grad_norm": 0.0632285624742508, "global_step": 35074, "epoch": 330, "lr": 2.5665968498055167e-05} +{"train_loss": 0.016337817534804344, "grad_norm": 0.06870999932289124, "global_step": 35075, "epoch": 330, "lr": 2.5663379453620707e-05} +{"train_loss": 0.01432124339044094, "grad_norm": 0.09061718732118607, "global_step": 35076, "epoch": 330, "lr": 2.566079049469455e-05} +{"train_loss": 0.0153246084228158, "grad_norm": 0.09776636958122253, "global_step": 35077, "epoch": 330, "lr": 2.5658201621285772e-05} +{"train_loss": 0.013203312642872334, "grad_norm": 0.0717225968837738, "global_step": 35078, "epoch": 330, "lr": 2.5655612833403487e-05} +{"train_loss": 0.014213824644684792, "grad_norm": 0.0872916728258133, "global_step": 35079, "epoch": 330, "lr": 2.5653024131056802e-05} +{"train_loss": 0.013787055388092995, "grad_norm": 0.07938572019338608, "global_step": 35080, "epoch": 330, "lr": 2.5650435514254793e-05} +{"train_loss": 0.015362763777375221, "grad_norm": 0.0665769875049591, "global_step": 35081, "epoch": 330, "lr": 2.5647846983006585e-05} +{"train_loss": 0.013693503104150295, "grad_norm": 0.10644637048244476, "global_step": 35082, "epoch": 330, "lr": 2.5645258537321227e-05} +{"train_loss": 0.015312539413571358, "grad_norm": 0.0945819765329361, "global_step": 35083, "epoch": 330, "lr": 2.564267017720786e-05} +{"train_loss": 0.016159258782863617, "grad_norm": 0.06312564015388489, "global_step": 35084, "epoch": 330, "lr": 2.5640081902675538e-05} +{"train_loss": 0.015153085558130493, "grad_norm": 0.10989335179328918, "global_step": 35085, "epoch": 330, "lr": 2.5637493713733374e-05, "train_batch_mse_error": 0.013485822826623917, "val_loss": 0.019830234348773956} +{"train_loss": 0.016171041876077652, "grad_norm": 0.09593595564365387, "global_step": 35086, "epoch": 331, "lr": 2.563490561039048e-05} +{"train_loss": 0.015644287690520287, "grad_norm": 0.06776998937129974, "global_step": 35087, "epoch": 331, "lr": 2.5632317592655934e-05} +{"train_loss": 0.014483612962067127, "grad_norm": 0.10076933354139328, "global_step": 35088, "epoch": 331, "lr": 2.562972966053881e-05} +{"train_loss": 0.016110777854919434, "grad_norm": 0.0704188272356987, "global_step": 35089, "epoch": 331, "lr": 2.5627141814048234e-05} +{"train_loss": 0.013717075809836388, "grad_norm": 0.09385491162538528, "global_step": 35090, "epoch": 331, "lr": 2.5624554053193262e-05} +{"train_loss": 0.013524052686989307, "grad_norm": 0.09412872046232224, "global_step": 35091, "epoch": 331, "lr": 2.5621966377983008e-05} +{"train_loss": 0.014754015952348709, "grad_norm": 0.07008031010627747, "global_step": 35092, "epoch": 331, "lr": 2.5619378788426586e-05} +{"train_loss": 0.016545835882425308, "grad_norm": 0.06210187077522278, "global_step": 35093, "epoch": 331, "lr": 2.5616791284533055e-05} +{"train_loss": 0.012495895847678185, "grad_norm": 0.05987963452935219, "global_step": 35094, "epoch": 331, "lr": 2.5614203866311503e-05} +{"train_loss": 0.013454952277243137, "grad_norm": 0.06006640940904617, "global_step": 35095, "epoch": 331, "lr": 2.5611616533771026e-05} +{"train_loss": 0.017043884843587875, "grad_norm": 0.10638194531202316, "global_step": 35096, "epoch": 331, "lr": 2.560902928692075e-05} +{"train_loss": 0.011898391880095005, "grad_norm": 0.07444372773170471, "global_step": 35097, "epoch": 331, "lr": 2.5606442125769713e-05} +{"train_loss": 0.013877371326088905, "grad_norm": 0.05441545322537422, "global_step": 35098, "epoch": 331, "lr": 2.5603855050327053e-05} +{"train_loss": 0.015200082212686539, "grad_norm": 0.08058038353919983, "global_step": 35099, "epoch": 331, "lr": 2.5601268060601813e-05} +{"train_loss": 0.015854384750127792, "grad_norm": 0.08962930738925934, "global_step": 35100, "epoch": 331, "lr": 2.5598681156603128e-05} +{"train_loss": 0.012945486232638359, "grad_norm": 0.09813375771045685, "global_step": 35101, "epoch": 331, "lr": 2.5596094338340047e-05} +{"train_loss": 0.017082959413528442, "grad_norm": 0.08549495786428452, "global_step": 35102, "epoch": 331, "lr": 2.5593507605821675e-05} +{"train_loss": 0.015903152525424957, "grad_norm": 0.09829548746347427, "global_step": 35103, "epoch": 331, "lr": 2.5590920959057124e-05} +{"train_loss": 0.016020243987441063, "grad_norm": 0.07956164330244064, "global_step": 35104, "epoch": 331, "lr": 2.5588334398055463e-05} +{"train_loss": 0.017371121793985367, "grad_norm": 0.07284552603960037, "global_step": 35105, "epoch": 331, "lr": 2.558574792282576e-05} +{"train_loss": 0.014592096209526062, "grad_norm": 0.0767938643693924, "global_step": 35106, "epoch": 331, "lr": 2.5583161533377137e-05} +{"train_loss": 0.01578124612569809, "grad_norm": 0.06957131624221802, "global_step": 35107, "epoch": 331, "lr": 2.558057522971865e-05} +{"train_loss": 0.01630060374736786, "grad_norm": 0.06588240712881088, "global_step": 35108, "epoch": 331, "lr": 2.5577989011859398e-05} +{"train_loss": 0.016101285815238953, "grad_norm": 0.06587406992912292, "global_step": 35109, "epoch": 331, "lr": 2.5575402879808495e-05} +{"train_loss": 0.01909361407160759, "grad_norm": 0.11311671137809753, "global_step": 35110, "epoch": 331, "lr": 2.5572816833574997e-05} +{"train_loss": 0.016041789203882217, "grad_norm": 0.07808973640203476, "global_step": 35111, "epoch": 331, "lr": 2.557023087316799e-05} +{"train_loss": 0.01519216038286686, "grad_norm": 0.13091543316841125, "global_step": 35112, "epoch": 331, "lr": 2.5567644998596563e-05} +{"train_loss": 0.017672188580036163, "grad_norm": 0.10813222080469131, "global_step": 35113, "epoch": 331, "lr": 2.556505920986982e-05} +{"train_loss": 0.014442026615142822, "grad_norm": 0.06650690734386444, "global_step": 35114, "epoch": 331, "lr": 2.556247350699682e-05} +{"train_loss": 0.01621203124523163, "grad_norm": 0.07627400010824203, "global_step": 35115, "epoch": 331, "lr": 2.5559887889986656e-05} +{"train_loss": 0.017643526196479797, "grad_norm": 0.08690371364355087, "global_step": 35116, "epoch": 331, "lr": 2.5557302358848434e-05} +{"train_loss": 0.013899190351366997, "grad_norm": 0.0824580267071724, "global_step": 35117, "epoch": 331, "lr": 2.5554716913591226e-05} +{"train_loss": 0.015296957455575466, "grad_norm": 0.0681043341755867, "global_step": 35118, "epoch": 331, "lr": 2.555213155422409e-05} +{"train_loss": 0.014720652252435684, "grad_norm": 0.12669646739959717, "global_step": 35119, "epoch": 331, "lr": 2.5549546280756137e-05} +{"train_loss": 0.01736448146402836, "grad_norm": 0.15421012043952942, "global_step": 35120, "epoch": 331, "lr": 2.5546961093196465e-05} +{"train_loss": 0.01649005338549614, "grad_norm": 0.06567979604005814, "global_step": 35121, "epoch": 331, "lr": 2.554437599155411e-05} +{"train_loss": 0.014870474115014076, "grad_norm": 0.07765033841133118, "global_step": 35122, "epoch": 331, "lr": 2.5541790975838208e-05} +{"train_loss": 0.015796352177858353, "grad_norm": 0.0721505731344223, "global_step": 35123, "epoch": 331, "lr": 2.553920604605781e-05} +{"train_loss": 0.013235550373792648, "grad_norm": 0.07318690419197083, "global_step": 35124, "epoch": 331, "lr": 2.5536621202221987e-05} +{"train_loss": 0.01428482960909605, "grad_norm": 0.10300538688898087, "global_step": 35125, "epoch": 331, "lr": 2.5534036444339842e-05} +{"train_loss": 0.014348801225423813, "grad_norm": 0.11672812700271606, "global_step": 35126, "epoch": 331, "lr": 2.5531451772420468e-05} +{"train_loss": 0.016686994582414627, "grad_norm": 0.08771022409200668, "global_step": 35127, "epoch": 331, "lr": 2.5528867186472916e-05} +{"train_loss": 0.014435049146413803, "grad_norm": 0.08092037588357925, "global_step": 35128, "epoch": 331, "lr": 2.5526282686506296e-05} +{"train_loss": 0.018745334818959236, "grad_norm": 0.0748818889260292, "global_step": 35129, "epoch": 331, "lr": 2.552369827252966e-05} +{"train_loss": 0.013704312033951283, "grad_norm": 0.09143134951591492, "global_step": 35130, "epoch": 331, "lr": 2.5521113944552126e-05} +{"train_loss": 0.015351465903222561, "grad_norm": 0.06905680149793625, "global_step": 35131, "epoch": 331, "lr": 2.551852970258273e-05} +{"train_loss": 0.013331092894077301, "grad_norm": 0.08083531260490417, "global_step": 35132, "epoch": 331, "lr": 2.5515945546630575e-05} +{"train_loss": 0.015537742525339127, "grad_norm": 0.0929049625992775, "global_step": 35133, "epoch": 331, "lr": 2.5513361476704754e-05} +{"train_loss": 0.015696100890636444, "grad_norm": 0.08180902898311615, "global_step": 35134, "epoch": 331, "lr": 2.5510777492814343e-05} +{"train_loss": 0.015299654565751553, "grad_norm": 0.05667611584067345, "global_step": 35135, "epoch": 331, "lr": 2.5508193594968378e-05} +{"train_loss": 0.014196922071278095, "grad_norm": 0.07595434039831161, "global_step": 35136, "epoch": 331, "lr": 2.5505609783175977e-05} +{"train_loss": 0.014500327408313751, "grad_norm": 0.09748449921607971, "global_step": 35137, "epoch": 331, "lr": 2.5503026057446222e-05} +{"train_loss": 0.014119279570877552, "grad_norm": 0.07230082899332047, "global_step": 35138, "epoch": 331, "lr": 2.550044241778817e-05} +{"train_loss": 0.012609483674168587, "grad_norm": 0.13135895133018494, "global_step": 35139, "epoch": 331, "lr": 2.5497858864210905e-05} +{"train_loss": 0.013957666233181953, "grad_norm": 0.1302119791507721, "global_step": 35140, "epoch": 331, "lr": 2.549527539672353e-05} +{"train_loss": 0.013969603925943375, "grad_norm": 0.058191850781440735, "global_step": 35141, "epoch": 331, "lr": 2.5492692015335097e-05} +{"train_loss": 0.015192429535090923, "grad_norm": 0.06409485638141632, "global_step": 35142, "epoch": 331, "lr": 2.549010872005467e-05} +{"train_loss": 0.016594499349594116, "grad_norm": 0.10422058403491974, "global_step": 35143, "epoch": 331, "lr": 2.5487525510891335e-05} +{"train_loss": 0.012253444641828537, "grad_norm": 0.1038566306233406, "global_step": 35144, "epoch": 331, "lr": 2.54849423878542e-05} +{"train_loss": 0.015199458226561546, "grad_norm": 0.11746690422296524, "global_step": 35145, "epoch": 331, "lr": 2.54823593509523e-05} +{"train_loss": 0.015005282126367092, "grad_norm": 0.082362100481987, "global_step": 35146, "epoch": 331, "lr": 2.5479776400194737e-05} +{"train_loss": 0.012524295598268509, "grad_norm": 0.10012444108724594, "global_step": 35147, "epoch": 331, "lr": 2.5477193535590582e-05} +{"train_loss": 0.01618371717631817, "grad_norm": 0.11341944336891174, "global_step": 35148, "epoch": 331, "lr": 2.5474610757148886e-05} +{"train_loss": 0.014603259041905403, "grad_norm": 0.09530756622552872, "global_step": 35149, "epoch": 331, "lr": 2.547202806487874e-05} +{"train_loss": 0.013987667858600616, "grad_norm": 0.07902246713638306, "global_step": 35150, "epoch": 331, "lr": 2.5469445458789243e-05} +{"train_loss": 0.014549260027706623, "grad_norm": 0.08362293243408203, "global_step": 35151, "epoch": 331, "lr": 2.546686293888944e-05} +{"train_loss": 0.014227299951016903, "grad_norm": 0.06817139685153961, "global_step": 35152, "epoch": 331, "lr": 2.5464280505188398e-05} +{"train_loss": 0.013431066647171974, "grad_norm": 0.0744672417640686, "global_step": 35153, "epoch": 331, "lr": 2.5461698157695202e-05} +{"train_loss": 0.017573703080415726, "grad_norm": 0.10086345672607422, "global_step": 35154, "epoch": 331, "lr": 2.5459115896418945e-05} +{"train_loss": 0.014257559552788734, "grad_norm": 0.09752901643514633, "global_step": 35155, "epoch": 331, "lr": 2.5456533721368663e-05} +{"train_loss": 0.01611444540321827, "grad_norm": 0.08502384275197983, "global_step": 35156, "epoch": 331, "lr": 2.545395163255345e-05} +{"train_loss": 0.014343277551233768, "grad_norm": 0.06261057406663895, "global_step": 35157, "epoch": 331, "lr": 2.545136962998239e-05} +{"train_loss": 0.015849795192480087, "grad_norm": 0.09967008978128433, "global_step": 35158, "epoch": 331, "lr": 2.544878771366455e-05} +{"train_loss": 0.01595160737633705, "grad_norm": 0.10817203670740128, "global_step": 35159, "epoch": 331, "lr": 2.544620588360897e-05} +{"train_loss": 0.01534268632531166, "grad_norm": 0.1345014125108719, "global_step": 35160, "epoch": 331, "lr": 2.544362413982474e-05} +{"train_loss": 0.013387452811002731, "grad_norm": 0.06671205163002014, "global_step": 35161, "epoch": 331, "lr": 2.5441042482320964e-05} +{"train_loss": 0.014393175020813942, "grad_norm": 0.0607965923845768, "global_step": 35162, "epoch": 331, "lr": 2.5438460911106654e-05} +{"train_loss": 0.014816159382462502, "grad_norm": 0.11205706000328064, "global_step": 35163, "epoch": 331, "lr": 2.543587942619094e-05} +{"train_loss": 0.01507135946303606, "grad_norm": 0.0615505650639534, "global_step": 35164, "epoch": 331, "lr": 2.543329802758286e-05} +{"train_loss": 0.014971235767006874, "grad_norm": 0.08166487514972687, "global_step": 35165, "epoch": 331, "lr": 2.5430716715291464e-05} +{"train_loss": 0.014383861795067787, "grad_norm": 0.09328487515449524, "global_step": 35166, "epoch": 331, "lr": 2.5428135489325845e-05} +{"train_loss": 0.016620587557554245, "grad_norm": 0.06842334568500519, "global_step": 35167, "epoch": 331, "lr": 2.5425554349695096e-05} +{"train_loss": 0.01556534506380558, "grad_norm": 0.12121269106864929, "global_step": 35168, "epoch": 331, "lr": 2.5422973296408236e-05} +{"train_loss": 0.01426325086504221, "grad_norm": 0.08455904573202133, "global_step": 35169, "epoch": 331, "lr": 2.5420392329474386e-05} +{"train_loss": 0.015040499158203602, "grad_norm": 0.05965448543429375, "global_step": 35170, "epoch": 331, "lr": 2.5417811448902562e-05} +{"train_loss": 0.015999851748347282, "grad_norm": 0.080958291888237, "global_step": 35171, "epoch": 331, "lr": 2.541523065470188e-05} +{"train_loss": 0.015635233372449875, "grad_norm": 0.06784301996231079, "global_step": 35172, "epoch": 331, "lr": 2.5412649946881373e-05} +{"train_loss": 0.016846852377057076, "grad_norm": 0.06783883273601532, "global_step": 35173, "epoch": 331, "lr": 2.5410069325450114e-05} +{"train_loss": 0.01585402712225914, "grad_norm": 0.09767352044582367, "global_step": 35174, "epoch": 331, "lr": 2.54074887904172e-05} +{"train_loss": 0.01370309479534626, "grad_norm": 0.0934615358710289, "global_step": 35175, "epoch": 331, "lr": 2.5404908341791672e-05} +{"train_loss": 0.016504906117916107, "grad_norm": 0.08368366956710815, "global_step": 35176, "epoch": 331, "lr": 2.5402327979582585e-05} +{"train_loss": 0.013794975355267525, "grad_norm": 0.0843672975897789, "global_step": 35177, "epoch": 331, "lr": 2.5399747703799014e-05} +{"train_loss": 0.013419104740023613, "grad_norm": 0.055812861770391464, "global_step": 35178, "epoch": 331, "lr": 2.5397167514450056e-05} +{"train_loss": 0.016811763867735863, "grad_norm": 0.08669215440750122, "global_step": 35179, "epoch": 331, "lr": 2.539458741154473e-05} +{"train_loss": 0.015596741810441017, "grad_norm": 0.07978547364473343, "global_step": 35180, "epoch": 331, "lr": 2.539200739509212e-05} +{"train_loss": 0.01638692617416382, "grad_norm": 0.09098204970359802, "global_step": 35181, "epoch": 331, "lr": 2.538942746510134e-05} +{"train_loss": 0.015918811783194542, "grad_norm": 0.06783096492290497, "global_step": 35182, "epoch": 331, "lr": 2.5386847621581367e-05} +{"train_loss": 0.015514891594648361, "grad_norm": 0.056068550795316696, "global_step": 35183, "epoch": 331, "lr": 2.538426786454131e-05} +{"train_loss": 0.015495209023356438, "grad_norm": 0.07874143123626709, "global_step": 35184, "epoch": 331, "lr": 2.538168819399025e-05} +{"train_loss": 0.01641327701508999, "grad_norm": 0.07908427715301514, "global_step": 35185, "epoch": 331, "lr": 2.5379108609937208e-05} +{"train_loss": 0.014043580740690231, "grad_norm": 0.15619833767414093, "global_step": 35186, "epoch": 331, "lr": 2.5376529112391266e-05} +{"train_loss": 0.013664636760950089, "grad_norm": 0.0702396109700203, "global_step": 35187, "epoch": 331, "lr": 2.537394970136152e-05} +{"train_loss": 0.015952218323946, "grad_norm": 0.08133845776319504, "global_step": 35188, "epoch": 331, "lr": 2.5371370376857007e-05} +{"train_loss": 0.016570590436458588, "grad_norm": 0.08953005820512772, "global_step": 35189, "epoch": 331, "lr": 2.536879113888676e-05} +{"train_loss": 0.012638524174690247, "grad_norm": 0.07996279001235962, "global_step": 35190, "epoch": 331, "lr": 2.5366211987459874e-05} +{"train_loss": 0.01515684493834961, "grad_norm": 0.10512107610702515, "global_step": 35191, "epoch": 331, "lr": 2.536363292258543e-05, "val_loss": 0.02003188244998455} +{"train_loss": 0.014904151670634747, "grad_norm": 0.10307244211435318, "global_step": 35192, "epoch": 332, "lr": 2.536105394427244e-05} +{"train_loss": 0.01520178560167551, "grad_norm": 0.09709591418504715, "global_step": 35193, "epoch": 332, "lr": 2.5358475052530018e-05} +{"train_loss": 0.017181621864438057, "grad_norm": 0.05654625967144966, "global_step": 35194, "epoch": 332, "lr": 2.535589624736717e-05} +{"train_loss": 0.015017326921224594, "grad_norm": 0.12127449363470078, "global_step": 35195, "epoch": 332, "lr": 2.535331752879301e-05} +{"train_loss": 0.016004882752895355, "grad_norm": 0.0696190893650055, "global_step": 35196, "epoch": 332, "lr": 2.535073889681655e-05} +{"train_loss": 0.015208302065730095, "grad_norm": 0.05449102818965912, "global_step": 35197, "epoch": 332, "lr": 2.5348160351446875e-05} +{"train_loss": 0.016142800450325012, "grad_norm": 0.11871914565563202, "global_step": 35198, "epoch": 332, "lr": 2.5345581892693066e-05} +{"train_loss": 0.01617124304175377, "grad_norm": 0.06786729395389557, "global_step": 35199, "epoch": 332, "lr": 2.5343003520564158e-05} +{"train_loss": 0.013412288390100002, "grad_norm": 0.06758321821689606, "global_step": 35200, "epoch": 332, "lr": 2.5340425235069194e-05} +{"train_loss": 0.015432466752827168, "grad_norm": 0.06740494817495346, "global_step": 35201, "epoch": 332, "lr": 2.5337847036217254e-05} +{"train_loss": 0.01883566752076149, "grad_norm": 0.07339288294315338, "global_step": 35202, "epoch": 332, "lr": 2.5335268924017407e-05} +{"train_loss": 0.015918951481580734, "grad_norm": 0.09272488951683044, "global_step": 35203, "epoch": 332, "lr": 2.5332690898478684e-05} +{"train_loss": 0.015207987278699875, "grad_norm": 0.09991225600242615, "global_step": 35204, "epoch": 332, "lr": 2.5330112959610175e-05} +{"train_loss": 0.013369761407375336, "grad_norm": 0.06755668669939041, "global_step": 35205, "epoch": 332, "lr": 2.5327535107420925e-05} +{"train_loss": 0.016050949692726135, "grad_norm": 0.06823571771383286, "global_step": 35206, "epoch": 332, "lr": 2.5324957341919965e-05} +{"train_loss": 0.01414898969233036, "grad_norm": 0.07483009994029999, "global_step": 35207, "epoch": 332, "lr": 2.5322379663116368e-05} +{"train_loss": 0.0144649688154459, "grad_norm": 0.0650469958782196, "global_step": 35208, "epoch": 332, "lr": 2.5319802071019227e-05} +{"train_loss": 0.016696345061063766, "grad_norm": 0.07409199327230453, "global_step": 35209, "epoch": 332, "lr": 2.5317224565637544e-05} +{"train_loss": 0.016022972762584686, "grad_norm": 0.06407463550567627, "global_step": 35210, "epoch": 332, "lr": 2.5314647146980415e-05} +{"train_loss": 0.015165921300649643, "grad_norm": 0.10709499567747116, "global_step": 35211, "epoch": 332, "lr": 2.5312069815056867e-05} +{"train_loss": 0.016953300684690475, "grad_norm": 0.12112259119749069, "global_step": 35212, "epoch": 332, "lr": 2.530949256987598e-05} +{"train_loss": 0.014572730287909508, "grad_norm": 0.06982392072677612, "global_step": 35213, "epoch": 332, "lr": 2.5306915411446787e-05} +{"train_loss": 0.014386147260665894, "grad_norm": 0.1390700489282608, "global_step": 35214, "epoch": 332, "lr": 2.5304338339778354e-05} +{"train_loss": 0.0162182729691267, "grad_norm": 0.08479227125644684, "global_step": 35215, "epoch": 332, "lr": 2.5301761354879756e-05} +{"train_loss": 0.016259629279375076, "grad_norm": 0.09596291929483414, "global_step": 35216, "epoch": 332, "lr": 2.529918445676003e-05} +{"train_loss": 0.017130712047219276, "grad_norm": 0.11860983073711395, "global_step": 35217, "epoch": 332, "lr": 2.52966076454282e-05} +{"train_loss": 0.014784988015890121, "grad_norm": 0.13446670770645142, "global_step": 35218, "epoch": 332, "lr": 2.5294030920893354e-05} +{"train_loss": 0.015508348122239113, "grad_norm": 0.07275161892175674, "global_step": 35219, "epoch": 332, "lr": 2.5291454283164552e-05} +{"train_loss": 0.015149921178817749, "grad_norm": 0.1561858355998993, "global_step": 35220, "epoch": 332, "lr": 2.528887773225082e-05} +{"train_loss": 0.01657203957438469, "grad_norm": 0.06536899507045746, "global_step": 35221, "epoch": 332, "lr": 2.5286301268161227e-05} +{"train_loss": 0.01622680388391018, "grad_norm": 0.0733017772436142, "global_step": 35222, "epoch": 332, "lr": 2.528372489090486e-05} +{"train_loss": 0.014680037274956703, "grad_norm": 0.0504843071103096, "global_step": 35223, "epoch": 332, "lr": 2.5281148600490688e-05} +{"train_loss": 0.01765100657939911, "grad_norm": 0.1840524673461914, "global_step": 35224, "epoch": 332, "lr": 2.5278572396927824e-05} +{"train_loss": 0.015218369662761688, "grad_norm": 0.10259249806404114, "global_step": 35225, "epoch": 332, "lr": 2.527599628022532e-05} +{"train_loss": 0.016164250671863556, "grad_norm": 0.0938062071800232, "global_step": 35226, "epoch": 332, "lr": 2.5273420250392188e-05} +{"train_loss": 0.015701504424214363, "grad_norm": 0.1364458203315735, "global_step": 35227, "epoch": 332, "lr": 2.527084430743751e-05} +{"train_loss": 0.014807883650064468, "grad_norm": 0.09260299801826477, "global_step": 35228, "epoch": 332, "lr": 2.5268268451370348e-05} +{"train_loss": 0.013972335495054722, "grad_norm": 0.06864644587039948, "global_step": 35229, "epoch": 332, "lr": 2.5265692682199732e-05} +{"train_loss": 0.015147710219025612, "grad_norm": 0.10998842120170593, "global_step": 35230, "epoch": 332, "lr": 2.52631169999347e-05} +{"train_loss": 0.01569439470767975, "grad_norm": 0.06857019662857056, "global_step": 35231, "epoch": 332, "lr": 2.5260541404584313e-05} +{"train_loss": 0.012078548781573772, "grad_norm": 0.058767374604940414, "global_step": 35232, "epoch": 332, "lr": 2.5257965896157648e-05} +{"train_loss": 0.013279840350151062, "grad_norm": 0.065533846616745, "global_step": 35233, "epoch": 332, "lr": 2.525539047466371e-05} +{"train_loss": 0.01580699533224106, "grad_norm": 0.08416277915239334, "global_step": 35234, "epoch": 332, "lr": 2.525281514011159e-05} +{"train_loss": 0.014066816307604313, "grad_norm": 0.09128350019454956, "global_step": 35235, "epoch": 332, "lr": 2.5250239892510296e-05} +{"train_loss": 0.014226706698536873, "grad_norm": 0.17625722289085388, "global_step": 35236, "epoch": 332, "lr": 2.5247664731868915e-05} +{"train_loss": 0.015476249158382416, "grad_norm": 0.06402341276407242, "global_step": 35237, "epoch": 332, "lr": 2.5245089658196452e-05} +{"train_loss": 0.012986605986952782, "grad_norm": 0.08906295895576477, "global_step": 35238, "epoch": 332, "lr": 2.5242514671501984e-05} +{"train_loss": 0.014321528375148773, "grad_norm": 0.08149678260087967, "global_step": 35239, "epoch": 332, "lr": 2.5239939771794563e-05} +{"train_loss": 0.015130925923585892, "grad_norm": 0.08403465151786804, "global_step": 35240, "epoch": 332, "lr": 2.523736495908323e-05} +{"train_loss": 0.018688509240746498, "grad_norm": 0.06185091286897659, "global_step": 35241, "epoch": 332, "lr": 2.5234790233377015e-05} +{"train_loss": 0.013074563816189766, "grad_norm": 0.07051119208335876, "global_step": 35242, "epoch": 332, "lr": 2.5232215594684973e-05} +{"train_loss": 0.01511678658425808, "grad_norm": 0.10615499317646027, "global_step": 35243, "epoch": 332, "lr": 2.5229641043016174e-05} +{"train_loss": 0.013665455393493176, "grad_norm": 0.07234979420900345, "global_step": 35244, "epoch": 332, "lr": 2.522706657837962e-05} +{"train_loss": 0.013893503695726395, "grad_norm": 0.17469404637813568, "global_step": 35245, "epoch": 332, "lr": 2.5224492200784383e-05} +{"train_loss": 0.015651635825634003, "grad_norm": 0.06056760251522064, "global_step": 35246, "epoch": 332, "lr": 2.5221917910239545e-05} +{"train_loss": 0.015616245567798615, "grad_norm": 0.13948149979114532, "global_step": 35247, "epoch": 332, "lr": 2.521934370675407e-05} +{"train_loss": 0.015507243573665619, "grad_norm": 0.05517352744936943, "global_step": 35248, "epoch": 332, "lr": 2.5216769590337052e-05} +{"train_loss": 0.013576822355389595, "grad_norm": 0.13661573827266693, "global_step": 35249, "epoch": 332, "lr": 2.5214195560997544e-05} +{"train_loss": 0.016435198485851288, "grad_norm": 0.09857355058193207, "global_step": 35250, "epoch": 332, "lr": 2.5211621618744553e-05} +{"train_loss": 0.015363695099949837, "grad_norm": 0.1494770050048828, "global_step": 35251, "epoch": 332, "lr": 2.520904776358716e-05} +{"train_loss": 0.015328090637922287, "grad_norm": 0.15624800324440002, "global_step": 35252, "epoch": 332, "lr": 2.5206473995534373e-05} +{"train_loss": 0.014148080721497536, "grad_norm": 0.1292937695980072, "global_step": 35253, "epoch": 332, "lr": 2.520390031459528e-05} +{"train_loss": 0.014270614832639694, "grad_norm": 0.06117480620741844, "global_step": 35254, "epoch": 332, "lr": 2.520132672077887e-05} +{"train_loss": 0.014650198630988598, "grad_norm": 0.09127216041088104, "global_step": 35255, "epoch": 332, "lr": 2.5198753214094223e-05} +{"train_loss": 0.015771538019180298, "grad_norm": 0.10818348824977875, "global_step": 35256, "epoch": 332, "lr": 2.5196179794550383e-05} +{"train_loss": 0.01578388549387455, "grad_norm": 0.11510922014713287, "global_step": 35257, "epoch": 332, "lr": 2.5193606462156384e-05} +{"train_loss": 0.015093261376023293, "grad_norm": 0.12926225364208221, "global_step": 35258, "epoch": 332, "lr": 2.5191033216921245e-05} +{"train_loss": 0.015645015984773636, "grad_norm": 0.0572262667119503, "global_step": 35259, "epoch": 332, "lr": 2.5188460058854023e-05} +{"train_loss": 0.014405027963221073, "grad_norm": 0.0917537733912468, "global_step": 35260, "epoch": 332, "lr": 2.518588698796378e-05} +{"train_loss": 0.012816740199923515, "grad_norm": 0.09884136915206909, "global_step": 35261, "epoch": 332, "lr": 2.5183314004259527e-05} +{"train_loss": 0.0148391742259264, "grad_norm": 0.098564013838768, "global_step": 35262, "epoch": 332, "lr": 2.5180741107750305e-05} +{"train_loss": 0.01532808504998684, "grad_norm": 0.10366944223642349, "global_step": 35263, "epoch": 332, "lr": 2.5178168298445216e-05} +{"train_loss": 0.01518368162214756, "grad_norm": 0.06992622464895248, "global_step": 35264, "epoch": 332, "lr": 2.5175595576353204e-05} +{"train_loss": 0.01487656868994236, "grad_norm": 0.1576947569847107, "global_step": 35265, "epoch": 332, "lr": 2.517302294148335e-05} +{"train_loss": 0.01448160782456398, "grad_norm": 0.08102740347385406, "global_step": 35266, "epoch": 332, "lr": 2.5170450393844725e-05} +{"train_loss": 0.015676282346248627, "grad_norm": 0.06805793941020966, "global_step": 35267, "epoch": 332, "lr": 2.516787793344631e-05} +{"train_loss": 0.015133112668991089, "grad_norm": 0.10024868696928024, "global_step": 35268, "epoch": 332, "lr": 2.516530556029718e-05} +{"train_loss": 0.014964282512664795, "grad_norm": 0.06402985751628876, "global_step": 35269, "epoch": 332, "lr": 2.5162733274406382e-05} +{"train_loss": 0.013381068594753742, "grad_norm": 0.10655423998832703, "global_step": 35270, "epoch": 332, "lr": 2.5160161075782933e-05} +{"train_loss": 0.014332272112369537, "grad_norm": 0.0717991292476654, "global_step": 35271, "epoch": 332, "lr": 2.5157588964435863e-05} +{"train_loss": 0.015103530138731003, "grad_norm": 0.16996267437934875, "global_step": 35272, "epoch": 332, "lr": 2.5155016940374222e-05} +{"train_loss": 0.01632864587008953, "grad_norm": 0.10248937457799911, "global_step": 35273, "epoch": 332, "lr": 2.5152445003607062e-05} +{"train_loss": 0.013817356899380684, "grad_norm": 0.06743641942739487, "global_step": 35274, "epoch": 332, "lr": 2.514987315414339e-05} +{"train_loss": 0.014424645341932774, "grad_norm": 0.1489998996257782, "global_step": 35275, "epoch": 332, "lr": 2.5147301391992274e-05} +{"train_loss": 0.015946023166179657, "grad_norm": 0.12713301181793213, "global_step": 35276, "epoch": 332, "lr": 2.5144729717162708e-05} +{"train_loss": 0.013646906241774559, "grad_norm": 0.08883346617221832, "global_step": 35277, "epoch": 332, "lr": 2.5142158129663773e-05} +{"train_loss": 0.015586618334054947, "grad_norm": 0.09486003965139389, "global_step": 35278, "epoch": 332, "lr": 2.5139586629504475e-05} +{"train_loss": 0.015275124460458755, "grad_norm": 0.09181853383779526, "global_step": 35279, "epoch": 332, "lr": 2.5137015216693848e-05} +{"train_loss": 0.014191832393407822, "grad_norm": 0.06182975322008133, "global_step": 35280, "epoch": 332, "lr": 2.5134443891240954e-05} +{"train_loss": 0.01519574224948883, "grad_norm": 0.062091607600450516, "global_step": 35281, "epoch": 332, "lr": 2.513187265315482e-05} +{"train_loss": 0.017658177763223648, "grad_norm": 0.11116278916597366, "global_step": 35282, "epoch": 332, "lr": 2.5129301502444447e-05} +{"train_loss": 0.01246945746243, "grad_norm": 0.06238118186593056, "global_step": 35283, "epoch": 332, "lr": 2.5126730439118912e-05} +{"train_loss": 0.01387691032141447, "grad_norm": 0.11405684053897858, "global_step": 35284, "epoch": 332, "lr": 2.5124159463187208e-05} +{"train_loss": 0.01623888872563839, "grad_norm": 0.06168168783187866, "global_step": 35285, "epoch": 332, "lr": 2.5121588574658394e-05} +{"train_loss": 0.0151273338124156, "grad_norm": 0.15776920318603516, "global_step": 35286, "epoch": 332, "lr": 2.511901777354151e-05} +{"train_loss": 0.014667588286101818, "grad_norm": 0.1002897396683693, "global_step": 35287, "epoch": 332, "lr": 2.5116447059845582e-05} +{"train_loss": 0.01649947091937065, "grad_norm": 0.11245894432067871, "global_step": 35288, "epoch": 332, "lr": 2.511387643357962e-05} +{"train_loss": 0.015147524885833263, "grad_norm": 0.07327350229024887, "global_step": 35289, "epoch": 332, "lr": 2.5111305894752668e-05} +{"train_loss": 0.013693232089281082, "grad_norm": 0.14054767787456512, "global_step": 35290, "epoch": 332, "lr": 2.5108735443373787e-05} +{"train_loss": 0.015188894234597683, "grad_norm": 0.09280598908662796, "global_step": 35291, "epoch": 332, "lr": 2.5106165079451966e-05} +{"train_loss": 0.01786879077553749, "grad_norm": 0.09203944355249405, "global_step": 35292, "epoch": 332, "lr": 2.5103594802996255e-05} +{"train_loss": 0.014267044141888618, "grad_norm": 0.0667373389005661, "global_step": 35293, "epoch": 332, "lr": 2.51010246140157e-05} +{"train_loss": 0.015641437843441963, "grad_norm": 0.12633532285690308, "global_step": 35294, "epoch": 332, "lr": 2.509845451251932e-05} +{"train_loss": 0.01592109352350235, "grad_norm": 0.07744552940130234, "global_step": 35295, "epoch": 332, "lr": 2.509588449851612e-05} +{"train_loss": 0.01608285680413246, "grad_norm": 0.05931859090924263, "global_step": 35296, "epoch": 332, "lr": 2.5093314572015147e-05} +{"train_loss": 0.015156534957295319, "grad_norm": 0.14820076525211334, "global_step": 35297, "epoch": 332, "lr": 2.509074473302546e-05, "val_loss": 0.019531331956386566} +{"train_loss": 0.01363234594464302, "grad_norm": 0.10002709180116653, "global_step": 35298, "epoch": 333, "lr": 2.5088174981556044e-05} +{"train_loss": 0.015930648893117905, "grad_norm": 0.07701493054628372, "global_step": 35299, "epoch": 333, "lr": 2.508560531761597e-05} +{"train_loss": 0.01533999852836132, "grad_norm": 0.10109353065490723, "global_step": 35300, "epoch": 333, "lr": 2.508303574121422e-05} +{"train_loss": 0.015349451452493668, "grad_norm": 0.08971842378377914, "global_step": 35301, "epoch": 333, "lr": 2.508046625235987e-05} +{"train_loss": 0.015549607574939728, "grad_norm": 0.07742439955472946, "global_step": 35302, "epoch": 333, "lr": 2.50778968510619e-05} +{"train_loss": 0.012259788811206818, "grad_norm": 0.07931533455848694, "global_step": 35303, "epoch": 333, "lr": 2.5075327537329364e-05} +{"train_loss": 0.01599876396358013, "grad_norm": 0.07427239418029785, "global_step": 35304, "epoch": 333, "lr": 2.507275831117132e-05} +{"train_loss": 0.015525773167610168, "grad_norm": 0.08220182359218597, "global_step": 35305, "epoch": 333, "lr": 2.507018917259673e-05} +{"train_loss": 0.01492416299879551, "grad_norm": 0.06388109177350998, "global_step": 35306, "epoch": 333, "lr": 2.5067620121614655e-05} +{"train_loss": 0.016500409692525864, "grad_norm": 0.07639184594154358, "global_step": 35307, "epoch": 333, "lr": 2.506505115823413e-05} +{"train_loss": 0.013423731550574303, "grad_norm": 0.055906374007463455, "global_step": 35308, "epoch": 333, "lr": 2.506248228246416e-05} +{"train_loss": 0.01569250226020813, "grad_norm": 0.08745607733726501, "global_step": 35309, "epoch": 333, "lr": 2.5059913494313787e-05} +{"train_loss": 0.013686475344002247, "grad_norm": 0.17095869779586792, "global_step": 35310, "epoch": 333, "lr": 2.505734479379205e-05} +{"train_loss": 0.018009353429079056, "grad_norm": 0.0938660278916359, "global_step": 35311, "epoch": 333, "lr": 2.5054776180907945e-05} +{"train_loss": 0.015308192931115627, "grad_norm": 0.12497271597385406, "global_step": 35312, "epoch": 333, "lr": 2.50522076556705e-05} +{"train_loss": 0.01835479959845543, "grad_norm": 0.08783704042434692, "global_step": 35313, "epoch": 333, "lr": 2.504963921808874e-05} +{"train_loss": 0.01439560204744339, "grad_norm": 0.05879083648324013, "global_step": 35314, "epoch": 333, "lr": 2.504707086817172e-05} +{"train_loss": 0.016601398587226868, "grad_norm": 0.0990014523267746, "global_step": 35315, "epoch": 333, "lr": 2.5044502605928423e-05} +{"train_loss": 0.01615249738097191, "grad_norm": 0.06677067279815674, "global_step": 35316, "epoch": 333, "lr": 2.5041934431367908e-05} +{"train_loss": 0.014929440803825855, "grad_norm": 0.05751051753759384, "global_step": 35317, "epoch": 333, "lr": 2.5039366344499165e-05} +{"train_loss": 0.014367305673658848, "grad_norm": 0.06771790236234665, "global_step": 35318, "epoch": 333, "lr": 2.5036798345331248e-05} +{"train_loss": 0.015479735098779202, "grad_norm": 0.14150886237621307, "global_step": 35319, "epoch": 333, "lr": 2.5034230433873153e-05} +{"train_loss": 0.013517078012228012, "grad_norm": 0.08147507160902023, "global_step": 35320, "epoch": 333, "lr": 2.503166261013391e-05} +{"train_loss": 0.015350496396422386, "grad_norm": 0.10709147900342941, "global_step": 35321, "epoch": 333, "lr": 2.5029094874122566e-05} +{"train_loss": 0.015267780981957912, "grad_norm": 0.144758403301239, "global_step": 35322, "epoch": 333, "lr": 2.5026527225848127e-05} +{"train_loss": 0.014525139704346657, "grad_norm": 0.1710583120584488, "global_step": 35323, "epoch": 333, "lr": 2.5023959665319587e-05} +{"train_loss": 0.015253052115440369, "grad_norm": 0.2532506287097931, "global_step": 35324, "epoch": 333, "lr": 2.502139219254601e-05} +{"train_loss": 0.01622498780488968, "grad_norm": 0.06274482607841492, "global_step": 35325, "epoch": 333, "lr": 2.5018824807536378e-05} +{"train_loss": 0.015282846987247467, "grad_norm": 0.08294316381216049, "global_step": 35326, "epoch": 333, "lr": 2.5016257510299734e-05} +{"train_loss": 0.013214530423283577, "grad_norm": 0.12530122697353363, "global_step": 35327, "epoch": 333, "lr": 2.5013690300845117e-05} +{"train_loss": 0.017393935471773148, "grad_norm": 0.12157741189002991, "global_step": 35328, "epoch": 333, "lr": 2.5011123179181524e-05} +{"train_loss": 0.01485536154359579, "grad_norm": 0.08790165930986404, "global_step": 35329, "epoch": 333, "lr": 2.5008556145317958e-05} +{"train_loss": 0.016559351235628128, "grad_norm": 0.07325038313865662, "global_step": 35330, "epoch": 333, "lr": 2.5005989199263458e-05} +{"train_loss": 0.01524890586733818, "grad_norm": 0.09558567404747009, "global_step": 35331, "epoch": 333, "lr": 2.500342234102706e-05} +{"train_loss": 0.01605130359530449, "grad_norm": 0.09007564932107925, "global_step": 35332, "epoch": 333, "lr": 2.500085557061774e-05} +{"train_loss": 0.013421202078461647, "grad_norm": 0.07181227952241898, "global_step": 35333, "epoch": 333, "lr": 2.4998288888044553e-05} +{"train_loss": 0.015319358557462692, "grad_norm": 0.11118015646934509, "global_step": 35334, "epoch": 333, "lr": 2.499572229331652e-05} +{"train_loss": 0.014660852961242199, "grad_norm": 0.07044260203838348, "global_step": 35335, "epoch": 333, "lr": 2.499315578644264e-05} +{"train_loss": 0.01539282500743866, "grad_norm": 0.09871149063110352, "global_step": 35336, "epoch": 333, "lr": 2.4990589367431923e-05} +{"train_loss": 0.014259123243391514, "grad_norm": 0.09736036509275436, "global_step": 35337, "epoch": 333, "lr": 2.4988023036293395e-05} +{"train_loss": 0.014488618820905685, "grad_norm": 0.07880879938602448, "global_step": 35338, "epoch": 333, "lr": 2.49854567930361e-05} +{"train_loss": 0.016325367614626884, "grad_norm": 0.06491578370332718, "global_step": 35339, "epoch": 333, "lr": 2.498289063766901e-05} +{"train_loss": 0.013640889897942543, "grad_norm": 0.06436396390199661, "global_step": 35340, "epoch": 333, "lr": 2.498032457020118e-05} +{"train_loss": 0.018186990171670914, "grad_norm": 0.06642023473978043, "global_step": 35341, "epoch": 333, "lr": 2.497775859064161e-05} +{"train_loss": 0.014498073607683182, "grad_norm": 0.07413914799690247, "global_step": 35342, "epoch": 333, "lr": 2.4975192698999295e-05} +{"train_loss": 0.011884592473506927, "grad_norm": 0.05173766613006592, "global_step": 35343, "epoch": 333, "lr": 2.4972626895283268e-05} +{"train_loss": 0.01587417721748352, "grad_norm": 0.12513327598571777, "global_step": 35344, "epoch": 333, "lr": 2.497006117950255e-05} +{"train_loss": 0.01506704930216074, "grad_norm": 0.09295131266117096, "global_step": 35345, "epoch": 333, "lr": 2.496749555166617e-05} +{"train_loss": 0.015683181583881378, "grad_norm": 0.09576410800218582, "global_step": 35346, "epoch": 333, "lr": 2.4964930011783124e-05} +{"train_loss": 0.015604129992425442, "grad_norm": 0.06533364206552505, "global_step": 35347, "epoch": 333, "lr": 2.4962364559862407e-05} +{"train_loss": 0.017858948558568954, "grad_norm": 0.07571031153202057, "global_step": 35348, "epoch": 333, "lr": 2.495979919591307e-05} +{"train_loss": 0.014719469472765923, "grad_norm": 0.052198708057403564, "global_step": 35349, "epoch": 333, "lr": 2.495723391994409e-05} +{"train_loss": 0.0153312673792243, "grad_norm": 0.13044629991054535, "global_step": 35350, "epoch": 333, "lr": 2.4954668731964496e-05} +{"train_loss": 0.013036347925662994, "grad_norm": 0.06984194368124008, "global_step": 35351, "epoch": 333, "lr": 2.4952103631983333e-05} +{"train_loss": 0.013855811208486557, "grad_norm": 0.10218977928161621, "global_step": 35352, "epoch": 333, "lr": 2.4949538620009577e-05} +{"train_loss": 0.017948346212506294, "grad_norm": 0.0947793573141098, "global_step": 35353, "epoch": 333, "lr": 2.4946973696052234e-05} +{"train_loss": 0.014198660850524902, "grad_norm": 0.11986615508794785, "global_step": 35354, "epoch": 333, "lr": 2.4944408860120328e-05} +{"train_loss": 0.01673581823706627, "grad_norm": 0.09591874480247498, "global_step": 35355, "epoch": 333, "lr": 2.494184411222289e-05} +{"train_loss": 0.014782488346099854, "grad_norm": 0.0810331404209137, "global_step": 35356, "epoch": 333, "lr": 2.4939279452368895e-05} +{"train_loss": 0.011740298010408878, "grad_norm": 0.07092862576246262, "global_step": 35357, "epoch": 333, "lr": 2.493671488056739e-05} +{"train_loss": 0.01633794605731964, "grad_norm": 0.09788687527179718, "global_step": 35358, "epoch": 333, "lr": 2.4934150396827356e-05} +{"train_loss": 0.012197406031191349, "grad_norm": 0.056929729878902435, "global_step": 35359, "epoch": 333, "lr": 2.4931586001157835e-05} +{"train_loss": 0.015412182547152042, "grad_norm": 0.09585949033498764, "global_step": 35360, "epoch": 333, "lr": 2.49290216935678e-05} +{"train_loss": 0.014833645895123482, "grad_norm": 0.0708211213350296, "global_step": 35361, "epoch": 333, "lr": 2.4926457474066277e-05} +{"train_loss": 0.014043983072042465, "grad_norm": 0.07277673482894897, "global_step": 35362, "epoch": 333, "lr": 2.4923893342662298e-05} +{"train_loss": 0.014497002586722374, "grad_norm": 0.1067129522562027, "global_step": 35363, "epoch": 333, "lr": 2.4921329299364853e-05} +{"train_loss": 0.01583719439804554, "grad_norm": 0.07813192903995514, "global_step": 35364, "epoch": 333, "lr": 2.491876534418293e-05} +{"train_loss": 0.014456966891884804, "grad_norm": 0.13631372153759003, "global_step": 35365, "epoch": 333, "lr": 2.4916201477125578e-05} +{"train_loss": 0.012577388435602188, "grad_norm": 0.08263839036226273, "global_step": 35366, "epoch": 333, "lr": 2.491363769820177e-05} +{"train_loss": 0.016671473160386086, "grad_norm": 0.07183077186346054, "global_step": 35367, "epoch": 333, "lr": 2.4911074007420525e-05} +{"train_loss": 0.016065314412117004, "grad_norm": 0.058623071759939194, "global_step": 35368, "epoch": 333, "lr": 2.4908510404790873e-05} +{"train_loss": 0.014175169169902802, "grad_norm": 0.05456918850541115, "global_step": 35369, "epoch": 333, "lr": 2.4905946890321807e-05} +{"train_loss": 0.014982197433710098, "grad_norm": 0.1006392389535904, "global_step": 35370, "epoch": 333, "lr": 2.490338346402231e-05} +{"train_loss": 0.014504853636026382, "grad_norm": 0.07574205845594406, "global_step": 35371, "epoch": 333, "lr": 2.4900820125901407e-05} +{"train_loss": 0.014841698110103607, "grad_norm": 0.07122723758220673, "global_step": 35372, "epoch": 333, "lr": 2.489825687596813e-05} +{"train_loss": 0.01489216834306717, "grad_norm": 0.06470030546188354, "global_step": 35373, "epoch": 333, "lr": 2.4895693714231443e-05} +{"train_loss": 0.017295818775892258, "grad_norm": 0.06634387373924255, "global_step": 35374, "epoch": 333, "lr": 2.489313064070037e-05} +{"train_loss": 0.014063854701817036, "grad_norm": 0.14139363169670105, "global_step": 35375, "epoch": 333, "lr": 2.4890567655383935e-05} +{"train_loss": 0.016154088079929352, "grad_norm": 0.09129656851291656, "global_step": 35376, "epoch": 333, "lr": 2.4888004758291122e-05} +{"train_loss": 0.01357470452785492, "grad_norm": 0.08722373098134995, "global_step": 35377, "epoch": 333, "lr": 2.4885441949430926e-05} +{"train_loss": 0.013717956840991974, "grad_norm": 0.09062699973583221, "global_step": 35378, "epoch": 333, "lr": 2.4882879228812362e-05} +{"train_loss": 0.01312239933758974, "grad_norm": 0.08505360782146454, "global_step": 35379, "epoch": 333, "lr": 2.4880316596444457e-05} +{"train_loss": 0.015005083754658699, "grad_norm": 0.119693823158741, "global_step": 35380, "epoch": 333, "lr": 2.487775405233618e-05} +{"train_loss": 0.013927765190601349, "grad_norm": 0.08711764216423035, "global_step": 35381, "epoch": 333, "lr": 2.487519159649656e-05} +{"train_loss": 0.013846427202224731, "grad_norm": 0.08761677891016006, "global_step": 35382, "epoch": 333, "lr": 2.4872629228934597e-05} +{"train_loss": 0.015226867981255054, "grad_norm": 0.19535377621650696, "global_step": 35383, "epoch": 333, "lr": 2.4870066949659264e-05} +{"train_loss": 0.01611759513616562, "grad_norm": 0.09185085445642471, "global_step": 35384, "epoch": 333, "lr": 2.4867504758679594e-05} +{"train_loss": 0.016919411718845367, "grad_norm": 0.12430217117071152, "global_step": 35385, "epoch": 333, "lr": 2.4864942656004596e-05} +{"train_loss": 0.015732474625110626, "grad_norm": 0.105353444814682, "global_step": 35386, "epoch": 333, "lr": 2.486238064164324e-05} +{"train_loss": 0.014454299584031105, "grad_norm": 0.08004333078861237, "global_step": 35387, "epoch": 333, "lr": 2.4859818715604565e-05} +{"train_loss": 0.01684260368347168, "grad_norm": 0.06856255978345871, "global_step": 35388, "epoch": 333, "lr": 2.485725687789754e-05} +{"train_loss": 0.017809268087148666, "grad_norm": 0.06554827094078064, "global_step": 35389, "epoch": 333, "lr": 2.4854695128531192e-05} +{"train_loss": 0.015800192952156067, "grad_norm": 0.061237286776304245, "global_step": 35390, "epoch": 333, "lr": 2.4852133467514495e-05} +{"train_loss": 0.013542599976062775, "grad_norm": 0.05674590542912483, "global_step": 35391, "epoch": 333, "lr": 2.4849571894856462e-05} +{"train_loss": 0.015079113654792309, "grad_norm": 0.06564242392778397, "global_step": 35392, "epoch": 333, "lr": 2.4847010410566118e-05} +{"train_loss": 0.017174117267131805, "grad_norm": 0.0778946727514267, "global_step": 35393, "epoch": 333, "lr": 2.4844449014652436e-05} +{"train_loss": 0.014552758075296879, "grad_norm": 0.10383698344230652, "global_step": 35394, "epoch": 333, "lr": 2.4841887707124407e-05} +{"train_loss": 0.014149919152259827, "grad_norm": 0.08827876299619675, "global_step": 35395, "epoch": 333, "lr": 2.4839326487991038e-05} +{"train_loss": 0.013514978811144829, "grad_norm": 0.08067701011896133, "global_step": 35396, "epoch": 333, "lr": 2.4836765357261353e-05} +{"train_loss": 0.014098623767495155, "grad_norm": 0.08581755310297012, "global_step": 35397, "epoch": 333, "lr": 2.4834204314944313e-05} +{"train_loss": 0.018398236483335495, "grad_norm": 0.10107800364494324, "global_step": 35398, "epoch": 333, "lr": 2.483164336104894e-05} +{"train_loss": 0.01544472761452198, "grad_norm": 0.05336504802107811, "global_step": 35399, "epoch": 333, "lr": 2.4829082495584242e-05} +{"train_loss": 0.015778809785842896, "grad_norm": 0.09278261661529541, "global_step": 35400, "epoch": 333, "lr": 2.4826521718559202e-05} +{"train_loss": 0.014372729696333408, "grad_norm": 0.09542027115821838, "global_step": 35401, "epoch": 333, "lr": 2.4823961029982796e-05} +{"train_loss": 0.01787424460053444, "grad_norm": 0.05663345009088516, "global_step": 35402, "epoch": 333, "lr": 2.4821400429864043e-05} +{"train_loss": 0.015151387528639357, "grad_norm": 0.09016895294189453, "global_step": 35403, "epoch": 333, "lr": 2.4818839918211962e-05, "val_loss": 0.019578557461500168} +{"train_loss": 0.015258556231856346, "grad_norm": 0.06346897780895233, "global_step": 35404, "epoch": 334, "lr": 2.4816279495035522e-05} +{"train_loss": 0.017329836264252663, "grad_norm": 0.06682829558849335, "global_step": 35405, "epoch": 334, "lr": 2.4813719160343708e-05} +{"train_loss": 0.014400414191186428, "grad_norm": 0.07520991563796997, "global_step": 35406, "epoch": 334, "lr": 2.481115891414555e-05} +{"train_loss": 0.017088470980525017, "grad_norm": 0.08435452729463577, "global_step": 35407, "epoch": 334, "lr": 2.4808598756450003e-05} +{"train_loss": 0.01658436842262745, "grad_norm": 0.06355894356966019, "global_step": 35408, "epoch": 334, "lr": 2.4806038687266087e-05} +{"train_loss": 0.015569016337394714, "grad_norm": 0.06143699586391449, "global_step": 35409, "epoch": 334, "lr": 2.4803478706602813e-05} +{"train_loss": 0.013090607710182667, "grad_norm": 0.10482098907232285, "global_step": 35410, "epoch": 334, "lr": 2.4800918814469153e-05} +{"train_loss": 0.015064614824950695, "grad_norm": 0.07880549132823944, "global_step": 35411, "epoch": 334, "lr": 2.4798359010874095e-05} +{"train_loss": 0.0159250907599926, "grad_norm": 0.0778917446732521, "global_step": 35412, "epoch": 334, "lr": 2.4795799295826638e-05} +{"train_loss": 0.015174953266978264, "grad_norm": 0.130470409989357, "global_step": 35413, "epoch": 334, "lr": 2.47932396693358e-05} +{"train_loss": 0.016562368720769882, "grad_norm": 0.21074917912483215, "global_step": 35414, "epoch": 334, "lr": 2.4790680131410536e-05} +{"train_loss": 0.01575610414147377, "grad_norm": 0.07863311469554901, "global_step": 35415, "epoch": 334, "lr": 2.478812068205987e-05} +{"train_loss": 0.014623505063354969, "grad_norm": 0.14261981844902039, "global_step": 35416, "epoch": 334, "lr": 2.478556132129279e-05} +{"train_loss": 0.014941358007490635, "grad_norm": 0.11063016951084137, "global_step": 35417, "epoch": 334, "lr": 2.4783002049118287e-05} +{"train_loss": 0.013079334050416946, "grad_norm": 0.10702753812074661, "global_step": 35418, "epoch": 334, "lr": 2.4780442865545322e-05} +{"train_loss": 0.014600086957216263, "grad_norm": 0.05224986374378204, "global_step": 35419, "epoch": 334, "lr": 2.477788377058292e-05} +{"train_loss": 0.016622796654701233, "grad_norm": 0.11512798815965652, "global_step": 35420, "epoch": 334, "lr": 2.4775324764240087e-05} +{"train_loss": 0.01395552046597004, "grad_norm": 0.09480152279138565, "global_step": 35421, "epoch": 334, "lr": 2.477276584652577e-05} +{"train_loss": 0.012668253853917122, "grad_norm": 0.07186608016490936, "global_step": 35422, "epoch": 334, "lr": 2.4770207017449e-05} +{"train_loss": 0.013723453506827354, "grad_norm": 0.06660769134759903, "global_step": 35423, "epoch": 334, "lr": 2.4767648277018752e-05} +{"train_loss": 0.013917931355535984, "grad_norm": 0.1348661333322525, "global_step": 35424, "epoch": 334, "lr": 2.4765089625243998e-05} +{"train_loss": 0.01592770218849182, "grad_norm": 0.07174547761678696, "global_step": 35425, "epoch": 334, "lr": 2.4762531062133743e-05} +{"train_loss": 0.01745218224823475, "grad_norm": 0.0773748904466629, "global_step": 35426, "epoch": 334, "lr": 2.4759972587696995e-05} +{"train_loss": 0.013774395920336246, "grad_norm": 0.10273879766464233, "global_step": 35427, "epoch": 334, "lr": 2.4757414201942713e-05} +{"train_loss": 0.014443494379520416, "grad_norm": 0.09812051057815552, "global_step": 35428, "epoch": 334, "lr": 2.4754855904879913e-05} +{"train_loss": 0.013249379582703114, "grad_norm": 0.06917176395654678, "global_step": 35429, "epoch": 334, "lr": 2.4752297696517556e-05} +{"train_loss": 0.014211980625987053, "grad_norm": 0.047443244606256485, "global_step": 35430, "epoch": 334, "lr": 2.474973957686466e-05} +{"train_loss": 0.015849413350224495, "grad_norm": 0.06692453473806381, "global_step": 35431, "epoch": 334, "lr": 2.474718154593018e-05} +{"train_loss": 0.01672440581023693, "grad_norm": 0.13441075384616852, "global_step": 35432, "epoch": 334, "lr": 2.4744623603723126e-05} +{"train_loss": 0.01608400046825409, "grad_norm": 0.11903964728116989, "global_step": 35433, "epoch": 334, "lr": 2.4742065750252497e-05} +{"train_loss": 0.015951385721564293, "grad_norm": 0.06524517387151718, "global_step": 35434, "epoch": 334, "lr": 2.4739507985527268e-05} +{"train_loss": 0.012943117879331112, "grad_norm": 0.06209230050444603, "global_step": 35435, "epoch": 334, "lr": 2.47369503095564e-05} +{"train_loss": 0.014721470884978771, "grad_norm": 0.11893972754478455, "global_step": 35436, "epoch": 334, "lr": 2.4734392722348904e-05} +{"train_loss": 0.014457069337368011, "grad_norm": 0.07071322202682495, "global_step": 35437, "epoch": 334, "lr": 2.4731835223913778e-05} +{"train_loss": 0.01480276882648468, "grad_norm": 0.06657439470291138, "global_step": 35438, "epoch": 334, "lr": 2.4729277814259983e-05} +{"train_loss": 0.01577472686767578, "grad_norm": 0.06638657301664352, "global_step": 35439, "epoch": 334, "lr": 2.4726720493396512e-05} +{"train_loss": 0.017025090754032135, "grad_norm": 0.08901514858007431, "global_step": 35440, "epoch": 334, "lr": 2.472416326133239e-05} +{"train_loss": 0.016549989581108093, "grad_norm": 0.18957817554473877, "global_step": 35441, "epoch": 334, "lr": 2.472160611807653e-05} +{"train_loss": 0.015364756807684898, "grad_norm": 0.06681949645280838, "global_step": 35442, "epoch": 334, "lr": 2.471904906363795e-05} +{"train_loss": 0.013870293274521828, "grad_norm": 0.05999679118394852, "global_step": 35443, "epoch": 334, "lr": 2.471649209802566e-05} +{"train_loss": 0.01423383504152298, "grad_norm": 0.05943235009908676, "global_step": 35444, "epoch": 334, "lr": 2.47139352212486e-05} +{"train_loss": 0.013945896178483963, "grad_norm": 0.10229511559009552, "global_step": 35445, "epoch": 334, "lr": 2.471137843331578e-05} +{"train_loss": 0.015540733933448792, "grad_norm": 0.07336732745170593, "global_step": 35446, "epoch": 334, "lr": 2.4708821734236193e-05} +{"train_loss": 0.018222596496343613, "grad_norm": 0.16028304398059845, "global_step": 35447, "epoch": 334, "lr": 2.470626512401881e-05} +{"train_loss": 0.015469429083168507, "grad_norm": 0.06271431595087051, "global_step": 35448, "epoch": 334, "lr": 2.4703708602672586e-05} +{"train_loss": 0.015834780409932137, "grad_norm": 0.12844215333461761, "global_step": 35449, "epoch": 334, "lr": 2.4701152170206536e-05} +{"train_loss": 0.013259490951895714, "grad_norm": 0.06338468939065933, "global_step": 35450, "epoch": 334, "lr": 2.469859582662965e-05} +{"train_loss": 0.014262959361076355, "grad_norm": 0.06045487895607948, "global_step": 35451, "epoch": 334, "lr": 2.4696039571950895e-05} +{"train_loss": 0.014559539034962654, "grad_norm": 0.176856130361557, "global_step": 35452, "epoch": 334, "lr": 2.469348340617924e-05} +{"train_loss": 0.01638275757431984, "grad_norm": 0.141059011220932, "global_step": 35453, "epoch": 334, "lr": 2.469092732932367e-05} +{"train_loss": 0.011576049029827118, "grad_norm": 0.0642627403140068, "global_step": 35454, "epoch": 334, "lr": 2.46883713413932e-05} +{"train_loss": 0.01626626029610634, "grad_norm": 0.06133823096752167, "global_step": 35455, "epoch": 334, "lr": 2.4685815442396765e-05} +{"train_loss": 0.016365472227334976, "grad_norm": 0.07632932811975479, "global_step": 35456, "epoch": 334, "lr": 2.4683259632343362e-05} +{"train_loss": 0.018477613106369972, "grad_norm": 0.0843103677034378, "global_step": 35457, "epoch": 334, "lr": 2.4680703911242e-05} +{"train_loss": 0.015205373987555504, "grad_norm": 0.060784995555877686, "global_step": 35458, "epoch": 334, "lr": 2.467814827910163e-05} +{"train_loss": 0.01562931388616562, "grad_norm": 0.10062749683856964, "global_step": 35459, "epoch": 334, "lr": 2.4675592735931213e-05} +{"train_loss": 0.012713585048913956, "grad_norm": 0.09114135801792145, "global_step": 35460, "epoch": 334, "lr": 2.4673037281739757e-05} +{"train_loss": 0.019051052629947662, "grad_norm": 0.14031627774238586, "global_step": 35461, "epoch": 334, "lr": 2.4670481916536248e-05} +{"train_loss": 0.014759626239538193, "grad_norm": 0.08575253188610077, "global_step": 35462, "epoch": 334, "lr": 2.466792664032963e-05} +{"train_loss": 0.015719804912805557, "grad_norm": 0.06249360367655754, "global_step": 35463, "epoch": 334, "lr": 2.466537145312892e-05} +{"train_loss": 0.013824116438627243, "grad_norm": 0.07602059841156006, "global_step": 35464, "epoch": 334, "lr": 2.466281635494308e-05} +{"train_loss": 0.015858041122555733, "grad_norm": 0.15154637396335602, "global_step": 35465, "epoch": 334, "lr": 2.4660261345781067e-05} +{"train_loss": 0.014726679772138596, "grad_norm": 0.07177554816007614, "global_step": 35466, "epoch": 334, "lr": 2.4657706425651867e-05} +{"train_loss": 0.014468482695519924, "grad_norm": 0.08950844407081604, "global_step": 35467, "epoch": 334, "lr": 2.465515159456449e-05} +{"train_loss": 0.013688620179891586, "grad_norm": 0.08185254782438278, "global_step": 35468, "epoch": 334, "lr": 2.465259685252787e-05} +{"train_loss": 0.015768248587846756, "grad_norm": 0.11633869260549545, "global_step": 35469, "epoch": 334, "lr": 2.4650042199551023e-05} +{"train_loss": 0.016452161595225334, "grad_norm": 0.0804324671626091, "global_step": 35470, "epoch": 334, "lr": 2.464748763564288e-05} +{"train_loss": 0.015237046405673027, "grad_norm": 0.057808198034763336, "global_step": 35471, "epoch": 334, "lr": 2.464493316081246e-05} +{"train_loss": 0.01627177558839321, "grad_norm": 0.05098375305533409, "global_step": 35472, "epoch": 334, "lr": 2.46423787750687e-05} +{"train_loss": 0.017350442707538605, "grad_norm": 0.05780085548758507, "global_step": 35473, "epoch": 334, "lr": 2.4639824478420593e-05} +{"train_loss": 0.013487081974744797, "grad_norm": 0.049865107983350754, "global_step": 35474, "epoch": 334, "lr": 2.463727027087713e-05} +{"train_loss": 0.014968054369091988, "grad_norm": 0.14123758673667908, "global_step": 35475, "epoch": 334, "lr": 2.463471615244727e-05} +{"train_loss": 0.014012182131409645, "grad_norm": 0.07313194870948792, "global_step": 35476, "epoch": 334, "lr": 2.4632162123139966e-05} +{"train_loss": 0.016916969791054726, "grad_norm": 0.06872659921646118, "global_step": 35477, "epoch": 334, "lr": 2.462960818296422e-05} +{"train_loss": 0.015265942551195621, "grad_norm": 0.06981544196605682, "global_step": 35478, "epoch": 334, "lr": 2.462705433192901e-05} +{"train_loss": 0.015087893232703209, "grad_norm": 0.07153958082199097, "global_step": 35479, "epoch": 334, "lr": 2.4624500570043278e-05} +{"train_loss": 0.014335919171571732, "grad_norm": 0.09121715277433395, "global_step": 35480, "epoch": 334, "lr": 2.4621946897316016e-05} +{"train_loss": 0.011053205467760563, "grad_norm": 0.07240677624940872, "global_step": 35481, "epoch": 334, "lr": 2.4619393313756238e-05} +{"train_loss": 0.015044305473566055, "grad_norm": 0.07368456572294235, "global_step": 35482, "epoch": 334, "lr": 2.4616839819372833e-05} +{"train_loss": 0.013664761558175087, "grad_norm": 0.09994108229875565, "global_step": 35483, "epoch": 334, "lr": 2.4614286414174816e-05} +{"train_loss": 0.013861162588000298, "grad_norm": 0.08755926787853241, "global_step": 35484, "epoch": 334, "lr": 2.461173309817117e-05} +{"train_loss": 0.014178631827235222, "grad_norm": 0.07808355242013931, "global_step": 35485, "epoch": 334, "lr": 2.4609179871370837e-05} +{"train_loss": 0.0148314805701375, "grad_norm": 0.07234091311693192, "global_step": 35486, "epoch": 334, "lr": 2.4606626733782806e-05} +{"train_loss": 0.017533695325255394, "grad_norm": 0.05849751830101013, "global_step": 35487, "epoch": 334, "lr": 2.4604073685416062e-05} +{"train_loss": 0.014766630716621876, "grad_norm": 0.10617624968290329, "global_step": 35488, "epoch": 334, "lr": 2.4601520726279557e-05} +{"train_loss": 0.016604363918304443, "grad_norm": 0.08161768317222595, "global_step": 35489, "epoch": 334, "lr": 2.459896785638225e-05} +{"train_loss": 0.017182735726237297, "grad_norm": 0.06632331758737564, "global_step": 35490, "epoch": 334, "lr": 2.4596415075733116e-05} +{"train_loss": 0.017133081331849098, "grad_norm": 0.07481693476438522, "global_step": 35491, "epoch": 334, "lr": 2.4593862384341154e-05} +{"train_loss": 0.012222573161125183, "grad_norm": 0.1019781082868576, "global_step": 35492, "epoch": 334, "lr": 2.4591309782215287e-05} +{"train_loss": 0.015073842369019985, "grad_norm": 0.12326764315366745, "global_step": 35493, "epoch": 334, "lr": 2.4588757269364537e-05} +{"train_loss": 0.014407455921173096, "grad_norm": 0.08893787115812302, "global_step": 35494, "epoch": 334, "lr": 2.4586204845797817e-05} +{"train_loss": 0.012849103659391403, "grad_norm": 0.09748052805662155, "global_step": 35495, "epoch": 334, "lr": 2.458365251152414e-05} +{"train_loss": 0.015058389864861965, "grad_norm": 0.07072758674621582, "global_step": 35496, "epoch": 334, "lr": 2.4581100266552436e-05} +{"train_loss": 0.015093499794602394, "grad_norm": 0.13086272776126862, "global_step": 35497, "epoch": 334, "lr": 2.457854811089169e-05} +{"train_loss": 0.017298322170972824, "grad_norm": 0.07623933255672455, "global_step": 35498, "epoch": 334, "lr": 2.4575996044550893e-05} +{"train_loss": 0.015885327011346817, "grad_norm": 0.05830267816781998, "global_step": 35499, "epoch": 334, "lr": 2.4573444067538986e-05} +{"train_loss": 0.016140662133693695, "grad_norm": 0.07788573950529099, "global_step": 35500, "epoch": 334, "lr": 2.4570892179864922e-05} +{"train_loss": 0.016604231670498848, "grad_norm": 0.09042138606309891, "global_step": 35501, "epoch": 334, "lr": 2.4568340381537686e-05} +{"train_loss": 0.01684538647532463, "grad_norm": 0.11116975545883179, "global_step": 35502, "epoch": 334, "lr": 2.456578867256626e-05} +{"train_loss": 0.016148213297128677, "grad_norm": 0.11246426403522491, "global_step": 35503, "epoch": 334, "lr": 2.456323705295957e-05} +{"train_loss": 0.012664170935750008, "grad_norm": 0.05952451378107071, "global_step": 35504, "epoch": 334, "lr": 2.4560685522726623e-05} +{"train_loss": 0.015709606930613518, "grad_norm": 0.08304223418235779, "global_step": 35505, "epoch": 334, "lr": 2.4558134081876367e-05} +{"train_loss": 0.015259422361850739, "grad_norm": 0.10939271748065948, "global_step": 35506, "epoch": 334, "lr": 2.4555582730417737e-05} +{"train_loss": 0.013554735109210014, "grad_norm": 0.1135319247841835, "global_step": 35507, "epoch": 334, "lr": 2.4553031468359728e-05} +{"train_loss": 0.014945263043045998, "grad_norm": 0.06952162832021713, "global_step": 35508, "epoch": 334, "lr": 2.455048029571132e-05} +{"train_loss": 0.01513972080681684, "grad_norm": 0.1170208677649498, "global_step": 35509, "epoch": 334, "lr": 2.4547929212481435e-05, "val_loss": 0.01947789452970028} +{"train_loss": 0.014833294786512852, "grad_norm": 0.10445614904165268, "global_step": 35510, "epoch": 335, "lr": 2.4545378218679076e-05} +{"train_loss": 0.014060946181416512, "grad_norm": 0.06834469735622406, "global_step": 35511, "epoch": 335, "lr": 2.4542827314313177e-05} +{"train_loss": 0.014189455658197403, "grad_norm": 0.09260895103216171, "global_step": 35512, "epoch": 335, "lr": 2.4540276499392723e-05} +{"train_loss": 0.013100512325763702, "grad_norm": 0.06201399862766266, "global_step": 35513, "epoch": 335, "lr": 2.453772577392665e-05} +{"train_loss": 0.01559531595557928, "grad_norm": 0.07258348166942596, "global_step": 35514, "epoch": 335, "lr": 2.4535175137923933e-05} +{"train_loss": 0.014621972106397152, "grad_norm": 0.05517391487956047, "global_step": 35515, "epoch": 335, "lr": 2.4532624591393554e-05} +{"train_loss": 0.01608995720744133, "grad_norm": 0.07750184834003448, "global_step": 35516, "epoch": 335, "lr": 2.453007413434446e-05} +{"train_loss": 0.014681019820272923, "grad_norm": 0.061318978667259216, "global_step": 35517, "epoch": 335, "lr": 2.4527523766785586e-05} +{"train_loss": 0.017277339473366737, "grad_norm": 0.10376709699630737, "global_step": 35518, "epoch": 335, "lr": 2.452497348872592e-05} +{"train_loss": 0.015102727338671684, "grad_norm": 0.07726860791444778, "global_step": 35519, "epoch": 335, "lr": 2.4522423300174437e-05} +{"train_loss": 0.014938807114958763, "grad_norm": 0.05963888391852379, "global_step": 35520, "epoch": 335, "lr": 2.451987320114006e-05} +{"train_loss": 0.01523741614073515, "grad_norm": 0.06876888126134872, "global_step": 35521, "epoch": 335, "lr": 2.4517323191631764e-05} +{"train_loss": 0.015586141496896744, "grad_norm": 0.0807604119181633, "global_step": 35522, "epoch": 335, "lr": 2.4514773271658554e-05} +{"train_loss": 0.014008460566401482, "grad_norm": 0.07940999418497086, "global_step": 35523, "epoch": 335, "lr": 2.4512223441229303e-05} +{"train_loss": 0.015104951336979866, "grad_norm": 0.07479899376630783, "global_step": 35524, "epoch": 335, "lr": 2.450967370035302e-05} +{"train_loss": 0.015862351283431053, "grad_norm": 0.06520446389913559, "global_step": 35525, "epoch": 335, "lr": 2.4507124049038678e-05} +{"train_loss": 0.01447216048836708, "grad_norm": 0.1255856156349182, "global_step": 35526, "epoch": 335, "lr": 2.4504574487295196e-05} +{"train_loss": 0.016265152022242546, "grad_norm": 0.06131483614444733, "global_step": 35527, "epoch": 335, "lr": 2.4502025015131548e-05} +{"train_loss": 0.012348235584795475, "grad_norm": 0.06364220380783081, "global_step": 35528, "epoch": 335, "lr": 2.4499475632556717e-05} +{"train_loss": 0.01664094068109989, "grad_norm": 0.13259606063365936, "global_step": 35529, "epoch": 335, "lr": 2.449692633957964e-05} +{"train_loss": 0.013698760420084, "grad_norm": 0.08821834623813629, "global_step": 35530, "epoch": 335, "lr": 2.449437713620925e-05} +{"train_loss": 0.013951933942735195, "grad_norm": 0.09593171626329422, "global_step": 35531, "epoch": 335, "lr": 2.4491828022454533e-05} +{"train_loss": 0.013972275890409946, "grad_norm": 0.05259525403380394, "global_step": 35532, "epoch": 335, "lr": 2.448927899832445e-05} +{"train_loss": 0.01595049723982811, "grad_norm": 0.07830037921667099, "global_step": 35533, "epoch": 335, "lr": 2.4486730063827933e-05} +{"train_loss": 0.014090897515416145, "grad_norm": 0.05915474146604538, "global_step": 35534, "epoch": 335, "lr": 2.4484181218973968e-05} +{"train_loss": 0.01628945767879486, "grad_norm": 0.08173949271440506, "global_step": 35535, "epoch": 335, "lr": 2.4481632463771474e-05} +{"train_loss": 0.017859164625406265, "grad_norm": 0.2681857645511627, "global_step": 35536, "epoch": 335, "lr": 2.4479083798229448e-05} +{"train_loss": 0.01565401442348957, "grad_norm": 0.10434583574533463, "global_step": 35537, "epoch": 335, "lr": 2.44765352223568e-05} +{"train_loss": 0.01517394371330738, "grad_norm": 0.09408345073461533, "global_step": 35538, "epoch": 335, "lr": 2.447398673616251e-05} +{"train_loss": 0.016227610409259796, "grad_norm": 0.05820709839463234, "global_step": 35539, "epoch": 335, "lr": 2.447143833965555e-05} +{"train_loss": 0.012963933870196342, "grad_norm": 0.09384860098361969, "global_step": 35540, "epoch": 335, "lr": 2.446889003284485e-05} +{"train_loss": 0.017017453908920288, "grad_norm": 0.08170780539512634, "global_step": 35541, "epoch": 335, "lr": 2.4466341815739348e-05} +{"train_loss": 0.015816230326890945, "grad_norm": 0.07523635774850845, "global_step": 35542, "epoch": 335, "lr": 2.4463793688348035e-05} +{"train_loss": 0.016582991927862167, "grad_norm": 0.07545726001262665, "global_step": 35543, "epoch": 335, "lr": 2.4461245650679833e-05} +{"train_loss": 0.017172273248434067, "grad_norm": 0.0770857036113739, "global_step": 35544, "epoch": 335, "lr": 2.44586977027437e-05} +{"train_loss": 0.017638497054576874, "grad_norm": 0.11648789793252945, "global_step": 35545, "epoch": 335, "lr": 2.4456149844548622e-05} +{"train_loss": 0.016420669853687286, "grad_norm": 0.06669845432043076, "global_step": 35546, "epoch": 335, "lr": 2.4453602076103516e-05} +{"train_loss": 0.015386896207928658, "grad_norm": 0.14280478656291962, "global_step": 35547, "epoch": 335, "lr": 2.4451054397417326e-05} +{"train_loss": 0.01792166382074356, "grad_norm": 0.07204518467187881, "global_step": 35548, "epoch": 335, "lr": 2.4448506808499028e-05} +{"train_loss": 0.016793664544820786, "grad_norm": 0.09626191109418869, "global_step": 35549, "epoch": 335, "lr": 2.4445959309357575e-05} +{"train_loss": 0.016603698953986168, "grad_norm": 0.10907024890184402, "global_step": 35550, "epoch": 335, "lr": 2.4443411900001896e-05} +{"train_loss": 0.01460191234946251, "grad_norm": 0.08643481880426407, "global_step": 35551, "epoch": 335, "lr": 2.444086458044095e-05} +{"train_loss": 0.01659097895026207, "grad_norm": 0.09856447577476501, "global_step": 35552, "epoch": 335, "lr": 2.4438317350683705e-05} +{"train_loss": 0.017736833542585373, "grad_norm": 0.1141619011759758, "global_step": 35553, "epoch": 335, "lr": 2.4435770210739102e-05} +{"train_loss": 0.01310875453054905, "grad_norm": 0.0711098313331604, "global_step": 35554, "epoch": 335, "lr": 2.443322316061606e-05} +{"train_loss": 0.014611741527915001, "grad_norm": 0.06730857491493225, "global_step": 35555, "epoch": 335, "lr": 2.4430676200323554e-05} +{"train_loss": 0.015694886445999146, "grad_norm": 0.1328933835029602, "global_step": 35556, "epoch": 335, "lr": 2.4428129329870558e-05} +{"train_loss": 0.015332009643316269, "grad_norm": 0.06689453125, "global_step": 35557, "epoch": 335, "lr": 2.4425582549265992e-05} +{"train_loss": 0.015713432803750038, "grad_norm": 0.06631425768136978, "global_step": 35558, "epoch": 335, "lr": 2.4423035858518784e-05} +{"train_loss": 0.016845783218741417, "grad_norm": 0.09952075779438019, "global_step": 35559, "epoch": 335, "lr": 2.4420489257637906e-05} +{"train_loss": 0.013866149820387363, "grad_norm": 0.09458927065134048, "global_step": 35560, "epoch": 335, "lr": 2.4417942746632326e-05} +{"train_loss": 0.013660535216331482, "grad_norm": 0.12072161585092545, "global_step": 35561, "epoch": 335, "lr": 2.441539632551095e-05} +{"train_loss": 0.01284861285239458, "grad_norm": 0.05541973561048508, "global_step": 35562, "epoch": 335, "lr": 2.4412849994282742e-05} +{"train_loss": 0.01513512060046196, "grad_norm": 0.05865706130862236, "global_step": 35563, "epoch": 335, "lr": 2.441030375295669e-05} +{"train_loss": 0.013611621223390102, "grad_norm": 0.06438256055116653, "global_step": 35564, "epoch": 335, "lr": 2.440775760154167e-05} +{"train_loss": 0.014225313439965248, "grad_norm": 0.10397428274154663, "global_step": 35565, "epoch": 335, "lr": 2.4405211540046657e-05} +{"train_loss": 0.016129540279507637, "grad_norm": 0.16590280830860138, "global_step": 35566, "epoch": 335, "lr": 2.4402665568480616e-05} +{"train_loss": 0.015019115060567856, "grad_norm": 0.06420749425888062, "global_step": 35567, "epoch": 335, "lr": 2.4400119686852457e-05} +{"train_loss": 0.014553576707839966, "grad_norm": 0.06289280951023102, "global_step": 35568, "epoch": 335, "lr": 2.439757389517115e-05} +{"train_loss": 0.018480535596609116, "grad_norm": 0.06846929341554642, "global_step": 35569, "epoch": 335, "lr": 2.4395028193445645e-05} +{"train_loss": 0.01447215210646391, "grad_norm": 0.05680660530924797, "global_step": 35570, "epoch": 335, "lr": 2.4392482581684882e-05} +{"train_loss": 0.015389054082334042, "grad_norm": 0.10105689615011215, "global_step": 35571, "epoch": 335, "lr": 2.438993705989778e-05} +{"train_loss": 0.013973291032016277, "grad_norm": 0.07614048570394516, "global_step": 35572, "epoch": 335, "lr": 2.43873916280933e-05} +{"train_loss": 0.012645150534808636, "grad_norm": 0.07348718494176865, "global_step": 35573, "epoch": 335, "lr": 2.4384846286280405e-05} +{"train_loss": 0.013237202540040016, "grad_norm": 0.08614356070756912, "global_step": 35574, "epoch": 335, "lr": 2.4382301034468e-05} +{"train_loss": 0.013874372467398643, "grad_norm": 0.08604414761066437, "global_step": 35575, "epoch": 335, "lr": 2.4379755872665072e-05} +{"train_loss": 0.016762077808380127, "grad_norm": 0.0669143795967102, "global_step": 35576, "epoch": 335, "lr": 2.437721080088052e-05} +{"train_loss": 0.017123371362686157, "grad_norm": 0.13283593952655792, "global_step": 35577, "epoch": 335, "lr": 2.437466581912332e-05} +{"train_loss": 0.013696543872356415, "grad_norm": 0.15043289959430695, "global_step": 35578, "epoch": 335, "lr": 2.4372120927402386e-05} +{"train_loss": 0.014300422742962837, "grad_norm": 0.05357932671904564, "global_step": 35579, "epoch": 335, "lr": 2.436957612572667e-05} +{"train_loss": 0.01391133014112711, "grad_norm": 0.05929629132151604, "global_step": 35580, "epoch": 335, "lr": 2.4367031414105142e-05} +{"train_loss": 0.017914749681949615, "grad_norm": 0.06461911648511887, "global_step": 35581, "epoch": 335, "lr": 2.4364486792546715e-05} +{"train_loss": 0.0140535281971097, "grad_norm": 0.06318093836307526, "global_step": 35582, "epoch": 335, "lr": 2.436194226106031e-05} +{"train_loss": 0.015438344329595566, "grad_norm": 0.05821068584918976, "global_step": 35583, "epoch": 335, "lr": 2.4359397819654912e-05} +{"train_loss": 0.015917537733912468, "grad_norm": 0.09404369443655014, "global_step": 35584, "epoch": 335, "lr": 2.4356853468339423e-05} +{"train_loss": 0.01423738431185484, "grad_norm": 0.07347923517227173, "global_step": 35585, "epoch": 335, "lr": 2.4354309207122794e-05} +{"train_loss": 0.012979861348867416, "grad_norm": 0.05989326164126396, "global_step": 35586, "epoch": 335, "lr": 2.4351765036013995e-05} +{"train_loss": 0.013265322893857956, "grad_norm": 0.07714881747961044, "global_step": 35587, "epoch": 335, "lr": 2.4349220955021935e-05} +{"train_loss": 0.01560170017182827, "grad_norm": 0.05906479060649872, "global_step": 35588, "epoch": 335, "lr": 2.434667696415554e-05} +{"train_loss": 0.016505658626556396, "grad_norm": 0.10859549045562744, "global_step": 35589, "epoch": 335, "lr": 2.4344133063423764e-05} +{"train_loss": 0.01567212864756584, "grad_norm": 0.10455793142318726, "global_step": 35590, "epoch": 335, "lr": 2.4341589252835568e-05} +{"train_loss": 0.013538504019379616, "grad_norm": 0.10162746906280518, "global_step": 35591, "epoch": 335, "lr": 2.4339045532399847e-05} +{"train_loss": 0.015271516516804695, "grad_norm": 0.05742825195193291, "global_step": 35592, "epoch": 335, "lr": 2.4336501902125565e-05} +{"train_loss": 0.012734919786453247, "grad_norm": 0.05496496334671974, "global_step": 35593, "epoch": 335, "lr": 2.4333958362021668e-05} +{"train_loss": 0.015304025262594223, "grad_norm": 0.08411245048046112, "global_step": 35594, "epoch": 335, "lr": 2.4331414912097082e-05} +{"train_loss": 0.014347699470818043, "grad_norm": 0.08472462743520737, "global_step": 35595, "epoch": 335, "lr": 2.4328871552360723e-05} +{"train_loss": 0.01756683737039566, "grad_norm": 0.07577763497829437, "global_step": 35596, "epoch": 335, "lr": 2.432632828282154e-05} +{"train_loss": 0.01522686704993248, "grad_norm": 0.07661347836256027, "global_step": 35597, "epoch": 335, "lr": 2.43237851034885e-05} +{"train_loss": 0.013526402413845062, "grad_norm": 0.05950304493308067, "global_step": 35598, "epoch": 335, "lr": 2.432124201437049e-05} +{"train_loss": 0.014367656782269478, "grad_norm": 0.05945170670747757, "global_step": 35599, "epoch": 335, "lr": 2.4318699015476493e-05} +{"train_loss": 0.015448510646820068, "grad_norm": 0.06765341758728027, "global_step": 35600, "epoch": 335, "lr": 2.4316156106815414e-05} +{"train_loss": 0.0152352936565876, "grad_norm": 0.07904588431119919, "global_step": 35601, "epoch": 335, "lr": 2.4313613288396175e-05} +{"train_loss": 0.01402975432574749, "grad_norm": 0.05464527755975723, "global_step": 35602, "epoch": 335, "lr": 2.4311070560227726e-05} +{"train_loss": 0.013516382314264774, "grad_norm": 0.09007678180932999, "global_step": 35603, "epoch": 335, "lr": 2.4308527922319008e-05} +{"train_loss": 0.015192540362477303, "grad_norm": 0.06733348220586777, "global_step": 35604, "epoch": 335, "lr": 2.4305985374678986e-05} +{"train_loss": 0.016593987122178078, "grad_norm": 0.07716768234968185, "global_step": 35605, "epoch": 335, "lr": 2.4303442917316516e-05} +{"train_loss": 0.015694959089159966, "grad_norm": 0.06702530384063721, "global_step": 35606, "epoch": 335, "lr": 2.4300900550240575e-05} +{"train_loss": 0.01633201539516449, "grad_norm": 0.09230683743953705, "global_step": 35607, "epoch": 335, "lr": 2.4298358273460114e-05} +{"train_loss": 0.015255434438586235, "grad_norm": 0.08735708147287369, "global_step": 35608, "epoch": 335, "lr": 2.4295816086984025e-05} +{"train_loss": 0.014972318895161152, "grad_norm": 0.06180824339389801, "global_step": 35609, "epoch": 335, "lr": 2.429327399082126e-05} +{"train_loss": 0.013087416999042034, "grad_norm": 0.06306097656488419, "global_step": 35610, "epoch": 335, "lr": 2.4290731984980763e-05} +{"train_loss": 0.014819925650954247, "grad_norm": 0.07519737631082535, "global_step": 35611, "epoch": 335, "lr": 2.4288190069471462e-05} +{"train_loss": 0.01590418443083763, "grad_norm": 0.1168287917971611, "global_step": 35612, "epoch": 335, "lr": 2.4285648244302257e-05} +{"train_loss": 0.014385339803993702, "grad_norm": 0.10342071950435638, "global_step": 35613, "epoch": 335, "lr": 2.42831065094821e-05} +{"train_loss": 0.015685640275478363, "grad_norm": 0.10298729687929153, "global_step": 35614, "epoch": 335, "lr": 2.4280564865019943e-05} +{"train_loss": 0.015144105455926005, "grad_norm": 0.1422041803598404, "global_step": 35615, "epoch": 335, "lr": 2.4278023310924673e-05, "val_loss": 0.019447406753897667} +{"train_loss": 0.014704952016472816, "grad_norm": 0.05029119923710823, "global_step": 35616, "epoch": 336, "lr": 2.4275481847205267e-05} +{"train_loss": 0.014390895143151283, "grad_norm": 0.04854995012283325, "global_step": 35617, "epoch": 336, "lr": 2.427294047387061e-05} +{"train_loss": 0.013705102726817131, "grad_norm": 0.1467210054397583, "global_step": 35618, "epoch": 336, "lr": 2.427039919092967e-05} +{"train_loss": 0.01609986647963524, "grad_norm": 0.06453211605548859, "global_step": 35619, "epoch": 336, "lr": 2.4267857998391342e-05} +{"train_loss": 0.01470585074275732, "grad_norm": 0.07493768632411957, "global_step": 35620, "epoch": 336, "lr": 2.4265316896264566e-05} +{"train_loss": 0.013597829267382622, "grad_norm": 0.06102627515792847, "global_step": 35621, "epoch": 336, "lr": 2.4262775884558298e-05} +{"train_loss": 0.01713371090590954, "grad_norm": 0.06514306366443634, "global_step": 35622, "epoch": 336, "lr": 2.426023496328144e-05} +{"train_loss": 0.01659145951271057, "grad_norm": 0.06458395719528198, "global_step": 35623, "epoch": 336, "lr": 2.4257694132442905e-05} +{"train_loss": 0.013529857620596886, "grad_norm": 0.09617143124341965, "global_step": 35624, "epoch": 336, "lr": 2.4255153392051655e-05} +{"train_loss": 0.014653945341706276, "grad_norm": 0.07270826399326324, "global_step": 35625, "epoch": 336, "lr": 2.4252612742116582e-05} +{"train_loss": 0.013989482074975967, "grad_norm": 0.09311100095510483, "global_step": 35626, "epoch": 336, "lr": 2.4250072182646626e-05} +{"train_loss": 0.017889028415083885, "grad_norm": 0.09087070822715759, "global_step": 35627, "epoch": 336, "lr": 2.4247531713650746e-05} +{"train_loss": 0.014592763967812061, "grad_norm": 0.0860142931342125, "global_step": 35628, "epoch": 336, "lr": 2.4244991335137833e-05} +{"train_loss": 0.014750258065760136, "grad_norm": 0.12937135994434357, "global_step": 35629, "epoch": 336, "lr": 2.4242451047116804e-05} +{"train_loss": 0.015223629772663116, "grad_norm": 0.06946907192468643, "global_step": 35630, "epoch": 336, "lr": 2.4239910849596596e-05} +{"train_loss": 0.015376046299934387, "grad_norm": 0.0965430811047554, "global_step": 35631, "epoch": 336, "lr": 2.4237370742586162e-05} +{"train_loss": 0.014807375147938728, "grad_norm": 0.0685303658246994, "global_step": 35632, "epoch": 336, "lr": 2.4234830726094382e-05} +{"train_loss": 0.014152386225759983, "grad_norm": 0.08376984298229218, "global_step": 35633, "epoch": 336, "lr": 2.4232290800130203e-05} +{"train_loss": 0.014771335758268833, "grad_norm": 0.09853915870189667, "global_step": 35634, "epoch": 336, "lr": 2.4229750964702567e-05} +{"train_loss": 0.015643086284399033, "grad_norm": 0.11434494704008102, "global_step": 35635, "epoch": 336, "lr": 2.4227211219820385e-05} +{"train_loss": 0.015129676088690758, "grad_norm": 0.07465403527021408, "global_step": 35636, "epoch": 336, "lr": 2.4224671565492548e-05} +{"train_loss": 0.014924447983503342, "grad_norm": 0.06878148764371872, "global_step": 35637, "epoch": 336, "lr": 2.4222132001728005e-05} +{"train_loss": 0.014718551188707352, "grad_norm": 0.06789233535528183, "global_step": 35638, "epoch": 336, "lr": 2.42195925285357e-05} +{"train_loss": 0.013905534520745277, "grad_norm": 0.05430665984749794, "global_step": 35639, "epoch": 336, "lr": 2.421705314592452e-05} +{"train_loss": 0.013628918677568436, "grad_norm": 0.056247442960739136, "global_step": 35640, "epoch": 336, "lr": 2.421451385390342e-05} +{"train_loss": 0.014165719971060753, "grad_norm": 0.0719190388917923, "global_step": 35641, "epoch": 336, "lr": 2.42119746524813e-05} +{"train_loss": 0.013529613614082336, "grad_norm": 0.10132166743278503, "global_step": 35642, "epoch": 336, "lr": 2.4209435541667068e-05} +{"train_loss": 0.01615328900516033, "grad_norm": 0.1359972357749939, "global_step": 35643, "epoch": 336, "lr": 2.4206896521469667e-05} +{"train_loss": 0.014005723409354687, "grad_norm": 0.07859775424003601, "global_step": 35644, "epoch": 336, "lr": 2.4204357591898035e-05} +{"train_loss": 0.015274139121174812, "grad_norm": 0.14604699611663818, "global_step": 35645, "epoch": 336, "lr": 2.4201818752961046e-05} +{"train_loss": 0.01535535417497158, "grad_norm": 0.09353407472372055, "global_step": 35646, "epoch": 336, "lr": 2.4199280004667672e-05} +{"train_loss": 0.017060602083802223, "grad_norm": 0.11974058300256729, "global_step": 35647, "epoch": 336, "lr": 2.419674134702679e-05} +{"train_loss": 0.013385575264692307, "grad_norm": 0.09427077323198318, "global_step": 35648, "epoch": 336, "lr": 2.4194202780047354e-05} +{"train_loss": 0.018700623884797096, "grad_norm": 0.05909974128007889, "global_step": 35649, "epoch": 336, "lr": 2.419166430373824e-05} +{"train_loss": 0.014599345624446869, "grad_norm": 0.11093717813491821, "global_step": 35650, "epoch": 336, "lr": 2.4189125918108407e-05} +{"train_loss": 0.014305750839412212, "grad_norm": 0.11103396862745285, "global_step": 35651, "epoch": 336, "lr": 2.4186587623166772e-05} +{"train_loss": 0.015527041628956795, "grad_norm": 0.07996832579374313, "global_step": 35652, "epoch": 336, "lr": 2.4184049418922243e-05} +{"train_loss": 0.01227562502026558, "grad_norm": 0.06397496908903122, "global_step": 35653, "epoch": 336, "lr": 2.4181511305383718e-05} +{"train_loss": 0.015635274350643158, "grad_norm": 0.1124151274561882, "global_step": 35654, "epoch": 336, "lr": 2.4178973282560135e-05} +{"train_loss": 0.014855585992336273, "grad_norm": 0.08011499792337418, "global_step": 35655, "epoch": 336, "lr": 2.417643535046043e-05} +{"train_loss": 0.014781536534428596, "grad_norm": 0.06369709223508835, "global_step": 35656, "epoch": 336, "lr": 2.417389750909348e-05} +{"train_loss": 0.014890773221850395, "grad_norm": 0.08310620486736298, "global_step": 35657, "epoch": 336, "lr": 2.4171359758468244e-05} +{"train_loss": 0.015518512576818466, "grad_norm": 0.059737835079431534, "global_step": 35658, "epoch": 336, "lr": 2.4168822098593597e-05} +{"train_loss": 0.01986056938767433, "grad_norm": 0.12863993644714355, "global_step": 35659, "epoch": 336, "lr": 2.4166284529478495e-05} +{"train_loss": 0.017402686178684235, "grad_norm": 0.15780501067638397, "global_step": 35660, "epoch": 336, "lr": 2.4163747051131817e-05} +{"train_loss": 0.01500992476940155, "grad_norm": 0.12360748648643494, "global_step": 35661, "epoch": 336, "lr": 2.4161209663562497e-05} +{"train_loss": 0.013814015313982964, "grad_norm": 0.07123509049415588, "global_step": 35662, "epoch": 336, "lr": 2.4158672366779462e-05} +{"train_loss": 0.016192611306905746, "grad_norm": 0.1136397197842598, "global_step": 35663, "epoch": 336, "lr": 2.4156135160791617e-05} +{"train_loss": 0.016218330711126328, "grad_norm": 0.12676575779914856, "global_step": 35664, "epoch": 336, "lr": 2.4153598045607855e-05} +{"train_loss": 0.017461366951465607, "grad_norm": 0.11767448484897614, "global_step": 35665, "epoch": 336, "lr": 2.4151061021237132e-05} +{"train_loss": 0.015258777886629105, "grad_norm": 0.07617148011922836, "global_step": 35666, "epoch": 336, "lr": 2.4148524087688313e-05} +{"train_loss": 0.01576860621571541, "grad_norm": 0.059239711612463, "global_step": 35667, "epoch": 336, "lr": 2.414598724497034e-05} +{"train_loss": 0.01975230872631073, "grad_norm": 0.09924624860286713, "global_step": 35668, "epoch": 336, "lr": 2.4143450493092146e-05} +{"train_loss": 0.01785130798816681, "grad_norm": 0.14973615109920502, "global_step": 35669, "epoch": 336, "lr": 2.4140913832062623e-05} +{"train_loss": 0.01312211062759161, "grad_norm": 0.07239331305027008, "global_step": 35670, "epoch": 336, "lr": 2.413837726189066e-05} +{"train_loss": 0.01676582545042038, "grad_norm": 0.10670559108257294, "global_step": 35671, "epoch": 336, "lr": 2.4135840782585195e-05} +{"train_loss": 0.013914735987782478, "grad_norm": 0.07139207422733307, "global_step": 35672, "epoch": 336, "lr": 2.413330439415516e-05} +{"train_loss": 0.015610515139997005, "grad_norm": 0.05371793359518051, "global_step": 35673, "epoch": 336, "lr": 2.413076809660942e-05} +{"train_loss": 0.013475763611495495, "grad_norm": 0.07443471997976303, "global_step": 35674, "epoch": 336, "lr": 2.412823188995691e-05} +{"train_loss": 0.013511445373296738, "grad_norm": 0.04988593980669975, "global_step": 35675, "epoch": 336, "lr": 2.4125695774206564e-05} +{"train_loss": 0.017662301659584045, "grad_norm": 0.08970431238412857, "global_step": 35676, "epoch": 336, "lr": 2.4123159749367268e-05} +{"train_loss": 0.014779994264245033, "grad_norm": 0.09011905640363693, "global_step": 35677, "epoch": 336, "lr": 2.412062381544792e-05} +{"train_loss": 0.01491038128733635, "grad_norm": 0.12382630258798599, "global_step": 35678, "epoch": 336, "lr": 2.4118087972457435e-05} +{"train_loss": 0.015492228791117668, "grad_norm": 0.07514345645904541, "global_step": 35679, "epoch": 336, "lr": 2.4115552220404765e-05} +{"train_loss": 0.016089454293251038, "grad_norm": 0.15942782163619995, "global_step": 35680, "epoch": 336, "lr": 2.4113016559298764e-05} +{"train_loss": 0.015479077585041523, "grad_norm": 0.07581640779972076, "global_step": 35681, "epoch": 336, "lr": 2.4110480989148383e-05} +{"train_loss": 0.013125856406986713, "grad_norm": 0.05773404613137245, "global_step": 35682, "epoch": 336, "lr": 2.4107945509962508e-05} +{"train_loss": 0.01521460060030222, "grad_norm": 0.0632571429014206, "global_step": 35683, "epoch": 336, "lr": 2.4105410121750045e-05} +{"train_loss": 0.01354159228503704, "grad_norm": 0.09987707436084747, "global_step": 35684, "epoch": 336, "lr": 2.41028748245199e-05} +{"train_loss": 0.013426070101559162, "grad_norm": 0.0984465628862381, "global_step": 35685, "epoch": 336, "lr": 2.4100339618281014e-05} +{"train_loss": 0.0154055031016469, "grad_norm": 0.08220940083265305, "global_step": 35686, "epoch": 336, "lr": 2.4097804503042248e-05} +{"train_loss": 0.014194879680871964, "grad_norm": 0.08934405446052551, "global_step": 35687, "epoch": 336, "lr": 2.4095269478812554e-05} +{"train_loss": 0.01322685182094574, "grad_norm": 0.05463740602135658, "global_step": 35688, "epoch": 336, "lr": 2.40927345456008e-05} +{"train_loss": 0.014474241062998772, "grad_norm": 0.08119119703769684, "global_step": 35689, "epoch": 336, "lr": 2.409019970341593e-05} +{"train_loss": 0.015401993878185749, "grad_norm": 0.06955774128437042, "global_step": 35690, "epoch": 336, "lr": 2.4087664952266804e-05} +{"train_loss": 0.012004041112959385, "grad_norm": 0.050632644444704056, "global_step": 35691, "epoch": 336, "lr": 2.408513029216236e-05} +{"train_loss": 0.0125843845307827, "grad_norm": 0.07315915822982788, "global_step": 35692, "epoch": 336, "lr": 2.4082595723111517e-05} +{"train_loss": 0.016727542504668236, "grad_norm": 0.18664482235908508, "global_step": 35693, "epoch": 336, "lr": 2.408006124512316e-05} +{"train_loss": 0.015580805018544197, "grad_norm": 0.14332152903079987, "global_step": 35694, "epoch": 336, "lr": 2.4077526858206178e-05} +{"train_loss": 0.01574666053056717, "grad_norm": 0.14360128343105316, "global_step": 35695, "epoch": 336, "lr": 2.407499256236949e-05} +{"train_loss": 0.015467524528503418, "grad_norm": 0.08645960688591003, "global_step": 35696, "epoch": 336, "lr": 2.4072458357622025e-05} +{"train_loss": 0.014993177726864815, "grad_norm": 0.07921036332845688, "global_step": 35697, "epoch": 336, "lr": 2.406992424397265e-05} +{"train_loss": 0.015058590099215508, "grad_norm": 0.143369659781456, "global_step": 35698, "epoch": 336, "lr": 2.406739022143028e-05} +{"train_loss": 0.014991875737905502, "grad_norm": 0.1342613399028778, "global_step": 35699, "epoch": 336, "lr": 2.406485629000386e-05} +{"train_loss": 0.015477482229471207, "grad_norm": 0.07604815810918808, "global_step": 35700, "epoch": 336, "lr": 2.4062322449702225e-05} +{"train_loss": 0.014505691826343536, "grad_norm": 0.10700857639312744, "global_step": 35701, "epoch": 336, "lr": 2.4059788700534303e-05} +{"train_loss": 0.017036043107509613, "grad_norm": 0.08906733989715576, "global_step": 35702, "epoch": 336, "lr": 2.405725504250902e-05} +{"train_loss": 0.01579114980995655, "grad_norm": 0.09919413179159164, "global_step": 35703, "epoch": 336, "lr": 2.4054721475635244e-05} +{"train_loss": 0.01646595261991024, "grad_norm": 0.06777843087911606, "global_step": 35704, "epoch": 336, "lr": 2.405218799992191e-05} +{"train_loss": 0.016070228070020676, "grad_norm": 0.06325023621320724, "global_step": 35705, "epoch": 336, "lr": 2.4049654615377888e-05} +{"train_loss": 0.01618209108710289, "grad_norm": 0.09308553487062454, "global_step": 35706, "epoch": 336, "lr": 2.404712132201211e-05} +{"train_loss": 0.01403347309678793, "grad_norm": 0.13105285167694092, "global_step": 35707, "epoch": 336, "lr": 2.4044588119833445e-05} +{"train_loss": 0.01698262058198452, "grad_norm": 0.1116958037018776, "global_step": 35708, "epoch": 336, "lr": 2.404205500885081e-05} +{"train_loss": 0.013929521664977074, "grad_norm": 0.09524696320295334, "global_step": 35709, "epoch": 336, "lr": 2.4039521989073122e-05} +{"train_loss": 0.012151630595326424, "grad_norm": 0.05913882702589035, "global_step": 35710, "epoch": 336, "lr": 2.4036989060509263e-05} +{"train_loss": 0.016786599531769753, "grad_norm": 0.06087542697787285, "global_step": 35711, "epoch": 336, "lr": 2.4034456223168118e-05} +{"train_loss": 0.01535133458673954, "grad_norm": 0.11439400166273117, "global_step": 35712, "epoch": 336, "lr": 2.4031923477058597e-05} +{"train_loss": 0.015767626464366913, "grad_norm": 0.09400878101587296, "global_step": 35713, "epoch": 336, "lr": 2.402939082218963e-05} +{"train_loss": 0.015721119940280914, "grad_norm": 0.13723161816596985, "global_step": 35714, "epoch": 336, "lr": 2.4026858258570062e-05} +{"train_loss": 0.015938904136419296, "grad_norm": 0.11473580449819565, "global_step": 35715, "epoch": 336, "lr": 2.4024325786208824e-05} +{"train_loss": 0.013480189256370068, "grad_norm": 0.06451433151960373, "global_step": 35716, "epoch": 336, "lr": 2.4021793405114824e-05} +{"train_loss": 0.017892150208353996, "grad_norm": 0.11999113112688065, "global_step": 35717, "epoch": 336, "lr": 2.401926111529694e-05} +{"train_loss": 0.014608586207032204, "grad_norm": 0.0949651226401329, "global_step": 35718, "epoch": 336, "lr": 2.4016728916764064e-05} +{"train_loss": 0.015054065734148026, "grad_norm": 0.06831176578998566, "global_step": 35719, "epoch": 336, "lr": 2.4014196809525102e-05} +{"train_loss": 0.012956303544342518, "grad_norm": 0.06402413547039032, "global_step": 35720, "epoch": 336, "lr": 2.4011664793588967e-05} +{"train_loss": 0.015154221912248517, "grad_norm": 0.07584966719150543, "global_step": 35721, "epoch": 336, "lr": 2.4009132868964522e-05, "val_loss": 0.02050126902759075} +{"train_loss": 0.012468129396438599, "grad_norm": 0.05889398977160454, "global_step": 35722, "epoch": 337, "lr": 2.4006601035660698e-05} +{"train_loss": 0.01614789292216301, "grad_norm": 0.06887536495923996, "global_step": 35723, "epoch": 337, "lr": 2.4004069293686375e-05} +{"train_loss": 0.01654958911240101, "grad_norm": 0.120480015873909, "global_step": 35724, "epoch": 337, "lr": 2.400153764305043e-05} +{"train_loss": 0.013980690389871597, "grad_norm": 0.10289561003446579, "global_step": 35725, "epoch": 337, "lr": 2.399900608376178e-05} +{"train_loss": 0.014855245128273964, "grad_norm": 0.08759292215108871, "global_step": 35726, "epoch": 337, "lr": 2.3996474615829323e-05} +{"train_loss": 0.016045449301600456, "grad_norm": 0.1179514080286026, "global_step": 35727, "epoch": 337, "lr": 2.3993943239261934e-05} +{"train_loss": 0.01603313907980919, "grad_norm": 0.06623008102178574, "global_step": 35728, "epoch": 337, "lr": 2.3991411954068532e-05} +{"train_loss": 0.01416570134460926, "grad_norm": 0.0753856897354126, "global_step": 35729, "epoch": 337, "lr": 2.398888076025798e-05} +{"train_loss": 0.013852131552994251, "grad_norm": 0.09774620085954666, "global_step": 35730, "epoch": 337, "lr": 2.3986349657839213e-05} +{"train_loss": 0.014841124415397644, "grad_norm": 0.07916427403688431, "global_step": 35731, "epoch": 337, "lr": 2.3983818646821076e-05} +{"train_loss": 0.016315560787916183, "grad_norm": 0.056035950779914856, "global_step": 35732, "epoch": 337, "lr": 2.398128772721249e-05} +{"train_loss": 0.016861945390701294, "grad_norm": 0.07507359236478806, "global_step": 35733, "epoch": 337, "lr": 2.3978756899022364e-05} +{"train_loss": 0.015505027025938034, "grad_norm": 0.09533131867647171, "global_step": 35734, "epoch": 337, "lr": 2.3976226162259563e-05} +{"train_loss": 0.016660597175359726, "grad_norm": 0.12149752676486969, "global_step": 35735, "epoch": 337, "lr": 2.397369551693297e-05} +{"train_loss": 0.014155568554997444, "grad_norm": 0.10265795886516571, "global_step": 35736, "epoch": 337, "lr": 2.3971164963051497e-05} +{"train_loss": 0.014406456612050533, "grad_norm": 0.06686591356992722, "global_step": 35737, "epoch": 337, "lr": 2.3968634500624044e-05} +{"train_loss": 0.015223879367113113, "grad_norm": 0.07024091482162476, "global_step": 35738, "epoch": 337, "lr": 2.3966104129659472e-05} +{"train_loss": 0.015554308891296387, "grad_norm": 0.08848990499973297, "global_step": 35739, "epoch": 337, "lr": 2.3963573850166686e-05} +{"train_loss": 0.014942686073482037, "grad_norm": 0.07148857414722443, "global_step": 35740, "epoch": 337, "lr": 2.396104366215462e-05} +{"train_loss": 0.014103539288043976, "grad_norm": 0.061683643609285355, "global_step": 35741, "epoch": 337, "lr": 2.3958513565632085e-05} +{"train_loss": 0.01557333953678608, "grad_norm": 0.08297817409038544, "global_step": 35742, "epoch": 337, "lr": 2.3955983560608003e-05} +{"train_loss": 0.01435836497694254, "grad_norm": 0.12206034362316132, "global_step": 35743, "epoch": 337, "lr": 2.395345364709129e-05} +{"train_loss": 0.015148752368986607, "grad_norm": 0.06384928524494171, "global_step": 35744, "epoch": 337, "lr": 2.39509238250908e-05} +{"train_loss": 0.015257232822477818, "grad_norm": 0.07287072390317917, "global_step": 35745, "epoch": 337, "lr": 2.394839409461543e-05} +{"train_loss": 0.014168012887239456, "grad_norm": 0.1459082067012787, "global_step": 35746, "epoch": 337, "lr": 2.394586445567409e-05} +{"train_loss": 0.0157121904194355, "grad_norm": 0.11352862417697906, "global_step": 35747, "epoch": 337, "lr": 2.3943334908275657e-05} +{"train_loss": 0.013985725119709969, "grad_norm": 0.05477827414870262, "global_step": 35748, "epoch": 337, "lr": 2.3940805452428988e-05} +{"train_loss": 0.012937469407916069, "grad_norm": 0.058727819472551346, "global_step": 35749, "epoch": 337, "lr": 2.3938276088143003e-05} +{"train_loss": 0.014613834209740162, "grad_norm": 0.09414072334766388, "global_step": 35750, "epoch": 337, "lr": 2.3935746815426595e-05} +{"train_loss": 0.01796240359544754, "grad_norm": 0.12242607027292252, "global_step": 35751, "epoch": 337, "lr": 2.3933217634288625e-05} +{"train_loss": 0.015768330544233322, "grad_norm": 0.08132436126470566, "global_step": 35752, "epoch": 337, "lr": 2.3930688544738006e-05} +{"train_loss": 0.014858968555927277, "grad_norm": 0.09653817862272263, "global_step": 35753, "epoch": 337, "lr": 2.392815954678359e-05} +{"train_loss": 0.014868635684251785, "grad_norm": 0.08634230494499207, "global_step": 35754, "epoch": 337, "lr": 2.3925630640434305e-05} +{"train_loss": 0.015703609213232994, "grad_norm": 0.10935831069946289, "global_step": 35755, "epoch": 337, "lr": 2.3923101825698997e-05} +{"train_loss": 0.014199978671967983, "grad_norm": 0.07771360874176025, "global_step": 35756, "epoch": 337, "lr": 2.3920573102586568e-05} +{"train_loss": 0.0150360818952322, "grad_norm": 0.05962048098444939, "global_step": 35757, "epoch": 337, "lr": 2.391804447110594e-05} +{"train_loss": 0.01511075533926487, "grad_norm": 0.07422097772359848, "global_step": 35758, "epoch": 337, "lr": 2.3915515931265918e-05} +{"train_loss": 0.013047556392848492, "grad_norm": 0.08775933086872101, "global_step": 35759, "epoch": 337, "lr": 2.391298748307544e-05} +{"train_loss": 0.014425309374928474, "grad_norm": 0.09324590861797333, "global_step": 35760, "epoch": 337, "lr": 2.3910459126543374e-05} +{"train_loss": 0.014600355178117752, "grad_norm": 0.13734862208366394, "global_step": 35761, "epoch": 337, "lr": 2.3907930861678622e-05} +{"train_loss": 0.016911432147026062, "grad_norm": 0.17944158613681793, "global_step": 35762, "epoch": 337, "lr": 2.3905402688490042e-05} +{"train_loss": 0.01689213700592518, "grad_norm": 0.11562221497297287, "global_step": 35763, "epoch": 337, "lr": 2.3902874606986548e-05} +{"train_loss": 0.013680924661457539, "grad_norm": 0.098911352455616, "global_step": 35764, "epoch": 337, "lr": 2.3900346617177e-05} +{"train_loss": 0.014223760925233364, "grad_norm": 0.04929071292281151, "global_step": 35765, "epoch": 337, "lr": 2.3897818719070263e-05} +{"train_loss": 0.016979046165943146, "grad_norm": 0.08272766321897507, "global_step": 35766, "epoch": 337, "lr": 2.3895290912675243e-05} +{"train_loss": 0.01623980887234211, "grad_norm": 0.06502419710159302, "global_step": 35767, "epoch": 337, "lr": 2.3892763198000834e-05} +{"train_loss": 0.013917066156864166, "grad_norm": 0.10358858108520508, "global_step": 35768, "epoch": 337, "lr": 2.3890235575055885e-05} +{"train_loss": 0.01644992269575596, "grad_norm": 0.07325862348079681, "global_step": 35769, "epoch": 337, "lr": 2.388770804384931e-05} +{"train_loss": 0.014701549895107746, "grad_norm": 0.09856126457452774, "global_step": 35770, "epoch": 337, "lr": 2.3885180604389963e-05} +{"train_loss": 0.014634815976023674, "grad_norm": 0.07670507580041885, "global_step": 35771, "epoch": 337, "lr": 2.3882653256686744e-05} +{"train_loss": 0.013350913301110268, "grad_norm": 0.07862494140863419, "global_step": 35772, "epoch": 337, "lr": 2.3880126000748502e-05} +{"train_loss": 0.017374176532030106, "grad_norm": 0.07485799491405487, "global_step": 35773, "epoch": 337, "lr": 2.3877598836584146e-05} +{"train_loss": 0.013900522142648697, "grad_norm": 0.09690432250499725, "global_step": 35774, "epoch": 337, "lr": 2.3875071764202563e-05} +{"train_loss": 0.014734994620084763, "grad_norm": 0.07439568638801575, "global_step": 35775, "epoch": 337, "lr": 2.3872544783612615e-05} +{"train_loss": 0.013083385303616524, "grad_norm": 0.09241440892219543, "global_step": 35776, "epoch": 337, "lr": 2.3870017894823166e-05} +{"train_loss": 0.01477725151926279, "grad_norm": 0.06868913024663925, "global_step": 35777, "epoch": 337, "lr": 2.386749109784311e-05} +{"train_loss": 0.013263850472867489, "grad_norm": 0.07818039506673813, "global_step": 35778, "epoch": 337, "lr": 2.3864964392681338e-05} +{"train_loss": 0.01453302800655365, "grad_norm": 0.06768911331892014, "global_step": 35779, "epoch": 337, "lr": 2.3862437779346702e-05} +{"train_loss": 0.016887549310922623, "grad_norm": 0.12674197554588318, "global_step": 35780, "epoch": 337, "lr": 2.3859911257848093e-05} +{"train_loss": 0.015013637021183968, "grad_norm": 0.10660252720117569, "global_step": 35781, "epoch": 337, "lr": 2.3857384828194422e-05} +{"train_loss": 0.01628982275724411, "grad_norm": 0.10187475383281708, "global_step": 35782, "epoch": 337, "lr": 2.3854858490394498e-05} +{"train_loss": 0.014562664553523064, "grad_norm": 0.08615949004888535, "global_step": 35783, "epoch": 337, "lr": 2.3852332244457232e-05} +{"train_loss": 0.013885416090488434, "grad_norm": 0.05276746302843094, "global_step": 35784, "epoch": 337, "lr": 2.3849806090391514e-05} +{"train_loss": 0.01458018459379673, "grad_norm": 0.05877608433365822, "global_step": 35785, "epoch": 337, "lr": 2.3847280028206187e-05} +{"train_loss": 0.01803603209555149, "grad_norm": 0.07067007571458817, "global_step": 35786, "epoch": 337, "lr": 2.3844754057910145e-05} +{"train_loss": 0.015903517603874207, "grad_norm": 0.07928574085235596, "global_step": 35787, "epoch": 337, "lr": 2.3842228179512282e-05} +{"train_loss": 0.015202957205474377, "grad_norm": 0.07178757339715958, "global_step": 35788, "epoch": 337, "lr": 2.3839702393021458e-05} +{"train_loss": 0.017118707299232483, "grad_norm": 0.07407461106777191, "global_step": 35789, "epoch": 337, "lr": 2.3837176698446516e-05} +{"train_loss": 0.015669284388422966, "grad_norm": 0.1082053855061531, "global_step": 35790, "epoch": 337, "lr": 2.3834651095796363e-05} +{"train_loss": 0.0185372456908226, "grad_norm": 0.06159596890211105, "global_step": 35791, "epoch": 337, "lr": 2.3832125585079884e-05} +{"train_loss": 0.014898264780640602, "grad_norm": 0.10434965044260025, "global_step": 35792, "epoch": 337, "lr": 2.3829600166305917e-05} +{"train_loss": 0.015190999954938889, "grad_norm": 0.1180722713470459, "global_step": 35793, "epoch": 337, "lr": 2.382707483948337e-05} +{"train_loss": 0.014891603961586952, "grad_norm": 0.0669582411646843, "global_step": 35794, "epoch": 337, "lr": 2.3824549604621082e-05} +{"train_loss": 0.015166562050580978, "grad_norm": 0.06532006710767746, "global_step": 35795, "epoch": 337, "lr": 2.3822024461727965e-05} +{"train_loss": 0.01650785282254219, "grad_norm": 0.11817096173763275, "global_step": 35796, "epoch": 337, "lr": 2.3819499410812845e-05} +{"train_loss": 0.014998971484601498, "grad_norm": 0.05017981678247452, "global_step": 35797, "epoch": 337, "lr": 2.381697445188462e-05} +{"train_loss": 0.01360442116856575, "grad_norm": 0.0714193657040596, "global_step": 35798, "epoch": 337, "lr": 2.3814449584952186e-05} +{"train_loss": 0.01371031440794468, "grad_norm": 0.06279715150594711, "global_step": 35799, "epoch": 337, "lr": 2.3811924810024384e-05} +{"train_loss": 0.016446862369775772, "grad_norm": 0.07308204472064972, "global_step": 35800, "epoch": 337, "lr": 2.380940012711007e-05} +{"train_loss": 0.015046322718262672, "grad_norm": 0.07729559391736984, "global_step": 35801, "epoch": 337, "lr": 2.3806875536218153e-05} +{"train_loss": 0.015409594401717186, "grad_norm": 0.07755178958177567, "global_step": 35802, "epoch": 337, "lr": 2.3804351037357464e-05} +{"train_loss": 0.013555191457271576, "grad_norm": 0.08984144032001495, "global_step": 35803, "epoch": 337, "lr": 2.3801826630536893e-05} +{"train_loss": 0.014287345111370087, "grad_norm": 0.12776367366313934, "global_step": 35804, "epoch": 337, "lr": 2.379930231576533e-05} +{"train_loss": 0.01461545005440712, "grad_norm": 0.07593989372253418, "global_step": 35805, "epoch": 337, "lr": 2.3796778093051626e-05} +{"train_loss": 0.014394927769899368, "grad_norm": 0.08715399354696274, "global_step": 35806, "epoch": 337, "lr": 2.3794253962404627e-05} +{"train_loss": 0.015372836962342262, "grad_norm": 0.10007021576166153, "global_step": 35807, "epoch": 337, "lr": 2.3791729923833222e-05} +{"train_loss": 0.013504582457244396, "grad_norm": 0.06557245552539825, "global_step": 35808, "epoch": 337, "lr": 2.3789205977346303e-05} +{"train_loss": 0.013936871662735939, "grad_norm": 0.06494034081697464, "global_step": 35809, "epoch": 337, "lr": 2.37866821229527e-05} +{"train_loss": 0.014727985486388206, "grad_norm": 0.06565451622009277, "global_step": 35810, "epoch": 337, "lr": 2.3784158360661303e-05} +{"train_loss": 0.01640734262764454, "grad_norm": 0.08016832917928696, "global_step": 35811, "epoch": 337, "lr": 2.3781634690480963e-05} +{"train_loss": 0.014605243690311909, "grad_norm": 0.08940044790506363, "global_step": 35812, "epoch": 337, "lr": 2.3779111112420572e-05} +{"train_loss": 0.01571718230843544, "grad_norm": 0.08201134204864502, "global_step": 35813, "epoch": 337, "lr": 2.377658762648896e-05} +{"train_loss": 0.016154423356056213, "grad_norm": 0.13261735439300537, "global_step": 35814, "epoch": 337, "lr": 2.377406423269502e-05} +{"train_loss": 0.0185479037463665, "grad_norm": 0.08551620692014694, "global_step": 35815, "epoch": 337, "lr": 2.3771540931047632e-05} +{"train_loss": 0.013328528963029385, "grad_norm": 0.10199294239282608, "global_step": 35816, "epoch": 337, "lr": 2.3769017721555643e-05} +{"train_loss": 0.014771176502108574, "grad_norm": 0.08823873847723007, "global_step": 35817, "epoch": 337, "lr": 2.3766494604227897e-05} +{"train_loss": 0.01512172818183899, "grad_norm": 0.12677626311779022, "global_step": 35818, "epoch": 337, "lr": 2.376397157907328e-05} +{"train_loss": 0.015541979111731052, "grad_norm": 0.12761884927749634, "global_step": 35819, "epoch": 337, "lr": 2.3761448646100676e-05} +{"train_loss": 0.018866481259465218, "grad_norm": 0.07281868904829025, "global_step": 35820, "epoch": 337, "lr": 2.375892580531891e-05} +{"train_loss": 0.015326078981161118, "grad_norm": 0.061626460403203964, "global_step": 35821, "epoch": 337, "lr": 2.375640305673687e-05} +{"train_loss": 0.013878554105758667, "grad_norm": 0.08920241892337799, "global_step": 35822, "epoch": 337, "lr": 2.3753880400363453e-05} +{"train_loss": 0.01726459711790085, "grad_norm": 0.081511951982975, "global_step": 35823, "epoch": 337, "lr": 2.375135783620745e-05} +{"train_loss": 0.014992629177868366, "grad_norm": 0.09469505399465561, "global_step": 35824, "epoch": 337, "lr": 2.3748835364277755e-05} +{"train_loss": 0.014855802059173584, "grad_norm": 0.09592224657535553, "global_step": 35825, "epoch": 337, "lr": 2.3746312984583257e-05} +{"train_loss": 0.013130364008247852, "grad_norm": 0.060053225606679916, "global_step": 35826, "epoch": 337, "lr": 2.3743790697132785e-05} +{"train_loss": 0.015132183340852553, "grad_norm": 0.18842682242393494, "global_step": 35827, "epoch": 337, "lr": 2.374126850193521e-05, "val_loss": 0.019037924706935883} +{"train_loss": 0.016554322093725204, "grad_norm": 0.061508744955062866, "global_step": 35828, "epoch": 338, "lr": 2.3738746398999416e-05} +{"train_loss": 0.014078324660658836, "grad_norm": 0.13652081787586212, "global_step": 35829, "epoch": 338, "lr": 2.3736224388334242e-05} +{"train_loss": 0.01741171069443226, "grad_norm": 0.05174216628074646, "global_step": 35830, "epoch": 338, "lr": 2.3733702469948545e-05} +{"train_loss": 0.01393946073949337, "grad_norm": 0.09155724942684174, "global_step": 35831, "epoch": 338, "lr": 2.3731180643851186e-05} +{"train_loss": 0.01208657119423151, "grad_norm": 0.05590042471885681, "global_step": 35832, "epoch": 338, "lr": 2.372865891005106e-05} +{"train_loss": 0.0165726151317358, "grad_norm": 0.06773591041564941, "global_step": 35833, "epoch": 338, "lr": 2.372613726855698e-05} +{"train_loss": 0.016263622790575027, "grad_norm": 0.06781839579343796, "global_step": 35834, "epoch": 338, "lr": 2.3723615719377844e-05} +{"train_loss": 0.013189525343477726, "grad_norm": 0.05114532634615898, "global_step": 35835, "epoch": 338, "lr": 2.3721094262522476e-05} +{"train_loss": 0.014833701774477959, "grad_norm": 0.07229172438383102, "global_step": 35836, "epoch": 338, "lr": 2.3718572897999774e-05} +{"train_loss": 0.014271978288888931, "grad_norm": 0.08232996612787247, "global_step": 35837, "epoch": 338, "lr": 2.3716051625818564e-05} +{"train_loss": 0.01618507131934166, "grad_norm": 0.052970971912145615, "global_step": 35838, "epoch": 338, "lr": 2.3713530445987707e-05} +{"train_loss": 0.0135298240929842, "grad_norm": 0.10473103076219559, "global_step": 35839, "epoch": 338, "lr": 2.3711009358516102e-05} +{"train_loss": 0.012157890945672989, "grad_norm": 0.06386791914701462, "global_step": 35840, "epoch": 338, "lr": 2.3708488363412573e-05} +{"train_loss": 0.015124257653951645, "grad_norm": 0.11150434613227844, "global_step": 35841, "epoch": 338, "lr": 2.3705967460685963e-05} +{"train_loss": 0.014795105904340744, "grad_norm": 0.12686192989349365, "global_step": 35842, "epoch": 338, "lr": 2.3703446650345168e-05} +{"train_loss": 0.018007608130574226, "grad_norm": 0.09024528414011002, "global_step": 35843, "epoch": 338, "lr": 2.3700925932399005e-05} +{"train_loss": 0.015778567641973495, "grad_norm": 0.08661993592977524, "global_step": 35844, "epoch": 338, "lr": 2.3698405306856354e-05} +{"train_loss": 0.013681957498192787, "grad_norm": 0.11432788521051407, "global_step": 35845, "epoch": 338, "lr": 2.3695884773726085e-05} +{"train_loss": 0.017517073079943657, "grad_norm": 0.09251652657985687, "global_step": 35846, "epoch": 338, "lr": 2.3693364333017038e-05} +{"train_loss": 0.015562405809760094, "grad_norm": 0.07793837785720825, "global_step": 35847, "epoch": 338, "lr": 2.369084398473805e-05} +{"train_loss": 0.014200047589838505, "grad_norm": 0.06052621826529503, "global_step": 35848, "epoch": 338, "lr": 2.368832372889799e-05} +{"train_loss": 0.013476947322487831, "grad_norm": 0.059537485241889954, "global_step": 35849, "epoch": 338, "lr": 2.3685803565505743e-05} +{"train_loss": 0.015287517569959164, "grad_norm": 0.06580843031406403, "global_step": 35850, "epoch": 338, "lr": 2.3683283494570115e-05} +{"train_loss": 0.015305736102163792, "grad_norm": 0.05726178362965584, "global_step": 35851, "epoch": 338, "lr": 2.3680763516099984e-05} +{"train_loss": 0.01698477938771248, "grad_norm": 0.05733437091112137, "global_step": 35852, "epoch": 338, "lr": 2.367824363010423e-05} +{"train_loss": 0.016135351732373238, "grad_norm": 0.07905025035142899, "global_step": 35853, "epoch": 338, "lr": 2.3675723836591674e-05} +{"train_loss": 0.015473934821784496, "grad_norm": 0.07537224888801575, "global_step": 35854, "epoch": 338, "lr": 2.367320413557116e-05} +{"train_loss": 0.014864420518279076, "grad_norm": 0.0956585630774498, "global_step": 35855, "epoch": 338, "lr": 2.3670684527051557e-05} +{"train_loss": 0.01534262951463461, "grad_norm": 0.07235226035118103, "global_step": 35856, "epoch": 338, "lr": 2.3668165011041738e-05} +{"train_loss": 0.014719108119606972, "grad_norm": 0.09791193157434464, "global_step": 35857, "epoch": 338, "lr": 2.3665645587550532e-05} +{"train_loss": 0.013964117504656315, "grad_norm": 0.13439346849918365, "global_step": 35858, "epoch": 338, "lr": 2.3663126256586783e-05} +{"train_loss": 0.01406472735106945, "grad_norm": 0.08230011910200119, "global_step": 35859, "epoch": 338, "lr": 2.3660607018159364e-05} +{"train_loss": 0.012378918007016182, "grad_norm": 0.05418660119175911, "global_step": 35860, "epoch": 338, "lr": 2.3658087872277108e-05} +{"train_loss": 0.01727653667330742, "grad_norm": 0.0929156094789505, "global_step": 35861, "epoch": 338, "lr": 2.365556881894887e-05} +{"train_loss": 0.01712777465581894, "grad_norm": 0.1284775733947754, "global_step": 35862, "epoch": 338, "lr": 2.365304985818351e-05} +{"train_loss": 0.01670168526470661, "grad_norm": 0.08550971746444702, "global_step": 35863, "epoch": 338, "lr": 2.3650530989989905e-05} +{"train_loss": 0.016468102112412453, "grad_norm": 0.07903770357370377, "global_step": 35864, "epoch": 338, "lr": 2.364801221437684e-05} +{"train_loss": 0.01557585597038269, "grad_norm": 0.10814967751502991, "global_step": 35865, "epoch": 338, "lr": 2.3645493531353207e-05} +{"train_loss": 0.016195358708500862, "grad_norm": 0.06923633068799973, "global_step": 35866, "epoch": 338, "lr": 2.364297494092786e-05} +{"train_loss": 0.01306790579110384, "grad_norm": 0.053900737315416336, "global_step": 35867, "epoch": 338, "lr": 2.364045644310962e-05} +{"train_loss": 0.016886495053768158, "grad_norm": 0.059503160417079926, "global_step": 35868, "epoch": 338, "lr": 2.3637938037907354e-05} +{"train_loss": 0.018885189667344093, "grad_norm": 0.11881827563047409, "global_step": 35869, "epoch": 338, "lr": 2.3635419725329928e-05} +{"train_loss": 0.01617046259343624, "grad_norm": 0.07817254960536957, "global_step": 35870, "epoch": 338, "lr": 2.3632901505386167e-05} +{"train_loss": 0.014336244203150272, "grad_norm": 0.0821288526058197, "global_step": 35871, "epoch": 338, "lr": 2.363038337808491e-05} +{"train_loss": 0.016626110300421715, "grad_norm": 0.09849674999713898, "global_step": 35872, "epoch": 338, "lr": 2.362786534343502e-05} +{"train_loss": 0.014355476014316082, "grad_norm": 0.06558802723884583, "global_step": 35873, "epoch": 338, "lr": 2.3625347401445354e-05} +{"train_loss": 0.013823515735566616, "grad_norm": 0.18196217715740204, "global_step": 35874, "epoch": 338, "lr": 2.362282955212473e-05} +{"train_loss": 0.014584817923605442, "grad_norm": 0.12244755029678345, "global_step": 35875, "epoch": 338, "lr": 2.3620311795482032e-05} +{"train_loss": 0.017434274777770042, "grad_norm": 0.11486674100160599, "global_step": 35876, "epoch": 338, "lr": 2.3617794131526066e-05} +{"train_loss": 0.016561975702643394, "grad_norm": 0.0652630478143692, "global_step": 35877, "epoch": 338, "lr": 2.361527656026572e-05} +{"train_loss": 0.017090430483222008, "grad_norm": 0.08681455254554749, "global_step": 35878, "epoch": 338, "lr": 2.3612759081709795e-05} +{"train_loss": 0.01324386429041624, "grad_norm": 0.08387395739555359, "global_step": 35879, "epoch": 338, "lr": 2.3610241695867163e-05} +{"train_loss": 0.015206137672066689, "grad_norm": 0.12126662582159042, "global_step": 35880, "epoch": 338, "lr": 2.3607724402746684e-05} +{"train_loss": 0.015359152108430862, "grad_norm": 0.07146543264389038, "global_step": 35881, "epoch": 338, "lr": 2.3605207202357182e-05} +{"train_loss": 0.01263104472309351, "grad_norm": 0.0640731081366539, "global_step": 35882, "epoch": 338, "lr": 2.3602690094707482e-05} +{"train_loss": 0.014749485068023205, "grad_norm": 0.0801747739315033, "global_step": 35883, "epoch": 338, "lr": 2.360017307980647e-05} +{"train_loss": 0.016859468072652817, "grad_norm": 0.06092679128050804, "global_step": 35884, "epoch": 338, "lr": 2.359765615766295e-05} +{"train_loss": 0.013705958612263203, "grad_norm": 0.07009786367416382, "global_step": 35885, "epoch": 338, "lr": 2.3595139328285786e-05} +{"train_loss": 0.015616502612829208, "grad_norm": 0.06293347477912903, "global_step": 35886, "epoch": 338, "lr": 2.3592622591683832e-05} +{"train_loss": 0.014748593792319298, "grad_norm": 0.0722610279917717, "global_step": 35887, "epoch": 338, "lr": 2.3590105947865924e-05} +{"train_loss": 0.01402213703840971, "grad_norm": 0.07959945499897003, "global_step": 35888, "epoch": 338, "lr": 2.3587589396840883e-05} +{"train_loss": 0.015406584367156029, "grad_norm": 0.08757442235946655, "global_step": 35889, "epoch": 338, "lr": 2.3585072938617557e-05} +{"train_loss": 0.01647251471877098, "grad_norm": 0.07781662791967392, "global_step": 35890, "epoch": 338, "lr": 2.3582556573204824e-05} +{"train_loss": 0.015068335458636284, "grad_norm": 0.10924045741558075, "global_step": 35891, "epoch": 338, "lr": 2.3580040300611472e-05} +{"train_loss": 0.013896120712161064, "grad_norm": 0.0943395271897316, "global_step": 35892, "epoch": 338, "lr": 2.357752412084637e-05} +{"train_loss": 0.013776624575257301, "grad_norm": 0.11854348331689835, "global_step": 35893, "epoch": 338, "lr": 2.3575008033918383e-05} +{"train_loss": 0.01470260787755251, "grad_norm": 0.09249430149793625, "global_step": 35894, "epoch": 338, "lr": 2.357249203983632e-05} +{"train_loss": 0.0133857736364007, "grad_norm": 0.06380659341812134, "global_step": 35895, "epoch": 338, "lr": 2.356997613860901e-05} +{"train_loss": 0.016120925545692444, "grad_norm": 0.08077550679445267, "global_step": 35896, "epoch": 338, "lr": 2.356746033024531e-05} +{"train_loss": 0.014216594398021698, "grad_norm": 0.07022149860858917, "global_step": 35897, "epoch": 338, "lr": 2.3564944614754076e-05} +{"train_loss": 0.01342556532472372, "grad_norm": 0.10377934575080872, "global_step": 35898, "epoch": 338, "lr": 2.3562428992144114e-05} +{"train_loss": 0.013116776943206787, "grad_norm": 0.06633099168539047, "global_step": 35899, "epoch": 338, "lr": 2.35599134624243e-05} +{"train_loss": 0.014257005415856838, "grad_norm": 0.10940445214509964, "global_step": 35900, "epoch": 338, "lr": 2.3557398025603445e-05} +{"train_loss": 0.01703587919473648, "grad_norm": 0.07839670032262802, "global_step": 35901, "epoch": 338, "lr": 2.3554882681690376e-05} +{"train_loss": 0.015237264335155487, "grad_norm": 0.1259172409772873, "global_step": 35902, "epoch": 338, "lr": 2.3552367430693954e-05} +{"train_loss": 0.01635771244764328, "grad_norm": 0.16510601341724396, "global_step": 35903, "epoch": 338, "lr": 2.3549852272623024e-05} +{"train_loss": 0.015454128384590149, "grad_norm": 0.10420094430446625, "global_step": 35904, "epoch": 338, "lr": 2.3547337207486397e-05} +{"train_loss": 0.015727737918496132, "grad_norm": 0.07277783006429672, "global_step": 35905, "epoch": 338, "lr": 2.354482223529294e-05} +{"train_loss": 0.014219165779650211, "grad_norm": 0.0899752601981163, "global_step": 35906, "epoch": 338, "lr": 2.354230735605145e-05} +{"train_loss": 0.017817936837673187, "grad_norm": 0.07028183341026306, "global_step": 35907, "epoch": 338, "lr": 2.353979256977081e-05} +{"train_loss": 0.013079995289444923, "grad_norm": 0.05127890780568123, "global_step": 35908, "epoch": 338, "lr": 2.3537277876459813e-05} +{"train_loss": 0.014694463461637497, "grad_norm": 0.05405140295624733, "global_step": 35909, "epoch": 338, "lr": 2.353476327612731e-05} +{"train_loss": 0.015370352193713188, "grad_norm": 0.08072492480278015, "global_step": 35910, "epoch": 338, "lr": 2.3532248768782156e-05} +{"train_loss": 0.012622293084859848, "grad_norm": 0.07150152325630188, "global_step": 35911, "epoch": 338, "lr": 2.3529734354433174e-05} +{"train_loss": 0.013584495522081852, "grad_norm": 0.053034380078315735, "global_step": 35912, "epoch": 338, "lr": 2.3527220033089175e-05} +{"train_loss": 0.017593465745449066, "grad_norm": 0.07627876102924347, "global_step": 35913, "epoch": 338, "lr": 2.3524705804759006e-05} +{"train_loss": 0.016121963039040565, "grad_norm": 0.09245481342077255, "global_step": 35914, "epoch": 338, "lr": 2.3522191669451533e-05} +{"train_loss": 0.01370043121278286, "grad_norm": 0.06766089797019958, "global_step": 35915, "epoch": 338, "lr": 2.351967762717554e-05} +{"train_loss": 0.014439491555094719, "grad_norm": 0.0577317513525486, "global_step": 35916, "epoch": 338, "lr": 2.3517163677939907e-05} +{"train_loss": 0.01615281030535698, "grad_norm": 0.14402319490909576, "global_step": 35917, "epoch": 338, "lr": 2.351464982175342e-05} +{"train_loss": 0.015198792330920696, "grad_norm": 0.08347205072641373, "global_step": 35918, "epoch": 338, "lr": 2.3512136058624963e-05} +{"train_loss": 0.015747951343655586, "grad_norm": 0.08099304884672165, "global_step": 35919, "epoch": 338, "lr": 2.3509622388563313e-05} +{"train_loss": 0.015697257593274117, "grad_norm": 0.07719254493713379, "global_step": 35920, "epoch": 338, "lr": 2.3507108811577333e-05} +{"train_loss": 0.013228270225226879, "grad_norm": 0.054309483617544174, "global_step": 35921, "epoch": 338, "lr": 2.3504595327675866e-05} +{"train_loss": 0.01718202792108059, "grad_norm": 0.11883176863193512, "global_step": 35922, "epoch": 338, "lr": 2.350208193686773e-05} +{"train_loss": 0.015578946098685265, "grad_norm": 0.08484023064374924, "global_step": 35923, "epoch": 338, "lr": 2.3499568639161738e-05} +{"train_loss": 0.018404599279165268, "grad_norm": 0.09393566101789474, "global_step": 35924, "epoch": 338, "lr": 2.3497055434566756e-05} +{"train_loss": 0.013670253567397594, "grad_norm": 0.11421572417020798, "global_step": 35925, "epoch": 338, "lr": 2.3494542323091578e-05} +{"train_loss": 0.016385024413466454, "grad_norm": 0.08193393796682358, "global_step": 35926, "epoch": 338, "lr": 2.349202930474505e-05} +{"train_loss": 0.015187369659543037, "grad_norm": 0.062227580696344376, "global_step": 35927, "epoch": 338, "lr": 2.348951637953602e-05} +{"train_loss": 0.013557402417063713, "grad_norm": 0.06457162648439407, "global_step": 35928, "epoch": 338, "lr": 2.3487003547473303e-05} +{"train_loss": 0.014313659630715847, "grad_norm": 0.12114263325929642, "global_step": 35929, "epoch": 338, "lr": 2.3484490808565703e-05} +{"train_loss": 0.01483858935534954, "grad_norm": 0.12542176246643066, "global_step": 35930, "epoch": 338, "lr": 2.348197816282207e-05} +{"train_loss": 0.013829800300300121, "grad_norm": 0.08168487250804901, "global_step": 35931, "epoch": 338, "lr": 2.3479465610251255e-05} +{"train_loss": 0.015211787074804306, "grad_norm": 0.07921474426984787, "global_step": 35932, "epoch": 338, "lr": 2.3476953150862047e-05} +{"train_loss": 0.01513318783375171, "grad_norm": 0.14036312699317932, "global_step": 35933, "epoch": 338, "lr": 2.347444078466329e-05, "val_loss": 0.019605932757258415} +{"train_loss": 0.01685166358947754, "grad_norm": 0.08043667674064636, "global_step": 35934, "epoch": 339, "lr": 2.3471928511663833e-05} +{"train_loss": 0.01578587293624878, "grad_norm": 0.05412190407514572, "global_step": 35935, "epoch": 339, "lr": 2.346941633187248e-05} +{"train_loss": 0.015884026885032654, "grad_norm": 0.07514218240976334, "global_step": 35936, "epoch": 339, "lr": 2.346690424529804e-05} +{"train_loss": 0.016297802329063416, "grad_norm": 0.05844473838806152, "global_step": 35937, "epoch": 339, "lr": 2.3464392251949357e-05} +{"train_loss": 0.014953927136957645, "grad_norm": 0.08814039826393127, "global_step": 35938, "epoch": 339, "lr": 2.3461880351835286e-05} +{"train_loss": 0.013385551050305367, "grad_norm": 0.06405309587717056, "global_step": 35939, "epoch": 339, "lr": 2.3459368544964607e-05} +{"train_loss": 0.015324709936976433, "grad_norm": 0.07508896291255951, "global_step": 35940, "epoch": 339, "lr": 2.3456856831346174e-05} +{"train_loss": 0.014111356809735298, "grad_norm": 0.055291254073381424, "global_step": 35941, "epoch": 339, "lr": 2.3454345210988816e-05} +{"train_loss": 0.013960057869553566, "grad_norm": 0.07921729981899261, "global_step": 35942, "epoch": 339, "lr": 2.3451833683901315e-05} +{"train_loss": 0.016980350017547607, "grad_norm": 0.07983777672052383, "global_step": 35943, "epoch": 339, "lr": 2.3449322250092532e-05} +{"train_loss": 0.0154138607904315, "grad_norm": 0.06348345428705215, "global_step": 35944, "epoch": 339, "lr": 2.34468109095713e-05} +{"train_loss": 0.016769561916589737, "grad_norm": 0.07981139421463013, "global_step": 35945, "epoch": 339, "lr": 2.3444299662346404e-05} +{"train_loss": 0.016904857009649277, "grad_norm": 0.06251873075962067, "global_step": 35946, "epoch": 339, "lr": 2.3441788508426716e-05} +{"train_loss": 0.017189498990774155, "grad_norm": 0.12705856561660767, "global_step": 35947, "epoch": 339, "lr": 2.343927744782101e-05} +{"train_loss": 0.013439692556858063, "grad_norm": 0.06726085394620895, "global_step": 35948, "epoch": 339, "lr": 2.343676648053815e-05} +{"train_loss": 0.012640077620744705, "grad_norm": 0.05744313821196556, "global_step": 35949, "epoch": 339, "lr": 2.3434255606586925e-05} +{"train_loss": 0.017110329121351242, "grad_norm": 0.15672741830348969, "global_step": 35950, "epoch": 339, "lr": 2.343174482597617e-05} +{"train_loss": 0.013261497020721436, "grad_norm": 0.1620967835187912, "global_step": 35951, "epoch": 339, "lr": 2.3429234138714732e-05} +{"train_loss": 0.01699940860271454, "grad_norm": 0.18220728635787964, "global_step": 35952, "epoch": 339, "lr": 2.342672354481141e-05} +{"train_loss": 0.01663137599825859, "grad_norm": 0.07255841046571732, "global_step": 35953, "epoch": 339, "lr": 2.3424213044275002e-05} +{"train_loss": 0.013858800753951073, "grad_norm": 0.10025656968355179, "global_step": 35954, "epoch": 339, "lr": 2.3421702637114356e-05} +{"train_loss": 0.014554611407220364, "grad_norm": 0.09482290595769882, "global_step": 35955, "epoch": 339, "lr": 2.3419192323338303e-05} +{"train_loss": 0.013193141669034958, "grad_norm": 0.07533152401447296, "global_step": 35956, "epoch": 339, "lr": 2.3416682102955635e-05} +{"train_loss": 0.015885483473539352, "grad_norm": 0.07860752940177917, "global_step": 35957, "epoch": 339, "lr": 2.3414171975975195e-05} +{"train_loss": 0.015933610498905182, "grad_norm": 0.06986703723669052, "global_step": 35958, "epoch": 339, "lr": 2.34116619424058e-05} +{"train_loss": 0.014563820324838161, "grad_norm": 0.09132419526576996, "global_step": 35959, "epoch": 339, "lr": 2.3409152002256247e-05} +{"train_loss": 0.013414628803730011, "grad_norm": 0.10601525008678436, "global_step": 35960, "epoch": 339, "lr": 2.3406642155535363e-05} +{"train_loss": 0.015287676826119423, "grad_norm": 0.07051549851894379, "global_step": 35961, "epoch": 339, "lr": 2.3404132402251992e-05} +{"train_loss": 0.014583476819097996, "grad_norm": 0.1265312135219574, "global_step": 35962, "epoch": 339, "lr": 2.3401622742414915e-05} +{"train_loss": 0.01508388202637434, "grad_norm": 0.07786434143781662, "global_step": 35963, "epoch": 339, "lr": 2.339911317603299e-05} +{"train_loss": 0.016607843339443207, "grad_norm": 0.07179280370473862, "global_step": 35964, "epoch": 339, "lr": 2.3396603703114996e-05} +{"train_loss": 0.013241762295365334, "grad_norm": 0.0874127671122551, "global_step": 35965, "epoch": 339, "lr": 2.339409432366978e-05} +{"train_loss": 0.0175495482981205, "grad_norm": 0.06955043226480484, "global_step": 35966, "epoch": 339, "lr": 2.3391585037706126e-05} +{"train_loss": 0.014201121404767036, "grad_norm": 0.08354882150888443, "global_step": 35967, "epoch": 339, "lr": 2.3389075845232878e-05} +{"train_loss": 0.0169413760304451, "grad_norm": 0.0904630497097969, "global_step": 35968, "epoch": 339, "lr": 2.338656674625886e-05} +{"train_loss": 0.015739720314741135, "grad_norm": 0.07053790986537933, "global_step": 35969, "epoch": 339, "lr": 2.338405774079287e-05} +{"train_loss": 0.01663024351000786, "grad_norm": 0.07746653258800507, "global_step": 35970, "epoch": 339, "lr": 2.3381548828843712e-05} +{"train_loss": 0.016640905290842056, "grad_norm": 0.13719530403614044, "global_step": 35971, "epoch": 339, "lr": 2.337904001042021e-05} +{"train_loss": 0.0159333236515522, "grad_norm": 0.09399569034576416, "global_step": 35972, "epoch": 339, "lr": 2.3376531285531205e-05} +{"train_loss": 0.014643959701061249, "grad_norm": 0.10187730938196182, "global_step": 35973, "epoch": 339, "lr": 2.3374022654185474e-05} +{"train_loss": 0.01579306274652481, "grad_norm": 0.07678689807653427, "global_step": 35974, "epoch": 339, "lr": 2.337151411639185e-05} +{"train_loss": 0.01448219083249569, "grad_norm": 0.06292957812547684, "global_step": 35975, "epoch": 339, "lr": 2.336900567215916e-05} +{"train_loss": 0.015017213299870491, "grad_norm": 0.12419290840625763, "global_step": 35976, "epoch": 339, "lr": 2.3366497321496206e-05} +{"train_loss": 0.013000350445508957, "grad_norm": 0.1396905481815338, "global_step": 35977, "epoch": 339, "lr": 2.3363989064411772e-05} +{"train_loss": 0.015263669192790985, "grad_norm": 0.05948774144053459, "global_step": 35978, "epoch": 339, "lr": 2.33614809009147e-05} +{"train_loss": 0.01295226439833641, "grad_norm": 0.07197198271751404, "global_step": 35979, "epoch": 339, "lr": 2.335897283101382e-05} +{"train_loss": 0.013493457809090614, "grad_norm": 0.07199735194444656, "global_step": 35980, "epoch": 339, "lr": 2.335646485471791e-05} +{"train_loss": 0.015103505924344063, "grad_norm": 0.06543387472629547, "global_step": 35981, "epoch": 339, "lr": 2.3353956972035806e-05} +{"train_loss": 0.014408882707357407, "grad_norm": 0.07212013751268387, "global_step": 35982, "epoch": 339, "lr": 2.335144918297631e-05} +{"train_loss": 0.014218835160136223, "grad_norm": 0.10827706754207611, "global_step": 35983, "epoch": 339, "lr": 2.3348941487548215e-05} +{"train_loss": 0.013618429191410542, "grad_norm": 0.09985087811946869, "global_step": 35984, "epoch": 339, "lr": 2.3346433885760348e-05} +{"train_loss": 0.012954382225871086, "grad_norm": 0.16904424130916595, "global_step": 35985, "epoch": 339, "lr": 2.3343926377621545e-05} +{"train_loss": 0.014356352388858795, "grad_norm": 0.08538120985031128, "global_step": 35986, "epoch": 339, "lr": 2.334141896314057e-05} +{"train_loss": 0.013659912161529064, "grad_norm": 0.07231884449720383, "global_step": 35987, "epoch": 339, "lr": 2.333891164232627e-05} +{"train_loss": 0.014279510825872421, "grad_norm": 0.14333103597164154, "global_step": 35988, "epoch": 339, "lr": 2.333640441518743e-05} +{"train_loss": 0.01428049523383379, "grad_norm": 0.06625853478908539, "global_step": 35989, "epoch": 339, "lr": 2.333389728173288e-05} +{"train_loss": 0.013303762301802635, "grad_norm": 0.09156268835067749, "global_step": 35990, "epoch": 339, "lr": 2.33313902419714e-05} +{"train_loss": 0.01562836393713951, "grad_norm": 0.194647878408432, "global_step": 35991, "epoch": 339, "lr": 2.3328883295911824e-05} +{"train_loss": 0.01462988555431366, "grad_norm": 0.06111268699169159, "global_step": 35992, "epoch": 339, "lr": 2.3326376443562964e-05} +{"train_loss": 0.014898235909640789, "grad_norm": 0.06673102080821991, "global_step": 35993, "epoch": 339, "lr": 2.332386968493362e-05} +{"train_loss": 0.016816396266222, "grad_norm": 0.058762967586517334, "global_step": 35994, "epoch": 339, "lr": 2.332136302003258e-05} +{"train_loss": 0.017281007021665573, "grad_norm": 0.0953156128525734, "global_step": 35995, "epoch": 339, "lr": 2.3318856448868665e-05} +{"train_loss": 0.015737956389784813, "grad_norm": 0.1912057250738144, "global_step": 35996, "epoch": 339, "lr": 2.3316349971450706e-05} +{"train_loss": 0.016881316900253296, "grad_norm": 0.11604434251785278, "global_step": 35997, "epoch": 339, "lr": 2.331384358778747e-05} +{"train_loss": 0.013535396195948124, "grad_norm": 0.05196733772754669, "global_step": 35998, "epoch": 339, "lr": 2.3311337297887782e-05} +{"train_loss": 0.016787905246019363, "grad_norm": 0.09993524104356766, "global_step": 35999, "epoch": 339, "lr": 2.3308831101760486e-05} +{"train_loss": 0.016268864274024963, "grad_norm": 0.10698337852954865, "global_step": 36000, "epoch": 339, "lr": 2.330632499941431e-05} +{"train_loss": 0.018039502203464508, "grad_norm": 0.08723003417253494, "global_step": 36001, "epoch": 339, "lr": 2.33038189908581e-05} +{"train_loss": 0.0156385637819767, "grad_norm": 0.10057162493467331, "global_step": 36002, "epoch": 339, "lr": 2.3301313076100678e-05} +{"train_loss": 0.015587512403726578, "grad_norm": 0.08268611133098602, "global_step": 36003, "epoch": 339, "lr": 2.3298807255150817e-05} +{"train_loss": 0.015392684377729893, "grad_norm": 0.06780943274497986, "global_step": 36004, "epoch": 339, "lr": 2.329630152801733e-05} +{"train_loss": 0.015074137598276138, "grad_norm": 0.060912441462278366, "global_step": 36005, "epoch": 339, "lr": 2.3293795894709047e-05} +{"train_loss": 0.016862384974956512, "grad_norm": 0.14094600081443787, "global_step": 36006, "epoch": 339, "lr": 2.329129035523475e-05} +{"train_loss": 0.015394138172268867, "grad_norm": 0.07041677832603455, "global_step": 36007, "epoch": 339, "lr": 2.328878490960323e-05} +{"train_loss": 0.014364119619131088, "grad_norm": 0.11272469907999039, "global_step": 36008, "epoch": 339, "lr": 2.32862795578233e-05} +{"train_loss": 0.016802020370960236, "grad_norm": 0.13243766129016876, "global_step": 36009, "epoch": 339, "lr": 2.3283774299903783e-05} +{"train_loss": 0.01810695230960846, "grad_norm": 0.09336253255605698, "global_step": 36010, "epoch": 339, "lr": 2.328126913585346e-05} +{"train_loss": 0.016924692317843437, "grad_norm": 0.08427674323320389, "global_step": 36011, "epoch": 339, "lr": 2.3278764065681123e-05} +{"train_loss": 0.015194438397884369, "grad_norm": 0.07491745799779892, "global_step": 36012, "epoch": 339, "lr": 2.327625908939559e-05} +{"train_loss": 0.016318421810865402, "grad_norm": 0.11603835225105286, "global_step": 36013, "epoch": 339, "lr": 2.3273754207005682e-05} +{"train_loss": 0.014914382249116898, "grad_norm": 0.08866184949874878, "global_step": 36014, "epoch": 339, "lr": 2.3271249418520157e-05} +{"train_loss": 0.017780384048819542, "grad_norm": 0.08188686519861221, "global_step": 36015, "epoch": 339, "lr": 2.326874472394784e-05} +{"train_loss": 0.015685562044382095, "grad_norm": 0.07356465607881546, "global_step": 36016, "epoch": 339, "lr": 2.3266240123297562e-05} +{"train_loss": 0.013960661366581917, "grad_norm": 0.06754005700349808, "global_step": 36017, "epoch": 339, "lr": 2.3263735616578052e-05} +{"train_loss": 0.018198981881141663, "grad_norm": 0.19819298386573792, "global_step": 36018, "epoch": 339, "lr": 2.326123120379815e-05} +{"train_loss": 0.016263209283351898, "grad_norm": 0.08436069637537003, "global_step": 36019, "epoch": 339, "lr": 2.3258726884966677e-05} +{"train_loss": 0.014764631167054176, "grad_norm": 0.06322884559631348, "global_step": 36020, "epoch": 339, "lr": 2.325622266009238e-05} +{"train_loss": 0.015020772814750671, "grad_norm": 0.08195691555738449, "global_step": 36021, "epoch": 339, "lr": 2.3253718529184095e-05} +{"train_loss": 0.015307077206671238, "grad_norm": 0.09197685122489929, "global_step": 36022, "epoch": 339, "lr": 2.3251214492250623e-05} +{"train_loss": 0.014656170271337032, "grad_norm": 0.08075965195894241, "global_step": 36023, "epoch": 339, "lr": 2.3248710549300755e-05} +{"train_loss": 0.01345797348767519, "grad_norm": 0.09426640719175339, "global_step": 36024, "epoch": 339, "lr": 2.3246206700343264e-05} +{"train_loss": 0.015470595099031925, "grad_norm": 0.09149590134620667, "global_step": 36025, "epoch": 339, "lr": 2.324370294538697e-05} +{"train_loss": 0.013961844146251678, "grad_norm": 0.0850689709186554, "global_step": 36026, "epoch": 339, "lr": 2.3241199284440685e-05} +{"train_loss": 0.01362176425755024, "grad_norm": 0.06765154749155045, "global_step": 36027, "epoch": 339, "lr": 2.3238695717513164e-05} +{"train_loss": 0.01475149393081665, "grad_norm": 0.08384960144758224, "global_step": 36028, "epoch": 339, "lr": 2.3236192244613253e-05} +{"train_loss": 0.01214101817458868, "grad_norm": 0.07024107873439789, "global_step": 36029, "epoch": 339, "lr": 2.3233688865749704e-05} +{"train_loss": 0.013880599290132523, "grad_norm": 0.13022847473621368, "global_step": 36030, "epoch": 339, "lr": 2.3231185580931354e-05} +{"train_loss": 0.014575544744729996, "grad_norm": 0.06659667938947678, "global_step": 36031, "epoch": 339, "lr": 2.322868239016695e-05} +{"train_loss": 0.013931820169091225, "grad_norm": 0.08216503262519836, "global_step": 36032, "epoch": 339, "lr": 2.3226179293465316e-05} +{"train_loss": 0.014238646253943443, "grad_norm": 0.07007203996181488, "global_step": 36033, "epoch": 339, "lr": 2.3223676290835268e-05} +{"train_loss": 0.015286391600966454, "grad_norm": 0.07977116852998734, "global_step": 36034, "epoch": 339, "lr": 2.3221173382285573e-05} +{"train_loss": 0.015092756599187851, "grad_norm": 0.0722663477063179, "global_step": 36035, "epoch": 339, "lr": 2.3218670567825006e-05} +{"train_loss": 0.012868581339716911, "grad_norm": 0.10606905817985535, "global_step": 36036, "epoch": 339, "lr": 2.3216167847462396e-05} +{"train_loss": 0.015102295204997063, "grad_norm": 0.06808454543352127, "global_step": 36037, "epoch": 339, "lr": 2.3213665221206532e-05} +{"train_loss": 0.014643444679677486, "grad_norm": 0.07753246277570724, "global_step": 36038, "epoch": 339, "lr": 2.3211162689066183e-05} +{"train_loss": 0.015147870520727252, "grad_norm": 0.1111740842461586, "global_step": 36039, "epoch": 339, "lr": 2.3208660251050158e-05, "val_loss": 0.019139720126986504} +{"train_loss": 0.012703727930784225, "grad_norm": 0.09545522928237915, "global_step": 36040, "epoch": 340, "lr": 2.3206157907167286e-05} +{"train_loss": 0.01547425240278244, "grad_norm": 0.10107172280550003, "global_step": 36041, "epoch": 340, "lr": 2.320365565742629e-05} +{"train_loss": 0.01765507087111473, "grad_norm": 0.11452813446521759, "global_step": 36042, "epoch": 340, "lr": 2.320115350183599e-05} +{"train_loss": 0.01470949873328209, "grad_norm": 0.12059963494539261, "global_step": 36043, "epoch": 340, "lr": 2.3198651440405207e-05} +{"train_loss": 0.01650351658463478, "grad_norm": 0.07322312146425247, "global_step": 36044, "epoch": 340, "lr": 2.319614947314268e-05} +{"train_loss": 0.017689276486635208, "grad_norm": 0.10135325789451599, "global_step": 36045, "epoch": 340, "lr": 2.3193647600057235e-05} +{"train_loss": 0.01588529348373413, "grad_norm": 0.13404609262943268, "global_step": 36046, "epoch": 340, "lr": 2.3191145821157668e-05} +{"train_loss": 0.013627666048705578, "grad_norm": 0.05328105390071869, "global_step": 36047, "epoch": 340, "lr": 2.318864413645276e-05} +{"train_loss": 0.017123648896813393, "grad_norm": 0.07405341416597366, "global_step": 36048, "epoch": 340, "lr": 2.3186142545951274e-05} +{"train_loss": 0.017184223979711533, "grad_norm": 0.07188884168863297, "global_step": 36049, "epoch": 340, "lr": 2.3183641049662018e-05} +{"train_loss": 0.01280160341411829, "grad_norm": 0.0906294658780098, "global_step": 36050, "epoch": 340, "lr": 2.3181139647593813e-05} +{"train_loss": 0.014383570291101933, "grad_norm": 0.09723887592554092, "global_step": 36051, "epoch": 340, "lr": 2.3178638339755397e-05} +{"train_loss": 0.01266099140048027, "grad_norm": 0.1116214171051979, "global_step": 36052, "epoch": 340, "lr": 2.31761371261556e-05} +{"train_loss": 0.01617223396897316, "grad_norm": 0.08693870902061462, "global_step": 36053, "epoch": 340, "lr": 2.3173636006803172e-05} +{"train_loss": 0.01667034812271595, "grad_norm": 0.08417771011590958, "global_step": 36054, "epoch": 340, "lr": 2.3171134981706943e-05} +{"train_loss": 0.013831009157001972, "grad_norm": 0.12593986093997955, "global_step": 36055, "epoch": 340, "lr": 2.316863405087566e-05} +{"train_loss": 0.014912113547325134, "grad_norm": 0.0782533511519432, "global_step": 36056, "epoch": 340, "lr": 2.3166133214318125e-05} +{"train_loss": 0.01657119207084179, "grad_norm": 0.07832545787096024, "global_step": 36057, "epoch": 340, "lr": 2.3163632472043144e-05} +{"train_loss": 0.014597272500395775, "grad_norm": 0.09173847734928131, "global_step": 36058, "epoch": 340, "lr": 2.3161131824059495e-05} +{"train_loss": 0.01577814109623432, "grad_norm": 0.10235096514225006, "global_step": 36059, "epoch": 340, "lr": 2.3158631270375936e-05} +{"train_loss": 0.01757483184337616, "grad_norm": 0.0854429081082344, "global_step": 36060, "epoch": 340, "lr": 2.3156130811001286e-05} +{"train_loss": 0.016007449477910995, "grad_norm": 0.07958726584911346, "global_step": 36061, "epoch": 340, "lr": 2.3153630445944307e-05} +{"train_loss": 0.012657951563596725, "grad_norm": 0.09799017757177353, "global_step": 36062, "epoch": 340, "lr": 2.315113017521379e-05} +{"train_loss": 0.013332165777683258, "grad_norm": 0.0732518881559372, "global_step": 36063, "epoch": 340, "lr": 2.314862999881855e-05} +{"train_loss": 0.014573105610907078, "grad_norm": 0.10010779649019241, "global_step": 36064, "epoch": 340, "lr": 2.3146129916767345e-05} +{"train_loss": 0.01563277654349804, "grad_norm": 0.08066193759441376, "global_step": 36065, "epoch": 340, "lr": 2.3143629929068938e-05} +{"train_loss": 0.013090485706925392, "grad_norm": 0.0694400817155838, "global_step": 36066, "epoch": 340, "lr": 2.314113003573214e-05} +{"train_loss": 0.017091581597924232, "grad_norm": 0.09184957295656204, "global_step": 36067, "epoch": 340, "lr": 2.3138630236765747e-05} +{"train_loss": 0.016192704439163208, "grad_norm": 0.14608612656593323, "global_step": 36068, "epoch": 340, "lr": 2.3136130532178508e-05} +{"train_loss": 0.015901725739240646, "grad_norm": 0.06580652296543121, "global_step": 36069, "epoch": 340, "lr": 2.3133630921979243e-05} +{"train_loss": 0.01549425721168518, "grad_norm": 0.10110265761613846, "global_step": 36070, "epoch": 340, "lr": 2.3131131406176697e-05} +{"train_loss": 0.01457739993929863, "grad_norm": 0.07323767244815826, "global_step": 36071, "epoch": 340, "lr": 2.3128631984779687e-05} +{"train_loss": 0.014215566217899323, "grad_norm": 0.06689002364873886, "global_step": 36072, "epoch": 340, "lr": 2.312613265779696e-05} +{"train_loss": 0.014196803793311119, "grad_norm": 0.051764167845249176, "global_step": 36073, "epoch": 340, "lr": 2.3123633425237316e-05} +{"train_loss": 0.013410454615950584, "grad_norm": 0.10998066514730453, "global_step": 36074, "epoch": 340, "lr": 2.3121134287109554e-05} +{"train_loss": 0.01367390900850296, "grad_norm": 0.09206872433423996, "global_step": 36075, "epoch": 340, "lr": 2.3118635243422437e-05} +{"train_loss": 0.01600053906440735, "grad_norm": 0.08927717059850693, "global_step": 36076, "epoch": 340, "lr": 2.3116136294184726e-05} +{"train_loss": 0.013837341219186783, "grad_norm": 0.06159476935863495, "global_step": 36077, "epoch": 340, "lr": 2.3113637439405222e-05} +{"train_loss": 0.013692107051610947, "grad_norm": 0.13276241719722748, "global_step": 36078, "epoch": 340, "lr": 2.311113867909272e-05} +{"train_loss": 0.0170372873544693, "grad_norm": 0.09654407203197479, "global_step": 36079, "epoch": 340, "lr": 2.3108640013255968e-05} +{"train_loss": 0.015317649580538273, "grad_norm": 0.06539247930049896, "global_step": 36080, "epoch": 340, "lr": 2.3106141441903758e-05} +{"train_loss": 0.013233177363872528, "grad_norm": 0.09950664639472961, "global_step": 36081, "epoch": 340, "lr": 2.310364296504491e-05} +{"train_loss": 0.01780875027179718, "grad_norm": 0.0828004851937294, "global_step": 36082, "epoch": 340, "lr": 2.3101144582688122e-05} +{"train_loss": 0.014922574162483215, "grad_norm": 0.0970844253897667, "global_step": 36083, "epoch": 340, "lr": 2.309864629484222e-05} +{"train_loss": 0.016049101948738098, "grad_norm": 0.13165706396102905, "global_step": 36084, "epoch": 340, "lr": 2.309614810151599e-05} +{"train_loss": 0.014030318707227707, "grad_norm": 0.06347737461328506, "global_step": 36085, "epoch": 340, "lr": 2.3093650002718175e-05} +{"train_loss": 0.01412191055715084, "grad_norm": 0.12119295448064804, "global_step": 36086, "epoch": 340, "lr": 2.309115199845758e-05} +{"train_loss": 0.017210889607667923, "grad_norm": 0.13820794224739075, "global_step": 36087, "epoch": 340, "lr": 2.308865408874299e-05} +{"train_loss": 0.015295341610908508, "grad_norm": 0.0910676047205925, "global_step": 36088, "epoch": 340, "lr": 2.3086156273583164e-05} +{"train_loss": 0.017253249883651733, "grad_norm": 0.18013520538806915, "global_step": 36089, "epoch": 340, "lr": 2.3083658552986865e-05} +{"train_loss": 0.01589425653219223, "grad_norm": 0.11762166023254395, "global_step": 36090, "epoch": 340, "lr": 2.3081160926962882e-05} +{"train_loss": 0.014525816775858402, "grad_norm": 0.07683567702770233, "global_step": 36091, "epoch": 340, "lr": 2.3078663395520007e-05} +{"train_loss": 0.016462722793221474, "grad_norm": 0.10426565259695053, "global_step": 36092, "epoch": 340, "lr": 2.307616595866699e-05} +{"train_loss": 0.01593177393078804, "grad_norm": 0.09371589869260788, "global_step": 36093, "epoch": 340, "lr": 2.307366861641263e-05} +{"train_loss": 0.016856689006090164, "grad_norm": 0.12850113213062286, "global_step": 36094, "epoch": 340, "lr": 2.307117136876567e-05} +{"train_loss": 0.016105983406305313, "grad_norm": 0.15300463140010834, "global_step": 36095, "epoch": 340, "lr": 2.3068674215734925e-05} +{"train_loss": 0.01403695996850729, "grad_norm": 0.06441552191972733, "global_step": 36096, "epoch": 340, "lr": 2.3066177157329126e-05} +{"train_loss": 0.014635393396019936, "grad_norm": 0.13523203134536743, "global_step": 36097, "epoch": 340, "lr": 2.306368019355707e-05} +{"train_loss": 0.016446691006422043, "grad_norm": 0.13160376250743866, "global_step": 36098, "epoch": 340, "lr": 2.3061183324427543e-05} +{"train_loss": 0.014953376725316048, "grad_norm": 0.06465443968772888, "global_step": 36099, "epoch": 340, "lr": 2.3058686549949306e-05} +{"train_loss": 0.014987832866609097, "grad_norm": 0.09667208045721054, "global_step": 36100, "epoch": 340, "lr": 2.3056189870131107e-05} +{"train_loss": 0.013518663123250008, "grad_norm": 0.08378688246011734, "global_step": 36101, "epoch": 340, "lr": 2.305369328498176e-05} +{"train_loss": 0.01473266165703535, "grad_norm": 0.11872424185276031, "global_step": 36102, "epoch": 340, "lr": 2.3051196794509992e-05} +{"train_loss": 0.01404247060418129, "grad_norm": 0.05659794434905052, "global_step": 36103, "epoch": 340, "lr": 2.3048700398724604e-05} +{"train_loss": 0.01563078537583351, "grad_norm": 0.05966152250766754, "global_step": 36104, "epoch": 340, "lr": 2.3046204097634378e-05} +{"train_loss": 0.01692294329404831, "grad_norm": 0.06899868696928024, "global_step": 36105, "epoch": 340, "lr": 2.3043707891248067e-05} +{"train_loss": 0.0173872672021389, "grad_norm": 0.07096011191606522, "global_step": 36106, "epoch": 340, "lr": 2.304121177957443e-05} +{"train_loss": 0.016369126737117767, "grad_norm": 0.08256027102470398, "global_step": 36107, "epoch": 340, "lr": 2.303871576262225e-05} +{"train_loss": 0.012352408841252327, "grad_norm": 0.09432218223810196, "global_step": 36108, "epoch": 340, "lr": 2.3036219840400313e-05} +{"train_loss": 0.015221280977129936, "grad_norm": 0.09541068971157074, "global_step": 36109, "epoch": 340, "lr": 2.3033724012917356e-05} +{"train_loss": 0.014510692097246647, "grad_norm": 0.10707242041826248, "global_step": 36110, "epoch": 340, "lr": 2.3031228280182182e-05} +{"train_loss": 0.015329881571233273, "grad_norm": 0.09508663415908813, "global_step": 36111, "epoch": 340, "lr": 2.3028732642203527e-05} +{"train_loss": 0.012846698053181171, "grad_norm": 0.09699451178312302, "global_step": 36112, "epoch": 340, "lr": 2.3026237098990194e-05} +{"train_loss": 0.013289299793541431, "grad_norm": 0.1075715646147728, "global_step": 36113, "epoch": 340, "lr": 2.3023741650550916e-05} +{"train_loss": 0.014223405160009861, "grad_norm": 0.06634590774774551, "global_step": 36114, "epoch": 340, "lr": 2.302124629689448e-05} +{"train_loss": 0.016405563801527023, "grad_norm": 0.059935685247182846, "global_step": 36115, "epoch": 340, "lr": 2.301875103802967e-05} +{"train_loss": 0.014655512757599354, "grad_norm": 0.08439112454652786, "global_step": 36116, "epoch": 340, "lr": 2.3016255873965233e-05} +{"train_loss": 0.016052309423685074, "grad_norm": 0.1587119996547699, "global_step": 36117, "epoch": 340, "lr": 2.3013760804709916e-05} +{"train_loss": 0.014943079091608524, "grad_norm": 0.06979573518037796, "global_step": 36118, "epoch": 340, "lr": 2.3011265830272532e-05} +{"train_loss": 0.01676366664469242, "grad_norm": 0.1590849906206131, "global_step": 36119, "epoch": 340, "lr": 2.30087709506618e-05} +{"train_loss": 0.015641022473573685, "grad_norm": 0.06845132261514664, "global_step": 36120, "epoch": 340, "lr": 2.3006276165886508e-05} +{"train_loss": 0.01595703884959221, "grad_norm": 0.1131896898150444, "global_step": 36121, "epoch": 340, "lr": 2.3003781475955443e-05} +{"train_loss": 0.014297828078269958, "grad_norm": 0.056986164301633835, "global_step": 36122, "epoch": 340, "lr": 2.3001286880877353e-05} +{"train_loss": 0.012331453152000904, "grad_norm": 0.0683591440320015, "global_step": 36123, "epoch": 340, "lr": 2.2998792380660976e-05} +{"train_loss": 0.01581587642431259, "grad_norm": 0.07945732772350311, "global_step": 36124, "epoch": 340, "lr": 2.29962979753151e-05} +{"train_loss": 0.012991217896342278, "grad_norm": 0.11171364039182663, "global_step": 36125, "epoch": 340, "lr": 2.299380366484851e-05} +{"train_loss": 0.0136381471529603, "grad_norm": 0.08359971642494202, "global_step": 36126, "epoch": 340, "lr": 2.2991309449269934e-05} +{"train_loss": 0.013414916582405567, "grad_norm": 0.06981834024190903, "global_step": 36127, "epoch": 340, "lr": 2.2988815328588143e-05} +{"train_loss": 0.014107203111052513, "grad_norm": 0.1535809487104416, "global_step": 36128, "epoch": 340, "lr": 2.2986321302811932e-05} +{"train_loss": 0.013601507991552353, "grad_norm": 0.059602439403533936, "global_step": 36129, "epoch": 340, "lr": 2.298382737195004e-05} +{"train_loss": 0.016029052436351776, "grad_norm": 0.08498547971248627, "global_step": 36130, "epoch": 340, "lr": 2.2981333536011207e-05} +{"train_loss": 0.01525680162012577, "grad_norm": 0.1193951740860939, "global_step": 36131, "epoch": 340, "lr": 2.297883979500422e-05} +{"train_loss": 0.016660552471876144, "grad_norm": 0.13492953777313232, "global_step": 36132, "epoch": 340, "lr": 2.297634614893786e-05} +{"train_loss": 0.012788210064172745, "grad_norm": 0.11762646585702896, "global_step": 36133, "epoch": 340, "lr": 2.2973852597820844e-05} +{"train_loss": 0.013254323974251747, "grad_norm": 0.12822259962558746, "global_step": 36134, "epoch": 340, "lr": 2.297135914166198e-05} +{"train_loss": 0.01645020954310894, "grad_norm": 0.06658011674880981, "global_step": 36135, "epoch": 340, "lr": 2.2968865780469985e-05} +{"train_loss": 0.018523672595620155, "grad_norm": 0.09691981971263885, "global_step": 36136, "epoch": 340, "lr": 2.296637251425366e-05} +{"train_loss": 0.01530323177576065, "grad_norm": 0.060923926532268524, "global_step": 36137, "epoch": 340, "lr": 2.296387934302172e-05} +{"train_loss": 0.01616794615983963, "grad_norm": 0.10037188977003098, "global_step": 36138, "epoch": 340, "lr": 2.296138626678296e-05} +{"train_loss": 0.014999435283243656, "grad_norm": 0.13982616364955902, "global_step": 36139, "epoch": 340, "lr": 2.2958893285546145e-05} +{"train_loss": 0.014373554848134518, "grad_norm": 0.09613121300935745, "global_step": 36140, "epoch": 340, "lr": 2.2956400399320015e-05} +{"train_loss": 0.015410097315907478, "grad_norm": 0.12545399367809296, "global_step": 36141, "epoch": 340, "lr": 2.2953907608113318e-05} +{"train_loss": 0.01654159277677536, "grad_norm": 0.12680426239967346, "global_step": 36142, "epoch": 340, "lr": 2.2951414911934848e-05} +{"train_loss": 0.014862707816064358, "grad_norm": 0.0708746463060379, "global_step": 36143, "epoch": 340, "lr": 2.294892231079332e-05} +{"train_loss": 0.013950113207101822, "grad_norm": 0.1099708080291748, "global_step": 36144, "epoch": 340, "lr": 2.2946429804697516e-05} +{"train_loss": 0.015117524806761517, "grad_norm": 0.08849528431892395, "global_step": 36145, "epoch": 340, "lr": 2.294393739365621e-05, "train_batch_mse_error": 0.013436131179332733, "val_loss": 0.019458822906017303} +{"train_loss": 0.015027161687612534, "grad_norm": 0.09624067693948746, "global_step": 36146, "epoch": 341, "lr": 2.294144507767815e-05} +{"train_loss": 0.013510366901755333, "grad_norm": 0.11636368185281754, "global_step": 36147, "epoch": 341, "lr": 2.2938952856772055e-05} +{"train_loss": 0.014900071546435356, "grad_norm": 0.0679299458861351, "global_step": 36148, "epoch": 341, "lr": 2.293646073094672e-05} +{"train_loss": 0.015392321161925793, "grad_norm": 0.06966274231672287, "global_step": 36149, "epoch": 341, "lr": 2.2933968700210906e-05} +{"train_loss": 0.016005558893084526, "grad_norm": 0.05490926653146744, "global_step": 36150, "epoch": 341, "lr": 2.2931476764573334e-05} +{"train_loss": 0.018728064373135567, "grad_norm": 0.07165832817554474, "global_step": 36151, "epoch": 341, "lr": 2.2928984924042784e-05} +{"train_loss": 0.012963039800524712, "grad_norm": 0.07234059274196625, "global_step": 36152, "epoch": 341, "lr": 2.292649317862803e-05} +{"train_loss": 0.012666217982769012, "grad_norm": 0.07876073569059372, "global_step": 36153, "epoch": 341, "lr": 2.2924001528337803e-05} +{"train_loss": 0.014392081648111343, "grad_norm": 0.06381627917289734, "global_step": 36154, "epoch": 341, "lr": 2.2921509973180838e-05} +{"train_loss": 0.013428157195448875, "grad_norm": 0.06618344783782959, "global_step": 36155, "epoch": 341, "lr": 2.2919018513165913e-05} +{"train_loss": 0.014041615650057793, "grad_norm": 0.06806845217943192, "global_step": 36156, "epoch": 341, "lr": 2.29165271483018e-05} +{"train_loss": 0.016874711960554123, "grad_norm": 0.06225685030221939, "global_step": 36157, "epoch": 341, "lr": 2.2914035878597207e-05} +{"train_loss": 0.015050185844302177, "grad_norm": 0.06102019548416138, "global_step": 36158, "epoch": 341, "lr": 2.2911544704060938e-05} +{"train_loss": 0.015825914219021797, "grad_norm": 0.070461206138134, "global_step": 36159, "epoch": 341, "lr": 2.2909053624701715e-05} +{"train_loss": 0.012656561098992825, "grad_norm": 0.06789221614599228, "global_step": 36160, "epoch": 341, "lr": 2.2906562640528284e-05} +{"train_loss": 0.016247671097517014, "grad_norm": 0.07083533704280853, "global_step": 36161, "epoch": 341, "lr": 2.2904071751549404e-05} +{"train_loss": 0.01445290632545948, "grad_norm": 0.07399838417768478, "global_step": 36162, "epoch": 341, "lr": 2.2901580957773855e-05} +{"train_loss": 0.015716638416051865, "grad_norm": 0.07711319625377655, "global_step": 36163, "epoch": 341, "lr": 2.2899090259210366e-05} +{"train_loss": 0.015165368095040321, "grad_norm": 0.1255672425031662, "global_step": 36164, "epoch": 341, "lr": 2.289659965586766e-05} +{"train_loss": 0.013632172718644142, "grad_norm": 0.14305919408798218, "global_step": 36165, "epoch": 341, "lr": 2.2894109147754526e-05} +{"train_loss": 0.01484757848083973, "grad_norm": 0.12047478556632996, "global_step": 36166, "epoch": 341, "lr": 2.2891618734879722e-05} +{"train_loss": 0.015619802288711071, "grad_norm": 0.07498667389154434, "global_step": 36167, "epoch": 341, "lr": 2.288912841725196e-05} +{"train_loss": 0.014588741585612297, "grad_norm": 0.1327764093875885, "global_step": 36168, "epoch": 341, "lr": 2.2886638194880007e-05} +{"train_loss": 0.012096543796360493, "grad_norm": 0.06448275595903397, "global_step": 36169, "epoch": 341, "lr": 2.288414806777264e-05} +{"train_loss": 0.014213120564818382, "grad_norm": 0.07290299981832504, "global_step": 36170, "epoch": 341, "lr": 2.288165803593858e-05} +{"train_loss": 0.01597636193037033, "grad_norm": 0.0703028216958046, "global_step": 36171, "epoch": 341, "lr": 2.2879168099386565e-05} +{"train_loss": 0.015510957688093185, "grad_norm": 0.08335430920124054, "global_step": 36172, "epoch": 341, "lr": 2.2876678258125356e-05} +{"train_loss": 0.014718988910317421, "grad_norm": 0.12414002418518066, "global_step": 36173, "epoch": 341, "lr": 2.2874188512163726e-05} +{"train_loss": 0.018162667751312256, "grad_norm": 0.1263824999332428, "global_step": 36174, "epoch": 341, "lr": 2.287169886151038e-05} +{"train_loss": 0.015641234815120697, "grad_norm": 0.08246475458145142, "global_step": 36175, "epoch": 341, "lr": 2.2869209306174105e-05} +{"train_loss": 0.013786238618195057, "grad_norm": 0.059440772980451584, "global_step": 36176, "epoch": 341, "lr": 2.286671984616363e-05} +{"train_loss": 0.015055174939334393, "grad_norm": 0.11434438079595566, "global_step": 36177, "epoch": 341, "lr": 2.2864230481487682e-05} +{"train_loss": 0.018905967473983765, "grad_norm": 0.08823361247777939, "global_step": 36178, "epoch": 341, "lr": 2.286174121215503e-05} +{"train_loss": 0.014368616044521332, "grad_norm": 0.057758878916502, "global_step": 36179, "epoch": 341, "lr": 2.2859252038174418e-05} +{"train_loss": 0.017360851168632507, "grad_norm": 0.11829554289579391, "global_step": 36180, "epoch": 341, "lr": 2.2856762959554607e-05} +{"train_loss": 0.016977116465568542, "grad_norm": 0.06577737629413605, "global_step": 36181, "epoch": 341, "lr": 2.2854273976304326e-05} +{"train_loss": 0.014919376000761986, "grad_norm": 0.12893067300319672, "global_step": 36182, "epoch": 341, "lr": 2.28517850884323e-05} +{"train_loss": 0.015209155157208443, "grad_norm": 0.1169213354587555, "global_step": 36183, "epoch": 341, "lr": 2.2849296295947315e-05} +{"train_loss": 0.01567094400525093, "grad_norm": 0.08803792297840118, "global_step": 36184, "epoch": 341, "lr": 2.284680759885808e-05} +{"train_loss": 0.014426106587052345, "grad_norm": 0.0645366832613945, "global_step": 36185, "epoch": 341, "lr": 2.2844318997173352e-05} +{"train_loss": 0.015994908288121223, "grad_norm": 0.07457859814167023, "global_step": 36186, "epoch": 341, "lr": 2.28418304909019e-05} +{"train_loss": 0.014660853892564774, "grad_norm": 0.08102817833423615, "global_step": 36187, "epoch": 341, "lr": 2.283934208005244e-05} +{"train_loss": 0.015714410692453384, "grad_norm": 0.0772850513458252, "global_step": 36188, "epoch": 341, "lr": 2.2836853764633704e-05} +{"train_loss": 0.014806521125137806, "grad_norm": 0.08368393778800964, "global_step": 36189, "epoch": 341, "lr": 2.2834365544654447e-05} +{"train_loss": 0.013610005378723145, "grad_norm": 0.07438156008720398, "global_step": 36190, "epoch": 341, "lr": 2.2831877420123437e-05} +{"train_loss": 0.012543927878141403, "grad_norm": 0.09027658402919769, "global_step": 36191, "epoch": 341, "lr": 2.2829389391049377e-05} +{"train_loss": 0.013884168118238449, "grad_norm": 0.07541611045598984, "global_step": 36192, "epoch": 341, "lr": 2.2826901457441023e-05} +{"train_loss": 0.014570262283086777, "grad_norm": 0.05893127992749214, "global_step": 36193, "epoch": 341, "lr": 2.2824413619307143e-05} +{"train_loss": 0.017300305888056755, "grad_norm": 0.06673435866832733, "global_step": 36194, "epoch": 341, "lr": 2.282192587665645e-05} +{"train_loss": 0.01244065910577774, "grad_norm": 0.06472501903772354, "global_step": 36195, "epoch": 341, "lr": 2.2819438229497676e-05} +{"train_loss": 0.01457950659096241, "grad_norm": 0.10955823212862015, "global_step": 36196, "epoch": 341, "lr": 2.2816950677839578e-05} +{"train_loss": 0.013464601710438728, "grad_norm": 0.05316948890686035, "global_step": 36197, "epoch": 341, "lr": 2.2814463221690906e-05} +{"train_loss": 0.01464011613279581, "grad_norm": 0.0804242268204689, "global_step": 36198, "epoch": 341, "lr": 2.2811975861060368e-05} +{"train_loss": 0.014067485928535461, "grad_norm": 0.07704632729291916, "global_step": 36199, "epoch": 341, "lr": 2.2809488595956745e-05} +{"train_loss": 0.015937335789203644, "grad_norm": 0.06767658144235611, "global_step": 36200, "epoch": 341, "lr": 2.280700142638876e-05} +{"train_loss": 0.014537477865815163, "grad_norm": 0.11664755642414093, "global_step": 36201, "epoch": 341, "lr": 2.280451435236512e-05} +{"train_loss": 0.015397640876471996, "grad_norm": 0.09377346932888031, "global_step": 36202, "epoch": 341, "lr": 2.280202737389459e-05} +{"train_loss": 0.013339092023670673, "grad_norm": 0.0589263029396534, "global_step": 36203, "epoch": 341, "lr": 2.2799540490985928e-05} +{"train_loss": 0.014956475235521793, "grad_norm": 0.06970266997814178, "global_step": 36204, "epoch": 341, "lr": 2.2797053703647837e-05} +{"train_loss": 0.016217516735196114, "grad_norm": 0.07995367795228958, "global_step": 36205, "epoch": 341, "lr": 2.279456701188908e-05} +{"train_loss": 0.013840118423104286, "grad_norm": 0.12194646894931793, "global_step": 36206, "epoch": 341, "lr": 2.279208041571837e-05} +{"train_loss": 0.014277328737080097, "grad_norm": 0.08516199886798859, "global_step": 36207, "epoch": 341, "lr": 2.278959391514447e-05} +{"train_loss": 0.01738029345870018, "grad_norm": 0.06342995166778564, "global_step": 36208, "epoch": 341, "lr": 2.2787107510176092e-05} +{"train_loss": 0.014863349497318268, "grad_norm": 0.1714422106742859, "global_step": 36209, "epoch": 341, "lr": 2.2784621200821977e-05} +{"train_loss": 0.015402024611830711, "grad_norm": 0.08826740086078644, "global_step": 36210, "epoch": 341, "lr": 2.278213498709089e-05} +{"train_loss": 0.01281660608947277, "grad_norm": 0.07937974482774734, "global_step": 36211, "epoch": 341, "lr": 2.2779648868991544e-05} +{"train_loss": 0.01599763333797455, "grad_norm": 0.06308489292860031, "global_step": 36212, "epoch": 341, "lr": 2.277716284653265e-05} +{"train_loss": 0.013059079647064209, "grad_norm": 0.05590590834617615, "global_step": 36213, "epoch": 341, "lr": 2.2774676919722975e-05} +{"train_loss": 0.015604354441165924, "grad_norm": 0.06384172290563583, "global_step": 36214, "epoch": 341, "lr": 2.277219108857126e-05} +{"train_loss": 0.01568768545985222, "grad_norm": 0.06095685809850693, "global_step": 36215, "epoch": 341, "lr": 2.2769705353086206e-05} +{"train_loss": 0.016352985054254532, "grad_norm": 0.08513451367616653, "global_step": 36216, "epoch": 341, "lr": 2.2767219713276582e-05} +{"train_loss": 0.01832295022904873, "grad_norm": 0.0747363269329071, "global_step": 36217, "epoch": 341, "lr": 2.276473416915111e-05} +{"train_loss": 0.015283050015568733, "grad_norm": 0.08991843461990356, "global_step": 36218, "epoch": 341, "lr": 2.2762248720718493e-05} +{"train_loss": 0.014190711081027985, "grad_norm": 0.11930514872074127, "global_step": 36219, "epoch": 341, "lr": 2.275976336798749e-05} +{"train_loss": 0.015508897602558136, "grad_norm": 0.0691017359495163, "global_step": 36220, "epoch": 341, "lr": 2.2757278110966855e-05} +{"train_loss": 0.014401092194020748, "grad_norm": 0.07476319372653961, "global_step": 36221, "epoch": 341, "lr": 2.2754792949665276e-05} +{"train_loss": 0.01656167395412922, "grad_norm": 0.06344792991876602, "global_step": 36222, "epoch": 341, "lr": 2.275230788409152e-05} +{"train_loss": 0.016444997861981392, "grad_norm": 0.08153454214334488, "global_step": 36223, "epoch": 341, "lr": 2.2749822914254292e-05} +{"train_loss": 0.01565011963248253, "grad_norm": 0.08728069067001343, "global_step": 36224, "epoch": 341, "lr": 2.2747338040162354e-05} +{"train_loss": 0.0134038794785738, "grad_norm": 0.06262312829494476, "global_step": 36225, "epoch": 341, "lr": 2.27448532618244e-05} +{"train_loss": 0.01678939163684845, "grad_norm": 0.07979528605937958, "global_step": 36226, "epoch": 341, "lr": 2.2742368579249173e-05} +{"train_loss": 0.01799885183572769, "grad_norm": 0.13053317368030548, "global_step": 36227, "epoch": 341, "lr": 2.2739883992445434e-05} +{"train_loss": 0.016379384323954582, "grad_norm": 0.08062374591827393, "global_step": 36228, "epoch": 341, "lr": 2.273739950142188e-05} +{"train_loss": 0.016751091927289963, "grad_norm": 0.06805587559938431, "global_step": 36229, "epoch": 341, "lr": 2.2734915106187238e-05} +{"train_loss": 0.01693265326321125, "grad_norm": 0.09504388272762299, "global_step": 36230, "epoch": 341, "lr": 2.2732430806750244e-05} +{"train_loss": 0.017889849841594696, "grad_norm": 0.13697727024555206, "global_step": 36231, "epoch": 341, "lr": 2.2729946603119646e-05} +{"train_loss": 0.014003724791109562, "grad_norm": 0.08453597128391266, "global_step": 36232, "epoch": 341, "lr": 2.2727462495304137e-05} +{"train_loss": 0.014297077432274818, "grad_norm": 0.05607832595705986, "global_step": 36233, "epoch": 341, "lr": 2.2724978483312463e-05} +{"train_loss": 0.013935505412518978, "grad_norm": 0.06598208099603653, "global_step": 36234, "epoch": 341, "lr": 2.2722494567153378e-05} +{"train_loss": 0.014559008181095123, "grad_norm": 0.056376006454229355, "global_step": 36235, "epoch": 341, "lr": 2.272001074683558e-05} +{"train_loss": 0.014216188341379166, "grad_norm": 0.13128389418125153, "global_step": 36236, "epoch": 341, "lr": 2.2717527022367787e-05} +{"train_loss": 0.014370905235409737, "grad_norm": 0.0985528826713562, "global_step": 36237, "epoch": 341, "lr": 2.2715043393758735e-05} +{"train_loss": 0.014433825388550758, "grad_norm": 0.07588256150484085, "global_step": 36238, "epoch": 341, "lr": 2.2712559861017168e-05} +{"train_loss": 0.015301747247576714, "grad_norm": 0.07112699002027512, "global_step": 36239, "epoch": 341, "lr": 2.2710076424151788e-05} +{"train_loss": 0.01318911463022232, "grad_norm": 0.05694657191634178, "global_step": 36240, "epoch": 341, "lr": 2.2707593083171348e-05} +{"train_loss": 0.016334110870957375, "grad_norm": 0.05692524090409279, "global_step": 36241, "epoch": 341, "lr": 2.2705109838084554e-05} +{"train_loss": 0.012758305296301842, "grad_norm": 0.05226133018732071, "global_step": 36242, "epoch": 341, "lr": 2.270262668890012e-05} +{"train_loss": 0.01673862338066101, "grad_norm": 0.06478149443864822, "global_step": 36243, "epoch": 341, "lr": 2.2700143635626782e-05} +{"train_loss": 0.015325241722166538, "grad_norm": 0.08331865817308426, "global_step": 36244, "epoch": 341, "lr": 2.2697660678273286e-05} +{"train_loss": 0.01641741767525673, "grad_norm": 0.0821177065372467, "global_step": 36245, "epoch": 341, "lr": 2.2695177816848318e-05} +{"train_loss": 0.014679917134344578, "grad_norm": 0.09832433611154556, "global_step": 36246, "epoch": 341, "lr": 2.2692695051360636e-05} +{"train_loss": 0.016257314011454582, "grad_norm": 0.10326797515153885, "global_step": 36247, "epoch": 341, "lr": 2.269021238181893e-05} +{"train_loss": 0.014614257961511612, "grad_norm": 0.08383071422576904, "global_step": 36248, "epoch": 341, "lr": 2.2687729808231966e-05} +{"train_loss": 0.014544053934514523, "grad_norm": 0.05728357285261154, "global_step": 36249, "epoch": 341, "lr": 2.2685247330608417e-05} +{"train_loss": 0.015407281927764416, "grad_norm": 0.08215684443712234, "global_step": 36250, "epoch": 341, "lr": 2.268276494895703e-05} +{"train_loss": 0.015129700276719511, "grad_norm": 0.09082029014825821, "global_step": 36251, "epoch": 341, "lr": 2.2680282663286552e-05, "val_loss": 0.02001619152724743} +{"train_loss": 0.01378270611166954, "grad_norm": 0.08441493660211563, "global_step": 36252, "epoch": 342, "lr": 2.267780047360567e-05} +{"train_loss": 0.017361324280500412, "grad_norm": 0.09123696386814117, "global_step": 36253, "epoch": 342, "lr": 2.2675318379923105e-05} +{"train_loss": 0.01410646177828312, "grad_norm": 0.062075424939394, "global_step": 36254, "epoch": 342, "lr": 2.2672836382247588e-05} +{"train_loss": 0.01575607620179653, "grad_norm": 0.07019923627376556, "global_step": 36255, "epoch": 342, "lr": 2.2670354480587853e-05} +{"train_loss": 0.017794787883758545, "grad_norm": 0.07711257040500641, "global_step": 36256, "epoch": 342, "lr": 2.266787267495259e-05} +{"train_loss": 0.016766058281064034, "grad_norm": 0.1260833889245987, "global_step": 36257, "epoch": 342, "lr": 2.2665390965350535e-05} +{"train_loss": 0.015050167217850685, "grad_norm": 0.11165696382522583, "global_step": 36258, "epoch": 342, "lr": 2.2662909351790446e-05} +{"train_loss": 0.015589124523103237, "grad_norm": 0.06282109022140503, "global_step": 36259, "epoch": 342, "lr": 2.2660427834280966e-05} +{"train_loss": 0.01629241369664669, "grad_norm": 0.0915793925523758, "global_step": 36260, "epoch": 342, "lr": 2.265794641283086e-05} +{"train_loss": 0.015669073909521103, "grad_norm": 0.08877797424793243, "global_step": 36261, "epoch": 342, "lr": 2.265546508744885e-05} +{"train_loss": 0.014959264546632767, "grad_norm": 0.10351165384054184, "global_step": 36262, "epoch": 342, "lr": 2.2652983858143633e-05} +{"train_loss": 0.014574864879250526, "grad_norm": 0.09958435595035553, "global_step": 36263, "epoch": 342, "lr": 2.265050272492395e-05} +{"train_loss": 0.01253918744623661, "grad_norm": 0.07743673026561737, "global_step": 36264, "epoch": 342, "lr": 2.2648021687798493e-05} +{"train_loss": 0.01569506898522377, "grad_norm": 0.10995855927467346, "global_step": 36265, "epoch": 342, "lr": 2.2645540746776005e-05} +{"train_loss": 0.013233551755547523, "grad_norm": 0.09741730242967606, "global_step": 36266, "epoch": 342, "lr": 2.2643059901865176e-05} +{"train_loss": 0.014363138005137444, "grad_norm": 0.07890189439058304, "global_step": 36267, "epoch": 342, "lr": 2.264057915307474e-05} +{"train_loss": 0.01854228973388672, "grad_norm": 0.09108611941337585, "global_step": 36268, "epoch": 342, "lr": 2.263809850041343e-05} +{"train_loss": 0.015780949965119362, "grad_norm": 0.09738245606422424, "global_step": 36269, "epoch": 342, "lr": 2.2635617943889935e-05} +{"train_loss": 0.013364467769861221, "grad_norm": 0.05963863804936409, "global_step": 36270, "epoch": 342, "lr": 2.2633137483512962e-05} +{"train_loss": 0.016281485557556152, "grad_norm": 0.11842639744281769, "global_step": 36271, "epoch": 342, "lr": 2.2630657119291244e-05} +{"train_loss": 0.016733508557081223, "grad_norm": 0.06373777985572815, "global_step": 36272, "epoch": 342, "lr": 2.262817685123351e-05} +{"train_loss": 0.016291704028844833, "grad_norm": 0.11505529284477234, "global_step": 36273, "epoch": 342, "lr": 2.2625696679348445e-05} +{"train_loss": 0.013525226153433323, "grad_norm": 0.07599719613790512, "global_step": 36274, "epoch": 342, "lr": 2.2623216603644777e-05} +{"train_loss": 0.01651683636009693, "grad_norm": 0.07045797258615494, "global_step": 36275, "epoch": 342, "lr": 2.2620736624131257e-05} +{"train_loss": 0.015637822449207306, "grad_norm": 0.0862688422203064, "global_step": 36276, "epoch": 342, "lr": 2.261825674081652e-05} +{"train_loss": 0.016071289777755737, "grad_norm": 0.05898364633321762, "global_step": 36277, "epoch": 342, "lr": 2.2615776953709322e-05} +{"train_loss": 0.01567835733294487, "grad_norm": 0.07661270350217819, "global_step": 36278, "epoch": 342, "lr": 2.2613297262818393e-05} +{"train_loss": 0.014001615345478058, "grad_norm": 0.0811784416437149, "global_step": 36279, "epoch": 342, "lr": 2.2610817668152413e-05} +{"train_loss": 0.013915621675550938, "grad_norm": 0.08555249869823456, "global_step": 36280, "epoch": 342, "lr": 2.2608338169720106e-05} +{"train_loss": 0.013647601939737797, "grad_norm": 0.06786108016967773, "global_step": 36281, "epoch": 342, "lr": 2.2605858767530203e-05} +{"train_loss": 0.013366570696234703, "grad_norm": 0.05778643861413002, "global_step": 36282, "epoch": 342, "lr": 2.26033794615914e-05} +{"train_loss": 0.01409140694886446, "grad_norm": 0.05640978366136551, "global_step": 36283, "epoch": 342, "lr": 2.2600900251912395e-05} +{"train_loss": 0.013184504583477974, "grad_norm": 0.08532882481813431, "global_step": 36284, "epoch": 342, "lr": 2.2598421138501902e-05} +{"train_loss": 0.014472187496721745, "grad_norm": 0.19793552160263062, "global_step": 36285, "epoch": 342, "lr": 2.2595942121368664e-05} +{"train_loss": 0.014363760128617287, "grad_norm": 0.06860500574111938, "global_step": 36286, "epoch": 342, "lr": 2.259346320052134e-05} +{"train_loss": 0.013628936372697353, "grad_norm": 0.06512384116649628, "global_step": 36287, "epoch": 342, "lr": 2.25909843759687e-05} +{"train_loss": 0.016167718917131424, "grad_norm": 0.09808049350976944, "global_step": 36288, "epoch": 342, "lr": 2.2588505647719392e-05} +{"train_loss": 0.012838141992688179, "grad_norm": 0.07162235677242279, "global_step": 36289, "epoch": 342, "lr": 2.258602701578218e-05} +{"train_loss": 0.01469484344124794, "grad_norm": 0.14734888076782227, "global_step": 36290, "epoch": 342, "lr": 2.258354848016572e-05} +{"train_loss": 0.01602315343916416, "grad_norm": 0.08813758939504623, "global_step": 36291, "epoch": 342, "lr": 2.2581070040878753e-05} +{"train_loss": 0.013876667246222496, "grad_norm": 0.06922262161970139, "global_step": 36292, "epoch": 342, "lr": 2.257859169793e-05} +{"train_loss": 0.016716938465833664, "grad_norm": 0.07898604869842529, "global_step": 36293, "epoch": 342, "lr": 2.257611345132815e-05} +{"train_loss": 0.013407869264483452, "grad_norm": 0.07280145585536957, "global_step": 36294, "epoch": 342, "lr": 2.2573635301081896e-05} +{"train_loss": 0.014512765221297741, "grad_norm": 0.10071022063493729, "global_step": 36295, "epoch": 342, "lr": 2.257115724719996e-05} +{"train_loss": 0.013352347537875175, "grad_norm": 0.07674280554056168, "global_step": 36296, "epoch": 342, "lr": 2.2568679289691064e-05} +{"train_loss": 0.015528824180364609, "grad_norm": 0.11155802756547928, "global_step": 36297, "epoch": 342, "lr": 2.2566201428563875e-05} +{"train_loss": 0.01402121689170599, "grad_norm": 0.06661335378885269, "global_step": 36298, "epoch": 342, "lr": 2.2563723663827136e-05} +{"train_loss": 0.012718701735138893, "grad_norm": 0.06489315629005432, "global_step": 36299, "epoch": 342, "lr": 2.2561245995489566e-05} +{"train_loss": 0.016872834414243698, "grad_norm": 0.07924674451351166, "global_step": 36300, "epoch": 342, "lr": 2.255876842355982e-05} +{"train_loss": 0.014151519164443016, "grad_norm": 0.06603612750768661, "global_step": 36301, "epoch": 342, "lr": 2.2556290948046616e-05} +{"train_loss": 0.01224412303417921, "grad_norm": 0.07975124567747116, "global_step": 36302, "epoch": 342, "lr": 2.255381356895869e-05} +{"train_loss": 0.015911350026726723, "grad_norm": 0.1615501046180725, "global_step": 36303, "epoch": 342, "lr": 2.2551336286304713e-05} +{"train_loss": 0.013724219053983688, "grad_norm": 0.08290217071771622, "global_step": 36304, "epoch": 342, "lr": 2.2548859100093407e-05} +{"train_loss": 0.01373311784118414, "grad_norm": 0.1022726520895958, "global_step": 36305, "epoch": 342, "lr": 2.2546382010333482e-05} +{"train_loss": 0.014616959728300571, "grad_norm": 0.16988086700439453, "global_step": 36306, "epoch": 342, "lr": 2.254390501703364e-05} +{"train_loss": 0.015510009601712227, "grad_norm": 0.08943545073270798, "global_step": 36307, "epoch": 342, "lr": 2.254142812020255e-05} +{"train_loss": 0.017951440066099167, "grad_norm": 0.06823917478322983, "global_step": 36308, "epoch": 342, "lr": 2.2538951319848943e-05} +{"train_loss": 0.014728154987096786, "grad_norm": 0.07798614352941513, "global_step": 36309, "epoch": 342, "lr": 2.253647461598154e-05} +{"train_loss": 0.014116477221250534, "grad_norm": 0.06750116497278214, "global_step": 36310, "epoch": 342, "lr": 2.2533998008608998e-05} +{"train_loss": 0.014719415456056595, "grad_norm": 0.06405580788850784, "global_step": 36311, "epoch": 342, "lr": 2.253152149774006e-05} +{"train_loss": 0.015269866213202477, "grad_norm": 0.06541326642036438, "global_step": 36312, "epoch": 342, "lr": 2.2529045083383398e-05} +{"train_loss": 0.017337875440716743, "grad_norm": 0.08819962292909622, "global_step": 36313, "epoch": 342, "lr": 2.252656876554774e-05} +{"train_loss": 0.014436930418014526, "grad_norm": 0.09846861660480499, "global_step": 36314, "epoch": 342, "lr": 2.2524092544241747e-05} +{"train_loss": 0.013595150783658028, "grad_norm": 0.05769668519496918, "global_step": 36315, "epoch": 342, "lr": 2.252161641947415e-05} +{"train_loss": 0.017013613134622574, "grad_norm": 0.07709979265928268, "global_step": 36316, "epoch": 342, "lr": 2.251914039125368e-05} +{"train_loss": 0.015336663462221622, "grad_norm": 0.06017312407493591, "global_step": 36317, "epoch": 342, "lr": 2.251666445958896e-05} +{"train_loss": 0.016053760424256325, "grad_norm": 0.06863642483949661, "global_step": 36318, "epoch": 342, "lr": 2.251418862448873e-05} +{"train_loss": 0.016027482226490974, "grad_norm": 0.08435818552970886, "global_step": 36319, "epoch": 342, "lr": 2.251171288596171e-05} +{"train_loss": 0.014647198840975761, "grad_norm": 0.06988221406936646, "global_step": 36320, "epoch": 342, "lr": 2.250923724401655e-05} +{"train_loss": 0.015088079497218132, "grad_norm": 0.07536269724369049, "global_step": 36321, "epoch": 342, "lr": 2.2506761698661983e-05} +{"train_loss": 0.01482260599732399, "grad_norm": 0.07864139974117279, "global_step": 36322, "epoch": 342, "lr": 2.2504286249906713e-05} +{"train_loss": 0.013367335312068462, "grad_norm": 0.05067911371588707, "global_step": 36323, "epoch": 342, "lr": 2.250181089775943e-05} +{"train_loss": 0.01582905277609825, "grad_norm": 0.14959610998630524, "global_step": 36324, "epoch": 342, "lr": 2.2499335642228796e-05} +{"train_loss": 0.015308476984500885, "grad_norm": 0.08556310087442398, "global_step": 36325, "epoch": 342, "lr": 2.2496860483323544e-05} +{"train_loss": 0.013329830020666122, "grad_norm": 0.1318269670009613, "global_step": 36326, "epoch": 342, "lr": 2.249438542105238e-05} +{"train_loss": 0.017067287117242813, "grad_norm": 0.06949123740196228, "global_step": 36327, "epoch": 342, "lr": 2.2491910455423965e-05} +{"train_loss": 0.016986068338155746, "grad_norm": 0.061730701476335526, "global_step": 36328, "epoch": 342, "lr": 2.2489435586447033e-05} +{"train_loss": 0.016331149265170097, "grad_norm": 0.10999219119548798, "global_step": 36329, "epoch": 342, "lr": 2.2486960814130238e-05} +{"train_loss": 0.015245016664266586, "grad_norm": 0.09438258409500122, "global_step": 36330, "epoch": 342, "lr": 2.2484486138482317e-05} +{"train_loss": 0.01567516103386879, "grad_norm": 0.07213229686021805, "global_step": 36331, "epoch": 342, "lr": 2.2482011559511932e-05} +{"train_loss": 0.015552416443824768, "grad_norm": 0.15879401564598083, "global_step": 36332, "epoch": 342, "lr": 2.247953707722779e-05} +{"train_loss": 0.01521236915141344, "grad_norm": 0.06408081203699112, "global_step": 36333, "epoch": 342, "lr": 2.24770626916386e-05} +{"train_loss": 0.014819134026765823, "grad_norm": 0.09150469303131104, "global_step": 36334, "epoch": 342, "lr": 2.247458840275305e-05} +{"train_loss": 0.012696217745542526, "grad_norm": 0.07553297281265259, "global_step": 36335, "epoch": 342, "lr": 2.2472114210579803e-05} +{"train_loss": 0.016267362982034683, "grad_norm": 0.12678736448287964, "global_step": 36336, "epoch": 342, "lr": 2.246964011512757e-05} +{"train_loss": 0.015533952973783016, "grad_norm": 0.10984650999307632, "global_step": 36337, "epoch": 342, "lr": 2.2467166116405076e-05} +{"train_loss": 0.013574780896306038, "grad_norm": 0.1146872416138649, "global_step": 36338, "epoch": 342, "lr": 2.2464692214420963e-05} +{"train_loss": 0.01688910275697708, "grad_norm": 0.131788969039917, "global_step": 36339, "epoch": 342, "lr": 2.246221840918395e-05} +{"train_loss": 0.015942230820655823, "grad_norm": 0.09159195423126221, "global_step": 36340, "epoch": 342, "lr": 2.2459744700702762e-05} +{"train_loss": 0.019611313939094543, "grad_norm": 0.07442215830087662, "global_step": 36341, "epoch": 342, "lr": 2.2457271088986015e-05} +{"train_loss": 0.016686532646417618, "grad_norm": 0.1096939817070961, "global_step": 36342, "epoch": 342, "lr": 2.245479757404244e-05} +{"train_loss": 0.015375388786196709, "grad_norm": 0.08817175775766373, "global_step": 36343, "epoch": 342, "lr": 2.2452324155880745e-05} +{"train_loss": 0.01779666543006897, "grad_norm": 0.12418016046285629, "global_step": 36344, "epoch": 342, "lr": 2.244985083450959e-05} +{"train_loss": 0.01637093722820282, "grad_norm": 0.10519763082265854, "global_step": 36345, "epoch": 342, "lr": 2.244737760993767e-05} +{"train_loss": 0.01407884806394577, "grad_norm": 0.164717435836792, "global_step": 36346, "epoch": 342, "lr": 2.2444904482173705e-05} +{"train_loss": 0.012498309835791588, "grad_norm": 0.07213398814201355, "global_step": 36347, "epoch": 342, "lr": 2.244243145122636e-05} +{"train_loss": 0.014955373480916023, "grad_norm": 0.07308836281299591, "global_step": 36348, "epoch": 342, "lr": 2.2439958517104308e-05} +{"train_loss": 0.014003023505210876, "grad_norm": 0.1669722944498062, "global_step": 36349, "epoch": 342, "lr": 2.2437485679816262e-05} +{"train_loss": 0.015609810128808022, "grad_norm": 0.12717796862125397, "global_step": 36350, "epoch": 342, "lr": 2.243501293937092e-05} +{"train_loss": 0.01512644998729229, "grad_norm": 0.14087049663066864, "global_step": 36351, "epoch": 342, "lr": 2.2432540295776934e-05} +{"train_loss": 0.014057701453566551, "grad_norm": 0.1530848890542984, "global_step": 36352, "epoch": 342, "lr": 2.2430067749043033e-05} +{"train_loss": 0.01702868938446045, "grad_norm": 0.15319401025772095, "global_step": 36353, "epoch": 342, "lr": 2.2427595299177866e-05} +{"train_loss": 0.013230497017502785, "grad_norm": 0.0691310316324234, "global_step": 36354, "epoch": 342, "lr": 2.2425122946190158e-05} +{"train_loss": 0.01644195429980755, "grad_norm": 0.07418374717235565, "global_step": 36355, "epoch": 342, "lr": 2.2422650690088555e-05} +{"train_loss": 0.01675775833427906, "grad_norm": 0.08390294760465622, "global_step": 36356, "epoch": 342, "lr": 2.2420178530881765e-05} +{"train_loss": 0.015145195981944507, "grad_norm": 0.1188451275229454, "global_step": 36357, "epoch": 342, "lr": 2.241770646857849e-05, "val_loss": 0.01908581331372261} +{"train_loss": 0.014953000470995903, "grad_norm": 0.11827731132507324, "global_step": 36358, "epoch": 343, "lr": 2.2415234503187398e-05} +{"train_loss": 0.01618374139070511, "grad_norm": 0.10410797595977783, "global_step": 36359, "epoch": 343, "lr": 2.241276263471716e-05} +{"train_loss": 0.014631622470915318, "grad_norm": 0.07633958756923676, "global_step": 36360, "epoch": 343, "lr": 2.2410290863176498e-05} +{"train_loss": 0.0139312744140625, "grad_norm": 0.08765453100204468, "global_step": 36361, "epoch": 343, "lr": 2.2407819188574052e-05} +{"train_loss": 0.014656184241175652, "grad_norm": 0.1069936752319336, "global_step": 36362, "epoch": 343, "lr": 2.2405347610918538e-05} +{"train_loss": 0.013451503589749336, "grad_norm": 0.09935451298952103, "global_step": 36363, "epoch": 343, "lr": 2.240287613021864e-05} +{"train_loss": 0.014502027072012424, "grad_norm": 0.0772484838962555, "global_step": 36364, "epoch": 343, "lr": 2.240040474648304e-05} +{"train_loss": 0.01670215278863907, "grad_norm": 0.0668422281742096, "global_step": 36365, "epoch": 343, "lr": 2.2397933459720395e-05} +{"train_loss": 0.015057445503771305, "grad_norm": 0.08238684386014938, "global_step": 36366, "epoch": 343, "lr": 2.239546226993941e-05} +{"train_loss": 0.016828086227178574, "grad_norm": 0.11283731460571289, "global_step": 36367, "epoch": 343, "lr": 2.239299117714878e-05} +{"train_loss": 0.014814828522503376, "grad_norm": 0.10893400758504868, "global_step": 36368, "epoch": 343, "lr": 2.2390520181357154e-05} +{"train_loss": 0.017333826050162315, "grad_norm": 0.10269074887037277, "global_step": 36369, "epoch": 343, "lr": 2.2388049282573258e-05} +{"train_loss": 0.016249697655439377, "grad_norm": 0.10192735493183136, "global_step": 36370, "epoch": 343, "lr": 2.2385578480805724e-05} +{"train_loss": 0.014941053465008736, "grad_norm": 0.09823685884475708, "global_step": 36371, "epoch": 343, "lr": 2.2383107776063277e-05} +{"train_loss": 0.013772612437605858, "grad_norm": 0.07913685590028763, "global_step": 36372, "epoch": 343, "lr": 2.238063716835457e-05} +{"train_loss": 0.014740519225597382, "grad_norm": 0.07400895655155182, "global_step": 36373, "epoch": 343, "lr": 2.2378166657688282e-05} +{"train_loss": 0.014454135671257973, "grad_norm": 0.09599766135215759, "global_step": 36374, "epoch": 343, "lr": 2.2375696244073125e-05} +{"train_loss": 0.014800041913986206, "grad_norm": 0.11184338480234146, "global_step": 36375, "epoch": 343, "lr": 2.2373225927517753e-05} +{"train_loss": 0.017239945009350777, "grad_norm": 0.06900440901517868, "global_step": 36376, "epoch": 343, "lr": 2.2370755708030842e-05} +{"train_loss": 0.016497697681188583, "grad_norm": 0.07568688690662384, "global_step": 36377, "epoch": 343, "lr": 2.236828558562109e-05} +{"train_loss": 0.015325107611715794, "grad_norm": 0.0905463695526123, "global_step": 36378, "epoch": 343, "lr": 2.2365815560297153e-05} +{"train_loss": 0.017205258831381798, "grad_norm": 0.07827319204807281, "global_step": 36379, "epoch": 343, "lr": 2.236334563206772e-05} +{"train_loss": 0.01395352091640234, "grad_norm": 0.05386117100715637, "global_step": 36380, "epoch": 343, "lr": 2.236087580094149e-05} +{"train_loss": 0.016046999022364616, "grad_norm": 0.06545556336641312, "global_step": 36381, "epoch": 343, "lr": 2.2358406066927125e-05} +{"train_loss": 0.018451901152729988, "grad_norm": 0.08068463206291199, "global_step": 36382, "epoch": 343, "lr": 2.235593643003328e-05} +{"train_loss": 0.016748547554016113, "grad_norm": 0.10065130144357681, "global_step": 36383, "epoch": 343, "lr": 2.235346689026865e-05} +{"train_loss": 0.014209738001227379, "grad_norm": 0.09393176436424255, "global_step": 36384, "epoch": 343, "lr": 2.2350997447641935e-05} +{"train_loss": 0.015590330585837364, "grad_norm": 0.07039874792098999, "global_step": 36385, "epoch": 343, "lr": 2.234852810216177e-05} +{"train_loss": 0.0159379281103611, "grad_norm": 0.07114125043153763, "global_step": 36386, "epoch": 343, "lr": 2.2346058853836856e-05} +{"train_loss": 0.01668507605791092, "grad_norm": 0.10967732220888138, "global_step": 36387, "epoch": 343, "lr": 2.2343589702675877e-05} +{"train_loss": 0.014246340841054916, "grad_norm": 0.0687645822763443, "global_step": 36388, "epoch": 343, "lr": 2.23411206486875e-05} +{"train_loss": 0.01551339216530323, "grad_norm": 0.1998254805803299, "global_step": 36389, "epoch": 343, "lr": 2.233865169188038e-05} +{"train_loss": 0.015929410234093666, "grad_norm": 0.05545554682612419, "global_step": 36390, "epoch": 343, "lr": 2.233618283226321e-05} +{"train_loss": 0.011938555166125298, "grad_norm": 0.12280307710170746, "global_step": 36391, "epoch": 343, "lr": 2.2333714069844674e-05} +{"train_loss": 0.015150496736168861, "grad_norm": 0.14362648129463196, "global_step": 36392, "epoch": 343, "lr": 2.2331245404633423e-05} +{"train_loss": 0.016101446002721786, "grad_norm": 0.07533241808414459, "global_step": 36393, "epoch": 343, "lr": 2.2328776836638155e-05} +{"train_loss": 0.0164668969810009, "grad_norm": 0.05747291073203087, "global_step": 36394, "epoch": 343, "lr": 2.232630836586752e-05} +{"train_loss": 0.013462206348776817, "grad_norm": 0.08482464402914047, "global_step": 36395, "epoch": 343, "lr": 2.2323839992330216e-05} +{"train_loss": 0.01533475425094366, "grad_norm": 0.12448615580797195, "global_step": 36396, "epoch": 343, "lr": 2.2321371716034884e-05} +{"train_loss": 0.011839495971798897, "grad_norm": 0.062233299016952515, "global_step": 36397, "epoch": 343, "lr": 2.2318903536990217e-05} +{"train_loss": 0.01538563147187233, "grad_norm": 0.07245799154043198, "global_step": 36398, "epoch": 343, "lr": 2.2316435455204902e-05} +{"train_loss": 0.014238579198718071, "grad_norm": 0.12339188903570175, "global_step": 36399, "epoch": 343, "lr": 2.2313967470687593e-05} +{"train_loss": 0.014187959022819996, "grad_norm": 0.14240485429763794, "global_step": 36400, "epoch": 343, "lr": 2.231149958344694e-05} +{"train_loss": 0.01651947945356369, "grad_norm": 0.08606407791376114, "global_step": 36401, "epoch": 343, "lr": 2.2309031793491663e-05} +{"train_loss": 0.013778558000922203, "grad_norm": 0.07786815613508224, "global_step": 36402, "epoch": 343, "lr": 2.2306564100830384e-05} +{"train_loss": 0.013563893735408783, "grad_norm": 0.07825291156768799, "global_step": 36403, "epoch": 343, "lr": 2.2304096505471795e-05} +{"train_loss": 0.017299987375736237, "grad_norm": 0.09337820112705231, "global_step": 36404, "epoch": 343, "lr": 2.2301629007424585e-05} +{"train_loss": 0.01433804165571928, "grad_norm": 0.07388272136449814, "global_step": 36405, "epoch": 343, "lr": 2.22991616066974e-05} +{"train_loss": 0.014966515824198723, "grad_norm": 0.13023902475833893, "global_step": 36406, "epoch": 343, "lr": 2.22966943032989e-05} +{"train_loss": 0.013505464419722557, "grad_norm": 0.1236424595117569, "global_step": 36407, "epoch": 343, "lr": 2.2294227097237768e-05} +{"train_loss": 0.014643622562289238, "grad_norm": 0.13590924441814423, "global_step": 36408, "epoch": 343, "lr": 2.2291759988522693e-05} +{"train_loss": 0.014810601249337196, "grad_norm": 0.0926278606057167, "global_step": 36409, "epoch": 343, "lr": 2.2289292977162302e-05} +{"train_loss": 0.012477297335863113, "grad_norm": 0.06593013554811478, "global_step": 36410, "epoch": 343, "lr": 2.228682606316529e-05} +{"train_loss": 0.014129573479294777, "grad_norm": 0.06085589528083801, "global_step": 36411, "epoch": 343, "lr": 2.228435924654033e-05} +{"train_loss": 0.014174532145261765, "grad_norm": 0.11216713488101959, "global_step": 36412, "epoch": 343, "lr": 2.2281892527296084e-05} +{"train_loss": 0.014537467621266842, "grad_norm": 0.0723777785897255, "global_step": 36413, "epoch": 343, "lr": 2.2279425905441192e-05} +{"train_loss": 0.014729170128703117, "grad_norm": 0.1745576411485672, "global_step": 36414, "epoch": 343, "lr": 2.2276959380984342e-05} +{"train_loss": 0.015757959336042404, "grad_norm": 0.08282852172851562, "global_step": 36415, "epoch": 343, "lr": 2.2274492953934224e-05} +{"train_loss": 0.014000598341226578, "grad_norm": 0.13011613488197327, "global_step": 36416, "epoch": 343, "lr": 2.2272026624299475e-05} +{"train_loss": 0.017846064642071724, "grad_norm": 0.10336866974830627, "global_step": 36417, "epoch": 343, "lr": 2.2269560392088744e-05} +{"train_loss": 0.01569506525993347, "grad_norm": 0.07163701951503754, "global_step": 36418, "epoch": 343, "lr": 2.2267094257310743e-05} +{"train_loss": 0.014182874001562595, "grad_norm": 0.0678725391626358, "global_step": 36419, "epoch": 343, "lr": 2.2264628219974094e-05} +{"train_loss": 0.015987182036042213, "grad_norm": 0.10696760565042496, "global_step": 36420, "epoch": 343, "lr": 2.2262162280087474e-05} +{"train_loss": 0.016843905672430992, "grad_norm": 0.09062401950359344, "global_step": 36421, "epoch": 343, "lr": 2.225969643765957e-05} +{"train_loss": 0.015175660140812397, "grad_norm": 0.060303695499897, "global_step": 36422, "epoch": 343, "lr": 2.2257230692699038e-05} +{"train_loss": 0.014008205384016037, "grad_norm": 0.08209860324859619, "global_step": 36423, "epoch": 343, "lr": 2.225476504521451e-05} +{"train_loss": 0.01248080562800169, "grad_norm": 0.08599162101745605, "global_step": 36424, "epoch": 343, "lr": 2.225229949521467e-05} +{"train_loss": 0.014611438848078251, "grad_norm": 0.06489346921443939, "global_step": 36425, "epoch": 343, "lr": 2.2249834042708202e-05} +{"train_loss": 0.017274893820285797, "grad_norm": 0.09006524085998535, "global_step": 36426, "epoch": 343, "lr": 2.2247368687703728e-05} +{"train_loss": 0.015500081703066826, "grad_norm": 0.07589573413133621, "global_step": 36427, "epoch": 343, "lr": 2.2244903430209934e-05} +{"train_loss": 0.015332236886024475, "grad_norm": 0.10330566763877869, "global_step": 36428, "epoch": 343, "lr": 2.2242438270235494e-05} +{"train_loss": 0.013859385624527931, "grad_norm": 0.07415271550416946, "global_step": 36429, "epoch": 343, "lr": 2.223997320778906e-05} +{"train_loss": 0.017827492207288742, "grad_norm": 0.08763965219259262, "global_step": 36430, "epoch": 343, "lr": 2.2237508242879268e-05} +{"train_loss": 0.015087259002029896, "grad_norm": 0.1691325604915619, "global_step": 36431, "epoch": 343, "lr": 2.2235043375514792e-05} +{"train_loss": 0.015518895350396633, "grad_norm": 0.06995905190706253, "global_step": 36432, "epoch": 343, "lr": 2.2232578605704324e-05} +{"train_loss": 0.01605401560664177, "grad_norm": 0.09106554090976715, "global_step": 36433, "epoch": 343, "lr": 2.2230113933456483e-05} +{"train_loss": 0.013945363461971283, "grad_norm": 0.059335313737392426, "global_step": 36434, "epoch": 343, "lr": 2.2227649358779956e-05} +{"train_loss": 0.013420145027339458, "grad_norm": 0.07750501483678818, "global_step": 36435, "epoch": 343, "lr": 2.2225184881683393e-05} +{"train_loss": 0.016310349106788635, "grad_norm": 0.06558512151241302, "global_step": 36436, "epoch": 343, "lr": 2.222272050217543e-05} +{"train_loss": 0.015551324002444744, "grad_norm": 0.07913364470005035, "global_step": 36437, "epoch": 343, "lr": 2.2220256220264756e-05} +{"train_loss": 0.01502656564116478, "grad_norm": 0.09471128135919571, "global_step": 36438, "epoch": 343, "lr": 2.2217792035960015e-05} +{"train_loss": 0.0157833993434906, "grad_norm": 0.05825820565223694, "global_step": 36439, "epoch": 343, "lr": 2.2215327949269886e-05} +{"train_loss": 0.015108522959053516, "grad_norm": 0.1461046189069748, "global_step": 36440, "epoch": 343, "lr": 2.2212863960203018e-05} +{"train_loss": 0.016127433627843857, "grad_norm": 0.058856453746557236, "global_step": 36441, "epoch": 343, "lr": 2.221040006876804e-05} +{"train_loss": 0.016869090497493744, "grad_norm": 0.11599476635456085, "global_step": 36442, "epoch": 343, "lr": 2.220793627497365e-05} +{"train_loss": 0.013851117342710495, "grad_norm": 0.07281984388828278, "global_step": 36443, "epoch": 343, "lr": 2.2205472578828468e-05} +{"train_loss": 0.0151580311357975, "grad_norm": 0.15037845075130463, "global_step": 36444, "epoch": 343, "lr": 2.2203008980341167e-05} +{"train_loss": 0.01579219102859497, "grad_norm": 0.15526896715164185, "global_step": 36445, "epoch": 343, "lr": 2.2200545479520424e-05} +{"train_loss": 0.012465467676520348, "grad_norm": 0.0876501277089119, "global_step": 36446, "epoch": 343, "lr": 2.219808207637487e-05} +{"train_loss": 0.014447538182139397, "grad_norm": 0.08509044349193573, "global_step": 36447, "epoch": 343, "lr": 2.2195618770913153e-05} +{"train_loss": 0.014483990147709846, "grad_norm": 0.0961754247546196, "global_step": 36448, "epoch": 343, "lr": 2.219315556314393e-05} +{"train_loss": 0.013225103728473186, "grad_norm": 0.05974894389510155, "global_step": 36449, "epoch": 343, "lr": 2.219069245307589e-05} +{"train_loss": 0.014650302939116955, "grad_norm": 0.07570905238389969, "global_step": 36450, "epoch": 343, "lr": 2.2188229440717644e-05} +{"train_loss": 0.01335536316037178, "grad_norm": 0.09121450036764145, "global_step": 36451, "epoch": 343, "lr": 2.2185766526077866e-05} +{"train_loss": 0.017462652176618576, "grad_norm": 0.18169286847114563, "global_step": 36452, "epoch": 343, "lr": 2.2183303709165226e-05} +{"train_loss": 0.014827501028776169, "grad_norm": 0.15199725329875946, "global_step": 36453, "epoch": 343, "lr": 2.218084098998836e-05} +{"train_loss": 0.014734371565282345, "grad_norm": 0.09380485862493515, "global_step": 36454, "epoch": 343, "lr": 2.2178378368555898e-05} +{"train_loss": 0.013612570241093636, "grad_norm": 0.06394310295581818, "global_step": 36455, "epoch": 343, "lr": 2.217591584487652e-05} +{"train_loss": 0.016591154038906097, "grad_norm": 0.09770212322473526, "global_step": 36456, "epoch": 343, "lr": 2.217345341895889e-05} +{"train_loss": 0.015463298186659813, "grad_norm": 0.09247808903455734, "global_step": 36457, "epoch": 343, "lr": 2.2170991090811626e-05} +{"train_loss": 0.018927861005067825, "grad_norm": 0.18240736424922943, "global_step": 36458, "epoch": 343, "lr": 2.216852886044341e-05} +{"train_loss": 0.015280014835298061, "grad_norm": 0.1111711859703064, "global_step": 36459, "epoch": 343, "lr": 2.2166066727862887e-05} +{"train_loss": 0.01644652709364891, "grad_norm": 0.07612905651330948, "global_step": 36460, "epoch": 343, "lr": 2.2163604693078678e-05} +{"train_loss": 0.014275392517447472, "grad_norm": 0.0632704496383667, "global_step": 36461, "epoch": 343, "lr": 2.2161142756099456e-05} +{"train_loss": 0.015871550887823105, "grad_norm": 0.07747769355773926, "global_step": 36462, "epoch": 343, "lr": 2.2158680916933894e-05} +{"train_loss": 0.015125761326964735, "grad_norm": 0.07546570152044296, "global_step": 36463, "epoch": 343, "lr": 2.215621917559062e-05, "val_loss": 0.019488511607050896} +{"train_loss": 0.0162726491689682, "grad_norm": 0.07483277469873428, "global_step": 36464, "epoch": 344, "lr": 2.215375753207826e-05} +{"train_loss": 0.015832237899303436, "grad_norm": 0.07109806686639786, "global_step": 36465, "epoch": 344, "lr": 2.2151295986405485e-05} +{"train_loss": 0.014384491369128227, "grad_norm": 0.09128770977258682, "global_step": 36466, "epoch": 344, "lr": 2.214883453858097e-05} +{"train_loss": 0.014087062329053879, "grad_norm": 0.07197223603725433, "global_step": 36467, "epoch": 344, "lr": 2.214637318861331e-05} +{"train_loss": 0.016440626233816147, "grad_norm": 0.05461167171597481, "global_step": 36468, "epoch": 344, "lr": 2.214391193651119e-05} +{"train_loss": 0.01349577121436596, "grad_norm": 0.12465719878673553, "global_step": 36469, "epoch": 344, "lr": 2.214145078228327e-05} +{"train_loss": 0.01570592075586319, "grad_norm": 0.1348496824502945, "global_step": 36470, "epoch": 344, "lr": 2.213898972593817e-05} +{"train_loss": 0.015339763835072517, "grad_norm": 0.08021198213100433, "global_step": 36471, "epoch": 344, "lr": 2.2136528767484523e-05} +{"train_loss": 0.014995994046330452, "grad_norm": 0.11585640907287598, "global_step": 36472, "epoch": 344, "lr": 2.2134067906931e-05} +{"train_loss": 0.01572491228580475, "grad_norm": 0.054608263075351715, "global_step": 36473, "epoch": 344, "lr": 2.2131607144286265e-05} +{"train_loss": 0.015730276703834534, "grad_norm": 0.06548818945884705, "global_step": 36474, "epoch": 344, "lr": 2.212914647955892e-05} +{"train_loss": 0.013981202617287636, "grad_norm": 0.19702082872390747, "global_step": 36475, "epoch": 344, "lr": 2.2126685912757654e-05} +{"train_loss": 0.014162094332277775, "grad_norm": 0.0761537179350853, "global_step": 36476, "epoch": 344, "lr": 2.2124225443891088e-05} +{"train_loss": 0.015748638659715652, "grad_norm": 0.07022584974765778, "global_step": 36477, "epoch": 344, "lr": 2.212176507296786e-05} +{"train_loss": 0.014065064489841461, "grad_norm": 0.05913301184773445, "global_step": 36478, "epoch": 344, "lr": 2.211930479999662e-05} +{"train_loss": 0.013908836990594864, "grad_norm": 0.05056339502334595, "global_step": 36479, "epoch": 344, "lr": 2.2116844624986037e-05} +{"train_loss": 0.013684220612049103, "grad_norm": 0.11805769801139832, "global_step": 36480, "epoch": 344, "lr": 2.211438454794472e-05} +{"train_loss": 0.013630174100399017, "grad_norm": 0.0887000784277916, "global_step": 36481, "epoch": 344, "lr": 2.2111924568881347e-05} +{"train_loss": 0.014053610153496265, "grad_norm": 0.08485615998506546, "global_step": 36482, "epoch": 344, "lr": 2.210946468780452e-05} +{"train_loss": 0.014876088127493858, "grad_norm": 0.134008526802063, "global_step": 36483, "epoch": 344, "lr": 2.210700490472293e-05} +{"train_loss": 0.016419745981693268, "grad_norm": 0.07293872535228729, "global_step": 36484, "epoch": 344, "lr": 2.210454521964517e-05} +{"train_loss": 0.017469462007284164, "grad_norm": 0.07786721736192703, "global_step": 36485, "epoch": 344, "lr": 2.210208563257991e-05} +{"train_loss": 0.018356790766119957, "grad_norm": 0.15507005155086517, "global_step": 36486, "epoch": 344, "lr": 2.2099626143535807e-05} +{"train_loss": 0.015149224549531937, "grad_norm": 0.06598775833845139, "global_step": 36487, "epoch": 344, "lr": 2.209716675252148e-05} +{"train_loss": 0.014620468020439148, "grad_norm": 0.08198723942041397, "global_step": 36488, "epoch": 344, "lr": 2.209470745954556e-05} +{"train_loss": 0.01421106606721878, "grad_norm": 0.05886504799127579, "global_step": 36489, "epoch": 344, "lr": 2.20922482646167e-05} +{"train_loss": 0.017027586698532104, "grad_norm": 0.08275993913412094, "global_step": 36490, "epoch": 344, "lr": 2.2089789167743562e-05} +{"train_loss": 0.015102509409189224, "grad_norm": 0.11822574585676193, "global_step": 36491, "epoch": 344, "lr": 2.208733016893475e-05} +{"train_loss": 0.013368142768740654, "grad_norm": 0.06767968088388443, "global_step": 36492, "epoch": 344, "lr": 2.208487126819892e-05} +{"train_loss": 0.014661389403045177, "grad_norm": 0.05708766356110573, "global_step": 36493, "epoch": 344, "lr": 2.2082412465544734e-05} +{"train_loss": 0.012289220467209816, "grad_norm": 0.08909361809492111, "global_step": 36494, "epoch": 344, "lr": 2.2079953760980803e-05} +{"train_loss": 0.01197333075106144, "grad_norm": 0.0845368281006813, "global_step": 36495, "epoch": 344, "lr": 2.207749515451576e-05} +{"train_loss": 0.014392506331205368, "grad_norm": 0.05411749333143234, "global_step": 36496, "epoch": 344, "lr": 2.2075036646158253e-05} +{"train_loss": 0.016394371166825294, "grad_norm": 0.08414909243583679, "global_step": 36497, "epoch": 344, "lr": 2.207257823591694e-05} +{"train_loss": 0.015326859429478645, "grad_norm": 0.12319013476371765, "global_step": 36498, "epoch": 344, "lr": 2.2070119923800427e-05} +{"train_loss": 0.016049165278673172, "grad_norm": 0.07761715352535248, "global_step": 36499, "epoch": 344, "lr": 2.2067661709817383e-05} +{"train_loss": 0.015252111479640007, "grad_norm": 0.06083107367157936, "global_step": 36500, "epoch": 344, "lr": 2.206520359397643e-05} +{"train_loss": 0.015078970231115818, "grad_norm": 0.11381667852401733, "global_step": 36501, "epoch": 344, "lr": 2.2062745576286182e-05} +{"train_loss": 0.016578149050474167, "grad_norm": 0.07171650975942612, "global_step": 36502, "epoch": 344, "lr": 2.2060287656755297e-05} +{"train_loss": 0.014055407606065273, "grad_norm": 0.07762530446052551, "global_step": 36503, "epoch": 344, "lr": 2.205782983539243e-05} +{"train_loss": 0.014021005481481552, "grad_norm": 0.11391344666481018, "global_step": 36504, "epoch": 344, "lr": 2.2055372112206176e-05} +{"train_loss": 0.017794419080018997, "grad_norm": 0.14365530014038086, "global_step": 36505, "epoch": 344, "lr": 2.205291448720521e-05} +{"train_loss": 0.013643995858728886, "grad_norm": 0.06975571811199188, "global_step": 36506, "epoch": 344, "lr": 2.205045696039813e-05} +{"train_loss": 0.015086607076227665, "grad_norm": 0.07693374902009964, "global_step": 36507, "epoch": 344, "lr": 2.204799953179361e-05} +{"train_loss": 0.01625126600265503, "grad_norm": 0.07490070909261703, "global_step": 36508, "epoch": 344, "lr": 2.204554220140025e-05} +{"train_loss": 0.016011860221624374, "grad_norm": 0.07301170378923416, "global_step": 36509, "epoch": 344, "lr": 2.204308496922669e-05} +{"train_loss": 0.0170260239392519, "grad_norm": 0.05816720798611641, "global_step": 36510, "epoch": 344, "lr": 2.2040627835281593e-05} +{"train_loss": 0.01195419579744339, "grad_norm": 0.09733854234218597, "global_step": 36511, "epoch": 344, "lr": 2.2038170799573572e-05} +{"train_loss": 0.01857679709792137, "grad_norm": 0.11283189058303833, "global_step": 36512, "epoch": 344, "lr": 2.2035713862111234e-05} +{"train_loss": 0.014475742354989052, "grad_norm": 0.05860559642314911, "global_step": 36513, "epoch": 344, "lr": 2.2033257022903243e-05} +{"train_loss": 0.01768869161605835, "grad_norm": 0.06920583546161652, "global_step": 36514, "epoch": 344, "lr": 2.2030800281958236e-05} +{"train_loss": 0.014809411950409412, "grad_norm": 0.10949669778347015, "global_step": 36515, "epoch": 344, "lr": 2.202834363928482e-05} +{"train_loss": 0.013517187908291817, "grad_norm": 0.05877465009689331, "global_step": 36516, "epoch": 344, "lr": 2.2025887094891657e-05} +{"train_loss": 0.015583477914333344, "grad_norm": 0.07676136493682861, "global_step": 36517, "epoch": 344, "lr": 2.2023430648787365e-05} +{"train_loss": 0.01367130596190691, "grad_norm": 0.10746495425701141, "global_step": 36518, "epoch": 344, "lr": 2.2020974300980547e-05} +{"train_loss": 0.015178998932242393, "grad_norm": 0.10290028899908066, "global_step": 36519, "epoch": 344, "lr": 2.201851805147986e-05} +{"train_loss": 0.01518290489912033, "grad_norm": 0.10519774258136749, "global_step": 36520, "epoch": 344, "lr": 2.201606190029395e-05} +{"train_loss": 0.015863973647356033, "grad_norm": 0.06780881434679031, "global_step": 36521, "epoch": 344, "lr": 2.2013605847431408e-05} +{"train_loss": 0.015452033840119839, "grad_norm": 0.1380089819431305, "global_step": 36522, "epoch": 344, "lr": 2.2011149892900907e-05} +{"train_loss": 0.013275181874632835, "grad_norm": 0.09308169782161713, "global_step": 36523, "epoch": 344, "lr": 2.200869403671103e-05} +{"train_loss": 0.016621846705675125, "grad_norm": 0.10512572526931763, "global_step": 36524, "epoch": 344, "lr": 2.200623827887045e-05} +{"train_loss": 0.013250688090920448, "grad_norm": 0.078209787607193, "global_step": 36525, "epoch": 344, "lr": 2.200378261938776e-05} +{"train_loss": 0.016390914097428322, "grad_norm": 0.09790019690990448, "global_step": 36526, "epoch": 344, "lr": 2.2001327058271594e-05} +{"train_loss": 0.014608414843678474, "grad_norm": 0.07757815718650818, "global_step": 36527, "epoch": 344, "lr": 2.199887159553061e-05} +{"train_loss": 0.014655804261565208, "grad_norm": 0.11053203046321869, "global_step": 36528, "epoch": 344, "lr": 2.1996416231173417e-05} +{"train_loss": 0.015385748818516731, "grad_norm": 0.06278578191995621, "global_step": 36529, "epoch": 344, "lr": 2.1993960965208616e-05} +{"train_loss": 0.014994421042501926, "grad_norm": 0.13798096776008606, "global_step": 36530, "epoch": 344, "lr": 2.1991505797644863e-05} +{"train_loss": 0.014929837547242641, "grad_norm": 0.11446034908294678, "global_step": 36531, "epoch": 344, "lr": 2.198905072849079e-05} +{"train_loss": 0.01361567247658968, "grad_norm": 0.04652911797165871, "global_step": 36532, "epoch": 344, "lr": 2.1986595757754997e-05} +{"train_loss": 0.016031714156270027, "grad_norm": 0.0768292248249054, "global_step": 36533, "epoch": 344, "lr": 2.198414088544612e-05} +{"train_loss": 0.016086608171463013, "grad_norm": 0.07908932864665985, "global_step": 36534, "epoch": 344, "lr": 2.198168611157283e-05} +{"train_loss": 0.016942350193858147, "grad_norm": 0.07907763123512268, "global_step": 36535, "epoch": 344, "lr": 2.197923143614367e-05} +{"train_loss": 0.013802168890833855, "grad_norm": 0.06265457719564438, "global_step": 36536, "epoch": 344, "lr": 2.1976776859167304e-05} +{"train_loss": 0.01537402719259262, "grad_norm": 0.09312079101800919, "global_step": 36537, "epoch": 344, "lr": 2.197432238065238e-05} +{"train_loss": 0.015577423386275768, "grad_norm": 0.10727965831756592, "global_step": 36538, "epoch": 344, "lr": 2.1971868000607476e-05} +{"train_loss": 0.01746891438961029, "grad_norm": 0.10514029115438461, "global_step": 36539, "epoch": 344, "lr": 2.1969413719041237e-05} +{"train_loss": 0.016019128262996674, "grad_norm": 0.06993843615055084, "global_step": 36540, "epoch": 344, "lr": 2.1966959535962313e-05} +{"train_loss": 0.015828337520360947, "grad_norm": 0.08448553830385208, "global_step": 36541, "epoch": 344, "lr": 2.1964505451379297e-05} +{"train_loss": 0.013558033853769302, "grad_norm": 0.09659382700920105, "global_step": 36542, "epoch": 344, "lr": 2.19620514653008e-05} +{"train_loss": 0.014124128967523575, "grad_norm": 0.06150691211223602, "global_step": 36543, "epoch": 344, "lr": 2.1959597577735465e-05} +{"train_loss": 0.014820979908108711, "grad_norm": 0.08163490146398544, "global_step": 36544, "epoch": 344, "lr": 2.195714378869192e-05} +{"train_loss": 0.01674124039709568, "grad_norm": 0.16230818629264832, "global_step": 36545, "epoch": 344, "lr": 2.195469009817876e-05} +{"train_loss": 0.015468316152691841, "grad_norm": 0.11379426717758179, "global_step": 36546, "epoch": 344, "lr": 2.1952236506204645e-05} +{"train_loss": 0.016198398545384407, "grad_norm": 0.07611827552318573, "global_step": 36547, "epoch": 344, "lr": 2.194978301277815e-05} +{"train_loss": 0.014751340262591839, "grad_norm": 0.06109223514795303, "global_step": 36548, "epoch": 344, "lr": 2.194732961790794e-05} +{"train_loss": 0.014640181325376034, "grad_norm": 0.062430690973997116, "global_step": 36549, "epoch": 344, "lr": 2.1944876321602592e-05} +{"train_loss": 0.01672104187309742, "grad_norm": 0.06113337725400925, "global_step": 36550, "epoch": 344, "lr": 2.1942423123870754e-05} +{"train_loss": 0.013401607051491737, "grad_norm": 0.06509875506162643, "global_step": 36551, "epoch": 344, "lr": 2.193997002472105e-05} +{"train_loss": 0.014726260676980019, "grad_norm": 0.15472841262817383, "global_step": 36552, "epoch": 344, "lr": 2.1937517024162097e-05} +{"train_loss": 0.013873834162950516, "grad_norm": 0.0968579351902008, "global_step": 36553, "epoch": 344, "lr": 2.193506412220248e-05} +{"train_loss": 0.012218491174280643, "grad_norm": 0.12008510529994965, "global_step": 36554, "epoch": 344, "lr": 2.193261131885085e-05} +{"train_loss": 0.015868892893195152, "grad_norm": 0.06279026716947556, "global_step": 36555, "epoch": 344, "lr": 2.1930158614115825e-05} +{"train_loss": 0.015059078112244606, "grad_norm": 0.11928024142980576, "global_step": 36556, "epoch": 344, "lr": 2.1927706008006006e-05} +{"train_loss": 0.014664258807897568, "grad_norm": 0.12383582442998886, "global_step": 36557, "epoch": 344, "lr": 2.1925253500530013e-05} +{"train_loss": 0.016543971374630928, "grad_norm": 0.11467426270246506, "global_step": 36558, "epoch": 344, "lr": 2.1922801091696516e-05} +{"train_loss": 0.017900500446558, "grad_norm": 0.10113334655761719, "global_step": 36559, "epoch": 344, "lr": 2.192034878151404e-05} +{"train_loss": 0.014305038377642632, "grad_norm": 0.08056779205799103, "global_step": 36560, "epoch": 344, "lr": 2.1917896569991247e-05} +{"train_loss": 0.015529322437942028, "grad_norm": 0.08583662658929825, "global_step": 36561, "epoch": 344, "lr": 2.1915444457136774e-05} +{"train_loss": 0.015819592401385307, "grad_norm": 0.09450475871562958, "global_step": 36562, "epoch": 344, "lr": 2.19129924429592e-05} +{"train_loss": 0.015606453642249107, "grad_norm": 0.08175428211688995, "global_step": 36563, "epoch": 344, "lr": 2.191054052746717e-05} +{"train_loss": 0.014668110758066177, "grad_norm": 0.08121434599161148, "global_step": 36564, "epoch": 344, "lr": 2.1908088710669267e-05} +{"train_loss": 0.01408166904002428, "grad_norm": 0.13243286311626434, "global_step": 36565, "epoch": 344, "lr": 2.1905636992574147e-05} +{"train_loss": 0.01725882850587368, "grad_norm": 0.09647565335035324, "global_step": 36566, "epoch": 344, "lr": 2.1903185373190377e-05} +{"train_loss": 0.01616203971207142, "grad_norm": 0.06263578683137894, "global_step": 36567, "epoch": 344, "lr": 2.1900733852526595e-05} +{"train_loss": 0.01327127031981945, "grad_norm": 0.09089039266109467, "global_step": 36568, "epoch": 344, "lr": 2.1898282430591432e-05} +{"train_loss": 0.015103834689998964, "grad_norm": 0.1459914743900299, "global_step": 36569, "epoch": 344, "lr": 2.1895831107393484e-05, "val_loss": 0.019354138523340225} +{"train_loss": 0.01659884676337242, "grad_norm": 0.06507926434278488, "global_step": 36570, "epoch": 345, "lr": 2.1893379882941346e-05} +{"train_loss": 0.013499213382601738, "grad_norm": 0.12506835162639618, "global_step": 36571, "epoch": 345, "lr": 2.189092875724365e-05} +{"train_loss": 0.015726076439023018, "grad_norm": 0.055218156427145004, "global_step": 36572, "epoch": 345, "lr": 2.188847773030902e-05} +{"train_loss": 0.012609554454684258, "grad_norm": 0.06059325486421585, "global_step": 36573, "epoch": 345, "lr": 2.1886026802146037e-05} +{"train_loss": 0.016333015635609627, "grad_norm": 0.11206474155187607, "global_step": 36574, "epoch": 345, "lr": 2.1883575972763333e-05} +{"train_loss": 0.013598423451185226, "grad_norm": 0.1561206579208374, "global_step": 36575, "epoch": 345, "lr": 2.1881125242169547e-05} +{"train_loss": 0.013901129364967346, "grad_norm": 0.0740811675786972, "global_step": 36576, "epoch": 345, "lr": 2.1878674610373222e-05} +{"train_loss": 0.015828557312488556, "grad_norm": 0.07621922343969345, "global_step": 36577, "epoch": 345, "lr": 2.1876224077383007e-05} +{"train_loss": 0.017701365053653717, "grad_norm": 0.07670637220144272, "global_step": 36578, "epoch": 345, "lr": 2.187377364320753e-05} +{"train_loss": 0.01707332208752632, "grad_norm": 0.08960051089525223, "global_step": 36579, "epoch": 345, "lr": 2.1871323307855356e-05} +{"train_loss": 0.012172202579677105, "grad_norm": 0.07446803897619247, "global_step": 36580, "epoch": 345, "lr": 2.1868873071335122e-05} +{"train_loss": 0.013911687768995762, "grad_norm": 0.0742233470082283, "global_step": 36581, "epoch": 345, "lr": 2.1866422933655457e-05} +{"train_loss": 0.014908449724316597, "grad_norm": 0.07159121334552765, "global_step": 36582, "epoch": 345, "lr": 2.186397289482494e-05} +{"train_loss": 0.016750281676650047, "grad_norm": 0.058462850749492645, "global_step": 36583, "epoch": 345, "lr": 2.1861522954852172e-05} +{"train_loss": 0.015472177416086197, "grad_norm": 0.06750395148992538, "global_step": 36584, "epoch": 345, "lr": 2.1859073113745776e-05} +{"train_loss": 0.014544899575412273, "grad_norm": 0.061170682311058044, "global_step": 36585, "epoch": 345, "lr": 2.1856623371514373e-05} +{"train_loss": 0.015340502373874187, "grad_norm": 0.08447790145874023, "global_step": 36586, "epoch": 345, "lr": 2.1854173728166545e-05} +{"train_loss": 0.014519944787025452, "grad_norm": 0.0853927806019783, "global_step": 36587, "epoch": 345, "lr": 2.185172418371092e-05} +{"train_loss": 0.015026319772005081, "grad_norm": 0.14288479089736938, "global_step": 36588, "epoch": 345, "lr": 2.1849274738156084e-05} +{"train_loss": 0.01502988487482071, "grad_norm": 0.08624505996704102, "global_step": 36589, "epoch": 345, "lr": 2.1846825391510667e-05} +{"train_loss": 0.014600452035665512, "grad_norm": 0.06746885180473328, "global_step": 36590, "epoch": 345, "lr": 2.1844376143783245e-05} +{"train_loss": 0.01695205084979534, "grad_norm": 0.07124518603086472, "global_step": 36591, "epoch": 345, "lr": 2.1841926994982443e-05} +{"train_loss": 0.016695253551006317, "grad_norm": 0.05661728233098984, "global_step": 36592, "epoch": 345, "lr": 2.1839477945116883e-05} +{"train_loss": 0.016538482159376144, "grad_norm": 0.06501226127147675, "global_step": 36593, "epoch": 345, "lr": 2.1837028994195148e-05} +{"train_loss": 0.01595783419907093, "grad_norm": 0.11144398897886276, "global_step": 36594, "epoch": 345, "lr": 2.1834580142225825e-05} +{"train_loss": 0.016652431339025497, "grad_norm": 0.08701622486114502, "global_step": 36595, "epoch": 345, "lr": 2.183213138921754e-05} +{"train_loss": 0.013943844474852085, "grad_norm": 0.06970295310020447, "global_step": 36596, "epoch": 345, "lr": 2.1829682735178912e-05} +{"train_loss": 0.015183614566922188, "grad_norm": 0.10082082450389862, "global_step": 36597, "epoch": 345, "lr": 2.1827234180118517e-05} +{"train_loss": 0.014495170675218105, "grad_norm": 0.08856245130300522, "global_step": 36598, "epoch": 345, "lr": 2.1824785724044967e-05} +{"train_loss": 0.01612355373799801, "grad_norm": 0.05971166118979454, "global_step": 36599, "epoch": 345, "lr": 2.1822337366966898e-05} +{"train_loss": 0.01633874699473381, "grad_norm": 0.1251707375049591, "global_step": 36600, "epoch": 345, "lr": 2.181988910889285e-05} +{"train_loss": 0.0148888835683465, "grad_norm": 0.07020312547683716, "global_step": 36601, "epoch": 345, "lr": 2.1817440949831453e-05} +{"train_loss": 0.01571238599717617, "grad_norm": 0.0698729008436203, "global_step": 36602, "epoch": 345, "lr": 2.181499288979133e-05} +{"train_loss": 0.01619354635477066, "grad_norm": 0.09170180559158325, "global_step": 36603, "epoch": 345, "lr": 2.1812544928781048e-05} +{"train_loss": 0.017076196148991585, "grad_norm": 0.06534174084663391, "global_step": 36604, "epoch": 345, "lr": 2.1810097066809227e-05} +{"train_loss": 0.014512646943330765, "grad_norm": 0.0991407111287117, "global_step": 36605, "epoch": 345, "lr": 2.180764930388448e-05} +{"train_loss": 0.016049090772867203, "grad_norm": 0.08691415935754776, "global_step": 36606, "epoch": 345, "lr": 2.1805201640015393e-05} +{"train_loss": 0.013875039294362068, "grad_norm": 0.08723155409097672, "global_step": 36607, "epoch": 345, "lr": 2.180275407521055e-05} +{"train_loss": 0.013688746839761734, "grad_norm": 0.06965120136737823, "global_step": 36608, "epoch": 345, "lr": 2.1800306609478564e-05} +{"train_loss": 0.013997441157698631, "grad_norm": 0.09681054204702377, "global_step": 36609, "epoch": 345, "lr": 2.179785924282806e-05} +{"train_loss": 0.013996303081512451, "grad_norm": 0.098549984395504, "global_step": 36610, "epoch": 345, "lr": 2.1795411975267594e-05} +{"train_loss": 0.014636455103754997, "grad_norm": 0.0773211345076561, "global_step": 36611, "epoch": 345, "lr": 2.17929648068058e-05} +{"train_loss": 0.014434047043323517, "grad_norm": 0.08301743865013123, "global_step": 36612, "epoch": 345, "lr": 2.1790517737451245e-05} +{"train_loss": 0.01499982364475727, "grad_norm": 0.09265415370464325, "global_step": 36613, "epoch": 345, "lr": 2.1788070767212554e-05} +{"train_loss": 0.0133441761136055, "grad_norm": 0.11421135067939758, "global_step": 36614, "epoch": 345, "lr": 2.17856238960983e-05} +{"train_loss": 0.015150605700910091, "grad_norm": 0.14015863835811615, "global_step": 36615, "epoch": 345, "lr": 2.1783177124117092e-05} +{"train_loss": 0.012290898710489273, "grad_norm": 0.06834016740322113, "global_step": 36616, "epoch": 345, "lr": 2.178073045127757e-05} +{"train_loss": 0.0195600762963295, "grad_norm": 0.17379841208457947, "global_step": 36617, "epoch": 345, "lr": 2.1778283877588253e-05} +{"train_loss": 0.01622602716088295, "grad_norm": 0.10769829154014587, "global_step": 36618, "epoch": 345, "lr": 2.1775837403057765e-05} +{"train_loss": 0.013430759310722351, "grad_norm": 0.07133359462022781, "global_step": 36619, "epoch": 345, "lr": 2.1773391027694734e-05} +{"train_loss": 0.016756461933255196, "grad_norm": 0.12277433276176453, "global_step": 36620, "epoch": 345, "lr": 2.177094475150771e-05} +{"train_loss": 0.014327197335660458, "grad_norm": 0.09255571663379669, "global_step": 36621, "epoch": 345, "lr": 2.1768498574505315e-05} +{"train_loss": 0.015218616463243961, "grad_norm": 0.08253416419029236, "global_step": 36622, "epoch": 345, "lr": 2.1766052496696153e-05} +{"train_loss": 0.015060300007462502, "grad_norm": 0.10674670338630676, "global_step": 36623, "epoch": 345, "lr": 2.1763606518088803e-05} +{"train_loss": 0.013469606637954712, "grad_norm": 0.0684707760810852, "global_step": 36624, "epoch": 345, "lr": 2.176116063869184e-05} +{"train_loss": 0.014219873584806919, "grad_norm": 0.06426224112510681, "global_step": 36625, "epoch": 345, "lr": 2.1758714858513883e-05} +{"train_loss": 0.013723958283662796, "grad_norm": 0.06758000701665878, "global_step": 36626, "epoch": 345, "lr": 2.1756269177563536e-05} +{"train_loss": 0.01681974157691002, "grad_norm": 0.08250658214092255, "global_step": 36627, "epoch": 345, "lr": 2.175382359584936e-05} +{"train_loss": 0.01353494729846716, "grad_norm": 0.08591306954622269, "global_step": 36628, "epoch": 345, "lr": 2.1751378113379978e-05} +{"train_loss": 0.01448035053908825, "grad_norm": 0.055869296193122864, "global_step": 36629, "epoch": 345, "lr": 2.174893273016395e-05} +{"train_loss": 0.01171334832906723, "grad_norm": 0.12891244888305664, "global_step": 36630, "epoch": 345, "lr": 2.1746487446209913e-05} +{"train_loss": 0.014907347038388252, "grad_norm": 0.11630001664161682, "global_step": 36631, "epoch": 345, "lr": 2.1744042261526405e-05} +{"train_loss": 0.01751795969903469, "grad_norm": 0.08841762691736221, "global_step": 36632, "epoch": 345, "lr": 2.1741597176122047e-05} +{"train_loss": 0.015136485919356346, "grad_norm": 0.07558170706033707, "global_step": 36633, "epoch": 345, "lr": 2.1739152190005447e-05} +{"train_loss": 0.014965428039431572, "grad_norm": 0.09026892483234406, "global_step": 36634, "epoch": 345, "lr": 2.173670730318517e-05} +{"train_loss": 0.014702314510941505, "grad_norm": 0.1288616955280304, "global_step": 36635, "epoch": 345, "lr": 2.17342625156698e-05} +{"train_loss": 0.01624229922890663, "grad_norm": 0.08438975363969803, "global_step": 36636, "epoch": 345, "lr": 2.173181782746796e-05} +{"train_loss": 0.015669142827391624, "grad_norm": 0.05809794366359711, "global_step": 36637, "epoch": 345, "lr": 2.172937323858819e-05} +{"train_loss": 0.018261857330799103, "grad_norm": 0.09414514899253845, "global_step": 36638, "epoch": 345, "lr": 2.1726928749039116e-05} +{"train_loss": 0.014876795932650566, "grad_norm": 0.060376547276973724, "global_step": 36639, "epoch": 345, "lr": 2.1724484358829338e-05} +{"train_loss": 0.012572775594890118, "grad_norm": 0.05085373669862747, "global_step": 36640, "epoch": 345, "lr": 2.172204006796742e-05} +{"train_loss": 0.017591405659914017, "grad_norm": 0.1513599008321762, "global_step": 36641, "epoch": 345, "lr": 2.171959587646194e-05} +{"train_loss": 0.014573425054550171, "grad_norm": 0.07487216591835022, "global_step": 36642, "epoch": 345, "lr": 2.17171517843215e-05} +{"train_loss": 0.015408273786306381, "grad_norm": 0.12216266989707947, "global_step": 36643, "epoch": 345, "lr": 2.1714707791554705e-05} +{"train_loss": 0.01528917346149683, "grad_norm": 0.06800223886966705, "global_step": 36644, "epoch": 345, "lr": 2.1712263898170116e-05} +{"train_loss": 0.014222227968275547, "grad_norm": 0.08601535111665726, "global_step": 36645, "epoch": 345, "lr": 2.170982010417632e-05} +{"train_loss": 0.017000699415802956, "grad_norm": 0.10268545895814896, "global_step": 36646, "epoch": 345, "lr": 2.1707376409581937e-05} +{"train_loss": 0.013182342052459717, "grad_norm": 0.061795491725206375, "global_step": 36647, "epoch": 345, "lr": 2.1704932814395528e-05} +{"train_loss": 0.016147203743457794, "grad_norm": 0.08482475578784943, "global_step": 36648, "epoch": 345, "lr": 2.1702489318625657e-05} +{"train_loss": 0.014486854895949364, "grad_norm": 0.0672636553645134, "global_step": 36649, "epoch": 345, "lr": 2.1700045922280937e-05} +{"train_loss": 0.01520277839154005, "grad_norm": 0.05389807000756264, "global_step": 36650, "epoch": 345, "lr": 2.1697602625369967e-05} +{"train_loss": 0.014358365908265114, "grad_norm": 0.05228833109140396, "global_step": 36651, "epoch": 345, "lr": 2.1695159427901295e-05} +{"train_loss": 0.016937267035245895, "grad_norm": 0.06807007640600204, "global_step": 36652, "epoch": 345, "lr": 2.1692716329883538e-05} +{"train_loss": 0.015665823593735695, "grad_norm": 0.07938893139362335, "global_step": 36653, "epoch": 345, "lr": 2.169027333132525e-05} +{"train_loss": 0.014074999839067459, "grad_norm": 0.11071495711803436, "global_step": 36654, "epoch": 345, "lr": 2.1687830432235046e-05} +{"train_loss": 0.014848851598799229, "grad_norm": 0.14109981060028076, "global_step": 36655, "epoch": 345, "lr": 2.168538763262148e-05} +{"train_loss": 0.01465899869799614, "grad_norm": 0.08937470614910126, "global_step": 36656, "epoch": 345, "lr": 2.1682944932493143e-05} +{"train_loss": 0.016304343938827515, "grad_norm": 0.11463896185159683, "global_step": 36657, "epoch": 345, "lr": 2.1680502331858644e-05} +{"train_loss": 0.017159905284643173, "grad_norm": 0.13304150104522705, "global_step": 36658, "epoch": 345, "lr": 2.1678059830726543e-05} +{"train_loss": 0.014422554522752762, "grad_norm": 0.07274777442216873, "global_step": 36659, "epoch": 345, "lr": 2.1675617429105403e-05} +{"train_loss": 0.015785621479153633, "grad_norm": 0.10511751472949982, "global_step": 36660, "epoch": 345, "lr": 2.167317512700384e-05} +{"train_loss": 0.0134909488260746, "grad_norm": 0.06530431658029556, "global_step": 36661, "epoch": 345, "lr": 2.167073292443041e-05} +{"train_loss": 0.0124114491045475, "grad_norm": 0.09974876046180725, "global_step": 36662, "epoch": 345, "lr": 2.1668290821393704e-05} +{"train_loss": 0.014915699139237404, "grad_norm": 0.06507227569818497, "global_step": 36663, "epoch": 345, "lr": 2.166584881790232e-05} +{"train_loss": 0.015069405548274517, "grad_norm": 0.15745684504508972, "global_step": 36664, "epoch": 345, "lr": 2.1663406913964816e-05} +{"train_loss": 0.01641702651977539, "grad_norm": 0.1175040751695633, "global_step": 36665, "epoch": 345, "lr": 2.1660965109589766e-05} +{"train_loss": 0.01552590448409319, "grad_norm": 0.0824880301952362, "global_step": 36666, "epoch": 345, "lr": 2.1658523404785752e-05} +{"train_loss": 0.015397628769278526, "grad_norm": 0.0970015898346901, "global_step": 36667, "epoch": 345, "lr": 2.1656081799561384e-05} +{"train_loss": 0.015139168128371239, "grad_norm": 0.07130763679742813, "global_step": 36668, "epoch": 345, "lr": 2.1653640293925205e-05} +{"train_loss": 0.013772333972156048, "grad_norm": 0.13713181018829346, "global_step": 36669, "epoch": 345, "lr": 2.1651198887885815e-05} +{"train_loss": 0.01714515872299671, "grad_norm": 0.07229314744472504, "global_step": 36670, "epoch": 345, "lr": 2.164875758145177e-05} +{"train_loss": 0.014559397473931313, "grad_norm": 0.09417225420475006, "global_step": 36671, "epoch": 345, "lr": 2.164631637463168e-05} +{"train_loss": 0.01680397242307663, "grad_norm": 0.07152687758207321, "global_step": 36672, "epoch": 345, "lr": 2.1643875267434088e-05} +{"train_loss": 0.012784812599420547, "grad_norm": 0.06872475892305374, "global_step": 36673, "epoch": 345, "lr": 2.1641434259867583e-05} +{"train_loss": 0.014667833223938942, "grad_norm": 0.08213333040475845, "global_step": 36674, "epoch": 345, "lr": 2.163899335194076e-05} +{"train_loss": 0.01511458268845981, "grad_norm": 0.11297806352376938, "global_step": 36675, "epoch": 345, "lr": 2.1636552543662186e-05, "val_loss": 0.01911546103656292} +{"train_loss": 0.013453120365738869, "grad_norm": 0.05980237200856209, "global_step": 36676, "epoch": 346, "lr": 2.1634111835040414e-05} +{"train_loss": 0.01579202711582184, "grad_norm": 0.09247776120901108, "global_step": 36677, "epoch": 346, "lr": 2.1631671226084053e-05} +{"train_loss": 0.013070200569927692, "grad_norm": 0.08906365931034088, "global_step": 36678, "epoch": 346, "lr": 2.1629230716801647e-05} +{"train_loss": 0.015050547197461128, "grad_norm": 0.11838753521442413, "global_step": 36679, "epoch": 346, "lr": 2.1626790307201778e-05} +{"train_loss": 0.01534975878894329, "grad_norm": 0.07226449251174927, "global_step": 36680, "epoch": 346, "lr": 2.1624349997293055e-05} +{"train_loss": 0.016722526401281357, "grad_norm": 0.0714985728263855, "global_step": 36681, "epoch": 346, "lr": 2.1621909787084023e-05} +{"train_loss": 0.014654875732958317, "grad_norm": 0.2068500816822052, "global_step": 36682, "epoch": 346, "lr": 2.1619469676583237e-05} +{"train_loss": 0.017074106261134148, "grad_norm": 0.09011668711900711, "global_step": 36683, "epoch": 346, "lr": 2.16170296657993e-05} +{"train_loss": 0.014575090259313583, "grad_norm": 0.11430300027132034, "global_step": 36684, "epoch": 346, "lr": 2.1614589754740787e-05} +{"train_loss": 0.013407682068645954, "grad_norm": 0.09207089990377426, "global_step": 36685, "epoch": 346, "lr": 2.161214994341625e-05} +{"train_loss": 0.013544674962759018, "grad_norm": 0.14400836825370789, "global_step": 36686, "epoch": 346, "lr": 2.1609710231834268e-05} +{"train_loss": 0.01514520961791277, "grad_norm": 0.11168672889471054, "global_step": 36687, "epoch": 346, "lr": 2.160727062000344e-05} +{"train_loss": 0.0169216301292181, "grad_norm": 0.06768349558115005, "global_step": 36688, "epoch": 346, "lr": 2.160483110793231e-05} +{"train_loss": 0.01408829353749752, "grad_norm": 0.10106406360864639, "global_step": 36689, "epoch": 346, "lr": 2.1602391695629437e-05} +{"train_loss": 0.014378014020621777, "grad_norm": 0.07457946985960007, "global_step": 36690, "epoch": 346, "lr": 2.1599952383103416e-05} +{"train_loss": 0.014435682445764542, "grad_norm": 0.10637848824262619, "global_step": 36691, "epoch": 346, "lr": 2.1597513170362825e-05} +{"train_loss": 0.014680193737149239, "grad_norm": 0.10127183794975281, "global_step": 36692, "epoch": 346, "lr": 2.1595074057416204e-05} +{"train_loss": 0.015657132491469383, "grad_norm": 0.1297803372144699, "global_step": 36693, "epoch": 346, "lr": 2.159263504427216e-05} +{"train_loss": 0.014260130003094673, "grad_norm": 0.05745555832982063, "global_step": 36694, "epoch": 346, "lr": 2.1590196130939233e-05} +{"train_loss": 0.015165131539106369, "grad_norm": 0.14339976012706757, "global_step": 36695, "epoch": 346, "lr": 2.1587757317425993e-05} +{"train_loss": 0.01739160716533661, "grad_norm": 0.05408267676830292, "global_step": 36696, "epoch": 346, "lr": 2.1585318603741016e-05} +{"train_loss": 0.0146161625161767, "grad_norm": 0.09769247472286224, "global_step": 36697, "epoch": 346, "lr": 2.158287998989287e-05} +{"train_loss": 0.013973299413919449, "grad_norm": 0.06507495045661926, "global_step": 36698, "epoch": 346, "lr": 2.1580441475890147e-05} +{"train_loss": 0.01340313907712698, "grad_norm": 0.058746203780174255, "global_step": 36699, "epoch": 346, "lr": 2.1578003061741388e-05} +{"train_loss": 0.012441780418157578, "grad_norm": 0.05939163640141487, "global_step": 36700, "epoch": 346, "lr": 2.1575564747455153e-05} +{"train_loss": 0.01486927643418312, "grad_norm": 0.0769231766462326, "global_step": 36701, "epoch": 346, "lr": 2.157312653304004e-05} +{"train_loss": 0.014296961948275566, "grad_norm": 0.055283647030591965, "global_step": 36702, "epoch": 346, "lr": 2.1570688418504577e-05} +{"train_loss": 0.017630605027079582, "grad_norm": 0.17372895777225494, "global_step": 36703, "epoch": 346, "lr": 2.156825040385735e-05} +{"train_loss": 0.013143612071871758, "grad_norm": 0.06689106673002243, "global_step": 36704, "epoch": 346, "lr": 2.1565812489106947e-05} +{"train_loss": 0.015837334096431732, "grad_norm": 0.09327424317598343, "global_step": 36705, "epoch": 346, "lr": 2.156337467426191e-05} +{"train_loss": 0.014750869013369083, "grad_norm": 0.053675126284360886, "global_step": 36706, "epoch": 346, "lr": 2.156093695933079e-05} +{"train_loss": 0.016290588304400444, "grad_norm": 0.09142868965864182, "global_step": 36707, "epoch": 346, "lr": 2.1558499344322174e-05} +{"train_loss": 0.01609804853796959, "grad_norm": 0.1257997304201126, "global_step": 36708, "epoch": 346, "lr": 2.155606182924464e-05} +{"train_loss": 0.015068013221025467, "grad_norm": 0.17141953110694885, "global_step": 36709, "epoch": 346, "lr": 2.1553624414106716e-05} +{"train_loss": 0.012972705997526646, "grad_norm": 0.056561559438705444, "global_step": 36710, "epoch": 346, "lr": 2.155118709891698e-05} +{"train_loss": 0.015108552761375904, "grad_norm": 0.09030084311962128, "global_step": 36711, "epoch": 346, "lr": 2.154874988368403e-05} +{"train_loss": 0.01213170401751995, "grad_norm": 0.06870479136705399, "global_step": 36712, "epoch": 346, "lr": 2.1546312768416387e-05} +{"train_loss": 0.01463986374437809, "grad_norm": 0.08207793533802032, "global_step": 36713, "epoch": 346, "lr": 2.1543875753122615e-05} +{"train_loss": 0.013746555894613266, "grad_norm": 0.09682241082191467, "global_step": 36714, "epoch": 346, "lr": 2.1541438837811283e-05} +{"train_loss": 0.013343763537704945, "grad_norm": 0.07551451027393341, "global_step": 36715, "epoch": 346, "lr": 2.1539002022490985e-05} +{"train_loss": 0.014702579006552696, "grad_norm": 0.13903509080410004, "global_step": 36716, "epoch": 346, "lr": 2.1536565307170255e-05} +{"train_loss": 0.016482042148709297, "grad_norm": 0.09619725495576859, "global_step": 36717, "epoch": 346, "lr": 2.153412869185763e-05} +{"train_loss": 0.013218006119132042, "grad_norm": 0.05780040845274925, "global_step": 36718, "epoch": 346, "lr": 2.1531692176561725e-05} +{"train_loss": 0.017658835276961327, "grad_norm": 0.06593558192253113, "global_step": 36719, "epoch": 346, "lr": 2.1529255761291052e-05} +{"train_loss": 0.0109458789229393, "grad_norm": 0.056911446154117584, "global_step": 36720, "epoch": 346, "lr": 2.1526819446054185e-05} +{"train_loss": 0.015497090294957161, "grad_norm": 0.06943035870790482, "global_step": 36721, "epoch": 346, "lr": 2.1524383230859718e-05} +{"train_loss": 0.015143965370953083, "grad_norm": 0.07963649183511734, "global_step": 36722, "epoch": 346, "lr": 2.1521947115716183e-05} +{"train_loss": 0.012712978757917881, "grad_norm": 0.05972115322947502, "global_step": 36723, "epoch": 346, "lr": 2.151951110063213e-05} +{"train_loss": 0.013155527412891388, "grad_norm": 0.1589837372303009, "global_step": 36724, "epoch": 346, "lr": 2.151707518561612e-05} +{"train_loss": 0.01683010160923004, "grad_norm": 0.08484118431806564, "global_step": 36725, "epoch": 346, "lr": 2.1514639370676738e-05} +{"train_loss": 0.01580050215125084, "grad_norm": 0.05348361283540726, "global_step": 36726, "epoch": 346, "lr": 2.1512203655822512e-05} +{"train_loss": 0.015550528652966022, "grad_norm": 0.10174515098333359, "global_step": 36727, "epoch": 346, "lr": 2.150976804106201e-05} +{"train_loss": 0.014498502016067505, "grad_norm": 0.14402464032173157, "global_step": 36728, "epoch": 346, "lr": 2.150733252640381e-05} +{"train_loss": 0.013658754527568817, "grad_norm": 0.061460260301828384, "global_step": 36729, "epoch": 346, "lr": 2.150489711185646e-05} +{"train_loss": 0.017299484461545944, "grad_norm": 0.0872100219130516, "global_step": 36730, "epoch": 346, "lr": 2.1502461797428485e-05} +{"train_loss": 0.013904981315135956, "grad_norm": 0.13265325129032135, "global_step": 36731, "epoch": 346, "lr": 2.1500026583128467e-05} +{"train_loss": 0.014961056411266327, "grad_norm": 0.1180499941110611, "global_step": 36732, "epoch": 346, "lr": 2.1497591468964983e-05} +{"train_loss": 0.016012415289878845, "grad_norm": 0.07772845029830933, "global_step": 36733, "epoch": 346, "lr": 2.1495156454946546e-05} +{"train_loss": 0.01396773848682642, "grad_norm": 0.11836381256580353, "global_step": 36734, "epoch": 346, "lr": 2.149272154108175e-05} +{"train_loss": 0.01612534001469612, "grad_norm": 0.10716883093118668, "global_step": 36735, "epoch": 346, "lr": 2.1490286727379133e-05} +{"train_loss": 0.013205152936279774, "grad_norm": 0.07419685274362564, "global_step": 36736, "epoch": 346, "lr": 2.148785201384723e-05} +{"train_loss": 0.014878403395414352, "grad_norm": 0.06860119104385376, "global_step": 36737, "epoch": 346, "lr": 2.1485417400494622e-05} +{"train_loss": 0.016797490417957306, "grad_norm": 0.09534493088722229, "global_step": 36738, "epoch": 346, "lr": 2.1482982887329874e-05} +{"train_loss": 0.014598573558032513, "grad_norm": 0.10256362706422806, "global_step": 36739, "epoch": 346, "lr": 2.1480548474361493e-05} +{"train_loss": 0.016494959592819214, "grad_norm": 0.0942855253815651, "global_step": 36740, "epoch": 346, "lr": 2.147811416159809e-05} +{"train_loss": 0.017126750200986862, "grad_norm": 0.07594989240169525, "global_step": 36741, "epoch": 346, "lr": 2.1475679949048173e-05} +{"train_loss": 0.013090122491121292, "grad_norm": 0.06635157018899918, "global_step": 36742, "epoch": 346, "lr": 2.1473245836720324e-05} +{"train_loss": 0.011496766470372677, "grad_norm": 0.07360400259494781, "global_step": 36743, "epoch": 346, "lr": 2.1470811824623072e-05} +{"train_loss": 0.013309034518897533, "grad_norm": 0.05612344667315483, "global_step": 36744, "epoch": 346, "lr": 2.1468377912764974e-05} +{"train_loss": 0.01611511781811714, "grad_norm": 0.06965386867523193, "global_step": 36745, "epoch": 346, "lr": 2.1465944101154607e-05} +{"train_loss": 0.01504366472363472, "grad_norm": 0.0609113983809948, "global_step": 36746, "epoch": 346, "lr": 2.1463510389800506e-05} +{"train_loss": 0.015565760433673859, "grad_norm": 0.10597017407417297, "global_step": 36747, "epoch": 346, "lr": 2.1461076778711196e-05} +{"train_loss": 0.01621129922568798, "grad_norm": 0.0918826013803482, "global_step": 36748, "epoch": 346, "lr": 2.1458643267895256e-05} +{"train_loss": 0.01626730151474476, "grad_norm": 0.08579187095165253, "global_step": 36749, "epoch": 346, "lr": 2.1456209857361248e-05} +{"train_loss": 0.016537602990865707, "grad_norm": 0.137159526348114, "global_step": 36750, "epoch": 346, "lr": 2.145377654711769e-05} +{"train_loss": 0.015150606632232666, "grad_norm": 0.0673169493675232, "global_step": 36751, "epoch": 346, "lr": 2.145134333717314e-05} +{"train_loss": 0.014423670247197151, "grad_norm": 0.15527130663394928, "global_step": 36752, "epoch": 346, "lr": 2.144891022753618e-05} +{"train_loss": 0.01512664183974266, "grad_norm": 0.09001469612121582, "global_step": 36753, "epoch": 346, "lr": 2.1446477218215327e-05} +{"train_loss": 0.01625940203666687, "grad_norm": 0.06479907780885696, "global_step": 36754, "epoch": 346, "lr": 2.144404430921912e-05} +{"train_loss": 0.01957857608795166, "grad_norm": 0.07512843608856201, "global_step": 36755, "epoch": 346, "lr": 2.1441611500556124e-05} +{"train_loss": 0.016445327550172806, "grad_norm": 0.14021305739879608, "global_step": 36756, "epoch": 346, "lr": 2.14391787922349e-05} +{"train_loss": 0.01444645132869482, "grad_norm": 0.09765070676803589, "global_step": 36757, "epoch": 346, "lr": 2.143674618426396e-05} +{"train_loss": 0.01417337916791439, "grad_norm": 0.11653049290180206, "global_step": 36758, "epoch": 346, "lr": 2.1434313676651902e-05} +{"train_loss": 0.014821207150816917, "grad_norm": 0.06404691189527512, "global_step": 36759, "epoch": 346, "lr": 2.1431881269407233e-05} +{"train_loss": 0.013034788891673088, "grad_norm": 0.0890781357884407, "global_step": 36760, "epoch": 346, "lr": 2.1429448962538497e-05} +{"train_loss": 0.01673024147748947, "grad_norm": 0.11053448170423508, "global_step": 36761, "epoch": 346, "lr": 2.1427016756054247e-05} +{"train_loss": 0.013991322368383408, "grad_norm": 0.13465553522109985, "global_step": 36762, "epoch": 346, "lr": 2.1424584649963052e-05} +{"train_loss": 0.013884078711271286, "grad_norm": 0.13911032676696777, "global_step": 36763, "epoch": 346, "lr": 2.1422152644273424e-05} +{"train_loss": 0.01743890345096588, "grad_norm": 0.11458200216293335, "global_step": 36764, "epoch": 346, "lr": 2.141972073899394e-05} +{"train_loss": 0.014871545135974884, "grad_norm": 0.055566590279340744, "global_step": 36765, "epoch": 346, "lr": 2.141728893413311e-05} +{"train_loss": 0.01824187859892845, "grad_norm": 0.13139653205871582, "global_step": 36766, "epoch": 346, "lr": 2.1414857229699516e-05} +{"train_loss": 0.015984730795025826, "grad_norm": 0.11561091989278793, "global_step": 36767, "epoch": 346, "lr": 2.1412425625701666e-05} +{"train_loss": 0.0153235774487257, "grad_norm": 0.10049035400152206, "global_step": 36768, "epoch": 346, "lr": 2.1409994122148114e-05} +{"train_loss": 0.014637301675975323, "grad_norm": 0.0740378201007843, "global_step": 36769, "epoch": 346, "lr": 2.1407562719047432e-05} +{"train_loss": 0.017882071435451508, "grad_norm": 0.09131386876106262, "global_step": 36770, "epoch": 346, "lr": 2.140513141640814e-05} +{"train_loss": 0.015399284660816193, "grad_norm": 0.08254975080490112, "global_step": 36771, "epoch": 346, "lr": 2.1402700214238764e-05} +{"train_loss": 0.013999414630234241, "grad_norm": 0.06789489090442657, "global_step": 36772, "epoch": 346, "lr": 2.1400269112547854e-05} +{"train_loss": 0.015289459377527237, "grad_norm": 0.08137369900941849, "global_step": 36773, "epoch": 346, "lr": 2.1397838111343986e-05} +{"train_loss": 0.015878979116678238, "grad_norm": 0.08557189255952835, "global_step": 36774, "epoch": 346, "lr": 2.1395407210635658e-05} +{"train_loss": 0.017216796055436134, "grad_norm": 0.07031992822885513, "global_step": 36775, "epoch": 346, "lr": 2.139297641043145e-05} +{"train_loss": 0.016338136047124863, "grad_norm": 0.09914343804121017, "global_step": 36776, "epoch": 346, "lr": 2.139054571073988e-05} +{"train_loss": 0.015993885695934296, "grad_norm": 0.09288114309310913, "global_step": 36777, "epoch": 346, "lr": 2.138811511156947e-05} +{"train_loss": 0.01776963658630848, "grad_norm": 0.07574155181646347, "global_step": 36778, "epoch": 346, "lr": 2.138568461292878e-05} +{"train_loss": 0.013757002539932728, "grad_norm": 0.07216186821460724, "global_step": 36779, "epoch": 346, "lr": 2.138325421482637e-05} +{"train_loss": 0.016052406281232834, "grad_norm": 0.0672827661037445, "global_step": 36780, "epoch": 346, "lr": 2.138082391727074e-05} +{"train_loss": 0.015125075486963088, "grad_norm": 0.08666174113750458, "global_step": 36781, "epoch": 346, "lr": 2.1378393720270468e-05, "val_loss": 0.019812192767858505} +{"train_loss": 0.016282012686133385, "grad_norm": 0.07293650507926941, "global_step": 36782, "epoch": 347, "lr": 2.1375963623834055e-05} +{"train_loss": 0.013552315533161163, "grad_norm": 0.07472600042819977, "global_step": 36783, "epoch": 347, "lr": 2.1373533627970076e-05} +{"train_loss": 0.014911298640072346, "grad_norm": 0.09739072620868683, "global_step": 36784, "epoch": 347, "lr": 2.137110373268703e-05} +{"train_loss": 0.014786776155233383, "grad_norm": 0.07377376407384872, "global_step": 36785, "epoch": 347, "lr": 2.1368673937993478e-05} +{"train_loss": 0.01355370320379734, "grad_norm": 0.11579722911119461, "global_step": 36786, "epoch": 347, "lr": 2.136624424389797e-05} +{"train_loss": 0.016402041539549828, "grad_norm": 0.09204714745283127, "global_step": 36787, "epoch": 347, "lr": 2.136381465040903e-05} +{"train_loss": 0.014892645180225372, "grad_norm": 0.06793154031038284, "global_step": 36788, "epoch": 347, "lr": 2.1361385157535173e-05} +{"train_loss": 0.014055581763386726, "grad_norm": 0.05603281408548355, "global_step": 36789, "epoch": 347, "lr": 2.1358955765284954e-05} +{"train_loss": 0.016647543758153915, "grad_norm": 0.107965849339962, "global_step": 36790, "epoch": 347, "lr": 2.1356526473666925e-05} +{"train_loss": 0.014114269986748695, "grad_norm": 0.06776637583971024, "global_step": 36791, "epoch": 347, "lr": 2.135409728268959e-05} +{"train_loss": 0.012259090319275856, "grad_norm": 0.1138935312628746, "global_step": 36792, "epoch": 347, "lr": 2.1351668192361497e-05} +{"train_loss": 0.015089268796145916, "grad_norm": 0.13027338683605194, "global_step": 36793, "epoch": 347, "lr": 2.134923920269122e-05} +{"train_loss": 0.013927724212408066, "grad_norm": 0.07061037421226501, "global_step": 36794, "epoch": 347, "lr": 2.134681031368722e-05} +{"train_loss": 0.015005972236394882, "grad_norm": 0.05947073549032211, "global_step": 36795, "epoch": 347, "lr": 2.134438152535807e-05} +{"train_loss": 0.01633966714143753, "grad_norm": 0.09281258285045624, "global_step": 36796, "epoch": 347, "lr": 2.1341952837712314e-05} +{"train_loss": 0.014482306316494942, "grad_norm": 0.07999992370605469, "global_step": 36797, "epoch": 347, "lr": 2.1339524250758463e-05} +{"train_loss": 0.012238179333508015, "grad_norm": 0.13391907513141632, "global_step": 36798, "epoch": 347, "lr": 2.1337095764505056e-05} +{"train_loss": 0.014120461419224739, "grad_norm": 0.10517293959856033, "global_step": 36799, "epoch": 347, "lr": 2.1334667378960644e-05} +{"train_loss": 0.013237036764621735, "grad_norm": 0.07905818521976471, "global_step": 36800, "epoch": 347, "lr": 2.1332239094133745e-05} +{"train_loss": 0.016172658652067184, "grad_norm": 0.056903544813394547, "global_step": 36801, "epoch": 347, "lr": 2.1329810910032877e-05} +{"train_loss": 0.01603478379547596, "grad_norm": 0.07854290306568146, "global_step": 36802, "epoch": 347, "lr": 2.132738282666658e-05} +{"train_loss": 0.015358424745500088, "grad_norm": 0.08004531264305115, "global_step": 36803, "epoch": 347, "lr": 2.132495484404341e-05} +{"train_loss": 0.01514723151922226, "grad_norm": 0.10432654619216919, "global_step": 36804, "epoch": 347, "lr": 2.1322526962171863e-05} +{"train_loss": 0.014903869479894638, "grad_norm": 0.06318742036819458, "global_step": 36805, "epoch": 347, "lr": 2.13200991810605e-05} +{"train_loss": 0.015280058607459068, "grad_norm": 0.05848143994808197, "global_step": 36806, "epoch": 347, "lr": 2.1317671500717816e-05} +{"train_loss": 0.017325278371572495, "grad_norm": 0.07068374752998352, "global_step": 36807, "epoch": 347, "lr": 2.1315243921152382e-05} +{"train_loss": 0.017020104452967644, "grad_norm": 0.05686827749013901, "global_step": 36808, "epoch": 347, "lr": 2.131281644237269e-05} +{"train_loss": 0.014637976884841919, "grad_norm": 0.08379838615655899, "global_step": 36809, "epoch": 347, "lr": 2.1310389064387282e-05} +{"train_loss": 0.017460722476243973, "grad_norm": 0.13774041831493378, "global_step": 36810, "epoch": 347, "lr": 2.130796178720471e-05} +{"train_loss": 0.016152221709489822, "grad_norm": 0.1281130611896515, "global_step": 36811, "epoch": 347, "lr": 2.1305534610833484e-05} +{"train_loss": 0.015437254682183266, "grad_norm": 0.11628705263137817, "global_step": 36812, "epoch": 347, "lr": 2.1303107535282107e-05} +{"train_loss": 0.014877203851938248, "grad_norm": 0.07541777938604355, "global_step": 36813, "epoch": 347, "lr": 2.1300680560559133e-05} +{"train_loss": 0.01686093583703041, "grad_norm": 0.08381656557321548, "global_step": 36814, "epoch": 347, "lr": 2.1298253686673107e-05} +{"train_loss": 0.01629544422030449, "grad_norm": 0.07154075056314468, "global_step": 36815, "epoch": 347, "lr": 2.1295826913632517e-05} +{"train_loss": 0.015467485412955284, "grad_norm": 0.06951398402452469, "global_step": 36816, "epoch": 347, "lr": 2.129340024144592e-05} +{"train_loss": 0.015607452020049095, "grad_norm": 0.055876296013593674, "global_step": 36817, "epoch": 347, "lr": 2.129097367012184e-05} +{"train_loss": 0.014125491492450237, "grad_norm": 0.07607527077198029, "global_step": 36818, "epoch": 347, "lr": 2.128854719966877e-05} +{"train_loss": 0.015125412493944168, "grad_norm": 0.12581604719161987, "global_step": 36819, "epoch": 347, "lr": 2.128612083009525e-05} +{"train_loss": 0.014344075694680214, "grad_norm": 0.0808788537979126, "global_step": 36820, "epoch": 347, "lr": 2.1283694561409844e-05} +{"train_loss": 0.014782129786908627, "grad_norm": 0.07017210125923157, "global_step": 36821, "epoch": 347, "lr": 2.128126839362103e-05} +{"train_loss": 0.011473934166133404, "grad_norm": 0.1296890676021576, "global_step": 36822, "epoch": 347, "lr": 2.1278842326737357e-05} +{"train_loss": 0.016436636447906494, "grad_norm": 0.06717830896377563, "global_step": 36823, "epoch": 347, "lr": 2.1276416360767333e-05} +{"train_loss": 0.016222596168518066, "grad_norm": 0.06623125076293945, "global_step": 36824, "epoch": 347, "lr": 2.1273990495719503e-05} +{"train_loss": 0.015230179764330387, "grad_norm": 0.07515744864940643, "global_step": 36825, "epoch": 347, "lr": 2.1271564731602363e-05} +{"train_loss": 0.018203768879175186, "grad_norm": 0.10731905698776245, "global_step": 36826, "epoch": 347, "lr": 2.126913906842445e-05} +{"train_loss": 0.018247896805405617, "grad_norm": 0.1398223489522934, "global_step": 36827, "epoch": 347, "lr": 2.126671350619431e-05} +{"train_loss": 0.013500802218914032, "grad_norm": 0.05911216139793396, "global_step": 36828, "epoch": 347, "lr": 2.1264288044920434e-05} +{"train_loss": 0.015233769081532955, "grad_norm": 0.0725732147693634, "global_step": 36829, "epoch": 347, "lr": 2.126186268461134e-05} +{"train_loss": 0.01332707516849041, "grad_norm": 0.06099332496523857, "global_step": 36830, "epoch": 347, "lr": 2.1259437425275564e-05} +{"train_loss": 0.015600241720676422, "grad_norm": 0.1166248768568039, "global_step": 36831, "epoch": 347, "lr": 2.125701226692164e-05} +{"train_loss": 0.015792666003108025, "grad_norm": 0.13159538805484772, "global_step": 36832, "epoch": 347, "lr": 2.1254587209558058e-05} +{"train_loss": 0.015059811994433403, "grad_norm": 0.07814966142177582, "global_step": 36833, "epoch": 347, "lr": 2.125216225319336e-05} +{"train_loss": 0.014657898806035519, "grad_norm": 0.05736684799194336, "global_step": 36834, "epoch": 347, "lr": 2.124973739783609e-05} +{"train_loss": 0.015255816280841827, "grad_norm": 0.0608760230243206, "global_step": 36835, "epoch": 347, "lr": 2.124731264349471e-05} +{"train_loss": 0.015409311279654503, "grad_norm": 0.08769983053207397, "global_step": 36836, "epoch": 347, "lr": 2.1244887990177763e-05} +{"train_loss": 0.015232020989060402, "grad_norm": 0.08403314650058746, "global_step": 36837, "epoch": 347, "lr": 2.12424634378938e-05} +{"train_loss": 0.013914238661527634, "grad_norm": 0.06869290769100189, "global_step": 36838, "epoch": 347, "lr": 2.124003898665129e-05} +{"train_loss": 0.014546191319823265, "grad_norm": 0.07460080832242966, "global_step": 36839, "epoch": 347, "lr": 2.123761463645878e-05} +{"train_loss": 0.01524195447564125, "grad_norm": 0.05928310379385948, "global_step": 36840, "epoch": 347, "lr": 2.1235190387324798e-05} +{"train_loss": 0.014123314060270786, "grad_norm": 0.0687846690416336, "global_step": 36841, "epoch": 347, "lr": 2.123276623925785e-05} +{"train_loss": 0.015598449856042862, "grad_norm": 0.08038275688886642, "global_step": 36842, "epoch": 347, "lr": 2.1230342192266434e-05} +{"train_loss": 0.01882690191268921, "grad_norm": 0.17626403272151947, "global_step": 36843, "epoch": 347, "lr": 2.122791824635908e-05} +{"train_loss": 0.013993676751852036, "grad_norm": 0.08639714121818542, "global_step": 36844, "epoch": 347, "lr": 2.1225494401544326e-05} +{"train_loss": 0.014910485595464706, "grad_norm": 0.09725897759199142, "global_step": 36845, "epoch": 347, "lr": 2.1223070657830657e-05} +{"train_loss": 0.01649540662765503, "grad_norm": 0.08018095046281815, "global_step": 36846, "epoch": 347, "lr": 2.122064701522662e-05} +{"train_loss": 0.013924759812653065, "grad_norm": 0.07982008159160614, "global_step": 36847, "epoch": 347, "lr": 2.1218223473740695e-05} +{"train_loss": 0.014466697350144386, "grad_norm": 0.11734439432621002, "global_step": 36848, "epoch": 347, "lr": 2.1215800033381434e-05} +{"train_loss": 0.01668950729072094, "grad_norm": 0.10319578647613525, "global_step": 36849, "epoch": 347, "lr": 2.1213376694157317e-05} +{"train_loss": 0.01646895706653595, "grad_norm": 0.06141155585646629, "global_step": 36850, "epoch": 347, "lr": 2.121095345607687e-05} +{"train_loss": 0.014502758160233498, "grad_norm": 0.058270350098609924, "global_step": 36851, "epoch": 347, "lr": 2.120853031914864e-05} +{"train_loss": 0.0174532700330019, "grad_norm": 0.08968488872051239, "global_step": 36852, "epoch": 347, "lr": 2.1206107283381106e-05} +{"train_loss": 0.014667271636426449, "grad_norm": 0.06476239860057831, "global_step": 36853, "epoch": 347, "lr": 2.120368434878277e-05} +{"train_loss": 0.014906449243426323, "grad_norm": 0.11332815885543823, "global_step": 36854, "epoch": 347, "lr": 2.120126151536217e-05} +{"train_loss": 0.01619253307580948, "grad_norm": 0.0661412924528122, "global_step": 36855, "epoch": 347, "lr": 2.1198838783127827e-05} +{"train_loss": 0.015451024286448956, "grad_norm": 0.07331396639347076, "global_step": 36856, "epoch": 347, "lr": 2.119641615208822e-05} +{"train_loss": 0.01782103255391121, "grad_norm": 0.0957769900560379, "global_step": 36857, "epoch": 347, "lr": 2.1193993622251885e-05} +{"train_loss": 0.01357240229845047, "grad_norm": 0.05755337327718735, "global_step": 36858, "epoch": 347, "lr": 2.1191571193627364e-05} +{"train_loss": 0.01355966366827488, "grad_norm": 0.07080982625484467, "global_step": 36859, "epoch": 347, "lr": 2.118914886622309e-05} +{"train_loss": 0.014947252348065376, "grad_norm": 0.07724196463823318, "global_step": 36860, "epoch": 347, "lr": 2.118672664004763e-05} +{"train_loss": 0.01653275266289711, "grad_norm": 0.08955259621143341, "global_step": 36861, "epoch": 347, "lr": 2.1184304515109488e-05} +{"train_loss": 0.015036724507808685, "grad_norm": 0.15068353712558746, "global_step": 36862, "epoch": 347, "lr": 2.1181882491417154e-05} +{"train_loss": 0.014391778036952019, "grad_norm": 0.08795627951622009, "global_step": 36863, "epoch": 347, "lr": 2.1179460568979153e-05} +{"train_loss": 0.014343726448714733, "grad_norm": 0.05761778727173805, "global_step": 36864, "epoch": 347, "lr": 2.117703874780401e-05} +{"train_loss": 0.013763703405857086, "grad_norm": 0.08436419814825058, "global_step": 36865, "epoch": 347, "lr": 2.1174617027900224e-05} +{"train_loss": 0.01633048802614212, "grad_norm": 0.07684159278869629, "global_step": 36866, "epoch": 347, "lr": 2.117219540927628e-05} +{"train_loss": 0.014740072190761566, "grad_norm": 0.07594375312328339, "global_step": 36867, "epoch": 347, "lr": 2.1169773891940697e-05} +{"train_loss": 0.018148068338632584, "grad_norm": 0.20058470964431763, "global_step": 36868, "epoch": 347, "lr": 2.1167352475902014e-05} +{"train_loss": 0.014600971713662148, "grad_norm": 0.08426769077777863, "global_step": 36869, "epoch": 347, "lr": 2.1164931161168712e-05} +{"train_loss": 0.015027728863060474, "grad_norm": 0.0990096852183342, "global_step": 36870, "epoch": 347, "lr": 2.116250994774929e-05} +{"train_loss": 0.012588052079081535, "grad_norm": 0.1157040074467659, "global_step": 36871, "epoch": 347, "lr": 2.1160088835652258e-05} +{"train_loss": 0.014225395396351814, "grad_norm": 0.07133281975984573, "global_step": 36872, "epoch": 347, "lr": 2.1157667824886158e-05} +{"train_loss": 0.013795404694974422, "grad_norm": 0.1438678503036499, "global_step": 36873, "epoch": 347, "lr": 2.1155246915459446e-05} +{"train_loss": 0.015846487134695053, "grad_norm": 0.14792446792125702, "global_step": 36874, "epoch": 347, "lr": 2.1152826107380653e-05} +{"train_loss": 0.014677191153168678, "grad_norm": 0.09301073104143143, "global_step": 36875, "epoch": 347, "lr": 2.1150405400658324e-05} +{"train_loss": 0.01740969344973564, "grad_norm": 0.07685571163892746, "global_step": 36876, "epoch": 347, "lr": 2.1147984795300885e-05} +{"train_loss": 0.013682939112186432, "grad_norm": 0.11620962619781494, "global_step": 36877, "epoch": 347, "lr": 2.1145564291316878e-05} +{"train_loss": 0.013577599078416824, "grad_norm": 0.05347074568271637, "global_step": 36878, "epoch": 347, "lr": 2.1143143888714834e-05} +{"train_loss": 0.01462264358997345, "grad_norm": 0.10749762505292892, "global_step": 36879, "epoch": 347, "lr": 2.114072358750321e-05} +{"train_loss": 0.014487363398075104, "grad_norm": 0.06258140504360199, "global_step": 36880, "epoch": 347, "lr": 2.1138303387690533e-05} +{"train_loss": 0.012769067659974098, "grad_norm": 0.07030323147773743, "global_step": 36881, "epoch": 347, "lr": 2.1135883289285325e-05} +{"train_loss": 0.015315932221710682, "grad_norm": 0.07764387130737305, "global_step": 36882, "epoch": 347, "lr": 2.113346329229607e-05} +{"train_loss": 0.014828336425125599, "grad_norm": 0.05781089514493942, "global_step": 36883, "epoch": 347, "lr": 2.1131043396731255e-05} +{"train_loss": 0.012185538187623024, "grad_norm": 0.08757314831018448, "global_step": 36884, "epoch": 347, "lr": 2.1128623602599397e-05} +{"train_loss": 0.017547253519296646, "grad_norm": 0.08285006880760193, "global_step": 36885, "epoch": 347, "lr": 2.1126203909909015e-05} +{"train_loss": 0.013989793136715889, "grad_norm": 0.061106979846954346, "global_step": 36886, "epoch": 347, "lr": 2.1123784318668583e-05} +{"train_loss": 0.0151079250763188, "grad_norm": 0.10993105173110962, "global_step": 36887, "epoch": 347, "lr": 2.112136482888663e-05, "val_loss": 0.019435055553913116} +{"train_loss": 0.01548327598720789, "grad_norm": 0.08840851485729218, "global_step": 36888, "epoch": 348, "lr": 2.1118945440571626e-05} +{"train_loss": 0.013867618516087532, "grad_norm": 0.05960671603679657, "global_step": 36889, "epoch": 348, "lr": 2.1116526153732104e-05} +{"train_loss": 0.015087527222931385, "grad_norm": 0.06414095312356949, "global_step": 36890, "epoch": 348, "lr": 2.111410696837653e-05} +{"train_loss": 0.015248194336891174, "grad_norm": 0.0721091777086258, "global_step": 36891, "epoch": 348, "lr": 2.111168788451342e-05} +{"train_loss": 0.01759323850274086, "grad_norm": 0.1574120819568634, "global_step": 36892, "epoch": 348, "lr": 2.1109268902151296e-05} +{"train_loss": 0.015567624941468239, "grad_norm": 0.10596047341823578, "global_step": 36893, "epoch": 348, "lr": 2.1106850021298636e-05} +{"train_loss": 0.01649445667862892, "grad_norm": 0.06535410135984421, "global_step": 36894, "epoch": 348, "lr": 2.1104431241963922e-05} +{"train_loss": 0.016633007675409317, "grad_norm": 0.10869956016540527, "global_step": 36895, "epoch": 348, "lr": 2.1102012564155695e-05} +{"train_loss": 0.013581651262938976, "grad_norm": 0.08927503228187561, "global_step": 36896, "epoch": 348, "lr": 2.1099593987882398e-05} +{"train_loss": 0.014819812960922718, "grad_norm": 0.06434963643550873, "global_step": 36897, "epoch": 348, "lr": 2.1097175513152567e-05} +{"train_loss": 0.015666302293539047, "grad_norm": 0.10806430876255035, "global_step": 36898, "epoch": 348, "lr": 2.1094757139974708e-05} +{"train_loss": 0.01608043722808361, "grad_norm": 0.05841781571507454, "global_step": 36899, "epoch": 348, "lr": 2.1092338868357302e-05} +{"train_loss": 0.015015410259366035, "grad_norm": 0.10725907236337662, "global_step": 36900, "epoch": 348, "lr": 2.108992069830883e-05} +{"train_loss": 0.014160321094095707, "grad_norm": 0.09219232201576233, "global_step": 36901, "epoch": 348, "lr": 2.1087502629837802e-05} +{"train_loss": 0.01278802752494812, "grad_norm": 0.08441746234893799, "global_step": 36902, "epoch": 348, "lr": 2.108508466295273e-05} +{"train_loss": 0.014856677502393723, "grad_norm": 0.09200379997491837, "global_step": 36903, "epoch": 348, "lr": 2.108266679766208e-05} +{"train_loss": 0.015972016379237175, "grad_norm": 0.06465709209442139, "global_step": 36904, "epoch": 348, "lr": 2.108024903397437e-05} +{"train_loss": 0.013539670035243034, "grad_norm": 0.07604239135980606, "global_step": 36905, "epoch": 348, "lr": 2.1077831371898098e-05} +{"train_loss": 0.01381102204322815, "grad_norm": 0.10947025567293167, "global_step": 36906, "epoch": 348, "lr": 2.107541381144175e-05} +{"train_loss": 0.015321082435548306, "grad_norm": 0.09523183107376099, "global_step": 36907, "epoch": 348, "lr": 2.1072996352613804e-05} +{"train_loss": 0.015024170279502869, "grad_norm": 0.06005887687206268, "global_step": 36908, "epoch": 348, "lr": 2.107057899542277e-05} +{"train_loss": 0.01522859651595354, "grad_norm": 0.07960959523916245, "global_step": 36909, "epoch": 348, "lr": 2.1068161739877153e-05} +{"train_loss": 0.014947669580578804, "grad_norm": 0.07683337479829788, "global_step": 36910, "epoch": 348, "lr": 2.1065744585985414e-05} +{"train_loss": 0.014504918828606606, "grad_norm": 0.1444620043039322, "global_step": 36911, "epoch": 348, "lr": 2.106332753375609e-05} +{"train_loss": 0.016544627025723457, "grad_norm": 0.1700143963098526, "global_step": 36912, "epoch": 348, "lr": 2.1060910583197625e-05} +{"train_loss": 0.014398720115423203, "grad_norm": 0.08534953743219376, "global_step": 36913, "epoch": 348, "lr": 2.1058493734318557e-05} +{"train_loss": 0.016161076724529266, "grad_norm": 0.08683360368013382, "global_step": 36914, "epoch": 348, "lr": 2.105607698712734e-05} +{"train_loss": 0.014015063643455505, "grad_norm": 0.05488361045718193, "global_step": 36915, "epoch": 348, "lr": 2.1053660341632476e-05} +{"train_loss": 0.015033794566988945, "grad_norm": 0.13435453176498413, "global_step": 36916, "epoch": 348, "lr": 2.1051243797842478e-05} +{"train_loss": 0.015015006996691227, "grad_norm": 0.05824959650635719, "global_step": 36917, "epoch": 348, "lr": 2.104882735576582e-05} +{"train_loss": 0.013560686260461807, "grad_norm": 0.07242953777313232, "global_step": 36918, "epoch": 348, "lr": 2.1046411015410984e-05} +{"train_loss": 0.013830499723553658, "grad_norm": 0.08289588242769241, "global_step": 36919, "epoch": 348, "lr": 2.1043994776786473e-05} +{"train_loss": 0.01393522322177887, "grad_norm": 0.09549272805452347, "global_step": 36920, "epoch": 348, "lr": 2.1041578639900762e-05} +{"train_loss": 0.015422623604536057, "grad_norm": 0.08871126919984818, "global_step": 36921, "epoch": 348, "lr": 2.1039162604762347e-05} +{"train_loss": 0.014850949868559837, "grad_norm": 0.11213377863168716, "global_step": 36922, "epoch": 348, "lr": 2.1036746671379738e-05} +{"train_loss": 0.014498534612357616, "grad_norm": 0.08815845847129822, "global_step": 36923, "epoch": 348, "lr": 2.1034330839761408e-05} +{"train_loss": 0.014528607949614525, "grad_norm": 0.056076716631650925, "global_step": 36924, "epoch": 348, "lr": 2.1031915109915822e-05} +{"train_loss": 0.014903143979609013, "grad_norm": 0.056501634418964386, "global_step": 36925, "epoch": 348, "lr": 2.102949948185149e-05} +{"train_loss": 0.013352278620004654, "grad_norm": 0.0994918942451477, "global_step": 36926, "epoch": 348, "lr": 2.102708395557691e-05} +{"train_loss": 0.01450405828654766, "grad_norm": 0.16698704659938812, "global_step": 36927, "epoch": 348, "lr": 2.1024668531100545e-05} +{"train_loss": 0.015162186697125435, "grad_norm": 0.08031285554170609, "global_step": 36928, "epoch": 348, "lr": 2.102225320843091e-05} +{"train_loss": 0.013572458177804947, "grad_norm": 0.10215522348880768, "global_step": 36929, "epoch": 348, "lr": 2.1019837987576458e-05} +{"train_loss": 0.013635004870593548, "grad_norm": 0.08303419500589371, "global_step": 36930, "epoch": 348, "lr": 2.101742286854571e-05} +{"train_loss": 0.01565994694828987, "grad_norm": 0.07269759476184845, "global_step": 36931, "epoch": 348, "lr": 2.1015007851347113e-05} +{"train_loss": 0.013912873342633247, "grad_norm": 0.0944194570183754, "global_step": 36932, "epoch": 348, "lr": 2.1012592935989178e-05} +{"train_loss": 0.01563144661486149, "grad_norm": 0.17441166937351227, "global_step": 36933, "epoch": 348, "lr": 2.10101781224804e-05} +{"train_loss": 0.015180004760622978, "grad_norm": 0.0731009840965271, "global_step": 36934, "epoch": 348, "lr": 2.100776341082925e-05} +{"train_loss": 0.017765434458851814, "grad_norm": 0.07930160313844681, "global_step": 36935, "epoch": 348, "lr": 2.1005348801044196e-05} +{"train_loss": 0.01627315580844879, "grad_norm": 0.13891443610191345, "global_step": 36936, "epoch": 348, "lr": 2.1002934293133747e-05} +{"train_loss": 0.015273390337824821, "grad_norm": 0.08567740768194199, "global_step": 36937, "epoch": 348, "lr": 2.1000519887106363e-05} +{"train_loss": 0.016429683193564415, "grad_norm": 0.08452346175909042, "global_step": 36938, "epoch": 348, "lr": 2.099810558297054e-05} +{"train_loss": 0.013115454465150833, "grad_norm": 0.09151536971330643, "global_step": 36939, "epoch": 348, "lr": 2.0995691380734783e-05} +{"train_loss": 0.01619407720863819, "grad_norm": 0.06131130829453468, "global_step": 36940, "epoch": 348, "lr": 2.0993277280407548e-05} +{"train_loss": 0.014758719131350517, "grad_norm": 0.10425899922847748, "global_step": 36941, "epoch": 348, "lr": 2.0990863281997307e-05} +{"train_loss": 0.016237083822488785, "grad_norm": 0.11903678625822067, "global_step": 36942, "epoch": 348, "lr": 2.0988449385512553e-05} +{"train_loss": 0.015858115628361702, "grad_norm": 0.055262163281440735, "global_step": 36943, "epoch": 348, "lr": 2.0986035590961793e-05} +{"train_loss": 0.01374884881079197, "grad_norm": 0.05707051232457161, "global_step": 36944, "epoch": 348, "lr": 2.0983621898353468e-05} +{"train_loss": 0.014439094811677933, "grad_norm": 0.09796781092882156, "global_step": 36945, "epoch": 348, "lr": 2.0981208307696072e-05} +{"train_loss": 0.016159430146217346, "grad_norm": 0.059597406536340714, "global_step": 36946, "epoch": 348, "lr": 2.097879481899811e-05} +{"train_loss": 0.018166955560445786, "grad_norm": 0.12741123139858246, "global_step": 36947, "epoch": 348, "lr": 2.0976381432268044e-05} +{"train_loss": 0.014632610604166985, "grad_norm": 0.07106579095125198, "global_step": 36948, "epoch": 348, "lr": 2.0973968147514334e-05} +{"train_loss": 0.016144409775733948, "grad_norm": 0.10193870961666107, "global_step": 36949, "epoch": 348, "lr": 2.0971554964745477e-05} +{"train_loss": 0.015738680958747864, "grad_norm": 0.12249091267585754, "global_step": 36950, "epoch": 348, "lr": 2.0969141883969968e-05} +{"train_loss": 0.014855905435979366, "grad_norm": 0.17293787002563477, "global_step": 36951, "epoch": 348, "lr": 2.0966728905196254e-05} +{"train_loss": 0.01423555612564087, "grad_norm": 0.0721038281917572, "global_step": 36952, "epoch": 348, "lr": 2.096431602843284e-05} +{"train_loss": 0.017512112855911255, "grad_norm": 0.08160928636789322, "global_step": 36953, "epoch": 348, "lr": 2.09619032536882e-05} +{"train_loss": 0.016851525753736496, "grad_norm": 0.10562671720981598, "global_step": 36954, "epoch": 348, "lr": 2.0959490580970782e-05} +{"train_loss": 0.014241400174796581, "grad_norm": 0.06927257031202316, "global_step": 36955, "epoch": 348, "lr": 2.095707801028908e-05} +{"train_loss": 0.01476109679788351, "grad_norm": 0.10716842114925385, "global_step": 36956, "epoch": 348, "lr": 2.0954665541651602e-05} +{"train_loss": 0.014474807307124138, "grad_norm": 0.06674118340015411, "global_step": 36957, "epoch": 348, "lr": 2.0952253175066773e-05} +{"train_loss": 0.015183264389634132, "grad_norm": 0.11241745948791504, "global_step": 36958, "epoch": 348, "lr": 2.094984091054312e-05} +{"train_loss": 0.013973314315080643, "grad_norm": 0.05402720347046852, "global_step": 36959, "epoch": 348, "lr": 2.0947428748089066e-05} +{"train_loss": 0.01830190233886242, "grad_norm": 0.09534437209367752, "global_step": 36960, "epoch": 348, "lr": 2.0945016687713132e-05} +{"train_loss": 0.015320396050810814, "grad_norm": 0.06841319054365158, "global_step": 36961, "epoch": 348, "lr": 2.094260472942376e-05} +{"train_loss": 0.014840531162917614, "grad_norm": 0.11790967732667923, "global_step": 36962, "epoch": 348, "lr": 2.0940192873229435e-05} +{"train_loss": 0.01550653949379921, "grad_norm": 0.10912476480007172, "global_step": 36963, "epoch": 348, "lr": 2.093778111913865e-05} +{"train_loss": 0.012579171918332577, "grad_norm": 0.0615723542869091, "global_step": 36964, "epoch": 348, "lr": 2.093536946715987e-05} +{"train_loss": 0.01646137237548828, "grad_norm": 0.23317641019821167, "global_step": 36965, "epoch": 348, "lr": 2.0932957917301537e-05} +{"train_loss": 0.01578264869749546, "grad_norm": 0.11259214580059052, "global_step": 36966, "epoch": 348, "lr": 2.0930546469572153e-05} +{"train_loss": 0.012068893760442734, "grad_norm": 0.06743298470973969, "global_step": 36967, "epoch": 348, "lr": 2.092813512398021e-05} +{"train_loss": 0.0159428920596838, "grad_norm": 0.08410856872797012, "global_step": 36968, "epoch": 348, "lr": 2.0925723880534132e-05} +{"train_loss": 0.015792714431881905, "grad_norm": 0.0921114832162857, "global_step": 36969, "epoch": 348, "lr": 2.0923312739242434e-05} +{"train_loss": 0.01570923812687397, "grad_norm": 0.11425580084323883, "global_step": 36970, "epoch": 348, "lr": 2.0920901700113555e-05} +{"train_loss": 0.01615484617650509, "grad_norm": 0.09512606263160706, "global_step": 36971, "epoch": 348, "lr": 2.0918490763155997e-05} +{"train_loss": 0.017981521785259247, "grad_norm": 0.07403559982776642, "global_step": 36972, "epoch": 348, "lr": 2.0916079928378197e-05} +{"train_loss": 0.014385581947863102, "grad_norm": 0.15972526371479034, "global_step": 36973, "epoch": 348, "lr": 2.0913669195788645e-05} +{"train_loss": 0.013711834326386452, "grad_norm": 0.0934029221534729, "global_step": 36974, "epoch": 348, "lr": 2.0911258565395826e-05} +{"train_loss": 0.014233110472559929, "grad_norm": 0.09494926780462265, "global_step": 36975, "epoch": 348, "lr": 2.0908848037208196e-05} +{"train_loss": 0.014254925772547722, "grad_norm": 0.09088405966758728, "global_step": 36976, "epoch": 348, "lr": 2.0906437611234203e-05} +{"train_loss": 0.01371838990598917, "grad_norm": 0.060906823724508286, "global_step": 36977, "epoch": 348, "lr": 2.0904027287482358e-05} +{"train_loss": 0.014391811564564705, "grad_norm": 0.10802660882472992, "global_step": 36978, "epoch": 348, "lr": 2.0901617065961088e-05} +{"train_loss": 0.016888394951820374, "grad_norm": 0.07052614539861679, "global_step": 36979, "epoch": 348, "lr": 2.089920694667888e-05} +{"train_loss": 0.017767034471035004, "grad_norm": 0.19509652256965637, "global_step": 36980, "epoch": 348, "lr": 2.0896796929644218e-05} +{"train_loss": 0.01591135561466217, "grad_norm": 0.15793125331401825, "global_step": 36981, "epoch": 348, "lr": 2.0894387014865556e-05} +{"train_loss": 0.013033764436841011, "grad_norm": 0.11332538723945618, "global_step": 36982, "epoch": 348, "lr": 2.089197720235135e-05} +{"train_loss": 0.01690566912293434, "grad_norm": 0.13708727061748505, "global_step": 36983, "epoch": 348, "lr": 2.088956749211007e-05} +{"train_loss": 0.016794830560684204, "grad_norm": 0.1097368597984314, "global_step": 36984, "epoch": 348, "lr": 2.088715788415021e-05} +{"train_loss": 0.014100894331932068, "grad_norm": 0.10046447068452835, "global_step": 36985, "epoch": 348, "lr": 2.0884748378480194e-05} +{"train_loss": 0.014893226325511932, "grad_norm": 0.0593569278717041, "global_step": 36986, "epoch": 348, "lr": 2.088233897510852e-05} +{"train_loss": 0.014588491059839725, "grad_norm": 0.05551804229617119, "global_step": 36987, "epoch": 348, "lr": 2.087992967404365e-05} +{"train_loss": 0.016369950026273727, "grad_norm": 0.0845753625035286, "global_step": 36988, "epoch": 348, "lr": 2.0877520475294055e-05} +{"train_loss": 0.014105765148997307, "grad_norm": 0.06848302483558655, "global_step": 36989, "epoch": 348, "lr": 2.087511137886816e-05} +{"train_loss": 0.014548877254128456, "grad_norm": 0.05236988142132759, "global_step": 36990, "epoch": 348, "lr": 2.0872702384774467e-05} +{"train_loss": 0.0133155956864357, "grad_norm": 0.06853239238262177, "global_step": 36991, "epoch": 348, "lr": 2.0870293493021443e-05} +{"train_loss": 0.015148727223277092, "grad_norm": 0.13741864264011383, "global_step": 36992, "epoch": 348, "lr": 2.0867884703617523e-05} +{"train_loss": 0.015095760676798955, "grad_norm": 0.09633968025445938, "global_step": 36993, "epoch": 348, "lr": 2.0865476016571207e-05, "val_loss": 0.020105699077248573} +{"train_loss": 0.01293899305164814, "grad_norm": 0.10849127918481827, "global_step": 36994, "epoch": 349, "lr": 2.0863067431890932e-05} +{"train_loss": 0.01367657445371151, "grad_norm": 0.060542166233062744, "global_step": 36995, "epoch": 349, "lr": 2.0860658949585147e-05} +{"train_loss": 0.01497240923345089, "grad_norm": 0.10736845433712006, "global_step": 36996, "epoch": 349, "lr": 2.0858250569662345e-05} +{"train_loss": 0.015441181138157845, "grad_norm": 0.07624264061450958, "global_step": 36997, "epoch": 349, "lr": 2.085584229213099e-05} +{"train_loss": 0.016141187399625778, "grad_norm": 0.07595160603523254, "global_step": 36998, "epoch": 349, "lr": 2.0853434116999508e-05} +{"train_loss": 0.016093870624899864, "grad_norm": 0.0641016960144043, "global_step": 36999, "epoch": 349, "lr": 2.0851026044276406e-05} +{"train_loss": 0.013536152429878712, "grad_norm": 0.07883410900831223, "global_step": 37000, "epoch": 349, "lr": 2.0848618073970105e-05} +{"train_loss": 0.013505235314369202, "grad_norm": 0.1066712811589241, "global_step": 37001, "epoch": 349, "lr": 2.0846210206089094e-05} +{"train_loss": 0.01576385088264942, "grad_norm": 0.09086434543132782, "global_step": 37002, "epoch": 349, "lr": 2.0843802440641814e-05} +{"train_loss": 0.015019349753856659, "grad_norm": 0.11185147613286972, "global_step": 37003, "epoch": 349, "lr": 2.0841394777636724e-05} +{"train_loss": 0.012755809351801872, "grad_norm": 0.10440928488969803, "global_step": 37004, "epoch": 349, "lr": 2.0838987217082313e-05} +{"train_loss": 0.01594916358590126, "grad_norm": 0.0826413705945015, "global_step": 37005, "epoch": 349, "lr": 2.083657975898702e-05} +{"train_loss": 0.01670524850487709, "grad_norm": 0.1494445949792862, "global_step": 37006, "epoch": 349, "lr": 2.083417240335928e-05} +{"train_loss": 0.014130666851997375, "grad_norm": 0.14244365692138672, "global_step": 37007, "epoch": 349, "lr": 2.083176515020758e-05} +{"train_loss": 0.014553449116647243, "grad_norm": 0.09602102637290955, "global_step": 37008, "epoch": 349, "lr": 2.082935799954039e-05} +{"train_loss": 0.0148584870621562, "grad_norm": 0.09446502476930618, "global_step": 37009, "epoch": 349, "lr": 2.0826950951366124e-05} +{"train_loss": 0.017491044476628304, "grad_norm": 0.05958768352866173, "global_step": 37010, "epoch": 349, "lr": 2.0824544005693275e-05} +{"train_loss": 0.014776035211980343, "grad_norm": 0.07295626401901245, "global_step": 37011, "epoch": 349, "lr": 2.0822137162530325e-05} +{"train_loss": 0.016656067222356796, "grad_norm": 0.10514748841524124, "global_step": 37012, "epoch": 349, "lr": 2.081973042188566e-05} +{"train_loss": 0.014562102034687996, "grad_norm": 0.08360302448272705, "global_step": 37013, "epoch": 349, "lr": 2.0817323783767766e-05} +{"train_loss": 0.015528887510299683, "grad_norm": 0.08997555822134018, "global_step": 37014, "epoch": 349, "lr": 2.0814917248185105e-05} +{"train_loss": 0.017097417265176773, "grad_norm": 0.0916762575507164, "global_step": 37015, "epoch": 349, "lr": 2.0812510815146157e-05} +{"train_loss": 0.013332082889974117, "grad_norm": 0.07514391094446182, "global_step": 37016, "epoch": 349, "lr": 2.081010448465933e-05} +{"train_loss": 0.014866006560623646, "grad_norm": 0.09389295428991318, "global_step": 37017, "epoch": 349, "lr": 2.0807698256733127e-05} +{"train_loss": 0.014997012913227081, "grad_norm": 0.06778985261917114, "global_step": 37018, "epoch": 349, "lr": 2.080529213137597e-05} +{"train_loss": 0.011969255283474922, "grad_norm": 0.06451199948787689, "global_step": 37019, "epoch": 349, "lr": 2.0802886108596304e-05} +{"train_loss": 0.014315263368189335, "grad_norm": 0.11792471259832382, "global_step": 37020, "epoch": 349, "lr": 2.0800480188402598e-05} +{"train_loss": 0.014189150184392929, "grad_norm": 0.07646997272968292, "global_step": 37021, "epoch": 349, "lr": 2.0798074370803322e-05} +{"train_loss": 0.016658037900924683, "grad_norm": 0.07056497037410736, "global_step": 37022, "epoch": 349, "lr": 2.0795668655806915e-05} +{"train_loss": 0.014021921902894974, "grad_norm": 0.09474111348390579, "global_step": 37023, "epoch": 349, "lr": 2.079326304342181e-05} +{"train_loss": 0.015426757745444775, "grad_norm": 0.08781329542398453, "global_step": 37024, "epoch": 349, "lr": 2.0790857533656476e-05} +{"train_loss": 0.01758711412549019, "grad_norm": 0.08520973473787308, "global_step": 37025, "epoch": 349, "lr": 2.0788452126519388e-05} +{"train_loss": 0.01461282279342413, "grad_norm": 0.10648024827241898, "global_step": 37026, "epoch": 349, "lr": 2.0786046822018956e-05} +{"train_loss": 0.01497451402246952, "grad_norm": 0.17293769121170044, "global_step": 37027, "epoch": 349, "lr": 2.0783641620163647e-05} +{"train_loss": 0.014387216418981552, "grad_norm": 0.08034387230873108, "global_step": 37028, "epoch": 349, "lr": 2.0781236520961934e-05} +{"train_loss": 0.018801450729370117, "grad_norm": 0.16995692253112793, "global_step": 37029, "epoch": 349, "lr": 2.0778831524422255e-05} +{"train_loss": 0.015970714390277863, "grad_norm": 0.09850557148456573, "global_step": 37030, "epoch": 349, "lr": 2.077642663055303e-05} +{"train_loss": 0.01418948546051979, "grad_norm": 0.07333997637033463, "global_step": 37031, "epoch": 349, "lr": 2.0774021839362734e-05} +{"train_loss": 0.015492396429181099, "grad_norm": 0.07110777497291565, "global_step": 37032, "epoch": 349, "lr": 2.0771617150859833e-05} +{"train_loss": 0.01642433926463127, "grad_norm": 0.09406161308288574, "global_step": 37033, "epoch": 349, "lr": 2.0769212565052742e-05} +{"train_loss": 0.016111936420202255, "grad_norm": 0.0854995846748352, "global_step": 37034, "epoch": 349, "lr": 2.076680808194994e-05} +{"train_loss": 0.016089223325252533, "grad_norm": 0.06725438684225082, "global_step": 37035, "epoch": 349, "lr": 2.0764403701559865e-05} +{"train_loss": 0.016016434878110886, "grad_norm": 0.10678146034479141, "global_step": 37036, "epoch": 349, "lr": 2.0761999423890943e-05} +{"train_loss": 0.015177476219832897, "grad_norm": 0.08529746532440186, "global_step": 37037, "epoch": 349, "lr": 2.075959524895164e-05} +{"train_loss": 0.015586389228701591, "grad_norm": 0.05784967541694641, "global_step": 37038, "epoch": 349, "lr": 2.0757191176750416e-05} +{"train_loss": 0.015280075371265411, "grad_norm": 0.07536207884550095, "global_step": 37039, "epoch": 349, "lr": 2.0754787207295696e-05} +{"train_loss": 0.016222812235355377, "grad_norm": 0.05810744687914848, "global_step": 37040, "epoch": 349, "lr": 2.0752383340595948e-05} +{"train_loss": 0.014971860684454441, "grad_norm": 0.06577299535274506, "global_step": 37041, "epoch": 349, "lr": 2.0749979576659585e-05} +{"train_loss": 0.015327944420278072, "grad_norm": 0.08167684078216553, "global_step": 37042, "epoch": 349, "lr": 2.0747575915495093e-05} +{"train_loss": 0.013223621062934399, "grad_norm": 0.060182105749845505, "global_step": 37043, "epoch": 349, "lr": 2.0745172357110882e-05} +{"train_loss": 0.015325087122619152, "grad_norm": 0.09123089909553528, "global_step": 37044, "epoch": 349, "lr": 2.0742768901515414e-05} +{"train_loss": 0.015656325966119766, "grad_norm": 0.0815071165561676, "global_step": 37045, "epoch": 349, "lr": 2.0740365548717145e-05} +{"train_loss": 0.017719801515340805, "grad_norm": 0.10551118850708008, "global_step": 37046, "epoch": 349, "lr": 2.0737962298724512e-05} +{"train_loss": 0.014992764219641685, "grad_norm": 0.10327735543251038, "global_step": 37047, "epoch": 349, "lr": 2.073555915154593e-05} +{"train_loss": 0.013759792782366276, "grad_norm": 0.05964245647192001, "global_step": 37048, "epoch": 349, "lr": 2.073315610718987e-05} +{"train_loss": 0.01518038846552372, "grad_norm": 0.0832652822136879, "global_step": 37049, "epoch": 349, "lr": 2.073075316566479e-05} +{"train_loss": 0.013554856181144714, "grad_norm": 0.05860128998756409, "global_step": 37050, "epoch": 349, "lr": 2.0728350326979095e-05} +{"train_loss": 0.01438598707318306, "grad_norm": 0.07292146235704422, "global_step": 37051, "epoch": 349, "lr": 2.0725947591141247e-05} +{"train_loss": 0.01756521314382553, "grad_norm": 0.08000627905130386, "global_step": 37052, "epoch": 349, "lr": 2.072354495815973e-05} +{"train_loss": 0.014587108045816422, "grad_norm": 0.07386497408151627, "global_step": 37053, "epoch": 349, "lr": 2.07211424280429e-05} +{"train_loss": 0.013046516105532646, "grad_norm": 0.10628671944141388, "global_step": 37054, "epoch": 349, "lr": 2.0718740000799248e-05} +{"train_loss": 0.015825781971216202, "grad_norm": 0.058847010135650635, "global_step": 37055, "epoch": 349, "lr": 2.0716337676437227e-05} +{"train_loss": 0.015661029145121574, "grad_norm": 0.05722016096115112, "global_step": 37056, "epoch": 349, "lr": 2.071393545496524e-05} +{"train_loss": 0.016353370621800423, "grad_norm": 0.07896106690168381, "global_step": 37057, "epoch": 349, "lr": 2.071153333639175e-05} +{"train_loss": 0.01607823558151722, "grad_norm": 0.0750240683555603, "global_step": 37058, "epoch": 349, "lr": 2.0709131320725206e-05} +{"train_loss": 0.013642404228448868, "grad_norm": 0.1296941339969635, "global_step": 37059, "epoch": 349, "lr": 2.070672940797404e-05} +{"train_loss": 0.016241181641817093, "grad_norm": 0.07333854585886002, "global_step": 37060, "epoch": 349, "lr": 2.0704327598146673e-05} +{"train_loss": 0.01663573458790779, "grad_norm": 0.07537240535020828, "global_step": 37061, "epoch": 349, "lr": 2.0701925891251554e-05} +{"train_loss": 0.015685519203543663, "grad_norm": 0.06682965159416199, "global_step": 37062, "epoch": 349, "lr": 2.0699524287297146e-05} +{"train_loss": 0.014277379959821701, "grad_norm": 0.0736146941781044, "global_step": 37063, "epoch": 349, "lr": 2.069712278629185e-05} +{"train_loss": 0.015476152300834656, "grad_norm": 0.07875840365886688, "global_step": 37064, "epoch": 349, "lr": 2.0694721388244136e-05} +{"train_loss": 0.013545975089073181, "grad_norm": 0.05306807905435562, "global_step": 37065, "epoch": 349, "lr": 2.0692320093162416e-05} +{"train_loss": 0.013554281555116177, "grad_norm": 0.14755065739154816, "global_step": 37066, "epoch": 349, "lr": 2.068991890105515e-05} +{"train_loss": 0.015545886941254139, "grad_norm": 0.06772736459970474, "global_step": 37067, "epoch": 349, "lr": 2.068751781193074e-05} +{"train_loss": 0.012857099063694477, "grad_norm": 0.12402886152267456, "global_step": 37068, "epoch": 349, "lr": 2.068511682579765e-05} +{"train_loss": 0.012993873097002506, "grad_norm": 0.09083754569292068, "global_step": 37069, "epoch": 349, "lr": 2.068271594266433e-05} +{"train_loss": 0.016756631433963776, "grad_norm": 0.05285840854048729, "global_step": 37070, "epoch": 349, "lr": 2.0680315162539192e-05} +{"train_loss": 0.01191047765314579, "grad_norm": 0.07512977719306946, "global_step": 37071, "epoch": 349, "lr": 2.067791448543066e-05} +{"train_loss": 0.014214645139873028, "grad_norm": 0.05892254784703255, "global_step": 37072, "epoch": 349, "lr": 2.0675513911347188e-05} +{"train_loss": 0.014425301924347878, "grad_norm": 0.05390461906790733, "global_step": 37073, "epoch": 349, "lr": 2.067311344029722e-05} +{"train_loss": 0.014472085051238537, "grad_norm": 0.07252869009971619, "global_step": 37074, "epoch": 349, "lr": 2.0670713072289167e-05} +{"train_loss": 0.015560215339064598, "grad_norm": 0.10293087363243103, "global_step": 37075, "epoch": 349, "lr": 2.0668312807331486e-05} +{"train_loss": 0.01559352409094572, "grad_norm": 0.08602427691221237, "global_step": 37076, "epoch": 349, "lr": 2.0665912645432595e-05} +{"train_loss": 0.011798007413744926, "grad_norm": 0.10500917583703995, "global_step": 37077, "epoch": 349, "lr": 2.0663512586600913e-05} +{"train_loss": 0.017215287312865257, "grad_norm": 0.0690443143248558, "global_step": 37078, "epoch": 349, "lr": 2.066111263084489e-05} +{"train_loss": 0.01393578015267849, "grad_norm": 0.062295250594615936, "global_step": 37079, "epoch": 349, "lr": 2.0658712778172982e-05} +{"train_loss": 0.015142153948545456, "grad_norm": 0.059282463043928146, "global_step": 37080, "epoch": 349, "lr": 2.0656313028593572e-05} +{"train_loss": 0.014688494615256786, "grad_norm": 0.06478741765022278, "global_step": 37081, "epoch": 349, "lr": 2.065391338211514e-05} +{"train_loss": 0.01511871162801981, "grad_norm": 0.09502051770687103, "global_step": 37082, "epoch": 349, "lr": 2.0651513838746074e-05} +{"train_loss": 0.016561580821871758, "grad_norm": 0.08579555153846741, "global_step": 37083, "epoch": 349, "lr": 2.064911439849484e-05} +{"train_loss": 0.014415234327316284, "grad_norm": 0.08062423765659332, "global_step": 37084, "epoch": 349, "lr": 2.0646715061369842e-05} +{"train_loss": 0.01477506011724472, "grad_norm": 0.12078731507062912, "global_step": 37085, "epoch": 349, "lr": 2.064431582737952e-05} +{"train_loss": 0.013946409337222576, "grad_norm": 0.05879891663789749, "global_step": 37086, "epoch": 349, "lr": 2.0641916696532316e-05} +{"train_loss": 0.01584988459944725, "grad_norm": 0.06027105078101158, "global_step": 37087, "epoch": 349, "lr": 2.0639517668836654e-05} +{"train_loss": 0.016289690509438515, "grad_norm": 0.08108874410390854, "global_step": 37088, "epoch": 349, "lr": 2.0637118744300944e-05} +{"train_loss": 0.016403481364250183, "grad_norm": 0.06796444207429886, "global_step": 37089, "epoch": 349, "lr": 2.0634719922933628e-05} +{"train_loss": 0.017078498378396034, "grad_norm": 0.05667321756482124, "global_step": 37090, "epoch": 349, "lr": 2.063232120474315e-05} +{"train_loss": 0.014856049790978432, "grad_norm": 0.058841343969106674, "global_step": 37091, "epoch": 349, "lr": 2.062992258973791e-05} +{"train_loss": 0.018511870875954628, "grad_norm": 0.10628905892372131, "global_step": 37092, "epoch": 349, "lr": 2.0627524077926348e-05} +{"train_loss": 0.01479773223400116, "grad_norm": 0.06252488493919373, "global_step": 37093, "epoch": 349, "lr": 2.0625125669316926e-05} +{"train_loss": 0.016649337485432625, "grad_norm": 0.10473154485225677, "global_step": 37094, "epoch": 349, "lr": 2.0622727363918004e-05} +{"train_loss": 0.013612471520900726, "grad_norm": 0.09148749709129333, "global_step": 37095, "epoch": 349, "lr": 2.062032916173804e-05} +{"train_loss": 0.017233338207006454, "grad_norm": 0.07505015283823013, "global_step": 37096, "epoch": 349, "lr": 2.0617931062785483e-05} +{"train_loss": 0.01604664884507656, "grad_norm": 0.08550560474395752, "global_step": 37097, "epoch": 349, "lr": 2.0615533067068722e-05} +{"train_loss": 0.011343935504555702, "grad_norm": 0.11260294169187546, "global_step": 37098, "epoch": 349, "lr": 2.0613135174596194e-05} +{"train_loss": 0.015104795709463223, "grad_norm": 0.10314621776342392, "global_step": 37099, "epoch": 349, "lr": 2.061073738537635e-05, "val_loss": 0.019369138404726982} +{"train_loss": 0.013571553863584995, "grad_norm": 0.07297372817993164, "global_step": 37100, "epoch": 350, "lr": 2.060833969941759e-05} +{"train_loss": 0.013105696998536587, "grad_norm": 0.11922761797904968, "global_step": 37101, "epoch": 350, "lr": 2.0605942116728327e-05} +{"train_loss": 0.017094995826482773, "grad_norm": 0.08821994066238403, "global_step": 37102, "epoch": 350, "lr": 2.0603544637317002e-05} +{"train_loss": 0.015070187859237194, "grad_norm": 0.13605357706546783, "global_step": 37103, "epoch": 350, "lr": 2.0601147261192056e-05} +{"train_loss": 0.01585819013416767, "grad_norm": 0.06398101896047592, "global_step": 37104, "epoch": 350, "lr": 2.0598749988361872e-05} +{"train_loss": 0.013061728328466415, "grad_norm": 0.053982168436050415, "global_step": 37105, "epoch": 350, "lr": 2.059635281883491e-05} +{"train_loss": 0.016427572816610336, "grad_norm": 0.09181525558233261, "global_step": 37106, "epoch": 350, "lr": 2.0593955752619565e-05} +{"train_loss": 0.014417098835110664, "grad_norm": 0.16433551907539368, "global_step": 37107, "epoch": 350, "lr": 2.0591558789724287e-05} +{"train_loss": 0.01247837208211422, "grad_norm": 0.05611864849925041, "global_step": 37108, "epoch": 350, "lr": 2.0589161930157457e-05} +{"train_loss": 0.015995290130376816, "grad_norm": 0.06732069700956345, "global_step": 37109, "epoch": 350, "lr": 2.058676517392753e-05} +{"train_loss": 0.015168927609920502, "grad_norm": 0.08535642176866531, "global_step": 37110, "epoch": 350, "lr": 2.0584368521042934e-05} +{"train_loss": 0.01498680841177702, "grad_norm": 0.0816430076956749, "global_step": 37111, "epoch": 350, "lr": 2.0581971971512077e-05} +{"train_loss": 0.01555162575095892, "grad_norm": 0.09227336198091507, "global_step": 37112, "epoch": 350, "lr": 2.0579575525343353e-05} +{"train_loss": 0.01678437739610672, "grad_norm": 0.08149543404579163, "global_step": 37113, "epoch": 350, "lr": 2.0577179182545232e-05} +{"train_loss": 0.014576148241758347, "grad_norm": 0.08742262423038483, "global_step": 37114, "epoch": 350, "lr": 2.057478294312608e-05} +{"train_loss": 0.014499726705253124, "grad_norm": 0.10937093943357468, "global_step": 37115, "epoch": 350, "lr": 2.0572386807094353e-05} +{"train_loss": 0.01667352393269539, "grad_norm": 0.07066461443901062, "global_step": 37116, "epoch": 350, "lr": 2.0569990774458453e-05} +{"train_loss": 0.015819499269127846, "grad_norm": 0.080286905169487, "global_step": 37117, "epoch": 350, "lr": 2.0567594845226852e-05} +{"train_loss": 0.01596778631210327, "grad_norm": 0.12454894930124283, "global_step": 37118, "epoch": 350, "lr": 2.0565199019407878e-05} +{"train_loss": 0.015732310712337494, "grad_norm": 0.08380086719989777, "global_step": 37119, "epoch": 350, "lr": 2.0562803297009996e-05} +{"train_loss": 0.014408085495233536, "grad_norm": 0.136368528008461, "global_step": 37120, "epoch": 350, "lr": 2.0560407678041644e-05} +{"train_loss": 0.014009056612849236, "grad_norm": 0.07631585001945496, "global_step": 37121, "epoch": 350, "lr": 2.0558012162511197e-05} +{"train_loss": 0.013696675188839436, "grad_norm": 0.05708900839090347, "global_step": 37122, "epoch": 350, "lr": 2.055561675042711e-05} +{"train_loss": 0.014321133494377136, "grad_norm": 0.10497194528579712, "global_step": 37123, "epoch": 350, "lr": 2.055322144179776e-05} +{"train_loss": 0.015393024310469627, "grad_norm": 0.12290602177381516, "global_step": 37124, "epoch": 350, "lr": 2.0550826236631597e-05} +{"train_loss": 0.014387200586497784, "grad_norm": 0.08721940219402313, "global_step": 37125, "epoch": 350, "lr": 2.0548431134937013e-05} +{"train_loss": 0.01634378731250763, "grad_norm": 0.1031704694032669, "global_step": 37126, "epoch": 350, "lr": 2.054603613672243e-05} +{"train_loss": 0.014960125088691711, "grad_norm": 0.07913191616535187, "global_step": 37127, "epoch": 350, "lr": 2.0543641241996286e-05} +{"train_loss": 0.015296336263418198, "grad_norm": 0.11846063286066055, "global_step": 37128, "epoch": 350, "lr": 2.0541246450766982e-05} +{"train_loss": 0.013111565262079239, "grad_norm": 0.05641692876815796, "global_step": 37129, "epoch": 350, "lr": 2.05388517630429e-05} +{"train_loss": 0.014545487239956856, "grad_norm": 0.07720597088336945, "global_step": 37130, "epoch": 350, "lr": 2.0536457178832485e-05} +{"train_loss": 0.014879407361149788, "grad_norm": 0.11749260872602463, "global_step": 37131, "epoch": 350, "lr": 2.0534062698144162e-05} +{"train_loss": 0.013145992532372475, "grad_norm": 0.059601735323667526, "global_step": 37132, "epoch": 350, "lr": 2.0531668320986315e-05} +{"train_loss": 0.013591503724455833, "grad_norm": 0.10983694344758987, "global_step": 37133, "epoch": 350, "lr": 2.0529274047367365e-05} +{"train_loss": 0.013636717572808266, "grad_norm": 0.08210422843694687, "global_step": 37134, "epoch": 350, "lr": 2.052687987729577e-05} +{"train_loss": 0.014154846780002117, "grad_norm": 0.060297030955553055, "global_step": 37135, "epoch": 350, "lr": 2.0524485810779858e-05} +{"train_loss": 0.014860226772725582, "grad_norm": 0.12845711410045624, "global_step": 37136, "epoch": 350, "lr": 2.052209184782809e-05} +{"train_loss": 0.013575868681073189, "grad_norm": 0.0780043751001358, "global_step": 37137, "epoch": 350, "lr": 2.0519697988448887e-05} +{"train_loss": 0.015029845759272575, "grad_norm": 0.06332899630069733, "global_step": 37138, "epoch": 350, "lr": 2.051730423265063e-05} +{"train_loss": 0.015939926728606224, "grad_norm": 0.08794374018907547, "global_step": 37139, "epoch": 350, "lr": 2.0514910580441744e-05} +{"train_loss": 0.015348576009273529, "grad_norm": 0.06919409334659576, "global_step": 37140, "epoch": 350, "lr": 2.0512517031830658e-05} +{"train_loss": 0.015837032347917557, "grad_norm": 0.10443709045648575, "global_step": 37141, "epoch": 350, "lr": 2.051012358682576e-05} +{"train_loss": 0.013114922679960728, "grad_norm": 0.07548484206199646, "global_step": 37142, "epoch": 350, "lr": 2.0507730245435446e-05} +{"train_loss": 0.015398707240819931, "grad_norm": 0.08271364122629166, "global_step": 37143, "epoch": 350, "lr": 2.050533700766814e-05} +{"train_loss": 0.015914134681224823, "grad_norm": 0.16417478024959564, "global_step": 37144, "epoch": 350, "lr": 2.0502943873532276e-05} +{"train_loss": 0.014935700222849846, "grad_norm": 0.10625585168600082, "global_step": 37145, "epoch": 350, "lr": 2.050055084303622e-05} +{"train_loss": 0.014274775050580502, "grad_norm": 0.051531825214624405, "global_step": 37146, "epoch": 350, "lr": 2.0498157916188416e-05} +{"train_loss": 0.013820834457874298, "grad_norm": 0.08904503285884857, "global_step": 37147, "epoch": 350, "lr": 2.0495765092997237e-05} +{"train_loss": 0.01555501390248537, "grad_norm": 0.10043796896934509, "global_step": 37148, "epoch": 350, "lr": 2.0493372373471127e-05} +{"train_loss": 0.01460428535938263, "grad_norm": 0.08168181777000427, "global_step": 37149, "epoch": 350, "lr": 2.0490979757618455e-05} +{"train_loss": 0.014195149764418602, "grad_norm": 0.08588480949401855, "global_step": 37150, "epoch": 350, "lr": 2.0488587245447648e-05} +{"train_loss": 0.01636289432644844, "grad_norm": 0.08912023901939392, "global_step": 37151, "epoch": 350, "lr": 2.0486194836967127e-05} +{"train_loss": 0.015013589523732662, "grad_norm": 0.11756904423236847, "global_step": 37152, "epoch": 350, "lr": 2.0483802532185286e-05} +{"train_loss": 0.015810616314411163, "grad_norm": 0.1159689798951149, "global_step": 37153, "epoch": 350, "lr": 2.0481410331110506e-05} +{"train_loss": 0.01535031758248806, "grad_norm": 0.06374131888151169, "global_step": 37154, "epoch": 350, "lr": 2.0479018233751228e-05} +{"train_loss": 0.01616097241640091, "grad_norm": 0.0836285725235939, "global_step": 37155, "epoch": 350, "lr": 2.047662624011583e-05} +{"train_loss": 0.015875939279794693, "grad_norm": 0.11279846727848053, "global_step": 37156, "epoch": 350, "lr": 2.047423435021273e-05} +{"train_loss": 0.014278200455009937, "grad_norm": 0.13600298762321472, "global_step": 37157, "epoch": 350, "lr": 2.0471842564050343e-05} +{"train_loss": 0.01513010635972023, "grad_norm": 0.10742972791194916, "global_step": 37158, "epoch": 350, "lr": 2.046945088163706e-05} +{"train_loss": 0.016315415501594543, "grad_norm": 0.07607005536556244, "global_step": 37159, "epoch": 350, "lr": 2.0467059302981263e-05} +{"train_loss": 0.01704290695488453, "grad_norm": 0.06292667239904404, "global_step": 37160, "epoch": 350, "lr": 2.0464667828091377e-05} +{"train_loss": 0.014985090121626854, "grad_norm": 0.1315765380859375, "global_step": 37161, "epoch": 350, "lr": 2.0462276456975827e-05} +{"train_loss": 0.014784036204218864, "grad_norm": 0.09155043959617615, "global_step": 37162, "epoch": 350, "lr": 2.0459885189642968e-05} +{"train_loss": 0.014671209268271923, "grad_norm": 0.10543297976255417, "global_step": 37163, "epoch": 350, "lr": 2.045749402610123e-05} +{"train_loss": 0.014843396842479706, "grad_norm": 0.12485262006521225, "global_step": 37164, "epoch": 350, "lr": 2.045510296635903e-05} +{"train_loss": 0.016504820436239243, "grad_norm": 0.11410924047231674, "global_step": 37165, "epoch": 350, "lr": 2.0452712010424747e-05} +{"train_loss": 0.013350537046790123, "grad_norm": 0.05209076777100563, "global_step": 37166, "epoch": 350, "lr": 2.0450321158306762e-05} +{"train_loss": 0.020398175343871117, "grad_norm": 0.06138802692294121, "global_step": 37167, "epoch": 350, "lr": 2.0447930410013498e-05} +{"train_loss": 0.014692584052681923, "grad_norm": 0.07368933409452438, "global_step": 37168, "epoch": 350, "lr": 2.0445539765553378e-05} +{"train_loss": 0.013311024755239487, "grad_norm": 0.16150809824466705, "global_step": 37169, "epoch": 350, "lr": 2.044314922493476e-05} +{"train_loss": 0.015880301594734192, "grad_norm": 0.10657406598329544, "global_step": 37170, "epoch": 350, "lr": 2.0440758788166076e-05} +{"train_loss": 0.012911037541925907, "grad_norm": 0.09040313959121704, "global_step": 37171, "epoch": 350, "lr": 2.0438368455255696e-05} +{"train_loss": 0.019881900399923325, "grad_norm": 0.13296210765838623, "global_step": 37172, "epoch": 350, "lr": 2.043597822621205e-05} +{"train_loss": 0.014662995003163815, "grad_norm": 0.17079603672027588, "global_step": 37173, "epoch": 350, "lr": 2.0433588101043506e-05} +{"train_loss": 0.014305811375379562, "grad_norm": 0.1403174251317978, "global_step": 37174, "epoch": 350, "lr": 2.043119807975847e-05} +{"train_loss": 0.01458175852894783, "grad_norm": 0.0733814537525177, "global_step": 37175, "epoch": 350, "lr": 2.0428808162365376e-05} +{"train_loss": 0.01905105635523796, "grad_norm": 0.20681053400039673, "global_step": 37176, "epoch": 350, "lr": 2.042641834887256e-05} +{"train_loss": 0.016044076532125473, "grad_norm": 0.13332833349704742, "global_step": 37177, "epoch": 350, "lr": 2.042402863928845e-05} +{"train_loss": 0.017168182879686356, "grad_norm": 0.07687952369451523, "global_step": 37178, "epoch": 350, "lr": 2.042163903362146e-05} +{"train_loss": 0.014415334910154343, "grad_norm": 0.05079862102866173, "global_step": 37179, "epoch": 350, "lr": 2.0419249531879943e-05} +{"train_loss": 0.017205756157636642, "grad_norm": 0.1237616240978241, "global_step": 37180, "epoch": 350, "lr": 2.0416860134072324e-05} +{"train_loss": 0.014922287315130234, "grad_norm": 0.12522314488887787, "global_step": 37181, "epoch": 350, "lr": 2.0414470840207007e-05} +{"train_loss": 0.018054403364658356, "grad_norm": 0.06537441164255142, "global_step": 37182, "epoch": 350, "lr": 2.0412081650292376e-05} +{"train_loss": 0.01379399560391903, "grad_norm": 0.0702696219086647, "global_step": 37183, "epoch": 350, "lr": 2.0409692564336806e-05} +{"train_loss": 0.013539972715079784, "grad_norm": 0.10290946066379547, "global_step": 37184, "epoch": 350, "lr": 2.0407303582348702e-05} +{"train_loss": 0.010338208638131618, "grad_norm": 0.05428566038608551, "global_step": 37185, "epoch": 350, "lr": 2.0404914704336485e-05} +{"train_loss": 0.014201805926859379, "grad_norm": 0.11773628741502762, "global_step": 37186, "epoch": 350, "lr": 2.0402525930308512e-05} +{"train_loss": 0.018322892487049103, "grad_norm": 0.1664290726184845, "global_step": 37187, "epoch": 350, "lr": 2.0400137260273207e-05} +{"train_loss": 0.017030252143740654, "grad_norm": 0.09787598997354507, "global_step": 37188, "epoch": 350, "lr": 2.039774869423893e-05} +{"train_loss": 0.013948844745755196, "grad_norm": 0.0764726847410202, "global_step": 37189, "epoch": 350, "lr": 2.0395360232214105e-05} +{"train_loss": 0.015589597634971142, "grad_norm": 0.0914464145898819, "global_step": 37190, "epoch": 350, "lr": 2.0392971874207096e-05} +{"train_loss": 0.014224501326680183, "grad_norm": 0.061596307903528214, "global_step": 37191, "epoch": 350, "lr": 2.0390583620226307e-05} +{"train_loss": 0.01482525747269392, "grad_norm": 0.09556595236063004, "global_step": 37192, "epoch": 350, "lr": 2.0388195470280148e-05} +{"train_loss": 0.015012325718998909, "grad_norm": 0.06807111203670502, "global_step": 37193, "epoch": 350, "lr": 2.038580742437699e-05} +{"train_loss": 0.017352327704429626, "grad_norm": 0.07763655483722687, "global_step": 37194, "epoch": 350, "lr": 2.038341948252521e-05} +{"train_loss": 0.016712376847863197, "grad_norm": 0.08280287683010101, "global_step": 37195, "epoch": 350, "lr": 2.0381031644733234e-05} +{"train_loss": 0.01709502376616001, "grad_norm": 0.13368387520313263, "global_step": 37196, "epoch": 350, "lr": 2.037864391100941e-05} +{"train_loss": 0.012192674912512302, "grad_norm": 0.09739664942026138, "global_step": 37197, "epoch": 350, "lr": 2.037625628136215e-05} +{"train_loss": 0.014823145233094692, "grad_norm": 0.07374328374862671, "global_step": 37198, "epoch": 350, "lr": 2.0373868755799863e-05} +{"train_loss": 0.015252446755766869, "grad_norm": 0.06220299378037453, "global_step": 37199, "epoch": 350, "lr": 2.0371481334330912e-05} +{"train_loss": 0.01641138084232807, "grad_norm": 0.1192854568362236, "global_step": 37200, "epoch": 350, "lr": 2.0369094016963676e-05} +{"train_loss": 0.013190625235438347, "grad_norm": 0.07338571548461914, "global_step": 37201, "epoch": 350, "lr": 2.036670680370656e-05} +{"train_loss": 0.018073465675115585, "grad_norm": 0.07614129781723022, "global_step": 37202, "epoch": 350, "lr": 2.036431969456796e-05} +{"train_loss": 0.013460902497172356, "grad_norm": 0.05677418410778046, "global_step": 37203, "epoch": 350, "lr": 2.036193268955624e-05} +{"train_loss": 0.014239758253097534, "grad_norm": 0.06445928663015366, "global_step": 37204, "epoch": 350, "lr": 2.0359545788679797e-05} +{"train_loss": 0.015105253928955996, "grad_norm": 0.13420230150222778, "global_step": 37205, "epoch": 350, "lr": 2.035715899194704e-05, "train_batch_mse_error": 0.01359952986240387, "val_loss": 0.019514909014105797} +{"train_loss": 0.011961903423070908, "grad_norm": 0.06474556773900986, "global_step": 37206, "epoch": 351, "lr": 2.035477229936633e-05} +{"train_loss": 0.014670206233859062, "grad_norm": 0.09082638472318649, "global_step": 37207, "epoch": 351, "lr": 2.035238571094605e-05} +{"train_loss": 0.01326829381287098, "grad_norm": 0.08554904162883759, "global_step": 37208, "epoch": 351, "lr": 2.0349999226694583e-05} +{"train_loss": 0.015099853277206421, "grad_norm": 0.07107328623533249, "global_step": 37209, "epoch": 351, "lr": 2.0347612846620345e-05} +{"train_loss": 0.013411572203040123, "grad_norm": 0.07773575931787491, "global_step": 37210, "epoch": 351, "lr": 2.0345226570731685e-05} +{"train_loss": 0.013310633599758148, "grad_norm": 0.04868510365486145, "global_step": 37211, "epoch": 351, "lr": 2.0342840399037015e-05} +{"train_loss": 0.015212921425700188, "grad_norm": 0.07401975244283676, "global_step": 37212, "epoch": 351, "lr": 2.034045433154471e-05} +{"train_loss": 0.013903223909437656, "grad_norm": 0.09071651101112366, "global_step": 37213, "epoch": 351, "lr": 2.0338068368263136e-05} +{"train_loss": 0.016138339415192604, "grad_norm": 0.08005373179912567, "global_step": 37214, "epoch": 351, "lr": 2.0335682509200683e-05} +{"train_loss": 0.013863308355212212, "grad_norm": 0.06849809736013412, "global_step": 37215, "epoch": 351, "lr": 2.033329675436576e-05} +{"train_loss": 0.014003245159983635, "grad_norm": 0.1572883278131485, "global_step": 37216, "epoch": 351, "lr": 2.0330911103766715e-05} +{"train_loss": 0.01594405248761177, "grad_norm": 0.11445306241512299, "global_step": 37217, "epoch": 351, "lr": 2.0328525557411964e-05} +{"train_loss": 0.01635296642780304, "grad_norm": 0.0558057464659214, "global_step": 37218, "epoch": 351, "lr": 2.032614011530985e-05} +{"train_loss": 0.016384603455662727, "grad_norm": 0.08176332712173462, "global_step": 37219, "epoch": 351, "lr": 2.0323754777468795e-05} +{"train_loss": 0.013186762109398842, "grad_norm": 0.09839311242103577, "global_step": 37220, "epoch": 351, "lr": 2.032136954389714e-05} +{"train_loss": 0.014222703874111176, "grad_norm": 0.10347900539636612, "global_step": 37221, "epoch": 351, "lr": 2.0318984414603288e-05} +{"train_loss": 0.0152822844684124, "grad_norm": 0.07074463367462158, "global_step": 37222, "epoch": 351, "lr": 2.031659938959563e-05} +{"train_loss": 0.01585071161389351, "grad_norm": 0.06975514441728592, "global_step": 37223, "epoch": 351, "lr": 2.0314214468882536e-05} +{"train_loss": 0.013545222580432892, "grad_norm": 0.06611430644989014, "global_step": 37224, "epoch": 351, "lr": 2.0311829652472363e-05} +{"train_loss": 0.015507003292441368, "grad_norm": 0.11568906903266907, "global_step": 37225, "epoch": 351, "lr": 2.030944494037351e-05} +{"train_loss": 0.015416407957673073, "grad_norm": 0.07533619552850723, "global_step": 37226, "epoch": 351, "lr": 2.030706033259438e-05} +{"train_loss": 0.015777401626110077, "grad_norm": 0.08816660940647125, "global_step": 37227, "epoch": 351, "lr": 2.03046758291433e-05} +{"train_loss": 0.0130006093531847, "grad_norm": 0.07827256619930267, "global_step": 37228, "epoch": 351, "lr": 2.0302291430028696e-05} +{"train_loss": 0.016375374048948288, "grad_norm": 0.06719820201396942, "global_step": 37229, "epoch": 351, "lr": 2.0299907135258904e-05} +{"train_loss": 0.01502541359513998, "grad_norm": 0.08342679589986801, "global_step": 37230, "epoch": 351, "lr": 2.0297522944842346e-05} +{"train_loss": 0.014712220057845116, "grad_norm": 0.0710865929722786, "global_step": 37231, "epoch": 351, "lr": 2.0295138858787354e-05} +{"train_loss": 0.0185604989528656, "grad_norm": 0.21670758724212646, "global_step": 37232, "epoch": 351, "lr": 2.0292754877102326e-05} +{"train_loss": 0.016015367582440376, "grad_norm": 0.08768267929553986, "global_step": 37233, "epoch": 351, "lr": 2.0290370999795654e-05} +{"train_loss": 0.01408841460943222, "grad_norm": 0.09083002060651779, "global_step": 37234, "epoch": 351, "lr": 2.02879872268757e-05} +{"train_loss": 0.014552436769008636, "grad_norm": 0.07188694924116135, "global_step": 37235, "epoch": 351, "lr": 2.028560355835082e-05} +{"train_loss": 0.015359356999397278, "grad_norm": 0.1231355145573616, "global_step": 37236, "epoch": 351, "lr": 2.0283219994229423e-05} +{"train_loss": 0.014772657305002213, "grad_norm": 0.10786810517311096, "global_step": 37237, "epoch": 351, "lr": 2.0280836534519853e-05} +{"train_loss": 0.01739683747291565, "grad_norm": 0.08567865192890167, "global_step": 37238, "epoch": 351, "lr": 2.0278453179230495e-05} +{"train_loss": 0.017087850719690323, "grad_norm": 0.1210923120379448, "global_step": 37239, "epoch": 351, "lr": 2.0276069928369746e-05} +{"train_loss": 0.014141825959086418, "grad_norm": 0.12991195917129517, "global_step": 37240, "epoch": 351, "lr": 2.027368678194596e-05} +{"train_loss": 0.015390460379421711, "grad_norm": 0.10917489230632782, "global_step": 37241, "epoch": 351, "lr": 2.0271303739967497e-05} +{"train_loss": 0.01490683201700449, "grad_norm": 0.056272946298122406, "global_step": 37242, "epoch": 351, "lr": 2.0268920802442737e-05} +{"train_loss": 0.013639481738209724, "grad_norm": 0.09483636170625687, "global_step": 37243, "epoch": 351, "lr": 2.0266537969380085e-05} +{"train_loss": 0.013404490426182747, "grad_norm": 0.09761523455381393, "global_step": 37244, "epoch": 351, "lr": 2.0264155240787864e-05} +{"train_loss": 0.014548499137163162, "grad_norm": 0.10588551312685013, "global_step": 37245, "epoch": 351, "lr": 2.0261772616674473e-05} +{"train_loss": 0.016294725239276886, "grad_norm": 0.12846849858760834, "global_step": 37246, "epoch": 351, "lr": 2.02593900970483e-05} +{"train_loss": 0.018717598170042038, "grad_norm": 0.06850339472293854, "global_step": 37247, "epoch": 351, "lr": 2.025700768191769e-05} +{"train_loss": 0.013248358853161335, "grad_norm": 0.08418890088796616, "global_step": 37248, "epoch": 351, "lr": 2.0254625371291003e-05} +{"train_loss": 0.01660984754562378, "grad_norm": 0.06868692487478256, "global_step": 37249, "epoch": 351, "lr": 2.0252243165176632e-05} +{"train_loss": 0.015107596293091774, "grad_norm": 0.05786444991827011, "global_step": 37250, "epoch": 351, "lr": 2.0249861063582952e-05} +{"train_loss": 0.01515567023307085, "grad_norm": 0.10696105659008026, "global_step": 37251, "epoch": 351, "lr": 2.024747906651831e-05} +{"train_loss": 0.015753567218780518, "grad_norm": 0.15271276235580444, "global_step": 37252, "epoch": 351, "lr": 2.0245097173991107e-05} +{"train_loss": 0.01536948699504137, "grad_norm": 0.09871488064527512, "global_step": 37253, "epoch": 351, "lr": 2.024271538600969e-05} +{"train_loss": 0.012580649927258492, "grad_norm": 0.06416299194097519, "global_step": 37254, "epoch": 351, "lr": 2.024033370258241e-05} +{"train_loss": 0.014069898054003716, "grad_norm": 0.09404736012220383, "global_step": 37255, "epoch": 351, "lr": 2.0237952123717656e-05} +{"train_loss": 0.014661970548331738, "grad_norm": 0.06973069161176682, "global_step": 37256, "epoch": 351, "lr": 2.0235570649423818e-05} +{"train_loss": 0.016110248863697052, "grad_norm": 0.08548387885093689, "global_step": 37257, "epoch": 351, "lr": 2.023318927970922e-05} +{"train_loss": 0.013986892066895962, "grad_norm": 0.06447511166334152, "global_step": 37258, "epoch": 351, "lr": 2.0230808014582263e-05} +{"train_loss": 0.01504584401845932, "grad_norm": 0.09191323816776276, "global_step": 37259, "epoch": 351, "lr": 2.0228426854051285e-05} +{"train_loss": 0.01601504161953926, "grad_norm": 0.08424891531467438, "global_step": 37260, "epoch": 351, "lr": 2.0226045798124688e-05} +{"train_loss": 0.016123929992318153, "grad_norm": 0.07423370331525803, "global_step": 37261, "epoch": 351, "lr": 2.0223664846810797e-05} +{"train_loss": 0.018501413986086845, "grad_norm": 0.11435792595148087, "global_step": 37262, "epoch": 351, "lr": 2.0221284000118e-05} +{"train_loss": 0.014026029035449028, "grad_norm": 0.06225345656275749, "global_step": 37263, "epoch": 351, "lr": 2.0218903258054678e-05} +{"train_loss": 0.014905019663274288, "grad_norm": 0.06082699075341225, "global_step": 37264, "epoch": 351, "lr": 2.021652262062918e-05} +{"train_loss": 0.014696482568979263, "grad_norm": 0.08640650659799576, "global_step": 37265, "epoch": 351, "lr": 2.0214142087849848e-05} +{"train_loss": 0.016187623143196106, "grad_norm": 0.09529990702867508, "global_step": 37266, "epoch": 351, "lr": 2.021176165972507e-05} +{"train_loss": 0.017710627987980843, "grad_norm": 0.07546798139810562, "global_step": 37267, "epoch": 351, "lr": 2.020938133626322e-05} +{"train_loss": 0.015677284449338913, "grad_norm": 0.05917823314666748, "global_step": 37268, "epoch": 351, "lr": 2.020700111747264e-05} +{"train_loss": 0.012215055525302887, "grad_norm": 0.0505346916615963, "global_step": 37269, "epoch": 351, "lr": 2.0204621003361695e-05} +{"train_loss": 0.014252051711082458, "grad_norm": 0.08616871386766434, "global_step": 37270, "epoch": 351, "lr": 2.0202240993938788e-05} +{"train_loss": 0.013010035268962383, "grad_norm": 0.19269727170467377, "global_step": 37271, "epoch": 351, "lr": 2.0199861089212213e-05} +{"train_loss": 0.01584414765238762, "grad_norm": 0.0628778487443924, "global_step": 37272, "epoch": 351, "lr": 2.0197481289190366e-05} +{"train_loss": 0.017783205956220627, "grad_norm": 0.09032753854990005, "global_step": 37273, "epoch": 351, "lr": 2.0195101593881615e-05} +{"train_loss": 0.014224158599972725, "grad_norm": 0.09963900595903397, "global_step": 37274, "epoch": 351, "lr": 2.019272200329433e-05} +{"train_loss": 0.014685338363051414, "grad_norm": 0.06900092959403992, "global_step": 37275, "epoch": 351, "lr": 2.0190342517436856e-05} +{"train_loss": 0.015221856534481049, "grad_norm": 0.07953610271215439, "global_step": 37276, "epoch": 351, "lr": 2.0187963136317535e-05} +{"train_loss": 0.014551163651049137, "grad_norm": 0.1103963777422905, "global_step": 37277, "epoch": 351, "lr": 2.018558385994476e-05} +{"train_loss": 0.015848929062485695, "grad_norm": 0.06821571290493011, "global_step": 37278, "epoch": 351, "lr": 2.018320468832687e-05} +{"train_loss": 0.014721864834427834, "grad_norm": 0.07508158683776855, "global_step": 37279, "epoch": 351, "lr": 2.0180825621472227e-05} +{"train_loss": 0.016279499977827072, "grad_norm": 0.06458128243684769, "global_step": 37280, "epoch": 351, "lr": 2.0178446659389212e-05} +{"train_loss": 0.015277300029993057, "grad_norm": 0.08086393028497696, "global_step": 37281, "epoch": 351, "lr": 2.017606780208617e-05} +{"train_loss": 0.01522507518529892, "grad_norm": 0.0853792130947113, "global_step": 37282, "epoch": 351, "lr": 2.017368904957143e-05} +{"train_loss": 0.01442708633840084, "grad_norm": 0.1073053851723671, "global_step": 37283, "epoch": 351, "lr": 2.0171310401853384e-05} +{"train_loss": 0.014978348277509212, "grad_norm": 0.06044449657201767, "global_step": 37284, "epoch": 351, "lr": 2.016893185894039e-05} +{"train_loss": 0.012983397580683231, "grad_norm": 0.08893825113773346, "global_step": 37285, "epoch": 351, "lr": 2.016655342084079e-05} +{"train_loss": 0.014574413187801838, "grad_norm": 0.1210906058549881, "global_step": 37286, "epoch": 351, "lr": 2.0164175087562943e-05} +{"train_loss": 0.016170881688594818, "grad_norm": 0.11416439712047577, "global_step": 37287, "epoch": 351, "lr": 2.016179685911522e-05} +{"train_loss": 0.016782253980636597, "grad_norm": 0.06479097157716751, "global_step": 37288, "epoch": 351, "lr": 2.0159418735505974e-05} +{"train_loss": 0.015121575444936752, "grad_norm": 0.05845057591795921, "global_step": 37289, "epoch": 351, "lr": 2.015704071674353e-05} +{"train_loss": 0.014653733931481838, "grad_norm": 0.0627213567495346, "global_step": 37290, "epoch": 351, "lr": 2.015466280283627e-05} +{"train_loss": 0.01490960456430912, "grad_norm": 0.08145781606435776, "global_step": 37291, "epoch": 351, "lr": 2.0152284993792566e-05} +{"train_loss": 0.01456825714558363, "grad_norm": 0.06891331076622009, "global_step": 37292, "epoch": 351, "lr": 2.0149907289620728e-05} +{"train_loss": 0.01508010271936655, "grad_norm": 0.06909523159265518, "global_step": 37293, "epoch": 351, "lr": 2.0147529690329158e-05} +{"train_loss": 0.018679561093449593, "grad_norm": 0.07569029182195663, "global_step": 37294, "epoch": 351, "lr": 2.014515219592618e-05} +{"train_loss": 0.012592273764312267, "grad_norm": 0.08063603192567825, "global_step": 37295, "epoch": 351, "lr": 2.0142774806420145e-05} +{"train_loss": 0.014430303126573563, "grad_norm": 0.0776120126247406, "global_step": 37296, "epoch": 351, "lr": 2.014039752181941e-05} +{"train_loss": 0.014365937560796738, "grad_norm": 0.06675240397453308, "global_step": 37297, "epoch": 351, "lr": 2.0138020342132346e-05} +{"train_loss": 0.014881102368235588, "grad_norm": 0.06055600196123123, "global_step": 37298, "epoch": 351, "lr": 2.0135643267367276e-05} +{"train_loss": 0.0153329037129879, "grad_norm": 0.08422297239303589, "global_step": 37299, "epoch": 351, "lr": 2.0133266297532587e-05} +{"train_loss": 0.01564391702413559, "grad_norm": 0.050066955387592316, "global_step": 37300, "epoch": 351, "lr": 2.0130889432636597e-05} +{"train_loss": 0.014490652829408646, "grad_norm": 0.07374975085258484, "global_step": 37301, "epoch": 351, "lr": 2.012851267268769e-05} +{"train_loss": 0.015730632469058037, "grad_norm": 0.06201651692390442, "global_step": 37302, "epoch": 351, "lr": 2.0126136017694175e-05} +{"train_loss": 0.014089818112552166, "grad_norm": 0.06216459348797798, "global_step": 37303, "epoch": 351, "lr": 2.0123759467664427e-05} +{"train_loss": 0.0134257972240448, "grad_norm": 0.0749983936548233, "global_step": 37304, "epoch": 351, "lr": 2.0121383022606815e-05} +{"train_loss": 0.016011975705623627, "grad_norm": 0.057481423020362854, "global_step": 37305, "epoch": 351, "lr": 2.0119006682529674e-05} +{"train_loss": 0.016533205285668373, "grad_norm": 0.09835734963417053, "global_step": 37306, "epoch": 351, "lr": 2.011663044744132e-05} +{"train_loss": 0.014152681455016136, "grad_norm": 0.06018529459834099, "global_step": 37307, "epoch": 351, "lr": 2.011425431735014e-05} +{"train_loss": 0.01714090071618557, "grad_norm": 0.12787316739559174, "global_step": 37308, "epoch": 351, "lr": 2.0111878292264486e-05} +{"train_loss": 0.015126565471291542, "grad_norm": 0.13180720806121826, "global_step": 37309, "epoch": 351, "lr": 2.010950237219268e-05} +{"train_loss": 0.017108162865042686, "grad_norm": 0.12095516920089722, "global_step": 37310, "epoch": 351, "lr": 2.010712655714308e-05} +{"train_loss": 0.015110380666435889, "grad_norm": 0.10959157347679138, "global_step": 37311, "epoch": 351, "lr": 2.0104750847124075e-05, "val_loss": 0.019989321008324623} +{"train_loss": 0.01684761233627796, "grad_norm": 0.18057148158550262, "global_step": 37312, "epoch": 352, "lr": 2.010237524214394e-05} +{"train_loss": 0.01705075241625309, "grad_norm": 0.07689353823661804, "global_step": 37313, "epoch": 352, "lr": 2.009999974221105e-05} +{"train_loss": 0.01466294378042221, "grad_norm": 0.09780681133270264, "global_step": 37314, "epoch": 352, "lr": 2.0097624347333783e-05} +{"train_loss": 0.01611311361193657, "grad_norm": 0.07952713221311569, "global_step": 37315, "epoch": 352, "lr": 2.0095249057520436e-05} +{"train_loss": 0.018668310716748238, "grad_norm": 0.12047042697668076, "global_step": 37316, "epoch": 352, "lr": 2.0092873872779383e-05} +{"train_loss": 0.013846023008227348, "grad_norm": 0.096535325050354, "global_step": 37317, "epoch": 352, "lr": 2.009049879311898e-05} +{"train_loss": 0.014933327212929726, "grad_norm": 0.07462475448846817, "global_step": 37318, "epoch": 352, "lr": 2.008812381854756e-05} +{"train_loss": 0.016072288155555725, "grad_norm": 0.10631527006626129, "global_step": 37319, "epoch": 352, "lr": 2.0085748949073446e-05} +{"train_loss": 0.01711437478661537, "grad_norm": 0.06477344781160355, "global_step": 37320, "epoch": 352, "lr": 2.0083374184704994e-05} +{"train_loss": 0.011724007315933704, "grad_norm": 0.08243925869464874, "global_step": 37321, "epoch": 352, "lr": 2.0080999525450584e-05} +{"train_loss": 0.016254626214504242, "grad_norm": 0.08874466270208359, "global_step": 37322, "epoch": 352, "lr": 2.0078624971318522e-05} +{"train_loss": 0.01409159041941166, "grad_norm": 0.07808640599250793, "global_step": 37323, "epoch": 352, "lr": 2.007625052231714e-05} +{"train_loss": 0.017222844064235687, "grad_norm": 0.08588097244501114, "global_step": 37324, "epoch": 352, "lr": 2.0073876178454808e-05} +{"train_loss": 0.0174851231276989, "grad_norm": 0.08979761600494385, "global_step": 37325, "epoch": 352, "lr": 2.0071501939739872e-05} +{"train_loss": 0.014400636777281761, "grad_norm": 0.08067449182271957, "global_step": 37326, "epoch": 352, "lr": 2.0069127806180648e-05} +{"train_loss": 0.015086748637259007, "grad_norm": 0.11492733657360077, "global_step": 37327, "epoch": 352, "lr": 2.0066753777785497e-05} +{"train_loss": 0.014518067240715027, "grad_norm": 0.09110989421606064, "global_step": 37328, "epoch": 352, "lr": 2.0064379854562765e-05} +{"train_loss": 0.015253698453307152, "grad_norm": 0.08372394740581512, "global_step": 37329, "epoch": 352, "lr": 2.0062006036520786e-05} +{"train_loss": 0.013879222795367241, "grad_norm": 0.10007822513580322, "global_step": 37330, "epoch": 352, "lr": 2.0059632323667882e-05} +{"train_loss": 0.014627307653427124, "grad_norm": 0.07304048538208008, "global_step": 37331, "epoch": 352, "lr": 2.0057258716012407e-05} +{"train_loss": 0.016063200309872627, "grad_norm": 0.0694313496351242, "global_step": 37332, "epoch": 352, "lr": 2.0054885213562724e-05} +{"train_loss": 0.018523138016462326, "grad_norm": 0.08125519007444382, "global_step": 37333, "epoch": 352, "lr": 2.005251181632713e-05} +{"train_loss": 0.016588471829891205, "grad_norm": 0.07204525172710419, "global_step": 37334, "epoch": 352, "lr": 2.0050138524314005e-05} +{"train_loss": 0.017550338059663773, "grad_norm": 0.06809922307729721, "global_step": 37335, "epoch": 352, "lr": 2.004776533753167e-05} +{"train_loss": 0.01645747944712639, "grad_norm": 0.08429663628339767, "global_step": 37336, "epoch": 352, "lr": 2.004539225598844e-05} +{"train_loss": 0.015266498550772667, "grad_norm": 0.11583888530731201, "global_step": 37337, "epoch": 352, "lr": 2.004301927969268e-05} +{"train_loss": 0.012732128612697124, "grad_norm": 0.062003590166568756, "global_step": 37338, "epoch": 352, "lr": 2.0040646408652734e-05} +{"train_loss": 0.015457039698958397, "grad_norm": 0.05522169545292854, "global_step": 37339, "epoch": 352, "lr": 2.0038273642876916e-05} +{"train_loss": 0.018540089949965477, "grad_norm": 0.12172888219356537, "global_step": 37340, "epoch": 352, "lr": 2.003590098237359e-05} +{"train_loss": 0.0174710750579834, "grad_norm": 0.06701476871967316, "global_step": 37341, "epoch": 352, "lr": 2.0033528427151062e-05} +{"train_loss": 0.014533350244164467, "grad_norm": 0.10782584547996521, "global_step": 37342, "epoch": 352, "lr": 2.0031155977217696e-05} +{"train_loss": 0.0156156150624156, "grad_norm": 0.07395528256893158, "global_step": 37343, "epoch": 352, "lr": 2.00287836325818e-05} +{"train_loss": 0.01464671641588211, "grad_norm": 0.06967902928590775, "global_step": 37344, "epoch": 352, "lr": 2.002641139325172e-05} +{"train_loss": 0.013336527161300182, "grad_norm": 0.0922873392701149, "global_step": 37345, "epoch": 352, "lr": 2.0024039259235816e-05} +{"train_loss": 0.016047926619648933, "grad_norm": 0.1169096827507019, "global_step": 37346, "epoch": 352, "lr": 2.00216672305424e-05} +{"train_loss": 0.01268504187464714, "grad_norm": 0.09462682157754898, "global_step": 37347, "epoch": 352, "lr": 2.001929530717979e-05} +{"train_loss": 0.015465321019291878, "grad_norm": 0.09743131697177887, "global_step": 37348, "epoch": 352, "lr": 2.001692348915634e-05} +{"train_loss": 0.01415560208261013, "grad_norm": 0.07373719662427902, "global_step": 37349, "epoch": 352, "lr": 2.00145517764804e-05} +{"train_loss": 0.014906346797943115, "grad_norm": 0.08269781619310379, "global_step": 37350, "epoch": 352, "lr": 2.0012180169160265e-05} +{"train_loss": 0.01393616758286953, "grad_norm": 0.06783940643072128, "global_step": 37351, "epoch": 352, "lr": 2.000980866720429e-05} +{"train_loss": 0.014579091221094131, "grad_norm": 0.055712781846523285, "global_step": 37352, "epoch": 352, "lr": 2.000743727062084e-05} +{"train_loss": 0.015469968318939209, "grad_norm": 0.09648288786411285, "global_step": 37353, "epoch": 352, "lr": 2.0005065979418176e-05} +{"train_loss": 0.015603653155267239, "grad_norm": 0.09185239672660828, "global_step": 37354, "epoch": 352, "lr": 2.0002694793604665e-05} +{"train_loss": 0.014867641031742096, "grad_norm": 0.09398677945137024, "global_step": 37355, "epoch": 352, "lr": 2.0000323713188655e-05} +{"train_loss": 0.014602363109588623, "grad_norm": 0.06684476882219315, "global_step": 37356, "epoch": 352, "lr": 1.999795273817845e-05} +{"train_loss": 0.014040937647223473, "grad_norm": 0.05275391787290573, "global_step": 37357, "epoch": 352, "lr": 1.9995581868582387e-05} +{"train_loss": 0.01604173518717289, "grad_norm": 0.09134959429502487, "global_step": 37358, "epoch": 352, "lr": 1.9993211104408815e-05} +{"train_loss": 0.012261081486940384, "grad_norm": 0.05922730639576912, "global_step": 37359, "epoch": 352, "lr": 1.9990840445666054e-05} +{"train_loss": 0.015147445723414421, "grad_norm": 0.054661162197589874, "global_step": 37360, "epoch": 352, "lr": 1.998846989236241e-05} +{"train_loss": 0.016016125679016113, "grad_norm": 0.11722621321678162, "global_step": 37361, "epoch": 352, "lr": 1.998609944450623e-05} +{"train_loss": 0.014433639124035835, "grad_norm": 0.08496934175491333, "global_step": 37362, "epoch": 352, "lr": 1.9983729102105864e-05} +{"train_loss": 0.01316136121749878, "grad_norm": 0.06724626570940018, "global_step": 37363, "epoch": 352, "lr": 1.9981358865169603e-05} +{"train_loss": 0.014906416647136211, "grad_norm": 0.09073340147733688, "global_step": 37364, "epoch": 352, "lr": 1.9978988733705807e-05} +{"train_loss": 0.012969211675226688, "grad_norm": 0.08781501650810242, "global_step": 37365, "epoch": 352, "lr": 1.997661870772277e-05} +{"train_loss": 0.015672948211431503, "grad_norm": 0.07636304944753647, "global_step": 37366, "epoch": 352, "lr": 1.997424878722886e-05} +{"train_loss": 0.015122143551707268, "grad_norm": 0.08877032995223999, "global_step": 37367, "epoch": 352, "lr": 1.997187897223236e-05} +{"train_loss": 0.01738005504012108, "grad_norm": 0.0955202728509903, "global_step": 37368, "epoch": 352, "lr": 1.996950926274162e-05} +{"train_loss": 0.015470555052161217, "grad_norm": 0.059643179178237915, "global_step": 37369, "epoch": 352, "lr": 1.996713965876498e-05} +{"train_loss": 0.013713687658309937, "grad_norm": 0.08241420239210129, "global_step": 37370, "epoch": 352, "lr": 1.9964770160310757e-05} +{"train_loss": 0.012250830419361591, "grad_norm": 0.05352317914366722, "global_step": 37371, "epoch": 352, "lr": 1.9962400767387253e-05} +{"train_loss": 0.015316775068640709, "grad_norm": 0.08848495781421661, "global_step": 37372, "epoch": 352, "lr": 1.9960031480002817e-05} +{"train_loss": 0.013716998510062695, "grad_norm": 0.07407569885253906, "global_step": 37373, "epoch": 352, "lr": 1.9957662298165758e-05} +{"train_loss": 0.013247120194137096, "grad_norm": 0.17388048768043518, "global_step": 37374, "epoch": 352, "lr": 1.9955293221884403e-05} +{"train_loss": 0.01666399836540222, "grad_norm": 0.074383445084095, "global_step": 37375, "epoch": 352, "lr": 1.9952924251167098e-05} +{"train_loss": 0.015757732093334198, "grad_norm": 0.07385393977165222, "global_step": 37376, "epoch": 352, "lr": 1.9950555386022156e-05} +{"train_loss": 0.014311600476503372, "grad_norm": 0.06539808213710785, "global_step": 37377, "epoch": 352, "lr": 1.994818662645787e-05} +{"train_loss": 0.01601901836693287, "grad_norm": 0.06503172963857651, "global_step": 37378, "epoch": 352, "lr": 1.9945817972482594e-05} +{"train_loss": 0.013449884951114655, "grad_norm": 0.06883694231510162, "global_step": 37379, "epoch": 352, "lr": 1.994344942410466e-05} +{"train_loss": 0.013814061880111694, "grad_norm": 0.056696850806474686, "global_step": 37380, "epoch": 352, "lr": 1.994108098133235e-05} +{"train_loss": 0.013716966845095158, "grad_norm": 0.06655634194612503, "global_step": 37381, "epoch": 352, "lr": 1.993871264417403e-05} +{"train_loss": 0.013717565685510635, "grad_norm": 0.07475801557302475, "global_step": 37382, "epoch": 352, "lr": 1.9936344412637987e-05} +{"train_loss": 0.017084484919905663, "grad_norm": 0.09949300438165665, "global_step": 37383, "epoch": 352, "lr": 1.9933976286732566e-05} +{"train_loss": 0.014956971630454063, "grad_norm": 0.0774407833814621, "global_step": 37384, "epoch": 352, "lr": 1.993160826646606e-05} +{"train_loss": 0.014887552708387375, "grad_norm": 0.05755825340747833, "global_step": 37385, "epoch": 352, "lr": 1.9929240351846812e-05} +{"train_loss": 0.01680554635822773, "grad_norm": 0.08326631784439087, "global_step": 37386, "epoch": 352, "lr": 1.9926872542883152e-05} +{"train_loss": 0.015899093821644783, "grad_norm": 0.07919158786535263, "global_step": 37387, "epoch": 352, "lr": 1.992450483958338e-05} +{"train_loss": 0.015718407928943634, "grad_norm": 0.10191778093576431, "global_step": 37388, "epoch": 352, "lr": 1.9922137241955803e-05} +{"train_loss": 0.012658771127462387, "grad_norm": 0.06556330621242523, "global_step": 37389, "epoch": 352, "lr": 1.9919769750008756e-05} +{"train_loss": 0.01730552688241005, "grad_norm": 0.08461388200521469, "global_step": 37390, "epoch": 352, "lr": 1.991740236375057e-05} +{"train_loss": 0.012792166322469711, "grad_norm": 0.05528998747467995, "global_step": 37391, "epoch": 352, "lr": 1.9915035083189536e-05} +{"train_loss": 0.01318447757512331, "grad_norm": 0.060176026076078415, "global_step": 37392, "epoch": 352, "lr": 1.9912667908333982e-05} +{"train_loss": 0.015264381654560566, "grad_norm": 0.07024477422237396, "global_step": 37393, "epoch": 352, "lr": 1.991030083919227e-05} +{"train_loss": 0.014246135950088501, "grad_norm": 0.05108342692255974, "global_step": 37394, "epoch": 352, "lr": 1.9907933875772633e-05} +{"train_loss": 0.015812290832400322, "grad_norm": 0.0806535854935646, "global_step": 37395, "epoch": 352, "lr": 1.9905567018083433e-05} +{"train_loss": 0.012801198288798332, "grad_norm": 0.06779631227254868, "global_step": 37396, "epoch": 352, "lr": 1.9903200266132998e-05} +{"train_loss": 0.01504325121641159, "grad_norm": 0.06776812672615051, "global_step": 37397, "epoch": 352, "lr": 1.9900833619929615e-05} +{"train_loss": 0.013983267359435558, "grad_norm": 0.05920371785759926, "global_step": 37398, "epoch": 352, "lr": 1.989846707948161e-05} +{"train_loss": 0.013529108837246895, "grad_norm": 0.0894785150885582, "global_step": 37399, "epoch": 352, "lr": 1.9896100644797317e-05} +{"train_loss": 0.014962390065193176, "grad_norm": 0.13756723701953888, "global_step": 37400, "epoch": 352, "lr": 1.989373431588504e-05} +{"train_loss": 0.014845484867691994, "grad_norm": 0.08764269948005676, "global_step": 37401, "epoch": 352, "lr": 1.9891368092753065e-05} +{"train_loss": 0.01671852543950081, "grad_norm": 0.10950706899166107, "global_step": 37402, "epoch": 352, "lr": 1.9889001975409726e-05} +{"train_loss": 0.01733175478875637, "grad_norm": 0.07749472558498383, "global_step": 37403, "epoch": 352, "lr": 1.9886635963863364e-05} +{"train_loss": 0.014692745171487331, "grad_norm": 0.10964307934045792, "global_step": 37404, "epoch": 352, "lr": 1.9884270058122245e-05} +{"train_loss": 0.014835823327302933, "grad_norm": 0.07488938421010971, "global_step": 37405, "epoch": 352, "lr": 1.9881904258194722e-05} +{"train_loss": 0.014695567078888416, "grad_norm": 0.1359388381242752, "global_step": 37406, "epoch": 352, "lr": 1.9879538564089075e-05} +{"train_loss": 0.017072876915335655, "grad_norm": 0.09790489077568054, "global_step": 37407, "epoch": 352, "lr": 1.9877172975813645e-05} +{"train_loss": 0.01522936299443245, "grad_norm": 0.062625452876091, "global_step": 37408, "epoch": 352, "lr": 1.9874807493376713e-05} +{"train_loss": 0.015047824010252953, "grad_norm": 0.08397931605577469, "global_step": 37409, "epoch": 352, "lr": 1.9872442116786603e-05} +{"train_loss": 0.014930270612239838, "grad_norm": 0.13136045634746552, "global_step": 37410, "epoch": 352, "lr": 1.9870076846051655e-05} +{"train_loss": 0.014248963445425034, "grad_norm": 0.06810209900140762, "global_step": 37411, "epoch": 352, "lr": 1.9867711681180145e-05} +{"train_loss": 0.015092278830707073, "grad_norm": 0.08947299420833588, "global_step": 37412, "epoch": 352, "lr": 1.986534662218038e-05} +{"train_loss": 0.013917009346187115, "grad_norm": 0.06411387026309967, "global_step": 37413, "epoch": 352, "lr": 1.9862981669060697e-05} +{"train_loss": 0.018199700862169266, "grad_norm": 0.09704544395208359, "global_step": 37414, "epoch": 352, "lr": 1.986061682182937e-05} +{"train_loss": 0.012419626116752625, "grad_norm": 0.10756320506334305, "global_step": 37415, "epoch": 352, "lr": 1.985825208049473e-05} +{"train_loss": 0.01584302820265293, "grad_norm": 0.05792660266160965, "global_step": 37416, "epoch": 352, "lr": 1.985588744506511e-05} +{"train_loss": 0.015085577077390451, "grad_norm": 0.09274391084909439, "global_step": 37417, "epoch": 352, "lr": 1.9853522915548777e-05, "val_loss": 0.019746188074350357} +{"train_loss": 0.01471182145178318, "grad_norm": 0.161432683467865, "global_step": 37418, "epoch": 353, "lr": 1.9851158491954048e-05} +{"train_loss": 0.015715843066573143, "grad_norm": 0.07204385101795197, "global_step": 37419, "epoch": 353, "lr": 1.9848794174289232e-05} +{"train_loss": 0.016037078574299812, "grad_norm": 0.08985567837953568, "global_step": 37420, "epoch": 353, "lr": 1.9846429962562657e-05} +{"train_loss": 0.014568952843546867, "grad_norm": 0.07126275449991226, "global_step": 37421, "epoch": 353, "lr": 1.98440658567826e-05} +{"train_loss": 0.014492455869913101, "grad_norm": 0.14215674996376038, "global_step": 37422, "epoch": 353, "lr": 1.9841701856957378e-05} +{"train_loss": 0.012662761844694614, "grad_norm": 0.04968304932117462, "global_step": 37423, "epoch": 353, "lr": 1.9839337963095322e-05} +{"train_loss": 0.014665830880403519, "grad_norm": 0.12699145078659058, "global_step": 37424, "epoch": 353, "lr": 1.9836974175204715e-05} +{"train_loss": 0.012554156593978405, "grad_norm": 0.05065886676311493, "global_step": 37425, "epoch": 353, "lr": 1.9834610493293844e-05} +{"train_loss": 0.014332333579659462, "grad_norm": 0.12540239095687866, "global_step": 37426, "epoch": 353, "lr": 1.9832246917371033e-05} +{"train_loss": 0.016410384327173233, "grad_norm": 0.08959260582923889, "global_step": 37427, "epoch": 353, "lr": 1.9829883447444604e-05} +{"train_loss": 0.013136067427694798, "grad_norm": 0.08329897373914719, "global_step": 37428, "epoch": 353, "lr": 1.9827520083522843e-05} +{"train_loss": 0.017203165218234062, "grad_norm": 0.1379765272140503, "global_step": 37429, "epoch": 353, "lr": 1.982515682561404e-05} +{"train_loss": 0.014813199639320374, "grad_norm": 0.0705953910946846, "global_step": 37430, "epoch": 353, "lr": 1.982279367372651e-05} +{"train_loss": 0.01526407990604639, "grad_norm": 0.05946556478738785, "global_step": 37431, "epoch": 353, "lr": 1.982043062786858e-05} +{"train_loss": 0.014020740985870361, "grad_norm": 0.1475827544927597, "global_step": 37432, "epoch": 353, "lr": 1.9818067688048508e-05} +{"train_loss": 0.015139291994273663, "grad_norm": 0.07431792467832565, "global_step": 37433, "epoch": 353, "lr": 1.981570485427463e-05} +{"train_loss": 0.014352998696267605, "grad_norm": 0.0548529215157032, "global_step": 37434, "epoch": 353, "lr": 1.9813342126555268e-05} +{"train_loss": 0.015597693622112274, "grad_norm": 0.0614924430847168, "global_step": 37435, "epoch": 353, "lr": 1.9810979504898657e-05} +{"train_loss": 0.014063536189496517, "grad_norm": 0.05954619497060776, "global_step": 37436, "epoch": 353, "lr": 1.980861698931314e-05} +{"train_loss": 0.01390707679092884, "grad_norm": 0.06589479744434357, "global_step": 37437, "epoch": 353, "lr": 1.9806254579807027e-05} +{"train_loss": 0.012712799943983555, "grad_norm": 0.08829439431428909, "global_step": 37438, "epoch": 353, "lr": 1.9803892276388585e-05} +{"train_loss": 0.014151185750961304, "grad_norm": 0.08948404341936111, "global_step": 37439, "epoch": 353, "lr": 1.9801530079066137e-05} +{"train_loss": 0.015081455931067467, "grad_norm": 0.10213050991296768, "global_step": 37440, "epoch": 353, "lr": 1.9799167987848e-05} +{"train_loss": 0.015379197895526886, "grad_norm": 0.06326751410961151, "global_step": 37441, "epoch": 353, "lr": 1.9796806002742452e-05} +{"train_loss": 0.018144546076655388, "grad_norm": 0.14506904780864716, "global_step": 37442, "epoch": 353, "lr": 1.979444412375777e-05} +{"train_loss": 0.01636260747909546, "grad_norm": 0.08704521507024765, "global_step": 37443, "epoch": 353, "lr": 1.979208235090228e-05} +{"train_loss": 0.013319836929440498, "grad_norm": 0.08764535188674927, "global_step": 37444, "epoch": 353, "lr": 1.9789720684184286e-05} +{"train_loss": 0.01335943304002285, "grad_norm": 0.052799686789512634, "global_step": 37445, "epoch": 353, "lr": 1.9787359123612066e-05} +{"train_loss": 0.01630658283829689, "grad_norm": 0.07294764369726181, "global_step": 37446, "epoch": 353, "lr": 1.9784997669193938e-05} +{"train_loss": 0.014469144865870476, "grad_norm": 0.0982082411646843, "global_step": 37447, "epoch": 353, "lr": 1.9782636320938176e-05} +{"train_loss": 0.01733444258570671, "grad_norm": 0.12497600167989731, "global_step": 37448, "epoch": 353, "lr": 1.9780275078853105e-05} +{"train_loss": 0.014111272990703583, "grad_norm": 0.09134664386510849, "global_step": 37449, "epoch": 353, "lr": 1.9777913942946985e-05} +{"train_loss": 0.01669136807322502, "grad_norm": 0.11909972876310349, "global_step": 37450, "epoch": 353, "lr": 1.9775552913228135e-05} +{"train_loss": 0.013639720156788826, "grad_norm": 0.05844606086611748, "global_step": 37451, "epoch": 353, "lr": 1.9773191989704864e-05} +{"train_loss": 0.01464616134762764, "grad_norm": 0.04977334663271904, "global_step": 37452, "epoch": 353, "lr": 1.977083117238545e-05} +{"train_loss": 0.015546540729701519, "grad_norm": 0.11543204635381699, "global_step": 37453, "epoch": 353, "lr": 1.976847046127818e-05} +{"train_loss": 0.016509519889950752, "grad_norm": 0.11739741265773773, "global_step": 37454, "epoch": 353, "lr": 1.9766109856391362e-05} +{"train_loss": 0.015008309856057167, "grad_norm": 0.07284417003393173, "global_step": 37455, "epoch": 353, "lr": 1.9763749357733275e-05} +{"train_loss": 0.016389954835176468, "grad_norm": 0.12448038160800934, "global_step": 37456, "epoch": 353, "lr": 1.9761388965312222e-05} +{"train_loss": 0.014266567304730415, "grad_norm": 0.10630805045366287, "global_step": 37457, "epoch": 353, "lr": 1.9759028679136514e-05} +{"train_loss": 0.0140877366065979, "grad_norm": 0.07832137495279312, "global_step": 37458, "epoch": 353, "lr": 1.9756668499214432e-05} +{"train_loss": 0.018352247774600983, "grad_norm": 0.09215313196182251, "global_step": 37459, "epoch": 353, "lr": 1.9754308425554242e-05} +{"train_loss": 0.013670943677425385, "grad_norm": 0.06639361381530762, "global_step": 37460, "epoch": 353, "lr": 1.975194845816426e-05} +{"train_loss": 0.015777824446558952, "grad_norm": 0.08049909770488739, "global_step": 37461, "epoch": 353, "lr": 1.974958859705279e-05} +{"train_loss": 0.017200924456119537, "grad_norm": 0.10263968259096146, "global_step": 37462, "epoch": 353, "lr": 1.9747228842228105e-05} +{"train_loss": 0.016658999025821686, "grad_norm": 0.08622939139604568, "global_step": 37463, "epoch": 353, "lr": 1.974486919369849e-05} +{"train_loss": 0.015827426686882973, "grad_norm": 0.14926451444625854, "global_step": 37464, "epoch": 353, "lr": 1.9742509651472268e-05} +{"train_loss": 0.014420239254832268, "grad_norm": 0.07421472668647766, "global_step": 37465, "epoch": 353, "lr": 1.9740150215557713e-05} +{"train_loss": 0.014564042910933495, "grad_norm": 0.06420520693063736, "global_step": 37466, "epoch": 353, "lr": 1.973779088596309e-05} +{"train_loss": 0.016596965491771698, "grad_norm": 0.09465900808572769, "global_step": 37467, "epoch": 353, "lr": 1.9735431662696703e-05} +{"train_loss": 0.016849732026457787, "grad_norm": 0.10911203920841217, "global_step": 37468, "epoch": 353, "lr": 1.9733072545766875e-05} +{"train_loss": 0.015376444905996323, "grad_norm": 0.07353300601243973, "global_step": 37469, "epoch": 353, "lr": 1.973071353518184e-05} +{"train_loss": 0.014633563347160816, "grad_norm": 0.05883454903960228, "global_step": 37470, "epoch": 353, "lr": 1.9728354630949936e-05} +{"train_loss": 0.013095121830701828, "grad_norm": 0.11782663315534592, "global_step": 37471, "epoch": 353, "lr": 1.972599583307943e-05} +{"train_loss": 0.01598382741212845, "grad_norm": 0.07892777025699615, "global_step": 37472, "epoch": 353, "lr": 1.972363714157859e-05} +{"train_loss": 0.016582438722252846, "grad_norm": 0.08286195993423462, "global_step": 37473, "epoch": 353, "lr": 1.972127855645572e-05} +{"train_loss": 0.016408944502472878, "grad_norm": 0.08095826208591461, "global_step": 37474, "epoch": 353, "lr": 1.9718920077719132e-05} +{"train_loss": 0.014532303437590599, "grad_norm": 0.07665114104747772, "global_step": 37475, "epoch": 353, "lr": 1.9716561705377085e-05} +{"train_loss": 0.015991464257240295, "grad_norm": 0.05750112980604172, "global_step": 37476, "epoch": 353, "lr": 1.9714203439437857e-05} +{"train_loss": 0.014316564425826073, "grad_norm": 0.140636146068573, "global_step": 37477, "epoch": 353, "lr": 1.9711845279909745e-05} +{"train_loss": 0.014187700115144253, "grad_norm": 0.10068823397159576, "global_step": 37478, "epoch": 353, "lr": 1.9709487226801053e-05} +{"train_loss": 0.015495220199227333, "grad_norm": 0.0916258841753006, "global_step": 37479, "epoch": 353, "lr": 1.9707129280120036e-05} +{"train_loss": 0.014884242787957191, "grad_norm": 0.06343218684196472, "global_step": 37480, "epoch": 353, "lr": 1.9704771439874993e-05} +{"train_loss": 0.014040115289390087, "grad_norm": 0.0900270938873291, "global_step": 37481, "epoch": 353, "lr": 1.970241370607422e-05} +{"train_loss": 0.013594631105661392, "grad_norm": 0.07967257499694824, "global_step": 37482, "epoch": 353, "lr": 1.970005607872599e-05} +{"train_loss": 0.017577625811100006, "grad_norm": 0.09124083071947098, "global_step": 37483, "epoch": 353, "lr": 1.9697698557838573e-05} +{"train_loss": 0.01544718537479639, "grad_norm": 0.0983852744102478, "global_step": 37484, "epoch": 353, "lr": 1.969534114342026e-05} +{"train_loss": 0.015146072953939438, "grad_norm": 0.06634896993637085, "global_step": 37485, "epoch": 353, "lr": 1.9692983835479363e-05} +{"train_loss": 0.014892700128257275, "grad_norm": 0.059256412088871, "global_step": 37486, "epoch": 353, "lr": 1.969062663402412e-05} +{"train_loss": 0.01614929735660553, "grad_norm": 0.07054470479488373, "global_step": 37487, "epoch": 353, "lr": 1.9688269539062848e-05} +{"train_loss": 0.013587789610028267, "grad_norm": 0.0618746280670166, "global_step": 37488, "epoch": 353, "lr": 1.9685912550603803e-05} +{"train_loss": 0.016525529325008392, "grad_norm": 0.11933723092079163, "global_step": 37489, "epoch": 353, "lr": 1.968355566865529e-05} +{"train_loss": 0.014184127561748028, "grad_norm": 0.07579226046800613, "global_step": 37490, "epoch": 353, "lr": 1.9681198893225562e-05} +{"train_loss": 0.014660986140370369, "grad_norm": 0.08925312012434006, "global_step": 37491, "epoch": 353, "lr": 1.9678842224322917e-05} +{"train_loss": 0.016142232343554497, "grad_norm": 0.06511420756578445, "global_step": 37492, "epoch": 353, "lr": 1.967648566195565e-05} +{"train_loss": 0.014022848568856716, "grad_norm": 0.09415356069803238, "global_step": 37493, "epoch": 353, "lr": 1.9674129206132032e-05} +{"train_loss": 0.013058163225650787, "grad_norm": 0.10484179109334946, "global_step": 37494, "epoch": 353, "lr": 1.967177285686031e-05} +{"train_loss": 0.014547714963555336, "grad_norm": 0.053484633564949036, "global_step": 37495, "epoch": 353, "lr": 1.9669416614148806e-05} +{"train_loss": 0.013485291972756386, "grad_norm": 0.08111046999692917, "global_step": 37496, "epoch": 353, "lr": 1.9667060478005765e-05} +{"train_loss": 0.013953094370663166, "grad_norm": 0.09022493660449982, "global_step": 37497, "epoch": 353, "lr": 1.9664704448439486e-05} +{"train_loss": 0.01691598817706108, "grad_norm": 0.11445052176713943, "global_step": 37498, "epoch": 353, "lr": 1.9662348525458262e-05} +{"train_loss": 0.012866896577179432, "grad_norm": 0.09058987349271774, "global_step": 37499, "epoch": 353, "lr": 1.9659992709070345e-05} +{"train_loss": 0.013116000220179558, "grad_norm": 0.11364921927452087, "global_step": 37500, "epoch": 353, "lr": 1.9657636999284007e-05} +{"train_loss": 0.01612059213221073, "grad_norm": 0.07874910533428192, "global_step": 37501, "epoch": 353, "lr": 1.9655281396107534e-05} +{"train_loss": 0.015189405530691147, "grad_norm": 0.06727136671543121, "global_step": 37502, "epoch": 353, "lr": 1.9652925899549225e-05} +{"train_loss": 0.01447613537311554, "grad_norm": 0.06581015139818192, "global_step": 37503, "epoch": 353, "lr": 1.9650570509617318e-05} +{"train_loss": 0.01525698509067297, "grad_norm": 0.11910264194011688, "global_step": 37504, "epoch": 353, "lr": 1.964821522632011e-05} +{"train_loss": 0.015801528468728065, "grad_norm": 0.08698344975709915, "global_step": 37505, "epoch": 353, "lr": 1.9645860049665887e-05} +{"train_loss": 0.013848839327692986, "grad_norm": 0.05869721621274948, "global_step": 37506, "epoch": 353, "lr": 1.9643504979662915e-05} +{"train_loss": 0.012864982709288597, "grad_norm": 0.06988685578107834, "global_step": 37507, "epoch": 353, "lr": 1.964115001631945e-05} +{"train_loss": 0.013485275208950043, "grad_norm": 0.06276462227106094, "global_step": 37508, "epoch": 353, "lr": 1.9638795159643775e-05} +{"train_loss": 0.016912782564759254, "grad_norm": 0.08475722372531891, "global_step": 37509, "epoch": 353, "lr": 1.9636440409644185e-05} +{"train_loss": 0.01575327292084694, "grad_norm": 0.06196371465921402, "global_step": 37510, "epoch": 353, "lr": 1.963408576632893e-05} +{"train_loss": 0.016362488269805908, "grad_norm": 0.06693378835916519, "global_step": 37511, "epoch": 353, "lr": 1.96317312297063e-05} +{"train_loss": 0.012723906897008419, "grad_norm": 0.05892050638794899, "global_step": 37512, "epoch": 353, "lr": 1.9629376799784572e-05} +{"train_loss": 0.016650695353746414, "grad_norm": 0.11741000413894653, "global_step": 37513, "epoch": 353, "lr": 1.9627022476571975e-05} +{"train_loss": 0.017559688538312912, "grad_norm": 0.07840393483638763, "global_step": 37514, "epoch": 353, "lr": 1.9624668260076822e-05} +{"train_loss": 0.016581010073423386, "grad_norm": 0.10308003425598145, "global_step": 37515, "epoch": 353, "lr": 1.9622314150307387e-05} +{"train_loss": 0.014757058583199978, "grad_norm": 0.06547331809997559, "global_step": 37516, "epoch": 353, "lr": 1.9619960147271917e-05} +{"train_loss": 0.01546757947653532, "grad_norm": 0.148209348320961, "global_step": 37517, "epoch": 353, "lr": 1.9617606250978708e-05} +{"train_loss": 0.014898634515702724, "grad_norm": 0.09090124815702438, "global_step": 37518, "epoch": 353, "lr": 1.9615252461436e-05} +{"train_loss": 0.01584036275744438, "grad_norm": 0.08633572608232498, "global_step": 37519, "epoch": 353, "lr": 1.96128987786521e-05} +{"train_loss": 0.0177563838660717, "grad_norm": 0.07634458690881729, "global_step": 37520, "epoch": 353, "lr": 1.961054520263524e-05} +{"train_loss": 0.013757266104221344, "grad_norm": 0.05614333227276802, "global_step": 37521, "epoch": 353, "lr": 1.960819173339371e-05} +{"train_loss": 0.01625162549316883, "grad_norm": 0.06851519644260406, "global_step": 37522, "epoch": 353, "lr": 1.960583837093579e-05} +{"train_loss": 0.01509799730946433, "grad_norm": 0.12775762379169464, "global_step": 37523, "epoch": 353, "lr": 1.9603485115269744e-05, "val_loss": 0.02017124369740486} +{"train_loss": 0.015625575557351112, "grad_norm": 0.13698598742485046, "global_step": 37524, "epoch": 354, "lr": 1.9601131966403812e-05} +{"train_loss": 0.013257771730422974, "grad_norm": 0.08127901703119278, "global_step": 37525, "epoch": 354, "lr": 1.959877892434628e-05} +{"train_loss": 0.01484072394669056, "grad_norm": 0.08329745382070541, "global_step": 37526, "epoch": 354, "lr": 1.9596425989105434e-05} +{"train_loss": 0.01608649082481861, "grad_norm": 0.0898766964673996, "global_step": 37527, "epoch": 354, "lr": 1.9594073160689508e-05} +{"train_loss": 0.01534249261021614, "grad_norm": 0.0777978003025055, "global_step": 37528, "epoch": 354, "lr": 1.959172043910681e-05} +{"train_loss": 0.014892456121742725, "grad_norm": 0.08059458434581757, "global_step": 37529, "epoch": 354, "lr": 1.9589367824365572e-05} +{"train_loss": 0.014382358640432358, "grad_norm": 0.06031249091029167, "global_step": 37530, "epoch": 354, "lr": 1.9587015316474055e-05} +{"train_loss": 0.014528250321745872, "grad_norm": 0.06506868451833725, "global_step": 37531, "epoch": 354, "lr": 1.9584662915440543e-05} +{"train_loss": 0.014845683239400387, "grad_norm": 0.07571528851985931, "global_step": 37532, "epoch": 354, "lr": 1.9582310621273298e-05} +{"train_loss": 0.014479849487543106, "grad_norm": 0.06674817949533463, "global_step": 37533, "epoch": 354, "lr": 1.9579958433980604e-05} +{"train_loss": 0.015351779758930206, "grad_norm": 0.07673951238393784, "global_step": 37534, "epoch": 354, "lr": 1.9577606353570705e-05} +{"train_loss": 0.014980403706431389, "grad_norm": 0.09484557062387466, "global_step": 37535, "epoch": 354, "lr": 1.9575254380051848e-05} +{"train_loss": 0.013782046735286713, "grad_norm": 0.09510202705860138, "global_step": 37536, "epoch": 354, "lr": 1.9572902513432334e-05} +{"train_loss": 0.015416608192026615, "grad_norm": 0.06758004426956177, "global_step": 37537, "epoch": 354, "lr": 1.957055075372039e-05} +{"train_loss": 0.013710815459489822, "grad_norm": 0.11297557502985, "global_step": 37538, "epoch": 354, "lr": 1.95681991009243e-05} +{"train_loss": 0.013358183205127716, "grad_norm": 0.08593334257602692, "global_step": 37539, "epoch": 354, "lr": 1.9565847555052342e-05} +{"train_loss": 0.01876159757375717, "grad_norm": 0.05828935280442238, "global_step": 37540, "epoch": 354, "lr": 1.9563496116112756e-05} +{"train_loss": 0.015435799956321716, "grad_norm": 0.06557440012693405, "global_step": 37541, "epoch": 354, "lr": 1.9561144784113793e-05} +{"train_loss": 0.013571326620876789, "grad_norm": 0.06782954931259155, "global_step": 37542, "epoch": 354, "lr": 1.955879355906373e-05} +{"train_loss": 0.01540000643581152, "grad_norm": 0.06391219794750214, "global_step": 37543, "epoch": 354, "lr": 1.9556442440970847e-05} +{"train_loss": 0.015519946813583374, "grad_norm": 0.1107991486787796, "global_step": 37544, "epoch": 354, "lr": 1.9554091429843364e-05} +{"train_loss": 0.015567612834274769, "grad_norm": 0.0783466249704361, "global_step": 37545, "epoch": 354, "lr": 1.9551740525689566e-05} +{"train_loss": 0.01513371616601944, "grad_norm": 0.0871986374258995, "global_step": 37546, "epoch": 354, "lr": 1.9549389728517726e-05} +{"train_loss": 0.014126677066087723, "grad_norm": 0.10010503232479095, "global_step": 37547, "epoch": 354, "lr": 1.954703903833609e-05} +{"train_loss": 0.0184539332985878, "grad_norm": 0.10448632389307022, "global_step": 37548, "epoch": 354, "lr": 1.9544688455152893e-05} +{"train_loss": 0.012176594696938992, "grad_norm": 0.0868951827287674, "global_step": 37549, "epoch": 354, "lr": 1.954233797897642e-05} +{"train_loss": 0.01642596535384655, "grad_norm": 0.1051177904009819, "global_step": 37550, "epoch": 354, "lr": 1.9539987609814947e-05} +{"train_loss": 0.012845203280448914, "grad_norm": 0.07100903987884521, "global_step": 37551, "epoch": 354, "lr": 1.9537637347676686e-05} +{"train_loss": 0.013232543133199215, "grad_norm": 0.05059659481048584, "global_step": 37552, "epoch": 354, "lr": 1.953528719256994e-05} +{"train_loss": 0.01459416188299656, "grad_norm": 0.1330055296421051, "global_step": 37553, "epoch": 354, "lr": 1.953293714450295e-05} +{"train_loss": 0.016132310032844543, "grad_norm": 0.09124084562063217, "global_step": 37554, "epoch": 354, "lr": 1.9530587203483948e-05} +{"train_loss": 0.014617031440138817, "grad_norm": 0.05719553679227829, "global_step": 37555, "epoch": 354, "lr": 1.9528237369521208e-05} +{"train_loss": 0.016544131562113762, "grad_norm": 0.13157586753368378, "global_step": 37556, "epoch": 354, "lr": 1.9525887642623013e-05} +{"train_loss": 0.01690365932881832, "grad_norm": 0.06455480307340622, "global_step": 37557, "epoch": 354, "lr": 1.9523538022797578e-05} +{"train_loss": 0.015702294185757637, "grad_norm": 0.08739796280860901, "global_step": 37558, "epoch": 354, "lr": 1.952118851005319e-05} +{"train_loss": 0.01647970825433731, "grad_norm": 0.07259110361337662, "global_step": 37559, "epoch": 354, "lr": 1.9518839104398075e-05} +{"train_loss": 0.016674278303980827, "grad_norm": 0.05937812104821205, "global_step": 37560, "epoch": 354, "lr": 1.951648980584052e-05} +{"train_loss": 0.017714472487568855, "grad_norm": 0.07818954437971115, "global_step": 37561, "epoch": 354, "lr": 1.9514140614388748e-05} +{"train_loss": 0.016605287790298462, "grad_norm": 0.058624207973480225, "global_step": 37562, "epoch": 354, "lr": 1.9511791530051027e-05} +{"train_loss": 0.013490084558725357, "grad_norm": 0.06894983351230621, "global_step": 37563, "epoch": 354, "lr": 1.950944255283563e-05} +{"train_loss": 0.015174468979239464, "grad_norm": 0.0948045626282692, "global_step": 37564, "epoch": 354, "lr": 1.9507093682750794e-05} +{"train_loss": 0.01522955670952797, "grad_norm": 0.058731067925691605, "global_step": 37565, "epoch": 354, "lr": 1.9504744919804746e-05} +{"train_loss": 0.016334421932697296, "grad_norm": 0.1357983499765396, "global_step": 37566, "epoch": 354, "lr": 1.9502396264005767e-05} +{"train_loss": 0.013439996168017387, "grad_norm": 0.09474137425422668, "global_step": 37567, "epoch": 354, "lr": 1.9500047715362118e-05} +{"train_loss": 0.013767522759735584, "grad_norm": 0.06294999271631241, "global_step": 37568, "epoch": 354, "lr": 1.9497699273882024e-05} +{"train_loss": 0.014586948789656162, "grad_norm": 0.06282956153154373, "global_step": 37569, "epoch": 354, "lr": 1.949535093957375e-05} +{"train_loss": 0.015846792608499527, "grad_norm": 0.06547088921070099, "global_step": 37570, "epoch": 354, "lr": 1.949300271244558e-05} +{"train_loss": 0.0155193991959095, "grad_norm": 0.13936768472194672, "global_step": 37571, "epoch": 354, "lr": 1.9490654592505696e-05} +{"train_loss": 0.014768751338124275, "grad_norm": 0.12467575818300247, "global_step": 37572, "epoch": 354, "lr": 1.9488306579762384e-05} +{"train_loss": 0.013130782172083855, "grad_norm": 0.08323700726032257, "global_step": 37573, "epoch": 354, "lr": 1.948595867422391e-05} +{"train_loss": 0.014728810638189316, "grad_norm": 0.06945782899856567, "global_step": 37574, "epoch": 354, "lr": 1.948361087589849e-05} +{"train_loss": 0.013452058658003807, "grad_norm": 0.06165019050240517, "global_step": 37575, "epoch": 354, "lr": 1.9481263184794408e-05} +{"train_loss": 0.01628796011209488, "grad_norm": 0.06121324375271797, "global_step": 37576, "epoch": 354, "lr": 1.947891560091988e-05} +{"train_loss": 0.014362961053848267, "grad_norm": 0.0673697218298912, "global_step": 37577, "epoch": 354, "lr": 1.9476568124283178e-05} +{"train_loss": 0.01467510499060154, "grad_norm": 0.05661437287926674, "global_step": 37578, "epoch": 354, "lr": 1.9474220754892536e-05} +{"train_loss": 0.01581171341240406, "grad_norm": 0.06790671497583389, "global_step": 37579, "epoch": 354, "lr": 1.9471873492756198e-05} +{"train_loss": 0.015219208784401417, "grad_norm": 0.08863529562950134, "global_step": 37580, "epoch": 354, "lr": 1.9469526337882444e-05} +{"train_loss": 0.015540927648544312, "grad_norm": 0.05466664209961891, "global_step": 37581, "epoch": 354, "lr": 1.946717929027949e-05} +{"train_loss": 0.014445746317505836, "grad_norm": 0.1070948988199234, "global_step": 37582, "epoch": 354, "lr": 1.9464832349955577e-05} +{"train_loss": 0.0164182186126709, "grad_norm": 0.11587798595428467, "global_step": 37583, "epoch": 354, "lr": 1.946248551691896e-05} +{"train_loss": 0.01306117419153452, "grad_norm": 0.06879490613937378, "global_step": 37584, "epoch": 354, "lr": 1.946013879117791e-05} +{"train_loss": 0.016041133552789688, "grad_norm": 0.08498457819223404, "global_step": 37585, "epoch": 354, "lr": 1.945779217274063e-05} +{"train_loss": 0.015778912231326103, "grad_norm": 0.07341241836547852, "global_step": 37586, "epoch": 354, "lr": 1.945544566161539e-05} +{"train_loss": 0.013412958942353725, "grad_norm": 0.07728338986635208, "global_step": 37587, "epoch": 354, "lr": 1.9453099257810448e-05} +{"train_loss": 0.017601408064365387, "grad_norm": 0.12146265804767609, "global_step": 37588, "epoch": 354, "lr": 1.9450752961334025e-05} +{"train_loss": 0.01728927716612816, "grad_norm": 0.08963048458099365, "global_step": 37589, "epoch": 354, "lr": 1.9448406772194356e-05} +{"train_loss": 0.013187352567911148, "grad_norm": 0.07363084703683853, "global_step": 37590, "epoch": 354, "lr": 1.9446060690399703e-05} +{"train_loss": 0.016927625983953476, "grad_norm": 0.1497235745191574, "global_step": 37591, "epoch": 354, "lr": 1.9443714715958315e-05} +{"train_loss": 0.015081524848937988, "grad_norm": 0.11207199096679688, "global_step": 37592, "epoch": 354, "lr": 1.944136884887841e-05} +{"train_loss": 0.01794789358973503, "grad_norm": 0.07684966176748276, "global_step": 37593, "epoch": 354, "lr": 1.9439023089168267e-05} +{"train_loss": 0.015614953823387623, "grad_norm": 0.056534573435783386, "global_step": 37594, "epoch": 354, "lr": 1.9436677436836104e-05} +{"train_loss": 0.01417846791446209, "grad_norm": 0.07129275053739548, "global_step": 37595, "epoch": 354, "lr": 1.9434331891890145e-05} +{"train_loss": 0.012201321311295033, "grad_norm": 0.0686466172337532, "global_step": 37596, "epoch": 354, "lr": 1.943198645433866e-05} +{"train_loss": 0.01697687804698944, "grad_norm": 0.08360739797353745, "global_step": 37597, "epoch": 354, "lr": 1.9429641124189884e-05} +{"train_loss": 0.014013160951435566, "grad_norm": 0.08795273303985596, "global_step": 37598, "epoch": 354, "lr": 1.942729590145205e-05} +{"train_loss": 0.016073016449809074, "grad_norm": 0.07157090306282043, "global_step": 37599, "epoch": 354, "lr": 1.942495078613341e-05} +{"train_loss": 0.013371609151363373, "grad_norm": 0.12071003019809723, "global_step": 37600, "epoch": 354, "lr": 1.9422605778242186e-05} +{"train_loss": 0.016165362671017647, "grad_norm": 0.0687236487865448, "global_step": 37601, "epoch": 354, "lr": 1.9420260877786638e-05} +{"train_loss": 0.012418241240084171, "grad_norm": 0.05570570006966591, "global_step": 37602, "epoch": 354, "lr": 1.9417916084774984e-05} +{"train_loss": 0.015279491432011127, "grad_norm": 0.06638526916503906, "global_step": 37603, "epoch": 354, "lr": 1.941557139921547e-05} +{"train_loss": 0.015195301733911037, "grad_norm": 0.08163486421108246, "global_step": 37604, "epoch": 354, "lr": 1.9413226821116355e-05} +{"train_loss": 0.01356348767876625, "grad_norm": 0.08517774939537048, "global_step": 37605, "epoch": 354, "lr": 1.941088235048586e-05} +{"train_loss": 0.016908885911107063, "grad_norm": 0.09416399151086807, "global_step": 37606, "epoch": 354, "lr": 1.9408537987332205e-05} +{"train_loss": 0.018689628690481186, "grad_norm": 0.22796592116355896, "global_step": 37607, "epoch": 354, "lr": 1.9406193731663643e-05} +{"train_loss": 0.014438746497035027, "grad_norm": 0.15542037785053253, "global_step": 37608, "epoch": 354, "lr": 1.9403849583488427e-05} +{"train_loss": 0.013697878457605839, "grad_norm": 0.05607247352600098, "global_step": 37609, "epoch": 354, "lr": 1.9401505542814756e-05} +{"train_loss": 0.013778896071016788, "grad_norm": 0.08135918527841568, "global_step": 37610, "epoch": 354, "lr": 1.9399161609650895e-05} +{"train_loss": 0.012430265545845032, "grad_norm": 0.05444451421499252, "global_step": 37611, "epoch": 354, "lr": 1.9396817784005107e-05} +{"train_loss": 0.01388264074921608, "grad_norm": 0.0990632101893425, "global_step": 37612, "epoch": 354, "lr": 1.9394474065885547e-05} +{"train_loss": 0.010603515431284904, "grad_norm": 0.06964437663555145, "global_step": 37613, "epoch": 354, "lr": 1.9392130455300506e-05} +{"train_loss": 0.014851389452815056, "grad_norm": 0.0939539223909378, "global_step": 37614, "epoch": 354, "lr": 1.938978695225822e-05} +{"train_loss": 0.015994541347026825, "grad_norm": 0.07472493499517441, "global_step": 37615, "epoch": 354, "lr": 1.93874435567669e-05} +{"train_loss": 0.016188431531190872, "grad_norm": 0.076233871281147, "global_step": 37616, "epoch": 354, "lr": 1.938510026883478e-05} +{"train_loss": 0.013960268348455429, "grad_norm": 0.10256323218345642, "global_step": 37617, "epoch": 354, "lr": 1.9382757088470134e-05} +{"train_loss": 0.016920432448387146, "grad_norm": 0.06733085960149765, "global_step": 37618, "epoch": 354, "lr": 1.9380414015681155e-05} +{"train_loss": 0.014376688748598099, "grad_norm": 0.06186169385910034, "global_step": 37619, "epoch": 354, "lr": 1.9378071050476066e-05} +{"train_loss": 0.014323417097330093, "grad_norm": 0.08540454506874084, "global_step": 37620, "epoch": 354, "lr": 1.9375728192863125e-05} +{"train_loss": 0.014743443578481674, "grad_norm": 0.09177487343549728, "global_step": 37621, "epoch": 354, "lr": 1.9373385442850572e-05} +{"train_loss": 0.014095919206738472, "grad_norm": 0.09378629177808762, "global_step": 37622, "epoch": 354, "lr": 1.9371042800446602e-05} +{"train_loss": 0.015296012163162231, "grad_norm": 0.09732352942228317, "global_step": 37623, "epoch": 354, "lr": 1.936870026565949e-05} +{"train_loss": 0.01638203300535679, "grad_norm": 0.12584078311920166, "global_step": 37624, "epoch": 354, "lr": 1.9366357838497424e-05} +{"train_loss": 0.0161057710647583, "grad_norm": 0.0807725116610527, "global_step": 37625, "epoch": 354, "lr": 1.936401551896867e-05} +{"train_loss": 0.018590983003377914, "grad_norm": 0.07951968163251877, "global_step": 37626, "epoch": 354, "lr": 1.9361673307081428e-05} +{"train_loss": 0.01590060442686081, "grad_norm": 0.088275246322155, "global_step": 37627, "epoch": 354, "lr": 1.9359331202843944e-05} +{"train_loss": 0.014945505186915398, "grad_norm": 0.05814112722873688, "global_step": 37628, "epoch": 354, "lr": 1.9356989206264474e-05} +{"train_loss": 0.015079922174578006, "grad_norm": 0.2009602189064026, "global_step": 37629, "epoch": 354, "lr": 1.9354647317351188e-05, "val_loss": 0.019875479862093925} +{"train_loss": 0.016428962349891663, "grad_norm": 0.11070438474416733, "global_step": 37630, "epoch": 355, "lr": 1.935230553611234e-05} +{"train_loss": 0.015050709247589111, "grad_norm": 0.1489376425743103, "global_step": 37631, "epoch": 355, "lr": 1.934996386255618e-05} +{"train_loss": 0.015986979007720947, "grad_norm": 0.08761592209339142, "global_step": 37632, "epoch": 355, "lr": 1.9347622296690902e-05} +{"train_loss": 0.016717512160539627, "grad_norm": 0.07805616408586502, "global_step": 37633, "epoch": 355, "lr": 1.934528083852475e-05} +{"train_loss": 0.01680745929479599, "grad_norm": 0.14548447728157043, "global_step": 37634, "epoch": 355, "lr": 1.9342939488065965e-05} +{"train_loss": 0.01768188178539276, "grad_norm": 0.1477622091770172, "global_step": 37635, "epoch": 355, "lr": 1.9340598245322755e-05} +{"train_loss": 0.015374742448329926, "grad_norm": 0.11727531254291534, "global_step": 37636, "epoch": 355, "lr": 1.9338257110303337e-05} +{"train_loss": 0.011674088425934315, "grad_norm": 0.05090491846203804, "global_step": 37637, "epoch": 355, "lr": 1.9335916083015947e-05} +{"train_loss": 0.015577493235468864, "grad_norm": 0.1543700397014618, "global_step": 37638, "epoch": 355, "lr": 1.933357516346883e-05} +{"train_loss": 0.014046188443899155, "grad_norm": 0.06200077757239342, "global_step": 37639, "epoch": 355, "lr": 1.933123435167018e-05} +{"train_loss": 0.015238983556628227, "grad_norm": 0.06060119718313217, "global_step": 37640, "epoch": 355, "lr": 1.932889364762825e-05} +{"train_loss": 0.01584138721227646, "grad_norm": 0.09306582808494568, "global_step": 37641, "epoch": 355, "lr": 1.9326553051351227e-05} +{"train_loss": 0.015320919454097748, "grad_norm": 0.11084012687206268, "global_step": 37642, "epoch": 355, "lr": 1.9324212562847376e-05} +{"train_loss": 0.013309139758348465, "grad_norm": 0.0628884881734848, "global_step": 37643, "epoch": 355, "lr": 1.932187218212489e-05} +{"train_loss": 0.01355877798050642, "grad_norm": 0.05735376104712486, "global_step": 37644, "epoch": 355, "lr": 1.9319531909191994e-05} +{"train_loss": 0.013673558831214905, "grad_norm": 0.13243062794208527, "global_step": 37645, "epoch": 355, "lr": 1.9317191744056943e-05} +{"train_loss": 0.01482473500072956, "grad_norm": 0.0558263398706913, "global_step": 37646, "epoch": 355, "lr": 1.9314851686727943e-05} +{"train_loss": 0.015642039477825165, "grad_norm": 0.17423325777053833, "global_step": 37647, "epoch": 355, "lr": 1.9312511737213186e-05} +{"train_loss": 0.014882776886224747, "grad_norm": 0.08490926027297974, "global_step": 37648, "epoch": 355, "lr": 1.931017189552092e-05} +{"train_loss": 0.013393129222095013, "grad_norm": 0.07249964773654938, "global_step": 37649, "epoch": 355, "lr": 1.930783216165937e-05} +{"train_loss": 0.017032407224178314, "grad_norm": 0.06500766426324844, "global_step": 37650, "epoch": 355, "lr": 1.9305492535636742e-05} +{"train_loss": 0.016044624149799347, "grad_norm": 0.11150917410850525, "global_step": 37651, "epoch": 355, "lr": 1.930315301746127e-05} +{"train_loss": 0.013783136382699013, "grad_norm": 0.059386298060417175, "global_step": 37652, "epoch": 355, "lr": 1.9300813607141194e-05} +{"train_loss": 0.015434633009135723, "grad_norm": 0.0818142518401146, "global_step": 37653, "epoch": 355, "lr": 1.929847430468467e-05} +{"train_loss": 0.015174636617302895, "grad_norm": 0.06672127544879913, "global_step": 37654, "epoch": 355, "lr": 1.9296135110099967e-05} +{"train_loss": 0.017231211066246033, "grad_norm": 0.1168467253446579, "global_step": 37655, "epoch": 355, "lr": 1.9293796023395305e-05} +{"train_loss": 0.013743197545409203, "grad_norm": 0.07738979160785675, "global_step": 37656, "epoch": 355, "lr": 1.929145704457887e-05} +{"train_loss": 0.015078870579600334, "grad_norm": 0.09638389199972153, "global_step": 37657, "epoch": 355, "lr": 1.9289118173658906e-05} +{"train_loss": 0.015628354623913765, "grad_norm": 0.07408096641302109, "global_step": 37658, "epoch": 355, "lr": 1.9286779410643635e-05} +{"train_loss": 0.015730025246739388, "grad_norm": 0.09018342196941376, "global_step": 37659, "epoch": 355, "lr": 1.9284440755541267e-05} +{"train_loss": 0.014146365225315094, "grad_norm": 0.09880296885967255, "global_step": 37660, "epoch": 355, "lr": 1.9282102208359997e-05} +{"train_loss": 0.01544950995594263, "grad_norm": 0.07739672809839249, "global_step": 37661, "epoch": 355, "lr": 1.9279763769108068e-05} +{"train_loss": 0.0159294456243515, "grad_norm": 0.08580564707517624, "global_step": 37662, "epoch": 355, "lr": 1.92774254377937e-05} +{"train_loss": 0.016269728541374207, "grad_norm": 0.06422978639602661, "global_step": 37663, "epoch": 355, "lr": 1.9275087214425085e-05} +{"train_loss": 0.01509031467139721, "grad_norm": 0.14574773609638214, "global_step": 37664, "epoch": 355, "lr": 1.9272749099010467e-05} +{"train_loss": 0.018051765859127045, "grad_norm": 0.11065024882555008, "global_step": 37665, "epoch": 355, "lr": 1.927041109155803e-05} +{"train_loss": 0.014325538650155067, "grad_norm": 0.07508539408445358, "global_step": 37666, "epoch": 355, "lr": 1.926807319207602e-05} +{"train_loss": 0.016798030585050583, "grad_norm": 0.06152854859828949, "global_step": 37667, "epoch": 355, "lr": 1.9265735400572614e-05} +{"train_loss": 0.01657874695956707, "grad_norm": 0.0839259922504425, "global_step": 37668, "epoch": 355, "lr": 1.9263397717056054e-05} +{"train_loss": 0.01544138416647911, "grad_norm": 0.17741920053958893, "global_step": 37669, "epoch": 355, "lr": 1.9261060141534565e-05} +{"train_loss": 0.01201619952917099, "grad_norm": 0.06117113307118416, "global_step": 37670, "epoch": 355, "lr": 1.9258722674016343e-05} +{"train_loss": 0.014891942963004112, "grad_norm": 0.08633729070425034, "global_step": 37671, "epoch": 355, "lr": 1.9256385314509577e-05} +{"train_loss": 0.01466383133083582, "grad_norm": 0.060329075902700424, "global_step": 37672, "epoch": 355, "lr": 1.9254048063022528e-05} +{"train_loss": 0.01694556325674057, "grad_norm": 0.065602146089077, "global_step": 37673, "epoch": 355, "lr": 1.925171091956336e-05} +{"train_loss": 0.01502661406993866, "grad_norm": 0.0734027773141861, "global_step": 37674, "epoch": 355, "lr": 1.924937388414031e-05} +{"train_loss": 0.01522180438041687, "grad_norm": 0.08242462575435638, "global_step": 37675, "epoch": 355, "lr": 1.9247036956761606e-05} +{"train_loss": 0.011962315067648888, "grad_norm": 0.07517058402299881, "global_step": 37676, "epoch": 355, "lr": 1.9244700137435438e-05} +{"train_loss": 0.01565694808959961, "grad_norm": 0.07577014714479446, "global_step": 37677, "epoch": 355, "lr": 1.924236342617e-05} +{"train_loss": 0.014576293528079987, "grad_norm": 0.11193422973155975, "global_step": 37678, "epoch": 355, "lr": 1.9240026822973522e-05} +{"train_loss": 0.01609809696674347, "grad_norm": 0.07705950736999512, "global_step": 37679, "epoch": 355, "lr": 1.9237690327854236e-05} +{"train_loss": 0.013100872747600079, "grad_norm": 0.12454316765069962, "global_step": 37680, "epoch": 355, "lr": 1.92353539408203e-05} +{"train_loss": 0.013854611665010452, "grad_norm": 0.057566821575164795, "global_step": 37681, "epoch": 355, "lr": 1.9233017661879975e-05} +{"train_loss": 0.01595461368560791, "grad_norm": 0.08609289675951004, "global_step": 37682, "epoch": 355, "lr": 1.9230681491041425e-05} +{"train_loss": 0.014122283086180687, "grad_norm": 0.09650389850139618, "global_step": 37683, "epoch": 355, "lr": 1.92283454283129e-05} +{"train_loss": 0.014520559459924698, "grad_norm": 0.06479398161172867, "global_step": 37684, "epoch": 355, "lr": 1.9226009473702566e-05} +{"train_loss": 0.014689214527606964, "grad_norm": 0.07076653838157654, "global_step": 37685, "epoch": 355, "lr": 1.9223673627218658e-05} +{"train_loss": 0.014820093289017677, "grad_norm": 0.09596819430589676, "global_step": 37686, "epoch": 355, "lr": 1.9221337888869385e-05} +{"train_loss": 0.014362633228302002, "grad_norm": 0.07278990745544434, "global_step": 37687, "epoch": 355, "lr": 1.921900225866295e-05} +{"train_loss": 0.014158200472593307, "grad_norm": 0.06055372953414917, "global_step": 37688, "epoch": 355, "lr": 1.9216666736607538e-05} +{"train_loss": 0.013582449406385422, "grad_norm": 0.08893996477127075, "global_step": 37689, "epoch": 355, "lr": 1.9214331322711372e-05} +{"train_loss": 0.013570969924330711, "grad_norm": 0.0969514325261116, "global_step": 37690, "epoch": 355, "lr": 1.921199601698267e-05} +{"train_loss": 0.015265286900103092, "grad_norm": 0.10327213257551193, "global_step": 37691, "epoch": 355, "lr": 1.9209660819429615e-05} +{"train_loss": 0.01534859836101532, "grad_norm": 0.16574133932590485, "global_step": 37692, "epoch": 355, "lr": 1.9207325730060414e-05} +{"train_loss": 0.014332752674818039, "grad_norm": 0.06504448503255844, "global_step": 37693, "epoch": 355, "lr": 1.9204990748883315e-05} +{"train_loss": 0.014989621937274933, "grad_norm": 0.1396273523569107, "global_step": 37694, "epoch": 355, "lr": 1.9202655875906456e-05} +{"train_loss": 0.013842318207025528, "grad_norm": 0.09434974938631058, "global_step": 37695, "epoch": 355, "lr": 1.920032111113807e-05} +{"train_loss": 0.017112676054239273, "grad_norm": 0.08396664261817932, "global_step": 37696, "epoch": 355, "lr": 1.9197986454586374e-05} +{"train_loss": 0.019862793385982513, "grad_norm": 0.1290302276611328, "global_step": 37697, "epoch": 355, "lr": 1.9195651906259543e-05} +{"train_loss": 0.016993243247270584, "grad_norm": 0.11406347155570984, "global_step": 37698, "epoch": 355, "lr": 1.91933174661658e-05} +{"train_loss": 0.016414783895015717, "grad_norm": 0.13164377212524414, "global_step": 37699, "epoch": 355, "lr": 1.919098313431335e-05} +{"train_loss": 0.013774416409432888, "grad_norm": 0.09005064517259598, "global_step": 37700, "epoch": 355, "lr": 1.9188648910710393e-05} +{"train_loss": 0.013941077515482903, "grad_norm": 0.07214215397834778, "global_step": 37701, "epoch": 355, "lr": 1.9186314795365108e-05} +{"train_loss": 0.013096889480948448, "grad_norm": 0.07917129248380661, "global_step": 37702, "epoch": 355, "lr": 1.918398078828571e-05} +{"train_loss": 0.016935624182224274, "grad_norm": 0.08928359299898148, "global_step": 37703, "epoch": 355, "lr": 1.9181646889480416e-05} +{"train_loss": 0.014670330099761486, "grad_norm": 0.11101555824279785, "global_step": 37704, "epoch": 355, "lr": 1.9179313098957396e-05} +{"train_loss": 0.015794316306710243, "grad_norm": 0.08125607669353485, "global_step": 37705, "epoch": 355, "lr": 1.9176979416724888e-05} +{"train_loss": 0.013272777199745178, "grad_norm": 0.07158885151147842, "global_step": 37706, "epoch": 355, "lr": 1.917464584279105e-05} +{"train_loss": 0.01467041950672865, "grad_norm": 0.10015599429607391, "global_step": 37707, "epoch": 355, "lr": 1.917231237716412e-05} +{"train_loss": 0.013400266878306866, "grad_norm": 0.11196895688772202, "global_step": 37708, "epoch": 355, "lr": 1.9169979019852257e-05} +{"train_loss": 0.013599585741758347, "grad_norm": 0.08210206776857376, "global_step": 37709, "epoch": 355, "lr": 1.9167645770863678e-05} +{"train_loss": 0.01369468867778778, "grad_norm": 0.07912342995405197, "global_step": 37710, "epoch": 355, "lr": 1.9165312630206604e-05} +{"train_loss": 0.016122667118906975, "grad_norm": 0.07474744319915771, "global_step": 37711, "epoch": 355, "lr": 1.9162979597889212e-05} +{"train_loss": 0.017209740355610847, "grad_norm": 0.07547362893819809, "global_step": 37712, "epoch": 355, "lr": 1.9160646673919684e-05} +{"train_loss": 0.013257106766104698, "grad_norm": 0.09739810973405838, "global_step": 37713, "epoch": 355, "lr": 1.915831385830625e-05} +{"train_loss": 0.015797723084688187, "grad_norm": 0.06128042936325073, "global_step": 37714, "epoch": 355, "lr": 1.9155981151057066e-05} +{"train_loss": 0.01680867373943329, "grad_norm": 0.10382957011461258, "global_step": 37715, "epoch": 355, "lr": 1.9153648552180352e-05} +{"train_loss": 0.015828639268875122, "grad_norm": 0.11682381480932236, "global_step": 37716, "epoch": 355, "lr": 1.9151316061684322e-05} +{"train_loss": 0.015684647485613823, "grad_norm": 0.15334385633468628, "global_step": 37717, "epoch": 355, "lr": 1.914898367957715e-05} +{"train_loss": 0.015568289905786514, "grad_norm": 0.08936077356338501, "global_step": 37718, "epoch": 355, "lr": 1.914665140586701e-05} +{"train_loss": 0.015192201361060143, "grad_norm": 0.07456526905298233, "global_step": 37719, "epoch": 355, "lr": 1.9144319240562125e-05} +{"train_loss": 0.015199370682239532, "grad_norm": 0.0633431226015091, "global_step": 37720, "epoch": 355, "lr": 1.91419871836707e-05} +{"train_loss": 0.017052922397851944, "grad_norm": 0.08407783508300781, "global_step": 37721, "epoch": 355, "lr": 1.9139655235200893e-05} +{"train_loss": 0.01431380957365036, "grad_norm": 0.09818717837333679, "global_step": 37722, "epoch": 355, "lr": 1.913732339516092e-05} +{"train_loss": 0.014967583119869232, "grad_norm": 0.06355536729097366, "global_step": 37723, "epoch": 355, "lr": 1.9134991663558988e-05} +{"train_loss": 0.01506391167640686, "grad_norm": 0.12288358807563782, "global_step": 37724, "epoch": 355, "lr": 1.9132660040403273e-05} +{"train_loss": 0.014893073588609695, "grad_norm": 0.05060582980513573, "global_step": 37725, "epoch": 355, "lr": 1.9130328525701944e-05} +{"train_loss": 0.016161030158400536, "grad_norm": 0.06750591099262238, "global_step": 37726, "epoch": 355, "lr": 1.912799711946322e-05} +{"train_loss": 0.015575231984257698, "grad_norm": 0.10660819709300995, "global_step": 37727, "epoch": 355, "lr": 1.9125665821695305e-05} +{"train_loss": 0.014150510542094707, "grad_norm": 0.10468350350856781, "global_step": 37728, "epoch": 355, "lr": 1.912333463240637e-05} +{"train_loss": 0.01538578700274229, "grad_norm": 0.07948131114244461, "global_step": 37729, "epoch": 355, "lr": 1.91210035516046e-05} +{"train_loss": 0.01378419529646635, "grad_norm": 0.09575601667165756, "global_step": 37730, "epoch": 355, "lr": 1.9118672579298203e-05} +{"train_loss": 0.015700437128543854, "grad_norm": 0.07121755182743073, "global_step": 37731, "epoch": 355, "lr": 1.9116341715495345e-05} +{"train_loss": 0.012675285339355469, "grad_norm": 0.056699659675359726, "global_step": 37732, "epoch": 355, "lr": 1.9114010960204238e-05} +{"train_loss": 0.01596006751060486, "grad_norm": 0.10479982197284698, "global_step": 37733, "epoch": 355, "lr": 1.9111680313433074e-05} +{"train_loss": 0.013590040616691113, "grad_norm": 0.059149257838726044, "global_step": 37734, "epoch": 355, "lr": 1.910934977519004e-05} +{"train_loss": 0.015075508821404205, "grad_norm": 0.11979833990335464, "global_step": 37735, "epoch": 355, "lr": 1.910701934548329e-05, "val_loss": 0.01970927231013775} +{"train_loss": 0.01313879992812872, "grad_norm": 0.1253783255815506, "global_step": 37736, "epoch": 356, "lr": 1.9104689024321047e-05} +{"train_loss": 0.016066670417785645, "grad_norm": 0.06516855955123901, "global_step": 37737, "epoch": 356, "lr": 1.9102358811711508e-05} +{"train_loss": 0.014615562744438648, "grad_norm": 0.11416416615247726, "global_step": 37738, "epoch": 356, "lr": 1.910002870766282e-05} +{"train_loss": 0.015547988936305046, "grad_norm": 0.10979951173067093, "global_step": 37739, "epoch": 356, "lr": 1.9097698712183194e-05} +{"train_loss": 0.014598645269870758, "grad_norm": 0.1317550390958786, "global_step": 37740, "epoch": 356, "lr": 1.9095368825280836e-05} +{"train_loss": 0.014018520712852478, "grad_norm": 0.12418002635240555, "global_step": 37741, "epoch": 356, "lr": 1.9093039046963913e-05} +{"train_loss": 0.015407606959342957, "grad_norm": 0.06792300194501877, "global_step": 37742, "epoch": 356, "lr": 1.909070937724059e-05} +{"train_loss": 0.01655671000480652, "grad_norm": 0.13814596831798553, "global_step": 37743, "epoch": 356, "lr": 1.908837981611907e-05} +{"train_loss": 0.015348426066339016, "grad_norm": 0.053819604218006134, "global_step": 37744, "epoch": 356, "lr": 1.908605036360756e-05} +{"train_loss": 0.015417873859405518, "grad_norm": 0.1147552877664566, "global_step": 37745, "epoch": 356, "lr": 1.9083721019714206e-05} +{"train_loss": 0.017765840515494347, "grad_norm": 0.11621133238077164, "global_step": 37746, "epoch": 356, "lr": 1.9081391784447233e-05} +{"train_loss": 0.013541070744395256, "grad_norm": 0.06707539409399033, "global_step": 37747, "epoch": 356, "lr": 1.9079062657814783e-05} +{"train_loss": 0.01777525804936886, "grad_norm": 0.056451987475156784, "global_step": 37748, "epoch": 356, "lr": 1.907673363982508e-05} +{"train_loss": 0.016943305730819702, "grad_norm": 0.07755640894174576, "global_step": 37749, "epoch": 356, "lr": 1.907440473048626e-05} +{"train_loss": 0.01454782672226429, "grad_norm": 0.13529632985591888, "global_step": 37750, "epoch": 356, "lr": 1.9072075929806543e-05} +{"train_loss": 0.014889556914567947, "grad_norm": 0.07024993747472763, "global_step": 37751, "epoch": 356, "lr": 1.9069747237794115e-05} +{"train_loss": 0.01362498290836811, "grad_norm": 0.11637349426746368, "global_step": 37752, "epoch": 356, "lr": 1.9067418654457143e-05} +{"train_loss": 0.01341073028743267, "grad_norm": 0.05722205713391304, "global_step": 37753, "epoch": 356, "lr": 1.906509017980379e-05} +{"train_loss": 0.01647236943244934, "grad_norm": 0.10004697740077972, "global_step": 37754, "epoch": 356, "lr": 1.906276181384228e-05} +{"train_loss": 0.016526540741324425, "grad_norm": 0.05565602704882622, "global_step": 37755, "epoch": 356, "lr": 1.9060433556580747e-05} +{"train_loss": 0.013518624007701874, "grad_norm": 0.04911413788795471, "global_step": 37756, "epoch": 356, "lr": 1.90581054080274e-05} +{"train_loss": 0.020656203851103783, "grad_norm": 0.1129019558429718, "global_step": 37757, "epoch": 356, "lr": 1.905577736819043e-05} +{"train_loss": 0.013988036662340164, "grad_norm": 0.07200219482183456, "global_step": 37758, "epoch": 356, "lr": 1.9053449437078003e-05} +{"train_loss": 0.016416305676102638, "grad_norm": 0.09097710996866226, "global_step": 37759, "epoch": 356, "lr": 1.9051121614698276e-05} +{"train_loss": 0.014681600034236908, "grad_norm": 0.08080488443374634, "global_step": 37760, "epoch": 356, "lr": 1.9048793901059454e-05} +{"train_loss": 0.014030300080776215, "grad_norm": 0.07661524415016174, "global_step": 37761, "epoch": 356, "lr": 1.9046466296169717e-05} +{"train_loss": 0.014119019731879234, "grad_norm": 0.053509216755628586, "global_step": 37762, "epoch": 356, "lr": 1.9044138800037225e-05} +{"train_loss": 0.017137203365564346, "grad_norm": 0.08344372361898422, "global_step": 37763, "epoch": 356, "lr": 1.9041811412670167e-05} +{"train_loss": 0.014634642750024796, "grad_norm": 0.07499800622463226, "global_step": 37764, "epoch": 356, "lr": 1.903948413407674e-05} +{"train_loss": 0.015775661915540695, "grad_norm": 0.07247269153594971, "global_step": 37765, "epoch": 356, "lr": 1.9037156964265097e-05} +{"train_loss": 0.013817698694765568, "grad_norm": 0.07573940604925156, "global_step": 37766, "epoch": 356, "lr": 1.90348299032434e-05} +{"train_loss": 0.014221781864762306, "grad_norm": 0.07149063050746918, "global_step": 37767, "epoch": 356, "lr": 1.9032502951019848e-05} +{"train_loss": 0.01670248433947563, "grad_norm": 0.06301919370889664, "global_step": 37768, "epoch": 356, "lr": 1.903017610760263e-05} +{"train_loss": 0.016405927017331123, "grad_norm": 0.12269728630781174, "global_step": 37769, "epoch": 356, "lr": 1.902784937299989e-05} +{"train_loss": 0.015600571408867836, "grad_norm": 0.06941749900579453, "global_step": 37770, "epoch": 356, "lr": 1.902552274721983e-05} +{"train_loss": 0.015747059136629105, "grad_norm": 0.10970786213874817, "global_step": 37771, "epoch": 356, "lr": 1.9023196230270616e-05} +{"train_loss": 0.014997825026512146, "grad_norm": 0.14928197860717773, "global_step": 37772, "epoch": 356, "lr": 1.9020869822160402e-05} +{"train_loss": 0.01395324058830738, "grad_norm": 0.06929833441972733, "global_step": 37773, "epoch": 356, "lr": 1.901854352289738e-05} +{"train_loss": 0.012371379882097244, "grad_norm": 0.10072311013936996, "global_step": 37774, "epoch": 356, "lr": 1.9016217332489738e-05} +{"train_loss": 0.01760776713490486, "grad_norm": 0.07040104269981384, "global_step": 37775, "epoch": 356, "lr": 1.9013891250945616e-05} +{"train_loss": 0.014261264353990555, "grad_norm": 0.08742992579936981, "global_step": 37776, "epoch": 356, "lr": 1.9011565278273224e-05} +{"train_loss": 0.017259716987609863, "grad_norm": 0.1036599799990654, "global_step": 37777, "epoch": 356, "lr": 1.9009239414480696e-05} +{"train_loss": 0.014171963557600975, "grad_norm": 0.06075939163565636, "global_step": 37778, "epoch": 356, "lr": 1.9006913659576236e-05} +{"train_loss": 0.014888331294059753, "grad_norm": 0.061365049332380295, "global_step": 37779, "epoch": 356, "lr": 1.9004588013567992e-05} +{"train_loss": 0.014567838981747627, "grad_norm": 0.1324794888496399, "global_step": 37780, "epoch": 356, "lr": 1.9002262476464143e-05} +{"train_loss": 0.015163956210017204, "grad_norm": 0.07414454966783524, "global_step": 37781, "epoch": 356, "lr": 1.8999937048272887e-05} +{"train_loss": 0.016005627810955048, "grad_norm": 0.10988906770944595, "global_step": 37782, "epoch": 356, "lr": 1.899761172900236e-05} +{"train_loss": 0.014971631579101086, "grad_norm": 0.09638907015323639, "global_step": 37783, "epoch": 356, "lr": 1.8995286518660733e-05} +{"train_loss": 0.01698577031493187, "grad_norm": 0.12510085105895996, "global_step": 37784, "epoch": 356, "lr": 1.8992961417256184e-05} +{"train_loss": 0.01434435322880745, "grad_norm": 0.07731108367443085, "global_step": 37785, "epoch": 356, "lr": 1.89906364247969e-05} +{"train_loss": 0.014308512210845947, "grad_norm": 0.07975506782531738, "global_step": 37786, "epoch": 356, "lr": 1.898831154129102e-05} +{"train_loss": 0.016764633357524872, "grad_norm": 0.10525533556938171, "global_step": 37787, "epoch": 356, "lr": 1.898598676674674e-05} +{"train_loss": 0.014508664608001709, "grad_norm": 0.07658039033412933, "global_step": 37788, "epoch": 356, "lr": 1.8983662101172216e-05} +{"train_loss": 0.01581120863556862, "grad_norm": 0.09375101327896118, "global_step": 37789, "epoch": 356, "lr": 1.8981337544575593e-05} +{"train_loss": 0.017147742211818695, "grad_norm": 0.1384313553571701, "global_step": 37790, "epoch": 356, "lr": 1.897901309696507e-05} +{"train_loss": 0.014966726303100586, "grad_norm": 0.06304486095905304, "global_step": 37791, "epoch": 356, "lr": 1.8976688758348816e-05} +{"train_loss": 0.013246671296656132, "grad_norm": 0.10567040741443634, "global_step": 37792, "epoch": 356, "lr": 1.8974364528734968e-05} +{"train_loss": 0.01146636251360178, "grad_norm": 0.09013371914625168, "global_step": 37793, "epoch": 356, "lr": 1.897204040813173e-05} +{"train_loss": 0.014250271953642368, "grad_norm": 0.06099741533398628, "global_step": 37794, "epoch": 356, "lr": 1.8969716396547225e-05} +{"train_loss": 0.015687257051467896, "grad_norm": 0.08030744642019272, "global_step": 37795, "epoch": 356, "lr": 1.8967392493989668e-05} +{"train_loss": 0.01705813966691494, "grad_norm": 0.05554972589015961, "global_step": 37796, "epoch": 356, "lr": 1.8965068700467175e-05} +{"train_loss": 0.015429215505719185, "grad_norm": 0.05955519154667854, "global_step": 37797, "epoch": 356, "lr": 1.8962745015987937e-05} +{"train_loss": 0.014768797904253006, "grad_norm": 0.08644174039363861, "global_step": 37798, "epoch": 356, "lr": 1.8960421440560135e-05} +{"train_loss": 0.012478143908083439, "grad_norm": 0.06347651034593582, "global_step": 37799, "epoch": 356, "lr": 1.8958097974191907e-05} +{"train_loss": 0.015820471569895744, "grad_norm": 0.08077402412891388, "global_step": 37800, "epoch": 356, "lr": 1.895577461689141e-05} +{"train_loss": 0.01377610769122839, "grad_norm": 0.055818844586610794, "global_step": 37801, "epoch": 356, "lr": 1.8953451368666814e-05} +{"train_loss": 0.01439625769853592, "grad_norm": 0.06495726853609085, "global_step": 37802, "epoch": 356, "lr": 1.895112822952631e-05} +{"train_loss": 0.013231138698756695, "grad_norm": 0.05854326859116554, "global_step": 37803, "epoch": 356, "lr": 1.8948805199478022e-05} +{"train_loss": 0.017191443592309952, "grad_norm": 0.08573491871356964, "global_step": 37804, "epoch": 356, "lr": 1.8946482278530132e-05} +{"train_loss": 0.014051154255867004, "grad_norm": 0.08179166167974472, "global_step": 37805, "epoch": 356, "lr": 1.894415946669081e-05} +{"train_loss": 0.015598745085299015, "grad_norm": 0.10686919838190079, "global_step": 37806, "epoch": 356, "lr": 1.8941836763968208e-05} +{"train_loss": 0.014815220609307289, "grad_norm": 0.09472472220659256, "global_step": 37807, "epoch": 356, "lr": 1.8939514170370464e-05} +{"train_loss": 0.014763322658836842, "grad_norm": 0.08710820227861404, "global_step": 37808, "epoch": 356, "lr": 1.8937191685905764e-05} +{"train_loss": 0.014938660897314548, "grad_norm": 0.12411592900753021, "global_step": 37809, "epoch": 356, "lr": 1.8934869310582275e-05} +{"train_loss": 0.015978828072547913, "grad_norm": 0.09768465161323547, "global_step": 37810, "epoch": 356, "lr": 1.8932547044408132e-05} +{"train_loss": 0.012790817767381668, "grad_norm": 0.07967942953109741, "global_step": 37811, "epoch": 356, "lr": 1.893022488739153e-05} +{"train_loss": 0.015137316659092903, "grad_norm": 0.08760206401348114, "global_step": 37812, "epoch": 356, "lr": 1.89279028395406e-05} +{"train_loss": 0.01354106143116951, "grad_norm": 0.0803997665643692, "global_step": 37813, "epoch": 356, "lr": 1.8925580900863486e-05} +{"train_loss": 0.014267221093177795, "grad_norm": 0.0994139313697815, "global_step": 37814, "epoch": 356, "lr": 1.8923259071368377e-05} +{"train_loss": 0.014376012608408928, "grad_norm": 0.10398635268211365, "global_step": 37815, "epoch": 356, "lr": 1.892093735106343e-05} +{"train_loss": 0.01764223724603653, "grad_norm": 0.12428686767816544, "global_step": 37816, "epoch": 356, "lr": 1.891861573995678e-05} +{"train_loss": 0.012911202386021614, "grad_norm": 0.10731273889541626, "global_step": 37817, "epoch": 356, "lr": 1.8916294238056608e-05} +{"train_loss": 0.014510143548250198, "grad_norm": 0.1249375119805336, "global_step": 37818, "epoch": 356, "lr": 1.8913972845371047e-05} +{"train_loss": 0.014486031606793404, "grad_norm": 0.06104641780257225, "global_step": 37819, "epoch": 356, "lr": 1.8911651561908287e-05} +{"train_loss": 0.016325585544109344, "grad_norm": 0.08103135228157043, "global_step": 37820, "epoch": 356, "lr": 1.8909330387676444e-05} +{"train_loss": 0.015391666442155838, "grad_norm": 0.08105356991291046, "global_step": 37821, "epoch": 356, "lr": 1.8907009322683695e-05} +{"train_loss": 0.01541786640882492, "grad_norm": 0.11219421029090881, "global_step": 37822, "epoch": 356, "lr": 1.8904688366938206e-05} +{"train_loss": 0.015035135671496391, "grad_norm": 0.07253650575876236, "global_step": 37823, "epoch": 356, "lr": 1.8902367520448122e-05} +{"train_loss": 0.014114577323198318, "grad_norm": 0.1296892613172531, "global_step": 37824, "epoch": 356, "lr": 1.8900046783221576e-05} +{"train_loss": 0.015771925449371338, "grad_norm": 0.16682200133800507, "global_step": 37825, "epoch": 356, "lr": 1.8897726155266738e-05} +{"train_loss": 0.01628296449780464, "grad_norm": 0.08168625086545944, "global_step": 37826, "epoch": 356, "lr": 1.8895405636591785e-05} +{"train_loss": 0.01404091902077198, "grad_norm": 0.05986891686916351, "global_step": 37827, "epoch": 356, "lr": 1.889308522720483e-05} +{"train_loss": 0.012708675116300583, "grad_norm": 0.07182034105062485, "global_step": 37828, "epoch": 356, "lr": 1.8890764927114068e-05} +{"train_loss": 0.015367455780506134, "grad_norm": 0.08316671848297119, "global_step": 37829, "epoch": 356, "lr": 1.8888444736327626e-05} +{"train_loss": 0.017338130623102188, "grad_norm": 0.07127729058265686, "global_step": 37830, "epoch": 356, "lr": 1.8886124654853637e-05} +{"train_loss": 0.01635625958442688, "grad_norm": 0.07330621778964996, "global_step": 37831, "epoch": 356, "lr": 1.888380468270028e-05} +{"train_loss": 0.0154105881229043, "grad_norm": 0.07705020159482956, "global_step": 37832, "epoch": 356, "lr": 1.888148481987572e-05} +{"train_loss": 0.01583726331591606, "grad_norm": 0.046535562723875046, "global_step": 37833, "epoch": 356, "lr": 1.8879165066388072e-05} +{"train_loss": 0.014254193753004074, "grad_norm": 0.08376488089561462, "global_step": 37834, "epoch": 356, "lr": 1.8876845422245516e-05} +{"train_loss": 0.01643979363143444, "grad_norm": 0.1726989597082138, "global_step": 37835, "epoch": 356, "lr": 1.8874525887456174e-05} +{"train_loss": 0.01576545462012291, "grad_norm": 0.13097244501113892, "global_step": 37836, "epoch": 356, "lr": 1.887220646202823e-05} +{"train_loss": 0.013822558335959911, "grad_norm": 0.08779789507389069, "global_step": 37837, "epoch": 356, "lr": 1.88698871459698e-05} +{"train_loss": 0.015895379707217216, "grad_norm": 0.1365204006433487, "global_step": 37838, "epoch": 356, "lr": 1.886756793928905e-05} +{"train_loss": 0.013052277266979218, "grad_norm": 0.06576188653707504, "global_step": 37839, "epoch": 356, "lr": 1.8865248841994143e-05} +{"train_loss": 0.012023303657770157, "grad_norm": 0.11029346287250519, "global_step": 37840, "epoch": 356, "lr": 1.886292985409321e-05} +{"train_loss": 0.015090249777543094, "grad_norm": 0.09127572178840637, "global_step": 37841, "epoch": 356, "lr": 1.8860610975594382e-05, "val_loss": 0.01905812695622444} +{"train_loss": 0.014192098751664162, "grad_norm": 0.06090182065963745, "global_step": 37842, "epoch": 357, "lr": 1.885829220650583e-05} +{"train_loss": 0.01414426602423191, "grad_norm": 0.1164361909031868, "global_step": 37843, "epoch": 357, "lr": 1.8855973546835702e-05} +{"train_loss": 0.017664272338151932, "grad_norm": 0.08750584721565247, "global_step": 37844, "epoch": 357, "lr": 1.8853654996592135e-05} +{"train_loss": 0.01620209962129593, "grad_norm": 0.10856841504573822, "global_step": 37845, "epoch": 357, "lr": 1.8851336555783273e-05} +{"train_loss": 0.013483243063092232, "grad_norm": 0.06560049951076508, "global_step": 37846, "epoch": 357, "lr": 1.8849018224417305e-05} +{"train_loss": 0.014316578395664692, "grad_norm": 0.09859748929738998, "global_step": 37847, "epoch": 357, "lr": 1.8846700002502298e-05} +{"train_loss": 0.016416169703006744, "grad_norm": 0.13654500246047974, "global_step": 37848, "epoch": 357, "lr": 1.8844381890046442e-05} +{"train_loss": 0.016507551074028015, "grad_norm": 0.07627488672733307, "global_step": 37849, "epoch": 357, "lr": 1.8842063887057882e-05} +{"train_loss": 0.017023630440235138, "grad_norm": 0.08945213258266449, "global_step": 37850, "epoch": 357, "lr": 1.8839745993544777e-05} +{"train_loss": 0.013659149408340454, "grad_norm": 0.07956689596176147, "global_step": 37851, "epoch": 357, "lr": 1.8837428209515233e-05} +{"train_loss": 0.015025829896330833, "grad_norm": 0.12590664625167847, "global_step": 37852, "epoch": 357, "lr": 1.8835110534977433e-05} +{"train_loss": 0.0137861967086792, "grad_norm": 0.13023534417152405, "global_step": 37853, "epoch": 357, "lr": 1.88327929699395e-05} +{"train_loss": 0.017003536224365234, "grad_norm": 0.11724129319190979, "global_step": 37854, "epoch": 357, "lr": 1.8830475514409564e-05} +{"train_loss": 0.014711055904626846, "grad_norm": 0.08062214404344559, "global_step": 37855, "epoch": 357, "lr": 1.8828158168395783e-05} +{"train_loss": 0.015869788825511932, "grad_norm": 0.10023655742406845, "global_step": 37856, "epoch": 357, "lr": 1.8825840931906307e-05} +{"train_loss": 0.014781715348362923, "grad_norm": 0.14184589684009552, "global_step": 37857, "epoch": 357, "lr": 1.882352380494926e-05} +{"train_loss": 0.01431233435869217, "grad_norm": 0.07517067342996597, "global_step": 37858, "epoch": 357, "lr": 1.88212067875328e-05} +{"train_loss": 0.015903931111097336, "grad_norm": 0.06085723266005516, "global_step": 37859, "epoch": 357, "lr": 1.881888987966505e-05} +{"train_loss": 0.015125369653105736, "grad_norm": 0.052478864789009094, "global_step": 37860, "epoch": 357, "lr": 1.881657308135418e-05} +{"train_loss": 0.013993564993143082, "grad_norm": 0.10434552282094955, "global_step": 37861, "epoch": 357, "lr": 1.8814256392608285e-05} +{"train_loss": 0.01479823887348175, "grad_norm": 0.06916137039661407, "global_step": 37862, "epoch": 357, "lr": 1.8811939813435543e-05} +{"train_loss": 0.014091435819864273, "grad_norm": 0.06344982981681824, "global_step": 37863, "epoch": 357, "lr": 1.880962334384409e-05} +{"train_loss": 0.014895555563271046, "grad_norm": 0.062219101935625076, "global_step": 37864, "epoch": 357, "lr": 1.8807306983842055e-05} +{"train_loss": 0.014457553625106812, "grad_norm": 0.1390879899263382, "global_step": 37865, "epoch": 357, "lr": 1.8804990733437565e-05} +{"train_loss": 0.014892861247062683, "grad_norm": 0.07585859298706055, "global_step": 37866, "epoch": 357, "lr": 1.880267459263877e-05} +{"train_loss": 0.014410685747861862, "grad_norm": 0.13118521869182587, "global_step": 37867, "epoch": 357, "lr": 1.8800358561453823e-05} +{"train_loss": 0.014834452420473099, "grad_norm": 0.20261864364147186, "global_step": 37868, "epoch": 357, "lr": 1.8798042639890833e-05} +{"train_loss": 0.016249777749180794, "grad_norm": 0.053710244596004486, "global_step": 37869, "epoch": 357, "lr": 1.8795726827957953e-05} +{"train_loss": 0.015051887370646, "grad_norm": 0.07594185322523117, "global_step": 37870, "epoch": 357, "lr": 1.879341112566335e-05} +{"train_loss": 0.01523633860051632, "grad_norm": 0.06509428471326828, "global_step": 37871, "epoch": 357, "lr": 1.8791095533015096e-05} +{"train_loss": 0.015298418700695038, "grad_norm": 0.06492777168750763, "global_step": 37872, "epoch": 357, "lr": 1.8788780050021365e-05} +{"train_loss": 0.01499447412788868, "grad_norm": 0.05851719528436661, "global_step": 37873, "epoch": 357, "lr": 1.8786464676690297e-05} +{"train_loss": 0.013216696679592133, "grad_norm": 0.067630834877491, "global_step": 37874, "epoch": 357, "lr": 1.8784149413030005e-05} +{"train_loss": 0.015592003241181374, "grad_norm": 0.12894123792648315, "global_step": 37875, "epoch": 357, "lr": 1.8781834259048635e-05} +{"train_loss": 0.014503316953778267, "grad_norm": 0.09377899765968323, "global_step": 37876, "epoch": 357, "lr": 1.877951921475434e-05} +{"train_loss": 0.015952374786138535, "grad_norm": 0.21493862569332123, "global_step": 37877, "epoch": 357, "lr": 1.8777204280155237e-05} +{"train_loss": 0.014717349782586098, "grad_norm": 0.13480298221111298, "global_step": 37878, "epoch": 357, "lr": 1.8774889455259442e-05} +{"train_loss": 0.015667814761400223, "grad_norm": 0.09157531708478928, "global_step": 37879, "epoch": 357, "lr": 1.877257474007511e-05} +{"train_loss": 0.018718089908361435, "grad_norm": 0.12462915480136871, "global_step": 37880, "epoch": 357, "lr": 1.8770260134610384e-05} +{"train_loss": 0.013858184218406677, "grad_norm": 0.093080535531044, "global_step": 37881, "epoch": 357, "lr": 1.876794563887338e-05} +{"train_loss": 0.013165969401597977, "grad_norm": 0.06882353127002716, "global_step": 37882, "epoch": 357, "lr": 1.8765631252872217e-05} +{"train_loss": 0.014687923714518547, "grad_norm": 0.12634679675102234, "global_step": 37883, "epoch": 357, "lr": 1.8763316976615043e-05} +{"train_loss": 0.013678786344826221, "grad_norm": 0.09247948974370956, "global_step": 37884, "epoch": 357, "lr": 1.8761002810110006e-05} +{"train_loss": 0.013320719823241234, "grad_norm": 0.061726778745651245, "global_step": 37885, "epoch": 357, "lr": 1.8758688753365206e-05} +{"train_loss": 0.014688855037093163, "grad_norm": 0.07584912329912186, "global_step": 37886, "epoch": 357, "lr": 1.8756374806388784e-05} +{"train_loss": 0.016344502568244934, "grad_norm": 0.144046351313591, "global_step": 37887, "epoch": 357, "lr": 1.8754060969188903e-05} +{"train_loss": 0.013703163713216782, "grad_norm": 0.08564954251050949, "global_step": 37888, "epoch": 357, "lr": 1.8751747241773636e-05} +{"train_loss": 0.014106020331382751, "grad_norm": 0.08250613510608673, "global_step": 37889, "epoch": 357, "lr": 1.8749433624151136e-05} +{"train_loss": 0.01967485621571541, "grad_norm": 0.13238462805747986, "global_step": 37890, "epoch": 357, "lr": 1.8747120116329554e-05} +{"train_loss": 0.016119327396154404, "grad_norm": 0.08910582959651947, "global_step": 37891, "epoch": 357, "lr": 1.874480671831698e-05} +{"train_loss": 0.01585758477449417, "grad_norm": 0.07696378231048584, "global_step": 37892, "epoch": 357, "lr": 1.8742493430121565e-05} +{"train_loss": 0.014566810801625252, "grad_norm": 0.06773362308740616, "global_step": 37893, "epoch": 357, "lr": 1.8740180251751456e-05} +{"train_loss": 0.01617196947336197, "grad_norm": 0.05805821344256401, "global_step": 37894, "epoch": 357, "lr": 1.8737867183214757e-05} +{"train_loss": 0.01455021183937788, "grad_norm": 0.10461803525686264, "global_step": 37895, "epoch": 357, "lr": 1.873555422451958e-05} +{"train_loss": 0.012843620032072067, "grad_norm": 0.07339175045490265, "global_step": 37896, "epoch": 357, "lr": 1.8733241375674065e-05} +{"train_loss": 0.014061950147151947, "grad_norm": 0.07103826850652695, "global_step": 37897, "epoch": 357, "lr": 1.8730928636686362e-05} +{"train_loss": 0.014984143897891045, "grad_norm": 0.07256051152944565, "global_step": 37898, "epoch": 357, "lr": 1.872861600756456e-05} +{"train_loss": 0.013472320511937141, "grad_norm": 0.19421473145484924, "global_step": 37899, "epoch": 357, "lr": 1.872630348831682e-05} +{"train_loss": 0.016266487538814545, "grad_norm": 0.1000261902809143, "global_step": 37900, "epoch": 357, "lr": 1.872399107895123e-05} +{"train_loss": 0.01760411635041237, "grad_norm": 0.06252147257328033, "global_step": 37901, "epoch": 357, "lr": 1.8721678779475955e-05} +{"train_loss": 0.017978083342313766, "grad_norm": 0.07677071541547775, "global_step": 37902, "epoch": 357, "lr": 1.8719366589899073e-05} +{"train_loss": 0.014980080537497997, "grad_norm": 0.1488533765077591, "global_step": 37903, "epoch": 357, "lr": 1.871705451022874e-05} +{"train_loss": 0.013991553336381912, "grad_norm": 0.10981403291225433, "global_step": 37904, "epoch": 357, "lr": 1.8714742540473086e-05} +{"train_loss": 0.016058053821325302, "grad_norm": 0.11668795347213745, "global_step": 37905, "epoch": 357, "lr": 1.8712430680640226e-05} +{"train_loss": 0.016363758593797684, "grad_norm": 0.0741971805691719, "global_step": 37906, "epoch": 357, "lr": 1.8710118930738256e-05} +{"train_loss": 0.014919211156666279, "grad_norm": 0.14160625636577606, "global_step": 37907, "epoch": 357, "lr": 1.870780729077532e-05} +{"train_loss": 0.01563769206404686, "grad_norm": 0.09237450361251831, "global_step": 37908, "epoch": 357, "lr": 1.870549576075956e-05} +{"train_loss": 0.013625839725136757, "grad_norm": 0.08495005965232849, "global_step": 37909, "epoch": 357, "lr": 1.8703184340699064e-05} +{"train_loss": 0.015059933066368103, "grad_norm": 0.0617985837161541, "global_step": 37910, "epoch": 357, "lr": 1.870087303060196e-05} +{"train_loss": 0.014947357587516308, "grad_norm": 0.10928871482610703, "global_step": 37911, "epoch": 357, "lr": 1.869856183047642e-05} +{"train_loss": 0.01655762270092964, "grad_norm": 0.129092276096344, "global_step": 37912, "epoch": 357, "lr": 1.8696250740330485e-05} +{"train_loss": 0.014560742303729057, "grad_norm": 0.07380553334951401, "global_step": 37913, "epoch": 357, "lr": 1.8693939760172308e-05} +{"train_loss": 0.01337361242622137, "grad_norm": 0.0685240626335144, "global_step": 37914, "epoch": 357, "lr": 1.8691628890010026e-05} +{"train_loss": 0.016633683815598488, "grad_norm": 0.05935505032539368, "global_step": 37915, "epoch": 357, "lr": 1.8689318129851736e-05} +{"train_loss": 0.014701604843139648, "grad_norm": 0.07701850682497025, "global_step": 37916, "epoch": 357, "lr": 1.8687007479705558e-05} +{"train_loss": 0.016405269503593445, "grad_norm": 0.14944201707839966, "global_step": 37917, "epoch": 357, "lr": 1.8684696939579638e-05} +{"train_loss": 0.014712030999362469, "grad_norm": 0.1083950623869896, "global_step": 37918, "epoch": 357, "lr": 1.8682386509482075e-05} +{"train_loss": 0.014308340847492218, "grad_norm": 0.14017881453037262, "global_step": 37919, "epoch": 357, "lr": 1.8680076189420964e-05} +{"train_loss": 0.013371866196393967, "grad_norm": 0.06021825224161148, "global_step": 37920, "epoch": 357, "lr": 1.8677765979404448e-05} +{"train_loss": 0.013310940936207771, "grad_norm": 0.0948907658457756, "global_step": 37921, "epoch": 357, "lr": 1.8675455879440663e-05} +{"train_loss": 0.015821004286408424, "grad_norm": 0.06338141858577728, "global_step": 37922, "epoch": 357, "lr": 1.867314588953768e-05} +{"train_loss": 0.016278065741062164, "grad_norm": 0.06226085126399994, "global_step": 37923, "epoch": 357, "lr": 1.8670836009703656e-05} +{"train_loss": 0.01435169205069542, "grad_norm": 0.0601433627307415, "global_step": 37924, "epoch": 357, "lr": 1.8668526239946677e-05} +{"train_loss": 0.01518921460956335, "grad_norm": 0.1441977322101593, "global_step": 37925, "epoch": 357, "lr": 1.866621658027488e-05} +{"train_loss": 0.015425939112901688, "grad_norm": 0.09202299267053604, "global_step": 37926, "epoch": 357, "lr": 1.8663907030696364e-05} +{"train_loss": 0.015962736681103706, "grad_norm": 0.0654420480132103, "global_step": 37927, "epoch": 357, "lr": 1.8661597591219242e-05} +{"train_loss": 0.016375062987208366, "grad_norm": 0.0962236300110817, "global_step": 37928, "epoch": 357, "lr": 1.865928826185166e-05} +{"train_loss": 0.015507846139371395, "grad_norm": 0.06298812478780746, "global_step": 37929, "epoch": 357, "lr": 1.865697904260171e-05} +{"train_loss": 0.015242377296090126, "grad_norm": 0.0879291221499443, "global_step": 37930, "epoch": 357, "lr": 1.8654669933477486e-05} +{"train_loss": 0.015843024477362633, "grad_norm": 0.06730862706899643, "global_step": 37931, "epoch": 357, "lr": 1.8652360934487136e-05} +{"train_loss": 0.013941140845417976, "grad_norm": 0.212297722697258, "global_step": 37932, "epoch": 357, "lr": 1.8650052045638743e-05} +{"train_loss": 0.013405858539044857, "grad_norm": 0.06732887029647827, "global_step": 37933, "epoch": 357, "lr": 1.8647743266940433e-05} +{"train_loss": 0.01472482644021511, "grad_norm": 0.061803556978702545, "global_step": 37934, "epoch": 357, "lr": 1.864543459840033e-05} +{"train_loss": 0.01783105544745922, "grad_norm": 0.10144300758838654, "global_step": 37935, "epoch": 357, "lr": 1.8643126040026537e-05} +{"train_loss": 0.012996440753340721, "grad_norm": 0.08958616852760315, "global_step": 37936, "epoch": 357, "lr": 1.8640817591827147e-05} +{"train_loss": 0.014248928055167198, "grad_norm": 0.06262294948101044, "global_step": 37937, "epoch": 357, "lr": 1.8638509253810288e-05} +{"train_loss": 0.015451177954673767, "grad_norm": 0.07026548683643341, "global_step": 37938, "epoch": 357, "lr": 1.863620102598408e-05} +{"train_loss": 0.014903733506798744, "grad_norm": 0.08646530658006668, "global_step": 37939, "epoch": 357, "lr": 1.863389290835661e-05} +{"train_loss": 0.014667152427136898, "grad_norm": 0.13649128377437592, "global_step": 37940, "epoch": 357, "lr": 1.8631584900936016e-05} +{"train_loss": 0.014035278931260109, "grad_norm": 0.08805689960718155, "global_step": 37941, "epoch": 357, "lr": 1.8629277003730372e-05} +{"train_loss": 0.015067998319864273, "grad_norm": 0.08158577233552933, "global_step": 37942, "epoch": 357, "lr": 1.8626969216747818e-05} +{"train_loss": 0.016041511669754982, "grad_norm": 0.08806327730417252, "global_step": 37943, "epoch": 357, "lr": 1.862466153999644e-05} +{"train_loss": 0.01518998946994543, "grad_norm": 0.07064786553382874, "global_step": 37944, "epoch": 357, "lr": 1.862235397348436e-05} +{"train_loss": 0.016018837690353394, "grad_norm": 0.05128873139619827, "global_step": 37945, "epoch": 357, "lr": 1.862004651721969e-05} +{"train_loss": 0.015815051272511482, "grad_norm": 0.164102703332901, "global_step": 37946, "epoch": 357, "lr": 1.8617739171210536e-05} +{"train_loss": 0.015091779712095575, "grad_norm": 0.07545733451843262, "global_step": 37947, "epoch": 357, "lr": 1.8615431935464982e-05, "val_loss": 0.019925115630030632} +{"train_loss": 0.016030820086598396, "grad_norm": 0.07336413860321045, "global_step": 37948, "epoch": 358, "lr": 1.861312480999117e-05} +{"train_loss": 0.012589074671268463, "grad_norm": 0.06552539020776749, "global_step": 37949, "epoch": 358, "lr": 1.8610817794797165e-05} +{"train_loss": 0.013548058457672596, "grad_norm": 0.06681124120950699, "global_step": 37950, "epoch": 358, "lr": 1.8608510889891094e-05} +{"train_loss": 0.014915592968463898, "grad_norm": 0.07397054880857468, "global_step": 37951, "epoch": 358, "lr": 1.8606204095281072e-05} +{"train_loss": 0.013759790919721127, "grad_norm": 0.08121099323034286, "global_step": 37952, "epoch": 358, "lr": 1.8603897410975224e-05} +{"train_loss": 0.01547409687191248, "grad_norm": 0.1072329431772232, "global_step": 37953, "epoch": 358, "lr": 1.860159083698159e-05} +{"train_loss": 0.015545256435871124, "grad_norm": 0.06739107519388199, "global_step": 37954, "epoch": 358, "lr": 1.8599284373308317e-05} +{"train_loss": 0.014433182775974274, "grad_norm": 0.06091456860303879, "global_step": 37955, "epoch": 358, "lr": 1.8596978019963518e-05} +{"train_loss": 0.016353793442249298, "grad_norm": 0.06611242145299911, "global_step": 37956, "epoch": 358, "lr": 1.8594671776955263e-05} +{"train_loss": 0.014191239140927792, "grad_norm": 0.08404491096735, "global_step": 37957, "epoch": 358, "lr": 1.8592365644291677e-05} +{"train_loss": 0.01633918657898903, "grad_norm": 0.09764360636472702, "global_step": 37958, "epoch": 358, "lr": 1.8590059621980875e-05} +{"train_loss": 0.016308430582284927, "grad_norm": 0.06674452126026154, "global_step": 37959, "epoch": 358, "lr": 1.858775371003094e-05} +{"train_loss": 0.013766194693744183, "grad_norm": 0.08771982043981552, "global_step": 37960, "epoch": 358, "lr": 1.8585447908449966e-05} +{"train_loss": 0.014561251737177372, "grad_norm": 0.12405119091272354, "global_step": 37961, "epoch": 358, "lr": 1.8583142217246065e-05} +{"train_loss": 0.013367220759391785, "grad_norm": 0.07225783914327621, "global_step": 37962, "epoch": 358, "lr": 1.8580836636427358e-05} +{"train_loss": 0.015209847129881382, "grad_norm": 0.0732412040233612, "global_step": 37963, "epoch": 358, "lr": 1.8578531166001907e-05} +{"train_loss": 0.01445087417960167, "grad_norm": 0.11784186214208603, "global_step": 37964, "epoch": 358, "lr": 1.857622580597785e-05} +{"train_loss": 0.013426505029201508, "grad_norm": 0.05665462836623192, "global_step": 37965, "epoch": 358, "lr": 1.857392055636325e-05} +{"train_loss": 0.015808768570423126, "grad_norm": 0.0686628669500351, "global_step": 37966, "epoch": 358, "lr": 1.857161541716625e-05} +{"train_loss": 0.018057169392704964, "grad_norm": 0.09382691234350204, "global_step": 37967, "epoch": 358, "lr": 1.8569310388394906e-05} +{"train_loss": 0.015795769169926643, "grad_norm": 0.08554653078317642, "global_step": 37968, "epoch": 358, "lr": 1.856700547005733e-05} +{"train_loss": 0.015147141180932522, "grad_norm": 0.08880192786455154, "global_step": 37969, "epoch": 358, "lr": 1.856470066216165e-05} +{"train_loss": 0.016341257840394974, "grad_norm": 0.07740712910890579, "global_step": 37970, "epoch": 358, "lr": 1.8562395964715944e-05} +{"train_loss": 0.016050519421696663, "grad_norm": 0.18359604477882385, "global_step": 37971, "epoch": 358, "lr": 1.8560091377728282e-05} +{"train_loss": 0.017269866541028023, "grad_norm": 0.07606088370084763, "global_step": 37972, "epoch": 358, "lr": 1.855778690120681e-05} +{"train_loss": 0.017962122336030006, "grad_norm": 0.09490690380334854, "global_step": 37973, "epoch": 358, "lr": 1.855548253515958e-05} +{"train_loss": 0.01538014318794012, "grad_norm": 0.07369814068078995, "global_step": 37974, "epoch": 358, "lr": 1.8553178279594714e-05} +{"train_loss": 0.013593866489827633, "grad_norm": 0.12867049872875214, "global_step": 37975, "epoch": 358, "lr": 1.8550874134520312e-05} +{"train_loss": 0.014570796862244606, "grad_norm": 0.06249934434890747, "global_step": 37976, "epoch": 358, "lr": 1.854857009994447e-05} +{"train_loss": 0.014810211956501007, "grad_norm": 0.06574587523937225, "global_step": 37977, "epoch": 358, "lr": 1.8546266175875254e-05} +{"train_loss": 0.015006348490715027, "grad_norm": 0.1166796013712883, "global_step": 37978, "epoch": 358, "lr": 1.8543962362320776e-05} +{"train_loss": 0.014310171827673912, "grad_norm": 0.05873946100473404, "global_step": 37979, "epoch": 358, "lr": 1.8541658659289156e-05} +{"train_loss": 0.017776960507035255, "grad_norm": 0.07585912942886353, "global_step": 37980, "epoch": 358, "lr": 1.853935506678845e-05} +{"train_loss": 0.015011417679488659, "grad_norm": 0.06778540462255478, "global_step": 37981, "epoch": 358, "lr": 1.8537051584826785e-05} +{"train_loss": 0.016580231487751007, "grad_norm": 0.09235987812280655, "global_step": 37982, "epoch": 358, "lr": 1.8534748213412223e-05} +{"train_loss": 0.014329783618450165, "grad_norm": 0.055862076580524445, "global_step": 37983, "epoch": 358, "lr": 1.8532444952552886e-05} +{"train_loss": 0.014896666631102562, "grad_norm": 0.057758938521146774, "global_step": 37984, "epoch": 358, "lr": 1.8530141802256834e-05} +{"train_loss": 0.014724042266607285, "grad_norm": 0.09411093592643738, "global_step": 37985, "epoch": 358, "lr": 1.8527838762532178e-05} +{"train_loss": 0.014458119869232178, "grad_norm": 0.16983842849731445, "global_step": 37986, "epoch": 358, "lr": 1.852553583338703e-05} +{"train_loss": 0.017061030492186546, "grad_norm": 0.12694481015205383, "global_step": 37987, "epoch": 358, "lr": 1.852323301482946e-05} +{"train_loss": 0.017685169354081154, "grad_norm": 0.08448106050491333, "global_step": 37988, "epoch": 358, "lr": 1.8520930306867544e-05} +{"train_loss": 0.015862809494137764, "grad_norm": 0.19450396299362183, "global_step": 37989, "epoch": 358, "lr": 1.8518627709509406e-05} +{"train_loss": 0.016263986006379128, "grad_norm": 0.07811571657657623, "global_step": 37990, "epoch": 358, "lr": 1.8516325222763105e-05} +{"train_loss": 0.015843849629163742, "grad_norm": 0.054187990725040436, "global_step": 37991, "epoch": 358, "lr": 1.8514022846636735e-05} +{"train_loss": 0.01465243473649025, "grad_norm": 0.09465131163597107, "global_step": 37992, "epoch": 358, "lr": 1.851172058113842e-05} +{"train_loss": 0.015431299805641174, "grad_norm": 0.07015540450811386, "global_step": 37993, "epoch": 358, "lr": 1.8509418426276228e-05} +{"train_loss": 0.014796881005167961, "grad_norm": 0.07026240974664688, "global_step": 37994, "epoch": 358, "lr": 1.850711638205822e-05} +{"train_loss": 0.013398376293480396, "grad_norm": 0.07198458164930344, "global_step": 37995, "epoch": 358, "lr": 1.850481444849252e-05} +{"train_loss": 0.0152024170383811, "grad_norm": 0.1242699921131134, "global_step": 37996, "epoch": 358, "lr": 1.850251262558721e-05} +{"train_loss": 0.015755504369735718, "grad_norm": 0.09808875620365143, "global_step": 37997, "epoch": 358, "lr": 1.8500210913350362e-05} +{"train_loss": 0.013177962973713875, "grad_norm": 0.0906267911195755, "global_step": 37998, "epoch": 358, "lr": 1.8497909311790072e-05} +{"train_loss": 0.015560917556285858, "grad_norm": 0.07727496325969696, "global_step": 37999, "epoch": 358, "lr": 1.849560782091445e-05} +{"train_loss": 0.015290698036551476, "grad_norm": 0.0823693573474884, "global_step": 38000, "epoch": 358, "lr": 1.8493306440731555e-05} +{"train_loss": 0.013435423374176025, "grad_norm": 0.0925208330154419, "global_step": 38001, "epoch": 358, "lr": 1.849100517124947e-05} +{"train_loss": 0.014751358889043331, "grad_norm": 0.06261327117681503, "global_step": 38002, "epoch": 358, "lr": 1.8488704012476283e-05} +{"train_loss": 0.016847625374794006, "grad_norm": 0.07050164043903351, "global_step": 38003, "epoch": 358, "lr": 1.8486402964420108e-05} +{"train_loss": 0.015578780323266983, "grad_norm": 0.07763548940420151, "global_step": 38004, "epoch": 358, "lr": 1.8484102027088985e-05} +{"train_loss": 0.0193149596452713, "grad_norm": 0.06338298320770264, "global_step": 38005, "epoch": 358, "lr": 1.8481801200491044e-05} +{"train_loss": 0.013978575356304646, "grad_norm": 0.15634793043136597, "global_step": 38006, "epoch": 358, "lr": 1.8479500484634327e-05} +{"train_loss": 0.0163288414478302, "grad_norm": 0.06433789432048798, "global_step": 38007, "epoch": 358, "lr": 1.8477199879526958e-05} +{"train_loss": 0.014995109289884567, "grad_norm": 0.12394341826438904, "global_step": 38008, "epoch": 358, "lr": 1.847489938517698e-05} +{"train_loss": 0.014540842734277248, "grad_norm": 0.09399701654911041, "global_step": 38009, "epoch": 358, "lr": 1.8472599001592494e-05} +{"train_loss": 0.01514504849910736, "grad_norm": 0.14236216247081757, "global_step": 38010, "epoch": 358, "lr": 1.8470298728781604e-05} +{"train_loss": 0.013568123802542686, "grad_norm": 0.06785689294338226, "global_step": 38011, "epoch": 358, "lr": 1.846799856675237e-05} +{"train_loss": 0.017533008009195328, "grad_norm": 0.08157739043235779, "global_step": 38012, "epoch": 358, "lr": 1.8465698515512857e-05} +{"train_loss": 0.015687964856624603, "grad_norm": 0.0711027979850769, "global_step": 38013, "epoch": 358, "lr": 1.8463398575071188e-05} +{"train_loss": 0.012987183406949043, "grad_norm": 0.08523590117692947, "global_step": 38014, "epoch": 358, "lr": 1.84610987454354e-05} +{"train_loss": 0.013153595849871635, "grad_norm": 0.07357311248779297, "global_step": 38015, "epoch": 358, "lr": 1.84587990266136e-05} +{"train_loss": 0.017062444239854813, "grad_norm": 0.13101357221603394, "global_step": 38016, "epoch": 358, "lr": 1.8456499418613877e-05} +{"train_loss": 0.012355362996459007, "grad_norm": 0.1507037878036499, "global_step": 38017, "epoch": 358, "lr": 1.8454199921444298e-05} +{"train_loss": 0.015827611088752747, "grad_norm": 0.0665082037448883, "global_step": 38018, "epoch": 358, "lr": 1.8451900535112928e-05} +{"train_loss": 0.016974397003650665, "grad_norm": 0.07846735417842865, "global_step": 38019, "epoch": 358, "lr": 1.8449601259627853e-05} +{"train_loss": 0.017215022817254066, "grad_norm": 0.0744820386171341, "global_step": 38020, "epoch": 358, "lr": 1.8447302094997182e-05} +{"train_loss": 0.01188617292791605, "grad_norm": 0.0565078966319561, "global_step": 38021, "epoch": 358, "lr": 1.8445003041228948e-05} +{"train_loss": 0.012231128290295601, "grad_norm": 0.12219733744859695, "global_step": 38022, "epoch": 358, "lr": 1.8442704098331248e-05} +{"train_loss": 0.015736252069473267, "grad_norm": 0.13571548461914062, "global_step": 38023, "epoch": 358, "lr": 1.8440405266312184e-05} +{"train_loss": 0.01795690506696701, "grad_norm": 0.07448969036340714, "global_step": 38024, "epoch": 358, "lr": 1.8438106545179812e-05} +{"train_loss": 0.012926280498504639, "grad_norm": 0.1325446516275406, "global_step": 38025, "epoch": 358, "lr": 1.8435807934942183e-05} +{"train_loss": 0.014241978526115417, "grad_norm": 0.05957665294408798, "global_step": 38026, "epoch": 358, "lr": 1.8433509435607405e-05} +{"train_loss": 0.016486691311001778, "grad_norm": 0.06999140977859497, "global_step": 38027, "epoch": 358, "lr": 1.843121104718356e-05} +{"train_loss": 0.014281881973147392, "grad_norm": 0.08843867480754852, "global_step": 38028, "epoch": 358, "lr": 1.8428912769678695e-05} +{"train_loss": 0.01729150116443634, "grad_norm": 0.12856720387935638, "global_step": 38029, "epoch": 358, "lr": 1.8426614603100922e-05} +{"train_loss": 0.013733448460698128, "grad_norm": 0.1396668404340744, "global_step": 38030, "epoch": 358, "lr": 1.8424316547458288e-05} +{"train_loss": 0.01570052094757557, "grad_norm": 0.07419102638959885, "global_step": 38031, "epoch": 358, "lr": 1.842201860275886e-05} +{"train_loss": 0.0171667467802763, "grad_norm": 0.059260573238134384, "global_step": 38032, "epoch": 358, "lr": 1.8419720769010724e-05} +{"train_loss": 0.014638648368418217, "grad_norm": 0.09606052935123444, "global_step": 38033, "epoch": 358, "lr": 1.8417423046221975e-05} +{"train_loss": 0.016796156764030457, "grad_norm": 0.09852010011672974, "global_step": 38034, "epoch": 358, "lr": 1.8415125434400665e-05} +{"train_loss": 0.014941608533263206, "grad_norm": 0.10403898358345032, "global_step": 38035, "epoch": 358, "lr": 1.841282793355485e-05} +{"train_loss": 0.012853972613811493, "grad_norm": 0.0721266120672226, "global_step": 38036, "epoch": 358, "lr": 1.8410530543692622e-05} +{"train_loss": 0.014050250872969627, "grad_norm": 0.09108021855354309, "global_step": 38037, "epoch": 358, "lr": 1.8408233264822067e-05} +{"train_loss": 0.01346739660948515, "grad_norm": 0.06266108900308609, "global_step": 38038, "epoch": 358, "lr": 1.8405936096951226e-05} +{"train_loss": 0.016952183097600937, "grad_norm": 0.10349169373512268, "global_step": 38039, "epoch": 358, "lr": 1.840363904008819e-05} +{"train_loss": 0.014840403571724892, "grad_norm": 0.10937810689210892, "global_step": 38040, "epoch": 358, "lr": 1.8401342094241036e-05} +{"train_loss": 0.015024365857243538, "grad_norm": 0.048686232417821884, "global_step": 38041, "epoch": 358, "lr": 1.8399045259417824e-05} +{"train_loss": 0.013519012369215488, "grad_norm": 0.07315580546855927, "global_step": 38042, "epoch": 358, "lr": 1.839674853562661e-05} +{"train_loss": 0.016487497836351395, "grad_norm": 0.09637217968702316, "global_step": 38043, "epoch": 358, "lr": 1.839445192287548e-05} +{"train_loss": 0.013663090765476227, "grad_norm": 0.12506215274333954, "global_step": 38044, "epoch": 358, "lr": 1.8392155421172513e-05} +{"train_loss": 0.014371082186698914, "grad_norm": 0.07857610285282135, "global_step": 38045, "epoch": 358, "lr": 1.8389859030525752e-05} +{"train_loss": 0.014008753001689911, "grad_norm": 0.11159399896860123, "global_step": 38046, "epoch": 358, "lr": 1.8387562750943293e-05} +{"train_loss": 0.015009867027401924, "grad_norm": 0.07554271072149277, "global_step": 38047, "epoch": 358, "lr": 1.8385266582433192e-05} +{"train_loss": 0.01233441848307848, "grad_norm": 0.05503711476922035, "global_step": 38048, "epoch": 358, "lr": 1.83829705250035e-05} +{"train_loss": 0.013434027321636677, "grad_norm": 0.05536044389009476, "global_step": 38049, "epoch": 358, "lr": 1.83806745786623e-05} +{"train_loss": 0.014726651832461357, "grad_norm": 0.07106699794530869, "global_step": 38050, "epoch": 358, "lr": 1.8378378743417673e-05} +{"train_loss": 0.015436914749443531, "grad_norm": 0.08964315801858902, "global_step": 38051, "epoch": 358, "lr": 1.8376083019277655e-05} +{"train_loss": 0.013295494019985199, "grad_norm": 0.07670023292303085, "global_step": 38052, "epoch": 358, "lr": 1.837378740625035e-05} +{"train_loss": 0.015079653303507926, "grad_norm": 0.0905788466334343, "global_step": 38053, "epoch": 358, "lr": 1.837149190434378e-05, "val_loss": 0.019323382526636124} +{"train_loss": 0.0161486454308033, "grad_norm": 0.07087408751249313, "global_step": 38054, "epoch": 359, "lr": 1.836919651356605e-05} +{"train_loss": 0.01605568826198578, "grad_norm": 0.10283952951431274, "global_step": 38055, "epoch": 359, "lr": 1.8366901233925192e-05} +{"train_loss": 0.015942523255944252, "grad_norm": 0.09941909462213516, "global_step": 38056, "epoch": 359, "lr": 1.8364606065429284e-05} +{"train_loss": 0.013455185107886791, "grad_norm": 0.0817451998591423, "global_step": 38057, "epoch": 359, "lr": 1.8362311008086412e-05} +{"train_loss": 0.01481664925813675, "grad_norm": 0.07236145436763763, "global_step": 38058, "epoch": 359, "lr": 1.836001606190462e-05} +{"train_loss": 0.015853151679039, "grad_norm": 0.10590475052595139, "global_step": 38059, "epoch": 359, "lr": 1.8357721226891954e-05} +{"train_loss": 0.016103338450193405, "grad_norm": 0.14198248088359833, "global_step": 38060, "epoch": 359, "lr": 1.8355426503056494e-05} +{"train_loss": 0.014354698359966278, "grad_norm": 0.08269619196653366, "global_step": 38061, "epoch": 359, "lr": 1.8353131890406323e-05} +{"train_loss": 0.016227178275585175, "grad_norm": 0.082113116979599, "global_step": 38062, "epoch": 359, "lr": 1.8350837388949464e-05} +{"train_loss": 0.016892464831471443, "grad_norm": 0.18773043155670166, "global_step": 38063, "epoch": 359, "lr": 1.8348542998694e-05} +{"train_loss": 0.016967114061117172, "grad_norm": 0.06621208786964417, "global_step": 38064, "epoch": 359, "lr": 1.8346248719648006e-05} +{"train_loss": 0.014881136827170849, "grad_norm": 0.10176945477724075, "global_step": 38065, "epoch": 359, "lr": 1.8343954551819532e-05} +{"train_loss": 0.011642476543784142, "grad_norm": 0.08245786279439926, "global_step": 38066, "epoch": 359, "lr": 1.8341660495216618e-05} +{"train_loss": 0.01893373392522335, "grad_norm": 0.10240956395864487, "global_step": 38067, "epoch": 359, "lr": 1.8339366549847338e-05} +{"train_loss": 0.015237627550959587, "grad_norm": 0.08101587742567062, "global_step": 38068, "epoch": 359, "lr": 1.8337072715719772e-05} +{"train_loss": 0.014176237396895885, "grad_norm": 0.11077825725078583, "global_step": 38069, "epoch": 359, "lr": 1.8334778992841943e-05} +{"train_loss": 0.016400881111621857, "grad_norm": 0.06863780319690704, "global_step": 38070, "epoch": 359, "lr": 1.8332485381221954e-05} +{"train_loss": 0.014289362356066704, "grad_norm": 0.1321098655462265, "global_step": 38071, "epoch": 359, "lr": 1.8330191880867836e-05} +{"train_loss": 0.013046767562627792, "grad_norm": 0.04837733507156372, "global_step": 38072, "epoch": 359, "lr": 1.8327898491787633e-05} +{"train_loss": 0.01306702196598053, "grad_norm": 0.10528039932250977, "global_step": 38073, "epoch": 359, "lr": 1.8325605213989423e-05} +{"train_loss": 0.016665007919073105, "grad_norm": 0.09974843263626099, "global_step": 38074, "epoch": 359, "lr": 1.832331204748128e-05} +{"train_loss": 0.02002551220357418, "grad_norm": 0.11077901721000671, "global_step": 38075, "epoch": 359, "lr": 1.8321018992271222e-05} +{"train_loss": 0.01550003606826067, "grad_norm": 0.07239741832017899, "global_step": 38076, "epoch": 359, "lr": 1.8318726048367342e-05} +{"train_loss": 0.013117348775267601, "grad_norm": 0.11270471662282944, "global_step": 38077, "epoch": 359, "lr": 1.831643321577767e-05} +{"train_loss": 0.0160372331738472, "grad_norm": 0.07116537541151047, "global_step": 38078, "epoch": 359, "lr": 1.831414049451028e-05} +{"train_loss": 0.0160149484872818, "grad_norm": 0.12481097877025604, "global_step": 38079, "epoch": 359, "lr": 1.831184788457321e-05} +{"train_loss": 0.015369033440947533, "grad_norm": 0.10247096419334412, "global_step": 38080, "epoch": 359, "lr": 1.8309555385974525e-05} +{"train_loss": 0.013774638995528221, "grad_norm": 0.06442642211914062, "global_step": 38081, "epoch": 359, "lr": 1.8307262998722297e-05} +{"train_loss": 0.015274347737431526, "grad_norm": 0.07556004077196121, "global_step": 38082, "epoch": 359, "lr": 1.8304970722824567e-05} +{"train_loss": 0.016394006088376045, "grad_norm": 0.08129526674747467, "global_step": 38083, "epoch": 359, "lr": 1.8302678558289366e-05} +{"train_loss": 0.014876286499202251, "grad_norm": 0.07421280443668365, "global_step": 38084, "epoch": 359, "lr": 1.8300386505124768e-05} +{"train_loss": 0.015158412978053093, "grad_norm": 0.07855647802352905, "global_step": 38085, "epoch": 359, "lr": 1.8298094563338842e-05} +{"train_loss": 0.01647711545228958, "grad_norm": 0.10022452473640442, "global_step": 38086, "epoch": 359, "lr": 1.829580273293961e-05} +{"train_loss": 0.014201817102730274, "grad_norm": 0.05918031930923462, "global_step": 38087, "epoch": 359, "lr": 1.8293511013935156e-05} +{"train_loss": 0.013723144307732582, "grad_norm": 0.08632326126098633, "global_step": 38088, "epoch": 359, "lr": 1.8291219406333516e-05} +{"train_loss": 0.013957876712083817, "grad_norm": 0.07762234658002853, "global_step": 38089, "epoch": 359, "lr": 1.8288927910142724e-05} +{"train_loss": 0.015160651877522469, "grad_norm": 0.12294352799654007, "global_step": 38090, "epoch": 359, "lr": 1.8286636525370848e-05} +{"train_loss": 0.018438752740621567, "grad_norm": 0.11579085141420364, "global_step": 38091, "epoch": 359, "lr": 1.828434525202596e-05} +{"train_loss": 0.014756416901946068, "grad_norm": 0.05557294189929962, "global_step": 38092, "epoch": 359, "lr": 1.8282054090116062e-05} +{"train_loss": 0.016121279448270798, "grad_norm": 0.11032083630561829, "global_step": 38093, "epoch": 359, "lr": 1.8279763039649263e-05} +{"train_loss": 0.01621956378221512, "grad_norm": 0.06751836836338043, "global_step": 38094, "epoch": 359, "lr": 1.8277472100633558e-05} +{"train_loss": 0.014603251591324806, "grad_norm": 0.08694013953208923, "global_step": 38095, "epoch": 359, "lr": 1.8275181273077034e-05} +{"train_loss": 0.015764310956001282, "grad_norm": 0.11136478185653687, "global_step": 38096, "epoch": 359, "lr": 1.8272890556987716e-05} +{"train_loss": 0.015528984367847443, "grad_norm": 0.061974480748176575, "global_step": 38097, "epoch": 359, "lr": 1.8270599952373656e-05} +{"train_loss": 0.015199248678982258, "grad_norm": 0.0601985789835453, "global_step": 38098, "epoch": 359, "lr": 1.826830945924293e-05} +{"train_loss": 0.012947946786880493, "grad_norm": 0.09864718466997147, "global_step": 38099, "epoch": 359, "lr": 1.826601907760357e-05} +{"train_loss": 0.01443685032427311, "grad_norm": 0.08425033092498779, "global_step": 38100, "epoch": 359, "lr": 1.8263728807463592e-05} +{"train_loss": 0.014456238597631454, "grad_norm": 0.1414949744939804, "global_step": 38101, "epoch": 359, "lr": 1.8261438648831075e-05} +{"train_loss": 0.012539174407720566, "grad_norm": 0.09686222672462463, "global_step": 38102, "epoch": 359, "lr": 1.8259148601714077e-05} +{"train_loss": 0.018352970480918884, "grad_norm": 0.08084601908922195, "global_step": 38103, "epoch": 359, "lr": 1.8256858666120608e-05} +{"train_loss": 0.014005133882164955, "grad_norm": 0.07465668767690659, "global_step": 38104, "epoch": 359, "lr": 1.8254568842058735e-05} +{"train_loss": 0.013500526547431946, "grad_norm": 0.06743825972080231, "global_step": 38105, "epoch": 359, "lr": 1.8252279129536532e-05} +{"train_loss": 0.013235637918114662, "grad_norm": 0.06836442649364471, "global_step": 38106, "epoch": 359, "lr": 1.824998952856198e-05} +{"train_loss": 0.012947624549269676, "grad_norm": 0.06070193275809288, "global_step": 38107, "epoch": 359, "lr": 1.8247700039143156e-05} +{"train_loss": 0.01664773002266884, "grad_norm": 0.0633043646812439, "global_step": 38108, "epoch": 359, "lr": 1.824541066128811e-05} +{"train_loss": 0.014670761302113533, "grad_norm": 0.13229960203170776, "global_step": 38109, "epoch": 359, "lr": 1.8243121395004893e-05} +{"train_loss": 0.014704618602991104, "grad_norm": 0.05969005078077316, "global_step": 38110, "epoch": 359, "lr": 1.8240832240301527e-05} +{"train_loss": 0.014895647764205933, "grad_norm": 0.11567895859479904, "global_step": 38111, "epoch": 359, "lr": 1.8238543197186074e-05} +{"train_loss": 0.013479093089699745, "grad_norm": 0.1035405695438385, "global_step": 38112, "epoch": 359, "lr": 1.823625426566657e-05} +{"train_loss": 0.01514262706041336, "grad_norm": 0.062035392969846725, "global_step": 38113, "epoch": 359, "lr": 1.823396544575104e-05} +{"train_loss": 0.014106418937444687, "grad_norm": 0.07217831164598465, "global_step": 38114, "epoch": 359, "lr": 1.823167673744754e-05} +{"train_loss": 0.01401571650058031, "grad_norm": 0.13554657995700836, "global_step": 38115, "epoch": 359, "lr": 1.8229388140764125e-05} +{"train_loss": 0.011915406212210655, "grad_norm": 0.0755876824259758, "global_step": 38116, "epoch": 359, "lr": 1.8227099655708812e-05} +{"train_loss": 0.012791477143764496, "grad_norm": 0.0652114599943161, "global_step": 38117, "epoch": 359, "lr": 1.8224811282289666e-05} +{"train_loss": 0.014734991826117039, "grad_norm": 0.06080213934183121, "global_step": 38118, "epoch": 359, "lr": 1.8222523020514698e-05} +{"train_loss": 0.016692452132701874, "grad_norm": 0.12138980627059937, "global_step": 38119, "epoch": 359, "lr": 1.8220234870391983e-05} +{"train_loss": 0.014555403031408787, "grad_norm": 0.12773603200912476, "global_step": 38120, "epoch": 359, "lr": 1.8217946831929526e-05} +{"train_loss": 0.014204604551196098, "grad_norm": 0.10792354494333267, "global_step": 38121, "epoch": 359, "lr": 1.821565890513538e-05} +{"train_loss": 0.015481078065931797, "grad_norm": 0.10682831704616547, "global_step": 38122, "epoch": 359, "lr": 1.8213371090017605e-05} +{"train_loss": 0.01610998809337616, "grad_norm": 0.08171014487743378, "global_step": 38123, "epoch": 359, "lr": 1.821108338658422e-05} +{"train_loss": 0.014302439987659454, "grad_norm": 0.06906785070896149, "global_step": 38124, "epoch": 359, "lr": 1.8208795794843247e-05} +{"train_loss": 0.015563826076686382, "grad_norm": 0.07169866561889648, "global_step": 38125, "epoch": 359, "lr": 1.820650831480274e-05} +{"train_loss": 0.015312234871089458, "grad_norm": 0.07031438499689102, "global_step": 38126, "epoch": 359, "lr": 1.8204220946470752e-05} +{"train_loss": 0.014314251020550728, "grad_norm": 0.05619322136044502, "global_step": 38127, "epoch": 359, "lr": 1.8201933689855284e-05} +{"train_loss": 0.014383552595973015, "grad_norm": 0.0866122767329216, "global_step": 38128, "epoch": 359, "lr": 1.8199646544964393e-05} +{"train_loss": 0.01494948286563158, "grad_norm": 0.06572570651769638, "global_step": 38129, "epoch": 359, "lr": 1.8197359511806154e-05} +{"train_loss": 0.015732474625110626, "grad_norm": 0.07670089602470398, "global_step": 38130, "epoch": 359, "lr": 1.8195072590388522e-05} +{"train_loss": 0.012434910051524639, "grad_norm": 0.06622390449047089, "global_step": 38131, "epoch": 359, "lr": 1.819278578071958e-05} +{"train_loss": 0.01579664833843708, "grad_norm": 0.07635574042797089, "global_step": 38132, "epoch": 359, "lr": 1.8190499082807373e-05} +{"train_loss": 0.013436335138976574, "grad_norm": 0.04600120708346367, "global_step": 38133, "epoch": 359, "lr": 1.8188212496659896e-05} +{"train_loss": 0.013253219425678253, "grad_norm": 0.055268023163080215, "global_step": 38134, "epoch": 359, "lr": 1.8185926022285226e-05} +{"train_loss": 0.01604025438427925, "grad_norm": 0.07348380237817764, "global_step": 38135, "epoch": 359, "lr": 1.818363965969136e-05} +{"train_loss": 0.01641983725130558, "grad_norm": 0.20472349226474762, "global_step": 38136, "epoch": 359, "lr": 1.8181353408886365e-05} +{"train_loss": 0.01725700870156288, "grad_norm": 0.14818041026592255, "global_step": 38137, "epoch": 359, "lr": 1.8179067269878236e-05} +{"train_loss": 0.014223843812942505, "grad_norm": 0.1147104948759079, "global_step": 38138, "epoch": 359, "lr": 1.8176781242675027e-05} +{"train_loss": 0.01354131568223238, "grad_norm": 0.06739542633295059, "global_step": 38139, "epoch": 359, "lr": 1.8174495327284786e-05} +{"train_loss": 0.01544104516506195, "grad_norm": 0.0801801010966301, "global_step": 38140, "epoch": 359, "lr": 1.8172209523715535e-05} +{"train_loss": 0.016275256872177124, "grad_norm": 0.1189841479063034, "global_step": 38141, "epoch": 359, "lr": 1.8169923831975273e-05} +{"train_loss": 0.014593547210097313, "grad_norm": 0.05925862118601799, "global_step": 38142, "epoch": 359, "lr": 1.8167638252072054e-05} +{"train_loss": 0.014928330667316914, "grad_norm": 0.11088287085294724, "global_step": 38143, "epoch": 359, "lr": 1.8165352784013932e-05} +{"train_loss": 0.016440128907561302, "grad_norm": 0.08648387342691422, "global_step": 38144, "epoch": 359, "lr": 1.81630674278089e-05} +{"train_loss": 0.017352072522044182, "grad_norm": 0.09095130115747452, "global_step": 38145, "epoch": 359, "lr": 1.8160782183465003e-05} +{"train_loss": 0.012288825586438179, "grad_norm": 0.09895624965429306, "global_step": 38146, "epoch": 359, "lr": 1.8158497050990307e-05} +{"train_loss": 0.019420746713876724, "grad_norm": 0.07602375000715256, "global_step": 38147, "epoch": 359, "lr": 1.8156212030392766e-05} +{"train_loss": 0.01755082607269287, "grad_norm": 0.07016324996948242, "global_step": 38148, "epoch": 359, "lr": 1.8153927121680453e-05} +{"train_loss": 0.01772569864988327, "grad_norm": 0.0643647238612175, "global_step": 38149, "epoch": 359, "lr": 1.81516423248614e-05} +{"train_loss": 0.013670118525624275, "grad_norm": 0.0716436356306076, "global_step": 38150, "epoch": 359, "lr": 1.8149357639943615e-05} +{"train_loss": 0.013828995637595654, "grad_norm": 0.09504474699497223, "global_step": 38151, "epoch": 359, "lr": 1.8147073066935133e-05} +{"train_loss": 0.012602295726537704, "grad_norm": 0.08156115561723709, "global_step": 38152, "epoch": 359, "lr": 1.8144788605844005e-05} +{"train_loss": 0.015289687551558018, "grad_norm": 0.09954453259706497, "global_step": 38153, "epoch": 359, "lr": 1.814250425667823e-05} +{"train_loss": 0.01615491509437561, "grad_norm": 0.10185148566961288, "global_step": 38154, "epoch": 359, "lr": 1.8140220019445824e-05} +{"train_loss": 0.014567714184522629, "grad_norm": 0.06619448959827423, "global_step": 38155, "epoch": 359, "lr": 1.8137935894154834e-05} +{"train_loss": 0.016272064298391342, "grad_norm": 0.09344851225614548, "global_step": 38156, "epoch": 359, "lr": 1.8135651880813298e-05} +{"train_loss": 0.015969131141901016, "grad_norm": 0.07048697769641876, "global_step": 38157, "epoch": 359, "lr": 1.81333679794292e-05} +{"train_loss": 0.015101175755262375, "grad_norm": 0.06446792930364609, "global_step": 38158, "epoch": 359, "lr": 1.813108419001061e-05} +{"train_loss": 0.015083930604511275, "grad_norm": 0.10507934540510178, "global_step": 38159, "epoch": 359, "lr": 1.8128800512565513e-05, "val_loss": 0.01960679702460766} +{"train_loss": 0.016371959820389748, "grad_norm": 0.05736055597662926, "global_step": 38160, "epoch": 360, "lr": 1.812651694710197e-05} +{"train_loss": 0.014417664147913456, "grad_norm": 0.10308317840099335, "global_step": 38161, "epoch": 360, "lr": 1.8124233493627968e-05} +{"train_loss": 0.011580714024603367, "grad_norm": 0.07535071671009064, "global_step": 38162, "epoch": 360, "lr": 1.8121950152151546e-05} +{"train_loss": 0.015974871814250946, "grad_norm": 0.07785067707300186, "global_step": 38163, "epoch": 360, "lr": 1.8119666922680744e-05} +{"train_loss": 0.015680823475122452, "grad_norm": 0.09564998000860214, "global_step": 38164, "epoch": 360, "lr": 1.8117383805223564e-05} +{"train_loss": 0.013779610395431519, "grad_norm": 0.14212913811206818, "global_step": 38165, "epoch": 360, "lr": 1.811510079978802e-05} +{"train_loss": 0.01722131296992302, "grad_norm": 0.10200832039117813, "global_step": 38166, "epoch": 360, "lr": 1.8112817906382147e-05} +{"train_loss": 0.015776101499795914, "grad_norm": 0.10284678637981415, "global_step": 38167, "epoch": 360, "lr": 1.8110535125013977e-05} +{"train_loss": 0.016688818112015724, "grad_norm": 0.159245103597641, "global_step": 38168, "epoch": 360, "lr": 1.8108252455691504e-05} +{"train_loss": 0.014420028775930405, "grad_norm": 0.10220412909984589, "global_step": 38169, "epoch": 360, "lr": 1.810596989842276e-05} +{"train_loss": 0.015709105879068375, "grad_norm": 0.07339079678058624, "global_step": 38170, "epoch": 360, "lr": 1.8103687453215805e-05} +{"train_loss": 0.013607488945126534, "grad_norm": 0.07243609428405762, "global_step": 38171, "epoch": 360, "lr": 1.8101405120078585e-05} +{"train_loss": 0.015168607234954834, "grad_norm": 0.07580253481864929, "global_step": 38172, "epoch": 360, "lr": 1.8099122899019155e-05} +{"train_loss": 0.01581634022295475, "grad_norm": 0.11439663171768188, "global_step": 38173, "epoch": 360, "lr": 1.809684079004555e-05} +{"train_loss": 0.016188813373446465, "grad_norm": 0.06720427423715591, "global_step": 38174, "epoch": 360, "lr": 1.8094558793165754e-05} +{"train_loss": 0.015233910642564297, "grad_norm": 0.08211463689804077, "global_step": 38175, "epoch": 360, "lr": 1.8092276908387806e-05} +{"train_loss": 0.013509416952729225, "grad_norm": 0.10188025236129761, "global_step": 38176, "epoch": 360, "lr": 1.808999513571974e-05} +{"train_loss": 0.01683558151125908, "grad_norm": 0.12271080911159515, "global_step": 38177, "epoch": 360, "lr": 1.8087713475169548e-05} +{"train_loss": 0.018207304179668427, "grad_norm": 0.07751091569662094, "global_step": 38178, "epoch": 360, "lr": 1.808543192674524e-05} +{"train_loss": 0.01532670110464096, "grad_norm": 0.06659189611673355, "global_step": 38179, "epoch": 360, "lr": 1.808315049045484e-05} +{"train_loss": 0.013691452331840992, "grad_norm": 0.08622436225414276, "global_step": 38180, "epoch": 360, "lr": 1.808086916630639e-05} +{"train_loss": 0.013502003625035286, "grad_norm": 0.10592078417539597, "global_step": 38181, "epoch": 360, "lr": 1.807858795430787e-05} +{"train_loss": 0.014127512462437153, "grad_norm": 0.10252659022808075, "global_step": 38182, "epoch": 360, "lr": 1.8076306854467323e-05} +{"train_loss": 0.014138709753751755, "grad_norm": 0.05971797928214073, "global_step": 38183, "epoch": 360, "lr": 1.8074025866792734e-05} +{"train_loss": 0.01468750461935997, "grad_norm": 0.11372469365596771, "global_step": 38184, "epoch": 360, "lr": 1.807174499129215e-05} +{"train_loss": 0.014428680762648582, "grad_norm": 0.06896715611219406, "global_step": 38185, "epoch": 360, "lr": 1.806946422797355e-05} +{"train_loss": 0.01413032691925764, "grad_norm": 0.09246032685041428, "global_step": 38186, "epoch": 360, "lr": 1.806718357684497e-05} +{"train_loss": 0.014571068808436394, "grad_norm": 0.09015355259180069, "global_step": 38187, "epoch": 360, "lr": 1.806490303791445e-05} +{"train_loss": 0.01291763223707676, "grad_norm": 0.09238283336162567, "global_step": 38188, "epoch": 360, "lr": 1.8062622611189944e-05} +{"train_loss": 0.016297750174999237, "grad_norm": 0.0669090747833252, "global_step": 38189, "epoch": 360, "lr": 1.806034229667949e-05} +{"train_loss": 0.014709718525409698, "grad_norm": 0.12044144421815872, "global_step": 38190, "epoch": 360, "lr": 1.805806209439112e-05} +{"train_loss": 0.014388103038072586, "grad_norm": 0.05690751224756241, "global_step": 38191, "epoch": 360, "lr": 1.8055782004332812e-05} +{"train_loss": 0.01699008420109749, "grad_norm": 0.10815462470054626, "global_step": 38192, "epoch": 360, "lr": 1.8053502026512593e-05} +{"train_loss": 0.016553115099668503, "grad_norm": 0.1390542834997177, "global_step": 38193, "epoch": 360, "lr": 1.805122216093849e-05} +{"train_loss": 0.013579707592725754, "grad_norm": 0.0659516453742981, "global_step": 38194, "epoch": 360, "lr": 1.8048942407618503e-05} +{"train_loss": 0.01554112508893013, "grad_norm": 0.06957828998565674, "global_step": 38195, "epoch": 360, "lr": 1.8046662766560617e-05} +{"train_loss": 0.013303242623806, "grad_norm": 0.09605982899665833, "global_step": 38196, "epoch": 360, "lr": 1.804438323777286e-05} +{"train_loss": 0.014010756276547909, "grad_norm": 0.07810506224632263, "global_step": 38197, "epoch": 360, "lr": 1.8042103821263264e-05} +{"train_loss": 0.014918822795152664, "grad_norm": 0.0644792839884758, "global_step": 38198, "epoch": 360, "lr": 1.80398245170398e-05} +{"train_loss": 0.01631590723991394, "grad_norm": 0.11572613567113876, "global_step": 38199, "epoch": 360, "lr": 1.8037545325110504e-05} +{"train_loss": 0.013729476369917393, "grad_norm": 0.09170664846897125, "global_step": 38200, "epoch": 360, "lr": 1.803526624548337e-05} +{"train_loss": 0.014672614634037018, "grad_norm": 0.05363739654421806, "global_step": 38201, "epoch": 360, "lr": 1.8032987278166412e-05} +{"train_loss": 0.020585227757692337, "grad_norm": 0.07917812466621399, "global_step": 38202, "epoch": 360, "lr": 1.8030708423167623e-05} +{"train_loss": 0.013513801619410515, "grad_norm": 0.0817909985780716, "global_step": 38203, "epoch": 360, "lr": 1.8028429680495022e-05} +{"train_loss": 0.0132647305727005, "grad_norm": 0.07376983761787415, "global_step": 38204, "epoch": 360, "lr": 1.8026151050156635e-05} +{"train_loss": 0.016012845560908318, "grad_norm": 0.0615546777844429, "global_step": 38205, "epoch": 360, "lr": 1.8023872532160445e-05} +{"train_loss": 0.01371760293841362, "grad_norm": 0.1427893340587616, "global_step": 38206, "epoch": 360, "lr": 1.8021594126514448e-05} +{"train_loss": 0.015037482604384422, "grad_norm": 0.07684748619794846, "global_step": 38207, "epoch": 360, "lr": 1.8019315833226676e-05} +{"train_loss": 0.016911320388317108, "grad_norm": 0.09427008032798767, "global_step": 38208, "epoch": 360, "lr": 1.8017037652305102e-05} +{"train_loss": 0.014699007384479046, "grad_norm": 0.06319473683834076, "global_step": 38209, "epoch": 360, "lr": 1.8014759583757747e-05} +{"train_loss": 0.013989198952913284, "grad_norm": 0.0780472531914711, "global_step": 38210, "epoch": 360, "lr": 1.801248162759262e-05} +{"train_loss": 0.017208285629749298, "grad_norm": 0.15089045464992523, "global_step": 38211, "epoch": 360, "lr": 1.8010203783817754e-05} +{"train_loss": 0.016732174903154373, "grad_norm": 0.13123272359371185, "global_step": 38212, "epoch": 360, "lr": 1.800792605244109e-05} +{"train_loss": 0.017508113756775856, "grad_norm": 0.07550125569105148, "global_step": 38213, "epoch": 360, "lr": 1.8005648433470656e-05} +{"train_loss": 0.016861164942383766, "grad_norm": 0.11230680346488953, "global_step": 38214, "epoch": 360, "lr": 1.8003370926914475e-05} +{"train_loss": 0.014432390220463276, "grad_norm": 0.12392957508563995, "global_step": 38215, "epoch": 360, "lr": 1.8001093532780522e-05} +{"train_loss": 0.015503563918173313, "grad_norm": 0.07655332237482071, "global_step": 38216, "epoch": 360, "lr": 1.7998816251076807e-05} +{"train_loss": 0.0166843980550766, "grad_norm": 0.06938940286636353, "global_step": 38217, "epoch": 360, "lr": 1.7996539081811354e-05} +{"train_loss": 0.01477053388953209, "grad_norm": 0.07639703154563904, "global_step": 38218, "epoch": 360, "lr": 1.799426202499214e-05} +{"train_loss": 0.015036994591355324, "grad_norm": 0.0874096155166626, "global_step": 38219, "epoch": 360, "lr": 1.799198508062716e-05} +{"train_loss": 0.014505471102893353, "grad_norm": 0.07876082509756088, "global_step": 38220, "epoch": 360, "lr": 1.798970824872442e-05} +{"train_loss": 0.01325777918100357, "grad_norm": 0.07936137914657593, "global_step": 38221, "epoch": 360, "lr": 1.7987431529291944e-05} +{"train_loss": 0.014168379828333855, "grad_norm": 0.08009466528892517, "global_step": 38222, "epoch": 360, "lr": 1.7985154922337692e-05} +{"train_loss": 0.017060261219739914, "grad_norm": 0.05643664300441742, "global_step": 38223, "epoch": 360, "lr": 1.7982878427869698e-05} +{"train_loss": 0.015926910564303398, "grad_norm": 0.12264171242713928, "global_step": 38224, "epoch": 360, "lr": 1.7980602045895932e-05} +{"train_loss": 0.014816942624747753, "grad_norm": 0.12501022219657898, "global_step": 38225, "epoch": 360, "lr": 1.7978325776424422e-05} +{"train_loss": 0.015276473015546799, "grad_norm": 0.07712194323539734, "global_step": 38226, "epoch": 360, "lr": 1.7976049619463132e-05} +{"train_loss": 0.015715645626187325, "grad_norm": 0.21223096549510956, "global_step": 38227, "epoch": 360, "lr": 1.7973773575020077e-05} +{"train_loss": 0.014672180637717247, "grad_norm": 0.07936156541109085, "global_step": 38228, "epoch": 360, "lr": 1.7971497643103274e-05} +{"train_loss": 0.016549523919820786, "grad_norm": 0.0952310636639595, "global_step": 38229, "epoch": 360, "lr": 1.7969221823720695e-05} +{"train_loss": 0.015489565208554268, "grad_norm": 0.10006192326545715, "global_step": 38230, "epoch": 360, "lr": 1.7966946116880323e-05} +{"train_loss": 0.016048280522227287, "grad_norm": 0.052069615572690964, "global_step": 38231, "epoch": 360, "lr": 1.7964670522590193e-05} +{"train_loss": 0.012951161712408066, "grad_norm": 0.08313097059726715, "global_step": 38232, "epoch": 360, "lr": 1.796239504085826e-05} +{"train_loss": 0.016432970762252808, "grad_norm": 0.06353291869163513, "global_step": 38233, "epoch": 360, "lr": 1.796011967169254e-05} +{"train_loss": 0.015006856992840767, "grad_norm": 0.06573475897312164, "global_step": 38234, "epoch": 360, "lr": 1.795784441510104e-05} +{"train_loss": 0.014094609767198563, "grad_norm": 0.05921467766165733, "global_step": 38235, "epoch": 360, "lr": 1.795556927109174e-05} +{"train_loss": 0.017495471984148026, "grad_norm": 0.09748128801584244, "global_step": 38236, "epoch": 360, "lr": 1.7953294239672614e-05} +{"train_loss": 0.015377260744571686, "grad_norm": 0.06737998872995377, "global_step": 38237, "epoch": 360, "lr": 1.7951019320851676e-05} +{"train_loss": 0.015150999650359154, "grad_norm": 0.1205022782087326, "global_step": 38238, "epoch": 360, "lr": 1.794874451463694e-05} +{"train_loss": 0.016227925196290016, "grad_norm": 0.06435836106538773, "global_step": 38239, "epoch": 360, "lr": 1.7946469821036354e-05} +{"train_loss": 0.01471603475511074, "grad_norm": 0.08275596797466278, "global_step": 38240, "epoch": 360, "lr": 1.7944195240057953e-05} +{"train_loss": 0.0157832819968462, "grad_norm": 0.0878131315112114, "global_step": 38241, "epoch": 360, "lr": 1.79419207717097e-05} +{"train_loss": 0.01475113071501255, "grad_norm": 0.1041000634431839, "global_step": 38242, "epoch": 360, "lr": 1.79396464159996e-05} +{"train_loss": 0.015015428885817528, "grad_norm": 0.07954636216163635, "global_step": 38243, "epoch": 360, "lr": 1.793737217293563e-05} +{"train_loss": 0.01350739598274231, "grad_norm": 0.11025243997573853, "global_step": 38244, "epoch": 360, "lr": 1.793509804252579e-05} +{"train_loss": 0.01612642966210842, "grad_norm": 0.05927080661058426, "global_step": 38245, "epoch": 360, "lr": 1.793282402477809e-05} +{"train_loss": 0.013884581625461578, "grad_norm": 0.0957733690738678, "global_step": 38246, "epoch": 360, "lr": 1.7930550119700494e-05} +{"train_loss": 0.014879519119858742, "grad_norm": 0.10490667074918747, "global_step": 38247, "epoch": 360, "lr": 1.7928276327300985e-05} +{"train_loss": 0.01666341722011566, "grad_norm": 0.067041777074337, "global_step": 38248, "epoch": 360, "lr": 1.7926002647587588e-05} +{"train_loss": 0.014867028221487999, "grad_norm": 0.0911339819431305, "global_step": 38249, "epoch": 360, "lr": 1.792372908056824e-05} +{"train_loss": 0.013739548623561859, "grad_norm": 0.0771983414888382, "global_step": 38250, "epoch": 360, "lr": 1.7921455626250965e-05} +{"train_loss": 0.01603762060403824, "grad_norm": 0.07393296808004379, "global_step": 38251, "epoch": 360, "lr": 1.7919182284643765e-05} +{"train_loss": 0.013243284076452255, "grad_norm": 0.07745541632175446, "global_step": 38252, "epoch": 360, "lr": 1.7916909055754605e-05} +{"train_loss": 0.013533013872802258, "grad_norm": 0.11788710206747055, "global_step": 38253, "epoch": 360, "lr": 1.791463593959145e-05} +{"train_loss": 0.013914048671722412, "grad_norm": 0.13127020001411438, "global_step": 38254, "epoch": 360, "lr": 1.791236293616232e-05} +{"train_loss": 0.013523001223802567, "grad_norm": 0.07435585558414459, "global_step": 38255, "epoch": 360, "lr": 1.7910090045475202e-05} +{"train_loss": 0.014008982107043266, "grad_norm": 0.07667803019285202, "global_step": 38256, "epoch": 360, "lr": 1.7907817267538058e-05} +{"train_loss": 0.015391811728477478, "grad_norm": 0.0750451534986496, "global_step": 38257, "epoch": 360, "lr": 1.790554460235889e-05} +{"train_loss": 0.013482878915965557, "grad_norm": 0.058771759271621704, "global_step": 38258, "epoch": 360, "lr": 1.7903272049945696e-05} +{"train_loss": 0.014986086636781693, "grad_norm": 0.09442868828773499, "global_step": 38259, "epoch": 360, "lr": 1.7900999610306447e-05} +{"train_loss": 0.016621511429548264, "grad_norm": 0.08711059391498566, "global_step": 38260, "epoch": 360, "lr": 1.7898727283449102e-05} +{"train_loss": 0.014325235038995743, "grad_norm": 0.09967181831598282, "global_step": 38261, "epoch": 360, "lr": 1.7896455069381672e-05} +{"train_loss": 0.013093840330839157, "grad_norm": 0.07240208983421326, "global_step": 38262, "epoch": 360, "lr": 1.789418296811216e-05} +{"train_loss": 0.013467708602547646, "grad_norm": 0.09772732853889465, "global_step": 38263, "epoch": 360, "lr": 1.7891910979648506e-05} +{"train_loss": 0.014228749088943005, "grad_norm": 0.06989181041717529, "global_step": 38264, "epoch": 360, "lr": 1.7889639103998728e-05} +{"train_loss": 0.015081990738662909, "grad_norm": 0.10112296789884567, "global_step": 38265, "epoch": 360, "lr": 1.788736734117078e-05, "train_batch_mse_error": 0.013571618124842644, "val_loss": 0.019356979057192802} +{"train_loss": 0.013564715161919594, "grad_norm": 0.08235391974449158, "global_step": 38266, "epoch": 361, "lr": 1.7885095691172677e-05} +{"train_loss": 0.01557965949177742, "grad_norm": 0.10527408868074417, "global_step": 38267, "epoch": 361, "lr": 1.7882824154012355e-05} +{"train_loss": 0.013790487311780453, "grad_norm": 0.13741040229797363, "global_step": 38268, "epoch": 361, "lr": 1.7880552729697832e-05} +{"train_loss": 0.012761619873344898, "grad_norm": 0.11054020375013351, "global_step": 38269, "epoch": 361, "lr": 1.7878281418237086e-05} +{"train_loss": 0.016663655638694763, "grad_norm": 0.07493425160646439, "global_step": 38270, "epoch": 361, "lr": 1.7876010219638097e-05} +{"train_loss": 0.01250404305756092, "grad_norm": 0.10676105320453644, "global_step": 38271, "epoch": 361, "lr": 1.787373913390882e-05} +{"train_loss": 0.017647022381424904, "grad_norm": 0.07131532579660416, "global_step": 38272, "epoch": 361, "lr": 1.7871468161057263e-05} +{"train_loss": 0.016591697931289673, "grad_norm": 0.0670618936419487, "global_step": 38273, "epoch": 361, "lr": 1.7869197301091383e-05} +{"train_loss": 0.015709204599261284, "grad_norm": 0.05624900385737419, "global_step": 38274, "epoch": 361, "lr": 1.786692655401917e-05} +{"train_loss": 0.013275155797600746, "grad_norm": 0.05539838969707489, "global_step": 38275, "epoch": 361, "lr": 1.786465591984862e-05} +{"train_loss": 0.014930807054042816, "grad_norm": 0.1343056857585907, "global_step": 38276, "epoch": 361, "lr": 1.7862385398587694e-05} +{"train_loss": 0.013608266599476337, "grad_norm": 0.05517778918147087, "global_step": 38277, "epoch": 361, "lr": 1.7860114990244352e-05} +{"train_loss": 0.014515794813632965, "grad_norm": 0.085515096783638, "global_step": 38278, "epoch": 361, "lr": 1.7857844694826586e-05} +{"train_loss": 0.01676914654672146, "grad_norm": 0.0843309536576271, "global_step": 38279, "epoch": 361, "lr": 1.7855574512342397e-05} +{"train_loss": 0.014883935451507568, "grad_norm": 0.07449831068515778, "global_step": 38280, "epoch": 361, "lr": 1.7853304442799722e-05} +{"train_loss": 0.014107304625213146, "grad_norm": 0.09043364971876144, "global_step": 38281, "epoch": 361, "lr": 1.7851034486206553e-05} +{"train_loss": 0.015369396656751633, "grad_norm": 0.06997796893119812, "global_step": 38282, "epoch": 361, "lr": 1.7848764642570885e-05} +{"train_loss": 0.015202667564153671, "grad_norm": 0.09846507012844086, "global_step": 38283, "epoch": 361, "lr": 1.7846494911900675e-05} +{"train_loss": 0.01560547947883606, "grad_norm": 0.07498046010732651, "global_step": 38284, "epoch": 361, "lr": 1.784422529420388e-05} +{"train_loss": 0.014797205105423927, "grad_norm": 0.12723495066165924, "global_step": 38285, "epoch": 361, "lr": 1.78419557894885e-05} +{"train_loss": 0.01601475104689598, "grad_norm": 0.14466047286987305, "global_step": 38286, "epoch": 361, "lr": 1.783968639776251e-05} +{"train_loss": 0.012997584417462349, "grad_norm": 0.07166770845651627, "global_step": 38287, "epoch": 361, "lr": 1.7837417119033882e-05} +{"train_loss": 0.01389506459236145, "grad_norm": 0.06187577545642853, "global_step": 38288, "epoch": 361, "lr": 1.7835147953310568e-05} +{"train_loss": 0.015827052295207977, "grad_norm": 0.10481154173612595, "global_step": 38289, "epoch": 361, "lr": 1.7832878900600568e-05} +{"train_loss": 0.017437048256397247, "grad_norm": 0.10366909950971603, "global_step": 38290, "epoch": 361, "lr": 1.783060996091183e-05} +{"train_loss": 0.01789276674389839, "grad_norm": 0.17892330884933472, "global_step": 38291, "epoch": 361, "lr": 1.7828341134252336e-05} +{"train_loss": 0.012762300670146942, "grad_norm": 0.08079536259174347, "global_step": 38292, "epoch": 361, "lr": 1.7826072420630075e-05} +{"train_loss": 0.016884971410036087, "grad_norm": 0.097292460501194, "global_step": 38293, "epoch": 361, "lr": 1.782380382005301e-05} +{"train_loss": 0.016918573528528214, "grad_norm": 0.07836364954710007, "global_step": 38294, "epoch": 361, "lr": 1.7821535332529083e-05} +{"train_loss": 0.013649341650307178, "grad_norm": 0.04780304804444313, "global_step": 38295, "epoch": 361, "lr": 1.7819266958066282e-05} +{"train_loss": 0.013191545382142067, "grad_norm": 0.07864727824926376, "global_step": 38296, "epoch": 361, "lr": 1.781699869667261e-05} +{"train_loss": 0.013016083277761936, "grad_norm": 0.07285269349813461, "global_step": 38297, "epoch": 361, "lr": 1.7814730548355984e-05} +{"train_loss": 0.013686509802937508, "grad_norm": 0.07130654156208038, "global_step": 38298, "epoch": 361, "lr": 1.78124625131244e-05} +{"train_loss": 0.01785268634557724, "grad_norm": 0.13051867485046387, "global_step": 38299, "epoch": 361, "lr": 1.781019459098584e-05} +{"train_loss": 0.01266616489738226, "grad_norm": 0.06728308647871017, "global_step": 38300, "epoch": 361, "lr": 1.7807926781948257e-05} +{"train_loss": 0.014397487975656986, "grad_norm": 0.06303755193948746, "global_step": 38301, "epoch": 361, "lr": 1.7805659086019603e-05} +{"train_loss": 0.015556770376861095, "grad_norm": 0.06928147375583649, "global_step": 38302, "epoch": 361, "lr": 1.7803391503207866e-05} +{"train_loss": 0.015898939222097397, "grad_norm": 0.12063147127628326, "global_step": 38303, "epoch": 361, "lr": 1.7801124033521028e-05} +{"train_loss": 0.017038023099303246, "grad_norm": 0.11831294000148773, "global_step": 38304, "epoch": 361, "lr": 1.7798856676967013e-05} +{"train_loss": 0.016269557178020477, "grad_norm": 0.07590235024690628, "global_step": 38305, "epoch": 361, "lr": 1.7796589433553833e-05} +{"train_loss": 0.012356409803032875, "grad_norm": 0.05864328518509865, "global_step": 38306, "epoch": 361, "lr": 1.779432230328943e-05} +{"train_loss": 0.014923300594091415, "grad_norm": 0.10548117011785507, "global_step": 38307, "epoch": 361, "lr": 1.779205528618176e-05} +{"train_loss": 0.01545235887169838, "grad_norm": 0.12467268109321594, "global_step": 38308, "epoch": 361, "lr": 1.7789788382238794e-05} +{"train_loss": 0.01584945246577263, "grad_norm": 0.0731346532702446, "global_step": 38309, "epoch": 361, "lr": 1.7787521591468524e-05} +{"train_loss": 0.01664295792579651, "grad_norm": 0.09416894614696503, "global_step": 38310, "epoch": 361, "lr": 1.7785254913878878e-05} +{"train_loss": 0.014181973412632942, "grad_norm": 0.07774849236011505, "global_step": 38311, "epoch": 361, "lr": 1.7782988349477852e-05} +{"train_loss": 0.014328308403491974, "grad_norm": 0.12354309111833572, "global_step": 38312, "epoch": 361, "lr": 1.7780721898273377e-05} +{"train_loss": 0.01675301231443882, "grad_norm": 0.07846014946699142, "global_step": 38313, "epoch": 361, "lr": 1.7778455560273456e-05} +{"train_loss": 0.014568483456969261, "grad_norm": 0.0832723006606102, "global_step": 38314, "epoch": 361, "lr": 1.777618933548601e-05} +{"train_loss": 0.016007550060749054, "grad_norm": 0.1511663943529129, "global_step": 38315, "epoch": 361, "lr": 1.7773923223919013e-05} +{"train_loss": 0.014834071509540081, "grad_norm": 0.07578616589307785, "global_step": 38316, "epoch": 361, "lr": 1.7771657225580463e-05} +{"train_loss": 0.014457163400948048, "grad_norm": 0.085071861743927, "global_step": 38317, "epoch": 361, "lr": 1.7769391340478285e-05} +{"train_loss": 0.01816558465361595, "grad_norm": 0.0637655109167099, "global_step": 38318, "epoch": 361, "lr": 1.776712556862044e-05} +{"train_loss": 0.018672160804271698, "grad_norm": 0.12390195578336716, "global_step": 38319, "epoch": 361, "lr": 1.7764859910014893e-05} +{"train_loss": 0.013672974891960621, "grad_norm": 0.07503725588321686, "global_step": 38320, "epoch": 361, "lr": 1.7762594364669627e-05} +{"train_loss": 0.012903658673167229, "grad_norm": 0.04283564165234566, "global_step": 38321, "epoch": 361, "lr": 1.776032893259257e-05} +{"train_loss": 0.013757891021668911, "grad_norm": 0.07608412951231003, "global_step": 38322, "epoch": 361, "lr": 1.7758063613791692e-05} +{"train_loss": 0.015797067433595657, "grad_norm": 0.10256075859069824, "global_step": 38323, "epoch": 361, "lr": 1.7755798408274978e-05} +{"train_loss": 0.015419193543493748, "grad_norm": 0.06546418368816376, "global_step": 38324, "epoch": 361, "lr": 1.7753533316050363e-05} +{"train_loss": 0.015762848779559135, "grad_norm": 0.0655437558889389, "global_step": 38325, "epoch": 361, "lr": 1.7751268337125794e-05} +{"train_loss": 0.015403429046273232, "grad_norm": 0.07111500948667526, "global_step": 38326, "epoch": 361, "lr": 1.7749003471509246e-05} +{"train_loss": 0.01389642059803009, "grad_norm": 0.07866773754358292, "global_step": 38327, "epoch": 361, "lr": 1.7746738719208684e-05} +{"train_loss": 0.01634293794631958, "grad_norm": 0.07472401857376099, "global_step": 38328, "epoch": 361, "lr": 1.7744474080232043e-05} +{"train_loss": 0.015053991228342056, "grad_norm": 0.09923525899648666, "global_step": 38329, "epoch": 361, "lr": 1.7742209554587306e-05} +{"train_loss": 0.01571057364344597, "grad_norm": 0.16708236932754517, "global_step": 38330, "epoch": 361, "lr": 1.7739945142282422e-05} +{"train_loss": 0.015227805823087692, "grad_norm": 0.15056250989437103, "global_step": 38331, "epoch": 361, "lr": 1.773768084332532e-05} +{"train_loss": 0.017122384160757065, "grad_norm": 0.09628846496343613, "global_step": 38332, "epoch": 361, "lr": 1.773541665772398e-05} +{"train_loss": 0.015560897998511791, "grad_norm": 0.09669963270425797, "global_step": 38333, "epoch": 361, "lr": 1.7733152585486368e-05} +{"train_loss": 0.015562765300273895, "grad_norm": 0.09595425426959991, "global_step": 38334, "epoch": 361, "lr": 1.7730888626620423e-05} +{"train_loss": 0.015834419056773186, "grad_norm": 0.07598969340324402, "global_step": 38335, "epoch": 361, "lr": 1.772862478113409e-05} +{"train_loss": 0.01401706412434578, "grad_norm": 0.05618257820606232, "global_step": 38336, "epoch": 361, "lr": 1.7726361049035334e-05} +{"train_loss": 0.013641657307744026, "grad_norm": 0.06682627648115158, "global_step": 38337, "epoch": 361, "lr": 1.7724097430332125e-05} +{"train_loss": 0.016319960355758667, "grad_norm": 0.10061559826135635, "global_step": 38338, "epoch": 361, "lr": 1.7721833925032387e-05} +{"train_loss": 0.016591347754001617, "grad_norm": 0.0668846145272255, "global_step": 38339, "epoch": 361, "lr": 1.7719570533144086e-05} +{"train_loss": 0.01394724752753973, "grad_norm": 0.1095450296998024, "global_step": 38340, "epoch": 361, "lr": 1.771730725467519e-05} +{"train_loss": 0.013612041249871254, "grad_norm": 0.06116308644413948, "global_step": 38341, "epoch": 361, "lr": 1.7715044089633642e-05} +{"train_loss": 0.01599445566534996, "grad_norm": 0.08124920725822449, "global_step": 38342, "epoch": 361, "lr": 1.771278103802737e-05} +{"train_loss": 0.015904881060123444, "grad_norm": 0.08305542916059494, "global_step": 38343, "epoch": 361, "lr": 1.7710518099864343e-05} +{"train_loss": 0.0144583685323596, "grad_norm": 0.07019900530576706, "global_step": 38344, "epoch": 361, "lr": 1.770825527515253e-05} +{"train_loss": 0.012100523337721825, "grad_norm": 0.0717882439494133, "global_step": 38345, "epoch": 361, "lr": 1.770599256389985e-05} +{"train_loss": 0.013403000310063362, "grad_norm": 0.051597293466329575, "global_step": 38346, "epoch": 361, "lr": 1.770372996611428e-05} +{"train_loss": 0.014589443802833557, "grad_norm": 0.061685048043727875, "global_step": 38347, "epoch": 361, "lr": 1.770146748180376e-05} +{"train_loss": 0.01838799938559532, "grad_norm": 0.06990072131156921, "global_step": 38348, "epoch": 361, "lr": 1.769920511097622e-05} +{"train_loss": 0.014200013130903244, "grad_norm": 0.08070448786020279, "global_step": 38349, "epoch": 361, "lr": 1.7696942853639626e-05} +{"train_loss": 0.015405295416712761, "grad_norm": 0.08078727126121521, "global_step": 38350, "epoch": 361, "lr": 1.769468070980195e-05} +{"train_loss": 0.01562926545739174, "grad_norm": 0.06753566116094589, "global_step": 38351, "epoch": 361, "lr": 1.769241867947109e-05} +{"train_loss": 0.01422836259007454, "grad_norm": 0.08363477140665054, "global_step": 38352, "epoch": 361, "lr": 1.769015676265504e-05} +{"train_loss": 0.015461893752217293, "grad_norm": 0.07021503895521164, "global_step": 38353, "epoch": 361, "lr": 1.768789495936171e-05} +{"train_loss": 0.015698419883847237, "grad_norm": 0.09120848774909973, "global_step": 38354, "epoch": 361, "lr": 1.768563326959909e-05} +{"train_loss": 0.015377610921859741, "grad_norm": 0.08036693185567856, "global_step": 38355, "epoch": 361, "lr": 1.7683371693375077e-05} +{"train_loss": 0.014363136142492294, "grad_norm": 0.057164885103702545, "global_step": 38356, "epoch": 361, "lr": 1.7681110230697646e-05} +{"train_loss": 0.013998007401823997, "grad_norm": 0.07365495711565018, "global_step": 38357, "epoch": 361, "lr": 1.767884888157475e-05} +{"train_loss": 0.013443363830447197, "grad_norm": 0.061785973608493805, "global_step": 38358, "epoch": 361, "lr": 1.7676587646014326e-05} +{"train_loss": 0.017238378524780273, "grad_norm": 0.14193971455097198, "global_step": 38359, "epoch": 361, "lr": 1.7674326524024298e-05} +{"train_loss": 0.019522611051797867, "grad_norm": 0.09964871406555176, "global_step": 38360, "epoch": 361, "lr": 1.767206551561263e-05} +{"train_loss": 0.014296023175120354, "grad_norm": 0.08881495893001556, "global_step": 38361, "epoch": 361, "lr": 1.7669804620787285e-05} +{"train_loss": 0.013551084324717522, "grad_norm": 0.10802518576383591, "global_step": 38362, "epoch": 361, "lr": 1.766754383955616e-05} +{"train_loss": 0.014645926654338837, "grad_norm": 0.08262717723846436, "global_step": 38363, "epoch": 361, "lr": 1.7665283171927233e-05} +{"train_loss": 0.015098903328180313, "grad_norm": 0.04883519932627678, "global_step": 38364, "epoch": 361, "lr": 1.766302261790847e-05} +{"train_loss": 0.013884387910366058, "grad_norm": 0.07788235694169998, "global_step": 38365, "epoch": 361, "lr": 1.766076217750775e-05} +{"train_loss": 0.013124705292284489, "grad_norm": 0.12409502267837524, "global_step": 38366, "epoch": 361, "lr": 1.7658501850733044e-05} +{"train_loss": 0.013819081708788872, "grad_norm": 0.10176881402730942, "global_step": 38367, "epoch": 361, "lr": 1.7656241637592302e-05} +{"train_loss": 0.014447346329689026, "grad_norm": 0.08178632706403732, "global_step": 38368, "epoch": 361, "lr": 1.7653981538093477e-05} +{"train_loss": 0.014939520508050919, "grad_norm": 0.0701228454709053, "global_step": 38369, "epoch": 361, "lr": 1.765172155224448e-05} +{"train_loss": 0.015497034415602684, "grad_norm": 0.08595737814903259, "global_step": 38370, "epoch": 361, "lr": 1.7649461680053274e-05} +{"train_loss": 0.015062897650630406, "grad_norm": 0.07966963201761246, "global_step": 38371, "epoch": 361, "lr": 1.76472019215278e-05, "val_loss": 0.019762562587857246} +{"train_loss": 0.013492685742676258, "grad_norm": 0.12027665227651596, "global_step": 38372, "epoch": 362, "lr": 1.7644942276675963e-05} +{"train_loss": 0.018081415444612503, "grad_norm": 0.10587327927350998, "global_step": 38373, "epoch": 362, "lr": 1.7642682745505735e-05} +{"train_loss": 0.014975468628108501, "grad_norm": 0.0946013554930687, "global_step": 38374, "epoch": 362, "lr": 1.7640423328025063e-05} +{"train_loss": 0.015327130444347858, "grad_norm": 0.06394978612661362, "global_step": 38375, "epoch": 362, "lr": 1.7638164024241854e-05} +{"train_loss": 0.013526284135878086, "grad_norm": 0.10801596939563751, "global_step": 38376, "epoch": 362, "lr": 1.7635904834164073e-05} +{"train_loss": 0.014540588483214378, "grad_norm": 0.04888957738876343, "global_step": 38377, "epoch": 362, "lr": 1.763364575779964e-05} +{"train_loss": 0.014798035845160484, "grad_norm": 0.07234874367713928, "global_step": 38378, "epoch": 362, "lr": 1.7631386795156513e-05} +{"train_loss": 0.018011286854743958, "grad_norm": 0.061938732862472534, "global_step": 38379, "epoch": 362, "lr": 1.76291279462426e-05} +{"train_loss": 0.013330274261534214, "grad_norm": 0.07001247256994247, "global_step": 38380, "epoch": 362, "lr": 1.7626869211065856e-05} +{"train_loss": 0.016837745904922485, "grad_norm": 0.08173535764217377, "global_step": 38381, "epoch": 362, "lr": 1.762461058963423e-05} +{"train_loss": 0.015580646693706512, "grad_norm": 0.07369234412908554, "global_step": 38382, "epoch": 362, "lr": 1.762235208195564e-05} +{"train_loss": 0.014123908244073391, "grad_norm": 0.05534176528453827, "global_step": 38383, "epoch": 362, "lr": 1.7620093688038007e-05} +{"train_loss": 0.015513300895690918, "grad_norm": 0.06787890195846558, "global_step": 38384, "epoch": 362, "lr": 1.7617835407889283e-05} +{"train_loss": 0.014841506257653236, "grad_norm": 0.09169414639472961, "global_step": 38385, "epoch": 362, "lr": 1.7615577241517423e-05} +{"train_loss": 0.015208374708890915, "grad_norm": 0.117818683385849, "global_step": 38386, "epoch": 362, "lr": 1.761331918893032e-05} +{"train_loss": 0.012884784489870071, "grad_norm": 0.061743784695863724, "global_step": 38387, "epoch": 362, "lr": 1.7611061250135953e-05} +{"train_loss": 0.01743767410516739, "grad_norm": 0.07177279889583588, "global_step": 38388, "epoch": 362, "lr": 1.760880342514223e-05} +{"train_loss": 0.016324225813150406, "grad_norm": 0.09232467412948608, "global_step": 38389, "epoch": 362, "lr": 1.7606545713957063e-05} +{"train_loss": 0.015898186713457108, "grad_norm": 0.056231897324323654, "global_step": 38390, "epoch": 362, "lr": 1.7604288116588407e-05} +{"train_loss": 0.015867311507463455, "grad_norm": 0.1373380571603775, "global_step": 38391, "epoch": 362, "lr": 1.7602030633044216e-05} +{"train_loss": 0.014906099066138268, "grad_norm": 0.058844611048698425, "global_step": 38392, "epoch": 362, "lr": 1.7599773263332375e-05} +{"train_loss": 0.01604285091161728, "grad_norm": 0.07686708122491837, "global_step": 38393, "epoch": 362, "lr": 1.7597516007460862e-05} +{"train_loss": 0.016339074820280075, "grad_norm": 0.06827697902917862, "global_step": 38394, "epoch": 362, "lr": 1.759525886543757e-05} +{"train_loss": 0.017494335770606995, "grad_norm": 0.20149505138397217, "global_step": 38395, "epoch": 362, "lr": 1.759300183727046e-05} +{"train_loss": 0.014621665701270103, "grad_norm": 0.10420695692300797, "global_step": 38396, "epoch": 362, "lr": 1.7590744922967433e-05} +{"train_loss": 0.015433110296726227, "grad_norm": 0.1019003614783287, "global_step": 38397, "epoch": 362, "lr": 1.7588488122536435e-05} +{"train_loss": 0.01416151411831379, "grad_norm": 0.08327759802341461, "global_step": 38398, "epoch": 362, "lr": 1.758623143598541e-05} +{"train_loss": 0.013772102072834969, "grad_norm": 0.08687695860862732, "global_step": 38399, "epoch": 362, "lr": 1.7583974863322274e-05} +{"train_loss": 0.015041835606098175, "grad_norm": 0.09240174293518066, "global_step": 38400, "epoch": 362, "lr": 1.758171840455493e-05} +{"train_loss": 0.018442964181303978, "grad_norm": 0.1434507966041565, "global_step": 38401, "epoch": 362, "lr": 1.7579462059691336e-05} +{"train_loss": 0.015911074355244637, "grad_norm": 0.07950083911418915, "global_step": 38402, "epoch": 362, "lr": 1.757720582873943e-05} +{"train_loss": 0.014731119386851788, "grad_norm": 0.08226054906845093, "global_step": 38403, "epoch": 362, "lr": 1.7574949711707102e-05} +{"train_loss": 0.01577139087021351, "grad_norm": 0.07644333690404892, "global_step": 38404, "epoch": 362, "lr": 1.75726937086023e-05} +{"train_loss": 0.012159585952758789, "grad_norm": 0.08571123331785202, "global_step": 38405, "epoch": 362, "lr": 1.7570437819432983e-05} +{"train_loss": 0.014500768855214119, "grad_norm": 0.07679147273302078, "global_step": 38406, "epoch": 362, "lr": 1.756818204420701e-05} +{"train_loss": 0.014347262680530548, "grad_norm": 0.09382853657007217, "global_step": 38407, "epoch": 362, "lr": 1.7565926382932342e-05} +{"train_loss": 0.013737148605287075, "grad_norm": 0.06534110009670258, "global_step": 38408, "epoch": 362, "lr": 1.756367083561692e-05} +{"train_loss": 0.012886563315987587, "grad_norm": 0.0687030628323555, "global_step": 38409, "epoch": 362, "lr": 1.7561415402268637e-05} +{"train_loss": 0.016162408515810966, "grad_norm": 0.06431988626718521, "global_step": 38410, "epoch": 362, "lr": 1.7559160082895432e-05} +{"train_loss": 0.01609007641673088, "grad_norm": 0.08368956297636032, "global_step": 38411, "epoch": 362, "lr": 1.7556904877505248e-05} +{"train_loss": 0.015167716890573502, "grad_norm": 0.09029033035039902, "global_step": 38412, "epoch": 362, "lr": 1.7554649786105993e-05} +{"train_loss": 0.015989381819963455, "grad_norm": 0.06721566617488861, "global_step": 38413, "epoch": 362, "lr": 1.755239480870557e-05} +{"train_loss": 0.013186508789658546, "grad_norm": 0.09908567368984222, "global_step": 38414, "epoch": 362, "lr": 1.7550139945311917e-05} +{"train_loss": 0.016000164672732353, "grad_norm": 0.07417982816696167, "global_step": 38415, "epoch": 362, "lr": 1.754788519593298e-05} +{"train_loss": 0.014234384521842003, "grad_norm": 0.06940367072820663, "global_step": 38416, "epoch": 362, "lr": 1.7545630560576642e-05} +{"train_loss": 0.0148479538038373, "grad_norm": 0.07341078668832779, "global_step": 38417, "epoch": 362, "lr": 1.7543376039250863e-05} +{"train_loss": 0.016842439770698547, "grad_norm": 0.06602026522159576, "global_step": 38418, "epoch": 362, "lr": 1.754112163196353e-05} +{"train_loss": 0.016133369877934456, "grad_norm": 0.10917957127094269, "global_step": 38419, "epoch": 362, "lr": 1.75388673387226e-05} +{"train_loss": 0.01435696892440319, "grad_norm": 0.05679966136813164, "global_step": 38420, "epoch": 362, "lr": 1.7536613159535948e-05} +{"train_loss": 0.014382593333721161, "grad_norm": 0.12911485135555267, "global_step": 38421, "epoch": 362, "lr": 1.7534359094411523e-05} +{"train_loss": 0.015435479581356049, "grad_norm": 0.11057765781879425, "global_step": 38422, "epoch": 362, "lr": 1.7532105143357258e-05} +{"train_loss": 0.016592558473348618, "grad_norm": 0.08938781172037125, "global_step": 38423, "epoch": 362, "lr": 1.7529851306381055e-05} +{"train_loss": 0.01578424498438835, "grad_norm": 0.07090095430612564, "global_step": 38424, "epoch": 362, "lr": 1.7527597583490822e-05} +{"train_loss": 0.014675421640276909, "grad_norm": 0.058685265481472015, "global_step": 38425, "epoch": 362, "lr": 1.7525343974694484e-05} +{"train_loss": 0.013859458267688751, "grad_norm": 0.055992066860198975, "global_step": 38426, "epoch": 362, "lr": 1.7523090479999982e-05} +{"train_loss": 0.015224299393594265, "grad_norm": 0.08782517910003662, "global_step": 38427, "epoch": 362, "lr": 1.7520837099415198e-05} +{"train_loss": 0.017168765887618065, "grad_norm": 0.09737507253885269, "global_step": 38428, "epoch": 362, "lr": 1.7518583832948072e-05} +{"train_loss": 0.012822140008211136, "grad_norm": 0.08906930685043335, "global_step": 38429, "epoch": 362, "lr": 1.7516330680606547e-05} +{"train_loss": 0.014751874841749668, "grad_norm": 0.050849977880716324, "global_step": 38430, "epoch": 362, "lr": 1.7514077642398474e-05} +{"train_loss": 0.014501815661787987, "grad_norm": 0.12459064275026321, "global_step": 38431, "epoch": 362, "lr": 1.751182471833181e-05} +{"train_loss": 0.016341237351298332, "grad_norm": 0.2045915722846985, "global_step": 38432, "epoch": 362, "lr": 1.750957190841448e-05} +{"train_loss": 0.014186924323439598, "grad_norm": 0.06236482411623001, "global_step": 38433, "epoch": 362, "lr": 1.7507319212654367e-05} +{"train_loss": 0.014759445562958717, "grad_norm": 0.06551309674978256, "global_step": 38434, "epoch": 362, "lr": 1.7505066631059403e-05} +{"train_loss": 0.015665747225284576, "grad_norm": 0.17194341123104095, "global_step": 38435, "epoch": 362, "lr": 1.7502814163637515e-05} +{"train_loss": 0.013022147119045258, "grad_norm": 0.07740969210863113, "global_step": 38436, "epoch": 362, "lr": 1.7500561810396614e-05} +{"train_loss": 0.013056132942438126, "grad_norm": 0.06438186019659042, "global_step": 38437, "epoch": 362, "lr": 1.7498309571344583e-05} +{"train_loss": 0.01286097802221775, "grad_norm": 0.10066012293100357, "global_step": 38438, "epoch": 362, "lr": 1.749605744648936e-05} +{"train_loss": 0.017716718837618828, "grad_norm": 0.07647179812192917, "global_step": 38439, "epoch": 362, "lr": 1.7493805435838874e-05} +{"train_loss": 0.013654890470206738, "grad_norm": 0.07727499306201935, "global_step": 38440, "epoch": 362, "lr": 1.749155353940101e-05} +{"train_loss": 0.014174885116517544, "grad_norm": 0.058752551674842834, "global_step": 38441, "epoch": 362, "lr": 1.7489301757183685e-05} +{"train_loss": 0.015203547663986683, "grad_norm": 0.06255100667476654, "global_step": 38442, "epoch": 362, "lr": 1.7487050089194806e-05} +{"train_loss": 0.014353553764522076, "grad_norm": 0.06448346376419067, "global_step": 38443, "epoch": 362, "lr": 1.7484798535442317e-05} +{"train_loss": 0.014682983048260212, "grad_norm": 0.15718714892864227, "global_step": 38444, "epoch": 362, "lr": 1.7482547095934084e-05} +{"train_loss": 0.010495522990822792, "grad_norm": 0.05949036404490471, "global_step": 38445, "epoch": 362, "lr": 1.748029577067804e-05} +{"train_loss": 0.01397007703781128, "grad_norm": 0.046433527022600174, "global_step": 38446, "epoch": 362, "lr": 1.7478044559682127e-05} +{"train_loss": 0.013771257363259792, "grad_norm": 0.0652371346950531, "global_step": 38447, "epoch": 362, "lr": 1.7475793462954193e-05} +{"train_loss": 0.018437838181853294, "grad_norm": 0.08372757583856583, "global_step": 38448, "epoch": 362, "lr": 1.747354248050217e-05} +{"train_loss": 0.015267640352249146, "grad_norm": 0.08024031668901443, "global_step": 38449, "epoch": 362, "lr": 1.7471291612333994e-05} +{"train_loss": 0.015205057337880135, "grad_norm": 0.11282209306955338, "global_step": 38450, "epoch": 362, "lr": 1.7469040858457536e-05} +{"train_loss": 0.015239782631397247, "grad_norm": 0.06911773979663849, "global_step": 38451, "epoch": 362, "lr": 1.746679021888072e-05} +{"train_loss": 0.01645546220242977, "grad_norm": 0.08579887449741364, "global_step": 38452, "epoch": 362, "lr": 1.746453969361147e-05} +{"train_loss": 0.015028553083539009, "grad_norm": 0.10135076195001602, "global_step": 38453, "epoch": 362, "lr": 1.746228928265768e-05} +{"train_loss": 0.017068741843104362, "grad_norm": 0.06879551708698273, "global_step": 38454, "epoch": 362, "lr": 1.746003898602724e-05} +{"train_loss": 0.011963240802288055, "grad_norm": 0.08140243589878082, "global_step": 38455, "epoch": 362, "lr": 1.745778880372807e-05} +{"train_loss": 0.01144033670425415, "grad_norm": 0.09197837114334106, "global_step": 38456, "epoch": 362, "lr": 1.7455538735768095e-05} +{"train_loss": 0.014231041073799133, "grad_norm": 0.0830506682395935, "global_step": 38457, "epoch": 362, "lr": 1.7453288782155185e-05} +{"train_loss": 0.014399539679288864, "grad_norm": 0.0699986070394516, "global_step": 38458, "epoch": 362, "lr": 1.7451038942897287e-05} +{"train_loss": 0.019405048340559006, "grad_norm": 0.06149426847696304, "global_step": 38459, "epoch": 362, "lr": 1.744878921800226e-05} +{"train_loss": 0.016863243654370308, "grad_norm": 0.07115744799375534, "global_step": 38460, "epoch": 362, "lr": 1.7446539607478047e-05} +{"train_loss": 0.01472857128828764, "grad_norm": 0.13876797258853912, "global_step": 38461, "epoch": 362, "lr": 1.7444290111332524e-05} +{"train_loss": 0.013314765878021717, "grad_norm": 0.058238912373781204, "global_step": 38462, "epoch": 362, "lr": 1.7442040729573604e-05} +{"train_loss": 0.01407746784389019, "grad_norm": 0.11828060448169708, "global_step": 38463, "epoch": 362, "lr": 1.7439791462209214e-05} +{"train_loss": 0.014558651484549046, "grad_norm": 0.07352443784475327, "global_step": 38464, "epoch": 362, "lr": 1.743754230924724e-05} +{"train_loss": 0.01348033081740141, "grad_norm": 0.06600247323513031, "global_step": 38465, "epoch": 362, "lr": 1.7435293270695558e-05} +{"train_loss": 0.016084853559732437, "grad_norm": 0.06060877814888954, "global_step": 38466, "epoch": 362, "lr": 1.7433044346562112e-05} +{"train_loss": 0.016117265447974205, "grad_norm": 0.08799377083778381, "global_step": 38467, "epoch": 362, "lr": 1.7430795536854773e-05} +{"train_loss": 0.01733645796775818, "grad_norm": 0.08079664409160614, "global_step": 38468, "epoch": 362, "lr": 1.742854684158145e-05} +{"train_loss": 0.016092371195554733, "grad_norm": 0.08385641127824783, "global_step": 38469, "epoch": 362, "lr": 1.742629826075006e-05} +{"train_loss": 0.015273170545697212, "grad_norm": 0.08226440101861954, "global_step": 38470, "epoch": 362, "lr": 1.7424049794368503e-05} +{"train_loss": 0.016391130164265633, "grad_norm": 0.0842033326625824, "global_step": 38471, "epoch": 362, "lr": 1.7421801442444647e-05} +{"train_loss": 0.013248506933450699, "grad_norm": 0.06112465262413025, "global_step": 38472, "epoch": 362, "lr": 1.741955320498641e-05} +{"train_loss": 0.017863480374217033, "grad_norm": 0.07108493894338608, "global_step": 38473, "epoch": 362, "lr": 1.741730508200171e-05} +{"train_loss": 0.016787145286798477, "grad_norm": 0.11118202656507492, "global_step": 38474, "epoch": 362, "lr": 1.741505707349842e-05} +{"train_loss": 0.013060281053185463, "grad_norm": 0.06992067396640778, "global_step": 38475, "epoch": 362, "lr": 1.741280917948444e-05} +{"train_loss": 0.013679626397788525, "grad_norm": 0.07845429331064224, "global_step": 38476, "epoch": 362, "lr": 1.74105613999677e-05} +{"train_loss": 0.015077243910504962, "grad_norm": 0.12627576291561127, "global_step": 38477, "epoch": 362, "lr": 1.740831373495607e-05, "val_loss": 0.018799467012286186} +{"train_loss": 0.013895228505134583, "grad_norm": 0.10260573029518127, "global_step": 38478, "epoch": 363, "lr": 1.7406066184457443e-05} +{"train_loss": 0.013782179914414883, "grad_norm": 0.13972389698028564, "global_step": 38479, "epoch": 363, "lr": 1.7403818748479715e-05} +{"train_loss": 0.015749571844935417, "grad_norm": 0.05429394543170929, "global_step": 38480, "epoch": 363, "lr": 1.7401571427030817e-05} +{"train_loss": 0.017842493951320648, "grad_norm": 0.10738050937652588, "global_step": 38481, "epoch": 363, "lr": 1.73993242201186e-05} +{"train_loss": 0.015926089137792587, "grad_norm": 0.11497558653354645, "global_step": 38482, "epoch": 363, "lr": 1.7397077127750993e-05} +{"train_loss": 0.015091543085873127, "grad_norm": 0.06642801314592361, "global_step": 38483, "epoch": 363, "lr": 1.739483014993587e-05} +{"train_loss": 0.014998341910541058, "grad_norm": 0.1491893231868744, "global_step": 38484, "epoch": 363, "lr": 1.7392583286681147e-05} +{"train_loss": 0.017479656264185905, "grad_norm": 0.05582208186388016, "global_step": 38485, "epoch": 363, "lr": 1.7390336537994684e-05} +{"train_loss": 0.01723097637295723, "grad_norm": 0.1233980804681778, "global_step": 38486, "epoch": 363, "lr": 1.7388089903884403e-05} +{"train_loss": 0.01590275950729847, "grad_norm": 0.0720193088054657, "global_step": 38487, "epoch": 363, "lr": 1.7385843384358225e-05} +{"train_loss": 0.01491628959774971, "grad_norm": 0.05800962820649147, "global_step": 38488, "epoch": 363, "lr": 1.7383596979423973e-05} +{"train_loss": 0.013244548812508583, "grad_norm": 0.05565132200717926, "global_step": 38489, "epoch": 363, "lr": 1.7381350689089576e-05} +{"train_loss": 0.012033089995384216, "grad_norm": 0.10286377370357513, "global_step": 38490, "epoch": 363, "lr": 1.7379104513362948e-05} +{"train_loss": 0.014964699745178223, "grad_norm": 0.11441070586442947, "global_step": 38491, "epoch": 363, "lr": 1.7376858452251942e-05} +{"train_loss": 0.015382889658212662, "grad_norm": 0.06496606767177582, "global_step": 38492, "epoch": 363, "lr": 1.737461250576447e-05} +{"train_loss": 0.015062293037772179, "grad_norm": 0.058450594544410706, "global_step": 38493, "epoch": 363, "lr": 1.7372366673908436e-05} +{"train_loss": 0.013391433283686638, "grad_norm": 0.10038432478904724, "global_step": 38494, "epoch": 363, "lr": 1.7370120956691716e-05} +{"train_loss": 0.016430016607046127, "grad_norm": 0.1425427794456482, "global_step": 38495, "epoch": 363, "lr": 1.7367875354122183e-05} +{"train_loss": 0.01588195562362671, "grad_norm": 0.06609658896923065, "global_step": 38496, "epoch": 363, "lr": 1.7365629866207745e-05} +{"train_loss": 0.014557652175426483, "grad_norm": 0.06901966780424118, "global_step": 38497, "epoch": 363, "lr": 1.736338449295631e-05} +{"train_loss": 0.015144828706979752, "grad_norm": 0.1450841873884201, "global_step": 38498, "epoch": 363, "lr": 1.736113923437573e-05} +{"train_loss": 0.012767845764756203, "grad_norm": 0.08147253096103668, "global_step": 38499, "epoch": 363, "lr": 1.7358894090473925e-05} +{"train_loss": 0.015042904764413834, "grad_norm": 0.07531565427780151, "global_step": 38500, "epoch": 363, "lr": 1.735664906125876e-05} +{"train_loss": 0.016708500683307648, "grad_norm": 0.10954651236534119, "global_step": 38501, "epoch": 363, "lr": 1.735440414673815e-05} +{"train_loss": 0.015601648017764091, "grad_norm": 0.06993772089481354, "global_step": 38502, "epoch": 363, "lr": 1.735215934691995e-05} +{"train_loss": 0.013527976348996162, "grad_norm": 0.11526557058095932, "global_step": 38503, "epoch": 363, "lr": 1.7349914661812063e-05} +{"train_loss": 0.015817441046237946, "grad_norm": 0.050210434943437576, "global_step": 38504, "epoch": 363, "lr": 1.734767009142239e-05} +{"train_loss": 0.014303910546004772, "grad_norm": 0.07446052879095078, "global_step": 38505, "epoch": 363, "lr": 1.734542563575881e-05} +{"train_loss": 0.01450410671532154, "grad_norm": 0.11351995915174484, "global_step": 38506, "epoch": 363, "lr": 1.7343181294829185e-05} +{"train_loss": 0.016933506354689598, "grad_norm": 0.15975181758403778, "global_step": 38507, "epoch": 363, "lr": 1.7340937068641432e-05} +{"train_loss": 0.014563683420419693, "grad_norm": 0.10623132437467575, "global_step": 38508, "epoch": 363, "lr": 1.733869295720341e-05} +{"train_loss": 0.016839198768138885, "grad_norm": 0.06111327186226845, "global_step": 38509, "epoch": 363, "lr": 1.7336448960523023e-05} +{"train_loss": 0.015618150122463703, "grad_norm": 0.10705304145812988, "global_step": 38510, "epoch": 363, "lr": 1.7334205078608155e-05} +{"train_loss": 0.014518176205456257, "grad_norm": 0.10738405585289001, "global_step": 38511, "epoch": 363, "lr": 1.733196131146669e-05} +{"train_loss": 0.014695326797664165, "grad_norm": 0.1169106736779213, "global_step": 38512, "epoch": 363, "lr": 1.7329717659106488e-05} +{"train_loss": 0.014459868893027306, "grad_norm": 0.0776803269982338, "global_step": 38513, "epoch": 363, "lr": 1.7327474121535447e-05} +{"train_loss": 0.018177906051278114, "grad_norm": 0.07778161764144897, "global_step": 38514, "epoch": 363, "lr": 1.7325230698761473e-05} +{"train_loss": 0.015580535866320133, "grad_norm": 0.08391176909208298, "global_step": 38515, "epoch": 363, "lr": 1.732298739079241e-05} +{"train_loss": 0.012193437665700912, "grad_norm": 0.07072127610445023, "global_step": 38516, "epoch": 363, "lr": 1.7320744197636163e-05} +{"train_loss": 0.01491545233875513, "grad_norm": 0.07943887263536453, "global_step": 38517, "epoch": 363, "lr": 1.7318501119300623e-05} +{"train_loss": 0.016304850578308105, "grad_norm": 0.12177242338657379, "global_step": 38518, "epoch": 363, "lr": 1.7316258155793656e-05} +{"train_loss": 0.015427385456860065, "grad_norm": 0.06458044052124023, "global_step": 38519, "epoch": 363, "lr": 1.7314015307123133e-05} +{"train_loss": 0.013204652816057205, "grad_norm": 0.11920671164989471, "global_step": 38520, "epoch": 363, "lr": 1.7311772573296936e-05} +{"train_loss": 0.015410315245389938, "grad_norm": 0.13309840857982635, "global_step": 38521, "epoch": 363, "lr": 1.730952995432298e-05} +{"train_loss": 0.016473842784762383, "grad_norm": 0.06254857778549194, "global_step": 38522, "epoch": 363, "lr": 1.7307287450209098e-05} +{"train_loss": 0.015430168248713017, "grad_norm": 0.0781228244304657, "global_step": 38523, "epoch": 363, "lr": 1.7305045060963205e-05} +{"train_loss": 0.011284681968390942, "grad_norm": 0.08183291554450989, "global_step": 38524, "epoch": 363, "lr": 1.7302802786593154e-05} +{"train_loss": 0.014941874891519547, "grad_norm": 0.07904578000307083, "global_step": 38525, "epoch": 363, "lr": 1.7300560627106842e-05} +{"train_loss": 0.015064148232340813, "grad_norm": 0.0704585388302803, "global_step": 38526, "epoch": 363, "lr": 1.729831858251213e-05} +{"train_loss": 0.015235194936394691, "grad_norm": 0.04771009087562561, "global_step": 38527, "epoch": 363, "lr": 1.7296076652816905e-05} +{"train_loss": 0.014782587066292763, "grad_norm": 0.07482673972845078, "global_step": 38528, "epoch": 363, "lr": 1.729383483802906e-05} +{"train_loss": 0.014048457145690918, "grad_norm": 0.09891125559806824, "global_step": 38529, "epoch": 363, "lr": 1.7291593138156454e-05} +{"train_loss": 0.014708046801388264, "grad_norm": 0.11734560132026672, "global_step": 38530, "epoch": 363, "lr": 1.7289351553206952e-05} +{"train_loss": 0.01310622226446867, "grad_norm": 0.0671301856637001, "global_step": 38531, "epoch": 363, "lr": 1.728711008318845e-05} +{"train_loss": 0.013968175277113914, "grad_norm": 0.07371113449335098, "global_step": 38532, "epoch": 363, "lr": 1.728486872810881e-05} +{"train_loss": 0.015349701046943665, "grad_norm": 0.08856557309627533, "global_step": 38533, "epoch": 363, "lr": 1.728262748797591e-05} +{"train_loss": 0.01583130471408367, "grad_norm": 0.0704730972647667, "global_step": 38534, "epoch": 363, "lr": 1.7280386362797652e-05} +{"train_loss": 0.01203516311943531, "grad_norm": 0.06843890249729156, "global_step": 38535, "epoch": 363, "lr": 1.7278145352581875e-05} +{"train_loss": 0.014661702327430248, "grad_norm": 0.10434166342020035, "global_step": 38536, "epoch": 363, "lr": 1.7275904457336455e-05} +{"train_loss": 0.014310335740447044, "grad_norm": 0.057411838322877884, "global_step": 38537, "epoch": 363, "lr": 1.727366367706928e-05} +{"train_loss": 0.014549060724675655, "grad_norm": 0.16992302238941193, "global_step": 38538, "epoch": 363, "lr": 1.727142301178823e-05} +{"train_loss": 0.016166824847459793, "grad_norm": 0.0749269649386406, "global_step": 38539, "epoch": 363, "lr": 1.726918246150115e-05} +{"train_loss": 0.01336116436868906, "grad_norm": 0.1537204533815384, "global_step": 38540, "epoch": 363, "lr": 1.7266942026215945e-05} +{"train_loss": 0.018042873591184616, "grad_norm": 0.11585503071546555, "global_step": 38541, "epoch": 363, "lr": 1.726470170594045e-05} +{"train_loss": 0.018099186941981316, "grad_norm": 0.14808307588100433, "global_step": 38542, "epoch": 363, "lr": 1.726246150068258e-05} +{"train_loss": 0.017137717455625534, "grad_norm": 0.095354825258255, "global_step": 38543, "epoch": 363, "lr": 1.7260221410450162e-05} +{"train_loss": 0.017966188490390778, "grad_norm": 0.13618887960910797, "global_step": 38544, "epoch": 363, "lr": 1.7257981435251093e-05} +{"train_loss": 0.015832722187042236, "grad_norm": 0.11424700170755386, "global_step": 38545, "epoch": 363, "lr": 1.7255741575093247e-05} +{"train_loss": 0.013367988169193268, "grad_norm": 0.06998180598020554, "global_step": 38546, "epoch": 363, "lr": 1.725350182998449e-05} +{"train_loss": 0.01708706095814705, "grad_norm": 0.16315120458602905, "global_step": 38547, "epoch": 363, "lr": 1.7251262199932672e-05} +{"train_loss": 0.014754465781152248, "grad_norm": 0.07379937171936035, "global_step": 38548, "epoch": 363, "lr": 1.724902268494569e-05} +{"train_loss": 0.014614912681281567, "grad_norm": 0.08820480108261108, "global_step": 38549, "epoch": 363, "lr": 1.724678328503138e-05} +{"train_loss": 0.014533699490129948, "grad_norm": 0.07024811208248138, "global_step": 38550, "epoch": 363, "lr": 1.724454400019763e-05} +{"train_loss": 0.014525825157761574, "grad_norm": 0.06265712529420853, "global_step": 38551, "epoch": 363, "lr": 1.7242304830452332e-05} +{"train_loss": 0.014976579695940018, "grad_norm": 0.10519152134656906, "global_step": 38552, "epoch": 363, "lr": 1.724006577580332e-05} +{"train_loss": 0.015887686982750893, "grad_norm": 0.10393010079860687, "global_step": 38553, "epoch": 363, "lr": 1.723782683625846e-05} +{"train_loss": 0.015216152183711529, "grad_norm": 0.09203743189573288, "global_step": 38554, "epoch": 363, "lr": 1.7235588011825625e-05} +{"train_loss": 0.015205573290586472, "grad_norm": 0.05758421868085861, "global_step": 38555, "epoch": 363, "lr": 1.72333493025127e-05} +{"train_loss": 0.014063827693462372, "grad_norm": 0.07163779437541962, "global_step": 38556, "epoch": 363, "lr": 1.7231110708327523e-05} +{"train_loss": 0.014176581054925919, "grad_norm": 0.11926772445440292, "global_step": 38557, "epoch": 363, "lr": 1.7228872229277972e-05} +{"train_loss": 0.015686821192502975, "grad_norm": 0.1048443540930748, "global_step": 38558, "epoch": 363, "lr": 1.7226633865371928e-05} +{"train_loss": 0.01547978911548853, "grad_norm": 0.0683324933052063, "global_step": 38559, "epoch": 363, "lr": 1.7224395616617235e-05} +{"train_loss": 0.01552589237689972, "grad_norm": 0.09301891177892685, "global_step": 38560, "epoch": 363, "lr": 1.722215748302175e-05} +{"train_loss": 0.017268827185034752, "grad_norm": 0.07860911637544632, "global_step": 38561, "epoch": 363, "lr": 1.721991946459335e-05} +{"train_loss": 0.013142403215169907, "grad_norm": 0.08781474828720093, "global_step": 38562, "epoch": 363, "lr": 1.721768156133991e-05} +{"train_loss": 0.017792172729969025, "grad_norm": 0.11281326413154602, "global_step": 38563, "epoch": 363, "lr": 1.7215443773269268e-05} +{"train_loss": 0.014759048819541931, "grad_norm": 0.07575535029172897, "global_step": 38564, "epoch": 363, "lr": 1.721320610038931e-05} +{"train_loss": 0.014769243076443672, "grad_norm": 0.051345206797122955, "global_step": 38565, "epoch": 363, "lr": 1.7210968542707895e-05} +{"train_loss": 0.014214425347745419, "grad_norm": 0.07913164794445038, "global_step": 38566, "epoch": 363, "lr": 1.7208731100232856e-05} +{"train_loss": 0.014867079444229603, "grad_norm": 0.05524902790784836, "global_step": 38567, "epoch": 363, "lr": 1.7206493772972072e-05} +{"train_loss": 0.013499287888407707, "grad_norm": 0.09318496286869049, "global_step": 38568, "epoch": 363, "lr": 1.720425656093343e-05} +{"train_loss": 0.015144720673561096, "grad_norm": 0.07536163926124573, "global_step": 38569, "epoch": 363, "lr": 1.7202019464124745e-05} +{"train_loss": 0.015467383898794651, "grad_norm": 0.07549353688955307, "global_step": 38570, "epoch": 363, "lr": 1.7199782482553923e-05} +{"train_loss": 0.012959436513483524, "grad_norm": 0.069590263068676, "global_step": 38571, "epoch": 363, "lr": 1.7197545616228784e-05} +{"train_loss": 0.014250274747610092, "grad_norm": 0.04947122558951378, "global_step": 38572, "epoch": 363, "lr": 1.7195308865157216e-05} +{"train_loss": 0.01391780748963356, "grad_norm": 0.09891431778669357, "global_step": 38573, "epoch": 363, "lr": 1.7193072229347052e-05} +{"train_loss": 0.015413539484143257, "grad_norm": 0.07153613865375519, "global_step": 38574, "epoch": 363, "lr": 1.7190835708806163e-05} +{"train_loss": 0.013770963996648788, "grad_norm": 0.0645742416381836, "global_step": 38575, "epoch": 363, "lr": 1.718859930354243e-05} +{"train_loss": 0.014946448616683483, "grad_norm": 0.06063053384423256, "global_step": 38576, "epoch": 363, "lr": 1.718636301356369e-05} +{"train_loss": 0.015897242352366447, "grad_norm": 0.06890472024679184, "global_step": 38577, "epoch": 363, "lr": 1.718412683887778e-05} +{"train_loss": 0.014513285830616951, "grad_norm": 0.05756848677992821, "global_step": 38578, "epoch": 363, "lr": 1.7181890779492578e-05} +{"train_loss": 0.015281984582543373, "grad_norm": 0.07094631344079971, "global_step": 38579, "epoch": 363, "lr": 1.7179654835415953e-05} +{"train_loss": 0.01521213073283434, "grad_norm": 0.09636008739471436, "global_step": 38580, "epoch": 363, "lr": 1.717741900665574e-05} +{"train_loss": 0.014951469376683235, "grad_norm": 0.18412697315216064, "global_step": 38581, "epoch": 363, "lr": 1.7175183293219792e-05} +{"train_loss": 0.017109621316194534, "grad_norm": 0.07410617917776108, "global_step": 38582, "epoch": 363, "lr": 1.7172947695115997e-05} +{"train_loss": 0.01506753536468407, "grad_norm": 0.10281773656606674, "global_step": 38583, "epoch": 363, "lr": 1.7170712212352187e-05, "val_loss": 0.020232554525136948} +{"train_loss": 0.016461867839097977, "grad_norm": 0.11225694417953491, "global_step": 38584, "epoch": 364, "lr": 1.71684768449362e-05} +{"train_loss": 0.01656484790146351, "grad_norm": 0.07164766639471054, "global_step": 38585, "epoch": 364, "lr": 1.7166241592875908e-05} +{"train_loss": 0.01381767075508833, "grad_norm": 0.09040141850709915, "global_step": 38586, "epoch": 364, "lr": 1.7164006456179182e-05} +{"train_loss": 0.01352005172520876, "grad_norm": 0.05021357908844948, "global_step": 38587, "epoch": 364, "lr": 1.7161771434853857e-05} +{"train_loss": 0.014013893902301788, "grad_norm": 0.09842666983604431, "global_step": 38588, "epoch": 364, "lr": 1.7159536528907765e-05} +{"train_loss": 0.01322015468031168, "grad_norm": 0.06408409029245377, "global_step": 38589, "epoch": 364, "lr": 1.7157301738348803e-05} +{"train_loss": 0.014619026333093643, "grad_norm": 0.06048217788338661, "global_step": 38590, "epoch": 364, "lr": 1.7155067063184776e-05} +{"train_loss": 0.016830913722515106, "grad_norm": 0.08575199544429779, "global_step": 38591, "epoch": 364, "lr": 1.7152832503423567e-05} +{"train_loss": 0.015122316777706146, "grad_norm": 0.05795784667134285, "global_step": 38592, "epoch": 364, "lr": 1.7150598059073032e-05} +{"train_loss": 0.013150546699762344, "grad_norm": 0.08982126414775848, "global_step": 38593, "epoch": 364, "lr": 1.714836373014101e-05} +{"train_loss": 0.014299698173999786, "grad_norm": 0.07518377155065536, "global_step": 38594, "epoch": 364, "lr": 1.7146129516635333e-05} +{"train_loss": 0.01534931268543005, "grad_norm": 0.06479115039110184, "global_step": 38595, "epoch": 364, "lr": 1.7143895418563872e-05} +{"train_loss": 0.017629697918891907, "grad_norm": 0.12333423644304276, "global_step": 38596, "epoch": 364, "lr": 1.7141661435934487e-05} +{"train_loss": 0.013697296380996704, "grad_norm": 0.08638772368431091, "global_step": 38597, "epoch": 364, "lr": 1.7139427568755002e-05} +{"train_loss": 0.0153591837733984, "grad_norm": 0.07154518365859985, "global_step": 38598, "epoch": 364, "lr": 1.7137193817033276e-05} +{"train_loss": 0.016065463423728943, "grad_norm": 0.07033023238182068, "global_step": 38599, "epoch": 364, "lr": 1.713496018077717e-05} +{"train_loss": 0.01393598411232233, "grad_norm": 0.14124764502048492, "global_step": 38600, "epoch": 364, "lr": 1.713272665999453e-05} +{"train_loss": 0.016047436743974686, "grad_norm": 0.09924178570508957, "global_step": 38601, "epoch": 364, "lr": 1.7130493254693174e-05} +{"train_loss": 0.0178692564368248, "grad_norm": 0.1088486984372139, "global_step": 38602, "epoch": 364, "lr": 1.712825996488097e-05} +{"train_loss": 0.014800368808209896, "grad_norm": 0.12975598871707916, "global_step": 38603, "epoch": 364, "lr": 1.7126026790565787e-05} +{"train_loss": 0.015648145228624344, "grad_norm": 0.0943186804652214, "global_step": 38604, "epoch": 364, "lr": 1.7123793731755423e-05} +{"train_loss": 0.015073220245540142, "grad_norm": 0.10790830850601196, "global_step": 38605, "epoch": 364, "lr": 1.7121560788457776e-05} +{"train_loss": 0.017311103641986847, "grad_norm": 0.08594619482755661, "global_step": 38606, "epoch": 364, "lr": 1.711932796068066e-05} +{"train_loss": 0.015381455421447754, "grad_norm": 0.06399717181921005, "global_step": 38607, "epoch": 364, "lr": 1.7117095248431913e-05} +{"train_loss": 0.015145380049943924, "grad_norm": 0.06339910626411438, "global_step": 38608, "epoch": 364, "lr": 1.7114862651719388e-05} +{"train_loss": 0.014988437294960022, "grad_norm": 0.07298540323972702, "global_step": 38609, "epoch": 364, "lr": 1.711263017055096e-05} +{"train_loss": 0.014845160767436028, "grad_norm": 0.076522096991539, "global_step": 38610, "epoch": 364, "lr": 1.711039780493442e-05} +{"train_loss": 0.016414344310760498, "grad_norm": 0.06653129309415817, "global_step": 38611, "epoch": 364, "lr": 1.7108165554877664e-05} +{"train_loss": 0.01369420811533928, "grad_norm": 0.13134290277957916, "global_step": 38612, "epoch": 364, "lr": 1.7105933420388488e-05} +{"train_loss": 0.013178211636841297, "grad_norm": 0.10951581597328186, "global_step": 38613, "epoch": 364, "lr": 1.7103701401474775e-05} +{"train_loss": 0.01423264853656292, "grad_norm": 0.10352744162082672, "global_step": 38614, "epoch": 364, "lr": 1.7101469498144336e-05} +{"train_loss": 0.016250938177108765, "grad_norm": 0.07533524930477142, "global_step": 38615, "epoch": 364, "lr": 1.709923771040502e-05} +{"train_loss": 0.01559748686850071, "grad_norm": 0.07428697496652603, "global_step": 38616, "epoch": 364, "lr": 1.709700603826469e-05} +{"train_loss": 0.01590241678059101, "grad_norm": 0.09862295538187027, "global_step": 38617, "epoch": 364, "lr": 1.709477448173118e-05} +{"train_loss": 0.013925165869295597, "grad_norm": 0.2026231437921524, "global_step": 38618, "epoch": 364, "lr": 1.7092543040812297e-05} +{"train_loss": 0.01575198769569397, "grad_norm": 0.05757866054773331, "global_step": 38619, "epoch": 364, "lr": 1.709031171551591e-05} +{"train_loss": 0.016019470989704132, "grad_norm": 0.09215563535690308, "global_step": 38620, "epoch": 364, "lr": 1.7088080505849874e-05} +{"train_loss": 0.01560719683766365, "grad_norm": 0.13137024641036987, "global_step": 38621, "epoch": 364, "lr": 1.7085849411821997e-05} +{"train_loss": 0.014098554849624634, "grad_norm": 0.08797960728406906, "global_step": 38622, "epoch": 364, "lr": 1.7083618433440125e-05} +{"train_loss": 0.01580175757408142, "grad_norm": 0.07450713962316513, "global_step": 38623, "epoch": 364, "lr": 1.7081387570712138e-05} +{"train_loss": 0.01694820076227188, "grad_norm": 0.13333366811275482, "global_step": 38624, "epoch": 364, "lr": 1.7079156823645802e-05} +{"train_loss": 0.015164289623498917, "grad_norm": 0.054923709481954575, "global_step": 38625, "epoch": 364, "lr": 1.7076926192248998e-05} +{"train_loss": 0.015538704581558704, "grad_norm": 0.18234312534332275, "global_step": 38626, "epoch": 364, "lr": 1.7074695676529573e-05} +{"train_loss": 0.014840800315141678, "grad_norm": 0.05566370114684105, "global_step": 38627, "epoch": 364, "lr": 1.7072465276495332e-05} +{"train_loss": 0.013611904345452785, "grad_norm": 0.06595908850431442, "global_step": 38628, "epoch": 364, "lr": 1.707023499215412e-05} +{"train_loss": 0.015043222345411777, "grad_norm": 0.0608169287443161, "global_step": 38629, "epoch": 364, "lr": 1.706800482351381e-05} +{"train_loss": 0.014838207513093948, "grad_norm": 0.0916067361831665, "global_step": 38630, "epoch": 364, "lr": 1.7065774770582204e-05} +{"train_loss": 0.014292877167463303, "grad_norm": 0.0910630077123642, "global_step": 38631, "epoch": 364, "lr": 1.7063544833367125e-05} +{"train_loss": 0.011939983814954758, "grad_norm": 0.08735306560993195, "global_step": 38632, "epoch": 364, "lr": 1.706131501187643e-05} +{"train_loss": 0.014275025576353073, "grad_norm": 0.07469739019870758, "global_step": 38633, "epoch": 364, "lr": 1.705908530611796e-05} +{"train_loss": 0.015210765413939953, "grad_norm": 0.05870625749230385, "global_step": 38634, "epoch": 364, "lr": 1.705685571609953e-05} +{"train_loss": 0.01648821495473385, "grad_norm": 0.05753720924258232, "global_step": 38635, "epoch": 364, "lr": 1.7054626241828993e-05} +{"train_loss": 0.015387043356895447, "grad_norm": 0.06639017164707184, "global_step": 38636, "epoch": 364, "lr": 1.7052396883314152e-05} +{"train_loss": 0.014918617904186249, "grad_norm": 0.10291461646556854, "global_step": 38637, "epoch": 364, "lr": 1.705016764056288e-05} +{"train_loss": 0.014478585682809353, "grad_norm": 0.10745774954557419, "global_step": 38638, "epoch": 364, "lr": 1.7047938513582978e-05} +{"train_loss": 0.01628493145108223, "grad_norm": 0.06274566054344177, "global_step": 38639, "epoch": 364, "lr": 1.7045709502382285e-05} +{"train_loss": 0.016708195209503174, "grad_norm": 0.12992653250694275, "global_step": 38640, "epoch": 364, "lr": 1.7043480606968654e-05} +{"train_loss": 0.015944384038448334, "grad_norm": 0.10078737139701843, "global_step": 38641, "epoch": 364, "lr": 1.7041251827349895e-05} +{"train_loss": 0.014465346932411194, "grad_norm": 0.05737532675266266, "global_step": 38642, "epoch": 364, "lr": 1.7039023163533836e-05} +{"train_loss": 0.015144926495850086, "grad_norm": 0.07120615988969803, "global_step": 38643, "epoch": 364, "lr": 1.7036794615528312e-05} +{"train_loss": 0.016681773588061333, "grad_norm": 0.10183000564575195, "global_step": 38644, "epoch": 364, "lr": 1.703456618334117e-05} +{"train_loss": 0.015199769288301468, "grad_norm": 0.0582922026515007, "global_step": 38645, "epoch": 364, "lr": 1.7032337866980215e-05} +{"train_loss": 0.018350550904870033, "grad_norm": 0.2322530448436737, "global_step": 38646, "epoch": 364, "lr": 1.7030109666453298e-05} +{"train_loss": 0.014764409512281418, "grad_norm": 0.06055357679724693, "global_step": 38647, "epoch": 364, "lr": 1.702788158176824e-05} +{"train_loss": 0.017235558480024338, "grad_norm": 0.1056063324213028, "global_step": 38648, "epoch": 364, "lr": 1.702565361293285e-05} +{"train_loss": 0.015370378270745277, "grad_norm": 0.10187902301549911, "global_step": 38649, "epoch": 364, "lr": 1.7023425759954974e-05} +{"train_loss": 0.014830770902335644, "grad_norm": 0.08236467838287354, "global_step": 38650, "epoch": 364, "lr": 1.7021198022842456e-05} +{"train_loss": 0.012778053060173988, "grad_norm": 0.10984643548727036, "global_step": 38651, "epoch": 364, "lr": 1.701897040160309e-05} +{"train_loss": 0.017149053514003754, "grad_norm": 0.09387952834367752, "global_step": 38652, "epoch": 364, "lr": 1.7016742896244736e-05} +{"train_loss": 0.012834828346967697, "grad_norm": 0.09049705415964127, "global_step": 38653, "epoch": 364, "lr": 1.701451550677518e-05} +{"train_loss": 0.0173180028796196, "grad_norm": 0.09268349409103394, "global_step": 38654, "epoch": 364, "lr": 1.7012288233202296e-05} +{"train_loss": 0.015705477446317673, "grad_norm": 0.09893345087766647, "global_step": 38655, "epoch": 364, "lr": 1.7010061075533867e-05} +{"train_loss": 0.014560081996023655, "grad_norm": 0.10210294276475906, "global_step": 38656, "epoch": 364, "lr": 1.7007834033777732e-05} +{"train_loss": 0.013136876747012138, "grad_norm": 0.1155601441860199, "global_step": 38657, "epoch": 364, "lr": 1.700560710794174e-05} +{"train_loss": 0.012976529076695442, "grad_norm": 0.09101881086826324, "global_step": 38658, "epoch": 364, "lr": 1.7003380298033695e-05} +{"train_loss": 0.01515754871070385, "grad_norm": 0.09719371795654297, "global_step": 38659, "epoch": 364, "lr": 1.7001153604061408e-05} +{"train_loss": 0.012976611964404583, "grad_norm": 0.060957446694374084, "global_step": 38660, "epoch": 364, "lr": 1.6998927026032712e-05} +{"train_loss": 0.016140494495630264, "grad_norm": 0.06791607290506363, "global_step": 38661, "epoch": 364, "lr": 1.6996700563955454e-05} +{"train_loss": 0.014514220878481865, "grad_norm": 0.160020649433136, "global_step": 38662, "epoch": 364, "lr": 1.699447421783742e-05} +{"train_loss": 0.014784939587116241, "grad_norm": 0.17479462921619415, "global_step": 38663, "epoch": 364, "lr": 1.699224798768645e-05} +{"train_loss": 0.014409517869353294, "grad_norm": 0.12226659059524536, "global_step": 38664, "epoch": 364, "lr": 1.6990021873510403e-05} +{"train_loss": 0.013829883188009262, "grad_norm": 0.05672186240553856, "global_step": 38665, "epoch": 364, "lr": 1.6987795875317024e-05} +{"train_loss": 0.015317702665925026, "grad_norm": 0.07984146475791931, "global_step": 38666, "epoch": 364, "lr": 1.6985569993114174e-05} +{"train_loss": 0.016676776111125946, "grad_norm": 0.09594765305519104, "global_step": 38667, "epoch": 364, "lr": 1.69833442269097e-05} +{"train_loss": 0.012424285523593426, "grad_norm": 0.12111175805330276, "global_step": 38668, "epoch": 364, "lr": 1.698111857671137e-05} +{"train_loss": 0.01548561081290245, "grad_norm": 0.0993620753288269, "global_step": 38669, "epoch": 364, "lr": 1.697889304252703e-05} +{"train_loss": 0.01509109977632761, "grad_norm": 0.10799281299114227, "global_step": 38670, "epoch": 364, "lr": 1.697666762436452e-05} +{"train_loss": 0.018212290480732918, "grad_norm": 0.08298860490322113, "global_step": 38671, "epoch": 364, "lr": 1.697444232223163e-05} +{"train_loss": 0.015066143125295639, "grad_norm": 0.06688541173934937, "global_step": 38672, "epoch": 364, "lr": 1.6972217136136175e-05} +{"train_loss": 0.015019647777080536, "grad_norm": 0.05185004323720932, "global_step": 38673, "epoch": 364, "lr": 1.6969992066085988e-05} +{"train_loss": 0.01467239111661911, "grad_norm": 0.10497591644525528, "global_step": 38674, "epoch": 364, "lr": 1.69677671120889e-05} +{"train_loss": 0.015767168253660202, "grad_norm": 0.07070077955722809, "global_step": 38675, "epoch": 364, "lr": 1.6965542274152697e-05} +{"train_loss": 0.015599595382809639, "grad_norm": 0.05653697997331619, "global_step": 38676, "epoch": 364, "lr": 1.6963317552285228e-05} +{"train_loss": 0.014534546993672848, "grad_norm": 0.07480408251285553, "global_step": 38677, "epoch": 364, "lr": 1.6961092946494277e-05} +{"train_loss": 0.01560347992926836, "grad_norm": 0.11132784187793732, "global_step": 38678, "epoch": 364, "lr": 1.6958868456787698e-05} +{"train_loss": 0.015668507665395737, "grad_norm": 0.09659070521593094, "global_step": 38679, "epoch": 364, "lr": 1.6956644083173262e-05} +{"train_loss": 0.015585800632834435, "grad_norm": 0.15937428176403046, "global_step": 38680, "epoch": 364, "lr": 1.6954419825658813e-05} +{"train_loss": 0.012381453067064285, "grad_norm": 0.09917592257261276, "global_step": 38681, "epoch": 364, "lr": 1.6952195684252175e-05} +{"train_loss": 0.014894741587340832, "grad_norm": 0.0644557848572731, "global_step": 38682, "epoch": 364, "lr": 1.6949971658961156e-05} +{"train_loss": 0.016050532460212708, "grad_norm": 0.10857972502708435, "global_step": 38683, "epoch": 364, "lr": 1.694774774979354e-05} +{"train_loss": 0.016818366944789886, "grad_norm": 0.06823206692934036, "global_step": 38684, "epoch": 364, "lr": 1.694552395675717e-05} +{"train_loss": 0.013477546162903309, "grad_norm": 0.0706000030040741, "global_step": 38685, "epoch": 364, "lr": 1.6943300279859863e-05} +{"train_loss": 0.01308707520365715, "grad_norm": 0.1123579740524292, "global_step": 38686, "epoch": 364, "lr": 1.694107671910941e-05} +{"train_loss": 0.012849625200033188, "grad_norm": 0.07532492280006409, "global_step": 38687, "epoch": 364, "lr": 1.6938853274513654e-05} +{"train_loss": 0.012938378378748894, "grad_norm": 0.05653874948620796, "global_step": 38688, "epoch": 364, "lr": 1.693662994608039e-05} +{"train_loss": 0.015076170420660725, "grad_norm": 0.12319831550121307, "global_step": 38689, "epoch": 364, "lr": 1.6934406733817414e-05, "val_loss": 0.019565967842936516} +{"train_loss": 0.015241325832903385, "grad_norm": 0.12756593525409698, "global_step": 38690, "epoch": 365, "lr": 1.6932183637732553e-05} +{"train_loss": 0.015732519328594208, "grad_norm": 0.08805302530527115, "global_step": 38691, "epoch": 365, "lr": 1.6929960657833633e-05} +{"train_loss": 0.014932607300579548, "grad_norm": 0.06070621684193611, "global_step": 38692, "epoch": 365, "lr": 1.692773779412843e-05} +{"train_loss": 0.017255663871765137, "grad_norm": 0.07505792379379272, "global_step": 38693, "epoch": 365, "lr": 1.692551504662479e-05} +{"train_loss": 0.014376714825630188, "grad_norm": 0.06994914263486862, "global_step": 38694, "epoch": 365, "lr": 1.6923292415330492e-05} +{"train_loss": 0.016506187617778778, "grad_norm": 0.09292203933000565, "global_step": 38695, "epoch": 365, "lr": 1.6921069900253377e-05} +{"train_loss": 0.01384031306952238, "grad_norm": 0.06422203779220581, "global_step": 38696, "epoch": 365, "lr": 1.6918847501401224e-05} +{"train_loss": 0.013612281531095505, "grad_norm": 0.09309641271829605, "global_step": 38697, "epoch": 365, "lr": 1.691662521878185e-05} +{"train_loss": 0.016083240509033203, "grad_norm": 0.07010718435049057, "global_step": 38698, "epoch": 365, "lr": 1.6914403052403082e-05} +{"train_loss": 0.012554895132780075, "grad_norm": 0.0716533437371254, "global_step": 38699, "epoch": 365, "lr": 1.6912181002272713e-05} +{"train_loss": 0.01657136157155037, "grad_norm": 0.11623216420412064, "global_step": 38700, "epoch": 365, "lr": 1.690995906839854e-05} +{"train_loss": 0.015845952555537224, "grad_norm": 0.09505698829889297, "global_step": 38701, "epoch": 365, "lr": 1.6907737250788375e-05} +{"train_loss": 0.012781248427927494, "grad_norm": 0.09431752562522888, "global_step": 38702, "epoch": 365, "lr": 1.6905515549450053e-05} +{"train_loss": 0.014566682279109955, "grad_norm": 0.054696209728717804, "global_step": 38703, "epoch": 365, "lr": 1.6903293964391333e-05} +{"train_loss": 0.016735831275582314, "grad_norm": 0.14674857258796692, "global_step": 38704, "epoch": 365, "lr": 1.6901072495620045e-05} +{"train_loss": 0.016431313008069992, "grad_norm": 0.06635510921478271, "global_step": 38705, "epoch": 365, "lr": 1.689885114314403e-05} +{"train_loss": 0.014413191005587578, "grad_norm": 0.07341113686561584, "global_step": 38706, "epoch": 365, "lr": 1.6896629906971024e-05} +{"train_loss": 0.013859626837074757, "grad_norm": 0.12369793653488159, "global_step": 38707, "epoch": 365, "lr": 1.6894408787108863e-05} +{"train_loss": 0.014663530513644218, "grad_norm": 0.06964164972305298, "global_step": 38708, "epoch": 365, "lr": 1.6892187783565367e-05} +{"train_loss": 0.016262365505099297, "grad_norm": 0.16679127514362335, "global_step": 38709, "epoch": 365, "lr": 1.688996689634831e-05} +{"train_loss": 0.01741069182753563, "grad_norm": 0.11494404077529907, "global_step": 38710, "epoch": 365, "lr": 1.688774612546551e-05} +{"train_loss": 0.015965769067406654, "grad_norm": 0.09384483098983765, "global_step": 38711, "epoch": 365, "lr": 1.688552547092479e-05} +{"train_loss": 0.016229478642344475, "grad_norm": 0.08032340556383133, "global_step": 38712, "epoch": 365, "lr": 1.6883304932733927e-05} +{"train_loss": 0.013449421152472496, "grad_norm": 0.0486149862408638, "global_step": 38713, "epoch": 365, "lr": 1.6881084510900712e-05} +{"train_loss": 0.014460098929703236, "grad_norm": 0.07304783165454865, "global_step": 38714, "epoch": 365, "lr": 1.6878864205432964e-05} +{"train_loss": 0.01499769277870655, "grad_norm": 0.10420665144920349, "global_step": 38715, "epoch": 365, "lr": 1.6876644016338504e-05} +{"train_loss": 0.012675665318965912, "grad_norm": 0.058781906962394714, "global_step": 38716, "epoch": 365, "lr": 1.687442394362509e-05} +{"train_loss": 0.016601093113422394, "grad_norm": 0.0771292895078659, "global_step": 38717, "epoch": 365, "lr": 1.6872203987300566e-05} +{"train_loss": 0.01486880611628294, "grad_norm": 0.0669223889708519, "global_step": 38718, "epoch": 365, "lr": 1.6869984147372693e-05} +{"train_loss": 0.014480242505669594, "grad_norm": 0.06475654989480972, "global_step": 38719, "epoch": 365, "lr": 1.6867764423849296e-05} +{"train_loss": 0.013827105984091759, "grad_norm": 0.13929453492164612, "global_step": 38720, "epoch": 365, "lr": 1.686554481673816e-05} +{"train_loss": 0.01754201389849186, "grad_norm": 0.13965994119644165, "global_step": 38721, "epoch": 365, "lr": 1.6863325326047085e-05} +{"train_loss": 0.015331760048866272, "grad_norm": 0.1060900166630745, "global_step": 38722, "epoch": 365, "lr": 1.686110595178389e-05} +{"train_loss": 0.01492554135620594, "grad_norm": 0.1420503556728363, "global_step": 38723, "epoch": 365, "lr": 1.6858886693956354e-05} +{"train_loss": 0.016648707911372185, "grad_norm": 0.09315766394138336, "global_step": 38724, "epoch": 365, "lr": 1.685666755257227e-05} +{"train_loss": 0.014361757785081863, "grad_norm": 0.07870347052812576, "global_step": 38725, "epoch": 365, "lr": 1.6854448527639456e-05} +{"train_loss": 0.015745971351861954, "grad_norm": 0.05840019881725311, "global_step": 38726, "epoch": 365, "lr": 1.6852229619165672e-05} +{"train_loss": 0.016790762543678284, "grad_norm": 0.09700565040111542, "global_step": 38727, "epoch": 365, "lr": 1.6850010827158742e-05} +{"train_loss": 0.015333358198404312, "grad_norm": 0.06170887127518654, "global_step": 38728, "epoch": 365, "lr": 1.6847792151626473e-05} +{"train_loss": 0.014381255954504013, "grad_norm": 0.0663275197148323, "global_step": 38729, "epoch": 365, "lr": 1.6845573592576647e-05} +{"train_loss": 0.015130162239074707, "grad_norm": 0.11296495795249939, "global_step": 38730, "epoch": 365, "lr": 1.6843355150017042e-05} +{"train_loss": 0.013692133128643036, "grad_norm": 0.12118028104305267, "global_step": 38731, "epoch": 365, "lr": 1.6841136823955462e-05} +{"train_loss": 0.016489354893565178, "grad_norm": 0.09410125762224197, "global_step": 38732, "epoch": 365, "lr": 1.683891861439973e-05} +{"train_loss": 0.013590453192591667, "grad_norm": 0.07412403076887131, "global_step": 38733, "epoch": 365, "lr": 1.6836700521357596e-05} +{"train_loss": 0.015176604501903057, "grad_norm": 0.19928811490535736, "global_step": 38734, "epoch": 365, "lr": 1.683448254483687e-05} +{"train_loss": 0.01508487481623888, "grad_norm": 0.06716717779636383, "global_step": 38735, "epoch": 365, "lr": 1.6832264684845372e-05} +{"train_loss": 0.013992192223668098, "grad_norm": 0.10649987310171127, "global_step": 38736, "epoch": 365, "lr": 1.683004694139087e-05} +{"train_loss": 0.014947950839996338, "grad_norm": 0.0811716690659523, "global_step": 38737, "epoch": 365, "lr": 1.6827829314481142e-05} +{"train_loss": 0.012416403740644455, "grad_norm": 0.05846409872174263, "global_step": 38738, "epoch": 365, "lr": 1.6825611804123993e-05} +{"train_loss": 0.01574721187353134, "grad_norm": 0.14718537032604218, "global_step": 38739, "epoch": 365, "lr": 1.6823394410327238e-05} +{"train_loss": 0.014679914340376854, "grad_norm": 0.0688343420624733, "global_step": 38740, "epoch": 365, "lr": 1.6821177133098642e-05} +{"train_loss": 0.01652393490076065, "grad_norm": 0.16509495675563812, "global_step": 38741, "epoch": 365, "lr": 1.6818959972445984e-05} +{"train_loss": 0.013971557840704918, "grad_norm": 0.0562058761715889, "global_step": 38742, "epoch": 365, "lr": 1.681674292837707e-05} +{"train_loss": 0.01550282072275877, "grad_norm": 0.06957361847162247, "global_step": 38743, "epoch": 365, "lr": 1.681452600089971e-05} +{"train_loss": 0.01335345208644867, "grad_norm": 0.08952594548463821, "global_step": 38744, "epoch": 365, "lr": 1.6812309190021657e-05} +{"train_loss": 0.015057229436933994, "grad_norm": 0.09774334728717804, "global_step": 38745, "epoch": 365, "lr": 1.6810092495750717e-05} +{"train_loss": 0.014765297994017601, "grad_norm": 0.06646724790334702, "global_step": 38746, "epoch": 365, "lr": 1.6807875918094712e-05} +{"train_loss": 0.013353342190384865, "grad_norm": 0.08045739680528641, "global_step": 38747, "epoch": 365, "lr": 1.6805659457061362e-05} +{"train_loss": 0.014008751139044762, "grad_norm": 0.14355586469173431, "global_step": 38748, "epoch": 365, "lr": 1.6803443112658486e-05} +{"train_loss": 0.015596112236380577, "grad_norm": 0.09813466668128967, "global_step": 38749, "epoch": 365, "lr": 1.6801226884893896e-05} +{"train_loss": 0.01668783649802208, "grad_norm": 0.07071355730295181, "global_step": 38750, "epoch": 365, "lr": 1.6799010773775338e-05} +{"train_loss": 0.014751977287232876, "grad_norm": 0.1411977857351303, "global_step": 38751, "epoch": 365, "lr": 1.6796794779310615e-05} +{"train_loss": 0.016931289806962013, "grad_norm": 0.05954277515411377, "global_step": 38752, "epoch": 365, "lr": 1.679457890150753e-05} +{"train_loss": 0.014289054088294506, "grad_norm": 0.08122169971466064, "global_step": 38753, "epoch": 365, "lr": 1.679236314037386e-05} +{"train_loss": 0.013925379142165184, "grad_norm": 0.0691838189959526, "global_step": 38754, "epoch": 365, "lr": 1.6790147495917363e-05} +{"train_loss": 0.01781032234430313, "grad_norm": 0.08739811182022095, "global_step": 38755, "epoch": 365, "lr": 1.678793196814585e-05} +{"train_loss": 0.015522858127951622, "grad_norm": 0.10656210035085678, "global_step": 38756, "epoch": 365, "lr": 1.678571655706711e-05} +{"train_loss": 0.01639533042907715, "grad_norm": 0.1400371789932251, "global_step": 38757, "epoch": 365, "lr": 1.678350126268891e-05} +{"train_loss": 0.014782221987843513, "grad_norm": 0.09640476107597351, "global_step": 38758, "epoch": 365, "lr": 1.6781286085019048e-05} +{"train_loss": 0.014121504500508308, "grad_norm": 0.08232229202985764, "global_step": 38759, "epoch": 365, "lr": 1.677907102406529e-05} +{"train_loss": 0.015051756985485554, "grad_norm": 0.08302843570709229, "global_step": 38760, "epoch": 365, "lr": 1.6776856079835445e-05} +{"train_loss": 0.015189587138593197, "grad_norm": 0.0810190886259079, "global_step": 38761, "epoch": 365, "lr": 1.6774641252337264e-05} +{"train_loss": 0.013676745817065239, "grad_norm": 0.0827179029583931, "global_step": 38762, "epoch": 365, "lr": 1.677242654157854e-05} +{"train_loss": 0.012649564072489738, "grad_norm": 0.06483198702335358, "global_step": 38763, "epoch": 365, "lr": 1.6770211947567082e-05} +{"train_loss": 0.016005512326955795, "grad_norm": 0.11394604295492172, "global_step": 38764, "epoch": 365, "lr": 1.6767997470310644e-05} +{"train_loss": 0.013477729633450508, "grad_norm": 0.11080866307020187, "global_step": 38765, "epoch": 365, "lr": 1.676578310981699e-05} +{"train_loss": 0.015336697921156883, "grad_norm": 0.1315748542547226, "global_step": 38766, "epoch": 365, "lr": 1.6763568866093942e-05} +{"train_loss": 0.016560068354010582, "grad_norm": 0.08561503142118454, "global_step": 38767, "epoch": 365, "lr": 1.6761354739149243e-05} +{"train_loss": 0.015805594623088837, "grad_norm": 0.05549915134906769, "global_step": 38768, "epoch": 365, "lr": 1.6759140728990686e-05} +{"train_loss": 0.015147965401411057, "grad_norm": 0.07367175817489624, "global_step": 38769, "epoch": 365, "lr": 1.6756926835626073e-05} +{"train_loss": 0.011800708249211311, "grad_norm": 0.05801306664943695, "global_step": 38770, "epoch": 365, "lr": 1.675471305906316e-05} +{"train_loss": 0.01557675376534462, "grad_norm": 0.13943584263324738, "global_step": 38771, "epoch": 365, "lr": 1.675249939930971e-05} +{"train_loss": 0.01605287566781044, "grad_norm": 0.05428997054696083, "global_step": 38772, "epoch": 365, "lr": 1.6750285856373515e-05} +{"train_loss": 0.016552504152059555, "grad_norm": 0.06833790987730026, "global_step": 38773, "epoch": 365, "lr": 1.6748072430262373e-05} +{"train_loss": 0.015878744423389435, "grad_norm": 0.11318711191415787, "global_step": 38774, "epoch": 365, "lr": 1.674585912098403e-05} +{"train_loss": 0.014314000494778156, "grad_norm": 0.09696123003959656, "global_step": 38775, "epoch": 365, "lr": 1.6743645928546275e-05} +{"train_loss": 0.013587643392384052, "grad_norm": 0.06453695148229599, "global_step": 38776, "epoch": 365, "lr": 1.6741432852956905e-05} +{"train_loss": 0.013828136026859283, "grad_norm": 0.12582045793533325, "global_step": 38777, "epoch": 365, "lr": 1.6739219894223673e-05} +{"train_loss": 0.01575164496898651, "grad_norm": 0.09839202463626862, "global_step": 38778, "epoch": 365, "lr": 1.6737007052354342e-05} +{"train_loss": 0.015301011502742767, "grad_norm": 0.07688567787408829, "global_step": 38779, "epoch": 365, "lr": 1.67347943273567e-05} +{"train_loss": 0.014544477686285973, "grad_norm": 0.08995263278484344, "global_step": 38780, "epoch": 365, "lr": 1.6732581719238543e-05} +{"train_loss": 0.01635849103331566, "grad_norm": 0.11758670210838318, "global_step": 38781, "epoch": 365, "lr": 1.6730369228007615e-05} +{"train_loss": 0.01461982075124979, "grad_norm": 0.08483457565307617, "global_step": 38782, "epoch": 365, "lr": 1.672815685367171e-05} +{"train_loss": 0.012362427078187466, "grad_norm": 0.05218067020177841, "global_step": 38783, "epoch": 365, "lr": 1.6725944596238596e-05} +{"train_loss": 0.014460661448538303, "grad_norm": 0.0739629939198494, "global_step": 38784, "epoch": 365, "lr": 1.6723732455716023e-05} +{"train_loss": 0.016769859939813614, "grad_norm": 0.10118789225816727, "global_step": 38785, "epoch": 365, "lr": 1.672152043211178e-05} +{"train_loss": 0.015170976519584656, "grad_norm": 0.07796908915042877, "global_step": 38786, "epoch": 365, "lr": 1.6719308525433646e-05} +{"train_loss": 0.01621219888329506, "grad_norm": 0.0720483735203743, "global_step": 38787, "epoch": 365, "lr": 1.6717096735689408e-05} +{"train_loss": 0.016661444678902626, "grad_norm": 0.0901489332318306, "global_step": 38788, "epoch": 365, "lr": 1.671488506288681e-05} +{"train_loss": 0.014748815447092056, "grad_norm": 0.06873816251754761, "global_step": 38789, "epoch": 365, "lr": 1.671267350703362e-05} +{"train_loss": 0.01618892326951027, "grad_norm": 0.05785860866308212, "global_step": 38790, "epoch": 365, "lr": 1.6710462068137632e-05} +{"train_loss": 0.013166694901883602, "grad_norm": 0.08771756291389465, "global_step": 38791, "epoch": 365, "lr": 1.6708250746206588e-05} +{"train_loss": 0.014159886166453362, "grad_norm": 0.07725398242473602, "global_step": 38792, "epoch": 365, "lr": 1.6706039541248276e-05} +{"train_loss": 0.017362672835588455, "grad_norm": 0.10679768025875092, "global_step": 38793, "epoch": 365, "lr": 1.6703828453270475e-05} +{"train_loss": 0.015340043231844902, "grad_norm": 0.11935735493898392, "global_step": 38794, "epoch": 365, "lr": 1.670161748228094e-05} +{"train_loss": 0.015072615948980147, "grad_norm": 0.12360749393701553, "global_step": 38795, "epoch": 365, "lr": 1.669940662828742e-05, "val_loss": 0.01949707604944706} +{"train_loss": 0.013677542097866535, "grad_norm": 0.057107407599687576, "global_step": 38796, "epoch": 366, "lr": 1.6697195891297713e-05} +{"train_loss": 0.012616690248250961, "grad_norm": 0.07184700667858124, "global_step": 38797, "epoch": 366, "lr": 1.6694985271319585e-05} +{"train_loss": 0.0133496830239892, "grad_norm": 0.07306421548128128, "global_step": 38798, "epoch": 366, "lr": 1.669277476836078e-05} +{"train_loss": 0.012700037099421024, "grad_norm": 0.05756320059299469, "global_step": 38799, "epoch": 366, "lr": 1.66905643824291e-05} +{"train_loss": 0.016543777659535408, "grad_norm": 0.07129580527544022, "global_step": 38800, "epoch": 366, "lr": 1.6688354113532273e-05} +{"train_loss": 0.01441190205514431, "grad_norm": 0.06896710395812988, "global_step": 38801, "epoch": 366, "lr": 1.6686143961678098e-05} +{"train_loss": 0.01512267254292965, "grad_norm": 0.07134599983692169, "global_step": 38802, "epoch": 366, "lr": 1.668393392687431e-05} +{"train_loss": 0.016458820551633835, "grad_norm": 0.056248463690280914, "global_step": 38803, "epoch": 366, "lr": 1.6681724009128686e-05} +{"train_loss": 0.016385599970817566, "grad_norm": 0.13892333209514618, "global_step": 38804, "epoch": 366, "lr": 1.667951420844901e-05} +{"train_loss": 0.013653299771249294, "grad_norm": 0.11771081387996674, "global_step": 38805, "epoch": 366, "lr": 1.6677304524843035e-05} +{"train_loss": 0.016426626592874527, "grad_norm": 0.08157341182231903, "global_step": 38806, "epoch": 366, "lr": 1.66750949583185e-05} +{"train_loss": 0.015465876087546349, "grad_norm": 0.1161346435546875, "global_step": 38807, "epoch": 366, "lr": 1.667288550888321e-05} +{"train_loss": 0.015958376228809357, "grad_norm": 0.08751379698514938, "global_step": 38808, "epoch": 366, "lr": 1.6670676176544886e-05} +{"train_loss": 0.01490586344152689, "grad_norm": 0.09999629855155945, "global_step": 38809, "epoch": 366, "lr": 1.6668466961311314e-05} +{"train_loss": 0.014174415729939938, "grad_norm": 0.06872358173131943, "global_step": 38810, "epoch": 366, "lr": 1.6666257863190266e-05} +{"train_loss": 0.013516724109649658, "grad_norm": 0.08528467267751694, "global_step": 38811, "epoch": 366, "lr": 1.6664048882189493e-05} +{"train_loss": 0.013549861498177052, "grad_norm": 0.05781328305602074, "global_step": 38812, "epoch": 366, "lr": 1.6661840018316738e-05} +{"train_loss": 0.015569974668323994, "grad_norm": 0.07092706859111786, "global_step": 38813, "epoch": 366, "lr": 1.6659631271579783e-05} +{"train_loss": 0.01763341575860977, "grad_norm": 0.06605862826108932, "global_step": 38814, "epoch": 366, "lr": 1.6657422641986397e-05} +{"train_loss": 0.01574093848466873, "grad_norm": 0.07974448055028915, "global_step": 38815, "epoch": 366, "lr": 1.6655214129544312e-05} +{"train_loss": 0.013865298591554165, "grad_norm": 0.08751694113016129, "global_step": 38816, "epoch": 366, "lr": 1.66530057342613e-05} +{"train_loss": 0.01585027202963829, "grad_norm": 0.08118273317813873, "global_step": 38817, "epoch": 366, "lr": 1.6650797456145147e-05} +{"train_loss": 0.014767687767744064, "grad_norm": 0.09684424847364426, "global_step": 38818, "epoch": 366, "lr": 1.664858929520358e-05} +{"train_loss": 0.01408037543296814, "grad_norm": 0.07106073200702667, "global_step": 38819, "epoch": 366, "lr": 1.6646381251444358e-05} +{"train_loss": 0.017051517963409424, "grad_norm": 0.10473126173019409, "global_step": 38820, "epoch": 366, "lr": 1.6644173324875247e-05} +{"train_loss": 0.015254603698849678, "grad_norm": 0.08108197897672653, "global_step": 38821, "epoch": 366, "lr": 1.6641965515504016e-05} +{"train_loss": 0.017268341034650803, "grad_norm": 0.11689493060112, "global_step": 38822, "epoch": 366, "lr": 1.66397578233384e-05} +{"train_loss": 0.01755993627011776, "grad_norm": 0.08069055527448654, "global_step": 38823, "epoch": 366, "lr": 1.6637550248386176e-05} +{"train_loss": 0.013402815908193588, "grad_norm": 0.06691233068704605, "global_step": 38824, "epoch": 366, "lr": 1.6635342790655096e-05} +{"train_loss": 0.016107773408293724, "grad_norm": 0.11823606491088867, "global_step": 38825, "epoch": 366, "lr": 1.6633135450152893e-05} +{"train_loss": 0.01588624157011509, "grad_norm": 0.0844479352235794, "global_step": 38826, "epoch": 366, "lr": 1.6630928226887343e-05} +{"train_loss": 0.01591777801513672, "grad_norm": 0.06871943175792694, "global_step": 38827, "epoch": 366, "lr": 1.662872112086622e-05} +{"train_loss": 0.015918977558612823, "grad_norm": 0.06093268841505051, "global_step": 38828, "epoch": 366, "lr": 1.6626514132097237e-05} +{"train_loss": 0.016648344695568085, "grad_norm": 0.09220569580793381, "global_step": 38829, "epoch": 366, "lr": 1.6624307260588184e-05} +{"train_loss": 0.017458707094192505, "grad_norm": 0.06737035512924194, "global_step": 38830, "epoch": 366, "lr": 1.6622100506346787e-05} +{"train_loss": 0.014399521984159946, "grad_norm": 0.0575421042740345, "global_step": 38831, "epoch": 366, "lr": 1.661989386938083e-05} +{"train_loss": 0.01792125031352043, "grad_norm": 0.10756964236497879, "global_step": 38832, "epoch": 366, "lr": 1.661768734969803e-05} +{"train_loss": 0.018236299976706505, "grad_norm": 0.07613469660282135, "global_step": 38833, "epoch": 366, "lr": 1.661548094730616e-05} +{"train_loss": 0.017778433859348297, "grad_norm": 0.08181587606668472, "global_step": 38834, "epoch": 366, "lr": 1.6613274662212992e-05} +{"train_loss": 0.016595806926488876, "grad_norm": 0.1794532835483551, "global_step": 38835, "epoch": 366, "lr": 1.661106849442625e-05} +{"train_loss": 0.014471516013145447, "grad_norm": 0.0627969354391098, "global_step": 38836, "epoch": 366, "lr": 1.6608862443953687e-05} +{"train_loss": 0.014914480037987232, "grad_norm": 0.07579885423183441, "global_step": 38837, "epoch": 366, "lr": 1.6606656510803054e-05} +{"train_loss": 0.016461949795484543, "grad_norm": 0.13887767493724823, "global_step": 38838, "epoch": 366, "lr": 1.660445069498212e-05} +{"train_loss": 0.014976851642131805, "grad_norm": 0.12234411388635635, "global_step": 38839, "epoch": 366, "lr": 1.660224499649861e-05} +{"train_loss": 0.016525641083717346, "grad_norm": 0.12072053551673889, "global_step": 38840, "epoch": 366, "lr": 1.660003941536031e-05} +{"train_loss": 0.01599956676363945, "grad_norm": 0.10926608741283417, "global_step": 38841, "epoch": 366, "lr": 1.659783395157492e-05} +{"train_loss": 0.01674262247979641, "grad_norm": 0.14869073033332825, "global_step": 38842, "epoch": 366, "lr": 1.6595628605150232e-05} +{"train_loss": 0.015966221690177917, "grad_norm": 0.09270655363798141, "global_step": 38843, "epoch": 366, "lr": 1.659342337609397e-05} +{"train_loss": 0.014988507144153118, "grad_norm": 0.059947602450847626, "global_step": 38844, "epoch": 366, "lr": 1.659121826441388e-05} +{"train_loss": 0.013785585761070251, "grad_norm": 0.06502801179885864, "global_step": 38845, "epoch": 366, "lr": 1.6589013270117742e-05} +{"train_loss": 0.016947822645306587, "grad_norm": 0.11343836039304733, "global_step": 38846, "epoch": 366, "lr": 1.658680839321328e-05} +{"train_loss": 0.014491613022983074, "grad_norm": 0.12095021456480026, "global_step": 38847, "epoch": 366, "lr": 1.658460363370823e-05} +{"train_loss": 0.012155613861978054, "grad_norm": 0.07369808107614517, "global_step": 38848, "epoch": 366, "lr": 1.658239899161036e-05} +{"train_loss": 0.013791040517389774, "grad_norm": 0.09070482105016708, "global_step": 38849, "epoch": 366, "lr": 1.6580194466927394e-05} +{"train_loss": 0.01573215238749981, "grad_norm": 0.06505990773439407, "global_step": 38850, "epoch": 366, "lr": 1.6577990059667093e-05} +{"train_loss": 0.012618491426110268, "grad_norm": 0.138124480843544, "global_step": 38851, "epoch": 366, "lr": 1.657578576983721e-05} +{"train_loss": 0.015862692147493362, "grad_norm": 0.12183830887079239, "global_step": 38852, "epoch": 366, "lr": 1.6573581597445485e-05} +{"train_loss": 0.013799924403429031, "grad_norm": 0.055509451776742935, "global_step": 38853, "epoch": 366, "lr": 1.6571377542499645e-05} +{"train_loss": 0.013103742152452469, "grad_norm": 0.08007273077964783, "global_step": 38854, "epoch": 366, "lr": 1.6569173605007438e-05} +{"train_loss": 0.01366175152361393, "grad_norm": 0.14410577714443207, "global_step": 38855, "epoch": 366, "lr": 1.6566969784976638e-05} +{"train_loss": 0.013244029134511948, "grad_norm": 0.07688798010349274, "global_step": 38856, "epoch": 366, "lr": 1.6564766082414952e-05} +{"train_loss": 0.01466107927262783, "grad_norm": 0.06219809129834175, "global_step": 38857, "epoch": 366, "lr": 1.6562562497330132e-05} +{"train_loss": 0.013825137168169022, "grad_norm": 0.06260280311107635, "global_step": 38858, "epoch": 366, "lr": 1.6560359029729944e-05} +{"train_loss": 0.012974102050065994, "grad_norm": 0.0791066512465477, "global_step": 38859, "epoch": 366, "lr": 1.655815567962211e-05} +{"train_loss": 0.01450092252343893, "grad_norm": 0.07374522089958191, "global_step": 38860, "epoch": 366, "lr": 1.655595244701436e-05} +{"train_loss": 0.014580528251826763, "grad_norm": 0.08972377330064774, "global_step": 38861, "epoch": 366, "lr": 1.655374933191445e-05} +{"train_loss": 0.013761883601546288, "grad_norm": 0.10449735075235367, "global_step": 38862, "epoch": 366, "lr": 1.655154633433013e-05} +{"train_loss": 0.014131455682218075, "grad_norm": 0.052629273384809494, "global_step": 38863, "epoch": 366, "lr": 1.654934345426912e-05} +{"train_loss": 0.015514323487877846, "grad_norm": 0.1055457592010498, "global_step": 38864, "epoch": 366, "lr": 1.6547140691739183e-05} +{"train_loss": 0.015619711950421333, "grad_norm": 0.07759974151849747, "global_step": 38865, "epoch": 366, "lr": 1.654493804674805e-05} +{"train_loss": 0.017224837094545364, "grad_norm": 0.08406504988670349, "global_step": 38866, "epoch": 366, "lr": 1.654273551930343e-05} +{"train_loss": 0.01607746258378029, "grad_norm": 0.1020011305809021, "global_step": 38867, "epoch": 366, "lr": 1.654053310941309e-05} +{"train_loss": 0.012869475409388542, "grad_norm": 0.09479229152202606, "global_step": 38868, "epoch": 366, "lr": 1.6538330817084785e-05} +{"train_loss": 0.014485666528344154, "grad_norm": 0.07916035503149033, "global_step": 38869, "epoch": 366, "lr": 1.653612864232622e-05} +{"train_loss": 0.013848498463630676, "grad_norm": 0.057919248938560486, "global_step": 38870, "epoch": 366, "lr": 1.6533926585145158e-05} +{"train_loss": 0.013837404549121857, "grad_norm": 0.06623682379722595, "global_step": 38871, "epoch": 366, "lr": 1.6531724645549307e-05} +{"train_loss": 0.015104060992598534, "grad_norm": 0.14723892509937286, "global_step": 38872, "epoch": 366, "lr": 1.652952282354644e-05} +{"train_loss": 0.015430772677063942, "grad_norm": 0.13688410818576813, "global_step": 38873, "epoch": 366, "lr": 1.652732111914425e-05} +{"train_loss": 0.013202622532844543, "grad_norm": 0.07763484865427017, "global_step": 38874, "epoch": 366, "lr": 1.6525119532350508e-05} +{"train_loss": 0.01295491959899664, "grad_norm": 0.07591573148965836, "global_step": 38875, "epoch": 366, "lr": 1.6522918063172947e-05} +{"train_loss": 0.017281310632824898, "grad_norm": 0.07342907041311264, "global_step": 38876, "epoch": 366, "lr": 1.6520716711619294e-05} +{"train_loss": 0.014914218336343765, "grad_norm": 0.06570903956890106, "global_step": 38877, "epoch": 366, "lr": 1.6518515477697267e-05} +{"train_loss": 0.016601134091615677, "grad_norm": 0.11325593292713165, "global_step": 38878, "epoch": 366, "lr": 1.6516314361414615e-05} +{"train_loss": 0.016023889183998108, "grad_norm": 0.07670196145772934, "global_step": 38879, "epoch": 366, "lr": 1.6514113362779092e-05} +{"train_loss": 0.014449099078774452, "grad_norm": 0.0797479972243309, "global_step": 38880, "epoch": 366, "lr": 1.651191248179839e-05} +{"train_loss": 0.014141442254185677, "grad_norm": 0.1214088425040245, "global_step": 38881, "epoch": 366, "lr": 1.650971171848027e-05} +{"train_loss": 0.013534247875213623, "grad_norm": 0.08848286420106888, "global_step": 38882, "epoch": 366, "lr": 1.6507511072832488e-05} +{"train_loss": 0.014119836501777172, "grad_norm": 0.09263394773006439, "global_step": 38883, "epoch": 366, "lr": 1.650531054486272e-05} +{"train_loss": 0.013345608487725258, "grad_norm": 0.11027049273252487, "global_step": 38884, "epoch": 366, "lr": 1.6503110134578714e-05} +{"train_loss": 0.016430892050266266, "grad_norm": 0.11512770503759384, "global_step": 38885, "epoch": 366, "lr": 1.6500909841988234e-05} +{"train_loss": 0.01567237637937069, "grad_norm": 0.07918681204319, "global_step": 38886, "epoch": 366, "lr": 1.649870966709897e-05} +{"train_loss": 0.01568474806845188, "grad_norm": 0.07993371039628983, "global_step": 38887, "epoch": 366, "lr": 1.6496509609918676e-05} +{"train_loss": 0.01602293737232685, "grad_norm": 0.07432142645120621, "global_step": 38888, "epoch": 366, "lr": 1.649430967045509e-05} +{"train_loss": 0.01473773829638958, "grad_norm": 0.07707060128450394, "global_step": 38889, "epoch": 366, "lr": 1.6492109848715926e-05} +{"train_loss": 0.016087928786873817, "grad_norm": 0.09528219699859619, "global_step": 38890, "epoch": 366, "lr": 1.6489910144708898e-05} +{"train_loss": 0.012948466464877129, "grad_norm": 0.060356780886650085, "global_step": 38891, "epoch": 366, "lr": 1.6487710558441753e-05} +{"train_loss": 0.015245538204908371, "grad_norm": 0.07792042940855026, "global_step": 38892, "epoch": 366, "lr": 1.648551108992224e-05} +{"train_loss": 0.016386263072490692, "grad_norm": 0.07699527591466904, "global_step": 38893, "epoch": 366, "lr": 1.648331173915807e-05} +{"train_loss": 0.013205735012888908, "grad_norm": 0.08603013306856155, "global_step": 38894, "epoch": 366, "lr": 1.648111250615694e-05} +{"train_loss": 0.013741958886384964, "grad_norm": 0.06931254267692566, "global_step": 38895, "epoch": 366, "lr": 1.6478913390926608e-05} +{"train_loss": 0.012897415086627007, "grad_norm": 0.08287449926137924, "global_step": 38896, "epoch": 366, "lr": 1.6476714393474813e-05} +{"train_loss": 0.016391992568969727, "grad_norm": 0.11894598603248596, "global_step": 38897, "epoch": 366, "lr": 1.6474515513809248e-05} +{"train_loss": 0.016033144667744637, "grad_norm": 0.09608829766511917, "global_step": 38898, "epoch": 366, "lr": 1.6472316751937655e-05} +{"train_loss": 0.017462074756622314, "grad_norm": 0.10344479233026505, "global_step": 38899, "epoch": 366, "lr": 1.6470118107867778e-05} +{"train_loss": 0.014667216688394547, "grad_norm": 0.09574229270219803, "global_step": 38900, "epoch": 366, "lr": 1.646791958160732e-05} +{"train_loss": 0.015058358366829607, "grad_norm": 0.08199846744537354, "global_step": 38901, "epoch": 366, "lr": 1.6465721173164002e-05, "val_loss": 0.019477788358926773} +{"train_loss": 0.017229383811354637, "grad_norm": 0.1300121247768402, "global_step": 38902, "epoch": 367, "lr": 1.646352288254555e-05} +{"train_loss": 0.013444319367408752, "grad_norm": 0.06768091022968292, "global_step": 38903, "epoch": 367, "lr": 1.646132470975971e-05} +{"train_loss": 0.0152530986815691, "grad_norm": 0.10600031912326813, "global_step": 38904, "epoch": 367, "lr": 1.645912665481418e-05} +{"train_loss": 0.01469959132373333, "grad_norm": 0.13070519268512726, "global_step": 38905, "epoch": 367, "lr": 1.6456928717716703e-05} +{"train_loss": 0.015845097601413727, "grad_norm": 0.0789123997092247, "global_step": 38906, "epoch": 367, "lr": 1.6454730898474992e-05} +{"train_loss": 0.013251201249659061, "grad_norm": 0.048059239983558655, "global_step": 38907, "epoch": 367, "lr": 1.645253319709675e-05} +{"train_loss": 0.015209638513624668, "grad_norm": 0.14286375045776367, "global_step": 38908, "epoch": 367, "lr": 1.645033561358971e-05} +{"train_loss": 0.01496632769703865, "grad_norm": 0.1340443342924118, "global_step": 38909, "epoch": 367, "lr": 1.644813814796163e-05} +{"train_loss": 0.015199127607047558, "grad_norm": 0.08955326676368713, "global_step": 38910, "epoch": 367, "lr": 1.6445940800220178e-05} +{"train_loss": 0.01737836003303528, "grad_norm": 0.11851323395967484, "global_step": 38911, "epoch": 367, "lr": 1.6443743570373116e-05} +{"train_loss": 0.013553283177316189, "grad_norm": 0.05889154225587845, "global_step": 38912, "epoch": 367, "lr": 1.644154645842813e-05} +{"train_loss": 0.016828587278723717, "grad_norm": 0.06267156451940536, "global_step": 38913, "epoch": 367, "lr": 1.6439349464392973e-05} +{"train_loss": 0.015573360957205296, "grad_norm": 0.0827883780002594, "global_step": 38914, "epoch": 367, "lr": 1.643715258827533e-05} +{"train_loss": 0.01522623561322689, "grad_norm": 0.09568031877279282, "global_step": 38915, "epoch": 367, "lr": 1.643495583008293e-05} +{"train_loss": 0.014244082383811474, "grad_norm": 0.09563497453927994, "global_step": 38916, "epoch": 367, "lr": 1.643275918982352e-05} +{"train_loss": 0.013858746737241745, "grad_norm": 0.0627978965640068, "global_step": 38917, "epoch": 367, "lr": 1.6430562667504796e-05} +{"train_loss": 0.013350054621696472, "grad_norm": 0.05994541570544243, "global_step": 38918, "epoch": 367, "lr": 1.6428366263134458e-05} +{"train_loss": 0.013249549083411694, "grad_norm": 0.09196209907531738, "global_step": 38919, "epoch": 367, "lr": 1.6426169976720245e-05} +{"train_loss": 0.014380344189703465, "grad_norm": 0.16794045269489288, "global_step": 38920, "epoch": 367, "lr": 1.6423973808269884e-05} +{"train_loss": 0.015551045536994934, "grad_norm": 0.06078966334462166, "global_step": 38921, "epoch": 367, "lr": 1.6421777757791063e-05} +{"train_loss": 0.01630985736846924, "grad_norm": 0.059801697731018066, "global_step": 38922, "epoch": 367, "lr": 1.6419581825291508e-05} +{"train_loss": 0.01602855697274208, "grad_norm": 0.13295988738536835, "global_step": 38923, "epoch": 367, "lr": 1.6417386010778973e-05} +{"train_loss": 0.016637727618217468, "grad_norm": 0.1053808331489563, "global_step": 38924, "epoch": 367, "lr": 1.6415190314261103e-05} +{"train_loss": 0.015219239518046379, "grad_norm": 0.10208303481340408, "global_step": 38925, "epoch": 367, "lr": 1.6412994735745653e-05} +{"train_loss": 0.014591134153306484, "grad_norm": 0.13815359771251678, "global_step": 38926, "epoch": 367, "lr": 1.641079927524035e-05} +{"train_loss": 0.015454587526619434, "grad_norm": 0.05556244030594826, "global_step": 38927, "epoch": 367, "lr": 1.640860393275287e-05} +{"train_loss": 0.015564212575554848, "grad_norm": 0.1500420719385147, "global_step": 38928, "epoch": 367, "lr": 1.6406408708290947e-05} +{"train_loss": 0.016259465366601944, "grad_norm": 0.05923990532755852, "global_step": 38929, "epoch": 367, "lr": 1.6404213601862317e-05} +{"train_loss": 0.015173528343439102, "grad_norm": 0.07264687120914459, "global_step": 38930, "epoch": 367, "lr": 1.6402018613474667e-05} +{"train_loss": 0.01563269831240177, "grad_norm": 0.06596619635820389, "global_step": 38931, "epoch": 367, "lr": 1.6399823743135694e-05} +{"train_loss": 0.015432426705956459, "grad_norm": 0.060961559414863586, "global_step": 38932, "epoch": 367, "lr": 1.639762899085313e-05} +{"train_loss": 0.01450624130666256, "grad_norm": 0.15363940596580505, "global_step": 38933, "epoch": 367, "lr": 1.6395434356634703e-05} +{"train_loss": 0.014707191847264767, "grad_norm": 0.07607889920473099, "global_step": 38934, "epoch": 367, "lr": 1.6393239840488084e-05} +{"train_loss": 0.012695631012320518, "grad_norm": 0.07819932699203491, "global_step": 38935, "epoch": 367, "lr": 1.6391045442421022e-05} +{"train_loss": 0.016843006014823914, "grad_norm": 0.09094584733247757, "global_step": 38936, "epoch": 367, "lr": 1.6388851162441198e-05} +{"train_loss": 0.014539049938321114, "grad_norm": 0.061754386872053146, "global_step": 38937, "epoch": 367, "lr": 1.638665700055635e-05} +{"train_loss": 0.015133874490857124, "grad_norm": 0.11933242529630661, "global_step": 38938, "epoch": 367, "lr": 1.6384462956774148e-05} +{"train_loss": 0.015570053830742836, "grad_norm": 0.09936060756444931, "global_step": 38939, "epoch": 367, "lr": 1.638226903110233e-05} +{"train_loss": 0.015653394162654877, "grad_norm": 0.06932145357131958, "global_step": 38940, "epoch": 367, "lr": 1.638007522354863e-05} +{"train_loss": 0.01609625853598118, "grad_norm": 0.05929693952202797, "global_step": 38941, "epoch": 367, "lr": 1.6377881534120693e-05} +{"train_loss": 0.012336819432675838, "grad_norm": 0.06656186282634735, "global_step": 38942, "epoch": 367, "lr": 1.637568796282626e-05} +{"train_loss": 0.01568443328142166, "grad_norm": 0.13442738354206085, "global_step": 38943, "epoch": 367, "lr": 1.637349450967303e-05} +{"train_loss": 0.018266800791025162, "grad_norm": 0.09245200455188751, "global_step": 38944, "epoch": 367, "lr": 1.637130117466874e-05} +{"train_loss": 0.015901943668723106, "grad_norm": 0.12142112106084824, "global_step": 38945, "epoch": 367, "lr": 1.636910795782105e-05} +{"train_loss": 0.016390830278396606, "grad_norm": 0.07771201431751251, "global_step": 38946, "epoch": 367, "lr": 1.636691485913771e-05} +{"train_loss": 0.015955984592437744, "grad_norm": 0.10583572089672089, "global_step": 38947, "epoch": 367, "lr": 1.6364721878626406e-05} +{"train_loss": 0.015158763155341148, "grad_norm": 0.13032527267932892, "global_step": 38948, "epoch": 367, "lr": 1.636252901629482e-05} +{"train_loss": 0.016279233619570732, "grad_norm": 0.09364538639783859, "global_step": 38949, "epoch": 367, "lr": 1.6360336272150685e-05} +{"train_loss": 0.01594974845647812, "grad_norm": 0.059861186891794205, "global_step": 38950, "epoch": 367, "lr": 1.6358143646201706e-05} +{"train_loss": 0.014292092993855476, "grad_norm": 0.16118939220905304, "global_step": 38951, "epoch": 367, "lr": 1.6355951138455572e-05} +{"train_loss": 0.014845659025013447, "grad_norm": 0.06759816408157349, "global_step": 38952, "epoch": 367, "lr": 1.6353758748920007e-05} +{"train_loss": 0.012963980436325073, "grad_norm": 0.07585092633962631, "global_step": 38953, "epoch": 367, "lr": 1.6351566477602687e-05} +{"train_loss": 0.015893694013357162, "grad_norm": 0.06957452744245529, "global_step": 38954, "epoch": 367, "lr": 1.6349374324511345e-05} +{"train_loss": 0.014443749561905861, "grad_norm": 0.08562233299016953, "global_step": 38955, "epoch": 367, "lr": 1.6347182289653652e-05} +{"train_loss": 0.013892989605665207, "grad_norm": 0.07188069075345993, "global_step": 38956, "epoch": 367, "lr": 1.634499037303732e-05} +{"train_loss": 0.015140186995267868, "grad_norm": 0.07780537754297256, "global_step": 38957, "epoch": 367, "lr": 1.6342798574670076e-05} +{"train_loss": 0.01511344313621521, "grad_norm": 0.13569706678390503, "global_step": 38958, "epoch": 367, "lr": 1.63406068945596e-05} +{"train_loss": 0.016398416832089424, "grad_norm": 0.14489108324050903, "global_step": 38959, "epoch": 367, "lr": 1.6338415332713576e-05} +{"train_loss": 0.01405548956245184, "grad_norm": 0.06055322289466858, "global_step": 38960, "epoch": 367, "lr": 1.6336223889139725e-05} +{"train_loss": 0.013217326253652573, "grad_norm": 0.06472283601760864, "global_step": 38961, "epoch": 367, "lr": 1.6334032563845754e-05} +{"train_loss": 0.014287080615758896, "grad_norm": 0.0644308552145958, "global_step": 38962, "epoch": 367, "lr": 1.6331841356839338e-05} +{"train_loss": 0.014324955642223358, "grad_norm": 0.09413090348243713, "global_step": 38963, "epoch": 367, "lr": 1.632965026812819e-05} +{"train_loss": 0.015460258349776268, "grad_norm": 0.09308519959449768, "global_step": 38964, "epoch": 367, "lr": 1.6327459297720043e-05} +{"train_loss": 0.014345088042318821, "grad_norm": 0.09152479469776154, "global_step": 38965, "epoch": 367, "lr": 1.6325268445622523e-05} +{"train_loss": 0.013364787213504314, "grad_norm": 0.07201240211725235, "global_step": 38966, "epoch": 367, "lr": 1.6323077711843365e-05} +{"train_loss": 0.013277052901685238, "grad_norm": 0.051742758601903915, "global_step": 38967, "epoch": 367, "lr": 1.6320887096390285e-05} +{"train_loss": 0.014381530694663525, "grad_norm": 0.09401635080575943, "global_step": 38968, "epoch": 367, "lr": 1.631869659927095e-05} +{"train_loss": 0.014874424785375595, "grad_norm": 0.06891069561243057, "global_step": 38969, "epoch": 367, "lr": 1.631650622049306e-05} +{"train_loss": 0.015414638444781303, "grad_norm": 0.158712700009346, "global_step": 38970, "epoch": 367, "lr": 1.6314315960064336e-05} +{"train_loss": 0.016036123037338257, "grad_norm": 0.09179986268281937, "global_step": 38971, "epoch": 367, "lr": 1.6312125817992458e-05} +{"train_loss": 0.015099938027560711, "grad_norm": 0.06999652832746506, "global_step": 38972, "epoch": 367, "lr": 1.6309935794285102e-05} +{"train_loss": 0.014668635092675686, "grad_norm": 0.056214071810245514, "global_step": 38973, "epoch": 367, "lr": 1.630774588894998e-05} +{"train_loss": 0.013844938948750496, "grad_norm": 0.11343418806791306, "global_step": 38974, "epoch": 367, "lr": 1.63055561019948e-05} +{"train_loss": 0.01464525330811739, "grad_norm": 0.055852871388196945, "global_step": 38975, "epoch": 367, "lr": 1.630336643342723e-05} +{"train_loss": 0.016409896314144135, "grad_norm": 0.0890720784664154, "global_step": 38976, "epoch": 367, "lr": 1.6301176883254993e-05} +{"train_loss": 0.013545150868594646, "grad_norm": 0.08001434057950974, "global_step": 38977, "epoch": 367, "lr": 1.6298987451485748e-05} +{"train_loss": 0.016795961186289787, "grad_norm": 0.09522334486246109, "global_step": 38978, "epoch": 367, "lr": 1.6296798138127222e-05} +{"train_loss": 0.01730378344655037, "grad_norm": 0.14580035209655762, "global_step": 38979, "epoch": 367, "lr": 1.6294608943187074e-05} +{"train_loss": 0.018040692433714867, "grad_norm": 0.09191610664129257, "global_step": 38980, "epoch": 367, "lr": 1.6292419866673007e-05} +{"train_loss": 0.017216112464666367, "grad_norm": 0.08092810958623886, "global_step": 38981, "epoch": 367, "lr": 1.629023090859274e-05} +{"train_loss": 0.013543108478188515, "grad_norm": 0.11264616250991821, "global_step": 38982, "epoch": 367, "lr": 1.628804206895394e-05} +{"train_loss": 0.013880765065550804, "grad_norm": 0.10415070503950119, "global_step": 38983, "epoch": 367, "lr": 1.6285853347764285e-05} +{"train_loss": 0.016060873866081238, "grad_norm": 0.08079929649829865, "global_step": 38984, "epoch": 367, "lr": 1.628366474503149e-05} +{"train_loss": 0.014653699472546577, "grad_norm": 0.06478215754032135, "global_step": 38985, "epoch": 367, "lr": 1.6281476260763224e-05} +{"train_loss": 0.014944038353860378, "grad_norm": 0.09075101464986801, "global_step": 38986, "epoch": 367, "lr": 1.627928789496718e-05} +{"train_loss": 0.016314368695020676, "grad_norm": 0.10875200480222702, "global_step": 38987, "epoch": 367, "lr": 1.6277099647651072e-05} +{"train_loss": 0.01439230889081955, "grad_norm": 0.10249662399291992, "global_step": 38988, "epoch": 367, "lr": 1.6274911518822573e-05} +{"train_loss": 0.012620635330677032, "grad_norm": 0.05148334056138992, "global_step": 38989, "epoch": 367, "lr": 1.6272723508489352e-05} +{"train_loss": 0.01434864941984415, "grad_norm": 0.07978616654872894, "global_step": 38990, "epoch": 367, "lr": 1.627053561665911e-05} +{"train_loss": 0.01759260520339012, "grad_norm": 0.08860205858945847, "global_step": 38991, "epoch": 367, "lr": 1.626834784333956e-05} +{"train_loss": 0.014804166741669178, "grad_norm": 0.05800558999180794, "global_step": 38992, "epoch": 367, "lr": 1.6266160188538344e-05} +{"train_loss": 0.015627574175596237, "grad_norm": 0.06459154933691025, "global_step": 38993, "epoch": 367, "lr": 1.626397265226319e-05} +{"train_loss": 0.016292087733745575, "grad_norm": 0.09667938947677612, "global_step": 38994, "epoch": 367, "lr": 1.6261785234521742e-05} +{"train_loss": 0.013237256556749344, "grad_norm": 0.07920840382575989, "global_step": 38995, "epoch": 367, "lr": 1.625959793532173e-05} +{"train_loss": 0.015360480174422264, "grad_norm": 0.10108104348182678, "global_step": 38996, "epoch": 367, "lr": 1.62574107546708e-05} +{"train_loss": 0.014527772553265095, "grad_norm": 0.0969940647482872, "global_step": 38997, "epoch": 367, "lr": 1.6255223692576655e-05} +{"train_loss": 0.014920096844434738, "grad_norm": 0.06551169604063034, "global_step": 38998, "epoch": 367, "lr": 1.6253036749046997e-05} +{"train_loss": 0.017121095210313797, "grad_norm": 0.06520899385213852, "global_step": 38999, "epoch": 367, "lr": 1.6250849924089484e-05} +{"train_loss": 0.013325984589755535, "grad_norm": 0.0737159252166748, "global_step": 39000, "epoch": 367, "lr": 1.6248663217711802e-05} +{"train_loss": 0.012597453780472279, "grad_norm": 0.08190115541219711, "global_step": 39001, "epoch": 367, "lr": 1.624647662992163e-05} +{"train_loss": 0.01671895943582058, "grad_norm": 0.0668741837143898, "global_step": 39002, "epoch": 367, "lr": 1.6244290160726683e-05} +{"train_loss": 0.013949093408882618, "grad_norm": 0.09948762506246567, "global_step": 39003, "epoch": 367, "lr": 1.6242103810134603e-05} +{"train_loss": 0.012977352365851402, "grad_norm": 0.10262976586818695, "global_step": 39004, "epoch": 367, "lr": 1.6239917578153085e-05} +{"train_loss": 0.013245251029729843, "grad_norm": 0.10727763175964355, "global_step": 39005, "epoch": 367, "lr": 1.6237731464789845e-05} +{"train_loss": 0.015307205729186535, "grad_norm": 0.09947366267442703, "global_step": 39006, "epoch": 367, "lr": 1.623554547005251e-05} +{"train_loss": 0.015057041450350915, "grad_norm": 0.12245544791221619, "global_step": 39007, "epoch": 367, "lr": 1.6233359593948777e-05, "val_loss": 0.019291618838906288} +{"train_loss": 0.014773115515708923, "grad_norm": 0.07986021786928177, "global_step": 39008, "epoch": 368, "lr": 1.6231173836486358e-05} +{"train_loss": 0.012806715443730354, "grad_norm": 0.08544052392244339, "global_step": 39009, "epoch": 368, "lr": 1.6228988197672886e-05} +{"train_loss": 0.015868429094552994, "grad_norm": 0.148936465382576, "global_step": 39010, "epoch": 368, "lr": 1.622680267751606e-05} +{"train_loss": 0.0156718697398901, "grad_norm": 0.06208770349621773, "global_step": 39011, "epoch": 368, "lr": 1.6224617276023586e-05} +{"train_loss": 0.012726993300020695, "grad_norm": 0.07334011793136597, "global_step": 39012, "epoch": 368, "lr": 1.6222431993203113e-05} +{"train_loss": 0.013626659289002419, "grad_norm": 0.06631340086460114, "global_step": 39013, "epoch": 368, "lr": 1.6220246829062313e-05} +{"train_loss": 0.014620974659919739, "grad_norm": 0.06686396151781082, "global_step": 39014, "epoch": 368, "lr": 1.6218061783608875e-05} +{"train_loss": 0.015066723339259624, "grad_norm": 0.07672028988599777, "global_step": 39015, "epoch": 368, "lr": 1.6215876856850485e-05} +{"train_loss": 0.016847770661115646, "grad_norm": 0.06516379863023758, "global_step": 39016, "epoch": 368, "lr": 1.6213692048794805e-05} +{"train_loss": 0.017733857035636902, "grad_norm": 0.09580738097429276, "global_step": 39017, "epoch": 368, "lr": 1.621150735944953e-05} +{"train_loss": 0.015167607925832272, "grad_norm": 0.09935523569583893, "global_step": 39018, "epoch": 368, "lr": 1.6209322788822306e-05} +{"train_loss": 0.014744078740477562, "grad_norm": 0.10979273915290833, "global_step": 39019, "epoch": 368, "lr": 1.620713833692084e-05} +{"train_loss": 0.01412665843963623, "grad_norm": 0.07905438542366028, "global_step": 39020, "epoch": 368, "lr": 1.6204954003752782e-05} +{"train_loss": 0.014028923586010933, "grad_norm": 0.06140284985303879, "global_step": 39021, "epoch": 368, "lr": 1.6202769789325818e-05} +{"train_loss": 0.01427113451063633, "grad_norm": 0.15551014244556427, "global_step": 39022, "epoch": 368, "lr": 1.620058569364764e-05} +{"train_loss": 0.01317192055284977, "grad_norm": 0.06673197448253632, "global_step": 39023, "epoch": 368, "lr": 1.6198401716725904e-05} +{"train_loss": 0.015489621087908745, "grad_norm": 0.09472738951444626, "global_step": 39024, "epoch": 368, "lr": 1.6196217858568268e-05} +{"train_loss": 0.016826637089252472, "grad_norm": 0.09278786927461624, "global_step": 39025, "epoch": 368, "lr": 1.619403411918244e-05} +{"train_loss": 0.015302064828574657, "grad_norm": 0.08170848339796066, "global_step": 39026, "epoch": 368, "lr": 1.6191850498576054e-05} +{"train_loss": 0.014529496431350708, "grad_norm": 0.06188989803195, "global_step": 39027, "epoch": 368, "lr": 1.6189666996756804e-05} +{"train_loss": 0.017825981602072716, "grad_norm": 0.0913519486784935, "global_step": 39028, "epoch": 368, "lr": 1.618748361373238e-05} +{"train_loss": 0.015506993979215622, "grad_norm": 0.09358606487512589, "global_step": 39029, "epoch": 368, "lr": 1.6185300349510428e-05} +{"train_loss": 0.016939565539360046, "grad_norm": 0.07717673480510712, "global_step": 39030, "epoch": 368, "lr": 1.6183117204098614e-05} +{"train_loss": 0.01675253175199032, "grad_norm": 0.15400491654872894, "global_step": 39031, "epoch": 368, "lr": 1.618093417750462e-05} +{"train_loss": 0.01716485247015953, "grad_norm": 0.06848832964897156, "global_step": 39032, "epoch": 368, "lr": 1.6178751269736125e-05} +{"train_loss": 0.016983315348625183, "grad_norm": 0.0878477618098259, "global_step": 39033, "epoch": 368, "lr": 1.617656848080078e-05} +{"train_loss": 0.017648369073867798, "grad_norm": 0.0616944395005703, "global_step": 39034, "epoch": 368, "lr": 1.617438581070626e-05} +{"train_loss": 0.01621653512120247, "grad_norm": 0.10613879561424255, "global_step": 39035, "epoch": 368, "lr": 1.6172203259460255e-05} +{"train_loss": 0.014088476076722145, "grad_norm": 0.07774369418621063, "global_step": 39036, "epoch": 368, "lr": 1.617002082707042e-05} +{"train_loss": 0.012937290593981743, "grad_norm": 0.07592275738716125, "global_step": 39037, "epoch": 368, "lr": 1.6167838513544402e-05} +{"train_loss": 0.015480313450098038, "grad_norm": 0.08613825589418411, "global_step": 39038, "epoch": 368, "lr": 1.6165656318889883e-05} +{"train_loss": 0.015086693689227104, "grad_norm": 0.06570813059806824, "global_step": 39039, "epoch": 368, "lr": 1.616347424311455e-05} +{"train_loss": 0.015844857320189476, "grad_norm": 0.0854850709438324, "global_step": 39040, "epoch": 368, "lr": 1.616129228622604e-05} +{"train_loss": 0.013217957690358162, "grad_norm": 0.07753857225179672, "global_step": 39041, "epoch": 368, "lr": 1.615911044823205e-05} +{"train_loss": 0.01669754460453987, "grad_norm": 0.08727862685918808, "global_step": 39042, "epoch": 368, "lr": 1.6156928729140226e-05} +{"train_loss": 0.014533079229295254, "grad_norm": 0.0887879952788353, "global_step": 39043, "epoch": 368, "lr": 1.615474712895822e-05} +{"train_loss": 0.013851282186806202, "grad_norm": 0.06641647219657898, "global_step": 39044, "epoch": 368, "lr": 1.6152565647693714e-05} +{"train_loss": 0.01365756243467331, "grad_norm": 0.11144096404314041, "global_step": 39045, "epoch": 368, "lr": 1.6150384285354375e-05} +{"train_loss": 0.015067541040480137, "grad_norm": 0.08343144506216049, "global_step": 39046, "epoch": 368, "lr": 1.6148203041947894e-05} +{"train_loss": 0.0174068845808506, "grad_norm": 0.1943022608757019, "global_step": 39047, "epoch": 368, "lr": 1.6146021917481875e-05} +{"train_loss": 0.013484841212630272, "grad_norm": 0.07915916293859482, "global_step": 39048, "epoch": 368, "lr": 1.6143840911964015e-05} +{"train_loss": 0.013981418684124947, "grad_norm": 0.07605796307325363, "global_step": 39049, "epoch": 368, "lr": 1.6141660025401985e-05} +{"train_loss": 0.014483386650681496, "grad_norm": 0.06760551780462265, "global_step": 39050, "epoch": 368, "lr": 1.6139479257803425e-05} +{"train_loss": 0.014571866020560265, "grad_norm": 0.08030401915311813, "global_step": 39051, "epoch": 368, "lr": 1.6137298609176006e-05} +{"train_loss": 0.016008323058485985, "grad_norm": 0.10555437952280045, "global_step": 39052, "epoch": 368, "lr": 1.6135118079527412e-05} +{"train_loss": 0.01686974987387657, "grad_norm": 0.05255116894841194, "global_step": 39053, "epoch": 368, "lr": 1.6132937668865283e-05} +{"train_loss": 0.01657254435122013, "grad_norm": 0.08615179359912872, "global_step": 39054, "epoch": 368, "lr": 1.613075737719727e-05} +{"train_loss": 0.014727856032550335, "grad_norm": 0.08620315045118332, "global_step": 39055, "epoch": 368, "lr": 1.6128577204531044e-05} +{"train_loss": 0.014611076563596725, "grad_norm": 0.06655669212341309, "global_step": 39056, "epoch": 368, "lr": 1.6126397150874285e-05} +{"train_loss": 0.01654360070824623, "grad_norm": 0.06793075799942017, "global_step": 39057, "epoch": 368, "lr": 1.6124217216234617e-05} +{"train_loss": 0.013000828213989735, "grad_norm": 0.12277019023895264, "global_step": 39058, "epoch": 368, "lr": 1.6122037400619733e-05} +{"train_loss": 0.013898350298404694, "grad_norm": 0.089299775660038, "global_step": 39059, "epoch": 368, "lr": 1.6119857704037266e-05} +{"train_loss": 0.01528515387326479, "grad_norm": 0.09930048882961273, "global_step": 39060, "epoch": 368, "lr": 1.6117678126494894e-05} +{"train_loss": 0.013818231411278248, "grad_norm": 0.07054793834686279, "global_step": 39061, "epoch": 368, "lr": 1.6115498668000255e-05} +{"train_loss": 0.014609385281801224, "grad_norm": 0.09146543592214584, "global_step": 39062, "epoch": 368, "lr": 1.6113319328561016e-05} +{"train_loss": 0.015327172353863716, "grad_norm": 0.09077517688274384, "global_step": 39063, "epoch": 368, "lr": 1.6111140108184852e-05} +{"train_loss": 0.015624692663550377, "grad_norm": 0.08644998073577881, "global_step": 39064, "epoch": 368, "lr": 1.6108961006879403e-05} +{"train_loss": 0.014700056985020638, "grad_norm": 0.10543062537908554, "global_step": 39065, "epoch": 368, "lr": 1.610678202465231e-05} +{"train_loss": 0.015415924601256847, "grad_norm": 0.06483688950538635, "global_step": 39066, "epoch": 368, "lr": 1.6104603161511263e-05} +{"train_loss": 0.016462672501802444, "grad_norm": 0.06292159110307693, "global_step": 39067, "epoch": 368, "lr": 1.610242441746388e-05} +{"train_loss": 0.015158627182245255, "grad_norm": 0.08190987259149551, "global_step": 39068, "epoch": 368, "lr": 1.6100245792517838e-05} +{"train_loss": 0.01363113708794117, "grad_norm": 0.08051975071430206, "global_step": 39069, "epoch": 368, "lr": 1.6098067286680805e-05} +{"train_loss": 0.013791986741125584, "grad_norm": 0.13923613727092743, "global_step": 39070, "epoch": 368, "lr": 1.6095888899960428e-05} +{"train_loss": 0.015841562300920486, "grad_norm": 0.0767771378159523, "global_step": 39071, "epoch": 368, "lr": 1.6093710632364328e-05} +{"train_loss": 0.014758256264030933, "grad_norm": 0.09849736839532852, "global_step": 39072, "epoch": 368, "lr": 1.6091532483900186e-05} +{"train_loss": 0.015488417819142342, "grad_norm": 0.08624397963285446, "global_step": 39073, "epoch": 368, "lr": 1.6089354454575667e-05} +{"train_loss": 0.01389646902680397, "grad_norm": 0.08603702485561371, "global_step": 39074, "epoch": 368, "lr": 1.608717654439839e-05} +{"train_loss": 0.014228025451302528, "grad_norm": 0.08091794699430466, "global_step": 39075, "epoch": 368, "lr": 1.6084998753376034e-05} +{"train_loss": 0.014717379584908485, "grad_norm": 0.11499320715665817, "global_step": 39076, "epoch": 368, "lr": 1.6082821081516254e-05} +{"train_loss": 0.017932455986738205, "grad_norm": 0.12222089618444443, "global_step": 39077, "epoch": 368, "lr": 1.6080643528826695e-05} +{"train_loss": 0.01563461869955063, "grad_norm": 0.09094586968421936, "global_step": 39078, "epoch": 368, "lr": 1.607846609531498e-05} +{"train_loss": 0.015086255967617035, "grad_norm": 0.14065265655517578, "global_step": 39079, "epoch": 368, "lr": 1.607628878098879e-05} +{"train_loss": 0.014169612899422646, "grad_norm": 0.06912380456924438, "global_step": 39080, "epoch": 368, "lr": 1.6074111585855782e-05} +{"train_loss": 0.01315229944884777, "grad_norm": 0.19389748573303223, "global_step": 39081, "epoch": 368, "lr": 1.6071934509923576e-05} +{"train_loss": 0.014850737527012825, "grad_norm": 0.06124386936426163, "global_step": 39082, "epoch": 368, "lr": 1.6069757553199856e-05} +{"train_loss": 0.015825407579541206, "grad_norm": 0.075809545814991, "global_step": 39083, "epoch": 368, "lr": 1.606758071569225e-05} +{"train_loss": 0.013940260745584965, "grad_norm": 0.058791451156139374, "global_step": 39084, "epoch": 368, "lr": 1.606540399740839e-05} +{"train_loss": 0.01695091649889946, "grad_norm": 0.14049096405506134, "global_step": 39085, "epoch": 368, "lr": 1.606322739835594e-05} +{"train_loss": 0.01421642117202282, "grad_norm": 0.09796858578920364, "global_step": 39086, "epoch": 368, "lr": 1.6061050918542577e-05} +{"train_loss": 0.014894061721861362, "grad_norm": 0.1090330109000206, "global_step": 39087, "epoch": 368, "lr": 1.6058874557975895e-05} +{"train_loss": 0.013876650482416153, "grad_norm": 0.09297648817300797, "global_step": 39088, "epoch": 368, "lr": 1.605669831666359e-05} +{"train_loss": 0.015201910398900509, "grad_norm": 0.09710793197154999, "global_step": 39089, "epoch": 368, "lr": 1.6054522194613264e-05} +{"train_loss": 0.013225135393440723, "grad_norm": 0.0896141305565834, "global_step": 39090, "epoch": 368, "lr": 1.60523461918326e-05} +{"train_loss": 0.015084320679306984, "grad_norm": 0.15575508773326874, "global_step": 39091, "epoch": 368, "lr": 1.6050170308329214e-05} +{"train_loss": 0.012389053590595722, "grad_norm": 0.059267688542604446, "global_step": 39092, "epoch": 368, "lr": 1.6047994544110767e-05} +{"train_loss": 0.014344356954097748, "grad_norm": 0.06474930793046951, "global_step": 39093, "epoch": 368, "lr": 1.6045818899184916e-05} +{"train_loss": 0.014014371670782566, "grad_norm": 0.06673422455787659, "global_step": 39094, "epoch": 368, "lr": 1.6043643373559298e-05} +{"train_loss": 0.014587410725653172, "grad_norm": 0.055408112704753876, "global_step": 39095, "epoch": 368, "lr": 1.6041467967241524e-05} +{"train_loss": 0.0142313027754426, "grad_norm": 0.07232014834880829, "global_step": 39096, "epoch": 368, "lr": 1.603929268023927e-05} +{"train_loss": 0.014091910794377327, "grad_norm": 0.08267508447170258, "global_step": 39097, "epoch": 368, "lr": 1.6037117512560185e-05} +{"train_loss": 0.01719384267926216, "grad_norm": 0.12160561978816986, "global_step": 39098, "epoch": 368, "lr": 1.6034942464211883e-05} +{"train_loss": 0.014938081614673138, "grad_norm": 0.1007281169295311, "global_step": 39099, "epoch": 368, "lr": 1.6032767535202043e-05} +{"train_loss": 0.013710010796785355, "grad_norm": 0.10298960655927658, "global_step": 39100, "epoch": 368, "lr": 1.603059272553826e-05} +{"train_loss": 0.014960074797272682, "grad_norm": 0.13798773288726807, "global_step": 39101, "epoch": 368, "lr": 1.6028418035228225e-05} +{"train_loss": 0.012156054377555847, "grad_norm": 0.0727323368191719, "global_step": 39102, "epoch": 368, "lr": 1.6026243464279538e-05} +{"train_loss": 0.014947079122066498, "grad_norm": 0.055501531809568405, "global_step": 39103, "epoch": 368, "lr": 1.6024069012699856e-05} +{"train_loss": 0.012821191921830177, "grad_norm": 0.13997751474380493, "global_step": 39104, "epoch": 368, "lr": 1.6021894680496834e-05} +{"train_loss": 0.014823048375546932, "grad_norm": 0.130185067653656, "global_step": 39105, "epoch": 368, "lr": 1.60197204676781e-05} +{"train_loss": 0.01791655272245407, "grad_norm": 0.15915611386299133, "global_step": 39106, "epoch": 368, "lr": 1.6017546374251275e-05} +{"train_loss": 0.01835257187485695, "grad_norm": 0.1306770294904709, "global_step": 39107, "epoch": 368, "lr": 1.6015372400224026e-05} +{"train_loss": 0.017850704491138458, "grad_norm": 0.09266628324985504, "global_step": 39108, "epoch": 368, "lr": 1.601319854560397e-05} +{"train_loss": 0.013991768471896648, "grad_norm": 0.18401256203651428, "global_step": 39109, "epoch": 368, "lr": 1.601102481039875e-05} +{"train_loss": 0.014952494762837887, "grad_norm": 0.068943552672863, "global_step": 39110, "epoch": 368, "lr": 1.600885119461602e-05} +{"train_loss": 0.015745103359222412, "grad_norm": 0.08664769679307938, "global_step": 39111, "epoch": 368, "lr": 1.6006677698263405e-05} +{"train_loss": 0.016941383481025696, "grad_norm": 0.06432559341192245, "global_step": 39112, "epoch": 368, "lr": 1.600450432134853e-05} +{"train_loss": 0.015072682608832728, "grad_norm": 0.130655899643898, "global_step": 39113, "epoch": 368, "lr": 1.600233106387904e-05, "val_loss": 0.019686292856931686} +{"train_loss": 0.01383668091148138, "grad_norm": 0.08718791604042053, "global_step": 39114, "epoch": 369, "lr": 1.6000157925862592e-05} +{"train_loss": 0.018550898879766464, "grad_norm": 0.14836452901363373, "global_step": 39115, "epoch": 369, "lr": 1.599798490730678e-05} +{"train_loss": 0.014162307605147362, "grad_norm": 0.07454174757003784, "global_step": 39116, "epoch": 369, "lr": 1.599581200821927e-05} +{"train_loss": 0.015225982293486595, "grad_norm": 0.06815487891435623, "global_step": 39117, "epoch": 369, "lr": 1.59936392286077e-05} +{"train_loss": 0.013563133776187897, "grad_norm": 0.07566496729850769, "global_step": 39118, "epoch": 369, "lr": 1.599146656847969e-05} +{"train_loss": 0.01590678095817566, "grad_norm": 0.09231431782245636, "global_step": 39119, "epoch": 369, "lr": 1.5989294027842867e-05} +{"train_loss": 0.01574399694800377, "grad_norm": 0.07684537768363953, "global_step": 39120, "epoch": 369, "lr": 1.5987121606704867e-05} +{"train_loss": 0.014240585267543793, "grad_norm": 0.07865238934755325, "global_step": 39121, "epoch": 369, "lr": 1.5984949305073354e-05} +{"train_loss": 0.016552971675992012, "grad_norm": 0.13344542682170868, "global_step": 39122, "epoch": 369, "lr": 1.5982777122955915e-05} +{"train_loss": 0.01597045361995697, "grad_norm": 0.07180032879114151, "global_step": 39123, "epoch": 369, "lr": 1.598060506036022e-05} +{"train_loss": 0.01683405414223671, "grad_norm": 0.06189775466918945, "global_step": 39124, "epoch": 369, "lr": 1.5978433117293883e-05} +{"train_loss": 0.014228340238332748, "grad_norm": 0.09352563321590424, "global_step": 39125, "epoch": 369, "lr": 1.5976261293764526e-05} +{"train_loss": 0.014211990870535374, "grad_norm": 0.08100226521492004, "global_step": 39126, "epoch": 369, "lr": 1.5974089589779785e-05} +{"train_loss": 0.01431342028081417, "grad_norm": 0.06007454916834831, "global_step": 39127, "epoch": 369, "lr": 1.5971918005347313e-05} +{"train_loss": 0.017075851559638977, "grad_norm": 0.1125616580247879, "global_step": 39128, "epoch": 369, "lr": 1.5969746540474705e-05} +{"train_loss": 0.015759063884615898, "grad_norm": 0.061723027378320694, "global_step": 39129, "epoch": 369, "lr": 1.5967575195169626e-05} +{"train_loss": 0.015515308827161789, "grad_norm": 0.12459364533424377, "global_step": 39130, "epoch": 369, "lr": 1.5965403969439674e-05} +{"train_loss": 0.015603492967784405, "grad_norm": 0.09107984602451324, "global_step": 39131, "epoch": 369, "lr": 1.5963232863292504e-05} +{"train_loss": 0.015939269214868546, "grad_norm": 0.06861988455057144, "global_step": 39132, "epoch": 369, "lr": 1.5961061876735718e-05} +{"train_loss": 0.015235370956361294, "grad_norm": 0.0687732920050621, "global_step": 39133, "epoch": 369, "lr": 1.5958891009776955e-05} +{"train_loss": 0.017534423619508743, "grad_norm": 0.06587184965610504, "global_step": 39134, "epoch": 369, "lr": 1.5956720262423862e-05} +{"train_loss": 0.014468556270003319, "grad_norm": 0.0947403833270073, "global_step": 39135, "epoch": 369, "lr": 1.5954549634684053e-05} +{"train_loss": 0.014025717973709106, "grad_norm": 0.05885016918182373, "global_step": 39136, "epoch": 369, "lr": 1.5952379126565126e-05} +{"train_loss": 0.015799321234226227, "grad_norm": 0.06846613436937332, "global_step": 39137, "epoch": 369, "lr": 1.5950208738074736e-05} +{"train_loss": 0.01420004665851593, "grad_norm": 0.06734348088502884, "global_step": 39138, "epoch": 369, "lr": 1.5948038469220523e-05} +{"train_loss": 0.012484692968428135, "grad_norm": 0.07730664312839508, "global_step": 39139, "epoch": 369, "lr": 1.5945868320010076e-05} +{"train_loss": 0.01575767621397972, "grad_norm": 0.08331597596406937, "global_step": 39140, "epoch": 369, "lr": 1.5943698290451036e-05} +{"train_loss": 0.01636516861617565, "grad_norm": 0.12545143067836761, "global_step": 39141, "epoch": 369, "lr": 1.5941528380551062e-05} +{"train_loss": 0.015307242050766945, "grad_norm": 0.1267160177230835, "global_step": 39142, "epoch": 369, "lr": 1.5939358590317715e-05} +{"train_loss": 0.016787562519311905, "grad_norm": 0.12794797122478485, "global_step": 39143, "epoch": 369, "lr": 1.5937188919758648e-05} +{"train_loss": 0.015783919021487236, "grad_norm": 0.05402328446507454, "global_step": 39144, "epoch": 369, "lr": 1.5935019368881504e-05} +{"train_loss": 0.015564991161227226, "grad_norm": 0.06363604962825775, "global_step": 39145, "epoch": 369, "lr": 1.5932849937693874e-05} +{"train_loss": 0.01569884642958641, "grad_norm": 0.08307678252458572, "global_step": 39146, "epoch": 369, "lr": 1.5930680626203403e-05} +{"train_loss": 0.015179354697465897, "grad_norm": 0.0951833501458168, "global_step": 39147, "epoch": 369, "lr": 1.592851143441769e-05} +{"train_loss": 0.014042165130376816, "grad_norm": 0.09371812641620636, "global_step": 39148, "epoch": 369, "lr": 1.5926342362344393e-05} +{"train_loss": 0.014469890855252743, "grad_norm": 0.09105212986469269, "global_step": 39149, "epoch": 369, "lr": 1.5924173409991087e-05} +{"train_loss": 0.014570568688213825, "grad_norm": 0.1053909882903099, "global_step": 39150, "epoch": 369, "lr": 1.5922004577365418e-05} +{"train_loss": 0.014683338813483715, "grad_norm": 0.049043022096157074, "global_step": 39151, "epoch": 369, "lr": 1.5919835864475025e-05} +{"train_loss": 0.012110952287912369, "grad_norm": 0.11823535710573196, "global_step": 39152, "epoch": 369, "lr": 1.59176672713275e-05} +{"train_loss": 0.015413534827530384, "grad_norm": 0.05993415787816048, "global_step": 39153, "epoch": 369, "lr": 1.5915498797930462e-05} +{"train_loss": 0.011139193549752235, "grad_norm": 0.09260191023349762, "global_step": 39154, "epoch": 369, "lr": 1.5913330444291537e-05} +{"train_loss": 0.015139623545110226, "grad_norm": 0.08684998005628586, "global_step": 39155, "epoch": 369, "lr": 1.5911162210418357e-05} +{"train_loss": 0.013525835238397121, "grad_norm": 0.06184957176446915, "global_step": 39156, "epoch": 369, "lr": 1.5908994096318515e-05} +{"train_loss": 0.015911690890789032, "grad_norm": 0.08263743668794632, "global_step": 39157, "epoch": 369, "lr": 1.590682610199964e-05} +{"train_loss": 0.011159414425492287, "grad_norm": 0.06808629631996155, "global_step": 39158, "epoch": 369, "lr": 1.5904658227469365e-05} +{"train_loss": 0.014661978930234909, "grad_norm": 0.12843069434165955, "global_step": 39159, "epoch": 369, "lr": 1.59024904727353e-05} +{"train_loss": 0.014757736586034298, "grad_norm": 0.05789068713784218, "global_step": 39160, "epoch": 369, "lr": 1.5900322837805033e-05} +{"train_loss": 0.014489050954580307, "grad_norm": 0.0784495547413826, "global_step": 39161, "epoch": 369, "lr": 1.5898155322686203e-05} +{"train_loss": 0.01552879624068737, "grad_norm": 0.10437532514333725, "global_step": 39162, "epoch": 369, "lr": 1.5895987927386446e-05} +{"train_loss": 0.014990500174462795, "grad_norm": 0.10512004047632217, "global_step": 39163, "epoch": 369, "lr": 1.5893820651913333e-05} +{"train_loss": 0.01470954343676567, "grad_norm": 0.08895393460988998, "global_step": 39164, "epoch": 369, "lr": 1.5891653496274524e-05} +{"train_loss": 0.014255051501095295, "grad_norm": 0.07550925761461258, "global_step": 39165, "epoch": 369, "lr": 1.588948646047761e-05} +{"train_loss": 0.016682449728250504, "grad_norm": 0.10834459960460663, "global_step": 39166, "epoch": 369, "lr": 1.5887319544530182e-05} +{"train_loss": 0.015012349002063274, "grad_norm": 0.07569935917854309, "global_step": 39167, "epoch": 369, "lr": 1.5885152748439885e-05} +{"train_loss": 0.015354948118329048, "grad_norm": 0.06698582321405411, "global_step": 39168, "epoch": 369, "lr": 1.5882986072214345e-05} +{"train_loss": 0.014807921834290028, "grad_norm": 0.07291098684072495, "global_step": 39169, "epoch": 369, "lr": 1.5880819515861135e-05} +{"train_loss": 0.015451676212251186, "grad_norm": 0.14344210922718048, "global_step": 39170, "epoch": 369, "lr": 1.58786530793879e-05} +{"train_loss": 0.014518879354000092, "grad_norm": 0.07212644815444946, "global_step": 39171, "epoch": 369, "lr": 1.5876486762802224e-05} +{"train_loss": 0.014676662161946297, "grad_norm": 0.1213163211941719, "global_step": 39172, "epoch": 369, "lr": 1.5874320566111756e-05} +{"train_loss": 0.013230936601758003, "grad_norm": 0.056244075298309326, "global_step": 39173, "epoch": 369, "lr": 1.587215448932406e-05} +{"train_loss": 0.014340430498123169, "grad_norm": 0.07636779546737671, "global_step": 39174, "epoch": 369, "lr": 1.5869988532446777e-05} +{"train_loss": 0.016047660261392593, "grad_norm": 0.068165123462677, "global_step": 39175, "epoch": 369, "lr": 1.5867822695487522e-05} +{"train_loss": 0.01587088778614998, "grad_norm": 0.11190706491470337, "global_step": 39176, "epoch": 369, "lr": 1.5865656978453903e-05} +{"train_loss": 0.014284854754805565, "grad_norm": 0.08821883052587509, "global_step": 39177, "epoch": 369, "lr": 1.5863491381353495e-05} +{"train_loss": 0.01620480790734291, "grad_norm": 0.09336283057928085, "global_step": 39178, "epoch": 369, "lr": 1.586132590419394e-05} +{"train_loss": 0.01444864273071289, "grad_norm": 0.11568382382392883, "global_step": 39179, "epoch": 369, "lr": 1.5859160546982855e-05} +{"train_loss": 0.013675950467586517, "grad_norm": 0.08650801330804825, "global_step": 39180, "epoch": 369, "lr": 1.5856995309727813e-05} +{"train_loss": 0.013683026656508446, "grad_norm": 0.059269979596138, "global_step": 39181, "epoch": 369, "lr": 1.5854830192436444e-05} +{"train_loss": 0.01727887988090515, "grad_norm": 0.08122556656599045, "global_step": 39182, "epoch": 369, "lr": 1.585266519511638e-05} +{"train_loss": 0.016656044870615005, "grad_norm": 0.11896184086799622, "global_step": 39183, "epoch": 369, "lr": 1.5850500317775168e-05} +{"train_loss": 0.014275433495640755, "grad_norm": 0.06891313195228577, "global_step": 39184, "epoch": 369, "lr": 1.584833556042045e-05} +{"train_loss": 0.01616656221449375, "grad_norm": 0.10136675089597702, "global_step": 39185, "epoch": 369, "lr": 1.5846170923059845e-05} +{"train_loss": 0.013972360640764236, "grad_norm": 0.09500959515571594, "global_step": 39186, "epoch": 369, "lr": 1.5844006405700927e-05} +{"train_loss": 0.01683793216943741, "grad_norm": 0.06996850669384003, "global_step": 39187, "epoch": 369, "lr": 1.5841842008351314e-05} +{"train_loss": 0.015191435813903809, "grad_norm": 0.08337262272834778, "global_step": 39188, "epoch": 369, "lr": 1.5839677731018633e-05} +{"train_loss": 0.015753544867038727, "grad_norm": 0.07948143780231476, "global_step": 39189, "epoch": 369, "lr": 1.5837513573710466e-05} +{"train_loss": 0.01361631415784359, "grad_norm": 0.11537216603755951, "global_step": 39190, "epoch": 369, "lr": 1.5835349536434403e-05} +{"train_loss": 0.01389298029243946, "grad_norm": 0.11682838201522827, "global_step": 39191, "epoch": 369, "lr": 1.583318561919806e-05} +{"train_loss": 0.01552940160036087, "grad_norm": 0.06772841513156891, "global_step": 39192, "epoch": 369, "lr": 1.583102182200907e-05} +{"train_loss": 0.015538453124463558, "grad_norm": 0.0685293972492218, "global_step": 39193, "epoch": 369, "lr": 1.5828858144874996e-05} +{"train_loss": 0.013342257589101791, "grad_norm": 0.1152813732624054, "global_step": 39194, "epoch": 369, "lr": 1.582669458780345e-05} +{"train_loss": 0.015575193800032139, "grad_norm": 0.08979149907827377, "global_step": 39195, "epoch": 369, "lr": 1.582453115080203e-05} +{"train_loss": 0.014139432460069656, "grad_norm": 0.09053413569927216, "global_step": 39196, "epoch": 369, "lr": 1.5822367833878366e-05} +{"train_loss": 0.013768880628049374, "grad_norm": 0.0803174078464508, "global_step": 39197, "epoch": 369, "lr": 1.582020463704002e-05} +{"train_loss": 0.01647287793457508, "grad_norm": 0.06217329949140549, "global_step": 39198, "epoch": 369, "lr": 1.581804156029461e-05} +{"train_loss": 0.016583148390054703, "grad_norm": 0.12506511807441711, "global_step": 39199, "epoch": 369, "lr": 1.581587860364977e-05} +{"train_loss": 0.014068933203816414, "grad_norm": 0.06583826243877411, "global_step": 39200, "epoch": 369, "lr": 1.5813715767113036e-05} +{"train_loss": 0.0145345414057374, "grad_norm": 0.0576002262532711, "global_step": 39201, "epoch": 369, "lr": 1.581155305069203e-05} +{"train_loss": 0.014915602281689644, "grad_norm": 0.12886856496334076, "global_step": 39202, "epoch": 369, "lr": 1.580939045439437e-05} +{"train_loss": 0.01692471280694008, "grad_norm": 0.12154950201511383, "global_step": 39203, "epoch": 369, "lr": 1.5807227978227645e-05} +{"train_loss": 0.01789899170398712, "grad_norm": 0.08136427402496338, "global_step": 39204, "epoch": 369, "lr": 1.5805065622199445e-05} +{"train_loss": 0.013759465888142586, "grad_norm": 0.08625436574220657, "global_step": 39205, "epoch": 369, "lr": 1.5802903386317384e-05} +{"train_loss": 0.01387157291173935, "grad_norm": 0.09159829467535019, "global_step": 39206, "epoch": 369, "lr": 1.5800741270589054e-05} +{"train_loss": 0.014753911644220352, "grad_norm": 0.07288775593042374, "global_step": 39207, "epoch": 369, "lr": 1.579857927502203e-05} +{"train_loss": 0.01572481170296669, "grad_norm": 0.15101881325244904, "global_step": 39208, "epoch": 369, "lr": 1.579641739962392e-05} +{"train_loss": 0.016779158264398575, "grad_norm": 0.07707805931568146, "global_step": 39209, "epoch": 369, "lr": 1.5794255644402343e-05} +{"train_loss": 0.017151817679405212, "grad_norm": 0.09386123716831207, "global_step": 39210, "epoch": 369, "lr": 1.5792094009364862e-05} +{"train_loss": 0.01457250863313675, "grad_norm": 0.0908743366599083, "global_step": 39211, "epoch": 369, "lr": 1.5789932494519095e-05} +{"train_loss": 0.013673270121216774, "grad_norm": 0.08373025804758072, "global_step": 39212, "epoch": 369, "lr": 1.5787771099872617e-05} +{"train_loss": 0.012264734134078026, "grad_norm": 0.09114623069763184, "global_step": 39213, "epoch": 369, "lr": 1.578560982543305e-05} +{"train_loss": 0.013545135036110878, "grad_norm": 0.0662417858839035, "global_step": 39214, "epoch": 369, "lr": 1.5783448671207952e-05} +{"train_loss": 0.015935465693473816, "grad_norm": 0.07748246937990189, "global_step": 39215, "epoch": 369, "lr": 1.5781287637204934e-05} +{"train_loss": 0.014543266966938972, "grad_norm": 0.05660821869969368, "global_step": 39216, "epoch": 369, "lr": 1.577912672343161e-05} +{"train_loss": 0.01643429696559906, "grad_norm": 0.06917601823806763, "global_step": 39217, "epoch": 369, "lr": 1.5776965929895547e-05} +{"train_loss": 0.018236463889479637, "grad_norm": 0.06751559674739838, "global_step": 39218, "epoch": 369, "lr": 1.577480525660433e-05} +{"train_loss": 0.015058438188203101, "grad_norm": 0.12173862010240555, "global_step": 39219, "epoch": 369, "lr": 1.5772644703565565e-05, "val_loss": 0.019370215013623238} +{"train_loss": 0.013742747716605663, "grad_norm": 0.07337596267461777, "global_step": 39220, "epoch": 370, "lr": 1.5770484270786855e-05} +{"train_loss": 0.01673160493373871, "grad_norm": 0.09916645288467407, "global_step": 39221, "epoch": 370, "lr": 1.5768323958275762e-05} +{"train_loss": 0.015370676293969154, "grad_norm": 0.0562426894903183, "global_step": 39222, "epoch": 370, "lr": 1.5766163766039888e-05} +{"train_loss": 0.016591038554906845, "grad_norm": 0.12751804292201996, "global_step": 39223, "epoch": 370, "lr": 1.5764003694086866e-05} +{"train_loss": 0.01389501802623272, "grad_norm": 0.1287015974521637, "global_step": 39224, "epoch": 370, "lr": 1.576184374242421e-05} +{"train_loss": 0.015281415544450283, "grad_norm": 0.07766040414571762, "global_step": 39225, "epoch": 370, "lr": 1.5759683911059542e-05} +{"train_loss": 0.013995454646646976, "grad_norm": 0.0660792738199234, "global_step": 39226, "epoch": 370, "lr": 1.5757524200000473e-05} +{"train_loss": 0.01444164291024208, "grad_norm": 0.05921093747019768, "global_step": 39227, "epoch": 370, "lr": 1.575536460925456e-05} +{"train_loss": 0.012873895466327667, "grad_norm": 0.0692233070731163, "global_step": 39228, "epoch": 370, "lr": 1.5753205138829395e-05} +{"train_loss": 0.01524185948073864, "grad_norm": 0.06474832445383072, "global_step": 39229, "epoch": 370, "lr": 1.5751045788732593e-05} +{"train_loss": 0.012858645990490913, "grad_norm": 0.07651827484369278, "global_step": 39230, "epoch": 370, "lr": 1.5748886558971725e-05} +{"train_loss": 0.015105592086911201, "grad_norm": 0.07359343767166138, "global_step": 39231, "epoch": 370, "lr": 1.5746727449554355e-05} +{"train_loss": 0.014726709574460983, "grad_norm": 0.09511303156614304, "global_step": 39232, "epoch": 370, "lr": 1.574456846048809e-05} +{"train_loss": 0.015916330739855766, "grad_norm": 0.07039022445678711, "global_step": 39233, "epoch": 370, "lr": 1.5742409591780532e-05} +{"train_loss": 0.0135884340852499, "grad_norm": 0.07771068811416626, "global_step": 39234, "epoch": 370, "lr": 1.574025084343923e-05} +{"train_loss": 0.01332828402519226, "grad_norm": 0.09402197599411011, "global_step": 39235, "epoch": 370, "lr": 1.57380922154718e-05} +{"train_loss": 0.013259527273476124, "grad_norm": 0.12822934985160828, "global_step": 39236, "epoch": 370, "lr": 1.5735933707885803e-05} +{"train_loss": 0.015534888952970505, "grad_norm": 0.09123705327510834, "global_step": 39237, "epoch": 370, "lr": 1.573377532068885e-05} +{"train_loss": 0.01817040704190731, "grad_norm": 0.08411502838134766, "global_step": 39238, "epoch": 370, "lr": 1.5731617053888494e-05} +{"train_loss": 0.016401086002588272, "grad_norm": 0.07947442680597305, "global_step": 39239, "epoch": 370, "lr": 1.5729458907492327e-05} +{"train_loss": 0.016346385702490807, "grad_norm": 0.08521074801683426, "global_step": 39240, "epoch": 370, "lr": 1.5727300881507955e-05} +{"train_loss": 0.01347924955189228, "grad_norm": 0.0994783267378807, "global_step": 39241, "epoch": 370, "lr": 1.572514297594294e-05} +{"train_loss": 0.014452437870204449, "grad_norm": 0.065497025847435, "global_step": 39242, "epoch": 370, "lr": 1.5722985190804855e-05} +{"train_loss": 0.015183614566922188, "grad_norm": 0.06525811553001404, "global_step": 39243, "epoch": 370, "lr": 1.5720827526101312e-05} +{"train_loss": 0.01457304134964943, "grad_norm": 0.1233452707529068, "global_step": 39244, "epoch": 370, "lr": 1.5718669981839858e-05} +{"train_loss": 0.014956359751522541, "grad_norm": 0.06632405519485474, "global_step": 39245, "epoch": 370, "lr": 1.5716512558028084e-05} +{"train_loss": 0.01593800075352192, "grad_norm": 0.1145017221570015, "global_step": 39246, "epoch": 370, "lr": 1.571435525467359e-05} +{"train_loss": 0.016738178208470345, "grad_norm": 0.09885980188846588, "global_step": 39247, "epoch": 370, "lr": 1.5712198071783945e-05} +{"train_loss": 0.01455529872328043, "grad_norm": 0.06092299893498421, "global_step": 39248, "epoch": 370, "lr": 1.571004100936671e-05} +{"train_loss": 0.015712430700659752, "grad_norm": 0.09270158410072327, "global_step": 39249, "epoch": 370, "lr": 1.5707884067429473e-05} +{"train_loss": 0.01467517577111721, "grad_norm": 0.09625241160392761, "global_step": 39250, "epoch": 370, "lr": 1.570572724597984e-05} +{"train_loss": 0.014556355774402618, "grad_norm": 0.08634073287248611, "global_step": 39251, "epoch": 370, "lr": 1.5703570545025347e-05} +{"train_loss": 0.013020405545830727, "grad_norm": 0.07610810548067093, "global_step": 39252, "epoch": 370, "lr": 1.5701413964573613e-05} +{"train_loss": 0.018622586503624916, "grad_norm": 0.11459361016750336, "global_step": 39253, "epoch": 370, "lr": 1.5699257504632177e-05} +{"train_loss": 0.012667917646467686, "grad_norm": 0.06951331347227097, "global_step": 39254, "epoch": 370, "lr": 1.5697101165208645e-05} +{"train_loss": 0.01608636975288391, "grad_norm": 0.0933426097035408, "global_step": 39255, "epoch": 370, "lr": 1.569494494631057e-05} +{"train_loss": 0.017323460429906845, "grad_norm": 0.07989226281642914, "global_step": 39256, "epoch": 370, "lr": 1.5692788847945534e-05} +{"train_loss": 0.01567794382572174, "grad_norm": 0.15205071866512299, "global_step": 39257, "epoch": 370, "lr": 1.5690632870121137e-05} +{"train_loss": 0.014160780236124992, "grad_norm": 0.17191116511821747, "global_step": 39258, "epoch": 370, "lr": 1.5688477012844937e-05} +{"train_loss": 0.015477732755243778, "grad_norm": 0.06675327569246292, "global_step": 39259, "epoch": 370, "lr": 1.5686321276124487e-05} +{"train_loss": 0.01271096058189869, "grad_norm": 0.05447598919272423, "global_step": 39260, "epoch": 370, "lr": 1.5684165659967387e-05} +{"train_loss": 0.01462164893746376, "grad_norm": 0.12760964035987854, "global_step": 39261, "epoch": 370, "lr": 1.568201016438122e-05} +{"train_loss": 0.015475569292902946, "grad_norm": 0.09375710040330887, "global_step": 39262, "epoch": 370, "lr": 1.5679854789373527e-05} +{"train_loss": 0.015950046479701996, "grad_norm": 0.061690106987953186, "global_step": 39263, "epoch": 370, "lr": 1.56776995349519e-05} +{"train_loss": 0.0150785306468606, "grad_norm": 0.07712909579277039, "global_step": 39264, "epoch": 370, "lr": 1.5675544401123944e-05} +{"train_loss": 0.01771269366145134, "grad_norm": 0.07806064933538437, "global_step": 39265, "epoch": 370, "lr": 1.5673389387897164e-05} +{"train_loss": 0.014605861157178879, "grad_norm": 0.06861602514982224, "global_step": 39266, "epoch": 370, "lr": 1.5671234495279165e-05} +{"train_loss": 0.014850249513983727, "grad_norm": 0.08903833478689194, "global_step": 39267, "epoch": 370, "lr": 1.566907972327753e-05} +{"train_loss": 0.015203703194856644, "grad_norm": 0.06011459231376648, "global_step": 39268, "epoch": 370, "lr": 1.5666925071899808e-05} +{"train_loss": 0.012932735495269299, "grad_norm": 0.07470743358135223, "global_step": 39269, "epoch": 370, "lr": 1.566477054115358e-05} +{"train_loss": 0.015019062906503677, "grad_norm": 0.1088661402463913, "global_step": 39270, "epoch": 370, "lr": 1.566261613104643e-05} +{"train_loss": 0.011064345017075539, "grad_norm": 0.05069456249475479, "global_step": 39271, "epoch": 370, "lr": 1.5660461841585923e-05} +{"train_loss": 0.015916597098112106, "grad_norm": 0.06696238368749619, "global_step": 39272, "epoch": 370, "lr": 1.5658307672779593e-05} +{"train_loss": 0.014035675674676895, "grad_norm": 0.06800024211406708, "global_step": 39273, "epoch": 370, "lr": 1.565615362463504e-05} +{"train_loss": 0.014863462187349796, "grad_norm": 0.09729015082120895, "global_step": 39274, "epoch": 370, "lr": 1.565399969715984e-05} +{"train_loss": 0.014326569624245167, "grad_norm": 0.06466852873563766, "global_step": 39275, "epoch": 370, "lr": 1.5651845890361533e-05} +{"train_loss": 0.015437101945281029, "grad_norm": 0.05536274611949921, "global_step": 39276, "epoch": 370, "lr": 1.564969220424772e-05} +{"train_loss": 0.016458911821246147, "grad_norm": 0.07704132050275803, "global_step": 39277, "epoch": 370, "lr": 1.5647538638825933e-05} +{"train_loss": 0.017962373793125153, "grad_norm": 0.0773981437087059, "global_step": 39278, "epoch": 370, "lr": 1.5645385194103773e-05} +{"train_loss": 0.017136504873633385, "grad_norm": 0.07559779286384583, "global_step": 39279, "epoch": 370, "lr": 1.5643231870088772e-05} +{"train_loss": 0.01588393561542034, "grad_norm": 0.10225888341665268, "global_step": 39280, "epoch": 370, "lr": 1.5641078666788512e-05} +{"train_loss": 0.01697302609682083, "grad_norm": 0.11328759044408798, "global_step": 39281, "epoch": 370, "lr": 1.5638925584210572e-05} +{"train_loss": 0.01251586526632309, "grad_norm": 0.08578161895275116, "global_step": 39282, "epoch": 370, "lr": 1.5636772622362512e-05} +{"train_loss": 0.01377335749566555, "grad_norm": 0.13496823608875275, "global_step": 39283, "epoch": 370, "lr": 1.563461978125187e-05} +{"train_loss": 0.016379937529563904, "grad_norm": 0.19595587253570557, "global_step": 39284, "epoch": 370, "lr": 1.5632467060886242e-05} +{"train_loss": 0.013532719574868679, "grad_norm": 0.0670158788561821, "global_step": 39285, "epoch": 370, "lr": 1.5630314461273167e-05} +{"train_loss": 0.014727081172168255, "grad_norm": 0.08124431222677231, "global_step": 39286, "epoch": 370, "lr": 1.5628161982420214e-05} +{"train_loss": 0.01448172889649868, "grad_norm": 0.06841550022363663, "global_step": 39287, "epoch": 370, "lr": 1.5626009624334975e-05} +{"train_loss": 0.015431889332830906, "grad_norm": 0.07483752816915512, "global_step": 39288, "epoch": 370, "lr": 1.562385738702498e-05} +{"train_loss": 0.01617565006017685, "grad_norm": 0.08532407879829407, "global_step": 39289, "epoch": 370, "lr": 1.5621705270497793e-05} +{"train_loss": 0.012728718109428883, "grad_norm": 0.06864757090806961, "global_step": 39290, "epoch": 370, "lr": 1.5619553274760978e-05} +{"train_loss": 0.012528683990240097, "grad_norm": 0.06354895979166031, "global_step": 39291, "epoch": 370, "lr": 1.5617401399822123e-05} +{"train_loss": 0.015627572312951088, "grad_norm": 0.07247411459684372, "global_step": 39292, "epoch": 370, "lr": 1.5615249645688747e-05} +{"train_loss": 0.014606280252337456, "grad_norm": 0.05901310592889786, "global_step": 39293, "epoch": 370, "lr": 1.5613098012368425e-05} +{"train_loss": 0.01392290461808443, "grad_norm": 0.1284235566854477, "global_step": 39294, "epoch": 370, "lr": 1.561094649986875e-05} +{"train_loss": 0.01637953519821167, "grad_norm": 0.15103751420974731, "global_step": 39295, "epoch": 370, "lr": 1.5608795108197243e-05} +{"train_loss": 0.017559899017214775, "grad_norm": 0.14567765593528748, "global_step": 39296, "epoch": 370, "lr": 1.5606643837361462e-05} +{"train_loss": 0.012922700494527817, "grad_norm": 0.06350353360176086, "global_step": 39297, "epoch": 370, "lr": 1.5604492687368976e-05} +{"train_loss": 0.01721714437007904, "grad_norm": 0.09197044372558594, "global_step": 39298, "epoch": 370, "lr": 1.5602341658227365e-05} +{"train_loss": 0.014602371491491795, "grad_norm": 0.06561825424432755, "global_step": 39299, "epoch": 370, "lr": 1.560019074994416e-05} +{"train_loss": 0.015150900930166245, "grad_norm": 0.06223250553011894, "global_step": 39300, "epoch": 370, "lr": 1.559803996252691e-05} +{"train_loss": 0.014840934425592422, "grad_norm": 0.08673646301031113, "global_step": 39301, "epoch": 370, "lr": 1.5595889295983206e-05} +{"train_loss": 0.01654321514070034, "grad_norm": 0.08097444474697113, "global_step": 39302, "epoch": 370, "lr": 1.559373875032057e-05} +{"train_loss": 0.012069483287632465, "grad_norm": 0.09217292070388794, "global_step": 39303, "epoch": 370, "lr": 1.5591588325546568e-05} +{"train_loss": 0.015744389966130257, "grad_norm": 0.07312707602977753, "global_step": 39304, "epoch": 370, "lr": 1.5589438021668777e-05} +{"train_loss": 0.015187889337539673, "grad_norm": 0.06626641750335693, "global_step": 39305, "epoch": 370, "lr": 1.558728783869474e-05} +{"train_loss": 0.014856071211397648, "grad_norm": 0.08160404115915298, "global_step": 39306, "epoch": 370, "lr": 1.5585137776631992e-05} +{"train_loss": 0.013604329898953438, "grad_norm": 0.10852645337581635, "global_step": 39307, "epoch": 370, "lr": 1.55829878354881e-05} +{"train_loss": 0.01671435870230198, "grad_norm": 0.0973411425948143, "global_step": 39308, "epoch": 370, "lr": 1.5580838015270638e-05} +{"train_loss": 0.01423853449523449, "grad_norm": 0.0752129927277565, "global_step": 39309, "epoch": 370, "lr": 1.557868831598713e-05} +{"train_loss": 0.015065749175846577, "grad_norm": 0.08463596552610397, "global_step": 39310, "epoch": 370, "lr": 1.5576538737645137e-05} +{"train_loss": 0.014781590551137924, "grad_norm": 0.1219283789396286, "global_step": 39311, "epoch": 370, "lr": 1.5574389280252232e-05} +{"train_loss": 0.015133114531636238, "grad_norm": 0.0667945072054863, "global_step": 39312, "epoch": 370, "lr": 1.557223994381595e-05} +{"train_loss": 0.015624010935425758, "grad_norm": 0.07090414315462112, "global_step": 39313, "epoch": 370, "lr": 1.557009072834384e-05} +{"train_loss": 0.014936837367713451, "grad_norm": 0.10739658027887344, "global_step": 39314, "epoch": 370, "lr": 1.5567941633843448e-05} +{"train_loss": 0.016529656946659088, "grad_norm": 0.07126015424728394, "global_step": 39315, "epoch": 370, "lr": 1.5565792660322353e-05} +{"train_loss": 0.01381128653883934, "grad_norm": 0.0539451539516449, "global_step": 39316, "epoch": 370, "lr": 1.5563643807788077e-05} +{"train_loss": 0.01614122837781906, "grad_norm": 0.06903936713933945, "global_step": 39317, "epoch": 370, "lr": 1.5561495076248188e-05} +{"train_loss": 0.015042527578771114, "grad_norm": 0.11008130013942719, "global_step": 39318, "epoch": 370, "lr": 1.5559346465710218e-05} +{"train_loss": 0.014120385982096195, "grad_norm": 0.07051597535610199, "global_step": 39319, "epoch": 370, "lr": 1.5557197976181744e-05} +{"train_loss": 0.013588898815214634, "grad_norm": 0.08368320763111115, "global_step": 39320, "epoch": 370, "lr": 1.5555049607670276e-05} +{"train_loss": 0.017573382705450058, "grad_norm": 0.07063225656747818, "global_step": 39321, "epoch": 370, "lr": 1.5552901360183387e-05} +{"train_loss": 0.016263514757156372, "grad_norm": 0.1698155552148819, "global_step": 39322, "epoch": 370, "lr": 1.5550753233728637e-05} +{"train_loss": 0.016165081411600113, "grad_norm": 0.06369050592184067, "global_step": 39323, "epoch": 370, "lr": 1.5548605228313556e-05} +{"train_loss": 0.015443596057593822, "grad_norm": 0.1206662654876709, "global_step": 39324, "epoch": 370, "lr": 1.554645734394568e-05} +{"train_loss": 0.015052006096702139, "grad_norm": 0.10774245113134384, "global_step": 39325, "epoch": 370, "lr": 1.554430958063259e-05, "train_batch_mse_error": 0.013495473191142082, "val_loss": 0.019818812608718872} +{"train_loss": 0.013630175963044167, "grad_norm": 0.1111971065402031, "global_step": 39326, "epoch": 371, "lr": 1.554216193838179e-05} +{"train_loss": 0.013504636473953724, "grad_norm": 0.10196342319250107, "global_step": 39327, "epoch": 371, "lr": 1.554001441720085e-05} +{"train_loss": 0.014216379262506962, "grad_norm": 0.07595297694206238, "global_step": 39328, "epoch": 371, "lr": 1.553786701709733e-05} +{"train_loss": 0.013935007154941559, "grad_norm": 0.08202749490737915, "global_step": 39329, "epoch": 371, "lr": 1.5535719738078757e-05} +{"train_loss": 0.015658259391784668, "grad_norm": 0.09432119876146317, "global_step": 39330, "epoch": 371, "lr": 1.5533572580152662e-05} +{"train_loss": 0.013489777222275734, "grad_norm": 0.09046770632266998, "global_step": 39331, "epoch": 371, "lr": 1.55314255433266e-05} +{"train_loss": 0.015841707587242126, "grad_norm": 0.10182853043079376, "global_step": 39332, "epoch": 371, "lr": 1.5529278627608134e-05} +{"train_loss": 0.012605457566678524, "grad_norm": 0.08370300382375717, "global_step": 39333, "epoch": 371, "lr": 1.552713183300478e-05} +{"train_loss": 0.01652921549975872, "grad_norm": 0.05628935247659683, "global_step": 39334, "epoch": 371, "lr": 1.552498515952409e-05} +{"train_loss": 0.015803957358002663, "grad_norm": 0.1518682986497879, "global_step": 39335, "epoch": 371, "lr": 1.5522838607173627e-05} +{"train_loss": 0.016662269830703735, "grad_norm": 0.06168288365006447, "global_step": 39336, "epoch": 371, "lr": 1.552069217596092e-05} +{"train_loss": 0.016582271084189415, "grad_norm": 0.09501951932907104, "global_step": 39337, "epoch": 371, "lr": 1.5518545865893485e-05} +{"train_loss": 0.01525929756462574, "grad_norm": 0.05469894036650658, "global_step": 39338, "epoch": 371, "lr": 1.5516399676978882e-05} +{"train_loss": 0.017330151051282883, "grad_norm": 0.07497282326221466, "global_step": 39339, "epoch": 371, "lr": 1.5514253609224676e-05} +{"train_loss": 0.014190346002578735, "grad_norm": 0.0657733604311943, "global_step": 39340, "epoch": 371, "lr": 1.5512107662638363e-05} +{"train_loss": 0.015541919507086277, "grad_norm": 0.06913940608501434, "global_step": 39341, "epoch": 371, "lr": 1.5509961837227527e-05} +{"train_loss": 0.013310575857758522, "grad_norm": 0.06967820227146149, "global_step": 39342, "epoch": 371, "lr": 1.5507816132999685e-05} +{"train_loss": 0.01250471081584692, "grad_norm": 0.06594660878181458, "global_step": 39343, "epoch": 371, "lr": 1.550567054996236e-05} +{"train_loss": 0.01391446590423584, "grad_norm": 0.050112634897232056, "global_step": 39344, "epoch": 371, "lr": 1.5503525088123105e-05} +{"train_loss": 0.01625915616750717, "grad_norm": 0.07201902568340302, "global_step": 39345, "epoch": 371, "lr": 1.550137974748948e-05} +{"train_loss": 0.01554687600582838, "grad_norm": 0.06814916431903839, "global_step": 39346, "epoch": 371, "lr": 1.5499234528069e-05} +{"train_loss": 0.01362715195864439, "grad_norm": 0.1394350677728653, "global_step": 39347, "epoch": 371, "lr": 1.5497089429869195e-05} +{"train_loss": 0.012108017690479755, "grad_norm": 0.05490998551249504, "global_step": 39348, "epoch": 371, "lr": 1.5494944452897607e-05} +{"train_loss": 0.01411464624106884, "grad_norm": 0.08365649729967117, "global_step": 39349, "epoch": 371, "lr": 1.5492799597161795e-05} +{"train_loss": 0.013802610337734222, "grad_norm": 0.08473162353038788, "global_step": 39350, "epoch": 371, "lr": 1.549065486266927e-05} +{"train_loss": 0.016337325796484947, "grad_norm": 0.08466193079948425, "global_step": 39351, "epoch": 371, "lr": 1.5488510249427567e-05} +{"train_loss": 0.014507835730910301, "grad_norm": 0.09399550408124924, "global_step": 39352, "epoch": 371, "lr": 1.548636575744425e-05} +{"train_loss": 0.015220286324620247, "grad_norm": 0.09199058264493942, "global_step": 39353, "epoch": 371, "lr": 1.548422138672683e-05} +{"train_loss": 0.013765469193458557, "grad_norm": 0.08949607610702515, "global_step": 39354, "epoch": 371, "lr": 1.5482077137282837e-05} +{"train_loss": 0.013354282826185226, "grad_norm": 0.12164397537708282, "global_step": 39355, "epoch": 371, "lr": 1.547993300911981e-05} +{"train_loss": 0.015997126698493958, "grad_norm": 0.13455769419670105, "global_step": 39356, "epoch": 371, "lr": 1.54777890022453e-05} +{"train_loss": 0.015329862013459206, "grad_norm": 0.07459856569766998, "global_step": 39357, "epoch": 371, "lr": 1.5475645116666816e-05} +{"train_loss": 0.012871094979345798, "grad_norm": 0.0600074827671051, "global_step": 39358, "epoch": 371, "lr": 1.5473501352391913e-05} +{"train_loss": 0.015126586891710758, "grad_norm": 0.06865835934877396, "global_step": 39359, "epoch": 371, "lr": 1.5471357709428093e-05} +{"train_loss": 0.016303563490509987, "grad_norm": 0.09447286278009415, "global_step": 39360, "epoch": 371, "lr": 1.5469214187782928e-05} +{"train_loss": 0.016667667776346207, "grad_norm": 0.08482729643583298, "global_step": 39361, "epoch": 371, "lr": 1.5467070787463906e-05} +{"train_loss": 0.015265380032360554, "grad_norm": 0.10084265470504761, "global_step": 39362, "epoch": 371, "lr": 1.5464927508478576e-05} +{"train_loss": 0.014997832477092743, "grad_norm": 0.12111223489046097, "global_step": 39363, "epoch": 371, "lr": 1.5462784350834497e-05} +{"train_loss": 0.013553488999605179, "grad_norm": 0.05553241819143295, "global_step": 39364, "epoch": 371, "lr": 1.546064131453917e-05} +{"train_loss": 0.017535507678985596, "grad_norm": 0.09354682266712189, "global_step": 39365, "epoch": 371, "lr": 1.5458498399600108e-05} +{"train_loss": 0.01884067803621292, "grad_norm": 0.08564221113920212, "global_step": 39366, "epoch": 371, "lr": 1.545635560602488e-05} +{"train_loss": 0.014696246013045311, "grad_norm": 0.09043079614639282, "global_step": 39367, "epoch": 371, "lr": 1.5454212933820985e-05} +{"train_loss": 0.015936918556690216, "grad_norm": 0.0767788290977478, "global_step": 39368, "epoch": 371, "lr": 1.5452070382995954e-05} +{"train_loss": 0.013663367368280888, "grad_norm": 0.11090445518493652, "global_step": 39369, "epoch": 371, "lr": 1.544992795355735e-05} +{"train_loss": 0.015364788472652435, "grad_norm": 0.06968904286623001, "global_step": 39370, "epoch": 371, "lr": 1.5447785645512663e-05} +{"train_loss": 0.016121361404657364, "grad_norm": 0.08025389909744263, "global_step": 39371, "epoch": 371, "lr": 1.5445643458869417e-05} +{"train_loss": 0.014200098812580109, "grad_norm": 0.10090100765228271, "global_step": 39372, "epoch": 371, "lr": 1.544350139363516e-05} +{"train_loss": 0.01613493263721466, "grad_norm": 0.09843653440475464, "global_step": 39373, "epoch": 371, "lr": 1.5441359449817418e-05} +{"train_loss": 0.014715176075696945, "grad_norm": 0.07425288110971451, "global_step": 39374, "epoch": 371, "lr": 1.5439217627423697e-05} +{"train_loss": 0.016318853944540024, "grad_norm": 0.05052117630839348, "global_step": 39375, "epoch": 371, "lr": 1.543707592646153e-05} +{"train_loss": 0.01650409586727619, "grad_norm": 0.0667320266366005, "global_step": 39376, "epoch": 371, "lr": 1.5434934346938474e-05} +{"train_loss": 0.015389226377010345, "grad_norm": 0.09216787666082382, "global_step": 39377, "epoch": 371, "lr": 1.543279288886202e-05} +{"train_loss": 0.013305200263857841, "grad_norm": 0.08036071062088013, "global_step": 39378, "epoch": 371, "lr": 1.5430651552239685e-05} +{"train_loss": 0.017568368464708328, "grad_norm": 0.062488406896591187, "global_step": 39379, "epoch": 371, "lr": 1.5428510337079005e-05} +{"train_loss": 0.015675820410251617, "grad_norm": 0.07688731700181961, "global_step": 39380, "epoch": 371, "lr": 1.5426369243387518e-05} +{"train_loss": 0.014199317432940006, "grad_norm": 0.06293494254350662, "global_step": 39381, "epoch": 371, "lr": 1.5424228271172726e-05} +{"train_loss": 0.013899479061365128, "grad_norm": 0.07209157198667526, "global_step": 39382, "epoch": 371, "lr": 1.5422087420442165e-05} +{"train_loss": 0.015029431320726871, "grad_norm": 0.05622640997171402, "global_step": 39383, "epoch": 371, "lr": 1.5419946691203358e-05} +{"train_loss": 0.013858424499630928, "grad_norm": 0.1437472105026245, "global_step": 39384, "epoch": 371, "lr": 1.5417806083463804e-05} +{"train_loss": 0.013713482767343521, "grad_norm": 0.07842054963111877, "global_step": 39385, "epoch": 371, "lr": 1.541566559723104e-05} +{"train_loss": 0.015562373213469982, "grad_norm": 0.0713033676147461, "global_step": 39386, "epoch": 371, "lr": 1.54135252325126e-05} +{"train_loss": 0.014911578968167305, "grad_norm": 0.06959064304828644, "global_step": 39387, "epoch": 371, "lr": 1.5411384989315975e-05} +{"train_loss": 0.016210203990340233, "grad_norm": 0.14464573562145233, "global_step": 39388, "epoch": 371, "lr": 1.540924486764872e-05} +{"train_loss": 0.014633371494710445, "grad_norm": 0.07477094978094101, "global_step": 39389, "epoch": 371, "lr": 1.540710486751832e-05} +{"train_loss": 0.016745978966355324, "grad_norm": 0.07069224119186401, "global_step": 39390, "epoch": 371, "lr": 1.540496498893232e-05} +{"train_loss": 0.01461928989738226, "grad_norm": 0.06734355539083481, "global_step": 39391, "epoch": 371, "lr": 1.5402825231898216e-05} +{"train_loss": 0.015856724232435226, "grad_norm": 0.08176803588867188, "global_step": 39392, "epoch": 371, "lr": 1.5400685596423537e-05} +{"train_loss": 0.0155250309035182, "grad_norm": 0.10885600000619888, "global_step": 39393, "epoch": 371, "lr": 1.539854608251582e-05} +{"train_loss": 0.01309429481625557, "grad_norm": 0.10148859769105911, "global_step": 39394, "epoch": 371, "lr": 1.5396406690182564e-05} +{"train_loss": 0.017419036477804184, "grad_norm": 0.09214680641889572, "global_step": 39395, "epoch": 371, "lr": 1.539426741943127e-05} +{"train_loss": 0.017299965023994446, "grad_norm": 0.10558507591485977, "global_step": 39396, "epoch": 371, "lr": 1.5392128270269468e-05} +{"train_loss": 0.015011454001069069, "grad_norm": 0.07807817310094833, "global_step": 39397, "epoch": 371, "lr": 1.5389989242704694e-05} +{"train_loss": 0.015632474794983864, "grad_norm": 0.12435392290353775, "global_step": 39398, "epoch": 371, "lr": 1.5387850336744432e-05} +{"train_loss": 0.015896223485469818, "grad_norm": 0.13706551492214203, "global_step": 39399, "epoch": 371, "lr": 1.5385711552396227e-05} +{"train_loss": 0.014392392709851265, "grad_norm": 0.08922002464532852, "global_step": 39400, "epoch": 371, "lr": 1.5383572889667574e-05} +{"train_loss": 0.014697135426104069, "grad_norm": 0.06393169611692429, "global_step": 39401, "epoch": 371, "lr": 1.5381434348565982e-05} +{"train_loss": 0.01446258183568716, "grad_norm": 0.07912543416023254, "global_step": 39402, "epoch": 371, "lr": 1.5379295929098968e-05} +{"train_loss": 0.014255253598093987, "grad_norm": 0.05328121408820152, "global_step": 39403, "epoch": 371, "lr": 1.5377157631274074e-05} +{"train_loss": 0.015399552881717682, "grad_norm": 0.0834445133805275, "global_step": 39404, "epoch": 371, "lr": 1.537501945509877e-05} +{"train_loss": 0.012011110782623291, "grad_norm": 0.057989560067653656, "global_step": 39405, "epoch": 371, "lr": 1.5372881400580608e-05} +{"train_loss": 0.014990721829235554, "grad_norm": 0.08530813455581665, "global_step": 39406, "epoch": 371, "lr": 1.5370743467727066e-05} +{"train_loss": 0.014832941815257072, "grad_norm": 0.06063145771622658, "global_step": 39407, "epoch": 371, "lr": 1.5368605656545687e-05} +{"train_loss": 0.012912672013044357, "grad_norm": 0.08339293301105499, "global_step": 39408, "epoch": 371, "lr": 1.5366467967043952e-05} +{"train_loss": 0.016553033143281937, "grad_norm": 0.10209950059652328, "global_step": 39409, "epoch": 371, "lr": 1.5364330399229386e-05} +{"train_loss": 0.013105301186442375, "grad_norm": 0.09145580977201462, "global_step": 39410, "epoch": 371, "lr": 1.5362192953109517e-05} +{"train_loss": 0.014728479087352753, "grad_norm": 0.06916269659996033, "global_step": 39411, "epoch": 371, "lr": 1.5360055628691837e-05} +{"train_loss": 0.015368234366178513, "grad_norm": 0.08849870413541794, "global_step": 39412, "epoch": 371, "lr": 1.535791842598384e-05} +{"train_loss": 0.015746621415019035, "grad_norm": 0.09796394407749176, "global_step": 39413, "epoch": 371, "lr": 1.5355781344993058e-05} +{"train_loss": 0.01533531490713358, "grad_norm": 0.05249058082699776, "global_step": 39414, "epoch": 371, "lr": 1.5353644385727005e-05} +{"train_loss": 0.017278458923101425, "grad_norm": 0.09354294091463089, "global_step": 39415, "epoch": 371, "lr": 1.5351507548193167e-05} +{"train_loss": 0.016491666436195374, "grad_norm": 0.0921313613653183, "global_step": 39416, "epoch": 371, "lr": 1.5349370832399058e-05} +{"train_loss": 0.014727252535521984, "grad_norm": 0.06720549613237381, "global_step": 39417, "epoch": 371, "lr": 1.534723423835221e-05} +{"train_loss": 0.01605617254972458, "grad_norm": 0.07579091936349869, "global_step": 39418, "epoch": 371, "lr": 1.5345097766060106e-05} +{"train_loss": 0.013853656128048897, "grad_norm": 0.06512218713760376, "global_step": 39419, "epoch": 371, "lr": 1.5342961415530242e-05} +{"train_loss": 0.01536074560135603, "grad_norm": 0.1764339804649353, "global_step": 39420, "epoch": 371, "lr": 1.5340825186770147e-05} +{"train_loss": 0.015063537284731865, "grad_norm": 0.07242940366268158, "global_step": 39421, "epoch": 371, "lr": 1.533868907978733e-05} +{"train_loss": 0.012877194210886955, "grad_norm": 0.07619520276784897, "global_step": 39422, "epoch": 371, "lr": 1.5336553094589273e-05} +{"train_loss": 0.014972140081226826, "grad_norm": 0.1445206105709076, "global_step": 39423, "epoch": 371, "lr": 1.5334417231183508e-05} +{"train_loss": 0.017273442819714546, "grad_norm": 0.06759490817785263, "global_step": 39424, "epoch": 371, "lr": 1.5332281489577522e-05} +{"train_loss": 0.014893626794219017, "grad_norm": 0.08485984802246094, "global_step": 39425, "epoch": 371, "lr": 1.5330145869778805e-05} +{"train_loss": 0.014682418666779995, "grad_norm": 0.10158557444810867, "global_step": 39426, "epoch": 371, "lr": 1.5328010371794887e-05} +{"train_loss": 0.01564883440732956, "grad_norm": 0.07091514766216278, "global_step": 39427, "epoch": 371, "lr": 1.532587499563327e-05} +{"train_loss": 0.015628915280103683, "grad_norm": 0.0933227390050888, "global_step": 39428, "epoch": 371, "lr": 1.5323739741301437e-05} +{"train_loss": 0.01419039722532034, "grad_norm": 0.11120322346687317, "global_step": 39429, "epoch": 371, "lr": 1.5321604608806916e-05} +{"train_loss": 0.017051173374056816, "grad_norm": 0.16142524778842926, "global_step": 39430, "epoch": 371, "lr": 1.531946959815718e-05} +{"train_loss": 0.015040833564510322, "grad_norm": 0.08525002747774124, "global_step": 39431, "epoch": 371, "lr": 1.531733470935976e-05, "val_loss": 0.019459614530205727} +{"train_loss": 0.016153624281287193, "grad_norm": 0.05866720527410507, "global_step": 39432, "epoch": 372, "lr": 1.531519994242213e-05} +{"train_loss": 0.014463216066360474, "grad_norm": 0.12684288620948792, "global_step": 39433, "epoch": 372, "lr": 1.5313065297351802e-05} +{"train_loss": 0.015075176022946835, "grad_norm": 0.060945041477680206, "global_step": 39434, "epoch": 372, "lr": 1.531093077415629e-05} +{"train_loss": 0.014432847499847412, "grad_norm": 0.08167524635791779, "global_step": 39435, "epoch": 372, "lr": 1.5308796372843087e-05} +{"train_loss": 0.013616325333714485, "grad_norm": 0.06818323582410812, "global_step": 39436, "epoch": 372, "lr": 1.530666209341967e-05} +{"train_loss": 0.014337092638015747, "grad_norm": 0.06853336095809937, "global_step": 39437, "epoch": 372, "lr": 1.5304527935893554e-05} +{"train_loss": 0.015350263565778732, "grad_norm": 0.06252206116914749, "global_step": 39438, "epoch": 372, "lr": 1.530239390027225e-05} +{"train_loss": 0.015233880840241909, "grad_norm": 0.06600980460643768, "global_step": 39439, "epoch": 372, "lr": 1.5300259986563232e-05} +{"train_loss": 0.013053127564489841, "grad_norm": 0.05881156027317047, "global_step": 39440, "epoch": 372, "lr": 1.529812619477401e-05} +{"train_loss": 0.014782779850065708, "grad_norm": 0.08170939981937408, "global_step": 39441, "epoch": 372, "lr": 1.5295992524912113e-05} +{"train_loss": 0.014484064653515816, "grad_norm": 0.06878269463777542, "global_step": 39442, "epoch": 372, "lr": 1.5293858976984977e-05} +{"train_loss": 0.014893025159835815, "grad_norm": 0.0765758529305458, "global_step": 39443, "epoch": 372, "lr": 1.529172555100012e-05} +{"train_loss": 0.016309339553117752, "grad_norm": 0.10259439796209335, "global_step": 39444, "epoch": 372, "lr": 1.5289592246965068e-05} +{"train_loss": 0.018145307898521423, "grad_norm": 0.06616736203432083, "global_step": 39445, "epoch": 372, "lr": 1.528745906488727e-05} +{"train_loss": 0.015368476510047913, "grad_norm": 0.08886376768350601, "global_step": 39446, "epoch": 372, "lr": 1.5285326004774263e-05} +{"train_loss": 0.014762450940907001, "grad_norm": 0.06672658026218414, "global_step": 39447, "epoch": 372, "lr": 1.528319306663351e-05} +{"train_loss": 0.015188828110694885, "grad_norm": 0.10919173806905746, "global_step": 39448, "epoch": 372, "lr": 1.528106025047254e-05} +{"train_loss": 0.01291086245328188, "grad_norm": 0.06190652772784233, "global_step": 39449, "epoch": 372, "lr": 1.52789275562988e-05} +{"train_loss": 0.014661945402622223, "grad_norm": 0.1231316477060318, "global_step": 39450, "epoch": 372, "lr": 1.527679498411981e-05} +{"train_loss": 0.013740502297878265, "grad_norm": 0.07427936792373657, "global_step": 39451, "epoch": 372, "lr": 1.5274662533943086e-05} +{"train_loss": 0.013703770935535431, "grad_norm": 0.12413936108350754, "global_step": 39452, "epoch": 372, "lr": 1.5272530205776086e-05} +{"train_loss": 0.01577293872833252, "grad_norm": 0.12247113138437271, "global_step": 39453, "epoch": 372, "lr": 1.5270397999626303e-05} +{"train_loss": 0.01547158882021904, "grad_norm": 0.09989416599273682, "global_step": 39454, "epoch": 372, "lr": 1.5268265915501234e-05} +{"train_loss": 0.014700105413794518, "grad_norm": 0.062299519777297974, "global_step": 39455, "epoch": 372, "lr": 1.5266133953408396e-05} +{"train_loss": 0.014698881655931473, "grad_norm": 0.06893290579319, "global_step": 39456, "epoch": 372, "lr": 1.5264002113355235e-05} +{"train_loss": 0.014668408781290054, "grad_norm": 0.08674173057079315, "global_step": 39457, "epoch": 372, "lr": 1.526187039534927e-05} +{"train_loss": 0.019588321447372437, "grad_norm": 0.09487173706293106, "global_step": 39458, "epoch": 372, "lr": 1.525973879939801e-05} +{"train_loss": 0.013411158695816994, "grad_norm": 0.06362985819578171, "global_step": 39459, "epoch": 372, "lr": 1.5257607325508894e-05} +{"train_loss": 0.015429766848683357, "grad_norm": 0.09558870643377304, "global_step": 39460, "epoch": 372, "lr": 1.5255475973689436e-05} +{"train_loss": 0.01572198048233986, "grad_norm": 0.1033787876367569, "global_step": 39461, "epoch": 372, "lr": 1.525334474394714e-05} +{"train_loss": 0.014707420021295547, "grad_norm": 0.0584978386759758, "global_step": 39462, "epoch": 372, "lr": 1.5251213636289464e-05} +{"train_loss": 0.012845590710639954, "grad_norm": 0.11193317919969559, "global_step": 39463, "epoch": 372, "lr": 1.5249082650723912e-05} +{"train_loss": 0.015500830486416817, "grad_norm": 0.08272115141153336, "global_step": 39464, "epoch": 372, "lr": 1.5246951787257985e-05} +{"train_loss": 0.016498534008860588, "grad_norm": 0.056034497916698456, "global_step": 39465, "epoch": 372, "lr": 1.5244821045899154e-05} +{"train_loss": 0.017891239374876022, "grad_norm": 0.12633845210075378, "global_step": 39466, "epoch": 372, "lr": 1.5242690426654893e-05} +{"train_loss": 0.015022925101220608, "grad_norm": 0.1078629344701767, "global_step": 39467, "epoch": 372, "lr": 1.5240559929532705e-05} +{"train_loss": 0.014245070517063141, "grad_norm": 0.07009803503751755, "global_step": 39468, "epoch": 372, "lr": 1.523842955454009e-05} +{"train_loss": 0.0157911516726017, "grad_norm": 0.09052606672048569, "global_step": 39469, "epoch": 372, "lr": 1.5236299301684498e-05} +{"train_loss": 0.012289369478821754, "grad_norm": 0.0600639171898365, "global_step": 39470, "epoch": 372, "lr": 1.5234169170973445e-05} +{"train_loss": 0.01399193610996008, "grad_norm": 0.09218215942382812, "global_step": 39471, "epoch": 372, "lr": 1.5232039162414391e-05} +{"train_loss": 0.01621915027499199, "grad_norm": 0.060112860053777695, "global_step": 39472, "epoch": 372, "lr": 1.5229909276014842e-05} +{"train_loss": 0.01578180305659771, "grad_norm": 0.08302280306816101, "global_step": 39473, "epoch": 372, "lr": 1.5227779511782258e-05} +{"train_loss": 0.015028445050120354, "grad_norm": 0.13270148634910583, "global_step": 39474, "epoch": 372, "lr": 1.5225649869724134e-05} +{"train_loss": 0.015250650234520435, "grad_norm": 0.12642818689346313, "global_step": 39475, "epoch": 372, "lr": 1.5223520349847969e-05} +{"train_loss": 0.01234412845224142, "grad_norm": 0.09046956151723862, "global_step": 39476, "epoch": 372, "lr": 1.522139095216123e-05} +{"train_loss": 0.015124181285500526, "grad_norm": 0.060130879282951355, "global_step": 39477, "epoch": 372, "lr": 1.521926167667138e-05} +{"train_loss": 0.015956265851855278, "grad_norm": 0.14175395667552948, "global_step": 39478, "epoch": 372, "lr": 1.5217132523385923e-05} +{"train_loss": 0.01526110153645277, "grad_norm": 0.07431590557098389, "global_step": 39479, "epoch": 372, "lr": 1.5215003492312351e-05} +{"train_loss": 0.013338852673768997, "grad_norm": 0.09195992350578308, "global_step": 39480, "epoch": 372, "lr": 1.5212874583458114e-05} +{"train_loss": 0.012544438242912292, "grad_norm": 0.06166299805045128, "global_step": 39481, "epoch": 372, "lr": 1.5210745796830705e-05} +{"train_loss": 0.01302403211593628, "grad_norm": 0.05044834315776825, "global_step": 39482, "epoch": 372, "lr": 1.5208617132437635e-05} +{"train_loss": 0.011502137407660484, "grad_norm": 0.10829794406890869, "global_step": 39483, "epoch": 372, "lr": 1.5206488590286327e-05} +{"train_loss": 0.0145042659714818, "grad_norm": 0.08559607714414597, "global_step": 39484, "epoch": 372, "lr": 1.5204360170384286e-05} +{"train_loss": 0.015863165259361267, "grad_norm": 0.10097315162420273, "global_step": 39485, "epoch": 372, "lr": 1.5202231872739003e-05} +{"train_loss": 0.01715919002890587, "grad_norm": 0.07334151864051819, "global_step": 39486, "epoch": 372, "lr": 1.5200103697357931e-05} +{"train_loss": 0.016813993453979492, "grad_norm": 0.08393300324678421, "global_step": 39487, "epoch": 372, "lr": 1.5197975644248563e-05} +{"train_loss": 0.01515529491007328, "grad_norm": 0.12160734087228775, "global_step": 39488, "epoch": 372, "lr": 1.5195847713418387e-05} +{"train_loss": 0.01410844549536705, "grad_norm": 0.09122169017791748, "global_step": 39489, "epoch": 372, "lr": 1.5193719904874864e-05} +{"train_loss": 0.016112398356199265, "grad_norm": 0.06121278926730156, "global_step": 39490, "epoch": 372, "lr": 1.5191592218625462e-05} +{"train_loss": 0.014169741421937943, "grad_norm": 0.11418268829584122, "global_step": 39491, "epoch": 372, "lr": 1.5189464654677666e-05} +{"train_loss": 0.015867741778492928, "grad_norm": 0.07362004369497299, "global_step": 39492, "epoch": 372, "lr": 1.5187337213038967e-05} +{"train_loss": 0.016679834574460983, "grad_norm": 0.08518023788928986, "global_step": 39493, "epoch": 372, "lr": 1.5185209893716807e-05} +{"train_loss": 0.016398685052990913, "grad_norm": 0.06138759106397629, "global_step": 39494, "epoch": 372, "lr": 1.5183082696718692e-05} +{"train_loss": 0.01894843578338623, "grad_norm": 0.09250157326459885, "global_step": 39495, "epoch": 372, "lr": 1.5180955622052073e-05} +{"train_loss": 0.017004597932100296, "grad_norm": 0.0629357323050499, "global_step": 39496, "epoch": 372, "lr": 1.517882866972445e-05} +{"train_loss": 0.01647709123790264, "grad_norm": 0.11353437602519989, "global_step": 39497, "epoch": 372, "lr": 1.5176701839743262e-05} +{"train_loss": 0.015939485281705856, "grad_norm": 0.09135444462299347, "global_step": 39498, "epoch": 372, "lr": 1.5174575132115998e-05} +{"train_loss": 0.01395657192915678, "grad_norm": 0.08633765578269958, "global_step": 39499, "epoch": 372, "lr": 1.5172448546850165e-05} +{"train_loss": 0.01425418071448803, "grad_norm": 0.10294506698846817, "global_step": 39500, "epoch": 372, "lr": 1.517032208395317e-05} +{"train_loss": 0.015262885019183159, "grad_norm": 0.09900590777397156, "global_step": 39501, "epoch": 372, "lr": 1.5168195743432516e-05} +{"train_loss": 0.01712847873568535, "grad_norm": 0.06706977635622025, "global_step": 39502, "epoch": 372, "lr": 1.5166069525295689e-05} +{"train_loss": 0.014848105609416962, "grad_norm": 0.14838720858097076, "global_step": 39503, "epoch": 372, "lr": 1.5163943429550131e-05} +{"train_loss": 0.014629168435931206, "grad_norm": 0.05412653088569641, "global_step": 39504, "epoch": 372, "lr": 1.5161817456203325e-05} +{"train_loss": 0.01353292167186737, "grad_norm": 0.07455674558877945, "global_step": 39505, "epoch": 372, "lr": 1.515969160526276e-05} +{"train_loss": 0.01598399505019188, "grad_norm": 0.05955887213349342, "global_step": 39506, "epoch": 372, "lr": 1.5157565876735885e-05} +{"train_loss": 0.014291943982243538, "grad_norm": 0.09471234679222107, "global_step": 39507, "epoch": 372, "lr": 1.5155440270630156e-05} +{"train_loss": 0.015409656800329685, "grad_norm": 0.08398998528718948, "global_step": 39508, "epoch": 372, "lr": 1.5153314786953054e-05} +{"train_loss": 0.015851669013500214, "grad_norm": 0.07959302514791489, "global_step": 39509, "epoch": 372, "lr": 1.5151189425712064e-05} +{"train_loss": 0.012574532069265842, "grad_norm": 0.08339923620223999, "global_step": 39510, "epoch": 372, "lr": 1.514906418691463e-05} +{"train_loss": 0.016931820660829544, "grad_norm": 0.10058367252349854, "global_step": 39511, "epoch": 372, "lr": 1.5146939070568238e-05} +{"train_loss": 0.014217238873243332, "grad_norm": 0.052457161247730255, "global_step": 39512, "epoch": 372, "lr": 1.514481407668033e-05} +{"train_loss": 0.013354882597923279, "grad_norm": 0.07860051840543747, "global_step": 39513, "epoch": 372, "lr": 1.5142689205258404e-05} +{"train_loss": 0.013462637551128864, "grad_norm": 0.04779892787337303, "global_step": 39514, "epoch": 372, "lr": 1.5140564456309891e-05} +{"train_loss": 0.014597758650779724, "grad_norm": 0.07576990127563477, "global_step": 39515, "epoch": 372, "lr": 1.5138439829842277e-05} +{"train_loss": 0.01424565538764, "grad_norm": 0.10964933782815933, "global_step": 39516, "epoch": 372, "lr": 1.5136315325863043e-05} +{"train_loss": 0.014806792140007019, "grad_norm": 0.09030435234308243, "global_step": 39517, "epoch": 372, "lr": 1.5134190944379634e-05} +{"train_loss": 0.017376461997628212, "grad_norm": 0.14243178069591522, "global_step": 39518, "epoch": 372, "lr": 1.5132066685399493e-05} +{"train_loss": 0.015229882672429085, "grad_norm": 0.07971613109111786, "global_step": 39519, "epoch": 372, "lr": 1.5129942548930115e-05} +{"train_loss": 0.015504692681133747, "grad_norm": 0.07288798689842224, "global_step": 39520, "epoch": 372, "lr": 1.5127818534978966e-05} +{"train_loss": 0.014738206751644611, "grad_norm": 0.07493146508932114, "global_step": 39521, "epoch": 372, "lr": 1.5125694643553478e-05} +{"train_loss": 0.013344330713152885, "grad_norm": 0.13559602200984955, "global_step": 39522, "epoch": 372, "lr": 1.5123570874661136e-05} +{"train_loss": 0.015261582098901272, "grad_norm": 0.08009994775056839, "global_step": 39523, "epoch": 372, "lr": 1.5121447228309433e-05} +{"train_loss": 0.015254451893270016, "grad_norm": 0.07216031849384308, "global_step": 39524, "epoch": 372, "lr": 1.5119323704505762e-05} +{"train_loss": 0.016052179038524628, "grad_norm": 0.13482926785945892, "global_step": 39525, "epoch": 372, "lr": 1.5117200303257617e-05} +{"train_loss": 0.014387508854269981, "grad_norm": 0.0600474551320076, "global_step": 39526, "epoch": 372, "lr": 1.5115077024572472e-05} +{"train_loss": 0.01519609335809946, "grad_norm": 0.0890481024980545, "global_step": 39527, "epoch": 372, "lr": 1.5112953868457764e-05} +{"train_loss": 0.013140428811311722, "grad_norm": 0.07931290566921234, "global_step": 39528, "epoch": 372, "lr": 1.5110830834920964e-05} +{"train_loss": 0.01468683872371912, "grad_norm": 0.13229048252105713, "global_step": 39529, "epoch": 372, "lr": 1.510870792396955e-05} +{"train_loss": 0.01354060135781765, "grad_norm": 0.09107746928930283, "global_step": 39530, "epoch": 372, "lr": 1.5106585135610962e-05} +{"train_loss": 0.014993177726864815, "grad_norm": 0.0789111852645874, "global_step": 39531, "epoch": 372, "lr": 1.5104462469852637e-05} +{"train_loss": 0.015636960044503212, "grad_norm": 0.07380970567464828, "global_step": 39532, "epoch": 372, "lr": 1.5102339926702064e-05} +{"train_loss": 0.01541368942707777, "grad_norm": 0.08548439294099808, "global_step": 39533, "epoch": 372, "lr": 1.5100217506166702e-05} +{"train_loss": 0.013756314292550087, "grad_norm": 0.10898017138242722, "global_step": 39534, "epoch": 372, "lr": 1.5098095208253982e-05} +{"train_loss": 0.018237628042697906, "grad_norm": 0.0755084976553917, "global_step": 39535, "epoch": 372, "lr": 1.5095973032971395e-05} +{"train_loss": 0.0165413748472929, "grad_norm": 0.11266152560710907, "global_step": 39536, "epoch": 372, "lr": 1.5093850980326363e-05} +{"train_loss": 0.015047605123089731, "grad_norm": 0.10902997851371765, "global_step": 39537, "epoch": 372, "lr": 1.5091729050326375e-05, "val_loss": 0.019538145512342453} +{"train_loss": 0.01527466718107462, "grad_norm": 0.13009138405323029, "global_step": 39538, "epoch": 373, "lr": 1.5089607242978854e-05} +{"train_loss": 0.015751507133245468, "grad_norm": 0.07549992203712463, "global_step": 39539, "epoch": 373, "lr": 1.508748555829127e-05} +{"train_loss": 0.01668628491461277, "grad_norm": 0.07069131731987, "global_step": 39540, "epoch": 373, "lr": 1.5085363996271096e-05} +{"train_loss": 0.014734601601958275, "grad_norm": 0.0715990737080574, "global_step": 39541, "epoch": 373, "lr": 1.5083242556925763e-05} +{"train_loss": 0.014607744291424751, "grad_norm": 0.11458011716604233, "global_step": 39542, "epoch": 373, "lr": 1.5081121240262725e-05} +{"train_loss": 0.012874407693743706, "grad_norm": 0.045764412730932236, "global_step": 39543, "epoch": 373, "lr": 1.5079000046289448e-05} +{"train_loss": 0.015512988902628422, "grad_norm": 0.14383827149868011, "global_step": 39544, "epoch": 373, "lr": 1.5076878975013364e-05} +{"train_loss": 0.013485020026564598, "grad_norm": 0.06848547607660294, "global_step": 39545, "epoch": 373, "lr": 1.5074758026441943e-05} +{"train_loss": 0.01527232676744461, "grad_norm": 0.06397469341754913, "global_step": 39546, "epoch": 373, "lr": 1.5072637200582652e-05} +{"train_loss": 0.015328282490372658, "grad_norm": 0.09680648893117905, "global_step": 39547, "epoch": 373, "lr": 1.5070516497442915e-05} +{"train_loss": 0.014662221074104309, "grad_norm": 0.05573219433426857, "global_step": 39548, "epoch": 373, "lr": 1.5068395917030187e-05} +{"train_loss": 0.014599715359508991, "grad_norm": 0.08131937682628632, "global_step": 39549, "epoch": 373, "lr": 1.506627545935192e-05} +{"train_loss": 0.01587238907814026, "grad_norm": 0.06953752040863037, "global_step": 39550, "epoch": 373, "lr": 1.506415512441558e-05} +{"train_loss": 0.01577013172209263, "grad_norm": 0.15575122833251953, "global_step": 39551, "epoch": 373, "lr": 1.5062034912228596e-05} +{"train_loss": 0.014180767349898815, "grad_norm": 0.0637989193201065, "global_step": 39552, "epoch": 373, "lr": 1.5059914822798443e-05} +{"train_loss": 0.016220148652791977, "grad_norm": 0.1108064278960228, "global_step": 39553, "epoch": 373, "lr": 1.5057794856132534e-05} +{"train_loss": 0.015257999300956726, "grad_norm": 0.14115218818187714, "global_step": 39554, "epoch": 373, "lr": 1.5055675012238358e-05} +{"train_loss": 0.01296276319772005, "grad_norm": 0.056572407484054565, "global_step": 39555, "epoch": 373, "lr": 1.5053555291123323e-05} +{"train_loss": 0.014830388128757477, "grad_norm": 0.10188817232847214, "global_step": 39556, "epoch": 373, "lr": 1.5051435692794902e-05} +{"train_loss": 0.01335166022181511, "grad_norm": 0.12274819612503052, "global_step": 39557, "epoch": 373, "lr": 1.504931621726055e-05} +{"train_loss": 0.014811980538070202, "grad_norm": 0.08200781047344208, "global_step": 39558, "epoch": 373, "lr": 1.5047196864527703e-05} +{"train_loss": 0.014869853854179382, "grad_norm": 0.07990115880966187, "global_step": 39559, "epoch": 373, "lr": 1.5045077634603787e-05} +{"train_loss": 0.018961962312459946, "grad_norm": 0.08261573314666748, "global_step": 39560, "epoch": 373, "lr": 1.5042958527496281e-05} +{"train_loss": 0.012871099635958672, "grad_norm": 0.08420970290899277, "global_step": 39561, "epoch": 373, "lr": 1.5040839543212604e-05} +{"train_loss": 0.014325356110930443, "grad_norm": 0.07869073003530502, "global_step": 39562, "epoch": 373, "lr": 1.5038720681760216e-05} +{"train_loss": 0.015052533708512783, "grad_norm": 0.08182361721992493, "global_step": 39563, "epoch": 373, "lr": 1.5036601943146572e-05} +{"train_loss": 0.014173561707139015, "grad_norm": 0.14334636926651, "global_step": 39564, "epoch": 373, "lr": 1.5034483327379101e-05} +{"train_loss": 0.016129907220602036, "grad_norm": 0.15005458891391754, "global_step": 39565, "epoch": 373, "lr": 1.503236483446524e-05} +{"train_loss": 0.015117742121219635, "grad_norm": 0.06463778764009476, "global_step": 39566, "epoch": 373, "lr": 1.5030246464412434e-05} +{"train_loss": 0.01556557696312666, "grad_norm": 0.09579405188560486, "global_step": 39567, "epoch": 373, "lr": 1.5028128217228155e-05} +{"train_loss": 0.016041938215494156, "grad_norm": 0.18458350002765656, "global_step": 39568, "epoch": 373, "lr": 1.5026010092919813e-05} +{"train_loss": 0.013319148682057858, "grad_norm": 0.12335988134145737, "global_step": 39569, "epoch": 373, "lr": 1.502389209149485e-05} +{"train_loss": 0.017245572060346603, "grad_norm": 0.0984104797244072, "global_step": 39570, "epoch": 373, "lr": 1.5021774212960744e-05} +{"train_loss": 0.014623485505580902, "grad_norm": 0.13368888199329376, "global_step": 39571, "epoch": 373, "lr": 1.501965645732491e-05} +{"train_loss": 0.014868319034576416, "grad_norm": 0.12205832451581955, "global_step": 39572, "epoch": 373, "lr": 1.5017538824594774e-05} +{"train_loss": 0.016647029668092728, "grad_norm": 0.1509333997964859, "global_step": 39573, "epoch": 373, "lr": 1.5015421314777795e-05} +{"train_loss": 0.013981256633996964, "grad_norm": 0.0964817926287651, "global_step": 39574, "epoch": 373, "lr": 1.5013303927881428e-05} +{"train_loss": 0.014815293252468109, "grad_norm": 0.08068858832120895, "global_step": 39575, "epoch": 373, "lr": 1.5011186663913074e-05} +{"train_loss": 0.014719274826347828, "grad_norm": 0.06242222338914871, "global_step": 39576, "epoch": 373, "lr": 1.5009069522880215e-05} +{"train_loss": 0.01698029600083828, "grad_norm": 0.16698643565177917, "global_step": 39577, "epoch": 373, "lr": 1.500695250479025e-05} +{"train_loss": 0.015160642564296722, "grad_norm": 0.14582203328609467, "global_step": 39578, "epoch": 373, "lr": 1.5004835609650652e-05} +{"train_loss": 0.01550364401191473, "grad_norm": 0.18230047821998596, "global_step": 39579, "epoch": 373, "lr": 1.5002718837468826e-05} +{"train_loss": 0.01151826698333025, "grad_norm": 0.08427277952432632, "global_step": 39580, "epoch": 373, "lr": 1.5000602188252228e-05} +{"train_loss": 0.014659363776445389, "grad_norm": 0.08455650508403778, "global_step": 39581, "epoch": 373, "lr": 1.499848566200831e-05} +{"train_loss": 0.014559922739863396, "grad_norm": 0.09822635352611542, "global_step": 39582, "epoch": 373, "lr": 1.4996369258744492e-05} +{"train_loss": 0.016045738011598587, "grad_norm": 0.09556490927934647, "global_step": 39583, "epoch": 373, "lr": 1.4994252978468193e-05} +{"train_loss": 0.012158932164311409, "grad_norm": 0.08654998242855072, "global_step": 39584, "epoch": 373, "lr": 1.4992136821186875e-05} +{"train_loss": 0.01402355544269085, "grad_norm": 0.09143009036779404, "global_step": 39585, "epoch": 373, "lr": 1.4990020786907955e-05} +{"train_loss": 0.013255774974822998, "grad_norm": 0.11712256819009781, "global_step": 39586, "epoch": 373, "lr": 1.4987904875638875e-05} +{"train_loss": 0.013565406203269958, "grad_norm": 0.07830434292554855, "global_step": 39587, "epoch": 373, "lr": 1.4985789087387086e-05} +{"train_loss": 0.01486331969499588, "grad_norm": 0.09047169238328934, "global_step": 39588, "epoch": 373, "lr": 1.4983673422160004e-05} +{"train_loss": 0.014458195306360722, "grad_norm": 0.07572990655899048, "global_step": 39589, "epoch": 373, "lr": 1.4981557879965058e-05} +{"train_loss": 0.016809463500976562, "grad_norm": 0.07608327269554138, "global_step": 39590, "epoch": 373, "lr": 1.4979442460809683e-05} +{"train_loss": 0.01499059610068798, "grad_norm": 0.07942487299442291, "global_step": 39591, "epoch": 373, "lr": 1.4977327164701332e-05} +{"train_loss": 0.01671256124973297, "grad_norm": 0.08969463407993317, "global_step": 39592, "epoch": 373, "lr": 1.497521199164741e-05} +{"train_loss": 0.013868125155568123, "grad_norm": 0.08005448430776596, "global_step": 39593, "epoch": 373, "lr": 1.4973096941655357e-05} +{"train_loss": 0.016841750591993332, "grad_norm": 0.1574287861585617, "global_step": 39594, "epoch": 373, "lr": 1.4970982014732626e-05} +{"train_loss": 0.015760347247123718, "grad_norm": 0.08922175318002701, "global_step": 39595, "epoch": 373, "lr": 1.4968867210886628e-05} +{"train_loss": 0.015319187194108963, "grad_norm": 0.07932532578706741, "global_step": 39596, "epoch": 373, "lr": 1.4966752530124783e-05} +{"train_loss": 0.015539422631263733, "grad_norm": 0.07618489861488342, "global_step": 39597, "epoch": 373, "lr": 1.4964637972454537e-05} +{"train_loss": 0.015650682151317596, "grad_norm": 0.11079933494329453, "global_step": 39598, "epoch": 373, "lr": 1.4962523537883327e-05} +{"train_loss": 0.014761319383978844, "grad_norm": 0.1252201646566391, "global_step": 39599, "epoch": 373, "lr": 1.4960409226418576e-05} +{"train_loss": 0.015512628480792046, "grad_norm": 0.07310637831687927, "global_step": 39600, "epoch": 373, "lr": 1.4958295038067694e-05} +{"train_loss": 0.016099337488412857, "grad_norm": 0.11137253791093826, "global_step": 39601, "epoch": 373, "lr": 1.4956180972838136e-05} +{"train_loss": 0.014592990279197693, "grad_norm": 0.058445319533348083, "global_step": 39602, "epoch": 373, "lr": 1.49540670307373e-05} +{"train_loss": 0.012850550934672356, "grad_norm": 0.05980994552373886, "global_step": 39603, "epoch": 373, "lr": 1.4951953211772635e-05} +{"train_loss": 0.013092691078782082, "grad_norm": 0.06284230947494507, "global_step": 39604, "epoch": 373, "lr": 1.494983951595158e-05} +{"train_loss": 0.013458794914186, "grad_norm": 0.08057552576065063, "global_step": 39605, "epoch": 373, "lr": 1.494772594328155e-05} +{"train_loss": 0.015932593494653702, "grad_norm": 0.07733124494552612, "global_step": 39606, "epoch": 373, "lr": 1.4945612493769944e-05} +{"train_loss": 0.017145389690995216, "grad_norm": 0.10491003096103668, "global_step": 39607, "epoch": 373, "lr": 1.4943499167424213e-05} +{"train_loss": 0.01680496335029602, "grad_norm": 0.07988203316926956, "global_step": 39608, "epoch": 373, "lr": 1.4941385964251797e-05} +{"train_loss": 0.015240047127008438, "grad_norm": 0.05181710049510002, "global_step": 39609, "epoch": 373, "lr": 1.4939272884260092e-05} +{"train_loss": 0.014262864366173744, "grad_norm": 0.06034436449408531, "global_step": 39610, "epoch": 373, "lr": 1.493715992745653e-05} +{"train_loss": 0.018288802355527878, "grad_norm": 0.22285886108875275, "global_step": 39611, "epoch": 373, "lr": 1.4935047093848554e-05} +{"train_loss": 0.012705564498901367, "grad_norm": 0.061806388199329376, "global_step": 39612, "epoch": 373, "lr": 1.4932934383443576e-05} +{"train_loss": 0.01580279879271984, "grad_norm": 0.0928778201341629, "global_step": 39613, "epoch": 373, "lr": 1.4930821796249e-05} +{"train_loss": 0.016282159835100174, "grad_norm": 0.11215069890022278, "global_step": 39614, "epoch": 373, "lr": 1.4928709332272266e-05} +{"train_loss": 0.015168992802500725, "grad_norm": 0.0603652149438858, "global_step": 39615, "epoch": 373, "lr": 1.4926596991520814e-05} +{"train_loss": 0.016611460596323013, "grad_norm": 0.07698166370391846, "global_step": 39616, "epoch": 373, "lr": 1.4924484774002023e-05} +{"train_loss": 0.0146200992166996, "grad_norm": 0.062334634363651276, "global_step": 39617, "epoch": 373, "lr": 1.4922372679723362e-05} +{"train_loss": 0.014458205550909042, "grad_norm": 0.06121682748198509, "global_step": 39618, "epoch": 373, "lr": 1.4920260708692224e-05} +{"train_loss": 0.01488580834120512, "grad_norm": 0.08048190921545029, "global_step": 39619, "epoch": 373, "lr": 1.491814886091602e-05} +{"train_loss": 0.014763515442609787, "grad_norm": 0.1269727349281311, "global_step": 39620, "epoch": 373, "lr": 1.4916037136402183e-05} +{"train_loss": 0.01727944239974022, "grad_norm": 0.1421341747045517, "global_step": 39621, "epoch": 373, "lr": 1.4913925535158135e-05} +{"train_loss": 0.01640922948718071, "grad_norm": 0.09070687741041183, "global_step": 39622, "epoch": 373, "lr": 1.4911814057191309e-05} +{"train_loss": 0.013482743874192238, "grad_norm": 0.12054726481437683, "global_step": 39623, "epoch": 373, "lr": 1.4909702702509116e-05} +{"train_loss": 0.015625765547156334, "grad_norm": 0.1167554259300232, "global_step": 39624, "epoch": 373, "lr": 1.4907591471118942e-05} +{"train_loss": 0.013162183575332165, "grad_norm": 0.06293708831071854, "global_step": 39625, "epoch": 373, "lr": 1.4905480363028252e-05} +{"train_loss": 0.012720681726932526, "grad_norm": 0.04907459765672684, "global_step": 39626, "epoch": 373, "lr": 1.490336937824442e-05} +{"train_loss": 0.014758587814867496, "grad_norm": 0.14394989609718323, "global_step": 39627, "epoch": 373, "lr": 1.490125851677489e-05} +{"train_loss": 0.012588589452207088, "grad_norm": 0.06500770896673203, "global_step": 39628, "epoch": 373, "lr": 1.4899147778627088e-05} +{"train_loss": 0.015977194532752037, "grad_norm": 0.07276760786771774, "global_step": 39629, "epoch": 373, "lr": 1.4897037163808414e-05} +{"train_loss": 0.015066519379615784, "grad_norm": 0.12774907052516937, "global_step": 39630, "epoch": 373, "lr": 1.489492667232627e-05} +{"train_loss": 0.014802776277065277, "grad_norm": 0.06696814298629761, "global_step": 39631, "epoch": 373, "lr": 1.489281630418809e-05} +{"train_loss": 0.013834095560014248, "grad_norm": 0.07313171029090881, "global_step": 39632, "epoch": 373, "lr": 1.4890706059401298e-05} +{"train_loss": 0.01391143724322319, "grad_norm": 0.10615558177232742, "global_step": 39633, "epoch": 373, "lr": 1.4888595937973282e-05} +{"train_loss": 0.016077112406492233, "grad_norm": 0.06756697595119476, "global_step": 39634, "epoch": 373, "lr": 1.4886485939911471e-05} +{"train_loss": 0.015656283125281334, "grad_norm": 0.06689418107271194, "global_step": 39635, "epoch": 373, "lr": 1.4884376065223294e-05} +{"train_loss": 0.013573313131928444, "grad_norm": 0.05195192992687225, "global_step": 39636, "epoch": 373, "lr": 1.4882266313916148e-05} +{"train_loss": 0.018298976123332977, "grad_norm": 0.10850747674703598, "global_step": 39637, "epoch": 373, "lr": 1.4880156685997432e-05} +{"train_loss": 0.015744950622320175, "grad_norm": 0.08082544803619385, "global_step": 39638, "epoch": 373, "lr": 1.4878047181474568e-05} +{"train_loss": 0.01450573094189167, "grad_norm": 0.0764750987291336, "global_step": 39639, "epoch": 373, "lr": 1.4875937800354988e-05} +{"train_loss": 0.015988962724804878, "grad_norm": 0.10376520454883575, "global_step": 39640, "epoch": 373, "lr": 1.4873828542646074e-05} +{"train_loss": 0.01770329847931862, "grad_norm": 0.08127019554376602, "global_step": 39641, "epoch": 373, "lr": 1.4871719408355267e-05} +{"train_loss": 0.016011618077754974, "grad_norm": 0.08085577934980392, "global_step": 39642, "epoch": 373, "lr": 1.4869610397489958e-05} +{"train_loss": 0.015057307518667207, "grad_norm": 0.11456208676099777, "global_step": 39643, "epoch": 373, "lr": 1.4867501510057546e-05, "val_loss": 0.019104132428765297} +{"train_loss": 0.013508230447769165, "grad_norm": 0.0861348956823349, "global_step": 39644, "epoch": 374, "lr": 1.4865392746065454e-05} +{"train_loss": 0.014545447193086147, "grad_norm": 0.062538281083107, "global_step": 39645, "epoch": 374, "lr": 1.4863284105521108e-05} +{"train_loss": 0.015688277781009674, "grad_norm": 0.06766016036272049, "global_step": 39646, "epoch": 374, "lr": 1.4861175588431885e-05} +{"train_loss": 0.016476020216941833, "grad_norm": 0.0937284305691719, "global_step": 39647, "epoch": 374, "lr": 1.4859067194805231e-05} +{"train_loss": 0.016583764925599098, "grad_norm": 0.09196792542934418, "global_step": 39648, "epoch": 374, "lr": 1.4856958924648511e-05} +{"train_loss": 0.01637738198041916, "grad_norm": 0.07208790630102158, "global_step": 39649, "epoch": 374, "lr": 1.485485077796917e-05} +{"train_loss": 0.016173146665096283, "grad_norm": 0.08572308719158173, "global_step": 39650, "epoch": 374, "lr": 1.4852742754774585e-05} +{"train_loss": 0.014494990929961205, "grad_norm": 0.06996233016252518, "global_step": 39651, "epoch": 374, "lr": 1.4850634855072177e-05} +{"train_loss": 0.014756720513105392, "grad_norm": 0.0710568055510521, "global_step": 39652, "epoch": 374, "lr": 1.4848527078869368e-05} +{"train_loss": 0.015532055869698524, "grad_norm": 0.07768514007329941, "global_step": 39653, "epoch": 374, "lr": 1.4846419426173548e-05} +{"train_loss": 0.013271735981106758, "grad_norm": 0.07170384377241135, "global_step": 39654, "epoch": 374, "lr": 1.4844311896992102e-05} +{"train_loss": 0.014189764857292175, "grad_norm": 0.10510063171386719, "global_step": 39655, "epoch": 374, "lr": 1.4842204491332462e-05} +{"train_loss": 0.013097170740365982, "grad_norm": 0.0870891660451889, "global_step": 39656, "epoch": 374, "lr": 1.4840097209202036e-05} +{"train_loss": 0.016072101891040802, "grad_norm": 0.12247555702924728, "global_step": 39657, "epoch": 374, "lr": 1.4837990050608208e-05} +{"train_loss": 0.01869146153330803, "grad_norm": 0.12069359421730042, "global_step": 39658, "epoch": 374, "lr": 1.48358830155584e-05} +{"train_loss": 0.014566821977496147, "grad_norm": 0.07298734784126282, "global_step": 39659, "epoch": 374, "lr": 1.483377610406001e-05} +{"train_loss": 0.01609845831990242, "grad_norm": 0.07785966247320175, "global_step": 39660, "epoch": 374, "lr": 1.4831669316120422e-05} +{"train_loss": 0.01720106229186058, "grad_norm": 0.07107505947351456, "global_step": 39661, "epoch": 374, "lr": 1.4829562651747054e-05} +{"train_loss": 0.013719949871301651, "grad_norm": 0.06281683593988419, "global_step": 39662, "epoch": 374, "lr": 1.4827456110947324e-05} +{"train_loss": 0.013983109965920448, "grad_norm": 0.0651521310210228, "global_step": 39663, "epoch": 374, "lr": 1.4825349693728602e-05} +{"train_loss": 0.014570070430636406, "grad_norm": 0.1276877075433731, "global_step": 39664, "epoch": 374, "lr": 1.4823243400098318e-05} +{"train_loss": 0.013258639723062515, "grad_norm": 0.06719706207513809, "global_step": 39665, "epoch": 374, "lr": 1.4821137230063843e-05} +{"train_loss": 0.012857316061854362, "grad_norm": 0.06723278760910034, "global_step": 39666, "epoch": 374, "lr": 1.4819031183632609e-05} +{"train_loss": 0.012852678075432777, "grad_norm": 0.055606093257665634, "global_step": 39667, "epoch": 374, "lr": 1.4816925260811982e-05} +{"train_loss": 0.016245678067207336, "grad_norm": 0.06398885697126389, "global_step": 39668, "epoch": 374, "lr": 1.4814819461609381e-05} +{"train_loss": 0.015598220750689507, "grad_norm": 0.08525527268648148, "global_step": 39669, "epoch": 374, "lr": 1.4812713786032218e-05} +{"train_loss": 0.0137053607031703, "grad_norm": 0.08411289751529694, "global_step": 39670, "epoch": 374, "lr": 1.4810608234087876e-05} +{"train_loss": 0.014630572870373726, "grad_norm": 0.08496356010437012, "global_step": 39671, "epoch": 374, "lr": 1.4808502805783736e-05} +{"train_loss": 0.01700381003320217, "grad_norm": 0.07816850394010544, "global_step": 39672, "epoch": 374, "lr": 1.4806397501127217e-05} +{"train_loss": 0.016096387058496475, "grad_norm": 0.07045004516839981, "global_step": 39673, "epoch": 374, "lr": 1.4804292320125723e-05} +{"train_loss": 0.015461131930351257, "grad_norm": 0.13283635675907135, "global_step": 39674, "epoch": 374, "lr": 1.4802187262786627e-05} +{"train_loss": 0.015586142428219318, "grad_norm": 0.09044001251459122, "global_step": 39675, "epoch": 374, "lr": 1.4800082329117337e-05} +{"train_loss": 0.014370065182447433, "grad_norm": 0.10043735057115555, "global_step": 39676, "epoch": 374, "lr": 1.4797977519125262e-05} +{"train_loss": 0.016187656670808792, "grad_norm": 0.12296789139509201, "global_step": 39677, "epoch": 374, "lr": 1.4795872832817791e-05} +{"train_loss": 0.015834901481866837, "grad_norm": 0.08838600665330887, "global_step": 39678, "epoch": 374, "lr": 1.479376827020229e-05} +{"train_loss": 0.01316303201019764, "grad_norm": 0.05038033798336983, "global_step": 39679, "epoch": 374, "lr": 1.4791663831286184e-05} +{"train_loss": 0.018399208784103394, "grad_norm": 0.08529293537139893, "global_step": 39680, "epoch": 374, "lr": 1.4789559516076872e-05} +{"train_loss": 0.016437338665127754, "grad_norm": 0.07155266404151917, "global_step": 39681, "epoch": 374, "lr": 1.4787455324581723e-05} +{"train_loss": 0.015261149033904076, "grad_norm": 0.15649867057800293, "global_step": 39682, "epoch": 374, "lr": 1.478535125680815e-05} +{"train_loss": 0.013663794845342636, "grad_norm": 0.060313962399959564, "global_step": 39683, "epoch": 374, "lr": 1.4783247312763542e-05} +{"train_loss": 0.013090908527374268, "grad_norm": 0.06469350308179855, "global_step": 39684, "epoch": 374, "lr": 1.4781143492455273e-05} +{"train_loss": 0.015591383911669254, "grad_norm": 0.0926424041390419, "global_step": 39685, "epoch": 374, "lr": 1.4779039795890753e-05} +{"train_loss": 0.015966249629855156, "grad_norm": 0.08733460307121277, "global_step": 39686, "epoch": 374, "lr": 1.4776936223077381e-05} +{"train_loss": 0.015786845237016678, "grad_norm": 0.10431214421987534, "global_step": 39687, "epoch": 374, "lr": 1.4774832774022523e-05} +{"train_loss": 0.01377446111291647, "grad_norm": 0.0719623863697052, "global_step": 39688, "epoch": 374, "lr": 1.4772729448733603e-05} +{"train_loss": 0.014184419997036457, "grad_norm": 0.10376036912202835, "global_step": 39689, "epoch": 374, "lr": 1.4770626247217967e-05} +{"train_loss": 0.01293928548693657, "grad_norm": 0.07629499584436417, "global_step": 39690, "epoch": 374, "lr": 1.476852316948305e-05} +{"train_loss": 0.015746859833598137, "grad_norm": 0.10050683468580246, "global_step": 39691, "epoch": 374, "lr": 1.4766420215536208e-05} +{"train_loss": 0.01841083914041519, "grad_norm": 0.06953047960996628, "global_step": 39692, "epoch": 374, "lr": 1.4764317385384841e-05} +{"train_loss": 0.014424708671867847, "grad_norm": 0.06874047964811325, "global_step": 39693, "epoch": 374, "lr": 1.4762214679036352e-05} +{"train_loss": 0.015105430968105793, "grad_norm": 0.077192023396492, "global_step": 39694, "epoch": 374, "lr": 1.4760112096498119e-05} +{"train_loss": 0.017231659963726997, "grad_norm": 0.09256322681903839, "global_step": 39695, "epoch": 374, "lr": 1.475800963777751e-05} +{"train_loss": 0.013308683410286903, "grad_norm": 0.0807061418890953, "global_step": 39696, "epoch": 374, "lr": 1.4755907302881927e-05} +{"train_loss": 0.013255643658339977, "grad_norm": 0.04832659661769867, "global_step": 39697, "epoch": 374, "lr": 1.4753805091818773e-05} +{"train_loss": 0.016271714121103287, "grad_norm": 0.07088249921798706, "global_step": 39698, "epoch": 374, "lr": 1.4751703004595407e-05} +{"train_loss": 0.016653480008244514, "grad_norm": 0.11799730360507965, "global_step": 39699, "epoch": 374, "lr": 1.4749601041219247e-05} +{"train_loss": 0.01607944816350937, "grad_norm": 0.07664615660905838, "global_step": 39700, "epoch": 374, "lr": 1.4747499201697652e-05} +{"train_loss": 0.013850219547748566, "grad_norm": 0.09050074964761734, "global_step": 39701, "epoch": 374, "lr": 1.4745397486038005e-05} +{"train_loss": 0.01469840295612812, "grad_norm": 0.06289936602115631, "global_step": 39702, "epoch": 374, "lr": 1.4743295894247693e-05} +{"train_loss": 0.01356775313615799, "grad_norm": 0.07400139421224594, "global_step": 39703, "epoch": 374, "lr": 1.4741194426334126e-05} +{"train_loss": 0.016780640929937363, "grad_norm": 0.06952174007892609, "global_step": 39704, "epoch": 374, "lr": 1.473909308230465e-05} +{"train_loss": 0.013512170873582363, "grad_norm": 0.055944763123989105, "global_step": 39705, "epoch": 374, "lr": 1.4736991862166682e-05} +{"train_loss": 0.01452879048883915, "grad_norm": 0.1783803105354309, "global_step": 39706, "epoch": 374, "lr": 1.4734890765927579e-05} +{"train_loss": 0.01484670676290989, "grad_norm": 0.07793624699115753, "global_step": 39707, "epoch": 374, "lr": 1.4732789793594742e-05} +{"train_loss": 0.015705406665802002, "grad_norm": 0.06746404618024826, "global_step": 39708, "epoch": 374, "lr": 1.4730688945175531e-05} +{"train_loss": 0.01616806536912918, "grad_norm": 0.059907134622335434, "global_step": 39709, "epoch": 374, "lr": 1.4728588220677342e-05} +{"train_loss": 0.013382280245423317, "grad_norm": 0.13656769692897797, "global_step": 39710, "epoch": 374, "lr": 1.472648762010757e-05} +{"train_loss": 0.018224095925688744, "grad_norm": 0.14304161071777344, "global_step": 39711, "epoch": 374, "lr": 1.4724387143473578e-05} +{"train_loss": 0.013386101461946964, "grad_norm": 0.08173990994691849, "global_step": 39712, "epoch": 374, "lr": 1.4722286790782736e-05} +{"train_loss": 0.011493058875203133, "grad_norm": 0.07343532890081406, "global_step": 39713, "epoch": 374, "lr": 1.4720186562042437e-05} +{"train_loss": 0.016281697899103165, "grad_norm": 0.09324494749307632, "global_step": 39714, "epoch": 374, "lr": 1.4718086457260078e-05} +{"train_loss": 0.015811333432793617, "grad_norm": 0.12356997281312943, "global_step": 39715, "epoch": 374, "lr": 1.4715986476442995e-05} +{"train_loss": 0.0157415010035038, "grad_norm": 0.08273743093013763, "global_step": 39716, "epoch": 374, "lr": 1.4713886619598599e-05} +{"train_loss": 0.01474026683717966, "grad_norm": 0.06765700876712799, "global_step": 39717, "epoch": 374, "lr": 1.4711786886734285e-05} +{"train_loss": 0.016007380560040474, "grad_norm": 0.12439284473657608, "global_step": 39718, "epoch": 374, "lr": 1.4709687277857376e-05} +{"train_loss": 0.016104450449347496, "grad_norm": 0.08147234469652176, "global_step": 39719, "epoch": 374, "lr": 1.470758779297528e-05} +{"train_loss": 0.011938134208321571, "grad_norm": 0.13269217312335968, "global_step": 39720, "epoch": 374, "lr": 1.4705488432095388e-05} +{"train_loss": 0.01585729792714119, "grad_norm": 0.07829036563634872, "global_step": 39721, "epoch": 374, "lr": 1.4703389195225049e-05} +{"train_loss": 0.014021916314959526, "grad_norm": 0.058219294995069504, "global_step": 39722, "epoch": 374, "lr": 1.4701290082371644e-05} +{"train_loss": 0.015306847169995308, "grad_norm": 0.05922224000096321, "global_step": 39723, "epoch": 374, "lr": 1.469919109354257e-05} +{"train_loss": 0.016635198146104813, "grad_norm": 0.1012345403432846, "global_step": 39724, "epoch": 374, "lr": 1.4697092228745191e-05} +{"train_loss": 0.015057109296321869, "grad_norm": 0.07544871419668198, "global_step": 39725, "epoch": 374, "lr": 1.4694993487986858e-05} +{"train_loss": 0.01659160479903221, "grad_norm": 0.06601354479789734, "global_step": 39726, "epoch": 374, "lr": 1.4692894871274964e-05} +{"train_loss": 0.015366261824965477, "grad_norm": 0.06490454077720642, "global_step": 39727, "epoch": 374, "lr": 1.46907963786169e-05} +{"train_loss": 0.01285877451300621, "grad_norm": 0.05266149342060089, "global_step": 39728, "epoch": 374, "lr": 1.4688698010020008e-05} +{"train_loss": 0.011767685413360596, "grad_norm": 0.09501209855079651, "global_step": 39729, "epoch": 374, "lr": 1.4686599765491687e-05} +{"train_loss": 0.014791004359722137, "grad_norm": 0.06057698279619217, "global_step": 39730, "epoch": 374, "lr": 1.4684501645039283e-05} +{"train_loss": 0.016132574528455734, "grad_norm": 0.13377980887889862, "global_step": 39731, "epoch": 374, "lr": 1.4682403648670196e-05} +{"train_loss": 0.016031142324209213, "grad_norm": 0.0664585754275322, "global_step": 39732, "epoch": 374, "lr": 1.4680305776391768e-05} +{"train_loss": 0.01394752785563469, "grad_norm": 0.058044031262397766, "global_step": 39733, "epoch": 374, "lr": 1.4678208028211387e-05} +{"train_loss": 0.013667349703609943, "grad_norm": 0.06757120043039322, "global_step": 39734, "epoch": 374, "lr": 1.4676110404136433e-05} +{"train_loss": 0.016143478453159332, "grad_norm": 0.07308631390333176, "global_step": 39735, "epoch": 374, "lr": 1.4674012904174273e-05} +{"train_loss": 0.012665452435612679, "grad_norm": 0.10701905190944672, "global_step": 39736, "epoch": 374, "lr": 1.4671915528332247e-05} +{"train_loss": 0.0151949068531394, "grad_norm": 0.060414768755435944, "global_step": 39737, "epoch": 374, "lr": 1.4669818276617747e-05} +{"train_loss": 0.015785884112119675, "grad_norm": 0.09277597814798355, "global_step": 39738, "epoch": 374, "lr": 1.466772114903816e-05} +{"train_loss": 0.016123607754707336, "grad_norm": 0.08672849833965302, "global_step": 39739, "epoch": 374, "lr": 1.4665624145600815e-05} +{"train_loss": 0.015497575514018536, "grad_norm": 0.06736729294061661, "global_step": 39740, "epoch": 374, "lr": 1.4663527266313105e-05} +{"train_loss": 0.015714706853032112, "grad_norm": 0.06482888758182526, "global_step": 39741, "epoch": 374, "lr": 1.4661430511182422e-05} +{"train_loss": 0.016101108863949776, "grad_norm": 0.11726368218660355, "global_step": 39742, "epoch": 374, "lr": 1.465933388021607e-05} +{"train_loss": 0.014177875593304634, "grad_norm": 0.07084588706493378, "global_step": 39743, "epoch": 374, "lr": 1.4657237373421451e-05} +{"train_loss": 0.013733583502471447, "grad_norm": 0.0833052471280098, "global_step": 39744, "epoch": 374, "lr": 1.4655140990805949e-05} +{"train_loss": 0.014666624367237091, "grad_norm": 0.09575800597667694, "global_step": 39745, "epoch": 374, "lr": 1.4653044732376891e-05} +{"train_loss": 0.018741942942142487, "grad_norm": 0.09600904583930969, "global_step": 39746, "epoch": 374, "lr": 1.4650948598141662e-05} +{"train_loss": 0.01397440955042839, "grad_norm": 0.14212526381015778, "global_step": 39747, "epoch": 374, "lr": 1.4648852588107642e-05} +{"train_loss": 0.011629346758127213, "grad_norm": 0.06766809523105621, "global_step": 39748, "epoch": 374, "lr": 1.4646756702282182e-05} +{"train_loss": 0.015039623512904037, "grad_norm": 0.06755312532186508, "global_step": 39749, "epoch": 374, "lr": 1.4644660940672627e-05, "val_loss": 0.019131971523165703} +{"train_loss": 0.015303125604987144, "grad_norm": 0.08128640800714493, "global_step": 39750, "epoch": 375, "lr": 1.464256530328636e-05} +{"train_loss": 0.015521387569606304, "grad_norm": 0.08686988055706024, "global_step": 39751, "epoch": 375, "lr": 1.4640469790130756e-05} +{"train_loss": 0.014777673408389091, "grad_norm": 0.07654134929180145, "global_step": 39752, "epoch": 375, "lr": 1.4638374401213162e-05} +{"train_loss": 0.012672880664467812, "grad_norm": 0.08002462983131409, "global_step": 39753, "epoch": 375, "lr": 1.4636279136540932e-05} +{"train_loss": 0.01621543988585472, "grad_norm": 0.0704546868801117, "global_step": 39754, "epoch": 375, "lr": 1.4634183996121432e-05} +{"train_loss": 0.01568460278213024, "grad_norm": 0.08559457957744598, "global_step": 39755, "epoch": 375, "lr": 1.4632088979962045e-05} +{"train_loss": 0.013927493244409561, "grad_norm": 0.05575380101799965, "global_step": 39756, "epoch": 375, "lr": 1.4629994088070104e-05} +{"train_loss": 0.014129064045846462, "grad_norm": 0.05551769584417343, "global_step": 39757, "epoch": 375, "lr": 1.462789932045298e-05} +{"train_loss": 0.014560684561729431, "grad_norm": 0.09461933374404907, "global_step": 39758, "epoch": 375, "lr": 1.4625804677118066e-05} +{"train_loss": 0.01331038773059845, "grad_norm": 0.10153492540121078, "global_step": 39759, "epoch": 375, "lr": 1.4623710158072663e-05} +{"train_loss": 0.015922661870718002, "grad_norm": 0.09941093623638153, "global_step": 39760, "epoch": 375, "lr": 1.4621615763324154e-05} +{"train_loss": 0.01686008647084236, "grad_norm": 0.09334047138690948, "global_step": 39761, "epoch": 375, "lr": 1.4619521492879917e-05} +{"train_loss": 0.01500235591083765, "grad_norm": 0.09556616842746735, "global_step": 39762, "epoch": 375, "lr": 1.4617427346747281e-05} +{"train_loss": 0.013760700821876526, "grad_norm": 0.1337253749370575, "global_step": 39763, "epoch": 375, "lr": 1.4615333324933616e-05} +{"train_loss": 0.015063787810504436, "grad_norm": 0.10241851210594177, "global_step": 39764, "epoch": 375, "lr": 1.4613239427446296e-05} +{"train_loss": 0.015166041441261768, "grad_norm": 0.07106372714042664, "global_step": 39765, "epoch": 375, "lr": 1.4611145654292668e-05} +{"train_loss": 0.01383849885314703, "grad_norm": 0.11948425322771072, "global_step": 39766, "epoch": 375, "lr": 1.4609052005480062e-05} +{"train_loss": 0.014217432588338852, "grad_norm": 0.09539026767015457, "global_step": 39767, "epoch": 375, "lr": 1.460695848101586e-05} +{"train_loss": 0.015623364597558975, "grad_norm": 0.1201605275273323, "global_step": 39768, "epoch": 375, "lr": 1.4604865080907432e-05} +{"train_loss": 0.013709552586078644, "grad_norm": 0.12075761705636978, "global_step": 39769, "epoch": 375, "lr": 1.4602771805162097e-05} +{"train_loss": 0.014811724424362183, "grad_norm": 0.06269891560077667, "global_step": 39770, "epoch": 375, "lr": 1.4600678653787243e-05} +{"train_loss": 0.015254257246851921, "grad_norm": 0.05917862057685852, "global_step": 39771, "epoch": 375, "lr": 1.459858562679019e-05} +{"train_loss": 0.013411620631814003, "grad_norm": 0.0786500945687294, "global_step": 39772, "epoch": 375, "lr": 1.4596492724178335e-05} +{"train_loss": 0.0162738598883152, "grad_norm": 0.09413953870534897, "global_step": 39773, "epoch": 375, "lr": 1.4594399945958987e-05} +{"train_loss": 0.016207218170166016, "grad_norm": 0.06802178919315338, "global_step": 39774, "epoch": 375, "lr": 1.4592307292139524e-05} +{"train_loss": 0.017740566283464432, "grad_norm": 0.1353720724582672, "global_step": 39775, "epoch": 375, "lr": 1.4590214762727305e-05} +{"train_loss": 0.01552818063646555, "grad_norm": 0.10838070511817932, "global_step": 39776, "epoch": 375, "lr": 1.4588122357729678e-05} +{"train_loss": 0.016144221648573875, "grad_norm": 0.08115990459918976, "global_step": 39777, "epoch": 375, "lr": 1.4586030077153967e-05} +{"train_loss": 0.01460324041545391, "grad_norm": 0.08160770684480667, "global_step": 39778, "epoch": 375, "lr": 1.458393792100755e-05} +{"train_loss": 0.014488688670098782, "grad_norm": 0.061240676790475845, "global_step": 39779, "epoch": 375, "lr": 1.4581845889297785e-05} +{"train_loss": 0.015789709985256195, "grad_norm": 0.05781678482890129, "global_step": 39780, "epoch": 375, "lr": 1.4579753982031997e-05} +{"train_loss": 0.014124633744359016, "grad_norm": 0.10388682782649994, "global_step": 39781, "epoch": 375, "lr": 1.457766219921755e-05} +{"train_loss": 0.014664242044091225, "grad_norm": 0.06668951362371445, "global_step": 39782, "epoch": 375, "lr": 1.4575570540861827e-05} +{"train_loss": 0.014572320505976677, "grad_norm": 0.07484273612499237, "global_step": 39783, "epoch": 375, "lr": 1.4573479006972107e-05} +{"train_loss": 0.014617565087974072, "grad_norm": 0.09073935449123383, "global_step": 39784, "epoch": 375, "lr": 1.457138759755578e-05} +{"train_loss": 0.013622386381030083, "grad_norm": 0.07019949704408646, "global_step": 39785, "epoch": 375, "lr": 1.4569296312620202e-05} +{"train_loss": 0.014969488605856895, "grad_norm": 0.13532115519046783, "global_step": 39786, "epoch": 375, "lr": 1.4567205152172691e-05} +{"train_loss": 0.01764170080423355, "grad_norm": 0.09265769273042679, "global_step": 39787, "epoch": 375, "lr": 1.4565114116220619e-05} +{"train_loss": 0.012043289840221405, "grad_norm": 0.061564795672893524, "global_step": 39788, "epoch": 375, "lr": 1.456302320477133e-05} +{"train_loss": 0.014494064263999462, "grad_norm": 0.10773363709449768, "global_step": 39789, "epoch": 375, "lr": 1.4560932417832178e-05} +{"train_loss": 0.014419689774513245, "grad_norm": 0.08959535509347916, "global_step": 39790, "epoch": 375, "lr": 1.4558841755410473e-05} +{"train_loss": 0.015648072585463524, "grad_norm": 0.07207680493593216, "global_step": 39791, "epoch": 375, "lr": 1.4556751217513592e-05} +{"train_loss": 0.01389410998672247, "grad_norm": 0.06824848800897598, "global_step": 39792, "epoch": 375, "lr": 1.4554660804148884e-05} +{"train_loss": 0.014075173065066338, "grad_norm": 0.17313988506793976, "global_step": 39793, "epoch": 375, "lr": 1.4552570515323671e-05} +{"train_loss": 0.016480544582009315, "grad_norm": 0.1052909642457962, "global_step": 39794, "epoch": 375, "lr": 1.4550480351045325e-05} +{"train_loss": 0.013804014772176743, "grad_norm": 0.14693988859653473, "global_step": 39795, "epoch": 375, "lr": 1.4548390311321158e-05} +{"train_loss": 0.012611709535121918, "grad_norm": 0.06799181550741196, "global_step": 39796, "epoch": 375, "lr": 1.4546300396158541e-05} +{"train_loss": 0.014258217066526413, "grad_norm": 0.07456684857606888, "global_step": 39797, "epoch": 375, "lr": 1.4544210605564795e-05} +{"train_loss": 0.014866113662719727, "grad_norm": 0.09825867414474487, "global_step": 39798, "epoch": 375, "lr": 1.4542120939547277e-05} +{"train_loss": 0.015660837292671204, "grad_norm": 0.12096456438302994, "global_step": 39799, "epoch": 375, "lr": 1.4540031398113335e-05} +{"train_loss": 0.017642837017774582, "grad_norm": 0.09596429765224457, "global_step": 39800, "epoch": 375, "lr": 1.4537941981270303e-05} +{"train_loss": 0.013138394802808762, "grad_norm": 0.052399661391973495, "global_step": 39801, "epoch": 375, "lr": 1.4535852689025508e-05} +{"train_loss": 0.015510311350226402, "grad_norm": 0.06454958021640778, "global_step": 39802, "epoch": 375, "lr": 1.4533763521386318e-05} +{"train_loss": 0.014761731028556824, "grad_norm": 0.06267479807138443, "global_step": 39803, "epoch": 375, "lr": 1.4531674478360042e-05} +{"train_loss": 0.016550924628973007, "grad_norm": 0.13226932287216187, "global_step": 39804, "epoch": 375, "lr": 1.452958555995404e-05} +{"train_loss": 0.014671443030238152, "grad_norm": 0.06525666266679764, "global_step": 39805, "epoch": 375, "lr": 1.4527496766175658e-05} +{"train_loss": 0.016781263053417206, "grad_norm": 0.11251400411128998, "global_step": 39806, "epoch": 375, "lr": 1.4525408097032234e-05} +{"train_loss": 0.016994940117001534, "grad_norm": 0.06840045005083084, "global_step": 39807, "epoch": 375, "lr": 1.4523319552531078e-05} +{"train_loss": 0.014661623165011406, "grad_norm": 0.07022003084421158, "global_step": 39808, "epoch": 375, "lr": 1.4521231132679546e-05} +{"train_loss": 0.01612911932170391, "grad_norm": 0.07993780821561813, "global_step": 39809, "epoch": 375, "lr": 1.4519142837485e-05} +{"train_loss": 0.0173442792147398, "grad_norm": 0.08446183055639267, "global_step": 39810, "epoch": 375, "lr": 1.4517054666954738e-05} +{"train_loss": 0.01636858657002449, "grad_norm": 0.07607711851596832, "global_step": 39811, "epoch": 375, "lr": 1.4514966621096132e-05} +{"train_loss": 0.01457932684570551, "grad_norm": 0.11898162215948105, "global_step": 39812, "epoch": 375, "lr": 1.4512878699916482e-05} +{"train_loss": 0.012952039018273354, "grad_norm": 0.08029469102621078, "global_step": 39813, "epoch": 375, "lr": 1.451079090342316e-05} +{"train_loss": 0.013621723279356956, "grad_norm": 0.13234728574752808, "global_step": 39814, "epoch": 375, "lr": 1.4508703231623472e-05} +{"train_loss": 0.014112100005149841, "grad_norm": 0.060575444251298904, "global_step": 39815, "epoch": 375, "lr": 1.4506615684524766e-05} +{"train_loss": 0.014583664946258068, "grad_norm": 0.061917588114738464, "global_step": 39816, "epoch": 375, "lr": 1.4504528262134393e-05} +{"train_loss": 0.015670593827962875, "grad_norm": 0.0773533508181572, "global_step": 39817, "epoch": 375, "lr": 1.4502440964459669e-05} +{"train_loss": 0.015174132771790028, "grad_norm": 0.06959502398967743, "global_step": 39818, "epoch": 375, "lr": 1.4500353791507909e-05} +{"train_loss": 0.013418283313512802, "grad_norm": 0.07441727072000504, "global_step": 39819, "epoch": 375, "lr": 1.4498266743286492e-05} +{"train_loss": 0.015636088326573372, "grad_norm": 0.06061724200844765, "global_step": 39820, "epoch": 375, "lr": 1.4496179819802708e-05} +{"train_loss": 0.013760710135102272, "grad_norm": 0.06533414125442505, "global_step": 39821, "epoch": 375, "lr": 1.4494093021063903e-05} +{"train_loss": 0.015316501259803772, "grad_norm": 0.07828934490680695, "global_step": 39822, "epoch": 375, "lr": 1.4492006347077436e-05} +{"train_loss": 0.01489414181560278, "grad_norm": 0.07250627130270004, "global_step": 39823, "epoch": 375, "lr": 1.4489919797850615e-05} +{"train_loss": 0.014355954714119434, "grad_norm": 0.08210098743438721, "global_step": 39824, "epoch": 375, "lr": 1.4487833373390759e-05} +{"train_loss": 0.015613192692399025, "grad_norm": 0.05471005663275719, "global_step": 39825, "epoch": 375, "lr": 1.4485747073705208e-05} +{"train_loss": 0.015084447339177132, "grad_norm": 0.23152069747447968, "global_step": 39826, "epoch": 375, "lr": 1.4483660898801316e-05} +{"train_loss": 0.01671169139444828, "grad_norm": 0.08292718976736069, "global_step": 39827, "epoch": 375, "lr": 1.4481574848686375e-05} +{"train_loss": 0.01423681154847145, "grad_norm": 0.08749837428331375, "global_step": 39828, "epoch": 375, "lr": 1.4479488923367734e-05} +{"train_loss": 0.014190565794706345, "grad_norm": 0.06488203257322311, "global_step": 39829, "epoch": 375, "lr": 1.447740312285274e-05} +{"train_loss": 0.015232227742671967, "grad_norm": 0.0921250730752945, "global_step": 39830, "epoch": 375, "lr": 1.4475317447148701e-05} +{"train_loss": 0.01426786556839943, "grad_norm": 0.13610579073429108, "global_step": 39831, "epoch": 375, "lr": 1.4473231896262935e-05} +{"train_loss": 0.015323733910918236, "grad_norm": 0.06754190474748611, "global_step": 39832, "epoch": 375, "lr": 1.4471146470202778e-05} +{"train_loss": 0.01425241306424141, "grad_norm": 0.08114149421453476, "global_step": 39833, "epoch": 375, "lr": 1.4469061168975579e-05} +{"train_loss": 0.01420387253165245, "grad_norm": 0.10153716057538986, "global_step": 39834, "epoch": 375, "lr": 1.4466975992588628e-05} +{"train_loss": 0.015664462000131607, "grad_norm": 0.07283829897642136, "global_step": 39835, "epoch": 375, "lr": 1.4464890941049292e-05} +{"train_loss": 0.01649506762623787, "grad_norm": 0.14668837189674377, "global_step": 39836, "epoch": 375, "lr": 1.4462806014364855e-05} +{"train_loss": 0.01653311401605606, "grad_norm": 0.05396309494972229, "global_step": 39837, "epoch": 375, "lr": 1.4460721212542684e-05} +{"train_loss": 0.017271213233470917, "grad_norm": 0.06836358457803726, "global_step": 39838, "epoch": 375, "lr": 1.4458636535590058e-05} +{"train_loss": 0.018695279955863953, "grad_norm": 0.07507466524839401, "global_step": 39839, "epoch": 375, "lr": 1.4456551983514333e-05} +{"train_loss": 0.014569064602255821, "grad_norm": 0.09717272222042084, "global_step": 39840, "epoch": 375, "lr": 1.4454467556322842e-05} +{"train_loss": 0.015047270804643631, "grad_norm": 0.14573892951011658, "global_step": 39841, "epoch": 375, "lr": 1.4452383254022895e-05} +{"train_loss": 0.012655941769480705, "grad_norm": 0.08197803795337677, "global_step": 39842, "epoch": 375, "lr": 1.4450299076621798e-05} +{"train_loss": 0.015879936516284943, "grad_norm": 0.08754212409257889, "global_step": 39843, "epoch": 375, "lr": 1.4448215024126904e-05} +{"train_loss": 0.01471768505871296, "grad_norm": 0.10109144449234009, "global_step": 39844, "epoch": 375, "lr": 1.4446131096545507e-05} +{"train_loss": 0.015763787552714348, "grad_norm": 0.09331846982240677, "global_step": 39845, "epoch": 375, "lr": 1.4444047293884938e-05} +{"train_loss": 0.01610318198800087, "grad_norm": 0.06396456062793732, "global_step": 39846, "epoch": 375, "lr": 1.4441963616152542e-05} +{"train_loss": 0.016198165714740753, "grad_norm": 0.057976916432380676, "global_step": 39847, "epoch": 375, "lr": 1.4439880063355626e-05} +{"train_loss": 0.014823307283222675, "grad_norm": 0.07096374779939651, "global_step": 39848, "epoch": 375, "lr": 1.4437796635501488e-05} +{"train_loss": 0.015213332138955593, "grad_norm": 0.0676664412021637, "global_step": 39849, "epoch": 375, "lr": 1.4435713332597472e-05} +{"train_loss": 0.016001375392079353, "grad_norm": 0.12197437137365341, "global_step": 39850, "epoch": 375, "lr": 1.4433630154650902e-05} +{"train_loss": 0.012917103245854378, "grad_norm": 0.07014310359954834, "global_step": 39851, "epoch": 375, "lr": 1.4431547101669074e-05} +{"train_loss": 0.015026520937681198, "grad_norm": 0.10379178076982498, "global_step": 39852, "epoch": 375, "lr": 1.442946417365934e-05} +{"train_loss": 0.01219257339835167, "grad_norm": 0.07719551771879196, "global_step": 39853, "epoch": 375, "lr": 1.4427381370628985e-05} +{"train_loss": 0.017476947978138924, "grad_norm": 0.11457053571939468, "global_step": 39854, "epoch": 375, "lr": 1.4425298692585355e-05} +{"train_loss": 0.015045458486057678, "grad_norm": 0.1067749410867691, "global_step": 39855, "epoch": 375, "lr": 1.4423216139535734e-05, "val_loss": 0.019679037854075432} +{"train_loss": 0.017192689701914787, "grad_norm": 0.10545018315315247, "global_step": 39856, "epoch": 376, "lr": 1.4421133711487467e-05} +{"train_loss": 0.01379203237593174, "grad_norm": 0.09497939795255661, "global_step": 39857, "epoch": 376, "lr": 1.4419051408447876e-05} +{"train_loss": 0.016963403671979904, "grad_norm": 0.10775037854909897, "global_step": 39858, "epoch": 376, "lr": 1.4416969230424259e-05} +{"train_loss": 0.01506771519780159, "grad_norm": 0.12927165627479553, "global_step": 39859, "epoch": 376, "lr": 1.441488717742393e-05} +{"train_loss": 0.014091823250055313, "grad_norm": 0.07682003080844879, "global_step": 39860, "epoch": 376, "lr": 1.4412805249454226e-05} +{"train_loss": 0.013792278245091438, "grad_norm": 0.06388024240732193, "global_step": 39861, "epoch": 376, "lr": 1.4410723446522428e-05} +{"train_loss": 0.015924552455544472, "grad_norm": 0.10393045842647552, "global_step": 39862, "epoch": 376, "lr": 1.4408641768635872e-05} +{"train_loss": 0.01434203889220953, "grad_norm": 0.07044456899166107, "global_step": 39863, "epoch": 376, "lr": 1.440656021580189e-05} +{"train_loss": 0.012519773095846176, "grad_norm": 0.0973469465970993, "global_step": 39864, "epoch": 376, "lr": 1.4404478788027776e-05} +{"train_loss": 0.013770416378974915, "grad_norm": 0.07135002315044403, "global_step": 39865, "epoch": 376, "lr": 1.4402397485320824e-05} +{"train_loss": 0.013272076845169067, "grad_norm": 0.07175882905721664, "global_step": 39866, "epoch": 376, "lr": 1.4400316307688366e-05} +{"train_loss": 0.015561612322926521, "grad_norm": 0.11817727982997894, "global_step": 39867, "epoch": 376, "lr": 1.4398235255137733e-05} +{"train_loss": 0.013946590945124626, "grad_norm": 0.1403440684080124, "global_step": 39868, "epoch": 376, "lr": 1.4396154327676204e-05} +{"train_loss": 0.0157475546002388, "grad_norm": 0.0944783166050911, "global_step": 39869, "epoch": 376, "lr": 1.4394073525311102e-05} +{"train_loss": 0.014925654977560043, "grad_norm": 0.08534485101699829, "global_step": 39870, "epoch": 376, "lr": 1.4391992848049756e-05} +{"train_loss": 0.014049977995455265, "grad_norm": 0.09143967181444168, "global_step": 39871, "epoch": 376, "lr": 1.4389912295899466e-05} +{"train_loss": 0.016084963455796242, "grad_norm": 0.06974317133426666, "global_step": 39872, "epoch": 376, "lr": 1.4387831868867518e-05} +{"train_loss": 0.014774652197957039, "grad_norm": 0.06286416947841644, "global_step": 39873, "epoch": 376, "lr": 1.4385751566961242e-05} +{"train_loss": 0.016726145520806313, "grad_norm": 0.12276766449213028, "global_step": 39874, "epoch": 376, "lr": 1.4383671390187963e-05} +{"train_loss": 0.014292475767433643, "grad_norm": 0.0778273493051529, "global_step": 39875, "epoch": 376, "lr": 1.4381591338554956e-05} +{"train_loss": 0.0170548427850008, "grad_norm": 0.05788286030292511, "global_step": 39876, "epoch": 376, "lr": 1.4379511412069569e-05} +{"train_loss": 0.014346141368150711, "grad_norm": 0.06271910667419434, "global_step": 39877, "epoch": 376, "lr": 1.4377431610739084e-05} +{"train_loss": 0.014784052968025208, "grad_norm": 0.10743513703346252, "global_step": 39878, "epoch": 376, "lr": 1.4375351934570796e-05} +{"train_loss": 0.013090761378407478, "grad_norm": 0.0696101039648056, "global_step": 39879, "epoch": 376, "lr": 1.4373272383572028e-05} +{"train_loss": 0.012382248416543007, "grad_norm": 0.08374256640672684, "global_step": 39880, "epoch": 376, "lr": 1.437119295775009e-05} +{"train_loss": 0.016786525025963783, "grad_norm": 0.06744161248207092, "global_step": 39881, "epoch": 376, "lr": 1.4369113657112299e-05} +{"train_loss": 0.017291855067014694, "grad_norm": 0.10696735233068466, "global_step": 39882, "epoch": 376, "lr": 1.4367034481665947e-05} +{"train_loss": 0.017024554312229156, "grad_norm": 0.07303514331579208, "global_step": 39883, "epoch": 376, "lr": 1.4364955431418326e-05} +{"train_loss": 0.015965372323989868, "grad_norm": 0.07183534651994705, "global_step": 39884, "epoch": 376, "lr": 1.4362876506376766e-05} +{"train_loss": 0.01551135815680027, "grad_norm": 0.08720705658197403, "global_step": 39885, "epoch": 376, "lr": 1.4360797706548545e-05} +{"train_loss": 0.01719418726861477, "grad_norm": 0.1608828902244568, "global_step": 39886, "epoch": 376, "lr": 1.4358719031940986e-05} +{"train_loss": 0.014986667782068253, "grad_norm": 0.054933495819568634, "global_step": 39887, "epoch": 376, "lr": 1.43566404825614e-05} +{"train_loss": 0.01656121388077736, "grad_norm": 0.10822468250989914, "global_step": 39888, "epoch": 376, "lr": 1.4354562058417082e-05} +{"train_loss": 0.014616291970014572, "grad_norm": 0.11778299510478973, "global_step": 39889, "epoch": 376, "lr": 1.4352483759515311e-05} +{"train_loss": 0.013025600463151932, "grad_norm": 0.10682689398527145, "global_step": 39890, "epoch": 376, "lr": 1.4350405585863414e-05} +{"train_loss": 0.014733990654349327, "grad_norm": 0.07288718968629837, "global_step": 39891, "epoch": 376, "lr": 1.4348327537468697e-05} +{"train_loss": 0.015422960743308067, "grad_norm": 0.06808413565158844, "global_step": 39892, "epoch": 376, "lr": 1.4346249614338442e-05} +{"train_loss": 0.013557666912674904, "grad_norm": 0.0966426208615303, "global_step": 39893, "epoch": 376, "lr": 1.4344171816479962e-05} +{"train_loss": 0.014934036880731583, "grad_norm": 0.09474845230579376, "global_step": 39894, "epoch": 376, "lr": 1.434209414390057e-05} +{"train_loss": 0.015328384935855865, "grad_norm": 0.06199399381875992, "global_step": 39895, "epoch": 376, "lr": 1.4340016596607547e-05} +{"train_loss": 0.017330622300505638, "grad_norm": 0.0970686674118042, "global_step": 39896, "epoch": 376, "lr": 1.4337939174608189e-05} +{"train_loss": 0.01594998687505722, "grad_norm": 0.07360955327749252, "global_step": 39897, "epoch": 376, "lr": 1.4335861877909801e-05} +{"train_loss": 0.013111080974340439, "grad_norm": 0.09392876923084259, "global_step": 39898, "epoch": 376, "lr": 1.4333784706519698e-05} +{"train_loss": 0.01569102331995964, "grad_norm": 0.11589237302541733, "global_step": 39899, "epoch": 376, "lr": 1.4331707660445155e-05} +{"train_loss": 0.013668803498148918, "grad_norm": 0.1448068618774414, "global_step": 39900, "epoch": 376, "lr": 1.432963073969349e-05} +{"train_loss": 0.013527005910873413, "grad_norm": 0.07289590686559677, "global_step": 39901, "epoch": 376, "lr": 1.4327553944271987e-05} +{"train_loss": 0.015112306922674179, "grad_norm": 0.11957315355539322, "global_step": 39902, "epoch": 376, "lr": 1.4325477274187937e-05} +{"train_loss": 0.017314767464995384, "grad_norm": 0.09734077006578445, "global_step": 39903, "epoch": 376, "lr": 1.4323400729448644e-05} +{"train_loss": 0.01616416871547699, "grad_norm": 0.06767790019512177, "global_step": 39904, "epoch": 376, "lr": 1.4321324310061418e-05} +{"train_loss": 0.015104832127690315, "grad_norm": 0.054147690534591675, "global_step": 39905, "epoch": 376, "lr": 1.4319248016033542e-05} +{"train_loss": 0.01757446676492691, "grad_norm": 0.06824908405542374, "global_step": 39906, "epoch": 376, "lr": 1.4317171847372302e-05} +{"train_loss": 0.015653783455491066, "grad_norm": 0.08570375293493271, "global_step": 39907, "epoch": 376, "lr": 1.4315095804085e-05} +{"train_loss": 0.01269704569131136, "grad_norm": 0.061335377395153046, "global_step": 39908, "epoch": 376, "lr": 1.4313019886178941e-05} +{"train_loss": 0.014296644367277622, "grad_norm": 0.06307835131883621, "global_step": 39909, "epoch": 376, "lr": 1.4310944093661399e-05} +{"train_loss": 0.017425980418920517, "grad_norm": 0.09167058020830154, "global_step": 39910, "epoch": 376, "lr": 1.4308868426539684e-05} +{"train_loss": 0.014479327946901321, "grad_norm": 0.05348074808716774, "global_step": 39911, "epoch": 376, "lr": 1.4306792884821091e-05} +{"train_loss": 0.01706658862531185, "grad_norm": 0.06040029972791672, "global_step": 39912, "epoch": 376, "lr": 1.4304717468512912e-05} +{"train_loss": 0.015449427999556065, "grad_norm": 0.0794079601764679, "global_step": 39913, "epoch": 376, "lr": 1.430264217762241e-05} +{"train_loss": 0.015970556065440178, "grad_norm": 0.09118014574050903, "global_step": 39914, "epoch": 376, "lr": 1.4300567012156906e-05} +{"train_loss": 0.014793808571994305, "grad_norm": 0.08139605820178986, "global_step": 39915, "epoch": 376, "lr": 1.4298491972123695e-05} +{"train_loss": 0.017322681844234467, "grad_norm": 0.065265953540802, "global_step": 39916, "epoch": 376, "lr": 1.4296417057530043e-05} +{"train_loss": 0.014044582843780518, "grad_norm": 0.11259628087282181, "global_step": 39917, "epoch": 376, "lr": 1.429434226838327e-05} +{"train_loss": 0.016096606850624084, "grad_norm": 0.08810314536094666, "global_step": 39918, "epoch": 376, "lr": 1.4292267604690651e-05} +{"train_loss": 0.017164258286356926, "grad_norm": 0.10451377183198929, "global_step": 39919, "epoch": 376, "lr": 1.4290193066459457e-05} +{"train_loss": 0.015443162992596626, "grad_norm": 0.0704144760966301, "global_step": 39920, "epoch": 376, "lr": 1.4288118653696997e-05} +{"train_loss": 0.015840347856283188, "grad_norm": 0.10782074928283691, "global_step": 39921, "epoch": 376, "lr": 1.4286044366410573e-05} +{"train_loss": 0.015133139677345753, "grad_norm": 0.0950501337647438, "global_step": 39922, "epoch": 376, "lr": 1.428397020460744e-05} +{"train_loss": 0.013324369676411152, "grad_norm": 0.05711432173848152, "global_step": 39923, "epoch": 376, "lr": 1.4281896168294917e-05} +{"train_loss": 0.015041390433907509, "grad_norm": 0.06661384552717209, "global_step": 39924, "epoch": 376, "lr": 1.4279822257480262e-05} +{"train_loss": 0.013067832216620445, "grad_norm": 0.08104147017002106, "global_step": 39925, "epoch": 376, "lr": 1.4277748472170798e-05} +{"train_loss": 0.014601768925786018, "grad_norm": 0.0846647247672081, "global_step": 39926, "epoch": 376, "lr": 1.4275674812373774e-05} +{"train_loss": 0.01305654738098383, "grad_norm": 0.1178329735994339, "global_step": 39927, "epoch": 376, "lr": 1.4273601278096493e-05} +{"train_loss": 0.013705207034945488, "grad_norm": 0.06474192440509796, "global_step": 39928, "epoch": 376, "lr": 1.4271527869346253e-05} +{"train_loss": 0.014869247563183308, "grad_norm": 0.06654291599988937, "global_step": 39929, "epoch": 376, "lr": 1.4269454586130327e-05} +{"train_loss": 0.01297491043806076, "grad_norm": 0.10993801802396774, "global_step": 39930, "epoch": 376, "lr": 1.4267381428455983e-05} +{"train_loss": 0.016421237960457802, "grad_norm": 0.0980888307094574, "global_step": 39931, "epoch": 376, "lr": 1.4265308396330518e-05} +{"train_loss": 0.013163791969418526, "grad_norm": 0.13288547098636627, "global_step": 39932, "epoch": 376, "lr": 1.426323548976124e-05} +{"train_loss": 0.013703066855669022, "grad_norm": 0.08704768121242523, "global_step": 39933, "epoch": 376, "lr": 1.4261162708755394e-05} +{"train_loss": 0.013843407854437828, "grad_norm": 0.06764613837003708, "global_step": 39934, "epoch": 376, "lr": 1.4259090053320278e-05} +{"train_loss": 0.015883609652519226, "grad_norm": 0.06887352466583252, "global_step": 39935, "epoch": 376, "lr": 1.4257017523463189e-05} +{"train_loss": 0.017574697732925415, "grad_norm": 0.11834665387868881, "global_step": 39936, "epoch": 376, "lr": 1.4254945119191404e-05} +{"train_loss": 0.01311773993074894, "grad_norm": 0.11093280464410782, "global_step": 39937, "epoch": 376, "lr": 1.4252872840512176e-05} +{"train_loss": 0.012780580669641495, "grad_norm": 0.07772495597600937, "global_step": 39938, "epoch": 376, "lr": 1.4250800687432808e-05} +{"train_loss": 0.014779042452573776, "grad_norm": 0.10283026099205017, "global_step": 39939, "epoch": 376, "lr": 1.4248728659960592e-05} +{"train_loss": 0.01423343550413847, "grad_norm": 0.07027707248926163, "global_step": 39940, "epoch": 376, "lr": 1.4246656758102784e-05} +{"train_loss": 0.0140359066426754, "grad_norm": 0.06686358153820038, "global_step": 39941, "epoch": 376, "lr": 1.4244584981866693e-05} +{"train_loss": 0.0161085557192564, "grad_norm": 0.08213695138692856, "global_step": 39942, "epoch": 376, "lr": 1.4242513331259576e-05} +{"train_loss": 0.013391772285103798, "grad_norm": 0.10594845563173294, "global_step": 39943, "epoch": 376, "lr": 1.4240441806288701e-05} +{"train_loss": 0.015168828889727592, "grad_norm": 0.06125277653336525, "global_step": 39944, "epoch": 376, "lr": 1.4238370406961366e-05} +{"train_loss": 0.014496929943561554, "grad_norm": 0.06730707734823227, "global_step": 39945, "epoch": 376, "lr": 1.4236299133284858e-05} +{"train_loss": 0.014118376187980175, "grad_norm": 0.08764277398586273, "global_step": 39946, "epoch": 376, "lr": 1.4234227985266424e-05} +{"train_loss": 0.015665873885154724, "grad_norm": 0.06827749311923981, "global_step": 39947, "epoch": 376, "lr": 1.4232156962913379e-05} +{"train_loss": 0.011890470050275326, "grad_norm": 0.07604868710041046, "global_step": 39948, "epoch": 376, "lr": 1.4230086066232962e-05} +{"train_loss": 0.016201579943299294, "grad_norm": 0.07371076196432114, "global_step": 39949, "epoch": 376, "lr": 1.4228015295232483e-05} +{"train_loss": 0.01382384542375803, "grad_norm": 0.07072265446186066, "global_step": 39950, "epoch": 376, "lr": 1.4225944649919182e-05} +{"train_loss": 0.01446869969367981, "grad_norm": 0.08163328468799591, "global_step": 39951, "epoch": 376, "lr": 1.4223874130300357e-05} +{"train_loss": 0.016739916056394577, "grad_norm": 0.06077193096280098, "global_step": 39952, "epoch": 376, "lr": 1.4221803736383293e-05} +{"train_loss": 0.01567292958498001, "grad_norm": 0.09947456419467926, "global_step": 39953, "epoch": 376, "lr": 1.4219733468175256e-05} +{"train_loss": 0.018240153789520264, "grad_norm": 0.09825945645570755, "global_step": 39954, "epoch": 376, "lr": 1.4217663325683494e-05} +{"train_loss": 0.017665594816207886, "grad_norm": 0.10841210931539536, "global_step": 39955, "epoch": 376, "lr": 1.4215593308915304e-05} +{"train_loss": 0.016658809036016464, "grad_norm": 0.09119986742734909, "global_step": 39956, "epoch": 376, "lr": 1.4213523417877972e-05} +{"train_loss": 0.013036888092756271, "grad_norm": 0.09075301140546799, "global_step": 39957, "epoch": 376, "lr": 1.4211453652578744e-05} +{"train_loss": 0.01606055535376072, "grad_norm": 0.08166418969631195, "global_step": 39958, "epoch": 376, "lr": 1.4209384013024913e-05} +{"train_loss": 0.013226182200014591, "grad_norm": 0.07141340523958206, "global_step": 39959, "epoch": 376, "lr": 1.4207314499223745e-05} +{"train_loss": 0.014006886631250381, "grad_norm": 0.0524301677942276, "global_step": 39960, "epoch": 376, "lr": 1.420524511118249e-05} +{"train_loss": 0.015044482301850364, "grad_norm": 0.23706747591495514, "global_step": 39961, "epoch": 376, "lr": 1.420317584890844e-05, "val_loss": 0.01907876878976822} +{"train_loss": 0.014768240973353386, "grad_norm": 0.05352305993437767, "global_step": 39962, "epoch": 377, "lr": 1.420110671240888e-05} +{"train_loss": 0.014524406753480434, "grad_norm": 0.0936533585190773, "global_step": 39963, "epoch": 377, "lr": 1.4199037701691042e-05} +{"train_loss": 0.01415395550429821, "grad_norm": 0.14007653295993805, "global_step": 39964, "epoch": 377, "lr": 1.4196968816762235e-05} +{"train_loss": 0.016702469438314438, "grad_norm": 0.12114254385232925, "global_step": 39965, "epoch": 377, "lr": 1.419490005762969e-05} +{"train_loss": 0.015045180916786194, "grad_norm": 0.09466776996850967, "global_step": 39966, "epoch": 377, "lr": 1.4192831424300718e-05} +{"train_loss": 0.015561136417090893, "grad_norm": 0.06995981186628342, "global_step": 39967, "epoch": 377, "lr": 1.419076291678254e-05} +{"train_loss": 0.01628788374364376, "grad_norm": 0.1360948234796524, "global_step": 39968, "epoch": 377, "lr": 1.4188694535082453e-05} +{"train_loss": 0.014992416836321354, "grad_norm": 0.0633237287402153, "global_step": 39969, "epoch": 377, "lr": 1.4186626279207731e-05} +{"train_loss": 0.016115916892886162, "grad_norm": 0.11089976876974106, "global_step": 39970, "epoch": 377, "lr": 1.4184558149165632e-05} +{"train_loss": 0.015417220070958138, "grad_norm": 0.05996380373835564, "global_step": 39971, "epoch": 377, "lr": 1.4182490144963407e-05} +{"train_loss": 0.016169913113117218, "grad_norm": 0.062400925904512405, "global_step": 39972, "epoch": 377, "lr": 1.4180422266608334e-05} +{"train_loss": 0.015968166291713715, "grad_norm": 0.1151568666100502, "global_step": 39973, "epoch": 377, "lr": 1.4178354514107695e-05} +{"train_loss": 0.014118730090558529, "grad_norm": 0.08652045577764511, "global_step": 39974, "epoch": 377, "lr": 1.417628688746872e-05} +{"train_loss": 0.014947045594453812, "grad_norm": 0.1602591872215271, "global_step": 39975, "epoch": 377, "lr": 1.4174219386698695e-05} +{"train_loss": 0.01437996607273817, "grad_norm": 0.09128647297620773, "global_step": 39976, "epoch": 377, "lr": 1.4172152011804919e-05} +{"train_loss": 0.015070619061589241, "grad_norm": 0.1040482223033905, "global_step": 39977, "epoch": 377, "lr": 1.4170084762794583e-05} +{"train_loss": 0.014488080516457558, "grad_norm": 0.05264229327440262, "global_step": 39978, "epoch": 377, "lr": 1.4168017639674991e-05} +{"train_loss": 0.014390800148248672, "grad_norm": 0.05899696424603462, "global_step": 39979, "epoch": 377, "lr": 1.4165950642453412e-05} +{"train_loss": 0.01554824784398079, "grad_norm": 0.10318169742822647, "global_step": 39980, "epoch": 377, "lr": 1.4163883771137088e-05} +{"train_loss": 0.014185622334480286, "grad_norm": 0.10440915822982788, "global_step": 39981, "epoch": 377, "lr": 1.4161817025733293e-05} +{"train_loss": 0.013865501619875431, "grad_norm": 0.08494502305984497, "global_step": 39982, "epoch": 377, "lr": 1.41597504062493e-05} +{"train_loss": 0.01580623909831047, "grad_norm": 0.07223235070705414, "global_step": 39983, "epoch": 377, "lr": 1.415768391269236e-05} +{"train_loss": 0.013834831304848194, "grad_norm": 0.05564097687602043, "global_step": 39984, "epoch": 377, "lr": 1.415561754506971e-05} +{"train_loss": 0.0145290307700634, "grad_norm": 0.06278171390295029, "global_step": 39985, "epoch": 377, "lr": 1.4153551303388635e-05} +{"train_loss": 0.014757417142391205, "grad_norm": 0.12253794819116592, "global_step": 39986, "epoch": 377, "lr": 1.415148518765641e-05} +{"train_loss": 0.01623731665313244, "grad_norm": 0.09377647191286087, "global_step": 39987, "epoch": 377, "lr": 1.414941919788026e-05} +{"train_loss": 0.01396026462316513, "grad_norm": 0.07872366905212402, "global_step": 39988, "epoch": 377, "lr": 1.4147353334067471e-05} +{"train_loss": 0.014541931450366974, "grad_norm": 0.05540751293301582, "global_step": 39989, "epoch": 377, "lr": 1.4145287596225276e-05} +{"train_loss": 0.013905839994549751, "grad_norm": 0.11019321531057358, "global_step": 39990, "epoch": 377, "lr": 1.4143221984360965e-05} +{"train_loss": 0.01619613729417324, "grad_norm": 0.059614162892103195, "global_step": 39991, "epoch": 377, "lr": 1.4141156498481762e-05} +{"train_loss": 0.016773462295532227, "grad_norm": 0.11648386716842651, "global_step": 39992, "epoch": 377, "lr": 1.413909113859494e-05} +{"train_loss": 0.016974616795778275, "grad_norm": 0.07580453902482986, "global_step": 39993, "epoch": 377, "lr": 1.4137025904707773e-05} +{"train_loss": 0.014243602752685547, "grad_norm": 0.06578077375888824, "global_step": 39994, "epoch": 377, "lr": 1.4134960796827502e-05} +{"train_loss": 0.01246449165046215, "grad_norm": 0.05161231756210327, "global_step": 39995, "epoch": 377, "lr": 1.413289581496136e-05} +{"train_loss": 0.014160865917801857, "grad_norm": 0.06735409051179886, "global_step": 39996, "epoch": 377, "lr": 1.4130830959116632e-05} +{"train_loss": 0.015376175753772259, "grad_norm": 0.09321051836013794, "global_step": 39997, "epoch": 377, "lr": 1.4128766229300577e-05} +{"train_loss": 0.01498803123831749, "grad_norm": 0.04928828775882721, "global_step": 39998, "epoch": 377, "lr": 1.4126701625520422e-05} +{"train_loss": 0.014515850692987442, "grad_norm": 0.07937508076429367, "global_step": 39999, "epoch": 377, "lr": 1.4124637147783432e-05} +{"train_loss": 0.012140759266912937, "grad_norm": 0.06603840738534927, "global_step": 40000, "epoch": 377, "lr": 1.41225727960969e-05} +{"train_loss": 0.013473086059093475, "grad_norm": 0.07241912931203842, "global_step": 40001, "epoch": 377, "lr": 1.4120508570468016e-05} +{"train_loss": 0.015187207609415054, "grad_norm": 0.16630050539970398, "global_step": 40002, "epoch": 377, "lr": 1.4118444470904058e-05} +{"train_loss": 0.014448304660618305, "grad_norm": 0.07012658566236496, "global_step": 40003, "epoch": 377, "lr": 1.4116380497412302e-05} +{"train_loss": 0.015111075714230537, "grad_norm": 0.07634250819683075, "global_step": 40004, "epoch": 377, "lr": 1.4114316649999958e-05} +{"train_loss": 0.016089150682091713, "grad_norm": 0.10498134791851044, "global_step": 40005, "epoch": 377, "lr": 1.4112252928674319e-05} +{"train_loss": 0.01771448366343975, "grad_norm": 0.07413828372955322, "global_step": 40006, "epoch": 377, "lr": 1.4110189333442598e-05} +{"train_loss": 0.01548328623175621, "grad_norm": 0.07406756281852722, "global_step": 40007, "epoch": 377, "lr": 1.4108125864312077e-05} +{"train_loss": 0.015765411779284477, "grad_norm": 0.092054083943367, "global_step": 40008, "epoch": 377, "lr": 1.4106062521289981e-05} +{"train_loss": 0.015293597243726254, "grad_norm": 0.1809028685092926, "global_step": 40009, "epoch": 377, "lr": 1.410399930438357e-05} +{"train_loss": 0.017093874514102936, "grad_norm": 0.08378330618143082, "global_step": 40010, "epoch": 377, "lr": 1.4101936213600108e-05} +{"train_loss": 0.015112249180674553, "grad_norm": 0.09661652147769928, "global_step": 40011, "epoch": 377, "lr": 1.4099873248946832e-05} +{"train_loss": 0.01578383892774582, "grad_norm": 0.09653294086456299, "global_step": 40012, "epoch": 377, "lr": 1.4097810410430978e-05} +{"train_loss": 0.013610901311039925, "grad_norm": 0.10841669142246246, "global_step": 40013, "epoch": 377, "lr": 1.4095747698059798e-05} +{"train_loss": 0.015666993334889412, "grad_norm": 0.061313774436712265, "global_step": 40014, "epoch": 377, "lr": 1.4093685111840566e-05} +{"train_loss": 0.01670815609395504, "grad_norm": 0.0666639432311058, "global_step": 40015, "epoch": 377, "lr": 1.409162265178049e-05} +{"train_loss": 0.014605134725570679, "grad_norm": 0.14223681390285492, "global_step": 40016, "epoch": 377, "lr": 1.408956031788684e-05} +{"train_loss": 0.015919316560029984, "grad_norm": 0.05484556779265404, "global_step": 40017, "epoch": 377, "lr": 1.4087498110166885e-05} +{"train_loss": 0.01406384538859129, "grad_norm": 0.09553225338459015, "global_step": 40018, "epoch": 377, "lr": 1.4085436028627819e-05} +{"train_loss": 0.0172154288738966, "grad_norm": 0.09046611934900284, "global_step": 40019, "epoch": 377, "lr": 1.4083374073276905e-05} +{"train_loss": 0.015047797933220863, "grad_norm": 0.1061154305934906, "global_step": 40020, "epoch": 377, "lr": 1.4081312244121414e-05} +{"train_loss": 0.013899249024689198, "grad_norm": 0.0576639287173748, "global_step": 40021, "epoch": 377, "lr": 1.4079250541168555e-05} +{"train_loss": 0.014923044480383396, "grad_norm": 0.09025581181049347, "global_step": 40022, "epoch": 377, "lr": 1.4077188964425586e-05} +{"train_loss": 0.016222314909100533, "grad_norm": 0.11172192543745041, "global_step": 40023, "epoch": 377, "lr": 1.4075127513899772e-05} +{"train_loss": 0.016604140400886536, "grad_norm": 0.18767747282981873, "global_step": 40024, "epoch": 377, "lr": 1.4073066189598333e-05} +{"train_loss": 0.01533186063170433, "grad_norm": 0.06947097927331924, "global_step": 40025, "epoch": 377, "lr": 1.4071004991528503e-05} +{"train_loss": 0.015118114650249481, "grad_norm": 0.07565084844827652, "global_step": 40026, "epoch": 377, "lr": 1.4068943919697537e-05} +{"train_loss": 0.015954867005348206, "grad_norm": 0.05639877915382385, "global_step": 40027, "epoch": 377, "lr": 1.4066882974112689e-05} +{"train_loss": 0.016696834936738014, "grad_norm": 0.08724121749401093, "global_step": 40028, "epoch": 377, "lr": 1.4064822154781171e-05} +{"train_loss": 0.014952133409678936, "grad_norm": 0.08786901831626892, "global_step": 40029, "epoch": 377, "lr": 1.4062761461710255e-05} +{"train_loss": 0.016516821458935738, "grad_norm": 0.07877380400896072, "global_step": 40030, "epoch": 377, "lr": 1.4060700894907154e-05} +{"train_loss": 0.012849068269133568, "grad_norm": 0.09010937809944153, "global_step": 40031, "epoch": 377, "lr": 1.4058640454379134e-05} +{"train_loss": 0.01368330791592598, "grad_norm": 0.07729868590831757, "global_step": 40032, "epoch": 377, "lr": 1.4056580140133407e-05} +{"train_loss": 0.017179107293486595, "grad_norm": 0.08506517112255096, "global_step": 40033, "epoch": 377, "lr": 1.4054519952177225e-05} +{"train_loss": 0.015465197153389454, "grad_norm": 0.08077192306518555, "global_step": 40034, "epoch": 377, "lr": 1.4052459890517844e-05} +{"train_loss": 0.014708875678479671, "grad_norm": 0.12078185379505157, "global_step": 40035, "epoch": 377, "lr": 1.4050399955162486e-05} +{"train_loss": 0.012885376811027527, "grad_norm": 0.1438310295343399, "global_step": 40036, "epoch": 377, "lr": 1.4048340146118372e-05} +{"train_loss": 0.015492106787860394, "grad_norm": 0.07593698799610138, "global_step": 40037, "epoch": 377, "lr": 1.4046280463392752e-05} +{"train_loss": 0.016541549935936928, "grad_norm": 0.06895463913679123, "global_step": 40038, "epoch": 377, "lr": 1.4044220906992888e-05} +{"train_loss": 0.01487906463444233, "grad_norm": 0.07511399686336517, "global_step": 40039, "epoch": 377, "lr": 1.404216147692598e-05} +{"train_loss": 0.01349550113081932, "grad_norm": 0.0766955241560936, "global_step": 40040, "epoch": 377, "lr": 1.4040102173199276e-05} +{"train_loss": 0.017225848510861397, "grad_norm": 0.07699238508939743, "global_step": 40041, "epoch": 377, "lr": 1.4038042995820044e-05} +{"train_loss": 0.016852151602506638, "grad_norm": 0.1210835725069046, "global_step": 40042, "epoch": 377, "lr": 1.4035983944795462e-05} +{"train_loss": 0.015449170023202896, "grad_norm": 0.06543698906898499, "global_step": 40043, "epoch": 377, "lr": 1.4033925020132787e-05} +{"train_loss": 0.016342975199222565, "grad_norm": 0.11248217523097992, "global_step": 40044, "epoch": 377, "lr": 1.403186622183928e-05} +{"train_loss": 0.01318732462823391, "grad_norm": 0.08377628773450851, "global_step": 40045, "epoch": 377, "lr": 1.4029807549922131e-05} +{"train_loss": 0.01564035192131996, "grad_norm": 0.06978106498718262, "global_step": 40046, "epoch": 377, "lr": 1.4027749004388602e-05} +{"train_loss": 0.01384183019399643, "grad_norm": 0.10184917598962784, "global_step": 40047, "epoch": 377, "lr": 1.4025690585245932e-05} +{"train_loss": 0.014683997258543968, "grad_norm": 0.07949940115213394, "global_step": 40048, "epoch": 377, "lr": 1.4023632292501337e-05} +{"train_loss": 0.013213133439421654, "grad_norm": 0.06500416249036789, "global_step": 40049, "epoch": 377, "lr": 1.4021574126162041e-05} +{"train_loss": 0.015095459297299385, "grad_norm": 0.07007074356079102, "global_step": 40050, "epoch": 377, "lr": 1.4019516086235285e-05} +{"train_loss": 0.018803423270583153, "grad_norm": 0.10123560577630997, "global_step": 40051, "epoch": 377, "lr": 1.401745817272832e-05} +{"train_loss": 0.015164446085691452, "grad_norm": 0.08074958622455597, "global_step": 40052, "epoch": 377, "lr": 1.4015400385648342e-05} +{"train_loss": 0.013587061315774918, "grad_norm": 0.08562314510345459, "global_step": 40053, "epoch": 377, "lr": 1.401334272500261e-05} +{"train_loss": 0.014747299253940582, "grad_norm": 0.1035473644733429, "global_step": 40054, "epoch": 377, "lr": 1.4011285190798328e-05} +{"train_loss": 0.013136894442141056, "grad_norm": 0.14174851775169373, "global_step": 40055, "epoch": 377, "lr": 1.4009227783042756e-05} +{"train_loss": 0.012595455162227154, "grad_norm": 0.06089998036623001, "global_step": 40056, "epoch": 377, "lr": 1.4007170501743088e-05} +{"train_loss": 0.014906220138072968, "grad_norm": 0.06949325650930405, "global_step": 40057, "epoch": 377, "lr": 1.4005113346906573e-05} +{"train_loss": 0.01714431494474411, "grad_norm": 0.19312337040901184, "global_step": 40058, "epoch": 377, "lr": 1.4003056318540465e-05} +{"train_loss": 0.015018097124993801, "grad_norm": 0.08785584568977356, "global_step": 40059, "epoch": 377, "lr": 1.4000999416651928e-05} +{"train_loss": 0.016522347927093506, "grad_norm": 0.09981241822242737, "global_step": 40060, "epoch": 377, "lr": 1.3998942641248225e-05} +{"train_loss": 0.012112175114452839, "grad_norm": 0.09163323044776917, "global_step": 40061, "epoch": 377, "lr": 1.39968859923366e-05} +{"train_loss": 0.012978868559002876, "grad_norm": 0.06428160518407822, "global_step": 40062, "epoch": 377, "lr": 1.3994829469924243e-05} +{"train_loss": 0.014120038598775864, "grad_norm": 0.16475199162960052, "global_step": 40063, "epoch": 377, "lr": 1.3992773074018395e-05} +{"train_loss": 0.013454600237309933, "grad_norm": 0.09218492358922958, "global_step": 40064, "epoch": 377, "lr": 1.3990716804626302e-05} +{"train_loss": 0.015535339713096619, "grad_norm": 0.16913828253746033, "global_step": 40065, "epoch": 377, "lr": 1.398866066175517e-05} +{"train_loss": 0.014127514325082302, "grad_norm": 0.1022559180855751, "global_step": 40066, "epoch": 377, "lr": 1.3986604645412204e-05} +{"train_loss": 0.015039424482240991, "grad_norm": 0.08067353069782257, "global_step": 40067, "epoch": 377, "lr": 1.3984548755604654e-05, "val_loss": 0.01947242021560669} +{"train_loss": 0.015788868069648743, "grad_norm": 0.0777614563703537, "global_step": 40068, "epoch": 378, "lr": 1.3982492992339747e-05} +{"train_loss": 0.013345686718821526, "grad_norm": 0.060998156666755676, "global_step": 40069, "epoch": 378, "lr": 1.3980437355624681e-05} +{"train_loss": 0.01274943072348833, "grad_norm": 0.07309287786483765, "global_step": 40070, "epoch": 378, "lr": 1.3978381845466715e-05} +{"train_loss": 0.015480329282581806, "grad_norm": 0.14240841567516327, "global_step": 40071, "epoch": 378, "lr": 1.3976326461873029e-05} +{"train_loss": 0.014670217409729958, "grad_norm": 0.08516863733530045, "global_step": 40072, "epoch": 378, "lr": 1.3974271204850881e-05} +{"train_loss": 0.014125430956482887, "grad_norm": 0.07207860797643661, "global_step": 40073, "epoch": 378, "lr": 1.3972216074407462e-05} +{"train_loss": 0.014495719224214554, "grad_norm": 0.0699291005730629, "global_step": 40074, "epoch": 378, "lr": 1.3970161070550008e-05} +{"train_loss": 0.015839610248804092, "grad_norm": 0.11151156574487686, "global_step": 40075, "epoch": 378, "lr": 1.3968106193285757e-05} +{"train_loss": 0.016473473981022835, "grad_norm": 0.08925489336252213, "global_step": 40076, "epoch": 378, "lr": 1.3966051442621908e-05} +{"train_loss": 0.012979051098227501, "grad_norm": 0.057616718113422394, "global_step": 40077, "epoch": 378, "lr": 1.3963996818565667e-05} +{"train_loss": 0.015571321360766888, "grad_norm": 0.07096979022026062, "global_step": 40078, "epoch": 378, "lr": 1.396194232112429e-05} +{"train_loss": 0.015175961889326572, "grad_norm": 0.07585270702838898, "global_step": 40079, "epoch": 378, "lr": 1.395988795030495e-05} +{"train_loss": 0.01366388238966465, "grad_norm": 0.08537870645523071, "global_step": 40080, "epoch": 378, "lr": 1.3957833706114897e-05} +{"train_loss": 0.016056789085268974, "grad_norm": 0.0894286036491394, "global_step": 40081, "epoch": 378, "lr": 1.3955779588561357e-05} +{"train_loss": 0.015103235840797424, "grad_norm": 0.07171045988798141, "global_step": 40082, "epoch": 378, "lr": 1.3953725597651535e-05} +{"train_loss": 0.01623213291168213, "grad_norm": 0.08528459817171097, "global_step": 40083, "epoch": 378, "lr": 1.3951671733392623e-05} +{"train_loss": 0.014937173575162888, "grad_norm": 0.0833364874124527, "global_step": 40084, "epoch": 378, "lr": 1.3949617995791864e-05} +{"train_loss": 0.013494548387825489, "grad_norm": 0.07166587561368942, "global_step": 40085, "epoch": 378, "lr": 1.3947564384856487e-05} +{"train_loss": 0.014129096642136574, "grad_norm": 0.08078407496213913, "global_step": 40086, "epoch": 378, "lr": 1.3945510900593673e-05} +{"train_loss": 0.015452445484697819, "grad_norm": 0.11959061771631241, "global_step": 40087, "epoch": 378, "lr": 1.3943457543010653e-05} +{"train_loss": 0.012983039952814579, "grad_norm": 0.12198187410831451, "global_step": 40088, "epoch": 378, "lr": 1.3941404312114659e-05} +{"train_loss": 0.013845240697264671, "grad_norm": 0.07846294343471527, "global_step": 40089, "epoch": 378, "lr": 1.3939351207912887e-05} +{"train_loss": 0.014005736447870731, "grad_norm": 0.10023016482591629, "global_step": 40090, "epoch": 378, "lr": 1.3937298230412537e-05} +{"train_loss": 0.014354733750224113, "grad_norm": 0.1837042272090912, "global_step": 40091, "epoch": 378, "lr": 1.3935245379620837e-05} +{"train_loss": 0.017997948452830315, "grad_norm": 0.1122342124581337, "global_step": 40092, "epoch": 378, "lr": 1.3933192655545019e-05} +{"train_loss": 0.01633305847644806, "grad_norm": 0.07655947655439377, "global_step": 40093, "epoch": 378, "lr": 1.393114005819226e-05} +{"train_loss": 0.013970754109323025, "grad_norm": 0.09532414376735687, "global_step": 40094, "epoch": 378, "lr": 1.3929087587569801e-05} +{"train_loss": 0.013909758999943733, "grad_norm": 0.06073015555739403, "global_step": 40095, "epoch": 378, "lr": 1.3927035243684833e-05} +{"train_loss": 0.014518686570227146, "grad_norm": 0.058887921273708344, "global_step": 40096, "epoch": 378, "lr": 1.3924983026544586e-05} +{"train_loss": 0.01644628867506981, "grad_norm": 0.07869068533182144, "global_step": 40097, "epoch": 378, "lr": 1.3922930936156248e-05} +{"train_loss": 0.014837050810456276, "grad_norm": 0.10787217319011688, "global_step": 40098, "epoch": 378, "lr": 1.3920878972527035e-05} +{"train_loss": 0.012017016299068928, "grad_norm": 0.07103678584098816, "global_step": 40099, "epoch": 378, "lr": 1.3918827135664187e-05} +{"train_loss": 0.016667576506733894, "grad_norm": 0.11519856005907059, "global_step": 40100, "epoch": 378, "lr": 1.3916775425574879e-05} +{"train_loss": 0.013055775314569473, "grad_norm": 0.07614726573228836, "global_step": 40101, "epoch": 378, "lr": 1.3914723842266319e-05} +{"train_loss": 0.014681491069495678, "grad_norm": 0.057703565806150436, "global_step": 40102, "epoch": 378, "lr": 1.3912672385745745e-05} +{"train_loss": 0.015462314710021019, "grad_norm": 0.08450190722942352, "global_step": 40103, "epoch": 378, "lr": 1.391062105602033e-05} +{"train_loss": 0.012960058636963367, "grad_norm": 0.15281452238559723, "global_step": 40104, "epoch": 378, "lr": 1.3908569853097298e-05} +{"train_loss": 0.01504629384726286, "grad_norm": 0.08409319818019867, "global_step": 40105, "epoch": 378, "lr": 1.3906518776983869e-05} +{"train_loss": 0.01448097638785839, "grad_norm": 0.05929957330226898, "global_step": 40106, "epoch": 378, "lr": 1.3904467827687235e-05} +{"train_loss": 0.01416132040321827, "grad_norm": 0.06781897693872452, "global_step": 40107, "epoch": 378, "lr": 1.3902417005214591e-05} +{"train_loss": 0.016632430255413055, "grad_norm": 0.14383788406848907, "global_step": 40108, "epoch": 378, "lr": 1.3900366309573155e-05} +{"train_loss": 0.016701385378837585, "grad_norm": 0.08671942353248596, "global_step": 40109, "epoch": 378, "lr": 1.3898315740770146e-05} +{"train_loss": 0.0176826361566782, "grad_norm": 0.1335340440273285, "global_step": 40110, "epoch": 378, "lr": 1.3896265298812745e-05} +{"train_loss": 0.01615920104086399, "grad_norm": 0.1213129311800003, "global_step": 40111, "epoch": 378, "lr": 1.3894214983708176e-05} +{"train_loss": 0.013351822271943092, "grad_norm": 0.08196485042572021, "global_step": 40112, "epoch": 378, "lr": 1.3892164795463618e-05} +{"train_loss": 0.016489317640662193, "grad_norm": 0.1807417869567871, "global_step": 40113, "epoch": 378, "lr": 1.3890114734086307e-05} +{"train_loss": 0.015054933726787567, "grad_norm": 0.12346260249614716, "global_step": 40114, "epoch": 378, "lr": 1.3888064799583417e-05} +{"train_loss": 0.016317404806613922, "grad_norm": 0.06323274970054626, "global_step": 40115, "epoch": 378, "lr": 1.388601499196216e-05} +{"train_loss": 0.017500897869467735, "grad_norm": 0.06371142715215683, "global_step": 40116, "epoch": 378, "lr": 1.3883965311229758e-05} +{"train_loss": 0.018022920936346054, "grad_norm": 0.1122388169169426, "global_step": 40117, "epoch": 378, "lr": 1.3881915757393393e-05} +{"train_loss": 0.01628420315682888, "grad_norm": 0.1123424768447876, "global_step": 40118, "epoch": 378, "lr": 1.3879866330460251e-05} +{"train_loss": 0.01349005289375782, "grad_norm": 0.058064695447683334, "global_step": 40119, "epoch": 378, "lr": 1.3877817030437568e-05} +{"train_loss": 0.015801724046468735, "grad_norm": 0.07506132125854492, "global_step": 40120, "epoch": 378, "lr": 1.387576785733251e-05} +{"train_loss": 0.01445958111435175, "grad_norm": 0.07794018089771271, "global_step": 40121, "epoch": 378, "lr": 1.3873718811152298e-05} +{"train_loss": 0.01721581630408764, "grad_norm": 0.06977180391550064, "global_step": 40122, "epoch": 378, "lr": 1.3871669891904143e-05} +{"train_loss": 0.014024105854332447, "grad_norm": 0.06978294253349304, "global_step": 40123, "epoch": 378, "lr": 1.3869621099595226e-05} +{"train_loss": 0.014416772872209549, "grad_norm": 0.12144657969474792, "global_step": 40124, "epoch": 378, "lr": 1.3867572434232729e-05} +{"train_loss": 0.01525814551860094, "grad_norm": 0.06763526052236557, "global_step": 40125, "epoch": 378, "lr": 1.3865523895823874e-05} +{"train_loss": 0.01708560436964035, "grad_norm": 0.058232903480529785, "global_step": 40126, "epoch": 378, "lr": 1.3863475484375866e-05} +{"train_loss": 0.0149720199406147, "grad_norm": 0.08141638338565826, "global_step": 40127, "epoch": 378, "lr": 1.3861427199895877e-05} +{"train_loss": 0.013434743508696556, "grad_norm": 0.06054585799574852, "global_step": 40128, "epoch": 378, "lr": 1.3859379042391113e-05} +{"train_loss": 0.015359248965978622, "grad_norm": 0.08110268414020538, "global_step": 40129, "epoch": 378, "lr": 1.385733101186879e-05} +{"train_loss": 0.015163091942667961, "grad_norm": 0.06612241268157959, "global_step": 40130, "epoch": 378, "lr": 1.3855283108336093e-05} +{"train_loss": 0.01716427691280842, "grad_norm": 0.07318218797445297, "global_step": 40131, "epoch": 378, "lr": 1.385323533180019e-05} +{"train_loss": 0.013369240798056126, "grad_norm": 0.06001513823866844, "global_step": 40132, "epoch": 378, "lr": 1.3851187682268302e-05} +{"train_loss": 0.016301224008202553, "grad_norm": 0.12000931799411774, "global_step": 40133, "epoch": 378, "lr": 1.3849140159747636e-05} +{"train_loss": 0.016422193497419357, "grad_norm": 0.07487630099058151, "global_step": 40134, "epoch": 378, "lr": 1.3847092764245351e-05} +{"train_loss": 0.014459008350968361, "grad_norm": 0.06764691323041916, "global_step": 40135, "epoch": 378, "lr": 1.3845045495768677e-05} +{"train_loss": 0.013599798083305359, "grad_norm": 0.09881392121315002, "global_step": 40136, "epoch": 378, "lr": 1.3842998354324787e-05} +{"train_loss": 0.013681362383067608, "grad_norm": 0.07643213123083115, "global_step": 40137, "epoch": 378, "lr": 1.3840951339920865e-05} +{"train_loss": 0.01212097518146038, "grad_norm": 0.06018558144569397, "global_step": 40138, "epoch": 378, "lr": 1.3838904452564117e-05} +{"train_loss": 0.017025072127580643, "grad_norm": 0.06443190574645996, "global_step": 40139, "epoch": 378, "lr": 1.383685769226174e-05} +{"train_loss": 0.01583593338727951, "grad_norm": 0.07820025086402893, "global_step": 40140, "epoch": 378, "lr": 1.3834811059020909e-05} +{"train_loss": 0.013912301510572433, "grad_norm": 0.08579975366592407, "global_step": 40141, "epoch": 378, "lr": 1.3832764552848837e-05} +{"train_loss": 0.01719873584806919, "grad_norm": 0.07871804386377335, "global_step": 40142, "epoch": 378, "lr": 1.3830718173752689e-05} +{"train_loss": 0.014900700189173222, "grad_norm": 0.09674949198961258, "global_step": 40143, "epoch": 378, "lr": 1.3828671921739684e-05} +{"train_loss": 0.015104511752724648, "grad_norm": 0.07471536099910736, "global_step": 40144, "epoch": 378, "lr": 1.3826625796816972e-05} +{"train_loss": 0.01485408004373312, "grad_norm": 0.11025942862033844, "global_step": 40145, "epoch": 378, "lr": 1.3824579798991772e-05} +{"train_loss": 0.01564532145857811, "grad_norm": 0.0699460357427597, "global_step": 40146, "epoch": 378, "lr": 1.3822533928271281e-05} +{"train_loss": 0.014741320163011551, "grad_norm": 0.09194549918174744, "global_step": 40147, "epoch": 378, "lr": 1.3820488184662672e-05} +{"train_loss": 0.015310639515519142, "grad_norm": 0.11563704162836075, "global_step": 40148, "epoch": 378, "lr": 1.3818442568173117e-05} +{"train_loss": 0.013654699549078941, "grad_norm": 0.06047216057777405, "global_step": 40149, "epoch": 378, "lr": 1.3816397078809823e-05} +{"train_loss": 0.018561266362667084, "grad_norm": 0.14474907517433167, "global_step": 40150, "epoch": 378, "lr": 1.3814351716579987e-05} +{"train_loss": 0.015276756137609482, "grad_norm": 0.15483152866363525, "global_step": 40151, "epoch": 378, "lr": 1.3812306481490766e-05} +{"train_loss": 0.01627018116414547, "grad_norm": 0.12017124146223068, "global_step": 40152, "epoch": 378, "lr": 1.3810261373549365e-05} +{"train_loss": 0.013165956363081932, "grad_norm": 0.07845285534858704, "global_step": 40153, "epoch": 378, "lr": 1.3808216392762979e-05} +{"train_loss": 0.013873644173145294, "grad_norm": 0.09963074326515198, "global_step": 40154, "epoch": 378, "lr": 1.3806171539138784e-05} +{"train_loss": 0.014559379778802395, "grad_norm": 0.06368834525346756, "global_step": 40155, "epoch": 378, "lr": 1.3804126812683943e-05} +{"train_loss": 0.015079489909112453, "grad_norm": 0.07270728051662445, "global_step": 40156, "epoch": 378, "lr": 1.3802082213405665e-05} +{"train_loss": 0.015360715799033642, "grad_norm": 0.09087907522916794, "global_step": 40157, "epoch": 378, "lr": 1.380003774131114e-05} +{"train_loss": 0.012700900435447693, "grad_norm": 0.06915021687746048, "global_step": 40158, "epoch": 378, "lr": 1.3797993396407538e-05} +{"train_loss": 0.013592648319900036, "grad_norm": 0.07309847325086594, "global_step": 40159, "epoch": 378, "lr": 1.3795949178702033e-05} +{"train_loss": 0.0171517264097929, "grad_norm": 0.09273853152990341, "global_step": 40160, "epoch": 378, "lr": 1.3793905088201831e-05} +{"train_loss": 0.017848175019025803, "grad_norm": 0.061133384704589844, "global_step": 40161, "epoch": 378, "lr": 1.3791861124914079e-05} +{"train_loss": 0.014242593199014664, "grad_norm": 0.10988330096006393, "global_step": 40162, "epoch": 378, "lr": 1.3789817288845992e-05} +{"train_loss": 0.016614846885204315, "grad_norm": 0.06188668683171272, "global_step": 40163, "epoch": 378, "lr": 1.3787773580004748e-05} +{"train_loss": 0.0160989947617054, "grad_norm": 0.08784444630146027, "global_step": 40164, "epoch": 378, "lr": 1.3785729998397518e-05} +{"train_loss": 0.012935852631926537, "grad_norm": 0.08450772613286972, "global_step": 40165, "epoch": 378, "lr": 1.378368654403147e-05} +{"train_loss": 0.012174449861049652, "grad_norm": 0.0715855285525322, "global_step": 40166, "epoch": 378, "lr": 1.3781643216913797e-05} +{"train_loss": 0.015070177614688873, "grad_norm": 0.0683024451136589, "global_step": 40167, "epoch": 378, "lr": 1.3779600017051698e-05} +{"train_loss": 0.015364767983555794, "grad_norm": 0.07632040977478027, "global_step": 40168, "epoch": 378, "lr": 1.3777556944452314e-05} +{"train_loss": 0.015538206323981285, "grad_norm": 0.07754914462566376, "global_step": 40169, "epoch": 378, "lr": 1.377551399912284e-05} +{"train_loss": 0.016256079077720642, "grad_norm": 0.06937111914157867, "global_step": 40170, "epoch": 378, "lr": 1.3773471181070479e-05} +{"train_loss": 0.015164988115429878, "grad_norm": 0.06951716542243958, "global_step": 40171, "epoch": 378, "lr": 1.3771428490302379e-05} +{"train_loss": 0.01395662035793066, "grad_norm": 0.09814032912254333, "global_step": 40172, "epoch": 378, "lr": 1.3769385926825712e-05} +{"train_loss": 0.015035589849400633, "grad_norm": 0.07164707034826279, "global_step": 40173, "epoch": 378, "lr": 1.3767343490647665e-05, "val_loss": 0.019674552604556084} +{"train_loss": 0.013417903333902359, "grad_norm": 0.07472660392522812, "global_step": 40174, "epoch": 379, "lr": 1.376530118177543e-05} +{"train_loss": 0.01606357842683792, "grad_norm": 0.06283917278051376, "global_step": 40175, "epoch": 379, "lr": 1.3763259000216155e-05} +{"train_loss": 0.015834029763936996, "grad_norm": 0.07206520438194275, "global_step": 40176, "epoch": 379, "lr": 1.3761216945977046e-05} +{"train_loss": 0.015832649543881416, "grad_norm": 0.08408485352993011, "global_step": 40177, "epoch": 379, "lr": 1.3759175019065257e-05} +{"train_loss": 0.014336933381855488, "grad_norm": 0.1074712872505188, "global_step": 40178, "epoch": 379, "lr": 1.375713321948795e-05} +{"train_loss": 0.013603891246020794, "grad_norm": 0.08445025980472565, "global_step": 40179, "epoch": 379, "lr": 1.3755091547252319e-05} +{"train_loss": 0.01574224978685379, "grad_norm": 0.07631266862154007, "global_step": 40180, "epoch": 379, "lr": 1.3753050002365547e-05} +{"train_loss": 0.015489589422941208, "grad_norm": 0.06652814894914627, "global_step": 40181, "epoch": 379, "lr": 1.3751008584834779e-05} +{"train_loss": 0.01536055188626051, "grad_norm": 0.08775027841329575, "global_step": 40182, "epoch": 379, "lr": 1.374896729466722e-05} +{"train_loss": 0.01443627942353487, "grad_norm": 0.09122948348522186, "global_step": 40183, "epoch": 379, "lr": 1.3746926131870008e-05} +{"train_loss": 0.014154437929391861, "grad_norm": 0.05776461213827133, "global_step": 40184, "epoch": 379, "lr": 1.374488509645035e-05} +{"train_loss": 0.014779428020119667, "grad_norm": 0.07780930399894714, "global_step": 40185, "epoch": 379, "lr": 1.3742844188415376e-05} +{"train_loss": 0.013614715076982975, "grad_norm": 0.10468362271785736, "global_step": 40186, "epoch": 379, "lr": 1.3740803407772285e-05} +{"train_loss": 0.019458075985312462, "grad_norm": 0.09655290842056274, "global_step": 40187, "epoch": 379, "lr": 1.3738762754528256e-05} +{"train_loss": 0.013532459735870361, "grad_norm": 0.10326369106769562, "global_step": 40188, "epoch": 379, "lr": 1.3736722228690446e-05} +{"train_loss": 0.018033526837825775, "grad_norm": 0.13098116219043732, "global_step": 40189, "epoch": 379, "lr": 1.373468183026601e-05} +{"train_loss": 0.018156979233026505, "grad_norm": 0.08855000138282776, "global_step": 40190, "epoch": 379, "lr": 1.3732641559262126e-05} +{"train_loss": 0.013103384524583817, "grad_norm": 0.11209339648485184, "global_step": 40191, "epoch": 379, "lr": 1.3730601415685989e-05} +{"train_loss": 0.01467850711196661, "grad_norm": 0.056030597537755966, "global_step": 40192, "epoch": 379, "lr": 1.3728561399544726e-05} +{"train_loss": 0.012351716868579388, "grad_norm": 0.0862560048699379, "global_step": 40193, "epoch": 379, "lr": 1.3726521510845525e-05} +{"train_loss": 0.015910224989056587, "grad_norm": 0.09541676938533783, "global_step": 40194, "epoch": 379, "lr": 1.3724481749595564e-05} +{"train_loss": 0.015108155086636543, "grad_norm": 0.062184520065784454, "global_step": 40195, "epoch": 379, "lr": 1.3722442115802004e-05} +{"train_loss": 0.014777825213968754, "grad_norm": 0.07163335382938385, "global_step": 40196, "epoch": 379, "lr": 1.3720402609471988e-05} +{"train_loss": 0.016335511580109596, "grad_norm": 0.0906764566898346, "global_step": 40197, "epoch": 379, "lr": 1.3718363230612697e-05} +{"train_loss": 0.01689322106540203, "grad_norm": 0.07396107167005539, "global_step": 40198, "epoch": 379, "lr": 1.3716323979231316e-05} +{"train_loss": 0.013821296393871307, "grad_norm": 0.050619956105947495, "global_step": 40199, "epoch": 379, "lr": 1.371428485533498e-05} +{"train_loss": 0.01427027303725481, "grad_norm": 0.09058279544115067, "global_step": 40200, "epoch": 379, "lr": 1.3712245858930884e-05} +{"train_loss": 0.014030888676643372, "grad_norm": 0.06697189807891846, "global_step": 40201, "epoch": 379, "lr": 1.371020699002617e-05} +{"train_loss": 0.013846013695001602, "grad_norm": 0.09762182831764221, "global_step": 40202, "epoch": 379, "lr": 1.3708168248627995e-05} +{"train_loss": 0.013906443491578102, "grad_norm": 0.11960899829864502, "global_step": 40203, "epoch": 379, "lr": 1.3706129634743531e-05} +{"train_loss": 0.014978091232478619, "grad_norm": 0.14666137099266052, "global_step": 40204, "epoch": 379, "lr": 1.370409114837996e-05} +{"train_loss": 0.014085829257965088, "grad_norm": 0.10290475189685822, "global_step": 40205, "epoch": 379, "lr": 1.3702052789544429e-05} +{"train_loss": 0.015471979975700378, "grad_norm": 0.12345931679010391, "global_step": 40206, "epoch": 379, "lr": 1.3700014558244085e-05} +{"train_loss": 0.014629974961280823, "grad_norm": 0.15231366455554962, "global_step": 40207, "epoch": 379, "lr": 1.36979764544861e-05} +{"train_loss": 0.01396014355123043, "grad_norm": 0.07002727687358856, "global_step": 40208, "epoch": 379, "lr": 1.3695938478277653e-05} +{"train_loss": 0.016265589743852615, "grad_norm": 0.057451363652944565, "global_step": 40209, "epoch": 379, "lr": 1.3693900629625878e-05} +{"train_loss": 0.012331517413258553, "grad_norm": 0.054250020533800125, "global_step": 40210, "epoch": 379, "lr": 1.3691862908537945e-05} +{"train_loss": 0.014155101031064987, "grad_norm": 0.08752305805683136, "global_step": 40211, "epoch": 379, "lr": 1.3689825315021032e-05} +{"train_loss": 0.01777030900120735, "grad_norm": 0.07016843557357788, "global_step": 40212, "epoch": 379, "lr": 1.3687787849082285e-05} +{"train_loss": 0.014353396371006966, "grad_norm": 0.07024148106575012, "global_step": 40213, "epoch": 379, "lr": 1.368575051072884e-05} +{"train_loss": 0.013728879392147064, "grad_norm": 0.06004855036735535, "global_step": 40214, "epoch": 379, "lr": 1.3683713299967877e-05} +{"train_loss": 0.015689324587583542, "grad_norm": 0.11509436368942261, "global_step": 40215, "epoch": 379, "lr": 1.3681676216806566e-05} +{"train_loss": 0.015119807794690132, "grad_norm": 0.059201277792453766, "global_step": 40216, "epoch": 379, "lr": 1.367963926125203e-05} +{"train_loss": 0.01540143508464098, "grad_norm": 0.07093854993581772, "global_step": 40217, "epoch": 379, "lr": 1.3677602433311471e-05} +{"train_loss": 0.014809587970376015, "grad_norm": 0.05490642040967941, "global_step": 40218, "epoch": 379, "lr": 1.3675565732992018e-05} +{"train_loss": 0.013503290712833405, "grad_norm": 0.09960480034351349, "global_step": 40219, "epoch": 379, "lr": 1.3673529160300814e-05} +{"train_loss": 0.013272466138005257, "grad_norm": 0.06991741806268692, "global_step": 40220, "epoch": 379, "lr": 1.3671492715245027e-05} +{"train_loss": 0.014395724050700665, "grad_norm": 0.11409962177276611, "global_step": 40221, "epoch": 379, "lr": 1.3669456397831837e-05} +{"train_loss": 0.01878698356449604, "grad_norm": 0.06251880526542664, "global_step": 40222, "epoch": 379, "lr": 1.366742020806836e-05} +{"train_loss": 0.014354160986840725, "grad_norm": 0.14073802530765533, "global_step": 40223, "epoch": 379, "lr": 1.3665384145961779e-05} +{"train_loss": 0.01606922782957554, "grad_norm": 0.13625729084014893, "global_step": 40224, "epoch": 379, "lr": 1.3663348211519228e-05} +{"train_loss": 0.017238765954971313, "grad_norm": 0.06105554476380348, "global_step": 40225, "epoch": 379, "lr": 1.3661312404747877e-05} +{"train_loss": 0.01649964042007923, "grad_norm": 0.06355186551809311, "global_step": 40226, "epoch": 379, "lr": 1.3659276725654863e-05} +{"train_loss": 0.01776835322380066, "grad_norm": 0.10319673269987106, "global_step": 40227, "epoch": 379, "lr": 1.3657241174247343e-05} +{"train_loss": 0.014794405549764633, "grad_norm": 0.1218656525015831, "global_step": 40228, "epoch": 379, "lr": 1.3655205750532486e-05} +{"train_loss": 0.014751550741493702, "grad_norm": 0.06335875391960144, "global_step": 40229, "epoch": 379, "lr": 1.3653170454517434e-05} +{"train_loss": 0.016398845240473747, "grad_norm": 0.0669812560081482, "global_step": 40230, "epoch": 379, "lr": 1.3651135286209315e-05} +{"train_loss": 0.013953001238405704, "grad_norm": 0.057547081261873245, "global_step": 40231, "epoch": 379, "lr": 1.3649100245615304e-05} +{"train_loss": 0.012336975894868374, "grad_norm": 0.08958113938570023, "global_step": 40232, "epoch": 379, "lr": 1.3647065332742554e-05} +{"train_loss": 0.01806768588721752, "grad_norm": 0.13530853390693665, "global_step": 40233, "epoch": 379, "lr": 1.3645030547598197e-05} +{"train_loss": 0.01311856135725975, "grad_norm": 0.14997632801532745, "global_step": 40234, "epoch": 379, "lr": 1.3642995890189392e-05} +{"train_loss": 0.01378670334815979, "grad_norm": 0.09495900571346283, "global_step": 40235, "epoch": 379, "lr": 1.3640961360523318e-05} +{"train_loss": 0.013118147850036621, "grad_norm": 0.06998075544834137, "global_step": 40236, "epoch": 379, "lr": 1.3638926958607061e-05} +{"train_loss": 0.012310652062296867, "grad_norm": 0.07145832479000092, "global_step": 40237, "epoch": 379, "lr": 1.3636892684447805e-05} +{"train_loss": 0.016488544642925262, "grad_norm": 0.10392376780509949, "global_step": 40238, "epoch": 379, "lr": 1.3634858538052708e-05} +{"train_loss": 0.015754826366901398, "grad_norm": 0.09949760884046555, "global_step": 40239, "epoch": 379, "lr": 1.3632824519428889e-05} +{"train_loss": 0.01723604090511799, "grad_norm": 0.09018256515264511, "global_step": 40240, "epoch": 379, "lr": 1.3630790628583507e-05} +{"train_loss": 0.015569336712360382, "grad_norm": 0.09759350121021271, "global_step": 40241, "epoch": 379, "lr": 1.362875686552373e-05} +{"train_loss": 0.016561517491936684, "grad_norm": 0.11014946550130844, "global_step": 40242, "epoch": 379, "lr": 1.3626723230256677e-05} +{"train_loss": 0.015641292557120323, "grad_norm": 0.09046295285224915, "global_step": 40243, "epoch": 379, "lr": 1.362468972278949e-05} +{"train_loss": 0.015932157635688782, "grad_norm": 0.14043524861335754, "global_step": 40244, "epoch": 379, "lr": 1.362265634312932e-05} +{"train_loss": 0.01363912783563137, "grad_norm": 0.1388881355524063, "global_step": 40245, "epoch": 379, "lr": 1.3620623091283336e-05} +{"train_loss": 0.014484966173768044, "grad_norm": 0.08542715013027191, "global_step": 40246, "epoch": 379, "lr": 1.3618589967258644e-05} +{"train_loss": 0.014934703707695007, "grad_norm": 0.053983595222234726, "global_step": 40247, "epoch": 379, "lr": 1.3616556971062422e-05} +{"train_loss": 0.016428157687187195, "grad_norm": 0.07899854332208633, "global_step": 40248, "epoch": 379, "lr": 1.3614524102701776e-05} +{"train_loss": 0.012029744684696198, "grad_norm": 0.09591838717460632, "global_step": 40249, "epoch": 379, "lr": 1.3612491362183888e-05} +{"train_loss": 0.016201794147491455, "grad_norm": 0.08592063933610916, "global_step": 40250, "epoch": 379, "lr": 1.3610458749515865e-05} +{"train_loss": 0.014171253889799118, "grad_norm": 0.07752692699432373, "global_step": 40251, "epoch": 379, "lr": 1.360842626470486e-05} +{"train_loss": 0.01605204865336418, "grad_norm": 0.11041251569986343, "global_step": 40252, "epoch": 379, "lr": 1.3606393907758042e-05} +{"train_loss": 0.016458887606859207, "grad_norm": 0.06338205933570862, "global_step": 40253, "epoch": 379, "lr": 1.360436167868252e-05} +{"train_loss": 0.01491512544453144, "grad_norm": 0.08337287604808807, "global_step": 40254, "epoch": 379, "lr": 1.3602329577485435e-05} +{"train_loss": 0.014563926495611668, "grad_norm": 0.0619489923119545, "global_step": 40255, "epoch": 379, "lr": 1.3600297604173933e-05} +{"train_loss": 0.013188217766582966, "grad_norm": 0.08853064477443695, "global_step": 40256, "epoch": 379, "lr": 1.359826575875517e-05} +{"train_loss": 0.015629567205905914, "grad_norm": 0.10748834908008575, "global_step": 40257, "epoch": 379, "lr": 1.3596234041236256e-05} +{"train_loss": 0.013582097366452217, "grad_norm": 0.0801505371928215, "global_step": 40258, "epoch": 379, "lr": 1.3594202451624356e-05} +{"train_loss": 0.0166500024497509, "grad_norm": 0.0701846107840538, "global_step": 40259, "epoch": 379, "lr": 1.35921709899266e-05} +{"train_loss": 0.016794780269265175, "grad_norm": 0.12027739733457565, "global_step": 40260, "epoch": 379, "lr": 1.3590139656150102e-05} +{"train_loss": 0.012742094695568085, "grad_norm": 0.12327190488576889, "global_step": 40261, "epoch": 379, "lr": 1.3588108450302023e-05} +{"train_loss": 0.014350878074765205, "grad_norm": 0.07180909067392349, "global_step": 40262, "epoch": 379, "lr": 1.3586077372389511e-05} +{"train_loss": 0.01605147123336792, "grad_norm": 0.08058767765760422, "global_step": 40263, "epoch": 379, "lr": 1.3584046422419673e-05} +{"train_loss": 0.015968941152095795, "grad_norm": 0.12108924239873886, "global_step": 40264, "epoch": 379, "lr": 1.3582015600399673e-05} +{"train_loss": 0.01438194327056408, "grad_norm": 0.06507830321788788, "global_step": 40265, "epoch": 379, "lr": 1.3579984906336612e-05} +{"train_loss": 0.014305228367447853, "grad_norm": 0.11930685490369797, "global_step": 40266, "epoch": 379, "lr": 1.3577954340237663e-05} +{"train_loss": 0.015345137566328049, "grad_norm": 0.10864244401454926, "global_step": 40267, "epoch": 379, "lr": 1.3575923902109933e-05} +{"train_loss": 0.01433072704821825, "grad_norm": 0.08156874030828476, "global_step": 40268, "epoch": 379, "lr": 1.3573893591960562e-05} +{"train_loss": 0.017110763117671013, "grad_norm": 0.14300739765167236, "global_step": 40269, "epoch": 379, "lr": 1.3571863409796703e-05} +{"train_loss": 0.016366221010684967, "grad_norm": 0.08513869345188141, "global_step": 40270, "epoch": 379, "lr": 1.3569833355625477e-05} +{"train_loss": 0.015802746638655663, "grad_norm": 0.06447505205869675, "global_step": 40271, "epoch": 379, "lr": 1.3567803429453996e-05} +{"train_loss": 0.01772160828113556, "grad_norm": 0.07631343603134155, "global_step": 40272, "epoch": 379, "lr": 1.3565773631289408e-05} +{"train_loss": 0.011966438964009285, "grad_norm": 0.07222669571638107, "global_step": 40273, "epoch": 379, "lr": 1.3563743961138864e-05} +{"train_loss": 0.01556218322366476, "grad_norm": 0.0632573813199997, "global_step": 40274, "epoch": 379, "lr": 1.3561714419009463e-05} +{"train_loss": 0.01350344531238079, "grad_norm": 0.09467428177595139, "global_step": 40275, "epoch": 379, "lr": 1.3559685004908346e-05} +{"train_loss": 0.015582157298922539, "grad_norm": 0.08122652024030685, "global_step": 40276, "epoch": 379, "lr": 1.3557655718842682e-05} +{"train_loss": 0.01340150460600853, "grad_norm": 0.08475861698389053, "global_step": 40277, "epoch": 379, "lr": 1.355562656081954e-05} +{"train_loss": 0.013434800319373608, "grad_norm": 0.09099788218736649, "global_step": 40278, "epoch": 379, "lr": 1.3553597530846074e-05} +{"train_loss": 0.015040170146538964, "grad_norm": 0.1591908186674118, "global_step": 40279, "epoch": 379, "lr": 1.3551568628929434e-05, "val_loss": 0.02007633075118065} +{"train_loss": 0.01457118894904852, "grad_norm": 0.07369331270456314, "global_step": 40280, "epoch": 380, "lr": 1.3549539855076716e-05} +{"train_loss": 0.01569483056664467, "grad_norm": 0.07880397140979767, "global_step": 40281, "epoch": 380, "lr": 1.354751120929506e-05} +{"train_loss": 0.013536574319005013, "grad_norm": 0.06410276144742966, "global_step": 40282, "epoch": 380, "lr": 1.354548269159161e-05} +{"train_loss": 0.015193217433989048, "grad_norm": 0.06866239011287689, "global_step": 40283, "epoch": 380, "lr": 1.3543454301973486e-05} +{"train_loss": 0.015339555218815804, "grad_norm": 0.08260437101125717, "global_step": 40284, "epoch": 380, "lr": 1.3541426040447792e-05} +{"train_loss": 0.01400203537195921, "grad_norm": 0.06914888322353363, "global_step": 40285, "epoch": 380, "lr": 1.3539397907021667e-05} +{"train_loss": 0.015201875008642673, "grad_norm": 0.07848463207483292, "global_step": 40286, "epoch": 380, "lr": 1.3537369901702263e-05} +{"train_loss": 0.014771947637200356, "grad_norm": 0.12478934228420258, "global_step": 40287, "epoch": 380, "lr": 1.3535342024496666e-05} +{"train_loss": 0.01719674840569496, "grad_norm": 0.15470701456069946, "global_step": 40288, "epoch": 380, "lr": 1.3533314275412034e-05} +{"train_loss": 0.015953004360198975, "grad_norm": 0.10934668779373169, "global_step": 40289, "epoch": 380, "lr": 1.353128665445546e-05} +{"train_loss": 0.017300084233283997, "grad_norm": 0.09575206786394119, "global_step": 40290, "epoch": 380, "lr": 1.35292591616341e-05} +{"train_loss": 0.015687329694628716, "grad_norm": 0.06978598237037659, "global_step": 40291, "epoch": 380, "lr": 1.3527231796955053e-05} +{"train_loss": 0.015932366251945496, "grad_norm": 0.08683229237794876, "global_step": 40292, "epoch": 380, "lr": 1.3525204560425448e-05} +{"train_loss": 0.0146905817091465, "grad_norm": 0.06051769107580185, "global_step": 40293, "epoch": 380, "lr": 1.3523177452052427e-05} +{"train_loss": 0.016911283135414124, "grad_norm": 0.07313819229602814, "global_step": 40294, "epoch": 380, "lr": 1.3521150471843097e-05} +{"train_loss": 0.015699001029133797, "grad_norm": 0.08357318490743637, "global_step": 40295, "epoch": 380, "lr": 1.3519123619804564e-05} +{"train_loss": 0.017630435526371002, "grad_norm": 0.08856943994760513, "global_step": 40296, "epoch": 380, "lr": 1.3517096895943981e-05} +{"train_loss": 0.014631439000368118, "grad_norm": 0.07002310454845428, "global_step": 40297, "epoch": 380, "lr": 1.351507030026844e-05} +{"train_loss": 0.017028000205755234, "grad_norm": 0.09318052977323532, "global_step": 40298, "epoch": 380, "lr": 1.3513043832785072e-05} +{"train_loss": 0.015932118520140648, "grad_norm": 0.12231215089559555, "global_step": 40299, "epoch": 380, "lr": 1.3511017493501005e-05} +{"train_loss": 0.016736585646867752, "grad_norm": 0.09459790587425232, "global_step": 40300, "epoch": 380, "lr": 1.3508991282423378e-05} +{"train_loss": 0.016180936247110367, "grad_norm": 0.09648260474205017, "global_step": 40301, "epoch": 380, "lr": 1.3506965199559262e-05} +{"train_loss": 0.015845797955989838, "grad_norm": 0.08498935401439667, "global_step": 40302, "epoch": 380, "lr": 1.3504939244915792e-05} +{"train_loss": 0.013921590521931648, "grad_norm": 0.06266918778419495, "global_step": 40303, "epoch": 380, "lr": 1.3502913418500119e-05} +{"train_loss": 0.013563038781285286, "grad_norm": 0.08969300985336304, "global_step": 40304, "epoch": 380, "lr": 1.3500887720319316e-05} +{"train_loss": 0.015064053237438202, "grad_norm": 0.08859024196863174, "global_step": 40305, "epoch": 380, "lr": 1.349886215038052e-05} +{"train_loss": 0.013481216505169868, "grad_norm": 0.07800660282373428, "global_step": 40306, "epoch": 380, "lr": 1.3496836708690864e-05} +{"train_loss": 0.0139808664098382, "grad_norm": 0.0662856474518776, "global_step": 40307, "epoch": 380, "lr": 1.349481139525745e-05} +{"train_loss": 0.01498381607234478, "grad_norm": 0.1052684634923935, "global_step": 40308, "epoch": 380, "lr": 1.3492786210087377e-05} +{"train_loss": 0.014855630695819855, "grad_norm": 0.06254440546035767, "global_step": 40309, "epoch": 380, "lr": 1.3490761153187775e-05} +{"train_loss": 0.014166193082928658, "grad_norm": 0.06985433399677277, "global_step": 40310, "epoch": 380, "lr": 1.3488736224565778e-05} +{"train_loss": 0.01455841213464737, "grad_norm": 0.1538475900888443, "global_step": 40311, "epoch": 380, "lr": 1.3486711424228487e-05} +{"train_loss": 0.016547560691833496, "grad_norm": 0.0735587626695633, "global_step": 40312, "epoch": 380, "lr": 1.3484686752182996e-05} +{"train_loss": 0.015128192491829395, "grad_norm": 0.10569636523723602, "global_step": 40313, "epoch": 380, "lr": 1.3482662208436431e-05} +{"train_loss": 0.014508066698908806, "grad_norm": 0.09686834365129471, "global_step": 40314, "epoch": 380, "lr": 1.3480637792995932e-05} +{"train_loss": 0.014223795384168625, "grad_norm": 0.07445866614580154, "global_step": 40315, "epoch": 380, "lr": 1.347861350586857e-05} +{"train_loss": 0.012066027149558067, "grad_norm": 0.05435827001929283, "global_step": 40316, "epoch": 380, "lr": 1.3476589347061475e-05} +{"train_loss": 0.015948805958032608, "grad_norm": 0.10071643441915512, "global_step": 40317, "epoch": 380, "lr": 1.3474565316581794e-05} +{"train_loss": 0.013783350586891174, "grad_norm": 0.05730434134602547, "global_step": 40318, "epoch": 380, "lr": 1.347254141443658e-05} +{"train_loss": 0.01907302439212799, "grad_norm": 0.08895089477300644, "global_step": 40319, "epoch": 380, "lr": 1.3470517640632969e-05} +{"train_loss": 0.012851063162088394, "grad_norm": 0.06270219385623932, "global_step": 40320, "epoch": 380, "lr": 1.3468493995178084e-05} +{"train_loss": 0.014935202896595001, "grad_norm": 0.060035400092601776, "global_step": 40321, "epoch": 380, "lr": 1.3466470478079007e-05} +{"train_loss": 0.012761138379573822, "grad_norm": 0.06735836714506149, "global_step": 40322, "epoch": 380, "lr": 1.3464447089342875e-05} +{"train_loss": 0.014269033446907997, "grad_norm": 0.08352483808994293, "global_step": 40323, "epoch": 380, "lr": 1.3462423828976795e-05} +{"train_loss": 0.01446782611310482, "grad_norm": 0.08342557400465012, "global_step": 40324, "epoch": 380, "lr": 1.3460400696987868e-05} +{"train_loss": 0.014170128852128983, "grad_norm": 0.057371217757463455, "global_step": 40325, "epoch": 380, "lr": 1.3458377693383183e-05} +{"train_loss": 0.012180645018815994, "grad_norm": 0.1110369861125946, "global_step": 40326, "epoch": 380, "lr": 1.3456354818169876e-05} +{"train_loss": 0.0137242516502738, "grad_norm": 0.07645301520824432, "global_step": 40327, "epoch": 380, "lr": 1.3454332071355052e-05} +{"train_loss": 0.013990424573421478, "grad_norm": 0.07947491854429245, "global_step": 40328, "epoch": 380, "lr": 1.3452309452945805e-05} +{"train_loss": 0.01342063955962658, "grad_norm": 0.07663270831108093, "global_step": 40329, "epoch": 380, "lr": 1.345028696294926e-05} +{"train_loss": 0.015506455674767494, "grad_norm": 0.07120291888713837, "global_step": 40330, "epoch": 380, "lr": 1.3448264601372496e-05} +{"train_loss": 0.014996528625488281, "grad_norm": 0.0850602388381958, "global_step": 40331, "epoch": 380, "lr": 1.344624236822265e-05} +{"train_loss": 0.01685664989054203, "grad_norm": 0.07657726854085922, "global_step": 40332, "epoch": 380, "lr": 1.3444220263506795e-05} +{"train_loss": 0.015942154452204704, "grad_norm": 0.08882348239421844, "global_step": 40333, "epoch": 380, "lr": 1.3442198287232055e-05} +{"train_loss": 0.014133008196949959, "grad_norm": 0.0745684802532196, "global_step": 40334, "epoch": 380, "lr": 1.3440176439405544e-05} +{"train_loss": 0.016706056892871857, "grad_norm": 0.09729534387588501, "global_step": 40335, "epoch": 380, "lr": 1.3438154720034357e-05} +{"train_loss": 0.015748366713523865, "grad_norm": 0.05382660776376724, "global_step": 40336, "epoch": 380, "lr": 1.3436133129125577e-05} +{"train_loss": 0.012620007619261742, "grad_norm": 0.09380689263343811, "global_step": 40337, "epoch": 380, "lr": 1.3434111666686339e-05} +{"train_loss": 0.013987056910991669, "grad_norm": 0.1062345877289772, "global_step": 40338, "epoch": 380, "lr": 1.3432090332723712e-05} +{"train_loss": 0.016324758529663086, "grad_norm": 0.13415804505348206, "global_step": 40339, "epoch": 380, "lr": 1.343006912724482e-05} +{"train_loss": 0.015112170949578285, "grad_norm": 0.10360239446163177, "global_step": 40340, "epoch": 380, "lr": 1.3428048050256776e-05} +{"train_loss": 0.0162124615162611, "grad_norm": 0.1520911604166031, "global_step": 40341, "epoch": 380, "lr": 1.3426027101766669e-05} +{"train_loss": 0.017992272973060608, "grad_norm": 0.15741997957229614, "global_step": 40342, "epoch": 380, "lr": 1.3424006281781577e-05} +{"train_loss": 0.014506885781884193, "grad_norm": 0.09842235594987869, "global_step": 40343, "epoch": 380, "lr": 1.3421985590308622e-05} +{"train_loss": 0.014783894643187523, "grad_norm": 0.07224396616220474, "global_step": 40344, "epoch": 380, "lr": 1.341996502735492e-05} +{"train_loss": 0.016282256692647934, "grad_norm": 0.10883325338363647, "global_step": 40345, "epoch": 380, "lr": 1.341794459292754e-05} +{"train_loss": 0.013971257023513317, "grad_norm": 0.06897499412298203, "global_step": 40346, "epoch": 380, "lr": 1.3415924287033588e-05} +{"train_loss": 0.015342258848249912, "grad_norm": 0.0795028954744339, "global_step": 40347, "epoch": 380, "lr": 1.341390410968018e-05} +{"train_loss": 0.014492174610495567, "grad_norm": 0.08835570514202118, "global_step": 40348, "epoch": 380, "lr": 1.3411884060874409e-05} +{"train_loss": 0.01305096410214901, "grad_norm": 0.06501943618059158, "global_step": 40349, "epoch": 380, "lr": 1.3409864140623351e-05} +{"train_loss": 0.016576897352933884, "grad_norm": 0.09282855689525604, "global_step": 40350, "epoch": 380, "lr": 1.3407844348934112e-05} +{"train_loss": 0.013896607793867588, "grad_norm": 0.12696075439453125, "global_step": 40351, "epoch": 380, "lr": 1.3405824685813818e-05} +{"train_loss": 0.01337558962404728, "grad_norm": 0.06424838304519653, "global_step": 40352, "epoch": 380, "lr": 1.3403805151269516e-05} +{"train_loss": 0.013694980181753635, "grad_norm": 0.061414092779159546, "global_step": 40353, "epoch": 380, "lr": 1.340178574530835e-05} +{"train_loss": 0.01639605686068535, "grad_norm": 0.07525600492954254, "global_step": 40354, "epoch": 380, "lr": 1.3399766467937375e-05} +{"train_loss": 0.012647580355405807, "grad_norm": 0.07323361933231354, "global_step": 40355, "epoch": 380, "lr": 1.3397747319163718e-05} +{"train_loss": 0.014598647132515907, "grad_norm": 0.07421717792749405, "global_step": 40356, "epoch": 380, "lr": 1.3395728298994447e-05} +{"train_loss": 0.016075363382697105, "grad_norm": 0.09369420260190964, "global_step": 40357, "epoch": 380, "lr": 1.3393709407436666e-05} +{"train_loss": 0.01641230843961239, "grad_norm": 0.10314755141735077, "global_step": 40358, "epoch": 380, "lr": 1.3391690644497502e-05} +{"train_loss": 0.014757247641682625, "grad_norm": 0.0737142264842987, "global_step": 40359, "epoch": 380, "lr": 1.3389672010183984e-05} +{"train_loss": 0.016021236777305603, "grad_norm": 0.12479764968156815, "global_step": 40360, "epoch": 380, "lr": 1.3387653504503239e-05} +{"train_loss": 0.012746304273605347, "grad_norm": 0.052648887038230896, "global_step": 40361, "epoch": 380, "lr": 1.338563512746237e-05} +{"train_loss": 0.015557928942143917, "grad_norm": 0.06537868827581406, "global_step": 40362, "epoch": 380, "lr": 1.3383616879068445e-05} +{"train_loss": 0.017251217737793922, "grad_norm": 0.07353532314300537, "global_step": 40363, "epoch": 380, "lr": 1.3381598759328563e-05} +{"train_loss": 0.013898376375436783, "grad_norm": 0.06216926500201225, "global_step": 40364, "epoch": 380, "lr": 1.3379580768249834e-05} +{"train_loss": 0.015029286965727806, "grad_norm": 0.07779909670352936, "global_step": 40365, "epoch": 380, "lr": 1.3377562905839331e-05} +{"train_loss": 0.012713714502751827, "grad_norm": 0.05643981322646141, "global_step": 40366, "epoch": 380, "lr": 1.3375545172104131e-05} +{"train_loss": 0.016645830124616623, "grad_norm": 0.082388736307621, "global_step": 40367, "epoch": 380, "lr": 1.3373527567051342e-05} +{"train_loss": 0.017227690666913986, "grad_norm": 0.12728291749954224, "global_step": 40368, "epoch": 380, "lr": 1.3371510090688055e-05} +{"train_loss": 0.01673533022403717, "grad_norm": 0.12266168743371964, "global_step": 40369, "epoch": 380, "lr": 1.3369492743021344e-05} +{"train_loss": 0.011199592612683773, "grad_norm": 0.05049757659435272, "global_step": 40370, "epoch": 380, "lr": 1.336747552405832e-05} +{"train_loss": 0.013872635550796986, "grad_norm": 0.08765576779842377, "global_step": 40371, "epoch": 380, "lr": 1.3365458433806038e-05} +{"train_loss": 0.01416599377989769, "grad_norm": 0.0710686519742012, "global_step": 40372, "epoch": 380, "lr": 1.3363441472271621e-05} +{"train_loss": 0.01568831130862236, "grad_norm": 0.07484567910432816, "global_step": 40373, "epoch": 380, "lr": 1.3361424639462122e-05} +{"train_loss": 0.013634703122079372, "grad_norm": 0.09477850049734116, "global_step": 40374, "epoch": 380, "lr": 1.3359407935384644e-05} +{"train_loss": 0.015002820640802383, "grad_norm": 0.07568187266588211, "global_step": 40375, "epoch": 380, "lr": 1.3357391360046285e-05} +{"train_loss": 0.015277761034667492, "grad_norm": 0.07303209602832794, "global_step": 40376, "epoch": 380, "lr": 1.3355374913454116e-05} +{"train_loss": 0.017024777829647064, "grad_norm": 0.11703314632177353, "global_step": 40377, "epoch": 380, "lr": 1.3353358595615212e-05} +{"train_loss": 0.014514172449707985, "grad_norm": 0.11270362883806229, "global_step": 40378, "epoch": 380, "lr": 1.3351342406536682e-05} +{"train_loss": 0.0158768892288208, "grad_norm": 0.0921383872628212, "global_step": 40379, "epoch": 380, "lr": 1.3349326346225577e-05} +{"train_loss": 0.015456106513738632, "grad_norm": 0.07565402239561081, "global_step": 40380, "epoch": 380, "lr": 1.3347310414689007e-05} +{"train_loss": 0.015509364195168018, "grad_norm": 0.07623616605997086, "global_step": 40381, "epoch": 380, "lr": 1.3345294611934061e-05} +{"train_loss": 0.015113838016986847, "grad_norm": 0.08686856180429459, "global_step": 40382, "epoch": 380, "lr": 1.3343278937967807e-05} +{"train_loss": 0.014332110062241554, "grad_norm": 0.10565628856420517, "global_step": 40383, "epoch": 380, "lr": 1.334126339279732e-05} +{"train_loss": 0.016882887110114098, "grad_norm": 0.10839738696813583, "global_step": 40384, "epoch": 380, "lr": 1.3339247976429686e-05} +{"train_loss": 0.015028808714013617, "grad_norm": 0.14333969354629517, "global_step": 40385, "epoch": 380, "lr": 1.3337232688872009e-05, "train_batch_mse_error": 0.013379615731537342, "val_loss": 0.01928289607167244} +{"train_loss": 0.01511736586689949, "grad_norm": 0.07520837336778641, "global_step": 40386, "epoch": 381, "lr": 1.3335217530131334e-05} +{"train_loss": 0.014445828273892403, "grad_norm": 0.07126552611589432, "global_step": 40387, "epoch": 381, "lr": 1.3333202500214758e-05} +{"train_loss": 0.013705984689295292, "grad_norm": 0.06334321200847626, "global_step": 40388, "epoch": 381, "lr": 1.3331187599129385e-05} +{"train_loss": 0.01714404486119747, "grad_norm": 0.08031412959098816, "global_step": 40389, "epoch": 381, "lr": 1.3329172826882263e-05} +{"train_loss": 0.01580231264233589, "grad_norm": 0.05315527319908142, "global_step": 40390, "epoch": 381, "lr": 1.332715818348047e-05} +{"train_loss": 0.01599704660475254, "grad_norm": 0.0694400817155838, "global_step": 40391, "epoch": 381, "lr": 1.3325143668931095e-05} +{"train_loss": 0.01366666704416275, "grad_norm": 0.1713295727968216, "global_step": 40392, "epoch": 381, "lr": 1.3323129283241232e-05} +{"train_loss": 0.01303952932357788, "grad_norm": 0.05434808135032654, "global_step": 40393, "epoch": 381, "lr": 1.3321115026417924e-05} +{"train_loss": 0.014943428337574005, "grad_norm": 0.13715912401676178, "global_step": 40394, "epoch": 381, "lr": 1.3319100898468284e-05} +{"train_loss": 0.014123210683465004, "grad_norm": 0.10353762656450272, "global_step": 40395, "epoch": 381, "lr": 1.3317086899399367e-05} +{"train_loss": 0.014751311391592026, "grad_norm": 0.06959911435842514, "global_step": 40396, "epoch": 381, "lr": 1.3315073029218244e-05} +{"train_loss": 0.014604831114411354, "grad_norm": 0.09921543300151825, "global_step": 40397, "epoch": 381, "lr": 1.3313059287932e-05} +{"train_loss": 0.01467911060899496, "grad_norm": 0.05950186401605606, "global_step": 40398, "epoch": 381, "lr": 1.3311045675547724e-05} +{"train_loss": 0.012832747772336006, "grad_norm": 0.0749727413058281, "global_step": 40399, "epoch": 381, "lr": 1.3309032192072463e-05} +{"train_loss": 0.01621491089463234, "grad_norm": 0.07161448895931244, "global_step": 40400, "epoch": 381, "lr": 1.330701883751332e-05} +{"train_loss": 0.014220542274415493, "grad_norm": 0.06824515759944916, "global_step": 40401, "epoch": 381, "lr": 1.3305005611877342e-05} +{"train_loss": 0.014179432764649391, "grad_norm": 0.0954955443739891, "global_step": 40402, "epoch": 381, "lr": 1.3302992515171631e-05} +{"train_loss": 0.016266807913780212, "grad_norm": 0.08855699747800827, "global_step": 40403, "epoch": 381, "lr": 1.330097954740323e-05} +{"train_loss": 0.014503221958875656, "grad_norm": 0.11002608388662338, "global_step": 40404, "epoch": 381, "lr": 1.3298966708579225e-05} +{"train_loss": 0.01562470942735672, "grad_norm": 0.07537767291069031, "global_step": 40405, "epoch": 381, "lr": 1.3296953998706707e-05} +{"train_loss": 0.01351558230817318, "grad_norm": 0.08585868030786514, "global_step": 40406, "epoch": 381, "lr": 1.3294941417792734e-05} +{"train_loss": 0.012126734480261803, "grad_norm": 0.13053156435489655, "global_step": 40407, "epoch": 381, "lr": 1.3292928965844354e-05} +{"train_loss": 0.016218774020671844, "grad_norm": 0.1234961599111557, "global_step": 40408, "epoch": 381, "lr": 1.3290916642868661e-05} +{"train_loss": 0.017255596816539764, "grad_norm": 0.1027495265007019, "global_step": 40409, "epoch": 381, "lr": 1.3288904448872735e-05} +{"train_loss": 0.01533016748726368, "grad_norm": 0.05712537467479706, "global_step": 40410, "epoch": 381, "lr": 1.3286892383863625e-05} +{"train_loss": 0.015350477769970894, "grad_norm": 0.10099165141582489, "global_step": 40411, "epoch": 381, "lr": 1.3284880447848419e-05} +{"train_loss": 0.015925534069538116, "grad_norm": 0.06122172996401787, "global_step": 40412, "epoch": 381, "lr": 1.3282868640834162e-05} +{"train_loss": 0.01927589625120163, "grad_norm": 0.08233866840600967, "global_step": 40413, "epoch": 381, "lr": 1.3280856962827948e-05} +{"train_loss": 0.013904109597206116, "grad_norm": 0.13017785549163818, "global_step": 40414, "epoch": 381, "lr": 1.3278845413836827e-05} +{"train_loss": 0.014777223579585552, "grad_norm": 0.06688720732927322, "global_step": 40415, "epoch": 381, "lr": 1.3276833993867865e-05} +{"train_loss": 0.013757706619799137, "grad_norm": 0.08866339176893234, "global_step": 40416, "epoch": 381, "lr": 1.3274822702928157e-05} +{"train_loss": 0.01339193619787693, "grad_norm": 0.09903616458177567, "global_step": 40417, "epoch": 381, "lr": 1.3272811541024749e-05} +{"train_loss": 0.016174476593732834, "grad_norm": 0.07211867719888687, "global_step": 40418, "epoch": 381, "lr": 1.32708005081647e-05} +{"train_loss": 0.01438663899898529, "grad_norm": 0.06838371604681015, "global_step": 40419, "epoch": 381, "lr": 1.3268789604355091e-05} +{"train_loss": 0.014282611198723316, "grad_norm": 0.06168821081519127, "global_step": 40420, "epoch": 381, "lr": 1.3266778829602972e-05} +{"train_loss": 0.01595338061451912, "grad_norm": 0.08222009241580963, "global_step": 40421, "epoch": 381, "lr": 1.3264768183915416e-05} +{"train_loss": 0.014732498675584793, "grad_norm": 0.06236719340085983, "global_step": 40422, "epoch": 381, "lr": 1.3262757667299508e-05} +{"train_loss": 0.014447318390011787, "grad_norm": 0.0700848400592804, "global_step": 40423, "epoch": 381, "lr": 1.3260747279762287e-05} +{"train_loss": 0.014065507799386978, "grad_norm": 0.055400293320417404, "global_step": 40424, "epoch": 381, "lr": 1.3258737021310807e-05} +{"train_loss": 0.016343172639608383, "grad_norm": 0.0961548238992691, "global_step": 40425, "epoch": 381, "lr": 1.3256726891952149e-05} +{"train_loss": 0.015639692544937134, "grad_norm": 0.06966055184602737, "global_step": 40426, "epoch": 381, "lr": 1.3254716891693392e-05} +{"train_loss": 0.016047189012169838, "grad_norm": 0.0930345356464386, "global_step": 40427, "epoch": 381, "lr": 1.325270702054156e-05} +{"train_loss": 0.012710551731288433, "grad_norm": 0.1072157472372055, "global_step": 40428, "epoch": 381, "lr": 1.3250697278503737e-05} +{"train_loss": 0.012695889919996262, "grad_norm": 0.059551794081926346, "global_step": 40429, "epoch": 381, "lr": 1.3248687665586996e-05} +{"train_loss": 0.015192795544862747, "grad_norm": 0.09355653077363968, "global_step": 40430, "epoch": 381, "lr": 1.3246678181798383e-05} +{"train_loss": 0.01752883940935135, "grad_norm": 0.08414101600646973, "global_step": 40431, "epoch": 381, "lr": 1.3244668827144945e-05} +{"train_loss": 0.015429355204105377, "grad_norm": 0.11060387641191483, "global_step": 40432, "epoch": 381, "lr": 1.3242659601633756e-05} +{"train_loss": 0.013175958767533302, "grad_norm": 0.102354034781456, "global_step": 40433, "epoch": 381, "lr": 1.324065050527189e-05} +{"train_loss": 0.014013610780239105, "grad_norm": 0.0665377825498581, "global_step": 40434, "epoch": 381, "lr": 1.3238641538066376e-05} +{"train_loss": 0.016306336969137192, "grad_norm": 0.13843105733394623, "global_step": 40435, "epoch": 381, "lr": 1.323663270002431e-05} +{"train_loss": 0.013601375743746758, "grad_norm": 0.08703162521123886, "global_step": 40436, "epoch": 381, "lr": 1.3234623991152723e-05} +{"train_loss": 0.01440505776554346, "grad_norm": 0.16785161197185516, "global_step": 40437, "epoch": 381, "lr": 1.3232615411458665e-05} +{"train_loss": 0.014474681578576565, "grad_norm": 0.07335881143808365, "global_step": 40438, "epoch": 381, "lr": 1.3230606960949205e-05} +{"train_loss": 0.013879122212529182, "grad_norm": 0.09539464116096497, "global_step": 40439, "epoch": 381, "lr": 1.3228598639631418e-05} +{"train_loss": 0.013575240969657898, "grad_norm": 0.08981883525848389, "global_step": 40440, "epoch": 381, "lr": 1.3226590447512333e-05} +{"train_loss": 0.01376404706388712, "grad_norm": 0.16753917932510376, "global_step": 40441, "epoch": 381, "lr": 1.3224582384599028e-05} +{"train_loss": 0.014014222659170628, "grad_norm": 0.05578666925430298, "global_step": 40442, "epoch": 381, "lr": 1.3222574450898528e-05} +{"train_loss": 0.016319824382662773, "grad_norm": 0.06991036236286163, "global_step": 40443, "epoch": 381, "lr": 1.3220566646417926e-05} +{"train_loss": 0.011930338107049465, "grad_norm": 0.07759112119674683, "global_step": 40444, "epoch": 381, "lr": 1.3218558971164246e-05} +{"train_loss": 0.014096520841121674, "grad_norm": 0.061520304530858994, "global_step": 40445, "epoch": 381, "lr": 1.3216551425144551e-05} +{"train_loss": 0.017643146216869354, "grad_norm": 0.09453794360160828, "global_step": 40446, "epoch": 381, "lr": 1.321454400836591e-05} +{"train_loss": 0.015021849423646927, "grad_norm": 0.0813809260725975, "global_step": 40447, "epoch": 381, "lr": 1.3212536720835366e-05} +{"train_loss": 0.016455253586173058, "grad_norm": 0.12732844054698944, "global_step": 40448, "epoch": 381, "lr": 1.3210529562559959e-05} +{"train_loss": 0.015349362976849079, "grad_norm": 0.07587870210409164, "global_step": 40449, "epoch": 381, "lr": 1.3208522533546747e-05} +{"train_loss": 0.015750236809253693, "grad_norm": 0.0841132178902626, "global_step": 40450, "epoch": 381, "lr": 1.3206515633802802e-05} +{"train_loss": 0.016861263662576675, "grad_norm": 0.10516144335269928, "global_step": 40451, "epoch": 381, "lr": 1.3204508863335147e-05} +{"train_loss": 0.015026066452264786, "grad_norm": 0.08211807906627655, "global_step": 40452, "epoch": 381, "lr": 1.3202502222150841e-05} +{"train_loss": 0.014703931286931038, "grad_norm": 0.07859355956315994, "global_step": 40453, "epoch": 381, "lr": 1.3200495710256972e-05} +{"train_loss": 0.013406220823526382, "grad_norm": 0.07253604382276535, "global_step": 40454, "epoch": 381, "lr": 1.3198489327660529e-05} +{"train_loss": 0.01779114082455635, "grad_norm": 0.09745858609676361, "global_step": 40455, "epoch": 381, "lr": 1.3196483074368593e-05} +{"train_loss": 0.01735539734363556, "grad_norm": 0.1256236433982849, "global_step": 40456, "epoch": 381, "lr": 1.3194476950388202e-05} +{"train_loss": 0.01723015494644642, "grad_norm": 0.08135201036930084, "global_step": 40457, "epoch": 381, "lr": 1.3192470955726433e-05} +{"train_loss": 0.015921518206596375, "grad_norm": 0.07000034302473068, "global_step": 40458, "epoch": 381, "lr": 1.3190465090390314e-05} +{"train_loss": 0.012437054887413979, "grad_norm": 0.07604983448982239, "global_step": 40459, "epoch": 381, "lr": 1.3188459354386878e-05} +{"train_loss": 0.016405316069722176, "grad_norm": 0.10678158700466156, "global_step": 40460, "epoch": 381, "lr": 1.31864537477232e-05} +{"train_loss": 0.014182488434016705, "grad_norm": 0.09460890293121338, "global_step": 40461, "epoch": 381, "lr": 1.3184448270406297e-05} +{"train_loss": 0.014637991786003113, "grad_norm": 0.053794655948877335, "global_step": 40462, "epoch": 381, "lr": 1.3182442922443233e-05} +{"train_loss": 0.014636063948273659, "grad_norm": 0.10609706491231918, "global_step": 40463, "epoch": 381, "lr": 1.318043770384107e-05} +{"train_loss": 0.014839138835668564, "grad_norm": 0.11115673929452896, "global_step": 40464, "epoch": 381, "lr": 1.3178432614606834e-05} +{"train_loss": 0.013312041759490967, "grad_norm": 0.0906304121017456, "global_step": 40465, "epoch": 381, "lr": 1.3176427654747564e-05} +{"train_loss": 0.014213766902685165, "grad_norm": 0.08527868241071701, "global_step": 40466, "epoch": 381, "lr": 1.3174422824270305e-05} +{"train_loss": 0.0155560914427042, "grad_norm": 0.15550807118415833, "global_step": 40467, "epoch": 381, "lr": 1.3172418123182128e-05} +{"train_loss": 0.017272964119911194, "grad_norm": 0.11332138627767563, "global_step": 40468, "epoch": 381, "lr": 1.317041355149004e-05} +{"train_loss": 0.016034286469221115, "grad_norm": 0.11706507205963135, "global_step": 40469, "epoch": 381, "lr": 1.3168409109201108e-05} +{"train_loss": 0.015405206009745598, "grad_norm": 0.09671149402856827, "global_step": 40470, "epoch": 381, "lr": 1.316640479632238e-05} +{"train_loss": 0.013514048419892788, "grad_norm": 0.06997925788164139, "global_step": 40471, "epoch": 381, "lr": 1.316440061286089e-05} +{"train_loss": 0.016442326828837395, "grad_norm": 0.08447570353746414, "global_step": 40472, "epoch": 381, "lr": 1.3162396558823653e-05} +{"train_loss": 0.01314995251595974, "grad_norm": 0.0883120596408844, "global_step": 40473, "epoch": 381, "lr": 1.3160392634217744e-05} +{"train_loss": 0.015235292725265026, "grad_norm": 0.07645069062709808, "global_step": 40474, "epoch": 381, "lr": 1.3158388839050207e-05} +{"train_loss": 0.015096901915967464, "grad_norm": 0.10936418175697327, "global_step": 40475, "epoch": 381, "lr": 1.3156385173328051e-05} +{"train_loss": 0.01447029784321785, "grad_norm": 0.1395946890115738, "global_step": 40476, "epoch": 381, "lr": 1.3154381637058349e-05} +{"train_loss": 0.018101777881383896, "grad_norm": 0.12820570170879364, "global_step": 40477, "epoch": 381, "lr": 1.315237823024813e-05} +{"train_loss": 0.015158193185925484, "grad_norm": 0.10728667676448822, "global_step": 40478, "epoch": 381, "lr": 1.3150374952904414e-05} +{"train_loss": 0.015226365998387337, "grad_norm": 0.09043844789266586, "global_step": 40479, "epoch": 381, "lr": 1.3148371805034253e-05} +{"train_loss": 0.01669435389339924, "grad_norm": 0.06853772699832916, "global_step": 40480, "epoch": 381, "lr": 1.3146368786644703e-05} +{"train_loss": 0.014092907309532166, "grad_norm": 0.08984830230474472, "global_step": 40481, "epoch": 381, "lr": 1.3144365897742772e-05} +{"train_loss": 0.013841621577739716, "grad_norm": 0.09843752533197403, "global_step": 40482, "epoch": 381, "lr": 1.3142363138335522e-05} +{"train_loss": 0.01626591756939888, "grad_norm": 0.06259596347808838, "global_step": 40483, "epoch": 381, "lr": 1.3140360508429972e-05} +{"train_loss": 0.01342802681028843, "grad_norm": 0.06023659184575081, "global_step": 40484, "epoch": 381, "lr": 1.3138358008033175e-05} +{"train_loss": 0.014987180940806866, "grad_norm": 0.06577423214912415, "global_step": 40485, "epoch": 381, "lr": 1.3136355637152142e-05} +{"train_loss": 0.014808226376771927, "grad_norm": 0.1380389779806137, "global_step": 40486, "epoch": 381, "lr": 1.3134353395793924e-05} +{"train_loss": 0.015292370691895485, "grad_norm": 0.05332919582724571, "global_step": 40487, "epoch": 381, "lr": 1.3132351283965572e-05} +{"train_loss": 0.015891434624791145, "grad_norm": 0.07364985346794128, "global_step": 40488, "epoch": 381, "lr": 1.31303493016741e-05} +{"train_loss": 0.018830465152859688, "grad_norm": 0.07667052000761032, "global_step": 40489, "epoch": 381, "lr": 1.3128347448926537e-05} +{"train_loss": 0.01522090844810009, "grad_norm": 0.08958908170461655, "global_step": 40490, "epoch": 381, "lr": 1.3126345725729921e-05} +{"train_loss": 0.015020189824391086, "grad_norm": 0.08988888561725616, "global_step": 40491, "epoch": 381, "lr": 1.312434413209131e-05, "val_loss": 0.019023217260837555} +{"train_loss": 0.01390616875141859, "grad_norm": 0.10361731797456741, "global_step": 40492, "epoch": 382, "lr": 1.31223426680177e-05} +{"train_loss": 0.01637701876461506, "grad_norm": 0.06183099374175072, "global_step": 40493, "epoch": 382, "lr": 1.3120341333516139e-05} +{"train_loss": 0.015662968158721924, "grad_norm": 0.06219892203807831, "global_step": 40494, "epoch": 382, "lr": 1.3118340128593693e-05} +{"train_loss": 0.0151672987267375, "grad_norm": 0.11831247806549072, "global_step": 40495, "epoch": 382, "lr": 1.3116339053257332e-05} +{"train_loss": 0.015572924166917801, "grad_norm": 0.09422433376312256, "global_step": 40496, "epoch": 382, "lr": 1.3114338107514107e-05} +{"train_loss": 0.014311710372567177, "grad_norm": 0.10948820412158966, "global_step": 40497, "epoch": 382, "lr": 1.311233729137108e-05} +{"train_loss": 0.01696433126926422, "grad_norm": 0.11711754649877548, "global_step": 40498, "epoch": 382, "lr": 1.3110336604835239e-05} +{"train_loss": 0.015610972419381142, "grad_norm": 0.06137334927916527, "global_step": 40499, "epoch": 382, "lr": 1.3108336047913633e-05} +{"train_loss": 0.014740286394953728, "grad_norm": 0.11652829498052597, "global_step": 40500, "epoch": 382, "lr": 1.3106335620613303e-05} +{"train_loss": 0.018431901931762695, "grad_norm": 0.06599162518978119, "global_step": 40501, "epoch": 382, "lr": 1.3104335322941269e-05} +{"train_loss": 0.015438435599207878, "grad_norm": 0.10108309239149094, "global_step": 40502, "epoch": 382, "lr": 1.3102335154904537e-05} +{"train_loss": 0.014398330822587013, "grad_norm": 0.07761713862419128, "global_step": 40503, "epoch": 382, "lr": 1.3100335116510154e-05} +{"train_loss": 0.014964021742343903, "grad_norm": 0.06223093718290329, "global_step": 40504, "epoch": 382, "lr": 1.3098335207765161e-05} +{"train_loss": 0.01567528210580349, "grad_norm": 0.08615045994520187, "global_step": 40505, "epoch": 382, "lr": 1.3096335428676554e-05} +{"train_loss": 0.01410211157053709, "grad_norm": 0.06999284029006958, "global_step": 40506, "epoch": 382, "lr": 1.3094335779251394e-05} +{"train_loss": 0.013231322169303894, "grad_norm": 0.06547585129737854, "global_step": 40507, "epoch": 382, "lr": 1.3092336259496668e-05} +{"train_loss": 0.014478779397904873, "grad_norm": 0.12445532530546188, "global_step": 40508, "epoch": 382, "lr": 1.309033686941944e-05} +{"train_loss": 0.015762414783239365, "grad_norm": 0.07537241280078888, "global_step": 40509, "epoch": 382, "lr": 1.3088337609026702e-05} +{"train_loss": 0.014654850587248802, "grad_norm": 0.07374469935894012, "global_step": 40510, "epoch": 382, "lr": 1.3086338478325493e-05} +{"train_loss": 0.015546748414635658, "grad_norm": 0.0809076651930809, "global_step": 40511, "epoch": 382, "lr": 1.3084339477322848e-05} +{"train_loss": 0.015522986650466919, "grad_norm": 0.07626349478960037, "global_step": 40512, "epoch": 382, "lr": 1.3082340606025783e-05} +{"train_loss": 0.014577750116586685, "grad_norm": 0.08603363484144211, "global_step": 40513, "epoch": 382, "lr": 1.3080341864441303e-05} +{"train_loss": 0.014906642027199268, "grad_norm": 0.07150860875844955, "global_step": 40514, "epoch": 382, "lr": 1.3078343252576447e-05} +{"train_loss": 0.018243256956338882, "grad_norm": 0.1768542230129242, "global_step": 40515, "epoch": 382, "lr": 1.3076344770438254e-05} +{"train_loss": 0.013625584542751312, "grad_norm": 0.06546930223703384, "global_step": 40516, "epoch": 382, "lr": 1.3074346418033706e-05} +{"train_loss": 0.012696836143732071, "grad_norm": 0.11411987990140915, "global_step": 40517, "epoch": 382, "lr": 1.3072348195369865e-05} +{"train_loss": 0.014581412076950073, "grad_norm": 0.09438052028417587, "global_step": 40518, "epoch": 382, "lr": 1.3070350102453732e-05} +{"train_loss": 0.014272546395659447, "grad_norm": 0.09561813622713089, "global_step": 40519, "epoch": 382, "lr": 1.3068352139292312e-05} +{"train_loss": 0.014826902188360691, "grad_norm": 0.10597214847803116, "global_step": 40520, "epoch": 382, "lr": 1.3066354305892647e-05} +{"train_loss": 0.015430759638547897, "grad_norm": 0.06251076608896255, "global_step": 40521, "epoch": 382, "lr": 1.3064356602261762e-05} +{"train_loss": 0.01544615812599659, "grad_norm": 0.06503316015005112, "global_step": 40522, "epoch": 382, "lr": 1.3062359028406646e-05} +{"train_loss": 0.015118665993213654, "grad_norm": 0.08939776569604874, "global_step": 40523, "epoch": 382, "lr": 1.3060361584334357e-05} +{"train_loss": 0.013457899913191795, "grad_norm": 0.07590112835168839, "global_step": 40524, "epoch": 382, "lr": 1.3058364270051876e-05} +{"train_loss": 0.01424156129360199, "grad_norm": 0.05626332014799118, "global_step": 40525, "epoch": 382, "lr": 1.305636708556625e-05} +{"train_loss": 0.016158422455191612, "grad_norm": 0.07770711183547974, "global_step": 40526, "epoch": 382, "lr": 1.3054370030884478e-05} +{"train_loss": 0.015430836006999016, "grad_norm": 0.10110966861248016, "global_step": 40527, "epoch": 382, "lr": 1.3052373106013571e-05} +{"train_loss": 0.018906008452177048, "grad_norm": 0.10450133681297302, "global_step": 40528, "epoch": 382, "lr": 1.3050376310960576e-05} +{"train_loss": 0.015517592430114746, "grad_norm": 0.09131090342998505, "global_step": 40529, "epoch": 382, "lr": 1.3048379645732496e-05} +{"train_loss": 0.013584259897470474, "grad_norm": 0.0825086310505867, "global_step": 40530, "epoch": 382, "lr": 1.3046383110336319e-05} +{"train_loss": 0.015826629474759102, "grad_norm": 0.07772055268287659, "global_step": 40531, "epoch": 382, "lr": 1.3044386704779083e-05} +{"train_loss": 0.01559649221599102, "grad_norm": 0.08347108960151672, "global_step": 40532, "epoch": 382, "lr": 1.3042390429067814e-05} +{"train_loss": 0.01515229232609272, "grad_norm": 0.07490360736846924, "global_step": 40533, "epoch": 382, "lr": 1.3040394283209494e-05} +{"train_loss": 0.01650567725300789, "grad_norm": 0.10813599824905396, "global_step": 40534, "epoch": 382, "lr": 1.3038398267211161e-05} +{"train_loss": 0.013618841767311096, "grad_norm": 0.05592354014515877, "global_step": 40535, "epoch": 382, "lr": 1.3036402381079848e-05} +{"train_loss": 0.014878932386636734, "grad_norm": 0.0994706004858017, "global_step": 40536, "epoch": 382, "lr": 1.3034406624822515e-05} +{"train_loss": 0.016439277678728104, "grad_norm": 0.08029946684837341, "global_step": 40537, "epoch": 382, "lr": 1.3032410998446199e-05} +{"train_loss": 0.01343470998108387, "grad_norm": 0.06723467260599136, "global_step": 40538, "epoch": 382, "lr": 1.3030415501957932e-05} +{"train_loss": 0.015961354598402977, "grad_norm": 0.10651755332946777, "global_step": 40539, "epoch": 382, "lr": 1.3028420135364695e-05} +{"train_loss": 0.01696706749498844, "grad_norm": 0.14804700016975403, "global_step": 40540, "epoch": 382, "lr": 1.30264248986735e-05} +{"train_loss": 0.01436874270439148, "grad_norm": 0.09152524173259735, "global_step": 40541, "epoch": 382, "lr": 1.3024429791891397e-05} +{"train_loss": 0.011349369771778584, "grad_norm": 0.06381099671125412, "global_step": 40542, "epoch": 382, "lr": 1.3022434815025358e-05} +{"train_loss": 0.012488024309277534, "grad_norm": 0.11783907562494278, "global_step": 40543, "epoch": 382, "lr": 1.3020439968082387e-05} +{"train_loss": 0.014807132072746754, "grad_norm": 0.0693177878856659, "global_step": 40544, "epoch": 382, "lr": 1.301844525106951e-05} +{"train_loss": 0.013199036940932274, "grad_norm": 0.12718720734119415, "global_step": 40545, "epoch": 382, "lr": 1.3016450663993752e-05} +{"train_loss": 0.015748031437397003, "grad_norm": 0.06889789551496506, "global_step": 40546, "epoch": 382, "lr": 1.3014456206862085e-05} +{"train_loss": 0.013745835050940514, "grad_norm": 0.08289012312889099, "global_step": 40547, "epoch": 382, "lr": 1.3012461879681547e-05} +{"train_loss": 0.014733407646417618, "grad_norm": 0.08271045237779617, "global_step": 40548, "epoch": 382, "lr": 1.3010467682459121e-05} +{"train_loss": 0.015237625688314438, "grad_norm": 0.1471116840839386, "global_step": 40549, "epoch": 382, "lr": 1.3008473615201838e-05} +{"train_loss": 0.015265184454619884, "grad_norm": 0.08074384927749634, "global_step": 40550, "epoch": 382, "lr": 1.3006479677916677e-05} +{"train_loss": 0.017700567841529846, "grad_norm": 0.1138920783996582, "global_step": 40551, "epoch": 382, "lr": 1.3004485870610661e-05} +{"train_loss": 0.016142161563038826, "grad_norm": 0.056710660457611084, "global_step": 40552, "epoch": 382, "lr": 1.3002492193290805e-05} +{"train_loss": 0.014209998771548271, "grad_norm": 0.062262192368507385, "global_step": 40553, "epoch": 382, "lr": 1.3000498645964105e-05} +{"train_loss": 0.014172248542308807, "grad_norm": 0.07264765352010727, "global_step": 40554, "epoch": 382, "lr": 1.2998505228637541e-05} +{"train_loss": 0.014693872071802616, "grad_norm": 0.10211734473705292, "global_step": 40555, "epoch": 382, "lr": 1.2996511941318157e-05} +{"train_loss": 0.01471453532576561, "grad_norm": 0.05670581012964249, "global_step": 40556, "epoch": 382, "lr": 1.2994518784012922e-05} +{"train_loss": 0.013185542076826096, "grad_norm": 0.08041056990623474, "global_step": 40557, "epoch": 382, "lr": 1.2992525756728857e-05} +{"train_loss": 0.014095071703195572, "grad_norm": 0.07327471673488617, "global_step": 40558, "epoch": 382, "lr": 1.2990532859472959e-05} +{"train_loss": 0.013873236253857613, "grad_norm": 0.05116257071495056, "global_step": 40559, "epoch": 382, "lr": 1.2988540092252266e-05} +{"train_loss": 0.013550717383623123, "grad_norm": 0.12521538138389587, "global_step": 40560, "epoch": 382, "lr": 1.2986547455073716e-05} +{"train_loss": 0.015888173133134842, "grad_norm": 0.056383825838565826, "global_step": 40561, "epoch": 382, "lr": 1.2984554947944338e-05} +{"train_loss": 0.01523946225643158, "grad_norm": 0.05423128232359886, "global_step": 40562, "epoch": 382, "lr": 1.298256257087116e-05} +{"train_loss": 0.013867665082216263, "grad_norm": 0.08257385343313217, "global_step": 40563, "epoch": 382, "lr": 1.2980570323861135e-05} +{"train_loss": 0.013171203434467316, "grad_norm": 0.05530034005641937, "global_step": 40564, "epoch": 382, "lr": 1.2978578206921304e-05} +{"train_loss": 0.01756187714636326, "grad_norm": 0.07437317073345184, "global_step": 40565, "epoch": 382, "lr": 1.2976586220058633e-05} +{"train_loss": 0.013897795230150223, "grad_norm": 0.11930454522371292, "global_step": 40566, "epoch": 382, "lr": 1.2974594363280157e-05} +{"train_loss": 0.015084725804626942, "grad_norm": 0.07363211363554001, "global_step": 40567, "epoch": 382, "lr": 1.297260263659284e-05} +{"train_loss": 0.013806009665131569, "grad_norm": 0.05707173049449921, "global_step": 40568, "epoch": 382, "lr": 1.2970611040003688e-05} +{"train_loss": 0.014645460061728954, "grad_norm": 0.0805007815361023, "global_step": 40569, "epoch": 382, "lr": 1.2968619573519725e-05} +{"train_loss": 0.014150582253932953, "grad_norm": 0.10988391935825348, "global_step": 40570, "epoch": 382, "lr": 1.2966628237147926e-05} +{"train_loss": 0.013748512603342533, "grad_norm": 0.0567476712167263, "global_step": 40571, "epoch": 382, "lr": 1.2964637030895277e-05} +{"train_loss": 0.015040233731269836, "grad_norm": 0.09796970337629318, "global_step": 40572, "epoch": 382, "lr": 1.2962645954768782e-05} +{"train_loss": 0.015316489152610302, "grad_norm": 0.05488945171236992, "global_step": 40573, "epoch": 382, "lr": 1.2960655008775463e-05} +{"train_loss": 0.012959565967321396, "grad_norm": 0.06063562631607056, "global_step": 40574, "epoch": 382, "lr": 1.2958664192922276e-05} +{"train_loss": 0.014721089974045753, "grad_norm": 0.07405275106430054, "global_step": 40575, "epoch": 382, "lr": 1.2956673507216233e-05} +{"train_loss": 0.014594350010156631, "grad_norm": 0.08307600021362305, "global_step": 40576, "epoch": 382, "lr": 1.2954682951664365e-05} +{"train_loss": 0.015945803374052048, "grad_norm": 0.09992101788520813, "global_step": 40577, "epoch": 382, "lr": 1.2952692526273597e-05} +{"train_loss": 0.015468752942979336, "grad_norm": 0.06944626569747925, "global_step": 40578, "epoch": 382, "lr": 1.2950702231050954e-05} +{"train_loss": 0.019493468105793, "grad_norm": 0.07681039720773697, "global_step": 40579, "epoch": 382, "lr": 1.294871206600345e-05} +{"train_loss": 0.012804752215743065, "grad_norm": 0.09349816292524338, "global_step": 40580, "epoch": 382, "lr": 1.2946722031138042e-05} +{"train_loss": 0.018217427656054497, "grad_norm": 0.10807182639837265, "global_step": 40581, "epoch": 382, "lr": 1.2944732126461739e-05} +{"train_loss": 0.014561263844370842, "grad_norm": 0.09182469546794891, "global_step": 40582, "epoch": 382, "lr": 1.2942742351981547e-05} +{"train_loss": 0.012947214767336845, "grad_norm": 0.06787066906690598, "global_step": 40583, "epoch": 382, "lr": 1.294075270770444e-05} +{"train_loss": 0.013520454056560993, "grad_norm": 0.0621306486427784, "global_step": 40584, "epoch": 382, "lr": 1.29387631936374e-05} +{"train_loss": 0.01566310040652752, "grad_norm": 0.06626013666391373, "global_step": 40585, "epoch": 382, "lr": 1.2936773809787429e-05} +{"train_loss": 0.014508849009871483, "grad_norm": 0.06072912737727165, "global_step": 40586, "epoch": 382, "lr": 1.2934784556161528e-05} +{"train_loss": 0.014232451096177101, "grad_norm": 0.0683896616101265, "global_step": 40587, "epoch": 382, "lr": 1.2932795432766665e-05} +{"train_loss": 0.015889015048742294, "grad_norm": 0.11324412375688553, "global_step": 40588, "epoch": 382, "lr": 1.2930806439609849e-05} +{"train_loss": 0.015600236132740974, "grad_norm": 0.07882578670978546, "global_step": 40589, "epoch": 382, "lr": 1.2928817576698044e-05} +{"train_loss": 0.01738521084189415, "grad_norm": 0.10810081660747528, "global_step": 40590, "epoch": 382, "lr": 1.2926828844038269e-05} +{"train_loss": 0.014685416594147682, "grad_norm": 0.1161712110042572, "global_step": 40591, "epoch": 382, "lr": 1.2924840241637481e-05} +{"train_loss": 0.014967270195484161, "grad_norm": 0.1206248551607132, "global_step": 40592, "epoch": 382, "lr": 1.2922851769502675e-05} +{"train_loss": 0.015813086181879044, "grad_norm": 0.13363777101039886, "global_step": 40593, "epoch": 382, "lr": 1.2920863427640867e-05} +{"train_loss": 0.01719892956316471, "grad_norm": 0.12337061762809753, "global_step": 40594, "epoch": 382, "lr": 1.2918875216059012e-05} +{"train_loss": 0.01564578153192997, "grad_norm": 0.06900312751531601, "global_step": 40595, "epoch": 382, "lr": 1.2916887134764088e-05} +{"train_loss": 0.015465633943676949, "grad_norm": 0.0683976486325264, "global_step": 40596, "epoch": 382, "lr": 1.2914899183763118e-05} +{"train_loss": 0.015035768900558634, "grad_norm": 0.13961143791675568, "global_step": 40597, "epoch": 382, "lr": 1.2912911363063046e-05, "val_loss": 0.019792554900050163} +{"train_loss": 0.013453424908220768, "grad_norm": 0.09568921476602554, "global_step": 40598, "epoch": 383, "lr": 1.2910923672670871e-05} +{"train_loss": 0.015237182378768921, "grad_norm": 0.06127684563398361, "global_step": 40599, "epoch": 383, "lr": 1.29089361125936e-05} +{"train_loss": 0.01610318012535572, "grad_norm": 0.06921199709177017, "global_step": 40600, "epoch": 383, "lr": 1.2906948682838194e-05} +{"train_loss": 0.013605600222945213, "grad_norm": 0.07150056213140488, "global_step": 40601, "epoch": 383, "lr": 1.2904961383411623e-05} +{"train_loss": 0.015662550926208496, "grad_norm": 0.06336139142513275, "global_step": 40602, "epoch": 383, "lr": 1.290297421432089e-05} +{"train_loss": 0.014262638986110687, "grad_norm": 0.06147058680653572, "global_step": 40603, "epoch": 383, "lr": 1.2900987175572988e-05} +{"train_loss": 0.013579011894762516, "grad_norm": 0.08995185792446136, "global_step": 40604, "epoch": 383, "lr": 1.2899000267174866e-05} +{"train_loss": 0.015448259189724922, "grad_norm": 0.09567520767450333, "global_step": 40605, "epoch": 383, "lr": 1.289701348913352e-05} +{"train_loss": 0.014877505600452423, "grad_norm": 0.13147160410881042, "global_step": 40606, "epoch": 383, "lr": 1.2895026841455949e-05} +{"train_loss": 0.016080642119050026, "grad_norm": 0.07535681873559952, "global_step": 40607, "epoch": 383, "lr": 1.2893040324149119e-05} +{"train_loss": 0.015091469511389732, "grad_norm": 0.0764860063791275, "global_step": 40608, "epoch": 383, "lr": 1.2891053937219993e-05} +{"train_loss": 0.014088761061429977, "grad_norm": 0.0722171738743782, "global_step": 40609, "epoch": 383, "lr": 1.2889067680675564e-05} +{"train_loss": 0.016565777361392975, "grad_norm": 0.0819065049290657, "global_step": 40610, "epoch": 383, "lr": 1.2887081554522823e-05} +{"train_loss": 0.016424652189016342, "grad_norm": 0.06471503525972366, "global_step": 40611, "epoch": 383, "lr": 1.2885095558768746e-05} +{"train_loss": 0.013308383524417877, "grad_norm": 0.06085565313696861, "global_step": 40612, "epoch": 383, "lr": 1.2883109693420282e-05} +{"train_loss": 0.015920106321573257, "grad_norm": 0.07037480175495148, "global_step": 40613, "epoch": 383, "lr": 1.2881123958484432e-05} +{"train_loss": 0.016579635441303253, "grad_norm": 0.09798447042703629, "global_step": 40614, "epoch": 383, "lr": 1.2879138353968178e-05} +{"train_loss": 0.014918327331542969, "grad_norm": 0.06487635523080826, "global_step": 40615, "epoch": 383, "lr": 1.2877152879878479e-05} +{"train_loss": 0.014717955142259598, "grad_norm": 0.07118982076644897, "global_step": 40616, "epoch": 383, "lr": 1.2875167536222322e-05} +{"train_loss": 0.014064814895391464, "grad_norm": 0.0851292535662651, "global_step": 40617, "epoch": 383, "lr": 1.2873182323006705e-05} +{"train_loss": 0.01757923513650894, "grad_norm": 0.13369201123714447, "global_step": 40618, "epoch": 383, "lr": 1.2871197240238552e-05} +{"train_loss": 0.015640320256352425, "grad_norm": 0.09023222327232361, "global_step": 40619, "epoch": 383, "lr": 1.286921228792487e-05} +{"train_loss": 0.013654494658112526, "grad_norm": 0.06112636625766754, "global_step": 40620, "epoch": 383, "lr": 1.2867227466072634e-05} +{"train_loss": 0.013708578422665596, "grad_norm": 0.06144673004746437, "global_step": 40621, "epoch": 383, "lr": 1.2865242774688807e-05} +{"train_loss": 0.01671186275780201, "grad_norm": 0.09834688901901245, "global_step": 40622, "epoch": 383, "lr": 1.286325821378036e-05} +{"train_loss": 0.015280180610716343, "grad_norm": 0.05796065926551819, "global_step": 40623, "epoch": 383, "lr": 1.2861273783354294e-05} +{"train_loss": 0.015232548117637634, "grad_norm": 0.12203027307987213, "global_step": 40624, "epoch": 383, "lr": 1.2859289483417558e-05} +{"train_loss": 0.015462322160601616, "grad_norm": 0.06147283688187599, "global_step": 40625, "epoch": 383, "lr": 1.2857305313977109e-05} +{"train_loss": 0.01614561676979065, "grad_norm": 0.08933316916227341, "global_step": 40626, "epoch": 383, "lr": 1.2855321275039939e-05} +{"train_loss": 0.013441354036331177, "grad_norm": 0.1296737641096115, "global_step": 40627, "epoch": 383, "lr": 1.2853337366613028e-05} +{"train_loss": 0.015135587193071842, "grad_norm": 0.09595327079296112, "global_step": 40628, "epoch": 383, "lr": 1.2851353588703325e-05} +{"train_loss": 0.012950755655765533, "grad_norm": 0.04796741530299187, "global_step": 40629, "epoch": 383, "lr": 1.2849369941317824e-05} +{"train_loss": 0.014581767842173576, "grad_norm": 0.10716074705123901, "global_step": 40630, "epoch": 383, "lr": 1.2847386424463465e-05} +{"train_loss": 0.012863153591752052, "grad_norm": 0.07965002208948135, "global_step": 40631, "epoch": 383, "lr": 1.2845403038147247e-05} +{"train_loss": 0.016713185235857964, "grad_norm": 0.0998578816652298, "global_step": 40632, "epoch": 383, "lr": 1.2843419782376109e-05} +{"train_loss": 0.01323956623673439, "grad_norm": 0.07664723694324493, "global_step": 40633, "epoch": 383, "lr": 1.2841436657157035e-05} +{"train_loss": 0.014077125117182732, "grad_norm": 0.06763914227485657, "global_step": 40634, "epoch": 383, "lr": 1.2839453662497009e-05} +{"train_loss": 0.019298072904348373, "grad_norm": 0.13638587296009064, "global_step": 40635, "epoch": 383, "lr": 1.2837470798402978e-05} +{"train_loss": 0.012642748653888702, "grad_norm": 0.1193457767367363, "global_step": 40636, "epoch": 383, "lr": 1.28354880648819e-05} +{"train_loss": 0.013654103502631187, "grad_norm": 0.10890081524848938, "global_step": 40637, "epoch": 383, "lr": 1.2833505461940775e-05} +{"train_loss": 0.015020417049527168, "grad_norm": 0.08168774843215942, "global_step": 40638, "epoch": 383, "lr": 1.2831522989586532e-05} +{"train_loss": 0.01461816020309925, "grad_norm": 0.0700642392039299, "global_step": 40639, "epoch": 383, "lr": 1.2829540647826149e-05} +{"train_loss": 0.01272747851908207, "grad_norm": 0.06654171645641327, "global_step": 40640, "epoch": 383, "lr": 1.2827558436666608e-05} +{"train_loss": 0.013983516953885555, "grad_norm": 0.07310016453266144, "global_step": 40641, "epoch": 383, "lr": 1.2825576356114866e-05} +{"train_loss": 0.017166689038276672, "grad_norm": 0.09980392456054688, "global_step": 40642, "epoch": 383, "lr": 1.2823594406177869e-05} +{"train_loss": 0.014723751693964005, "grad_norm": 0.061458148062229156, "global_step": 40643, "epoch": 383, "lr": 1.2821612586862585e-05} +{"train_loss": 0.018152151256799698, "grad_norm": 0.06443680822849274, "global_step": 40644, "epoch": 383, "lr": 1.2819630898176011e-05} +{"train_loss": 0.01384691707789898, "grad_norm": 0.11694376170635223, "global_step": 40645, "epoch": 383, "lr": 1.2817649340125065e-05} +{"train_loss": 0.015531562268733978, "grad_norm": 0.0911736860871315, "global_step": 40646, "epoch": 383, "lr": 1.2815667912716728e-05} +{"train_loss": 0.014966167509555817, "grad_norm": 0.07095501571893692, "global_step": 40647, "epoch": 383, "lr": 1.2813686615957976e-05} +{"train_loss": 0.015774086117744446, "grad_norm": 0.0881219357252121, "global_step": 40648, "epoch": 383, "lr": 1.281170544985576e-05} +{"train_loss": 0.01701723411679268, "grad_norm": 0.10796868056058884, "global_step": 40649, "epoch": 383, "lr": 1.2809724414417023e-05} +{"train_loss": 0.013931446708738804, "grad_norm": 0.09692816436290741, "global_step": 40650, "epoch": 383, "lr": 1.2807743509648745e-05} +{"train_loss": 0.01434292271733284, "grad_norm": 0.10971762984991074, "global_step": 40651, "epoch": 383, "lr": 1.2805762735557892e-05} +{"train_loss": 0.01632261648774147, "grad_norm": 0.09669064730405807, "global_step": 40652, "epoch": 383, "lr": 1.2803782092151401e-05} +{"train_loss": 0.015558595769107342, "grad_norm": 0.0751134380698204, "global_step": 40653, "epoch": 383, "lr": 1.2801801579436257e-05} +{"train_loss": 0.013997637666761875, "grad_norm": 0.1020016297698021, "global_step": 40654, "epoch": 383, "lr": 1.2799821197419404e-05} +{"train_loss": 0.016427254304289818, "grad_norm": 0.07397696375846863, "global_step": 40655, "epoch": 383, "lr": 1.2797840946107786e-05} +{"train_loss": 0.01375641580671072, "grad_norm": 0.06856139749288559, "global_step": 40656, "epoch": 383, "lr": 1.2795860825508376e-05} +{"train_loss": 0.015872253105044365, "grad_norm": 0.05684361606836319, "global_step": 40657, "epoch": 383, "lr": 1.2793880835628153e-05} +{"train_loss": 0.014029931277036667, "grad_norm": 0.07557922601699829, "global_step": 40658, "epoch": 383, "lr": 1.279190097647403e-05} +{"train_loss": 0.013222699984908104, "grad_norm": 0.06927452236413956, "global_step": 40659, "epoch": 383, "lr": 1.2789921248052999e-05} +{"train_loss": 0.014009177684783936, "grad_norm": 0.054104920476675034, "global_step": 40660, "epoch": 383, "lr": 1.2787941650371992e-05} +{"train_loss": 0.0151812843978405, "grad_norm": 0.10306765884160995, "global_step": 40661, "epoch": 383, "lr": 1.2785962183437994e-05} +{"train_loss": 0.018287569284439087, "grad_norm": 0.08136926591396332, "global_step": 40662, "epoch": 383, "lr": 1.278398284725792e-05} +{"train_loss": 0.015612456016242504, "grad_norm": 0.0591818206012249, "global_step": 40663, "epoch": 383, "lr": 1.2782003641838747e-05} +{"train_loss": 0.016767142340540886, "grad_norm": 0.12190289795398712, "global_step": 40664, "epoch": 383, "lr": 1.2780024567187443e-05} +{"train_loss": 0.013141188770532608, "grad_norm": 0.09099693596363068, "global_step": 40665, "epoch": 383, "lr": 1.2778045623310941e-05} +{"train_loss": 0.017167411744594574, "grad_norm": 0.09337177127599716, "global_step": 40666, "epoch": 383, "lr": 1.2776066810216193e-05} +{"train_loss": 0.014255779795348644, "grad_norm": 0.11372070014476776, "global_step": 40667, "epoch": 383, "lr": 1.277408812791015e-05} +{"train_loss": 0.015850307419896126, "grad_norm": 0.08785822242498398, "global_step": 40668, "epoch": 383, "lr": 1.2772109576399794e-05} +{"train_loss": 0.015546589158475399, "grad_norm": 0.1419181376695633, "global_step": 40669, "epoch": 383, "lr": 1.2770131155692033e-05} +{"train_loss": 0.01538837980479002, "grad_norm": 0.11599101126194, "global_step": 40670, "epoch": 383, "lr": 1.276815286579386e-05} +{"train_loss": 0.012593693099915981, "grad_norm": 0.11884836107492447, "global_step": 40671, "epoch": 383, "lr": 1.2766174706712192e-05} +{"train_loss": 0.01206132024526596, "grad_norm": 0.06677994132041931, "global_step": 40672, "epoch": 383, "lr": 1.2764196678454005e-05} +{"train_loss": 0.014582350850105286, "grad_norm": 0.11096323281526566, "global_step": 40673, "epoch": 383, "lr": 1.2762218781026225e-05} +{"train_loss": 0.014378339983522892, "grad_norm": 0.10904305428266525, "global_step": 40674, "epoch": 383, "lr": 1.2760241014435814e-05} +{"train_loss": 0.01218230091035366, "grad_norm": 0.06536320596933365, "global_step": 40675, "epoch": 383, "lr": 1.2758263378689733e-05} +{"train_loss": 0.01413267757743597, "grad_norm": 0.06622394174337387, "global_step": 40676, "epoch": 383, "lr": 1.275628587379492e-05} +{"train_loss": 0.012203603982925415, "grad_norm": 0.07085121423006058, "global_step": 40677, "epoch": 383, "lr": 1.275430849975831e-05} +{"train_loss": 0.01712615415453911, "grad_norm": 0.09119465947151184, "global_step": 40678, "epoch": 383, "lr": 1.2752331256586874e-05} +{"train_loss": 0.01560304407030344, "grad_norm": 0.0857996940612793, "global_step": 40679, "epoch": 383, "lr": 1.275035414428753e-05} +{"train_loss": 0.014975495636463165, "grad_norm": 0.09551184624433517, "global_step": 40680, "epoch": 383, "lr": 1.2748377162867248e-05} +{"train_loss": 0.014641663059592247, "grad_norm": 0.08720763027667999, "global_step": 40681, "epoch": 383, "lr": 1.274640031233298e-05} +{"train_loss": 0.014217954128980637, "grad_norm": 0.07626252621412277, "global_step": 40682, "epoch": 383, "lr": 1.2744423592691656e-05} +{"train_loss": 0.013029578141868114, "grad_norm": 0.09583449363708496, "global_step": 40683, "epoch": 383, "lr": 1.274244700395022e-05} +{"train_loss": 0.014243383891880512, "grad_norm": 0.0740419402718544, "global_step": 40684, "epoch": 383, "lr": 1.2740470546115613e-05} +{"train_loss": 0.0149165578186512, "grad_norm": 0.060371771454811096, "global_step": 40685, "epoch": 383, "lr": 1.2738494219194807e-05} +{"train_loss": 0.016157105565071106, "grad_norm": 0.11836637556552887, "global_step": 40686, "epoch": 383, "lr": 1.2736518023194715e-05} +{"train_loss": 0.013720845803618431, "grad_norm": 0.06311538070440292, "global_step": 40687, "epoch": 383, "lr": 1.2734541958122293e-05} +{"train_loss": 0.014740674756467342, "grad_norm": 0.060565147548913956, "global_step": 40688, "epoch": 383, "lr": 1.27325660239845e-05} +{"train_loss": 0.016678694635629654, "grad_norm": 0.0779334008693695, "global_step": 40689, "epoch": 383, "lr": 1.2730590220788258e-05} +{"train_loss": 0.016456363722682, "grad_norm": 0.08702126145362854, "global_step": 40690, "epoch": 383, "lr": 1.2728614548540502e-05} +{"train_loss": 0.014735784381628036, "grad_norm": 0.0719439834356308, "global_step": 40691, "epoch": 383, "lr": 1.2726639007248187e-05} +{"train_loss": 0.01652788370847702, "grad_norm": 0.08387907594442368, "global_step": 40692, "epoch": 383, "lr": 1.2724663596918262e-05} +{"train_loss": 0.015537180937826633, "grad_norm": 0.06423284858465195, "global_step": 40693, "epoch": 383, "lr": 1.2722688317557651e-05} +{"train_loss": 0.017711356282234192, "grad_norm": 0.12381619960069656, "global_step": 40694, "epoch": 383, "lr": 1.2720713169173315e-05} +{"train_loss": 0.01623038947582245, "grad_norm": 0.10514894872903824, "global_step": 40695, "epoch": 383, "lr": 1.2718738151772175e-05} +{"train_loss": 0.01611587405204773, "grad_norm": 0.05230351537466049, "global_step": 40696, "epoch": 383, "lr": 1.2716763265361164e-05} +{"train_loss": 0.01570727676153183, "grad_norm": 0.05158419907093048, "global_step": 40697, "epoch": 383, "lr": 1.2714788509947234e-05} +{"train_loss": 0.01654951088130474, "grad_norm": 0.12172737717628479, "global_step": 40698, "epoch": 383, "lr": 1.2712813885537339e-05} +{"train_loss": 0.014163753017783165, "grad_norm": 0.09705270081758499, "global_step": 40699, "epoch": 383, "lr": 1.2710839392138386e-05} +{"train_loss": 0.013786116614937782, "grad_norm": 0.0711834505200386, "global_step": 40700, "epoch": 383, "lr": 1.2708865029757339e-05} +{"train_loss": 0.01743411459028721, "grad_norm": 0.0944332405924797, "global_step": 40701, "epoch": 383, "lr": 1.2706890798401106e-05} +{"train_loss": 0.013726073317229748, "grad_norm": 0.07452266663312912, "global_step": 40702, "epoch": 383, "lr": 1.2704916698076657e-05} +{"train_loss": 0.015028679075668443, "grad_norm": 0.07310103625059128, "global_step": 40703, "epoch": 383, "lr": 1.2702942728790895e-05, "val_loss": 0.01903962343931198} +{"train_loss": 0.01540110632777214, "grad_norm": 0.08801617473363876, "global_step": 40704, "epoch": 384, "lr": 1.2700968890550774e-05} +{"train_loss": 0.01621144637465477, "grad_norm": 0.10481750965118408, "global_step": 40705, "epoch": 384, "lr": 1.2698995183363244e-05} +{"train_loss": 0.014227722771465778, "grad_norm": 0.07065680623054504, "global_step": 40706, "epoch": 384, "lr": 1.2697021607235221e-05} +{"train_loss": 0.015691101551055908, "grad_norm": 0.06666980683803558, "global_step": 40707, "epoch": 384, "lr": 1.2695048162173629e-05} +{"train_loss": 0.014293565414845943, "grad_norm": 0.09826093912124634, "global_step": 40708, "epoch": 384, "lr": 1.269307484818541e-05} +{"train_loss": 0.01609310321509838, "grad_norm": 0.07925444841384888, "global_step": 40709, "epoch": 384, "lr": 1.2691101665277516e-05} +{"train_loss": 0.013531274162232876, "grad_norm": 0.07150956243276596, "global_step": 40710, "epoch": 384, "lr": 1.2689128613456858e-05} +{"train_loss": 0.012146509252488613, "grad_norm": 0.12210090458393097, "global_step": 40711, "epoch": 384, "lr": 1.2687155692730385e-05} +{"train_loss": 0.015105048194527626, "grad_norm": 0.08916395157575607, "global_step": 40712, "epoch": 384, "lr": 1.268518290310502e-05} +{"train_loss": 0.015200743451714516, "grad_norm": 0.0906534492969513, "global_step": 40713, "epoch": 384, "lr": 1.2683210244587679e-05} +{"train_loss": 0.014494099654257298, "grad_norm": 0.10379218310117722, "global_step": 40714, "epoch": 384, "lr": 1.2681237717185313e-05} +{"train_loss": 0.014484452083706856, "grad_norm": 0.1132098138332367, "global_step": 40715, "epoch": 384, "lr": 1.2679265320904842e-05} +{"train_loss": 0.014155112206935883, "grad_norm": 0.12012673169374466, "global_step": 40716, "epoch": 384, "lr": 1.2677293055753225e-05} +{"train_loss": 0.01334250345826149, "grad_norm": 0.11475694924592972, "global_step": 40717, "epoch": 384, "lr": 1.267532092173736e-05} +{"train_loss": 0.016903724521398544, "grad_norm": 0.057172685861587524, "global_step": 40718, "epoch": 384, "lr": 1.2673348918864174e-05} +{"train_loss": 0.014450066722929478, "grad_norm": 0.1098996251821518, "global_step": 40719, "epoch": 384, "lr": 1.2671377047140625e-05} +{"train_loss": 0.017574500292539597, "grad_norm": 0.07783392071723938, "global_step": 40720, "epoch": 384, "lr": 1.2669405306573607e-05} +{"train_loss": 0.014873780310153961, "grad_norm": 0.07817800343036652, "global_step": 40721, "epoch": 384, "lr": 1.2667433697170061e-05} +{"train_loss": 0.013347831554710865, "grad_norm": 0.0754719004034996, "global_step": 40722, "epoch": 384, "lr": 1.2665462218936936e-05} +{"train_loss": 0.01534432452172041, "grad_norm": 0.094612255692482, "global_step": 40723, "epoch": 384, "lr": 1.2663490871881139e-05} +{"train_loss": 0.01700492575764656, "grad_norm": 0.06802433729171753, "global_step": 40724, "epoch": 384, "lr": 1.2661519656009585e-05} +{"train_loss": 0.014600185677409172, "grad_norm": 0.06933290511369705, "global_step": 40725, "epoch": 384, "lr": 1.2659548571329211e-05} +{"train_loss": 0.01273508183658123, "grad_norm": 0.07537676393985748, "global_step": 40726, "epoch": 384, "lr": 1.2657577617846961e-05} +{"train_loss": 0.01615862175822258, "grad_norm": 0.07433824241161346, "global_step": 40727, "epoch": 384, "lr": 1.2655606795569725e-05} +{"train_loss": 0.01304006390273571, "grad_norm": 0.06447754800319672, "global_step": 40728, "epoch": 384, "lr": 1.2653636104504445e-05} +{"train_loss": 0.017200594767928123, "grad_norm": 0.06213672459125519, "global_step": 40729, "epoch": 384, "lr": 1.2651665544658064e-05} +{"train_loss": 0.015367977321147919, "grad_norm": 0.06780723482370377, "global_step": 40730, "epoch": 384, "lr": 1.2649695116037491e-05} +{"train_loss": 0.012949083000421524, "grad_norm": 0.06972155719995499, "global_step": 40731, "epoch": 384, "lr": 1.2647724818649625e-05} +{"train_loss": 0.014599377289414406, "grad_norm": 0.05157528072595596, "global_step": 40732, "epoch": 384, "lr": 1.2645754652501418e-05} +{"train_loss": 0.012250838801264763, "grad_norm": 0.06069808825850487, "global_step": 40733, "epoch": 384, "lr": 1.2643784617599791e-05} +{"train_loss": 0.012574559077620506, "grad_norm": 0.05143284797668457, "global_step": 40734, "epoch": 384, "lr": 1.264181471395165e-05} +{"train_loss": 0.015657873824238777, "grad_norm": 0.09610781818628311, "global_step": 40735, "epoch": 384, "lr": 1.263984494156394e-05} +{"train_loss": 0.014351220801472664, "grad_norm": 0.06145360320806503, "global_step": 40736, "epoch": 384, "lr": 1.2637875300443564e-05} +{"train_loss": 0.01418411172926426, "grad_norm": 0.1637522131204605, "global_step": 40737, "epoch": 384, "lr": 1.263590579059743e-05} +{"train_loss": 0.015132756903767586, "grad_norm": 0.06566791981458664, "global_step": 40738, "epoch": 384, "lr": 1.2633936412032477e-05} +{"train_loss": 0.017632070928812027, "grad_norm": 0.11563681811094284, "global_step": 40739, "epoch": 384, "lr": 1.2631967164755632e-05} +{"train_loss": 0.013870621100068092, "grad_norm": 0.0774647518992424, "global_step": 40740, "epoch": 384, "lr": 1.2629998048773794e-05} +{"train_loss": 0.016194600611925125, "grad_norm": 0.08486075699329376, "global_step": 40741, "epoch": 384, "lr": 1.2628029064093905e-05} +{"train_loss": 0.01528910081833601, "grad_norm": 0.09390119463205338, "global_step": 40742, "epoch": 384, "lr": 1.262606021072285e-05} +{"train_loss": 0.014257311820983887, "grad_norm": 0.0735369622707367, "global_step": 40743, "epoch": 384, "lr": 1.2624091488667584e-05} +{"train_loss": 0.016977861523628235, "grad_norm": 0.08825460076332092, "global_step": 40744, "epoch": 384, "lr": 1.262212289793499e-05} +{"train_loss": 0.014197871088981628, "grad_norm": 0.09622419625520706, "global_step": 40745, "epoch": 384, "lr": 1.2620154438532e-05} +{"train_loss": 0.013463595882058144, "grad_norm": 0.08697953075170517, "global_step": 40746, "epoch": 384, "lr": 1.2618186110465552e-05} +{"train_loss": 0.016649838536977768, "grad_norm": 0.08527681231498718, "global_step": 40747, "epoch": 384, "lr": 1.2616217913742534e-05} +{"train_loss": 0.013403273187577724, "grad_norm": 0.08674580603837967, "global_step": 40748, "epoch": 384, "lr": 1.2614249848369858e-05} +{"train_loss": 0.018689977005124092, "grad_norm": 0.10114993900060654, "global_step": 40749, "epoch": 384, "lr": 1.2612281914354451e-05} +{"train_loss": 0.0158150065690279, "grad_norm": 0.056915443390607834, "global_step": 40750, "epoch": 384, "lr": 1.2610314111703236e-05} +{"train_loss": 0.012267347425222397, "grad_norm": 0.08274998515844345, "global_step": 40751, "epoch": 384, "lr": 1.2608346440423102e-05} +{"train_loss": 0.012187412939965725, "grad_norm": 0.05096200108528137, "global_step": 40752, "epoch": 384, "lr": 1.2606378900520982e-05} +{"train_loss": 0.01447291485965252, "grad_norm": 0.07217101752758026, "global_step": 40753, "epoch": 384, "lr": 1.260441149200381e-05} +{"train_loss": 0.013923530466854572, "grad_norm": 0.10373260825872421, "global_step": 40754, "epoch": 384, "lr": 1.2602444214878445e-05} +{"train_loss": 0.015014293603599072, "grad_norm": 0.08126931637525558, "global_step": 40755, "epoch": 384, "lr": 1.2600477069151829e-05} +{"train_loss": 0.014346875250339508, "grad_norm": 0.08060406148433685, "global_step": 40756, "epoch": 384, "lr": 1.2598510054830886e-05} +{"train_loss": 0.015187522396445274, "grad_norm": 0.06891584396362305, "global_step": 40757, "epoch": 384, "lr": 1.2596543171922492e-05} +{"train_loss": 0.018500447273254395, "grad_norm": 0.12919990718364716, "global_step": 40758, "epoch": 384, "lr": 1.259457642043359e-05} +{"train_loss": 0.012402634136378765, "grad_norm": 0.0862211361527443, "global_step": 40759, "epoch": 384, "lr": 1.2592609800371086e-05} +{"train_loss": 0.01770034246146679, "grad_norm": 0.07487472891807556, "global_step": 40760, "epoch": 384, "lr": 1.2590643311741884e-05} +{"train_loss": 0.012839719653129578, "grad_norm": 0.05670525133609772, "global_step": 40761, "epoch": 384, "lr": 1.258867695455288e-05} +{"train_loss": 0.015378819778561592, "grad_norm": 0.14849971234798431, "global_step": 40762, "epoch": 384, "lr": 1.2586710728810997e-05} +{"train_loss": 0.01614702120423317, "grad_norm": 0.06672532856464386, "global_step": 40763, "epoch": 384, "lr": 1.2584744634523155e-05} +{"train_loss": 0.013879469595849514, "grad_norm": 0.07837670296430588, "global_step": 40764, "epoch": 384, "lr": 1.2582778671696244e-05} +{"train_loss": 0.015022587962448597, "grad_norm": 0.1139829233288765, "global_step": 40765, "epoch": 384, "lr": 1.258081284033717e-05} +{"train_loss": 0.015556620433926582, "grad_norm": 0.05957062169909477, "global_step": 40766, "epoch": 384, "lr": 1.2578847140452837e-05} +{"train_loss": 0.01622956246137619, "grad_norm": 0.07131025940179825, "global_step": 40767, "epoch": 384, "lr": 1.2576881572050187e-05} +{"train_loss": 0.015610345639288425, "grad_norm": 0.0876060202717781, "global_step": 40768, "epoch": 384, "lr": 1.2574916135136078e-05} +{"train_loss": 0.013107238337397575, "grad_norm": 0.07875018566846848, "global_step": 40769, "epoch": 384, "lr": 1.2572950829717439e-05} +{"train_loss": 0.01382522378116846, "grad_norm": 0.06630893796682358, "global_step": 40770, "epoch": 384, "lr": 1.2570985655801188e-05} +{"train_loss": 0.015254604630172253, "grad_norm": 0.06314527243375778, "global_step": 40771, "epoch": 384, "lr": 1.2569020613394217e-05} +{"train_loss": 0.013678440824151039, "grad_norm": 0.07553243637084961, "global_step": 40772, "epoch": 384, "lr": 1.2567055702503416e-05} +{"train_loss": 0.015936724841594696, "grad_norm": 0.2128765881061554, "global_step": 40773, "epoch": 384, "lr": 1.2565090923135704e-05} +{"train_loss": 0.014059997163712978, "grad_norm": 0.09419813007116318, "global_step": 40774, "epoch": 384, "lr": 1.2563126275297993e-05} +{"train_loss": 0.01642359420657158, "grad_norm": 0.0633058026432991, "global_step": 40775, "epoch": 384, "lr": 1.2561161758997165e-05} +{"train_loss": 0.014332584105432034, "grad_norm": 0.05424968898296356, "global_step": 40776, "epoch": 384, "lr": 1.2559197374240139e-05} +{"train_loss": 0.013874556869268417, "grad_norm": 0.08979985117912292, "global_step": 40777, "epoch": 384, "lr": 1.2557233121033818e-05} +{"train_loss": 0.014519922435283661, "grad_norm": 0.11181825399398804, "global_step": 40778, "epoch": 384, "lr": 1.2555268999385083e-05} +{"train_loss": 0.018316933885216713, "grad_norm": 0.06716500222682953, "global_step": 40779, "epoch": 384, "lr": 1.2553305009300841e-05} +{"train_loss": 0.01610996387898922, "grad_norm": 0.09337896853685379, "global_step": 40780, "epoch": 384, "lr": 1.2551341150788021e-05} +{"train_loss": 0.013092972338199615, "grad_norm": 0.11994791030883789, "global_step": 40781, "epoch": 384, "lr": 1.2549377423853487e-05} +{"train_loss": 0.016301341354846954, "grad_norm": 0.05604179948568344, "global_step": 40782, "epoch": 384, "lr": 1.2547413828504168e-05} +{"train_loss": 0.015835359692573547, "grad_norm": 0.08232785761356354, "global_step": 40783, "epoch": 384, "lr": 1.2545450364746936e-05} +{"train_loss": 0.013137304224073887, "grad_norm": 0.08001434803009033, "global_step": 40784, "epoch": 384, "lr": 1.2543487032588714e-05} +{"train_loss": 0.014310724101960659, "grad_norm": 0.08293873071670532, "global_step": 40785, "epoch": 384, "lr": 1.2541523832036378e-05} +{"train_loss": 0.016071340069174767, "grad_norm": 0.06750716269016266, "global_step": 40786, "epoch": 384, "lr": 1.253956076309683e-05} +{"train_loss": 0.011632265523076057, "grad_norm": 0.09386686235666275, "global_step": 40787, "epoch": 384, "lr": 1.2537597825776998e-05} +{"train_loss": 0.01811271160840988, "grad_norm": 0.09422311931848526, "global_step": 40788, "epoch": 384, "lr": 1.2535635020083752e-05} +{"train_loss": 0.014945532195270061, "grad_norm": 0.08143321424722672, "global_step": 40789, "epoch": 384, "lr": 1.2533672346023972e-05} +{"train_loss": 0.013594319112598896, "grad_norm": 0.08684708923101425, "global_step": 40790, "epoch": 384, "lr": 1.253170980360458e-05} +{"train_loss": 0.01585402525961399, "grad_norm": 0.07099690288305283, "global_step": 40791, "epoch": 384, "lr": 1.252974739283248e-05} +{"train_loss": 0.017222438007593155, "grad_norm": 0.06735021620988846, "global_step": 40792, "epoch": 384, "lr": 1.2527785113714535e-05} +{"train_loss": 0.01728319376707077, "grad_norm": 0.07656222581863403, "global_step": 40793, "epoch": 384, "lr": 1.2525822966257656e-05} +{"train_loss": 0.015250831842422485, "grad_norm": 0.1965564489364624, "global_step": 40794, "epoch": 384, "lr": 1.2523860950468768e-05} +{"train_loss": 0.01858489029109478, "grad_norm": 0.0974983498454094, "global_step": 40795, "epoch": 384, "lr": 1.2521899066354708e-05} +{"train_loss": 0.0150966327637434, "grad_norm": 0.07901514321565628, "global_step": 40796, "epoch": 384, "lr": 1.2519937313922392e-05} +{"train_loss": 0.016156213358044624, "grad_norm": 0.09552421420812607, "global_step": 40797, "epoch": 384, "lr": 1.2517975693178736e-05} +{"train_loss": 0.015715934336185455, "grad_norm": 0.09195981174707413, "global_step": 40798, "epoch": 384, "lr": 1.2516014204130593e-05} +{"train_loss": 0.015811365097761154, "grad_norm": 0.0833657756447792, "global_step": 40799, "epoch": 384, "lr": 1.251405284678488e-05} +{"train_loss": 0.013645239174365997, "grad_norm": 0.12913431227207184, "global_step": 40800, "epoch": 384, "lr": 1.2512091621148492e-05} +{"train_loss": 0.01547993067651987, "grad_norm": 0.0619988851249218, "global_step": 40801, "epoch": 384, "lr": 1.2510130527228314e-05} +{"train_loss": 0.015138094313442707, "grad_norm": 0.08585137128829956, "global_step": 40802, "epoch": 384, "lr": 1.2508169565031219e-05} +{"train_loss": 0.015125472098588943, "grad_norm": 0.09908391535282135, "global_step": 40803, "epoch": 384, "lr": 1.2506208734564112e-05} +{"train_loss": 0.012697872705757618, "grad_norm": 0.07663760334253311, "global_step": 40804, "epoch": 384, "lr": 1.2504248035833892e-05} +{"train_loss": 0.01487877406179905, "grad_norm": 0.09045298397541046, "global_step": 40805, "epoch": 384, "lr": 1.2502287468847429e-05} +{"train_loss": 0.0184248685836792, "grad_norm": 0.10952653735876083, "global_step": 40806, "epoch": 384, "lr": 1.2500327033611636e-05} +{"train_loss": 0.01601605862379074, "grad_norm": 0.11643464863300323, "global_step": 40807, "epoch": 384, "lr": 1.2498366730133365e-05} +{"train_loss": 0.015545256435871124, "grad_norm": 0.05994902551174164, "global_step": 40808, "epoch": 384, "lr": 1.2496406558419543e-05} +{"train_loss": 0.015017153589792972, "grad_norm": 0.07505207508802414, "global_step": 40809, "epoch": 384, "lr": 1.2494446518477022e-05, "val_loss": 0.019285785034298897} +{"train_loss": 0.0160362608730793, "grad_norm": 0.06382859498262405, "global_step": 40810, "epoch": 385, "lr": 1.249248661031271e-05} +{"train_loss": 0.01321287453174591, "grad_norm": 0.07923323661088943, "global_step": 40811, "epoch": 385, "lr": 1.2490526833933498e-05} +{"train_loss": 0.015503453090786934, "grad_norm": 0.07323645055294037, "global_step": 40812, "epoch": 385, "lr": 1.2488567189346262e-05} +{"train_loss": 0.011978157795965672, "grad_norm": 0.07544118911027908, "global_step": 40813, "epoch": 385, "lr": 1.2486607676557876e-05} +{"train_loss": 0.01326785609126091, "grad_norm": 0.1357850432395935, "global_step": 40814, "epoch": 385, "lr": 1.2484648295575246e-05} +{"train_loss": 0.015131285414099693, "grad_norm": 0.06458018720149994, "global_step": 40815, "epoch": 385, "lr": 1.2482689046405239e-05} +{"train_loss": 0.01758209615945816, "grad_norm": 0.08711866289377213, "global_step": 40816, "epoch": 385, "lr": 1.248072992905474e-05} +{"train_loss": 0.01373017393052578, "grad_norm": 0.05709099769592285, "global_step": 40817, "epoch": 385, "lr": 1.247877094353066e-05} +{"train_loss": 0.014883220195770264, "grad_norm": 0.09910249710083008, "global_step": 40818, "epoch": 385, "lr": 1.2476812089839857e-05} +{"train_loss": 0.01350843720138073, "grad_norm": 0.0568152479827404, "global_step": 40819, "epoch": 385, "lr": 1.2474853367989208e-05} +{"train_loss": 0.01563350111246109, "grad_norm": 0.10032496601343155, "global_step": 40820, "epoch": 385, "lr": 1.24728947779856e-05} +{"train_loss": 0.014946065843105316, "grad_norm": 0.08448123186826706, "global_step": 40821, "epoch": 385, "lr": 1.247093631983594e-05} +{"train_loss": 0.019075244665145874, "grad_norm": 0.18767152726650238, "global_step": 40822, "epoch": 385, "lr": 1.2468977993547071e-05} +{"train_loss": 0.016196012496948242, "grad_norm": 0.09605973958969116, "global_step": 40823, "epoch": 385, "lr": 1.2467019799125907e-05} +{"train_loss": 0.016528137028217316, "grad_norm": 0.08547528088092804, "global_step": 40824, "epoch": 385, "lr": 1.2465061736579292e-05} +{"train_loss": 0.01808660477399826, "grad_norm": 0.09016787260770798, "global_step": 40825, "epoch": 385, "lr": 1.246310380591415e-05} +{"train_loss": 0.014576805755496025, "grad_norm": 0.17318250238895416, "global_step": 40826, "epoch": 385, "lr": 1.2461146007137314e-05} +{"train_loss": 0.015082691796123981, "grad_norm": 0.08929228782653809, "global_step": 40827, "epoch": 385, "lr": 1.245918834025569e-05} +{"train_loss": 0.014359691180288792, "grad_norm": 0.0814412385225296, "global_step": 40828, "epoch": 385, "lr": 1.2457230805276171e-05} +{"train_loss": 0.01745031028985977, "grad_norm": 0.07272597402334213, "global_step": 40829, "epoch": 385, "lr": 1.2455273402205608e-05} +{"train_loss": 0.015510751865804195, "grad_norm": 0.08809777349233627, "global_step": 40830, "epoch": 385, "lr": 1.245331613105088e-05} +{"train_loss": 0.013758514076471329, "grad_norm": 0.0552990697324276, "global_step": 40831, "epoch": 385, "lr": 1.2451358991818861e-05} +{"train_loss": 0.01636935956776142, "grad_norm": 0.07184350490570068, "global_step": 40832, "epoch": 385, "lr": 1.2449401984516462e-05} +{"train_loss": 0.014794765040278435, "grad_norm": 0.08841904997825623, "global_step": 40833, "epoch": 385, "lr": 1.2447445109150512e-05} +{"train_loss": 0.013575796037912369, "grad_norm": 0.05650751665234566, "global_step": 40834, "epoch": 385, "lr": 1.2445488365727908e-05} +{"train_loss": 0.014851163141429424, "grad_norm": 0.07932546734809875, "global_step": 40835, "epoch": 385, "lr": 1.244353175425556e-05} +{"train_loss": 0.0135744484141469, "grad_norm": 0.10215124487876892, "global_step": 40836, "epoch": 385, "lr": 1.2441575274740281e-05} +{"train_loss": 0.016545774415135384, "grad_norm": 0.1259123682975769, "global_step": 40837, "epoch": 385, "lr": 1.2439618927188967e-05} +{"train_loss": 0.016316892579197884, "grad_norm": 0.06301110237836838, "global_step": 40838, "epoch": 385, "lr": 1.2437662711608517e-05} +{"train_loss": 0.012060804292559624, "grad_norm": 0.06993117928504944, "global_step": 40839, "epoch": 385, "lr": 1.2435706628005767e-05} +{"train_loss": 0.014438329264521599, "grad_norm": 0.05644123628735542, "global_step": 40840, "epoch": 385, "lr": 1.2433750676387607e-05} +{"train_loss": 0.015603609383106232, "grad_norm": 0.07537487894296646, "global_step": 40841, "epoch": 385, "lr": 1.2431794856760926e-05} +{"train_loss": 0.01609305664896965, "grad_norm": 0.08346204459667206, "global_step": 40842, "epoch": 385, "lr": 1.2429839169132579e-05} +{"train_loss": 0.014195853844285011, "grad_norm": 0.08860635757446289, "global_step": 40843, "epoch": 385, "lr": 1.2427883613509422e-05} +{"train_loss": 0.013901281170547009, "grad_norm": 0.059790581464767456, "global_step": 40844, "epoch": 385, "lr": 1.2425928189898345e-05} +{"train_loss": 0.016235116869211197, "grad_norm": 0.07378561049699783, "global_step": 40845, "epoch": 385, "lr": 1.242397289830623e-05} +{"train_loss": 0.015095086768269539, "grad_norm": 0.07256089150905609, "global_step": 40846, "epoch": 385, "lr": 1.2422017738739921e-05} +{"train_loss": 0.014954869635403156, "grad_norm": 0.08371816575527191, "global_step": 40847, "epoch": 385, "lr": 1.2420062711206309e-05} +{"train_loss": 0.01379193551838398, "grad_norm": 0.06738809496164322, "global_step": 40848, "epoch": 385, "lr": 1.2418107815712244e-05} +{"train_loss": 0.014884954318404198, "grad_norm": 0.07490251213312149, "global_step": 40849, "epoch": 385, "lr": 1.2416153052264618e-05} +{"train_loss": 0.012815296649932861, "grad_norm": 0.0790763720870018, "global_step": 40850, "epoch": 385, "lr": 1.2414198420870265e-05} +{"train_loss": 0.01452687382698059, "grad_norm": 0.15865151584148407, "global_step": 40851, "epoch": 385, "lr": 1.2412243921536077e-05} +{"train_loss": 0.014069771394133568, "grad_norm": 0.06389355659484863, "global_step": 40852, "epoch": 385, "lr": 1.2410289554268933e-05} +{"train_loss": 0.015381156466901302, "grad_norm": 0.07090446352958679, "global_step": 40853, "epoch": 385, "lr": 1.2408335319075682e-05} +{"train_loss": 0.01570873335003853, "grad_norm": 0.0778149962425232, "global_step": 40854, "epoch": 385, "lr": 1.2406381215963181e-05} +{"train_loss": 0.017788458615541458, "grad_norm": 0.08902323991060257, "global_step": 40855, "epoch": 385, "lr": 1.2404427244938321e-05} +{"train_loss": 0.015595860779285431, "grad_norm": 0.08299718797206879, "global_step": 40856, "epoch": 385, "lr": 1.2402473406007936e-05} +{"train_loss": 0.017791641876101494, "grad_norm": 0.08529619872570038, "global_step": 40857, "epoch": 385, "lr": 1.2400519699178914e-05} +{"train_loss": 0.014316997490823269, "grad_norm": 0.06282493472099304, "global_step": 40858, "epoch": 385, "lr": 1.2398566124458123e-05} +{"train_loss": 0.016179371625185013, "grad_norm": 0.09123148024082184, "global_step": 40859, "epoch": 385, "lr": 1.239661268185242e-05} +{"train_loss": 0.014700992964208126, "grad_norm": 0.08072518557310104, "global_step": 40860, "epoch": 385, "lr": 1.2394659371368656e-05} +{"train_loss": 0.015072699636220932, "grad_norm": 0.12659022212028503, "global_step": 40861, "epoch": 385, "lr": 1.2392706193013698e-05} +{"train_loss": 0.013805015943944454, "grad_norm": 0.07745259255170822, "global_step": 40862, "epoch": 385, "lr": 1.2390753146794437e-05} +{"train_loss": 0.015880992636084557, "grad_norm": 0.054509297013282776, "global_step": 40863, "epoch": 385, "lr": 1.2388800232717695e-05} +{"train_loss": 0.013927467167377472, "grad_norm": 0.09883401542901993, "global_step": 40864, "epoch": 385, "lr": 1.2386847450790367e-05} +{"train_loss": 0.01598176918923855, "grad_norm": 0.10195537656545639, "global_step": 40865, "epoch": 385, "lr": 1.2384894801019286e-05} +{"train_loss": 0.015162808820605278, "grad_norm": 0.07346328347921371, "global_step": 40866, "epoch": 385, "lr": 1.2382942283411342e-05} +{"train_loss": 0.014372220262885094, "grad_norm": 0.07845305651426315, "global_step": 40867, "epoch": 385, "lr": 1.2380989897973367e-05} +{"train_loss": 0.014184532687067986, "grad_norm": 0.08610299974679947, "global_step": 40868, "epoch": 385, "lr": 1.2379037644712227e-05} +{"train_loss": 0.015590507537126541, "grad_norm": 0.10229317098855972, "global_step": 40869, "epoch": 385, "lr": 1.2377085523634807e-05} +{"train_loss": 0.014500493183732033, "grad_norm": 0.09465836733579636, "global_step": 40870, "epoch": 385, "lr": 1.2375133534747946e-05} +{"train_loss": 0.015980523079633713, "grad_norm": 0.06202540546655655, "global_step": 40871, "epoch": 385, "lr": 1.2373181678058487e-05} +{"train_loss": 0.014305916614830494, "grad_norm": 0.07565155625343323, "global_step": 40872, "epoch": 385, "lr": 1.2371229953573304e-05} +{"train_loss": 0.01405092142522335, "grad_norm": 0.10312800109386444, "global_step": 40873, "epoch": 385, "lr": 1.236927836129927e-05} +{"train_loss": 0.016899673268198967, "grad_norm": 0.0835205465555191, "global_step": 40874, "epoch": 385, "lr": 1.2367326901243214e-05} +{"train_loss": 0.015596595592796803, "grad_norm": 0.08252647519111633, "global_step": 40875, "epoch": 385, "lr": 1.2365375573412003e-05} +{"train_loss": 0.013602805323898792, "grad_norm": 0.10767225921154022, "global_step": 40876, "epoch": 385, "lr": 1.2363424377812522e-05} +{"train_loss": 0.012654963880777359, "grad_norm": 0.05974750220775604, "global_step": 40877, "epoch": 385, "lr": 1.2361473314451578e-05} +{"train_loss": 0.01623750478029251, "grad_norm": 0.10103007405996323, "global_step": 40878, "epoch": 385, "lr": 1.2359522383336042e-05} +{"train_loss": 0.014265074394643307, "grad_norm": 0.06799643486738205, "global_step": 40879, "epoch": 385, "lr": 1.2357571584472794e-05} +{"train_loss": 0.014360976405441761, "grad_norm": 0.06607038527727127, "global_step": 40880, "epoch": 385, "lr": 1.235562091786865e-05} +{"train_loss": 0.014075309038162231, "grad_norm": 0.06103626638650894, "global_step": 40881, "epoch": 385, "lr": 1.2353670383530486e-05} +{"train_loss": 0.014241477474570274, "grad_norm": 0.1060100570321083, "global_step": 40882, "epoch": 385, "lr": 1.2351719981465164e-05} +{"train_loss": 0.012924741953611374, "grad_norm": 0.06736528128385544, "global_step": 40883, "epoch": 385, "lr": 1.2349769711679527e-05} +{"train_loss": 0.013496983796358109, "grad_norm": 0.06385021656751633, "global_step": 40884, "epoch": 385, "lr": 1.2347819574180414e-05} +{"train_loss": 0.0147818299010396, "grad_norm": 0.08695018291473389, "global_step": 40885, "epoch": 385, "lr": 1.2345869568974682e-05} +{"train_loss": 0.016238965094089508, "grad_norm": 0.15107177197933197, "global_step": 40886, "epoch": 385, "lr": 1.2343919696069212e-05} +{"train_loss": 0.015847153961658478, "grad_norm": 0.0868186429142952, "global_step": 40887, "epoch": 385, "lr": 1.2341969955470812e-05} +{"train_loss": 0.014451994560658932, "grad_norm": 0.10963813960552216, "global_step": 40888, "epoch": 385, "lr": 1.2340020347186366e-05} +{"train_loss": 0.013885987922549248, "grad_norm": 0.12455430626869202, "global_step": 40889, "epoch": 385, "lr": 1.2338070871222696e-05} +{"train_loss": 0.014545378275215626, "grad_norm": 0.08464846760034561, "global_step": 40890, "epoch": 385, "lr": 1.2336121527586687e-05} +{"train_loss": 0.016172785311937332, "grad_norm": 0.0657731145620346, "global_step": 40891, "epoch": 385, "lr": 1.2334172316285148e-05} +{"train_loss": 0.01607893407344818, "grad_norm": 0.05762966349720955, "global_step": 40892, "epoch": 385, "lr": 1.2332223237324947e-05} +{"train_loss": 0.013813034631311893, "grad_norm": 0.08892887085676193, "global_step": 40893, "epoch": 385, "lr": 1.2330274290712946e-05} +{"train_loss": 0.01568334735929966, "grad_norm": 0.06684048473834991, "global_step": 40894, "epoch": 385, "lr": 1.2328325476455982e-05} +{"train_loss": 0.016449688002467155, "grad_norm": 0.05654546990990639, "global_step": 40895, "epoch": 385, "lr": 1.232637679456088e-05} +{"train_loss": 0.014323724433779716, "grad_norm": 0.06958770751953125, "global_step": 40896, "epoch": 385, "lr": 1.2324428245034524e-05} +{"train_loss": 0.015361793339252472, "grad_norm": 0.1299847513437271, "global_step": 40897, "epoch": 385, "lr": 1.2322479827883726e-05} +{"train_loss": 0.015705231577157974, "grad_norm": 0.14919428527355194, "global_step": 40898, "epoch": 385, "lr": 1.2320531543115354e-05} +{"train_loss": 0.01621568202972412, "grad_norm": 0.07250459492206573, "global_step": 40899, "epoch": 385, "lr": 1.2318583390736254e-05} +{"train_loss": 0.0163880605250597, "grad_norm": 0.15828052163124084, "global_step": 40900, "epoch": 385, "lr": 1.231663537075327e-05} +{"train_loss": 0.01702611893415451, "grad_norm": 0.08964213728904724, "global_step": 40901, "epoch": 385, "lr": 1.2314687483173226e-05} +{"train_loss": 0.012773459777235985, "grad_norm": 0.08524607121944427, "global_step": 40902, "epoch": 385, "lr": 1.2312739728002982e-05} +{"train_loss": 0.013361244462430477, "grad_norm": 0.1004524901509285, "global_step": 40903, "epoch": 385, "lr": 1.2310792105249397e-05} +{"train_loss": 0.01444951444864273, "grad_norm": 0.08288928866386414, "global_step": 40904, "epoch": 385, "lr": 1.2308844614919279e-05} +{"train_loss": 0.013746448792517185, "grad_norm": 0.11349523067474365, "global_step": 40905, "epoch": 385, "lr": 1.2306897257019496e-05} +{"train_loss": 0.01328255794942379, "grad_norm": 0.09643906354904175, "global_step": 40906, "epoch": 385, "lr": 1.2304950031556888e-05} +{"train_loss": 0.016766414046287537, "grad_norm": 0.08148637413978577, "global_step": 40907, "epoch": 385, "lr": 1.2303002938538299e-05} +{"train_loss": 0.015630289912223816, "grad_norm": 0.07088650017976761, "global_step": 40908, "epoch": 385, "lr": 1.230105597797055e-05} +{"train_loss": 0.0168321430683136, "grad_norm": 0.18561597168445587, "global_step": 40909, "epoch": 385, "lr": 1.2299109149860495e-05} +{"train_loss": 0.01493983156979084, "grad_norm": 0.051349252462387085, "global_step": 40910, "epoch": 385, "lr": 1.2297162454214988e-05} +{"train_loss": 0.016113560646772385, "grad_norm": 0.08349382877349854, "global_step": 40911, "epoch": 385, "lr": 1.229521589104084e-05} +{"train_loss": 0.013998492620885372, "grad_norm": 0.05283831059932709, "global_step": 40912, "epoch": 385, "lr": 1.2293269460344919e-05} +{"train_loss": 0.014295345172286034, "grad_norm": 0.07014741003513336, "global_step": 40913, "epoch": 385, "lr": 1.2291323162134056e-05} +{"train_loss": 0.014859646558761597, "grad_norm": 0.06187313795089722, "global_step": 40914, "epoch": 385, "lr": 1.228937699641507e-05} +{"train_loss": 0.015014598638100445, "grad_norm": 0.09170965105295181, "global_step": 40915, "epoch": 385, "lr": 1.2287430963194807e-05, "val_loss": 0.01950223185122013} +{"train_loss": 0.014834655448794365, "grad_norm": 0.10693197697401047, "global_step": 40916, "epoch": 386, "lr": 1.2285485062480129e-05} +{"train_loss": 0.01452797930687666, "grad_norm": 0.06437645852565765, "global_step": 40917, "epoch": 386, "lr": 1.2283539294277857e-05} +{"train_loss": 0.01347985491156578, "grad_norm": 0.06803571432828903, "global_step": 40918, "epoch": 386, "lr": 1.2281593658594808e-05} +{"train_loss": 0.015516954474151134, "grad_norm": 0.13641145825386047, "global_step": 40919, "epoch": 386, "lr": 1.227964815543784e-05} +{"train_loss": 0.014659246429800987, "grad_norm": 0.12507440149784088, "global_step": 40920, "epoch": 386, "lr": 1.2277702784813793e-05} +{"train_loss": 0.01617385447025299, "grad_norm": 0.08126582950353622, "global_step": 40921, "epoch": 386, "lr": 1.2275757546729482e-05} +{"train_loss": 0.015511109493672848, "grad_norm": 0.08922215551137924, "global_step": 40922, "epoch": 386, "lr": 1.2273812441191751e-05} +{"train_loss": 0.01301388256251812, "grad_norm": 0.08399608731269836, "global_step": 40923, "epoch": 386, "lr": 1.2271867468207455e-05} +{"train_loss": 0.014370309188961983, "grad_norm": 0.09401260316371918, "global_step": 40924, "epoch": 386, "lr": 1.2269922627783403e-05} +{"train_loss": 0.0167263001203537, "grad_norm": 0.06969192624092102, "global_step": 40925, "epoch": 386, "lr": 1.2267977919926421e-05} +{"train_loss": 0.014834282919764519, "grad_norm": 0.06389081478118896, "global_step": 40926, "epoch": 386, "lr": 1.2266033344643364e-05} +{"train_loss": 0.017266895622015, "grad_norm": 0.05367984622716904, "global_step": 40927, "epoch": 386, "lr": 1.2264088901941062e-05} +{"train_loss": 0.01614682376384735, "grad_norm": 0.057812873274087906, "global_step": 40928, "epoch": 386, "lr": 1.2262144591826324e-05} +{"train_loss": 0.014315498992800713, "grad_norm": 0.07417470216751099, "global_step": 40929, "epoch": 386, "lr": 1.2260200414306017e-05} +{"train_loss": 0.012318797409534454, "grad_norm": 0.059714704751968384, "global_step": 40930, "epoch": 386, "lr": 1.2258256369386939e-05} +{"train_loss": 0.016417497768998146, "grad_norm": 0.07390483468770981, "global_step": 40931, "epoch": 386, "lr": 1.2256312457075952e-05} +{"train_loss": 0.013685936108231544, "grad_norm": 0.06249300017952919, "global_step": 40932, "epoch": 386, "lr": 1.225436867737985e-05} +{"train_loss": 0.014896105974912643, "grad_norm": 0.11568894237279892, "global_step": 40933, "epoch": 386, "lr": 1.2252425030305482e-05} +{"train_loss": 0.013622627593576908, "grad_norm": 0.08161766827106476, "global_step": 40934, "epoch": 386, "lr": 1.225048151585969e-05} +{"train_loss": 0.013520842418074608, "grad_norm": 0.08939714729785919, "global_step": 40935, "epoch": 386, "lr": 1.2248538134049292e-05} +{"train_loss": 0.01697530969977379, "grad_norm": 0.1007230281829834, "global_step": 40936, "epoch": 386, "lr": 1.2246594884881096e-05} +{"train_loss": 0.01574701815843582, "grad_norm": 0.1195983961224556, "global_step": 40937, "epoch": 386, "lr": 1.2244651768361964e-05} +{"train_loss": 0.013780949637293816, "grad_norm": 0.0550851970911026, "global_step": 40938, "epoch": 386, "lr": 1.2242708784498691e-05} +{"train_loss": 0.01601332053542137, "grad_norm": 0.07580360770225525, "global_step": 40939, "epoch": 386, "lr": 1.2240765933298116e-05} +{"train_loss": 0.017166316509246826, "grad_norm": 0.0891949012875557, "global_step": 40940, "epoch": 386, "lr": 1.2238823214767086e-05} +{"train_loss": 0.016308512538671494, "grad_norm": 0.07517524808645248, "global_step": 40941, "epoch": 386, "lr": 1.2236880628912412e-05} +{"train_loss": 0.015881160274147987, "grad_norm": 0.10633432865142822, "global_step": 40942, "epoch": 386, "lr": 1.2234938175740896e-05} +{"train_loss": 0.013713685795664787, "grad_norm": 0.06751453876495361, "global_step": 40943, "epoch": 386, "lr": 1.2232995855259388e-05} +{"train_loss": 0.01406827662140131, "grad_norm": 0.07965504378080368, "global_step": 40944, "epoch": 386, "lr": 1.2231053667474724e-05} +{"train_loss": 0.015693124383687973, "grad_norm": 0.07983275502920151, "global_step": 40945, "epoch": 386, "lr": 1.222911161239369e-05} +{"train_loss": 0.015460205264389515, "grad_norm": 0.06258754432201385, "global_step": 40946, "epoch": 386, "lr": 1.2227169690023138e-05} +{"train_loss": 0.014473994262516499, "grad_norm": 0.06444361060857773, "global_step": 40947, "epoch": 386, "lr": 1.2225227900369895e-05} +{"train_loss": 0.014489223249256611, "grad_norm": 0.07593537867069244, "global_step": 40948, "epoch": 386, "lr": 1.222328624344078e-05} +{"train_loss": 0.014979630708694458, "grad_norm": 0.06392702460289001, "global_step": 40949, "epoch": 386, "lr": 1.2221344719242589e-05} +{"train_loss": 0.016856811940670013, "grad_norm": 0.10150010883808136, "global_step": 40950, "epoch": 386, "lr": 1.2219403327782164e-05} +{"train_loss": 0.013914404436945915, "grad_norm": 0.08537083119153976, "global_step": 40951, "epoch": 386, "lr": 1.2217462069066344e-05} +{"train_loss": 0.01425166241824627, "grad_norm": 0.0770316794514656, "global_step": 40952, "epoch": 386, "lr": 1.221552094310191e-05} +{"train_loss": 0.015798183158040047, "grad_norm": 0.06358964741230011, "global_step": 40953, "epoch": 386, "lr": 1.221357994989572e-05} +{"train_loss": 0.015204336494207382, "grad_norm": 0.10905889421701431, "global_step": 40954, "epoch": 386, "lr": 1.2211639089454579e-05} +{"train_loss": 0.015081437304615974, "grad_norm": 0.07834074646234512, "global_step": 40955, "epoch": 386, "lr": 1.2209698361785287e-05} +{"train_loss": 0.014174437150359154, "grad_norm": 0.09256191551685333, "global_step": 40956, "epoch": 386, "lr": 1.2207757766894683e-05} +{"train_loss": 0.013151858001947403, "grad_norm": 0.11984842270612717, "global_step": 40957, "epoch": 386, "lr": 1.2205817304789596e-05} +{"train_loss": 0.014242986217141151, "grad_norm": 0.07512374967336655, "global_step": 40958, "epoch": 386, "lr": 1.220387697547682e-05} +{"train_loss": 0.014843798242509365, "grad_norm": 0.058880798518657684, "global_step": 40959, "epoch": 386, "lr": 1.2201936778963192e-05} +{"train_loss": 0.01422420796006918, "grad_norm": 0.0616530105471611, "global_step": 40960, "epoch": 386, "lr": 1.2199996715255508e-05} +{"train_loss": 0.012872874736785889, "grad_norm": 0.08048568665981293, "global_step": 40961, "epoch": 386, "lr": 1.2198056784360613e-05} +{"train_loss": 0.014339825138449669, "grad_norm": 0.05601237341761589, "global_step": 40962, "epoch": 386, "lr": 1.2196116986285288e-05} +{"train_loss": 0.015566755086183548, "grad_norm": 0.07835692167282104, "global_step": 40963, "epoch": 386, "lr": 1.2194177321036371e-05} +{"train_loss": 0.013962982222437859, "grad_norm": 0.05399169400334358, "global_step": 40964, "epoch": 386, "lr": 1.2192237788620692e-05} +{"train_loss": 0.013122593984007835, "grad_norm": 0.08296995609998703, "global_step": 40965, "epoch": 386, "lr": 1.219029838904504e-05} +{"train_loss": 0.014717288315296173, "grad_norm": 0.058156587183475494, "global_step": 40966, "epoch": 386, "lr": 1.2188359122316223e-05} +{"train_loss": 0.014116082340478897, "grad_norm": 0.11374877393245697, "global_step": 40967, "epoch": 386, "lr": 1.2186419988441072e-05} +{"train_loss": 0.015547000803053379, "grad_norm": 0.07474798709154129, "global_step": 40968, "epoch": 386, "lr": 1.218448098742641e-05} +{"train_loss": 0.013601402752101421, "grad_norm": 0.06681874394416809, "global_step": 40969, "epoch": 386, "lr": 1.2182542119279018e-05} +{"train_loss": 0.014255608431994915, "grad_norm": 0.10067751258611679, "global_step": 40970, "epoch": 386, "lr": 1.2180603384005746e-05} +{"train_loss": 0.01714412309229374, "grad_norm": 0.12584346532821655, "global_step": 40971, "epoch": 386, "lr": 1.217866478161338e-05} +{"train_loss": 0.01482471078634262, "grad_norm": 0.054715290665626526, "global_step": 40972, "epoch": 386, "lr": 1.2176726312108733e-05} +{"train_loss": 0.015118520706892014, "grad_norm": 0.07767403870820999, "global_step": 40973, "epoch": 386, "lr": 1.2174787975498614e-05} +{"train_loss": 0.013723740354180336, "grad_norm": 0.09341555833816528, "global_step": 40974, "epoch": 386, "lr": 1.2172849771789857e-05} +{"train_loss": 0.013591073453426361, "grad_norm": 0.11119557172060013, "global_step": 40975, "epoch": 386, "lr": 1.2170911700989235e-05} +{"train_loss": 0.01489312294870615, "grad_norm": 0.06637803465127945, "global_step": 40976, "epoch": 386, "lr": 1.2168973763103598e-05} +{"train_loss": 0.016917694360017776, "grad_norm": 0.05645148456096649, "global_step": 40977, "epoch": 386, "lr": 1.2167035958139722e-05} +{"train_loss": 0.014443449676036835, "grad_norm": 0.14369651675224304, "global_step": 40978, "epoch": 386, "lr": 1.2165098286104442e-05} +{"train_loss": 0.015506868250668049, "grad_norm": 0.09415919333696365, "global_step": 40979, "epoch": 386, "lr": 1.2163160747004537e-05} +{"train_loss": 0.015427706763148308, "grad_norm": 0.07610463351011276, "global_step": 40980, "epoch": 386, "lr": 1.2161223340846828e-05} +{"train_loss": 0.015738843008875847, "grad_norm": 0.05103710666298866, "global_step": 40981, "epoch": 386, "lr": 1.2159286067638143e-05} +{"train_loss": 0.014523016288876534, "grad_norm": 0.08797501027584076, "global_step": 40982, "epoch": 386, "lr": 1.2157348927385265e-05} +{"train_loss": 0.01378462091088295, "grad_norm": 0.0861305445432663, "global_step": 40983, "epoch": 386, "lr": 1.2155411920094994e-05} +{"train_loss": 0.015565016306936741, "grad_norm": 0.06281708180904388, "global_step": 40984, "epoch": 386, "lr": 1.2153475045774148e-05} +{"train_loss": 0.015082341618835926, "grad_norm": 0.05706600472331047, "global_step": 40985, "epoch": 386, "lr": 1.215153830442955e-05} +{"train_loss": 0.018018154427409172, "grad_norm": 0.14268308877944946, "global_step": 40986, "epoch": 386, "lr": 1.214960169606797e-05} +{"train_loss": 0.01651834510266781, "grad_norm": 0.118087999522686, "global_step": 40987, "epoch": 386, "lr": 1.2147665220696225e-05} +{"train_loss": 0.014660734683275223, "grad_norm": 0.06974107772111893, "global_step": 40988, "epoch": 386, "lr": 1.214572887832114e-05} +{"train_loss": 0.015296596102416515, "grad_norm": 0.08280125260353088, "global_step": 40989, "epoch": 386, "lr": 1.21437926689495e-05} +{"train_loss": 0.015948068350553513, "grad_norm": 0.08019894361495972, "global_step": 40990, "epoch": 386, "lr": 1.2141856592588096e-05} +{"train_loss": 0.01645687222480774, "grad_norm": 0.06293774396181107, "global_step": 40991, "epoch": 386, "lr": 1.213992064924374e-05} +{"train_loss": 0.013334114104509354, "grad_norm": 0.06496734172105789, "global_step": 40992, "epoch": 386, "lr": 1.2137984838923255e-05} +{"train_loss": 0.01575121469795704, "grad_norm": 0.08629101514816284, "global_step": 40993, "epoch": 386, "lr": 1.2136049161633417e-05} +{"train_loss": 0.01554872002452612, "grad_norm": 0.0731426551938057, "global_step": 40994, "epoch": 386, "lr": 1.2134113617381043e-05} +{"train_loss": 0.013385165482759476, "grad_norm": 0.061816975474357605, "global_step": 40995, "epoch": 386, "lr": 1.2132178206172929e-05} +{"train_loss": 0.013629714958369732, "grad_norm": 0.074716717004776, "global_step": 40996, "epoch": 386, "lr": 1.2130242928015856e-05} +{"train_loss": 0.016643328592181206, "grad_norm": 0.075843945145607, "global_step": 40997, "epoch": 386, "lr": 1.2128307782916642e-05} +{"train_loss": 0.013608884066343307, "grad_norm": 0.06380604952573776, "global_step": 40998, "epoch": 386, "lr": 1.21263727708821e-05} +{"train_loss": 0.0163282360881567, "grad_norm": 0.07991337031126022, "global_step": 40999, "epoch": 386, "lr": 1.2124437891918993e-05} +{"train_loss": 0.015290413983166218, "grad_norm": 0.07765503227710724, "global_step": 41000, "epoch": 386, "lr": 1.2122503146034165e-05} +{"train_loss": 0.016707714647054672, "grad_norm": 0.06351610273122787, "global_step": 41001, "epoch": 386, "lr": 1.2120568533234367e-05} +{"train_loss": 0.01475018821656704, "grad_norm": 0.10139605402946472, "global_step": 41002, "epoch": 386, "lr": 1.2118634053526429e-05} +{"train_loss": 0.015570783987641335, "grad_norm": 0.08436161279678345, "global_step": 41003, "epoch": 386, "lr": 1.2116699706917129e-05} +{"train_loss": 0.014354226179420948, "grad_norm": 0.0949779748916626, "global_step": 41004, "epoch": 386, "lr": 1.2114765493413265e-05} +{"train_loss": 0.013791721314191818, "grad_norm": 0.09851144999265671, "global_step": 41005, "epoch": 386, "lr": 1.2112831413021658e-05} +{"train_loss": 0.014731883071362972, "grad_norm": 0.07562283426523209, "global_step": 41006, "epoch": 386, "lr": 1.2110897465749082e-05} +{"train_loss": 0.014870915561914444, "grad_norm": 0.15035949647426605, "global_step": 41007, "epoch": 386, "lr": 1.2108963651602323e-05} +{"train_loss": 0.016998589038848877, "grad_norm": 0.15143883228302002, "global_step": 41008, "epoch": 386, "lr": 1.2107029970588185e-05} +{"train_loss": 0.013849283568561077, "grad_norm": 0.10920049250125885, "global_step": 41009, "epoch": 386, "lr": 1.210509642271348e-05} +{"train_loss": 0.013871227391064167, "grad_norm": 0.075148805975914, "global_step": 41010, "epoch": 386, "lr": 1.2103163007984975e-05} +{"train_loss": 0.013435857370495796, "grad_norm": 0.06234770268201828, "global_step": 41011, "epoch": 386, "lr": 1.2101229726409468e-05} +{"train_loss": 0.013556189835071564, "grad_norm": 0.13104423880577087, "global_step": 41012, "epoch": 386, "lr": 1.209929657799379e-05} +{"train_loss": 0.014721641317009926, "grad_norm": 0.06259500235319138, "global_step": 41013, "epoch": 386, "lr": 1.2097363562744674e-05} +{"train_loss": 0.015748662874102592, "grad_norm": 0.06416802853345871, "global_step": 41014, "epoch": 386, "lr": 1.209543068066894e-05} +{"train_loss": 0.014773286879062653, "grad_norm": 0.0867481604218483, "global_step": 41015, "epoch": 386, "lr": 1.209349793177339e-05} +{"train_loss": 0.016517018899321556, "grad_norm": 0.07037962973117828, "global_step": 41016, "epoch": 386, "lr": 1.2091565316064785e-05} +{"train_loss": 0.016146764159202576, "grad_norm": 0.06365977227687836, "global_step": 41017, "epoch": 386, "lr": 1.2089632833549952e-05} +{"train_loss": 0.016700638458132744, "grad_norm": 0.11078432202339172, "global_step": 41018, "epoch": 386, "lr": 1.2087700484235647e-05} +{"train_loss": 0.020309116691350937, "grad_norm": 0.1220008060336113, "global_step": 41019, "epoch": 386, "lr": 1.2085768268128694e-05} +{"train_loss": 0.016469409689307213, "grad_norm": 0.10215581953525543, "global_step": 41020, "epoch": 386, "lr": 1.2083836185235842e-05} +{"train_loss": 0.0150119171583287, "grad_norm": 0.08848527073860168, "global_step": 41021, "epoch": 386, "lr": 1.2081904235563906e-05, "val_loss": 0.020025698468089104} +{"train_loss": 0.01405930332839489, "grad_norm": 0.08667689561843872, "global_step": 41022, "epoch": 387, "lr": 1.2079972419119678e-05} +{"train_loss": 0.014869176782667637, "grad_norm": 0.12336689233779907, "global_step": 41023, "epoch": 387, "lr": 1.207804073590994e-05} +{"train_loss": 0.012971590273082256, "grad_norm": 0.08261732012033463, "global_step": 41024, "epoch": 387, "lr": 1.207610918594146e-05} +{"train_loss": 0.013695343397557735, "grad_norm": 0.07215098291635513, "global_step": 41025, "epoch": 387, "lr": 1.2074177769221034e-05} +{"train_loss": 0.01313523668795824, "grad_norm": 0.05674343928694725, "global_step": 41026, "epoch": 387, "lr": 1.2072246485755472e-05} +{"train_loss": 0.014592139981687069, "grad_norm": 0.1257714480161667, "global_step": 41027, "epoch": 387, "lr": 1.2070315335551529e-05} +{"train_loss": 0.014197111129760742, "grad_norm": 0.07027105242013931, "global_step": 41028, "epoch": 387, "lr": 1.2068384318616e-05} +{"train_loss": 0.0164029598236084, "grad_norm": 0.26801374554634094, "global_step": 41029, "epoch": 387, "lr": 1.2066453434955688e-05} +{"train_loss": 0.016904767602682114, "grad_norm": 0.10382549464702606, "global_step": 41030, "epoch": 387, "lr": 1.2064522684577362e-05} +{"train_loss": 0.015659961849451065, "grad_norm": 0.09325700253248215, "global_step": 41031, "epoch": 387, "lr": 1.2062592067487794e-05} +{"train_loss": 0.014750795438885689, "grad_norm": 0.08238485455513, "global_step": 41032, "epoch": 387, "lr": 1.2060661583693771e-05} +{"train_loss": 0.014852634631097317, "grad_norm": 0.13081926107406616, "global_step": 41033, "epoch": 387, "lr": 1.2058731233202102e-05} +{"train_loss": 0.01727551966905594, "grad_norm": 0.06123403459787369, "global_step": 41034, "epoch": 387, "lr": 1.2056801016019542e-05} +{"train_loss": 0.014100844971835613, "grad_norm": 0.09918714314699173, "global_step": 41035, "epoch": 387, "lr": 1.2054870932152889e-05} +{"train_loss": 0.015592962503433228, "grad_norm": 0.08962354063987732, "global_step": 41036, "epoch": 387, "lr": 1.2052940981608924e-05} +{"train_loss": 0.013697367161512375, "grad_norm": 0.07790791243314743, "global_step": 41037, "epoch": 387, "lr": 1.2051011164394405e-05} +{"train_loss": 0.015614354982972145, "grad_norm": 0.08012578636407852, "global_step": 41038, "epoch": 387, "lr": 1.2049081480516129e-05} +{"train_loss": 0.01513321977108717, "grad_norm": 0.07889120280742645, "global_step": 41039, "epoch": 387, "lr": 1.2047151929980888e-05} +{"train_loss": 0.014755010604858398, "grad_norm": 0.07571876794099808, "global_step": 41040, "epoch": 387, "lr": 1.204522251279544e-05} +{"train_loss": 0.014803392812609673, "grad_norm": 0.16343554854393005, "global_step": 41041, "epoch": 387, "lr": 1.2043293228966585e-05} +{"train_loss": 0.016799017786979675, "grad_norm": 0.08585938066244125, "global_step": 41042, "epoch": 387, "lr": 1.2041364078501077e-05} +{"train_loss": 0.01320816669613123, "grad_norm": 0.08542976528406143, "global_step": 41043, "epoch": 387, "lr": 1.2039435061405724e-05} +{"train_loss": 0.017384888604283333, "grad_norm": 0.08135004341602325, "global_step": 41044, "epoch": 387, "lr": 1.203750617768727e-05} +{"train_loss": 0.012287994846701622, "grad_norm": 0.11779603362083435, "global_step": 41045, "epoch": 387, "lr": 1.2035577427352512e-05} +{"train_loss": 0.015109606087207794, "grad_norm": 0.14830470085144043, "global_step": 41046, "epoch": 387, "lr": 1.2033648810408238e-05} +{"train_loss": 0.013619267381727695, "grad_norm": 0.0986412912607193, "global_step": 41047, "epoch": 387, "lr": 1.2031720326861212e-05} +{"train_loss": 0.015875082463026047, "grad_norm": 0.07842011004686356, "global_step": 41048, "epoch": 387, "lr": 1.2029791976718191e-05} +{"train_loss": 0.0147557957097888, "grad_norm": 0.1041804701089859, "global_step": 41049, "epoch": 387, "lr": 1.2027863759985974e-05} +{"train_loss": 0.014902877621352673, "grad_norm": 0.1111266240477562, "global_step": 41050, "epoch": 387, "lr": 1.202593567667134e-05} +{"train_loss": 0.017093878239393234, "grad_norm": 0.1033884808421135, "global_step": 41051, "epoch": 387, "lr": 1.202400772678104e-05} +{"train_loss": 0.013938087970018387, "grad_norm": 0.09248464554548264, "global_step": 41052, "epoch": 387, "lr": 1.2022079910321866e-05} +{"train_loss": 0.014917748980224133, "grad_norm": 0.07326121628284454, "global_step": 41053, "epoch": 387, "lr": 1.2020152227300608e-05} +{"train_loss": 0.014130826108157635, "grad_norm": 0.07502058893442154, "global_step": 41054, "epoch": 387, "lr": 1.2018224677723994e-05} +{"train_loss": 0.01470712199807167, "grad_norm": 0.06414586305618286, "global_step": 41055, "epoch": 387, "lr": 1.2016297261598818e-05} +{"train_loss": 0.013339495286345482, "grad_norm": 0.06510850042104721, "global_step": 41056, "epoch": 387, "lr": 1.2014369978931872e-05} +{"train_loss": 0.01767710968852043, "grad_norm": 0.1383424997329712, "global_step": 41057, "epoch": 387, "lr": 1.2012442829729897e-05} +{"train_loss": 0.012723177671432495, "grad_norm": 0.07063090801239014, "global_step": 41058, "epoch": 387, "lr": 1.201051581399968e-05} +{"train_loss": 0.015398930758237839, "grad_norm": 0.08668231964111328, "global_step": 41059, "epoch": 387, "lr": 1.2008588931747993e-05} +{"train_loss": 0.013444137759506702, "grad_norm": 0.07031244784593582, "global_step": 41060, "epoch": 387, "lr": 1.200666218298161e-05} +{"train_loss": 0.014535754919052124, "grad_norm": 0.07325972616672516, "global_step": 41061, "epoch": 387, "lr": 1.2004735567707281e-05} +{"train_loss": 0.014976870268583298, "grad_norm": 0.06137392297387123, "global_step": 41062, "epoch": 387, "lr": 1.2002809085931781e-05} +{"train_loss": 0.01598507910966873, "grad_norm": 0.09097962081432343, "global_step": 41063, "epoch": 387, "lr": 1.2000882737661906e-05} +{"train_loss": 0.01652459055185318, "grad_norm": 0.060713768005371094, "global_step": 41064, "epoch": 387, "lr": 1.1998956522904397e-05} +{"train_loss": 0.011825493536889553, "grad_norm": 0.07742796093225479, "global_step": 41065, "epoch": 387, "lr": 1.1997030441666019e-05} +{"train_loss": 0.016083519905805588, "grad_norm": 0.07118602097034454, "global_step": 41066, "epoch": 387, "lr": 1.1995104493953545e-05} +{"train_loss": 0.01814727857708931, "grad_norm": 0.08351889252662659, "global_step": 41067, "epoch": 387, "lr": 1.1993178679773765e-05} +{"train_loss": 0.016356628388166428, "grad_norm": 0.0865161269903183, "global_step": 41068, "epoch": 387, "lr": 1.1991252999133412e-05} +{"train_loss": 0.017524514347314835, "grad_norm": 0.07215029001235962, "global_step": 41069, "epoch": 387, "lr": 1.1989327452039262e-05} +{"train_loss": 0.014149577356874943, "grad_norm": 0.1610797494649887, "global_step": 41070, "epoch": 387, "lr": 1.1987402038498119e-05} +{"train_loss": 0.013250436633825302, "grad_norm": 0.056108616292476654, "global_step": 41071, "epoch": 387, "lr": 1.1985476758516678e-05} +{"train_loss": 0.014537209644913673, "grad_norm": 0.07220960408449173, "global_step": 41072, "epoch": 387, "lr": 1.1983551612101745e-05} +{"train_loss": 0.013675160706043243, "grad_norm": 0.11961878836154938, "global_step": 41073, "epoch": 387, "lr": 1.1981626599260093e-05} +{"train_loss": 0.014052489772439003, "grad_norm": 0.09296667575836182, "global_step": 41074, "epoch": 387, "lr": 1.1979701719998453e-05} +{"train_loss": 0.013857023790478706, "grad_norm": 0.09033995121717453, "global_step": 41075, "epoch": 387, "lr": 1.1977776974323612e-05} +{"train_loss": 0.017764534801244736, "grad_norm": 0.12536734342575073, "global_step": 41076, "epoch": 387, "lr": 1.1975852362242341e-05} +{"train_loss": 0.014441059902310371, "grad_norm": 0.07023625820875168, "global_step": 41077, "epoch": 387, "lr": 1.1973927883761388e-05} +{"train_loss": 0.015312543138861656, "grad_norm": 0.12422773987054825, "global_step": 41078, "epoch": 387, "lr": 1.1972003538887499e-05} +{"train_loss": 0.014487363398075104, "grad_norm": 0.07538452744483948, "global_step": 41079, "epoch": 387, "lr": 1.1970079327627453e-05} +{"train_loss": 0.01637832447886467, "grad_norm": 0.09630793333053589, "global_step": 41080, "epoch": 387, "lr": 1.1968155249988022e-05} +{"train_loss": 0.017562350258231163, "grad_norm": 0.08623147755861282, "global_step": 41081, "epoch": 387, "lr": 1.196623130597594e-05} +{"train_loss": 0.015877682715654373, "grad_norm": 0.09128183871507645, "global_step": 41082, "epoch": 387, "lr": 1.1964307495597999e-05} +{"train_loss": 0.014980094507336617, "grad_norm": 0.10917236655950546, "global_step": 41083, "epoch": 387, "lr": 1.1962383818860917e-05} +{"train_loss": 0.014062995091080666, "grad_norm": 0.06759726256132126, "global_step": 41084, "epoch": 387, "lr": 1.19604602757715e-05} +{"train_loss": 0.015426579862833023, "grad_norm": 0.1365356743335724, "global_step": 41085, "epoch": 387, "lr": 1.1958536866336461e-05} +{"train_loss": 0.015743112191557884, "grad_norm": 0.11147015541791916, "global_step": 41086, "epoch": 387, "lr": 1.1956613590562588e-05} +{"train_loss": 0.01611362397670746, "grad_norm": 0.07149263471364975, "global_step": 41087, "epoch": 387, "lr": 1.1954690448456634e-05} +{"train_loss": 0.017430616542696953, "grad_norm": 0.0889761671423912, "global_step": 41088, "epoch": 387, "lr": 1.1952767440025359e-05} +{"train_loss": 0.01492508314549923, "grad_norm": 0.1353069692850113, "global_step": 41089, "epoch": 387, "lr": 1.1950844565275499e-05} +{"train_loss": 0.015316221863031387, "grad_norm": 0.0748978927731514, "global_step": 41090, "epoch": 387, "lr": 1.1948921824213821e-05} +{"train_loss": 0.014651265926659107, "grad_norm": 0.09763937443494797, "global_step": 41091, "epoch": 387, "lr": 1.1946999216847099e-05} +{"train_loss": 0.01345323957502842, "grad_norm": 0.08606315404176712, "global_step": 41092, "epoch": 387, "lr": 1.194507674318206e-05} +{"train_loss": 0.014029441401362419, "grad_norm": 0.09548000246286392, "global_step": 41093, "epoch": 387, "lr": 1.1943154403225471e-05} +{"train_loss": 0.017349034547805786, "grad_norm": 0.09169935435056686, "global_step": 41094, "epoch": 387, "lr": 1.1941232196984115e-05} +{"train_loss": 0.013901004567742348, "grad_norm": 0.06824707984924316, "global_step": 41095, "epoch": 387, "lr": 1.1939310124464692e-05} +{"train_loss": 0.015126085840165615, "grad_norm": 0.10038265585899353, "global_step": 41096, "epoch": 387, "lr": 1.1937388185673975e-05} +{"train_loss": 0.013258794322609901, "grad_norm": 0.07495556771755219, "global_step": 41097, "epoch": 387, "lr": 1.1935466380618742e-05} +{"train_loss": 0.01381261833012104, "grad_norm": 0.06346158683300018, "global_step": 41098, "epoch": 387, "lr": 1.193354470930571e-05} +{"train_loss": 0.016595911234617233, "grad_norm": 0.12129520624876022, "global_step": 41099, "epoch": 387, "lr": 1.1931623171741652e-05} +{"train_loss": 0.015948042273521423, "grad_norm": 0.07048695534467697, "global_step": 41100, "epoch": 387, "lr": 1.192970176793332e-05} +{"train_loss": 0.013378534466028214, "grad_norm": 0.058018218725919724, "global_step": 41101, "epoch": 387, "lr": 1.1927780497887469e-05} +{"train_loss": 0.017843592911958694, "grad_norm": 0.06527398526668549, "global_step": 41102, "epoch": 387, "lr": 1.192585936161082e-05} +{"train_loss": 0.015336154028773308, "grad_norm": 0.18248985707759857, "global_step": 41103, "epoch": 387, "lr": 1.1923938359110143e-05} +{"train_loss": 0.016116883605718613, "grad_norm": 0.05416860058903694, "global_step": 41104, "epoch": 387, "lr": 1.1922017490392207e-05} +{"train_loss": 0.01734025403857231, "grad_norm": 0.0764918103814125, "global_step": 41105, "epoch": 387, "lr": 1.1920096755463722e-05} +{"train_loss": 0.01593530923128128, "grad_norm": 0.10975658148527145, "global_step": 41106, "epoch": 387, "lr": 1.1918176154331473e-05} +{"train_loss": 0.015185346826910973, "grad_norm": 0.07434646040201187, "global_step": 41107, "epoch": 387, "lr": 1.191625568700218e-05} +{"train_loss": 0.014383957721292973, "grad_norm": 0.10118327289819717, "global_step": 41108, "epoch": 387, "lr": 1.1914335353482613e-05} +{"train_loss": 0.015508769080042839, "grad_norm": 0.08265095204114914, "global_step": 41109, "epoch": 387, "lr": 1.1912415153779493e-05} +{"train_loss": 0.014822661876678467, "grad_norm": 0.09835413843393326, "global_step": 41110, "epoch": 387, "lr": 1.1910495087899587e-05} +{"train_loss": 0.015590596944093704, "grad_norm": 0.0648510679602623, "global_step": 41111, "epoch": 387, "lr": 1.1908575155849649e-05} +{"train_loss": 0.014217114076018333, "grad_norm": 0.09384150058031082, "global_step": 41112, "epoch": 387, "lr": 1.1906655357636415e-05} +{"train_loss": 0.014928224496543407, "grad_norm": 0.09739840775728226, "global_step": 41113, "epoch": 387, "lr": 1.1904735693266605e-05} +{"train_loss": 0.014220936223864555, "grad_norm": 0.08842422813177109, "global_step": 41114, "epoch": 387, "lr": 1.190281616274701e-05} +{"train_loss": 0.015556611120700836, "grad_norm": 0.12946853041648865, "global_step": 41115, "epoch": 387, "lr": 1.190089676608433e-05} +{"train_loss": 0.012388773262500763, "grad_norm": 0.06926321983337402, "global_step": 41116, "epoch": 387, "lr": 1.1898977503285335e-05} +{"train_loss": 0.016268443316221237, "grad_norm": 0.1528737097978592, "global_step": 41117, "epoch": 387, "lr": 1.1897058374356774e-05} +{"train_loss": 0.015328911133110523, "grad_norm": 0.05902237072587013, "global_step": 41118, "epoch": 387, "lr": 1.1895139379305381e-05} +{"train_loss": 0.015903078019618988, "grad_norm": 0.13291819393634796, "global_step": 41119, "epoch": 387, "lr": 1.1893220518137882e-05} +{"train_loss": 0.014756815508008003, "grad_norm": 0.08068503439426422, "global_step": 41120, "epoch": 387, "lr": 1.1891301790861038e-05} +{"train_loss": 0.014644338749349117, "grad_norm": 0.06717739999294281, "global_step": 41121, "epoch": 387, "lr": 1.1889383197481597e-05} +{"train_loss": 0.013927232474088669, "grad_norm": 0.0927470251917839, "global_step": 41122, "epoch": 387, "lr": 1.188746473800627e-05} +{"train_loss": 0.016044089570641518, "grad_norm": 0.09421467036008835, "global_step": 41123, "epoch": 387, "lr": 1.1885546412441839e-05} +{"train_loss": 0.014872952364385128, "grad_norm": 0.05734911188483238, "global_step": 41124, "epoch": 387, "lr": 1.1883628220795007e-05} +{"train_loss": 0.013617053627967834, "grad_norm": 0.06780952960252762, "global_step": 41125, "epoch": 387, "lr": 1.188171016307254e-05} +{"train_loss": 0.01222531683743, "grad_norm": 0.0751219317317009, "global_step": 41126, "epoch": 387, "lr": 1.1879792239281157e-05} +{"train_loss": 0.015031045812340278, "grad_norm": 0.15648147463798523, "global_step": 41127, "epoch": 387, "lr": 1.18778744494276e-05, "val_loss": 0.019277961924672127} +{"train_loss": 0.016783159226179123, "grad_norm": 0.05619828775525093, "global_step": 41128, "epoch": 388, "lr": 1.1875956793518633e-05} +{"train_loss": 0.01623157411813736, "grad_norm": 0.09245731681585312, "global_step": 41129, "epoch": 388, "lr": 1.1874039271560972e-05} +{"train_loss": 0.015576980076730251, "grad_norm": 0.10804005712270737, "global_step": 41130, "epoch": 388, "lr": 1.1872121883561343e-05} +{"train_loss": 0.015778115019202232, "grad_norm": 0.09184613078832626, "global_step": 41131, "epoch": 388, "lr": 1.187020462952651e-05} +{"train_loss": 0.01324193924665451, "grad_norm": 0.054883621633052826, "global_step": 41132, "epoch": 388, "lr": 1.1868287509463183e-05} +{"train_loss": 0.013561410829424858, "grad_norm": 0.06211308389902115, "global_step": 41133, "epoch": 388, "lr": 1.1866370523378106e-05} +{"train_loss": 0.014853967353701591, "grad_norm": 0.06875704228878021, "global_step": 41134, "epoch": 388, "lr": 1.186445367127802e-05} +{"train_loss": 0.011469867080450058, "grad_norm": 0.09136705100536346, "global_step": 41135, "epoch": 388, "lr": 1.186253695316969e-05} +{"train_loss": 0.01612187922000885, "grad_norm": 0.05950668826699257, "global_step": 41136, "epoch": 388, "lr": 1.1860620369059795e-05} +{"train_loss": 0.012413540855050087, "grad_norm": 0.049442119896411896, "global_step": 41137, "epoch": 388, "lr": 1.1858703918955088e-05} +{"train_loss": 0.013642319478094578, "grad_norm": 0.07537490874528885, "global_step": 41138, "epoch": 388, "lr": 1.1856787602862323e-05} +{"train_loss": 0.014507865533232689, "grad_norm": 0.14497415721416473, "global_step": 41139, "epoch": 388, "lr": 1.1854871420788206e-05} +{"train_loss": 0.016657935455441475, "grad_norm": 0.06750961393117905, "global_step": 41140, "epoch": 388, "lr": 1.1852955372739478e-05} +{"train_loss": 0.018008360639214516, "grad_norm": 0.07626177370548248, "global_step": 41141, "epoch": 388, "lr": 1.1851039458722896e-05} +{"train_loss": 0.016783438622951508, "grad_norm": 0.12572038173675537, "global_step": 41142, "epoch": 388, "lr": 1.1849123678745166e-05} +{"train_loss": 0.020761456340551376, "grad_norm": 0.11533990502357483, "global_step": 41143, "epoch": 388, "lr": 1.1847208032813007e-05} +{"train_loss": 0.014893914572894573, "grad_norm": 0.08494854718446732, "global_step": 41144, "epoch": 388, "lr": 1.1845292520933172e-05} +{"train_loss": 0.013736463151872158, "grad_norm": 0.10456350445747375, "global_step": 41145, "epoch": 388, "lr": 1.18433771431124e-05} +{"train_loss": 0.0154147082939744, "grad_norm": 0.09508563578128815, "global_step": 41146, "epoch": 388, "lr": 1.1841461899357393e-05} +{"train_loss": 0.015186546370387077, "grad_norm": 0.09054309129714966, "global_step": 41147, "epoch": 388, "lr": 1.1839546789674905e-05} +{"train_loss": 0.01576334796845913, "grad_norm": 0.0741957426071167, "global_step": 41148, "epoch": 388, "lr": 1.1837631814071642e-05} +{"train_loss": 0.01350867748260498, "grad_norm": 0.05269279703497887, "global_step": 41149, "epoch": 388, "lr": 1.1835716972554356e-05} +{"train_loss": 0.012794982641935349, "grad_norm": 0.11150544136762619, "global_step": 41150, "epoch": 388, "lr": 1.1833802265129751e-05} +{"train_loss": 0.014964040368795395, "grad_norm": 0.06380677968263626, "global_step": 41151, "epoch": 388, "lr": 1.1831887691804567e-05} +{"train_loss": 0.01802052929997444, "grad_norm": 0.07151518017053604, "global_step": 41152, "epoch": 388, "lr": 1.1829973252585546e-05} +{"train_loss": 0.014809289947152138, "grad_norm": 0.06794597953557968, "global_step": 41153, "epoch": 388, "lr": 1.1828058947479398e-05} +{"train_loss": 0.01611579954624176, "grad_norm": 0.07280226051807404, "global_step": 41154, "epoch": 388, "lr": 1.1826144776492837e-05} +{"train_loss": 0.014709196984767914, "grad_norm": 0.06620428711175919, "global_step": 41155, "epoch": 388, "lr": 1.1824230739632618e-05} +{"train_loss": 0.015084570273756981, "grad_norm": 0.06867998093366623, "global_step": 41156, "epoch": 388, "lr": 1.1822316836905434e-05} +{"train_loss": 0.014867840334773064, "grad_norm": 0.06630498170852661, "global_step": 41157, "epoch": 388, "lr": 1.1820403068318026e-05} +{"train_loss": 0.015283472836017609, "grad_norm": 0.09450557082891464, "global_step": 41158, "epoch": 388, "lr": 1.1818489433877129e-05} +{"train_loss": 0.01648303121328354, "grad_norm": 0.09593746811151505, "global_step": 41159, "epoch": 388, "lr": 1.1816575933589458e-05} +{"train_loss": 0.015769708901643753, "grad_norm": 0.0993104949593544, "global_step": 41160, "epoch": 388, "lr": 1.1814662567461715e-05} +{"train_loss": 0.012130718678236008, "grad_norm": 0.053587764501571655, "global_step": 41161, "epoch": 388, "lr": 1.181274933550065e-05} +{"train_loss": 0.014739053323864937, "grad_norm": 0.0693669319152832, "global_step": 41162, "epoch": 388, "lr": 1.1810836237712985e-05} +{"train_loss": 0.016531653702259064, "grad_norm": 0.10854317247867584, "global_step": 41163, "epoch": 388, "lr": 1.1808923274105421e-05} +{"train_loss": 0.015457695350050926, "grad_norm": 0.155186727643013, "global_step": 41164, "epoch": 388, "lr": 1.180701044468469e-05} +{"train_loss": 0.016012966632843018, "grad_norm": 0.08180487155914307, "global_step": 41165, "epoch": 388, "lr": 1.1805097749457528e-05} +{"train_loss": 0.016040194779634476, "grad_norm": 0.10875038057565689, "global_step": 41166, "epoch": 388, "lr": 1.1803185188430643e-05} +{"train_loss": 0.013253683224320412, "grad_norm": 0.07219479233026505, "global_step": 41167, "epoch": 388, "lr": 1.1801272761610744e-05} +{"train_loss": 0.01628280058503151, "grad_norm": 0.08718808740377426, "global_step": 41168, "epoch": 388, "lr": 1.1799360469004555e-05} +{"train_loss": 0.01420474424958229, "grad_norm": 0.06295658648014069, "global_step": 41169, "epoch": 388, "lr": 1.179744831061882e-05} +{"train_loss": 0.016898278146982193, "grad_norm": 0.06982791423797607, "global_step": 41170, "epoch": 388, "lr": 1.179553628646023e-05} +{"train_loss": 0.013273902237415314, "grad_norm": 0.0771246999502182, "global_step": 41171, "epoch": 388, "lr": 1.1793624396535502e-05} +{"train_loss": 0.015135168097913265, "grad_norm": 0.0768619254231453, "global_step": 41172, "epoch": 388, "lr": 1.1791712640851376e-05} +{"train_loss": 0.017194513231515884, "grad_norm": 0.09032509475946426, "global_step": 41173, "epoch": 388, "lr": 1.1789801019414537e-05} +{"train_loss": 0.015123007819056511, "grad_norm": 0.0567220114171505, "global_step": 41174, "epoch": 388, "lr": 1.1787889532231727e-05} +{"train_loss": 0.01663561537861824, "grad_norm": 0.07741639763116837, "global_step": 41175, "epoch": 388, "lr": 1.1785978179309665e-05} +{"train_loss": 0.01395849883556366, "grad_norm": 0.09009596705436707, "global_step": 41176, "epoch": 388, "lr": 1.178406696065506e-05} +{"train_loss": 0.014151494950056076, "grad_norm": 0.09377098083496094, "global_step": 41177, "epoch": 388, "lr": 1.1782155876274604e-05} +{"train_loss": 0.019133495166897774, "grad_norm": 0.0865505039691925, "global_step": 41178, "epoch": 388, "lr": 1.1780244926175039e-05} +{"train_loss": 0.015041445381939411, "grad_norm": 0.07118186354637146, "global_step": 41179, "epoch": 388, "lr": 1.1778334110363082e-05} +{"train_loss": 0.014816545881330967, "grad_norm": 0.09630721807479858, "global_step": 41180, "epoch": 388, "lr": 1.1776423428845423e-05} +{"train_loss": 0.01659875176846981, "grad_norm": 0.1097545400261879, "global_step": 41181, "epoch": 388, "lr": 1.1774512881628785e-05} +{"train_loss": 0.014981602318584919, "grad_norm": 0.10432872176170349, "global_step": 41182, "epoch": 388, "lr": 1.1772602468719907e-05} +{"train_loss": 0.01370929554104805, "grad_norm": 0.09558525681495667, "global_step": 41183, "epoch": 388, "lr": 1.1770692190125477e-05} +{"train_loss": 0.012477211654186249, "grad_norm": 0.07302004843950272, "global_step": 41184, "epoch": 388, "lr": 1.176878204585219e-05} +{"train_loss": 0.016200270503759384, "grad_norm": 0.11253228783607483, "global_step": 41185, "epoch": 388, "lr": 1.1766872035906785e-05} +{"train_loss": 0.014538388699293137, "grad_norm": 0.14904260635375977, "global_step": 41186, "epoch": 388, "lr": 1.1764962160295972e-05} +{"train_loss": 0.011760490946471691, "grad_norm": 0.09694432467222214, "global_step": 41187, "epoch": 388, "lr": 1.1763052419026438e-05} +{"train_loss": 0.015056246891617775, "grad_norm": 0.09324516355991364, "global_step": 41188, "epoch": 388, "lr": 1.176114281210493e-05} +{"train_loss": 0.017064524814486504, "grad_norm": 0.11337997019290924, "global_step": 41189, "epoch": 388, "lr": 1.1759233339538117e-05} +{"train_loss": 0.01618225686252117, "grad_norm": 0.12688158452510834, "global_step": 41190, "epoch": 388, "lr": 1.1757324001332743e-05} +{"train_loss": 0.015036485157907009, "grad_norm": 0.0932021290063858, "global_step": 41191, "epoch": 388, "lr": 1.175541479749549e-05} +{"train_loss": 0.014393649995326996, "grad_norm": 0.055379271507263184, "global_step": 41192, "epoch": 388, "lr": 1.1753505728033076e-05} +{"train_loss": 0.014770752750337124, "grad_norm": 0.0907282680273056, "global_step": 41193, "epoch": 388, "lr": 1.1751596792952225e-05} +{"train_loss": 0.015042115934193134, "grad_norm": 0.0866539403796196, "global_step": 41194, "epoch": 388, "lr": 1.1749687992259622e-05} +{"train_loss": 0.014893891289830208, "grad_norm": 0.07898388803005219, "global_step": 41195, "epoch": 388, "lr": 1.1747779325961977e-05} +{"train_loss": 0.011344394646584988, "grad_norm": 0.09758392721414566, "global_step": 41196, "epoch": 388, "lr": 1.1745870794066005e-05} +{"train_loss": 0.015145130455493927, "grad_norm": 0.09424301236867905, "global_step": 41197, "epoch": 388, "lr": 1.17439623965784e-05} +{"train_loss": 0.017122138291597366, "grad_norm": 0.07814547419548035, "global_step": 41198, "epoch": 388, "lr": 1.174205413350587e-05} +{"train_loss": 0.015284335240721703, "grad_norm": 0.10546955466270447, "global_step": 41199, "epoch": 388, "lr": 1.174014600485514e-05} +{"train_loss": 0.012911412864923477, "grad_norm": 0.0669870674610138, "global_step": 41200, "epoch": 388, "lr": 1.1738238010632896e-05} +{"train_loss": 0.016037771478295326, "grad_norm": 0.09932485967874527, "global_step": 41201, "epoch": 388, "lr": 1.1736330150845831e-05} +{"train_loss": 0.01589377224445343, "grad_norm": 0.16297179460525513, "global_step": 41202, "epoch": 388, "lr": 1.1734422425500658e-05} +{"train_loss": 0.011829772964119911, "grad_norm": 0.119256392121315, "global_step": 41203, "epoch": 388, "lr": 1.1732514834604102e-05} +{"train_loss": 0.015235615894198418, "grad_norm": 0.12777285277843475, "global_step": 41204, "epoch": 388, "lr": 1.1730607378162833e-05} +{"train_loss": 0.014156682416796684, "grad_norm": 0.08305607736110687, "global_step": 41205, "epoch": 388, "lr": 1.1728700056183562e-05} +{"train_loss": 0.014816077426075935, "grad_norm": 0.0671549141407013, "global_step": 41206, "epoch": 388, "lr": 1.1726792868673015e-05} +{"train_loss": 0.015971750020980835, "grad_norm": 0.08348756283521652, "global_step": 41207, "epoch": 388, "lr": 1.172488581563787e-05} +{"train_loss": 0.01500056404620409, "grad_norm": 0.12940986454486847, "global_step": 41208, "epoch": 388, "lr": 1.1722978897084818e-05} +{"train_loss": 0.014421564526855946, "grad_norm": 0.07188250124454498, "global_step": 41209, "epoch": 388, "lr": 1.172107211302057e-05} +{"train_loss": 0.014341762289404869, "grad_norm": 0.11807992309331894, "global_step": 41210, "epoch": 388, "lr": 1.1719165463451841e-05} +{"train_loss": 0.014899535104632378, "grad_norm": 0.1183023601770401, "global_step": 41211, "epoch": 388, "lr": 1.1717258948385307e-05} +{"train_loss": 0.016363631933927536, "grad_norm": 0.08080045878887177, "global_step": 41212, "epoch": 388, "lr": 1.1715352567827686e-05} +{"train_loss": 0.014016298577189445, "grad_norm": 0.05916757136583328, "global_step": 41213, "epoch": 388, "lr": 1.171344632178567e-05} +{"train_loss": 0.01237858273088932, "grad_norm": 0.06718490272760391, "global_step": 41214, "epoch": 388, "lr": 1.1711540210265937e-05} +{"train_loss": 0.014973336830735207, "grad_norm": 0.0510559007525444, "global_step": 41215, "epoch": 388, "lr": 1.1709634233275201e-05} +{"train_loss": 0.013822494074702263, "grad_norm": 0.09284719079732895, "global_step": 41216, "epoch": 388, "lr": 1.1707728390820171e-05} +{"train_loss": 0.014210431836545467, "grad_norm": 0.08191089332103729, "global_step": 41217, "epoch": 388, "lr": 1.1705822682907536e-05} +{"train_loss": 0.015209950506687164, "grad_norm": 0.07786804437637329, "global_step": 41218, "epoch": 388, "lr": 1.1703917109543965e-05} +{"train_loss": 0.015195057727396488, "grad_norm": 0.05905919149518013, "global_step": 41219, "epoch": 388, "lr": 1.170201167073618e-05} +{"train_loss": 0.01396193727850914, "grad_norm": 0.07888007909059525, "global_step": 41220, "epoch": 388, "lr": 1.1700106366490881e-05} +{"train_loss": 0.015226582996547222, "grad_norm": 0.09112250059843063, "global_step": 41221, "epoch": 388, "lr": 1.169820119681474e-05} +{"train_loss": 0.014417174272239208, "grad_norm": 0.0917324498295784, "global_step": 41222, "epoch": 388, "lr": 1.1696296161714466e-05} +{"train_loss": 0.015292232856154442, "grad_norm": 0.06870413571596146, "global_step": 41223, "epoch": 388, "lr": 1.1694391261196758e-05} +{"train_loss": 0.01556471362709999, "grad_norm": 0.08913762122392654, "global_step": 41224, "epoch": 388, "lr": 1.1692486495268306e-05} +{"train_loss": 0.012858315370976925, "grad_norm": 0.12398867309093475, "global_step": 41225, "epoch": 388, "lr": 1.1690581863935779e-05} +{"train_loss": 0.01567157730460167, "grad_norm": 0.07448412477970123, "global_step": 41226, "epoch": 388, "lr": 1.1688677367205885e-05} +{"train_loss": 0.015302365645766258, "grad_norm": 0.07480233162641525, "global_step": 41227, "epoch": 388, "lr": 1.1686773005085333e-05} +{"train_loss": 0.015812339261174202, "grad_norm": 0.084987111389637, "global_step": 41228, "epoch": 388, "lr": 1.1684868777580787e-05} +{"train_loss": 0.017946936190128326, "grad_norm": 0.07346416264772415, "global_step": 41229, "epoch": 388, "lr": 1.1682964684698966e-05} +{"train_loss": 0.012222347781062126, "grad_norm": 0.06501120328903198, "global_step": 41230, "epoch": 388, "lr": 1.1681060726446536e-05} +{"train_loss": 0.014630195684731007, "grad_norm": 0.05834326148033142, "global_step": 41231, "epoch": 388, "lr": 1.1679156902830185e-05} +{"train_loss": 0.014801601879298687, "grad_norm": 0.11166451126337051, "global_step": 41232, "epoch": 388, "lr": 1.1677253213856609e-05} +{"train_loss": 0.015019744880638033, "grad_norm": 0.086418516933918, "global_step": 41233, "epoch": 388, "lr": 1.1675349659532513e-05, "val_loss": 0.019412320107221603} +{"train_loss": 0.012890777550637722, "grad_norm": 0.083284392952919, "global_step": 41234, "epoch": 389, "lr": 1.167344623986456e-05} +{"train_loss": 0.013800700195133686, "grad_norm": 0.10801298171281815, "global_step": 41235, "epoch": 389, "lr": 1.1671542954859465e-05} +{"train_loss": 0.013553222641348839, "grad_norm": 0.09361924231052399, "global_step": 41236, "epoch": 389, "lr": 1.166963980452388e-05} +{"train_loss": 0.01469445787370205, "grad_norm": 0.09496141970157623, "global_step": 41237, "epoch": 389, "lr": 1.166773678886453e-05} +{"train_loss": 0.015340633690357208, "grad_norm": 0.05815544351935387, "global_step": 41238, "epoch": 389, "lr": 1.1665833907888063e-05} +{"train_loss": 0.01615351065993309, "grad_norm": 0.0924055427312851, "global_step": 41239, "epoch": 389, "lr": 1.1663931161601188e-05} +{"train_loss": 0.015627209097146988, "grad_norm": 0.0757550299167633, "global_step": 41240, "epoch": 389, "lr": 1.1662028550010601e-05} +{"train_loss": 0.014414247125387192, "grad_norm": 0.0810665637254715, "global_step": 41241, "epoch": 389, "lr": 1.1660126073122974e-05} +{"train_loss": 0.01692754030227661, "grad_norm": 0.11584074795246124, "global_step": 41242, "epoch": 389, "lr": 1.1658223730944972e-05} +{"train_loss": 0.01462271623313427, "grad_norm": 0.12280076742172241, "global_step": 41243, "epoch": 389, "lr": 1.1656321523483305e-05} +{"train_loss": 0.014185816049575806, "grad_norm": 0.12105078250169754, "global_step": 41244, "epoch": 389, "lr": 1.1654419450744658e-05} +{"train_loss": 0.015745341777801514, "grad_norm": 0.07502849400043488, "global_step": 41245, "epoch": 389, "lr": 1.165251751273569e-05} +{"train_loss": 0.014244378544390202, "grad_norm": 0.062140706926584244, "global_step": 41246, "epoch": 389, "lr": 1.1650615709463097e-05} +{"train_loss": 0.016209125518798828, "grad_norm": 0.09269534051418304, "global_step": 41247, "epoch": 389, "lr": 1.1648714040933585e-05} +{"train_loss": 0.015851439908146858, "grad_norm": 0.0782674252986908, "global_step": 41248, "epoch": 389, "lr": 1.1646812507153804e-05} +{"train_loss": 0.01399182714521885, "grad_norm": 0.11523714661598206, "global_step": 41249, "epoch": 389, "lr": 1.1644911108130435e-05} +{"train_loss": 0.014291014522314072, "grad_norm": 0.10409970581531525, "global_step": 41250, "epoch": 389, "lr": 1.1643009843870167e-05} +{"train_loss": 0.014716295525431633, "grad_norm": 0.12010747194290161, "global_step": 41251, "epoch": 389, "lr": 1.1641108714379695e-05} +{"train_loss": 0.016040470451116562, "grad_norm": 0.16964422166347504, "global_step": 41252, "epoch": 389, "lr": 1.1639207719665673e-05} +{"train_loss": 0.014262832701206207, "grad_norm": 0.08955641090869904, "global_step": 41253, "epoch": 389, "lr": 1.1637306859734809e-05} +{"train_loss": 0.014855800196528435, "grad_norm": 0.08120646327733994, "global_step": 41254, "epoch": 389, "lr": 1.1635406134593763e-05} +{"train_loss": 0.013835540972650051, "grad_norm": 0.08402613550424576, "global_step": 41255, "epoch": 389, "lr": 1.1633505544249196e-05} +{"train_loss": 0.015136124566197395, "grad_norm": 0.07854367047548294, "global_step": 41256, "epoch": 389, "lr": 1.1631605088707808e-05} +{"train_loss": 0.015664979815483093, "grad_norm": 0.06499271094799042, "global_step": 41257, "epoch": 389, "lr": 1.162970476797629e-05} +{"train_loss": 0.017269723117351532, "grad_norm": 0.1270957887172699, "global_step": 41258, "epoch": 389, "lr": 1.1627804582061292e-05} +{"train_loss": 0.014347204938530922, "grad_norm": 0.06350631266832352, "global_step": 41259, "epoch": 389, "lr": 1.162590453096951e-05} +{"train_loss": 0.01357421837747097, "grad_norm": 0.06656353175640106, "global_step": 41260, "epoch": 389, "lr": 1.16240046147076e-05} +{"train_loss": 0.013712017796933651, "grad_norm": 0.06259402632713318, "global_step": 41261, "epoch": 389, "lr": 1.162210483328226e-05} +{"train_loss": 0.014115065336227417, "grad_norm": 0.05440835654735565, "global_step": 41262, "epoch": 389, "lr": 1.1620205186700139e-05} +{"train_loss": 0.015680257230997086, "grad_norm": 0.06550291925668716, "global_step": 41263, "epoch": 389, "lr": 1.1618305674967927e-05} +{"train_loss": 0.013761896640062332, "grad_norm": 0.07701573520898819, "global_step": 41264, "epoch": 389, "lr": 1.1616406298092313e-05} +{"train_loss": 0.016039792448282242, "grad_norm": 0.08504823595285416, "global_step": 41265, "epoch": 389, "lr": 1.1614507056079954e-05} +{"train_loss": 0.015584137290716171, "grad_norm": 0.061361126601696014, "global_step": 41266, "epoch": 389, "lr": 1.1612607948937505e-05} +{"train_loss": 0.012766646221280098, "grad_norm": 0.0528360940515995, "global_step": 41267, "epoch": 389, "lr": 1.1610708976671664e-05} +{"train_loss": 0.013902978971600533, "grad_norm": 0.0783642902970314, "global_step": 41268, "epoch": 389, "lr": 1.1608810139289105e-05} +{"train_loss": 0.016419563442468643, "grad_norm": 0.14925149083137512, "global_step": 41269, "epoch": 389, "lr": 1.160691143679648e-05} +{"train_loss": 0.013955207541584969, "grad_norm": 0.103583924472332, "global_step": 41270, "epoch": 389, "lr": 1.1605012869200488e-05} +{"train_loss": 0.013817121274769306, "grad_norm": 0.1283060908317566, "global_step": 41271, "epoch": 389, "lr": 1.1603114436507778e-05} +{"train_loss": 0.014385447837412357, "grad_norm": 0.06644398719072342, "global_step": 41272, "epoch": 389, "lr": 1.1601216138725019e-05} +{"train_loss": 0.01343325525522232, "grad_norm": 0.09622164070606232, "global_step": 41273, "epoch": 389, "lr": 1.1599317975858882e-05} +{"train_loss": 0.01733509637415409, "grad_norm": 0.06916496157646179, "global_step": 41274, "epoch": 389, "lr": 1.1597419947916056e-05} +{"train_loss": 0.014323443174362183, "grad_norm": 0.07254113256931305, "global_step": 41275, "epoch": 389, "lr": 1.159552205490318e-05} +{"train_loss": 0.019330665469169617, "grad_norm": 0.08684445172548294, "global_step": 41276, "epoch": 389, "lr": 1.1593624296826954e-05} +{"train_loss": 0.015105105936527252, "grad_norm": 0.09301937371492386, "global_step": 41277, "epoch": 389, "lr": 1.1591726673694015e-05} +{"train_loss": 0.014408537186682224, "grad_norm": 0.14416295289993286, "global_step": 41278, "epoch": 389, "lr": 1.1589829185511059e-05} +{"train_loss": 0.017320875078439713, "grad_norm": 0.06546303629875183, "global_step": 41279, "epoch": 389, "lr": 1.1587931832284726e-05} +{"train_loss": 0.01740642637014389, "grad_norm": 0.15220049023628235, "global_step": 41280, "epoch": 389, "lr": 1.1586034614021695e-05} +{"train_loss": 0.013118168339133263, "grad_norm": 0.0895746499300003, "global_step": 41281, "epoch": 389, "lr": 1.1584137530728645e-05} +{"train_loss": 0.01502107735723257, "grad_norm": 0.09539525955915451, "global_step": 41282, "epoch": 389, "lr": 1.1582240582412235e-05} +{"train_loss": 0.015487276948988438, "grad_norm": 0.06222601234912872, "global_step": 41283, "epoch": 389, "lr": 1.1580343769079105e-05} +{"train_loss": 0.012974830344319344, "grad_norm": 0.09389092028141022, "global_step": 41284, "epoch": 389, "lr": 1.1578447090735939e-05} +{"train_loss": 0.014975477010011673, "grad_norm": 0.0903821811079979, "global_step": 41285, "epoch": 389, "lr": 1.1576550547389415e-05} +{"train_loss": 0.0158601813018322, "grad_norm": 0.06826648116111755, "global_step": 41286, "epoch": 389, "lr": 1.1574654139046171e-05} +{"train_loss": 0.015572239644825459, "grad_norm": 0.11674638092517853, "global_step": 41287, "epoch": 389, "lr": 1.157275786571288e-05} +{"train_loss": 0.015351928770542145, "grad_norm": 0.10635891556739807, "global_step": 41288, "epoch": 389, "lr": 1.1570861727396231e-05} +{"train_loss": 0.017560862004756927, "grad_norm": 0.100922591984272, "global_step": 41289, "epoch": 389, "lr": 1.1568965724102837e-05} +{"train_loss": 0.015276417136192322, "grad_norm": 0.1400899887084961, "global_step": 41290, "epoch": 389, "lr": 1.1567069855839386e-05} +{"train_loss": 0.01792721264064312, "grad_norm": 0.0653630942106247, "global_step": 41291, "epoch": 389, "lr": 1.1565174122612532e-05} +{"train_loss": 0.012346148490905762, "grad_norm": 0.0666579008102417, "global_step": 41292, "epoch": 389, "lr": 1.1563278524428962e-05} +{"train_loss": 0.014031589031219482, "grad_norm": 0.0881996676325798, "global_step": 41293, "epoch": 389, "lr": 1.1561383061295295e-05} +{"train_loss": 0.014946823939681053, "grad_norm": 0.11030018329620361, "global_step": 41294, "epoch": 389, "lr": 1.1559487733218227e-05} +{"train_loss": 0.01743750274181366, "grad_norm": 0.0700240284204483, "global_step": 41295, "epoch": 389, "lr": 1.1557592540204404e-05} +{"train_loss": 0.016421692445874214, "grad_norm": 0.12623517215251923, "global_step": 41296, "epoch": 389, "lr": 1.1555697482260469e-05} +{"train_loss": 0.01569255068898201, "grad_norm": 0.06508881598711014, "global_step": 41297, "epoch": 389, "lr": 1.1553802559393096e-05} +{"train_loss": 0.013981333002448082, "grad_norm": 0.057692840695381165, "global_step": 41298, "epoch": 389, "lr": 1.155190777160895e-05} +{"train_loss": 0.01494248304516077, "grad_norm": 0.09985637664794922, "global_step": 41299, "epoch": 389, "lr": 1.1550013118914666e-05} +{"train_loss": 0.015581673011183739, "grad_norm": 0.07441588491201401, "global_step": 41300, "epoch": 389, "lr": 1.1548118601316932e-05} +{"train_loss": 0.015658210963010788, "grad_norm": 0.08256281912326813, "global_step": 41301, "epoch": 389, "lr": 1.1546224218822365e-05} +{"train_loss": 0.016923800110816956, "grad_norm": 0.09564458578824997, "global_step": 41302, "epoch": 389, "lr": 1.1544329971437667e-05} +{"train_loss": 0.01321666594594717, "grad_norm": 0.10154252499341965, "global_step": 41303, "epoch": 389, "lr": 1.1542435859169448e-05} +{"train_loss": 0.015124109573662281, "grad_norm": 0.07174408435821533, "global_step": 41304, "epoch": 389, "lr": 1.1540541882024391e-05} +{"train_loss": 0.015294692479074001, "grad_norm": 0.1013258770108223, "global_step": 41305, "epoch": 389, "lr": 1.1538648040009154e-05} +{"train_loss": 0.014009525999426842, "grad_norm": 0.07427319884300232, "global_step": 41306, "epoch": 389, "lr": 1.1536754333130384e-05} +{"train_loss": 0.01451248861849308, "grad_norm": 0.0663948729634285, "global_step": 41307, "epoch": 389, "lr": 1.1534860761394717e-05} +{"train_loss": 0.013708202168345451, "grad_norm": 0.09424003213644028, "global_step": 41308, "epoch": 389, "lr": 1.1532967324808818e-05} +{"train_loss": 0.013654937967658043, "grad_norm": 0.06928236782550812, "global_step": 41309, "epoch": 389, "lr": 1.1531074023379362e-05} +{"train_loss": 0.01498632412403822, "grad_norm": 0.06855092942714691, "global_step": 41310, "epoch": 389, "lr": 1.1529180857112965e-05} +{"train_loss": 0.01178484782576561, "grad_norm": 0.05378545820713043, "global_step": 41311, "epoch": 389, "lr": 1.1527287826016298e-05} +{"train_loss": 0.015559074468910694, "grad_norm": 0.07806172221899033, "global_step": 41312, "epoch": 389, "lr": 1.1525394930096034e-05} +{"train_loss": 0.013411463238298893, "grad_norm": 0.0966145396232605, "global_step": 41313, "epoch": 389, "lr": 1.1523502169358779e-05} +{"train_loss": 0.015308219939470291, "grad_norm": 0.07622933387756348, "global_step": 41314, "epoch": 389, "lr": 1.1521609543811201e-05} +{"train_loss": 0.01780012622475624, "grad_norm": 0.07022850960493088, "global_step": 41315, "epoch": 389, "lr": 1.1519717053459967e-05} +{"train_loss": 0.01493110042065382, "grad_norm": 0.0727623924612999, "global_step": 41316, "epoch": 389, "lr": 1.1517824698311696e-05} +{"train_loss": 0.014554500579833984, "grad_norm": 0.06369038671255112, "global_step": 41317, "epoch": 389, "lr": 1.151593247837307e-05} +{"train_loss": 0.015799248591065407, "grad_norm": 0.074812151491642, "global_step": 41318, "epoch": 389, "lr": 1.1514040393650705e-05} +{"train_loss": 0.014159001410007477, "grad_norm": 0.12197386473417282, "global_step": 41319, "epoch": 389, "lr": 1.1512148444151283e-05} +{"train_loss": 0.014645079150795937, "grad_norm": 0.09250330179929733, "global_step": 41320, "epoch": 389, "lr": 1.1510256629881416e-05} +{"train_loss": 0.013933910056948662, "grad_norm": 0.07943198084831238, "global_step": 41321, "epoch": 389, "lr": 1.1508364950847771e-05} +{"train_loss": 0.01796136423945427, "grad_norm": 0.07341158390045166, "global_step": 41322, "epoch": 389, "lr": 1.1506473407057006e-05} +{"train_loss": 0.016599182039499283, "grad_norm": 0.07394188642501831, "global_step": 41323, "epoch": 389, "lr": 1.1504581998515752e-05} +{"train_loss": 0.01393203530460596, "grad_norm": 0.06480485945940018, "global_step": 41324, "epoch": 389, "lr": 1.1502690725230641e-05} +{"train_loss": 0.015082460828125477, "grad_norm": 0.06574194133281708, "global_step": 41325, "epoch": 389, "lr": 1.1500799587208332e-05} +{"train_loss": 0.015138846822082996, "grad_norm": 0.12263571470975876, "global_step": 41326, "epoch": 389, "lr": 1.1498908584455486e-05} +{"train_loss": 0.013089647516608238, "grad_norm": 0.09937003999948502, "global_step": 41327, "epoch": 389, "lr": 1.1497017716978719e-05} +{"train_loss": 0.015677273273468018, "grad_norm": 0.09568903595209122, "global_step": 41328, "epoch": 389, "lr": 1.1495126984784688e-05} +{"train_loss": 0.014931528829038143, "grad_norm": 0.05466359481215477, "global_step": 41329, "epoch": 389, "lr": 1.1493236387880058e-05} +{"train_loss": 0.012663566507399082, "grad_norm": 0.05362527444958687, "global_step": 41330, "epoch": 389, "lr": 1.1491345926271424e-05} +{"train_loss": 0.013311374001204967, "grad_norm": 0.1028437465429306, "global_step": 41331, "epoch": 389, "lr": 1.1489455599965453e-05} +{"train_loss": 0.01704580895602703, "grad_norm": 0.13617682456970215, "global_step": 41332, "epoch": 389, "lr": 1.1487565408968804e-05} +{"train_loss": 0.015287176705896854, "grad_norm": 0.07399996370077133, "global_step": 41333, "epoch": 389, "lr": 1.1485675353288083e-05} +{"train_loss": 0.014080880209803581, "grad_norm": 0.0589391253888607, "global_step": 41334, "epoch": 389, "lr": 1.148378543292995e-05} +{"train_loss": 0.016116563230752945, "grad_norm": 0.08776217699050903, "global_step": 41335, "epoch": 389, "lr": 1.1481895647901064e-05} +{"train_loss": 0.018271295353770256, "grad_norm": 0.17683149874210358, "global_step": 41336, "epoch": 389, "lr": 1.1480005998208042e-05} +{"train_loss": 0.01549883745610714, "grad_norm": 0.09188295155763626, "global_step": 41337, "epoch": 389, "lr": 1.1478116483857515e-05} +{"train_loss": 0.01456054113805294, "grad_norm": 0.1333833932876587, "global_step": 41338, "epoch": 389, "lr": 1.147622710485613e-05} +{"train_loss": 0.015012916001789976, "grad_norm": 0.09650889784097672, "global_step": 41339, "epoch": 389, "lr": 1.1474337861210543e-05, "val_loss": 0.020090270787477493} +{"train_loss": 0.014689112082123756, "grad_norm": 0.09694869071245193, "global_step": 41340, "epoch": 390, "lr": 1.1472448752927367e-05} +{"train_loss": 0.014373423531651497, "grad_norm": 0.06533320993185043, "global_step": 41341, "epoch": 390, "lr": 1.1470559780013267e-05} +{"train_loss": 0.013704768382012844, "grad_norm": 0.09061478078365326, "global_step": 41342, "epoch": 390, "lr": 1.1468670942474846e-05} +{"train_loss": 0.014909503981471062, "grad_norm": 0.06650855392217636, "global_step": 41343, "epoch": 390, "lr": 1.1466782240318774e-05} +{"train_loss": 0.01534406840801239, "grad_norm": 0.09500185400247574, "global_step": 41344, "epoch": 390, "lr": 1.1464893673551657e-05} +{"train_loss": 0.016754478216171265, "grad_norm": 0.06513526290655136, "global_step": 41345, "epoch": 390, "lr": 1.1463005242180147e-05} +{"train_loss": 0.018321486189961433, "grad_norm": 0.10483809560537338, "global_step": 41346, "epoch": 390, "lr": 1.1461116946210893e-05} +{"train_loss": 0.0153731070458889, "grad_norm": 0.07044392824172974, "global_step": 41347, "epoch": 390, "lr": 1.1459228785650511e-05} +{"train_loss": 0.013602839782834053, "grad_norm": 0.09204212576150894, "global_step": 41348, "epoch": 390, "lr": 1.1457340760505625e-05} +{"train_loss": 0.01488739438354969, "grad_norm": 0.0722031518816948, "global_step": 41349, "epoch": 390, "lr": 1.145545287078288e-05} +{"train_loss": 0.015386993065476418, "grad_norm": 0.057568375021219254, "global_step": 41350, "epoch": 390, "lr": 1.1453565116488924e-05} +{"train_loss": 0.015393134206533432, "grad_norm": 0.0549515001475811, "global_step": 41351, "epoch": 390, "lr": 1.1451677497630364e-05} +{"train_loss": 0.014492877759039402, "grad_norm": 0.07418818026781082, "global_step": 41352, "epoch": 390, "lr": 1.1449790014213846e-05} +{"train_loss": 0.013750755228102207, "grad_norm": 0.07355071604251862, "global_step": 41353, "epoch": 390, "lr": 1.144790266624603e-05} +{"train_loss": 0.016066543757915497, "grad_norm": 0.12960204482078552, "global_step": 41354, "epoch": 390, "lr": 1.144601545373349e-05} +{"train_loss": 0.015419780276715755, "grad_norm": 0.17041727900505066, "global_step": 41355, "epoch": 390, "lr": 1.1444128376682883e-05} +{"train_loss": 0.012671804055571556, "grad_norm": 0.0639437735080719, "global_step": 41356, "epoch": 390, "lr": 1.1442241435100853e-05} +{"train_loss": 0.014298858121037483, "grad_norm": 0.1305510401725769, "global_step": 41357, "epoch": 390, "lr": 1.144035462899401e-05} +{"train_loss": 0.015602274797856808, "grad_norm": 0.11443319171667099, "global_step": 41358, "epoch": 390, "lr": 1.1438467958368986e-05} +{"train_loss": 0.014440687373280525, "grad_norm": 0.07986193895339966, "global_step": 41359, "epoch": 390, "lr": 1.1436581423232434e-05} +{"train_loss": 0.01530927699059248, "grad_norm": 0.08381112664937973, "global_step": 41360, "epoch": 390, "lr": 1.1434695023590964e-05} +{"train_loss": 0.012818735092878342, "grad_norm": 0.07131238281726837, "global_step": 41361, "epoch": 390, "lr": 1.1432808759451191e-05} +{"train_loss": 0.015097561292350292, "grad_norm": 0.08041125535964966, "global_step": 41362, "epoch": 390, "lr": 1.1430922630819752e-05} +{"train_loss": 0.015542801469564438, "grad_norm": 0.06652297079563141, "global_step": 41363, "epoch": 390, "lr": 1.1429036637703295e-05} +{"train_loss": 0.01497698575258255, "grad_norm": 0.06526526063680649, "global_step": 41364, "epoch": 390, "lr": 1.1427150780108415e-05} +{"train_loss": 0.01392450649291277, "grad_norm": 0.14379216730594635, "global_step": 41365, "epoch": 390, "lr": 1.1425265058041762e-05} +{"train_loss": 0.01366637833416462, "grad_norm": 0.06200188025832176, "global_step": 41366, "epoch": 390, "lr": 1.1423379471509942e-05} +{"train_loss": 0.014002609997987747, "grad_norm": 0.11012443900108337, "global_step": 41367, "epoch": 390, "lr": 1.1421494020519603e-05} +{"train_loss": 0.01485701184719801, "grad_norm": 0.06719978898763657, "global_step": 41368, "epoch": 390, "lr": 1.1419608705077339e-05} +{"train_loss": 0.014277642592787743, "grad_norm": 0.09759721159934998, "global_step": 41369, "epoch": 390, "lr": 1.1417723525189794e-05} +{"train_loss": 0.01459906343370676, "grad_norm": 0.09024088829755783, "global_step": 41370, "epoch": 390, "lr": 1.1415838480863622e-05} +{"train_loss": 0.016020670533180237, "grad_norm": 0.07974221557378769, "global_step": 41371, "epoch": 390, "lr": 1.141395357210538e-05} +{"train_loss": 0.015692129731178284, "grad_norm": 0.0961751788854599, "global_step": 41372, "epoch": 390, "lr": 1.1412068798921726e-05} +{"train_loss": 0.016949884593486786, "grad_norm": 0.06900374591350555, "global_step": 41373, "epoch": 390, "lr": 1.1410184161319298e-05} +{"train_loss": 0.01450085174292326, "grad_norm": 0.06715239584445953, "global_step": 41374, "epoch": 390, "lr": 1.1408299659304684e-05} +{"train_loss": 0.016341550275683403, "grad_norm": 0.08135067671537399, "global_step": 41375, "epoch": 390, "lr": 1.140641529288452e-05} +{"train_loss": 0.01655116304755211, "grad_norm": 0.1407545655965805, "global_step": 41376, "epoch": 390, "lr": 1.1404531062065449e-05} +{"train_loss": 0.01865941658616066, "grad_norm": 0.08313301205635071, "global_step": 41377, "epoch": 390, "lr": 1.1402646966854064e-05} +{"train_loss": 0.016501251608133316, "grad_norm": 0.09665359556674957, "global_step": 41378, "epoch": 390, "lr": 1.1400763007256982e-05} +{"train_loss": 0.014244724065065384, "grad_norm": 0.07258326560258865, "global_step": 41379, "epoch": 390, "lr": 1.1398879183280831e-05} +{"train_loss": 0.016388021409511566, "grad_norm": 0.08185949921607971, "global_step": 41380, "epoch": 390, "lr": 1.1396995494932245e-05} +{"train_loss": 0.01520839799195528, "grad_norm": 0.13490846753120422, "global_step": 41381, "epoch": 390, "lr": 1.1395111942217818e-05} +{"train_loss": 0.015780311077833176, "grad_norm": 0.06358832120895386, "global_step": 41382, "epoch": 390, "lr": 1.139322852514419e-05} +{"train_loss": 0.017192410305142403, "grad_norm": 0.09304403513669968, "global_step": 41383, "epoch": 390, "lr": 1.1391345243717954e-05} +{"train_loss": 0.014878816902637482, "grad_norm": 0.09905959665775299, "global_step": 41384, "epoch": 390, "lr": 1.1389462097945753e-05} +{"train_loss": 0.01425677165389061, "grad_norm": 0.06085994094610214, "global_step": 41385, "epoch": 390, "lr": 1.1387579087834178e-05} +{"train_loss": 0.0152007807046175, "grad_norm": 0.09095879644155502, "global_step": 41386, "epoch": 390, "lr": 1.1385696213389857e-05} +{"train_loss": 0.014903958886861801, "grad_norm": 0.08737828582525253, "global_step": 41387, "epoch": 390, "lr": 1.1383813474619426e-05} +{"train_loss": 0.01682369038462639, "grad_norm": 0.08690402656793594, "global_step": 41388, "epoch": 390, "lr": 1.1381930871529473e-05} +{"train_loss": 0.015536477789282799, "grad_norm": 0.11920775473117828, "global_step": 41389, "epoch": 390, "lr": 1.1380048404126608e-05} +{"train_loss": 0.014641491696238518, "grad_norm": 0.05846384912729263, "global_step": 41390, "epoch": 390, "lr": 1.137816607241748e-05} +{"train_loss": 0.015950806438922882, "grad_norm": 0.060463469475507736, "global_step": 41391, "epoch": 390, "lr": 1.1376283876408655e-05} +{"train_loss": 0.014909634366631508, "grad_norm": 0.0633750632405281, "global_step": 41392, "epoch": 390, "lr": 1.1374401816106778e-05} +{"train_loss": 0.013421179726719856, "grad_norm": 0.07586121559143066, "global_step": 41393, "epoch": 390, "lr": 1.1372519891518452e-05} +{"train_loss": 0.015394734218716621, "grad_norm": 0.08131831139326096, "global_step": 41394, "epoch": 390, "lr": 1.1370638102650322e-05} +{"train_loss": 0.016302859410643578, "grad_norm": 0.14875906705856323, "global_step": 41395, "epoch": 390, "lr": 1.1368756449508943e-05} +{"train_loss": 0.016530286520719528, "grad_norm": 0.11846105009317398, "global_step": 41396, "epoch": 390, "lr": 1.136687493210095e-05} +{"train_loss": 0.014104683883488178, "grad_norm": 0.08705822378396988, "global_step": 41397, "epoch": 390, "lr": 1.1364993550432973e-05} +{"train_loss": 0.014065911062061787, "grad_norm": 0.06933509558439255, "global_step": 41398, "epoch": 390, "lr": 1.1363112304511598e-05} +{"train_loss": 0.015387531369924545, "grad_norm": 0.08403631299734116, "global_step": 41399, "epoch": 390, "lr": 1.1361231194343436e-05} +{"train_loss": 0.01630447804927826, "grad_norm": 0.06591086089611053, "global_step": 41400, "epoch": 390, "lr": 1.1359350219935122e-05} +{"train_loss": 0.014696547761559486, "grad_norm": 0.06488413363695145, "global_step": 41401, "epoch": 390, "lr": 1.1357469381293246e-05} +{"train_loss": 0.014335609041154385, "grad_norm": 0.08332808315753937, "global_step": 41402, "epoch": 390, "lr": 1.1355588678424406e-05} +{"train_loss": 0.01732223853468895, "grad_norm": 0.14190945029258728, "global_step": 41403, "epoch": 390, "lr": 1.1353708111335215e-05} +{"train_loss": 0.01341741532087326, "grad_norm": 0.058910418301820755, "global_step": 41404, "epoch": 390, "lr": 1.1351827680032307e-05} +{"train_loss": 0.01521773636341095, "grad_norm": 0.10928256809711456, "global_step": 41405, "epoch": 390, "lr": 1.1349947384522253e-05} +{"train_loss": 0.013543358072638512, "grad_norm": 0.10440268367528915, "global_step": 41406, "epoch": 390, "lr": 1.1348067224811687e-05} +{"train_loss": 0.014112213626503944, "grad_norm": 0.07200980186462402, "global_step": 41407, "epoch": 390, "lr": 1.1346187200907194e-05} +{"train_loss": 0.017474090680480003, "grad_norm": 0.12180378288030624, "global_step": 41408, "epoch": 390, "lr": 1.13443073128154e-05} +{"train_loss": 0.014952808618545532, "grad_norm": 0.067974753677845, "global_step": 41409, "epoch": 390, "lr": 1.1342427560542885e-05} +{"train_loss": 0.014301375485956669, "grad_norm": 0.06186928227543831, "global_step": 41410, "epoch": 390, "lr": 1.134054794409627e-05} +{"train_loss": 0.014619866386055946, "grad_norm": 0.07192021608352661, "global_step": 41411, "epoch": 390, "lr": 1.1338668463482166e-05} +{"train_loss": 0.013618245720863342, "grad_norm": 0.12942293286323547, "global_step": 41412, "epoch": 390, "lr": 1.1336789118707174e-05} +{"train_loss": 0.016259530559182167, "grad_norm": 0.07988239079713821, "global_step": 41413, "epoch": 390, "lr": 1.1334909909777875e-05} +{"train_loss": 0.01773536577820778, "grad_norm": 0.06883379071950912, "global_step": 41414, "epoch": 390, "lr": 1.1333030836700897e-05} +{"train_loss": 0.01529183890670538, "grad_norm": 0.11581624299287796, "global_step": 41415, "epoch": 390, "lr": 1.1331151899482828e-05} +{"train_loss": 0.014120185747742653, "grad_norm": 0.12174461036920547, "global_step": 41416, "epoch": 390, "lr": 1.1329273098130266e-05} +{"train_loss": 0.01526567805558443, "grad_norm": 0.10266099870204926, "global_step": 41417, "epoch": 390, "lr": 1.132739443264984e-05} +{"train_loss": 0.016818681731820107, "grad_norm": 0.06355307251214981, "global_step": 41418, "epoch": 390, "lr": 1.1325515903048128e-05} +{"train_loss": 0.015614442527294159, "grad_norm": 0.08276217430830002, "global_step": 41419, "epoch": 390, "lr": 1.132363750933172e-05} +{"train_loss": 0.016142290085554123, "grad_norm": 0.0663076713681221, "global_step": 41420, "epoch": 390, "lr": 1.1321759251507235e-05} +{"train_loss": 0.013291674666106701, "grad_norm": 0.17790883779525757, "global_step": 41421, "epoch": 390, "lr": 1.1319881129581278e-05} +{"train_loss": 0.01500517874956131, "grad_norm": 0.07240065187215805, "global_step": 41422, "epoch": 390, "lr": 1.1318003143560423e-05} +{"train_loss": 0.015232531353831291, "grad_norm": 0.06524691730737686, "global_step": 41423, "epoch": 390, "lr": 1.1316125293451297e-05} +{"train_loss": 0.014679646119475365, "grad_norm": 0.06314274668693542, "global_step": 41424, "epoch": 390, "lr": 1.1314247579260467e-05} +{"train_loss": 0.014426211826503277, "grad_norm": 0.08234928548336029, "global_step": 41425, "epoch": 390, "lr": 1.1312370000994566e-05} +{"train_loss": 0.0151141919195652, "grad_norm": 0.0709538608789444, "global_step": 41426, "epoch": 390, "lr": 1.1310492558660151e-05} +{"train_loss": 0.013509077951312065, "grad_norm": 0.09483719617128372, "global_step": 41427, "epoch": 390, "lr": 1.1308615252263843e-05} +{"train_loss": 0.014866607263684273, "grad_norm": 0.07235969603061676, "global_step": 41428, "epoch": 390, "lr": 1.1306738081812251e-05} +{"train_loss": 0.013586698099970818, "grad_norm": 0.07004547864198685, "global_step": 41429, "epoch": 390, "lr": 1.1304861047311954e-05} +{"train_loss": 0.014271749183535576, "grad_norm": 0.11211515218019485, "global_step": 41430, "epoch": 390, "lr": 1.1302984148769529e-05} +{"train_loss": 0.016885530203580856, "grad_norm": 0.09417807310819626, "global_step": 41431, "epoch": 390, "lr": 1.1301107386191606e-05} +{"train_loss": 0.01327445823699236, "grad_norm": 0.07876276224851608, "global_step": 41432, "epoch": 390, "lr": 1.129923075958475e-05} +{"train_loss": 0.01331547275185585, "grad_norm": 0.12296365946531296, "global_step": 41433, "epoch": 390, "lr": 1.1297354268955562e-05} +{"train_loss": 0.013250697404146194, "grad_norm": 0.0805111676454544, "global_step": 41434, "epoch": 390, "lr": 1.129547791431066e-05} +{"train_loss": 0.013616930693387985, "grad_norm": 0.0968068391084671, "global_step": 41435, "epoch": 390, "lr": 1.1293601695656609e-05} +{"train_loss": 0.013235419988632202, "grad_norm": 0.05467066913843155, "global_step": 41436, "epoch": 390, "lr": 1.1291725612999998e-05} +{"train_loss": 0.015033960342407227, "grad_norm": 0.07735941559076309, "global_step": 41437, "epoch": 390, "lr": 1.128984966634743e-05} +{"train_loss": 0.013406144455075264, "grad_norm": 0.12901918590068817, "global_step": 41438, "epoch": 390, "lr": 1.1287973855705508e-05} +{"train_loss": 0.014024551957845688, "grad_norm": 0.0780666396021843, "global_step": 41439, "epoch": 390, "lr": 1.1286098181080795e-05} +{"train_loss": 0.016422370448708534, "grad_norm": 0.08447560667991638, "global_step": 41440, "epoch": 390, "lr": 1.1284222642479903e-05} +{"train_loss": 0.012690153904259205, "grad_norm": 0.05131290480494499, "global_step": 41441, "epoch": 390, "lr": 1.1282347239909425e-05} +{"train_loss": 0.014527222141623497, "grad_norm": 0.09201078861951828, "global_step": 41442, "epoch": 390, "lr": 1.1280471973375938e-05} +{"train_loss": 0.014433045871555805, "grad_norm": 0.08291833847761154, "global_step": 41443, "epoch": 390, "lr": 1.1278596842886024e-05} +{"train_loss": 0.016276512295007706, "grad_norm": 0.08302349597215652, "global_step": 41444, "epoch": 390, "lr": 1.1276721848446282e-05} +{"train_loss": 0.015026015888477833, "grad_norm": 0.1255795955657959, "global_step": 41445, "epoch": 390, "lr": 1.1274846990063315e-05, "train_batch_mse_error": 0.013395660556852818, "val_loss": 0.01958506368100643} +{"train_loss": 0.016316279768943787, "grad_norm": 0.07459066808223724, "global_step": 41446, "epoch": 391, "lr": 1.1272972267743676e-05} +{"train_loss": 0.014640232548117638, "grad_norm": 0.08112838864326477, "global_step": 41447, "epoch": 391, "lr": 1.1271097681493986e-05} +{"train_loss": 0.016819369047880173, "grad_norm": 0.07464054971933365, "global_step": 41448, "epoch": 391, "lr": 1.1269223231320802e-05} +{"train_loss": 0.016824252903461456, "grad_norm": 0.07760978490114212, "global_step": 41449, "epoch": 391, "lr": 1.1267348917230736e-05} +{"train_loss": 0.01363278366625309, "grad_norm": 0.09364335983991623, "global_step": 41450, "epoch": 391, "lr": 1.1265474739230352e-05} +{"train_loss": 0.015890583395957947, "grad_norm": 0.07408292591571808, "global_step": 41451, "epoch": 391, "lr": 1.1263600697326244e-05} +{"train_loss": 0.016888484358787537, "grad_norm": 0.0811459943652153, "global_step": 41452, "epoch": 391, "lr": 1.1261726791525007e-05} +{"train_loss": 0.013384738937020302, "grad_norm": 0.06685862690210342, "global_step": 41453, "epoch": 391, "lr": 1.1259853021833216e-05} +{"train_loss": 0.01653192564845085, "grad_norm": 0.11411844938993454, "global_step": 41454, "epoch": 391, "lr": 1.1257979388257439e-05} +{"train_loss": 0.014447338879108429, "grad_norm": 0.07570572197437286, "global_step": 41455, "epoch": 391, "lr": 1.1256105890804292e-05} +{"train_loss": 0.016243282705545425, "grad_norm": 0.09218164533376694, "global_step": 41456, "epoch": 391, "lr": 1.1254232529480325e-05} +{"train_loss": 0.012775175273418427, "grad_norm": 0.10472220182418823, "global_step": 41457, "epoch": 391, "lr": 1.1252359304292131e-05} +{"train_loss": 0.013850790448486805, "grad_norm": 0.08466851711273193, "global_step": 41458, "epoch": 391, "lr": 1.1250486215246314e-05} +{"train_loss": 0.01657935231924057, "grad_norm": 0.07857795804738998, "global_step": 41459, "epoch": 391, "lr": 1.1248613262349433e-05} +{"train_loss": 0.01402001827955246, "grad_norm": 0.13581329584121704, "global_step": 41460, "epoch": 391, "lr": 1.1246740445608062e-05} +{"train_loss": 0.015558942221105099, "grad_norm": 0.08072731643915176, "global_step": 41461, "epoch": 391, "lr": 1.1244867765028788e-05} +{"train_loss": 0.013190025463700294, "grad_norm": 0.05545589327812195, "global_step": 41462, "epoch": 391, "lr": 1.1242995220618214e-05} +{"train_loss": 0.014063357375562191, "grad_norm": 0.11752460151910782, "global_step": 41463, "epoch": 391, "lr": 1.1241122812382881e-05} +{"train_loss": 0.016991466283798218, "grad_norm": 0.0989190861582756, "global_step": 41464, "epoch": 391, "lr": 1.1239250540329387e-05} +{"train_loss": 0.016708917915821075, "grad_norm": 0.09298425912857056, "global_step": 41465, "epoch": 391, "lr": 1.1237378404464328e-05} +{"train_loss": 0.014419102109968662, "grad_norm": 0.13737133145332336, "global_step": 41466, "epoch": 391, "lr": 1.123550640479426e-05} +{"train_loss": 0.01663881167769432, "grad_norm": 0.08446254581212997, "global_step": 41467, "epoch": 391, "lr": 1.123363454132575e-05} +{"train_loss": 0.015206238254904747, "grad_norm": 0.08278350532054901, "global_step": 41468, "epoch": 391, "lr": 1.1231762814065389e-05} +{"train_loss": 0.017660299316048622, "grad_norm": 0.06428776681423187, "global_step": 41469, "epoch": 391, "lr": 1.1229891223019773e-05} +{"train_loss": 0.01346585899591446, "grad_norm": 0.08362124860286713, "global_step": 41470, "epoch": 391, "lr": 1.1228019768195452e-05} +{"train_loss": 0.013013901188969612, "grad_norm": 0.11199266463518143, "global_step": 41471, "epoch": 391, "lr": 1.1226148449598994e-05} +{"train_loss": 0.017044806852936745, "grad_norm": 0.14328710734844208, "global_step": 41472, "epoch": 391, "lr": 1.1224277267237005e-05} +{"train_loss": 0.015363759361207485, "grad_norm": 0.11199609190225601, "global_step": 41473, "epoch": 391, "lr": 1.1222406221116027e-05} +{"train_loss": 0.01739325001835823, "grad_norm": 0.06913092732429504, "global_step": 41474, "epoch": 391, "lr": 1.1220535311242652e-05} +{"train_loss": 0.014885293319821358, "grad_norm": 0.10408340394496918, "global_step": 41475, "epoch": 391, "lr": 1.1218664537623463e-05} +{"train_loss": 0.013967279344797134, "grad_norm": 0.07029865682125092, "global_step": 41476, "epoch": 391, "lr": 1.1216793900265028e-05} +{"train_loss": 0.013650108128786087, "grad_norm": 0.07696148753166199, "global_step": 41477, "epoch": 391, "lr": 1.1214923399173888e-05} +{"train_loss": 0.013830240815877914, "grad_norm": 0.13954296708106995, "global_step": 41478, "epoch": 391, "lr": 1.1213053034356647e-05} +{"train_loss": 0.01667102985084057, "grad_norm": 0.057070616632699966, "global_step": 41479, "epoch": 391, "lr": 1.1211182805819881e-05} +{"train_loss": 0.01406930573284626, "grad_norm": 0.09543642401695251, "global_step": 41480, "epoch": 391, "lr": 1.1209312713570136e-05} +{"train_loss": 0.01540873758494854, "grad_norm": 0.09480147808790207, "global_step": 41481, "epoch": 391, "lr": 1.1207442757614e-05} +{"train_loss": 0.012417696416378021, "grad_norm": 0.049320634454488754, "global_step": 41482, "epoch": 391, "lr": 1.1205572937958048e-05} +{"train_loss": 0.01661384105682373, "grad_norm": 0.10344414412975311, "global_step": 41483, "epoch": 391, "lr": 1.1203703254608843e-05} +{"train_loss": 0.018132109194993973, "grad_norm": 0.08859618008136749, "global_step": 41484, "epoch": 391, "lr": 1.120183370757294e-05} +{"train_loss": 0.014942926354706287, "grad_norm": 0.14451761543750763, "global_step": 41485, "epoch": 391, "lr": 1.1199964296856924e-05} +{"train_loss": 0.014444617554545403, "grad_norm": 0.06102649122476578, "global_step": 41486, "epoch": 391, "lr": 1.119809502246737e-05} +{"train_loss": 0.01770934648811817, "grad_norm": 0.0681014284491539, "global_step": 41487, "epoch": 391, "lr": 1.1196225884410822e-05} +{"train_loss": 0.016538510099053383, "grad_norm": 0.06593097001314163, "global_step": 41488, "epoch": 391, "lr": 1.1194356882693873e-05} +{"train_loss": 0.017602451145648956, "grad_norm": 0.09436043351888657, "global_step": 41489, "epoch": 391, "lr": 1.1192488017323078e-05} +{"train_loss": 0.013952689245343208, "grad_norm": 0.14164987206459045, "global_step": 41490, "epoch": 391, "lr": 1.1190619288304993e-05} +{"train_loss": 0.015073412097990513, "grad_norm": 0.0869249701499939, "global_step": 41491, "epoch": 391, "lr": 1.1188750695646188e-05} +{"train_loss": 0.018282398581504822, "grad_norm": 0.10274064540863037, "global_step": 41492, "epoch": 391, "lr": 1.1186882239353253e-05} +{"train_loss": 0.012568170204758644, "grad_norm": 0.078085757791996, "global_step": 41493, "epoch": 391, "lr": 1.1185013919432718e-05} +{"train_loss": 0.01482695434242487, "grad_norm": 0.08661771565675735, "global_step": 41494, "epoch": 391, "lr": 1.1183145735891182e-05} +{"train_loss": 0.016200890764594078, "grad_norm": 0.07571709156036377, "global_step": 41495, "epoch": 391, "lr": 1.1181277688735176e-05} +{"train_loss": 0.012798886746168137, "grad_norm": 0.06809531897306442, "global_step": 41496, "epoch": 391, "lr": 1.117940977797129e-05} +{"train_loss": 0.01393373403698206, "grad_norm": 0.05540252849459648, "global_step": 41497, "epoch": 391, "lr": 1.1177542003606062e-05} +{"train_loss": 0.013522080145776272, "grad_norm": 0.08022214472293854, "global_step": 41498, "epoch": 391, "lr": 1.1175674365646066e-05} +{"train_loss": 0.012185228057205677, "grad_norm": 0.10710655152797699, "global_step": 41499, "epoch": 391, "lr": 1.1173806864097886e-05} +{"train_loss": 0.015518543310463428, "grad_norm": 0.07950194925069809, "global_step": 41500, "epoch": 391, "lr": 1.117193949896806e-05} +{"train_loss": 0.013889127410948277, "grad_norm": 0.10430462658405304, "global_step": 41501, "epoch": 391, "lr": 1.1170072270263138e-05} +{"train_loss": 0.016783736646175385, "grad_norm": 0.11690574884414673, "global_step": 41502, "epoch": 391, "lr": 1.1168205177989694e-05} +{"train_loss": 0.015022849664092064, "grad_norm": 0.08646128326654434, "global_step": 41503, "epoch": 391, "lr": 1.1166338222154305e-05} +{"train_loss": 0.01494886539876461, "grad_norm": 0.08537789434194565, "global_step": 41504, "epoch": 391, "lr": 1.11644714027635e-05} +{"train_loss": 0.015121936798095703, "grad_norm": 0.07040729373693466, "global_step": 41505, "epoch": 391, "lr": 1.1162604719823855e-05} +{"train_loss": 0.015209460631012917, "grad_norm": 0.08900411427021027, "global_step": 41506, "epoch": 391, "lr": 1.1160738173341939e-05} +{"train_loss": 0.01437533088028431, "grad_norm": 0.06676023453474045, "global_step": 41507, "epoch": 391, "lr": 1.1158871763324303e-05} +{"train_loss": 0.015749294310808182, "grad_norm": 0.08930838108062744, "global_step": 41508, "epoch": 391, "lr": 1.1157005489777483e-05} +{"train_loss": 0.013097856193780899, "grad_norm": 0.11079517006874084, "global_step": 41509, "epoch": 391, "lr": 1.1155139352708056e-05} +{"train_loss": 0.01491236500442028, "grad_norm": 0.0931781530380249, "global_step": 41510, "epoch": 391, "lr": 1.1153273352122584e-05} +{"train_loss": 0.013719502836465836, "grad_norm": 0.06997635960578918, "global_step": 41511, "epoch": 391, "lr": 1.1151407488027605e-05} +{"train_loss": 0.01624847948551178, "grad_norm": 0.09513434022665024, "global_step": 41512, "epoch": 391, "lr": 1.11495417604297e-05} +{"train_loss": 0.015020963735878468, "grad_norm": 0.07648029178380966, "global_step": 41513, "epoch": 391, "lr": 1.1147676169335408e-05} +{"train_loss": 0.016253367066383362, "grad_norm": 0.10450766235589981, "global_step": 41514, "epoch": 391, "lr": 1.1145810714751276e-05} +{"train_loss": 0.014988461509346962, "grad_norm": 0.08416479080915451, "global_step": 41515, "epoch": 391, "lr": 1.1143945396683864e-05} +{"train_loss": 0.013959743082523346, "grad_norm": 0.1171555444598198, "global_step": 41516, "epoch": 391, "lr": 1.1142080215139745e-05} +{"train_loss": 0.013388562016189098, "grad_norm": 0.08873505145311356, "global_step": 41517, "epoch": 391, "lr": 1.1140215170125439e-05} +{"train_loss": 0.015488126315176487, "grad_norm": 0.05901390314102173, "global_step": 41518, "epoch": 391, "lr": 1.113835026164754e-05} +{"train_loss": 0.014962868764996529, "grad_norm": 0.09688816964626312, "global_step": 41519, "epoch": 391, "lr": 1.1136485489712556e-05} +{"train_loss": 0.01621786318719387, "grad_norm": 0.10226760804653168, "global_step": 41520, "epoch": 391, "lr": 1.1134620854327076e-05} +{"train_loss": 0.014021147042512894, "grad_norm": 0.07683375477790833, "global_step": 41521, "epoch": 391, "lr": 1.1132756355497625e-05} +{"train_loss": 0.013486593961715698, "grad_norm": 0.07842633873224258, "global_step": 41522, "epoch": 391, "lr": 1.1130891993230764e-05} +{"train_loss": 0.01428578607738018, "grad_norm": 0.07575333118438721, "global_step": 41523, "epoch": 391, "lr": 1.1129027767533057e-05} +{"train_loss": 0.015633942559361458, "grad_norm": 0.06826511770486832, "global_step": 41524, "epoch": 391, "lr": 1.1127163678411045e-05} +{"train_loss": 0.01715352013707161, "grad_norm": 0.0764985904097557, "global_step": 41525, "epoch": 391, "lr": 1.1125299725871258e-05} +{"train_loss": 0.01491487119346857, "grad_norm": 0.1396971344947815, "global_step": 41526, "epoch": 391, "lr": 1.112343590992026e-05} +{"train_loss": 0.01802634261548519, "grad_norm": 0.06899811327457428, "global_step": 41527, "epoch": 391, "lr": 1.112157223056462e-05} +{"train_loss": 0.012376999482512474, "grad_norm": 0.10206824541091919, "global_step": 41528, "epoch": 391, "lr": 1.1119708687810853e-05} +{"train_loss": 0.015655308961868286, "grad_norm": 0.07353458553552628, "global_step": 41529, "epoch": 391, "lr": 1.1117845281665535e-05} +{"train_loss": 0.015217294916510582, "grad_norm": 0.08401709794998169, "global_step": 41530, "epoch": 391, "lr": 1.1115982012135196e-05} +{"train_loss": 0.016283003613352776, "grad_norm": 0.10818474739789963, "global_step": 41531, "epoch": 391, "lr": 1.111411887922637e-05} +{"train_loss": 0.013822121545672417, "grad_norm": 0.06465697288513184, "global_step": 41532, "epoch": 391, "lr": 1.1112255882945627e-05} +{"train_loss": 0.015016697347164154, "grad_norm": 0.11498887091875076, "global_step": 41533, "epoch": 391, "lr": 1.1110393023299515e-05} +{"train_loss": 0.01411939412355423, "grad_norm": 0.10312352329492569, "global_step": 41534, "epoch": 391, "lr": 1.1108530300294551e-05} +{"train_loss": 0.014004942029714584, "grad_norm": 0.06761197745800018, "global_step": 41535, "epoch": 391, "lr": 1.1106667713937318e-05} +{"train_loss": 0.017474979162216187, "grad_norm": 0.11066295951604843, "global_step": 41536, "epoch": 391, "lr": 1.1104805264234324e-05} +{"train_loss": 0.013048186898231506, "grad_norm": 0.10304205119609833, "global_step": 41537, "epoch": 391, "lr": 1.1102942951192141e-05} +{"train_loss": 0.01443510502576828, "grad_norm": 0.07164804637432098, "global_step": 41538, "epoch": 391, "lr": 1.1101080774817291e-05} +{"train_loss": 0.015538211911916733, "grad_norm": 0.11843252927064896, "global_step": 41539, "epoch": 391, "lr": 1.1099218735116323e-05} +{"train_loss": 0.012262802571058273, "grad_norm": 0.07106326520442963, "global_step": 41540, "epoch": 391, "lr": 1.10973568320958e-05} +{"train_loss": 0.012046176008880138, "grad_norm": 0.06011684611439705, "global_step": 41541, "epoch": 391, "lr": 1.1095495065762246e-05} +{"train_loss": 0.015763193368911743, "grad_norm": 0.12963524460792542, "global_step": 41542, "epoch": 391, "lr": 1.1093633436122186e-05} +{"train_loss": 0.015168756246566772, "grad_norm": 0.07178863137960434, "global_step": 41543, "epoch": 391, "lr": 1.1091771943182183e-05} +{"train_loss": 0.013932611793279648, "grad_norm": 0.0800490602850914, "global_step": 41544, "epoch": 391, "lr": 1.1089910586948787e-05} +{"train_loss": 0.015523710288107395, "grad_norm": 0.08610869944095612, "global_step": 41545, "epoch": 391, "lr": 1.1088049367428505e-05} +{"train_loss": 0.013357609510421753, "grad_norm": 0.10087892413139343, "global_step": 41546, "epoch": 391, "lr": 1.1086188284627896e-05} +{"train_loss": 0.012995285913348198, "grad_norm": 0.06714164465665817, "global_step": 41547, "epoch": 391, "lr": 1.108432733855353e-05} +{"train_loss": 0.017889615148305893, "grad_norm": 0.13111147284507751, "global_step": 41548, "epoch": 391, "lr": 1.1082466529211882e-05} +{"train_loss": 0.013342048972845078, "grad_norm": 0.05729609355330467, "global_step": 41549, "epoch": 391, "lr": 1.1080605856609521e-05} +{"train_loss": 0.013187232427299023, "grad_norm": 0.06952471286058426, "global_step": 41550, "epoch": 391, "lr": 1.1078745320752987e-05} +{"train_loss": 0.01502689835176153, "grad_norm": 0.1586173176765442, "global_step": 41551, "epoch": 391, "lr": 1.1076884921648834e-05, "val_loss": 0.020194608718156815} +{"train_loss": 0.01362615916877985, "grad_norm": 0.06635631620883942, "global_step": 41552, "epoch": 392, "lr": 1.1075024659303562e-05} +{"train_loss": 0.015400419011712074, "grad_norm": 0.0803200900554657, "global_step": 41553, "epoch": 392, "lr": 1.107316453372374e-05} +{"train_loss": 0.015015552751719952, "grad_norm": 0.05908338725566864, "global_step": 41554, "epoch": 392, "lr": 1.1071304544915883e-05} +{"train_loss": 0.017338760197162628, "grad_norm": 0.0840061604976654, "global_step": 41555, "epoch": 392, "lr": 1.1069444692886527e-05} +{"train_loss": 0.014050059020519257, "grad_norm": 0.054193586111068726, "global_step": 41556, "epoch": 392, "lr": 1.1067584977642209e-05} +{"train_loss": 0.015976469963788986, "grad_norm": 0.08723975718021393, "global_step": 41557, "epoch": 392, "lr": 1.106572539918948e-05} +{"train_loss": 0.017859483137726784, "grad_norm": 0.08345016092061996, "global_step": 41558, "epoch": 392, "lr": 1.1063865957534846e-05} +{"train_loss": 0.01576678827404976, "grad_norm": 0.09272117912769318, "global_step": 41559, "epoch": 392, "lr": 1.1062006652684864e-05} +{"train_loss": 0.014874164015054703, "grad_norm": 0.09895120561122894, "global_step": 41560, "epoch": 392, "lr": 1.106014748464605e-05} +{"train_loss": 0.014899120666086674, "grad_norm": 0.08382666110992432, "global_step": 41561, "epoch": 392, "lr": 1.1058288453424954e-05} +{"train_loss": 0.012329958379268646, "grad_norm": 0.10192792117595673, "global_step": 41562, "epoch": 392, "lr": 1.1056429559028082e-05} +{"train_loss": 0.016511209309101105, "grad_norm": 0.05900995805859566, "global_step": 41563, "epoch": 392, "lr": 1.1054570801461984e-05} +{"train_loss": 0.016661375761032104, "grad_norm": 0.11654987931251526, "global_step": 41564, "epoch": 392, "lr": 1.1052712180733193e-05} +{"train_loss": 0.01349298469722271, "grad_norm": 0.0726608857512474, "global_step": 41565, "epoch": 392, "lr": 1.105085369684824e-05} +{"train_loss": 0.014699138700962067, "grad_norm": 0.06069334223866463, "global_step": 41566, "epoch": 392, "lr": 1.1048995349813635e-05} +{"train_loss": 0.014119191095232964, "grad_norm": 0.07686084508895874, "global_step": 41567, "epoch": 392, "lr": 1.1047137139635922e-05} +{"train_loss": 0.013143263757228851, "grad_norm": 0.09084822982549667, "global_step": 41568, "epoch": 392, "lr": 1.104527906632164e-05} +{"train_loss": 0.01474075112491846, "grad_norm": 0.11526474356651306, "global_step": 41569, "epoch": 392, "lr": 1.104342112987729e-05} +{"train_loss": 0.017133232206106186, "grad_norm": 0.10855134576559067, "global_step": 41570, "epoch": 392, "lr": 1.1041563330309434e-05} +{"train_loss": 0.017187610268592834, "grad_norm": 0.08715134114027023, "global_step": 41571, "epoch": 392, "lr": 1.1039705667624583e-05} +{"train_loss": 0.014013494364917278, "grad_norm": 0.11972756683826447, "global_step": 41572, "epoch": 392, "lr": 1.1037848141829249e-05} +{"train_loss": 0.011980063281953335, "grad_norm": 0.050875745713710785, "global_step": 41573, "epoch": 392, "lr": 1.103599075292997e-05} +{"train_loss": 0.015299547463655472, "grad_norm": 0.06291501969099045, "global_step": 41574, "epoch": 392, "lr": 1.1034133500933291e-05} +{"train_loss": 0.01720644347369671, "grad_norm": 0.12087865173816681, "global_step": 41575, "epoch": 392, "lr": 1.1032276385845702e-05} +{"train_loss": 0.01677524857223034, "grad_norm": 0.07261057943105698, "global_step": 41576, "epoch": 392, "lr": 1.1030419407673765e-05} +{"train_loss": 0.015582803636789322, "grad_norm": 0.07641562074422836, "global_step": 41577, "epoch": 392, "lr": 1.1028562566423972e-05} +{"train_loss": 0.014155169948935509, "grad_norm": 0.08321346342563629, "global_step": 41578, "epoch": 392, "lr": 1.1026705862102876e-05} +{"train_loss": 0.015069487504661083, "grad_norm": 0.060691095888614655, "global_step": 41579, "epoch": 392, "lr": 1.1024849294716976e-05} +{"train_loss": 0.017397206276655197, "grad_norm": 0.12767669558525085, "global_step": 41580, "epoch": 392, "lr": 1.1022992864272803e-05} +{"train_loss": 0.014347659423947334, "grad_norm": 0.058900754898786545, "global_step": 41581, "epoch": 392, "lr": 1.102113657077689e-05} +{"train_loss": 0.015366829931735992, "grad_norm": 0.10844805091619492, "global_step": 41582, "epoch": 392, "lr": 1.101928041423576e-05} +{"train_loss": 0.016830969601869583, "grad_norm": 0.07684118300676346, "global_step": 41583, "epoch": 392, "lr": 1.1017424394655913e-05} +{"train_loss": 0.016549786552786827, "grad_norm": 0.07823297381401062, "global_step": 41584, "epoch": 392, "lr": 1.1015568512043877e-05} +{"train_loss": 0.01636751927435398, "grad_norm": 0.07063397020101547, "global_step": 41585, "epoch": 392, "lr": 1.1013712766406197e-05} +{"train_loss": 0.01394391804933548, "grad_norm": 0.0675579160451889, "global_step": 41586, "epoch": 392, "lr": 1.101185715774936e-05} +{"train_loss": 0.016233431175351143, "grad_norm": 0.07012369483709335, "global_step": 41587, "epoch": 392, "lr": 1.1010001686079907e-05} +{"train_loss": 0.01429934985935688, "grad_norm": 0.16369491815567017, "global_step": 41588, "epoch": 392, "lr": 1.1008146351404374e-05} +{"train_loss": 0.015145591460168362, "grad_norm": 0.06879480928182602, "global_step": 41589, "epoch": 392, "lr": 1.1006291153729231e-05} +{"train_loss": 0.01297655701637268, "grad_norm": 0.1073366329073906, "global_step": 41590, "epoch": 392, "lr": 1.1004436093061027e-05} +{"train_loss": 0.013142090290784836, "grad_norm": 0.057848669588565826, "global_step": 41591, "epoch": 392, "lr": 1.1002581169406285e-05} +{"train_loss": 0.016955913975834846, "grad_norm": 0.07914583384990692, "global_step": 41592, "epoch": 392, "lr": 1.1000726382771498e-05} +{"train_loss": 0.01569429412484169, "grad_norm": 0.0642741397023201, "global_step": 41593, "epoch": 392, "lr": 1.0998871733163203e-05} +{"train_loss": 0.015295655466616154, "grad_norm": 0.16012389957904816, "global_step": 41594, "epoch": 392, "lr": 1.0997017220587925e-05} +{"train_loss": 0.0171346515417099, "grad_norm": 0.07488267868757248, "global_step": 41595, "epoch": 392, "lr": 1.099516284505216e-05} +{"train_loss": 0.015046972781419754, "grad_norm": 0.07974933087825775, "global_step": 41596, "epoch": 392, "lr": 1.0993308606562425e-05} +{"train_loss": 0.012965977191925049, "grad_norm": 0.11772140860557556, "global_step": 41597, "epoch": 392, "lr": 1.0991454505125232e-05} +{"train_loss": 0.01595289632678032, "grad_norm": 0.08457532525062561, "global_step": 41598, "epoch": 392, "lr": 1.0989600540747125e-05} +{"train_loss": 0.013488522730767727, "grad_norm": 0.08553431183099747, "global_step": 41599, "epoch": 392, "lr": 1.0987746713434576e-05} +{"train_loss": 0.015209869481623173, "grad_norm": 0.12206760793924332, "global_step": 41600, "epoch": 392, "lr": 1.0985893023194137e-05} +{"train_loss": 0.016081437468528748, "grad_norm": 0.10849063843488693, "global_step": 41601, "epoch": 392, "lr": 1.0984039470032288e-05} +{"train_loss": 0.014616986736655235, "grad_norm": 0.06449683010578156, "global_step": 41602, "epoch": 392, "lr": 1.0982186053955568e-05} +{"train_loss": 0.01571929268538952, "grad_norm": 0.07866980880498886, "global_step": 41603, "epoch": 392, "lr": 1.098033277497047e-05} +{"train_loss": 0.01421702466905117, "grad_norm": 0.07334108650684357, "global_step": 41604, "epoch": 392, "lr": 1.097847963308351e-05} +{"train_loss": 0.017214901745319366, "grad_norm": 0.07098835706710815, "global_step": 41605, "epoch": 392, "lr": 1.0976626628301217e-05} +{"train_loss": 0.014193625189363956, "grad_norm": 0.06962395459413528, "global_step": 41606, "epoch": 392, "lr": 1.0974773760630086e-05} +{"train_loss": 0.015252240933477879, "grad_norm": 0.09534952044487, "global_step": 41607, "epoch": 392, "lr": 1.0972921030076616e-05} +{"train_loss": 0.012077449820935726, "grad_norm": 0.0888608768582344, "global_step": 41608, "epoch": 392, "lr": 1.0971068436647326e-05} +{"train_loss": 0.013002613559365273, "grad_norm": 0.08825910836458206, "global_step": 41609, "epoch": 392, "lr": 1.0969215980348746e-05} +{"train_loss": 0.012635793536901474, "grad_norm": 0.07147987186908722, "global_step": 41610, "epoch": 392, "lr": 1.0967363661187352e-05} +{"train_loss": 0.01506895013153553, "grad_norm": 0.11385228484869003, "global_step": 41611, "epoch": 392, "lr": 1.0965511479169666e-05} +{"train_loss": 0.014209497720003128, "grad_norm": 0.11844180524349213, "global_step": 41612, "epoch": 392, "lr": 1.0963659434302226e-05} +{"train_loss": 0.017247090116143227, "grad_norm": 0.11112777888774872, "global_step": 41613, "epoch": 392, "lr": 1.0961807526591477e-05} +{"train_loss": 0.014275839552283287, "grad_norm": 0.07263031601905823, "global_step": 41614, "epoch": 392, "lr": 1.0959955756043965e-05} +{"train_loss": 0.015127411112189293, "grad_norm": 0.10457015782594681, "global_step": 41615, "epoch": 392, "lr": 1.0958104122666207e-05} +{"train_loss": 0.014532340690493584, "grad_norm": 0.08871488273143768, "global_step": 41616, "epoch": 392, "lr": 1.0956252626464674e-05} +{"train_loss": 0.013748537749052048, "grad_norm": 0.06284898519515991, "global_step": 41617, "epoch": 392, "lr": 1.095440126744589e-05} +{"train_loss": 0.014372654259204865, "grad_norm": 0.06901296973228455, "global_step": 41618, "epoch": 392, "lr": 1.0952550045616372e-05} +{"train_loss": 0.013449965044856071, "grad_norm": 0.07197372615337372, "global_step": 41619, "epoch": 392, "lr": 1.0950698960982614e-05} +{"train_loss": 0.014872109517455101, "grad_norm": 0.14160215854644775, "global_step": 41620, "epoch": 392, "lr": 1.0948848013551105e-05} +{"train_loss": 0.015829291194677353, "grad_norm": 0.09171346575021744, "global_step": 41621, "epoch": 392, "lr": 1.0946997203328357e-05} +{"train_loss": 0.014410833828151226, "grad_norm": 0.06146789714694023, "global_step": 41622, "epoch": 392, "lr": 1.0945146530320893e-05} +{"train_loss": 0.0142561886459589, "grad_norm": 0.10325382649898529, "global_step": 41623, "epoch": 392, "lr": 1.0943295994535202e-05} +{"train_loss": 0.01737918332219124, "grad_norm": 0.11041898280382156, "global_step": 41624, "epoch": 392, "lr": 1.0941445595977767e-05} +{"train_loss": 0.013848747126758099, "grad_norm": 0.11472257226705551, "global_step": 41625, "epoch": 392, "lr": 1.0939595334655107e-05} +{"train_loss": 0.013607184402644634, "grad_norm": 0.09112038463354111, "global_step": 41626, "epoch": 392, "lr": 1.0937745210573736e-05} +{"train_loss": 0.015532681718468666, "grad_norm": 0.07625770568847656, "global_step": 41627, "epoch": 392, "lr": 1.0935895223740127e-05} +{"train_loss": 0.013063674792647362, "grad_norm": 0.12713584303855896, "global_step": 41628, "epoch": 392, "lr": 1.0934045374160794e-05} +{"train_loss": 0.015546953305602074, "grad_norm": 0.0825568288564682, "global_step": 41629, "epoch": 392, "lr": 1.0932195661842265e-05} +{"train_loss": 0.012305406853556633, "grad_norm": 0.09790603816509247, "global_step": 41630, "epoch": 392, "lr": 1.0930346086790977e-05} +{"train_loss": 0.017515383660793304, "grad_norm": 0.07274473458528519, "global_step": 41631, "epoch": 392, "lr": 1.092849664901347e-05} +{"train_loss": 0.015005986206233501, "grad_norm": 0.07702622562646866, "global_step": 41632, "epoch": 392, "lr": 1.092664734851624e-05} +{"train_loss": 0.013860685750842094, "grad_norm": 0.05935506895184517, "global_step": 41633, "epoch": 392, "lr": 1.0924798185305768e-05} +{"train_loss": 0.014516250230371952, "grad_norm": 0.07025173306465149, "global_step": 41634, "epoch": 392, "lr": 1.0922949159388562e-05} +{"train_loss": 0.0174203272908926, "grad_norm": 0.07258794456720352, "global_step": 41635, "epoch": 392, "lr": 1.0921100270771134e-05} +{"train_loss": 0.013409807346761227, "grad_norm": 0.05728163570165634, "global_step": 41636, "epoch": 392, "lr": 1.0919251519459967e-05} +{"train_loss": 0.014373427256941795, "grad_norm": 0.08150090277194977, "global_step": 41637, "epoch": 392, "lr": 1.0917402905461533e-05} +{"train_loss": 0.015981875360012054, "grad_norm": 0.07572982460260391, "global_step": 41638, "epoch": 392, "lr": 1.0915554428782354e-05} +{"train_loss": 0.01373794861137867, "grad_norm": 0.09472698718309402, "global_step": 41639, "epoch": 392, "lr": 1.0913706089428932e-05} +{"train_loss": 0.015898849815130234, "grad_norm": 0.08579018712043762, "global_step": 41640, "epoch": 392, "lr": 1.0911857887407734e-05} +{"train_loss": 0.0131490807980299, "grad_norm": 0.09960861504077911, "global_step": 41641, "epoch": 392, "lr": 1.091000982272528e-05} +{"train_loss": 0.01444107573479414, "grad_norm": 0.10408688336610794, "global_step": 41642, "epoch": 392, "lr": 1.0908161895388042e-05} +{"train_loss": 0.016723832115530968, "grad_norm": 0.058901622891426086, "global_step": 41643, "epoch": 392, "lr": 1.0906314105402538e-05} +{"train_loss": 0.014799372293055058, "grad_norm": 0.06891930848360062, "global_step": 41644, "epoch": 392, "lr": 1.0904466452775225e-05} +{"train_loss": 0.015965204685926437, "grad_norm": 0.07501547038555145, "global_step": 41645, "epoch": 392, "lr": 1.0902618937512621e-05} +{"train_loss": 0.016109811142086983, "grad_norm": 0.09964010119438171, "global_step": 41646, "epoch": 392, "lr": 1.090077155962122e-05} +{"train_loss": 0.015748534351587296, "grad_norm": 0.08167126774787903, "global_step": 41647, "epoch": 392, "lr": 1.0898924319107507e-05} +{"train_loss": 0.015302912332117558, "grad_norm": 0.09895139932632446, "global_step": 41648, "epoch": 392, "lr": 1.0897077215977951e-05} +{"train_loss": 0.013073878362774849, "grad_norm": 0.09394090622663498, "global_step": 41649, "epoch": 392, "lr": 1.0895230250239075e-05} +{"train_loss": 0.01427236944437027, "grad_norm": 0.07596056163311005, "global_step": 41650, "epoch": 392, "lr": 1.0893383421897346e-05} +{"train_loss": 0.015481838956475258, "grad_norm": 0.060566846281290054, "global_step": 41651, "epoch": 392, "lr": 1.0891536730959256e-05} +{"train_loss": 0.015233457088470459, "grad_norm": 0.07782445847988129, "global_step": 41652, "epoch": 392, "lr": 1.0889690177431294e-05} +{"train_loss": 0.01558757945895195, "grad_norm": 0.11208260804414749, "global_step": 41653, "epoch": 392, "lr": 1.0887843761319988e-05} +{"train_loss": 0.015620686113834381, "grad_norm": 0.059381045401096344, "global_step": 41654, "epoch": 392, "lr": 1.088599748263176e-05} +{"train_loss": 0.014986002817749977, "grad_norm": 0.06048976257443428, "global_step": 41655, "epoch": 392, "lr": 1.0884151341373122e-05} +{"train_loss": 0.014496116898953915, "grad_norm": 0.056403614580631256, "global_step": 41656, "epoch": 392, "lr": 1.0882305337550581e-05} +{"train_loss": 0.015005801417776998, "grad_norm": 0.12419776618480682, "global_step": 41657, "epoch": 392, "lr": 1.0880459471170596e-05, "val_loss": 0.019109077751636505} +{"train_loss": 0.01517091877758503, "grad_norm": 0.12716592848300934, "global_step": 41658, "epoch": 393, "lr": 1.0878613742239662e-05} +{"train_loss": 0.016882022842764854, "grad_norm": 0.08600974828004837, "global_step": 41659, "epoch": 393, "lr": 1.0876768150764277e-05} +{"train_loss": 0.014745410531759262, "grad_norm": 0.06466174870729446, "global_step": 41660, "epoch": 393, "lr": 1.0874922696750917e-05} +{"train_loss": 0.015971891582012177, "grad_norm": 0.14020812511444092, "global_step": 41661, "epoch": 393, "lr": 1.0873077380206054e-05} +{"train_loss": 0.014978373423218727, "grad_norm": 0.11573607474565506, "global_step": 41662, "epoch": 393, "lr": 1.0871232201136177e-05} +{"train_loss": 0.013177015818655491, "grad_norm": 0.0821278989315033, "global_step": 41663, "epoch": 393, "lr": 1.0869387159547784e-05} +{"train_loss": 0.017873749136924744, "grad_norm": 0.18256671726703644, "global_step": 41664, "epoch": 393, "lr": 1.0867542255447338e-05} +{"train_loss": 0.012093689292669296, "grad_norm": 0.05701465904712677, "global_step": 41665, "epoch": 393, "lr": 1.0865697488841342e-05} +{"train_loss": 0.014468035660684109, "grad_norm": 0.054341819137334824, "global_step": 41666, "epoch": 393, "lr": 1.0863852859736256e-05} +{"train_loss": 0.014785561710596085, "grad_norm": 0.08724291622638702, "global_step": 41667, "epoch": 393, "lr": 1.0862008368138582e-05} +{"train_loss": 0.013851378113031387, "grad_norm": 0.08674413710832596, "global_step": 41668, "epoch": 393, "lr": 1.0860164014054775e-05} +{"train_loss": 0.014105896465480328, "grad_norm": 0.06460542976856232, "global_step": 41669, "epoch": 393, "lr": 1.0858319797491328e-05} +{"train_loss": 0.014757685363292694, "grad_norm": 0.06252548843622208, "global_step": 41670, "epoch": 393, "lr": 1.085647571845474e-05} +{"train_loss": 0.017167706042528152, "grad_norm": 0.06979718059301376, "global_step": 41671, "epoch": 393, "lr": 1.0854631776951473e-05} +{"train_loss": 0.01342724822461605, "grad_norm": 0.09854109585285187, "global_step": 41672, "epoch": 393, "lr": 1.0852787972987993e-05} +{"train_loss": 0.016250021755695343, "grad_norm": 0.05330418050289154, "global_step": 41673, "epoch": 393, "lr": 1.0850944306570804e-05} +{"train_loss": 0.012332659214735031, "grad_norm": 0.056109294295310974, "global_step": 41674, "epoch": 393, "lr": 1.0849100777706361e-05} +{"train_loss": 0.015022344887256622, "grad_norm": 0.10195174068212509, "global_step": 41675, "epoch": 393, "lr": 1.084725738640115e-05} +{"train_loss": 0.015548794530332088, "grad_norm": 0.1443164199590683, "global_step": 41676, "epoch": 393, "lr": 1.0845414132661658e-05} +{"train_loss": 0.014259659685194492, "grad_norm": 0.0634625032544136, "global_step": 41677, "epoch": 393, "lr": 1.0843571016494359e-05} +{"train_loss": 0.01345841959118843, "grad_norm": 0.10173282772302628, "global_step": 41678, "epoch": 393, "lr": 1.0841728037905707e-05} +{"train_loss": 0.01602918654680252, "grad_norm": 0.07855147868394852, "global_step": 41679, "epoch": 393, "lr": 1.0839885196902194e-05} +{"train_loss": 0.013947195373475552, "grad_norm": 0.11422217637300491, "global_step": 41680, "epoch": 393, "lr": 1.0838042493490308e-05} +{"train_loss": 0.0135090546682477, "grad_norm": 0.07984347641468048, "global_step": 41681, "epoch": 393, "lr": 1.0836199927676493e-05} +{"train_loss": 0.01654895395040512, "grad_norm": 0.08622495085000992, "global_step": 41682, "epoch": 393, "lr": 1.0834357499467256e-05} +{"train_loss": 0.011749455705285072, "grad_norm": 0.07010050117969513, "global_step": 41683, "epoch": 393, "lr": 1.0832515208869037e-05} +{"train_loss": 0.016133850440382957, "grad_norm": 0.08039022982120514, "global_step": 41684, "epoch": 393, "lr": 1.083067305588834e-05} +{"train_loss": 0.01607714779675007, "grad_norm": 0.08794420212507248, "global_step": 41685, "epoch": 393, "lr": 1.0828831040531618e-05} +{"train_loss": 0.014946450479328632, "grad_norm": 0.05342395603656769, "global_step": 41686, "epoch": 393, "lr": 1.082698916280534e-05} +{"train_loss": 0.01746203377842903, "grad_norm": 0.06117171421647072, "global_step": 41687, "epoch": 393, "lr": 1.0825147422716003e-05} +{"train_loss": 0.015359137207269669, "grad_norm": 0.10972122102975845, "global_step": 41688, "epoch": 393, "lr": 1.0823305820270058e-05} +{"train_loss": 0.01712595298886299, "grad_norm": 0.06265376508235931, "global_step": 41689, "epoch": 393, "lr": 1.0821464355473971e-05} +{"train_loss": 0.01296914741396904, "grad_norm": 0.05518139526247978, "global_step": 41690, "epoch": 393, "lr": 1.081962302833423e-05} +{"train_loss": 0.014562293887138367, "grad_norm": 0.09049974381923676, "global_step": 41691, "epoch": 393, "lr": 1.0817781838857283e-05} +{"train_loss": 0.017745235934853554, "grad_norm": 0.21215498447418213, "global_step": 41692, "epoch": 393, "lr": 1.0815940787049605e-05} +{"train_loss": 0.01401710044592619, "grad_norm": 0.09361903369426727, "global_step": 41693, "epoch": 393, "lr": 1.0814099872917694e-05} +{"train_loss": 0.014495132490992546, "grad_norm": 0.07416346669197083, "global_step": 41694, "epoch": 393, "lr": 1.0812259096467985e-05} +{"train_loss": 0.011301619932055473, "grad_norm": 0.083537258207798, "global_step": 41695, "epoch": 393, "lr": 1.0810418457706945e-05} +{"train_loss": 0.013975546695291996, "grad_norm": 0.08351481705904007, "global_step": 41696, "epoch": 393, "lr": 1.0808577956641053e-05} +{"train_loss": 0.015163795091211796, "grad_norm": 0.14005905389785767, "global_step": 41697, "epoch": 393, "lr": 1.080673759327679e-05} +{"train_loss": 0.015048611909151077, "grad_norm": 0.0716935247182846, "global_step": 41698, "epoch": 393, "lr": 1.0804897367620587e-05} +{"train_loss": 0.014464248903095722, "grad_norm": 0.07240650057792664, "global_step": 41699, "epoch": 393, "lr": 1.080305727967893e-05} +{"train_loss": 0.015027882531285286, "grad_norm": 0.09066358208656311, "global_step": 41700, "epoch": 393, "lr": 1.0801217329458302e-05} +{"train_loss": 0.015221250243484974, "grad_norm": 0.07113946229219437, "global_step": 41701, "epoch": 393, "lr": 1.0799377516965142e-05} +{"train_loss": 0.013586251996457577, "grad_norm": 0.056975748389959335, "global_step": 41702, "epoch": 393, "lr": 1.0797537842205912e-05} +{"train_loss": 0.01382121630012989, "grad_norm": 0.05349007993936539, "global_step": 41703, "epoch": 393, "lr": 1.0795698305187086e-05} +{"train_loss": 0.01612185873091221, "grad_norm": 0.09055963903665543, "global_step": 41704, "epoch": 393, "lr": 1.0793858905915139e-05} +{"train_loss": 0.013917602598667145, "grad_norm": 0.05849558860063553, "global_step": 41705, "epoch": 393, "lr": 1.0792019644396511e-05} +{"train_loss": 0.018318718299269676, "grad_norm": 0.17759475111961365, "global_step": 41706, "epoch": 393, "lr": 1.0790180520637682e-05} +{"train_loss": 0.01518199685961008, "grad_norm": 0.06010035052895546, "global_step": 41707, "epoch": 393, "lr": 1.0788341534645096e-05} +{"train_loss": 0.015001997351646423, "grad_norm": 0.08287827670574188, "global_step": 41708, "epoch": 393, "lr": 1.0786502686425242e-05} +{"train_loss": 0.014096684753894806, "grad_norm": 0.10592298954725266, "global_step": 41709, "epoch": 393, "lr": 1.078466397598455e-05} +{"train_loss": 0.019303040578961372, "grad_norm": 0.1475204974412918, "global_step": 41710, "epoch": 393, "lr": 1.0782825403329488e-05} +{"train_loss": 0.013336609117686749, "grad_norm": 0.1043497696518898, "global_step": 41711, "epoch": 393, "lr": 1.0780986968466544e-05} +{"train_loss": 0.015721367672085762, "grad_norm": 0.08307164162397385, "global_step": 41712, "epoch": 393, "lr": 1.077914867140215e-05} +{"train_loss": 0.013953171670436859, "grad_norm": 0.07091581076383591, "global_step": 41713, "epoch": 393, "lr": 1.0777310512142762e-05} +{"train_loss": 0.013741627335548401, "grad_norm": 0.0819256380200386, "global_step": 41714, "epoch": 393, "lr": 1.077547249069486e-05} +{"train_loss": 0.014848023653030396, "grad_norm": 0.11045195162296295, "global_step": 41715, "epoch": 393, "lr": 1.0773634607064875e-05} +{"train_loss": 0.012694142758846283, "grad_norm": 0.07718035578727722, "global_step": 41716, "epoch": 393, "lr": 1.0771796861259275e-05} +{"train_loss": 0.018269646912813187, "grad_norm": 0.10359609872102737, "global_step": 41717, "epoch": 393, "lr": 1.076995925328454e-05} +{"train_loss": 0.013916232623159885, "grad_norm": 0.060522571206092834, "global_step": 41718, "epoch": 393, "lr": 1.0768121783147105e-05} +{"train_loss": 0.014068603515625, "grad_norm": 0.0762404352426529, "global_step": 41719, "epoch": 393, "lr": 1.0766284450853415e-05} +{"train_loss": 0.015701424330472946, "grad_norm": 0.0958082526922226, "global_step": 41720, "epoch": 393, "lr": 1.0764447256409938e-05} +{"train_loss": 0.014756472781300545, "grad_norm": 0.05795970559120178, "global_step": 41721, "epoch": 393, "lr": 1.0762610199823148e-05} +{"train_loss": 0.013598712161183357, "grad_norm": 0.07289115339517593, "global_step": 41722, "epoch": 393, "lr": 1.0760773281099462e-05} +{"train_loss": 0.014469882473349571, "grad_norm": 0.07084070891141891, "global_step": 41723, "epoch": 393, "lr": 1.0758936500245376e-05} +{"train_loss": 0.01509045995771885, "grad_norm": 0.08049877732992172, "global_step": 41724, "epoch": 393, "lr": 1.07570998572673e-05} +{"train_loss": 0.017347581684589386, "grad_norm": 0.11776228994131088, "global_step": 41725, "epoch": 393, "lr": 1.0755263352171724e-05} +{"train_loss": 0.017299102619290352, "grad_norm": 0.08626613020896912, "global_step": 41726, "epoch": 393, "lr": 1.0753426984965075e-05} +{"train_loss": 0.011829367838799953, "grad_norm": 0.09127553552389145, "global_step": 41727, "epoch": 393, "lr": 1.0751590755653818e-05} +{"train_loss": 0.014908203855156898, "grad_norm": 0.13532009720802307, "global_step": 41728, "epoch": 393, "lr": 1.074975466424441e-05} +{"train_loss": 0.01617809757590294, "grad_norm": 0.07661893218755722, "global_step": 41729, "epoch": 393, "lr": 1.0747918710743298e-05} +{"train_loss": 0.01623149961233139, "grad_norm": 0.06719903647899628, "global_step": 41730, "epoch": 393, "lr": 1.0746082895156912e-05} +{"train_loss": 0.01496883761137724, "grad_norm": 0.17013747990131378, "global_step": 41731, "epoch": 393, "lr": 1.0744247217491738e-05} +{"train_loss": 0.013276873156428337, "grad_norm": 0.08631718158721924, "global_step": 41732, "epoch": 393, "lr": 1.0742411677754188e-05} +{"train_loss": 0.012870092876255512, "grad_norm": 0.07307979464530945, "global_step": 41733, "epoch": 393, "lr": 1.0740576275950737e-05} +{"train_loss": 0.016142457723617554, "grad_norm": 0.06303779035806656, "global_step": 41734, "epoch": 393, "lr": 1.0738741012087838e-05} +{"train_loss": 0.01686178520321846, "grad_norm": 0.1784939467906952, "global_step": 41735, "epoch": 393, "lr": 1.0736905886171933e-05} +{"train_loss": 0.015148657374083996, "grad_norm": 0.08912607282400131, "global_step": 41736, "epoch": 393, "lr": 1.0735070898209442e-05} +{"train_loss": 0.016006793826818466, "grad_norm": 0.10010869801044464, "global_step": 41737, "epoch": 393, "lr": 1.0733236048206847e-05} +{"train_loss": 0.013840211555361748, "grad_norm": 0.09901918470859528, "global_step": 41738, "epoch": 393, "lr": 1.0731401336170593e-05} +{"train_loss": 0.015304718166589737, "grad_norm": 0.08137068152427673, "global_step": 41739, "epoch": 393, "lr": 1.0729566762107101e-05} +{"train_loss": 0.01661456748843193, "grad_norm": 0.08850380778312683, "global_step": 41740, "epoch": 393, "lr": 1.0727732326022837e-05} +{"train_loss": 0.013741509988904, "grad_norm": 0.10511116683483124, "global_step": 41741, "epoch": 393, "lr": 1.0725898027924253e-05} +{"train_loss": 0.015413926914334297, "grad_norm": 0.08250703662633896, "global_step": 41742, "epoch": 393, "lr": 1.0724063867817786e-05} +{"train_loss": 0.0147951515391469, "grad_norm": 0.0752897784113884, "global_step": 41743, "epoch": 393, "lr": 1.0722229845709863e-05} +{"train_loss": 0.01874222606420517, "grad_norm": 0.08021921664476395, "global_step": 41744, "epoch": 393, "lr": 1.0720395961606943e-05} +{"train_loss": 0.015901336446404457, "grad_norm": 0.06833220273256302, "global_step": 41745, "epoch": 393, "lr": 1.0718562215515487e-05} +{"train_loss": 0.017670929431915283, "grad_norm": 0.1339181512594223, "global_step": 41746, "epoch": 393, "lr": 1.07167286074419e-05} +{"train_loss": 0.016751114279031754, "grad_norm": 0.06184746325016022, "global_step": 41747, "epoch": 393, "lr": 1.0714895137392662e-05} +{"train_loss": 0.013362947851419449, "grad_norm": 0.05752601847052574, "global_step": 41748, "epoch": 393, "lr": 1.0713061805374197e-05} +{"train_loss": 0.016651228070259094, "grad_norm": 0.09373042732477188, "global_step": 41749, "epoch": 393, "lr": 1.0711228611392937e-05} +{"train_loss": 0.015833625569939613, "grad_norm": 0.07170625776052475, "global_step": 41750, "epoch": 393, "lr": 1.0709395555455326e-05} +{"train_loss": 0.012336920946836472, "grad_norm": 0.07691116631031036, "global_step": 41751, "epoch": 393, "lr": 1.070756263756783e-05} +{"train_loss": 0.015733692795038223, "grad_norm": 0.06362045556306839, "global_step": 41752, "epoch": 393, "lr": 1.0705729857736863e-05} +{"train_loss": 0.01418630313128233, "grad_norm": 0.08684194087982178, "global_step": 41753, "epoch": 393, "lr": 1.0703897215968878e-05} +{"train_loss": 0.015558598563075066, "grad_norm": 0.0651974081993103, "global_step": 41754, "epoch": 393, "lr": 1.0702064712270299e-05} +{"train_loss": 0.01603413000702858, "grad_norm": 0.11882118135690689, "global_step": 41755, "epoch": 393, "lr": 1.0700232346647587e-05} +{"train_loss": 0.013982947915792465, "grad_norm": 0.07988309115171432, "global_step": 41756, "epoch": 393, "lr": 1.0698400119107155e-05} +{"train_loss": 0.01353304460644722, "grad_norm": 0.11008754372596741, "global_step": 41757, "epoch": 393, "lr": 1.0696568029655451e-05} +{"train_loss": 0.015951549634337425, "grad_norm": 0.11571094393730164, "global_step": 41758, "epoch": 393, "lr": 1.0694736078298927e-05} +{"train_loss": 0.017309753224253654, "grad_norm": 0.06591247022151947, "global_step": 41759, "epoch": 393, "lr": 1.0692904265044012e-05} +{"train_loss": 0.013886417262256145, "grad_norm": 0.0603613555431366, "global_step": 41760, "epoch": 393, "lr": 1.0691072589897123e-05} +{"train_loss": 0.0129835344851017, "grad_norm": 0.13216783106327057, "global_step": 41761, "epoch": 393, "lr": 1.0689241052864702e-05} +{"train_loss": 0.013902699574828148, "grad_norm": 0.06829781830310822, "global_step": 41762, "epoch": 393, "lr": 1.0687409653953212e-05} +{"train_loss": 0.015012627766239192, "grad_norm": 0.09116482734680176, "global_step": 41763, "epoch": 393, "lr": 1.0685578393169055e-05, "val_loss": 0.01945364847779274} +{"train_loss": 0.012684821151196957, "grad_norm": 0.06451068073511124, "global_step": 41764, "epoch": 394, "lr": 1.0683747270518679e-05} +{"train_loss": 0.013627552427351475, "grad_norm": 0.06619902700185776, "global_step": 41765, "epoch": 394, "lr": 1.0681916286008526e-05} +{"train_loss": 0.017280127853155136, "grad_norm": 0.0974678248167038, "global_step": 41766, "epoch": 394, "lr": 1.0680085439645021e-05} +{"train_loss": 0.01546386443078518, "grad_norm": 0.08253291249275208, "global_step": 41767, "epoch": 394, "lr": 1.0678254731434583e-05} +{"train_loss": 0.016834210604429245, "grad_norm": 0.08339161425828934, "global_step": 41768, "epoch": 394, "lr": 1.0676424161383658e-05} +{"train_loss": 0.01603000983595848, "grad_norm": 0.18567104637622833, "global_step": 41769, "epoch": 394, "lr": 1.0674593729498695e-05} +{"train_loss": 0.01646355167031288, "grad_norm": 0.09113408625125885, "global_step": 41770, "epoch": 394, "lr": 1.0672763435786087e-05} +{"train_loss": 0.01702996902167797, "grad_norm": 0.14165621995925903, "global_step": 41771, "epoch": 394, "lr": 1.0670933280252304e-05} +{"train_loss": 0.013851776719093323, "grad_norm": 0.07195605337619781, "global_step": 41772, "epoch": 394, "lr": 1.066910326290375e-05} +{"train_loss": 0.014204783365130424, "grad_norm": 0.1172897070646286, "global_step": 41773, "epoch": 394, "lr": 1.0667273383746857e-05} +{"train_loss": 0.0150430453941226, "grad_norm": 0.061434704810380936, "global_step": 41774, "epoch": 394, "lr": 1.066544364278806e-05} +{"train_loss": 0.015061356127262115, "grad_norm": 0.06129660829901695, "global_step": 41775, "epoch": 394, "lr": 1.0663614040033798e-05} +{"train_loss": 0.013101925142109394, "grad_norm": 0.14543858170509338, "global_step": 41776, "epoch": 394, "lr": 1.0661784575490491e-05} +{"train_loss": 0.016009896993637085, "grad_norm": 0.08083022385835648, "global_step": 41777, "epoch": 394, "lr": 1.065995524916455e-05} +{"train_loss": 0.016502324491739273, "grad_norm": 0.08242670446634293, "global_step": 41778, "epoch": 394, "lr": 1.0658126061062417e-05} +{"train_loss": 0.01568831317126751, "grad_norm": 0.06407827138900757, "global_step": 41779, "epoch": 394, "lr": 1.0656297011190535e-05} +{"train_loss": 0.014422859996557236, "grad_norm": 0.09197726100683212, "global_step": 41780, "epoch": 394, "lr": 1.0654468099555304e-05} +{"train_loss": 0.01363650057464838, "grad_norm": 0.07598769664764404, "global_step": 41781, "epoch": 394, "lr": 1.0652639326163156e-05} +{"train_loss": 0.014413346536457539, "grad_norm": 0.0678255558013916, "global_step": 41782, "epoch": 394, "lr": 1.0650810691020536e-05} +{"train_loss": 0.0169994980096817, "grad_norm": 0.07484977692365646, "global_step": 41783, "epoch": 394, "lr": 1.0648982194133861e-05} +{"train_loss": 0.0123700350522995, "grad_norm": 0.05415109172463417, "global_step": 41784, "epoch": 394, "lr": 1.0647153835509527e-05} +{"train_loss": 0.013796774670481682, "grad_norm": 0.07162424921989441, "global_step": 41785, "epoch": 394, "lr": 1.0645325615153984e-05} +{"train_loss": 0.015438674949109554, "grad_norm": 0.09740695357322693, "global_step": 41786, "epoch": 394, "lr": 1.064349753307367e-05} +{"train_loss": 0.013737962581217289, "grad_norm": 0.05536436289548874, "global_step": 41787, "epoch": 394, "lr": 1.0641669589274972e-05} +{"train_loss": 0.017016304656863213, "grad_norm": 0.08149363845586777, "global_step": 41788, "epoch": 394, "lr": 1.0639841783764343e-05} +{"train_loss": 0.01621076464653015, "grad_norm": 0.10305929183959961, "global_step": 41789, "epoch": 394, "lr": 1.0638014116548195e-05} +{"train_loss": 0.013724349439144135, "grad_norm": 0.08687052875757217, "global_step": 41790, "epoch": 394, "lr": 1.0636186587632934e-05} +{"train_loss": 0.015559794381260872, "grad_norm": 0.10245266556739807, "global_step": 41791, "epoch": 394, "lr": 1.063435919702499e-05} +{"train_loss": 0.014029375277459621, "grad_norm": 0.058687176555395126, "global_step": 41792, "epoch": 394, "lr": 1.0632531944730807e-05} +{"train_loss": 0.013975275680422783, "grad_norm": 0.1505313515663147, "global_step": 41793, "epoch": 394, "lr": 1.0630704830756771e-05} +{"train_loss": 0.015841713175177574, "grad_norm": 0.08816593140363693, "global_step": 41794, "epoch": 394, "lr": 1.0628877855109331e-05} +{"train_loss": 0.014568532817065716, "grad_norm": 0.11855112016201019, "global_step": 41795, "epoch": 394, "lr": 1.0627051017794876e-05} +{"train_loss": 0.01521390862762928, "grad_norm": 0.12151726335287094, "global_step": 41796, "epoch": 394, "lr": 1.0625224318819859e-05} +{"train_loss": 0.016112325713038445, "grad_norm": 0.070731520652771, "global_step": 41797, "epoch": 394, "lr": 1.0623397758190662e-05} +{"train_loss": 0.016004052013158798, "grad_norm": 0.07646939158439636, "global_step": 41798, "epoch": 394, "lr": 1.0621571335913722e-05} +{"train_loss": 0.013425296172499657, "grad_norm": 0.07875878363847733, "global_step": 41799, "epoch": 394, "lr": 1.0619745051995472e-05} +{"train_loss": 0.015232282690703869, "grad_norm": 0.10024138540029526, "global_step": 41800, "epoch": 394, "lr": 1.061791890644231e-05} +{"train_loss": 0.015679046511650085, "grad_norm": 0.08026955276727676, "global_step": 41801, "epoch": 394, "lr": 1.0616092899260638e-05} +{"train_loss": 0.01342456229031086, "grad_norm": 0.13557523488998413, "global_step": 41802, "epoch": 394, "lr": 1.0614267030456893e-05} +{"train_loss": 0.015337714925408363, "grad_norm": 0.05390586704015732, "global_step": 41803, "epoch": 394, "lr": 1.0612441300037495e-05} +{"train_loss": 0.017351165413856506, "grad_norm": 0.07324574142694473, "global_step": 41804, "epoch": 394, "lr": 1.0610615708008836e-05} +{"train_loss": 0.013986563310027122, "grad_norm": 0.08452826738357544, "global_step": 41805, "epoch": 394, "lr": 1.0608790254377343e-05} +{"train_loss": 0.01416392158716917, "grad_norm": 0.10032685101032257, "global_step": 41806, "epoch": 394, "lr": 1.0606964939149456e-05} +{"train_loss": 0.015252779237926006, "grad_norm": 0.07617326825857162, "global_step": 41807, "epoch": 394, "lr": 1.060513976233154e-05} +{"train_loss": 0.012341480702161789, "grad_norm": 0.08424872905015945, "global_step": 41808, "epoch": 394, "lr": 1.0603314723930024e-05} +{"train_loss": 0.016615521162748337, "grad_norm": 0.08862902969121933, "global_step": 41809, "epoch": 394, "lr": 1.0601489823951349e-05} +{"train_loss": 0.014533762820065022, "grad_norm": 0.08353488147258759, "global_step": 41810, "epoch": 394, "lr": 1.0599665062401887e-05} +{"train_loss": 0.01312306895852089, "grad_norm": 0.05001920461654663, "global_step": 41811, "epoch": 394, "lr": 1.0597840439288064e-05} +{"train_loss": 0.013970973901450634, "grad_norm": 0.06451371312141418, "global_step": 41812, "epoch": 394, "lr": 1.0596015954616312e-05} +{"train_loss": 0.017043663188815117, "grad_norm": 0.10811343044042587, "global_step": 41813, "epoch": 394, "lr": 1.0594191608393022e-05} +{"train_loss": 0.014983979985117912, "grad_norm": 0.0990515723824501, "global_step": 41814, "epoch": 394, "lr": 1.0592367400624592e-05} +{"train_loss": 0.015277680940926075, "grad_norm": 0.06995468586683273, "global_step": 41815, "epoch": 394, "lr": 1.0590543331317443e-05} +{"train_loss": 0.015560883097350597, "grad_norm": 0.07993384450674057, "global_step": 41816, "epoch": 394, "lr": 1.0588719400478004e-05} +{"train_loss": 0.016079843044281006, "grad_norm": 0.06276963651180267, "global_step": 41817, "epoch": 394, "lr": 1.058689560811265e-05} +{"train_loss": 0.015535588376224041, "grad_norm": 0.0481313057243824, "global_step": 41818, "epoch": 394, "lr": 1.0585071954227815e-05} +{"train_loss": 0.014766894280910492, "grad_norm": 0.07410307228565216, "global_step": 41819, "epoch": 394, "lr": 1.0583248438829891e-05} +{"train_loss": 0.014287559315562248, "grad_norm": 0.11616717278957367, "global_step": 41820, "epoch": 394, "lr": 1.0581425061925293e-05} +{"train_loss": 0.01692967116832733, "grad_norm": 0.08776094019412994, "global_step": 41821, "epoch": 394, "lr": 1.057960182352042e-05} +{"train_loss": 0.01434384100139141, "grad_norm": 0.06598269939422607, "global_step": 41822, "epoch": 394, "lr": 1.0577778723621679e-05} +{"train_loss": 0.01512451283633709, "grad_norm": 0.0716763362288475, "global_step": 41823, "epoch": 394, "lr": 1.057595576223549e-05} +{"train_loss": 0.01608762890100479, "grad_norm": 0.13714253902435303, "global_step": 41824, "epoch": 394, "lr": 1.0574132939368253e-05} +{"train_loss": 0.012949882075190544, "grad_norm": 0.06923127919435501, "global_step": 41825, "epoch": 394, "lr": 1.0572310255026346e-05} +{"train_loss": 0.015298504382371902, "grad_norm": 0.16432765126228333, "global_step": 41826, "epoch": 394, "lr": 1.0570487709216204e-05} +{"train_loss": 0.014215873554348946, "grad_norm": 0.10600267350673676, "global_step": 41827, "epoch": 394, "lr": 1.0568665301944225e-05} +{"train_loss": 0.015158098191022873, "grad_norm": 0.08367403596639633, "global_step": 41828, "epoch": 394, "lr": 1.05668430332168e-05} +{"train_loss": 0.016346400603652, "grad_norm": 0.0667918398976326, "global_step": 41829, "epoch": 394, "lr": 1.056502090304035e-05} +{"train_loss": 0.01487815659493208, "grad_norm": 0.06647561490535736, "global_step": 41830, "epoch": 394, "lr": 1.0563198911421268e-05} +{"train_loss": 0.013217251747846603, "grad_norm": 0.06524772942066193, "global_step": 41831, "epoch": 394, "lr": 1.056137705836594e-05} +{"train_loss": 0.018103700131177902, "grad_norm": 0.1310909241437912, "global_step": 41832, "epoch": 394, "lr": 1.0559555343880784e-05} +{"train_loss": 0.014698874205350876, "grad_norm": 0.06257499009370804, "global_step": 41833, "epoch": 394, "lr": 1.055773376797221e-05} +{"train_loss": 0.016149431467056274, "grad_norm": 0.061237625777721405, "global_step": 41834, "epoch": 394, "lr": 1.0555912330646595e-05} +{"train_loss": 0.01588362827897072, "grad_norm": 0.06852496415376663, "global_step": 41835, "epoch": 394, "lr": 1.0554091031910362e-05} +{"train_loss": 0.013716744258999825, "grad_norm": 0.10556788742542267, "global_step": 41836, "epoch": 394, "lr": 1.0552269871769888e-05} +{"train_loss": 0.013671686872839928, "grad_norm": 0.07056879997253418, "global_step": 41837, "epoch": 394, "lr": 1.0550448850231592e-05} +{"train_loss": 0.013184264302253723, "grad_norm": 0.05988387018442154, "global_step": 41838, "epoch": 394, "lr": 1.0548627967301849e-05} +{"train_loss": 0.016056949272751808, "grad_norm": 0.059356238692998886, "global_step": 41839, "epoch": 394, "lr": 1.0546807222987071e-05} +{"train_loss": 0.013904199004173279, "grad_norm": 0.0611727312207222, "global_step": 41840, "epoch": 394, "lr": 1.0544986617293668e-05} +{"train_loss": 0.013708432205021381, "grad_norm": 0.08947301656007767, "global_step": 41841, "epoch": 394, "lr": 1.0543166150228023e-05} +{"train_loss": 0.017829740419983864, "grad_norm": 0.0628271996974945, "global_step": 41842, "epoch": 394, "lr": 1.0541345821796522e-05} +{"train_loss": 0.014014746993780136, "grad_norm": 0.06866522878408432, "global_step": 41843, "epoch": 394, "lr": 1.053952563200557e-05} +{"train_loss": 0.013715657405555248, "grad_norm": 0.1467449814081192, "global_step": 41844, "epoch": 394, "lr": 1.053770558086158e-05} +{"train_loss": 0.01447899080812931, "grad_norm": 0.1120845302939415, "global_step": 41845, "epoch": 394, "lr": 1.0535885668370914e-05} +{"train_loss": 0.01577448658645153, "grad_norm": 0.08918435126543045, "global_step": 41846, "epoch": 394, "lr": 1.0534065894539985e-05} +{"train_loss": 0.016592297703027725, "grad_norm": 0.11068326234817505, "global_step": 41847, "epoch": 394, "lr": 1.0532246259375212e-05} +{"train_loss": 0.015012077987194061, "grad_norm": 0.08265294134616852, "global_step": 41848, "epoch": 394, "lr": 1.0530426762882934e-05} +{"train_loss": 0.012283452786505222, "grad_norm": 0.08054977655410767, "global_step": 41849, "epoch": 394, "lr": 1.0528607405069574e-05} +{"train_loss": 0.012852062471210957, "grad_norm": 0.05270307511091232, "global_step": 41850, "epoch": 394, "lr": 1.0526788185941533e-05} +{"train_loss": 0.015253816731274128, "grad_norm": 0.08029663562774658, "global_step": 41851, "epoch": 394, "lr": 1.0524969105505177e-05} +{"train_loss": 0.014901398681104183, "grad_norm": 0.15890148282051086, "global_step": 41852, "epoch": 394, "lr": 1.0523150163766915e-05} +{"train_loss": 0.013386883772909641, "grad_norm": 0.09856344014406204, "global_step": 41853, "epoch": 394, "lr": 1.0521331360733145e-05} +{"train_loss": 0.015814121812582016, "grad_norm": 0.06945076584815979, "global_step": 41854, "epoch": 394, "lr": 1.051951269641025e-05} +{"train_loss": 0.01617247797548771, "grad_norm": 0.08614175766706467, "global_step": 41855, "epoch": 394, "lr": 1.0517694170804603e-05} +{"train_loss": 0.01619521528482437, "grad_norm": 0.06638047844171524, "global_step": 41856, "epoch": 394, "lr": 1.0515875783922613e-05} +{"train_loss": 0.01480894535779953, "grad_norm": 0.08932527899742126, "global_step": 41857, "epoch": 394, "lr": 1.0514057535770671e-05} +{"train_loss": 0.013495462015271187, "grad_norm": 0.07226377725601196, "global_step": 41858, "epoch": 394, "lr": 1.0512239426355153e-05} +{"train_loss": 0.01534067653119564, "grad_norm": 0.07618281245231628, "global_step": 41859, "epoch": 394, "lr": 1.0510421455682463e-05} +{"train_loss": 0.013467521406710148, "grad_norm": 0.0864691212773323, "global_step": 41860, "epoch": 394, "lr": 1.0508603623758962e-05} +{"train_loss": 0.014706152491271496, "grad_norm": 0.05799496918916702, "global_step": 41861, "epoch": 394, "lr": 1.0506785930591068e-05} +{"train_loss": 0.015678543597459793, "grad_norm": 0.0892530083656311, "global_step": 41862, "epoch": 394, "lr": 1.0504968376185143e-05} +{"train_loss": 0.01721709966659546, "grad_norm": 0.12085683643817902, "global_step": 41863, "epoch": 394, "lr": 1.0503150960547581e-05} +{"train_loss": 0.014031654223799706, "grad_norm": 0.14011560380458832, "global_step": 41864, "epoch": 394, "lr": 1.0501333683684784e-05} +{"train_loss": 0.015802819281816483, "grad_norm": 0.09231705963611603, "global_step": 41865, "epoch": 394, "lr": 1.0499516545603122e-05} +{"train_loss": 0.01561786886304617, "grad_norm": 0.10348619520664215, "global_step": 41866, "epoch": 394, "lr": 1.049769954630897e-05} +{"train_loss": 0.016665251925587654, "grad_norm": 0.14479704201221466, "global_step": 41867, "epoch": 394, "lr": 1.0495882685808716e-05} +{"train_loss": 0.012820353731513023, "grad_norm": 0.045412223786115646, "global_step": 41868, "epoch": 394, "lr": 1.0494065964108768e-05} +{"train_loss": 0.015021942459257706, "grad_norm": 0.10213616490364075, "global_step": 41869, "epoch": 394, "lr": 1.049224938121548e-05, "val_loss": 0.019373416900634766} +{"train_loss": 0.014423678629100323, "grad_norm": 0.07854960858821869, "global_step": 41870, "epoch": 395, "lr": 1.0490432937135241e-05} +{"train_loss": 0.014965645968914032, "grad_norm": 0.12446162104606628, "global_step": 41871, "epoch": 395, "lr": 1.048861663187447e-05} +{"train_loss": 0.013261253014206886, "grad_norm": 0.09768464416265488, "global_step": 41872, "epoch": 395, "lr": 1.048680046543949e-05} +{"train_loss": 0.013375429436564445, "grad_norm": 0.07270430773496628, "global_step": 41873, "epoch": 395, "lr": 1.0484984437836709e-05} +{"train_loss": 0.01490759663283825, "grad_norm": 0.06386936455965042, "global_step": 41874, "epoch": 395, "lr": 1.0483168549072519e-05} +{"train_loss": 0.01318553276360035, "grad_norm": 0.09246734529733658, "global_step": 41875, "epoch": 395, "lr": 1.0481352799153276e-05} +{"train_loss": 0.014164860360324383, "grad_norm": 0.09390612691640854, "global_step": 41876, "epoch": 395, "lr": 1.0479537188085392e-05} +{"train_loss": 0.014453173615038395, "grad_norm": 0.06582175940275192, "global_step": 41877, "epoch": 395, "lr": 1.047772171587521e-05} +{"train_loss": 0.016802947968244553, "grad_norm": 0.10282102972269058, "global_step": 41878, "epoch": 395, "lr": 1.0475906382529138e-05} +{"train_loss": 0.013708099722862244, "grad_norm": 0.060918278992176056, "global_step": 41879, "epoch": 395, "lr": 1.047409118805353e-05} +{"train_loss": 0.013988557271659374, "grad_norm": 0.08436114341020584, "global_step": 41880, "epoch": 395, "lr": 1.0472276132454778e-05} +{"train_loss": 0.014173166826367378, "grad_norm": 0.07451954483985901, "global_step": 41881, "epoch": 395, "lr": 1.0470461215739269e-05} +{"train_loss": 0.015766575932502747, "grad_norm": 0.05972423776984215, "global_step": 41882, "epoch": 395, "lr": 1.0468646437913372e-05} +{"train_loss": 0.013419313356280327, "grad_norm": 0.052540723234415054, "global_step": 41883, "epoch": 395, "lr": 1.0466831798983439e-05} +{"train_loss": 0.014280143193900585, "grad_norm": 0.05212351307272911, "global_step": 41884, "epoch": 395, "lr": 1.046501729895587e-05} +{"train_loss": 0.015182660892605782, "grad_norm": 0.0633438304066658, "global_step": 41885, "epoch": 395, "lr": 1.0463202937837052e-05} +{"train_loss": 0.01438823714852333, "grad_norm": 0.0740317776799202, "global_step": 41886, "epoch": 395, "lr": 1.0461388715633331e-05} +{"train_loss": 0.015249506570398808, "grad_norm": 0.06548789888620377, "global_step": 41887, "epoch": 395, "lr": 1.0459574632351093e-05} +{"train_loss": 0.01423637941479683, "grad_norm": 0.09924037009477615, "global_step": 41888, "epoch": 395, "lr": 1.0457760687996742e-05} +{"train_loss": 0.015638312324881554, "grad_norm": 0.06480707228183746, "global_step": 41889, "epoch": 395, "lr": 1.0455946882576601e-05} +{"train_loss": 0.016665367409586906, "grad_norm": 0.0999307632446289, "global_step": 41890, "epoch": 395, "lr": 1.0454133216097055e-05} +{"train_loss": 0.01714809611439705, "grad_norm": 0.08664604276418686, "global_step": 41891, "epoch": 395, "lr": 1.045231968856451e-05} +{"train_loss": 0.0153249092400074, "grad_norm": 0.05812197923660278, "global_step": 41892, "epoch": 395, "lr": 1.0450506299985296e-05} +{"train_loss": 0.015736958011984825, "grad_norm": 0.06166267767548561, "global_step": 41893, "epoch": 395, "lr": 1.044869305036581e-05} +{"train_loss": 0.012988703325390816, "grad_norm": 0.07923945784568787, "global_step": 41894, "epoch": 395, "lr": 1.0446879939712423e-05} +{"train_loss": 0.013979533687233925, "grad_norm": 0.0728207603096962, "global_step": 41895, "epoch": 395, "lr": 1.0445066968031503e-05} +{"train_loss": 0.014670291915535927, "grad_norm": 0.12502822279930115, "global_step": 41896, "epoch": 395, "lr": 1.04432541353294e-05} +{"train_loss": 0.015006659552454948, "grad_norm": 0.0914430171251297, "global_step": 41897, "epoch": 395, "lr": 1.0441441441612498e-05} +{"train_loss": 0.015693150460720062, "grad_norm": 0.07060395926237106, "global_step": 41898, "epoch": 395, "lr": 1.0439628886887181e-05} +{"train_loss": 0.013274867087602615, "grad_norm": 0.10843658447265625, "global_step": 41899, "epoch": 395, "lr": 1.0437816471159789e-05} +{"train_loss": 0.014000609517097473, "grad_norm": 0.08435725420713425, "global_step": 41900, "epoch": 395, "lr": 1.0436004194436722e-05} +{"train_loss": 0.01518496498465538, "grad_norm": 0.0999293252825737, "global_step": 41901, "epoch": 395, "lr": 1.0434192056724313e-05} +{"train_loss": 0.014922413975000381, "grad_norm": 0.09883289039134979, "global_step": 41902, "epoch": 395, "lr": 1.0432380058028962e-05} +{"train_loss": 0.016140157356858253, "grad_norm": 0.05621001869440079, "global_step": 41903, "epoch": 395, "lr": 1.043056819835701e-05} +{"train_loss": 0.016947012394666672, "grad_norm": 0.05954618752002716, "global_step": 41904, "epoch": 395, "lr": 1.042875647771483e-05} +{"train_loss": 0.013909872621297836, "grad_norm": 0.1003129705786705, "global_step": 41905, "epoch": 395, "lr": 1.0426944896108803e-05} +{"train_loss": 0.015280243009328842, "grad_norm": 0.1491643786430359, "global_step": 41906, "epoch": 395, "lr": 1.0425133453545282e-05} +{"train_loss": 0.017801016569137573, "grad_norm": 0.10608168691396713, "global_step": 41907, "epoch": 395, "lr": 1.0423322150030618e-05} +{"train_loss": 0.016799628734588623, "grad_norm": 0.09225114434957504, "global_step": 41908, "epoch": 395, "lr": 1.0421510985571204e-05} +{"train_loss": 0.014904720708727837, "grad_norm": 0.05855632945895195, "global_step": 41909, "epoch": 395, "lr": 1.0419699960173374e-05} +{"train_loss": 0.013988731428980827, "grad_norm": 0.1110031008720398, "global_step": 41910, "epoch": 395, "lr": 1.0417889073843501e-05} +{"train_loss": 0.014188081957399845, "grad_norm": 0.09724628925323486, "global_step": 41911, "epoch": 395, "lr": 1.041607832658797e-05} +{"train_loss": 0.013758374378085136, "grad_norm": 0.08542835712432861, "global_step": 41912, "epoch": 395, "lr": 1.0414267718413123e-05} +{"train_loss": 0.018143191933631897, "grad_norm": 0.07658831775188446, "global_step": 41913, "epoch": 395, "lr": 1.0412457249325313e-05} +{"train_loss": 0.014203927479684353, "grad_norm": 0.05904022604227066, "global_step": 41914, "epoch": 395, "lr": 1.0410646919330908e-05} +{"train_loss": 0.013687746599316597, "grad_norm": 0.056404538452625275, "global_step": 41915, "epoch": 395, "lr": 1.040883672843629e-05} +{"train_loss": 0.015202758833765984, "grad_norm": 0.075741708278656, "global_step": 41916, "epoch": 395, "lr": 1.0407026676647786e-05} +{"train_loss": 0.015433892607688904, "grad_norm": 0.0645458847284317, "global_step": 41917, "epoch": 395, "lr": 1.0405216763971775e-05} +{"train_loss": 0.016628148034214973, "grad_norm": 0.05568188056349754, "global_step": 41918, "epoch": 395, "lr": 1.0403406990414621e-05} +{"train_loss": 0.01591753587126732, "grad_norm": 0.07307613641023636, "global_step": 41919, "epoch": 395, "lr": 1.0401597355982678e-05} +{"train_loss": 0.015189571306109428, "grad_norm": 0.120132215321064, "global_step": 41920, "epoch": 395, "lr": 1.0399787860682291e-05} +{"train_loss": 0.013294847682118416, "grad_norm": 0.06841419637203217, "global_step": 41921, "epoch": 395, "lr": 1.0397978504519823e-05} +{"train_loss": 0.013485698029398918, "grad_norm": 0.0667383223772049, "global_step": 41922, "epoch": 395, "lr": 1.039616928750165e-05} +{"train_loss": 0.015017997473478317, "grad_norm": 0.06437040865421295, "global_step": 41923, "epoch": 395, "lr": 1.0394360209634102e-05} +{"train_loss": 0.014241905882954597, "grad_norm": 0.12949830293655396, "global_step": 41924, "epoch": 395, "lr": 1.0392551270923566e-05} +{"train_loss": 0.014795148745179176, "grad_norm": 0.13940854370594025, "global_step": 41925, "epoch": 395, "lr": 1.0390742471376363e-05} +{"train_loss": 0.015654556453227997, "grad_norm": 0.07312142848968506, "global_step": 41926, "epoch": 395, "lr": 1.0388933810998875e-05} +{"train_loss": 0.015521518886089325, "grad_norm": 0.07120824605226517, "global_step": 41927, "epoch": 395, "lr": 1.038712528979744e-05} +{"train_loss": 0.01510959304869175, "grad_norm": 0.09056197851896286, "global_step": 41928, "epoch": 395, "lr": 1.0385316907778414e-05} +{"train_loss": 0.015444350428879261, "grad_norm": 0.06463898718357086, "global_step": 41929, "epoch": 395, "lr": 1.0383508664948189e-05} +{"train_loss": 0.015940070152282715, "grad_norm": 0.13639970123767853, "global_step": 41930, "epoch": 395, "lr": 1.0381700561313056e-05} +{"train_loss": 0.015408634208142757, "grad_norm": 0.06740257143974304, "global_step": 41931, "epoch": 395, "lr": 1.0379892596879398e-05} +{"train_loss": 0.014754951000213623, "grad_norm": 0.07954246550798416, "global_step": 41932, "epoch": 395, "lr": 1.0378084771653579e-05} +{"train_loss": 0.015943802893161774, "grad_norm": 0.15522132813930511, "global_step": 41933, "epoch": 395, "lr": 1.0376277085641927e-05} +{"train_loss": 0.014765318483114243, "grad_norm": 0.07120981067419052, "global_step": 41934, "epoch": 395, "lr": 1.0374469538850806e-05} +{"train_loss": 0.015183906070888042, "grad_norm": 0.08486338704824448, "global_step": 41935, "epoch": 395, "lr": 1.0372662131286576e-05} +{"train_loss": 0.016393907368183136, "grad_norm": 0.05532696843147278, "global_step": 41936, "epoch": 395, "lr": 1.0370854862955581e-05} +{"train_loss": 0.015072263777256012, "grad_norm": 0.1257794350385666, "global_step": 41937, "epoch": 395, "lr": 1.0369047733864157e-05} +{"train_loss": 0.013886075466871262, "grad_norm": 0.1305178552865982, "global_step": 41938, "epoch": 395, "lr": 1.0367240744018658e-05} +{"train_loss": 0.012566182762384415, "grad_norm": 0.06289294362068176, "global_step": 41939, "epoch": 395, "lr": 1.0365433893425458e-05} +{"train_loss": 0.015589965507388115, "grad_norm": 0.08196380734443665, "global_step": 41940, "epoch": 395, "lr": 1.0363627182090868e-05} +{"train_loss": 0.015087896026670933, "grad_norm": 0.07494542747735977, "global_step": 41941, "epoch": 395, "lr": 1.0361820610021273e-05} +{"train_loss": 0.012316086329519749, "grad_norm": 0.06077340617775917, "global_step": 41942, "epoch": 395, "lr": 1.0360014177222988e-05} +{"train_loss": 0.015704892575740814, "grad_norm": 0.06162222847342491, "global_step": 41943, "epoch": 395, "lr": 1.0358207883702387e-05} +{"train_loss": 0.013012886978685856, "grad_norm": 0.08498861640691757, "global_step": 41944, "epoch": 395, "lr": 1.0356401729465786e-05} +{"train_loss": 0.01420237310230732, "grad_norm": 0.09692779183387756, "global_step": 41945, "epoch": 395, "lr": 1.0354595714519554e-05} +{"train_loss": 0.01657424122095108, "grad_norm": 0.07605132460594177, "global_step": 41946, "epoch": 395, "lr": 1.0352789838870047e-05} +{"train_loss": 0.014964563772082329, "grad_norm": 0.08540506660938263, "global_step": 41947, "epoch": 395, "lr": 1.0350984102523586e-05} +{"train_loss": 0.015801629051566124, "grad_norm": 0.12180210649967194, "global_step": 41948, "epoch": 395, "lr": 1.0349178505486517e-05} +{"train_loss": 0.01800105720758438, "grad_norm": 0.11923687160015106, "global_step": 41949, "epoch": 395, "lr": 1.0347373047765202e-05} +{"train_loss": 0.013822815380990505, "grad_norm": 0.05377107113599777, "global_step": 41950, "epoch": 395, "lr": 1.0345567729365962e-05} +{"train_loss": 0.016477834433317184, "grad_norm": 0.09845321625471115, "global_step": 41951, "epoch": 395, "lr": 1.034376255029515e-05} +{"train_loss": 0.014426584355533123, "grad_norm": 0.10830248892307281, "global_step": 41952, "epoch": 395, "lr": 1.0341957510559125e-05} +{"train_loss": 0.017055107280611992, "grad_norm": 0.06181182712316513, "global_step": 41953, "epoch": 395, "lr": 1.0340152610164217e-05} +{"train_loss": 0.01636739820241928, "grad_norm": 0.11878238618373871, "global_step": 41954, "epoch": 395, "lr": 1.0338347849116747e-05} +{"train_loss": 0.014931149780750275, "grad_norm": 0.07676935195922852, "global_step": 41955, "epoch": 395, "lr": 1.0336543227423074e-05} +{"train_loss": 0.01495555229485035, "grad_norm": 0.06564250588417053, "global_step": 41956, "epoch": 395, "lr": 1.0334738745089556e-05} +{"train_loss": 0.012393995188176632, "grad_norm": 0.1005324050784111, "global_step": 41957, "epoch": 395, "lr": 1.0332934402122502e-05} +{"train_loss": 0.016732243821024895, "grad_norm": 0.0966903418302536, "global_step": 41958, "epoch": 395, "lr": 1.0331130198528265e-05} +{"train_loss": 0.015033634379506111, "grad_norm": 0.0887736901640892, "global_step": 41959, "epoch": 395, "lr": 1.03293261343132e-05} +{"train_loss": 0.013910754583775997, "grad_norm": 0.12284395098686218, "global_step": 41960, "epoch": 395, "lr": 1.0327522209483626e-05} +{"train_loss": 0.0142722949385643, "grad_norm": 0.08299306035041809, "global_step": 41961, "epoch": 395, "lr": 1.0325718424045877e-05} +{"train_loss": 0.014845858328044415, "grad_norm": 0.08784262835979462, "global_step": 41962, "epoch": 395, "lr": 1.0323914778006294e-05} +{"train_loss": 0.013482004404067993, "grad_norm": 0.09485027939081192, "global_step": 41963, "epoch": 395, "lr": 1.0322111271371237e-05} +{"train_loss": 0.015555515885353088, "grad_norm": 0.106033556163311, "global_step": 41964, "epoch": 395, "lr": 1.0320307904147014e-05} +{"train_loss": 0.013301372528076172, "grad_norm": 0.07723990827798843, "global_step": 41965, "epoch": 395, "lr": 1.0318504676339985e-05} +{"train_loss": 0.014990422874689102, "grad_norm": 0.07401793450117111, "global_step": 41966, "epoch": 395, "lr": 1.0316701587956474e-05} +{"train_loss": 0.0147250946611166, "grad_norm": 0.08570019155740738, "global_step": 41967, "epoch": 395, "lr": 1.0314898639002802e-05} +{"train_loss": 0.016358882188796997, "grad_norm": 0.06704911589622498, "global_step": 41968, "epoch": 395, "lr": 1.0313095829485314e-05} +{"train_loss": 0.017170194536447525, "grad_norm": 0.10574411600828171, "global_step": 41969, "epoch": 395, "lr": 1.0311293159410352e-05} +{"train_loss": 0.014721320010721684, "grad_norm": 0.06625235825777054, "global_step": 41970, "epoch": 395, "lr": 1.0309490628784257e-05} +{"train_loss": 0.015981052070856094, "grad_norm": 0.05172714963555336, "global_step": 41971, "epoch": 395, "lr": 1.030768823761335e-05} +{"train_loss": 0.014923356473445892, "grad_norm": 0.09774205088615417, "global_step": 41972, "epoch": 395, "lr": 1.0305885985903956e-05} +{"train_loss": 0.015159685164690018, "grad_norm": 0.07188955694437027, "global_step": 41973, "epoch": 395, "lr": 1.0304083873662423e-05} +{"train_loss": 0.019181273877620697, "grad_norm": 0.0726391077041626, "global_step": 41974, "epoch": 395, "lr": 1.0302281900895072e-05} +{"train_loss": 0.014998419544185107, "grad_norm": 0.06748586893081665, "global_step": 41975, "epoch": 395, "lr": 1.0300480067608231e-05, "val_loss": 0.019775371998548508} +{"train_loss": 0.01408058125525713, "grad_norm": 0.12982429563999176, "global_step": 41976, "epoch": 396, "lr": 1.0298678373808252e-05} +{"train_loss": 0.015852537006139755, "grad_norm": 0.15575644373893738, "global_step": 41977, "epoch": 396, "lr": 1.0296876819501455e-05} +{"train_loss": 0.019078031182289124, "grad_norm": 0.08117405325174332, "global_step": 41978, "epoch": 396, "lr": 1.0295075404694149e-05} +{"train_loss": 0.014131826348602772, "grad_norm": 0.12909553945064545, "global_step": 41979, "epoch": 396, "lr": 1.0293274129392683e-05} +{"train_loss": 0.01630270853638649, "grad_norm": 0.11992529779672623, "global_step": 41980, "epoch": 396, "lr": 1.0291472993603396e-05} +{"train_loss": 0.014735283330082893, "grad_norm": 0.06994836032390594, "global_step": 41981, "epoch": 396, "lr": 1.0289671997332589e-05} +{"train_loss": 0.012742138467729092, "grad_norm": 0.08990536630153656, "global_step": 41982, "epoch": 396, "lr": 1.0287871140586625e-05} +{"train_loss": 0.015036729164421558, "grad_norm": 0.08194807916879654, "global_step": 41983, "epoch": 396, "lr": 1.0286070423371791e-05} +{"train_loss": 0.014238719828426838, "grad_norm": 0.12918579578399658, "global_step": 41984, "epoch": 396, "lr": 1.028426984569445e-05} +{"train_loss": 0.015023864805698395, "grad_norm": 0.08109769225120544, "global_step": 41985, "epoch": 396, "lr": 1.0282469407560903e-05} +{"train_loss": 0.016866080462932587, "grad_norm": 0.10964155197143555, "global_step": 41986, "epoch": 396, "lr": 1.028066910897748e-05} +{"train_loss": 0.013513391837477684, "grad_norm": 0.06385276466608047, "global_step": 41987, "epoch": 396, "lr": 1.0278868949950532e-05} +{"train_loss": 0.01511632651090622, "grad_norm": 0.10808200389146805, "global_step": 41988, "epoch": 396, "lr": 1.0277068930486361e-05} +{"train_loss": 0.01492839865386486, "grad_norm": 0.11388526856899261, "global_step": 41989, "epoch": 396, "lr": 1.0275269050591279e-05} +{"train_loss": 0.011872144415974617, "grad_norm": 0.13976266980171204, "global_step": 41990, "epoch": 396, "lr": 1.0273469310271644e-05} +{"train_loss": 0.01564226672053337, "grad_norm": 0.07184363156557083, "global_step": 41991, "epoch": 396, "lr": 1.0271669709533744e-05} +{"train_loss": 0.01579907350242138, "grad_norm": 0.06529726833105087, "global_step": 41992, "epoch": 396, "lr": 1.0269870248383917e-05} +{"train_loss": 0.016271930187940598, "grad_norm": 0.10234740376472473, "global_step": 41993, "epoch": 396, "lr": 1.0268070926828505e-05} +{"train_loss": 0.01408868283033371, "grad_norm": 0.1491188406944275, "global_step": 41994, "epoch": 396, "lr": 1.0266271744873812e-05} +{"train_loss": 0.011234253644943237, "grad_norm": 0.0744844600558281, "global_step": 41995, "epoch": 396, "lr": 1.0264472702526151e-05} +{"train_loss": 0.017044173553586006, "grad_norm": 0.07687967270612717, "global_step": 41996, "epoch": 396, "lr": 1.026267379979185e-05} +{"train_loss": 0.014944951981306076, "grad_norm": 0.07042884826660156, "global_step": 41997, "epoch": 396, "lr": 1.0260875036677242e-05} +{"train_loss": 0.015455661341547966, "grad_norm": 0.13496577739715576, "global_step": 41998, "epoch": 396, "lr": 1.0259076413188624e-05} +{"train_loss": 0.015385458245873451, "grad_norm": 0.0646277442574501, "global_step": 41999, "epoch": 396, "lr": 1.0257277929332332e-05} +{"train_loss": 0.013281945139169693, "grad_norm": 0.06180308014154434, "global_step": 42000, "epoch": 396, "lr": 1.0255479585114691e-05} +{"train_loss": 0.014627460390329361, "grad_norm": 0.06617745012044907, "global_step": 42001, "epoch": 396, "lr": 1.0253681380542013e-05} +{"train_loss": 0.013582659885287285, "grad_norm": 0.08325990289449692, "global_step": 42002, "epoch": 396, "lr": 1.0251883315620603e-05} +{"train_loss": 0.01342247985303402, "grad_norm": 0.05568784847855568, "global_step": 42003, "epoch": 396, "lr": 1.0250085390356785e-05} +{"train_loss": 0.01827431097626686, "grad_norm": 0.07290420681238174, "global_step": 42004, "epoch": 396, "lr": 1.0248287604756895e-05} +{"train_loss": 0.015458737500011921, "grad_norm": 0.061331503093242645, "global_step": 42005, "epoch": 396, "lr": 1.0246489958827221e-05} +{"train_loss": 0.013816741295158863, "grad_norm": 0.06599988043308258, "global_step": 42006, "epoch": 396, "lr": 1.0244692452574107e-05} +{"train_loss": 0.01740313693881035, "grad_norm": 0.0791100561618805, "global_step": 42007, "epoch": 396, "lr": 1.024289508600385e-05} +{"train_loss": 0.014642816036939621, "grad_norm": 0.06212037801742554, "global_step": 42008, "epoch": 396, "lr": 1.0241097859122761e-05} +{"train_loss": 0.013973908498883247, "grad_norm": 0.06852451711893082, "global_step": 42009, "epoch": 396, "lr": 1.0239300771937166e-05} +{"train_loss": 0.014739113859832287, "grad_norm": 0.10283531248569489, "global_step": 42010, "epoch": 396, "lr": 1.0237503824453381e-05} +{"train_loss": 0.012132180854678154, "grad_norm": 0.07369748502969742, "global_step": 42011, "epoch": 396, "lr": 1.023570701667771e-05} +{"train_loss": 0.011862112209200859, "grad_norm": 0.08696470409631729, "global_step": 42012, "epoch": 396, "lr": 1.0233910348616482e-05} +{"train_loss": 0.014489477500319481, "grad_norm": 0.0910165086388588, "global_step": 42013, "epoch": 396, "lr": 1.0232113820275985e-05} +{"train_loss": 0.0147347217425704, "grad_norm": 0.0608036182820797, "global_step": 42014, "epoch": 396, "lr": 1.0230317431662561e-05} +{"train_loss": 0.014011207967996597, "grad_norm": 0.0545317679643631, "global_step": 42015, "epoch": 396, "lr": 1.0228521182782491e-05} +{"train_loss": 0.015486270189285278, "grad_norm": 0.08647053688764572, "global_step": 42016, "epoch": 396, "lr": 1.0226725073642101e-05} +{"train_loss": 0.015397516079246998, "grad_norm": 0.1552211195230484, "global_step": 42017, "epoch": 396, "lr": 1.0224929104247721e-05} +{"train_loss": 0.016330812126398087, "grad_norm": 0.06433898955583572, "global_step": 42018, "epoch": 396, "lr": 1.0223133274605634e-05} +{"train_loss": 0.01682904176414013, "grad_norm": 0.07087858766317368, "global_step": 42019, "epoch": 396, "lr": 1.0221337584722152e-05} +{"train_loss": 0.017703630030155182, "grad_norm": 0.06750411540269852, "global_step": 42020, "epoch": 396, "lr": 1.0219542034603585e-05} +{"train_loss": 0.015286928974092007, "grad_norm": 0.09187904000282288, "global_step": 42021, "epoch": 396, "lr": 1.0217746624256263e-05} +{"train_loss": 0.0133502297103405, "grad_norm": 0.05722557380795479, "global_step": 42022, "epoch": 396, "lr": 1.0215951353686465e-05} +{"train_loss": 0.013749298639595509, "grad_norm": 0.07808932662010193, "global_step": 42023, "epoch": 396, "lr": 1.0214156222900512e-05} +{"train_loss": 0.014265100471675396, "grad_norm": 0.10782545804977417, "global_step": 42024, "epoch": 396, "lr": 1.0212361231904726e-05} +{"train_loss": 0.015055431984364986, "grad_norm": 0.1503298431634903, "global_step": 42025, "epoch": 396, "lr": 1.02105663807054e-05} +{"train_loss": 0.014570718631148338, "grad_norm": 0.1311711221933365, "global_step": 42026, "epoch": 396, "lr": 1.0208771669308825e-05} +{"train_loss": 0.014061953872442245, "grad_norm": 0.0911780595779419, "global_step": 42027, "epoch": 396, "lr": 1.0206977097721321e-05} +{"train_loss": 0.013618217781186104, "grad_norm": 0.09677937626838684, "global_step": 42028, "epoch": 396, "lr": 1.0205182665949208e-05} +{"train_loss": 0.016282469034194946, "grad_norm": 0.09300314635038376, "global_step": 42029, "epoch": 396, "lr": 1.0203388373998779e-05} +{"train_loss": 0.015523570589721203, "grad_norm": 0.09325090050697327, "global_step": 42030, "epoch": 396, "lr": 1.020159422187632e-05} +{"train_loss": 0.015483543276786804, "grad_norm": 0.051255036145448685, "global_step": 42031, "epoch": 396, "lr": 1.0199800209588167e-05} +{"train_loss": 0.014679290354251862, "grad_norm": 0.08234387636184692, "global_step": 42032, "epoch": 396, "lr": 1.0198006337140591e-05} +{"train_loss": 0.014735372737050056, "grad_norm": 0.0706593245267868, "global_step": 42033, "epoch": 396, "lr": 1.0196212604539912e-05} +{"train_loss": 0.012136486358940601, "grad_norm": 0.06590969115495682, "global_step": 42034, "epoch": 396, "lr": 1.0194419011792444e-05} +{"train_loss": 0.014259561896324158, "grad_norm": 0.05989997461438179, "global_step": 42035, "epoch": 396, "lr": 1.0192625558904478e-05} +{"train_loss": 0.016204189509153366, "grad_norm": 0.07407081872224808, "global_step": 42036, "epoch": 396, "lr": 1.0190832245882304e-05} +{"train_loss": 0.015416156500577927, "grad_norm": 0.06912101805210114, "global_step": 42037, "epoch": 396, "lr": 1.0189039072732226e-05} +{"train_loss": 0.018005024641752243, "grad_norm": 0.06457982212305069, "global_step": 42038, "epoch": 396, "lr": 1.0187246039460569e-05} +{"train_loss": 0.015291661024093628, "grad_norm": 0.08069486171007156, "global_step": 42039, "epoch": 396, "lr": 1.0185453146073604e-05} +{"train_loss": 0.012298684567213058, "grad_norm": 0.07178369164466858, "global_step": 42040, "epoch": 396, "lr": 1.0183660392577637e-05} +{"train_loss": 0.017173487693071365, "grad_norm": 0.07033392041921616, "global_step": 42041, "epoch": 396, "lr": 1.0181867778978988e-05} +{"train_loss": 0.01524255983531475, "grad_norm": 0.06712671369314194, "global_step": 42042, "epoch": 396, "lr": 1.0180075305283936e-05} +{"train_loss": 0.014941063709557056, "grad_norm": 0.06254581362009048, "global_step": 42043, "epoch": 396, "lr": 1.0178282971498775e-05} +{"train_loss": 0.016763074323534966, "grad_norm": 0.058251816779375076, "global_step": 42044, "epoch": 396, "lr": 1.0176490777629804e-05} +{"train_loss": 0.01612941175699234, "grad_norm": 0.12139204144477844, "global_step": 42045, "epoch": 396, "lr": 1.017469872368334e-05} +{"train_loss": 0.015572819858789444, "grad_norm": 0.06663641333580017, "global_step": 42046, "epoch": 396, "lr": 1.0172906809665655e-05} +{"train_loss": 0.016081804409623146, "grad_norm": 0.08300993591547012, "global_step": 42047, "epoch": 396, "lr": 1.0171115035583067e-05} +{"train_loss": 0.011685805395245552, "grad_norm": 0.08764351904392242, "global_step": 42048, "epoch": 396, "lr": 1.016932340144186e-05} +{"train_loss": 0.01634201407432556, "grad_norm": 0.07836242765188217, "global_step": 42049, "epoch": 396, "lr": 1.0167531907248311e-05} +{"train_loss": 0.015002432279288769, "grad_norm": 0.0994872897863388, "global_step": 42050, "epoch": 396, "lr": 1.0165740553008735e-05} +{"train_loss": 0.01580119878053665, "grad_norm": 0.06066840887069702, "global_step": 42051, "epoch": 396, "lr": 1.016394933872944e-05} +{"train_loss": 0.01574660651385784, "grad_norm": 0.06119230017066002, "global_step": 42052, "epoch": 396, "lr": 1.0162158264416682e-05} +{"train_loss": 0.015913214534521103, "grad_norm": 0.05228849872946739, "global_step": 42053, "epoch": 396, "lr": 1.0160367330076793e-05} +{"train_loss": 0.01751997321844101, "grad_norm": 0.08410541713237762, "global_step": 42054, "epoch": 396, "lr": 1.0158576535716036e-05} +{"train_loss": 0.015236645936965942, "grad_norm": 0.06010330840945244, "global_step": 42055, "epoch": 396, "lr": 1.0156785881340724e-05} +{"train_loss": 0.016589976847171783, "grad_norm": 0.08011113852262497, "global_step": 42056, "epoch": 396, "lr": 1.0154995366957126e-05} +{"train_loss": 0.013327660970389843, "grad_norm": 0.0753316730260849, "global_step": 42057, "epoch": 396, "lr": 1.0153204992571546e-05} +{"train_loss": 0.014350077137351036, "grad_norm": 0.05828772112727165, "global_step": 42058, "epoch": 396, "lr": 1.0151414758190291e-05} +{"train_loss": 0.014643153175711632, "grad_norm": 0.08320604264736176, "global_step": 42059, "epoch": 396, "lr": 1.0149624663819635e-05} +{"train_loss": 0.014187321066856384, "grad_norm": 0.08275492489337921, "global_step": 42060, "epoch": 396, "lr": 1.0147834709465853e-05} +{"train_loss": 0.015575550496578217, "grad_norm": 0.06812511384487152, "global_step": 42061, "epoch": 396, "lr": 1.0146044895135248e-05} +{"train_loss": 0.014629105105996132, "grad_norm": 0.08461419492959976, "global_step": 42062, "epoch": 396, "lr": 1.0144255220834121e-05} +{"train_loss": 0.015219058841466904, "grad_norm": 0.08843192458152771, "global_step": 42063, "epoch": 396, "lr": 1.0142465686568736e-05} +{"train_loss": 0.013720419257879257, "grad_norm": 0.0562613271176815, "global_step": 42064, "epoch": 396, "lr": 1.014067629234539e-05} +{"train_loss": 0.013745147734880447, "grad_norm": 0.08058453351259232, "global_step": 42065, "epoch": 396, "lr": 1.0138887038170408e-05} +{"train_loss": 0.01382363960146904, "grad_norm": 0.08081167936325073, "global_step": 42066, "epoch": 396, "lr": 1.013709792405001e-05} +{"train_loss": 0.015911396592855453, "grad_norm": 0.07596974819898605, "global_step": 42067, "epoch": 396, "lr": 1.0135308949990513e-05} +{"train_loss": 0.014569958671927452, "grad_norm": 0.11287985742092133, "global_step": 42068, "epoch": 396, "lr": 1.0133520115998218e-05} +{"train_loss": 0.015560770407319069, "grad_norm": 0.05909503623843193, "global_step": 42069, "epoch": 396, "lr": 1.0131731422079382e-05} +{"train_loss": 0.014187279157340527, "grad_norm": 0.06575724482536316, "global_step": 42070, "epoch": 396, "lr": 1.0129942868240305e-05} +{"train_loss": 0.015654247254133224, "grad_norm": 0.06191415712237358, "global_step": 42071, "epoch": 396, "lr": 1.0128154454487281e-05} +{"train_loss": 0.01585240662097931, "grad_norm": 0.060735072940588, "global_step": 42072, "epoch": 396, "lr": 1.0126366180826585e-05} +{"train_loss": 0.013296311721205711, "grad_norm": 0.08669164031744003, "global_step": 42073, "epoch": 396, "lr": 1.0124578047264482e-05} +{"train_loss": 0.014345590025186539, "grad_norm": 0.07311896979808807, "global_step": 42074, "epoch": 396, "lr": 1.0122790053807273e-05} +{"train_loss": 0.017706435173749924, "grad_norm": 0.06834983825683594, "global_step": 42075, "epoch": 396, "lr": 1.0121002200461249e-05} +{"train_loss": 0.014313227497041225, "grad_norm": 0.06830243021249771, "global_step": 42076, "epoch": 396, "lr": 1.0119214487232682e-05} +{"train_loss": 0.015664052218198776, "grad_norm": 0.06205669417977333, "global_step": 42077, "epoch": 396, "lr": 1.0117426914127837e-05} +{"train_loss": 0.01746147871017456, "grad_norm": 0.11889111995697021, "global_step": 42078, "epoch": 396, "lr": 1.011563948115301e-05} +{"train_loss": 0.015854522585868835, "grad_norm": 0.0898929312825203, "global_step": 42079, "epoch": 396, "lr": 1.0113852188314493e-05} +{"train_loss": 0.01531876903027296, "grad_norm": 0.07966238260269165, "global_step": 42080, "epoch": 396, "lr": 1.011206503561854e-05} +{"train_loss": 0.014995126449063702, "grad_norm": 0.07167913019657135, "global_step": 42081, "epoch": 396, "lr": 1.0110278023071446e-05, "val_loss": 0.020315878093242645} +{"train_loss": 0.013328228145837784, "grad_norm": 0.05445576459169388, "global_step": 42082, "epoch": 397, "lr": 1.01084911506795e-05} +{"train_loss": 0.014059026725590229, "grad_norm": 0.06860500574111938, "global_step": 42083, "epoch": 397, "lr": 1.0106704418448964e-05} +{"train_loss": 0.01470506563782692, "grad_norm": 0.05910134315490723, "global_step": 42084, "epoch": 397, "lr": 1.0104917826386106e-05} +{"train_loss": 0.016013724729418755, "grad_norm": 0.07051901519298553, "global_step": 42085, "epoch": 397, "lr": 1.0103131374497221e-05} +{"train_loss": 0.01510668359696865, "grad_norm": 0.11264780908823013, "global_step": 42086, "epoch": 397, "lr": 1.0101345062788598e-05} +{"train_loss": 0.016328932717442513, "grad_norm": 0.10685130208730698, "global_step": 42087, "epoch": 397, "lr": 1.0099558891266476e-05} +{"train_loss": 0.014695127494633198, "grad_norm": 0.08478939533233643, "global_step": 42088, "epoch": 397, "lr": 1.009777285993717e-05} +{"train_loss": 0.0142205860465765, "grad_norm": 0.06761468201875687, "global_step": 42089, "epoch": 397, "lr": 1.0095986968806937e-05} +{"train_loss": 0.016661737114191055, "grad_norm": 0.15414442121982574, "global_step": 42090, "epoch": 397, "lr": 1.0094201217882038e-05} +{"train_loss": 0.015175591222941875, "grad_norm": 0.05814836546778679, "global_step": 42091, "epoch": 397, "lr": 1.0092415607168759e-05} +{"train_loss": 0.014871561899781227, "grad_norm": 0.09844666719436646, "global_step": 42092, "epoch": 397, "lr": 1.009063013667339e-05} +{"train_loss": 0.014724396169185638, "grad_norm": 0.06897471845149994, "global_step": 42093, "epoch": 397, "lr": 1.008884480640218e-05} +{"train_loss": 0.01486910879611969, "grad_norm": 0.059056881815195084, "global_step": 42094, "epoch": 397, "lr": 1.0087059616361422e-05} +{"train_loss": 0.017113886773586273, "grad_norm": 0.10605936497449875, "global_step": 42095, "epoch": 397, "lr": 1.0085274566557367e-05} +{"train_loss": 0.014243071898818016, "grad_norm": 0.08123686909675598, "global_step": 42096, "epoch": 397, "lr": 1.0083489656996314e-05} +{"train_loss": 0.01559881679713726, "grad_norm": 0.145954892039299, "global_step": 42097, "epoch": 397, "lr": 1.0081704887684506e-05} +{"train_loss": 0.014785198494791985, "grad_norm": 0.07117082923650742, "global_step": 42098, "epoch": 397, "lr": 1.0079920258628223e-05} +{"train_loss": 0.014533264562487602, "grad_norm": 0.05015282332897186, "global_step": 42099, "epoch": 397, "lr": 1.0078135769833758e-05} +{"train_loss": 0.014672892168164253, "grad_norm": 0.08947114646434784, "global_step": 42100, "epoch": 397, "lr": 1.0076351421307356e-05} +{"train_loss": 0.014956574887037277, "grad_norm": 0.08817210793495178, "global_step": 42101, "epoch": 397, "lr": 1.0074567213055286e-05} +{"train_loss": 0.015079435892403126, "grad_norm": 0.05496765673160553, "global_step": 42102, "epoch": 397, "lr": 1.007278314508382e-05} +{"train_loss": 0.014866958372294903, "grad_norm": 0.11639861017465591, "global_step": 42103, "epoch": 397, "lr": 1.0070999217399245e-05} +{"train_loss": 0.013815049082040787, "grad_norm": 0.10512003302574158, "global_step": 42104, "epoch": 397, "lr": 1.00692154300078e-05} +{"train_loss": 0.013925481587648392, "grad_norm": 0.0903799906373024, "global_step": 42105, "epoch": 397, "lr": 1.0067431782915764e-05} +{"train_loss": 0.01393876876682043, "grad_norm": 0.08403262495994568, "global_step": 42106, "epoch": 397, "lr": 1.0065648276129436e-05} +{"train_loss": 0.01652476191520691, "grad_norm": 0.07314672321081161, "global_step": 42107, "epoch": 397, "lr": 1.006386490965503e-05} +{"train_loss": 0.015400954522192478, "grad_norm": 0.06769610941410065, "global_step": 42108, "epoch": 397, "lr": 1.006208168349883e-05} +{"train_loss": 0.014097122475504875, "grad_norm": 0.06408818066120148, "global_step": 42109, "epoch": 397, "lr": 1.0060298597667117e-05} +{"train_loss": 0.013611510396003723, "grad_norm": 0.07057245820760727, "global_step": 42110, "epoch": 397, "lr": 1.005851565216614e-05} +{"train_loss": 0.013533355668187141, "grad_norm": 0.088587187230587, "global_step": 42111, "epoch": 397, "lr": 1.0056732847002165e-05} +{"train_loss": 0.015735633671283722, "grad_norm": 0.0705806314945221, "global_step": 42112, "epoch": 397, "lr": 1.0054950182181472e-05} +{"train_loss": 0.014581567607820034, "grad_norm": 0.11015024036169052, "global_step": 42113, "epoch": 397, "lr": 1.0053167657710311e-05} +{"train_loss": 0.014869865030050278, "grad_norm": 0.09688222408294678, "global_step": 42114, "epoch": 397, "lr": 1.0051385273594938e-05} +{"train_loss": 0.016783714294433594, "grad_norm": 0.08534970879554749, "global_step": 42115, "epoch": 397, "lr": 1.004960302984162e-05} +{"train_loss": 0.014657037332654, "grad_norm": 0.08354637771844864, "global_step": 42116, "epoch": 397, "lr": 1.0047820926456635e-05} +{"train_loss": 0.014049742370843887, "grad_norm": 0.07401349395513535, "global_step": 42117, "epoch": 397, "lr": 1.0046038963446225e-05} +{"train_loss": 0.013607954606413841, "grad_norm": 0.06523066759109497, "global_step": 42118, "epoch": 397, "lr": 1.0044257140816672e-05} +{"train_loss": 0.015937823802232742, "grad_norm": 0.10350339859724045, "global_step": 42119, "epoch": 397, "lr": 1.0042475458574202e-05} +{"train_loss": 0.014079244807362556, "grad_norm": 0.0794435515999794, "global_step": 42120, "epoch": 397, "lr": 1.0040693916725113e-05} +{"train_loss": 0.01420909259468317, "grad_norm": 0.08971983939409256, "global_step": 42121, "epoch": 397, "lr": 1.0038912515275638e-05} +{"train_loss": 0.014363574795424938, "grad_norm": 0.10169710963964462, "global_step": 42122, "epoch": 397, "lr": 1.003713125423204e-05} +{"train_loss": 0.016645804047584534, "grad_norm": 0.06434361636638641, "global_step": 42123, "epoch": 397, "lr": 1.0035350133600601e-05} +{"train_loss": 0.013120923191308975, "grad_norm": 0.11569961905479431, "global_step": 42124, "epoch": 397, "lr": 1.0033569153387562e-05} +{"train_loss": 0.014130834490060806, "grad_norm": 0.0542806051671505, "global_step": 42125, "epoch": 397, "lr": 1.0031788313599166e-05} +{"train_loss": 0.013090526685118675, "grad_norm": 0.11150510609149933, "global_step": 42126, "epoch": 397, "lr": 1.0030007614241683e-05} +{"train_loss": 0.014373798854649067, "grad_norm": 0.051622506231069565, "global_step": 42127, "epoch": 397, "lr": 1.0028227055321388e-05} +{"train_loss": 0.015243216417729855, "grad_norm": 0.05529932677745819, "global_step": 42128, "epoch": 397, "lr": 1.00264466368445e-05} +{"train_loss": 0.013773510232567787, "grad_norm": 0.06105758994817734, "global_step": 42129, "epoch": 397, "lr": 1.0024666358817308e-05} +{"train_loss": 0.014572321437299252, "grad_norm": 0.08184705674648285, "global_step": 42130, "epoch": 397, "lr": 1.0022886221246058e-05} +{"train_loss": 0.01565239205956459, "grad_norm": 0.0887347087264061, "global_step": 42131, "epoch": 397, "lr": 1.0021106224136989e-05} +{"train_loss": 0.014753615483641624, "grad_norm": 0.07439117878675461, "global_step": 42132, "epoch": 397, "lr": 1.0019326367496363e-05} +{"train_loss": 0.01768343709409237, "grad_norm": 0.11405156552791595, "global_step": 42133, "epoch": 397, "lr": 1.0017546651330445e-05} +{"train_loss": 0.013129395432770252, "grad_norm": 0.07994089275598526, "global_step": 42134, "epoch": 397, "lr": 1.0015767075645471e-05} +{"train_loss": 0.014600558206439018, "grad_norm": 0.15767642855644226, "global_step": 42135, "epoch": 397, "lr": 1.0013987640447719e-05} +{"train_loss": 0.01336703635752201, "grad_norm": 0.10428863763809204, "global_step": 42136, "epoch": 397, "lr": 1.0012208345743412e-05} +{"train_loss": 0.01487540639936924, "grad_norm": 0.11675908416509628, "global_step": 42137, "epoch": 397, "lr": 1.0010429191538823e-05} +{"train_loss": 0.014603335410356522, "grad_norm": 0.054023995995521545, "global_step": 42138, "epoch": 397, "lr": 1.0008650177840184e-05} +{"train_loss": 0.013643782585859299, "grad_norm": 0.0598611906170845, "global_step": 42139, "epoch": 397, "lr": 1.000687130465376e-05} +{"train_loss": 0.014555288478732109, "grad_norm": 0.07718922942876816, "global_step": 42140, "epoch": 397, "lr": 1.0005092571985808e-05} +{"train_loss": 0.013384411111474037, "grad_norm": 0.1186075359582901, "global_step": 42141, "epoch": 397, "lr": 1.0003313979842565e-05} +{"train_loss": 0.01868031546473503, "grad_norm": 0.06994946300983429, "global_step": 42142, "epoch": 397, "lr": 1.0001535528230271e-05} +{"train_loss": 0.01395810954272747, "grad_norm": 0.06554923206567764, "global_step": 42143, "epoch": 397, "lr": 9.999757217155186e-06} +{"train_loss": 0.014994394965469837, "grad_norm": 0.07887349277734756, "global_step": 42144, "epoch": 397, "lr": 9.997979046623574e-06} +{"train_loss": 0.016703851521015167, "grad_norm": 0.06964920461177826, "global_step": 42145, "epoch": 397, "lr": 9.996201016641648e-06} +{"train_loss": 0.016260892152786255, "grad_norm": 0.06017843633890152, "global_step": 42146, "epoch": 397, "lr": 9.99442312721568e-06} +{"train_loss": 0.015644418075680733, "grad_norm": 0.08372730761766434, "global_step": 42147, "epoch": 397, "lr": 9.99264537835194e-06} +{"train_loss": 0.015874380245804787, "grad_norm": 0.10354213416576385, "global_step": 42148, "epoch": 397, "lr": 9.990867770056612e-06} +{"train_loss": 0.015059143304824829, "grad_norm": 0.12450893223285675, "global_step": 42149, "epoch": 397, "lr": 9.98909030233598e-06} +{"train_loss": 0.015603023581206799, "grad_norm": 0.13341553509235382, "global_step": 42150, "epoch": 397, "lr": 9.987312975196294e-06} +{"train_loss": 0.014069173485040665, "grad_norm": 0.09672622382640839, "global_step": 42151, "epoch": 397, "lr": 9.98553578864378e-06} +{"train_loss": 0.01392448041588068, "grad_norm": 0.058490075170993805, "global_step": 42152, "epoch": 397, "lr": 9.98375874268469e-06} +{"train_loss": 0.017156725749373436, "grad_norm": 0.0678158849477768, "global_step": 42153, "epoch": 397, "lr": 9.981981837325282e-06} +{"train_loss": 0.013466953299939632, "grad_norm": 0.08644302189350128, "global_step": 42154, "epoch": 397, "lr": 9.980205072571786e-06} +{"train_loss": 0.015364591032266617, "grad_norm": 0.08145719021558762, "global_step": 42155, "epoch": 397, "lr": 9.978428448430432e-06} +{"train_loss": 0.014229253865778446, "grad_norm": 0.13178296387195587, "global_step": 42156, "epoch": 397, "lr": 9.976651964907473e-06} +{"train_loss": 0.018872832879424095, "grad_norm": 0.0746733620762825, "global_step": 42157, "epoch": 397, "lr": 9.974875622009172e-06} +{"train_loss": 0.014976410195231438, "grad_norm": 0.059466686099767685, "global_step": 42158, "epoch": 397, "lr": 9.973099419741732e-06} +{"train_loss": 0.01429129857569933, "grad_norm": 0.11663347482681274, "global_step": 42159, "epoch": 397, "lr": 9.971323358111434e-06} +{"train_loss": 0.016431212425231934, "grad_norm": 0.06844370812177658, "global_step": 42160, "epoch": 397, "lr": 9.969547437124477e-06} +{"train_loss": 0.018078137189149857, "grad_norm": 0.11302490532398224, "global_step": 42161, "epoch": 397, "lr": 9.967771656787139e-06} +{"train_loss": 0.01809360831975937, "grad_norm": 0.0974663719534874, "global_step": 42162, "epoch": 397, "lr": 9.965996017105627e-06} +{"train_loss": 0.013143439777195454, "grad_norm": 0.07104972004890442, "global_step": 42163, "epoch": 397, "lr": 9.964220518086198e-06} +{"train_loss": 0.012155445292592049, "grad_norm": 0.07672175765037537, "global_step": 42164, "epoch": 397, "lr": 9.962445159735096e-06} +{"train_loss": 0.014106404036283493, "grad_norm": 0.06011256203055382, "global_step": 42165, "epoch": 397, "lr": 9.960669942058554e-06} +{"train_loss": 0.015699582174420357, "grad_norm": 0.13620691001415253, "global_step": 42166, "epoch": 397, "lr": 9.95889486506279e-06} +{"train_loss": 0.015301772393286228, "grad_norm": 0.05313826724886894, "global_step": 42167, "epoch": 397, "lr": 9.957119928754066e-06} +{"train_loss": 0.015713699162006378, "grad_norm": 0.07964827120304108, "global_step": 42168, "epoch": 397, "lr": 9.9553451331386e-06} +{"train_loss": 0.016824297606945038, "grad_norm": 0.12379761785268784, "global_step": 42169, "epoch": 397, "lr": 9.953570478222634e-06} +{"train_loss": 0.015118159353733063, "grad_norm": 0.09235816448926926, "global_step": 42170, "epoch": 397, "lr": 9.951795964012422e-06} +{"train_loss": 0.017832785844802856, "grad_norm": 0.07252394407987595, "global_step": 42171, "epoch": 397, "lr": 9.95002159051418e-06} +{"train_loss": 0.017329439520835876, "grad_norm": 0.06662407517433167, "global_step": 42172, "epoch": 397, "lr": 9.94824735773413e-06} +{"train_loss": 0.013981536030769348, "grad_norm": 0.07291523367166519, "global_step": 42173, "epoch": 397, "lr": 9.946473265678525e-06} +{"train_loss": 0.018269218504428864, "grad_norm": 0.09916706383228302, "global_step": 42174, "epoch": 397, "lr": 9.9446993143536e-06} +{"train_loss": 0.015610437840223312, "grad_norm": 0.0835927352309227, "global_step": 42175, "epoch": 397, "lr": 9.942925503765577e-06} +{"train_loss": 0.013166734017431736, "grad_norm": 0.10330301523208618, "global_step": 42176, "epoch": 397, "lr": 9.941151833920687e-06} +{"train_loss": 0.017870523035526276, "grad_norm": 0.08819045126438141, "global_step": 42177, "epoch": 397, "lr": 9.939378304825187e-06} +{"train_loss": 0.015324980020523071, "grad_norm": 0.12204530835151672, "global_step": 42178, "epoch": 397, "lr": 9.937604916485282e-06} +{"train_loss": 0.012035954743623734, "grad_norm": 0.0708153247833252, "global_step": 42179, "epoch": 397, "lr": 9.935831668907203e-06} +{"train_loss": 0.0148350540548563, "grad_norm": 0.08254477381706238, "global_step": 42180, "epoch": 397, "lr": 9.934058562097187e-06} +{"train_loss": 0.015897564589977264, "grad_norm": 0.11112836748361588, "global_step": 42181, "epoch": 397, "lr": 9.93228559606148e-06} +{"train_loss": 0.012380126863718033, "grad_norm": 0.08205225318670273, "global_step": 42182, "epoch": 397, "lr": 9.930512770806294e-06} +{"train_loss": 0.014853004366159439, "grad_norm": 0.09309224784374237, "global_step": 42183, "epoch": 397, "lr": 9.928740086337851e-06} +{"train_loss": 0.01348489336669445, "grad_norm": 0.06535239517688751, "global_step": 42184, "epoch": 397, "lr": 9.926967542662386e-06} +{"train_loss": 0.015956945717334747, "grad_norm": 0.10433107614517212, "global_step": 42185, "epoch": 397, "lr": 9.925195139786143e-06} +{"train_loss": 0.01490207388997078, "grad_norm": 0.06506768614053726, "global_step": 42186, "epoch": 397, "lr": 9.923422877715322e-06} +{"train_loss": 0.014994410508013558, "grad_norm": 0.18814943730831146, "global_step": 42187, "epoch": 397, "lr": 9.921650756456164e-06, "val_loss": 0.019599052146077156} +{"train_loss": 0.013114383444190025, "grad_norm": 0.06654779613018036, "global_step": 42188, "epoch": 398, "lr": 9.919878776014924e-06} +{"train_loss": 0.017001960426568985, "grad_norm": 0.10731956362724304, "global_step": 42189, "epoch": 398, "lr": 9.918106936397775e-06} +{"train_loss": 0.015086444094777107, "grad_norm": 0.06463732570409775, "global_step": 42190, "epoch": 398, "lr": 9.916335237610962e-06} +{"train_loss": 0.012233680114150047, "grad_norm": 0.056593284010887146, "global_step": 42191, "epoch": 398, "lr": 9.914563679660731e-06} +{"train_loss": 0.01297803781926632, "grad_norm": 0.07392188906669617, "global_step": 42192, "epoch": 398, "lr": 9.912792262553277e-06} +{"train_loss": 0.014010917395353317, "grad_norm": 0.09796568006277084, "global_step": 42193, "epoch": 398, "lr": 9.911020986294839e-06} +{"train_loss": 0.015533837489783764, "grad_norm": 0.06600061058998108, "global_step": 42194, "epoch": 398, "lr": 9.90924985089165e-06} +{"train_loss": 0.0162123404443264, "grad_norm": 0.10672494024038315, "global_step": 42195, "epoch": 398, "lr": 9.907478856349923e-06} +{"train_loss": 0.014678273350000381, "grad_norm": 0.07951180636882782, "global_step": 42196, "epoch": 398, "lr": 9.905708002675868e-06} +{"train_loss": 0.012561701238155365, "grad_norm": 0.06167597323656082, "global_step": 42197, "epoch": 398, "lr": 9.903937289875715e-06} +{"train_loss": 0.015804167836904526, "grad_norm": 0.08198609203100204, "global_step": 42198, "epoch": 398, "lr": 9.902166717955702e-06} +{"train_loss": 0.014719225466251373, "grad_norm": 0.0807049497961998, "global_step": 42199, "epoch": 398, "lr": 9.900396286922026e-06} +{"train_loss": 0.01608879677951336, "grad_norm": 0.049966856837272644, "global_step": 42200, "epoch": 398, "lr": 9.89862599678093e-06} +{"train_loss": 0.01366492360830307, "grad_norm": 0.08986201882362366, "global_step": 42201, "epoch": 398, "lr": 9.896855847538611e-06} +{"train_loss": 0.014544475823640823, "grad_norm": 0.09319142252206802, "global_step": 42202, "epoch": 398, "lr": 9.895085839201318e-06} +{"train_loss": 0.014091242104768753, "grad_norm": 0.08207913488149643, "global_step": 42203, "epoch": 398, "lr": 9.89331597177523e-06} +{"train_loss": 0.013047192245721817, "grad_norm": 0.07455231249332428, "global_step": 42204, "epoch": 398, "lr": 9.891546245266592e-06} +{"train_loss": 0.01600867509841919, "grad_norm": 0.07174172252416611, "global_step": 42205, "epoch": 398, "lr": 9.889776659681632e-06} +{"train_loss": 0.01752631366252899, "grad_norm": 0.09950628876686096, "global_step": 42206, "epoch": 398, "lr": 9.88800721502655e-06} +{"train_loss": 0.0134779317304492, "grad_norm": 0.057480473071336746, "global_step": 42207, "epoch": 398, "lr": 9.88623791130756e-06} +{"train_loss": 0.015200598165392876, "grad_norm": 0.06966967135667801, "global_step": 42208, "epoch": 398, "lr": 9.884468748530895e-06} +{"train_loss": 0.014288687147200108, "grad_norm": 0.08087364584207535, "global_step": 42209, "epoch": 398, "lr": 9.882699726702744e-06} +{"train_loss": 0.014956473372876644, "grad_norm": 0.06165057048201561, "global_step": 42210, "epoch": 398, "lr": 9.880930845829345e-06} +{"train_loss": 0.016505161300301552, "grad_norm": 0.0888369232416153, "global_step": 42211, "epoch": 398, "lr": 9.879162105916918e-06} +{"train_loss": 0.016374794766306877, "grad_norm": 0.1216546967625618, "global_step": 42212, "epoch": 398, "lr": 9.877393506971666e-06} +{"train_loss": 0.016597501933574677, "grad_norm": 0.07173538208007812, "global_step": 42213, "epoch": 398, "lr": 9.875625048999793e-06} +{"train_loss": 0.013437209650874138, "grad_norm": 0.08174465596675873, "global_step": 42214, "epoch": 398, "lr": 9.87385673200752e-06} +{"train_loss": 0.01380277518182993, "grad_norm": 0.0576646514236927, "global_step": 42215, "epoch": 398, "lr": 9.87208855600108e-06} +{"train_loss": 0.01637095957994461, "grad_norm": 0.10556069761514664, "global_step": 42216, "epoch": 398, "lr": 9.87032052098666e-06} +{"train_loss": 0.014605632051825523, "grad_norm": 0.07674713432788849, "global_step": 42217, "epoch": 398, "lr": 9.868552626970478e-06} +{"train_loss": 0.016224626451730728, "grad_norm": 0.09642823785543442, "global_step": 42218, "epoch": 398, "lr": 9.866784873958762e-06} +{"train_loss": 0.01519493293017149, "grad_norm": 0.1077241525053978, "global_step": 42219, "epoch": 398, "lr": 9.865017261957715e-06} +{"train_loss": 0.01493285596370697, "grad_norm": 0.08505143970251083, "global_step": 42220, "epoch": 398, "lr": 9.863249790973528e-06} +{"train_loss": 0.012499263510107994, "grad_norm": 0.07620266824960709, "global_step": 42221, "epoch": 398, "lr": 9.861482461012423e-06} +{"train_loss": 0.014299474656581879, "grad_norm": 0.09459280222654343, "global_step": 42222, "epoch": 398, "lr": 9.859715272080628e-06} +{"train_loss": 0.013870038092136383, "grad_norm": 0.05636339634656906, "global_step": 42223, "epoch": 398, "lr": 9.85794822418432e-06} +{"train_loss": 0.013810234144330025, "grad_norm": 0.07261185348033905, "global_step": 42224, "epoch": 398, "lr": 9.856181317329744e-06} +{"train_loss": 0.013767082244157791, "grad_norm": 0.05335475131869316, "global_step": 42225, "epoch": 398, "lr": 9.85441455152309e-06} +{"train_loss": 0.017327096313238144, "grad_norm": 0.08670191466808319, "global_step": 42226, "epoch": 398, "lr": 9.852647926770547e-06} +{"train_loss": 0.014543155208230019, "grad_norm": 0.0689491555094719, "global_step": 42227, "epoch": 398, "lr": 9.850881443078341e-06} +{"train_loss": 0.01659952476620674, "grad_norm": 0.08695700764656067, "global_step": 42228, "epoch": 398, "lr": 9.849115100452678e-06} +{"train_loss": 0.013944512233138084, "grad_norm": 0.07359128445386887, "global_step": 42229, "epoch": 398, "lr": 9.847348898899783e-06} +{"train_loss": 0.01607414335012436, "grad_norm": 0.1045590490102768, "global_step": 42230, "epoch": 398, "lr": 9.84558283842582e-06} +{"train_loss": 0.017061984166502953, "grad_norm": 0.08560405671596527, "global_step": 42231, "epoch": 398, "lr": 9.843816919037018e-06} +{"train_loss": 0.01192576251924038, "grad_norm": 0.05939362570643425, "global_step": 42232, "epoch": 398, "lr": 9.842051140739594e-06} +{"train_loss": 0.014508794993162155, "grad_norm": 0.11669909954071045, "global_step": 42233, "epoch": 398, "lr": 9.840285503539725e-06} +{"train_loss": 0.015329302288591862, "grad_norm": 0.10857465863227844, "global_step": 42234, "epoch": 398, "lr": 9.838520007443624e-06} +{"train_loss": 0.017169807106256485, "grad_norm": 0.16665485501289368, "global_step": 42235, "epoch": 398, "lr": 9.836754652457514e-06} +{"train_loss": 0.0164293572306633, "grad_norm": 0.10025203973054886, "global_step": 42236, "epoch": 398, "lr": 9.834989438587578e-06} +{"train_loss": 0.018793143332004547, "grad_norm": 0.10375170409679413, "global_step": 42237, "epoch": 398, "lr": 9.83322436584001e-06} +{"train_loss": 0.013174697756767273, "grad_norm": 0.10938321053981781, "global_step": 42238, "epoch": 398, "lr": 9.83145943422103e-06} +{"train_loss": 0.015423791483044624, "grad_norm": 0.07388731837272644, "global_step": 42239, "epoch": 398, "lr": 9.829694643736836e-06} +{"train_loss": 0.015653058886528015, "grad_norm": 0.09741954505443573, "global_step": 42240, "epoch": 398, "lr": 9.82792999439362e-06} +{"train_loss": 0.015212217345833778, "grad_norm": 0.07441286742687225, "global_step": 42241, "epoch": 398, "lr": 9.826165486197598e-06} +{"train_loss": 0.016098828986287117, "grad_norm": 0.074303537607193, "global_step": 42242, "epoch": 398, "lr": 9.824401119154947e-06} +{"train_loss": 0.015480972826480865, "grad_norm": 0.23254446685314178, "global_step": 42243, "epoch": 398, "lr": 9.822636893271886e-06} +{"train_loss": 0.01584804803133011, "grad_norm": 0.12004780024290085, "global_step": 42244, "epoch": 398, "lr": 9.820872808554598e-06} +{"train_loss": 0.015057827346026897, "grad_norm": 0.11956078559160233, "global_step": 42245, "epoch": 398, "lr": 9.819108865009291e-06} +{"train_loss": 0.016451727598905563, "grad_norm": 0.07404205948114395, "global_step": 42246, "epoch": 398, "lr": 9.817345062642174e-06} +{"train_loss": 0.01569807156920433, "grad_norm": 0.08686556667089462, "global_step": 42247, "epoch": 398, "lr": 9.815581401459428e-06} +{"train_loss": 0.013932093977928162, "grad_norm": 0.0957968607544899, "global_step": 42248, "epoch": 398, "lr": 9.813817881467241e-06} +{"train_loss": 0.01544975396245718, "grad_norm": 0.08232644200325012, "global_step": 42249, "epoch": 398, "lr": 9.812054502671836e-06} +{"train_loss": 0.013230420649051666, "grad_norm": 0.0708155706524849, "global_step": 42250, "epoch": 398, "lr": 9.81029126507938e-06} +{"train_loss": 0.014465565793216228, "grad_norm": 0.06273792684078217, "global_step": 42251, "epoch": 398, "lr": 9.808528168696086e-06} +{"train_loss": 0.015364844352006912, "grad_norm": 0.09511857479810715, "global_step": 42252, "epoch": 398, "lr": 9.806765213528158e-06} +{"train_loss": 0.013057575561106205, "grad_norm": 0.11895528435707092, "global_step": 42253, "epoch": 398, "lr": 9.80500239958177e-06} +{"train_loss": 0.016150064766407013, "grad_norm": 0.1851007044315338, "global_step": 42254, "epoch": 398, "lr": 9.803239726863117e-06} +{"train_loss": 0.0154008399695158, "grad_norm": 0.10392844676971436, "global_step": 42255, "epoch": 398, "lr": 9.801477195378395e-06} +{"train_loss": 0.014855378307402134, "grad_norm": 0.07115728408098221, "global_step": 42256, "epoch": 398, "lr": 9.799714805133814e-06} +{"train_loss": 0.017133869230747223, "grad_norm": 0.07077725976705551, "global_step": 42257, "epoch": 398, "lr": 9.797952556135542e-06} +{"train_loss": 0.015036286786198616, "grad_norm": 0.07904600352048874, "global_step": 42258, "epoch": 398, "lr": 9.79619044838978e-06} +{"train_loss": 0.01499965786933899, "grad_norm": 0.07968340814113617, "global_step": 42259, "epoch": 398, "lr": 9.794428481902735e-06} +{"train_loss": 0.01455543003976345, "grad_norm": 0.07057148218154907, "global_step": 42260, "epoch": 398, "lr": 9.792666656680582e-06} +{"train_loss": 0.013540856540203094, "grad_norm": 0.08295824378728867, "global_step": 42261, "epoch": 398, "lr": 9.790904972729498e-06} +{"train_loss": 0.015341239050030708, "grad_norm": 0.08918528258800507, "global_step": 42262, "epoch": 398, "lr": 9.789143430055687e-06} +{"train_loss": 0.01585906185209751, "grad_norm": 0.08275680243968964, "global_step": 42263, "epoch": 398, "lr": 9.787382028665355e-06} +{"train_loss": 0.015260965563356876, "grad_norm": 0.08672478049993515, "global_step": 42264, "epoch": 398, "lr": 9.785620768564663e-06} +{"train_loss": 0.014883974567055702, "grad_norm": 0.06488721072673798, "global_step": 42265, "epoch": 398, "lr": 9.783859649759818e-06} +{"train_loss": 0.016151221469044685, "grad_norm": 0.08651027828454971, "global_step": 42266, "epoch": 398, "lr": 9.782098672257006e-06} +{"train_loss": 0.015249978750944138, "grad_norm": 0.07774819433689117, "global_step": 42267, "epoch": 398, "lr": 9.780337836062391e-06} +{"train_loss": 0.014616785570979118, "grad_norm": 0.06203127279877663, "global_step": 42268, "epoch": 398, "lr": 9.778577141182183e-06} +{"train_loss": 0.014511983841657639, "grad_norm": 0.08557282388210297, "global_step": 42269, "epoch": 398, "lr": 9.776816587622573e-06} +{"train_loss": 0.01651465892791748, "grad_norm": 0.07871060073375702, "global_step": 42270, "epoch": 398, "lr": 9.775056175389724e-06} +{"train_loss": 0.015797363594174385, "grad_norm": 0.0550578273832798, "global_step": 42271, "epoch": 398, "lr": 9.773295904489848e-06} +{"train_loss": 0.015737289562821388, "grad_norm": 0.11934729665517807, "global_step": 42272, "epoch": 398, "lr": 9.771535774929102e-06} +{"train_loss": 0.016168851405382156, "grad_norm": 0.07688536494970322, "global_step": 42273, "epoch": 398, "lr": 9.769775786713697e-06} +{"train_loss": 0.013414809480309486, "grad_norm": 0.13789092004299164, "global_step": 42274, "epoch": 398, "lr": 9.768015939849796e-06} +{"train_loss": 0.014406001195311546, "grad_norm": 0.07590880244970322, "global_step": 42275, "epoch": 398, "lr": 9.766256234343586e-06} +{"train_loss": 0.014492835849523544, "grad_norm": 0.06975485384464264, "global_step": 42276, "epoch": 398, "lr": 9.764496670201267e-06} +{"train_loss": 0.014297259040176868, "grad_norm": 0.14584754407405853, "global_step": 42277, "epoch": 398, "lr": 9.76273724742901e-06} +{"train_loss": 0.016344966366887093, "grad_norm": 0.10303891450166702, "global_step": 42278, "epoch": 398, "lr": 9.760977966032981e-06} +{"train_loss": 0.014792617410421371, "grad_norm": 0.06868335604667664, "global_step": 42279, "epoch": 398, "lr": 9.759218826019378e-06} +{"train_loss": 0.014179401099681854, "grad_norm": 0.08120471239089966, "global_step": 42280, "epoch": 398, "lr": 9.757459827394394e-06} +{"train_loss": 0.015682492405176163, "grad_norm": 0.10004456341266632, "global_step": 42281, "epoch": 398, "lr": 9.755700970164177e-06} +{"train_loss": 0.011236142367124557, "grad_norm": 0.10365453362464905, "global_step": 42282, "epoch": 398, "lr": 9.75394225433494e-06} +{"train_loss": 0.015335625037550926, "grad_norm": 0.18813957273960114, "global_step": 42283, "epoch": 398, "lr": 9.75218367991284e-06} +{"train_loss": 0.012591386213898659, "grad_norm": 0.11944947391748428, "global_step": 42284, "epoch": 398, "lr": 9.75042524690407e-06} +{"train_loss": 0.014381011947989464, "grad_norm": 0.09753897041082382, "global_step": 42285, "epoch": 398, "lr": 9.748666955314789e-06} +{"train_loss": 0.015987548977136612, "grad_norm": 0.06358286738395691, "global_step": 42286, "epoch": 398, "lr": 9.74690880515119e-06} +{"train_loss": 0.013298764824867249, "grad_norm": 0.06628451496362686, "global_step": 42287, "epoch": 398, "lr": 9.745150796419462e-06} +{"train_loss": 0.012621496804058552, "grad_norm": 0.05687067657709122, "global_step": 42288, "epoch": 398, "lr": 9.743392929125766e-06} +{"train_loss": 0.017425980418920517, "grad_norm": 0.09895505011081696, "global_step": 42289, "epoch": 398, "lr": 9.741635203276267e-06} +{"train_loss": 0.013446632772684097, "grad_norm": 0.08854648470878601, "global_step": 42290, "epoch": 398, "lr": 9.739877618877169e-06} +{"train_loss": 0.02062269300222397, "grad_norm": 0.1005481630563736, "global_step": 42291, "epoch": 398, "lr": 9.738120175934612e-06} +{"train_loss": 0.015457418747246265, "grad_norm": 0.060222726315259933, "global_step": 42292, "epoch": 398, "lr": 9.736362874454797e-06} +{"train_loss": 0.015000569607781351, "grad_norm": 0.09270334988832474, "global_step": 42293, "epoch": 398, "lr": 9.734605714443906e-06, "val_loss": 0.01967974193394184} +{"train_loss": 0.014628085307776928, "grad_norm": 0.08594856411218643, "global_step": 42294, "epoch": 399, "lr": 9.732848695908098e-06} +{"train_loss": 0.016657479107379913, "grad_norm": 0.07060358673334122, "global_step": 42295, "epoch": 399, "lr": 9.731091818853538e-06} +{"train_loss": 0.0158627238124609, "grad_norm": 0.09568358957767487, "global_step": 42296, "epoch": 399, "lr": 9.729335083286406e-06} +{"train_loss": 0.016297314316034317, "grad_norm": 0.08940097689628601, "global_step": 42297, "epoch": 399, "lr": 9.72757848921289e-06} +{"train_loss": 0.013812610879540443, "grad_norm": 0.18925096094608307, "global_step": 42298, "epoch": 399, "lr": 9.725822036639138e-06} +{"train_loss": 0.014869209378957748, "grad_norm": 0.09500224888324738, "global_step": 42299, "epoch": 399, "lr": 9.72406572557133e-06} +{"train_loss": 0.017383888363838196, "grad_norm": 0.0643935352563858, "global_step": 42300, "epoch": 399, "lr": 9.722309556015657e-06} +{"train_loss": 0.016182873398065567, "grad_norm": 0.07955781370401382, "global_step": 42301, "epoch": 399, "lr": 9.720553527978265e-06} +{"train_loss": 0.014669572934508324, "grad_norm": 0.06435400247573853, "global_step": 42302, "epoch": 399, "lr": 9.718797641465326e-06} +{"train_loss": 0.013542406260967255, "grad_norm": 0.128473699092865, "global_step": 42303, "epoch": 399, "lr": 9.71704189648301e-06} +{"train_loss": 0.015534773468971252, "grad_norm": 0.061733923852443695, "global_step": 42304, "epoch": 399, "lr": 9.7152862930375e-06} +{"train_loss": 0.016123071312904358, "grad_norm": 0.07008504867553711, "global_step": 42305, "epoch": 399, "lr": 9.713530831134948e-06} +{"train_loss": 0.015605440363287926, "grad_norm": 0.13536760210990906, "global_step": 42306, "epoch": 399, "lr": 9.71177551078154e-06} +{"train_loss": 0.01600375398993492, "grad_norm": 0.13486061990261078, "global_step": 42307, "epoch": 399, "lr": 9.710020331983427e-06} +{"train_loss": 0.013369168154895306, "grad_norm": 0.07949960976839066, "global_step": 42308, "epoch": 399, "lr": 9.708265294746772e-06} +{"train_loss": 0.013167060911655426, "grad_norm": 0.10434170067310333, "global_step": 42309, "epoch": 399, "lr": 9.70651039907775e-06} +{"train_loss": 0.017027340829372406, "grad_norm": 0.14509496092796326, "global_step": 42310, "epoch": 399, "lr": 9.704755644982539e-06} +{"train_loss": 0.01499104406684637, "grad_norm": 0.05364590883255005, "global_step": 42311, "epoch": 399, "lr": 9.703001032467279e-06} +{"train_loss": 0.016829531639814377, "grad_norm": 0.07705748081207275, "global_step": 42312, "epoch": 399, "lr": 9.701246561538163e-06} +{"train_loss": 0.017065173014998436, "grad_norm": 0.09442063421010971, "global_step": 42313, "epoch": 399, "lr": 9.699492232201329e-06} +{"train_loss": 0.014293443411588669, "grad_norm": 0.10003524273633957, "global_step": 42314, "epoch": 399, "lr": 9.697738044462968e-06} +{"train_loss": 0.014320909045636654, "grad_norm": 0.0770108625292778, "global_step": 42315, "epoch": 399, "lr": 9.695983998329212e-06} +{"train_loss": 0.014716283418238163, "grad_norm": 0.07013068348169327, "global_step": 42316, "epoch": 399, "lr": 9.694230093806244e-06} +{"train_loss": 0.015041529200971127, "grad_norm": 0.06118037924170494, "global_step": 42317, "epoch": 399, "lr": 9.69247633090023e-06} +{"train_loss": 0.014686623588204384, "grad_norm": 0.12353154271841049, "global_step": 42318, "epoch": 399, "lr": 9.690722709617327e-06} +{"train_loss": 0.015718741342425346, "grad_norm": 0.06033455953001976, "global_step": 42319, "epoch": 399, "lr": 9.688969229963685e-06} +{"train_loss": 0.015445029363036156, "grad_norm": 0.06956576555967331, "global_step": 42320, "epoch": 399, "lr": 9.687215891945472e-06} +{"train_loss": 0.015434612520039082, "grad_norm": 0.04860509932041168, "global_step": 42321, "epoch": 399, "lr": 9.685462695568864e-06} +{"train_loss": 0.013994690962135792, "grad_norm": 0.08810955286026001, "global_step": 42322, "epoch": 399, "lr": 9.683709640839994e-06} +{"train_loss": 0.012804955244064331, "grad_norm": 0.09006381779909134, "global_step": 42323, "epoch": 399, "lr": 9.681956727765034e-06} +{"train_loss": 0.014764541760087013, "grad_norm": 0.06675013154745102, "global_step": 42324, "epoch": 399, "lr": 9.680203956350175e-06} +{"train_loss": 0.013922339305281639, "grad_norm": 0.0614079125225544, "global_step": 42325, "epoch": 399, "lr": 9.678451326601518e-06} +{"train_loss": 0.012799780815839767, "grad_norm": 0.07809752970933914, "global_step": 42326, "epoch": 399, "lr": 9.676698838525245e-06} +{"train_loss": 0.018314199522137642, "grad_norm": 0.11450020968914032, "global_step": 42327, "epoch": 399, "lr": 9.674946492127528e-06} +{"train_loss": 0.014335446059703827, "grad_norm": 0.0877646878361702, "global_step": 42328, "epoch": 399, "lr": 9.673194287414499e-06} +{"train_loss": 0.013890424743294716, "grad_norm": 0.07575228810310364, "global_step": 42329, "epoch": 399, "lr": 9.671442224392342e-06} +{"train_loss": 0.016947217285633087, "grad_norm": 0.06770766526460648, "global_step": 42330, "epoch": 399, "lr": 9.669690303067187e-06} +{"train_loss": 0.015877021476626396, "grad_norm": 0.07287826389074326, "global_step": 42331, "epoch": 399, "lr": 9.667938523445214e-06} +{"train_loss": 0.013896960765123367, "grad_norm": 0.0648319199681282, "global_step": 42332, "epoch": 399, "lr": 9.666186885532552e-06} +{"train_loss": 0.015318779274821281, "grad_norm": 0.0783185213804245, "global_step": 42333, "epoch": 399, "lr": 9.664435389335363e-06} +{"train_loss": 0.012001095339655876, "grad_norm": 0.06520998477935791, "global_step": 42334, "epoch": 399, "lr": 9.662684034859825e-06} +{"train_loss": 0.012348847463726997, "grad_norm": 0.08470330387353897, "global_step": 42335, "epoch": 399, "lr": 9.660932822112073e-06} +{"train_loss": 0.014080913737416267, "grad_norm": 0.09869659692049026, "global_step": 42336, "epoch": 399, "lr": 9.659181751098245e-06} +{"train_loss": 0.014526214450597763, "grad_norm": 0.06441648304462433, "global_step": 42337, "epoch": 399, "lr": 9.657430821824509e-06} +{"train_loss": 0.013648904860019684, "grad_norm": 0.14756731688976288, "global_step": 42338, "epoch": 399, "lr": 9.655680034297032e-06} +{"train_loss": 0.014889411628246307, "grad_norm": 0.08250408619642258, "global_step": 42339, "epoch": 399, "lr": 9.653929388521938e-06} +{"train_loss": 0.018128525465726852, "grad_norm": 0.12195507436990738, "global_step": 42340, "epoch": 399, "lr": 9.652178884505387e-06} +{"train_loss": 0.014137038961052895, "grad_norm": 0.0600532703101635, "global_step": 42341, "epoch": 399, "lr": 9.650428522253546e-06} +{"train_loss": 0.016428669914603233, "grad_norm": 0.05483807995915413, "global_step": 42342, "epoch": 399, "lr": 9.648678301772551e-06} +{"train_loss": 0.014245633967220783, "grad_norm": 0.09136726707220078, "global_step": 42343, "epoch": 399, "lr": 9.646928223068541e-06} +{"train_loss": 0.01586434617638588, "grad_norm": 0.08241701871156693, "global_step": 42344, "epoch": 399, "lr": 9.64517828614767e-06} +{"train_loss": 0.013872053474187851, "grad_norm": 0.09152337908744812, "global_step": 42345, "epoch": 399, "lr": 9.643428491016116e-06} +{"train_loss": 0.014656715095043182, "grad_norm": 0.10363218933343887, "global_step": 42346, "epoch": 399, "lr": 9.641678837679985e-06} +{"train_loss": 0.014142772182822227, "grad_norm": 0.08970007300376892, "global_step": 42347, "epoch": 399, "lr": 9.639929326145459e-06} +{"train_loss": 0.01781681552529335, "grad_norm": 0.17022646963596344, "global_step": 42348, "epoch": 399, "lr": 9.638179956418663e-06} +{"train_loss": 0.01479842234402895, "grad_norm": 0.06925859302282333, "global_step": 42349, "epoch": 399, "lr": 9.636430728505742e-06} +{"train_loss": 0.015903428196907043, "grad_norm": 0.125136598944664, "global_step": 42350, "epoch": 399, "lr": 9.634681642412846e-06} +{"train_loss": 0.01618954911828041, "grad_norm": 0.08678015321493149, "global_step": 42351, "epoch": 399, "lr": 9.63293269814614e-06} +{"train_loss": 0.012423431500792503, "grad_norm": 0.05057666823267937, "global_step": 42352, "epoch": 399, "lr": 9.631183895711743e-06} +{"train_loss": 0.011808758601546288, "grad_norm": 0.09769514948129654, "global_step": 42353, "epoch": 399, "lr": 9.62943523511582e-06} +{"train_loss": 0.014769360423088074, "grad_norm": 0.12278322875499725, "global_step": 42354, "epoch": 399, "lr": 9.627686716364487e-06} +{"train_loss": 0.013944001868367195, "grad_norm": 0.08235082775354385, "global_step": 42355, "epoch": 399, "lr": 9.625938339463925e-06} +{"train_loss": 0.013489721342921257, "grad_norm": 0.06963163614273071, "global_step": 42356, "epoch": 399, "lr": 9.624190104420244e-06} +{"train_loss": 0.013603268191218376, "grad_norm": 0.11376699060201645, "global_step": 42357, "epoch": 399, "lr": 9.622442011239596e-06} +{"train_loss": 0.015314694494009018, "grad_norm": 0.12073459476232529, "global_step": 42358, "epoch": 399, "lr": 9.620694059928143e-06} +{"train_loss": 0.01687530428171158, "grad_norm": 0.0855366438627243, "global_step": 42359, "epoch": 399, "lr": 9.618946250492011e-06} +{"train_loss": 0.016306772828102112, "grad_norm": 0.07132838666439056, "global_step": 42360, "epoch": 399, "lr": 9.617198582937325e-06} +{"train_loss": 0.014688309282064438, "grad_norm": 0.06737472116947174, "global_step": 42361, "epoch": 399, "lr": 9.615451057270241e-06} +{"train_loss": 0.014764627441763878, "grad_norm": 0.09802287071943283, "global_step": 42362, "epoch": 399, "lr": 9.613703673496915e-06} +{"train_loss": 0.014510808512568474, "grad_norm": 0.07515111565589905, "global_step": 42363, "epoch": 399, "lr": 9.611956431623458e-06} +{"train_loss": 0.013366357423365116, "grad_norm": 0.09156233072280884, "global_step": 42364, "epoch": 399, "lr": 9.610209331656023e-06} +{"train_loss": 0.01468387059867382, "grad_norm": 0.08347280323505402, "global_step": 42365, "epoch": 399, "lr": 9.608462373600768e-06} +{"train_loss": 0.014109042473137379, "grad_norm": 0.07021918892860413, "global_step": 42366, "epoch": 399, "lr": 9.60671555746379e-06} +{"train_loss": 0.016106320545077324, "grad_norm": 0.0890720933675766, "global_step": 42367, "epoch": 399, "lr": 9.604968883251242e-06} +{"train_loss": 0.015602631494402885, "grad_norm": 0.08316488564014435, "global_step": 42368, "epoch": 399, "lr": 9.603222350969282e-06} +{"train_loss": 0.016224287450313568, "grad_norm": 0.12748229503631592, "global_step": 42369, "epoch": 399, "lr": 9.601475960624017e-06} +{"train_loss": 0.015624075196683407, "grad_norm": 0.12004578858613968, "global_step": 42370, "epoch": 399, "lr": 9.5997297122216e-06} +{"train_loss": 0.015660621225833893, "grad_norm": 0.059985946863889694, "global_step": 42371, "epoch": 399, "lr": 9.597983605768169e-06} +{"train_loss": 0.014025741256773472, "grad_norm": 0.08507804572582245, "global_step": 42372, "epoch": 399, "lr": 9.596237641269856e-06} +{"train_loss": 0.011953910812735558, "grad_norm": 0.08457288891077042, "global_step": 42373, "epoch": 399, "lr": 9.594491818732782e-06} +{"train_loss": 0.013159976340830326, "grad_norm": 0.05761845409870148, "global_step": 42374, "epoch": 399, "lr": 9.592746138163094e-06} +{"train_loss": 0.016496261581778526, "grad_norm": 0.09009487181901932, "global_step": 42375, "epoch": 399, "lr": 9.591000599566929e-06} +{"train_loss": 0.01761903055012226, "grad_norm": 0.06848820298910141, "global_step": 42376, "epoch": 399, "lr": 9.589255202950409e-06} +{"train_loss": 0.015578373335301876, "grad_norm": 0.20584923028945923, "global_step": 42377, "epoch": 399, "lr": 9.587509948319679e-06} +{"train_loss": 0.014034006744623184, "grad_norm": 0.09142264723777771, "global_step": 42378, "epoch": 399, "lr": 9.585764835680855e-06} +{"train_loss": 0.01653299853205681, "grad_norm": 0.08182033896446228, "global_step": 42379, "epoch": 399, "lr": 9.584019865040089e-06} +{"train_loss": 0.016024142503738403, "grad_norm": 0.09945256263017654, "global_step": 42380, "epoch": 399, "lr": 9.582275036403487e-06} +{"train_loss": 0.012323739938437939, "grad_norm": 0.14189206063747406, "global_step": 42381, "epoch": 399, "lr": 9.580530349777195e-06} +{"train_loss": 0.014125320129096508, "grad_norm": 0.10613204538822174, "global_step": 42382, "epoch": 399, "lr": 9.578785805167368e-06} +{"train_loss": 0.01682855188846588, "grad_norm": 0.06289955228567123, "global_step": 42383, "epoch": 399, "lr": 9.577041402580083e-06} +{"train_loss": 0.015226379968225956, "grad_norm": 0.13506537675857544, "global_step": 42384, "epoch": 399, "lr": 9.575297142021494e-06} +{"train_loss": 0.017100516706705093, "grad_norm": 0.06971247494220734, "global_step": 42385, "epoch": 399, "lr": 9.573553023497728e-06} +{"train_loss": 0.014994152821600437, "grad_norm": 0.05872593820095062, "global_step": 42386, "epoch": 399, "lr": 9.571809047014935e-06} +{"train_loss": 0.015347251668572426, "grad_norm": 0.08286383748054504, "global_step": 42387, "epoch": 399, "lr": 9.570065212579204e-06} +{"train_loss": 0.014731992036104202, "grad_norm": 0.059087034314870834, "global_step": 42388, "epoch": 399, "lr": 9.5683215201967e-06} +{"train_loss": 0.014012230560183525, "grad_norm": 0.06714808195829391, "global_step": 42389, "epoch": 399, "lr": 9.566577969873524e-06} +{"train_loss": 0.01568995788693428, "grad_norm": 0.0859212651848793, "global_step": 42390, "epoch": 399, "lr": 9.564834561615799e-06} +{"train_loss": 0.016531407833099365, "grad_norm": 0.07970967143774033, "global_step": 42391, "epoch": 399, "lr": 9.563091295429661e-06} +{"train_loss": 0.016208108514547348, "grad_norm": 0.0755603164434433, "global_step": 42392, "epoch": 399, "lr": 9.561348171321249e-06} +{"train_loss": 0.014954822137951851, "grad_norm": 0.05675636604428291, "global_step": 42393, "epoch": 399, "lr": 9.55960518929666e-06} +{"train_loss": 0.013637358322739601, "grad_norm": 0.11113456636667252, "global_step": 42394, "epoch": 399, "lr": 9.55786234936204e-06} +{"train_loss": 0.015095542185008526, "grad_norm": 0.05775969475507736, "global_step": 42395, "epoch": 399, "lr": 9.556119651523498e-06} +{"train_loss": 0.014993558637797832, "grad_norm": 0.0843629390001297, "global_step": 42396, "epoch": 399, "lr": 9.55437709578717e-06} +{"train_loss": 0.015466822311282158, "grad_norm": 0.10718125104904175, "global_step": 42397, "epoch": 399, "lr": 9.552634682159162e-06} +{"train_loss": 0.014031103812158108, "grad_norm": 0.054184041917324066, "global_step": 42398, "epoch": 399, "lr": 9.550892410645601e-06} +{"train_loss": 0.015008502332317942, "grad_norm": 0.1613374799489975, "global_step": 42399, "epoch": 399, "lr": 9.549150281252633e-06, "val_loss": 0.01944383978843689} +{"train_loss": 0.013811256736516953, "grad_norm": 0.05824004113674164, "global_step": 42400, "epoch": 400, "lr": 9.547408293986354e-06} +{"train_loss": 0.015478779561817646, "grad_norm": 0.06757719069719315, "global_step": 42401, "epoch": 400, "lr": 9.54566644885288e-06} +{"train_loss": 0.017969274893403053, "grad_norm": 0.06381809711456299, "global_step": 42402, "epoch": 400, "lr": 9.54392474585834e-06} +{"train_loss": 0.014504262246191502, "grad_norm": 0.06190215051174164, "global_step": 42403, "epoch": 400, "lr": 9.542183185008869e-06} +{"train_loss": 0.015274888835847378, "grad_norm": 0.06689966470003128, "global_step": 42404, "epoch": 400, "lr": 9.540441766310553e-06} +{"train_loss": 0.014426157809793949, "grad_norm": 0.0786362811923027, "global_step": 42405, "epoch": 400, "lr": 9.538700489769536e-06} +{"train_loss": 0.016039542853832245, "grad_norm": 0.13117414712905884, "global_step": 42406, "epoch": 400, "lr": 9.53695935539195e-06} +{"train_loss": 0.013800404965877533, "grad_norm": 0.060034412890672684, "global_step": 42407, "epoch": 400, "lr": 9.535218363183868e-06} +{"train_loss": 0.013562057167291641, "grad_norm": 0.0587182492017746, "global_step": 42408, "epoch": 400, "lr": 9.533477513151435e-06} +{"train_loss": 0.015392902307212353, "grad_norm": 0.09608116000890732, "global_step": 42409, "epoch": 400, "lr": 9.531736805300773e-06} +{"train_loss": 0.015674564987421036, "grad_norm": 0.08961458504199982, "global_step": 42410, "epoch": 400, "lr": 9.529996239637973e-06} +{"train_loss": 0.013933682814240456, "grad_norm": 0.08127117902040482, "global_step": 42411, "epoch": 400, "lr": 9.528255816169167e-06} +{"train_loss": 0.01560051366686821, "grad_norm": 0.13787850737571716, "global_step": 42412, "epoch": 400, "lr": 9.526515534900488e-06} +{"train_loss": 0.01628967933356762, "grad_norm": 0.10902139544487, "global_step": 42413, "epoch": 400, "lr": 9.524775395838026e-06} +{"train_loss": 0.016608674079179764, "grad_norm": 0.07816799730062485, "global_step": 42414, "epoch": 400, "lr": 9.52303539898789e-06} +{"train_loss": 0.01435733400285244, "grad_norm": 0.06796174496412277, "global_step": 42415, "epoch": 400, "lr": 9.521295544356208e-06} +{"train_loss": 0.01582992449402809, "grad_norm": 0.10140016674995422, "global_step": 42416, "epoch": 400, "lr": 9.519555831949101e-06} +{"train_loss": 0.016511285677552223, "grad_norm": 0.12741245329380035, "global_step": 42417, "epoch": 400, "lr": 9.517816261772655e-06} +{"train_loss": 0.01744873821735382, "grad_norm": 0.10361266881227493, "global_step": 42418, "epoch": 400, "lr": 9.516076833833009e-06} +{"train_loss": 0.016788171604275703, "grad_norm": 0.0896795392036438, "global_step": 42419, "epoch": 400, "lr": 9.514337548136255e-06} +{"train_loss": 0.016767483204603195, "grad_norm": 0.0839390680193901, "global_step": 42420, "epoch": 400, "lr": 9.512598404688523e-06} +{"train_loss": 0.015456190332770348, "grad_norm": 0.07023470103740692, "global_step": 42421, "epoch": 400, "lr": 9.510859403495898e-06} +{"train_loss": 0.016843287274241447, "grad_norm": 0.060506805777549744, "global_step": 42422, "epoch": 400, "lr": 9.509120544564504e-06} +{"train_loss": 0.0150894895195961, "grad_norm": 0.07517866045236588, "global_step": 42423, "epoch": 400, "lr": 9.50738182790047e-06} +{"train_loss": 0.014419453218579292, "grad_norm": 0.11034231632947922, "global_step": 42424, "epoch": 400, "lr": 9.505643253509884e-06} +{"train_loss": 0.016312457621097565, "grad_norm": 0.07625478506088257, "global_step": 42425, "epoch": 400, "lr": 9.50390482139884e-06} +{"train_loss": 0.016988901421427727, "grad_norm": 0.11328103393316269, "global_step": 42426, "epoch": 400, "lr": 9.50216653157348e-06} +{"train_loss": 0.013232877478003502, "grad_norm": 0.10577457398176193, "global_step": 42427, "epoch": 400, "lr": 9.500428384039883e-06} +{"train_loss": 0.014014367014169693, "grad_norm": 0.07803647220134735, "global_step": 42428, "epoch": 400, "lr": 9.498690378804164e-06} +{"train_loss": 0.012927720323204994, "grad_norm": 0.06852427870035172, "global_step": 42429, "epoch": 400, "lr": 9.496952515872448e-06} +{"train_loss": 0.016008902341127396, "grad_norm": 0.05862337723374367, "global_step": 42430, "epoch": 400, "lr": 9.495214795250828e-06} +{"train_loss": 0.015711873769760132, "grad_norm": 0.09177344292402267, "global_step": 42431, "epoch": 400, "lr": 9.49347721694539e-06} +{"train_loss": 0.013388829305768013, "grad_norm": 0.05192539095878601, "global_step": 42432, "epoch": 400, "lr": 9.491739780962266e-06} +{"train_loss": 0.01210428774356842, "grad_norm": 0.06283789873123169, "global_step": 42433, "epoch": 400, "lr": 9.49000248730756e-06} +{"train_loss": 0.014882927760481834, "grad_norm": 0.09141504764556885, "global_step": 42434, "epoch": 400, "lr": 9.488265335987351e-06} +{"train_loss": 0.014093123376369476, "grad_norm": 0.09448101371526718, "global_step": 42435, "epoch": 400, "lr": 9.486528327007777e-06} +{"train_loss": 0.016909092664718628, "grad_norm": 0.08513688296079636, "global_step": 42436, "epoch": 400, "lr": 9.484791460374914e-06} +{"train_loss": 0.015076037496328354, "grad_norm": 0.06980293989181519, "global_step": 42437, "epoch": 400, "lr": 9.483054736094882e-06} +{"train_loss": 0.014069287106394768, "grad_norm": 0.09468323737382889, "global_step": 42438, "epoch": 400, "lr": 9.481318154173763e-06} +{"train_loss": 0.013045871630311012, "grad_norm": 0.05897130072116852, "global_step": 42439, "epoch": 400, "lr": 9.479581714617668e-06} +{"train_loss": 0.01586793176829815, "grad_norm": 0.06855161488056183, "global_step": 42440, "epoch": 400, "lr": 9.477845417432723e-06} +{"train_loss": 0.012989608570933342, "grad_norm": 0.07046309858560562, "global_step": 42441, "epoch": 400, "lr": 9.476109262625e-06} +{"train_loss": 0.013789936900138855, "grad_norm": 0.07239459455013275, "global_step": 42442, "epoch": 400, "lr": 9.474373250200591e-06} +{"train_loss": 0.013818925246596336, "grad_norm": 0.059253036975860596, "global_step": 42443, "epoch": 400, "lr": 9.472637380165616e-06} +{"train_loss": 0.01583843119442463, "grad_norm": 0.06604400277137756, "global_step": 42444, "epoch": 400, "lr": 9.470901652526177e-06} +{"train_loss": 0.01557128131389618, "grad_norm": 0.05843956395983696, "global_step": 42445, "epoch": 400, "lr": 9.469166067288354e-06} +{"train_loss": 0.01476554200053215, "grad_norm": 0.11363149434328079, "global_step": 42446, "epoch": 400, "lr": 9.467430624458251e-06} +{"train_loss": 0.017472371459007263, "grad_norm": 0.0628218799829483, "global_step": 42447, "epoch": 400, "lr": 9.465695324042001e-06} +{"train_loss": 0.012306442484259605, "grad_norm": 0.06419125199317932, "global_step": 42448, "epoch": 400, "lr": 9.463960166045637e-06} +{"train_loss": 0.014855822548270226, "grad_norm": 0.13045485317707062, "global_step": 42449, "epoch": 400, "lr": 9.462225150475295e-06} +{"train_loss": 0.016130119562149048, "grad_norm": 0.10498697310686111, "global_step": 42450, "epoch": 400, "lr": 9.460490277337076e-06} +{"train_loss": 0.016351081430912018, "grad_norm": 0.06504108756780624, "global_step": 42451, "epoch": 400, "lr": 9.45875554663705e-06} +{"train_loss": 0.01775493286550045, "grad_norm": 0.08195805549621582, "global_step": 42452, "epoch": 400, "lr": 9.457020958381324e-06} +{"train_loss": 0.01514105498790741, "grad_norm": 0.07139405608177185, "global_step": 42453, "epoch": 400, "lr": 9.455286512576005e-06} +{"train_loss": 0.01457340270280838, "grad_norm": 0.058218155056238174, "global_step": 42454, "epoch": 400, "lr": 9.453552209227185e-06} +{"train_loss": 0.014872146770358086, "grad_norm": 0.07270188629627228, "global_step": 42455, "epoch": 400, "lr": 9.451818048340932e-06} +{"train_loss": 0.015446530655026436, "grad_norm": 0.12147458642721176, "global_step": 42456, "epoch": 400, "lr": 9.450084029923356e-06} +{"train_loss": 0.015556908212602139, "grad_norm": 0.08385459333658218, "global_step": 42457, "epoch": 400, "lr": 9.448350153980562e-06} +{"train_loss": 0.01632484793663025, "grad_norm": 0.11617980152368546, "global_step": 42458, "epoch": 400, "lr": 9.446616420518623e-06} +{"train_loss": 0.016002196818590164, "grad_norm": 0.07349060475826263, "global_step": 42459, "epoch": 400, "lr": 9.444882829543644e-06} +{"train_loss": 0.014507700689136982, "grad_norm": 0.09407413005828857, "global_step": 42460, "epoch": 400, "lr": 9.4431493810617e-06} +{"train_loss": 0.014525553211569786, "grad_norm": 0.07271252572536469, "global_step": 42461, "epoch": 400, "lr": 9.441416075078901e-06} +{"train_loss": 0.016455840319395065, "grad_norm": 0.08396340906620026, "global_step": 42462, "epoch": 400, "lr": 9.439682911601321e-06} +{"train_loss": 0.015691690146923065, "grad_norm": 0.06823507696390152, "global_step": 42463, "epoch": 400, "lr": 9.437949890635051e-06} +{"train_loss": 0.013285810127854347, "grad_norm": 0.11626042425632477, "global_step": 42464, "epoch": 400, "lr": 9.4362170121862e-06} +{"train_loss": 0.014694387093186378, "grad_norm": 0.06967078894376755, "global_step": 42465, "epoch": 400, "lr": 9.434484276260841e-06} +{"train_loss": 0.014680333435535431, "grad_norm": 0.054901979863643646, "global_step": 42466, "epoch": 400, "lr": 9.432751682865048e-06} +{"train_loss": 0.012186333537101746, "grad_norm": 0.05539451912045479, "global_step": 42467, "epoch": 400, "lr": 9.43101923200494e-06} +{"train_loss": 0.015379566699266434, "grad_norm": 0.1708676517009735, "global_step": 42468, "epoch": 400, "lr": 9.429286923686576e-06} +{"train_loss": 0.015407426282763481, "grad_norm": 0.06045841798186302, "global_step": 42469, "epoch": 400, "lr": 9.427554757916047e-06} +{"train_loss": 0.015117747709155083, "grad_norm": 0.09016400575637817, "global_step": 42470, "epoch": 400, "lr": 9.425822734699464e-06} +{"train_loss": 0.012390169315040112, "grad_norm": 0.0585348978638649, "global_step": 42471, "epoch": 400, "lr": 9.424090854042895e-06} +{"train_loss": 0.012962691485881805, "grad_norm": 0.09375403821468353, "global_step": 42472, "epoch": 400, "lr": 9.422359115952406e-06} +{"train_loss": 0.015518026426434517, "grad_norm": 0.07427027076482773, "global_step": 42473, "epoch": 400, "lr": 9.420627520434105e-06} +{"train_loss": 0.016289034858345985, "grad_norm": 0.06934481859207153, "global_step": 42474, "epoch": 400, "lr": 9.418896067494087e-06} +{"train_loss": 0.016130097210407257, "grad_norm": 0.09509052336215973, "global_step": 42475, "epoch": 400, "lr": 9.417164757138403e-06} +{"train_loss": 0.012752048671245575, "grad_norm": 0.11447323113679886, "global_step": 42476, "epoch": 400, "lr": 9.415433589373151e-06} +{"train_loss": 0.015385974198579788, "grad_norm": 0.09984178841114044, "global_step": 42477, "epoch": 400, "lr": 9.413702564204435e-06} +{"train_loss": 0.012287608347833157, "grad_norm": 0.09246833622455597, "global_step": 42478, "epoch": 400, "lr": 9.411971681638315e-06} +{"train_loss": 0.014773773960769176, "grad_norm": 0.06433955579996109, "global_step": 42479, "epoch": 400, "lr": 9.410240941680859e-06} +{"train_loss": 0.013766469433903694, "grad_norm": 0.09981066733598709, "global_step": 42480, "epoch": 400, "lr": 9.408510344338167e-06} +{"train_loss": 0.013007503002882004, "grad_norm": 0.09147049486637115, "global_step": 42481, "epoch": 400, "lr": 9.40677988961633e-06} +{"train_loss": 0.013785529881715775, "grad_norm": 0.13267074525356293, "global_step": 42482, "epoch": 400, "lr": 9.405049577521413e-06} +{"train_loss": 0.014057327061891556, "grad_norm": 0.08496024459600449, "global_step": 42483, "epoch": 400, "lr": 9.403319408059485e-06} +{"train_loss": 0.017291933298110962, "grad_norm": 0.05537937954068184, "global_step": 42484, "epoch": 400, "lr": 9.401589381236653e-06} +{"train_loss": 0.014192391186952591, "grad_norm": 0.09857445955276489, "global_step": 42485, "epoch": 400, "lr": 9.399859497058967e-06} +{"train_loss": 0.013960132375359535, "grad_norm": 0.09709946066141129, "global_step": 42486, "epoch": 400, "lr": 9.39812975553252e-06} +{"train_loss": 0.014694321900606155, "grad_norm": 0.07449638098478317, "global_step": 42487, "epoch": 400, "lr": 9.396400156663382e-06} +{"train_loss": 0.015081264078617096, "grad_norm": 0.07378166168928146, "global_step": 42488, "epoch": 400, "lr": 9.394670700457665e-06} +{"train_loss": 0.014645416289567947, "grad_norm": 0.10203015059232712, "global_step": 42489, "epoch": 400, "lr": 9.39294138692139e-06} +{"train_loss": 0.015390239655971527, "grad_norm": 0.0949515849351883, "global_step": 42490, "epoch": 400, "lr": 9.391212216060658e-06} +{"train_loss": 0.013439719565212727, "grad_norm": 0.09623735398054123, "global_step": 42491, "epoch": 400, "lr": 9.38948318788156e-06} +{"train_loss": 0.016184531152248383, "grad_norm": 0.08676357567310333, "global_step": 42492, "epoch": 400, "lr": 9.387754302390145e-06} +{"train_loss": 0.016403138637542725, "grad_norm": 0.07479041069746017, "global_step": 42493, "epoch": 400, "lr": 9.3860255595925e-06} +{"train_loss": 0.013500289991497993, "grad_norm": 0.11244018375873566, "global_step": 42494, "epoch": 400, "lr": 9.38429695949471e-06} +{"train_loss": 0.015510772354900837, "grad_norm": 0.0732005313038826, "global_step": 42495, "epoch": 400, "lr": 9.38256850210284e-06} +{"train_loss": 0.016391601413488388, "grad_norm": 0.0932927131652832, "global_step": 42496, "epoch": 400, "lr": 9.38084018742294e-06} +{"train_loss": 0.016902314499020576, "grad_norm": 0.11373983323574066, "global_step": 42497, "epoch": 400, "lr": 9.379112015461112e-06} +{"train_loss": 0.015619105659425259, "grad_norm": 0.09068979322910309, "global_step": 42498, "epoch": 400, "lr": 9.377383986223426e-06} +{"train_loss": 0.013624889776110649, "grad_norm": 0.10991647094488144, "global_step": 42499, "epoch": 400, "lr": 9.375656099715934e-06} +{"train_loss": 0.014861276373267174, "grad_norm": 0.10103655606508255, "global_step": 42500, "epoch": 400, "lr": 9.373928355944734e-06} +{"train_loss": 0.014568716287612915, "grad_norm": 0.12187222391366959, "global_step": 42501, "epoch": 400, "lr": 9.372200754915866e-06} +{"train_loss": 0.01643216609954834, "grad_norm": 0.16231383383274078, "global_step": 42502, "epoch": 400, "lr": 9.370473296635435e-06} +{"train_loss": 0.014465649612247944, "grad_norm": 0.0778091698884964, "global_step": 42503, "epoch": 400, "lr": 9.368745981109473e-06} +{"train_loss": 0.015503820031881332, "grad_norm": 0.07608287781476974, "global_step": 42504, "epoch": 400, "lr": 9.36701880834407e-06} +{"train_loss": 0.01500519415152804, "grad_norm": 0.09544723480939865, "global_step": 42505, "epoch": 400, "lr": 9.365291778345303e-06, "train_batch_mse_error": 0.01340741291642189, "val_loss": 0.019507281482219696} +{"train_loss": 0.014445435255765915, "grad_norm": 0.09224049746990204, "global_step": 42506, "epoch": 401, "lr": 9.363564891119231e-06} +{"train_loss": 0.014323899522423744, "grad_norm": 0.07425647228956223, "global_step": 42507, "epoch": 401, "lr": 9.361838146671908e-06} +{"train_loss": 0.014886450953781605, "grad_norm": 0.059123694896698, "global_step": 42508, "epoch": 401, "lr": 9.360111545009425e-06} +{"train_loss": 0.014734853059053421, "grad_norm": 0.06408633291721344, "global_step": 42509, "epoch": 401, "lr": 9.358385086137822e-06} +{"train_loss": 0.01362469419836998, "grad_norm": 0.16328111290931702, "global_step": 42510, "epoch": 401, "lr": 9.356658770063177e-06} +{"train_loss": 0.01799698919057846, "grad_norm": 0.0938485860824585, "global_step": 42511, "epoch": 401, "lr": 9.354932596791571e-06} +{"train_loss": 0.014655454084277153, "grad_norm": 0.11228809505701065, "global_step": 42512, "epoch": 401, "lr": 9.35320656632906e-06} +{"train_loss": 0.01612037792801857, "grad_norm": 0.07788525521755219, "global_step": 42513, "epoch": 401, "lr": 9.35148067868169e-06} +{"train_loss": 0.017178736627101898, "grad_norm": 0.14077194035053253, "global_step": 42514, "epoch": 401, "lr": 9.349754933855536e-06} +{"train_loss": 0.01198733039200306, "grad_norm": 0.08695171028375626, "global_step": 42515, "epoch": 401, "lr": 9.348029331856684e-06} +{"train_loss": 0.013431366533041, "grad_norm": 0.08683052659034729, "global_step": 42516, "epoch": 401, "lr": 9.34630387269116e-06} +{"train_loss": 0.013600097969174385, "grad_norm": 0.1286008059978485, "global_step": 42517, "epoch": 401, "lr": 9.344578556365047e-06} +{"train_loss": 0.014791660010814667, "grad_norm": 0.11812017858028412, "global_step": 42518, "epoch": 401, "lr": 9.342853382884415e-06} +{"train_loss": 0.014160551130771637, "grad_norm": 0.0752716213464737, "global_step": 42519, "epoch": 401, "lr": 9.341128352255313e-06} +{"train_loss": 0.013200294226408005, "grad_norm": 0.08273813128471375, "global_step": 42520, "epoch": 401, "lr": 9.339403464483799e-06} +{"train_loss": 0.019137009978294373, "grad_norm": 0.10327545553445816, "global_step": 42521, "epoch": 401, "lr": 9.337678719575932e-06} +{"train_loss": 0.013977402821183205, "grad_norm": 0.06816808134317398, "global_step": 42522, "epoch": 401, "lr": 9.335954117537793e-06} +{"train_loss": 0.013261523097753525, "grad_norm": 0.08004007488489151, "global_step": 42523, "epoch": 401, "lr": 9.33422965837541e-06} +{"train_loss": 0.015276243910193443, "grad_norm": 0.07713685929775238, "global_step": 42524, "epoch": 401, "lr": 9.332505342094877e-06} +{"train_loss": 0.014935796149075031, "grad_norm": 0.07575865089893341, "global_step": 42525, "epoch": 401, "lr": 9.33078116870223e-06} +{"train_loss": 0.013163981959223747, "grad_norm": 0.07863471657037735, "global_step": 42526, "epoch": 401, "lr": 9.329057138203523e-06} +{"train_loss": 0.014686539769172668, "grad_norm": 0.11801797896623611, "global_step": 42527, "epoch": 401, "lr": 9.32733325060482e-06} +{"train_loss": 0.01379598118364811, "grad_norm": 0.0684623196721077, "global_step": 42528, "epoch": 401, "lr": 9.325609505912192e-06} +{"train_loss": 0.016954606398940086, "grad_norm": 0.06588438153266907, "global_step": 42529, "epoch": 401, "lr": 9.323885904131673e-06} +{"train_loss": 0.015189867466688156, "grad_norm": 0.1181497797369957, "global_step": 42530, "epoch": 401, "lr": 9.322162445269338e-06} +{"train_loss": 0.01471604686230421, "grad_norm": 0.07698075473308563, "global_step": 42531, "epoch": 401, "lr": 9.320439129331226e-06} +{"train_loss": 0.016437452286481857, "grad_norm": 0.07300572842359543, "global_step": 42532, "epoch": 401, "lr": 9.31871595632341e-06} +{"train_loss": 0.01275162398815155, "grad_norm": 0.12668445706367493, "global_step": 42533, "epoch": 401, "lr": 9.316992926251921e-06} +{"train_loss": 0.014560097828507423, "grad_norm": 0.08413903415203094, "global_step": 42534, "epoch": 401, "lr": 9.315270039122826e-06} +{"train_loss": 0.013809355907142162, "grad_norm": 0.07221979647874832, "global_step": 42535, "epoch": 401, "lr": 9.313547294942193e-06} +{"train_loss": 0.013128080405294895, "grad_norm": 0.08657743036746979, "global_step": 42536, "epoch": 401, "lr": 9.311824693716059e-06} +{"train_loss": 0.014447759836912155, "grad_norm": 0.07408563047647476, "global_step": 42537, "epoch": 401, "lr": 9.310102235450468e-06} +{"train_loss": 0.013930059969425201, "grad_norm": 0.06080440804362297, "global_step": 42538, "epoch": 401, "lr": 9.308379920151484e-06} +{"train_loss": 0.014650776982307434, "grad_norm": 0.07583603262901306, "global_step": 42539, "epoch": 401, "lr": 9.306657747825165e-06} +{"train_loss": 0.01334192231297493, "grad_norm": 0.09095904976129532, "global_step": 42540, "epoch": 401, "lr": 9.304935718477543e-06} +{"train_loss": 0.014339003711938858, "grad_norm": 0.0636407881975174, "global_step": 42541, "epoch": 401, "lr": 9.303213832114688e-06} +{"train_loss": 0.014495640061795712, "grad_norm": 0.08337681740522385, "global_step": 42542, "epoch": 401, "lr": 9.301492088742631e-06} +{"train_loss": 0.01831941306591034, "grad_norm": 0.08438051491975784, "global_step": 42543, "epoch": 401, "lr": 9.299770488367449e-06} +{"train_loss": 0.013346996158361435, "grad_norm": 0.0902571752667427, "global_step": 42544, "epoch": 401, "lr": 9.298049030995154e-06} +{"train_loss": 0.01612536795437336, "grad_norm": 0.10326898097991943, "global_step": 42545, "epoch": 401, "lr": 9.296327716631813e-06} +{"train_loss": 0.012757701799273491, "grad_norm": 0.09596726298332214, "global_step": 42546, "epoch": 401, "lr": 9.294606545283491e-06} +{"train_loss": 0.014227653853595257, "grad_norm": 0.0887245237827301, "global_step": 42547, "epoch": 401, "lr": 9.292885516956223e-06} +{"train_loss": 0.015766333788633347, "grad_norm": 0.1497088074684143, "global_step": 42548, "epoch": 401, "lr": 9.291164631656035e-06} +{"train_loss": 0.015535250306129456, "grad_norm": 0.060162052512168884, "global_step": 42549, "epoch": 401, "lr": 9.289443889389005e-06} +{"train_loss": 0.015468573197722435, "grad_norm": 0.08147912472486496, "global_step": 42550, "epoch": 401, "lr": 9.287723290161148e-06} +{"train_loss": 0.014351978898048401, "grad_norm": 0.09113233536481857, "global_step": 42551, "epoch": 401, "lr": 9.28600283397853e-06} +{"train_loss": 0.012337920255959034, "grad_norm": 0.05015707015991211, "global_step": 42552, "epoch": 401, "lr": 9.284282520847204e-06} +{"train_loss": 0.016897939145565033, "grad_norm": 0.05968538299202919, "global_step": 42553, "epoch": 401, "lr": 9.2825623507732e-06} +{"train_loss": 0.01554609090089798, "grad_norm": 0.11575677990913391, "global_step": 42554, "epoch": 401, "lr": 9.280842323762556e-06} +{"train_loss": 0.01188827957957983, "grad_norm": 0.08783738315105438, "global_step": 42555, "epoch": 401, "lr": 9.27912243982132e-06} +{"train_loss": 0.015053743496537209, "grad_norm": 0.06936468183994293, "global_step": 42556, "epoch": 401, "lr": 9.277402698955551e-06} +{"train_loss": 0.014279857277870178, "grad_norm": 0.09353451430797577, "global_step": 42557, "epoch": 401, "lr": 9.275683101171267e-06} +{"train_loss": 0.01873142644762993, "grad_norm": 0.11556743085384369, "global_step": 42558, "epoch": 401, "lr": 9.273963646474526e-06} +{"train_loss": 0.012637969106435776, "grad_norm": 0.06581471115350723, "global_step": 42559, "epoch": 401, "lr": 9.272244334871377e-06} +{"train_loss": 0.01677451655268669, "grad_norm": 0.09963611513376236, "global_step": 42560, "epoch": 401, "lr": 9.270525166367844e-06} +{"train_loss": 0.013954861089587212, "grad_norm": 0.05521427094936371, "global_step": 42561, "epoch": 401, "lr": 9.268806140969965e-06} +{"train_loss": 0.015147747471928596, "grad_norm": 0.12454704940319061, "global_step": 42562, "epoch": 401, "lr": 9.267087258683787e-06} +{"train_loss": 0.01580999791622162, "grad_norm": 0.08532240241765976, "global_step": 42563, "epoch": 401, "lr": 9.265368519515361e-06} +{"train_loss": 0.01747295819222927, "grad_norm": 0.08615632355213165, "global_step": 42564, "epoch": 401, "lr": 9.263649923470708e-06} +{"train_loss": 0.016265248879790306, "grad_norm": 0.10392609983682632, "global_step": 42565, "epoch": 401, "lr": 9.26193147055588e-06} +{"train_loss": 0.01492992602288723, "grad_norm": 0.0977468192577362, "global_step": 42566, "epoch": 401, "lr": 9.26021316077691e-06} +{"train_loss": 0.016313018277287483, "grad_norm": 0.14446169137954712, "global_step": 42567, "epoch": 401, "lr": 9.258494994139822e-06} +{"train_loss": 0.013155167922377586, "grad_norm": 0.0649011880159378, "global_step": 42568, "epoch": 401, "lr": 9.25677697065066e-06} +{"train_loss": 0.014557862654328346, "grad_norm": 0.06426338851451874, "global_step": 42569, "epoch": 401, "lr": 9.255059090315483e-06} +{"train_loss": 0.013667517341673374, "grad_norm": 0.0709698423743248, "global_step": 42570, "epoch": 401, "lr": 9.253341353140293e-06} +{"train_loss": 0.01470260415226221, "grad_norm": 0.06529790908098221, "global_step": 42571, "epoch": 401, "lr": 9.251623759131155e-06} +{"train_loss": 0.01405926700681448, "grad_norm": 0.09389863163232803, "global_step": 42572, "epoch": 401, "lr": 9.249906308294081e-06} +{"train_loss": 0.016587309539318085, "grad_norm": 0.06995422393083572, "global_step": 42573, "epoch": 401, "lr": 9.248189000635121e-06} +{"train_loss": 0.01714644953608513, "grad_norm": 0.11709421873092651, "global_step": 42574, "epoch": 401, "lr": 9.246471836160293e-06} +{"train_loss": 0.016531476750969887, "grad_norm": 0.11061414331197739, "global_step": 42575, "epoch": 401, "lr": 9.244754814875639e-06} +{"train_loss": 0.017031867057085037, "grad_norm": 0.053314149379730225, "global_step": 42576, "epoch": 401, "lr": 9.243037936787209e-06} +{"train_loss": 0.015174944885075092, "grad_norm": 0.09652161598205566, "global_step": 42577, "epoch": 401, "lr": 9.241321201901016e-06} +{"train_loss": 0.014111258089542389, "grad_norm": 0.06593379378318787, "global_step": 42578, "epoch": 401, "lr": 9.239604610223079e-06} +{"train_loss": 0.012248463928699493, "grad_norm": 0.07250364869832993, "global_step": 42579, "epoch": 401, "lr": 9.23788816175945e-06} +{"train_loss": 0.017175616696476936, "grad_norm": 0.08216521143913269, "global_step": 42580, "epoch": 401, "lr": 9.236171856516163e-06} +{"train_loss": 0.015386475250124931, "grad_norm": 0.06824686378240585, "global_step": 42581, "epoch": 401, "lr": 9.234455694499233e-06} +{"train_loss": 0.017043408006429672, "grad_norm": 0.10368416458368301, "global_step": 42582, "epoch": 401, "lr": 9.232739675714708e-06} +{"train_loss": 0.01527299452573061, "grad_norm": 0.06076158210635185, "global_step": 42583, "epoch": 401, "lr": 9.231023800168603e-06} +{"train_loss": 0.012791464105248451, "grad_norm": 0.08146119117736816, "global_step": 42584, "epoch": 401, "lr": 9.229308067866943e-06} +{"train_loss": 0.018939761444926262, "grad_norm": 0.11262445151805878, "global_step": 42585, "epoch": 401, "lr": 9.227592478815761e-06} +{"train_loss": 0.015708670020103455, "grad_norm": 0.08227145671844482, "global_step": 42586, "epoch": 401, "lr": 9.225877033021096e-06} +{"train_loss": 0.011407031677663326, "grad_norm": 0.11530068516731262, "global_step": 42587, "epoch": 401, "lr": 9.224161730488961e-06} +{"train_loss": 0.01857050508260727, "grad_norm": 0.111578568816185, "global_step": 42588, "epoch": 401, "lr": 9.222446571225402e-06} +{"train_loss": 0.01644759066402912, "grad_norm": 0.10507124662399292, "global_step": 42589, "epoch": 401, "lr": 9.220731555236412e-06} +{"train_loss": 0.012466580606997013, "grad_norm": 0.1379481554031372, "global_step": 42590, "epoch": 401, "lr": 9.21901668252806e-06} +{"train_loss": 0.012252746149897575, "grad_norm": 0.06204276531934738, "global_step": 42591, "epoch": 401, "lr": 9.217301953106328e-06} +{"train_loss": 0.01402339618653059, "grad_norm": 0.09291984885931015, "global_step": 42592, "epoch": 401, "lr": 9.215587366977263e-06} +{"train_loss": 0.01853552833199501, "grad_norm": 0.10912727564573288, "global_step": 42593, "epoch": 401, "lr": 9.2138729241469e-06} +{"train_loss": 0.014928599819540977, "grad_norm": 0.09830481559038162, "global_step": 42594, "epoch": 401, "lr": 9.212158624621248e-06} +{"train_loss": 0.016894295811653137, "grad_norm": 0.12028888612985611, "global_step": 42595, "epoch": 401, "lr": 9.21044446840632e-06} +{"train_loss": 0.01671304926276207, "grad_norm": 0.07148025184869766, "global_step": 42596, "epoch": 401, "lr": 9.208730455508152e-06} +{"train_loss": 0.013728621415793896, "grad_norm": 0.10141710191965103, "global_step": 42597, "epoch": 401, "lr": 9.207016585932782e-06} +{"train_loss": 0.014880646951496601, "grad_norm": 0.058834679424762726, "global_step": 42598, "epoch": 401, "lr": 9.205302859686198e-06} +{"train_loss": 0.017599139362573624, "grad_norm": 0.10280146449804306, "global_step": 42599, "epoch": 401, "lr": 9.203589276774439e-06} +{"train_loss": 0.016993213444948196, "grad_norm": 0.0684814304113388, "global_step": 42600, "epoch": 401, "lr": 9.201875837203539e-06} +{"train_loss": 0.01637972891330719, "grad_norm": 0.08769414573907852, "global_step": 42601, "epoch": 401, "lr": 9.200162540979506e-06} +{"train_loss": 0.016829825937747955, "grad_norm": 0.1308431327342987, "global_step": 42602, "epoch": 401, "lr": 9.198449388108344e-06} +{"train_loss": 0.015414947643876076, "grad_norm": 0.09781580418348312, "global_step": 42603, "epoch": 401, "lr": 9.196736378596088e-06} +{"train_loss": 0.012070303782820702, "grad_norm": 0.07557535916566849, "global_step": 42604, "epoch": 401, "lr": 9.195023512448769e-06} +{"train_loss": 0.014788385480642319, "grad_norm": 0.07140368223190308, "global_step": 42605, "epoch": 401, "lr": 9.193310789672377e-06} +{"train_loss": 0.01232738047838211, "grad_norm": 0.09165888279676437, "global_step": 42606, "epoch": 401, "lr": 9.191598210272956e-06} +{"train_loss": 0.01638224720954895, "grad_norm": 0.1308526247739792, "global_step": 42607, "epoch": 401, "lr": 9.189885774256512e-06} +{"train_loss": 0.014938421547412872, "grad_norm": 0.06095650792121887, "global_step": 42608, "epoch": 401, "lr": 9.188173481629053e-06} +{"train_loss": 0.015269432216882706, "grad_norm": 0.07520078122615814, "global_step": 42609, "epoch": 401, "lr": 9.1864613323966e-06} +{"train_loss": 0.014910290949046612, "grad_norm": 0.06399908661842346, "global_step": 42610, "epoch": 401, "lr": 9.18474932656519e-06} +{"train_loss": 0.015000508957595195, "grad_norm": 0.10676644742488861, "global_step": 42611, "epoch": 401, "lr": 9.183037464140804e-06, "val_loss": 0.019545404240489006} +{"train_loss": 0.01564692333340645, "grad_norm": 0.06654728949069977, "global_step": 42612, "epoch": 402, "lr": 9.181325745129488e-06} +{"train_loss": 0.014719071798026562, "grad_norm": 0.0754924938082695, "global_step": 42613, "epoch": 402, "lr": 9.179614169537227e-06} +{"train_loss": 0.018381746485829353, "grad_norm": 0.06444396823644638, "global_step": 42614, "epoch": 402, "lr": 9.17790273737006e-06} +{"train_loss": 0.014905491843819618, "grad_norm": 0.08989860117435455, "global_step": 42615, "epoch": 402, "lr": 9.176191448633981e-06} +{"train_loss": 0.013542593456804752, "grad_norm": 0.0716046541929245, "global_step": 42616, "epoch": 402, "lr": 9.17448030333501e-06} +{"train_loss": 0.013702990487217903, "grad_norm": 0.05417979881167412, "global_step": 42617, "epoch": 402, "lr": 9.172769301479173e-06} +{"train_loss": 0.014670170843601227, "grad_norm": 0.06360966712236404, "global_step": 42618, "epoch": 402, "lr": 9.171058443072472e-06} +{"train_loss": 0.018779419362545013, "grad_norm": 0.09609610587358475, "global_step": 42619, "epoch": 402, "lr": 9.169347728120903e-06} +{"train_loss": 0.01399361714720726, "grad_norm": 0.07322781533002853, "global_step": 42620, "epoch": 402, "lr": 9.167637156630488e-06} +{"train_loss": 0.016308050602674484, "grad_norm": 0.18612930178642273, "global_step": 42621, "epoch": 402, "lr": 9.165926728607249e-06} +{"train_loss": 0.014543913304805756, "grad_norm": 0.1331019252538681, "global_step": 42622, "epoch": 402, "lr": 9.164216444057172e-06} +{"train_loss": 0.017497114837169647, "grad_norm": 0.21613101661205292, "global_step": 42623, "epoch": 402, "lr": 9.162506302986284e-06} +{"train_loss": 0.012988081201910973, "grad_norm": 0.09291037172079086, "global_step": 42624, "epoch": 402, "lr": 9.16079630540061e-06} +{"train_loss": 0.014145677909255028, "grad_norm": 0.10026420652866364, "global_step": 42625, "epoch": 402, "lr": 9.159086451306115e-06} +{"train_loss": 0.017109474167227745, "grad_norm": 0.06706593185663223, "global_step": 42626, "epoch": 402, "lr": 9.157376740708829e-06} +{"train_loss": 0.01464170403778553, "grad_norm": 0.08423765748739243, "global_step": 42627, "epoch": 402, "lr": 9.155667173614769e-06} +{"train_loss": 0.014097839593887329, "grad_norm": 0.08382444828748703, "global_step": 42628, "epoch": 402, "lr": 9.153957750029924e-06} +{"train_loss": 0.01246696338057518, "grad_norm": 0.08821634948253632, "global_step": 42629, "epoch": 402, "lr": 9.152248469960301e-06} +{"train_loss": 0.014642300084233284, "grad_norm": 0.11876699328422546, "global_step": 42630, "epoch": 402, "lr": 9.150539333411934e-06} +{"train_loss": 0.013259295374155045, "grad_norm": 0.07440032064914703, "global_step": 42631, "epoch": 402, "lr": 9.148830340390801e-06} +{"train_loss": 0.013510293327271938, "grad_norm": 0.06258850544691086, "global_step": 42632, "epoch": 402, "lr": 9.1471214909029e-06} +{"train_loss": 0.016600314527750015, "grad_norm": 0.06034289672970772, "global_step": 42633, "epoch": 402, "lr": 9.14541278495425e-06} +{"train_loss": 0.016023900359869003, "grad_norm": 0.1074117124080658, "global_step": 42634, "epoch": 402, "lr": 9.143704222550864e-06} +{"train_loss": 0.014982311055064201, "grad_norm": 0.07597474753856659, "global_step": 42635, "epoch": 402, "lr": 9.141995803698733e-06} +{"train_loss": 0.016061022877693176, "grad_norm": 0.12352053076028824, "global_step": 42636, "epoch": 402, "lr": 9.140287528403846e-06} +{"train_loss": 0.016258925199508667, "grad_norm": 0.07727178931236267, "global_step": 42637, "epoch": 402, "lr": 9.13857939667222e-06} +{"train_loss": 0.012652426958084106, "grad_norm": 0.07074543833732605, "global_step": 42638, "epoch": 402, "lr": 9.136871408509867e-06} +{"train_loss": 0.015839958563447, "grad_norm": 0.08743676543235779, "global_step": 42639, "epoch": 402, "lr": 9.135163563922766e-06} +{"train_loss": 0.012804921716451645, "grad_norm": 0.05342443659901619, "global_step": 42640, "epoch": 402, "lr": 9.13345586291693e-06} +{"train_loss": 0.01441040076315403, "grad_norm": 0.05414886772632599, "global_step": 42641, "epoch": 402, "lr": 9.131748305498383e-06} +{"train_loss": 0.016338050365447998, "grad_norm": 0.10298164933919907, "global_step": 42642, "epoch": 402, "lr": 9.130040891673074e-06} +{"train_loss": 0.013634053990244865, "grad_norm": 0.09329546242952347, "global_step": 42643, "epoch": 402, "lr": 9.128333621447027e-06} +{"train_loss": 0.01467216294258833, "grad_norm": 0.10793031752109528, "global_step": 42644, "epoch": 402, "lr": 9.12662649482625e-06} +{"train_loss": 0.014624129980802536, "grad_norm": 0.06726562976837158, "global_step": 42645, "epoch": 402, "lr": 9.124919511816726e-06} +{"train_loss": 0.01340770348906517, "grad_norm": 0.062139660120010376, "global_step": 42646, "epoch": 402, "lr": 9.123212672424453e-06} +{"train_loss": 0.016071729362010956, "grad_norm": 0.10028615593910217, "global_step": 42647, "epoch": 402, "lr": 9.121505976655447e-06} +{"train_loss": 0.014780358411371708, "grad_norm": 0.07868234068155289, "global_step": 42648, "epoch": 402, "lr": 9.119799424515696e-06} +{"train_loss": 0.014550645835697651, "grad_norm": 0.10626883804798126, "global_step": 42649, "epoch": 402, "lr": 9.118093016011175e-06} +{"train_loss": 0.014551110565662384, "grad_norm": 0.07672679424285889, "global_step": 42650, "epoch": 402, "lr": 9.116386751147899e-06} +{"train_loss": 0.015766175463795662, "grad_norm": 0.066102534532547, "global_step": 42651, "epoch": 402, "lr": 9.11468062993187e-06} +{"train_loss": 0.013864943757653236, "grad_norm": 0.059115514159202576, "global_step": 42652, "epoch": 402, "lr": 9.112974652369061e-06} +{"train_loss": 0.015260430052876472, "grad_norm": 0.07099330425262451, "global_step": 42653, "epoch": 402, "lr": 9.11126881846549e-06} +{"train_loss": 0.014695519581437111, "grad_norm": 0.17628586292266846, "global_step": 42654, "epoch": 402, "lr": 9.109563128227128e-06} +{"train_loss": 0.013919820077717304, "grad_norm": 0.09583248943090439, "global_step": 42655, "epoch": 402, "lr": 9.107857581659985e-06} +{"train_loss": 0.014805166982114315, "grad_norm": 0.0748182013630867, "global_step": 42656, "epoch": 402, "lr": 9.106152178770039e-06} +{"train_loss": 0.015053413808345795, "grad_norm": 0.07824063301086426, "global_step": 42657, "epoch": 402, "lr": 9.104446919563292e-06} +{"train_loss": 0.01542949490249157, "grad_norm": 0.13169759511947632, "global_step": 42658, "epoch": 402, "lr": 9.10274180404574e-06} +{"train_loss": 0.014542755670845509, "grad_norm": 0.06634482741355896, "global_step": 42659, "epoch": 402, "lr": 9.101036832223376e-06} +{"train_loss": 0.015266431495547295, "grad_norm": 0.08569475263357162, "global_step": 42660, "epoch": 402, "lr": 9.099332004102162e-06} +{"train_loss": 0.018989069387316704, "grad_norm": 0.060609087347984314, "global_step": 42661, "epoch": 402, "lr": 9.097627319688113e-06} +{"train_loss": 0.015537906438112259, "grad_norm": 0.10479956120252609, "global_step": 42662, "epoch": 402, "lr": 9.09592277898722e-06} +{"train_loss": 0.014557022601366043, "grad_norm": 0.06701050698757172, "global_step": 42663, "epoch": 402, "lr": 9.09421838200546e-06} +{"train_loss": 0.014095032587647438, "grad_norm": 0.06288706511259079, "global_step": 42664, "epoch": 402, "lr": 9.09251412874882e-06} +{"train_loss": 0.015021344646811485, "grad_norm": 0.10614300519227982, "global_step": 42665, "epoch": 402, "lr": 9.090810019223328e-06} +{"train_loss": 0.015450471080839634, "grad_norm": 0.10778927057981491, "global_step": 42666, "epoch": 402, "lr": 9.08910605343491e-06} +{"train_loss": 0.01532904151827097, "grad_norm": 0.054936155676841736, "global_step": 42667, "epoch": 402, "lr": 9.08740223138958e-06} +{"train_loss": 0.0151872169226408, "grad_norm": 0.12007226794958115, "global_step": 42668, "epoch": 402, "lr": 9.08569855309334e-06} +{"train_loss": 0.01403624378144741, "grad_norm": 0.09179869294166565, "global_step": 42669, "epoch": 402, "lr": 9.083995018552149e-06} +{"train_loss": 0.015142889693379402, "grad_norm": 0.08703653514385223, "global_step": 42670, "epoch": 402, "lr": 9.082291627772004e-06} +{"train_loss": 0.013598717749118805, "grad_norm": 0.07429714500904083, "global_step": 42671, "epoch": 402, "lr": 9.080588380758909e-06} +{"train_loss": 0.013929765671491623, "grad_norm": 0.06651502847671509, "global_step": 42672, "epoch": 402, "lr": 9.078885277518828e-06} +{"train_loss": 0.016410043463110924, "grad_norm": 0.0718454122543335, "global_step": 42673, "epoch": 402, "lr": 9.077182318057737e-06} +{"train_loss": 0.013288173824548721, "grad_norm": 0.08319208025932312, "global_step": 42674, "epoch": 402, "lr": 9.075479502381628e-06} +{"train_loss": 0.01323721744120121, "grad_norm": 0.07171120494604111, "global_step": 42675, "epoch": 402, "lr": 9.0737768304965e-06} +{"train_loss": 0.014027515426278114, "grad_norm": 0.07734392583370209, "global_step": 42676, "epoch": 402, "lr": 9.072074302408312e-06} +{"train_loss": 0.01610569655895233, "grad_norm": 0.07621543109416962, "global_step": 42677, "epoch": 402, "lr": 9.070371918123067e-06} +{"train_loss": 0.013666432350873947, "grad_norm": 0.06290413439273834, "global_step": 42678, "epoch": 402, "lr": 9.068669677646719e-06} +{"train_loss": 0.01488522905856371, "grad_norm": 0.12876810133457184, "global_step": 42679, "epoch": 402, "lr": 9.06696758098528e-06} +{"train_loss": 0.015831096097826958, "grad_norm": 0.06043320894241333, "global_step": 42680, "epoch": 402, "lr": 9.06526562814471e-06} +{"train_loss": 0.01395803689956665, "grad_norm": 0.13264378905296326, "global_step": 42681, "epoch": 402, "lr": 9.063563819130982e-06} +{"train_loss": 0.014057885855436325, "grad_norm": 0.06390578299760818, "global_step": 42682, "epoch": 402, "lr": 9.061862153950107e-06} +{"train_loss": 0.012657858431339264, "grad_norm": 0.0953681617975235, "global_step": 42683, "epoch": 402, "lr": 9.060160632608045e-06} +{"train_loss": 0.013490467332303524, "grad_norm": 0.06107382848858833, "global_step": 42684, "epoch": 402, "lr": 9.058459255110763e-06} +{"train_loss": 0.015251124277710915, "grad_norm": 0.09400483965873718, "global_step": 42685, "epoch": 402, "lr": 9.056758021464262e-06} +{"train_loss": 0.012535309419035912, "grad_norm": 0.06995343416929245, "global_step": 42686, "epoch": 402, "lr": 9.055056931674494e-06} +{"train_loss": 0.013946295715868473, "grad_norm": 0.05209176987409592, "global_step": 42687, "epoch": 402, "lr": 9.053355985747448e-06} +{"train_loss": 0.01586398296058178, "grad_norm": 0.0839647725224495, "global_step": 42688, "epoch": 402, "lr": 9.051655183689117e-06} +{"train_loss": 0.013048248365521431, "grad_norm": 0.05914505571126938, "global_step": 42689, "epoch": 402, "lr": 9.04995452550546e-06} +{"train_loss": 0.016580350697040558, "grad_norm": 0.06700216233730316, "global_step": 42690, "epoch": 402, "lr": 9.048254011202445e-06} +{"train_loss": 0.014158172532916069, "grad_norm": 0.08864912390708923, "global_step": 42691, "epoch": 402, "lr": 9.046553640786048e-06} +{"train_loss": 0.01813974231481552, "grad_norm": 0.13044679164886475, "global_step": 42692, "epoch": 402, "lr": 9.044853414262271e-06} +{"train_loss": 0.01660297065973282, "grad_norm": 0.07561909407377243, "global_step": 42693, "epoch": 402, "lr": 9.043153331637055e-06} +{"train_loss": 0.014721367508172989, "grad_norm": 0.06904705613851547, "global_step": 42694, "epoch": 402, "lr": 9.0414533929164e-06} +{"train_loss": 0.015420468524098396, "grad_norm": 0.10261262953281403, "global_step": 42695, "epoch": 402, "lr": 9.039753598106248e-06} +{"train_loss": 0.014862417243421078, "grad_norm": 0.08061382919549942, "global_step": 42696, "epoch": 402, "lr": 9.038053947212605e-06} +{"train_loss": 0.01573239080607891, "grad_norm": 0.09522263705730438, "global_step": 42697, "epoch": 402, "lr": 9.036354440241413e-06} +{"train_loss": 0.015289575792849064, "grad_norm": 0.1274840533733368, "global_step": 42698, "epoch": 402, "lr": 9.034655077198656e-06} +{"train_loss": 0.016089096665382385, "grad_norm": 0.07224897295236588, "global_step": 42699, "epoch": 402, "lr": 9.03295585809032e-06} +{"train_loss": 0.014572877436876297, "grad_norm": 0.08401654660701752, "global_step": 42700, "epoch": 402, "lr": 9.031256782922359e-06} +{"train_loss": 0.017756158486008644, "grad_norm": 0.1398259848356247, "global_step": 42701, "epoch": 402, "lr": 9.029557851700732e-06} +{"train_loss": 0.016072724014520645, "grad_norm": 0.08721943199634552, "global_step": 42702, "epoch": 402, "lr": 9.027859064431421e-06} +{"train_loss": 0.01569867506623268, "grad_norm": 0.09423428773880005, "global_step": 42703, "epoch": 402, "lr": 9.026160421120411e-06} +{"train_loss": 0.016236955299973488, "grad_norm": 0.061727024614810944, "global_step": 42704, "epoch": 402, "lr": 9.024461921773642e-06} +{"train_loss": 0.017318055033683777, "grad_norm": 0.08207565546035767, "global_step": 42705, "epoch": 402, "lr": 9.022763566397091e-06} +{"train_loss": 0.012662352994084358, "grad_norm": 0.06457913666963577, "global_step": 42706, "epoch": 402, "lr": 9.02106535499675e-06} +{"train_loss": 0.014390163123607635, "grad_norm": 0.08574520796537399, "global_step": 42707, "epoch": 402, "lr": 9.019367287578545e-06} +{"train_loss": 0.013937913812696934, "grad_norm": 0.10495860129594803, "global_step": 42708, "epoch": 402, "lr": 9.017669364148456e-06} +{"train_loss": 0.015510227531194687, "grad_norm": 0.0948244035243988, "global_step": 42709, "epoch": 402, "lr": 9.015971584712473e-06} +{"train_loss": 0.01565174013376236, "grad_norm": 0.07662095129489899, "global_step": 42710, "epoch": 402, "lr": 9.014273949276525e-06} +{"train_loss": 0.015285618603229523, "grad_norm": 0.07819585502147675, "global_step": 42711, "epoch": 402, "lr": 9.0125764578466e-06} +{"train_loss": 0.015586183406412601, "grad_norm": 0.08189758658409119, "global_step": 42712, "epoch": 402, "lr": 9.01087911042866e-06} +{"train_loss": 0.014662586152553558, "grad_norm": 0.11251933872699738, "global_step": 42713, "epoch": 402, "lr": 9.009181907028674e-06} +{"train_loss": 0.014335254207253456, "grad_norm": 0.051645826548337936, "global_step": 42714, "epoch": 402, "lr": 9.007484847652575e-06} +{"train_loss": 0.015117457136511803, "grad_norm": 0.10166850686073303, "global_step": 42715, "epoch": 402, "lr": 9.005787932306353e-06} +{"train_loss": 0.017990602180361748, "grad_norm": 0.078715480864048, "global_step": 42716, "epoch": 402, "lr": 9.004091160995975e-06} +{"train_loss": 0.01500220823470714, "grad_norm": 0.13728506863117218, "global_step": 42717, "epoch": 402, "lr": 9.002394533727382e-06, "val_loss": 0.019242998212575912} +{"train_loss": 0.013694529421627522, "grad_norm": 0.10611927509307861, "global_step": 42718, "epoch": 403, "lr": 9.000698050506557e-06} +{"train_loss": 0.016223769634962082, "grad_norm": 0.09364688396453857, "global_step": 42719, "epoch": 403, "lr": 8.999001711339434e-06} +{"train_loss": 0.012649784795939922, "grad_norm": 0.06569568812847137, "global_step": 42720, "epoch": 403, "lr": 8.997305516232002e-06} +{"train_loss": 0.016412319615483284, "grad_norm": 0.09643638879060745, "global_step": 42721, "epoch": 403, "lr": 8.995609465190197e-06} +{"train_loss": 0.01481197401881218, "grad_norm": 0.07151520997285843, "global_step": 42722, "epoch": 403, "lr": 8.993913558219985e-06} +{"train_loss": 0.015601607970893383, "grad_norm": 0.09054218977689743, "global_step": 42723, "epoch": 403, "lr": 8.99221779532734e-06} +{"train_loss": 0.018903836607933044, "grad_norm": 0.09554292261600494, "global_step": 42724, "epoch": 403, "lr": 8.990522176518207e-06} +{"train_loss": 0.01666889525949955, "grad_norm": 0.15066872537136078, "global_step": 42725, "epoch": 403, "lr": 8.98882670179853e-06} +{"train_loss": 0.01465996727347374, "grad_norm": 0.05605386570096016, "global_step": 42726, "epoch": 403, "lr": 8.9871313711743e-06} +{"train_loss": 0.014589119702577591, "grad_norm": 0.11144962906837463, "global_step": 42727, "epoch": 403, "lr": 8.985436184651436e-06} +{"train_loss": 0.015331913717091084, "grad_norm": 0.070128433406353, "global_step": 42728, "epoch": 403, "lr": 8.983741142235918e-06} +{"train_loss": 0.016589360311627388, "grad_norm": 0.08098036050796509, "global_step": 42729, "epoch": 403, "lr": 8.9820462439337e-06} +{"train_loss": 0.017098475247621536, "grad_norm": 0.10723123699426651, "global_step": 42730, "epoch": 403, "lr": 8.980351489750744e-06} +{"train_loss": 0.014733187854290009, "grad_norm": 0.09675341844558716, "global_step": 42731, "epoch": 403, "lr": 8.978656879692976e-06} +{"train_loss": 0.013817957602441311, "grad_norm": 0.05568945035338402, "global_step": 42732, "epoch": 403, "lr": 8.976962413766365e-06} +{"train_loss": 0.016582917422056198, "grad_norm": 0.08159777522087097, "global_step": 42733, "epoch": 403, "lr": 8.975268091976885e-06} +{"train_loss": 0.011793438345193863, "grad_norm": 0.06595338881015778, "global_step": 42734, "epoch": 403, "lr": 8.973573914330458e-06} +{"train_loss": 0.01533430814743042, "grad_norm": 0.054382551461458206, "global_step": 42735, "epoch": 403, "lr": 8.971879880833062e-06} +{"train_loss": 0.015749327838420868, "grad_norm": 0.13288934528827667, "global_step": 42736, "epoch": 403, "lr": 8.970185991490627e-06} +{"train_loss": 0.01304646022617817, "grad_norm": 0.054403968155384064, "global_step": 42737, "epoch": 403, "lr": 8.968492246309124e-06} +{"train_loss": 0.014021939598023891, "grad_norm": 0.05774450674653053, "global_step": 42738, "epoch": 403, "lr": 8.966798645294483e-06} +{"train_loss": 0.012512192130088806, "grad_norm": 0.05871725454926491, "global_step": 42739, "epoch": 403, "lr": 8.965105188452666e-06} +{"train_loss": 0.015890099108219147, "grad_norm": 0.08009129762649536, "global_step": 42740, "epoch": 403, "lr": 8.963411875789641e-06} +{"train_loss": 0.013591679744422436, "grad_norm": 0.0889623761177063, "global_step": 42741, "epoch": 403, "lr": 8.961718707311339e-06} +{"train_loss": 0.014180148020386696, "grad_norm": 0.05831611156463623, "global_step": 42742, "epoch": 403, "lr": 8.960025683023693e-06} +{"train_loss": 0.013803706504404545, "grad_norm": 0.0804378017783165, "global_step": 42743, "epoch": 403, "lr": 8.958332802932685e-06} +{"train_loss": 0.013526011258363724, "grad_norm": 0.048803213983774185, "global_step": 42744, "epoch": 403, "lr": 8.956640067044236e-06} +{"train_loss": 0.015098728239536285, "grad_norm": 0.07576461881399155, "global_step": 42745, "epoch": 403, "lr": 8.9549474753643e-06} +{"train_loss": 0.015132954344153404, "grad_norm": 0.09485320001840591, "global_step": 42746, "epoch": 403, "lr": 8.95325502789885e-06} +{"train_loss": 0.017104072496294975, "grad_norm": 0.08036606013774872, "global_step": 42747, "epoch": 403, "lr": 8.9515627246538e-06} +{"train_loss": 0.016161277890205383, "grad_norm": 0.09376392513513565, "global_step": 42748, "epoch": 403, "lr": 8.949870565635104e-06} +{"train_loss": 0.01686231791973114, "grad_norm": 0.06594917178153992, "global_step": 42749, "epoch": 403, "lr": 8.948178550848702e-06} +{"train_loss": 0.016396408900618553, "grad_norm": 0.09604686498641968, "global_step": 42750, "epoch": 403, "lr": 8.946486680300564e-06} +{"train_loss": 0.01409151591360569, "grad_norm": 0.07281848788261414, "global_step": 42751, "epoch": 403, "lr": 8.94479495399661e-06} +{"train_loss": 0.013118244707584381, "grad_norm": 0.0727548822760582, "global_step": 42752, "epoch": 403, "lr": 8.943103371942785e-06} +{"train_loss": 0.013224047608673573, "grad_norm": 0.09121979027986526, "global_step": 42753, "epoch": 403, "lr": 8.941411934145055e-06} +{"train_loss": 0.01590551994740963, "grad_norm": 0.08928555995225906, "global_step": 42754, "epoch": 403, "lr": 8.939720640609351e-06} +{"train_loss": 0.013250365853309631, "grad_norm": 0.059375207871198654, "global_step": 42755, "epoch": 403, "lr": 8.938029491341592e-06} +{"train_loss": 0.015341738238930702, "grad_norm": 0.10991431027650833, "global_step": 42756, "epoch": 403, "lr": 8.936338486347745e-06} +{"train_loss": 0.01304149255156517, "grad_norm": 0.11049320548772812, "global_step": 42757, "epoch": 403, "lr": 8.934647625633757e-06} +{"train_loss": 0.016265425831079483, "grad_norm": 0.08828627318143845, "global_step": 42758, "epoch": 403, "lr": 8.932956909205547e-06} +{"train_loss": 0.012929440476000309, "grad_norm": 0.09359991550445557, "global_step": 42759, "epoch": 403, "lr": 8.931266337069083e-06} +{"train_loss": 0.017498541623353958, "grad_norm": 0.11730673164129257, "global_step": 42760, "epoch": 403, "lr": 8.92957590923027e-06} +{"train_loss": 0.012622617185115814, "grad_norm": 0.07896698266267776, "global_step": 42761, "epoch": 403, "lr": 8.927885625695082e-06} +{"train_loss": 0.017059125006198883, "grad_norm": 0.08016616106033325, "global_step": 42762, "epoch": 403, "lr": 8.92619548646943e-06} +{"train_loss": 0.01667090132832527, "grad_norm": 0.09705793857574463, "global_step": 42763, "epoch": 403, "lr": 8.924505491559265e-06} +{"train_loss": 0.014926551841199398, "grad_norm": 0.08565320074558258, "global_step": 42764, "epoch": 403, "lr": 8.922815640970534e-06} +{"train_loss": 0.015167949721217155, "grad_norm": 0.07119519263505936, "global_step": 42765, "epoch": 403, "lr": 8.921125934709168e-06} +{"train_loss": 0.015047727152705193, "grad_norm": 0.05656180903315544, "global_step": 42766, "epoch": 403, "lr": 8.919436372781087e-06} +{"train_loss": 0.016889970749616623, "grad_norm": 0.051696859300136566, "global_step": 42767, "epoch": 403, "lr": 8.917746955192251e-06} +{"train_loss": 0.015970738604664803, "grad_norm": 0.08670610934495926, "global_step": 42768, "epoch": 403, "lr": 8.916057681948575e-06} +{"train_loss": 0.015856876969337463, "grad_norm": 0.0885864868760109, "global_step": 42769, "epoch": 403, "lr": 8.914368553056007e-06} +{"train_loss": 0.016190676018595695, "grad_norm": 0.07837186753749847, "global_step": 42770, "epoch": 403, "lr": 8.912679568520494e-06} +{"train_loss": 0.015730051323771477, "grad_norm": 0.10085013508796692, "global_step": 42771, "epoch": 403, "lr": 8.91099072834795e-06} +{"train_loss": 0.016406651586294174, "grad_norm": 0.08037111163139343, "global_step": 42772, "epoch": 403, "lr": 8.909302032544309e-06} +{"train_loss": 0.014284972101449966, "grad_norm": 0.05331725999712944, "global_step": 42773, "epoch": 403, "lr": 8.907613481115512e-06} +{"train_loss": 0.014102188870310783, "grad_norm": 0.06711139529943466, "global_step": 42774, "epoch": 403, "lr": 8.9059250740675e-06} +{"train_loss": 0.015044976957142353, "grad_norm": 0.07286038994789124, "global_step": 42775, "epoch": 403, "lr": 8.90423681140618e-06} +{"train_loss": 0.01664476841688156, "grad_norm": 0.08722339570522308, "global_step": 42776, "epoch": 403, "lr": 8.902548693137502e-06} +{"train_loss": 0.014503201469779015, "grad_norm": 0.05991879478096962, "global_step": 42777, "epoch": 403, "lr": 8.900860719267413e-06} +{"train_loss": 0.013673873618245125, "grad_norm": 0.11444370448589325, "global_step": 42778, "epoch": 403, "lr": 8.89917288980182e-06} +{"train_loss": 0.015029193833470345, "grad_norm": 0.0735757127404213, "global_step": 42779, "epoch": 403, "lr": 8.897485204746647e-06} +{"train_loss": 0.015140388160943985, "grad_norm": 0.09880685806274414, "global_step": 42780, "epoch": 403, "lr": 8.895797664107841e-06} +{"train_loss": 0.013586142100393772, "grad_norm": 0.07939432561397552, "global_step": 42781, "epoch": 403, "lr": 8.894110267891336e-06} +{"train_loss": 0.01375298760831356, "grad_norm": 0.11192396283149719, "global_step": 42782, "epoch": 403, "lr": 8.892423016103036e-06} +{"train_loss": 0.014402201399207115, "grad_norm": 0.07418099790811539, "global_step": 42783, "epoch": 403, "lr": 8.890735908748898e-06} +{"train_loss": 0.015800610184669495, "grad_norm": 0.067750483751297, "global_step": 42784, "epoch": 403, "lr": 8.889048945834838e-06} +{"train_loss": 0.016213061287999153, "grad_norm": 0.14616797864437103, "global_step": 42785, "epoch": 403, "lr": 8.887362127366765e-06} +{"train_loss": 0.016466861590743065, "grad_norm": 0.11738098412752151, "global_step": 42786, "epoch": 403, "lr": 8.885675453350628e-06} +{"train_loss": 0.01424398086965084, "grad_norm": 0.09852518886327744, "global_step": 42787, "epoch": 403, "lr": 8.883988923792358e-06} +{"train_loss": 0.016625292599201202, "grad_norm": 0.06685791164636612, "global_step": 42788, "epoch": 403, "lr": 8.882302538697867e-06} +{"train_loss": 0.013812167569994926, "grad_norm": 0.0663178488612175, "global_step": 42789, "epoch": 403, "lr": 8.880616298073069e-06} +{"train_loss": 0.014281246811151505, "grad_norm": 0.05835571140050888, "global_step": 42790, "epoch": 403, "lr": 8.878930201923908e-06} +{"train_loss": 0.014506638050079346, "grad_norm": 0.05327257513999939, "global_step": 42791, "epoch": 403, "lr": 8.877244250256317e-06} +{"train_loss": 0.013040533289313316, "grad_norm": 0.060281891375780106, "global_step": 42792, "epoch": 403, "lr": 8.875558443076188e-06} +{"train_loss": 0.013643607497215271, "grad_norm": 0.10219249874353409, "global_step": 42793, "epoch": 403, "lr": 8.873872780389464e-06} +{"train_loss": 0.013199102133512497, "grad_norm": 0.08369576930999756, "global_step": 42794, "epoch": 403, "lr": 8.87218726220208e-06} +{"train_loss": 0.015554622747004032, "grad_norm": 0.0632094219326973, "global_step": 42795, "epoch": 403, "lr": 8.870501888519944e-06} +{"train_loss": 0.015840604901313782, "grad_norm": 0.18295453488826752, "global_step": 42796, "epoch": 403, "lr": 8.868816659348962e-06} +{"train_loss": 0.013402607291936874, "grad_norm": 0.11060435324907303, "global_step": 42797, "epoch": 403, "lr": 8.86713157469507e-06} +{"train_loss": 0.013862550258636475, "grad_norm": 0.07412378489971161, "global_step": 42798, "epoch": 403, "lr": 8.865446634564206e-06} +{"train_loss": 0.015038751997053623, "grad_norm": 0.06016915664076805, "global_step": 42799, "epoch": 403, "lr": 8.86376183896226e-06} +{"train_loss": 0.016365109011530876, "grad_norm": 0.07495499402284622, "global_step": 42800, "epoch": 403, "lr": 8.862077187895179e-06} +{"train_loss": 0.01571911759674549, "grad_norm": 0.08434554934501648, "global_step": 42801, "epoch": 403, "lr": 8.860392681368868e-06} +{"train_loss": 0.015578210353851318, "grad_norm": 0.05250127241015434, "global_step": 42802, "epoch": 403, "lr": 8.85870831938923e-06} +{"train_loss": 0.014432663097977638, "grad_norm": 0.07237757742404938, "global_step": 42803, "epoch": 403, "lr": 8.857024101962202e-06} +{"train_loss": 0.015464755706489086, "grad_norm": 0.10621313005685806, "global_step": 42804, "epoch": 403, "lr": 8.855340029093695e-06} +{"train_loss": 0.013608468696475029, "grad_norm": 0.08384009450674057, "global_step": 42805, "epoch": 403, "lr": 8.853656100789647e-06} +{"train_loss": 0.014039138332009315, "grad_norm": 0.06261631101369858, "global_step": 42806, "epoch": 403, "lr": 8.851972317055956e-06} +{"train_loss": 0.013735208660364151, "grad_norm": 0.09918472915887833, "global_step": 42807, "epoch": 403, "lr": 8.850288677898528e-06} +{"train_loss": 0.015812022611498833, "grad_norm": 0.07629068940877914, "global_step": 42808, "epoch": 403, "lr": 8.848605183323299e-06} +{"train_loss": 0.016147278249263763, "grad_norm": 0.10482024401426315, "global_step": 42809, "epoch": 403, "lr": 8.846921833336164e-06} +{"train_loss": 0.016214855015277863, "grad_norm": 0.06431809067726135, "global_step": 42810, "epoch": 403, "lr": 8.845238627943048e-06} +{"train_loss": 0.014984071254730225, "grad_norm": 0.10136998444795609, "global_step": 42811, "epoch": 403, "lr": 8.84355556714988e-06} +{"train_loss": 0.018532561138272285, "grad_norm": 0.06909853219985962, "global_step": 42812, "epoch": 403, "lr": 8.841872650962563e-06} +{"train_loss": 0.016946304589509964, "grad_norm": 0.09835478663444519, "global_step": 42813, "epoch": 403, "lr": 8.840189879386989e-06} +{"train_loss": 0.014629507437348366, "grad_norm": 0.06033261492848396, "global_step": 42814, "epoch": 403, "lr": 8.838507252429084e-06} +{"train_loss": 0.015352901071310043, "grad_norm": 0.10835717618465424, "global_step": 42815, "epoch": 403, "lr": 8.836824770094782e-06} +{"train_loss": 0.014419608749449253, "grad_norm": 0.05462545529007912, "global_step": 42816, "epoch": 403, "lr": 8.835142432389965e-06} +{"train_loss": 0.014258932322263718, "grad_norm": 0.07407371699810028, "global_step": 42817, "epoch": 403, "lr": 8.833460239320552e-06} +{"train_loss": 0.013948367908596992, "grad_norm": 0.06422651559114456, "global_step": 42818, "epoch": 403, "lr": 8.83177819089247e-06} +{"train_loss": 0.014451928436756134, "grad_norm": 0.08280085772275925, "global_step": 42819, "epoch": 403, "lr": 8.830096287111617e-06} +{"train_loss": 0.014991693198680878, "grad_norm": 0.10775132477283478, "global_step": 42820, "epoch": 403, "lr": 8.828414527983886e-06} +{"train_loss": 0.013379060663282871, "grad_norm": 0.0937773734331131, "global_step": 42821, "epoch": 403, "lr": 8.826732913515207e-06} +{"train_loss": 0.01445765234529972, "grad_norm": 0.06351467221975327, "global_step": 42822, "epoch": 403, "lr": 8.825051443711491e-06} +{"train_loss": 0.014990665958667139, "grad_norm": 0.13484810292720795, "global_step": 42823, "epoch": 403, "lr": 8.823370118578628e-06, "val_loss": 0.019505921751260757} +{"train_loss": 0.014281217008829117, "grad_norm": 0.06645900756120682, "global_step": 42824, "epoch": 404, "lr": 8.821688938122541e-06} +{"train_loss": 0.014889225363731384, "grad_norm": 0.0951094701886177, "global_step": 42825, "epoch": 404, "lr": 8.820007902349136e-06} +{"train_loss": 0.015867041423916817, "grad_norm": 0.06996724754571915, "global_step": 42826, "epoch": 404, "lr": 8.818327011264305e-06} +{"train_loss": 0.0165665652602911, "grad_norm": 0.10972902178764343, "global_step": 42827, "epoch": 404, "lr": 8.816646264873957e-06} +{"train_loss": 0.01541338674724102, "grad_norm": 0.08976669609546661, "global_step": 42828, "epoch": 404, "lr": 8.814965663184021e-06} +{"train_loss": 0.015040623024106026, "grad_norm": 0.06742649525403976, "global_step": 42829, "epoch": 404, "lr": 8.813285206200372e-06} +{"train_loss": 0.01574891060590744, "grad_norm": 0.1280917078256607, "global_step": 42830, "epoch": 404, "lr": 8.811604893928938e-06} +{"train_loss": 0.015027912333607674, "grad_norm": 0.0891878679394722, "global_step": 42831, "epoch": 404, "lr": 8.809924726375602e-06} +{"train_loss": 0.014063969254493713, "grad_norm": 0.08166143298149109, "global_step": 42832, "epoch": 404, "lr": 8.808244703546286e-06} +{"train_loss": 0.01593264937400818, "grad_norm": 0.11501439660787582, "global_step": 42833, "epoch": 404, "lr": 8.806564825446873e-06} +{"train_loss": 0.0166008360683918, "grad_norm": 0.07817284762859344, "global_step": 42834, "epoch": 404, "lr": 8.804885092083276e-06} +{"train_loss": 0.014832290820777416, "grad_norm": 0.11310756951570511, "global_step": 42835, "epoch": 404, "lr": 8.803205503461414e-06} +{"train_loss": 0.014794062823057175, "grad_norm": 0.12896597385406494, "global_step": 42836, "epoch": 404, "lr": 8.801526059587167e-06} +{"train_loss": 0.015451915562152863, "grad_norm": 0.09694507718086243, "global_step": 42837, "epoch": 404, "lr": 8.799846760466435e-06} +{"train_loss": 0.01271731499582529, "grad_norm": 0.09378746896982193, "global_step": 42838, "epoch": 404, "lr": 8.798167606105118e-06} +{"train_loss": 0.01490064337849617, "grad_norm": 0.1101427972316742, "global_step": 42839, "epoch": 404, "lr": 8.796488596509133e-06} +{"train_loss": 0.01547127403318882, "grad_norm": 0.09312701970338821, "global_step": 42840, "epoch": 404, "lr": 8.794809731684362e-06} +{"train_loss": 0.01220976747572422, "grad_norm": 0.07791676372289658, "global_step": 42841, "epoch": 404, "lr": 8.793131011636719e-06} +{"train_loss": 0.014068035408854485, "grad_norm": 0.06920593976974487, "global_step": 42842, "epoch": 404, "lr": 8.791452436372088e-06} +{"train_loss": 0.016541926190257072, "grad_norm": 0.09888073801994324, "global_step": 42843, "epoch": 404, "lr": 8.789774005896368e-06} +{"train_loss": 0.01551793608814478, "grad_norm": 0.09346609562635422, "global_step": 42844, "epoch": 404, "lr": 8.788095720215455e-06} +{"train_loss": 0.01527198776602745, "grad_norm": 0.12073231488466263, "global_step": 42845, "epoch": 404, "lr": 8.786417579335265e-06} +{"train_loss": 0.013950161635875702, "grad_norm": 0.0819043219089508, "global_step": 42846, "epoch": 404, "lr": 8.784739583261669e-06} +{"train_loss": 0.017323464155197144, "grad_norm": 0.06610960513353348, "global_step": 42847, "epoch": 404, "lr": 8.783061732000586e-06} +{"train_loss": 0.014792874455451965, "grad_norm": 0.0723743885755539, "global_step": 42848, "epoch": 404, "lr": 8.781384025557888e-06} +{"train_loss": 0.015736861154437065, "grad_norm": 0.07359359413385391, "global_step": 42849, "epoch": 404, "lr": 8.779706463939491e-06} +{"train_loss": 0.013201335445046425, "grad_norm": 0.05447326600551605, "global_step": 42850, "epoch": 404, "lr": 8.77802904715127e-06} +{"train_loss": 0.015085470862686634, "grad_norm": 0.07725948095321655, "global_step": 42851, "epoch": 404, "lr": 8.776351775199126e-06} +{"train_loss": 0.016960419714450836, "grad_norm": 0.08886580169200897, "global_step": 42852, "epoch": 404, "lr": 8.774674648088965e-06} +{"train_loss": 0.016026534140110016, "grad_norm": 0.0809582769870758, "global_step": 42853, "epoch": 404, "lr": 8.772997665826671e-06} +{"train_loss": 0.01443653367459774, "grad_norm": 0.06335917860269547, "global_step": 42854, "epoch": 404, "lr": 8.771320828418117e-06} +{"train_loss": 0.01710580475628376, "grad_norm": 0.06178181618452072, "global_step": 42855, "epoch": 404, "lr": 8.76964413586922e-06} +{"train_loss": 0.015200046822428703, "grad_norm": 0.0845642015337944, "global_step": 42856, "epoch": 404, "lr": 8.767967588185872e-06} +{"train_loss": 0.014250729233026505, "grad_norm": 0.1009361669421196, "global_step": 42857, "epoch": 404, "lr": 8.766291185373938e-06} +{"train_loss": 0.014983322471380234, "grad_norm": 0.08699575066566467, "global_step": 42858, "epoch": 404, "lr": 8.76461492743933e-06} +{"train_loss": 0.014344091527163982, "grad_norm": 0.0928550735116005, "global_step": 42859, "epoch": 404, "lr": 8.76293881438794e-06} +{"train_loss": 0.01608613133430481, "grad_norm": 0.11876227706670761, "global_step": 42860, "epoch": 404, "lr": 8.761262846225653e-06} +{"train_loss": 0.0144168920814991, "grad_norm": 0.052574027329683304, "global_step": 42861, "epoch": 404, "lr": 8.759587022958343e-06} +{"train_loss": 0.015143468976020813, "grad_norm": 0.06396868079900742, "global_step": 42862, "epoch": 404, "lr": 8.757911344591907e-06} +{"train_loss": 0.015406090766191483, "grad_norm": 0.08668576180934906, "global_step": 42863, "epoch": 404, "lr": 8.756235811132247e-06} +{"train_loss": 0.015730075538158417, "grad_norm": 0.08459614962339401, "global_step": 42864, "epoch": 404, "lr": 8.754560422585223e-06} +{"train_loss": 0.015942588448524475, "grad_norm": 0.0822049081325531, "global_step": 42865, "epoch": 404, "lr": 8.752885178956755e-06} +{"train_loss": 0.019203361123800278, "grad_norm": 0.07756848633289337, "global_step": 42866, "epoch": 404, "lr": 8.751210080252703e-06} +{"train_loss": 0.015537173487246037, "grad_norm": 0.0691426545381546, "global_step": 42867, "epoch": 404, "lr": 8.749535126478953e-06} +{"train_loss": 0.015560644678771496, "grad_norm": 0.09782803803682327, "global_step": 42868, "epoch": 404, "lr": 8.7478603176414e-06} +{"train_loss": 0.01781662367284298, "grad_norm": 0.07798115164041519, "global_step": 42869, "epoch": 404, "lr": 8.74618565374593e-06} +{"train_loss": 0.018089178949594498, "grad_norm": 0.08810969442129135, "global_step": 42870, "epoch": 404, "lr": 8.744511134798416e-06} +{"train_loss": 0.012875032611191273, "grad_norm": 0.05864638835191727, "global_step": 42871, "epoch": 404, "lr": 8.742836760804762e-06} +{"train_loss": 0.016140246763825417, "grad_norm": 0.09111382067203522, "global_step": 42872, "epoch": 404, "lr": 8.74116253177082e-06} +{"train_loss": 0.017186351120471954, "grad_norm": 0.0526273213326931, "global_step": 42873, "epoch": 404, "lr": 8.739488447702504e-06} +{"train_loss": 0.015591636300086975, "grad_norm": 0.07999718934297562, "global_step": 42874, "epoch": 404, "lr": 8.737814508605674e-06} +{"train_loss": 0.014871158637106419, "grad_norm": 0.07358718663454056, "global_step": 42875, "epoch": 404, "lr": 8.736140714486213e-06} +{"train_loss": 0.016782570630311966, "grad_norm": 0.07951435446739197, "global_step": 42876, "epoch": 404, "lr": 8.734467065350022e-06} +{"train_loss": 0.012059366330504417, "grad_norm": 0.05548742413520813, "global_step": 42877, "epoch": 404, "lr": 8.732793561202973e-06} +{"train_loss": 0.014187254011631012, "grad_norm": 0.1236044391989708, "global_step": 42878, "epoch": 404, "lr": 8.731120202050924e-06} +{"train_loss": 0.01382935419678688, "grad_norm": 0.06318242847919464, "global_step": 42879, "epoch": 404, "lr": 8.72944698789977e-06} +{"train_loss": 0.016267254948616028, "grad_norm": 0.05437125638127327, "global_step": 42880, "epoch": 404, "lr": 8.727773918755405e-06} +{"train_loss": 0.013820316642522812, "grad_norm": 0.07542460411787033, "global_step": 42881, "epoch": 404, "lr": 8.726100994623682e-06} +{"train_loss": 0.014381477609276772, "grad_norm": 0.06962494552135468, "global_step": 42882, "epoch": 404, "lr": 8.724428215510488e-06} +{"train_loss": 0.015115849673748016, "grad_norm": 0.059278447180986404, "global_step": 42883, "epoch": 404, "lr": 8.722755581421732e-06} +{"train_loss": 0.01350812241435051, "grad_norm": 0.06705980002880096, "global_step": 42884, "epoch": 404, "lr": 8.721083092363236e-06} +{"train_loss": 0.012348391115665436, "grad_norm": 0.07844056189060211, "global_step": 42885, "epoch": 404, "lr": 8.7194107483409e-06} +{"train_loss": 0.01742197386920452, "grad_norm": 0.10695305466651917, "global_step": 42886, "epoch": 404, "lr": 8.717738549360616e-06} +{"train_loss": 0.013098325580358505, "grad_norm": 0.11124516278505325, "global_step": 42887, "epoch": 404, "lr": 8.716066495428232e-06} +{"train_loss": 0.014522373676300049, "grad_norm": 0.10864683985710144, "global_step": 42888, "epoch": 404, "lr": 8.714394586549651e-06} +{"train_loss": 0.01676986925303936, "grad_norm": 0.13798411190509796, "global_step": 42889, "epoch": 404, "lr": 8.712722822730717e-06} +{"train_loss": 0.014513779431581497, "grad_norm": 0.11462181061506271, "global_step": 42890, "epoch": 404, "lr": 8.71105120397734e-06} +{"train_loss": 0.0139896459877491, "grad_norm": 0.07812736183404922, "global_step": 42891, "epoch": 404, "lr": 8.709379730295354e-06} +{"train_loss": 0.0143730528652668, "grad_norm": 0.07001016288995743, "global_step": 42892, "epoch": 404, "lr": 8.70770840169065e-06} +{"train_loss": 0.011840395629405975, "grad_norm": 0.0595594197511673, "global_step": 42893, "epoch": 404, "lr": 8.706037218169116e-06} +{"train_loss": 0.013869274407625198, "grad_norm": 0.08188758045434952, "global_step": 42894, "epoch": 404, "lr": 8.704366179736606e-06} +{"train_loss": 0.01581544242799282, "grad_norm": 0.0735773965716362, "global_step": 42895, "epoch": 404, "lr": 8.702695286398987e-06} +{"train_loss": 0.014740167185664177, "grad_norm": 0.07624704390764236, "global_step": 42896, "epoch": 404, "lr": 8.701024538162133e-06} +{"train_loss": 0.01362510584294796, "grad_norm": 0.04791427403688431, "global_step": 42897, "epoch": 404, "lr": 8.699353935031934e-06} +{"train_loss": 0.013692975975573063, "grad_norm": 0.08675599098205566, "global_step": 42898, "epoch": 404, "lr": 8.69768347701423e-06} +{"train_loss": 0.013994799926877022, "grad_norm": 0.0498933345079422, "global_step": 42899, "epoch": 404, "lr": 8.696013164114902e-06} +{"train_loss": 0.013126742094755173, "grad_norm": 0.12372633814811707, "global_step": 42900, "epoch": 404, "lr": 8.694342996339849e-06} +{"train_loss": 0.01686924695968628, "grad_norm": 0.07629837840795517, "global_step": 42901, "epoch": 404, "lr": 8.692672973694882e-06} +{"train_loss": 0.015018641948699951, "grad_norm": 0.0999138355255127, "global_step": 42902, "epoch": 404, "lr": 8.6910030961859e-06} +{"train_loss": 0.015910588204860687, "grad_norm": 0.088003970682621, "global_step": 42903, "epoch": 404, "lr": 8.689333363818786e-06} +{"train_loss": 0.016865679994225502, "grad_norm": 0.0837865099310875, "global_step": 42904, "epoch": 404, "lr": 8.68766377659937e-06} +{"train_loss": 0.014558176510035992, "grad_norm": 0.1206086128950119, "global_step": 42905, "epoch": 404, "lr": 8.685994334533543e-06} +{"train_loss": 0.013279950246214867, "grad_norm": 0.054426684975624084, "global_step": 42906, "epoch": 404, "lr": 8.684325037627173e-06} +{"train_loss": 0.015559541992843151, "grad_norm": 0.06359611451625824, "global_step": 42907, "epoch": 404, "lr": 8.68265588588612e-06} +{"train_loss": 0.013107482343912125, "grad_norm": 0.08766482770442963, "global_step": 42908, "epoch": 404, "lr": 8.680986879316233e-06} +{"train_loss": 0.015390228480100632, "grad_norm": 0.08526677638292313, "global_step": 42909, "epoch": 404, "lr": 8.679318017923388e-06} +{"train_loss": 0.013779008761048317, "grad_norm": 0.08733391016721725, "global_step": 42910, "epoch": 404, "lr": 8.677649301713458e-06} +{"train_loss": 0.015290290117263794, "grad_norm": 0.05500783026218414, "global_step": 42911, "epoch": 404, "lr": 8.67598073069229e-06} +{"train_loss": 0.01631815917789936, "grad_norm": 0.09188991039991379, "global_step": 42912, "epoch": 404, "lr": 8.674312304865767e-06} +{"train_loss": 0.016619708389043808, "grad_norm": 0.12089244276285172, "global_step": 42913, "epoch": 404, "lr": 8.672644024239724e-06} +{"train_loss": 0.013517412357032299, "grad_norm": 0.061778612434864044, "global_step": 42914, "epoch": 404, "lr": 8.67097588882005e-06} +{"train_loss": 0.010904552415013313, "grad_norm": 0.08180932700634003, "global_step": 42915, "epoch": 404, "lr": 8.669307898612582e-06} +{"train_loss": 0.016661910340189934, "grad_norm": 0.08798666298389435, "global_step": 42916, "epoch": 404, "lr": 8.667640053623194e-06} +{"train_loss": 0.013805367052555084, "grad_norm": 0.059793174266815186, "global_step": 42917, "epoch": 404, "lr": 8.66597235385775e-06} +{"train_loss": 0.014328640885651112, "grad_norm": 0.06558898836374283, "global_step": 42918, "epoch": 404, "lr": 8.66430479932211e-06} +{"train_loss": 0.014634869992733002, "grad_norm": 0.11806057393550873, "global_step": 42919, "epoch": 404, "lr": 8.66263739002211e-06} +{"train_loss": 0.015290005132555962, "grad_norm": 0.08112955838441849, "global_step": 42920, "epoch": 404, "lr": 8.660970125963625e-06} +{"train_loss": 0.013176944106817245, "grad_norm": 0.05229966714978218, "global_step": 42921, "epoch": 404, "lr": 8.659303007152524e-06} +{"train_loss": 0.014401329681277275, "grad_norm": 0.09516685456037521, "global_step": 42922, "epoch": 404, "lr": 8.657636033594646e-06} +{"train_loss": 0.013991360552608967, "grad_norm": 0.06637740135192871, "global_step": 42923, "epoch": 404, "lr": 8.655969205295855e-06} +{"train_loss": 0.013082343153655529, "grad_norm": 0.06344777345657349, "global_step": 42924, "epoch": 404, "lr": 8.654302522262031e-06} +{"train_loss": 0.016144126653671265, "grad_norm": 0.1029694527387619, "global_step": 42925, "epoch": 404, "lr": 8.65263598449898e-06} +{"train_loss": 0.017323818057775497, "grad_norm": 0.0616525262594223, "global_step": 42926, "epoch": 404, "lr": 8.650969592012587e-06} +{"train_loss": 0.01593679189682007, "grad_norm": 0.07828060537576675, "global_step": 42927, "epoch": 404, "lr": 8.649303344808717e-06} +{"train_loss": 0.015836864709854126, "grad_norm": 0.08563327044248581, "global_step": 42928, "epoch": 404, "lr": 8.647637242893202e-06} +{"train_loss": 0.014986282346313292, "grad_norm": 0.0950096920132637, "global_step": 42929, "epoch": 404, "lr": 8.645971286271904e-06, "val_loss": 0.019253334030508995} +{"train_loss": 0.015703421086072922, "grad_norm": 0.06560596078634262, "global_step": 42930, "epoch": 405, "lr": 8.644305474950693e-06} +{"train_loss": 0.015959318727254868, "grad_norm": 0.10339680314064026, "global_step": 42931, "epoch": 405, "lr": 8.642639808935405e-06} +{"train_loss": 0.01396113820374012, "grad_norm": 0.05756969749927521, "global_step": 42932, "epoch": 405, "lr": 8.640974288231884e-06} +{"train_loss": 0.013644907623529434, "grad_norm": 0.06194506213068962, "global_step": 42933, "epoch": 405, "lr": 8.63930891284599e-06} +{"train_loss": 0.016375551000237465, "grad_norm": 0.11245850473642349, "global_step": 42934, "epoch": 405, "lr": 8.637643682783592e-06} +{"train_loss": 0.01624520868062973, "grad_norm": 0.07741830497980118, "global_step": 42935, "epoch": 405, "lr": 8.635978598050514e-06} +{"train_loss": 0.014014236629009247, "grad_norm": 0.05301695689558983, "global_step": 42936, "epoch": 405, "lr": 8.634313658652627e-06} +{"train_loss": 0.015510368160903454, "grad_norm": 0.07626605778932571, "global_step": 42937, "epoch": 405, "lr": 8.632648864595766e-06} +{"train_loss": 0.014558611437678337, "grad_norm": 0.06527430564165115, "global_step": 42938, "epoch": 405, "lr": 8.630984215885795e-06} +{"train_loss": 0.015988029539585114, "grad_norm": 0.07428308576345444, "global_step": 42939, "epoch": 405, "lr": 8.629319712528544e-06} +{"train_loss": 0.014266856014728546, "grad_norm": 0.08880605548620224, "global_step": 42940, "epoch": 405, "lr": 8.627655354529867e-06} +{"train_loss": 0.013135984539985657, "grad_norm": 0.1268363893032074, "global_step": 42941, "epoch": 405, "lr": 8.625991141895651e-06} +{"train_loss": 0.014766181819140911, "grad_norm": 0.17119111120700836, "global_step": 42942, "epoch": 405, "lr": 8.624327074631677e-06} +{"train_loss": 0.014101449400186539, "grad_norm": 0.0741288810968399, "global_step": 42943, "epoch": 405, "lr": 8.622663152743827e-06} +{"train_loss": 0.014338813722133636, "grad_norm": 0.07239198684692383, "global_step": 42944, "epoch": 405, "lr": 8.620999376237954e-06} +{"train_loss": 0.015163691714406013, "grad_norm": 0.090915746986866, "global_step": 42945, "epoch": 405, "lr": 8.619335745119882e-06} +{"train_loss": 0.012172142043709755, "grad_norm": 0.10380498319864273, "global_step": 42946, "epoch": 405, "lr": 8.61767225939547e-06} +{"train_loss": 0.014239574782550335, "grad_norm": 0.11627528816461563, "global_step": 42947, "epoch": 405, "lr": 8.616008919070572e-06} +{"train_loss": 0.01743398979306221, "grad_norm": 0.10292467474937439, "global_step": 42948, "epoch": 405, "lr": 8.614345724151018e-06} +{"train_loss": 0.013598164543509483, "grad_norm": 0.0656188353896141, "global_step": 42949, "epoch": 405, "lr": 8.612682674642647e-06} +{"train_loss": 0.015498154796659946, "grad_norm": 0.07133181393146515, "global_step": 42950, "epoch": 405, "lr": 8.61101977055131e-06} +{"train_loss": 0.015094469301402569, "grad_norm": 0.059402383863925934, "global_step": 42951, "epoch": 405, "lr": 8.609357011882863e-06} +{"train_loss": 0.017733626067638397, "grad_norm": 0.07254515588283539, "global_step": 42952, "epoch": 405, "lr": 8.607694398643123e-06} +{"train_loss": 0.014382396824657917, "grad_norm": 0.05738050863146782, "global_step": 42953, "epoch": 405, "lr": 8.606031930837955e-06} +{"train_loss": 0.01728903129696846, "grad_norm": 0.07813308387994766, "global_step": 42954, "epoch": 405, "lr": 8.604369608473172e-06} +{"train_loss": 0.013412237167358398, "grad_norm": 0.10783478617668152, "global_step": 42955, "epoch": 405, "lr": 8.602707431554653e-06} +{"train_loss": 0.01592995785176754, "grad_norm": 0.06703118234872818, "global_step": 42956, "epoch": 405, "lr": 8.601045400088197e-06} +{"train_loss": 0.013414817862212658, "grad_norm": 0.06317086517810822, "global_step": 42957, "epoch": 405, "lr": 8.599383514079667e-06} +{"train_loss": 0.012094343081116676, "grad_norm": 0.08370472490787506, "global_step": 42958, "epoch": 405, "lr": 8.59772177353491e-06} +{"train_loss": 0.01520621683448553, "grad_norm": 0.0669412612915039, "global_step": 42959, "epoch": 405, "lr": 8.596060178459758e-06} +{"train_loss": 0.012956171296536922, "grad_norm": 0.11268387734889984, "global_step": 42960, "epoch": 405, "lr": 8.594398728860026e-06} +{"train_loss": 0.01430493500083685, "grad_norm": 0.09342695027589798, "global_step": 42961, "epoch": 405, "lr": 8.59273742474157e-06} +{"train_loss": 0.014118464663624763, "grad_norm": 0.07695543020963669, "global_step": 42962, "epoch": 405, "lr": 8.591076266110243e-06} +{"train_loss": 0.014212370850145817, "grad_norm": 0.06560590863227844, "global_step": 42963, "epoch": 405, "lr": 8.58941525297185e-06} +{"train_loss": 0.015056213364005089, "grad_norm": 0.09867726266384125, "global_step": 42964, "epoch": 405, "lr": 8.587754385332247e-06} +{"train_loss": 0.01659226417541504, "grad_norm": 0.10108910501003265, "global_step": 42965, "epoch": 405, "lr": 8.586093663197287e-06} +{"train_loss": 0.014649668708443642, "grad_norm": 0.05429665371775627, "global_step": 42966, "epoch": 405, "lr": 8.584433086572758e-06} +{"train_loss": 0.012640321627259254, "grad_norm": 0.09173557907342911, "global_step": 42967, "epoch": 405, "lr": 8.582772655464522e-06} +{"train_loss": 0.014811739325523376, "grad_norm": 0.06788307428359985, "global_step": 42968, "epoch": 405, "lr": 8.581112369878425e-06} +{"train_loss": 0.014389153569936752, "grad_norm": 0.06168973818421364, "global_step": 42969, "epoch": 405, "lr": 8.579452229820273e-06} +{"train_loss": 0.01266708318144083, "grad_norm": 0.06101314350962639, "global_step": 42970, "epoch": 405, "lr": 8.577792235295911e-06} +{"train_loss": 0.012968866154551506, "grad_norm": 0.07522962987422943, "global_step": 42971, "epoch": 405, "lr": 8.576132386311187e-06} +{"train_loss": 0.014119750820100307, "grad_norm": 0.06157851591706276, "global_step": 42972, "epoch": 405, "lr": 8.574472682871915e-06} +{"train_loss": 0.01457703486084938, "grad_norm": 0.06696715205907822, "global_step": 42973, "epoch": 405, "lr": 8.572813124983925e-06} +{"train_loss": 0.016364509239792824, "grad_norm": 0.05475049093365669, "global_step": 42974, "epoch": 405, "lr": 8.571153712653045e-06} +{"train_loss": 0.012655469588935375, "grad_norm": 0.06798522174358368, "global_step": 42975, "epoch": 405, "lr": 8.569494445885135e-06} +{"train_loss": 0.017846230417490005, "grad_norm": 0.08141028881072998, "global_step": 42976, "epoch": 405, "lr": 8.567835324685986e-06} +{"train_loss": 0.015261434018611908, "grad_norm": 0.06897632032632828, "global_step": 42977, "epoch": 405, "lr": 8.566176349061461e-06} +{"train_loss": 0.015228796750307083, "grad_norm": 0.088286854326725, "global_step": 42978, "epoch": 405, "lr": 8.564517519017357e-06} +{"train_loss": 0.012596912682056427, "grad_norm": 0.06749343872070312, "global_step": 42979, "epoch": 405, "lr": 8.562858834559534e-06} +{"train_loss": 0.014963964000344276, "grad_norm": 0.12518465518951416, "global_step": 42980, "epoch": 405, "lr": 8.561200295693788e-06} +{"train_loss": 0.013433351181447506, "grad_norm": 0.06566327065229416, "global_step": 42981, "epoch": 405, "lr": 8.559541902425967e-06} +{"train_loss": 0.015337866730988026, "grad_norm": 0.11005118489265442, "global_step": 42982, "epoch": 405, "lr": 8.557883654761906e-06} +{"train_loss": 0.018069829791784286, "grad_norm": 0.08705092966556549, "global_step": 42983, "epoch": 405, "lr": 8.55622555270742e-06} +{"train_loss": 0.015175776556134224, "grad_norm": 0.10452117025852203, "global_step": 42984, "epoch": 405, "lr": 8.55456759626832e-06} +{"train_loss": 0.016344793140888214, "grad_norm": 0.08460575342178345, "global_step": 42985, "epoch": 405, "lr": 8.552909785450452e-06} +{"train_loss": 0.015310336835682392, "grad_norm": 0.09329868853092194, "global_step": 42986, "epoch": 405, "lr": 8.551252120259628e-06} +{"train_loss": 0.015893222764134407, "grad_norm": 0.08334905654191971, "global_step": 42987, "epoch": 405, "lr": 8.549594600701682e-06} +{"train_loss": 0.013047993183135986, "grad_norm": 0.09425097703933716, "global_step": 42988, "epoch": 405, "lr": 8.547937226782443e-06} +{"train_loss": 0.014448476023972034, "grad_norm": 0.10590433329343796, "global_step": 42989, "epoch": 405, "lr": 8.546279998507723e-06} +{"train_loss": 0.016406290233135223, "grad_norm": 0.062215328216552734, "global_step": 42990, "epoch": 405, "lr": 8.544622915883337e-06} +{"train_loss": 0.015786385163664818, "grad_norm": 0.05339983478188515, "global_step": 42991, "epoch": 405, "lr": 8.542965978915119e-06} +{"train_loss": 0.015116792172193527, "grad_norm": 0.11575130373239517, "global_step": 42992, "epoch": 405, "lr": 8.541309187608898e-06} +{"train_loss": 0.01451418362557888, "grad_norm": 0.12977811694145203, "global_step": 42993, "epoch": 405, "lr": 8.539652541970478e-06} +{"train_loss": 0.012985732406377792, "grad_norm": 0.08235899358987808, "global_step": 42994, "epoch": 405, "lr": 8.537996042005696e-06} +{"train_loss": 0.014914518222212791, "grad_norm": 0.07471581548452377, "global_step": 42995, "epoch": 405, "lr": 8.536339687720358e-06} +{"train_loss": 0.01608879491686821, "grad_norm": 0.10227568447589874, "global_step": 42996, "epoch": 405, "lr": 8.534683479120299e-06} +{"train_loss": 0.017492331564426422, "grad_norm": 0.1404121220111847, "global_step": 42997, "epoch": 405, "lr": 8.53302741621132e-06} +{"train_loss": 0.015232335776090622, "grad_norm": 0.106593556702137, "global_step": 42998, "epoch": 405, "lr": 8.531371498999247e-06} +{"train_loss": 0.012980005703866482, "grad_norm": 0.07762447744607925, "global_step": 42999, "epoch": 405, "lr": 8.529715727489912e-06} +{"train_loss": 0.014152337796986103, "grad_norm": 0.057791534811258316, "global_step": 43000, "epoch": 405, "lr": 8.528060101689122e-06} +{"train_loss": 0.017008429393172264, "grad_norm": 0.10500706732273102, "global_step": 43001, "epoch": 405, "lr": 8.52640462160268e-06} +{"train_loss": 0.015851853415369987, "grad_norm": 0.10030076652765274, "global_step": 43002, "epoch": 405, "lr": 8.524749287236434e-06} +{"train_loss": 0.014709670096635818, "grad_norm": 0.0691237598657608, "global_step": 43003, "epoch": 405, "lr": 8.523094098596163e-06} +{"train_loss": 0.014445699751377106, "grad_norm": 0.1087748184800148, "global_step": 43004, "epoch": 405, "lr": 8.521439055687703e-06} +{"train_loss": 0.013338621705770493, "grad_norm": 0.06833779066801071, "global_step": 43005, "epoch": 405, "lr": 8.51978415851688e-06} +{"train_loss": 0.014308009296655655, "grad_norm": 0.05813935399055481, "global_step": 43006, "epoch": 405, "lr": 8.518129407089493e-06} +{"train_loss": 0.014922220259904861, "grad_norm": 0.07675933837890625, "global_step": 43007, "epoch": 405, "lr": 8.51647480141135e-06} +{"train_loss": 0.015516461804509163, "grad_norm": 0.1058616042137146, "global_step": 43008, "epoch": 405, "lr": 8.514820341488278e-06} +{"train_loss": 0.012686539441347122, "grad_norm": 0.06848392635583878, "global_step": 43009, "epoch": 405, "lr": 8.513166027326091e-06} +{"train_loss": 0.013766275718808174, "grad_norm": 0.11301568895578384, "global_step": 43010, "epoch": 405, "lr": 8.511511858930587e-06} +{"train_loss": 0.017877932637929916, "grad_norm": 0.15923340618610382, "global_step": 43011, "epoch": 405, "lr": 8.50985783630759e-06} +{"train_loss": 0.01794256456196308, "grad_norm": 0.08303379267454147, "global_step": 43012, "epoch": 405, "lr": 8.508203959462918e-06} +{"train_loss": 0.017680414021015167, "grad_norm": 0.0809604823589325, "global_step": 43013, "epoch": 405, "lr": 8.50655022840237e-06} +{"train_loss": 0.014987405389547348, "grad_norm": 0.08434553444385529, "global_step": 43014, "epoch": 405, "lr": 8.504896643131754e-06} +{"train_loss": 0.014619238674640656, "grad_norm": 0.07252325862646103, "global_step": 43015, "epoch": 405, "lr": 8.503243203656886e-06} +{"train_loss": 0.01677425019443035, "grad_norm": 0.09180320799350739, "global_step": 43016, "epoch": 405, "lr": 8.501589909983582e-06} +{"train_loss": 0.017487913370132446, "grad_norm": 0.08483501523733139, "global_step": 43017, "epoch": 405, "lr": 8.499936762117633e-06} +{"train_loss": 0.014801114797592163, "grad_norm": 0.10165958851575851, "global_step": 43018, "epoch": 405, "lr": 8.498283760064867e-06} +{"train_loss": 0.01803845912218094, "grad_norm": 0.09749060869216919, "global_step": 43019, "epoch": 405, "lr": 8.496630903831077e-06} +{"train_loss": 0.017789259552955627, "grad_norm": 0.0850650817155838, "global_step": 43020, "epoch": 405, "lr": 8.494978193422087e-06} +{"train_loss": 0.014706790447235107, "grad_norm": 0.08972427248954773, "global_step": 43021, "epoch": 405, "lr": 8.493325628843684e-06} +{"train_loss": 0.014260638505220413, "grad_norm": 0.07186370342969894, "global_step": 43022, "epoch": 405, "lr": 8.49167321010168e-06} +{"train_loss": 0.013188756071031094, "grad_norm": 0.055674146860837936, "global_step": 43023, "epoch": 405, "lr": 8.490020937201898e-06} +{"train_loss": 0.014875774271786213, "grad_norm": 0.09094832092523575, "global_step": 43024, "epoch": 405, "lr": 8.488368810150132e-06} +{"train_loss": 0.015441438183188438, "grad_norm": 0.08609575033187866, "global_step": 43025, "epoch": 405, "lr": 8.486716828952168e-06} +{"train_loss": 0.015258449129760265, "grad_norm": 0.07244479656219482, "global_step": 43026, "epoch": 405, "lr": 8.485064993613845e-06} +{"train_loss": 0.012730799615383148, "grad_norm": 0.11887476593255997, "global_step": 43027, "epoch": 405, "lr": 8.483413304140935e-06} +{"train_loss": 0.015362255275249481, "grad_norm": 0.07242152839899063, "global_step": 43028, "epoch": 405, "lr": 8.481761760539253e-06} +{"train_loss": 0.014245199039578438, "grad_norm": 0.200005441904068, "global_step": 43029, "epoch": 405, "lr": 8.480110362814619e-06} +{"train_loss": 0.013919140212237835, "grad_norm": 0.1404440999031067, "global_step": 43030, "epoch": 405, "lr": 8.478459110972825e-06} +{"train_loss": 0.01822488382458687, "grad_norm": 0.08444702625274658, "global_step": 43031, "epoch": 405, "lr": 8.476808005019648e-06} +{"train_loss": 0.015194721519947052, "grad_norm": 0.08413342386484146, "global_step": 43032, "epoch": 405, "lr": 8.475157044960913e-06} +{"train_loss": 0.014148823916912079, "grad_norm": 0.09044014662504196, "global_step": 43033, "epoch": 405, "lr": 8.47350623080243e-06} +{"train_loss": 0.01718050241470337, "grad_norm": 0.10007734596729279, "global_step": 43034, "epoch": 405, "lr": 8.471855562549979e-06} +{"train_loss": 0.014994445204172493, "grad_norm": 0.10886053740978241, "global_step": 43035, "epoch": 405, "lr": 8.470205040209361e-06, "val_loss": 0.019369401037693024} +{"train_loss": 0.014616472646594048, "grad_norm": 0.08071482181549072, "global_step": 43036, "epoch": 406, "lr": 8.468554663786398e-06} +{"train_loss": 0.012633672915399075, "grad_norm": 0.0697719007730484, "global_step": 43037, "epoch": 406, "lr": 8.466904433286871e-06} +{"train_loss": 0.015938570722937584, "grad_norm": 0.09162242710590363, "global_step": 43038, "epoch": 406, "lr": 8.465254348716567e-06} +{"train_loss": 0.014645801857113838, "grad_norm": 0.1218666359782219, "global_step": 43039, "epoch": 406, "lr": 8.463604410081293e-06} +{"train_loss": 0.018126560375094414, "grad_norm": 0.06456899642944336, "global_step": 43040, "epoch": 406, "lr": 8.46195461738687e-06} +{"train_loss": 0.014862625859677792, "grad_norm": 0.05826462432742119, "global_step": 43041, "epoch": 406, "lr": 8.460304970639066e-06} +{"train_loss": 0.013970272615551949, "grad_norm": 0.05938590690493584, "global_step": 43042, "epoch": 406, "lr": 8.458655469843673e-06} +{"train_loss": 0.013881877064704895, "grad_norm": 0.07438068091869354, "global_step": 43043, "epoch": 406, "lr": 8.457006115006515e-06} +{"train_loss": 0.014688306488096714, "grad_norm": 0.06859790533781052, "global_step": 43044, "epoch": 406, "lr": 8.45535690613336e-06} +{"train_loss": 0.015542995184659958, "grad_norm": 0.05999782681465149, "global_step": 43045, "epoch": 406, "lr": 8.453707843230007e-06} +{"train_loss": 0.0157078318297863, "grad_norm": 0.07344647496938705, "global_step": 43046, "epoch": 406, "lr": 8.452058926302275e-06} +{"train_loss": 0.01415518019348383, "grad_norm": 0.08253366500139236, "global_step": 43047, "epoch": 406, "lr": 8.450410155355936e-06} +{"train_loss": 0.015706773847341537, "grad_norm": 0.09077291935682297, "global_step": 43048, "epoch": 406, "lr": 8.448761530396776e-06} +{"train_loss": 0.01825120486319065, "grad_norm": 0.07134491950273514, "global_step": 43049, "epoch": 406, "lr": 8.4471130514306e-06} +{"train_loss": 0.014429135248064995, "grad_norm": 0.09795460850000381, "global_step": 43050, "epoch": 406, "lr": 8.44546471846321e-06} +{"train_loss": 0.01294827088713646, "grad_norm": 0.05724477767944336, "global_step": 43051, "epoch": 406, "lr": 8.443816531500365e-06} +{"train_loss": 0.01515901554375887, "grad_norm": 0.06885722279548645, "global_step": 43052, "epoch": 406, "lr": 8.44216849054788e-06} +{"train_loss": 0.017093811184167862, "grad_norm": 0.06975038349628448, "global_step": 43053, "epoch": 406, "lr": 8.440520595611557e-06} +{"train_loss": 0.016276728361845016, "grad_norm": 0.14049117267131805, "global_step": 43054, "epoch": 406, "lr": 8.438872846697171e-06} +{"train_loss": 0.013304224237799644, "grad_norm": 0.08886437863111496, "global_step": 43055, "epoch": 406, "lr": 8.437225243810498e-06} +{"train_loss": 0.014312835410237312, "grad_norm": 0.1420065313577652, "global_step": 43056, "epoch": 406, "lr": 8.435577786957339e-06} +{"train_loss": 0.015176055021584034, "grad_norm": 0.09731334447860718, "global_step": 43057, "epoch": 406, "lr": 8.433930476143497e-06} +{"train_loss": 0.015945639461278915, "grad_norm": 0.09078551083803177, "global_step": 43058, "epoch": 406, "lr": 8.432283311374738e-06} +{"train_loss": 0.01656937599182129, "grad_norm": 0.06544925272464752, "global_step": 43059, "epoch": 406, "lr": 8.430636292656868e-06} +{"train_loss": 0.016306370496749878, "grad_norm": 0.08895155787467957, "global_step": 43060, "epoch": 406, "lr": 8.42898941999566e-06} +{"train_loss": 0.013929873704910278, "grad_norm": 0.07969633489847183, "global_step": 43061, "epoch": 406, "lr": 8.427342693396895e-06} +{"train_loss": 0.013499820604920387, "grad_norm": 0.0759236067533493, "global_step": 43062, "epoch": 406, "lr": 8.425696112866371e-06} +{"train_loss": 0.011942287907004356, "grad_norm": 0.09184935688972473, "global_step": 43063, "epoch": 406, "lr": 8.424049678409868e-06} +{"train_loss": 0.01652807928621769, "grad_norm": 0.07338064163923264, "global_step": 43064, "epoch": 406, "lr": 8.422403390033184e-06} +{"train_loss": 0.011951923370361328, "grad_norm": 0.10144016891717911, "global_step": 43065, "epoch": 406, "lr": 8.420757247742095e-06} +{"train_loss": 0.01584434136748314, "grad_norm": 0.15421754121780396, "global_step": 43066, "epoch": 406, "lr": 8.419111251542372e-06} +{"train_loss": 0.012713760137557983, "grad_norm": 0.06992781907320023, "global_step": 43067, "epoch": 406, "lr": 8.417465401439816e-06} +{"train_loss": 0.014619359746575356, "grad_norm": 0.05742891877889633, "global_step": 43068, "epoch": 406, "lr": 8.415819697440197e-06} +{"train_loss": 0.016459623351693153, "grad_norm": 0.11154720187187195, "global_step": 43069, "epoch": 406, "lr": 8.414174139549302e-06} +{"train_loss": 0.015886448323726654, "grad_norm": 0.06578473746776581, "global_step": 43070, "epoch": 406, "lr": 8.412528727772922e-06} +{"train_loss": 0.013668465428054333, "grad_norm": 0.09511919319629669, "global_step": 43071, "epoch": 406, "lr": 8.41088346211683e-06} +{"train_loss": 0.01426917128264904, "grad_norm": 0.16079290211200714, "global_step": 43072, "epoch": 406, "lr": 8.4092383425868e-06} +{"train_loss": 0.013748183846473694, "grad_norm": 0.11411920189857483, "global_step": 43073, "epoch": 406, "lr": 8.407593369188615e-06} +{"train_loss": 0.014210676774382591, "grad_norm": 0.052921708673238754, "global_step": 43074, "epoch": 406, "lr": 8.405948541928071e-06} +{"train_loss": 0.015271982178092003, "grad_norm": 0.13649533689022064, "global_step": 43075, "epoch": 406, "lr": 8.404303860810924e-06} +{"train_loss": 0.014367252588272095, "grad_norm": 0.09988274425268173, "global_step": 43076, "epoch": 406, "lr": 8.402659325842966e-06} +{"train_loss": 0.013555888086557388, "grad_norm": 0.0803104043006897, "global_step": 43077, "epoch": 406, "lr": 8.401014937029982e-06} +{"train_loss": 0.013726308010518551, "grad_norm": 0.09377047419548035, "global_step": 43078, "epoch": 406, "lr": 8.399370694377745e-06} +{"train_loss": 0.01306100469082594, "grad_norm": 0.11659165471792221, "global_step": 43079, "epoch": 406, "lr": 8.397726597892008e-06} +{"train_loss": 0.013151646591722965, "grad_norm": 0.08480362594127655, "global_step": 43080, "epoch": 406, "lr": 8.396082647578574e-06} +{"train_loss": 0.017611496150493622, "grad_norm": 0.13051845133304596, "global_step": 43081, "epoch": 406, "lr": 8.39443884344322e-06} +{"train_loss": 0.012592468410730362, "grad_norm": 0.05662590265274048, "global_step": 43082, "epoch": 406, "lr": 8.392795185491708e-06} +{"train_loss": 0.015361310914158821, "grad_norm": 0.09274212270975113, "global_step": 43083, "epoch": 406, "lr": 8.391151673729824e-06} +{"train_loss": 0.015385174192488194, "grad_norm": 0.12046027183532715, "global_step": 43084, "epoch": 406, "lr": 8.389508308163341e-06} +{"train_loss": 0.016820736229419708, "grad_norm": 0.12320791184902191, "global_step": 43085, "epoch": 406, "lr": 8.387865088798019e-06} +{"train_loss": 0.015916690230369568, "grad_norm": 0.06820296496152878, "global_step": 43086, "epoch": 406, "lr": 8.386222015639645e-06} +{"train_loss": 0.013944912701845169, "grad_norm": 0.08911650627851486, "global_step": 43087, "epoch": 406, "lr": 8.384579088693994e-06} +{"train_loss": 0.013371238484978676, "grad_norm": 0.1298208385705948, "global_step": 43088, "epoch": 406, "lr": 8.382936307966838e-06} +{"train_loss": 0.015070813708007336, "grad_norm": 0.07712776958942413, "global_step": 43089, "epoch": 406, "lr": 8.381293673463936e-06} +{"train_loss": 0.016909128054976463, "grad_norm": 0.09754446893930435, "global_step": 43090, "epoch": 406, "lr": 8.379651185191067e-06} +{"train_loss": 0.013585802167654037, "grad_norm": 0.08594310283660889, "global_step": 43091, "epoch": 406, "lr": 8.378008843154011e-06} +{"train_loss": 0.016857460141181946, "grad_norm": 0.16212856769561768, "global_step": 43092, "epoch": 406, "lr": 8.376366647358525e-06} +{"train_loss": 0.015343833714723587, "grad_norm": 0.07430242002010345, "global_step": 43093, "epoch": 406, "lr": 8.374724597810379e-06} +{"train_loss": 0.015197651460766792, "grad_norm": 0.06723059713840485, "global_step": 43094, "epoch": 406, "lr": 8.373082694515367e-06} +{"train_loss": 0.014522392302751541, "grad_norm": 0.05343824252486229, "global_step": 43095, "epoch": 406, "lr": 8.371440937479236e-06} +{"train_loss": 0.013888410292565823, "grad_norm": 0.10140804201364517, "global_step": 43096, "epoch": 406, "lr": 8.369799326707745e-06} +{"train_loss": 0.01692119799554348, "grad_norm": 0.15517278015613556, "global_step": 43097, "epoch": 406, "lr": 8.368157862206671e-06} +{"train_loss": 0.016062069684267044, "grad_norm": 0.0897974744439125, "global_step": 43098, "epoch": 406, "lr": 8.366516543981807e-06} +{"train_loss": 0.015796147286891937, "grad_norm": 0.08356401324272156, "global_step": 43099, "epoch": 406, "lr": 8.364875372038876e-06} +{"train_loss": 0.015591111034154892, "grad_norm": 0.0585654117166996, "global_step": 43100, "epoch": 406, "lr": 8.363234346383686e-06} +{"train_loss": 0.012433122843503952, "grad_norm": 0.09076395630836487, "global_step": 43101, "epoch": 406, "lr": 8.361593467021977e-06} +{"train_loss": 0.015166773460805416, "grad_norm": 0.11544334888458252, "global_step": 43102, "epoch": 406, "lr": 8.359952733959514e-06} +{"train_loss": 0.015417481772601604, "grad_norm": 0.10613427311182022, "global_step": 43103, "epoch": 406, "lr": 8.358312147202063e-06} +{"train_loss": 0.016066856682300568, "grad_norm": 0.15430159866809845, "global_step": 43104, "epoch": 406, "lr": 8.356671706755409e-06} +{"train_loss": 0.016972798854112625, "grad_norm": 0.06516430526971817, "global_step": 43105, "epoch": 406, "lr": 8.355031412625285e-06} +{"train_loss": 0.015216922387480736, "grad_norm": 0.09297282993793488, "global_step": 43106, "epoch": 406, "lr": 8.35339126481749e-06} +{"train_loss": 0.015843570232391357, "grad_norm": 0.08143341541290283, "global_step": 43107, "epoch": 406, "lr": 8.351751263337749e-06} +{"train_loss": 0.01428224891424179, "grad_norm": 0.0754297599196434, "global_step": 43108, "epoch": 406, "lr": 8.350111408191852e-06} +{"train_loss": 0.01357890572398901, "grad_norm": 0.06352085620164871, "global_step": 43109, "epoch": 406, "lr": 8.34847169938554e-06} +{"train_loss": 0.015555476769804955, "grad_norm": 0.10772670805454254, "global_step": 43110, "epoch": 406, "lr": 8.346832136924582e-06} +{"train_loss": 0.014760235324501991, "grad_norm": 0.11737612634897232, "global_step": 43111, "epoch": 406, "lr": 8.34519272081476e-06} +{"train_loss": 0.011378670111298561, "grad_norm": 0.07402821630239487, "global_step": 43112, "epoch": 406, "lr": 8.343553451061808e-06} +{"train_loss": 0.014939382672309875, "grad_norm": 0.0894094929099083, "global_step": 43113, "epoch": 406, "lr": 8.341914327671486e-06} +{"train_loss": 0.012856513261795044, "grad_norm": 0.08299338817596436, "global_step": 43114, "epoch": 406, "lr": 8.340275350649562e-06} +{"train_loss": 0.015052171424031258, "grad_norm": 0.11521082371473312, "global_step": 43115, "epoch": 406, "lr": 8.338636520001802e-06} +{"train_loss": 0.013469692319631577, "grad_norm": 0.07635144889354706, "global_step": 43116, "epoch": 406, "lr": 8.33699783573394e-06} +{"train_loss": 0.020243674516677856, "grad_norm": 0.09168020635843277, "global_step": 43117, "epoch": 406, "lr": 8.335359297851753e-06} +{"train_loss": 0.015198642387986183, "grad_norm": 0.07113867253065109, "global_step": 43118, "epoch": 406, "lr": 8.33372090636101e-06} +{"train_loss": 0.01699400693178177, "grad_norm": 0.07247035205364227, "global_step": 43119, "epoch": 406, "lr": 8.332082661267443e-06} +{"train_loss": 0.014976443722844124, "grad_norm": 0.06410925835371017, "global_step": 43120, "epoch": 406, "lr": 8.330444562576811e-06} +{"train_loss": 0.015027564018964767, "grad_norm": 0.1312185376882553, "global_step": 43121, "epoch": 406, "lr": 8.32880661029487e-06} +{"train_loss": 0.015498028136789799, "grad_norm": 0.14338789880275726, "global_step": 43122, "epoch": 406, "lr": 8.327168804427398e-06} +{"train_loss": 0.016257263720035553, "grad_norm": 0.11343434453010559, "global_step": 43123, "epoch": 406, "lr": 8.325531144980114e-06} +{"train_loss": 0.016324952244758606, "grad_norm": 0.14541411399841309, "global_step": 43124, "epoch": 406, "lr": 8.323893631958806e-06} +{"train_loss": 0.016998272389173508, "grad_norm": 0.08529580384492874, "global_step": 43125, "epoch": 406, "lr": 8.32225626536921e-06} +{"train_loss": 0.015687594190239906, "grad_norm": 0.07270403951406479, "global_step": 43126, "epoch": 406, "lr": 8.32061904521707e-06} +{"train_loss": 0.016042638570070267, "grad_norm": 0.08405592292547226, "global_step": 43127, "epoch": 406, "lr": 8.318981971508145e-06} +{"train_loss": 0.015869230031967163, "grad_norm": 0.1767989844083786, "global_step": 43128, "epoch": 406, "lr": 8.317345044248204e-06} +{"train_loss": 0.015357148833572865, "grad_norm": 0.0911237820982933, "global_step": 43129, "epoch": 406, "lr": 8.315708263442974e-06} +{"train_loss": 0.013956321403384209, "grad_norm": 0.06186172366142273, "global_step": 43130, "epoch": 406, "lr": 8.31407162909823e-06} +{"train_loss": 0.014618647284805775, "grad_norm": 0.09648021310567856, "global_step": 43131, "epoch": 406, "lr": 8.312435141219699e-06} +{"train_loss": 0.016031749546527863, "grad_norm": 0.08737839013338089, "global_step": 43132, "epoch": 406, "lr": 8.310798799813152e-06} +{"train_loss": 0.013507746160030365, "grad_norm": 0.05722208693623543, "global_step": 43133, "epoch": 406, "lr": 8.309162604884313e-06} +{"train_loss": 0.013941168785095215, "grad_norm": 0.08220457285642624, "global_step": 43134, "epoch": 406, "lr": 8.30752655643895e-06} +{"train_loss": 0.013418816961348057, "grad_norm": 0.07580389827489853, "global_step": 43135, "epoch": 406, "lr": 8.305890654482811e-06} +{"train_loss": 0.01717311143875122, "grad_norm": 0.09245257079601288, "global_step": 43136, "epoch": 406, "lr": 8.304254899021651e-06} +{"train_loss": 0.015981337055563927, "grad_norm": 0.08499378710985184, "global_step": 43137, "epoch": 406, "lr": 8.302619290061187e-06} +{"train_loss": 0.013401944190263748, "grad_norm": 0.08547849208116531, "global_step": 43138, "epoch": 406, "lr": 8.30098382760719e-06} +{"train_loss": 0.013863570988178253, "grad_norm": 0.08113213628530502, "global_step": 43139, "epoch": 406, "lr": 8.299348511665412e-06} +{"train_loss": 0.016045495867729187, "grad_norm": 0.07102923095226288, "global_step": 43140, "epoch": 406, "lr": 8.29771334224157e-06} +{"train_loss": 0.01499595442119072, "grad_norm": 0.13684214651584625, "global_step": 43141, "epoch": 406, "lr": 8.296078319341443e-06, "val_loss": 0.019500209018588066} +{"train_loss": 0.013978367671370506, "grad_norm": 0.0743328258395195, "global_step": 43142, "epoch": 407, "lr": 8.294443442970768e-06} +{"train_loss": 0.01607002690434456, "grad_norm": 0.10728499293327332, "global_step": 43143, "epoch": 407, "lr": 8.292808713135258e-06} +{"train_loss": 0.015160767361521721, "grad_norm": 0.1339089423418045, "global_step": 43144, "epoch": 407, "lr": 8.291174129840684e-06} +{"train_loss": 0.014085222035646439, "grad_norm": 0.06385401636362076, "global_step": 43145, "epoch": 407, "lr": 8.2895396930928e-06} +{"train_loss": 0.015733223408460617, "grad_norm": 0.11571619659662247, "global_step": 43146, "epoch": 407, "lr": 8.287905402897323e-06} +{"train_loss": 0.012822091579437256, "grad_norm": 0.12328947335481644, "global_step": 43147, "epoch": 407, "lr": 8.286271259260015e-06} +{"train_loss": 0.015791263431310654, "grad_norm": 0.0675051361322403, "global_step": 43148, "epoch": 407, "lr": 8.2846372621866e-06} +{"train_loss": 0.017179053276777267, "grad_norm": 0.09063856303691864, "global_step": 43149, "epoch": 407, "lr": 8.283003411682839e-06} +{"train_loss": 0.014207536354660988, "grad_norm": 0.05984728783369064, "global_step": 43150, "epoch": 407, "lr": 8.28136970775445e-06} +{"train_loss": 0.01428409107029438, "grad_norm": 0.06719564646482468, "global_step": 43151, "epoch": 407, "lr": 8.27973615040718e-06} +{"train_loss": 0.014887341298162937, "grad_norm": 0.06559380143880844, "global_step": 43152, "epoch": 407, "lr": 8.278102739646793e-06} +{"train_loss": 0.013443481177091599, "grad_norm": 0.1634664237499237, "global_step": 43153, "epoch": 407, "lr": 8.276469475479004e-06} +{"train_loss": 0.014263206161558628, "grad_norm": 0.06061511114239693, "global_step": 43154, "epoch": 407, "lr": 8.274836357909543e-06} +{"train_loss": 0.014545351266860962, "grad_norm": 0.08322066813707352, "global_step": 43155, "epoch": 407, "lr": 8.273203386944161e-06} +{"train_loss": 0.014985667541623116, "grad_norm": 0.08712638914585114, "global_step": 43156, "epoch": 407, "lr": 8.271570562588614e-06} +{"train_loss": 0.01801304519176483, "grad_norm": 0.08978506177663803, "global_step": 43157, "epoch": 407, "lr": 8.269937884848606e-06} +{"train_loss": 0.016415739431977272, "grad_norm": 0.06468293070793152, "global_step": 43158, "epoch": 407, "lr": 8.268305353729882e-06} +{"train_loss": 0.011863138526678085, "grad_norm": 0.06024878844618797, "global_step": 43159, "epoch": 407, "lr": 8.266672969238216e-06} +{"train_loss": 0.012583687901496887, "grad_norm": 0.06819919496774673, "global_step": 43160, "epoch": 407, "lr": 8.265040731379286e-06} +{"train_loss": 0.013364596292376518, "grad_norm": 0.06974547356367111, "global_step": 43161, "epoch": 407, "lr": 8.263408640158855e-06} +{"train_loss": 0.01463987585157156, "grad_norm": 0.05884983763098717, "global_step": 43162, "epoch": 407, "lr": 8.261776695582662e-06} +{"train_loss": 0.012981022708117962, "grad_norm": 0.06419263780117035, "global_step": 43163, "epoch": 407, "lr": 8.260144897656425e-06} +{"train_loss": 0.017568880692124367, "grad_norm": 0.1306534856557846, "global_step": 43164, "epoch": 407, "lr": 8.258513246385885e-06} +{"train_loss": 0.015014808624982834, "grad_norm": 0.11378195881843567, "global_step": 43165, "epoch": 407, "lr": 8.25688174177679e-06} +{"train_loss": 0.014641305431723595, "grad_norm": 0.09132514894008636, "global_step": 43166, "epoch": 407, "lr": 8.255250383834862e-06} +{"train_loss": 0.014196409843862057, "grad_norm": 0.10214156657457352, "global_step": 43167, "epoch": 407, "lr": 8.253619172565813e-06} +{"train_loss": 0.014323337934911251, "grad_norm": 0.06655226647853851, "global_step": 43168, "epoch": 407, "lr": 8.251988107975395e-06} +{"train_loss": 0.014746779575943947, "grad_norm": 0.07098276913166046, "global_step": 43169, "epoch": 407, "lr": 8.250357190069347e-06} +{"train_loss": 0.01573076844215393, "grad_norm": 0.10488856583833694, "global_step": 43170, "epoch": 407, "lr": 8.24872641885337e-06} +{"train_loss": 0.013280822895467281, "grad_norm": 0.08429263532161713, "global_step": 43171, "epoch": 407, "lr": 8.247095794333226e-06} +{"train_loss": 0.016164284199476242, "grad_norm": 0.06945433467626572, "global_step": 43172, "epoch": 407, "lr": 8.245465316514622e-06} +{"train_loss": 0.016812022775411606, "grad_norm": 0.05802328139543533, "global_step": 43173, "epoch": 407, "lr": 8.243834985403298e-06} +{"train_loss": 0.01423543319106102, "grad_norm": 0.08182525634765625, "global_step": 43174, "epoch": 407, "lr": 8.242204801004972e-06} +{"train_loss": 0.014523166231811047, "grad_norm": 0.1252780705690384, "global_step": 43175, "epoch": 407, "lr": 8.240574763325381e-06} +{"train_loss": 0.01754995435476303, "grad_norm": 0.12798698246479034, "global_step": 43176, "epoch": 407, "lr": 8.238944872370258e-06} +{"train_loss": 0.015531210228800774, "grad_norm": 0.06975675374269485, "global_step": 43177, "epoch": 407, "lr": 8.237315128145318e-06} +{"train_loss": 0.011977728456258774, "grad_norm": 0.13031868636608124, "global_step": 43178, "epoch": 407, "lr": 8.235685530656284e-06} +{"train_loss": 0.01858200505375862, "grad_norm": 0.06052779406309128, "global_step": 43179, "epoch": 407, "lr": 8.234056079908886e-06} +{"train_loss": 0.013760443776845932, "grad_norm": 0.06476274877786636, "global_step": 43180, "epoch": 407, "lr": 8.232426775908863e-06} +{"train_loss": 0.015862513333559036, "grad_norm": 0.09868797659873962, "global_step": 43181, "epoch": 407, "lr": 8.230797618661917e-06} +{"train_loss": 0.014248359948396683, "grad_norm": 0.10389160364866257, "global_step": 43182, "epoch": 407, "lr": 8.229168608173781e-06} +{"train_loss": 0.012588843703269958, "grad_norm": 0.052799470722675323, "global_step": 43183, "epoch": 407, "lr": 8.227539744450207e-06} +{"train_loss": 0.015803251415491104, "grad_norm": 0.1787015199661255, "global_step": 43184, "epoch": 407, "lr": 8.225911027496868e-06} +{"train_loss": 0.014086230657994747, "grad_norm": 0.12400570511817932, "global_step": 43185, "epoch": 407, "lr": 8.22428245731951e-06} +{"train_loss": 0.015423818491399288, "grad_norm": 0.10450439900159836, "global_step": 43186, "epoch": 407, "lr": 8.22265403392387e-06} +{"train_loss": 0.01629096083343029, "grad_norm": 0.06513073295354843, "global_step": 43187, "epoch": 407, "lr": 8.221025757315642e-06} +{"train_loss": 0.017930353060364723, "grad_norm": 0.06275783479213715, "global_step": 43188, "epoch": 407, "lr": 8.219397627500569e-06} +{"train_loss": 0.013920661993324757, "grad_norm": 0.0709969699382782, "global_step": 43189, "epoch": 407, "lr": 8.217769644484352e-06} +{"train_loss": 0.014795975759625435, "grad_norm": 0.06777399778366089, "global_step": 43190, "epoch": 407, "lr": 8.216141808272738e-06} +{"train_loss": 0.015341139398515224, "grad_norm": 0.08345742523670197, "global_step": 43191, "epoch": 407, "lr": 8.214514118871414e-06} +{"train_loss": 0.017527110874652863, "grad_norm": 0.0883949026465416, "global_step": 43192, "epoch": 407, "lr": 8.212886576286117e-06} +{"train_loss": 0.015131562948226929, "grad_norm": 0.12663449347019196, "global_step": 43193, "epoch": 407, "lr": 8.211259180522579e-06} +{"train_loss": 0.015867913141846657, "grad_norm": 0.0714949294924736, "global_step": 43194, "epoch": 407, "lr": 8.209631931586498e-06} +{"train_loss": 0.015161113813519478, "grad_norm": 0.06859283894300461, "global_step": 43195, "epoch": 407, "lr": 8.208004829483584e-06} +{"train_loss": 0.01578698679804802, "grad_norm": 0.10344232618808746, "global_step": 43196, "epoch": 407, "lr": 8.206377874219568e-06} +{"train_loss": 0.01496730837970972, "grad_norm": 0.07029436528682709, "global_step": 43197, "epoch": 407, "lr": 8.204751065800176e-06} +{"train_loss": 0.011557121761143208, "grad_norm": 0.07267626374959946, "global_step": 43198, "epoch": 407, "lr": 8.203124404231106e-06} +{"train_loss": 0.01696188747882843, "grad_norm": 0.06621333211660385, "global_step": 43199, "epoch": 407, "lr": 8.201497889518073e-06} +{"train_loss": 0.016264259815216064, "grad_norm": 0.06438510864973068, "global_step": 43200, "epoch": 407, "lr": 8.199871521666825e-06} +{"train_loss": 0.017975904047489166, "grad_norm": 0.1305704265832901, "global_step": 43201, "epoch": 407, "lr": 8.198245300683022e-06} +{"train_loss": 0.016010602936148643, "grad_norm": 0.10246512293815613, "global_step": 43202, "epoch": 407, "lr": 8.19661922657241e-06} +{"train_loss": 0.014866015873849392, "grad_norm": 0.10924433171749115, "global_step": 43203, "epoch": 407, "lr": 8.194993299340708e-06} +{"train_loss": 0.015269462019205093, "grad_norm": 0.07270436733961105, "global_step": 43204, "epoch": 407, "lr": 8.19336751899361e-06} +{"train_loss": 0.016033343970775604, "grad_norm": 0.12315338104963303, "global_step": 43205, "epoch": 407, "lr": 8.191741885536836e-06} +{"train_loss": 0.014155545271933079, "grad_norm": 0.08813319355249405, "global_step": 43206, "epoch": 407, "lr": 8.190116398976112e-06} +{"train_loss": 0.014872638508677483, "grad_norm": 0.06783298403024673, "global_step": 43207, "epoch": 407, "lr": 8.188491059317138e-06} +{"train_loss": 0.013027162291109562, "grad_norm": 0.09051205217838287, "global_step": 43208, "epoch": 407, "lr": 8.186865866565607e-06} +{"train_loss": 0.016392765566706657, "grad_norm": 0.094429150223732, "global_step": 43209, "epoch": 407, "lr": 8.185240820727247e-06} +{"train_loss": 0.013109451159834862, "grad_norm": 0.07238324731588364, "global_step": 43210, "epoch": 407, "lr": 8.183615921807781e-06} +{"train_loss": 0.01275664009153843, "grad_norm": 0.05095658451318741, "global_step": 43211, "epoch": 407, "lr": 8.181991169812887e-06} +{"train_loss": 0.015056896023452282, "grad_norm": 0.0766364187002182, "global_step": 43212, "epoch": 407, "lr": 8.180366564748304e-06} +{"train_loss": 0.015399330295622349, "grad_norm": 0.11150536686182022, "global_step": 43213, "epoch": 407, "lr": 8.178742106619713e-06} +{"train_loss": 0.01656264439225197, "grad_norm": 0.0786222591996193, "global_step": 43214, "epoch": 407, "lr": 8.177117795432843e-06} +{"train_loss": 0.014307903125882149, "grad_norm": 0.05120459198951721, "global_step": 43215, "epoch": 407, "lr": 8.175493631193388e-06} +{"train_loss": 0.014021620154380798, "grad_norm": 0.06228305399417877, "global_step": 43216, "epoch": 407, "lr": 8.173869613907054e-06} +{"train_loss": 0.015444912947714329, "grad_norm": 0.0684003010392189, "global_step": 43217, "epoch": 407, "lr": 8.172245743579564e-06} +{"train_loss": 0.014440554194152355, "grad_norm": 0.10284345597028732, "global_step": 43218, "epoch": 407, "lr": 8.170622020216618e-06} +{"train_loss": 0.015194285660982132, "grad_norm": 0.11570318788290024, "global_step": 43219, "epoch": 407, "lr": 8.1689984438239e-06} +{"train_loss": 0.015298400074243546, "grad_norm": 0.06361322849988937, "global_step": 43220, "epoch": 407, "lr": 8.167375014407125e-06} +{"train_loss": 0.015034018084406853, "grad_norm": 0.12510965764522552, "global_step": 43221, "epoch": 407, "lr": 8.16575173197202e-06} +{"train_loss": 0.013432879000902176, "grad_norm": 0.06423044204711914, "global_step": 43222, "epoch": 407, "lr": 8.164128596524256e-06} +{"train_loss": 0.014640383422374725, "grad_norm": 0.11791805177927017, "global_step": 43223, "epoch": 407, "lr": 8.162505608069548e-06} +{"train_loss": 0.014880155213177204, "grad_norm": 0.06884235143661499, "global_step": 43224, "epoch": 407, "lr": 8.160882766613631e-06} +{"train_loss": 0.012840259820222855, "grad_norm": 0.056035660207271576, "global_step": 43225, "epoch": 407, "lr": 8.159260072162144e-06} +{"train_loss": 0.01622384786605835, "grad_norm": 0.07492444664239883, "global_step": 43226, "epoch": 407, "lr": 8.15763752472083e-06} +{"train_loss": 0.016331756487488747, "grad_norm": 0.10373810678720474, "global_step": 43227, "epoch": 407, "lr": 8.156015124295386e-06} +{"train_loss": 0.015861637890338898, "grad_norm": 0.07771681249141693, "global_step": 43228, "epoch": 407, "lr": 8.154392870891498e-06} +{"train_loss": 0.015349783934652805, "grad_norm": 0.08797377347946167, "global_step": 43229, "epoch": 407, "lr": 8.152770764514873e-06} +{"train_loss": 0.015303023159503937, "grad_norm": 0.10126833617687225, "global_step": 43230, "epoch": 407, "lr": 8.151148805171227e-06} +{"train_loss": 0.015203993767499924, "grad_norm": 0.07806544750928879, "global_step": 43231, "epoch": 407, "lr": 8.149526992866247e-06} +{"train_loss": 0.013589761219918728, "grad_norm": 0.11089601367712021, "global_step": 43232, "epoch": 407, "lr": 8.147905327605616e-06} +{"train_loss": 0.01646934263408184, "grad_norm": 0.09991437941789627, "global_step": 43233, "epoch": 407, "lr": 8.146283809395039e-06} +{"train_loss": 0.015519492328166962, "grad_norm": 0.11955033987760544, "global_step": 43234, "epoch": 407, "lr": 8.144662438240235e-06} +{"train_loss": 0.015534251928329468, "grad_norm": 0.12643255293369293, "global_step": 43235, "epoch": 407, "lr": 8.143041214146873e-06} +{"train_loss": 0.01375944260507822, "grad_norm": 0.08316175639629364, "global_step": 43236, "epoch": 407, "lr": 8.14142013712067e-06} +{"train_loss": 0.01403820887207985, "grad_norm": 0.0760219395160675, "global_step": 43237, "epoch": 407, "lr": 8.139799207167303e-06} +{"train_loss": 0.01551513560116291, "grad_norm": 0.06353908777236938, "global_step": 43238, "epoch": 407, "lr": 8.138178424292492e-06} +{"train_loss": 0.011728053912520409, "grad_norm": 0.053002893924713135, "global_step": 43239, "epoch": 407, "lr": 8.136557788501903e-06} +{"train_loss": 0.017749207094311714, "grad_norm": 0.08825601637363434, "global_step": 43240, "epoch": 407, "lr": 8.134937299801244e-06} +{"train_loss": 0.014204785227775574, "grad_norm": 0.06304112076759338, "global_step": 43241, "epoch": 407, "lr": 8.13331695819623e-06} +{"train_loss": 0.01399246696382761, "grad_norm": 0.08645402640104294, "global_step": 43242, "epoch": 407, "lr": 8.13169676369251e-06} +{"train_loss": 0.016749422997236252, "grad_norm": 0.0789778083562851, "global_step": 43243, "epoch": 407, "lr": 8.130076716295804e-06} +{"train_loss": 0.01575373113155365, "grad_norm": 0.16731923818588257, "global_step": 43244, "epoch": 407, "lr": 8.128456816011804e-06} +{"train_loss": 0.014969521202147007, "grad_norm": 0.07862290740013123, "global_step": 43245, "epoch": 407, "lr": 8.126837062846193e-06} +{"train_loss": 0.01216842234134674, "grad_norm": 0.07019320875406265, "global_step": 43246, "epoch": 407, "lr": 8.125217456804663e-06} +{"train_loss": 0.014996201880627646, "grad_norm": 0.10733973234891891, "global_step": 43247, "epoch": 407, "lr": 8.123597997892918e-06, "val_loss": 0.019388608634471893} +{"train_loss": 0.014275554567575455, "grad_norm": 0.10776452720165253, "global_step": 43248, "epoch": 408, "lr": 8.121978686116638e-06} +{"train_loss": 0.014848808757960796, "grad_norm": 0.07780084013938904, "global_step": 43249, "epoch": 408, "lr": 8.120359521481501e-06} +{"train_loss": 0.014603028073906898, "grad_norm": 0.0663084164261818, "global_step": 43250, "epoch": 408, "lr": 8.118740503993211e-06} +{"train_loss": 0.015617466531693935, "grad_norm": 0.0905894935131073, "global_step": 43251, "epoch": 408, "lr": 8.117121633657465e-06} +{"train_loss": 0.01622377336025238, "grad_norm": 0.06017247587442398, "global_step": 43252, "epoch": 408, "lr": 8.115502910479927e-06} +{"train_loss": 0.013492242433130741, "grad_norm": 0.13455088436603546, "global_step": 43253, "epoch": 408, "lr": 8.113884334466304e-06} +{"train_loss": 0.014660939574241638, "grad_norm": 0.1352614313364029, "global_step": 43254, "epoch": 408, "lr": 8.112265905622268e-06} +{"train_loss": 0.015973374247550964, "grad_norm": 0.06803970783948898, "global_step": 43255, "epoch": 408, "lr": 8.110647623953527e-06} +{"train_loss": 0.016526006162166595, "grad_norm": 0.1325976401567459, "global_step": 43256, "epoch": 408, "lr": 8.109029489465736e-06} +{"train_loss": 0.012771423906087875, "grad_norm": 0.05595230683684349, "global_step": 43257, "epoch": 408, "lr": 8.107411502164597e-06} +{"train_loss": 0.014407833106815815, "grad_norm": 0.05944235622882843, "global_step": 43258, "epoch": 408, "lr": 8.10579366205581e-06} +{"train_loss": 0.015761371701955795, "grad_norm": 0.10414701700210571, "global_step": 43259, "epoch": 408, "lr": 8.104175969145045e-06} +{"train_loss": 0.012887004762887955, "grad_norm": 0.05816534534096718, "global_step": 43260, "epoch": 408, "lr": 8.102558423437972e-06} +{"train_loss": 0.014137043617665768, "grad_norm": 0.11129432916641235, "global_step": 43261, "epoch": 408, "lr": 8.100941024940307e-06} +{"train_loss": 0.014578181318938732, "grad_norm": 0.07905981689691544, "global_step": 43262, "epoch": 408, "lr": 8.099323773657692e-06} +{"train_loss": 0.013717327266931534, "grad_norm": 0.09423774480819702, "global_step": 43263, "epoch": 408, "lr": 8.097706669595833e-06} +{"train_loss": 0.014976824633777142, "grad_norm": 0.0753956213593483, "global_step": 43264, "epoch": 408, "lr": 8.09608971276043e-06} +{"train_loss": 0.01347593404352665, "grad_norm": 0.09457667917013168, "global_step": 43265, "epoch": 408, "lr": 8.094472903157135e-06} +{"train_loss": 0.013743549585342407, "grad_norm": 0.06293169409036636, "global_step": 43266, "epoch": 408, "lr": 8.092856240791629e-06} +{"train_loss": 0.01448308490216732, "grad_norm": 0.11483141034841537, "global_step": 43267, "epoch": 408, "lr": 8.091239725669602e-06} +{"train_loss": 0.013881785795092583, "grad_norm": 0.07465874403715134, "global_step": 43268, "epoch": 408, "lr": 8.089623357796743e-06} +{"train_loss": 0.014704765751957893, "grad_norm": 0.07157999277114868, "global_step": 43269, "epoch": 408, "lr": 8.08800713717871e-06} +{"train_loss": 0.01272430457174778, "grad_norm": 0.07266521453857422, "global_step": 43270, "epoch": 408, "lr": 8.086391063821196e-06} +{"train_loss": 0.01602162793278694, "grad_norm": 0.07336880266666412, "global_step": 43271, "epoch": 408, "lr": 8.084775137729889e-06} +{"train_loss": 0.013804967515170574, "grad_norm": 0.0965198278427124, "global_step": 43272, "epoch": 408, "lr": 8.083159358910447e-06} +{"train_loss": 0.015095043927431107, "grad_norm": 0.0682135671377182, "global_step": 43273, "epoch": 408, "lr": 8.081543727368545e-06} +{"train_loss": 0.012905038893222809, "grad_norm": 0.0686812773346901, "global_step": 43274, "epoch": 408, "lr": 8.079928243109869e-06} +{"train_loss": 0.013445310294628143, "grad_norm": 0.07532674819231033, "global_step": 43275, "epoch": 408, "lr": 8.078312906140107e-06} +{"train_loss": 0.015162823721766472, "grad_norm": 0.07724594324827194, "global_step": 43276, "epoch": 408, "lr": 8.076697716464909e-06} +{"train_loss": 0.016700919717550278, "grad_norm": 0.10168953239917755, "global_step": 43277, "epoch": 408, "lr": 8.075082674089979e-06} +{"train_loss": 0.016567939892411232, "grad_norm": 0.06724945455789566, "global_step": 43278, "epoch": 408, "lr": 8.07346777902096e-06} +{"train_loss": 0.014990847557783127, "grad_norm": 0.0705300122499466, "global_step": 43279, "epoch": 408, "lr": 8.071853031263554e-06} +{"train_loss": 0.015018893405795097, "grad_norm": 0.0865422934293747, "global_step": 43280, "epoch": 408, "lr": 8.070238430823407e-06} +{"train_loss": 0.016508787870407104, "grad_norm": 0.07001728564500809, "global_step": 43281, "epoch": 408, "lr": 8.068623977706213e-06} +{"train_loss": 0.01483125053346157, "grad_norm": 0.08364338427782059, "global_step": 43282, "epoch": 408, "lr": 8.067009671917648e-06} +{"train_loss": 0.015294861979782581, "grad_norm": 0.07840131968259811, "global_step": 43283, "epoch": 408, "lr": 8.065395513463375e-06} +{"train_loss": 0.01601548306643963, "grad_norm": 0.06808457523584366, "global_step": 43284, "epoch": 408, "lr": 8.063781502349054e-06} +{"train_loss": 0.01513027772307396, "grad_norm": 0.07360347360372543, "global_step": 43285, "epoch": 408, "lr": 8.062167638580382e-06} +{"train_loss": 0.01429230347275734, "grad_norm": 0.062306150794029236, "global_step": 43286, "epoch": 408, "lr": 8.060553922163e-06} +{"train_loss": 0.013871070928871632, "grad_norm": 0.0892028883099556, "global_step": 43287, "epoch": 408, "lr": 8.05894035310259e-06} +{"train_loss": 0.016238730400800705, "grad_norm": 0.14176347851753235, "global_step": 43288, "epoch": 408, "lr": 8.05732693140484e-06} +{"train_loss": 0.017531614750623703, "grad_norm": 0.10506082326173782, "global_step": 43289, "epoch": 408, "lr": 8.055713657075403e-06} +{"train_loss": 0.014624813571572304, "grad_norm": 0.14067909121513367, "global_step": 43290, "epoch": 408, "lr": 8.054100530119935e-06} +{"train_loss": 0.015846755355596542, "grad_norm": 0.07308994233608246, "global_step": 43291, "epoch": 408, "lr": 8.052487550544113e-06} +{"train_loss": 0.012651141732931137, "grad_norm": 0.08319868892431259, "global_step": 43292, "epoch": 408, "lr": 8.050874718353618e-06} +{"train_loss": 0.013836611062288284, "grad_norm": 0.07017699629068375, "global_step": 43293, "epoch": 408, "lr": 8.049262033554101e-06} +{"train_loss": 0.014800197444856167, "grad_norm": 0.08956773579120636, "global_step": 43294, "epoch": 408, "lr": 8.047649496151238e-06} +{"train_loss": 0.014949165284633636, "grad_norm": 0.09639778733253479, "global_step": 43295, "epoch": 408, "lr": 8.046037106150684e-06} +{"train_loss": 0.017715226858854294, "grad_norm": 0.09227712452411652, "global_step": 43296, "epoch": 408, "lr": 8.044424863558114e-06} +{"train_loss": 0.015239223837852478, "grad_norm": 0.053202997893095016, "global_step": 43297, "epoch": 408, "lr": 8.042812768379187e-06} +{"train_loss": 0.011813700199127197, "grad_norm": 0.07917670905590057, "global_step": 43298, "epoch": 408, "lr": 8.041200820619565e-06} +{"train_loss": 0.018094949424266815, "grad_norm": 0.07745558023452759, "global_step": 43299, "epoch": 408, "lr": 8.039589020284926e-06} +{"train_loss": 0.014330188743770123, "grad_norm": 0.09820137172937393, "global_step": 43300, "epoch": 408, "lr": 8.037977367380922e-06} +{"train_loss": 0.014588071033358574, "grad_norm": 0.11525367200374603, "global_step": 43301, "epoch": 408, "lr": 8.03636586191321e-06} +{"train_loss": 0.015399966388940811, "grad_norm": 0.08443521708250046, "global_step": 43302, "epoch": 408, "lr": 8.034754503887465e-06} +{"train_loss": 0.01705227419734001, "grad_norm": 0.07650832831859589, "global_step": 43303, "epoch": 408, "lr": 8.033143293309336e-06} +{"train_loss": 0.014538583345711231, "grad_norm": 0.15242859721183777, "global_step": 43304, "epoch": 408, "lr": 8.03153223018449e-06} +{"train_loss": 0.016248812898993492, "grad_norm": 0.07556995004415512, "global_step": 43305, "epoch": 408, "lr": 8.029921314518595e-06} +{"train_loss": 0.019150493666529655, "grad_norm": 0.08430441468954086, "global_step": 43306, "epoch": 408, "lr": 8.028310546317308e-06} +{"train_loss": 0.01450004056096077, "grad_norm": 0.059077195823192596, "global_step": 43307, "epoch": 408, "lr": 8.026699925586272e-06} +{"train_loss": 0.016920531168580055, "grad_norm": 0.06699780374765396, "global_step": 43308, "epoch": 408, "lr": 8.025089452331164e-06} +{"train_loss": 0.017195917665958405, "grad_norm": 0.09482980519533157, "global_step": 43309, "epoch": 408, "lr": 8.023479126557642e-06} +{"train_loss": 0.015349917113780975, "grad_norm": 0.09419263899326324, "global_step": 43310, "epoch": 408, "lr": 8.02186894827135e-06} +{"train_loss": 0.013692858628928661, "grad_norm": 0.11076178401708603, "global_step": 43311, "epoch": 408, "lr": 8.020258917477952e-06} +{"train_loss": 0.01683536171913147, "grad_norm": 0.08509568870067596, "global_step": 43312, "epoch": 408, "lr": 8.018649034183123e-06} +{"train_loss": 0.016100384294986725, "grad_norm": 0.11660031974315643, "global_step": 43313, "epoch": 408, "lr": 8.017039298392504e-06} +{"train_loss": 0.017320938408374786, "grad_norm": 0.07748620212078094, "global_step": 43314, "epoch": 408, "lr": 8.015429710111739e-06} +{"train_loss": 0.015729226171970367, "grad_norm": 0.06866783648729324, "global_step": 43315, "epoch": 408, "lr": 8.013820269346495e-06} +{"train_loss": 0.015906095504760742, "grad_norm": 0.08737713098526001, "global_step": 43316, "epoch": 408, "lr": 8.012210976102435e-06} +{"train_loss": 0.013244220986962318, "grad_norm": 0.09059698134660721, "global_step": 43317, "epoch": 408, "lr": 8.010601830385201e-06} +{"train_loss": 0.015676986426115036, "grad_norm": 0.14231672883033752, "global_step": 43318, "epoch": 408, "lr": 8.008992832200457e-06} +{"train_loss": 0.014247095212340355, "grad_norm": 0.11345086991786957, "global_step": 43319, "epoch": 408, "lr": 8.007383981553857e-06} +{"train_loss": 0.014202957972884178, "grad_norm": 0.10633182525634766, "global_step": 43320, "epoch": 408, "lr": 8.005775278451028e-06} +{"train_loss": 0.014156905934214592, "grad_norm": 0.08071578294038773, "global_step": 43321, "epoch": 408, "lr": 8.004166722897649e-06} +{"train_loss": 0.015093078836798668, "grad_norm": 0.06880079954862595, "global_step": 43322, "epoch": 408, "lr": 8.00255831489936e-06} +{"train_loss": 0.013884466141462326, "grad_norm": 0.07488397508859634, "global_step": 43323, "epoch": 408, "lr": 8.000950054461832e-06} +{"train_loss": 0.014852406457066536, "grad_norm": 0.10621090233325958, "global_step": 43324, "epoch": 408, "lr": 7.999341941590698e-06} +{"train_loss": 0.015946708619594574, "grad_norm": 0.08855768293142319, "global_step": 43325, "epoch": 408, "lr": 7.9977339762916e-06} +{"train_loss": 0.01766521856188774, "grad_norm": 0.14775419235229492, "global_step": 43326, "epoch": 408, "lr": 7.996126158570211e-06} +{"train_loss": 0.015084398910403252, "grad_norm": 0.11086469143629074, "global_step": 43327, "epoch": 408, "lr": 7.994518488432157e-06} +{"train_loss": 0.015322251245379448, "grad_norm": 0.10081300139427185, "global_step": 43328, "epoch": 408, "lr": 7.992910965883093e-06} +{"train_loss": 0.014204638078808784, "grad_norm": 0.07102243602275848, "global_step": 43329, "epoch": 408, "lr": 7.991303590928684e-06} +{"train_loss": 0.013383720070123672, "grad_norm": 0.0974113941192627, "global_step": 43330, "epoch": 408, "lr": 7.989696363574567e-06} +{"train_loss": 0.014705189503729343, "grad_norm": 0.06416616588830948, "global_step": 43331, "epoch": 408, "lr": 7.988089283826377e-06} +{"train_loss": 0.015248489566147327, "grad_norm": 0.05993717908859253, "global_step": 43332, "epoch": 408, "lr": 7.986482351689767e-06} +{"train_loss": 0.012273231521248817, "grad_norm": 0.06981271505355835, "global_step": 43333, "epoch": 408, "lr": 7.984875567170402e-06} +{"train_loss": 0.0132498349994421, "grad_norm": 0.0656745582818985, "global_step": 43334, "epoch": 408, "lr": 7.983268930273902e-06} +{"train_loss": 0.013414217159152031, "grad_norm": 0.06308750063180923, "global_step": 43335, "epoch": 408, "lr": 7.981662441005916e-06} +{"train_loss": 0.01374820712953806, "grad_norm": 0.07259458303451538, "global_step": 43336, "epoch": 408, "lr": 7.980056099372113e-06} +{"train_loss": 0.013141696341335773, "grad_norm": 0.07016277313232422, "global_step": 43337, "epoch": 408, "lr": 7.978449905378115e-06} +{"train_loss": 0.017394158989191055, "grad_norm": 0.08010409772396088, "global_step": 43338, "epoch": 408, "lr": 7.976843859029558e-06} +{"train_loss": 0.01531762070953846, "grad_norm": 0.10610092431306839, "global_step": 43339, "epoch": 408, "lr": 7.975237960332099e-06} +{"train_loss": 0.01814144104719162, "grad_norm": 0.11332260072231293, "global_step": 43340, "epoch": 408, "lr": 7.973632209291387e-06} +{"train_loss": 0.01460960227996111, "grad_norm": 0.1278325915336609, "global_step": 43341, "epoch": 408, "lr": 7.972026605913052e-06} +{"train_loss": 0.014356499537825584, "grad_norm": 0.07243292778730392, "global_step": 43342, "epoch": 408, "lr": 7.970421150202733e-06} +{"train_loss": 0.01308370940387249, "grad_norm": 0.05058250203728676, "global_step": 43343, "epoch": 408, "lr": 7.968815842166082e-06} +{"train_loss": 0.013562343083322048, "grad_norm": 0.05428936704993248, "global_step": 43344, "epoch": 408, "lr": 7.967210681808723e-06} +{"train_loss": 0.014358957298099995, "grad_norm": 0.06463392823934555, "global_step": 43345, "epoch": 408, "lr": 7.965605669136305e-06} +{"train_loss": 0.015019100159406662, "grad_norm": 0.08146093785762787, "global_step": 43346, "epoch": 408, "lr": 7.964000804154486e-06} +{"train_loss": 0.015798885375261307, "grad_norm": 0.08348669856786728, "global_step": 43347, "epoch": 408, "lr": 7.962396086868884e-06} +{"train_loss": 0.014003010466694832, "grad_norm": 0.12573491036891937, "global_step": 43348, "epoch": 408, "lr": 7.960791517285127e-06} +{"train_loss": 0.014374801889061928, "grad_norm": 0.05935567244887352, "global_step": 43349, "epoch": 408, "lr": 7.959187095408866e-06} +{"train_loss": 0.015620316378772259, "grad_norm": 0.06603078544139862, "global_step": 43350, "epoch": 408, "lr": 7.957582821245751e-06} +{"train_loss": 0.01754193753004074, "grad_norm": 0.09062079340219498, "global_step": 43351, "epoch": 408, "lr": 7.955978694801392e-06} +{"train_loss": 0.015112446621060371, "grad_norm": 0.09817500412464142, "global_step": 43352, "epoch": 408, "lr": 7.954374716081442e-06} +{"train_loss": 0.014995291055935734, "grad_norm": 0.10716146975755692, "global_step": 43353, "epoch": 408, "lr": 7.952770885091548e-06, "val_loss": 0.01940491981804371} +{"train_loss": 0.015014659613370895, "grad_norm": 0.08296240866184235, "global_step": 43354, "epoch": 409, "lr": 7.951167201837329e-06} +{"train_loss": 0.01332150585949421, "grad_norm": 0.06357737630605698, "global_step": 43355, "epoch": 409, "lr": 7.94956366632441e-06} +{"train_loss": 0.014536494389176369, "grad_norm": 0.08290422707796097, "global_step": 43356, "epoch": 409, "lr": 7.947960278558436e-06} +{"train_loss": 0.014433692209422588, "grad_norm": 0.11516797542572021, "global_step": 43357, "epoch": 409, "lr": 7.946357038545054e-06} +{"train_loss": 0.014380628243088722, "grad_norm": 0.061597179621458054, "global_step": 43358, "epoch": 409, "lr": 7.94475394628988e-06} +{"train_loss": 0.014193207025527954, "grad_norm": 0.08853072673082352, "global_step": 43359, "epoch": 409, "lr": 7.943151001798554e-06} +{"train_loss": 0.014803921803832054, "grad_norm": 0.07882677763700485, "global_step": 43360, "epoch": 409, "lr": 7.94154820507671e-06} +{"train_loss": 0.01465076208114624, "grad_norm": 0.08800575137138367, "global_step": 43361, "epoch": 409, "lr": 7.939945556129957e-06} +{"train_loss": 0.014870920218527317, "grad_norm": 0.09486203640699387, "global_step": 43362, "epoch": 409, "lr": 7.938343054963948e-06} +{"train_loss": 0.016998957842588425, "grad_norm": 0.0881292074918747, "global_step": 43363, "epoch": 409, "lr": 7.936740701584321e-06} +{"train_loss": 0.015282711945474148, "grad_norm": 0.07923239469528198, "global_step": 43364, "epoch": 409, "lr": 7.935138495996685e-06} +{"train_loss": 0.012938681989908218, "grad_norm": 0.07498550415039062, "global_step": 43365, "epoch": 409, "lr": 7.93353643820669e-06} +{"train_loss": 0.014571135863661766, "grad_norm": 0.08742470294237137, "global_step": 43366, "epoch": 409, "lr": 7.931934528219936e-06} +{"train_loss": 0.014406004920601845, "grad_norm": 0.0861053317785263, "global_step": 43367, "epoch": 409, "lr": 7.930332766042086e-06} +{"train_loss": 0.01605420745909214, "grad_norm": 0.07189144194126129, "global_step": 43368, "epoch": 409, "lr": 7.928731151678743e-06} +{"train_loss": 0.015255511738359928, "grad_norm": 0.09417584538459778, "global_step": 43369, "epoch": 409, "lr": 7.927129685135537e-06} +{"train_loss": 0.01437779888510704, "grad_norm": 0.08873052150011063, "global_step": 43370, "epoch": 409, "lr": 7.925528366418112e-06} +{"train_loss": 0.014350784942507744, "grad_norm": 0.06760725378990173, "global_step": 43371, "epoch": 409, "lr": 7.923927195532083e-06} +{"train_loss": 0.014567273668944836, "grad_norm": 0.08187956362962723, "global_step": 43372, "epoch": 409, "lr": 7.922326172483064e-06} +{"train_loss": 0.014218220487236977, "grad_norm": 0.06226499751210213, "global_step": 43373, "epoch": 409, "lr": 7.920725297276693e-06} +{"train_loss": 0.013919247314333916, "grad_norm": 0.07336084544658661, "global_step": 43374, "epoch": 409, "lr": 7.919124569918601e-06} +{"train_loss": 0.01725047267973423, "grad_norm": 0.06249101087450981, "global_step": 43375, "epoch": 409, "lr": 7.9175239904144e-06} +{"train_loss": 0.013281870633363724, "grad_norm": 0.06121020019054413, "global_step": 43376, "epoch": 409, "lr": 7.915923558769712e-06} +{"train_loss": 0.015140562318265438, "grad_norm": 0.09092686325311661, "global_step": 43377, "epoch": 409, "lr": 7.914323274990182e-06} +{"train_loss": 0.01812427118420601, "grad_norm": 0.10411670058965683, "global_step": 43378, "epoch": 409, "lr": 7.912723139081418e-06} +{"train_loss": 0.014596959576010704, "grad_norm": 0.10609974712133408, "global_step": 43379, "epoch": 409, "lr": 7.911123151049026e-06} +{"train_loss": 0.014498185366392136, "grad_norm": 0.057831451296806335, "global_step": 43380, "epoch": 409, "lr": 7.909523310898648e-06} +{"train_loss": 0.017323937267065048, "grad_norm": 0.06429605931043625, "global_step": 43381, "epoch": 409, "lr": 7.90792361863591e-06} +{"train_loss": 0.017719510942697525, "grad_norm": 0.06903053820133209, "global_step": 43382, "epoch": 409, "lr": 7.906324074266409e-06} +{"train_loss": 0.014924051240086555, "grad_norm": 0.06545792520046234, "global_step": 43383, "epoch": 409, "lr": 7.904724677795795e-06} +{"train_loss": 0.014413822442293167, "grad_norm": 0.055469684302806854, "global_step": 43384, "epoch": 409, "lr": 7.90312542922967e-06} +{"train_loss": 0.014423573389649391, "grad_norm": 0.12424419820308685, "global_step": 43385, "epoch": 409, "lr": 7.901526328573645e-06} +{"train_loss": 0.015485798940062523, "grad_norm": 0.07667317241430283, "global_step": 43386, "epoch": 409, "lr": 7.899927375833344e-06} +{"train_loss": 0.01414249837398529, "grad_norm": 0.05888655036687851, "global_step": 43387, "epoch": 409, "lr": 7.898328571014402e-06} +{"train_loss": 0.013905507512390614, "grad_norm": 0.08029626309871674, "global_step": 43388, "epoch": 409, "lr": 7.896729914122413e-06} +{"train_loss": 0.01586611568927765, "grad_norm": 0.06292465329170227, "global_step": 43389, "epoch": 409, "lr": 7.89513140516302e-06} +{"train_loss": 0.015532644465565681, "grad_norm": 0.07781604677438736, "global_step": 43390, "epoch": 409, "lr": 7.89353304414181e-06} +{"train_loss": 0.015436023473739624, "grad_norm": 0.10223609209060669, "global_step": 43391, "epoch": 409, "lr": 7.891934831064424e-06} +{"train_loss": 0.015729427337646484, "grad_norm": 0.08285271376371384, "global_step": 43392, "epoch": 409, "lr": 7.890336765936458e-06} +{"train_loss": 0.014203608967363834, "grad_norm": 0.09134966880083084, "global_step": 43393, "epoch": 409, "lr": 7.888738848763532e-06} +{"train_loss": 0.013064993545413017, "grad_norm": 0.11545290052890778, "global_step": 43394, "epoch": 409, "lr": 7.887141079551274e-06} +{"train_loss": 0.013978255912661552, "grad_norm": 0.08329805731773376, "global_step": 43395, "epoch": 409, "lr": 7.885543458305294e-06} +{"train_loss": 0.014445260167121887, "grad_norm": 0.08679430931806564, "global_step": 43396, "epoch": 409, "lr": 7.883945985031183e-06} +{"train_loss": 0.01563430018723011, "grad_norm": 0.05464716628193855, "global_step": 43397, "epoch": 409, "lr": 7.882348659734573e-06} +{"train_loss": 0.015194286592304707, "grad_norm": 0.06228771433234215, "global_step": 43398, "epoch": 409, "lr": 7.880751482421083e-06} +{"train_loss": 0.013504356145858765, "grad_norm": 0.06918717920780182, "global_step": 43399, "epoch": 409, "lr": 7.879154453096304e-06} +{"train_loss": 0.013979141600430012, "grad_norm": 0.0610438697040081, "global_step": 43400, "epoch": 409, "lr": 7.87755757176587e-06} +{"train_loss": 0.013820021413266659, "grad_norm": 0.14623712003231049, "global_step": 43401, "epoch": 409, "lr": 7.875960838435381e-06} +{"train_loss": 0.013426017016172409, "grad_norm": 0.06528154015541077, "global_step": 43402, "epoch": 409, "lr": 7.874364253110433e-06} +{"train_loss": 0.015104877762496471, "grad_norm": 0.06388024985790253, "global_step": 43403, "epoch": 409, "lr": 7.87276781579665e-06} +{"train_loss": 0.014222816564142704, "grad_norm": 0.10844167321920395, "global_step": 43404, "epoch": 409, "lr": 7.871171526499648e-06} +{"train_loss": 0.018112169578671455, "grad_norm": 0.07719381153583527, "global_step": 43405, "epoch": 409, "lr": 7.86957538522502e-06} +{"train_loss": 0.017223920673131943, "grad_norm": 0.10402843356132507, "global_step": 43406, "epoch": 409, "lr": 7.867979391978397e-06} +{"train_loss": 0.014545307494699955, "grad_norm": 0.10236470401287079, "global_step": 43407, "epoch": 409, "lr": 7.866383546765354e-06} +{"train_loss": 0.014109589159488678, "grad_norm": 0.0863950178027153, "global_step": 43408, "epoch": 409, "lr": 7.864787849591531e-06} +{"train_loss": 0.017279580235481262, "grad_norm": 0.09698081016540527, "global_step": 43409, "epoch": 409, "lr": 7.863192300462508e-06} +{"train_loss": 0.012994566932320595, "grad_norm": 0.06673615425825119, "global_step": 43410, "epoch": 409, "lr": 7.8615968993839e-06} +{"train_loss": 0.016447439789772034, "grad_norm": 0.08988533914089203, "global_step": 43411, "epoch": 409, "lr": 7.860001646361325e-06} +{"train_loss": 0.014817934483289719, "grad_norm": 0.0939382016658783, "global_step": 43412, "epoch": 409, "lr": 7.858406541400382e-06} +{"train_loss": 0.016155418008565903, "grad_norm": 0.09437020868062973, "global_step": 43413, "epoch": 409, "lr": 7.856811584506663e-06} +{"train_loss": 0.016667647287249565, "grad_norm": 0.12297491729259491, "global_step": 43414, "epoch": 409, "lr": 7.855216775685776e-06} +{"train_loss": 0.015216664411127567, "grad_norm": 0.1462114006280899, "global_step": 43415, "epoch": 409, "lr": 7.853622114943337e-06} +{"train_loss": 0.014294978231191635, "grad_norm": 0.09695811569690704, "global_step": 43416, "epoch": 409, "lr": 7.852027602284933e-06} +{"train_loss": 0.014611095190048218, "grad_norm": 0.06414705514907837, "global_step": 43417, "epoch": 409, "lr": 7.850433237716175e-06} +{"train_loss": 0.0176458228379488, "grad_norm": 0.1176476776599884, "global_step": 43418, "epoch": 409, "lr": 7.848839021242687e-06} +{"train_loss": 0.015594306401908398, "grad_norm": 0.08275827020406723, "global_step": 43419, "epoch": 409, "lr": 7.847244952870026e-06} +{"train_loss": 0.015393136069178581, "grad_norm": 0.07542936503887177, "global_step": 43420, "epoch": 409, "lr": 7.845651032603812e-06} +{"train_loss": 0.016276778653264046, "grad_norm": 0.12582333385944366, "global_step": 43421, "epoch": 409, "lr": 7.84405726044966e-06} +{"train_loss": 0.01467808336019516, "grad_norm": 0.06776700913906097, "global_step": 43422, "epoch": 409, "lr": 7.84246363641315e-06} +{"train_loss": 0.01648678630590439, "grad_norm": 0.0935979038476944, "global_step": 43423, "epoch": 409, "lr": 7.84087016049988e-06} +{"train_loss": 0.015909560024738312, "grad_norm": 0.09939856082201004, "global_step": 43424, "epoch": 409, "lr": 7.839276832715475e-06} +{"train_loss": 0.01633610762655735, "grad_norm": 0.05222531408071518, "global_step": 43425, "epoch": 409, "lr": 7.837683653065519e-06} +{"train_loss": 0.012715861201286316, "grad_norm": 0.10085190832614899, "global_step": 43426, "epoch": 409, "lr": 7.836090621555593e-06} +{"train_loss": 0.013643954880535603, "grad_norm": 0.08790121227502823, "global_step": 43427, "epoch": 409, "lr": 7.834497738191305e-06} +{"train_loss": 0.014550385996699333, "grad_norm": 0.06025600805878639, "global_step": 43428, "epoch": 409, "lr": 7.832905002978269e-06} +{"train_loss": 0.015182917937636375, "grad_norm": 0.09406956285238266, "global_step": 43429, "epoch": 409, "lr": 7.831312415922054e-06} +{"train_loss": 0.01574622467160225, "grad_norm": 0.07391610741615295, "global_step": 43430, "epoch": 409, "lr": 7.829719977028277e-06} +{"train_loss": 0.017005501314997673, "grad_norm": 0.11116816103458405, "global_step": 43431, "epoch": 409, "lr": 7.828127686302517e-06} +{"train_loss": 0.015889424830675125, "grad_norm": 0.07571721076965332, "global_step": 43432, "epoch": 409, "lr": 7.826535543750391e-06} +{"train_loss": 0.014406785368919373, "grad_norm": 0.11605645716190338, "global_step": 43433, "epoch": 409, "lr": 7.824943549377461e-06} +{"train_loss": 0.01562068797647953, "grad_norm": 0.08390943706035614, "global_step": 43434, "epoch": 409, "lr": 7.82335170318934e-06} +{"train_loss": 0.012819038704037666, "grad_norm": 0.07490129768848419, "global_step": 43435, "epoch": 409, "lr": 7.82176000519163e-06} +{"train_loss": 0.01520664431154728, "grad_norm": 0.22834986448287964, "global_step": 43436, "epoch": 409, "lr": 7.820168455389914e-06} +{"train_loss": 0.014372391626238823, "grad_norm": 0.08485864102840424, "global_step": 43437, "epoch": 409, "lr": 7.818577053789766e-06} +{"train_loss": 0.015020220540463924, "grad_norm": 0.1435508131980896, "global_step": 43438, "epoch": 409, "lr": 7.816985800396803e-06} +{"train_loss": 0.014689800329506397, "grad_norm": 0.05725222826004028, "global_step": 43439, "epoch": 409, "lr": 7.81539469521661e-06} +{"train_loss": 0.01576613262295723, "grad_norm": 0.07106992602348328, "global_step": 43440, "epoch": 409, "lr": 7.813803738254765e-06} +{"train_loss": 0.013094990514218807, "grad_norm": 0.14356891810894012, "global_step": 43441, "epoch": 409, "lr": 7.812212929516883e-06} +{"train_loss": 0.01450185477733612, "grad_norm": 0.08465573936700821, "global_step": 43442, "epoch": 409, "lr": 7.810622269008533e-06} +{"train_loss": 0.014418511651456356, "grad_norm": 0.12324601411819458, "global_step": 43443, "epoch": 409, "lr": 7.809031756735302e-06} +{"train_loss": 0.014672596007585526, "grad_norm": 0.07652212679386139, "global_step": 43444, "epoch": 409, "lr": 7.80744139270278e-06} +{"train_loss": 0.0152061115950346, "grad_norm": 0.07941776514053345, "global_step": 43445, "epoch": 409, "lr": 7.805851176916574e-06} +{"train_loss": 0.0147504648193717, "grad_norm": 0.11741780489683151, "global_step": 43446, "epoch": 409, "lr": 7.804261109382244e-06} +{"train_loss": 0.014674952253699303, "grad_norm": 0.13234597444534302, "global_step": 43447, "epoch": 409, "lr": 7.802671190105405e-06} +{"train_loss": 0.012428469024598598, "grad_norm": 0.09785519540309906, "global_step": 43448, "epoch": 409, "lr": 7.801081419091616e-06} +{"train_loss": 0.015247009694576263, "grad_norm": 0.0777340978384018, "global_step": 43449, "epoch": 409, "lr": 7.799491796346486e-06} +{"train_loss": 0.015060918405652046, "grad_norm": 0.08955099433660507, "global_step": 43450, "epoch": 409, "lr": 7.797902321875577e-06} +{"train_loss": 0.01630963757634163, "grad_norm": 0.08779999613761902, "global_step": 43451, "epoch": 409, "lr": 7.796312995684485e-06} +{"train_loss": 0.013622231781482697, "grad_norm": 0.07371124625205994, "global_step": 43452, "epoch": 409, "lr": 7.794723817778809e-06} +{"train_loss": 0.016153234988451004, "grad_norm": 0.08112096041440964, "global_step": 43453, "epoch": 409, "lr": 7.793134788164113e-06} +{"train_loss": 0.015303539112210274, "grad_norm": 0.11644919961690903, "global_step": 43454, "epoch": 409, "lr": 7.791545906845982e-06} +{"train_loss": 0.013226576149463654, "grad_norm": 0.10169903188943863, "global_step": 43455, "epoch": 409, "lr": 7.789957173829998e-06} +{"train_loss": 0.016282740980386734, "grad_norm": 0.06669483333826065, "global_step": 43456, "epoch": 409, "lr": 7.788368589121765e-06} +{"train_loss": 0.015568889677524567, "grad_norm": 0.08135581761598587, "global_step": 43457, "epoch": 409, "lr": 7.78678015272683e-06} +{"train_loss": 0.015325228683650494, "grad_norm": 0.11910288780927658, "global_step": 43458, "epoch": 409, "lr": 7.785191864650792e-06} +{"train_loss": 0.014980637406895184, "grad_norm": 0.06828787177801132, "global_step": 43459, "epoch": 409, "lr": 7.783603724899257e-06, "val_loss": 0.01938525028526783} +{"train_loss": 0.01522074919193983, "grad_norm": 0.0631522536277771, "global_step": 43460, "epoch": 410, "lr": 7.782015733477754e-06} +{"train_loss": 0.01654190942645073, "grad_norm": 0.12649168074131012, "global_step": 43461, "epoch": 410, "lr": 7.78042789039189e-06} +{"train_loss": 0.0127573162317276, "grad_norm": 0.08284711092710495, "global_step": 43462, "epoch": 410, "lr": 7.778840195647252e-06} +{"train_loss": 0.01554830651730299, "grad_norm": 0.08144179731607437, "global_step": 43463, "epoch": 410, "lr": 7.777252649249395e-06} +{"train_loss": 0.015816176310181618, "grad_norm": 0.10896030813455582, "global_step": 43464, "epoch": 410, "lr": 7.775665251203912e-06} +{"train_loss": 0.013256722129881382, "grad_norm": 0.08834606409072876, "global_step": 43465, "epoch": 410, "lr": 7.774078001516388e-06} +{"train_loss": 0.013424146920442581, "grad_norm": 0.05551869794726372, "global_step": 43466, "epoch": 410, "lr": 7.772490900192391e-06} +{"train_loss": 0.01468598935753107, "grad_norm": 0.07641847431659698, "global_step": 43467, "epoch": 410, "lr": 7.77090394723749e-06} +{"train_loss": 0.014924821443855762, "grad_norm": 0.06662235409021378, "global_step": 43468, "epoch": 410, "lr": 7.769317142657262e-06} +{"train_loss": 0.014070016331970692, "grad_norm": 0.09748684614896774, "global_step": 43469, "epoch": 410, "lr": 7.767730486457297e-06} +{"train_loss": 0.014632739126682281, "grad_norm": 0.0862874835729599, "global_step": 43470, "epoch": 410, "lr": 7.766143978643154e-06} +{"train_loss": 0.013060132041573524, "grad_norm": 0.08029615134000778, "global_step": 43471, "epoch": 410, "lr": 7.764557619220425e-06} +{"train_loss": 0.016745414584875107, "grad_norm": 0.08081911504268646, "global_step": 43472, "epoch": 410, "lr": 7.762971408194663e-06} +{"train_loss": 0.016086602583527565, "grad_norm": 0.07790883630514145, "global_step": 43473, "epoch": 410, "lr": 7.761385345571459e-06} +{"train_loss": 0.014893470332026482, "grad_norm": 0.11977523565292358, "global_step": 43474, "epoch": 410, "lr": 7.759799431356369e-06} +{"train_loss": 0.015279507264494896, "grad_norm": 0.10139647871255875, "global_step": 43475, "epoch": 410, "lr": 7.75821366555497e-06} +{"train_loss": 0.014353247359395027, "grad_norm": 0.07009921222925186, "global_step": 43476, "epoch": 410, "lr": 7.756628048172854e-06} +{"train_loss": 0.014167293906211853, "grad_norm": 0.08400049805641174, "global_step": 43477, "epoch": 410, "lr": 7.75504257921557e-06} +{"train_loss": 0.013002301566302776, "grad_norm": 0.10394016653299332, "global_step": 43478, "epoch": 410, "lr": 7.75345725868869e-06} +{"train_loss": 0.014975485391914845, "grad_norm": 0.126817524433136, "global_step": 43479, "epoch": 410, "lr": 7.751872086597783e-06} +{"train_loss": 0.014165863394737244, "grad_norm": 0.17525315284729004, "global_step": 43480, "epoch": 410, "lr": 7.750287062948441e-06} +{"train_loss": 0.01569109596312046, "grad_norm": 0.1401488333940506, "global_step": 43481, "epoch": 410, "lr": 7.7487021877462e-06} +{"train_loss": 0.012102138251066208, "grad_norm": 0.14601072669029236, "global_step": 43482, "epoch": 410, "lr": 7.74711746099665e-06} +{"train_loss": 0.017742019146680832, "grad_norm": 0.14648906886577606, "global_step": 43483, "epoch": 410, "lr": 7.74553288270537e-06} +{"train_loss": 0.017481114715337753, "grad_norm": 0.11645860224962234, "global_step": 43484, "epoch": 410, "lr": 7.743948452877897e-06} +{"train_loss": 0.014239078387618065, "grad_norm": 0.07335571944713593, "global_step": 43485, "epoch": 410, "lr": 7.742364171519807e-06} +{"train_loss": 0.013338787481188774, "grad_norm": 0.06850041449069977, "global_step": 43486, "epoch": 410, "lr": 7.74078003863668e-06} +{"train_loss": 0.01617615669965744, "grad_norm": 0.06688546389341354, "global_step": 43487, "epoch": 410, "lr": 7.73919605423407e-06} +{"train_loss": 0.014070870354771614, "grad_norm": 0.09324108064174652, "global_step": 43488, "epoch": 410, "lr": 7.737612218317541e-06} +{"train_loss": 0.015096975490450859, "grad_norm": 0.14188207685947418, "global_step": 43489, "epoch": 410, "lr": 7.73602853089268e-06} +{"train_loss": 0.016529813408851624, "grad_norm": 0.0908685252070427, "global_step": 43490, "epoch": 410, "lr": 7.734444991965028e-06} +{"train_loss": 0.012990262359380722, "grad_norm": 0.10146136581897736, "global_step": 43491, "epoch": 410, "lr": 7.732861601540148e-06} +{"train_loss": 0.015799716114997864, "grad_norm": 0.11531278491020203, "global_step": 43492, "epoch": 410, "lr": 7.731278359623611e-06} +{"train_loss": 0.013086285442113876, "grad_norm": 0.09400996565818787, "global_step": 43493, "epoch": 410, "lr": 7.729695266220988e-06} +{"train_loss": 0.017348647117614746, "grad_norm": 0.09446011483669281, "global_step": 43494, "epoch": 410, "lr": 7.728112321337838e-06} +{"train_loss": 0.016075801104307175, "grad_norm": 0.14470037817955017, "global_step": 43495, "epoch": 410, "lr": 7.726529524979703e-06} +{"train_loss": 0.01379998866468668, "grad_norm": 0.06909751892089844, "global_step": 43496, "epoch": 410, "lr": 7.724946877152162e-06} +{"train_loss": 0.017316598445177078, "grad_norm": 0.18843872845172882, "global_step": 43497, "epoch": 410, "lr": 7.72336437786078e-06} +{"train_loss": 0.015762045979499817, "grad_norm": 0.09255150705575943, "global_step": 43498, "epoch": 410, "lr": 7.721782027111102e-06} +{"train_loss": 0.014624875970184803, "grad_norm": 0.07486255466938019, "global_step": 43499, "epoch": 410, "lr": 7.720199824908692e-06} +{"train_loss": 0.016731973737478256, "grad_norm": 0.07550140470266342, "global_step": 43500, "epoch": 410, "lr": 7.71861777125914e-06} +{"train_loss": 0.013559314422309399, "grad_norm": 0.08712281286716461, "global_step": 43501, "epoch": 410, "lr": 7.717035866167948e-06} +{"train_loss": 0.014101332053542137, "grad_norm": 0.08389806002378464, "global_step": 43502, "epoch": 410, "lr": 7.715454109640707e-06} +{"train_loss": 0.014908523298799992, "grad_norm": 0.05362153798341751, "global_step": 43503, "epoch": 410, "lr": 7.713872501682984e-06} +{"train_loss": 0.014358202926814556, "grad_norm": 0.06398782879114151, "global_step": 43504, "epoch": 410, "lr": 7.712291042300313e-06} +{"train_loss": 0.016588907688856125, "grad_norm": 0.09106127917766571, "global_step": 43505, "epoch": 410, "lr": 7.710709731498262e-06} +{"train_loss": 0.01592016965150833, "grad_norm": 0.07314455509185791, "global_step": 43506, "epoch": 410, "lr": 7.709128569282393e-06} +{"train_loss": 0.015889901667833328, "grad_norm": 0.1018557921051979, "global_step": 43507, "epoch": 410, "lr": 7.707547555658257e-06} +{"train_loss": 0.015711400657892227, "grad_norm": 0.07137571275234222, "global_step": 43508, "epoch": 410, "lr": 7.705966690631389e-06} +{"train_loss": 0.015531880781054497, "grad_norm": 0.08823420852422714, "global_step": 43509, "epoch": 410, "lr": 7.704385974207368e-06} +{"train_loss": 0.013553568162024021, "grad_norm": 0.10550706088542938, "global_step": 43510, "epoch": 410, "lr": 7.70280540639175e-06} +{"train_loss": 0.014323084615170956, "grad_norm": 0.09625933319330215, "global_step": 43511, "epoch": 410, "lr": 7.70122498719007e-06} +{"train_loss": 0.014972656965255737, "grad_norm": 0.08180543780326843, "global_step": 43512, "epoch": 410, "lr": 7.699644716607895e-06} +{"train_loss": 0.0155381103977561, "grad_norm": 0.09539168328046799, "global_step": 43513, "epoch": 410, "lr": 7.698064594650766e-06} +{"train_loss": 0.013763501308858395, "grad_norm": 0.06353159248828888, "global_step": 43514, "epoch": 410, "lr": 7.696484621324252e-06} +{"train_loss": 0.017099684104323387, "grad_norm": 0.11560230702161789, "global_step": 43515, "epoch": 410, "lr": 7.694904796633885e-06} +{"train_loss": 0.014304619282484055, "grad_norm": 0.10413981974124908, "global_step": 43516, "epoch": 410, "lr": 7.693325120585221e-06} +{"train_loss": 0.01657593995332718, "grad_norm": 0.09408776462078094, "global_step": 43517, "epoch": 410, "lr": 7.691745593183825e-06} +{"train_loss": 0.018133996054530144, "grad_norm": 0.06589073687791824, "global_step": 43518, "epoch": 410, "lr": 7.69016621443524e-06} +{"train_loss": 0.015273252502083778, "grad_norm": 0.08709632605314255, "global_step": 43519, "epoch": 410, "lr": 7.688586984344992e-06} +{"train_loss": 0.015110891312360764, "grad_norm": 0.06628535687923431, "global_step": 43520, "epoch": 410, "lr": 7.687007902918669e-06} +{"train_loss": 0.012605028226971626, "grad_norm": 0.06944610923528671, "global_step": 43521, "epoch": 410, "lr": 7.685428970161779e-06} +{"train_loss": 0.01409968826919794, "grad_norm": 0.12272309511899948, "global_step": 43522, "epoch": 410, "lr": 7.683850186079894e-06} +{"train_loss": 0.014824628829956055, "grad_norm": 0.10888795554637909, "global_step": 43523, "epoch": 410, "lr": 7.682271550678565e-06} +{"train_loss": 0.014232197776436806, "grad_norm": 0.11737661063671112, "global_step": 43524, "epoch": 410, "lr": 7.680693063963334e-06} +{"train_loss": 0.015580044128000736, "grad_norm": 0.05760467052459717, "global_step": 43525, "epoch": 410, "lr": 7.679114725939734e-06} +{"train_loss": 0.01546633243560791, "grad_norm": 0.13741041719913483, "global_step": 43526, "epoch": 410, "lr": 7.677536536613316e-06} +{"train_loss": 0.01378560345619917, "grad_norm": 0.10636446624994278, "global_step": 43527, "epoch": 410, "lr": 7.675958495989638e-06} +{"train_loss": 0.014639992266893387, "grad_norm": 0.10122449696063995, "global_step": 43528, "epoch": 410, "lr": 7.674380604074227e-06} +{"train_loss": 0.01742485910654068, "grad_norm": 0.08648087829351425, "global_step": 43529, "epoch": 410, "lr": 7.672802860872635e-06} +{"train_loss": 0.015769589692354202, "grad_norm": 0.09588892012834549, "global_step": 43530, "epoch": 410, "lr": 7.671225266390419e-06} +{"train_loss": 0.016351930797100067, "grad_norm": 0.08803803473711014, "global_step": 43531, "epoch": 410, "lr": 7.669647820633108e-06} +{"train_loss": 0.013093112967908382, "grad_norm": 0.05570301413536072, "global_step": 43532, "epoch": 410, "lr": 7.668070523606235e-06} +{"train_loss": 0.012020913884043694, "grad_norm": 0.08515272289514542, "global_step": 43533, "epoch": 410, "lr": 7.666493375315353e-06} +{"train_loss": 0.015035808086395264, "grad_norm": 0.06753301620483398, "global_step": 43534, "epoch": 410, "lr": 7.664916375766013e-06} +{"train_loss": 0.014177992939949036, "grad_norm": 0.12566782534122467, "global_step": 43535, "epoch": 410, "lr": 7.663339524963736e-06} +{"train_loss": 0.015078222379088402, "grad_norm": 0.07588416337966919, "global_step": 43536, "epoch": 410, "lr": 7.661762822914082e-06} +{"train_loss": 0.017743978649377823, "grad_norm": 0.09419964998960495, "global_step": 43537, "epoch": 410, "lr": 7.660186269622571e-06} +{"train_loss": 0.012271950021386147, "grad_norm": 0.066621333360672, "global_step": 43538, "epoch": 410, "lr": 7.658609865094762e-06} +{"train_loss": 0.016278833150863647, "grad_norm": 0.08385904878377914, "global_step": 43539, "epoch": 410, "lr": 7.657033609336173e-06} +{"train_loss": 0.013371452689170837, "grad_norm": 0.06061522290110588, "global_step": 43540, "epoch": 410, "lr": 7.655457502352353e-06} +{"train_loss": 0.01557072065770626, "grad_norm": 0.12061215192079544, "global_step": 43541, "epoch": 410, "lr": 7.653881544148856e-06} +{"train_loss": 0.012583187781274319, "grad_norm": 0.08062063157558441, "global_step": 43542, "epoch": 410, "lr": 7.652305734731196e-06} +{"train_loss": 0.016750570386648178, "grad_norm": 0.10234183818101883, "global_step": 43543, "epoch": 410, "lr": 7.650730074104912e-06} +{"train_loss": 0.015316714532673359, "grad_norm": 0.05485165864229202, "global_step": 43544, "epoch": 410, "lr": 7.649154562275556e-06} +{"train_loss": 0.014194137416779995, "grad_norm": 0.1635051965713501, "global_step": 43545, "epoch": 410, "lr": 7.64757919924864e-06} +{"train_loss": 0.01445554569363594, "grad_norm": 0.06553694605827332, "global_step": 43546, "epoch": 410, "lr": 7.64600398502971e-06} +{"train_loss": 0.015836605802178383, "grad_norm": 0.15255215764045715, "global_step": 43547, "epoch": 410, "lr": 7.644428919624319e-06} +{"train_loss": 0.014704260975122452, "grad_norm": 0.06908800452947617, "global_step": 43548, "epoch": 410, "lr": 7.64285400303798e-06} +{"train_loss": 0.016801320016384125, "grad_norm": 0.079537533223629, "global_step": 43549, "epoch": 410, "lr": 7.641279235276222e-06} +{"train_loss": 0.014101316221058369, "grad_norm": 0.1409357637166977, "global_step": 43550, "epoch": 410, "lr": 7.63970461634459e-06} +{"train_loss": 0.014978012070059776, "grad_norm": 0.09382784366607666, "global_step": 43551, "epoch": 410, "lr": 7.638130146248623e-06} +{"train_loss": 0.014874309301376343, "grad_norm": 0.11574121564626694, "global_step": 43552, "epoch": 410, "lr": 7.63655582499383e-06} +{"train_loss": 0.01538365613669157, "grad_norm": 0.06584278494119644, "global_step": 43553, "epoch": 410, "lr": 7.63498165258577e-06} +{"train_loss": 0.013533838093280792, "grad_norm": 0.06830353289842606, "global_step": 43554, "epoch": 410, "lr": 7.633407629029948e-06} +{"train_loss": 0.015227187424898148, "grad_norm": 0.06743013858795166, "global_step": 43555, "epoch": 410, "lr": 7.631833754331924e-06} +{"train_loss": 0.016451653093099594, "grad_norm": 0.08054826408624649, "global_step": 43556, "epoch": 410, "lr": 7.630260028497193e-06} +{"train_loss": 0.015102298930287361, "grad_norm": 0.08317697793245316, "global_step": 43557, "epoch": 410, "lr": 7.628686451531303e-06} +{"train_loss": 0.015520874410867691, "grad_norm": 0.11456739157438278, "global_step": 43558, "epoch": 410, "lr": 7.627113023439797e-06} +{"train_loss": 0.013692297972738743, "grad_norm": 0.05841945484280586, "global_step": 43559, "epoch": 410, "lr": 7.6255397442281825e-06} +{"train_loss": 0.015005631372332573, "grad_norm": 0.06520114094018936, "global_step": 43560, "epoch": 410, "lr": 7.623966613901989e-06} +{"train_loss": 0.014578773640096188, "grad_norm": 0.09012005478143692, "global_step": 43561, "epoch": 410, "lr": 7.622393632466751e-06} +{"train_loss": 0.015539765357971191, "grad_norm": 0.06636634469032288, "global_step": 43562, "epoch": 410, "lr": 7.620820799927986e-06} +{"train_loss": 0.01585085503757, "grad_norm": 0.10121486335992813, "global_step": 43563, "epoch": 410, "lr": 7.619248116291228e-06} +{"train_loss": 0.014637247659265995, "grad_norm": 0.07785563915967941, "global_step": 43564, "epoch": 410, "lr": 7.617675581562012e-06} +{"train_loss": 0.015004476435693368, "grad_norm": 0.08576203137636185, "global_step": 43565, "epoch": 410, "lr": 7.6161031957458494e-06, "train_batch_mse_error": 0.01338600181043148, "val_loss": 0.019547412171959877} +{"train_loss": 0.016146980226039886, "grad_norm": 0.06274755299091339, "global_step": 43566, "epoch": 411, "lr": 7.614530958848259e-06} +{"train_loss": 0.015099992975592613, "grad_norm": 0.12754294276237488, "global_step": 43567, "epoch": 411, "lr": 7.61295887087477e-06} +{"train_loss": 0.015140458941459656, "grad_norm": 0.07674610614776611, "global_step": 43568, "epoch": 411, "lr": 7.611386931830927e-06} +{"train_loss": 0.013773109763860703, "grad_norm": 0.06477890908718109, "global_step": 43569, "epoch": 411, "lr": 7.609815141722221e-06} +{"train_loss": 0.013424145057797432, "grad_norm": 0.1054387241601944, "global_step": 43570, "epoch": 411, "lr": 7.608243500554191e-06} +{"train_loss": 0.013503823429346085, "grad_norm": 0.11297143250703812, "global_step": 43571, "epoch": 411, "lr": 7.606672008332366e-06} +{"train_loss": 0.013850953429937363, "grad_norm": 0.06033565104007721, "global_step": 43572, "epoch": 411, "lr": 7.605100665062265e-06} +{"train_loss": 0.014647824689745903, "grad_norm": 0.0533580556511879, "global_step": 43573, "epoch": 411, "lr": 7.6035294707493875e-06} +{"train_loss": 0.014681818895041943, "grad_norm": 0.0851832777261734, "global_step": 43574, "epoch": 411, "lr": 7.601958425399269e-06} +{"train_loss": 0.013879813253879547, "grad_norm": 0.07924457639455795, "global_step": 43575, "epoch": 411, "lr": 7.600387529017444e-06} +{"train_loss": 0.016103221103549004, "grad_norm": 0.06647910177707672, "global_step": 43576, "epoch": 411, "lr": 7.598816781609403e-06} +{"train_loss": 0.013665549457073212, "grad_norm": 0.07622936367988586, "global_step": 43577, "epoch": 411, "lr": 7.597246183180695e-06} +{"train_loss": 0.015334540978074074, "grad_norm": 0.07365372776985168, "global_step": 43578, "epoch": 411, "lr": 7.595675733736824e-06} +{"train_loss": 0.016424832865595818, "grad_norm": 0.09437572956085205, "global_step": 43579, "epoch": 411, "lr": 7.594105433283294e-06} +{"train_loss": 0.014533722773194313, "grad_norm": 0.10490548610687256, "global_step": 43580, "epoch": 411, "lr": 7.592535281825635e-06} +{"train_loss": 0.015489501878619194, "grad_norm": 0.12546499073505402, "global_step": 43581, "epoch": 411, "lr": 7.590965279369377e-06} +{"train_loss": 0.014527116902172565, "grad_norm": 0.06866966933012009, "global_step": 43582, "epoch": 411, "lr": 7.589395425920015e-06} +{"train_loss": 0.013287501409649849, "grad_norm": 0.10959994792938232, "global_step": 43583, "epoch": 411, "lr": 7.587825721483082e-06} +{"train_loss": 0.015037976205348969, "grad_norm": 0.05733387544751167, "global_step": 43584, "epoch": 411, "lr": 7.5862561660640765e-06} +{"train_loss": 0.013741850852966309, "grad_norm": 0.09668970853090286, "global_step": 43585, "epoch": 411, "lr": 7.58468675966853e-06} +{"train_loss": 0.014893028885126114, "grad_norm": 0.09560177475214005, "global_step": 43586, "epoch": 411, "lr": 7.583117502301934e-06} +{"train_loss": 0.017391299828886986, "grad_norm": 0.11671557277441025, "global_step": 43587, "epoch": 411, "lr": 7.581548393969823e-06} +{"train_loss": 0.016215883195400238, "grad_norm": 0.0803011804819107, "global_step": 43588, "epoch": 411, "lr": 7.5799794346777144e-06} +{"train_loss": 0.014968784525990486, "grad_norm": 0.11158586293458939, "global_step": 43589, "epoch": 411, "lr": 7.578410624431104e-06} +{"train_loss": 0.014114242047071457, "grad_norm": 0.059628553688526154, "global_step": 43590, "epoch": 411, "lr": 7.576841963235504e-06} +{"train_loss": 0.01352726574987173, "grad_norm": 0.07269154489040375, "global_step": 43591, "epoch": 411, "lr": 7.575273451096432e-06} +{"train_loss": 0.012796706520020962, "grad_norm": 0.072522833943367, "global_step": 43592, "epoch": 411, "lr": 7.573705088019406e-06} +{"train_loss": 0.01373792439699173, "grad_norm": 0.09053921699523926, "global_step": 43593, "epoch": 411, "lr": 7.572136874009922e-06} +{"train_loss": 0.014801252633333206, "grad_norm": 0.08491161465644836, "global_step": 43594, "epoch": 411, "lr": 7.570568809073508e-06} +{"train_loss": 0.014109546318650246, "grad_norm": 0.10043691843748093, "global_step": 43595, "epoch": 411, "lr": 7.56900089321565e-06} +{"train_loss": 0.014819000847637653, "grad_norm": 0.07094473391771317, "global_step": 43596, "epoch": 411, "lr": 7.567433126441886e-06} +{"train_loss": 0.016814973205327988, "grad_norm": 0.08935943245887756, "global_step": 43597, "epoch": 411, "lr": 7.565865508757691e-06} +{"train_loss": 0.01690378040075302, "grad_norm": 0.0681491568684578, "global_step": 43598, "epoch": 411, "lr": 7.564298040168594e-06} +{"train_loss": 0.014040665701031685, "grad_norm": 0.08712112158536911, "global_step": 43599, "epoch": 411, "lr": 7.562730720680112e-06} +{"train_loss": 0.016774870455265045, "grad_norm": 0.1201123520731926, "global_step": 43600, "epoch": 411, "lr": 7.5611635502977345e-06} +{"train_loss": 0.016932763159275055, "grad_norm": 0.12930917739868164, "global_step": 43601, "epoch": 411, "lr": 7.559596529026964e-06} +{"train_loss": 0.016587162390351295, "grad_norm": 0.07041162252426147, "global_step": 43602, "epoch": 411, "lr": 7.558029656873328e-06} +{"train_loss": 0.014443155378103256, "grad_norm": 0.05788063257932663, "global_step": 43603, "epoch": 411, "lr": 7.556462933842301e-06} +{"train_loss": 0.014635298401117325, "grad_norm": 0.07792891561985016, "global_step": 43604, "epoch": 411, "lr": 7.5548963599394064e-06} +{"train_loss": 0.014225969091057777, "grad_norm": 0.07865828275680542, "global_step": 43605, "epoch": 411, "lr": 7.553329935170161e-06} +{"train_loss": 0.014585950411856174, "grad_norm": 0.05754111334681511, "global_step": 43606, "epoch": 411, "lr": 7.551763659540056e-06} +{"train_loss": 0.014767294749617577, "grad_norm": 0.06019698455929756, "global_step": 43607, "epoch": 411, "lr": 7.5501975330545815e-06} +{"train_loss": 0.01660824753344059, "grad_norm": 0.08233004808425903, "global_step": 43608, "epoch": 411, "lr": 7.548631555719249e-06} +{"train_loss": 0.014170557260513306, "grad_norm": 0.08696496486663818, "global_step": 43609, "epoch": 411, "lr": 7.547065727539576e-06} +{"train_loss": 0.012928321957588196, "grad_norm": 0.0683140903711319, "global_step": 43610, "epoch": 411, "lr": 7.545500048521037e-06} +{"train_loss": 0.013395369984209538, "grad_norm": 0.06882026791572571, "global_step": 43611, "epoch": 411, "lr": 7.543934518669154e-06} +{"train_loss": 0.01638861559331417, "grad_norm": 0.09744306653738022, "global_step": 43612, "epoch": 411, "lr": 7.54236913798943e-06} +{"train_loss": 0.015101388096809387, "grad_norm": 0.08490829914808273, "global_step": 43613, "epoch": 411, "lr": 7.540803906487353e-06} +{"train_loss": 0.013870660215616226, "grad_norm": 0.0830831527709961, "global_step": 43614, "epoch": 411, "lr": 7.539238824168421e-06} +{"train_loss": 0.015288450755178928, "grad_norm": 0.1376274973154068, "global_step": 43615, "epoch": 411, "lr": 7.537673891038133e-06} +{"train_loss": 0.01403075736016035, "grad_norm": 0.07554111629724503, "global_step": 43616, "epoch": 411, "lr": 7.536109107102007e-06} +{"train_loss": 0.015205495059490204, "grad_norm": 0.0605892539024353, "global_step": 43617, "epoch": 411, "lr": 7.534544472365518e-06} +{"train_loss": 0.013695931993424892, "grad_norm": 0.10672112554311752, "global_step": 43618, "epoch": 411, "lr": 7.532979986834176e-06} +{"train_loss": 0.016900327056646347, "grad_norm": 0.07850824296474457, "global_step": 43619, "epoch": 411, "lr": 7.531415650513479e-06} +{"train_loss": 0.01674477942287922, "grad_norm": 0.05927363038063049, "global_step": 43620, "epoch": 411, "lr": 7.529851463408905e-06} +{"train_loss": 0.015720630064606667, "grad_norm": 0.09834660589694977, "global_step": 43621, "epoch": 411, "lr": 7.528287425525965e-06} +{"train_loss": 0.014929287135601044, "grad_norm": 0.0866638645529747, "global_step": 43622, "epoch": 411, "lr": 7.526723536870161e-06} +{"train_loss": 0.013910401612520218, "grad_norm": 0.07050430774688721, "global_step": 43623, "epoch": 411, "lr": 7.5251597974469676e-06} +{"train_loss": 0.015055729076266289, "grad_norm": 0.08189259469509125, "global_step": 43624, "epoch": 411, "lr": 7.523596207261907e-06} +{"train_loss": 0.015115451067686081, "grad_norm": 0.06269990652799606, "global_step": 43625, "epoch": 411, "lr": 7.522032766320442e-06} +{"train_loss": 0.015498997643589973, "grad_norm": 0.06869063526391983, "global_step": 43626, "epoch": 411, "lr": 7.520469474628095e-06} +{"train_loss": 0.01660931296646595, "grad_norm": 0.08313099294900894, "global_step": 43627, "epoch": 411, "lr": 7.518906332190329e-06} +{"train_loss": 0.014147994108498096, "grad_norm": 0.15477734804153442, "global_step": 43628, "epoch": 411, "lr": 7.517343339012656e-06} +{"train_loss": 0.014951987192034721, "grad_norm": 0.0994453951716423, "global_step": 43629, "epoch": 411, "lr": 7.515780495100572e-06} +{"train_loss": 0.015189931727945805, "grad_norm": 0.1347746104001999, "global_step": 43630, "epoch": 411, "lr": 7.514217800459555e-06} +{"train_loss": 0.01267993450164795, "grad_norm": 0.07723399996757507, "global_step": 43631, "epoch": 411, "lr": 7.5126552550950965e-06} +{"train_loss": 0.01503149513155222, "grad_norm": 0.1830606460571289, "global_step": 43632, "epoch": 411, "lr": 7.5110928590126864e-06} +{"train_loss": 0.011568824760615826, "grad_norm": 0.11051177233457565, "global_step": 43633, "epoch": 411, "lr": 7.50953061221783e-06} +{"train_loss": 0.016570501029491425, "grad_norm": 0.05662437900900841, "global_step": 43634, "epoch": 411, "lr": 7.507968514715991e-06} +{"train_loss": 0.0162302665412426, "grad_norm": 0.053651582449674606, "global_step": 43635, "epoch": 411, "lr": 7.5064065665126745e-06} +{"train_loss": 0.01469343900680542, "grad_norm": 0.08367778360843658, "global_step": 43636, "epoch": 411, "lr": 7.504844767613389e-06} +{"train_loss": 0.017215508967638016, "grad_norm": 0.08476068824529648, "global_step": 43637, "epoch": 411, "lr": 7.503283118023574e-06} +{"train_loss": 0.012598901987075806, "grad_norm": 0.0774095207452774, "global_step": 43638, "epoch": 411, "lr": 7.501721617748736e-06} +{"train_loss": 0.015162038616836071, "grad_norm": 0.13663312792778015, "global_step": 43639, "epoch": 411, "lr": 7.500160266794371e-06} +{"train_loss": 0.014567187055945396, "grad_norm": 0.1464904099702835, "global_step": 43640, "epoch": 411, "lr": 7.4985990651659735e-06} +{"train_loss": 0.015597658231854439, "grad_norm": 0.08706113696098328, "global_step": 43641, "epoch": 411, "lr": 7.497038012869001e-06} +{"train_loss": 0.015500066801905632, "grad_norm": 0.056792307645082474, "global_step": 43642, "epoch": 411, "lr": 7.495477109908966e-06} +{"train_loss": 0.015393221750855446, "grad_norm": 0.07487744837999344, "global_step": 43643, "epoch": 411, "lr": 7.493916356291336e-06} +{"train_loss": 0.014195804484188557, "grad_norm": 0.1537080705165863, "global_step": 43644, "epoch": 411, "lr": 7.49235575202159e-06} +{"train_loss": 0.017485622316598892, "grad_norm": 0.07181146740913391, "global_step": 43645, "epoch": 411, "lr": 7.490795297105219e-06} +{"train_loss": 0.014805206097662449, "grad_norm": 0.09882086515426636, "global_step": 43646, "epoch": 411, "lr": 7.489234991547717e-06} +{"train_loss": 0.016069691628217697, "grad_norm": 0.10028437525033951, "global_step": 43647, "epoch": 411, "lr": 7.487674835354558e-06} +{"train_loss": 0.012054730206727982, "grad_norm": 0.17575585842132568, "global_step": 43648, "epoch": 411, "lr": 7.48611482853121e-06} +{"train_loss": 0.01255466602742672, "grad_norm": 0.11021026223897934, "global_step": 43649, "epoch": 411, "lr": 7.484554971083169e-06} +{"train_loss": 0.018495630472898483, "grad_norm": 0.060588203370571136, "global_step": 43650, "epoch": 411, "lr": 7.482995263015919e-06} +{"train_loss": 0.01437353901565075, "grad_norm": 0.12618790566921234, "global_step": 43651, "epoch": 411, "lr": 7.481435704334922e-06} +{"train_loss": 0.015139663591980934, "grad_norm": 0.08664621412754059, "global_step": 43652, "epoch": 411, "lr": 7.479876295045668e-06} +{"train_loss": 0.017163962125778198, "grad_norm": 0.08316273242235184, "global_step": 43653, "epoch": 411, "lr": 7.478317035153654e-06} +{"train_loss": 0.01757078990340233, "grad_norm": 0.09018298983573914, "global_step": 43654, "epoch": 411, "lr": 7.476757924664335e-06} +{"train_loss": 0.014453033916652203, "grad_norm": 0.06289348006248474, "global_step": 43655, "epoch": 411, "lr": 7.475198963583191e-06} +{"train_loss": 0.012531843967735767, "grad_norm": 0.0802493467926979, "global_step": 43656, "epoch": 411, "lr": 7.473640151915701e-06} +{"train_loss": 0.01504618488252163, "grad_norm": 0.06704241782426834, "global_step": 43657, "epoch": 411, "lr": 7.4720814896673545e-06} +{"train_loss": 0.014728291891515255, "grad_norm": 0.07447796314954758, "global_step": 43658, "epoch": 411, "lr": 7.470522976843608e-06} +{"train_loss": 0.014665941707789898, "grad_norm": 0.09373299777507782, "global_step": 43659, "epoch": 411, "lr": 7.468964613449964e-06} +{"train_loss": 0.01506084855645895, "grad_norm": 0.10167639702558517, "global_step": 43660, "epoch": 411, "lr": 7.467406399491878e-06} +{"train_loss": 0.01703646034002304, "grad_norm": 0.0751003548502922, "global_step": 43661, "epoch": 411, "lr": 7.465848334974818e-06} +{"train_loss": 0.016351880505681038, "grad_norm": 0.17457839846611023, "global_step": 43662, "epoch": 411, "lr": 7.4642904199042686e-06} +{"train_loss": 0.016289640218019485, "grad_norm": 0.10610167682170868, "global_step": 43663, "epoch": 411, "lr": 7.46273265428572e-06} +{"train_loss": 0.014131821691989899, "grad_norm": 0.10315500944852829, "global_step": 43664, "epoch": 411, "lr": 7.461175038124612e-06} +{"train_loss": 0.017957231029868126, "grad_norm": 0.09626130759716034, "global_step": 43665, "epoch": 411, "lr": 7.459617571426452e-06} +{"train_loss": 0.011307256296277046, "grad_norm": 0.07324361056089401, "global_step": 43666, "epoch": 411, "lr": 7.458060254196686e-06} +{"train_loss": 0.017298366874456406, "grad_norm": 0.09449909627437592, "global_step": 43667, "epoch": 411, "lr": 7.456503086440803e-06} +{"train_loss": 0.0134896170347929, "grad_norm": 0.07264109700918198, "global_step": 43668, "epoch": 411, "lr": 7.4549460681642535e-06} +{"train_loss": 0.01705874130129814, "grad_norm": 0.07285121828317642, "global_step": 43669, "epoch": 411, "lr": 7.453389199372523e-06} +{"train_loss": 0.014664757996797562, "grad_norm": 0.09743909537792206, "global_step": 43670, "epoch": 411, "lr": 7.451832480071091e-06} +{"train_loss": 0.014981578956445996, "grad_norm": 0.0881020650267601, "global_step": 43671, "epoch": 411, "lr": 7.450275910265414e-06, "val_loss": 0.019385062158107758} +{"train_loss": 0.018347734585404396, "grad_norm": 0.06793626397848129, "global_step": 43672, "epoch": 412, "lr": 7.448719489960959e-06} +{"train_loss": 0.012292170897126198, "grad_norm": 0.08092931658029556, "global_step": 43673, "epoch": 412, "lr": 7.447163219163189e-06} +{"train_loss": 0.01457283180207014, "grad_norm": 0.18467901647090912, "global_step": 43674, "epoch": 412, "lr": 7.445607097877599e-06} +{"train_loss": 0.01654820889234543, "grad_norm": 0.09553217887878418, "global_step": 43675, "epoch": 412, "lr": 7.444051126109624e-06} +{"train_loss": 0.018441610038280487, "grad_norm": 0.07754874974489212, "global_step": 43676, "epoch": 412, "lr": 7.442495303864749e-06} +{"train_loss": 0.01414601132273674, "grad_norm": 0.06453695893287659, "global_step": 43677, "epoch": 412, "lr": 7.440939631148463e-06} +{"train_loss": 0.014602357521653175, "grad_norm": 0.09182349592447281, "global_step": 43678, "epoch": 412, "lr": 7.4393841079661795e-06} +{"train_loss": 0.015004528686404228, "grad_norm": 0.058357350528240204, "global_step": 43679, "epoch": 412, "lr": 7.437828734323393e-06} +{"train_loss": 0.015743358060717583, "grad_norm": 0.1041826456785202, "global_step": 43680, "epoch": 412, "lr": 7.436273510225578e-06} +{"train_loss": 0.016342468559741974, "grad_norm": 0.0646611675620079, "global_step": 43681, "epoch": 412, "lr": 7.434718435678179e-06} +{"train_loss": 0.016880495473742485, "grad_norm": 0.07359539717435837, "global_step": 43682, "epoch": 412, "lr": 7.433163510686664e-06} +{"train_loss": 0.01446120161563158, "grad_norm": 0.0985264927148819, "global_step": 43683, "epoch": 412, "lr": 7.431608735256512e-06} +{"train_loss": 0.018090637400746346, "grad_norm": 0.09327655285596848, "global_step": 43684, "epoch": 412, "lr": 7.4300541093931754e-06} +{"train_loss": 0.01356352400034666, "grad_norm": 0.08952672779560089, "global_step": 43685, "epoch": 412, "lr": 7.428499633102104e-06} +{"train_loss": 0.012704722583293915, "grad_norm": 0.07833719253540039, "global_step": 43686, "epoch": 412, "lr": 7.426945306388772e-06} +{"train_loss": 0.016848964616656303, "grad_norm": 0.07810042798519135, "global_step": 43687, "epoch": 412, "lr": 7.425391129258646e-06} +{"train_loss": 0.01603185385465622, "grad_norm": 0.13268744945526123, "global_step": 43688, "epoch": 412, "lr": 7.423837101717174e-06} +{"train_loss": 0.013713421300053596, "grad_norm": 0.09947271645069122, "global_step": 43689, "epoch": 412, "lr": 7.4222832237698326e-06} +{"train_loss": 0.014306709170341492, "grad_norm": 0.09188118577003479, "global_step": 43690, "epoch": 412, "lr": 7.420729495422063e-06} +{"train_loss": 0.015704747289419174, "grad_norm": 0.06738854199647903, "global_step": 43691, "epoch": 412, "lr": 7.4191759166793386e-06} +{"train_loss": 0.017065836116671562, "grad_norm": 0.08919418603181839, "global_step": 43692, "epoch": 412, "lr": 7.417622487547104e-06} +{"train_loss": 0.015000810846686363, "grad_norm": 0.10674162209033966, "global_step": 43693, "epoch": 412, "lr": 7.416069208030829e-06} +{"train_loss": 0.013874182477593422, "grad_norm": 0.10732848942279816, "global_step": 43694, "epoch": 412, "lr": 7.414516078135974e-06} +{"train_loss": 0.014404884539544582, "grad_norm": 0.05943773314356804, "global_step": 43695, "epoch": 412, "lr": 7.412963097867992e-06} +{"train_loss": 0.014307470060884953, "grad_norm": 0.08514440059661865, "global_step": 43696, "epoch": 412, "lr": 7.411410267232322e-06} +{"train_loss": 0.013479562476277351, "grad_norm": 0.06797096878290176, "global_step": 43697, "epoch": 412, "lr": 7.409857586234437e-06} +{"train_loss": 0.013958773575723171, "grad_norm": 0.09078384935855865, "global_step": 43698, "epoch": 412, "lr": 7.408305054879799e-06} +{"train_loss": 0.016443327069282532, "grad_norm": 0.07612911611795425, "global_step": 43699, "epoch": 412, "lr": 7.40675267317385e-06} +{"train_loss": 0.013988789170980453, "grad_norm": 0.12359178811311722, "global_step": 43700, "epoch": 412, "lr": 7.405200441122057e-06} +{"train_loss": 0.013959887437522411, "grad_norm": 0.07305203378200531, "global_step": 43701, "epoch": 412, "lr": 7.403648358729865e-06} +{"train_loss": 0.014887653291225433, "grad_norm": 0.05913097783923149, "global_step": 43702, "epoch": 412, "lr": 7.402096426002714e-06} +{"train_loss": 0.014658395200967789, "grad_norm": 0.06469473242759705, "global_step": 43703, "epoch": 412, "lr": 7.400544642946072e-06} +{"train_loss": 0.014113513752818108, "grad_norm": 0.10621310025453568, "global_step": 43704, "epoch": 412, "lr": 7.398993009565408e-06} +{"train_loss": 0.013215608894824982, "grad_norm": 0.1025020107626915, "global_step": 43705, "epoch": 412, "lr": 7.397441525866139e-06} +{"train_loss": 0.01327992882579565, "grad_norm": 0.07199008762836456, "global_step": 43706, "epoch": 412, "lr": 7.3958901918537434e-06} +{"train_loss": 0.01601634919643402, "grad_norm": 0.08554007112979889, "global_step": 43707, "epoch": 412, "lr": 7.394339007533651e-06} +{"train_loss": 0.013753334060311317, "grad_norm": 0.0993923544883728, "global_step": 43708, "epoch": 412, "lr": 7.392787972911336e-06} +{"train_loss": 0.014854982495307922, "grad_norm": 0.05003915727138519, "global_step": 43709, "epoch": 412, "lr": 7.3912370879922195e-06} +{"train_loss": 0.013150186277925968, "grad_norm": 0.08349350839853287, "global_step": 43710, "epoch": 412, "lr": 7.389686352781772e-06} +{"train_loss": 0.012578843161463737, "grad_norm": 0.056827738881111145, "global_step": 43711, "epoch": 412, "lr": 7.388135767285442e-06} +{"train_loss": 0.013528338633477688, "grad_norm": 0.06160569190979004, "global_step": 43712, "epoch": 412, "lr": 7.386585331508672e-06} +{"train_loss": 0.014940162189304829, "grad_norm": 0.06175344064831734, "global_step": 43713, "epoch": 412, "lr": 7.385035045456901e-06} +{"train_loss": 0.015560541301965714, "grad_norm": 0.08666741847991943, "global_step": 43714, "epoch": 412, "lr": 7.383484909135579e-06} +{"train_loss": 0.015595175325870514, "grad_norm": 0.10399695485830307, "global_step": 43715, "epoch": 412, "lr": 7.3819349225501715e-06} +{"train_loss": 0.014467772096395493, "grad_norm": 0.10532012581825256, "global_step": 43716, "epoch": 412, "lr": 7.380385085706098e-06} +{"train_loss": 0.015031266026198864, "grad_norm": 0.05968913808465004, "global_step": 43717, "epoch": 412, "lr": 7.378835398608819e-06} +{"train_loss": 0.0143070537596941, "grad_norm": 0.06299299746751785, "global_step": 43718, "epoch": 412, "lr": 7.3772858612638e-06} +{"train_loss": 0.014618187211453915, "grad_norm": 0.055465515702962875, "global_step": 43719, "epoch": 412, "lr": 7.375736473676442e-06} +{"train_loss": 0.014130153693258762, "grad_norm": 0.13783814013004303, "global_step": 43720, "epoch": 412, "lr": 7.374187235852203e-06} +{"train_loss": 0.014905430376529694, "grad_norm": 0.09869915246963501, "global_step": 43721, "epoch": 412, "lr": 7.3726381477965455e-06} +{"train_loss": 0.015507306903600693, "grad_norm": 0.12910108268260956, "global_step": 43722, "epoch": 412, "lr": 7.371089209514892e-06} +{"train_loss": 0.019087255001068115, "grad_norm": 0.06513067334890366, "global_step": 43723, "epoch": 412, "lr": 7.369540421012688e-06} +{"train_loss": 0.015340789221227169, "grad_norm": 0.13234856724739075, "global_step": 43724, "epoch": 412, "lr": 7.367991782295391e-06} +{"train_loss": 0.01320702489465475, "grad_norm": 0.05644197016954422, "global_step": 43725, "epoch": 412, "lr": 7.36644329336843e-06} +{"train_loss": 0.013168707489967346, "grad_norm": 0.09051414579153061, "global_step": 43726, "epoch": 412, "lr": 7.364894954237239e-06} +{"train_loss": 0.013200394809246063, "grad_norm": 0.0655297115445137, "global_step": 43727, "epoch": 412, "lr": 7.363346764907264e-06} +{"train_loss": 0.015139076858758926, "grad_norm": 0.13321848213672638, "global_step": 43728, "epoch": 412, "lr": 7.361798725383956e-06} +{"train_loss": 0.015385557897388935, "grad_norm": 0.10042529553174973, "global_step": 43729, "epoch": 412, "lr": 7.360250835672733e-06} +{"train_loss": 0.015438899397850037, "grad_norm": 0.06197469308972359, "global_step": 43730, "epoch": 412, "lr": 7.358703095779062e-06} +{"train_loss": 0.016790183261036873, "grad_norm": 0.07969443500041962, "global_step": 43731, "epoch": 412, "lr": 7.357155505708346e-06} +{"train_loss": 0.017230724915862083, "grad_norm": 0.1690770983695984, "global_step": 43732, "epoch": 412, "lr": 7.355608065466057e-06} +{"train_loss": 0.013098448514938354, "grad_norm": 0.08603234589099884, "global_step": 43733, "epoch": 412, "lr": 7.3540607750575975e-06} +{"train_loss": 0.014836655929684639, "grad_norm": 0.09487131237983704, "global_step": 43734, "epoch": 412, "lr": 7.3525136344884284e-06} +{"train_loss": 0.01599625125527382, "grad_norm": 0.056053560227155685, "global_step": 43735, "epoch": 412, "lr": 7.3509666437639905e-06} +{"train_loss": 0.015524109825491905, "grad_norm": 0.11486561596393585, "global_step": 43736, "epoch": 412, "lr": 7.349419802889706e-06} +{"train_loss": 0.014348152093589306, "grad_norm": 0.06593631207942963, "global_step": 43737, "epoch": 412, "lr": 7.3478731118710006e-06} +{"train_loss": 0.015648990869522095, "grad_norm": 0.0761457160115242, "global_step": 43738, "epoch": 412, "lr": 7.34632657071333e-06} +{"train_loss": 0.01602766290307045, "grad_norm": 0.0827178880572319, "global_step": 43739, "epoch": 412, "lr": 7.344780179422112e-06} +{"train_loss": 0.012424327433109283, "grad_norm": 0.0969398096203804, "global_step": 43740, "epoch": 412, "lr": 7.34323393800278e-06} +{"train_loss": 0.017617758363485336, "grad_norm": 0.07156992703676224, "global_step": 43741, "epoch": 412, "lr": 7.341687846460776e-06} +{"train_loss": 0.015699386596679688, "grad_norm": 0.07362649589776993, "global_step": 43742, "epoch": 412, "lr": 7.340141904801556e-06} +{"train_loss": 0.011669641360640526, "grad_norm": 0.08753334730863571, "global_step": 43743, "epoch": 412, "lr": 7.338596113030499e-06} +{"train_loss": 0.014568057842552662, "grad_norm": 0.09424831718206406, "global_step": 43744, "epoch": 412, "lr": 7.337050471153062e-06} +{"train_loss": 0.015474364161491394, "grad_norm": 0.09904172271490097, "global_step": 43745, "epoch": 412, "lr": 7.335504979174684e-06} +{"train_loss": 0.013822002336382866, "grad_norm": 0.06363719701766968, "global_step": 43746, "epoch": 412, "lr": 7.333959637100785e-06} +{"train_loss": 0.015535340644419193, "grad_norm": 0.13332684338092804, "global_step": 43747, "epoch": 412, "lr": 7.332414444936797e-06} +{"train_loss": 0.015591881237924099, "grad_norm": 0.09143196046352386, "global_step": 43748, "epoch": 412, "lr": 7.330869402688145e-06} +{"train_loss": 0.016329709440469742, "grad_norm": 0.1454833596944809, "global_step": 43749, "epoch": 412, "lr": 7.329324510360269e-06} +{"train_loss": 0.017898641526699066, "grad_norm": 0.08122143149375916, "global_step": 43750, "epoch": 412, "lr": 7.327779767958582e-06} +{"train_loss": 0.015938621014356613, "grad_norm": 0.07988240569829941, "global_step": 43751, "epoch": 412, "lr": 7.326235175488516e-06} +{"train_loss": 0.017001919448375702, "grad_norm": 0.14351454377174377, "global_step": 43752, "epoch": 412, "lr": 7.3246907329555194e-06} +{"train_loss": 0.015505409799516201, "grad_norm": 0.06561624258756638, "global_step": 43753, "epoch": 412, "lr": 7.323146440364992e-06} +{"train_loss": 0.016929758712649345, "grad_norm": 0.06692223995923996, "global_step": 43754, "epoch": 412, "lr": 7.321602297722363e-06} +{"train_loss": 0.013561483472585678, "grad_norm": 0.066193126142025, "global_step": 43755, "epoch": 412, "lr": 7.320058305033057e-06} +{"train_loss": 0.013527686707675457, "grad_norm": 0.10121214389801025, "global_step": 43756, "epoch": 412, "lr": 7.3185144623025225e-06} +{"train_loss": 0.014017445966601372, "grad_norm": 0.05767899751663208, "global_step": 43757, "epoch": 412, "lr": 7.316970769536152e-06} +{"train_loss": 0.014603884890675545, "grad_norm": 0.09827642142772675, "global_step": 43758, "epoch": 412, "lr": 7.315427226739385e-06} +{"train_loss": 0.015465580858290195, "grad_norm": 0.07394669950008392, "global_step": 43759, "epoch": 412, "lr": 7.3138838339176675e-06} +{"train_loss": 0.01470721885561943, "grad_norm": 0.0551958866417408, "global_step": 43760, "epoch": 412, "lr": 7.312340591076372e-06} +{"train_loss": 0.01623992621898651, "grad_norm": 0.154478058218956, "global_step": 43761, "epoch": 412, "lr": 7.310797498220951e-06} +{"train_loss": 0.016178904101252556, "grad_norm": 0.1781047284603119, "global_step": 43762, "epoch": 412, "lr": 7.309254555356831e-06} +{"train_loss": 0.013120247051119804, "grad_norm": 0.07030762732028961, "global_step": 43763, "epoch": 412, "lr": 7.307711762489416e-06} +{"train_loss": 0.0161065012216568, "grad_norm": 0.10375092178583145, "global_step": 43764, "epoch": 412, "lr": 7.3061691196241345e-06} +{"train_loss": 0.014000178314745426, "grad_norm": 0.07919999212026596, "global_step": 43765, "epoch": 412, "lr": 7.30462662676642e-06} +{"train_loss": 0.013162334449589252, "grad_norm": 0.07937651127576828, "global_step": 43766, "epoch": 412, "lr": 7.303084283921674e-06} +{"train_loss": 0.013012446463108063, "grad_norm": 0.08638337254524231, "global_step": 43767, "epoch": 412, "lr": 7.301542091095315e-06} +{"train_loss": 0.014002839103341103, "grad_norm": 0.0733746886253357, "global_step": 43768, "epoch": 412, "lr": 7.300000048292771e-06} +{"train_loss": 0.017366666346788406, "grad_norm": 0.06978556513786316, "global_step": 43769, "epoch": 412, "lr": 7.298458155519466e-06} +{"train_loss": 0.014785028994083405, "grad_norm": 0.060288894921541214, "global_step": 43770, "epoch": 412, "lr": 7.296916412780797e-06} +{"train_loss": 0.014412185177206993, "grad_norm": 0.07666079699993134, "global_step": 43771, "epoch": 412, "lr": 7.295374820082201e-06} +{"train_loss": 0.014962181448936462, "grad_norm": 0.0706486850976944, "global_step": 43772, "epoch": 412, "lr": 7.293833377429077e-06} +{"train_loss": 0.013881378807127476, "grad_norm": 0.07107508927583694, "global_step": 43773, "epoch": 412, "lr": 7.292292084826863e-06} +{"train_loss": 0.01648428663611412, "grad_norm": 0.07969406992197037, "global_step": 43774, "epoch": 412, "lr": 7.29075094228095e-06} +{"train_loss": 0.01698564551770687, "grad_norm": 0.06173235923051834, "global_step": 43775, "epoch": 412, "lr": 7.289209949796761e-06} +{"train_loss": 0.01372747402638197, "grad_norm": 0.05301232635974884, "global_step": 43776, "epoch": 412, "lr": 7.2876691073797295e-06} +{"train_loss": 0.014985798251867856, "grad_norm": 0.11468978971242905, "global_step": 43777, "epoch": 412, "lr": 7.286128415035248e-06, "val_loss": 0.019494930282235146} +{"train_loss": 0.015319263562560081, "grad_norm": 0.07500671595335007, "global_step": 43778, "epoch": 413, "lr": 7.284587872768728e-06} +{"train_loss": 0.015097163617610931, "grad_norm": 0.06818258762359619, "global_step": 43779, "epoch": 413, "lr": 7.283047480585603e-06} +{"train_loss": 0.015015261247754097, "grad_norm": 0.05732487514615059, "global_step": 43780, "epoch": 413, "lr": 7.281507238491259e-06} +{"train_loss": 0.014192670583724976, "grad_norm": 0.07052213698625565, "global_step": 43781, "epoch": 413, "lr": 7.279967146491118e-06} +{"train_loss": 0.014397243037819862, "grad_norm": 0.07917487621307373, "global_step": 43782, "epoch": 413, "lr": 7.27842720459061e-06} +{"train_loss": 0.015896901488304138, "grad_norm": 0.055588774383068085, "global_step": 43783, "epoch": 413, "lr": 7.2768874127951245e-06} +{"train_loss": 0.014011898078024387, "grad_norm": 0.06789323687553406, "global_step": 43784, "epoch": 413, "lr": 7.27534777111007e-06} +{"train_loss": 0.013951405882835388, "grad_norm": 0.0645248293876648, "global_step": 43785, "epoch": 413, "lr": 7.273808279540862e-06} +{"train_loss": 0.017042046412825584, "grad_norm": 0.05017274245619774, "global_step": 43786, "epoch": 413, "lr": 7.272268938092919e-06} +{"train_loss": 0.01642531156539917, "grad_norm": 0.10168448835611343, "global_step": 43787, "epoch": 413, "lr": 7.270729746771632e-06} +{"train_loss": 0.013601113110780716, "grad_norm": 0.061735089868307114, "global_step": 43788, "epoch": 413, "lr": 7.2691907055824184e-06} +{"train_loss": 0.013096615672111511, "grad_norm": 0.0775425136089325, "global_step": 43789, "epoch": 413, "lr": 7.267651814530696e-06} +{"train_loss": 0.013712748885154724, "grad_norm": 0.07568222284317017, "global_step": 43790, "epoch": 413, "lr": 7.266113073621866e-06} +{"train_loss": 0.013421326875686646, "grad_norm": 0.07067347317934036, "global_step": 43791, "epoch": 413, "lr": 7.264574482861314e-06} +{"train_loss": 0.014230343513190746, "grad_norm": 0.06500758230686188, "global_step": 43792, "epoch": 413, "lr": 7.263036042254462e-06} +{"train_loss": 0.015273520722985268, "grad_norm": 0.06768098473548889, "global_step": 43793, "epoch": 413, "lr": 7.261497751806728e-06} +{"train_loss": 0.014926129952073097, "grad_norm": 0.08600263297557831, "global_step": 43794, "epoch": 413, "lr": 7.259959611523493e-06} +{"train_loss": 0.015328596346080303, "grad_norm": 0.1295676827430725, "global_step": 43795, "epoch": 413, "lr": 7.258421621410183e-06} +{"train_loss": 0.014755605719983578, "grad_norm": 0.06413072347640991, "global_step": 43796, "epoch": 413, "lr": 7.256883781472179e-06} +{"train_loss": 0.014732125215232372, "grad_norm": 0.07818009704351425, "global_step": 43797, "epoch": 413, "lr": 7.2553460917149155e-06} +{"train_loss": 0.01408575288951397, "grad_norm": 0.06395872682332993, "global_step": 43798, "epoch": 413, "lr": 7.25380855214376e-06} +{"train_loss": 0.01564096286892891, "grad_norm": 0.07710635662078857, "global_step": 43799, "epoch": 413, "lr": 7.252271162764129e-06} +{"train_loss": 0.013240965083241463, "grad_norm": 0.07321890443563461, "global_step": 43800, "epoch": 413, "lr": 7.2507339235814545e-06} +{"train_loss": 0.01605742610991001, "grad_norm": 0.09331867843866348, "global_step": 43801, "epoch": 413, "lr": 7.249196834601085e-06} +{"train_loss": 0.012773998081684113, "grad_norm": 0.10075530409812927, "global_step": 43802, "epoch": 413, "lr": 7.247659895828446e-06} +{"train_loss": 0.01569191738963127, "grad_norm": 0.07136541604995728, "global_step": 43803, "epoch": 413, "lr": 7.246123107268948e-06} +{"train_loss": 0.015357676893472672, "grad_norm": 0.060482654720544815, "global_step": 43804, "epoch": 413, "lr": 7.244586468927967e-06} +{"train_loss": 0.016202399507164955, "grad_norm": 0.10743444412946701, "global_step": 43805, "epoch": 413, "lr": 7.2430499808109176e-06} +{"train_loss": 0.016897672787308693, "grad_norm": 0.06518284976482391, "global_step": 43806, "epoch": 413, "lr": 7.2415136429232085e-06} +{"train_loss": 0.01693592593073845, "grad_norm": 0.08987288177013397, "global_step": 43807, "epoch": 413, "lr": 7.239977455270225e-06} +{"train_loss": 0.012018884532153606, "grad_norm": 0.12963664531707764, "global_step": 43808, "epoch": 413, "lr": 7.2384414178573555e-06} +{"train_loss": 0.01575460657477379, "grad_norm": 0.11734753102064133, "global_step": 43809, "epoch": 413, "lr": 7.236905530690003e-06} +{"train_loss": 0.01509657222777605, "grad_norm": 0.10397221148014069, "global_step": 43810, "epoch": 413, "lr": 7.235369793773578e-06} +{"train_loss": 0.015167025849223137, "grad_norm": 0.06815280020236969, "global_step": 43811, "epoch": 413, "lr": 7.233834207113454e-06} +{"train_loss": 0.015188141725957394, "grad_norm": 0.07856200635433197, "global_step": 43812, "epoch": 413, "lr": 7.232298770715051e-06} +{"train_loss": 0.014408967457711697, "grad_norm": 0.05610993504524231, "global_step": 43813, "epoch": 413, "lr": 7.230763484583742e-06} +{"train_loss": 0.015016838908195496, "grad_norm": 0.06945276260375977, "global_step": 43814, "epoch": 413, "lr": 7.2292283487249426e-06} +{"train_loss": 0.015275560319423676, "grad_norm": 0.08666247129440308, "global_step": 43815, "epoch": 413, "lr": 7.227693363144017e-06} +{"train_loss": 0.014437664300203323, "grad_norm": 0.10207894444465637, "global_step": 43816, "epoch": 413, "lr": 7.226158527846377e-06} +{"train_loss": 0.01294146291911602, "grad_norm": 0.09553097933530807, "global_step": 43817, "epoch": 413, "lr": 7.22462384283743e-06} +{"train_loss": 0.01613905094563961, "grad_norm": 0.08208414167165756, "global_step": 43818, "epoch": 413, "lr": 7.223089308122549e-06} +{"train_loss": 0.014809955842792988, "grad_norm": 0.09046212583780289, "global_step": 43819, "epoch": 413, "lr": 7.221554923707119e-06} +{"train_loss": 0.013912444934248924, "grad_norm": 0.10389306396245956, "global_step": 43820, "epoch": 413, "lr": 7.220020689596552e-06} +{"train_loss": 0.015534568578004837, "grad_norm": 0.14054153859615326, "global_step": 43821, "epoch": 413, "lr": 7.218486605796215e-06} +{"train_loss": 0.014477180317044258, "grad_norm": 0.07014372199773788, "global_step": 43822, "epoch": 413, "lr": 7.216952672311511e-06} +{"train_loss": 0.014174322597682476, "grad_norm": 0.10662373900413513, "global_step": 43823, "epoch": 413, "lr": 7.21541888914784e-06} +{"train_loss": 0.016212869435548782, "grad_norm": 0.09210758656263351, "global_step": 43824, "epoch": 413, "lr": 7.213885256310582e-06} +{"train_loss": 0.014249308034777641, "grad_norm": 0.0826648473739624, "global_step": 43825, "epoch": 413, "lr": 7.212351773805115e-06} +{"train_loss": 0.013830289244651794, "grad_norm": 0.0676681250333786, "global_step": 43826, "epoch": 413, "lr": 7.210818441636835e-06} +{"train_loss": 0.01268935389816761, "grad_norm": 0.10838330537080765, "global_step": 43827, "epoch": 413, "lr": 7.209285259811138e-06} +{"train_loss": 0.016320504248142242, "grad_norm": 0.07295338064432144, "global_step": 43828, "epoch": 413, "lr": 7.207752228333392e-06} +{"train_loss": 0.017364272847771645, "grad_norm": 0.057420115917921066, "global_step": 43829, "epoch": 413, "lr": 7.2062193472089975e-06} +{"train_loss": 0.012696714140474796, "grad_norm": 0.07075303047895432, "global_step": 43830, "epoch": 413, "lr": 7.204686616443351e-06} +{"train_loss": 0.014214640483260155, "grad_norm": 0.057578615844249725, "global_step": 43831, "epoch": 413, "lr": 7.20315403604182e-06} +{"train_loss": 0.015141509473323822, "grad_norm": 0.0924714207649231, "global_step": 43832, "epoch": 413, "lr": 7.20162160600979e-06} +{"train_loss": 0.01647140085697174, "grad_norm": 0.07952215522527695, "global_step": 43833, "epoch": 413, "lr": 7.200089326352644e-06} +{"train_loss": 0.015801049768924713, "grad_norm": 0.07980453968048096, "global_step": 43834, "epoch": 413, "lr": 7.198557197075778e-06} +{"train_loss": 0.015678133815526962, "grad_norm": 0.0559234581887722, "global_step": 43835, "epoch": 413, "lr": 7.1970252181845664e-06} +{"train_loss": 0.01571432128548622, "grad_norm": 0.06487390398979187, "global_step": 43836, "epoch": 413, "lr": 7.195493389684399e-06} +{"train_loss": 0.01703101396560669, "grad_norm": 0.06912319362163544, "global_step": 43837, "epoch": 413, "lr": 7.193961711580654e-06} +{"train_loss": 0.015734966844320297, "grad_norm": 0.1324014514684677, "global_step": 43838, "epoch": 413, "lr": 7.192430183878696e-06} +{"train_loss": 0.015035446733236313, "grad_norm": 0.08924902975559235, "global_step": 43839, "epoch": 413, "lr": 7.190898806583929e-06} +{"train_loss": 0.013018018566071987, "grad_norm": 0.08429893851280212, "global_step": 43840, "epoch": 413, "lr": 7.189367579701733e-06} +{"train_loss": 0.017845969647169113, "grad_norm": 0.06477294117212296, "global_step": 43841, "epoch": 413, "lr": 7.187836503237472e-06} +{"train_loss": 0.014398491941392422, "grad_norm": 0.07284563034772873, "global_step": 43842, "epoch": 413, "lr": 7.186305577196545e-06} +{"train_loss": 0.01413794606924057, "grad_norm": 0.06665829569101334, "global_step": 43843, "epoch": 413, "lr": 7.184774801584309e-06} +{"train_loss": 0.01358285453170538, "grad_norm": 0.08709458261728287, "global_step": 43844, "epoch": 413, "lr": 7.183244176406168e-06} +{"train_loss": 0.013896171003580093, "grad_norm": 0.10374192148447037, "global_step": 43845, "epoch": 413, "lr": 7.181713701667475e-06} +{"train_loss": 0.013478009030222893, "grad_norm": 0.07099080085754395, "global_step": 43846, "epoch": 413, "lr": 7.180183377373617e-06} +{"train_loss": 0.014818862080574036, "grad_norm": 0.06058858707547188, "global_step": 43847, "epoch": 413, "lr": 7.1786532035299895e-06} +{"train_loss": 0.01772058941423893, "grad_norm": 0.0845998227596283, "global_step": 43848, "epoch": 413, "lr": 7.177123180141948e-06} +{"train_loss": 0.01546229887753725, "grad_norm": 0.12323079258203506, "global_step": 43849, "epoch": 413, "lr": 7.175593307214862e-06} +{"train_loss": 0.01951705664396286, "grad_norm": 0.0884351059794426, "global_step": 43850, "epoch": 413, "lr": 7.174063584754115e-06} +{"train_loss": 0.014881938695907593, "grad_norm": 0.16673003137111664, "global_step": 43851, "epoch": 413, "lr": 7.172534012765097e-06} +{"train_loss": 0.015259809792041779, "grad_norm": 0.07576088607311249, "global_step": 43852, "epoch": 413, "lr": 7.171004591253155e-06} +{"train_loss": 0.013967781327664852, "grad_norm": 0.089973583817482, "global_step": 43853, "epoch": 413, "lr": 7.1694753202236954e-06} +{"train_loss": 0.01298239454627037, "grad_norm": 0.08310823142528534, "global_step": 43854, "epoch": 413, "lr": 7.167946199682057e-06} +{"train_loss": 0.016262337565422058, "grad_norm": 0.06257439404726028, "global_step": 43855, "epoch": 413, "lr": 7.166417229633638e-06} +{"train_loss": 0.016317766159772873, "grad_norm": 0.10702849179506302, "global_step": 43856, "epoch": 413, "lr": 7.164888410083792e-06} +{"train_loss": 0.015202579088509083, "grad_norm": 0.06568010151386261, "global_step": 43857, "epoch": 413, "lr": 7.163359741037901e-06} +{"train_loss": 0.014445658773183823, "grad_norm": 0.08875660598278046, "global_step": 43858, "epoch": 413, "lr": 7.161831222501342e-06} +{"train_loss": 0.0178513303399086, "grad_norm": 0.09725863486528397, "global_step": 43859, "epoch": 413, "lr": 7.160302854479484e-06} +{"train_loss": 0.016594599932432175, "grad_norm": 0.10852351039648056, "global_step": 43860, "epoch": 413, "lr": 7.1587746369776775e-06} +{"train_loss": 0.01607609912753105, "grad_norm": 0.06591033190488815, "global_step": 43861, "epoch": 413, "lr": 7.157246570001314e-06} +{"train_loss": 0.016475845128297806, "grad_norm": 0.11187034100294113, "global_step": 43862, "epoch": 413, "lr": 7.155718653555748e-06} +{"train_loss": 0.01415335200726986, "grad_norm": 0.09847980737686157, "global_step": 43863, "epoch": 413, "lr": 7.15419088764635e-06} +{"train_loss": 0.016337500885128975, "grad_norm": 0.07537809014320374, "global_step": 43864, "epoch": 413, "lr": 7.1526632722785035e-06} +{"train_loss": 0.01520782895386219, "grad_norm": 0.06665313988924026, "global_step": 43865, "epoch": 413, "lr": 7.1511358074575655e-06} +{"train_loss": 0.013478131033480167, "grad_norm": 0.09538847953081131, "global_step": 43866, "epoch": 413, "lr": 7.149608493188892e-06} +{"train_loss": 0.012574753724038601, "grad_norm": 0.05864657089114189, "global_step": 43867, "epoch": 413, "lr": 7.148081329477857e-06} +{"train_loss": 0.016141945496201515, "grad_norm": 0.07632547616958618, "global_step": 43868, "epoch": 413, "lr": 7.14655431632984e-06} +{"train_loss": 0.012329403311014175, "grad_norm": 0.06720009446144104, "global_step": 43869, "epoch": 413, "lr": 7.145027453750186e-06} +{"train_loss": 0.013882206752896309, "grad_norm": 0.056685976684093475, "global_step": 43870, "epoch": 413, "lr": 7.143500741744269e-06} +{"train_loss": 0.015639595687389374, "grad_norm": 0.08923714607954025, "global_step": 43871, "epoch": 413, "lr": 7.1419741803174564e-06} +{"train_loss": 0.014262612909078598, "grad_norm": 0.08849863708019257, "global_step": 43872, "epoch": 413, "lr": 7.140447769475117e-06} +{"train_loss": 0.015686331316828728, "grad_norm": 0.06494440138339996, "global_step": 43873, "epoch": 413, "lr": 7.1389215092225905e-06} +{"train_loss": 0.01515407208353281, "grad_norm": 0.1023019328713417, "global_step": 43874, "epoch": 413, "lr": 7.13739539956525e-06} +{"train_loss": 0.014587758108973503, "grad_norm": 0.1169414073228836, "global_step": 43875, "epoch": 413, "lr": 7.1358694405084805e-06} +{"train_loss": 0.012839100323617458, "grad_norm": 0.09287744760513306, "global_step": 43876, "epoch": 413, "lr": 7.134343632057605e-06} +{"train_loss": 0.014842703007161617, "grad_norm": 0.06460978835821152, "global_step": 43877, "epoch": 413, "lr": 7.13281797421802e-06} +{"train_loss": 0.015427367761731148, "grad_norm": 0.06989774107933044, "global_step": 43878, "epoch": 413, "lr": 7.131292466995071e-06} +{"train_loss": 0.014887899160385132, "grad_norm": 0.13843940198421478, "global_step": 43879, "epoch": 413, "lr": 7.1297671103941035e-06} +{"train_loss": 0.01580532267689705, "grad_norm": 0.07258367538452148, "global_step": 43880, "epoch": 413, "lr": 7.128241904420491e-06} +{"train_loss": 0.015738146379590034, "grad_norm": 0.0712704062461853, "global_step": 43881, "epoch": 413, "lr": 7.126716849079606e-06} +{"train_loss": 0.013788399286568165, "grad_norm": 0.06170162931084633, "global_step": 43882, "epoch": 413, "lr": 7.12519194437678e-06} +{"train_loss": 0.0149849166584043, "grad_norm": 0.10530305653810501, "global_step": 43883, "epoch": 413, "lr": 7.123667190317396e-06, "val_loss": 0.019488144665956497} +{"train_loss": 0.016107972711324692, "grad_norm": 0.07026665657758713, "global_step": 43884, "epoch": 414, "lr": 7.122142586906783e-06} +{"train_loss": 0.013158692046999931, "grad_norm": 0.061074480414390564, "global_step": 43885, "epoch": 414, "lr": 7.1206181341503305e-06} +{"train_loss": 0.015038739889860153, "grad_norm": 0.07966944575309753, "global_step": 43886, "epoch": 414, "lr": 7.119093832053364e-06} +{"train_loss": 0.014858296141028404, "grad_norm": 0.06194373592734337, "global_step": 43887, "epoch": 414, "lr": 7.117569680621256e-06} +{"train_loss": 0.015033065341413021, "grad_norm": 0.09814750403165817, "global_step": 43888, "epoch": 414, "lr": 7.116045679859368e-06} +{"train_loss": 0.015019230544567108, "grad_norm": 0.059773944318294525, "global_step": 43889, "epoch": 414, "lr": 7.114521829773047e-06} +{"train_loss": 0.0157783105969429, "grad_norm": 0.07843805104494095, "global_step": 43890, "epoch": 414, "lr": 7.112998130367632e-06} +{"train_loss": 0.013463503681123257, "grad_norm": 0.09846721589565277, "global_step": 43891, "epoch": 414, "lr": 7.111474581648491e-06} +{"train_loss": 0.012381169945001602, "grad_norm": 0.09603036940097809, "global_step": 43892, "epoch": 414, "lr": 7.109951183620994e-06} +{"train_loss": 0.01540474034845829, "grad_norm": 0.08317328244447708, "global_step": 43893, "epoch": 414, "lr": 7.108427936290457e-06} +{"train_loss": 0.014469055458903313, "grad_norm": 0.11437114328145981, "global_step": 43894, "epoch": 414, "lr": 7.106904839662254e-06} +{"train_loss": 0.014326557517051697, "grad_norm": 0.1624595820903778, "global_step": 43895, "epoch": 414, "lr": 7.105381893741758e-06} +{"train_loss": 0.013767820782959461, "grad_norm": 0.11387494206428528, "global_step": 43896, "epoch": 414, "lr": 7.1038590985342715e-06} +{"train_loss": 0.01563524827361107, "grad_norm": 0.10505647212266922, "global_step": 43897, "epoch": 414, "lr": 7.102336454045172e-06} +{"train_loss": 0.014326779171824455, "grad_norm": 0.09192267060279846, "global_step": 43898, "epoch": 414, "lr": 7.100813960279801e-06} +{"train_loss": 0.0143548883497715, "grad_norm": 0.14764955639839172, "global_step": 43899, "epoch": 414, "lr": 7.099291617243526e-06} +{"train_loss": 0.01670743152499199, "grad_norm": 0.06849276274442673, "global_step": 43900, "epoch": 414, "lr": 7.097769424941686e-06} +{"train_loss": 0.012468989007174969, "grad_norm": 0.061573706567287445, "global_step": 43901, "epoch": 414, "lr": 7.096247383379617e-06} +{"train_loss": 0.016925515606999397, "grad_norm": 0.07679842412471771, "global_step": 43902, "epoch": 414, "lr": 7.094725492562681e-06} +{"train_loss": 0.013551623560488224, "grad_norm": 0.1042252779006958, "global_step": 43903, "epoch": 414, "lr": 7.093203752496213e-06} +{"train_loss": 0.014917060732841492, "grad_norm": 0.06716486811637878, "global_step": 43904, "epoch": 414, "lr": 7.0916821631855705e-06} +{"train_loss": 0.013536054641008377, "grad_norm": 0.0654953345656395, "global_step": 43905, "epoch": 414, "lr": 7.090160724636102e-06} +{"train_loss": 0.017273787409067154, "grad_norm": 0.18919667601585388, "global_step": 43906, "epoch": 414, "lr": 7.088639436853151e-06} +{"train_loss": 0.013492729514837265, "grad_norm": 0.08775917440652847, "global_step": 43907, "epoch": 414, "lr": 7.08711829984205e-06} +{"train_loss": 0.014450355432927608, "grad_norm": 0.08308064192533493, "global_step": 43908, "epoch": 414, "lr": 7.08559731360815e-06} +{"train_loss": 0.014174015261232853, "grad_norm": 0.06798805296421051, "global_step": 43909, "epoch": 414, "lr": 7.08407647815681e-06} +{"train_loss": 0.015120834112167358, "grad_norm": 0.15220868587493896, "global_step": 43910, "epoch": 414, "lr": 7.082555793493345e-06} +{"train_loss": 0.013241825625300407, "grad_norm": 0.061522915959358215, "global_step": 43911, "epoch": 414, "lr": 7.08103525962312e-06} +{"train_loss": 0.01237561646848917, "grad_norm": 0.06187549978494644, "global_step": 43912, "epoch": 414, "lr": 7.079514876551485e-06} +{"train_loss": 0.014255560003221035, "grad_norm": 0.0752788707613945, "global_step": 43913, "epoch": 414, "lr": 7.07799464428377e-06} +{"train_loss": 0.01496403943747282, "grad_norm": 0.06169624626636505, "global_step": 43914, "epoch": 414, "lr": 7.076474562825297e-06} +{"train_loss": 0.01640189066529274, "grad_norm": 0.06875138729810715, "global_step": 43915, "epoch": 414, "lr": 7.07495463218143e-06} +{"train_loss": 0.013788556680083275, "grad_norm": 0.08892049640417099, "global_step": 43916, "epoch": 414, "lr": 7.073434852357519e-06} +{"train_loss": 0.016482222825288773, "grad_norm": 0.07925165444612503, "global_step": 43917, "epoch": 414, "lr": 7.071915223358877e-06} +{"train_loss": 0.01472650095820427, "grad_norm": 0.10115204006433487, "global_step": 43918, "epoch": 414, "lr": 7.070395745190861e-06} +{"train_loss": 0.016097471117973328, "grad_norm": 0.0767841637134552, "global_step": 43919, "epoch": 414, "lr": 7.068876417858811e-06} +{"train_loss": 0.01612912118434906, "grad_norm": 0.07308351248502731, "global_step": 43920, "epoch": 414, "lr": 7.067357241368045e-06} +{"train_loss": 0.01415271870791912, "grad_norm": 0.0749673992395401, "global_step": 43921, "epoch": 414, "lr": 7.065838215723919e-06} +{"train_loss": 0.01522693783044815, "grad_norm": 0.13508452475070953, "global_step": 43922, "epoch": 414, "lr": 7.064319340931774e-06} +{"train_loss": 0.015061557292938232, "grad_norm": 0.0599081888794899, "global_step": 43923, "epoch": 414, "lr": 7.062800616996928e-06} +{"train_loss": 0.016070187091827393, "grad_norm": 0.11419092118740082, "global_step": 43924, "epoch": 414, "lr": 7.061282043924739e-06} +{"train_loss": 0.014095727354288101, "grad_norm": 0.08116121590137482, "global_step": 43925, "epoch": 414, "lr": 7.0597636217205165e-06} +{"train_loss": 0.015056304633617401, "grad_norm": 0.06618348509073257, "global_step": 43926, "epoch": 414, "lr": 7.058245350389625e-06} +{"train_loss": 0.01574578322470188, "grad_norm": 0.07215192168951035, "global_step": 43927, "epoch": 414, "lr": 7.056727229937371e-06} +{"train_loss": 0.015414606779813766, "grad_norm": 0.08434303849935532, "global_step": 43928, "epoch": 414, "lr": 7.0552092603691056e-06} +{"train_loss": 0.01567060500383377, "grad_norm": 0.08149000257253647, "global_step": 43929, "epoch": 414, "lr": 7.053691441690169e-06} +{"train_loss": 0.014411453157663345, "grad_norm": 0.07771671563386917, "global_step": 43930, "epoch": 414, "lr": 7.0521737739058844e-06} +{"train_loss": 0.013229399919509888, "grad_norm": 0.08377952128648758, "global_step": 43931, "epoch": 414, "lr": 7.050656257021571e-06} +{"train_loss": 0.01563212275505066, "grad_norm": 0.16884690523147583, "global_step": 43932, "epoch": 414, "lr": 7.0491388910425724e-06} +{"train_loss": 0.013588037341833115, "grad_norm": 0.11030333489179611, "global_step": 43933, "epoch": 414, "lr": 7.047621675974231e-06} +{"train_loss": 0.014321817085146904, "grad_norm": 0.04836907982826233, "global_step": 43934, "epoch": 414, "lr": 7.0461046118218575e-06} +{"train_loss": 0.01752743311226368, "grad_norm": 0.10174153000116348, "global_step": 43935, "epoch": 414, "lr": 7.044587698590793e-06} +{"train_loss": 0.016751006245613098, "grad_norm": 0.1206163838505745, "global_step": 43936, "epoch": 414, "lr": 7.0430709362863945e-06} +{"train_loss": 0.014929166994988918, "grad_norm": 0.05911363661289215, "global_step": 43937, "epoch": 414, "lr": 7.0415543249139345e-06} +{"train_loss": 0.014677537605166435, "grad_norm": 0.14598453044891357, "global_step": 43938, "epoch": 414, "lr": 7.0400378644787715e-06} +{"train_loss": 0.013103407807648182, "grad_norm": 0.0900009498000145, "global_step": 43939, "epoch": 414, "lr": 7.038521554986244e-06} +{"train_loss": 0.012328384444117546, "grad_norm": 0.15230585634708405, "global_step": 43940, "epoch": 414, "lr": 7.037005396441654e-06} +{"train_loss": 0.01674194075167179, "grad_norm": 0.17695502936840057, "global_step": 43941, "epoch": 414, "lr": 7.035489388850347e-06} +{"train_loss": 0.014012940227985382, "grad_norm": 0.05495227500796318, "global_step": 43942, "epoch": 414, "lr": 7.033973532217653e-06} +{"train_loss": 0.014060833491384983, "grad_norm": 0.09857578575611115, "global_step": 43943, "epoch": 414, "lr": 7.0324578265488936e-06} +{"train_loss": 0.013503915630280972, "grad_norm": 0.08816271275281906, "global_step": 43944, "epoch": 414, "lr": 7.030942271849383e-06} +{"train_loss": 0.01591549441218376, "grad_norm": 0.08283673971891403, "global_step": 43945, "epoch": 414, "lr": 7.02942686812445e-06} +{"train_loss": 0.015139290131628513, "grad_norm": 0.0838375985622406, "global_step": 43946, "epoch": 414, "lr": 7.027911615379435e-06} +{"train_loss": 0.016257300972938538, "grad_norm": 0.06373430043458939, "global_step": 43947, "epoch": 414, "lr": 7.026396513619648e-06} +{"train_loss": 0.013915935531258583, "grad_norm": 0.06969434022903442, "global_step": 43948, "epoch": 414, "lr": 7.02488156285041e-06} +{"train_loss": 0.01446451898664236, "grad_norm": 0.10867299884557724, "global_step": 43949, "epoch": 414, "lr": 7.023366763077044e-06} +{"train_loss": 0.017012815922498703, "grad_norm": 0.07661568373441696, "global_step": 43950, "epoch": 414, "lr": 7.021852114304894e-06} +{"train_loss": 0.01484520360827446, "grad_norm": 0.08246022462844849, "global_step": 43951, "epoch": 414, "lr": 7.020337616539246e-06} +{"train_loss": 0.013086317107081413, "grad_norm": 0.12686611711978912, "global_step": 43952, "epoch": 414, "lr": 7.018823269785446e-06} +{"train_loss": 0.016515197232365608, "grad_norm": 0.09318700432777405, "global_step": 43953, "epoch": 414, "lr": 7.0173090740488164e-06} +{"train_loss": 0.014569563791155815, "grad_norm": 0.14267823100090027, "global_step": 43954, "epoch": 414, "lr": 7.015795029334677e-06} +{"train_loss": 0.014503752812743187, "grad_norm": 0.060870211571455, "global_step": 43955, "epoch": 414, "lr": 7.0142811356483205e-06} +{"train_loss": 0.014998424798250198, "grad_norm": 0.07383476197719574, "global_step": 43956, "epoch": 414, "lr": 7.0127673929950895e-06} +{"train_loss": 0.015347721055150032, "grad_norm": 0.06692187488079071, "global_step": 43957, "epoch": 414, "lr": 7.011253801380313e-06} +{"train_loss": 0.017712416127324104, "grad_norm": 0.09186007082462311, "global_step": 43958, "epoch": 414, "lr": 7.0097403608092805e-06} +{"train_loss": 0.014628011733293533, "grad_norm": 0.09389892220497131, "global_step": 43959, "epoch": 414, "lr": 7.008227071287338e-06} +{"train_loss": 0.013159843161702156, "grad_norm": 0.06411967426538467, "global_step": 43960, "epoch": 414, "lr": 7.006713932819792e-06} +{"train_loss": 0.014855682849884033, "grad_norm": 0.08088821172714233, "global_step": 43961, "epoch": 414, "lr": 7.005200945411938e-06} +{"train_loss": 0.014614446088671684, "grad_norm": 0.10182426124811172, "global_step": 43962, "epoch": 414, "lr": 7.003688109069112e-06} +{"train_loss": 0.013303546234965324, "grad_norm": 0.11991561949253082, "global_step": 43963, "epoch": 414, "lr": 7.002175423796642e-06} +{"train_loss": 0.017381735146045685, "grad_norm": 0.10517042875289917, "global_step": 43964, "epoch": 414, "lr": 7.000662889599813e-06} +{"train_loss": 0.015568813309073448, "grad_norm": 0.07337582111358643, "global_step": 43965, "epoch": 414, "lr": 6.999150506483965e-06} +{"train_loss": 0.014998428523540497, "grad_norm": 0.06944144517183304, "global_step": 43966, "epoch": 414, "lr": 6.997638274454394e-06} +{"train_loss": 0.014956200495362282, "grad_norm": 0.059361983090639114, "global_step": 43967, "epoch": 414, "lr": 6.996126193516434e-06} +{"train_loss": 0.014832442626357079, "grad_norm": 0.05806795507669449, "global_step": 43968, "epoch": 414, "lr": 6.9946142636753695e-06} +{"train_loss": 0.016307126730680466, "grad_norm": 0.11292708665132523, "global_step": 43969, "epoch": 414, "lr": 6.9931024849365255e-06} +{"train_loss": 0.017671138048171997, "grad_norm": 0.073237843811512, "global_step": 43970, "epoch": 414, "lr": 6.99159085730523e-06} +{"train_loss": 0.014853116124868393, "grad_norm": 0.08570434898138046, "global_step": 43971, "epoch": 414, "lr": 6.990079380786785e-06} +{"train_loss": 0.01486972812563181, "grad_norm": 0.08456079661846161, "global_step": 43972, "epoch": 414, "lr": 6.988568055386479e-06} +{"train_loss": 0.017535898834466934, "grad_norm": 0.09157863259315491, "global_step": 43973, "epoch": 414, "lr": 6.9870568811096435e-06} +{"train_loss": 0.013069840148091316, "grad_norm": 0.07119867950677872, "global_step": 43974, "epoch": 414, "lr": 6.985545857961595e-06} +{"train_loss": 0.01656172052025795, "grad_norm": 0.08713620156049728, "global_step": 43975, "epoch": 414, "lr": 6.984034985947619e-06} +{"train_loss": 0.014923561364412308, "grad_norm": 0.05932294949889183, "global_step": 43976, "epoch": 414, "lr": 6.982524265073037e-06} +{"train_loss": 0.016530439257621765, "grad_norm": 0.10949195176362991, "global_step": 43977, "epoch": 414, "lr": 6.9810136953431805e-06} +{"train_loss": 0.015553539618849754, "grad_norm": 0.07682514190673828, "global_step": 43978, "epoch": 414, "lr": 6.979503276763311e-06} +{"train_loss": 0.016266142949461937, "grad_norm": 0.1221122220158577, "global_step": 43979, "epoch": 414, "lr": 6.977993009338757e-06} +{"train_loss": 0.014764462597668171, "grad_norm": 0.060694821178913116, "global_step": 43980, "epoch": 414, "lr": 6.976482893074837e-06} +{"train_loss": 0.015481855720281601, "grad_norm": 0.06237228214740753, "global_step": 43981, "epoch": 414, "lr": 6.9749729279768304e-06} +{"train_loss": 0.013240678235888481, "grad_norm": 0.10042595863342285, "global_step": 43982, "epoch": 414, "lr": 6.9734631140500665e-06} +{"train_loss": 0.014878423884510994, "grad_norm": 0.0902896597981453, "global_step": 43983, "epoch": 414, "lr": 6.9719534512998456e-06} +{"train_loss": 0.016219697892665863, "grad_norm": 0.06485520303249359, "global_step": 43984, "epoch": 414, "lr": 6.970443939731469e-06} +{"train_loss": 0.015586317516863346, "grad_norm": 0.07549566775560379, "global_step": 43985, "epoch": 414, "lr": 6.968934579350233e-06} +{"train_loss": 0.016046661883592606, "grad_norm": 0.10584766417741776, "global_step": 43986, "epoch": 414, "lr": 6.96742537016144e-06} +{"train_loss": 0.016021771356463432, "grad_norm": 0.07153071463108063, "global_step": 43987, "epoch": 414, "lr": 6.9659163121704175e-06} +{"train_loss": 0.014914145693182945, "grad_norm": 0.07895640283823013, "global_step": 43988, "epoch": 414, "lr": 6.964407405382439e-06} +{"train_loss": 0.014988572881469186, "grad_norm": 0.09255000203847885, "global_step": 43989, "epoch": 414, "lr": 6.962898649802823e-06, "val_loss": 0.020015660673379898} +{"train_loss": 0.014511343091726303, "grad_norm": 0.06147903576493263, "global_step": 43990, "epoch": 415, "lr": 6.961390045436855e-06} +{"train_loss": 0.014322190545499325, "grad_norm": 0.06211930140852928, "global_step": 43991, "epoch": 415, "lr": 6.959881592289857e-06} +{"train_loss": 0.01707959547638893, "grad_norm": 0.0825214758515358, "global_step": 43992, "epoch": 415, "lr": 6.95837329036711e-06} +{"train_loss": 0.01460410188883543, "grad_norm": 0.07362325489521027, "global_step": 43993, "epoch": 415, "lr": 6.956865139673918e-06} +{"train_loss": 0.01688740774989128, "grad_norm": 0.11973737180233002, "global_step": 43994, "epoch": 415, "lr": 6.955357140215596e-06} +{"train_loss": 0.01620676927268505, "grad_norm": 0.10604611039161682, "global_step": 43995, "epoch": 415, "lr": 6.9538492919974275e-06} +{"train_loss": 0.01450316235423088, "grad_norm": 0.07288998365402222, "global_step": 43996, "epoch": 415, "lr": 6.9523415950246975e-06} +{"train_loss": 0.01611214317381382, "grad_norm": 0.11880486458539963, "global_step": 43997, "epoch": 415, "lr": 6.950834049302734e-06} +{"train_loss": 0.01814349554479122, "grad_norm": 0.1325242668390274, "global_step": 43998, "epoch": 415, "lr": 6.949326654836808e-06} +{"train_loss": 0.014046160504221916, "grad_norm": 0.07876657694578171, "global_step": 43999, "epoch": 415, "lr": 6.947819411632223e-06} +{"train_loss": 0.013432394713163376, "grad_norm": 0.0602777823805809, "global_step": 44000, "epoch": 415, "lr": 6.946312319694287e-06} +{"train_loss": 0.01356971263885498, "grad_norm": 0.074318528175354, "global_step": 44001, "epoch": 415, "lr": 6.944805379028291e-06} +{"train_loss": 0.014713970012962818, "grad_norm": 0.10063450038433075, "global_step": 44002, "epoch": 415, "lr": 6.943298589639513e-06} +{"train_loss": 0.01254582405090332, "grad_norm": 0.0812288448214531, "global_step": 44003, "epoch": 415, "lr": 6.941791951533255e-06} +{"train_loss": 0.01480904035270214, "grad_norm": 0.0736108273267746, "global_step": 44004, "epoch": 415, "lr": 6.94028546471483e-06} +{"train_loss": 0.015532311052083969, "grad_norm": 0.12458344548940659, "global_step": 44005, "epoch": 415, "lr": 6.938779129189499e-06} +{"train_loss": 0.017319083213806152, "grad_norm": 0.0707821175456047, "global_step": 44006, "epoch": 415, "lr": 6.937272944962586e-06} +{"train_loss": 0.0155709870159626, "grad_norm": 0.12801669538021088, "global_step": 44007, "epoch": 415, "lr": 6.93576691203936e-06} +{"train_loss": 0.016640542075037956, "grad_norm": 0.11756233870983124, "global_step": 44008, "epoch": 415, "lr": 6.934261030425127e-06} +{"train_loss": 0.015351982787251472, "grad_norm": 0.06504109501838684, "global_step": 44009, "epoch": 415, "lr": 6.932755300125165e-06} +{"train_loss": 0.01604236662387848, "grad_norm": 0.10144350677728653, "global_step": 44010, "epoch": 415, "lr": 6.9312497211447665e-06} +{"train_loss": 0.014704473316669464, "grad_norm": 0.06938903033733368, "global_step": 44011, "epoch": 415, "lr": 6.929744293489243e-06} +{"train_loss": 0.016577592119574547, "grad_norm": 0.06773149967193604, "global_step": 44012, "epoch": 415, "lr": 6.928239017163862e-06} +{"train_loss": 0.015280859544873238, "grad_norm": 0.0645027682185173, "global_step": 44013, "epoch": 415, "lr": 6.926733892173909e-06} +{"train_loss": 0.01362229697406292, "grad_norm": 0.052314382046461105, "global_step": 44014, "epoch": 415, "lr": 6.925228918524685e-06} +{"train_loss": 0.014424357563257217, "grad_norm": 0.0599907748401165, "global_step": 44015, "epoch": 415, "lr": 6.923724096221479e-06} +{"train_loss": 0.017826110124588013, "grad_norm": 0.08452143520116806, "global_step": 44016, "epoch": 415, "lr": 6.922219425269566e-06} +{"train_loss": 0.014471947215497494, "grad_norm": 0.06145859137177467, "global_step": 44017, "epoch": 415, "lr": 6.920714905674236e-06} +{"train_loss": 0.014162136241793633, "grad_norm": 0.0699106827378273, "global_step": 44018, "epoch": 415, "lr": 6.919210537440807e-06} +{"train_loss": 0.015120375901460648, "grad_norm": 0.150916188955307, "global_step": 44019, "epoch": 415, "lr": 6.917706320574508e-06} +{"train_loss": 0.012884975410997868, "grad_norm": 0.07309292256832123, "global_step": 44020, "epoch": 415, "lr": 6.916202255080656e-06} +{"train_loss": 0.014145314693450928, "grad_norm": 0.09230158478021622, "global_step": 44021, "epoch": 415, "lr": 6.914698340964543e-06} +{"train_loss": 0.014198077842593193, "grad_norm": 0.08059096336364746, "global_step": 44022, "epoch": 415, "lr": 6.913194578231435e-06} +{"train_loss": 0.014850478619337082, "grad_norm": 0.06871432065963745, "global_step": 44023, "epoch": 415, "lr": 6.911690966886619e-06} +{"train_loss": 0.016205284744501114, "grad_norm": 0.10879776626825333, "global_step": 44024, "epoch": 415, "lr": 6.910187506935395e-06} +{"train_loss": 0.014481330290436745, "grad_norm": 0.11020760983228683, "global_step": 44025, "epoch": 415, "lr": 6.908684198383036e-06} +{"train_loss": 0.013632385060191154, "grad_norm": 0.06350169330835342, "global_step": 44026, "epoch": 415, "lr": 6.907181041234806e-06} +{"train_loss": 0.014084039255976677, "grad_norm": 0.06603623926639557, "global_step": 44027, "epoch": 415, "lr": 6.9056780354959995e-06} +{"train_loss": 0.015685996040701866, "grad_norm": 0.11968942731618881, "global_step": 44028, "epoch": 415, "lr": 6.9041751811719125e-06} +{"train_loss": 0.01586967334151268, "grad_norm": 0.0649399533867836, "global_step": 44029, "epoch": 415, "lr": 6.902672478267802e-06} +{"train_loss": 0.016890063881874084, "grad_norm": 0.09190673381090164, "global_step": 44030, "epoch": 415, "lr": 6.901169926788969e-06} +{"train_loss": 0.014716335572302341, "grad_norm": 0.11259964853525162, "global_step": 44031, "epoch": 415, "lr": 6.899667526740672e-06} +{"train_loss": 0.014148896560072899, "grad_norm": 0.06191514432430267, "global_step": 44032, "epoch": 415, "lr": 6.898165278128204e-06} +{"train_loss": 0.014540312811732292, "grad_norm": 0.06556589901447296, "global_step": 44033, "epoch": 415, "lr": 6.896663180956836e-06} +{"train_loss": 0.015148960053920746, "grad_norm": 0.055266816169023514, "global_step": 44034, "epoch": 415, "lr": 6.895161235231845e-06} +{"train_loss": 0.01530006155371666, "grad_norm": 0.07452541589736938, "global_step": 44035, "epoch": 415, "lr": 6.893659440958522e-06} +{"train_loss": 0.016309810802340508, "grad_norm": 0.08395347744226456, "global_step": 44036, "epoch": 415, "lr": 6.892157798142129e-06} +{"train_loss": 0.016668181866407394, "grad_norm": 0.05851884186267853, "global_step": 44037, "epoch": 415, "lr": 6.89065630678794e-06} +{"train_loss": 0.01630622148513794, "grad_norm": 0.09026457369327545, "global_step": 44038, "epoch": 415, "lr": 6.889154966901251e-06} +{"train_loss": 0.012928519397974014, "grad_norm": 0.054466210305690765, "global_step": 44039, "epoch": 415, "lr": 6.887653778487307e-06} +{"train_loss": 0.0130220428109169, "grad_norm": 0.10377242416143417, "global_step": 44040, "epoch": 415, "lr": 6.8861527415514e-06} +{"train_loss": 0.013952058739960194, "grad_norm": 0.07667956501245499, "global_step": 44041, "epoch": 415, "lr": 6.884651856098817e-06} +{"train_loss": 0.015646912157535553, "grad_norm": 0.12004221975803375, "global_step": 44042, "epoch": 415, "lr": 6.883151122134812e-06} +{"train_loss": 0.01672256365418434, "grad_norm": 0.07355432212352753, "global_step": 44043, "epoch": 415, "lr": 6.881650539664652e-06} +{"train_loss": 0.014463558793067932, "grad_norm": 0.11325757205486298, "global_step": 44044, "epoch": 415, "lr": 6.880150108693622e-06} +{"train_loss": 0.012690523639321327, "grad_norm": 0.09446533769369125, "global_step": 44045, "epoch": 415, "lr": 6.878649829227007e-06} +{"train_loss": 0.0162515752017498, "grad_norm": 0.1456877738237381, "global_step": 44046, "epoch": 415, "lr": 6.877149701270047e-06} +{"train_loss": 0.016417227685451508, "grad_norm": 0.09669411927461624, "global_step": 44047, "epoch": 415, "lr": 6.875649724828032e-06} +{"train_loss": 0.01399166975170374, "grad_norm": 0.074283167719841, "global_step": 44048, "epoch": 415, "lr": 6.87414989990624e-06} +{"train_loss": 0.014245398342609406, "grad_norm": 0.06890065968036652, "global_step": 44049, "epoch": 415, "lr": 6.87265022650993e-06} +{"train_loss": 0.014061505906283855, "grad_norm": 0.08886127918958664, "global_step": 44050, "epoch": 415, "lr": 6.871150704644358e-06} +{"train_loss": 0.013241956941783428, "grad_norm": 0.11121819168329239, "global_step": 44051, "epoch": 415, "lr": 6.869651334314809e-06} +{"train_loss": 0.015831714496016502, "grad_norm": 0.14164245128631592, "global_step": 44052, "epoch": 415, "lr": 6.86815211552656e-06} +{"train_loss": 0.015886640176177025, "grad_norm": 0.09310892224311829, "global_step": 44053, "epoch": 415, "lr": 6.866653048284865e-06} +{"train_loss": 0.017589855939149857, "grad_norm": 0.13499167561531067, "global_step": 44054, "epoch": 415, "lr": 6.865154132594986e-06} +{"train_loss": 0.015582273714244366, "grad_norm": 0.06419268995523453, "global_step": 44055, "epoch": 415, "lr": 6.863655368462191e-06} +{"train_loss": 0.014118621125817299, "grad_norm": 0.12136770784854889, "global_step": 44056, "epoch": 415, "lr": 6.862156755891769e-06} +{"train_loss": 0.017133723944425583, "grad_norm": 0.0798962265253067, "global_step": 44057, "epoch": 415, "lr": 6.86065829488895e-06} +{"train_loss": 0.014386128634214401, "grad_norm": 0.11208764463663101, "global_step": 44058, "epoch": 415, "lr": 6.8591599854590185e-06} +{"train_loss": 0.012923214584589005, "grad_norm": 0.0698227807879448, "global_step": 44059, "epoch": 415, "lr": 6.857661827607264e-06} +{"train_loss": 0.014891479164361954, "grad_norm": 0.10446202009916306, "global_step": 44060, "epoch": 415, "lr": 6.856163821338896e-06} +{"train_loss": 0.013450996950268745, "grad_norm": 0.06071428209543228, "global_step": 44061, "epoch": 415, "lr": 6.854665966659207e-06} +{"train_loss": 0.014039331115782261, "grad_norm": 0.0589115135371685, "global_step": 44062, "epoch": 415, "lr": 6.853168263573473e-06} +{"train_loss": 0.017790259793400764, "grad_norm": 0.11111120134592056, "global_step": 44063, "epoch": 415, "lr": 6.851670712086927e-06} +{"train_loss": 0.01449492760002613, "grad_norm": 0.09519606083631516, "global_step": 44064, "epoch": 415, "lr": 6.85017331220485e-06} +{"train_loss": 0.015628525987267494, "grad_norm": 0.09683932363986969, "global_step": 44065, "epoch": 415, "lr": 6.848676063932502e-06} +{"train_loss": 0.014432240277528763, "grad_norm": 0.07083411514759064, "global_step": 44066, "epoch": 415, "lr": 6.8471789672751475e-06} +{"train_loss": 0.013313288800418377, "grad_norm": 0.06986350566148758, "global_step": 44067, "epoch": 415, "lr": 6.845682022238026e-06} +{"train_loss": 0.01542692631483078, "grad_norm": 0.0794290229678154, "global_step": 44068, "epoch": 415, "lr": 6.844185228826411e-06} +{"train_loss": 0.014132753014564514, "grad_norm": 0.06701558828353882, "global_step": 44069, "epoch": 415, "lr": 6.84268858704557e-06} +{"train_loss": 0.016218725591897964, "grad_norm": 0.12256074696779251, "global_step": 44070, "epoch": 415, "lr": 6.841192096900739e-06} +{"train_loss": 0.014882449992001057, "grad_norm": 0.0765279158949852, "global_step": 44071, "epoch": 415, "lr": 6.839695758397208e-06} +{"train_loss": 0.011708627454936504, "grad_norm": 0.10851921141147614, "global_step": 44072, "epoch": 415, "lr": 6.838199571540199e-06} +{"train_loss": 0.016534149646759033, "grad_norm": 0.07276534289121628, "global_step": 44073, "epoch": 415, "lr": 6.836703536335004e-06} +{"train_loss": 0.01473411824554205, "grad_norm": 0.10401942580938339, "global_step": 44074, "epoch": 415, "lr": 6.835207652786846e-06} +{"train_loss": 0.016596371307969093, "grad_norm": 0.07468375563621521, "global_step": 44075, "epoch": 415, "lr": 6.833711920900998e-06} +{"train_loss": 0.016372667625546455, "grad_norm": 0.10894801467657089, "global_step": 44076, "epoch": 415, "lr": 6.832216340682723e-06} +{"train_loss": 0.01685049943625927, "grad_norm": 0.10192204266786575, "global_step": 44077, "epoch": 415, "lr": 6.830720912137267e-06} +{"train_loss": 0.01595381274819374, "grad_norm": 0.09572051465511322, "global_step": 44078, "epoch": 415, "lr": 6.82922563526987e-06} +{"train_loss": 0.012784341350197792, "grad_norm": 0.07177311927080154, "global_step": 44079, "epoch": 415, "lr": 6.827730510085817e-06} +{"train_loss": 0.01678805984556675, "grad_norm": 0.08296119421720505, "global_step": 44080, "epoch": 415, "lr": 6.82623553659033e-06} +{"train_loss": 0.012679457664489746, "grad_norm": 0.07937867939472198, "global_step": 44081, "epoch": 415, "lr": 6.824740714788674e-06} +{"train_loss": 0.013747422024607658, "grad_norm": 0.11373133212327957, "global_step": 44082, "epoch": 415, "lr": 6.823246044686116e-06} +{"train_loss": 0.014119710773229599, "grad_norm": 0.09270065277814865, "global_step": 44083, "epoch": 415, "lr": 6.82175152628789e-06} +{"train_loss": 0.01294262707233429, "grad_norm": 0.12111963331699371, "global_step": 44084, "epoch": 415, "lr": 6.820257159599247e-06} +{"train_loss": 0.014224207028746605, "grad_norm": 0.05700811371207237, "global_step": 44085, "epoch": 415, "lr": 6.81876294462544e-06} +{"train_loss": 0.01408818457275629, "grad_norm": 0.10095501691102982, "global_step": 44086, "epoch": 415, "lr": 6.81726888137173e-06} +{"train_loss": 0.014691771939396858, "grad_norm": 0.0842798501253128, "global_step": 44087, "epoch": 415, "lr": 6.815774969843347e-06} +{"train_loss": 0.014332103542983532, "grad_norm": 0.08646237105131149, "global_step": 44088, "epoch": 415, "lr": 6.814281210045548e-06} +{"train_loss": 0.014312900602817535, "grad_norm": 0.09148499369621277, "global_step": 44089, "epoch": 415, "lr": 6.8127876019836005e-06} +{"train_loss": 0.01506035402417183, "grad_norm": 0.07288643717765808, "global_step": 44090, "epoch": 415, "lr": 6.811294145662728e-06} +{"train_loss": 0.014514360576868057, "grad_norm": 0.0812097042798996, "global_step": 44091, "epoch": 415, "lr": 6.809800841088182e-06} +{"train_loss": 0.015608569607138634, "grad_norm": 0.11176685243844986, "global_step": 44092, "epoch": 415, "lr": 6.808307688265208e-06} +{"train_loss": 0.014625240117311478, "grad_norm": 0.09627986699342728, "global_step": 44093, "epoch": 415, "lr": 6.80681468719907e-06} +{"train_loss": 0.01618773117661476, "grad_norm": 0.0883258655667305, "global_step": 44094, "epoch": 415, "lr": 6.8053218378949905e-06} +{"train_loss": 0.014984896617396822, "grad_norm": 0.13156674802303314, "global_step": 44095, "epoch": 415, "lr": 6.803829140358237e-06, "val_loss": 0.019932154566049576} +{"train_loss": 0.015499601140618324, "grad_norm": 0.0631038025021553, "global_step": 44096, "epoch": 416, "lr": 6.802336594594044e-06} +{"train_loss": 0.016960598528385162, "grad_norm": 0.10202590376138687, "global_step": 44097, "epoch": 416, "lr": 6.8008442006076364e-06} +{"train_loss": 0.013902857899665833, "grad_norm": 0.07832077890634537, "global_step": 44098, "epoch": 416, "lr": 6.799351958404282e-06} +{"train_loss": 0.01795203424990177, "grad_norm": 0.12242389470338821, "global_step": 44099, "epoch": 416, "lr": 6.797859867989226e-06} +{"train_loss": 0.016111239790916443, "grad_norm": 0.08595985174179077, "global_step": 44100, "epoch": 416, "lr": 6.796367929367703e-06} +{"train_loss": 0.015337794087827206, "grad_norm": 0.06760889291763306, "global_step": 44101, "epoch": 416, "lr": 6.794876142544943e-06} +{"train_loss": 0.01796388067305088, "grad_norm": 0.07832933962345123, "global_step": 44102, "epoch": 416, "lr": 6.793384507526196e-06} +{"train_loss": 0.015713807195425034, "grad_norm": 0.07589124888181686, "global_step": 44103, "epoch": 416, "lr": 6.79189302431672e-06} +{"train_loss": 0.015342635102570057, "grad_norm": 0.09168838709592819, "global_step": 44104, "epoch": 416, "lr": 6.790401692921728e-06} +{"train_loss": 0.015345938503742218, "grad_norm": 0.0996292233467102, "global_step": 44105, "epoch": 416, "lr": 6.788910513346474e-06} +{"train_loss": 0.013091208413243294, "grad_norm": 0.0689210295677185, "global_step": 44106, "epoch": 416, "lr": 6.787419485596208e-06} +{"train_loss": 0.01355176791548729, "grad_norm": 0.10018105059862137, "global_step": 44107, "epoch": 416, "lr": 6.785928609676156e-06} +{"train_loss": 0.01393619179725647, "grad_norm": 0.10283481329679489, "global_step": 44108, "epoch": 416, "lr": 6.784437885591549e-06} +{"train_loss": 0.01498232688754797, "grad_norm": 0.06739101558923721, "global_step": 44109, "epoch": 416, "lr": 6.7829473133476315e-06} +{"train_loss": 0.015587549656629562, "grad_norm": 0.09013471752405167, "global_step": 44110, "epoch": 416, "lr": 6.781456892949656e-06} +{"train_loss": 0.011874381452798843, "grad_norm": 0.06656405329704285, "global_step": 44111, "epoch": 416, "lr": 6.779966624402834e-06} +{"train_loss": 0.013111853040754795, "grad_norm": 0.09653839468955994, "global_step": 44112, "epoch": 416, "lr": 6.7784765077124225e-06} +{"train_loss": 0.013623179867863655, "grad_norm": 0.10129570960998535, "global_step": 44113, "epoch": 416, "lr": 6.77698654288364e-06} +{"train_loss": 0.016174130141735077, "grad_norm": 0.11755338311195374, "global_step": 44114, "epoch": 416, "lr": 6.775496729921738e-06} +{"train_loss": 0.018182804808020592, "grad_norm": 0.07565590739250183, "global_step": 44115, "epoch": 416, "lr": 6.774007068831928e-06} +{"train_loss": 0.015097560361027718, "grad_norm": 0.10971464216709137, "global_step": 44116, "epoch": 416, "lr": 6.772517559619462e-06} +{"train_loss": 0.01489889807999134, "grad_norm": 0.10861991345882416, "global_step": 44117, "epoch": 416, "lr": 6.771028202289587e-06} +{"train_loss": 0.016202827915549278, "grad_norm": 0.07013461738824844, "global_step": 44118, "epoch": 416, "lr": 6.769538996847513e-06} +{"train_loss": 0.013647472485899925, "grad_norm": 0.07348942756652832, "global_step": 44119, "epoch": 416, "lr": 6.7680499432984654e-06} +{"train_loss": 0.014837034046649933, "grad_norm": 0.08929380774497986, "global_step": 44120, "epoch": 416, "lr": 6.766561041647707e-06} +{"train_loss": 0.016755402088165283, "grad_norm": 0.08254758268594742, "global_step": 44121, "epoch": 416, "lr": 6.765072291900432e-06} +{"train_loss": 0.013488580472767353, "grad_norm": 0.12155143171548843, "global_step": 44122, "epoch": 416, "lr": 6.763583694061898e-06} +{"train_loss": 0.015881631523370743, "grad_norm": 0.12126867473125458, "global_step": 44123, "epoch": 416, "lr": 6.762095248137334e-06} +{"train_loss": 0.015388816595077515, "grad_norm": 0.10208051651716232, "global_step": 44124, "epoch": 416, "lr": 6.760606954131965e-06} +{"train_loss": 0.014575459994375706, "grad_norm": 0.06877420842647552, "global_step": 44125, "epoch": 416, "lr": 6.759118812051013e-06} +{"train_loss": 0.015345280058681965, "grad_norm": 0.0601264163851738, "global_step": 44126, "epoch": 416, "lr": 6.757630821899708e-06} +{"train_loss": 0.014856484718620777, "grad_norm": 0.06424643844366074, "global_step": 44127, "epoch": 416, "lr": 6.756142983683289e-06} +{"train_loss": 0.01214748527854681, "grad_norm": 0.06249493360519409, "global_step": 44128, "epoch": 416, "lr": 6.7546552974069754e-06} +{"train_loss": 0.013628151267766953, "grad_norm": 0.07828919589519501, "global_step": 44129, "epoch": 416, "lr": 6.753167763075991e-06} +{"train_loss": 0.01489056646823883, "grad_norm": 0.1565774530172348, "global_step": 44130, "epoch": 416, "lr": 6.75168038069558e-06} +{"train_loss": 0.012721389532089233, "grad_norm": 0.07716170698404312, "global_step": 44131, "epoch": 416, "lr": 6.750193150270956e-06} +{"train_loss": 0.018451936542987823, "grad_norm": 0.09951857477426529, "global_step": 44132, "epoch": 416, "lr": 6.748706071807332e-06} +{"train_loss": 0.014195184223353863, "grad_norm": 0.10716700553894043, "global_step": 44133, "epoch": 416, "lr": 6.747219145309946e-06} +{"train_loss": 0.016396811231970787, "grad_norm": 0.1101612001657486, "global_step": 44134, "epoch": 416, "lr": 6.745732370784031e-06} +{"train_loss": 0.012427080422639847, "grad_norm": 0.12999317049980164, "global_step": 44135, "epoch": 416, "lr": 6.744245748234795e-06} +{"train_loss": 0.015106391161680222, "grad_norm": 0.06164368987083435, "global_step": 44136, "epoch": 416, "lr": 6.742759277667476e-06} +{"train_loss": 0.011852730065584183, "grad_norm": 0.13690102100372314, "global_step": 44137, "epoch": 416, "lr": 6.741272959087286e-06} +{"train_loss": 0.016700921580195427, "grad_norm": 0.09689464420080185, "global_step": 44138, "epoch": 416, "lr": 6.739786792499448e-06} +{"train_loss": 0.013969261199235916, "grad_norm": 0.07196847349405289, "global_step": 44139, "epoch": 416, "lr": 6.738300777909179e-06} +{"train_loss": 0.013096134178340435, "grad_norm": 0.0622403584420681, "global_step": 44140, "epoch": 416, "lr": 6.736814915321715e-06} +{"train_loss": 0.014313827268779278, "grad_norm": 0.08201874792575836, "global_step": 44141, "epoch": 416, "lr": 6.735329204742264e-06} +{"train_loss": 0.01583174616098404, "grad_norm": 0.07142090052366257, "global_step": 44142, "epoch": 416, "lr": 6.733843646176063e-06} +{"train_loss": 0.015421221032738686, "grad_norm": 0.08142583072185516, "global_step": 44143, "epoch": 416, "lr": 6.732358239628306e-06} +{"train_loss": 0.013319268822669983, "grad_norm": 0.0894206315279007, "global_step": 44144, "epoch": 416, "lr": 6.730872985104237e-06} +{"train_loss": 0.0145280621945858, "grad_norm": 0.05949018523097038, "global_step": 44145, "epoch": 416, "lr": 6.729387882609045e-06} +{"train_loss": 0.017638204619288445, "grad_norm": 0.09348788857460022, "global_step": 44146, "epoch": 416, "lr": 6.727902932147973e-06} +{"train_loss": 0.014691517688333988, "grad_norm": 0.09890905022621155, "global_step": 44147, "epoch": 416, "lr": 6.726418133726242e-06} +{"train_loss": 0.01346549205482006, "grad_norm": 0.07028177380561829, "global_step": 44148, "epoch": 416, "lr": 6.724933487349061e-06} +{"train_loss": 0.014308685436844826, "grad_norm": 0.0829383134841919, "global_step": 44149, "epoch": 416, "lr": 6.723448993021631e-06} +{"train_loss": 0.016434280201792717, "grad_norm": 0.0803135335445404, "global_step": 44150, "epoch": 416, "lr": 6.72196465074918e-06} +{"train_loss": 0.01395273208618164, "grad_norm": 0.07019829005002975, "global_step": 44151, "epoch": 416, "lr": 6.720480460536932e-06} +{"train_loss": 0.014170532114803791, "grad_norm": 0.13425849378108978, "global_step": 44152, "epoch": 416, "lr": 6.7189964223900895e-06} +{"train_loss": 0.015559559687972069, "grad_norm": 0.08468781411647797, "global_step": 44153, "epoch": 416, "lr": 6.717512536313875e-06} +{"train_loss": 0.015450866892933846, "grad_norm": 0.08278960734605789, "global_step": 44154, "epoch": 416, "lr": 6.716028802313501e-06} +{"train_loss": 0.014322226867079735, "grad_norm": 0.07691878080368042, "global_step": 44155, "epoch": 416, "lr": 6.714545220394169e-06} +{"train_loss": 0.01328654307872057, "grad_norm": 0.07113809138536453, "global_step": 44156, "epoch": 416, "lr": 6.713061790561098e-06} +{"train_loss": 0.016194477677345276, "grad_norm": 0.07024118304252625, "global_step": 44157, "epoch": 416, "lr": 6.7115785128195055e-06} +{"train_loss": 0.018843935802578926, "grad_norm": 0.09021089971065521, "global_step": 44158, "epoch": 416, "lr": 6.710095387174614e-06} +{"train_loss": 0.013554036617279053, "grad_norm": 0.09795722365379333, "global_step": 44159, "epoch": 416, "lr": 6.708612413631615e-06} +{"train_loss": 0.015597041696310043, "grad_norm": 0.06098159775137901, "global_step": 44160, "epoch": 416, "lr": 6.707129592195716e-06} +{"train_loss": 0.014120698906481266, "grad_norm": 0.06014975532889366, "global_step": 44161, "epoch": 416, "lr": 6.7056469228721496e-06} +{"train_loss": 0.015470569021999836, "grad_norm": 0.11799710243940353, "global_step": 44162, "epoch": 416, "lr": 6.704164405666097e-06} +{"train_loss": 0.01393546350300312, "grad_norm": 0.07545728236436844, "global_step": 44163, "epoch": 416, "lr": 6.702682040582786e-06} +{"train_loss": 0.01438051462173462, "grad_norm": 0.08034850656986237, "global_step": 44164, "epoch": 416, "lr": 6.70119982762743e-06} +{"train_loss": 0.0166747085750103, "grad_norm": 0.06655818969011307, "global_step": 44165, "epoch": 416, "lr": 6.699717766805225e-06} +{"train_loss": 0.016468707472085953, "grad_norm": 0.0758371502161026, "global_step": 44166, "epoch": 416, "lr": 6.698235858121371e-06} +{"train_loss": 0.013515636324882507, "grad_norm": 0.07369258999824524, "global_step": 44167, "epoch": 416, "lr": 6.696754101581088e-06} +{"train_loss": 0.015557556413114071, "grad_norm": 0.11084248870611191, "global_step": 44168, "epoch": 416, "lr": 6.695272497189586e-06} +{"train_loss": 0.015196554362773895, "grad_norm": 0.07244624197483063, "global_step": 44169, "epoch": 416, "lr": 6.693791044952058e-06} +{"train_loss": 0.015077418647706509, "grad_norm": 0.1143108606338501, "global_step": 44170, "epoch": 416, "lr": 6.6923097448737095e-06} +{"train_loss": 0.01974661648273468, "grad_norm": 0.0848349779844284, "global_step": 44171, "epoch": 416, "lr": 6.690828596959764e-06} +{"train_loss": 0.014770249836146832, "grad_norm": 0.10420684516429901, "global_step": 44172, "epoch": 416, "lr": 6.689347601215412e-06} +{"train_loss": 0.015214645303785801, "grad_norm": 0.0692894235253334, "global_step": 44173, "epoch": 416, "lr": 6.687866757645844e-06} +{"train_loss": 0.016293853521347046, "grad_norm": 0.06728484481573105, "global_step": 44174, "epoch": 416, "lr": 6.686386066256278e-06} +{"train_loss": 0.014396434649825096, "grad_norm": 0.07852336764335632, "global_step": 44175, "epoch": 416, "lr": 6.684905527051927e-06} +{"train_loss": 0.015312543138861656, "grad_norm": 0.06953836232423782, "global_step": 44176, "epoch": 416, "lr": 6.683425140037969e-06} +{"train_loss": 0.01496068760752678, "grad_norm": 0.06364046037197113, "global_step": 44177, "epoch": 416, "lr": 6.681944905219628e-06} +{"train_loss": 0.01506735198199749, "grad_norm": 0.14691221714019775, "global_step": 44178, "epoch": 416, "lr": 6.680464822602095e-06} +{"train_loss": 0.01295811403542757, "grad_norm": 0.06365238130092621, "global_step": 44179, "epoch": 416, "lr": 6.678984892190559e-06} +{"train_loss": 0.01425692718476057, "grad_norm": 0.05635449290275574, "global_step": 44180, "epoch": 416, "lr": 6.677505113990229e-06} +{"train_loss": 0.015108797699213028, "grad_norm": 0.07282307744026184, "global_step": 44181, "epoch": 416, "lr": 6.676025488006316e-06} +{"train_loss": 0.01487550139427185, "grad_norm": 0.11863677948713303, "global_step": 44182, "epoch": 416, "lr": 6.674546014243993e-06} +{"train_loss": 0.012875616550445557, "grad_norm": 0.07169994711875916, "global_step": 44183, "epoch": 416, "lr": 6.673066692708491e-06} +{"train_loss": 0.013712389394640923, "grad_norm": 0.06219564750790596, "global_step": 44184, "epoch": 416, "lr": 6.671587523404976e-06} +{"train_loss": 0.01720060221850872, "grad_norm": 0.08716978132724762, "global_step": 44185, "epoch": 416, "lr": 6.6701085063386735e-06} +{"train_loss": 0.014552794396877289, "grad_norm": 0.09121312201023102, "global_step": 44186, "epoch": 416, "lr": 6.6686296415147566e-06} +{"train_loss": 0.013212635181844234, "grad_norm": 0.1162177249789238, "global_step": 44187, "epoch": 416, "lr": 6.667150928938426e-06} +{"train_loss": 0.0174773707985878, "grad_norm": 0.11958548426628113, "global_step": 44188, "epoch": 416, "lr": 6.665672368614894e-06} +{"train_loss": 0.014420200139284134, "grad_norm": 0.07028642296791077, "global_step": 44189, "epoch": 416, "lr": 6.664193960549342e-06} +{"train_loss": 0.016068577766418457, "grad_norm": 0.12557625770568848, "global_step": 44190, "epoch": 416, "lr": 6.662715704746958e-06} +{"train_loss": 0.014550401829183102, "grad_norm": 0.1166052371263504, "global_step": 44191, "epoch": 416, "lr": 6.661237601212944e-06} +{"train_loss": 0.014711359515786171, "grad_norm": 0.13690489530563354, "global_step": 44192, "epoch": 416, "lr": 6.659759649952507e-06} +{"train_loss": 0.016936462372541428, "grad_norm": 0.05506844446063042, "global_step": 44193, "epoch": 416, "lr": 6.65828185097081e-06} +{"train_loss": 0.017280016094446182, "grad_norm": 0.08228878676891327, "global_step": 44194, "epoch": 416, "lr": 6.656804204273065e-06} +{"train_loss": 0.012646802701056004, "grad_norm": 0.06971524655818939, "global_step": 44195, "epoch": 416, "lr": 6.655326709864484e-06} +{"train_loss": 0.01687183789908886, "grad_norm": 0.055509231984615326, "global_step": 44196, "epoch": 416, "lr": 6.653849367750209e-06} +{"train_loss": 0.015001941472291946, "grad_norm": 0.132232204079628, "global_step": 44197, "epoch": 416, "lr": 6.652372177935457e-06} +{"train_loss": 0.014102943241596222, "grad_norm": 0.1027812734246254, "global_step": 44198, "epoch": 416, "lr": 6.650895140425434e-06} +{"train_loss": 0.01571020670235157, "grad_norm": 0.10488034784793854, "global_step": 44199, "epoch": 416, "lr": 6.649418255225298e-06} +{"train_loss": 0.01197846420109272, "grad_norm": 0.05197690799832344, "global_step": 44200, "epoch": 416, "lr": 6.647941522340262e-06} +{"train_loss": 0.014972838740092967, "grad_norm": 0.08635810762643814, "global_step": 44201, "epoch": 416, "lr": 6.646464941775499e-06, "val_loss": 0.019550157710909843} +{"train_loss": 0.012128299102187157, "grad_norm": 0.07574966549873352, "global_step": 44202, "epoch": 417, "lr": 6.644988513536216e-06} +{"train_loss": 0.015710793435573578, "grad_norm": 0.08610352873802185, "global_step": 44203, "epoch": 417, "lr": 6.643512237627575e-06} +{"train_loss": 0.016400985419750214, "grad_norm": 0.08211667090654373, "global_step": 44204, "epoch": 417, "lr": 6.642036114054778e-06} +{"train_loss": 0.014317771419882774, "grad_norm": 0.11297351866960526, "global_step": 44205, "epoch": 417, "lr": 6.640560142823021e-06} +{"train_loss": 0.014195915311574936, "grad_norm": 0.07869099080562592, "global_step": 44206, "epoch": 417, "lr": 6.639084323937478e-06} +{"train_loss": 0.01448296569287777, "grad_norm": 0.13778282701969147, "global_step": 44207, "epoch": 417, "lr": 6.637608657403327e-06} +{"train_loss": 0.016467083245515823, "grad_norm": 0.10984884202480316, "global_step": 44208, "epoch": 417, "lr": 6.636133143225764e-06} +{"train_loss": 0.01481255330145359, "grad_norm": 0.06758414208889008, "global_step": 44209, "epoch": 417, "lr": 6.634657781409975e-06} +{"train_loss": 0.017543628811836243, "grad_norm": 0.09121152013540268, "global_step": 44210, "epoch": 417, "lr": 6.633182571961133e-06} +{"train_loss": 0.014353709295392036, "grad_norm": 0.07335970550775528, "global_step": 44211, "epoch": 417, "lr": 6.631707514884427e-06} +{"train_loss": 0.012111106887459755, "grad_norm": 0.10564317554235458, "global_step": 44212, "epoch": 417, "lr": 6.63023261018505e-06} +{"train_loss": 0.014272738248109818, "grad_norm": 0.07789120823144913, "global_step": 44213, "epoch": 417, "lr": 6.628757857868173e-06} +{"train_loss": 0.014714160934090614, "grad_norm": 0.05555994063615799, "global_step": 44214, "epoch": 417, "lr": 6.627283257938971e-06} +{"train_loss": 0.013951539993286133, "grad_norm": 0.06863311678171158, "global_step": 44215, "epoch": 417, "lr": 6.625808810402634e-06} +{"train_loss": 0.013205097988247871, "grad_norm": 0.06633670628070831, "global_step": 44216, "epoch": 417, "lr": 6.624334515264352e-06} +{"train_loss": 0.013669157400727272, "grad_norm": 0.07010705024003983, "global_step": 44217, "epoch": 417, "lr": 6.6228603725292825e-06} +{"train_loss": 0.016612399369478226, "grad_norm": 0.08130733668804169, "global_step": 44218, "epoch": 417, "lr": 6.621386382202632e-06} +{"train_loss": 0.016098763793706894, "grad_norm": 0.08777692914009094, "global_step": 44219, "epoch": 417, "lr": 6.619912544289558e-06} +{"train_loss": 0.01540543045848608, "grad_norm": 0.1436864733695984, "global_step": 44220, "epoch": 417, "lr": 6.61843885879524e-06} +{"train_loss": 0.013447782956063747, "grad_norm": 0.08064502477645874, "global_step": 44221, "epoch": 417, "lr": 6.616965325724861e-06} +{"train_loss": 0.01544024609029293, "grad_norm": 0.07757460325956345, "global_step": 44222, "epoch": 417, "lr": 6.615491945083613e-06} +{"train_loss": 0.016343072056770325, "grad_norm": 0.11109688133001328, "global_step": 44223, "epoch": 417, "lr": 6.614018716876646e-06} +{"train_loss": 0.014330154284834862, "grad_norm": 0.057752422988414764, "global_step": 44224, "epoch": 417, "lr": 6.612545641109158e-06} +{"train_loss": 0.01668054237961769, "grad_norm": 0.14258234202861786, "global_step": 44225, "epoch": 417, "lr": 6.611072717786304e-06} +{"train_loss": 0.014889980666339397, "grad_norm": 0.09732510149478912, "global_step": 44226, "epoch": 417, "lr": 6.609599946913286e-06} +{"train_loss": 0.013587906956672668, "grad_norm": 0.06267018616199493, "global_step": 44227, "epoch": 417, "lr": 6.60812732849525e-06} +{"train_loss": 0.015781961381435394, "grad_norm": 0.07953610271215439, "global_step": 44228, "epoch": 417, "lr": 6.606654862537387e-06} +{"train_loss": 0.0158066488802433, "grad_norm": 0.06400330364704132, "global_step": 44229, "epoch": 417, "lr": 6.605182549044875e-06} +{"train_loss": 0.015349158085882664, "grad_norm": 0.07031923532485962, "global_step": 44230, "epoch": 417, "lr": 6.6037103880228826e-06} +{"train_loss": 0.012609738856554031, "grad_norm": 0.09692610800266266, "global_step": 44231, "epoch": 417, "lr": 6.602238379476567e-06} +{"train_loss": 0.013437516987323761, "grad_norm": 0.055444296449422836, "global_step": 44232, "epoch": 417, "lr": 6.600766523411112e-06} +{"train_loss": 0.014166864566504955, "grad_norm": 0.07036673277616501, "global_step": 44233, "epoch": 417, "lr": 6.599294819831697e-06} +{"train_loss": 0.014159953221678734, "grad_norm": 0.06706836819648743, "global_step": 44234, "epoch": 417, "lr": 6.5978232687434805e-06} +{"train_loss": 0.013743588700890541, "grad_norm": 0.1021309643983841, "global_step": 44235, "epoch": 417, "lr": 6.596351870151634e-06} +{"train_loss": 0.015579501166939735, "grad_norm": 0.06412970274686813, "global_step": 44236, "epoch": 417, "lr": 6.594880624061356e-06} +{"train_loss": 0.014548446983098984, "grad_norm": 0.058828700333833694, "global_step": 44237, "epoch": 417, "lr": 6.593409530477768e-06} +{"train_loss": 0.013210272416472435, "grad_norm": 0.10880596935749054, "global_step": 44238, "epoch": 417, "lr": 6.5919385894060655e-06} +{"train_loss": 0.014790606684982777, "grad_norm": 0.06856080144643784, "global_step": 44239, "epoch": 417, "lr": 6.590467800851419e-06} +{"train_loss": 0.012782704085111618, "grad_norm": 0.06778548657894135, "global_step": 44240, "epoch": 417, "lr": 6.588997164818983e-06} +{"train_loss": 0.015576843172311783, "grad_norm": 0.08741720765829086, "global_step": 44241, "epoch": 417, "lr": 6.587526681313927e-06} +{"train_loss": 0.015016329474747181, "grad_norm": 0.1244087740778923, "global_step": 44242, "epoch": 417, "lr": 6.5860563503414404e-06} +{"train_loss": 0.015948109328746796, "grad_norm": 0.17710717022418976, "global_step": 44243, "epoch": 417, "lr": 6.5845861719066644e-06} +{"train_loss": 0.014275597408413887, "grad_norm": 0.0894012525677681, "global_step": 44244, "epoch": 417, "lr": 6.583116146014768e-06} +{"train_loss": 0.016663748770952225, "grad_norm": 0.15796855092048645, "global_step": 44245, "epoch": 417, "lr": 6.581646272670916e-06} +{"train_loss": 0.015680570155382156, "grad_norm": 0.06476103514432907, "global_step": 44246, "epoch": 417, "lr": 6.580176551880285e-06} +{"train_loss": 0.013242741115391254, "grad_norm": 0.07176566869020462, "global_step": 44247, "epoch": 417, "lr": 6.578706983648025e-06} +{"train_loss": 0.015155363827943802, "grad_norm": 0.07499473541975021, "global_step": 44248, "epoch": 417, "lr": 6.577237567979311e-06} +{"train_loss": 0.013361537829041481, "grad_norm": 0.08679962903261185, "global_step": 44249, "epoch": 417, "lr": 6.575768304879293e-06} +{"train_loss": 0.01527248602360487, "grad_norm": 0.10432391613721848, "global_step": 44250, "epoch": 417, "lr": 6.574299194353151e-06} +{"train_loss": 0.01562553644180298, "grad_norm": 0.07013674825429916, "global_step": 44251, "epoch": 417, "lr": 6.572830236406024e-06} +{"train_loss": 0.014685951173305511, "grad_norm": 0.08341994136571884, "global_step": 44252, "epoch": 417, "lr": 6.571361431043088e-06} +{"train_loss": 0.017097804695367813, "grad_norm": 0.07840156555175781, "global_step": 44253, "epoch": 417, "lr": 6.569892778269521e-06} +{"train_loss": 0.014158949255943298, "grad_norm": 0.05853426828980446, "global_step": 44254, "epoch": 417, "lr": 6.568424278090446e-06} +{"train_loss": 0.01722833514213562, "grad_norm": 0.08217538148164749, "global_step": 44255, "epoch": 417, "lr": 6.566955930511037e-06} +{"train_loss": 0.016329150646924973, "grad_norm": 0.07791322469711304, "global_step": 44256, "epoch": 417, "lr": 6.565487735536463e-06} +{"train_loss": 0.014863040298223495, "grad_norm": 0.0810793936252594, "global_step": 44257, "epoch": 417, "lr": 6.564019693171869e-06} +{"train_loss": 0.013766864314675331, "grad_norm": 0.10234116017818451, "global_step": 44258, "epoch": 417, "lr": 6.562551803422423e-06} +{"train_loss": 0.015206955373287201, "grad_norm": 0.07078243046998978, "global_step": 44259, "epoch": 417, "lr": 6.561084066293288e-06} +{"train_loss": 0.015782231464982033, "grad_norm": 0.07556486874818802, "global_step": 44260, "epoch": 417, "lr": 6.55961648178961e-06} +{"train_loss": 0.015911053866147995, "grad_norm": 0.08611088991165161, "global_step": 44261, "epoch": 417, "lr": 6.55814904991654e-06} +{"train_loss": 0.015572139993309975, "grad_norm": 0.08276952058076859, "global_step": 44262, "epoch": 417, "lr": 6.556681770679241e-06} +{"train_loss": 0.015415419824421406, "grad_norm": 0.11811801046133041, "global_step": 44263, "epoch": 417, "lr": 6.55521464408288e-06} +{"train_loss": 0.015213501639664173, "grad_norm": 0.09264768660068512, "global_step": 44264, "epoch": 417, "lr": 6.553747670132587e-06} +{"train_loss": 0.013525400310754776, "grad_norm": 0.12717314064502716, "global_step": 44265, "epoch": 417, "lr": 6.552280848833547e-06} +{"train_loss": 0.013941222801804543, "grad_norm": 0.12926001846790314, "global_step": 44266, "epoch": 417, "lr": 6.550814180190884e-06} +{"train_loss": 0.0136191938072443, "grad_norm": 0.10121610760688782, "global_step": 44267, "epoch": 417, "lr": 6.549347664209776e-06} +{"train_loss": 0.01539615262299776, "grad_norm": 0.061702027916908264, "global_step": 44268, "epoch": 417, "lr": 6.547881300895354e-06} +{"train_loss": 0.015918880701065063, "grad_norm": 0.09366507828235626, "global_step": 44269, "epoch": 417, "lr": 6.5464150902527775e-06} +{"train_loss": 0.013783219270408154, "grad_norm": 0.04766734316945076, "global_step": 44270, "epoch": 417, "lr": 6.544949032287212e-06} +{"train_loss": 0.017746899276971817, "grad_norm": 0.1135375127196312, "global_step": 44271, "epoch": 417, "lr": 6.543483127003797e-06} +{"train_loss": 0.014592583291232586, "grad_norm": 0.09984079748392105, "global_step": 44272, "epoch": 417, "lr": 6.542017374407678e-06} +{"train_loss": 0.01887669414281845, "grad_norm": 0.10406868159770966, "global_step": 44273, "epoch": 417, "lr": 6.540551774504006e-06} +{"train_loss": 0.014079693704843521, "grad_norm": 0.05687759071588516, "global_step": 44274, "epoch": 417, "lr": 6.5390863272979445e-06} +{"train_loss": 0.012929690070450306, "grad_norm": 0.09268324822187424, "global_step": 44275, "epoch": 417, "lr": 6.537621032794622e-06} +{"train_loss": 0.013326335698366165, "grad_norm": 0.07860158383846283, "global_step": 44276, "epoch": 417, "lr": 6.536155890999202e-06} +{"train_loss": 0.013563687913119793, "grad_norm": 0.08023571223020554, "global_step": 44277, "epoch": 417, "lr": 6.534690901916846e-06} +{"train_loss": 0.015520773828029633, "grad_norm": 0.054211873561143875, "global_step": 44278, "epoch": 417, "lr": 6.5332260655526665e-06} +{"train_loss": 0.01633196882903576, "grad_norm": 0.09139551967382431, "global_step": 44279, "epoch": 417, "lr": 6.531761381911827e-06} +{"train_loss": 0.01653105393052101, "grad_norm": 0.11577991396188736, "global_step": 44280, "epoch": 417, "lr": 6.530296850999479e-06} +{"train_loss": 0.017717676237225533, "grad_norm": 0.07370483875274658, "global_step": 44281, "epoch": 417, "lr": 6.528832472820756e-06} +{"train_loss": 0.013634136877954006, "grad_norm": 0.061642833054065704, "global_step": 44282, "epoch": 417, "lr": 6.527368247380811e-06} +{"train_loss": 0.017063230276107788, "grad_norm": 0.0603780522942543, "global_step": 44283, "epoch": 417, "lr": 6.5259041746848e-06} +{"train_loss": 0.013040139339864254, "grad_norm": 0.07149191200733185, "global_step": 44284, "epoch": 417, "lr": 6.524440254737857e-06} +{"train_loss": 0.014871513471007347, "grad_norm": 0.06690342724323273, "global_step": 44285, "epoch": 417, "lr": 6.522976487545107e-06} +{"train_loss": 0.015306507237255573, "grad_norm": 0.09455934166908264, "global_step": 44286, "epoch": 417, "lr": 6.521512873111718e-06} +{"train_loss": 0.01412900723516941, "grad_norm": 0.1252855658531189, "global_step": 44287, "epoch": 417, "lr": 6.52004941144283e-06} +{"train_loss": 0.014460578560829163, "grad_norm": 0.15970389544963837, "global_step": 44288, "epoch": 417, "lr": 6.518586102543567e-06} +{"train_loss": 0.016027426347136497, "grad_norm": 0.06903557479381561, "global_step": 44289, "epoch": 417, "lr": 6.5171229464190965e-06} +{"train_loss": 0.015558433718979359, "grad_norm": 0.06566355377435684, "global_step": 44290, "epoch": 417, "lr": 6.5156599430745366e-06} +{"train_loss": 0.0156704094260931, "grad_norm": 0.0783400759100914, "global_step": 44291, "epoch": 417, "lr": 6.5141970925150454e-06} +{"train_loss": 0.015845660120248795, "grad_norm": 0.06913311779499054, "global_step": 44292, "epoch": 417, "lr": 6.51273439474574e-06} +{"train_loss": 0.01413116417825222, "grad_norm": 0.05396931990981102, "global_step": 44293, "epoch": 417, "lr": 6.511271849771777e-06} +{"train_loss": 0.015529731288552284, "grad_norm": 0.07096350938081741, "global_step": 44294, "epoch": 417, "lr": 6.509809457598304e-06} +{"train_loss": 0.017026826739311218, "grad_norm": 0.09171460568904877, "global_step": 44295, "epoch": 417, "lr": 6.508347218230448e-06} +{"train_loss": 0.013375087641179562, "grad_norm": 0.05620872601866722, "global_step": 44296, "epoch": 417, "lr": 6.506885131673335e-06} +{"train_loss": 0.01687248982489109, "grad_norm": 0.06548696011304855, "global_step": 44297, "epoch": 417, "lr": 6.50542319793212e-06} +{"train_loss": 0.014881066046655178, "grad_norm": 0.06483156234025955, "global_step": 44298, "epoch": 417, "lr": 6.503961417011922e-06} +{"train_loss": 0.01489326823502779, "grad_norm": 0.08569642156362534, "global_step": 44299, "epoch": 417, "lr": 6.502499788917893e-06} +{"train_loss": 0.015461675822734833, "grad_norm": 0.08022690564393997, "global_step": 44300, "epoch": 417, "lr": 6.501038313655172e-06} +{"train_loss": 0.014351246878504753, "grad_norm": 0.11219937354326248, "global_step": 44301, "epoch": 417, "lr": 6.499576991228884e-06} +{"train_loss": 0.0151364142075181, "grad_norm": 0.1034785583615303, "global_step": 44302, "epoch": 417, "lr": 6.498115821644152e-06} +{"train_loss": 0.016442421823740005, "grad_norm": 0.061544906347990036, "global_step": 44303, "epoch": 417, "lr": 6.496654804906122e-06} +{"train_loss": 0.014801849611103535, "grad_norm": 0.08246107399463654, "global_step": 44304, "epoch": 417, "lr": 6.495193941019945e-06} +{"train_loss": 0.01342350710183382, "grad_norm": 0.10536716133356094, "global_step": 44305, "epoch": 417, "lr": 6.493733229990717e-06} +{"train_loss": 0.015085969120264053, "grad_norm": 0.09685064852237701, "global_step": 44306, "epoch": 417, "lr": 6.492272671823607e-06} +{"train_loss": 0.014984272596127581, "grad_norm": 0.11727927625179291, "global_step": 44307, "epoch": 417, "lr": 6.490812266523716e-06, "val_loss": 0.019758909940719604} +{"train_loss": 0.01236695982515812, "grad_norm": 0.05129191651940346, "global_step": 44308, "epoch": 418, "lr": 6.489352014096206e-06} +{"train_loss": 0.015696242451667786, "grad_norm": 0.1003987118601799, "global_step": 44309, "epoch": 418, "lr": 6.487891914546174e-06} +{"train_loss": 0.014164932072162628, "grad_norm": 0.06316088140010834, "global_step": 44310, "epoch": 418, "lr": 6.486431967878764e-06} +{"train_loss": 0.014958555810153484, "grad_norm": 0.11799696832895279, "global_step": 44311, "epoch": 418, "lr": 6.484972174099125e-06} +{"train_loss": 0.014946658164262772, "grad_norm": 0.12064484506845474, "global_step": 44312, "epoch": 418, "lr": 6.483512533212372e-06} +{"train_loss": 0.015715358778834343, "grad_norm": 0.08259081095457077, "global_step": 44313, "epoch": 418, "lr": 6.4820530452236124e-06} +{"train_loss": 0.014387094415724277, "grad_norm": 0.09951314330101013, "global_step": 44314, "epoch": 418, "lr": 6.480593710138e-06} +{"train_loss": 0.01429379265755415, "grad_norm": 0.058196716010570526, "global_step": 44315, "epoch": 418, "lr": 6.4791345279606615e-06} +{"train_loss": 0.015079855918884277, "grad_norm": 0.07574216276407242, "global_step": 44316, "epoch": 418, "lr": 6.477675498696711e-06} +{"train_loss": 0.015679918229579926, "grad_norm": 0.09850077331066132, "global_step": 44317, "epoch": 418, "lr": 6.4762166223512776e-06} +{"train_loss": 0.01556704007089138, "grad_norm": 0.06035320833325386, "global_step": 44318, "epoch": 418, "lr": 6.474757898929512e-06} +{"train_loss": 0.01402503065764904, "grad_norm": 0.06770822405815125, "global_step": 44319, "epoch": 418, "lr": 6.473299328436499e-06} +{"train_loss": 0.0146424425765872, "grad_norm": 0.0667976513504982, "global_step": 44320, "epoch": 418, "lr": 6.4718409108773855e-06} +{"train_loss": 0.016123972833156586, "grad_norm": 0.060141995549201965, "global_step": 44321, "epoch": 418, "lr": 6.4703826462573005e-06} +{"train_loss": 0.0182428490370512, "grad_norm": 0.08865219354629517, "global_step": 44322, "epoch": 418, "lr": 6.4689245345813495e-06} +{"train_loss": 0.01627267710864544, "grad_norm": 0.0745127722620964, "global_step": 44323, "epoch": 418, "lr": 6.467466575854664e-06} +{"train_loss": 0.014144406653940678, "grad_norm": 0.103823222219944, "global_step": 44324, "epoch": 418, "lr": 6.466008770082388e-06} +{"train_loss": 0.014991761185228825, "grad_norm": 0.09447620064020157, "global_step": 44325, "epoch": 418, "lr": 6.4645511172696186e-06} +{"train_loss": 0.014416310004889965, "grad_norm": 0.06130747124552727, "global_step": 44326, "epoch": 418, "lr": 6.463093617421473e-06} +{"train_loss": 0.015925494953989983, "grad_norm": 0.05987180024385452, "global_step": 44327, "epoch": 418, "lr": 6.461636270543081e-06} +{"train_loss": 0.014259584248065948, "grad_norm": 0.07240666449069977, "global_step": 44328, "epoch": 418, "lr": 6.460179076639583e-06} +{"train_loss": 0.013791910372674465, "grad_norm": 0.052589382976293564, "global_step": 44329, "epoch": 418, "lr": 6.458722035716064e-06} +{"train_loss": 0.013289958238601685, "grad_norm": 0.08835400640964508, "global_step": 44330, "epoch": 418, "lr": 6.457265147777669e-06} +{"train_loss": 0.017981592565774918, "grad_norm": 0.08095560222864151, "global_step": 44331, "epoch": 418, "lr": 6.455808412829501e-06} +{"train_loss": 0.013278271071612835, "grad_norm": 0.05449318885803223, "global_step": 44332, "epoch": 418, "lr": 6.454351830876698e-06} +{"train_loss": 0.016781765967607498, "grad_norm": 0.0658872202038765, "global_step": 44333, "epoch": 418, "lr": 6.452895401924353e-06} +{"train_loss": 0.011806472204625607, "grad_norm": 0.07192488014698029, "global_step": 44334, "epoch": 418, "lr": 6.451439125977593e-06} +{"train_loss": 0.01639898121356964, "grad_norm": 0.07383879274129868, "global_step": 44335, "epoch": 418, "lr": 6.449983003041549e-06} +{"train_loss": 0.016755912452936172, "grad_norm": 0.10712168365716934, "global_step": 44336, "epoch": 418, "lr": 6.448527033121327e-06} +{"train_loss": 0.014643114991486073, "grad_norm": 0.060544125735759735, "global_step": 44337, "epoch": 418, "lr": 6.447071216222028e-06} +{"train_loss": 0.012992504984140396, "grad_norm": 0.13743676245212555, "global_step": 44338, "epoch": 418, "lr": 6.445615552348794e-06} +{"train_loss": 0.01706288382411003, "grad_norm": 0.069574736058712, "global_step": 44339, "epoch": 418, "lr": 6.444160041506708e-06} +{"train_loss": 0.014340377412736416, "grad_norm": 0.06193867325782776, "global_step": 44340, "epoch": 418, "lr": 6.4427046837009065e-06} +{"train_loss": 0.016215473413467407, "grad_norm": 0.08980635553598404, "global_step": 44341, "epoch": 418, "lr": 6.441249478936512e-06} +{"train_loss": 0.01500063855201006, "grad_norm": 0.17760884761810303, "global_step": 44342, "epoch": 418, "lr": 6.439794427218621e-06} +{"train_loss": 0.013753710314631462, "grad_norm": 0.06990547478199005, "global_step": 44343, "epoch": 418, "lr": 6.438339528552334e-06} +{"train_loss": 0.016691390424966812, "grad_norm": 0.13147012889385223, "global_step": 44344, "epoch": 418, "lr": 6.436884782942782e-06} +{"train_loss": 0.015802443027496338, "grad_norm": 0.09607233852148056, "global_step": 44345, "epoch": 418, "lr": 6.435430190395081e-06} +{"train_loss": 0.014276410453021526, "grad_norm": 0.11374038457870483, "global_step": 44346, "epoch": 418, "lr": 6.433975750914323e-06} +{"train_loss": 0.014762334525585175, "grad_norm": 0.08055362105369568, "global_step": 44347, "epoch": 418, "lr": 6.432521464505625e-06} +{"train_loss": 0.015819843858480453, "grad_norm": 0.11474978178739548, "global_step": 44348, "epoch": 418, "lr": 6.431067331174112e-06} +{"train_loss": 0.013963544741272926, "grad_norm": 0.05901985242962837, "global_step": 44349, "epoch": 418, "lr": 6.429613350924879e-06} +{"train_loss": 0.015009805560112, "grad_norm": 0.09312267601490021, "global_step": 44350, "epoch": 418, "lr": 6.428159523763028e-06} +{"train_loss": 0.013488375581800938, "grad_norm": 0.08462686091661453, "global_step": 44351, "epoch": 418, "lr": 6.426705849693676e-06} +{"train_loss": 0.014581204392015934, "grad_norm": 0.11180565506219864, "global_step": 44352, "epoch": 418, "lr": 6.425252328721942e-06} +{"train_loss": 0.014158724807202816, "grad_norm": 0.10370253026485443, "global_step": 44353, "epoch": 418, "lr": 6.423798960852917e-06} +{"train_loss": 0.012761427089571953, "grad_norm": 0.050162047147750854, "global_step": 44354, "epoch": 418, "lr": 6.4223457460917015e-06} +{"train_loss": 0.01830059103667736, "grad_norm": 0.11972316354513168, "global_step": 44355, "epoch": 418, "lr": 6.420892684443425e-06} +{"train_loss": 0.01667880080640316, "grad_norm": 0.10806043446063995, "global_step": 44356, "epoch": 418, "lr": 6.419439775913167e-06} +{"train_loss": 0.016928808763623238, "grad_norm": 0.07104665040969849, "global_step": 44357, "epoch": 418, "lr": 6.41798702050605e-06} +{"train_loss": 0.01584668457508087, "grad_norm": 0.06840790808200836, "global_step": 44358, "epoch": 418, "lr": 6.416534418227177e-06} +{"train_loss": 0.015428109094500542, "grad_norm": 0.06099523603916168, "global_step": 44359, "epoch": 418, "lr": 6.415081969081649e-06} +{"train_loss": 0.014409930445253849, "grad_norm": 0.08974021673202515, "global_step": 44360, "epoch": 418, "lr": 6.413629673074561e-06} +{"train_loss": 0.015673402696847916, "grad_norm": 0.1224123165011406, "global_step": 44361, "epoch": 418, "lr": 6.412177530211022e-06} +{"train_loss": 0.015931855887174606, "grad_norm": 0.059724342077970505, "global_step": 44362, "epoch": 418, "lr": 6.410725540496143e-06} +{"train_loss": 0.014586769044399261, "grad_norm": 0.09254292398691177, "global_step": 44363, "epoch": 418, "lr": 6.409273703935015e-06} +{"train_loss": 0.01909472420811653, "grad_norm": 0.06824444234371185, "global_step": 44364, "epoch": 418, "lr": 6.4078220205327335e-06} +{"train_loss": 0.01438889466226101, "grad_norm": 0.08936161547899246, "global_step": 44365, "epoch": 418, "lr": 6.406370490294422e-06} +{"train_loss": 0.015345498919487, "grad_norm": 0.14676132798194885, "global_step": 44366, "epoch": 418, "lr": 6.404919113225166e-06} +{"train_loss": 0.015093729831278324, "grad_norm": 0.055943477898836136, "global_step": 44367, "epoch": 418, "lr": 6.403467889330056e-06} +{"train_loss": 0.01783897541463375, "grad_norm": 0.08681930601596832, "global_step": 44368, "epoch": 418, "lr": 6.4020168186141985e-06} +{"train_loss": 0.01486243586987257, "grad_norm": 0.07998831570148468, "global_step": 44369, "epoch": 418, "lr": 6.400565901082706e-06} +{"train_loss": 0.014391432516276836, "grad_norm": 0.07909739762544632, "global_step": 44370, "epoch": 418, "lr": 6.3991151367406525e-06} +{"train_loss": 0.016448426991701126, "grad_norm": 0.07062828540802002, "global_step": 44371, "epoch": 418, "lr": 6.397664525593156e-06} +{"train_loss": 0.012940861284732819, "grad_norm": 0.07783529907464981, "global_step": 44372, "epoch": 418, "lr": 6.396214067645295e-06} +{"train_loss": 0.014416727237403393, "grad_norm": 0.07709893584251404, "global_step": 44373, "epoch": 418, "lr": 6.394763762902184e-06} +{"train_loss": 0.014724291861057281, "grad_norm": 0.06619042903184891, "global_step": 44374, "epoch": 418, "lr": 6.3933136113689e-06} +{"train_loss": 0.016046755015850067, "grad_norm": 0.07573456317186356, "global_step": 44375, "epoch": 418, "lr": 6.391863613050547e-06} +{"train_loss": 0.015692196786403656, "grad_norm": 0.088656947016716, "global_step": 44376, "epoch": 418, "lr": 6.390413767952236e-06} +{"train_loss": 0.015959877520799637, "grad_norm": 0.05941697210073471, "global_step": 44377, "epoch": 418, "lr": 6.388964076079041e-06} +{"train_loss": 0.016395458951592445, "grad_norm": 0.10619507730007172, "global_step": 44378, "epoch": 418, "lr": 6.387514537436051e-06} +{"train_loss": 0.013509462587535381, "grad_norm": 0.15544281899929047, "global_step": 44379, "epoch": 418, "lr": 6.386065152028375e-06} +{"train_loss": 0.013997994363307953, "grad_norm": 0.06591757386922836, "global_step": 44380, "epoch": 418, "lr": 6.3846159198610925e-06} +{"train_loss": 0.014929749071598053, "grad_norm": 0.060716722160577774, "global_step": 44381, "epoch": 418, "lr": 6.383166840939303e-06} +{"train_loss": 0.013880586251616478, "grad_norm": 0.05438472703099251, "global_step": 44382, "epoch": 418, "lr": 6.381717915268104e-06} +{"train_loss": 0.015336080454289913, "grad_norm": 0.09694822877645493, "global_step": 44383, "epoch": 418, "lr": 6.380269142852585e-06} +{"train_loss": 0.013803349807858467, "grad_norm": 0.08107736706733704, "global_step": 44384, "epoch": 418, "lr": 6.378820523697815e-06} +{"train_loss": 0.014252685941755772, "grad_norm": 0.14133229851722717, "global_step": 44385, "epoch": 418, "lr": 6.3773720578089e-06} +{"train_loss": 0.014885514974594116, "grad_norm": 0.06587289273738861, "global_step": 44386, "epoch": 418, "lr": 6.375923745190937e-06} +{"train_loss": 0.013284435495734215, "grad_norm": 0.12699037790298462, "global_step": 44387, "epoch": 418, "lr": 6.374475585848999e-06} +{"train_loss": 0.014837674796581268, "grad_norm": 0.10336386412382126, "global_step": 44388, "epoch": 418, "lr": 6.373027579788182e-06} +{"train_loss": 0.013338328339159489, "grad_norm": 0.07281167805194855, "global_step": 44389, "epoch": 418, "lr": 6.371579727013582e-06} +{"train_loss": 0.012572724372148514, "grad_norm": 0.0834331139922142, "global_step": 44390, "epoch": 418, "lr": 6.370132027530279e-06} +{"train_loss": 0.01324655395001173, "grad_norm": 0.08071567863225937, "global_step": 44391, "epoch": 418, "lr": 6.368684481343345e-06} +{"train_loss": 0.017584633082151413, "grad_norm": 0.09018181264400482, "global_step": 44392, "epoch": 418, "lr": 6.367237088457878e-06} +{"train_loss": 0.01663857512176037, "grad_norm": 0.12200725078582764, "global_step": 44393, "epoch": 418, "lr": 6.365789848878978e-06} +{"train_loss": 0.015467002056539059, "grad_norm": 0.13387802243232727, "global_step": 44394, "epoch": 418, "lr": 6.364342762611708e-06} +{"train_loss": 0.012871812097728252, "grad_norm": 0.08002849668264389, "global_step": 44395, "epoch": 418, "lr": 6.362895829661164e-06} +{"train_loss": 0.01434624195098877, "grad_norm": 0.056632719933986664, "global_step": 44396, "epoch": 418, "lr": 6.361449050032437e-06} +{"train_loss": 0.013822441920638084, "grad_norm": 0.1425313651561737, "global_step": 44397, "epoch": 418, "lr": 6.360002423730582e-06} +{"train_loss": 0.014390811324119568, "grad_norm": 0.09433300048112869, "global_step": 44398, "epoch": 418, "lr": 6.358555950760703e-06} +{"train_loss": 0.014186810702085495, "grad_norm": 0.06660454720258713, "global_step": 44399, "epoch": 418, "lr": 6.357109631127889e-06} +{"train_loss": 0.016520783305168152, "grad_norm": 0.08617126941680908, "global_step": 44400, "epoch": 418, "lr": 6.355663464837197e-06} +{"train_loss": 0.013691676780581474, "grad_norm": 0.11565754562616348, "global_step": 44401, "epoch": 418, "lr": 6.35421745189374e-06} +{"train_loss": 0.015393448993563652, "grad_norm": 0.10009972751140594, "global_step": 44402, "epoch": 418, "lr": 6.35277159230257e-06} +{"train_loss": 0.01569841243326664, "grad_norm": 0.07109227776527405, "global_step": 44403, "epoch": 418, "lr": 6.351325886068787e-06} +{"train_loss": 0.013642714358866215, "grad_norm": 0.06118537113070488, "global_step": 44404, "epoch": 418, "lr": 6.349880333197456e-06} +{"train_loss": 0.014814485795795918, "grad_norm": 0.10515985637903214, "global_step": 44405, "epoch": 418, "lr": 6.348434933693659e-06} +{"train_loss": 0.013879344798624516, "grad_norm": 0.08282068371772766, "global_step": 44406, "epoch": 418, "lr": 6.346989687562488e-06} +{"train_loss": 0.01764022558927536, "grad_norm": 0.07740806043148041, "global_step": 44407, "epoch": 418, "lr": 6.345544594809011e-06} +{"train_loss": 0.014418432489037514, "grad_norm": 0.0630955770611763, "global_step": 44408, "epoch": 418, "lr": 6.344099655438296e-06} +{"train_loss": 0.015528753399848938, "grad_norm": 0.07127202302217484, "global_step": 44409, "epoch": 418, "lr": 6.3426548694554335e-06} +{"train_loss": 0.0139309698715806, "grad_norm": 0.07910830527544022, "global_step": 44410, "epoch": 418, "lr": 6.341210236865502e-06} +{"train_loss": 0.014604085125029087, "grad_norm": 0.07918982952833176, "global_step": 44411, "epoch": 418, "lr": 6.339765757673561e-06} +{"train_loss": 0.01332648191601038, "grad_norm": 0.0634913370013237, "global_step": 44412, "epoch": 418, "lr": 6.338321431884708e-06} +{"train_loss": 0.014974703687190448, "grad_norm": 0.09793242812156677, "global_step": 44413, "epoch": 418, "lr": 6.336877259504004e-06, "val_loss": 0.020155098289251328} +{"train_loss": 0.014353850856423378, "grad_norm": 0.0797848030924797, "global_step": 44414, "epoch": 419, "lr": 6.335433240536515e-06} +{"train_loss": 0.015411549247801304, "grad_norm": 0.08818435668945312, "global_step": 44415, "epoch": 419, "lr": 6.33398937498732e-06} +{"train_loss": 0.012644990347325802, "grad_norm": 0.10115966200828552, "global_step": 44416, "epoch": 419, "lr": 6.332545662861511e-06} +{"train_loss": 0.014499368146061897, "grad_norm": 0.05386687070131302, "global_step": 44417, "epoch": 419, "lr": 6.331102104164138e-06} +{"train_loss": 0.01555121410638094, "grad_norm": 0.07263489812612534, "global_step": 44418, "epoch": 419, "lr": 6.329658698900287e-06} +{"train_loss": 0.012750184163451195, "grad_norm": 0.15015169978141785, "global_step": 44419, "epoch": 419, "lr": 6.3282154470750146e-06} +{"train_loss": 0.016005542129278183, "grad_norm": 0.06743476539850235, "global_step": 44420, "epoch": 419, "lr": 6.32677234869341e-06} +{"train_loss": 0.01336679607629776, "grad_norm": 0.06656145304441452, "global_step": 44421, "epoch": 419, "lr": 6.3253294037605274e-06} +{"train_loss": 0.015878131613135338, "grad_norm": 0.10520526766777039, "global_step": 44422, "epoch": 419, "lr": 6.3238866122814435e-06} +{"train_loss": 0.014623955823481083, "grad_norm": 0.10319146513938904, "global_step": 44423, "epoch": 419, "lr": 6.322443974261233e-06} +{"train_loss": 0.014176092110574245, "grad_norm": 0.10402460396289825, "global_step": 44424, "epoch": 419, "lr": 6.321001489704964e-06} +{"train_loss": 0.017226163297891617, "grad_norm": 0.07633566111326218, "global_step": 44425, "epoch": 419, "lr": 6.3195591586176885e-06} +{"train_loss": 0.015674494206905365, "grad_norm": 0.08007783442735672, "global_step": 44426, "epoch": 419, "lr": 6.318116981004485e-06} +{"train_loss": 0.013966148719191551, "grad_norm": 0.06952304393053055, "global_step": 44427, "epoch": 419, "lr": 6.316674956870438e-06} +{"train_loss": 0.01566937193274498, "grad_norm": 0.09306781738996506, "global_step": 44428, "epoch": 419, "lr": 6.3152330862205785e-06} +{"train_loss": 0.012846122495830059, "grad_norm": 0.06313060969114304, "global_step": 44429, "epoch": 419, "lr": 6.313791369059996e-06} +{"train_loss": 0.013442374765872955, "grad_norm": 0.142161026597023, "global_step": 44430, "epoch": 419, "lr": 6.3123498053937634e-06} +{"train_loss": 0.01472659595310688, "grad_norm": 0.06948959082365036, "global_step": 44431, "epoch": 419, "lr": 6.310908395226939e-06} +{"train_loss": 0.01573892869055271, "grad_norm": 0.08581966906785965, "global_step": 44432, "epoch": 419, "lr": 6.309467138564562e-06} +{"train_loss": 0.016353121027350426, "grad_norm": 0.08187069743871689, "global_step": 44433, "epoch": 419, "lr": 6.308026035411724e-06} +{"train_loss": 0.012723726220428944, "grad_norm": 0.05438052490353584, "global_step": 44434, "epoch": 419, "lr": 6.306585085773487e-06} +{"train_loss": 0.016371577978134155, "grad_norm": 0.12157923728227615, "global_step": 44435, "epoch": 419, "lr": 6.305144289654902e-06} +{"train_loss": 0.013276660814881325, "grad_norm": 0.13585834205150604, "global_step": 44436, "epoch": 419, "lr": 6.303703647061049e-06} +{"train_loss": 0.0176557470113039, "grad_norm": 0.09262252599000931, "global_step": 44437, "epoch": 419, "lr": 6.302263157996974e-06} +{"train_loss": 0.015131969004869461, "grad_norm": 0.10768691450357437, "global_step": 44438, "epoch": 419, "lr": 6.300822822467733e-06} +{"train_loss": 0.015477280132472515, "grad_norm": 0.10053331404924393, "global_step": 44439, "epoch": 419, "lr": 6.2993826404783965e-06} +{"train_loss": 0.015316488221287727, "grad_norm": 0.07214796543121338, "global_step": 44440, "epoch": 419, "lr": 6.297942612034036e-06} +{"train_loss": 0.01511984970420599, "grad_norm": 0.05979298800230026, "global_step": 44441, "epoch": 419, "lr": 6.2965027371396925e-06} +{"train_loss": 0.014465371146798134, "grad_norm": 0.06888388097286224, "global_step": 44442, "epoch": 419, "lr": 6.29506301580044e-06} +{"train_loss": 0.01456029899418354, "grad_norm": 0.0653296485543251, "global_step": 44443, "epoch": 419, "lr": 6.293623448021313e-06} +{"train_loss": 0.014940284192562103, "grad_norm": 0.07173304259777069, "global_step": 44444, "epoch": 419, "lr": 6.2921840338074025e-06} +{"train_loss": 0.012732538394629955, "grad_norm": 0.06011960655450821, "global_step": 44445, "epoch": 419, "lr": 6.290744773163737e-06} +{"train_loss": 0.016601620241999626, "grad_norm": 0.06525176018476486, "global_step": 44446, "epoch": 419, "lr": 6.289305666095385e-06} +{"train_loss": 0.016859155148267746, "grad_norm": 0.08297418802976608, "global_step": 44447, "epoch": 419, "lr": 6.287866712607416e-06} +{"train_loss": 0.014122305437922478, "grad_norm": 0.15314148366451263, "global_step": 44448, "epoch": 419, "lr": 6.286427912704873e-06} +{"train_loss": 0.013644900172948837, "grad_norm": 0.10424640774726868, "global_step": 44449, "epoch": 419, "lr": 6.2849892663928045e-06} +{"train_loss": 0.013277687132358551, "grad_norm": 0.08858465403318405, "global_step": 44450, "epoch": 419, "lr": 6.283550773676266e-06} +{"train_loss": 0.014951842837035656, "grad_norm": 0.09436745941638947, "global_step": 44451, "epoch": 419, "lr": 6.282112434560333e-06} +{"train_loss": 0.014299864880740643, "grad_norm": 0.10160958021879196, "global_step": 44452, "epoch": 419, "lr": 6.280674249050033e-06} +{"train_loss": 0.014503848738968372, "grad_norm": 0.06936383992433548, "global_step": 44453, "epoch": 419, "lr": 6.279236217150442e-06} +{"train_loss": 0.015404503792524338, "grad_norm": 0.1286417692899704, "global_step": 44454, "epoch": 419, "lr": 6.277798338866603e-06} +{"train_loss": 0.013132713735103607, "grad_norm": 0.0714142695069313, "global_step": 44455, "epoch": 419, "lr": 6.276360614203552e-06} +{"train_loss": 0.015217332169413567, "grad_norm": 0.08356944471597672, "global_step": 44456, "epoch": 419, "lr": 6.274923043166353e-06} +{"train_loss": 0.0128629170358181, "grad_norm": 0.10473122447729111, "global_step": 44457, "epoch": 419, "lr": 6.273485625760072e-06} +{"train_loss": 0.016462573781609535, "grad_norm": 0.1116207093000412, "global_step": 44458, "epoch": 419, "lr": 6.272048361989741e-06} +{"train_loss": 0.015111352317035198, "grad_norm": 0.06274618953466415, "global_step": 44459, "epoch": 419, "lr": 6.2706112518604194e-06} +{"train_loss": 0.01780001074075699, "grad_norm": 0.21346396207809448, "global_step": 44460, "epoch": 419, "lr": 6.2691742953771445e-06} +{"train_loss": 0.0158220324665308, "grad_norm": 0.12665817141532898, "global_step": 44461, "epoch": 419, "lr": 6.267737492544984e-06} +{"train_loss": 0.016333196312189102, "grad_norm": 0.09492766112089157, "global_step": 44462, "epoch": 419, "lr": 6.266300843368961e-06} +{"train_loss": 0.012361198663711548, "grad_norm": 0.0668712854385376, "global_step": 44463, "epoch": 419, "lr": 6.264864347854138e-06} +{"train_loss": 0.014843422919511795, "grad_norm": 0.06374845653772354, "global_step": 44464, "epoch": 419, "lr": 6.263428006005578e-06} +{"train_loss": 0.014911177568137646, "grad_norm": 0.09874147921800613, "global_step": 44465, "epoch": 419, "lr": 6.261991817828306e-06} +{"train_loss": 0.016460729762911797, "grad_norm": 0.09155996888875961, "global_step": 44466, "epoch": 419, "lr": 6.2605557833273656e-06} +{"train_loss": 0.01676723174750805, "grad_norm": 0.06538058072328568, "global_step": 44467, "epoch": 419, "lr": 6.259119902507809e-06} +{"train_loss": 0.014619804918766022, "grad_norm": 0.07283113896846771, "global_step": 44468, "epoch": 419, "lr": 6.257684175374695e-06} +{"train_loss": 0.01606491208076477, "grad_norm": 0.09479741007089615, "global_step": 44469, "epoch": 419, "lr": 6.256248601933046e-06} +{"train_loss": 0.014094507321715355, "grad_norm": 0.0743371918797493, "global_step": 44470, "epoch": 419, "lr": 6.254813182187913e-06} +{"train_loss": 0.016569709405303, "grad_norm": 0.10668826103210449, "global_step": 44471, "epoch": 419, "lr": 6.253377916144365e-06} +{"train_loss": 0.013386405073106289, "grad_norm": 0.06557747721672058, "global_step": 44472, "epoch": 419, "lr": 6.251942803807398e-06} +{"train_loss": 0.014954334124922752, "grad_norm": 0.10800058394670486, "global_step": 44473, "epoch": 419, "lr": 6.25050784518208e-06} +{"train_loss": 0.013234928250312805, "grad_norm": 0.09196770936250687, "global_step": 44474, "epoch": 419, "lr": 6.249073040273451e-06} +{"train_loss": 0.014089947566390038, "grad_norm": 0.09181658923625946, "global_step": 44475, "epoch": 419, "lr": 6.247638389086569e-06} +{"train_loss": 0.01349029503762722, "grad_norm": 0.05763296037912369, "global_step": 44476, "epoch": 419, "lr": 6.2462038916264396e-06} +{"train_loss": 0.014918779954314232, "grad_norm": 0.08396045863628387, "global_step": 44477, "epoch": 419, "lr": 6.244769547898138e-06} +{"train_loss": 0.015681616961956024, "grad_norm": 0.06797303259372711, "global_step": 44478, "epoch": 419, "lr": 6.243335357906682e-06} +{"train_loss": 0.014711057767271996, "grad_norm": 0.07780758291482925, "global_step": 44479, "epoch": 419, "lr": 6.241901321657112e-06} +{"train_loss": 0.014599411748349667, "grad_norm": 0.06378452479839325, "global_step": 44480, "epoch": 419, "lr": 6.240467439154468e-06} +{"train_loss": 0.014638261869549751, "grad_norm": 0.09680056571960449, "global_step": 44481, "epoch": 419, "lr": 6.239033710403796e-06} +{"train_loss": 0.0149796511977911, "grad_norm": 0.06122644245624542, "global_step": 44482, "epoch": 419, "lr": 6.237600135410127e-06} +{"train_loss": 0.014515459537506104, "grad_norm": 0.0772739052772522, "global_step": 44483, "epoch": 419, "lr": 6.2361667141785044e-06} +{"train_loss": 0.014799468219280243, "grad_norm": 0.12210162729024887, "global_step": 44484, "epoch": 419, "lr": 6.234733446713953e-06} +{"train_loss": 0.015256663784384727, "grad_norm": 0.07014565169811249, "global_step": 44485, "epoch": 419, "lr": 6.233300333021519e-06} +{"train_loss": 0.014691866002976894, "grad_norm": 0.06653741747140884, "global_step": 44486, "epoch": 419, "lr": 6.231867373106231e-06} +{"train_loss": 0.014928006567060947, "grad_norm": 0.10009420663118362, "global_step": 44487, "epoch": 419, "lr": 6.2304345669731205e-06} +{"train_loss": 0.014222982339560986, "grad_norm": 0.1054002195596695, "global_step": 44488, "epoch": 419, "lr": 6.229001914627241e-06} +{"train_loss": 0.014581995084881783, "grad_norm": 0.09206829965114594, "global_step": 44489, "epoch": 419, "lr": 6.227569416073614e-06} +{"train_loss": 0.012026384472846985, "grad_norm": 0.06806391477584839, "global_step": 44490, "epoch": 419, "lr": 6.226137071317262e-06} +{"train_loss": 0.01580739952623844, "grad_norm": 0.10121309012174606, "global_step": 44491, "epoch": 419, "lr": 6.22470488036323e-06} +{"train_loss": 0.015594886615872383, "grad_norm": 0.0809754729270935, "global_step": 44492, "epoch": 419, "lr": 6.223272843216554e-06} +{"train_loss": 0.013924261555075645, "grad_norm": 0.06742671132087708, "global_step": 44493, "epoch": 419, "lr": 6.221840959882247e-06} +{"train_loss": 0.017802197486162186, "grad_norm": 0.08116855472326279, "global_step": 44494, "epoch": 419, "lr": 6.22040923036536e-06} +{"train_loss": 0.013679800555109978, "grad_norm": 0.05846003070473671, "global_step": 44495, "epoch": 419, "lr": 6.218977654670932e-06} +{"train_loss": 0.016233615577220917, "grad_norm": 0.06258155405521393, "global_step": 44496, "epoch": 419, "lr": 6.2175462328039505e-06} +{"train_loss": 0.013970224186778069, "grad_norm": 0.07917433232069016, "global_step": 44497, "epoch": 419, "lr": 6.216114964769476e-06} +{"train_loss": 0.01451817899942398, "grad_norm": 0.06212581694126129, "global_step": 44498, "epoch": 419, "lr": 6.214683850572545e-06} +{"train_loss": 0.014417802914977074, "grad_norm": 0.06074133142828941, "global_step": 44499, "epoch": 419, "lr": 6.213252890218163e-06} +{"train_loss": 0.013241305015981197, "grad_norm": 0.051699843257665634, "global_step": 44500, "epoch": 419, "lr": 6.211822083711366e-06} +{"train_loss": 0.016511157155036926, "grad_norm": 0.08401480317115784, "global_step": 44501, "epoch": 419, "lr": 6.210391431057189e-06} +{"train_loss": 0.019975267350673676, "grad_norm": 0.1507219821214676, "global_step": 44502, "epoch": 419, "lr": 6.20896093226066e-06} +{"train_loss": 0.015420455485582352, "grad_norm": 0.12960246205329895, "global_step": 44503, "epoch": 419, "lr": 6.207530587326782e-06} +{"train_loss": 0.016221825033426285, "grad_norm": 0.06941789388656616, "global_step": 44504, "epoch": 419, "lr": 6.2061003962606e-06} +{"train_loss": 0.014177295379340649, "grad_norm": 0.15119311213493347, "global_step": 44505, "epoch": 419, "lr": 6.204670359067149e-06} +{"train_loss": 0.01459345780313015, "grad_norm": 0.12857568264007568, "global_step": 44506, "epoch": 419, "lr": 6.203240475751437e-06} +{"train_loss": 0.012505501508712769, "grad_norm": 0.16567444801330566, "global_step": 44507, "epoch": 419, "lr": 6.201810746318481e-06} +{"train_loss": 0.016247669234871864, "grad_norm": 0.0719335526227951, "global_step": 44508, "epoch": 419, "lr": 6.2003811707733115e-06} +{"train_loss": 0.016282299533486366, "grad_norm": 0.11626590043306351, "global_step": 44509, "epoch": 419, "lr": 6.198951749120968e-06} +{"train_loss": 0.013599734753370285, "grad_norm": 0.07609762251377106, "global_step": 44510, "epoch": 419, "lr": 6.197522481366452e-06} +{"train_loss": 0.013387387618422508, "grad_norm": 0.09435354173183441, "global_step": 44511, "epoch": 419, "lr": 6.196093367514788e-06} +{"train_loss": 0.01565689966082573, "grad_norm": 0.08058756589889526, "global_step": 44512, "epoch": 419, "lr": 6.194664407571027e-06} +{"train_loss": 0.01703503355383873, "grad_norm": 0.08705680817365646, "global_step": 44513, "epoch": 419, "lr": 6.193235601540143e-06} +{"train_loss": 0.014775896444916725, "grad_norm": 0.08020316064357758, "global_step": 44514, "epoch": 419, "lr": 6.191806949427176e-06} +{"train_loss": 0.018095605075359344, "grad_norm": 0.05918271094560623, "global_step": 44515, "epoch": 419, "lr": 6.190378451237156e-06} +{"train_loss": 0.016284342855215073, "grad_norm": 0.09874136745929718, "global_step": 44516, "epoch": 419, "lr": 6.188950106975089e-06} +{"train_loss": 0.016386356204748154, "grad_norm": 0.08626885712146759, "global_step": 44517, "epoch": 419, "lr": 6.187521916645994e-06} +{"train_loss": 0.015728266909718513, "grad_norm": 0.15173262357711792, "global_step": 44518, "epoch": 419, "lr": 6.186093880254906e-06} +{"train_loss": 0.01499144493971231, "grad_norm": 0.12987321615219116, "global_step": 44519, "epoch": 419, "lr": 6.184665997806832e-06, "val_loss": 0.019419120624661446} +{"train_loss": 0.014082547277212143, "grad_norm": 0.06564980745315552, "global_step": 44520, "epoch": 420, "lr": 6.1832382693067735e-06} +{"train_loss": 0.015311057679355145, "grad_norm": 0.10166607797145844, "global_step": 44521, "epoch": 420, "lr": 6.18181069475976e-06} +{"train_loss": 0.013664535246789455, "grad_norm": 0.06803236901760101, "global_step": 44522, "epoch": 420, "lr": 6.18038327417082e-06} +{"train_loss": 0.0142957903444767, "grad_norm": 0.07649093866348267, "global_step": 44523, "epoch": 420, "lr": 6.178956007544945e-06} +{"train_loss": 0.014631533995270729, "grad_norm": 0.056603800505399704, "global_step": 44524, "epoch": 420, "lr": 6.177528894887169e-06} +{"train_loss": 0.01445431262254715, "grad_norm": 0.07641703635454178, "global_step": 44525, "epoch": 420, "lr": 6.1761019362024945e-06} +{"train_loss": 0.014688463881611824, "grad_norm": 0.061258286237716675, "global_step": 44526, "epoch": 420, "lr": 6.174675131495944e-06} +{"train_loss": 0.013515522703528404, "grad_norm": 0.08159005641937256, "global_step": 44527, "epoch": 420, "lr": 6.17324848077252e-06} +{"train_loss": 0.015013298019766808, "grad_norm": 0.11579468846321106, "global_step": 44528, "epoch": 420, "lr": 6.171821984037235e-06} +{"train_loss": 0.016421888023614883, "grad_norm": 0.10380324721336365, "global_step": 44529, "epoch": 420, "lr": 6.170395641295123e-06} +{"train_loss": 0.013456413522362709, "grad_norm": 0.06993129849433899, "global_step": 44530, "epoch": 420, "lr": 6.1689694525511765e-06} +{"train_loss": 0.014582169242203236, "grad_norm": 0.07887911796569824, "global_step": 44531, "epoch": 420, "lr": 6.1675434178104005e-06} +{"train_loss": 0.01608777791261673, "grad_norm": 0.0976436585187912, "global_step": 44532, "epoch": 420, "lr": 6.16611753707782e-06} +{"train_loss": 0.01359599083662033, "grad_norm": 0.10951263457536697, "global_step": 44533, "epoch": 420, "lr": 6.164691810358442e-06} +{"train_loss": 0.014824205078184605, "grad_norm": 0.0632912740111351, "global_step": 44534, "epoch": 420, "lr": 6.163266237657267e-06} +{"train_loss": 0.015450265258550644, "grad_norm": 0.09910939633846283, "global_step": 44535, "epoch": 420, "lr": 6.161840818979309e-06} +{"train_loss": 0.01481616124510765, "grad_norm": 0.066957488656044, "global_step": 44536, "epoch": 420, "lr": 6.1604155543296015e-06} +{"train_loss": 0.014658736065030098, "grad_norm": 0.06337404996156693, "global_step": 44537, "epoch": 420, "lr": 6.158990443713103e-06} +{"train_loss": 0.0160115584731102, "grad_norm": 0.05547253042459488, "global_step": 44538, "epoch": 420, "lr": 6.157565487134848e-06} +{"train_loss": 0.014079570770263672, "grad_norm": 0.17251569032669067, "global_step": 44539, "epoch": 420, "lr": 6.1561406845998535e-06} +{"train_loss": 0.015970781445503235, "grad_norm": 0.08389641344547272, "global_step": 44540, "epoch": 420, "lr": 6.154716036113101e-06} +{"train_loss": 0.013017231598496437, "grad_norm": 0.07962438464164734, "global_step": 44541, "epoch": 420, "lr": 6.153291541679612e-06} +{"train_loss": 0.014079812914133072, "grad_norm": 0.07365819066762924, "global_step": 44542, "epoch": 420, "lr": 6.151867201304395e-06} +{"train_loss": 0.013893804512917995, "grad_norm": 0.12915228307247162, "global_step": 44543, "epoch": 420, "lr": 6.1504430149924465e-06} +{"train_loss": 0.014285517856478691, "grad_norm": 0.13665272295475006, "global_step": 44544, "epoch": 420, "lr": 6.149018982748761e-06} +{"train_loss": 0.015231698751449585, "grad_norm": 0.07060578465461731, "global_step": 44545, "epoch": 420, "lr": 6.147595104578358e-06} +{"train_loss": 0.01410752721130848, "grad_norm": 0.08462947607040405, "global_step": 44546, "epoch": 420, "lr": 6.146171380486243e-06} +{"train_loss": 0.012329253368079662, "grad_norm": 0.10687882453203201, "global_step": 44547, "epoch": 420, "lr": 6.144747810477397e-06} +{"train_loss": 0.015097696334123611, "grad_norm": 0.06535104662179947, "global_step": 44548, "epoch": 420, "lr": 6.143324394556848e-06} +{"train_loss": 0.014487843960523605, "grad_norm": 0.07732342183589935, "global_step": 44549, "epoch": 420, "lr": 6.141901132729571e-06} +{"train_loss": 0.014538243412971497, "grad_norm": 0.08545278012752533, "global_step": 44550, "epoch": 420, "lr": 6.14047802500059e-06} +{"train_loss": 0.01544745173305273, "grad_norm": 0.09270132333040237, "global_step": 44551, "epoch": 420, "lr": 6.139055071374889e-06} +{"train_loss": 0.014598771929740906, "grad_norm": 0.08506270498037338, "global_step": 44552, "epoch": 420, "lr": 6.1376322718574745e-06} +{"train_loss": 0.015736203640699387, "grad_norm": 0.06309137493371964, "global_step": 44553, "epoch": 420, "lr": 6.136209626453349e-06} +{"train_loss": 0.013797091320157051, "grad_norm": 0.06552736461162567, "global_step": 44554, "epoch": 420, "lr": 6.134787135167513e-06} +{"train_loss": 0.01412212010473013, "grad_norm": 0.08863046765327454, "global_step": 44555, "epoch": 420, "lr": 6.133364798004948e-06} +{"train_loss": 0.015485694631934166, "grad_norm": 0.11050920933485031, "global_step": 44556, "epoch": 420, "lr": 6.1319426149706704e-06} +{"train_loss": 0.017027810215950012, "grad_norm": 0.12172485888004303, "global_step": 44557, "epoch": 420, "lr": 6.130520586069655e-06} +{"train_loss": 0.014632167294621468, "grad_norm": 0.08565099537372589, "global_step": 44558, "epoch": 420, "lr": 6.1290987113069186e-06} +{"train_loss": 0.014328531920909882, "grad_norm": 0.08947164565324783, "global_step": 44559, "epoch": 420, "lr": 6.127676990687453e-06} +{"train_loss": 0.0142215546220541, "grad_norm": 0.14868557453155518, "global_step": 44560, "epoch": 420, "lr": 6.1262554242162596e-06} +{"train_loss": 0.01218029111623764, "grad_norm": 0.13738782703876495, "global_step": 44561, "epoch": 420, "lr": 6.124834011898306e-06} +{"train_loss": 0.016654711216688156, "grad_norm": 0.08484841138124466, "global_step": 44562, "epoch": 420, "lr": 6.123412753738611e-06} +{"train_loss": 0.01223829947412014, "grad_norm": 0.08664217591285706, "global_step": 44563, "epoch": 420, "lr": 6.12199164974217e-06} +{"train_loss": 0.01839781552553177, "grad_norm": 0.12004441022872925, "global_step": 44564, "epoch": 420, "lr": 6.120570699913958e-06} +{"train_loss": 0.014811726287007332, "grad_norm": 0.09124469757080078, "global_step": 44565, "epoch": 420, "lr": 6.1191499042589915e-06} +{"train_loss": 0.01811002939939499, "grad_norm": 0.08631338179111481, "global_step": 44566, "epoch": 420, "lr": 6.1177292627822344e-06} +{"train_loss": 0.014110114425420761, "grad_norm": 0.1073671206831932, "global_step": 44567, "epoch": 420, "lr": 6.116308775488705e-06} +{"train_loss": 0.016285501420497894, "grad_norm": 0.10697925835847855, "global_step": 44568, "epoch": 420, "lr": 6.11488844238337e-06} +{"train_loss": 0.018001843243837357, "grad_norm": 0.08055569231510162, "global_step": 44569, "epoch": 420, "lr": 6.113468263471234e-06} +{"train_loss": 0.018741870298981667, "grad_norm": 0.12816700339317322, "global_step": 44570, "epoch": 420, "lr": 6.11204823875729e-06} +{"train_loss": 0.015450704842805862, "grad_norm": 0.08500142395496368, "global_step": 44571, "epoch": 420, "lr": 6.110628368246529e-06} +{"train_loss": 0.017345519736409187, "grad_norm": 0.09249824285507202, "global_step": 44572, "epoch": 420, "lr": 6.109208651943921e-06} +{"train_loss": 0.015518061816692352, "grad_norm": 0.10021138936281204, "global_step": 44573, "epoch": 420, "lr": 6.107789089854471e-06} +{"train_loss": 0.01597174070775509, "grad_norm": 0.10248372703790665, "global_step": 44574, "epoch": 420, "lr": 6.10636968198316e-06} +{"train_loss": 0.015233693644404411, "grad_norm": 0.10370314866304398, "global_step": 44575, "epoch": 420, "lr": 6.104950428334971e-06} +{"train_loss": 0.013038493692874908, "grad_norm": 0.06530198454856873, "global_step": 44576, "epoch": 420, "lr": 6.103531328914897e-06} +{"train_loss": 0.015971269458532333, "grad_norm": 0.07641280442476273, "global_step": 44577, "epoch": 420, "lr": 6.102112383727948e-06} +{"train_loss": 0.015015669167041779, "grad_norm": 0.0738062933087349, "global_step": 44578, "epoch": 420, "lr": 6.100693592779061e-06} +{"train_loss": 0.01638651080429554, "grad_norm": 0.07923955470323563, "global_step": 44579, "epoch": 420, "lr": 6.099274956073248e-06} +{"train_loss": 0.015142329037189484, "grad_norm": 0.07717661559581757, "global_step": 44580, "epoch": 420, "lr": 6.097856473615504e-06} +{"train_loss": 0.015161195769906044, "grad_norm": 0.06400489062070847, "global_step": 44581, "epoch": 420, "lr": 6.0964381454107826e-06} +{"train_loss": 0.01391720399260521, "grad_norm": 0.12153996527194977, "global_step": 44582, "epoch": 420, "lr": 6.095019971464089e-06} +{"train_loss": 0.015302671119570732, "grad_norm": 0.07052762806415558, "global_step": 44583, "epoch": 420, "lr": 6.093601951780409e-06} +{"train_loss": 0.01236571092158556, "grad_norm": 0.10320118814706802, "global_step": 44584, "epoch": 420, "lr": 6.092184086364721e-06} +{"train_loss": 0.014016895554959774, "grad_norm": 0.139428049325943, "global_step": 44585, "epoch": 420, "lr": 6.09076637522199e-06} +{"train_loss": 0.015632741153240204, "grad_norm": 0.06724068522453308, "global_step": 44586, "epoch": 420, "lr": 6.08934881835721e-06} +{"train_loss": 0.017596503719687462, "grad_norm": 0.08555963635444641, "global_step": 44587, "epoch": 420, "lr": 6.087931415775377e-06} +{"train_loss": 0.016410084441304207, "grad_norm": 0.07250625640153885, "global_step": 44588, "epoch": 420, "lr": 6.086514167481439e-06} +{"train_loss": 0.014753911644220352, "grad_norm": 0.09015066176652908, "global_step": 44589, "epoch": 420, "lr": 6.085097073480406e-06} +{"train_loss": 0.015253566205501556, "grad_norm": 0.0703919529914856, "global_step": 44590, "epoch": 420, "lr": 6.083680133777236e-06} +{"train_loss": 0.014994043856859207, "grad_norm": 0.07172255963087082, "global_step": 44591, "epoch": 420, "lr": 6.082263348376921e-06} +{"train_loss": 0.017935853451490402, "grad_norm": 0.0790194496512413, "global_step": 44592, "epoch": 420, "lr": 6.080846717284427e-06} +{"train_loss": 0.012842636555433273, "grad_norm": 0.057616859674453735, "global_step": 44593, "epoch": 420, "lr": 6.079430240504735e-06} +{"train_loss": 0.017680490389466286, "grad_norm": 0.14152337610721588, "global_step": 44594, "epoch": 420, "lr": 6.078013918042835e-06} +{"train_loss": 0.016374103724956512, "grad_norm": 0.06687402725219727, "global_step": 44595, "epoch": 420, "lr": 6.076597749903695e-06} +{"train_loss": 0.01595585234463215, "grad_norm": 0.06470353156328201, "global_step": 44596, "epoch": 420, "lr": 6.075181736092278e-06} +{"train_loss": 0.013278057798743248, "grad_norm": 0.052969057112932205, "global_step": 44597, "epoch": 420, "lr": 6.073765876613579e-06} +{"train_loss": 0.014561348594725132, "grad_norm": 0.08615221083164215, "global_step": 44598, "epoch": 420, "lr": 6.072350171472557e-06} +{"train_loss": 0.014484546147286892, "grad_norm": 0.05906239151954651, "global_step": 44599, "epoch": 420, "lr": 6.07093462067419e-06} +{"train_loss": 0.014126697555184364, "grad_norm": 0.0840616226196289, "global_step": 44600, "epoch": 420, "lr": 6.069519224223463e-06} +{"train_loss": 0.011901790276169777, "grad_norm": 0.06320314109325409, "global_step": 44601, "epoch": 420, "lr": 6.068103982125345e-06} +{"train_loss": 0.015179621055722237, "grad_norm": 0.06710722297430038, "global_step": 44602, "epoch": 420, "lr": 6.066688894384786e-06} +{"train_loss": 0.015611773356795311, "grad_norm": 0.07181628048419952, "global_step": 44603, "epoch": 420, "lr": 6.065273961006784e-06} +{"train_loss": 0.016906725242733955, "grad_norm": 0.06927941739559174, "global_step": 44604, "epoch": 420, "lr": 6.063859181996312e-06} +{"train_loss": 0.013926571235060692, "grad_norm": 0.28165140748023987, "global_step": 44605, "epoch": 420, "lr": 6.062444557358316e-06} +{"train_loss": 0.014229567721486092, "grad_norm": 0.05733805149793625, "global_step": 44606, "epoch": 420, "lr": 6.061030087097796e-06} +{"train_loss": 0.015844780951738358, "grad_norm": 0.08251258730888367, "global_step": 44607, "epoch": 420, "lr": 6.059615771219696e-06} +{"train_loss": 0.015299029648303986, "grad_norm": 0.05892445147037506, "global_step": 44608, "epoch": 420, "lr": 6.058201609729008e-06} +{"train_loss": 0.01458040252327919, "grad_norm": 0.08457788825035095, "global_step": 44609, "epoch": 420, "lr": 6.056787602630681e-06} +{"train_loss": 0.016340279951691628, "grad_norm": 0.09342151135206223, "global_step": 44610, "epoch": 420, "lr": 6.055373749929693e-06} +{"train_loss": 0.015157547779381275, "grad_norm": 0.07170122861862183, "global_step": 44611, "epoch": 420, "lr": 6.053960051631019e-06} +{"train_loss": 0.013935812748968601, "grad_norm": 0.05839579552412033, "global_step": 44612, "epoch": 420, "lr": 6.052546507739615e-06} +{"train_loss": 0.015459883958101273, "grad_norm": 0.09519504755735397, "global_step": 44613, "epoch": 420, "lr": 6.051133118260444e-06} +{"train_loss": 0.01462301891297102, "grad_norm": 0.09561461955308914, "global_step": 44614, "epoch": 420, "lr": 6.0497198831984845e-06} +{"train_loss": 0.014712626114487648, "grad_norm": 0.0800851434469223, "global_step": 44615, "epoch": 420, "lr": 6.0483068025586895e-06} +{"train_loss": 0.014810264110565186, "grad_norm": 0.09995663911104202, "global_step": 44616, "epoch": 420, "lr": 6.046893876346027e-06} +{"train_loss": 0.0152351763099432, "grad_norm": 0.08370621502399445, "global_step": 44617, "epoch": 420, "lr": 6.045481104565481e-06} +{"train_loss": 0.015711763873696327, "grad_norm": 0.20512987673282623, "global_step": 44618, "epoch": 420, "lr": 6.044068487221993e-06} +{"train_loss": 0.016254348680377007, "grad_norm": 0.07276353985071182, "global_step": 44619, "epoch": 420, "lr": 6.042656024320525e-06} +{"train_loss": 0.013639761134982109, "grad_norm": 0.05790119245648384, "global_step": 44620, "epoch": 420, "lr": 6.0412437158660464e-06} +{"train_loss": 0.013395329937338829, "grad_norm": 0.14131836593151093, "global_step": 44621, "epoch": 420, "lr": 6.0398315618635295e-06} +{"train_loss": 0.016216348856687546, "grad_norm": 0.09757186472415924, "global_step": 44622, "epoch": 420, "lr": 6.038419562317915e-06} +{"train_loss": 0.012833252549171448, "grad_norm": 0.07573466002941132, "global_step": 44623, "epoch": 420, "lr": 6.037007717234177e-06} +{"train_loss": 0.014326278120279312, "grad_norm": 0.08503689616918564, "global_step": 44624, "epoch": 420, "lr": 6.0355960266172906e-06} +{"train_loss": 0.014978699605771393, "grad_norm": 0.09272823482751846, "global_step": 44625, "epoch": 420, "lr": 6.034184490472195e-06, "train_batch_mse_error": 0.013389943167567253, "val_loss": 0.019176490604877472} +{"train_loss": 0.01253604982048273, "grad_norm": 0.07511212676763535, "global_step": 44626, "epoch": 421, "lr": 6.0327731088038465e-06} +{"train_loss": 0.014159563928842545, "grad_norm": 0.05682051554322243, "global_step": 44627, "epoch": 421, "lr": 6.031361881617209e-06} +{"train_loss": 0.013216981664299965, "grad_norm": 0.0701933428645134, "global_step": 44628, "epoch": 421, "lr": 6.029950808917256e-06} +{"train_loss": 0.014485547319054604, "grad_norm": 0.0688304528594017, "global_step": 44629, "epoch": 421, "lr": 6.028539890708929e-06} +{"train_loss": 0.013062766753137112, "grad_norm": 0.07177262008190155, "global_step": 44630, "epoch": 421, "lr": 6.027129126997194e-06} +{"train_loss": 0.016203196719288826, "grad_norm": 0.062367022037506104, "global_step": 44631, "epoch": 421, "lr": 6.0257185177869935e-06} +{"train_loss": 0.013826999813318253, "grad_norm": 0.07393153011798859, "global_step": 44632, "epoch": 421, "lr": 6.024308063083312e-06} +{"train_loss": 0.014747422188520432, "grad_norm": 0.07462558150291443, "global_step": 44633, "epoch": 421, "lr": 6.0228977628910665e-06} +{"train_loss": 0.015298874117434025, "grad_norm": 0.05925934389233589, "global_step": 44634, "epoch": 421, "lr": 6.021487617215238e-06} +{"train_loss": 0.013178203254938126, "grad_norm": 0.07094540446996689, "global_step": 44635, "epoch": 421, "lr": 6.020077626060788e-06} +{"train_loss": 0.014519513584673405, "grad_norm": 0.178517684340477, "global_step": 44636, "epoch": 421, "lr": 6.018667789432658e-06} +{"train_loss": 0.014510463923215866, "grad_norm": 0.06883189082145691, "global_step": 44637, "epoch": 421, "lr": 6.017258107335793e-06} +{"train_loss": 0.013329840265214443, "grad_norm": 0.09849707037210464, "global_step": 44638, "epoch": 421, "lr": 6.015848579775163e-06} +{"train_loss": 0.016774501651525497, "grad_norm": 0.08890710026025772, "global_step": 44639, "epoch": 421, "lr": 6.014439206755706e-06} +{"train_loss": 0.01719864085316658, "grad_norm": 0.10051983594894409, "global_step": 44640, "epoch": 421, "lr": 6.013029988282381e-06} +{"train_loss": 0.016831470653414726, "grad_norm": 0.06789040565490723, "global_step": 44641, "epoch": 421, "lr": 6.011620924360145e-06} +{"train_loss": 0.017533916980028152, "grad_norm": 0.07646644115447998, "global_step": 44642, "epoch": 421, "lr": 6.0102120149939485e-06} +{"train_loss": 0.016375843435525894, "grad_norm": 0.08491118997335434, "global_step": 44643, "epoch": 421, "lr": 6.008803260188723e-06} +{"train_loss": 0.013540957123041153, "grad_norm": 0.0681607648730278, "global_step": 44644, "epoch": 421, "lr": 6.007394659949428e-06} +{"train_loss": 0.015880288556218147, "grad_norm": 0.09253107756376266, "global_step": 44645, "epoch": 421, "lr": 6.0059862142810345e-06} +{"train_loss": 0.015520105138421059, "grad_norm": 0.08652787655591965, "global_step": 44646, "epoch": 421, "lr": 6.0045779231884595e-06} +{"train_loss": 0.013721831142902374, "grad_norm": 0.06457656621932983, "global_step": 44647, "epoch": 421, "lr": 6.00316978667666e-06} +{"train_loss": 0.014963243156671524, "grad_norm": 0.07847736775875092, "global_step": 44648, "epoch": 421, "lr": 6.001761804750605e-06} +{"train_loss": 0.014739634469151497, "grad_norm": 0.0915629044175148, "global_step": 44649, "epoch": 421, "lr": 6.0003539774152185e-06} +{"train_loss": 0.014367202296853065, "grad_norm": 0.07079539448022842, "global_step": 44650, "epoch": 421, "lr": 5.998946304675446e-06} +{"train_loss": 0.016790591180324554, "grad_norm": 0.11076760292053223, "global_step": 44651, "epoch": 421, "lr": 5.997538786536244e-06} +{"train_loss": 0.013751329854130745, "grad_norm": 0.06130911037325859, "global_step": 44652, "epoch": 421, "lr": 5.99613142300256e-06} +{"train_loss": 0.015024869702756405, "grad_norm": 0.08370501548051834, "global_step": 44653, "epoch": 421, "lr": 5.9947242140793224e-06} +{"train_loss": 0.014773692935705185, "grad_norm": 0.08384115993976593, "global_step": 44654, "epoch": 421, "lr": 5.9933171597715e-06} +{"train_loss": 0.01633254997432232, "grad_norm": 0.09906802326440811, "global_step": 44655, "epoch": 421, "lr": 5.991910260084022e-06} +{"train_loss": 0.015609854832291603, "grad_norm": 0.0618484690785408, "global_step": 44656, "epoch": 421, "lr": 5.990503515021823e-06} +{"train_loss": 0.013783464208245277, "grad_norm": 0.059718672186136246, "global_step": 44657, "epoch": 421, "lr": 5.9890969245898545e-06} +{"train_loss": 0.015553085133433342, "grad_norm": 0.09268582612276077, "global_step": 44658, "epoch": 421, "lr": 5.987690488793074e-06} +{"train_loss": 0.014211639761924744, "grad_norm": 0.06688426434993744, "global_step": 44659, "epoch": 421, "lr": 5.986284207636405e-06} +{"train_loss": 0.013962415978312492, "grad_norm": 0.10087157785892487, "global_step": 44660, "epoch": 421, "lr": 5.984878081124779e-06} +{"train_loss": 0.012195258401334286, "grad_norm": 0.06951494514942169, "global_step": 44661, "epoch": 421, "lr": 5.983472109263155e-06} +{"train_loss": 0.014467228204011917, "grad_norm": 0.0695185661315918, "global_step": 44662, "epoch": 421, "lr": 5.982066292056476e-06} +{"train_loss": 0.012807287275791168, "grad_norm": 0.061197783797979355, "global_step": 44663, "epoch": 421, "lr": 5.98066062950966e-06} +{"train_loss": 0.014673550613224506, "grad_norm": 0.11936615407466888, "global_step": 44664, "epoch": 421, "lr": 5.979255121627664e-06} +{"train_loss": 0.014906889759004116, "grad_norm": 0.056092604994773865, "global_step": 44665, "epoch": 421, "lr": 5.977849768415428e-06} +{"train_loss": 0.015585877932608128, "grad_norm": 0.07070588320493698, "global_step": 44666, "epoch": 421, "lr": 5.976444569877887e-06} +{"train_loss": 0.017062954604625702, "grad_norm": 0.17031040787696838, "global_step": 44667, "epoch": 421, "lr": 5.9750395260199595e-06} +{"train_loss": 0.014006524346768856, "grad_norm": 0.0683780312538147, "global_step": 44668, "epoch": 421, "lr": 5.973634636846598e-06} +{"train_loss": 0.01479578111320734, "grad_norm": 0.07058173418045044, "global_step": 44669, "epoch": 421, "lr": 5.9722299023627525e-06} +{"train_loss": 0.013468803837895393, "grad_norm": 0.12039943784475327, "global_step": 44670, "epoch": 421, "lr": 5.970825322573331e-06} +{"train_loss": 0.01566522940993309, "grad_norm": 0.11346009373664856, "global_step": 44671, "epoch": 421, "lr": 5.9694208974832914e-06} +{"train_loss": 0.013957055285573006, "grad_norm": 0.08349942415952682, "global_step": 44672, "epoch": 421, "lr": 5.968016627097556e-06} +{"train_loss": 0.014082245528697968, "grad_norm": 0.08098981529474258, "global_step": 44673, "epoch": 421, "lr": 5.966612511421055e-06} +{"train_loss": 0.014554549008607864, "grad_norm": 0.06670982390642166, "global_step": 44674, "epoch": 421, "lr": 5.965208550458723e-06} +{"train_loss": 0.015624800696969032, "grad_norm": 0.07873009890317917, "global_step": 44675, "epoch": 421, "lr": 5.963804744215512e-06} +{"train_loss": 0.015455363318324089, "grad_norm": 0.11646223813295364, "global_step": 44676, "epoch": 421, "lr": 5.962401092696329e-06} +{"train_loss": 0.014732060953974724, "grad_norm": 0.055099233984947205, "global_step": 44677, "epoch": 421, "lr": 5.960997595906131e-06} +{"train_loss": 0.014872581698000431, "grad_norm": 0.0926792174577713, "global_step": 44678, "epoch": 421, "lr": 5.95959425384982e-06} +{"train_loss": 0.013744333758950233, "grad_norm": 0.06795100122690201, "global_step": 44679, "epoch": 421, "lr": 5.958191066532354e-06} +{"train_loss": 0.014539649710059166, "grad_norm": 0.07107861340045929, "global_step": 44680, "epoch": 421, "lr": 5.9567880339586435e-06} +{"train_loss": 0.013406509533524513, "grad_norm": 0.09426827728748322, "global_step": 44681, "epoch": 421, "lr": 5.955385156133625e-06} +{"train_loss": 0.014916274696588516, "grad_norm": 0.07862111181020737, "global_step": 44682, "epoch": 421, "lr": 5.9539824330622385e-06} +{"train_loss": 0.016717175021767616, "grad_norm": 0.07192949950695038, "global_step": 44683, "epoch": 421, "lr": 5.9525798647494024e-06} +{"train_loss": 0.016949797049164772, "grad_norm": 0.08244878053665161, "global_step": 44684, "epoch": 421, "lr": 5.951177451200035e-06} +{"train_loss": 0.017523374408483505, "grad_norm": 0.1625758707523346, "global_step": 44685, "epoch": 421, "lr": 5.949775192419071e-06} +{"train_loss": 0.01367977261543274, "grad_norm": 0.11260316520929337, "global_step": 44686, "epoch": 421, "lr": 5.948373088411452e-06} +{"train_loss": 0.0157400444149971, "grad_norm": 0.11294777691364288, "global_step": 44687, "epoch": 421, "lr": 5.946971139182084e-06} +{"train_loss": 0.014354727230966091, "grad_norm": 0.06299242377281189, "global_step": 44688, "epoch": 421, "lr": 5.945569344735896e-06} +{"train_loss": 0.01574181206524372, "grad_norm": 0.09579651802778244, "global_step": 44689, "epoch": 421, "lr": 5.944167705077835e-06} +{"train_loss": 0.01630397140979767, "grad_norm": 0.08742597699165344, "global_step": 44690, "epoch": 421, "lr": 5.9427662202128085e-06} +{"train_loss": 0.016237841919064522, "grad_norm": 0.12376867979764938, "global_step": 44691, "epoch": 421, "lr": 5.9413648901457285e-06} +{"train_loss": 0.01581145077943802, "grad_norm": 0.06778208911418915, "global_step": 44692, "epoch": 421, "lr": 5.93996371488153e-06} +{"train_loss": 0.016626683995127678, "grad_norm": 0.08797550946474075, "global_step": 44693, "epoch": 421, "lr": 5.9385626944251535e-06} +{"train_loss": 0.018324438482522964, "grad_norm": 0.07718545198440552, "global_step": 44694, "epoch": 421, "lr": 5.93716182878149e-06} +{"train_loss": 0.013005156069993973, "grad_norm": 0.12790890038013458, "global_step": 44695, "epoch": 421, "lr": 5.935761117955491e-06} +{"train_loss": 0.013929564505815506, "grad_norm": 0.13326792418956757, "global_step": 44696, "epoch": 421, "lr": 5.934360561952068e-06} +{"train_loss": 0.01301560364663601, "grad_norm": 0.08114328235387802, "global_step": 44697, "epoch": 421, "lr": 5.932960160776119e-06} +{"train_loss": 0.015001092106103897, "grad_norm": 0.06599850952625275, "global_step": 44698, "epoch": 421, "lr": 5.931559914432588e-06} +{"train_loss": 0.013885210268199444, "grad_norm": 0.09645888209342957, "global_step": 44699, "epoch": 421, "lr": 5.9301598229264064e-06} +{"train_loss": 0.013196185231208801, "grad_norm": 0.06278163939714432, "global_step": 44700, "epoch": 421, "lr": 5.928759886262464e-06} +{"train_loss": 0.016860056668519974, "grad_norm": 0.07339047640562057, "global_step": 44701, "epoch": 421, "lr": 5.927360104445701e-06} +{"train_loss": 0.014270218089222908, "grad_norm": 0.069150410592556, "global_step": 44702, "epoch": 421, "lr": 5.925960477481018e-06} +{"train_loss": 0.015337606891989708, "grad_norm": 0.17172612249851227, "global_step": 44703, "epoch": 421, "lr": 5.9245610053733525e-06} +{"train_loss": 0.013423118740320206, "grad_norm": 0.11779976636171341, "global_step": 44704, "epoch": 421, "lr": 5.923161688127604e-06} +{"train_loss": 0.013599539175629616, "grad_norm": 0.053615618497133255, "global_step": 44705, "epoch": 421, "lr": 5.921762525748698e-06} +{"train_loss": 0.014749746769666672, "grad_norm": 0.08447197079658508, "global_step": 44706, "epoch": 421, "lr": 5.920363518241556e-06} +{"train_loss": 0.01391607616096735, "grad_norm": 0.08944177627563477, "global_step": 44707, "epoch": 421, "lr": 5.918964665611087e-06} +{"train_loss": 0.016581492498517036, "grad_norm": 0.14833176136016846, "global_step": 44708, "epoch": 421, "lr": 5.917565967862199e-06} +{"train_loss": 0.01811143569648266, "grad_norm": 0.07993699610233307, "global_step": 44709, "epoch": 421, "lr": 5.916167424999808e-06} +{"train_loss": 0.014729108661413193, "grad_norm": 0.11126967519521713, "global_step": 44710, "epoch": 421, "lr": 5.914769037028845e-06} +{"train_loss": 0.01616200990974903, "grad_norm": 0.07984676957130432, "global_step": 44711, "epoch": 421, "lr": 5.913370803954205e-06} +{"train_loss": 0.014039669185876846, "grad_norm": 0.08237475901842117, "global_step": 44712, "epoch": 421, "lr": 5.911972725780812e-06} +{"train_loss": 0.014746787026524544, "grad_norm": 0.08528643846511841, "global_step": 44713, "epoch": 421, "lr": 5.91057480251358e-06} +{"train_loss": 0.011832451447844505, "grad_norm": 0.0656452402472496, "global_step": 44714, "epoch": 421, "lr": 5.909177034157398e-06} +{"train_loss": 0.016147229820489883, "grad_norm": 0.12014291435480118, "global_step": 44715, "epoch": 421, "lr": 5.907779420717196e-06} +{"train_loss": 0.014192972332239151, "grad_norm": 0.15654230117797852, "global_step": 44716, "epoch": 421, "lr": 5.9063819621978935e-06} +{"train_loss": 0.016404051333665848, "grad_norm": 0.08681891858577728, "global_step": 44717, "epoch": 421, "lr": 5.9049846586043735e-06} +{"train_loss": 0.01626504771411419, "grad_norm": 0.07762826234102249, "global_step": 44718, "epoch": 421, "lr": 5.903587509941577e-06} +{"train_loss": 0.016241438686847687, "grad_norm": 0.0901791974902153, "global_step": 44719, "epoch": 421, "lr": 5.902190516214384e-06} +{"train_loss": 0.01503563392907381, "grad_norm": 0.09895572066307068, "global_step": 44720, "epoch": 421, "lr": 5.900793677427724e-06} +{"train_loss": 0.014715529978275299, "grad_norm": 0.07091198116540909, "global_step": 44721, "epoch": 421, "lr": 5.899396993586493e-06} +{"train_loss": 0.016455359756946564, "grad_norm": 0.09672091901302338, "global_step": 44722, "epoch": 421, "lr": 5.898000464695596e-06} +{"train_loss": 0.015193495899438858, "grad_norm": 0.05902475491166115, "global_step": 44723, "epoch": 421, "lr": 5.89660409075996e-06} +{"train_loss": 0.014759834855794907, "grad_norm": 0.12833836674690247, "global_step": 44724, "epoch": 421, "lr": 5.895207871784475e-06} +{"train_loss": 0.014947225339710712, "grad_norm": 0.07820571213960648, "global_step": 44725, "epoch": 421, "lr": 5.893811807774041e-06} +{"train_loss": 0.014673089608550072, "grad_norm": 0.06475087255239487, "global_step": 44726, "epoch": 421, "lr": 5.8924158987335665e-06} +{"train_loss": 0.015027911402285099, "grad_norm": 0.07270577549934387, "global_step": 44727, "epoch": 421, "lr": 5.891020144667975e-06} +{"train_loss": 0.01661975309252739, "grad_norm": 0.11328265815973282, "global_step": 44728, "epoch": 421, "lr": 5.889624545582145e-06} +{"train_loss": 0.016422340646386147, "grad_norm": 0.10875922441482544, "global_step": 44729, "epoch": 421, "lr": 5.888229101480991e-06} +{"train_loss": 0.014518226496875286, "grad_norm": 0.11314628273248672, "global_step": 44730, "epoch": 421, "lr": 5.8868338123694355e-06} +{"train_loss": 0.014970982570271447, "grad_norm": 0.08046486228704453, "global_step": 44731, "epoch": 421, "lr": 5.885438678252342e-06, "val_loss": 0.019568635150790215} +{"train_loss": 0.015391917899250984, "grad_norm": 0.08126667886972427, "global_step": 44732, "epoch": 422, "lr": 5.8840436991346335e-06} +{"train_loss": 0.014741459861397743, "grad_norm": 0.08739491552114487, "global_step": 44733, "epoch": 422, "lr": 5.882648875021207e-06} +{"train_loss": 0.01554589718580246, "grad_norm": 0.08836632966995239, "global_step": 44734, "epoch": 422, "lr": 5.881254205916981e-06} +{"train_loss": 0.014245802536606789, "grad_norm": 0.061787087470293045, "global_step": 44735, "epoch": 422, "lr": 5.879859691826828e-06} +{"train_loss": 0.012688105925917625, "grad_norm": 0.1686955839395523, "global_step": 44736, "epoch": 422, "lr": 5.8784653327556675e-06} +{"train_loss": 0.015391713939607143, "grad_norm": 0.11580225825309753, "global_step": 44737, "epoch": 422, "lr": 5.877071128708394e-06} +{"train_loss": 0.015045645646750927, "grad_norm": 0.0536968857049942, "global_step": 44738, "epoch": 422, "lr": 5.875677079689901e-06} +{"train_loss": 0.015068739652633667, "grad_norm": 0.07547054439783096, "global_step": 44739, "epoch": 422, "lr": 5.874283185705081e-06} +{"train_loss": 0.013471694663167, "grad_norm": 0.1008041575551033, "global_step": 44740, "epoch": 422, "lr": 5.87288944675885e-06} +{"train_loss": 0.01308630034327507, "grad_norm": 0.08048078417778015, "global_step": 44741, "epoch": 422, "lr": 5.871495862856091e-06} +{"train_loss": 0.01401603501290083, "grad_norm": 0.10566994547843933, "global_step": 44742, "epoch": 422, "lr": 5.870102434001712e-06} +{"train_loss": 0.01641152612864971, "grad_norm": 0.0879969373345375, "global_step": 44743, "epoch": 422, "lr": 5.868709160200591e-06} +{"train_loss": 0.01567632332444191, "grad_norm": 0.11303270608186722, "global_step": 44744, "epoch": 422, "lr": 5.867316041457643e-06} +{"train_loss": 0.015149434097111225, "grad_norm": 0.15807472169399261, "global_step": 44745, "epoch": 422, "lr": 5.865923077777746e-06} +{"train_loss": 0.017297400161623955, "grad_norm": 0.06730323284864426, "global_step": 44746, "epoch": 422, "lr": 5.8645302691657965e-06} +{"train_loss": 0.01583542488515377, "grad_norm": 0.12261077016592026, "global_step": 44747, "epoch": 422, "lr": 5.863137615626707e-06} +{"train_loss": 0.014953907579183578, "grad_norm": 0.0714036151766777, "global_step": 44748, "epoch": 422, "lr": 5.861745117165362e-06} +{"train_loss": 0.018747590482234955, "grad_norm": 0.1554800122976303, "global_step": 44749, "epoch": 422, "lr": 5.860352773786631e-06} +{"train_loss": 0.01525520533323288, "grad_norm": 0.060038331896066666, "global_step": 44750, "epoch": 422, "lr": 5.858960585495432e-06} +{"train_loss": 0.012777132913470268, "grad_norm": 0.07118505239486694, "global_step": 44751, "epoch": 422, "lr": 5.857568552296655e-06} +{"train_loss": 0.015259986743330956, "grad_norm": 0.11000791937112808, "global_step": 44752, "epoch": 422, "lr": 5.856176674195174e-06} +{"train_loss": 0.015095251612365246, "grad_norm": 0.12468259036540985, "global_step": 44753, "epoch": 422, "lr": 5.854784951195896e-06} +{"train_loss": 0.0160398930311203, "grad_norm": 0.07732920348644257, "global_step": 44754, "epoch": 422, "lr": 5.853393383303724e-06} +{"train_loss": 0.012972919270396233, "grad_norm": 0.13196256756782532, "global_step": 44755, "epoch": 422, "lr": 5.852001970523502e-06} +{"train_loss": 0.018000029027462006, "grad_norm": 0.07932672649621964, "global_step": 44756, "epoch": 422, "lr": 5.8506107128601495e-06} +{"train_loss": 0.016995107755064964, "grad_norm": 0.07303038239479065, "global_step": 44757, "epoch": 422, "lr": 5.849219610318563e-06} +{"train_loss": 0.016376186162233353, "grad_norm": 0.0650150403380394, "global_step": 44758, "epoch": 422, "lr": 5.84782866290361e-06} +{"train_loss": 0.016820523887872696, "grad_norm": 0.09290513396263123, "global_step": 44759, "epoch": 422, "lr": 5.846437870620192e-06} +{"train_loss": 0.018483906984329224, "grad_norm": 0.07184003293514252, "global_step": 44760, "epoch": 422, "lr": 5.845047233473178e-06} +{"train_loss": 0.014308644458651543, "grad_norm": 0.09606476128101349, "global_step": 44761, "epoch": 422, "lr": 5.84365675146748e-06} +{"train_loss": 0.01603236421942711, "grad_norm": 0.10129868239164352, "global_step": 44762, "epoch": 422, "lr": 5.842266424607956e-06} +{"train_loss": 0.014379318803548813, "grad_norm": 0.05304442718625069, "global_step": 44763, "epoch": 422, "lr": 5.840876252899502e-06} +{"train_loss": 0.015516098588705063, "grad_norm": 0.07995064556598663, "global_step": 44764, "epoch": 422, "lr": 5.8394862363470194e-06} +{"train_loss": 0.014274423941969872, "grad_norm": 0.0769307017326355, "global_step": 44765, "epoch": 422, "lr": 5.838096374955376e-06} +{"train_loss": 0.016525039449334145, "grad_norm": 0.09853485226631165, "global_step": 44766, "epoch": 422, "lr": 5.8367066687294415e-06} +{"train_loss": 0.012280753813683987, "grad_norm": 0.08203853666782379, "global_step": 44767, "epoch": 422, "lr": 5.835317117674121e-06} +{"train_loss": 0.014387699775397778, "grad_norm": 0.10858830064535141, "global_step": 44768, "epoch": 422, "lr": 5.833927721794297e-06} +{"train_loss": 0.014874571934342384, "grad_norm": 0.13593150675296783, "global_step": 44769, "epoch": 422, "lr": 5.832538481094829e-06} +{"train_loss": 0.015463548712432384, "grad_norm": 0.07305330783128738, "global_step": 44770, "epoch": 422, "lr": 5.8311493955806204e-06} +{"train_loss": 0.015453722327947617, "grad_norm": 0.07783785462379456, "global_step": 44771, "epoch": 422, "lr": 5.829760465256562e-06} +{"train_loss": 0.016031833365559578, "grad_norm": 0.06301789730787277, "global_step": 44772, "epoch": 422, "lr": 5.828371690127493e-06} +{"train_loss": 0.0130671551451087, "grad_norm": 0.0709514394402504, "global_step": 44773, "epoch": 422, "lr": 5.826983070198317e-06} +{"train_loss": 0.017796382308006287, "grad_norm": 0.08707079291343689, "global_step": 44774, "epoch": 422, "lr": 5.825594605473922e-06} +{"train_loss": 0.016125615686178207, "grad_norm": 0.11628162860870361, "global_step": 44775, "epoch": 422, "lr": 5.824206295959167e-06} +{"train_loss": 0.01636020839214325, "grad_norm": 0.06888959556818008, "global_step": 44776, "epoch": 422, "lr": 5.8228181416589425e-06} +{"train_loss": 0.012696091085672379, "grad_norm": 0.08703155815601349, "global_step": 44777, "epoch": 422, "lr": 5.821430142578132e-06} +{"train_loss": 0.01475679874420166, "grad_norm": 0.06258618086576462, "global_step": 44778, "epoch": 422, "lr": 5.820042298721601e-06} +{"train_loss": 0.013744844123721123, "grad_norm": 0.061068832874298096, "global_step": 44779, "epoch": 422, "lr": 5.818654610094221e-06} +{"train_loss": 0.01466378103941679, "grad_norm": 0.09926773607730865, "global_step": 44780, "epoch": 422, "lr": 5.817267076700872e-06} +{"train_loss": 0.015681326389312744, "grad_norm": 0.10127876698970795, "global_step": 44781, "epoch": 422, "lr": 5.815879698546439e-06} +{"train_loss": 0.014588050544261932, "grad_norm": 0.11096522212028503, "global_step": 44782, "epoch": 422, "lr": 5.814492475635785e-06} +{"train_loss": 0.012406769208610058, "grad_norm": 0.059339217841625214, "global_step": 44783, "epoch": 422, "lr": 5.813105407973796e-06} +{"train_loss": 0.015388354659080505, "grad_norm": 0.06781701743602753, "global_step": 44784, "epoch": 422, "lr": 5.811718495565327e-06} +{"train_loss": 0.01309177465736866, "grad_norm": 0.10082488507032394, "global_step": 44785, "epoch": 422, "lr": 5.8103317384152755e-06} +{"train_loss": 0.014234937727451324, "grad_norm": 0.10014011710882187, "global_step": 44786, "epoch": 422, "lr": 5.808945136528493e-06} +{"train_loss": 0.01449831947684288, "grad_norm": 0.07548397034406662, "global_step": 44787, "epoch": 422, "lr": 5.807558689909854e-06} +{"train_loss": 0.011857783421874046, "grad_norm": 0.08212258666753769, "global_step": 44788, "epoch": 422, "lr": 5.806172398564247e-06} +{"train_loss": 0.014173557981848717, "grad_norm": 0.07979381829500198, "global_step": 44789, "epoch": 422, "lr": 5.804786262496531e-06} +{"train_loss": 0.016133008524775505, "grad_norm": 0.10491743683815002, "global_step": 44790, "epoch": 422, "lr": 5.803400281711563e-06} +{"train_loss": 0.015213239938020706, "grad_norm": 0.05919330194592476, "global_step": 44791, "epoch": 422, "lr": 5.802014456214228e-06} +{"train_loss": 0.01420624926686287, "grad_norm": 0.0828738585114479, "global_step": 44792, "epoch": 422, "lr": 5.800628786009405e-06} +{"train_loss": 0.01582980342209339, "grad_norm": 0.12370191514492035, "global_step": 44793, "epoch": 422, "lr": 5.79924327110194e-06} +{"train_loss": 0.015350347384810448, "grad_norm": 0.09931971877813339, "global_step": 44794, "epoch": 422, "lr": 5.797857911496707e-06} +{"train_loss": 0.013726521283388138, "grad_norm": 0.08055581897497177, "global_step": 44795, "epoch": 422, "lr": 5.796472707198603e-06} +{"train_loss": 0.016417067497968674, "grad_norm": 0.09404061734676361, "global_step": 44796, "epoch": 422, "lr": 5.79508765821245e-06} +{"train_loss": 0.013833236880600452, "grad_norm": 0.10996399074792862, "global_step": 44797, "epoch": 422, "lr": 5.793702764543135e-06} +{"train_loss": 0.01871521770954132, "grad_norm": 0.10907229781150818, "global_step": 44798, "epoch": 422, "lr": 5.792318026195531e-06} +{"train_loss": 0.014860727824270725, "grad_norm": 0.08484172821044922, "global_step": 44799, "epoch": 422, "lr": 5.79093344317449e-06} +{"train_loss": 0.015374964103102684, "grad_norm": 0.053423695266246796, "global_step": 44800, "epoch": 422, "lr": 5.78954901548488e-06} +{"train_loss": 0.01295781135559082, "grad_norm": 0.09190887212753296, "global_step": 44801, "epoch": 422, "lr": 5.78816474313158e-06} +{"train_loss": 0.01468702219426632, "grad_norm": 0.057918451726436615, "global_step": 44802, "epoch": 422, "lr": 5.786780626119437e-06} +{"train_loss": 0.015127912163734436, "grad_norm": 0.05871275067329407, "global_step": 44803, "epoch": 422, "lr": 5.785396664453313e-06} +{"train_loss": 0.014703189954161644, "grad_norm": 0.0742194727063179, "global_step": 44804, "epoch": 422, "lr": 5.784012858138077e-06} +{"train_loss": 0.014955406077206135, "grad_norm": 0.11260690540075302, "global_step": 44805, "epoch": 422, "lr": 5.782629207178603e-06} +{"train_loss": 0.016102930530905724, "grad_norm": 0.1323523074388504, "global_step": 44806, "epoch": 422, "lr": 5.781245711579725e-06} +{"train_loss": 0.015131054446101189, "grad_norm": 0.08458433300256729, "global_step": 44807, "epoch": 422, "lr": 5.779862371346334e-06} +{"train_loss": 0.01591731235384941, "grad_norm": 0.06596377491950989, "global_step": 44808, "epoch": 422, "lr": 5.77847918648326e-06} +{"train_loss": 0.016198456287384033, "grad_norm": 0.0658988207578659, "global_step": 44809, "epoch": 422, "lr": 5.777096156995393e-06} +{"train_loss": 0.015599117614328861, "grad_norm": 0.12264394015073776, "global_step": 44810, "epoch": 422, "lr": 5.775713282887568e-06} +{"train_loss": 0.016367079690098763, "grad_norm": 0.08270663768053055, "global_step": 44811, "epoch": 422, "lr": 5.774330564164654e-06} +{"train_loss": 0.0160505473613739, "grad_norm": 0.08174902945756912, "global_step": 44812, "epoch": 422, "lr": 5.77294800083153e-06} +{"train_loss": 0.01344283577054739, "grad_norm": 0.05706185847520828, "global_step": 44813, "epoch": 422, "lr": 5.771565592893008e-06} +{"train_loss": 0.017246514558792114, "grad_norm": 0.06613035500049591, "global_step": 44814, "epoch": 422, "lr": 5.770183340353974e-06} +{"train_loss": 0.01418408565223217, "grad_norm": 0.08107441663742065, "global_step": 44815, "epoch": 422, "lr": 5.768801243219291e-06} +{"train_loss": 0.013103332370519638, "grad_norm": 0.07036646455526352, "global_step": 44816, "epoch": 422, "lr": 5.767419301493793e-06} +{"train_loss": 0.013726937584578991, "grad_norm": 0.06364881247282028, "global_step": 44817, "epoch": 422, "lr": 5.766037515182349e-06} +{"train_loss": 0.011565277352929115, "grad_norm": 0.07290313392877579, "global_step": 44818, "epoch": 422, "lr": 5.764655884289821e-06} +{"train_loss": 0.013881227932870388, "grad_norm": 0.0774322971701622, "global_step": 44819, "epoch": 422, "lr": 5.763274408821057e-06} +{"train_loss": 0.01455344632267952, "grad_norm": 0.07272966206073761, "global_step": 44820, "epoch": 422, "lr": 5.761893088780895e-06} +{"train_loss": 0.014210491441190243, "grad_norm": 0.08252911269664764, "global_step": 44821, "epoch": 422, "lr": 5.760511924174206e-06} +{"train_loss": 0.014304907992482185, "grad_norm": 0.16036172211170197, "global_step": 44822, "epoch": 422, "lr": 5.759130915005845e-06} +{"train_loss": 0.016646601259708405, "grad_norm": 0.0718630999326706, "global_step": 44823, "epoch": 422, "lr": 5.7577500612806535e-06} +{"train_loss": 0.01493285782635212, "grad_norm": 0.09103614091873169, "global_step": 44824, "epoch": 422, "lr": 5.7563693630034944e-06} +{"train_loss": 0.01394101232290268, "grad_norm": 0.0635410025715828, "global_step": 44825, "epoch": 422, "lr": 5.754988820179202e-06} +{"train_loss": 0.01430155336856842, "grad_norm": 0.07685083895921707, "global_step": 44826, "epoch": 422, "lr": 5.753608432812646e-06} +{"train_loss": 0.016062598675489426, "grad_norm": 0.11084263771772385, "global_step": 44827, "epoch": 422, "lr": 5.75222820090866e-06} +{"train_loss": 0.01403433084487915, "grad_norm": 0.10705479234457016, "global_step": 44828, "epoch": 422, "lr": 5.750848124472102e-06} +{"train_loss": 0.015214299783110619, "grad_norm": 0.07083418220281601, "global_step": 44829, "epoch": 422, "lr": 5.74946820350783e-06} +{"train_loss": 0.01468009315431118, "grad_norm": 0.0974845141172409, "global_step": 44830, "epoch": 422, "lr": 5.748088438020677e-06} +{"train_loss": 0.01569885015487671, "grad_norm": 0.07763401418924332, "global_step": 44831, "epoch": 422, "lr": 5.74670882801549e-06} +{"train_loss": 0.013770009391009808, "grad_norm": 0.06655742228031158, "global_step": 44832, "epoch": 422, "lr": 5.7453293734971385e-06} +{"train_loss": 0.013813503086566925, "grad_norm": 0.07756877690553665, "global_step": 44833, "epoch": 422, "lr": 5.743950074470439e-06} +{"train_loss": 0.015614410862326622, "grad_norm": 0.12296482920646667, "global_step": 44834, "epoch": 422, "lr": 5.74257093094025e-06} +{"train_loss": 0.014522253535687923, "grad_norm": 0.09294365346431732, "global_step": 44835, "epoch": 422, "lr": 5.741191942911423e-06} +{"train_loss": 0.013574814423918724, "grad_norm": 0.07633969932794571, "global_step": 44836, "epoch": 422, "lr": 5.7398131103888195e-06} +{"train_loss": 0.014973685408959974, "grad_norm": 0.1107906699180603, "global_step": 44837, "epoch": 422, "lr": 5.738434433377243e-06, "val_loss": 0.01989969238638878} +{"train_loss": 0.016637001186609268, "grad_norm": 0.08804735541343689, "global_step": 44838, "epoch": 423, "lr": 5.737055911881562e-06} +{"train_loss": 0.015640664845705032, "grad_norm": 0.07800288498401642, "global_step": 44839, "epoch": 423, "lr": 5.735677545906626e-06} +{"train_loss": 0.015992509201169014, "grad_norm": 0.06213296949863434, "global_step": 44840, "epoch": 423, "lr": 5.734299335457255e-06} +{"train_loss": 0.015206494368612766, "grad_norm": 0.07801035791635513, "global_step": 44841, "epoch": 423, "lr": 5.732921280538311e-06} +{"train_loss": 0.015116848982870579, "grad_norm": 0.06742732226848602, "global_step": 44842, "epoch": 423, "lr": 5.731543381154642e-06} +{"train_loss": 0.01353597640991211, "grad_norm": 0.1142025738954544, "global_step": 44843, "epoch": 423, "lr": 5.730165637311069e-06} +{"train_loss": 0.013907212764024734, "grad_norm": 0.07057787477970123, "global_step": 44844, "epoch": 423, "lr": 5.72878804901244e-06} +{"train_loss": 0.015303224325180054, "grad_norm": 0.06567135453224182, "global_step": 44845, "epoch": 423, "lr": 5.72741061626359e-06} +{"train_loss": 0.016134751960635185, "grad_norm": 0.17462274432182312, "global_step": 44846, "epoch": 423, "lr": 5.726033339069381e-06} +{"train_loss": 0.0151588786393404, "grad_norm": 0.06634673476219177, "global_step": 44847, "epoch": 423, "lr": 5.7246562174346265e-06} +{"train_loss": 0.013664530590176582, "grad_norm": 0.08675646036863327, "global_step": 44848, "epoch": 423, "lr": 5.723279251364183e-06} +{"train_loss": 0.012956101447343826, "grad_norm": 0.10322417318820953, "global_step": 44849, "epoch": 423, "lr": 5.72190244086287e-06} +{"train_loss": 0.015426812693476677, "grad_norm": 0.09744106233119965, "global_step": 44850, "epoch": 423, "lr": 5.720525785935543e-06} +{"train_loss": 0.015456560999155045, "grad_norm": 0.07896973192691803, "global_step": 44851, "epoch": 423, "lr": 5.719149286587028e-06} +{"train_loss": 0.017378026619553566, "grad_norm": 0.07635582983493805, "global_step": 44852, "epoch": 423, "lr": 5.717772942822164e-06} +{"train_loss": 0.015567058697342873, "grad_norm": 0.07635710388422012, "global_step": 44853, "epoch": 423, "lr": 5.716396754645792e-06} +{"train_loss": 0.015369866043329239, "grad_norm": 0.1858779489994049, "global_step": 44854, "epoch": 423, "lr": 5.715020722062752e-06} +{"train_loss": 0.015277529135346413, "grad_norm": 0.13175871968269348, "global_step": 44855, "epoch": 423, "lr": 5.713644845077853e-06} +{"train_loss": 0.01603078655898571, "grad_norm": 0.11263052374124527, "global_step": 44856, "epoch": 423, "lr": 5.712269123695962e-06} +{"train_loss": 0.014894701540470123, "grad_norm": 0.1302311271429062, "global_step": 44857, "epoch": 423, "lr": 5.710893557921881e-06} +{"train_loss": 0.013528076931834221, "grad_norm": 0.0703100711107254, "global_step": 44858, "epoch": 423, "lr": 5.709518147760468e-06} +{"train_loss": 0.018385592848062515, "grad_norm": 0.12663935124874115, "global_step": 44859, "epoch": 423, "lr": 5.70814289321655e-06} +{"train_loss": 0.014442815445363522, "grad_norm": 0.12785962224006653, "global_step": 44860, "epoch": 423, "lr": 5.70676779429496e-06} +{"train_loss": 0.015901759266853333, "grad_norm": 0.09023158997297287, "global_step": 44861, "epoch": 423, "lr": 5.705392851000507e-06} +{"train_loss": 0.013030050322413445, "grad_norm": 0.13448825478553772, "global_step": 44862, "epoch": 423, "lr": 5.704018063338052e-06} +{"train_loss": 0.016278350725769997, "grad_norm": 0.1454523503780365, "global_step": 44863, "epoch": 423, "lr": 5.702643431312415e-06} +{"train_loss": 0.014574732631444931, "grad_norm": 0.09221427887678146, "global_step": 44864, "epoch": 423, "lr": 5.701268954928418e-06} +{"train_loss": 0.014620894566178322, "grad_norm": 0.07143422961235046, "global_step": 44865, "epoch": 423, "lr": 5.699894634190905e-06} +{"train_loss": 0.016568206250667572, "grad_norm": 0.07296907156705856, "global_step": 44866, "epoch": 423, "lr": 5.69852046910469e-06} +{"train_loss": 0.015282254666090012, "grad_norm": 0.07289966940879822, "global_step": 44867, "epoch": 423, "lr": 5.697146459674613e-06} +{"train_loss": 0.014377095736563206, "grad_norm": 0.1874961107969284, "global_step": 44868, "epoch": 423, "lr": 5.695772605905492e-06} +{"train_loss": 0.016368135809898376, "grad_norm": 0.1076236143708229, "global_step": 44869, "epoch": 423, "lr": 5.694398907802157e-06} +{"train_loss": 0.01596830040216446, "grad_norm": 0.08744833618402481, "global_step": 44870, "epoch": 423, "lr": 5.693025365369442e-06} +{"train_loss": 0.01338798739016056, "grad_norm": 0.0601402148604393, "global_step": 44871, "epoch": 423, "lr": 5.691651978612173e-06} +{"train_loss": 0.012425151653587818, "grad_norm": 0.05240248516201973, "global_step": 44872, "epoch": 423, "lr": 5.690278747535155e-06} +{"train_loss": 0.015242515131831169, "grad_norm": 0.10405083000659943, "global_step": 44873, "epoch": 423, "lr": 5.688905672143241e-06} +{"train_loss": 0.017053741961717606, "grad_norm": 0.07703891396522522, "global_step": 44874, "epoch": 423, "lr": 5.6875327524412326e-06} +{"train_loss": 0.014273520559072495, "grad_norm": 0.06966153532266617, "global_step": 44875, "epoch": 423, "lr": 5.686159988433959e-06} +{"train_loss": 0.015964189544320107, "grad_norm": 0.059391703456640244, "global_step": 44876, "epoch": 423, "lr": 5.684787380126261e-06} +{"train_loss": 0.013695242814719677, "grad_norm": 0.06938617676496506, "global_step": 44877, "epoch": 423, "lr": 5.6834149275229456e-06} +{"train_loss": 0.013872506096959114, "grad_norm": 0.05920799821615219, "global_step": 44878, "epoch": 423, "lr": 5.682042630628831e-06} +{"train_loss": 0.015350403264164925, "grad_norm": 0.08848792314529419, "global_step": 44879, "epoch": 423, "lr": 5.680670489448742e-06} +{"train_loss": 0.015545599162578583, "grad_norm": 0.08101017028093338, "global_step": 44880, "epoch": 423, "lr": 5.679298503987518e-06} +{"train_loss": 0.014529184438288212, "grad_norm": 0.08271311968564987, "global_step": 44881, "epoch": 423, "lr": 5.67792667424995e-06} +{"train_loss": 0.014591203071177006, "grad_norm": 0.06294132769107819, "global_step": 44882, "epoch": 423, "lr": 5.6765550002408685e-06} +{"train_loss": 0.01731642708182335, "grad_norm": 0.08412688970565796, "global_step": 44883, "epoch": 423, "lr": 5.675183481965113e-06} +{"train_loss": 0.015733500942587852, "grad_norm": 0.09682421386241913, "global_step": 44884, "epoch": 423, "lr": 5.673812119427485e-06} +{"train_loss": 0.016130251809954643, "grad_norm": 0.10153637826442719, "global_step": 44885, "epoch": 423, "lr": 5.672440912632792e-06} +{"train_loss": 0.01420062780380249, "grad_norm": 0.09106381237506866, "global_step": 44886, "epoch": 423, "lr": 5.67106986158587e-06} +{"train_loss": 0.015672758221626282, "grad_norm": 0.0648871660232544, "global_step": 44887, "epoch": 423, "lr": 5.669698966291537e-06} +{"train_loss": 0.013279448263347149, "grad_norm": 0.07449960708618164, "global_step": 44888, "epoch": 423, "lr": 5.668328226754594e-06} +{"train_loss": 0.012068802490830421, "grad_norm": 0.08474607765674591, "global_step": 44889, "epoch": 423, "lr": 5.666957642979875e-06} +{"train_loss": 0.015009693801403046, "grad_norm": 0.08581074327230453, "global_step": 44890, "epoch": 423, "lr": 5.665587214972174e-06} +{"train_loss": 0.01577480509877205, "grad_norm": 0.08118399977684021, "global_step": 44891, "epoch": 423, "lr": 5.664216942736333e-06} +{"train_loss": 0.013169706799089909, "grad_norm": 0.07878638803958893, "global_step": 44892, "epoch": 423, "lr": 5.66284682627714e-06} +{"train_loss": 0.014580540359020233, "grad_norm": 0.07806577533483505, "global_step": 44893, "epoch": 423, "lr": 5.661476865599424e-06} +{"train_loss": 0.014732318930327892, "grad_norm": 0.07451855391263962, "global_step": 44894, "epoch": 423, "lr": 5.6601070607080035e-06} +{"train_loss": 0.015837490558624268, "grad_norm": 0.07721240073442459, "global_step": 44895, "epoch": 423, "lr": 5.658737411607689e-06} +{"train_loss": 0.015508397482335567, "grad_norm": 0.08196556568145752, "global_step": 44896, "epoch": 423, "lr": 5.657367918303274e-06} +{"train_loss": 0.016964009031653404, "grad_norm": 0.10146261751651764, "global_step": 44897, "epoch": 423, "lr": 5.655998580799593e-06} +{"train_loss": 0.01267823576927185, "grad_norm": 0.06343187391757965, "global_step": 44898, "epoch": 423, "lr": 5.654629399101441e-06} +{"train_loss": 0.012427148409187794, "grad_norm": 0.0860009714961052, "global_step": 44899, "epoch": 423, "lr": 5.653260373213632e-06} +{"train_loss": 0.014477039687335491, "grad_norm": 0.06273715943098068, "global_step": 44900, "epoch": 423, "lr": 5.651891503140994e-06} +{"train_loss": 0.015532882884144783, "grad_norm": 0.06762050837278366, "global_step": 44901, "epoch": 423, "lr": 5.650522788888324e-06} +{"train_loss": 0.01445891335606575, "grad_norm": 0.0734747126698494, "global_step": 44902, "epoch": 423, "lr": 5.649154230460413e-06} +{"train_loss": 0.015392588451504707, "grad_norm": 0.1240200474858284, "global_step": 44903, "epoch": 423, "lr": 5.64778582786209e-06} +{"train_loss": 0.013990238308906555, "grad_norm": 0.11930791288614273, "global_step": 44904, "epoch": 423, "lr": 5.646417581098174e-06} +{"train_loss": 0.013898390345275402, "grad_norm": 0.06401023268699646, "global_step": 44905, "epoch": 423, "lr": 5.645049490173443e-06} +{"train_loss": 0.012843996286392212, "grad_norm": 0.13762889802455902, "global_step": 44906, "epoch": 423, "lr": 5.643681555092717e-06} +{"train_loss": 0.012454479932785034, "grad_norm": 0.11058346927165985, "global_step": 44907, "epoch": 423, "lr": 5.642313775860819e-06} +{"train_loss": 0.016142575070261955, "grad_norm": 0.086204394698143, "global_step": 44908, "epoch": 423, "lr": 5.6409461524825345e-06} +{"train_loss": 0.01631881110370159, "grad_norm": 0.07650286704301834, "global_step": 44909, "epoch": 423, "lr": 5.639578684962665e-06} +{"train_loss": 0.01685776747763157, "grad_norm": 0.09876766055822372, "global_step": 44910, "epoch": 423, "lr": 5.638211373306029e-06} +{"train_loss": 0.013966836035251617, "grad_norm": 0.07500416785478592, "global_step": 44911, "epoch": 423, "lr": 5.636844217517434e-06} +{"train_loss": 0.014493979513645172, "grad_norm": 0.10306897014379501, "global_step": 44912, "epoch": 423, "lr": 5.63547721760167e-06} +{"train_loss": 0.01602886989712715, "grad_norm": 0.1266050934791565, "global_step": 44913, "epoch": 423, "lr": 5.634110373563539e-06} +{"train_loss": 0.015375013463199139, "grad_norm": 0.09835352748632431, "global_step": 44914, "epoch": 423, "lr": 5.632743685407859e-06} +{"train_loss": 0.016237517818808556, "grad_norm": 0.11937061697244644, "global_step": 44915, "epoch": 423, "lr": 5.631377153139417e-06} +{"train_loss": 0.014855417422950268, "grad_norm": 0.08549979329109192, "global_step": 44916, "epoch": 423, "lr": 5.630010776763017e-06} +{"train_loss": 0.015574721619486809, "grad_norm": 0.07096504420042038, "global_step": 44917, "epoch": 423, "lr": 5.628644556283474e-06} +{"train_loss": 0.014391962438821793, "grad_norm": 0.07716511189937592, "global_step": 44918, "epoch": 423, "lr": 5.627278491705579e-06} +{"train_loss": 0.014213912189006805, "grad_norm": 0.13605603575706482, "global_step": 44919, "epoch": 423, "lr": 5.625912583034115e-06} +{"train_loss": 0.012364612892270088, "grad_norm": 0.08498135209083557, "global_step": 44920, "epoch": 423, "lr": 5.6245468302739024e-06} +{"train_loss": 0.01557021401822567, "grad_norm": 0.07245258241891861, "global_step": 44921, "epoch": 423, "lr": 5.623181233429742e-06} +{"train_loss": 0.016322454437613487, "grad_norm": 0.16243967413902283, "global_step": 44922, "epoch": 423, "lr": 5.621815792506413e-06} +{"train_loss": 0.01552609447389841, "grad_norm": 0.0822966918349266, "global_step": 44923, "epoch": 423, "lr": 5.620450507508723e-06} +{"train_loss": 0.01402122899889946, "grad_norm": 0.08255943655967712, "global_step": 44924, "epoch": 423, "lr": 5.61908537844148e-06} +{"train_loss": 0.01412126049399376, "grad_norm": 0.08453154563903809, "global_step": 44925, "epoch": 423, "lr": 5.6177204053094675e-06} +{"train_loss": 0.015998173505067825, "grad_norm": 0.07151374220848083, "global_step": 44926, "epoch": 423, "lr": 5.616355588117478e-06} +{"train_loss": 0.0139039047062397, "grad_norm": 0.09374579042196274, "global_step": 44927, "epoch": 423, "lr": 5.614990926870312e-06} +{"train_loss": 0.014363003894686699, "grad_norm": 0.08464103937149048, "global_step": 44928, "epoch": 423, "lr": 5.613626421572776e-06} +{"train_loss": 0.01509777270257473, "grad_norm": 0.08916039764881134, "global_step": 44929, "epoch": 423, "lr": 5.612262072229646e-06} +{"train_loss": 0.01499263383448124, "grad_norm": 0.060388628393411636, "global_step": 44930, "epoch": 423, "lr": 5.610897878845728e-06} +{"train_loss": 0.012863436713814735, "grad_norm": 0.10589049011468887, "global_step": 44931, "epoch": 423, "lr": 5.6095338414258124e-06} +{"train_loss": 0.015352071262896061, "grad_norm": 0.0772532969713211, "global_step": 44932, "epoch": 423, "lr": 5.60816995997468e-06} +{"train_loss": 0.015163792297244072, "grad_norm": 0.08348459750413895, "global_step": 44933, "epoch": 423, "lr": 5.606806234497136e-06} +{"train_loss": 0.014097945764660835, "grad_norm": 0.0618947334587574, "global_step": 44934, "epoch": 423, "lr": 5.605442664997973e-06} +{"train_loss": 0.016536327078938484, "grad_norm": 0.09367918223142624, "global_step": 44935, "epoch": 423, "lr": 5.604079251481975e-06} +{"train_loss": 0.013463573530316353, "grad_norm": 0.07219754159450531, "global_step": 44936, "epoch": 423, "lr": 5.602715993953938e-06} +{"train_loss": 0.01688561961054802, "grad_norm": 0.0791708379983902, "global_step": 44937, "epoch": 423, "lr": 5.6013528924186436e-06} +{"train_loss": 0.015995441004633904, "grad_norm": 0.08569466322660446, "global_step": 44938, "epoch": 423, "lr": 5.599989946880896e-06} +{"train_loss": 0.016362199559807777, "grad_norm": 0.07264787703752518, "global_step": 44939, "epoch": 423, "lr": 5.598627157345465e-06} +{"train_loss": 0.014799388125538826, "grad_norm": 0.11288721859455109, "global_step": 44940, "epoch": 423, "lr": 5.597264523817147e-06} +{"train_loss": 0.01446638535708189, "grad_norm": 0.0902429148554802, "global_step": 44941, "epoch": 423, "lr": 5.595902046300744e-06} +{"train_loss": 0.014729119837284088, "grad_norm": 0.0542181134223938, "global_step": 44942, "epoch": 423, "lr": 5.594539724801029e-06} +{"train_loss": 0.014979187205855577, "grad_norm": 0.11006047576665878, "global_step": 44943, "epoch": 423, "lr": 5.593177559322777e-06, "val_loss": 0.019274085760116577} +{"train_loss": 0.013961415737867355, "grad_norm": 0.09455570578575134, "global_step": 44944, "epoch": 424, "lr": 5.591815549870788e-06} +{"train_loss": 0.014024726115167141, "grad_norm": 0.19407571852207184, "global_step": 44945, "epoch": 424, "lr": 5.59045369644986e-06} +{"train_loss": 0.014138232916593552, "grad_norm": 0.07495535910129547, "global_step": 44946, "epoch": 424, "lr": 5.589091999064755e-06} +{"train_loss": 0.01488974317908287, "grad_norm": 0.09232399612665176, "global_step": 44947, "epoch": 424, "lr": 5.587730457720264e-06} +{"train_loss": 0.014169209636747837, "grad_norm": 0.08593728393316269, "global_step": 44948, "epoch": 424, "lr": 5.586369072421183e-06} +{"train_loss": 0.013174818828701973, "grad_norm": 0.0542445182800293, "global_step": 44949, "epoch": 424, "lr": 5.585007843172285e-06} +{"train_loss": 0.01557464525103569, "grad_norm": 0.08679556101560593, "global_step": 44950, "epoch": 424, "lr": 5.583646769978346e-06} +{"train_loss": 0.01665688306093216, "grad_norm": 0.07583416998386383, "global_step": 44951, "epoch": 424, "lr": 5.5822858528441605e-06} +{"train_loss": 0.014724904671311378, "grad_norm": 0.06849082559347153, "global_step": 44952, "epoch": 424, "lr": 5.580925091774508e-06} +{"train_loss": 0.015288755297660828, "grad_norm": 0.08421580493450165, "global_step": 44953, "epoch": 424, "lr": 5.579564486774164e-06} +{"train_loss": 0.014648549258708954, "grad_norm": 0.10132475942373276, "global_step": 44954, "epoch": 424, "lr": 5.578204037847918e-06} +{"train_loss": 0.012832637876272202, "grad_norm": 0.12758606672286987, "global_step": 44955, "epoch": 424, "lr": 5.576843745000543e-06} +{"train_loss": 0.01494657900184393, "grad_norm": 0.09651875495910645, "global_step": 44956, "epoch": 424, "lr": 5.575483608236814e-06} +{"train_loss": 0.014364208094775677, "grad_norm": 0.0615360401570797, "global_step": 44957, "epoch": 424, "lr": 5.574123627561518e-06} +{"train_loss": 0.016490088775753975, "grad_norm": 0.06757473945617676, "global_step": 44958, "epoch": 424, "lr": 5.572763802979436e-06} +{"train_loss": 0.016509532928466797, "grad_norm": 0.08845394849777222, "global_step": 44959, "epoch": 424, "lr": 5.5714041344953445e-06} +{"train_loss": 0.01588868349790573, "grad_norm": 0.06375187635421753, "global_step": 44960, "epoch": 424, "lr": 5.570044622114002e-06} +{"train_loss": 0.015133647248148918, "grad_norm": 0.18989868462085724, "global_step": 44961, "epoch": 424, "lr": 5.568685265840207e-06} +{"train_loss": 0.017887461930513382, "grad_norm": 0.10100549459457397, "global_step": 44962, "epoch": 424, "lr": 5.567326065678735e-06} +{"train_loss": 0.015224272385239601, "grad_norm": 0.06302382051944733, "global_step": 44963, "epoch": 424, "lr": 5.565967021634349e-06} +{"train_loss": 0.018054036423563957, "grad_norm": 0.07315253466367722, "global_step": 44964, "epoch": 424, "lr": 5.564608133711829e-06} +{"train_loss": 0.01509246975183487, "grad_norm": 0.11235972493886948, "global_step": 44965, "epoch": 424, "lr": 5.563249401915966e-06} +{"train_loss": 0.013246869668364525, "grad_norm": 0.06661166995763779, "global_step": 44966, "epoch": 424, "lr": 5.561890826251514e-06} +{"train_loss": 0.018002189695835114, "grad_norm": 0.08491100370883942, "global_step": 44967, "epoch": 424, "lr": 5.560532406723245e-06} +{"train_loss": 0.013107123784720898, "grad_norm": 0.16347967088222504, "global_step": 44968, "epoch": 424, "lr": 5.5591741433359375e-06} +{"train_loss": 0.013129059225320816, "grad_norm": 0.10414169728755951, "global_step": 44969, "epoch": 424, "lr": 5.557816036094376e-06} +{"train_loss": 0.01659468375146389, "grad_norm": 0.09277762472629547, "global_step": 44970, "epoch": 424, "lr": 5.556458085003308e-06} +{"train_loss": 0.016254590824246407, "grad_norm": 0.07425466179847717, "global_step": 44971, "epoch": 424, "lr": 5.5551002900675335e-06} +{"train_loss": 0.014921579509973526, "grad_norm": 0.10234851390123367, "global_step": 44972, "epoch": 424, "lr": 5.553742651291804e-06} +{"train_loss": 0.014894943684339523, "grad_norm": 0.08375765383243561, "global_step": 44973, "epoch": 424, "lr": 5.552385168680885e-06} +{"train_loss": 0.012824490666389465, "grad_norm": 0.06371378898620605, "global_step": 44974, "epoch": 424, "lr": 5.5510278422395586e-06} +{"train_loss": 0.014793604612350464, "grad_norm": 0.07220938801765442, "global_step": 44975, "epoch": 424, "lr": 5.549670671972596e-06} +{"train_loss": 0.014694785699248314, "grad_norm": 0.08774710446596146, "global_step": 44976, "epoch": 424, "lr": 5.548313657884752e-06} +{"train_loss": 0.01399405300617218, "grad_norm": 0.06370174884796143, "global_step": 44977, "epoch": 424, "lr": 5.546956799980807e-06} +{"train_loss": 0.013421926647424698, "grad_norm": 0.08130935579538345, "global_step": 44978, "epoch": 424, "lr": 5.545600098265519e-06} +{"train_loss": 0.015698034316301346, "grad_norm": 0.060998234897851944, "global_step": 44979, "epoch": 424, "lr": 5.544243552743672e-06} +{"train_loss": 0.013590056449174881, "grad_norm": 0.07487981766462326, "global_step": 44980, "epoch": 424, "lr": 5.542887163420002e-06} +{"train_loss": 0.014734085649251938, "grad_norm": 0.08017529547214508, "global_step": 44981, "epoch": 424, "lr": 5.541530930299299e-06} +{"train_loss": 0.01274799183011055, "grad_norm": 0.15425646305084229, "global_step": 44982, "epoch": 424, "lr": 5.540174853386326e-06} +{"train_loss": 0.017048172652721405, "grad_norm": 0.08985418826341629, "global_step": 44983, "epoch": 424, "lr": 5.538818932685852e-06} +{"train_loss": 0.014602210372686386, "grad_norm": 0.08230554312467575, "global_step": 44984, "epoch": 424, "lr": 5.537463168202617e-06} +{"train_loss": 0.017070550471544266, "grad_norm": 0.06743454933166504, "global_step": 44985, "epoch": 424, "lr": 5.5361075599414e-06} +{"train_loss": 0.01625894382596016, "grad_norm": 0.07704805582761765, "global_step": 44986, "epoch": 424, "lr": 5.534752107906971e-06} +{"train_loss": 0.013083157129585743, "grad_norm": 0.07586126029491425, "global_step": 44987, "epoch": 424, "lr": 5.53339681210408e-06} +{"train_loss": 0.014692217111587524, "grad_norm": 0.06293267011642456, "global_step": 44988, "epoch": 424, "lr": 5.532041672537497e-06} +{"train_loss": 0.015474689193069935, "grad_norm": 0.09588982164859772, "global_step": 44989, "epoch": 424, "lr": 5.530686689211994e-06} +{"train_loss": 0.01650877669453621, "grad_norm": 0.06048008054494858, "global_step": 44990, "epoch": 424, "lr": 5.5293318621323086e-06} +{"train_loss": 0.013851776719093323, "grad_norm": 0.10594403743743896, "global_step": 44991, "epoch": 424, "lr": 5.527977191303202e-06} +{"train_loss": 0.014717001467943192, "grad_norm": 0.07524718344211578, "global_step": 44992, "epoch": 424, "lr": 5.526622676729448e-06} +{"train_loss": 0.015399875119328499, "grad_norm": 0.07079223543405533, "global_step": 44993, "epoch": 424, "lr": 5.52526831841581e-06} +{"train_loss": 0.015675825998187065, "grad_norm": 0.09073412418365479, "global_step": 44994, "epoch": 424, "lr": 5.523914116367029e-06} +{"train_loss": 0.013389039784669876, "grad_norm": 0.07714608311653137, "global_step": 44995, "epoch": 424, "lr": 5.522560070587879e-06} +{"train_loss": 0.014891035854816437, "grad_norm": 0.10629618167877197, "global_step": 44996, "epoch": 424, "lr": 5.521206181083111e-06} +{"train_loss": 0.015043660998344421, "grad_norm": 0.0918034091591835, "global_step": 44997, "epoch": 424, "lr": 5.519852447857471e-06} +{"train_loss": 0.015255709178745747, "grad_norm": 0.09451641142368317, "global_step": 44998, "epoch": 424, "lr": 5.518498870915722e-06} +{"train_loss": 0.01515025645494461, "grad_norm": 0.08167754858732224, "global_step": 44999, "epoch": 424, "lr": 5.51714545026264e-06} +{"train_loss": 0.014107373543083668, "grad_norm": 0.09907504916191101, "global_step": 45000, "epoch": 424, "lr": 5.515792185902951e-06} +{"train_loss": 0.016080589964985847, "grad_norm": 0.06420200318098068, "global_step": 45001, "epoch": 424, "lr": 5.514439077841433e-06} +{"train_loss": 0.017514267936348915, "grad_norm": 0.1032654345035553, "global_step": 45002, "epoch": 424, "lr": 5.5130861260828225e-06} +{"train_loss": 0.016653694212436676, "grad_norm": 0.09436763823032379, "global_step": 45003, "epoch": 424, "lr": 5.511733330631885e-06} +{"train_loss": 0.015309775248169899, "grad_norm": 0.07296744734048843, "global_step": 45004, "epoch": 424, "lr": 5.510380691493361e-06} +{"train_loss": 0.014890089631080627, "grad_norm": 0.10608880966901779, "global_step": 45005, "epoch": 424, "lr": 5.509028208672012e-06} +{"train_loss": 0.0138887669891119, "grad_norm": 0.05906618386507034, "global_step": 45006, "epoch": 424, "lr": 5.507675882172597e-06} +{"train_loss": 0.014090999960899353, "grad_norm": 0.08002495020627975, "global_step": 45007, "epoch": 424, "lr": 5.506323711999867e-06} +{"train_loss": 0.015392504632472992, "grad_norm": 0.08361275494098663, "global_step": 45008, "epoch": 424, "lr": 5.5049716981585504e-06} +{"train_loss": 0.015144900418817997, "grad_norm": 0.08226914703845978, "global_step": 45009, "epoch": 424, "lr": 5.503619840653412e-06} +{"train_loss": 0.014106471091508865, "grad_norm": 0.08654236793518066, "global_step": 45010, "epoch": 424, "lr": 5.502268139489214e-06} +{"train_loss": 0.016031306236982346, "grad_norm": 0.07610537111759186, "global_step": 45011, "epoch": 424, "lr": 5.50091659467068e-06} +{"train_loss": 0.016735892742872238, "grad_norm": 0.07740600407123566, "global_step": 45012, "epoch": 424, "lr": 5.499565206202589e-06} +{"train_loss": 0.013080179691314697, "grad_norm": 0.07126107811927795, "global_step": 45013, "epoch": 424, "lr": 5.498213974089672e-06} +{"train_loss": 0.015632178634405136, "grad_norm": 0.0824165791273117, "global_step": 45014, "epoch": 424, "lr": 5.496862898336669e-06} +{"train_loss": 0.01383562944829464, "grad_norm": 0.06629832834005356, "global_step": 45015, "epoch": 424, "lr": 5.4955119789483315e-06} +{"train_loss": 0.014963414520025253, "grad_norm": 0.082114078104496, "global_step": 45016, "epoch": 424, "lr": 5.494161215929422e-06} +{"train_loss": 0.015280184336006641, "grad_norm": 0.100951187312603, "global_step": 45017, "epoch": 424, "lr": 5.492810609284666e-06} +{"train_loss": 0.013941502198576927, "grad_norm": 0.05727895721793175, "global_step": 45018, "epoch": 424, "lr": 5.491460159018824e-06} +{"train_loss": 0.013729149475693703, "grad_norm": 0.06347595155239105, "global_step": 45019, "epoch": 424, "lr": 5.490109865136628e-06} +{"train_loss": 0.01639433577656746, "grad_norm": 0.06407354772090912, "global_step": 45020, "epoch": 424, "lr": 5.4887597276428396e-06} +{"train_loss": 0.0148547338321805, "grad_norm": 0.09536876529455185, "global_step": 45021, "epoch": 424, "lr": 5.487409746542182e-06} +{"train_loss": 0.012361191213130951, "grad_norm": 0.07495345920324326, "global_step": 45022, "epoch": 424, "lr": 5.486059921839409e-06} +{"train_loss": 0.014450590126216412, "grad_norm": 0.07347570359706879, "global_step": 45023, "epoch": 424, "lr": 5.484710253539271e-06} +{"train_loss": 0.01387277152389288, "grad_norm": 0.07914141565561295, "global_step": 45024, "epoch": 424, "lr": 5.4833607416465035e-06} +{"train_loss": 0.016327591612935066, "grad_norm": 0.09491249173879623, "global_step": 45025, "epoch": 424, "lr": 5.482011386165836e-06} +{"train_loss": 0.016003869473934174, "grad_norm": 0.07388180494308472, "global_step": 45026, "epoch": 424, "lr": 5.48066218710202e-06} +{"train_loss": 0.016654539853334427, "grad_norm": 0.08845244348049164, "global_step": 45027, "epoch": 424, "lr": 5.479313144459808e-06} +{"train_loss": 0.018100418150424957, "grad_norm": 0.10936643928289413, "global_step": 45028, "epoch": 424, "lr": 5.477964258243917e-06} +{"train_loss": 0.015293685719370842, "grad_norm": 0.06841567903757095, "global_step": 45029, "epoch": 424, "lr": 5.476615528459095e-06} +{"train_loss": 0.013816313818097115, "grad_norm": 0.09147273004055023, "global_step": 45030, "epoch": 424, "lr": 5.475266955110103e-06} +{"train_loss": 0.01572832092642784, "grad_norm": 0.09196745604276657, "global_step": 45031, "epoch": 424, "lr": 5.473918538201644e-06} +{"train_loss": 0.013686710968613625, "grad_norm": 0.056636687368154526, "global_step": 45032, "epoch": 424, "lr": 5.47257027773847e-06} +{"train_loss": 0.013765651732683182, "grad_norm": 0.12431660294532776, "global_step": 45033, "epoch": 424, "lr": 5.471222173725326e-06} +{"train_loss": 0.014643974602222443, "grad_norm": 0.09028146415948868, "global_step": 45034, "epoch": 424, "lr": 5.469874226166938e-06} +{"train_loss": 0.012945745140314102, "grad_norm": 0.10650575160980225, "global_step": 45035, "epoch": 424, "lr": 5.468526435068044e-06} +{"train_loss": 0.015759993344545364, "grad_norm": 0.06857504695653915, "global_step": 45036, "epoch": 424, "lr": 5.467178800433392e-06} +{"train_loss": 0.013128363527357578, "grad_norm": 0.077389195561409, "global_step": 45037, "epoch": 424, "lr": 5.465831322267706e-06} +{"train_loss": 0.015775753185153008, "grad_norm": 0.0739632323384285, "global_step": 45038, "epoch": 424, "lr": 5.464484000575709e-06} +{"train_loss": 0.016652394086122513, "grad_norm": 0.07117766886949539, "global_step": 45039, "epoch": 424, "lr": 5.463136835362148e-06} +{"train_loss": 0.014325717464089394, "grad_norm": 0.08641888201236725, "global_step": 45040, "epoch": 424, "lr": 5.461789826631769e-06} +{"train_loss": 0.015869155526161194, "grad_norm": 0.07973330467939377, "global_step": 45041, "epoch": 424, "lr": 5.460442974389279e-06} +{"train_loss": 0.015566742978990078, "grad_norm": 0.05907149985432625, "global_step": 45042, "epoch": 424, "lr": 5.459096278639436e-06} +{"train_loss": 0.013857337646186352, "grad_norm": 0.11071980744600296, "global_step": 45043, "epoch": 424, "lr": 5.457749739386941e-06} +{"train_loss": 0.014859875664114952, "grad_norm": 0.058226943016052246, "global_step": 45044, "epoch": 424, "lr": 5.456403356636558e-06} +{"train_loss": 0.014534522779285908, "grad_norm": 0.06769832968711853, "global_step": 45045, "epoch": 424, "lr": 5.4550571303929934e-06} +{"train_loss": 0.01516399160027504, "grad_norm": 0.05984368175268173, "global_step": 45046, "epoch": 424, "lr": 5.453711060660982e-06} +{"train_loss": 0.014525636099278927, "grad_norm": 0.0917133092880249, "global_step": 45047, "epoch": 424, "lr": 5.452365147445271e-06} +{"train_loss": 0.014473779127001762, "grad_norm": 0.09129945188760757, "global_step": 45048, "epoch": 424, "lr": 5.451019390750572e-06} +{"train_loss": 0.014972041176526615, "grad_norm": 0.1304897516965866, "global_step": 45049, "epoch": 424, "lr": 5.449673790581611e-06, "val_loss": 0.019519617781043053} +{"train_loss": 0.01537062507122755, "grad_norm": 0.06533078849315643, "global_step": 45050, "epoch": 425, "lr": 5.448328346943115e-06} +{"train_loss": 0.01420111209154129, "grad_norm": 0.08418742567300797, "global_step": 45051, "epoch": 425, "lr": 5.446983059839833e-06} +{"train_loss": 0.016850866377353668, "grad_norm": 0.08569321036338806, "global_step": 45052, "epoch": 425, "lr": 5.4456379292764706e-06} +{"train_loss": 0.014592503197491169, "grad_norm": 0.0645938590168953, "global_step": 45053, "epoch": 425, "lr": 5.444292955257757e-06} +{"train_loss": 0.013580993749201298, "grad_norm": 0.08204025775194168, "global_step": 45054, "epoch": 425, "lr": 5.44294813778844e-06} +{"train_loss": 0.018005451187491417, "grad_norm": 0.10624903440475464, "global_step": 45055, "epoch": 425, "lr": 5.441603476873203e-06} +{"train_loss": 0.014507833868265152, "grad_norm": 0.09372533857822418, "global_step": 45056, "epoch": 425, "lr": 5.440258972516799e-06} +{"train_loss": 0.014697474427521229, "grad_norm": 0.08419543504714966, "global_step": 45057, "epoch": 425, "lr": 5.438914624723957e-06} +{"train_loss": 0.01710769347846508, "grad_norm": 0.10507582873106003, "global_step": 45058, "epoch": 425, "lr": 5.4375704334993726e-06} +{"train_loss": 0.013975774869322777, "grad_norm": 0.1186944767832756, "global_step": 45059, "epoch": 425, "lr": 5.4362263988477936e-06} +{"train_loss": 0.01598968729376793, "grad_norm": 0.0747302696108818, "global_step": 45060, "epoch": 425, "lr": 5.434882520773943e-06} +{"train_loss": 0.01650485396385193, "grad_norm": 0.10059177130460739, "global_step": 45061, "epoch": 425, "lr": 5.433538799282533e-06} +{"train_loss": 0.013411037623882294, "grad_norm": 0.04822815582156181, "global_step": 45062, "epoch": 425, "lr": 5.432195234378274e-06} +{"train_loss": 0.014270028099417686, "grad_norm": 0.08751325309276581, "global_step": 45063, "epoch": 425, "lr": 5.430851826065903e-06} +{"train_loss": 0.015572543255984783, "grad_norm": 0.07619926333427429, "global_step": 45064, "epoch": 425, "lr": 5.429508574350145e-06} +{"train_loss": 0.015411695465445518, "grad_norm": 0.05481519550085068, "global_step": 45065, "epoch": 425, "lr": 5.428165479235709e-06} +{"train_loss": 0.012678956612944603, "grad_norm": 0.05768256634473801, "global_step": 45066, "epoch": 425, "lr": 5.426822540727305e-06} +{"train_loss": 0.01521329302340746, "grad_norm": 0.047884467989206314, "global_step": 45067, "epoch": 425, "lr": 5.425479758829666e-06} +{"train_loss": 0.015791703015565872, "grad_norm": 0.14612169563770294, "global_step": 45068, "epoch": 425, "lr": 5.424137133547513e-06} +{"train_loss": 0.013221140019595623, "grad_norm": 0.06023809313774109, "global_step": 45069, "epoch": 425, "lr": 5.42279466488555e-06} +{"train_loss": 0.015599112957715988, "grad_norm": 0.1414046734571457, "global_step": 45070, "epoch": 425, "lr": 5.4214523528485e-06} +{"train_loss": 0.013263670727610588, "grad_norm": 0.06987910717725754, "global_step": 45071, "epoch": 425, "lr": 5.420110197441097e-06} +{"train_loss": 0.01570373773574829, "grad_norm": 0.1234474927186966, "global_step": 45072, "epoch": 425, "lr": 5.418768198668017e-06} +{"train_loss": 0.013273129239678383, "grad_norm": 0.061618510633707047, "global_step": 45073, "epoch": 425, "lr": 5.4174263565340055e-06} +{"train_loss": 0.01638251170516014, "grad_norm": 0.07326360791921616, "global_step": 45074, "epoch": 425, "lr": 5.416084671043775e-06} +{"train_loss": 0.01824052631855011, "grad_norm": 0.08252928406000137, "global_step": 45075, "epoch": 425, "lr": 5.4147431422020225e-06} +{"train_loss": 0.015347693115472794, "grad_norm": 0.10396254807710648, "global_step": 45076, "epoch": 425, "lr": 5.413401770013471e-06} +{"train_loss": 0.013353908434510231, "grad_norm": 0.10699767619371414, "global_step": 45077, "epoch": 425, "lr": 5.41206055448285e-06} +{"train_loss": 0.013182971626520157, "grad_norm": 0.07534758746623993, "global_step": 45078, "epoch": 425, "lr": 5.410719495614858e-06} +{"train_loss": 0.016702573746442795, "grad_norm": 0.06310845166444778, "global_step": 45079, "epoch": 425, "lr": 5.409378593414194e-06} +{"train_loss": 0.013964585959911346, "grad_norm": 0.07632094621658325, "global_step": 45080, "epoch": 425, "lr": 5.4080378478855834e-06} +{"train_loss": 0.01343841478228569, "grad_norm": 0.059972621500492096, "global_step": 45081, "epoch": 425, "lr": 5.406697259033744e-06} +{"train_loss": 0.015803851187229156, "grad_norm": 0.09017153829336166, "global_step": 45082, "epoch": 425, "lr": 5.405356826863367e-06} +{"train_loss": 0.016308682039380074, "grad_norm": 0.11443737894296646, "global_step": 45083, "epoch": 425, "lr": 5.404016551379182e-06} +{"train_loss": 0.014993153512477875, "grad_norm": 0.05873521417379379, "global_step": 45084, "epoch": 425, "lr": 5.402676432585879e-06} +{"train_loss": 0.013720562681555748, "grad_norm": 0.09055604040622711, "global_step": 45085, "epoch": 425, "lr": 5.401336470488189e-06} +{"train_loss": 0.015267627313733101, "grad_norm": 0.11907067149877548, "global_step": 45086, "epoch": 425, "lr": 5.399996665090795e-06} +{"train_loss": 0.014568556100130081, "grad_norm": 0.07881742715835571, "global_step": 45087, "epoch": 425, "lr": 5.398657016398418e-06} +{"train_loss": 0.014937859028577805, "grad_norm": 0.0699705109000206, "global_step": 45088, "epoch": 425, "lr": 5.397317524415768e-06} +{"train_loss": 0.015468333847820759, "grad_norm": 0.06914360076189041, "global_step": 45089, "epoch": 425, "lr": 5.395978189147555e-06} +{"train_loss": 0.017224900424480438, "grad_norm": 0.08318692445755005, "global_step": 45090, "epoch": 425, "lr": 5.3946390105984625e-06} +{"train_loss": 0.01361231692135334, "grad_norm": 0.06703902035951614, "global_step": 45091, "epoch": 425, "lr": 5.393299988773221e-06} +{"train_loss": 0.012677576392889023, "grad_norm": 0.10840070247650146, "global_step": 45092, "epoch": 425, "lr": 5.391961123676514e-06} +{"train_loss": 0.018200106918811798, "grad_norm": 0.08895213156938553, "global_step": 45093, "epoch": 425, "lr": 5.390622415313057e-06} +{"train_loss": 0.015044060535728931, "grad_norm": 0.09033656120300293, "global_step": 45094, "epoch": 425, "lr": 5.389283863687561e-06} +{"train_loss": 0.014427246525883675, "grad_norm": 0.08211302012205124, "global_step": 45095, "epoch": 425, "lr": 5.387945468804723e-06} +{"train_loss": 0.014456622302532196, "grad_norm": 0.07632164657115936, "global_step": 45096, "epoch": 425, "lr": 5.386607230669228e-06} +{"train_loss": 0.013707256875932217, "grad_norm": 0.0641765147447586, "global_step": 45097, "epoch": 425, "lr": 5.385269149285799e-06} +{"train_loss": 0.013430239632725716, "grad_norm": 0.07849294692277908, "global_step": 45098, "epoch": 425, "lr": 5.383931224659139e-06} +{"train_loss": 0.013571495190262794, "grad_norm": 0.08139914274215698, "global_step": 45099, "epoch": 425, "lr": 5.382593456793933e-06} +{"train_loss": 0.015386005863547325, "grad_norm": 0.06104531139135361, "global_step": 45100, "epoch": 425, "lr": 5.381255845694888e-06} +{"train_loss": 0.01572469435632229, "grad_norm": 0.08410630375146866, "global_step": 45101, "epoch": 425, "lr": 5.379918391366723e-06} +{"train_loss": 0.015198865905404091, "grad_norm": 0.07355540990829468, "global_step": 45102, "epoch": 425, "lr": 5.378581093814111e-06} +{"train_loss": 0.01259339414536953, "grad_norm": 0.13060833513736725, "global_step": 45103, "epoch": 425, "lr": 5.377243953041755e-06} +{"train_loss": 0.013261578977108002, "grad_norm": 0.06491784751415253, "global_step": 45104, "epoch": 425, "lr": 5.3759069690543565e-06} +{"train_loss": 0.014047030359506607, "grad_norm": 0.09217625111341476, "global_step": 45105, "epoch": 425, "lr": 5.3745701418566215e-06} +{"train_loss": 0.014919331297278404, "grad_norm": 0.09395688027143478, "global_step": 45106, "epoch": 425, "lr": 5.373233471453232e-06} +{"train_loss": 0.014397494494915009, "grad_norm": 0.07291679084300995, "global_step": 45107, "epoch": 425, "lr": 5.371896957848904e-06} +{"train_loss": 0.014461224898695946, "grad_norm": 0.10449600964784622, "global_step": 45108, "epoch": 425, "lr": 5.3705606010483135e-06} +{"train_loss": 0.014851056039333344, "grad_norm": 0.05949273705482483, "global_step": 45109, "epoch": 425, "lr": 5.369224401056172e-06} +{"train_loss": 0.013156905770301819, "grad_norm": 0.10667961835861206, "global_step": 45110, "epoch": 425, "lr": 5.367888357877154e-06} +{"train_loss": 0.018927160650491714, "grad_norm": 0.08554080873727798, "global_step": 45111, "epoch": 425, "lr": 5.366552471515967e-06} +{"train_loss": 0.016856368631124496, "grad_norm": 0.12570659816265106, "global_step": 45112, "epoch": 425, "lr": 5.365216741977325e-06} +{"train_loss": 0.013750063255429268, "grad_norm": 0.06234559789299965, "global_step": 45113, "epoch": 425, "lr": 5.363881169265883e-06} +{"train_loss": 0.01677706278860569, "grad_norm": 0.07194064557552338, "global_step": 45114, "epoch": 425, "lr": 5.36254575338635e-06} +{"train_loss": 0.01503071840852499, "grad_norm": 0.07250785827636719, "global_step": 45115, "epoch": 425, "lr": 5.3612104943434276e-06} +{"train_loss": 0.0180031880736351, "grad_norm": 0.16033905744552612, "global_step": 45116, "epoch": 425, "lr": 5.359875392141783e-06} +{"train_loss": 0.01575050875544548, "grad_norm": 0.08279674500226974, "global_step": 45117, "epoch": 425, "lr": 5.358540446786131e-06} +{"train_loss": 0.01649581640958786, "grad_norm": 0.08610492944717407, "global_step": 45118, "epoch": 425, "lr": 5.357205658281162e-06} +{"train_loss": 0.01681569218635559, "grad_norm": 0.09497963637113571, "global_step": 45119, "epoch": 425, "lr": 5.355871026631554e-06} +{"train_loss": 0.015190397389233112, "grad_norm": 0.10224717855453491, "global_step": 45120, "epoch": 425, "lr": 5.354536551841993e-06} +{"train_loss": 0.013237068429589272, "grad_norm": 0.09675832837820053, "global_step": 45121, "epoch": 425, "lr": 5.353202233917176e-06} +{"train_loss": 0.015493512153625488, "grad_norm": 0.06478084623813629, "global_step": 45122, "epoch": 425, "lr": 5.351868072861799e-06} +{"train_loss": 0.01601487770676613, "grad_norm": 0.07666224241256714, "global_step": 45123, "epoch": 425, "lr": 5.350534068680529e-06} +{"train_loss": 0.013298716396093369, "grad_norm": 0.06297855079174042, "global_step": 45124, "epoch": 425, "lr": 5.349200221378076e-06} +{"train_loss": 0.015049805864691734, "grad_norm": 0.08761375397443771, "global_step": 45125, "epoch": 425, "lr": 5.347866530959106e-06} +{"train_loss": 0.016192354261875153, "grad_norm": 0.09557250142097473, "global_step": 45126, "epoch": 425, "lr": 5.346532997428321e-06} +{"train_loss": 0.016614247113466263, "grad_norm": 0.06987860053777695, "global_step": 45127, "epoch": 425, "lr": 5.345199620790392e-06} +{"train_loss": 0.013732349500060081, "grad_norm": 0.0629827231168747, "global_step": 45128, "epoch": 425, "lr": 5.343866401050012e-06} +{"train_loss": 0.01464888732880354, "grad_norm": 0.09749764204025269, "global_step": 45129, "epoch": 425, "lr": 5.342533338211875e-06} +{"train_loss": 0.014875547960400581, "grad_norm": 0.08062295615673065, "global_step": 45130, "epoch": 425, "lr": 5.341200432280657e-06} +{"train_loss": 0.014306359924376011, "grad_norm": 0.1323385238647461, "global_step": 45131, "epoch": 425, "lr": 5.339867683261024e-06} +{"train_loss": 0.015043837018311024, "grad_norm": 0.05957719683647156, "global_step": 45132, "epoch": 425, "lr": 5.338535091157687e-06} +{"train_loss": 0.011441787704825401, "grad_norm": 0.08634906262159348, "global_step": 45133, "epoch": 425, "lr": 5.337202655975304e-06} +{"train_loss": 0.014996184036135674, "grad_norm": 0.07813716679811478, "global_step": 45134, "epoch": 425, "lr": 5.3358703777185715e-06} +{"train_loss": 0.01596079394221306, "grad_norm": 0.06654208153486252, "global_step": 45135, "epoch": 425, "lr": 5.334538256392169e-06} +{"train_loss": 0.013023357838392258, "grad_norm": 0.09790142625570297, "global_step": 45136, "epoch": 425, "lr": 5.3332062920007805e-06} +{"train_loss": 0.014926798641681671, "grad_norm": 0.10089421272277832, "global_step": 45137, "epoch": 425, "lr": 5.331874484549071e-06} +{"train_loss": 0.013794738799333572, "grad_norm": 0.1153956949710846, "global_step": 45138, "epoch": 425, "lr": 5.330542834041724e-06} +{"train_loss": 0.014025485143065453, "grad_norm": 0.09911314398050308, "global_step": 45139, "epoch": 425, "lr": 5.329211340483436e-06} +{"train_loss": 0.013537981547415257, "grad_norm": 0.08889568597078323, "global_step": 45140, "epoch": 425, "lr": 5.32788000387886e-06} +{"train_loss": 0.016001194715499878, "grad_norm": 0.07430262863636017, "global_step": 45141, "epoch": 425, "lr": 5.326548824232691e-06} +{"train_loss": 0.013139923103153706, "grad_norm": 0.06704779714345932, "global_step": 45142, "epoch": 425, "lr": 5.325217801549609e-06} +{"train_loss": 0.016061946749687195, "grad_norm": 0.09188605099916458, "global_step": 45143, "epoch": 425, "lr": 5.323886935834282e-06} +{"train_loss": 0.013740628957748413, "grad_norm": 0.1720857471227646, "global_step": 45144, "epoch": 425, "lr": 5.322556227091374e-06} +{"train_loss": 0.01466237474232912, "grad_norm": 0.07033749669790268, "global_step": 45145, "epoch": 425, "lr": 5.321225675325581e-06} +{"train_loss": 0.014158649370074272, "grad_norm": 0.05846032127737999, "global_step": 45146, "epoch": 425, "lr": 5.319895280541575e-06} +{"train_loss": 0.014990748837590218, "grad_norm": 0.11639925837516785, "global_step": 45147, "epoch": 425, "lr": 5.318565042744017e-06} +{"train_loss": 0.013855453580617905, "grad_norm": 0.09487254172563553, "global_step": 45148, "epoch": 425, "lr": 5.317234961937606e-06} +{"train_loss": 0.016832228749990463, "grad_norm": 0.07259120792150497, "global_step": 45149, "epoch": 425, "lr": 5.315905038126984e-06} +{"train_loss": 0.016324235126376152, "grad_norm": 0.07837880402803421, "global_step": 45150, "epoch": 425, "lr": 5.314575271316846e-06} +{"train_loss": 0.013371055945754051, "grad_norm": 0.0781351774930954, "global_step": 45151, "epoch": 425, "lr": 5.313245661511851e-06} +{"train_loss": 0.015445159748196602, "grad_norm": 0.10321756452322006, "global_step": 45152, "epoch": 425, "lr": 5.311916208716677e-06} +{"train_loss": 0.015527579002082348, "grad_norm": 0.07901270687580109, "global_step": 45153, "epoch": 425, "lr": 5.310586912936006e-06} +{"train_loss": 0.016493387520313263, "grad_norm": 0.09798521548509598, "global_step": 45154, "epoch": 425, "lr": 5.309257774174498e-06} +{"train_loss": 0.014987454600280748, "grad_norm": 0.1137736514210701, "global_step": 45155, "epoch": 425, "lr": 5.307928792436811e-06, "val_loss": 0.019421597942709923} +{"train_loss": 0.014641291461884975, "grad_norm": 0.091106116771698, "global_step": 45156, "epoch": 426, "lr": 5.306599967727638e-06} +{"train_loss": 0.0128469942137599, "grad_norm": 0.06370507180690765, "global_step": 45157, "epoch": 426, "lr": 5.305271300051623e-06} +{"train_loss": 0.014218103140592575, "grad_norm": 0.0893189087510109, "global_step": 45158, "epoch": 426, "lr": 5.303942789413452e-06} +{"train_loss": 0.014663683250546455, "grad_norm": 0.06474975496530533, "global_step": 45159, "epoch": 426, "lr": 5.302614435817793e-06} +{"train_loss": 0.01490093395113945, "grad_norm": 0.1266578882932663, "global_step": 45160, "epoch": 426, "lr": 5.301286239269315e-06} +{"train_loss": 0.015948349609971046, "grad_norm": 0.08157634735107422, "global_step": 45161, "epoch": 426, "lr": 5.299958199772664e-06} +{"train_loss": 0.014140959829092026, "grad_norm": 0.09131091833114624, "global_step": 45162, "epoch": 426, "lr": 5.29863031733252e-06} +{"train_loss": 0.015115160495042801, "grad_norm": 0.07650412619113922, "global_step": 45163, "epoch": 426, "lr": 5.2973025919535565e-06} +{"train_loss": 0.014492552727460861, "grad_norm": 0.08172433078289032, "global_step": 45164, "epoch": 426, "lr": 5.295975023640426e-06} +{"train_loss": 0.015244175679981709, "grad_norm": 0.06351979076862335, "global_step": 45165, "epoch": 426, "lr": 5.294647612397807e-06} +{"train_loss": 0.014974035322666168, "grad_norm": 0.06839220225811005, "global_step": 45166, "epoch": 426, "lr": 5.2933203582303414e-06} +{"train_loss": 0.016374867409467697, "grad_norm": 0.055564213544130325, "global_step": 45167, "epoch": 426, "lr": 5.2919932611427195e-06} +{"train_loss": 0.016060832887887955, "grad_norm": 0.13134197890758514, "global_step": 45168, "epoch": 426, "lr": 5.290666321139576e-06} +{"train_loss": 0.018137603998184204, "grad_norm": 0.08647000044584274, "global_step": 45169, "epoch": 426, "lr": 5.289339538225591e-06} +{"train_loss": 0.01598435640335083, "grad_norm": 0.09426244348287582, "global_step": 45170, "epoch": 426, "lr": 5.288012912405433e-06} +{"train_loss": 0.013266546651721, "grad_norm": 0.08691304922103882, "global_step": 45171, "epoch": 426, "lr": 5.286686443683753e-06} +{"train_loss": 0.015557866543531418, "grad_norm": 0.08183945715427399, "global_step": 45172, "epoch": 426, "lr": 5.285360132065198e-06} +{"train_loss": 0.013725237920880318, "grad_norm": 0.05636507272720337, "global_step": 45173, "epoch": 426, "lr": 5.284033977554453e-06} +{"train_loss": 0.014772615395486355, "grad_norm": 0.09575802832841873, "global_step": 45174, "epoch": 426, "lr": 5.28270798015616e-06} +{"train_loss": 0.01796063780784607, "grad_norm": 0.06972982734441757, "global_step": 45175, "epoch": 426, "lr": 5.28138213987498e-06} +{"train_loss": 0.015277130529284477, "grad_norm": 0.0682235062122345, "global_step": 45176, "epoch": 426, "lr": 5.280056456715582e-06} +{"train_loss": 0.015605822205543518, "grad_norm": 0.09377135336399078, "global_step": 45177, "epoch": 426, "lr": 5.278730930682624e-06} +{"train_loss": 0.01622462086379528, "grad_norm": 0.18080587685108185, "global_step": 45178, "epoch": 426, "lr": 5.277405561780746e-06} +{"train_loss": 0.014849533326923847, "grad_norm": 0.10247910767793655, "global_step": 45179, "epoch": 426, "lr": 5.276080350014612e-06} +{"train_loss": 0.01554131880402565, "grad_norm": 0.05624900758266449, "global_step": 45180, "epoch": 426, "lr": 5.274755295388895e-06} +{"train_loss": 0.015646737068891525, "grad_norm": 0.1267111450433731, "global_step": 45181, "epoch": 426, "lr": 5.2734303979082245e-06} +{"train_loss": 0.015656467527151108, "grad_norm": 0.09430496394634247, "global_step": 45182, "epoch": 426, "lr": 5.2721056575772694e-06} +{"train_loss": 0.016903910785913467, "grad_norm": 0.0755995586514473, "global_step": 45183, "epoch": 426, "lr": 5.270781074400699e-06} +{"train_loss": 0.019684210419654846, "grad_norm": 0.14657855033874512, "global_step": 45184, "epoch": 426, "lr": 5.269456648383147e-06} +{"train_loss": 0.01403137668967247, "grad_norm": 0.1358281373977661, "global_step": 45185, "epoch": 426, "lr": 5.268132379529261e-06} +{"train_loss": 0.01688797026872635, "grad_norm": 0.13347488641738892, "global_step": 45186, "epoch": 426, "lr": 5.266808267843704e-06} +{"train_loss": 0.013997549191117287, "grad_norm": 0.12539717555046082, "global_step": 45187, "epoch": 426, "lr": 5.265484313331137e-06} +{"train_loss": 0.01647019013762474, "grad_norm": 0.10378269851207733, "global_step": 45188, "epoch": 426, "lr": 5.264160515996197e-06} +{"train_loss": 0.015339276753365993, "grad_norm": 0.1291394680738449, "global_step": 45189, "epoch": 426, "lr": 5.262836875843552e-06} +{"train_loss": 0.01673520728945732, "grad_norm": 0.13486844301223755, "global_step": 45190, "epoch": 426, "lr": 5.261513392877843e-06} +{"train_loss": 0.016209296882152557, "grad_norm": 0.05994397774338722, "global_step": 45191, "epoch": 426, "lr": 5.2601900671037096e-06} +{"train_loss": 0.014221595600247383, "grad_norm": 0.06097109615802765, "global_step": 45192, "epoch": 426, "lr": 5.258866898525811e-06} +{"train_loss": 0.013361133635044098, "grad_norm": 0.0613659992814064, "global_step": 45193, "epoch": 426, "lr": 5.257543887148802e-06} +{"train_loss": 0.013811195269227028, "grad_norm": 0.096063993871212, "global_step": 45194, "epoch": 426, "lr": 5.256221032977321e-06} +{"train_loss": 0.013283726759254932, "grad_norm": 0.08638269454240799, "global_step": 45195, "epoch": 426, "lr": 5.254898336016029e-06} +{"train_loss": 0.016679946333169937, "grad_norm": 0.07409854233264923, "global_step": 45196, "epoch": 426, "lr": 5.253575796269555e-06} +{"train_loss": 0.0142825273796916, "grad_norm": 0.053074683994054794, "global_step": 45197, "epoch": 426, "lr": 5.252253413742569e-06} +{"train_loss": 0.01573193445801735, "grad_norm": 0.06333581358194351, "global_step": 45198, "epoch": 426, "lr": 5.25093118843969e-06} +{"train_loss": 0.01281738094985485, "grad_norm": 0.12455498427152634, "global_step": 45199, "epoch": 426, "lr": 5.249609120365578e-06} +{"train_loss": 0.014007395133376122, "grad_norm": 0.06872936338186264, "global_step": 45200, "epoch": 426, "lr": 5.2482872095248935e-06} +{"train_loss": 0.013011320494115353, "grad_norm": 0.060513824224472046, "global_step": 45201, "epoch": 426, "lr": 5.246965455922259e-06} +{"train_loss": 0.016180410981178284, "grad_norm": 0.08038116991519928, "global_step": 45202, "epoch": 426, "lr": 5.245643859562322e-06} +{"train_loss": 0.01685289479792118, "grad_norm": 0.11897604167461395, "global_step": 45203, "epoch": 426, "lr": 5.2443224204497266e-06} +{"train_loss": 0.015155825763940811, "grad_norm": 0.06216336414217949, "global_step": 45204, "epoch": 426, "lr": 5.243001138589127e-06} +{"train_loss": 0.01380578987300396, "grad_norm": 0.06079939752817154, "global_step": 45205, "epoch": 426, "lr": 5.2416800139851454e-06} +{"train_loss": 0.016146011650562286, "grad_norm": 0.08075679838657379, "global_step": 45206, "epoch": 426, "lr": 5.24035904664244e-06} +{"train_loss": 0.015495818108320236, "grad_norm": 0.053427621722221375, "global_step": 45207, "epoch": 426, "lr": 5.239038236565652e-06} +{"train_loss": 0.01350255124270916, "grad_norm": 0.09850133210420609, "global_step": 45208, "epoch": 426, "lr": 5.2377175837594216e-06} +{"train_loss": 0.012376225553452969, "grad_norm": 0.06438985466957092, "global_step": 45209, "epoch": 426, "lr": 5.236397088228373e-06} +{"train_loss": 0.014364991337060928, "grad_norm": 0.0776674672961235, "global_step": 45210, "epoch": 426, "lr": 5.235076749977164e-06} +{"train_loss": 0.013190500438213348, "grad_norm": 0.10536207258701324, "global_step": 45211, "epoch": 426, "lr": 5.233756569010429e-06} +{"train_loss": 0.01523849368095398, "grad_norm": 0.0705670714378357, "global_step": 45212, "epoch": 426, "lr": 5.2324365453328085e-06} +{"train_loss": 0.01535741612315178, "grad_norm": 0.07197485864162445, "global_step": 45213, "epoch": 426, "lr": 5.231116678948927e-06} +{"train_loss": 0.014059774577617645, "grad_norm": 0.07506529241800308, "global_step": 45214, "epoch": 426, "lr": 5.229796969863443e-06} +{"train_loss": 0.015855032950639725, "grad_norm": 0.06626780331134796, "global_step": 45215, "epoch": 426, "lr": 5.228477418080974e-06} +{"train_loss": 0.012962043285369873, "grad_norm": 0.07751726359128952, "global_step": 45216, "epoch": 426, "lr": 5.22715802360616e-06} +{"train_loss": 0.014072209596633911, "grad_norm": 0.08064967393875122, "global_step": 45217, "epoch": 426, "lr": 5.2258387864436546e-06} +{"train_loss": 0.014567590318620205, "grad_norm": 0.0920347198843956, "global_step": 45218, "epoch": 426, "lr": 5.22451970659808e-06} +{"train_loss": 0.013971604406833649, "grad_norm": 0.06109561026096344, "global_step": 45219, "epoch": 426, "lr": 5.223200784074062e-06} +{"train_loss": 0.013193851336836815, "grad_norm": 0.12172883749008179, "global_step": 45220, "epoch": 426, "lr": 5.22188201887624e-06} +{"train_loss": 0.013461398892104626, "grad_norm": 0.05875163897871971, "global_step": 45221, "epoch": 426, "lr": 5.220563411009261e-06} +{"train_loss": 0.016757424920797348, "grad_norm": 0.06678280234336853, "global_step": 45222, "epoch": 426, "lr": 5.219244960477743e-06} +{"train_loss": 0.018017208203673363, "grad_norm": 0.09633267670869827, "global_step": 45223, "epoch": 426, "lr": 5.217926667286316e-06} +{"train_loss": 0.016685940325260162, "grad_norm": 0.07891703397035599, "global_step": 45224, "epoch": 426, "lr": 5.216608531439637e-06} +{"train_loss": 0.01379586011171341, "grad_norm": 0.08332110941410065, "global_step": 45225, "epoch": 426, "lr": 5.215290552942314e-06} +{"train_loss": 0.015440646559000015, "grad_norm": 0.07793384790420532, "global_step": 45226, "epoch": 426, "lr": 5.213972731798977e-06} +{"train_loss": 0.015485311858355999, "grad_norm": 0.10157331079244614, "global_step": 45227, "epoch": 426, "lr": 5.212655068014261e-06} +{"train_loss": 0.013318551704287529, "grad_norm": 0.10312401503324509, "global_step": 45228, "epoch": 426, "lr": 5.211337561592811e-06} +{"train_loss": 0.013056576251983643, "grad_norm": 0.062060099095106125, "global_step": 45229, "epoch": 426, "lr": 5.21002021253923e-06} +{"train_loss": 0.012226765044033527, "grad_norm": 0.06245064735412598, "global_step": 45230, "epoch": 426, "lr": 5.208703020858169e-06} +{"train_loss": 0.015962231904268265, "grad_norm": 0.07217153906822205, "global_step": 45231, "epoch": 426, "lr": 5.207385986554248e-06} +{"train_loss": 0.013657906092703342, "grad_norm": 0.08975336700677872, "global_step": 45232, "epoch": 426, "lr": 5.206069109632084e-06} +{"train_loss": 0.014366919174790382, "grad_norm": 0.07418244332075119, "global_step": 45233, "epoch": 426, "lr": 5.204752390096307e-06} +{"train_loss": 0.014068758115172386, "grad_norm": 0.06607452034950256, "global_step": 45234, "epoch": 426, "lr": 5.203435827951564e-06} +{"train_loss": 0.016093233600258827, "grad_norm": 0.06396067142486572, "global_step": 45235, "epoch": 426, "lr": 5.2021194232024565e-06} +{"train_loss": 0.01617865264415741, "grad_norm": 0.06120827794075012, "global_step": 45236, "epoch": 426, "lr": 5.20080317585363e-06} +{"train_loss": 0.01374329999089241, "grad_norm": 0.05688049644231796, "global_step": 45237, "epoch": 426, "lr": 5.1994870859096825e-06} +{"train_loss": 0.014580706134438515, "grad_norm": 0.11280180513858795, "global_step": 45238, "epoch": 426, "lr": 5.19817115337527e-06} +{"train_loss": 0.012566533870995045, "grad_norm": 0.13710005581378937, "global_step": 45239, "epoch": 426, "lr": 5.196855378254989e-06} +{"train_loss": 0.015438685193657875, "grad_norm": 0.05805276334285736, "global_step": 45240, "epoch": 426, "lr": 5.19553976055347e-06} +{"train_loss": 0.014169459231197834, "grad_norm": 0.08287562429904938, "global_step": 45241, "epoch": 426, "lr": 5.1942243002753524e-06} +{"train_loss": 0.013731494545936584, "grad_norm": 0.06214781478047371, "global_step": 45242, "epoch": 426, "lr": 5.192908997425245e-06} +{"train_loss": 0.014978758990764618, "grad_norm": 0.08721638470888138, "global_step": 45243, "epoch": 426, "lr": 5.1915938520077585e-06} +{"train_loss": 0.014685582369565964, "grad_norm": 0.10765308141708374, "global_step": 45244, "epoch": 426, "lr": 5.190278864027526e-06} +{"train_loss": 0.014182835817337036, "grad_norm": 0.14324109256267548, "global_step": 45245, "epoch": 426, "lr": 5.188964033489179e-06} +{"train_loss": 0.014680545777082443, "grad_norm": 0.10919556766748428, "global_step": 45246, "epoch": 426, "lr": 5.187649360397312e-06} +{"train_loss": 0.01441105268895626, "grad_norm": 0.0667896568775177, "global_step": 45247, "epoch": 426, "lr": 5.186334844756552e-06} +{"train_loss": 0.017577551305294037, "grad_norm": 0.08137443661689758, "global_step": 45248, "epoch": 426, "lr": 5.185020486571551e-06} +{"train_loss": 0.013133573345839977, "grad_norm": 0.14659538865089417, "global_step": 45249, "epoch": 426, "lr": 5.1837062858468736e-06} +{"train_loss": 0.01780047081410885, "grad_norm": 0.08644933998584747, "global_step": 45250, "epoch": 426, "lr": 5.182392242587158e-06} +{"train_loss": 0.01585153117775917, "grad_norm": 0.056107256561517715, "global_step": 45251, "epoch": 426, "lr": 5.181078356797042e-06} +{"train_loss": 0.015903843566775322, "grad_norm": 0.14440235495567322, "global_step": 45252, "epoch": 426, "lr": 5.1797646284811155e-06} +{"train_loss": 0.015370827168226242, "grad_norm": 0.07097893208265305, "global_step": 45253, "epoch": 426, "lr": 5.178451057644001e-06} +{"train_loss": 0.014544788748025894, "grad_norm": 0.06470902264118195, "global_step": 45254, "epoch": 426, "lr": 5.17713764429033e-06} +{"train_loss": 0.012821577489376068, "grad_norm": 0.09565125405788422, "global_step": 45255, "epoch": 426, "lr": 5.175824388424699e-06} +{"train_loss": 0.01421067863702774, "grad_norm": 0.09573758393526077, "global_step": 45256, "epoch": 426, "lr": 5.174511290051726e-06} +{"train_loss": 0.014493320137262344, "grad_norm": 0.05853743106126785, "global_step": 45257, "epoch": 426, "lr": 5.173198349176017e-06} +{"train_loss": 0.017063122242689133, "grad_norm": 0.07879084348678589, "global_step": 45258, "epoch": 426, "lr": 5.171885565802209e-06} +{"train_loss": 0.017901092767715454, "grad_norm": 0.06506789475679398, "global_step": 45259, "epoch": 426, "lr": 5.170572939934893e-06} +{"train_loss": 0.0134904058650136, "grad_norm": 0.12288059294223785, "global_step": 45260, "epoch": 426, "lr": 5.1692604715786904e-06} +{"train_loss": 0.014974849237093949, "grad_norm": 0.20663093030452728, "global_step": 45261, "epoch": 426, "lr": 5.1679481607382065e-06, "val_loss": 0.01889212429523468} +{"train_loss": 0.014124438166618347, "grad_norm": 0.10303301364183426, "global_step": 45262, "epoch": 427, "lr": 5.166636007418063e-06} +{"train_loss": 0.011272226460278034, "grad_norm": 0.0918651595711708, "global_step": 45263, "epoch": 427, "lr": 5.1653240116228575e-06} +{"train_loss": 0.013802092522382736, "grad_norm": 0.07274313271045685, "global_step": 45264, "epoch": 427, "lr": 5.164012173357196e-06} +{"train_loss": 0.013972090557217598, "grad_norm": 0.09272240847349167, "global_step": 45265, "epoch": 427, "lr": 5.162700492625716e-06} +{"train_loss": 0.01343025267124176, "grad_norm": 0.13848285377025604, "global_step": 45266, "epoch": 427, "lr": 5.161388969433001e-06} +{"train_loss": 0.01319284550845623, "grad_norm": 0.0759473443031311, "global_step": 45267, "epoch": 427, "lr": 5.160077603783658e-06} +{"train_loss": 0.014982370659708977, "grad_norm": 0.06437192112207413, "global_step": 45268, "epoch": 427, "lr": 5.158766395682308e-06} +{"train_loss": 0.013730067759752274, "grad_norm": 0.07393603026866913, "global_step": 45269, "epoch": 427, "lr": 5.157455345133555e-06} +{"train_loss": 0.014693263918161392, "grad_norm": 0.06742795556783676, "global_step": 45270, "epoch": 427, "lr": 5.156144452141992e-06} +{"train_loss": 0.015062849037349224, "grad_norm": 0.06735498458147049, "global_step": 45271, "epoch": 427, "lr": 5.154833716712248e-06} +{"train_loss": 0.013312704861164093, "grad_norm": 0.0837794616818428, "global_step": 45272, "epoch": 427, "lr": 5.153523138848915e-06} +{"train_loss": 0.013797461055219173, "grad_norm": 0.09361416101455688, "global_step": 45273, "epoch": 427, "lr": 5.152212718556587e-06} +{"train_loss": 0.013168999925255775, "grad_norm": 0.057201605290174484, "global_step": 45274, "epoch": 427, "lr": 5.150902455839879e-06} +{"train_loss": 0.013713032007217407, "grad_norm": 0.08003085851669312, "global_step": 45275, "epoch": 427, "lr": 5.1495923507034086e-06} +{"train_loss": 0.013276432640850544, "grad_norm": 0.06961541622877121, "global_step": 45276, "epoch": 427, "lr": 5.148282403151755e-06} +{"train_loss": 0.015262177214026451, "grad_norm": 0.11549802124500275, "global_step": 45277, "epoch": 427, "lr": 5.146972613189538e-06} +{"train_loss": 0.015197593718767166, "grad_norm": 0.06965745240449905, "global_step": 45278, "epoch": 427, "lr": 5.145662980821347e-06} +{"train_loss": 0.013468815013766289, "grad_norm": 0.09276346117258072, "global_step": 45279, "epoch": 427, "lr": 5.144353506051797e-06} +{"train_loss": 0.01701243221759796, "grad_norm": 0.0723460465669632, "global_step": 45280, "epoch": 427, "lr": 5.143044188885477e-06} +{"train_loss": 0.015696773305535316, "grad_norm": 0.07457218319177628, "global_step": 45281, "epoch": 427, "lr": 5.141735029326989e-06} +{"train_loss": 0.014333391562104225, "grad_norm": 0.12672299146652222, "global_step": 45282, "epoch": 427, "lr": 5.14042602738094e-06} +{"train_loss": 0.014223495498299599, "grad_norm": 0.07296549528837204, "global_step": 45283, "epoch": 427, "lr": 5.139117183051934e-06} +{"train_loss": 0.015164419077336788, "grad_norm": 0.06193045526742935, "global_step": 45284, "epoch": 427, "lr": 5.137808496344543e-06} +{"train_loss": 0.015301673673093319, "grad_norm": 0.09020915627479553, "global_step": 45285, "epoch": 427, "lr": 5.136499967263386e-06} +{"train_loss": 0.013453654013574123, "grad_norm": 0.06471362709999084, "global_step": 45286, "epoch": 427, "lr": 5.135191595813072e-06} +{"train_loss": 0.01593189314007759, "grad_norm": 0.10203135758638382, "global_step": 45287, "epoch": 427, "lr": 5.133883381998167e-06} +{"train_loss": 0.018435083329677582, "grad_norm": 0.07139331847429276, "global_step": 45288, "epoch": 427, "lr": 5.1325753258232855e-06} +{"train_loss": 0.015130038373172283, "grad_norm": 0.10896417498588562, "global_step": 45289, "epoch": 427, "lr": 5.131267427293046e-06} +{"train_loss": 0.01560711394995451, "grad_norm": 0.1632644236087799, "global_step": 45290, "epoch": 427, "lr": 5.129959686411995e-06} +{"train_loss": 0.01381446048617363, "grad_norm": 0.0816824808716774, "global_step": 45291, "epoch": 427, "lr": 5.128652103184756e-06} +{"train_loss": 0.016046924516558647, "grad_norm": 0.04836076870560646, "global_step": 45292, "epoch": 427, "lr": 5.127344677615925e-06} +{"train_loss": 0.017045797780156136, "grad_norm": 0.07368454337120056, "global_step": 45293, "epoch": 427, "lr": 5.126037409710082e-06} +{"train_loss": 0.013195980340242386, "grad_norm": 0.06175953149795532, "global_step": 45294, "epoch": 427, "lr": 5.1247302994718305e-06} +{"train_loss": 0.01638881117105484, "grad_norm": 0.1241726279258728, "global_step": 45295, "epoch": 427, "lr": 5.12342334690577e-06} +{"train_loss": 0.013062214478850365, "grad_norm": 0.07678806781768799, "global_step": 45296, "epoch": 427, "lr": 5.122116552016487e-06} +{"train_loss": 0.014034904539585114, "grad_norm": 0.07250981777906418, "global_step": 45297, "epoch": 427, "lr": 5.120809914808555e-06} +{"train_loss": 0.014268936589360237, "grad_norm": 0.128173828125, "global_step": 45298, "epoch": 427, "lr": 5.119503435286576e-06} +{"train_loss": 0.015662185847759247, "grad_norm": 0.06023985147476196, "global_step": 45299, "epoch": 427, "lr": 5.1181971134551646e-06} +{"train_loss": 0.014887544326484203, "grad_norm": 0.08888272941112518, "global_step": 45300, "epoch": 427, "lr": 5.11689094931887e-06} +{"train_loss": 0.019257865846157074, "grad_norm": 0.06845197826623917, "global_step": 45301, "epoch": 427, "lr": 5.115584942882317e-06} +{"train_loss": 0.013611650094389915, "grad_norm": 0.07330328226089478, "global_step": 45302, "epoch": 427, "lr": 5.1142790941500655e-06} +{"train_loss": 0.01564411260187626, "grad_norm": 0.09075305610895157, "global_step": 45303, "epoch": 427, "lr": 5.112973403126731e-06} +{"train_loss": 0.016403190791606903, "grad_norm": 0.07799351215362549, "global_step": 45304, "epoch": 427, "lr": 5.111667869816872e-06} +{"train_loss": 0.012519998475909233, "grad_norm": 0.06280648708343506, "global_step": 45305, "epoch": 427, "lr": 5.110362494225096e-06} +{"train_loss": 0.015058722347021103, "grad_norm": 0.09579093009233475, "global_step": 45306, "epoch": 427, "lr": 5.109057276355989e-06} +{"train_loss": 0.015553154982626438, "grad_norm": 0.12800323963165283, "global_step": 45307, "epoch": 427, "lr": 5.107752216214134e-06} +{"train_loss": 0.01412738487124443, "grad_norm": 0.06886453181505203, "global_step": 45308, "epoch": 427, "lr": 5.106447313804108e-06} +{"train_loss": 0.015366003848612309, "grad_norm": 0.09180274605751038, "global_step": 45309, "epoch": 427, "lr": 5.105142569130494e-06} +{"train_loss": 0.017298823222517967, "grad_norm": 0.1605129837989807, "global_step": 45310, "epoch": 427, "lr": 5.103837982197901e-06} +{"train_loss": 0.01539075467735529, "grad_norm": 0.06065499782562256, "global_step": 45311, "epoch": 427, "lr": 5.102533553010891e-06} +{"train_loss": 0.013955261558294296, "grad_norm": 0.06609630584716797, "global_step": 45312, "epoch": 427, "lr": 5.101229281574055e-06} +{"train_loss": 0.015524778515100479, "grad_norm": 0.06378234922885895, "global_step": 45313, "epoch": 427, "lr": 5.099925167891978e-06} +{"train_loss": 0.012017764151096344, "grad_norm": 0.09206797927618027, "global_step": 45314, "epoch": 427, "lr": 5.098621211969223e-06} +{"train_loss": 0.01393821369856596, "grad_norm": 0.09181445837020874, "global_step": 45315, "epoch": 427, "lr": 5.097317413810388e-06} +{"train_loss": 0.017663083970546722, "grad_norm": 0.07883349061012268, "global_step": 45316, "epoch": 427, "lr": 5.0960137734200605e-06} +{"train_loss": 0.014665279537439346, "grad_norm": 0.08168212324380875, "global_step": 45317, "epoch": 427, "lr": 5.094710290802806e-06} +{"train_loss": 0.017741311341524124, "grad_norm": 0.08463788777589798, "global_step": 45318, "epoch": 427, "lr": 5.093406965963221e-06} +{"train_loss": 0.013660278171300888, "grad_norm": 0.085667684674263, "global_step": 45319, "epoch": 427, "lr": 5.0921037989058614e-06} +{"train_loss": 0.014132557436823845, "grad_norm": 0.09297806024551392, "global_step": 45320, "epoch": 427, "lr": 5.0908007896353304e-06} +{"train_loss": 0.015903327614068985, "grad_norm": 0.08449599146842957, "global_step": 45321, "epoch": 427, "lr": 5.0894979381561845e-06} +{"train_loss": 0.016534794121980667, "grad_norm": 0.06818763166666031, "global_step": 45322, "epoch": 427, "lr": 5.088195244473009e-06} +{"train_loss": 0.014705062843859196, "grad_norm": 0.056648965924978256, "global_step": 45323, "epoch": 427, "lr": 5.086892708590396e-06} +{"train_loss": 0.014660093933343887, "grad_norm": 0.08594498038291931, "global_step": 45324, "epoch": 427, "lr": 5.085590330512913e-06} +{"train_loss": 0.012484174221754074, "grad_norm": 0.06292205303907394, "global_step": 45325, "epoch": 427, "lr": 5.084288110245122e-06} +{"train_loss": 0.013637511059641838, "grad_norm": 0.10097895562648773, "global_step": 45326, "epoch": 427, "lr": 5.082986047791605e-06} +{"train_loss": 0.014412561431527138, "grad_norm": 0.07048450410366058, "global_step": 45327, "epoch": 427, "lr": 5.081684143156956e-06} +{"train_loss": 0.016956374049186707, "grad_norm": 0.060286495834589005, "global_step": 45328, "epoch": 427, "lr": 5.080382396345723e-06} +{"train_loss": 0.015815895050764084, "grad_norm": 0.06500057876110077, "global_step": 45329, "epoch": 427, "lr": 5.079080807362491e-06} +{"train_loss": 0.015575623139739037, "grad_norm": 0.06552023440599442, "global_step": 45330, "epoch": 427, "lr": 5.077779376211849e-06} +{"train_loss": 0.01507275179028511, "grad_norm": 0.07445870339870453, "global_step": 45331, "epoch": 427, "lr": 5.076478102898341e-06} +{"train_loss": 0.015151186846196651, "grad_norm": 0.07202695310115814, "global_step": 45332, "epoch": 427, "lr": 5.0751769874265495e-06} +{"train_loss": 0.014022385701537132, "grad_norm": 0.06311115622520447, "global_step": 45333, "epoch": 427, "lr": 5.073876029801056e-06} +{"train_loss": 0.012897582724690437, "grad_norm": 0.06591453403234482, "global_step": 45334, "epoch": 427, "lr": 5.072575230026416e-06} +{"train_loss": 0.014919833280146122, "grad_norm": 0.07073641568422318, "global_step": 45335, "epoch": 427, "lr": 5.071274588107211e-06} +{"train_loss": 0.01486210711300373, "grad_norm": 0.10083742439746857, "global_step": 45336, "epoch": 427, "lr": 5.069974104048014e-06} +{"train_loss": 0.015520529821515083, "grad_norm": 0.09119860082864761, "global_step": 45337, "epoch": 427, "lr": 5.0686737778533945e-06} +{"train_loss": 0.013133898377418518, "grad_norm": 0.061680618673563004, "global_step": 45338, "epoch": 427, "lr": 5.0673736095278975e-06} +{"train_loss": 0.013968873769044876, "grad_norm": 0.09077063947916031, "global_step": 45339, "epoch": 427, "lr": 5.066073599076115e-06} +{"train_loss": 0.01678093522787094, "grad_norm": 0.11039536446332932, "global_step": 45340, "epoch": 427, "lr": 5.064773746502615e-06} +{"train_loss": 0.013560160994529724, "grad_norm": 0.09557390213012695, "global_step": 45341, "epoch": 427, "lr": 5.06347405181195e-06} +{"train_loss": 0.014533665031194687, "grad_norm": 0.10357686877250671, "global_step": 45342, "epoch": 427, "lr": 5.062174515008705e-06} +{"train_loss": 0.015242896974086761, "grad_norm": 0.06028943508863449, "global_step": 45343, "epoch": 427, "lr": 5.060875136097426e-06} +{"train_loss": 0.016460614278912544, "grad_norm": 0.07036350667476654, "global_step": 45344, "epoch": 427, "lr": 5.059575915082698e-06} +{"train_loss": 0.01586499810218811, "grad_norm": 0.11472292989492416, "global_step": 45345, "epoch": 427, "lr": 5.058276851969063e-06} +{"train_loss": 0.016045337542891502, "grad_norm": 0.08535432070493698, "global_step": 45346, "epoch": 427, "lr": 5.056977946761104e-06} +{"train_loss": 0.015973545610904694, "grad_norm": 0.11413375288248062, "global_step": 45347, "epoch": 427, "lr": 5.0556791994633865e-06} +{"train_loss": 0.015989622101187706, "grad_norm": 0.10124611854553223, "global_step": 45348, "epoch": 427, "lr": 5.054380610080467e-06} +{"train_loss": 0.013942526653409004, "grad_norm": 0.06122412905097008, "global_step": 45349, "epoch": 427, "lr": 5.053082178616897e-06} +{"train_loss": 0.01563936471939087, "grad_norm": 0.10619675368070602, "global_step": 45350, "epoch": 427, "lr": 5.051783905077256e-06} +{"train_loss": 0.01377684623003006, "grad_norm": 0.07169271260499954, "global_step": 45351, "epoch": 427, "lr": 5.050485789466097e-06} +{"train_loss": 0.016932563856244087, "grad_norm": 0.1474144160747528, "global_step": 45352, "epoch": 427, "lr": 5.049187831787977e-06} +{"train_loss": 0.01604369282722473, "grad_norm": 0.09108416736125946, "global_step": 45353, "epoch": 427, "lr": 5.047890032047475e-06} +{"train_loss": 0.015728997066617012, "grad_norm": 0.13047243654727936, "global_step": 45354, "epoch": 427, "lr": 5.046592390249133e-06} +{"train_loss": 0.014714520424604416, "grad_norm": 0.07744701951742172, "global_step": 45355, "epoch": 427, "lr": 5.045294906397513e-06} +{"train_loss": 0.013714957982301712, "grad_norm": 0.08162034302949905, "global_step": 45356, "epoch": 427, "lr": 5.043997580497173e-06} +{"train_loss": 0.015323777683079243, "grad_norm": 0.08587629348039627, "global_step": 45357, "epoch": 427, "lr": 5.042700412552681e-06} +{"train_loss": 0.01664257049560547, "grad_norm": 0.06486325711011887, "global_step": 45358, "epoch": 427, "lr": 5.041403402568584e-06} +{"train_loss": 0.014577431604266167, "grad_norm": 0.12608693540096283, "global_step": 45359, "epoch": 427, "lr": 5.0401065505494445e-06} +{"train_loss": 0.01475951261818409, "grad_norm": 0.06464799493551254, "global_step": 45360, "epoch": 427, "lr": 5.038809856499821e-06} +{"train_loss": 0.0153062604367733, "grad_norm": 0.08782701194286346, "global_step": 45361, "epoch": 427, "lr": 5.03751332042427e-06} +{"train_loss": 0.01711270585656166, "grad_norm": 0.08288829028606415, "global_step": 45362, "epoch": 427, "lr": 5.0362169423273376e-06} +{"train_loss": 0.016182612627744675, "grad_norm": 0.07815362513065338, "global_step": 45363, "epoch": 427, "lr": 5.034920722213582e-06} +{"train_loss": 0.01513512060046196, "grad_norm": 0.0731499046087265, "global_step": 45364, "epoch": 427, "lr": 5.033624660087566e-06} +{"train_loss": 0.015862448140978813, "grad_norm": 0.06872016936540604, "global_step": 45365, "epoch": 427, "lr": 5.0323287559538416e-06} +{"train_loss": 0.016699645668268204, "grad_norm": 0.07353240996599197, "global_step": 45366, "epoch": 427, "lr": 5.031033009816949e-06} +{"train_loss": 0.014981367625296116, "grad_norm": 0.1494453102350235, "global_step": 45367, "epoch": 427, "lr": 5.029737421681446e-06, "val_loss": 0.019666185602545738} +{"train_loss": 0.012247399426996708, "grad_norm": 0.09538162499666214, "global_step": 45368, "epoch": 428, "lr": 5.0284419915519025e-06} +{"train_loss": 0.018017522990703583, "grad_norm": 0.11519283056259155, "global_step": 45369, "epoch": 428, "lr": 5.027146719432846e-06} +{"train_loss": 0.016477052122354507, "grad_norm": 0.08561781793832779, "global_step": 45370, "epoch": 428, "lr": 5.025851605328835e-06} +{"train_loss": 0.014827929437160492, "grad_norm": 0.06587546318769455, "global_step": 45371, "epoch": 428, "lr": 5.024556649244444e-06} +{"train_loss": 0.013052472844719887, "grad_norm": 0.11314865946769714, "global_step": 45372, "epoch": 428, "lr": 5.02326185118418e-06} +{"train_loss": 0.015757404267787933, "grad_norm": 0.07005175203084946, "global_step": 45373, "epoch": 428, "lr": 5.021967211152618e-06} +{"train_loss": 0.01551102101802826, "grad_norm": 0.08101782947778702, "global_step": 45374, "epoch": 428, "lr": 5.020672729154308e-06} +{"train_loss": 0.013173136860132217, "grad_norm": 0.07627061754465103, "global_step": 45375, "epoch": 428, "lr": 5.019378405193787e-06} +{"train_loss": 0.01450345292687416, "grad_norm": 0.06866207718849182, "global_step": 45376, "epoch": 428, "lr": 5.0180842392756054e-06} +{"train_loss": 0.01380370557308197, "grad_norm": 0.05706169456243515, "global_step": 45377, "epoch": 428, "lr": 5.0167902314043284e-06} +{"train_loss": 0.014337172731757164, "grad_norm": 0.09686440229415894, "global_step": 45378, "epoch": 428, "lr": 5.015496381584483e-06} +{"train_loss": 0.01305975392460823, "grad_norm": 0.05854716897010803, "global_step": 45379, "epoch": 428, "lr": 5.014202689820613e-06} +{"train_loss": 0.016964081674814224, "grad_norm": 0.10904815047979355, "global_step": 45380, "epoch": 428, "lr": 5.012909156117268e-06} +{"train_loss": 0.014266682788729668, "grad_norm": 0.10644462704658508, "global_step": 45381, "epoch": 428, "lr": 5.011615780479012e-06} +{"train_loss": 0.014339160174131393, "grad_norm": 0.07344430685043335, "global_step": 45382, "epoch": 428, "lr": 5.0103225629103576e-06} +{"train_loss": 0.015857629477977753, "grad_norm": 0.07118570804595947, "global_step": 45383, "epoch": 428, "lr": 5.009029503415874e-06} +{"train_loss": 0.01437840424478054, "grad_norm": 0.06475649029016495, "global_step": 45384, "epoch": 428, "lr": 5.0077366020000906e-06} +{"train_loss": 0.014673139899969101, "grad_norm": 0.08438576757907867, "global_step": 45385, "epoch": 428, "lr": 5.006443858667559e-06} +{"train_loss": 0.013440695591270924, "grad_norm": 0.10950752347707748, "global_step": 45386, "epoch": 428, "lr": 5.005151273422804e-06} +{"train_loss": 0.017870306968688965, "grad_norm": 0.11948616802692413, "global_step": 45387, "epoch": 428, "lr": 5.003858846270387e-06} +{"train_loss": 0.015608097426593304, "grad_norm": 0.06324369460344315, "global_step": 45388, "epoch": 428, "lr": 5.002566577214845e-06} +{"train_loss": 0.016488276422023773, "grad_norm": 0.12614311277866364, "global_step": 45389, "epoch": 428, "lr": 5.0012744662607195e-06} +{"train_loss": 0.012207917869091034, "grad_norm": 0.05869537219405174, "global_step": 45390, "epoch": 428, "lr": 4.999982513412532e-06} +{"train_loss": 0.012239949777722359, "grad_norm": 0.0561591312289238, "global_step": 45391, "epoch": 428, "lr": 4.998690718674848e-06} +{"train_loss": 0.013631032779812813, "grad_norm": 0.07119542360305786, "global_step": 45392, "epoch": 428, "lr": 4.9973990820521845e-06} +{"train_loss": 0.0169413760304451, "grad_norm": 0.06551466137170792, "global_step": 45393, "epoch": 428, "lr": 4.996107603549094e-06} +{"train_loss": 0.016961492598056793, "grad_norm": 0.10201229900121689, "global_step": 45394, "epoch": 428, "lr": 4.994816283170112e-06} +{"train_loss": 0.015682075172662735, "grad_norm": 0.10708920657634735, "global_step": 45395, "epoch": 428, "lr": 4.993525120919779e-06} +{"train_loss": 0.014690055511891842, "grad_norm": 0.07460135221481323, "global_step": 45396, "epoch": 428, "lr": 4.992234116802614e-06} +{"train_loss": 0.015191336162388325, "grad_norm": 0.13008049130439758, "global_step": 45397, "epoch": 428, "lr": 4.9909432708231614e-06} +{"train_loss": 0.014357230626046658, "grad_norm": 0.08938770741224289, "global_step": 45398, "epoch": 428, "lr": 4.989652582985976e-06} +{"train_loss": 0.014939336106181145, "grad_norm": 0.12416118383407593, "global_step": 45399, "epoch": 428, "lr": 4.988362053295564e-06} +{"train_loss": 0.016521291807293892, "grad_norm": 0.10921116173267365, "global_step": 45400, "epoch": 428, "lr": 4.9870716817564726e-06} +{"train_loss": 0.014583607204258442, "grad_norm": 0.07766157388687134, "global_step": 45401, "epoch": 428, "lr": 4.985781468373247e-06} +{"train_loss": 0.016025550663471222, "grad_norm": 0.0959799587726593, "global_step": 45402, "epoch": 428, "lr": 4.9844914131504106e-06} +{"train_loss": 0.014160980470478535, "grad_norm": 0.06496751308441162, "global_step": 45403, "epoch": 428, "lr": 4.983201516092484e-06} +{"train_loss": 0.0153430737555027, "grad_norm": 0.06385717540979385, "global_step": 45404, "epoch": 428, "lr": 4.9819117772040066e-06} +{"train_loss": 0.015276556834578514, "grad_norm": 0.07106190174818039, "global_step": 45405, "epoch": 428, "lr": 4.9806221964895264e-06} +{"train_loss": 0.016350001096725464, "grad_norm": 0.17440958321094513, "global_step": 45406, "epoch": 428, "lr": 4.979332773953549e-06} +{"train_loss": 0.016123393550515175, "grad_norm": 0.09815532714128494, "global_step": 45407, "epoch": 428, "lr": 4.9780435096006275e-06} +{"train_loss": 0.014040585607290268, "grad_norm": 0.08025752007961273, "global_step": 45408, "epoch": 428, "lr": 4.976754403435285e-06} +{"train_loss": 0.015099839307367802, "grad_norm": 0.06881587207317352, "global_step": 45409, "epoch": 428, "lr": 4.975465455462031e-06} +{"train_loss": 0.013383735902607441, "grad_norm": 0.08531035482883453, "global_step": 45410, "epoch": 428, "lr": 4.974176665685415e-06} +{"train_loss": 0.01466755848377943, "grad_norm": 0.08487054705619812, "global_step": 45411, "epoch": 428, "lr": 4.9728880341099625e-06} +{"train_loss": 0.015161501243710518, "grad_norm": 0.09744793176651001, "global_step": 45412, "epoch": 428, "lr": 4.971599560740209e-06} +{"train_loss": 0.01791245862841606, "grad_norm": 0.08527732640504837, "global_step": 45413, "epoch": 428, "lr": 4.970311245580672e-06} +{"train_loss": 0.012792868539690971, "grad_norm": 0.10975444316864014, "global_step": 45414, "epoch": 428, "lr": 4.969023088635866e-06} +{"train_loss": 0.012413917109370232, "grad_norm": 0.08028049767017365, "global_step": 45415, "epoch": 428, "lr": 4.9677350899103415e-06} +{"train_loss": 0.015484883449971676, "grad_norm": 0.20578911900520325, "global_step": 45416, "epoch": 428, "lr": 4.966447249408601e-06} +{"train_loss": 0.0158750768750906, "grad_norm": 0.07029984891414642, "global_step": 45417, "epoch": 428, "lr": 4.96515956713518e-06} +{"train_loss": 0.01576959528028965, "grad_norm": 0.050877053290605545, "global_step": 45418, "epoch": 428, "lr": 4.963872043094614e-06} +{"train_loss": 0.013193469494581223, "grad_norm": 0.07629460096359253, "global_step": 45419, "epoch": 428, "lr": 4.962584677291416e-06} +{"train_loss": 0.017943764105439186, "grad_norm": 0.08572869002819061, "global_step": 45420, "epoch": 428, "lr": 4.961297469730097e-06} +{"train_loss": 0.013018439523875713, "grad_norm": 0.06740787625312805, "global_step": 45421, "epoch": 428, "lr": 4.960010420415195e-06} +{"train_loss": 0.015310168266296387, "grad_norm": 0.0683492124080658, "global_step": 45422, "epoch": 428, "lr": 4.958723529351239e-06} +{"train_loss": 0.015196914784610271, "grad_norm": 0.08177600800991058, "global_step": 45423, "epoch": 428, "lr": 4.957436796542731e-06} +{"train_loss": 0.014331434853374958, "grad_norm": 0.06469381600618362, "global_step": 45424, "epoch": 428, "lr": 4.95615022199421e-06} +{"train_loss": 0.013154961168766022, "grad_norm": 0.102483831346035, "global_step": 45425, "epoch": 428, "lr": 4.954863805710175e-06} +{"train_loss": 0.01842454820871353, "grad_norm": 0.07340233772993088, "global_step": 45426, "epoch": 428, "lr": 4.953577547695176e-06} +{"train_loss": 0.014980852603912354, "grad_norm": 0.06055064499378204, "global_step": 45427, "epoch": 428, "lr": 4.9522914479537e-06} +{"train_loss": 0.015787463635206223, "grad_norm": 0.12011875957250595, "global_step": 45428, "epoch": 428, "lr": 4.9510055064902804e-06} +{"train_loss": 0.014871738851070404, "grad_norm": 0.06455565243959427, "global_step": 45429, "epoch": 428, "lr": 4.949719723309448e-06} +{"train_loss": 0.013943130150437355, "grad_norm": 0.16234135627746582, "global_step": 45430, "epoch": 428, "lr": 4.948434098415711e-06} +{"train_loss": 0.015393326990306377, "grad_norm": 0.079356350004673, "global_step": 45431, "epoch": 428, "lr": 4.9471486318135755e-06} +{"train_loss": 0.012759539298713207, "grad_norm": 0.06645751744508743, "global_step": 45432, "epoch": 428, "lr": 4.945863323507571e-06} +{"train_loss": 0.015360300429165363, "grad_norm": 0.08899645507335663, "global_step": 45433, "epoch": 428, "lr": 4.944578173502201e-06} +{"train_loss": 0.014514021575450897, "grad_norm": 0.10551923513412476, "global_step": 45434, "epoch": 428, "lr": 4.943293181801989e-06} +{"train_loss": 0.013237269595265388, "grad_norm": 0.10690727829933167, "global_step": 45435, "epoch": 428, "lr": 4.9420083484114645e-06} +{"train_loss": 0.012029709294438362, "grad_norm": 0.06073475629091263, "global_step": 45436, "epoch": 428, "lr": 4.940723673335123e-06} +{"train_loss": 0.01484400313347578, "grad_norm": 0.08405828475952148, "global_step": 45437, "epoch": 428, "lr": 4.939439156577475e-06} +{"train_loss": 0.01355756726115942, "grad_norm": 0.10214879363775253, "global_step": 45438, "epoch": 428, "lr": 4.938154798143035e-06} +{"train_loss": 0.014431340619921684, "grad_norm": 0.09984932839870453, "global_step": 45439, "epoch": 428, "lr": 4.9368705980363415e-06} +{"train_loss": 0.016214169561862946, "grad_norm": 0.08942088484764099, "global_step": 45440, "epoch": 428, "lr": 4.935586556261873e-06} +{"train_loss": 0.012364238500595093, "grad_norm": 0.09884792566299438, "global_step": 45441, "epoch": 428, "lr": 4.934302672824154e-06} +{"train_loss": 0.01537555642426014, "grad_norm": 0.14206911623477936, "global_step": 45442, "epoch": 428, "lr": 4.9330189477277074e-06} +{"train_loss": 0.014917878434062004, "grad_norm": 0.10371913015842438, "global_step": 45443, "epoch": 428, "lr": 4.931735380977032e-06} +{"train_loss": 0.014820625074207783, "grad_norm": 0.07194892317056656, "global_step": 45444, "epoch": 428, "lr": 4.930451972576633e-06} +{"train_loss": 0.016357846558094025, "grad_norm": 0.06240539252758026, "global_step": 45445, "epoch": 428, "lr": 4.9291687225310244e-06} +{"train_loss": 0.015044234693050385, "grad_norm": 0.11659036576747894, "global_step": 45446, "epoch": 428, "lr": 4.927885630844725e-06} +{"train_loss": 0.012206673622131348, "grad_norm": 0.06223565340042114, "global_step": 45447, "epoch": 428, "lr": 4.926602697522226e-06} +{"train_loss": 0.01370924897491932, "grad_norm": 0.1696571558713913, "global_step": 45448, "epoch": 428, "lr": 4.92531992256805e-06} +{"train_loss": 0.016315853223204613, "grad_norm": 0.10878199338912964, "global_step": 45449, "epoch": 428, "lr": 4.924037305986695e-06} +{"train_loss": 0.015387767925858498, "grad_norm": 0.1516103893518448, "global_step": 45450, "epoch": 428, "lr": 4.922754847782668e-06} +{"train_loss": 0.01561332494020462, "grad_norm": 0.10994428396224976, "global_step": 45451, "epoch": 428, "lr": 4.921472547960471e-06} +{"train_loss": 0.019723325967788696, "grad_norm": 0.09266575425863266, "global_step": 45452, "epoch": 428, "lr": 4.920190406524628e-06} +{"train_loss": 0.013868999667465687, "grad_norm": 0.06997188925743103, "global_step": 45453, "epoch": 428, "lr": 4.918908423479618e-06} +{"train_loss": 0.015422997996211052, "grad_norm": 0.07027784734964371, "global_step": 45454, "epoch": 428, "lr": 4.917626598829972e-06} +{"train_loss": 0.014758903533220291, "grad_norm": 0.0594438761472702, "global_step": 45455, "epoch": 428, "lr": 4.916344932580169e-06} +{"train_loss": 0.01516745239496231, "grad_norm": 0.1415276825428009, "global_step": 45456, "epoch": 428, "lr": 4.915063424734734e-06} +{"train_loss": 0.014909526333212852, "grad_norm": 0.0615798681974411, "global_step": 45457, "epoch": 428, "lr": 4.913782075298146e-06} +{"train_loss": 0.015599753707647324, "grad_norm": 0.09083713591098785, "global_step": 45458, "epoch": 428, "lr": 4.912500884274923e-06} +{"train_loss": 0.0141653036698699, "grad_norm": 0.07829678803682327, "global_step": 45459, "epoch": 428, "lr": 4.911219851669574e-06} +{"train_loss": 0.013439388945698738, "grad_norm": 0.11179419606924057, "global_step": 45460, "epoch": 428, "lr": 4.909938977486589e-06} +{"train_loss": 0.017887283116579056, "grad_norm": 0.07375667244195938, "global_step": 45461, "epoch": 428, "lr": 4.9086582617304645e-06} +{"train_loss": 0.014558176510035992, "grad_norm": 0.10423935949802399, "global_step": 45462, "epoch": 428, "lr": 4.907377704405697e-06} +{"train_loss": 0.014649859629571438, "grad_norm": 0.09213199466466904, "global_step": 45463, "epoch": 428, "lr": 4.906097305516811e-06} +{"train_loss": 0.014896588400006294, "grad_norm": 0.0663527175784111, "global_step": 45464, "epoch": 428, "lr": 4.904817065068274e-06} +{"train_loss": 0.01676974818110466, "grad_norm": 0.06790319830179214, "global_step": 45465, "epoch": 428, "lr": 4.9035369830646106e-06} +{"train_loss": 0.016931409016251564, "grad_norm": 0.10710618644952774, "global_step": 45466, "epoch": 428, "lr": 4.902257059510296e-06} +{"train_loss": 0.017118269577622414, "grad_norm": 0.09246402233839035, "global_step": 45467, "epoch": 428, "lr": 4.900977294409842e-06} +{"train_loss": 0.0139303645119071, "grad_norm": 0.07077296823263168, "global_step": 45468, "epoch": 428, "lr": 4.89969768776774e-06} +{"train_loss": 0.01477484218776226, "grad_norm": 0.07394732534885406, "global_step": 45469, "epoch": 428, "lr": 4.898418239588481e-06} +{"train_loss": 0.017032310366630554, "grad_norm": 0.08232077211141586, "global_step": 45470, "epoch": 428, "lr": 4.897138949876578e-06} +{"train_loss": 0.017584051936864853, "grad_norm": 0.08092288672924042, "global_step": 45471, "epoch": 428, "lr": 4.895859818636511e-06} +{"train_loss": 0.013179854489862919, "grad_norm": 0.0763881579041481, "global_step": 45472, "epoch": 428, "lr": 4.8945808458727704e-06} +{"train_loss": 0.014964058898600205, "grad_norm": 0.11997665464878082, "global_step": 45473, "epoch": 428, "lr": 4.893302031589864e-06, "val_loss": 0.019938813522458076} +{"train_loss": 0.015902355313301086, "grad_norm": 0.06617046147584915, "global_step": 45474, "epoch": 429, "lr": 4.89202337579227e-06} +{"train_loss": 0.016140930354595184, "grad_norm": 0.12371127307415009, "global_step": 45475, "epoch": 429, "lr": 4.8907448784844935e-06} +{"train_loss": 0.014952404424548149, "grad_norm": 0.09813663363456726, "global_step": 45476, "epoch": 429, "lr": 4.889466539671022e-06} +{"train_loss": 0.014672743156552315, "grad_norm": 0.05915239825844765, "global_step": 45477, "epoch": 429, "lr": 4.888188359356355e-06} +{"train_loss": 0.014708156697452068, "grad_norm": 0.06695234030485153, "global_step": 45478, "epoch": 429, "lr": 4.886910337544964e-06} +{"train_loss": 0.016428396105766296, "grad_norm": 0.0942499116063118, "global_step": 45479, "epoch": 429, "lr": 4.885632474241347e-06} +{"train_loss": 0.014524346217513084, "grad_norm": 0.1003015860915184, "global_step": 45480, "epoch": 429, "lr": 4.884354769450012e-06} +{"train_loss": 0.015553583391010761, "grad_norm": 0.07944949716329575, "global_step": 45481, "epoch": 429, "lr": 4.8830772231754256e-06} +{"train_loss": 0.012854167260229588, "grad_norm": 0.054479923099279404, "global_step": 45482, "epoch": 429, "lr": 4.88179983542208e-06} +{"train_loss": 0.014908719807863235, "grad_norm": 0.1153659075498581, "global_step": 45483, "epoch": 429, "lr": 4.880522606194482e-06} +{"train_loss": 0.014585339464247227, "grad_norm": 0.10084904730319977, "global_step": 45484, "epoch": 429, "lr": 4.879245535497101e-06} +{"train_loss": 0.015232502482831478, "grad_norm": 0.1110239028930664, "global_step": 45485, "epoch": 429, "lr": 4.877968623334422e-06} +{"train_loss": 0.012737685814499855, "grad_norm": 0.07551497220993042, "global_step": 45486, "epoch": 429, "lr": 4.876691869710942e-06} +{"train_loss": 0.012272270396351814, "grad_norm": 0.06675903499126434, "global_step": 45487, "epoch": 429, "lr": 4.8754152746311455e-06} +{"train_loss": 0.013406602665781975, "grad_norm": 0.06345917284488678, "global_step": 45488, "epoch": 429, "lr": 4.874138838099512e-06} +{"train_loss": 0.013461587950587273, "grad_norm": 0.07966817915439606, "global_step": 45489, "epoch": 429, "lr": 4.872862560120534e-06} +{"train_loss": 0.015299820341169834, "grad_norm": 0.06024272367358208, "global_step": 45490, "epoch": 429, "lr": 4.871586440698695e-06} +{"train_loss": 0.014738550409674644, "grad_norm": 0.08151739835739136, "global_step": 45491, "epoch": 429, "lr": 4.870310479838469e-06} +{"train_loss": 0.013456476852297783, "grad_norm": 0.07674987614154816, "global_step": 45492, "epoch": 429, "lr": 4.869034677544337e-06} +{"train_loss": 0.016504645347595215, "grad_norm": 0.07407379895448685, "global_step": 45493, "epoch": 429, "lr": 4.8677590338208064e-06} +{"train_loss": 0.012180890887975693, "grad_norm": 0.10212710499763489, "global_step": 45494, "epoch": 429, "lr": 4.866483548672335e-06} +{"train_loss": 0.013707072474062443, "grad_norm": 0.058978017419576645, "global_step": 45495, "epoch": 429, "lr": 4.865208222103418e-06} +{"train_loss": 0.01732456684112549, "grad_norm": 0.16692495346069336, "global_step": 45496, "epoch": 429, "lr": 4.863933054118519e-06} +{"train_loss": 0.01331360824406147, "grad_norm": 0.14786851406097412, "global_step": 45497, "epoch": 429, "lr": 4.862658044722146e-06} +{"train_loss": 0.015642359852790833, "grad_norm": 0.06826969981193542, "global_step": 45498, "epoch": 429, "lr": 4.8613831939187505e-06} +{"train_loss": 0.01745608076453209, "grad_norm": 0.08857733011245728, "global_step": 45499, "epoch": 429, "lr": 4.860108501712824e-06} +{"train_loss": 0.01623564213514328, "grad_norm": 0.049444373697042465, "global_step": 45500, "epoch": 429, "lr": 4.85883396810885e-06} +{"train_loss": 0.01520818006247282, "grad_norm": 0.09595557302236557, "global_step": 45501, "epoch": 429, "lr": 4.8575595931113104e-06} +{"train_loss": 0.014606336131691933, "grad_norm": 0.07140902429819107, "global_step": 45502, "epoch": 429, "lr": 4.856285376724662e-06} +{"train_loss": 0.01627180352807045, "grad_norm": 0.06282614916563034, "global_step": 45503, "epoch": 429, "lr": 4.85501131895339e-06} +{"train_loss": 0.014763050712645054, "grad_norm": 0.073349229991436, "global_step": 45504, "epoch": 429, "lr": 4.85373741980199e-06} +{"train_loss": 0.014078988693654537, "grad_norm": 0.0754127949476242, "global_step": 45505, "epoch": 429, "lr": 4.85246367927491e-06} +{"train_loss": 0.01466031651943922, "grad_norm": 0.09120848029851913, "global_step": 45506, "epoch": 429, "lr": 4.85119009737664e-06} +{"train_loss": 0.014107933267951012, "grad_norm": 0.05079717934131622, "global_step": 45507, "epoch": 429, "lr": 4.849916674111676e-06} +{"train_loss": 0.013704841025173664, "grad_norm": 0.08222459256649017, "global_step": 45508, "epoch": 429, "lr": 4.848643409484443e-06} +{"train_loss": 0.015376420691609383, "grad_norm": 0.10375899076461792, "global_step": 45509, "epoch": 429, "lr": 4.847370303499449e-06} +{"train_loss": 0.01591811701655388, "grad_norm": 0.0774572491645813, "global_step": 45510, "epoch": 429, "lr": 4.846097356161167e-06} +{"train_loss": 0.014564881101250648, "grad_norm": 0.09938319772481918, "global_step": 45511, "epoch": 429, "lr": 4.844824567474049e-06} +{"train_loss": 0.016618002206087112, "grad_norm": 0.10193610191345215, "global_step": 45512, "epoch": 429, "lr": 4.843551937442581e-06} +{"train_loss": 0.014524328522384167, "grad_norm": 0.10761696845293045, "global_step": 45513, "epoch": 429, "lr": 4.842279466071242e-06} +{"train_loss": 0.016631929203867912, "grad_norm": 0.12817440927028656, "global_step": 45514, "epoch": 429, "lr": 4.841007153364496e-06} +{"train_loss": 0.015737898647785187, "grad_norm": 0.07739842683076859, "global_step": 45515, "epoch": 429, "lr": 4.839734999326806e-06} +{"train_loss": 0.013409401290118694, "grad_norm": 0.07293175905942917, "global_step": 45516, "epoch": 429, "lr": 4.83846300396264e-06} +{"train_loss": 0.012063616886734962, "grad_norm": 0.05383562296628952, "global_step": 45517, "epoch": 429, "lr": 4.837191167276489e-06} +{"train_loss": 0.014992957003414631, "grad_norm": 0.07899688184261322, "global_step": 45518, "epoch": 429, "lr": 4.835919489272811e-06} +{"train_loss": 0.015622948296368122, "grad_norm": 0.0934867113828659, "global_step": 45519, "epoch": 429, "lr": 4.834647969956052e-06} +{"train_loss": 0.017684098333120346, "grad_norm": 0.06755521148443222, "global_step": 45520, "epoch": 429, "lr": 4.833376609330703e-06} +{"train_loss": 0.014494170434772968, "grad_norm": 0.08905918896198273, "global_step": 45521, "epoch": 429, "lr": 4.832105407401238e-06} +{"train_loss": 0.012197883799672127, "grad_norm": 0.07038140296936035, "global_step": 45522, "epoch": 429, "lr": 4.830834364172099e-06} +{"train_loss": 0.014125908724963665, "grad_norm": 0.0945739597082138, "global_step": 45523, "epoch": 429, "lr": 4.829563479647764e-06} +{"train_loss": 0.014510421082377434, "grad_norm": 0.10832992196083069, "global_step": 45524, "epoch": 429, "lr": 4.828292753832708e-06} +{"train_loss": 0.01226385124027729, "grad_norm": 0.06342918425798416, "global_step": 45525, "epoch": 429, "lr": 4.8270221867313896e-06} +{"train_loss": 0.015484355390071869, "grad_norm": 0.07717761397361755, "global_step": 45526, "epoch": 429, "lr": 4.825751778348258e-06} +{"train_loss": 0.017870109528303146, "grad_norm": 0.09252416342496872, "global_step": 45527, "epoch": 429, "lr": 4.824481528687791e-06} +{"train_loss": 0.014773378148674965, "grad_norm": 0.06477490812540054, "global_step": 45528, "epoch": 429, "lr": 4.8232114377544604e-06} +{"train_loss": 0.014753198251128197, "grad_norm": 0.057660240679979324, "global_step": 45529, "epoch": 429, "lr": 4.821941505552707e-06} +{"train_loss": 0.016898680478334427, "grad_norm": 0.11167959123849869, "global_step": 45530, "epoch": 429, "lr": 4.820671732087012e-06} +{"train_loss": 0.014019285328686237, "grad_norm": 0.06695195287466049, "global_step": 45531, "epoch": 429, "lr": 4.819402117361826e-06} +{"train_loss": 0.014172679744660854, "grad_norm": 0.06477588415145874, "global_step": 45532, "epoch": 429, "lr": 4.818132661381608e-06} +{"train_loss": 0.015049499459564686, "grad_norm": 0.08813749998807907, "global_step": 45533, "epoch": 429, "lr": 4.81686336415082e-06} +{"train_loss": 0.013103501871228218, "grad_norm": 0.1520518660545349, "global_step": 45534, "epoch": 429, "lr": 4.8155942256739365e-06} +{"train_loss": 0.015085053630173206, "grad_norm": 0.06143850460648537, "global_step": 45535, "epoch": 429, "lr": 4.814325245955392e-06} +{"train_loss": 0.017720254138112068, "grad_norm": 0.08334626257419586, "global_step": 45536, "epoch": 429, "lr": 4.813056424999668e-06} +{"train_loss": 0.017025522887706757, "grad_norm": 0.12315376847982407, "global_step": 45537, "epoch": 429, "lr": 4.811787762811204e-06} +{"train_loss": 0.018803495913743973, "grad_norm": 0.14663328230381012, "global_step": 45538, "epoch": 429, "lr": 4.810519259394475e-06} +{"train_loss": 0.016279473900794983, "grad_norm": 0.11955954879522324, "global_step": 45539, "epoch": 429, "lr": 4.8092509147539155e-06} +{"train_loss": 0.012871943414211273, "grad_norm": 0.06338052451610565, "global_step": 45540, "epoch": 429, "lr": 4.8079827288940005e-06} +{"train_loss": 0.012054428458213806, "grad_norm": 0.04769972711801529, "global_step": 45541, "epoch": 429, "lr": 4.806714701819193e-06} +{"train_loss": 0.013120303861796856, "grad_norm": 0.06300723552703857, "global_step": 45542, "epoch": 429, "lr": 4.805446833533928e-06} +{"train_loss": 0.014478888362646103, "grad_norm": 0.0777774304151535, "global_step": 45543, "epoch": 429, "lr": 4.804179124042662e-06} +{"train_loss": 0.013951236382126808, "grad_norm": 0.06865038722753525, "global_step": 45544, "epoch": 429, "lr": 4.802911573349856e-06} +{"train_loss": 0.013095221482217312, "grad_norm": 0.09123729169368744, "global_step": 45545, "epoch": 429, "lr": 4.8016441814599744e-06} +{"train_loss": 0.015176799148321152, "grad_norm": 0.06860344856977463, "global_step": 45546, "epoch": 429, "lr": 4.800376948377449e-06} +{"train_loss": 0.014526605606079102, "grad_norm": 0.06958184391260147, "global_step": 45547, "epoch": 429, "lr": 4.799109874106738e-06} +{"train_loss": 0.0154513418674469, "grad_norm": 0.1229461282491684, "global_step": 45548, "epoch": 429, "lr": 4.797842958652321e-06} +{"train_loss": 0.013756725937128067, "grad_norm": 0.11631391197443008, "global_step": 45549, "epoch": 429, "lr": 4.796576202018605e-06} +{"train_loss": 0.01478886790573597, "grad_norm": 0.07878603786230087, "global_step": 45550, "epoch": 429, "lr": 4.7953096042100635e-06} +{"train_loss": 0.015424461103975773, "grad_norm": 0.12431249767541885, "global_step": 45551, "epoch": 429, "lr": 4.794043165231154e-06} +{"train_loss": 0.019211743026971817, "grad_norm": 0.1354595571756363, "global_step": 45552, "epoch": 429, "lr": 4.792776885086303e-06} +{"train_loss": 0.014920794405043125, "grad_norm": 0.06530579179525375, "global_step": 45553, "epoch": 429, "lr": 4.791510763779983e-06} +{"train_loss": 0.013330403715372086, "grad_norm": 0.14284898340702057, "global_step": 45554, "epoch": 429, "lr": 4.790244801316634e-06} +{"train_loss": 0.013930892571806908, "grad_norm": 0.16992726922035217, "global_step": 45555, "epoch": 429, "lr": 4.788978997700711e-06} +{"train_loss": 0.012806420214474201, "grad_norm": 0.058258503675460815, "global_step": 45556, "epoch": 429, "lr": 4.787713352936641e-06} +{"train_loss": 0.016860703006386757, "grad_norm": 0.10507403314113617, "global_step": 45557, "epoch": 429, "lr": 4.786447867028882e-06} +{"train_loss": 0.017473118379712105, "grad_norm": 0.14131461083889008, "global_step": 45558, "epoch": 429, "lr": 4.785182539981891e-06} +{"train_loss": 0.018276426941156387, "grad_norm": 0.17978355288505554, "global_step": 45559, "epoch": 429, "lr": 4.783917371800101e-06} +{"train_loss": 0.017200376838445663, "grad_norm": 0.16999836266040802, "global_step": 45560, "epoch": 429, "lr": 4.782652362487966e-06} +{"train_loss": 0.013794670812785625, "grad_norm": 0.0648455023765564, "global_step": 45561, "epoch": 429, "lr": 4.781387512049923e-06} +{"train_loss": 0.013888095505535603, "grad_norm": 0.07779005169868469, "global_step": 45562, "epoch": 429, "lr": 4.780122820490424e-06} +{"train_loss": 0.015079669654369354, "grad_norm": 0.05733611434698105, "global_step": 45563, "epoch": 429, "lr": 4.778858287813898e-06} +{"train_loss": 0.015644490718841553, "grad_norm": 0.08005614578723907, "global_step": 45564, "epoch": 429, "lr": 4.777593914024797e-06} +{"train_loss": 0.014277860522270203, "grad_norm": 0.06324774771928787, "global_step": 45565, "epoch": 429, "lr": 4.776329699127585e-06} +{"train_loss": 0.015462380833923817, "grad_norm": 0.10312361270189285, "global_step": 45566, "epoch": 429, "lr": 4.775065643126664e-06} +{"train_loss": 0.01348177157342434, "grad_norm": 0.08161196112632751, "global_step": 45567, "epoch": 429, "lr": 4.773801746026496e-06} +{"train_loss": 0.016066044569015503, "grad_norm": 0.06218830868601799, "global_step": 45568, "epoch": 429, "lr": 4.772538007831523e-06} +{"train_loss": 0.017577210441231728, "grad_norm": 0.06477046757936478, "global_step": 45569, "epoch": 429, "lr": 4.771274428546185e-06} +{"train_loss": 0.015803376212716103, "grad_norm": 0.09765705466270447, "global_step": 45570, "epoch": 429, "lr": 4.7700110081749125e-06} +{"train_loss": 0.014169612899422646, "grad_norm": 0.09235996752977371, "global_step": 45571, "epoch": 429, "lr": 4.768747746722163e-06} +{"train_loss": 0.015284519642591476, "grad_norm": 0.07312649488449097, "global_step": 45572, "epoch": 429, "lr": 4.767484644192361e-06} +{"train_loss": 0.013464286923408508, "grad_norm": 0.056225843727588654, "global_step": 45573, "epoch": 429, "lr": 4.76622170058994e-06} +{"train_loss": 0.015825094655156136, "grad_norm": 0.08633524924516678, "global_step": 45574, "epoch": 429, "lr": 4.764958915919338e-06} +{"train_loss": 0.0135801387950778, "grad_norm": 0.07205025106668472, "global_step": 45575, "epoch": 429, "lr": 4.763696290185015e-06} +{"train_loss": 0.01483746524900198, "grad_norm": 0.08514948189258575, "global_step": 45576, "epoch": 429, "lr": 4.7624338233913766e-06} +{"train_loss": 0.01831589639186859, "grad_norm": 0.09320063143968582, "global_step": 45577, "epoch": 429, "lr": 4.761171515542884e-06} +{"train_loss": 0.012783419340848923, "grad_norm": 0.058511435985565186, "global_step": 45578, "epoch": 429, "lr": 4.7599093666439494e-06} +{"train_loss": 0.014981435506411319, "grad_norm": 0.1906052976846695, "global_step": 45579, "epoch": 429, "lr": 4.758647376699032e-06, "val_loss": 0.019015012308955193} +{"train_loss": 0.014607500284910202, "grad_norm": 0.10861341655254364, "global_step": 45580, "epoch": 430, "lr": 4.757385545712539e-06} +{"train_loss": 0.01580914296209812, "grad_norm": 0.10103806108236313, "global_step": 45581, "epoch": 430, "lr": 4.756123873688917e-06} +{"train_loss": 0.015521623194217682, "grad_norm": 0.07821595668792725, "global_step": 45582, "epoch": 430, "lr": 4.7548623606326115e-06} +{"train_loss": 0.011833790689706802, "grad_norm": 0.07044952362775803, "global_step": 45583, "epoch": 430, "lr": 4.753601006548047e-06} +{"train_loss": 0.012786535546183586, "grad_norm": 0.058578282594680786, "global_step": 45584, "epoch": 430, "lr": 4.752339811439638e-06} +{"train_loss": 0.0146118039265275, "grad_norm": 0.06884053349494934, "global_step": 45585, "epoch": 430, "lr": 4.751078775311829e-06} +{"train_loss": 0.01666235364973545, "grad_norm": 0.08200721442699432, "global_step": 45586, "epoch": 430, "lr": 4.749817898169062e-06} +{"train_loss": 0.013711211271584034, "grad_norm": 0.08419569581747055, "global_step": 45587, "epoch": 430, "lr": 4.7485571800157445e-06} +{"train_loss": 0.01265166699886322, "grad_norm": 0.06287246942520142, "global_step": 45588, "epoch": 430, "lr": 4.7472966208563176e-06} +{"train_loss": 0.017874937504529953, "grad_norm": 0.07801131904125214, "global_step": 45589, "epoch": 430, "lr": 4.746036220695232e-06} +{"train_loss": 0.014961505308747292, "grad_norm": 0.10596463084220886, "global_step": 45590, "epoch": 430, "lr": 4.744775979536875e-06} +{"train_loss": 0.01554903108626604, "grad_norm": 0.07714392244815826, "global_step": 45591, "epoch": 430, "lr": 4.743515897385692e-06} +{"train_loss": 0.014563502743840218, "grad_norm": 0.06577999889850616, "global_step": 45592, "epoch": 430, "lr": 4.742255974246123e-06} +{"train_loss": 0.015629321336746216, "grad_norm": 0.06938339024782181, "global_step": 45593, "epoch": 430, "lr": 4.740996210122578e-06} +{"train_loss": 0.016653556376695633, "grad_norm": 0.11433815211057663, "global_step": 45594, "epoch": 430, "lr": 4.7397366050194845e-06} +{"train_loss": 0.016992758959531784, "grad_norm": 0.11200050264596939, "global_step": 45595, "epoch": 430, "lr": 4.738477158941285e-06} +{"train_loss": 0.016881611198186874, "grad_norm": 0.07030381262302399, "global_step": 45596, "epoch": 430, "lr": 4.737217871892397e-06} +{"train_loss": 0.01648668199777603, "grad_norm": 0.08852142840623856, "global_step": 45597, "epoch": 430, "lr": 4.735958743877228e-06} +{"train_loss": 0.016364632174372673, "grad_norm": 0.13753117620944977, "global_step": 45598, "epoch": 430, "lr": 4.734699774900214e-06} +{"train_loss": 0.01500683929771185, "grad_norm": 0.05564168840646744, "global_step": 45599, "epoch": 430, "lr": 4.733440964965791e-06} +{"train_loss": 0.016083110123872757, "grad_norm": 0.06939846277236938, "global_step": 45600, "epoch": 430, "lr": 4.732182314078359e-06} +{"train_loss": 0.014016465283930302, "grad_norm": 0.0759618729352951, "global_step": 45601, "epoch": 430, "lr": 4.730923822242367e-06} +{"train_loss": 0.014062687754631042, "grad_norm": 0.10894536226987839, "global_step": 45602, "epoch": 430, "lr": 4.729665489462204e-06} +{"train_loss": 0.01192840002477169, "grad_norm": 0.08738474547863007, "global_step": 45603, "epoch": 430, "lr": 4.728407315742323e-06} +{"train_loss": 0.014591041021049023, "grad_norm": 0.09302355349063873, "global_step": 45604, "epoch": 430, "lr": 4.727149301087125e-06} +{"train_loss": 0.013189651072025299, "grad_norm": 0.06672924757003784, "global_step": 45605, "epoch": 430, "lr": 4.7258914455010294e-06} +{"train_loss": 0.015031132847070694, "grad_norm": 0.14587366580963135, "global_step": 45606, "epoch": 430, "lr": 4.724633748988472e-06} +{"train_loss": 0.012806745246052742, "grad_norm": 0.12523147463798523, "global_step": 45607, "epoch": 430, "lr": 4.723376211553865e-06} +{"train_loss": 0.014145036228001118, "grad_norm": 0.10453879833221436, "global_step": 45608, "epoch": 430, "lr": 4.722118833201611e-06} +{"train_loss": 0.015999194234609604, "grad_norm": 0.0709214136004448, "global_step": 45609, "epoch": 430, "lr": 4.7208616139361564e-06} +{"train_loss": 0.015321469865739346, "grad_norm": 0.053736716508865356, "global_step": 45610, "epoch": 430, "lr": 4.719604553761886e-06} +{"train_loss": 0.014642680063843727, "grad_norm": 0.0803290456533432, "global_step": 45611, "epoch": 430, "lr": 4.718347652683236e-06} +{"train_loss": 0.014917283318936825, "grad_norm": 0.07283534854650497, "global_step": 45612, "epoch": 430, "lr": 4.71709091070463e-06} +{"train_loss": 0.013856434263288975, "grad_norm": 0.10456583648920059, "global_step": 45613, "epoch": 430, "lr": 4.71583432783047e-06} +{"train_loss": 0.014332801103591919, "grad_norm": 0.06709486991167068, "global_step": 45614, "epoch": 430, "lr": 4.714577904065171e-06} +{"train_loss": 0.01370272971689701, "grad_norm": 0.09246985614299774, "global_step": 45615, "epoch": 430, "lr": 4.7133216394131485e-06} +{"train_loss": 0.014970456250011921, "grad_norm": 0.16133619844913483, "global_step": 45616, "epoch": 430, "lr": 4.712065533878829e-06} +{"train_loss": 0.01582743600010872, "grad_norm": 0.0773250013589859, "global_step": 45617, "epoch": 430, "lr": 4.7108095874666025e-06} +{"train_loss": 0.01432325690984726, "grad_norm": 0.11887603253126144, "global_step": 45618, "epoch": 430, "lr": 4.70955380018091e-06} +{"train_loss": 0.013995060697197914, "grad_norm": 0.09885212033987045, "global_step": 45619, "epoch": 430, "lr": 4.708298172026138e-06} +{"train_loss": 0.013068989850580692, "grad_norm": 0.07145252078771591, "global_step": 45620, "epoch": 430, "lr": 4.70704270300672e-06} +{"train_loss": 0.014418946579098701, "grad_norm": 0.07527266442775726, "global_step": 45621, "epoch": 430, "lr": 4.7057873931270425e-06} +{"train_loss": 0.015163135714828968, "grad_norm": 0.09046351164579391, "global_step": 45622, "epoch": 430, "lr": 4.704532242391535e-06} +{"train_loss": 0.014693964272737503, "grad_norm": 0.07449725270271301, "global_step": 45623, "epoch": 430, "lr": 4.703277250804611e-06} +{"train_loss": 0.017202306538820267, "grad_norm": 0.12856487929821014, "global_step": 45624, "epoch": 430, "lr": 4.702022418370672e-06} +{"train_loss": 0.01539318822324276, "grad_norm": 0.060777030885219574, "global_step": 45625, "epoch": 430, "lr": 4.700767745094115e-06} +{"train_loss": 0.01459433976560831, "grad_norm": 0.08145904541015625, "global_step": 45626, "epoch": 430, "lr": 4.699513230979364e-06} +{"train_loss": 0.013690716587007046, "grad_norm": 0.06521778553724289, "global_step": 45627, "epoch": 430, "lr": 4.698258876030825e-06} +{"train_loss": 0.014931297861039639, "grad_norm": 0.07317408174276352, "global_step": 45628, "epoch": 430, "lr": 4.697004680252898e-06} +{"train_loss": 0.014267542399466038, "grad_norm": 0.1391730159521103, "global_step": 45629, "epoch": 430, "lr": 4.6957506436499985e-06} +{"train_loss": 0.015170680359005928, "grad_norm": 0.10732707381248474, "global_step": 45630, "epoch": 430, "lr": 4.6944967662265415e-06} +{"train_loss": 0.014010746031999588, "grad_norm": 0.07242587208747864, "global_step": 45631, "epoch": 430, "lr": 4.693243047986906e-06} +{"train_loss": 0.01402673963457346, "grad_norm": 0.0701291412115097, "global_step": 45632, "epoch": 430, "lr": 4.691989488935511e-06} +{"train_loss": 0.01241219136863947, "grad_norm": 0.0850866287946701, "global_step": 45633, "epoch": 430, "lr": 4.690736089076769e-06} +{"train_loss": 0.015861311927437782, "grad_norm": 0.08102118223905563, "global_step": 45634, "epoch": 430, "lr": 4.689482848415067e-06} +{"train_loss": 0.014844780787825584, "grad_norm": 0.09560756385326385, "global_step": 45635, "epoch": 430, "lr": 4.688229766954816e-06} +{"train_loss": 0.014809368178248405, "grad_norm": 0.18361744284629822, "global_step": 45636, "epoch": 430, "lr": 4.686976844700425e-06} +{"train_loss": 0.014729330316185951, "grad_norm": 0.06846670061349869, "global_step": 45637, "epoch": 430, "lr": 4.685724081656301e-06} +{"train_loss": 0.013994161039590836, "grad_norm": 0.11935703456401825, "global_step": 45638, "epoch": 430, "lr": 4.684471477826818e-06} +{"train_loss": 0.013992950320243835, "grad_norm": 0.08158577233552933, "global_step": 45639, "epoch": 430, "lr": 4.683219033216402e-06} +{"train_loss": 0.017886191606521606, "grad_norm": 0.1465473622083664, "global_step": 45640, "epoch": 430, "lr": 4.681966747829453e-06} +{"train_loss": 0.014171584509313107, "grad_norm": 0.08484353870153427, "global_step": 45641, "epoch": 430, "lr": 4.6807146216703515e-06} +{"train_loss": 0.01378648355603218, "grad_norm": 0.06545299291610718, "global_step": 45642, "epoch": 430, "lr": 4.679462654743521e-06} +{"train_loss": 0.015165967866778374, "grad_norm": 0.09071941673755646, "global_step": 45643, "epoch": 430, "lr": 4.678210847053344e-06} +{"train_loss": 0.01616317220032215, "grad_norm": 0.08034731447696686, "global_step": 45644, "epoch": 430, "lr": 4.676959198604231e-06} +{"train_loss": 0.018573278561234474, "grad_norm": 0.05951530113816261, "global_step": 45645, "epoch": 430, "lr": 4.675707709400562e-06} +{"train_loss": 0.012751108966767788, "grad_norm": 0.08085006475448608, "global_step": 45646, "epoch": 430, "lr": 4.674456379446751e-06} +{"train_loss": 0.015320521779358387, "grad_norm": 0.0720902532339096, "global_step": 45647, "epoch": 430, "lr": 4.673205208747189e-06} +{"train_loss": 0.013462692499160767, "grad_norm": 0.09432677179574966, "global_step": 45648, "epoch": 430, "lr": 4.671954197306278e-06} +{"train_loss": 0.013857598416507244, "grad_norm": 0.06916108727455139, "global_step": 45649, "epoch": 430, "lr": 4.670703345128396e-06} +{"train_loss": 0.01286410354077816, "grad_norm": 0.10999526083469391, "global_step": 45650, "epoch": 430, "lr": 4.669452652217959e-06} +{"train_loss": 0.01593800261616707, "grad_norm": 0.06405134499073029, "global_step": 45651, "epoch": 430, "lr": 4.668202118579346e-06} +{"train_loss": 0.015187829732894897, "grad_norm": 0.1458672434091568, "global_step": 45652, "epoch": 430, "lr": 4.666951744216952e-06} +{"train_loss": 0.01530851423740387, "grad_norm": 0.09168198704719543, "global_step": 45653, "epoch": 430, "lr": 4.665701529135186e-06} +{"train_loss": 0.015624670311808586, "grad_norm": 0.0865316092967987, "global_step": 45654, "epoch": 430, "lr": 4.664451473338427e-06} +{"train_loss": 0.016905948519706726, "grad_norm": 0.07259345054626465, "global_step": 45655, "epoch": 430, "lr": 4.663201576831061e-06} +{"train_loss": 0.019234750419855118, "grad_norm": 0.12283976376056671, "global_step": 45656, "epoch": 430, "lr": 4.66195183961749e-06} +{"train_loss": 0.01601540297269821, "grad_norm": 0.08208979666233063, "global_step": 45657, "epoch": 430, "lr": 4.6607022617021104e-06} +{"train_loss": 0.017053838819265366, "grad_norm": 0.0735754668712616, "global_step": 45658, "epoch": 430, "lr": 4.659452843089296e-06} +{"train_loss": 0.01642392948269844, "grad_norm": 0.05445732921361923, "global_step": 45659, "epoch": 430, "lr": 4.658203583783449e-06} +{"train_loss": 0.014916221611201763, "grad_norm": 0.1142563745379448, "global_step": 45660, "epoch": 430, "lr": 4.656954483788967e-06} +{"train_loss": 0.015269739553332329, "grad_norm": 0.09542138874530792, "global_step": 45661, "epoch": 430, "lr": 4.655705543110222e-06} +{"train_loss": 0.015473762527108192, "grad_norm": 0.0532132051885128, "global_step": 45662, "epoch": 430, "lr": 4.654456761751602e-06} +{"train_loss": 0.014214741066098213, "grad_norm": 0.0758318305015564, "global_step": 45663, "epoch": 430, "lr": 4.653208139717502e-06} +{"train_loss": 0.01348574087023735, "grad_norm": 0.057030610740184784, "global_step": 45664, "epoch": 430, "lr": 4.651959677012313e-06} +{"train_loss": 0.012518560513854027, "grad_norm": 0.14219941198825836, "global_step": 45665, "epoch": 430, "lr": 4.650711373640409e-06} +{"train_loss": 0.012359295040369034, "grad_norm": 0.08810172975063324, "global_step": 45666, "epoch": 430, "lr": 4.649463229606193e-06} +{"train_loss": 0.015061086975038052, "grad_norm": 0.08816748112440109, "global_step": 45667, "epoch": 430, "lr": 4.64821524491404e-06} +{"train_loss": 0.017154842615127563, "grad_norm": 0.14386247098445892, "global_step": 45668, "epoch": 430, "lr": 4.646967419568326e-06} +{"train_loss": 0.014824792742729187, "grad_norm": 0.12209141999483109, "global_step": 45669, "epoch": 430, "lr": 4.645719753573447e-06} +{"train_loss": 0.017252612859010696, "grad_norm": 0.08035115152597427, "global_step": 45670, "epoch": 430, "lr": 4.644472246933784e-06} +{"train_loss": 0.01627933420240879, "grad_norm": 0.13803556561470032, "global_step": 45671, "epoch": 430, "lr": 4.643224899653742e-06} +{"train_loss": 0.014969476498663425, "grad_norm": 0.09719723463058472, "global_step": 45672, "epoch": 430, "lr": 4.641977711737661e-06} +{"train_loss": 0.016543379053473473, "grad_norm": 0.13052745163440704, "global_step": 45673, "epoch": 430, "lr": 4.640730683189942e-06} +{"train_loss": 0.012658385559916496, "grad_norm": 0.1034342423081398, "global_step": 45674, "epoch": 430, "lr": 4.639483814014983e-06} +{"train_loss": 0.017049934715032578, "grad_norm": 0.1309175342321396, "global_step": 45675, "epoch": 430, "lr": 4.638237104217141e-06} +{"train_loss": 0.016971975564956665, "grad_norm": 0.10642000287771225, "global_step": 45676, "epoch": 430, "lr": 4.636990553800802e-06} +{"train_loss": 0.016035743057727814, "grad_norm": 0.05775028094649315, "global_step": 45677, "epoch": 430, "lr": 4.6357441627703666e-06} +{"train_loss": 0.015566376969218254, "grad_norm": 0.12658153474330902, "global_step": 45678, "epoch": 430, "lr": 4.634497931130194e-06} +{"train_loss": 0.013864567503333092, "grad_norm": 0.10484123975038528, "global_step": 45679, "epoch": 430, "lr": 4.633251858884657e-06} +{"train_loss": 0.013844684697687626, "grad_norm": 0.08049555122852325, "global_step": 45680, "epoch": 430, "lr": 4.632005946038143e-06} +{"train_loss": 0.016142280772328377, "grad_norm": 0.09710193425416946, "global_step": 45681, "epoch": 430, "lr": 4.630760192595041e-06} +{"train_loss": 0.012405455112457275, "grad_norm": 0.08646921813488007, "global_step": 45682, "epoch": 430, "lr": 4.6295145985597085e-06} +{"train_loss": 0.013034796342253685, "grad_norm": 0.0658147856593132, "global_step": 45683, "epoch": 430, "lr": 4.628269163936538e-06} +{"train_loss": 0.016287341713905334, "grad_norm": 0.07772636413574219, "global_step": 45684, "epoch": 430, "lr": 4.627023888729892e-06} +{"train_loss": 0.014979164493885243, "grad_norm": 0.10943082720041275, "global_step": 45685, "epoch": 430, "lr": 4.625778772944156e-06, "train_batch_mse_error": 0.013385452330112457, "val_loss": 0.019495639950037003} +{"train_loss": 0.014009320177137852, "grad_norm": 0.12131154537200928, "global_step": 45686, "epoch": 431, "lr": 4.624533816583698e-06} +{"train_loss": 0.0167531855404377, "grad_norm": 0.11491737514734268, "global_step": 45687, "epoch": 431, "lr": 4.623289019652887e-06} +{"train_loss": 0.016933085396885872, "grad_norm": 0.0951605811715126, "global_step": 45688, "epoch": 431, "lr": 4.622044382156121e-06} +{"train_loss": 0.015453454107046127, "grad_norm": 0.0734737291932106, "global_step": 45689, "epoch": 431, "lr": 4.620799904097756e-06} +{"train_loss": 0.012842909432947636, "grad_norm": 0.08486319333314896, "global_step": 45690, "epoch": 431, "lr": 4.619555585482149e-06} +{"train_loss": 0.017961077392101288, "grad_norm": 0.07011991739273071, "global_step": 45691, "epoch": 431, "lr": 4.618311426313704e-06} +{"train_loss": 0.015310368500649929, "grad_norm": 0.06734491884708405, "global_step": 45692, "epoch": 431, "lr": 4.6170674265967665e-06} +{"train_loss": 0.014318355359137058, "grad_norm": 0.07602538168430328, "global_step": 45693, "epoch": 431, "lr": 4.615823586335716e-06} +{"train_loss": 0.01596948504447937, "grad_norm": 0.07721517235040665, "global_step": 45694, "epoch": 431, "lr": 4.614579905534933e-06} +{"train_loss": 0.015445934608578682, "grad_norm": 0.10373350977897644, "global_step": 45695, "epoch": 431, "lr": 4.613336384198779e-06} +{"train_loss": 0.01747049018740654, "grad_norm": 0.08001967519521713, "global_step": 45696, "epoch": 431, "lr": 4.612093022331615e-06} +{"train_loss": 0.015075258910655975, "grad_norm": 0.09574331343173981, "global_step": 45697, "epoch": 431, "lr": 4.610849819937818e-06} +{"train_loss": 0.017337337136268616, "grad_norm": 0.06769502907991409, "global_step": 45698, "epoch": 431, "lr": 4.609606777021763e-06} +{"train_loss": 0.014975171536207199, "grad_norm": 0.12450214475393295, "global_step": 45699, "epoch": 431, "lr": 4.608363893587803e-06} +{"train_loss": 0.012076018378138542, "grad_norm": 0.09362918883562088, "global_step": 45700, "epoch": 431, "lr": 4.607121169640311e-06} +{"train_loss": 0.012262122705578804, "grad_norm": 0.07449760288000107, "global_step": 45701, "epoch": 431, "lr": 4.605878605183661e-06} +{"train_loss": 0.016149012371897697, "grad_norm": 0.11081872135400772, "global_step": 45702, "epoch": 431, "lr": 4.604636200222217e-06} +{"train_loss": 0.015349654480814934, "grad_norm": 0.12287788093090057, "global_step": 45703, "epoch": 431, "lr": 4.603393954760327e-06} +{"train_loss": 0.01445799320936203, "grad_norm": 0.06963035464286804, "global_step": 45704, "epoch": 431, "lr": 4.602151868802368e-06} +{"train_loss": 0.014625487849116325, "grad_norm": 0.06322957575321198, "global_step": 45705, "epoch": 431, "lr": 4.600909942352716e-06} +{"train_loss": 0.015981359407305717, "grad_norm": 0.061603110283613205, "global_step": 45706, "epoch": 431, "lr": 4.5996681754157115e-06} +{"train_loss": 0.014527378603816032, "grad_norm": 0.11251910775899887, "global_step": 45707, "epoch": 431, "lr": 4.598426567995734e-06} +{"train_loss": 0.01486968994140625, "grad_norm": 0.08972105383872986, "global_step": 45708, "epoch": 431, "lr": 4.597185120097147e-06} +{"train_loss": 0.016162611544132233, "grad_norm": 0.09175171703100204, "global_step": 45709, "epoch": 431, "lr": 4.595943831724298e-06} +{"train_loss": 0.01441233791410923, "grad_norm": 0.09215822070837021, "global_step": 45710, "epoch": 431, "lr": 4.594702702881554e-06} +{"train_loss": 0.017088908702135086, "grad_norm": 0.08004742860794067, "global_step": 45711, "epoch": 431, "lr": 4.593461733573285e-06} +{"train_loss": 0.014728618785738945, "grad_norm": 0.06833930313587189, "global_step": 45712, "epoch": 431, "lr": 4.5922209238038415e-06} +{"train_loss": 0.015023836866021156, "grad_norm": 0.13491816818714142, "global_step": 45713, "epoch": 431, "lr": 4.590980273577589e-06} +{"train_loss": 0.011366978287696838, "grad_norm": 0.08525276184082031, "global_step": 45714, "epoch": 431, "lr": 4.589739782898883e-06} +{"train_loss": 0.016460716724395752, "grad_norm": 0.06361806392669678, "global_step": 45715, "epoch": 431, "lr": 4.588499451772082e-06} +{"train_loss": 0.01657540537416935, "grad_norm": 0.10584784299135208, "global_step": 45716, "epoch": 431, "lr": 4.5872592802015445e-06} +{"train_loss": 0.015967799350619316, "grad_norm": 0.07417870312929153, "global_step": 45717, "epoch": 431, "lr": 4.586019268191627e-06} +{"train_loss": 0.014844304881989956, "grad_norm": 0.09434159100055695, "global_step": 45718, "epoch": 431, "lr": 4.584779415746693e-06} +{"train_loss": 0.01768277958035469, "grad_norm": 0.10481852293014526, "global_step": 45719, "epoch": 431, "lr": 4.583539722871094e-06} +{"train_loss": 0.015086453408002853, "grad_norm": 0.08236312121152878, "global_step": 45720, "epoch": 431, "lr": 4.582300189569177e-06} +{"train_loss": 0.016091594472527504, "grad_norm": 0.1591525375843048, "global_step": 45721, "epoch": 431, "lr": 4.581060815845306e-06} +{"train_loss": 0.012447704561054707, "grad_norm": 0.09725665301084518, "global_step": 45722, "epoch": 431, "lr": 4.5798216017038485e-06} +{"train_loss": 0.013671964406967163, "grad_norm": 0.08226355165243149, "global_step": 45723, "epoch": 431, "lr": 4.5785825471491295e-06} +{"train_loss": 0.013695178553462029, "grad_norm": 0.056650348007678986, "global_step": 45724, "epoch": 431, "lr": 4.577343652185534e-06} +{"train_loss": 0.015288587659597397, "grad_norm": 0.059059202671051025, "global_step": 45725, "epoch": 431, "lr": 4.576104916817381e-06} +{"train_loss": 0.015855953097343445, "grad_norm": 0.12122140824794769, "global_step": 45726, "epoch": 431, "lr": 4.574866341049056e-06} +{"train_loss": 0.012294751591980457, "grad_norm": 0.06347595900297165, "global_step": 45727, "epoch": 431, "lr": 4.5736279248848875e-06} +{"train_loss": 0.01570235565304756, "grad_norm": 0.05905332416296005, "global_step": 45728, "epoch": 431, "lr": 4.57238966832923e-06} +{"train_loss": 0.01421988196671009, "grad_norm": 0.08791472762823105, "global_step": 45729, "epoch": 431, "lr": 4.571151571386456e-06} +{"train_loss": 0.015587221831083298, "grad_norm": 0.06542481482028961, "global_step": 45730, "epoch": 431, "lr": 4.569913634060891e-06} +{"train_loss": 0.012803875841200352, "grad_norm": 0.06564775854349136, "global_step": 45731, "epoch": 431, "lr": 4.568675856356891e-06} +{"train_loss": 0.0165354423224926, "grad_norm": 0.08137615770101547, "global_step": 45732, "epoch": 431, "lr": 4.567438238278809e-06} +{"train_loss": 0.015548521652817726, "grad_norm": 0.06408856809139252, "global_step": 45733, "epoch": 431, "lr": 4.566200779830987e-06} +{"train_loss": 0.014434183947741985, "grad_norm": 0.06629787385463715, "global_step": 45734, "epoch": 431, "lr": 4.56496348101777e-06} +{"train_loss": 0.015054434537887573, "grad_norm": 0.13498425483703613, "global_step": 45735, "epoch": 431, "lr": 4.563726341843527e-06} +{"train_loss": 0.01436749380081892, "grad_norm": 0.08973454684019089, "global_step": 45736, "epoch": 431, "lr": 4.562489362312588e-06} +{"train_loss": 0.015795987099409103, "grad_norm": 0.12187200039625168, "global_step": 45737, "epoch": 431, "lr": 4.561252542429295e-06} +{"train_loss": 0.01746370457112789, "grad_norm": 0.08008001744747162, "global_step": 45738, "epoch": 431, "lr": 4.560015882197993e-06} +{"train_loss": 0.014724018052220345, "grad_norm": 0.06578823924064636, "global_step": 45739, "epoch": 431, "lr": 4.558779381623052e-06} +{"train_loss": 0.01411910355091095, "grad_norm": 0.07044554501771927, "global_step": 45740, "epoch": 431, "lr": 4.557543040708784e-06} +{"train_loss": 0.015389848500490189, "grad_norm": 0.11549292504787445, "global_step": 45741, "epoch": 431, "lr": 4.556306859459547e-06} +{"train_loss": 0.014310646802186966, "grad_norm": 0.05769700929522514, "global_step": 45742, "epoch": 431, "lr": 4.5550708378796935e-06} +{"train_loss": 0.013905378058552742, "grad_norm": 0.05747034028172493, "global_step": 45743, "epoch": 431, "lr": 4.553834975973559e-06} +{"train_loss": 0.014775916934013367, "grad_norm": 0.06471648812294006, "global_step": 45744, "epoch": 431, "lr": 4.5525992737454725e-06} +{"train_loss": 0.017120856791734695, "grad_norm": 0.10644324123859406, "global_step": 45745, "epoch": 431, "lr": 4.551363731199792e-06} +{"train_loss": 0.01513335108757019, "grad_norm": 0.07421913743019104, "global_step": 45746, "epoch": 431, "lr": 4.550128348340859e-06} +{"train_loss": 0.012948263436555862, "grad_norm": 0.051599230617284775, "global_step": 45747, "epoch": 431, "lr": 4.548893125173004e-06} +{"train_loss": 0.013811817392706871, "grad_norm": 0.0795079842209816, "global_step": 45748, "epoch": 431, "lr": 4.5476580617005825e-06} +{"train_loss": 0.014919636771082878, "grad_norm": 0.13625402748584747, "global_step": 45749, "epoch": 431, "lr": 4.546423157927921e-06} +{"train_loss": 0.012296313419938087, "grad_norm": 0.07326096296310425, "global_step": 45750, "epoch": 431, "lr": 4.5451884138593525e-06} +{"train_loss": 0.014473024755716324, "grad_norm": 0.08699963241815567, "global_step": 45751, "epoch": 431, "lr": 4.543953829499225e-06} +{"train_loss": 0.014404899440705776, "grad_norm": 0.07251983135938644, "global_step": 45752, "epoch": 431, "lr": 4.542719404851881e-06} +{"train_loss": 0.016620073467493057, "grad_norm": 0.06876867264509201, "global_step": 45753, "epoch": 431, "lr": 4.541485139921647e-06} +{"train_loss": 0.017152555286884308, "grad_norm": 0.0939018502831459, "global_step": 45754, "epoch": 431, "lr": 4.540251034712873e-06} +{"train_loss": 0.016212932765483856, "grad_norm": 0.06444499641656876, "global_step": 45755, "epoch": 431, "lr": 4.53901708922988e-06} +{"train_loss": 0.014840520918369293, "grad_norm": 0.07336382567882538, "global_step": 45756, "epoch": 431, "lr": 4.537783303477017e-06} +{"train_loss": 0.013992009684443474, "grad_norm": 0.07195211946964264, "global_step": 45757, "epoch": 431, "lr": 4.536549677458602e-06} +{"train_loss": 0.0156831294298172, "grad_norm": 0.08978866040706635, "global_step": 45758, "epoch": 431, "lr": 4.535316211178981e-06} +{"train_loss": 0.017250467091798782, "grad_norm": 0.10165289044380188, "global_step": 45759, "epoch": 431, "lr": 4.534082904642495e-06} +{"train_loss": 0.013646906241774559, "grad_norm": 0.12484946846961975, "global_step": 45760, "epoch": 431, "lr": 4.532849757853475e-06} +{"train_loss": 0.01504469569772482, "grad_norm": 0.07216218113899231, "global_step": 45761, "epoch": 431, "lr": 4.531616770816233e-06} +{"train_loss": 0.01398387085646391, "grad_norm": 0.07942331582307816, "global_step": 45762, "epoch": 431, "lr": 4.530383943535116e-06} +{"train_loss": 0.013198520988225937, "grad_norm": 0.12982012331485748, "global_step": 45763, "epoch": 431, "lr": 4.529151276014465e-06} +{"train_loss": 0.01684003323316574, "grad_norm": 0.10562723129987717, "global_step": 45764, "epoch": 431, "lr": 4.527918768258593e-06} +{"train_loss": 0.013454752042889595, "grad_norm": 0.06970347464084625, "global_step": 45765, "epoch": 431, "lr": 4.526686420271842e-06} +{"train_loss": 0.013875197619199753, "grad_norm": 0.08409253507852554, "global_step": 45766, "epoch": 431, "lr": 4.525454232058557e-06} +{"train_loss": 0.014537381939589977, "grad_norm": 0.0666390061378479, "global_step": 45767, "epoch": 431, "lr": 4.524222203623024e-06} +{"train_loss": 0.015082070603966713, "grad_norm": 0.10024202615022659, "global_step": 45768, "epoch": 431, "lr": 4.522990334969607e-06} +{"train_loss": 0.016364920884370804, "grad_norm": 0.06930027157068253, "global_step": 45769, "epoch": 431, "lr": 4.521758626102629e-06} +{"train_loss": 0.014769131317734718, "grad_norm": 0.06747642159461975, "global_step": 45770, "epoch": 431, "lr": 4.520527077026404e-06} +{"train_loss": 0.014534180983901024, "grad_norm": 0.06571609526872635, "global_step": 45771, "epoch": 431, "lr": 4.519295687745278e-06} +{"train_loss": 0.015324750915169716, "grad_norm": 0.061179615557193756, "global_step": 45772, "epoch": 431, "lr": 4.518064458263555e-06} +{"train_loss": 0.016731707379221916, "grad_norm": 0.07345865666866302, "global_step": 45773, "epoch": 431, "lr": 4.5168333885855895e-06} +{"train_loss": 0.013140425086021423, "grad_norm": 0.07986507564783096, "global_step": 45774, "epoch": 431, "lr": 4.5156024787156805e-06} +{"train_loss": 0.016766823828220367, "grad_norm": 0.09213533997535706, "global_step": 45775, "epoch": 431, "lr": 4.514371728658162e-06} +{"train_loss": 0.012230602093040943, "grad_norm": 0.052950676530599594, "global_step": 45776, "epoch": 431, "lr": 4.51314113841737e-06} +{"train_loss": 0.013111590407788754, "grad_norm": 0.06406500935554504, "global_step": 45777, "epoch": 431, "lr": 4.511910707997618e-06} +{"train_loss": 0.013512993231415749, "grad_norm": 0.06510359793901443, "global_step": 45778, "epoch": 431, "lr": 4.51068043740322e-06} +{"train_loss": 0.015187980607151985, "grad_norm": 0.12084171175956726, "global_step": 45779, "epoch": 431, "lr": 4.509450326638509e-06} +{"train_loss": 0.015350621193647385, "grad_norm": 0.08810125291347504, "global_step": 45780, "epoch": 431, "lr": 4.5082203757078114e-06} +{"train_loss": 0.013017305172979832, "grad_norm": 0.08048281818628311, "global_step": 45781, "epoch": 431, "lr": 4.506990584615439e-06} +{"train_loss": 0.014458425343036652, "grad_norm": 0.08650831878185272, "global_step": 45782, "epoch": 431, "lr": 4.505760953365718e-06} +{"train_loss": 0.01416279561817646, "grad_norm": 0.06617918610572815, "global_step": 45783, "epoch": 431, "lr": 4.504531481962976e-06} +{"train_loss": 0.01566525176167488, "grad_norm": 0.08387419581413269, "global_step": 45784, "epoch": 431, "lr": 4.503302170411522e-06} +{"train_loss": 0.013747288845479488, "grad_norm": 0.06685055792331696, "global_step": 45785, "epoch": 431, "lr": 4.502073018715675e-06} +{"train_loss": 0.016050871461629868, "grad_norm": 0.0849778801202774, "global_step": 45786, "epoch": 431, "lr": 4.5008440268797526e-06} +{"train_loss": 0.013965900056064129, "grad_norm": 0.07821154594421387, "global_step": 45787, "epoch": 431, "lr": 4.499615194908085e-06} +{"train_loss": 0.016186870634555817, "grad_norm": 0.07998456060886383, "global_step": 45788, "epoch": 431, "lr": 4.498386522804976e-06} +{"train_loss": 0.012787017971277237, "grad_norm": 0.07850771397352219, "global_step": 45789, "epoch": 431, "lr": 4.497158010574759e-06} +{"train_loss": 0.015706900507211685, "grad_norm": 0.07229387015104294, "global_step": 45790, "epoch": 431, "lr": 4.495929658221737e-06} +{"train_loss": 0.014985124114900827, "grad_norm": 0.09359496086835861, "global_step": 45791, "epoch": 431, "lr": 4.494701465750217e-06, "val_loss": 0.019496610388159752} +{"train_loss": 0.011447157710790634, "grad_norm": 0.12718403339385986, "global_step": 45792, "epoch": 432, "lr": 4.493473433164535e-06} +{"train_loss": 0.01595364511013031, "grad_norm": 0.12657976150512695, "global_step": 45793, "epoch": 432, "lr": 4.492245560469e-06} +{"train_loss": 0.014996696263551712, "grad_norm": 0.07575026899576187, "global_step": 45794, "epoch": 432, "lr": 4.491017847667917e-06} +{"train_loss": 0.015071345493197441, "grad_norm": 0.09564071893692017, "global_step": 45795, "epoch": 432, "lr": 4.489790294765611e-06} +{"train_loss": 0.015028391033411026, "grad_norm": 0.0863097533583641, "global_step": 45796, "epoch": 432, "lr": 4.488562901766386e-06} +{"train_loss": 0.013686012476682663, "grad_norm": 0.08241543918848038, "global_step": 45797, "epoch": 432, "lr": 4.487335668674564e-06} +{"train_loss": 0.01659255102276802, "grad_norm": 0.10115832090377808, "global_step": 45798, "epoch": 432, "lr": 4.486108595494443e-06} +{"train_loss": 0.01509331539273262, "grad_norm": 0.0820801854133606, "global_step": 45799, "epoch": 432, "lr": 4.484881682230341e-06} +{"train_loss": 0.014903559349477291, "grad_norm": 0.06808539479970932, "global_step": 45800, "epoch": 432, "lr": 4.4836549288865825e-06} +{"train_loss": 0.015457972884178162, "grad_norm": 0.0680570900440216, "global_step": 45801, "epoch": 432, "lr": 4.48242833546747e-06} +{"train_loss": 0.015447849407792091, "grad_norm": 0.05185996741056442, "global_step": 45802, "epoch": 432, "lr": 4.481201901977294e-06} +{"train_loss": 0.016066351905465126, "grad_norm": 0.10238438099622726, "global_step": 45803, "epoch": 432, "lr": 4.4799756284203795e-06} +{"train_loss": 0.014496324583888054, "grad_norm": 0.07921797782182693, "global_step": 45804, "epoch": 432, "lr": 4.478749514801045e-06} +{"train_loss": 0.01641741581261158, "grad_norm": 0.06557545810937881, "global_step": 45805, "epoch": 432, "lr": 4.477523561123581e-06} +{"train_loss": 0.015006471425294876, "grad_norm": 0.07593067735433578, "global_step": 45806, "epoch": 432, "lr": 4.4762977673923015e-06} +{"train_loss": 0.015769613906741142, "grad_norm": 0.07411394268274307, "global_step": 45807, "epoch": 432, "lr": 4.47507213361153e-06} +{"train_loss": 0.013567673973739147, "grad_norm": 0.06264886260032654, "global_step": 45808, "epoch": 432, "lr": 4.473846659785535e-06} +{"train_loss": 0.015916841104626656, "grad_norm": 0.06417039781808853, "global_step": 45809, "epoch": 432, "lr": 4.472621345918649e-06} +{"train_loss": 0.016251467168331146, "grad_norm": 0.0777025818824768, "global_step": 45810, "epoch": 432, "lr": 4.471396192015181e-06} +{"train_loss": 0.016032708808779716, "grad_norm": 0.09803358465433121, "global_step": 45811, "epoch": 432, "lr": 4.47017119807942e-06} +{"train_loss": 0.018150944262742996, "grad_norm": 0.07120808959007263, "global_step": 45812, "epoch": 432, "lr": 4.468946364115672e-06} +{"train_loss": 0.01580287702381611, "grad_norm": 0.06264512240886688, "global_step": 45813, "epoch": 432, "lr": 4.467721690128257e-06} +{"train_loss": 0.015967246145009995, "grad_norm": 0.08030465245246887, "global_step": 45814, "epoch": 432, "lr": 4.46649717612147e-06} +{"train_loss": 0.01432503666728735, "grad_norm": 0.0705643892288208, "global_step": 45815, "epoch": 432, "lr": 4.465272822099598e-06} +{"train_loss": 0.016315147280693054, "grad_norm": 0.07348044961690903, "global_step": 45816, "epoch": 432, "lr": 4.464048628066953e-06} +{"train_loss": 0.014490748755633831, "grad_norm": 0.0563623271882534, "global_step": 45817, "epoch": 432, "lr": 4.462824594027848e-06} +{"train_loss": 0.014950900338590145, "grad_norm": 0.10864882916212082, "global_step": 45818, "epoch": 432, "lr": 4.461600719986575e-06} +{"train_loss": 0.015079677104949951, "grad_norm": 0.07098744809627533, "global_step": 45819, "epoch": 432, "lr": 4.460377005947419e-06} +{"train_loss": 0.017459416761994362, "grad_norm": 0.06980834156274796, "global_step": 45820, "epoch": 432, "lr": 4.459153451914699e-06} +{"train_loss": 0.016578607261180878, "grad_norm": 0.06247193366289139, "global_step": 45821, "epoch": 432, "lr": 4.457930057892717e-06} +{"train_loss": 0.0149473175406456, "grad_norm": 0.08918603509664536, "global_step": 45822, "epoch": 432, "lr": 4.456706823885753e-06} +{"train_loss": 0.015987742692232132, "grad_norm": 0.12903380393981934, "global_step": 45823, "epoch": 432, "lr": 4.455483749898115e-06} +{"train_loss": 0.015823930501937866, "grad_norm": 0.07422886043787003, "global_step": 45824, "epoch": 432, "lr": 4.454260835934121e-06} +{"train_loss": 0.017066337168216705, "grad_norm": 0.07820736616849899, "global_step": 45825, "epoch": 432, "lr": 4.453038081998023e-06} +{"train_loss": 0.01735874079167843, "grad_norm": 0.084426149725914, "global_step": 45826, "epoch": 432, "lr": 4.451815488094147e-06} +{"train_loss": 0.012711566872894764, "grad_norm": 0.07543383538722992, "global_step": 45827, "epoch": 432, "lr": 4.450593054226782e-06} +{"train_loss": 0.015715081244707108, "grad_norm": 0.08268757909536362, "global_step": 45828, "epoch": 432, "lr": 4.44937078040023e-06} +{"train_loss": 0.01454052422195673, "grad_norm": 0.06183096766471863, "global_step": 45829, "epoch": 432, "lr": 4.448148666618774e-06} +{"train_loss": 0.013065936043858528, "grad_norm": 0.07266934216022491, "global_step": 45830, "epoch": 432, "lr": 4.446926712886717e-06} +{"train_loss": 0.014047695323824883, "grad_norm": 0.06452328711748123, "global_step": 45831, "epoch": 432, "lr": 4.445704919208354e-06} +{"train_loss": 0.014109982177615166, "grad_norm": 0.06490398943424225, "global_step": 45832, "epoch": 432, "lr": 4.4444832855879635e-06} +{"train_loss": 0.014490890316665173, "grad_norm": 0.060449302196502686, "global_step": 45833, "epoch": 432, "lr": 4.443261812029847e-06} +{"train_loss": 0.0142755676060915, "grad_norm": 0.1036423072218895, "global_step": 45834, "epoch": 432, "lr": 4.442040498538308e-06} +{"train_loss": 0.015572560019791126, "grad_norm": 0.07294439524412155, "global_step": 45835, "epoch": 432, "lr": 4.440819345117614e-06} +{"train_loss": 0.01409500278532505, "grad_norm": 0.07660622149705887, "global_step": 45836, "epoch": 432, "lr": 4.4395983517720784e-06} +{"train_loss": 0.012615760788321495, "grad_norm": 0.07570487260818481, "global_step": 45837, "epoch": 432, "lr": 4.438377518505965e-06} +{"train_loss": 0.01409339439123869, "grad_norm": 0.08028703927993774, "global_step": 45838, "epoch": 432, "lr": 4.437156845323598e-06} +{"train_loss": 0.0146554009988904, "grad_norm": 0.08102066069841385, "global_step": 45839, "epoch": 432, "lr": 4.435936332229229e-06} +{"train_loss": 0.015671107918024063, "grad_norm": 0.1251193732023239, "global_step": 45840, "epoch": 432, "lr": 4.434715979227172e-06} +{"train_loss": 0.015308482572436333, "grad_norm": 0.08556298166513443, "global_step": 45841, "epoch": 432, "lr": 4.433495786321712e-06} +{"train_loss": 0.014164187014102936, "grad_norm": 0.12062595039606094, "global_step": 45842, "epoch": 432, "lr": 4.432275753517129e-06} +{"train_loss": 0.015499897301197052, "grad_norm": 0.0730283185839653, "global_step": 45843, "epoch": 432, "lr": 4.4310558808177085e-06} +{"train_loss": 0.01588577777147293, "grad_norm": 0.06203638017177582, "global_step": 45844, "epoch": 432, "lr": 4.429836168227735e-06} +{"train_loss": 0.014539996162056923, "grad_norm": 0.0629739984869957, "global_step": 45845, "epoch": 432, "lr": 4.428616615751513e-06} +{"train_loss": 0.014505140483379364, "grad_norm": 0.058741334825754166, "global_step": 45846, "epoch": 432, "lr": 4.427397223393304e-06} +{"train_loss": 0.013590414077043533, "grad_norm": 0.06652791053056717, "global_step": 45847, "epoch": 432, "lr": 4.426177991157399e-06} +{"train_loss": 0.01460123062133789, "grad_norm": 0.07953884452581406, "global_step": 45848, "epoch": 432, "lr": 4.424958919048105e-06} +{"train_loss": 0.014375357888638973, "grad_norm": 0.0994524210691452, "global_step": 45849, "epoch": 432, "lr": 4.4237400070696656e-06} +{"train_loss": 0.015420119278132915, "grad_norm": 0.08784275501966476, "global_step": 45850, "epoch": 432, "lr": 4.422521255226386e-06} +{"train_loss": 0.012845104560256004, "grad_norm": 0.05442557483911514, "global_step": 45851, "epoch": 432, "lr": 4.421302663522553e-06} +{"train_loss": 0.014234304428100586, "grad_norm": 0.0837506353855133, "global_step": 45852, "epoch": 432, "lr": 4.420084231962435e-06} +{"train_loss": 0.01609194464981556, "grad_norm": 0.1046619638800621, "global_step": 45853, "epoch": 432, "lr": 4.418865960550317e-06} +{"train_loss": 0.01648956909775734, "grad_norm": 0.09347905218601227, "global_step": 45854, "epoch": 432, "lr": 4.417647849290485e-06} +{"train_loss": 0.01389092206954956, "grad_norm": 0.10458449274301529, "global_step": 45855, "epoch": 432, "lr": 4.416429898187224e-06} +{"train_loss": 0.013065599836409092, "grad_norm": 0.07705224305391312, "global_step": 45856, "epoch": 432, "lr": 4.415212107244787e-06} +{"train_loss": 0.01444670557975769, "grad_norm": 0.06516466289758682, "global_step": 45857, "epoch": 432, "lr": 4.413994476467475e-06} +{"train_loss": 0.013102245517075062, "grad_norm": 0.0560419000685215, "global_step": 45858, "epoch": 432, "lr": 4.41277700585957e-06} +{"train_loss": 0.01576758548617363, "grad_norm": 0.08010804653167725, "global_step": 45859, "epoch": 432, "lr": 4.411559695425333e-06} +{"train_loss": 0.015099659562110901, "grad_norm": 0.06208522990345955, "global_step": 45860, "epoch": 432, "lr": 4.410342545169061e-06} +{"train_loss": 0.015141520649194717, "grad_norm": 0.078313909471035, "global_step": 45861, "epoch": 432, "lr": 4.409125555095006e-06} +{"train_loss": 0.013656098395586014, "grad_norm": 0.13181397318840027, "global_step": 45862, "epoch": 432, "lr": 4.4079087252074654e-06} +{"train_loss": 0.014759372919797897, "grad_norm": 0.09290560334920883, "global_step": 45863, "epoch": 432, "lr": 4.4066920555106975e-06} +{"train_loss": 0.01890053227543831, "grad_norm": 0.07425554096698761, "global_step": 45864, "epoch": 432, "lr": 4.405475546008986e-06} +{"train_loss": 0.01351170800626278, "grad_norm": 0.11116276681423187, "global_step": 45865, "epoch": 432, "lr": 4.404259196706617e-06} +{"train_loss": 0.016306385397911072, "grad_norm": 0.08056536316871643, "global_step": 45866, "epoch": 432, "lr": 4.4030430076078485e-06} +{"train_loss": 0.014935893006622791, "grad_norm": 0.08035342395305634, "global_step": 45867, "epoch": 432, "lr": 4.40182697871695e-06} +{"train_loss": 0.016590159386396408, "grad_norm": 0.07526044547557831, "global_step": 45868, "epoch": 432, "lr": 4.400611110038205e-06} +{"train_loss": 0.014026753604412079, "grad_norm": 0.07320292294025421, "global_step": 45869, "epoch": 432, "lr": 4.3993954015758786e-06} +{"train_loss": 0.012730363756418228, "grad_norm": 0.051241159439086914, "global_step": 45870, "epoch": 432, "lr": 4.3981798533342445e-06} +{"train_loss": 0.0133592514321208, "grad_norm": 0.06633064150810242, "global_step": 45871, "epoch": 432, "lr": 4.396964465317582e-06} +{"train_loss": 0.012720300815999508, "grad_norm": 0.07104010134935379, "global_step": 45872, "epoch": 432, "lr": 4.395749237530161e-06} +{"train_loss": 0.014241054654121399, "grad_norm": 0.07578828930854797, "global_step": 45873, "epoch": 432, "lr": 4.394534169976227e-06} +{"train_loss": 0.014688698574900627, "grad_norm": 0.061119213700294495, "global_step": 45874, "epoch": 432, "lr": 4.393319262660073e-06} +{"train_loss": 0.016763679683208466, "grad_norm": 0.1079443171620369, "global_step": 45875, "epoch": 432, "lr": 4.392104515585965e-06} +{"train_loss": 0.013650815933942795, "grad_norm": 0.11457665264606476, "global_step": 45876, "epoch": 432, "lr": 4.390889928758163e-06} +{"train_loss": 0.015490030869841576, "grad_norm": 0.11980806291103363, "global_step": 45877, "epoch": 432, "lr": 4.38967550218094e-06} +{"train_loss": 0.014700613915920258, "grad_norm": 0.14083437621593475, "global_step": 45878, "epoch": 432, "lr": 4.38846123585856e-06} +{"train_loss": 0.0150733832269907, "grad_norm": 0.08298288285732269, "global_step": 45879, "epoch": 432, "lr": 4.3872471297952965e-06} +{"train_loss": 0.0136664854362607, "grad_norm": 0.05006641149520874, "global_step": 45880, "epoch": 432, "lr": 4.386033183995408e-06} +{"train_loss": 0.017937038093805313, "grad_norm": 0.09956160187721252, "global_step": 45881, "epoch": 432, "lr": 4.384819398463153e-06} +{"train_loss": 0.015569273382425308, "grad_norm": 0.07210490107536316, "global_step": 45882, "epoch": 432, "lr": 4.38360577320282e-06} +{"train_loss": 0.01602463610470295, "grad_norm": 0.06363038718700409, "global_step": 45883, "epoch": 432, "lr": 4.382392308218658e-06} +{"train_loss": 0.015501720830798149, "grad_norm": 0.09342970699071884, "global_step": 45884, "epoch": 432, "lr": 4.381179003514918e-06} +{"train_loss": 0.014293354004621506, "grad_norm": 0.07042687386274338, "global_step": 45885, "epoch": 432, "lr": 4.37996585909588e-06} +{"train_loss": 0.01763872615993023, "grad_norm": 0.0890730693936348, "global_step": 45886, "epoch": 432, "lr": 4.378752874965808e-06} +{"train_loss": 0.014433023519814014, "grad_norm": 0.05928296968340874, "global_step": 45887, "epoch": 432, "lr": 4.377540051128953e-06} +{"train_loss": 0.013467431999742985, "grad_norm": 0.04765024036169052, "global_step": 45888, "epoch": 432, "lr": 4.376327387589579e-06} +{"train_loss": 0.014700175262987614, "grad_norm": 0.07318572700023651, "global_step": 45889, "epoch": 432, "lr": 4.375114884351966e-06} +{"train_loss": 0.013705611228942871, "grad_norm": 0.06650571525096893, "global_step": 45890, "epoch": 432, "lr": 4.373902541420344e-06} +{"train_loss": 0.014815209433436394, "grad_norm": 0.10830068588256836, "global_step": 45891, "epoch": 432, "lr": 4.372690358798981e-06} +{"train_loss": 0.016320014372467995, "grad_norm": 0.06894361227750778, "global_step": 45892, "epoch": 432, "lr": 4.371478336492152e-06} +{"train_loss": 0.014451769180595875, "grad_norm": 0.09241867810487747, "global_step": 45893, "epoch": 432, "lr": 4.370266474504098e-06} +{"train_loss": 0.010290160775184631, "grad_norm": 0.05406247079372406, "global_step": 45894, "epoch": 432, "lr": 4.369054772839082e-06} +{"train_loss": 0.01599542424082756, "grad_norm": 0.08315781503915787, "global_step": 45895, "epoch": 432, "lr": 4.367843231501373e-06} +{"train_loss": 0.014677124097943306, "grad_norm": 0.09520315378904343, "global_step": 45896, "epoch": 432, "lr": 4.366631850495218e-06} +{"train_loss": 0.014970364690459561, "grad_norm": 0.08430942893028259, "global_step": 45897, "epoch": 432, "lr": 4.365420629824863e-06, "val_loss": 0.019275741651654243} +{"train_loss": 0.014252971857786179, "grad_norm": 0.16826707124710083, "global_step": 45898, "epoch": 433, "lr": 4.3642095694945766e-06} +{"train_loss": 0.013100964948534966, "grad_norm": 0.06232574209570885, "global_step": 45899, "epoch": 433, "lr": 4.362998669508617e-06} +{"train_loss": 0.01580028608441353, "grad_norm": 0.1101934164762497, "global_step": 45900, "epoch": 433, "lr": 4.361787929871225e-06} +{"train_loss": 0.01550072431564331, "grad_norm": 0.06245386600494385, "global_step": 45901, "epoch": 433, "lr": 4.36057735058667e-06} +{"train_loss": 0.01544523797929287, "grad_norm": 0.10759439319372177, "global_step": 45902, "epoch": 433, "lr": 4.359366931659192e-06} +{"train_loss": 0.017241688445210457, "grad_norm": 0.11223410069942474, "global_step": 45903, "epoch": 433, "lr": 4.358156673093061e-06} +{"train_loss": 0.015654517337679863, "grad_norm": 0.06513173878192902, "global_step": 45904, "epoch": 433, "lr": 4.356946574892506e-06} +{"train_loss": 0.012777569703757763, "grad_norm": 0.10801718384027481, "global_step": 45905, "epoch": 433, "lr": 4.355736637061791e-06} +{"train_loss": 0.015575828962028027, "grad_norm": 0.10794009268283844, "global_step": 45906, "epoch": 433, "lr": 4.354526859605173e-06} +{"train_loss": 0.014290251769125462, "grad_norm": 0.06312969326972961, "global_step": 45907, "epoch": 433, "lr": 4.353317242526905e-06} +{"train_loss": 0.01374734379351139, "grad_norm": 0.07679605484008789, "global_step": 45908, "epoch": 433, "lr": 4.352107785831216e-06} +{"train_loss": 0.016451401636004448, "grad_norm": 0.06306274980306625, "global_step": 45909, "epoch": 433, "lr": 4.350898489522376e-06} +{"train_loss": 0.014771305955946445, "grad_norm": 0.14719603955745697, "global_step": 45910, "epoch": 433, "lr": 4.349689353604619e-06} +{"train_loss": 0.01495741680264473, "grad_norm": 0.07163648307323456, "global_step": 45911, "epoch": 433, "lr": 4.348480378082198e-06} +{"train_loss": 0.01550763938575983, "grad_norm": 0.06445393711328506, "global_step": 45912, "epoch": 433, "lr": 4.3472715629593765e-06} +{"train_loss": 0.012713751755654812, "grad_norm": 0.06180733069777489, "global_step": 45913, "epoch": 433, "lr": 4.346062908240389e-06} +{"train_loss": 0.016708774492144585, "grad_norm": 0.14653988182544708, "global_step": 45914, "epoch": 433, "lr": 4.344854413929472e-06} +{"train_loss": 0.015299815684556961, "grad_norm": 0.0690639391541481, "global_step": 45915, "epoch": 433, "lr": 4.343646080030883e-06} +{"train_loss": 0.016074445098638535, "grad_norm": 0.07089181244373322, "global_step": 45916, "epoch": 433, "lr": 4.342437906548874e-06} +{"train_loss": 0.016261687502264977, "grad_norm": 0.08722761273384094, "global_step": 45917, "epoch": 433, "lr": 4.341229893487669e-06} +{"train_loss": 0.014619678258895874, "grad_norm": 0.053367309272289276, "global_step": 45918, "epoch": 433, "lr": 4.340022040851533e-06} +{"train_loss": 0.013462438248097897, "grad_norm": 0.07066630572080612, "global_step": 45919, "epoch": 433, "lr": 4.3388143486447045e-06} +{"train_loss": 0.015261917375028133, "grad_norm": 0.08885788917541504, "global_step": 45920, "epoch": 433, "lr": 4.337606816871431e-06} +{"train_loss": 0.01562926359474659, "grad_norm": 0.09734687954187393, "global_step": 45921, "epoch": 433, "lr": 4.336399445535938e-06} +{"train_loss": 0.014534097164869308, "grad_norm": 0.08955520391464233, "global_step": 45922, "epoch": 433, "lr": 4.335192234642477e-06} +{"train_loss": 0.015794169157743454, "grad_norm": 0.11207180470228195, "global_step": 45923, "epoch": 433, "lr": 4.3339851841952995e-06} +{"train_loss": 0.013120214454829693, "grad_norm": 0.07342663407325745, "global_step": 45924, "epoch": 433, "lr": 4.332778294198642e-06} +{"train_loss": 0.014949826523661613, "grad_norm": 0.06196404993534088, "global_step": 45925, "epoch": 433, "lr": 4.331571564656733e-06} +{"train_loss": 0.017056133598089218, "grad_norm": 0.09013058245182037, "global_step": 45926, "epoch": 433, "lr": 4.330364995573827e-06} +{"train_loss": 0.01679094135761261, "grad_norm": 0.09731573611497879, "global_step": 45927, "epoch": 433, "lr": 4.329158586954152e-06} +{"train_loss": 0.015900926664471626, "grad_norm": 0.06109267473220825, "global_step": 45928, "epoch": 433, "lr": 4.327952338801944e-06} +{"train_loss": 0.013126925565302372, "grad_norm": 0.058281924575567245, "global_step": 45929, "epoch": 433, "lr": 4.326746251121466e-06} +{"train_loss": 0.014506375417113304, "grad_norm": 0.1575256586074829, "global_step": 45930, "epoch": 433, "lr": 4.325540323916938e-06} +{"train_loss": 0.013734361156821251, "grad_norm": 0.06929445266723633, "global_step": 45931, "epoch": 433, "lr": 4.324334557192583e-06} +{"train_loss": 0.014197614043951035, "grad_norm": 0.1082773208618164, "global_step": 45932, "epoch": 433, "lr": 4.323128950952659e-06} +{"train_loss": 0.013719161041080952, "grad_norm": 0.10296367853879929, "global_step": 45933, "epoch": 433, "lr": 4.3219235052014025e-06} +{"train_loss": 0.01555736642330885, "grad_norm": 0.08633556962013245, "global_step": 45934, "epoch": 433, "lr": 4.3207182199430315e-06} +{"train_loss": 0.01320695411413908, "grad_norm": 0.06396719068288803, "global_step": 45935, "epoch": 433, "lr": 4.319513095181793e-06} +{"train_loss": 0.015494728460907936, "grad_norm": 0.09253539890050888, "global_step": 45936, "epoch": 433, "lr": 4.318308130921928e-06} +{"train_loss": 0.01624242216348648, "grad_norm": 0.07666400820016861, "global_step": 45937, "epoch": 433, "lr": 4.3171033271676655e-06} +{"train_loss": 0.013286890462040901, "grad_norm": 0.07027404755353928, "global_step": 45938, "epoch": 433, "lr": 4.315898683923219e-06} +{"train_loss": 0.015037992969155312, "grad_norm": 0.07645044475793839, "global_step": 45939, "epoch": 433, "lr": 4.314694201192842e-06} +{"train_loss": 0.015372645109891891, "grad_norm": 0.055201347917318344, "global_step": 45940, "epoch": 433, "lr": 4.3134898789807695e-06} +{"train_loss": 0.015758056193590164, "grad_norm": 0.07353027909994125, "global_step": 45941, "epoch": 433, "lr": 4.312285717291215e-06} +{"train_loss": 0.01260161679238081, "grad_norm": 0.0832333043217659, "global_step": 45942, "epoch": 433, "lr": 4.311081716128429e-06} +{"train_loss": 0.015702201053500175, "grad_norm": 0.11338871717453003, "global_step": 45943, "epoch": 433, "lr": 4.309877875496621e-06} +{"train_loss": 0.013210701756179333, "grad_norm": 0.12968795001506805, "global_step": 45944, "epoch": 433, "lr": 4.308674195400042e-06} +{"train_loss": 0.01600837893784046, "grad_norm": 0.12527760863304138, "global_step": 45945, "epoch": 433, "lr": 4.3074706758429065e-06} +{"train_loss": 0.015474517829716206, "grad_norm": 0.0928797796368599, "global_step": 45946, "epoch": 433, "lr": 4.306267316829443e-06} +{"train_loss": 0.013637719675898552, "grad_norm": 0.08030954748392105, "global_step": 45947, "epoch": 433, "lr": 4.305064118363894e-06} +{"train_loss": 0.0118020698428154, "grad_norm": 0.07676298171281815, "global_step": 45948, "epoch": 433, "lr": 4.303861080450483e-06} +{"train_loss": 0.014347027987241745, "grad_norm": 0.070161834359169, "global_step": 45949, "epoch": 433, "lr": 4.302658203093418e-06} +{"train_loss": 0.014655403792858124, "grad_norm": 0.15335294604301453, "global_step": 45950, "epoch": 433, "lr": 4.301455486296946e-06} +{"train_loss": 0.018345361575484276, "grad_norm": 0.13901996612548828, "global_step": 45951, "epoch": 433, "lr": 4.3002529300652794e-06} +{"train_loss": 0.015476902946829796, "grad_norm": 0.0687212198972702, "global_step": 45952, "epoch": 433, "lr": 4.2990505344026485e-06} +{"train_loss": 0.016658347100019455, "grad_norm": 0.08750804513692856, "global_step": 45953, "epoch": 433, "lr": 4.297848299313289e-06} +{"train_loss": 0.014953523874282837, "grad_norm": 0.09971238672733307, "global_step": 45954, "epoch": 433, "lr": 4.296646224801415e-06} +{"train_loss": 0.011723666451871395, "grad_norm": 0.14945897459983826, "global_step": 45955, "epoch": 433, "lr": 4.295444310871244e-06} +{"train_loss": 0.013995936140418053, "grad_norm": 0.06712716817855835, "global_step": 45956, "epoch": 433, "lr": 4.294242557527001e-06} +{"train_loss": 0.01249515451490879, "grad_norm": 0.11240880191326141, "global_step": 45957, "epoch": 433, "lr": 4.293040964772921e-06} +{"train_loss": 0.014212314039468765, "grad_norm": 0.06517861038446426, "global_step": 45958, "epoch": 433, "lr": 4.291839532613212e-06} +{"train_loss": 0.01293899491429329, "grad_norm": 0.06290972232818604, "global_step": 45959, "epoch": 433, "lr": 4.290638261052099e-06} +{"train_loss": 0.011203635483980179, "grad_norm": 0.05914812535047531, "global_step": 45960, "epoch": 433, "lr": 4.289437150093817e-06} +{"train_loss": 0.014009755104780197, "grad_norm": 0.07524731010198593, "global_step": 45961, "epoch": 433, "lr": 4.288236199742568e-06} +{"train_loss": 0.01858162134885788, "grad_norm": 0.10712587833404541, "global_step": 45962, "epoch": 433, "lr": 4.287035410002577e-06} +{"train_loss": 0.013217849656939507, "grad_norm": 0.07510188221931458, "global_step": 45963, "epoch": 433, "lr": 4.2858347808780564e-06} +{"train_loss": 0.01609155721962452, "grad_norm": 0.07056400924921036, "global_step": 45964, "epoch": 433, "lr": 4.284634312373242e-06} +{"train_loss": 0.015308316797018051, "grad_norm": 0.08756943047046661, "global_step": 45965, "epoch": 433, "lr": 4.283434004492332e-06} +{"train_loss": 0.016962159425020218, "grad_norm": 0.13004490733146667, "global_step": 45966, "epoch": 433, "lr": 4.282233857239565e-06} +{"train_loss": 0.015111641958355904, "grad_norm": 0.06200885772705078, "global_step": 45967, "epoch": 433, "lr": 4.281033870619144e-06} +{"train_loss": 0.014210434630513191, "grad_norm": 0.06969914585351944, "global_step": 45968, "epoch": 433, "lr": 4.2798340446352784e-06} +{"train_loss": 0.014688491821289062, "grad_norm": 0.09653536975383759, "global_step": 45969, "epoch": 433, "lr": 4.278634379292195e-06} +{"train_loss": 0.015185367316007614, "grad_norm": 0.06281519681215286, "global_step": 45970, "epoch": 433, "lr": 4.2774348745941165e-06} +{"train_loss": 0.01322668232023716, "grad_norm": 0.09853174537420273, "global_step": 45971, "epoch": 433, "lr": 4.276235530545242e-06} +{"train_loss": 0.014573676511645317, "grad_norm": 0.061193324625492096, "global_step": 45972, "epoch": 433, "lr": 4.2750363471497855e-06} +{"train_loss": 0.013566801324486732, "grad_norm": 0.08352821320295334, "global_step": 45973, "epoch": 433, "lr": 4.273837324411967e-06} +{"train_loss": 0.014598012901842594, "grad_norm": 0.05913021042943001, "global_step": 45974, "epoch": 433, "lr": 4.272638462336004e-06} +{"train_loss": 0.015329640358686447, "grad_norm": 0.06848523765802383, "global_step": 45975, "epoch": 433, "lr": 4.271439760926099e-06} +{"train_loss": 0.015412587672472, "grad_norm": 0.07384255528450012, "global_step": 45976, "epoch": 433, "lr": 4.270241220186466e-06} +{"train_loss": 0.015767116099596024, "grad_norm": 0.07035800814628601, "global_step": 45977, "epoch": 433, "lr": 4.269042840121324e-06} +{"train_loss": 0.014774307608604431, "grad_norm": 0.0895792618393898, "global_step": 45978, "epoch": 433, "lr": 4.267844620734879e-06} +{"train_loss": 0.012927769683301449, "grad_norm": 0.08314754068851471, "global_step": 45979, "epoch": 433, "lr": 4.266646562031334e-06} +{"train_loss": 0.014069268479943275, "grad_norm": 0.07138080894947052, "global_step": 45980, "epoch": 433, "lr": 4.265448664014899e-06} +{"train_loss": 0.01620795950293541, "grad_norm": 0.10388273000717163, "global_step": 45981, "epoch": 433, "lr": 4.2642509266898e-06} +{"train_loss": 0.015134861692786217, "grad_norm": 0.096436507999897, "global_step": 45982, "epoch": 433, "lr": 4.263053350060225e-06} +{"train_loss": 0.014775299467146397, "grad_norm": 0.05760085955262184, "global_step": 45983, "epoch": 433, "lr": 4.2618559341303984e-06} +{"train_loss": 0.014613174833357334, "grad_norm": 0.06193958967924118, "global_step": 45984, "epoch": 433, "lr": 4.260658678904511e-06} +{"train_loss": 0.013315736316144466, "grad_norm": 0.08707991242408752, "global_step": 45985, "epoch": 433, "lr": 4.2594615843867815e-06} +{"train_loss": 0.017115481197834015, "grad_norm": 0.16079245507717133, "global_step": 45986, "epoch": 433, "lr": 4.258264650581406e-06} +{"train_loss": 0.016254039481282234, "grad_norm": 0.06437964737415314, "global_step": 45987, "epoch": 433, "lr": 4.2570678774925985e-06} +{"train_loss": 0.012877221219241619, "grad_norm": 0.06604617089033127, "global_step": 45988, "epoch": 433, "lr": 4.255871265124567e-06} +{"train_loss": 0.015035199001431465, "grad_norm": 0.06347353011369705, "global_step": 45989, "epoch": 433, "lr": 4.254674813481513e-06} +{"train_loss": 0.016553469002246857, "grad_norm": 0.13319678604602814, "global_step": 45990, "epoch": 433, "lr": 4.253478522567628e-06} +{"train_loss": 0.017126070335507393, "grad_norm": 0.08359399437904358, "global_step": 45991, "epoch": 433, "lr": 4.25228239238713e-06} +{"train_loss": 0.016435949131846428, "grad_norm": 0.06866133958101273, "global_step": 45992, "epoch": 433, "lr": 4.251086422944211e-06} +{"train_loss": 0.01630379632115364, "grad_norm": 0.07575181126594543, "global_step": 45993, "epoch": 433, "lr": 4.24989061424308e-06} +{"train_loss": 0.016303017735481262, "grad_norm": 0.08221883326768875, "global_step": 45994, "epoch": 433, "lr": 4.2486949662879415e-06} +{"train_loss": 0.01921570673584938, "grad_norm": 0.1496981978416443, "global_step": 45995, "epoch": 433, "lr": 4.247499479083001e-06} +{"train_loss": 0.01692069321870804, "grad_norm": 0.13888424634933472, "global_step": 45996, "epoch": 433, "lr": 4.24630415263243e-06} +{"train_loss": 0.016836242750287056, "grad_norm": 0.08516792207956314, "global_step": 45997, "epoch": 433, "lr": 4.245108986940455e-06} +{"train_loss": 0.014683883637189865, "grad_norm": 0.07633891701698303, "global_step": 45998, "epoch": 433, "lr": 4.243913982011277e-06} +{"train_loss": 0.014134544879198074, "grad_norm": 0.08277569711208344, "global_step": 45999, "epoch": 433, "lr": 4.242719137849077e-06} +{"train_loss": 0.014097046107053757, "grad_norm": 0.07738494127988815, "global_step": 46000, "epoch": 433, "lr": 4.241524454458063e-06} +{"train_loss": 0.015160512179136276, "grad_norm": 0.08049878478050232, "global_step": 46001, "epoch": 433, "lr": 4.2403299318424405e-06} +{"train_loss": 0.01580650731921196, "grad_norm": 0.0805918499827385, "global_step": 46002, "epoch": 433, "lr": 4.2391355700063975e-06} +{"train_loss": 0.014971678101299506, "grad_norm": 0.07998499274253845, "global_step": 46003, "epoch": 433, "lr": 4.237941368954124e-06, "val_loss": 0.01933169551193714} +{"train_loss": 0.015004703775048256, "grad_norm": 0.06043560802936554, "global_step": 46004, "epoch": 434, "lr": 4.236747328689822e-06} +{"train_loss": 0.015573028475046158, "grad_norm": 0.0941057875752449, "global_step": 46005, "epoch": 434, "lr": 4.2355534492177e-06} +{"train_loss": 0.013931527733802795, "grad_norm": 0.05807216465473175, "global_step": 46006, "epoch": 434, "lr": 4.234359730541926e-06} +{"train_loss": 0.016549507156014442, "grad_norm": 0.06937927007675171, "global_step": 46007, "epoch": 434, "lr": 4.233166172666719e-06} +{"train_loss": 0.014675624668598175, "grad_norm": 0.06654927879571915, "global_step": 46008, "epoch": 434, "lr": 4.231972775596266e-06} +{"train_loss": 0.016160357743501663, "grad_norm": 0.12173572182655334, "global_step": 46009, "epoch": 434, "lr": 4.230779539334745e-06} +{"train_loss": 0.013555943965911865, "grad_norm": 0.05570609122514725, "global_step": 46010, "epoch": 434, "lr": 4.229586463886365e-06} +{"train_loss": 0.013226425275206566, "grad_norm": 0.08521754294633865, "global_step": 46011, "epoch": 434, "lr": 4.228393549255316e-06} +{"train_loss": 0.012506622821092606, "grad_norm": 0.07639485597610474, "global_step": 46012, "epoch": 434, "lr": 4.2272007954457794e-06} +{"train_loss": 0.015966134145855904, "grad_norm": 0.05832383781671524, "global_step": 46013, "epoch": 434, "lr": 4.226008202461967e-06} +{"train_loss": 0.014255352318286896, "grad_norm": 0.04753556847572327, "global_step": 46014, "epoch": 434, "lr": 4.224815770308038e-06} +{"train_loss": 0.014358711428940296, "grad_norm": 0.1019596979022026, "global_step": 46015, "epoch": 434, "lr": 4.223623498988216e-06} +{"train_loss": 0.012383498251438141, "grad_norm": 0.08123230189085007, "global_step": 46016, "epoch": 434, "lr": 4.222431388506659e-06} +{"train_loss": 0.0163780115544796, "grad_norm": 0.08729385584592819, "global_step": 46017, "epoch": 434, "lr": 4.221239438867569e-06} +{"train_loss": 0.01446584239602089, "grad_norm": 0.07983757555484772, "global_step": 46018, "epoch": 434, "lr": 4.220047650075149e-06} +{"train_loss": 0.016063373535871506, "grad_norm": 0.145024836063385, "global_step": 46019, "epoch": 434, "lr": 4.2188560221335715e-06} +{"train_loss": 0.0172134917229414, "grad_norm": 0.062447842210531235, "global_step": 46020, "epoch": 434, "lr": 4.217664555047013e-06} +{"train_loss": 0.013907819986343384, "grad_norm": 0.06686235964298248, "global_step": 46021, "epoch": 434, "lr": 4.216473248819669e-06} +{"train_loss": 0.014321006834506989, "grad_norm": 0.07166282087564468, "global_step": 46022, "epoch": 434, "lr": 4.215282103455742e-06} +{"train_loss": 0.015710799023509026, "grad_norm": 0.0726262554526329, "global_step": 46023, "epoch": 434, "lr": 4.21409111895939e-06} +{"train_loss": 0.013937879353761673, "grad_norm": 0.06049036234617233, "global_step": 46024, "epoch": 434, "lr": 4.212900295334821e-06} +{"train_loss": 0.01439758762717247, "grad_norm": 0.05926050245761871, "global_step": 46025, "epoch": 434, "lr": 4.211709632586203e-06} +{"train_loss": 0.01359343808144331, "grad_norm": 0.1289687156677246, "global_step": 46026, "epoch": 434, "lr": 4.210519130717722e-06} +{"train_loss": 0.015345275402069092, "grad_norm": 0.053557150065898895, "global_step": 46027, "epoch": 434, "lr": 4.2093287897335574e-06} +{"train_loss": 0.01566215045750141, "grad_norm": 0.08618271350860596, "global_step": 46028, "epoch": 434, "lr": 4.2081386096379126e-06} +{"train_loss": 0.016613997519016266, "grad_norm": 0.08054827153682709, "global_step": 46029, "epoch": 434, "lr": 4.206948590434944e-06} +{"train_loss": 0.015073515474796295, "grad_norm": 0.07586225867271423, "global_step": 46030, "epoch": 434, "lr": 4.205758732128851e-06} +{"train_loss": 0.0173017680644989, "grad_norm": 0.08059850335121155, "global_step": 46031, "epoch": 434, "lr": 4.2045690347237935e-06} +{"train_loss": 0.015207656659185886, "grad_norm": 0.10774834454059601, "global_step": 46032, "epoch": 434, "lr": 4.20337949822398e-06} +{"train_loss": 0.015353307127952576, "grad_norm": 0.07825501263141632, "global_step": 46033, "epoch": 434, "lr": 4.202190122633565e-06} +{"train_loss": 0.016714947298169136, "grad_norm": 0.07373277097940445, "global_step": 46034, "epoch": 434, "lr": 4.2010009079567314e-06} +{"train_loss": 0.013781119138002396, "grad_norm": 0.0627298504114151, "global_step": 46035, "epoch": 434, "lr": 4.199811854197677e-06} +{"train_loss": 0.014182781800627708, "grad_norm": 0.0790199488401413, "global_step": 46036, "epoch": 434, "lr": 4.1986229613605655e-06} +{"train_loss": 0.015872878953814507, "grad_norm": 0.09442434459924698, "global_step": 46037, "epoch": 434, "lr": 4.1974342294495596e-06} +{"train_loss": 0.014978338964283466, "grad_norm": 0.09060727059841156, "global_step": 46038, "epoch": 434, "lr": 4.196245658468856e-06} +{"train_loss": 0.013405769132077694, "grad_norm": 0.14508482813835144, "global_step": 46039, "epoch": 434, "lr": 4.1950572484226345e-06} +{"train_loss": 0.01279645785689354, "grad_norm": 0.11081143468618393, "global_step": 46040, "epoch": 434, "lr": 4.1938689993150536e-06} +{"train_loss": 0.014562489464879036, "grad_norm": 0.08573409169912338, "global_step": 46041, "epoch": 434, "lr": 4.192680911150293e-06} +{"train_loss": 0.012939144857227802, "grad_norm": 0.11642225831747055, "global_step": 46042, "epoch": 434, "lr": 4.1914929839325435e-06} +{"train_loss": 0.014056031592190266, "grad_norm": 0.07550667971372604, "global_step": 46043, "epoch": 434, "lr": 4.190305217665963e-06} +{"train_loss": 0.015827320516109467, "grad_norm": 0.057222895324230194, "global_step": 46044, "epoch": 434, "lr": 4.189117612354721e-06} +{"train_loss": 0.014007993042469025, "grad_norm": 0.08320432156324387, "global_step": 46045, "epoch": 434, "lr": 4.1879301680029954e-06} +{"train_loss": 0.013698122464120388, "grad_norm": 0.06353066861629486, "global_step": 46046, "epoch": 434, "lr": 4.186742884614969e-06} +{"train_loss": 0.014074856415390968, "grad_norm": 0.052871450781822205, "global_step": 46047, "epoch": 434, "lr": 4.1855557621947986e-06} +{"train_loss": 0.015669040381908417, "grad_norm": 0.10897161811590195, "global_step": 46048, "epoch": 434, "lr": 4.1843688007466685e-06} +{"train_loss": 0.014787057414650917, "grad_norm": 0.07494539767503738, "global_step": 46049, "epoch": 434, "lr": 4.1831820002747436e-06} +{"train_loss": 0.01662348583340645, "grad_norm": 0.09232845902442932, "global_step": 46050, "epoch": 434, "lr": 4.181995360783175e-06} +{"train_loss": 0.013435600325465202, "grad_norm": 0.08036915957927704, "global_step": 46051, "epoch": 434, "lr": 4.180808882276155e-06} +{"train_loss": 0.01263735443353653, "grad_norm": 0.08769097924232483, "global_step": 46052, "epoch": 434, "lr": 4.179622564757857e-06} +{"train_loss": 0.01460556872189045, "grad_norm": 0.08765441179275513, "global_step": 46053, "epoch": 434, "lr": 4.178436408232428e-06} +{"train_loss": 0.014254949986934662, "grad_norm": 0.0646924376487732, "global_step": 46054, "epoch": 434, "lr": 4.177250412704054e-06} +{"train_loss": 0.015398533083498478, "grad_norm": 0.09736485779285431, "global_step": 46055, "epoch": 434, "lr": 4.176064578176886e-06} +{"train_loss": 0.015169112011790276, "grad_norm": 0.05694675073027611, "global_step": 46056, "epoch": 434, "lr": 4.174878904655105e-06} +{"train_loss": 0.011853519827127457, "grad_norm": 0.06963200867176056, "global_step": 46057, "epoch": 434, "lr": 4.173693392142869e-06} +{"train_loss": 0.012705493718385696, "grad_norm": 0.07729142159223557, "global_step": 46058, "epoch": 434, "lr": 4.17250804064434e-06} +{"train_loss": 0.014894017949700356, "grad_norm": 0.09659378230571747, "global_step": 46059, "epoch": 434, "lr": 4.171322850163695e-06} +{"train_loss": 0.015348165296018124, "grad_norm": 0.06891006976366043, "global_step": 46060, "epoch": 434, "lr": 4.1701378207050944e-06} +{"train_loss": 0.013475347310304642, "grad_norm": 0.06815048307180405, "global_step": 46061, "epoch": 434, "lr": 4.168952952272687e-06} +{"train_loss": 0.01622941344976425, "grad_norm": 0.07365649193525314, "global_step": 46062, "epoch": 434, "lr": 4.167768244870651e-06} +{"train_loss": 0.016446802765130997, "grad_norm": 0.120406873524189, "global_step": 46063, "epoch": 434, "lr": 4.166583698503151e-06} +{"train_loss": 0.014738347381353378, "grad_norm": 0.1247839629650116, "global_step": 46064, "epoch": 434, "lr": 4.165399313174334e-06} +{"train_loss": 0.016373496502637863, "grad_norm": 0.06308858841657639, "global_step": 46065, "epoch": 434, "lr": 4.164215088888374e-06} +{"train_loss": 0.015670519322156906, "grad_norm": 0.07856647670269012, "global_step": 46066, "epoch": 434, "lr": 4.16303102564945e-06} +{"train_loss": 0.01542850024998188, "grad_norm": 0.11463640630245209, "global_step": 46067, "epoch": 434, "lr": 4.161847123461677e-06} +{"train_loss": 0.01566227152943611, "grad_norm": 0.050943341106176376, "global_step": 46068, "epoch": 434, "lr": 4.1606633823292385e-06} +{"train_loss": 0.01625668816268444, "grad_norm": 0.09192942827939987, "global_step": 46069, "epoch": 434, "lr": 4.159479802256305e-06} +{"train_loss": 0.01574789360165596, "grad_norm": 0.12340220808982849, "global_step": 46070, "epoch": 434, "lr": 4.158296383247018e-06} +{"train_loss": 0.015088826417922974, "grad_norm": 0.07024522125720978, "global_step": 46071, "epoch": 434, "lr": 4.157113125305545e-06} +{"train_loss": 0.016287675127387047, "grad_norm": 0.062362976372241974, "global_step": 46072, "epoch": 434, "lr": 4.1559300284360335e-06} +{"train_loss": 0.01454309280961752, "grad_norm": 0.0730435699224472, "global_step": 46073, "epoch": 434, "lr": 4.154747092642653e-06} +{"train_loss": 0.017067335546016693, "grad_norm": 0.12689366936683655, "global_step": 46074, "epoch": 434, "lr": 4.153564317929548e-06} +{"train_loss": 0.01324186660349369, "grad_norm": 0.06047280877828598, "global_step": 46075, "epoch": 434, "lr": 4.152381704300879e-06} +{"train_loss": 0.016458775848150253, "grad_norm": 0.07254840433597565, "global_step": 46076, "epoch": 434, "lr": 4.151199251760812e-06} +{"train_loss": 0.01575125753879547, "grad_norm": 0.09074012190103531, "global_step": 46077, "epoch": 434, "lr": 4.150016960313491e-06} +{"train_loss": 0.01394940447062254, "grad_norm": 0.06275495886802673, "global_step": 46078, "epoch": 434, "lr": 4.148834829963061e-06} +{"train_loss": 0.014413172379136086, "grad_norm": 0.07695022225379944, "global_step": 46079, "epoch": 434, "lr": 4.147652860713685e-06} +{"train_loss": 0.014834991656243801, "grad_norm": 0.1226402297616005, "global_step": 46080, "epoch": 434, "lr": 4.1464710525695285e-06} +{"train_loss": 0.01564847305417061, "grad_norm": 0.06353069841861725, "global_step": 46081, "epoch": 434, "lr": 4.145289405534719e-06} +{"train_loss": 0.01613088697195053, "grad_norm": 0.06621035933494568, "global_step": 46082, "epoch": 434, "lr": 4.1441079196134216e-06} +{"train_loss": 0.01759917661547661, "grad_norm": 0.07645228505134583, "global_step": 46083, "epoch": 434, "lr": 4.142926594809804e-06} +{"train_loss": 0.013982160948216915, "grad_norm": 0.10021145641803741, "global_step": 46084, "epoch": 434, "lr": 4.141745431127986e-06} +{"train_loss": 0.017243027687072754, "grad_norm": 0.07536633312702179, "global_step": 46085, "epoch": 434, "lr": 4.140564428572125e-06} +{"train_loss": 0.015156060457229614, "grad_norm": 0.07203949242830276, "global_step": 46086, "epoch": 434, "lr": 4.139383587146389e-06} +{"train_loss": 0.018444126471877098, "grad_norm": 0.09510757029056549, "global_step": 46087, "epoch": 434, "lr": 4.138202906854904e-06} +{"train_loss": 0.012786687351763248, "grad_norm": 0.06707905232906342, "global_step": 46088, "epoch": 434, "lr": 4.137022387701833e-06} +{"train_loss": 0.013496460393071175, "grad_norm": 0.08144408464431763, "global_step": 46089, "epoch": 434, "lr": 4.135842029691328e-06} +{"train_loss": 0.014502455480396748, "grad_norm": 0.09724149107933044, "global_step": 46090, "epoch": 434, "lr": 4.134661832827524e-06} +{"train_loss": 0.01331289578229189, "grad_norm": 0.07846415042877197, "global_step": 46091, "epoch": 434, "lr": 4.133481797114569e-06} +{"train_loss": 0.017750531435012817, "grad_norm": 0.07752491533756256, "global_step": 46092, "epoch": 434, "lr": 4.1323019225566084e-06} +{"train_loss": 0.0127413310110569, "grad_norm": 0.06881245970726013, "global_step": 46093, "epoch": 434, "lr": 4.1311222091578e-06} +{"train_loss": 0.01667175628244877, "grad_norm": 0.07879804819822311, "global_step": 46094, "epoch": 434, "lr": 4.1299426569222755e-06} +{"train_loss": 0.01597626507282257, "grad_norm": 0.07393692433834076, "global_step": 46095, "epoch": 434, "lr": 4.128763265854191e-06} +{"train_loss": 0.01486852765083313, "grad_norm": 0.05536311864852905, "global_step": 46096, "epoch": 434, "lr": 4.127584035957671e-06} +{"train_loss": 0.015508956275880337, "grad_norm": 0.10596180707216263, "global_step": 46097, "epoch": 434, "lr": 4.126404967236885e-06} +{"train_loss": 0.018246421590447426, "grad_norm": 0.06162892282009125, "global_step": 46098, "epoch": 434, "lr": 4.1252260596959524e-06} +{"train_loss": 0.013041649013757706, "grad_norm": 0.07242091000080109, "global_step": 46099, "epoch": 434, "lr": 4.124047313339025e-06} +{"train_loss": 0.016383377835154533, "grad_norm": 0.12053622305393219, "global_step": 46100, "epoch": 434, "lr": 4.1228687281702486e-06} +{"train_loss": 0.013318995013833046, "grad_norm": 0.10329943150281906, "global_step": 46101, "epoch": 434, "lr": 4.121690304193765e-06} +{"train_loss": 0.012912165373563766, "grad_norm": 0.06741277873516083, "global_step": 46102, "epoch": 434, "lr": 4.120512041413699e-06} +{"train_loss": 0.015064283274114132, "grad_norm": 0.09397508203983307, "global_step": 46103, "epoch": 434, "lr": 4.119333939834202e-06} +{"train_loss": 0.015583550557494164, "grad_norm": 0.05752263590693474, "global_step": 46104, "epoch": 434, "lr": 4.118155999459422e-06} +{"train_loss": 0.013349346816539764, "grad_norm": 0.13742506504058838, "global_step": 46105, "epoch": 434, "lr": 4.116978220293482e-06} +{"train_loss": 0.015797289088368416, "grad_norm": 0.08614388108253479, "global_step": 46106, "epoch": 434, "lr": 4.115800602340519e-06} +{"train_loss": 0.015297958627343178, "grad_norm": 0.08520429581403732, "global_step": 46107, "epoch": 434, "lr": 4.114623145604701e-06} +{"train_loss": 0.01618826575577259, "grad_norm": 0.1067444384098053, "global_step": 46108, "epoch": 434, "lr": 4.1134458500901255e-06} +{"train_loss": 0.01496249414488392, "grad_norm": 0.18619218468666077, "global_step": 46109, "epoch": 434, "lr": 4.112268715800943e-06, "val_loss": 0.019694644957780838} +{"train_loss": 0.017841089516878128, "grad_norm": 0.08330661058425903, "global_step": 46110, "epoch": 435, "lr": 4.111091742741302e-06} +{"train_loss": 0.017229484394192696, "grad_norm": 0.06852702051401138, "global_step": 46111, "epoch": 435, "lr": 4.109914930915321e-06} +{"train_loss": 0.016469944268465042, "grad_norm": 0.1229868084192276, "global_step": 46112, "epoch": 435, "lr": 4.10873828032714e-06} +{"train_loss": 0.013743984512984753, "grad_norm": 0.07068467140197754, "global_step": 46113, "epoch": 435, "lr": 4.107561790980901e-06} +{"train_loss": 0.014820506796240807, "grad_norm": 0.1706686019897461, "global_step": 46114, "epoch": 435, "lr": 4.106385462880735e-06} +{"train_loss": 0.01446590106934309, "grad_norm": 0.07856196910142899, "global_step": 46115, "epoch": 435, "lr": 4.105209296030765e-06} +{"train_loss": 0.0140081737190485, "grad_norm": 0.1067287027835846, "global_step": 46116, "epoch": 435, "lr": 4.104033290435128e-06} +{"train_loss": 0.01406528614461422, "grad_norm": 0.09449511766433716, "global_step": 46117, "epoch": 435, "lr": 4.102857446097969e-06} +{"train_loss": 0.014036156237125397, "grad_norm": 0.11516904830932617, "global_step": 46118, "epoch": 435, "lr": 4.101681763023396e-06} +{"train_loss": 0.014198046177625656, "grad_norm": 0.11234497278928757, "global_step": 46119, "epoch": 435, "lr": 4.100506241215562e-06} +{"train_loss": 0.012823140248656273, "grad_norm": 0.0910264328122139, "global_step": 46120, "epoch": 435, "lr": 4.099330880678581e-06} +{"train_loss": 0.012588588520884514, "grad_norm": 0.08807570487260818, "global_step": 46121, "epoch": 435, "lr": 4.098155681416599e-06} +{"train_loss": 0.01533106341958046, "grad_norm": 0.07564222067594528, "global_step": 46122, "epoch": 435, "lr": 4.096980643433728e-06} +{"train_loss": 0.01663927175104618, "grad_norm": 0.07701314240694046, "global_step": 46123, "epoch": 435, "lr": 4.095805766734101e-06} +{"train_loss": 0.014064942486584187, "grad_norm": 0.06896639615297318, "global_step": 46124, "epoch": 435, "lr": 4.094631051321873e-06} +{"train_loss": 0.015397585928440094, "grad_norm": 0.06887819617986679, "global_step": 46125, "epoch": 435, "lr": 4.093456497201126e-06} +{"train_loss": 0.01325155422091484, "grad_norm": 0.06247911602258682, "global_step": 46126, "epoch": 435, "lr": 4.092282104376005e-06} +{"train_loss": 0.013818798586726189, "grad_norm": 0.0986269861459732, "global_step": 46127, "epoch": 435, "lr": 4.091107872850653e-06} +{"train_loss": 0.014455794356763363, "grad_norm": 0.06354349106550217, "global_step": 46128, "epoch": 435, "lr": 4.089933802629175e-06} +{"train_loss": 0.013557789847254753, "grad_norm": 0.20143404603004456, "global_step": 46129, "epoch": 435, "lr": 4.088759893715699e-06} +{"train_loss": 0.01606728509068489, "grad_norm": 0.07273399084806442, "global_step": 46130, "epoch": 435, "lr": 4.08758614611437e-06} +{"train_loss": 0.015712711960077286, "grad_norm": 0.11505565047264099, "global_step": 46131, "epoch": 435, "lr": 4.086412559829289e-06} +{"train_loss": 0.01695874333381653, "grad_norm": 0.10693275928497314, "global_step": 46132, "epoch": 435, "lr": 4.085239134864583e-06} +{"train_loss": 0.015118596144020557, "grad_norm": 0.09643945097923279, "global_step": 46133, "epoch": 435, "lr": 4.08406587122438e-06} +{"train_loss": 0.015384873375296593, "grad_norm": 0.06660987436771393, "global_step": 46134, "epoch": 435, "lr": 4.0828927689128046e-06} +{"train_loss": 0.015774056315422058, "grad_norm": 0.10131997615098953, "global_step": 46135, "epoch": 435, "lr": 4.081719827933972e-06} +{"train_loss": 0.016594644635915756, "grad_norm": 0.07263675332069397, "global_step": 46136, "epoch": 435, "lr": 4.08054704829201e-06} +{"train_loss": 0.015510202385485172, "grad_norm": 0.06519465893507004, "global_step": 46137, "epoch": 435, "lr": 4.079374429991028e-06} +{"train_loss": 0.014627563767135143, "grad_norm": 0.056563254445791245, "global_step": 46138, "epoch": 435, "lr": 4.078201973035168e-06} +{"train_loss": 0.01285950094461441, "grad_norm": 0.08728314936161041, "global_step": 46139, "epoch": 435, "lr": 4.077029677428523e-06} +{"train_loss": 0.015239759348332882, "grad_norm": 0.10703147947788239, "global_step": 46140, "epoch": 435, "lr": 4.075857543175227e-06} +{"train_loss": 0.01461000181734562, "grad_norm": 0.06745092570781708, "global_step": 46141, "epoch": 435, "lr": 4.0746855702794015e-06} +{"train_loss": 0.01505151204764843, "grad_norm": 0.1515330672264099, "global_step": 46142, "epoch": 435, "lr": 4.073513758745156e-06} +{"train_loss": 0.01694282703101635, "grad_norm": 0.08876963704824448, "global_step": 46143, "epoch": 435, "lr": 4.072342108576605e-06} +{"train_loss": 0.01583806984126568, "grad_norm": 0.07414482533931732, "global_step": 46144, "epoch": 435, "lr": 4.07117061977787e-06} +{"train_loss": 0.016472961753606796, "grad_norm": 0.09387357532978058, "global_step": 46145, "epoch": 435, "lr": 4.069999292353078e-06} +{"train_loss": 0.0132681829854846, "grad_norm": 0.08202074468135834, "global_step": 46146, "epoch": 435, "lr": 4.068828126306323e-06} +{"train_loss": 0.014482471160590649, "grad_norm": 0.07176219671964645, "global_step": 46147, "epoch": 435, "lr": 4.067657121641732e-06} +{"train_loss": 0.014833010733127594, "grad_norm": 0.05405508354306221, "global_step": 46148, "epoch": 435, "lr": 4.066486278363435e-06} +{"train_loss": 0.014478905126452446, "grad_norm": 0.07247213274240494, "global_step": 46149, "epoch": 435, "lr": 4.065315596475511e-06} +{"train_loss": 0.01666468381881714, "grad_norm": 0.07786928862333298, "global_step": 46150, "epoch": 435, "lr": 4.064145075982096e-06} +{"train_loss": 0.012216377072036266, "grad_norm": 0.060980424284935, "global_step": 46151, "epoch": 435, "lr": 4.062974716887303e-06} +{"train_loss": 0.01576797291636467, "grad_norm": 0.096941739320755, "global_step": 46152, "epoch": 435, "lr": 4.061804519195228e-06} +{"train_loss": 0.014877520501613617, "grad_norm": 0.12938934564590454, "global_step": 46153, "epoch": 435, "lr": 4.060634482909997e-06} +{"train_loss": 0.014650722034275532, "grad_norm": 0.05438093841075897, "global_step": 46154, "epoch": 435, "lr": 4.059464608035729e-06} +{"train_loss": 0.014764836058020592, "grad_norm": 0.07783738523721695, "global_step": 46155, "epoch": 435, "lr": 4.0582948945765195e-06} +{"train_loss": 0.016721203923225403, "grad_norm": 0.0717020183801651, "global_step": 46156, "epoch": 435, "lr": 4.057125342536477e-06} +{"train_loss": 0.01666468009352684, "grad_norm": 0.10866233706474304, "global_step": 46157, "epoch": 435, "lr": 4.055955951919716e-06} +{"train_loss": 0.01408209279179573, "grad_norm": 0.0880347341299057, "global_step": 46158, "epoch": 435, "lr": 4.054786722730353e-06} +{"train_loss": 0.01356991846114397, "grad_norm": 0.0972571149468422, "global_step": 46159, "epoch": 435, "lr": 4.0536176549724806e-06} +{"train_loss": 0.012571676634252071, "grad_norm": 0.06861220300197601, "global_step": 46160, "epoch": 435, "lr": 4.052448748650217e-06} +{"train_loss": 0.01876196265220642, "grad_norm": 0.10198232531547546, "global_step": 46161, "epoch": 435, "lr": 4.051280003767666e-06} +{"train_loss": 0.01595318876206875, "grad_norm": 0.14278601109981537, "global_step": 46162, "epoch": 435, "lr": 4.050111420328939e-06} +{"train_loss": 0.014347472228109837, "grad_norm": 0.07200971990823746, "global_step": 46163, "epoch": 435, "lr": 4.048942998338129e-06} +{"train_loss": 0.014356876723468304, "grad_norm": 0.08827134221792221, "global_step": 46164, "epoch": 435, "lr": 4.047774737799348e-06} +{"train_loss": 0.014088008552789688, "grad_norm": 0.05965128168463707, "global_step": 46165, "epoch": 435, "lr": 4.04660663871671e-06} +{"train_loss": 0.011931091547012329, "grad_norm": 0.0730142742395401, "global_step": 46166, "epoch": 435, "lr": 4.045438701094317e-06} +{"train_loss": 0.01615549810230732, "grad_norm": 0.07115579396486282, "global_step": 46167, "epoch": 435, "lr": 4.0442709249362555e-06} +{"train_loss": 0.015013364143669605, "grad_norm": 0.0686577782034874, "global_step": 46168, "epoch": 435, "lr": 4.043103310246649e-06} +{"train_loss": 0.016236966475844383, "grad_norm": 0.11311527341604233, "global_step": 46169, "epoch": 435, "lr": 4.041935857029583e-06} +{"train_loss": 0.016969693824648857, "grad_norm": 0.0705176442861557, "global_step": 46170, "epoch": 435, "lr": 4.040768565289172e-06} +{"train_loss": 0.013738940469920635, "grad_norm": 0.06826060265302658, "global_step": 46171, "epoch": 435, "lr": 4.039601435029516e-06} +{"train_loss": 0.01639203354716301, "grad_norm": 0.06477849185466766, "global_step": 46172, "epoch": 435, "lr": 4.038434466254714e-06} +{"train_loss": 0.014913009479641914, "grad_norm": 0.10000308603048325, "global_step": 46173, "epoch": 435, "lr": 4.037267658968857e-06} +{"train_loss": 0.01667908765375614, "grad_norm": 0.09096334874629974, "global_step": 46174, "epoch": 435, "lr": 4.036101013176053e-06} +{"train_loss": 0.01430799812078476, "grad_norm": 0.15170307457447052, "global_step": 46175, "epoch": 435, "lr": 4.034934528880413e-06} +{"train_loss": 0.014012552797794342, "grad_norm": 0.06857843697071075, "global_step": 46176, "epoch": 435, "lr": 4.033768206086008e-06} +{"train_loss": 0.016225557774305344, "grad_norm": 0.07326576858758926, "global_step": 46177, "epoch": 435, "lr": 4.032602044796963e-06} +{"train_loss": 0.017019230872392654, "grad_norm": 0.04509210214018822, "global_step": 46178, "epoch": 435, "lr": 4.031436045017356e-06} +{"train_loss": 0.013390660285949707, "grad_norm": 0.11114499717950821, "global_step": 46179, "epoch": 435, "lr": 4.030270206751302e-06} +{"train_loss": 0.017047423869371414, "grad_norm": 0.06426159292459488, "global_step": 46180, "epoch": 435, "lr": 4.029104530002875e-06} +{"train_loss": 0.013061607256531715, "grad_norm": 0.06992075592279434, "global_step": 46181, "epoch": 435, "lr": 4.027939014776183e-06} +{"train_loss": 0.013515781611204147, "grad_norm": 0.0805404782295227, "global_step": 46182, "epoch": 435, "lr": 4.0267736610753285e-06} +{"train_loss": 0.013815399259328842, "grad_norm": 0.09181693941354752, "global_step": 46183, "epoch": 435, "lr": 4.0256084689044015e-06} +{"train_loss": 0.012932094745337963, "grad_norm": 0.08036034554243088, "global_step": 46184, "epoch": 435, "lr": 4.024443438267478e-06} +{"train_loss": 0.013478988781571388, "grad_norm": 0.09702517092227936, "global_step": 46185, "epoch": 435, "lr": 4.023278569168681e-06} +{"train_loss": 0.014612063765525818, "grad_norm": 0.11563084274530411, "global_step": 46186, "epoch": 435, "lr": 4.0221138616120755e-06} +{"train_loss": 0.014352969825267792, "grad_norm": 0.07275792211294174, "global_step": 46187, "epoch": 435, "lr": 4.020949315601769e-06} +{"train_loss": 0.014601918868720531, "grad_norm": 0.06652118265628815, "global_step": 46188, "epoch": 435, "lr": 4.0197849311418565e-06} +{"train_loss": 0.017119958996772766, "grad_norm": 0.07521649450063705, "global_step": 46189, "epoch": 435, "lr": 4.018620708236426e-06} +{"train_loss": 0.01675068587064743, "grad_norm": 0.14826558530330658, "global_step": 46190, "epoch": 435, "lr": 4.017456646889555e-06} +{"train_loss": 0.016475427895784378, "grad_norm": 0.0995444506406784, "global_step": 46191, "epoch": 435, "lr": 4.016292747105349e-06} +{"train_loss": 0.013281217776238918, "grad_norm": 0.0899047777056694, "global_step": 46192, "epoch": 435, "lr": 4.0151290088878964e-06} +{"train_loss": 0.016406487673521042, "grad_norm": 0.08538846671581268, "global_step": 46193, "epoch": 435, "lr": 4.013965432241273e-06} +{"train_loss": 0.015169261023402214, "grad_norm": 0.06041081249713898, "global_step": 46194, "epoch": 435, "lr": 4.012802017169581e-06} +{"train_loss": 0.013791065663099289, "grad_norm": 0.07285510748624802, "global_step": 46195, "epoch": 435, "lr": 4.011638763676906e-06} +{"train_loss": 0.01528282929211855, "grad_norm": 0.08842748403549194, "global_step": 46196, "epoch": 435, "lr": 4.0104756717673385e-06} +{"train_loss": 0.014110518619418144, "grad_norm": 0.07275392860174179, "global_step": 46197, "epoch": 435, "lr": 4.009312741444948e-06} +{"train_loss": 0.014557795599102974, "grad_norm": 0.07564239203929901, "global_step": 46198, "epoch": 435, "lr": 4.008149972713831e-06} +{"train_loss": 0.015239492990076542, "grad_norm": 0.14547914266586304, "global_step": 46199, "epoch": 435, "lr": 4.00698736557808e-06} +{"train_loss": 0.014966323971748352, "grad_norm": 0.06320947408676147, "global_step": 46200, "epoch": 435, "lr": 4.005824920041768e-06} +{"train_loss": 0.015269861556589603, "grad_norm": 0.09738809615373611, "global_step": 46201, "epoch": 435, "lr": 4.0046626361089915e-06} +{"train_loss": 0.015117846429347992, "grad_norm": 0.09572666138410568, "global_step": 46202, "epoch": 435, "lr": 4.003500513783825e-06} +{"train_loss": 0.013675743713974953, "grad_norm": 0.08051446825265884, "global_step": 46203, "epoch": 435, "lr": 4.0023385530703564e-06} +{"train_loss": 0.014641737565398216, "grad_norm": 0.0849330797791481, "global_step": 46204, "epoch": 435, "lr": 4.001176753972658e-06} +{"train_loss": 0.016948869451880455, "grad_norm": 0.11239486932754517, "global_step": 46205, "epoch": 435, "lr": 4.0000151164948265e-06} +{"train_loss": 0.015431014820933342, "grad_norm": 0.09219792485237122, "global_step": 46206, "epoch": 435, "lr": 3.9988536406409375e-06} +{"train_loss": 0.015011644922196865, "grad_norm": 0.0577705092728138, "global_step": 46207, "epoch": 435, "lr": 3.997692326415076e-06} +{"train_loss": 0.015313205309212208, "grad_norm": 0.08554672449827194, "global_step": 46208, "epoch": 435, "lr": 3.99653117382131e-06} +{"train_loss": 0.01682940684258938, "grad_norm": 0.059871941804885864, "global_step": 46209, "epoch": 435, "lr": 3.995370182863733e-06} +{"train_loss": 0.012720824219286442, "grad_norm": 0.12411151826381683, "global_step": 46210, "epoch": 435, "lr": 3.9942093535464076e-06} +{"train_loss": 0.01635625772178173, "grad_norm": 0.07398727536201477, "global_step": 46211, "epoch": 435, "lr": 3.99304868587343e-06} +{"train_loss": 0.013700896874070168, "grad_norm": 0.11209259927272797, "global_step": 46212, "epoch": 435, "lr": 3.991888179848874e-06} +{"train_loss": 0.015530124306678772, "grad_norm": 0.12783417105674744, "global_step": 46213, "epoch": 435, "lr": 3.990727835476815e-06} +{"train_loss": 0.012697670608758926, "grad_norm": 0.11587482690811157, "global_step": 46214, "epoch": 435, "lr": 3.989567652761322e-06} +{"train_loss": 0.014960014779682992, "grad_norm": 0.09840068966150284, "global_step": 46215, "epoch": 435, "lr": 3.9884076317064814e-06, "val_loss": 0.01978306658565998} +{"train_loss": 0.01587574928998947, "grad_norm": 0.09178970754146576, "global_step": 46216, "epoch": 436, "lr": 3.987247772316377e-06} +{"train_loss": 0.013339819386601448, "grad_norm": 0.0872916430234909, "global_step": 46217, "epoch": 436, "lr": 3.986088074595057e-06} +{"train_loss": 0.013360457494854927, "grad_norm": 0.05166996270418167, "global_step": 46218, "epoch": 436, "lr": 3.984928538546617e-06} +{"train_loss": 0.013653322122991085, "grad_norm": 0.0544835701584816, "global_step": 46219, "epoch": 436, "lr": 3.983769164175139e-06} +{"train_loss": 0.015614868141710758, "grad_norm": 0.07632306218147278, "global_step": 46220, "epoch": 436, "lr": 3.982609951484678e-06} +{"train_loss": 0.014703254215419292, "grad_norm": 0.1903296411037445, "global_step": 46221, "epoch": 436, "lr": 3.9814509004793056e-06} +{"train_loss": 0.014638016000390053, "grad_norm": 0.08239197731018066, "global_step": 46222, "epoch": 436, "lr": 3.980292011163106e-06} +{"train_loss": 0.014997539110481739, "grad_norm": 0.12227248400449753, "global_step": 46223, "epoch": 436, "lr": 3.979133283540148e-06} +{"train_loss": 0.014282293617725372, "grad_norm": 0.06171296164393425, "global_step": 46224, "epoch": 436, "lr": 3.9779747176145075e-06} +{"train_loss": 0.018171627074480057, "grad_norm": 0.12897755205631256, "global_step": 46225, "epoch": 436, "lr": 3.976816313390242e-06} +{"train_loss": 0.016304586082696915, "grad_norm": 0.07591060549020767, "global_step": 46226, "epoch": 436, "lr": 3.975658070871436e-06} +{"train_loss": 0.01591194048523903, "grad_norm": 0.12469364702701569, "global_step": 46227, "epoch": 436, "lr": 3.974499990062142e-06} +{"train_loss": 0.017812881618738174, "grad_norm": 0.08071795850992203, "global_step": 46228, "epoch": 436, "lr": 3.973342070966435e-06} +{"train_loss": 0.011195356026291847, "grad_norm": 0.0689312070608139, "global_step": 46229, "epoch": 436, "lr": 3.972184313588401e-06} +{"train_loss": 0.014661738649010658, "grad_norm": 0.1826602816581726, "global_step": 46230, "epoch": 436, "lr": 3.971026717932097e-06} +{"train_loss": 0.015571987256407738, "grad_norm": 0.07849975675344467, "global_step": 46231, "epoch": 436, "lr": 3.969869284001576e-06} +{"train_loss": 0.01574038341641426, "grad_norm": 0.07061850279569626, "global_step": 46232, "epoch": 436, "lr": 3.968712011800912e-06} +{"train_loss": 0.01574459671974182, "grad_norm": 0.09017237275838852, "global_step": 46233, "epoch": 436, "lr": 3.9675549013341915e-06} +{"train_loss": 0.015038413926959038, "grad_norm": 0.08890333771705627, "global_step": 46234, "epoch": 436, "lr": 3.96639795260545e-06} +{"train_loss": 0.01452646404504776, "grad_norm": 0.10980545729398727, "global_step": 46235, "epoch": 436, "lr": 3.965241165618772e-06} +{"train_loss": 0.013947520405054092, "grad_norm": 0.10238683223724365, "global_step": 46236, "epoch": 436, "lr": 3.9640845403782215e-06} +{"train_loss": 0.015616129152476788, "grad_norm": 0.05751638486981392, "global_step": 46237, "epoch": 436, "lr": 3.962928076887856e-06} +{"train_loss": 0.013385489583015442, "grad_norm": 0.08292444795370102, "global_step": 46238, "epoch": 436, "lr": 3.961771775151735e-06} +{"train_loss": 0.014360923320055008, "grad_norm": 0.07016493380069733, "global_step": 46239, "epoch": 436, "lr": 3.960615635173925e-06} +{"train_loss": 0.015765368938446045, "grad_norm": 0.08390697091817856, "global_step": 46240, "epoch": 436, "lr": 3.959459656958503e-06} +{"train_loss": 0.015689168125391006, "grad_norm": 0.07021155208349228, "global_step": 46241, "epoch": 436, "lr": 3.958303840509503e-06} +{"train_loss": 0.01482413150370121, "grad_norm": 0.0848507434129715, "global_step": 46242, "epoch": 436, "lr": 3.957148185831011e-06} +{"train_loss": 0.015749726444482803, "grad_norm": 0.09982606023550034, "global_step": 46243, "epoch": 436, "lr": 3.955992692927074e-06} +{"train_loss": 0.015956513583660126, "grad_norm": 0.22680091857910156, "global_step": 46244, "epoch": 436, "lr": 3.95483736180175e-06} +{"train_loss": 0.015801750123500824, "grad_norm": 0.08451712876558304, "global_step": 46245, "epoch": 436, "lr": 3.953682192459096e-06} +{"train_loss": 0.014437112025916576, "grad_norm": 0.09579706192016602, "global_step": 46246, "epoch": 436, "lr": 3.952527184903187e-06} +{"train_loss": 0.01593729853630066, "grad_norm": 0.13442020118236542, "global_step": 46247, "epoch": 436, "lr": 3.951372339138076e-06} +{"train_loss": 0.016205742955207825, "grad_norm": 0.1825421303510666, "global_step": 46248, "epoch": 436, "lr": 3.950217655167815e-06} +{"train_loss": 0.012865619733929634, "grad_norm": 0.07165943086147308, "global_step": 46249, "epoch": 436, "lr": 3.949063132996455e-06} +{"train_loss": 0.014916088432073593, "grad_norm": 0.05360129103064537, "global_step": 46250, "epoch": 436, "lr": 3.947908772628072e-06} +{"train_loss": 0.013689893297851086, "grad_norm": 0.09924820810556412, "global_step": 46251, "epoch": 436, "lr": 3.946754574066697e-06} +{"train_loss": 0.01565537415444851, "grad_norm": 0.0747745931148529, "global_step": 46252, "epoch": 436, "lr": 3.945600537316402e-06} +{"train_loss": 0.013818962499499321, "grad_norm": 0.06408350169658661, "global_step": 46253, "epoch": 436, "lr": 3.944446662381246e-06} +{"train_loss": 0.015189582481980324, "grad_norm": 0.1464117020368576, "global_step": 46254, "epoch": 436, "lr": 3.943292949265276e-06} +{"train_loss": 0.014858461916446686, "grad_norm": 0.08647967875003815, "global_step": 46255, "epoch": 436, "lr": 3.942139397972533e-06} +{"train_loss": 0.011022582650184631, "grad_norm": 0.08912874013185501, "global_step": 46256, "epoch": 436, "lr": 3.940986008507086e-06} +{"train_loss": 0.014377261511981487, "grad_norm": 0.06225184351205826, "global_step": 46257, "epoch": 436, "lr": 3.939832780872988e-06} +{"train_loss": 0.014767627231776714, "grad_norm": 0.1014501303434372, "global_step": 46258, "epoch": 436, "lr": 3.938679715074283e-06} +{"train_loss": 0.014268781989812851, "grad_norm": 0.06800007075071335, "global_step": 46259, "epoch": 436, "lr": 3.937526811115022e-06} +{"train_loss": 0.0172438807785511, "grad_norm": 0.11266808956861496, "global_step": 46260, "epoch": 436, "lr": 3.9363740689992715e-06} +{"train_loss": 0.014526141807436943, "grad_norm": 0.07057720422744751, "global_step": 46261, "epoch": 436, "lr": 3.9352214887310666e-06} +{"train_loss": 0.016670336946845055, "grad_norm": 0.18446335196495056, "global_step": 46262, "epoch": 436, "lr": 3.934069070314456e-06} +{"train_loss": 0.014303368516266346, "grad_norm": 0.06483855098485947, "global_step": 46263, "epoch": 436, "lr": 3.93291681375349e-06} +{"train_loss": 0.01414178404957056, "grad_norm": 0.10142723470926285, "global_step": 46264, "epoch": 436, "lr": 3.931764719052228e-06} +{"train_loss": 0.01532298605889082, "grad_norm": 0.07958132028579712, "global_step": 46265, "epoch": 436, "lr": 3.930612786214699e-06} +{"train_loss": 0.015196442604064941, "grad_norm": 0.0899694636464119, "global_step": 46266, "epoch": 436, "lr": 3.929461015244979e-06} +{"train_loss": 0.013979123905301094, "grad_norm": 0.07792917639017105, "global_step": 46267, "epoch": 436, "lr": 3.92830940614709e-06} +{"train_loss": 0.012640472501516342, "grad_norm": 0.1311555951833725, "global_step": 46268, "epoch": 436, "lr": 3.9271579589250816e-06} +{"train_loss": 0.015348430722951889, "grad_norm": 0.101340152323246, "global_step": 46269, "epoch": 436, "lr": 3.926006673582999e-06} +{"train_loss": 0.015388202853500843, "grad_norm": 0.0857449322938919, "global_step": 46270, "epoch": 436, "lr": 3.924855550124906e-06} +{"train_loss": 0.018631234765052795, "grad_norm": 0.10523203760385513, "global_step": 46271, "epoch": 436, "lr": 3.923704588554822e-06} +{"train_loss": 0.01810508593916893, "grad_norm": 0.12882113456726074, "global_step": 46272, "epoch": 436, "lr": 3.9225537888768096e-06} +{"train_loss": 0.014562014490365982, "grad_norm": 0.08966493606567383, "global_step": 46273, "epoch": 436, "lr": 3.921403151094899e-06} +{"train_loss": 0.015833880752325058, "grad_norm": 0.055817101150751114, "global_step": 46274, "epoch": 436, "lr": 3.920252675213143e-06} +{"train_loss": 0.015011210925877094, "grad_norm": 0.06426158547401428, "global_step": 46275, "epoch": 436, "lr": 3.919102361235572e-06} +{"train_loss": 0.016343051567673683, "grad_norm": 0.08545458316802979, "global_step": 46276, "epoch": 436, "lr": 3.917952209166237e-06} +{"train_loss": 0.012289539910852909, "grad_norm": 0.062449242919683456, "global_step": 46277, "epoch": 436, "lr": 3.916802219009186e-06} +{"train_loss": 0.01499877031892538, "grad_norm": 0.058868926018476486, "global_step": 46278, "epoch": 436, "lr": 3.9156523907684485e-06} +{"train_loss": 0.01581360213458538, "grad_norm": 0.07721506059169769, "global_step": 46279, "epoch": 436, "lr": 3.914502724448061e-06} +{"train_loss": 0.01523961964994669, "grad_norm": 0.05711328983306885, "global_step": 46280, "epoch": 436, "lr": 3.913353220052069e-06} +{"train_loss": 0.018266629427671432, "grad_norm": 0.07396505028009415, "global_step": 46281, "epoch": 436, "lr": 3.91220387758452e-06} +{"train_loss": 0.015638666227459908, "grad_norm": 0.10437945276498795, "global_step": 46282, "epoch": 436, "lr": 3.911054697049432e-06} +{"train_loss": 0.014180291444063187, "grad_norm": 0.06382927298545837, "global_step": 46283, "epoch": 436, "lr": 3.909905678450865e-06} +{"train_loss": 0.015298337675631046, "grad_norm": 0.10256835073232651, "global_step": 46284, "epoch": 436, "lr": 3.908756821792842e-06} +{"train_loss": 0.013277260586619377, "grad_norm": 0.06633062660694122, "global_step": 46285, "epoch": 436, "lr": 3.907608127079398e-06} +{"train_loss": 0.017112694680690765, "grad_norm": 0.08475185930728912, "global_step": 46286, "epoch": 436, "lr": 3.9064595943145766e-06} +{"train_loss": 0.012875603511929512, "grad_norm": 0.1034284234046936, "global_step": 46287, "epoch": 436, "lr": 3.905311223502411e-06} +{"train_loss": 0.015662159770727158, "grad_norm": 0.10300570726394653, "global_step": 46288, "epoch": 436, "lr": 3.904163014646933e-06} +{"train_loss": 0.01691664569079876, "grad_norm": 0.09688661992549896, "global_step": 46289, "epoch": 436, "lr": 3.9030149677521885e-06} +{"train_loss": 0.015154039487242699, "grad_norm": 0.06464093178510666, "global_step": 46290, "epoch": 436, "lr": 3.901867082822192e-06} +{"train_loss": 0.014887696132063866, "grad_norm": 0.09958899766206741, "global_step": 46291, "epoch": 436, "lr": 3.900719359860999e-06} +{"train_loss": 0.015876000747084618, "grad_norm": 0.07539774477481842, "global_step": 46292, "epoch": 436, "lr": 3.8995717988726145e-06} +{"train_loss": 0.015867765992879868, "grad_norm": 0.06956633180379868, "global_step": 46293, "epoch": 436, "lr": 3.898424399861089e-06} +{"train_loss": 0.01781594008207321, "grad_norm": 0.1360296756029129, "global_step": 46294, "epoch": 436, "lr": 3.897277162830465e-06} +{"train_loss": 0.01591392233967781, "grad_norm": 0.06591179221868515, "global_step": 46295, "epoch": 436, "lr": 3.896130087784755e-06} +{"train_loss": 0.015033380128443241, "grad_norm": 0.07138755172491074, "global_step": 46296, "epoch": 436, "lr": 3.894983174727984e-06} +{"train_loss": 0.014364214614033699, "grad_norm": 0.06911465525627136, "global_step": 46297, "epoch": 436, "lr": 3.893836423664194e-06} +{"train_loss": 0.013213861733675003, "grad_norm": 0.08864365518093109, "global_step": 46298, "epoch": 436, "lr": 3.8926898345974185e-06} +{"train_loss": 0.011562585830688477, "grad_norm": 0.10906775295734406, "global_step": 46299, "epoch": 436, "lr": 3.891543407531673e-06} +{"train_loss": 0.015002207830548286, "grad_norm": 0.1276107281446457, "global_step": 46300, "epoch": 436, "lr": 3.890397142470992e-06} +{"train_loss": 0.015936851501464844, "grad_norm": 0.0613490492105484, "global_step": 46301, "epoch": 436, "lr": 3.889251039419412e-06} +{"train_loss": 0.016574349254369736, "grad_norm": 0.11264503747224808, "global_step": 46302, "epoch": 436, "lr": 3.888105098380945e-06} +{"train_loss": 0.015315938740968704, "grad_norm": 0.06739380210638046, "global_step": 46303, "epoch": 436, "lr": 3.886959319359623e-06} +{"train_loss": 0.014390699565410614, "grad_norm": 0.08910524100065231, "global_step": 46304, "epoch": 436, "lr": 3.885813702359464e-06} +{"train_loss": 0.013501801528036594, "grad_norm": 0.05991201102733612, "global_step": 46305, "epoch": 436, "lr": 3.884668247384516e-06} +{"train_loss": 0.01707247458398342, "grad_norm": 0.11661901324987411, "global_step": 46306, "epoch": 436, "lr": 3.8835229544387785e-06} +{"train_loss": 0.016109086573123932, "grad_norm": 0.08549244701862335, "global_step": 46307, "epoch": 436, "lr": 3.8823778235262895e-06} +{"train_loss": 0.01342657208442688, "grad_norm": 0.07411690056324005, "global_step": 46308, "epoch": 436, "lr": 3.881232854651074e-06} +{"train_loss": 0.01457242202013731, "grad_norm": 0.07325074076652527, "global_step": 46309, "epoch": 436, "lr": 3.880088047817143e-06} +{"train_loss": 0.015563318505883217, "grad_norm": 0.09858374297618866, "global_step": 46310, "epoch": 436, "lr": 3.878943403028517e-06} +{"train_loss": 0.012502890080213547, "grad_norm": 0.061419758945703506, "global_step": 46311, "epoch": 436, "lr": 3.877798920289244e-06} +{"train_loss": 0.012935005128383636, "grad_norm": 0.1411535143852234, "global_step": 46312, "epoch": 436, "lr": 3.8766545996033134e-06} +{"train_loss": 0.014551976695656776, "grad_norm": 0.0694480836391449, "global_step": 46313, "epoch": 436, "lr": 3.8755104409747725e-06} +{"train_loss": 0.014751626178622246, "grad_norm": 0.11671175062656403, "global_step": 46314, "epoch": 436, "lr": 3.874366444407618e-06} +{"train_loss": 0.013685546815395355, "grad_norm": 0.08673758804798126, "global_step": 46315, "epoch": 436, "lr": 3.873222609905891e-06} +{"train_loss": 0.014218547381460667, "grad_norm": 0.08259798586368561, "global_step": 46316, "epoch": 436, "lr": 3.872078937473589e-06} +{"train_loss": 0.016110146418213844, "grad_norm": 0.07942547649145126, "global_step": 46317, "epoch": 436, "lr": 3.870935427114741e-06} +{"train_loss": 0.013042539358139038, "grad_norm": 0.08599773794412613, "global_step": 46318, "epoch": 436, "lr": 3.869792078833373e-06} +{"train_loss": 0.01349685899913311, "grad_norm": 0.07621774077415466, "global_step": 46319, "epoch": 436, "lr": 3.868648892633497e-06} +{"train_loss": 0.015190092846751213, "grad_norm": 0.11701947450637817, "global_step": 46320, "epoch": 436, "lr": 3.867505868519117e-06} +{"train_loss": 0.014961185636666586, "grad_norm": 0.07639195024967194, "global_step": 46321, "epoch": 436, "lr": 3.866363006494255e-06, "val_loss": 0.019741086289286613} +{"train_loss": 0.012377722188830376, "grad_norm": 0.06848651170730591, "global_step": 46322, "epoch": 437, "lr": 3.8652203065629445e-06} +{"train_loss": 0.015300389379262924, "grad_norm": 0.14267878234386444, "global_step": 46323, "epoch": 437, "lr": 3.864077768729174e-06} +{"train_loss": 0.015498626977205276, "grad_norm": 0.08439264446496964, "global_step": 46324, "epoch": 437, "lr": 3.86293539299698e-06} +{"train_loss": 0.017185617238283157, "grad_norm": 0.10534420609474182, "global_step": 46325, "epoch": 437, "lr": 3.861793179370371e-06} +{"train_loss": 0.01700664311647415, "grad_norm": 0.10047557204961777, "global_step": 46326, "epoch": 437, "lr": 3.860651127853343e-06} +{"train_loss": 0.014412537217140198, "grad_norm": 0.17209558188915253, "global_step": 46327, "epoch": 437, "lr": 3.859509238449921e-06} +{"train_loss": 0.015964597463607788, "grad_norm": 0.06696899980306625, "global_step": 46328, "epoch": 437, "lr": 3.858367511164123e-06} +{"train_loss": 0.015234741382300854, "grad_norm": 0.08537551015615463, "global_step": 46329, "epoch": 437, "lr": 3.857225945999953e-06} +{"train_loss": 0.013719883747398853, "grad_norm": 0.04912504926323891, "global_step": 46330, "epoch": 437, "lr": 3.85608454296143e-06} +{"train_loss": 0.01652611419558525, "grad_norm": 0.14550743997097015, "global_step": 46331, "epoch": 437, "lr": 3.85494330205255e-06} +{"train_loss": 0.014897028915584087, "grad_norm": 0.12705768644809723, "global_step": 46332, "epoch": 437, "lr": 3.853802223277336e-06} +{"train_loss": 0.012701855972409248, "grad_norm": 0.056354839354753494, "global_step": 46333, "epoch": 437, "lr": 3.852661306639793e-06} +{"train_loss": 0.016346223652362823, "grad_norm": 0.1045653373003006, "global_step": 46334, "epoch": 437, "lr": 3.851520552143922e-06} +{"train_loss": 0.01471872441470623, "grad_norm": 0.08019232004880905, "global_step": 46335, "epoch": 437, "lr": 3.850379959793748e-06} +{"train_loss": 0.0154200354591012, "grad_norm": 0.07196962088346481, "global_step": 46336, "epoch": 437, "lr": 3.849239529593268e-06} +{"train_loss": 0.017798226326704025, "grad_norm": 0.09458830207586288, "global_step": 46337, "epoch": 437, "lr": 3.848099261546484e-06} +{"train_loss": 0.012361159548163414, "grad_norm": 0.07825423032045364, "global_step": 46338, "epoch": 437, "lr": 3.846959155657409e-06} +{"train_loss": 0.012965233996510506, "grad_norm": 0.05528559908270836, "global_step": 46339, "epoch": 437, "lr": 3.8458192119300515e-06} +{"train_loss": 0.01762222684919834, "grad_norm": 0.12470173090696335, "global_step": 46340, "epoch": 437, "lr": 3.8446794303684084e-06} +{"train_loss": 0.015029679983854294, "grad_norm": 0.07780974358320236, "global_step": 46341, "epoch": 437, "lr": 3.843539810976493e-06} +{"train_loss": 0.01669059321284294, "grad_norm": 0.05530086159706116, "global_step": 46342, "epoch": 437, "lr": 3.842400353758319e-06} +{"train_loss": 0.01336399745196104, "grad_norm": 0.06739280372858047, "global_step": 46343, "epoch": 437, "lr": 3.8412610587178595e-06} +{"train_loss": 0.013355939649045467, "grad_norm": 0.10838638246059418, "global_step": 46344, "epoch": 437, "lr": 3.840121925859136e-06} +{"train_loss": 0.015886105597019196, "grad_norm": 0.06143894046545029, "global_step": 46345, "epoch": 437, "lr": 3.838982955186154e-06} +{"train_loss": 0.015384607948362827, "grad_norm": 0.09248282760381699, "global_step": 46346, "epoch": 437, "lr": 3.837844146702907e-06} +{"train_loss": 0.014321185648441315, "grad_norm": 0.07376214116811752, "global_step": 46347, "epoch": 437, "lr": 3.836705500413395e-06} +{"train_loss": 0.015128226019442081, "grad_norm": 0.056728966534137726, "global_step": 46348, "epoch": 437, "lr": 3.8355670163216395e-06} +{"train_loss": 0.015860550105571747, "grad_norm": 0.07209448516368866, "global_step": 46349, "epoch": 437, "lr": 3.8344286944316245e-06} +{"train_loss": 0.012231563217937946, "grad_norm": 0.0843360498547554, "global_step": 46350, "epoch": 437, "lr": 3.8332905347473354e-06} +{"train_loss": 0.014163171872496605, "grad_norm": 0.08209876716136932, "global_step": 46351, "epoch": 437, "lr": 3.8321525372727925e-06} +{"train_loss": 0.013108784332871437, "grad_norm": 0.09564828872680664, "global_step": 46352, "epoch": 437, "lr": 3.831014702011993e-06} +{"train_loss": 0.014754233881831169, "grad_norm": 0.09956537187099457, "global_step": 46353, "epoch": 437, "lr": 3.82987702896892e-06} +{"train_loss": 0.014449812471866608, "grad_norm": 0.0826651081442833, "global_step": 46354, "epoch": 437, "lr": 3.828739518147589e-06} +{"train_loss": 0.013360736891627312, "grad_norm": 0.0515889897942543, "global_step": 46355, "epoch": 437, "lr": 3.82760216955198e-06} +{"train_loss": 0.014985714107751846, "grad_norm": 0.11680284887552261, "global_step": 46356, "epoch": 437, "lr": 3.826464983186107e-06} +{"train_loss": 0.014960657805204391, "grad_norm": 0.07168202102184296, "global_step": 46357, "epoch": 437, "lr": 3.825327959053948e-06} +{"train_loss": 0.017408572137355804, "grad_norm": 0.08336086571216583, "global_step": 46358, "epoch": 437, "lr": 3.824191097159507e-06} +{"train_loss": 0.016811076551675797, "grad_norm": 0.1332794725894928, "global_step": 46359, "epoch": 437, "lr": 3.823054397506781e-06} +{"train_loss": 0.013496637344360352, "grad_norm": 0.08044536411762238, "global_step": 46360, "epoch": 437, "lr": 3.821917860099761e-06} +{"train_loss": 0.015033808536827564, "grad_norm": 0.0632484182715416, "global_step": 46361, "epoch": 437, "lr": 3.820781484942432e-06} +{"train_loss": 0.011994671076536179, "grad_norm": 0.06645756959915161, "global_step": 46362, "epoch": 437, "lr": 3.819645272038796e-06} +{"train_loss": 0.013994313776493073, "grad_norm": 0.0652477815747261, "global_step": 46363, "epoch": 437, "lr": 3.8185092213928465e-06} +{"train_loss": 0.013683049008250237, "grad_norm": 0.06705275923013687, "global_step": 46364, "epoch": 437, "lr": 3.817373333008567e-06} +{"train_loss": 0.014333551749587059, "grad_norm": 0.0849401131272316, "global_step": 46365, "epoch": 437, "lr": 3.816237606889955e-06} +{"train_loss": 0.014349615201354027, "grad_norm": 0.0819418728351593, "global_step": 46366, "epoch": 437, "lr": 3.815102043041013e-06} +{"train_loss": 0.017110880464315414, "grad_norm": 0.10274114459753036, "global_step": 46367, "epoch": 437, "lr": 3.8139666414656994e-06} +{"train_loss": 0.015057149343192577, "grad_norm": 0.09094561636447906, "global_step": 46368, "epoch": 437, "lr": 3.8128314021680267e-06} +{"train_loss": 0.015013542957603931, "grad_norm": 0.08925385028123856, "global_step": 46369, "epoch": 437, "lr": 3.811696325151981e-06} +{"train_loss": 0.015553705394268036, "grad_norm": 0.07393675297498703, "global_step": 46370, "epoch": 437, "lr": 3.8105614104215424e-06} +{"train_loss": 0.013920046389102936, "grad_norm": 0.08736477047204971, "global_step": 46371, "epoch": 437, "lr": 3.8094266579807027e-06} +{"train_loss": 0.016696270555257797, "grad_norm": 0.11887228488922119, "global_step": 46372, "epoch": 437, "lr": 3.8082920678334577e-06} +{"train_loss": 0.015178605914115906, "grad_norm": 0.1409354954957962, "global_step": 46373, "epoch": 437, "lr": 3.8071576399837825e-06} +{"train_loss": 0.01745188981294632, "grad_norm": 0.09213193506002426, "global_step": 46374, "epoch": 437, "lr": 3.8060233744356633e-06} +{"train_loss": 0.012187700718641281, "grad_norm": 0.06988807022571564, "global_step": 46375, "epoch": 437, "lr": 3.8048892711930904e-06} +{"train_loss": 0.016001757234334946, "grad_norm": 0.07192590832710266, "global_step": 46376, "epoch": 437, "lr": 3.8037553302600504e-06} +{"train_loss": 0.015108020976185799, "grad_norm": 0.07218708097934723, "global_step": 46377, "epoch": 437, "lr": 3.802621551640528e-06} +{"train_loss": 0.013277750462293625, "grad_norm": 0.10251650959253311, "global_step": 46378, "epoch": 437, "lr": 3.8014879353384938e-06} +{"train_loss": 0.014272544533014297, "grad_norm": 0.12086131423711777, "global_step": 46379, "epoch": 437, "lr": 3.800354481357937e-06} +{"train_loss": 0.01520026009529829, "grad_norm": 0.1026758924126625, "global_step": 46380, "epoch": 437, "lr": 3.7992211897028507e-06} +{"train_loss": 0.014559590257704258, "grad_norm": 0.11386018246412277, "global_step": 46381, "epoch": 437, "lr": 3.798088060377203e-06} +{"train_loss": 0.015421207994222641, "grad_norm": 0.06024817004799843, "global_step": 46382, "epoch": 437, "lr": 3.7969550933849797e-06} +{"train_loss": 0.0143471360206604, "grad_norm": 0.10308217257261276, "global_step": 46383, "epoch": 437, "lr": 3.795822288730183e-06} +{"train_loss": 0.013726240023970604, "grad_norm": 0.053844988346099854, "global_step": 46384, "epoch": 437, "lr": 3.794689646416755e-06} +{"train_loss": 0.015507865697145462, "grad_norm": 0.06128283590078354, "global_step": 46385, "epoch": 437, "lr": 3.7935571664486914e-06} +{"train_loss": 0.017863092944025993, "grad_norm": 0.10586944967508316, "global_step": 46386, "epoch": 437, "lr": 3.792424848829984e-06} +{"train_loss": 0.013403275981545448, "grad_norm": 0.0831213891506195, "global_step": 46387, "epoch": 437, "lr": 3.7912926935645907e-06} +{"train_loss": 0.016495265066623688, "grad_norm": 0.07056053727865219, "global_step": 46388, "epoch": 437, "lr": 3.7901607006564974e-06} +{"train_loss": 0.017020652070641518, "grad_norm": 0.07147981971502304, "global_step": 46389, "epoch": 437, "lr": 3.7890288701096954e-06} +{"train_loss": 0.01575857773423195, "grad_norm": 0.08616763353347778, "global_step": 46390, "epoch": 437, "lr": 3.787897201928148e-06} +{"train_loss": 0.014443745836615562, "grad_norm": 0.0747884139418602, "global_step": 46391, "epoch": 437, "lr": 3.7867656961158237e-06} +{"train_loss": 0.016551131382584572, "grad_norm": 0.08553431928157806, "global_step": 46392, "epoch": 437, "lr": 3.785634352676709e-06} +{"train_loss": 0.01607501693069935, "grad_norm": 0.08786056935787201, "global_step": 46393, "epoch": 437, "lr": 3.784503171614784e-06} +{"train_loss": 0.014035303145647049, "grad_norm": 0.11601261794567108, "global_step": 46394, "epoch": 437, "lr": 3.783372152934006e-06} +{"train_loss": 0.01564639061689377, "grad_norm": 0.09836187958717346, "global_step": 46395, "epoch": 437, "lr": 3.7822412966383726e-06} +{"train_loss": 0.01651691645383835, "grad_norm": 0.07346063107252121, "global_step": 46396, "epoch": 437, "lr": 3.78111060273183e-06} +{"train_loss": 0.01601465791463852, "grad_norm": 0.07839848101139069, "global_step": 46397, "epoch": 437, "lr": 3.7799800712183753e-06} +{"train_loss": 0.014032557606697083, "grad_norm": 0.11196970194578171, "global_step": 46398, "epoch": 437, "lr": 3.778849702101961e-06} +{"train_loss": 0.014360930770635605, "grad_norm": 0.06713997572660446, "global_step": 46399, "epoch": 437, "lr": 3.7777194953865667e-06} +{"train_loss": 0.014686760492622852, "grad_norm": 0.07771551609039307, "global_step": 46400, "epoch": 437, "lr": 3.776589451076179e-06} +{"train_loss": 0.016552843153476715, "grad_norm": 0.13488925993442535, "global_step": 46401, "epoch": 437, "lr": 3.7754595691747495e-06} +{"train_loss": 0.015211179852485657, "grad_norm": 0.11403680592775345, "global_step": 46402, "epoch": 437, "lr": 3.774329849686242e-06} +{"train_loss": 0.01318722777068615, "grad_norm": 0.08577346056699753, "global_step": 46403, "epoch": 437, "lr": 3.773200292614637e-06} +{"train_loss": 0.013932399451732635, "grad_norm": 0.07302926480770111, "global_step": 46404, "epoch": 437, "lr": 3.7720708979639142e-06} +{"train_loss": 0.01479418110102415, "grad_norm": 0.09915830194950104, "global_step": 46405, "epoch": 437, "lr": 3.7709416657380203e-06} +{"train_loss": 0.014631971716880798, "grad_norm": 0.06281771510839462, "global_step": 46406, "epoch": 437, "lr": 3.769812595940936e-06} +{"train_loss": 0.01777837611734867, "grad_norm": 0.10333950072526932, "global_step": 46407, "epoch": 437, "lr": 3.768683688576635e-06} +{"train_loss": 0.013465614058077335, "grad_norm": 0.07524876296520233, "global_step": 46408, "epoch": 437, "lr": 3.7675549436490596e-06} +{"train_loss": 0.013367783278226852, "grad_norm": 0.0949302464723587, "global_step": 46409, "epoch": 437, "lr": 3.7664263611621952e-06} +{"train_loss": 0.0143280616030097, "grad_norm": 0.06154589354991913, "global_step": 46410, "epoch": 437, "lr": 3.7652979411200053e-06} +{"train_loss": 0.015068928711116314, "grad_norm": 0.08071263134479523, "global_step": 46411, "epoch": 437, "lr": 3.764169683526447e-06} +{"train_loss": 0.014899216592311859, "grad_norm": 0.06992486119270325, "global_step": 46412, "epoch": 437, "lr": 3.7630415883854854e-06} +{"train_loss": 0.014658362604677677, "grad_norm": 0.10587145388126373, "global_step": 46413, "epoch": 437, "lr": 3.7619136557010994e-06} +{"train_loss": 0.015395918861031532, "grad_norm": 0.07847557961940765, "global_step": 46414, "epoch": 437, "lr": 3.760785885477236e-06} +{"train_loss": 0.01609860546886921, "grad_norm": 0.1275632381439209, "global_step": 46415, "epoch": 437, "lr": 3.759658277717859e-06} +{"train_loss": 0.013133423402905464, "grad_norm": 0.059071436524391174, "global_step": 46416, "epoch": 437, "lr": 3.758530832426932e-06} +{"train_loss": 0.015210927464067936, "grad_norm": 0.06287068873643875, "global_step": 46417, "epoch": 437, "lr": 3.7574035496084236e-06} +{"train_loss": 0.01219768077135086, "grad_norm": 0.09859117120504379, "global_step": 46418, "epoch": 437, "lr": 3.756276429266281e-06} +{"train_loss": 0.015457254834473133, "grad_norm": 0.058712609112262726, "global_step": 46419, "epoch": 437, "lr": 3.7551494714044843e-06} +{"train_loss": 0.014360989443957806, "grad_norm": 0.15530672669410706, "global_step": 46420, "epoch": 437, "lr": 3.754022676026969e-06} +{"train_loss": 0.017490815371274948, "grad_norm": 0.07493255287408829, "global_step": 46421, "epoch": 437, "lr": 3.752896043137716e-06} +{"train_loss": 0.013923859223723412, "grad_norm": 0.0842178538441658, "global_step": 46422, "epoch": 437, "lr": 3.75176957274066e-06} +{"train_loss": 0.01765609346330166, "grad_norm": 0.05936574935913086, "global_step": 46423, "epoch": 437, "lr": 3.7506432648397815e-06} +{"train_loss": 0.014203310012817383, "grad_norm": 0.10319014638662338, "global_step": 46424, "epoch": 437, "lr": 3.7495171194390335e-06} +{"train_loss": 0.01423050556331873, "grad_norm": 0.07288859784603119, "global_step": 46425, "epoch": 437, "lr": 3.7483911365423675e-06} +{"train_loss": 0.0165322907269001, "grad_norm": 0.1064702570438385, "global_step": 46426, "epoch": 437, "lr": 3.747265316153736e-06} +{"train_loss": 0.014958169258568646, "grad_norm": 0.12598629295825958, "global_step": 46427, "epoch": 437, "lr": 3.7461396582771035e-06, "val_loss": 0.019729916006326675} +{"train_loss": 0.01412251591682434, "grad_norm": 0.06800411641597748, "global_step": 46428, "epoch": 438, "lr": 3.745014162916416e-06} +{"train_loss": 0.014636553823947906, "grad_norm": 0.06270669400691986, "global_step": 46429, "epoch": 438, "lr": 3.7438888300756313e-06} +{"train_loss": 0.012086601927876472, "grad_norm": 0.0556449256837368, "global_step": 46430, "epoch": 438, "lr": 3.7427636597587136e-06} +{"train_loss": 0.01638728380203247, "grad_norm": 0.0705910325050354, "global_step": 46431, "epoch": 438, "lr": 3.741638651969609e-06} +{"train_loss": 0.015002467669546604, "grad_norm": 0.07562749087810516, "global_step": 46432, "epoch": 438, "lr": 3.7405138067122593e-06} +{"train_loss": 0.01346861943602562, "grad_norm": 0.06922699511051178, "global_step": 46433, "epoch": 438, "lr": 3.739389123990622e-06} +{"train_loss": 0.017339065670967102, "grad_norm": 0.125726118683815, "global_step": 46434, "epoch": 438, "lr": 3.7382646038086665e-06} +{"train_loss": 0.015787281095981598, "grad_norm": 0.058828793466091156, "global_step": 46435, "epoch": 438, "lr": 3.737140246170323e-06} +{"train_loss": 0.015052445232868195, "grad_norm": 0.1163751557469368, "global_step": 46436, "epoch": 438, "lr": 3.736016051079555e-06} +{"train_loss": 0.014847524464130402, "grad_norm": 0.0528811551630497, "global_step": 46437, "epoch": 438, "lr": 3.7348920185403038e-06} +{"train_loss": 0.01612856611609459, "grad_norm": 0.1274893879890442, "global_step": 46438, "epoch": 438, "lr": 3.733768148556521e-06} +{"train_loss": 0.017650267109274864, "grad_norm": 0.06622528284788132, "global_step": 46439, "epoch": 438, "lr": 3.7326444411321547e-06} +{"train_loss": 0.014233903028070927, "grad_norm": 0.06743866205215454, "global_step": 46440, "epoch": 438, "lr": 3.7315208962711566e-06} +{"train_loss": 0.015925878658890724, "grad_norm": 0.06484279781579971, "global_step": 46441, "epoch": 438, "lr": 3.730397513977474e-06} +{"train_loss": 0.016298409551382065, "grad_norm": 0.0988708883523941, "global_step": 46442, "epoch": 438, "lr": 3.7292742942550583e-06} +{"train_loss": 0.014933433383703232, "grad_norm": 0.10749106109142303, "global_step": 46443, "epoch": 438, "lr": 3.728151237107835e-06} +{"train_loss": 0.015716318041086197, "grad_norm": 0.11345146596431732, "global_step": 46444, "epoch": 438, "lr": 3.727028342539779e-06} +{"train_loss": 0.014591431245207787, "grad_norm": 0.07218733429908752, "global_step": 46445, "epoch": 438, "lr": 3.725905610554814e-06} +{"train_loss": 0.017765939235687256, "grad_norm": 0.1354263871908188, "global_step": 46446, "epoch": 438, "lr": 3.724783041156893e-06} +{"train_loss": 0.013597107492387295, "grad_norm": 0.056298986077308655, "global_step": 46447, "epoch": 438, "lr": 3.7236606343499624e-06} +{"train_loss": 0.014484617859125137, "grad_norm": 0.06551574170589447, "global_step": 46448, "epoch": 438, "lr": 3.7225383901379695e-06} +{"train_loss": 0.015130698680877686, "grad_norm": 0.08030008524656296, "global_step": 46449, "epoch": 438, "lr": 3.721416308524839e-06} +{"train_loss": 0.012816104106605053, "grad_norm": 0.09349823743104935, "global_step": 46450, "epoch": 438, "lr": 3.720294389514528e-06} +{"train_loss": 0.01599758304655552, "grad_norm": 0.09032022207975388, "global_step": 46451, "epoch": 438, "lr": 3.7191726331109842e-06} +{"train_loss": 0.012997165322303772, "grad_norm": 0.08321656286716461, "global_step": 46452, "epoch": 438, "lr": 3.7180510393181324e-06} +{"train_loss": 0.016252368688583374, "grad_norm": 0.06437527388334274, "global_step": 46453, "epoch": 438, "lr": 3.7169296081399186e-06} +{"train_loss": 0.014436742290854454, "grad_norm": 0.053059231489896774, "global_step": 46454, "epoch": 438, "lr": 3.7158083395802954e-06} +{"train_loss": 0.012966827489435673, "grad_norm": 0.17174158990383148, "global_step": 46455, "epoch": 438, "lr": 3.7146872336431936e-06} +{"train_loss": 0.014326179400086403, "grad_norm": 0.08409959077835083, "global_step": 46456, "epoch": 438, "lr": 3.7135662903325487e-06} +{"train_loss": 0.013301745057106018, "grad_norm": 0.08343926817178726, "global_step": 46457, "epoch": 438, "lr": 3.712445509652296e-06} +{"train_loss": 0.01495825033634901, "grad_norm": 0.06949866563081741, "global_step": 46458, "epoch": 438, "lr": 3.711324891606388e-06} +{"train_loss": 0.01562936045229435, "grad_norm": 0.06584172695875168, "global_step": 46459, "epoch": 438, "lr": 3.71020443619875e-06} +{"train_loss": 0.01595785841345787, "grad_norm": 0.08963635563850403, "global_step": 46460, "epoch": 438, "lr": 3.709084143433328e-06} +{"train_loss": 0.0157150961458683, "grad_norm": 0.10185783356428146, "global_step": 46461, "epoch": 438, "lr": 3.707964013314047e-06} +{"train_loss": 0.015900855883955956, "grad_norm": 0.06078462675213814, "global_step": 46462, "epoch": 438, "lr": 3.7068440458448538e-06} +{"train_loss": 0.013642821460962296, "grad_norm": 0.06659777462482452, "global_step": 46463, "epoch": 438, "lr": 3.7057242410296734e-06} +{"train_loss": 0.0164711344987154, "grad_norm": 0.06739603728055954, "global_step": 46464, "epoch": 438, "lr": 3.704604598872441e-06} +{"train_loss": 0.012675262056291103, "grad_norm": 0.062874935567379, "global_step": 46465, "epoch": 438, "lr": 3.7034851193771093e-06} +{"train_loss": 0.015652615576982498, "grad_norm": 0.04867888614535332, "global_step": 46466, "epoch": 438, "lr": 3.7023658025475917e-06} +{"train_loss": 0.016670208424329758, "grad_norm": 0.06925707310438156, "global_step": 46467, "epoch": 438, "lr": 3.7012466483878183e-06} +{"train_loss": 0.014603479765355587, "grad_norm": 0.052049070596694946, "global_step": 46468, "epoch": 438, "lr": 3.700127656901742e-06} +{"train_loss": 0.016173891723155975, "grad_norm": 0.11289119720458984, "global_step": 46469, "epoch": 438, "lr": 3.69900882809327e-06} +{"train_loss": 0.01536181103438139, "grad_norm": 0.05568268150091171, "global_step": 46470, "epoch": 438, "lr": 3.6978901619663443e-06} +{"train_loss": 0.01499854028224945, "grad_norm": 0.09264685213565826, "global_step": 46471, "epoch": 438, "lr": 3.6967716585249057e-06} +{"train_loss": 0.015450451523065567, "grad_norm": 0.07015611231327057, "global_step": 46472, "epoch": 438, "lr": 3.6956533177728735e-06} +{"train_loss": 0.017486244440078735, "grad_norm": 0.08187016099691391, "global_step": 46473, "epoch": 438, "lr": 3.6945351397141725e-06} +{"train_loss": 0.01598263904452324, "grad_norm": 0.06338600069284439, "global_step": 46474, "epoch": 438, "lr": 3.6934171243527383e-06} +{"train_loss": 0.016645584255456924, "grad_norm": 0.07108461856842041, "global_step": 46475, "epoch": 438, "lr": 3.6922992716925064e-06} +{"train_loss": 0.013678641989827156, "grad_norm": 0.07547274231910706, "global_step": 46476, "epoch": 438, "lr": 3.6911815817373853e-06} +{"train_loss": 0.014111708849668503, "grad_norm": 0.0681285560131073, "global_step": 46477, "epoch": 438, "lr": 3.6900640544913213e-06} +{"train_loss": 0.018277011811733246, "grad_norm": 0.07277484238147736, "global_step": 46478, "epoch": 438, "lr": 3.6889466899582228e-06} +{"train_loss": 0.013854866847395897, "grad_norm": 0.061389997601509094, "global_step": 46479, "epoch": 438, "lr": 3.6878294881420363e-06} +{"train_loss": 0.013219843618571758, "grad_norm": 0.05801963806152344, "global_step": 46480, "epoch": 438, "lr": 3.6867124490466698e-06} +{"train_loss": 0.01511378400027752, "grad_norm": 0.0859341248869896, "global_step": 46481, "epoch": 438, "lr": 3.6855955726760538e-06} +{"train_loss": 0.012951463460922241, "grad_norm": 0.05332290008664131, "global_step": 46482, "epoch": 438, "lr": 3.684478859034118e-06} +{"train_loss": 0.015451619401574135, "grad_norm": 0.11737880110740662, "global_step": 46483, "epoch": 438, "lr": 3.6833623081247826e-06} +{"train_loss": 0.013127760961651802, "grad_norm": 0.08737868815660477, "global_step": 46484, "epoch": 438, "lr": 3.6822459199519654e-06} +{"train_loss": 0.014367027208209038, "grad_norm": 0.09114152938127518, "global_step": 46485, "epoch": 438, "lr": 3.6811296945195973e-06} +{"train_loss": 0.01579231768846512, "grad_norm": 0.15403980016708374, "global_step": 46486, "epoch": 438, "lr": 3.680013631831586e-06} +{"train_loss": 0.014269000850617886, "grad_norm": 0.07939714193344116, "global_step": 46487, "epoch": 438, "lr": 3.6788977318918672e-06} +{"train_loss": 0.012890946120023727, "grad_norm": 0.06553375720977783, "global_step": 46488, "epoch": 438, "lr": 3.6777819947043603e-06} +{"train_loss": 0.013492610305547714, "grad_norm": 0.0998297929763794, "global_step": 46489, "epoch": 438, "lr": 3.6766664202729894e-06} +{"train_loss": 0.012443043291568756, "grad_norm": 0.05809812620282173, "global_step": 46490, "epoch": 438, "lr": 3.675551008601652e-06} +{"train_loss": 0.01588287577033043, "grad_norm": 0.07972966879606247, "global_step": 46491, "epoch": 438, "lr": 3.674435759694289e-06} +{"train_loss": 0.016365889459848404, "grad_norm": 0.1305980682373047, "global_step": 46492, "epoch": 438, "lr": 3.673320673554814e-06} +{"train_loss": 0.012231209315359592, "grad_norm": 0.09659293293952942, "global_step": 46493, "epoch": 438, "lr": 3.672205750187141e-06} +{"train_loss": 0.0159505195915699, "grad_norm": 0.07386072725057602, "global_step": 46494, "epoch": 438, "lr": 3.6710909895951886e-06} +{"train_loss": 0.015087769366800785, "grad_norm": 0.1547074168920517, "global_step": 46495, "epoch": 438, "lr": 3.6699763917828812e-06} +{"train_loss": 0.015471206977963448, "grad_norm": 0.06155341863632202, "global_step": 46496, "epoch": 438, "lr": 3.668861956754127e-06} +{"train_loss": 0.016601914539933205, "grad_norm": 0.09522319585084915, "global_step": 46497, "epoch": 438, "lr": 3.6677476845128346e-06} +{"train_loss": 0.01632477343082428, "grad_norm": 0.06071515008807182, "global_step": 46498, "epoch": 438, "lr": 3.666633575062928e-06} +{"train_loss": 0.015921827405691147, "grad_norm": 0.10518436878919601, "global_step": 46499, "epoch": 438, "lr": 3.6655196284083317e-06} +{"train_loss": 0.016202908009290695, "grad_norm": 0.06273158639669418, "global_step": 46500, "epoch": 438, "lr": 3.6644058445529372e-06} +{"train_loss": 0.011858870275318623, "grad_norm": 0.07382167130708694, "global_step": 46501, "epoch": 438, "lr": 3.6632922235006807e-06} +{"train_loss": 0.01741485670208931, "grad_norm": 0.09898363798856735, "global_step": 46502, "epoch": 438, "lr": 3.662178765255464e-06} +{"train_loss": 0.014912529848515987, "grad_norm": 0.06704629957675934, "global_step": 46503, "epoch": 438, "lr": 3.66106546982119e-06} +{"train_loss": 0.013820292428135872, "grad_norm": 0.09121784567832947, "global_step": 46504, "epoch": 438, "lr": 3.659952337201783e-06} +{"train_loss": 0.01742095686495304, "grad_norm": 0.09590642899274826, "global_step": 46505, "epoch": 438, "lr": 3.6588393674011457e-06} +{"train_loss": 0.014644145034253597, "grad_norm": 0.07836858183145523, "global_step": 46506, "epoch": 438, "lr": 3.6577265604232026e-06} +{"train_loss": 0.018522312864661217, "grad_norm": 0.10676952451467514, "global_step": 46507, "epoch": 438, "lr": 3.656613916271856e-06} +{"train_loss": 0.013243518769741058, "grad_norm": 0.07356448471546173, "global_step": 46508, "epoch": 438, "lr": 3.6555014349510087e-06} +{"train_loss": 0.015435278415679932, "grad_norm": 0.06725355237722397, "global_step": 46509, "epoch": 438, "lr": 3.6543891164645794e-06} +{"train_loss": 0.013625719584524632, "grad_norm": 0.0641784742474556, "global_step": 46510, "epoch": 438, "lr": 3.6532769608164653e-06} +{"train_loss": 0.013877836987376213, "grad_norm": 0.07085876911878586, "global_step": 46511, "epoch": 438, "lr": 3.65216496801058e-06} +{"train_loss": 0.015391990542411804, "grad_norm": 0.06073040887713432, "global_step": 46512, "epoch": 438, "lr": 3.651053138050836e-06} +{"train_loss": 0.013802297413349152, "grad_norm": 0.08295898139476776, "global_step": 46513, "epoch": 438, "lr": 3.6499414709411373e-06} +{"train_loss": 0.01562882959842682, "grad_norm": 0.10888399183750153, "global_step": 46514, "epoch": 438, "lr": 3.64882996668538e-06} +{"train_loss": 0.015066670253872871, "grad_norm": 0.08125082403421402, "global_step": 46515, "epoch": 438, "lr": 3.647718625287472e-06} +{"train_loss": 0.01410181075334549, "grad_norm": 0.07393879443407059, "global_step": 46516, "epoch": 438, "lr": 3.6466074467513323e-06} +{"train_loss": 0.013822712004184723, "grad_norm": 0.0865563377737999, "global_step": 46517, "epoch": 438, "lr": 3.6454964310808526e-06} +{"train_loss": 0.01462310180068016, "grad_norm": 0.10771851986646652, "global_step": 46518, "epoch": 438, "lr": 3.6443855782799353e-06} +{"train_loss": 0.01649927720427513, "grad_norm": 0.08355160802602768, "global_step": 46519, "epoch": 438, "lr": 3.6432748883524935e-06} +{"train_loss": 0.016736643388867378, "grad_norm": 0.08938045054674149, "global_step": 46520, "epoch": 438, "lr": 3.642164361302425e-06} +{"train_loss": 0.014631585218012333, "grad_norm": 0.10162445902824402, "global_step": 46521, "epoch": 438, "lr": 3.6410539971336254e-06} +{"train_loss": 0.013775376603007317, "grad_norm": 0.07251759618520737, "global_step": 46522, "epoch": 438, "lr": 3.6399437958499982e-06} +{"train_loss": 0.015089081600308418, "grad_norm": 0.05494223162531853, "global_step": 46523, "epoch": 438, "lr": 3.6388337574554566e-06} +{"train_loss": 0.01472310721874237, "grad_norm": 0.06451723724603653, "global_step": 46524, "epoch": 438, "lr": 3.637723881953886e-06} +{"train_loss": 0.01403727661818266, "grad_norm": 0.1008504331111908, "global_step": 46525, "epoch": 438, "lr": 3.636614169349195e-06} +{"train_loss": 0.012967046350240707, "grad_norm": 0.07288744300603867, "global_step": 46526, "epoch": 438, "lr": 3.6355046196452804e-06} +{"train_loss": 0.016567029058933258, "grad_norm": 0.0839478150010109, "global_step": 46527, "epoch": 438, "lr": 3.6343952328460275e-06} +{"train_loss": 0.013795075938105583, "grad_norm": 0.11355605721473694, "global_step": 46528, "epoch": 438, "lr": 3.63328600895535e-06} +{"train_loss": 0.015405325219035149, "grad_norm": 0.08896157145500183, "global_step": 46529, "epoch": 438, "lr": 3.6321769479771504e-06} +{"train_loss": 0.010488745756447315, "grad_norm": 0.06966540962457657, "global_step": 46530, "epoch": 438, "lr": 3.631068049915315e-06} +{"train_loss": 0.015536212362349033, "grad_norm": 0.0746600553393364, "global_step": 46531, "epoch": 438, "lr": 3.6299593147737286e-06} +{"train_loss": 0.016135726124048233, "grad_norm": 0.11476355791091919, "global_step": 46532, "epoch": 438, "lr": 3.6288507425563058e-06} +{"train_loss": 0.014962003988055687, "grad_norm": 0.1227957010269165, "global_step": 46533, "epoch": 438, "lr": 3.627742333266937e-06, "val_loss": 0.020404674112796783} +{"train_loss": 0.013722419738769531, "grad_norm": 0.0988323912024498, "global_step": 46534, "epoch": 439, "lr": 3.626634086909514e-06} +{"train_loss": 0.013462094590067863, "grad_norm": 0.08189485222101212, "global_step": 46535, "epoch": 439, "lr": 3.625526003487928e-06} +{"train_loss": 0.015088481828570366, "grad_norm": 0.07457299530506134, "global_step": 46536, "epoch": 439, "lr": 3.6244180830060813e-06} +{"train_loss": 0.01759951189160347, "grad_norm": 0.0813215896487236, "global_step": 46537, "epoch": 439, "lr": 3.6233103254678656e-06} +{"train_loss": 0.013701558113098145, "grad_norm": 0.09244014322757721, "global_step": 46538, "epoch": 439, "lr": 3.6222027308771557e-06} +{"train_loss": 0.014188040047883987, "grad_norm": 0.07559603452682495, "global_step": 46539, "epoch": 439, "lr": 3.6210952992378586e-06} +{"train_loss": 0.015600526705384254, "grad_norm": 0.0657348483800888, "global_step": 46540, "epoch": 439, "lr": 3.6199880305538724e-06} +{"train_loss": 0.016149021685123444, "grad_norm": 0.07973963767290115, "global_step": 46541, "epoch": 439, "lr": 3.6188809248290712e-06} +{"train_loss": 0.014785869047045708, "grad_norm": 0.06686562299728394, "global_step": 46542, "epoch": 439, "lr": 3.617773982067357e-06} +{"train_loss": 0.016462089493870735, "grad_norm": 0.07593818753957748, "global_step": 46543, "epoch": 439, "lr": 3.6166672022726166e-06} +{"train_loss": 0.014720575883984566, "grad_norm": 0.16364973783493042, "global_step": 46544, "epoch": 439, "lr": 3.615560585448724e-06} +{"train_loss": 0.014349639415740967, "grad_norm": 0.10348260402679443, "global_step": 46545, "epoch": 439, "lr": 3.6144541315995816e-06} +{"train_loss": 0.013634863309562206, "grad_norm": 0.06459540128707886, "global_step": 46546, "epoch": 439, "lr": 3.6133478407290867e-06} +{"train_loss": 0.014643970876932144, "grad_norm": 0.06999959796667099, "global_step": 46547, "epoch": 439, "lr": 3.6122417128411025e-06} +{"train_loss": 0.01648980751633644, "grad_norm": 0.0838969275355339, "global_step": 46548, "epoch": 439, "lr": 3.6111357479395368e-06} +{"train_loss": 0.01531771570444107, "grad_norm": 0.07638286054134369, "global_step": 46549, "epoch": 439, "lr": 3.610029946028254e-06} +{"train_loss": 0.01357970479875803, "grad_norm": 0.12165044993162155, "global_step": 46550, "epoch": 439, "lr": 3.6089243071111613e-06} +{"train_loss": 0.014981891959905624, "grad_norm": 0.08359923958778381, "global_step": 46551, "epoch": 439, "lr": 3.607818831192128e-06} +{"train_loss": 0.014717446640133858, "grad_norm": 0.10686857253313065, "global_step": 46552, "epoch": 439, "lr": 3.6067135182750454e-06} +{"train_loss": 0.014905672520399094, "grad_norm": 0.12771296501159668, "global_step": 46553, "epoch": 439, "lr": 3.6056083683638e-06} +{"train_loss": 0.015027882531285286, "grad_norm": 0.06453805416822433, "global_step": 46554, "epoch": 439, "lr": 3.604503381462271e-06} +{"train_loss": 0.015439353883266449, "grad_norm": 0.07388471066951752, "global_step": 46555, "epoch": 439, "lr": 3.6033985575743335e-06} +{"train_loss": 0.016372013837099075, "grad_norm": 0.12276003509759903, "global_step": 46556, "epoch": 439, "lr": 3.6022938967038735e-06} +{"train_loss": 0.014306183904409409, "grad_norm": 0.15319836139678955, "global_step": 46557, "epoch": 439, "lr": 3.6011893988547873e-06} +{"train_loss": 0.0173475444316864, "grad_norm": 0.1403971165418625, "global_step": 46558, "epoch": 439, "lr": 3.6000850640309337e-06} +{"train_loss": 0.012814321555197239, "grad_norm": 0.1031203642487526, "global_step": 46559, "epoch": 439, "lr": 3.598980892236198e-06} +{"train_loss": 0.015732167288661003, "grad_norm": 0.09257898479700089, "global_step": 46560, "epoch": 439, "lr": 3.597876883474477e-06} +{"train_loss": 0.01632985845208168, "grad_norm": 0.12100592255592346, "global_step": 46561, "epoch": 439, "lr": 3.5967730377496347e-06} +{"train_loss": 0.0166049525141716, "grad_norm": 0.0780610740184784, "global_step": 46562, "epoch": 439, "lr": 3.595669355065545e-06} +{"train_loss": 0.016430146992206573, "grad_norm": 0.14192400872707367, "global_step": 46563, "epoch": 439, "lr": 3.594565835426089e-06} +{"train_loss": 0.014990797266364098, "grad_norm": 0.0747910737991333, "global_step": 46564, "epoch": 439, "lr": 3.593462478835158e-06} +{"train_loss": 0.013873820193111897, "grad_norm": 0.12655149400234222, "global_step": 46565, "epoch": 439, "lr": 3.592359285296615e-06} +{"train_loss": 0.01577078364789486, "grad_norm": 0.10709021985530853, "global_step": 46566, "epoch": 439, "lr": 3.59125625481434e-06} +{"train_loss": 0.014662687666714191, "grad_norm": 0.11065342277288437, "global_step": 46567, "epoch": 439, "lr": 3.5901533873922143e-06} +{"train_loss": 0.014999181032180786, "grad_norm": 0.07512959092855453, "global_step": 46568, "epoch": 439, "lr": 3.5890506830340897e-06} +{"train_loss": 0.012663990259170532, "grad_norm": 0.06205277517437935, "global_step": 46569, "epoch": 439, "lr": 3.587948141743863e-06} +{"train_loss": 0.016462579369544983, "grad_norm": 0.1136789470911026, "global_step": 46570, "epoch": 439, "lr": 3.5868457635254147e-06} +{"train_loss": 0.01284278929233551, "grad_norm": 0.08818109333515167, "global_step": 46571, "epoch": 439, "lr": 3.585743548382592e-06} +{"train_loss": 0.015615729615092278, "grad_norm": 0.10123506188392639, "global_step": 46572, "epoch": 439, "lr": 3.584641496319291e-06} +{"train_loss": 0.013184787705540657, "grad_norm": 0.10471300780773163, "global_step": 46573, "epoch": 439, "lr": 3.5835396073393645e-06} +{"train_loss": 0.015288200229406357, "grad_norm": 0.0840226411819458, "global_step": 46574, "epoch": 439, "lr": 3.5824378814467043e-06} +{"train_loss": 0.017271699383854866, "grad_norm": 0.07158875465393066, "global_step": 46575, "epoch": 439, "lr": 3.581336318645162e-06} +{"train_loss": 0.015562200918793678, "grad_norm": 0.08220316469669342, "global_step": 46576, "epoch": 439, "lr": 3.5802349189386187e-06} +{"train_loss": 0.014314617961645126, "grad_norm": 0.08249495178461075, "global_step": 46577, "epoch": 439, "lr": 3.5791336823309486e-06} +{"train_loss": 0.01573197916150093, "grad_norm": 0.08158081769943237, "global_step": 46578, "epoch": 439, "lr": 3.578032608826015e-06} +{"train_loss": 0.013816081918776035, "grad_norm": 0.057526301592588425, "global_step": 46579, "epoch": 439, "lr": 3.5769316984276767e-06} +{"train_loss": 0.013352528214454651, "grad_norm": 0.1438157558441162, "global_step": 46580, "epoch": 439, "lr": 3.5758309511398137e-06} +{"train_loss": 0.01418828684836626, "grad_norm": 0.06343906372785568, "global_step": 46581, "epoch": 439, "lr": 3.5747303669663e-06} +{"train_loss": 0.015601551160216331, "grad_norm": 0.0654495358467102, "global_step": 46582, "epoch": 439, "lr": 3.5736299459109836e-06} +{"train_loss": 0.014493675902485847, "grad_norm": 0.10093192756175995, "global_step": 46583, "epoch": 439, "lr": 3.572529687977749e-06} +{"train_loss": 0.016608648002147675, "grad_norm": 0.09105989336967468, "global_step": 46584, "epoch": 439, "lr": 3.5714295931704555e-06} +{"train_loss": 0.015446679666638374, "grad_norm": 0.057719435542821884, "global_step": 46585, "epoch": 439, "lr": 3.5703296614929603e-06} +{"train_loss": 0.014573000371456146, "grad_norm": 0.10243528336286545, "global_step": 46586, "epoch": 439, "lr": 3.569229892949133e-06} +{"train_loss": 0.014338342472910881, "grad_norm": 0.05759356543421745, "global_step": 46587, "epoch": 439, "lr": 3.568130287542848e-06} +{"train_loss": 0.013752326369285583, "grad_norm": 0.05826181545853615, "global_step": 46588, "epoch": 439, "lr": 3.5670308452779575e-06} +{"train_loss": 0.015481488779187202, "grad_norm": 0.076244056224823, "global_step": 46589, "epoch": 439, "lr": 3.5659315661583316e-06} +{"train_loss": 0.015284263528883457, "grad_norm": 0.10911506414413452, "global_step": 46590, "epoch": 439, "lr": 3.564832450187816e-06} +{"train_loss": 0.01548640988767147, "grad_norm": 0.0857061967253685, "global_step": 46591, "epoch": 439, "lr": 3.5637334973702975e-06} +{"train_loss": 0.015436775982379913, "grad_norm": 0.055603887885808945, "global_step": 46592, "epoch": 439, "lr": 3.562634707709617e-06} +{"train_loss": 0.0136429313570261, "grad_norm": 0.08915767818689346, "global_step": 46593, "epoch": 439, "lr": 3.5615360812096432e-06} +{"train_loss": 0.015459293499588966, "grad_norm": 0.06251244246959686, "global_step": 46594, "epoch": 439, "lr": 3.560437617874246e-06} +{"train_loss": 0.01555330865085125, "grad_norm": 0.07672956585884094, "global_step": 46595, "epoch": 439, "lr": 3.559339317707272e-06} +{"train_loss": 0.016083676367998123, "grad_norm": 0.059439919888973236, "global_step": 46596, "epoch": 439, "lr": 3.5582411807125736e-06} +{"train_loss": 0.014271320775151253, "grad_norm": 0.09045927971601486, "global_step": 46597, "epoch": 439, "lr": 3.5571432068940204e-06} +{"train_loss": 0.013586001470685005, "grad_norm": 0.08517777174711227, "global_step": 46598, "epoch": 439, "lr": 3.556045396255475e-06} +{"train_loss": 0.01477124635130167, "grad_norm": 0.06108458340167999, "global_step": 46599, "epoch": 439, "lr": 3.5549477488007854e-06} +{"train_loss": 0.015026681125164032, "grad_norm": 0.05724311247467995, "global_step": 46600, "epoch": 439, "lr": 3.5538502645338035e-06} +{"train_loss": 0.015147073194384575, "grad_norm": 0.09119083732366562, "global_step": 46601, "epoch": 439, "lr": 3.552752943458415e-06} +{"train_loss": 0.014996493235230446, "grad_norm": 0.06256761401891708, "global_step": 46602, "epoch": 439, "lr": 3.551655785578434e-06} +{"train_loss": 0.016473006457090378, "grad_norm": 0.08593703806400299, "global_step": 46603, "epoch": 439, "lr": 3.55055879089774e-06} +{"train_loss": 0.015476353466510773, "grad_norm": 0.05665287375450134, "global_step": 46604, "epoch": 439, "lr": 3.549461959420186e-06} +{"train_loss": 0.014801856130361557, "grad_norm": 0.06140675023198128, "global_step": 46605, "epoch": 439, "lr": 3.5483652911496133e-06} +{"train_loss": 0.014543995261192322, "grad_norm": 0.09066029638051987, "global_step": 46606, "epoch": 439, "lr": 3.5472687860898803e-06} +{"train_loss": 0.012878240086138248, "grad_norm": 0.1219760999083519, "global_step": 46607, "epoch": 439, "lr": 3.5461724442448552e-06} +{"train_loss": 0.015263247303664684, "grad_norm": 0.07159235328435898, "global_step": 46608, "epoch": 439, "lr": 3.54507626561838e-06} +{"train_loss": 0.016814017668366432, "grad_norm": 0.11349251866340637, "global_step": 46609, "epoch": 439, "lr": 3.543980250214296e-06} +{"train_loss": 0.01461594458669424, "grad_norm": 0.08411014080047607, "global_step": 46610, "epoch": 439, "lr": 3.542884398036456e-06} +{"train_loss": 0.013444487005472183, "grad_norm": 0.09978750348091125, "global_step": 46611, "epoch": 439, "lr": 3.541788709088728e-06} +{"train_loss": 0.015011105686426163, "grad_norm": 0.08476130664348602, "global_step": 46612, "epoch": 439, "lr": 3.5406931833749436e-06} +{"train_loss": 0.015676181763410568, "grad_norm": 0.08439268171787262, "global_step": 46613, "epoch": 439, "lr": 3.5395978208989655e-06} +{"train_loss": 0.017526529729366302, "grad_norm": 0.11452795565128326, "global_step": 46614, "epoch": 439, "lr": 3.53850262166463e-06} +{"train_loss": 0.01581595093011856, "grad_norm": 0.0800434798002243, "global_step": 46615, "epoch": 439, "lr": 3.5374075856757947e-06} +{"train_loss": 0.01235174760222435, "grad_norm": 0.1247163712978363, "global_step": 46616, "epoch": 439, "lr": 3.5363127129363016e-06} +{"train_loss": 0.01680033653974533, "grad_norm": 0.0679684430360794, "global_step": 46617, "epoch": 439, "lr": 3.5352180034499916e-06} +{"train_loss": 0.014581315219402313, "grad_norm": 0.06878693401813507, "global_step": 46618, "epoch": 439, "lr": 3.5341234572207284e-06} +{"train_loss": 0.015001627616584301, "grad_norm": 0.07372402399778366, "global_step": 46619, "epoch": 439, "lr": 3.5330290742523532e-06} +{"train_loss": 0.014782669022679329, "grad_norm": 0.08623377978801727, "global_step": 46620, "epoch": 439, "lr": 3.5319348545486906e-06} +{"train_loss": 0.015123608522117138, "grad_norm": 0.06717004626989365, "global_step": 46621, "epoch": 439, "lr": 3.530840798113605e-06} +{"train_loss": 0.016378970816731453, "grad_norm": 0.11207417398691177, "global_step": 46622, "epoch": 439, "lr": 3.5297469049509425e-06} +{"train_loss": 0.01297627855092287, "grad_norm": 0.09559499472379684, "global_step": 46623, "epoch": 439, "lr": 3.5286531750645336e-06} +{"train_loss": 0.01179485209286213, "grad_norm": 0.12218163162469864, "global_step": 46624, "epoch": 439, "lr": 3.5275596084582253e-06} +{"train_loss": 0.015939760953187943, "grad_norm": 0.062415678054094315, "global_step": 46625, "epoch": 439, "lr": 3.526466205135881e-06} +{"train_loss": 0.012483375146985054, "grad_norm": 0.061950523406267166, "global_step": 46626, "epoch": 439, "lr": 3.5253729651013033e-06} +{"train_loss": 0.01324920542538166, "grad_norm": 0.12067443877458572, "global_step": 46627, "epoch": 439, "lr": 3.524279888358356e-06} +{"train_loss": 0.014947640709578991, "grad_norm": 0.07328805327415466, "global_step": 46628, "epoch": 439, "lr": 3.523186974910886e-06} +{"train_loss": 0.015835553407669067, "grad_norm": 0.09481161087751389, "global_step": 46629, "epoch": 439, "lr": 3.5220942247627175e-06} +{"train_loss": 0.01606018655002117, "grad_norm": 0.07887796312570572, "global_step": 46630, "epoch": 439, "lr": 3.5210016379177036e-06} +{"train_loss": 0.013329688459634781, "grad_norm": 0.1032169833779335, "global_step": 46631, "epoch": 439, "lr": 3.5199092143796687e-06} +{"train_loss": 0.015051991678774357, "grad_norm": 0.07860882580280304, "global_step": 46632, "epoch": 439, "lr": 3.518816954152465e-06} +{"train_loss": 0.013576781377196312, "grad_norm": 0.06651246547698975, "global_step": 46633, "epoch": 439, "lr": 3.517724857239918e-06} +{"train_loss": 0.014739208854734898, "grad_norm": 0.0717645213007927, "global_step": 46634, "epoch": 439, "lr": 3.516632923645874e-06} +{"train_loss": 0.01652149297297001, "grad_norm": 0.11550827324390411, "global_step": 46635, "epoch": 439, "lr": 3.5155411533741744e-06} +{"train_loss": 0.014676826074719429, "grad_norm": 0.0679665356874466, "global_step": 46636, "epoch": 439, "lr": 3.514449546428644e-06} +{"train_loss": 0.016703901812434196, "grad_norm": 0.0828789621591568, "global_step": 46637, "epoch": 439, "lr": 3.513358102813119e-06} +{"train_loss": 0.013703498058021069, "grad_norm": 0.07887640595436096, "global_step": 46638, "epoch": 439, "lr": 3.5122668225314348e-06} +{"train_loss": 0.014972790873627056, "grad_norm": 0.10354724526405334, "global_step": 46639, "epoch": 439, "lr": 3.511175705587433e-06, "val_loss": 0.019481953233480453} +{"train_loss": 0.01611027494072914, "grad_norm": 0.09501434117555618, "global_step": 46640, "epoch": 440, "lr": 3.510084751984938e-06} +{"train_loss": 0.014184831641614437, "grad_norm": 0.09958191215991974, "global_step": 46641, "epoch": 440, "lr": 3.5089939617277857e-06} +{"train_loss": 0.016094684600830078, "grad_norm": 0.07258438318967819, "global_step": 46642, "epoch": 440, "lr": 3.5079033348198287e-06} +{"train_loss": 0.018354490399360657, "grad_norm": 0.06195949763059616, "global_step": 46643, "epoch": 440, "lr": 3.5068128712648586e-06} +{"train_loss": 0.016259748488664627, "grad_norm": 0.07338552922010422, "global_step": 46644, "epoch": 440, "lr": 3.505722571066733e-06} +{"train_loss": 0.014063600450754166, "grad_norm": 0.11966497451066971, "global_step": 46645, "epoch": 440, "lr": 3.504632434229288e-06} +{"train_loss": 0.017245452851057053, "grad_norm": 0.06702955067157745, "global_step": 46646, "epoch": 440, "lr": 3.5035424607563317e-06} +{"train_loss": 0.013660595752298832, "grad_norm": 0.08585026115179062, "global_step": 46647, "epoch": 440, "lr": 3.502452650651711e-06} +{"train_loss": 0.015064517967402935, "grad_norm": 0.07352156192064285, "global_step": 46648, "epoch": 440, "lr": 3.501363003919256e-06} +{"train_loss": 0.013422786258161068, "grad_norm": 0.09612344950437546, "global_step": 46649, "epoch": 440, "lr": 3.5002735205627914e-06} +{"train_loss": 0.017146263271570206, "grad_norm": 0.061605650931596756, "global_step": 46650, "epoch": 440, "lr": 3.499184200586131e-06} +{"train_loss": 0.01564439758658409, "grad_norm": 0.07613939046859741, "global_step": 46651, "epoch": 440, "lr": 3.4980950439931215e-06} +{"train_loss": 0.012929685413837433, "grad_norm": 0.11043545603752136, "global_step": 46652, "epoch": 440, "lr": 3.4970060507875825e-06} +{"train_loss": 0.01548832468688488, "grad_norm": 0.0796787366271019, "global_step": 46653, "epoch": 440, "lr": 3.4959172209733383e-06} +{"train_loss": 0.01520207617431879, "grad_norm": 0.07844123989343643, "global_step": 46654, "epoch": 440, "lr": 3.4948285545542247e-06} +{"train_loss": 0.01639164239168167, "grad_norm": 0.08547703176736832, "global_step": 46655, "epoch": 440, "lr": 3.4937400515340503e-06} +{"train_loss": 0.017784666270017624, "grad_norm": 0.11706901341676712, "global_step": 46656, "epoch": 440, "lr": 3.4926517119166556e-06} +{"train_loss": 0.015219494700431824, "grad_norm": 0.1210850402712822, "global_step": 46657, "epoch": 440, "lr": 3.4915635357058496e-06} +{"train_loss": 0.014356869272887707, "grad_norm": 0.0862048789858818, "global_step": 46658, "epoch": 440, "lr": 3.490475522905462e-06} +{"train_loss": 0.013445699587464333, "grad_norm": 0.06206747889518738, "global_step": 46659, "epoch": 440, "lr": 3.4893876735193287e-06} +{"train_loss": 0.01520091574639082, "grad_norm": 0.13317593932151794, "global_step": 46660, "epoch": 440, "lr": 3.4882999875512577e-06} +{"train_loss": 0.0158168263733387, "grad_norm": 0.07004024088382721, "global_step": 46661, "epoch": 440, "lr": 3.4872124650050687e-06} +{"train_loss": 0.014994310215115547, "grad_norm": 0.10283200442790985, "global_step": 46662, "epoch": 440, "lr": 3.48612510588458e-06} +{"train_loss": 0.015199677087366581, "grad_norm": 0.13453900814056396, "global_step": 46663, "epoch": 440, "lr": 3.4850379101936336e-06} +{"train_loss": 0.015538256615400314, "grad_norm": 0.07159775495529175, "global_step": 46664, "epoch": 440, "lr": 3.4839508779360265e-06} +{"train_loss": 0.01668757200241089, "grad_norm": 0.07100316882133484, "global_step": 46665, "epoch": 440, "lr": 3.4828640091155827e-06} +{"train_loss": 0.012932734563946724, "grad_norm": 0.07215455174446106, "global_step": 46666, "epoch": 440, "lr": 3.4817773037361447e-06} +{"train_loss": 0.013601861894130707, "grad_norm": 0.07718497514724731, "global_step": 46667, "epoch": 440, "lr": 3.4806907618014916e-06} +{"train_loss": 0.01588217169046402, "grad_norm": 0.1163710281252861, "global_step": 46668, "epoch": 440, "lr": 3.4796043833154656e-06} +{"train_loss": 0.016876105219125748, "grad_norm": 0.1607966125011444, "global_step": 46669, "epoch": 440, "lr": 3.4785181682818804e-06} +{"train_loss": 0.015316318720579147, "grad_norm": 0.09486950933933258, "global_step": 46670, "epoch": 440, "lr": 3.477432116704549e-06} +{"train_loss": 0.014913828112185001, "grad_norm": 0.0836053192615509, "global_step": 46671, "epoch": 440, "lr": 3.476346228587285e-06} +{"train_loss": 0.015334553085267544, "grad_norm": 0.09426102042198181, "global_step": 46672, "epoch": 440, "lr": 3.4752605039339135e-06} +{"train_loss": 0.012844081036746502, "grad_norm": 0.07517117261886597, "global_step": 46673, "epoch": 440, "lr": 3.4741749427482482e-06} +{"train_loss": 0.013559499755501747, "grad_norm": 0.08421576768159866, "global_step": 46674, "epoch": 440, "lr": 3.4730895450340917e-06} +{"train_loss": 0.013163408264517784, "grad_norm": 0.08853957802057266, "global_step": 46675, "epoch": 440, "lr": 3.472004310795257e-06} +{"train_loss": 0.014674879610538483, "grad_norm": 0.09905455261468887, "global_step": 46676, "epoch": 440, "lr": 3.4709192400355805e-06} +{"train_loss": 0.014217587187886238, "grad_norm": 0.06946554780006409, "global_step": 46677, "epoch": 440, "lr": 3.4698343327588477e-06} +{"train_loss": 0.016684096306562424, "grad_norm": 0.0774681568145752, "global_step": 46678, "epoch": 440, "lr": 3.468749588968889e-06} +{"train_loss": 0.015786662697792053, "grad_norm": 0.08833047747612, "global_step": 46679, "epoch": 440, "lr": 3.4676650086695016e-06} +{"train_loss": 0.016503164544701576, "grad_norm": 0.08091765642166138, "global_step": 46680, "epoch": 440, "lr": 3.46658059186451e-06} +{"train_loss": 0.014662767760455608, "grad_norm": 0.0955803319811821, "global_step": 46681, "epoch": 440, "lr": 3.4654963385577112e-06} +{"train_loss": 0.015357332304120064, "grad_norm": 0.06250160187482834, "global_step": 46682, "epoch": 440, "lr": 3.4644122487529185e-06} +{"train_loss": 0.014625167474150658, "grad_norm": 0.07708714157342911, "global_step": 46683, "epoch": 440, "lr": 3.4633283224539626e-06} +{"train_loss": 0.014763496816158295, "grad_norm": 0.07812502235174179, "global_step": 46684, "epoch": 440, "lr": 3.4622445596646126e-06} +{"train_loss": 0.012123214080929756, "grad_norm": 0.06437267363071442, "global_step": 46685, "epoch": 440, "lr": 3.4611609603886986e-06} +{"train_loss": 0.016923587769269943, "grad_norm": 0.1306825876235962, "global_step": 46686, "epoch": 440, "lr": 3.4600775246300345e-06} +{"train_loss": 0.014647082425653934, "grad_norm": 0.06462309509515762, "global_step": 46687, "epoch": 440, "lr": 3.4589942523924115e-06} +{"train_loss": 0.014379139989614487, "grad_norm": 0.05830041319131851, "global_step": 46688, "epoch": 440, "lr": 3.4579111436796375e-06} +{"train_loss": 0.01617659442126751, "grad_norm": 0.08695484697818756, "global_step": 46689, "epoch": 440, "lr": 3.456828198495532e-06} +{"train_loss": 0.014012577943503857, "grad_norm": 0.0899493396282196, "global_step": 46690, "epoch": 440, "lr": 3.4557454168438974e-06} +{"train_loss": 0.013925493694841862, "grad_norm": 0.08893883973360062, "global_step": 46691, "epoch": 440, "lr": 3.4546627987285196e-06} +{"train_loss": 0.01603693887591362, "grad_norm": 0.0770392045378685, "global_step": 46692, "epoch": 440, "lr": 3.4535803441532123e-06} +{"train_loss": 0.01573369838297367, "grad_norm": 0.06268198788166046, "global_step": 46693, "epoch": 440, "lr": 3.4524980531217945e-06} +{"train_loss": 0.017120689153671265, "grad_norm": 0.09219369292259216, "global_step": 46694, "epoch": 440, "lr": 3.451415925638041e-06} +{"train_loss": 0.014355378225445747, "grad_norm": 0.08326496183872223, "global_step": 46695, "epoch": 440, "lr": 3.450333961705782e-06} +{"train_loss": 0.017843620851635933, "grad_norm": 0.10209200531244278, "global_step": 46696, "epoch": 440, "lr": 3.4492521613287976e-06} +{"train_loss": 0.016327213495969772, "grad_norm": 0.08364243805408478, "global_step": 46697, "epoch": 440, "lr": 3.448170524510902e-06} +{"train_loss": 0.0141160748898983, "grad_norm": 0.06682849675416946, "global_step": 46698, "epoch": 440, "lr": 3.447089051255881e-06} +{"train_loss": 0.015061754733324051, "grad_norm": 0.13560223579406738, "global_step": 46699, "epoch": 440, "lr": 3.4460077415675474e-06} +{"train_loss": 0.014178239740431309, "grad_norm": 0.06256517022848129, "global_step": 46700, "epoch": 440, "lr": 3.4449265954497045e-06} +{"train_loss": 0.012744717299938202, "grad_norm": 0.08217527717351913, "global_step": 46701, "epoch": 440, "lr": 3.443845612906144e-06} +{"train_loss": 0.014879126101732254, "grad_norm": 0.07121860235929489, "global_step": 46702, "epoch": 440, "lr": 3.4427647939406504e-06} +{"train_loss": 0.015021422877907753, "grad_norm": 0.08001237362623215, "global_step": 46703, "epoch": 440, "lr": 3.4416841385570443e-06} +{"train_loss": 0.0126766012981534, "grad_norm": 0.06465155631303787, "global_step": 46704, "epoch": 440, "lr": 3.440603646759105e-06} +{"train_loss": 0.015777310356497765, "grad_norm": 0.11880616843700409, "global_step": 46705, "epoch": 440, "lr": 3.4395233185506357e-06} +{"train_loss": 0.013810812495648861, "grad_norm": 0.060367289930582047, "global_step": 46706, "epoch": 440, "lr": 3.4384431539354443e-06} +{"train_loss": 0.015516765415668488, "grad_norm": 0.07293238490819931, "global_step": 46707, "epoch": 440, "lr": 3.4373631529173113e-06} +{"train_loss": 0.015889789909124374, "grad_norm": 0.11014845222234726, "global_step": 46708, "epoch": 440, "lr": 3.4362833155000274e-06} +{"train_loss": 0.013900349847972393, "grad_norm": 0.09919149428606033, "global_step": 46709, "epoch": 440, "lr": 3.4352036416873957e-06} +{"train_loss": 0.014059234410524368, "grad_norm": 0.09035912901163101, "global_step": 46710, "epoch": 440, "lr": 3.434124131483213e-06} +{"train_loss": 0.017558319494128227, "grad_norm": 0.06007295846939087, "global_step": 46711, "epoch": 440, "lr": 3.4330447848912593e-06} +{"train_loss": 0.014660010114312172, "grad_norm": 0.0669265016913414, "global_step": 46712, "epoch": 440, "lr": 3.431965601915338e-06} +{"train_loss": 0.014088793657720089, "grad_norm": 0.10497234761714935, "global_step": 46713, "epoch": 440, "lr": 3.4308865825592394e-06} +{"train_loss": 0.016267025843262672, "grad_norm": 0.08073346316814423, "global_step": 46714, "epoch": 440, "lr": 3.4298077268267614e-06} +{"train_loss": 0.015191472135484219, "grad_norm": 0.13224469125270844, "global_step": 46715, "epoch": 440, "lr": 3.428729034721673e-06} +{"train_loss": 0.015026663430035114, "grad_norm": 0.07614608108997345, "global_step": 46716, "epoch": 440, "lr": 3.427650506247776e-06} +{"train_loss": 0.013537906110286713, "grad_norm": 0.06948088109493256, "global_step": 46717, "epoch": 440, "lr": 3.426572141408874e-06} +{"train_loss": 0.014566983096301556, "grad_norm": 0.0828394964337349, "global_step": 46718, "epoch": 440, "lr": 3.4254939402087294e-06} +{"train_loss": 0.014119520783424377, "grad_norm": 0.20476892590522766, "global_step": 46719, "epoch": 440, "lr": 3.4244159026511566e-06} +{"train_loss": 0.013838013634085655, "grad_norm": 0.0658487007021904, "global_step": 46720, "epoch": 440, "lr": 3.4233380287399254e-06} +{"train_loss": 0.01576114445924759, "grad_norm": 0.065792977809906, "global_step": 46721, "epoch": 440, "lr": 3.4222603184788316e-06} +{"train_loss": 0.014268440194427967, "grad_norm": 0.05707301199436188, "global_step": 46722, "epoch": 440, "lr": 3.4211827718716505e-06} +{"train_loss": 0.01648014411330223, "grad_norm": 0.08712105453014374, "global_step": 46723, "epoch": 440, "lr": 3.420105388922179e-06} +{"train_loss": 0.01613636314868927, "grad_norm": 0.0718667209148407, "global_step": 46724, "epoch": 440, "lr": 3.4190281696342085e-06} +{"train_loss": 0.014446835033595562, "grad_norm": 0.061503276228904724, "global_step": 46725, "epoch": 440, "lr": 3.4179511140115193e-06} +{"train_loss": 0.01435281615704298, "grad_norm": 0.0517231784760952, "global_step": 46726, "epoch": 440, "lr": 3.4168742220578754e-06} +{"train_loss": 0.01435792725533247, "grad_norm": 0.10472060739994049, "global_step": 46727, "epoch": 440, "lr": 3.4157974937770897e-06} +{"train_loss": 0.012178356759250164, "grad_norm": 0.07607048749923706, "global_step": 46728, "epoch": 440, "lr": 3.4147209291729265e-06} +{"train_loss": 0.01621238701045513, "grad_norm": 0.08164774626493454, "global_step": 46729, "epoch": 440, "lr": 3.4136445282491712e-06} +{"train_loss": 0.013400740921497345, "grad_norm": 0.08303695917129517, "global_step": 46730, "epoch": 440, "lr": 3.4125682910096155e-06} +{"train_loss": 0.014415377750992775, "grad_norm": 0.06306192278862, "global_step": 46731, "epoch": 440, "lr": 3.4114922174580397e-06} +{"train_loss": 0.01509823091328144, "grad_norm": 0.08394214510917664, "global_step": 46732, "epoch": 440, "lr": 3.4104163075982077e-06} +{"train_loss": 0.015819896012544632, "grad_norm": 0.06278420239686966, "global_step": 46733, "epoch": 440, "lr": 3.4093405614339157e-06} +{"train_loss": 0.011950600892305374, "grad_norm": 0.07077610492706299, "global_step": 46734, "epoch": 440, "lr": 3.4082649789689446e-06} +{"train_loss": 0.01628803461790085, "grad_norm": 0.09349673986434937, "global_step": 46735, "epoch": 440, "lr": 3.407189560207058e-06} +{"train_loss": 0.016026295721530914, "grad_norm": 0.0664326474070549, "global_step": 46736, "epoch": 440, "lr": 3.4061143051520526e-06} +{"train_loss": 0.01470204722136259, "grad_norm": 0.061930101364851, "global_step": 46737, "epoch": 440, "lr": 3.4050392138076923e-06} +{"train_loss": 0.01442692894488573, "grad_norm": 0.06534293293952942, "global_step": 46738, "epoch": 440, "lr": 3.4039642861777687e-06} +{"train_loss": 0.011863009072840214, "grad_norm": 0.06396529078483582, "global_step": 46739, "epoch": 440, "lr": 3.402889522266045e-06} +{"train_loss": 0.015692396089434624, "grad_norm": 0.155409038066864, "global_step": 46740, "epoch": 440, "lr": 3.401814922076302e-06} +{"train_loss": 0.016867414116859436, "grad_norm": 0.06592853367328644, "global_step": 46741, "epoch": 440, "lr": 3.4007404856123194e-06} +{"train_loss": 0.015742357820272446, "grad_norm": 0.11293447017669678, "global_step": 46742, "epoch": 440, "lr": 3.3996662128778722e-06} +{"train_loss": 0.013974334113299847, "grad_norm": 0.060727089643478394, "global_step": 46743, "epoch": 440, "lr": 3.398592103876724e-06} +{"train_loss": 0.012166336178779602, "grad_norm": 0.04773896932601929, "global_step": 46744, "epoch": 440, "lr": 3.397518158612667e-06} +{"train_loss": 0.014955477340955217, "grad_norm": 0.06754667311906815, "global_step": 46745, "epoch": 440, "lr": 3.3964443770894528e-06, "train_batch_mse_error": 0.013372980058193207, "val_loss": 0.019540470093488693} +{"train_loss": 0.014321082271635532, "grad_norm": 0.06970859318971634, "global_step": 46746, "epoch": 441, "lr": 3.3953707593108675e-06} +{"train_loss": 0.016044659540057182, "grad_norm": 0.07503067702054977, "global_step": 46747, "epoch": 441, "lr": 3.3942973052806916e-06} +{"train_loss": 0.016827601939439774, "grad_norm": 0.06178811565041542, "global_step": 46748, "epoch": 441, "lr": 3.393224015002683e-06} +{"train_loss": 0.01700429618358612, "grad_norm": 0.12835024297237396, "global_step": 46749, "epoch": 441, "lr": 3.3921508884806108e-06} +{"train_loss": 0.014859338290989399, "grad_norm": 0.08754517883062363, "global_step": 46750, "epoch": 441, "lr": 3.39107792571825e-06} +{"train_loss": 0.014236338436603546, "grad_norm": 0.11920931935310364, "global_step": 46751, "epoch": 441, "lr": 3.390005126719381e-06} +{"train_loss": 0.015261521562933922, "grad_norm": 0.0687793642282486, "global_step": 46752, "epoch": 441, "lr": 3.3889324914877564e-06} +{"train_loss": 0.01731269434094429, "grad_norm": 0.08157902956008911, "global_step": 46753, "epoch": 441, "lr": 3.3878600200271504e-06} +{"train_loss": 0.014943217858672142, "grad_norm": 0.09523192793130875, "global_step": 46754, "epoch": 441, "lr": 3.3867877123413384e-06} +{"train_loss": 0.012808150611817837, "grad_norm": 0.05123627185821533, "global_step": 46755, "epoch": 441, "lr": 3.385715568434089e-06} +{"train_loss": 0.01497756689786911, "grad_norm": 0.06592364609241486, "global_step": 46756, "epoch": 441, "lr": 3.38464358830915e-06} +{"train_loss": 0.012987843714654446, "grad_norm": 0.05809970200061798, "global_step": 46757, "epoch": 441, "lr": 3.3835717719703007e-06} +{"train_loss": 0.01572282798588276, "grad_norm": 0.0660293772816658, "global_step": 46758, "epoch": 441, "lr": 3.3825001194213226e-06} +{"train_loss": 0.014490931294858456, "grad_norm": 0.07889214158058167, "global_step": 46759, "epoch": 441, "lr": 3.3814286306659502e-06} +{"train_loss": 0.01426538173109293, "grad_norm": 0.1293388158082962, "global_step": 46760, "epoch": 441, "lr": 3.3803573057079763e-06} +{"train_loss": 0.01322094164788723, "grad_norm": 0.07510057091712952, "global_step": 46761, "epoch": 441, "lr": 3.379286144551147e-06} +{"train_loss": 0.0160367451608181, "grad_norm": 0.06658441573381424, "global_step": 46762, "epoch": 441, "lr": 3.378215147199226e-06} +{"train_loss": 0.012498253956437111, "grad_norm": 0.06369150429964066, "global_step": 46763, "epoch": 441, "lr": 3.3771443136559833e-06} +{"train_loss": 0.013696779496967793, "grad_norm": 0.13302867114543915, "global_step": 46764, "epoch": 441, "lr": 3.3760736439251874e-06} +{"train_loss": 0.014327940531075, "grad_norm": 0.07776566594839096, "global_step": 46765, "epoch": 441, "lr": 3.375003138010585e-06} +{"train_loss": 0.01797950267791748, "grad_norm": 0.06815272569656372, "global_step": 46766, "epoch": 441, "lr": 3.373932795915952e-06} +{"train_loss": 0.01413706224411726, "grad_norm": 0.09115590155124664, "global_step": 46767, "epoch": 441, "lr": 3.372862617645034e-06} +{"train_loss": 0.015630269423127174, "grad_norm": 0.09440618753433228, "global_step": 46768, "epoch": 441, "lr": 3.3717926032016066e-06} +{"train_loss": 0.015904128551483154, "grad_norm": 0.06730633229017258, "global_step": 46769, "epoch": 441, "lr": 3.3707227525894114e-06} +{"train_loss": 0.014691809192299843, "grad_norm": 0.06894160062074661, "global_step": 46770, "epoch": 441, "lr": 3.369653065812223e-06} +{"train_loss": 0.013602804392576218, "grad_norm": 0.08507315814495087, "global_step": 46771, "epoch": 441, "lr": 3.368583542873799e-06} +{"train_loss": 0.01531351637095213, "grad_norm": 0.10294806957244873, "global_step": 46772, "epoch": 441, "lr": 3.367514183777898e-06} +{"train_loss": 0.01649622619152069, "grad_norm": 0.0743790715932846, "global_step": 46773, "epoch": 441, "lr": 3.366444988528261e-06} +{"train_loss": 0.01533563993871212, "grad_norm": 0.13150502741336823, "global_step": 46774, "epoch": 441, "lr": 3.3653759571286524e-06} +{"train_loss": 0.016922637820243835, "grad_norm": 0.14525911211967468, "global_step": 46775, "epoch": 441, "lr": 3.3643070895828467e-06} +{"train_loss": 0.01273046713322401, "grad_norm": 0.10234098881483078, "global_step": 46776, "epoch": 441, "lr": 3.363238385894568e-06} +{"train_loss": 0.01366465725004673, "grad_norm": 0.09491240978240967, "global_step": 46777, "epoch": 441, "lr": 3.3621698460675977e-06} +{"train_loss": 0.014542147517204285, "grad_norm": 0.0674005076289177, "global_step": 46778, "epoch": 441, "lr": 3.361101470105682e-06} +{"train_loss": 0.014617041684687138, "grad_norm": 0.11256549507379532, "global_step": 46779, "epoch": 441, "lr": 3.3600332580125735e-06} +{"train_loss": 0.015839211642742157, "grad_norm": 0.07030482590198517, "global_step": 46780, "epoch": 441, "lr": 3.3589652097920197e-06} +{"train_loss": 0.015748564153909683, "grad_norm": 0.0765969380736351, "global_step": 46781, "epoch": 441, "lr": 3.357897325447773e-06} +{"train_loss": 0.016103340312838554, "grad_norm": 0.059670671820640564, "global_step": 46782, "epoch": 441, "lr": 3.356829604983602e-06} +{"train_loss": 0.013663622550666332, "grad_norm": 0.055039383471012115, "global_step": 46783, "epoch": 441, "lr": 3.355762048403249e-06} +{"train_loss": 0.017889544367790222, "grad_norm": 0.08271462470293045, "global_step": 46784, "epoch": 441, "lr": 3.3546946557104553e-06} +{"train_loss": 0.014275159686803818, "grad_norm": 0.05769050866365433, "global_step": 46785, "epoch": 441, "lr": 3.3536274269089842e-06} +{"train_loss": 0.015245920047163963, "grad_norm": 0.07228036969900131, "global_step": 46786, "epoch": 441, "lr": 3.352560362002577e-06} +{"train_loss": 0.015530925244092941, "grad_norm": 0.0653769001364708, "global_step": 46787, "epoch": 441, "lr": 3.3514934609949812e-06} +{"train_loss": 0.012218017131090164, "grad_norm": 0.07709673792123795, "global_step": 46788, "epoch": 441, "lr": 3.35042672388996e-06} +{"train_loss": 0.016934558749198914, "grad_norm": 0.08900576829910278, "global_step": 46789, "epoch": 441, "lr": 3.3493601506912498e-06} +{"train_loss": 0.015051696449518204, "grad_norm": 0.09648319333791733, "global_step": 46790, "epoch": 441, "lr": 3.348293741402597e-06} +{"train_loss": 0.015470251441001892, "grad_norm": 0.08147112280130386, "global_step": 46791, "epoch": 441, "lr": 3.3472274960277493e-06} +{"train_loss": 0.01607150211930275, "grad_norm": 0.10597258061170578, "global_step": 46792, "epoch": 441, "lr": 3.3461614145704647e-06} +{"train_loss": 0.014372944831848145, "grad_norm": 0.06700237840414047, "global_step": 46793, "epoch": 441, "lr": 3.345095497034467e-06} +{"train_loss": 0.015281366184353828, "grad_norm": 0.07649561762809753, "global_step": 46794, "epoch": 441, "lr": 3.3440297434235212e-06} +{"train_loss": 0.014360547065734863, "grad_norm": 0.054049454629421234, "global_step": 46795, "epoch": 441, "lr": 3.342964153741368e-06} +{"train_loss": 0.013607695698738098, "grad_norm": 0.06597001105546951, "global_step": 46796, "epoch": 441, "lr": 3.3418987279917492e-06} +{"train_loss": 0.014155764132738113, "grad_norm": 0.06633318215608597, "global_step": 46797, "epoch": 441, "lr": 3.3408334661784003e-06} +{"train_loss": 0.014389862306416035, "grad_norm": 0.0994371697306633, "global_step": 46798, "epoch": 441, "lr": 3.3397683683050685e-06} +{"train_loss": 0.014058823697268963, "grad_norm": 0.07528598606586456, "global_step": 46799, "epoch": 441, "lr": 3.3387034343755065e-06} +{"train_loss": 0.013754636980593204, "grad_norm": 0.0803041085600853, "global_step": 46800, "epoch": 441, "lr": 3.3376386643934444e-06} +{"train_loss": 0.013874124735593796, "grad_norm": 0.10219627618789673, "global_step": 46801, "epoch": 441, "lr": 3.336574058362635e-06} +{"train_loss": 0.013603472150862217, "grad_norm": 0.0668724924325943, "global_step": 46802, "epoch": 441, "lr": 3.335509616286808e-06} +{"train_loss": 0.015230967663228512, "grad_norm": 0.1004682406783104, "global_step": 46803, "epoch": 441, "lr": 3.334445338169695e-06} +{"train_loss": 0.013517516665160656, "grad_norm": 0.07622920721769333, "global_step": 46804, "epoch": 441, "lr": 3.333381224015053e-06} +{"train_loss": 0.016041047871112823, "grad_norm": 0.06572464853525162, "global_step": 46805, "epoch": 441, "lr": 3.3323172738266183e-06} +{"train_loss": 0.014113610610365868, "grad_norm": 0.1290651559829712, "global_step": 46806, "epoch": 441, "lr": 3.3312534876081215e-06} +{"train_loss": 0.015351674519479275, "grad_norm": 0.07506837695837021, "global_step": 46807, "epoch": 441, "lr": 3.330189865363309e-06} +{"train_loss": 0.01485553290694952, "grad_norm": 0.06632982939481735, "global_step": 46808, "epoch": 441, "lr": 3.329126407095906e-06} +{"train_loss": 0.013327092863619328, "grad_norm": 0.08818952739238739, "global_step": 46809, "epoch": 441, "lr": 3.3280631128096652e-06} +{"train_loss": 0.015009293332695961, "grad_norm": 0.09837089478969574, "global_step": 46810, "epoch": 441, "lr": 3.3269999825083055e-06} +{"train_loss": 0.01493847742676735, "grad_norm": 0.09639349579811096, "global_step": 46811, "epoch": 441, "lr": 3.3259370161955683e-06} +{"train_loss": 0.014103091321885586, "grad_norm": 0.04946649819612503, "global_step": 46812, "epoch": 441, "lr": 3.324874213875201e-06} +{"train_loss": 0.014228520914912224, "grad_norm": 0.09723155945539474, "global_step": 46813, "epoch": 441, "lr": 3.323811575550928e-06} +{"train_loss": 0.016582153737545013, "grad_norm": 0.07725222408771515, "global_step": 46814, "epoch": 441, "lr": 3.3227491012264688e-06} +{"train_loss": 0.013469073921442032, "grad_norm": 0.09074316173791885, "global_step": 46815, "epoch": 441, "lr": 3.321686790905576e-06} +{"train_loss": 0.015781516209244728, "grad_norm": 0.0639735609292984, "global_step": 46816, "epoch": 441, "lr": 3.32062464459198e-06} +{"train_loss": 0.015418590977787971, "grad_norm": 0.06915154308080673, "global_step": 46817, "epoch": 441, "lr": 3.319562662289399e-06} +{"train_loss": 0.015312498435378075, "grad_norm": 0.10104318708181381, "global_step": 46818, "epoch": 441, "lr": 3.3185008440015765e-06} +{"train_loss": 0.016225852072238922, "grad_norm": 0.0717010572552681, "global_step": 46819, "epoch": 441, "lr": 3.317439189732252e-06} +{"train_loss": 0.01402391865849495, "grad_norm": 0.06217966228723526, "global_step": 46820, "epoch": 441, "lr": 3.3163776994851405e-06} +{"train_loss": 0.015091671608388424, "grad_norm": 0.06901797652244568, "global_step": 46821, "epoch": 441, "lr": 3.3153163732639713e-06} +{"train_loss": 0.017674699425697327, "grad_norm": 0.11201870441436768, "global_step": 46822, "epoch": 441, "lr": 3.3142552110724756e-06} +{"train_loss": 0.01571846194565296, "grad_norm": 0.08308735489845276, "global_step": 46823, "epoch": 441, "lr": 3.313194212914389e-06} +{"train_loss": 0.015460850670933723, "grad_norm": 0.08405209332704544, "global_step": 46824, "epoch": 441, "lr": 3.3121333787934303e-06} +{"train_loss": 0.019432134926319122, "grad_norm": 0.09463156014680862, "global_step": 46825, "epoch": 441, "lr": 3.3110727087133365e-06} +{"train_loss": 0.013204848393797874, "grad_norm": 0.11855240166187286, "global_step": 46826, "epoch": 441, "lr": 3.3100122026778314e-06} +{"train_loss": 0.01407155953347683, "grad_norm": 0.06172751635313034, "global_step": 46827, "epoch": 441, "lr": 3.308951860690629e-06} +{"train_loss": 0.015001673251390457, "grad_norm": 0.0677097961306572, "global_step": 46828, "epoch": 441, "lr": 3.307891682755465e-06} +{"train_loss": 0.01353183388710022, "grad_norm": 0.1276419758796692, "global_step": 46829, "epoch": 441, "lr": 3.3068316688760704e-06} +{"train_loss": 0.017229557037353516, "grad_norm": 0.07946397364139557, "global_step": 46830, "epoch": 441, "lr": 3.305771819056164e-06} +{"train_loss": 0.01405133306980133, "grad_norm": 0.10518208146095276, "global_step": 46831, "epoch": 441, "lr": 3.304712133299459e-06} +{"train_loss": 0.016193963587284088, "grad_norm": 0.17010731995105743, "global_step": 46832, "epoch": 441, "lr": 3.3036526116096867e-06} +{"train_loss": 0.012852245941758156, "grad_norm": 0.09086870402097702, "global_step": 46833, "epoch": 441, "lr": 3.3025932539905825e-06} +{"train_loss": 0.015039998106658459, "grad_norm": 0.08142667263746262, "global_step": 46834, "epoch": 441, "lr": 3.301534060445849e-06} +{"train_loss": 0.017107753083109856, "grad_norm": 0.061389703303575516, "global_step": 46835, "epoch": 441, "lr": 3.3004750309792165e-06} +{"train_loss": 0.01472412422299385, "grad_norm": 0.1329955905675888, "global_step": 46836, "epoch": 441, "lr": 3.2994161655944157e-06} +{"train_loss": 0.011838700622320175, "grad_norm": 0.09300170093774796, "global_step": 46837, "epoch": 441, "lr": 3.298357464295154e-06} +{"train_loss": 0.013770310208201408, "grad_norm": 0.058049507439136505, "global_step": 46838, "epoch": 441, "lr": 3.297298927085146e-06} +{"train_loss": 0.013784052804112434, "grad_norm": 0.0654957965016365, "global_step": 46839, "epoch": 441, "lr": 3.2962405539681217e-06} +{"train_loss": 0.013217194005846977, "grad_norm": 0.08579723536968231, "global_step": 46840, "epoch": 441, "lr": 3.2951823449478004e-06} +{"train_loss": 0.01665569469332695, "grad_norm": 0.08893848210573196, "global_step": 46841, "epoch": 441, "lr": 3.2941243000278953e-06} +{"train_loss": 0.014668222516775131, "grad_norm": 0.07751080393791199, "global_step": 46842, "epoch": 441, "lr": 3.293066419212132e-06} +{"train_loss": 0.015798207372426987, "grad_norm": 0.1164473444223404, "global_step": 46843, "epoch": 441, "lr": 3.292008702504218e-06} +{"train_loss": 0.01416579820215702, "grad_norm": 0.08063473552465439, "global_step": 46844, "epoch": 441, "lr": 3.2909511499078673e-06} +{"train_loss": 0.015557875856757164, "grad_norm": 0.06906318664550781, "global_step": 46845, "epoch": 441, "lr": 3.289893761426799e-06} +{"train_loss": 0.015183242037892342, "grad_norm": 0.06519259512424469, "global_step": 46846, "epoch": 441, "lr": 3.288836537064738e-06} +{"train_loss": 0.018376007676124573, "grad_norm": 0.12456345558166504, "global_step": 46847, "epoch": 441, "lr": 3.2877794768253813e-06} +{"train_loss": 0.014939183369278908, "grad_norm": 0.07435428351163864, "global_step": 46848, "epoch": 441, "lr": 3.286722580712465e-06} +{"train_loss": 0.015199251472949982, "grad_norm": 0.07762369513511658, "global_step": 46849, "epoch": 441, "lr": 3.28566584872968e-06} +{"train_loss": 0.014449920505285263, "grad_norm": 0.08223403245210648, "global_step": 46850, "epoch": 441, "lr": 3.2846092808807627e-06} +{"train_loss": 0.014960017406715537, "grad_norm": 0.11249065399169922, "global_step": 46851, "epoch": 441, "lr": 3.2835528771693992e-06, "val_loss": 0.019265709444880486} +{"train_loss": 0.016189144924283028, "grad_norm": 0.07630744576454163, "global_step": 46852, "epoch": 442, "lr": 3.282496637599314e-06} +{"train_loss": 0.015341145917773247, "grad_norm": 0.10444962233304977, "global_step": 46853, "epoch": 442, "lr": 3.2814405621742317e-06} +{"train_loss": 0.01498479675501585, "grad_norm": 0.09110242128372192, "global_step": 46854, "epoch": 442, "lr": 3.2803846508978443e-06} +{"train_loss": 0.014333406463265419, "grad_norm": 0.05357597395777702, "global_step": 46855, "epoch": 442, "lr": 3.279328903773865e-06} +{"train_loss": 0.01471967063844204, "grad_norm": 0.08580777049064636, "global_step": 46856, "epoch": 442, "lr": 3.2782733208060025e-06} +{"train_loss": 0.011870373971760273, "grad_norm": 0.06647620350122452, "global_step": 46857, "epoch": 442, "lr": 3.2772179019979752e-06} +{"train_loss": 0.013705998659133911, "grad_norm": 0.06940000504255295, "global_step": 46858, "epoch": 442, "lr": 3.276162647353481e-06} +{"train_loss": 0.017009809613227844, "grad_norm": 0.11142035573720932, "global_step": 46859, "epoch": 442, "lr": 3.275107556876228e-06} +{"train_loss": 0.012790627777576447, "grad_norm": 0.11151435226202011, "global_step": 46860, "epoch": 442, "lr": 3.274052630569946e-06} +{"train_loss": 0.016381587833166122, "grad_norm": 0.05720022693276405, "global_step": 46861, "epoch": 442, "lr": 3.2729978684383045e-06} +{"train_loss": 0.015111330896615982, "grad_norm": 0.0935935527086258, "global_step": 46862, "epoch": 442, "lr": 3.2719432704850284e-06} +{"train_loss": 0.015925660729408264, "grad_norm": 0.08584021776914597, "global_step": 46863, "epoch": 442, "lr": 3.2708888367138256e-06} +{"train_loss": 0.015237787738442421, "grad_norm": 0.1477225124835968, "global_step": 46864, "epoch": 442, "lr": 3.269834567128394e-06} +{"train_loss": 0.016376353800296783, "grad_norm": 0.085634745657444, "global_step": 46865, "epoch": 442, "lr": 3.26878046173244e-06} +{"train_loss": 0.014855347573757172, "grad_norm": 0.059483692049980164, "global_step": 46866, "epoch": 442, "lr": 3.267726520529679e-06} +{"train_loss": 0.01347328256815672, "grad_norm": 0.11638887226581573, "global_step": 46867, "epoch": 442, "lr": 3.266672743523802e-06} +{"train_loss": 0.013322131708264351, "grad_norm": 0.08806287497282028, "global_step": 46868, "epoch": 442, "lr": 3.2656191307185057e-06} +{"train_loss": 0.016219858080148697, "grad_norm": 0.07336176931858063, "global_step": 46869, "epoch": 442, "lr": 3.2645656821174986e-06} +{"train_loss": 0.01365726813673973, "grad_norm": 0.08517757058143616, "global_step": 46870, "epoch": 442, "lr": 3.2635123977244886e-06} +{"train_loss": 0.012721952982246876, "grad_norm": 0.06632872670888901, "global_step": 46871, "epoch": 442, "lr": 3.262459277543167e-06} +{"train_loss": 0.015491417609155178, "grad_norm": 0.07263600826263428, "global_step": 46872, "epoch": 442, "lr": 3.2614063215772485e-06} +{"train_loss": 0.013947388157248497, "grad_norm": 0.08815348148345947, "global_step": 46873, "epoch": 442, "lr": 3.2603535298304066e-06} +{"train_loss": 0.01799171045422554, "grad_norm": 0.12211571633815765, "global_step": 46874, "epoch": 442, "lr": 3.259300902306367e-06} +{"train_loss": 0.016227971762418747, "grad_norm": 0.11518553644418716, "global_step": 46875, "epoch": 442, "lr": 3.2582484390088096e-06} +{"train_loss": 0.013931610621511936, "grad_norm": 0.08860094845294952, "global_step": 46876, "epoch": 442, "lr": 3.257196139941443e-06} +{"train_loss": 0.015361061319708824, "grad_norm": 0.06207851320505142, "global_step": 46877, "epoch": 442, "lr": 3.256144005107964e-06} +{"train_loss": 0.014589138329029083, "grad_norm": 0.06307408958673477, "global_step": 46878, "epoch": 442, "lr": 3.2550920345120696e-06} +{"train_loss": 0.01641194522380829, "grad_norm": 0.0925465077161789, "global_step": 46879, "epoch": 442, "lr": 3.25404022815744e-06} +{"train_loss": 0.015523454174399376, "grad_norm": 0.07236508280038834, "global_step": 46880, "epoch": 442, "lr": 3.2529885860477895e-06} +{"train_loss": 0.015826713293790817, "grad_norm": 0.0746506005525589, "global_step": 46881, "epoch": 442, "lr": 3.2519371081868145e-06} +{"train_loss": 0.014173343777656555, "grad_norm": 0.1099979504942894, "global_step": 46882, "epoch": 442, "lr": 3.25088579457819e-06} +{"train_loss": 0.0157890934497118, "grad_norm": 0.06205678731203079, "global_step": 46883, "epoch": 442, "lr": 3.2498346452256356e-06} +{"train_loss": 0.015112257562577724, "grad_norm": 0.0777946189045906, "global_step": 46884, "epoch": 442, "lr": 3.248783660132826e-06} +{"train_loss": 0.01352787110954523, "grad_norm": 0.12343785911798477, "global_step": 46885, "epoch": 442, "lr": 3.2477328393034577e-06} +{"train_loss": 0.015727533027529716, "grad_norm": 0.07708851248025894, "global_step": 46886, "epoch": 442, "lr": 3.2466821827412175e-06} +{"train_loss": 0.015173809602856636, "grad_norm": 0.09490858018398285, "global_step": 46887, "epoch": 442, "lr": 3.2456316904498184e-06} +{"train_loss": 0.014206405729055405, "grad_norm": 0.06277525424957275, "global_step": 46888, "epoch": 442, "lr": 3.2445813624329247e-06} +{"train_loss": 0.013302702456712723, "grad_norm": 0.06508445739746094, "global_step": 46889, "epoch": 442, "lr": 3.243531198694244e-06} +{"train_loss": 0.013152184896171093, "grad_norm": 0.0581781342625618, "global_step": 46890, "epoch": 442, "lr": 3.242481199237457e-06} +{"train_loss": 0.01532596442848444, "grad_norm": 0.08687489479780197, "global_step": 46891, "epoch": 442, "lr": 3.2414313640662607e-06} +{"train_loss": 0.015492560341954231, "grad_norm": 0.10265189409255981, "global_step": 46892, "epoch": 442, "lr": 3.2403816931843354e-06} +{"train_loss": 0.013803001493215561, "grad_norm": 0.05191877856850624, "global_step": 46893, "epoch": 442, "lr": 3.239332186595373e-06} +{"train_loss": 0.0162623543292284, "grad_norm": 0.06836649775505066, "global_step": 46894, "epoch": 442, "lr": 3.23828284430307e-06} +{"train_loss": 0.015206428244709969, "grad_norm": 0.11328523606061935, "global_step": 46895, "epoch": 442, "lr": 3.2372336663111068e-06} +{"train_loss": 0.017150208353996277, "grad_norm": 0.09507612138986588, "global_step": 46896, "epoch": 442, "lr": 3.236184652623153e-06} +{"train_loss": 0.014894891530275345, "grad_norm": 0.07928716391324997, "global_step": 46897, "epoch": 442, "lr": 3.235135803242917e-06} +{"train_loss": 0.014415298588573933, "grad_norm": 0.08583877235651016, "global_step": 46898, "epoch": 442, "lr": 3.2340871181740786e-06} +{"train_loss": 0.016232948750257492, "grad_norm": 0.08458787947893143, "global_step": 46899, "epoch": 442, "lr": 3.233038597420318e-06} +{"train_loss": 0.01635553687810898, "grad_norm": 0.06367877125740051, "global_step": 46900, "epoch": 442, "lr": 3.2319902409853165e-06} +{"train_loss": 0.01555364765226841, "grad_norm": 0.12526334822177887, "global_step": 46901, "epoch": 442, "lr": 3.230942048872776e-06} +{"train_loss": 0.015567934140563011, "grad_norm": 0.08310776948928833, "global_step": 46902, "epoch": 442, "lr": 3.229894021086355e-06} +{"train_loss": 0.016462991014122963, "grad_norm": 0.09910929948091507, "global_step": 46903, "epoch": 442, "lr": 3.228846157629745e-06} +{"train_loss": 0.01513045746833086, "grad_norm": 0.0920657142996788, "global_step": 46904, "epoch": 442, "lr": 3.2277984585066366e-06} +{"train_loss": 0.016018856316804886, "grad_norm": 0.05578669160604477, "global_step": 46905, "epoch": 442, "lr": 3.2267509237207007e-06} +{"train_loss": 0.014315370470285416, "grad_norm": 0.11484608799219131, "global_step": 46906, "epoch": 442, "lr": 3.2257035532756164e-06} +{"train_loss": 0.014836044982075691, "grad_norm": 0.08736017346382141, "global_step": 46907, "epoch": 442, "lr": 3.2246563471750757e-06} +{"train_loss": 0.016467180103063583, "grad_norm": 0.07674740999937057, "global_step": 46908, "epoch": 442, "lr": 3.2236093054227534e-06} +{"train_loss": 0.014237210154533386, "grad_norm": 0.07498335093259811, "global_step": 46909, "epoch": 442, "lr": 3.2225624280223133e-06} +{"train_loss": 0.014219371601939201, "grad_norm": 0.05588715896010399, "global_step": 46910, "epoch": 442, "lr": 3.221515714977447e-06} +{"train_loss": 0.0136937340721488, "grad_norm": 0.07983199506998062, "global_step": 46911, "epoch": 442, "lr": 3.2204691662918398e-06} +{"train_loss": 0.013333207927644253, "grad_norm": 0.06942671537399292, "global_step": 46912, "epoch": 442, "lr": 3.2194227819691556e-06} +{"train_loss": 0.01604273170232773, "grad_norm": 0.10204045474529266, "global_step": 46913, "epoch": 442, "lr": 3.2183765620130814e-06} +{"train_loss": 0.016986194998025894, "grad_norm": 0.09986797720193863, "global_step": 46914, "epoch": 442, "lr": 3.217330506427274e-06} +{"train_loss": 0.01653738133609295, "grad_norm": 0.10576217621564865, "global_step": 46915, "epoch": 442, "lr": 3.2162846152154367e-06} +{"train_loss": 0.01778528094291687, "grad_norm": 0.09729568660259247, "global_step": 46916, "epoch": 442, "lr": 3.2152388883812167e-06} +{"train_loss": 0.01213099341839552, "grad_norm": 0.07066354900598526, "global_step": 46917, "epoch": 442, "lr": 3.2141933259283052e-06} +{"train_loss": 0.01421597134321928, "grad_norm": 0.08501968532800674, "global_step": 46918, "epoch": 442, "lr": 3.213147927860377e-06} +{"train_loss": 0.013259869068861008, "grad_norm": 0.06821316480636597, "global_step": 46919, "epoch": 442, "lr": 3.2121026941811015e-06} +{"train_loss": 0.013622066006064415, "grad_norm": 0.08116913586854935, "global_step": 46920, "epoch": 442, "lr": 3.2110576248941425e-06} +{"train_loss": 0.015089577063918114, "grad_norm": 0.10757849365472794, "global_step": 46921, "epoch": 442, "lr": 3.2100127200031803e-06} +{"train_loss": 0.016700075939297676, "grad_norm": 0.07646314054727554, "global_step": 46922, "epoch": 442, "lr": 3.208967979511884e-06} +{"train_loss": 0.01645985059440136, "grad_norm": 0.08763809502124786, "global_step": 46923, "epoch": 442, "lr": 3.2079234034239236e-06} +{"train_loss": 0.01531622838228941, "grad_norm": 0.07330943644046783, "global_step": 46924, "epoch": 442, "lr": 3.206878991742973e-06} +{"train_loss": 0.014064631424844265, "grad_norm": 0.07894661277532578, "global_step": 46925, "epoch": 442, "lr": 3.205834744472713e-06} +{"train_loss": 0.01369231566786766, "grad_norm": 0.07092539221048355, "global_step": 46926, "epoch": 442, "lr": 3.2047906616167854e-06} +{"train_loss": 0.015237038023769855, "grad_norm": 0.06940746307373047, "global_step": 46927, "epoch": 442, "lr": 3.2037467431788702e-06} +{"train_loss": 0.014429077506065369, "grad_norm": 0.0698404610157013, "global_step": 46928, "epoch": 442, "lr": 3.202702989162648e-06} +{"train_loss": 0.014158878475427628, "grad_norm": 0.07891255617141724, "global_step": 46929, "epoch": 442, "lr": 3.201659399571766e-06} +{"train_loss": 0.017799081280827522, "grad_norm": 0.06949851661920547, "global_step": 46930, "epoch": 442, "lr": 3.2006159744098983e-06} +{"train_loss": 0.012779329903423786, "grad_norm": 0.06870134174823761, "global_step": 46931, "epoch": 442, "lr": 3.1995727136807263e-06} +{"train_loss": 0.015151359140872955, "grad_norm": 0.08497294038534164, "global_step": 46932, "epoch": 442, "lr": 3.198529617387902e-06} +{"train_loss": 0.017228223383426666, "grad_norm": 0.09918873012065887, "global_step": 46933, "epoch": 442, "lr": 3.1974866855350836e-06} +{"train_loss": 0.013765119016170502, "grad_norm": 0.05599721893668175, "global_step": 46934, "epoch": 442, "lr": 3.196443918125941e-06} +{"train_loss": 0.013753922656178474, "grad_norm": 0.07214492559432983, "global_step": 46935, "epoch": 442, "lr": 3.1954013151641484e-06} +{"train_loss": 0.01414639875292778, "grad_norm": 0.061486952006816864, "global_step": 46936, "epoch": 442, "lr": 3.1943588766533637e-06} +{"train_loss": 0.016460087150335312, "grad_norm": 0.09100815653800964, "global_step": 46937, "epoch": 442, "lr": 3.193316602597235e-06} +{"train_loss": 0.014564961194992065, "grad_norm": 0.10907384008169174, "global_step": 46938, "epoch": 442, "lr": 3.192274492999442e-06} +{"train_loss": 0.012250976637005806, "grad_norm": 0.110537089407444, "global_step": 46939, "epoch": 442, "lr": 3.191232547863643e-06} +{"train_loss": 0.01511826366186142, "grad_norm": 0.07988640666007996, "global_step": 46940, "epoch": 442, "lr": 3.1901907671934907e-06} +{"train_loss": 0.015196844935417175, "grad_norm": 0.07467656582593918, "global_step": 46941, "epoch": 442, "lr": 3.1891491509926484e-06} +{"train_loss": 0.013929374516010284, "grad_norm": 0.07306837290525436, "global_step": 46942, "epoch": 442, "lr": 3.188107699264792e-06} +{"train_loss": 0.01641400344669819, "grad_norm": 0.07763834297657013, "global_step": 46943, "epoch": 442, "lr": 3.1870664120135564e-06} +{"train_loss": 0.013521712273359299, "grad_norm": 0.09084898233413696, "global_step": 46944, "epoch": 442, "lr": 3.1860252892426112e-06} +{"train_loss": 0.014783328399062157, "grad_norm": 0.09155459702014923, "global_step": 46945, "epoch": 442, "lr": 3.1849843309556206e-06} +{"train_loss": 0.016157280653715134, "grad_norm": 0.10733643174171448, "global_step": 46946, "epoch": 442, "lr": 3.1839435371562255e-06} +{"train_loss": 0.014206836000084877, "grad_norm": 0.10477413982152939, "global_step": 46947, "epoch": 442, "lr": 3.1829029078480952e-06} +{"train_loss": 0.014058675616979599, "grad_norm": 0.08106852322816849, "global_step": 46948, "epoch": 442, "lr": 3.181862443034894e-06} +{"train_loss": 0.013988017104566097, "grad_norm": 0.06738535314798355, "global_step": 46949, "epoch": 442, "lr": 3.180822142720263e-06} +{"train_loss": 0.014374900609254837, "grad_norm": 0.06941613554954529, "global_step": 46950, "epoch": 442, "lr": 3.1797820069078556e-06} +{"train_loss": 0.016457322984933853, "grad_norm": 0.08193647861480713, "global_step": 46951, "epoch": 442, "lr": 3.1787420356013344e-06} +{"train_loss": 0.01454838179051876, "grad_norm": 0.0725288912653923, "global_step": 46952, "epoch": 442, "lr": 3.177702228804358e-06} +{"train_loss": 0.015917010605335236, "grad_norm": 0.0833306610584259, "global_step": 46953, "epoch": 442, "lr": 3.176662586520568e-06} +{"train_loss": 0.014760213904082775, "grad_norm": 0.1696445494890213, "global_step": 46954, "epoch": 442, "lr": 3.175623108753628e-06} +{"train_loss": 0.012398878112435341, "grad_norm": 0.05821623653173447, "global_step": 46955, "epoch": 442, "lr": 3.17458379550718e-06} +{"train_loss": 0.0173720121383667, "grad_norm": 0.0790051594376564, "global_step": 46956, "epoch": 442, "lr": 3.173544646784887e-06} +{"train_loss": 0.01495774955717179, "grad_norm": 0.12463957816362381, "global_step": 46957, "epoch": 442, "lr": 3.172505662590386e-06, "val_loss": 0.020054833963513374} +{"train_loss": 0.01701226457953453, "grad_norm": 0.1205776184797287, "global_step": 46958, "epoch": 443, "lr": 3.171466842927334e-06} +{"train_loss": 0.016239549964666367, "grad_norm": 0.07509271055459976, "global_step": 46959, "epoch": 443, "lr": 3.1704281877993903e-06} +{"train_loss": 0.015586980618536472, "grad_norm": 0.08199544996023178, "global_step": 46960, "epoch": 443, "lr": 3.1693896972101953e-06} +{"train_loss": 0.01621837541460991, "grad_norm": 0.07650910317897797, "global_step": 46961, "epoch": 443, "lr": 3.1683513711633917e-06} +{"train_loss": 0.01518956758081913, "grad_norm": 0.07885462045669556, "global_step": 46962, "epoch": 443, "lr": 3.167313209662648e-06} +{"train_loss": 0.017788801342248917, "grad_norm": 0.09152467548847198, "global_step": 46963, "epoch": 443, "lr": 3.1662752127115834e-06} +{"train_loss": 0.013143425807356834, "grad_norm": 0.06319186836481094, "global_step": 46964, "epoch": 443, "lr": 3.1652373803138624e-06} +{"train_loss": 0.01463403832167387, "grad_norm": 0.08962569385766983, "global_step": 46965, "epoch": 443, "lr": 3.164199712473137e-06} +{"train_loss": 0.013809866271913052, "grad_norm": 0.08559567481279373, "global_step": 46966, "epoch": 443, "lr": 3.163162209193049e-06} +{"train_loss": 0.01637205295264721, "grad_norm": 0.14041151106357574, "global_step": 46967, "epoch": 443, "lr": 3.162124870477229e-06} +{"train_loss": 0.012876171618700027, "grad_norm": 0.09560109674930573, "global_step": 46968, "epoch": 443, "lr": 3.1610876963293344e-06} +{"train_loss": 0.014808819629251957, "grad_norm": 0.07184844464063644, "global_step": 46969, "epoch": 443, "lr": 3.1600506867530132e-06} +{"train_loss": 0.013579877093434334, "grad_norm": 0.06381794065237045, "global_step": 46970, "epoch": 443, "lr": 3.1590138417519013e-06} +{"train_loss": 0.014444427564740181, "grad_norm": 0.05852118879556656, "global_step": 46971, "epoch": 443, "lr": 3.1579771613296395e-06} +{"train_loss": 0.01573915220797062, "grad_norm": 0.08706555515527725, "global_step": 46972, "epoch": 443, "lr": 3.1569406454898863e-06} +{"train_loss": 0.014088686555624008, "grad_norm": 0.07453061640262604, "global_step": 46973, "epoch": 443, "lr": 3.155904294236267e-06} +{"train_loss": 0.015960441902279854, "grad_norm": 0.09554151445627213, "global_step": 46974, "epoch": 443, "lr": 3.154868107572423e-06} +{"train_loss": 0.013622457161545753, "grad_norm": 0.12415987253189087, "global_step": 46975, "epoch": 443, "lr": 3.153832085502001e-06} +{"train_loss": 0.01611662283539772, "grad_norm": 0.08945249766111374, "global_step": 46976, "epoch": 443, "lr": 3.1527962280286483e-06} +{"train_loss": 0.01544232852756977, "grad_norm": 0.07157319039106369, "global_step": 46977, "epoch": 443, "lr": 3.1517605351559897e-06} +{"train_loss": 0.016994832083582878, "grad_norm": 0.10447879135608673, "global_step": 46978, "epoch": 443, "lr": 3.1507250068876728e-06} +{"train_loss": 0.015728428959846497, "grad_norm": 0.07481902837753296, "global_step": 46979, "epoch": 443, "lr": 3.1496896432273325e-06} +{"train_loss": 0.013821814209222794, "grad_norm": 0.1112266331911087, "global_step": 46980, "epoch": 443, "lr": 3.1486544441786112e-06} +{"train_loss": 0.01454872451722622, "grad_norm": 0.08735864609479904, "global_step": 46981, "epoch": 443, "lr": 3.1476194097451393e-06} +{"train_loss": 0.014364211820065975, "grad_norm": 0.06648427993059158, "global_step": 46982, "epoch": 443, "lr": 3.1465845399305525e-06} +{"train_loss": 0.013841869309544563, "grad_norm": 0.056677043437957764, "global_step": 46983, "epoch": 443, "lr": 3.1455498347385093e-06} +{"train_loss": 0.015549324452877045, "grad_norm": 0.056349195539951324, "global_step": 46984, "epoch": 443, "lr": 3.144515294172612e-06} +{"train_loss": 0.013767393305897713, "grad_norm": 0.10644838958978653, "global_step": 46985, "epoch": 443, "lr": 3.1434809182365133e-06} +{"train_loss": 0.016079308465123177, "grad_norm": 0.08485855907201767, "global_step": 46986, "epoch": 443, "lr": 3.142446706933849e-06} +{"train_loss": 0.01510695368051529, "grad_norm": 0.05922972038388252, "global_step": 46987, "epoch": 443, "lr": 3.1414126602682446e-06} +{"train_loss": 0.015391077846288681, "grad_norm": 0.07113105058670044, "global_step": 46988, "epoch": 443, "lr": 3.1403787782433415e-06} +{"train_loss": 0.012985923327505589, "grad_norm": 0.07892701029777527, "global_step": 46989, "epoch": 443, "lr": 3.1393450608627693e-06} +{"train_loss": 0.014300726354122162, "grad_norm": 0.06930845230817795, "global_step": 46990, "epoch": 443, "lr": 3.1383115081301596e-06} +{"train_loss": 0.016811490058898926, "grad_norm": 0.09544672816991806, "global_step": 46991, "epoch": 443, "lr": 3.1372781200491417e-06} +{"train_loss": 0.016674116253852844, "grad_norm": 0.09180402755737305, "global_step": 46992, "epoch": 443, "lr": 3.1362448966233417e-06} +{"train_loss": 0.016541101038455963, "grad_norm": 0.08426835387945175, "global_step": 46993, "epoch": 443, "lr": 3.135211837856411e-06} +{"train_loss": 0.015422344207763672, "grad_norm": 0.08775003999471664, "global_step": 46994, "epoch": 443, "lr": 3.134178943751953e-06} +{"train_loss": 0.01614474505186081, "grad_norm": 0.06077065318822861, "global_step": 46995, "epoch": 443, "lr": 3.13314621431362e-06} +{"train_loss": 0.01703479513525963, "grad_norm": 0.10196184366941452, "global_step": 46996, "epoch": 443, "lr": 3.132113649545015e-06} +{"train_loss": 0.016841931268572807, "grad_norm": 0.09239082783460617, "global_step": 46997, "epoch": 443, "lr": 3.1310812494497966e-06} +{"train_loss": 0.013882096856832504, "grad_norm": 0.0580558106303215, "global_step": 46998, "epoch": 443, "lr": 3.1300490140315607e-06} +{"train_loss": 0.014206591993570328, "grad_norm": 0.10648797452449799, "global_step": 46999, "epoch": 443, "lr": 3.1290169432939553e-06} +{"train_loss": 0.013059288263320923, "grad_norm": 0.09897708892822266, "global_step": 47000, "epoch": 443, "lr": 3.127985037240605e-06} +{"train_loss": 0.012885872274637222, "grad_norm": 0.06174987554550171, "global_step": 47001, "epoch": 443, "lr": 3.126953295875129e-06} +{"train_loss": 0.013978364877402782, "grad_norm": 0.10103359818458557, "global_step": 47002, "epoch": 443, "lr": 3.1259217192011527e-06} +{"train_loss": 0.012832575477659702, "grad_norm": 0.09860304743051529, "global_step": 47003, "epoch": 443, "lr": 3.1248903072223055e-06} +{"train_loss": 0.014078814536333084, "grad_norm": 0.09338286519050598, "global_step": 47004, "epoch": 443, "lr": 3.1238590599422023e-06} +{"train_loss": 0.015643268823623657, "grad_norm": 0.08230739831924438, "global_step": 47005, "epoch": 443, "lr": 3.122827977364473e-06} +{"train_loss": 0.013602630235254765, "grad_norm": 0.10783953219652176, "global_step": 47006, "epoch": 443, "lr": 3.1217970594927427e-06} +{"train_loss": 0.014783398248255253, "grad_norm": 0.07857911288738251, "global_step": 47007, "epoch": 443, "lr": 3.1207663063306357e-06} +{"train_loss": 0.014488368295133114, "grad_norm": 0.12619638442993164, "global_step": 47008, "epoch": 443, "lr": 3.119735717881761e-06} +{"train_loss": 0.012788421474397182, "grad_norm": 0.07322666794061661, "global_step": 47009, "epoch": 443, "lr": 3.1187052941497485e-06} +{"train_loss": 0.01610873080790043, "grad_norm": 0.07477652281522751, "global_step": 47010, "epoch": 443, "lr": 3.1176750351382235e-06} +{"train_loss": 0.014653547666966915, "grad_norm": 0.06957651674747467, "global_step": 47011, "epoch": 443, "lr": 3.116644940850788e-06} +{"train_loss": 0.01678282767534256, "grad_norm": 0.0733426883816719, "global_step": 47012, "epoch": 443, "lr": 3.115615011291079e-06} +{"train_loss": 0.013827359303832054, "grad_norm": 0.08781833946704865, "global_step": 47013, "epoch": 443, "lr": 3.1145852464627145e-06} +{"train_loss": 0.014780400320887566, "grad_norm": 0.0789245143532753, "global_step": 47014, "epoch": 443, "lr": 3.1135556463693093e-06} +{"train_loss": 0.014995193108916283, "grad_norm": 0.1311093270778656, "global_step": 47015, "epoch": 443, "lr": 3.1125262110144713e-06} +{"train_loss": 0.014891471713781357, "grad_norm": 0.09952668100595474, "global_step": 47016, "epoch": 443, "lr": 3.1114969404018203e-06} +{"train_loss": 0.01586504839360714, "grad_norm": 0.08040730655193329, "global_step": 47017, "epoch": 443, "lr": 3.110467834534991e-06} +{"train_loss": 0.018366456031799316, "grad_norm": 0.09580845385789871, "global_step": 47018, "epoch": 443, "lr": 3.109438893417577e-06} +{"train_loss": 0.01656327024102211, "grad_norm": 0.12925107777118683, "global_step": 47019, "epoch": 443, "lr": 3.1084101170532122e-06} +{"train_loss": 0.015241887420415878, "grad_norm": 0.11536943167448044, "global_step": 47020, "epoch": 443, "lr": 3.1073815054455e-06} +{"train_loss": 0.014923881739377975, "grad_norm": 0.06902172416448593, "global_step": 47021, "epoch": 443, "lr": 3.106353058598044e-06} +{"train_loss": 0.015899483114480972, "grad_norm": 0.06979361921548843, "global_step": 47022, "epoch": 443, "lr": 3.1053247765144732e-06} +{"train_loss": 0.01488301157951355, "grad_norm": 0.08095870167016983, "global_step": 47023, "epoch": 443, "lr": 3.104296659198408e-06} +{"train_loss": 0.014079129323363304, "grad_norm": 0.08938144147396088, "global_step": 47024, "epoch": 443, "lr": 3.1032687066534395e-06} +{"train_loss": 0.014071803539991379, "grad_norm": 0.09411397576332092, "global_step": 47025, "epoch": 443, "lr": 3.102240918883198e-06} +{"train_loss": 0.014312506653368473, "grad_norm": 0.1290634423494339, "global_step": 47026, "epoch": 443, "lr": 3.1012132958912808e-06} +{"train_loss": 0.015061451122164726, "grad_norm": 0.13328471779823303, "global_step": 47027, "epoch": 443, "lr": 3.1001858376813077e-06} +{"train_loss": 0.015008749440312386, "grad_norm": 0.06638356298208237, "global_step": 47028, "epoch": 443, "lr": 3.0991585442568804e-06} +{"train_loss": 0.016015954315662384, "grad_norm": 0.07022246718406677, "global_step": 47029, "epoch": 443, "lr": 3.0981314156216135e-06} +{"train_loss": 0.01789417676627636, "grad_norm": 0.0659317672252655, "global_step": 47030, "epoch": 443, "lr": 3.0971044517791205e-06} +{"train_loss": 0.016172269359230995, "grad_norm": 0.06663087755441666, "global_step": 47031, "epoch": 443, "lr": 3.0960776527330093e-06} +{"train_loss": 0.01477070152759552, "grad_norm": 0.16275547444820404, "global_step": 47032, "epoch": 443, "lr": 3.095051018486872e-06} +{"train_loss": 0.014459624886512756, "grad_norm": 0.07536523044109344, "global_step": 47033, "epoch": 443, "lr": 3.0940245490443277e-06} +{"train_loss": 0.015995733439922333, "grad_norm": 0.1643257886171341, "global_step": 47034, "epoch": 443, "lr": 3.09299824440899e-06} +{"train_loss": 0.015206573531031609, "grad_norm": 0.0728657990694046, "global_step": 47035, "epoch": 443, "lr": 3.091972104584451e-06} +{"train_loss": 0.01433348935097456, "grad_norm": 0.07453550398349762, "global_step": 47036, "epoch": 443, "lr": 3.0909461295743293e-06} +{"train_loss": 0.013791926205158234, "grad_norm": 0.0667370930314064, "global_step": 47037, "epoch": 443, "lr": 3.089920319382217e-06} +{"train_loss": 0.014168353751301765, "grad_norm": 0.06942200660705566, "global_step": 47038, "epoch": 443, "lr": 3.0888946740117273e-06} +{"train_loss": 0.014251681044697762, "grad_norm": 0.10544001311063766, "global_step": 47039, "epoch": 443, "lr": 3.087869193466458e-06} +{"train_loss": 0.013028543442487717, "grad_norm": 0.06386896222829819, "global_step": 47040, "epoch": 443, "lr": 3.0868438777500117e-06} +{"train_loss": 0.01638714224100113, "grad_norm": 0.10030277818441391, "global_step": 47041, "epoch": 443, "lr": 3.085818726866002e-06} +{"train_loss": 0.013455402106046677, "grad_norm": 0.06277113407850266, "global_step": 47042, "epoch": 443, "lr": 3.0847937408180206e-06} +{"train_loss": 0.012321436777710915, "grad_norm": 0.06813089549541473, "global_step": 47043, "epoch": 443, "lr": 3.0837689196096643e-06} +{"train_loss": 0.012192449532449245, "grad_norm": 0.13723984360694885, "global_step": 47044, "epoch": 443, "lr": 3.0827442632445523e-06} +{"train_loss": 0.014318384230136871, "grad_norm": 0.08748146146535873, "global_step": 47045, "epoch": 443, "lr": 3.0817197717262603e-06} +{"train_loss": 0.015815194696187973, "grad_norm": 0.0785992443561554, "global_step": 47046, "epoch": 443, "lr": 3.080695445058407e-06} +{"train_loss": 0.015170040540397167, "grad_norm": 0.06597046554088593, "global_step": 47047, "epoch": 443, "lr": 3.0796712832445897e-06} +{"train_loss": 0.014859628863632679, "grad_norm": 0.05918571725487709, "global_step": 47048, "epoch": 443, "lr": 3.0786472862883997e-06} +{"train_loss": 0.015385681763291359, "grad_norm": 0.05961748585104942, "global_step": 47049, "epoch": 443, "lr": 3.0776234541934345e-06} +{"train_loss": 0.01582484133541584, "grad_norm": 0.07253957539796829, "global_step": 47050, "epoch": 443, "lr": 3.0765997869632914e-06} +{"train_loss": 0.015215186402201653, "grad_norm": 0.11123283207416534, "global_step": 47051, "epoch": 443, "lr": 3.075576284601578e-06} +{"train_loss": 0.014459344558417797, "grad_norm": 0.1386251449584961, "global_step": 47052, "epoch": 443, "lr": 3.074552947111875e-06} +{"train_loss": 0.015224218368530273, "grad_norm": 0.061273690313100815, "global_step": 47053, "epoch": 443, "lr": 3.0735297744977853e-06} +{"train_loss": 0.016256188973784447, "grad_norm": 0.12518420815467834, "global_step": 47054, "epoch": 443, "lr": 3.0725067667629116e-06} +{"train_loss": 0.014225094579160213, "grad_norm": 0.07465075701475143, "global_step": 47055, "epoch": 443, "lr": 3.071483923910845e-06} +{"train_loss": 0.016540680080652237, "grad_norm": 0.16255779564380646, "global_step": 47056, "epoch": 443, "lr": 3.0704612459451664e-06} +{"train_loss": 0.014680958352982998, "grad_norm": 0.09449170529842377, "global_step": 47057, "epoch": 443, "lr": 3.069438732869473e-06} +{"train_loss": 0.013431981205940247, "grad_norm": 0.06666763126850128, "global_step": 47058, "epoch": 443, "lr": 3.0684163846873726e-06} +{"train_loss": 0.015439499169588089, "grad_norm": 0.06703019142150879, "global_step": 47059, "epoch": 443, "lr": 3.0673942014024403e-06} +{"train_loss": 0.014379028230905533, "grad_norm": 0.12645509839057922, "global_step": 47060, "epoch": 443, "lr": 3.0663721830182845e-06} +{"train_loss": 0.013868005946278572, "grad_norm": 0.06269575655460358, "global_step": 47061, "epoch": 443, "lr": 3.06535032953848e-06} +{"train_loss": 0.0137605220079422, "grad_norm": 0.08614776283502579, "global_step": 47062, "epoch": 443, "lr": 3.0643286409666185e-06} +{"train_loss": 0.014960720845199418, "grad_norm": 0.12573619186878204, "global_step": 47063, "epoch": 443, "lr": 3.0633071173062967e-06, "val_loss": 0.019920293241739273} +{"train_loss": 0.015923351049423218, "grad_norm": 0.07240660488605499, "global_step": 47064, "epoch": 444, "lr": 3.0622857585611065e-06} +{"train_loss": 0.013440623879432678, "grad_norm": 0.08643029630184174, "global_step": 47065, "epoch": 444, "lr": 3.061264564734623e-06} +{"train_loss": 0.016690272837877274, "grad_norm": 0.07601744681596756, "global_step": 47066, "epoch": 444, "lr": 3.0602435358304535e-06} +{"train_loss": 0.01435938011854887, "grad_norm": 0.06907866895198822, "global_step": 47067, "epoch": 444, "lr": 3.0592226718521635e-06} +{"train_loss": 0.017115367576479912, "grad_norm": 0.07758829742670059, "global_step": 47068, "epoch": 444, "lr": 3.0582019728033597e-06} +{"train_loss": 0.01577356643974781, "grad_norm": 0.07800980657339096, "global_step": 47069, "epoch": 444, "lr": 3.057181438687612e-06} +{"train_loss": 0.016556397080421448, "grad_norm": 0.07731412351131439, "global_step": 47070, "epoch": 444, "lr": 3.0561610695085173e-06} +{"train_loss": 0.015867169946432114, "grad_norm": 0.07851342856884003, "global_step": 47071, "epoch": 444, "lr": 3.0551408652696623e-06} +{"train_loss": 0.01173336710780859, "grad_norm": 0.07861556112766266, "global_step": 47072, "epoch": 444, "lr": 3.0541208259746213e-06} +{"train_loss": 0.014630882069468498, "grad_norm": 0.06588525325059891, "global_step": 47073, "epoch": 444, "lr": 3.0531009516269803e-06} +{"train_loss": 0.01543436013162136, "grad_norm": 0.07454448938369751, "global_step": 47074, "epoch": 444, "lr": 3.0520812422303314e-06} +{"train_loss": 0.016297457739710808, "grad_norm": 0.06367959827184677, "global_step": 47075, "epoch": 444, "lr": 3.0510616977882545e-06} +{"train_loss": 0.014629121869802475, "grad_norm": 0.11067914217710495, "global_step": 47076, "epoch": 444, "lr": 3.050042318304319e-06} +{"train_loss": 0.014617716893553734, "grad_norm": 0.06651929765939713, "global_step": 47077, "epoch": 444, "lr": 3.049023103782117e-06} +{"train_loss": 0.013669627718627453, "grad_norm": 0.07991555333137512, "global_step": 47078, "epoch": 444, "lr": 3.048004054225251e-06} +{"train_loss": 0.013740934431552887, "grad_norm": 0.0756087526679039, "global_step": 47079, "epoch": 444, "lr": 3.0469851696372564e-06} +{"train_loss": 0.015323193743824959, "grad_norm": 0.11583195626735687, "global_step": 47080, "epoch": 444, "lr": 3.0459664500217423e-06} +{"train_loss": 0.01391393318772316, "grad_norm": 0.05598375201225281, "global_step": 47081, "epoch": 444, "lr": 3.044947895382283e-06} +{"train_loss": 0.016265925019979477, "grad_norm": 0.07582863420248032, "global_step": 47082, "epoch": 444, "lr": 3.0439295057224594e-06} +{"train_loss": 0.014649579301476479, "grad_norm": 0.057312510907649994, "global_step": 47083, "epoch": 444, "lr": 3.0429112810458514e-06} +{"train_loss": 0.012931068427860737, "grad_norm": 0.06401680409908295, "global_step": 47084, "epoch": 444, "lr": 3.0418932213560235e-06} +{"train_loss": 0.013193320482969284, "grad_norm": 0.06604282557964325, "global_step": 47085, "epoch": 444, "lr": 3.040875326656567e-06} +{"train_loss": 0.013368912972509861, "grad_norm": 0.1573486626148224, "global_step": 47086, "epoch": 444, "lr": 3.039857596951046e-06} +{"train_loss": 0.013688694685697556, "grad_norm": 0.07386364042758942, "global_step": 47087, "epoch": 444, "lr": 3.03884003224304e-06} +{"train_loss": 0.014680672436952591, "grad_norm": 0.07686135172843933, "global_step": 47088, "epoch": 444, "lr": 3.0378226325361412e-06} +{"train_loss": 0.01631689816713333, "grad_norm": 0.0971507802605629, "global_step": 47089, "epoch": 444, "lr": 3.0368053978339085e-06} +{"train_loss": 0.014949455857276917, "grad_norm": 0.0655774250626564, "global_step": 47090, "epoch": 444, "lr": 3.0357883281399047e-06} +{"train_loss": 0.015055352821946144, "grad_norm": 0.07040384411811829, "global_step": 47091, "epoch": 444, "lr": 3.0347714234577217e-06} +{"train_loss": 0.014146954752504826, "grad_norm": 0.09776497632265091, "global_step": 47092, "epoch": 444, "lr": 3.033754683790935e-06} +{"train_loss": 0.016403980553150177, "grad_norm": 0.09286721795797348, "global_step": 47093, "epoch": 444, "lr": 3.032738109143102e-06} +{"train_loss": 0.014025012031197548, "grad_norm": 0.06190796196460724, "global_step": 47094, "epoch": 444, "lr": 3.0317216995177978e-06} +{"train_loss": 0.014129741117358208, "grad_norm": 0.06635966151952744, "global_step": 47095, "epoch": 444, "lr": 3.0307054549186086e-06} +{"train_loss": 0.01603679731488228, "grad_norm": 0.0939822569489479, "global_step": 47096, "epoch": 444, "lr": 3.029689375349093e-06} +{"train_loss": 0.013765955343842506, "grad_norm": 0.06363095343112946, "global_step": 47097, "epoch": 444, "lr": 3.0286734608128143e-06} +{"train_loss": 0.013462260365486145, "grad_norm": 0.07554923743009567, "global_step": 47098, "epoch": 444, "lr": 3.0276577113133474e-06} +{"train_loss": 0.013887962326407433, "grad_norm": 0.06191300228238106, "global_step": 47099, "epoch": 444, "lr": 3.0266421268542735e-06} +{"train_loss": 0.014076704159379005, "grad_norm": 0.13414236903190613, "global_step": 47100, "epoch": 444, "lr": 3.025626707439144e-06} +{"train_loss": 0.01565793715417385, "grad_norm": 0.06924400478601456, "global_step": 47101, "epoch": 444, "lr": 3.024611453071541e-06} +{"train_loss": 0.015584235079586506, "grad_norm": 0.19012148678302765, "global_step": 47102, "epoch": 444, "lr": 3.0235963637550267e-06} +{"train_loss": 0.014351706951856613, "grad_norm": 0.07741183042526245, "global_step": 47103, "epoch": 444, "lr": 3.022581439493155e-06} +{"train_loss": 0.013317390345036983, "grad_norm": 0.10623041540384293, "global_step": 47104, "epoch": 444, "lr": 3.0215666802895005e-06} +{"train_loss": 0.016546912491321564, "grad_norm": 0.06523056328296661, "global_step": 47105, "epoch": 444, "lr": 3.020552086147638e-06} +{"train_loss": 0.0178853627294302, "grad_norm": 0.10428718477487564, "global_step": 47106, "epoch": 444, "lr": 3.0195376570711196e-06} +{"train_loss": 0.01567007228732109, "grad_norm": 0.07529241591691971, "global_step": 47107, "epoch": 444, "lr": 3.0185233930635216e-06} +{"train_loss": 0.014766319654881954, "grad_norm": 0.09408965706825256, "global_step": 47108, "epoch": 444, "lr": 3.0175092941283957e-06} +{"train_loss": 0.01332276500761509, "grad_norm": 0.14302296936511993, "global_step": 47109, "epoch": 444, "lr": 3.0164953602693115e-06} +{"train_loss": 0.01554703339934349, "grad_norm": 0.0557992085814476, "global_step": 47110, "epoch": 444, "lr": 3.0154815914898273e-06} +{"train_loss": 0.015812702476978302, "grad_norm": 0.06210794299840927, "global_step": 47111, "epoch": 444, "lr": 3.0144679877935065e-06} +{"train_loss": 0.013966286554932594, "grad_norm": 0.09542705863714218, "global_step": 47112, "epoch": 444, "lr": 3.0134545491839195e-06} +{"train_loss": 0.0164458267390728, "grad_norm": 0.08361513167619705, "global_step": 47113, "epoch": 444, "lr": 3.012441275664618e-06} +{"train_loss": 0.018970146775245667, "grad_norm": 0.12898032367229462, "global_step": 47114, "epoch": 444, "lr": 3.0114281672391554e-06} +{"train_loss": 0.013150610029697418, "grad_norm": 0.07504136860370636, "global_step": 47115, "epoch": 444, "lr": 3.0104152239111007e-06} +{"train_loss": 0.013397686183452606, "grad_norm": 0.0756443440914154, "global_step": 47116, "epoch": 444, "lr": 3.0094024456840174e-06} +{"train_loss": 0.014329429715871811, "grad_norm": 0.08385875821113586, "global_step": 47117, "epoch": 444, "lr": 3.0083898325614535e-06} +{"train_loss": 0.01316996943205595, "grad_norm": 0.08012552559375763, "global_step": 47118, "epoch": 444, "lr": 3.0073773845469723e-06} +{"train_loss": 0.015564724802970886, "grad_norm": 0.06735056638717651, "global_step": 47119, "epoch": 444, "lr": 3.0063651016441428e-06} +{"train_loss": 0.015475424006581306, "grad_norm": 0.14116108417510986, "global_step": 47120, "epoch": 444, "lr": 3.0053529838564963e-06} +{"train_loss": 0.016575859859585762, "grad_norm": 0.09164424240589142, "global_step": 47121, "epoch": 444, "lr": 3.0043410311876018e-06} +{"train_loss": 0.014317900873720646, "grad_norm": 0.06615940481424332, "global_step": 47122, "epoch": 444, "lr": 3.0033292436410176e-06} +{"train_loss": 0.015860125422477722, "grad_norm": 0.07778092473745346, "global_step": 47123, "epoch": 444, "lr": 3.0023176212202962e-06} +{"train_loss": 0.015529615804553032, "grad_norm": 0.11050110310316086, "global_step": 47124, "epoch": 444, "lr": 3.001306163928985e-06} +{"train_loss": 0.013869097456336021, "grad_norm": 0.05582907423377037, "global_step": 47125, "epoch": 444, "lr": 3.000294871770659e-06} +{"train_loss": 0.015849895775318146, "grad_norm": 0.1039520651102066, "global_step": 47126, "epoch": 444, "lr": 2.9992837447488543e-06} +{"train_loss": 0.014739764854311943, "grad_norm": 0.06115918606519699, "global_step": 47127, "epoch": 444, "lr": 2.9982727828671177e-06} +{"train_loss": 0.0141211599111557, "grad_norm": 0.053983911871910095, "global_step": 47128, "epoch": 444, "lr": 2.997261986129013e-06} +{"train_loss": 0.016480794176459312, "grad_norm": 0.067771777510643, "global_step": 47129, "epoch": 444, "lr": 2.9962513545380933e-06} +{"train_loss": 0.013771843165159225, "grad_norm": 0.09632561355829239, "global_step": 47130, "epoch": 444, "lr": 2.9952408880978944e-06} +{"train_loss": 0.014526879414916039, "grad_norm": 0.10338038206100464, "global_step": 47131, "epoch": 444, "lr": 2.9942305868119915e-06} +{"train_loss": 0.01401202566921711, "grad_norm": 0.09492892771959305, "global_step": 47132, "epoch": 444, "lr": 2.993220450683909e-06} +{"train_loss": 0.014714697375893593, "grad_norm": 0.1337183117866516, "global_step": 47133, "epoch": 444, "lr": 2.9922104797172166e-06} +{"train_loss": 0.014877310022711754, "grad_norm": 0.16114889085292816, "global_step": 47134, "epoch": 444, "lr": 2.9912006739154453e-06} +{"train_loss": 0.015653207898139954, "grad_norm": 0.07962771505117416, "global_step": 47135, "epoch": 444, "lr": 2.990191033282147e-06} +{"train_loss": 0.017255090177059174, "grad_norm": 0.13443027436733246, "global_step": 47136, "epoch": 444, "lr": 2.989181557820886e-06} +{"train_loss": 0.013495991006493568, "grad_norm": 0.06265497207641602, "global_step": 47137, "epoch": 444, "lr": 2.9881722475351925e-06} +{"train_loss": 0.015889689326286316, "grad_norm": 0.08489682525396347, "global_step": 47138, "epoch": 444, "lr": 2.987163102428614e-06} +{"train_loss": 0.01467050053179264, "grad_norm": 0.06604879349470139, "global_step": 47139, "epoch": 444, "lr": 2.9861541225046975e-06} +{"train_loss": 0.014169808477163315, "grad_norm": 0.09025723487138748, "global_step": 47140, "epoch": 444, "lr": 2.9851453077669957e-06} +{"train_loss": 0.015844620764255524, "grad_norm": 0.06415186077356339, "global_step": 47141, "epoch": 444, "lr": 2.984136658219039e-06} +{"train_loss": 0.01570240780711174, "grad_norm": 0.13830958306789398, "global_step": 47142, "epoch": 444, "lr": 2.983128173864391e-06} +{"train_loss": 0.013763673603534698, "grad_norm": 0.06536255776882172, "global_step": 47143, "epoch": 444, "lr": 2.9821198547065775e-06} +{"train_loss": 0.01590830646455288, "grad_norm": 0.10115117579698563, "global_step": 47144, "epoch": 444, "lr": 2.981111700749145e-06} +{"train_loss": 0.014474700205028057, "grad_norm": 0.06294672191143036, "global_step": 47145, "epoch": 444, "lr": 2.9801037119956356e-06} +{"train_loss": 0.0184048879891634, "grad_norm": 0.08825641125440598, "global_step": 47146, "epoch": 444, "lr": 2.979095888449601e-06} +{"train_loss": 0.013888055458664894, "grad_norm": 0.05659692361950874, "global_step": 47147, "epoch": 444, "lr": 2.9780882301145673e-06} +{"train_loss": 0.014171022921800613, "grad_norm": 0.08895882219076157, "global_step": 47148, "epoch": 444, "lr": 2.977080736994092e-06} +{"train_loss": 0.01439820695668459, "grad_norm": 0.07952884584665298, "global_step": 47149, "epoch": 444, "lr": 2.9760734090916953e-06} +{"train_loss": 0.012585578486323357, "grad_norm": 0.1463300734758377, "global_step": 47150, "epoch": 444, "lr": 2.9750662464109345e-06} +{"train_loss": 0.01333657931536436, "grad_norm": 0.06809189170598984, "global_step": 47151, "epoch": 444, "lr": 2.97405924895533e-06} +{"train_loss": 0.020222298800945282, "grad_norm": 0.0805601254105568, "global_step": 47152, "epoch": 444, "lr": 2.9730524167284337e-06} +{"train_loss": 0.01710563525557518, "grad_norm": 0.10138747841119766, "global_step": 47153, "epoch": 444, "lr": 2.972045749733787e-06} +{"train_loss": 0.01593504287302494, "grad_norm": 0.09192147105932236, "global_step": 47154, "epoch": 444, "lr": 2.971039247974916e-06} +{"train_loss": 0.015632756054401398, "grad_norm": 0.11688166856765747, "global_step": 47155, "epoch": 444, "lr": 2.970032911455356e-06} +{"train_loss": 0.015632737427949905, "grad_norm": 0.08370860666036606, "global_step": 47156, "epoch": 444, "lr": 2.969026740178649e-06} +{"train_loss": 0.014208532869815826, "grad_norm": 0.0700492113828659, "global_step": 47157, "epoch": 444, "lr": 2.9680207341483356e-06} +{"train_loss": 0.016416111961007118, "grad_norm": 0.12496299296617508, "global_step": 47158, "epoch": 444, "lr": 2.9670148933679366e-06} +{"train_loss": 0.015408912673592567, "grad_norm": 0.11391766369342804, "global_step": 47159, "epoch": 444, "lr": 2.9660092178409927e-06} +{"train_loss": 0.013490661978721619, "grad_norm": 0.09488531202077866, "global_step": 47160, "epoch": 444, "lr": 2.9650037075710514e-06} +{"train_loss": 0.012484494596719742, "grad_norm": 0.06306792795658112, "global_step": 47161, "epoch": 444, "lr": 2.9639983625616206e-06} +{"train_loss": 0.01509915478527546, "grad_norm": 0.06516611576080322, "global_step": 47162, "epoch": 444, "lr": 2.9629931828162373e-06} +{"train_loss": 0.015954088419675827, "grad_norm": 0.1031833365559578, "global_step": 47163, "epoch": 444, "lr": 2.9619881683384532e-06} +{"train_loss": 0.015753157436847687, "grad_norm": 0.08254339545965195, "global_step": 47164, "epoch": 444, "lr": 2.960983319131777e-06} +{"train_loss": 0.01541473250836134, "grad_norm": 0.08774802833795547, "global_step": 47165, "epoch": 444, "lr": 2.9599786351997504e-06} +{"train_loss": 0.012953761965036392, "grad_norm": 0.061395302414894104, "global_step": 47166, "epoch": 444, "lr": 2.9589741165459094e-06} +{"train_loss": 0.013573067262768745, "grad_norm": 0.12308283150196075, "global_step": 47167, "epoch": 444, "lr": 2.9579697631737735e-06} +{"train_loss": 0.01461193710565567, "grad_norm": 0.08689692616462708, "global_step": 47168, "epoch": 444, "lr": 2.9569655750868674e-06} +{"train_loss": 0.014960692466219078, "grad_norm": 0.09998609125614166, "global_step": 47169, "epoch": 444, "lr": 2.9559615522887273e-06, "val_loss": 0.019200244918465614} +{"train_loss": 0.015398284420371056, "grad_norm": 0.06843339651823044, "global_step": 47170, "epoch": 445, "lr": 2.954957694782884e-06} +{"train_loss": 0.013073429465293884, "grad_norm": 0.06834877282381058, "global_step": 47171, "epoch": 445, "lr": 2.9539540025728507e-06} +{"train_loss": 0.012483019381761551, "grad_norm": 0.0972762405872345, "global_step": 47172, "epoch": 445, "lr": 2.952950475662175e-06} +{"train_loss": 0.014254332520067692, "grad_norm": 0.057838574051856995, "global_step": 47173, "epoch": 445, "lr": 2.951947114054365e-06} +{"train_loss": 0.0133647620677948, "grad_norm": 0.12623897194862366, "global_step": 47174, "epoch": 445, "lr": 2.9509439177529576e-06} +{"train_loss": 0.01259351335465908, "grad_norm": 0.10500871390104294, "global_step": 47175, "epoch": 445, "lr": 2.9499408867614595e-06} +{"train_loss": 0.013949898071587086, "grad_norm": 0.13238994777202606, "global_step": 47176, "epoch": 445, "lr": 2.9489380210834137e-06} +{"train_loss": 0.014733774587512016, "grad_norm": 0.13005490601062775, "global_step": 47177, "epoch": 445, "lr": 2.9479353207223446e-06} +{"train_loss": 0.012920645996928215, "grad_norm": 0.06115691736340523, "global_step": 47178, "epoch": 445, "lr": 2.946932785681766e-06} +{"train_loss": 0.013667987659573555, "grad_norm": 0.07171168178319931, "global_step": 47179, "epoch": 445, "lr": 2.945930415965198e-06} +{"train_loss": 0.018909413367509842, "grad_norm": 0.1742040067911148, "global_step": 47180, "epoch": 445, "lr": 2.9449282115761754e-06} +{"train_loss": 0.016583438962697983, "grad_norm": 0.06358785182237625, "global_step": 47181, "epoch": 445, "lr": 2.943926172518202e-06} +{"train_loss": 0.012168982066214085, "grad_norm": 0.05911756679415703, "global_step": 47182, "epoch": 445, "lr": 2.942924298794808e-06} +{"train_loss": 0.014227374456822872, "grad_norm": 0.07263367623090744, "global_step": 47183, "epoch": 445, "lr": 2.9419225904095126e-06} +{"train_loss": 0.013236472383141518, "grad_norm": 0.06678018718957901, "global_step": 47184, "epoch": 445, "lr": 2.940921047365852e-06} +{"train_loss": 0.015116434544324875, "grad_norm": 0.09208272397518158, "global_step": 47185, "epoch": 445, "lr": 2.9399196696673126e-06} +{"train_loss": 0.01290593110024929, "grad_norm": 0.06497158110141754, "global_step": 47186, "epoch": 445, "lr": 2.93891845731743e-06} +{"train_loss": 0.01618461310863495, "grad_norm": 0.0897815078496933, "global_step": 47187, "epoch": 445, "lr": 2.9379174103197347e-06} +{"train_loss": 0.014454510062932968, "grad_norm": 0.05824436992406845, "global_step": 47188, "epoch": 445, "lr": 2.936916528677719e-06} +{"train_loss": 0.014586016535758972, "grad_norm": 0.09830455482006073, "global_step": 47189, "epoch": 445, "lr": 2.935915812394918e-06} +{"train_loss": 0.015404822304844856, "grad_norm": 0.0903380811214447, "global_step": 47190, "epoch": 445, "lr": 2.9349152614748355e-06} +{"train_loss": 0.014731308445334435, "grad_norm": 0.07153580337762833, "global_step": 47191, "epoch": 445, "lr": 2.9339148759209957e-06} +{"train_loss": 0.014560751616954803, "grad_norm": 0.0885394960641861, "global_step": 47192, "epoch": 445, "lr": 2.932914655736907e-06} +{"train_loss": 0.01483924314379692, "grad_norm": 0.13949179649353027, "global_step": 47193, "epoch": 445, "lr": 2.931914600926089e-06} +{"train_loss": 0.015349740162491798, "grad_norm": 0.08452508598566055, "global_step": 47194, "epoch": 445, "lr": 2.9309147114920608e-06} +{"train_loss": 0.013413570821285248, "grad_norm": 0.10790283977985382, "global_step": 47195, "epoch": 445, "lr": 2.929914987438326e-06} +{"train_loss": 0.013447592966258526, "grad_norm": 0.06705398857593536, "global_step": 47196, "epoch": 445, "lr": 2.928915428768392e-06} +{"train_loss": 0.013987822458148003, "grad_norm": 0.09825041145086288, "global_step": 47197, "epoch": 445, "lr": 2.9279160354857783e-06} +{"train_loss": 0.014424378983676434, "grad_norm": 0.08715518563985825, "global_step": 47198, "epoch": 445, "lr": 2.92691680759401e-06} +{"train_loss": 0.016739023849368095, "grad_norm": 0.12494460493326187, "global_step": 47199, "epoch": 445, "lr": 2.9259177450965682e-06} +{"train_loss": 0.015314795076847076, "grad_norm": 0.10070367157459259, "global_step": 47200, "epoch": 445, "lr": 2.924918847996988e-06} +{"train_loss": 0.015416420996189117, "grad_norm": 0.0825122818350792, "global_step": 47201, "epoch": 445, "lr": 2.9239201162987784e-06} +{"train_loss": 0.014448709785938263, "grad_norm": 0.09367072582244873, "global_step": 47202, "epoch": 445, "lr": 2.9229215500054308e-06} +{"train_loss": 0.015793710947036743, "grad_norm": 0.12034852057695389, "global_step": 47203, "epoch": 445, "lr": 2.921923149120459e-06} +{"train_loss": 0.015941757708787918, "grad_norm": 0.08699880540370941, "global_step": 47204, "epoch": 445, "lr": 2.920924913647388e-06} +{"train_loss": 0.014786393381655216, "grad_norm": 0.07832514494657516, "global_step": 47205, "epoch": 445, "lr": 2.9199268435896987e-06} +{"train_loss": 0.01650279387831688, "grad_norm": 0.089869923889637, "global_step": 47206, "epoch": 445, "lr": 2.9189289389509155e-06} +{"train_loss": 0.016548801213502884, "grad_norm": 0.07205726206302643, "global_step": 47207, "epoch": 445, "lr": 2.9179311997345527e-06} +{"train_loss": 0.013659194111824036, "grad_norm": 0.06296898424625397, "global_step": 47208, "epoch": 445, "lr": 2.916933625944096e-06} +{"train_loss": 0.014316758140921593, "grad_norm": 0.07199293375015259, "global_step": 47209, "epoch": 445, "lr": 2.9159362175830597e-06} +{"train_loss": 0.015498636290431023, "grad_norm": 0.07771054655313492, "global_step": 47210, "epoch": 445, "lr": 2.9149389746549407e-06} +{"train_loss": 0.014287851750850677, "grad_norm": 0.06880658119916916, "global_step": 47211, "epoch": 445, "lr": 2.913941897163258e-06} +{"train_loss": 0.01561763882637024, "grad_norm": 0.07564043253660202, "global_step": 47212, "epoch": 445, "lr": 2.912944985111499e-06} +{"train_loss": 0.013749255798757076, "grad_norm": 0.05853227898478508, "global_step": 47213, "epoch": 445, "lr": 2.9119482385031815e-06} +{"train_loss": 0.014824941754341125, "grad_norm": 0.07377966493368149, "global_step": 47214, "epoch": 445, "lr": 2.9109516573417926e-06} +{"train_loss": 0.01329024974256754, "grad_norm": 0.11474930495023727, "global_step": 47215, "epoch": 445, "lr": 2.9099552416308518e-06} +{"train_loss": 0.014822104014456272, "grad_norm": 0.06906010955572128, "global_step": 47216, "epoch": 445, "lr": 2.9089589913738393e-06} +{"train_loss": 0.015418956987559795, "grad_norm": 0.126878023147583, "global_step": 47217, "epoch": 445, "lr": 2.907962906574263e-06} +{"train_loss": 0.013115206733345985, "grad_norm": 0.06514991074800491, "global_step": 47218, "epoch": 445, "lr": 2.9069669872356374e-06} +{"train_loss": 0.014360497705638409, "grad_norm": 0.10833821445703506, "global_step": 47219, "epoch": 445, "lr": 2.905971233361449e-06} +{"train_loss": 0.015359265729784966, "grad_norm": 0.059165455400943756, "global_step": 47220, "epoch": 445, "lr": 2.9049756449551824e-06} +{"train_loss": 0.01594480872154236, "grad_norm": 0.06583847850561142, "global_step": 47221, "epoch": 445, "lr": 2.9039802220203636e-06} +{"train_loss": 0.016109399497509003, "grad_norm": 0.061124131083488464, "global_step": 47222, "epoch": 445, "lr": 2.9029849645604733e-06} +{"train_loss": 0.016433022916316986, "grad_norm": 0.0675029382109642, "global_step": 47223, "epoch": 445, "lr": 2.901989872579003e-06} +{"train_loss": 0.016948312520980835, "grad_norm": 0.09995385259389877, "global_step": 47224, "epoch": 445, "lr": 2.9009949460794715e-06} +{"train_loss": 0.015480441972613335, "grad_norm": 0.17841756343841553, "global_step": 47225, "epoch": 445, "lr": 2.9000001850653604e-06} +{"train_loss": 0.01250367984175682, "grad_norm": 0.09143044799566269, "global_step": 47226, "epoch": 445, "lr": 2.8990055895401546e-06} +{"train_loss": 0.015591492876410484, "grad_norm": 0.13797888159751892, "global_step": 47227, "epoch": 445, "lr": 2.8980111595073634e-06} +{"train_loss": 0.014731768518686295, "grad_norm": 0.08029275387525558, "global_step": 47228, "epoch": 445, "lr": 2.8970168949704834e-06} +{"train_loss": 0.015812890604138374, "grad_norm": 0.17306199669837952, "global_step": 47229, "epoch": 445, "lr": 2.89602279593299e-06} +{"train_loss": 0.015621173195540905, "grad_norm": 0.10324735939502716, "global_step": 47230, "epoch": 445, "lr": 2.895028862398391e-06} +{"train_loss": 0.016828708350658417, "grad_norm": 0.20733729004859924, "global_step": 47231, "epoch": 445, "lr": 2.8940350943701845e-06} +{"train_loss": 0.015274512581527233, "grad_norm": 0.1337079405784607, "global_step": 47232, "epoch": 445, "lr": 2.89304149185185e-06} +{"train_loss": 0.014748598448932171, "grad_norm": 0.06427358090877533, "global_step": 47233, "epoch": 445, "lr": 2.892048054846874e-06} +{"train_loss": 0.013705744408071041, "grad_norm": 0.07263989746570587, "global_step": 47234, "epoch": 445, "lr": 2.891054783358754e-06} +{"train_loss": 0.015495151281356812, "grad_norm": 0.09761779010295868, "global_step": 47235, "epoch": 445, "lr": 2.890061677390993e-06} +{"train_loss": 0.013339947909116745, "grad_norm": 0.08547262102365494, "global_step": 47236, "epoch": 445, "lr": 2.889068736947059e-06} +{"train_loss": 0.013855380937457085, "grad_norm": 0.10779328644275665, "global_step": 47237, "epoch": 445, "lr": 2.888075962030451e-06} +{"train_loss": 0.014726020395755768, "grad_norm": 0.07834755629301071, "global_step": 47238, "epoch": 445, "lr": 2.8870833526446484e-06} +{"train_loss": 0.014336694031953812, "grad_norm": 0.08230867236852646, "global_step": 47239, "epoch": 445, "lr": 2.8860909087931543e-06} +{"train_loss": 0.015816640108823776, "grad_norm": 0.08411110937595367, "global_step": 47240, "epoch": 445, "lr": 2.8850986304794435e-06} +{"train_loss": 0.014205977320671082, "grad_norm": 0.0747542530298233, "global_step": 47241, "epoch": 445, "lr": 2.884106517707008e-06} +{"train_loss": 0.015983525663614273, "grad_norm": 0.09996113926172256, "global_step": 47242, "epoch": 445, "lr": 2.8831145704793394e-06} +{"train_loss": 0.014081675559282303, "grad_norm": 0.06967951357364655, "global_step": 47243, "epoch": 445, "lr": 2.882122788799907e-06} +{"train_loss": 0.012983478605747223, "grad_norm": 0.10543454438447952, "global_step": 47244, "epoch": 445, "lr": 2.881131172672202e-06} +{"train_loss": 0.013907966203987598, "grad_norm": 0.07369904965162277, "global_step": 47245, "epoch": 445, "lr": 2.8801397220997174e-06} +{"train_loss": 0.015742160379886627, "grad_norm": 0.08215377479791641, "global_step": 47246, "epoch": 445, "lr": 2.8791484370859268e-06} +{"train_loss": 0.015886850655078888, "grad_norm": 0.08919395506381989, "global_step": 47247, "epoch": 445, "lr": 2.878157317634311e-06} +{"train_loss": 0.014889195561408997, "grad_norm": 0.07574225962162018, "global_step": 47248, "epoch": 445, "lr": 2.8771663637483682e-06} +{"train_loss": 0.017891669645905495, "grad_norm": 0.06288326531648636, "global_step": 47249, "epoch": 445, "lr": 2.8761755754315668e-06} +{"train_loss": 0.01443401351571083, "grad_norm": 0.09291959553956985, "global_step": 47250, "epoch": 445, "lr": 2.875184952687382e-06} +{"train_loss": 0.01479722186923027, "grad_norm": 0.11171343922615051, "global_step": 47251, "epoch": 445, "lr": 2.8741944955193112e-06} +{"train_loss": 0.01467345841228962, "grad_norm": 0.12131918221712112, "global_step": 47252, "epoch": 445, "lr": 2.873204203930824e-06} +{"train_loss": 0.015244148671627045, "grad_norm": 0.08462842553853989, "global_step": 47253, "epoch": 445, "lr": 2.8722140779254004e-06} +{"train_loss": 0.016102738678455353, "grad_norm": 0.09152870625257492, "global_step": 47254, "epoch": 445, "lr": 2.871224117506527e-06} +{"train_loss": 0.01353004015982151, "grad_norm": 0.09352526813745499, "global_step": 47255, "epoch": 445, "lr": 2.8702343226776728e-06} +{"train_loss": 0.014758461154997349, "grad_norm": 0.15839378535747528, "global_step": 47256, "epoch": 445, "lr": 2.8692446934423246e-06} +{"train_loss": 0.017135269939899445, "grad_norm": 0.06250259280204773, "global_step": 47257, "epoch": 445, "lr": 2.8682552298039455e-06} +{"train_loss": 0.017218545079231262, "grad_norm": 0.07234326004981995, "global_step": 47258, "epoch": 445, "lr": 2.8672659317660166e-06} +{"train_loss": 0.016736218705773354, "grad_norm": 0.09098536521196365, "global_step": 47259, "epoch": 445, "lr": 2.8662767993320295e-06} +{"train_loss": 0.01582302711904049, "grad_norm": 0.07178681343793869, "global_step": 47260, "epoch": 445, "lr": 2.8652878325054477e-06} +{"train_loss": 0.016652625054121017, "grad_norm": 0.07177623361349106, "global_step": 47261, "epoch": 445, "lr": 2.864299031289741e-06} +{"train_loss": 0.015150770545005798, "grad_norm": 0.0571686252951622, "global_step": 47262, "epoch": 445, "lr": 2.86331039568839e-06} +{"train_loss": 0.015511120669543743, "grad_norm": 0.11460404098033905, "global_step": 47263, "epoch": 445, "lr": 2.862321925704864e-06} +{"train_loss": 0.01795577071607113, "grad_norm": 0.07426650077104568, "global_step": 47264, "epoch": 445, "lr": 2.861333621342638e-06} +{"train_loss": 0.015987752005457878, "grad_norm": 0.08089130371809006, "global_step": 47265, "epoch": 445, "lr": 2.8603454826051923e-06} +{"train_loss": 0.015825171023607254, "grad_norm": 0.09251008927822113, "global_step": 47266, "epoch": 445, "lr": 2.859357509495991e-06} +{"train_loss": 0.014760918915271759, "grad_norm": 0.07025756686925888, "global_step": 47267, "epoch": 445, "lr": 2.858369702018504e-06} +{"train_loss": 0.01448110956698656, "grad_norm": 0.08245792984962463, "global_step": 47268, "epoch": 445, "lr": 2.8573820601762003e-06} +{"train_loss": 0.016431856900453568, "grad_norm": 0.11353085935115814, "global_step": 47269, "epoch": 445, "lr": 2.856394583972566e-06} +{"train_loss": 0.014218856580555439, "grad_norm": 0.11818768829107285, "global_step": 47270, "epoch": 445, "lr": 2.8554072734110484e-06} +{"train_loss": 0.014884283766150475, "grad_norm": 0.07275737822055817, "global_step": 47271, "epoch": 445, "lr": 2.854420128495122e-06} +{"train_loss": 0.01540853176265955, "grad_norm": 0.0947449579834938, "global_step": 47272, "epoch": 445, "lr": 2.853433149228274e-06} +{"train_loss": 0.012787602841854095, "grad_norm": 0.06782654672861099, "global_step": 47273, "epoch": 445, "lr": 2.852446335613956e-06} +{"train_loss": 0.016873741522431374, "grad_norm": 0.11409412324428558, "global_step": 47274, "epoch": 445, "lr": 2.8514596876556333e-06} +{"train_loss": 0.014956757593197081, "grad_norm": 0.10975170880556107, "global_step": 47275, "epoch": 445, "lr": 2.850473205356774e-06, "val_loss": 0.019070686772465706} +{"train_loss": 0.015533342026174068, "grad_norm": 0.0893963947892189, "global_step": 47276, "epoch": 446, "lr": 2.8494868887208537e-06} +{"train_loss": 0.017221637070178986, "grad_norm": 0.0737631618976593, "global_step": 47277, "epoch": 446, "lr": 2.848500737751325e-06} +{"train_loss": 0.014534746296703815, "grad_norm": 0.05750090628862381, "global_step": 47278, "epoch": 446, "lr": 2.847514752451669e-06} +{"train_loss": 0.01690029725432396, "grad_norm": 0.11688455194234848, "global_step": 47279, "epoch": 446, "lr": 2.8465289328253376e-06} +{"train_loss": 0.01785886287689209, "grad_norm": 0.08134270459413528, "global_step": 47280, "epoch": 446, "lr": 2.8455432788757895e-06} +{"train_loss": 0.015412853099405766, "grad_norm": 0.07491043210029602, "global_step": 47281, "epoch": 446, "lr": 2.8445577906064945e-06} +{"train_loss": 0.015260079875588417, "grad_norm": 0.06370798498392105, "global_step": 47282, "epoch": 446, "lr": 2.8435724680209275e-06} +{"train_loss": 0.013790739700198174, "grad_norm": 0.07708688080310822, "global_step": 47283, "epoch": 446, "lr": 2.84258731112253e-06} +{"train_loss": 0.015832703560590744, "grad_norm": 0.06623362004756927, "global_step": 47284, "epoch": 446, "lr": 2.8416023199147766e-06} +{"train_loss": 0.014274866320192814, "grad_norm": 0.07501116394996643, "global_step": 47285, "epoch": 446, "lr": 2.840617494401121e-06} +{"train_loss": 0.01421377994120121, "grad_norm": 0.08733939379453659, "global_step": 47286, "epoch": 446, "lr": 2.839632834585032e-06} +{"train_loss": 0.015295131132006645, "grad_norm": 0.05640533193945885, "global_step": 47287, "epoch": 446, "lr": 2.838648340469957e-06} +{"train_loss": 0.015416573733091354, "grad_norm": 0.08153112977743149, "global_step": 47288, "epoch": 446, "lr": 2.83766401205936e-06} +{"train_loss": 0.016946658492088318, "grad_norm": 0.0928688645362854, "global_step": 47289, "epoch": 446, "lr": 2.8366798493567105e-06} +{"train_loss": 0.01429595798254013, "grad_norm": 0.08055272698402405, "global_step": 47290, "epoch": 446, "lr": 2.8356958523654553e-06} +{"train_loss": 0.017589973285794258, "grad_norm": 0.06193923205137253, "global_step": 47291, "epoch": 446, "lr": 2.8347120210890478e-06} +{"train_loss": 0.012649234384298325, "grad_norm": 0.10744816809892654, "global_step": 47292, "epoch": 446, "lr": 2.8337283555309458e-06} +{"train_loss": 0.016623210161924362, "grad_norm": 0.12606564164161682, "global_step": 47293, "epoch": 446, "lr": 2.8327448556946244e-06} +{"train_loss": 0.01610133796930313, "grad_norm": 0.1289711743593216, "global_step": 47294, "epoch": 446, "lr": 2.831761521583515e-06} +{"train_loss": 0.014700166881084442, "grad_norm": 0.06396038830280304, "global_step": 47295, "epoch": 446, "lr": 2.8307783532010913e-06} +{"train_loss": 0.015445937402546406, "grad_norm": 0.11081642657518387, "global_step": 47296, "epoch": 446, "lr": 2.8297953505507902e-06} +{"train_loss": 0.016110695898532867, "grad_norm": 0.08878004550933838, "global_step": 47297, "epoch": 446, "lr": 2.828812513636081e-06} +{"train_loss": 0.01557885855436325, "grad_norm": 0.12073390930891037, "global_step": 47298, "epoch": 446, "lr": 2.8278298424604055e-06} +{"train_loss": 0.015851490199565887, "grad_norm": 0.056907206773757935, "global_step": 47299, "epoch": 446, "lr": 2.826847337027222e-06} +{"train_loss": 0.016161911189556122, "grad_norm": 0.0767369493842125, "global_step": 47300, "epoch": 446, "lr": 2.8258649973399887e-06} +{"train_loss": 0.0163942351937294, "grad_norm": 0.08178035914897919, "global_step": 47301, "epoch": 446, "lr": 2.8248828234021474e-06} +{"train_loss": 0.015176473185420036, "grad_norm": 0.08053936064243317, "global_step": 47302, "epoch": 446, "lr": 2.8239008152171452e-06} +{"train_loss": 0.014132272452116013, "grad_norm": 0.07161857932806015, "global_step": 47303, "epoch": 446, "lr": 2.8229189727884465e-06} +{"train_loss": 0.014797130599617958, "grad_norm": 0.10023926943540573, "global_step": 47304, "epoch": 446, "lr": 2.821937296119487e-06} +{"train_loss": 0.014241911470890045, "grad_norm": 0.0626615509390831, "global_step": 47305, "epoch": 446, "lr": 2.820955785213725e-06} +{"train_loss": 0.01463788840919733, "grad_norm": 0.061266642063856125, "global_step": 47306, "epoch": 446, "lr": 2.819974440074613e-06} +{"train_loss": 0.01647094264626503, "grad_norm": 0.08835793286561966, "global_step": 47307, "epoch": 446, "lr": 2.8189932607055934e-06} +{"train_loss": 0.01607230119407177, "grad_norm": 0.07382325828075409, "global_step": 47308, "epoch": 446, "lr": 2.8180122471101024e-06} +{"train_loss": 0.01746668480336666, "grad_norm": 0.10126591473817825, "global_step": 47309, "epoch": 446, "lr": 2.8170313992915974e-06} +{"train_loss": 0.014448177069425583, "grad_norm": 0.07434472441673279, "global_step": 47310, "epoch": 446, "lr": 2.8160507172535376e-06} +{"train_loss": 0.01632455736398697, "grad_norm": 0.12864108383655548, "global_step": 47311, "epoch": 446, "lr": 2.8150702009993424e-06} +{"train_loss": 0.01316993311047554, "grad_norm": 0.09293525665998459, "global_step": 47312, "epoch": 446, "lr": 2.8140898505324755e-06} +{"train_loss": 0.012177368625998497, "grad_norm": 0.1489952653646469, "global_step": 47313, "epoch": 446, "lr": 2.8131096658563783e-06} +{"train_loss": 0.012785936705768108, "grad_norm": 0.09315815567970276, "global_step": 47314, "epoch": 446, "lr": 2.8121296469744986e-06} +{"train_loss": 0.013530373573303223, "grad_norm": 0.058390628546476364, "global_step": 47315, "epoch": 446, "lr": 2.811149793890261e-06} +{"train_loss": 0.014762196689844131, "grad_norm": 0.0725984275341034, "global_step": 47316, "epoch": 446, "lr": 2.810170106607124e-06} +{"train_loss": 0.013894179835915565, "grad_norm": 0.07737264782190323, "global_step": 47317, "epoch": 446, "lr": 2.809190585128535e-06} +{"train_loss": 0.014381492510437965, "grad_norm": 0.08045994490385056, "global_step": 47318, "epoch": 446, "lr": 2.8082112294579188e-06} +{"train_loss": 0.01275267731398344, "grad_norm": 0.077652707695961, "global_step": 47319, "epoch": 446, "lr": 2.8072320395987285e-06} +{"train_loss": 0.014252165332436562, "grad_norm": 0.11654022336006165, "global_step": 47320, "epoch": 446, "lr": 2.806253015554405e-06} +{"train_loss": 0.014835881069302559, "grad_norm": 0.08884577453136444, "global_step": 47321, "epoch": 446, "lr": 2.80527415732838e-06} +{"train_loss": 0.014624767936766148, "grad_norm": 0.08268243074417114, "global_step": 47322, "epoch": 446, "lr": 2.8042954649240948e-06} +{"train_loss": 0.013875285163521767, "grad_norm": 0.07776730507612228, "global_step": 47323, "epoch": 446, "lr": 2.8033169383449962e-06} +{"train_loss": 0.01340609509497881, "grad_norm": 0.0802178606390953, "global_step": 47324, "epoch": 446, "lr": 2.80233857759451e-06} +{"train_loss": 0.015532001852989197, "grad_norm": 0.13298337161540985, "global_step": 47325, "epoch": 446, "lr": 2.801360382676088e-06} +{"train_loss": 0.015298687852919102, "grad_norm": 0.06683225184679031, "global_step": 47326, "epoch": 446, "lr": 2.800382353593156e-06} +{"train_loss": 0.015700936317443848, "grad_norm": 0.14216966927051544, "global_step": 47327, "epoch": 446, "lr": 2.7994044903491557e-06} +{"train_loss": 0.014240861870348454, "grad_norm": 0.0849761962890625, "global_step": 47328, "epoch": 446, "lr": 2.7984267929475173e-06} +{"train_loss": 0.012245763093233109, "grad_norm": 0.10068567097187042, "global_step": 47329, "epoch": 446, "lr": 2.797449261391677e-06} +{"train_loss": 0.014421968720853329, "grad_norm": 0.09379859268665314, "global_step": 47330, "epoch": 446, "lr": 2.7964718956850765e-06} +{"train_loss": 0.01483135111629963, "grad_norm": 0.09652591496706009, "global_step": 47331, "epoch": 446, "lr": 2.7954946958311524e-06} +{"train_loss": 0.012485935352742672, "grad_norm": 0.15106533467769623, "global_step": 47332, "epoch": 446, "lr": 2.794517661833318e-06} +{"train_loss": 0.01450654398649931, "grad_norm": 0.08167808502912521, "global_step": 47333, "epoch": 446, "lr": 2.7935407936950266e-06} +{"train_loss": 0.01575568877160549, "grad_norm": 0.08909104764461517, "global_step": 47334, "epoch": 446, "lr": 2.7925640914197027e-06} +{"train_loss": 0.013365015387535095, "grad_norm": 0.051983222365379333, "global_step": 47335, "epoch": 446, "lr": 2.791587555010777e-06} +{"train_loss": 0.013556132093071938, "grad_norm": 0.12008782476186752, "global_step": 47336, "epoch": 446, "lr": 2.7906111844716863e-06} +{"train_loss": 0.014853274449706078, "grad_norm": 0.09688791632652283, "global_step": 47337, "epoch": 446, "lr": 2.7896349798058607e-06} +{"train_loss": 0.016257040202617645, "grad_norm": 0.06212092563509941, "global_step": 47338, "epoch": 446, "lr": 2.788658941016714e-06} +{"train_loss": 0.014180812984704971, "grad_norm": 0.06847439706325531, "global_step": 47339, "epoch": 446, "lr": 2.787683068107694e-06} +{"train_loss": 0.015493934974074364, "grad_norm": 0.06415639072656631, "global_step": 47340, "epoch": 446, "lr": 2.7867073610822192e-06} +{"train_loss": 0.016858506947755814, "grad_norm": 0.08935333788394928, "global_step": 47341, "epoch": 446, "lr": 2.785731819943732e-06} +{"train_loss": 0.013367637991905212, "grad_norm": 0.09397713840007782, "global_step": 47342, "epoch": 446, "lr": 2.7847564446956465e-06} +{"train_loss": 0.01535040233284235, "grad_norm": 0.07822109758853912, "global_step": 47343, "epoch": 446, "lr": 2.783781235341387e-06} +{"train_loss": 0.016166651621460915, "grad_norm": 0.06060061231255531, "global_step": 47344, "epoch": 446, "lr": 2.782806191884396e-06} +{"train_loss": 0.015837565064430237, "grad_norm": 0.07478852570056915, "global_step": 47345, "epoch": 446, "lr": 2.7818313143280873e-06} +{"train_loss": 0.013930018991231918, "grad_norm": 0.12717467546463013, "global_step": 47346, "epoch": 446, "lr": 2.7808566026758796e-06} +{"train_loss": 0.013653283007442951, "grad_norm": 0.11615967005491257, "global_step": 47347, "epoch": 446, "lr": 2.7798820569312212e-06} +{"train_loss": 0.01649712212383747, "grad_norm": 0.12115460634231567, "global_step": 47348, "epoch": 446, "lr": 2.7789076770975142e-06} +{"train_loss": 0.01606588065624237, "grad_norm": 0.14929473400115967, "global_step": 47349, "epoch": 446, "lr": 2.7779334631781895e-06} +{"train_loss": 0.015315393917262554, "grad_norm": 0.0624130554497242, "global_step": 47350, "epoch": 446, "lr": 2.7769594151766666e-06} +{"train_loss": 0.015079188160598278, "grad_norm": 0.08363015204668045, "global_step": 47351, "epoch": 446, "lr": 2.7759855330963815e-06} +{"train_loss": 0.014166053384542465, "grad_norm": 0.08097776025533676, "global_step": 47352, "epoch": 446, "lr": 2.775011816940737e-06} +{"train_loss": 0.015598207712173462, "grad_norm": 0.07097387313842773, "global_step": 47353, "epoch": 446, "lr": 2.7740382667131638e-06} +{"train_loss": 0.012896200641989708, "grad_norm": 0.07219664007425308, "global_step": 47354, "epoch": 446, "lr": 2.7730648824170867e-06} +{"train_loss": 0.015065465122461319, "grad_norm": 0.062123991549015045, "global_step": 47355, "epoch": 446, "lr": 2.7720916640559257e-06} +{"train_loss": 0.014388497918844223, "grad_norm": 0.1963662952184677, "global_step": 47356, "epoch": 446, "lr": 2.7711186116330833e-06} +{"train_loss": 0.01402951404452324, "grad_norm": 0.09864803403615952, "global_step": 47357, "epoch": 446, "lr": 2.770145725151996e-06} +{"train_loss": 0.015127965249121189, "grad_norm": 0.07455957680940628, "global_step": 47358, "epoch": 446, "lr": 2.7691730046160822e-06} +{"train_loss": 0.012984590604901314, "grad_norm": 0.055159423500299454, "global_step": 47359, "epoch": 446, "lr": 2.7682004500287464e-06} +{"train_loss": 0.015736186876893044, "grad_norm": 0.10522890090942383, "global_step": 47360, "epoch": 446, "lr": 2.7672280613934176e-06} +{"train_loss": 0.01419051829725504, "grad_norm": 0.10045503079891205, "global_step": 47361, "epoch": 446, "lr": 2.766255838713516e-06} +{"train_loss": 0.017448969185352325, "grad_norm": 0.10501082986593246, "global_step": 47362, "epoch": 446, "lr": 2.7652837819924337e-06} +{"train_loss": 0.013542707078158855, "grad_norm": 0.07171382755041122, "global_step": 47363, "epoch": 446, "lr": 2.764311891233612e-06} +{"train_loss": 0.013257035985589027, "grad_norm": 0.06050410121679306, "global_step": 47364, "epoch": 446, "lr": 2.763340166440459e-06} +{"train_loss": 0.01803300902247429, "grad_norm": 0.10355820506811142, "global_step": 47365, "epoch": 446, "lr": 2.762368607616378e-06} +{"train_loss": 0.01665075495839119, "grad_norm": 0.09926123172044754, "global_step": 47366, "epoch": 446, "lr": 2.761397214764794e-06} +{"train_loss": 0.01602786034345627, "grad_norm": 0.10680978745222092, "global_step": 47367, "epoch": 446, "lr": 2.7604259878891146e-06} +{"train_loss": 0.012993715703487396, "grad_norm": 0.10755178332328796, "global_step": 47368, "epoch": 446, "lr": 2.7594549269927605e-06} +{"train_loss": 0.013369903899729252, "grad_norm": 0.13949772715568542, "global_step": 47369, "epoch": 446, "lr": 2.7584840320791337e-06} +{"train_loss": 0.014893297106027603, "grad_norm": 0.10341131687164307, "global_step": 47370, "epoch": 446, "lr": 2.757513303151643e-06} +{"train_loss": 0.013236109167337418, "grad_norm": 0.07010474801063538, "global_step": 47371, "epoch": 446, "lr": 2.756542740213719e-06} +{"train_loss": 0.014269396662712097, "grad_norm": 0.08224338293075562, "global_step": 47372, "epoch": 446, "lr": 2.755572343268753e-06} +{"train_loss": 0.017303183674812317, "grad_norm": 0.1414390504360199, "global_step": 47373, "epoch": 446, "lr": 2.7546021123201538e-06} +{"train_loss": 0.01631065644323826, "grad_norm": 0.12980680167675018, "global_step": 47374, "epoch": 446, "lr": 2.753632047371335e-06} +{"train_loss": 0.014761016704142094, "grad_norm": 0.07438083738088608, "global_step": 47375, "epoch": 446, "lr": 2.7526621484257165e-06} +{"train_loss": 0.015483686700463295, "grad_norm": 0.10259471088647842, "global_step": 47376, "epoch": 446, "lr": 2.751692415486684e-06} +{"train_loss": 0.013801359571516514, "grad_norm": 0.06036384031176567, "global_step": 47377, "epoch": 446, "lr": 2.750722848557663e-06} +{"train_loss": 0.015679288655519485, "grad_norm": 0.07233741134405136, "global_step": 47378, "epoch": 446, "lr": 2.749753447642067e-06} +{"train_loss": 0.016422316431999207, "grad_norm": 0.06737420707941055, "global_step": 47379, "epoch": 446, "lr": 2.7487842127432716e-06} +{"train_loss": 0.014635283499956131, "grad_norm": 0.0883597880601883, "global_step": 47380, "epoch": 446, "lr": 2.747815143864696e-06} +{"train_loss": 0.01495775611157406, "grad_norm": 0.1266440600156784, "global_step": 47381, "epoch": 446, "lr": 2.746846241009765e-06, "val_loss": 0.019239168614149094} +{"train_loss": 0.015906192362308502, "grad_norm": 0.07072947174310684, "global_step": 47382, "epoch": 447, "lr": 2.745877504181854e-06} +{"train_loss": 0.011789729818701744, "grad_norm": 0.15994477272033691, "global_step": 47383, "epoch": 447, "lr": 2.744908933384377e-06} +{"train_loss": 0.01614704355597496, "grad_norm": 0.11552770435810089, "global_step": 47384, "epoch": 447, "lr": 2.7439405286207474e-06} +{"train_loss": 0.014512712135910988, "grad_norm": 0.07165295630693436, "global_step": 47385, "epoch": 447, "lr": 2.7429722898943634e-06} +{"train_loss": 0.017234452068805695, "grad_norm": 0.09466195106506348, "global_step": 47386, "epoch": 447, "lr": 2.742004217208616e-06} +{"train_loss": 0.018825529143214226, "grad_norm": 0.12816347181797028, "global_step": 47387, "epoch": 447, "lr": 2.741036310566908e-06} +{"train_loss": 0.016445623710751534, "grad_norm": 0.06111957132816315, "global_step": 47388, "epoch": 447, "lr": 2.740068569972659e-06} +{"train_loss": 0.015510523691773415, "grad_norm": 0.09691836684942245, "global_step": 47389, "epoch": 447, "lr": 2.7391009954292556e-06} +{"train_loss": 0.014890198595821857, "grad_norm": 0.08489246666431427, "global_step": 47390, "epoch": 447, "lr": 2.7381335869400894e-06} +{"train_loss": 0.01679062657058239, "grad_norm": 0.08807387948036194, "global_step": 47391, "epoch": 447, "lr": 2.7371663445085684e-06} +{"train_loss": 0.013592192903161049, "grad_norm": 0.08806739002466202, "global_step": 47392, "epoch": 447, "lr": 2.7361992681380956e-06} +{"train_loss": 0.01190679706633091, "grad_norm": 0.05433575436472893, "global_step": 47393, "epoch": 447, "lr": 2.735232357832057e-06} +{"train_loss": 0.01410618145018816, "grad_norm": 0.10502813011407852, "global_step": 47394, "epoch": 447, "lr": 2.7342656135938615e-06} +{"train_loss": 0.011943954974412918, "grad_norm": 0.11284495890140533, "global_step": 47395, "epoch": 447, "lr": 2.733299035426906e-06} +{"train_loss": 0.013793762773275375, "grad_norm": 0.08355061709880829, "global_step": 47396, "epoch": 447, "lr": 2.7323326233345826e-06} +{"train_loss": 0.01554735004901886, "grad_norm": 0.11380760371685028, "global_step": 47397, "epoch": 447, "lr": 2.731366377320277e-06} +{"train_loss": 0.013154895976185799, "grad_norm": 0.05660512298345566, "global_step": 47398, "epoch": 447, "lr": 2.7304002973873976e-06} +{"train_loss": 0.013626709580421448, "grad_norm": 0.10176614671945572, "global_step": 47399, "epoch": 447, "lr": 2.7294343835393368e-06} +{"train_loss": 0.015207244083285332, "grad_norm": 0.08741839975118637, "global_step": 47400, "epoch": 447, "lr": 2.7284686357794796e-06} +{"train_loss": 0.016777532175183296, "grad_norm": 0.07867827266454697, "global_step": 47401, "epoch": 447, "lr": 2.7275030541112358e-06} +{"train_loss": 0.015381783246994019, "grad_norm": 0.08376168459653854, "global_step": 47402, "epoch": 447, "lr": 2.726537638537985e-06} +{"train_loss": 0.017529163509607315, "grad_norm": 0.06768399477005005, "global_step": 47403, "epoch": 447, "lr": 2.725572389063119e-06} +{"train_loss": 0.014339745044708252, "grad_norm": 0.08113383501768112, "global_step": 47404, "epoch": 447, "lr": 2.72460730569003e-06} +{"train_loss": 0.015193380415439606, "grad_norm": 0.08342934399843216, "global_step": 47405, "epoch": 447, "lr": 2.7236423884221207e-06} +{"train_loss": 0.014800402335822582, "grad_norm": 0.08555731177330017, "global_step": 47406, "epoch": 447, "lr": 2.7226776372627605e-06} +{"train_loss": 0.01412487868219614, "grad_norm": 0.07575571537017822, "global_step": 47407, "epoch": 447, "lr": 2.7217130522153577e-06} +{"train_loss": 0.015220346860587597, "grad_norm": 0.09310173243284225, "global_step": 47408, "epoch": 447, "lr": 2.7207486332832933e-06} +{"train_loss": 0.01569730043411255, "grad_norm": 0.06789495795965195, "global_step": 47409, "epoch": 447, "lr": 2.7197843804699587e-06} +{"train_loss": 0.015520796179771423, "grad_norm": 0.13151513040065765, "global_step": 47410, "epoch": 447, "lr": 2.7188202937787345e-06} +{"train_loss": 0.016228608787059784, "grad_norm": 0.1113763302564621, "global_step": 47411, "epoch": 447, "lr": 2.717856373213018e-06} +{"train_loss": 0.012629600241780281, "grad_norm": 0.08249720185995102, "global_step": 47412, "epoch": 447, "lr": 2.7168926187761955e-06} +{"train_loss": 0.015979159623384476, "grad_norm": 0.08805778622627258, "global_step": 47413, "epoch": 447, "lr": 2.7159290304716476e-06} +{"train_loss": 0.01657092198729515, "grad_norm": 0.08039012551307678, "global_step": 47414, "epoch": 447, "lr": 2.7149656083027554e-06} +{"train_loss": 0.015842892229557037, "grad_norm": 0.07295145839452744, "global_step": 47415, "epoch": 447, "lr": 2.7140023522729096e-06} +{"train_loss": 0.012970434501767159, "grad_norm": 0.05546814203262329, "global_step": 47416, "epoch": 447, "lr": 2.7130392623855084e-06} +{"train_loss": 0.01557699590921402, "grad_norm": 0.09280657023191452, "global_step": 47417, "epoch": 447, "lr": 2.7120763386439096e-06} +{"train_loss": 0.012984761968255043, "grad_norm": 0.07151743769645691, "global_step": 47418, "epoch": 447, "lr": 2.711113581051511e-06} +{"train_loss": 0.01571817137300968, "grad_norm": 0.07886523753404617, "global_step": 47419, "epoch": 447, "lr": 2.7101509896117093e-06} +{"train_loss": 0.01592463068664074, "grad_norm": 0.07219118624925613, "global_step": 47420, "epoch": 447, "lr": 2.709188564327858e-06} +{"train_loss": 0.017183881253004074, "grad_norm": 0.10489380359649658, "global_step": 47421, "epoch": 447, "lr": 2.7082263052033487e-06} +{"train_loss": 0.016369305551052094, "grad_norm": 0.07643081247806549, "global_step": 47422, "epoch": 447, "lr": 2.7072642122415726e-06} +{"train_loss": 0.016568973660469055, "grad_norm": 0.08440302312374115, "global_step": 47423, "epoch": 447, "lr": 2.7063022854458997e-06} +{"train_loss": 0.016214944422245026, "grad_norm": 0.07247420400381088, "global_step": 47424, "epoch": 447, "lr": 2.7053405248197104e-06} +{"train_loss": 0.014613734558224678, "grad_norm": 0.08174484968185425, "global_step": 47425, "epoch": 447, "lr": 2.704378930366391e-06} +{"train_loss": 0.011451814323663712, "grad_norm": 0.09668837487697601, "global_step": 47426, "epoch": 447, "lr": 2.703417502089317e-06} +{"train_loss": 0.015226787887513638, "grad_norm": 0.09427682310342789, "global_step": 47427, "epoch": 447, "lr": 2.702456239991863e-06} +{"train_loss": 0.015463868156075478, "grad_norm": 0.1011442244052887, "global_step": 47428, "epoch": 447, "lr": 2.701495144077404e-06} +{"train_loss": 0.015741650015115738, "grad_norm": 0.06495285779237747, "global_step": 47429, "epoch": 447, "lr": 2.7005342143493317e-06} +{"train_loss": 0.015392628498375416, "grad_norm": 0.17399780452251434, "global_step": 47430, "epoch": 447, "lr": 2.6995734508109993e-06} +{"train_loss": 0.013842303305864334, "grad_norm": 0.10248459130525589, "global_step": 47431, "epoch": 447, "lr": 2.6986128534658095e-06} +{"train_loss": 0.014165655709803104, "grad_norm": 0.06607471406459808, "global_step": 47432, "epoch": 447, "lr": 2.6976524223171097e-06} +{"train_loss": 0.015125165693461895, "grad_norm": 0.054432328790426254, "global_step": 47433, "epoch": 447, "lr": 2.6966921573683024e-06} +{"train_loss": 0.014672012068331242, "grad_norm": 0.09693850576877594, "global_step": 47434, "epoch": 447, "lr": 2.695732058622735e-06} +{"train_loss": 0.015352502465248108, "grad_norm": 0.06514142453670502, "global_step": 47435, "epoch": 447, "lr": 2.6947721260837943e-06} +{"train_loss": 0.013661419972777367, "grad_norm": 0.10873083025217056, "global_step": 47436, "epoch": 447, "lr": 2.6938123597548605e-06} +{"train_loss": 0.014635806903243065, "grad_norm": 0.11392123252153397, "global_step": 47437, "epoch": 447, "lr": 2.6928527596392917e-06} +{"train_loss": 0.013749338686466217, "grad_norm": 0.10913991183042526, "global_step": 47438, "epoch": 447, "lr": 2.6918933257404633e-06} +{"train_loss": 0.015264836139976978, "grad_norm": 0.08114147186279297, "global_step": 47439, "epoch": 447, "lr": 2.690934058061756e-06} +{"train_loss": 0.015102934092283249, "grad_norm": 0.06732841581106186, "global_step": 47440, "epoch": 447, "lr": 2.6899749566065225e-06} +{"train_loss": 0.013605637475848198, "grad_norm": 0.056731052696704865, "global_step": 47441, "epoch": 447, "lr": 2.689016021378138e-06} +{"train_loss": 0.014767254702746868, "grad_norm": 0.1183583214879036, "global_step": 47442, "epoch": 447, "lr": 2.6880572523799884e-06} +{"train_loss": 0.014300830662250519, "grad_norm": 0.07591388374567032, "global_step": 47443, "epoch": 447, "lr": 2.6870986496154327e-06} +{"train_loss": 0.014280011877417564, "grad_norm": 0.14759166538715363, "global_step": 47444, "epoch": 447, "lr": 2.6861402130878234e-06} +{"train_loss": 0.016766496002674103, "grad_norm": 0.1014164611697197, "global_step": 47445, "epoch": 447, "lr": 2.6851819428005466e-06} +{"train_loss": 0.01697527803480625, "grad_norm": 0.09050021320581436, "global_step": 47446, "epoch": 447, "lr": 2.684223838756966e-06} +{"train_loss": 0.014011463150382042, "grad_norm": 0.10635402798652649, "global_step": 47447, "epoch": 447, "lr": 2.683265900960441e-06} +{"train_loss": 0.011392883956432343, "grad_norm": 0.07395505905151367, "global_step": 47448, "epoch": 447, "lr": 2.6823081294143514e-06} +{"train_loss": 0.015201311558485031, "grad_norm": 0.06910598278045654, "global_step": 47449, "epoch": 447, "lr": 2.6813505241220447e-06} +{"train_loss": 0.01775284670293331, "grad_norm": 0.0728265717625618, "global_step": 47450, "epoch": 447, "lr": 2.6803930850869018e-06} +{"train_loss": 0.014407425187528133, "grad_norm": 0.06765177845954895, "global_step": 47451, "epoch": 447, "lr": 2.6794358123122697e-06} +{"train_loss": 0.017796192318201065, "grad_norm": 0.09436461329460144, "global_step": 47452, "epoch": 447, "lr": 2.678478705801518e-06} +{"train_loss": 0.01591724343597889, "grad_norm": 0.07124015688896179, "global_step": 47453, "epoch": 447, "lr": 2.6775217655580277e-06} +{"train_loss": 0.015792347490787506, "grad_norm": 0.0604156032204628, "global_step": 47454, "epoch": 447, "lr": 2.6765649915851398e-06} +{"train_loss": 0.014087874442338943, "grad_norm": 0.1525125503540039, "global_step": 47455, "epoch": 447, "lr": 2.675608383886213e-06} +{"train_loss": 0.015262724831700325, "grad_norm": 0.10270471125841141, "global_step": 47456, "epoch": 447, "lr": 2.6746519424646233e-06} +{"train_loss": 0.013797570019960403, "grad_norm": 0.07917388528585434, "global_step": 47457, "epoch": 447, "lr": 2.673695667323728e-06} +{"train_loss": 0.013842624612152576, "grad_norm": 0.05795136094093323, "global_step": 47458, "epoch": 447, "lr": 2.6727395584668803e-06} +{"train_loss": 0.015213435515761375, "grad_norm": 0.09298849105834961, "global_step": 47459, "epoch": 447, "lr": 2.671783615897444e-06} +{"train_loss": 0.016421005129814148, "grad_norm": 0.06988519430160522, "global_step": 47460, "epoch": 447, "lr": 2.670827839618789e-06} +{"train_loss": 0.016187310218811035, "grad_norm": 0.08336514979600906, "global_step": 47461, "epoch": 447, "lr": 2.6698722296342515e-06} +{"train_loss": 0.013596853241324425, "grad_norm": 0.13232046365737915, "global_step": 47462, "epoch": 447, "lr": 2.6689167859472007e-06} +{"train_loss": 0.01334189809858799, "grad_norm": 0.08590495586395264, "global_step": 47463, "epoch": 447, "lr": 2.6679615085609957e-06} +{"train_loss": 0.012872220948338509, "grad_norm": 0.06510581821203232, "global_step": 47464, "epoch": 447, "lr": 2.6670063974789826e-06} +{"train_loss": 0.01642989180982113, "grad_norm": 0.07436317950487137, "global_step": 47465, "epoch": 447, "lr": 2.6660514527045267e-06} +{"train_loss": 0.014497417956590652, "grad_norm": 0.08120173215866089, "global_step": 47466, "epoch": 447, "lr": 2.6650966742409857e-06} +{"train_loss": 0.01741141267120838, "grad_norm": 0.09040389955043793, "global_step": 47467, "epoch": 447, "lr": 2.6641420620917124e-06} +{"train_loss": 0.01592196151614189, "grad_norm": 0.07660277187824249, "global_step": 47468, "epoch": 447, "lr": 2.663187616260049e-06} +{"train_loss": 0.014671946875751019, "grad_norm": 0.12051571160554886, "global_step": 47469, "epoch": 447, "lr": 2.662233336749359e-06} +{"train_loss": 0.0157647542655468, "grad_norm": 0.11730595678091049, "global_step": 47470, "epoch": 447, "lr": 2.661279223563001e-06} +{"train_loss": 0.014673720113933086, "grad_norm": 0.0672653317451477, "global_step": 47471, "epoch": 447, "lr": 2.6603252767043117e-06} +{"train_loss": 0.01392359845340252, "grad_norm": 0.06660521030426025, "global_step": 47472, "epoch": 447, "lr": 2.6593714961766604e-06} +{"train_loss": 0.014172002673149109, "grad_norm": 0.07368015497922897, "global_step": 47473, "epoch": 447, "lr": 2.6584178819833826e-06} +{"train_loss": 0.015751920640468597, "grad_norm": 0.10064209252595901, "global_step": 47474, "epoch": 447, "lr": 2.6574644341278485e-06} +{"train_loss": 0.01236991211771965, "grad_norm": 0.0540955550968647, "global_step": 47475, "epoch": 447, "lr": 2.6565111526133833e-06} +{"train_loss": 0.014290907420217991, "grad_norm": 0.10431062430143356, "global_step": 47476, "epoch": 447, "lr": 2.6555580374433453e-06} +{"train_loss": 0.016446217894554138, "grad_norm": 0.08992455899715424, "global_step": 47477, "epoch": 447, "lr": 2.654605088621098e-06} +{"train_loss": 0.014413641765713692, "grad_norm": 0.07520138472318649, "global_step": 47478, "epoch": 447, "lr": 2.6536523061499783e-06} +{"train_loss": 0.015111686661839485, "grad_norm": 0.06402240693569183, "global_step": 47479, "epoch": 447, "lr": 2.6526996900333277e-06} +{"train_loss": 0.012830211780965328, "grad_norm": 0.06146731227636337, "global_step": 47480, "epoch": 447, "lr": 2.6517472402745105e-06} +{"train_loss": 0.014419738203287125, "grad_norm": 0.1262510120868683, "global_step": 47481, "epoch": 447, "lr": 2.650794956876845e-06} +{"train_loss": 0.01460456196218729, "grad_norm": 0.09096585959196091, "global_step": 47482, "epoch": 447, "lr": 2.6498428398437025e-06} +{"train_loss": 0.014380289241671562, "grad_norm": 0.11054761707782745, "global_step": 47483, "epoch": 447, "lr": 2.6488908891784237e-06} +{"train_loss": 0.01659465581178665, "grad_norm": 0.08004441112279892, "global_step": 47484, "epoch": 447, "lr": 2.6479391048843503e-06} +{"train_loss": 0.013344917446374893, "grad_norm": 0.05748867243528366, "global_step": 47485, "epoch": 447, "lr": 2.6469874869648247e-06} +{"train_loss": 0.013944493606686592, "grad_norm": 0.07370150834321976, "global_step": 47486, "epoch": 447, "lr": 2.646036035423183e-06} +{"train_loss": 0.01496863643691506, "grad_norm": 0.0805397778749466, "global_step": 47487, "epoch": 447, "lr": 2.6450847502627884e-06, "val_loss": 0.020042816177010536} +{"train_loss": 0.014709504321217537, "grad_norm": 0.10489416122436523, "global_step": 47488, "epoch": 448, "lr": 2.644133631486967e-06} +{"train_loss": 0.014347709715366364, "grad_norm": 0.06543759256601334, "global_step": 47489, "epoch": 448, "lr": 2.6431826790990655e-06} +{"train_loss": 0.016709255054593086, "grad_norm": 0.10272262245416641, "global_step": 47490, "epoch": 448, "lr": 2.642231893102426e-06} +{"train_loss": 0.01706538163125515, "grad_norm": 0.07187794148921967, "global_step": 47491, "epoch": 448, "lr": 2.64128127350039e-06} +{"train_loss": 0.01783822476863861, "grad_norm": 0.14160360395908356, "global_step": 47492, "epoch": 448, "lr": 2.640330820296294e-06} +{"train_loss": 0.014128231443464756, "grad_norm": 0.0784463882446289, "global_step": 47493, "epoch": 448, "lr": 2.639380533493474e-06} +{"train_loss": 0.015723472461104393, "grad_norm": 0.07396332919597626, "global_step": 47494, "epoch": 448, "lr": 2.638430413095283e-06} +{"train_loss": 0.017318420112133026, "grad_norm": 0.0976141095161438, "global_step": 47495, "epoch": 448, "lr": 2.6374804591050517e-06} +{"train_loss": 0.01534200832247734, "grad_norm": 0.1996902972459793, "global_step": 47496, "epoch": 448, "lr": 2.636530671526105e-06} +{"train_loss": 0.01698809303343296, "grad_norm": 0.09136148542165756, "global_step": 47497, "epoch": 448, "lr": 2.635581050361796e-06} +{"train_loss": 0.013153167441487312, "grad_norm": 0.103342704474926, "global_step": 47498, "epoch": 448, "lr": 2.6346315956154665e-06} +{"train_loss": 0.015461411327123642, "grad_norm": 0.06918652355670929, "global_step": 47499, "epoch": 448, "lr": 2.6336823072904304e-06} +{"train_loss": 0.0150456503033638, "grad_norm": 0.10135631263256073, "global_step": 47500, "epoch": 448, "lr": 2.632733185390035e-06} +{"train_loss": 0.015336476266384125, "grad_norm": 0.07106080651283264, "global_step": 47501, "epoch": 448, "lr": 2.6317842299176334e-06} +{"train_loss": 0.01309218443930149, "grad_norm": 0.09861753135919571, "global_step": 47502, "epoch": 448, "lr": 2.630835440876528e-06} +{"train_loss": 0.016233723610639572, "grad_norm": 0.05198675021529198, "global_step": 47503, "epoch": 448, "lr": 2.6298868182700666e-06} +{"train_loss": 0.013375731185078621, "grad_norm": 0.09827281534671783, "global_step": 47504, "epoch": 448, "lr": 2.6289383621015906e-06} +{"train_loss": 0.014837326481938362, "grad_norm": 0.12358713150024414, "global_step": 47505, "epoch": 448, "lr": 2.6279900723744145e-06} +{"train_loss": 0.013934696093201637, "grad_norm": 0.07439319044351578, "global_step": 47506, "epoch": 448, "lr": 2.62704194909188e-06} +{"train_loss": 0.013666544109582901, "grad_norm": 0.0920945554971695, "global_step": 47507, "epoch": 448, "lr": 2.6260939922573226e-06} +{"train_loss": 0.015487195923924446, "grad_norm": 0.08276604861021042, "global_step": 47508, "epoch": 448, "lr": 2.6251462018740735e-06} +{"train_loss": 0.012107981368899345, "grad_norm": 0.0719841718673706, "global_step": 47509, "epoch": 448, "lr": 2.624198577945447e-06} +{"train_loss": 0.013522264547646046, "grad_norm": 0.07088205963373184, "global_step": 47510, "epoch": 448, "lr": 2.6232511204747845e-06} +{"train_loss": 0.014207672327756882, "grad_norm": 0.09677845239639282, "global_step": 47511, "epoch": 448, "lr": 2.6223038294654223e-06} +{"train_loss": 0.014093324542045593, "grad_norm": 0.11037331074476242, "global_step": 47512, "epoch": 448, "lr": 2.6213567049206745e-06} +{"train_loss": 0.01455176156014204, "grad_norm": 0.09482639282941818, "global_step": 47513, "epoch": 448, "lr": 2.6204097468438772e-06} +{"train_loss": 0.013851329684257507, "grad_norm": 0.10016390681266785, "global_step": 47514, "epoch": 448, "lr": 2.6194629552383556e-06} +{"train_loss": 0.015024317428469658, "grad_norm": 0.0903698280453682, "global_step": 47515, "epoch": 448, "lr": 2.6185163301074344e-06} +{"train_loss": 0.014414720237255096, "grad_norm": 0.06498479098081589, "global_step": 47516, "epoch": 448, "lr": 2.6175698714544393e-06} +{"train_loss": 0.01417633704841137, "grad_norm": 0.12903665006160736, "global_step": 47517, "epoch": 448, "lr": 2.6166235792826953e-06} +{"train_loss": 0.014341937378048897, "grad_norm": 0.12247000634670258, "global_step": 47518, "epoch": 448, "lr": 2.6156774535955388e-06} +{"train_loss": 0.014520708471536636, "grad_norm": 0.07830735296010971, "global_step": 47519, "epoch": 448, "lr": 2.614731494396283e-06} +{"train_loss": 0.014060338959097862, "grad_norm": 0.05651659145951271, "global_step": 47520, "epoch": 448, "lr": 2.613785701688248e-06} +{"train_loss": 0.014624650590121746, "grad_norm": 0.09971436113119125, "global_step": 47521, "epoch": 448, "lr": 2.6128400754747704e-06} +{"train_loss": 0.016856443136930466, "grad_norm": 0.07355867326259613, "global_step": 47522, "epoch": 448, "lr": 2.611894615759153e-06} +{"train_loss": 0.01651664450764656, "grad_norm": 0.061210937798023224, "global_step": 47523, "epoch": 448, "lr": 2.6109493225447314e-06} +{"train_loss": 0.01419040560722351, "grad_norm": 0.08382822573184967, "global_step": 47524, "epoch": 448, "lr": 2.610004195834831e-06} +{"train_loss": 0.013569895178079605, "grad_norm": 0.09030499309301376, "global_step": 47525, "epoch": 448, "lr": 2.6090592356327715e-06} +{"train_loss": 0.016596585512161255, "grad_norm": 0.13040602207183838, "global_step": 47526, "epoch": 448, "lr": 2.6081144419418556e-06} +{"train_loss": 0.014221464283764362, "grad_norm": 0.07965092360973358, "global_step": 47527, "epoch": 448, "lr": 2.60716981476542e-06} +{"train_loss": 0.013401767238974571, "grad_norm": 0.06662376970052719, "global_step": 47528, "epoch": 448, "lr": 2.6062253541067837e-06} +{"train_loss": 0.013669680804014206, "grad_norm": 0.08936365693807602, "global_step": 47529, "epoch": 448, "lr": 2.60528105996925e-06} +{"train_loss": 0.014532716944813728, "grad_norm": 0.05084198713302612, "global_step": 47530, "epoch": 448, "lr": 2.6043369323561493e-06} +{"train_loss": 0.014651387929916382, "grad_norm": 0.06797317415475845, "global_step": 47531, "epoch": 448, "lr": 2.603392971270807e-06} +{"train_loss": 0.015547234565019608, "grad_norm": 0.11608928442001343, "global_step": 47532, "epoch": 448, "lr": 2.602449176716526e-06} +{"train_loss": 0.013949351385235786, "grad_norm": 0.056856561452150345, "global_step": 47533, "epoch": 448, "lr": 2.60150554869662e-06} +{"train_loss": 0.012749634683132172, "grad_norm": 0.07991206645965576, "global_step": 47534, "epoch": 448, "lr": 2.6005620872144145e-06} +{"train_loss": 0.019128769636154175, "grad_norm": 0.11163330078125, "global_step": 47535, "epoch": 448, "lr": 2.599618792273223e-06} +{"train_loss": 0.014189529232680798, "grad_norm": 0.06103530526161194, "global_step": 47536, "epoch": 448, "lr": 2.598675663876349e-06} +{"train_loss": 0.015098724514245987, "grad_norm": 0.08445911109447479, "global_step": 47537, "epoch": 448, "lr": 2.5977327020271224e-06} +{"train_loss": 0.013445200398564339, "grad_norm": 0.07941887527704239, "global_step": 47538, "epoch": 448, "lr": 2.5967899067288527e-06} +{"train_loss": 0.01383441872894764, "grad_norm": 0.04581276699900627, "global_step": 47539, "epoch": 448, "lr": 2.5958472779848364e-06} +{"train_loss": 0.011827594600617886, "grad_norm": 0.11255062371492386, "global_step": 47540, "epoch": 448, "lr": 2.594904815798399e-06} +{"train_loss": 0.013049285858869553, "grad_norm": 0.07847806066274643, "global_step": 47541, "epoch": 448, "lr": 2.5939625201728545e-06} +{"train_loss": 0.016011135652661324, "grad_norm": 0.11392664909362793, "global_step": 47542, "epoch": 448, "lr": 2.593020391111517e-06} +{"train_loss": 0.015562048181891441, "grad_norm": 0.10497859120368958, "global_step": 47543, "epoch": 448, "lr": 2.5920784286176723e-06} +{"train_loss": 0.014677280560135841, "grad_norm": 0.06914398074150085, "global_step": 47544, "epoch": 448, "lr": 2.5911366326946517e-06} +{"train_loss": 0.01695443131029606, "grad_norm": 0.07146088033914566, "global_step": 47545, "epoch": 448, "lr": 2.590195003345769e-06} +{"train_loss": 0.015226610004901886, "grad_norm": 0.09152303636074066, "global_step": 47546, "epoch": 448, "lr": 2.5892535405743103e-06} +{"train_loss": 0.013454688712954521, "grad_norm": 0.06664086133241653, "global_step": 47547, "epoch": 448, "lr": 2.5883122443836007e-06} +{"train_loss": 0.014555178582668304, "grad_norm": 0.05797949060797691, "global_step": 47548, "epoch": 448, "lr": 2.5873711147769486e-06} +{"train_loss": 0.014837921597063541, "grad_norm": 0.06539212167263031, "global_step": 47549, "epoch": 448, "lr": 2.586430151757657e-06} +{"train_loss": 0.014279952272772789, "grad_norm": 0.12349098920822144, "global_step": 47550, "epoch": 448, "lr": 2.585489355329018e-06} +{"train_loss": 0.014921383932232857, "grad_norm": 0.06506944447755814, "global_step": 47551, "epoch": 448, "lr": 2.5845487254943513e-06} +{"train_loss": 0.016102522611618042, "grad_norm": 0.08949105441570282, "global_step": 47552, "epoch": 448, "lr": 2.58360826225697e-06} +{"train_loss": 0.014353582635521889, "grad_norm": 0.08477582037448883, "global_step": 47553, "epoch": 448, "lr": 2.5826679656201615e-06} +{"train_loss": 0.016495469957590103, "grad_norm": 0.1195531040430069, "global_step": 47554, "epoch": 448, "lr": 2.5817278355872387e-06} +{"train_loss": 0.014027529396116734, "grad_norm": 0.10185343772172928, "global_step": 47555, "epoch": 448, "lr": 2.5807878721615e-06} +{"train_loss": 0.014561080373823643, "grad_norm": 0.07611174881458282, "global_step": 47556, "epoch": 448, "lr": 2.5798480753462585e-06} +{"train_loss": 0.014826824888586998, "grad_norm": 0.08210611343383789, "global_step": 47557, "epoch": 448, "lr": 2.5789084451447952e-06} +{"train_loss": 0.013421121053397655, "grad_norm": 0.07115333527326584, "global_step": 47558, "epoch": 448, "lr": 2.57796898156043e-06} +{"train_loss": 0.014379876665771008, "grad_norm": 0.08877868950366974, "global_step": 47559, "epoch": 448, "lr": 2.577029684596466e-06} +{"train_loss": 0.01599080301821232, "grad_norm": 0.06544145196676254, "global_step": 47560, "epoch": 448, "lr": 2.576090554256194e-06} +{"train_loss": 0.012626959942281246, "grad_norm": 0.08878574520349503, "global_step": 47561, "epoch": 448, "lr": 2.575151590542907e-06} +{"train_loss": 0.016583988443017006, "grad_norm": 0.07997237145900726, "global_step": 47562, "epoch": 448, "lr": 2.5742127934599236e-06} +{"train_loss": 0.016067078337073326, "grad_norm": 0.06516910344362259, "global_step": 47563, "epoch": 448, "lr": 2.5732741630105197e-06} +{"train_loss": 0.015917565673589706, "grad_norm": 0.06719382852315903, "global_step": 47564, "epoch": 448, "lr": 2.5723356991980087e-06} +{"train_loss": 0.015102459117770195, "grad_norm": 0.07341179996728897, "global_step": 47565, "epoch": 448, "lr": 2.5713974020256936e-06} +{"train_loss": 0.01712767407298088, "grad_norm": 0.0868208184838295, "global_step": 47566, "epoch": 448, "lr": 2.5704592714968555e-06} +{"train_loss": 0.01588955521583557, "grad_norm": 0.09165481477975845, "global_step": 47567, "epoch": 448, "lr": 2.5695213076147973e-06} +{"train_loss": 0.015942499041557312, "grad_norm": 0.08171705901622772, "global_step": 47568, "epoch": 448, "lr": 2.5685835103828106e-06} +{"train_loss": 0.01539582759141922, "grad_norm": 0.09754478186368942, "global_step": 47569, "epoch": 448, "lr": 2.567645879804198e-06} +{"train_loss": 0.014474894851446152, "grad_norm": 0.07399071007966995, "global_step": 47570, "epoch": 448, "lr": 2.5667084158822463e-06} +{"train_loss": 0.015739236027002335, "grad_norm": 0.08746074140071869, "global_step": 47571, "epoch": 448, "lr": 2.5657711186202527e-06} +{"train_loss": 0.01709483191370964, "grad_norm": 0.06943891197443008, "global_step": 47572, "epoch": 448, "lr": 2.5648339880215143e-06} +{"train_loss": 0.013372976332902908, "grad_norm": 0.06290856003761292, "global_step": 47573, "epoch": 448, "lr": 2.563897024089329e-06} +{"train_loss": 0.015782775357365608, "grad_norm": 0.05907314270734787, "global_step": 47574, "epoch": 448, "lr": 2.562960226826966e-06} +{"train_loss": 0.017651796340942383, "grad_norm": 0.09761228412389755, "global_step": 47575, "epoch": 448, "lr": 2.5620235962377336e-06} +{"train_loss": 0.017697077244520187, "grad_norm": 0.08393540233373642, "global_step": 47576, "epoch": 448, "lr": 2.5610871323249242e-06} +{"train_loss": 0.014859938994050026, "grad_norm": 0.06848273426294327, "global_step": 47577, "epoch": 448, "lr": 2.560150835091818e-06} +{"train_loss": 0.016069840639829636, "grad_norm": 0.08332563191652298, "global_step": 47578, "epoch": 448, "lr": 2.559214704541718e-06} +{"train_loss": 0.01454601064324379, "grad_norm": 0.09454122930765152, "global_step": 47579, "epoch": 448, "lr": 2.558278740677905e-06} +{"train_loss": 0.016412099823355675, "grad_norm": 0.09659180790185928, "global_step": 47580, "epoch": 448, "lr": 2.5573429435036654e-06} +{"train_loss": 0.013069480657577515, "grad_norm": 0.10955289006233215, "global_step": 47581, "epoch": 448, "lr": 2.556407313022291e-06} +{"train_loss": 0.014482482336461544, "grad_norm": 0.06484384834766388, "global_step": 47582, "epoch": 448, "lr": 2.5554718492370733e-06} +{"train_loss": 0.015884118154644966, "grad_norm": 0.11139468103647232, "global_step": 47583, "epoch": 448, "lr": 2.554536552151282e-06} +{"train_loss": 0.01726679503917694, "grad_norm": 0.14134293794631958, "global_step": 47584, "epoch": 448, "lr": 2.5536014217682313e-06} +{"train_loss": 0.014535375870764256, "grad_norm": 0.07001787424087524, "global_step": 47585, "epoch": 448, "lr": 2.5526664580911795e-06} +{"train_loss": 0.01698753423988819, "grad_norm": 0.08882100135087967, "global_step": 47586, "epoch": 448, "lr": 2.5517316611234297e-06} +{"train_loss": 0.013204868882894516, "grad_norm": 0.07463531941175461, "global_step": 47587, "epoch": 448, "lr": 2.5507970308682573e-06} +{"train_loss": 0.014098463580012321, "grad_norm": 0.05903565511107445, "global_step": 47588, "epoch": 448, "lr": 2.549862567328948e-06} +{"train_loss": 0.01566205359995365, "grad_norm": 0.07187793403863907, "global_step": 47589, "epoch": 448, "lr": 2.5489282705087938e-06} +{"train_loss": 0.016187656670808792, "grad_norm": 0.11692295968532562, "global_step": 47590, "epoch": 448, "lr": 2.547994140411064e-06} +{"train_loss": 0.013390075415372849, "grad_norm": 0.0983869880437851, "global_step": 47591, "epoch": 448, "lr": 2.5470601770390457e-06} +{"train_loss": 0.014787895604968071, "grad_norm": 0.08198552578687668, "global_step": 47592, "epoch": 448, "lr": 2.546126380396019e-06} +{"train_loss": 0.014941222192543858, "grad_norm": 0.08667400479316711, "global_step": 47593, "epoch": 448, "lr": 2.5451927504852757e-06, "val_loss": 0.019882382825016975} +{"train_loss": 0.01515927817672491, "grad_norm": 0.07672856748104095, "global_step": 47594, "epoch": 449, "lr": 2.5442592873100744e-06} +{"train_loss": 0.011829626746475697, "grad_norm": 0.11401601880788803, "global_step": 47595, "epoch": 449, "lr": 2.543325990873724e-06} +{"train_loss": 0.014549256302416325, "grad_norm": 0.09950307011604309, "global_step": 47596, "epoch": 449, "lr": 2.5423928611794823e-06} +{"train_loss": 0.01574329100549221, "grad_norm": 0.05513995513319969, "global_step": 47597, "epoch": 449, "lr": 2.5414598982306248e-06} +{"train_loss": 0.013197831809520721, "grad_norm": 0.0754118412733078, "global_step": 47598, "epoch": 449, "lr": 2.540527102030443e-06} +{"train_loss": 0.01590900681912899, "grad_norm": 0.08304336667060852, "global_step": 47599, "epoch": 449, "lr": 2.539594472582213e-06} +{"train_loss": 0.01248860266059637, "grad_norm": 0.08511212468147278, "global_step": 47600, "epoch": 449, "lr": 2.5386620098892034e-06} +{"train_loss": 0.015030018985271454, "grad_norm": 0.08462853729724884, "global_step": 47601, "epoch": 449, "lr": 2.5377297139547005e-06} +{"train_loss": 0.016727786511182785, "grad_norm": 0.11281470209360123, "global_step": 47602, "epoch": 449, "lr": 2.5367975847819636e-06} +{"train_loss": 0.017216909676790237, "grad_norm": 0.07926372438669205, "global_step": 47603, "epoch": 449, "lr": 2.535865622374289e-06} +{"train_loss": 0.014977553859353065, "grad_norm": 0.13300225138664246, "global_step": 47604, "epoch": 449, "lr": 2.5349338267349364e-06} +{"train_loss": 0.013697313144803047, "grad_norm": 0.19000715017318726, "global_step": 47605, "epoch": 449, "lr": 2.5340021978671856e-06} +{"train_loss": 0.01571555808186531, "grad_norm": 0.0727994441986084, "global_step": 47606, "epoch": 449, "lr": 2.5330707357743124e-06} +{"train_loss": 0.0157725028693676, "grad_norm": 0.0906400978565216, "global_step": 47607, "epoch": 449, "lr": 2.5321394404595855e-06} +{"train_loss": 0.014692872762680054, "grad_norm": 0.15060633420944214, "global_step": 47608, "epoch": 449, "lr": 2.5312083119262697e-06} +{"train_loss": 0.013685295358300209, "grad_norm": 0.08483358472585678, "global_step": 47609, "epoch": 449, "lr": 2.5302773501776456e-06} +{"train_loss": 0.013123612850904465, "grad_norm": 0.058090221136808395, "global_step": 47610, "epoch": 449, "lr": 2.5293465552169883e-06} +{"train_loss": 0.014664176851511002, "grad_norm": 0.06446263194084167, "global_step": 47611, "epoch": 449, "lr": 2.5284159270475614e-06} +{"train_loss": 0.014889889396727085, "grad_norm": 0.06866440922021866, "global_step": 47612, "epoch": 449, "lr": 2.5274854656726298e-06} +{"train_loss": 0.013568629510700703, "grad_norm": 0.0547291599214077, "global_step": 47613, "epoch": 449, "lr": 2.5265551710954795e-06} +{"train_loss": 0.017257118597626686, "grad_norm": 0.07964815199375153, "global_step": 47614, "epoch": 449, "lr": 2.5256250433193686e-06} +{"train_loss": 0.015423443168401718, "grad_norm": 0.07366752624511719, "global_step": 47615, "epoch": 449, "lr": 2.5246950823475557e-06} +{"train_loss": 0.014718690887093544, "grad_norm": 0.09564178436994553, "global_step": 47616, "epoch": 449, "lr": 2.5237652881833163e-06} +{"train_loss": 0.015033159404993057, "grad_norm": 0.077692411839962, "global_step": 47617, "epoch": 449, "lr": 2.522835660829931e-06} +{"train_loss": 0.016726359724998474, "grad_norm": 0.07599113136529922, "global_step": 47618, "epoch": 449, "lr": 2.5219062002906466e-06} +{"train_loss": 0.015284115448594093, "grad_norm": 0.08802825212478638, "global_step": 47619, "epoch": 449, "lr": 2.5209769065687394e-06} +{"train_loss": 0.015017542988061905, "grad_norm": 0.09169705212116241, "global_step": 47620, "epoch": 449, "lr": 2.5200477796674728e-06} +{"train_loss": 0.013915104791522026, "grad_norm": 0.11137653887271881, "global_step": 47621, "epoch": 449, "lr": 2.519118819590105e-06} +{"train_loss": 0.01565423794090748, "grad_norm": 0.06666620075702667, "global_step": 47622, "epoch": 449, "lr": 2.5181900263399007e-06} +{"train_loss": 0.013474729843437672, "grad_norm": 0.052659980952739716, "global_step": 47623, "epoch": 449, "lr": 2.5172613999201343e-06} +{"train_loss": 0.014325717464089394, "grad_norm": 0.06433995813131332, "global_step": 47624, "epoch": 449, "lr": 2.5163329403340596e-06} +{"train_loss": 0.015028937719762325, "grad_norm": 0.09476266801357269, "global_step": 47625, "epoch": 449, "lr": 2.515404647584946e-06} +{"train_loss": 0.012769349850714207, "grad_norm": 0.09834104031324387, "global_step": 47626, "epoch": 449, "lr": 2.5144765216760404e-06} +{"train_loss": 0.013582861050963402, "grad_norm": 0.04931911081075668, "global_step": 47627, "epoch": 449, "lr": 2.513548562610624e-06} +{"train_loss": 0.013863470405340195, "grad_norm": 0.0812792107462883, "global_step": 47628, "epoch": 449, "lr": 2.512620770391938e-06} +{"train_loss": 0.016634542495012283, "grad_norm": 0.07942551374435425, "global_step": 47629, "epoch": 449, "lr": 2.5116931450232473e-06} +{"train_loss": 0.017742479220032692, "grad_norm": 0.11524117738008499, "global_step": 47630, "epoch": 449, "lr": 2.5107656865078267e-06} +{"train_loss": 0.014678853563964367, "grad_norm": 0.07448076456785202, "global_step": 47631, "epoch": 449, "lr": 2.5098383948489233e-06} +{"train_loss": 0.014449968934059143, "grad_norm": 0.13515980541706085, "global_step": 47632, "epoch": 449, "lr": 2.508911270049785e-06} +{"train_loss": 0.013050675392150879, "grad_norm": 0.06734388321638107, "global_step": 47633, "epoch": 449, "lr": 2.5079843121136814e-06} +{"train_loss": 0.012941837310791016, "grad_norm": 0.05659238249063492, "global_step": 47634, "epoch": 449, "lr": 2.507057521043871e-06} +{"train_loss": 0.014632580801844597, "grad_norm": 0.062587670981884, "global_step": 47635, "epoch": 449, "lr": 2.5061308968436002e-06} +{"train_loss": 0.01355364266782999, "grad_norm": 0.07397468388080597, "global_step": 47636, "epoch": 449, "lr": 2.5052044395161343e-06} +{"train_loss": 0.014704217202961445, "grad_norm": 0.07339071482419968, "global_step": 47637, "epoch": 449, "lr": 2.5042781490647317e-06} +{"train_loss": 0.015934409573674202, "grad_norm": 0.09514377266168594, "global_step": 47638, "epoch": 449, "lr": 2.5033520254926335e-06} +{"train_loss": 0.01576177589595318, "grad_norm": 0.0719667598605156, "global_step": 47639, "epoch": 449, "lr": 2.5024260688030987e-06} +{"train_loss": 0.014226742088794708, "grad_norm": 0.06402892619371414, "global_step": 47640, "epoch": 449, "lr": 2.5015002789993914e-06} +{"train_loss": 0.013659879565238953, "grad_norm": 0.1684027910232544, "global_step": 47641, "epoch": 449, "lr": 2.500574656084742e-06} +{"train_loss": 0.015226510353386402, "grad_norm": 0.11673138290643692, "global_step": 47642, "epoch": 449, "lr": 2.4996492000624318e-06} +{"train_loss": 0.016010195016860962, "grad_norm": 0.07942541688680649, "global_step": 47643, "epoch": 449, "lr": 2.498723910935685e-06} +{"train_loss": 0.015104438178241253, "grad_norm": 0.07004491984844208, "global_step": 47644, "epoch": 449, "lr": 2.4977987887077726e-06} +{"train_loss": 0.0148728396743536, "grad_norm": 0.05576903373003006, "global_step": 47645, "epoch": 449, "lr": 2.4968738333819297e-06} +{"train_loss": 0.013753103092312813, "grad_norm": 0.0830657109618187, "global_step": 47646, "epoch": 449, "lr": 2.4959490449614096e-06} +{"train_loss": 0.0164997149258852, "grad_norm": 0.0773257464170456, "global_step": 47647, "epoch": 449, "lr": 2.4950244234494768e-06} +{"train_loss": 0.01329964492470026, "grad_norm": 0.07480672746896744, "global_step": 47648, "epoch": 449, "lr": 2.4940999688493727e-06} +{"train_loss": 0.01482016034424305, "grad_norm": 0.10678916424512863, "global_step": 47649, "epoch": 449, "lr": 2.493175681164328e-06} +{"train_loss": 0.014712532982230186, "grad_norm": 0.08649057894945145, "global_step": 47650, "epoch": 449, "lr": 2.492251560397607e-06} +{"train_loss": 0.01576315239071846, "grad_norm": 0.07627108693122864, "global_step": 47651, "epoch": 449, "lr": 2.4913276065524626e-06} +{"train_loss": 0.016828831285238266, "grad_norm": 0.10406915098428726, "global_step": 47652, "epoch": 449, "lr": 2.49040381963212e-06} +{"train_loss": 0.01778918318450451, "grad_norm": 0.07583046704530716, "global_step": 47653, "epoch": 449, "lr": 2.4894801996398376e-06} +{"train_loss": 0.014528501778841019, "grad_norm": 0.06546351313591003, "global_step": 47654, "epoch": 449, "lr": 2.4885567465788684e-06} +{"train_loss": 0.014080846682190895, "grad_norm": 0.06339116394519806, "global_step": 47655, "epoch": 449, "lr": 2.4876334604524485e-06} +{"train_loss": 0.013782085850834846, "grad_norm": 0.10104269534349442, "global_step": 47656, "epoch": 449, "lr": 2.4867103412638202e-06} +{"train_loss": 0.017514575272798538, "grad_norm": 0.0722440704703331, "global_step": 47657, "epoch": 449, "lr": 2.4857873890162197e-06} +{"train_loss": 0.01466807909309864, "grad_norm": 0.06734167784452438, "global_step": 47658, "epoch": 449, "lr": 2.484864603712911e-06} +{"train_loss": 0.016942257061600685, "grad_norm": 0.08556868135929108, "global_step": 47659, "epoch": 449, "lr": 2.483941985357119e-06} +{"train_loss": 0.013450193218886852, "grad_norm": 0.07721313089132309, "global_step": 47660, "epoch": 449, "lr": 2.483019533952091e-06} +{"train_loss": 0.01329445093870163, "grad_norm": 0.07155188918113708, "global_step": 47661, "epoch": 449, "lr": 2.482097249501075e-06} +{"train_loss": 0.014994269236922264, "grad_norm": 0.07071582227945328, "global_step": 47662, "epoch": 449, "lr": 2.4811751320072964e-06} +{"train_loss": 0.01613347977399826, "grad_norm": 0.08561201393604279, "global_step": 47663, "epoch": 449, "lr": 2.480253181473996e-06} +{"train_loss": 0.01561975572258234, "grad_norm": 0.0777164176106453, "global_step": 47664, "epoch": 449, "lr": 2.479331397904433e-06} +{"train_loss": 0.013086671940982342, "grad_norm": 0.1209985539317131, "global_step": 47665, "epoch": 449, "lr": 2.478409781301827e-06} +{"train_loss": 0.013776551932096481, "grad_norm": 0.06612806767225266, "global_step": 47666, "epoch": 449, "lr": 2.477488331669425e-06} +{"train_loss": 0.01458085048943758, "grad_norm": 0.08350888639688492, "global_step": 47667, "epoch": 449, "lr": 2.4765670490104587e-06} +{"train_loss": 0.014471519738435745, "grad_norm": 0.06100265681743622, "global_step": 47668, "epoch": 449, "lr": 2.4756459333281746e-06} +{"train_loss": 0.014839304611086845, "grad_norm": 0.06413057446479797, "global_step": 47669, "epoch": 449, "lr": 2.474724984625798e-06} +{"train_loss": 0.014474949799478054, "grad_norm": 0.11756905913352966, "global_step": 47670, "epoch": 449, "lr": 2.4738042029065655e-06} +{"train_loss": 0.012775013223290443, "grad_norm": 0.06549108773469925, "global_step": 47671, "epoch": 449, "lr": 2.4728835881737246e-06} +{"train_loss": 0.01493244618177414, "grad_norm": 0.05542236566543579, "global_step": 47672, "epoch": 449, "lr": 2.4719631404305057e-06} +{"train_loss": 0.015211077407002449, "grad_norm": 0.09486305713653564, "global_step": 47673, "epoch": 449, "lr": 2.4710428596801284e-06} +{"train_loss": 0.015494432300329208, "grad_norm": 0.08740144968032837, "global_step": 47674, "epoch": 449, "lr": 2.4701227459258346e-06} +{"train_loss": 0.016942579299211502, "grad_norm": 0.10296197235584259, "global_step": 47675, "epoch": 449, "lr": 2.4692027991708666e-06} +{"train_loss": 0.014242764562368393, "grad_norm": 0.07700041681528091, "global_step": 47676, "epoch": 449, "lr": 2.4682830194184437e-06} +{"train_loss": 0.01641809567809105, "grad_norm": 0.08408065140247345, "global_step": 47677, "epoch": 449, "lr": 2.4673634066718076e-06} +{"train_loss": 0.01571093499660492, "grad_norm": 0.10246121883392334, "global_step": 47678, "epoch": 449, "lr": 2.4664439609341948e-06} +{"train_loss": 0.015670545399188995, "grad_norm": 0.06997600197792053, "global_step": 47679, "epoch": 449, "lr": 2.465524682208814e-06} +{"train_loss": 0.01384706236422062, "grad_norm": 0.09437184780836105, "global_step": 47680, "epoch": 449, "lr": 2.464605570498907e-06} +{"train_loss": 0.015314599499106407, "grad_norm": 0.09913251549005508, "global_step": 47681, "epoch": 449, "lr": 2.4636866258077042e-06} +{"train_loss": 0.01370220072567463, "grad_norm": 0.09022481739521027, "global_step": 47682, "epoch": 449, "lr": 2.4627678481384363e-06} +{"train_loss": 0.016657687723636627, "grad_norm": 0.1360858976840973, "global_step": 47683, "epoch": 449, "lr": 2.4618492374943234e-06} +{"train_loss": 0.015498632565140724, "grad_norm": 0.07875161617994308, "global_step": 47684, "epoch": 449, "lr": 2.4609307938786076e-06} +{"train_loss": 0.014500342309474945, "grad_norm": 0.05088622868061066, "global_step": 47685, "epoch": 449, "lr": 2.460012517294502e-06} +{"train_loss": 0.016460347920656204, "grad_norm": 0.09943341463804245, "global_step": 47686, "epoch": 449, "lr": 2.4590944077452383e-06} +{"train_loss": 0.014020652510225773, "grad_norm": 0.08920832723379135, "global_step": 47687, "epoch": 449, "lr": 2.4581764652340357e-06} +{"train_loss": 0.015429848805069923, "grad_norm": 0.07689867168664932, "global_step": 47688, "epoch": 449, "lr": 2.4572586897641357e-06} +{"train_loss": 0.013989347033202648, "grad_norm": 0.06168144941329956, "global_step": 47689, "epoch": 449, "lr": 2.4563410813387534e-06} +{"train_loss": 0.01628120057284832, "grad_norm": 0.07128549367189407, "global_step": 47690, "epoch": 449, "lr": 2.4554236399611017e-06} +{"train_loss": 0.013709675520658493, "grad_norm": 0.14213818311691284, "global_step": 47691, "epoch": 449, "lr": 2.454506365634418e-06} +{"train_loss": 0.017151325941085815, "grad_norm": 0.1089659109711647, "global_step": 47692, "epoch": 449, "lr": 2.453589258361927e-06} +{"train_loss": 0.014292299747467041, "grad_norm": 0.08962737023830414, "global_step": 47693, "epoch": 449, "lr": 2.4526723181468425e-06} +{"train_loss": 0.015232536941766739, "grad_norm": 0.11600406467914581, "global_step": 47694, "epoch": 449, "lr": 2.4517555449923847e-06} +{"train_loss": 0.01723892241716385, "grad_norm": 0.142134428024292, "global_step": 47695, "epoch": 449, "lr": 2.450838938901795e-06} +{"train_loss": 0.015440301969647408, "grad_norm": 0.08513665944337845, "global_step": 47696, "epoch": 449, "lr": 2.4499224998782656e-06} +{"train_loss": 0.016943350434303284, "grad_norm": 0.0676388144493103, "global_step": 47697, "epoch": 449, "lr": 2.4490062279250324e-06} +{"train_loss": 0.014105936512351036, "grad_norm": 0.10302188992500305, "global_step": 47698, "epoch": 449, "lr": 2.4480901230453156e-06} +{"train_loss": 0.01497017734525901, "grad_norm": 0.14567139744758606, "global_step": 47699, "epoch": 449, "lr": 2.4471741852423237e-06, "val_loss": 0.019452018663287163} +{"train_loss": 0.01655326783657074, "grad_norm": 0.0653475672006607, "global_step": 47700, "epoch": 450, "lr": 2.446258414519281e-06} +{"train_loss": 0.01613686978816986, "grad_norm": 0.10533392429351807, "global_step": 47701, "epoch": 450, "lr": 2.445342810879414e-06} +{"train_loss": 0.014571420848369598, "grad_norm": 0.08899109810590744, "global_step": 47702, "epoch": 450, "lr": 2.4444273743259305e-06} +{"train_loss": 0.014118017628788948, "grad_norm": 0.05966378375887871, "global_step": 47703, "epoch": 450, "lr": 2.4435121048620393e-06} +{"train_loss": 0.01304610911756754, "grad_norm": 0.1111844852566719, "global_step": 47704, "epoch": 450, "lr": 2.4425970024909704e-06} +{"train_loss": 0.01334120612591505, "grad_norm": 0.06258756667375565, "global_step": 47705, "epoch": 450, "lr": 2.4416820672159334e-06} +{"train_loss": 0.017652543261647224, "grad_norm": 0.10498609393835068, "global_step": 47706, "epoch": 450, "lr": 2.440767299040142e-06} +{"train_loss": 0.015345167368650436, "grad_norm": 0.0922781303524971, "global_step": 47707, "epoch": 450, "lr": 2.4398526979668157e-06} +{"train_loss": 0.012851862236857414, "grad_norm": 0.07309050858020782, "global_step": 47708, "epoch": 450, "lr": 2.4389382639991576e-06} +{"train_loss": 0.014103958383202553, "grad_norm": 0.10177876055240631, "global_step": 47709, "epoch": 450, "lr": 2.438023997140393e-06} +{"train_loss": 0.014588998630642891, "grad_norm": 0.09066581726074219, "global_step": 47710, "epoch": 450, "lr": 2.437109897393719e-06} +{"train_loss": 0.015934409573674202, "grad_norm": 0.11247341334819794, "global_step": 47711, "epoch": 450, "lr": 2.4361959647623612e-06} +{"train_loss": 0.01473439671099186, "grad_norm": 0.08994559198617935, "global_step": 47712, "epoch": 450, "lr": 2.435282199249528e-06} +{"train_loss": 0.016343504190444946, "grad_norm": 0.07235413044691086, "global_step": 47713, "epoch": 450, "lr": 2.4343686008584276e-06} +{"train_loss": 0.016677087172865868, "grad_norm": 0.10106490552425385, "global_step": 47714, "epoch": 450, "lr": 2.4334551695922636e-06} +{"train_loss": 0.013078748248517513, "grad_norm": 0.052870310842990875, "global_step": 47715, "epoch": 450, "lr": 2.432541905454255e-06} +{"train_loss": 0.019314322620630264, "grad_norm": 0.07805139571428299, "global_step": 47716, "epoch": 450, "lr": 2.4316288084476112e-06} +{"train_loss": 0.013932551257312298, "grad_norm": 0.07995672523975372, "global_step": 47717, "epoch": 450, "lr": 2.430715878575535e-06} +{"train_loss": 0.016120724380016327, "grad_norm": 0.08193210512399673, "global_step": 47718, "epoch": 450, "lr": 2.4298031158412337e-06} +{"train_loss": 0.016309447586536407, "grad_norm": 0.15898792445659637, "global_step": 47719, "epoch": 450, "lr": 2.4288905202479283e-06} +{"train_loss": 0.01566741243004799, "grad_norm": 0.07073619216680527, "global_step": 47720, "epoch": 450, "lr": 2.427978091798799e-06} +{"train_loss": 0.014413826167583466, "grad_norm": 0.05754455551505089, "global_step": 47721, "epoch": 450, "lr": 2.427065830497066e-06} +{"train_loss": 0.014487380161881447, "grad_norm": 0.0687560886144638, "global_step": 47722, "epoch": 450, "lr": 2.4261537363459486e-06} +{"train_loss": 0.015068726614117622, "grad_norm": 0.05219613388180733, "global_step": 47723, "epoch": 450, "lr": 2.425241809348622e-06} +{"train_loss": 0.011693163774907589, "grad_norm": 0.08071621507406235, "global_step": 47724, "epoch": 450, "lr": 2.424330049508311e-06} +{"train_loss": 0.01455751620233059, "grad_norm": 0.0744904950261116, "global_step": 47725, "epoch": 450, "lr": 2.423418456828219e-06} +{"train_loss": 0.013687733560800552, "grad_norm": 0.0835518166422844, "global_step": 47726, "epoch": 450, "lr": 2.4225070313115437e-06} +{"train_loss": 0.013744214549660683, "grad_norm": 0.07489088922739029, "global_step": 47727, "epoch": 450, "lr": 2.4215957729614825e-06} +{"train_loss": 0.01784483902156353, "grad_norm": 0.11092893779277802, "global_step": 47728, "epoch": 450, "lr": 2.4206846817812432e-06} +{"train_loss": 0.015112826600670815, "grad_norm": 0.07376685738563538, "global_step": 47729, "epoch": 450, "lr": 2.41977375777403e-06} +{"train_loss": 0.015050786547362804, "grad_norm": 0.06357087939977646, "global_step": 47730, "epoch": 450, "lr": 2.4188630009430334e-06} +{"train_loss": 0.015157138928771019, "grad_norm": 0.18166857957839966, "global_step": 47731, "epoch": 450, "lr": 2.4179524112914685e-06} +{"train_loss": 0.014535993337631226, "grad_norm": 0.06193013861775398, "global_step": 47732, "epoch": 450, "lr": 2.4170419888225217e-06} +{"train_loss": 0.01668349839746952, "grad_norm": 0.13934099674224854, "global_step": 47733, "epoch": 450, "lr": 2.416131733539395e-06} +{"train_loss": 0.017217330634593964, "grad_norm": 0.18570545315742493, "global_step": 47734, "epoch": 450, "lr": 2.415221645445287e-06} +{"train_loss": 0.012134653516113758, "grad_norm": 0.11512506008148193, "global_step": 47735, "epoch": 450, "lr": 2.4143117245433946e-06} +{"train_loss": 0.017677251249551773, "grad_norm": 0.1624366194009781, "global_step": 47736, "epoch": 450, "lr": 2.4134019708369204e-06} +{"train_loss": 0.018523935228586197, "grad_norm": 0.15385781228542328, "global_step": 47737, "epoch": 450, "lr": 2.4124923843290627e-06} +{"train_loss": 0.01909310556948185, "grad_norm": 0.09913421422243118, "global_step": 47738, "epoch": 450, "lr": 2.4115829650230015e-06} +{"train_loss": 0.014791866764426231, "grad_norm": 0.11168131232261658, "global_step": 47739, "epoch": 450, "lr": 2.410673712921946e-06} +{"train_loss": 0.014649897813796997, "grad_norm": 0.06902267038822174, "global_step": 47740, "epoch": 450, "lr": 2.4097646280290873e-06} +{"train_loss": 0.016062363982200623, "grad_norm": 0.09943694621324539, "global_step": 47741, "epoch": 450, "lr": 2.408855710347613e-06} +{"train_loss": 0.014102221466600895, "grad_norm": 0.051688291132450104, "global_step": 47742, "epoch": 450, "lr": 2.40794695988073e-06} +{"train_loss": 0.014729888178408146, "grad_norm": 0.07801847159862518, "global_step": 47743, "epoch": 450, "lr": 2.4070383766316264e-06} +{"train_loss": 0.015520572662353516, "grad_norm": 0.1115405410528183, "global_step": 47744, "epoch": 450, "lr": 2.4061299606034872e-06} +{"train_loss": 0.013970430940389633, "grad_norm": 0.10682081431150436, "global_step": 47745, "epoch": 450, "lr": 2.4052217117995046e-06} +{"train_loss": 0.014870316721498966, "grad_norm": 0.06296209990978241, "global_step": 47746, "epoch": 450, "lr": 2.404313630222882e-06} +{"train_loss": 0.014481190592050552, "grad_norm": 0.10211463272571564, "global_step": 47747, "epoch": 450, "lr": 2.4034057158767996e-06} +{"train_loss": 0.014729846268892288, "grad_norm": 0.07124757021665573, "global_step": 47748, "epoch": 450, "lr": 2.4024979687644554e-06} +{"train_loss": 0.014319796115159988, "grad_norm": 0.08873956650495529, "global_step": 47749, "epoch": 450, "lr": 2.4015903888890245e-06} +{"train_loss": 0.01506765466183424, "grad_norm": 0.07800406962633133, "global_step": 47750, "epoch": 450, "lr": 2.4006829762537154e-06} +{"train_loss": 0.01349341869354248, "grad_norm": 0.12411075830459595, "global_step": 47751, "epoch": 450, "lr": 2.3997757308616973e-06} +{"train_loss": 0.015647120773792267, "grad_norm": 0.07501079142093658, "global_step": 47752, "epoch": 450, "lr": 2.3988686527161687e-06} +{"train_loss": 0.016647297888994217, "grad_norm": 0.15661513805389404, "global_step": 47753, "epoch": 450, "lr": 2.3979617418203148e-06} +{"train_loss": 0.011828398331999779, "grad_norm": 0.05266749486327171, "global_step": 47754, "epoch": 450, "lr": 2.3970549981773284e-06} +{"train_loss": 0.015406087972223759, "grad_norm": 0.09642349928617477, "global_step": 47755, "epoch": 450, "lr": 2.3961484217903783e-06} +{"train_loss": 0.012787790969014168, "grad_norm": 0.08112423866987228, "global_step": 47756, "epoch": 450, "lr": 2.3952420126626685e-06} +{"train_loss": 0.01652435027062893, "grad_norm": 0.08005881309509277, "global_step": 47757, "epoch": 450, "lr": 2.394335770797368e-06} +{"train_loss": 0.014842540957033634, "grad_norm": 0.0810365080833435, "global_step": 47758, "epoch": 450, "lr": 2.3934296961976687e-06} +{"train_loss": 0.014307443052530289, "grad_norm": 0.06348129361867905, "global_step": 47759, "epoch": 450, "lr": 2.3925237888667572e-06} +{"train_loss": 0.01301681436598301, "grad_norm": 0.0673195943236351, "global_step": 47760, "epoch": 450, "lr": 2.3916180488078254e-06} +{"train_loss": 0.012799570336937904, "grad_norm": 0.08245379477739334, "global_step": 47761, "epoch": 450, "lr": 2.3907124760240263e-06} +{"train_loss": 0.013947557657957077, "grad_norm": 0.1515452265739441, "global_step": 47762, "epoch": 450, "lr": 2.389807070518563e-06} +{"train_loss": 0.013068389147520065, "grad_norm": 0.0552373081445694, "global_step": 47763, "epoch": 450, "lr": 2.3889018322946154e-06} +{"train_loss": 0.014014482498168945, "grad_norm": 0.07283083349466324, "global_step": 47764, "epoch": 450, "lr": 2.3879967613553543e-06} +{"train_loss": 0.013216321356594563, "grad_norm": 0.08328870683908463, "global_step": 47765, "epoch": 450, "lr": 2.387091857703971e-06} +{"train_loss": 0.015537520870566368, "grad_norm": 0.06985022872686386, "global_step": 47766, "epoch": 450, "lr": 2.3861871213436416e-06} +{"train_loss": 0.014358633197844028, "grad_norm": 0.12418507784605026, "global_step": 47767, "epoch": 450, "lr": 2.385282552277551e-06} +{"train_loss": 0.014607703313231468, "grad_norm": 0.10318053513765335, "global_step": 47768, "epoch": 450, "lr": 2.384378150508859e-06} +{"train_loss": 0.014524292200803757, "grad_norm": 0.07539262622594833, "global_step": 47769, "epoch": 450, "lr": 2.383473916040757e-06} +{"train_loss": 0.014577198773622513, "grad_norm": 0.09758516401052475, "global_step": 47770, "epoch": 450, "lr": 2.382569848876426e-06} +{"train_loss": 0.012206336483359337, "grad_norm": 0.07801708579063416, "global_step": 47771, "epoch": 450, "lr": 2.381665949019024e-06} +{"train_loss": 0.01310223899781704, "grad_norm": 0.06232850253582001, "global_step": 47772, "epoch": 450, "lr": 2.380762216471755e-06} +{"train_loss": 0.015030991286039352, "grad_norm": 0.09116285294294357, "global_step": 47773, "epoch": 450, "lr": 2.379858651237765e-06} +{"train_loss": 0.014150666072964668, "grad_norm": 0.14243029057979584, "global_step": 47774, "epoch": 450, "lr": 2.378955253320253e-06} +{"train_loss": 0.015215389430522919, "grad_norm": 0.09982618689537048, "global_step": 47775, "epoch": 450, "lr": 2.3780520227223713e-06} +{"train_loss": 0.013160347938537598, "grad_norm": 0.08694560080766678, "global_step": 47776, "epoch": 450, "lr": 2.377148959447306e-06} +{"train_loss": 0.014532243832945824, "grad_norm": 0.0536566898226738, "global_step": 47777, "epoch": 450, "lr": 2.3762460634982385e-06} +{"train_loss": 0.012853430584073067, "grad_norm": 0.08253505825996399, "global_step": 47778, "epoch": 450, "lr": 2.3753433348783327e-06} +{"train_loss": 0.014322556555271149, "grad_norm": 0.06296809762716293, "global_step": 47779, "epoch": 450, "lr": 2.374440773590747e-06} +{"train_loss": 0.014661203138530254, "grad_norm": 0.06061689555644989, "global_step": 47780, "epoch": 450, "lr": 2.3735383796386732e-06} +{"train_loss": 0.017546743154525757, "grad_norm": 0.15597033500671387, "global_step": 47781, "epoch": 450, "lr": 2.372636153025265e-06} +{"train_loss": 0.016484234482049942, "grad_norm": 0.09314564615488052, "global_step": 47782, "epoch": 450, "lr": 2.3717340937537026e-06} +{"train_loss": 0.01671774312853813, "grad_norm": 0.09171359986066818, "global_step": 47783, "epoch": 450, "lr": 2.370832201827161e-06} +{"train_loss": 0.016840698197484016, "grad_norm": 0.09974885731935501, "global_step": 47784, "epoch": 450, "lr": 2.3699304772487997e-06} +{"train_loss": 0.017753683030605316, "grad_norm": 0.06463232636451721, "global_step": 47785, "epoch": 450, "lr": 2.3690289200217874e-06} +{"train_loss": 0.014339342713356018, "grad_norm": 0.08577627688646317, "global_step": 47786, "epoch": 450, "lr": 2.368127530149289e-06} +{"train_loss": 0.014540838077664375, "grad_norm": 0.15064533054828644, "global_step": 47787, "epoch": 450, "lr": 2.3672263076344846e-06} +{"train_loss": 0.01499820128083229, "grad_norm": 0.11754003167152405, "global_step": 47788, "epoch": 450, "lr": 2.3663252524805224e-06} +{"train_loss": 0.015505816787481308, "grad_norm": 0.12945665419101715, "global_step": 47789, "epoch": 450, "lr": 2.365424364690577e-06} +{"train_loss": 0.01148255169391632, "grad_norm": 0.06558895111083984, "global_step": 47790, "epoch": 450, "lr": 2.3645236442678244e-06} +{"train_loss": 0.013853127136826515, "grad_norm": 0.07360891997814178, "global_step": 47791, "epoch": 450, "lr": 2.3636230912154167e-06} +{"train_loss": 0.016352413222193718, "grad_norm": 0.0783081203699112, "global_step": 47792, "epoch": 450, "lr": 2.3627227055365187e-06} +{"train_loss": 0.018594246357679367, "grad_norm": 0.18847046792507172, "global_step": 47793, "epoch": 450, "lr": 2.3618224872342943e-06} +{"train_loss": 0.015498179011046886, "grad_norm": 0.09010721743106842, "global_step": 47794, "epoch": 450, "lr": 2.3609224363119133e-06} +{"train_loss": 0.013577228412032127, "grad_norm": 0.07047674804925919, "global_step": 47795, "epoch": 450, "lr": 2.3600225527725338e-06} +{"train_loss": 0.01380440779030323, "grad_norm": 0.14994341135025024, "global_step": 47796, "epoch": 450, "lr": 2.359122836619315e-06} +{"train_loss": 0.015435958281159401, "grad_norm": 0.08949032425880432, "global_step": 47797, "epoch": 450, "lr": 2.358223287855421e-06} +{"train_loss": 0.015683773905038834, "grad_norm": 0.09829015284776688, "global_step": 47798, "epoch": 450, "lr": 2.3573239064840048e-06} +{"train_loss": 0.016070038080215454, "grad_norm": 0.07766930758953094, "global_step": 47799, "epoch": 450, "lr": 2.3564246925082357e-06} +{"train_loss": 0.016341764479875565, "grad_norm": 0.09204579144716263, "global_step": 47800, "epoch": 450, "lr": 2.355525645931278e-06} +{"train_loss": 0.012457049451768398, "grad_norm": 0.08768338710069656, "global_step": 47801, "epoch": 450, "lr": 2.3546267667562795e-06} +{"train_loss": 0.015413971617817879, "grad_norm": 0.14621080458164215, "global_step": 47802, "epoch": 450, "lr": 2.353728054986398e-06} +{"train_loss": 0.017128119245171547, "grad_norm": 0.1378355771303177, "global_step": 47803, "epoch": 450, "lr": 2.3528295106247934e-06} +{"train_loss": 0.015130464918911457, "grad_norm": 0.05845651775598526, "global_step": 47804, "epoch": 450, "lr": 2.3519311336746343e-06} +{"train_loss": 0.014966276043780008, "grad_norm": 0.13774023950099945, "global_step": 47805, "epoch": 450, "lr": 2.351032924139063e-06, "train_batch_mse_error": 0.013393737375736237, "val_loss": 0.020386261865496635} +{"train_loss": 0.01243673637509346, "grad_norm": 0.046388134360313416, "global_step": 47806, "epoch": 451, "lr": 2.350134882021232e-06} +{"train_loss": 0.012128423899412155, "grad_norm": 0.0901479572057724, "global_step": 47807, "epoch": 451, "lr": 2.3492370073243175e-06} +{"train_loss": 0.014522122219204903, "grad_norm": 0.09417615830898285, "global_step": 47808, "epoch": 451, "lr": 2.348339300051461e-06} +{"train_loss": 0.013636044226586819, "grad_norm": 0.06986851990222931, "global_step": 47809, "epoch": 451, "lr": 2.3474417602058096e-06} +{"train_loss": 0.013707824051380157, "grad_norm": 0.12635760009288788, "global_step": 47810, "epoch": 451, "lr": 2.3465443877905224e-06} +{"train_loss": 0.01457834430038929, "grad_norm": 0.06685636937618256, "global_step": 47811, "epoch": 451, "lr": 2.3456471828087635e-06} +{"train_loss": 0.01563354581594467, "grad_norm": 0.09778435528278351, "global_step": 47812, "epoch": 451, "lr": 2.344750145263669e-06} +{"train_loss": 0.014718512073159218, "grad_norm": 0.0673028975725174, "global_step": 47813, "epoch": 451, "lr": 2.343853275158403e-06} +{"train_loss": 0.012953381054103374, "grad_norm": 0.057996995747089386, "global_step": 47814, "epoch": 451, "lr": 2.3429565724961024e-06} +{"train_loss": 0.0147653017193079, "grad_norm": 0.10113420337438583, "global_step": 47815, "epoch": 451, "lr": 2.342060037279936e-06} +{"train_loss": 0.01254146546125412, "grad_norm": 0.07858102768659592, "global_step": 47816, "epoch": 451, "lr": 2.341163669513041e-06} +{"train_loss": 0.013269167393445969, "grad_norm": 0.10367506742477417, "global_step": 47817, "epoch": 451, "lr": 2.3402674691985703e-06} +{"train_loss": 0.014444932341575623, "grad_norm": 0.11673025786876678, "global_step": 47818, "epoch": 451, "lr": 2.339371436339677e-06} +{"train_loss": 0.014944346621632576, "grad_norm": 0.059881292283535004, "global_step": 47819, "epoch": 451, "lr": 2.3384755709395025e-06} +{"train_loss": 0.014622421003878117, "grad_norm": 0.07102621346712112, "global_step": 47820, "epoch": 451, "lr": 2.3375798730011943e-06} +{"train_loss": 0.013388055376708508, "grad_norm": 0.06195201724767685, "global_step": 47821, "epoch": 451, "lr": 2.3366843425279118e-06} +{"train_loss": 0.014215325936675072, "grad_norm": 0.0871603861451149, "global_step": 47822, "epoch": 451, "lr": 2.335788979522785e-06} +{"train_loss": 0.012423943728208542, "grad_norm": 0.10226759314537048, "global_step": 47823, "epoch": 451, "lr": 2.3348937839889617e-06} +{"train_loss": 0.016799869015812874, "grad_norm": 0.07085244357585907, "global_step": 47824, "epoch": 451, "lr": 2.3339987559296062e-06} +{"train_loss": 0.014471405185759068, "grad_norm": 0.10179764032363892, "global_step": 47825, "epoch": 451, "lr": 2.3331038953478434e-06} +{"train_loss": 0.016099777072668076, "grad_norm": 0.05859440192580223, "global_step": 47826, "epoch": 451, "lr": 2.332209202246821e-06} +{"train_loss": 0.015103043057024479, "grad_norm": 0.11289842426776886, "global_step": 47827, "epoch": 451, "lr": 2.3313146766296812e-06} +{"train_loss": 0.013864357024431229, "grad_norm": 0.07078859210014343, "global_step": 47828, "epoch": 451, "lr": 2.330420318499582e-06} +{"train_loss": 0.012579767033457756, "grad_norm": 0.06354360282421112, "global_step": 47829, "epoch": 451, "lr": 2.329526127859649e-06} +{"train_loss": 0.014154208824038506, "grad_norm": 0.07645969837903976, "global_step": 47830, "epoch": 451, "lr": 2.32863210471303e-06} +{"train_loss": 0.014871173538267612, "grad_norm": 0.08460116386413574, "global_step": 47831, "epoch": 451, "lr": 2.3277382490628663e-06} +{"train_loss": 0.01496885810047388, "grad_norm": 0.12795262038707733, "global_step": 47832, "epoch": 451, "lr": 2.326844560912306e-06} +{"train_loss": 0.015125786885619164, "grad_norm": 0.15602336823940277, "global_step": 47833, "epoch": 451, "lr": 2.325951040264468e-06} +{"train_loss": 0.016429461538791656, "grad_norm": 0.0843762755393982, "global_step": 47834, "epoch": 451, "lr": 2.325057687122512e-06} +{"train_loss": 0.016792524605989456, "grad_norm": 0.07233031094074249, "global_step": 47835, "epoch": 451, "lr": 2.324164501489573e-06} +{"train_loss": 0.013100489974021912, "grad_norm": 0.05964319035410881, "global_step": 47836, "epoch": 451, "lr": 2.3232714833687772e-06} +{"train_loss": 0.01551094837486744, "grad_norm": 0.06528251618146896, "global_step": 47837, "epoch": 451, "lr": 2.3223786327632834e-06} +{"train_loss": 0.01672232523560524, "grad_norm": 0.06562367081642151, "global_step": 47838, "epoch": 451, "lr": 2.321485949676211e-06} +{"train_loss": 0.014069043099880219, "grad_norm": 0.05278276652097702, "global_step": 47839, "epoch": 451, "lr": 2.320593434110696e-06} +{"train_loss": 0.018400434404611588, "grad_norm": 0.11878693103790283, "global_step": 47840, "epoch": 451, "lr": 2.3197010860698865e-06} +{"train_loss": 0.014130971394479275, "grad_norm": 0.05681339278817177, "global_step": 47841, "epoch": 451, "lr": 2.318808905556913e-06} +{"train_loss": 0.014679864048957825, "grad_norm": 0.07297180593013763, "global_step": 47842, "epoch": 451, "lr": 2.317916892574912e-06} +{"train_loss": 0.014135140925645828, "grad_norm": 0.08048149198293686, "global_step": 47843, "epoch": 451, "lr": 2.3170250471270028e-06} +{"train_loss": 0.015673354268074036, "grad_norm": 0.07407175749540329, "global_step": 47844, "epoch": 451, "lr": 2.316133369216339e-06} +{"train_loss": 0.015345499850809574, "grad_norm": 0.08228300511837006, "global_step": 47845, "epoch": 451, "lr": 2.3152418588460456e-06} +{"train_loss": 0.015653099864721298, "grad_norm": 0.07509132474660873, "global_step": 47846, "epoch": 451, "lr": 2.314350516019248e-06} +{"train_loss": 0.01622806116938591, "grad_norm": 0.06409549713134766, "global_step": 47847, "epoch": 451, "lr": 2.313459340739088e-06} +{"train_loss": 0.013401584699749947, "grad_norm": 0.0776359960436821, "global_step": 47848, "epoch": 451, "lr": 2.3125683330087023e-06} +{"train_loss": 0.013215761631727219, "grad_norm": 0.08331942558288574, "global_step": 47849, "epoch": 451, "lr": 2.3116774928312048e-06} +{"train_loss": 0.01760907471179962, "grad_norm": 0.11822625994682312, "global_step": 47850, "epoch": 451, "lr": 2.3107868202097317e-06} +{"train_loss": 0.014179980382323265, "grad_norm": 0.06120448559522629, "global_step": 47851, "epoch": 451, "lr": 2.309896315147414e-06} +{"train_loss": 0.019825560972094536, "grad_norm": 0.09264924377202988, "global_step": 47852, "epoch": 451, "lr": 2.3090059776473884e-06} +{"train_loss": 0.012763570062816143, "grad_norm": 0.08929522335529327, "global_step": 47853, "epoch": 451, "lr": 2.3081158077127683e-06} +{"train_loss": 0.014566316269338131, "grad_norm": 0.08082357048988342, "global_step": 47854, "epoch": 451, "lr": 2.3072258053466965e-06} +{"train_loss": 0.01840526983141899, "grad_norm": 0.08802015334367752, "global_step": 47855, "epoch": 451, "lr": 2.3063359705522923e-06} +{"train_loss": 0.014396442100405693, "grad_norm": 0.06946961581707001, "global_step": 47856, "epoch": 451, "lr": 2.30544630333267e-06} +{"train_loss": 0.012267709709703922, "grad_norm": 0.063747338950634, "global_step": 47857, "epoch": 451, "lr": 2.3045568036909714e-06} +{"train_loss": 0.014449809677898884, "grad_norm": 0.06991302222013474, "global_step": 47858, "epoch": 451, "lr": 2.3036674716303275e-06} +{"train_loss": 0.014908034354448318, "grad_norm": 0.07709861546754837, "global_step": 47859, "epoch": 451, "lr": 2.302778307153841e-06} +{"train_loss": 0.01587977446615696, "grad_norm": 0.1005028560757637, "global_step": 47860, "epoch": 451, "lr": 2.30188931026466e-06} +{"train_loss": 0.017306819558143616, "grad_norm": 0.11207572370767593, "global_step": 47861, "epoch": 451, "lr": 2.301000480965887e-06} +{"train_loss": 0.014481723308563232, "grad_norm": 0.08591768145561218, "global_step": 47862, "epoch": 451, "lr": 2.3001118192606642e-06} +{"train_loss": 0.015075994655489922, "grad_norm": 0.09579005837440491, "global_step": 47863, "epoch": 451, "lr": 2.2992233251520947e-06} +{"train_loss": 0.015271520242094994, "grad_norm": 0.09656663239002228, "global_step": 47864, "epoch": 451, "lr": 2.298334998643309e-06} +{"train_loss": 0.015657437965273857, "grad_norm": 0.06702379882335663, "global_step": 47865, "epoch": 451, "lr": 2.2974468397374383e-06} +{"train_loss": 0.018302861601114273, "grad_norm": 0.11073806136846542, "global_step": 47866, "epoch": 451, "lr": 2.2965588484375967e-06} +{"train_loss": 0.014867305755615234, "grad_norm": 0.0824093371629715, "global_step": 47867, "epoch": 451, "lr": 2.2956710247468927e-06} +{"train_loss": 0.013172974810004234, "grad_norm": 0.13506387174129486, "global_step": 47868, "epoch": 451, "lr": 2.2947833686684516e-06} +{"train_loss": 0.013678541406989098, "grad_norm": 0.0762474462389946, "global_step": 47869, "epoch": 451, "lr": 2.2938958802054043e-06} +{"train_loss": 0.016445521265268326, "grad_norm": 0.13898570835590363, "global_step": 47870, "epoch": 451, "lr": 2.293008559360854e-06} +{"train_loss": 0.015513529069721699, "grad_norm": 0.06877488642930984, "global_step": 47871, "epoch": 451, "lr": 2.2921214061379202e-06} +{"train_loss": 0.015689581632614136, "grad_norm": 0.057001400738954544, "global_step": 47872, "epoch": 451, "lr": 2.2912344205397397e-06} +{"train_loss": 0.016935037449002266, "grad_norm": 0.07347124814987183, "global_step": 47873, "epoch": 451, "lr": 2.290347602569404e-06} +{"train_loss": 0.015750078484416008, "grad_norm": 0.07014688849449158, "global_step": 47874, "epoch": 451, "lr": 2.289460952230038e-06} +{"train_loss": 0.01478347834199667, "grad_norm": 0.14057126641273499, "global_step": 47875, "epoch": 451, "lr": 2.288574469524757e-06} +{"train_loss": 0.014695960097014904, "grad_norm": 0.06015768274664879, "global_step": 47876, "epoch": 451, "lr": 2.2876881544566797e-06} +{"train_loss": 0.014727887697517872, "grad_norm": 0.13470381498336792, "global_step": 47877, "epoch": 451, "lr": 2.2868020070289097e-06} +{"train_loss": 0.01351405493915081, "grad_norm": 0.06982231885194778, "global_step": 47878, "epoch": 451, "lr": 2.285916027244578e-06} +{"train_loss": 0.015139011666178703, "grad_norm": 0.06050247326493263, "global_step": 47879, "epoch": 451, "lr": 2.2850302151067814e-06} +{"train_loss": 0.01651633158326149, "grad_norm": 0.06457063555717468, "global_step": 47880, "epoch": 451, "lr": 2.2841445706186402e-06} +{"train_loss": 0.01581132598221302, "grad_norm": 0.08230619877576828, "global_step": 47881, "epoch": 451, "lr": 2.2832590937832577e-06} +{"train_loss": 0.016056422144174576, "grad_norm": 0.06519307941198349, "global_step": 47882, "epoch": 451, "lr": 2.2823737846037585e-06} +{"train_loss": 0.01658041402697563, "grad_norm": 0.06688867509365082, "global_step": 47883, "epoch": 451, "lr": 2.2814886430832406e-06} +{"train_loss": 0.013640644028782845, "grad_norm": 0.06531190127134323, "global_step": 47884, "epoch": 451, "lr": 2.280603669224823e-06} +{"train_loss": 0.01328766718506813, "grad_norm": 0.09726233035326004, "global_step": 47885, "epoch": 451, "lr": 2.279718863031605e-06} +{"train_loss": 0.015514278784394264, "grad_norm": 0.06839866936206818, "global_step": 47886, "epoch": 451, "lr": 2.278834224506715e-06} +{"train_loss": 0.013883495703339577, "grad_norm": 0.060357898473739624, "global_step": 47887, "epoch": 451, "lr": 2.277949753653236e-06} +{"train_loss": 0.015914883464574814, "grad_norm": 0.055644333362579346, "global_step": 47888, "epoch": 451, "lr": 2.277065450474286e-06} +{"train_loss": 0.014842275530099869, "grad_norm": 0.08497371524572372, "global_step": 47889, "epoch": 451, "lr": 2.276181314972986e-06} +{"train_loss": 0.015085302293300629, "grad_norm": 0.06652095168828964, "global_step": 47890, "epoch": 451, "lr": 2.2752973471524277e-06} +{"train_loss": 0.015893608331680298, "grad_norm": 0.08763618767261505, "global_step": 47891, "epoch": 451, "lr": 2.2744135470157136e-06} +{"train_loss": 0.01468035951256752, "grad_norm": 0.10378537327051163, "global_step": 47892, "epoch": 451, "lr": 2.273529914565953e-06} +{"train_loss": 0.013670800253748894, "grad_norm": 0.06707534193992615, "global_step": 47893, "epoch": 451, "lr": 2.272646449806259e-06} +{"train_loss": 0.014775874093174934, "grad_norm": 0.05869719013571739, "global_step": 47894, "epoch": 451, "lr": 2.2717631527397255e-06} +{"train_loss": 0.013431143946945667, "grad_norm": 0.08184675127267838, "global_step": 47895, "epoch": 451, "lr": 2.270880023369465e-06} +{"train_loss": 0.014103618450462818, "grad_norm": 0.09920509159564972, "global_step": 47896, "epoch": 451, "lr": 2.2699970616985757e-06} +{"train_loss": 0.015061505138874054, "grad_norm": 0.08166629076004028, "global_step": 47897, "epoch": 451, "lr": 2.269114267730149e-06} +{"train_loss": 0.01561829075217247, "grad_norm": 0.07439655810594559, "global_step": 47898, "epoch": 451, "lr": 2.2682316414673056e-06} +{"train_loss": 0.014966312795877457, "grad_norm": 0.0915941447019577, "global_step": 47899, "epoch": 451, "lr": 2.2673491829131364e-06} +{"train_loss": 0.015443416312336922, "grad_norm": 0.07910099625587463, "global_step": 47900, "epoch": 451, "lr": 2.26646689207074e-06} +{"train_loss": 0.01578558422625065, "grad_norm": 0.10164162516593933, "global_step": 47901, "epoch": 451, "lr": 2.26558476894323e-06} +{"train_loss": 0.0165106188505888, "grad_norm": 0.11398342996835709, "global_step": 47902, "epoch": 451, "lr": 2.2647028135336867e-06} +{"train_loss": 0.0170778576284647, "grad_norm": 0.09531109780073166, "global_step": 47903, "epoch": 451, "lr": 2.263821025845225e-06} +{"train_loss": 0.014756012707948685, "grad_norm": 0.08878745883703232, "global_step": 47904, "epoch": 451, "lr": 2.2629394058809317e-06} +{"train_loss": 0.013623961247503757, "grad_norm": 0.07784189283847809, "global_step": 47905, "epoch": 451, "lr": 2.2620579536439036e-06} +{"train_loss": 0.014976431615650654, "grad_norm": 0.11991552263498306, "global_step": 47906, "epoch": 451, "lr": 2.261176669137255e-06} +{"train_loss": 0.014962130226194859, "grad_norm": 0.08337873220443726, "global_step": 47907, "epoch": 451, "lr": 2.2602955523640668e-06} +{"train_loss": 0.015261187218129635, "grad_norm": 0.11163748800754547, "global_step": 47908, "epoch": 451, "lr": 2.2594146033274367e-06} +{"train_loss": 0.013558538630604744, "grad_norm": 0.07554043084383011, "global_step": 47909, "epoch": 451, "lr": 2.2585338220304565e-06} +{"train_loss": 0.01878582127392292, "grad_norm": 0.10816962271928787, "global_step": 47910, "epoch": 451, "lr": 2.2576532084762346e-06} +{"train_loss": 0.014952699618660054, "grad_norm": 0.08775625377893448, "global_step": 47911, "epoch": 451, "lr": 2.2567727626678527e-06, "val_loss": 0.019632261246442795} +{"train_loss": 0.014524411410093307, "grad_norm": 0.05316058173775673, "global_step": 47912, "epoch": 452, "lr": 2.2558924846084074e-06} +{"train_loss": 0.013963106088340282, "grad_norm": 0.08637560904026031, "global_step": 47913, "epoch": 452, "lr": 2.2550123743010023e-06} +{"train_loss": 0.014356808736920357, "grad_norm": 0.07400061190128326, "global_step": 47914, "epoch": 452, "lr": 2.254132431748712e-06} +{"train_loss": 0.01633566990494728, "grad_norm": 0.06641077995300293, "global_step": 47915, "epoch": 452, "lr": 2.2532526569546297e-06} +{"train_loss": 0.016794681549072266, "grad_norm": 0.05859238654375076, "global_step": 47916, "epoch": 452, "lr": 2.2523730499218575e-06} +{"train_loss": 0.01418030634522438, "grad_norm": 0.08874624222517014, "global_step": 47917, "epoch": 452, "lr": 2.251493610653488e-06} +{"train_loss": 0.01505262777209282, "grad_norm": 0.09243965148925781, "global_step": 47918, "epoch": 452, "lr": 2.250614339152601e-06} +{"train_loss": 0.014123977161943913, "grad_norm": 0.07936819642782211, "global_step": 47919, "epoch": 452, "lr": 2.2497352354222902e-06} +{"train_loss": 0.01522020436823368, "grad_norm": 0.061646465212106705, "global_step": 47920, "epoch": 452, "lr": 2.248856299465646e-06} +{"train_loss": 0.012797854840755463, "grad_norm": 0.10186292231082916, "global_step": 47921, "epoch": 452, "lr": 2.2479775312857498e-06} +{"train_loss": 0.015015599317848682, "grad_norm": 0.11240855604410172, "global_step": 47922, "epoch": 452, "lr": 2.2470989308856883e-06} +{"train_loss": 0.012510479427874088, "grad_norm": 0.06139605864882469, "global_step": 47923, "epoch": 452, "lr": 2.2462204982685643e-06} +{"train_loss": 0.01569356396794319, "grad_norm": 0.13071879744529724, "global_step": 47924, "epoch": 452, "lr": 2.2453422334374473e-06} +{"train_loss": 0.014373047277331352, "grad_norm": 0.06853339076042175, "global_step": 47925, "epoch": 452, "lr": 2.2444641363954354e-06} +{"train_loss": 0.015784168615937233, "grad_norm": 0.06308089196681976, "global_step": 47926, "epoch": 452, "lr": 2.2435862071456037e-06} +{"train_loss": 0.014100069180130959, "grad_norm": 0.061739567667245865, "global_step": 47927, "epoch": 452, "lr": 2.2427084456910495e-06} +{"train_loss": 0.017011407762765884, "grad_norm": 0.08189794421195984, "global_step": 47928, "epoch": 452, "lr": 2.2418308520348376e-06} +{"train_loss": 0.01592290960252285, "grad_norm": 0.13218006491661072, "global_step": 47929, "epoch": 452, "lr": 2.240953426180065e-06} +{"train_loss": 0.0159207284450531, "grad_norm": 0.07831841707229614, "global_step": 47930, "epoch": 452, "lr": 2.240076168129818e-06} +{"train_loss": 0.013783768750727177, "grad_norm": 0.0938442274928093, "global_step": 47931, "epoch": 452, "lr": 2.2391990778871717e-06} +{"train_loss": 0.013862067833542824, "grad_norm": 0.06467597931623459, "global_step": 47932, "epoch": 452, "lr": 2.238322155455208e-06} +{"train_loss": 0.014196401461958885, "grad_norm": 0.08271218091249466, "global_step": 47933, "epoch": 452, "lr": 2.2374454008370014e-06} +{"train_loss": 0.014652086421847343, "grad_norm": 0.09208189696073532, "global_step": 47934, "epoch": 452, "lr": 2.23656881403565e-06} +{"train_loss": 0.01713407039642334, "grad_norm": 0.07690469920635223, "global_step": 47935, "epoch": 452, "lr": 2.235692395054223e-06} +{"train_loss": 0.013762944377958775, "grad_norm": 0.08690719306468964, "global_step": 47936, "epoch": 452, "lr": 2.2348161438957958e-06} +{"train_loss": 0.014506438747048378, "grad_norm": 0.09683732688426971, "global_step": 47937, "epoch": 452, "lr": 2.2339400605634607e-06} +{"train_loss": 0.012762695550918579, "grad_norm": 0.06838054209947586, "global_step": 47938, "epoch": 452, "lr": 2.2330641450602818e-06} +{"train_loss": 0.015494514256715775, "grad_norm": 0.11544828116893768, "global_step": 47939, "epoch": 452, "lr": 2.2321883973893397e-06} +{"train_loss": 0.014429549686610699, "grad_norm": 0.11101003736257553, "global_step": 47940, "epoch": 452, "lr": 2.231312817553721e-06} +{"train_loss": 0.011922701261937618, "grad_norm": 0.14271984994411469, "global_step": 47941, "epoch": 452, "lr": 2.230437405556485e-06} +{"train_loss": 0.015687044709920883, "grad_norm": 0.11245132237672806, "global_step": 47942, "epoch": 452, "lr": 2.229562161400728e-06} +{"train_loss": 0.015371796675026417, "grad_norm": 0.09417696297168732, "global_step": 47943, "epoch": 452, "lr": 2.2286870850895094e-06} +{"train_loss": 0.016228459775447845, "grad_norm": 0.07281682640314102, "global_step": 47944, "epoch": 452, "lr": 2.2278121766259097e-06} +{"train_loss": 0.014899822883307934, "grad_norm": 0.07540279626846313, "global_step": 47945, "epoch": 452, "lr": 2.226937436012999e-06} +{"train_loss": 0.01424618810415268, "grad_norm": 0.09136343747377396, "global_step": 47946, "epoch": 452, "lr": 2.226062863253858e-06} +{"train_loss": 0.014534210786223412, "grad_norm": 0.06833577156066895, "global_step": 47947, "epoch": 452, "lr": 2.225188458351557e-06} +{"train_loss": 0.017381086945533752, "grad_norm": 0.06487873196601868, "global_step": 47948, "epoch": 452, "lr": 2.2243142213091707e-06} +{"train_loss": 0.013174602761864662, "grad_norm": 0.08598670363426208, "global_step": 47949, "epoch": 452, "lr": 2.2234401521297575e-06} +{"train_loss": 0.012629720382392406, "grad_norm": 0.06326168030500412, "global_step": 47950, "epoch": 452, "lr": 2.222566250816399e-06} +{"train_loss": 0.012955335900187492, "grad_norm": 0.06015613302588463, "global_step": 47951, "epoch": 452, "lr": 2.2216925173721703e-06} +{"train_loss": 0.014462009072303772, "grad_norm": 0.12264469265937805, "global_step": 47952, "epoch": 452, "lr": 2.2208189518001355e-06} +{"train_loss": 0.01586691103875637, "grad_norm": 0.07846073806285858, "global_step": 47953, "epoch": 452, "lr": 2.2199455541033588e-06} +{"train_loss": 0.014254698529839516, "grad_norm": 0.0719253197312355, "global_step": 47954, "epoch": 452, "lr": 2.2190723242849267e-06} +{"train_loss": 0.015528536401689053, "grad_norm": 0.10578972101211548, "global_step": 47955, "epoch": 452, "lr": 2.218199262347881e-06} +{"train_loss": 0.013273772783577442, "grad_norm": 0.07204774022102356, "global_step": 47956, "epoch": 452, "lr": 2.217326368295308e-06} +{"train_loss": 0.012842698954045773, "grad_norm": 0.10415008664131165, "global_step": 47957, "epoch": 452, "lr": 2.216453642130273e-06} +{"train_loss": 0.015526819974184036, "grad_norm": 0.06224557012319565, "global_step": 47958, "epoch": 452, "lr": 2.215581083855839e-06} +{"train_loss": 0.015012407675385475, "grad_norm": 0.11388178169727325, "global_step": 47959, "epoch": 452, "lr": 2.214708693475065e-06} +{"train_loss": 0.01194989588111639, "grad_norm": 0.06793619692325592, "global_step": 47960, "epoch": 452, "lr": 2.213836470991032e-06} +{"train_loss": 0.015117431990802288, "grad_norm": 0.07725635915994644, "global_step": 47961, "epoch": 452, "lr": 2.2129644164067985e-06} +{"train_loss": 0.015329962596297264, "grad_norm": 0.06658805161714554, "global_step": 47962, "epoch": 452, "lr": 2.212092529725418e-06} +{"train_loss": 0.014242572709918022, "grad_norm": 0.10344738513231277, "global_step": 47963, "epoch": 452, "lr": 2.2112208109499597e-06} +{"train_loss": 0.013600017875432968, "grad_norm": 0.07223670929670334, "global_step": 47964, "epoch": 452, "lr": 2.210349260083494e-06} +{"train_loss": 0.014534000307321548, "grad_norm": 0.08224301785230637, "global_step": 47965, "epoch": 452, "lr": 2.2094778771290792e-06} +{"train_loss": 0.012965556234121323, "grad_norm": 0.16117461025714874, "global_step": 47966, "epoch": 452, "lr": 2.2086066620897738e-06} +{"train_loss": 0.014203363098204136, "grad_norm": 0.13824905455112457, "global_step": 47967, "epoch": 452, "lr": 2.2077356149686367e-06} +{"train_loss": 0.016409816220402718, "grad_norm": 0.10566841810941696, "global_step": 47968, "epoch": 452, "lr": 2.2068647357687435e-06} +{"train_loss": 0.016958270221948624, "grad_norm": 0.13024014234542847, "global_step": 47969, "epoch": 452, "lr": 2.20599402449313e-06} +{"train_loss": 0.016944345086812973, "grad_norm": 0.0896911472082138, "global_step": 47970, "epoch": 452, "lr": 2.2051234811448717e-06} +{"train_loss": 0.01803150773048401, "grad_norm": 0.09515055269002914, "global_step": 47971, "epoch": 452, "lr": 2.2042531057270277e-06} +{"train_loss": 0.01425702404230833, "grad_norm": 0.0743907243013382, "global_step": 47972, "epoch": 452, "lr": 2.2033828982426563e-06} +{"train_loss": 0.013766169548034668, "grad_norm": 0.07632668316364288, "global_step": 47973, "epoch": 452, "lr": 2.202512858694805e-06} +{"train_loss": 0.014926206320524216, "grad_norm": 0.11983831226825714, "global_step": 47974, "epoch": 452, "lr": 2.201642987086533e-06} +{"train_loss": 0.01402700413018465, "grad_norm": 0.08271750062704086, "global_step": 47975, "epoch": 452, "lr": 2.2007732834209092e-06} +{"train_loss": 0.01486942358314991, "grad_norm": 0.08018113672733307, "global_step": 47976, "epoch": 452, "lr": 2.1999037477009766e-06} +{"train_loss": 0.014316163957118988, "grad_norm": 0.09363333135843277, "global_step": 47977, "epoch": 452, "lr": 2.1990343799297876e-06} +{"train_loss": 0.01416760217398405, "grad_norm": 0.08561824262142181, "global_step": 47978, "epoch": 452, "lr": 2.198165180110423e-06} +{"train_loss": 0.0163277518004179, "grad_norm": 0.0925154834985733, "global_step": 47979, "epoch": 452, "lr": 2.197296148245903e-06} +{"train_loss": 0.01705722138285637, "grad_norm": 0.11066169291734695, "global_step": 47980, "epoch": 452, "lr": 2.196427284339292e-06} +{"train_loss": 0.01555588562041521, "grad_norm": 0.15448014438152313, "global_step": 47981, "epoch": 452, "lr": 2.1955585883936593e-06} +{"train_loss": 0.016981851309537888, "grad_norm": 0.07873308658599854, "global_step": 47982, "epoch": 452, "lr": 2.1946900604120357e-06} +{"train_loss": 0.015974974259734154, "grad_norm": 0.10009811073541641, "global_step": 47983, "epoch": 452, "lr": 2.193821700397475e-06} +{"train_loss": 0.01493835635483265, "grad_norm": 0.06967789679765701, "global_step": 47984, "epoch": 452, "lr": 2.1929535083530462e-06} +{"train_loss": 0.01764698512852192, "grad_norm": 0.11680950224399567, "global_step": 47985, "epoch": 452, "lr": 2.1920854842817918e-06} +{"train_loss": 0.013707565143704414, "grad_norm": 0.0798746794462204, "global_step": 47986, "epoch": 452, "lr": 2.1912176281867427e-06} +{"train_loss": 0.015107627026736736, "grad_norm": 0.061735719442367554, "global_step": 47987, "epoch": 452, "lr": 2.190349940070968e-06} +{"train_loss": 0.014324406161904335, "grad_norm": 0.07256224006414413, "global_step": 47988, "epoch": 452, "lr": 2.1894824199375217e-06} +{"train_loss": 0.01607396826148033, "grad_norm": 0.15827171504497528, "global_step": 47989, "epoch": 452, "lr": 2.188615067789429e-06} +{"train_loss": 0.016195613890886307, "grad_norm": 0.0780869722366333, "global_step": 47990, "epoch": 452, "lr": 2.1877478836297592e-06} +{"train_loss": 0.014029137790203094, "grad_norm": 0.07876090705394745, "global_step": 47991, "epoch": 452, "lr": 2.186880867461544e-06} +{"train_loss": 0.01366075873374939, "grad_norm": 0.06887681782245636, "global_step": 47992, "epoch": 452, "lr": 2.1860140192878464e-06} +{"train_loss": 0.0165947787463665, "grad_norm": 0.11094524711370468, "global_step": 47993, "epoch": 452, "lr": 2.185147339111687e-06} +{"train_loss": 0.01352701522409916, "grad_norm": 0.11631496995687485, "global_step": 47994, "epoch": 452, "lr": 2.184280826936136e-06} +{"train_loss": 0.01542646624147892, "grad_norm": 0.06836075335741043, "global_step": 47995, "epoch": 452, "lr": 2.1834144827642346e-06} +{"train_loss": 0.020033804699778557, "grad_norm": 0.08029884099960327, "global_step": 47996, "epoch": 452, "lr": 2.1825483065990026e-06} +{"train_loss": 0.014571604318916798, "grad_norm": 0.06687990576028824, "global_step": 47997, "epoch": 452, "lr": 2.1816822984435104e-06} +{"train_loss": 0.01314385887235403, "grad_norm": 0.07897662371397018, "global_step": 47998, "epoch": 452, "lr": 2.1808164583007883e-06} +{"train_loss": 0.01401720754802227, "grad_norm": 0.08838661760091782, "global_step": 47999, "epoch": 452, "lr": 2.179950786173879e-06} +{"train_loss": 0.013958154246211052, "grad_norm": 0.07109769433736801, "global_step": 48000, "epoch": 452, "lr": 2.179085282065829e-06} +{"train_loss": 0.014885326847434044, "grad_norm": 0.1272929310798645, "global_step": 48001, "epoch": 452, "lr": 2.1782199459796758e-06} +{"train_loss": 0.01695096492767334, "grad_norm": 0.059832002967596054, "global_step": 48002, "epoch": 452, "lr": 2.1773547779184665e-06} +{"train_loss": 0.015747597441077232, "grad_norm": 0.12451592832803726, "global_step": 48003, "epoch": 452, "lr": 2.176489777885227e-06} +{"train_loss": 0.014760366640985012, "grad_norm": 0.06792818754911423, "global_step": 48004, "epoch": 452, "lr": 2.1756249458830036e-06} +{"train_loss": 0.017026163637638092, "grad_norm": 0.10845864564180374, "global_step": 48005, "epoch": 452, "lr": 2.174760281914845e-06} +{"train_loss": 0.01589592546224594, "grad_norm": 0.06769522279500961, "global_step": 48006, "epoch": 452, "lr": 2.1738957859837707e-06} +{"train_loss": 0.015611868351697922, "grad_norm": 0.07040553539991379, "global_step": 48007, "epoch": 452, "lr": 2.1730314580928333e-06} +{"train_loss": 0.016097474843263626, "grad_norm": 0.0655910074710846, "global_step": 48008, "epoch": 452, "lr": 2.1721672982450592e-06} +{"train_loss": 0.01585930585861206, "grad_norm": 0.09146698564291, "global_step": 48009, "epoch": 452, "lr": 2.1713033064435007e-06} +{"train_loss": 0.013000939041376114, "grad_norm": 0.05966776981949806, "global_step": 48010, "epoch": 452, "lr": 2.170439482691167e-06} +{"train_loss": 0.01685611717402935, "grad_norm": 0.09953194856643677, "global_step": 48011, "epoch": 452, "lr": 2.169575826991116e-06} +{"train_loss": 0.013149620965123177, "grad_norm": 0.11694533377885818, "global_step": 48012, "epoch": 452, "lr": 2.168712339346374e-06} +{"train_loss": 0.015132136642932892, "grad_norm": 0.08113010972738266, "global_step": 48013, "epoch": 452, "lr": 2.167849019759982e-06} +{"train_loss": 0.015805214643478394, "grad_norm": 0.10841362923383713, "global_step": 48014, "epoch": 452, "lr": 2.1669858682349552e-06} +{"train_loss": 0.015426535159349442, "grad_norm": 0.07735656946897507, "global_step": 48015, "epoch": 452, "lr": 2.1661228847743464e-06} +{"train_loss": 0.014935007318854332, "grad_norm": 0.07370224595069885, "global_step": 48016, "epoch": 452, "lr": 2.165260069381175e-06} +{"train_loss": 0.014951644781625495, "grad_norm": 0.10314997285604477, "global_step": 48017, "epoch": 452, "lr": 2.164397422058473e-06, "val_loss": 0.01947084255516529} +{"train_loss": 0.017049383372068405, "grad_norm": 0.07357995957136154, "global_step": 48018, "epoch": 453, "lr": 2.163534942809281e-06} +{"train_loss": 0.014666395261883736, "grad_norm": 0.05945619195699692, "global_step": 48019, "epoch": 453, "lr": 2.162672631636631e-06} +{"train_loss": 0.017267782241106033, "grad_norm": 0.06910204887390137, "global_step": 48020, "epoch": 453, "lr": 2.1618104885435365e-06} +{"train_loss": 0.014007003977894783, "grad_norm": 0.05708061531186104, "global_step": 48021, "epoch": 453, "lr": 2.1609485135330345e-06} +{"train_loss": 0.014629609882831573, "grad_norm": 0.06624332070350647, "global_step": 48022, "epoch": 453, "lr": 2.160086706608161e-06} +{"train_loss": 0.016411272808909416, "grad_norm": 0.11045064777135849, "global_step": 48023, "epoch": 453, "lr": 2.1592250677719308e-06} +{"train_loss": 0.014285347424447536, "grad_norm": 0.1103583425283432, "global_step": 48024, "epoch": 453, "lr": 2.15836359702738e-06} +{"train_loss": 0.014046983793377876, "grad_norm": 0.16986683011054993, "global_step": 48025, "epoch": 453, "lr": 2.1575022943775338e-06} +{"train_loss": 0.015169132500886917, "grad_norm": 0.0805845782160759, "global_step": 48026, "epoch": 453, "lr": 2.156641159825423e-06} +{"train_loss": 0.017089810222387314, "grad_norm": 0.11573311686515808, "global_step": 48027, "epoch": 453, "lr": 2.155780193374063e-06} +{"train_loss": 0.015123549848794937, "grad_norm": 0.0827246904373169, "global_step": 48028, "epoch": 453, "lr": 2.1549193950264834e-06} +{"train_loss": 0.014960123226046562, "grad_norm": 0.06911790370941162, "global_step": 48029, "epoch": 453, "lr": 2.1540587647857215e-06} +{"train_loss": 0.016539476811885834, "grad_norm": 0.12121950834989548, "global_step": 48030, "epoch": 453, "lr": 2.1531983026547742e-06} +{"train_loss": 0.014787279069423676, "grad_norm": 0.10670720785856247, "global_step": 48031, "epoch": 453, "lr": 2.1523380086366896e-06} +{"train_loss": 0.014720559120178223, "grad_norm": 0.11271867156028748, "global_step": 48032, "epoch": 453, "lr": 2.1514778827344763e-06} +{"train_loss": 0.013797161169350147, "grad_norm": 0.1086481437087059, "global_step": 48033, "epoch": 453, "lr": 2.1506179249511706e-06} +{"train_loss": 0.01507692039012909, "grad_norm": 0.06171155720949173, "global_step": 48034, "epoch": 453, "lr": 2.14975813528977e-06} +{"train_loss": 0.013688432052731514, "grad_norm": 0.08596733212471008, "global_step": 48035, "epoch": 453, "lr": 2.1488985137533167e-06} +{"train_loss": 0.01539093628525734, "grad_norm": 0.13679566979408264, "global_step": 48036, "epoch": 453, "lr": 2.1480390603448252e-06} +{"train_loss": 0.019419072195887566, "grad_norm": 0.08871549367904663, "global_step": 48037, "epoch": 453, "lr": 2.147179775067315e-06} +{"train_loss": 0.01877705007791519, "grad_norm": 0.07837104052305222, "global_step": 48038, "epoch": 453, "lr": 2.1463206579238004e-06} +{"train_loss": 0.016187356784939766, "grad_norm": 0.07964035123586655, "global_step": 48039, "epoch": 453, "lr": 2.145461708917312e-06} +{"train_loss": 0.013908781111240387, "grad_norm": 0.0591379851102829, "global_step": 48040, "epoch": 453, "lr": 2.1446029280508485e-06} +{"train_loss": 0.0177106112241745, "grad_norm": 0.09310245513916016, "global_step": 48041, "epoch": 453, "lr": 2.1437443153274395e-06} +{"train_loss": 0.013881739228963852, "grad_norm": 0.0913134515285492, "global_step": 48042, "epoch": 453, "lr": 2.142885870750111e-06} +{"train_loss": 0.01499488577246666, "grad_norm": 0.06601132452487946, "global_step": 48043, "epoch": 453, "lr": 2.142027594321866e-06} +{"train_loss": 0.01678684912621975, "grad_norm": 0.0628311038017273, "global_step": 48044, "epoch": 453, "lr": 2.1411694860457186e-06} +{"train_loss": 0.015431491658091545, "grad_norm": 0.12588872015476227, "global_step": 48045, "epoch": 453, "lr": 2.1403115459246838e-06} +{"train_loss": 0.014573730528354645, "grad_norm": 0.10918467491865158, "global_step": 48046, "epoch": 453, "lr": 2.1394537739617914e-06} +{"train_loss": 0.013330873101949692, "grad_norm": 0.061488378793001175, "global_step": 48047, "epoch": 453, "lr": 2.1385961701600397e-06} +{"train_loss": 0.015068389475345612, "grad_norm": 0.08011379092931747, "global_step": 48048, "epoch": 453, "lr": 2.1377387345224486e-06} +{"train_loss": 0.012287681922316551, "grad_norm": 0.09623196721076965, "global_step": 48049, "epoch": 453, "lr": 2.1368814670520265e-06} +{"train_loss": 0.014891961589455605, "grad_norm": 0.07640030980110168, "global_step": 48050, "epoch": 453, "lr": 2.136024367751788e-06} +{"train_loss": 0.01581461727619171, "grad_norm": 0.08375677466392517, "global_step": 48051, "epoch": 453, "lr": 2.1351674366247465e-06} +{"train_loss": 0.014088267460465431, "grad_norm": 0.06674046069383621, "global_step": 48052, "epoch": 453, "lr": 2.134310673673906e-06} +{"train_loss": 0.01612761989235878, "grad_norm": 0.06725268065929413, "global_step": 48053, "epoch": 453, "lr": 2.1334540789022862e-06} +{"train_loss": 0.013424968346953392, "grad_norm": 0.06213485077023506, "global_step": 48054, "epoch": 453, "lr": 2.1325976523128953e-06} +{"train_loss": 0.016073692589998245, "grad_norm": 0.07004941254854202, "global_step": 48055, "epoch": 453, "lr": 2.1317413939087316e-06} +{"train_loss": 0.014922917820513248, "grad_norm": 0.06919950991868973, "global_step": 48056, "epoch": 453, "lr": 2.1308853036928145e-06} +{"train_loss": 0.012657711282372475, "grad_norm": 0.07645627856254578, "global_step": 48057, "epoch": 453, "lr": 2.1300293816681473e-06} +{"train_loss": 0.014065123163163662, "grad_norm": 0.07576987892389297, "global_step": 48058, "epoch": 453, "lr": 2.1291736278377385e-06} +{"train_loss": 0.014352602884173393, "grad_norm": 0.16171079874038696, "global_step": 48059, "epoch": 453, "lr": 2.1283180422045966e-06} +{"train_loss": 0.009748810902237892, "grad_norm": 0.09208587557077408, "global_step": 48060, "epoch": 453, "lr": 2.127462624771731e-06} +{"train_loss": 0.014892684295773506, "grad_norm": 0.084439218044281, "global_step": 48061, "epoch": 453, "lr": 2.1266073755421336e-06} +{"train_loss": 0.015397307462990284, "grad_norm": 0.05963394045829773, "global_step": 48062, "epoch": 453, "lr": 2.125752294518818e-06} +{"train_loss": 0.014197103679180145, "grad_norm": 0.06619234383106232, "global_step": 48063, "epoch": 453, "lr": 2.124897381704799e-06} +{"train_loss": 0.018690552562475204, "grad_norm": 0.07360268384218216, "global_step": 48064, "epoch": 453, "lr": 2.124042637103063e-06} +{"train_loss": 0.014883710071444511, "grad_norm": 0.05697346478700638, "global_step": 48065, "epoch": 453, "lr": 2.123188060716619e-06} +{"train_loss": 0.012633787468075752, "grad_norm": 0.09824404865503311, "global_step": 48066, "epoch": 453, "lr": 2.122333652548475e-06} +{"train_loss": 0.014865197241306305, "grad_norm": 0.06449060142040253, "global_step": 48067, "epoch": 453, "lr": 2.1214794126016346e-06} +{"train_loss": 0.011421818286180496, "grad_norm": 0.05862262099981308, "global_step": 48068, "epoch": 453, "lr": 2.1206253408790843e-06} +{"train_loss": 0.01458512432873249, "grad_norm": 0.05863795056939125, "global_step": 48069, "epoch": 453, "lr": 2.119771437383833e-06} +{"train_loss": 0.016909010708332062, "grad_norm": 0.10142114013433456, "global_step": 48070, "epoch": 453, "lr": 2.118917702118889e-06} +{"train_loss": 0.013502142392098904, "grad_norm": 0.12948490679264069, "global_step": 48071, "epoch": 453, "lr": 2.1180641350872386e-06} +{"train_loss": 0.01250919047743082, "grad_norm": 0.05857020989060402, "global_step": 48072, "epoch": 453, "lr": 2.117210736291897e-06} +{"train_loss": 0.015075073577463627, "grad_norm": 0.10625872015953064, "global_step": 48073, "epoch": 453, "lr": 2.1163575057358443e-06} +{"train_loss": 0.013068285770714283, "grad_norm": 0.06999097764492035, "global_step": 48074, "epoch": 453, "lr": 2.1155044434220948e-06} +{"train_loss": 0.013161951676011086, "grad_norm": 0.06992589682340622, "global_step": 48075, "epoch": 453, "lr": 2.1146515493536356e-06} +{"train_loss": 0.016058338806033134, "grad_norm": 0.09864264726638794, "global_step": 48076, "epoch": 453, "lr": 2.1137988235334584e-06} +{"train_loss": 0.016210494562983513, "grad_norm": 0.07053160667419434, "global_step": 48077, "epoch": 453, "lr": 2.112946265964577e-06} +{"train_loss": 0.013406665064394474, "grad_norm": 0.09219064563512802, "global_step": 48078, "epoch": 453, "lr": 2.1120938766499787e-06} +{"train_loss": 0.016904711723327637, "grad_norm": 0.0601290762424469, "global_step": 48079, "epoch": 453, "lr": 2.1112416555926497e-06} +{"train_loss": 0.015912216156721115, "grad_norm": 0.07031217217445374, "global_step": 48080, "epoch": 453, "lr": 2.1103896027955926e-06} +{"train_loss": 0.01322631724178791, "grad_norm": 0.06328815966844559, "global_step": 48081, "epoch": 453, "lr": 2.1095377182618e-06} +{"train_loss": 0.015410361811518669, "grad_norm": 0.1106576919555664, "global_step": 48082, "epoch": 453, "lr": 2.108686001994259e-06} +{"train_loss": 0.01658109575510025, "grad_norm": 0.11584306508302689, "global_step": 48083, "epoch": 453, "lr": 2.1078344539959772e-06} +{"train_loss": 0.016544509679079056, "grad_norm": 0.06820867210626602, "global_step": 48084, "epoch": 453, "lr": 2.1069830742699415e-06} +{"train_loss": 0.014770891517400742, "grad_norm": 0.07752451300621033, "global_step": 48085, "epoch": 453, "lr": 2.1061318628191272e-06} +{"train_loss": 0.013134414330124855, "grad_norm": 0.1339179277420044, "global_step": 48086, "epoch": 453, "lr": 2.105280819646538e-06} +{"train_loss": 0.016291698440909386, "grad_norm": 0.07366151362657547, "global_step": 48087, "epoch": 453, "lr": 2.104429944755171e-06} +{"train_loss": 0.014915421605110168, "grad_norm": 0.08378655463457108, "global_step": 48088, "epoch": 453, "lr": 2.103579238148001e-06} +{"train_loss": 0.014750656671822071, "grad_norm": 0.05706328526139259, "global_step": 48089, "epoch": 453, "lr": 2.1027286998280215e-06} +{"train_loss": 0.015735585242509842, "grad_norm": 0.10339101403951645, "global_step": 48090, "epoch": 453, "lr": 2.1018783297982347e-06} +{"train_loss": 0.01448983233422041, "grad_norm": 0.08832085132598877, "global_step": 48091, "epoch": 453, "lr": 2.1010281280616164e-06} +{"train_loss": 0.013176757842302322, "grad_norm": 0.0634179636836052, "global_step": 48092, "epoch": 453, "lr": 2.1001780946211413e-06} +{"train_loss": 0.015633510425686836, "grad_norm": 0.0879560261964798, "global_step": 48093, "epoch": 453, "lr": 2.099328229479819e-06} +{"train_loss": 0.012526370584964752, "grad_norm": 0.1038123369216919, "global_step": 48094, "epoch": 453, "lr": 2.098478532640624e-06} +{"train_loss": 0.014993098564445972, "grad_norm": 0.10394147038459778, "global_step": 48095, "epoch": 453, "lr": 2.0976290041065493e-06} +{"train_loss": 0.013295425102114677, "grad_norm": 0.05837458744645119, "global_step": 48096, "epoch": 453, "lr": 2.0967796438805644e-06} +{"train_loss": 0.014495490118861198, "grad_norm": 0.11301788687705994, "global_step": 48097, "epoch": 453, "lr": 2.095930451965672e-06} +{"train_loss": 0.016068503260612488, "grad_norm": 0.07845734804868698, "global_step": 48098, "epoch": 453, "lr": 2.0950814283648366e-06} +{"train_loss": 0.01643885299563408, "grad_norm": 0.08145105093717575, "global_step": 48099, "epoch": 453, "lr": 2.0942325730810565e-06} +{"train_loss": 0.014211558736860752, "grad_norm": 0.07924500107765198, "global_step": 48100, "epoch": 453, "lr": 2.0933838861173115e-06} +{"train_loss": 0.01594688557088375, "grad_norm": 0.11356110870838165, "global_step": 48101, "epoch": 453, "lr": 2.0925353674765835e-06} +{"train_loss": 0.014694435521960258, "grad_norm": 0.06479169428348541, "global_step": 48102, "epoch": 453, "lr": 2.0916870171618473e-06} +{"train_loss": 0.015499971807003021, "grad_norm": 0.0906129702925682, "global_step": 48103, "epoch": 453, "lr": 2.0908388351760845e-06} +{"train_loss": 0.015407351776957512, "grad_norm": 0.06539473682641983, "global_step": 48104, "epoch": 453, "lr": 2.089990821522292e-06} +{"train_loss": 0.016463905572891235, "grad_norm": 0.1098046749830246, "global_step": 48105, "epoch": 453, "lr": 2.089142976203423e-06} +{"train_loss": 0.015453946776688099, "grad_norm": 0.0808526799082756, "global_step": 48106, "epoch": 453, "lr": 2.08829529922247e-06} +{"train_loss": 0.016553236171603203, "grad_norm": 0.10758674144744873, "global_step": 48107, "epoch": 453, "lr": 2.0874477905824195e-06} +{"train_loss": 0.014740798622369766, "grad_norm": 0.10217370092868805, "global_step": 48108, "epoch": 453, "lr": 2.0866004502862414e-06} +{"train_loss": 0.013972594402730465, "grad_norm": 0.06186624616384506, "global_step": 48109, "epoch": 453, "lr": 2.085753278336905e-06} +{"train_loss": 0.015161932446062565, "grad_norm": 0.08983805775642395, "global_step": 48110, "epoch": 453, "lr": 2.084906274737397e-06} +{"train_loss": 0.015547254122793674, "grad_norm": 0.08871053159236908, "global_step": 48111, "epoch": 453, "lr": 2.084059439490693e-06} +{"train_loss": 0.01478955615311861, "grad_norm": 0.0931168794631958, "global_step": 48112, "epoch": 453, "lr": 2.083212772599763e-06} +{"train_loss": 0.015141937881708145, "grad_norm": 0.0617607980966568, "global_step": 48113, "epoch": 453, "lr": 2.0823662740675877e-06} +{"train_loss": 0.01440556813031435, "grad_norm": 0.06604784727096558, "global_step": 48114, "epoch": 453, "lr": 2.081519943897142e-06} +{"train_loss": 0.013407016173005104, "grad_norm": 0.061418697237968445, "global_step": 48115, "epoch": 453, "lr": 2.0806737820913858e-06} +{"train_loss": 0.01534188911318779, "grad_norm": 0.0741768479347229, "global_step": 48116, "epoch": 453, "lr": 2.0798277886533046e-06} +{"train_loss": 0.01396232657134533, "grad_norm": 0.057232100516557693, "global_step": 48117, "epoch": 453, "lr": 2.0789819635858743e-06} +{"train_loss": 0.016373472288250923, "grad_norm": 0.05921275541186333, "global_step": 48118, "epoch": 453, "lr": 2.078136306892059e-06} +{"train_loss": 0.014404169283807278, "grad_norm": 0.09775850921869278, "global_step": 48119, "epoch": 453, "lr": 2.077290818574834e-06} +{"train_loss": 0.014608565717935562, "grad_norm": 0.08754726499319077, "global_step": 48120, "epoch": 453, "lr": 2.0764454986371583e-06} +{"train_loss": 0.01591617614030838, "grad_norm": 0.062406379729509354, "global_step": 48121, "epoch": 453, "lr": 2.075600347082024e-06} +{"train_loss": 0.01545141264796257, "grad_norm": 0.07516239583492279, "global_step": 48122, "epoch": 453, "lr": 2.0747553639123786e-06} +{"train_loss": 0.014944744507237425, "grad_norm": 0.13897380232810974, "global_step": 48123, "epoch": 453, "lr": 2.0739105491312027e-06, "val_loss": 0.019850440323352814} +{"train_loss": 0.0136032123118639, "grad_norm": 0.07825607061386108, "global_step": 48124, "epoch": 454, "lr": 2.073065902741472e-06} +{"train_loss": 0.015006262809038162, "grad_norm": 0.05950440466403961, "global_step": 48125, "epoch": 454, "lr": 2.0722214247461404e-06} +{"train_loss": 0.013898124918341637, "grad_norm": 0.0951184406876564, "global_step": 48126, "epoch": 454, "lr": 2.071377115148171e-06} +{"train_loss": 0.015525702387094498, "grad_norm": 0.08807151764631271, "global_step": 48127, "epoch": 454, "lr": 2.0705329739505454e-06} +{"train_loss": 0.016003694385290146, "grad_norm": 0.07506243884563446, "global_step": 48128, "epoch": 454, "lr": 2.069689001156222e-06} +{"train_loss": 0.01538810133934021, "grad_norm": 0.12304632365703583, "global_step": 48129, "epoch": 454, "lr": 2.0688451967681653e-06} +{"train_loss": 0.0142898540943861, "grad_norm": 0.06021479144692421, "global_step": 48130, "epoch": 454, "lr": 2.06800156078934e-06} +{"train_loss": 0.014801973477005959, "grad_norm": 0.0713067427277565, "global_step": 48131, "epoch": 454, "lr": 2.0671580932227208e-06} +{"train_loss": 0.014583288691937923, "grad_norm": 0.10459103435277939, "global_step": 48132, "epoch": 454, "lr": 2.0663147940712614e-06} +{"train_loss": 0.014433344826102257, "grad_norm": 0.10099025815725327, "global_step": 48133, "epoch": 454, "lr": 2.06547166333792e-06} +{"train_loss": 0.01711311936378479, "grad_norm": 0.102445088326931, "global_step": 48134, "epoch": 454, "lr": 2.064628701025667e-06} +{"train_loss": 0.015758223831653595, "grad_norm": 0.09136631339788437, "global_step": 48135, "epoch": 454, "lr": 2.0637859071374665e-06} +{"train_loss": 0.013727441430091858, "grad_norm": 0.0622064583003521, "global_step": 48136, "epoch": 454, "lr": 2.062943281676266e-06} +{"train_loss": 0.014167962595820427, "grad_norm": 0.08003131300210953, "global_step": 48137, "epoch": 454, "lr": 2.0621008246450467e-06} +{"train_loss": 0.014375898987054825, "grad_norm": 0.056769777089357376, "global_step": 48138, "epoch": 454, "lr": 2.0612585360467562e-06} +{"train_loss": 0.015565077774226665, "grad_norm": 0.05553169175982475, "global_step": 48139, "epoch": 454, "lr": 2.0604164158843474e-06} +{"train_loss": 0.013791348785161972, "grad_norm": 0.06913229823112488, "global_step": 48140, "epoch": 454, "lr": 2.059574464160785e-06} +{"train_loss": 0.016271445900201797, "grad_norm": 0.12476566433906555, "global_step": 48141, "epoch": 454, "lr": 2.0587326808790388e-06} +{"train_loss": 0.014302784577012062, "grad_norm": 0.05905300751328468, "global_step": 48142, "epoch": 454, "lr": 2.0578910660420503e-06} +{"train_loss": 0.01653120666742325, "grad_norm": 0.061390116810798645, "global_step": 48143, "epoch": 454, "lr": 2.05704961965279e-06} +{"train_loss": 0.015862930566072464, "grad_norm": 0.0958137959241867, "global_step": 48144, "epoch": 454, "lr": 2.0562083417142e-06} +{"train_loss": 0.01274942234158516, "grad_norm": 0.07649319618940353, "global_step": 48145, "epoch": 454, "lr": 2.0553672322292494e-06} +{"train_loss": 0.012262744829058647, "grad_norm": 0.055102311074733734, "global_step": 48146, "epoch": 454, "lr": 2.0545262912008812e-06} +{"train_loss": 0.013734623789787292, "grad_norm": 0.09050330519676208, "global_step": 48147, "epoch": 454, "lr": 2.053685518632059e-06} +{"train_loss": 0.014654643833637238, "grad_norm": 0.12480613589286804, "global_step": 48148, "epoch": 454, "lr": 2.0528449145257366e-06} +{"train_loss": 0.014817272312939167, "grad_norm": 0.058135729283094406, "global_step": 48149, "epoch": 454, "lr": 2.0520044788848724e-06} +{"train_loss": 0.015327688306570053, "grad_norm": 0.1567947119474411, "global_step": 48150, "epoch": 454, "lr": 2.0511642117124032e-06} +{"train_loss": 0.015628743916749954, "grad_norm": 0.09581420570611954, "global_step": 48151, "epoch": 454, "lr": 2.050324113011287e-06} +{"train_loss": 0.014130696654319763, "grad_norm": 0.12465853244066238, "global_step": 48152, "epoch": 454, "lr": 2.049484182784489e-06} +{"train_loss": 0.01770714856684208, "grad_norm": 0.17506372928619385, "global_step": 48153, "epoch": 454, "lr": 2.04864442103494e-06} +{"train_loss": 0.016235139220952988, "grad_norm": 0.08092914521694183, "global_step": 48154, "epoch": 454, "lr": 2.047804827765615e-06} +{"train_loss": 0.015530699864029884, "grad_norm": 0.06429888308048248, "global_step": 48155, "epoch": 454, "lr": 2.046965402979445e-06} +{"train_loss": 0.013318798504769802, "grad_norm": 0.10011143237352371, "global_step": 48156, "epoch": 454, "lr": 2.046126146679378e-06} +{"train_loss": 0.015322921797633171, "grad_norm": 0.13794024288654327, "global_step": 48157, "epoch": 454, "lr": 2.045287058868367e-06} +{"train_loss": 0.0187794491648674, "grad_norm": 0.12736523151397705, "global_step": 48158, "epoch": 454, "lr": 2.044448139549371e-06} +{"train_loss": 0.01400865986943245, "grad_norm": 0.05658154934644699, "global_step": 48159, "epoch": 454, "lr": 2.043609388725326e-06} +{"train_loss": 0.01296575739979744, "grad_norm": 0.07595250010490417, "global_step": 48160, "epoch": 454, "lr": 2.042770806399186e-06} +{"train_loss": 0.018158096820116043, "grad_norm": 0.08688770979642868, "global_step": 48161, "epoch": 454, "lr": 2.041932392573881e-06} +{"train_loss": 0.013806742615997791, "grad_norm": 0.0890214666724205, "global_step": 48162, "epoch": 454, "lr": 2.0410941472523814e-06} +{"train_loss": 0.016117412596940994, "grad_norm": 0.10717735439538956, "global_step": 48163, "epoch": 454, "lr": 2.0402560704376127e-06} +{"train_loss": 0.013057449832558632, "grad_norm": 0.06479118764400482, "global_step": 48164, "epoch": 454, "lr": 2.039418162132528e-06} +{"train_loss": 0.013894779607653618, "grad_norm": 0.08481280505657196, "global_step": 48165, "epoch": 454, "lr": 2.038580422340075e-06} +{"train_loss": 0.016019307076931, "grad_norm": 0.07479126751422882, "global_step": 48166, "epoch": 454, "lr": 2.0377428510631955e-06} +{"train_loss": 0.013514192774891853, "grad_norm": 0.08641984313726425, "global_step": 48167, "epoch": 454, "lr": 2.036905448304821e-06} +{"train_loss": 0.014165658503770828, "grad_norm": 0.10459861904382706, "global_step": 48168, "epoch": 454, "lr": 2.0360682140679045e-06} +{"train_loss": 0.015162402763962746, "grad_norm": 0.059938687831163406, "global_step": 48169, "epoch": 454, "lr": 2.035231148355388e-06} +{"train_loss": 0.015591510571539402, "grad_norm": 0.08498270064592361, "global_step": 48170, "epoch": 454, "lr": 2.0343942511702074e-06} +{"train_loss": 0.015591765753924847, "grad_norm": 0.08589490503072739, "global_step": 48171, "epoch": 454, "lr": 2.0335575225153003e-06} +{"train_loss": 0.013622750528156757, "grad_norm": 0.0845007598400116, "global_step": 48172, "epoch": 454, "lr": 2.0327209623936307e-06} +{"train_loss": 0.015461009927093983, "grad_norm": 0.07326032221317291, "global_step": 48173, "epoch": 454, "lr": 2.0318845708081013e-06} +{"train_loss": 0.014074393548071384, "grad_norm": 0.0847671777009964, "global_step": 48174, "epoch": 454, "lr": 2.0310483477616714e-06} +{"train_loss": 0.01591036655008793, "grad_norm": 0.09393715113401413, "global_step": 48175, "epoch": 454, "lr": 2.0302122932572774e-06} +{"train_loss": 0.013912811875343323, "grad_norm": 0.05855613946914673, "global_step": 48176, "epoch": 454, "lr": 2.0293764072978618e-06} +{"train_loss": 0.01642005145549774, "grad_norm": 0.11327806860208511, "global_step": 48177, "epoch": 454, "lr": 2.0285406898863436e-06} +{"train_loss": 0.016420818865299225, "grad_norm": 0.08443737030029297, "global_step": 48178, "epoch": 454, "lr": 2.0277051410256875e-06} +{"train_loss": 0.014515634626150131, "grad_norm": 0.0645487979054451, "global_step": 48179, "epoch": 454, "lr": 2.0268697607188023e-06} +{"train_loss": 0.015634119510650635, "grad_norm": 0.07038144022226334, "global_step": 48180, "epoch": 454, "lr": 2.026034548968636e-06} +{"train_loss": 0.012521447613835335, "grad_norm": 0.08684402704238892, "global_step": 48181, "epoch": 454, "lr": 2.025199505778119e-06} +{"train_loss": 0.014578873291611671, "grad_norm": 0.08685912936925888, "global_step": 48182, "epoch": 454, "lr": 2.024364631150194e-06} +{"train_loss": 0.014664699323475361, "grad_norm": 0.07199432700872421, "global_step": 48183, "epoch": 454, "lr": 2.023529925087775e-06} +{"train_loss": 0.01394887175410986, "grad_norm": 0.10391518473625183, "global_step": 48184, "epoch": 454, "lr": 2.0226953875938205e-06} +{"train_loss": 0.01430769544094801, "grad_norm": 0.0770275816321373, "global_step": 48185, "epoch": 454, "lr": 2.0218610186712396e-06} +{"train_loss": 0.018640484660863876, "grad_norm": 0.07426364719867706, "global_step": 48186, "epoch": 454, "lr": 2.0210268183229854e-06} +{"train_loss": 0.014919617213308811, "grad_norm": 0.10043928772211075, "global_step": 48187, "epoch": 454, "lr": 2.0201927865519665e-06} +{"train_loss": 0.015077812597155571, "grad_norm": 0.079499252140522, "global_step": 48188, "epoch": 454, "lr": 2.0193589233611256e-06} +{"train_loss": 0.015270453877747059, "grad_norm": 0.06554978340864182, "global_step": 48189, "epoch": 454, "lr": 2.0185252287533984e-06} +{"train_loss": 0.013073749840259552, "grad_norm": 0.11375366896390915, "global_step": 48190, "epoch": 454, "lr": 2.0176917027317055e-06} +{"train_loss": 0.01390185859054327, "grad_norm": 0.06546951085329056, "global_step": 48191, "epoch": 454, "lr": 2.016858345298972e-06} +{"train_loss": 0.014230364933609962, "grad_norm": 0.07166421413421631, "global_step": 48192, "epoch": 454, "lr": 2.016025156458129e-06} +{"train_loss": 0.015909776091575623, "grad_norm": 0.07438631355762482, "global_step": 48193, "epoch": 454, "lr": 2.0151921362121128e-06} +{"train_loss": 0.01574058085680008, "grad_norm": 0.0701172947883606, "global_step": 48194, "epoch": 454, "lr": 2.0143592845638383e-06} +{"train_loss": 0.015520155429840088, "grad_norm": 0.07251090556383133, "global_step": 48195, "epoch": 454, "lr": 2.0135266015162413e-06} +{"train_loss": 0.014235551469027996, "grad_norm": 0.0646171122789383, "global_step": 48196, "epoch": 454, "lr": 2.0126940870722477e-06} +{"train_loss": 0.013653479516506195, "grad_norm": 0.09594767540693283, "global_step": 48197, "epoch": 454, "lr": 2.0118617412347664e-06} +{"train_loss": 0.013184208422899246, "grad_norm": 0.04876105859875679, "global_step": 48198, "epoch": 454, "lr": 2.0110295640067335e-06} +{"train_loss": 0.016162533313035965, "grad_norm": 0.07769494503736496, "global_step": 48199, "epoch": 454, "lr": 2.01019755539108e-06} +{"train_loss": 0.014136230573058128, "grad_norm": 0.10970992594957352, "global_step": 48200, "epoch": 454, "lr": 2.0093657153907206e-06} +{"train_loss": 0.013659682124853134, "grad_norm": 0.08354344964027405, "global_step": 48201, "epoch": 454, "lr": 2.008534044008581e-06} +{"train_loss": 0.015366748906672001, "grad_norm": 0.06352292001247406, "global_step": 48202, "epoch": 454, "lr": 2.0077025412475747e-06} +{"train_loss": 0.014545531943440437, "grad_norm": 0.06693298369646072, "global_step": 48203, "epoch": 454, "lr": 2.0068712071106387e-06} +{"train_loss": 0.015701899304986, "grad_norm": 0.0707143023610115, "global_step": 48204, "epoch": 454, "lr": 2.006040041600682e-06} +{"train_loss": 0.0168165173381567, "grad_norm": 0.10929585248231888, "global_step": 48205, "epoch": 454, "lr": 2.005209044720624e-06} +{"train_loss": 0.01496860384941101, "grad_norm": 0.12346950173377991, "global_step": 48206, "epoch": 454, "lr": 2.004378216473396e-06} +{"train_loss": 0.014967561699450016, "grad_norm": 0.09457137435674667, "global_step": 48207, "epoch": 454, "lr": 2.003547556861912e-06} +{"train_loss": 0.013330354355275631, "grad_norm": 0.0941108986735344, "global_step": 48208, "epoch": 454, "lr": 2.0027170658890816e-06} +{"train_loss": 0.017192436382174492, "grad_norm": 0.0783480852842331, "global_step": 48209, "epoch": 454, "lr": 2.0018867435578294e-06} +{"train_loss": 0.013525099493563175, "grad_norm": 0.0608602799475193, "global_step": 48210, "epoch": 454, "lr": 2.001056589871081e-06} +{"train_loss": 0.01601284369826317, "grad_norm": 0.13925422728061676, "global_step": 48211, "epoch": 454, "lr": 2.0002266048317396e-06} +{"train_loss": 0.014512249268591404, "grad_norm": 0.11502768844366074, "global_step": 48212, "epoch": 454, "lr": 1.9993967884427256e-06} +{"train_loss": 0.015818079933524132, "grad_norm": 0.08707839250564575, "global_step": 48213, "epoch": 454, "lr": 1.9985671407069693e-06} +{"train_loss": 0.013862095773220062, "grad_norm": 0.06736703962087631, "global_step": 48214, "epoch": 454, "lr": 1.997737661627358e-06} +{"train_loss": 0.0153639642521739, "grad_norm": 0.07983812689781189, "global_step": 48215, "epoch": 454, "lr": 1.9969083512068275e-06} +{"train_loss": 0.01497999019920826, "grad_norm": 0.07764558494091034, "global_step": 48216, "epoch": 454, "lr": 1.9960792094482872e-06} +{"train_loss": 0.01812065578997135, "grad_norm": 0.10658256709575653, "global_step": 48217, "epoch": 454, "lr": 1.99525023635464e-06} +{"train_loss": 0.014491555280983448, "grad_norm": 0.0978931337594986, "global_step": 48218, "epoch": 454, "lr": 1.9944214319288113e-06} +{"train_loss": 0.013726729899644852, "grad_norm": 0.09347942471504211, "global_step": 48219, "epoch": 454, "lr": 1.99359279617371e-06} +{"train_loss": 0.015125580132007599, "grad_norm": 0.06489135324954987, "global_step": 48220, "epoch": 454, "lr": 1.9927643290922502e-06} +{"train_loss": 0.01801072061061859, "grad_norm": 0.06415966898202896, "global_step": 48221, "epoch": 454, "lr": 1.9919360306873304e-06} +{"train_loss": 0.015847548842430115, "grad_norm": 0.08169955015182495, "global_step": 48222, "epoch": 454, "lr": 1.9911079009618696e-06} +{"train_loss": 0.013189194723963737, "grad_norm": 0.100834921002388, "global_step": 48223, "epoch": 454, "lr": 1.990279939918782e-06} +{"train_loss": 0.014496790245175362, "grad_norm": 0.07289966940879822, "global_step": 48224, "epoch": 454, "lr": 1.989452147560966e-06} +{"train_loss": 0.014996165409684181, "grad_norm": 0.08690881729125977, "global_step": 48225, "epoch": 454, "lr": 1.988624523891347e-06} +{"train_loss": 0.015777554363012314, "grad_norm": 0.06194349378347397, "global_step": 48226, "epoch": 454, "lr": 1.987797068912811e-06} +{"train_loss": 0.012400570325553417, "grad_norm": 0.07289745658636093, "global_step": 48227, "epoch": 454, "lr": 1.9869697826282842e-06} +{"train_loss": 0.014744644984602928, "grad_norm": 0.07743065804243088, "global_step": 48228, "epoch": 454, "lr": 1.986142665040658e-06} +{"train_loss": 0.01496448774628763, "grad_norm": 0.156488835811615, "global_step": 48229, "epoch": 454, "lr": 1.985315716152847e-06, "val_loss": 0.01916530542075634} +{"train_loss": 0.01363038644194603, "grad_norm": 0.06419064104557037, "global_step": 48230, "epoch": 455, "lr": 1.9844889359677597e-06} +{"train_loss": 0.01608356088399887, "grad_norm": 0.06447254866361618, "global_step": 48231, "epoch": 455, "lr": 1.9836623244882945e-06} +{"train_loss": 0.016225796192884445, "grad_norm": 0.06927529722452164, "global_step": 48232, "epoch": 455, "lr": 1.9828358817173542e-06} +{"train_loss": 0.013246450573205948, "grad_norm": 0.08739485591650009, "global_step": 48233, "epoch": 455, "lr": 1.9820096076578478e-06} +{"train_loss": 0.01652122475206852, "grad_norm": 0.10380967706441879, "global_step": 48234, "epoch": 455, "lr": 1.9811835023126833e-06} +{"train_loss": 0.013688327744603157, "grad_norm": 0.0680496022105217, "global_step": 48235, "epoch": 455, "lr": 1.980357565684754e-06} +{"train_loss": 0.016139404848217964, "grad_norm": 0.0753645971417427, "global_step": 48236, "epoch": 455, "lr": 1.9795317977769625e-06} +{"train_loss": 0.016166776418685913, "grad_norm": 0.07052168995141983, "global_step": 48237, "epoch": 455, "lr": 1.978706198592223e-06} +{"train_loss": 0.014884551987051964, "grad_norm": 0.10928696393966675, "global_step": 48238, "epoch": 455, "lr": 1.9778807681334164e-06} +{"train_loss": 0.016083884984254837, "grad_norm": 0.07439908385276794, "global_step": 48239, "epoch": 455, "lr": 1.9770555064034467e-06} +{"train_loss": 0.013984454795718193, "grad_norm": 0.13315872848033905, "global_step": 48240, "epoch": 455, "lr": 1.9762304134052336e-06} +{"train_loss": 0.014553861692547798, "grad_norm": 0.07170738279819489, "global_step": 48241, "epoch": 455, "lr": 1.975405489141652e-06} +{"train_loss": 0.014065857045352459, "grad_norm": 0.06066470593214035, "global_step": 48242, "epoch": 455, "lr": 1.974580733615611e-06} +{"train_loss": 0.015994781628251076, "grad_norm": 0.06509481370449066, "global_step": 48243, "epoch": 455, "lr": 1.973756146830008e-06} +{"train_loss": 0.01484947744756937, "grad_norm": 0.10123158991336823, "global_step": 48244, "epoch": 455, "lr": 1.972931728787747e-06} +{"train_loss": 0.014535818248987198, "grad_norm": 0.06490501761436462, "global_step": 48245, "epoch": 455, "lr": 1.9721074794917084e-06} +{"train_loss": 0.013814630918204784, "grad_norm": 0.08005804568529129, "global_step": 48246, "epoch": 455, "lr": 1.971283398944801e-06} +{"train_loss": 0.01355687715113163, "grad_norm": 0.05079168081283569, "global_step": 48247, "epoch": 455, "lr": 1.9704594871499173e-06} +{"train_loss": 0.015542314387857914, "grad_norm": 0.05996456742286682, "global_step": 48248, "epoch": 455, "lr": 1.9696357441099544e-06} +{"train_loss": 0.017531517893075943, "grad_norm": 0.15727397799491882, "global_step": 48249, "epoch": 455, "lr": 1.9688121698277995e-06} +{"train_loss": 0.012564636766910553, "grad_norm": 0.08780747652053833, "global_step": 48250, "epoch": 455, "lr": 1.9679887643063442e-06} +{"train_loss": 0.016014572232961655, "grad_norm": 0.08752116560935974, "global_step": 48251, "epoch": 455, "lr": 1.967165527548498e-06} +{"train_loss": 0.016383469104766846, "grad_norm": 0.09326549619436264, "global_step": 48252, "epoch": 455, "lr": 1.9663424595571356e-06} +{"train_loss": 0.017325598746538162, "grad_norm": 0.0796750858426094, "global_step": 48253, "epoch": 455, "lr": 1.9655195603351606e-06} +{"train_loss": 0.014168517664074898, "grad_norm": 0.1319984495639801, "global_step": 48254, "epoch": 455, "lr": 1.9646968298854707e-06} +{"train_loss": 0.01575043611228466, "grad_norm": 0.060557398945093155, "global_step": 48255, "epoch": 455, "lr": 1.96387426821093e-06} +{"train_loss": 0.015603158622980118, "grad_norm": 0.0709555447101593, "global_step": 48256, "epoch": 455, "lr": 1.9630518753144478e-06} +{"train_loss": 0.014640210196375847, "grad_norm": 0.08242304623126984, "global_step": 48257, "epoch": 455, "lr": 1.9622296511989156e-06} +{"train_loss": 0.01414645928889513, "grad_norm": 0.0807691290974617, "global_step": 48258, "epoch": 455, "lr": 1.9614075958672153e-06} +{"train_loss": 0.01526815164834261, "grad_norm": 0.08789495378732681, "global_step": 48259, "epoch": 455, "lr": 1.9605857093222326e-06} +{"train_loss": 0.014207983389496803, "grad_norm": 0.07169762253761292, "global_step": 48260, "epoch": 455, "lr": 1.959763991566865e-06} +{"train_loss": 0.016013197600841522, "grad_norm": 0.0646880492568016, "global_step": 48261, "epoch": 455, "lr": 1.9589424426040006e-06} +{"train_loss": 0.014715943485498428, "grad_norm": 0.06325923651456833, "global_step": 48262, "epoch": 455, "lr": 1.9581210624365077e-06} +{"train_loss": 0.012944426387548447, "grad_norm": 0.10024285316467285, "global_step": 48263, "epoch": 455, "lr": 1.9572998510672903e-06} +{"train_loss": 0.013737204484641552, "grad_norm": 0.10386881977319717, "global_step": 48264, "epoch": 455, "lr": 1.956478808499229e-06} +{"train_loss": 0.016385210677981377, "grad_norm": 0.06355676800012589, "global_step": 48265, "epoch": 455, "lr": 1.9556579347352054e-06} +{"train_loss": 0.013971095904707909, "grad_norm": 0.059566203504800797, "global_step": 48266, "epoch": 455, "lr": 1.954837229778106e-06} +{"train_loss": 0.014777664095163345, "grad_norm": 0.09256183356046677, "global_step": 48267, "epoch": 455, "lr": 1.9540166936308114e-06} +{"train_loss": 0.013367054983973503, "grad_norm": 0.12694023549556732, "global_step": 48268, "epoch": 455, "lr": 1.9531963262962195e-06} +{"train_loss": 0.013357983902096748, "grad_norm": 0.08376847207546234, "global_step": 48269, "epoch": 455, "lr": 1.952376127777189e-06} +{"train_loss": 0.01532864011824131, "grad_norm": 0.09645737707614899, "global_step": 48270, "epoch": 455, "lr": 1.9515560980766123e-06} +{"train_loss": 0.01442575454711914, "grad_norm": 0.07412445545196533, "global_step": 48271, "epoch": 455, "lr": 1.9507362371973814e-06} +{"train_loss": 0.015258710831403732, "grad_norm": 0.08476939797401428, "global_step": 48272, "epoch": 455, "lr": 1.9499165451423608e-06} +{"train_loss": 0.013126427307724953, "grad_norm": 0.10566306114196777, "global_step": 48273, "epoch": 455, "lr": 1.9490970219144366e-06} +{"train_loss": 0.013931989669799805, "grad_norm": 0.09790099412202835, "global_step": 48274, "epoch": 455, "lr": 1.948277667516496e-06} +{"train_loss": 0.013278213329613209, "grad_norm": 0.08629197627305984, "global_step": 48275, "epoch": 455, "lr": 1.947458481951403e-06} +{"train_loss": 0.01641746424138546, "grad_norm": 0.07065022736787796, "global_step": 48276, "epoch": 455, "lr": 1.9466394652220387e-06} +{"train_loss": 0.01357356645166874, "grad_norm": 0.07986044138669968, "global_step": 48277, "epoch": 455, "lr": 1.94582061733129e-06} +{"train_loss": 0.015976570546627045, "grad_norm": 0.07522077858448029, "global_step": 48278, "epoch": 455, "lr": 1.945001938282043e-06} +{"train_loss": 0.01630999520421028, "grad_norm": 0.20481087267398834, "global_step": 48279, "epoch": 455, "lr": 1.944183428077145e-06} +{"train_loss": 0.014691954478621483, "grad_norm": 0.06280341744422913, "global_step": 48280, "epoch": 455, "lr": 1.9433650867194898e-06} +{"train_loss": 0.015234099701046944, "grad_norm": 0.11093249917030334, "global_step": 48281, "epoch": 455, "lr": 1.9425469142119566e-06} +{"train_loss": 0.014498312957584858, "grad_norm": 0.0911526158452034, "global_step": 48282, "epoch": 455, "lr": 1.9417289105574053e-06} +{"train_loss": 0.016291461884975433, "grad_norm": 0.08245960623025894, "global_step": 48283, "epoch": 455, "lr": 1.9409110757587166e-06} +{"train_loss": 0.016201557591557503, "grad_norm": 0.14397060871124268, "global_step": 48284, "epoch": 455, "lr": 1.940093409818777e-06} +{"train_loss": 0.016390599310398102, "grad_norm": 0.08364816755056381, "global_step": 48285, "epoch": 455, "lr": 1.939275912740446e-06} +{"train_loss": 0.014889121055603027, "grad_norm": 0.06852632761001587, "global_step": 48286, "epoch": 455, "lr": 1.9384585845265977e-06} +{"train_loss": 0.014956165105104446, "grad_norm": 0.07382982224225998, "global_step": 48287, "epoch": 455, "lr": 1.937641425180098e-06} +{"train_loss": 0.012784060090780258, "grad_norm": 0.05895243212580681, "global_step": 48288, "epoch": 455, "lr": 1.9368244347038324e-06} +{"train_loss": 0.017508022487163544, "grad_norm": 0.06506306678056717, "global_step": 48289, "epoch": 455, "lr": 1.93600761310066e-06} +{"train_loss": 0.014355801045894623, "grad_norm": 0.11568016558885574, "global_step": 48290, "epoch": 455, "lr": 1.9351909603734563e-06} +{"train_loss": 0.014746563509106636, "grad_norm": 0.11885696649551392, "global_step": 48291, "epoch": 455, "lr": 1.934374476525086e-06} +{"train_loss": 0.013606146909296513, "grad_norm": 0.08007880300283432, "global_step": 48292, "epoch": 455, "lr": 1.933558161558424e-06} +{"train_loss": 0.013577474281191826, "grad_norm": 0.06699894368648529, "global_step": 48293, "epoch": 455, "lr": 1.93274201547633e-06} +{"train_loss": 0.012354893609881401, "grad_norm": 0.099025659263134, "global_step": 48294, "epoch": 455, "lr": 1.9319260382816786e-06} +{"train_loss": 0.014236639253795147, "grad_norm": 0.13194389641284943, "global_step": 48295, "epoch": 455, "lr": 1.93111022997734e-06} +{"train_loss": 0.01265142485499382, "grad_norm": 0.10154358297586441, "global_step": 48296, "epoch": 455, "lr": 1.930294590566173e-06} +{"train_loss": 0.017177008092403412, "grad_norm": 0.09080827981233597, "global_step": 48297, "epoch": 455, "lr": 1.929479120051042e-06} +{"train_loss": 0.0156874880194664, "grad_norm": 0.07303082942962646, "global_step": 48298, "epoch": 455, "lr": 1.9286638184348228e-06} +{"train_loss": 0.01424410566687584, "grad_norm": 0.08169399946928024, "global_step": 48299, "epoch": 455, "lr": 1.927848685720368e-06} +{"train_loss": 0.015636034309864044, "grad_norm": 0.09363701194524765, "global_step": 48300, "epoch": 455, "lr": 1.927033721910543e-06} +{"train_loss": 0.019109424203634262, "grad_norm": 0.06893719732761383, "global_step": 48301, "epoch": 455, "lr": 1.926218927008222e-06} +{"train_loss": 0.016166549175977707, "grad_norm": 0.08671759068965912, "global_step": 48302, "epoch": 455, "lr": 1.9254043010162648e-06} +{"train_loss": 0.01599165052175522, "grad_norm": 0.10284692049026489, "global_step": 48303, "epoch": 455, "lr": 1.924589843937519e-06} +{"train_loss": 0.013481494039297104, "grad_norm": 0.09597717225551605, "global_step": 48304, "epoch": 455, "lr": 1.9237755557748594e-06} +{"train_loss": 0.01571669802069664, "grad_norm": 0.07373148202896118, "global_step": 48305, "epoch": 455, "lr": 1.922961436531151e-06} +{"train_loss": 0.012804936617612839, "grad_norm": 0.08646946400403976, "global_step": 48306, "epoch": 455, "lr": 1.9221474862092416e-06} +{"train_loss": 0.013052355498075485, "grad_norm": 0.07951625436544418, "global_step": 48307, "epoch": 455, "lr": 1.921333704812006e-06} +{"train_loss": 0.016853850334882736, "grad_norm": 0.07961919903755188, "global_step": 48308, "epoch": 455, "lr": 1.920520092342282e-06} +{"train_loss": 0.01245185174047947, "grad_norm": 0.08182583004236221, "global_step": 48309, "epoch": 455, "lr": 1.9197066488029546e-06} +{"train_loss": 0.012281226925551891, "grad_norm": 0.06697419285774231, "global_step": 48310, "epoch": 455, "lr": 1.9188933741968562e-06} +{"train_loss": 0.015302728861570358, "grad_norm": 0.055867891758680344, "global_step": 48311, "epoch": 455, "lr": 1.9180802685268616e-06} +{"train_loss": 0.012790853157639503, "grad_norm": 0.09778451919555664, "global_step": 48312, "epoch": 455, "lr": 1.9172673317958243e-06} +{"train_loss": 0.015035519376397133, "grad_norm": 0.08099857717752457, "global_step": 48313, "epoch": 455, "lr": 1.9164545640066036e-06} +{"train_loss": 0.015922002494335175, "grad_norm": 0.11660496890544891, "global_step": 48314, "epoch": 455, "lr": 1.9156419651620404e-06} +{"train_loss": 0.01350063644349575, "grad_norm": 0.06430427730083466, "global_step": 48315, "epoch": 455, "lr": 1.9148295352650116e-06} +{"train_loss": 0.01980283111333847, "grad_norm": 0.08318371325731277, "global_step": 48316, "epoch": 455, "lr": 1.914017274318347e-06} +{"train_loss": 0.020292453467845917, "grad_norm": 0.08058007806539536, "global_step": 48317, "epoch": 455, "lr": 1.9132051823249177e-06} +{"train_loss": 0.015703709796071053, "grad_norm": 0.15142427384853363, "global_step": 48318, "epoch": 455, "lr": 1.9123932592875816e-06} +{"train_loss": 0.015110294334590435, "grad_norm": 0.0814000740647316, "global_step": 48319, "epoch": 455, "lr": 1.911581505209176e-06} +{"train_loss": 0.01578560099005699, "grad_norm": 0.06951715052127838, "global_step": 48320, "epoch": 455, "lr": 1.910769920092559e-06} +{"train_loss": 0.015169372782111168, "grad_norm": 0.07188062369823456, "global_step": 48321, "epoch": 455, "lr": 1.9099585039405787e-06} +{"train_loss": 0.016168711706995964, "grad_norm": 0.08459387719631195, "global_step": 48322, "epoch": 455, "lr": 1.9091472567560997e-06} +{"train_loss": 0.014321734197437763, "grad_norm": 0.08799056708812714, "global_step": 48323, "epoch": 455, "lr": 1.9083361785419585e-06} +{"train_loss": 0.013387689366936684, "grad_norm": 0.07353444397449493, "global_step": 48324, "epoch": 455, "lr": 1.907525269301008e-06} +{"train_loss": 0.016853347420692444, "grad_norm": 0.09996042400598526, "global_step": 48325, "epoch": 455, "lr": 1.9067145290361022e-06} +{"train_loss": 0.012615926563739777, "grad_norm": 0.0503121092915535, "global_step": 48326, "epoch": 455, "lr": 1.9059039577500937e-06} +{"train_loss": 0.015023057349026203, "grad_norm": 0.0742555782198906, "global_step": 48327, "epoch": 455, "lr": 1.9050935554458083e-06} +{"train_loss": 0.014244625344872475, "grad_norm": 0.10889635980129242, "global_step": 48328, "epoch": 455, "lr": 1.9042833221261158e-06} +{"train_loss": 0.014036276377737522, "grad_norm": 0.10057560354471207, "global_step": 48329, "epoch": 455, "lr": 1.903473257793853e-06} +{"train_loss": 0.015149290673434734, "grad_norm": 0.07734997570514679, "global_step": 48330, "epoch": 455, "lr": 1.9026633624518674e-06} +{"train_loss": 0.013733136467635632, "grad_norm": 0.06626902520656586, "global_step": 48331, "epoch": 455, "lr": 1.9018536361030127e-06} +{"train_loss": 0.016233358532190323, "grad_norm": 0.06558427214622498, "global_step": 48332, "epoch": 455, "lr": 1.9010440787501194e-06} +{"train_loss": 0.013308257795870304, "grad_norm": 0.0840824693441391, "global_step": 48333, "epoch": 455, "lr": 1.9002346903960466e-06} +{"train_loss": 0.015398657880723476, "grad_norm": 0.08632321655750275, "global_step": 48334, "epoch": 455, "lr": 1.8994254710436255e-06} +{"train_loss": 0.014962195770217563, "grad_norm": 0.11027317494153976, "global_step": 48335, "epoch": 455, "lr": 1.8986164206957035e-06, "val_loss": 0.019588619470596313} +{"train_loss": 0.012565599754452705, "grad_norm": 0.09091388434171677, "global_step": 48336, "epoch": 456, "lr": 1.8978075393551287e-06} +{"train_loss": 0.01548790279775858, "grad_norm": 0.0586216077208519, "global_step": 48337, "epoch": 456, "lr": 1.896998827024743e-06} +{"train_loss": 0.015612877905368805, "grad_norm": 0.0772273987531662, "global_step": 48338, "epoch": 456, "lr": 1.8961902837073775e-06} +{"train_loss": 0.013399314135313034, "grad_norm": 0.1488567292690277, "global_step": 48339, "epoch": 456, "lr": 1.8953819094058855e-06} +{"train_loss": 0.01647568680346012, "grad_norm": 0.10410083830356598, "global_step": 48340, "epoch": 456, "lr": 1.8945737041230927e-06} +{"train_loss": 0.013907495886087418, "grad_norm": 0.08682110905647278, "global_step": 48341, "epoch": 456, "lr": 1.8937656678618465e-06} +{"train_loss": 0.016762498766183853, "grad_norm": 0.06603018194437027, "global_step": 48342, "epoch": 456, "lr": 1.8929578006249948e-06} +{"train_loss": 0.014528820291161537, "grad_norm": 0.08501239866018295, "global_step": 48343, "epoch": 456, "lr": 1.8921501024153688e-06} +{"train_loss": 0.014626745134592056, "grad_norm": 0.06611338257789612, "global_step": 48344, "epoch": 456, "lr": 1.8913425732357993e-06} +{"train_loss": 0.011886201798915863, "grad_norm": 0.10178133100271225, "global_step": 48345, "epoch": 456, "lr": 1.890535213089134e-06} +{"train_loss": 0.01647772639989853, "grad_norm": 0.06447423994541168, "global_step": 48346, "epoch": 456, "lr": 1.88972802197821e-06} +{"train_loss": 0.014317735098302364, "grad_norm": 0.07861451059579849, "global_step": 48347, "epoch": 456, "lr": 1.8889209999058465e-06} +{"train_loss": 0.014926486648619175, "grad_norm": 0.07739398628473282, "global_step": 48348, "epoch": 456, "lr": 1.8881141468749031e-06} +{"train_loss": 0.014065800234675407, "grad_norm": 0.058204907923936844, "global_step": 48349, "epoch": 456, "lr": 1.8873074628881993e-06} +{"train_loss": 0.01252269372344017, "grad_norm": 0.08078905194997787, "global_step": 48350, "epoch": 456, "lr": 1.8865009479485774e-06} +{"train_loss": 0.01585056446492672, "grad_norm": 0.07045615464448929, "global_step": 48351, "epoch": 456, "lr": 1.8856946020588573e-06} +{"train_loss": 0.01689515821635723, "grad_norm": 0.11732277274131775, "global_step": 48352, "epoch": 456, "lr": 1.8848884252218868e-06} +{"train_loss": 0.016734572127461433, "grad_norm": 0.11292383819818497, "global_step": 48353, "epoch": 456, "lr": 1.8840824174404969e-06} +{"train_loss": 0.015409104526042938, "grad_norm": 0.10475891083478928, "global_step": 48354, "epoch": 456, "lr": 1.8832765787175188e-06} +{"train_loss": 0.012329721823334694, "grad_norm": 0.08747260272502899, "global_step": 48355, "epoch": 456, "lr": 1.8824709090557725e-06} +{"train_loss": 0.0153506426140666, "grad_norm": 0.10474507510662079, "global_step": 48356, "epoch": 456, "lr": 1.8816654084581053e-06} +{"train_loss": 0.014100784435868263, "grad_norm": 0.08790972083806992, "global_step": 48357, "epoch": 456, "lr": 1.8808600769273377e-06} +{"train_loss": 0.014568804763257504, "grad_norm": 0.06165453791618347, "global_step": 48358, "epoch": 456, "lr": 1.880054914466295e-06} +{"train_loss": 0.015629541128873825, "grad_norm": 0.06575444340705872, "global_step": 48359, "epoch": 456, "lr": 1.8792499210778191e-06} +{"train_loss": 0.013686318881809711, "grad_norm": 0.09100186079740524, "global_step": 48360, "epoch": 456, "lr": 1.878445096764736e-06} +{"train_loss": 0.016992580145597458, "grad_norm": 0.07459001243114471, "global_step": 48361, "epoch": 456, "lr": 1.8776404415298599e-06} +{"train_loss": 0.018076706677675247, "grad_norm": 0.09349390864372253, "global_step": 48362, "epoch": 456, "lr": 1.8768359553760272e-06} +{"train_loss": 0.01495453529059887, "grad_norm": 0.08430952578783035, "global_step": 48363, "epoch": 456, "lr": 1.876031638306075e-06} +{"train_loss": 0.014503675512969494, "grad_norm": 0.06567781418561935, "global_step": 48364, "epoch": 456, "lr": 1.8752274903228062e-06} +{"train_loss": 0.014362114481627941, "grad_norm": 0.08327138423919678, "global_step": 48365, "epoch": 456, "lr": 1.8744235114290631e-06} +{"train_loss": 0.015053920447826385, "grad_norm": 0.08714709430932999, "global_step": 48366, "epoch": 456, "lr": 1.8736197016276768e-06} +{"train_loss": 0.013903077691793442, "grad_norm": 0.1419169306755066, "global_step": 48367, "epoch": 456, "lr": 1.8728160609214561e-06} +{"train_loss": 0.016607176512479782, "grad_norm": 0.09831563383340836, "global_step": 48368, "epoch": 456, "lr": 1.8720125893132212e-06} +{"train_loss": 0.012617658823728561, "grad_norm": 0.06714829057455063, "global_step": 48369, "epoch": 456, "lr": 1.8712092868058085e-06} +{"train_loss": 0.014389409683644772, "grad_norm": 0.05421825125813484, "global_step": 48370, "epoch": 456, "lr": 1.8704061534020434e-06} +{"train_loss": 0.015112392604351044, "grad_norm": 0.14183349907398224, "global_step": 48371, "epoch": 456, "lr": 1.8696031891047294e-06} +{"train_loss": 0.01576545648276806, "grad_norm": 0.10943859815597534, "global_step": 48372, "epoch": 456, "lr": 1.8688003939167032e-06} +{"train_loss": 0.015915654599666595, "grad_norm": 0.06363038718700409, "global_step": 48373, "epoch": 456, "lr": 1.8679977678407846e-06} +{"train_loss": 0.013030605390667915, "grad_norm": 0.07974181324243546, "global_step": 48374, "epoch": 456, "lr": 1.8671953108797824e-06} +{"train_loss": 0.015377910807728767, "grad_norm": 0.08679421246051788, "global_step": 48375, "epoch": 456, "lr": 1.8663930230365224e-06} +{"train_loss": 0.016167905181646347, "grad_norm": 0.06937000155448914, "global_step": 48376, "epoch": 456, "lr": 1.8655909043138353e-06} +{"train_loss": 0.013687685132026672, "grad_norm": 0.06332196295261383, "global_step": 48377, "epoch": 456, "lr": 1.8647889547145192e-06} +{"train_loss": 0.013674614951014519, "grad_norm": 0.09665046632289886, "global_step": 48378, "epoch": 456, "lr": 1.863987174241405e-06} +{"train_loss": 0.016099838539958, "grad_norm": 0.06519667059183121, "global_step": 48379, "epoch": 456, "lr": 1.8631855628973015e-06} +{"train_loss": 0.012559682130813599, "grad_norm": 0.05290830507874489, "global_step": 48380, "epoch": 456, "lr": 1.8623841206850402e-06} +{"train_loss": 0.01442679949104786, "grad_norm": 0.08798635005950928, "global_step": 48381, "epoch": 456, "lr": 1.8615828476074182e-06} +{"train_loss": 0.015045640990138054, "grad_norm": 0.06234583258628845, "global_step": 48382, "epoch": 456, "lr": 1.860781743667256e-06} +{"train_loss": 0.016868956387043, "grad_norm": 0.08439677208662033, "global_step": 48383, "epoch": 456, "lr": 1.8599808088673787e-06} +{"train_loss": 0.017167173326015472, "grad_norm": 0.10220038890838623, "global_step": 48384, "epoch": 456, "lr": 1.8591800432105955e-06} +{"train_loss": 0.014065037481486797, "grad_norm": 0.0759398490190506, "global_step": 48385, "epoch": 456, "lr": 1.8583794466997096e-06} +{"train_loss": 0.016513949260115623, "grad_norm": 0.08538179844617844, "global_step": 48386, "epoch": 456, "lr": 1.8575790193375465e-06} +{"train_loss": 0.017051756381988525, "grad_norm": 0.0646372139453888, "global_step": 48387, "epoch": 456, "lr": 1.856778761126915e-06} +{"train_loss": 0.015487279742956161, "grad_norm": 0.07207120954990387, "global_step": 48388, "epoch": 456, "lr": 1.8559786720706184e-06} +{"train_loss": 0.015591809526085854, "grad_norm": 0.06392895430326462, "global_step": 48389, "epoch": 456, "lr": 1.8551787521714825e-06} +{"train_loss": 0.01786571741104126, "grad_norm": 0.1494891345500946, "global_step": 48390, "epoch": 456, "lr": 1.8543790014323103e-06} +{"train_loss": 0.016408098861575127, "grad_norm": 0.09507083147764206, "global_step": 48391, "epoch": 456, "lr": 1.8535794198559165e-06} +{"train_loss": 0.01191116776317358, "grad_norm": 0.05406048521399498, "global_step": 48392, "epoch": 456, "lr": 1.8527800074450985e-06} +{"train_loss": 0.014163202606141567, "grad_norm": 0.08693816512823105, "global_step": 48393, "epoch": 456, "lr": 1.851980764202671e-06} +{"train_loss": 0.013761060312390327, "grad_norm": 0.0802258849143982, "global_step": 48394, "epoch": 456, "lr": 1.8511816901314538e-06} +{"train_loss": 0.014576973393559456, "grad_norm": 0.08290696889162064, "global_step": 48395, "epoch": 456, "lr": 1.8503827852342391e-06} +{"train_loss": 0.013851193711161613, "grad_norm": 0.07974668592214584, "global_step": 48396, "epoch": 456, "lr": 1.8495840495138416e-06} +{"train_loss": 0.012979501858353615, "grad_norm": 0.04809381440281868, "global_step": 48397, "epoch": 456, "lr": 1.84878548297307e-06} +{"train_loss": 0.016520041972398758, "grad_norm": 0.07253462076187134, "global_step": 48398, "epoch": 456, "lr": 1.8479870856147164e-06} +{"train_loss": 0.016560034826397896, "grad_norm": 0.10518480092287064, "global_step": 48399, "epoch": 456, "lr": 1.8471888574415951e-06} +{"train_loss": 0.017182055860757828, "grad_norm": 0.08302700519561768, "global_step": 48400, "epoch": 456, "lr": 1.846390798456521e-06} +{"train_loss": 0.014612764120101929, "grad_norm": 0.09236764162778854, "global_step": 48401, "epoch": 456, "lr": 1.8455929086622858e-06} +{"train_loss": 0.013514847494661808, "grad_norm": 0.0627516433596611, "global_step": 48402, "epoch": 456, "lr": 1.8447951880616875e-06} +{"train_loss": 0.01504259742796421, "grad_norm": 0.06391122937202454, "global_step": 48403, "epoch": 456, "lr": 1.8439976366575406e-06} +{"train_loss": 0.013772486709058285, "grad_norm": 0.08862572908401489, "global_step": 48404, "epoch": 456, "lr": 1.8432002544526483e-06} +{"train_loss": 0.01350429654121399, "grad_norm": 0.05434846878051758, "global_step": 48405, "epoch": 456, "lr": 1.8424030414497972e-06} +{"train_loss": 0.014821581542491913, "grad_norm": 0.09452859312295914, "global_step": 48406, "epoch": 456, "lr": 1.841605997651802e-06} +{"train_loss": 0.016029439866542816, "grad_norm": 0.11084166914224625, "global_step": 48407, "epoch": 456, "lr": 1.840809123061471e-06} +{"train_loss": 0.01306096650660038, "grad_norm": 0.05427268519997597, "global_step": 48408, "epoch": 456, "lr": 1.8400124176815858e-06} +{"train_loss": 0.013385937549173832, "grad_norm": 0.09659016132354736, "global_step": 48409, "epoch": 456, "lr": 1.8392158815149497e-06} +{"train_loss": 0.015894455835223198, "grad_norm": 0.05917362496256828, "global_step": 48410, "epoch": 456, "lr": 1.8384195145643602e-06} +{"train_loss": 0.016445869579911232, "grad_norm": 0.08061178028583527, "global_step": 48411, "epoch": 456, "lr": 1.8376233168326318e-06} +{"train_loss": 0.014204784296452999, "grad_norm": 0.05829257890582085, "global_step": 48412, "epoch": 456, "lr": 1.8368272883225402e-06} +{"train_loss": 0.016360819339752197, "grad_norm": 0.06559508293867111, "global_step": 48413, "epoch": 456, "lr": 1.8360314290368996e-06} +{"train_loss": 0.014582485891878605, "grad_norm": 0.12375106662511826, "global_step": 48414, "epoch": 456, "lr": 1.835235738978497e-06} +{"train_loss": 0.01689399965107441, "grad_norm": 0.06779545545578003, "global_step": 48415, "epoch": 456, "lr": 1.834440218150124e-06} +{"train_loss": 0.016461433842778206, "grad_norm": 0.06737881898880005, "global_step": 48416, "epoch": 456, "lr": 1.8336448665545792e-06} +{"train_loss": 0.015475822612643242, "grad_norm": 0.13604995608329773, "global_step": 48417, "epoch": 456, "lr": 1.8328496841946707e-06} +{"train_loss": 0.014348074793815613, "grad_norm": 0.18386486172676086, "global_step": 48418, "epoch": 456, "lr": 1.832054671073169e-06} +{"train_loss": 0.014668505638837814, "grad_norm": 0.06852752715349197, "global_step": 48419, "epoch": 456, "lr": 1.8312598271928883e-06} +{"train_loss": 0.014204224571585655, "grad_norm": 0.06304392218589783, "global_step": 48420, "epoch": 456, "lr": 1.8304651525566096e-06} +{"train_loss": 0.01622442901134491, "grad_norm": 0.10387739539146423, "global_step": 48421, "epoch": 456, "lr": 1.829670647167131e-06} +{"train_loss": 0.013569097965955734, "grad_norm": 0.07980111241340637, "global_step": 48422, "epoch": 456, "lr": 1.8288763110272333e-06} +{"train_loss": 0.016042586416006088, "grad_norm": 0.08555984497070312, "global_step": 48423, "epoch": 456, "lr": 1.82808214413972e-06} +{"train_loss": 0.01651960425078869, "grad_norm": 0.10423042625188828, "global_step": 48424, "epoch": 456, "lr": 1.8272881465073777e-06} +{"train_loss": 0.013758499175310135, "grad_norm": 0.06149277836084366, "global_step": 48425, "epoch": 456, "lr": 1.8264943181329985e-06} +{"train_loss": 0.01575615257024765, "grad_norm": 0.08867708593606949, "global_step": 48426, "epoch": 456, "lr": 1.8257006590193637e-06} +{"train_loss": 0.01246424950659275, "grad_norm": 0.05564088001847267, "global_step": 48427, "epoch": 456, "lr": 1.8249071691692598e-06} +{"train_loss": 0.015381526201963425, "grad_norm": 0.10243955254554749, "global_step": 48428, "epoch": 456, "lr": 1.8241138485854902e-06} +{"train_loss": 0.013441545888781548, "grad_norm": 0.15555451810359955, "global_step": 48429, "epoch": 456, "lr": 1.8233206972708305e-06} +{"train_loss": 0.015193765051662922, "grad_norm": 0.061960622668266296, "global_step": 48430, "epoch": 456, "lr": 1.8225277152280674e-06} +{"train_loss": 0.014882364310324192, "grad_norm": 0.07487097382545471, "global_step": 48431, "epoch": 456, "lr": 1.821734902460004e-06} +{"train_loss": 0.015187960118055344, "grad_norm": 0.09856615960597992, "global_step": 48432, "epoch": 456, "lr": 1.8209422589693991e-06} +{"train_loss": 0.01704438403248787, "grad_norm": 0.0792769119143486, "global_step": 48433, "epoch": 456, "lr": 1.820149784759051e-06} +{"train_loss": 0.014232050627470016, "grad_norm": 0.10174812376499176, "global_step": 48434, "epoch": 456, "lr": 1.8193574798317458e-06} +{"train_loss": 0.012451503425836563, "grad_norm": 0.05971814692020416, "global_step": 48435, "epoch": 456, "lr": 1.818565344190265e-06} +{"train_loss": 0.015665598213672638, "grad_norm": 0.08642860502004623, "global_step": 48436, "epoch": 456, "lr": 1.8177733778373896e-06} +{"train_loss": 0.01592819020152092, "grad_norm": 0.12876136600971222, "global_step": 48437, "epoch": 456, "lr": 1.8169815807759116e-06} +{"train_loss": 0.013644525781273842, "grad_norm": 0.0568721704185009, "global_step": 48438, "epoch": 456, "lr": 1.8161899530086014e-06} +{"train_loss": 0.01568189263343811, "grad_norm": 0.15989743173122406, "global_step": 48439, "epoch": 456, "lr": 1.8153984945382452e-06} +{"train_loss": 0.015230165794491768, "grad_norm": 0.0865241065621376, "global_step": 48440, "epoch": 456, "lr": 1.8146072053676188e-06} +{"train_loss": 0.014950163381279639, "grad_norm": 0.08414579927921295, "global_step": 48441, "epoch": 456, "lr": 1.8138160854995145e-06, "val_loss": 0.019047100096940994} +{"train_loss": 0.014787295833230019, "grad_norm": 0.15292102098464966, "global_step": 48442, "epoch": 457, "lr": 1.8130251349367021e-06} +{"train_loss": 0.014198314398527145, "grad_norm": 0.1355181634426117, "global_step": 48443, "epoch": 457, "lr": 1.8122343536819687e-06} +{"train_loss": 0.018399059772491455, "grad_norm": 0.10673140734434128, "global_step": 48444, "epoch": 457, "lr": 1.8114437417380781e-06} +{"train_loss": 0.016144296154379845, "grad_norm": 0.06567765027284622, "global_step": 48445, "epoch": 457, "lr": 1.8106532991078283e-06} +{"train_loss": 0.01621415838599205, "grad_norm": 0.07340735197067261, "global_step": 48446, "epoch": 457, "lr": 1.8098630257939787e-06} +{"train_loss": 0.01699993386864662, "grad_norm": 0.06776634603738785, "global_step": 48447, "epoch": 457, "lr": 1.8090729217993097e-06} +{"train_loss": 0.015430906787514687, "grad_norm": 0.11243011057376862, "global_step": 48448, "epoch": 457, "lr": 1.8082829871266083e-06} +{"train_loss": 0.013494924642145634, "grad_norm": 0.0600556805729866, "global_step": 48449, "epoch": 457, "lr": 1.8074932217786445e-06} +{"train_loss": 0.016203371807932854, "grad_norm": 0.11179786920547485, "global_step": 48450, "epoch": 457, "lr": 1.8067036257581827e-06} +{"train_loss": 0.013500673696398735, "grad_norm": 0.06479257345199585, "global_step": 48451, "epoch": 457, "lr": 1.805914199068004e-06} +{"train_loss": 0.013097681105136871, "grad_norm": 0.05362397059798241, "global_step": 48452, "epoch": 457, "lr": 1.805124941710895e-06} +{"train_loss": 0.015005145221948624, "grad_norm": 0.14490368962287903, "global_step": 48453, "epoch": 457, "lr": 1.8043358536896093e-06} +{"train_loss": 0.01551646739244461, "grad_norm": 0.08827365934848785, "global_step": 48454, "epoch": 457, "lr": 1.8035469350069335e-06} +{"train_loss": 0.015219330787658691, "grad_norm": 0.06581219285726547, "global_step": 48455, "epoch": 457, "lr": 1.8027581856656317e-06} +{"train_loss": 0.01251387782394886, "grad_norm": 0.11705466359853745, "global_step": 48456, "epoch": 457, "lr": 1.8019696056684742e-06} +{"train_loss": 0.015356279909610748, "grad_norm": 0.13886240124702454, "global_step": 48457, "epoch": 457, "lr": 1.8011811950182312e-06} +{"train_loss": 0.012980607338249683, "grad_norm": 0.06428417563438416, "global_step": 48458, "epoch": 457, "lr": 1.8003929537176833e-06} +{"train_loss": 0.017601870000362396, "grad_norm": 0.09077514708042145, "global_step": 48459, "epoch": 457, "lr": 1.79960488176959e-06} +{"train_loss": 0.01400783285498619, "grad_norm": 0.05846240371465683, "global_step": 48460, "epoch": 457, "lr": 1.7988169791767261e-06} +{"train_loss": 0.015409901738166809, "grad_norm": 0.07207553833723068, "global_step": 48461, "epoch": 457, "lr": 1.7980292459418569e-06} +{"train_loss": 0.015235411003232002, "grad_norm": 0.09115903824567795, "global_step": 48462, "epoch": 457, "lr": 1.7972416820677518e-06} +{"train_loss": 0.013429112732410431, "grad_norm": 0.06400115042924881, "global_step": 48463, "epoch": 457, "lr": 1.79645428755717e-06} +{"train_loss": 0.013391165062785149, "grad_norm": 0.05824071168899536, "global_step": 48464, "epoch": 457, "lr": 1.7956670624128868e-06} +{"train_loss": 0.014520054683089256, "grad_norm": 0.067799873650074, "global_step": 48465, "epoch": 457, "lr": 1.794880006637667e-06} +{"train_loss": 0.013167185708880424, "grad_norm": 0.08018535375595093, "global_step": 48466, "epoch": 457, "lr": 1.7940931202342803e-06} +{"train_loss": 0.01537411194294691, "grad_norm": 0.18091364204883575, "global_step": 48467, "epoch": 457, "lr": 1.7933064032054803e-06} +{"train_loss": 0.014830327592790127, "grad_norm": 0.09123893082141876, "global_step": 48468, "epoch": 457, "lr": 1.7925198555540367e-06} +{"train_loss": 0.01535789854824543, "grad_norm": 0.06408122926950455, "global_step": 48469, "epoch": 457, "lr": 1.7917334772827198e-06} +{"train_loss": 0.017076600342988968, "grad_norm": 0.08691468089818954, "global_step": 48470, "epoch": 457, "lr": 1.7909472683942773e-06} +{"train_loss": 0.014629899524152279, "grad_norm": 0.07109387964010239, "global_step": 48471, "epoch": 457, "lr": 1.7901612288914793e-06} +{"train_loss": 0.01523754931986332, "grad_norm": 0.09863435477018356, "global_step": 48472, "epoch": 457, "lr": 1.7893753587771067e-06} +{"train_loss": 0.01512715220451355, "grad_norm": 0.061225924640893936, "global_step": 48473, "epoch": 457, "lr": 1.7885896580538851e-06} +{"train_loss": 0.013651534914970398, "grad_norm": 0.07257804274559021, "global_step": 48474, "epoch": 457, "lr": 1.787804126724596e-06} +{"train_loss": 0.01403544656932354, "grad_norm": 0.08458588272333145, "global_step": 48475, "epoch": 457, "lr": 1.7870187647920034e-06} +{"train_loss": 0.014947681687772274, "grad_norm": 0.0928320437669754, "global_step": 48476, "epoch": 457, "lr": 1.7862335722588496e-06} +{"train_loss": 0.012680823914706707, "grad_norm": 0.06147218123078346, "global_step": 48477, "epoch": 457, "lr": 1.7854485491279106e-06} +{"train_loss": 0.013870013877749443, "grad_norm": 0.07657742500305176, "global_step": 48478, "epoch": 457, "lr": 1.7846636954019336e-06} +{"train_loss": 0.015785960480570793, "grad_norm": 0.05522777885198593, "global_step": 48479, "epoch": 457, "lr": 1.783879011083689e-06} +{"train_loss": 0.017610859125852585, "grad_norm": 0.09544128179550171, "global_step": 48480, "epoch": 457, "lr": 1.7830944961759134e-06} +{"train_loss": 0.01671263389289379, "grad_norm": 0.07491185516119003, "global_step": 48481, "epoch": 457, "lr": 1.7823101506813766e-06} +{"train_loss": 0.016177572309970856, "grad_norm": 0.10078750550746918, "global_step": 48482, "epoch": 457, "lr": 1.7815259746028378e-06} +{"train_loss": 0.016279879957437515, "grad_norm": 0.09376507997512817, "global_step": 48483, "epoch": 457, "lr": 1.7807419679430448e-06} +{"train_loss": 0.014844788238406181, "grad_norm": 0.11201772093772888, "global_step": 48484, "epoch": 457, "lr": 1.7799581307047565e-06} +{"train_loss": 0.015274984762072563, "grad_norm": 0.10762163251638412, "global_step": 48485, "epoch": 457, "lr": 1.7791744628907203e-06} +{"train_loss": 0.013891628012061119, "grad_norm": 0.10461100935935974, "global_step": 48486, "epoch": 457, "lr": 1.7783909645036956e-06} +{"train_loss": 0.015991276130080223, "grad_norm": 0.059636108577251434, "global_step": 48487, "epoch": 457, "lr": 1.7776076355464356e-06} +{"train_loss": 0.01483165193349123, "grad_norm": 0.07890662550926208, "global_step": 48488, "epoch": 457, "lr": 1.7768244760216878e-06} +{"train_loss": 0.015204735100269318, "grad_norm": 0.08760125190019608, "global_step": 48489, "epoch": 457, "lr": 1.7760414859322117e-06} +{"train_loss": 0.012340942397713661, "grad_norm": 0.07520457357168198, "global_step": 48490, "epoch": 457, "lr": 1.7752586652807546e-06} +{"train_loss": 0.015162723138928413, "grad_norm": 0.05775050073862076, "global_step": 48491, "epoch": 457, "lr": 1.7744760140700644e-06} +{"train_loss": 0.013839002698659897, "grad_norm": 0.10086655616760254, "global_step": 48492, "epoch": 457, "lr": 1.7736935323028891e-06} +{"train_loss": 0.016354801133275032, "grad_norm": 0.0852641835808754, "global_step": 48493, "epoch": 457, "lr": 1.7729112199819875e-06} +{"train_loss": 0.016876626759767532, "grad_norm": 0.12080743163824081, "global_step": 48494, "epoch": 457, "lr": 1.7721290771100961e-06} +{"train_loss": 0.015432192012667656, "grad_norm": 0.0706266239285469, "global_step": 48495, "epoch": 457, "lr": 1.7713471036899686e-06} +{"train_loss": 0.01704804226756096, "grad_norm": 0.1525975912809372, "global_step": 48496, "epoch": 457, "lr": 1.7705652997243638e-06} +{"train_loss": 0.014684278517961502, "grad_norm": 0.11714692413806915, "global_step": 48497, "epoch": 457, "lr": 1.7697836652160126e-06} +{"train_loss": 0.017653189599514008, "grad_norm": 0.08274916559457779, "global_step": 48498, "epoch": 457, "lr": 1.7690022001676632e-06} +{"train_loss": 0.014468494802713394, "grad_norm": 0.06944318860769272, "global_step": 48499, "epoch": 457, "lr": 1.7682209045820686e-06} +{"train_loss": 0.015146366320550442, "grad_norm": 0.09275982528924942, "global_step": 48500, "epoch": 457, "lr": 1.7674397784619657e-06} +{"train_loss": 0.01541370153427124, "grad_norm": 0.06610938161611557, "global_step": 48501, "epoch": 457, "lr": 1.7666588218101133e-06} +{"train_loss": 0.01595137268304825, "grad_norm": 0.07680963724851608, "global_step": 48502, "epoch": 457, "lr": 1.765878034629237e-06} +{"train_loss": 0.012480312958359718, "grad_norm": 0.06756973266601562, "global_step": 48503, "epoch": 457, "lr": 1.7650974169220957e-06} +{"train_loss": 0.013578648678958416, "grad_norm": 0.11952265352010727, "global_step": 48504, "epoch": 457, "lr": 1.7643169686914151e-06} +{"train_loss": 0.01329470332711935, "grad_norm": 0.06227593496441841, "global_step": 48505, "epoch": 457, "lr": 1.763536689939954e-06} +{"train_loss": 0.01561286486685276, "grad_norm": 0.06690248847007751, "global_step": 48506, "epoch": 457, "lr": 1.7627565806704494e-06} +{"train_loss": 0.011563354171812534, "grad_norm": 0.05433277040719986, "global_step": 48507, "epoch": 457, "lr": 1.7619766408856375e-06} +{"train_loss": 0.01351243257522583, "grad_norm": 0.08070230484008789, "global_step": 48508, "epoch": 457, "lr": 1.7611968705882608e-06} +{"train_loss": 0.014966616407036781, "grad_norm": 0.06795915216207504, "global_step": 48509, "epoch": 457, "lr": 1.7604172697810618e-06} +{"train_loss": 0.016452731564641, "grad_norm": 0.1640886664390564, "global_step": 48510, "epoch": 457, "lr": 1.7596378384667767e-06} +{"train_loss": 0.015420881099998951, "grad_norm": 0.0802396833896637, "global_step": 48511, "epoch": 457, "lr": 1.7588585766481425e-06} +{"train_loss": 0.014721579849720001, "grad_norm": 0.060899991542100906, "global_step": 48512, "epoch": 457, "lr": 1.7580794843279013e-06} +{"train_loss": 0.014433414675295353, "grad_norm": 0.1088341623544693, "global_step": 48513, "epoch": 457, "lr": 1.7573005615087957e-06} +{"train_loss": 0.016953516751527786, "grad_norm": 0.09437422454357147, "global_step": 48514, "epoch": 457, "lr": 1.7565218081935508e-06} +{"train_loss": 0.014665510505437851, "grad_norm": 0.06848784536123276, "global_step": 48515, "epoch": 457, "lr": 1.7557432243849037e-06} +{"train_loss": 0.013198677450418472, "grad_norm": 0.047723934054374695, "global_step": 48516, "epoch": 457, "lr": 1.754964810085602e-06} +{"train_loss": 0.017199596390128136, "grad_norm": 0.14054307341575623, "global_step": 48517, "epoch": 457, "lr": 1.7541865652983658e-06} +{"train_loss": 0.017699524760246277, "grad_norm": 0.07516582310199738, "global_step": 48518, "epoch": 457, "lr": 1.7534084900259318e-06} +{"train_loss": 0.013504566624760628, "grad_norm": 0.044374506920576096, "global_step": 48519, "epoch": 457, "lr": 1.7526305842710532e-06} +{"train_loss": 0.01570175029337406, "grad_norm": 0.0694272369146347, "global_step": 48520, "epoch": 457, "lr": 1.7518528480364393e-06} +{"train_loss": 0.01413559541106224, "grad_norm": 0.057508524507284164, "global_step": 48521, "epoch": 457, "lr": 1.7510752813248321e-06} +{"train_loss": 0.01366655807942152, "grad_norm": 0.07050422579050064, "global_step": 48522, "epoch": 457, "lr": 1.7502978841389627e-06} +{"train_loss": 0.014884617179632187, "grad_norm": 0.07068667560815811, "global_step": 48523, "epoch": 457, "lr": 1.749520656481568e-06} +{"train_loss": 0.013419633731245995, "grad_norm": 0.10429634153842926, "global_step": 48524, "epoch": 457, "lr": 1.7487435983553679e-06} +{"train_loss": 0.019566714763641357, "grad_norm": 0.07464415580034256, "global_step": 48525, "epoch": 457, "lr": 1.7479667097631048e-06} +{"train_loss": 0.015370750799775124, "grad_norm": 0.0721232146024704, "global_step": 48526, "epoch": 457, "lr": 1.7471899907074985e-06} +{"train_loss": 0.015023228712379932, "grad_norm": 0.0722615197300911, "global_step": 48527, "epoch": 457, "lr": 1.7464134411912857e-06} +{"train_loss": 0.015981808304786682, "grad_norm": 0.1300545632839203, "global_step": 48528, "epoch": 457, "lr": 1.7456370612171868e-06} +{"train_loss": 0.013366675935685635, "grad_norm": 0.0737517923116684, "global_step": 48529, "epoch": 457, "lr": 1.7448608507879328e-06} +{"train_loss": 0.014878037385642529, "grad_norm": 0.07345573604106903, "global_step": 48530, "epoch": 457, "lr": 1.74408480990626e-06} +{"train_loss": 0.014858745969831944, "grad_norm": 0.07717811316251755, "global_step": 48531, "epoch": 457, "lr": 1.743308938574889e-06} +{"train_loss": 0.01445537805557251, "grad_norm": 0.0674833357334137, "global_step": 48532, "epoch": 457, "lr": 1.7425332367965342e-06} +{"train_loss": 0.01434516254812479, "grad_norm": 0.11210745573043823, "global_step": 48533, "epoch": 457, "lr": 1.7417577045739375e-06} +{"train_loss": 0.013718146830797195, "grad_norm": 0.07245256751775742, "global_step": 48534, "epoch": 457, "lr": 1.7409823419098082e-06} +{"train_loss": 0.01354683842509985, "grad_norm": 0.054479241371154785, "global_step": 48535, "epoch": 457, "lr": 1.7402071488068882e-06} +{"train_loss": 0.01316386740654707, "grad_norm": 0.05145907402038574, "global_step": 48536, "epoch": 457, "lr": 1.7394321252678924e-06} +{"train_loss": 0.014428401365876198, "grad_norm": 0.07586526870727539, "global_step": 48537, "epoch": 457, "lr": 1.7386572712955462e-06} +{"train_loss": 0.016712497919797897, "grad_norm": 0.10375223308801651, "global_step": 48538, "epoch": 457, "lr": 1.7378825868925585e-06} +{"train_loss": 0.015190386213362217, "grad_norm": 0.08413102477788925, "global_step": 48539, "epoch": 457, "lr": 1.7371080720616717e-06} +{"train_loss": 0.01276189461350441, "grad_norm": 0.04645383730530739, "global_step": 48540, "epoch": 457, "lr": 1.7363337268056e-06} +{"train_loss": 0.013856337405741215, "grad_norm": 0.09715554863214493, "global_step": 48541, "epoch": 457, "lr": 1.735559551127053e-06} +{"train_loss": 0.015465686097741127, "grad_norm": 0.08397112041711807, "global_step": 48542, "epoch": 457, "lr": 1.7347855450287664e-06} +{"train_loss": 0.014045885764062405, "grad_norm": 0.12021110951900482, "global_step": 48543, "epoch": 457, "lr": 1.7340117085134554e-06} +{"train_loss": 0.015642082318663597, "grad_norm": 0.11985574662685394, "global_step": 48544, "epoch": 457, "lr": 1.7332380415838344e-06} +{"train_loss": 0.014085624366998672, "grad_norm": 0.12064137309789658, "global_step": 48545, "epoch": 457, "lr": 1.7324645442426235e-06} +{"train_loss": 0.014629470184445381, "grad_norm": 0.08863403648138046, "global_step": 48546, "epoch": 457, "lr": 1.7316912164925424e-06} +{"train_loss": 0.01496865031889306, "grad_norm": 0.14227674901485443, "global_step": 48547, "epoch": 457, "lr": 1.730918058336306e-06, "val_loss": 0.019023939967155457} +{"train_loss": 0.0138460174202919, "grad_norm": 0.08359099179506302, "global_step": 48548, "epoch": 458, "lr": 1.7301450697766342e-06} +{"train_loss": 0.01405065506696701, "grad_norm": 0.10344872623682022, "global_step": 48549, "epoch": 458, "lr": 1.7293722508162412e-06} +{"train_loss": 0.01409042626619339, "grad_norm": 0.08569695800542831, "global_step": 48550, "epoch": 458, "lr": 1.7285996014578365e-06} +{"train_loss": 0.014075923711061478, "grad_norm": 0.09314493089914322, "global_step": 48551, "epoch": 458, "lr": 1.7278271217041453e-06} +{"train_loss": 0.014611219987273216, "grad_norm": 0.17335407435894012, "global_step": 48552, "epoch": 458, "lr": 1.7270548115578712e-06} +{"train_loss": 0.01588742434978485, "grad_norm": 0.14404885470867157, "global_step": 48553, "epoch": 458, "lr": 1.7262826710217338e-06} +{"train_loss": 0.01411338523030281, "grad_norm": 0.06019477918744087, "global_step": 48554, "epoch": 458, "lr": 1.7255107000984594e-06} +{"train_loss": 0.014420855790376663, "grad_norm": 0.08428676426410675, "global_step": 48555, "epoch": 458, "lr": 1.7247388987907286e-06} +{"train_loss": 0.013374879024922848, "grad_norm": 0.10862943530082703, "global_step": 48556, "epoch": 458, "lr": 1.7239672671012785e-06} +{"train_loss": 0.014854668639600277, "grad_norm": 0.10370713472366333, "global_step": 48557, "epoch": 458, "lr": 1.7231958050328124e-06} +{"train_loss": 0.01450935285538435, "grad_norm": 0.05558932572603226, "global_step": 48558, "epoch": 458, "lr": 1.7224245125880389e-06} +{"train_loss": 0.01434002909809351, "grad_norm": 0.06540942937135696, "global_step": 48559, "epoch": 458, "lr": 1.7216533897696675e-06} +{"train_loss": 0.014149169437587261, "grad_norm": 0.06843899190425873, "global_step": 48560, "epoch": 458, "lr": 1.7208824365804176e-06} +{"train_loss": 0.016439229249954224, "grad_norm": 0.1457136571407318, "global_step": 48561, "epoch": 458, "lr": 1.7201116530229932e-06} +{"train_loss": 0.016128621995449066, "grad_norm": 0.07217424362897873, "global_step": 48562, "epoch": 458, "lr": 1.7193410391000919e-06} +{"train_loss": 0.013529709540307522, "grad_norm": 0.150498628616333, "global_step": 48563, "epoch": 458, "lr": 1.7185705948144281e-06} +{"train_loss": 0.013478429988026619, "grad_norm": 0.08194079250097275, "global_step": 48564, "epoch": 458, "lr": 1.7178003201687165e-06} +{"train_loss": 0.015240022912621498, "grad_norm": 0.08063487708568573, "global_step": 48565, "epoch": 458, "lr": 1.7170302151656547e-06} +{"train_loss": 0.013057292439043522, "grad_norm": 0.07781216502189636, "global_step": 48566, "epoch": 458, "lr": 1.716260279807952e-06} +{"train_loss": 0.013888070359826088, "grad_norm": 0.06503502279520035, "global_step": 48567, "epoch": 458, "lr": 1.7154905140983113e-06} +{"train_loss": 0.016139531508088112, "grad_norm": 0.0846782997250557, "global_step": 48568, "epoch": 458, "lr": 1.7147209180394418e-06} +{"train_loss": 0.013572605326771736, "grad_norm": 0.08270546793937683, "global_step": 48569, "epoch": 458, "lr": 1.7139514916340416e-06} +{"train_loss": 0.016112716868519783, "grad_norm": 0.09882985055446625, "global_step": 48570, "epoch": 458, "lr": 1.7131822348848137e-06} +{"train_loss": 0.015371005982160568, "grad_norm": 0.06499351561069489, "global_step": 48571, "epoch": 458, "lr": 1.7124131477944727e-06} +{"train_loss": 0.014695198275148869, "grad_norm": 0.11449787765741348, "global_step": 48572, "epoch": 458, "lr": 1.7116442303657054e-06} +{"train_loss": 0.015397725626826286, "grad_norm": 0.06560591608285904, "global_step": 48573, "epoch": 458, "lr": 1.7108754826012208e-06} +{"train_loss": 0.016744496300816536, "grad_norm": 0.10429073125123978, "global_step": 48574, "epoch": 458, "lr": 1.7101069045037222e-06} +{"train_loss": 0.016113735735416412, "grad_norm": 0.16150926053524017, "global_step": 48575, "epoch": 458, "lr": 1.7093384960759018e-06} +{"train_loss": 0.015016298741102219, "grad_norm": 0.08242081105709076, "global_step": 48576, "epoch": 458, "lr": 1.7085702573204688e-06} +{"train_loss": 0.017216075211763382, "grad_norm": 0.09481269866228104, "global_step": 48577, "epoch": 458, "lr": 1.7078021882401208e-06} +{"train_loss": 0.015078607946634293, "grad_norm": 0.06504274904727936, "global_step": 48578, "epoch": 458, "lr": 1.7070342888375557e-06} +{"train_loss": 0.012278164736926556, "grad_norm": 0.09689989686012268, "global_step": 48579, "epoch": 458, "lr": 1.7062665591154603e-06} +{"train_loss": 0.012927808798849583, "grad_norm": 0.10931891947984695, "global_step": 48580, "epoch": 458, "lr": 1.705498999076549e-06} +{"train_loss": 0.01423281617462635, "grad_norm": 0.07607104629278183, "global_step": 48581, "epoch": 458, "lr": 1.7047316087235087e-06} +{"train_loss": 0.015928035601973534, "grad_norm": 0.10820987820625305, "global_step": 48582, "epoch": 458, "lr": 1.7039643880590372e-06} +{"train_loss": 0.014515629038214684, "grad_norm": 0.09742744266986847, "global_step": 48583, "epoch": 458, "lr": 1.7031973370858322e-06} +{"train_loss": 0.01314723864197731, "grad_norm": 0.06079007685184479, "global_step": 48584, "epoch": 458, "lr": 1.7024304558065918e-06} +{"train_loss": 0.015791352838277817, "grad_norm": 0.19598573446273804, "global_step": 48585, "epoch": 458, "lr": 1.701663744224008e-06} +{"train_loss": 0.016429107636213303, "grad_norm": 0.09108773618936539, "global_step": 48586, "epoch": 458, "lr": 1.7008972023407676e-06} +{"train_loss": 0.016005098819732666, "grad_norm": 0.12251336127519608, "global_step": 48587, "epoch": 458, "lr": 1.7001308301595686e-06} +{"train_loss": 0.01581677421927452, "grad_norm": 0.06745102256536484, "global_step": 48588, "epoch": 458, "lr": 1.6993646276831144e-06} +{"train_loss": 0.01548517681658268, "grad_norm": 0.11837338656187057, "global_step": 48589, "epoch": 458, "lr": 1.6985985949140804e-06} +{"train_loss": 0.015887044370174408, "grad_norm": 0.08045697212219238, "global_step": 48590, "epoch": 458, "lr": 1.6978327318551646e-06} +{"train_loss": 0.018144797533750534, "grad_norm": 0.058724723756313324, "global_step": 48591, "epoch": 458, "lr": 1.6970670385090648e-06} +{"train_loss": 0.017146533355116844, "grad_norm": 0.12114723771810532, "global_step": 48592, "epoch": 458, "lr": 1.6963015148784568e-06} +{"train_loss": 0.015228243544697762, "grad_norm": 0.081192746758461, "global_step": 48593, "epoch": 458, "lr": 1.6955361609660436e-06} +{"train_loss": 0.015665564686059952, "grad_norm": 0.06527592241764069, "global_step": 48594, "epoch": 458, "lr": 1.6947709767745013e-06} +{"train_loss": 0.0138894347473979, "grad_norm": 0.11578552424907684, "global_step": 48595, "epoch": 458, "lr": 1.6940059623065385e-06} +{"train_loss": 0.015059787780046463, "grad_norm": 0.12070073932409286, "global_step": 48596, "epoch": 458, "lr": 1.6932411175648254e-06} +{"train_loss": 0.014559971168637276, "grad_norm": 0.07455786317586899, "global_step": 48597, "epoch": 458, "lr": 1.6924764425520546e-06} +{"train_loss": 0.015978727489709854, "grad_norm": 0.0888756737112999, "global_step": 48598, "epoch": 458, "lr": 1.6917119372709178e-06} +{"train_loss": 0.01670650579035282, "grad_norm": 0.10235070437192917, "global_step": 48599, "epoch": 458, "lr": 1.6909476017240912e-06} +{"train_loss": 0.015135718509554863, "grad_norm": 0.06350137293338776, "global_step": 48600, "epoch": 458, "lr": 1.690183435914261e-06} +{"train_loss": 0.013573408126831055, "grad_norm": 0.08846614509820938, "global_step": 48601, "epoch": 458, "lr": 1.6894194398441255e-06} +{"train_loss": 0.013914174400269985, "grad_norm": 0.0597456619143486, "global_step": 48602, "epoch": 458, "lr": 1.68865561351636e-06} +{"train_loss": 0.016430266201496124, "grad_norm": 0.08594071865081787, "global_step": 48603, "epoch": 458, "lr": 1.6878919569336459e-06} +{"train_loss": 0.01638958975672722, "grad_norm": 0.09196364879608154, "global_step": 48604, "epoch": 458, "lr": 1.6871284700986646e-06} +{"train_loss": 0.015210428275167942, "grad_norm": 0.05806087329983711, "global_step": 48605, "epoch": 458, "lr": 1.6863651530141134e-06} +{"train_loss": 0.014065666124224663, "grad_norm": 0.11755682528018951, "global_step": 48606, "epoch": 458, "lr": 1.685602005682657e-06} +{"train_loss": 0.015048839151859283, "grad_norm": 0.14870795607566833, "global_step": 48607, "epoch": 458, "lr": 1.6848390281069882e-06} +{"train_loss": 0.016080476343631744, "grad_norm": 0.1349211186170578, "global_step": 48608, "epoch": 458, "lr": 1.6840762202897764e-06} +{"train_loss": 0.014507802203297615, "grad_norm": 0.07608505338430405, "global_step": 48609, "epoch": 458, "lr": 1.6833135822337144e-06} +{"train_loss": 0.013097532093524933, "grad_norm": 0.0529959611594677, "global_step": 48610, "epoch": 458, "lr": 1.6825511139414718e-06} +{"train_loss": 0.015536071732640266, "grad_norm": 0.0767320916056633, "global_step": 48611, "epoch": 458, "lr": 1.6817888154157358e-06} +{"train_loss": 0.01321463007479906, "grad_norm": 0.07436343282461166, "global_step": 48612, "epoch": 458, "lr": 1.6810266866591817e-06} +{"train_loss": 0.014420317485928535, "grad_norm": 0.06883490085601807, "global_step": 48613, "epoch": 458, "lr": 1.6802647276744854e-06} +{"train_loss": 0.015155012719333172, "grad_norm": 0.09619449824094772, "global_step": 48614, "epoch": 458, "lr": 1.6795029384643223e-06} +{"train_loss": 0.017138060182332993, "grad_norm": 0.08737000823020935, "global_step": 48615, "epoch": 458, "lr": 1.6787413190313794e-06} +{"train_loss": 0.014440692961215973, "grad_norm": 0.06588798016309738, "global_step": 48616, "epoch": 458, "lr": 1.6779798693783156e-06} +{"train_loss": 0.012682223692536354, "grad_norm": 0.05846376717090607, "global_step": 48617, "epoch": 458, "lr": 1.6772185895078174e-06} +{"train_loss": 0.013902084901928902, "grad_norm": 0.12267810851335526, "global_step": 48618, "epoch": 458, "lr": 1.6764574794225608e-06} +{"train_loss": 0.014942559413611889, "grad_norm": 0.08856258541345596, "global_step": 48619, "epoch": 458, "lr": 1.675696539125221e-06} +{"train_loss": 0.01618899032473564, "grad_norm": 0.12092995643615723, "global_step": 48620, "epoch": 458, "lr": 1.674935768618463e-06} +{"train_loss": 0.013656048104166985, "grad_norm": 0.0614137202501297, "global_step": 48621, "epoch": 458, "lr": 1.6741751679049622e-06} +{"train_loss": 0.015529107302427292, "grad_norm": 0.17143943905830383, "global_step": 48622, "epoch": 458, "lr": 1.6734147369873999e-06} +{"train_loss": 0.017329249531030655, "grad_norm": 0.13907204568386078, "global_step": 48623, "epoch": 458, "lr": 1.6726544758684347e-06} +{"train_loss": 0.014626352116465569, "grad_norm": 0.07582222670316696, "global_step": 48624, "epoch": 458, "lr": 1.671894384550743e-06} +{"train_loss": 0.015980437397956848, "grad_norm": 0.076331727206707, "global_step": 48625, "epoch": 458, "lr": 1.6711344630370051e-06} +{"train_loss": 0.015547375194728374, "grad_norm": 0.0631227120757103, "global_step": 48626, "epoch": 458, "lr": 1.6703747113298863e-06} +{"train_loss": 0.015152506530284882, "grad_norm": 0.0970534011721611, "global_step": 48627, "epoch": 458, "lr": 1.669615129432045e-06} +{"train_loss": 0.015754815191030502, "grad_norm": 0.09481687098741531, "global_step": 48628, "epoch": 458, "lr": 1.6688557173461572e-06} +{"train_loss": 0.014594132080674171, "grad_norm": 0.07386278361082077, "global_step": 48629, "epoch": 458, "lr": 1.668096475074893e-06} +{"train_loss": 0.014102187007665634, "grad_norm": 0.0561295710504055, "global_step": 48630, "epoch": 458, "lr": 1.667337402620922e-06} +{"train_loss": 0.014736766926944256, "grad_norm": 0.08418469876050949, "global_step": 48631, "epoch": 458, "lr": 1.6665784999869093e-06} +{"train_loss": 0.014376959763467312, "grad_norm": 0.0701989084482193, "global_step": 48632, "epoch": 458, "lr": 1.6658197671755194e-06} +{"train_loss": 0.014893211424350739, "grad_norm": 0.0778612419962883, "global_step": 48633, "epoch": 458, "lr": 1.665061204189411e-06} +{"train_loss": 0.015847966074943542, "grad_norm": 0.08374354243278503, "global_step": 48634, "epoch": 458, "lr": 1.6643028110312598e-06} +{"train_loss": 0.01365666277706623, "grad_norm": 0.0520688034594059, "global_step": 48635, "epoch": 458, "lr": 1.6635445877037358e-06} +{"train_loss": 0.015151049941778183, "grad_norm": 0.06722597032785416, "global_step": 48636, "epoch": 458, "lr": 1.6627865342094872e-06} +{"train_loss": 0.01671891286969185, "grad_norm": 0.13047635555267334, "global_step": 48637, "epoch": 458, "lr": 1.6620286505511894e-06} +{"train_loss": 0.01707901619374752, "grad_norm": 0.07609453797340393, "global_step": 48638, "epoch": 458, "lr": 1.661270936731496e-06} +{"train_loss": 0.015333489514887333, "grad_norm": 0.1288694590330124, "global_step": 48639, "epoch": 458, "lr": 1.6605133927530825e-06} +{"train_loss": 0.015013480558991432, "grad_norm": 0.06224043294787407, "global_step": 48640, "epoch": 458, "lr": 1.6597560186185967e-06} +{"train_loss": 0.013553054071962833, "grad_norm": 0.07363340258598328, "global_step": 48641, "epoch": 458, "lr": 1.6589988143307033e-06} +{"train_loss": 0.014414309523999691, "grad_norm": 0.11161277443170547, "global_step": 48642, "epoch": 458, "lr": 1.6582417798920723e-06} +{"train_loss": 0.014605021104216576, "grad_norm": 0.09833776205778122, "global_step": 48643, "epoch": 458, "lr": 1.6574849153053573e-06} +{"train_loss": 0.01307755894958973, "grad_norm": 0.060099873691797256, "global_step": 48644, "epoch": 458, "lr": 1.6567282205732116e-06} +{"train_loss": 0.01532149501144886, "grad_norm": 0.08258048444986343, "global_step": 48645, "epoch": 458, "lr": 1.6559716956982995e-06} +{"train_loss": 0.013564263470470905, "grad_norm": 0.09339528530836105, "global_step": 48646, "epoch": 458, "lr": 1.6552153406832805e-06} +{"train_loss": 0.015217145904898643, "grad_norm": 0.08984829485416412, "global_step": 48647, "epoch": 458, "lr": 1.6544591555308076e-06} +{"train_loss": 0.017232758924365044, "grad_norm": 0.10368365049362183, "global_step": 48648, "epoch": 458, "lr": 1.6537031402435399e-06} +{"train_loss": 0.013104084879159927, "grad_norm": 0.06997331976890564, "global_step": 48649, "epoch": 458, "lr": 1.6529472948241365e-06} +{"train_loss": 0.01679571345448494, "grad_norm": 0.08416300266981125, "global_step": 48650, "epoch": 458, "lr": 1.6521916192752563e-06} +{"train_loss": 0.014043405652046204, "grad_norm": 0.07588440179824829, "global_step": 48651, "epoch": 458, "lr": 1.6514361135995416e-06} +{"train_loss": 0.012797573581337929, "grad_norm": 0.07133177667856216, "global_step": 48652, "epoch": 458, "lr": 1.6506807777996514e-06} +{"train_loss": 0.014950991739993388, "grad_norm": 0.08747407793998718, "global_step": 48653, "epoch": 458, "lr": 1.6499256118782503e-06, "val_loss": 0.019148068502545357} +{"train_loss": 0.0145538579672575, "grad_norm": 0.0606834776699543, "global_step": 48654, "epoch": 459, "lr": 1.6491706158379805e-06} +{"train_loss": 0.01385258324444294, "grad_norm": 0.0785968154668808, "global_step": 48655, "epoch": 459, "lr": 1.6484157896814956e-06} +{"train_loss": 0.01385621726512909, "grad_norm": 0.105898417532444, "global_step": 48656, "epoch": 459, "lr": 1.6476611334114544e-06} +{"train_loss": 0.015113066881895065, "grad_norm": 0.07624717801809311, "global_step": 48657, "epoch": 459, "lr": 1.6469066470304995e-06} +{"train_loss": 0.01670476421713829, "grad_norm": 0.08230384439229965, "global_step": 48658, "epoch": 459, "lr": 1.646152330541284e-06} +{"train_loss": 0.014925984665751457, "grad_norm": 0.05524410307407379, "global_step": 48659, "epoch": 459, "lr": 1.6453981839464672e-06} +{"train_loss": 0.01643826626241207, "grad_norm": 0.11871126294136047, "global_step": 48660, "epoch": 459, "lr": 1.6446442072486968e-06} +{"train_loss": 0.01662716083228588, "grad_norm": 0.06234940141439438, "global_step": 48661, "epoch": 459, "lr": 1.643890400450604e-06} +{"train_loss": 0.015589114278554916, "grad_norm": 0.06385143101215363, "global_step": 48662, "epoch": 459, "lr": 1.643136763554859e-06} +{"train_loss": 0.01601700484752655, "grad_norm": 0.07520755380392075, "global_step": 48663, "epoch": 459, "lr": 1.642383296564104e-06} +{"train_loss": 0.01581939123570919, "grad_norm": 0.0909094288945198, "global_step": 48664, "epoch": 459, "lr": 1.6416299994809758e-06} +{"train_loss": 0.015243971720337868, "grad_norm": 0.05852377414703369, "global_step": 48665, "epoch": 459, "lr": 1.6408768723081336e-06} +{"train_loss": 0.015769148245453835, "grad_norm": 0.10457523912191391, "global_step": 48666, "epoch": 459, "lr": 1.6401239150482195e-06} +{"train_loss": 0.01426590047776699, "grad_norm": 0.127757266163826, "global_step": 48667, "epoch": 459, "lr": 1.6393711277038815e-06} +{"train_loss": 0.013698844239115715, "grad_norm": 0.07599950581789017, "global_step": 48668, "epoch": 459, "lr": 1.6386185102777562e-06} +{"train_loss": 0.013399338349699974, "grad_norm": 0.05238223820924759, "global_step": 48669, "epoch": 459, "lr": 1.6378660627724919e-06} +{"train_loss": 0.0136333629488945, "grad_norm": 0.07608883827924728, "global_step": 48670, "epoch": 459, "lr": 1.6371137851907413e-06} +{"train_loss": 0.014511508867144585, "grad_norm": 0.07555586099624634, "global_step": 48671, "epoch": 459, "lr": 1.6363616775351365e-06} +{"train_loss": 0.013093785382807255, "grad_norm": 0.09297602623701096, "global_step": 48672, "epoch": 459, "lr": 1.6356097398083302e-06} +{"train_loss": 0.012848146259784698, "grad_norm": 0.08320663124322891, "global_step": 48673, "epoch": 459, "lr": 1.6348579720129541e-06} +{"train_loss": 0.01582401990890503, "grad_norm": 0.10596159100532532, "global_step": 48674, "epoch": 459, "lr": 1.6341063741516504e-06} +{"train_loss": 0.017125796526670456, "grad_norm": 0.06807385385036469, "global_step": 48675, "epoch": 459, "lr": 1.6333549462270615e-06} +{"train_loss": 0.014734144322574139, "grad_norm": 0.13575749099254608, "global_step": 48676, "epoch": 459, "lr": 1.6326036882418405e-06} +{"train_loss": 0.015169277787208557, "grad_norm": 0.07515960931777954, "global_step": 48677, "epoch": 459, "lr": 1.6318526001986023e-06} +{"train_loss": 0.014209364540874958, "grad_norm": 0.06656097620725632, "global_step": 48678, "epoch": 459, "lr": 1.6311016821000114e-06} +{"train_loss": 0.017299890518188477, "grad_norm": 0.12399160116910934, "global_step": 48679, "epoch": 459, "lr": 1.6303509339486823e-06} +{"train_loss": 0.015982620418071747, "grad_norm": 0.06314940750598907, "global_step": 48680, "epoch": 459, "lr": 1.6296003557472738e-06} +{"train_loss": 0.013659290969371796, "grad_norm": 0.0840531438589096, "global_step": 48681, "epoch": 459, "lr": 1.628849947498412e-06} +{"train_loss": 0.014719266444444656, "grad_norm": 0.08796968311071396, "global_step": 48682, "epoch": 459, "lr": 1.6280997092047279e-06} +{"train_loss": 0.014134637080132961, "grad_norm": 0.14962467551231384, "global_step": 48683, "epoch": 459, "lr": 1.627349640868875e-06} +{"train_loss": 0.015122586861252785, "grad_norm": 0.10009945929050446, "global_step": 48684, "epoch": 459, "lr": 1.6265997424934788e-06} +{"train_loss": 0.013854876160621643, "grad_norm": 0.07172498852014542, "global_step": 48685, "epoch": 459, "lr": 1.625850014081165e-06} +{"train_loss": 0.014465417712926865, "grad_norm": 0.13132333755493164, "global_step": 48686, "epoch": 459, "lr": 1.6251004556345816e-06} +{"train_loss": 0.01483527384698391, "grad_norm": 0.07050250470638275, "global_step": 48687, "epoch": 459, "lr": 1.62435106715636e-06} +{"train_loss": 0.012591131962835789, "grad_norm": 0.06756909936666489, "global_step": 48688, "epoch": 459, "lr": 1.6236018486491255e-06} +{"train_loss": 0.01311008632183075, "grad_norm": 0.13621923327445984, "global_step": 48689, "epoch": 459, "lr": 1.6228528001155152e-06} +{"train_loss": 0.01311450731009245, "grad_norm": 0.05851425230503082, "global_step": 48690, "epoch": 459, "lr": 1.6221039215581767e-06} +{"train_loss": 0.017387041822075844, "grad_norm": 0.11143019050359726, "global_step": 48691, "epoch": 459, "lr": 1.6213552129797139e-06} +{"train_loss": 0.017000790685415268, "grad_norm": 0.06167377158999443, "global_step": 48692, "epoch": 459, "lr": 1.620606674382763e-06} +{"train_loss": 0.016719482839107513, "grad_norm": 0.13023057579994202, "global_step": 48693, "epoch": 459, "lr": 1.6198583057699724e-06} +{"train_loss": 0.012770840898156166, "grad_norm": 0.059349317103624344, "global_step": 48694, "epoch": 459, "lr": 1.6191101071439507e-06} +{"train_loss": 0.016591809689998627, "grad_norm": 0.09191827476024628, "global_step": 48695, "epoch": 459, "lr": 1.6183620785073405e-06} +{"train_loss": 0.016733184456825256, "grad_norm": 0.07755767554044724, "global_step": 48696, "epoch": 459, "lr": 1.6176142198627674e-06} +{"train_loss": 0.01262705959379673, "grad_norm": 0.06528379023075104, "global_step": 48697, "epoch": 459, "lr": 1.6168665312128572e-06} +{"train_loss": 0.012597855180501938, "grad_norm": 0.11216752976179123, "global_step": 48698, "epoch": 459, "lr": 1.61611901256023e-06} +{"train_loss": 0.0153343565762043, "grad_norm": 0.06240692362189293, "global_step": 48699, "epoch": 459, "lr": 1.6153716639075222e-06} +{"train_loss": 0.015445709228515625, "grad_norm": 0.07594525814056396, "global_step": 48700, "epoch": 459, "lr": 1.61462448525736e-06} +{"train_loss": 0.01557723805308342, "grad_norm": 0.07473886013031006, "global_step": 48701, "epoch": 459, "lr": 1.6138774766123688e-06} +{"train_loss": 0.016376955434679985, "grad_norm": 0.1310839056968689, "global_step": 48702, "epoch": 459, "lr": 1.6131306379751576e-06} +{"train_loss": 0.01603802852332592, "grad_norm": 0.07060869038105011, "global_step": 48703, "epoch": 459, "lr": 1.612383969348369e-06} +{"train_loss": 0.014542141929268837, "grad_norm": 0.05553359165787697, "global_step": 48704, "epoch": 459, "lr": 1.6116374707346227e-06} +{"train_loss": 0.0155791025608778, "grad_norm": 0.06764601171016693, "global_step": 48705, "epoch": 459, "lr": 1.6108911421365281e-06} +{"train_loss": 0.015519486740231514, "grad_norm": 0.08766575157642365, "global_step": 48706, "epoch": 459, "lr": 1.6101449835567273e-06} +{"train_loss": 0.014779306016862392, "grad_norm": 0.1352001428604126, "global_step": 48707, "epoch": 459, "lr": 1.6093989949978293e-06} +{"train_loss": 0.013409590348601341, "grad_norm": 0.11515261977910995, "global_step": 48708, "epoch": 459, "lr": 1.6086531764624657e-06} +{"train_loss": 0.012853385880589485, "grad_norm": 0.0737563744187355, "global_step": 48709, "epoch": 459, "lr": 1.6079075279532396e-06} +{"train_loss": 0.012424296699464321, "grad_norm": 0.11924493312835693, "global_step": 48710, "epoch": 459, "lr": 1.6071620494727879e-06} +{"train_loss": 0.016249509528279305, "grad_norm": 0.12192926555871964, "global_step": 48711, "epoch": 459, "lr": 1.6064167410237197e-06} +{"train_loss": 0.015465790405869484, "grad_norm": 0.09422880411148071, "global_step": 48712, "epoch": 459, "lr": 1.6056716026086606e-06} +{"train_loss": 0.013793138787150383, "grad_norm": 0.06853887438774109, "global_step": 48713, "epoch": 459, "lr": 1.604926634230225e-06} +{"train_loss": 0.012658541090786457, "grad_norm": 0.05492318049073219, "global_step": 48714, "epoch": 459, "lr": 1.6041818358910332e-06} +{"train_loss": 0.011635292321443558, "grad_norm": 0.07519835233688354, "global_step": 48715, "epoch": 459, "lr": 1.6034372075936942e-06} +{"train_loss": 0.01586415059864521, "grad_norm": 0.14175541698932648, "global_step": 48716, "epoch": 459, "lr": 1.6026927493408285e-06} +{"train_loss": 0.01564401388168335, "grad_norm": 0.06656288355588913, "global_step": 48717, "epoch": 459, "lr": 1.601948461135061e-06} +{"train_loss": 0.015029627829790115, "grad_norm": 0.09203408658504486, "global_step": 48718, "epoch": 459, "lr": 1.6012043429789902e-06} +{"train_loss": 0.014092894271016121, "grad_norm": 0.12234519422054291, "global_step": 48719, "epoch": 459, "lr": 1.6004603948752473e-06} +{"train_loss": 0.015464426949620247, "grad_norm": 0.0587114617228508, "global_step": 48720, "epoch": 459, "lr": 1.59971661682643e-06} +{"train_loss": 0.01719995029270649, "grad_norm": 0.10054612904787064, "global_step": 48721, "epoch": 459, "lr": 1.598973008835164e-06} +{"train_loss": 0.014470834285020828, "grad_norm": 0.11945797502994537, "global_step": 48722, "epoch": 459, "lr": 1.5982295709040585e-06} +{"train_loss": 0.01590988039970398, "grad_norm": 0.08876930177211761, "global_step": 48723, "epoch": 459, "lr": 1.5974863030357223e-06} +{"train_loss": 0.019442308694124222, "grad_norm": 0.06532815843820572, "global_step": 48724, "epoch": 459, "lr": 1.5967432052327702e-06} +{"train_loss": 0.016816578805446625, "grad_norm": 0.07934839278459549, "global_step": 48725, "epoch": 459, "lr": 1.5960002774978166e-06} +{"train_loss": 0.013769794255495071, "grad_norm": 0.12236370891332626, "global_step": 48726, "epoch": 459, "lr": 1.5952575198334651e-06} +{"train_loss": 0.014012910425662994, "grad_norm": 0.08944972604513168, "global_step": 48727, "epoch": 459, "lr": 1.5945149322423248e-06} +{"train_loss": 0.01525852270424366, "grad_norm": 0.13391676545143127, "global_step": 48728, "epoch": 459, "lr": 1.59377251472701e-06} +{"train_loss": 0.01755429059267044, "grad_norm": 0.12272686511278152, "global_step": 48729, "epoch": 459, "lr": 1.5930302672901298e-06} +{"train_loss": 0.013401476666331291, "grad_norm": 0.052204061299562454, "global_step": 48730, "epoch": 459, "lr": 1.5922881899342822e-06} +{"train_loss": 0.013049141503870487, "grad_norm": 0.0733649879693985, "global_step": 48731, "epoch": 459, "lr": 1.5915462826620986e-06} +{"train_loss": 0.01613786444067955, "grad_norm": 0.07151982188224792, "global_step": 48732, "epoch": 459, "lr": 1.5908045454761544e-06} +{"train_loss": 0.01456555724143982, "grad_norm": 0.08893997967243195, "global_step": 48733, "epoch": 459, "lr": 1.5900629783790699e-06} +{"train_loss": 0.016203517094254494, "grad_norm": 0.07304096221923828, "global_step": 48734, "epoch": 459, "lr": 1.5893215813734596e-06} +{"train_loss": 0.01620730571448803, "grad_norm": 0.11314930766820908, "global_step": 48735, "epoch": 459, "lr": 1.5885803544619104e-06} +{"train_loss": 0.013752768747508526, "grad_norm": 0.07439856976270676, "global_step": 48736, "epoch": 459, "lr": 1.5878392976470423e-06} +{"train_loss": 0.016602274030447006, "grad_norm": 0.06815403699874878, "global_step": 48737, "epoch": 459, "lr": 1.5870984109314535e-06} +{"train_loss": 0.015302359126508236, "grad_norm": 0.07284096628427505, "global_step": 48738, "epoch": 459, "lr": 1.586357694317747e-06} +{"train_loss": 0.013788377866148949, "grad_norm": 0.11555487662553787, "global_step": 48739, "epoch": 459, "lr": 1.5856171478085213e-06} +{"train_loss": 0.013810206204652786, "grad_norm": 0.10627557337284088, "global_step": 48740, "epoch": 459, "lr": 1.584876771406385e-06} +{"train_loss": 0.015311266295611858, "grad_norm": 0.1002335324883461, "global_step": 48741, "epoch": 459, "lr": 1.584136565113936e-06} +{"train_loss": 0.015211129561066628, "grad_norm": 0.05578000098466873, "global_step": 48742, "epoch": 459, "lr": 1.5833965289337726e-06} +{"train_loss": 0.01727861352264881, "grad_norm": 0.08008487522602081, "global_step": 48743, "epoch": 459, "lr": 1.5826566628685036e-06} +{"train_loss": 0.016316179186105728, "grad_norm": 0.06831101328134537, "global_step": 48744, "epoch": 459, "lr": 1.5819169669207212e-06} +{"train_loss": 0.01296231709420681, "grad_norm": 0.06260067969560623, "global_step": 48745, "epoch": 459, "lr": 1.5811774410930235e-06} +{"train_loss": 0.012188397347927094, "grad_norm": 0.06315966695547104, "global_step": 48746, "epoch": 459, "lr": 1.580438085388014e-06} +{"train_loss": 0.015055282041430473, "grad_norm": 0.06158352270722389, "global_step": 48747, "epoch": 459, "lr": 1.579698899808285e-06} +{"train_loss": 0.01536178681999445, "grad_norm": 0.07039345800876617, "global_step": 48748, "epoch": 459, "lr": 1.5789598843564401e-06} +{"train_loss": 0.016479618847370148, "grad_norm": 0.08536065369844437, "global_step": 48749, "epoch": 459, "lr": 1.5782210390350716e-06} +{"train_loss": 0.01577473059296608, "grad_norm": 0.07180442661046982, "global_step": 48750, "epoch": 459, "lr": 1.5774823638467774e-06} +{"train_loss": 0.013247133232653141, "grad_norm": 0.052789878100156784, "global_step": 48751, "epoch": 459, "lr": 1.5767438587941441e-06} +{"train_loss": 0.01472560502588749, "grad_norm": 0.07777166366577148, "global_step": 48752, "epoch": 459, "lr": 1.576005523879781e-06} +{"train_loss": 0.0139992143958807, "grad_norm": 0.09180319309234619, "global_step": 48753, "epoch": 459, "lr": 1.575267359106275e-06} +{"train_loss": 0.014426423236727715, "grad_norm": 0.12872757017612457, "global_step": 48754, "epoch": 459, "lr": 1.5745293644762239e-06} +{"train_loss": 0.016836602240800858, "grad_norm": 0.07432544231414795, "global_step": 48755, "epoch": 459, "lr": 1.5737915399922144e-06} +{"train_loss": 0.01497576106339693, "grad_norm": 0.11723773926496506, "global_step": 48756, "epoch": 459, "lr": 1.5730538856568388e-06} +{"train_loss": 0.016318585723638535, "grad_norm": 0.10504575073719025, "global_step": 48757, "epoch": 459, "lr": 1.57231640147269e-06} +{"train_loss": 0.014000125229358673, "grad_norm": 0.06364976614713669, "global_step": 48758, "epoch": 459, "lr": 1.5715790874423653e-06} +{"train_loss": 0.01495481527603741, "grad_norm": 0.13320544362068176, "global_step": 48759, "epoch": 459, "lr": 1.5708419435684462e-06, "val_loss": 0.01921785995364189} +{"train_loss": 0.014532234519720078, "grad_norm": 0.08178050816059113, "global_step": 48760, "epoch": 460, "lr": 1.5701049698535309e-06} +{"train_loss": 0.015051262453198433, "grad_norm": 0.06600721180438995, "global_step": 48761, "epoch": 460, "lr": 1.5693681663002002e-06} +{"train_loss": 0.015276244841516018, "grad_norm": 0.05529307574033737, "global_step": 48762, "epoch": 460, "lr": 1.568631532911058e-06} +{"train_loss": 0.01319158636033535, "grad_norm": 0.15821512043476105, "global_step": 48763, "epoch": 460, "lr": 1.5678950696886686e-06} +{"train_loss": 0.015815766528248787, "grad_norm": 0.09830841422080994, "global_step": 48764, "epoch": 460, "lr": 1.567158776635641e-06} +{"train_loss": 0.015755873173475266, "grad_norm": 0.07314594835042953, "global_step": 48765, "epoch": 460, "lr": 1.5664226537545567e-06} +{"train_loss": 0.01477758213877678, "grad_norm": 0.10314390808343887, "global_step": 48766, "epoch": 460, "lr": 1.5656867010480025e-06} +{"train_loss": 0.015192557126283646, "grad_norm": 0.10076291859149933, "global_step": 48767, "epoch": 460, "lr": 1.564950918518554e-06} +{"train_loss": 0.012488085776567459, "grad_norm": 0.06740181893110275, "global_step": 48768, "epoch": 460, "lr": 1.5642153061688035e-06} +{"train_loss": 0.01488784234970808, "grad_norm": 0.05426228791475296, "global_step": 48769, "epoch": 460, "lr": 1.5634798640013383e-06} +{"train_loss": 0.015782853588461876, "grad_norm": 0.06854566931724548, "global_step": 48770, "epoch": 460, "lr": 1.562744592018739e-06} +{"train_loss": 0.01492296438664198, "grad_norm": 0.07475630939006805, "global_step": 48771, "epoch": 460, "lr": 1.5620094902235927e-06} +{"train_loss": 0.01422320306301117, "grad_norm": 0.10559281706809998, "global_step": 48772, "epoch": 460, "lr": 1.5612745586184863e-06} +{"train_loss": 0.013703320175409317, "grad_norm": 0.07626617699861526, "global_step": 48773, "epoch": 460, "lr": 1.56053979720599e-06} +{"train_loss": 0.01762194186449051, "grad_norm": 0.059914980083703995, "global_step": 48774, "epoch": 460, "lr": 1.5598052059886903e-06} +{"train_loss": 0.015527890995144844, "grad_norm": 0.05410778149962425, "global_step": 48775, "epoch": 460, "lr": 1.5590707849691689e-06} +{"train_loss": 0.014136994257569313, "grad_norm": 0.12215335667133331, "global_step": 48776, "epoch": 460, "lr": 1.5583365341500067e-06} +{"train_loss": 0.014481445774435997, "grad_norm": 0.0756632536649704, "global_step": 48777, "epoch": 460, "lr": 1.5576024535337797e-06} +{"train_loss": 0.014981187880039215, "grad_norm": 0.08927134424448013, "global_step": 48778, "epoch": 460, "lr": 1.5568685431230745e-06} +{"train_loss": 0.01340264081954956, "grad_norm": 0.11963184922933578, "global_step": 48779, "epoch": 460, "lr": 1.5561348029204725e-06} +{"train_loss": 0.014086290262639523, "grad_norm": 0.059116508811712265, "global_step": 48780, "epoch": 460, "lr": 1.5554012329285328e-06} +{"train_loss": 0.01493713166564703, "grad_norm": 0.0657457485795021, "global_step": 48781, "epoch": 460, "lr": 1.5546678331498476e-06} +{"train_loss": 0.014138982631266117, "grad_norm": 0.10177813470363617, "global_step": 48782, "epoch": 460, "lr": 1.5539346035869984e-06} +{"train_loss": 0.014798802323639393, "grad_norm": 0.09248103946447372, "global_step": 48783, "epoch": 460, "lr": 1.5532015442425497e-06} +{"train_loss": 0.016487136483192444, "grad_norm": 0.08931870758533478, "global_step": 48784, "epoch": 460, "lr": 1.5524686551190825e-06} +{"train_loss": 0.012111845426261425, "grad_norm": 0.06389717757701874, "global_step": 48785, "epoch": 460, "lr": 1.5517359362191675e-06} +{"train_loss": 0.017275992780923843, "grad_norm": 0.0853099524974823, "global_step": 48786, "epoch": 460, "lr": 1.5510033875453911e-06} +{"train_loss": 0.013264736160635948, "grad_norm": 0.13840748369693756, "global_step": 48787, "epoch": 460, "lr": 1.5502710091003126e-06} +{"train_loss": 0.013709094375371933, "grad_norm": 0.05660464987158775, "global_step": 48788, "epoch": 460, "lr": 1.549538800886513e-06} +{"train_loss": 0.016495339572429657, "grad_norm": 0.11570650339126587, "global_step": 48789, "epoch": 460, "lr": 1.5488067629065627e-06} +{"train_loss": 0.016461081802845, "grad_norm": 0.06805098801851273, "global_step": 48790, "epoch": 460, "lr": 1.5480748951630375e-06} +{"train_loss": 0.017602546140551567, "grad_norm": 0.10339250415563583, "global_step": 48791, "epoch": 460, "lr": 1.5473431976585017e-06} +{"train_loss": 0.013722484931349754, "grad_norm": 0.07210814952850342, "global_step": 48792, "epoch": 460, "lr": 1.546611670395537e-06} +{"train_loss": 0.013929142616689205, "grad_norm": 0.06298182904720306, "global_step": 48793, "epoch": 460, "lr": 1.545880313376702e-06} +{"train_loss": 0.017257822677493095, "grad_norm": 0.1539856642484665, "global_step": 48794, "epoch": 460, "lr": 1.545149126604567e-06} +{"train_loss": 0.016113974153995514, "grad_norm": 0.09402014315128326, "global_step": 48795, "epoch": 460, "lr": 1.5444181100817135e-06} +{"train_loss": 0.017442386597394943, "grad_norm": 0.11079677194356918, "global_step": 48796, "epoch": 460, "lr": 1.5436872638107058e-06} +{"train_loss": 0.017714589834213257, "grad_norm": 0.1261916607618332, "global_step": 48797, "epoch": 460, "lr": 1.5429565877940977e-06} +{"train_loss": 0.015422839671373367, "grad_norm": 0.11001887172460556, "global_step": 48798, "epoch": 460, "lr": 1.5422260820344647e-06} +{"train_loss": 0.012875000014901161, "grad_norm": 0.069513238966465, "global_step": 48799, "epoch": 460, "lr": 1.5414957465343882e-06} +{"train_loss": 0.01490507647395134, "grad_norm": 0.05643925070762634, "global_step": 48800, "epoch": 460, "lr": 1.5407655812964106e-06} +{"train_loss": 0.014413334429264069, "grad_norm": 0.0697113573551178, "global_step": 48801, "epoch": 460, "lr": 1.5400355863231075e-06} +{"train_loss": 0.013028180226683617, "grad_norm": 0.05240620672702789, "global_step": 48802, "epoch": 460, "lr": 1.5393057616170547e-06} +{"train_loss": 0.012672808021306992, "grad_norm": 0.0749274343252182, "global_step": 48803, "epoch": 460, "lr": 1.5385761071808058e-06} +{"train_loss": 0.015434646047651768, "grad_norm": 0.11979265511035919, "global_step": 48804, "epoch": 460, "lr": 1.5378466230169141e-06} +{"train_loss": 0.015863997861742973, "grad_norm": 0.07146085798740387, "global_step": 48805, "epoch": 460, "lr": 1.5371173091279611e-06} +{"train_loss": 0.01660989411175251, "grad_norm": 0.08876413851976395, "global_step": 48806, "epoch": 460, "lr": 1.5363881655165058e-06} +{"train_loss": 0.01618645153939724, "grad_norm": 0.08480881154537201, "global_step": 48807, "epoch": 460, "lr": 1.5356591921851016e-06} +{"train_loss": 0.013369932770729065, "grad_norm": 0.07562851160764694, "global_step": 48808, "epoch": 460, "lr": 1.5349303891363131e-06} +{"train_loss": 0.016025152057409286, "grad_norm": 0.06790545582771301, "global_step": 48809, "epoch": 460, "lr": 1.5342017563726997e-06} +{"train_loss": 0.015071584843099117, "grad_norm": 0.1281391829252243, "global_step": 48810, "epoch": 460, "lr": 1.533473293896831e-06} +{"train_loss": 0.013368592597544193, "grad_norm": 0.06895487755537033, "global_step": 48811, "epoch": 460, "lr": 1.5327450017112555e-06} +{"train_loss": 0.014251007698476315, "grad_norm": 0.06524021178483963, "global_step": 48812, "epoch": 460, "lr": 1.532016879818532e-06} +{"train_loss": 0.014856046997010708, "grad_norm": 0.07434181123971939, "global_step": 48813, "epoch": 460, "lr": 1.5312889282212362e-06} +{"train_loss": 0.016614273190498352, "grad_norm": 0.07763495296239853, "global_step": 48814, "epoch": 460, "lr": 1.530561146921905e-06} +{"train_loss": 0.01247390080243349, "grad_norm": 0.0785977840423584, "global_step": 48815, "epoch": 460, "lr": 1.5298335359230975e-06} +{"train_loss": 0.01594376750290394, "grad_norm": 0.05153120681643486, "global_step": 48816, "epoch": 460, "lr": 1.529106095227384e-06} +{"train_loss": 0.013528035953640938, "grad_norm": 0.09103351831436157, "global_step": 48817, "epoch": 460, "lr": 1.5283788248373066e-06} +{"train_loss": 0.015486977994441986, "grad_norm": 0.06269140541553497, "global_step": 48818, "epoch": 460, "lr": 1.5276517247554302e-06} +{"train_loss": 0.014622695744037628, "grad_norm": 0.08384834975004196, "global_step": 48819, "epoch": 460, "lr": 1.5269247949843079e-06} +{"train_loss": 0.015216872096061707, "grad_norm": 0.07945120334625244, "global_step": 48820, "epoch": 460, "lr": 1.5261980355264882e-06} +{"train_loss": 0.015484503470361233, "grad_norm": 0.0677831843495369, "global_step": 48821, "epoch": 460, "lr": 1.5254714463845244e-06} +{"train_loss": 0.014095905236899853, "grad_norm": 0.06848202645778656, "global_step": 48822, "epoch": 460, "lr": 1.5247450275609755e-06} +{"train_loss": 0.014603785239160061, "grad_norm": 0.10850559175014496, "global_step": 48823, "epoch": 460, "lr": 1.5240187790583949e-06} +{"train_loss": 0.014632590115070343, "grad_norm": 0.11663181334733963, "global_step": 48824, "epoch": 460, "lr": 1.523292700879325e-06} +{"train_loss": 0.01315171830356121, "grad_norm": 0.09109245240688324, "global_step": 48825, "epoch": 460, "lr": 1.522566793026331e-06} +{"train_loss": 0.015792451798915863, "grad_norm": 0.06171495467424393, "global_step": 48826, "epoch": 460, "lr": 1.5218410555019436e-06} +{"train_loss": 0.015190010890364647, "grad_norm": 0.09564313292503357, "global_step": 48827, "epoch": 460, "lr": 1.5211154883087331e-06} +{"train_loss": 0.01405823603272438, "grad_norm": 0.0674818605184555, "global_step": 48828, "epoch": 460, "lr": 1.5203900914492365e-06} +{"train_loss": 0.016200751066207886, "grad_norm": 0.0727100595831871, "global_step": 48829, "epoch": 460, "lr": 1.5196648649260015e-06} +{"train_loss": 0.015431199222803116, "grad_norm": 0.0838756412267685, "global_step": 48830, "epoch": 460, "lr": 1.5189398087415875e-06} +{"train_loss": 0.014473577961325645, "grad_norm": 0.0776773989200592, "global_step": 48831, "epoch": 460, "lr": 1.5182149228985364e-06} +{"train_loss": 0.014983112923800945, "grad_norm": 0.1023210659623146, "global_step": 48832, "epoch": 460, "lr": 1.5174902073993858e-06} +{"train_loss": 0.015377809293568134, "grad_norm": 0.07295116037130356, "global_step": 48833, "epoch": 460, "lr": 1.5167656622466996e-06} +{"train_loss": 0.01508711464703083, "grad_norm": 0.08555814623832703, "global_step": 48834, "epoch": 460, "lr": 1.5160412874430096e-06} +{"train_loss": 0.015933619812130928, "grad_norm": 0.08836683630943298, "global_step": 48835, "epoch": 460, "lr": 1.5153170829908581e-06} +{"train_loss": 0.018886059522628784, "grad_norm": 0.08121515810489655, "global_step": 48836, "epoch": 460, "lr": 1.5145930488928094e-06} +{"train_loss": 0.016359394416213036, "grad_norm": 0.09802840650081635, "global_step": 48837, "epoch": 460, "lr": 1.5138691851513897e-06} +{"train_loss": 0.012767644599080086, "grad_norm": 0.10881246626377106, "global_step": 48838, "epoch": 460, "lr": 1.5131454917691467e-06} +{"train_loss": 0.015286488458514214, "grad_norm": 0.077820785343647, "global_step": 48839, "epoch": 460, "lr": 1.512421968748623e-06} +{"train_loss": 0.014429823495447636, "grad_norm": 0.07150297611951828, "global_step": 48840, "epoch": 460, "lr": 1.5116986160923663e-06} +{"train_loss": 0.013456772081553936, "grad_norm": 0.050495296716690063, "global_step": 48841, "epoch": 460, "lr": 1.510975433802908e-06} +{"train_loss": 0.01563044637441635, "grad_norm": 0.08809805661439896, "global_step": 48842, "epoch": 460, "lr": 1.5102524218827962e-06} +{"train_loss": 0.01612718589603901, "grad_norm": 0.10195425897836685, "global_step": 48843, "epoch": 460, "lr": 1.5095295803345732e-06} +{"train_loss": 0.015211870893836021, "grad_norm": 0.08755962550640106, "global_step": 48844, "epoch": 460, "lr": 1.508806909160776e-06} +{"train_loss": 0.014729523099958897, "grad_norm": 0.08681592345237732, "global_step": 48845, "epoch": 460, "lr": 1.5080844083639355e-06} +{"train_loss": 0.013772092759609222, "grad_norm": 0.06995801627635956, "global_step": 48846, "epoch": 460, "lr": 1.5073620779466003e-06} +{"train_loss": 0.014653606340289116, "grad_norm": 0.07903873920440674, "global_step": 48847, "epoch": 460, "lr": 1.5066399179113121e-06} +{"train_loss": 0.013580976985394955, "grad_norm": 0.07539518177509308, "global_step": 48848, "epoch": 460, "lr": 1.5059179282605974e-06} +{"train_loss": 0.013683614321053028, "grad_norm": 0.05808939039707184, "global_step": 48849, "epoch": 460, "lr": 1.5051961089969979e-06} +{"train_loss": 0.015477429144084454, "grad_norm": 0.13007965683937073, "global_step": 48850, "epoch": 460, "lr": 1.5044744601230566e-06} +{"train_loss": 0.016346793621778488, "grad_norm": 0.06610430777072906, "global_step": 48851, "epoch": 460, "lr": 1.5037529816412877e-06} +{"train_loss": 0.016740113496780396, "grad_norm": 0.12206956744194031, "global_step": 48852, "epoch": 460, "lr": 1.5030316735542449e-06} +{"train_loss": 0.013879301026463509, "grad_norm": 0.07804350554943085, "global_step": 48853, "epoch": 460, "lr": 1.5023105358644596e-06} +{"train_loss": 0.01467314176261425, "grad_norm": 0.056535039097070694, "global_step": 48854, "epoch": 460, "lr": 1.501589568574474e-06} +{"train_loss": 0.014983057044446468, "grad_norm": 0.0992746353149414, "global_step": 48855, "epoch": 460, "lr": 1.5008687716868031e-06} +{"train_loss": 0.013758450746536255, "grad_norm": 0.09845753759145737, "global_step": 48856, "epoch": 460, "lr": 1.5001481452039834e-06} +{"train_loss": 0.015387633815407753, "grad_norm": 0.09231502562761307, "global_step": 48857, "epoch": 460, "lr": 1.4994276891285575e-06} +{"train_loss": 0.015685148537158966, "grad_norm": 0.06919613480567932, "global_step": 48858, "epoch": 460, "lr": 1.4987074034630456e-06} +{"train_loss": 0.018620915710926056, "grad_norm": 0.1806928515434265, "global_step": 48859, "epoch": 460, "lr": 1.4979872882099788e-06} +{"train_loss": 0.014003443531692028, "grad_norm": 0.0941973552107811, "global_step": 48860, "epoch": 460, "lr": 1.4972673433719053e-06} +{"train_loss": 0.014718289487063885, "grad_norm": 0.09921236336231232, "global_step": 48861, "epoch": 460, "lr": 1.4965475689513342e-06} +{"train_loss": 0.013601092621684074, "grad_norm": 0.07593022286891937, "global_step": 48862, "epoch": 460, "lr": 1.4958279649507968e-06} +{"train_loss": 0.014680818654596806, "grad_norm": 0.05771986022591591, "global_step": 48863, "epoch": 460, "lr": 1.4951085313728296e-06} +{"train_loss": 0.013269523158669472, "grad_norm": 0.06283499300479889, "global_step": 48864, "epoch": 460, "lr": 1.4943892682199589e-06} +{"train_loss": 0.014949521585806922, "grad_norm": 0.11971548944711685, "global_step": 48865, "epoch": 460, "lr": 1.4936701754947101e-06, "train_batch_mse_error": 0.013370761647820473, "val_loss": 0.019460223615169525} +{"train_loss": 0.016096940264105797, "grad_norm": 0.11462973803281784, "global_step": 48866, "epoch": 461, "lr": 1.4929512531996093e-06} +{"train_loss": 0.014013969339430332, "grad_norm": 0.056078314781188965, "global_step": 48867, "epoch": 461, "lr": 1.4922325013371762e-06} +{"train_loss": 0.016201995313167572, "grad_norm": 0.06941215693950653, "global_step": 48868, "epoch": 461, "lr": 1.4915139199099538e-06} +{"train_loss": 0.0139026353135705, "grad_norm": 0.058047104626894, "global_step": 48869, "epoch": 461, "lr": 1.490795508920445e-06} +{"train_loss": 0.013893496245145798, "grad_norm": 0.07565566152334213, "global_step": 48870, "epoch": 461, "lr": 1.4900772683711873e-06} +{"train_loss": 0.01630149409174919, "grad_norm": 0.09974130243062973, "global_step": 48871, "epoch": 461, "lr": 1.489359198264706e-06} +{"train_loss": 0.013342145830392838, "grad_norm": 0.065998375415802, "global_step": 48872, "epoch": 461, "lr": 1.4886412986035215e-06} +{"train_loss": 0.016065318137407303, "grad_norm": 0.08367141336202621, "global_step": 48873, "epoch": 461, "lr": 1.4879235693901428e-06} +{"train_loss": 0.013176796957850456, "grad_norm": 0.06530781090259552, "global_step": 48874, "epoch": 461, "lr": 1.487206010627118e-06} +{"train_loss": 0.015213927254080772, "grad_norm": 0.060671038925647736, "global_step": 48875, "epoch": 461, "lr": 1.4864886223169395e-06} +{"train_loss": 0.014256203547120094, "grad_norm": 0.10420003533363342, "global_step": 48876, "epoch": 461, "lr": 1.4857714044621495e-06} +{"train_loss": 0.015296360477805138, "grad_norm": 0.10581506788730621, "global_step": 48877, "epoch": 461, "lr": 1.4850543570652632e-06} +{"train_loss": 0.014504693448543549, "grad_norm": 0.11236637085676193, "global_step": 48878, "epoch": 461, "lr": 1.4843374801287947e-06} +{"train_loss": 0.015213060192763805, "grad_norm": 0.06300276517868042, "global_step": 48879, "epoch": 461, "lr": 1.4836207736552642e-06} +{"train_loss": 0.016981229186058044, "grad_norm": 0.08353689312934875, "global_step": 48880, "epoch": 461, "lr": 1.4829042376471868e-06} +{"train_loss": 0.015448296442627907, "grad_norm": 0.09765373170375824, "global_step": 48881, "epoch": 461, "lr": 1.4821878721070936e-06} +{"train_loss": 0.01502283662557602, "grad_norm": 0.07561253011226654, "global_step": 48882, "epoch": 461, "lr": 1.4814716770374827e-06} +{"train_loss": 0.015365404076874256, "grad_norm": 0.1330007165670395, "global_step": 48883, "epoch": 461, "lr": 1.480755652440885e-06} +{"train_loss": 0.01350137498229742, "grad_norm": 0.08064336329698563, "global_step": 48884, "epoch": 461, "lr": 1.4800397983198155e-06} +{"train_loss": 0.013611717149615288, "grad_norm": 0.060396987944841385, "global_step": 48885, "epoch": 461, "lr": 1.4793241146767833e-06} +{"train_loss": 0.014410117641091347, "grad_norm": 0.11630039662122726, "global_step": 48886, "epoch": 461, "lr": 1.478608601514303e-06} +{"train_loss": 0.015640731900930405, "grad_norm": 0.09864486008882523, "global_step": 48887, "epoch": 461, "lr": 1.4778932588348837e-06} +{"train_loss": 0.012956884689629078, "grad_norm": 0.06988417357206345, "global_step": 48888, "epoch": 461, "lr": 1.4771780866410567e-06} +{"train_loss": 0.017977675423026085, "grad_norm": 0.08370955288410187, "global_step": 48889, "epoch": 461, "lr": 1.4764630849353201e-06} +{"train_loss": 0.014886426739394665, "grad_norm": 0.16365396976470947, "global_step": 48890, "epoch": 461, "lr": 1.475748253720194e-06} +{"train_loss": 0.016521785408258438, "grad_norm": 0.06705109775066376, "global_step": 48891, "epoch": 461, "lr": 1.4750335929981817e-06} +{"train_loss": 0.012860190123319626, "grad_norm": 0.08343857526779175, "global_step": 48892, "epoch": 461, "lr": 1.4743191027717985e-06} +{"train_loss": 0.015474654734134674, "grad_norm": 0.15071530640125275, "global_step": 48893, "epoch": 461, "lr": 1.4736047830435528e-06} +{"train_loss": 0.014443302527070045, "grad_norm": 0.0864652693271637, "global_step": 48894, "epoch": 461, "lr": 1.4728906338159599e-06} +{"train_loss": 0.014001259580254555, "grad_norm": 0.09549067169427872, "global_step": 48895, "epoch": 461, "lr": 1.4721766550915227e-06} +{"train_loss": 0.015610482543706894, "grad_norm": 0.0725473016500473, "global_step": 48896, "epoch": 461, "lr": 1.4714628468727564e-06} +{"train_loss": 0.015046592801809311, "grad_norm": 0.06250477582216263, "global_step": 48897, "epoch": 461, "lr": 1.4707492091621588e-06} +{"train_loss": 0.013778679072856903, "grad_norm": 0.11203321814537048, "global_step": 48898, "epoch": 461, "lr": 1.47003574196225e-06} +{"train_loss": 0.013735579326748848, "grad_norm": 0.12753771245479584, "global_step": 48899, "epoch": 461, "lr": 1.4693224452755282e-06} +{"train_loss": 0.01434735581278801, "grad_norm": 0.05671950802206993, "global_step": 48900, "epoch": 461, "lr": 1.468609319104497e-06} +{"train_loss": 0.015016404911875725, "grad_norm": 0.06770137697458267, "global_step": 48901, "epoch": 461, "lr": 1.4678963634516763e-06} +{"train_loss": 0.01274668239057064, "grad_norm": 0.08074693381786346, "global_step": 48902, "epoch": 461, "lr": 1.4671835783195587e-06} +{"train_loss": 0.018701115623116493, "grad_norm": 0.07288974523544312, "global_step": 48903, "epoch": 461, "lr": 1.4664709637106478e-06} +{"train_loss": 0.01634928397834301, "grad_norm": 0.07436778396368027, "global_step": 48904, "epoch": 461, "lr": 1.4657585196274526e-06} +{"train_loss": 0.014971498399972916, "grad_norm": 0.06849180907011032, "global_step": 48905, "epoch": 461, "lr": 1.4650462460724768e-06} +{"train_loss": 0.012874830514192581, "grad_norm": 0.07464710623025894, "global_step": 48906, "epoch": 461, "lr": 1.4643341430482182e-06} +{"train_loss": 0.01652715913951397, "grad_norm": 0.0975099727511406, "global_step": 48907, "epoch": 461, "lr": 1.4636222105571863e-06} +{"train_loss": 0.014329329133033752, "grad_norm": 0.05713557451963425, "global_step": 48908, "epoch": 461, "lr": 1.462910448601873e-06} +{"train_loss": 0.01653926819562912, "grad_norm": 0.0984797403216362, "global_step": 48909, "epoch": 461, "lr": 1.4621988571847878e-06} +{"train_loss": 0.015805348753929138, "grad_norm": 0.0653635561466217, "global_step": 48910, "epoch": 461, "lr": 1.4614874363084285e-06} +{"train_loss": 0.016737811267375946, "grad_norm": 0.08196751028299332, "global_step": 48911, "epoch": 461, "lr": 1.4607761859752877e-06} +{"train_loss": 0.015667390078306198, "grad_norm": 0.12171215564012527, "global_step": 48912, "epoch": 461, "lr": 1.4600651061878745e-06} +{"train_loss": 0.015498239547014236, "grad_norm": 0.09950398653745651, "global_step": 48913, "epoch": 461, "lr": 1.4593541969486869e-06} +{"train_loss": 0.01604403182864189, "grad_norm": 0.0906788557767868, "global_step": 48914, "epoch": 461, "lr": 1.4586434582602115e-06} +{"train_loss": 0.015248590148985386, "grad_norm": 0.07476016134023666, "global_step": 48915, "epoch": 461, "lr": 1.4579328901249578e-06} +{"train_loss": 0.0144033282995224, "grad_norm": 0.06347883492708206, "global_step": 48916, "epoch": 461, "lr": 1.4572224925454126e-06} +{"train_loss": 0.014077393338084221, "grad_norm": 0.08507063239812851, "global_step": 48917, "epoch": 461, "lr": 1.4565122655240793e-06} +{"train_loss": 0.013996670953929424, "grad_norm": 0.06954582780599594, "global_step": 48918, "epoch": 461, "lr": 1.4558022090634504e-06} +{"train_loss": 0.012645119801163673, "grad_norm": 0.13070732355117798, "global_step": 48919, "epoch": 461, "lr": 1.455092323166024e-06} +{"train_loss": 0.01696554198861122, "grad_norm": 0.07392772287130356, "global_step": 48920, "epoch": 461, "lr": 1.4543826078342926e-06} +{"train_loss": 0.016366152092814445, "grad_norm": 0.07958362996578217, "global_step": 48921, "epoch": 461, "lr": 1.4536730630707429e-06} +{"train_loss": 0.014977728947997093, "grad_norm": 0.09567593783140182, "global_step": 48922, "epoch": 461, "lr": 1.4529636888778786e-06} +{"train_loss": 0.016334500163793564, "grad_norm": 0.09809703379869461, "global_step": 48923, "epoch": 461, "lr": 1.452254485258181e-06} +{"train_loss": 0.017221281304955482, "grad_norm": 0.06587919592857361, "global_step": 48924, "epoch": 461, "lr": 1.4515454522141536e-06} +{"train_loss": 0.016011398285627365, "grad_norm": 0.0693686231970787, "global_step": 48925, "epoch": 461, "lr": 1.4508365897482833e-06} +{"train_loss": 0.016132093966007233, "grad_norm": 0.07395877689123154, "global_step": 48926, "epoch": 461, "lr": 1.4501278978630573e-06} +{"train_loss": 0.014645969495177269, "grad_norm": 0.07299332320690155, "global_step": 48927, "epoch": 461, "lr": 1.4494193765609675e-06} +{"train_loss": 0.014655856415629387, "grad_norm": 0.05605004355311394, "global_step": 48928, "epoch": 461, "lr": 1.4487110258445014e-06} +{"train_loss": 0.013151869177818298, "grad_norm": 0.13065820932388306, "global_step": 48929, "epoch": 461, "lr": 1.4480028457161566e-06} +{"train_loss": 0.01741619221866131, "grad_norm": 0.0802590399980545, "global_step": 48930, "epoch": 461, "lr": 1.4472948361784144e-06} +{"train_loss": 0.013184109702706337, "grad_norm": 0.10120078176259995, "global_step": 48931, "epoch": 461, "lr": 1.446586997233762e-06} +{"train_loss": 0.01350066252052784, "grad_norm": 0.08046534657478333, "global_step": 48932, "epoch": 461, "lr": 1.4458793288846918e-06} +{"train_loss": 0.018542872741818428, "grad_norm": 0.15377403795719147, "global_step": 48933, "epoch": 461, "lr": 1.4451718311336793e-06} +{"train_loss": 0.012809268198907375, "grad_norm": 0.05933689326047897, "global_step": 48934, "epoch": 461, "lr": 1.4444645039832117e-06} +{"train_loss": 0.014907462522387505, "grad_norm": 0.12781092524528503, "global_step": 48935, "epoch": 461, "lr": 1.4437573474357923e-06} +{"train_loss": 0.016215886920690536, "grad_norm": 0.06700875610113144, "global_step": 48936, "epoch": 461, "lr": 1.443050361493886e-06} +{"train_loss": 0.014750231057405472, "grad_norm": 0.06498458236455917, "global_step": 48937, "epoch": 461, "lr": 1.4423435461599854e-06} +{"train_loss": 0.0158892460167408, "grad_norm": 0.06051504611968994, "global_step": 48938, "epoch": 461, "lr": 1.4416369014365716e-06} +{"train_loss": 0.014298086985945702, "grad_norm": 0.126902237534523, "global_step": 48939, "epoch": 461, "lr": 1.4409304273261314e-06} +{"train_loss": 0.016015244647860527, "grad_norm": 0.06642178446054459, "global_step": 48940, "epoch": 461, "lr": 1.4402241238311409e-06} +{"train_loss": 0.013490035198628902, "grad_norm": 0.06736364960670471, "global_step": 48941, "epoch": 461, "lr": 1.4395179909540812e-06} +{"train_loss": 0.015305662527680397, "grad_norm": 0.0846996009349823, "global_step": 48942, "epoch": 461, "lr": 1.438812028697445e-06} +{"train_loss": 0.013052771799266338, "grad_norm": 0.055543623864650726, "global_step": 48943, "epoch": 461, "lr": 1.438106237063702e-06} +{"train_loss": 0.013774799183011055, "grad_norm": 0.08479078114032745, "global_step": 48944, "epoch": 461, "lr": 1.4374006160553287e-06} +{"train_loss": 0.013836013153195381, "grad_norm": 0.08196838945150375, "global_step": 48945, "epoch": 461, "lr": 1.4366951656748173e-06} +{"train_loss": 0.0142283346503973, "grad_norm": 0.11332149803638458, "global_step": 48946, "epoch": 461, "lr": 1.4359898859246378e-06} +{"train_loss": 0.01754099875688553, "grad_norm": 0.06493975222110748, "global_step": 48947, "epoch": 461, "lr": 1.4352847768072663e-06} +{"train_loss": 0.014398280531167984, "grad_norm": 0.0564284548163414, "global_step": 48948, "epoch": 461, "lr": 1.4345798383251841e-06} +{"train_loss": 0.011655474081635475, "grad_norm": 0.05078205093741417, "global_step": 48949, "epoch": 461, "lr": 1.433875070480878e-06} +{"train_loss": 0.013728915713727474, "grad_norm": 0.06966043263673782, "global_step": 48950, "epoch": 461, "lr": 1.433170473276807e-06} +{"train_loss": 0.015856176614761353, "grad_norm": 0.07110315561294556, "global_step": 48951, "epoch": 461, "lr": 1.4324660467154471e-06} +{"train_loss": 0.015254592522978783, "grad_norm": 0.07351796329021454, "global_step": 48952, "epoch": 461, "lr": 1.4317617907992909e-06} +{"train_loss": 0.011597787961363792, "grad_norm": 0.05824706703424454, "global_step": 48953, "epoch": 461, "lr": 1.4310577055307972e-06} +{"train_loss": 0.01550487894564867, "grad_norm": 0.15265785157680511, "global_step": 48954, "epoch": 461, "lr": 1.4303537909124476e-06} +{"train_loss": 0.014347618445754051, "grad_norm": 0.06116119772195816, "global_step": 48955, "epoch": 461, "lr": 1.4296500469467066e-06} +{"train_loss": 0.012903117574751377, "grad_norm": 0.07998891174793243, "global_step": 48956, "epoch": 461, "lr": 1.4289464736360558e-06} +{"train_loss": 0.013818569481372833, "grad_norm": 0.10381647944450378, "global_step": 48957, "epoch": 461, "lr": 1.4282430709829653e-06} +{"train_loss": 0.014912053942680359, "grad_norm": 0.08025240898132324, "global_step": 48958, "epoch": 461, "lr": 1.4275398389898996e-06} +{"train_loss": 0.01716504618525505, "grad_norm": 0.07536435127258301, "global_step": 48959, "epoch": 461, "lr": 1.4268367776593405e-06} +{"train_loss": 0.014498618431389332, "grad_norm": 0.06768829375505447, "global_step": 48960, "epoch": 461, "lr": 1.426133886993758e-06} +{"train_loss": 0.017206883057951927, "grad_norm": 0.08254808932542801, "global_step": 48961, "epoch": 461, "lr": 1.4254311669956056e-06} +{"train_loss": 0.0161062590777874, "grad_norm": 0.09917762875556946, "global_step": 48962, "epoch": 461, "lr": 1.42472861766737e-06} +{"train_loss": 0.013549293391406536, "grad_norm": 0.06066693738102913, "global_step": 48963, "epoch": 461, "lr": 1.4240262390115167e-06} +{"train_loss": 0.016849346458911896, "grad_norm": 0.06939642876386642, "global_step": 48964, "epoch": 461, "lr": 1.4233240310304985e-06} +{"train_loss": 0.013782545924186707, "grad_norm": 0.1051497757434845, "global_step": 48965, "epoch": 461, "lr": 1.4226219937268025e-06} +{"train_loss": 0.014548632316291332, "grad_norm": 0.08587189018726349, "global_step": 48966, "epoch": 461, "lr": 1.4219201271028882e-06} +{"train_loss": 0.016769416630268097, "grad_norm": 0.09401942044496536, "global_step": 48967, "epoch": 461, "lr": 1.4212184311612197e-06} +{"train_loss": 0.01422747503966093, "grad_norm": 0.048464223742485046, "global_step": 48968, "epoch": 461, "lr": 1.4205169059042623e-06} +{"train_loss": 0.014510909095406532, "grad_norm": 0.09013792127370834, "global_step": 48969, "epoch": 461, "lr": 1.4198155513344746e-06} +{"train_loss": 0.014028487727046013, "grad_norm": 0.06083459407091141, "global_step": 48970, "epoch": 461, "lr": 1.4191143674543383e-06} +{"train_loss": 0.014952604272314962, "grad_norm": 0.09746143221855164, "global_step": 48971, "epoch": 461, "lr": 1.4184133542663014e-06, "val_loss": 0.01953023672103882} +{"train_loss": 0.016794484108686447, "grad_norm": 0.07227808982133865, "global_step": 48972, "epoch": 462, "lr": 1.4177125117728396e-06} +{"train_loss": 0.01426011323928833, "grad_norm": 0.05921049788594246, "global_step": 48973, "epoch": 462, "lr": 1.4170118399764064e-06} +{"train_loss": 0.014766525477170944, "grad_norm": 0.11437427997589111, "global_step": 48974, "epoch": 462, "lr": 1.4163113388794558e-06} +{"train_loss": 0.015421566553413868, "grad_norm": 0.07542313635349274, "global_step": 48975, "epoch": 462, "lr": 1.4156110084844632e-06} +{"train_loss": 0.014733769930899143, "grad_norm": 0.08836274594068527, "global_step": 48976, "epoch": 462, "lr": 1.4149108487938877e-06} +{"train_loss": 0.01415138877928257, "grad_norm": 0.07389449328184128, "global_step": 48977, "epoch": 462, "lr": 1.4142108598101833e-06} +{"train_loss": 0.01680801436305046, "grad_norm": 0.08937472105026245, "global_step": 48978, "epoch": 462, "lr": 1.41351104153582e-06} +{"train_loss": 0.015220385044813156, "grad_norm": 0.05479129031300545, "global_step": 48979, "epoch": 462, "lr": 1.4128113939732402e-06} +{"train_loss": 0.015205280855298042, "grad_norm": 0.06158941239118576, "global_step": 48980, "epoch": 462, "lr": 1.4121119171249142e-06} +{"train_loss": 0.013505102135241032, "grad_norm": 0.07735620439052582, "global_step": 48981, "epoch": 462, "lr": 1.4114126109932958e-06} +{"train_loss": 0.014357036910951138, "grad_norm": 0.07058095186948776, "global_step": 48982, "epoch": 462, "lr": 1.4107134755808437e-06} +{"train_loss": 0.012985620647668839, "grad_norm": 0.07924412935972214, "global_step": 48983, "epoch": 462, "lr": 1.4100145108900175e-06} +{"train_loss": 0.016699910163879395, "grad_norm": 0.09911392629146576, "global_step": 48984, "epoch": 462, "lr": 1.4093157169232651e-06} +{"train_loss": 0.01508188433945179, "grad_norm": 0.06186864525079727, "global_step": 48985, "epoch": 462, "lr": 1.4086170936830457e-06} +{"train_loss": 0.013288384303450584, "grad_norm": 0.11318375170230865, "global_step": 48986, "epoch": 462, "lr": 1.4079186411718127e-06} +{"train_loss": 0.013259051367640495, "grad_norm": 0.0581943653523922, "global_step": 48987, "epoch": 462, "lr": 1.4072203593920252e-06} +{"train_loss": 0.015385139733552933, "grad_norm": 0.09418012201786041, "global_step": 48988, "epoch": 462, "lr": 1.406522248346126e-06} +{"train_loss": 0.015150394290685654, "grad_norm": 0.07569557428359985, "global_step": 48989, "epoch": 462, "lr": 1.4058243080365796e-06} +{"train_loss": 0.0159613024443388, "grad_norm": 0.1255776584148407, "global_step": 48990, "epoch": 462, "lr": 1.4051265384658396e-06} +{"train_loss": 0.012611474841833115, "grad_norm": 0.08131228387355804, "global_step": 48991, "epoch": 462, "lr": 1.4044289396363431e-06} +{"train_loss": 0.015646561980247498, "grad_norm": 0.07515382021665573, "global_step": 48992, "epoch": 462, "lr": 1.403731511550549e-06} +{"train_loss": 0.014548512175679207, "grad_norm": 0.09494827687740326, "global_step": 48993, "epoch": 462, "lr": 1.4030342542109164e-06} +{"train_loss": 0.01531461626291275, "grad_norm": 0.0670427531003952, "global_step": 48994, "epoch": 462, "lr": 1.402337167619877e-06} +{"train_loss": 0.013546520844101906, "grad_norm": 0.07383847236633301, "global_step": 48995, "epoch": 462, "lr": 1.4016402517798955e-06} +{"train_loss": 0.01561826467514038, "grad_norm": 0.07540489733219147, "global_step": 48996, "epoch": 462, "lr": 1.4009435066934196e-06} +{"train_loss": 0.015281546860933304, "grad_norm": 0.1104658842086792, "global_step": 48997, "epoch": 462, "lr": 1.400246932362892e-06} +{"train_loss": 0.014952577650547028, "grad_norm": 0.11202847957611084, "global_step": 48998, "epoch": 462, "lr": 1.399550528790755e-06} +{"train_loss": 0.013610437512397766, "grad_norm": 0.05761731043457985, "global_step": 48999, "epoch": 462, "lr": 1.3988542959794627e-06} +{"train_loss": 0.012442701496183872, "grad_norm": 0.10651182383298874, "global_step": 49000, "epoch": 462, "lr": 1.3981582339314625e-06} +{"train_loss": 0.015052475035190582, "grad_norm": 0.07030747085809708, "global_step": 49001, "epoch": 462, "lr": 1.3974623426491973e-06} +{"train_loss": 0.014046902768313885, "grad_norm": 0.09289678186178207, "global_step": 49002, "epoch": 462, "lr": 1.396766622135115e-06} +{"train_loss": 0.014469951391220093, "grad_norm": 0.08814568817615509, "global_step": 49003, "epoch": 462, "lr": 1.3960710723916525e-06} +{"train_loss": 0.016046369448304176, "grad_norm": 0.08821216225624084, "global_step": 49004, "epoch": 462, "lr": 1.3953756934212636e-06} +{"train_loss": 0.018254440277814865, "grad_norm": 0.1149919405579567, "global_step": 49005, "epoch": 462, "lr": 1.394680485226385e-06} +{"train_loss": 0.015018796548247337, "grad_norm": 0.10427109897136688, "global_step": 49006, "epoch": 462, "lr": 1.3939854478094594e-06} +{"train_loss": 0.014865974895656109, "grad_norm": 0.061803240329027176, "global_step": 49007, "epoch": 462, "lr": 1.3932905811729403e-06} +{"train_loss": 0.0139199523255229, "grad_norm": 0.1104675680398941, "global_step": 49008, "epoch": 462, "lr": 1.3925958853192478e-06} +{"train_loss": 0.018804918974637985, "grad_norm": 0.0779055505990982, "global_step": 49009, "epoch": 462, "lr": 1.3919013602508412e-06} +{"train_loss": 0.013859635218977928, "grad_norm": 0.06747490912675858, "global_step": 49010, "epoch": 462, "lr": 1.3912070059701464e-06} +{"train_loss": 0.016352638602256775, "grad_norm": 0.0786035880446434, "global_step": 49011, "epoch": 462, "lr": 1.3905128224796227e-06} +{"train_loss": 0.013745367527008057, "grad_norm": 0.07168310135602951, "global_step": 49012, "epoch": 462, "lr": 1.38981880978169e-06} +{"train_loss": 0.014951040968298912, "grad_norm": 0.10262653231620789, "global_step": 49013, "epoch": 462, "lr": 1.3891249678787965e-06} +{"train_loss": 0.014756139367818832, "grad_norm": 0.09759340435266495, "global_step": 49014, "epoch": 462, "lr": 1.3884312967733793e-06} +{"train_loss": 0.016378218308091164, "grad_norm": 0.15911145508289337, "global_step": 49015, "epoch": 462, "lr": 1.387737796467864e-06} +{"train_loss": 0.01300906203687191, "grad_norm": 0.0831223800778389, "global_step": 49016, "epoch": 462, "lr": 1.3870444669647042e-06} +{"train_loss": 0.017665240913629532, "grad_norm": 0.0679132267832756, "global_step": 49017, "epoch": 462, "lr": 1.3863513082663315e-06} +{"train_loss": 0.01668647676706314, "grad_norm": 0.0613446906208992, "global_step": 49018, "epoch": 462, "lr": 1.3856583203751716e-06} +{"train_loss": 0.017151115462183952, "grad_norm": 0.07297301292419434, "global_step": 49019, "epoch": 462, "lr": 1.3849655032936782e-06} +{"train_loss": 0.014133643358945847, "grad_norm": 0.10317102819681168, "global_step": 49020, "epoch": 462, "lr": 1.3842728570242602e-06} +{"train_loss": 0.015101214870810509, "grad_norm": 0.10310174524784088, "global_step": 49021, "epoch": 462, "lr": 1.383580381569377e-06} +{"train_loss": 0.01396972592920065, "grad_norm": 0.11657775193452835, "global_step": 49022, "epoch": 462, "lr": 1.382888076931438e-06} +{"train_loss": 0.01398172415792942, "grad_norm": 0.0911245346069336, "global_step": 49023, "epoch": 462, "lr": 1.3821959431128905e-06} +{"train_loss": 0.013377714902162552, "grad_norm": 0.09865261614322662, "global_step": 49024, "epoch": 462, "lr": 1.3815039801161721e-06} +{"train_loss": 0.01578308828175068, "grad_norm": 0.08055956661701202, "global_step": 49025, "epoch": 462, "lr": 1.380812187943703e-06} +{"train_loss": 0.014921322464942932, "grad_norm": 0.08320550620555878, "global_step": 49026, "epoch": 462, "lr": 1.380120566597909e-06} +{"train_loss": 0.01700594462454319, "grad_norm": 0.12821735441684723, "global_step": 49027, "epoch": 462, "lr": 1.3794291160812267e-06} +{"train_loss": 0.01382039301097393, "grad_norm": 0.07465720176696777, "global_step": 49028, "epoch": 462, "lr": 1.3787378363960934e-06} +{"train_loss": 0.014720126986503601, "grad_norm": 0.09078293293714523, "global_step": 49029, "epoch": 462, "lr": 1.3780467275449238e-06} +{"train_loss": 0.015447456389665604, "grad_norm": 0.09367960691452026, "global_step": 49030, "epoch": 462, "lr": 1.3773557895301492e-06} +{"train_loss": 0.014854636043310165, "grad_norm": 0.06383811682462692, "global_step": 49031, "epoch": 462, "lr": 1.3766650223542176e-06} +{"train_loss": 0.0140951257199049, "grad_norm": 0.06149589642882347, "global_step": 49032, "epoch": 462, "lr": 1.3759744260195273e-06} +{"train_loss": 0.014099976979196072, "grad_norm": 0.07019831240177155, "global_step": 49033, "epoch": 462, "lr": 1.375284000528515e-06} +{"train_loss": 0.014035655185580254, "grad_norm": 0.08795128762722015, "global_step": 49034, "epoch": 462, "lr": 1.3745937458836123e-06} +{"train_loss": 0.01697571948170662, "grad_norm": 0.10269048064947128, "global_step": 49035, "epoch": 462, "lr": 1.3739036620872391e-06} +{"train_loss": 0.014661258086562157, "grad_norm": 0.0635441392660141, "global_step": 49036, "epoch": 462, "lr": 1.3732137491418217e-06} +{"train_loss": 0.015140041708946228, "grad_norm": 0.11217521876096725, "global_step": 49037, "epoch": 462, "lr": 1.3725240070497858e-06} +{"train_loss": 0.014867369085550308, "grad_norm": 0.05290273576974869, "global_step": 49038, "epoch": 462, "lr": 1.3718344358135572e-06} +{"train_loss": 0.015239027328789234, "grad_norm": 0.08768915385007858, "global_step": 49039, "epoch": 462, "lr": 1.3711450354355449e-06} +{"train_loss": 0.0135636106133461, "grad_norm": 0.09315986186265945, "global_step": 49040, "epoch": 462, "lr": 1.3704558059181805e-06} +{"train_loss": 0.015117213129997253, "grad_norm": 0.09255111217498779, "global_step": 49041, "epoch": 462, "lr": 1.3697667472638954e-06} +{"train_loss": 0.015357865020632744, "grad_norm": 0.09924928843975067, "global_step": 49042, "epoch": 462, "lr": 1.369077859475093e-06} +{"train_loss": 0.01602023094892502, "grad_norm": 0.08782585710287094, "global_step": 49043, "epoch": 462, "lr": 1.3683891425542106e-06} +{"train_loss": 0.013747408986091614, "grad_norm": 0.059107135981321335, "global_step": 49044, "epoch": 462, "lr": 1.3677005965036516e-06} +{"train_loss": 0.013760432600975037, "grad_norm": 0.07500023394823074, "global_step": 49045, "epoch": 462, "lr": 1.3670122213258474e-06} +{"train_loss": 0.012489070184528828, "grad_norm": 0.130707249045372, "global_step": 49046, "epoch": 462, "lr": 1.3663240170232071e-06} +{"train_loss": 0.014715306460857391, "grad_norm": 0.0960342288017273, "global_step": 49047, "epoch": 462, "lr": 1.365635983598157e-06} +{"train_loss": 0.016798710450530052, "grad_norm": 0.09491921216249466, "global_step": 49048, "epoch": 462, "lr": 1.3649481210531168e-06} +{"train_loss": 0.015060880221426487, "grad_norm": 0.1373308002948761, "global_step": 49049, "epoch": 462, "lr": 1.3642604293904959e-06} +{"train_loss": 0.014146873727440834, "grad_norm": 0.05645091459155083, "global_step": 49050, "epoch": 462, "lr": 1.3635729086127035e-06} +{"train_loss": 0.015661083161830902, "grad_norm": 0.12822873890399933, "global_step": 49051, "epoch": 462, "lr": 1.3628855587221767e-06} +{"train_loss": 0.012927348725497723, "grad_norm": 0.08799388259649277, "global_step": 49052, "epoch": 462, "lr": 1.3621983797213079e-06} +{"train_loss": 0.013832544907927513, "grad_norm": 0.1137847751379013, "global_step": 49053, "epoch": 462, "lr": 1.3615113716125283e-06} +{"train_loss": 0.013744911178946495, "grad_norm": 0.09771566838026047, "global_step": 49054, "epoch": 462, "lr": 1.3608245343982474e-06} +{"train_loss": 0.013575541786849499, "grad_norm": 0.07215014100074768, "global_step": 49055, "epoch": 462, "lr": 1.3601378680808741e-06} +{"train_loss": 0.015170300379395485, "grad_norm": 0.060081228613853455, "global_step": 49056, "epoch": 462, "lr": 1.3594513726628177e-06} +{"train_loss": 0.01573517546057701, "grad_norm": 0.055001355707645416, "global_step": 49057, "epoch": 462, "lr": 1.3587650481464987e-06} +{"train_loss": 0.01874689757823944, "grad_norm": 0.09700065851211548, "global_step": 49058, "epoch": 462, "lr": 1.3580788945343315e-06} +{"train_loss": 0.01704135537147522, "grad_norm": 0.07398372888565063, "global_step": 49059, "epoch": 462, "lr": 1.3573929118287143e-06} +{"train_loss": 0.013064438477158546, "grad_norm": 0.11465217173099518, "global_step": 49060, "epoch": 462, "lr": 1.3567071000320731e-06} +{"train_loss": 0.014167883433401585, "grad_norm": 0.1063264012336731, "global_step": 49061, "epoch": 462, "lr": 1.3560214591468002e-06} +{"train_loss": 0.013710396364331245, "grad_norm": 0.05610545724630356, "global_step": 49062, "epoch": 462, "lr": 1.3553359891753159e-06} +{"train_loss": 0.015094738453626633, "grad_norm": 0.05960656702518463, "global_step": 49063, "epoch": 462, "lr": 1.354650690120024e-06} +{"train_loss": 0.017273345962166786, "grad_norm": 0.07109975814819336, "global_step": 49064, "epoch": 462, "lr": 1.3539655619833337e-06} +{"train_loss": 0.015325641259551048, "grad_norm": 0.14040032029151917, "global_step": 49065, "epoch": 462, "lr": 1.3532806047676594e-06} +{"train_loss": 0.012560717761516571, "grad_norm": 0.0656767264008522, "global_step": 49066, "epoch": 462, "lr": 1.3525958184753994e-06} +{"train_loss": 0.012658304534852505, "grad_norm": 0.04892317205667496, "global_step": 49067, "epoch": 462, "lr": 1.3519112031089577e-06} +{"train_loss": 0.015054058283567429, "grad_norm": 0.10218099504709244, "global_step": 49068, "epoch": 462, "lr": 1.3512267586707427e-06} +{"train_loss": 0.014695178717374802, "grad_norm": 0.09393438696861267, "global_step": 49069, "epoch": 462, "lr": 1.3505424851631643e-06} +{"train_loss": 0.015256487764418125, "grad_norm": 0.05315544083714485, "global_step": 49070, "epoch": 462, "lr": 1.3498583825886201e-06} +{"train_loss": 0.0156259648501873, "grad_norm": 0.10201887786388397, "global_step": 49071, "epoch": 462, "lr": 1.3491744509495142e-06} +{"train_loss": 0.01605764962732792, "grad_norm": 0.06567326933145523, "global_step": 49072, "epoch": 462, "lr": 1.348490690248261e-06} +{"train_loss": 0.018710825592279434, "grad_norm": 0.062454041093587875, "global_step": 49073, "epoch": 462, "lr": 1.3478071004872472e-06} +{"train_loss": 0.017247457057237625, "grad_norm": 0.12521278858184814, "global_step": 49074, "epoch": 462, "lr": 1.347123681668877e-06} +{"train_loss": 0.014371076598763466, "grad_norm": 0.09435734897851944, "global_step": 49075, "epoch": 462, "lr": 1.3464404337955649e-06} +{"train_loss": 0.012612110003829002, "grad_norm": 0.06234988570213318, "global_step": 49076, "epoch": 462, "lr": 1.3457573568696923e-06} +{"train_loss": 0.014959207419657483, "grad_norm": 0.09942931681871414, "global_step": 49077, "epoch": 462, "lr": 1.3450744508936686e-06, "val_loss": 0.020174063742160797} +{"train_loss": 0.013083649799227715, "grad_norm": 0.10574745386838913, "global_step": 49078, "epoch": 463, "lr": 1.3443917158699027e-06} +{"train_loss": 0.01489641796797514, "grad_norm": 0.06147220730781555, "global_step": 49079, "epoch": 463, "lr": 1.3437091518007816e-06} +{"train_loss": 0.017401188611984253, "grad_norm": 0.13060973584651947, "global_step": 49080, "epoch": 463, "lr": 1.3430267586887036e-06} +{"train_loss": 0.010329792276024818, "grad_norm": 0.11900259554386139, "global_step": 49081, "epoch": 463, "lr": 1.3423445365360721e-06} +{"train_loss": 0.015707839280366898, "grad_norm": 0.07889743894338608, "global_step": 49082, "epoch": 463, "lr": 1.3416624853452796e-06} +{"train_loss": 0.017359290271997452, "grad_norm": 0.07512780278921127, "global_step": 49083, "epoch": 463, "lr": 1.3409806051187246e-06} +{"train_loss": 0.016346685588359833, "grad_norm": 0.071005679666996, "global_step": 49084, "epoch": 463, "lr": 1.3402988958588048e-06} +{"train_loss": 0.015578191727399826, "grad_norm": 0.15337136387825012, "global_step": 49085, "epoch": 463, "lr": 1.3396173575679072e-06} +{"train_loss": 0.01477908156812191, "grad_norm": 0.12799213826656342, "global_step": 49086, "epoch": 463, "lr": 1.338935990248441e-06} +{"train_loss": 0.016290979459881783, "grad_norm": 0.06179991737008095, "global_step": 49087, "epoch": 463, "lr": 1.3382547939027823e-06} +{"train_loss": 0.013444801792502403, "grad_norm": 0.06762302666902542, "global_step": 49088, "epoch": 463, "lr": 1.3375737685333345e-06} +{"train_loss": 0.013452693819999695, "grad_norm": 0.1076945811510086, "global_step": 49089, "epoch": 463, "lr": 1.3368929141424957e-06} +{"train_loss": 0.014687355607748032, "grad_norm": 0.05561614781618118, "global_step": 49090, "epoch": 463, "lr": 1.336212230732653e-06} +{"train_loss": 0.014264335855841637, "grad_norm": 0.07228349149227142, "global_step": 49091, "epoch": 463, "lr": 1.3355317183061933e-06} +{"train_loss": 0.018074125051498413, "grad_norm": 0.07072409242391586, "global_step": 49092, "epoch": 463, "lr": 1.3348513768655147e-06} +{"train_loss": 0.01608285680413246, "grad_norm": 0.06625708937644958, "global_step": 49093, "epoch": 463, "lr": 1.3341712064129985e-06} +{"train_loss": 0.014331458136439323, "grad_norm": 0.07728513330221176, "global_step": 49094, "epoch": 463, "lr": 1.333491206951043e-06} +{"train_loss": 0.014245694503188133, "grad_norm": 0.118345707654953, "global_step": 49095, "epoch": 463, "lr": 1.3328113784820407e-06} +{"train_loss": 0.01525274571031332, "grad_norm": 0.07470173388719559, "global_step": 49096, "epoch": 463, "lr": 1.3321317210083728e-06} +{"train_loss": 0.014667196199297905, "grad_norm": 0.17630581557750702, "global_step": 49097, "epoch": 463, "lr": 1.331452234532421e-06} +{"train_loss": 0.014400070533156395, "grad_norm": 0.08799243718385696, "global_step": 49098, "epoch": 463, "lr": 1.3307729190565832e-06} +{"train_loss": 0.014303948730230331, "grad_norm": 0.08415300399065018, "global_step": 49099, "epoch": 463, "lr": 1.3300937745832521e-06} +{"train_loss": 0.013103429228067398, "grad_norm": 0.08140914142131805, "global_step": 49100, "epoch": 463, "lr": 1.3294148011147922e-06} +{"train_loss": 0.01580951362848282, "grad_norm": 0.07054693251848221, "global_step": 49101, "epoch": 463, "lr": 1.3287359986536075e-06} +{"train_loss": 0.016246318817138672, "grad_norm": 0.08181614428758621, "global_step": 49102, "epoch": 463, "lr": 1.3280573672020846e-06} +{"train_loss": 0.016598127782344818, "grad_norm": 0.10166341811418533, "global_step": 49103, "epoch": 463, "lr": 1.3273789067625998e-06} +{"train_loss": 0.012692907825112343, "grad_norm": 0.1200193464756012, "global_step": 49104, "epoch": 463, "lr": 1.3267006173375341e-06} +{"train_loss": 0.013313233852386475, "grad_norm": 0.056546296924352646, "global_step": 49105, "epoch": 463, "lr": 1.3260224989292747e-06} +{"train_loss": 0.014861397445201874, "grad_norm": 0.09790083765983582, "global_step": 49106, "epoch": 463, "lr": 1.3253445515402086e-06} +{"train_loss": 0.01752130500972271, "grad_norm": 0.07302332669496536, "global_step": 49107, "epoch": 463, "lr": 1.3246667751727116e-06} +{"train_loss": 0.014234626665711403, "grad_norm": 0.05388278514146805, "global_step": 49108, "epoch": 463, "lr": 1.3239891698291651e-06} +{"train_loss": 0.013385245576500893, "grad_norm": 0.11511530727148056, "global_step": 49109, "epoch": 463, "lr": 1.3233117355119562e-06} +{"train_loss": 0.016457900404930115, "grad_norm": 0.08419974893331528, "global_step": 49110, "epoch": 463, "lr": 1.322634472223455e-06} +{"train_loss": 0.014253322966396809, "grad_norm": 0.06571436673402786, "global_step": 49111, "epoch": 463, "lr": 1.321957379966049e-06} +{"train_loss": 0.014670044183731079, "grad_norm": 0.06232338026165962, "global_step": 49112, "epoch": 463, "lr": 1.3212804587421135e-06} +{"train_loss": 0.015975866466760635, "grad_norm": 0.07662662118673325, "global_step": 49113, "epoch": 463, "lr": 1.3206037085540358e-06} +{"train_loss": 0.0131938885897398, "grad_norm": 0.07690006494522095, "global_step": 49114, "epoch": 463, "lr": 1.3199271294041803e-06} +{"train_loss": 0.014394882135093212, "grad_norm": 0.09107264131307602, "global_step": 49115, "epoch": 463, "lr": 1.3192507212949346e-06} +{"train_loss": 0.014821629971265793, "grad_norm": 0.12813329696655273, "global_step": 49116, "epoch": 463, "lr": 1.3185744842286686e-06} +{"train_loss": 0.017255399376153946, "grad_norm": 0.07017497718334198, "global_step": 49117, "epoch": 463, "lr": 1.317898418207758e-06} +{"train_loss": 0.014606432989239693, "grad_norm": 0.07226506620645523, "global_step": 49118, "epoch": 463, "lr": 1.3172225232345846e-06} +{"train_loss": 0.016251562163233757, "grad_norm": 0.06007957085967064, "global_step": 49119, "epoch": 463, "lr": 1.3165467993115244e-06} +{"train_loss": 0.017143307253718376, "grad_norm": 0.0625683143734932, "global_step": 49120, "epoch": 463, "lr": 1.3158712464409417e-06} +{"train_loss": 0.015918057411909103, "grad_norm": 0.06497592478990555, "global_step": 49121, "epoch": 463, "lr": 1.3151958646252128e-06} +{"train_loss": 0.01382268313318491, "grad_norm": 0.09252865612506866, "global_step": 49122, "epoch": 463, "lr": 1.3145206538667133e-06} +{"train_loss": 0.01311553455889225, "grad_norm": 0.0613161139190197, "global_step": 49123, "epoch": 463, "lr": 1.3138456141678247e-06} +{"train_loss": 0.01688563823699951, "grad_norm": 0.06480325013399124, "global_step": 49124, "epoch": 463, "lr": 1.3131707455309006e-06} +{"train_loss": 0.015129427425563335, "grad_norm": 0.07422647625207901, "global_step": 49125, "epoch": 463, "lr": 1.3124960479583226e-06} +{"train_loss": 0.012223035097122192, "grad_norm": 0.08236537128686905, "global_step": 49126, "epoch": 463, "lr": 1.311821521452461e-06} +{"train_loss": 0.01672416925430298, "grad_norm": 0.0898670107126236, "global_step": 49127, "epoch": 463, "lr": 1.311147166015686e-06} +{"train_loss": 0.014617346227169037, "grad_norm": 0.06425149738788605, "global_step": 49128, "epoch": 463, "lr": 1.3104729816503624e-06} +{"train_loss": 0.014630576595664024, "grad_norm": 0.07900356501340866, "global_step": 49129, "epoch": 463, "lr": 1.3097989683588608e-06} +{"train_loss": 0.012095054611563683, "grad_norm": 0.0741574838757515, "global_step": 49130, "epoch": 463, "lr": 1.3091251261435566e-06} +{"train_loss": 0.017470331862568855, "grad_norm": 0.07031282037496567, "global_step": 49131, "epoch": 463, "lr": 1.3084514550068095e-06} +{"train_loss": 0.013427656143903732, "grad_norm": 0.06000243499875069, "global_step": 49132, "epoch": 463, "lr": 1.3077779549509894e-06} +{"train_loss": 0.015200542286038399, "grad_norm": 0.22589468955993652, "global_step": 49133, "epoch": 463, "lr": 1.3071046259784614e-06} +{"train_loss": 0.01501318160444498, "grad_norm": 0.07088805735111237, "global_step": 49134, "epoch": 463, "lr": 1.3064314680915902e-06} +{"train_loss": 0.01408527884632349, "grad_norm": 0.06111563742160797, "global_step": 49135, "epoch": 463, "lr": 1.3057584812927404e-06} +{"train_loss": 0.013966443948447704, "grad_norm": 0.13022805750370026, "global_step": 49136, "epoch": 463, "lr": 1.3050856655842825e-06} +{"train_loss": 0.012639162130653858, "grad_norm": 0.06348313391208649, "global_step": 49137, "epoch": 463, "lr": 1.3044130209685813e-06} +{"train_loss": 0.014909780584275723, "grad_norm": 0.06363943964242935, "global_step": 49138, "epoch": 463, "lr": 1.3037405474479846e-06} +{"train_loss": 0.015001866035163403, "grad_norm": 0.1357170194387436, "global_step": 49139, "epoch": 463, "lr": 1.3030682450248689e-06} +{"train_loss": 0.01689532771706581, "grad_norm": 0.13062697649002075, "global_step": 49140, "epoch": 463, "lr": 1.3023961137015982e-06} +{"train_loss": 0.01440894789993763, "grad_norm": 0.06282883137464523, "global_step": 49141, "epoch": 463, "lr": 1.3017241534805269e-06} +{"train_loss": 0.015286222100257874, "grad_norm": 0.10174544900655746, "global_step": 49142, "epoch": 463, "lr": 1.3010523643640193e-06} +{"train_loss": 0.012939726002514362, "grad_norm": 0.07187202572822571, "global_step": 49143, "epoch": 463, "lr": 1.300380746354435e-06} +{"train_loss": 0.016203196719288826, "grad_norm": 0.07451368123292923, "global_step": 49144, "epoch": 463, "lr": 1.2997092994541382e-06} +{"train_loss": 0.01694074273109436, "grad_norm": 0.0793214738368988, "global_step": 49145, "epoch": 463, "lr": 1.2990380236654775e-06} +{"train_loss": 0.013323845341801643, "grad_norm": 0.09191405028104782, "global_step": 49146, "epoch": 463, "lr": 1.2983669189908176e-06} +{"train_loss": 0.013693210668861866, "grad_norm": 0.06540760397911072, "global_step": 49147, "epoch": 463, "lr": 1.2976959854325177e-06} +{"train_loss": 0.015288272872567177, "grad_norm": 0.10504129528999329, "global_step": 49148, "epoch": 463, "lr": 1.297025222992937e-06} +{"train_loss": 0.015220968052744865, "grad_norm": 0.05733421444892883, "global_step": 49149, "epoch": 463, "lr": 1.296354631674429e-06} +{"train_loss": 0.014302188530564308, "grad_norm": 0.08259190618991852, "global_step": 49150, "epoch": 463, "lr": 1.2956842114793533e-06} +{"train_loss": 0.015038888901472092, "grad_norm": 0.08776222914457321, "global_step": 49151, "epoch": 463, "lr": 1.2950139624100522e-06} +{"train_loss": 0.015457019209861755, "grad_norm": 0.08496159315109253, "global_step": 49152, "epoch": 463, "lr": 1.294343884468896e-06} +{"train_loss": 0.01605181023478508, "grad_norm": 0.08696140348911285, "global_step": 49153, "epoch": 463, "lr": 1.293673977658233e-06} +{"train_loss": 0.015417266637086868, "grad_norm": 0.08825923502445221, "global_step": 49154, "epoch": 463, "lr": 1.2930042419804167e-06} +{"train_loss": 0.014829268679022789, "grad_norm": 0.11666350066661835, "global_step": 49155, "epoch": 463, "lr": 1.2923346774378063e-06} +{"train_loss": 0.015033164992928505, "grad_norm": 0.11559514701366425, "global_step": 49156, "epoch": 463, "lr": 1.2916652840327447e-06} +{"train_loss": 0.014446953311562538, "grad_norm": 0.09315700829029083, "global_step": 49157, "epoch": 463, "lr": 1.2909960617675965e-06} +{"train_loss": 0.013774815946817398, "grad_norm": 0.08434727787971497, "global_step": 49158, "epoch": 463, "lr": 1.290327010644693e-06} +{"train_loss": 0.01599004678428173, "grad_norm": 0.13733841478824615, "global_step": 49159, "epoch": 463, "lr": 1.2896581306664047e-06} +{"train_loss": 0.015275832265615463, "grad_norm": 0.06106835976243019, "global_step": 49160, "epoch": 463, "lr": 1.288989421835074e-06} +{"train_loss": 0.015720408409833908, "grad_norm": 0.07436739653348923, "global_step": 49161, "epoch": 463, "lr": 1.2883208841530493e-06} +{"train_loss": 0.014236129820346832, "grad_norm": 0.07120963931083679, "global_step": 49162, "epoch": 463, "lr": 1.287652517622684e-06} +{"train_loss": 0.014811892062425613, "grad_norm": 0.07485876977443695, "global_step": 49163, "epoch": 463, "lr": 1.286984322246315e-06} +{"train_loss": 0.01438232883810997, "grad_norm": 0.07868488132953644, "global_step": 49164, "epoch": 463, "lr": 1.2863162980263077e-06} +{"train_loss": 0.015555249527096748, "grad_norm": 0.0578208789229393, "global_step": 49165, "epoch": 463, "lr": 1.285648444964993e-06} +{"train_loss": 0.013367308303713799, "grad_norm": 0.06962019205093384, "global_step": 49166, "epoch": 463, "lr": 1.2849807630647304e-06} +{"train_loss": 0.01585577055811882, "grad_norm": 0.1304197907447815, "global_step": 49167, "epoch": 463, "lr": 1.2843132523278512e-06} +{"train_loss": 0.012604799121618271, "grad_norm": 0.09487614780664444, "global_step": 49168, "epoch": 463, "lr": 1.2836459127567202e-06} +{"train_loss": 0.01630929857492447, "grad_norm": 0.08761046826839447, "global_step": 49169, "epoch": 463, "lr": 1.2829787443536634e-06} +{"train_loss": 0.014700738713145256, "grad_norm": 0.06625767797231674, "global_step": 49170, "epoch": 463, "lr": 1.2823117471210344e-06} +{"train_loss": 0.01621388830244541, "grad_norm": 0.13029050827026367, "global_step": 49171, "epoch": 463, "lr": 1.2816449210611758e-06} +{"train_loss": 0.014283576980233192, "grad_norm": 0.07571587711572647, "global_step": 49172, "epoch": 463, "lr": 1.2809782661764357e-06} +{"train_loss": 0.016190893948078156, "grad_norm": 0.057175494730472565, "global_step": 49173, "epoch": 463, "lr": 1.2803117824691402e-06} +{"train_loss": 0.014699984341859818, "grad_norm": 0.0618082657456398, "global_step": 49174, "epoch": 463, "lr": 1.2796454699416483e-06} +{"train_loss": 0.01446615345776081, "grad_norm": 0.09991758316755295, "global_step": 49175, "epoch": 463, "lr": 1.2789793285962914e-06} +{"train_loss": 0.015062167309224606, "grad_norm": 0.07428465783596039, "global_step": 49176, "epoch": 463, "lr": 1.2783133584354123e-06} +{"train_loss": 0.014859949238598347, "grad_norm": 0.0675048753619194, "global_step": 49177, "epoch": 463, "lr": 1.2776475594613535e-06} +{"train_loss": 0.015099635347723961, "grad_norm": 0.1079179123044014, "global_step": 49178, "epoch": 463, "lr": 1.2769819316764575e-06} +{"train_loss": 0.012957240454852581, "grad_norm": 0.07398631423711777, "global_step": 49179, "epoch": 463, "lr": 1.2763164750830502e-06} +{"train_loss": 0.015867553651332855, "grad_norm": 0.07434125244617462, "global_step": 49180, "epoch": 463, "lr": 1.2756511896834745e-06} +{"train_loss": 0.016970155760645866, "grad_norm": 0.0745825469493866, "global_step": 49181, "epoch": 463, "lr": 1.274986075480078e-06} +{"train_loss": 0.016892720013856888, "grad_norm": 0.08238770067691803, "global_step": 49182, "epoch": 463, "lr": 1.274321132475187e-06} +{"train_loss": 0.014954691357418613, "grad_norm": 0.07746165990829468, "global_step": 49183, "epoch": 463, "lr": 1.2736563606711382e-06, "val_loss": 0.019955895841121674} +{"train_loss": 0.014896044507622719, "grad_norm": 0.1142505556344986, "global_step": 49184, "epoch": 464, "lr": 1.2729917600702801e-06} +{"train_loss": 0.015722138807177544, "grad_norm": 0.058599215000867844, "global_step": 49185, "epoch": 464, "lr": 1.272327330674933e-06} +{"train_loss": 0.013782736845314503, "grad_norm": 0.09180048853158951, "global_step": 49186, "epoch": 464, "lr": 1.271663072487439e-06} +{"train_loss": 0.016643060371279716, "grad_norm": 0.0741051435470581, "global_step": 49187, "epoch": 464, "lr": 1.2709989855101245e-06} +{"train_loss": 0.015174665488302708, "grad_norm": 0.0671742856502533, "global_step": 49188, "epoch": 464, "lr": 1.270335069745332e-06} +{"train_loss": 0.01565726101398468, "grad_norm": 0.11271163076162338, "global_step": 49189, "epoch": 464, "lr": 1.2696713251953874e-06} +{"train_loss": 0.014754867181181908, "grad_norm": 0.07702772319316864, "global_step": 49190, "epoch": 464, "lr": 1.2690077518626276e-06} +{"train_loss": 0.012871840037405491, "grad_norm": 0.06288870424032211, "global_step": 49191, "epoch": 464, "lr": 1.268344349749384e-06} +{"train_loss": 0.015437118709087372, "grad_norm": 0.05601708963513374, "global_step": 49192, "epoch": 464, "lr": 1.2676811188579773e-06} +{"train_loss": 0.015660516917705536, "grad_norm": 0.07191415131092072, "global_step": 49193, "epoch": 464, "lr": 1.26701805919075e-06} +{"train_loss": 0.01377780456095934, "grad_norm": 0.08923162519931793, "global_step": 49194, "epoch": 464, "lr": 1.2663551707500331e-06} +{"train_loss": 0.014453957788646221, "grad_norm": 0.07519209384918213, "global_step": 49195, "epoch": 464, "lr": 1.2656924535381421e-06} +{"train_loss": 0.014745054766535759, "grad_norm": 0.05449709668755531, "global_step": 49196, "epoch": 464, "lr": 1.2650299075574245e-06} +{"train_loss": 0.016305334866046906, "grad_norm": 0.07113943994045258, "global_step": 49197, "epoch": 464, "lr": 1.26436753281019e-06} +{"train_loss": 0.01501352246850729, "grad_norm": 0.14081211388111115, "global_step": 49198, "epoch": 464, "lr": 1.2637053292987754e-06} +{"train_loss": 0.014787634834647179, "grad_norm": 0.07521501183509827, "global_step": 49199, "epoch": 464, "lr": 1.2630432970255013e-06} +{"train_loss": 0.014785796403884888, "grad_norm": 0.08880601823329926, "global_step": 49200, "epoch": 464, "lr": 1.262381435992699e-06} +{"train_loss": 0.014402923174202442, "grad_norm": 0.08039595931768417, "global_step": 49201, "epoch": 464, "lr": 1.2617197462026998e-06} +{"train_loss": 0.016365841031074524, "grad_norm": 0.07591088861227036, "global_step": 49202, "epoch": 464, "lr": 1.2610582276578188e-06} +{"train_loss": 0.015235114842653275, "grad_norm": 0.09379283338785172, "global_step": 49203, "epoch": 464, "lr": 1.2603968803603815e-06} +{"train_loss": 0.014319255948066711, "grad_norm": 0.055256977677345276, "global_step": 49204, "epoch": 464, "lr": 1.259735704312709e-06} +{"train_loss": 0.01382342167198658, "grad_norm": 0.08841399103403091, "global_step": 49205, "epoch": 464, "lr": 1.2590746995171377e-06} +{"train_loss": 0.013069882988929749, "grad_norm": 0.10353948175907135, "global_step": 49206, "epoch": 464, "lr": 1.2584138659759714e-06} +{"train_loss": 0.013752997852861881, "grad_norm": 0.07819337397813797, "global_step": 49207, "epoch": 464, "lr": 1.2577532036915474e-06} +{"train_loss": 0.014833087101578712, "grad_norm": 0.07334121316671371, "global_step": 49208, "epoch": 464, "lr": 1.2570927126661857e-06} +{"train_loss": 0.016178224235773087, "grad_norm": 0.10556943714618683, "global_step": 49209, "epoch": 464, "lr": 1.2564323929021959e-06} +{"train_loss": 0.01459435373544693, "grad_norm": 0.10760877281427383, "global_step": 49210, "epoch": 464, "lr": 1.2557722444019038e-06} +{"train_loss": 0.014507303014397621, "grad_norm": 0.10930754244327545, "global_step": 49211, "epoch": 464, "lr": 1.2551122671676297e-06} +{"train_loss": 0.016064060851931572, "grad_norm": 0.09351539611816406, "global_step": 49212, "epoch": 464, "lr": 1.2544524612016938e-06} +{"train_loss": 0.01529091876000166, "grad_norm": 0.1207069382071495, "global_step": 49213, "epoch": 464, "lr": 1.2537928265064168e-06} +{"train_loss": 0.01355844084173441, "grad_norm": 0.056928399950265884, "global_step": 49214, "epoch": 464, "lr": 1.253133363084108e-06} +{"train_loss": 0.017658397555351257, "grad_norm": 0.0706384927034378, "global_step": 49215, "epoch": 464, "lr": 1.252474070937093e-06} +{"train_loss": 0.014735320582985878, "grad_norm": 0.10310352593660355, "global_step": 49216, "epoch": 464, "lr": 1.2518149500676812e-06} +{"train_loss": 0.01445369329303503, "grad_norm": 0.06780440360307693, "global_step": 49217, "epoch": 464, "lr": 1.2511560004781874e-06} +{"train_loss": 0.013130735605955124, "grad_norm": 0.07214298099279404, "global_step": 49218, "epoch": 464, "lr": 1.2504972221709431e-06} +{"train_loss": 0.01608630269765854, "grad_norm": 0.09107659012079239, "global_step": 49219, "epoch": 464, "lr": 1.2498386151482467e-06} +{"train_loss": 0.01619160734117031, "grad_norm": 0.06171141937375069, "global_step": 49220, "epoch": 464, "lr": 1.249180179412407e-06} +{"train_loss": 0.01703450083732605, "grad_norm": 0.09769651293754578, "global_step": 49221, "epoch": 464, "lr": 1.2485219149657557e-06} +{"train_loss": 0.016895456239581108, "grad_norm": 0.102199986577034, "global_step": 49222, "epoch": 464, "lr": 1.2478638218105964e-06} +{"train_loss": 0.01302916556596756, "grad_norm": 0.06573142111301422, "global_step": 49223, "epoch": 464, "lr": 1.2472058999492387e-06} +{"train_loss": 0.014576390385627747, "grad_norm": 0.06930463761091232, "global_step": 49224, "epoch": 464, "lr": 1.246548149383997e-06} +{"train_loss": 0.014923427253961563, "grad_norm": 0.07270664721727371, "global_step": 49225, "epoch": 464, "lr": 1.2458905701171863e-06} +{"train_loss": 0.015560099855065346, "grad_norm": 0.062275685369968414, "global_step": 49226, "epoch": 464, "lr": 1.245233162151116e-06} +{"train_loss": 0.015120603144168854, "grad_norm": 0.08090919256210327, "global_step": 49227, "epoch": 464, "lr": 1.2445759254880895e-06} +{"train_loss": 0.014050992205739021, "grad_norm": 0.07368788123130798, "global_step": 49228, "epoch": 464, "lr": 1.2439188601304163e-06} +{"train_loss": 0.014539754949510098, "grad_norm": 0.09406695514917374, "global_step": 49229, "epoch": 464, "lr": 1.2432619660804168e-06} +{"train_loss": 0.016225341707468033, "grad_norm": 0.05722435191273689, "global_step": 49230, "epoch": 464, "lr": 1.2426052433403834e-06} +{"train_loss": 0.01780327409505844, "grad_norm": 0.12298507243394852, "global_step": 49231, "epoch": 464, "lr": 1.2419486919126365e-06} +{"train_loss": 0.017418134957551956, "grad_norm": 0.14776760339736938, "global_step": 49232, "epoch": 464, "lr": 1.2412923117994801e-06} +{"train_loss": 0.014122243970632553, "grad_norm": 0.05564126744866371, "global_step": 49233, "epoch": 464, "lr": 1.2406361030032122e-06} +{"train_loss": 0.013053128495812416, "grad_norm": 0.07151702046394348, "global_step": 49234, "epoch": 464, "lr": 1.2399800655261417e-06} +{"train_loss": 0.01407218724489212, "grad_norm": 0.05780411884188652, "global_step": 49235, "epoch": 464, "lr": 1.239324199370584e-06} +{"train_loss": 0.014907021075487137, "grad_norm": 0.06220034882426262, "global_step": 49236, "epoch": 464, "lr": 1.2386685045388313e-06} +{"train_loss": 0.013968158513307571, "grad_norm": 0.07590549439191818, "global_step": 49237, "epoch": 464, "lr": 1.2380129810331931e-06} +{"train_loss": 0.015022069215774536, "grad_norm": 0.09743910282850266, "global_step": 49238, "epoch": 464, "lr": 1.2373576288559675e-06} +{"train_loss": 0.015191799029707909, "grad_norm": 0.09621047228574753, "global_step": 49239, "epoch": 464, "lr": 1.2367024480094691e-06} +{"train_loss": 0.01396876946091652, "grad_norm": 0.09004779905080795, "global_step": 49240, "epoch": 464, "lr": 1.2360474384959797e-06} +{"train_loss": 0.014729082584381104, "grad_norm": 0.07698412239551544, "global_step": 49241, "epoch": 464, "lr": 1.2353926003178196e-06} +{"train_loss": 0.01414276659488678, "grad_norm": 0.06620656698942184, "global_step": 49242, "epoch": 464, "lr": 1.2347379334772868e-06} +{"train_loss": 0.01916712149977684, "grad_norm": 0.13577549159526825, "global_step": 49243, "epoch": 464, "lr": 1.2340834379766742e-06} +{"train_loss": 0.013659716583788395, "grad_norm": 0.07182399183511734, "global_step": 49244, "epoch": 464, "lr": 1.2334291138182797e-06} +{"train_loss": 0.01434934139251709, "grad_norm": 0.13622714579105377, "global_step": 49245, "epoch": 464, "lr": 1.232774961004407e-06} +{"train_loss": 0.01573873683810234, "grad_norm": 0.08129339665174484, "global_step": 49246, "epoch": 464, "lr": 1.2321209795373655e-06} +{"train_loss": 0.016102967783808708, "grad_norm": 0.0913689136505127, "global_step": 49247, "epoch": 464, "lr": 1.2314671694194312e-06} +{"train_loss": 0.013446325436234474, "grad_norm": 0.059079304337501526, "global_step": 49248, "epoch": 464, "lr": 1.2308135306529134e-06} +{"train_loss": 0.013502020388841629, "grad_norm": 0.0935978814959526, "global_step": 49249, "epoch": 464, "lr": 1.230160063240121e-06} +{"train_loss": 0.014467692002654076, "grad_norm": 0.07538000494241714, "global_step": 49250, "epoch": 464, "lr": 1.229506767183325e-06} +{"train_loss": 0.014242611825466156, "grad_norm": 0.06598642468452454, "global_step": 49251, "epoch": 464, "lr": 1.2288536424848285e-06} +{"train_loss": 0.014636777341365814, "grad_norm": 0.060252752155065536, "global_step": 49252, "epoch": 464, "lr": 1.2282006891469412e-06} +{"train_loss": 0.01736644096672535, "grad_norm": 0.08034073561429977, "global_step": 49253, "epoch": 464, "lr": 1.2275479071719387e-06} +{"train_loss": 0.0144636370241642, "grad_norm": 0.06907345354557037, "global_step": 49254, "epoch": 464, "lr": 1.2268952965621249e-06} +{"train_loss": 0.01616095006465912, "grad_norm": 0.09447610378265381, "global_step": 49255, "epoch": 464, "lr": 1.2262428573197926e-06} +{"train_loss": 0.015498723834753036, "grad_norm": 0.06744486093521118, "global_step": 49256, "epoch": 464, "lr": 1.2255905894472285e-06} +{"train_loss": 0.013404110446572304, "grad_norm": 0.05541520193219185, "global_step": 49257, "epoch": 464, "lr": 1.2249384929467255e-06} +{"train_loss": 0.01567242294549942, "grad_norm": 0.11878958344459534, "global_step": 49258, "epoch": 464, "lr": 1.2242865678205818e-06} +{"train_loss": 0.013672012835741043, "grad_norm": 0.06923551857471466, "global_step": 49259, "epoch": 464, "lr": 1.223634814071084e-06} +{"train_loss": 0.018430441617965698, "grad_norm": 0.13079623878002167, "global_step": 49260, "epoch": 464, "lr": 1.2229832317005196e-06} +{"train_loss": 0.015208837576210499, "grad_norm": 0.05391782522201538, "global_step": 49261, "epoch": 464, "lr": 1.2223318207111756e-06} +{"train_loss": 0.014936378225684166, "grad_norm": 0.13442544639110565, "global_step": 49262, "epoch": 464, "lr": 1.2216805811053445e-06} +{"train_loss": 0.014037733897566795, "grad_norm": 0.09524841606616974, "global_step": 49263, "epoch": 464, "lr": 1.2210295128853188e-06} +{"train_loss": 0.014368506148457527, "grad_norm": 0.14764906466007233, "global_step": 49264, "epoch": 464, "lr": 1.2203786160533804e-06} +{"train_loss": 0.014566820114850998, "grad_norm": 0.061476122587919235, "global_step": 49265, "epoch": 464, "lr": 1.2197278906118158e-06} +{"train_loss": 0.014501238241791725, "grad_norm": 0.09922301769256592, "global_step": 49266, "epoch": 464, "lr": 1.2190773365629238e-06} +{"train_loss": 0.01571711152791977, "grad_norm": 0.12070252746343613, "global_step": 49267, "epoch": 464, "lr": 1.2184269539089687e-06} +{"train_loss": 0.01655624806880951, "grad_norm": 0.10500221699476242, "global_step": 49268, "epoch": 464, "lr": 1.217776742652249e-06} +{"train_loss": 0.015245704911649227, "grad_norm": 0.06414929032325745, "global_step": 49269, "epoch": 464, "lr": 1.2171267027950516e-06} +{"train_loss": 0.01630794070661068, "grad_norm": 0.093857541680336, "global_step": 49270, "epoch": 464, "lr": 1.216476834339647e-06} +{"train_loss": 0.012888718396425247, "grad_norm": 0.09078939259052277, "global_step": 49271, "epoch": 464, "lr": 1.2158271372883333e-06} +{"train_loss": 0.016132663935422897, "grad_norm": 0.0774049460887909, "global_step": 49272, "epoch": 464, "lr": 1.2151776116433866e-06} +{"train_loss": 0.013255334459245205, "grad_norm": 0.07443944364786148, "global_step": 49273, "epoch": 464, "lr": 1.2145282574070937e-06} +{"train_loss": 0.013345934450626373, "grad_norm": 0.07765598595142365, "global_step": 49274, "epoch": 464, "lr": 1.2138790745817253e-06} +{"train_loss": 0.013731824234127998, "grad_norm": 0.0905948206782341, "global_step": 49275, "epoch": 464, "lr": 1.2132300631695681e-06} +{"train_loss": 0.014724457636475563, "grad_norm": 0.09386724978685379, "global_step": 49276, "epoch": 464, "lr": 1.212581223172915e-06} +{"train_loss": 0.014009496197104454, "grad_norm": 0.06080351769924164, "global_step": 49277, "epoch": 464, "lr": 1.2119325545940252e-06} +{"train_loss": 0.016201741993427277, "grad_norm": 0.07389234751462936, "global_step": 49278, "epoch": 464, "lr": 1.2112840574351914e-06} +{"train_loss": 0.014201724901795387, "grad_norm": 0.060522809624671936, "global_step": 49279, "epoch": 464, "lr": 1.2106357316986838e-06} +{"train_loss": 0.01327202096581459, "grad_norm": 0.0845528393983841, "global_step": 49280, "epoch": 464, "lr": 1.209987577386784e-06} +{"train_loss": 0.014048948884010315, "grad_norm": 0.07630807161331177, "global_step": 49281, "epoch": 464, "lr": 1.209339594501774e-06} +{"train_loss": 0.01568075269460678, "grad_norm": 0.09968558698892593, "global_step": 49282, "epoch": 464, "lr": 1.2086917830459178e-06} +{"train_loss": 0.014461527578532696, "grad_norm": 0.06874505430459976, "global_step": 49283, "epoch": 464, "lr": 1.2080441430215083e-06} +{"train_loss": 0.014798860996961594, "grad_norm": 0.08995895832777023, "global_step": 49284, "epoch": 464, "lr": 1.2073966744308163e-06} +{"train_loss": 0.01447156723588705, "grad_norm": 0.1466057300567627, "global_step": 49285, "epoch": 464, "lr": 1.2067493772761062e-06} +{"train_loss": 0.015309728682041168, "grad_norm": 0.12366043031215668, "global_step": 49286, "epoch": 464, "lr": 1.2061022515596544e-06} +{"train_loss": 0.014278793707489967, "grad_norm": 0.17445924878120422, "global_step": 49287, "epoch": 464, "lr": 1.2054552972837474e-06} +{"train_loss": 0.01458345353603363, "grad_norm": 0.08158138394355774, "global_step": 49288, "epoch": 464, "lr": 1.2048085144506449e-06} +{"train_loss": 0.014962313521020818, "grad_norm": 0.10340062528848648, "global_step": 49289, "epoch": 464, "lr": 1.2041619030626284e-06, "val_loss": 0.01933220960199833} +{"train_loss": 0.01752459816634655, "grad_norm": 0.06592722237110138, "global_step": 49290, "epoch": 465, "lr": 1.2035154631219737e-06} +{"train_loss": 0.015114666894078255, "grad_norm": 0.09692821651697159, "global_step": 49291, "epoch": 465, "lr": 1.2028691946309345e-06} +{"train_loss": 0.015358242206275463, "grad_norm": 0.0737554207444191, "global_step": 49292, "epoch": 465, "lr": 1.202223097591787e-06} +{"train_loss": 0.0165092870593071, "grad_norm": 0.08092988282442093, "global_step": 49293, "epoch": 465, "lr": 1.201577172006818e-06} +{"train_loss": 0.01386356819421053, "grad_norm": 0.06728620082139969, "global_step": 49294, "epoch": 465, "lr": 1.200931417878276e-06} +{"train_loss": 0.014199083670973778, "grad_norm": 0.0874452292919159, "global_step": 49295, "epoch": 465, "lr": 1.2002858352084423e-06} +{"train_loss": 0.014245845377445221, "grad_norm": 0.0827954038977623, "global_step": 49296, "epoch": 465, "lr": 1.1996404239995818e-06} +{"train_loss": 0.013355705887079239, "grad_norm": 0.06541755050420761, "global_step": 49297, "epoch": 465, "lr": 1.1989951842539648e-06} +{"train_loss": 0.013502340763807297, "grad_norm": 0.07572474330663681, "global_step": 49298, "epoch": 465, "lr": 1.1983501159738507e-06} +{"train_loss": 0.01449146494269371, "grad_norm": 0.07033570855855942, "global_step": 49299, "epoch": 465, "lr": 1.1977052191615156e-06} +{"train_loss": 0.015500567853450775, "grad_norm": 0.058675091713666916, "global_step": 49300, "epoch": 465, "lr": 1.1970604938192186e-06} +{"train_loss": 0.014133041724562645, "grad_norm": 0.05713168904185295, "global_step": 49301, "epoch": 465, "lr": 1.1964159399492247e-06} +{"train_loss": 0.015380887314677238, "grad_norm": 0.0703476145863533, "global_step": 49302, "epoch": 465, "lr": 1.1957715575538042e-06} +{"train_loss": 0.014214197173714638, "grad_norm": 0.1040935069322586, "global_step": 49303, "epoch": 465, "lr": 1.1951273466352164e-06} +{"train_loss": 0.015406699851155281, "grad_norm": 0.07012313604354858, "global_step": 49304, "epoch": 465, "lr": 1.1944833071957318e-06} +{"train_loss": 0.018261659890413284, "grad_norm": 0.12012603878974915, "global_step": 49305, "epoch": 465, "lr": 1.1938394392375984e-06} +{"train_loss": 0.01645372435450554, "grad_norm": 0.09347838908433914, "global_step": 49306, "epoch": 465, "lr": 1.1931957427630925e-06} +{"train_loss": 0.014409516006708145, "grad_norm": 0.07705140858888626, "global_step": 49307, "epoch": 465, "lr": 1.192552217774473e-06} +{"train_loss": 0.015068169683218002, "grad_norm": 0.1439201831817627, "global_step": 49308, "epoch": 465, "lr": 1.1919088642740051e-06} +{"train_loss": 0.014416125603020191, "grad_norm": 0.05888404697179794, "global_step": 49309, "epoch": 465, "lr": 1.1912656822639312e-06} +{"train_loss": 0.015891505405306816, "grad_norm": 0.07236020267009735, "global_step": 49310, "epoch": 465, "lr": 1.190622671746533e-06} +{"train_loss": 0.01595345512032509, "grad_norm": 0.06757903844118118, "global_step": 49311, "epoch": 465, "lr": 1.189979832724053e-06} +{"train_loss": 0.014619240537285805, "grad_norm": 0.0941244512796402, "global_step": 49312, "epoch": 465, "lr": 1.189337165198756e-06} +{"train_loss": 0.012944537214934826, "grad_norm": 0.06574973464012146, "global_step": 49313, "epoch": 465, "lr": 1.1886946691729074e-06} +{"train_loss": 0.013931604102253914, "grad_norm": 0.07616022974252701, "global_step": 49314, "epoch": 465, "lr": 1.1880523446487601e-06} +{"train_loss": 0.012536490336060524, "grad_norm": 0.06656739860773087, "global_step": 49315, "epoch": 465, "lr": 1.187410191628563e-06} +{"train_loss": 0.015639115124940872, "grad_norm": 0.1352171003818512, "global_step": 49316, "epoch": 465, "lr": 1.1867682101145749e-06} +{"train_loss": 0.016095589846372604, "grad_norm": 0.09954731911420822, "global_step": 49317, "epoch": 465, "lr": 1.186126400109061e-06} +{"train_loss": 0.014485200867056847, "grad_norm": 0.08423389494419098, "global_step": 49318, "epoch": 465, "lr": 1.1854847616142694e-06} +{"train_loss": 0.01364317536354065, "grad_norm": 0.09993594884872437, "global_step": 49319, "epoch": 465, "lr": 1.1848432946324594e-06} +{"train_loss": 0.013556997291743755, "grad_norm": 0.07216636091470718, "global_step": 49320, "epoch": 465, "lr": 1.1842019991658738e-06} +{"train_loss": 0.014016096480190754, "grad_norm": 0.08489742130041122, "global_step": 49321, "epoch": 465, "lr": 1.183560875216777e-06} +{"train_loss": 0.014945317059755325, "grad_norm": 0.0737343579530716, "global_step": 49322, "epoch": 465, "lr": 1.1829199227874177e-06} +{"train_loss": 0.016312388703227043, "grad_norm": 0.06685029715299606, "global_step": 49323, "epoch": 465, "lr": 1.1822791418800439e-06} +{"train_loss": 0.015615992248058319, "grad_norm": 0.06709855049848557, "global_step": 49324, "epoch": 465, "lr": 1.181638532496915e-06} +{"train_loss": 0.013477951288223267, "grad_norm": 0.06380662322044373, "global_step": 49325, "epoch": 465, "lr": 1.1809980946402788e-06} +{"train_loss": 0.014168514870107174, "grad_norm": 0.06508681178092957, "global_step": 49326, "epoch": 465, "lr": 1.1803578283123784e-06} +{"train_loss": 0.014626680873334408, "grad_norm": 0.08358920365571976, "global_step": 49327, "epoch": 465, "lr": 1.179717733515473e-06} +{"train_loss": 0.01570136472582817, "grad_norm": 0.06781038641929626, "global_step": 49328, "epoch": 465, "lr": 1.1790778102518107e-06} +{"train_loss": 0.01590019091963768, "grad_norm": 0.08432246744632721, "global_step": 49329, "epoch": 465, "lr": 1.1784380585236343e-06} +{"train_loss": 0.01585787534713745, "grad_norm": 0.07749104499816895, "global_step": 49330, "epoch": 465, "lr": 1.1777984783331918e-06} +{"train_loss": 0.014060964807868004, "grad_norm": 0.12950973212718964, "global_step": 49331, "epoch": 465, "lr": 1.1771590696827429e-06} +{"train_loss": 0.01936301402747631, "grad_norm": 0.09201262146234512, "global_step": 49332, "epoch": 465, "lr": 1.1765198325745185e-06} +{"train_loss": 0.01667868345975876, "grad_norm": 0.16200782358646393, "global_step": 49333, "epoch": 465, "lr": 1.1758807670107675e-06} +{"train_loss": 0.013932009227573872, "grad_norm": 0.05676252022385597, "global_step": 49334, "epoch": 465, "lr": 1.1752418729937431e-06} +{"train_loss": 0.01523355022072792, "grad_norm": 0.12060783803462982, "global_step": 49335, "epoch": 465, "lr": 1.1746031505256828e-06} +{"train_loss": 0.016353081911802292, "grad_norm": 0.06815466284751892, "global_step": 49336, "epoch": 465, "lr": 1.1739645996088288e-06} +{"train_loss": 0.013634510338306427, "grad_norm": 0.08217767626047134, "global_step": 49337, "epoch": 465, "lr": 1.173326220245441e-06} +{"train_loss": 0.016845228150486946, "grad_norm": 0.06635738909244537, "global_step": 49338, "epoch": 465, "lr": 1.1726880124377449e-06} +{"train_loss": 0.015742935240268707, "grad_norm": 0.08985072374343872, "global_step": 49339, "epoch": 465, "lr": 1.1720499761879888e-06} +{"train_loss": 0.014546437188982964, "grad_norm": 0.09383026510477066, "global_step": 49340, "epoch": 465, "lr": 1.17141211149841e-06} +{"train_loss": 0.013045024126768112, "grad_norm": 0.08344987779855728, "global_step": 49341, "epoch": 465, "lr": 1.1707744183712622e-06} +{"train_loss": 0.015177140012383461, "grad_norm": 0.08711378276348114, "global_step": 49342, "epoch": 465, "lr": 1.1701368968087712e-06} +{"train_loss": 0.01485222578048706, "grad_norm": 0.08277860283851624, "global_step": 49343, "epoch": 465, "lr": 1.1694995468131853e-06} +{"train_loss": 0.014902370050549507, "grad_norm": 0.056658972054719925, "global_step": 49344, "epoch": 465, "lr": 1.1688623683867417e-06} +{"train_loss": 0.01319289579987526, "grad_norm": 0.08128198981285095, "global_step": 49345, "epoch": 465, "lr": 1.1682253615316829e-06} +{"train_loss": 0.01487940177321434, "grad_norm": 0.08833496272563934, "global_step": 49346, "epoch": 465, "lr": 1.1675885262502406e-06} +{"train_loss": 0.014812327921390533, "grad_norm": 0.06163116544485092, "global_step": 49347, "epoch": 465, "lr": 1.1669518625446573e-06} +{"train_loss": 0.014287514612078667, "grad_norm": 0.08088533580303192, "global_step": 49348, "epoch": 465, "lr": 1.1663153704171703e-06} +{"train_loss": 0.014677909202873707, "grad_norm": 0.10846976190805435, "global_step": 49349, "epoch": 465, "lr": 1.1656790498700109e-06} +{"train_loss": 0.012878168374300003, "grad_norm": 0.05672064051032066, "global_step": 49350, "epoch": 465, "lr": 1.1650429009054164e-06} +{"train_loss": 0.01506551168859005, "grad_norm": 0.10114243626594543, "global_step": 49351, "epoch": 465, "lr": 1.1644069235256295e-06} +{"train_loss": 0.013799389824271202, "grad_norm": 0.08157467097043991, "global_step": 49352, "epoch": 465, "lr": 1.1637711177328702e-06} +{"train_loss": 0.01574447937309742, "grad_norm": 0.05897099897265434, "global_step": 49353, "epoch": 465, "lr": 1.1631354835293818e-06} +{"train_loss": 0.015014266595244408, "grad_norm": 0.07699766755104065, "global_step": 49354, "epoch": 465, "lr": 1.1625000209174063e-06} +{"train_loss": 0.01544807106256485, "grad_norm": 0.0871506929397583, "global_step": 49355, "epoch": 465, "lr": 1.1618647298991593e-06} +{"train_loss": 0.01580580696463585, "grad_norm": 0.0864333063364029, "global_step": 49356, "epoch": 465, "lr": 1.1612296104768828e-06} +{"train_loss": 0.01554534025490284, "grad_norm": 0.09806864708662033, "global_step": 49357, "epoch": 465, "lr": 1.160594662652803e-06} +{"train_loss": 0.013175014406442642, "grad_norm": 0.0893353745341301, "global_step": 49358, "epoch": 465, "lr": 1.1599598864291572e-06} +{"train_loss": 0.01746935397386551, "grad_norm": 0.09387384355068207, "global_step": 49359, "epoch": 465, "lr": 1.1593252818081658e-06} +{"train_loss": 0.014291808009147644, "grad_norm": 0.05897262319922447, "global_step": 49360, "epoch": 465, "lr": 1.1586908487920767e-06} +{"train_loss": 0.019113097339868546, "grad_norm": 0.07981109619140625, "global_step": 49361, "epoch": 465, "lr": 1.1580565873830996e-06} +{"train_loss": 0.01340552233159542, "grad_norm": 0.052202027291059494, "global_step": 49362, "epoch": 465, "lr": 1.1574224975834713e-06} +{"train_loss": 0.01714501529932022, "grad_norm": 0.06793483346700668, "global_step": 49363, "epoch": 465, "lr": 1.1567885793954181e-06} +{"train_loss": 0.014270618557929993, "grad_norm": 0.09090438485145569, "global_step": 49364, "epoch": 465, "lr": 1.1561548328211712e-06} +{"train_loss": 0.014451025053858757, "grad_norm": 0.08919298648834229, "global_step": 49365, "epoch": 465, "lr": 1.155521257862957e-06} +{"train_loss": 0.01590000092983246, "grad_norm": 0.0818326398730278, "global_step": 49366, "epoch": 465, "lr": 1.1548878545229958e-06} +{"train_loss": 0.016372939571738243, "grad_norm": 0.08647626638412476, "global_step": 49367, "epoch": 465, "lr": 1.1542546228035133e-06} +{"train_loss": 0.0130319157615304, "grad_norm": 0.1058950275182724, "global_step": 49368, "epoch": 465, "lr": 1.153621562706747e-06} +{"train_loss": 0.016896110028028488, "grad_norm": 0.06184815615415573, "global_step": 49369, "epoch": 465, "lr": 1.1529886742349005e-06} +{"train_loss": 0.013541042804718018, "grad_norm": 0.08665908128023148, "global_step": 49370, "epoch": 465, "lr": 1.1523559573902164e-06} +{"train_loss": 0.014723588712513447, "grad_norm": 0.10085900127887726, "global_step": 49371, "epoch": 465, "lr": 1.15172341217491e-06} +{"train_loss": 0.016185732558369637, "grad_norm": 0.13329213857650757, "global_step": 49372, "epoch": 465, "lr": 1.151091038591201e-06} +{"train_loss": 0.013565835542976856, "grad_norm": 0.09132545441389084, "global_step": 49373, "epoch": 465, "lr": 1.150458836641316e-06} +{"train_loss": 0.01626431941986084, "grad_norm": 0.07644559442996979, "global_step": 49374, "epoch": 465, "lr": 1.1498268063274697e-06} +{"train_loss": 0.013144943863153458, "grad_norm": 0.0859953761100769, "global_step": 49375, "epoch": 465, "lr": 1.1491949476518936e-06} +{"train_loss": 0.013793127611279488, "grad_norm": 0.13243599236011505, "global_step": 49376, "epoch": 465, "lr": 1.1485632606167974e-06} +{"train_loss": 0.012532256543636322, "grad_norm": 0.06183580681681633, "global_step": 49377, "epoch": 465, "lr": 1.1479317452244065e-06} +{"train_loss": 0.014281743206083775, "grad_norm": 0.07279030233621597, "global_step": 49378, "epoch": 465, "lr": 1.1473004014769417e-06} +{"train_loss": 0.014311324805021286, "grad_norm": 0.05443646013736725, "global_step": 49379, "epoch": 465, "lr": 1.146669229376618e-06} +{"train_loss": 0.016100311651825905, "grad_norm": 0.09752289950847626, "global_step": 49380, "epoch": 465, "lr": 1.14603822892565e-06} +{"train_loss": 0.014502888545393944, "grad_norm": 0.053735390305519104, "global_step": 49381, "epoch": 465, "lr": 1.1454074001262526e-06} +{"train_loss": 0.01385523658245802, "grad_norm": 0.054760389029979706, "global_step": 49382, "epoch": 465, "lr": 1.1447767429806576e-06} +{"train_loss": 0.0152380196377635, "grad_norm": 0.07765273004770279, "global_step": 49383, "epoch": 465, "lr": 1.1441462574910633e-06} +{"train_loss": 0.015592886134982109, "grad_norm": 0.058483947068452835, "global_step": 49384, "epoch": 465, "lr": 1.1435159436596954e-06} +{"train_loss": 0.013518116436898708, "grad_norm": 0.09474548697471619, "global_step": 49385, "epoch": 465, "lr": 1.1428858014887634e-06} +{"train_loss": 0.013907684944570065, "grad_norm": 0.11688839644193649, "global_step": 49386, "epoch": 465, "lr": 1.1422558309804876e-06} +{"train_loss": 0.020140090957283974, "grad_norm": 0.09379953145980835, "global_step": 49387, "epoch": 465, "lr": 1.141626032137072e-06} +{"train_loss": 0.014630195684731007, "grad_norm": 0.05551102012395859, "global_step": 49388, "epoch": 465, "lr": 1.1409964049607313e-06} +{"train_loss": 0.01682020165026188, "grad_norm": 0.13319461047649384, "global_step": 49389, "epoch": 465, "lr": 1.1403669494536918e-06} +{"train_loss": 0.014539692550897598, "grad_norm": 0.06568165868520737, "global_step": 49390, "epoch": 465, "lr": 1.1397376656181513e-06} +{"train_loss": 0.01599152386188507, "grad_norm": 0.08966643363237381, "global_step": 49391, "epoch": 465, "lr": 1.1391085534563196e-06} +{"train_loss": 0.012896731495857239, "grad_norm": 0.0881802961230278, "global_step": 49392, "epoch": 465, "lr": 1.1384796129704113e-06} +{"train_loss": 0.012943631038069725, "grad_norm": 0.058450110256671906, "global_step": 49393, "epoch": 465, "lr": 1.1378508441626413e-06} +{"train_loss": 0.014791546389460564, "grad_norm": 0.09252557903528214, "global_step": 49394, "epoch": 465, "lr": 1.1372222470352078e-06} +{"train_loss": 0.014947609580561237, "grad_norm": 0.12334973365068436, "global_step": 49395, "epoch": 465, "lr": 1.136593821590326e-06, "val_loss": 0.018988411873579025} +{"train_loss": 0.01553382445126772, "grad_norm": 0.1063469797372818, "global_step": 49396, "epoch": 466, "lr": 1.1359655678302106e-06} +{"train_loss": 0.014550598338246346, "grad_norm": 0.07723850011825562, "global_step": 49397, "epoch": 466, "lr": 1.1353374857570542e-06} +{"train_loss": 0.01559794694185257, "grad_norm": 0.08273579180240631, "global_step": 49398, "epoch": 466, "lr": 1.1347095753730719e-06} +{"train_loss": 0.014412629418075085, "grad_norm": 0.0863809883594513, "global_step": 49399, "epoch": 466, "lr": 1.1340818366804729e-06} +{"train_loss": 0.01568765565752983, "grad_norm": 0.061873190104961395, "global_step": 49400, "epoch": 466, "lr": 1.1334542696814554e-06} +{"train_loss": 0.014605779200792313, "grad_norm": 0.05130110681056976, "global_step": 49401, "epoch": 466, "lr": 1.1328268743782289e-06} +{"train_loss": 0.015098435804247856, "grad_norm": 0.07565411180257797, "global_step": 49402, "epoch": 466, "lr": 1.1321996507730026e-06} +{"train_loss": 0.015066850930452347, "grad_norm": 0.08908192068338394, "global_step": 49403, "epoch": 466, "lr": 1.1315725988679693e-06} +{"train_loss": 0.01637088507413864, "grad_norm": 0.09238120168447495, "global_step": 49404, "epoch": 466, "lr": 1.1309457186653383e-06} +{"train_loss": 0.01715061627328396, "grad_norm": 0.14366945624351501, "global_step": 49405, "epoch": 466, "lr": 1.1303190101673078e-06} +{"train_loss": 0.012800473719835281, "grad_norm": 0.07872457802295685, "global_step": 49406, "epoch": 466, "lr": 1.1296924733760927e-06} +{"train_loss": 0.013997985050082207, "grad_norm": 0.0734172984957695, "global_step": 49407, "epoch": 466, "lr": 1.1290661082938747e-06} +{"train_loss": 0.016035549342632294, "grad_norm": 0.06017821654677391, "global_step": 49408, "epoch": 466, "lr": 1.1284399149228742e-06} +{"train_loss": 0.015240540727972984, "grad_norm": 0.07516037672758102, "global_step": 49409, "epoch": 466, "lr": 1.1278138932652781e-06} +{"train_loss": 0.012383341789245605, "grad_norm": 0.0738876610994339, "global_step": 49410, "epoch": 466, "lr": 1.1271880433232906e-06} +{"train_loss": 0.016115248203277588, "grad_norm": 0.0784858763217926, "global_step": 49411, "epoch": 466, "lr": 1.1265623650991042e-06} +{"train_loss": 0.012945802882313728, "grad_norm": 0.11592340469360352, "global_step": 49412, "epoch": 466, "lr": 1.1259368585949337e-06} +{"train_loss": 0.016912519931793213, "grad_norm": 0.05578264221549034, "global_step": 49413, "epoch": 466, "lr": 1.1253115238129609e-06} +{"train_loss": 0.015885110944509506, "grad_norm": 0.06617743521928787, "global_step": 49414, "epoch": 466, "lr": 1.1246863607553891e-06} +{"train_loss": 0.01306011900305748, "grad_norm": 0.07277915626764297, "global_step": 49415, "epoch": 466, "lr": 1.1240613694244117e-06} +{"train_loss": 0.015956314280629158, "grad_norm": 0.056422848254442215, "global_step": 49416, "epoch": 466, "lr": 1.123436549822232e-06} +{"train_loss": 0.016958685591816902, "grad_norm": 0.17077240347862244, "global_step": 49417, "epoch": 466, "lr": 1.122811901951032e-06} +{"train_loss": 0.015976818278431892, "grad_norm": 0.07290343195199966, "global_step": 49418, "epoch": 466, "lr": 1.1221874258130206e-06} +{"train_loss": 0.015036641620099545, "grad_norm": 0.06647312641143799, "global_step": 49419, "epoch": 466, "lr": 1.121563121410385e-06} +{"train_loss": 0.014729572460055351, "grad_norm": 0.1183648481965065, "global_step": 49420, "epoch": 466, "lr": 1.1209389887453237e-06} +{"train_loss": 0.01612689718604088, "grad_norm": 0.06964336335659027, "global_step": 49421, "epoch": 466, "lr": 1.1203150278200237e-06} +{"train_loss": 0.019062465056777, "grad_norm": 0.06376046687364578, "global_step": 49422, "epoch": 466, "lr": 1.119691238636672e-06} +{"train_loss": 0.017844507470726967, "grad_norm": 0.08714242279529572, "global_step": 49423, "epoch": 466, "lr": 1.1190676211974783e-06} +{"train_loss": 0.01554795354604721, "grad_norm": 0.06308849900960922, "global_step": 49424, "epoch": 466, "lr": 1.1184441755046182e-06} +{"train_loss": 0.015251914970576763, "grad_norm": 0.07770335674285889, "global_step": 49425, "epoch": 466, "lr": 1.1178209015602903e-06} +{"train_loss": 0.014603139832615852, "grad_norm": 0.10445267707109451, "global_step": 49426, "epoch": 466, "lr": 1.1171977993666817e-06} +{"train_loss": 0.014393007382750511, "grad_norm": 0.06430978327989578, "global_step": 49427, "epoch": 466, "lr": 1.116574868925979e-06} +{"train_loss": 0.015512166544795036, "grad_norm": 0.0563470833003521, "global_step": 49428, "epoch": 466, "lr": 1.1159521102403703e-06} +{"train_loss": 0.013860993087291718, "grad_norm": 0.06325837224721909, "global_step": 49429, "epoch": 466, "lr": 1.1153295233120476e-06} +{"train_loss": 0.01564495638012886, "grad_norm": 0.09699850529432297, "global_step": 49430, "epoch": 466, "lr": 1.1147071081432037e-06} +{"train_loss": 0.014564858749508858, "grad_norm": 0.06043057516217232, "global_step": 49431, "epoch": 466, "lr": 1.1140848647360147e-06} +{"train_loss": 0.014695310033857822, "grad_norm": 0.0761689692735672, "global_step": 49432, "epoch": 466, "lr": 1.1134627930926733e-06} +{"train_loss": 0.015084002166986465, "grad_norm": 0.16071657836437225, "global_step": 49433, "epoch": 466, "lr": 1.112840893215361e-06} +{"train_loss": 0.015254450961947441, "grad_norm": 0.14450152218341827, "global_step": 49434, "epoch": 466, "lr": 1.1122191651062652e-06} +{"train_loss": 0.01491322461515665, "grad_norm": 0.06635981053113937, "global_step": 49435, "epoch": 466, "lr": 1.1115976087675729e-06} +{"train_loss": 0.017057020217180252, "grad_norm": 0.06273176521062851, "global_step": 49436, "epoch": 466, "lr": 1.1109762242014655e-06} +{"train_loss": 0.016068294644355774, "grad_norm": 0.10071752965450287, "global_step": 49437, "epoch": 466, "lr": 1.1103550114101301e-06} +{"train_loss": 0.01687171868979931, "grad_norm": 0.11175606399774551, "global_step": 49438, "epoch": 466, "lr": 1.1097339703957377e-06} +{"train_loss": 0.014869455248117447, "grad_norm": 0.09272640943527222, "global_step": 49439, "epoch": 466, "lr": 1.1091131011604804e-06} +{"train_loss": 0.015031052753329277, "grad_norm": 0.05564730241894722, "global_step": 49440, "epoch": 466, "lr": 1.1084924037065404e-06} +{"train_loss": 0.013683492317795753, "grad_norm": 0.06360288709402084, "global_step": 49441, "epoch": 466, "lr": 1.107871878036093e-06} +{"train_loss": 0.012261973693966866, "grad_norm": 0.0931163877248764, "global_step": 49442, "epoch": 466, "lr": 1.1072515241513204e-06} +{"train_loss": 0.017447659745812416, "grad_norm": 0.0759897381067276, "global_step": 49443, "epoch": 466, "lr": 1.1066313420544095e-06} +{"train_loss": 0.012855641543865204, "grad_norm": 0.11534924060106277, "global_step": 49444, "epoch": 466, "lr": 1.1060113317475307e-06} +{"train_loss": 0.011558892205357552, "grad_norm": 0.06756435334682465, "global_step": 49445, "epoch": 466, "lr": 1.1053914932328601e-06} +{"train_loss": 0.012777438387274742, "grad_norm": 0.08532710373401642, "global_step": 49446, "epoch": 466, "lr": 1.1047718265125795e-06} +{"train_loss": 0.013072446919977665, "grad_norm": 0.05794687941670418, "global_step": 49447, "epoch": 466, "lr": 1.1041523315888759e-06} +{"train_loss": 0.01306040771305561, "grad_norm": 0.08907339721918106, "global_step": 49448, "epoch": 466, "lr": 1.1035330084639083e-06} +{"train_loss": 0.015668291598558426, "grad_norm": 0.05660703778266907, "global_step": 49449, "epoch": 466, "lr": 1.1029138571398644e-06} +{"train_loss": 0.013331936672329903, "grad_norm": 0.07318912446498871, "global_step": 49450, "epoch": 466, "lr": 1.1022948776189202e-06} +{"train_loss": 0.015084478072822094, "grad_norm": 0.06518737971782684, "global_step": 49451, "epoch": 466, "lr": 1.1016760699032402e-06} +{"train_loss": 0.01705262064933777, "grad_norm": 0.10982117056846619, "global_step": 49452, "epoch": 466, "lr": 1.1010574339950064e-06} +{"train_loss": 0.013330994173884392, "grad_norm": 0.1075470894575119, "global_step": 49453, "epoch": 466, "lr": 1.1004389698963945e-06} +{"train_loss": 0.015590480528771877, "grad_norm": 0.0917755737900734, "global_step": 49454, "epoch": 466, "lr": 1.0998206776095698e-06} +{"train_loss": 0.016022568568587303, "grad_norm": 0.07278236001729965, "global_step": 49455, "epoch": 466, "lr": 1.0992025571367137e-06} +{"train_loss": 0.014890729449689388, "grad_norm": 0.10017592459917068, "global_step": 49456, "epoch": 466, "lr": 1.0985846084799856e-06} +{"train_loss": 0.01580212265253067, "grad_norm": 0.062127359211444855, "global_step": 49457, "epoch": 466, "lr": 1.0979668316415726e-06} +{"train_loss": 0.0154498852789402, "grad_norm": 0.07313887774944305, "global_step": 49458, "epoch": 466, "lr": 1.097349226623634e-06} +{"train_loss": 0.015949828550219536, "grad_norm": 0.07609620690345764, "global_step": 49459, "epoch": 466, "lr": 1.0967317934283406e-06} +{"train_loss": 0.013795310631394386, "grad_norm": 0.06883877515792847, "global_step": 49460, "epoch": 466, "lr": 1.096114532057868e-06} +{"train_loss": 0.015778688713908195, "grad_norm": 0.08331038057804108, "global_step": 49461, "epoch": 466, "lr": 1.0954974425143815e-06} +{"train_loss": 0.015872158110141754, "grad_norm": 0.05736352875828743, "global_step": 49462, "epoch": 466, "lr": 1.094880524800046e-06} +{"train_loss": 0.013377764262259007, "grad_norm": 0.08209863305091858, "global_step": 49463, "epoch": 466, "lr": 1.0942637789170317e-06} +{"train_loss": 0.01548054814338684, "grad_norm": 0.07585366070270538, "global_step": 49464, "epoch": 466, "lr": 1.0936472048675095e-06} +{"train_loss": 0.01664520427584648, "grad_norm": 0.07576063275337219, "global_step": 49465, "epoch": 466, "lr": 1.093030802653644e-06} +{"train_loss": 0.014767961576581001, "grad_norm": 0.10917215049266815, "global_step": 49466, "epoch": 466, "lr": 1.0924145722775948e-06} +{"train_loss": 0.013560221530497074, "grad_norm": 0.07398573309183121, "global_step": 49467, "epoch": 466, "lr": 1.0917985137415376e-06} +{"train_loss": 0.014399388805031776, "grad_norm": 0.16935908794403076, "global_step": 49468, "epoch": 466, "lr": 1.091182627047621e-06} +{"train_loss": 0.01458526961505413, "grad_norm": 0.12033507227897644, "global_step": 49469, "epoch": 466, "lr": 1.090566912198021e-06} +{"train_loss": 0.016667788848280907, "grad_norm": 0.12044547498226166, "global_step": 49470, "epoch": 466, "lr": 1.0899513691949025e-06} +{"train_loss": 0.014722009189426899, "grad_norm": 0.1286589503288269, "global_step": 49471, "epoch": 466, "lr": 1.0893359980404249e-06} +{"train_loss": 0.017372865229845047, "grad_norm": 0.12022432684898376, "global_step": 49472, "epoch": 466, "lr": 1.0887207987367477e-06} +{"train_loss": 0.013613182120025158, "grad_norm": 0.12730175256729126, "global_step": 49473, "epoch": 466, "lr": 1.0881057712860355e-06} +{"train_loss": 0.013663874939084053, "grad_norm": 0.07592974603176117, "global_step": 49474, "epoch": 466, "lr": 1.0874909156904478e-06} +{"train_loss": 0.014228727668523788, "grad_norm": 0.05822291597723961, "global_step": 49475, "epoch": 466, "lr": 1.0868762319521442e-06} +{"train_loss": 0.017867907881736755, "grad_norm": 0.07140613347291946, "global_step": 49476, "epoch": 466, "lr": 1.086261720073284e-06} +{"train_loss": 0.01799847185611725, "grad_norm": 0.08147644251585007, "global_step": 49477, "epoch": 466, "lr": 1.0856473800560373e-06} +{"train_loss": 0.016994573175907135, "grad_norm": 0.11744705587625504, "global_step": 49478, "epoch": 466, "lr": 1.0850332119025476e-06} +{"train_loss": 0.015430310741066933, "grad_norm": 0.09092696011066437, "global_step": 49479, "epoch": 466, "lr": 1.084419215614979e-06} +{"train_loss": 0.01541637908667326, "grad_norm": 0.09218870848417282, "global_step": 49480, "epoch": 466, "lr": 1.083805391195486e-06} +{"train_loss": 0.013528440147638321, "grad_norm": 0.12142607569694519, "global_step": 49481, "epoch": 466, "lr": 1.0831917386462332e-06} +{"train_loss": 0.014465096406638622, "grad_norm": 0.07953143119812012, "global_step": 49482, "epoch": 466, "lr": 1.0825782579693634e-06} +{"train_loss": 0.012403301894664764, "grad_norm": 0.07289914041757584, "global_step": 49483, "epoch": 466, "lr": 1.0819649491670414e-06} +{"train_loss": 0.012075390666723251, "grad_norm": 0.05048659071326256, "global_step": 49484, "epoch": 466, "lr": 1.0813518122414269e-06} +{"train_loss": 0.011306033469736576, "grad_norm": 0.05485847592353821, "global_step": 49485, "epoch": 466, "lr": 1.0807388471946677e-06} +{"train_loss": 0.014407893642783165, "grad_norm": 0.07139062881469727, "global_step": 49486, "epoch": 466, "lr": 1.0801260540289127e-06} +{"train_loss": 0.014521731063723564, "grad_norm": 0.07995619624853134, "global_step": 49487, "epoch": 466, "lr": 1.0795134327463207e-06} +{"train_loss": 0.016006477177143097, "grad_norm": 0.07735249400138855, "global_step": 49488, "epoch": 466, "lr": 1.078900983349046e-06} +{"train_loss": 0.016083048656582832, "grad_norm": 0.11615019291639328, "global_step": 49489, "epoch": 466, "lr": 1.0782887058392366e-06} +{"train_loss": 0.013956969603896141, "grad_norm": 0.07003884017467499, "global_step": 49490, "epoch": 466, "lr": 1.0776766002190464e-06} +{"train_loss": 0.013156896457076073, "grad_norm": 0.06804177910089493, "global_step": 49491, "epoch": 466, "lr": 1.0770646664906292e-06} +{"train_loss": 0.015148301608860493, "grad_norm": 0.06948574632406235, "global_step": 49492, "epoch": 466, "lr": 1.0764529046561223e-06} +{"train_loss": 0.015620551072061062, "grad_norm": 0.06853806972503662, "global_step": 49493, "epoch": 466, "lr": 1.075841314717685e-06} +{"train_loss": 0.016336558386683464, "grad_norm": 0.10940301418304443, "global_step": 49494, "epoch": 466, "lr": 1.075229896677471e-06} +{"train_loss": 0.014433354139328003, "grad_norm": 0.0659908652305603, "global_step": 49495, "epoch": 466, "lr": 1.0746186505376122e-06} +{"train_loss": 0.014278542250394821, "grad_norm": 0.09006480127573013, "global_step": 49496, "epoch": 466, "lr": 1.0740075763002789e-06} +{"train_loss": 0.01289147138595581, "grad_norm": 0.08228164911270142, "global_step": 49497, "epoch": 466, "lr": 1.0733966739675971e-06} +{"train_loss": 0.014957059174776077, "grad_norm": 0.09716228395700455, "global_step": 49498, "epoch": 466, "lr": 1.0727859435417266e-06} +{"train_loss": 0.011819199658930302, "grad_norm": 0.09893640130758286, "global_step": 49499, "epoch": 466, "lr": 1.0721753850247984e-06} +{"train_loss": 0.015384847298264503, "grad_norm": 0.11609257757663727, "global_step": 49500, "epoch": 466, "lr": 1.0715649984189725e-06} +{"train_loss": 0.014949253224328442, "grad_norm": 0.10094142705202103, "global_step": 49501, "epoch": 466, "lr": 1.0709547837263966e-06, "val_loss": 0.019769173115491867} +{"train_loss": 0.01335641648620367, "grad_norm": 0.07666239887475967, "global_step": 49502, "epoch": 467, "lr": 1.070344740949203e-06} +{"train_loss": 0.014825298450887203, "grad_norm": 0.09268714487552643, "global_step": 49503, "epoch": 467, "lr": 1.0697348700895337e-06} +{"train_loss": 0.014606676995754242, "grad_norm": 0.06996721774339676, "global_step": 49504, "epoch": 467, "lr": 1.0691251711495432e-06} +{"train_loss": 0.017041398212313652, "grad_norm": 0.13116393983364105, "global_step": 49505, "epoch": 467, "lr": 1.0685156441313625e-06} +{"train_loss": 0.015112163498997688, "grad_norm": 0.06087139993906021, "global_step": 49506, "epoch": 467, "lr": 1.0679062890371405e-06} +{"train_loss": 0.015951523557305336, "grad_norm": 0.11328354477882385, "global_step": 49507, "epoch": 467, "lr": 1.067297105869014e-06} +{"train_loss": 0.01569504663348198, "grad_norm": 0.08982744812965393, "global_step": 49508, "epoch": 467, "lr": 1.0666880946291313e-06} +{"train_loss": 0.014552569016814232, "grad_norm": 0.06954112648963928, "global_step": 49509, "epoch": 467, "lr": 1.0660792553196241e-06} +{"train_loss": 0.013642212375998497, "grad_norm": 0.13635577261447906, "global_step": 49510, "epoch": 467, "lr": 1.0654705879426241e-06} +{"train_loss": 0.014988236129283905, "grad_norm": 0.11088774353265762, "global_step": 49511, "epoch": 467, "lr": 1.0648620925002906e-06} +{"train_loss": 0.013833864592015743, "grad_norm": 0.10810624808073044, "global_step": 49512, "epoch": 467, "lr": 1.064253768994744e-06} +{"train_loss": 0.013168979436159134, "grad_norm": 0.12691093981266022, "global_step": 49513, "epoch": 467, "lr": 1.0636456174281329e-06} +{"train_loss": 0.013292471878230572, "grad_norm": 0.08924084901809692, "global_step": 49514, "epoch": 467, "lr": 1.063037637802583e-06} +{"train_loss": 0.01473784539848566, "grad_norm": 0.05759739130735397, "global_step": 49515, "epoch": 467, "lr": 1.0624298301202429e-06} +{"train_loss": 0.01214491855353117, "grad_norm": 0.0651763379573822, "global_step": 49516, "epoch": 467, "lr": 1.0618221943832386e-06} +{"train_loss": 0.01642705872654915, "grad_norm": 0.16340003907680511, "global_step": 49517, "epoch": 467, "lr": 1.061214730593707e-06} +{"train_loss": 0.014237936586141586, "grad_norm": 0.057882510125637054, "global_step": 49518, "epoch": 467, "lr": 1.0606074387537913e-06} +{"train_loss": 0.013730124570429325, "grad_norm": 0.16959765553474426, "global_step": 49519, "epoch": 467, "lr": 1.0600003188656117e-06} +{"train_loss": 0.017496002838015556, "grad_norm": 0.08175094425678253, "global_step": 49520, "epoch": 467, "lr": 1.0593933709313109e-06} +{"train_loss": 0.017507044598460197, "grad_norm": 0.0665745735168457, "global_step": 49521, "epoch": 467, "lr": 1.0587865949530097e-06} +{"train_loss": 0.014908288605511189, "grad_norm": 0.07464560121297836, "global_step": 49522, "epoch": 467, "lr": 1.0581799909328617e-06} +{"train_loss": 0.013384539633989334, "grad_norm": 0.09461983293294907, "global_step": 49523, "epoch": 467, "lr": 1.0575735588729706e-06} +{"train_loss": 0.017733797430992126, "grad_norm": 0.08147411048412323, "global_step": 49524, "epoch": 467, "lr": 1.0569672987754908e-06} +{"train_loss": 0.016010699793696404, "grad_norm": 0.091374471783638, "global_step": 49525, "epoch": 467, "lr": 1.056361210642548e-06} +{"train_loss": 0.013603148981928825, "grad_norm": 0.10163417458534241, "global_step": 49526, "epoch": 467, "lr": 1.0557552944762571e-06} +{"train_loss": 0.013830358162522316, "grad_norm": 0.052049197256565094, "global_step": 49527, "epoch": 467, "lr": 1.0551495502787556e-06} +{"train_loss": 0.015678446739912033, "grad_norm": 0.0641520619392395, "global_step": 49528, "epoch": 467, "lr": 1.0545439780521805e-06} +{"train_loss": 0.015510466881096363, "grad_norm": 0.08633919060230255, "global_step": 49529, "epoch": 467, "lr": 1.0539385777986467e-06} +{"train_loss": 0.016391010954976082, "grad_norm": 0.06903392821550369, "global_step": 49530, "epoch": 467, "lr": 1.0533333495202856e-06} +{"train_loss": 0.016440974548459053, "grad_norm": 0.09157794713973999, "global_step": 49531, "epoch": 467, "lr": 1.0527282932192294e-06} +{"train_loss": 0.01624239608645439, "grad_norm": 0.07406323403120041, "global_step": 49532, "epoch": 467, "lr": 1.052123408897604e-06} +{"train_loss": 0.01625930517911911, "grad_norm": 0.09841607511043549, "global_step": 49533, "epoch": 467, "lr": 1.0515186965575186e-06} +{"train_loss": 0.017688505351543427, "grad_norm": 0.08864807337522507, "global_step": 49534, "epoch": 467, "lr": 1.050914156201116e-06} +{"train_loss": 0.013449310325086117, "grad_norm": 0.061242397874593735, "global_step": 49535, "epoch": 467, "lr": 1.0503097878305113e-06} +{"train_loss": 0.01644756644964218, "grad_norm": 0.08825843036174774, "global_step": 49536, "epoch": 467, "lr": 1.0497055914478305e-06} +{"train_loss": 0.01613771729171276, "grad_norm": 0.10775906592607498, "global_step": 49537, "epoch": 467, "lr": 1.0491015670551996e-06} +{"train_loss": 0.015241567976772785, "grad_norm": 0.05095241218805313, "global_step": 49538, "epoch": 467, "lr": 1.0484977146547338e-06} +{"train_loss": 0.013916066847741604, "grad_norm": 0.08973010629415512, "global_step": 49539, "epoch": 467, "lr": 1.0478940342485588e-06} +{"train_loss": 0.015452138148248196, "grad_norm": 0.12419896572828293, "global_step": 49540, "epoch": 467, "lr": 1.0472905258387955e-06} +{"train_loss": 0.011849748902022839, "grad_norm": 0.058928776532411575, "global_step": 49541, "epoch": 467, "lr": 1.0466871894275643e-06} +{"train_loss": 0.014802908524870872, "grad_norm": 0.0781317800283432, "global_step": 49542, "epoch": 467, "lr": 1.0460840250169913e-06} +{"train_loss": 0.015022139064967632, "grad_norm": 0.061199694871902466, "global_step": 49543, "epoch": 467, "lr": 1.0454810326091857e-06} +{"train_loss": 0.012893805280327797, "grad_norm": 0.08313029259443283, "global_step": 49544, "epoch": 467, "lr": 1.0448782122062683e-06} +{"train_loss": 0.013653431087732315, "grad_norm": 0.08098890632390976, "global_step": 49545, "epoch": 467, "lr": 1.0442755638103542e-06} +{"train_loss": 0.016809487715363503, "grad_norm": 0.062087222933769226, "global_step": 49546, "epoch": 467, "lr": 1.0436730874235744e-06} +{"train_loss": 0.014736294746398926, "grad_norm": 0.10121379047632217, "global_step": 49547, "epoch": 467, "lr": 1.043070783048028e-06} +{"train_loss": 0.013162009418010712, "grad_norm": 0.14226026833057404, "global_step": 49548, "epoch": 467, "lr": 1.0424686506858461e-06} +{"train_loss": 0.01771194487810135, "grad_norm": 0.0835777223110199, "global_step": 49549, "epoch": 467, "lr": 1.041866690339144e-06} +{"train_loss": 0.013934407383203506, "grad_norm": 0.08051671832799911, "global_step": 49550, "epoch": 467, "lr": 1.0412649020100196e-06} +{"train_loss": 0.013337809592485428, "grad_norm": 0.059472691267728806, "global_step": 49551, "epoch": 467, "lr": 1.0406632857006048e-06} +{"train_loss": 0.014513333328068256, "grad_norm": 0.1006707176566124, "global_step": 49552, "epoch": 467, "lr": 1.0400618414130092e-06} +{"train_loss": 0.011877290904521942, "grad_norm": 0.11172445118427277, "global_step": 49553, "epoch": 467, "lr": 1.0394605691493365e-06} +{"train_loss": 0.01571001298725605, "grad_norm": 0.17544880509376526, "global_step": 49554, "epoch": 467, "lr": 1.0388594689117071e-06} +{"train_loss": 0.015100858174264431, "grad_norm": 0.09967531263828278, "global_step": 49555, "epoch": 467, "lr": 1.0382585407022416e-06} +{"train_loss": 0.016655363142490387, "grad_norm": 0.13404472172260284, "global_step": 49556, "epoch": 467, "lr": 1.0376577845230384e-06} +{"train_loss": 0.017873268574476242, "grad_norm": 0.1015017032623291, "global_step": 49557, "epoch": 467, "lr": 1.0370572003762124e-06} +{"train_loss": 0.015459692105650902, "grad_norm": 0.07983168959617615, "global_step": 49558, "epoch": 467, "lr": 1.036456788263873e-06} +{"train_loss": 0.0148500120267272, "grad_norm": 0.06327837705612183, "global_step": 49559, "epoch": 467, "lr": 1.0358565481881356e-06} +{"train_loss": 0.015751821920275688, "grad_norm": 0.07943231612443924, "global_step": 49560, "epoch": 467, "lr": 1.0352564801510977e-06} +{"train_loss": 0.01682785153388977, "grad_norm": 0.08210617303848267, "global_step": 49561, "epoch": 467, "lr": 1.0346565841548805e-06} +{"train_loss": 0.011730640195310116, "grad_norm": 0.062377844005823135, "global_step": 49562, "epoch": 467, "lr": 1.0340568602015765e-06} +{"train_loss": 0.014240055344998837, "grad_norm": 0.06253146380186081, "global_step": 49563, "epoch": 467, "lr": 1.033457308293312e-06} +{"train_loss": 0.01615634374320507, "grad_norm": 0.08612614870071411, "global_step": 49564, "epoch": 467, "lr": 1.0328579284321792e-06} +{"train_loss": 0.01605706289410591, "grad_norm": 0.07823187857866287, "global_step": 49565, "epoch": 467, "lr": 1.0322587206202827e-06} +{"train_loss": 0.015316586941480637, "grad_norm": 0.08311624825000763, "global_step": 49566, "epoch": 467, "lr": 1.0316596848597483e-06} +{"train_loss": 0.014395667240023613, "grad_norm": 0.07365016639232635, "global_step": 49567, "epoch": 467, "lr": 1.0310608211526574e-06} +{"train_loss": 0.014683688059449196, "grad_norm": 0.08165111392736435, "global_step": 49568, "epoch": 467, "lr": 1.0304621295011196e-06} +{"train_loss": 0.01719547063112259, "grad_norm": 0.09444279968738556, "global_step": 49569, "epoch": 467, "lr": 1.0298636099072446e-06} +{"train_loss": 0.015103106386959553, "grad_norm": 0.07821688801050186, "global_step": 49570, "epoch": 467, "lr": 1.0292652623731303e-06} +{"train_loss": 0.018122553825378418, "grad_norm": 0.10946804285049438, "global_step": 49571, "epoch": 467, "lr": 1.0286670869008808e-06} +{"train_loss": 0.01156747154891491, "grad_norm": 0.05217937007546425, "global_step": 49572, "epoch": 467, "lr": 1.0280690834926055e-06} +{"train_loss": 0.0131072998046875, "grad_norm": 0.07053449749946594, "global_step": 49573, "epoch": 467, "lr": 1.0274712521503916e-06} +{"train_loss": 0.015008094720542431, "grad_norm": 0.06883186846971512, "global_step": 49574, "epoch": 467, "lr": 1.0268735928763485e-06} +{"train_loss": 0.01516176201403141, "grad_norm": 0.07550261169672012, "global_step": 49575, "epoch": 467, "lr": 1.026276105672569e-06} +{"train_loss": 0.0160782802850008, "grad_norm": 0.07713280618190765, "global_step": 49576, "epoch": 467, "lr": 1.0256787905411625e-06} +{"train_loss": 0.012511195614933968, "grad_norm": 0.04520031437277794, "global_step": 49577, "epoch": 467, "lr": 1.0250816474842163e-06} +{"train_loss": 0.014243140816688538, "grad_norm": 0.05019115284085274, "global_step": 49578, "epoch": 467, "lr": 1.0244846765038453e-06} +{"train_loss": 0.015741635113954544, "grad_norm": 0.08385566622018814, "global_step": 49579, "epoch": 467, "lr": 1.0238878776021255e-06} +{"train_loss": 0.013949369080364704, "grad_norm": 0.0881149023771286, "global_step": 49580, "epoch": 467, "lr": 1.0232912507811721e-06} +{"train_loss": 0.011321283876895905, "grad_norm": 0.058437369763851166, "global_step": 49581, "epoch": 467, "lr": 1.0226947960430667e-06} +{"train_loss": 0.01296053733676672, "grad_norm": 0.05469517782330513, "global_step": 49582, "epoch": 467, "lr": 1.0220985133899131e-06} +{"train_loss": 0.016121666878461838, "grad_norm": 0.06131671369075775, "global_step": 49583, "epoch": 467, "lr": 1.0215024028238153e-06} +{"train_loss": 0.015197517350316048, "grad_norm": 0.09469963610172272, "global_step": 49584, "epoch": 467, "lr": 1.0209064643468492e-06} +{"train_loss": 0.013547730632126331, "grad_norm": 0.09023558348417282, "global_step": 49585, "epoch": 467, "lr": 1.0203106979611187e-06} +{"train_loss": 0.016325868666172028, "grad_norm": 0.07991055399179459, "global_step": 49586, "epoch": 467, "lr": 1.0197151036687113e-06} +{"train_loss": 0.015278782695531845, "grad_norm": 0.09108598530292511, "global_step": 49587, "epoch": 467, "lr": 1.0191196814717364e-06} +{"train_loss": 0.012664390727877617, "grad_norm": 0.07948184758424759, "global_step": 49588, "epoch": 467, "lr": 1.0185244313722586e-06} +{"train_loss": 0.013532254844903946, "grad_norm": 0.06846056878566742, "global_step": 49589, "epoch": 467, "lr": 1.0179293533723933e-06} +{"train_loss": 0.012817307375371456, "grad_norm": 0.056273218244314194, "global_step": 49590, "epoch": 467, "lr": 1.0173344474742274e-06} +{"train_loss": 0.015864789485931396, "grad_norm": 0.14075057208538055, "global_step": 49591, "epoch": 467, "lr": 1.0167397136798374e-06} +{"train_loss": 0.013678332790732384, "grad_norm": 0.05935431271791458, "global_step": 49592, "epoch": 467, "lr": 1.0161451519913213e-06} +{"train_loss": 0.018445570021867752, "grad_norm": 0.10067678242921829, "global_step": 49593, "epoch": 467, "lr": 1.0155507624107775e-06} +{"train_loss": 0.015299771912395954, "grad_norm": 0.07524832338094711, "global_step": 49594, "epoch": 467, "lr": 1.0149565449402764e-06} +{"train_loss": 0.015110870823264122, "grad_norm": 0.06927665323019028, "global_step": 49595, "epoch": 467, "lr": 1.0143624995819167e-06} +{"train_loss": 0.016245407983660698, "grad_norm": 0.05585695058107376, "global_step": 49596, "epoch": 467, "lr": 1.0137686263377856e-06} +{"train_loss": 0.012983856722712517, "grad_norm": 0.058398544788360596, "global_step": 49597, "epoch": 467, "lr": 1.0131749252099698e-06} +{"train_loss": 0.015864204615354538, "grad_norm": 0.10165011137723923, "global_step": 49598, "epoch": 467, "lr": 1.0125813962005515e-06} +{"train_loss": 0.015339668840169907, "grad_norm": 0.07804077118635178, "global_step": 49599, "epoch": 467, "lr": 1.0119880393116176e-06} +{"train_loss": 0.012997506186366081, "grad_norm": 0.06423011422157288, "global_step": 49600, "epoch": 467, "lr": 1.0113948545452556e-06} +{"train_loss": 0.01533284317702055, "grad_norm": 0.08489783108234406, "global_step": 49601, "epoch": 467, "lr": 1.0108018419035414e-06} +{"train_loss": 0.01729491353034973, "grad_norm": 0.1215229406952858, "global_step": 49602, "epoch": 467, "lr": 1.0102090013885734e-06} +{"train_loss": 0.017797376960515976, "grad_norm": 0.17123423516750336, "global_step": 49603, "epoch": 467, "lr": 1.0096163330024222e-06} +{"train_loss": 0.013552555814385414, "grad_norm": 0.11137021332979202, "global_step": 49604, "epoch": 467, "lr": 1.009023836747175e-06} +{"train_loss": 0.015435419976711273, "grad_norm": 0.13242819905281067, "global_step": 49605, "epoch": 467, "lr": 1.0084315126249078e-06} +{"train_loss": 0.014310328289866447, "grad_norm": 0.059481166303157806, "global_step": 49606, "epoch": 467, "lr": 1.007839360637708e-06} +{"train_loss": 0.01495545071040122, "grad_norm": 0.09992683678865433, "global_step": 49607, "epoch": 467, "lr": 1.007247380787657e-06, "val_loss": 0.01959524303674698} +{"train_loss": 0.014229970052838326, "grad_norm": 0.06664612889289856, "global_step": 49608, "epoch": 468, "lr": 1.0066555730768313e-06} +{"train_loss": 0.012735876254737377, "grad_norm": 0.06365299969911575, "global_step": 49609, "epoch": 468, "lr": 1.006063937507312e-06} +{"train_loss": 0.015657540410757065, "grad_norm": 0.06824447959661484, "global_step": 49610, "epoch": 468, "lr": 1.005472474081176e-06} +{"train_loss": 0.015503140166401863, "grad_norm": 0.062423333525657654, "global_step": 49611, "epoch": 468, "lr": 1.0048811828005044e-06} +{"train_loss": 0.014699658378958702, "grad_norm": 0.08987977355718613, "global_step": 49612, "epoch": 468, "lr": 1.004290063667368e-06} +{"train_loss": 0.016232524067163467, "grad_norm": 0.08275283128023148, "global_step": 49613, "epoch": 468, "lr": 1.003699116683854e-06} +{"train_loss": 0.013302478939294815, "grad_norm": 0.09272019565105438, "global_step": 49614, "epoch": 468, "lr": 1.0031083418520326e-06} +{"train_loss": 0.015852943062782288, "grad_norm": 0.10321086645126343, "global_step": 49615, "epoch": 468, "lr": 1.0025177391739804e-06} +{"train_loss": 0.015564652159810066, "grad_norm": 0.09409650415182114, "global_step": 49616, "epoch": 468, "lr": 1.0019273086517678e-06} +{"train_loss": 0.013891739770770073, "grad_norm": 0.06534399837255478, "global_step": 49617, "epoch": 468, "lr": 1.001337050287482e-06} +{"train_loss": 0.014496605843305588, "grad_norm": 0.07063458114862442, "global_step": 49618, "epoch": 468, "lr": 1.0007469640831824e-06} +{"train_loss": 0.015077054500579834, "grad_norm": 0.07730863988399506, "global_step": 49619, "epoch": 468, "lr": 1.0001570500409507e-06} +{"train_loss": 0.014827234670519829, "grad_norm": 0.09152673184871674, "global_step": 49620, "epoch": 468, "lr": 9.995673081628577e-07} +{"train_loss": 0.017662808299064636, "grad_norm": 0.05774286389350891, "global_step": 49621, "epoch": 468, "lr": 9.989777384509735e-07} +{"train_loss": 0.015114318579435349, "grad_norm": 0.0577840656042099, "global_step": 49622, "epoch": 468, "lr": 9.983883409073746e-07} +{"train_loss": 0.016782067716121674, "grad_norm": 0.07271374762058258, "global_step": 49623, "epoch": 468, "lr": 9.97799115534126e-07} +{"train_loss": 0.018595222383737564, "grad_norm": 0.08220839500427246, "global_step": 49624, "epoch": 468, "lr": 9.972100623333037e-07} +{"train_loss": 0.0153169184923172, "grad_norm": 0.10196080803871155, "global_step": 49625, "epoch": 468, "lr": 9.966211813069726e-07} +{"train_loss": 0.014189524576067924, "grad_norm": 0.13341015577316284, "global_step": 49626, "epoch": 468, "lr": 9.960324724572035e-07} +{"train_loss": 0.014072881080210209, "grad_norm": 0.08936743438243866, "global_step": 49627, "epoch": 468, "lr": 9.954439357860668e-07} +{"train_loss": 0.014814525842666626, "grad_norm": 0.09123101830482483, "global_step": 49628, "epoch": 468, "lr": 9.948555712956275e-07} +{"train_loss": 0.015301959589123726, "grad_norm": 0.07019948959350586, "global_step": 49629, "epoch": 468, "lr": 9.942673789879508e-07} +{"train_loss": 0.015151271596550941, "grad_norm": 0.16259540617465973, "global_step": 49630, "epoch": 468, "lr": 9.936793588651129e-07} +{"train_loss": 0.01597219705581665, "grad_norm": 0.06478775292634964, "global_step": 49631, "epoch": 468, "lr": 9.930915109291727e-07} +{"train_loss": 0.013554738834500313, "grad_norm": 0.0646706148982048, "global_step": 49632, "epoch": 468, "lr": 9.925038351821903e-07} +{"train_loss": 0.011888587847352028, "grad_norm": 0.05948428064584732, "global_step": 49633, "epoch": 468, "lr": 9.919163316262414e-07} +{"train_loss": 0.014889325946569443, "grad_norm": 0.1153964102268219, "global_step": 49634, "epoch": 468, "lr": 9.913290002633856e-07} +{"train_loss": 0.01263589970767498, "grad_norm": 0.05386854708194733, "global_step": 49635, "epoch": 468, "lr": 9.907418410956825e-07} +{"train_loss": 0.014078634791076183, "grad_norm": 0.10522972047328949, "global_step": 49636, "epoch": 468, "lr": 9.90154854125197e-07} +{"train_loss": 0.01456495188176632, "grad_norm": 0.09194067865610123, "global_step": 49637, "epoch": 468, "lr": 9.895680393540052e-07} +{"train_loss": 0.015330835245549679, "grad_norm": 0.0857221782207489, "global_step": 49638, "epoch": 468, "lr": 9.8898139678415e-07} +{"train_loss": 0.014734119176864624, "grad_norm": 0.07914615422487259, "global_step": 49639, "epoch": 468, "lr": 9.883949264176962e-07} +{"train_loss": 0.013660633005201817, "grad_norm": 0.15268975496292114, "global_step": 49640, "epoch": 468, "lr": 9.878086282567146e-07} +{"train_loss": 0.015736911445856094, "grad_norm": 0.10109451413154602, "global_step": 49641, "epoch": 468, "lr": 9.872225023032534e-07} +{"train_loss": 0.01621812768280506, "grad_norm": 0.10736332088708878, "global_step": 49642, "epoch": 468, "lr": 9.866365485593776e-07} +{"train_loss": 0.014403371140360832, "grad_norm": 0.10066567361354828, "global_step": 49643, "epoch": 468, "lr": 9.860507670271525e-07} +{"train_loss": 0.014746280387043953, "grad_norm": 0.1064314991235733, "global_step": 49644, "epoch": 468, "lr": 9.854651577086204e-07} +{"train_loss": 0.014810164459049702, "grad_norm": 0.05910177528858185, "global_step": 49645, "epoch": 468, "lr": 9.84879720605858e-07} +{"train_loss": 0.014753220602869987, "grad_norm": 0.14087150990962982, "global_step": 49646, "epoch": 468, "lr": 9.842944557209023e-07} +{"train_loss": 0.0168754942715168, "grad_norm": 0.09729301929473877, "global_step": 49647, "epoch": 468, "lr": 9.837093630558237e-07} +{"train_loss": 0.012154987081885338, "grad_norm": 0.10530103743076324, "global_step": 49648, "epoch": 468, "lr": 9.831244426126762e-07} +{"train_loss": 0.01547754555940628, "grad_norm": 0.0770038589835167, "global_step": 49649, "epoch": 468, "lr": 9.825396943935083e-07} +{"train_loss": 0.012564055621623993, "grad_norm": 0.07206065207719803, "global_step": 49650, "epoch": 468, "lr": 9.819551184003795e-07} +{"train_loss": 0.01564178243279457, "grad_norm": 0.07363925129175186, "global_step": 49651, "epoch": 468, "lr": 9.81370714635349e-07} +{"train_loss": 0.013433478772640228, "grad_norm": 0.049444541335105896, "global_step": 49652, "epoch": 468, "lr": 9.807864831004543e-07} +{"train_loss": 0.012655086815357208, "grad_norm": 0.0894361361861229, "global_step": 49653, "epoch": 468, "lr": 9.802024237977604e-07} +{"train_loss": 0.014362575486302376, "grad_norm": 0.08641990274190903, "global_step": 49654, "epoch": 468, "lr": 9.79618536729321e-07} +{"train_loss": 0.015505462884902954, "grad_norm": 0.08531983196735382, "global_step": 49655, "epoch": 468, "lr": 9.790348218971846e-07} +{"train_loss": 0.016373328864574432, "grad_norm": 0.060013506561517715, "global_step": 49656, "epoch": 468, "lr": 9.784512793033996e-07} +{"train_loss": 0.01530023105442524, "grad_norm": 0.07164940237998962, "global_step": 49657, "epoch": 468, "lr": 9.778679089500142e-07} +{"train_loss": 0.013387689366936684, "grad_norm": 0.0504976250231266, "global_step": 49658, "epoch": 468, "lr": 9.772847108390882e-07} +{"train_loss": 0.016804620623588562, "grad_norm": 0.0595720075070858, "global_step": 49659, "epoch": 468, "lr": 9.767016849726584e-07} +{"train_loss": 0.015087615698575974, "grad_norm": 0.10418324917554855, "global_step": 49660, "epoch": 468, "lr": 9.761188313527791e-07} +{"train_loss": 0.013219010084867477, "grad_norm": 0.06566503643989563, "global_step": 49661, "epoch": 468, "lr": 9.75536149981504e-07} +{"train_loss": 0.014945289120078087, "grad_norm": 0.06391647458076477, "global_step": 49662, "epoch": 468, "lr": 9.749536408608707e-07} +{"train_loss": 0.01554438378661871, "grad_norm": 0.06949478387832642, "global_step": 49663, "epoch": 468, "lr": 9.743713039929269e-07} +{"train_loss": 0.01562611013650894, "grad_norm": 0.09355196356773376, "global_step": 49664, "epoch": 468, "lr": 9.737891393797272e-07} +{"train_loss": 0.015126083046197891, "grad_norm": 0.09451484680175781, "global_step": 49665, "epoch": 468, "lr": 9.732071470233085e-07} +{"train_loss": 0.012793799862265587, "grad_norm": 0.10972615331411362, "global_step": 49666, "epoch": 468, "lr": 9.726253269257246e-07} +{"train_loss": 0.01732584834098816, "grad_norm": 0.06464795768260956, "global_step": 49667, "epoch": 468, "lr": 9.720436790890074e-07} +{"train_loss": 0.012302989140152931, "grad_norm": 0.09176602959632874, "global_step": 49668, "epoch": 468, "lr": 9.71462203515211e-07} +{"train_loss": 0.0169697105884552, "grad_norm": 0.1164856031537056, "global_step": 49669, "epoch": 468, "lr": 9.708809002063667e-07} +{"train_loss": 0.012709008529782295, "grad_norm": 0.05845685303211212, "global_step": 49670, "epoch": 468, "lr": 9.70299769164529e-07} +{"train_loss": 0.016135983169078827, "grad_norm": 0.07265305519104004, "global_step": 49671, "epoch": 468, "lr": 9.697188103917399e-07} +{"train_loss": 0.015783894807100296, "grad_norm": 0.05715438351035118, "global_step": 49672, "epoch": 468, "lr": 9.691380238900372e-07} +{"train_loss": 0.014612767845392227, "grad_norm": 0.06496311724185944, "global_step": 49673, "epoch": 468, "lr": 9.685574096614524e-07} +{"train_loss": 0.015286510810256004, "grad_norm": 0.11805533617734909, "global_step": 49674, "epoch": 468, "lr": 9.67976967708034e-07} +{"train_loss": 0.015657225623726845, "grad_norm": 0.07623103260993958, "global_step": 49675, "epoch": 468, "lr": 9.673966980318306e-07} +{"train_loss": 0.014407165348529816, "grad_norm": 0.07377801835536957, "global_step": 49676, "epoch": 468, "lr": 9.668166006348623e-07} +{"train_loss": 0.014735860750079155, "grad_norm": 0.08385331928730011, "global_step": 49677, "epoch": 468, "lr": 9.662366755191775e-07} +{"train_loss": 0.017299732193350792, "grad_norm": 0.10562307387590408, "global_step": 49678, "epoch": 468, "lr": 9.656569226868196e-07} +{"train_loss": 0.015035971999168396, "grad_norm": 0.1338415890932083, "global_step": 49679, "epoch": 468, "lr": 9.65077342139814e-07} +{"train_loss": 0.015262331813573837, "grad_norm": 0.07934629917144775, "global_step": 49680, "epoch": 468, "lr": 9.644979338801986e-07} +{"train_loss": 0.015036417171359062, "grad_norm": 0.10123718529939651, "global_step": 49681, "epoch": 468, "lr": 9.639186979100157e-07} +{"train_loss": 0.015257737599313259, "grad_norm": 0.0669073835015297, "global_step": 49682, "epoch": 468, "lr": 9.633396342313028e-07} +{"train_loss": 0.012266046367585659, "grad_norm": 0.06633476912975311, "global_step": 49683, "epoch": 468, "lr": 9.627607428460805e-07} +{"train_loss": 0.014171235263347626, "grad_norm": 0.09876669943332672, "global_step": 49684, "epoch": 468, "lr": 9.621820237563972e-07} +{"train_loss": 0.014294307678937912, "grad_norm": 0.06076132506132126, "global_step": 49685, "epoch": 468, "lr": 9.61603476964279e-07} +{"train_loss": 0.014897942543029785, "grad_norm": 0.06361697614192963, "global_step": 49686, "epoch": 468, "lr": 9.610251024717575e-07} +{"train_loss": 0.016730647534132004, "grad_norm": 0.16062262654304504, "global_step": 49687, "epoch": 468, "lr": 9.604469002808647e-07} +{"train_loss": 0.016454750671982765, "grad_norm": 0.12702757120132446, "global_step": 49688, "epoch": 468, "lr": 9.598688703936377e-07} +{"train_loss": 0.012901168316602707, "grad_norm": 0.08655231446027756, "global_step": 49689, "epoch": 468, "lr": 9.59291012812108e-07} +{"train_loss": 0.016093740239739418, "grad_norm": 0.09888356924057007, "global_step": 49690, "epoch": 468, "lr": 9.587133275382964e-07} +{"train_loss": 0.015385396778583527, "grad_norm": 0.12018182128667831, "global_step": 49691, "epoch": 468, "lr": 9.581358145742404e-07} +{"train_loss": 0.01720063015818596, "grad_norm": 0.1333954930305481, "global_step": 49692, "epoch": 468, "lr": 9.575584739219712e-07} +{"train_loss": 0.013984264805912971, "grad_norm": 0.06811874359846115, "global_step": 49693, "epoch": 468, "lr": 9.569813055835097e-07} +{"train_loss": 0.016058113425970078, "grad_norm": 0.09177010506391525, "global_step": 49694, "epoch": 468, "lr": 9.564043095608876e-07} +{"train_loss": 0.014323575422167778, "grad_norm": 0.11501871794462204, "global_step": 49695, "epoch": 468, "lr": 9.558274858561367e-07} +{"train_loss": 0.015866070985794067, "grad_norm": 0.07913053780794144, "global_step": 49696, "epoch": 468, "lr": 9.552508344712774e-07} +{"train_loss": 0.013981569558382034, "grad_norm": 0.0625077486038208, "global_step": 49697, "epoch": 468, "lr": 9.54674355408336e-07} +{"train_loss": 0.016032686457037926, "grad_norm": 0.06794676929712296, "global_step": 49698, "epoch": 468, "lr": 9.540980486693386e-07} +{"train_loss": 0.015071986243128777, "grad_norm": 0.07927945256233215, "global_step": 49699, "epoch": 468, "lr": 9.535219142563168e-07} +{"train_loss": 0.014088636264204979, "grad_norm": 0.06264787912368774, "global_step": 49700, "epoch": 468, "lr": 9.529459521712803e-07} +{"train_loss": 0.013903165236115456, "grad_norm": 0.13493701815605164, "global_step": 49701, "epoch": 468, "lr": 9.523701624162662e-07} +{"train_loss": 0.01628914102911949, "grad_norm": 0.06853388994932175, "global_step": 49702, "epoch": 468, "lr": 9.517945449932953e-07} +{"train_loss": 0.016477664932608604, "grad_norm": 0.08688127249479294, "global_step": 49703, "epoch": 468, "lr": 9.512190999043879e-07} +{"train_loss": 0.01611648127436638, "grad_norm": 0.08439670503139496, "global_step": 49704, "epoch": 468, "lr": 9.50643827151565e-07} +{"train_loss": 0.01356353610754013, "grad_norm": 0.07337123155593872, "global_step": 49705, "epoch": 468, "lr": 9.500687267368413e-07} +{"train_loss": 0.014291536994278431, "grad_norm": 0.09426011145114899, "global_step": 49706, "epoch": 468, "lr": 9.494937986622543e-07} +{"train_loss": 0.012749148532748222, "grad_norm": 0.08428165316581726, "global_step": 49707, "epoch": 468, "lr": 9.489190429298133e-07} +{"train_loss": 0.014160295017063618, "grad_norm": 0.07518414407968521, "global_step": 49708, "epoch": 468, "lr": 9.483444595415392e-07} +{"train_loss": 0.017193863168358803, "grad_norm": 0.12144169956445694, "global_step": 49709, "epoch": 468, "lr": 9.477700484994523e-07} +{"train_loss": 0.015681598335504532, "grad_norm": 0.09410949796438217, "global_step": 49710, "epoch": 468, "lr": 9.471958098055623e-07} +{"train_loss": 0.015428517013788223, "grad_norm": 0.08693835884332657, "global_step": 49711, "epoch": 468, "lr": 9.466217434618952e-07} +{"train_loss": 0.01561641227453947, "grad_norm": 0.06509216129779816, "global_step": 49712, "epoch": 468, "lr": 9.460478494704716e-07} +{"train_loss": 0.01495484287305823, "grad_norm": 0.1223127469420433, "global_step": 49713, "epoch": 468, "lr": 9.454741278333012e-07, "val_loss": 0.019566871225833893} +{"train_loss": 0.014126665890216827, "grad_norm": 0.06564629822969437, "global_step": 49714, "epoch": 469, "lr": 9.44900578552399e-07} +{"train_loss": 0.014690193347632885, "grad_norm": 0.06864006072282791, "global_step": 49715, "epoch": 469, "lr": 9.443272016297799e-07} +{"train_loss": 0.015327515080571175, "grad_norm": 0.07940173894166946, "global_step": 49716, "epoch": 469, "lr": 9.437539970674646e-07} +{"train_loss": 0.014526726678013802, "grad_norm": 0.09622348099946976, "global_step": 49717, "epoch": 469, "lr": 9.431809648674628e-07} +{"train_loss": 0.01666230335831642, "grad_norm": 0.1035841554403305, "global_step": 49718, "epoch": 469, "lr": 9.426081050317892e-07} +{"train_loss": 0.014214466325938702, "grad_norm": 0.15968821942806244, "global_step": 49719, "epoch": 469, "lr": 9.420354175624591e-07} +{"train_loss": 0.015256380662322044, "grad_norm": 0.08463267982006073, "global_step": 49720, "epoch": 469, "lr": 9.414629024614819e-07} +{"train_loss": 0.012559659779071808, "grad_norm": 0.14705504477024078, "global_step": 49721, "epoch": 469, "lr": 9.408905597308671e-07} +{"train_loss": 0.016089756041765213, "grad_norm": 0.1148204430937767, "global_step": 49722, "epoch": 469, "lr": 9.403183893726242e-07} +{"train_loss": 0.012751694768667221, "grad_norm": 0.0983952209353447, "global_step": 49723, "epoch": 469, "lr": 9.397463913887738e-07} +{"train_loss": 0.0168661717325449, "grad_norm": 0.07696563750505447, "global_step": 49724, "epoch": 469, "lr": 9.391745657813144e-07} +{"train_loss": 0.016628805547952652, "grad_norm": 0.07730232924222946, "global_step": 49725, "epoch": 469, "lr": 9.386029125522611e-07} +{"train_loss": 0.015503001399338245, "grad_norm": 0.06836839765310287, "global_step": 49726, "epoch": 469, "lr": 9.380314317036232e-07} +{"train_loss": 0.01609811931848526, "grad_norm": 0.0610661655664444, "global_step": 49727, "epoch": 469, "lr": 9.374601232374047e-07} +{"train_loss": 0.015146549791097641, "grad_norm": 0.1090058982372284, "global_step": 49728, "epoch": 469, "lr": 9.368889871556096e-07} +{"train_loss": 0.014852803200483322, "grad_norm": 0.08096301555633545, "global_step": 49729, "epoch": 469, "lr": 9.363180234602531e-07} +{"train_loss": 0.014457464218139648, "grad_norm": 0.09079928696155548, "global_step": 49730, "epoch": 469, "lr": 9.357472321533389e-07} +{"train_loss": 0.014738114550709724, "grad_norm": 0.06911765784025192, "global_step": 49731, "epoch": 469, "lr": 9.351766132368711e-07} +{"train_loss": 0.01420171745121479, "grad_norm": 0.07897384464740753, "global_step": 49732, "epoch": 469, "lr": 9.346061667128537e-07} +{"train_loss": 0.011223874054849148, "grad_norm": 0.06576137244701385, "global_step": 49733, "epoch": 469, "lr": 9.340358925832959e-07} +{"train_loss": 0.01522120088338852, "grad_norm": 0.08880344778299332, "global_step": 49734, "epoch": 469, "lr": 9.334657908501909e-07} +{"train_loss": 0.015669971704483032, "grad_norm": 0.09719561785459518, "global_step": 49735, "epoch": 469, "lr": 9.328958615155536e-07} +{"train_loss": 0.012696869671344757, "grad_norm": 0.08833001554012299, "global_step": 49736, "epoch": 469, "lr": 9.323261045813824e-07} +{"train_loss": 0.015881262719631195, "grad_norm": 0.08034355938434601, "global_step": 49737, "epoch": 469, "lr": 9.317565200496814e-07} +{"train_loss": 0.01700408384203911, "grad_norm": 0.07873699814081192, "global_step": 49738, "epoch": 469, "lr": 9.311871079224377e-07} +{"train_loss": 0.01319885067641735, "grad_norm": 0.06934111565351486, "global_step": 49739, "epoch": 469, "lr": 9.306178682016719e-07} +{"train_loss": 0.01441621221601963, "grad_norm": 0.06500454992055893, "global_step": 49740, "epoch": 469, "lr": 9.300488008893716e-07} +{"train_loss": 0.012546390295028687, "grad_norm": 0.1040269285440445, "global_step": 49741, "epoch": 469, "lr": 9.294799059875403e-07} +{"train_loss": 0.014329769648611546, "grad_norm": 0.07856931537389755, "global_step": 49742, "epoch": 469, "lr": 9.289111834981767e-07} +{"train_loss": 0.015516829676926136, "grad_norm": 0.15101848542690277, "global_step": 49743, "epoch": 469, "lr": 9.283426334232792e-07} +{"train_loss": 0.015558364801108837, "grad_norm": 0.13690584897994995, "global_step": 49744, "epoch": 469, "lr": 9.277742557648461e-07} +{"train_loss": 0.014159871265292168, "grad_norm": 0.09510475397109985, "global_step": 49745, "epoch": 469, "lr": 9.272060505248703e-07} +{"train_loss": 0.014078542590141296, "grad_norm": 0.06365091353654861, "global_step": 49746, "epoch": 469, "lr": 9.266380177053502e-07} +{"train_loss": 0.015957506373524666, "grad_norm": 0.07320073246955872, "global_step": 49747, "epoch": 469, "lr": 9.260701573082897e-07} +{"train_loss": 0.014800877310335636, "grad_norm": 0.07427923381328583, "global_step": 49748, "epoch": 469, "lr": 9.255024693356706e-07} +{"train_loss": 0.014968414790928364, "grad_norm": 0.05269530043005943, "global_step": 49749, "epoch": 469, "lr": 9.249349537894969e-07} +{"train_loss": 0.017348431050777435, "grad_norm": 0.08699165284633636, "global_step": 49750, "epoch": 469, "lr": 9.243676106717613e-07} +{"train_loss": 0.014109157025814056, "grad_norm": 0.09029904752969742, "global_step": 49751, "epoch": 469, "lr": 9.238004399844513e-07} +{"train_loss": 0.01559494435787201, "grad_norm": 0.13280147314071655, "global_step": 49752, "epoch": 469, "lr": 9.232334417295651e-07} +{"train_loss": 0.015591432340443134, "grad_norm": 0.06440085172653198, "global_step": 49753, "epoch": 469, "lr": 9.226666159090958e-07} +{"train_loss": 0.013528501614928246, "grad_norm": 0.10870025306940079, "global_step": 49754, "epoch": 469, "lr": 9.220999625250304e-07} +{"train_loss": 0.017121173441410065, "grad_norm": 0.07614847272634506, "global_step": 49755, "epoch": 469, "lr": 9.215334815793675e-07} +{"train_loss": 0.017033806070685387, "grad_norm": 0.09969072788953781, "global_step": 49756, "epoch": 469, "lr": 9.209671730740832e-07} +{"train_loss": 0.01508621871471405, "grad_norm": 0.1389203667640686, "global_step": 49757, "epoch": 469, "lr": 9.204010370111815e-07} +{"train_loss": 0.015242952853441238, "grad_norm": 0.08407071232795715, "global_step": 49758, "epoch": 469, "lr": 9.198350733926442e-07} +{"train_loss": 0.014843493700027466, "grad_norm": 0.10089732706546783, "global_step": 49759, "epoch": 469, "lr": 9.19269282220464e-07} +{"train_loss": 0.014596318826079369, "grad_norm": 0.06869561225175858, "global_step": 49760, "epoch": 469, "lr": 9.187036634966284e-07} +{"train_loss": 0.015008669346570969, "grad_norm": 0.07979723066091537, "global_step": 49761, "epoch": 469, "lr": 9.181382172231246e-07} +{"train_loss": 0.014588544145226479, "grad_norm": 0.09819042682647705, "global_step": 49762, "epoch": 469, "lr": 9.175729434019342e-07} +{"train_loss": 0.014240184798836708, "grad_norm": 0.06603042781352997, "global_step": 49763, "epoch": 469, "lr": 9.170078420350448e-07} +{"train_loss": 0.01444520615041256, "grad_norm": 0.0854785144329071, "global_step": 49764, "epoch": 469, "lr": 9.16442913124449e-07} +{"train_loss": 0.016114676371216774, "grad_norm": 0.08773549646139145, "global_step": 49765, "epoch": 469, "lr": 9.15878156672123e-07} +{"train_loss": 0.015315033495426178, "grad_norm": 0.07558649778366089, "global_step": 49766, "epoch": 469, "lr": 9.153135726800599e-07} +{"train_loss": 0.017861340194940567, "grad_norm": 0.10904808342456818, "global_step": 49767, "epoch": 469, "lr": 9.147491611502357e-07} +{"train_loss": 0.013486024923622608, "grad_norm": 0.0936233252286911, "global_step": 49768, "epoch": 469, "lr": 9.141849220846377e-07} +{"train_loss": 0.015879441052675247, "grad_norm": 0.107046939432621, "global_step": 49769, "epoch": 469, "lr": 9.136208554852421e-07} +{"train_loss": 0.01415790244936943, "grad_norm": 0.060291096568107605, "global_step": 49770, "epoch": 469, "lr": 9.130569613540418e-07} +{"train_loss": 0.016173694282770157, "grad_norm": 0.10315856337547302, "global_step": 49771, "epoch": 469, "lr": 9.124932396930075e-07} +{"train_loss": 0.014926482923328876, "grad_norm": 0.07999494671821594, "global_step": 49772, "epoch": 469, "lr": 9.119296905041264e-07} +{"train_loss": 0.013326434418559074, "grad_norm": 0.07916431128978729, "global_step": 49773, "epoch": 469, "lr": 9.113663137893746e-07} +{"train_loss": 0.016514254733920097, "grad_norm": 0.07666712254285812, "global_step": 49774, "epoch": 469, "lr": 9.108031095507396e-07} +{"train_loss": 0.016464389860630035, "grad_norm": 0.08511035889387131, "global_step": 49775, "epoch": 469, "lr": 9.102400777901865e-07} +{"train_loss": 0.01644684001803398, "grad_norm": 0.07769306749105453, "global_step": 49776, "epoch": 469, "lr": 9.096772185097025e-07} +{"train_loss": 0.017633214592933655, "grad_norm": 0.07279407978057861, "global_step": 49777, "epoch": 469, "lr": 9.091145317112692e-07} +{"train_loss": 0.013681279495358467, "grad_norm": 0.07730530202388763, "global_step": 49778, "epoch": 469, "lr": 9.085520173968576e-07} +{"train_loss": 0.014860973693430424, "grad_norm": 0.095134437084198, "global_step": 49779, "epoch": 469, "lr": 9.079896755684437e-07} +{"train_loss": 0.012252844870090485, "grad_norm": 0.06980106979608536, "global_step": 49780, "epoch": 469, "lr": 9.07427506227998e-07} +{"train_loss": 0.013485242612659931, "grad_norm": 0.08942481875419617, "global_step": 49781, "epoch": 469, "lr": 9.068655093775081e-07} +{"train_loss": 0.013695742934942245, "grad_norm": 0.060764946043491364, "global_step": 49782, "epoch": 469, "lr": 9.063036850189444e-07} +{"train_loss": 0.014374298974871635, "grad_norm": 0.08121750503778458, "global_step": 49783, "epoch": 469, "lr": 9.057420331542721e-07} +{"train_loss": 0.014229369349777699, "grad_norm": 0.0671471506357193, "global_step": 49784, "epoch": 469, "lr": 9.05180553785484e-07} +{"train_loss": 0.01646934449672699, "grad_norm": 0.09085448086261749, "global_step": 49785, "epoch": 469, "lr": 9.046192469145343e-07} +{"train_loss": 0.013989100232720375, "grad_norm": 0.08285640925168991, "global_step": 49786, "epoch": 469, "lr": 9.040581125433989e-07} +{"train_loss": 0.015298978425562382, "grad_norm": 0.12780769169330597, "global_step": 49787, "epoch": 469, "lr": 9.034971506740542e-07} +{"train_loss": 0.01404748298227787, "grad_norm": 0.08556175976991653, "global_step": 49788, "epoch": 469, "lr": 9.029363613084652e-07} +{"train_loss": 0.017264921218156815, "grad_norm": 0.063034288585186, "global_step": 49789, "epoch": 469, "lr": 9.023757444486081e-07} +{"train_loss": 0.014711756259202957, "grad_norm": 0.10971441864967346, "global_step": 49790, "epoch": 469, "lr": 9.018153000964536e-07} +{"train_loss": 0.015327041037380695, "grad_norm": 0.058058857917785645, "global_step": 49791, "epoch": 469, "lr": 9.012550282539722e-07} +{"train_loss": 0.013965621590614319, "grad_norm": 0.06849528849124908, "global_step": 49792, "epoch": 469, "lr": 9.006949289231181e-07} +{"train_loss": 0.015212710946798325, "grad_norm": 0.06534268707036972, "global_step": 49793, "epoch": 469, "lr": 9.00135002105873e-07} +{"train_loss": 0.013744974508881569, "grad_norm": 0.11114826053380966, "global_step": 49794, "epoch": 469, "lr": 8.995752478042019e-07} +{"train_loss": 0.016092706471681595, "grad_norm": 0.11750099062919617, "global_step": 49795, "epoch": 469, "lr": 8.990156660200699e-07} +{"train_loss": 0.015876930207014084, "grad_norm": 0.08270946890115738, "global_step": 49796, "epoch": 469, "lr": 8.984562567554422e-07} +{"train_loss": 0.014869483187794685, "grad_norm": 0.08574418723583221, "global_step": 49797, "epoch": 469, "lr": 8.978970200122838e-07} +{"train_loss": 0.01839071698486805, "grad_norm": 0.06429807096719742, "global_step": 49798, "epoch": 469, "lr": 8.973379557925655e-07} +{"train_loss": 0.014127696864306927, "grad_norm": 0.0900624543428421, "global_step": 49799, "epoch": 469, "lr": 8.967790640982465e-07} +{"train_loss": 0.013192780315876007, "grad_norm": 0.08114995062351227, "global_step": 49800, "epoch": 469, "lr": 8.962203449312867e-07} +{"train_loss": 0.014268631115555763, "grad_norm": 0.06944522261619568, "global_step": 49801, "epoch": 469, "lr": 8.956617982936622e-07} +{"train_loss": 0.014355404302477837, "grad_norm": 0.08058936893939972, "global_step": 49802, "epoch": 469, "lr": 8.951034241873269e-07} +{"train_loss": 0.015212910249829292, "grad_norm": 0.13083893060684204, "global_step": 49803, "epoch": 469, "lr": 8.945452226142348e-07} +{"train_loss": 0.01370338536798954, "grad_norm": 0.09989669919013977, "global_step": 49804, "epoch": 469, "lr": 8.939871935763566e-07} +{"train_loss": 0.014426947571337223, "grad_norm": 0.09869690239429474, "global_step": 49805, "epoch": 469, "lr": 8.934293370756574e-07} +{"train_loss": 0.0129020344465971, "grad_norm": 0.08601819723844528, "global_step": 49806, "epoch": 469, "lr": 8.928716531140857e-07} +{"train_loss": 0.016188958659768105, "grad_norm": 0.08774785697460175, "global_step": 49807, "epoch": 469, "lr": 8.923141416936121e-07} +{"train_loss": 0.013663774356245995, "grad_norm": 0.07516330480575562, "global_step": 49808, "epoch": 469, "lr": 8.917568028161904e-07} +{"train_loss": 0.01417287066578865, "grad_norm": 0.0651632621884346, "global_step": 49809, "epoch": 469, "lr": 8.911996364837749e-07} +{"train_loss": 0.016114354133605957, "grad_norm": 0.09082459658384323, "global_step": 49810, "epoch": 469, "lr": 8.906426426983306e-07} +{"train_loss": 0.01536525133997202, "grad_norm": 0.06538403779268265, "global_step": 49811, "epoch": 469, "lr": 8.900858214618113e-07} +{"train_loss": 0.012805311009287834, "grad_norm": 0.07634514570236206, "global_step": 49812, "epoch": 469, "lr": 8.895291727761657e-07} +{"train_loss": 0.015730150043964386, "grad_norm": 0.07369399815797806, "global_step": 49813, "epoch": 469, "lr": 8.889726966433642e-07} +{"train_loss": 0.013615049421787262, "grad_norm": 0.08038946986198425, "global_step": 49814, "epoch": 469, "lr": 8.8841639306535e-07} +{"train_loss": 0.014820992015302181, "grad_norm": 0.0989546850323677, "global_step": 49815, "epoch": 469, "lr": 8.878602620440879e-07} +{"train_loss": 0.01710980013012886, "grad_norm": 0.09293628484010696, "global_step": 49816, "epoch": 469, "lr": 8.873043035815209e-07} +{"train_loss": 0.015597136691212654, "grad_norm": 0.08814193308353424, "global_step": 49817, "epoch": 469, "lr": 8.867485176796142e-07} +{"train_loss": 0.015306856483221054, "grad_norm": 0.07599848508834839, "global_step": 49818, "epoch": 469, "lr": 8.861929043403106e-07} +{"train_loss": 0.014952914727816605, "grad_norm": 0.09708938747644424, "global_step": 49819, "epoch": 469, "lr": 8.856374635655695e-07, "val_loss": 0.019732056185603142} +{"train_loss": 0.01591034047305584, "grad_norm": 0.07865037769079208, "global_step": 49820, "epoch": 470, "lr": 8.850821953573341e-07} +{"train_loss": 0.014310827478766441, "grad_norm": 0.07924244552850723, "global_step": 49821, "epoch": 470, "lr": 8.845270997175636e-07} +{"train_loss": 0.014203646220266819, "grad_norm": 0.09091569483280182, "global_step": 49822, "epoch": 470, "lr": 8.839721766482012e-07} +{"train_loss": 0.016655385494232178, "grad_norm": 0.10753876715898514, "global_step": 49823, "epoch": 470, "lr": 8.834174261512063e-07} +{"train_loss": 0.012763193808495998, "grad_norm": 0.05645211413502693, "global_step": 49824, "epoch": 470, "lr": 8.828628482285162e-07} +{"train_loss": 0.014378249645233154, "grad_norm": 0.09702646732330322, "global_step": 49825, "epoch": 470, "lr": 8.82308442882096e-07} +{"train_loss": 0.014138842932879925, "grad_norm": 0.0835779532790184, "global_step": 49826, "epoch": 470, "lr": 8.81754210113872e-07} +{"train_loss": 0.016245104372501373, "grad_norm": 0.11697900295257568, "global_step": 49827, "epoch": 470, "lr": 8.812001499258038e-07} +{"train_loss": 0.01863567903637886, "grad_norm": 0.089601531624794, "global_step": 49828, "epoch": 470, "lr": 8.806462623198452e-07} +{"train_loss": 0.014819735661149025, "grad_norm": 0.08784284442663193, "global_step": 49829, "epoch": 470, "lr": 8.800925472979282e-07} +{"train_loss": 0.012337285093963146, "grad_norm": 0.06941600888967514, "global_step": 49830, "epoch": 470, "lr": 8.795390048620011e-07} +{"train_loss": 0.011651802808046341, "grad_norm": 0.07704851031303406, "global_step": 49831, "epoch": 470, "lr": 8.78985635014018e-07} +{"train_loss": 0.0158251766115427, "grad_norm": 0.09711217135190964, "global_step": 49832, "epoch": 470, "lr": 8.784324377559161e-07} +{"train_loss": 0.014132820069789886, "grad_norm": 0.11708331108093262, "global_step": 49833, "epoch": 470, "lr": 8.778794130896384e-07} +{"train_loss": 0.015571309253573418, "grad_norm": 0.07467992603778839, "global_step": 49834, "epoch": 470, "lr": 8.773265610171277e-07} +{"train_loss": 0.013463864102959633, "grad_norm": 0.06071992963552475, "global_step": 49835, "epoch": 470, "lr": 8.767738815403381e-07} +{"train_loss": 0.011174244806170464, "grad_norm": 0.05451246723532677, "global_step": 49836, "epoch": 470, "lr": 8.762213746611902e-07} +{"train_loss": 0.015089818276464939, "grad_norm": 0.11749796569347382, "global_step": 49837, "epoch": 470, "lr": 8.756690403816492e-07} +{"train_loss": 0.012582449242472649, "grad_norm": 0.09404551237821579, "global_step": 49838, "epoch": 470, "lr": 8.751168787036301e-07} +{"train_loss": 0.015818903222680092, "grad_norm": 0.1045001670718193, "global_step": 49839, "epoch": 470, "lr": 8.745648896290981e-07} +{"train_loss": 0.015490611083805561, "grad_norm": 0.07466684281826019, "global_step": 49840, "epoch": 470, "lr": 8.740130731599738e-07} +{"train_loss": 0.014706835150718689, "grad_norm": 0.07440488785505295, "global_step": 49841, "epoch": 470, "lr": 8.734614292982057e-07} +{"train_loss": 0.01562536135315895, "grad_norm": 0.1086547002196312, "global_step": 49842, "epoch": 470, "lr": 8.729099580457312e-07} +{"train_loss": 0.0159891527146101, "grad_norm": 0.09788251668214798, "global_step": 49843, "epoch": 470, "lr": 8.723586594044874e-07} +{"train_loss": 0.012585107237100601, "grad_norm": 0.12493825703859329, "global_step": 49844, "epoch": 470, "lr": 8.718075333764009e-07} +{"train_loss": 0.01633410155773163, "grad_norm": 0.14448793232440948, "global_step": 49845, "epoch": 470, "lr": 8.712565799634254e-07} +{"train_loss": 0.015207278542220592, "grad_norm": 0.13817596435546875, "global_step": 49846, "epoch": 470, "lr": 8.707057991674872e-07} +{"train_loss": 0.014325588941574097, "grad_norm": 0.0782504603266716, "global_step": 49847, "epoch": 470, "lr": 8.701551909905236e-07} +{"train_loss": 0.01504360418766737, "grad_norm": 0.07804553210735321, "global_step": 49848, "epoch": 470, "lr": 8.696047554344666e-07} +{"train_loss": 0.017433159053325653, "grad_norm": 0.12976396083831787, "global_step": 49849, "epoch": 470, "lr": 8.690544925012589e-07} +{"train_loss": 0.015545490197837353, "grad_norm": 0.06455384939908981, "global_step": 49850, "epoch": 470, "lr": 8.685044021928212e-07} +{"train_loss": 0.012600334361195564, "grad_norm": 0.07976044714450836, "global_step": 49851, "epoch": 470, "lr": 8.679544845110909e-07} +{"train_loss": 0.0151398666203022, "grad_norm": 0.08258787542581558, "global_step": 49852, "epoch": 470, "lr": 8.674047394580054e-07} +{"train_loss": 0.015234842896461487, "grad_norm": 0.14055800437927246, "global_step": 49853, "epoch": 470, "lr": 8.668551670354852e-07} +{"train_loss": 0.017746856436133385, "grad_norm": 0.08699549734592438, "global_step": 49854, "epoch": 470, "lr": 8.663057672454732e-07} +{"train_loss": 0.015985604375600815, "grad_norm": 0.1410127431154251, "global_step": 49855, "epoch": 470, "lr": 8.657565400899015e-07} +{"train_loss": 0.017270654439926147, "grad_norm": 0.08609309792518616, "global_step": 49856, "epoch": 470, "lr": 8.652074855706904e-07} +{"train_loss": 0.012745527550578117, "grad_norm": 0.0841173380613327, "global_step": 49857, "epoch": 470, "lr": 8.646586036897664e-07} +{"train_loss": 0.015012889169156551, "grad_norm": 0.051287099719047546, "global_step": 49858, "epoch": 470, "lr": 8.641098944490666e-07} +{"train_loss": 0.014977462589740753, "grad_norm": 0.06884423643350601, "global_step": 49859, "epoch": 470, "lr": 8.635613578505175e-07} +{"train_loss": 0.014276135712862015, "grad_norm": 0.08731260895729065, "global_step": 49860, "epoch": 470, "lr": 8.630129938960396e-07} +{"train_loss": 0.01586315780878067, "grad_norm": 0.10654184967279434, "global_step": 49861, "epoch": 470, "lr": 8.624648025875704e-07} +{"train_loss": 0.01376691460609436, "grad_norm": 0.06333853304386139, "global_step": 49862, "epoch": 470, "lr": 8.619167839270303e-07} +{"train_loss": 0.013981565833091736, "grad_norm": 0.07535933703184128, "global_step": 49863, "epoch": 470, "lr": 8.613689379163459e-07} +{"train_loss": 0.016206422820687294, "grad_norm": 0.08912597596645355, "global_step": 49864, "epoch": 470, "lr": 8.608212645574376e-07} +{"train_loss": 0.014804543927311897, "grad_norm": 0.07601133733987808, "global_step": 49865, "epoch": 470, "lr": 8.602737638522318e-07} +{"train_loss": 0.01534836646169424, "grad_norm": 0.06540478020906448, "global_step": 49866, "epoch": 470, "lr": 8.597264358026602e-07} +{"train_loss": 0.013048957102000713, "grad_norm": 0.07446961849927902, "global_step": 49867, "epoch": 470, "lr": 8.59179280410638e-07} +{"train_loss": 0.015870515257120132, "grad_norm": 0.09240424633026123, "global_step": 49868, "epoch": 470, "lr": 8.586322976780858e-07} +{"train_loss": 0.014186060056090355, "grad_norm": 0.08057812601327896, "global_step": 49869, "epoch": 470, "lr": 8.5808548760693e-07} +{"train_loss": 0.013171277940273285, "grad_norm": 0.1238529160618782, "global_step": 49870, "epoch": 470, "lr": 8.575388501990911e-07} +{"train_loss": 0.01521306112408638, "grad_norm": 0.06715548783540726, "global_step": 49871, "epoch": 470, "lr": 8.569923854564843e-07} +{"train_loss": 0.016417834907770157, "grad_norm": 0.06388577073812485, "global_step": 49872, "epoch": 470, "lr": 8.564460933810415e-07} +{"train_loss": 0.016155540943145752, "grad_norm": 0.08103900402784348, "global_step": 49873, "epoch": 470, "lr": 8.558999739746721e-07} +{"train_loss": 0.01626170240342617, "grad_norm": 0.15117548406124115, "global_step": 49874, "epoch": 470, "lr": 8.553540272392969e-07} +{"train_loss": 0.014372996054589748, "grad_norm": 0.07448866963386536, "global_step": 49875, "epoch": 470, "lr": 8.548082531768365e-07} +{"train_loss": 0.014154617674648762, "grad_norm": 0.07814633101224899, "global_step": 49876, "epoch": 470, "lr": 8.542626517892116e-07} +{"train_loss": 0.014237308874726295, "grad_norm": 0.07927889376878738, "global_step": 49877, "epoch": 470, "lr": 8.537172230783264e-07} +{"train_loss": 0.012977108359336853, "grad_norm": 0.05414401367306709, "global_step": 49878, "epoch": 470, "lr": 8.531719670461125e-07} +{"train_loss": 0.012626275420188904, "grad_norm": 0.08259961009025574, "global_step": 49879, "epoch": 470, "lr": 8.52626883694474e-07} +{"train_loss": 0.013580426573753357, "grad_norm": 0.07975159585475922, "global_step": 49880, "epoch": 470, "lr": 8.520819730253371e-07} +{"train_loss": 0.013897579163312912, "grad_norm": 0.07878474146127701, "global_step": 49881, "epoch": 470, "lr": 8.515372350406004e-07} +{"train_loss": 0.015307608060538769, "grad_norm": 0.07982544600963593, "global_step": 49882, "epoch": 470, "lr": 8.509926697421955e-07} +{"train_loss": 0.018361855298280716, "grad_norm": 0.0948529839515686, "global_step": 49883, "epoch": 470, "lr": 8.504482771320265e-07} +{"train_loss": 0.013482663780450821, "grad_norm": 0.05694223567843437, "global_step": 49884, "epoch": 470, "lr": 8.499040572120088e-07} +{"train_loss": 0.016097191721200943, "grad_norm": 0.11419419944286346, "global_step": 49885, "epoch": 470, "lr": 8.493600099840515e-07} +{"train_loss": 0.01654021628201008, "grad_norm": 0.1313619762659073, "global_step": 49886, "epoch": 470, "lr": 8.488161354500701e-07} +{"train_loss": 0.011199390515685081, "grad_norm": 0.10206861793994904, "global_step": 49887, "epoch": 470, "lr": 8.48272433611974e-07} +{"train_loss": 0.016180025413632393, "grad_norm": 0.09542981535196304, "global_step": 49888, "epoch": 470, "lr": 8.477289044716674e-07} +{"train_loss": 0.01685928925871849, "grad_norm": 0.06757570058107376, "global_step": 49889, "epoch": 470, "lr": 8.471855480310709e-07} +{"train_loss": 0.015377639792859554, "grad_norm": 0.06228822469711304, "global_step": 49890, "epoch": 470, "lr": 8.466423642920884e-07} +{"train_loss": 0.015456783585250378, "grad_norm": 0.08358369767665863, "global_step": 49891, "epoch": 470, "lr": 8.460993532566241e-07} +{"train_loss": 0.015140681527554989, "grad_norm": 0.05985640361905098, "global_step": 49892, "epoch": 470, "lr": 8.455565149265932e-07} +{"train_loss": 0.01605447381734848, "grad_norm": 0.07902514189481735, "global_step": 49893, "epoch": 470, "lr": 8.450138493038994e-07} +{"train_loss": 0.014618420042097569, "grad_norm": 0.057750243693590164, "global_step": 49894, "epoch": 470, "lr": 8.444713563904527e-07} +{"train_loss": 0.01524067297577858, "grad_norm": 0.07646338641643524, "global_step": 49895, "epoch": 470, "lr": 8.439290361881514e-07} +{"train_loss": 0.014248952269554138, "grad_norm": 0.14609111845493317, "global_step": 49896, "epoch": 470, "lr": 8.433868886989105e-07} +{"train_loss": 0.016815148293972015, "grad_norm": 0.10623563081026077, "global_step": 49897, "epoch": 470, "lr": 8.428449139246287e-07} +{"train_loss": 0.014008402824401855, "grad_norm": 0.15522199869155884, "global_step": 49898, "epoch": 470, "lr": 8.423031118672098e-07} +{"train_loss": 0.017594587057828903, "grad_norm": 0.07564294338226318, "global_step": 49899, "epoch": 470, "lr": 8.417614825285636e-07} +{"train_loss": 0.01651204377412796, "grad_norm": 0.08426471054553986, "global_step": 49900, "epoch": 470, "lr": 8.412200259105885e-07} +{"train_loss": 0.015081080608069897, "grad_norm": 0.1830596625804901, "global_step": 49901, "epoch": 470, "lr": 8.406787420151829e-07} +{"train_loss": 0.01612960547208786, "grad_norm": 0.07526304572820663, "global_step": 49902, "epoch": 470, "lr": 8.401376308442621e-07} +{"train_loss": 0.01363877672702074, "grad_norm": 0.09457974880933762, "global_step": 49903, "epoch": 470, "lr": 8.395966923997134e-07} +{"train_loss": 0.016403764486312866, "grad_norm": 0.08201973140239716, "global_step": 49904, "epoch": 470, "lr": 8.390559266834408e-07} +{"train_loss": 0.015794608741998672, "grad_norm": 0.10747089236974716, "global_step": 49905, "epoch": 470, "lr": 8.385153336973484e-07} +{"train_loss": 0.016233600676059723, "grad_norm": 0.0832986831665039, "global_step": 49906, "epoch": 470, "lr": 8.37974913443329e-07} +{"train_loss": 0.01368789654225111, "grad_norm": 0.054603006690740585, "global_step": 49907, "epoch": 470, "lr": 8.374346659232924e-07} +{"train_loss": 0.013405500911176205, "grad_norm": 0.06306291371583939, "global_step": 49908, "epoch": 470, "lr": 8.368945911391313e-07} +{"train_loss": 0.01556072011590004, "grad_norm": 0.11652785539627075, "global_step": 49909, "epoch": 470, "lr": 8.363546890927387e-07} +{"train_loss": 0.01614215038716793, "grad_norm": 0.10610386729240417, "global_step": 49910, "epoch": 470, "lr": 8.358149597860132e-07} +{"train_loss": 0.014723595231771469, "grad_norm": 0.09636207669973373, "global_step": 49911, "epoch": 470, "lr": 8.352754032208531e-07} +{"train_loss": 0.014033520594239235, "grad_norm": 0.1312018483877182, "global_step": 49912, "epoch": 470, "lr": 8.347360193991571e-07} +{"train_loss": 0.016792673617601395, "grad_norm": 0.07493270933628082, "global_step": 49913, "epoch": 470, "lr": 8.34196808322818e-07} +{"train_loss": 0.013054072856903076, "grad_norm": 0.09003511816263199, "global_step": 49914, "epoch": 470, "lr": 8.336577699937286e-07} +{"train_loss": 0.015120347961783409, "grad_norm": 0.08446915447711945, "global_step": 49915, "epoch": 470, "lr": 8.331189044137821e-07} +{"train_loss": 0.01587846502661705, "grad_norm": 0.061453677713871, "global_step": 49916, "epoch": 470, "lr": 8.325802115848714e-07} +{"train_loss": 0.013359473086893559, "grad_norm": 0.10705109685659409, "global_step": 49917, "epoch": 470, "lr": 8.320416915088946e-07} +{"train_loss": 0.014612551778554916, "grad_norm": 0.07979664206504822, "global_step": 49918, "epoch": 470, "lr": 8.315033441877395e-07} +{"train_loss": 0.016251830384135246, "grad_norm": 0.17337864637374878, "global_step": 49919, "epoch": 470, "lr": 8.309651696233045e-07} +{"train_loss": 0.01394298393279314, "grad_norm": 0.12558557093143463, "global_step": 49920, "epoch": 470, "lr": 8.304271678174658e-07} +{"train_loss": 0.014996932819485664, "grad_norm": 0.08515042811632156, "global_step": 49921, "epoch": 470, "lr": 8.29889338772133e-07} +{"train_loss": 0.013998344540596008, "grad_norm": 0.06640893220901489, "global_step": 49922, "epoch": 470, "lr": 8.293516824891767e-07} +{"train_loss": 0.013446679338812828, "grad_norm": 0.07105144113302231, "global_step": 49923, "epoch": 470, "lr": 8.288141989704956e-07} +{"train_loss": 0.015517506748437881, "grad_norm": 0.074270099401474, "global_step": 49924, "epoch": 470, "lr": 8.282768882179825e-07} +{"train_loss": 0.014959976683318053, "grad_norm": 0.10474281758069992, "global_step": 49925, "epoch": 470, "lr": 8.277397502335194e-07, "train_batch_mse_error": 0.01335921697318554, "val_loss": 0.019399769604206085} +{"train_loss": 0.013185083866119385, "grad_norm": 0.0698138177394867, "global_step": 49926, "epoch": 471, "lr": 8.272027850189879e-07} +{"train_loss": 0.01741563156247139, "grad_norm": 0.09164375066757202, "global_step": 49927, "epoch": 471, "lr": 8.26665992576292e-07} +{"train_loss": 0.015410153195261955, "grad_norm": 0.16060888767242432, "global_step": 49928, "epoch": 471, "lr": 8.261293729072972e-07} +{"train_loss": 0.016296379268169403, "grad_norm": 0.076886385679245, "global_step": 49929, "epoch": 471, "lr": 8.255929260139017e-07} +{"train_loss": 0.013374777510762215, "grad_norm": 0.055014077574014664, "global_step": 49930, "epoch": 471, "lr": 8.250566518979874e-07} +{"train_loss": 0.0144534632563591, "grad_norm": 0.12330867350101471, "global_step": 49931, "epoch": 471, "lr": 8.245205505614417e-07} +{"train_loss": 0.015032983385026455, "grad_norm": 0.06297842413187027, "global_step": 49932, "epoch": 471, "lr": 8.239846220061409e-07} +{"train_loss": 0.013584909960627556, "grad_norm": 0.14017587900161743, "global_step": 49933, "epoch": 471, "lr": 8.234488662339667e-07} +{"train_loss": 0.015238822437822819, "grad_norm": 0.07671023905277252, "global_step": 49934, "epoch": 471, "lr": 8.229132832468178e-07} +{"train_loss": 0.01417453307658434, "grad_norm": 0.09993439167737961, "global_step": 49935, "epoch": 471, "lr": 8.223778730465592e-07} +{"train_loss": 0.0152682363986969, "grad_norm": 0.08985640853643417, "global_step": 49936, "epoch": 471, "lr": 8.218426356350784e-07} +{"train_loss": 0.015362524427473545, "grad_norm": 0.06434879451990128, "global_step": 49937, "epoch": 471, "lr": 8.213075710142571e-07} +{"train_loss": 0.016043053939938545, "grad_norm": 0.07115059345960617, "global_step": 49938, "epoch": 471, "lr": 8.207726791859771e-07} +{"train_loss": 0.015716470777988434, "grad_norm": 0.09127239137887955, "global_step": 49939, "epoch": 471, "lr": 8.202379601521038e-07} +{"train_loss": 0.013005131855607033, "grad_norm": 0.09552762657403946, "global_step": 49940, "epoch": 471, "lr": 8.197034139145354e-07} +{"train_loss": 0.01502387784421444, "grad_norm": 0.10290157049894333, "global_step": 49941, "epoch": 471, "lr": 8.191690404751374e-07} +{"train_loss": 0.014746304601430893, "grad_norm": 0.058637481182813644, "global_step": 49942, "epoch": 471, "lr": 8.186348398357913e-07} +{"train_loss": 0.016701912507414818, "grad_norm": 0.11215156316757202, "global_step": 49943, "epoch": 471, "lr": 8.18100811998379e-07} +{"train_loss": 0.013590510934591293, "grad_norm": 0.11467621475458145, "global_step": 49944, "epoch": 471, "lr": 8.175669569647715e-07} +{"train_loss": 0.014912251383066177, "grad_norm": 0.060402028262615204, "global_step": 49945, "epoch": 471, "lr": 8.170332747368392e-07} +{"train_loss": 0.014687001705169678, "grad_norm": 0.07907836139202118, "global_step": 49946, "epoch": 471, "lr": 8.164997653164641e-07} +{"train_loss": 0.01682959869503975, "grad_norm": 0.0885496512055397, "global_step": 49947, "epoch": 471, "lr": 8.159664287055169e-07} +{"train_loss": 0.015717821195721626, "grad_norm": 0.09310062229633331, "global_step": 49948, "epoch": 471, "lr": 8.154332649058793e-07} +{"train_loss": 0.01199211087077856, "grad_norm": 0.06500426679849625, "global_step": 49949, "epoch": 471, "lr": 8.149002739194222e-07} +{"train_loss": 0.015097714960575104, "grad_norm": 0.07559385150671005, "global_step": 49950, "epoch": 471, "lr": 8.143674557480108e-07} +{"train_loss": 0.01468610018491745, "grad_norm": 0.07945321500301361, "global_step": 49951, "epoch": 471, "lr": 8.138348103935211e-07} +{"train_loss": 0.013823263347148895, "grad_norm": 0.08717046678066254, "global_step": 49952, "epoch": 471, "lr": 8.133023378578242e-07} +{"train_loss": 0.014113686978816986, "grad_norm": 0.12149901688098907, "global_step": 49953, "epoch": 471, "lr": 8.12770038142796e-07} +{"train_loss": 0.012283898890018463, "grad_norm": 0.07745154947042465, "global_step": 49954, "epoch": 471, "lr": 8.122379112503021e-07} +{"train_loss": 0.014438922517001629, "grad_norm": 0.06487325578927994, "global_step": 49955, "epoch": 471, "lr": 8.117059571822128e-07} +{"train_loss": 0.014070605859160423, "grad_norm": 0.07426047325134277, "global_step": 49956, "epoch": 471, "lr": 8.111741759403934e-07} +{"train_loss": 0.014548406936228275, "grad_norm": 0.0904422402381897, "global_step": 49957, "epoch": 471, "lr": 8.106425675267204e-07} +{"train_loss": 0.015680436044931412, "grad_norm": 0.07874362170696259, "global_step": 49958, "epoch": 471, "lr": 8.101111319430588e-07} +{"train_loss": 0.012717068195343018, "grad_norm": 0.053895074874162674, "global_step": 49959, "epoch": 471, "lr": 8.095798691912737e-07} +{"train_loss": 0.015429847873747349, "grad_norm": 0.07613114267587662, "global_step": 49960, "epoch": 471, "lr": 8.090487792732249e-07} +{"train_loss": 0.014164973050355911, "grad_norm": 0.055451758205890656, "global_step": 49961, "epoch": 471, "lr": 8.085178621907996e-07} +{"train_loss": 0.014679480344057083, "grad_norm": 0.0930408164858818, "global_step": 49962, "epoch": 471, "lr": 8.079871179458409e-07} +{"train_loss": 0.013184905983507633, "grad_norm": 0.08300068974494934, "global_step": 49963, "epoch": 471, "lr": 8.07456546540225e-07} +{"train_loss": 0.01585952192544937, "grad_norm": 0.13187038898468018, "global_step": 49964, "epoch": 471, "lr": 8.069261479758116e-07} +{"train_loss": 0.013231033459305763, "grad_norm": 0.061560481786727905, "global_step": 49965, "epoch": 471, "lr": 8.063959222544715e-07} +{"train_loss": 0.014651855453848839, "grad_norm": 0.06635711342096329, "global_step": 49966, "epoch": 471, "lr": 8.058658693780585e-07} +{"train_loss": 0.014254063367843628, "grad_norm": 0.11151203513145447, "global_step": 49967, "epoch": 471, "lr": 8.053359893484381e-07} +{"train_loss": 0.015083007514476776, "grad_norm": 0.06480158120393753, "global_step": 49968, "epoch": 471, "lr": 8.048062821674751e-07} +{"train_loss": 0.013898852281272411, "grad_norm": 0.06785955280065536, "global_step": 49969, "epoch": 471, "lr": 8.042767478370239e-07} +{"train_loss": 0.013853633776307106, "grad_norm": 0.10159533470869064, "global_step": 49970, "epoch": 471, "lr": 8.037473863589496e-07} +{"train_loss": 0.014498063363134861, "grad_norm": 0.08951106667518616, "global_step": 49971, "epoch": 471, "lr": 8.032181977351172e-07} +{"train_loss": 0.013598989695310593, "grad_norm": 0.07080662995576859, "global_step": 49972, "epoch": 471, "lr": 8.02689181967381e-07} +{"train_loss": 0.013943184167146683, "grad_norm": 0.07147380709648132, "global_step": 49973, "epoch": 471, "lr": 8.021603390575949e-07} +{"train_loss": 0.017288431525230408, "grad_norm": 0.0753854289650917, "global_step": 49974, "epoch": 471, "lr": 8.016316690076242e-07} +{"train_loss": 0.016441602259874344, "grad_norm": 0.09904427826404572, "global_step": 49975, "epoch": 471, "lr": 8.011031718193285e-07} +{"train_loss": 0.014193926006555557, "grad_norm": 0.08525330573320389, "global_step": 49976, "epoch": 471, "lr": 8.005748474945507e-07} +{"train_loss": 0.016007082536816597, "grad_norm": 0.11239275336265564, "global_step": 49977, "epoch": 471, "lr": 8.000466960351616e-07} +{"train_loss": 0.015927419066429138, "grad_norm": 0.08429683744907379, "global_step": 49978, "epoch": 471, "lr": 7.995187174430152e-07} +{"train_loss": 0.017133628949522972, "grad_norm": 0.08176405727863312, "global_step": 49979, "epoch": 471, "lr": 7.989909117199601e-07} +{"train_loss": 0.015295138582587242, "grad_norm": 0.10337997227907181, "global_step": 49980, "epoch": 471, "lr": 7.984632788678503e-07} +{"train_loss": 0.01835767552256584, "grad_norm": 0.06543672829866409, "global_step": 49981, "epoch": 471, "lr": 7.979358188885455e-07} +{"train_loss": 0.011340109631419182, "grad_norm": 0.07653629034757614, "global_step": 49982, "epoch": 471, "lr": 7.974085317839053e-07} +{"train_loss": 0.016386058181524277, "grad_norm": 0.09153318405151367, "global_step": 49983, "epoch": 471, "lr": 7.968814175557615e-07} +{"train_loss": 0.014642259106040001, "grad_norm": 0.08680592477321625, "global_step": 49984, "epoch": 471, "lr": 7.963544762059849e-07} +{"train_loss": 0.014025991782546043, "grad_norm": 0.07251142710447311, "global_step": 49985, "epoch": 471, "lr": 7.95827707736424e-07} +{"train_loss": 0.015625931322574615, "grad_norm": 0.14433741569519043, "global_step": 49986, "epoch": 471, "lr": 7.953011121489218e-07} +{"train_loss": 0.012917332351207733, "grad_norm": 0.06976215541362762, "global_step": 49987, "epoch": 471, "lr": 7.947746894453323e-07} +{"train_loss": 0.016324853524565697, "grad_norm": 0.08211596310138702, "global_step": 49988, "epoch": 471, "lr": 7.94248439627504e-07} +{"train_loss": 0.014839833602309227, "grad_norm": 0.18897980451583862, "global_step": 49989, "epoch": 471, "lr": 7.937223626972912e-07} +{"train_loss": 0.015181254595518112, "grad_norm": 0.08165048807859421, "global_step": 49990, "epoch": 471, "lr": 7.931964586565422e-07} +{"train_loss": 0.015103533864021301, "grad_norm": 0.09911704063415527, "global_step": 49991, "epoch": 471, "lr": 7.926707275070943e-07} +{"train_loss": 0.01804763823747635, "grad_norm": 0.17294734716415405, "global_step": 49992, "epoch": 471, "lr": 7.921451692508075e-07} +{"train_loss": 0.01381237618625164, "grad_norm": 0.17454388737678528, "global_step": 49993, "epoch": 471, "lr": 7.916197838895134e-07} +{"train_loss": 0.014752335846424103, "grad_norm": 0.08107393980026245, "global_step": 49994, "epoch": 471, "lr": 7.910945714250717e-07} +{"train_loss": 0.014807669445872307, "grad_norm": 0.10411330312490463, "global_step": 49995, "epoch": 471, "lr": 7.905695318593309e-07} +{"train_loss": 0.013636600226163864, "grad_norm": 0.0722876563668251, "global_step": 49996, "epoch": 471, "lr": 7.900446651941229e-07} +{"train_loss": 0.014719306491315365, "grad_norm": 0.06539291143417358, "global_step": 49997, "epoch": 471, "lr": 7.895199714312962e-07} +{"train_loss": 0.01265413872897625, "grad_norm": 0.1306394785642624, "global_step": 49998, "epoch": 471, "lr": 7.889954505726938e-07} +{"train_loss": 0.017826562747359276, "grad_norm": 0.06799149513244629, "global_step": 49999, "epoch": 471, "lr": 7.884711026201585e-07} +{"train_loss": 0.015951743349432945, "grad_norm": 0.07249225676059723, "global_step": 50000, "epoch": 471, "lr": 7.879469275755391e-07} +{"train_loss": 0.01668725535273552, "grad_norm": 0.09677808731794357, "global_step": 50001, "epoch": 471, "lr": 7.874229254406673e-07} +{"train_loss": 0.01428175251930952, "grad_norm": 0.07685421407222748, "global_step": 50002, "epoch": 471, "lr": 7.868990962173972e-07} +{"train_loss": 0.01873016729950905, "grad_norm": 0.11694904416799545, "global_step": 50003, "epoch": 471, "lr": 7.863754399075607e-07} +{"train_loss": 0.015443507581949234, "grad_norm": 0.07577474415302277, "global_step": 50004, "epoch": 471, "lr": 7.858519565129896e-07} +{"train_loss": 0.014444160275161266, "grad_norm": 0.08056429028511047, "global_step": 50005, "epoch": 471, "lr": 7.85328646035538e-07} +{"train_loss": 0.01558389700949192, "grad_norm": 0.0826287493109703, "global_step": 50006, "epoch": 471, "lr": 7.848055084770378e-07} +{"train_loss": 0.01571515202522278, "grad_norm": 0.10175462812185287, "global_step": 50007, "epoch": 471, "lr": 7.842825438393265e-07} +{"train_loss": 0.017446327954530716, "grad_norm": 0.077715203166008, "global_step": 50008, "epoch": 471, "lr": 7.837597521242468e-07} +{"train_loss": 0.01269497349858284, "grad_norm": 0.1095433160662651, "global_step": 50009, "epoch": 471, "lr": 7.832371333336308e-07} +{"train_loss": 0.012034794315695763, "grad_norm": 0.09249214082956314, "global_step": 50010, "epoch": 471, "lr": 7.827146874693158e-07} +{"train_loss": 0.014287447556853294, "grad_norm": 0.0712689533829689, "global_step": 50011, "epoch": 471, "lr": 7.821924145331339e-07} +{"train_loss": 0.015517167747020721, "grad_norm": 0.0593632273375988, "global_step": 50012, "epoch": 471, "lr": 7.816703145269277e-07} +{"train_loss": 0.016309333965182304, "grad_norm": 0.07293196022510529, "global_step": 50013, "epoch": 471, "lr": 7.811483874525239e-07} +{"train_loss": 0.014517078176140785, "grad_norm": 0.10955784469842911, "global_step": 50014, "epoch": 471, "lr": 7.806266333117595e-07} +{"train_loss": 0.01462824922055006, "grad_norm": 0.12421130388975143, "global_step": 50015, "epoch": 471, "lr": 7.801050521064723e-07} +{"train_loss": 0.01763157919049263, "grad_norm": 0.06580033898353577, "global_step": 50016, "epoch": 471, "lr": 7.795836438384885e-07} +{"train_loss": 0.017019666731357574, "grad_norm": 0.12303883582353592, "global_step": 50017, "epoch": 471, "lr": 7.790624085096398e-07} +{"train_loss": 0.017130305990576744, "grad_norm": 0.09411273896694183, "global_step": 50018, "epoch": 471, "lr": 7.785413461217638e-07} +{"train_loss": 0.013618926517665386, "grad_norm": 0.07515574246644974, "global_step": 50019, "epoch": 471, "lr": 7.780204566766869e-07} +{"train_loss": 0.015036877244710922, "grad_norm": 0.11658121645450592, "global_step": 50020, "epoch": 471, "lr": 7.774997401762463e-07} +{"train_loss": 0.013421755284070969, "grad_norm": 0.08728723973035812, "global_step": 50021, "epoch": 471, "lr": 7.769791966222573e-07} +{"train_loss": 0.013706804253160954, "grad_norm": 0.07503800839185715, "global_step": 50022, "epoch": 471, "lr": 7.764588260165517e-07} +{"train_loss": 0.01595916599035263, "grad_norm": 0.06871788948774338, "global_step": 50023, "epoch": 471, "lr": 7.759386283609726e-07} +{"train_loss": 0.014223203994333744, "grad_norm": 0.06992452591657639, "global_step": 50024, "epoch": 471, "lr": 7.754186036573352e-07} +{"train_loss": 0.01644774153828621, "grad_norm": 0.11722386628389359, "global_step": 50025, "epoch": 471, "lr": 7.748987519074713e-07} +{"train_loss": 0.013795848935842514, "grad_norm": 0.09911011159420013, "global_step": 50026, "epoch": 471, "lr": 7.743790731132073e-07} +{"train_loss": 0.015821292996406555, "grad_norm": 0.06459762156009674, "global_step": 50027, "epoch": 471, "lr": 7.738595672763638e-07} +{"train_loss": 0.012379568070173264, "grad_norm": 0.10735723376274109, "global_step": 50028, "epoch": 471, "lr": 7.733402343987672e-07} +{"train_loss": 0.01648852787911892, "grad_norm": 0.0829438641667366, "global_step": 50029, "epoch": 471, "lr": 7.728210744822495e-07} +{"train_loss": 0.014450018294155598, "grad_norm": 0.053904157131910324, "global_step": 50030, "epoch": 471, "lr": 7.723020875286313e-07} +{"train_loss": 0.014958386283086718, "grad_norm": 0.17404113709926605, "global_step": 50031, "epoch": 471, "lr": 7.717832735397335e-07, "val_loss": 0.01972445659339428} +{"train_loss": 0.0160670205950737, "grad_norm": 0.10592890530824661, "global_step": 50032, "epoch": 472, "lr": 7.712646325173767e-07} +{"train_loss": 0.014703531749546528, "grad_norm": 0.09269026666879654, "global_step": 50033, "epoch": 472, "lr": 7.707461644633928e-07} +{"train_loss": 0.014417426660656929, "grad_norm": 0.10077004134654999, "global_step": 50034, "epoch": 472, "lr": 7.702278693795917e-07} +{"train_loss": 0.015090327709913254, "grad_norm": 0.09532495588064194, "global_step": 50035, "epoch": 472, "lr": 7.697097472678049e-07} +{"train_loss": 0.01480773650109768, "grad_norm": 0.06995843350887299, "global_step": 50036, "epoch": 472, "lr": 7.69191798129848e-07} +{"train_loss": 0.012555085122585297, "grad_norm": 0.0955238938331604, "global_step": 50037, "epoch": 472, "lr": 7.686740219675414e-07} +{"train_loss": 0.014508670195937157, "grad_norm": 0.09041019529104233, "global_step": 50038, "epoch": 472, "lr": 7.681564187827006e-07} +{"train_loss": 0.013398146256804466, "grad_norm": 0.07024749368429184, "global_step": 50039, "epoch": 472, "lr": 7.676389885771518e-07} +{"train_loss": 0.015057982876896858, "grad_norm": 0.14278282225131989, "global_step": 50040, "epoch": 472, "lr": 7.671217313527101e-07} +{"train_loss": 0.015560725703835487, "grad_norm": 0.151427760720253, "global_step": 50041, "epoch": 472, "lr": 7.666046471111854e-07} +{"train_loss": 0.01575615629553795, "grad_norm": 0.09772511571645737, "global_step": 50042, "epoch": 472, "lr": 7.660877358544039e-07} +{"train_loss": 0.015458356589078903, "grad_norm": 0.07623692601919174, "global_step": 50043, "epoch": 472, "lr": 7.655709975841863e-07} +{"train_loss": 0.01637359708547592, "grad_norm": 0.08377708494663239, "global_step": 50044, "epoch": 472, "lr": 7.650544323023312e-07} +{"train_loss": 0.013036243617534637, "grad_norm": 0.08518704026937485, "global_step": 50045, "epoch": 472, "lr": 7.64538040010665e-07} +{"train_loss": 0.017945479601621628, "grad_norm": 0.09303393959999084, "global_step": 50046, "epoch": 472, "lr": 7.640218207110028e-07} +{"train_loss": 0.014765968546271324, "grad_norm": 0.06524448096752167, "global_step": 50047, "epoch": 472, "lr": 7.635057744051543e-07} +{"train_loss": 0.014482418075203896, "grad_norm": 0.09550490975379944, "global_step": 50048, "epoch": 472, "lr": 7.629899010949292e-07} +{"train_loss": 0.015676245093345642, "grad_norm": 0.07900639623403549, "global_step": 50049, "epoch": 472, "lr": 7.624742007821539e-07} +{"train_loss": 0.013450710102915764, "grad_norm": 0.12383247166872025, "global_step": 50050, "epoch": 472, "lr": 7.619586734686268e-07} +{"train_loss": 0.01730026677250862, "grad_norm": 0.0745241791009903, "global_step": 50051, "epoch": 472, "lr": 7.614433191561632e-07} +{"train_loss": 0.014070555567741394, "grad_norm": 0.0704273208975792, "global_step": 50052, "epoch": 472, "lr": 7.609281378465727e-07} +{"train_loss": 0.018404755741357803, "grad_norm": 0.17038211226463318, "global_step": 50053, "epoch": 472, "lr": 7.604131295416761e-07} +{"train_loss": 0.01569412276148796, "grad_norm": 0.10854911804199219, "global_step": 50054, "epoch": 472, "lr": 7.598982942432665e-07} +{"train_loss": 0.014268649742007256, "grad_norm": 0.08468137681484222, "global_step": 50055, "epoch": 472, "lr": 7.593836319531644e-07} +{"train_loss": 0.014655672013759613, "grad_norm": 0.08008453249931335, "global_step": 50056, "epoch": 472, "lr": 7.588691426731687e-07} +{"train_loss": 0.015784135088324547, "grad_norm": 0.06936807930469513, "global_step": 50057, "epoch": 472, "lr": 7.583548264050944e-07} +{"train_loss": 0.016293756663799286, "grad_norm": 0.08493932336568832, "global_step": 50058, "epoch": 472, "lr": 7.578406831507457e-07} +{"train_loss": 0.015551213175058365, "grad_norm": 0.08876772969961166, "global_step": 50059, "epoch": 472, "lr": 7.573267129119321e-07} +{"train_loss": 0.014955819584429264, "grad_norm": 0.10127075761556625, "global_step": 50060, "epoch": 472, "lr": 7.56812915690458e-07} +{"train_loss": 0.014002202078700066, "grad_norm": 0.06279850751161575, "global_step": 50061, "epoch": 472, "lr": 7.562992914881328e-07} +{"train_loss": 0.016309939324855804, "grad_norm": 0.08946015685796738, "global_step": 50062, "epoch": 472, "lr": 7.557858403067441e-07} +{"train_loss": 0.01498312409967184, "grad_norm": 0.0771072655916214, "global_step": 50063, "epoch": 472, "lr": 7.552725621481182e-07} +{"train_loss": 0.01600445620715618, "grad_norm": 0.08548015356063843, "global_step": 50064, "epoch": 472, "lr": 7.547594570140426e-07} +{"train_loss": 0.014899290166795254, "grad_norm": 0.055022694170475006, "global_step": 50065, "epoch": 472, "lr": 7.542465249063269e-07} +{"train_loss": 0.014788215979933739, "grad_norm": 0.09566164016723633, "global_step": 50066, "epoch": 472, "lr": 7.537337658267807e-07} +{"train_loss": 0.014270009472966194, "grad_norm": 0.12544305622577667, "global_step": 50067, "epoch": 472, "lr": 7.532211797771915e-07} +{"train_loss": 0.015745632350444794, "grad_norm": 0.06844360381364822, "global_step": 50068, "epoch": 472, "lr": 7.527087667593635e-07} +{"train_loss": 0.014005579054355621, "grad_norm": 0.06897835433483124, "global_step": 50069, "epoch": 472, "lr": 7.521965267751008e-07} +{"train_loss": 0.012906143441796303, "grad_norm": 0.07097051292657852, "global_step": 50070, "epoch": 472, "lr": 7.516844598262074e-07} +{"train_loss": 0.015628324821591377, "grad_norm": 0.062219250947237015, "global_step": 50071, "epoch": 472, "lr": 7.511725659144709e-07} +{"train_loss": 0.013921122997999191, "grad_norm": 0.07805511355400085, "global_step": 50072, "epoch": 472, "lr": 7.506608450417007e-07} +{"train_loss": 0.015448357909917831, "grad_norm": 0.07906409353017807, "global_step": 50073, "epoch": 472, "lr": 7.501492972096846e-07} +{"train_loss": 0.015623098239302635, "grad_norm": 0.061801809817552567, "global_step": 50074, "epoch": 472, "lr": 7.496379224202321e-07} +{"train_loss": 0.01698652096092701, "grad_norm": 0.1854746788740158, "global_step": 50075, "epoch": 472, "lr": 7.491267206751307e-07} +{"train_loss": 0.011049043387174606, "grad_norm": 0.07758381962776184, "global_step": 50076, "epoch": 472, "lr": 7.486156919761734e-07} +{"train_loss": 0.016310568898916245, "grad_norm": 0.07207492738962173, "global_step": 50077, "epoch": 472, "lr": 7.481048363251697e-07} +{"train_loss": 0.01401345431804657, "grad_norm": 0.06973407417535782, "global_step": 50078, "epoch": 472, "lr": 7.475941537239073e-07} +{"train_loss": 0.016044091433286667, "grad_norm": 0.07604249566793442, "global_step": 50079, "epoch": 472, "lr": 7.470836441741736e-07} +{"train_loss": 0.01355801522731781, "grad_norm": 0.09994139522314072, "global_step": 50080, "epoch": 472, "lr": 7.465733076777726e-07} +{"train_loss": 0.017730416730046272, "grad_norm": 0.09021145850419998, "global_step": 50081, "epoch": 472, "lr": 7.460631442364918e-07} +{"train_loss": 0.015084177255630493, "grad_norm": 0.06742126494646072, "global_step": 50082, "epoch": 472, "lr": 7.455531538521244e-07} +{"train_loss": 0.013640344142913818, "grad_norm": 0.09154078364372253, "global_step": 50083, "epoch": 472, "lr": 7.450433365264631e-07} +{"train_loss": 0.015490522608160973, "grad_norm": 0.11492393910884857, "global_step": 50084, "epoch": 472, "lr": 7.445336922613067e-07} +{"train_loss": 0.014433558098971844, "grad_norm": 0.08488429337739944, "global_step": 50085, "epoch": 472, "lr": 7.440242210584314e-07} +{"train_loss": 0.014616232365369797, "grad_norm": 0.07168019562959671, "global_step": 50086, "epoch": 472, "lr": 7.435149229196358e-07} +{"train_loss": 0.015773363411426544, "grad_norm": 0.07991810142993927, "global_step": 50087, "epoch": 472, "lr": 7.430057978467076e-07} +{"train_loss": 0.014520512893795967, "grad_norm": 0.08235300332307816, "global_step": 50088, "epoch": 472, "lr": 7.424968458414338e-07} +{"train_loss": 0.013191052712500095, "grad_norm": 0.1005367636680603, "global_step": 50089, "epoch": 472, "lr": 7.419880669056023e-07} +{"train_loss": 0.016049128025770187, "grad_norm": 0.08015088737010956, "global_step": 50090, "epoch": 472, "lr": 7.414794610410059e-07} +{"train_loss": 0.014529004693031311, "grad_norm": 0.0792042538523674, "global_step": 50091, "epoch": 472, "lr": 7.40971028249432e-07} +{"train_loss": 0.014926637522876263, "grad_norm": 0.09211653470993042, "global_step": 50092, "epoch": 472, "lr": 7.404627685326571e-07} +{"train_loss": 0.014465663582086563, "grad_norm": 0.06021415814757347, "global_step": 50093, "epoch": 472, "lr": 7.399546818924796e-07} +{"train_loss": 0.014342311769723892, "grad_norm": 0.11565785109996796, "global_step": 50094, "epoch": 472, "lr": 7.39446768330676e-07} +{"train_loss": 0.01363745890557766, "grad_norm": 0.052125249058008194, "global_step": 50095, "epoch": 472, "lr": 7.389390278490338e-07} +{"train_loss": 0.013609864749014378, "grad_norm": 0.06289556622505188, "global_step": 50096, "epoch": 472, "lr": 7.384314604493403e-07} +{"train_loss": 0.012786410748958588, "grad_norm": 0.09427813440561295, "global_step": 50097, "epoch": 472, "lr": 7.379240661333664e-07} +{"train_loss": 0.015886986628174782, "grad_norm": 0.09003785252571106, "global_step": 50098, "epoch": 472, "lr": 7.374168449029106e-07} +{"train_loss": 0.014964548870921135, "grad_norm": 0.05913778394460678, "global_step": 50099, "epoch": 472, "lr": 7.369097967597493e-07} +{"train_loss": 0.017105616629123688, "grad_norm": 0.11545772105455399, "global_step": 50100, "epoch": 472, "lr": 7.36402921705659e-07} +{"train_loss": 0.015436231158673763, "grad_norm": 0.09729525446891785, "global_step": 50101, "epoch": 472, "lr": 7.358962197424269e-07} +{"train_loss": 0.012712270952761173, "grad_norm": 0.08692767471075058, "global_step": 50102, "epoch": 472, "lr": 7.35389690871835e-07} +{"train_loss": 0.013033157214522362, "grad_norm": 0.0528329499065876, "global_step": 50103, "epoch": 472, "lr": 7.348833350956486e-07} +{"train_loss": 0.015442458912730217, "grad_norm": 0.10179571062326431, "global_step": 50104, "epoch": 472, "lr": 7.343771524156662e-07} +{"train_loss": 0.012854700908064842, "grad_norm": 0.07370809465646744, "global_step": 50105, "epoch": 472, "lr": 7.338711428336531e-07} +{"train_loss": 0.015449346043169498, "grad_norm": 0.09521126747131348, "global_step": 50106, "epoch": 472, "lr": 7.333653063513913e-07} +{"train_loss": 0.01433259341865778, "grad_norm": 0.07598954439163208, "global_step": 50107, "epoch": 472, "lr": 7.328596429706624e-07} +{"train_loss": 0.014666825532913208, "grad_norm": 0.07843363285064697, "global_step": 50108, "epoch": 472, "lr": 7.323541526932431e-07} +{"train_loss": 0.01580282673239708, "grad_norm": 0.10951564460992813, "global_step": 50109, "epoch": 472, "lr": 7.31848835520893e-07} +{"train_loss": 0.01570182852447033, "grad_norm": 0.14165638387203217, "global_step": 50110, "epoch": 472, "lr": 7.31343691455405e-07} +{"train_loss": 0.017011981457471848, "grad_norm": 0.06981360912322998, "global_step": 50111, "epoch": 472, "lr": 7.308387204985557e-07} +{"train_loss": 0.014391533099114895, "grad_norm": 0.07477107644081116, "global_step": 50112, "epoch": 472, "lr": 7.303339226521044e-07} +{"train_loss": 0.013826864771544933, "grad_norm": 0.0876232236623764, "global_step": 50113, "epoch": 472, "lr": 7.298292979178334e-07} +{"train_loss": 0.016198933124542236, "grad_norm": 0.05726569518446922, "global_step": 50114, "epoch": 472, "lr": 7.293248462975188e-07} +{"train_loss": 0.014866122044622898, "grad_norm": 0.08972986042499542, "global_step": 50115, "epoch": 472, "lr": 7.28820567792926e-07} +{"train_loss": 0.017062637954950333, "grad_norm": 0.06669192761182785, "global_step": 50116, "epoch": 472, "lr": 7.283164624058314e-07} +{"train_loss": 0.01443316601216793, "grad_norm": 0.07743310183286667, "global_step": 50117, "epoch": 472, "lr": 7.278125301380057e-07} +{"train_loss": 0.017263349145650864, "grad_norm": 0.09050293266773224, "global_step": 50118, "epoch": 472, "lr": 7.273087709912196e-07} +{"train_loss": 0.013093427754938602, "grad_norm": 0.07136652618646622, "global_step": 50119, "epoch": 472, "lr": 7.268051849672441e-07} +{"train_loss": 0.015036042779684067, "grad_norm": 0.08532023429870605, "global_step": 50120, "epoch": 472, "lr": 7.263017720678445e-07} +{"train_loss": 0.014976732432842255, "grad_norm": 0.07119026780128479, "global_step": 50121, "epoch": 472, "lr": 7.257985322947858e-07} +{"train_loss": 0.014756587333977222, "grad_norm": 0.08220620453357697, "global_step": 50122, "epoch": 472, "lr": 7.252954656498501e-07} +{"train_loss": 0.0171511247754097, "grad_norm": 0.08163761347532272, "global_step": 50123, "epoch": 472, "lr": 7.24792572134797e-07} +{"train_loss": 0.013904327526688576, "grad_norm": 0.07306196540594101, "global_step": 50124, "epoch": 472, "lr": 7.242898517513863e-07} +{"train_loss": 0.015497036278247833, "grad_norm": 0.1036401093006134, "global_step": 50125, "epoch": 472, "lr": 7.237873045014055e-07} +{"train_loss": 0.013298004865646362, "grad_norm": 0.05064726620912552, "global_step": 50126, "epoch": 472, "lr": 7.232849303865974e-07} +{"train_loss": 0.013666416518390179, "grad_norm": 0.14845936000347137, "global_step": 50127, "epoch": 472, "lr": 7.22782729408733e-07} +{"train_loss": 0.016719073057174683, "grad_norm": 0.08958559483289719, "global_step": 50128, "epoch": 472, "lr": 7.222807015695887e-07} +{"train_loss": 0.013599826954305172, "grad_norm": 0.11014453321695328, "global_step": 50129, "epoch": 472, "lr": 7.21778846870913e-07} +{"train_loss": 0.01606498286128044, "grad_norm": 0.06970667093992233, "global_step": 50130, "epoch": 472, "lr": 7.212771653144768e-07} +{"train_loss": 0.014002551324665546, "grad_norm": 0.05369674414396286, "global_step": 50131, "epoch": 472, "lr": 7.207756569020396e-07} +{"train_loss": 0.016114557161927223, "grad_norm": 0.0820685401558876, "global_step": 50132, "epoch": 472, "lr": 7.202743216353725e-07} +{"train_loss": 0.014656098559498787, "grad_norm": 0.0541217103600502, "global_step": 50133, "epoch": 472, "lr": 7.197731595162238e-07} +{"train_loss": 0.016681650653481483, "grad_norm": 0.0631684958934784, "global_step": 50134, "epoch": 472, "lr": 7.192721705463645e-07} +{"train_loss": 0.010580241680145264, "grad_norm": 0.09638173133134842, "global_step": 50135, "epoch": 472, "lr": 7.187713547275542e-07} +{"train_loss": 0.01603171043097973, "grad_norm": 0.08176660537719727, "global_step": 50136, "epoch": 472, "lr": 7.182707120615417e-07} +{"train_loss": 0.014949583852628493, "grad_norm": 0.09581111371517181, "global_step": 50137, "epoch": 472, "lr": 7.177702425500976e-07, "val_loss": 0.01997082680463791} +{"train_loss": 0.012921893037855625, "grad_norm": 0.12870293855667114, "global_step": 50138, "epoch": 473, "lr": 7.172699461949761e-07} +{"train_loss": 0.01679588109254837, "grad_norm": 0.0603933148086071, "global_step": 50139, "epoch": 473, "lr": 7.167698229979425e-07} +{"train_loss": 0.017093593254685402, "grad_norm": 0.06668663769960403, "global_step": 50140, "epoch": 473, "lr": 7.162698729607342e-07} +{"train_loss": 0.014971410855650902, "grad_norm": 0.07962268590927124, "global_step": 50141, "epoch": 473, "lr": 7.157700960851277e-07} +{"train_loss": 0.01586349867284298, "grad_norm": 0.05723600089550018, "global_step": 50142, "epoch": 473, "lr": 7.152704923728715e-07} +{"train_loss": 0.014552479609847069, "grad_norm": 0.06195372715592384, "global_step": 50143, "epoch": 473, "lr": 7.147710618257253e-07} +{"train_loss": 0.016005288809537888, "grad_norm": 0.0537402480840683, "global_step": 50144, "epoch": 473, "lr": 7.142718044454378e-07} +{"train_loss": 0.013884973712265491, "grad_norm": 0.08517616987228394, "global_step": 50145, "epoch": 473, "lr": 7.13772720233763e-07} +{"train_loss": 0.014494221657514572, "grad_norm": 0.06665079295635223, "global_step": 50146, "epoch": 473, "lr": 7.132738091924551e-07} +{"train_loss": 0.014641208574175835, "grad_norm": 0.14313726127147675, "global_step": 50147, "epoch": 473, "lr": 7.127750713232684e-07} +{"train_loss": 0.013870986178517342, "grad_norm": 0.055752791464328766, "global_step": 50148, "epoch": 473, "lr": 7.122765066279624e-07} +{"train_loss": 0.014986516907811165, "grad_norm": 0.08225180953741074, "global_step": 50149, "epoch": 473, "lr": 7.117781151082803e-07} +{"train_loss": 0.01569388061761856, "grad_norm": 0.0632060095667839, "global_step": 50150, "epoch": 473, "lr": 7.112798967659706e-07} +{"train_loss": 0.014940986409783363, "grad_norm": 0.06057091802358627, "global_step": 50151, "epoch": 473, "lr": 7.107818516027876e-07} +{"train_loss": 0.012510839849710464, "grad_norm": 0.0810413658618927, "global_step": 50152, "epoch": 473, "lr": 7.102839796204907e-07} +{"train_loss": 0.013641188852488995, "grad_norm": 0.07438773661851883, "global_step": 50153, "epoch": 473, "lr": 7.097862808208123e-07} +{"train_loss": 0.012078484520316124, "grad_norm": 0.0567227341234684, "global_step": 50154, "epoch": 473, "lr": 7.092887552055061e-07} +{"train_loss": 0.014494486153125763, "grad_norm": 0.12693719565868378, "global_step": 50155, "epoch": 473, "lr": 7.087914027763321e-07} +{"train_loss": 0.013344698585569859, "grad_norm": 0.10606064647436142, "global_step": 50156, "epoch": 473, "lr": 7.082942235350276e-07} +{"train_loss": 0.014450684189796448, "grad_norm": 0.07587843388319016, "global_step": 50157, "epoch": 473, "lr": 7.077972174833358e-07} +{"train_loss": 0.015251724049448967, "grad_norm": 0.12408456951379776, "global_step": 50158, "epoch": 473, "lr": 7.073003846230108e-07} +{"train_loss": 0.01352771557867527, "grad_norm": 0.06782099604606628, "global_step": 50159, "epoch": 473, "lr": 7.068037249557957e-07} +{"train_loss": 0.013280652463436127, "grad_norm": 0.07360319048166275, "global_step": 50160, "epoch": 473, "lr": 7.063072384834335e-07} +{"train_loss": 0.014334352687001228, "grad_norm": 0.08612438291311264, "global_step": 50161, "epoch": 473, "lr": 7.058109252076728e-07} +{"train_loss": 0.019035885110497475, "grad_norm": 0.09682545065879822, "global_step": 50162, "epoch": 473, "lr": 7.053147851302566e-07} +{"train_loss": 0.01495364960283041, "grad_norm": 0.08233758062124252, "global_step": 50163, "epoch": 473, "lr": 7.048188182529225e-07} +{"train_loss": 0.014207877218723297, "grad_norm": 0.10628723353147507, "global_step": 50164, "epoch": 473, "lr": 7.04323024577419e-07} +{"train_loss": 0.01715593971312046, "grad_norm": 0.0807599350810051, "global_step": 50165, "epoch": 473, "lr": 7.038274041054838e-07} +{"train_loss": 0.014791260473430157, "grad_norm": 0.05833083391189575, "global_step": 50166, "epoch": 473, "lr": 7.033319568388652e-07} +{"train_loss": 0.015200736001133919, "grad_norm": 0.17391152679920197, "global_step": 50167, "epoch": 473, "lr": 7.02836682779301e-07} +{"train_loss": 0.01472003199160099, "grad_norm": 0.05497393012046814, "global_step": 50168, "epoch": 473, "lr": 7.023415819285228e-07} +{"train_loss": 0.016804739832878113, "grad_norm": 0.14320604503154755, "global_step": 50169, "epoch": 473, "lr": 7.018466542882796e-07} +{"train_loss": 0.01769646257162094, "grad_norm": 0.14474482834339142, "global_step": 50170, "epoch": 473, "lr": 7.013518998603086e-07} +{"train_loss": 0.014058944769203663, "grad_norm": 0.08974092453718185, "global_step": 50171, "epoch": 473, "lr": 7.008573186463474e-07} +{"train_loss": 0.01820879988372326, "grad_norm": 0.08368747681379318, "global_step": 50172, "epoch": 473, "lr": 7.00362910648139e-07} +{"train_loss": 0.01550927385687828, "grad_norm": 0.08156527578830719, "global_step": 50173, "epoch": 473, "lr": 6.9986867586741e-07} +{"train_loss": 0.01619117520749569, "grad_norm": 0.11719592660665512, "global_step": 50174, "epoch": 473, "lr": 6.993746143059033e-07} +{"train_loss": 0.015896741300821304, "grad_norm": 0.12122488021850586, "global_step": 50175, "epoch": 473, "lr": 6.988807259653507e-07} +{"train_loss": 0.015540098771452904, "grad_norm": 0.07247911393642426, "global_step": 50176, "epoch": 473, "lr": 6.983870108474955e-07} +{"train_loss": 0.014795254915952682, "grad_norm": 0.06786169856786728, "global_step": 50177, "epoch": 473, "lr": 6.978934689540695e-07} +{"train_loss": 0.012252441607415676, "grad_norm": 0.10131316632032394, "global_step": 50178, "epoch": 473, "lr": 6.974001002867991e-07} +{"train_loss": 0.015163617208600044, "grad_norm": 0.07153801620006561, "global_step": 50179, "epoch": 473, "lr": 6.969069048474275e-07} +{"train_loss": 0.011587975546717644, "grad_norm": 0.08589664846658707, "global_step": 50180, "epoch": 473, "lr": 6.964138826376865e-07} +{"train_loss": 0.015642132610082626, "grad_norm": 0.1301083266735077, "global_step": 50181, "epoch": 473, "lr": 6.959210336592969e-07} +{"train_loss": 0.01438157819211483, "grad_norm": 0.06210140138864517, "global_step": 50182, "epoch": 473, "lr": 6.954283579140019e-07} +{"train_loss": 0.015345389023423195, "grad_norm": 0.10946907848119736, "global_step": 50183, "epoch": 473, "lr": 6.949358554035334e-07} +{"train_loss": 0.013598188757896423, "grad_norm": 0.12438367307186127, "global_step": 50184, "epoch": 473, "lr": 6.944435261296178e-07} +{"train_loss": 0.01327458955347538, "grad_norm": 0.11396894603967667, "global_step": 50185, "epoch": 473, "lr": 6.939513700939815e-07} +{"train_loss": 0.016185056418180466, "grad_norm": 0.1465708166360855, "global_step": 50186, "epoch": 473, "lr": 6.934593872983619e-07} +{"train_loss": 0.015878556296229362, "grad_norm": 0.07643083482980728, "global_step": 50187, "epoch": 473, "lr": 6.929675777444799e-07} +{"train_loss": 0.013707929290831089, "grad_norm": 0.0802551880478859, "global_step": 50188, "epoch": 473, "lr": 6.924759414340676e-07} +{"train_loss": 0.015274515375494957, "grad_norm": 0.08492249995470047, "global_step": 50189, "epoch": 473, "lr": 6.919844783688512e-07} +{"train_loss": 0.014403070323169231, "grad_norm": 0.09003221243619919, "global_step": 50190, "epoch": 473, "lr": 6.914931885505627e-07} +{"train_loss": 0.017270684242248535, "grad_norm": 0.1043286994099617, "global_step": 50191, "epoch": 473, "lr": 6.910020719809173e-07} +{"train_loss": 0.01525968499481678, "grad_norm": 0.08030740171670914, "global_step": 50192, "epoch": 473, "lr": 6.905111286616472e-07} +{"train_loss": 0.015425747260451317, "grad_norm": 0.10774748027324677, "global_step": 50193, "epoch": 473, "lr": 6.900203585944786e-07} +{"train_loss": 0.014704987406730652, "grad_norm": 0.10953472554683685, "global_step": 50194, "epoch": 473, "lr": 6.895297617811325e-07} +{"train_loss": 0.015224523842334747, "grad_norm": 0.07900908589363098, "global_step": 50195, "epoch": 473, "lr": 6.890393382233351e-07} +{"train_loss": 0.013636764138936996, "grad_norm": 0.07227955758571625, "global_step": 50196, "epoch": 473, "lr": 6.885490879228074e-07} +{"train_loss": 0.012213502079248428, "grad_norm": 0.055432870984077454, "global_step": 50197, "epoch": 473, "lr": 6.880590108812756e-07} +{"train_loss": 0.013038957491517067, "grad_norm": 0.10984785854816437, "global_step": 50198, "epoch": 473, "lr": 6.875691071004608e-07} +{"train_loss": 0.011866504326462746, "grad_norm": 0.06442463397979736, "global_step": 50199, "epoch": 473, "lr": 6.870793765820782e-07} +{"train_loss": 0.01515834778547287, "grad_norm": 0.09009509533643723, "global_step": 50200, "epoch": 473, "lr": 6.865898193278542e-07} +{"train_loss": 0.014049232006072998, "grad_norm": 0.08113056421279907, "global_step": 50201, "epoch": 473, "lr": 6.861004353395095e-07} +{"train_loss": 0.016767822206020355, "grad_norm": 0.10465311259031296, "global_step": 50202, "epoch": 473, "lr": 6.856112246187652e-07} +{"train_loss": 0.015356382355093956, "grad_norm": 0.07889930158853531, "global_step": 50203, "epoch": 473, "lr": 6.851221871673308e-07} +{"train_loss": 0.01786927506327629, "grad_norm": 0.13427434861660004, "global_step": 50204, "epoch": 473, "lr": 6.846333229869329e-07} +{"train_loss": 0.015008477494120598, "grad_norm": 0.09808716922998428, "global_step": 50205, "epoch": 473, "lr": 6.841446320792811e-07} +{"train_loss": 0.0147703243419528, "grad_norm": 0.09382703900337219, "global_step": 50206, "epoch": 473, "lr": 6.836561144461073e-07} +{"train_loss": 0.011441830545663834, "grad_norm": 0.11524980515241623, "global_step": 50207, "epoch": 473, "lr": 6.83167770089116e-07} +{"train_loss": 0.014634122140705585, "grad_norm": 0.09304069727659225, "global_step": 50208, "epoch": 473, "lr": 6.826795990100221e-07} +{"train_loss": 0.016311023384332657, "grad_norm": 0.07788160443305969, "global_step": 50209, "epoch": 473, "lr": 6.821916012105467e-07} +{"train_loss": 0.015997707843780518, "grad_norm": 0.10656227171421051, "global_step": 50210, "epoch": 473, "lr": 6.817037766924051e-07} +{"train_loss": 0.013657537288963795, "grad_norm": 0.058806344866752625, "global_step": 50211, "epoch": 473, "lr": 6.812161254573068e-07} +{"train_loss": 0.016460122540593147, "grad_norm": 0.06314875930547714, "global_step": 50212, "epoch": 473, "lr": 6.807286475069619e-07} +{"train_loss": 0.014327555894851685, "grad_norm": 0.09029734879732132, "global_step": 50213, "epoch": 473, "lr": 6.802413428430965e-07} +{"train_loss": 0.01470896229147911, "grad_norm": 0.06197977438569069, "global_step": 50214, "epoch": 473, "lr": 6.797542114674094e-07} +{"train_loss": 0.014246813952922821, "grad_norm": 0.10336059331893921, "global_step": 50215, "epoch": 473, "lr": 6.792672533816214e-07} +{"train_loss": 0.015037594363093376, "grad_norm": 0.09090578556060791, "global_step": 50216, "epoch": 473, "lr": 6.787804685874311e-07} +{"train_loss": 0.017724785953760147, "grad_norm": 0.07303646951913834, "global_step": 50217, "epoch": 473, "lr": 6.78293857086565e-07} +{"train_loss": 0.014691216871142387, "grad_norm": 0.07761278003454208, "global_step": 50218, "epoch": 473, "lr": 6.778074188807216e-07} +{"train_loss": 0.014216594398021698, "grad_norm": 0.10600662231445312, "global_step": 50219, "epoch": 473, "lr": 6.773211539716107e-07} +{"train_loss": 0.016674751415848732, "grad_norm": 0.07353553175926208, "global_step": 50220, "epoch": 473, "lr": 6.768350623609476e-07} +{"train_loss": 0.015200920403003693, "grad_norm": 0.08260391652584076, "global_step": 50221, "epoch": 473, "lr": 6.763491440504366e-07} +{"train_loss": 0.01612507738173008, "grad_norm": 0.07325295358896255, "global_step": 50222, "epoch": 473, "lr": 6.758633990417818e-07} +{"train_loss": 0.015107800252735615, "grad_norm": 0.14933966100215912, "global_step": 50223, "epoch": 473, "lr": 6.753778273366873e-07} +{"train_loss": 0.01565360650420189, "grad_norm": 0.07711444050073624, "global_step": 50224, "epoch": 473, "lr": 6.748924289368741e-07} +{"train_loss": 0.014645805582404137, "grad_norm": 0.09221651405096054, "global_step": 50225, "epoch": 473, "lr": 6.744072038440352e-07} +{"train_loss": 0.01540263369679451, "grad_norm": 0.07758508622646332, "global_step": 50226, "epoch": 473, "lr": 6.739221520598749e-07} +{"train_loss": 0.01605807989835739, "grad_norm": 0.086187444627285, "global_step": 50227, "epoch": 473, "lr": 6.734372735861083e-07} +{"train_loss": 0.015409227460622787, "grad_norm": 0.08755313605070114, "global_step": 50228, "epoch": 473, "lr": 6.729525684244231e-07} +{"train_loss": 0.01387137733399868, "grad_norm": 0.0709863230586052, "global_step": 50229, "epoch": 473, "lr": 6.724680365765346e-07} +{"train_loss": 0.014913158491253853, "grad_norm": 0.13121606409549713, "global_step": 50230, "epoch": 473, "lr": 6.719836780441413e-07} +{"train_loss": 0.016184810549020767, "grad_norm": 0.10302332788705826, "global_step": 50231, "epoch": 473, "lr": 6.714994928289475e-07} +{"train_loss": 0.0158919058740139, "grad_norm": 0.09029508382081985, "global_step": 50232, "epoch": 473, "lr": 6.710154809326463e-07} +{"train_loss": 0.014428069815039635, "grad_norm": 0.05611845850944519, "global_step": 50233, "epoch": 473, "lr": 6.705316423569474e-07} +{"train_loss": 0.014738840982317924, "grad_norm": 0.09106788784265518, "global_step": 50234, "epoch": 473, "lr": 6.700479771035495e-07} +{"train_loss": 0.01751076430082321, "grad_norm": 0.07864274829626083, "global_step": 50235, "epoch": 473, "lr": 6.695644851741456e-07} +{"train_loss": 0.01831194758415222, "grad_norm": 0.06986535340547562, "global_step": 50236, "epoch": 473, "lr": 6.6908116657044e-07} +{"train_loss": 0.01360725611448288, "grad_norm": 0.06077153980731964, "global_step": 50237, "epoch": 473, "lr": 6.685980212941312e-07} +{"train_loss": 0.0154756810516119, "grad_norm": 0.10402297228574753, "global_step": 50238, "epoch": 473, "lr": 6.68115049346918e-07} +{"train_loss": 0.014824659563601017, "grad_norm": 0.05352139472961426, "global_step": 50239, "epoch": 473, "lr": 6.676322507304877e-07} +{"train_loss": 0.011618772521615028, "grad_norm": 0.06419765204191208, "global_step": 50240, "epoch": 473, "lr": 6.671496254465448e-07} +{"train_loss": 0.014763127081096172, "grad_norm": 0.06526979804039001, "global_step": 50241, "epoch": 473, "lr": 6.666671734967877e-07} +{"train_loss": 0.01335253193974495, "grad_norm": 0.06356622278690338, "global_step": 50242, "epoch": 473, "lr": 6.661848948829042e-07} +{"train_loss": 0.014947512564864361, "grad_norm": 0.0991915687918663, "global_step": 50243, "epoch": 473, "lr": 6.657027896065982e-07, "val_loss": 0.019077515229582787} +{"train_loss": 0.01491361204534769, "grad_norm": 0.10988742113113403, "global_step": 50244, "epoch": 474, "lr": 6.652208576695574e-07} +{"train_loss": 0.015189167112112045, "grad_norm": 0.13470733165740967, "global_step": 50245, "epoch": 474, "lr": 6.647390990734692e-07} +{"train_loss": 0.013771376572549343, "grad_norm": 0.0575019046664238, "global_step": 50246, "epoch": 474, "lr": 6.642575138200325e-07} +{"train_loss": 0.01571531593799591, "grad_norm": 0.06526220589876175, "global_step": 50247, "epoch": 474, "lr": 6.637761019109456e-07} +{"train_loss": 0.014728428795933723, "grad_norm": 0.07132468372583389, "global_step": 50248, "epoch": 474, "lr": 6.632948633478853e-07} +{"train_loss": 0.01394298579543829, "grad_norm": 0.07287822663784027, "global_step": 50249, "epoch": 474, "lr": 6.628137981325611e-07} +{"train_loss": 0.01543539110571146, "grad_norm": 0.12504614889621735, "global_step": 50250, "epoch": 474, "lr": 6.623329062666439e-07} +{"train_loss": 0.016970403492450714, "grad_norm": 0.08346953988075256, "global_step": 50251, "epoch": 474, "lr": 6.61852187751838e-07} +{"train_loss": 0.014927685260772705, "grad_norm": 0.18866589665412903, "global_step": 50252, "epoch": 474, "lr": 6.613716425898253e-07} +{"train_loss": 0.015690337866544724, "grad_norm": 0.07366820424795151, "global_step": 50253, "epoch": 474, "lr": 6.608912707822934e-07} +{"train_loss": 0.015344901010394096, "grad_norm": 0.09390687197446823, "global_step": 50254, "epoch": 474, "lr": 6.604110723309353e-07} +{"train_loss": 0.01338367909193039, "grad_norm": 0.06613335013389587, "global_step": 50255, "epoch": 474, "lr": 6.599310472374387e-07} +{"train_loss": 0.016331546008586884, "grad_norm": 0.10181203484535217, "global_step": 50256, "epoch": 474, "lr": 6.594511955034799e-07} +{"train_loss": 0.013878883793950081, "grad_norm": 0.0680290088057518, "global_step": 50257, "epoch": 474, "lr": 6.58971517130752e-07} +{"train_loss": 0.016915788874030113, "grad_norm": 0.1709972769021988, "global_step": 50258, "epoch": 474, "lr": 6.584920121209481e-07} +{"train_loss": 0.016518330201506615, "grad_norm": 0.06408568471670151, "global_step": 50259, "epoch": 474, "lr": 6.580126804757392e-07} +{"train_loss": 0.015226570889353752, "grad_norm": 0.05891101807355881, "global_step": 50260, "epoch": 474, "lr": 6.575335221968127e-07} +{"train_loss": 0.017339598387479782, "grad_norm": 0.09276984632015228, "global_step": 50261, "epoch": 474, "lr": 6.570545372858672e-07} +{"train_loss": 0.015446952544152737, "grad_norm": 0.13815832138061523, "global_step": 50262, "epoch": 474, "lr": 6.565757257445626e-07} +{"train_loss": 0.014531737193465233, "grad_norm": 0.08252883702516556, "global_step": 50263, "epoch": 474, "lr": 6.56097087574592e-07} +{"train_loss": 0.013416657224297523, "grad_norm": 0.08617383986711502, "global_step": 50264, "epoch": 474, "lr": 6.556186227776373e-07} +{"train_loss": 0.016243286430835724, "grad_norm": 0.0720977634191513, "global_step": 50265, "epoch": 474, "lr": 6.551403313553861e-07} +{"train_loss": 0.01697147637605667, "grad_norm": 0.07435322552919388, "global_step": 50266, "epoch": 474, "lr": 6.546622133095093e-07} +{"train_loss": 0.014668677002191544, "grad_norm": 0.06249954178929329, "global_step": 50267, "epoch": 474, "lr": 6.541842686416888e-07} +{"train_loss": 0.013957146555185318, "grad_norm": 0.07670778781175613, "global_step": 50268, "epoch": 474, "lr": 6.537064973536067e-07} +{"train_loss": 0.014904603362083435, "grad_norm": 0.08822644501924515, "global_step": 50269, "epoch": 474, "lr": 6.532288994469338e-07} +{"train_loss": 0.012091794982552528, "grad_norm": 0.07215826958417892, "global_step": 50270, "epoch": 474, "lr": 6.527514749233577e-07} +{"train_loss": 0.01623142696917057, "grad_norm": 0.0652727261185646, "global_step": 50271, "epoch": 474, "lr": 6.522742237845547e-07} +{"train_loss": 0.016902059316635132, "grad_norm": 0.06864006817340851, "global_step": 50272, "epoch": 474, "lr": 6.517971460322014e-07} +{"train_loss": 0.014949055388569832, "grad_norm": 0.05641145259141922, "global_step": 50273, "epoch": 474, "lr": 6.513202416679687e-07} +{"train_loss": 0.018916837871074677, "grad_norm": 0.083096444606781, "global_step": 50274, "epoch": 474, "lr": 6.508435106935328e-07} +{"train_loss": 0.016555488109588623, "grad_norm": 0.08452068269252777, "global_step": 50275, "epoch": 474, "lr": 6.50366953110576e-07} +{"train_loss": 0.015782874077558517, "grad_norm": 0.1174619197845459, "global_step": 50276, "epoch": 474, "lr": 6.498905689207635e-07} +{"train_loss": 0.0161966010928154, "grad_norm": 0.09425649046897888, "global_step": 50277, "epoch": 474, "lr": 6.494143581257717e-07} +{"train_loss": 0.016835413873195648, "grad_norm": 0.06368359178304672, "global_step": 50278, "epoch": 474, "lr": 6.489383207272825e-07} +{"train_loss": 0.016686419025063515, "grad_norm": 0.09365349262952805, "global_step": 50279, "epoch": 474, "lr": 6.484624567269615e-07} +{"train_loss": 0.016849009320139885, "grad_norm": 0.10136353969573975, "global_step": 50280, "epoch": 474, "lr": 6.479867661264793e-07} +{"train_loss": 0.013039818964898586, "grad_norm": 0.11765977740287781, "global_step": 50281, "epoch": 474, "lr": 6.47511248927507e-07} +{"train_loss": 0.018278498202562332, "grad_norm": 0.11314954608678818, "global_step": 50282, "epoch": 474, "lr": 6.47035905131721e-07} +{"train_loss": 0.013789213262498379, "grad_norm": 0.10648933798074722, "global_step": 50283, "epoch": 474, "lr": 6.46560734740781e-07} +{"train_loss": 0.013166943565011024, "grad_norm": 0.07031406462192535, "global_step": 50284, "epoch": 474, "lr": 6.460857377563745e-07} +{"train_loss": 0.013302674517035484, "grad_norm": 0.11860091984272003, "global_step": 50285, "epoch": 474, "lr": 6.456109141801503e-07} +{"train_loss": 0.011980907991528511, "grad_norm": 0.08265814185142517, "global_step": 50286, "epoch": 474, "lr": 6.451362640137904e-07} +{"train_loss": 0.017551785334944725, "grad_norm": 0.12236042320728302, "global_step": 50287, "epoch": 474, "lr": 6.446617872589544e-07} +{"train_loss": 0.016906023025512695, "grad_norm": 0.0897601917386055, "global_step": 50288, "epoch": 474, "lr": 6.441874839173134e-07} +{"train_loss": 0.014544742181897163, "grad_norm": 0.08974325656890869, "global_step": 50289, "epoch": 474, "lr": 6.437133539905326e-07} +{"train_loss": 0.015583905391395092, "grad_norm": 0.1314448118209839, "global_step": 50290, "epoch": 474, "lr": 6.43239397480283e-07} +{"train_loss": 0.016550295054912567, "grad_norm": 0.08321421593427658, "global_step": 50291, "epoch": 474, "lr": 6.427656143882188e-07} +{"train_loss": 0.014615565538406372, "grad_norm": 0.09205405414104462, "global_step": 50292, "epoch": 474, "lr": 6.422920047160164e-07} +{"train_loss": 0.013200094923377037, "grad_norm": 0.06103721261024475, "global_step": 50293, "epoch": 474, "lr": 6.418185684653355e-07} +{"train_loss": 0.01438227854669094, "grad_norm": 0.06484505534172058, "global_step": 50294, "epoch": 474, "lr": 6.413453056378361e-07} +{"train_loss": 0.014265783131122589, "grad_norm": 0.10150128602981567, "global_step": 50295, "epoch": 474, "lr": 6.408722162351832e-07} +{"train_loss": 0.012399358674883842, "grad_norm": 0.06034481152892113, "global_step": 50296, "epoch": 474, "lr": 6.403993002590425e-07} +{"train_loss": 0.017078205943107605, "grad_norm": 0.10727988928556442, "global_step": 50297, "epoch": 474, "lr": 6.399265577110681e-07} +{"train_loss": 0.016159681603312492, "grad_norm": 0.06571783870458603, "global_step": 50298, "epoch": 474, "lr": 6.394539885929307e-07} +{"train_loss": 0.014715158380568027, "grad_norm": 0.06641115248203278, "global_step": 50299, "epoch": 474, "lr": 6.389815929062848e-07} +{"train_loss": 0.016698289662599564, "grad_norm": 0.0727631151676178, "global_step": 50300, "epoch": 474, "lr": 6.385093706527901e-07} +{"train_loss": 0.01652960106730461, "grad_norm": 0.11201059073209763, "global_step": 50301, "epoch": 474, "lr": 6.380373218341007e-07} +{"train_loss": 0.013677661307156086, "grad_norm": 0.11224853992462158, "global_step": 50302, "epoch": 474, "lr": 6.375654464518932e-07} +{"train_loss": 0.014376037754118443, "grad_norm": 0.07454527169466019, "global_step": 50303, "epoch": 474, "lr": 6.370937445078051e-07} +{"train_loss": 0.013556556776165962, "grad_norm": 0.08287820965051651, "global_step": 50304, "epoch": 474, "lr": 6.366222160035018e-07} +{"train_loss": 0.017389802262187004, "grad_norm": 0.08976113796234131, "global_step": 50305, "epoch": 474, "lr": 6.361508609406486e-07} +{"train_loss": 0.013271142728626728, "grad_norm": 0.06162893772125244, "global_step": 50306, "epoch": 474, "lr": 6.356796793208831e-07} +{"train_loss": 0.013885150663554668, "grad_norm": 0.06591734290122986, "global_step": 50307, "epoch": 474, "lr": 6.35208671145876e-07} +{"train_loss": 0.013734494335949421, "grad_norm": 0.08102058619260788, "global_step": 50308, "epoch": 474, "lr": 6.347378364172817e-07} +{"train_loss": 0.014901133254170418, "grad_norm": 0.14716434478759766, "global_step": 50309, "epoch": 474, "lr": 6.342671751367491e-07} +{"train_loss": 0.013215345330536366, "grad_norm": 0.11940605193376541, "global_step": 50310, "epoch": 474, "lr": 6.337966873059264e-07} +{"train_loss": 0.017081430181860924, "grad_norm": 0.11416938900947571, "global_step": 50311, "epoch": 474, "lr": 6.333263729264738e-07} +{"train_loss": 0.014559955336153507, "grad_norm": 0.06072657182812691, "global_step": 50312, "epoch": 474, "lr": 6.328562320000509e-07} +{"train_loss": 0.01669343188405037, "grad_norm": 0.10189411789178848, "global_step": 50313, "epoch": 474, "lr": 6.323862645282952e-07} +{"train_loss": 0.011773886159062386, "grad_norm": 0.07159806787967682, "global_step": 50314, "epoch": 474, "lr": 6.319164705128667e-07} +{"train_loss": 0.013656668365001678, "grad_norm": 0.06223561242222786, "global_step": 50315, "epoch": 474, "lr": 6.314468499554138e-07} +{"train_loss": 0.015740536153316498, "grad_norm": 0.13695433735847473, "global_step": 50316, "epoch": 474, "lr": 6.30977402857591e-07} +{"train_loss": 0.01599055528640747, "grad_norm": 0.0953577384352684, "global_step": 50317, "epoch": 474, "lr": 6.305081292210413e-07} +{"train_loss": 0.01613837294280529, "grad_norm": 0.12771454453468323, "global_step": 50318, "epoch": 474, "lr": 6.300390290474078e-07} +{"train_loss": 0.01595735363662243, "grad_norm": 0.08331002295017242, "global_step": 50319, "epoch": 474, "lr": 6.29570102338356e-07} +{"train_loss": 0.014614942483603954, "grad_norm": 0.14571526646614075, "global_step": 50320, "epoch": 474, "lr": 6.291013490955289e-07} +{"train_loss": 0.013880433514714241, "grad_norm": 0.10347706824541092, "global_step": 50321, "epoch": 474, "lr": 6.286327693205585e-07} +{"train_loss": 0.013752510771155357, "grad_norm": 0.10578539222478867, "global_step": 50322, "epoch": 474, "lr": 6.281643630151046e-07} +{"train_loss": 0.013455105014145374, "grad_norm": 0.10445582866668701, "global_step": 50323, "epoch": 474, "lr": 6.276961301808104e-07} +{"train_loss": 0.016060952097177505, "grad_norm": 0.07830725610256195, "global_step": 50324, "epoch": 474, "lr": 6.27228070819319e-07} +{"train_loss": 0.01191885955631733, "grad_norm": 0.07601652294397354, "global_step": 50325, "epoch": 474, "lr": 6.26760184932279e-07} +{"train_loss": 0.014001859351992607, "grad_norm": 0.06617553532123566, "global_step": 50326, "epoch": 474, "lr": 6.262924725213337e-07} +{"train_loss": 0.016817796975374222, "grad_norm": 0.06505008041858673, "global_step": 50327, "epoch": 474, "lr": 6.258249335881205e-07} +{"train_loss": 0.01208348385989666, "grad_norm": 0.0960778072476387, "global_step": 50328, "epoch": 474, "lr": 6.253575681342827e-07} +{"train_loss": 0.01506087277084589, "grad_norm": 0.0668601468205452, "global_step": 50329, "epoch": 474, "lr": 6.248903761614744e-07} +{"train_loss": 0.014577513560652733, "grad_norm": 0.10915354639291763, "global_step": 50330, "epoch": 474, "lr": 6.244233576713221e-07} +{"train_loss": 0.013234414160251617, "grad_norm": 0.06930748373270035, "global_step": 50331, "epoch": 474, "lr": 6.239565126654801e-07} +{"train_loss": 0.014524132013320923, "grad_norm": 0.11666297167539597, "global_step": 50332, "epoch": 474, "lr": 6.234898411455747e-07} +{"train_loss": 0.014983869157731533, "grad_norm": 0.0630364790558815, "global_step": 50333, "epoch": 474, "lr": 6.230233431132548e-07} +{"train_loss": 0.01563527062535286, "grad_norm": 0.08905564993619919, "global_step": 50334, "epoch": 474, "lr": 6.225570185701579e-07} +{"train_loss": 0.013970806263387203, "grad_norm": 0.08643580228090286, "global_step": 50335, "epoch": 474, "lr": 6.220908675179215e-07} +{"train_loss": 0.014135217294096947, "grad_norm": 0.0695747658610344, "global_step": 50336, "epoch": 474, "lr": 6.21624889958189e-07} +{"train_loss": 0.01416788436472416, "grad_norm": 0.05846618860960007, "global_step": 50337, "epoch": 474, "lr": 6.211590858925864e-07} +{"train_loss": 0.015804782509803772, "grad_norm": 0.09007467329502106, "global_step": 50338, "epoch": 474, "lr": 6.206934553227573e-07} +{"train_loss": 0.012666655704379082, "grad_norm": 0.06056083366274834, "global_step": 50339, "epoch": 474, "lr": 6.202279982503389e-07} +{"train_loss": 0.013744687661528587, "grad_norm": 0.11130594462156296, "global_step": 50340, "epoch": 474, "lr": 6.197627146769636e-07} +{"train_loss": 0.015296990051865578, "grad_norm": 0.1010449081659317, "global_step": 50341, "epoch": 474, "lr": 6.192976046042687e-07} +{"train_loss": 0.015273773111402988, "grad_norm": 0.14540570974349976, "global_step": 50342, "epoch": 474, "lr": 6.188326680338807e-07} +{"train_loss": 0.01431288756430149, "grad_norm": 0.158689484000206, "global_step": 50343, "epoch": 474, "lr": 6.183679049674484e-07} +{"train_loss": 0.013826579786837101, "grad_norm": 0.05380633845925331, "global_step": 50344, "epoch": 474, "lr": 6.179033154065928e-07} +{"train_loss": 0.015163589268922806, "grad_norm": 0.11160428822040558, "global_step": 50345, "epoch": 474, "lr": 6.174388993529456e-07} +{"train_loss": 0.012847475707530975, "grad_norm": 0.07778825610876083, "global_step": 50346, "epoch": 474, "lr": 6.169746568081447e-07} +{"train_loss": 0.013231223449110985, "grad_norm": 0.05623089522123337, "global_step": 50347, "epoch": 474, "lr": 6.16510587773822e-07} +{"train_loss": 0.016320012509822845, "grad_norm": 0.06147710978984833, "global_step": 50348, "epoch": 474, "lr": 6.160466922515984e-07} +{"train_loss": 0.014945614959974334, "grad_norm": 0.09796342998743057, "global_step": 50349, "epoch": 474, "lr": 6.15582970243117e-07, "val_loss": 0.019206462427973747} +{"train_loss": 0.0160692036151886, "grad_norm": 0.059280507266521454, "global_step": 50350, "epoch": 475, "lr": 6.15119421749999e-07} +{"train_loss": 0.01701105758547783, "grad_norm": 0.07603399455547333, "global_step": 50351, "epoch": 475, "lr": 6.146560467738649e-07} +{"train_loss": 0.014665933325886726, "grad_norm": 0.08083584904670715, "global_step": 50352, "epoch": 475, "lr": 6.141928453163581e-07} +{"train_loss": 0.01352935004979372, "grad_norm": 0.06657739728689194, "global_step": 50353, "epoch": 475, "lr": 6.137298173791051e-07} +{"train_loss": 0.018912000581622124, "grad_norm": 0.08500190079212189, "global_step": 50354, "epoch": 475, "lr": 6.132669629637267e-07} +{"train_loss": 0.014148898422718048, "grad_norm": 0.14448633790016174, "global_step": 50355, "epoch": 475, "lr": 6.128042820718494e-07} +{"train_loss": 0.015011962503194809, "grad_norm": 0.07871372997760773, "global_step": 50356, "epoch": 475, "lr": 6.123417747050941e-07} +{"train_loss": 0.01421896182000637, "grad_norm": 0.0737995058298111, "global_step": 50357, "epoch": 475, "lr": 6.118794408650985e-07} +{"train_loss": 0.0173049233853817, "grad_norm": 0.09817212074995041, "global_step": 50358, "epoch": 475, "lr": 6.114172805534779e-07} +{"train_loss": 0.012424401938915253, "grad_norm": 0.09316445142030716, "global_step": 50359, "epoch": 475, "lr": 6.109552937718588e-07} +{"train_loss": 0.013061696663498878, "grad_norm": 0.08651771396398544, "global_step": 50360, "epoch": 475, "lr": 6.104934805218621e-07} +{"train_loss": 0.01519585307687521, "grad_norm": 0.08010535687208176, "global_step": 50361, "epoch": 475, "lr": 6.100318408051198e-07} +{"train_loss": 0.015126224607229233, "grad_norm": 0.11214561760425568, "global_step": 50362, "epoch": 475, "lr": 6.095703746232362e-07} +{"train_loss": 0.013723695650696754, "grad_norm": 0.09142006933689117, "global_step": 50363, "epoch": 475, "lr": 6.091090819778545e-07} +{"train_loss": 0.01494397222995758, "grad_norm": 0.07559376955032349, "global_step": 50364, "epoch": 475, "lr": 6.086479628705733e-07} +{"train_loss": 0.01557808555662632, "grad_norm": 0.23537924885749817, "global_step": 50365, "epoch": 475, "lr": 6.081870173030246e-07} +{"train_loss": 0.015458378940820694, "grad_norm": 0.06393855065107346, "global_step": 50366, "epoch": 475, "lr": 6.077262452768296e-07} +{"train_loss": 0.015140334144234657, "grad_norm": 0.0661911591887474, "global_step": 50367, "epoch": 475, "lr": 6.072656467936089e-07} +{"train_loss": 0.01406944915652275, "grad_norm": 0.05497325584292412, "global_step": 50368, "epoch": 475, "lr": 6.068052218549724e-07} +{"train_loss": 0.015768008306622505, "grad_norm": 0.0796842947602272, "global_step": 50369, "epoch": 475, "lr": 6.063449704625357e-07} +{"train_loss": 0.016999289393424988, "grad_norm": 0.10733410716056824, "global_step": 50370, "epoch": 475, "lr": 6.058848926179306e-07} +{"train_loss": 0.01621224917471409, "grad_norm": 0.07300934195518494, "global_step": 50371, "epoch": 475, "lr": 6.054249883227614e-07} +{"train_loss": 0.0154335405677557, "grad_norm": 0.08536368608474731, "global_step": 50372, "epoch": 475, "lr": 6.049652575786491e-07} +{"train_loss": 0.012967094779014587, "grad_norm": 0.14711971580982208, "global_step": 50373, "epoch": 475, "lr": 6.04505700387209e-07} +{"train_loss": 0.014724497683346272, "grad_norm": 0.14118720591068268, "global_step": 50374, "epoch": 475, "lr": 6.04046316750051e-07} +{"train_loss": 0.013263242319226265, "grad_norm": 0.09155826270580292, "global_step": 50375, "epoch": 475, "lr": 6.035871066687904e-07} +{"train_loss": 0.016620974987745285, "grad_norm": 0.08745992183685303, "global_step": 50376, "epoch": 475, "lr": 6.031280701450481e-07} +{"train_loss": 0.014713268727064133, "grad_norm": 0.0688362568616867, "global_step": 50377, "epoch": 475, "lr": 6.026692071804286e-07} +{"train_loss": 0.013418282382190228, "grad_norm": 0.08881861716508865, "global_step": 50378, "epoch": 475, "lr": 6.022105177765525e-07} +{"train_loss": 0.014957096427679062, "grad_norm": 0.06858542561531067, "global_step": 50379, "epoch": 475, "lr": 6.017520019350187e-07} +{"train_loss": 0.013617363758385181, "grad_norm": 0.09355966001749039, "global_step": 50380, "epoch": 475, "lr": 6.012936596574481e-07} +{"train_loss": 0.014368733391165733, "grad_norm": 0.057999566197395325, "global_step": 50381, "epoch": 475, "lr": 6.008354909454506e-07} +{"train_loss": 0.015936467796564102, "grad_norm": 0.1855725347995758, "global_step": 50382, "epoch": 475, "lr": 6.003774958006303e-07} +{"train_loss": 0.012062343768775463, "grad_norm": 0.07115600258111954, "global_step": 50383, "epoch": 475, "lr": 5.999196742246027e-07} +{"train_loss": 0.014313487336039543, "grad_norm": 0.08993955701589584, "global_step": 50384, "epoch": 475, "lr": 5.994620262189776e-07} +{"train_loss": 0.0166560597717762, "grad_norm": 0.12990377843379974, "global_step": 50385, "epoch": 475, "lr": 5.990045517853538e-07} +{"train_loss": 0.013722487725317478, "grad_norm": 0.10418885201215744, "global_step": 50386, "epoch": 475, "lr": 5.985472509253465e-07} +{"train_loss": 0.01236776728183031, "grad_norm": 0.10209986567497253, "global_step": 50387, "epoch": 475, "lr": 5.980901236405601e-07} +{"train_loss": 0.014430331066250801, "grad_norm": 0.10086048394441605, "global_step": 50388, "epoch": 475, "lr": 5.976331699325988e-07} +{"train_loss": 0.014618189074099064, "grad_norm": 0.10319226235151291, "global_step": 50389, "epoch": 475, "lr": 5.971763898030669e-07} +{"train_loss": 0.014855416491627693, "grad_norm": 0.0764685571193695, "global_step": 50390, "epoch": 475, "lr": 5.967197832535798e-07} +{"train_loss": 0.013389740139245987, "grad_norm": 0.06477522850036621, "global_step": 50391, "epoch": 475, "lr": 5.962633502857306e-07} +{"train_loss": 0.016621088609099388, "grad_norm": 0.08901390433311462, "global_step": 50392, "epoch": 475, "lr": 5.958070909011293e-07} +{"train_loss": 0.013664258643984795, "grad_norm": 0.05685396119952202, "global_step": 50393, "epoch": 475, "lr": 5.953510051013689e-07} +{"train_loss": 0.012283774092793465, "grad_norm": 0.05961029604077339, "global_step": 50394, "epoch": 475, "lr": 5.948950928880703e-07} +{"train_loss": 0.0130680613219738, "grad_norm": 0.07883361726999283, "global_step": 50395, "epoch": 475, "lr": 5.944393542628157e-07} +{"train_loss": 0.015364641323685646, "grad_norm": 0.08366470038890839, "global_step": 50396, "epoch": 475, "lr": 5.939837892272204e-07} +{"train_loss": 0.015404053963720798, "grad_norm": 0.08538494259119034, "global_step": 50397, "epoch": 475, "lr": 5.935283977828776e-07} +{"train_loss": 0.013728276826441288, "grad_norm": 0.059103529900312424, "global_step": 50398, "epoch": 475, "lr": 5.930731799313971e-07} +{"train_loss": 0.014921796508133411, "grad_norm": 0.0813225656747818, "global_step": 50399, "epoch": 475, "lr": 5.92618135674361e-07} +{"train_loss": 0.016465242952108383, "grad_norm": 0.053358446806669235, "global_step": 50400, "epoch": 475, "lr": 5.921632650133791e-07} +{"train_loss": 0.01639309525489807, "grad_norm": 0.10530690848827362, "global_step": 50401, "epoch": 475, "lr": 5.917085679500556e-07} +{"train_loss": 0.015830576419830322, "grad_norm": 0.064551442861557, "global_step": 50402, "epoch": 475, "lr": 5.912540444859782e-07} +{"train_loss": 0.013641402125358582, "grad_norm": 0.10058367997407913, "global_step": 50403, "epoch": 475, "lr": 5.907996946227457e-07} +{"train_loss": 0.013687757775187492, "grad_norm": 0.105772964656353, "global_step": 50404, "epoch": 475, "lr": 5.903455183619566e-07} +{"train_loss": 0.015131469815969467, "grad_norm": 0.08064990490674973, "global_step": 50405, "epoch": 475, "lr": 5.898915157051987e-07} +{"train_loss": 0.016041971743106842, "grad_norm": 0.10524182766675949, "global_step": 50406, "epoch": 475, "lr": 5.894376866540818e-07} +{"train_loss": 0.013525946997106075, "grad_norm": 0.06331279873847961, "global_step": 50407, "epoch": 475, "lr": 5.889840312101935e-07} +{"train_loss": 0.016703443601727486, "grad_norm": 0.094458669424057, "global_step": 50408, "epoch": 475, "lr": 5.885305493751214e-07} +{"train_loss": 0.014615498483181, "grad_norm": 0.1053517684340477, "global_step": 50409, "epoch": 475, "lr": 5.880772411504699e-07} +{"train_loss": 0.01917414553463459, "grad_norm": 0.11059290170669556, "global_step": 50410, "epoch": 475, "lr": 5.876241065378207e-07} +{"train_loss": 0.017809167504310608, "grad_norm": 0.08835744112730026, "global_step": 50411, "epoch": 475, "lr": 5.871711455387785e-07} +{"train_loss": 0.014898916706442833, "grad_norm": 0.11234674602746964, "global_step": 50412, "epoch": 475, "lr": 5.867183581549196e-07} +{"train_loss": 0.013445476070046425, "grad_norm": 0.053572796285152435, "global_step": 50413, "epoch": 475, "lr": 5.862657443878483e-07} +{"train_loss": 0.016732297837734222, "grad_norm": 0.06656452268362045, "global_step": 50414, "epoch": 475, "lr": 5.858133042391467e-07} +{"train_loss": 0.013959339819848537, "grad_norm": 0.08969465643167496, "global_step": 50415, "epoch": 475, "lr": 5.853610377104135e-07} +{"train_loss": 0.012846751138567924, "grad_norm": 0.07388706505298615, "global_step": 50416, "epoch": 475, "lr": 5.849089448032252e-07} +{"train_loss": 0.014169203117489815, "grad_norm": 0.08280527591705322, "global_step": 50417, "epoch": 475, "lr": 5.844570255191806e-07} +{"train_loss": 0.015238162130117416, "grad_norm": 0.08230139315128326, "global_step": 50418, "epoch": 475, "lr": 5.840052798598616e-07} +{"train_loss": 0.017849614843726158, "grad_norm": 0.06642528623342514, "global_step": 50419, "epoch": 475, "lr": 5.83553707826856e-07} +{"train_loss": 0.015062937512993813, "grad_norm": 0.0700918585062027, "global_step": 50420, "epoch": 475, "lr": 5.83102309421757e-07} +{"train_loss": 0.014746423810720444, "grad_norm": 0.09121980518102646, "global_step": 50421, "epoch": 475, "lr": 5.826510846461408e-07} +{"train_loss": 0.014692127704620361, "grad_norm": 0.07892128825187683, "global_step": 50422, "epoch": 475, "lr": 5.822000335016065e-07} +{"train_loss": 0.01714531145989895, "grad_norm": 0.09046125411987305, "global_step": 50423, "epoch": 475, "lr": 5.817491559897192e-07} +{"train_loss": 0.01658552512526512, "grad_norm": 0.06622257083654404, "global_step": 50424, "epoch": 475, "lr": 5.812984521120779e-07} +{"train_loss": 0.014451372437179089, "grad_norm": 0.07611308246850967, "global_step": 50425, "epoch": 475, "lr": 5.8084792187027e-07} +{"train_loss": 0.013833538629114628, "grad_norm": 0.06519923359155655, "global_step": 50426, "epoch": 475, "lr": 5.803975652658611e-07} +{"train_loss": 0.014356717467308044, "grad_norm": 0.16545385122299194, "global_step": 50427, "epoch": 475, "lr": 5.799473823004442e-07} +{"train_loss": 0.013900191523134708, "grad_norm": 0.06012113764882088, "global_step": 50428, "epoch": 475, "lr": 5.794973729756014e-07} +{"train_loss": 0.015293183736503124, "grad_norm": 0.09855999052524567, "global_step": 50429, "epoch": 475, "lr": 5.790475372929093e-07} +{"train_loss": 0.016842851415276527, "grad_norm": 0.11761032789945602, "global_step": 50430, "epoch": 475, "lr": 5.785978752539556e-07} +{"train_loss": 0.016022318974137306, "grad_norm": 0.07794712483882904, "global_step": 50431, "epoch": 475, "lr": 5.781483868603165e-07} +{"train_loss": 0.015740539878606796, "grad_norm": 0.06597296893596649, "global_step": 50432, "epoch": 475, "lr": 5.776990721135689e-07} +{"train_loss": 0.01596425659954548, "grad_norm": 0.09665396064519882, "global_step": 50433, "epoch": 475, "lr": 5.772499310152946e-07} +{"train_loss": 0.014476735144853592, "grad_norm": 0.0601608008146286, "global_step": 50434, "epoch": 475, "lr": 5.768009635670702e-07} +{"train_loss": 0.014725979417562485, "grad_norm": 0.10917777568101883, "global_step": 50435, "epoch": 475, "lr": 5.763521697704721e-07} +{"train_loss": 0.015557855367660522, "grad_norm": 0.0877041444182396, "global_step": 50436, "epoch": 475, "lr": 5.759035496270826e-07} +{"train_loss": 0.011560400947928429, "grad_norm": 0.07495829463005066, "global_step": 50437, "epoch": 475, "lr": 5.754551031384725e-07} +{"train_loss": 0.014847977086901665, "grad_norm": 0.10163435339927673, "global_step": 50438, "epoch": 475, "lr": 5.750068303062184e-07} +{"train_loss": 0.013317212462425232, "grad_norm": 0.05565456673502922, "global_step": 50439, "epoch": 475, "lr": 5.745587311318968e-07} +{"train_loss": 0.015604685060679913, "grad_norm": 0.1028565987944603, "global_step": 50440, "epoch": 475, "lr": 5.741108056170785e-07} +{"train_loss": 0.015487316064536572, "grad_norm": 0.06754449754953384, "global_step": 50441, "epoch": 475, "lr": 5.736630537633403e-07} +{"train_loss": 0.014672860503196716, "grad_norm": 0.06038197875022888, "global_step": 50442, "epoch": 475, "lr": 5.732154755722585e-07} +{"train_loss": 0.01574120670557022, "grad_norm": 0.07105260342359543, "global_step": 50443, "epoch": 475, "lr": 5.727680710454042e-07} +{"train_loss": 0.014539157971739769, "grad_norm": 0.07722126692533493, "global_step": 50444, "epoch": 475, "lr": 5.723208401843371e-07} +{"train_loss": 0.015525241382420063, "grad_norm": 0.0670577734708786, "global_step": 50445, "epoch": 475, "lr": 5.718737829906506e-07} +{"train_loss": 0.014201359823346138, "grad_norm": 0.08847962319850922, "global_step": 50446, "epoch": 475, "lr": 5.714268994658989e-07} +{"train_loss": 0.01667511835694313, "grad_norm": 0.08287931978702545, "global_step": 50447, "epoch": 475, "lr": 5.709801896116529e-07} +{"train_loss": 0.014527538791298866, "grad_norm": 0.05657678470015526, "global_step": 50448, "epoch": 475, "lr": 5.705336534294947e-07} +{"train_loss": 0.013905785977840424, "grad_norm": 0.06768616288900375, "global_step": 50449, "epoch": 475, "lr": 5.700872909209842e-07} +{"train_loss": 0.014999069273471832, "grad_norm": 0.07163545489311218, "global_step": 50450, "epoch": 475, "lr": 5.696411020876868e-07} +{"train_loss": 0.015157695859670639, "grad_norm": 0.09162379056215286, "global_step": 50451, "epoch": 475, "lr": 5.691950869311736e-07} +{"train_loss": 0.015073966234922409, "grad_norm": 0.06352629512548447, "global_step": 50452, "epoch": 475, "lr": 5.687492454530097e-07} +{"train_loss": 0.014074454084038734, "grad_norm": 0.08490247279405594, "global_step": 50453, "epoch": 475, "lr": 5.683035776547663e-07} +{"train_loss": 0.016222944483160973, "grad_norm": 0.07028449326753616, "global_step": 50454, "epoch": 475, "lr": 5.678580835380087e-07} +{"train_loss": 0.014947451203007181, "grad_norm": 0.23668010532855988, "global_step": 50455, "epoch": 475, "lr": 5.674127631043025e-07, "val_loss": 0.01926458068192005} +{"train_loss": 0.014478160068392754, "grad_norm": 0.05752090364694595, "global_step": 50456, "epoch": 476, "lr": 5.669676163552074e-07} +{"train_loss": 0.01418991293758154, "grad_norm": 0.07927361130714417, "global_step": 50457, "epoch": 476, "lr": 5.665226432922888e-07} +{"train_loss": 0.01497168280184269, "grad_norm": 0.07614126056432724, "global_step": 50458, "epoch": 476, "lr": 5.660778439171122e-07} +{"train_loss": 0.015022268518805504, "grad_norm": 0.06434779614210129, "global_step": 50459, "epoch": 476, "lr": 5.656332182312374e-07} +{"train_loss": 0.015441949479281902, "grad_norm": 0.12737269699573517, "global_step": 50460, "epoch": 476, "lr": 5.651887662362299e-07} +{"train_loss": 0.014281794428825378, "grad_norm": 0.07906144857406616, "global_step": 50461, "epoch": 476, "lr": 5.64744487933655e-07} +{"train_loss": 0.014804456382989883, "grad_norm": 0.07680095732212067, "global_step": 50462, "epoch": 476, "lr": 5.643003833250671e-07} +{"train_loss": 0.01719936728477478, "grad_norm": 0.09805865585803986, "global_step": 50463, "epoch": 476, "lr": 5.638564524120316e-07} +{"train_loss": 0.01623518578708172, "grad_norm": 0.10121459513902664, "global_step": 50464, "epoch": 476, "lr": 5.634126951960972e-07} +{"train_loss": 0.013485925272107124, "grad_norm": 0.10381012409925461, "global_step": 50465, "epoch": 476, "lr": 5.629691116788405e-07} +{"train_loss": 0.015449116006493568, "grad_norm": 0.09749886393547058, "global_step": 50466, "epoch": 476, "lr": 5.625257018618046e-07} +{"train_loss": 0.014265637844800949, "grad_norm": 0.12360105663537979, "global_step": 50467, "epoch": 476, "lr": 5.62082465746555e-07} +{"train_loss": 0.017990289255976677, "grad_norm": 0.06269258260726929, "global_step": 50468, "epoch": 476, "lr": 5.616394033346461e-07} +{"train_loss": 0.01915571093559265, "grad_norm": 0.08678583800792694, "global_step": 50469, "epoch": 476, "lr": 5.611965146276376e-07} +{"train_loss": 0.012916460633277893, "grad_norm": 0.054485127329826355, "global_step": 50470, "epoch": 476, "lr": 5.60753799627084e-07} +{"train_loss": 0.012953018769621849, "grad_norm": 0.09365473687648773, "global_step": 50471, "epoch": 476, "lr": 5.603112583345394e-07} +{"train_loss": 0.015614569187164307, "grad_norm": 0.11306329071521759, "global_step": 50472, "epoch": 476, "lr": 5.598688907515637e-07} +{"train_loss": 0.013874993659555912, "grad_norm": 0.08843503892421722, "global_step": 50473, "epoch": 476, "lr": 5.594266968797113e-07} +{"train_loss": 0.014462942257523537, "grad_norm": 0.0899556577205658, "global_step": 50474, "epoch": 476, "lr": 5.589846767205254e-07} +{"train_loss": 0.014125053770840168, "grad_norm": 0.08272713422775269, "global_step": 50475, "epoch": 476, "lr": 5.585428302755657e-07} +{"train_loss": 0.015783987939357758, "grad_norm": 0.07642436027526855, "global_step": 50476, "epoch": 476, "lr": 5.581011575463868e-07} +{"train_loss": 0.015549973584711552, "grad_norm": 0.056888647377491, "global_step": 50477, "epoch": 476, "lr": 5.576596585345372e-07} +{"train_loss": 0.014273807406425476, "grad_norm": 0.05554727092385292, "global_step": 50478, "epoch": 476, "lr": 5.57218333241577e-07} +{"train_loss": 0.015508804470300674, "grad_norm": 0.07090231776237488, "global_step": 50479, "epoch": 476, "lr": 5.567771816690381e-07} +{"train_loss": 0.01677677407860756, "grad_norm": 0.0847260057926178, "global_step": 50480, "epoch": 476, "lr": 5.563362038184917e-07} +{"train_loss": 0.016841357573866844, "grad_norm": 0.0722382590174675, "global_step": 50481, "epoch": 476, "lr": 5.558953996914695e-07} +{"train_loss": 0.013460598886013031, "grad_norm": 0.057819999754428864, "global_step": 50482, "epoch": 476, "lr": 5.554547692895318e-07} +{"train_loss": 0.015868205577135086, "grad_norm": 0.05820852145552635, "global_step": 50483, "epoch": 476, "lr": 5.550143126142271e-07} +{"train_loss": 0.014761287719011307, "grad_norm": 0.06415490806102753, "global_step": 50484, "epoch": 476, "lr": 5.545740296670987e-07} +{"train_loss": 0.01386386901140213, "grad_norm": 0.09103536605834961, "global_step": 50485, "epoch": 476, "lr": 5.541339204496898e-07} +{"train_loss": 0.01614191196858883, "grad_norm": 0.06123846396803856, "global_step": 50486, "epoch": 476, "lr": 5.536939849635492e-07} +{"train_loss": 0.017743341624736786, "grad_norm": 0.1390426754951477, "global_step": 50487, "epoch": 476, "lr": 5.532542232102256e-07} +{"train_loss": 0.014482397586107254, "grad_norm": 0.0662909522652626, "global_step": 50488, "epoch": 476, "lr": 5.528146351912622e-07} +{"train_loss": 0.014485970139503479, "grad_norm": 0.0921890139579773, "global_step": 50489, "epoch": 476, "lr": 5.523752209082078e-07} +{"train_loss": 0.013497978448867798, "grad_norm": 0.0964510589838028, "global_step": 50490, "epoch": 476, "lr": 5.519359803626056e-07} +{"train_loss": 0.01851499453186989, "grad_norm": 0.0659722313284874, "global_step": 50491, "epoch": 476, "lr": 5.514969135559877e-07} +{"train_loss": 0.016829941421747208, "grad_norm": 0.12853316962718964, "global_step": 50492, "epoch": 476, "lr": 5.510580204899085e-07} +{"train_loss": 0.013722906820476055, "grad_norm": 0.06657067686319351, "global_step": 50493, "epoch": 476, "lr": 5.506193011659111e-07} +{"train_loss": 0.014956481754779816, "grad_norm": 0.07976184040307999, "global_step": 50494, "epoch": 476, "lr": 5.501807555855221e-07} +{"train_loss": 0.012625799514353275, "grad_norm": 0.060593292117118835, "global_step": 50495, "epoch": 476, "lr": 5.497423837503013e-07} +{"train_loss": 0.014324769377708435, "grad_norm": 0.09518275409936905, "global_step": 50496, "epoch": 476, "lr": 5.493041856617809e-07} +{"train_loss": 0.013185430318117142, "grad_norm": 0.05839342623949051, "global_step": 50497, "epoch": 476, "lr": 5.488661613214985e-07} +{"train_loss": 0.014580285176634789, "grad_norm": 0.08594830334186554, "global_step": 50498, "epoch": 476, "lr": 5.484283107309918e-07} +{"train_loss": 0.018176615238189697, "grad_norm": 0.1000097319483757, "global_step": 50499, "epoch": 476, "lr": 5.479906338917984e-07} +{"train_loss": 0.014594011008739471, "grad_norm": 0.09973849356174469, "global_step": 50500, "epoch": 476, "lr": 5.475531308054671e-07} +{"train_loss": 0.017573164775967598, "grad_norm": 0.06281064450740814, "global_step": 50501, "epoch": 476, "lr": 5.471158014735245e-07} +{"train_loss": 0.015255779027938843, "grad_norm": 0.07487403601408005, "global_step": 50502, "epoch": 476, "lr": 5.466786458975137e-07} +{"train_loss": 0.01264101080596447, "grad_norm": 0.11011172086000443, "global_step": 50503, "epoch": 476, "lr": 5.46241664078967e-07} +{"train_loss": 0.016053149476647377, "grad_norm": 0.14622442424297333, "global_step": 50504, "epoch": 476, "lr": 5.458048560194162e-07} +{"train_loss": 0.012791931629180908, "grad_norm": 0.0781840980052948, "global_step": 50505, "epoch": 476, "lr": 5.453682217204048e-07} +{"train_loss": 0.016522116959095, "grad_norm": 0.07562791556119919, "global_step": 50506, "epoch": 476, "lr": 5.449317611834593e-07} +{"train_loss": 0.012647800147533417, "grad_norm": 0.04804183915257454, "global_step": 50507, "epoch": 476, "lr": 5.444954744101172e-07} +{"train_loss": 0.013691953383386135, "grad_norm": 0.06884486973285675, "global_step": 50508, "epoch": 476, "lr": 5.440593614019108e-07} +{"train_loss": 0.015842337161302567, "grad_norm": 0.05973280966281891, "global_step": 50509, "epoch": 476, "lr": 5.436234221603719e-07} +{"train_loss": 0.016637511551380157, "grad_norm": 0.06572505086660385, "global_step": 50510, "epoch": 476, "lr": 5.431876566870331e-07} +{"train_loss": 0.01605043187737465, "grad_norm": 0.06585568934679031, "global_step": 50511, "epoch": 476, "lr": 5.427520649834206e-07} +{"train_loss": 0.013594402000308037, "grad_norm": 0.07340206950902939, "global_step": 50512, "epoch": 476, "lr": 5.423166470510721e-07} +{"train_loss": 0.014059754088521004, "grad_norm": 0.13181179761886597, "global_step": 50513, "epoch": 476, "lr": 5.418814028915143e-07} +{"train_loss": 0.014847006648778915, "grad_norm": 0.07212886214256287, "global_step": 50514, "epoch": 476, "lr": 5.414463325062791e-07} +{"train_loss": 0.01820533350110054, "grad_norm": 0.08011993020772934, "global_step": 50515, "epoch": 476, "lr": 5.410114358968877e-07} +{"train_loss": 0.014265395700931549, "grad_norm": 0.10097827762365341, "global_step": 50516, "epoch": 476, "lr": 5.40576713064872e-07} +{"train_loss": 0.017052819952368736, "grad_norm": 0.1606646180152893, "global_step": 50517, "epoch": 476, "lr": 5.401421640117643e-07} +{"train_loss": 0.01694313809275627, "grad_norm": 0.09854070097208023, "global_step": 50518, "epoch": 476, "lr": 5.397077887390856e-07} +{"train_loss": 0.016039539128541946, "grad_norm": 0.09061215072870255, "global_step": 50519, "epoch": 476, "lr": 5.392735872483623e-07} +{"train_loss": 0.014588067308068275, "grad_norm": 0.06131052225828171, "global_step": 50520, "epoch": 476, "lr": 5.388395595411266e-07} +{"train_loss": 0.013271511532366276, "grad_norm": 0.06863712519407272, "global_step": 50521, "epoch": 476, "lr": 5.384057056188996e-07} +{"train_loss": 0.015002629719674587, "grad_norm": 0.0744541808962822, "global_step": 50522, "epoch": 476, "lr": 5.379720254831965e-07} +{"train_loss": 0.01288003008812666, "grad_norm": 0.05614393576979637, "global_step": 50523, "epoch": 476, "lr": 5.375385191355554e-07} +{"train_loss": 0.018426697701215744, "grad_norm": 0.06515967100858688, "global_step": 50524, "epoch": 476, "lr": 5.371051865774912e-07} +{"train_loss": 0.017362985759973526, "grad_norm": 0.07508333027362823, "global_step": 50525, "epoch": 476, "lr": 5.366720278105309e-07} +{"train_loss": 0.015968535095453262, "grad_norm": 0.06277148425579071, "global_step": 50526, "epoch": 476, "lr": 5.362390428361897e-07} +{"train_loss": 0.01648227870464325, "grad_norm": 0.06327348202466965, "global_step": 50527, "epoch": 476, "lr": 5.358062316559998e-07} +{"train_loss": 0.017069034278392792, "grad_norm": 0.06387374550104141, "global_step": 50528, "epoch": 476, "lr": 5.35373594271471e-07} +{"train_loss": 0.015788603574037552, "grad_norm": 0.08510740101337433, "global_step": 50529, "epoch": 476, "lr": 5.349411306841246e-07} +{"train_loss": 0.01674676313996315, "grad_norm": 0.09128674119710922, "global_step": 50530, "epoch": 476, "lr": 5.345088408954868e-07} +{"train_loss": 0.015037569217383862, "grad_norm": 0.07469560205936432, "global_step": 50531, "epoch": 476, "lr": 5.340767249070789e-07} +{"train_loss": 0.013709035702049732, "grad_norm": 0.057451359927654266, "global_step": 50532, "epoch": 476, "lr": 5.33644782720405e-07} +{"train_loss": 0.014890534803271294, "grad_norm": 0.059699539095163345, "global_step": 50533, "epoch": 476, "lr": 5.332130143369918e-07} +{"train_loss": 0.012912440113723278, "grad_norm": 0.062000371515750885, "global_step": 50534, "epoch": 476, "lr": 5.327814197583603e-07} +{"train_loss": 0.0138709656894207, "grad_norm": 0.055900268256664276, "global_step": 50535, "epoch": 476, "lr": 5.323499989860148e-07} +{"train_loss": 0.016019415110349655, "grad_norm": 0.10252780467271805, "global_step": 50536, "epoch": 476, "lr": 5.319187520214819e-07} +{"train_loss": 0.01457309816032648, "grad_norm": 0.08817549794912338, "global_step": 50537, "epoch": 476, "lr": 5.31487678866277e-07} +{"train_loss": 0.015511208213865757, "grad_norm": 0.055851906538009644, "global_step": 50538, "epoch": 476, "lr": 5.310567795219102e-07} +{"train_loss": 0.015507589094340801, "grad_norm": 0.08464084565639496, "global_step": 50539, "epoch": 476, "lr": 5.306260539898911e-07} +{"train_loss": 0.012753825634717941, "grad_norm": 0.08652272075414658, "global_step": 50540, "epoch": 476, "lr": 5.301955022717409e-07} +{"train_loss": 0.012247591279447079, "grad_norm": 0.0902879536151886, "global_step": 50541, "epoch": 476, "lr": 5.297651243689695e-07} +{"train_loss": 0.013985582627356052, "grad_norm": 0.08237162977457047, "global_step": 50542, "epoch": 476, "lr": 5.293349202830922e-07} +{"train_loss": 0.015142679214477539, "grad_norm": 0.06355756521224976, "global_step": 50543, "epoch": 476, "lr": 5.289048900156191e-07} +{"train_loss": 0.014999987557530403, "grad_norm": 0.07039287686347961, "global_step": 50544, "epoch": 476, "lr": 5.284750335680543e-07} +{"train_loss": 0.012290488928556442, "grad_norm": 0.09264590591192245, "global_step": 50545, "epoch": 476, "lr": 5.280453509419192e-07} +{"train_loss": 0.013434536755084991, "grad_norm": 0.07917050272226334, "global_step": 50546, "epoch": 476, "lr": 5.276158421387122e-07} +{"train_loss": 0.01468786969780922, "grad_norm": 0.097405344247818, "global_step": 50547, "epoch": 476, "lr": 5.271865071599546e-07} +{"train_loss": 0.01480424776673317, "grad_norm": 0.047276757657527924, "global_step": 50548, "epoch": 476, "lr": 5.267573460071396e-07} +{"train_loss": 0.013637152500450611, "grad_norm": 0.06617490202188492, "global_step": 50549, "epoch": 476, "lr": 5.263283586817935e-07} +{"train_loss": 0.011596053838729858, "grad_norm": 0.056770630180835724, "global_step": 50550, "epoch": 476, "lr": 5.258995451854099e-07} +{"train_loss": 0.015227441675961018, "grad_norm": 0.10170131921768188, "global_step": 50551, "epoch": 476, "lr": 5.254709055194984e-07} +{"train_loss": 0.013184340670704842, "grad_norm": 0.06969641149044037, "global_step": 50552, "epoch": 476, "lr": 5.250424396855635e-07} +{"train_loss": 0.014632352627813816, "grad_norm": 0.09197863191366196, "global_step": 50553, "epoch": 476, "lr": 5.246141476851207e-07} +{"train_loss": 0.0172037985175848, "grad_norm": 0.0784715861082077, "global_step": 50554, "epoch": 476, "lr": 5.241860295196632e-07} +{"train_loss": 0.014214301481842995, "grad_norm": 0.06531146913766861, "global_step": 50555, "epoch": 476, "lr": 5.23758085190701e-07} +{"train_loss": 0.013997621834278107, "grad_norm": 0.11440032720565796, "global_step": 50556, "epoch": 476, "lr": 5.233303146997326e-07} +{"train_loss": 0.0131035465747118, "grad_norm": 0.06442071497440338, "global_step": 50557, "epoch": 476, "lr": 5.229027180482682e-07} +{"train_loss": 0.014839300885796547, "grad_norm": 0.06996031850576401, "global_step": 50558, "epoch": 476, "lr": 5.224752952378064e-07} +{"train_loss": 0.011110013350844383, "grad_norm": 0.06940425932407379, "global_step": 50559, "epoch": 476, "lr": 5.220480462698462e-07} +{"train_loss": 0.014300277456641197, "grad_norm": 0.07482703030109406, "global_step": 50560, "epoch": 476, "lr": 5.216209711458975e-07} +{"train_loss": 0.014950554668551908, "grad_norm": 0.10485243052244186, "global_step": 50561, "epoch": 476, "lr": 5.211940698674534e-07, "val_loss": 0.020161427557468414} +{"train_loss": 0.015401134267449379, "grad_norm": 0.05895640701055527, "global_step": 50562, "epoch": 477, "lr": 5.207673424360127e-07} +{"train_loss": 0.013366551138460636, "grad_norm": 0.06398463994264603, "global_step": 50563, "epoch": 477, "lr": 5.203407888530798e-07} +{"train_loss": 0.014392614364624023, "grad_norm": 0.0781337171792984, "global_step": 50564, "epoch": 477, "lr": 5.199144091201536e-07} +{"train_loss": 0.014470269903540611, "grad_norm": 0.07533689588308334, "global_step": 50565, "epoch": 477, "lr": 5.194882032387216e-07} +{"train_loss": 0.0166042260825634, "grad_norm": 0.0977974683046341, "global_step": 50566, "epoch": 477, "lr": 5.190621712102939e-07} +{"train_loss": 0.014721914194524288, "grad_norm": 0.13107547163963318, "global_step": 50567, "epoch": 477, "lr": 5.186363130363637e-07} +{"train_loss": 0.014555168338119984, "grad_norm": 0.06665139645338058, "global_step": 50568, "epoch": 477, "lr": 5.182106287184296e-07} +{"train_loss": 0.014694372192025185, "grad_norm": 0.2112097293138504, "global_step": 50569, "epoch": 477, "lr": 5.177851182579796e-07} +{"train_loss": 0.017317652702331543, "grad_norm": 0.059288401156663895, "global_step": 50570, "epoch": 477, "lr": 5.173597816565179e-07} +{"train_loss": 0.0158399511128664, "grad_norm": 0.09601669013500214, "global_step": 50571, "epoch": 477, "lr": 5.169346189155322e-07} +{"train_loss": 0.014020266942679882, "grad_norm": 0.14476145803928375, "global_step": 50572, "epoch": 477, "lr": 5.165096300365214e-07} +{"train_loss": 0.013409445993602276, "grad_norm": 0.06482166051864624, "global_step": 50573, "epoch": 477, "lr": 5.160848150209673e-07} +{"train_loss": 0.015243327245116234, "grad_norm": 0.0894261971116066, "global_step": 50574, "epoch": 477, "lr": 5.156601738703748e-07} +{"train_loss": 0.015393003821372986, "grad_norm": 0.07383739948272705, "global_step": 50575, "epoch": 477, "lr": 5.152357065862368e-07} +{"train_loss": 0.014743290841579437, "grad_norm": 0.11531071364879608, "global_step": 50576, "epoch": 477, "lr": 5.148114131700354e-07} +{"train_loss": 0.01645883172750473, "grad_norm": 0.07367614656686783, "global_step": 50577, "epoch": 477, "lr": 5.143872936232697e-07} +{"train_loss": 0.017206799238920212, "grad_norm": 0.06220686435699463, "global_step": 50578, "epoch": 477, "lr": 5.139633479474271e-07} +{"train_loss": 0.017940236255526543, "grad_norm": 0.10168134421110153, "global_step": 50579, "epoch": 477, "lr": 5.135395761439954e-07} +{"train_loss": 0.01481139101088047, "grad_norm": 0.08601796627044678, "global_step": 50580, "epoch": 477, "lr": 5.131159782144623e-07} +{"train_loss": 0.012545140460133553, "grad_norm": 0.05870102345943451, "global_step": 50581, "epoch": 477, "lr": 5.126925541603156e-07} +{"train_loss": 0.015260735526680946, "grad_norm": 0.08377081900835037, "global_step": 50582, "epoch": 477, "lr": 5.122693039830539e-07} +{"train_loss": 0.014490964822471142, "grad_norm": 0.07295255362987518, "global_step": 50583, "epoch": 477, "lr": 5.118462276841485e-07} +{"train_loss": 0.015235070139169693, "grad_norm": 0.06151988357305527, "global_step": 50584, "epoch": 477, "lr": 5.114233252650979e-07} +{"train_loss": 0.014997386373579502, "grad_norm": 0.08553065359592438, "global_step": 50585, "epoch": 477, "lr": 5.110005967273846e-07} +{"train_loss": 0.01300233043730259, "grad_norm": 0.07820524275302887, "global_step": 50586, "epoch": 477, "lr": 5.105780420724904e-07} +{"train_loss": 0.015401200391352177, "grad_norm": 0.08529777824878693, "global_step": 50587, "epoch": 477, "lr": 5.101556613019032e-07} +{"train_loss": 0.016068417578935623, "grad_norm": 0.11234361678361893, "global_step": 50588, "epoch": 477, "lr": 5.097334544171107e-07} +{"train_loss": 0.01963043212890625, "grad_norm": 0.08641793578863144, "global_step": 50589, "epoch": 477, "lr": 5.093114214195838e-07} +{"train_loss": 0.014867672696709633, "grad_norm": 0.08618281781673431, "global_step": 50590, "epoch": 477, "lr": 5.088895623108214e-07} +{"train_loss": 0.01304563321173191, "grad_norm": 0.08063127845525742, "global_step": 50591, "epoch": 477, "lr": 5.084678770922946e-07} +{"train_loss": 0.013249466195702553, "grad_norm": 0.09902019798755646, "global_step": 50592, "epoch": 477, "lr": 5.080463657654855e-07} +{"train_loss": 0.013403794728219509, "grad_norm": 0.09322068840265274, "global_step": 50593, "epoch": 477, "lr": 5.076250283318818e-07} +{"train_loss": 0.017208829522132874, "grad_norm": 0.12104880809783936, "global_step": 50594, "epoch": 477, "lr": 5.0720386479296e-07} +{"train_loss": 0.014441240578889847, "grad_norm": 0.07424034923315048, "global_step": 50595, "epoch": 477, "lr": 5.067828751501968e-07} +{"train_loss": 0.013931136578321457, "grad_norm": 0.07750944048166275, "global_step": 50596, "epoch": 477, "lr": 5.0636205940508e-07} +{"train_loss": 0.013798888772726059, "grad_norm": 0.08391593396663666, "global_step": 50597, "epoch": 477, "lr": 5.059414175590804e-07} +{"train_loss": 0.01423749327659607, "grad_norm": 0.09194532781839371, "global_step": 50598, "epoch": 477, "lr": 5.055209496136748e-07} +{"train_loss": 0.016110321506857872, "grad_norm": 0.09237689524888992, "global_step": 50599, "epoch": 477, "lr": 5.051006555703453e-07} +{"train_loss": 0.016466893255710602, "grad_norm": 0.08685636520385742, "global_step": 50600, "epoch": 477, "lr": 5.046805354305684e-07} +{"train_loss": 0.014835602603852749, "grad_norm": 0.12013666331768036, "global_step": 50601, "epoch": 477, "lr": 5.042605891958207e-07} +{"train_loss": 0.015737487003207207, "grad_norm": 0.06482043862342834, "global_step": 50602, "epoch": 477, "lr": 5.03840816867579e-07} +{"train_loss": 0.014605882577598095, "grad_norm": 0.14411769807338715, "global_step": 50603, "epoch": 477, "lr": 5.034212184473086e-07} +{"train_loss": 0.013441333547234535, "grad_norm": 0.07419117540121078, "global_step": 50604, "epoch": 477, "lr": 5.030017939364917e-07} +{"train_loss": 0.013014749623835087, "grad_norm": 0.137889564037323, "global_step": 50605, "epoch": 477, "lr": 5.025825433366049e-07} +{"train_loss": 0.016707517206668854, "grad_norm": 0.06261055916547775, "global_step": 50606, "epoch": 477, "lr": 5.021634666491137e-07} +{"train_loss": 0.01776275411248207, "grad_norm": 0.09273526072502136, "global_step": 50607, "epoch": 477, "lr": 5.017445638754892e-07} +{"train_loss": 0.014452682808041573, "grad_norm": 0.11877221614122391, "global_step": 50608, "epoch": 477, "lr": 5.01325835017219e-07} +{"train_loss": 0.014188711531460285, "grad_norm": 0.059030793607234955, "global_step": 50609, "epoch": 477, "lr": 5.009072800757574e-07} +{"train_loss": 0.015423993580043316, "grad_norm": 0.09323158860206604, "global_step": 50610, "epoch": 477, "lr": 5.004888990525814e-07} +{"train_loss": 0.011986766010522842, "grad_norm": 0.09543514996767044, "global_step": 50611, "epoch": 477, "lr": 5.000706919491616e-07} +{"train_loss": 0.016774866729974747, "grad_norm": 0.10920090228319168, "global_step": 50612, "epoch": 477, "lr": 4.996526587669636e-07} +{"train_loss": 0.015822526067495346, "grad_norm": 0.149619460105896, "global_step": 50613, "epoch": 477, "lr": 4.992347995074643e-07} +{"train_loss": 0.014477092772722244, "grad_norm": 0.05688446760177612, "global_step": 50614, "epoch": 477, "lr": 4.988171141721232e-07} +{"train_loss": 0.015490822494029999, "grad_norm": 0.08457380533218384, "global_step": 50615, "epoch": 477, "lr": 4.983996027624116e-07} +{"train_loss": 0.014210740104317665, "grad_norm": 0.06254700571298599, "global_step": 50616, "epoch": 477, "lr": 4.97982265279795e-07} +{"train_loss": 0.015866534784436226, "grad_norm": 0.1054701954126358, "global_step": 50617, "epoch": 477, "lr": 4.975651017257443e-07} +{"train_loss": 0.014862892217934132, "grad_norm": 0.06753383576869965, "global_step": 50618, "epoch": 477, "lr": 4.971481121017141e-07} +{"train_loss": 0.01504126563668251, "grad_norm": 0.14017264544963837, "global_step": 50619, "epoch": 477, "lr": 4.967312964091863e-07} +{"train_loss": 0.01462436281144619, "grad_norm": 0.124335378408432, "global_step": 50620, "epoch": 477, "lr": 4.963146546496155e-07} +{"train_loss": 0.012255655601620674, "grad_norm": 0.054014090448617935, "global_step": 50621, "epoch": 477, "lr": 4.958981868244616e-07} +{"train_loss": 0.015209602192044258, "grad_norm": 0.08077344298362732, "global_step": 50622, "epoch": 477, "lr": 4.954818929351957e-07} +{"train_loss": 0.014062399044632912, "grad_norm": 0.09984520822763443, "global_step": 50623, "epoch": 477, "lr": 4.950657729832775e-07} +{"train_loss": 0.014284949749708176, "grad_norm": 0.12831439077854156, "global_step": 50624, "epoch": 477, "lr": 4.946498269701616e-07} +{"train_loss": 0.016215210780501366, "grad_norm": 0.11658032238483429, "global_step": 50625, "epoch": 477, "lr": 4.942340548973301e-07} +{"train_loss": 0.014736352488398552, "grad_norm": 0.10894562304019928, "global_step": 50626, "epoch": 477, "lr": 4.938184567662208e-07} +{"train_loss": 0.015920627862215042, "grad_norm": 0.07486756891012192, "global_step": 50627, "epoch": 477, "lr": 4.934030325783102e-07} +{"train_loss": 0.016104277223348618, "grad_norm": 0.07794268429279327, "global_step": 50628, "epoch": 477, "lr": 4.929877823350471e-07} +{"train_loss": 0.01665922999382019, "grad_norm": 0.07530906051397324, "global_step": 50629, "epoch": 477, "lr": 4.92572706037897e-07} +{"train_loss": 0.016735710203647614, "grad_norm": 0.11404776573181152, "global_step": 50630, "epoch": 477, "lr": 4.92157803688309e-07} +{"train_loss": 0.016066621989011765, "grad_norm": 0.11318923532962799, "global_step": 50631, "epoch": 477, "lr": 4.917430752877594e-07} +{"train_loss": 0.014298325404524803, "grad_norm": 0.0681409165263176, "global_step": 50632, "epoch": 477, "lr": 4.913285208376861e-07} +{"train_loss": 0.013885295018553734, "grad_norm": 0.12169782817363739, "global_step": 50633, "epoch": 477, "lr": 4.9091414033956e-07} +{"train_loss": 0.014552549459040165, "grad_norm": 0.049899034202098846, "global_step": 50634, "epoch": 477, "lr": 4.904999337948247e-07} +{"train_loss": 0.014277667738497257, "grad_norm": 0.11480981856584549, "global_step": 50635, "epoch": 477, "lr": 4.900859012049397e-07} +{"train_loss": 0.014213882386684418, "grad_norm": 0.09498234838247299, "global_step": 50636, "epoch": 477, "lr": 4.896720425713708e-07} +{"train_loss": 0.013407839462161064, "grad_norm": 0.0935729593038559, "global_step": 50637, "epoch": 477, "lr": 4.892583578955612e-07} +{"train_loss": 0.014059734530746937, "grad_norm": 0.12039850652217865, "global_step": 50638, "epoch": 477, "lr": 4.888448471789598e-07} +{"train_loss": 0.014977633021771908, "grad_norm": 0.06340962648391724, "global_step": 50639, "epoch": 477, "lr": 4.884315104230264e-07} +{"train_loss": 0.017411665990948677, "grad_norm": 0.16047890484333038, "global_step": 50640, "epoch": 477, "lr": 4.880183476292156e-07} +{"train_loss": 0.014987793751060963, "grad_norm": 0.09104892611503601, "global_step": 50641, "epoch": 477, "lr": 4.87605358798976e-07} +{"train_loss": 0.018203556537628174, "grad_norm": 0.05610184371471405, "global_step": 50642, "epoch": 477, "lr": 4.871925439337565e-07} +{"train_loss": 0.014898648485541344, "grad_norm": 0.10842397809028625, "global_step": 50643, "epoch": 477, "lr": 4.867799030350173e-07} +{"train_loss": 0.016095727682113647, "grad_norm": 0.07940807193517685, "global_step": 50644, "epoch": 477, "lr": 4.863674361041903e-07} +{"train_loss": 0.012939538806676865, "grad_norm": 0.0732756182551384, "global_step": 50645, "epoch": 477, "lr": 4.859551431427411e-07} +{"train_loss": 0.014080051332712173, "grad_norm": 0.10845615714788437, "global_step": 50646, "epoch": 477, "lr": 4.855430241521131e-07} +{"train_loss": 0.015119192190468311, "grad_norm": 0.08979814499616623, "global_step": 50647, "epoch": 477, "lr": 4.851310791337494e-07} +{"train_loss": 0.01432815846055746, "grad_norm": 0.05940144136548042, "global_step": 50648, "epoch": 477, "lr": 4.847193080891044e-07} +{"train_loss": 0.014283502474427223, "grad_norm": 0.15622247755527496, "global_step": 50649, "epoch": 477, "lr": 4.843077110196215e-07} +{"train_loss": 0.012034747749567032, "grad_norm": 0.055454377084970474, "global_step": 50650, "epoch": 477, "lr": 4.838962879267495e-07} +{"train_loss": 0.015581187792122364, "grad_norm": 0.06023203954100609, "global_step": 50651, "epoch": 477, "lr": 4.834850388119316e-07} +{"train_loss": 0.01576944813132286, "grad_norm": 0.08709146827459335, "global_step": 50652, "epoch": 477, "lr": 4.830739636766113e-07} +{"train_loss": 0.015013661235570908, "grad_norm": 0.11294309794902802, "global_step": 50653, "epoch": 477, "lr": 4.826630625222372e-07} +{"train_loss": 0.012335434556007385, "grad_norm": 0.05624598264694214, "global_step": 50654, "epoch": 477, "lr": 4.822523353502473e-07} +{"train_loss": 0.013648511841893196, "grad_norm": 0.08162138611078262, "global_step": 50655, "epoch": 477, "lr": 4.818417821620902e-07} +{"train_loss": 0.016739843413233757, "grad_norm": 0.1181066632270813, "global_step": 50656, "epoch": 477, "lr": 4.814314029592038e-07} +{"train_loss": 0.015004636719822884, "grad_norm": 0.07941983640193939, "global_step": 50657, "epoch": 477, "lr": 4.810211977430367e-07} +{"train_loss": 0.015194841660559177, "grad_norm": 0.1511097550392151, "global_step": 50658, "epoch": 477, "lr": 4.806111665150215e-07} +{"train_loss": 0.016744468361139297, "grad_norm": 0.08296304196119308, "global_step": 50659, "epoch": 477, "lr": 4.802013092766011e-07} +{"train_loss": 0.016416752710938454, "grad_norm": 0.11723281443119049, "global_step": 50660, "epoch": 477, "lr": 4.797916260292246e-07} +{"train_loss": 0.013018177822232246, "grad_norm": 0.070695661008358, "global_step": 50661, "epoch": 477, "lr": 4.79382116774324e-07} +{"train_loss": 0.015007935464382172, "grad_norm": 0.06925772875547409, "global_step": 50662, "epoch": 477, "lr": 4.789727815133315e-07} +{"train_loss": 0.013235708698630333, "grad_norm": 0.06316409260034561, "global_step": 50663, "epoch": 477, "lr": 4.785636202477018e-07} +{"train_loss": 0.013625595718622208, "grad_norm": 0.07216336578130722, "global_step": 50664, "epoch": 477, "lr": 4.781546329788555e-07} +{"train_loss": 0.014694353565573692, "grad_norm": 0.06540559232234955, "global_step": 50665, "epoch": 477, "lr": 4.777458197082418e-07} +{"train_loss": 0.013882966712117195, "grad_norm": 0.09861604869365692, "global_step": 50666, "epoch": 477, "lr": 4.773371804372928e-07} +{"train_loss": 0.014948258176445961, "grad_norm": 0.08954411745071411, "global_step": 50667, "epoch": 477, "lr": 4.769287151674406e-07, "val_loss": 0.01908545382320881} +{"train_loss": 0.014197257347404957, "grad_norm": 0.09811888635158539, "global_step": 50668, "epoch": 478, "lr": 4.765204239001231e-07} +{"train_loss": 0.014994516968727112, "grad_norm": 0.09780912846326828, "global_step": 50669, "epoch": 478, "lr": 4.7611230663677787e-07} +{"train_loss": 0.013562986627221107, "grad_norm": 0.05982733890414238, "global_step": 50670, "epoch": 478, "lr": 4.7570436337884274e-07} +{"train_loss": 0.014767115004360676, "grad_norm": 0.0735590010881424, "global_step": 50671, "epoch": 478, "lr": 4.752965941277332e-07} +{"train_loss": 0.014369400218129158, "grad_norm": 0.08538495749235153, "global_step": 50672, "epoch": 478, "lr": 4.748889988849037e-07} +{"train_loss": 0.01495006401091814, "grad_norm": 0.06517433375120163, "global_step": 50673, "epoch": 478, "lr": 4.744815776517697e-07} +{"train_loss": 0.015385081991553307, "grad_norm": 0.08508381992578506, "global_step": 50674, "epoch": 478, "lr": 4.7407433042977456e-07} +{"train_loss": 0.013579994440078735, "grad_norm": 0.07195185869932175, "global_step": 50675, "epoch": 478, "lr": 4.7366725722033935e-07} +{"train_loss": 0.01777905784547329, "grad_norm": 0.1357709914445877, "global_step": 50676, "epoch": 478, "lr": 4.732603580249018e-07} +{"train_loss": 0.0158783420920372, "grad_norm": 0.06893008202314377, "global_step": 50677, "epoch": 478, "lr": 4.7285363284488847e-07} +{"train_loss": 0.013372335582971573, "grad_norm": 0.07474122196435928, "global_step": 50678, "epoch": 478, "lr": 4.724470816817317e-07} +{"train_loss": 0.013309594243764877, "grad_norm": 0.10919203609228134, "global_step": 50679, "epoch": 478, "lr": 4.7204070453685244e-07} +{"train_loss": 0.015121683478355408, "grad_norm": 0.0703105553984642, "global_step": 50680, "epoch": 478, "lr": 4.7163450141168296e-07} +{"train_loss": 0.016927950084209442, "grad_norm": 0.07398197054862976, "global_step": 50681, "epoch": 478, "lr": 4.712284723076554e-07} +{"train_loss": 0.014236727729439735, "grad_norm": 0.06744592636823654, "global_step": 50682, "epoch": 478, "lr": 4.7082261722618537e-07} +{"train_loss": 0.015243414789438248, "grad_norm": 0.07762123644351959, "global_step": 50683, "epoch": 478, "lr": 4.70416936168705e-07} +{"train_loss": 0.015835274010896683, "grad_norm": 0.09274943917989731, "global_step": 50684, "epoch": 478, "lr": 4.7001142913664643e-07} +{"train_loss": 0.01686679944396019, "grad_norm": 0.06601648032665253, "global_step": 50685, "epoch": 478, "lr": 4.696060961314197e-07} +{"train_loss": 0.014512963593006134, "grad_norm": 0.0776909664273262, "global_step": 50686, "epoch": 478, "lr": 4.6920093715446255e-07} +{"train_loss": 0.013036864809691906, "grad_norm": 0.06965036690235138, "global_step": 50687, "epoch": 478, "lr": 4.687959522071905e-07} +{"train_loss": 0.01318126916885376, "grad_norm": 0.0616624653339386, "global_step": 50688, "epoch": 478, "lr": 4.6839114129103023e-07} +{"train_loss": 0.014328105375170708, "grad_norm": 0.07183065265417099, "global_step": 50689, "epoch": 478, "lr": 4.6798650440739725e-07} +{"train_loss": 0.016098059713840485, "grad_norm": 0.13364842534065247, "global_step": 50690, "epoch": 478, "lr": 4.675820415577292e-07} +{"train_loss": 0.01461067795753479, "grad_norm": 0.10677530616521835, "global_step": 50691, "epoch": 478, "lr": 4.671777527434307e-07} +{"train_loss": 0.013917702250182629, "grad_norm": 0.09371419996023178, "global_step": 50692, "epoch": 478, "lr": 4.667736379659282e-07} +{"train_loss": 0.014850394800305367, "grad_norm": 0.055616166442632675, "global_step": 50693, "epoch": 478, "lr": 4.6636969722663735e-07} +{"train_loss": 0.01364358514547348, "grad_norm": 0.08819211274385452, "global_step": 50694, "epoch": 478, "lr": 4.659659305269903e-07} +{"train_loss": 0.016567785292863846, "grad_norm": 0.10728869587182999, "global_step": 50695, "epoch": 478, "lr": 4.655623378683915e-07} +{"train_loss": 0.016006920486688614, "grad_norm": 0.06635788083076477, "global_step": 50696, "epoch": 478, "lr": 4.651589192522676e-07} +{"train_loss": 0.01590564474463463, "grad_norm": 0.06906744837760925, "global_step": 50697, "epoch": 478, "lr": 4.647556746800341e-07} +{"train_loss": 0.013145066797733307, "grad_norm": 0.06887418776750565, "global_step": 50698, "epoch": 478, "lr": 4.643526041531066e-07} +{"train_loss": 0.014066159725189209, "grad_norm": 0.08959917724132538, "global_step": 50699, "epoch": 478, "lr": 4.639497076728949e-07} +{"train_loss": 0.018143512308597565, "grad_norm": 0.07190259546041489, "global_step": 50700, "epoch": 478, "lr": 4.6354698524082583e-07} +{"train_loss": 0.015565144829452038, "grad_norm": 0.11287621408700943, "global_step": 50701, "epoch": 478, "lr": 4.631444368583149e-07} +{"train_loss": 0.016495075076818466, "grad_norm": 0.14484964311122894, "global_step": 50702, "epoch": 478, "lr": 4.6274206252676644e-07} +{"train_loss": 0.015197142958641052, "grad_norm": 0.0918695405125618, "global_step": 50703, "epoch": 478, "lr": 4.6233986224759607e-07} +{"train_loss": 0.015613707713782787, "grad_norm": 0.08110377937555313, "global_step": 50704, "epoch": 478, "lr": 4.619378360222304e-07} +{"train_loss": 0.015899457037448883, "grad_norm": 0.08511398732662201, "global_step": 50705, "epoch": 478, "lr": 4.6153598385206276e-07} +{"train_loss": 0.013886057771742344, "grad_norm": 0.07864899933338165, "global_step": 50706, "epoch": 478, "lr": 4.6113430573851426e-07} +{"train_loss": 0.01592083089053631, "grad_norm": 0.1362987458705902, "global_step": 50707, "epoch": 478, "lr": 4.607328016830004e-07} +{"train_loss": 0.01262008585035801, "grad_norm": 0.04792274162173271, "global_step": 50708, "epoch": 478, "lr": 4.603314716869256e-07} +{"train_loss": 0.01715305633842945, "grad_norm": 0.0867120698094368, "global_step": 50709, "epoch": 478, "lr": 4.5993031575169986e-07} +{"train_loss": 0.013202786445617676, "grad_norm": 0.09172308444976807, "global_step": 50710, "epoch": 478, "lr": 4.595293338787332e-07} +{"train_loss": 0.012563792988657951, "grad_norm": 0.09524891525506973, "global_step": 50711, "epoch": 478, "lr": 4.5912852606944115e-07} +{"train_loss": 0.012836406007409096, "grad_norm": 0.08595603704452515, "global_step": 50712, "epoch": 478, "lr": 4.587278923252225e-07} +{"train_loss": 0.015118978917598724, "grad_norm": 0.061974771320819855, "global_step": 50713, "epoch": 478, "lr": 4.5832743264748734e-07} +{"train_loss": 0.013915589079260826, "grad_norm": 0.05179031938314438, "global_step": 50714, "epoch": 478, "lr": 4.5792714703765114e-07} +{"train_loss": 0.014963995665311813, "grad_norm": 0.09496867656707764, "global_step": 50715, "epoch": 478, "lr": 4.575270354971073e-07} +{"train_loss": 0.014876817353069782, "grad_norm": 0.10276217758655548, "global_step": 50716, "epoch": 478, "lr": 4.5712709802727126e-07} +{"train_loss": 0.014638010412454605, "grad_norm": 0.11576832830905914, "global_step": 50717, "epoch": 478, "lr": 4.56727334629542e-07} +{"train_loss": 0.012110710144042969, "grad_norm": 0.12014322727918625, "global_step": 50718, "epoch": 478, "lr": 4.56327745305335e-07} +{"train_loss": 0.01535039208829403, "grad_norm": 0.08739781379699707, "global_step": 50719, "epoch": 478, "lr": 4.5592833005603796e-07} +{"train_loss": 0.015164251439273357, "grad_norm": 0.07479149103164673, "global_step": 50720, "epoch": 478, "lr": 4.5552908888306655e-07} +{"train_loss": 0.01473336573690176, "grad_norm": 0.06875170767307281, "global_step": 50721, "epoch": 478, "lr": 4.551300217878196e-07} +{"train_loss": 0.015353919938206673, "grad_norm": 0.1252630203962326, "global_step": 50722, "epoch": 478, "lr": 4.54731128771696e-07} +{"train_loss": 0.016642719507217407, "grad_norm": 0.08657331019639969, "global_step": 50723, "epoch": 478, "lr": 4.5433240983610016e-07} +{"train_loss": 0.013579005375504494, "grad_norm": 0.10432608425617218, "global_step": 50724, "epoch": 478, "lr": 4.5393386498244205e-07} +{"train_loss": 0.015666835010051727, "grad_norm": 0.08962684124708176, "global_step": 50725, "epoch": 478, "lr": 4.5353549421210953e-07} +{"train_loss": 0.016661344096064568, "grad_norm": 0.09473308175802231, "global_step": 50726, "epoch": 478, "lr": 4.5313729752650137e-07} +{"train_loss": 0.014018958434462547, "grad_norm": 0.06579525768756866, "global_step": 50727, "epoch": 478, "lr": 4.5273927492702206e-07} +{"train_loss": 0.012872094288468361, "grad_norm": 0.06483186036348343, "global_step": 50728, "epoch": 478, "lr": 4.5234142641507047e-07} +{"train_loss": 0.016106419265270233, "grad_norm": 0.07944320887327194, "global_step": 50729, "epoch": 478, "lr": 4.5194375199204553e-07} +{"train_loss": 0.01504205260425806, "grad_norm": 0.08257589489221573, "global_step": 50730, "epoch": 478, "lr": 4.515462516593405e-07} +{"train_loss": 0.01510632038116455, "grad_norm": 0.07411520183086395, "global_step": 50731, "epoch": 478, "lr": 4.511489254183543e-07} +{"train_loss": 0.015336167998611927, "grad_norm": 0.08306334167718887, "global_step": 50732, "epoch": 478, "lr": 4.507517732704858e-07} +{"train_loss": 0.014328600838780403, "grad_norm": 0.06593184173107147, "global_step": 50733, "epoch": 478, "lr": 4.5035479521712274e-07} +{"train_loss": 0.014993131160736084, "grad_norm": 0.06608402729034424, "global_step": 50734, "epoch": 478, "lr": 4.499579912596641e-07} +{"train_loss": 0.015012705698609352, "grad_norm": 0.06254483759403229, "global_step": 50735, "epoch": 478, "lr": 4.4956136139950867e-07} +{"train_loss": 0.015631452202796936, "grad_norm": 0.0670284852385521, "global_step": 50736, "epoch": 478, "lr": 4.491649056380387e-07} +{"train_loss": 0.014144552871584892, "grad_norm": 0.10450801998376846, "global_step": 50737, "epoch": 478, "lr": 4.4876862397665863e-07} +{"train_loss": 0.013275062665343285, "grad_norm": 0.06708676367998123, "global_step": 50738, "epoch": 478, "lr": 4.4837251641675626e-07} +{"train_loss": 0.013095300644636154, "grad_norm": 0.08873113989830017, "global_step": 50739, "epoch": 478, "lr": 4.4797658295972487e-07} +{"train_loss": 0.015081770718097687, "grad_norm": 0.07468865066766739, "global_step": 50740, "epoch": 478, "lr": 4.475808236069523e-07} +{"train_loss": 0.017290577292442322, "grad_norm": 0.061543166637420654, "global_step": 50741, "epoch": 478, "lr": 4.4718523835982627e-07} +{"train_loss": 0.01483827643096447, "grad_norm": 0.06316976249217987, "global_step": 50742, "epoch": 478, "lr": 4.4678982721975126e-07} +{"train_loss": 0.01473331544548273, "grad_norm": 0.09742560982704163, "global_step": 50743, "epoch": 478, "lr": 4.4639459018810393e-07} +{"train_loss": 0.015668565407395363, "grad_norm": 0.06809753179550171, "global_step": 50744, "epoch": 478, "lr": 4.4599952726627203e-07} +{"train_loss": 0.013854341581463814, "grad_norm": 0.08417320996522903, "global_step": 50745, "epoch": 478, "lr": 4.4560463845564894e-07} +{"train_loss": 0.015263723209500313, "grad_norm": 0.051867663860321045, "global_step": 50746, "epoch": 478, "lr": 4.4520992375761685e-07} +{"train_loss": 0.012065334245562553, "grad_norm": 0.05874372646212578, "global_step": 50747, "epoch": 478, "lr": 4.448153831735691e-07} +{"train_loss": 0.015335620380938053, "grad_norm": 0.08454152196645737, "global_step": 50748, "epoch": 478, "lr": 4.4442101670489344e-07} +{"train_loss": 0.017138035967946053, "grad_norm": 0.09234882891178131, "global_step": 50749, "epoch": 478, "lr": 4.4402682435296663e-07} +{"train_loss": 0.018071824684739113, "grad_norm": 0.07379033416509628, "global_step": 50750, "epoch": 478, "lr": 4.436328061191819e-07} +{"train_loss": 0.015670988708734512, "grad_norm": 0.06092602014541626, "global_step": 50751, "epoch": 478, "lr": 4.43238962004916e-07} +{"train_loss": 0.016401872038841248, "grad_norm": 0.059948548674583435, "global_step": 50752, "epoch": 478, "lr": 4.4284529201155665e-07} +{"train_loss": 0.014312567189335823, "grad_norm": 0.14445438981056213, "global_step": 50753, "epoch": 478, "lr": 4.4245179614048613e-07} +{"train_loss": 0.016005443409085274, "grad_norm": 0.07546433806419373, "global_step": 50754, "epoch": 478, "lr": 4.420584743930922e-07} +{"train_loss": 0.013815149664878845, "grad_norm": 0.0864209532737732, "global_step": 50755, "epoch": 478, "lr": 4.4166532677075154e-07} +{"train_loss": 0.015162164345383644, "grad_norm": 0.06042717024683952, "global_step": 50756, "epoch": 478, "lr": 4.412723532748464e-07} +{"train_loss": 0.015088757500052452, "grad_norm": 0.09874259680509567, "global_step": 50757, "epoch": 478, "lr": 4.4087955390675897e-07} +{"train_loss": 0.014778046868741512, "grad_norm": 0.06555742770433426, "global_step": 50758, "epoch": 478, "lr": 4.4048692866786035e-07} +{"train_loss": 0.014320124872028828, "grad_norm": 0.09523748606443405, "global_step": 50759, "epoch": 478, "lr": 4.4009447755954944e-07} +{"train_loss": 0.015578340739011765, "grad_norm": 0.08061285316944122, "global_step": 50760, "epoch": 478, "lr": 4.3970220058318635e-07} +{"train_loss": 0.01569702848792076, "grad_norm": 0.10083670914173126, "global_step": 50761, "epoch": 478, "lr": 4.3931009774015874e-07} +{"train_loss": 0.01602628082036972, "grad_norm": 0.07784832268953323, "global_step": 50762, "epoch": 478, "lr": 4.3891816903184335e-07} +{"train_loss": 0.015340144746005535, "grad_norm": 0.10526877641677856, "global_step": 50763, "epoch": 478, "lr": 4.385264144596113e-07} +{"train_loss": 0.013809067197144032, "grad_norm": 0.06309762597084045, "global_step": 50764, "epoch": 478, "lr": 4.381348340248448e-07} +{"train_loss": 0.018115602433681488, "grad_norm": 0.08998401463031769, "global_step": 50765, "epoch": 478, "lr": 4.377434277289205e-07} +{"train_loss": 0.013815416023135185, "grad_norm": 0.0842757523059845, "global_step": 50766, "epoch": 478, "lr": 4.373521955732096e-07} +{"train_loss": 0.01545816008001566, "grad_norm": 0.11417181044816971, "global_step": 50767, "epoch": 478, "lr": 4.3696113755909427e-07} +{"train_loss": 0.015483666211366653, "grad_norm": 0.08288157731294632, "global_step": 50768, "epoch": 478, "lr": 4.365702536879346e-07} +{"train_loss": 0.013269942253828049, "grad_norm": 0.1248004212975502, "global_step": 50769, "epoch": 478, "lr": 4.3617954396111824e-07} +{"train_loss": 0.01562030054628849, "grad_norm": 0.12896686792373657, "global_step": 50770, "epoch": 478, "lr": 4.357890083800109e-07} +{"train_loss": 0.01525326631963253, "grad_norm": 0.07946842163801193, "global_step": 50771, "epoch": 478, "lr": 4.353986469459892e-07} +{"train_loss": 0.013938135467469692, "grad_norm": 0.11405432224273682, "global_step": 50772, "epoch": 478, "lr": 4.350084596604187e-07} +{"train_loss": 0.014948876161690591, "grad_norm": 0.10046317428350449, "global_step": 50773, "epoch": 478, "lr": 4.3461844652467607e-07, "val_loss": 0.019642528146505356} +{"train_loss": 0.014145627617835999, "grad_norm": 0.06314258277416229, "global_step": 50774, "epoch": 479, "lr": 4.342286075401214e-07} +{"train_loss": 0.015087660402059555, "grad_norm": 0.060444802045822144, "global_step": 50775, "epoch": 479, "lr": 4.3383894270813685e-07} +{"train_loss": 0.015854183584451675, "grad_norm": 0.07020241767168045, "global_step": 50776, "epoch": 479, "lr": 4.33449452030088e-07} +{"train_loss": 0.012918727472424507, "grad_norm": 0.06871166825294495, "global_step": 50777, "epoch": 479, "lr": 4.330601355073405e-07} +{"train_loss": 0.01715550385415554, "grad_norm": 0.13095501065254211, "global_step": 50778, "epoch": 479, "lr": 4.326709931412598e-07} +{"train_loss": 0.014873363077640533, "grad_norm": 0.10343778878450394, "global_step": 50779, "epoch": 479, "lr": 4.3228202493322267e-07} +{"train_loss": 0.014367698691785336, "grad_norm": 0.05691937357187271, "global_step": 50780, "epoch": 479, "lr": 4.318932308845891e-07} +{"train_loss": 0.015472659841179848, "grad_norm": 0.08847154676914215, "global_step": 50781, "epoch": 479, "lr": 4.315046109967247e-07} +{"train_loss": 0.01488950289785862, "grad_norm": 0.06825977563858032, "global_step": 50782, "epoch": 479, "lr": 4.31116165270995e-07} +{"train_loss": 0.014732751064002514, "grad_norm": 0.06157286837697029, "global_step": 50783, "epoch": 479, "lr": 4.3072789370876554e-07} +{"train_loss": 0.014895405620336533, "grad_norm": 0.11876681447029114, "global_step": 50784, "epoch": 479, "lr": 4.303397963114075e-07} +{"train_loss": 0.013566205278038979, "grad_norm": 0.09317851066589355, "global_step": 50785, "epoch": 479, "lr": 4.2995187308026984e-07} +{"train_loss": 0.01477092131972313, "grad_norm": 0.07737834006547928, "global_step": 50786, "epoch": 479, "lr": 4.295641240167292e-07} +{"train_loss": 0.015210539102554321, "grad_norm": 0.09432592242956161, "global_step": 50787, "epoch": 479, "lr": 4.2917654912214e-07} +{"train_loss": 0.01712030917406082, "grad_norm": 0.06443056464195251, "global_step": 50788, "epoch": 479, "lr": 4.287891483978679e-07} +{"train_loss": 0.017306197434663773, "grad_norm": 0.12446869164705276, "global_step": 50789, "epoch": 479, "lr": 4.2840192184527284e-07} +{"train_loss": 0.015388879925012589, "grad_norm": 0.09343210607767105, "global_step": 50790, "epoch": 479, "lr": 4.280148694657149e-07} +{"train_loss": 0.013233764097094536, "grad_norm": 0.059247542172670364, "global_step": 50791, "epoch": 479, "lr": 4.276279912605541e-07} +{"train_loss": 0.014225218445062637, "grad_norm": 0.11413843184709549, "global_step": 50792, "epoch": 479, "lr": 4.2724128723115044e-07} +{"train_loss": 0.015118066221475601, "grad_norm": 0.08704327791929245, "global_step": 50793, "epoch": 479, "lr": 4.2685475737886395e-07} +{"train_loss": 0.013755000196397305, "grad_norm": 0.11044242978096008, "global_step": 50794, "epoch": 479, "lr": 4.264684017050491e-07} +{"train_loss": 0.015531443059444427, "grad_norm": 0.07535234838724136, "global_step": 50795, "epoch": 479, "lr": 4.260822202110659e-07} +{"train_loss": 0.014866489917039871, "grad_norm": 0.058436211198568344, "global_step": 50796, "epoch": 479, "lr": 4.256962128982689e-07} +{"train_loss": 0.014862889423966408, "grad_norm": 0.06088920682668686, "global_step": 50797, "epoch": 479, "lr": 4.25310379768018e-07} +{"train_loss": 0.01458823587745428, "grad_norm": 0.06895075738430023, "global_step": 50798, "epoch": 479, "lr": 4.249247208216678e-07} +{"train_loss": 0.01562269777059555, "grad_norm": 0.16604174673557281, "global_step": 50799, "epoch": 479, "lr": 4.2453923606057265e-07} +{"train_loss": 0.016986239701509476, "grad_norm": 0.08537297695875168, "global_step": 50800, "epoch": 479, "lr": 4.241539254860871e-07} +{"train_loss": 0.01380106434226036, "grad_norm": 0.06724277138710022, "global_step": 50801, "epoch": 479, "lr": 4.237687890995601e-07} +{"train_loss": 0.017221828922629356, "grad_norm": 0.08380121737718582, "global_step": 50802, "epoch": 479, "lr": 4.233838269023516e-07} +{"train_loss": 0.013604124076664448, "grad_norm": 0.05294940993189812, "global_step": 50803, "epoch": 479, "lr": 4.2299903889581603e-07} +{"train_loss": 0.012924907729029655, "grad_norm": 0.06566952913999557, "global_step": 50804, "epoch": 479, "lr": 4.226144250812969e-07} +{"train_loss": 0.013705217279493809, "grad_norm": 0.07088381052017212, "global_step": 50805, "epoch": 479, "lr": 4.2222998546014857e-07} +{"train_loss": 0.015651505440473557, "grad_norm": 0.11388996243476868, "global_step": 50806, "epoch": 479, "lr": 4.2184572003373115e-07} +{"train_loss": 0.011577870696783066, "grad_norm": 0.05017964541912079, "global_step": 50807, "epoch": 479, "lr": 4.214616288033768e-07} +{"train_loss": 0.018195072188973427, "grad_norm": 0.08763254433870316, "global_step": 50808, "epoch": 479, "lr": 4.2107771177045674e-07} +{"train_loss": 0.016988014802336693, "grad_norm": 0.1444154977798462, "global_step": 50809, "epoch": 479, "lr": 4.2069396893629766e-07} +{"train_loss": 0.016263781115412712, "grad_norm": 0.07939130812883377, "global_step": 50810, "epoch": 479, "lr": 4.203104003022651e-07} +{"train_loss": 0.015694959089159966, "grad_norm": 0.08695938438177109, "global_step": 50811, "epoch": 479, "lr": 4.199270058696969e-07} +{"train_loss": 0.01433255709707737, "grad_norm": 0.12980252504348755, "global_step": 50812, "epoch": 479, "lr": 4.19543785639942e-07} +{"train_loss": 0.013662286102771759, "grad_norm": 0.06812101602554321, "global_step": 50813, "epoch": 479, "lr": 4.1916073961434933e-07} +{"train_loss": 0.01752784103155136, "grad_norm": 0.06527819484472275, "global_step": 50814, "epoch": 479, "lr": 4.1877786779426776e-07} +{"train_loss": 0.014603568241000175, "grad_norm": 0.15070199966430664, "global_step": 50815, "epoch": 479, "lr": 4.183951701810351e-07} +{"train_loss": 0.01539421547204256, "grad_norm": 0.06521260738372803, "global_step": 50816, "epoch": 479, "lr": 4.180126467760004e-07} +{"train_loss": 0.012582987546920776, "grad_norm": 0.05324868485331535, "global_step": 50817, "epoch": 479, "lr": 4.1763029758050686e-07} +{"train_loss": 0.01342119462788105, "grad_norm": 0.08439628034830093, "global_step": 50818, "epoch": 479, "lr": 4.172481225958924e-07} +{"train_loss": 0.016906576231122017, "grad_norm": 0.19631674885749817, "global_step": 50819, "epoch": 479, "lr": 4.168661218235115e-07} +{"train_loss": 0.01319381408393383, "grad_norm": 0.0851716473698616, "global_step": 50820, "epoch": 479, "lr": 4.1648429526470187e-07} +{"train_loss": 0.015058931894600391, "grad_norm": 0.10976710170507431, "global_step": 50821, "epoch": 479, "lr": 4.1610264292079594e-07} +{"train_loss": 0.016161229461431503, "grad_norm": 0.0668688639998436, "global_step": 50822, "epoch": 479, "lr": 4.157211647931425e-07} +{"train_loss": 0.014303360134363174, "grad_norm": 0.10638896375894547, "global_step": 50823, "epoch": 479, "lr": 4.15339860883085e-07} +{"train_loss": 0.014138268306851387, "grad_norm": 0.11056248098611832, "global_step": 50824, "epoch": 479, "lr": 4.149587311919612e-07} +{"train_loss": 0.01562919281423092, "grad_norm": 0.07163500785827637, "global_step": 50825, "epoch": 479, "lr": 4.1457777572110337e-07} +{"train_loss": 0.015382010489702225, "grad_norm": 0.08009105920791626, "global_step": 50826, "epoch": 479, "lr": 4.1419699447186045e-07} +{"train_loss": 0.015632132068276405, "grad_norm": 0.09086502343416214, "global_step": 50827, "epoch": 479, "lr": 4.138163874455592e-07} +{"train_loss": 0.014132375828921795, "grad_norm": 0.055434249341487885, "global_step": 50828, "epoch": 479, "lr": 4.134359546435429e-07} +{"train_loss": 0.013137181289494038, "grad_norm": 0.07501890510320663, "global_step": 50829, "epoch": 479, "lr": 4.1305569606714946e-07} +{"train_loss": 0.01391911506652832, "grad_norm": 0.10692478716373444, "global_step": 50830, "epoch": 479, "lr": 4.126756117177166e-07} +{"train_loss": 0.014330592006444931, "grad_norm": 0.07877335697412491, "global_step": 50831, "epoch": 479, "lr": 4.122957015965767e-07} +{"train_loss": 0.01622084155678749, "grad_norm": 0.08218937367200851, "global_step": 50832, "epoch": 479, "lr": 4.11915965705062e-07} +{"train_loss": 0.014574726112186909, "grad_norm": 0.07680997997522354, "global_step": 50833, "epoch": 479, "lr": 4.1153640404451023e-07} +{"train_loss": 0.012914754450321198, "grad_norm": 0.0667162612080574, "global_step": 50834, "epoch": 479, "lr": 4.111570166162537e-07} +{"train_loss": 0.014696608297526836, "grad_norm": 0.05237411707639694, "global_step": 50835, "epoch": 479, "lr": 4.1077780342162473e-07} +{"train_loss": 0.015258468687534332, "grad_norm": 0.06305529177188873, "global_step": 50836, "epoch": 479, "lr": 4.1039876446195556e-07} +{"train_loss": 0.012127991765737534, "grad_norm": 0.08844289928674698, "global_step": 50837, "epoch": 479, "lr": 4.10019899738584e-07} +{"train_loss": 0.017645560204982758, "grad_norm": 0.05971913784742355, "global_step": 50838, "epoch": 479, "lr": 4.096412092528368e-07} +{"train_loss": 0.015694836154580116, "grad_norm": 0.08700098097324371, "global_step": 50839, "epoch": 479, "lr": 4.0926269300603503e-07} +{"train_loss": 0.015026023611426353, "grad_norm": 0.06147732958197594, "global_step": 50840, "epoch": 479, "lr": 4.0888435099952217e-07} +{"train_loss": 0.013960660435259342, "grad_norm": 0.12923789024353027, "global_step": 50841, "epoch": 479, "lr": 4.085061832346304e-07} +{"train_loss": 0.015588398091495037, "grad_norm": 0.07362812757492065, "global_step": 50842, "epoch": 479, "lr": 4.0812818971266984e-07} +{"train_loss": 0.014406757429242134, "grad_norm": 0.0899210274219513, "global_step": 50843, "epoch": 479, "lr": 4.0775037043498943e-07} +{"train_loss": 0.014880284667015076, "grad_norm": 0.07257380336523056, "global_step": 50844, "epoch": 479, "lr": 4.0737272540290473e-07} +{"train_loss": 0.014032291248440742, "grad_norm": 0.09907880425453186, "global_step": 50845, "epoch": 479, "lr": 4.0699525461774247e-07} +{"train_loss": 0.015279106795787811, "grad_norm": 0.12507759034633636, "global_step": 50846, "epoch": 479, "lr": 4.0661795808082936e-07} +{"train_loss": 0.016368312761187553, "grad_norm": 0.07257266342639923, "global_step": 50847, "epoch": 479, "lr": 4.062408357934977e-07} +{"train_loss": 0.01330637652426958, "grad_norm": 0.05981118977069855, "global_step": 50848, "epoch": 479, "lr": 4.0586388775706865e-07} +{"train_loss": 0.015314990654587746, "grad_norm": 0.067350834608078, "global_step": 50849, "epoch": 479, "lr": 4.054871139728633e-07} +{"train_loss": 0.017217393964529037, "grad_norm": 0.09354060143232346, "global_step": 50850, "epoch": 479, "lr": 4.051105144422085e-07} +{"train_loss": 0.01563994772732258, "grad_norm": 0.0653839111328125, "global_step": 50851, "epoch": 479, "lr": 4.047340891664253e-07} +{"train_loss": 0.01283396314829588, "grad_norm": 0.11183229833841324, "global_step": 50852, "epoch": 479, "lr": 4.0435783814684046e-07} +{"train_loss": 0.01355480682104826, "grad_norm": 0.10962829738855362, "global_step": 50853, "epoch": 479, "lr": 4.039817613847696e-07} +{"train_loss": 0.011669648811221123, "grad_norm": 0.04891384392976761, "global_step": 50854, "epoch": 479, "lr": 4.03605858881545e-07} +{"train_loss": 0.015568142756819725, "grad_norm": 0.059996575117111206, "global_step": 50855, "epoch": 479, "lr": 4.0323013063848225e-07} +{"train_loss": 0.01602359674870968, "grad_norm": 0.08468642830848694, "global_step": 50856, "epoch": 479, "lr": 4.0285457665689143e-07} +{"train_loss": 0.016095008701086044, "grad_norm": 0.08261413872241974, "global_step": 50857, "epoch": 479, "lr": 4.024791969380992e-07} +{"train_loss": 0.012604044750332832, "grad_norm": 0.07090306282043457, "global_step": 50858, "epoch": 479, "lr": 4.0210399148343236e-07} +{"train_loss": 0.017698684707283974, "grad_norm": 0.07823008298873901, "global_step": 50859, "epoch": 479, "lr": 4.0172896029420093e-07} +{"train_loss": 0.01676865667104721, "grad_norm": 0.0800679549574852, "global_step": 50860, "epoch": 479, "lr": 4.013541033717205e-07} +{"train_loss": 0.016832947731018066, "grad_norm": 0.08833848685026169, "global_step": 50861, "epoch": 479, "lr": 4.009794207173234e-07} +{"train_loss": 0.014590785838663578, "grad_norm": 0.0884690135717392, "global_step": 50862, "epoch": 479, "lr": 4.00604912332303e-07} +{"train_loss": 0.015522311441600323, "grad_norm": 0.07179713994264603, "global_step": 50863, "epoch": 479, "lr": 4.0023057821798606e-07} +{"train_loss": 0.017412737011909485, "grad_norm": 0.06248936429619789, "global_step": 50864, "epoch": 479, "lr": 3.9985641837569364e-07} +{"train_loss": 0.013483229093253613, "grad_norm": 0.0633101612329483, "global_step": 50865, "epoch": 479, "lr": 3.994824328067359e-07} +{"train_loss": 0.013337205164134502, "grad_norm": 0.06569826602935791, "global_step": 50866, "epoch": 479, "lr": 3.991086215124229e-07} +{"train_loss": 0.018620755523443222, "grad_norm": 0.08218370378017426, "global_step": 50867, "epoch": 479, "lr": 3.9873498449407575e-07} +{"train_loss": 0.014963279478251934, "grad_norm": 0.0770568698644638, "global_step": 50868, "epoch": 479, "lr": 3.98361521752999e-07} +{"train_loss": 0.0149910319596529, "grad_norm": 0.07856029272079468, "global_step": 50869, "epoch": 479, "lr": 3.979882332905083e-07} +{"train_loss": 0.012108715251088142, "grad_norm": 0.05989845097064972, "global_step": 50870, "epoch": 479, "lr": 3.976151191079136e-07} +{"train_loss": 0.014968948438763618, "grad_norm": 0.07142750173807144, "global_step": 50871, "epoch": 479, "lr": 3.9724217920653616e-07} +{"train_loss": 0.015749070793390274, "grad_norm": 0.07047943025827408, "global_step": 50872, "epoch": 479, "lr": 3.9686941358766936e-07} +{"train_loss": 0.014272395521402359, "grad_norm": 0.06591969728469849, "global_step": 50873, "epoch": 479, "lr": 3.964968222526344e-07} +{"train_loss": 0.015320328995585442, "grad_norm": 0.10619303584098816, "global_step": 50874, "epoch": 479, "lr": 3.961244052027413e-07} +{"train_loss": 0.01566598378121853, "grad_norm": 0.1003129631280899, "global_step": 50875, "epoch": 479, "lr": 3.95752162439289e-07} +{"train_loss": 0.013847362250089645, "grad_norm": 0.06181439757347107, "global_step": 50876, "epoch": 479, "lr": 3.953800939635932e-07} +{"train_loss": 0.01547155436128378, "grad_norm": 0.059144094586372375, "global_step": 50877, "epoch": 479, "lr": 3.950081997769639e-07} +{"train_loss": 0.014811189845204353, "grad_norm": 0.06290840357542038, "global_step": 50878, "epoch": 479, "lr": 3.9463647988070005e-07} +{"train_loss": 0.014941161603860135, "grad_norm": 0.0843530148267746, "global_step": 50879, "epoch": 479, "lr": 3.9426493427611177e-07, "val_loss": 0.01899409480392933} +{"train_loss": 0.015331842936575413, "grad_norm": 0.11416946351528168, "global_step": 50880, "epoch": 480, "lr": 3.93893562964498e-07} +{"train_loss": 0.0141066275537014, "grad_norm": 0.11326304078102112, "global_step": 50881, "epoch": 480, "lr": 3.935223659471743e-07} +{"train_loss": 0.01811099611222744, "grad_norm": 0.08511784672737122, "global_step": 50882, "epoch": 480, "lr": 3.9315134322543967e-07} +{"train_loss": 0.014236848801374435, "grad_norm": 0.08252596110105515, "global_step": 50883, "epoch": 480, "lr": 3.927804948005931e-07} +{"train_loss": 0.013661381788551807, "grad_norm": 0.12203548848628998, "global_step": 50884, "epoch": 480, "lr": 3.9240982067395017e-07} +{"train_loss": 0.01430690847337246, "grad_norm": 0.08048328757286072, "global_step": 50885, "epoch": 480, "lr": 3.920393208468043e-07} +{"train_loss": 0.016133654862642288, "grad_norm": 0.06529507786035538, "global_step": 50886, "epoch": 480, "lr": 3.916689953204489e-07} +{"train_loss": 0.01652909815311432, "grad_norm": 0.11214431375265121, "global_step": 50887, "epoch": 480, "lr": 3.9129884409619955e-07} +{"train_loss": 0.013873565010726452, "grad_norm": 0.06557192653417587, "global_step": 50888, "epoch": 480, "lr": 3.909288671753553e-07} +{"train_loss": 0.015045981854200363, "grad_norm": 0.0964931771159172, "global_step": 50889, "epoch": 480, "lr": 3.905590645592094e-07} +{"train_loss": 0.015263601206243038, "grad_norm": 0.06243211403489113, "global_step": 50890, "epoch": 480, "lr": 3.9018943624906654e-07} +{"train_loss": 0.013824015855789185, "grad_norm": 0.08060812205076218, "global_step": 50891, "epoch": 480, "lr": 3.8981998224622006e-07} +{"train_loss": 0.01656651869416237, "grad_norm": 0.08293864876031876, "global_step": 50892, "epoch": 480, "lr": 3.8945070255197446e-07} +{"train_loss": 0.015386268496513367, "grad_norm": 0.0868031233549118, "global_step": 50893, "epoch": 480, "lr": 3.8908159716762317e-07} +{"train_loss": 0.01592077687382698, "grad_norm": 0.07749883830547333, "global_step": 50894, "epoch": 480, "lr": 3.887126660944651e-07} +{"train_loss": 0.0130893150344491, "grad_norm": 0.07335349172353745, "global_step": 50895, "epoch": 480, "lr": 3.883439093337937e-07} +{"train_loss": 0.015123981982469559, "grad_norm": 0.051046982407569885, "global_step": 50896, "epoch": 480, "lr": 3.8797532688690797e-07} +{"train_loss": 0.015120120719075203, "grad_norm": 0.05581997334957123, "global_step": 50897, "epoch": 480, "lr": 3.876069187550957e-07} +{"train_loss": 0.017802083864808083, "grad_norm": 0.08804892748594284, "global_step": 50898, "epoch": 480, "lr": 3.872386849396614e-07} +{"train_loss": 0.013734288513660431, "grad_norm": 0.07010684162378311, "global_step": 50899, "epoch": 480, "lr": 3.86870625441893e-07} +{"train_loss": 0.01438738964498043, "grad_norm": 0.09364353865385056, "global_step": 50900, "epoch": 480, "lr": 3.8650274026308384e-07} +{"train_loss": 0.014680626802146435, "grad_norm": 0.07767762243747711, "global_step": 50901, "epoch": 480, "lr": 3.8613502940453294e-07} +{"train_loss": 0.017082449048757553, "grad_norm": 0.10879439115524292, "global_step": 50902, "epoch": 480, "lr": 3.8576749286752813e-07} +{"train_loss": 0.016005046665668488, "grad_norm": 0.12714265286922455, "global_step": 50903, "epoch": 480, "lr": 3.8540013065335166e-07} +{"train_loss": 0.01597278192639351, "grad_norm": 0.08175090700387955, "global_step": 50904, "epoch": 480, "lr": 3.850329427633026e-07} +{"train_loss": 0.015448037534952164, "grad_norm": 0.07263275235891342, "global_step": 50905, "epoch": 480, "lr": 3.8466592919867985e-07} +{"train_loss": 0.013581735081970692, "grad_norm": 0.11990223079919815, "global_step": 50906, "epoch": 480, "lr": 3.842990899607546e-07} +{"train_loss": 0.012297545559704304, "grad_norm": 0.05807681381702423, "global_step": 50907, "epoch": 480, "lr": 3.839324250508258e-07} +{"train_loss": 0.015553001314401627, "grad_norm": 0.06167490407824516, "global_step": 50908, "epoch": 480, "lr": 3.8356593447018694e-07} +{"train_loss": 0.016394011676311493, "grad_norm": 0.0953637957572937, "global_step": 50909, "epoch": 480, "lr": 3.831996182201203e-07} +{"train_loss": 0.012252699583768845, "grad_norm": 0.09040512889623642, "global_step": 50910, "epoch": 480, "lr": 3.828334763019081e-07} +{"train_loss": 0.015849122777581215, "grad_norm": 0.06300438195466995, "global_step": 50911, "epoch": 480, "lr": 3.8246750871683834e-07} +{"train_loss": 0.013031666167080402, "grad_norm": 0.06379131227731705, "global_step": 50912, "epoch": 480, "lr": 3.8210171546620433e-07} +{"train_loss": 0.015530361793935299, "grad_norm": 0.07575522363185883, "global_step": 50913, "epoch": 480, "lr": 3.817360965512828e-07} +{"train_loss": 0.015434067696332932, "grad_norm": 0.06560998409986496, "global_step": 50914, "epoch": 480, "lr": 3.813706519733673e-07} +{"train_loss": 0.014382658526301384, "grad_norm": 0.09467656165361404, "global_step": 50915, "epoch": 480, "lr": 3.810053817337345e-07} +{"train_loss": 0.013990992680191994, "grad_norm": 0.08935631066560745, "global_step": 50916, "epoch": 480, "lr": 3.806402858336722e-07} +{"train_loss": 0.014117652550339699, "grad_norm": 0.06627832353115082, "global_step": 50917, "epoch": 480, "lr": 3.802753642744572e-07} +{"train_loss": 0.015152511186897755, "grad_norm": 0.08559145033359528, "global_step": 50918, "epoch": 480, "lr": 3.799106170573774e-07} +{"train_loss": 0.01457519456744194, "grad_norm": 0.08190006017684937, "global_step": 50919, "epoch": 480, "lr": 3.795460441837095e-07} +{"train_loss": 0.013912715017795563, "grad_norm": 0.10144539922475815, "global_step": 50920, "epoch": 480, "lr": 3.7918164565474146e-07} +{"train_loss": 0.014656947925686836, "grad_norm": 0.1226171925663948, "global_step": 50921, "epoch": 480, "lr": 3.7881742147174993e-07} +{"train_loss": 0.013523288071155548, "grad_norm": 0.11886914074420929, "global_step": 50922, "epoch": 480, "lr": 3.784533716360117e-07} +{"train_loss": 0.018752221018075943, "grad_norm": 0.10460364073514938, "global_step": 50923, "epoch": 480, "lr": 3.7808949614880907e-07} +{"train_loss": 0.014743820764124393, "grad_norm": 0.07211430370807648, "global_step": 50924, "epoch": 480, "lr": 3.777257950114188e-07} +{"train_loss": 0.01339871995151043, "grad_norm": 0.13859844207763672, "global_step": 50925, "epoch": 480, "lr": 3.773622682251232e-07} +{"train_loss": 0.015201536938548088, "grad_norm": 0.07781004905700684, "global_step": 50926, "epoch": 480, "lr": 3.769989157911935e-07} +{"train_loss": 0.011620947159826756, "grad_norm": 0.0632244274020195, "global_step": 50927, "epoch": 480, "lr": 3.76635737710912e-07} +{"train_loss": 0.018120503053069115, "grad_norm": 0.11916671693325043, "global_step": 50928, "epoch": 480, "lr": 3.7627273398554987e-07} +{"train_loss": 0.015627436339855194, "grad_norm": 0.11209101229906082, "global_step": 50929, "epoch": 480, "lr": 3.7590990461638387e-07} +{"train_loss": 0.014792980626225471, "grad_norm": 0.07113374769687653, "global_step": 50930, "epoch": 480, "lr": 3.755472496046908e-07} +{"train_loss": 0.017991110682487488, "grad_norm": 0.09275037795305252, "global_step": 50931, "epoch": 480, "lr": 3.751847689517418e-07} +{"train_loss": 0.015484550967812538, "grad_norm": 0.08715697377920151, "global_step": 50932, "epoch": 480, "lr": 3.748224626588137e-07} +{"train_loss": 0.014672045595943928, "grad_norm": 0.1273145228624344, "global_step": 50933, "epoch": 480, "lr": 3.744603307271777e-07} +{"train_loss": 0.01601329818367958, "grad_norm": 0.08901271969079971, "global_step": 50934, "epoch": 480, "lr": 3.740983731581049e-07} +{"train_loss": 0.01529998704791069, "grad_norm": 0.06503486633300781, "global_step": 50935, "epoch": 480, "lr": 3.737365899528722e-07} +{"train_loss": 0.013678660616278648, "grad_norm": 0.05898372456431389, "global_step": 50936, "epoch": 480, "lr": 3.7337498111274514e-07} +{"train_loss": 0.016816575080156326, "grad_norm": 0.12598171830177307, "global_step": 50937, "epoch": 480, "lr": 3.7301354663900057e-07} +{"train_loss": 0.013869062066078186, "grad_norm": 0.08196456730365753, "global_step": 50938, "epoch": 480, "lr": 3.726522865328985e-07} +{"train_loss": 0.01408359408378601, "grad_norm": 0.06324409693479538, "global_step": 50939, "epoch": 480, "lr": 3.722912007957158e-07} +{"train_loss": 0.01324850507080555, "grad_norm": 0.06174511834979057, "global_step": 50940, "epoch": 480, "lr": 3.71930289428718e-07} +{"train_loss": 0.01534869521856308, "grad_norm": 0.09426205605268478, "global_step": 50941, "epoch": 480, "lr": 3.7156955243317636e-07} +{"train_loss": 0.013592458330094814, "grad_norm": 0.05864249914884567, "global_step": 50942, "epoch": 480, "lr": 3.71208989810351e-07} +{"train_loss": 0.01639309898018837, "grad_norm": 0.07005500793457031, "global_step": 50943, "epoch": 480, "lr": 3.708486015615242e-07} +{"train_loss": 0.013643029145896435, "grad_norm": 0.07444148510694504, "global_step": 50944, "epoch": 480, "lr": 3.704883876879506e-07} +{"train_loss": 0.01592029444873333, "grad_norm": 0.08747300505638123, "global_step": 50945, "epoch": 480, "lr": 3.7012834819089013e-07} +{"train_loss": 0.016616005450487137, "grad_norm": 0.11122146248817444, "global_step": 50946, "epoch": 480, "lr": 3.697684830716197e-07} +{"train_loss": 0.015177404507994652, "grad_norm": 0.07536735385656357, "global_step": 50947, "epoch": 480, "lr": 3.6940879233139935e-07} +{"train_loss": 0.014922340400516987, "grad_norm": 0.07884179800748825, "global_step": 50948, "epoch": 480, "lr": 3.690492759714892e-07} +{"train_loss": 0.013917598873376846, "grad_norm": 0.10709177702665329, "global_step": 50949, "epoch": 480, "lr": 3.6868993399316045e-07} +{"train_loss": 0.01585102640092373, "grad_norm": 0.11500820517539978, "global_step": 50950, "epoch": 480, "lr": 3.6833076639767317e-07} +{"train_loss": 0.015418712981045246, "grad_norm": 0.08207927644252777, "global_step": 50951, "epoch": 480, "lr": 3.67971773186282e-07} +{"train_loss": 0.013156440109014511, "grad_norm": 0.059376657009124756, "global_step": 50952, "epoch": 480, "lr": 3.6761295436025245e-07} +{"train_loss": 0.015656813979148865, "grad_norm": 0.07671647518873215, "global_step": 50953, "epoch": 480, "lr": 3.672543099208503e-07} +{"train_loss": 0.013338984921574593, "grad_norm": 0.11530855298042297, "global_step": 50954, "epoch": 480, "lr": 3.6689583986933006e-07} +{"train_loss": 0.015566429123282433, "grad_norm": 0.1311052143573761, "global_step": 50955, "epoch": 480, "lr": 3.665375442069574e-07} +{"train_loss": 0.014767480082809925, "grad_norm": 0.07325401157140732, "global_step": 50956, "epoch": 480, "lr": 3.661794229349813e-07} +{"train_loss": 0.014048184268176556, "grad_norm": 0.07043467462062836, "global_step": 50957, "epoch": 480, "lr": 3.6582147605466744e-07} +{"train_loss": 0.015165247023105621, "grad_norm": 0.0964529737830162, "global_step": 50958, "epoch": 480, "lr": 3.6546370356727033e-07} +{"train_loss": 0.012460004538297653, "grad_norm": 0.0840359628200531, "global_step": 50959, "epoch": 480, "lr": 3.651061054740501e-07} +{"train_loss": 0.013968952000141144, "grad_norm": 0.056717175990343094, "global_step": 50960, "epoch": 480, "lr": 3.647486817762613e-07} +{"train_loss": 0.015236989594995975, "grad_norm": 0.10940851271152496, "global_step": 50961, "epoch": 480, "lr": 3.643914324751585e-07} +{"train_loss": 0.01623459719121456, "grad_norm": 0.07712358981370926, "global_step": 50962, "epoch": 480, "lr": 3.6403435757200176e-07} +{"train_loss": 0.016703542321920395, "grad_norm": 0.1346338838338852, "global_step": 50963, "epoch": 480, "lr": 3.636774570680401e-07} +{"train_loss": 0.011798818595707417, "grad_norm": 0.06598470360040665, "global_step": 50964, "epoch": 480, "lr": 3.633207309645281e-07} +{"train_loss": 0.011722567491233349, "grad_norm": 0.07381041347980499, "global_step": 50965, "epoch": 480, "lr": 3.629641792627203e-07} +{"train_loss": 0.01577027328312397, "grad_norm": 0.12623071670532227, "global_step": 50966, "epoch": 480, "lr": 3.6260780196387125e-07} +{"train_loss": 0.014903769828379154, "grad_norm": 0.06354465335607529, "global_step": 50967, "epoch": 480, "lr": 3.622515990692299e-07} +{"train_loss": 0.011767694726586342, "grad_norm": 0.062483541667461395, "global_step": 50968, "epoch": 480, "lr": 3.6189557058005086e-07} +{"train_loss": 0.014793668873608112, "grad_norm": 0.13515323400497437, "global_step": 50969, "epoch": 480, "lr": 3.615397164975831e-07} +{"train_loss": 0.013717064633965492, "grad_norm": 0.09218066930770874, "global_step": 50970, "epoch": 480, "lr": 3.611840368230812e-07} +{"train_loss": 0.01751641556620598, "grad_norm": 0.10254323482513428, "global_step": 50971, "epoch": 480, "lr": 3.608285315577886e-07} +{"train_loss": 0.014661077409982681, "grad_norm": 0.07759726047515869, "global_step": 50972, "epoch": 480, "lr": 3.604732007029543e-07} +{"train_loss": 0.014270314946770668, "grad_norm": 0.11072227358818054, "global_step": 50973, "epoch": 480, "lr": 3.601180442598329e-07} +{"train_loss": 0.014871570281684399, "grad_norm": 0.11068334430456161, "global_step": 50974, "epoch": 480, "lr": 3.597630622296677e-07} +{"train_loss": 0.01679803803563118, "grad_norm": 0.07321574538946152, "global_step": 50975, "epoch": 480, "lr": 3.5940825461370785e-07} +{"train_loss": 0.013180190697312355, "grad_norm": 0.07738261669874191, "global_step": 50976, "epoch": 480, "lr": 3.5905362141319674e-07} +{"train_loss": 0.015206621028482914, "grad_norm": 0.10586589574813843, "global_step": 50977, "epoch": 480, "lr": 3.58699162629389e-07} +{"train_loss": 0.01633453741669655, "grad_norm": 0.12099529802799225, "global_step": 50978, "epoch": 480, "lr": 3.583448782635168e-07} +{"train_loss": 0.015386861748993397, "grad_norm": 0.08401165902614594, "global_step": 50979, "epoch": 480, "lr": 3.5799076831683487e-07} +{"train_loss": 0.015770278871059418, "grad_norm": 0.10927528142929077, "global_step": 50980, "epoch": 480, "lr": 3.576368327905866e-07} +{"train_loss": 0.01586124859750271, "grad_norm": 0.07810013741254807, "global_step": 50981, "epoch": 480, "lr": 3.572830716860154e-07} +{"train_loss": 0.018591206520795822, "grad_norm": 0.11166072636842728, "global_step": 50982, "epoch": 480, "lr": 3.5692948500435364e-07} +{"train_loss": 0.01477049570530653, "grad_norm": 0.06606888771057129, "global_step": 50983, "epoch": 480, "lr": 3.5657607274686143e-07} +{"train_loss": 0.014315878972411156, "grad_norm": 0.0665988102555275, "global_step": 50984, "epoch": 480, "lr": 3.5622283491477113e-07} +{"train_loss": 0.014947889451780971, "grad_norm": 0.09137731045484543, "global_step": 50985, "epoch": 480, "lr": 3.558697715093207e-07, "train_batch_mse_error": 0.013366125524044037, "val_loss": 0.019002538174390793} +{"train_loss": 0.014479359611868858, "grad_norm": 0.06673764437437057, "global_step": 50986, "epoch": 481, "lr": 3.5551688253175343e-07} +{"train_loss": 0.016214843839406967, "grad_norm": 0.08080486208200455, "global_step": 50987, "epoch": 481, "lr": 3.5516416798331286e-07} +{"train_loss": 0.01654522493481636, "grad_norm": 0.09663999825716019, "global_step": 50988, "epoch": 481, "lr": 3.5481162786523136e-07} +{"train_loss": 0.01586814410984516, "grad_norm": 0.07326789945363998, "global_step": 50989, "epoch": 481, "lr": 3.5445926217875236e-07} +{"train_loss": 0.013764165341854095, "grad_norm": 0.08137254416942596, "global_step": 50990, "epoch": 481, "lr": 3.5410707092511377e-07} +{"train_loss": 0.015379857271909714, "grad_norm": 0.11025384813547134, "global_step": 50991, "epoch": 481, "lr": 3.5375505410555896e-07} +{"train_loss": 0.019844748079776764, "grad_norm": 0.12411829084157944, "global_step": 50992, "epoch": 481, "lr": 3.534032117213093e-07} +{"train_loss": 0.015760647132992744, "grad_norm": 0.07159318774938583, "global_step": 50993, "epoch": 481, "lr": 3.530515437736137e-07} +{"train_loss": 0.014539415016770363, "grad_norm": 0.09054850041866302, "global_step": 50994, "epoch": 481, "lr": 3.5270005026370454e-07} +{"train_loss": 0.016692176461219788, "grad_norm": 0.11254200339317322, "global_step": 50995, "epoch": 481, "lr": 3.523487311928142e-07} +{"train_loss": 0.013303177431225777, "grad_norm": 0.05220865085721016, "global_step": 50996, "epoch": 481, "lr": 3.5199758656218604e-07} +{"train_loss": 0.017022399231791496, "grad_norm": 0.07938230782747269, "global_step": 50997, "epoch": 481, "lr": 3.516466163730414e-07} +{"train_loss": 0.014013579115271568, "grad_norm": 0.10285783559083939, "global_step": 50998, "epoch": 481, "lr": 3.5129582062661815e-07} +{"train_loss": 0.014634821563959122, "grad_norm": 0.08701090514659882, "global_step": 50999, "epoch": 481, "lr": 3.5094519932415417e-07} +{"train_loss": 0.0146078672260046, "grad_norm": 0.05855010449886322, "global_step": 51000, "epoch": 481, "lr": 3.505947524668707e-07} +{"train_loss": 0.015829460695385933, "grad_norm": 0.08320151269435883, "global_step": 51001, "epoch": 481, "lr": 3.5024448005601116e-07} +{"train_loss": 0.013991010375320911, "grad_norm": 0.10531249642372131, "global_step": 51002, "epoch": 481, "lr": 3.498943820927969e-07} +{"train_loss": 0.013940625824034214, "grad_norm": 0.07024472206830978, "global_step": 51003, "epoch": 481, "lr": 3.4954445857846576e-07} +{"train_loss": 0.01428112480789423, "grad_norm": 0.06918641924858093, "global_step": 51004, "epoch": 481, "lr": 3.4919470951423894e-07} +{"train_loss": 0.012552795931696892, "grad_norm": 0.06827401369810104, "global_step": 51005, "epoch": 481, "lr": 3.4884513490134886e-07} +{"train_loss": 0.016606315970420837, "grad_norm": 0.05913715064525604, "global_step": 51006, "epoch": 481, "lr": 3.484957347410223e-07} +{"train_loss": 0.014358662068843842, "grad_norm": 0.09564803540706635, "global_step": 51007, "epoch": 481, "lr": 3.481465090344971e-07} +{"train_loss": 0.01225193776190281, "grad_norm": 0.060704220086336136, "global_step": 51008, "epoch": 481, "lr": 3.47797457782989e-07} +{"train_loss": 0.015122979879379272, "grad_norm": 0.1401222050189972, "global_step": 51009, "epoch": 481, "lr": 3.474485809877248e-07} +{"train_loss": 0.015982646495103836, "grad_norm": 0.07010180503129959, "global_step": 51010, "epoch": 481, "lr": 3.470998786499313e-07} +{"train_loss": 0.016833817586302757, "grad_norm": 0.1949407309293747, "global_step": 51011, "epoch": 481, "lr": 3.467513507708409e-07} +{"train_loss": 0.014727369882166386, "grad_norm": 0.07565056532621384, "global_step": 51012, "epoch": 481, "lr": 3.464029973516691e-07} +{"train_loss": 0.01408034935593605, "grad_norm": 0.12229727953672409, "global_step": 51013, "epoch": 481, "lr": 3.460548183936374e-07} +{"train_loss": 0.017056560143828392, "grad_norm": 0.16640682518482208, "global_step": 51014, "epoch": 481, "lr": 3.4570681389798354e-07} +{"train_loss": 0.01306604128330946, "grad_norm": 0.05767691135406494, "global_step": 51015, "epoch": 481, "lr": 3.4535898386592324e-07} +{"train_loss": 0.016795482486486435, "grad_norm": 0.07109402120113373, "global_step": 51016, "epoch": 481, "lr": 3.450113282986722e-07} +{"train_loss": 0.015925996005535126, "grad_norm": 0.10747691988945007, "global_step": 51017, "epoch": 481, "lr": 3.4466384719745725e-07} +{"train_loss": 0.014315946027636528, "grad_norm": 0.07534851878881454, "global_step": 51018, "epoch": 481, "lr": 3.4431654056350513e-07} +{"train_loss": 0.01378375943750143, "grad_norm": 0.10489187389612198, "global_step": 51019, "epoch": 481, "lr": 3.439694083980205e-07} +{"train_loss": 0.013936884701251984, "grad_norm": 0.13094954192638397, "global_step": 51020, "epoch": 481, "lr": 3.436224507022412e-07} +{"train_loss": 0.015511381439864635, "grad_norm": 0.1055118665099144, "global_step": 51021, "epoch": 481, "lr": 3.4327566747737183e-07} +{"train_loss": 0.015523909591138363, "grad_norm": 0.0920042172074318, "global_step": 51022, "epoch": 481, "lr": 3.429290587246392e-07} +{"train_loss": 0.01600532978773117, "grad_norm": 0.0745103731751442, "global_step": 51023, "epoch": 481, "lr": 3.42582624445259e-07} +{"train_loss": 0.016363153234124184, "grad_norm": 0.11031737178564072, "global_step": 51024, "epoch": 481, "lr": 3.422363646404469e-07} +{"train_loss": 0.01384769193828106, "grad_norm": 0.07177221775054932, "global_step": 51025, "epoch": 481, "lr": 3.418902793114187e-07} +{"train_loss": 0.016292810440063477, "grad_norm": 0.07469147443771362, "global_step": 51026, "epoch": 481, "lr": 3.415443684593955e-07} +{"train_loss": 0.015970364212989807, "grad_norm": 0.07517239451408386, "global_step": 51027, "epoch": 481, "lr": 3.4119863208558757e-07} +{"train_loss": 0.015505041927099228, "grad_norm": 0.08533352613449097, "global_step": 51028, "epoch": 481, "lr": 3.4085307019121605e-07} +{"train_loss": 0.015267737209796906, "grad_norm": 0.11564905196428299, "global_step": 51029, "epoch": 481, "lr": 3.405076827774856e-07} +{"train_loss": 0.010920450091362, "grad_norm": 0.05992971733212471, "global_step": 51030, "epoch": 481, "lr": 3.4016246984561184e-07} +{"train_loss": 0.015253065153956413, "grad_norm": 0.11306829750537872, "global_step": 51031, "epoch": 481, "lr": 3.398174313968161e-07} +{"train_loss": 0.015528143383562565, "grad_norm": 0.06207932531833649, "global_step": 51032, "epoch": 481, "lr": 3.3947256743230847e-07} +{"train_loss": 0.018215831369161606, "grad_norm": 0.06777971982955933, "global_step": 51033, "epoch": 481, "lr": 3.3912787795329357e-07} +{"train_loss": 0.015478379093110561, "grad_norm": 0.08572875708341599, "global_step": 51034, "epoch": 481, "lr": 3.387833629609816e-07} +{"train_loss": 0.014359915629029274, "grad_norm": 0.07208321988582611, "global_step": 51035, "epoch": 481, "lr": 3.3843902245659364e-07} +{"train_loss": 0.016366640105843544, "grad_norm": 0.124455526471138, "global_step": 51036, "epoch": 481, "lr": 3.3809485644132885e-07} +{"train_loss": 0.01668817177414894, "grad_norm": 0.09837045520544052, "global_step": 51037, "epoch": 481, "lr": 3.377508649164085e-07} +{"train_loss": 0.013494920916855335, "grad_norm": 0.12242050468921661, "global_step": 51038, "epoch": 481, "lr": 3.3740704788303157e-07} +{"train_loss": 0.015045477077364922, "grad_norm": 0.08337820321321487, "global_step": 51039, "epoch": 481, "lr": 3.370634053424082e-07} +{"train_loss": 0.014938665553927422, "grad_norm": 0.06127787008881569, "global_step": 51040, "epoch": 481, "lr": 3.3671993729574303e-07} +{"train_loss": 0.014297256246209145, "grad_norm": 0.09736639261245728, "global_step": 51041, "epoch": 481, "lr": 3.363766437442517e-07} +{"train_loss": 0.01438777893781662, "grad_norm": 0.0716504156589508, "global_step": 51042, "epoch": 481, "lr": 3.3603352468912774e-07} +{"train_loss": 0.01445217989385128, "grad_norm": 0.06632725149393082, "global_step": 51043, "epoch": 481, "lr": 3.3569058013158683e-07} +{"train_loss": 0.01544981636106968, "grad_norm": 0.06772024184465408, "global_step": 51044, "epoch": 481, "lr": 3.3534781007283355e-07} +{"train_loss": 0.017392471432685852, "grad_norm": 0.15122786164283752, "global_step": 51045, "epoch": 481, "lr": 3.35005214514067e-07} +{"train_loss": 0.014773381873965263, "grad_norm": 0.11779316514730453, "global_step": 51046, "epoch": 481, "lr": 3.3466279345649165e-07} +{"train_loss": 0.016180014237761497, "grad_norm": 0.1088748425245285, "global_step": 51047, "epoch": 481, "lr": 3.343205469013122e-07} +{"train_loss": 0.014540541917085648, "grad_norm": 0.10907737165689468, "global_step": 51048, "epoch": 481, "lr": 3.3397847484973324e-07} +{"train_loss": 0.0164203979074955, "grad_norm": 0.07636845856904984, "global_step": 51049, "epoch": 481, "lr": 3.3363657730295927e-07} +{"train_loss": 0.015802806243300438, "grad_norm": 0.061729516834020615, "global_step": 51050, "epoch": 481, "lr": 3.332948542621783e-07} +{"train_loss": 0.013997670263051987, "grad_norm": 0.07570330798625946, "global_step": 51051, "epoch": 481, "lr": 3.3295330572860604e-07} +{"train_loss": 0.014161242172122002, "grad_norm": 0.07247612625360489, "global_step": 51052, "epoch": 481, "lr": 3.3261193170343596e-07} +{"train_loss": 0.01615915074944496, "grad_norm": 0.13345423340797424, "global_step": 51053, "epoch": 481, "lr": 3.3227073218786154e-07} +{"train_loss": 0.01723865047097206, "grad_norm": 0.06332820653915405, "global_step": 51054, "epoch": 481, "lr": 3.319297071830929e-07} +{"train_loss": 0.01506731566041708, "grad_norm": 0.05249343812465668, "global_step": 51055, "epoch": 481, "lr": 3.315888566903236e-07} +{"train_loss": 0.014313718304038048, "grad_norm": 0.058173440396785736, "global_step": 51056, "epoch": 481, "lr": 3.3124818071075257e-07} +{"train_loss": 0.014180486090481281, "grad_norm": 0.08059094101190567, "global_step": 51057, "epoch": 481, "lr": 3.3090767924556786e-07} +{"train_loss": 0.012698417529463768, "grad_norm": 0.06843946129083633, "global_step": 51058, "epoch": 481, "lr": 3.305673522959796e-07} +{"train_loss": 0.015611500479280949, "grad_norm": 0.06356970220804214, "global_step": 51059, "epoch": 481, "lr": 3.302271998631756e-07} +{"train_loss": 0.015751302242279053, "grad_norm": 0.07559385895729065, "global_step": 51060, "epoch": 481, "lr": 3.2988722194834955e-07} +{"train_loss": 0.014360260218381882, "grad_norm": 0.11973133683204651, "global_step": 51061, "epoch": 481, "lr": 3.2954741855270587e-07} +{"train_loss": 0.014469385147094727, "grad_norm": 0.05643720552325249, "global_step": 51062, "epoch": 481, "lr": 3.2920778967742703e-07} +{"train_loss": 0.01507937628775835, "grad_norm": 0.06605375558137894, "global_step": 51063, "epoch": 481, "lr": 3.2886833532370656e-07} +{"train_loss": 0.01389225572347641, "grad_norm": 0.10607884079217911, "global_step": 51064, "epoch": 481, "lr": 3.2852905549274337e-07} +{"train_loss": 0.016110148280858994, "grad_norm": 0.10139025747776031, "global_step": 51065, "epoch": 481, "lr": 3.281899501857311e-07} +{"train_loss": 0.016188638284802437, "grad_norm": 0.11927954107522964, "global_step": 51066, "epoch": 481, "lr": 3.2785101940385753e-07} +{"train_loss": 0.01484186016023159, "grad_norm": 0.08761761337518692, "global_step": 51067, "epoch": 481, "lr": 3.2751226314831076e-07} +{"train_loss": 0.014042289927601814, "grad_norm": 0.0700092762708664, "global_step": 51068, "epoch": 481, "lr": 3.2717368142028414e-07} +{"train_loss": 0.014430699869990349, "grad_norm": 0.06056297942996025, "global_step": 51069, "epoch": 481, "lr": 3.2683527422097124e-07} +{"train_loss": 0.012897145934402943, "grad_norm": 0.0589105598628521, "global_step": 51070, "epoch": 481, "lr": 3.264970415515489e-07} +{"train_loss": 0.01420475635677576, "grad_norm": 0.06339457631111145, "global_step": 51071, "epoch": 481, "lr": 3.261589834132217e-07} +{"train_loss": 0.015339208766818047, "grad_norm": 0.08456927537918091, "global_step": 51072, "epoch": 481, "lr": 3.2582109980716645e-07} +{"train_loss": 0.012370482087135315, "grad_norm": 0.07406658679246902, "global_step": 51073, "epoch": 481, "lr": 3.254833907345711e-07} +{"train_loss": 0.014703442342579365, "grad_norm": 0.10789556801319122, "global_step": 51074, "epoch": 481, "lr": 3.2514585619662364e-07} +{"train_loss": 0.013778883032500744, "grad_norm": 0.05357806012034416, "global_step": 51075, "epoch": 481, "lr": 3.248084961945119e-07} +{"train_loss": 0.016525674611330032, "grad_norm": 0.06662691384553909, "global_step": 51076, "epoch": 481, "lr": 3.2447131072942394e-07} +{"train_loss": 0.01521901786327362, "grad_norm": 0.0700293555855751, "global_step": 51077, "epoch": 481, "lr": 3.241342998025365e-07} +{"train_loss": 0.016266178339719772, "grad_norm": 0.09669096767902374, "global_step": 51078, "epoch": 481, "lr": 3.237974634150376e-07} +{"train_loss": 0.01396524254232645, "grad_norm": 0.050597816705703735, "global_step": 51079, "epoch": 481, "lr": 3.234608015681151e-07} +{"train_loss": 0.014796353876590729, "grad_norm": 0.06193864345550537, "global_step": 51080, "epoch": 481, "lr": 3.2312431426294034e-07} +{"train_loss": 0.013822436332702637, "grad_norm": 0.09290590137243271, "global_step": 51081, "epoch": 481, "lr": 3.2278800150070674e-07} +{"train_loss": 0.01532592624425888, "grad_norm": 0.09353083372116089, "global_step": 51082, "epoch": 481, "lr": 3.224518632825913e-07} +{"train_loss": 0.013793069869279861, "grad_norm": 0.06413856148719788, "global_step": 51083, "epoch": 481, "lr": 3.221158996097762e-07} +{"train_loss": 0.015198608860373497, "grad_norm": 0.105209581553936, "global_step": 51084, "epoch": 481, "lr": 3.2178011048344394e-07} +{"train_loss": 0.013692593201994896, "grad_norm": 0.10773645341396332, "global_step": 51085, "epoch": 481, "lr": 3.214444959047713e-07} +{"train_loss": 0.014558617025613785, "grad_norm": 0.08557592332363129, "global_step": 51086, "epoch": 481, "lr": 3.211090558749352e-07} +{"train_loss": 0.014578474685549736, "grad_norm": 0.0797937735915184, "global_step": 51087, "epoch": 481, "lr": 3.20773790395118e-07} +{"train_loss": 0.012259101495146751, "grad_norm": 0.09191340953111649, "global_step": 51088, "epoch": 481, "lr": 3.2043869946649653e-07} +{"train_loss": 0.014475190080702305, "grad_norm": 0.06954041868448257, "global_step": 51089, "epoch": 481, "lr": 3.201037830902476e-07} +{"train_loss": 0.012302994728088379, "grad_norm": 0.0905173271894455, "global_step": 51090, "epoch": 481, "lr": 3.197690412675536e-07} +{"train_loss": 0.014938037482283588, "grad_norm": 0.08058269321918488, "global_step": 51091, "epoch": 481, "lr": 3.1943447399958027e-07, "val_loss": 0.01959299109876156} +{"train_loss": 0.013959024101495743, "grad_norm": 0.06341195851564407, "global_step": 51092, "epoch": 482, "lr": 3.1910008128751e-07} +{"train_loss": 0.014032399281859398, "grad_norm": 0.12765352427959442, "global_step": 51093, "epoch": 482, "lr": 3.1876586313251964e-07} +{"train_loss": 0.016405228525400162, "grad_norm": 0.07413651049137115, "global_step": 51094, "epoch": 482, "lr": 3.184318195357749e-07} +{"train_loss": 0.013447915203869343, "grad_norm": 0.07941791415214539, "global_step": 51095, "epoch": 482, "lr": 3.1809795049845824e-07} +{"train_loss": 0.01496409997344017, "grad_norm": 0.056755099445581436, "global_step": 51096, "epoch": 482, "lr": 3.177642560217409e-07} +{"train_loss": 0.015405835583806038, "grad_norm": 0.09220641851425171, "global_step": 51097, "epoch": 482, "lr": 3.1743073610678854e-07} +{"train_loss": 0.01576978527009487, "grad_norm": 0.0854516476392746, "global_step": 51098, "epoch": 482, "lr": 3.170973907547836e-07} +{"train_loss": 0.013137186877429485, "grad_norm": 0.061311133205890656, "global_step": 51099, "epoch": 482, "lr": 3.167642199668863e-07} +{"train_loss": 0.014075970277190208, "grad_norm": 0.07852417230606079, "global_step": 51100, "epoch": 482, "lr": 3.1643122374427903e-07} +{"train_loss": 0.016225405037403107, "grad_norm": 0.0907488539814949, "global_step": 51101, "epoch": 482, "lr": 3.160984020881219e-07} +{"train_loss": 0.013659419491887093, "grad_norm": 0.09176743030548096, "global_step": 51102, "epoch": 482, "lr": 3.1576575499958626e-07} +{"train_loss": 0.012498809024691582, "grad_norm": 0.061449117958545685, "global_step": 51103, "epoch": 482, "lr": 3.1543328247984894e-07} +{"train_loss": 0.012960385531187057, "grad_norm": 0.09463836997747421, "global_step": 51104, "epoch": 482, "lr": 3.151009845300645e-07} +{"train_loss": 0.015865951776504517, "grad_norm": 0.09188809990882874, "global_step": 51105, "epoch": 482, "lr": 3.1476886115141545e-07} +{"train_loss": 0.013264136388897896, "grad_norm": 0.07760151475667953, "global_step": 51106, "epoch": 482, "lr": 3.1443691234505636e-07} +{"train_loss": 0.014531437307596207, "grad_norm": 0.061254046857357025, "global_step": 51107, "epoch": 482, "lr": 3.141051381121585e-07} +{"train_loss": 0.014228248968720436, "grad_norm": 0.05986151471734047, "global_step": 51108, "epoch": 482, "lr": 3.137735384538876e-07} +{"train_loss": 0.011743934825062752, "grad_norm": 0.07028131932020187, "global_step": 51109, "epoch": 482, "lr": 3.134421133714094e-07} +{"train_loss": 0.014511572197079659, "grad_norm": 0.075650155544281, "global_step": 51110, "epoch": 482, "lr": 3.1311086286588407e-07} +{"train_loss": 0.01481546275317669, "grad_norm": 0.06861884146928787, "global_step": 51111, "epoch": 482, "lr": 3.1277978693848296e-07} +{"train_loss": 0.015095176175236702, "grad_norm": 0.06826843321323395, "global_step": 51112, "epoch": 482, "lr": 3.124488855903607e-07} +{"train_loss": 0.01422127801924944, "grad_norm": 0.09305594116449356, "global_step": 51113, "epoch": 482, "lr": 3.121181588226885e-07} +{"train_loss": 0.014200234785676003, "grad_norm": 0.054079048335552216, "global_step": 51114, "epoch": 482, "lr": 3.1178760663662653e-07} +{"train_loss": 0.014005307108163834, "grad_norm": 0.06209387257695198, "global_step": 51115, "epoch": 482, "lr": 3.1145722903332955e-07} +{"train_loss": 0.01580275595188141, "grad_norm": 0.10899914056062698, "global_step": 51116, "epoch": 482, "lr": 3.1112702601396315e-07} +{"train_loss": 0.015339165925979614, "grad_norm": 0.07365807145833969, "global_step": 51117, "epoch": 482, "lr": 3.1079699757969317e-07} +{"train_loss": 0.017655152827501297, "grad_norm": 0.0756622850894928, "global_step": 51118, "epoch": 482, "lr": 3.104671437316686e-07} +{"train_loss": 0.016484305262565613, "grad_norm": 0.06682158261537552, "global_step": 51119, "epoch": 482, "lr": 3.1013746447104975e-07} +{"train_loss": 0.01592104136943817, "grad_norm": 0.12190405279397964, "global_step": 51120, "epoch": 482, "lr": 3.0980795979900777e-07} +{"train_loss": 0.016519926488399506, "grad_norm": 0.06629990041255951, "global_step": 51121, "epoch": 482, "lr": 3.094786297166807e-07} +{"train_loss": 0.014225184917449951, "grad_norm": 0.06590704619884491, "global_step": 51122, "epoch": 482, "lr": 3.0914947422523986e-07} +{"train_loss": 0.016809672117233276, "grad_norm": 0.10527080297470093, "global_step": 51123, "epoch": 482, "lr": 3.0882049332583983e-07} +{"train_loss": 0.013154979795217514, "grad_norm": 0.06495817750692368, "global_step": 51124, "epoch": 482, "lr": 3.0849168701962973e-07} +{"train_loss": 0.013673873618245125, "grad_norm": 0.062404293566942215, "global_step": 51125, "epoch": 482, "lr": 3.0816305530777524e-07} +{"train_loss": 0.014682681299746037, "grad_norm": 0.07032723724842072, "global_step": 51126, "epoch": 482, "lr": 3.078345981914199e-07} +{"train_loss": 0.015020952560007572, "grad_norm": 0.10307720303535461, "global_step": 51127, "epoch": 482, "lr": 3.075063156717295e-07} +{"train_loss": 0.01596461422741413, "grad_norm": 0.10617688298225403, "global_step": 51128, "epoch": 482, "lr": 3.071782077498475e-07} +{"train_loss": 0.015179065987467766, "grad_norm": 0.060741107910871506, "global_step": 51129, "epoch": 482, "lr": 3.068502744269286e-07} +{"train_loss": 0.016421129927039146, "grad_norm": 0.1726546436548233, "global_step": 51130, "epoch": 482, "lr": 3.0652251570412736e-07} +{"train_loss": 0.014306247234344482, "grad_norm": 0.06252125650644302, "global_step": 51131, "epoch": 482, "lr": 3.061949315825985e-07} +{"train_loss": 0.015046438202261925, "grad_norm": 0.08602310717105865, "global_step": 51132, "epoch": 482, "lr": 3.0586752206348547e-07} +{"train_loss": 0.018646765500307083, "grad_norm": 0.09475922584533691, "global_step": 51133, "epoch": 482, "lr": 3.0554028714794844e-07} +{"train_loss": 0.01381203904747963, "grad_norm": 0.06918317824602127, "global_step": 51134, "epoch": 482, "lr": 3.052132268371255e-07} +{"train_loss": 0.01461133360862732, "grad_norm": 0.08866985887289047, "global_step": 51135, "epoch": 482, "lr": 3.0488634113217117e-07} +{"train_loss": 0.014302991330623627, "grad_norm": 0.05841325968503952, "global_step": 51136, "epoch": 482, "lr": 3.045596300342346e-07} +{"train_loss": 0.014702586457133293, "grad_norm": 0.11773618310689926, "global_step": 51137, "epoch": 482, "lr": 3.0423309354447036e-07} +{"train_loss": 0.012496009469032288, "grad_norm": 0.07182967662811279, "global_step": 51138, "epoch": 482, "lr": 3.039067316640165e-07} +{"train_loss": 0.014980526641011238, "grad_norm": 0.06570303440093994, "global_step": 51139, "epoch": 482, "lr": 3.035805443940165e-07} +{"train_loss": 0.014966106042265892, "grad_norm": 0.10212157666683197, "global_step": 51140, "epoch": 482, "lr": 3.03254531735625e-07} +{"train_loss": 0.013453862629830837, "grad_norm": 0.05518251657485962, "global_step": 51141, "epoch": 482, "lr": 3.0292869368998557e-07} +{"train_loss": 0.015523197129368782, "grad_norm": 0.06476938724517822, "global_step": 51142, "epoch": 482, "lr": 3.0260303025824166e-07} +{"train_loss": 0.016527311876416206, "grad_norm": 0.11207785457372665, "global_step": 51143, "epoch": 482, "lr": 3.0227754144153684e-07} +{"train_loss": 0.01256688591092825, "grad_norm": 0.05488864332437515, "global_step": 51144, "epoch": 482, "lr": 3.019522272410202e-07} +{"train_loss": 0.015502391383051872, "grad_norm": 0.05827591195702553, "global_step": 51145, "epoch": 482, "lr": 3.016270876578242e-07} +{"train_loss": 0.017863648012280464, "grad_norm": 0.08654104173183441, "global_step": 51146, "epoch": 482, "lr": 3.013021226930979e-07} +{"train_loss": 0.01726188324391842, "grad_norm": 0.10070456564426422, "global_step": 51147, "epoch": 482, "lr": 3.0097733234798474e-07} +{"train_loss": 0.015606755390763283, "grad_norm": 0.07414145767688751, "global_step": 51148, "epoch": 482, "lr": 3.006527166236228e-07} +{"train_loss": 0.017016587778925896, "grad_norm": 0.13469594717025757, "global_step": 51149, "epoch": 482, "lr": 3.0032827552115564e-07} +{"train_loss": 0.013961710967123508, "grad_norm": 0.07083747535943985, "global_step": 51150, "epoch": 482, "lr": 3.0000400904171555e-07} +{"train_loss": 0.016573242843151093, "grad_norm": 0.10428453236818314, "global_step": 51151, "epoch": 482, "lr": 2.9967991718645727e-07} +{"train_loss": 0.016034038737416267, "grad_norm": 0.08961036056280136, "global_step": 51152, "epoch": 482, "lr": 2.993559999565021e-07} +{"train_loss": 0.012814247980713844, "grad_norm": 0.08643878996372223, "global_step": 51153, "epoch": 482, "lr": 2.9903225735299355e-07} +{"train_loss": 0.014016102068126202, "grad_norm": 0.1495763063430786, "global_step": 51154, "epoch": 482, "lr": 2.987086893770752e-07} +{"train_loss": 0.01547420583665371, "grad_norm": 0.09014599025249481, "global_step": 51155, "epoch": 482, "lr": 2.9838529602988496e-07} +{"train_loss": 0.011202935129404068, "grad_norm": 0.04845694452524185, "global_step": 51156, "epoch": 482, "lr": 2.980620773125442e-07} +{"train_loss": 0.014049213379621506, "grad_norm": 0.054172929376363754, "global_step": 51157, "epoch": 482, "lr": 2.97739033226202e-07} +{"train_loss": 0.014060769230127335, "grad_norm": 0.06676489859819412, "global_step": 51158, "epoch": 482, "lr": 2.9741616377199634e-07} +{"train_loss": 0.014561374671757221, "grad_norm": 0.06965900212526321, "global_step": 51159, "epoch": 482, "lr": 2.970934689510485e-07} +{"train_loss": 0.014788005501031876, "grad_norm": 0.06394153833389282, "global_step": 51160, "epoch": 482, "lr": 2.967709487644965e-07} +{"train_loss": 0.015720216557383537, "grad_norm": 0.05904201790690422, "global_step": 51161, "epoch": 482, "lr": 2.964486032134839e-07} +{"train_loss": 0.01565338671207428, "grad_norm": 0.08773703873157501, "global_step": 51162, "epoch": 482, "lr": 2.9612643229913193e-07} +{"train_loss": 0.01692892052233219, "grad_norm": 0.11556234955787659, "global_step": 51163, "epoch": 482, "lr": 2.9580443602257315e-07} +{"train_loss": 0.015230403281748295, "grad_norm": 0.06408815830945969, "global_step": 51164, "epoch": 482, "lr": 2.954826143849454e-07} +{"train_loss": 0.014648482203483582, "grad_norm": 0.07746360450983047, "global_step": 51165, "epoch": 482, "lr": 2.9516096738737566e-07} +{"train_loss": 0.014586793258786201, "grad_norm": 0.0951918438076973, "global_step": 51166, "epoch": 482, "lr": 2.9483949503099074e-07} +{"train_loss": 0.013282597064971924, "grad_norm": 0.07213716208934784, "global_step": 51167, "epoch": 482, "lr": 2.9451819731692864e-07} +{"train_loss": 0.015689067542552948, "grad_norm": 0.06173653155565262, "global_step": 51168, "epoch": 482, "lr": 2.941970742463107e-07} +{"train_loss": 0.013780666515231133, "grad_norm": 0.13767793774604797, "global_step": 51169, "epoch": 482, "lr": 2.9387612582026936e-07} +{"train_loss": 0.014995124191045761, "grad_norm": 0.07694297283887863, "global_step": 51170, "epoch": 482, "lr": 2.935553520399259e-07} +{"train_loss": 0.01073798630386591, "grad_norm": 0.05651068314909935, "global_step": 51171, "epoch": 482, "lr": 2.9323475290641834e-07} +{"train_loss": 0.01566362753510475, "grad_norm": 0.09460330754518509, "global_step": 51172, "epoch": 482, "lr": 2.929143284208624e-07} +{"train_loss": 0.013163596391677856, "grad_norm": 0.10471484065055847, "global_step": 51173, "epoch": 482, "lr": 2.925940785843906e-07} +{"train_loss": 0.016596263274550438, "grad_norm": 0.07168354094028473, "global_step": 51174, "epoch": 482, "lr": 2.9227400339812414e-07} +{"train_loss": 0.014238450676202774, "grad_norm": 0.08629090338945389, "global_step": 51175, "epoch": 482, "lr": 2.919541028631956e-07} +{"train_loss": 0.016764964908361435, "grad_norm": 0.10698222368955612, "global_step": 51176, "epoch": 482, "lr": 2.9163437698071504e-07} +{"train_loss": 0.01634790003299713, "grad_norm": 0.08167452365159988, "global_step": 51177, "epoch": 482, "lr": 2.91314825751815e-07} +{"train_loss": 0.016189171001315117, "grad_norm": 0.05540182814002037, "global_step": 51178, "epoch": 482, "lr": 2.9099544917762234e-07} +{"train_loss": 0.01712290570139885, "grad_norm": 0.10027210414409637, "global_step": 51179, "epoch": 482, "lr": 2.906762472592528e-07} +{"train_loss": 0.015917610377073288, "grad_norm": 0.072090283036232, "global_step": 51180, "epoch": 482, "lr": 2.9035721999782216e-07} +{"train_loss": 0.014891964383423328, "grad_norm": 0.08602922409772873, "global_step": 51181, "epoch": 482, "lr": 2.900383673944629e-07} +{"train_loss": 0.014113976620137691, "grad_norm": 0.1039450541138649, "global_step": 51182, "epoch": 482, "lr": 2.897196894502907e-07} +{"train_loss": 0.015044017694890499, "grad_norm": 0.06205208972096443, "global_step": 51183, "epoch": 482, "lr": 2.8940118616642143e-07} +{"train_loss": 0.01418537087738514, "grad_norm": 0.12060479074716568, "global_step": 51184, "epoch": 482, "lr": 2.890828575439819e-07} +{"train_loss": 0.01756313070654869, "grad_norm": 0.06905996799468994, "global_step": 51185, "epoch": 482, "lr": 2.8876470358408794e-07} +{"train_loss": 0.016354883089661598, "grad_norm": 0.0812850147485733, "global_step": 51186, "epoch": 482, "lr": 2.8844672428785525e-07} +{"train_loss": 0.01785147935152054, "grad_norm": 0.1007334291934967, "global_step": 51187, "epoch": 482, "lr": 2.881289196563941e-07} +{"train_loss": 0.016836397349834442, "grad_norm": 0.07409331202507019, "global_step": 51188, "epoch": 482, "lr": 2.8781128969083695e-07} +{"train_loss": 0.012981991283595562, "grad_norm": 0.05943140387535095, "global_step": 51189, "epoch": 482, "lr": 2.874938343922884e-07} +{"train_loss": 0.015151618048548698, "grad_norm": 0.10134634375572205, "global_step": 51190, "epoch": 482, "lr": 2.871765537618698e-07} +{"train_loss": 0.01366267167031765, "grad_norm": 0.1123049259185791, "global_step": 51191, "epoch": 482, "lr": 2.868594478006914e-07} +{"train_loss": 0.015457839705049992, "grad_norm": 0.09922917932271957, "global_step": 51192, "epoch": 482, "lr": 2.86542516509869e-07} +{"train_loss": 0.013134531676769257, "grad_norm": 0.09193426370620728, "global_step": 51193, "epoch": 482, "lr": 2.8622575989051825e-07} +{"train_loss": 0.015952855348587036, "grad_norm": 0.07470310479402542, "global_step": 51194, "epoch": 482, "lr": 2.8590917794374397e-07} +{"train_loss": 0.015089437365531921, "grad_norm": 0.10712294280529022, "global_step": 51195, "epoch": 482, "lr": 2.855927706706729e-07} +{"train_loss": 0.016827143728733063, "grad_norm": 0.09876430779695511, "global_step": 51196, "epoch": 482, "lr": 2.852765380724043e-07} +{"train_loss": 0.014937912263327613, "grad_norm": 0.08265286684036255, "global_step": 51197, "epoch": 482, "lr": 2.849604801500538e-07, "val_loss": 0.01950577087700367} +{"train_loss": 0.0175358597189188, "grad_norm": 0.08580256253480911, "global_step": 51198, "epoch": 483, "lr": 2.846445969047373e-07} +{"train_loss": 0.014507698826491833, "grad_norm": 0.05744956061244011, "global_step": 51199, "epoch": 483, "lr": 2.843288883375539e-07} +{"train_loss": 0.012104073539376259, "grad_norm": 0.09869100153446198, "global_step": 51200, "epoch": 483, "lr": 2.840133544496137e-07} +{"train_loss": 0.01742035523056984, "grad_norm": 0.09723295271396637, "global_step": 51201, "epoch": 483, "lr": 2.836979952420327e-07} +{"train_loss": 0.014329894445836544, "grad_norm": 0.1407146006822586, "global_step": 51202, "epoch": 483, "lr": 2.833828107159209e-07} +{"train_loss": 0.015718311071395874, "grad_norm": 0.10607407987117767, "global_step": 51203, "epoch": 483, "lr": 2.830678008723775e-07} +{"train_loss": 0.01560885924845934, "grad_norm": 0.09282498061656952, "global_step": 51204, "epoch": 483, "lr": 2.8275296571251276e-07} +{"train_loss": 0.015430357307195663, "grad_norm": 0.08288607001304626, "global_step": 51205, "epoch": 483, "lr": 2.824383052374313e-07} +{"train_loss": 0.015442957170307636, "grad_norm": 0.12085049599409103, "global_step": 51206, "epoch": 483, "lr": 2.8212381944823784e-07} +{"train_loss": 0.013062769547104836, "grad_norm": 0.06169948726892471, "global_step": 51207, "epoch": 483, "lr": 2.818095083460426e-07} +{"train_loss": 0.01610158197581768, "grad_norm": 0.06226474791765213, "global_step": 51208, "epoch": 483, "lr": 2.814953719319502e-07} +{"train_loss": 0.013380010612308979, "grad_norm": 0.10536600649356842, "global_step": 51209, "epoch": 483, "lr": 2.811814102070598e-07} +{"train_loss": 0.014791127294301987, "grad_norm": 0.06965769827365875, "global_step": 51210, "epoch": 483, "lr": 2.808676231724705e-07} +{"train_loss": 0.017573390156030655, "grad_norm": 0.11821266263723373, "global_step": 51211, "epoch": 483, "lr": 2.805540108292926e-07} +{"train_loss": 0.014382313936948776, "grad_norm": 0.10279307514429092, "global_step": 51212, "epoch": 483, "lr": 2.8024057317862505e-07} +{"train_loss": 0.014255369082093239, "grad_norm": 0.06381194293498993, "global_step": 51213, "epoch": 483, "lr": 2.799273102215727e-07} +{"train_loss": 0.01649545133113861, "grad_norm": 0.08824451267719269, "global_step": 51214, "epoch": 483, "lr": 2.796142219592346e-07} +{"train_loss": 0.01398560032248497, "grad_norm": 0.06794867664575577, "global_step": 51215, "epoch": 483, "lr": 2.7930130839270427e-07} +{"train_loss": 0.014256754890084267, "grad_norm": 0.061675213277339935, "global_step": 51216, "epoch": 483, "lr": 2.789885695230865e-07} +{"train_loss": 0.01688956841826439, "grad_norm": 0.09477952122688293, "global_step": 51217, "epoch": 483, "lr": 2.786760053514803e-07} +{"train_loss": 0.011653326451778412, "grad_norm": 0.0748494565486908, "global_step": 51218, "epoch": 483, "lr": 2.7836361587898483e-07} +{"train_loss": 0.015275469981133938, "grad_norm": 0.10256566852331161, "global_step": 51219, "epoch": 483, "lr": 2.780514011066937e-07} +{"train_loss": 0.014605788514018059, "grad_norm": 0.08811470866203308, "global_step": 51220, "epoch": 483, "lr": 2.777393610357115e-07} +{"train_loss": 0.01544084120541811, "grad_norm": 0.060069598257541656, "global_step": 51221, "epoch": 483, "lr": 2.7742749566712636e-07} +{"train_loss": 0.015558161772787571, "grad_norm": 0.06528971344232559, "global_step": 51222, "epoch": 483, "lr": 2.771158050020373e-07} +{"train_loss": 0.014350632205605507, "grad_norm": 0.09242796152830124, "global_step": 51223, "epoch": 483, "lr": 2.768042890415379e-07} +{"train_loss": 0.013029396533966064, "grad_norm": 0.09056929498910904, "global_step": 51224, "epoch": 483, "lr": 2.764929477867217e-07} +{"train_loss": 0.014843170531094074, "grad_norm": 0.07281414419412613, "global_step": 51225, "epoch": 483, "lr": 2.7618178123869354e-07} +{"train_loss": 0.01578974351286888, "grad_norm": 0.08030865341424942, "global_step": 51226, "epoch": 483, "lr": 2.758707893985357e-07} +{"train_loss": 0.012920577079057693, "grad_norm": 0.08341893553733826, "global_step": 51227, "epoch": 483, "lr": 2.755599722673363e-07} +{"train_loss": 0.014006414450705051, "grad_norm": 0.08474230766296387, "global_step": 51228, "epoch": 483, "lr": 2.752493298462e-07} +{"train_loss": 0.013614507392048836, "grad_norm": 0.1268681287765503, "global_step": 51229, "epoch": 483, "lr": 2.7493886213621476e-07} +{"train_loss": 0.017069410532712936, "grad_norm": 0.08514165133237839, "global_step": 51230, "epoch": 483, "lr": 2.746285691384631e-07} +{"train_loss": 0.01463920995593071, "grad_norm": 0.08327825367450714, "global_step": 51231, "epoch": 483, "lr": 2.7431845085404974e-07} +{"train_loss": 0.01361253671348095, "grad_norm": 0.06452183425426483, "global_step": 51232, "epoch": 483, "lr": 2.740085072840515e-07} +{"train_loss": 0.014853543601930141, "grad_norm": 0.10377433151006699, "global_step": 51233, "epoch": 483, "lr": 2.7369873842956196e-07} +{"train_loss": 0.015007393434643745, "grad_norm": 0.061249107122421265, "global_step": 51234, "epoch": 483, "lr": 2.733891442916692e-07} +{"train_loss": 0.017613381147384644, "grad_norm": 0.1081884577870369, "global_step": 51235, "epoch": 483, "lr": 2.730797248714612e-07} +{"train_loss": 0.01503976434469223, "grad_norm": 0.08193762600421906, "global_step": 51236, "epoch": 483, "lr": 2.727704801700315e-07} +{"train_loss": 0.014952602796256542, "grad_norm": 0.147373229265213, "global_step": 51237, "epoch": 483, "lr": 2.724614101884571e-07} +{"train_loss": 0.016107764095067978, "grad_norm": 0.0832998976111412, "global_step": 51238, "epoch": 483, "lr": 2.721525149278259e-07} +{"train_loss": 0.014559616334736347, "grad_norm": 0.09632263332605362, "global_step": 51239, "epoch": 483, "lr": 2.71843794389226e-07} +{"train_loss": 0.014502773061394691, "grad_norm": 0.08591342717409134, "global_step": 51240, "epoch": 483, "lr": 2.715352485737399e-07} +{"train_loss": 0.016597045585513115, "grad_norm": 0.11738482117652893, "global_step": 51241, "epoch": 483, "lr": 2.7122687748245004e-07} +{"train_loss": 0.014873260632157326, "grad_norm": 0.09797964245080948, "global_step": 51242, "epoch": 483, "lr": 2.7091868111644435e-07} +{"train_loss": 0.015012497082352638, "grad_norm": 0.11136379092931747, "global_step": 51243, "epoch": 483, "lr": 2.706106594768054e-07} +{"train_loss": 0.015134625136852264, "grad_norm": 0.08132275193929672, "global_step": 51244, "epoch": 483, "lr": 2.7030281256461566e-07} +{"train_loss": 0.015856904909014702, "grad_norm": 0.06747730821371078, "global_step": 51245, "epoch": 483, "lr": 2.6999514038095196e-07} +{"train_loss": 0.016684118658304214, "grad_norm": 0.06073250621557236, "global_step": 51246, "epoch": 483, "lr": 2.696876429268969e-07} +{"train_loss": 0.015053845942020416, "grad_norm": 0.08842815458774567, "global_step": 51247, "epoch": 483, "lr": 2.6938032020353274e-07} +{"train_loss": 0.013118581846356392, "grad_norm": 0.12500576674938202, "global_step": 51248, "epoch": 483, "lr": 2.690731722119422e-07} +{"train_loss": 0.012634875252842903, "grad_norm": 0.06123877316713333, "global_step": 51249, "epoch": 483, "lr": 2.687661989531964e-07} +{"train_loss": 0.016377784311771393, "grad_norm": 0.10505664348602295, "global_step": 51250, "epoch": 483, "lr": 2.684594004283836e-07} +{"train_loss": 0.014624801464378834, "grad_norm": 0.07503543794155121, "global_step": 51251, "epoch": 483, "lr": 2.6815277663857495e-07} +{"train_loss": 0.01578593999147415, "grad_norm": 0.13171790540218353, "global_step": 51252, "epoch": 483, "lr": 2.6784632758484753e-07} +{"train_loss": 0.01395563967525959, "grad_norm": 0.06682086735963821, "global_step": 51253, "epoch": 483, "lr": 2.6754005326828366e-07} +{"train_loss": 0.015621302649378777, "grad_norm": 0.06694035232067108, "global_step": 51254, "epoch": 483, "lr": 2.6723395368995485e-07} +{"train_loss": 0.014554057270288467, "grad_norm": 0.08049174398183823, "global_step": 51255, "epoch": 483, "lr": 2.6692802885093794e-07} +{"train_loss": 0.015290508046746254, "grad_norm": 0.07914840430021286, "global_step": 51256, "epoch": 483, "lr": 2.6662227875230427e-07} +{"train_loss": 0.013825426809489727, "grad_norm": 0.0597805418074131, "global_step": 51257, "epoch": 483, "lr": 2.663167033951364e-07} +{"train_loss": 0.013324564322829247, "grad_norm": 0.0782603770494461, "global_step": 51258, "epoch": 483, "lr": 2.660113027804945e-07} +{"train_loss": 0.01472553238272667, "grad_norm": 0.1119343489408493, "global_step": 51259, "epoch": 483, "lr": 2.657060769094666e-07} +{"train_loss": 0.014573950320482254, "grad_norm": 0.07048632949590683, "global_step": 51260, "epoch": 483, "lr": 2.6540102578311853e-07} +{"train_loss": 0.01293342374265194, "grad_norm": 0.061380188912153244, "global_step": 51261, "epoch": 483, "lr": 2.6509614940252723e-07} +{"train_loss": 0.01589076779782772, "grad_norm": 0.13884320855140686, "global_step": 51262, "epoch": 483, "lr": 2.6479144776875295e-07} +{"train_loss": 0.014871124178171158, "grad_norm": 0.07324255257844925, "global_step": 51263, "epoch": 483, "lr": 2.644869208828726e-07} +{"train_loss": 0.01513610128313303, "grad_norm": 0.06002281978726387, "global_step": 51264, "epoch": 483, "lr": 2.6418256874595203e-07} +{"train_loss": 0.013990789651870728, "grad_norm": 0.06079469248652458, "global_step": 51265, "epoch": 483, "lr": 2.638783913590681e-07} +{"train_loss": 0.014246433973312378, "grad_norm": 0.06163853406906128, "global_step": 51266, "epoch": 483, "lr": 2.635743887232867e-07} +{"train_loss": 0.01615864783525467, "grad_norm": 0.07138733565807343, "global_step": 51267, "epoch": 483, "lr": 2.6327056083967903e-07} +{"train_loss": 0.019178807735443115, "grad_norm": 0.15113110840320587, "global_step": 51268, "epoch": 483, "lr": 2.6296690770929997e-07} +{"train_loss": 0.015800783410668373, "grad_norm": 0.18888723850250244, "global_step": 51269, "epoch": 483, "lr": 2.6266342933323197e-07} +{"train_loss": 0.016621679067611694, "grad_norm": 0.07426145672798157, "global_step": 51270, "epoch": 483, "lr": 2.623601257125297e-07} +{"train_loss": 0.013385096564888954, "grad_norm": 0.11621281504631042, "global_step": 51271, "epoch": 483, "lr": 2.6205699684826444e-07} +{"train_loss": 0.014909602701663971, "grad_norm": 0.07359862327575684, "global_step": 51272, "epoch": 483, "lr": 2.617540427415022e-07} +{"train_loss": 0.013996630907058716, "grad_norm": 0.06955555081367493, "global_step": 51273, "epoch": 483, "lr": 2.614512633933086e-07} +{"train_loss": 0.014502937905490398, "grad_norm": 0.06009485945105553, "global_step": 51274, "epoch": 483, "lr": 2.61148658804744e-07} +{"train_loss": 0.015452814288437366, "grad_norm": 0.0563998743891716, "global_step": 51275, "epoch": 483, "lr": 2.608462289768632e-07} +{"train_loss": 0.016426460817456245, "grad_norm": 0.09822108596563339, "global_step": 51276, "epoch": 483, "lr": 2.6054397391074294e-07} +{"train_loss": 0.01628202572464943, "grad_norm": 0.09720101952552795, "global_step": 51277, "epoch": 483, "lr": 2.6024189360744353e-07} +{"train_loss": 0.015223726630210876, "grad_norm": 0.09959852695465088, "global_step": 51278, "epoch": 483, "lr": 2.5993998806801976e-07} +{"train_loss": 0.014816241338849068, "grad_norm": 0.07841663807630539, "global_step": 51279, "epoch": 483, "lr": 2.596382572935374e-07} +{"train_loss": 0.016297150403261185, "grad_norm": 0.14335517585277557, "global_step": 51280, "epoch": 483, "lr": 2.593367012850567e-07} +{"train_loss": 0.01753959245979786, "grad_norm": 0.0821971520781517, "global_step": 51281, "epoch": 483, "lr": 2.5903532004362684e-07} +{"train_loss": 0.013850442133843899, "grad_norm": 0.11389525979757309, "global_step": 51282, "epoch": 483, "lr": 2.587341135703192e-07} +{"train_loss": 0.0128925871104002, "grad_norm": 0.08301842212677002, "global_step": 51283, "epoch": 483, "lr": 2.5843308186619397e-07} +{"train_loss": 0.015156494453549385, "grad_norm": 0.07978895306587219, "global_step": 51284, "epoch": 483, "lr": 2.581322249323004e-07} +{"train_loss": 0.015108313411474228, "grad_norm": 0.06890033185482025, "global_step": 51285, "epoch": 483, "lr": 2.5783154276969313e-07} +{"train_loss": 0.016017014160752296, "grad_norm": 0.07239018380641937, "global_step": 51286, "epoch": 483, "lr": 2.575310353794325e-07} +{"train_loss": 0.01605827361345291, "grad_norm": 0.06652053445577621, "global_step": 51287, "epoch": 483, "lr": 2.5723070276257867e-07} +{"train_loss": 0.016255270689725876, "grad_norm": 0.06803004443645477, "global_step": 51288, "epoch": 483, "lr": 2.5693054492018644e-07} +{"train_loss": 0.01471615768969059, "grad_norm": 0.08732521533966064, "global_step": 51289, "epoch": 483, "lr": 2.5663056185329936e-07} +{"train_loss": 0.015424302779138088, "grad_norm": 0.13202287256717682, "global_step": 51290, "epoch": 483, "lr": 2.5633075356298886e-07} +{"train_loss": 0.015050534158945084, "grad_norm": 0.06362367421388626, "global_step": 51291, "epoch": 483, "lr": 2.560311200502985e-07} +{"train_loss": 0.019903631880879402, "grad_norm": 0.11495102941989899, "global_step": 51292, "epoch": 483, "lr": 2.5573166131627746e-07} +{"train_loss": 0.013136381283402443, "grad_norm": 0.06258394569158554, "global_step": 51293, "epoch": 483, "lr": 2.5543237736198043e-07} +{"train_loss": 0.01529247872531414, "grad_norm": 0.06155605986714363, "global_step": 51294, "epoch": 483, "lr": 2.5513326818846216e-07} +{"train_loss": 0.013534150086343288, "grad_norm": 0.15013328194618225, "global_step": 51295, "epoch": 483, "lr": 2.548343337967773e-07} +{"train_loss": 0.013764958828687668, "grad_norm": 0.10585026443004608, "global_step": 51296, "epoch": 483, "lr": 2.54535574187964e-07} +{"train_loss": 0.01427849568426609, "grad_norm": 0.07696205377578735, "global_step": 51297, "epoch": 483, "lr": 2.5423698936308805e-07} +{"train_loss": 0.013319788500666618, "grad_norm": 0.08607720583677292, "global_step": 51298, "epoch": 483, "lr": 2.5393857932318186e-07} +{"train_loss": 0.01198870874941349, "grad_norm": 0.08004403859376907, "global_step": 51299, "epoch": 483, "lr": 2.536403440693003e-07} +{"train_loss": 0.013030183501541615, "grad_norm": 0.07890773564577103, "global_step": 51300, "epoch": 483, "lr": 2.533422836024979e-07} +{"train_loss": 0.012982012704014778, "grad_norm": 0.09566820412874222, "global_step": 51301, "epoch": 483, "lr": 2.5304439792381284e-07} +{"train_loss": 0.013055147603154182, "grad_norm": 0.07824545353651047, "global_step": 51302, "epoch": 483, "lr": 2.527466870342943e-07} +{"train_loss": 0.014954148000984822, "grad_norm": 0.10121949762105942, "global_step": 51303, "epoch": 483, "lr": 2.5244915093499134e-07, "val_loss": 0.01973295956850052} +{"train_loss": 0.014835966750979424, "grad_norm": 0.08803072571754456, "global_step": 51304, "epoch": 484, "lr": 2.5215178962694766e-07} +{"train_loss": 0.01341748796403408, "grad_norm": 0.14784346520900726, "global_step": 51305, "epoch": 484, "lr": 2.518546031112068e-07} +{"train_loss": 0.01300220936536789, "grad_norm": 0.11490939557552338, "global_step": 51306, "epoch": 484, "lr": 2.5155759138881793e-07} +{"train_loss": 0.014868107624351978, "grad_norm": 0.06734228879213333, "global_step": 51307, "epoch": 484, "lr": 2.512607544608192e-07} +{"train_loss": 0.016290027648210526, "grad_norm": 0.056922368705272675, "global_step": 51308, "epoch": 484, "lr": 2.509640923282541e-07} +{"train_loss": 0.014162247069180012, "grad_norm": 0.10617455095052719, "global_step": 51309, "epoch": 484, "lr": 2.506676049921608e-07} +{"train_loss": 0.014722602441906929, "grad_norm": 0.11033080518245697, "global_step": 51310, "epoch": 484, "lr": 2.5037129245359394e-07} +{"train_loss": 0.015169174410402775, "grad_norm": 0.07599396258592606, "global_step": 51311, "epoch": 484, "lr": 2.5007515471358044e-07} +{"train_loss": 0.01690506935119629, "grad_norm": 0.08649833500385284, "global_step": 51312, "epoch": 484, "lr": 2.4977919177316954e-07} +{"train_loss": 0.01509887259453535, "grad_norm": 0.06298603862524033, "global_step": 51313, "epoch": 484, "lr": 2.4948340363339926e-07} +{"train_loss": 0.015559516847133636, "grad_norm": 0.09132244437932968, "global_step": 51314, "epoch": 484, "lr": 2.4918779029531327e-07} +{"train_loss": 0.014492501504719257, "grad_norm": 0.1040220558643341, "global_step": 51315, "epoch": 484, "lr": 2.4889235175993843e-07} +{"train_loss": 0.015170557424426079, "grad_norm": 0.1411772519350052, "global_step": 51316, "epoch": 484, "lr": 2.4859708802832395e-07} +{"train_loss": 0.012789549306035042, "grad_norm": 0.099220409989357, "global_step": 51317, "epoch": 484, "lr": 2.4830199910149675e-07} +{"train_loss": 0.013456366956233978, "grad_norm": 0.07666202634572983, "global_step": 51318, "epoch": 484, "lr": 2.480070849805061e-07} +{"train_loss": 0.01463065855205059, "grad_norm": 0.08114666491746902, "global_step": 51319, "epoch": 484, "lr": 2.477123456663788e-07} +{"train_loss": 0.018574975430965424, "grad_norm": 0.13711883127689362, "global_step": 51320, "epoch": 484, "lr": 2.4741778116015855e-07} +{"train_loss": 0.015372822061181068, "grad_norm": 0.08919104933738708, "global_step": 51321, "epoch": 484, "lr": 2.4712339146287235e-07} +{"train_loss": 0.013825858943164349, "grad_norm": 0.07305117696523666, "global_step": 51322, "epoch": 484, "lr": 2.468291765755526e-07} +{"train_loss": 0.016193291172385216, "grad_norm": 0.07065718621015549, "global_step": 51323, "epoch": 484, "lr": 2.46535136499243e-07} +{"train_loss": 0.015326553955674171, "grad_norm": 0.10252171009778976, "global_step": 51324, "epoch": 484, "lr": 2.4624127123497045e-07} +{"train_loss": 0.017215076833963394, "grad_norm": 0.09233511984348297, "global_step": 51325, "epoch": 484, "lr": 2.4594758078376743e-07} +{"train_loss": 0.01558731496334076, "grad_norm": 0.08552902936935425, "global_step": 51326, "epoch": 484, "lr": 2.4565406514667213e-07} +{"train_loss": 0.01670287363231182, "grad_norm": 0.1692219227552414, "global_step": 51327, "epoch": 484, "lr": 2.453607243247058e-07} +{"train_loss": 0.014590470120310783, "grad_norm": 0.1265372335910797, "global_step": 51328, "epoch": 484, "lr": 2.450675583189066e-07} +{"train_loss": 0.014887657016515732, "grad_norm": 0.05892070755362511, "global_step": 51329, "epoch": 484, "lr": 2.4477456713030144e-07} +{"train_loss": 0.014857813715934753, "grad_norm": 0.06595002859830856, "global_step": 51330, "epoch": 484, "lr": 2.4448175075992283e-07} +{"train_loss": 0.014833003282546997, "grad_norm": 0.10129953920841217, "global_step": 51331, "epoch": 484, "lr": 2.441891092087922e-07} +{"train_loss": 0.014390749856829643, "grad_norm": 0.09612525254487991, "global_step": 51332, "epoch": 484, "lr": 2.4389664247794766e-07} +{"train_loss": 0.016580387949943542, "grad_norm": 0.11126250773668289, "global_step": 51333, "epoch": 484, "lr": 2.4360435056841045e-07} +{"train_loss": 0.016042230650782585, "grad_norm": 0.10590566694736481, "global_step": 51334, "epoch": 484, "lr": 2.433122334812077e-07} +{"train_loss": 0.015342630445957184, "grad_norm": 0.08556671440601349, "global_step": 51335, "epoch": 484, "lr": 2.430202912173718e-07} +{"train_loss": 0.013966187834739685, "grad_norm": 0.05922134593129158, "global_step": 51336, "epoch": 484, "lr": 2.4272852377791865e-07} +{"train_loss": 0.01337340846657753, "grad_norm": 0.05982285365462303, "global_step": 51337, "epoch": 484, "lr": 2.424369311638808e-07} +{"train_loss": 0.012993094511330128, "grad_norm": 0.0733305811882019, "global_step": 51338, "epoch": 484, "lr": 2.421455133762796e-07} +{"train_loss": 0.015178592875599861, "grad_norm": 0.08084806799888611, "global_step": 51339, "epoch": 484, "lr": 2.4185427041613637e-07} +{"train_loss": 0.015368646942079067, "grad_norm": 0.07107233256101608, "global_step": 51340, "epoch": 484, "lr": 2.415632022844838e-07} +{"train_loss": 0.01441442035138607, "grad_norm": 0.07976769655942917, "global_step": 51341, "epoch": 484, "lr": 2.4127230898233767e-07} +{"train_loss": 0.013549145311117172, "grad_norm": 0.06843557953834534, "global_step": 51342, "epoch": 484, "lr": 2.4098159051071934e-07} +{"train_loss": 0.014812425710260868, "grad_norm": 0.07913463562726974, "global_step": 51343, "epoch": 484, "lr": 2.406910468706558e-07} +{"train_loss": 0.013780660927295685, "grad_norm": 0.09011273086071014, "global_step": 51344, "epoch": 484, "lr": 2.4040067806315735e-07} +{"train_loss": 0.015100961551070213, "grad_norm": 0.13218198716640472, "global_step": 51345, "epoch": 484, "lr": 2.401104840892565e-07} +{"train_loss": 0.01375825610011816, "grad_norm": 0.06386080384254456, "global_step": 51346, "epoch": 484, "lr": 2.3982046494996357e-07} +{"train_loss": 0.013428490608930588, "grad_norm": 0.11843150109052658, "global_step": 51347, "epoch": 484, "lr": 2.395306206463055e-07} +{"train_loss": 0.014562021940946579, "grad_norm": 0.11193668842315674, "global_step": 51348, "epoch": 484, "lr": 2.392409511792981e-07} +{"train_loss": 0.013055624440312386, "grad_norm": 0.08683216571807861, "global_step": 51349, "epoch": 484, "lr": 2.3895145654995175e-07} +{"train_loss": 0.014082768000662327, "grad_norm": 0.08444743603467941, "global_step": 51350, "epoch": 484, "lr": 2.3866213675929337e-07} +{"train_loss": 0.013170476071536541, "grad_norm": 0.06702151894569397, "global_step": 51351, "epoch": 484, "lr": 2.383729918083333e-07} +{"train_loss": 0.01493364479392767, "grad_norm": 0.09551455080509186, "global_step": 51352, "epoch": 484, "lr": 2.3808402169809284e-07} +{"train_loss": 0.013956023380160332, "grad_norm": 0.05568409338593483, "global_step": 51353, "epoch": 484, "lr": 2.377952264295824e-07} +{"train_loss": 0.014707032591104507, "grad_norm": 0.06546671688556671, "global_step": 51354, "epoch": 484, "lr": 2.375066060038178e-07} +{"train_loss": 0.018464013934135437, "grad_norm": 0.11699537932872772, "global_step": 51355, "epoch": 484, "lr": 2.372181604218149e-07} +{"train_loss": 0.017191054299473763, "grad_norm": 0.10151077806949615, "global_step": 51356, "epoch": 484, "lr": 2.3692988968458395e-07} +{"train_loss": 0.014769715256989002, "grad_norm": 0.09881072491407394, "global_step": 51357, "epoch": 484, "lr": 2.366417937931409e-07} +{"train_loss": 0.017693834379315376, "grad_norm": 0.08926765620708466, "global_step": 51358, "epoch": 484, "lr": 2.3635387274849598e-07} +{"train_loss": 0.0169672928750515, "grad_norm": 0.06591235101222992, "global_step": 51359, "epoch": 484, "lr": 2.3606612655166504e-07} +{"train_loss": 0.01348608173429966, "grad_norm": 0.08711135387420654, "global_step": 51360, "epoch": 484, "lr": 2.3577855520365288e-07} +{"train_loss": 0.014838997274637222, "grad_norm": 0.0784672200679779, "global_step": 51361, "epoch": 484, "lr": 2.354911587054809e-07} +{"train_loss": 0.014983761124312878, "grad_norm": 0.08120809495449066, "global_step": 51362, "epoch": 484, "lr": 2.352039370581427e-07} +{"train_loss": 0.013140778057277203, "grad_norm": 0.12204965204000473, "global_step": 51363, "epoch": 484, "lr": 2.3491689026265974e-07} +{"train_loss": 0.014225628226995468, "grad_norm": 0.09059400856494904, "global_step": 51364, "epoch": 484, "lr": 2.3463001832003117e-07} +{"train_loss": 0.014192135073244572, "grad_norm": 0.1263025999069214, "global_step": 51365, "epoch": 484, "lr": 2.343433212312729e-07} +{"train_loss": 0.014533594250679016, "grad_norm": 0.07824628800153732, "global_step": 51366, "epoch": 484, "lr": 2.3405679899738963e-07} +{"train_loss": 0.015249132178723812, "grad_norm": 0.08578921109437943, "global_step": 51367, "epoch": 484, "lr": 2.337704516193917e-07} +{"train_loss": 0.015149617567658424, "grad_norm": 0.12792059779167175, "global_step": 51368, "epoch": 484, "lr": 2.334842790982783e-07} +{"train_loss": 0.017136225476861, "grad_norm": 0.12200076133012772, "global_step": 51369, "epoch": 484, "lr": 2.3319828143505974e-07} +{"train_loss": 0.015763137489557266, "grad_norm": 0.09965143352746964, "global_step": 51370, "epoch": 484, "lr": 2.3291245863074074e-07} +{"train_loss": 0.017061837017536163, "grad_norm": 0.10735266655683517, "global_step": 51371, "epoch": 484, "lr": 2.3262681068632052e-07} +{"train_loss": 0.015477406792342663, "grad_norm": 0.10266745835542679, "global_step": 51372, "epoch": 484, "lr": 2.323413376028094e-07} +{"train_loss": 0.014717376790940762, "grad_norm": 0.09264840930700302, "global_step": 51373, "epoch": 484, "lr": 2.3205603938120658e-07} +{"train_loss": 0.016029085963964462, "grad_norm": 0.08573270589113235, "global_step": 51374, "epoch": 484, "lr": 2.3177091602251676e-07} +{"train_loss": 0.013531981967389584, "grad_norm": 0.10337650030851364, "global_step": 51375, "epoch": 484, "lr": 2.3148596752773366e-07} +{"train_loss": 0.013350062072277069, "grad_norm": 0.06686541438102722, "global_step": 51376, "epoch": 484, "lr": 2.312011938978731e-07} +{"train_loss": 0.015477288514375687, "grad_norm": 0.12328971922397614, "global_step": 51377, "epoch": 484, "lr": 2.3091659513392315e-07} +{"train_loss": 0.01331299263983965, "grad_norm": 0.06217917799949646, "global_step": 51378, "epoch": 484, "lr": 2.3063217123688863e-07} +{"train_loss": 0.01352410577237606, "grad_norm": 0.09033571928739548, "global_step": 51379, "epoch": 484, "lr": 2.3034792220777424e-07} +{"train_loss": 0.015718407928943634, "grad_norm": 0.07501684129238129, "global_step": 51380, "epoch": 484, "lr": 2.300638480475681e-07} +{"train_loss": 0.015589186921715736, "grad_norm": 0.11201353371143341, "global_step": 51381, "epoch": 484, "lr": 2.2977994875726938e-07} +{"train_loss": 0.015068983659148216, "grad_norm": 0.10740963369607925, "global_step": 51382, "epoch": 484, "lr": 2.2949622433788287e-07} +{"train_loss": 0.014022806659340858, "grad_norm": 0.07783299684524536, "global_step": 51383, "epoch": 484, "lr": 2.2921267479040221e-07} +{"train_loss": 0.015463195741176605, "grad_norm": 0.06480968743562698, "global_step": 51384, "epoch": 484, "lr": 2.289293001158266e-07} +{"train_loss": 0.015502363443374634, "grad_norm": 0.06905880570411682, "global_step": 51385, "epoch": 484, "lr": 2.286461003151441e-07} +{"train_loss": 0.018796607851982117, "grad_norm": 0.10049755126237869, "global_step": 51386, "epoch": 484, "lr": 2.2836307538935952e-07} +{"train_loss": 0.015502449125051498, "grad_norm": 0.11540047824382782, "global_step": 51387, "epoch": 484, "lr": 2.280802253394554e-07} +{"train_loss": 0.015040744096040726, "grad_norm": 0.07668005675077438, "global_step": 51388, "epoch": 484, "lr": 2.2779755016643644e-07} +{"train_loss": 0.016182344406843185, "grad_norm": 0.07310431450605392, "global_step": 51389, "epoch": 484, "lr": 2.275150498712908e-07} +{"train_loss": 0.013704853132367134, "grad_norm": 0.06746571511030197, "global_step": 51390, "epoch": 484, "lr": 2.2723272445501208e-07} +{"train_loss": 0.015938689932227135, "grad_norm": 0.09021782875061035, "global_step": 51391, "epoch": 484, "lr": 2.2695057391858843e-07} +{"train_loss": 0.013525108806788921, "grad_norm": 0.052082423120737076, "global_step": 51392, "epoch": 484, "lr": 2.26668598263019e-07} +{"train_loss": 0.01690993644297123, "grad_norm": 0.0685468539595604, "global_step": 51393, "epoch": 484, "lr": 2.2638679748928636e-07} +{"train_loss": 0.013695918023586273, "grad_norm": 0.055999673902988434, "global_step": 51394, "epoch": 484, "lr": 2.2610517159838972e-07} +{"train_loss": 0.014938733540475368, "grad_norm": 0.08560355007648468, "global_step": 51395, "epoch": 484, "lr": 2.2582372059130607e-07} +{"train_loss": 0.015684843063354492, "grad_norm": 0.09625406563282013, "global_step": 51396, "epoch": 484, "lr": 2.2554244446904017e-07} +{"train_loss": 0.013836262747645378, "grad_norm": 0.07863730192184448, "global_step": 51397, "epoch": 484, "lr": 2.2526134323256898e-07} +{"train_loss": 0.01472912635654211, "grad_norm": 0.12514762580394745, "global_step": 51398, "epoch": 484, "lr": 2.2498041688288062e-07} +{"train_loss": 0.013324106112122536, "grad_norm": 0.10587143898010254, "global_step": 51399, "epoch": 484, "lr": 2.2469966542096322e-07} +{"train_loss": 0.01569824479520321, "grad_norm": 0.08901140838861465, "global_step": 51400, "epoch": 484, "lr": 2.2441908884781039e-07} +{"train_loss": 0.015403925441205502, "grad_norm": 0.19264350831508636, "global_step": 51401, "epoch": 484, "lr": 2.2413868716439358e-07} +{"train_loss": 0.013858758844435215, "grad_norm": 0.09661120921373367, "global_step": 51402, "epoch": 484, "lr": 2.2385846037171198e-07} +{"train_loss": 0.01616969145834446, "grad_norm": 0.06650980561971664, "global_step": 51403, "epoch": 484, "lr": 2.2357840847074819e-07} +{"train_loss": 0.014277685433626175, "grad_norm": 0.0625448077917099, "global_step": 51404, "epoch": 484, "lr": 2.2329853146247915e-07} +{"train_loss": 0.01455905381590128, "grad_norm": 0.0761890858411789, "global_step": 51405, "epoch": 484, "lr": 2.2301882934788742e-07} +{"train_loss": 0.01339193806052208, "grad_norm": 0.06507010757923126, "global_step": 51406, "epoch": 484, "lr": 2.2273930212796667e-07} +{"train_loss": 0.014630615711212158, "grad_norm": 0.12908625602722168, "global_step": 51407, "epoch": 484, "lr": 2.2245994980368833e-07} +{"train_loss": 0.014611156657338142, "grad_norm": 0.10453183948993683, "global_step": 51408, "epoch": 484, "lr": 2.221807723760405e-07} +{"train_loss": 0.014949638528292472, "grad_norm": 0.16635897755622864, "global_step": 51409, "epoch": 484, "lr": 2.219017698460002e-07, "val_loss": 0.019709687680006027} +{"train_loss": 0.013542991131544113, "grad_norm": 0.07359439134597778, "global_step": 51410, "epoch": 485, "lr": 2.2162294221454993e-07} +{"train_loss": 0.01499096304178238, "grad_norm": 0.07663413137197495, "global_step": 51411, "epoch": 485, "lr": 2.2134428948266673e-07} +{"train_loss": 0.01582694798707962, "grad_norm": 0.08607643842697144, "global_step": 51412, "epoch": 485, "lr": 2.210658116513331e-07} +{"train_loss": 0.014237786643207073, "grad_norm": 0.08666759729385376, "global_step": 51413, "epoch": 485, "lr": 2.2078750872152608e-07} +{"train_loss": 0.01867802068591118, "grad_norm": 0.06565339118242264, "global_step": 51414, "epoch": 485, "lr": 2.2050938069422268e-07} +{"train_loss": 0.013681380078196526, "grad_norm": 0.07588434964418411, "global_step": 51415, "epoch": 485, "lr": 2.2023142757039982e-07} +{"train_loss": 0.01306372694671154, "grad_norm": 0.12009749561548233, "global_step": 51416, "epoch": 485, "lr": 2.1995364935104014e-07} +{"train_loss": 0.014647439122200012, "grad_norm": 0.14870408177375793, "global_step": 51417, "epoch": 485, "lr": 2.1967604603710946e-07} +{"train_loss": 0.015950385481119156, "grad_norm": 0.07422693073749542, "global_step": 51418, "epoch": 485, "lr": 2.1939861762959036e-07} +{"train_loss": 0.015332539565861225, "grad_norm": 0.10439915210008621, "global_step": 51419, "epoch": 485, "lr": 2.1912136412945429e-07} +{"train_loss": 0.015027139335870743, "grad_norm": 0.050680018961429596, "global_step": 51420, "epoch": 485, "lr": 2.1884428553767822e-07} +{"train_loss": 0.01570052094757557, "grad_norm": 0.06568238884210587, "global_step": 51421, "epoch": 485, "lr": 2.185673818552336e-07} +{"train_loss": 0.014454742893576622, "grad_norm": 0.05378180742263794, "global_step": 51422, "epoch": 485, "lr": 2.1829065308309194e-07} +{"train_loss": 0.01723070628941059, "grad_norm": 0.10882627218961716, "global_step": 51423, "epoch": 485, "lr": 2.1801409922223016e-07} +{"train_loss": 0.013251337222754955, "grad_norm": 0.06677216291427612, "global_step": 51424, "epoch": 485, "lr": 2.1773772027361417e-07} +{"train_loss": 0.014044856652617455, "grad_norm": 0.09160574525594711, "global_step": 51425, "epoch": 485, "lr": 2.1746151623821543e-07} +{"train_loss": 0.015362050384283066, "grad_norm": 0.11859583109617233, "global_step": 51426, "epoch": 485, "lr": 2.1718548711701648e-07} +{"train_loss": 0.014612484723329544, "grad_norm": 0.0924677848815918, "global_step": 51427, "epoch": 485, "lr": 2.169096329109721e-07} +{"train_loss": 0.014397552236914635, "grad_norm": 0.07692159712314606, "global_step": 51428, "epoch": 485, "lr": 2.1663395362105932e-07} +{"train_loss": 0.015648696571588516, "grad_norm": 0.20726856589317322, "global_step": 51429, "epoch": 485, "lr": 2.1635844924823844e-07} +{"train_loss": 0.01682823896408081, "grad_norm": 0.08234941959381104, "global_step": 51430, "epoch": 485, "lr": 2.1608311979349205e-07} +{"train_loss": 0.013423952274024487, "grad_norm": 0.07118044048547745, "global_step": 51431, "epoch": 485, "lr": 2.1580796525777492e-07} +{"train_loss": 0.016467193141579628, "grad_norm": 0.06783799082040787, "global_step": 51432, "epoch": 485, "lr": 2.155329856420585e-07} +{"train_loss": 0.014991668984293938, "grad_norm": 0.08845826983451843, "global_step": 51433, "epoch": 485, "lr": 2.152581809473031e-07} +{"train_loss": 0.012231030501425266, "grad_norm": 0.09095045179128647, "global_step": 51434, "epoch": 485, "lr": 2.1498355117449133e-07} +{"train_loss": 0.013777704909443855, "grad_norm": 0.0585734061896801, "global_step": 51435, "epoch": 485, "lr": 2.1470909632456682e-07} +{"train_loss": 0.013805401511490345, "grad_norm": 0.06694355607032776, "global_step": 51436, "epoch": 485, "lr": 2.144348163985066e-07} +{"train_loss": 0.014601564966142178, "grad_norm": 0.1012270376086235, "global_step": 51437, "epoch": 485, "lr": 2.1416071139727102e-07} +{"train_loss": 0.015802236273884773, "grad_norm": 0.10865359008312225, "global_step": 51438, "epoch": 485, "lr": 2.1388678132182595e-07} +{"train_loss": 0.017520278692245483, "grad_norm": 0.09565014392137527, "global_step": 51439, "epoch": 485, "lr": 2.1361302617312619e-07} +{"train_loss": 0.012607553042471409, "grad_norm": 0.0967821255326271, "global_step": 51440, "epoch": 485, "lr": 2.1333944595214317e-07} +{"train_loss": 0.014813292771577835, "grad_norm": 0.08410784602165222, "global_step": 51441, "epoch": 485, "lr": 2.130660406598317e-07} +{"train_loss": 0.014463184401392937, "grad_norm": 0.11890837550163269, "global_step": 51442, "epoch": 485, "lr": 2.1279281029715215e-07} +{"train_loss": 0.015064572915434837, "grad_norm": 0.06503592431545258, "global_step": 51443, "epoch": 485, "lr": 2.1251975486507036e-07} +{"train_loss": 0.016099225729703903, "grad_norm": 0.06568913161754608, "global_step": 51444, "epoch": 485, "lr": 2.122468743645467e-07} +{"train_loss": 0.015296433120965958, "grad_norm": 0.058599457144737244, "global_step": 51445, "epoch": 485, "lr": 2.1197416879652487e-07} +{"train_loss": 0.015202781185507774, "grad_norm": 0.0618567168712616, "global_step": 51446, "epoch": 485, "lr": 2.1170163816198185e-07} +{"train_loss": 0.016079716384410858, "grad_norm": 0.05686383321881294, "global_step": 51447, "epoch": 485, "lr": 2.1142928246186133e-07} +{"train_loss": 0.01644875481724739, "grad_norm": 0.1527772843837738, "global_step": 51448, "epoch": 485, "lr": 2.1115710169712922e-07} +{"train_loss": 0.015208648517727852, "grad_norm": 0.10504182428121567, "global_step": 51449, "epoch": 485, "lr": 2.1088509586874028e-07} +{"train_loss": 0.014471798203885555, "grad_norm": 0.08093994110822678, "global_step": 51450, "epoch": 485, "lr": 2.1061326497764377e-07} +{"train_loss": 0.014325067400932312, "grad_norm": 0.08280019462108612, "global_step": 51451, "epoch": 485, "lr": 2.1034160902480004e-07} +{"train_loss": 0.015221293084323406, "grad_norm": 0.08345772325992584, "global_step": 51452, "epoch": 485, "lr": 2.1007012801116388e-07} +{"train_loss": 0.014075808227062225, "grad_norm": 0.0635603815317154, "global_step": 51453, "epoch": 485, "lr": 2.0979882193769006e-07} +{"train_loss": 0.015705537050962448, "grad_norm": 0.09668567776679993, "global_step": 51454, "epoch": 485, "lr": 2.0952769080532786e-07} +{"train_loss": 0.016675271093845367, "grad_norm": 0.13481999933719635, "global_step": 51455, "epoch": 485, "lr": 2.0925673461503205e-07} +{"train_loss": 0.01622142642736435, "grad_norm": 0.06857357174158096, "global_step": 51456, "epoch": 485, "lr": 2.0898595336775738e-07} +{"train_loss": 0.013250605203211308, "grad_norm": 0.07125959545373917, "global_step": 51457, "epoch": 485, "lr": 2.0871534706444763e-07} +{"train_loss": 0.011891789734363556, "grad_norm": 0.06094082444906235, "global_step": 51458, "epoch": 485, "lr": 2.0844491570606305e-07} +{"train_loss": 0.015266073867678642, "grad_norm": 0.07187829911708832, "global_step": 51459, "epoch": 485, "lr": 2.081746592935474e-07} +{"train_loss": 0.01489375252276659, "grad_norm": 0.13743934035301208, "global_step": 51460, "epoch": 485, "lr": 2.0790457782784988e-07} +{"train_loss": 0.012549048289656639, "grad_norm": 0.09182190895080566, "global_step": 51461, "epoch": 485, "lr": 2.0763467130992532e-07} +{"train_loss": 0.015521824359893799, "grad_norm": 0.11826733499765396, "global_step": 51462, "epoch": 485, "lr": 2.0736493974071736e-07} +{"train_loss": 0.01415218599140644, "grad_norm": 0.08911093324422836, "global_step": 51463, "epoch": 485, "lr": 2.0709538312117527e-07} +{"train_loss": 0.01869516633450985, "grad_norm": 0.09662463515996933, "global_step": 51464, "epoch": 485, "lr": 2.0682600145224273e-07} +{"train_loss": 0.01546396967023611, "grad_norm": 0.09724146127700806, "global_step": 51465, "epoch": 485, "lr": 2.0655679473486899e-07} +{"train_loss": 0.013229124248027802, "grad_norm": 0.06722980737686157, "global_step": 51466, "epoch": 485, "lr": 2.062877629700033e-07} +{"train_loss": 0.01574617624282837, "grad_norm": 0.06037168949842453, "global_step": 51467, "epoch": 485, "lr": 2.060189061585893e-07} +{"train_loss": 0.014646491967141628, "grad_norm": 0.08494192361831665, "global_step": 51468, "epoch": 485, "lr": 2.0575022430156521e-07} +{"train_loss": 0.014446690678596497, "grad_norm": 0.08300434798002243, "global_step": 51469, "epoch": 485, "lr": 2.054817173998802e-07} +{"train_loss": 0.013957232236862183, "grad_norm": 0.06362351775169373, "global_step": 51470, "epoch": 485, "lr": 2.0521338545448354e-07} +{"train_loss": 0.01324701402336359, "grad_norm": 0.10451904684305191, "global_step": 51471, "epoch": 485, "lr": 2.049452284663078e-07} +{"train_loss": 0.016625799238681793, "grad_norm": 0.1675974279642105, "global_step": 51472, "epoch": 485, "lr": 2.046772464362967e-07} +{"train_loss": 0.01458272896707058, "grad_norm": 0.11035799980163574, "global_step": 51473, "epoch": 485, "lr": 2.0440943936539947e-07} +{"train_loss": 0.015634674578905106, "grad_norm": 0.06737358123064041, "global_step": 51474, "epoch": 485, "lr": 2.041418072545487e-07} +{"train_loss": 0.014214706607162952, "grad_norm": 0.06846827268600464, "global_step": 51475, "epoch": 485, "lr": 2.0387435010468802e-07} +{"train_loss": 0.014067716896533966, "grad_norm": 0.073818638920784, "global_step": 51476, "epoch": 485, "lr": 2.0360706791675565e-07} +{"train_loss": 0.015078937634825706, "grad_norm": 0.08103403449058533, "global_step": 51477, "epoch": 485, "lr": 2.0333996069169526e-07} +{"train_loss": 0.01572929322719574, "grad_norm": 0.06413036584854126, "global_step": 51478, "epoch": 485, "lr": 2.030730284304394e-07} +{"train_loss": 0.016174234449863434, "grad_norm": 0.06134891137480736, "global_step": 51479, "epoch": 485, "lr": 2.028062711339318e-07} +{"train_loss": 0.015846144407987595, "grad_norm": 0.11216691136360168, "global_step": 51480, "epoch": 485, "lr": 2.0253968880310504e-07} +{"train_loss": 0.013972306624054909, "grad_norm": 0.0779966339468956, "global_step": 51481, "epoch": 485, "lr": 2.0227328143889723e-07} +{"train_loss": 0.016683287918567657, "grad_norm": 0.07312373071908951, "global_step": 51482, "epoch": 485, "lr": 2.0200704904224655e-07} +{"train_loss": 0.015475569292902946, "grad_norm": 0.0609474815428257, "global_step": 51483, "epoch": 485, "lr": 2.0174099161408e-07} +{"train_loss": 0.012813810259103775, "grad_norm": 0.06566406786441803, "global_step": 51484, "epoch": 485, "lr": 2.0147510915534684e-07} +{"train_loss": 0.014315884560346603, "grad_norm": 0.08282443135976791, "global_step": 51485, "epoch": 485, "lr": 2.012094016669741e-07} +{"train_loss": 0.014526715502142906, "grad_norm": 0.0696045458316803, "global_step": 51486, "epoch": 485, "lr": 2.0094386914988884e-07} +{"train_loss": 0.01467012520879507, "grad_norm": 0.058838341385126114, "global_step": 51487, "epoch": 485, "lr": 2.0067851160502914e-07} +{"train_loss": 0.01427812036126852, "grad_norm": 0.057279348373413086, "global_step": 51488, "epoch": 485, "lr": 2.0041332903333322e-07} +{"train_loss": 0.016928110271692276, "grad_norm": 0.0950617641210556, "global_step": 51489, "epoch": 485, "lr": 2.0014832143572248e-07} +{"train_loss": 0.015965793281793594, "grad_norm": 0.06589486449956894, "global_step": 51490, "epoch": 485, "lr": 1.9988348881314068e-07} +{"train_loss": 0.014634042978286743, "grad_norm": 0.07852569222450256, "global_step": 51491, "epoch": 485, "lr": 1.996188311665037e-07} +{"train_loss": 0.015616564080119133, "grad_norm": 0.08797373622655869, "global_step": 51492, "epoch": 485, "lr": 1.9935434849674972e-07} +{"train_loss": 0.015076929703354836, "grad_norm": 0.06357110291719437, "global_step": 51493, "epoch": 485, "lr": 1.990900408048113e-07} +{"train_loss": 0.015469852834939957, "grad_norm": 0.10083423554897308, "global_step": 51494, "epoch": 485, "lr": 1.9882590809160438e-07} +{"train_loss": 0.015965480357408524, "grad_norm": 0.07771077007055283, "global_step": 51495, "epoch": 485, "lr": 1.9856195035807268e-07} +{"train_loss": 0.015697520226240158, "grad_norm": 0.08293228596448898, "global_step": 51496, "epoch": 485, "lr": 1.9829816760513764e-07} +{"train_loss": 0.015307145193219185, "grad_norm": 0.06114377826452255, "global_step": 51497, "epoch": 485, "lr": 1.980345598337152e-07} +{"train_loss": 0.017309635877609253, "grad_norm": 0.09101619571447372, "global_step": 51498, "epoch": 485, "lr": 1.9777112704474908e-07} +{"train_loss": 0.013678645715117455, "grad_norm": 0.06315529346466064, "global_step": 51499, "epoch": 485, "lr": 1.975078692391552e-07} +{"train_loss": 0.013111688196659088, "grad_norm": 0.055295269936323166, "global_step": 51500, "epoch": 485, "lr": 1.9724478641785505e-07} +{"train_loss": 0.01470787450671196, "grad_norm": 0.09762536734342575, "global_step": 51501, "epoch": 485, "lr": 1.9698187858178674e-07} +{"train_loss": 0.01687697321176529, "grad_norm": 0.08672413975000381, "global_step": 51502, "epoch": 485, "lr": 1.9671914573186069e-07} +{"train_loss": 0.013548500835895538, "grad_norm": 0.09453805536031723, "global_step": 51503, "epoch": 485, "lr": 1.964565878690039e-07} +{"train_loss": 0.015083513222634792, "grad_norm": 0.07814424484968185, "global_step": 51504, "epoch": 485, "lr": 1.9619420499413786e-07} +{"train_loss": 0.015260858461260796, "grad_norm": 0.07523752003908157, "global_step": 51505, "epoch": 485, "lr": 1.9593199710818966e-07} +{"train_loss": 0.01174240279942751, "grad_norm": 0.0578736737370491, "global_step": 51506, "epoch": 485, "lr": 1.9566996421208072e-07} +{"train_loss": 0.01409995649009943, "grad_norm": 0.09603995829820633, "global_step": 51507, "epoch": 485, "lr": 1.9540810630672147e-07} +{"train_loss": 0.014755005948245525, "grad_norm": 0.06548912078142166, "global_step": 51508, "epoch": 485, "lr": 1.9514642339304446e-07} +{"train_loss": 0.01410827785730362, "grad_norm": 0.05526871234178543, "global_step": 51509, "epoch": 485, "lr": 1.9488491547196008e-07} +{"train_loss": 0.01371280662715435, "grad_norm": 0.07233844697475433, "global_step": 51510, "epoch": 485, "lr": 1.9462358254438985e-07} +{"train_loss": 0.014160603284835815, "grad_norm": 0.13865603506565094, "global_step": 51511, "epoch": 485, "lr": 1.9436242461124966e-07} +{"train_loss": 0.017490293830633163, "grad_norm": 0.07360115647315979, "global_step": 51512, "epoch": 485, "lr": 1.9410144167346656e-07} +{"train_loss": 0.015487819910049438, "grad_norm": 0.08930745720863342, "global_step": 51513, "epoch": 485, "lr": 1.9384063373194538e-07} +{"train_loss": 0.015236271545290947, "grad_norm": 0.07072894275188446, "global_step": 51514, "epoch": 485, "lr": 1.9358000078761318e-07} +{"train_loss": 0.0149449808962362, "grad_norm": 0.09676814824342728, "global_step": 51515, "epoch": 485, "lr": 1.9331954284137476e-07, "val_loss": 0.02004813961684704} +{"train_loss": 0.015456954017281532, "grad_norm": 0.07338743656873703, "global_step": 51516, "epoch": 486, "lr": 1.930592598941572e-07} +{"train_loss": 0.01380709744989872, "grad_norm": 0.08455949276685715, "global_step": 51517, "epoch": 486, "lr": 1.9279915194686525e-07} +{"train_loss": 0.014977432787418365, "grad_norm": 0.10269585251808167, "global_step": 51518, "epoch": 486, "lr": 1.9253921900041494e-07} +{"train_loss": 0.01253684051334858, "grad_norm": 0.07963120937347412, "global_step": 51519, "epoch": 486, "lr": 1.922794610557277e-07} +{"train_loss": 0.014460711739957333, "grad_norm": 0.08179349452257156, "global_step": 51520, "epoch": 486, "lr": 1.9201987811370835e-07} +{"train_loss": 0.014383325353264809, "grad_norm": 0.13638511300086975, "global_step": 51521, "epoch": 486, "lr": 1.9176047017526179e-07} +{"train_loss": 0.015328270383179188, "grad_norm": 0.0522560179233551, "global_step": 51522, "epoch": 486, "lr": 1.9150123724131498e-07} +{"train_loss": 0.014639534056186676, "grad_norm": 0.108823262155056, "global_step": 51523, "epoch": 486, "lr": 1.9124217931277277e-07} +{"train_loss": 0.014060689136385918, "grad_norm": 0.08074445277452469, "global_step": 51524, "epoch": 486, "lr": 1.9098329639054002e-07} +{"train_loss": 0.012136264704167843, "grad_norm": 0.14210009574890137, "global_step": 51525, "epoch": 486, "lr": 1.9072458847553266e-07} +{"train_loss": 0.014113377779722214, "grad_norm": 0.08543187379837036, "global_step": 51526, "epoch": 486, "lr": 1.9046605556866104e-07} +{"train_loss": 0.015149980783462524, "grad_norm": 0.08223038166761398, "global_step": 51527, "epoch": 486, "lr": 1.9020769767082446e-07} +{"train_loss": 0.016126137226819992, "grad_norm": 0.09096590429544449, "global_step": 51528, "epoch": 486, "lr": 1.8994951478293888e-07} +{"train_loss": 0.01660710573196411, "grad_norm": 0.08740804344415665, "global_step": 51529, "epoch": 486, "lr": 1.8969150690590908e-07} +{"train_loss": 0.016521278768777847, "grad_norm": 0.08676094561815262, "global_step": 51530, "epoch": 486, "lr": 1.8943367404063994e-07} +{"train_loss": 0.014370286837220192, "grad_norm": 0.13604003190994263, "global_step": 51531, "epoch": 486, "lr": 1.8917601618803626e-07} +{"train_loss": 0.014874683693051338, "grad_norm": 0.1464771330356598, "global_step": 51532, "epoch": 486, "lr": 1.8891853334900844e-07} +{"train_loss": 0.016685619950294495, "grad_norm": 0.08531724661588669, "global_step": 51533, "epoch": 486, "lr": 1.886612255244613e-07} +{"train_loss": 0.013723216950893402, "grad_norm": 0.05842391774058342, "global_step": 51534, "epoch": 486, "lr": 1.8840409271529414e-07} +{"train_loss": 0.01570945605635643, "grad_norm": 0.08019720762968063, "global_step": 51535, "epoch": 486, "lr": 1.8814713492241177e-07} +{"train_loss": 0.015861352905631065, "grad_norm": 0.11698674410581589, "global_step": 51536, "epoch": 486, "lr": 1.8789035214671903e-07} +{"train_loss": 0.013937747105956078, "grad_norm": 0.0749332383275032, "global_step": 51537, "epoch": 486, "lr": 1.8763374438912074e-07} +{"train_loss": 0.01625257171690464, "grad_norm": 0.10130294412374496, "global_step": 51538, "epoch": 486, "lr": 1.873773116505051e-07} +{"train_loss": 0.015904460102319717, "grad_norm": 0.09072597324848175, "global_step": 51539, "epoch": 486, "lr": 1.8712105393178802e-07} +{"train_loss": 0.014049756340682507, "grad_norm": 0.10401461273431778, "global_step": 51540, "epoch": 486, "lr": 1.8686497123386881e-07} +{"train_loss": 0.012986714020371437, "grad_norm": 0.06245623901486397, "global_step": 51541, "epoch": 486, "lr": 1.8660906355763563e-07} +{"train_loss": 0.014337560161948204, "grad_norm": 0.11222147941589355, "global_step": 51542, "epoch": 486, "lr": 1.8635333090399886e-07} +{"train_loss": 0.01707291230559349, "grad_norm": 0.06165153160691261, "global_step": 51543, "epoch": 486, "lr": 1.8609777327385224e-07} +{"train_loss": 0.015953589230775833, "grad_norm": 0.09958559274673462, "global_step": 51544, "epoch": 486, "lr": 1.8584239066809506e-07} +{"train_loss": 0.01654619723558426, "grad_norm": 0.06793773174285889, "global_step": 51545, "epoch": 486, "lr": 1.8558718308762101e-07} +{"train_loss": 0.015706345438957214, "grad_norm": 0.12527255713939667, "global_step": 51546, "epoch": 486, "lr": 1.8533215053333498e-07} +{"train_loss": 0.015796208754181862, "grad_norm": 0.07054063677787781, "global_step": 51547, "epoch": 486, "lr": 1.850772930061251e-07} +{"train_loss": 0.016121601685881615, "grad_norm": 0.13955648243427277, "global_step": 51548, "epoch": 486, "lr": 1.8482261050688508e-07} +{"train_loss": 0.013918984681367874, "grad_norm": 0.0688767209649086, "global_step": 51549, "epoch": 486, "lr": 1.8456810303651984e-07} +{"train_loss": 0.011973172426223755, "grad_norm": 0.10627954453229904, "global_step": 51550, "epoch": 486, "lr": 1.8431377059592302e-07} +{"train_loss": 0.012438012287020683, "grad_norm": 0.08883001655340195, "global_step": 51551, "epoch": 486, "lr": 1.840596131859773e-07} +{"train_loss": 0.01487521082162857, "grad_norm": 0.07889365404844284, "global_step": 51552, "epoch": 486, "lr": 1.838056308075764e-07} +{"train_loss": 0.01595645770430565, "grad_norm": 0.0562155656516552, "global_step": 51553, "epoch": 486, "lr": 1.8355182346162514e-07} +{"train_loss": 0.014883408322930336, "grad_norm": 0.06806401908397675, "global_step": 51554, "epoch": 486, "lr": 1.8329819114900615e-07} +{"train_loss": 0.012827299535274506, "grad_norm": 0.060942139476537704, "global_step": 51555, "epoch": 486, "lr": 1.830447338706187e-07} +{"train_loss": 0.017046578228473663, "grad_norm": 0.09440767019987106, "global_step": 51556, "epoch": 486, "lr": 1.827914516273399e-07} +{"train_loss": 0.01348719373345375, "grad_norm": 0.058061178773641586, "global_step": 51557, "epoch": 486, "lr": 1.8253834442006902e-07} +{"train_loss": 0.01686711236834526, "grad_norm": 0.07714403420686722, "global_step": 51558, "epoch": 486, "lr": 1.822854122496942e-07} +{"train_loss": 0.014081957750022411, "grad_norm": 0.06898941099643707, "global_step": 51559, "epoch": 486, "lr": 1.8203265511710365e-07} +{"train_loss": 0.014786632731556892, "grad_norm": 0.08227989822626114, "global_step": 51560, "epoch": 486, "lr": 1.8178007302318001e-07} +{"train_loss": 0.01628764159977436, "grad_norm": 0.1256883293390274, "global_step": 51561, "epoch": 486, "lr": 1.8152766596882253e-07} +{"train_loss": 0.01378411240875721, "grad_norm": 0.06840164959430695, "global_step": 51562, "epoch": 486, "lr": 1.812754339549083e-07} +{"train_loss": 0.01573094166815281, "grad_norm": 0.06302483379840851, "global_step": 51563, "epoch": 486, "lr": 1.8102337698232553e-07} +{"train_loss": 0.016752034425735474, "grad_norm": 0.07662002742290497, "global_step": 51564, "epoch": 486, "lr": 1.8077149505196234e-07} +{"train_loss": 0.01594046875834465, "grad_norm": 0.06151289865374565, "global_step": 51565, "epoch": 486, "lr": 1.8051978816470138e-07} +{"train_loss": 0.016850801184773445, "grad_norm": 0.08245434612035751, "global_step": 51566, "epoch": 486, "lr": 1.8026825632143086e-07} +{"train_loss": 0.011437789537012577, "grad_norm": 0.056274425238370895, "global_step": 51567, "epoch": 486, "lr": 1.800168995230278e-07} +{"train_loss": 0.015302296727895737, "grad_norm": 0.08234687894582748, "global_step": 51568, "epoch": 486, "lr": 1.7976571777038044e-07} +{"train_loss": 0.01567605510354042, "grad_norm": 0.05932562053203583, "global_step": 51569, "epoch": 486, "lr": 1.7951471106437134e-07} +{"train_loss": 0.020839400589466095, "grad_norm": 0.08634860813617706, "global_step": 51570, "epoch": 486, "lr": 1.7926387940587764e-07} +{"train_loss": 0.015249356627464294, "grad_norm": 0.08719364553689957, "global_step": 51571, "epoch": 486, "lr": 1.7901322279578748e-07} +{"train_loss": 0.014253092929720879, "grad_norm": 0.10790306329727173, "global_step": 51572, "epoch": 486, "lr": 1.7876274123497794e-07} +{"train_loss": 0.015870850533246994, "grad_norm": 0.06319043040275574, "global_step": 51573, "epoch": 486, "lr": 1.7851243472433167e-07} +{"train_loss": 0.014321064576506615, "grad_norm": 0.09209669381380081, "global_step": 51574, "epoch": 486, "lr": 1.7826230326472016e-07} +{"train_loss": 0.01628163829445839, "grad_norm": 0.11252616345882416, "global_step": 51575, "epoch": 486, "lr": 1.7801234685703162e-07} +{"train_loss": 0.014668898656964302, "grad_norm": 0.1526464968919754, "global_step": 51576, "epoch": 486, "lr": 1.7776256550214309e-07} +{"train_loss": 0.016189029440283775, "grad_norm": 0.09215541929006577, "global_step": 51577, "epoch": 486, "lr": 1.7751295920092613e-07} +{"train_loss": 0.014622253365814686, "grad_norm": 0.06055956333875656, "global_step": 51578, "epoch": 486, "lr": 1.7726352795426337e-07} +{"train_loss": 0.015077478252351284, "grad_norm": 0.07215926051139832, "global_step": 51579, "epoch": 486, "lr": 1.7701427176302631e-07} +{"train_loss": 0.014962448738515377, "grad_norm": 0.1268826574087143, "global_step": 51580, "epoch": 486, "lr": 1.767651906280976e-07} +{"train_loss": 0.016119282692670822, "grad_norm": 0.07032271474599838, "global_step": 51581, "epoch": 486, "lr": 1.7651628455034318e-07} +{"train_loss": 0.013452740386128426, "grad_norm": 0.10327835381031036, "global_step": 51582, "epoch": 486, "lr": 1.7626755353064573e-07} +{"train_loss": 0.015758540481328964, "grad_norm": 0.06751857697963715, "global_step": 51583, "epoch": 486, "lr": 1.7601899756988228e-07} +{"train_loss": 0.015036788769066334, "grad_norm": 0.12598319351673126, "global_step": 51584, "epoch": 486, "lr": 1.7577061666891325e-07} +{"train_loss": 0.01400616392493248, "grad_norm": 0.10720837861299515, "global_step": 51585, "epoch": 486, "lr": 1.7552241082862132e-07} +{"train_loss": 0.014008162543177605, "grad_norm": 0.061529431492090225, "global_step": 51586, "epoch": 486, "lr": 1.752743800498724e-07} +{"train_loss": 0.014021522365510464, "grad_norm": 0.0751107707619667, "global_step": 51587, "epoch": 486, "lr": 1.7502652433354362e-07} +{"train_loss": 0.014463799074292183, "grad_norm": 0.09275307506322861, "global_step": 51588, "epoch": 486, "lr": 1.7477884368050092e-07} +{"train_loss": 0.014371942728757858, "grad_norm": 0.06823365390300751, "global_step": 51589, "epoch": 486, "lr": 1.7453133809161582e-07} +{"train_loss": 0.015030857175588608, "grad_norm": 0.10798334330320358, "global_step": 51590, "epoch": 486, "lr": 1.7428400756776542e-07} +{"train_loss": 0.014381228014826775, "grad_norm": 0.060531578958034515, "global_step": 51591, "epoch": 486, "lr": 1.7403685210981012e-07} +{"train_loss": 0.015666745603084564, "grad_norm": 0.08211685717105865, "global_step": 51592, "epoch": 486, "lr": 1.7378987171861593e-07} +{"train_loss": 0.015999246388673782, "grad_norm": 0.06803518533706665, "global_step": 51593, "epoch": 486, "lr": 1.735430663950599e-07} +{"train_loss": 0.012262002564966679, "grad_norm": 0.06995350122451782, "global_step": 51594, "epoch": 486, "lr": 1.7329643614000246e-07} +{"train_loss": 0.01523931697010994, "grad_norm": 0.057234641164541245, "global_step": 51595, "epoch": 486, "lr": 1.7304998095430958e-07} +{"train_loss": 0.014132741838693619, "grad_norm": 0.0957152470946312, "global_step": 51596, "epoch": 486, "lr": 1.7280370083885833e-07} +{"train_loss": 0.015145299956202507, "grad_norm": 0.07028542459011078, "global_step": 51597, "epoch": 486, "lr": 1.7255759579449805e-07} +{"train_loss": 0.013107303529977798, "grad_norm": 0.0763661339879036, "global_step": 51598, "epoch": 486, "lr": 1.723116658221058e-07} +{"train_loss": 0.014295186847448349, "grad_norm": 0.08535533398389816, "global_step": 51599, "epoch": 486, "lr": 1.7206591092253642e-07} +{"train_loss": 0.016086548566818237, "grad_norm": 0.07228852063417435, "global_step": 51600, "epoch": 486, "lr": 1.7182033109666152e-07} +{"train_loss": 0.013747459277510643, "grad_norm": 0.05663248896598816, "global_step": 51601, "epoch": 486, "lr": 1.7157492634533588e-07} +{"train_loss": 0.013458394445478916, "grad_norm": 0.09027078002691269, "global_step": 51602, "epoch": 486, "lr": 1.713296966694311e-07} +{"train_loss": 0.01614650897681713, "grad_norm": 0.08858546614646912, "global_step": 51603, "epoch": 486, "lr": 1.7108464206980202e-07} +{"train_loss": 0.01722768321633339, "grad_norm": 0.07470352947711945, "global_step": 51604, "epoch": 486, "lr": 1.7083976254730904e-07} +{"train_loss": 0.014670400880277157, "grad_norm": 0.08775564283132553, "global_step": 51605, "epoch": 486, "lr": 1.7059505810281818e-07} +{"train_loss": 0.014668715186417103, "grad_norm": 0.09249153733253479, "global_step": 51606, "epoch": 486, "lr": 1.7035052873718426e-07} +{"train_loss": 0.014208202250301838, "grad_norm": 0.10510201007127762, "global_step": 51607, "epoch": 486, "lr": 1.7010617445126776e-07} +{"train_loss": 0.014054995030164719, "grad_norm": 0.09834922105073929, "global_step": 51608, "epoch": 486, "lr": 1.6986199524592905e-07} +{"train_loss": 0.014783959835767746, "grad_norm": 0.09455275535583496, "global_step": 51609, "epoch": 486, "lr": 1.696179911220286e-07} +{"train_loss": 0.016568666324019432, "grad_norm": 0.14099763333797455, "global_step": 51610, "epoch": 486, "lr": 1.6937416208041013e-07} +{"train_loss": 0.014452412724494934, "grad_norm": 0.06536532938480377, "global_step": 51611, "epoch": 486, "lr": 1.6913050812195074e-07} +{"train_loss": 0.014015097171068192, "grad_norm": 0.07533932477235794, "global_step": 51612, "epoch": 486, "lr": 1.6888702924748868e-07} +{"train_loss": 0.013199861161410809, "grad_norm": 0.08182117342948914, "global_step": 51613, "epoch": 486, "lr": 1.6864372545788986e-07} +{"train_loss": 0.014673503115773201, "grad_norm": 0.11135528236627579, "global_step": 51614, "epoch": 486, "lr": 1.6840059675400365e-07} +{"train_loss": 0.015079835429787636, "grad_norm": 0.07245659083127975, "global_step": 51615, "epoch": 486, "lr": 1.6815764313669046e-07} +{"train_loss": 0.01566215232014656, "grad_norm": 0.0767105370759964, "global_step": 51616, "epoch": 486, "lr": 1.6791486460679406e-07} +{"train_loss": 0.013462906703352928, "grad_norm": 0.06293948739767075, "global_step": 51617, "epoch": 486, "lr": 1.676722611651804e-07} +{"train_loss": 0.015454741194844246, "grad_norm": 0.06348546594381332, "global_step": 51618, "epoch": 486, "lr": 1.6742983281268776e-07} +{"train_loss": 0.016207700595259666, "grad_norm": 0.07675135135650635, "global_step": 51619, "epoch": 486, "lr": 1.6718757955018205e-07} +{"train_loss": 0.01567596197128296, "grad_norm": 0.09816356748342514, "global_step": 51620, "epoch": 486, "lr": 1.6694550137850152e-07} +{"train_loss": 0.014947157933042859, "grad_norm": 0.14499343931674957, "global_step": 51621, "epoch": 486, "lr": 1.6670359829850657e-07, "val_loss": 0.019142232835292816} +{"train_loss": 0.016196567565202713, "grad_norm": 0.09611639380455017, "global_step": 51622, "epoch": 487, "lr": 1.6646187031104098e-07} +{"train_loss": 0.01551367063075304, "grad_norm": 0.09333544224500656, "global_step": 51623, "epoch": 487, "lr": 1.6622031741695965e-07} +{"train_loss": 0.014581099152565002, "grad_norm": 0.10448100417852402, "global_step": 51624, "epoch": 487, "lr": 1.659789396171063e-07} +{"train_loss": 0.012861419469118118, "grad_norm": 0.07474012672901154, "global_step": 51625, "epoch": 487, "lr": 1.6573773691232475e-07} +{"train_loss": 0.012345874682068825, "grad_norm": 0.05573166906833649, "global_step": 51626, "epoch": 487, "lr": 1.6549670930347538e-07} +{"train_loss": 0.01609528809785843, "grad_norm": 0.11152464896440506, "global_step": 51627, "epoch": 487, "lr": 1.6525585679139642e-07} +{"train_loss": 0.014581983909010887, "grad_norm": 0.1006179228425026, "global_step": 51628, "epoch": 487, "lr": 1.6501517937693722e-07} +{"train_loss": 0.016268856823444366, "grad_norm": 0.09938042610883713, "global_step": 51629, "epoch": 487, "lr": 1.6477467706094152e-07} +{"train_loss": 0.014403719455003738, "grad_norm": 0.12277931720018387, "global_step": 51630, "epoch": 487, "lr": 1.645343498442531e-07} +{"train_loss": 0.01528492197394371, "grad_norm": 0.059010401368141174, "global_step": 51631, "epoch": 487, "lr": 1.6429419772772126e-07} +{"train_loss": 0.014329648576676846, "grad_norm": 0.09611884504556656, "global_step": 51632, "epoch": 487, "lr": 1.6405422071218424e-07} +{"train_loss": 0.01633596234023571, "grad_norm": 0.14624756574630737, "global_step": 51633, "epoch": 487, "lr": 1.638144187984858e-07} +{"train_loss": 0.015028034336864948, "grad_norm": 0.05895393714308739, "global_step": 51634, "epoch": 487, "lr": 1.6357479198746973e-07} +{"train_loss": 0.015172535553574562, "grad_norm": 0.06835377961397171, "global_step": 51635, "epoch": 487, "lr": 1.6333534027998533e-07} +{"train_loss": 0.015384270809590816, "grad_norm": 0.0772034153342247, "global_step": 51636, "epoch": 487, "lr": 1.630960636768597e-07} +{"train_loss": 0.014772122725844383, "grad_norm": 0.06600785255432129, "global_step": 51637, "epoch": 487, "lr": 1.6285696217894774e-07} +{"train_loss": 0.012757711112499237, "grad_norm": 0.09028366208076477, "global_step": 51638, "epoch": 487, "lr": 1.6261803578707657e-07} +{"train_loss": 0.01666395366191864, "grad_norm": 0.11559312790632248, "global_step": 51639, "epoch": 487, "lr": 1.623792845020955e-07} +{"train_loss": 0.013701492920517921, "grad_norm": 0.1324022263288498, "global_step": 51640, "epoch": 487, "lr": 1.6214070832483718e-07} +{"train_loss": 0.01596594974398613, "grad_norm": 0.09402111917734146, "global_step": 51641, "epoch": 487, "lr": 1.619023072561454e-07} +{"train_loss": 0.016116760671138763, "grad_norm": 0.06332200020551682, "global_step": 51642, "epoch": 487, "lr": 1.6166408129685285e-07} +{"train_loss": 0.01492423377931118, "grad_norm": 0.10810840129852295, "global_step": 51643, "epoch": 487, "lr": 1.6142603044780325e-07} +{"train_loss": 0.013865799643099308, "grad_norm": 0.06505396217107773, "global_step": 51644, "epoch": 487, "lr": 1.6118815470982372e-07} +{"train_loss": 0.01673118956387043, "grad_norm": 0.0877494141459465, "global_step": 51645, "epoch": 487, "lr": 1.6095045408375253e-07} +{"train_loss": 0.012606197968125343, "grad_norm": 0.06531887501478195, "global_step": 51646, "epoch": 487, "lr": 1.6071292857042785e-07} +{"train_loss": 0.016820577904582024, "grad_norm": 0.14003844559192657, "global_step": 51647, "epoch": 487, "lr": 1.6047557817068793e-07} +{"train_loss": 0.014568443410098553, "grad_norm": 0.11030502617359161, "global_step": 51648, "epoch": 487, "lr": 1.6023840288535985e-07} +{"train_loss": 0.012988807633519173, "grad_norm": 0.07979337126016617, "global_step": 51649, "epoch": 487, "lr": 1.6000140271528185e-07} +{"train_loss": 0.012763306498527527, "grad_norm": 0.06365711241960526, "global_step": 51650, "epoch": 487, "lr": 1.5976457766128106e-07} +{"train_loss": 0.015743698924779892, "grad_norm": 0.1290503591299057, "global_step": 51651, "epoch": 487, "lr": 1.595279277241901e-07} +{"train_loss": 0.014365687035024166, "grad_norm": 0.11715515702962875, "global_step": 51652, "epoch": 487, "lr": 1.5929145290484725e-07} +{"train_loss": 0.016105037182569504, "grad_norm": 0.07080883532762527, "global_step": 51653, "epoch": 487, "lr": 1.590551532040796e-07} +{"train_loss": 0.015175988897681236, "grad_norm": 0.16700860857963562, "global_step": 51654, "epoch": 487, "lr": 1.5881902862271425e-07} +{"train_loss": 0.013972875662147999, "grad_norm": 0.13633383810520172, "global_step": 51655, "epoch": 487, "lr": 1.5858307916158388e-07} +{"train_loss": 0.016222208738327026, "grad_norm": 0.07178042083978653, "global_step": 51656, "epoch": 487, "lr": 1.583473048215156e-07} +{"train_loss": 0.01750286854803562, "grad_norm": 0.06378181278705597, "global_step": 51657, "epoch": 487, "lr": 1.581117056033421e-07} +{"train_loss": 0.014921438880264759, "grad_norm": 0.14398181438446045, "global_step": 51658, "epoch": 487, "lr": 1.5787628150789047e-07} +{"train_loss": 0.016179043799638748, "grad_norm": 0.10767918825149536, "global_step": 51659, "epoch": 487, "lr": 1.576410325359823e-07} +{"train_loss": 0.012609787285327911, "grad_norm": 0.07532129436731339, "global_step": 51660, "epoch": 487, "lr": 1.5740595868844464e-07} +{"train_loss": 0.015358919277787209, "grad_norm": 0.09672404080629349, "global_step": 51661, "epoch": 487, "lr": 1.5717105996611027e-07} +{"train_loss": 0.017441924661397934, "grad_norm": 0.08620552718639374, "global_step": 51662, "epoch": 487, "lr": 1.5693633636980066e-07} +{"train_loss": 0.015575363300740719, "grad_norm": 0.06761308014392853, "global_step": 51663, "epoch": 487, "lr": 1.5670178790034297e-07} +{"train_loss": 0.01683541014790535, "grad_norm": 0.08268187940120697, "global_step": 51664, "epoch": 487, "lr": 1.5646741455855318e-07} +{"train_loss": 0.013529127463698387, "grad_norm": 0.08541171997785568, "global_step": 51665, "epoch": 487, "lr": 1.56233216345264e-07} +{"train_loss": 0.012599797919392586, "grad_norm": 0.07335184514522552, "global_step": 51666, "epoch": 487, "lr": 1.5599919326129696e-07} +{"train_loss": 0.0147653017193079, "grad_norm": 0.08442618697881699, "global_step": 51667, "epoch": 487, "lr": 1.5576534530746812e-07} +{"train_loss": 0.013860711827874184, "grad_norm": 0.08195891976356506, "global_step": 51668, "epoch": 487, "lr": 1.5553167248460454e-07} +{"train_loss": 0.013630329631268978, "grad_norm": 0.12572531402111053, "global_step": 51669, "epoch": 487, "lr": 1.5529817479352782e-07} +{"train_loss": 0.016337597742676735, "grad_norm": 0.07975368946790695, "global_step": 51670, "epoch": 487, "lr": 1.5506485223505396e-07} +{"train_loss": 0.015175881795585155, "grad_norm": 0.06645781546831131, "global_step": 51671, "epoch": 487, "lr": 1.5483170481000454e-07} +{"train_loss": 0.014863313175737858, "grad_norm": 0.11760693788528442, "global_step": 51672, "epoch": 487, "lr": 1.545987325192011e-07} +{"train_loss": 0.015496638603508472, "grad_norm": 0.11235013604164124, "global_step": 51673, "epoch": 487, "lr": 1.5436593536346522e-07} +{"train_loss": 0.01443999633193016, "grad_norm": 0.06645483523607254, "global_step": 51674, "epoch": 487, "lr": 1.5413331334360182e-07} +{"train_loss": 0.013834322802722454, "grad_norm": 0.1180073618888855, "global_step": 51675, "epoch": 487, "lr": 1.53900866460438e-07} +{"train_loss": 0.015442395582795143, "grad_norm": 0.08736760914325714, "global_step": 51676, "epoch": 487, "lr": 1.5366859471479534e-07} +{"train_loss": 0.01561159547418356, "grad_norm": 0.0645681619644165, "global_step": 51677, "epoch": 487, "lr": 1.5343649810747874e-07} +{"train_loss": 0.012623844668269157, "grad_norm": 0.05340813845396042, "global_step": 51678, "epoch": 487, "lr": 1.532045766393042e-07} +{"train_loss": 0.015416251495480537, "grad_norm": 0.0781189575791359, "global_step": 51679, "epoch": 487, "lr": 1.529728303110989e-07} +{"train_loss": 0.015204602852463722, "grad_norm": 0.057395581156015396, "global_step": 51680, "epoch": 487, "lr": 1.5274125912366765e-07} +{"train_loss": 0.013455637730658054, "grad_norm": 0.09863783419132233, "global_step": 51681, "epoch": 487, "lr": 1.52509863077821e-07} +{"train_loss": 0.014157801866531372, "grad_norm": 0.058723993599414825, "global_step": 51682, "epoch": 487, "lr": 1.5227864217438048e-07} +{"train_loss": 0.013075729832053185, "grad_norm": 0.052529484033584595, "global_step": 51683, "epoch": 487, "lr": 1.5204759641415655e-07} +{"train_loss": 0.014759154058992863, "grad_norm": 0.05938877537846565, "global_step": 51684, "epoch": 487, "lr": 1.5181672579795413e-07} +{"train_loss": 0.014469998888671398, "grad_norm": 0.07357670366764069, "global_step": 51685, "epoch": 487, "lr": 1.515860303265948e-07} +{"train_loss": 0.014032987877726555, "grad_norm": 0.06596585363149643, "global_step": 51686, "epoch": 487, "lr": 1.513555100008779e-07} +{"train_loss": 0.017056241631507874, "grad_norm": 0.16425864398479462, "global_step": 51687, "epoch": 487, "lr": 1.5112516482161943e-07} +{"train_loss": 0.014698977582156658, "grad_norm": 0.11578020453453064, "global_step": 51688, "epoch": 487, "lr": 1.5089499478962988e-07} +{"train_loss": 0.016489926725625992, "grad_norm": 0.062538281083107, "global_step": 51689, "epoch": 487, "lr": 1.506649999057197e-07} +{"train_loss": 0.015856193378567696, "grad_norm": 0.07319381088018417, "global_step": 51690, "epoch": 487, "lr": 1.504351801706938e-07} +{"train_loss": 0.015426350757479668, "grad_norm": 0.10921695083379745, "global_step": 51691, "epoch": 487, "lr": 1.502055355853571e-07} +{"train_loss": 0.015141032636165619, "grad_norm": 0.0835423395037651, "global_step": 51692, "epoch": 487, "lr": 1.4997606615052008e-07} +{"train_loss": 0.0147395059466362, "grad_norm": 0.08048835396766663, "global_step": 51693, "epoch": 487, "lr": 1.497467718669876e-07} +{"train_loss": 0.015122871845960617, "grad_norm": 0.07346653938293457, "global_step": 51694, "epoch": 487, "lr": 1.4951765273556461e-07} +{"train_loss": 0.015966663137078285, "grad_norm": 0.09003759175539017, "global_step": 51695, "epoch": 487, "lr": 1.4928870875705603e-07} +{"train_loss": 0.015165787190198898, "grad_norm": 0.10266362875699997, "global_step": 51696, "epoch": 487, "lr": 1.4905993993227228e-07} +{"train_loss": 0.014453394338488579, "grad_norm": 0.10763918608427048, "global_step": 51697, "epoch": 487, "lr": 1.4883134626200722e-07} +{"train_loss": 0.015923693776130676, "grad_norm": 0.06026844680309296, "global_step": 51698, "epoch": 487, "lr": 1.486029277470713e-07} +{"train_loss": 0.014924999326467514, "grad_norm": 0.08405911922454834, "global_step": 51699, "epoch": 487, "lr": 1.4837468438826386e-07} +{"train_loss": 0.017165804281830788, "grad_norm": 0.127389058470726, "global_step": 51700, "epoch": 487, "lr": 1.481466161863898e-07} +{"train_loss": 0.016268452629446983, "grad_norm": 0.0638144239783287, "global_step": 51701, "epoch": 487, "lr": 1.4791872314224297e-07} +{"train_loss": 0.016747666522860527, "grad_norm": 0.0534980446100235, "global_step": 51702, "epoch": 487, "lr": 1.4769100525663938e-07} +{"train_loss": 0.015424652025103569, "grad_norm": 0.0960642471909523, "global_step": 51703, "epoch": 487, "lr": 1.474634625303617e-07} +{"train_loss": 0.01560607273131609, "grad_norm": 0.09584003686904907, "global_step": 51704, "epoch": 487, "lr": 1.472360949642204e-07} +{"train_loss": 0.013852892443537712, "grad_norm": 0.11211550235748291, "global_step": 51705, "epoch": 487, "lr": 1.4700890255900935e-07} +{"train_loss": 0.014582044444978237, "grad_norm": 0.10781337320804596, "global_step": 51706, "epoch": 487, "lr": 1.4678188531552782e-07} +{"train_loss": 0.014314934611320496, "grad_norm": 0.1320711076259613, "global_step": 51707, "epoch": 487, "lr": 1.4655504323457524e-07} +{"train_loss": 0.014814922586083412, "grad_norm": 0.07872019708156586, "global_step": 51708, "epoch": 487, "lr": 1.4632837631695097e-07} +{"train_loss": 0.014470506459474564, "grad_norm": 0.10488509386777878, "global_step": 51709, "epoch": 487, "lr": 1.4610188456344322e-07} +{"train_loss": 0.01623997837305069, "grad_norm": 0.06789536029100418, "global_step": 51710, "epoch": 487, "lr": 1.458755679748569e-07} +{"train_loss": 0.015265416353940964, "grad_norm": 0.13173668086528778, "global_step": 51711, "epoch": 487, "lr": 1.4564942655198032e-07} +{"train_loss": 0.01325744017958641, "grad_norm": 0.06825682520866394, "global_step": 51712, "epoch": 487, "lr": 1.4542346029560726e-07} +{"train_loss": 0.014999481849372387, "grad_norm": 0.1413615196943283, "global_step": 51713, "epoch": 487, "lr": 1.4519766920654265e-07} +{"train_loss": 0.013960465788841248, "grad_norm": 0.0952010378241539, "global_step": 51714, "epoch": 487, "lr": 1.4497205328556918e-07} +{"train_loss": 0.012887895107269287, "grad_norm": 0.08257712423801422, "global_step": 51715, "epoch": 487, "lr": 1.4474661253348066e-07} +{"train_loss": 0.015320787206292152, "grad_norm": 0.1220134049654007, "global_step": 51716, "epoch": 487, "lr": 1.445213469510709e-07} +{"train_loss": 0.012819572351872921, "grad_norm": 0.07076819241046906, "global_step": 51717, "epoch": 487, "lr": 1.4429625653912815e-07} +{"train_loss": 0.015380012802779675, "grad_norm": 0.07694081962108612, "global_step": 51718, "epoch": 487, "lr": 1.4407134129845178e-07} +{"train_loss": 0.015408236533403397, "grad_norm": 0.09194429218769073, "global_step": 51719, "epoch": 487, "lr": 1.4384660122983007e-07} +{"train_loss": 0.017302371561527252, "grad_norm": 0.08301530033349991, "global_step": 51720, "epoch": 487, "lr": 1.436220363340457e-07} +{"train_loss": 0.014372710138559341, "grad_norm": 0.08716072142124176, "global_step": 51721, "epoch": 487, "lr": 1.4339764661188692e-07} +{"train_loss": 0.013556580990552902, "grad_norm": 0.06806916743516922, "global_step": 51722, "epoch": 487, "lr": 1.4317343206415312e-07} +{"train_loss": 0.012955933809280396, "grad_norm": 0.058317143470048904, "global_step": 51723, "epoch": 487, "lr": 1.4294939269162144e-07} +{"train_loss": 0.01705842837691307, "grad_norm": 0.07071855664253235, "global_step": 51724, "epoch": 487, "lr": 1.427255284950857e-07} +{"train_loss": 0.016069957986474037, "grad_norm": 0.0945190042257309, "global_step": 51725, "epoch": 487, "lr": 1.425018394753286e-07} +{"train_loss": 0.015152632258832455, "grad_norm": 0.12006314843893051, "global_step": 51726, "epoch": 487, "lr": 1.422783256331439e-07} +{"train_loss": 0.014939384332875599, "grad_norm": 0.14341413974761963, "global_step": 51727, "epoch": 487, "lr": 1.420549869693033e-07, "val_loss": 0.019111517816781998} +{"train_loss": 0.016053875908255577, "grad_norm": 0.05147814750671387, "global_step": 51728, "epoch": 488, "lr": 1.4183182348460056e-07} +{"train_loss": 0.01347270980477333, "grad_norm": 0.07453013956546783, "global_step": 51729, "epoch": 488, "lr": 1.416088351798184e-07} +{"train_loss": 0.014650162309408188, "grad_norm": 0.16634340584278107, "global_step": 51730, "epoch": 488, "lr": 1.4138602205573947e-07} +{"train_loss": 0.014076147228479385, "grad_norm": 0.06822211295366287, "global_step": 51731, "epoch": 488, "lr": 1.411633841131521e-07} +{"train_loss": 0.016403991729021072, "grad_norm": 0.09652221202850342, "global_step": 51732, "epoch": 488, "lr": 1.409409213528279e-07} +{"train_loss": 0.015487991273403168, "grad_norm": 0.10308505594730377, "global_step": 51733, "epoch": 488, "lr": 1.4071863377555506e-07} +{"train_loss": 0.015084333717823029, "grad_norm": 0.08464866131544113, "global_step": 51734, "epoch": 488, "lr": 1.4049652138211634e-07} +{"train_loss": 0.013853654265403748, "grad_norm": 0.0733613669872284, "global_step": 51735, "epoch": 488, "lr": 1.4027458417328333e-07} +{"train_loss": 0.016215376555919647, "grad_norm": 0.06058372184634209, "global_step": 51736, "epoch": 488, "lr": 1.4005282214984982e-07} +{"train_loss": 0.01513463445007801, "grad_norm": 0.06537134200334549, "global_step": 51737, "epoch": 488, "lr": 1.398312353125819e-07} +{"train_loss": 0.016229163855314255, "grad_norm": 0.131052628159523, "global_step": 51738, "epoch": 488, "lr": 1.3960982366226783e-07} +{"train_loss": 0.01386620756238699, "grad_norm": 0.09626775979995728, "global_step": 51739, "epoch": 488, "lr": 1.3938858719967363e-07} +{"train_loss": 0.01794213429093361, "grad_norm": 0.07971518486738205, "global_step": 51740, "epoch": 488, "lr": 1.3916752592559313e-07} +{"train_loss": 0.014427559450268745, "grad_norm": 0.05544649809598923, "global_step": 51741, "epoch": 488, "lr": 1.3894663984078681e-07} +{"train_loss": 0.015969350934028625, "grad_norm": 0.06527091562747955, "global_step": 51742, "epoch": 488, "lr": 1.3872592894604297e-07} +{"train_loss": 0.015832271426916122, "grad_norm": 0.09734217822551727, "global_step": 51743, "epoch": 488, "lr": 1.3850539324212763e-07} +{"train_loss": 0.013699210248887539, "grad_norm": 0.08663129061460495, "global_step": 51744, "epoch": 488, "lr": 1.3828503272982353e-07} +{"train_loss": 0.01428647805005312, "grad_norm": 0.12591682374477386, "global_step": 51745, "epoch": 488, "lr": 1.3806484740990223e-07} +{"train_loss": 0.017385166138410568, "grad_norm": 0.13860465586185455, "global_step": 51746, "epoch": 488, "lr": 1.3784483728313537e-07} +{"train_loss": 0.014485720545053482, "grad_norm": 0.06987850368022919, "global_step": 51747, "epoch": 488, "lr": 1.3762500235029452e-07} +{"train_loss": 0.013724394142627716, "grad_norm": 0.07005354017019272, "global_step": 51748, "epoch": 488, "lr": 1.3740534261215688e-07} +{"train_loss": 0.013819726184010506, "grad_norm": 0.061852678656578064, "global_step": 51749, "epoch": 488, "lr": 1.3718585806949403e-07} +{"train_loss": 0.016117004677653313, "grad_norm": 0.124770388007164, "global_step": 51750, "epoch": 488, "lr": 1.3696654872307202e-07} +{"train_loss": 0.01610979065299034, "grad_norm": 0.05782392621040344, "global_step": 51751, "epoch": 488, "lr": 1.36747414573668e-07} +{"train_loss": 0.014051832258701324, "grad_norm": 0.07894779741764069, "global_step": 51752, "epoch": 488, "lr": 1.3652845562204808e-07} +{"train_loss": 0.014750040136277676, "grad_norm": 0.11269431561231613, "global_step": 51753, "epoch": 488, "lr": 1.3630967186897825e-07} +{"train_loss": 0.014832068234682083, "grad_norm": 0.08521568775177002, "global_step": 51754, "epoch": 488, "lr": 1.360910633152357e-07} +{"train_loss": 0.01488715223968029, "grad_norm": 0.07309354841709137, "global_step": 51755, "epoch": 488, "lr": 1.358726299615809e-07} +{"train_loss": 0.015895845368504524, "grad_norm": 0.08325483649969101, "global_step": 51756, "epoch": 488, "lr": 1.356543718087855e-07} +{"train_loss": 0.014111429452896118, "grad_norm": 0.055586982518434525, "global_step": 51757, "epoch": 488, "lr": 1.354362888576155e-07} +{"train_loss": 0.013754738494753838, "grad_norm": 0.10290724784135818, "global_step": 51758, "epoch": 488, "lr": 1.35218381108837e-07} +{"train_loss": 0.014407167211174965, "grad_norm": 0.1280117928981781, "global_step": 51759, "epoch": 488, "lr": 1.3500064856321603e-07} +{"train_loss": 0.015993718057870865, "grad_norm": 0.11630940437316895, "global_step": 51760, "epoch": 488, "lr": 1.3478309122151866e-07} +{"train_loss": 0.013091441243886948, "grad_norm": 0.08041313290596008, "global_step": 51761, "epoch": 488, "lr": 1.345657090845054e-07} +{"train_loss": 0.0179123654961586, "grad_norm": 0.07378873974084854, "global_step": 51762, "epoch": 488, "lr": 1.3434850215294226e-07} +{"train_loss": 0.01574636436998844, "grad_norm": 0.11608967185020447, "global_step": 51763, "epoch": 488, "lr": 1.3413147042758979e-07} +{"train_loss": 0.014663501642644405, "grad_norm": 0.11253447085618973, "global_step": 51764, "epoch": 488, "lr": 1.33914613909214e-07} +{"train_loss": 0.0140323955565691, "grad_norm": 0.14090432226657867, "global_step": 51765, "epoch": 488, "lr": 1.3369793259858098e-07} +{"train_loss": 0.015682969242334366, "grad_norm": 0.09582873433828354, "global_step": 51766, "epoch": 488, "lr": 1.3348142649644012e-07} +{"train_loss": 0.01397764403373003, "grad_norm": 0.07033924758434296, "global_step": 51767, "epoch": 488, "lr": 1.332650956035686e-07} +{"train_loss": 0.013773411512374878, "grad_norm": 0.08505985140800476, "global_step": 51768, "epoch": 488, "lr": 1.330489399207102e-07} +{"train_loss": 0.012899298220872879, "grad_norm": 0.059111110866069794, "global_step": 51769, "epoch": 488, "lr": 1.3283295944863105e-07} +{"train_loss": 0.01297823153436184, "grad_norm": 0.09081383049488068, "global_step": 51770, "epoch": 488, "lr": 1.326171541880916e-07} +{"train_loss": 0.013561633415520191, "grad_norm": 0.08034791797399521, "global_step": 51771, "epoch": 488, "lr": 1.324015241398524e-07} +{"train_loss": 0.013947000727057457, "grad_norm": 0.06829135119915009, "global_step": 51772, "epoch": 488, "lr": 1.3218606930466282e-07} +{"train_loss": 0.014849746599793434, "grad_norm": 0.07503359019756317, "global_step": 51773, "epoch": 488, "lr": 1.3197078968328336e-07} +{"train_loss": 0.012569698505103588, "grad_norm": 0.06785925477743149, "global_step": 51774, "epoch": 488, "lr": 1.3175568527647452e-07} +{"train_loss": 0.013733655214309692, "grad_norm": 0.09963209927082062, "global_step": 51775, "epoch": 488, "lr": 1.3154075608499127e-07} +{"train_loss": 0.015602189116179943, "grad_norm": 0.1203840970993042, "global_step": 51776, "epoch": 488, "lr": 1.31326002109583e-07} +{"train_loss": 0.015256160870194435, "grad_norm": 0.08172906190156937, "global_step": 51777, "epoch": 488, "lr": 1.3111142335100467e-07} +{"train_loss": 0.017176050692796707, "grad_norm": 0.11305626481771469, "global_step": 51778, "epoch": 488, "lr": 1.308970198100168e-07} +{"train_loss": 0.015207120217382908, "grad_norm": 0.1378239542245865, "global_step": 51779, "epoch": 488, "lr": 1.3068279148737428e-07} +{"train_loss": 0.017741795629262924, "grad_norm": 0.06785277277231216, "global_step": 51780, "epoch": 488, "lr": 1.3046873838381546e-07} +{"train_loss": 0.01432366669178009, "grad_norm": 0.1284416764974594, "global_step": 51781, "epoch": 488, "lr": 1.3025486050011193e-07} +{"train_loss": 0.01468975841999054, "grad_norm": 0.07982540875673294, "global_step": 51782, "epoch": 488, "lr": 1.3004115783699643e-07} +{"train_loss": 0.014263146556913853, "grad_norm": 0.1615167260169983, "global_step": 51783, "epoch": 488, "lr": 1.298276303952295e-07} +{"train_loss": 0.015657002106308937, "grad_norm": 0.0972118005156517, "global_step": 51784, "epoch": 488, "lr": 1.2961427817556604e-07} +{"train_loss": 0.016176000237464905, "grad_norm": 0.07161105424165726, "global_step": 51785, "epoch": 488, "lr": 1.2940110117874436e-07} +{"train_loss": 0.015604681335389614, "grad_norm": 0.17985601723194122, "global_step": 51786, "epoch": 488, "lr": 1.29188099405525e-07} +{"train_loss": 0.01571228913962841, "grad_norm": 0.10289429128170013, "global_step": 51787, "epoch": 488, "lr": 1.289752728566407e-07} +{"train_loss": 0.014208403415977955, "grad_norm": 0.10231084376573563, "global_step": 51788, "epoch": 488, "lr": 1.2876262153285744e-07} +{"train_loss": 0.015047474764287472, "grad_norm": 0.09063167124986649, "global_step": 51789, "epoch": 488, "lr": 1.285501454349136e-07} +{"train_loss": 0.013186539523303509, "grad_norm": 0.07444608956575394, "global_step": 51790, "epoch": 488, "lr": 1.2833784456355303e-07} +{"train_loss": 0.015145154669880867, "grad_norm": 0.09691145271062851, "global_step": 51791, "epoch": 488, "lr": 1.2812571891952507e-07} +{"train_loss": 0.015911683440208435, "grad_norm": 0.0656195729970932, "global_step": 51792, "epoch": 488, "lr": 1.2791376850357363e-07} +{"train_loss": 0.012660016305744648, "grad_norm": 0.06508680433034897, "global_step": 51793, "epoch": 488, "lr": 1.2770199331644806e-07} +{"train_loss": 0.014182301238179207, "grad_norm": 0.082975834608078, "global_step": 51794, "epoch": 488, "lr": 1.2749039335888114e-07} +{"train_loss": 0.013931100256741047, "grad_norm": 0.0902789756655693, "global_step": 51795, "epoch": 488, "lr": 1.2727896863163334e-07} +{"train_loss": 0.015306984074413776, "grad_norm": 0.07552745938301086, "global_step": 51796, "epoch": 488, "lr": 1.270677191354319e-07} +{"train_loss": 0.01510408241301775, "grad_norm": 0.07971262186765671, "global_step": 51797, "epoch": 488, "lr": 1.2685664487103176e-07} +{"train_loss": 0.014080625958740711, "grad_norm": 0.07820772379636765, "global_step": 51798, "epoch": 488, "lr": 1.266457458391601e-07} +{"train_loss": 0.017013316974043846, "grad_norm": 0.057799458503723145, "global_step": 51799, "epoch": 488, "lr": 1.264350220405719e-07} +{"train_loss": 0.015822486951947212, "grad_norm": 0.08542450517416, "global_step": 51800, "epoch": 488, "lr": 1.2622447347599985e-07} +{"train_loss": 0.01538865827023983, "grad_norm": 0.07039904594421387, "global_step": 51801, "epoch": 488, "lr": 1.2601410014618232e-07} +{"train_loss": 0.014323124662041664, "grad_norm": 0.08407294005155563, "global_step": 51802, "epoch": 488, "lr": 1.2580390205186864e-07} +{"train_loss": 0.014374848455190659, "grad_norm": 0.09731142967939377, "global_step": 51803, "epoch": 488, "lr": 1.2559387919378607e-07} +{"train_loss": 0.014018045738339424, "grad_norm": 0.07078319787979126, "global_step": 51804, "epoch": 488, "lr": 1.2538403157267287e-07} +{"train_loss": 0.014167026616632938, "grad_norm": 0.10199537873268127, "global_step": 51805, "epoch": 488, "lr": 1.2517435918927289e-07} +{"train_loss": 0.015044864267110825, "grad_norm": 0.0638030394911766, "global_step": 51806, "epoch": 488, "lr": 1.2496486204432445e-07} +{"train_loss": 0.015423059463500977, "grad_norm": 0.04983996972441673, "global_step": 51807, "epoch": 488, "lr": 1.2475554013855472e-07} +{"train_loss": 0.015244908630847931, "grad_norm": 0.11979460716247559, "global_step": 51808, "epoch": 488, "lr": 1.2454639347270757e-07} +{"train_loss": 0.015781104564666748, "grad_norm": 0.09760969877243042, "global_step": 51809, "epoch": 488, "lr": 1.243374220475102e-07} +{"train_loss": 0.014695354737341404, "grad_norm": 0.06312067061662674, "global_step": 51810, "epoch": 488, "lr": 1.241286258637009e-07} +{"train_loss": 0.015947818756103516, "grad_norm": 0.12644053995609283, "global_step": 51811, "epoch": 488, "lr": 1.2392000492201794e-07} +{"train_loss": 0.0128993671387434, "grad_norm": 0.07322552800178528, "global_step": 51812, "epoch": 488, "lr": 1.2371155922318855e-07} +{"train_loss": 0.016954710707068443, "grad_norm": 0.08884814381599426, "global_step": 51813, "epoch": 488, "lr": 1.2350328876793993e-07} +{"train_loss": 0.01511000283062458, "grad_norm": 0.06339382380247116, "global_step": 51814, "epoch": 488, "lr": 1.232951935570159e-07} +{"train_loss": 0.015740010887384415, "grad_norm": 0.07047458738088608, "global_step": 51815, "epoch": 488, "lr": 1.230872735911437e-07} +{"train_loss": 0.017513377591967583, "grad_norm": 0.08251627534627914, "global_step": 51816, "epoch": 488, "lr": 1.2287952887104494e-07} +{"train_loss": 0.016059022396802902, "grad_norm": 0.11846961081027985, "global_step": 51817, "epoch": 488, "lr": 1.2267195939745791e-07} +{"train_loss": 0.01579124480485916, "grad_norm": 0.09493168443441391, "global_step": 51818, "epoch": 488, "lr": 1.224645651711154e-07} +{"train_loss": 0.013714488595724106, "grad_norm": 0.07302545756101608, "global_step": 51819, "epoch": 488, "lr": 1.2225734619273898e-07} +{"train_loss": 0.013607112690806389, "grad_norm": 0.05828240141272545, "global_step": 51820, "epoch": 488, "lr": 1.2205030246305594e-07} +{"train_loss": 0.014301665127277374, "grad_norm": 0.08425273001194, "global_step": 51821, "epoch": 488, "lr": 1.2184343398279896e-07} +{"train_loss": 0.016468072310090065, "grad_norm": 0.10337112098932266, "global_step": 51822, "epoch": 488, "lr": 1.2163674075269526e-07} +{"train_loss": 0.013132058084011078, "grad_norm": 0.07044123113155365, "global_step": 51823, "epoch": 488, "lr": 1.214302227734665e-07} +{"train_loss": 0.016007252037525177, "grad_norm": 0.09864483028650284, "global_step": 51824, "epoch": 488, "lr": 1.2122388004583985e-07} +{"train_loss": 0.012553751468658447, "grad_norm": 0.0797274112701416, "global_step": 51825, "epoch": 488, "lr": 1.2101771257054251e-07} +{"train_loss": 0.014535405673086643, "grad_norm": 0.04779436066746712, "global_step": 51826, "epoch": 488, "lr": 1.2081172034829614e-07} +{"train_loss": 0.016417745500802994, "grad_norm": 0.09360170364379883, "global_step": 51827, "epoch": 488, "lr": 1.2060590337982237e-07} +{"train_loss": 0.01582307368516922, "grad_norm": 0.06550220400094986, "global_step": 51828, "epoch": 488, "lr": 1.2040026166584838e-07} +{"train_loss": 0.01450965367257595, "grad_norm": 0.06512155383825302, "global_step": 51829, "epoch": 488, "lr": 1.2019479520710143e-07} +{"train_loss": 0.015530676580965519, "grad_norm": 0.09110009670257568, "global_step": 51830, "epoch": 488, "lr": 1.1998950400429197e-07} +{"train_loss": 0.013985910452902317, "grad_norm": 0.08076721429824829, "global_step": 51831, "epoch": 488, "lr": 1.197843880581473e-07} +{"train_loss": 0.01568765379488468, "grad_norm": 0.10955481231212616, "global_step": 51832, "epoch": 488, "lr": 1.1957944736938897e-07} +{"train_loss": 0.014942063440410595, "grad_norm": 0.07303529232740402, "global_step": 51833, "epoch": 488, "lr": 1.193746819387387e-07, "val_loss": 0.019878946244716644} +{"train_loss": 0.014266032725572586, "grad_norm": 0.08223436772823334, "global_step": 51834, "epoch": 489, "lr": 1.1917009176691251e-07} +{"train_loss": 0.01272985152900219, "grad_norm": 0.05875837802886963, "global_step": 51835, "epoch": 489, "lr": 1.1896567685462657e-07} +{"train_loss": 0.014834806323051453, "grad_norm": 0.06350834667682648, "global_step": 51836, "epoch": 489, "lr": 1.1876143720260802e-07} +{"train_loss": 0.014826610684394836, "grad_norm": 0.06866073608398438, "global_step": 51837, "epoch": 489, "lr": 1.1855737281156742e-07} +{"train_loss": 0.015592613257467747, "grad_norm": 0.0641583725810051, "global_step": 51838, "epoch": 489, "lr": 1.1835348368222088e-07} +{"train_loss": 0.014301721006631851, "grad_norm": 0.07339394092559814, "global_step": 51839, "epoch": 489, "lr": 1.1814976981529002e-07} +{"train_loss": 0.01385343074798584, "grad_norm": 0.07939191907644272, "global_step": 51840, "epoch": 489, "lr": 1.179462312114854e-07} +{"train_loss": 0.014989945106208324, "grad_norm": 0.0666334331035614, "global_step": 51841, "epoch": 489, "lr": 1.1774286787152311e-07} +{"train_loss": 0.013137077912688255, "grad_norm": 0.09341107308864594, "global_step": 51842, "epoch": 489, "lr": 1.1753967979611925e-07} +{"train_loss": 0.01530049741268158, "grad_norm": 0.10809732973575592, "global_step": 51843, "epoch": 489, "lr": 1.1733666698599543e-07} +{"train_loss": 0.014857465401291847, "grad_norm": 0.07176955044269562, "global_step": 51844, "epoch": 489, "lr": 1.1713382944184559e-07} +{"train_loss": 0.01591986045241356, "grad_norm": 0.08662991225719452, "global_step": 51845, "epoch": 489, "lr": 1.1693116716440244e-07} +{"train_loss": 0.0164475254714489, "grad_norm": 0.06569068878889084, "global_step": 51846, "epoch": 489, "lr": 1.1672868015436544e-07} +{"train_loss": 0.014127472415566444, "grad_norm": 0.0630728080868721, "global_step": 51847, "epoch": 489, "lr": 1.1652636841245068e-07} +{"train_loss": 0.012809492647647858, "grad_norm": 0.08789458125829697, "global_step": 51848, "epoch": 489, "lr": 1.163242319393687e-07} +{"train_loss": 0.014773562550544739, "grad_norm": 0.05711574852466583, "global_step": 51849, "epoch": 489, "lr": 1.1612227073583559e-07} +{"train_loss": 0.017174899578094482, "grad_norm": 0.07972241938114166, "global_step": 51850, "epoch": 489, "lr": 1.1592048480254525e-07} +{"train_loss": 0.014574036933481693, "grad_norm": 0.0674375593662262, "global_step": 51851, "epoch": 489, "lr": 1.1571887414022486e-07} +{"train_loss": 0.01431517954915762, "grad_norm": 0.12294575572013855, "global_step": 51852, "epoch": 489, "lr": 1.1551743874956834e-07} +{"train_loss": 0.015167582780122757, "grad_norm": 0.12261425703763962, "global_step": 51853, "epoch": 489, "lr": 1.1531617863129173e-07} +{"train_loss": 0.015328298322856426, "grad_norm": 0.0881422907114029, "global_step": 51854, "epoch": 489, "lr": 1.1511509378610009e-07} +{"train_loss": 0.014567861333489418, "grad_norm": 0.12937389314174652, "global_step": 51855, "epoch": 489, "lr": 1.1491418421469835e-07} +{"train_loss": 0.016260627657175064, "grad_norm": 0.08384916186332703, "global_step": 51856, "epoch": 489, "lr": 1.1471344991779154e-07} +{"train_loss": 0.013701323419809341, "grad_norm": 0.06354765594005585, "global_step": 51857, "epoch": 489, "lr": 1.1451289089609018e-07} +{"train_loss": 0.01544768176972866, "grad_norm": 0.13350938260555267, "global_step": 51858, "epoch": 489, "lr": 1.1431250715029374e-07} +{"train_loss": 0.015671931207180023, "grad_norm": 0.06598417460918427, "global_step": 51859, "epoch": 489, "lr": 1.1411229868111272e-07} +{"train_loss": 0.015261078253388405, "grad_norm": 0.07766149193048477, "global_step": 51860, "epoch": 489, "lr": 1.1391226548924106e-07} +{"train_loss": 0.014506442472338676, "grad_norm": 0.07842308282852173, "global_step": 51861, "epoch": 489, "lr": 1.137124075753948e-07} +{"train_loss": 0.012675691395998001, "grad_norm": 0.08339223265647888, "global_step": 51862, "epoch": 489, "lr": 1.1351272494026233e-07} +{"train_loss": 0.015709679573774338, "grad_norm": 0.07222562283277512, "global_step": 51863, "epoch": 489, "lr": 1.133132175845486e-07} +{"train_loss": 0.012954095378518105, "grad_norm": 0.05262099951505661, "global_step": 51864, "epoch": 489, "lr": 1.1311388550895863e-07} +{"train_loss": 0.014513158239424229, "grad_norm": 0.08395350724458694, "global_step": 51865, "epoch": 489, "lr": 1.1291472871419184e-07} +{"train_loss": 0.015612196177244186, "grad_norm": 0.09310838580131531, "global_step": 51866, "epoch": 489, "lr": 1.1271574720095323e-07} +{"train_loss": 0.013485590927302837, "grad_norm": 0.08123968541622162, "global_step": 51867, "epoch": 489, "lr": 1.1251694096993115e-07} +{"train_loss": 0.015536237508058548, "grad_norm": 0.12262666970491409, "global_step": 51868, "epoch": 489, "lr": 1.1231831002182502e-07} +{"train_loss": 0.014165293425321579, "grad_norm": 0.10934630781412125, "global_step": 51869, "epoch": 489, "lr": 1.1211985435733985e-07} +{"train_loss": 0.015342806465923786, "grad_norm": 0.08810056000947952, "global_step": 51870, "epoch": 489, "lr": 1.1192157397717506e-07} +{"train_loss": 0.014577270485460758, "grad_norm": 0.08414363116025925, "global_step": 51871, "epoch": 489, "lr": 1.1172346888201347e-07} +{"train_loss": 0.016382534056901932, "grad_norm": 0.10401028394699097, "global_step": 51872, "epoch": 489, "lr": 1.1152553907256557e-07} +{"train_loss": 0.011797787621617317, "grad_norm": 0.14916083216667175, "global_step": 51873, "epoch": 489, "lr": 1.1132778454951976e-07} +{"train_loss": 0.014594123698771, "grad_norm": 0.057022128254175186, "global_step": 51874, "epoch": 489, "lr": 1.1113020531357543e-07} +{"train_loss": 0.015609664842486382, "grad_norm": 0.05872002989053726, "global_step": 51875, "epoch": 489, "lr": 1.109328013654154e-07} +{"train_loss": 0.014411656185984612, "grad_norm": 0.10697390884160995, "global_step": 51876, "epoch": 489, "lr": 1.107355727057502e-07} +{"train_loss": 0.01632920280098915, "grad_norm": 0.07033433020114899, "global_step": 51877, "epoch": 489, "lr": 1.1053851933525705e-07} +{"train_loss": 0.015152832493185997, "grad_norm": 0.08274658024311066, "global_step": 51878, "epoch": 489, "lr": 1.1034164125463542e-07} +{"train_loss": 0.016995634883642197, "grad_norm": 0.13721121847629547, "global_step": 51879, "epoch": 489, "lr": 1.1014493846457919e-07} +{"train_loss": 0.014615709893405437, "grad_norm": 0.12100470066070557, "global_step": 51880, "epoch": 489, "lr": 1.0994841096577669e-07} +{"train_loss": 0.015569687820971012, "grad_norm": 0.0702282190322876, "global_step": 51881, "epoch": 489, "lr": 1.097520587589218e-07} +{"train_loss": 0.01417195238173008, "grad_norm": 0.06534913182258606, "global_step": 51882, "epoch": 489, "lr": 1.0955588184469734e-07} +{"train_loss": 0.014095993712544441, "grad_norm": 0.08468220382928848, "global_step": 51883, "epoch": 489, "lr": 1.0935988022379718e-07} +{"train_loss": 0.013952605426311493, "grad_norm": 0.05187716707587242, "global_step": 51884, "epoch": 489, "lr": 1.0916405389690964e-07} +{"train_loss": 0.012622218579053879, "grad_norm": 0.06449870020151138, "global_step": 51885, "epoch": 489, "lr": 1.0896840286472865e-07} +{"train_loss": 0.016486093401908875, "grad_norm": 0.07700639218091965, "global_step": 51886, "epoch": 489, "lr": 1.0877292712792585e-07} +{"train_loss": 0.017423709854483604, "grad_norm": 0.1289626657962799, "global_step": 51887, "epoch": 489, "lr": 1.0857762668720628e-07} +{"train_loss": 0.0128086619079113, "grad_norm": 0.05517858639359474, "global_step": 51888, "epoch": 489, "lr": 1.0838250154324159e-07} +{"train_loss": 0.013299131765961647, "grad_norm": 0.06712232530117035, "global_step": 51889, "epoch": 489, "lr": 1.0818755169672568e-07} +{"train_loss": 0.015432802960276604, "grad_norm": 0.11042039096355438, "global_step": 51890, "epoch": 489, "lr": 1.0799277714834133e-07} +{"train_loss": 0.014062101021409035, "grad_norm": 0.13408838212490082, "global_step": 51891, "epoch": 489, "lr": 1.0779817789877688e-07} +{"train_loss": 0.015800083056092262, "grad_norm": 0.09690164029598236, "global_step": 51892, "epoch": 489, "lr": 1.0760375394870404e-07} +{"train_loss": 0.014039197936654091, "grad_norm": 0.09251797944307327, "global_step": 51893, "epoch": 489, "lr": 1.0740950529882221e-07} +{"train_loss": 0.01428876630961895, "grad_norm": 0.0772065743803978, "global_step": 51894, "epoch": 489, "lr": 1.0721543194979755e-07} +{"train_loss": 0.014716120436787605, "grad_norm": 0.07789492607116699, "global_step": 51895, "epoch": 489, "lr": 1.0702153390232394e-07} +{"train_loss": 0.015515170991420746, "grad_norm": 0.07794040441513062, "global_step": 51896, "epoch": 489, "lr": 1.0682781115708418e-07} +{"train_loss": 0.015378923155367374, "grad_norm": 0.1570262908935547, "global_step": 51897, "epoch": 489, "lr": 1.0663426371474993e-07} +{"train_loss": 0.015145793557167053, "grad_norm": 0.09695777297019958, "global_step": 51898, "epoch": 489, "lr": 1.0644089157599845e-07} +{"train_loss": 0.016985151916742325, "grad_norm": 0.13524827361106873, "global_step": 51899, "epoch": 489, "lr": 1.0624769474152363e-07} +{"train_loss": 0.013779645785689354, "grad_norm": 0.05319581925868988, "global_step": 51900, "epoch": 489, "lr": 1.0605467321199158e-07} +{"train_loss": 0.015787940472364426, "grad_norm": 0.10974280536174774, "global_step": 51901, "epoch": 489, "lr": 1.0586182698808511e-07} +{"train_loss": 0.01739506982266903, "grad_norm": 0.10878460109233856, "global_step": 51902, "epoch": 489, "lr": 1.05669156070487e-07} +{"train_loss": 0.01671348698437214, "grad_norm": 0.06108896806836128, "global_step": 51903, "epoch": 489, "lr": 1.0547666045986337e-07} +{"train_loss": 0.016083544120192528, "grad_norm": 0.1080726608633995, "global_step": 51904, "epoch": 489, "lr": 1.0528434015689703e-07} +{"train_loss": 0.01366179995238781, "grad_norm": 0.06102731451392174, "global_step": 51905, "epoch": 489, "lr": 1.0509219516226521e-07} +{"train_loss": 0.014284866861999035, "grad_norm": 0.08247080445289612, "global_step": 51906, "epoch": 489, "lr": 1.0490022547663958e-07} +{"train_loss": 0.015347953885793686, "grad_norm": 0.08021959662437439, "global_step": 51907, "epoch": 489, "lr": 1.047084311006974e-07} +{"train_loss": 0.014163403771817684, "grad_norm": 0.06246708706021309, "global_step": 51908, "epoch": 489, "lr": 1.0451681203510477e-07} +{"train_loss": 0.016360586509108543, "grad_norm": 0.08578867465257645, "global_step": 51909, "epoch": 489, "lr": 1.0432536828054451e-07} +{"train_loss": 0.014161424711346626, "grad_norm": 0.0862979143857956, "global_step": 51910, "epoch": 489, "lr": 1.041340998376883e-07} +{"train_loss": 0.015328348614275455, "grad_norm": 0.07020533084869385, "global_step": 51911, "epoch": 489, "lr": 1.0394300670720781e-07} +{"train_loss": 0.015150964260101318, "grad_norm": 0.14352722465991974, "global_step": 51912, "epoch": 489, "lr": 1.0375208888976362e-07} +{"train_loss": 0.01656722091138363, "grad_norm": 0.06082412227988243, "global_step": 51913, "epoch": 489, "lr": 1.0356134638604409e-07} +{"train_loss": 0.013903316110372543, "grad_norm": 0.0912286564707756, "global_step": 51914, "epoch": 489, "lr": 1.0337077919670424e-07} +{"train_loss": 0.015895776450634003, "grad_norm": 0.07036273181438446, "global_step": 51915, "epoch": 489, "lr": 1.0318038732242685e-07} +{"train_loss": 0.016762452200055122, "grad_norm": 0.10808908939361572, "global_step": 51916, "epoch": 489, "lr": 1.0299017076386697e-07} +{"train_loss": 0.0142667843028903, "grad_norm": 0.06783609837293625, "global_step": 51917, "epoch": 489, "lr": 1.0280012952170182e-07} +{"train_loss": 0.014102727174758911, "grad_norm": 0.07558752596378326, "global_step": 51918, "epoch": 489, "lr": 1.0261026359659754e-07} +{"train_loss": 0.015065109357237816, "grad_norm": 0.0780356302857399, "global_step": 51919, "epoch": 489, "lr": 1.0242057298922581e-07} +{"train_loss": 0.01607365533709526, "grad_norm": 0.07074781507253647, "global_step": 51920, "epoch": 489, "lr": 1.0223105770024167e-07} +{"train_loss": 0.01394036877900362, "grad_norm": 0.0804438665509224, "global_step": 51921, "epoch": 489, "lr": 1.0204171773032234e-07} +{"train_loss": 0.014437089674174786, "grad_norm": 0.06614074856042862, "global_step": 51922, "epoch": 489, "lr": 1.0185255308012287e-07} +{"train_loss": 0.015028892084956169, "grad_norm": 0.08203792572021484, "global_step": 51923, "epoch": 489, "lr": 1.0166356375032049e-07} +{"train_loss": 0.0161278173327446, "grad_norm": 0.1564648449420929, "global_step": 51924, "epoch": 489, "lr": 1.0147474974156468e-07} +{"train_loss": 0.016135573387145996, "grad_norm": 0.06849916279315948, "global_step": 51925, "epoch": 489, "lr": 1.0128611105453267e-07} +{"train_loss": 0.014002654701471329, "grad_norm": 0.10396434366703033, "global_step": 51926, "epoch": 489, "lr": 1.0109764768987951e-07} +{"train_loss": 0.013871023431420326, "grad_norm": 0.10745726525783539, "global_step": 51927, "epoch": 489, "lr": 1.0090935964826576e-07} +{"train_loss": 0.014970320276916027, "grad_norm": 0.05827174708247185, "global_step": 51928, "epoch": 489, "lr": 1.0072124693035756e-07} +{"train_loss": 0.014966008253395557, "grad_norm": 0.06233803927898407, "global_step": 51929, "epoch": 489, "lr": 1.005333095368155e-07} +{"train_loss": 0.019191864877939224, "grad_norm": 0.11803451180458069, "global_step": 51930, "epoch": 489, "lr": 1.0034554746829461e-07} +{"train_loss": 0.01590139791369438, "grad_norm": 0.0856964960694313, "global_step": 51931, "epoch": 489, "lr": 1.0015796072546657e-07} +{"train_loss": 0.014226190745830536, "grad_norm": 0.07306195050477982, "global_step": 51932, "epoch": 489, "lr": 9.997054930897531e-08} +{"train_loss": 0.01476382091641426, "grad_norm": 0.05763797461986542, "global_step": 51933, "epoch": 489, "lr": 9.978331321948697e-08} +{"train_loss": 0.015880081802606583, "grad_norm": 0.059509288519620895, "global_step": 51934, "epoch": 489, "lr": 9.959625245766213e-08} +{"train_loss": 0.015659337863326073, "grad_norm": 0.06883500516414642, "global_step": 51935, "epoch": 489, "lr": 9.940936702415583e-08} +{"train_loss": 0.015002204105257988, "grad_norm": 0.11142574995756149, "global_step": 51936, "epoch": 489, "lr": 9.922265691962307e-08} +{"train_loss": 0.01409436110407114, "grad_norm": 0.06868911534547806, "global_step": 51937, "epoch": 489, "lr": 9.903612214471891e-08} +{"train_loss": 0.01400685589760542, "grad_norm": 0.08138235658407211, "global_step": 51938, "epoch": 489, "lr": 9.884976270009838e-08} +{"train_loss": 0.014942915723571237, "grad_norm": 0.13252834975719452, "global_step": 51939, "epoch": 489, "lr": 9.866357858642205e-08, "val_loss": 0.019775183871388435} +{"train_loss": 0.014787277206778526, "grad_norm": 0.10071828216314316, "global_step": 51940, "epoch": 490, "lr": 9.847756980433942e-08} +{"train_loss": 0.016904819756746292, "grad_norm": 0.13145723938941956, "global_step": 51941, "epoch": 490, "lr": 9.829173635450551e-08} +{"train_loss": 0.01591338962316513, "grad_norm": 0.09036362171173096, "global_step": 51942, "epoch": 490, "lr": 9.810607823756979e-08} +{"train_loss": 0.014892373234033585, "grad_norm": 0.11965291947126389, "global_step": 51943, "epoch": 490, "lr": 9.792059545419285e-08} +{"train_loss": 0.017319422215223312, "grad_norm": 0.08061953634023666, "global_step": 51944, "epoch": 490, "lr": 9.773528800501864e-08} +{"train_loss": 0.016107669100165367, "grad_norm": 0.11237439513206482, "global_step": 51945, "epoch": 490, "lr": 9.75501558906966e-08} +{"train_loss": 0.014386827126145363, "grad_norm": 0.07081106305122375, "global_step": 51946, "epoch": 490, "lr": 9.736519911188735e-08} +{"train_loss": 0.013610603287816048, "grad_norm": 0.1268823742866516, "global_step": 51947, "epoch": 490, "lr": 9.718041766922925e-08} +{"train_loss": 0.013911599293351173, "grad_norm": 0.08264392614364624, "global_step": 51948, "epoch": 490, "lr": 9.699581156338289e-08} +{"train_loss": 0.014892985112965107, "grad_norm": 0.15673144161701202, "global_step": 51949, "epoch": 490, "lr": 9.681138079498665e-08} +{"train_loss": 0.014799350872635841, "grad_norm": 0.08102414757013321, "global_step": 51950, "epoch": 490, "lr": 9.662712536469554e-08} +{"train_loss": 0.011667373590171337, "grad_norm": 0.0987035408616066, "global_step": 51951, "epoch": 490, "lr": 9.64430452731535e-08} +{"train_loss": 0.01489582285284996, "grad_norm": 0.09101353585720062, "global_step": 51952, "epoch": 490, "lr": 9.625914052101004e-08} +{"train_loss": 0.012976307421922684, "grad_norm": 0.07889755815267563, "global_step": 51953, "epoch": 490, "lr": 9.60754111089035e-08} +{"train_loss": 0.014644552022218704, "grad_norm": 0.09860827028751373, "global_step": 51954, "epoch": 490, "lr": 9.589185703748893e-08} +{"train_loss": 0.014610919170081615, "grad_norm": 0.05481347441673279, "global_step": 51955, "epoch": 490, "lr": 9.570847830741025e-08} +{"train_loss": 0.013756144791841507, "grad_norm": 0.07169133424758911, "global_step": 51956, "epoch": 490, "lr": 9.552527491930585e-08} +{"train_loss": 0.013886569067835808, "grad_norm": 0.10558813810348511, "global_step": 51957, "epoch": 490, "lr": 9.534224687382521e-08} +{"train_loss": 0.015344193205237389, "grad_norm": 0.0962572693824768, "global_step": 51958, "epoch": 490, "lr": 9.515939417160669e-08} +{"train_loss": 0.013535019010305405, "grad_norm": 0.116838239133358, "global_step": 51959, "epoch": 490, "lr": 9.497671681329423e-08} +{"train_loss": 0.016624759882688522, "grad_norm": 0.07236099243164062, "global_step": 51960, "epoch": 490, "lr": 9.479421479953177e-08} +{"train_loss": 0.01640116609632969, "grad_norm": 0.12943093478679657, "global_step": 51961, "epoch": 490, "lr": 9.461188813095768e-08} +{"train_loss": 0.015641245990991592, "grad_norm": 0.11916980147361755, "global_step": 51962, "epoch": 490, "lr": 9.442973680821588e-08} +{"train_loss": 0.013887914828956127, "grad_norm": 0.07353609055280685, "global_step": 51963, "epoch": 490, "lr": 9.424776083194475e-08} +{"train_loss": 0.01597973145544529, "grad_norm": 0.07320225238800049, "global_step": 51964, "epoch": 490, "lr": 9.406596020278268e-08} +{"train_loss": 0.014731300994753838, "grad_norm": 0.0965990275144577, "global_step": 51965, "epoch": 490, "lr": 9.38843349213736e-08} +{"train_loss": 0.01521976850926876, "grad_norm": 0.08957267552614212, "global_step": 51966, "epoch": 490, "lr": 9.370288498834478e-08} +{"train_loss": 0.01603035256266594, "grad_norm": 0.11985089629888535, "global_step": 51967, "epoch": 490, "lr": 9.35216104043457e-08} +{"train_loss": 0.01442032027989626, "grad_norm": 0.07030785083770752, "global_step": 51968, "epoch": 490, "lr": 9.334051117000364e-08} +{"train_loss": 0.014049902558326721, "grad_norm": 0.12370457500219345, "global_step": 51969, "epoch": 490, "lr": 9.315958728596252e-08} +{"train_loss": 0.015843285247683525, "grad_norm": 0.061895791441202164, "global_step": 51970, "epoch": 490, "lr": 9.297883875284963e-08} +{"train_loss": 0.01699564792215824, "grad_norm": 0.08394446969032288, "global_step": 51971, "epoch": 490, "lr": 9.279826557130889e-08} +{"train_loss": 0.014897698536515236, "grad_norm": 0.06318220496177673, "global_step": 51972, "epoch": 490, "lr": 9.261786774197312e-08} +{"train_loss": 0.013922015205025673, "grad_norm": 0.13865183293819427, "global_step": 51973, "epoch": 490, "lr": 9.243764526547515e-08} +{"train_loss": 0.012939037755131721, "grad_norm": 0.11812987178564072, "global_step": 51974, "epoch": 490, "lr": 9.225759814244228e-08} +{"train_loss": 0.014419661834836006, "grad_norm": 0.07296416163444519, "global_step": 51975, "epoch": 490, "lr": 9.207772637351286e-08} +{"train_loss": 0.012676196172833443, "grad_norm": 0.06138119846582413, "global_step": 51976, "epoch": 490, "lr": 9.189802995931973e-08} +{"train_loss": 0.015788786113262177, "grad_norm": 0.07935312390327454, "global_step": 51977, "epoch": 490, "lr": 9.171850890049017e-08} +{"train_loss": 0.018829263746738434, "grad_norm": 0.15114563703536987, "global_step": 51978, "epoch": 490, "lr": 9.153916319765698e-08} +{"train_loss": 0.012714569456875324, "grad_norm": 0.05896000936627388, "global_step": 51979, "epoch": 490, "lr": 9.135999285145302e-08} +{"train_loss": 0.016029540449380875, "grad_norm": 0.06666138768196106, "global_step": 51980, "epoch": 490, "lr": 9.118099786250001e-08} +{"train_loss": 0.015465926378965378, "grad_norm": 0.11670719087123871, "global_step": 51981, "epoch": 490, "lr": 9.10021782314363e-08} +{"train_loss": 0.014398826286196709, "grad_norm": 0.07005178928375244, "global_step": 51982, "epoch": 490, "lr": 9.08235339588892e-08} +{"train_loss": 0.014237241819500923, "grad_norm": 0.1176326647400856, "global_step": 51983, "epoch": 490, "lr": 9.064506504548043e-08} +{"train_loss": 0.01473703607916832, "grad_norm": 0.12674707174301147, "global_step": 51984, "epoch": 490, "lr": 9.046677149183724e-08} +{"train_loss": 0.014756642282009125, "grad_norm": 0.07133451104164124, "global_step": 51985, "epoch": 490, "lr": 9.028865329859249e-08} +{"train_loss": 0.014547396451234818, "grad_norm": 0.06378735601902008, "global_step": 51986, "epoch": 490, "lr": 9.011071046636787e-08} +{"train_loss": 0.013124809600412846, "grad_norm": 0.06439638882875443, "global_step": 51987, "epoch": 490, "lr": 8.993294299578514e-08} +{"train_loss": 0.015658270567655563, "grad_norm": 0.07213094085454941, "global_step": 51988, "epoch": 490, "lr": 8.975535088747711e-08} +{"train_loss": 0.015751443803310394, "grad_norm": 0.09873834997415543, "global_step": 51989, "epoch": 490, "lr": 8.957793414205995e-08} +{"train_loss": 0.014803841710090637, "grad_norm": 0.08657346665859222, "global_step": 51990, "epoch": 490, "lr": 8.940069276016649e-08} +{"train_loss": 0.012332463636994362, "grad_norm": 0.12236066907644272, "global_step": 51991, "epoch": 490, "lr": 8.922362674240736e-08} +{"train_loss": 0.014101732522249222, "grad_norm": 0.07848288118839264, "global_step": 51992, "epoch": 490, "lr": 8.904673608940983e-08} +{"train_loss": 0.014406684786081314, "grad_norm": 0.07725891470909119, "global_step": 51993, "epoch": 490, "lr": 8.887002080180118e-08} +{"train_loss": 0.018161874264478683, "grad_norm": 0.07185645401477814, "global_step": 51994, "epoch": 490, "lr": 8.869348088019203e-08} +{"train_loss": 0.016030415892601013, "grad_norm": 0.10265568643808365, "global_step": 51995, "epoch": 490, "lr": 8.85171163252152e-08} +{"train_loss": 0.014825057238340378, "grad_norm": 0.11643047630786896, "global_step": 51996, "epoch": 490, "lr": 8.834092713748132e-08} +{"train_loss": 0.015180448070168495, "grad_norm": 0.07529965043067932, "global_step": 51997, "epoch": 490, "lr": 8.816491331760657e-08} +{"train_loss": 0.01318762730807066, "grad_norm": 0.06330403685569763, "global_step": 51998, "epoch": 490, "lr": 8.798907486621822e-08} +{"train_loss": 0.016616422683000565, "grad_norm": 0.09257493168115616, "global_step": 51999, "epoch": 490, "lr": 8.781341178393244e-08} +{"train_loss": 0.014828981831669807, "grad_norm": 0.08191053569316864, "global_step": 52000, "epoch": 490, "lr": 8.763792407135985e-08} +{"train_loss": 0.015320378355681896, "grad_norm": 0.0897732824087143, "global_step": 52001, "epoch": 490, "lr": 8.746261172912217e-08} +{"train_loss": 0.01614069752395153, "grad_norm": 0.11614222079515457, "global_step": 52002, "epoch": 490, "lr": 8.72874747578356e-08} +{"train_loss": 0.013777341693639755, "grad_norm": 0.07076668739318848, "global_step": 52003, "epoch": 490, "lr": 8.711251315811076e-08} +{"train_loss": 0.014721480198204517, "grad_norm": 0.06936173886060715, "global_step": 52004, "epoch": 490, "lr": 8.693772693056934e-08} +{"train_loss": 0.015056719072163105, "grad_norm": 0.05624917522072792, "global_step": 52005, "epoch": 490, "lr": 8.676311607582199e-08} +{"train_loss": 0.013278952799737453, "grad_norm": 0.05843305587768555, "global_step": 52006, "epoch": 490, "lr": 8.658868059447933e-08} +{"train_loss": 0.014652134850621223, "grad_norm": 0.0928645133972168, "global_step": 52007, "epoch": 490, "lr": 8.641442048716308e-08} +{"train_loss": 0.015318457037210464, "grad_norm": 0.07127202302217484, "global_step": 52008, "epoch": 490, "lr": 8.624033575447831e-08} +{"train_loss": 0.015236668288707733, "grad_norm": 0.08811109513044357, "global_step": 52009, "epoch": 490, "lr": 8.606642639703566e-08} +{"train_loss": 0.015666473656892776, "grad_norm": 0.08552829176187515, "global_step": 52010, "epoch": 490, "lr": 8.589269241544573e-08} +{"train_loss": 0.01490507461130619, "grad_norm": 0.10296942293643951, "global_step": 52011, "epoch": 490, "lr": 8.57191338103247e-08} +{"train_loss": 0.01660131849348545, "grad_norm": 0.11849730461835861, "global_step": 52012, "epoch": 490, "lr": 8.55457505822832e-08} +{"train_loss": 0.01526382751762867, "grad_norm": 0.07370192557573318, "global_step": 52013, "epoch": 490, "lr": 8.537254273192075e-08} +{"train_loss": 0.016041025519371033, "grad_norm": 0.07063039392232895, "global_step": 52014, "epoch": 490, "lr": 8.519951025985906e-08} +{"train_loss": 0.01605869084596634, "grad_norm": 0.10353107750415802, "global_step": 52015, "epoch": 490, "lr": 8.502665316669211e-08} +{"train_loss": 0.014270504005253315, "grad_norm": 0.0861765593290329, "global_step": 52016, "epoch": 490, "lr": 8.485397145303609e-08} +{"train_loss": 0.017026733607053757, "grad_norm": 0.10699713230133057, "global_step": 52017, "epoch": 490, "lr": 8.468146511949604e-08} +{"train_loss": 0.015201195143163204, "grad_norm": 0.13159415125846863, "global_step": 52018, "epoch": 490, "lr": 8.450913416667705e-08} +{"train_loss": 0.014835325069725513, "grad_norm": 0.0931842252612114, "global_step": 52019, "epoch": 490, "lr": 8.433697859518975e-08} +{"train_loss": 0.014574543572962284, "grad_norm": 0.12285038828849792, "global_step": 52020, "epoch": 490, "lr": 8.416499840562808e-08} +{"train_loss": 0.012572405859827995, "grad_norm": 0.07191513478755951, "global_step": 52021, "epoch": 490, "lr": 8.399319359860824e-08} +{"train_loss": 0.015317118726670742, "grad_norm": 0.10311257839202881, "global_step": 52022, "epoch": 490, "lr": 8.38215641747242e-08} +{"train_loss": 0.014296742156147957, "grad_norm": 0.06025930121541023, "global_step": 52023, "epoch": 490, "lr": 8.365011013458657e-08} +{"train_loss": 0.013723589479923248, "grad_norm": 0.06956210732460022, "global_step": 52024, "epoch": 490, "lr": 8.347883147878932e-08} +{"train_loss": 0.01596599817276001, "grad_norm": 0.09508700668811798, "global_step": 52025, "epoch": 490, "lr": 8.330772820794308e-08} +{"train_loss": 0.015402533113956451, "grad_norm": 0.07508828490972519, "global_step": 52026, "epoch": 490, "lr": 8.313680032264737e-08} +{"train_loss": 0.014480454847216606, "grad_norm": 0.12348224222660065, "global_step": 52027, "epoch": 490, "lr": 8.296604782349615e-08} +{"train_loss": 0.016007140278816223, "grad_norm": 0.12635846436023712, "global_step": 52028, "epoch": 490, "lr": 8.27954707110945e-08} +{"train_loss": 0.015612524002790451, "grad_norm": 0.08581417053937912, "global_step": 52029, "epoch": 490, "lr": 8.262506898604194e-08} +{"train_loss": 0.015751374885439873, "grad_norm": 0.06742341071367264, "global_step": 52030, "epoch": 490, "lr": 8.245484264893799e-08} +{"train_loss": 0.014096636325120926, "grad_norm": 0.08677054196596146, "global_step": 52031, "epoch": 490, "lr": 8.228479170037661e-08} +{"train_loss": 0.014766933396458626, "grad_norm": 0.09857480227947235, "global_step": 52032, "epoch": 490, "lr": 8.21149161409629e-08} +{"train_loss": 0.014600388705730438, "grad_norm": 0.0831378847360611, "global_step": 52033, "epoch": 490, "lr": 8.194521597128524e-08} +{"train_loss": 0.017168140038847923, "grad_norm": 0.08883047848939896, "global_step": 52034, "epoch": 490, "lr": 8.177569119194317e-08} +{"train_loss": 0.013834713026881218, "grad_norm": 0.06272928416728973, "global_step": 52035, "epoch": 490, "lr": 8.160634180353066e-08} +{"train_loss": 0.01413873489946127, "grad_norm": 0.06123126670718193, "global_step": 52036, "epoch": 490, "lr": 8.143716780664722e-08} +{"train_loss": 0.012529062107205391, "grad_norm": 0.057887397706508636, "global_step": 52037, "epoch": 490, "lr": 8.126816920188129e-08} +{"train_loss": 0.017112769186496735, "grad_norm": 0.07044706493616104, "global_step": 52038, "epoch": 490, "lr": 8.109934598983238e-08} +{"train_loss": 0.01473136804997921, "grad_norm": 0.06111275777220726, "global_step": 52039, "epoch": 490, "lr": 8.093069817109445e-08} +{"train_loss": 0.012859418988227844, "grad_norm": 0.05179216340184212, "global_step": 52040, "epoch": 490, "lr": 8.076222574625036e-08} +{"train_loss": 0.01693132519721985, "grad_norm": 0.07490136474370956, "global_step": 52041, "epoch": 490, "lr": 8.059392871590521e-08} +{"train_loss": 0.016322065144777298, "grad_norm": 0.11390658468008041, "global_step": 52042, "epoch": 490, "lr": 8.042580708063629e-08} +{"train_loss": 0.013559233397245407, "grad_norm": 0.09261046350002289, "global_step": 52043, "epoch": 490, "lr": 8.02578608410487e-08} +{"train_loss": 0.01293628104031086, "grad_norm": 0.058679260313510895, "global_step": 52044, "epoch": 490, "lr": 8.009008999771971e-08} +{"train_loss": 0.014964104261038438, "grad_norm": 0.0892491415143013, "global_step": 52045, "epoch": 490, "lr": 7.99224945512489e-08, "train_batch_mse_error": 0.01336434856057167, "val_loss": 0.019548067823052406} +{"train_loss": 0.013746645301580429, "grad_norm": 0.056880105286836624, "global_step": 52046, "epoch": 491, "lr": 7.975507450221909e-08} +{"train_loss": 0.014014698565006256, "grad_norm": 0.05781923234462738, "global_step": 52047, "epoch": 491, "lr": 7.95878298512187e-08} +{"train_loss": 0.014346776530146599, "grad_norm": 0.06955263018608093, "global_step": 52048, "epoch": 491, "lr": 7.942076059883618e-08} +{"train_loss": 0.014635254628956318, "grad_norm": 0.0695635974407196, "global_step": 52049, "epoch": 491, "lr": 7.925386674566549e-08} +{"train_loss": 0.013958072289824486, "grad_norm": 0.054007723927497864, "global_step": 52050, "epoch": 491, "lr": 7.908714829227837e-08} +{"train_loss": 0.014007056131958961, "grad_norm": 0.06217079609632492, "global_step": 52051, "epoch": 491, "lr": 7.892060523927436e-08} +{"train_loss": 0.013068333268165588, "grad_norm": 0.0787082090973854, "global_step": 52052, "epoch": 491, "lr": 7.875423758723077e-08} +{"train_loss": 0.013300672173500061, "grad_norm": 0.07390962541103363, "global_step": 52053, "epoch": 491, "lr": 7.858804533673602e-08} +{"train_loss": 0.014590837061405182, "grad_norm": 0.05521424487233162, "global_step": 52054, "epoch": 491, "lr": 7.842202848837299e-08} +{"train_loss": 0.014148116111755371, "grad_norm": 0.07927112281322479, "global_step": 52055, "epoch": 491, "lr": 7.825618704272453e-08} +{"train_loss": 0.01538480632007122, "grad_norm": 0.13379424810409546, "global_step": 52056, "epoch": 491, "lr": 7.80905210003735e-08} +{"train_loss": 0.015116013586521149, "grad_norm": 0.08661632984876633, "global_step": 52057, "epoch": 491, "lr": 7.79250303619028e-08} +{"train_loss": 0.01594572328031063, "grad_norm": 0.12671169638633728, "global_step": 52058, "epoch": 491, "lr": 7.775971512788971e-08} +{"train_loss": 0.013987405225634575, "grad_norm": 0.0788579136133194, "global_step": 52059, "epoch": 491, "lr": 7.759457529892267e-08} +{"train_loss": 0.014912092126905918, "grad_norm": 0.06033102795481682, "global_step": 52060, "epoch": 491, "lr": 7.742961087557343e-08} +{"train_loss": 0.01596800982952118, "grad_norm": 0.06364276260137558, "global_step": 52061, "epoch": 491, "lr": 7.726482185843042e-08} +{"train_loss": 0.012986263260245323, "grad_norm": 0.06926491111516953, "global_step": 52062, "epoch": 491, "lr": 7.710020824806541e-08} +{"train_loss": 0.012665513902902603, "grad_norm": 0.1059238389134407, "global_step": 52063, "epoch": 491, "lr": 7.69357700450557e-08} +{"train_loss": 0.016579298302531242, "grad_norm": 0.06266925483942032, "global_step": 52064, "epoch": 491, "lr": 7.677150724998971e-08} +{"train_loss": 0.0156989898532629, "grad_norm": 0.1703629046678543, "global_step": 52065, "epoch": 491, "lr": 7.660741986343367e-08} +{"train_loss": 0.01363641582429409, "grad_norm": 0.08610168099403381, "global_step": 52066, "epoch": 491, "lr": 7.644350788596489e-08} +{"train_loss": 0.0158558152616024, "grad_norm": 0.06359899789094925, "global_step": 52067, "epoch": 491, "lr": 7.627977131816622e-08} +{"train_loss": 0.012867939658463001, "grad_norm": 0.07410632818937302, "global_step": 52068, "epoch": 491, "lr": 7.611621016060944e-08} +{"train_loss": 0.015404476784169674, "grad_norm": 0.06529926508665085, "global_step": 52069, "epoch": 491, "lr": 7.595282441386631e-08} +{"train_loss": 0.014046937227249146, "grad_norm": 0.07622366398572922, "global_step": 52070, "epoch": 491, "lr": 7.57896140785197e-08} +{"train_loss": 0.01566174067556858, "grad_norm": 0.11014635860919952, "global_step": 52071, "epoch": 491, "lr": 7.562657915513027e-08} +{"train_loss": 0.01900481805205345, "grad_norm": 0.10486506670713425, "global_step": 52072, "epoch": 491, "lr": 7.54637196442809e-08} +{"train_loss": 0.015182813629508018, "grad_norm": 0.05859482288360596, "global_step": 52073, "epoch": 491, "lr": 7.530103554653778e-08} +{"train_loss": 0.014772061258554459, "grad_norm": 0.0643988624215126, "global_step": 52074, "epoch": 491, "lr": 7.513852686247825e-08} +{"train_loss": 0.015260142274200916, "grad_norm": 0.0738593116402626, "global_step": 52075, "epoch": 491, "lr": 7.49761935926685e-08} +{"train_loss": 0.01485393289476633, "grad_norm": 0.10333336889743805, "global_step": 52076, "epoch": 491, "lr": 7.481403573768031e-08} +{"train_loss": 0.012377273291349411, "grad_norm": 0.05670488625764847, "global_step": 52077, "epoch": 491, "lr": 7.465205329808545e-08} +{"train_loss": 0.015217681415379047, "grad_norm": 0.059887707233428955, "global_step": 52078, "epoch": 491, "lr": 7.449024627445012e-08} +{"train_loss": 0.016498707234859467, "grad_norm": 0.06145934760570526, "global_step": 52079, "epoch": 491, "lr": 7.432861466734608e-08} +{"train_loss": 0.014040028676390648, "grad_norm": 0.052635323256254196, "global_step": 52080, "epoch": 491, "lr": 7.416715847733958e-08} +{"train_loss": 0.013440923765301704, "grad_norm": 0.10496055334806442, "global_step": 52081, "epoch": 491, "lr": 7.400587770499124e-08} +{"train_loss": 0.015387605875730515, "grad_norm": 0.0973840057849884, "global_step": 52082, "epoch": 491, "lr": 7.384477235087839e-08} +{"train_loss": 0.013662654906511307, "grad_norm": 0.05720331519842148, "global_step": 52083, "epoch": 491, "lr": 7.36838424155617e-08} +{"train_loss": 0.01478264108300209, "grad_norm": 0.08611470460891724, "global_step": 52084, "epoch": 491, "lr": 7.352308789961293e-08} +{"train_loss": 0.014821101911365986, "grad_norm": 0.10101162642240524, "global_step": 52085, "epoch": 491, "lr": 7.336250880358719e-08} +{"train_loss": 0.01795221120119095, "grad_norm": 0.0972660481929779, "global_step": 52086, "epoch": 491, "lr": 7.32021051280507e-08} +{"train_loss": 0.01441357284784317, "grad_norm": 0.0866575688123703, "global_step": 52087, "epoch": 491, "lr": 7.304187687356967e-08} +{"train_loss": 0.014480536803603172, "grad_norm": 0.10387036949396133, "global_step": 52088, "epoch": 491, "lr": 7.288182404071031e-08} +{"train_loss": 0.014249442145228386, "grad_norm": 0.09245482832193375, "global_step": 52089, "epoch": 491, "lr": 7.272194663002774e-08} +{"train_loss": 0.013505334034562111, "grad_norm": 0.09177900105714798, "global_step": 52090, "epoch": 491, "lr": 7.256224464209372e-08} +{"train_loss": 0.013508068397641182, "grad_norm": 0.08454329520463943, "global_step": 52091, "epoch": 491, "lr": 7.24027180774578e-08} +{"train_loss": 0.016359703615307808, "grad_norm": 0.08955320715904236, "global_step": 52092, "epoch": 491, "lr": 7.224336693668621e-08} +{"train_loss": 0.0127305518835783, "grad_norm": 0.062449827790260315, "global_step": 52093, "epoch": 491, "lr": 7.20841912203396e-08} +{"train_loss": 0.015115944668650627, "grad_norm": 0.054286327213048935, "global_step": 52094, "epoch": 491, "lr": 7.192519092897309e-08} +{"train_loss": 0.015135745517909527, "grad_norm": 0.0797928124666214, "global_step": 52095, "epoch": 491, "lr": 7.17663660631529e-08} +{"train_loss": 0.01407633163034916, "grad_norm": 0.08557654917240143, "global_step": 52096, "epoch": 491, "lr": 7.160771662342857e-08} +{"train_loss": 0.015028305351734161, "grad_norm": 0.07902169227600098, "global_step": 52097, "epoch": 491, "lr": 7.144924261036634e-08} +{"train_loss": 0.015620887279510498, "grad_norm": 0.07690548896789551, "global_step": 52098, "epoch": 491, "lr": 7.129094402451575e-08} +{"train_loss": 0.015510182827711105, "grad_norm": 0.07890985161066055, "global_step": 52099, "epoch": 491, "lr": 7.113282086643191e-08} +{"train_loss": 0.016488291323184967, "grad_norm": 0.1283864825963974, "global_step": 52100, "epoch": 491, "lr": 7.09748731366755e-08} +{"train_loss": 0.01729811355471611, "grad_norm": 0.09927580505609512, "global_step": 52101, "epoch": 491, "lr": 7.081710083580162e-08} +{"train_loss": 0.015697820112109184, "grad_norm": 0.09071867167949677, "global_step": 52102, "epoch": 491, "lr": 7.065950396436538e-08} +{"train_loss": 0.014681551605463028, "grad_norm": 0.056897521018981934, "global_step": 52103, "epoch": 491, "lr": 7.05020825229108e-08} +{"train_loss": 0.015270860865712166, "grad_norm": 0.08245516568422318, "global_step": 52104, "epoch": 491, "lr": 7.034483651200407e-08} +{"train_loss": 0.015520622953772545, "grad_norm": 0.12511689960956573, "global_step": 52105, "epoch": 491, "lr": 7.018776593218923e-08} +{"train_loss": 0.013464441522955894, "grad_norm": 0.0999990925192833, "global_step": 52106, "epoch": 491, "lr": 7.003087078402138e-08} +{"train_loss": 0.015446875244379044, "grad_norm": 0.13823093473911285, "global_step": 52107, "epoch": 491, "lr": 6.987415106805007e-08} +{"train_loss": 0.015179445967078209, "grad_norm": 0.08762049674987793, "global_step": 52108, "epoch": 491, "lr": 6.971760678483041e-08} +{"train_loss": 0.013613875955343246, "grad_norm": 0.06196364015340805, "global_step": 52109, "epoch": 491, "lr": 6.956123793490643e-08} +{"train_loss": 0.01831250637769699, "grad_norm": 0.07775288075208664, "global_step": 52110, "epoch": 491, "lr": 6.940504451882768e-08} +{"train_loss": 0.018841585144400597, "grad_norm": 0.08240261673927307, "global_step": 52111, "epoch": 491, "lr": 6.924902653714927e-08} +{"train_loss": 0.015369454398751259, "grad_norm": 0.14507628977298737, "global_step": 52112, "epoch": 491, "lr": 6.90931839904152e-08} +{"train_loss": 0.014186406508088112, "grad_norm": 0.11372881382703781, "global_step": 52113, "epoch": 491, "lr": 6.893751687916949e-08} +{"train_loss": 0.016400426626205444, "grad_norm": 0.07082828134298325, "global_step": 52114, "epoch": 491, "lr": 6.878202520396727e-08} +{"train_loss": 0.013726945966482162, "grad_norm": 0.060621120035648346, "global_step": 52115, "epoch": 491, "lr": 6.862670896534696e-08} +{"train_loss": 0.015589487738907337, "grad_norm": 0.08747053891420364, "global_step": 52116, "epoch": 491, "lr": 6.847156816385814e-08} +{"train_loss": 0.015181534923613071, "grad_norm": 0.07344435155391693, "global_step": 52117, "epoch": 491, "lr": 6.831660280004481e-08} +{"train_loss": 0.013904742896556854, "grad_norm": 0.06038610264658928, "global_step": 52118, "epoch": 491, "lr": 6.816181287445656e-08} +{"train_loss": 0.012178434990346432, "grad_norm": 0.05289753898978233, "global_step": 52119, "epoch": 491, "lr": 6.800719838763182e-08} +{"train_loss": 0.015166077762842178, "grad_norm": 0.06523016095161438, "global_step": 52120, "epoch": 491, "lr": 6.78527593401146e-08} +{"train_loss": 0.015389267355203629, "grad_norm": 0.07992570102214813, "global_step": 52121, "epoch": 491, "lr": 6.769849573244336e-08} +{"train_loss": 0.016824228689074516, "grad_norm": 0.08508843928575516, "global_step": 52122, "epoch": 491, "lr": 6.754440756517321e-08} +{"train_loss": 0.01507265493273735, "grad_norm": 0.06724059581756592, "global_step": 52123, "epoch": 491, "lr": 6.739049483883153e-08} +{"train_loss": 0.015446682460606098, "grad_norm": 0.10928994417190552, "global_step": 52124, "epoch": 491, "lr": 6.72367575539623e-08} +{"train_loss": 0.01333516277372837, "grad_norm": 0.04544563218951225, "global_step": 52125, "epoch": 491, "lr": 6.708319571110954e-08} +{"train_loss": 0.01588781550526619, "grad_norm": 0.11873593181371689, "global_step": 52126, "epoch": 491, "lr": 6.692980931081727e-08} +{"train_loss": 0.015479588881134987, "grad_norm": 0.0933053195476532, "global_step": 52127, "epoch": 491, "lr": 6.677659835361283e-08} +{"train_loss": 0.015123278833925724, "grad_norm": 0.04993751272559166, "global_step": 52128, "epoch": 491, "lr": 6.662356284004023e-08} +{"train_loss": 0.015092437155544758, "grad_norm": 0.08296263217926025, "global_step": 52129, "epoch": 491, "lr": 6.647070277063794e-08} +{"train_loss": 0.015876980498433113, "grad_norm": 0.055335063487291336, "global_step": 52130, "epoch": 491, "lr": 6.631801814593885e-08} +{"train_loss": 0.016926974058151245, "grad_norm": 0.11825087666511536, "global_step": 52131, "epoch": 491, "lr": 6.616550896648699e-08} +{"train_loss": 0.014747371897101402, "grad_norm": 0.09405572712421417, "global_step": 52132, "epoch": 491, "lr": 6.60131752328097e-08} +{"train_loss": 0.013891912996768951, "grad_norm": 0.06167781352996826, "global_step": 52133, "epoch": 491, "lr": 6.586101694545099e-08} +{"train_loss": 0.016656549647450447, "grad_norm": 0.09597519785165787, "global_step": 52134, "epoch": 491, "lr": 6.570903410493822e-08} +{"train_loss": 0.014059264212846756, "grad_norm": 0.11430045962333679, "global_step": 52135, "epoch": 491, "lr": 6.555722671180986e-08} +{"train_loss": 0.01559047307819128, "grad_norm": 0.0739544928073883, "global_step": 52136, "epoch": 491, "lr": 6.540559476659325e-08} +{"train_loss": 0.01673281565308571, "grad_norm": 0.10586245357990265, "global_step": 52137, "epoch": 491, "lr": 6.525413826982685e-08} +{"train_loss": 0.013301857747137547, "grad_norm": 0.07777263969182968, "global_step": 52138, "epoch": 491, "lr": 6.510285722204357e-08} +{"train_loss": 0.016263727098703384, "grad_norm": 0.07608017325401306, "global_step": 52139, "epoch": 491, "lr": 6.495175162377631e-08} +{"train_loss": 0.01613173447549343, "grad_norm": 0.07517373561859131, "global_step": 52140, "epoch": 491, "lr": 6.480082147554689e-08} +{"train_loss": 0.014764550141990185, "grad_norm": 0.05447027459740639, "global_step": 52141, "epoch": 491, "lr": 6.465006677789376e-08} +{"train_loss": 0.0132021214812994, "grad_norm": 0.06652595847845078, "global_step": 52142, "epoch": 491, "lr": 6.449948753133872e-08} +{"train_loss": 0.01743628829717636, "grad_norm": 0.07207895815372467, "global_step": 52143, "epoch": 491, "lr": 6.434908373642578e-08} +{"train_loss": 0.01625591143965721, "grad_norm": 0.12373847514390945, "global_step": 52144, "epoch": 491, "lr": 6.419885539366566e-08} +{"train_loss": 0.013927323743700981, "grad_norm": 0.09561832994222641, "global_step": 52145, "epoch": 491, "lr": 6.40488025035968e-08} +{"train_loss": 0.015103192999958992, "grad_norm": 0.10041096806526184, "global_step": 52146, "epoch": 491, "lr": 6.3898925066741e-08} +{"train_loss": 0.01414930447936058, "grad_norm": 0.06552603095769882, "global_step": 52147, "epoch": 491, "lr": 6.374922308363119e-08} +{"train_loss": 0.013668954372406006, "grad_norm": 0.0824742466211319, "global_step": 52148, "epoch": 491, "lr": 6.359969655478915e-08} +{"train_loss": 0.014626809395849705, "grad_norm": 0.14342732727527618, "global_step": 52149, "epoch": 491, "lr": 6.345034548074224e-08} +{"train_loss": 0.013841977342963219, "grad_norm": 0.0662979707121849, "global_step": 52150, "epoch": 491, "lr": 6.330116986201228e-08} +{"train_loss": 0.014944056725516072, "grad_norm": 0.08049342036247253, "global_step": 52151, "epoch": 491, "lr": 6.315216969912663e-08, "val_loss": 0.019432809203863144} +{"train_loss": 0.015458841808140278, "grad_norm": 0.07043688744306564, "global_step": 52152, "epoch": 492, "lr": 6.300334499260708e-08} +{"train_loss": 0.015759799629449844, "grad_norm": 0.06114458292722702, "global_step": 52153, "epoch": 492, "lr": 6.285469574297542e-08} +{"train_loss": 0.014572052285075188, "grad_norm": 0.08096518367528915, "global_step": 52154, "epoch": 492, "lr": 6.270622195075348e-08} +{"train_loss": 0.014496084302663803, "grad_norm": 0.09573197364807129, "global_step": 52155, "epoch": 492, "lr": 6.255792361646862e-08} +{"train_loss": 0.013815023936331272, "grad_norm": 0.08759848773479462, "global_step": 52156, "epoch": 492, "lr": 6.240980074063708e-08} +{"train_loss": 0.01486880425363779, "grad_norm": 0.11418928951025009, "global_step": 52157, "epoch": 492, "lr": 6.226185332378065e-08} +{"train_loss": 0.014849777333438396, "grad_norm": 0.07107409089803696, "global_step": 52158, "epoch": 492, "lr": 6.211408136641561e-08} +{"train_loss": 0.015750031918287277, "grad_norm": 0.06768577545881271, "global_step": 52159, "epoch": 492, "lr": 6.196648486906375e-08} +{"train_loss": 0.014867870137095451, "grad_norm": 0.12672507762908936, "global_step": 52160, "epoch": 492, "lr": 6.18190638322469e-08} +{"train_loss": 0.017043953761458397, "grad_norm": 0.06690123677253723, "global_step": 52161, "epoch": 492, "lr": 6.167181825648127e-08} +{"train_loss": 0.01574045978486538, "grad_norm": 0.06764146685600281, "global_step": 52162, "epoch": 492, "lr": 6.152474814228315e-08} +{"train_loss": 0.013655491173267365, "grad_norm": 0.06167704612016678, "global_step": 52163, "epoch": 492, "lr": 6.137785349016877e-08} +{"train_loss": 0.014163235202431679, "grad_norm": 0.06709206849336624, "global_step": 52164, "epoch": 492, "lr": 6.123113430065441e-08} +{"train_loss": 0.014213546179234982, "grad_norm": 0.07590039819478989, "global_step": 52165, "epoch": 492, "lr": 6.108459057425631e-08} +{"train_loss": 0.015840575098991394, "grad_norm": 0.1100984662771225, "global_step": 52166, "epoch": 492, "lr": 6.093822231148515e-08} +{"train_loss": 0.017969943583011627, "grad_norm": 0.07671230286359787, "global_step": 52167, "epoch": 492, "lr": 6.079202951286833e-08} +{"train_loss": 0.013988086022436619, "grad_norm": 0.11569943279027939, "global_step": 52168, "epoch": 492, "lr": 6.064601217889987e-08} +{"train_loss": 0.016202354803681374, "grad_norm": 0.08394425362348557, "global_step": 52169, "epoch": 492, "lr": 6.050017031011269e-08} +{"train_loss": 0.014434373006224632, "grad_norm": 0.06980687379837036, "global_step": 52170, "epoch": 492, "lr": 6.035450390700081e-08} +{"train_loss": 0.013509264215826988, "grad_norm": 0.06598955392837524, "global_step": 52171, "epoch": 492, "lr": 6.020901297009162e-08} +{"train_loss": 0.013575906865298748, "grad_norm": 0.05183020606637001, "global_step": 52172, "epoch": 492, "lr": 6.006369749988471e-08} +{"train_loss": 0.014933886006474495, "grad_norm": 0.07647577673196793, "global_step": 52173, "epoch": 492, "lr": 5.991855749690189e-08} +{"train_loss": 0.015619121491909027, "grad_norm": 0.07020336389541626, "global_step": 52174, "epoch": 492, "lr": 5.977359296164276e-08} +{"train_loss": 0.014610303565859795, "grad_norm": 0.07052630931138992, "global_step": 52175, "epoch": 492, "lr": 5.962880389462356e-08} +{"train_loss": 0.01492992416024208, "grad_norm": 0.05691862106323242, "global_step": 52176, "epoch": 492, "lr": 5.948419029634944e-08} +{"train_loss": 0.016994576901197433, "grad_norm": 0.08538839966058731, "global_step": 52177, "epoch": 492, "lr": 5.933975216732557e-08} +{"train_loss": 0.013179243542253971, "grad_norm": 0.06399450451135635, "global_step": 52178, "epoch": 492, "lr": 5.919548950806819e-08} +{"train_loss": 0.012893655337393284, "grad_norm": 0.08043448626995087, "global_step": 52179, "epoch": 492, "lr": 5.905140231907691e-08} +{"train_loss": 0.013210274279117584, "grad_norm": 0.08511996269226074, "global_step": 52180, "epoch": 492, "lr": 5.8907490600856873e-08} +{"train_loss": 0.014870798215270042, "grad_norm": 0.05620889738202095, "global_step": 52181, "epoch": 492, "lr": 5.8763754353924336e-08} +{"train_loss": 0.01477246917784214, "grad_norm": 0.09960130602121353, "global_step": 52182, "epoch": 492, "lr": 5.8620193578773355e-08} +{"train_loss": 0.013575725257396698, "grad_norm": 0.11542408168315887, "global_step": 52183, "epoch": 492, "lr": 5.8476808275909064e-08} +{"train_loss": 0.014915076084434986, "grad_norm": 0.08538077026605606, "global_step": 52184, "epoch": 492, "lr": 5.8333598445842187e-08} +{"train_loss": 0.01338669192045927, "grad_norm": 0.08576236665248871, "global_step": 52185, "epoch": 492, "lr": 5.8190564089072305e-08} +{"train_loss": 0.01565023884177208, "grad_norm": 0.07618162035942078, "global_step": 52186, "epoch": 492, "lr": 5.804770520609904e-08} +{"train_loss": 0.012669863179326057, "grad_norm": 0.08694427460432053, "global_step": 52187, "epoch": 492, "lr": 5.790502179742752e-08} +{"train_loss": 0.015072537586092949, "grad_norm": 0.10871993005275726, "global_step": 52188, "epoch": 492, "lr": 5.7762513863557355e-08} +{"train_loss": 0.014308243989944458, "grad_norm": 0.0725000873208046, "global_step": 52189, "epoch": 492, "lr": 5.76201814049937e-08} +{"train_loss": 0.013102593831717968, "grad_norm": 0.07531864196062088, "global_step": 52190, "epoch": 492, "lr": 5.747802442223615e-08} +{"train_loss": 0.014275871217250824, "grad_norm": 0.09323537349700928, "global_step": 52191, "epoch": 492, "lr": 5.7336042915773216e-08} +{"train_loss": 0.013074513524770737, "grad_norm": 0.10909126698970795, "global_step": 52192, "epoch": 492, "lr": 5.719423688611558e-08} +{"train_loss": 0.012837998569011688, "grad_norm": 0.11728741973638535, "global_step": 52193, "epoch": 492, "lr": 5.705260633375731e-08} +{"train_loss": 0.015878571197390556, "grad_norm": 0.10147065669298172, "global_step": 52194, "epoch": 492, "lr": 5.6911151259197996e-08} +{"train_loss": 0.014691075310111046, "grad_norm": 0.0690118744969368, "global_step": 52195, "epoch": 492, "lr": 5.6769871662931686e-08} +{"train_loss": 0.015773529186844826, "grad_norm": 0.0541810467839241, "global_step": 52196, "epoch": 492, "lr": 5.662876754545798e-08} +{"train_loss": 0.015922751277685165, "grad_norm": 0.10628034174442291, "global_step": 52197, "epoch": 492, "lr": 5.648783890727094e-08} +{"train_loss": 0.01564924418926239, "grad_norm": 0.1057094931602478, "global_step": 52198, "epoch": 492, "lr": 5.63470857488646e-08} +{"train_loss": 0.013758478686213493, "grad_norm": 0.07990089058876038, "global_step": 52199, "epoch": 492, "lr": 5.620650807073857e-08} +{"train_loss": 0.015552900731563568, "grad_norm": 0.06998343020677567, "global_step": 52200, "epoch": 492, "lr": 5.6066105873375795e-08} +{"train_loss": 0.014571189880371094, "grad_norm": 0.0951543003320694, "global_step": 52201, "epoch": 492, "lr": 5.5925879157281424e-08} +{"train_loss": 0.012768035754561424, "grad_norm": 0.07820088416337967, "global_step": 52202, "epoch": 492, "lr": 5.5785827922943956e-08} +{"train_loss": 0.016567986458539963, "grad_norm": 0.0725022703409195, "global_step": 52203, "epoch": 492, "lr": 5.564595217085189e-08} +{"train_loss": 0.014033995568752289, "grad_norm": 0.07138024270534515, "global_step": 52204, "epoch": 492, "lr": 5.550625190150483e-08} +{"train_loss": 0.015002220869064331, "grad_norm": 0.11615210026502609, "global_step": 52205, "epoch": 492, "lr": 5.536672711538016e-08} +{"train_loss": 0.01430806890130043, "grad_norm": 0.05874348431825638, "global_step": 52206, "epoch": 492, "lr": 5.5227377812983037e-08} +{"train_loss": 0.015922242775559425, "grad_norm": 0.0651097446680069, "global_step": 52207, "epoch": 492, "lr": 5.5088203994796414e-08} +{"train_loss": 0.015536203980445862, "grad_norm": 0.09401129186153412, "global_step": 52208, "epoch": 492, "lr": 5.494920566130879e-08} +{"train_loss": 0.014733275398612022, "grad_norm": 0.06707978248596191, "global_step": 52209, "epoch": 492, "lr": 5.48103828130031e-08} +{"train_loss": 0.01816590316593647, "grad_norm": 0.16524848341941833, "global_step": 52210, "epoch": 492, "lr": 5.467173545037896e-08} +{"train_loss": 0.014594687148928642, "grad_norm": 0.06808382272720337, "global_step": 52211, "epoch": 492, "lr": 5.45332635739193e-08} +{"train_loss": 0.014973469078540802, "grad_norm": 0.07425782829523087, "global_step": 52212, "epoch": 492, "lr": 5.439496718410153e-08} +{"train_loss": 0.014663798734545708, "grad_norm": 0.0719841793179512, "global_step": 52213, "epoch": 492, "lr": 5.4256846281425245e-08} +{"train_loss": 0.014363521710038185, "grad_norm": 0.06953418254852295, "global_step": 52214, "epoch": 492, "lr": 5.411890086637339e-08} +{"train_loss": 0.016322176903486252, "grad_norm": 0.08421266078948975, "global_step": 52215, "epoch": 492, "lr": 5.398113093941781e-08} +{"train_loss": 0.01469404250383377, "grad_norm": 0.10938581079244614, "global_step": 52216, "epoch": 492, "lr": 5.384353650105811e-08} +{"train_loss": 0.01431947946548462, "grad_norm": 0.08724967390298843, "global_step": 52217, "epoch": 492, "lr": 5.3706117551766135e-08} +{"train_loss": 0.014332696795463562, "grad_norm": 0.09335167706012726, "global_step": 52218, "epoch": 492, "lr": 5.3568874092030376e-08} +{"train_loss": 0.013624711893498898, "grad_norm": 0.06601931154727936, "global_step": 52219, "epoch": 492, "lr": 5.343180612233378e-08} +{"train_loss": 0.012505477294325829, "grad_norm": 0.10875139385461807, "global_step": 52220, "epoch": 492, "lr": 5.329491364315931e-08} +{"train_loss": 0.012821954675018787, "grad_norm": 0.1008228287100792, "global_step": 52221, "epoch": 492, "lr": 5.3158196654984335e-08} +{"train_loss": 0.012314913794398308, "grad_norm": 0.07454163581132889, "global_step": 52222, "epoch": 492, "lr": 5.3021655158286274e-08} +{"train_loss": 0.015495764091610909, "grad_norm": 0.08458425849676132, "global_step": 52223, "epoch": 492, "lr": 5.2885289153553616e-08} +{"train_loss": 0.015767155215144157, "grad_norm": 0.169404998421669, "global_step": 52224, "epoch": 492, "lr": 5.274909864125821e-08} +{"train_loss": 0.015345394611358643, "grad_norm": 0.06421520560979843, "global_step": 52225, "epoch": 492, "lr": 5.2613083621883e-08} +{"train_loss": 0.013795936480164528, "grad_norm": 0.053586702793836594, "global_step": 52226, "epoch": 492, "lr": 5.2477244095899825e-08} +{"train_loss": 0.016543282195925713, "grad_norm": 0.1185273677110672, "global_step": 52227, "epoch": 492, "lr": 5.234158006379164e-08} +{"train_loss": 0.014620617963373661, "grad_norm": 0.06202484294772148, "global_step": 52228, "epoch": 492, "lr": 5.220609152603584e-08} +{"train_loss": 0.014063352718949318, "grad_norm": 0.09499563276767731, "global_step": 52229, "epoch": 492, "lr": 5.207077848310427e-08} +{"train_loss": 0.018019355833530426, "grad_norm": 0.07956632226705551, "global_step": 52230, "epoch": 492, "lr": 5.193564093547432e-08} +{"train_loss": 0.012420738115906715, "grad_norm": 0.09028591215610504, "global_step": 52231, "epoch": 492, "lr": 5.1800678883623386e-08} +{"train_loss": 0.013567756861448288, "grad_norm": 0.08468227833509445, "global_step": 52232, "epoch": 492, "lr": 5.166589232801777e-08} +{"train_loss": 0.015725187957286835, "grad_norm": 0.06555705517530441, "global_step": 52233, "epoch": 492, "lr": 5.153128126914042e-08} +{"train_loss": 0.013829968869686127, "grad_norm": 0.06531643867492676, "global_step": 52234, "epoch": 492, "lr": 5.139684570745762e-08} +{"train_loss": 0.01743161305785179, "grad_norm": 0.09911644458770752, "global_step": 52235, "epoch": 492, "lr": 5.126258564344677e-08} +{"train_loss": 0.014382291585206985, "grad_norm": 0.07539286464452744, "global_step": 52236, "epoch": 492, "lr": 5.1128501077574166e-08} +{"train_loss": 0.015289661474525928, "grad_norm": 0.11236359179019928, "global_step": 52237, "epoch": 492, "lr": 5.09945920103172e-08} +{"train_loss": 0.016750503331422806, "grad_norm": 0.06930506229400635, "global_step": 52238, "epoch": 492, "lr": 5.0860858442136616e-08} +{"train_loss": 0.01637098379433155, "grad_norm": 0.09176448732614517, "global_step": 52239, "epoch": 492, "lr": 5.072730037351536e-08} +{"train_loss": 0.01604483276605606, "grad_norm": 0.07456421852111816, "global_step": 52240, "epoch": 492, "lr": 5.059391780491418e-08} +{"train_loss": 0.016442399471998215, "grad_norm": 0.05579940602183342, "global_step": 52241, "epoch": 492, "lr": 5.046071073679936e-08} +{"train_loss": 0.01852709800004959, "grad_norm": 0.12573722004890442, "global_step": 52242, "epoch": 492, "lr": 5.032767916964831e-08} +{"train_loss": 0.015330808237195015, "grad_norm": 0.08012799173593521, "global_step": 52243, "epoch": 492, "lr": 5.0194823103921765e-08} +{"train_loss": 0.016850803047418594, "grad_norm": 0.12793752551078796, "global_step": 52244, "epoch": 492, "lr": 5.006214254008601e-08} +{"train_loss": 0.014974603429436684, "grad_norm": 0.07696866244077682, "global_step": 52245, "epoch": 492, "lr": 4.99296374786129e-08} +{"train_loss": 0.01646900177001953, "grad_norm": 0.08321021497249603, "global_step": 52246, "epoch": 492, "lr": 4.979730791996317e-08} +{"train_loss": 0.014816176146268845, "grad_norm": 0.10579400509595871, "global_step": 52247, "epoch": 492, "lr": 4.966515386460313e-08} +{"train_loss": 0.015982946380972862, "grad_norm": 0.08189339190721512, "global_step": 52248, "epoch": 492, "lr": 4.953317531299351e-08} +{"train_loss": 0.014253266155719757, "grad_norm": 0.06430333852767944, "global_step": 52249, "epoch": 492, "lr": 4.940137226560615e-08} +{"train_loss": 0.014480959624052048, "grad_norm": 0.07996688783168793, "global_step": 52250, "epoch": 492, "lr": 4.926974472289625e-08} +{"train_loss": 0.01265038549900055, "grad_norm": 0.050349362194538116, "global_step": 52251, "epoch": 492, "lr": 4.91382926853301e-08} +{"train_loss": 0.015471130609512329, "grad_norm": 0.11081540584564209, "global_step": 52252, "epoch": 492, "lr": 4.9007016153368446e-08} +{"train_loss": 0.01592857949435711, "grad_norm": 0.09357963502407074, "global_step": 52253, "epoch": 492, "lr": 4.887591512747758e-08} +{"train_loss": 0.015894388779997826, "grad_norm": 0.0663001611828804, "global_step": 52254, "epoch": 492, "lr": 4.874498960810714e-08} +{"train_loss": 0.015703871846199036, "grad_norm": 0.10721784085035324, "global_step": 52255, "epoch": 492, "lr": 4.8614239595728973e-08} +{"train_loss": 0.018022701144218445, "grad_norm": 0.12306901067495346, "global_step": 52256, "epoch": 492, "lr": 4.8483665090792715e-08} +{"train_loss": 0.014947542788917726, "grad_norm": 0.09184923768043518, "global_step": 52257, "epoch": 492, "lr": 4.835326609376467e-08, "val_loss": 0.01958608627319336} +{"train_loss": 0.014668554067611694, "grad_norm": 0.09833203256130219, "global_step": 52258, "epoch": 493, "lr": 4.822304260510002e-08} +{"train_loss": 0.01656584069132805, "grad_norm": 0.06748104095458984, "global_step": 52259, "epoch": 493, "lr": 4.8092994625253964e-08} +{"train_loss": 0.020177438855171204, "grad_norm": 0.10034886747598648, "global_step": 52260, "epoch": 493, "lr": 4.7963122154681686e-08} +{"train_loss": 0.014388903975486755, "grad_norm": 0.12968049943447113, "global_step": 52261, "epoch": 493, "lr": 4.783342519384948e-08} +{"train_loss": 0.012977206148207188, "grad_norm": 0.06562144309282303, "global_step": 52262, "epoch": 493, "lr": 4.770390374320699e-08} +{"train_loss": 0.013325473293662071, "grad_norm": 0.07823219895362854, "global_step": 52263, "epoch": 493, "lr": 4.757455780320941e-08} +{"train_loss": 0.014304440468549728, "grad_norm": 0.059783197939395905, "global_step": 52264, "epoch": 493, "lr": 4.7445387374306374e-08} +{"train_loss": 0.015850186347961426, "grad_norm": 0.1095777377486229, "global_step": 52265, "epoch": 493, "lr": 4.731639245695862e-08} +{"train_loss": 0.017176635563373566, "grad_norm": 0.13992491364479065, "global_step": 52266, "epoch": 493, "lr": 4.7187573051621356e-08} +{"train_loss": 0.015685509890317917, "grad_norm": 0.08870436996221542, "global_step": 52267, "epoch": 493, "lr": 4.7058929158738665e-08} +{"train_loss": 0.014553194865584373, "grad_norm": 0.07172248512506485, "global_step": 52268, "epoch": 493, "lr": 4.693046077876573e-08} +{"train_loss": 0.015543299727141857, "grad_norm": 0.18162919580936432, "global_step": 52269, "epoch": 493, "lr": 4.68021679121633e-08} +{"train_loss": 0.015382302924990654, "grad_norm": 0.11389324814081192, "global_step": 52270, "epoch": 493, "lr": 4.667405055936436e-08} +{"train_loss": 0.012218553572893143, "grad_norm": 0.10780085623264313, "global_step": 52271, "epoch": 493, "lr": 4.65461087208352e-08} +{"train_loss": 0.015188915655016899, "grad_norm": 0.055700164288282394, "global_step": 52272, "epoch": 493, "lr": 4.641834239701437e-08} +{"train_loss": 0.016315143555402756, "grad_norm": 0.1524215191602707, "global_step": 52273, "epoch": 493, "lr": 4.629075158835705e-08} +{"train_loss": 0.013687552884221077, "grad_norm": 0.08718904852867126, "global_step": 52274, "epoch": 493, "lr": 4.616333629530733e-08} +{"train_loss": 0.013443268835544586, "grad_norm": 0.11041416227817535, "global_step": 52275, "epoch": 493, "lr": 4.603609651832042e-08} +{"train_loss": 0.015270611271262169, "grad_norm": 0.07985465228557587, "global_step": 52276, "epoch": 493, "lr": 4.5909032257834825e-08} +{"train_loss": 0.0145790446549654, "grad_norm": 0.0895366296172142, "global_step": 52277, "epoch": 493, "lr": 4.5782143514300216e-08} +{"train_loss": 0.016274984925985336, "grad_norm": 0.0681697204709053, "global_step": 52278, "epoch": 493, "lr": 4.565543028816621e-08} +{"train_loss": 0.01342798676341772, "grad_norm": 0.057797256857156754, "global_step": 52279, "epoch": 493, "lr": 4.552889257987136e-08} +{"train_loss": 0.014160921797156334, "grad_norm": 0.09288940578699112, "global_step": 52280, "epoch": 493, "lr": 4.5402530389865306e-08} +{"train_loss": 0.014147482812404633, "grad_norm": 0.07428062707185745, "global_step": 52281, "epoch": 493, "lr": 4.5276343718586576e-08} +{"train_loss": 0.013492768630385399, "grad_norm": 0.06462644040584564, "global_step": 52282, "epoch": 493, "lr": 4.515033256648482e-08} +{"train_loss": 0.013691870495676994, "grad_norm": 0.1149408370256424, "global_step": 52283, "epoch": 493, "lr": 4.5024496933998574e-08} +{"train_loss": 0.015138767659664154, "grad_norm": 0.08091338723897934, "global_step": 52284, "epoch": 493, "lr": 4.489883682157192e-08} +{"train_loss": 0.01536781806498766, "grad_norm": 0.06623244285583496, "global_step": 52285, "epoch": 493, "lr": 4.4773352229643403e-08} +{"train_loss": 0.01427144929766655, "grad_norm": 0.07276289910078049, "global_step": 52286, "epoch": 493, "lr": 4.4648043158662666e-08} +{"train_loss": 0.014676136896014214, "grad_norm": 0.09210366755723953, "global_step": 52287, "epoch": 493, "lr": 4.452290960905714e-08} +{"train_loss": 0.015570659190416336, "grad_norm": 0.11515108495950699, "global_step": 52288, "epoch": 493, "lr": 4.4397951581276465e-08} +{"train_loss": 0.0139781404286623, "grad_norm": 0.06638345867395401, "global_step": 52289, "epoch": 493, "lr": 4.427316907575918e-08} +{"train_loss": 0.01623884029686451, "grad_norm": 0.06631326675415039, "global_step": 52290, "epoch": 493, "lr": 4.414856209293272e-08} +{"train_loss": 0.0177436713129282, "grad_norm": 0.11753788590431213, "global_step": 52291, "epoch": 493, "lr": 4.402413063325228e-08} +{"train_loss": 0.016953088343143463, "grad_norm": 0.10375824570655823, "global_step": 52292, "epoch": 493, "lr": 4.389987469713974e-08} +{"train_loss": 0.013882098719477654, "grad_norm": 0.0974179208278656, "global_step": 52293, "epoch": 493, "lr": 4.3775794285039195e-08} +{"train_loss": 0.013386142440140247, "grad_norm": 0.05901213362812996, "global_step": 52294, "epoch": 493, "lr": 4.365188939738363e-08} +{"train_loss": 0.016985461115837097, "grad_norm": 0.08541948348283768, "global_step": 52295, "epoch": 493, "lr": 4.3528160034611575e-08} +{"train_loss": 0.015564936213195324, "grad_norm": 0.1054932177066803, "global_step": 52296, "epoch": 493, "lr": 4.340460619715603e-08} +{"train_loss": 0.01454631332308054, "grad_norm": 0.08337438106536865, "global_step": 52297, "epoch": 493, "lr": 4.328122788544997e-08} +{"train_loss": 0.013645853847265244, "grad_norm": 0.07313385605812073, "global_step": 52298, "epoch": 493, "lr": 4.315802509993194e-08} +{"train_loss": 0.016248300671577454, "grad_norm": 0.0725279226899147, "global_step": 52299, "epoch": 493, "lr": 4.303499784102383e-08} +{"train_loss": 0.014328593388199806, "grad_norm": 0.10226982086896896, "global_step": 52300, "epoch": 493, "lr": 4.2912146109169717e-08} +{"train_loss": 0.013170980848371983, "grad_norm": 0.056014031171798706, "global_step": 52301, "epoch": 493, "lr": 4.278946990479704e-08} +{"train_loss": 0.014133022166788578, "grad_norm": 0.07642924040555954, "global_step": 52302, "epoch": 493, "lr": 4.266696922833324e-08} +{"train_loss": 0.015435922890901566, "grad_norm": 0.08054730296134949, "global_step": 52303, "epoch": 493, "lr": 4.254464408021685e-08} +{"train_loss": 0.016035251319408417, "grad_norm": 0.11531142890453339, "global_step": 52304, "epoch": 493, "lr": 4.2422494460869766e-08} +{"train_loss": 0.014856318943202496, "grad_norm": 0.11118219047784805, "global_step": 52305, "epoch": 493, "lr": 4.230052037072496e-08} +{"train_loss": 0.0157562717795372, "grad_norm": 0.09460313618183136, "global_step": 52306, "epoch": 493, "lr": 4.2178721810209874e-08} +{"train_loss": 0.014971114695072174, "grad_norm": 0.06644397974014282, "global_step": 52307, "epoch": 493, "lr": 4.2057098779751944e-08} +{"train_loss": 0.012852830812335014, "grad_norm": 0.06744720786809921, "global_step": 52308, "epoch": 493, "lr": 4.193565127977861e-08} +{"train_loss": 0.014729824848473072, "grad_norm": 0.055200643837451935, "global_step": 52309, "epoch": 493, "lr": 4.18143793107173e-08} +{"train_loss": 0.01563764549791813, "grad_norm": 0.07803761214017868, "global_step": 52310, "epoch": 493, "lr": 4.169328287299545e-08} +{"train_loss": 0.012250246480107307, "grad_norm": 0.08104882389307022, "global_step": 52311, "epoch": 493, "lr": 4.157236196703495e-08} +{"train_loss": 0.012765422463417053, "grad_norm": 0.07292332500219345, "global_step": 52312, "epoch": 493, "lr": 4.145161659326324e-08} +{"train_loss": 0.017658445984125137, "grad_norm": 0.13795164227485657, "global_step": 52313, "epoch": 493, "lr": 4.133104675210775e-08} +{"train_loss": 0.01588604226708412, "grad_norm": 0.08955724537372589, "global_step": 52314, "epoch": 493, "lr": 4.121065244398481e-08} +{"train_loss": 0.01509217917919159, "grad_norm": 0.07919204235076904, "global_step": 52315, "epoch": 493, "lr": 4.1090433669321857e-08} +{"train_loss": 0.01663101837038994, "grad_norm": 0.06950271874666214, "global_step": 52316, "epoch": 493, "lr": 4.097039042854078e-08} +{"train_loss": 0.017163287848234177, "grad_norm": 0.06235945597290993, "global_step": 52317, "epoch": 493, "lr": 4.0850522722063466e-08} +{"train_loss": 0.013831118121743202, "grad_norm": 0.13460321724414825, "global_step": 52318, "epoch": 493, "lr": 4.0730830550311795e-08} +{"train_loss": 0.01293411199003458, "grad_norm": 0.07365428656339645, "global_step": 52319, "epoch": 493, "lr": 4.06113139137021e-08} +{"train_loss": 0.014740467071533203, "grad_norm": 0.07672742754220963, "global_step": 52320, "epoch": 493, "lr": 4.049197281266181e-08} +{"train_loss": 0.013087288476526737, "grad_norm": 0.05760882422327995, "global_step": 52321, "epoch": 493, "lr": 4.037280724760173e-08} +{"train_loss": 0.015748389065265656, "grad_norm": 0.07181306183338165, "global_step": 52322, "epoch": 493, "lr": 4.0253817218943725e-08} +{"train_loss": 0.01396351307630539, "grad_norm": 0.11377160996198654, "global_step": 52323, "epoch": 493, "lr": 4.013500272710968e-08} +{"train_loss": 0.016905004158616066, "grad_norm": 0.12410537898540497, "global_step": 52324, "epoch": 493, "lr": 4.0016363772515944e-08} +{"train_loss": 0.015938568860292435, "grad_norm": 0.09728357940912247, "global_step": 52325, "epoch": 493, "lr": 3.989790035557328e-08} +{"train_loss": 0.015422647818922997, "grad_norm": 0.1047527864575386, "global_step": 52326, "epoch": 493, "lr": 3.9779612476709135e-08} +{"train_loss": 0.01497685257345438, "grad_norm": 0.06612184643745422, "global_step": 52327, "epoch": 493, "lr": 3.966150013632874e-08} +{"train_loss": 0.013242962770164013, "grad_norm": 0.10136538743972778, "global_step": 52328, "epoch": 493, "lr": 3.954356333484843e-08} +{"train_loss": 0.017117680981755257, "grad_norm": 0.058790236711502075, "global_step": 52329, "epoch": 493, "lr": 3.9425802072684534e-08} +{"train_loss": 0.01497576106339693, "grad_norm": 0.08437097072601318, "global_step": 52330, "epoch": 493, "lr": 3.930821635025339e-08} +{"train_loss": 0.017083153128623962, "grad_norm": 0.06698068976402283, "global_step": 52331, "epoch": 493, "lr": 3.9190806167965774e-08} +{"train_loss": 0.013475009240210056, "grad_norm": 0.09388270229101181, "global_step": 52332, "epoch": 493, "lr": 3.9073571526232475e-08} +{"train_loss": 0.015689391642808914, "grad_norm": 0.10430897772312164, "global_step": 52333, "epoch": 493, "lr": 3.8956512425469824e-08} +{"train_loss": 0.016171222552657127, "grad_norm": 0.07352641969919205, "global_step": 52334, "epoch": 493, "lr": 3.8839628866088604e-08} +{"train_loss": 0.014131870120763779, "grad_norm": 0.06993668526411057, "global_step": 52335, "epoch": 493, "lr": 3.872292084849405e-08} +{"train_loss": 0.015766602009534836, "grad_norm": 0.13586965203285217, "global_step": 52336, "epoch": 493, "lr": 3.8606388373096936e-08} +{"train_loss": 0.014103714376688004, "grad_norm": 0.0722249448299408, "global_step": 52337, "epoch": 493, "lr": 3.84900314403136e-08} +{"train_loss": 0.012941693887114525, "grad_norm": 0.07008932530879974, "global_step": 52338, "epoch": 493, "lr": 3.8373850050549274e-08} +{"train_loss": 0.013904237188398838, "grad_norm": 0.0914064273238182, "global_step": 52339, "epoch": 493, "lr": 3.8257844204203643e-08} +{"train_loss": 0.015498164109885693, "grad_norm": 0.14995023608207703, "global_step": 52340, "epoch": 493, "lr": 3.814201390169858e-08} +{"train_loss": 0.017129015177488327, "grad_norm": 0.06730739772319794, "global_step": 52341, "epoch": 493, "lr": 3.802635914343378e-08} +{"train_loss": 0.014735544100403786, "grad_norm": 0.0916697308421135, "global_step": 52342, "epoch": 493, "lr": 3.791087992982001e-08} +{"train_loss": 0.017639415338635445, "grad_norm": 0.07970909029245377, "global_step": 52343, "epoch": 493, "lr": 3.7795576261256957e-08} +{"train_loss": 0.015550723299384117, "grad_norm": 0.06699714064598083, "global_step": 52344, "epoch": 493, "lr": 3.768044813814986e-08} +{"train_loss": 0.013484170660376549, "grad_norm": 0.052620746195316315, "global_step": 52345, "epoch": 493, "lr": 3.756549556090394e-08} +{"train_loss": 0.012953009456396103, "grad_norm": 0.07723960280418396, "global_step": 52346, "epoch": 493, "lr": 3.745071852992998e-08} +{"train_loss": 0.017992716282606125, "grad_norm": 0.06974686682224274, "global_step": 52347, "epoch": 493, "lr": 3.733611704562212e-08} +{"train_loss": 0.01452118530869484, "grad_norm": 0.07112576812505722, "global_step": 52348, "epoch": 493, "lr": 3.722169110839113e-08} +{"train_loss": 0.015578515827655792, "grad_norm": 0.07431522756814957, "global_step": 52349, "epoch": 493, "lr": 3.710744071863115e-08} +{"train_loss": 0.01099121943116188, "grad_norm": 0.07045580446720123, "global_step": 52350, "epoch": 493, "lr": 3.699336587675295e-08} +{"train_loss": 0.015117356553673744, "grad_norm": 0.07933510839939117, "global_step": 52351, "epoch": 493, "lr": 3.687946658314512e-08} +{"train_loss": 0.014743493869900703, "grad_norm": 0.0696353018283844, "global_step": 52352, "epoch": 493, "lr": 3.676574283821843e-08} +{"train_loss": 0.015593986958265305, "grad_norm": 0.07524902373552322, "global_step": 52353, "epoch": 493, "lr": 3.665219464236702e-08} +{"train_loss": 0.014083890244364738, "grad_norm": 0.075954869389534, "global_step": 52354, "epoch": 493, "lr": 3.6538821995990566e-08} +{"train_loss": 0.013904422521591187, "grad_norm": 0.061376821249723434, "global_step": 52355, "epoch": 493, "lr": 3.6425624899488755e-08} +{"train_loss": 0.014940144494175911, "grad_norm": 0.11940862238407135, "global_step": 52356, "epoch": 493, "lr": 3.6312603353255704e-08} +{"train_loss": 0.014310978353023529, "grad_norm": 0.075982965528965, "global_step": 52357, "epoch": 493, "lr": 3.619975735769665e-08} +{"train_loss": 0.013894268311560154, "grad_norm": 0.0972248762845993, "global_step": 52358, "epoch": 493, "lr": 3.6087086913200175e-08} +{"train_loss": 0.014301333576440811, "grad_norm": 0.07655202597379684, "global_step": 52359, "epoch": 493, "lr": 3.5974592020165954e-08} +{"train_loss": 0.013725295662879944, "grad_norm": 0.09173925966024399, "global_step": 52360, "epoch": 493, "lr": 3.5862272678988117e-08} +{"train_loss": 0.01387072540819645, "grad_norm": 0.06490031629800797, "global_step": 52361, "epoch": 493, "lr": 3.57501288900608e-08} +{"train_loss": 0.01851021870970726, "grad_norm": 0.08612213283777237, "global_step": 52362, "epoch": 493, "lr": 3.563816065378367e-08} +{"train_loss": 0.014947657526101707, "grad_norm": 0.1962023228406906, "global_step": 52363, "epoch": 493, "lr": 3.5526367970539765e-08, "val_loss": 0.01938747614622116} +{"train_loss": 0.01662810891866684, "grad_norm": 0.07800666242837906, "global_step": 52364, "epoch": 494, "lr": 3.541475084072876e-08} +{"train_loss": 0.014119577594101429, "grad_norm": 0.10817544162273407, "global_step": 52365, "epoch": 494, "lr": 3.530330926473924e-08} +{"train_loss": 0.014579132199287415, "grad_norm": 0.08717948198318481, "global_step": 52366, "epoch": 494, "lr": 3.5192043242965325e-08} +{"train_loss": 0.01567971147596836, "grad_norm": 0.06428224593400955, "global_step": 52367, "epoch": 494, "lr": 3.5080952775795597e-08} +{"train_loss": 0.015305956825613976, "grad_norm": 0.07938280701637268, "global_step": 52368, "epoch": 494, "lr": 3.4970037863629734e-08} +{"train_loss": 0.01401941105723381, "grad_norm": 0.06582318991422653, "global_step": 52369, "epoch": 494, "lr": 3.4859298506845216e-08} +{"train_loss": 0.013316506519913673, "grad_norm": 0.07802770286798477, "global_step": 52370, "epoch": 494, "lr": 3.474873470583062e-08} +{"train_loss": 0.01429884135723114, "grad_norm": 0.09092520922422409, "global_step": 52371, "epoch": 494, "lr": 3.4638346460985625e-08} +{"train_loss": 0.015145202167332172, "grad_norm": 0.06550765782594681, "global_step": 52372, "epoch": 494, "lr": 3.452813377269326e-08} +{"train_loss": 0.013043960556387901, "grad_norm": 0.096222423017025, "global_step": 52373, "epoch": 494, "lr": 3.441809664133655e-08} +{"train_loss": 0.014474445953965187, "grad_norm": 0.057490017265081406, "global_step": 52374, "epoch": 494, "lr": 3.430823506730962e-08} +{"train_loss": 0.014455365017056465, "grad_norm": 0.0638994500041008, "global_step": 52375, "epoch": 494, "lr": 3.419854905098996e-08} +{"train_loss": 0.014871545135974884, "grad_norm": 0.056615110486745834, "global_step": 52376, "epoch": 494, "lr": 3.408903859276613e-08} +{"train_loss": 0.012603369541466236, "grad_norm": 0.049939174205064774, "global_step": 52377, "epoch": 494, "lr": 3.397970369302672e-08} +{"train_loss": 0.01361894141882658, "grad_norm": 0.11015332490205765, "global_step": 52378, "epoch": 494, "lr": 3.38705443521492e-08} +{"train_loss": 0.015216413885354996, "grad_norm": 0.07871616631746292, "global_step": 52379, "epoch": 494, "lr": 3.3761560570527704e-08} +{"train_loss": 0.014585183933377266, "grad_norm": 0.08102066814899445, "global_step": 52380, "epoch": 494, "lr": 3.3652752348534157e-08} +{"train_loss": 0.013988968916237354, "grad_norm": 0.0723872184753418, "global_step": 52381, "epoch": 494, "lr": 3.354411968655158e-08} +{"train_loss": 0.016092758625745773, "grad_norm": 0.08903511613607407, "global_step": 52382, "epoch": 494, "lr": 3.343566258496855e-08} +{"train_loss": 0.01581709459424019, "grad_norm": 0.08041024208068848, "global_step": 52383, "epoch": 494, "lr": 3.3327381044162554e-08} +{"train_loss": 0.01626507379114628, "grad_norm": 0.10048029571771622, "global_step": 52384, "epoch": 494, "lr": 3.3219275064516607e-08} +{"train_loss": 0.016886452212929726, "grad_norm": 0.07885055989027023, "global_step": 52385, "epoch": 494, "lr": 3.311134464640819e-08} +{"train_loss": 0.012932632118463516, "grad_norm": 0.14234685897827148, "global_step": 52386, "epoch": 494, "lr": 3.300358979020923e-08} +{"train_loss": 0.014405903406441212, "grad_norm": 0.1249304786324501, "global_step": 52387, "epoch": 494, "lr": 3.2896010496313855e-08} +{"train_loss": 0.01301539782434702, "grad_norm": 0.08422019332647324, "global_step": 52388, "epoch": 494, "lr": 3.2788606765088436e-08} +{"train_loss": 0.014491314068436623, "grad_norm": 0.06769811362028122, "global_step": 52389, "epoch": 494, "lr": 3.268137859691045e-08} +{"train_loss": 0.01512809470295906, "grad_norm": 0.1652131825685501, "global_step": 52390, "epoch": 494, "lr": 3.257432599216292e-08} +{"train_loss": 0.015181520022451878, "grad_norm": 0.05858495458960533, "global_step": 52391, "epoch": 494, "lr": 3.246744895122333e-08} +{"train_loss": 0.014775880612432957, "grad_norm": 0.07133562117815018, "global_step": 52392, "epoch": 494, "lr": 3.236074747445805e-08} +{"train_loss": 0.014799989759922028, "grad_norm": 0.06752332299947739, "global_step": 52393, "epoch": 494, "lr": 3.2254221562244555e-08} +{"train_loss": 0.013486592099070549, "grad_norm": 0.10804887115955353, "global_step": 52394, "epoch": 494, "lr": 3.214787121496588e-08} +{"train_loss": 0.015318270772695541, "grad_norm": 0.06736677885055542, "global_step": 52395, "epoch": 494, "lr": 3.204169643298283e-08} +{"train_loss": 0.013943782076239586, "grad_norm": 0.059041813015937805, "global_step": 52396, "epoch": 494, "lr": 3.193569721667844e-08} +{"train_loss": 0.014439575374126434, "grad_norm": 0.06672342866659164, "global_step": 52397, "epoch": 494, "lr": 3.182987356641909e-08} +{"train_loss": 0.013739680871367455, "grad_norm": 0.076807901263237, "global_step": 52398, "epoch": 494, "lr": 3.172422548258225e-08} +{"train_loss": 0.015640614554286003, "grad_norm": 0.14192792773246765, "global_step": 52399, "epoch": 494, "lr": 3.161875296553429e-08} +{"train_loss": 0.01336184423416853, "grad_norm": 0.0706968829035759, "global_step": 52400, "epoch": 494, "lr": 3.1513456015647146e-08} +{"train_loss": 0.014431895688176155, "grad_norm": 0.09945763647556305, "global_step": 52401, "epoch": 494, "lr": 3.140833463329273e-08} +{"train_loss": 0.014341101050376892, "grad_norm": 0.1285773664712906, "global_step": 52402, "epoch": 494, "lr": 3.1303388818837434e-08} +{"train_loss": 0.015011010691523552, "grad_norm": 0.0824904590845108, "global_step": 52403, "epoch": 494, "lr": 3.1198618572647607e-08} +{"train_loss": 0.014519236981868744, "grad_norm": 0.1633347123861313, "global_step": 52404, "epoch": 494, "lr": 3.109402389509519e-08} +{"train_loss": 0.01737123541533947, "grad_norm": 0.07584037631750107, "global_step": 52405, "epoch": 494, "lr": 3.098960478655211e-08} +{"train_loss": 0.01739201508462429, "grad_norm": 0.1005295142531395, "global_step": 52406, "epoch": 494, "lr": 3.0885361247379176e-08} +{"train_loss": 0.015175264328718185, "grad_norm": 0.05961019545793533, "global_step": 52407, "epoch": 494, "lr": 3.078129327794277e-08} +{"train_loss": 0.013243114575743675, "grad_norm": 0.07801708579063416, "global_step": 52408, "epoch": 494, "lr": 3.067740087860926e-08} +{"train_loss": 0.014048002660274506, "grad_norm": 0.06454386562108994, "global_step": 52409, "epoch": 494, "lr": 3.0573684049750584e-08} +{"train_loss": 0.015826966613531113, "grad_norm": 0.10824508219957352, "global_step": 52410, "epoch": 494, "lr": 3.047014279171645e-08} +{"train_loss": 0.015083425678312778, "grad_norm": 0.06859083473682404, "global_step": 52411, "epoch": 494, "lr": 3.036677710488434e-08} +{"train_loss": 0.014176261611282825, "grad_norm": 0.07444988936185837, "global_step": 52412, "epoch": 494, "lr": 3.0263586989609516e-08} +{"train_loss": 0.01658717915415764, "grad_norm": 0.07190682739019394, "global_step": 52413, "epoch": 494, "lr": 3.016057244625836e-08} +{"train_loss": 0.013632018119096756, "grad_norm": 0.09926196187734604, "global_step": 52414, "epoch": 494, "lr": 3.00577334751917e-08} +{"train_loss": 0.014725811779499054, "grad_norm": 0.10121973603963852, "global_step": 52415, "epoch": 494, "lr": 2.995507007677034e-08} +{"train_loss": 0.015369242057204247, "grad_norm": 0.07033983618021011, "global_step": 52416, "epoch": 494, "lr": 2.985258225135512e-08} +{"train_loss": 0.015525147318840027, "grad_norm": 0.07996116578578949, "global_step": 52417, "epoch": 494, "lr": 2.9750269999306857e-08} +{"train_loss": 0.014645153656601906, "grad_norm": 0.1755376160144806, "global_step": 52418, "epoch": 494, "lr": 2.9648133320986373e-08} +{"train_loss": 0.01738683320581913, "grad_norm": 0.08592089265584946, "global_step": 52419, "epoch": 494, "lr": 2.9546172216748936e-08} +{"train_loss": 0.015800101682543755, "grad_norm": 0.08566256612539291, "global_step": 52420, "epoch": 494, "lr": 2.9444386686955373e-08} +{"train_loss": 0.01571463793516159, "grad_norm": 0.08166113495826721, "global_step": 52421, "epoch": 494, "lr": 2.9342776731960952e-08} +{"train_loss": 0.016132719814777374, "grad_norm": 0.08690819889307022, "global_step": 52422, "epoch": 494, "lr": 2.92413423521265e-08} +{"train_loss": 0.011220477521419525, "grad_norm": 0.06495370715856552, "global_step": 52423, "epoch": 494, "lr": 2.9140083547807284e-08} +{"train_loss": 0.0157165564596653, "grad_norm": 0.10320240259170532, "global_step": 52424, "epoch": 494, "lr": 2.9039000319353027e-08} +{"train_loss": 0.01323024183511734, "grad_norm": 0.08933989703655243, "global_step": 52425, "epoch": 494, "lr": 2.89380926671301e-08} +{"train_loss": 0.01838294416666031, "grad_norm": 0.14459121227264404, "global_step": 52426, "epoch": 494, "lr": 2.8837360591482675e-08} +{"train_loss": 0.015382861718535423, "grad_norm": 0.07779064774513245, "global_step": 52427, "epoch": 494, "lr": 2.8736804092771573e-08} +{"train_loss": 0.014788257889449596, "grad_norm": 0.08224383741617203, "global_step": 52428, "epoch": 494, "lr": 2.8636423171340966e-08} +{"train_loss": 0.01518397219479084, "grad_norm": 0.06787962466478348, "global_step": 52429, "epoch": 494, "lr": 2.8536217827557223e-08} +{"train_loss": 0.014741484075784683, "grad_norm": 0.0899328961968422, "global_step": 52430, "epoch": 494, "lr": 2.8436188061764512e-08} +{"train_loss": 0.015162588097155094, "grad_norm": 0.10096446424722672, "global_step": 52431, "epoch": 494, "lr": 2.833633387431256e-08} +{"train_loss": 0.015294370241463184, "grad_norm": 0.07422976940870285, "global_step": 52432, "epoch": 494, "lr": 2.8236655265556634e-08} +{"train_loss": 0.015000401064753532, "grad_norm": 0.07068315893411636, "global_step": 52433, "epoch": 494, "lr": 2.8137152235840903e-08} +{"train_loss": 0.014334414154291153, "grad_norm": 0.10983826220035553, "global_step": 52434, "epoch": 494, "lr": 2.803782478552064e-08} +{"train_loss": 0.01867556944489479, "grad_norm": 0.06452526152133942, "global_step": 52435, "epoch": 494, "lr": 2.7938672914945562e-08} +{"train_loss": 0.014123933389782906, "grad_norm": 0.07265032082796097, "global_step": 52436, "epoch": 494, "lr": 2.7839696624459842e-08} +{"train_loss": 0.01578136533498764, "grad_norm": 0.12098690867424011, "global_step": 52437, "epoch": 494, "lr": 2.77408959144132e-08} +{"train_loss": 0.012034750543534756, "grad_norm": 0.07771506160497665, "global_step": 52438, "epoch": 494, "lr": 2.7642270785155354e-08} +{"train_loss": 0.016049940139055252, "grad_norm": 0.08874883502721786, "global_step": 52439, "epoch": 494, "lr": 2.7543821237030475e-08} +{"train_loss": 0.015589479357004166, "grad_norm": 0.09433355182409286, "global_step": 52440, "epoch": 494, "lr": 2.744554727038273e-08} +{"train_loss": 0.01322791539132595, "grad_norm": 0.12254948168992996, "global_step": 52441, "epoch": 494, "lr": 2.7347448885556294e-08} +{"train_loss": 0.013429689221084118, "grad_norm": 0.11953282356262207, "global_step": 52442, "epoch": 494, "lr": 2.724952608290088e-08} +{"train_loss": 0.01584862545132637, "grad_norm": 0.09258832037448883, "global_step": 52443, "epoch": 494, "lr": 2.7151778862755106e-08} +{"train_loss": 0.013448981568217278, "grad_norm": 0.05503510311245918, "global_step": 52444, "epoch": 494, "lr": 2.7054207225468698e-08} +{"train_loss": 0.017815161496400833, "grad_norm": 0.07158400118350983, "global_step": 52445, "epoch": 494, "lr": 2.695681117138027e-08} +{"train_loss": 0.015902342274785042, "grad_norm": 0.1314578503370285, "global_step": 52446, "epoch": 494, "lr": 2.685959070083399e-08} +{"train_loss": 0.014066125266253948, "grad_norm": 0.09296274185180664, "global_step": 52447, "epoch": 494, "lr": 2.6762545814168483e-08} +{"train_loss": 0.014396024867892265, "grad_norm": 0.12906935811042786, "global_step": 52448, "epoch": 494, "lr": 2.666567651172236e-08} +{"train_loss": 0.01708691567182541, "grad_norm": 0.06371129304170609, "global_step": 52449, "epoch": 494, "lr": 2.6568982793850894e-08} +{"train_loss": 0.01613098941743374, "grad_norm": 0.11120034009218216, "global_step": 52450, "epoch": 494, "lr": 2.647246466087605e-08} +{"train_loss": 0.014600005000829697, "grad_norm": 0.12105699628591537, "global_step": 52451, "epoch": 494, "lr": 2.6376122113142e-08} +{"train_loss": 0.016038846224546432, "grad_norm": 0.06673074513673782, "global_step": 52452, "epoch": 494, "lr": 2.6279955150992907e-08} +{"train_loss": 0.013787785544991493, "grad_norm": 0.08185312151908875, "global_step": 52453, "epoch": 494, "lr": 2.6183963774761842e-08} +{"train_loss": 0.01643754169344902, "grad_norm": 0.1903718262910843, "global_step": 52454, "epoch": 494, "lr": 2.6088147984787427e-08} +{"train_loss": 0.014571614563465118, "grad_norm": 0.0628945454955101, "global_step": 52455, "epoch": 494, "lr": 2.599250778140827e-08} +{"train_loss": 0.016384294256567955, "grad_norm": 0.06397616118192673, "global_step": 52456, "epoch": 494, "lr": 2.5897043164957447e-08} +{"train_loss": 0.014266700483858585, "grad_norm": 0.08456949144601822, "global_step": 52457, "epoch": 494, "lr": 2.580175413577357e-08} +{"train_loss": 0.016304094344377518, "grad_norm": 0.1695062816143036, "global_step": 52458, "epoch": 494, "lr": 2.570664069418416e-08} +{"train_loss": 0.013878711499273777, "grad_norm": 0.1029282733798027, "global_step": 52459, "epoch": 494, "lr": 2.561170284052783e-08} +{"train_loss": 0.0130497757345438, "grad_norm": 0.06709150224924088, "global_step": 52460, "epoch": 494, "lr": 2.55169405751432e-08} +{"train_loss": 0.015577908605337143, "grad_norm": 0.0578964501619339, "global_step": 52461, "epoch": 494, "lr": 2.5422353898357788e-08} +{"train_loss": 0.014925580471754074, "grad_norm": 0.06984472274780273, "global_step": 52462, "epoch": 494, "lr": 2.5327942810504656e-08} +{"train_loss": 0.015063196420669556, "grad_norm": 0.06604712456464767, "global_step": 52463, "epoch": 494, "lr": 2.5233707311911325e-08} +{"train_loss": 0.015174264088273048, "grad_norm": 0.05701738968491554, "global_step": 52464, "epoch": 494, "lr": 2.513964740292196e-08} +{"train_loss": 0.015462055802345276, "grad_norm": 0.10257736593484879, "global_step": 52465, "epoch": 494, "lr": 2.504576308385298e-08} +{"train_loss": 0.01454296987503767, "grad_norm": 0.07967682182788849, "global_step": 52466, "epoch": 494, "lr": 2.4952054355037448e-08} +{"train_loss": 0.015995565801858902, "grad_norm": 0.09099143743515015, "global_step": 52467, "epoch": 494, "lr": 2.4858521216808427e-08} +{"train_loss": 0.014724042266607285, "grad_norm": 0.06684475392103195, "global_step": 52468, "epoch": 494, "lr": 2.476516366949344e-08} +{"train_loss": 0.014951587768869018, "grad_norm": 0.085132896900177, "global_step": 52469, "epoch": 494, "lr": 2.467198171342e-08, "val_loss": 0.01954340562224388} +{"train_loss": 0.01498293224722147, "grad_norm": 0.088116355240345, "global_step": 52470, "epoch": 495, "lr": 2.4578975348921174e-08} +{"train_loss": 0.015389248728752136, "grad_norm": 0.09063980728387833, "global_step": 52471, "epoch": 495, "lr": 2.4486144576307825e-08} +{"train_loss": 0.014840992167592049, "grad_norm": 0.12986811995506287, "global_step": 52472, "epoch": 495, "lr": 2.439348939592412e-08} +{"train_loss": 0.013051236048340797, "grad_norm": 0.06436058133840561, "global_step": 52473, "epoch": 495, "lr": 2.430100980808647e-08} +{"train_loss": 0.01690332591533661, "grad_norm": 0.11633225530385971, "global_step": 52474, "epoch": 495, "lr": 2.420870581311685e-08} +{"train_loss": 0.013001663610339165, "grad_norm": 0.07851444184780121, "global_step": 52475, "epoch": 495, "lr": 2.4116577411348317e-08} +{"train_loss": 0.013582437299191952, "grad_norm": 0.06466653943061829, "global_step": 52476, "epoch": 495, "lr": 2.402462460309729e-08} +{"train_loss": 0.014778103679418564, "grad_norm": 0.08035251498222351, "global_step": 52477, "epoch": 495, "lr": 2.393284738869128e-08} +{"train_loss": 0.014220508746802807, "grad_norm": 0.09799407422542572, "global_step": 52478, "epoch": 495, "lr": 2.384124576845226e-08} +{"train_loss": 0.015071464702486992, "grad_norm": 0.0640501007437706, "global_step": 52479, "epoch": 495, "lr": 2.3749819742702185e-08} +{"train_loss": 0.017269019037485123, "grad_norm": 0.07320605218410492, "global_step": 52480, "epoch": 495, "lr": 2.3658569311757473e-08} +{"train_loss": 0.012695906683802605, "grad_norm": 0.05909081920981407, "global_step": 52481, "epoch": 495, "lr": 2.356749447594564e-08} +{"train_loss": 0.013566941022872925, "grad_norm": 0.05843553692102432, "global_step": 52482, "epoch": 495, "lr": 2.34765952355831e-08} +{"train_loss": 0.017020121216773987, "grad_norm": 0.08254318684339523, "global_step": 52483, "epoch": 495, "lr": 2.3385871590986263e-08} +{"train_loss": 0.013930637389421463, "grad_norm": 0.06159244105219841, "global_step": 52484, "epoch": 495, "lr": 2.3295323542477098e-08} +{"train_loss": 0.015111291781067848, "grad_norm": 0.12009240686893463, "global_step": 52485, "epoch": 495, "lr": 2.3204951090377568e-08} +{"train_loss": 0.015500463545322418, "grad_norm": 0.06096542254090309, "global_step": 52486, "epoch": 495, "lr": 2.3114754235004086e-08} +{"train_loss": 0.01253508310765028, "grad_norm": 0.06117359548807144, "global_step": 52487, "epoch": 495, "lr": 2.3024732976661966e-08} +{"train_loss": 0.016671236604452133, "grad_norm": 0.10179456323385239, "global_step": 52488, "epoch": 495, "lr": 2.2934887315684273e-08} +{"train_loss": 0.016167398542165756, "grad_norm": 0.08879013359546661, "global_step": 52489, "epoch": 495, "lr": 2.2845217252376316e-08} +{"train_loss": 0.012092551216483116, "grad_norm": 0.11929760873317719, "global_step": 52490, "epoch": 495, "lr": 2.275572278705451e-08} +{"train_loss": 0.016880860552191734, "grad_norm": 0.09656843543052673, "global_step": 52491, "epoch": 495, "lr": 2.266640392003527e-08} +{"train_loss": 0.013990829698741436, "grad_norm": 0.0651312842965126, "global_step": 52492, "epoch": 495, "lr": 2.2577260651635014e-08} +{"train_loss": 0.015398233197629452, "grad_norm": 0.07958410680294037, "global_step": 52493, "epoch": 495, "lr": 2.2488292982159044e-08} +{"train_loss": 0.01662871614098549, "grad_norm": 0.08004196733236313, "global_step": 52494, "epoch": 495, "lr": 2.2399500911923776e-08} +{"train_loss": 0.014229841530323029, "grad_norm": 0.06933791190385818, "global_step": 52495, "epoch": 495, "lr": 2.2310884441245627e-08} +{"train_loss": 0.013880506157875061, "grad_norm": 0.12239384651184082, "global_step": 52496, "epoch": 495, "lr": 2.2222443570435458e-08} +{"train_loss": 0.012924771755933762, "grad_norm": 0.05984494090080261, "global_step": 52497, "epoch": 495, "lr": 2.213417829979303e-08} +{"train_loss": 0.013496835716068745, "grad_norm": 0.0721425861120224, "global_step": 52498, "epoch": 495, "lr": 2.2046088629640306e-08} +{"train_loss": 0.014107613824307919, "grad_norm": 0.06634514033794403, "global_step": 52499, "epoch": 495, "lr": 2.1958174560282595e-08} +{"train_loss": 0.015749163925647736, "grad_norm": 0.07218301296234131, "global_step": 52500, "epoch": 495, "lr": 2.1870436092030765e-08} +{"train_loss": 0.016623951494693756, "grad_norm": 0.07718860357999802, "global_step": 52501, "epoch": 495, "lr": 2.1782873225184576e-08} +{"train_loss": 0.015341022983193398, "grad_norm": 0.11062155663967133, "global_step": 52502, "epoch": 495, "lr": 2.1695485960065986e-08} +{"train_loss": 0.016789104789495468, "grad_norm": 0.07214881479740143, "global_step": 52503, "epoch": 495, "lr": 2.160827429696921e-08} +{"train_loss": 0.016935579478740692, "grad_norm": 0.09040424227714539, "global_step": 52504, "epoch": 495, "lr": 2.1521238236210663e-08} +{"train_loss": 0.01259811781346798, "grad_norm": 0.10082808881998062, "global_step": 52505, "epoch": 495, "lr": 2.1434377778090096e-08} +{"train_loss": 0.014107544906437397, "grad_norm": 0.07983026653528214, "global_step": 52506, "epoch": 495, "lr": 2.1347692922912834e-08} +{"train_loss": 0.015512178651988506, "grad_norm": 0.13555559515953064, "global_step": 52507, "epoch": 495, "lr": 2.1261183670984177e-08} +{"train_loss": 0.018308494240045547, "grad_norm": 0.0796947106719017, "global_step": 52508, "epoch": 495, "lr": 2.1174850022609438e-08} +{"train_loss": 0.01497515570372343, "grad_norm": 0.058856915682554245, "global_step": 52509, "epoch": 495, "lr": 2.108869197808838e-08} +{"train_loss": 0.011923126876354218, "grad_norm": 0.061635665595531464, "global_step": 52510, "epoch": 495, "lr": 2.1002709537731868e-08} +{"train_loss": 0.015560991130769253, "grad_norm": 0.06323522329330444, "global_step": 52511, "epoch": 495, "lr": 2.0916902701834106e-08} +{"train_loss": 0.015342837199568748, "grad_norm": 0.11442746222019196, "global_step": 52512, "epoch": 495, "lr": 2.083127147069486e-08} +{"train_loss": 0.014519797638058662, "grad_norm": 0.07252629846334457, "global_step": 52513, "epoch": 495, "lr": 2.074581584462498e-08} +{"train_loss": 0.014895528554916382, "grad_norm": 0.06862524896860123, "global_step": 52514, "epoch": 495, "lr": 2.066053582391869e-08} +{"train_loss": 0.014915323816239834, "grad_norm": 0.0797441229224205, "global_step": 52515, "epoch": 495, "lr": 2.0575431408870194e-08} +{"train_loss": 0.01403868105262518, "grad_norm": 0.11966894567012787, "global_step": 52516, "epoch": 495, "lr": 2.049050259979035e-08} +{"train_loss": 0.01337966788560152, "grad_norm": 0.1452798992395401, "global_step": 52517, "epoch": 495, "lr": 2.040574939696782e-08} +{"train_loss": 0.015087227337062359, "grad_norm": 0.08634081482887268, "global_step": 52518, "epoch": 495, "lr": 2.0321171800702365e-08} +{"train_loss": 0.014987746253609657, "grad_norm": 0.07772234082221985, "global_step": 52519, "epoch": 495, "lr": 2.0236769811299294e-08} +{"train_loss": 0.013967124745249748, "grad_norm": 0.07378894090652466, "global_step": 52520, "epoch": 495, "lr": 2.0152543429041714e-08} +{"train_loss": 0.014746008440852165, "grad_norm": 0.07054417580366135, "global_step": 52521, "epoch": 495, "lr": 2.0068492654234937e-08} +{"train_loss": 0.013777133077383041, "grad_norm": 0.07524711638689041, "global_step": 52522, "epoch": 495, "lr": 1.9984617487173174e-08} +{"train_loss": 0.012042462825775146, "grad_norm": 0.0783567950129509, "global_step": 52523, "epoch": 495, "lr": 1.990091792815063e-08} +{"train_loss": 0.015676287934184074, "grad_norm": 0.07681762427091599, "global_step": 52524, "epoch": 495, "lr": 1.9817393977455968e-08} +{"train_loss": 0.014812910929322243, "grad_norm": 0.13640262186527252, "global_step": 52525, "epoch": 495, "lr": 1.9734045635388942e-08} +{"train_loss": 0.014462849125266075, "grad_norm": 0.0519406832754612, "global_step": 52526, "epoch": 495, "lr": 1.9650872902243766e-08} +{"train_loss": 0.014253336936235428, "grad_norm": 0.10866156220436096, "global_step": 52527, "epoch": 495, "lr": 1.9567875778303546e-08} +{"train_loss": 0.01549367792904377, "grad_norm": 0.11827044934034348, "global_step": 52528, "epoch": 495, "lr": 1.9485054263873593e-08} +{"train_loss": 0.013517786748707294, "grad_norm": 0.12782149016857147, "global_step": 52529, "epoch": 495, "lr": 1.9402408359231462e-08} +{"train_loss": 0.01574387028813362, "grad_norm": 0.0749659538269043, "global_step": 52530, "epoch": 495, "lr": 1.9319938064676912e-08} +{"train_loss": 0.017572596669197083, "grad_norm": 0.07159367948770523, "global_step": 52531, "epoch": 495, "lr": 1.9237643380493055e-08} +{"train_loss": 0.014805443584918976, "grad_norm": 0.08433051407337189, "global_step": 52532, "epoch": 495, "lr": 1.9155524306974094e-08} +{"train_loss": 0.015229066833853722, "grad_norm": 0.12130037695169449, "global_step": 52533, "epoch": 495, "lr": 1.907358084440314e-08} +{"train_loss": 0.01454569399356842, "grad_norm": 0.08231807500123978, "global_step": 52534, "epoch": 495, "lr": 1.89918129930744e-08} +{"train_loss": 0.013620548881590366, "grad_norm": 0.0706355944275856, "global_step": 52535, "epoch": 495, "lr": 1.891022075327098e-08} +{"train_loss": 0.01325956080108881, "grad_norm": 0.09109286218881607, "global_step": 52536, "epoch": 495, "lr": 1.882880412528154e-08} +{"train_loss": 0.016351930797100067, "grad_norm": 0.07020927965641022, "global_step": 52537, "epoch": 495, "lr": 1.874756310939474e-08} +{"train_loss": 0.015281230211257935, "grad_norm": 0.06003662571310997, "global_step": 52538, "epoch": 495, "lr": 1.866649770588813e-08} +{"train_loss": 0.013853652402758598, "grad_norm": 0.06922329217195511, "global_step": 52539, "epoch": 495, "lr": 1.8585607915050373e-08} +{"train_loss": 0.015376072376966476, "grad_norm": 0.0804503932595253, "global_step": 52540, "epoch": 495, "lr": 1.8504893737170125e-08} +{"train_loss": 0.014905109070241451, "grad_norm": 0.081448033452034, "global_step": 52541, "epoch": 495, "lr": 1.8424355172524942e-08} +{"train_loss": 0.015190728940069675, "grad_norm": 0.07485496252775192, "global_step": 52542, "epoch": 495, "lr": 1.8343992221403483e-08} +{"train_loss": 0.016745049506425858, "grad_norm": 0.10290665924549103, "global_step": 52543, "epoch": 495, "lr": 1.82638048840833e-08} +{"train_loss": 0.015938853845000267, "grad_norm": 0.14253166317939758, "global_step": 52544, "epoch": 495, "lr": 1.8183793160841954e-08} +{"train_loss": 0.014405773021280766, "grad_norm": 0.10164079070091248, "global_step": 52545, "epoch": 495, "lr": 1.810395705197365e-08} +{"train_loss": 0.014556784182786942, "grad_norm": 0.07770024240016937, "global_step": 52546, "epoch": 495, "lr": 1.80242965577504e-08} +{"train_loss": 0.017714280635118484, "grad_norm": 0.08049997687339783, "global_step": 52547, "epoch": 495, "lr": 1.7944811678449745e-08} +{"train_loss": 0.01761254295706749, "grad_norm": 0.08573560416698456, "global_step": 52548, "epoch": 495, "lr": 1.7865502414354806e-08} +{"train_loss": 0.015481850132346153, "grad_norm": 0.08593107014894485, "global_step": 52549, "epoch": 495, "lr": 1.7786368765748684e-08} +{"train_loss": 0.012755462899804115, "grad_norm": 0.07359243184328079, "global_step": 52550, "epoch": 495, "lr": 1.7707410732897834e-08} +{"train_loss": 0.013427482917904854, "grad_norm": 0.060293979942798615, "global_step": 52551, "epoch": 495, "lr": 1.762862831609091e-08} +{"train_loss": 0.015514761209487915, "grad_norm": 0.07406576722860336, "global_step": 52552, "epoch": 495, "lr": 1.7550021515599925e-08} +{"train_loss": 0.014733536168932915, "grad_norm": 0.08482930809259415, "global_step": 52553, "epoch": 495, "lr": 1.7471590331702424e-08} +{"train_loss": 0.013980627991259098, "grad_norm": 0.06660273671150208, "global_step": 52554, "epoch": 495, "lr": 1.739333476467042e-08} +{"train_loss": 0.014833500608801842, "grad_norm": 0.06951161473989487, "global_step": 52555, "epoch": 495, "lr": 1.7315254814781468e-08} +{"train_loss": 0.013851388357579708, "grad_norm": 0.06234241649508476, "global_step": 52556, "epoch": 495, "lr": 1.7237350482307568e-08} +{"train_loss": 0.01522668357938528, "grad_norm": 0.1008487194776535, "global_step": 52557, "epoch": 495, "lr": 1.715962176752628e-08} +{"train_loss": 0.01520952396094799, "grad_norm": 0.0716405138373375, "global_step": 52558, "epoch": 495, "lr": 1.7082068670715157e-08} +{"train_loss": 0.016315406188368797, "grad_norm": 0.09583684056997299, "global_step": 52559, "epoch": 495, "lr": 1.7004691192135104e-08} +{"train_loss": 0.014452550560235977, "grad_norm": 0.08489671349525452, "global_step": 52560, "epoch": 495, "lr": 1.6927489332063673e-08} +{"train_loss": 0.013782900758087635, "grad_norm": 0.07157821953296661, "global_step": 52561, "epoch": 495, "lr": 1.6850463090772872e-08} +{"train_loss": 0.015383010730147362, "grad_norm": 0.0804561972618103, "global_step": 52562, "epoch": 495, "lr": 1.6773612468529155e-08} +{"train_loss": 0.01614009402692318, "grad_norm": 0.10110285878181458, "global_step": 52563, "epoch": 495, "lr": 1.6696937465610074e-08} +{"train_loss": 0.013770174235105515, "grad_norm": 0.07965648174285889, "global_step": 52564, "epoch": 495, "lr": 1.6620438082276535e-08} +{"train_loss": 0.017209235578775406, "grad_norm": 0.16505593061447144, "global_step": 52565, "epoch": 495, "lr": 1.654411431880609e-08} +{"train_loss": 0.017368502914905548, "grad_norm": 0.1422847956418991, "global_step": 52566, "epoch": 495, "lr": 1.6467966175459647e-08} +{"train_loss": 0.015139959752559662, "grad_norm": 0.11658578366041183, "global_step": 52567, "epoch": 495, "lr": 1.6391993652509207e-08} +{"train_loss": 0.017319202423095703, "grad_norm": 0.08367320895195007, "global_step": 52568, "epoch": 495, "lr": 1.6316196750221225e-08} +{"train_loss": 0.015004144981503487, "grad_norm": 0.14371122419834137, "global_step": 52569, "epoch": 495, "lr": 1.624057546886215e-08} +{"train_loss": 0.01730785146355629, "grad_norm": 0.09018834680318832, "global_step": 52570, "epoch": 495, "lr": 1.6165129808692892e-08} +{"train_loss": 0.014912128448486328, "grad_norm": 0.0638248473405838, "global_step": 52571, "epoch": 495, "lr": 1.6089859769979897e-08} +{"train_loss": 0.01522815227508545, "grad_norm": 0.12378887832164764, "global_step": 52572, "epoch": 495, "lr": 1.6014765352989625e-08} +{"train_loss": 0.014388611540198326, "grad_norm": 0.09350002557039261, "global_step": 52573, "epoch": 495, "lr": 1.5939846557988526e-08} +{"train_loss": 0.012130720540881157, "grad_norm": 0.09845808148384094, "global_step": 52574, "epoch": 495, "lr": 1.58651033852375e-08} +{"train_loss": 0.01495661486361949, "grad_norm": 0.10811545699834824, "global_step": 52575, "epoch": 495, "lr": 1.5790535835003008e-08, "val_loss": 0.019427547231316566} +{"train_loss": 0.01263231411576271, "grad_norm": 0.09075150638818741, "global_step": 52576, "epoch": 496, "lr": 1.5716143907534842e-08} +{"train_loss": 0.015219890512526035, "grad_norm": 0.05773797631263733, "global_step": 52577, "epoch": 496, "lr": 1.5641927603110562e-08} +{"train_loss": 0.01416282169520855, "grad_norm": 0.06694725900888443, "global_step": 52578, "epoch": 496, "lr": 1.5567886921974416e-08} +{"train_loss": 0.013444628566503525, "grad_norm": 0.06614166498184204, "global_step": 52579, "epoch": 496, "lr": 1.549402186440396e-08} +{"train_loss": 0.015334799885749817, "grad_norm": 0.15137672424316406, "global_step": 52580, "epoch": 496, "lr": 1.5420332430643447e-08} +{"train_loss": 0.01660880446434021, "grad_norm": 0.09540007263422012, "global_step": 52581, "epoch": 496, "lr": 1.5346818620959324e-08} +{"train_loss": 0.015535522252321243, "grad_norm": 0.07672998309135437, "global_step": 52582, "epoch": 496, "lr": 1.5273480435612496e-08} +{"train_loss": 0.014492141082882881, "grad_norm": 0.09664494544267654, "global_step": 52583, "epoch": 496, "lr": 1.5200317874852764e-08} +{"train_loss": 0.013898014090955257, "grad_norm": 0.06260393559932709, "global_step": 52584, "epoch": 496, "lr": 1.512733093894103e-08} +{"train_loss": 0.015075810253620148, "grad_norm": 0.06123970076441765, "global_step": 52585, "epoch": 496, "lr": 1.5054519628132645e-08} +{"train_loss": 0.013848599046468735, "grad_norm": 0.0966063067317009, "global_step": 52586, "epoch": 496, "lr": 1.4981883942688514e-08} +{"train_loss": 0.015556668862700462, "grad_norm": 0.05870236083865166, "global_step": 52587, "epoch": 496, "lr": 1.490942388285843e-08} +{"train_loss": 0.01505091693252325, "grad_norm": 0.10778164118528366, "global_step": 52588, "epoch": 496, "lr": 1.4837139448903304e-08} +{"train_loss": 0.01349676214158535, "grad_norm": 0.09093066304922104, "global_step": 52589, "epoch": 496, "lr": 1.4765030641067378e-08} +{"train_loss": 0.015794605016708374, "grad_norm": 0.0632563903927803, "global_step": 52590, "epoch": 496, "lr": 1.4693097459611561e-08} +{"train_loss": 0.014347406104207039, "grad_norm": 0.052954018115997314, "global_step": 52591, "epoch": 496, "lr": 1.4621339904785647e-08} +{"train_loss": 0.01618344895541668, "grad_norm": 0.11550960689783096, "global_step": 52592, "epoch": 496, "lr": 1.454975797683944e-08} +{"train_loss": 0.016156667843461037, "grad_norm": 0.07706069946289062, "global_step": 52593, "epoch": 496, "lr": 1.447835167602829e-08} +{"train_loss": 0.014227604493498802, "grad_norm": 0.11029257625341415, "global_step": 52594, "epoch": 496, "lr": 1.4407121002601998e-08} +{"train_loss": 0.018266910687088966, "grad_norm": 0.13600075244903564, "global_step": 52595, "epoch": 496, "lr": 1.4336065956815914e-08} +{"train_loss": 0.018030494451522827, "grad_norm": 0.09031791239976883, "global_step": 52596, "epoch": 496, "lr": 1.4265186538908737e-08} +{"train_loss": 0.015824120491743088, "grad_norm": 0.07578281313180923, "global_step": 52597, "epoch": 496, "lr": 1.4194482749141369e-08} +{"train_loss": 0.016877474263310432, "grad_norm": 0.10107995569705963, "global_step": 52598, "epoch": 496, "lr": 1.4123954587752507e-08} +{"train_loss": 0.014847522601485252, "grad_norm": 0.06491521745920181, "global_step": 52599, "epoch": 496, "lr": 1.4053602054991955e-08} +{"train_loss": 0.015006257221102715, "grad_norm": 0.10421720147132874, "global_step": 52600, "epoch": 496, "lr": 1.3983425151109508e-08} +{"train_loss": 0.01346474327147007, "grad_norm": 0.06485012918710709, "global_step": 52601, "epoch": 496, "lr": 1.3913423876354969e-08} +{"train_loss": 0.01575079932808876, "grad_norm": 0.08969414234161377, "global_step": 52602, "epoch": 496, "lr": 1.3843598230967037e-08} +{"train_loss": 0.018019186332821846, "grad_norm": 0.0754503384232521, "global_step": 52603, "epoch": 496, "lr": 1.377394821519551e-08} +{"train_loss": 0.014452685602009296, "grad_norm": 0.11242827028036118, "global_step": 52604, "epoch": 496, "lr": 1.3704473829279085e-08} +{"train_loss": 0.01281621865928173, "grad_norm": 0.055306728929281235, "global_step": 52605, "epoch": 496, "lr": 1.3635175073473116e-08} +{"train_loss": 0.013923309743404388, "grad_norm": 0.05793510004878044, "global_step": 52606, "epoch": 496, "lr": 1.356605194801075e-08} +{"train_loss": 0.018383309245109558, "grad_norm": 0.08585721254348755, "global_step": 52607, "epoch": 496, "lr": 1.3497104453141785e-08} +{"train_loss": 0.015596933662891388, "grad_norm": 0.11000492423772812, "global_step": 52608, "epoch": 496, "lr": 1.342833258910492e-08} +{"train_loss": 0.016279064118862152, "grad_norm": 0.07394278049468994, "global_step": 52609, "epoch": 496, "lr": 1.3359736356138853e-08} +{"train_loss": 0.01555466465651989, "grad_norm": 0.14434318244457245, "global_step": 52610, "epoch": 496, "lr": 1.3291315754493383e-08} +{"train_loss": 0.014550114050507545, "grad_norm": 0.08744984120130539, "global_step": 52611, "epoch": 496, "lr": 1.3223070784396108e-08} +{"train_loss": 0.01616739109158516, "grad_norm": 0.07153376936912537, "global_step": 52612, "epoch": 496, "lr": 1.3155001446102377e-08} +{"train_loss": 0.014589973725378513, "grad_norm": 0.09373554587364197, "global_step": 52613, "epoch": 496, "lr": 1.3087107739839787e-08} +{"train_loss": 0.016331270337104797, "grad_norm": 0.1396971195936203, "global_step": 52614, "epoch": 496, "lr": 1.3019389665847037e-08} +{"train_loss": 0.013100014999508858, "grad_norm": 0.1408863067626953, "global_step": 52615, "epoch": 496, "lr": 1.2951847224368374e-08} +{"train_loss": 0.0131786298006773, "grad_norm": 0.0928228348493576, "global_step": 52616, "epoch": 496, "lr": 1.2884480415642497e-08} +{"train_loss": 0.016346121206879616, "grad_norm": 0.07303141802549362, "global_step": 52617, "epoch": 496, "lr": 1.2817289239897002e-08} +{"train_loss": 0.014785528182983398, "grad_norm": 0.09497960656881332, "global_step": 52618, "epoch": 496, "lr": 1.2750273697370584e-08} +{"train_loss": 0.013356922194361687, "grad_norm": 0.06459252536296844, "global_step": 52619, "epoch": 496, "lr": 1.2683433788307497e-08} +{"train_loss": 0.013919953256845474, "grad_norm": 0.09800004959106445, "global_step": 52620, "epoch": 496, "lr": 1.2616769512929783e-08} +{"train_loss": 0.01423659361898899, "grad_norm": 0.12615017592906952, "global_step": 52621, "epoch": 496, "lr": 1.255028087148169e-08} +{"train_loss": 0.013718661852180958, "grad_norm": 0.09875111281871796, "global_step": 52622, "epoch": 496, "lr": 1.2483967864190816e-08} +{"train_loss": 0.014154783450067043, "grad_norm": 0.07768874615430832, "global_step": 52623, "epoch": 496, "lr": 1.2417830491290306e-08} +{"train_loss": 0.017696836963295937, "grad_norm": 0.07566450536251068, "global_step": 52624, "epoch": 496, "lr": 1.2351868753018858e-08} +{"train_loss": 0.013021932914853096, "grad_norm": 0.08105851709842682, "global_step": 52625, "epoch": 496, "lr": 1.2286082649604069e-08} +{"train_loss": 0.01537029817700386, "grad_norm": 0.12045804411172867, "global_step": 52626, "epoch": 496, "lr": 1.2220472181273534e-08} +{"train_loss": 0.015926005318760872, "grad_norm": 0.07768239080905914, "global_step": 52627, "epoch": 496, "lr": 1.2155037348260401e-08} +{"train_loss": 0.012908171862363815, "grad_norm": 0.07808104902505875, "global_step": 52628, "epoch": 496, "lr": 1.2089778150797815e-08} +{"train_loss": 0.013630234636366367, "grad_norm": 0.15386924147605896, "global_step": 52629, "epoch": 496, "lr": 1.2024694589113372e-08} +{"train_loss": 0.01587417908012867, "grad_norm": 0.07682976871728897, "global_step": 52630, "epoch": 496, "lr": 1.1959786663434669e-08} +{"train_loss": 0.014088276773691177, "grad_norm": 0.07905066013336182, "global_step": 52631, "epoch": 496, "lr": 1.18950543739893e-08} +{"train_loss": 0.014291306957602501, "grad_norm": 0.04897488281130791, "global_step": 52632, "epoch": 496, "lr": 1.1830497721004863e-08} +{"train_loss": 0.01286490447819233, "grad_norm": 0.05871717631816864, "global_step": 52633, "epoch": 496, "lr": 1.1766116704708952e-08} +{"train_loss": 0.01420834381133318, "grad_norm": 0.08192569017410278, "global_step": 52634, "epoch": 496, "lr": 1.1701911325329163e-08} +{"train_loss": 0.01472482830286026, "grad_norm": 0.08185157179832458, "global_step": 52635, "epoch": 496, "lr": 1.163788158308754e-08} +{"train_loss": 0.014153121039271355, "grad_norm": 0.0729612484574318, "global_step": 52636, "epoch": 496, "lr": 1.157402747821168e-08} +{"train_loss": 0.013943884521722794, "grad_norm": 0.13467048108577728, "global_step": 52637, "epoch": 496, "lr": 1.1510349010923627e-08} +{"train_loss": 0.016453906893730164, "grad_norm": 0.09246634691953659, "global_step": 52638, "epoch": 496, "lr": 1.1446846181450977e-08} +{"train_loss": 0.014138623140752316, "grad_norm": 0.056952737271785736, "global_step": 52639, "epoch": 496, "lr": 1.1383518990015773e-08} +{"train_loss": 0.01653684675693512, "grad_norm": 0.14159254729747772, "global_step": 52640, "epoch": 496, "lr": 1.1320367436834511e-08} +{"train_loss": 0.016232343390583992, "grad_norm": 0.07293131202459335, "global_step": 52641, "epoch": 496, "lr": 1.1257391522140336e-08} +{"train_loss": 0.014541610144078732, "grad_norm": 0.08632969111204147, "global_step": 52642, "epoch": 496, "lr": 1.1194591246144193e-08} +{"train_loss": 0.01360524632036686, "grad_norm": 0.07056145370006561, "global_step": 52643, "epoch": 496, "lr": 1.1131966609068122e-08} +{"train_loss": 0.014652788639068604, "grad_norm": 0.18735846877098083, "global_step": 52644, "epoch": 496, "lr": 1.1069517611139723e-08} +{"train_loss": 0.016676681116223335, "grad_norm": 0.0957946628332138, "global_step": 52645, "epoch": 496, "lr": 1.1007244252575488e-08} +{"train_loss": 0.012579860165715218, "grad_norm": 0.06450638920068741, "global_step": 52646, "epoch": 496, "lr": 1.0945146533586359e-08} +{"train_loss": 0.013084844686090946, "grad_norm": 0.07500642538070679, "global_step": 52647, "epoch": 496, "lr": 1.0883224454399931e-08} +{"train_loss": 0.015361442230641842, "grad_norm": 0.11615525186061859, "global_step": 52648, "epoch": 496, "lr": 1.082147801522715e-08} +{"train_loss": 0.013332953676581383, "grad_norm": 0.07671330124139786, "global_step": 52649, "epoch": 496, "lr": 1.0759907216290055e-08} +{"train_loss": 0.015798136591911316, "grad_norm": 0.09540387243032455, "global_step": 52650, "epoch": 496, "lr": 1.0698512057799593e-08} +{"train_loss": 0.016671622171998024, "grad_norm": 0.13416585326194763, "global_step": 52651, "epoch": 496, "lr": 1.0637292539977806e-08} +{"train_loss": 0.016744762659072876, "grad_norm": 0.07035790383815765, "global_step": 52652, "epoch": 496, "lr": 1.0576248663041188e-08} +{"train_loss": 0.012815207242965698, "grad_norm": 0.06536436080932617, "global_step": 52653, "epoch": 496, "lr": 1.0515380427195132e-08} +{"train_loss": 0.019202496856451035, "grad_norm": 0.08766308426856995, "global_step": 52654, "epoch": 496, "lr": 1.0454687832656129e-08} +{"train_loss": 0.015562204644083977, "grad_norm": 0.13156156241893768, "global_step": 52655, "epoch": 496, "lr": 1.0394170879640675e-08} +{"train_loss": 0.013109741732478142, "grad_norm": 0.06368856877088547, "global_step": 52656, "epoch": 496, "lr": 1.033382956835971e-08} +{"train_loss": 0.013694231398403645, "grad_norm": 0.0777115672826767, "global_step": 52657, "epoch": 496, "lr": 1.0273663899029729e-08} +{"train_loss": 0.014129141345620155, "grad_norm": 0.07327155023813248, "global_step": 52658, "epoch": 496, "lr": 1.0213673871856123e-08} +{"train_loss": 0.01625659689307213, "grad_norm": 0.09415899962186813, "global_step": 52659, "epoch": 496, "lr": 1.0153859487049833e-08} +{"train_loss": 0.015470937825739384, "grad_norm": 0.09941595047712326, "global_step": 52660, "epoch": 496, "lr": 1.0094220744827354e-08} +{"train_loss": 0.014553619548678398, "grad_norm": 0.08563674241304398, "global_step": 52661, "epoch": 496, "lr": 1.0034757645388526e-08} +{"train_loss": 0.014985756948590279, "grad_norm": 0.06878861039876938, "global_step": 52662, "epoch": 496, "lr": 9.975470188949842e-09} +{"train_loss": 0.01612546667456627, "grad_norm": 0.06244466453790665, "global_step": 52663, "epoch": 496, "lr": 9.916358375716695e-09} +{"train_loss": 0.01676124520599842, "grad_norm": 0.1017778068780899, "global_step": 52664, "epoch": 496, "lr": 9.857422205894473e-09} +{"train_loss": 0.016316987574100494, "grad_norm": 0.1112506240606308, "global_step": 52665, "epoch": 496, "lr": 9.798661679699673e-09} +{"train_loss": 0.01472559291869402, "grad_norm": 0.07198282331228256, "global_step": 52666, "epoch": 496, "lr": 9.74007679732658e-09} +{"train_loss": 0.013683954253792763, "grad_norm": 0.11136186867952347, "global_step": 52667, "epoch": 496, "lr": 9.681667558991691e-09} +{"train_loss": 0.018027035519480705, "grad_norm": 0.11009836941957474, "global_step": 52668, "epoch": 496, "lr": 9.623433964889295e-09} +{"train_loss": 0.013789044693112373, "grad_norm": 0.09347329288721085, "global_step": 52669, "epoch": 496, "lr": 9.565376015235883e-09} +{"train_loss": 0.012269617058336735, "grad_norm": 0.07117939740419388, "global_step": 52670, "epoch": 496, "lr": 9.507493710225746e-09} +{"train_loss": 0.015661586076021194, "grad_norm": 0.0853169783949852, "global_step": 52671, "epoch": 496, "lr": 9.449787050064275e-09} +{"train_loss": 0.015065412037074566, "grad_norm": 0.11171193420886993, "global_step": 52672, "epoch": 496, "lr": 9.39225603496241e-09} +{"train_loss": 0.01443781703710556, "grad_norm": 0.0805671289563179, "global_step": 52673, "epoch": 496, "lr": 9.334900665108892e-09} +{"train_loss": 0.013074226677417755, "grad_norm": 0.1543121039867401, "global_step": 52674, "epoch": 496, "lr": 9.277720940714663e-09} +{"train_loss": 0.014107891358435154, "grad_norm": 0.057030923664569855, "global_step": 52675, "epoch": 496, "lr": 9.22071686197956e-09} +{"train_loss": 0.01576242223381996, "grad_norm": 0.06936679780483246, "global_step": 52676, "epoch": 496, "lr": 9.16388842910343e-09} +{"train_loss": 0.014338072389364243, "grad_norm": 0.07393672317266464, "global_step": 52677, "epoch": 496, "lr": 9.107235642280554e-09} +{"train_loss": 0.013684706762433052, "grad_norm": 0.10316070169210434, "global_step": 52678, "epoch": 496, "lr": 9.050758501716327e-09} +{"train_loss": 0.016255147755146027, "grad_norm": 0.08720631897449493, "global_step": 52679, "epoch": 496, "lr": 8.99445700761059e-09} +{"train_loss": 0.015953171998262405, "grad_norm": 0.09974972158670425, "global_step": 52680, "epoch": 496, "lr": 8.938331160157632e-09} +{"train_loss": 0.01493314476156572, "grad_norm": 0.07204800099134445, "global_step": 52681, "epoch": 496, "lr": 8.882380959551739e-09, "val_loss": 0.018961463123559952} +{"train_loss": 0.013533124700188637, "grad_norm": 0.05989842489361763, "global_step": 52682, "epoch": 497, "lr": 8.826606405992755e-09} +{"train_loss": 0.0155362943187356, "grad_norm": 0.07433998584747314, "global_step": 52683, "epoch": 497, "lr": 8.771007499674966e-09} +{"train_loss": 0.014967938885092735, "grad_norm": 0.14145714044570923, "global_step": 52684, "epoch": 497, "lr": 8.715584240798214e-09} +{"train_loss": 0.014931689947843552, "grad_norm": 0.10852010548114777, "global_step": 52685, "epoch": 497, "lr": 8.660336629551235e-09} +{"train_loss": 0.01612912490963936, "grad_norm": 0.08665762841701508, "global_step": 52686, "epoch": 497, "lr": 8.605264666133872e-09} +{"train_loss": 0.01641961932182312, "grad_norm": 0.07214326411485672, "global_step": 52687, "epoch": 497, "lr": 8.55036835073486e-09} +{"train_loss": 0.01500758621841669, "grad_norm": 0.09444476664066315, "global_step": 52688, "epoch": 497, "lr": 8.495647683554042e-09} +{"train_loss": 0.014323314651846886, "grad_norm": 0.06658180058002472, "global_step": 52689, "epoch": 497, "lr": 8.441102664774604e-09} +{"train_loss": 0.014852432534098625, "grad_norm": 0.07359950244426727, "global_step": 52690, "epoch": 497, "lr": 8.386733294590831e-09} +{"train_loss": 0.013476299121975899, "grad_norm": 0.13163718581199646, "global_step": 52691, "epoch": 497, "lr": 8.332539573197019e-09} +{"train_loss": 0.01623278111219406, "grad_norm": 0.08198532462120056, "global_step": 52692, "epoch": 497, "lr": 8.2785215007819e-09} +{"train_loss": 0.013160835951566696, "grad_norm": 0.0634939968585968, "global_step": 52693, "epoch": 497, "lr": 8.224679077539764e-09} +{"train_loss": 0.015011098235845566, "grad_norm": 0.06299342215061188, "global_step": 52694, "epoch": 497, "lr": 8.171012303648252e-09} +{"train_loss": 0.01513217855244875, "grad_norm": 0.07862330228090286, "global_step": 52695, "epoch": 497, "lr": 8.117521179307197e-09} +{"train_loss": 0.01478726789355278, "grad_norm": 0.08735909312963486, "global_step": 52696, "epoch": 497, "lr": 8.06420570469979e-09} +{"train_loss": 0.01389479823410511, "grad_norm": 0.08743638545274734, "global_step": 52697, "epoch": 497, "lr": 8.011065880009216e-09} +{"train_loss": 0.01439064834266901, "grad_norm": 0.06119301915168762, "global_step": 52698, "epoch": 497, "lr": 7.958101705435316e-09} +{"train_loss": 0.015621468424797058, "grad_norm": 0.06307563185691833, "global_step": 52699, "epoch": 497, "lr": 7.905313181150176e-09} +{"train_loss": 0.014800140634179115, "grad_norm": 0.08566851913928986, "global_step": 52700, "epoch": 497, "lr": 7.852700307342531e-09} +{"train_loss": 0.0171655286103487, "grad_norm": 0.0891837328672409, "global_step": 52701, "epoch": 497, "lr": 7.80026308420112e-09} +{"train_loss": 0.017627086490392685, "grad_norm": 0.08918789029121399, "global_step": 52702, "epoch": 497, "lr": 7.748001511914683e-09} +{"train_loss": 0.015372044406831264, "grad_norm": 0.061089787632226944, "global_step": 52703, "epoch": 497, "lr": 7.6959155906553e-09} +{"train_loss": 0.013532804325222969, "grad_norm": 0.1067269816994667, "global_step": 52704, "epoch": 497, "lr": 7.644005320611713e-09} +{"train_loss": 0.016056619584560394, "grad_norm": 0.07420000433921814, "global_step": 52705, "epoch": 497, "lr": 7.592270701967108e-09} +{"train_loss": 0.016535937786102295, "grad_norm": 0.0812564343214035, "global_step": 52706, "epoch": 497, "lr": 7.54071173489912e-09} +{"train_loss": 0.01471499539911747, "grad_norm": 0.058569975197315216, "global_step": 52707, "epoch": 497, "lr": 7.489328419596487e-09} +{"train_loss": 0.013884452171623707, "grad_norm": 0.06129831820726395, "global_step": 52708, "epoch": 497, "lr": 7.438120756231293e-09} +{"train_loss": 0.017358120530843735, "grad_norm": 0.12076161801815033, "global_step": 52709, "epoch": 497, "lr": 7.387088744986725e-09} +{"train_loss": 0.014885935001075268, "grad_norm": 0.0745529979467392, "global_step": 52710, "epoch": 497, "lr": 7.336232386045972e-09} +{"train_loss": 0.012905334122478962, "grad_norm": 0.08099127560853958, "global_step": 52711, "epoch": 497, "lr": 7.2855516795866665e-09} +{"train_loss": 0.012976080179214478, "grad_norm": 0.07756004482507706, "global_step": 52712, "epoch": 497, "lr": 7.235046625780894e-09} +{"train_loss": 0.01725875213742256, "grad_norm": 0.06073484569787979, "global_step": 52713, "epoch": 497, "lr": 7.1847172248062924e-09} +{"train_loss": 0.013258817605674267, "grad_norm": 0.08935572952032089, "global_step": 52714, "epoch": 497, "lr": 7.134563476851597e-09} +{"train_loss": 0.01506668608635664, "grad_norm": 0.09286275506019592, "global_step": 52715, "epoch": 497, "lr": 7.084585382077791e-09} +{"train_loss": 0.01600637100636959, "grad_norm": 0.13698086142539978, "global_step": 52716, "epoch": 497, "lr": 7.034782940668061e-09} +{"train_loss": 0.015507997944951057, "grad_norm": 0.06627357006072998, "global_step": 52717, "epoch": 497, "lr": 6.985156152800043e-09} +{"train_loss": 0.016820989549160004, "grad_norm": 0.06450101733207703, "global_step": 52718, "epoch": 497, "lr": 6.935705018645822e-09} +{"train_loss": 0.014882983639836311, "grad_norm": 0.06444217264652252, "global_step": 52719, "epoch": 497, "lr": 6.886429538377481e-09} +{"train_loss": 0.01351470872759819, "grad_norm": 0.06116994842886925, "global_step": 52720, "epoch": 497, "lr": 6.837329712167107e-09} +{"train_loss": 0.014154158532619476, "grad_norm": 0.05575970932841301, "global_step": 52721, "epoch": 497, "lr": 6.788405540186782e-09} +{"train_loss": 0.01421102974563837, "grad_norm": 0.10847878456115723, "global_step": 52722, "epoch": 497, "lr": 6.739657022614143e-09} +{"train_loss": 0.01743209920823574, "grad_norm": 0.07424869388341904, "global_step": 52723, "epoch": 497, "lr": 6.691084159615724e-09} +{"train_loss": 0.014701169915497303, "grad_norm": 0.07275649905204773, "global_step": 52724, "epoch": 497, "lr": 6.642686951363608e-09} +{"train_loss": 0.015443826094269753, "grad_norm": 0.05731469392776489, "global_step": 52725, "epoch": 497, "lr": 6.59446539802433e-09} +{"train_loss": 0.01475728489458561, "grad_norm": 0.07425082474946976, "global_step": 52726, "epoch": 497, "lr": 6.546419499775525e-09} +{"train_loss": 0.014707352966070175, "grad_norm": 0.08046794682741165, "global_step": 52727, "epoch": 497, "lr": 6.498549256778175e-09} +{"train_loss": 0.0144706591963768, "grad_norm": 0.08808371424674988, "global_step": 52728, "epoch": 497, "lr": 6.4508546692043646e-09} +{"train_loss": 0.013879489153623581, "grad_norm": 0.06966488063335419, "global_step": 52729, "epoch": 497, "lr": 6.4033357372206285e-09} +{"train_loss": 0.014395218342542648, "grad_norm": 0.06490785628557205, "global_step": 52730, "epoch": 497, "lr": 6.355992460993498e-09} +{"train_loss": 0.016207372769713402, "grad_norm": 0.09319654852151871, "global_step": 52731, "epoch": 497, "lr": 6.308824840689509e-09} +{"train_loss": 0.015006421133875847, "grad_norm": 0.06473101675510406, "global_step": 52732, "epoch": 497, "lr": 6.261832876475193e-09} +{"train_loss": 0.014186741784214973, "grad_norm": 0.05718938633799553, "global_step": 52733, "epoch": 497, "lr": 6.215016568511534e-09} +{"train_loss": 0.014394211582839489, "grad_norm": 0.0656924843788147, "global_step": 52734, "epoch": 497, "lr": 6.1683759169706146e-09} +{"train_loss": 0.012552877888083458, "grad_norm": 0.12224648892879486, "global_step": 52735, "epoch": 497, "lr": 6.121910922013418e-09} +{"train_loss": 0.014869273640215397, "grad_norm": 0.06905090808868408, "global_step": 52736, "epoch": 497, "lr": 6.0756215837953765e-09} +{"train_loss": 0.014829840511083603, "grad_norm": 0.12214698642492294, "global_step": 52737, "epoch": 497, "lr": 6.0295079024941244e-09} +{"train_loss": 0.017360679805278778, "grad_norm": 0.10174081474542618, "global_step": 52738, "epoch": 497, "lr": 5.983569878259543e-09} +{"train_loss": 0.0166408009827137, "grad_norm": 0.09734521061182022, "global_step": 52739, "epoch": 497, "lr": 5.9378075112581645e-09} +{"train_loss": 0.013868642970919609, "grad_norm": 0.10737708956003189, "global_step": 52740, "epoch": 497, "lr": 5.89222080164542e-09} +{"train_loss": 0.015013198368251324, "grad_norm": 0.08895178139209747, "global_step": 52741, "epoch": 497, "lr": 5.846809749587845e-09} +{"train_loss": 0.01320883072912693, "grad_norm": 0.09779205918312073, "global_step": 52742, "epoch": 497, "lr": 5.80157435524642e-09} +{"train_loss": 0.013838130049407482, "grad_norm": 0.09881175309419632, "global_step": 52743, "epoch": 497, "lr": 5.756514618771025e-09} +{"train_loss": 0.017195070162415504, "grad_norm": 0.0676068514585495, "global_step": 52744, "epoch": 497, "lr": 5.711630540328195e-09} +{"train_loss": 0.012458560988307, "grad_norm": 0.09400682896375656, "global_step": 52745, "epoch": 497, "lr": 5.666922120073359e-09} +{"train_loss": 0.018325746059417725, "grad_norm": 0.10912955552339554, "global_step": 52746, "epoch": 497, "lr": 5.622389358161951e-09} +{"train_loss": 0.016166530549526215, "grad_norm": 0.0780469998717308, "global_step": 52747, "epoch": 497, "lr": 5.578032254749399e-09} +{"train_loss": 0.01365690678358078, "grad_norm": 0.08958832919597626, "global_step": 52748, "epoch": 497, "lr": 5.5338508099966875e-09} +{"train_loss": 0.013827389106154442, "grad_norm": 0.07761748880147934, "global_step": 52749, "epoch": 497, "lr": 5.489845024053697e-09} +{"train_loss": 0.014449482783675194, "grad_norm": 0.10514965653419495, "global_step": 52750, "epoch": 497, "lr": 5.446014897081408e-09} +{"train_loss": 0.015952758491039276, "grad_norm": 0.05991669371724129, "global_step": 52751, "epoch": 497, "lr": 5.402360429224151e-09} +{"train_loss": 0.014840517193078995, "grad_norm": 0.08169552683830261, "global_step": 52752, "epoch": 497, "lr": 5.358881620642908e-09} +{"train_loss": 0.012252725660800934, "grad_norm": 0.08375327289104462, "global_step": 52753, "epoch": 497, "lr": 5.31557847149311e-09} +{"train_loss": 0.013896306045353413, "grad_norm": 0.06222335994243622, "global_step": 52754, "epoch": 497, "lr": 5.2724509819135344e-09} +{"train_loss": 0.01398053951561451, "grad_norm": 0.107872873544693, "global_step": 52755, "epoch": 497, "lr": 5.229499152070716e-09} +{"train_loss": 0.014892124570906162, "grad_norm": 0.11667562276124954, "global_step": 52756, "epoch": 497, "lr": 5.186722982108982e-09} +{"train_loss": 0.01602228917181492, "grad_norm": 0.07945641875267029, "global_step": 52757, "epoch": 497, "lr": 5.144122472178214e-09} +{"train_loss": 0.015140699222683907, "grad_norm": 0.059112802147865295, "global_step": 52758, "epoch": 497, "lr": 5.101697622428292e-09} +{"train_loss": 0.016046546399593353, "grad_norm": 0.13672828674316406, "global_step": 52759, "epoch": 497, "lr": 5.0594484330090955e-09} +{"train_loss": 0.014401871711015701, "grad_norm": 0.09783784300088882, "global_step": 52760, "epoch": 497, "lr": 5.0173749040705044e-09} +{"train_loss": 0.015559788793325424, "grad_norm": 0.10176600515842438, "global_step": 52761, "epoch": 497, "lr": 4.975477035756848e-09} +{"train_loss": 0.015996893867850304, "grad_norm": 0.07287776470184326, "global_step": 52762, "epoch": 497, "lr": 4.933754828212456e-09} +{"train_loss": 0.014738664031028748, "grad_norm": 0.06600024551153183, "global_step": 52763, "epoch": 497, "lr": 4.892208281592758e-09} +{"train_loss": 0.015889551490545273, "grad_norm": 0.08840014785528183, "global_step": 52764, "epoch": 497, "lr": 4.850837396042085e-09} +{"train_loss": 0.014709332957863808, "grad_norm": 0.07620631158351898, "global_step": 52765, "epoch": 497, "lr": 4.809642171699214e-09} +{"train_loss": 0.013093488290905952, "grad_norm": 0.05992591008543968, "global_step": 52766, "epoch": 497, "lr": 4.768622608719575e-09} +{"train_loss": 0.012665822170674801, "grad_norm": 0.06392429023981094, "global_step": 52767, "epoch": 497, "lr": 4.727778707230846e-09} +{"train_loss": 0.015572340227663517, "grad_norm": 0.09352052211761475, "global_step": 52768, "epoch": 497, "lr": 4.6871104673940065e-09} +{"train_loss": 0.0147725660353899, "grad_norm": 0.08271130174398422, "global_step": 52769, "epoch": 497, "lr": 4.6466178893422865e-09} +{"train_loss": 0.013618994504213333, "grad_norm": 0.0776795744895935, "global_step": 52770, "epoch": 497, "lr": 4.606300973220013e-09} +{"train_loss": 0.013767339289188385, "grad_norm": 0.05963243916630745, "global_step": 52771, "epoch": 497, "lr": 4.566159719165963e-09} +{"train_loss": 0.014986866153776646, "grad_norm": 0.0711071789264679, "global_step": 52772, "epoch": 497, "lr": 4.526194127324468e-09} +{"train_loss": 0.014541002921760082, "grad_norm": 0.08828090131282806, "global_step": 52773, "epoch": 497, "lr": 4.486404197839855e-09} +{"train_loss": 0.015653815120458603, "grad_norm": 0.09392843395471573, "global_step": 52774, "epoch": 497, "lr": 4.446789930845352e-09} +{"train_loss": 0.0130165359005332, "grad_norm": 0.10491658002138138, "global_step": 52775, "epoch": 497, "lr": 4.407351326479736e-09} +{"train_loss": 0.019082367420196533, "grad_norm": 0.14822308719158173, "global_step": 52776, "epoch": 497, "lr": 4.3680883848873365e-09} +{"train_loss": 0.015036523342132568, "grad_norm": 0.08816087990999222, "global_step": 52777, "epoch": 497, "lr": 4.32900110620138e-09} +{"train_loss": 0.013615981675684452, "grad_norm": 0.10051575303077698, "global_step": 52778, "epoch": 497, "lr": 4.2900894905606444e-09} +{"train_loss": 0.014483533799648285, "grad_norm": 0.08491622656583786, "global_step": 52779, "epoch": 497, "lr": 4.251353538103908e-09} +{"train_loss": 0.01365470141172409, "grad_norm": 0.06230783462524414, "global_step": 52780, "epoch": 497, "lr": 4.212793248964398e-09} +{"train_loss": 0.016337603330612183, "grad_norm": 0.11631415039300919, "global_step": 52781, "epoch": 497, "lr": 4.174408623275339e-09} +{"train_loss": 0.017299840226769447, "grad_norm": 0.06196970120072365, "global_step": 52782, "epoch": 497, "lr": 4.136199661175511e-09} +{"train_loss": 0.014924272894859314, "grad_norm": 0.08442799746990204, "global_step": 52783, "epoch": 497, "lr": 4.09816636280369e-09} +{"train_loss": 0.015505118295550346, "grad_norm": 0.1307828724384308, "global_step": 52784, "epoch": 497, "lr": 4.060308728282003e-09} +{"train_loss": 0.01508872490376234, "grad_norm": 0.08671078830957413, "global_step": 52785, "epoch": 497, "lr": 4.022626757754777e-09} +{"train_loss": 0.015815667808055878, "grad_norm": 0.14464907348155975, "global_step": 52786, "epoch": 497, "lr": 3.985120451349688e-09} +{"train_loss": 0.014938995549631006, "grad_norm": 0.07440453767776489, "global_step": 52787, "epoch": 497, "lr": 3.947789809194414e-09, "val_loss": 0.019865091890096664} +{"train_loss": 0.012925493530929089, "grad_norm": 0.0626903846859932, "global_step": 52788, "epoch": 498, "lr": 3.910634831422177e-09} +{"train_loss": 0.013718076050281525, "grad_norm": 0.07250252366065979, "global_step": 52789, "epoch": 498, "lr": 3.873655518171759e-09} +{"train_loss": 0.016331976279616356, "grad_norm": 0.06939995288848877, "global_step": 52790, "epoch": 498, "lr": 3.836851869565283e-09} +{"train_loss": 0.01696981117129326, "grad_norm": 0.09714572876691818, "global_step": 52791, "epoch": 498, "lr": 3.800223885730425e-09} +{"train_loss": 0.013190273195505142, "grad_norm": 0.072507843375206, "global_step": 52792, "epoch": 498, "lr": 3.763771566800411e-09} +{"train_loss": 0.013545410707592964, "grad_norm": 0.07459834218025208, "global_step": 52793, "epoch": 498, "lr": 3.727494912897367e-09} +{"train_loss": 0.012778748758137226, "grad_norm": 0.1264462172985077, "global_step": 52794, "epoch": 498, "lr": 3.6913939241600692e-09} +{"train_loss": 0.016442136839032173, "grad_norm": 0.08171258121728897, "global_step": 52795, "epoch": 498, "lr": 3.655468600705092e-09} +{"train_loss": 0.01290272455662489, "grad_norm": 0.06786154210567474, "global_step": 52796, "epoch": 498, "lr": 3.619718942660111e-09} +{"train_loss": 0.01388917863368988, "grad_norm": 0.0874277874827385, "global_step": 52797, "epoch": 498, "lr": 3.5841449501528013e-09} +{"train_loss": 0.013305334374308586, "grad_norm": 0.11690449714660645, "global_step": 52798, "epoch": 498, "lr": 3.5487466233108392e-09} +{"train_loss": 0.014909213408827782, "grad_norm": 0.09338045120239258, "global_step": 52799, "epoch": 498, "lr": 3.513523962256349e-09} +{"train_loss": 0.014714998193085194, "grad_norm": 0.09646832942962646, "global_step": 52800, "epoch": 498, "lr": 3.4784769671059036e-09} +{"train_loss": 0.014705612324178219, "grad_norm": 0.07897740602493286, "global_step": 52801, "epoch": 498, "lr": 3.443605637998282e-09} +{"train_loss": 0.01430335734039545, "grad_norm": 0.07374241203069687, "global_step": 52802, "epoch": 498, "lr": 3.408909975038954e-09} +{"train_loss": 0.017121030017733574, "grad_norm": 0.09440846741199493, "global_step": 52803, "epoch": 498, "lr": 3.374389978361148e-09} +{"train_loss": 0.012266959995031357, "grad_norm": 0.08528096228837967, "global_step": 52804, "epoch": 498, "lr": 3.340045648086987e-09} +{"train_loss": 0.014702863059937954, "grad_norm": 0.055999625474214554, "global_step": 52805, "epoch": 498, "lr": 3.3058769843274938e-09} +{"train_loss": 0.014001686125993729, "grad_norm": 0.06797536462545395, "global_step": 52806, "epoch": 498, "lr": 3.2718839872103444e-09} +{"train_loss": 0.01448500994592905, "grad_norm": 0.05496729910373688, "global_step": 52807, "epoch": 498, "lr": 3.2380666568521122e-09} +{"train_loss": 0.014879686757922173, "grad_norm": 0.0668732225894928, "global_step": 52808, "epoch": 498, "lr": 3.2044249933693706e-09} +{"train_loss": 0.014338653534650803, "grad_norm": 0.09441091865301132, "global_step": 52809, "epoch": 498, "lr": 3.1709589968842436e-09} +{"train_loss": 0.017605382949113846, "grad_norm": 0.07213111221790314, "global_step": 52810, "epoch": 498, "lr": 3.137668667518856e-09} +{"train_loss": 0.016799475997686386, "grad_norm": 0.05650324374437332, "global_step": 52811, "epoch": 498, "lr": 3.104554005378679e-09} +{"train_loss": 0.0150150703266263, "grad_norm": 0.07858873903751373, "global_step": 52812, "epoch": 498, "lr": 3.0716150105858375e-09} +{"train_loss": 0.016218937933444977, "grad_norm": 0.11501073092222214, "global_step": 52813, "epoch": 498, "lr": 3.038851683262456e-09} +{"train_loss": 0.016299676150083542, "grad_norm": 0.07546967267990112, "global_step": 52814, "epoch": 498, "lr": 3.0062640235140047e-09} +{"train_loss": 0.01645982638001442, "grad_norm": 0.10807449370622635, "global_step": 52815, "epoch": 498, "lr": 2.9738520314570586e-09} +{"train_loss": 0.015164894983172417, "grad_norm": 0.1030680239200592, "global_step": 52816, "epoch": 498, "lr": 2.9416157072081895e-09} +{"train_loss": 0.015878722071647644, "grad_norm": 0.10404432564973831, "global_step": 52817, "epoch": 498, "lr": 2.909555050878421e-09} +{"train_loss": 0.013097567483782768, "grad_norm": 0.08174800872802734, "global_step": 52818, "epoch": 498, "lr": 2.8776700625787743e-09} +{"train_loss": 0.013127717189490795, "grad_norm": 0.08223042637109756, "global_step": 52819, "epoch": 498, "lr": 2.845960742431375e-09} +{"train_loss": 0.013809197582304478, "grad_norm": 0.05461841821670532, "global_step": 52820, "epoch": 498, "lr": 2.8144270905305915e-09} +{"train_loss": 0.01590622216463089, "grad_norm": 0.10101445019245148, "global_step": 52821, "epoch": 498, "lr": 2.7830691070040994e-09} +{"train_loss": 0.01694430224597454, "grad_norm": 0.12615464627742767, "global_step": 52822, "epoch": 498, "lr": 2.7518867919518188e-09} +{"train_loss": 0.015154223889112473, "grad_norm": 0.06650885194540024, "global_step": 52823, "epoch": 498, "lr": 2.720880145484772e-09} +{"train_loss": 0.013250218704342842, "grad_norm": 0.07158062607049942, "global_step": 52824, "epoch": 498, "lr": 2.6900491677195326e-09} +{"train_loss": 0.01656845211982727, "grad_norm": 0.056241437792778015, "global_step": 52825, "epoch": 498, "lr": 2.6593938587504697e-09} +{"train_loss": 0.014976716600358486, "grad_norm": 0.10262072831392288, "global_step": 52826, "epoch": 498, "lr": 2.628914218699707e-09} +{"train_loss": 0.015009739436209202, "grad_norm": 0.10587441921234131, "global_step": 52827, "epoch": 498, "lr": 2.5986102476616147e-09} +{"train_loss": 0.015627410262823105, "grad_norm": 0.1081971749663353, "global_step": 52828, "epoch": 498, "lr": 2.5684819457527655e-09} +{"train_loss": 0.01391194574534893, "grad_norm": 0.06607386469841003, "global_step": 52829, "epoch": 498, "lr": 2.5385293130730793e-09} +{"train_loss": 0.0136469267308712, "grad_norm": 0.047747790813446045, "global_step": 52830, "epoch": 498, "lr": 2.5087523497280275e-09} +{"train_loss": 0.01293252408504486, "grad_norm": 0.06562028080224991, "global_step": 52831, "epoch": 498, "lr": 2.4791510558286322e-09} +{"train_loss": 0.012652399949729443, "grad_norm": 0.07960204780101776, "global_step": 52832, "epoch": 498, "lr": 2.449725431469263e-09} +{"train_loss": 0.015406185761094093, "grad_norm": 0.14359769225120544, "global_step": 52833, "epoch": 498, "lr": 2.4204754767609415e-09} +{"train_loss": 0.01629030704498291, "grad_norm": 0.10745368897914886, "global_step": 52834, "epoch": 498, "lr": 2.3914011918035883e-09} +{"train_loss": 0.017911959439516068, "grad_norm": 0.06979593634605408, "global_step": 52835, "epoch": 498, "lr": 2.3625025767026744e-09} +{"train_loss": 0.015460507944226265, "grad_norm": 0.08980447798967361, "global_step": 52836, "epoch": 498, "lr": 2.3337796315525684e-09} +{"train_loss": 0.014966974034905434, "grad_norm": 0.08058275282382965, "global_step": 52837, "epoch": 498, "lr": 2.305232356458742e-09} +{"train_loss": 0.016655344516038895, "grad_norm": 0.05904446169734001, "global_step": 52838, "epoch": 498, "lr": 2.276860751521115e-09} +{"train_loss": 0.012914691120386124, "grad_norm": 0.0650109350681305, "global_step": 52839, "epoch": 498, "lr": 2.2486648168396075e-09} +{"train_loss": 0.016472917050123215, "grad_norm": 0.07665925472974777, "global_step": 52840, "epoch": 498, "lr": 2.2206445525085883e-09} +{"train_loss": 0.014268295839428902, "grad_norm": 0.08021838963031769, "global_step": 52841, "epoch": 498, "lr": 2.192799958633529e-09} +{"train_loss": 0.014467122033238411, "grad_norm": 0.05573156476020813, "global_step": 52842, "epoch": 498, "lr": 2.1651310353143494e-09} +{"train_loss": 0.013522550463676453, "grad_norm": 0.08578769862651825, "global_step": 52843, "epoch": 498, "lr": 2.1376377826343164e-09} +{"train_loss": 0.015581317245960236, "grad_norm": 0.1583370715379715, "global_step": 52844, "epoch": 498, "lr": 2.110320200704452e-09} +{"train_loss": 0.013962049037218094, "grad_norm": 0.09502988308668137, "global_step": 52845, "epoch": 498, "lr": 2.0831782896191254e-09} +{"train_loss": 0.014747563749551773, "grad_norm": 0.07383787631988525, "global_step": 52846, "epoch": 498, "lr": 2.056212049461603e-09} +{"train_loss": 0.012913359329104424, "grad_norm": 0.0665246993303299, "global_step": 52847, "epoch": 498, "lr": 2.029421480342908e-09} +{"train_loss": 0.01528517808765173, "grad_norm": 0.06161634996533394, "global_step": 52848, "epoch": 498, "lr": 2.002806582346306e-09} +{"train_loss": 0.015586433932185173, "grad_norm": 0.08113566786050797, "global_step": 52849, "epoch": 498, "lr": 1.9763673555661664e-09} +{"train_loss": 0.013536164537072182, "grad_norm": 0.06364675611257553, "global_step": 52850, "epoch": 498, "lr": 1.9501038000968584e-09} +{"train_loss": 0.014120841398835182, "grad_norm": 0.09001481533050537, "global_step": 52851, "epoch": 498, "lr": 1.924015916038302e-09} +{"train_loss": 0.016699520871043205, "grad_norm": 0.12147105485200882, "global_step": 52852, "epoch": 498, "lr": 1.8981037034682126e-09} +{"train_loss": 0.012349131517112255, "grad_norm": 0.0501619353890419, "global_step": 52853, "epoch": 498, "lr": 1.8723671624865103e-09} +{"train_loss": 0.013370400294661522, "grad_norm": 0.05328957363963127, "global_step": 52854, "epoch": 498, "lr": 1.8468062931820128e-09} +{"train_loss": 0.014651471748948097, "grad_norm": 0.09612131863832474, "global_step": 52855, "epoch": 498, "lr": 1.8214210956435384e-09} +{"train_loss": 0.01697367988526821, "grad_norm": 0.12818783521652222, "global_step": 52856, "epoch": 498, "lr": 1.7962115699599046e-09} +{"train_loss": 0.014795355498790741, "grad_norm": 0.07825496792793274, "global_step": 52857, "epoch": 498, "lr": 1.7711777162254806e-09} +{"train_loss": 0.014810323715209961, "grad_norm": 0.06955981999635696, "global_step": 52858, "epoch": 498, "lr": 1.7463195345179817e-09} +{"train_loss": 0.01618974842131138, "grad_norm": 0.09481306374073029, "global_step": 52859, "epoch": 498, "lr": 1.721637024931777e-09} +{"train_loss": 0.016681481152772903, "grad_norm": 0.054897863417863846, "global_step": 52860, "epoch": 498, "lr": 1.6971301875556845e-09} +{"train_loss": 0.016126297414302826, "grad_norm": 0.06087614968419075, "global_step": 52861, "epoch": 498, "lr": 1.6727990224674194e-09} +{"train_loss": 0.01402333378791809, "grad_norm": 0.064218670129776, "global_step": 52862, "epoch": 498, "lr": 1.6486435297613512e-09} +{"train_loss": 0.015541676431894302, "grad_norm": 0.08731652051210403, "global_step": 52863, "epoch": 498, "lr": 1.624663709515195e-09} +{"train_loss": 0.014826798811554909, "grad_norm": 0.12460102885961533, "global_step": 52864, "epoch": 498, "lr": 1.600859561817769e-09} +{"train_loss": 0.015902135521173477, "grad_norm": 0.09151516854763031, "global_step": 52865, "epoch": 498, "lr": 1.5772310867523399e-09} +{"train_loss": 0.015347467735409737, "grad_norm": 0.10570045560598373, "global_step": 52866, "epoch": 498, "lr": 1.5537782843966231e-09} +{"train_loss": 0.013123051263391972, "grad_norm": 0.1009366363286972, "global_step": 52867, "epoch": 498, "lr": 1.5305011548394365e-09} +{"train_loss": 0.015865754336118698, "grad_norm": 0.08205537497997284, "global_step": 52868, "epoch": 498, "lr": 1.507399698164047e-09} +{"train_loss": 0.01586749218404293, "grad_norm": 0.06628333777189255, "global_step": 52869, "epoch": 498, "lr": 1.484473914442619e-09} +{"train_loss": 0.01479540579020977, "grad_norm": 0.0672207772731781, "global_step": 52870, "epoch": 498, "lr": 1.4617238037639702e-09} +{"train_loss": 0.015576709993183613, "grad_norm": 0.10546936094760895, "global_step": 52871, "epoch": 498, "lr": 1.4391493662002652e-09} +{"train_loss": 0.01405315101146698, "grad_norm": 0.08971969038248062, "global_step": 52872, "epoch": 498, "lr": 1.416750601840322e-09} +{"train_loss": 0.014994567260146141, "grad_norm": 0.09214483201503754, "global_step": 52873, "epoch": 498, "lr": 1.394527510756305e-09} +{"train_loss": 0.0145618487149477, "grad_norm": 0.1092272400856018, "global_step": 52874, "epoch": 498, "lr": 1.3724800930314807e-09} +{"train_loss": 0.0180964395403862, "grad_norm": 0.09924901276826859, "global_step": 52875, "epoch": 498, "lr": 1.3506083487380138e-09} +{"train_loss": 0.014999188482761383, "grad_norm": 0.07261929661035538, "global_step": 52876, "epoch": 498, "lr": 1.32891227795362e-09} +{"train_loss": 0.01639597862958908, "grad_norm": 0.08692969381809235, "global_step": 52877, "epoch": 498, "lr": 1.3073918807560148e-09} +{"train_loss": 0.014922957867383957, "grad_norm": 0.0691784992814064, "global_step": 52878, "epoch": 498, "lr": 1.2860471572173626e-09} +{"train_loss": 0.015047100372612476, "grad_norm": 0.0707874745130539, "global_step": 52879, "epoch": 498, "lr": 1.2648781074209304e-09} +{"train_loss": 0.01624538004398346, "grad_norm": 0.08377902954816818, "global_step": 52880, "epoch": 498, "lr": 1.2438847314333314e-09} +{"train_loss": 0.014416148886084557, "grad_norm": 0.09477580338716507, "global_step": 52881, "epoch": 498, "lr": 1.22306702932673e-09} +{"train_loss": 0.014876552857458591, "grad_norm": 0.07027160376310349, "global_step": 52882, "epoch": 498, "lr": 1.202425001184393e-09} +{"train_loss": 0.016429783776402473, "grad_norm": 0.07681997120380402, "global_step": 52883, "epoch": 498, "lr": 1.1819586470729338e-09} +{"train_loss": 0.013576175086200237, "grad_norm": 0.05787273868918419, "global_step": 52884, "epoch": 498, "lr": 1.1616679670589658e-09} +{"train_loss": 0.01711149513721466, "grad_norm": 0.08835732936859131, "global_step": 52885, "epoch": 498, "lr": 1.1415529612257557e-09} +{"train_loss": 0.014520576223731041, "grad_norm": 0.08464190363883972, "global_step": 52886, "epoch": 498, "lr": 1.121613629628815e-09} +{"train_loss": 0.013740962371230125, "grad_norm": 0.05619050934910774, "global_step": 52887, "epoch": 498, "lr": 1.1018499723514097e-09} +{"train_loss": 0.012645913287997246, "grad_norm": 0.12273906916379929, "global_step": 52888, "epoch": 498, "lr": 1.0822619894601537e-09} +{"train_loss": 0.01594276912510395, "grad_norm": 0.08993936330080032, "global_step": 52889, "epoch": 498, "lr": 1.0628496810216604e-09} +{"train_loss": 0.018521098420023918, "grad_norm": 0.10031864792108536, "global_step": 52890, "epoch": 498, "lr": 1.0436130471025429e-09} +{"train_loss": 0.015178465284407139, "grad_norm": 0.20342516899108887, "global_step": 52891, "epoch": 498, "lr": 1.0245520877694148e-09} +{"train_loss": 0.01396794430911541, "grad_norm": 0.0655190497636795, "global_step": 52892, "epoch": 498, "lr": 1.0056668030944406e-09} +{"train_loss": 0.014940808149370944, "grad_norm": 0.12527592480182648, "global_step": 52893, "epoch": 498, "lr": 9.869571931442334e-10, "val_loss": 0.019914517179131508} +{"train_loss": 0.013854939490556717, "grad_norm": 0.06690125167369843, "global_step": 52894, "epoch": 499, "lr": 9.684232579798557e-10} +{"train_loss": 0.013794492930173874, "grad_norm": 0.06881251931190491, "global_step": 52895, "epoch": 499, "lr": 9.50064997667921e-10} +{"train_loss": 0.016173869371414185, "grad_norm": 0.18368642032146454, "global_step": 52896, "epoch": 499, "lr": 9.318824122750425e-10} +{"train_loss": 0.016013771295547485, "grad_norm": 0.06420490890741348, "global_step": 52897, "epoch": 499, "lr": 9.138755018622824e-10} +{"train_loss": 0.014769808389246464, "grad_norm": 0.08994381874799728, "global_step": 52898, "epoch": 499, "lr": 8.960442664907031e-10} +{"train_loss": 0.015721939504146576, "grad_norm": 0.09088654071092606, "global_step": 52899, "epoch": 499, "lr": 8.783887062324692e-10} +{"train_loss": 0.016142670065164566, "grad_norm": 0.06787104904651642, "global_step": 52900, "epoch": 499, "lr": 8.609088211375404e-10} +{"train_loss": 0.014089391566812992, "grad_norm": 0.07178155332803726, "global_step": 52901, "epoch": 499, "lr": 8.436046112780815e-10} +{"train_loss": 0.013413477689027786, "grad_norm": 0.06605727970600128, "global_step": 52902, "epoch": 499, "lr": 8.264760767096036e-10} +{"train_loss": 0.014378143474459648, "grad_norm": 0.06822000443935394, "global_step": 52903, "epoch": 499, "lr": 8.095232174931689e-10} +{"train_loss": 0.013686222955584526, "grad_norm": 0.1056957021355629, "global_step": 52904, "epoch": 499, "lr": 7.927460336842885e-10} +{"train_loss": 0.017118006944656372, "grad_norm": 0.0651266872882843, "global_step": 52905, "epoch": 499, "lr": 7.761445253495758e-10} +{"train_loss": 0.01605594903230667, "grad_norm": 0.06680923700332642, "global_step": 52906, "epoch": 499, "lr": 7.59718692538991e-10} +{"train_loss": 0.012543403543531895, "grad_norm": 0.09110139310359955, "global_step": 52907, "epoch": 499, "lr": 7.434685353191473e-10} +{"train_loss": 0.013749569654464722, "grad_norm": 0.09484288841485977, "global_step": 52908, "epoch": 499, "lr": 7.273940537400048e-10} +{"train_loss": 0.013969965279102325, "grad_norm": 0.061572760343551636, "global_step": 52909, "epoch": 499, "lr": 7.114952478626258e-10} +{"train_loss": 0.014654533937573433, "grad_norm": 0.07068806141614914, "global_step": 52910, "epoch": 499, "lr": 6.957721177369702e-10} +{"train_loss": 0.013677669689059258, "grad_norm": 0.06202266737818718, "global_step": 52911, "epoch": 499, "lr": 6.802246634296516e-10} +{"train_loss": 0.012543469667434692, "grad_norm": 0.050674956291913986, "global_step": 52912, "epoch": 499, "lr": 6.648528849795277e-10} +{"train_loss": 0.01448478177189827, "grad_norm": 0.12522269785404205, "global_step": 52913, "epoch": 499, "lr": 6.496567824532119e-10} +{"train_loss": 0.01464779581874609, "grad_norm": 0.07059931755065918, "global_step": 52914, "epoch": 499, "lr": 6.34636355900664e-10} +{"train_loss": 0.016337858512997627, "grad_norm": 0.09506836533546448, "global_step": 52915, "epoch": 499, "lr": 6.197916053718445e-10} +{"train_loss": 0.016080539673566818, "grad_norm": 0.05906286463141441, "global_step": 52916, "epoch": 499, "lr": 6.051225309222641e-10} +{"train_loss": 0.013418283313512802, "grad_norm": 0.06213448941707611, "global_step": 52917, "epoch": 499, "lr": 5.906291326018831e-10} +{"train_loss": 0.016919001936912537, "grad_norm": 0.06264419853687286, "global_step": 52918, "epoch": 499, "lr": 5.763114104606615e-10} +{"train_loss": 0.015095947310328484, "grad_norm": 0.05850264057517052, "global_step": 52919, "epoch": 499, "lr": 5.621693645541104e-10} +{"train_loss": 0.01657075807452202, "grad_norm": 0.07531709969043732, "global_step": 52920, "epoch": 499, "lr": 5.482029949266387e-10} +{"train_loss": 0.01465049758553505, "grad_norm": 0.06705804914236069, "global_step": 52921, "epoch": 499, "lr": 5.344123016226554e-10} +{"train_loss": 0.018035748973488808, "grad_norm": 0.10219435393810272, "global_step": 52922, "epoch": 499, "lr": 5.207972847032228e-10} +{"train_loss": 0.015826020389795303, "grad_norm": 0.07085723429918289, "global_step": 52923, "epoch": 499, "lr": 5.073579442071985e-10} +{"train_loss": 0.01601380854845047, "grad_norm": 0.09179136902093887, "global_step": 52924, "epoch": 499, "lr": 4.940942801845427e-10} +{"train_loss": 0.016905467957258224, "grad_norm": 0.06990401446819305, "global_step": 52925, "epoch": 499, "lr": 4.810062926796643e-10} +{"train_loss": 0.017884477972984314, "grad_norm": 0.07659710943698883, "global_step": 52926, "epoch": 499, "lr": 4.680939817425234e-10} +{"train_loss": 0.015244029462337494, "grad_norm": 0.08568063378334045, "global_step": 52927, "epoch": 499, "lr": 4.553573474119777e-10} +{"train_loss": 0.012499482370913029, "grad_norm": 0.05874462053179741, "global_step": 52928, "epoch": 499, "lr": 4.427963897435383e-10} +{"train_loss": 0.01535212155431509, "grad_norm": 0.06823138892650604, "global_step": 52929, "epoch": 499, "lr": 4.3041110877051203e-10} +{"train_loss": 0.01530302781611681, "grad_norm": 0.07657922059297562, "global_step": 52930, "epoch": 499, "lr": 4.182015045428589e-10} +{"train_loss": 0.015497103333473206, "grad_norm": 0.12187867611646652, "global_step": 52931, "epoch": 499, "lr": 4.0616757709943667e-10} +{"train_loss": 0.013381540775299072, "grad_norm": 0.06907995045185089, "global_step": 52932, "epoch": 499, "lr": 3.9430932648465426e-10} +{"train_loss": 0.015406912192702293, "grad_norm": 0.1251087188720703, "global_step": 52933, "epoch": 499, "lr": 3.8262675274292057e-10} +{"train_loss": 0.014682909473776817, "grad_norm": 0.09048233181238174, "global_step": 52934, "epoch": 499, "lr": 3.711198559075424e-10} +{"train_loss": 0.014239855110645294, "grad_norm": 0.07310749590396881, "global_step": 52935, "epoch": 499, "lr": 3.597886360284797e-10} +{"train_loss": 0.015131296589970589, "grad_norm": 0.08226277679204941, "global_step": 52936, "epoch": 499, "lr": 3.4863309313903915e-10} +{"train_loss": 0.013688147999346256, "grad_norm": 0.07139902561903, "global_step": 52937, "epoch": 499, "lr": 3.376532272836297e-10} +{"train_loss": 0.013430045917630196, "grad_norm": 0.07785437256097794, "global_step": 52938, "epoch": 499, "lr": 3.2684903849000695e-10} +{"train_loss": 0.01377265527844429, "grad_norm": 0.12350720167160034, "global_step": 52939, "epoch": 499, "lr": 3.162205268081309e-10} +{"train_loss": 0.01596606895327568, "grad_norm": 0.07508277893066406, "global_step": 52940, "epoch": 499, "lr": 3.057676922713082e-10} +{"train_loss": 0.011076636612415314, "grad_norm": 0.057589780539274216, "global_step": 52941, "epoch": 499, "lr": 2.954905349128456e-10} +{"train_loss": 0.015499931760132313, "grad_norm": 0.13224759697914124, "global_step": 52942, "epoch": 499, "lr": 2.853890547771521e-10} +{"train_loss": 0.014285372570157051, "grad_norm": 0.09541137516498566, "global_step": 52943, "epoch": 499, "lr": 2.7546325188643197e-10} +{"train_loss": 0.016232063993811607, "grad_norm": 0.07558439671993256, "global_step": 52944, "epoch": 499, "lr": 2.657131262906454e-10} +{"train_loss": 0.014722954481840134, "grad_norm": 0.07641705125570297, "global_step": 52945, "epoch": 499, "lr": 2.5613867801199674e-10} +{"train_loss": 0.013648230582475662, "grad_norm": 0.06967395544052124, "global_step": 52946, "epoch": 499, "lr": 2.4673990708934393e-10} +{"train_loss": 0.01434120163321495, "grad_norm": 0.0988263338804245, "global_step": 52947, "epoch": 499, "lr": 2.3751681356154465e-10} +{"train_loss": 0.01564599573612213, "grad_norm": 0.10522247105836868, "global_step": 52948, "epoch": 499, "lr": 2.2846939744525232e-10} +{"train_loss": 0.013783671893179417, "grad_norm": 0.05751842260360718, "global_step": 52949, "epoch": 499, "lr": 2.1959765879042694e-10} +{"train_loss": 0.0187167227268219, "grad_norm": 0.07718101888895035, "global_step": 52950, "epoch": 499, "lr": 2.1090159761372185e-10} +{"train_loss": 0.012857028283178806, "grad_norm": 0.062213100492954254, "global_step": 52951, "epoch": 499, "lr": 2.023812139539949e-10} +{"train_loss": 0.014867467805743217, "grad_norm": 0.11124538630247116, "global_step": 52952, "epoch": 499, "lr": 1.9403650783900163e-10} +{"train_loss": 0.013948860578238964, "grad_norm": 0.073478102684021, "global_step": 52953, "epoch": 499, "lr": 1.8586747929649762e-10} +{"train_loss": 0.014287436380982399, "grad_norm": 0.10663820058107376, "global_step": 52954, "epoch": 499, "lr": 1.7787412835978956e-10} +{"train_loss": 0.01577625796198845, "grad_norm": 0.05670441687107086, "global_step": 52955, "epoch": 499, "lr": 1.7005645505108193e-10} +{"train_loss": 0.015433273278176785, "grad_norm": 0.09358447045087814, "global_step": 52956, "epoch": 499, "lr": 1.6241445939813028e-10} +{"train_loss": 0.015214409679174423, "grad_norm": 0.08337794989347458, "global_step": 52957, "epoch": 499, "lr": 1.549481414342413e-10} +{"train_loss": 0.01607654057443142, "grad_norm": 0.1167629212141037, "global_step": 52958, "epoch": 499, "lr": 1.4765750117606835e-10} +{"train_loss": 0.015661099925637245, "grad_norm": 0.07018245756626129, "global_step": 52959, "epoch": 499, "lr": 1.4054253866246925e-10} +{"train_loss": 0.01577722281217575, "grad_norm": 0.06809970736503601, "global_step": 52960, "epoch": 499, "lr": 1.336032539045462e-10} +{"train_loss": 0.016347497701644897, "grad_norm": 0.14153216779232025, "global_step": 52961, "epoch": 499, "lr": 1.268396469356059e-10} +{"train_loss": 0.014894360676407814, "grad_norm": 0.11051511019468307, "global_step": 52962, "epoch": 499, "lr": 1.202517177723017e-10} +{"train_loss": 0.014783980324864388, "grad_norm": 0.1503383219242096, "global_step": 52963, "epoch": 499, "lr": 1.138394664479403e-10} +{"train_loss": 0.013101656921207905, "grad_norm": 0.09399069100618362, "global_step": 52964, "epoch": 499, "lr": 1.0760289297917503e-10} +{"train_loss": 0.016787957400083542, "grad_norm": 0.06801591068506241, "global_step": 52965, "epoch": 499, "lr": 1.0154199738265923e-10} +{"train_loss": 0.01584670878946781, "grad_norm": 0.09918919205665588, "global_step": 52966, "epoch": 499, "lr": 9.565677968614851e-11} +{"train_loss": 0.013515669852495193, "grad_norm": 0.05823581665754318, "global_step": 52967, "epoch": 499, "lr": 8.994723991184728e-11} +{"train_loss": 0.015552221797406673, "grad_norm": 0.10889938473701477, "global_step": 52968, "epoch": 499, "lr": 8.441337807640892e-11} +{"train_loss": 0.015350380912423134, "grad_norm": 0.0752127394080162, "global_step": 52969, "epoch": 499, "lr": 7.905519420203788e-11} +{"train_loss": 0.013997398316860199, "grad_norm": 0.09913033246994019, "global_step": 52970, "epoch": 499, "lr": 7.38726882998364e-11} +{"train_loss": 0.014485559426248074, "grad_norm": 0.10000509768724442, "global_step": 52971, "epoch": 499, "lr": 6.886586039756004e-11} +{"train_loss": 0.014927579089999199, "grad_norm": 0.1306077390909195, "global_step": 52972, "epoch": 499, "lr": 6.403471050631105e-11} +{"train_loss": 0.015008281916379929, "grad_norm": 0.11893440783023834, "global_step": 52973, "epoch": 499, "lr": 5.937923864829387e-11} +{"train_loss": 0.0176667682826519, "grad_norm": 0.06498827040195465, "global_step": 52974, "epoch": 499, "lr": 5.489944483461074e-11} +{"train_loss": 0.014281690120697021, "grad_norm": 0.06195396929979324, "global_step": 52975, "epoch": 499, "lr": 5.0595329087466114e-11} +{"train_loss": 0.011948633939027786, "grad_norm": 0.05328576639294624, "global_step": 52976, "epoch": 499, "lr": 4.646689141796223e-11} +{"train_loss": 0.014338001608848572, "grad_norm": 0.1349128633737564, "global_step": 52977, "epoch": 499, "lr": 4.251413183720132e-11} +{"train_loss": 0.013198294676840305, "grad_norm": 0.11721865832805634, "global_step": 52978, "epoch": 499, "lr": 3.873705036183673e-11} +{"train_loss": 0.01601238362491131, "grad_norm": 0.11160469055175781, "global_step": 52979, "epoch": 499, "lr": 3.513564700852179e-11} +{"train_loss": 0.015249316580593586, "grad_norm": 0.10039109736680984, "global_step": 52980, "epoch": 499, "lr": 3.1709921788358744e-11} +{"train_loss": 0.015346558764576912, "grad_norm": 0.0603741854429245, "global_step": 52981, "epoch": 499, "lr": 2.8459874712449818e-11} +{"train_loss": 0.01678888499736786, "grad_norm": 0.06452804803848267, "global_step": 52982, "epoch": 499, "lr": 2.5385505791897245e-11} +{"train_loss": 0.014350321143865585, "grad_norm": 0.06455035507678986, "global_step": 52983, "epoch": 499, "lr": 2.248681503225214e-11} +{"train_loss": 0.014385429210960865, "grad_norm": 0.11340080201625824, "global_step": 52984, "epoch": 499, "lr": 1.976380245571896e-11} +{"train_loss": 0.014436442404985428, "grad_norm": 0.07349316775798798, "global_step": 52985, "epoch": 499, "lr": 1.721646806229771e-11} +{"train_loss": 0.012331368401646614, "grad_norm": 0.06491052359342575, "global_step": 52986, "epoch": 499, "lr": 1.4844811868641728e-11} +{"train_loss": 0.015994291752576828, "grad_norm": 0.06525395065546036, "global_step": 52987, "epoch": 499, "lr": 1.2648833869199906e-11} +{"train_loss": 0.012589335441589355, "grad_norm": 0.08156180381774902, "global_step": 52988, "epoch": 499, "lr": 1.06285340861767e-11} +{"train_loss": 0.013942417688667774, "grad_norm": 0.07688703387975693, "global_step": 52989, "epoch": 499, "lr": 8.783912525123227e-12} +{"train_loss": 0.01626778021454811, "grad_norm": 0.06996141374111176, "global_step": 52990, "epoch": 499, "lr": 7.114969186039488e-12} +{"train_loss": 0.015074502676725388, "grad_norm": 0.0814257487654686, "global_step": 52991, "epoch": 499, "lr": 5.621704074476597e-12} +{"train_loss": 0.014557518064975739, "grad_norm": 0.06939258426427841, "global_step": 52992, "epoch": 499, "lr": 4.304117201536784e-12} +{"train_loss": 0.015738749876618385, "grad_norm": 0.0777270570397377, "global_step": 52993, "epoch": 499, "lr": 3.1622085672200486e-12} +{"train_loss": 0.015139339491724968, "grad_norm": 0.09375626593828201, "global_step": 52994, "epoch": 499, "lr": 2.1959781770775068e-12} +{"train_loss": 0.015300719998776913, "grad_norm": 0.05964365974068642, "global_step": 52995, "epoch": 499, "lr": 1.4054260366602735e-12} +{"train_loss": 0.014975311234593391, "grad_norm": 0.10169051587581635, "global_step": 52996, "epoch": 499, "lr": 7.905521459683485e-13} +{"train_loss": 0.017425810918211937, "grad_norm": 0.1294352263212204, "global_step": 52997, "epoch": 499, "lr": 3.513565105528471e-13} +{"train_loss": 0.01650330051779747, "grad_norm": 0.07211646437644958, "global_step": 52998, "epoch": 499, "lr": 8.783913041376935e-14} +{"train_loss": 0.014948459992290667, "grad_norm": 0.10184671729803085, "global_step": 52999, "epoch": 499, "lr": 0.0, "val_loss": 0.01939132623374462} diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/train_rbc_reward_workspace.log b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/train_rbc_reward_workspace.log new file mode 100644 index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391 diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/debug-internal.log b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/debug-internal.log new file mode 100644 index 0000000000000000000000000000000000000000..1cfc9797fb752b17a2d07b65a6616212bd3f4b98 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/debug-internal.log @@ -0,0 +1,16 @@ +{"time":"2026-04-10T14:18:55.241047338+08:00","level":"INFO","msg":"using version","core version":"0.18.6"} +{"time":"2026-04-10T14:18:55.241055452+08:00","level":"INFO","msg":"created symlink","path":"/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug-core.log"} +{"time":"2026-04-10T14:18:55.347087887+08:00","level":"INFO","msg":"created new stream","id":"1qgdq8b7"} +{"time":"2026-04-10T14:18:55.347106369+08:00","level":"INFO","msg":"stream: started","id":"1qgdq8b7"} +{"time":"2026-04-10T14:18:55.347167122+08:00","level":"INFO","msg":"sender: started","stream_id":"1qgdq8b7"} +{"time":"2026-04-10T14:18:55.347153014+08:00","level":"INFO","msg":"writer: Do: started","stream_id":{"value":"1qgdq8b7"}} +{"time":"2026-04-10T14:18:55.347269256+08:00","level":"INFO","msg":"handler: started","stream_id":{"value":"1qgdq8b7"}} +{"time":"2026-04-10T14:18:56.300007778+08:00","level":"INFO","msg":"Starting system monitor"} +{"time":"2026-04-10T14:22:11.169724622+08:00","level":"INFO","msg":"stream: closing","id":"1qgdq8b7"} +{"time":"2026-04-10T14:22:11.169761642+08:00","level":"INFO","msg":"Stopping system monitor"} +{"time":"2026-04-10T14:22:11.170790625+08:00","level":"INFO","msg":"Stopped system monitor"} +{"time":"2026-04-10T14:22:14.98946979+08:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"} +{"time":"2026-04-10T14:22:15.339141849+08:00","level":"INFO","msg":"handler: closed","stream_id":{"value":"1qgdq8b7"}} +{"time":"2026-04-10T14:22:15.33923456+08:00","level":"INFO","msg":"sender: closed","stream_id":"1qgdq8b7"} +{"time":"2026-04-10T14:22:15.339227628+08:00","level":"INFO","msg":"writer: Close: closed","stream_id":{"value":"1qgdq8b7"}} +{"time":"2026-04-10T14:22:15.339394236+08:00","level":"INFO","msg":"stream: closed","id":"1qgdq8b7"} diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/debug.log b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/debug.log new file mode 100644 index 0000000000000000000000000000000000000000..fe6de2740b93be8709b60458cb9423c6b96be001 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/debug.log @@ -0,0 +1,27 @@ +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Current SDK version is 0.18.6 +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Configure stats pid to 234263 +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Loading settings from /home/chyang/.config/wandb/settings +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Loading settings from /tmp2/chyang/workspace/LLM-BC/wandb/settings +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Loading settings from environment variables: {} +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Applying setup settings: {'mode': 'online', '_disable_service': None} +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Inferring run settings from compute environment: {'program_relpath': 'llmbc/workspace/train_rbc_reward_workspace.py', 'program_abspath': '/tmp2/chyang/workspace/LLM-BC/llmbc/workspace/train_rbc_reward_workspace.py', 'program': '/tmp2/chyang/workspace/LLM-BC/./llmbc/workspace/train_rbc_reward_workspace.py'} +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Applying login settings: {} +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:_log_setup():533] Logging user logs to /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug.log +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:_log_setup():534] Logging internal logs to /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug-internal.log +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:init():619] calling init triggers +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:init():626] wandb.init called with sweep_config: {} +config: {'name': 'train_rbc_reward_model', '_target_': 'llmbc.workspace.train_rbc_reward_workspace.TrainRBCRewardWorkspace', 'obs_dim': 39, 'action_dim': 30, 'task_name': 'adroit-hand-relocate-v1', 'exp_name': 'default', 'horizon': 1, 'n_obs_steps': 1, 'n_action_steps': 1, 'n_latency_steps': 0, 'past_action_visible': False, 'reward_model': {'_target_': 'llmbc.policy.rbc_lowdim_policy.RBCLowdimRewardModel', 'input_size': 69, 'hidden_size': [128], 'obs_dim': 39, 'action_dim': 30, 'n_obs_steps': 1, 'n_action_steps': 1}, 'dataloader': {'batch_size': 256, 'num_workers': 0, 'shuffle': True, 'pin_memory': False, 'persistent_workers': False}, 'val_dataloader': {'batch_size': 256, 'num_workers': 0, 'shuffle': True, 'pin_memory': False, 'persistent_workers': False}, 'optimizer': {'_target_': 'torch.optim.Adam', 'lr': 0.0001, 'betas': [0.95, 0.999], 'eps': 1e-08, 'weight_decay': 1e-06}, 'training': {'device': 'cuda', 'seed': 42, 'debug': False, 'resume': True, 'lr_scheduler': 'cosine', 'lr_warmup_steps': 0, 'num_epochs': 500, 'gradient_accumulate_every': 1, 'grad_norm_clip': 0.5, 'checkpoint_every': 10, 'val_every': 1, 'sample_every': 10, 'sample_max_batch': 1000, 'max_train_steps': None, 'max_val_steps': None, 'tqdm_interval_sec': 1.0}, 'logging': {'project': 'rbc_module_training', 'resume': True, 'mode': 'online', 'name': '2026.04.10-14.18.54_train_rbc_reward_model_adroit-hand-relocate-v1', 'tags': ['train_rbc_reward_model', 'adroit-hand-relocate-v1', 'default'], 'id': None, 'group': None}, 'checkpoint': {'topk': {'monitor_key': 'val_loss', 'mode': 'min', 'k': 5, 'format_str': 'epoch={epoch:04d}-val_loss={val_loss:.3f}.ckpt'}, 'save_last_ckpt': True, 'save_last_snapshot': False}, 'multi_run': {'run_dir': 'data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1', 'wandb_name_base': '2026.04.10-14.18.54_train_rbc_reward_model_adroit-hand-relocate-v1'}, 'task': {'name': 'adroit-hand-relocate-v1', 'obs_dim': 39, 'action_dim': 30, 'env_runner': {'_target_': 'llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner', 'env_name': 'llf-adroit-adroit-hand-relocate-v1', 'n_train': 10, 'n_test': 50, 'n_envs': 10, 'max_steps': 150, 'n_obs_steps': 1, 'n_action_steps': 1, 'instruction_type': 'b', 'feedback_type': ['hp', 'hn', 'fp'], 'visual': False, 'discount': 0.99}, 'dataset': {'_target_': 'llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset', 'data_path': 'datasets/adroit-hand-relocate-v1-general-reward.pt', 'data_path2': 'datasets/adroit-hand-relocate-v1-reward.pt', 'horizon': 1, 'pad_before': 0, 'pad_after': 0, 'obs_eef_target': True, 'use_manual_normalizer': False, 'val_ratio': 0.1, 'dummy_normalizer': True}, 'instructor': {'_target_': 'llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor'}}} +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:init():669] starting backend +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:init():673] sending inform_init request +2026-04-10 14:18:55,236 INFO MainThread:234263 [backend.py:_multiprocessing_setup():104] multiprocessing start_methods=fork,spawn,forkserver, using: spawn +2026-04-10 14:18:55,236 INFO MainThread:234263 [wandb_init.py:init():686] backend started and connected +2026-04-10 14:18:55,239 INFO MainThread:234263 [wandb_init.py:init():781] updated telemetry +2026-04-10 14:18:55,256 INFO MainThread:234263 [wandb_init.py:init():814] communicating run to backend with 90.0 second timeout +2026-04-10 14:18:56,297 INFO MainThread:234263 [wandb_init.py:init():867] starting run threads in backend +2026-04-10 14:18:56,455 INFO MainThread:234263 [wandb_run.py:_console_start():2451] atexit reg +2026-04-10 14:18:56,455 INFO MainThread:234263 [wandb_run.py:_redirect():2299] redirect: wrap_raw +2026-04-10 14:18:56,455 INFO MainThread:234263 [wandb_run.py:_redirect():2364] Wrapping output streams. +2026-04-10 14:18:56,455 INFO MainThread:234263 [wandb_run.py:_redirect():2389] Redirects installed. +2026-04-10 14:18:56,457 INFO MainThread:234263 [wandb_init.py:init():911] run started, returning control to user process +2026-04-10 14:18:56,457 INFO MainThread:234263 [wandb_run.py:_config_callback():1389] config_cb None None {'output_dir': '/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1'} +2026-04-10 14:22:11,169 WARNING MsgRouterThr:234263 [router.py:message_loop():75] message_loop has been closed diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/config.yaml b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..a41efa7bc7e96929805bdb639aebf51d68ef47c0 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/config.yaml @@ -0,0 +1,170 @@ +_target_: + value: llmbc.workspace.train_rbc_reward_workspace.TrainRBCRewardWorkspace +_wandb: + value: + cli_version: 0.18.6 + m: [] + python_version: 3.9.23 + t: + "1": + - 1 + - 11 + - 49 + - 50 + - 55 + - 83 + - 105 + "2": + - 1 + - 11 + - 49 + - 50 + - 55 + - 83 + - 105 + "3": + - 13 + - 15 + - 16 + - 23 + - 55 + - 61 + "4": 3.9.23 + "5": 0.18.6 + "6": 4.47.1 + "8": + - 5 + "12": 0.18.6 + "13": linux-x86_64 +action_dim: + value: 30 +checkpoint: + value: + save_last_ckpt: true + save_last_snapshot: false + topk: + format_str: epoch={epoch:04d}-val_loss={val_loss:.3f}.ckpt + k: 5 + mode: min + monitor_key: val_loss +dataloader: + value: + batch_size: 256 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true +exp_name: + value: default +horizon: + value: 1 +logging: + value: + group: null + id: null + mode: online + name: 2026.04.10-14.18.54_train_rbc_reward_model_adroit-hand-relocate-v1 + project: rbc_module_training + resume: true + tags: + - train_rbc_reward_model + - adroit-hand-relocate-v1 + - default +multi_run: + value: + run_dir: data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1 + wandb_name_base: 2026.04.10-14.18.54_train_rbc_reward_model_adroit-hand-relocate-v1 +n_action_steps: + value: 1 +n_latency_steps: + value: 0 +n_obs_steps: + value: 1 +name: + value: train_rbc_reward_model +obs_dim: + value: 39 +optimizer: + value: + _target_: torch.optim.Adam + betas: + - 0.95 + - 0.999 + eps: 1e-08 + lr: 0.0001 + weight_decay: 1e-06 +output_dir: + value: /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1 +past_action_visible: + value: false +reward_model: + value: + _target_: llmbc.policy.rbc_lowdim_policy.RBCLowdimRewardModel + action_dim: 30 + hidden_size: + - 128 + input_size: 69 + n_action_steps: 1 + n_obs_steps: 1 + obs_dim: 39 +task: + value: + action_dim: 30 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-general-reward.pt + data_path2: datasets/adroit-hand-relocate-v1-reward.pt + dummy_normalizer: true + horizon: 1 + obs_eef_target: true + pad_after: 0 + pad_before: 0 + use_manual_normalizer: false + val_ratio: 0.1 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + discount: 0.99 + env_name: llf-adroit-adroit-hand-relocate-v1 + feedback_type: + - hp + - hn + - fp + instruction_type: b + max_steps: 150 + n_action_steps: 1 + n_envs: 10 + n_obs_steps: 1 + n_test: 50 + n_train: 10 + visual: false + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor + name: adroit-hand-relocate-v1 + obs_dim: 39 +task_name: + value: adroit-hand-relocate-v1 +training: + value: + checkpoint_every: 10 + debug: false + device: cuda + grad_norm_clip: 0.5 + gradient_accumulate_every: 1 + lr_scheduler: cosine + lr_warmup_steps: 0 + max_train_steps: null + max_val_steps: null + num_epochs: 500 + resume: true + sample_every: 10 + sample_max_batch: 1000 + seed: 42 + tqdm_interval_sec: 1 + val_every: 1 +val_dataloader: + value: + batch_size: 256 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/output.log b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/output.log new file mode 100644 index 0000000000000000000000000000000000000000..8d058b9f83d0533bec6b8645486acc7a220b28f9 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/output.log @@ -0,0 +1 @@ + diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/wandb-metadata.json b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/wandb-metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..4f7e6d19439b9379cfa68d466501e6eb4791843e --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/wandb-metadata.json @@ -0,0 +1,49 @@ +{ + "os": "Linux-5.15.0-139-generic-x86_64-with-glibc2.31", + "python": "3.9.23", + "startedAt": "2026-04-10T06:18:55.236440Z", + "program": "/tmp2/chyang/workspace/LLM-BC/./llmbc/workspace/train_rbc_reward_workspace.py", + "codePath": "llmbc/workspace/train_rbc_reward_workspace.py", + "git": { + "remote": "https://github.com/CHYang25/LLM-BC.git", + "commit": "91d54e3c1656137d4b85c2bf26f3663614dd84b0" + }, + "email": "chris920325@gmail.com", + "root": "/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1", + "host": "rllab1015", + "username": "chyang", + "executable": "/home/chyang/miniconda3/envs/llm-bc/bin/python3", + "codePathLocal": "llmbc/workspace/train_rbc_reward_workspace.py", + "cpu_count": 20, + "cpu_count_logical": 40, + "gpu": "NVIDIA GeForce RTX 4090", + "gpu_count": 2, + "disk": { + "/": { + "total": "1967317549056", + "used": "1342829142016" + } + }, + "memory": { + "total": "66889547776" + }, + "cpu": { + "count": 20, + "countLogical": 40 + }, + "gpu_nvidia": [ + { + "name": "NVIDIA GeForce RTX 4090", + "memoryTotal": "25757220864", + "cudaCores": 16384, + "architecture": "Ada" + }, + { + "name": "NVIDIA GeForce RTX 4090", + "memoryTotal": "24146608128", + "cudaCores": 16384, + "architecture": "Ada" + } + ], + "cudaVersion": "12.8" +} \ No newline at end of file diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/wandb-summary.json b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/wandb-summary.json new file mode 100644 index 0000000000000000000000000000000000000000..d923a01336b2a1fefaf16fa651f978348f139c9c --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/files/wandb-summary.json @@ -0,0 +1 @@ +{"train_batch_mse_error":0.01336434856057167,"_step":52999,"_runtime":195.933306463,"train_loss":0.014948459992290667,"global_step":52999,"lr":0,"grad_norm":0.10184671729803085,"_timestamp":1.7758021311685236e+09,"_wandb":{"runtime":195},"epoch":499,"val_loss":0.01939132623374462} \ No newline at end of file diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug-core.log b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug-core.log new file mode 100644 index 0000000000000000000000000000000000000000..c26f57139c3355ab032889d458ff419ad86af979 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug-core.log @@ -0,0 +1,14 @@ +{"time":"2026-04-10T14:18:54.59861994+08:00","level":"INFO","msg":"started logging, with flags","port-filename":"/tmp/tmpqflr44pu/port-234263.txt","pid":234263,"debug":false,"disable-analytics":false} +{"time":"2026-04-10T14:18:54.598743322+08:00","level":"INFO","msg":"FeatureState","shutdownOnParentExitEnabled":false} +{"time":"2026-04-10T14:18:54.60083844+08:00","level":"INFO","msg":"Will exit if parent process dies.","ppid":234263} +{"time":"2026-04-10T14:18:54.600789793+08:00","level":"INFO","msg":"server is running","addr":{"IP":"127.0.0.1","Port":36403,"Zone":""}} +{"time":"2026-04-10T14:18:54.778119801+08:00","level":"INFO","msg":"connection: ManageConnectionData: new connection created","id":"127.0.0.1:37858"} +{"time":"2026-04-10T14:18:55.240802078+08:00","level":"INFO","msg":"handleInformInit: received","streamId":"1qgdq8b7","id":"127.0.0.1:37858"} +{"time":"2026-04-10T14:18:55.347109517+08:00","level":"INFO","msg":"handleInformInit: stream started","streamId":"1qgdq8b7","id":"127.0.0.1:37858"} +{"time":"2026-04-10T14:22:11.169648712+08:00","level":"INFO","msg":"handleInformTeardown: server teardown initiated","id":"127.0.0.1:37858"} +{"time":"2026-04-10T14:22:11.169734383+08:00","level":"INFO","msg":"server is shutting down"} +{"time":"2026-04-10T14:22:11.169726788+08:00","level":"INFO","msg":"connection: Close: initiating connection closure","id":"127.0.0.1:37858"} +{"time":"2026-04-10T14:22:11.169887799+08:00","level":"INFO","msg":"connection: Close: connection successfully closed","id":"127.0.0.1:37858"} +{"time":"2026-04-10T14:22:15.339508523+08:00","level":"INFO","msg":"handleInformTeardown: server shutdown complete","id":"127.0.0.1:37858"} +{"time":"2026-04-10T14:22:15.339553802+08:00","level":"INFO","msg":"connection: ManageConnectionData: connection closed","id":"127.0.0.1:37858"} +{"time":"2026-04-10T14:22:15.339588642+08:00","level":"INFO","msg":"server is closed"} diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug-internal.log b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug-internal.log new file mode 100644 index 0000000000000000000000000000000000000000..1cfc9797fb752b17a2d07b65a6616212bd3f4b98 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug-internal.log @@ -0,0 +1,16 @@ +{"time":"2026-04-10T14:18:55.241047338+08:00","level":"INFO","msg":"using version","core version":"0.18.6"} +{"time":"2026-04-10T14:18:55.241055452+08:00","level":"INFO","msg":"created symlink","path":"/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug-core.log"} +{"time":"2026-04-10T14:18:55.347087887+08:00","level":"INFO","msg":"created new stream","id":"1qgdq8b7"} +{"time":"2026-04-10T14:18:55.347106369+08:00","level":"INFO","msg":"stream: started","id":"1qgdq8b7"} +{"time":"2026-04-10T14:18:55.347167122+08:00","level":"INFO","msg":"sender: started","stream_id":"1qgdq8b7"} +{"time":"2026-04-10T14:18:55.347153014+08:00","level":"INFO","msg":"writer: Do: started","stream_id":{"value":"1qgdq8b7"}} +{"time":"2026-04-10T14:18:55.347269256+08:00","level":"INFO","msg":"handler: started","stream_id":{"value":"1qgdq8b7"}} +{"time":"2026-04-10T14:18:56.300007778+08:00","level":"INFO","msg":"Starting system monitor"} +{"time":"2026-04-10T14:22:11.169724622+08:00","level":"INFO","msg":"stream: closing","id":"1qgdq8b7"} +{"time":"2026-04-10T14:22:11.169761642+08:00","level":"INFO","msg":"Stopping system monitor"} +{"time":"2026-04-10T14:22:11.170790625+08:00","level":"INFO","msg":"Stopped system monitor"} +{"time":"2026-04-10T14:22:14.98946979+08:00","level":"INFO","msg":"fileTransfer: Close: file transfer manager closed"} +{"time":"2026-04-10T14:22:15.339141849+08:00","level":"INFO","msg":"handler: closed","stream_id":{"value":"1qgdq8b7"}} +{"time":"2026-04-10T14:22:15.33923456+08:00","level":"INFO","msg":"sender: closed","stream_id":"1qgdq8b7"} +{"time":"2026-04-10T14:22:15.339227628+08:00","level":"INFO","msg":"writer: Close: closed","stream_id":{"value":"1qgdq8b7"}} +{"time":"2026-04-10T14:22:15.339394236+08:00","level":"INFO","msg":"stream: closed","id":"1qgdq8b7"} diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug.log b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug.log new file mode 100644 index 0000000000000000000000000000000000000000..fe6de2740b93be8709b60458cb9423c6b96be001 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug.log @@ -0,0 +1,27 @@ +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Current SDK version is 0.18.6 +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Configure stats pid to 234263 +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Loading settings from /home/chyang/.config/wandb/settings +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Loading settings from /tmp2/chyang/workspace/LLM-BC/wandb/settings +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Loading settings from environment variables: {} +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Applying setup settings: {'mode': 'online', '_disable_service': None} +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Inferring run settings from compute environment: {'program_relpath': 'llmbc/workspace/train_rbc_reward_workspace.py', 'program_abspath': '/tmp2/chyang/workspace/LLM-BC/llmbc/workspace/train_rbc_reward_workspace.py', 'program': '/tmp2/chyang/workspace/LLM-BC/./llmbc/workspace/train_rbc_reward_workspace.py'} +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_setup.py:_flush():79] Applying login settings: {} +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:_log_setup():533] Logging user logs to /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug.log +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:_log_setup():534] Logging internal logs to /tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/logs/debug-internal.log +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:init():619] calling init triggers +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:init():626] wandb.init called with sweep_config: {} +config: {'name': 'train_rbc_reward_model', '_target_': 'llmbc.workspace.train_rbc_reward_workspace.TrainRBCRewardWorkspace', 'obs_dim': 39, 'action_dim': 30, 'task_name': 'adroit-hand-relocate-v1', 'exp_name': 'default', 'horizon': 1, 'n_obs_steps': 1, 'n_action_steps': 1, 'n_latency_steps': 0, 'past_action_visible': False, 'reward_model': {'_target_': 'llmbc.policy.rbc_lowdim_policy.RBCLowdimRewardModel', 'input_size': 69, 'hidden_size': [128], 'obs_dim': 39, 'action_dim': 30, 'n_obs_steps': 1, 'n_action_steps': 1}, 'dataloader': {'batch_size': 256, 'num_workers': 0, 'shuffle': True, 'pin_memory': False, 'persistent_workers': False}, 'val_dataloader': {'batch_size': 256, 'num_workers': 0, 'shuffle': True, 'pin_memory': False, 'persistent_workers': False}, 'optimizer': {'_target_': 'torch.optim.Adam', 'lr': 0.0001, 'betas': [0.95, 0.999], 'eps': 1e-08, 'weight_decay': 1e-06}, 'training': {'device': 'cuda', 'seed': 42, 'debug': False, 'resume': True, 'lr_scheduler': 'cosine', 'lr_warmup_steps': 0, 'num_epochs': 500, 'gradient_accumulate_every': 1, 'grad_norm_clip': 0.5, 'checkpoint_every': 10, 'val_every': 1, 'sample_every': 10, 'sample_max_batch': 1000, 'max_train_steps': None, 'max_val_steps': None, 'tqdm_interval_sec': 1.0}, 'logging': {'project': 'rbc_module_training', 'resume': True, 'mode': 'online', 'name': '2026.04.10-14.18.54_train_rbc_reward_model_adroit-hand-relocate-v1', 'tags': ['train_rbc_reward_model', 'adroit-hand-relocate-v1', 'default'], 'id': None, 'group': None}, 'checkpoint': {'topk': {'monitor_key': 'val_loss', 'mode': 'min', 'k': 5, 'format_str': 'epoch={epoch:04d}-val_loss={val_loss:.3f}.ckpt'}, 'save_last_ckpt': True, 'save_last_snapshot': False}, 'multi_run': {'run_dir': 'data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1', 'wandb_name_base': '2026.04.10-14.18.54_train_rbc_reward_model_adroit-hand-relocate-v1'}, 'task': {'name': 'adroit-hand-relocate-v1', 'obs_dim': 39, 'action_dim': 30, 'env_runner': {'_target_': 'llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner', 'env_name': 'llf-adroit-adroit-hand-relocate-v1', 'n_train': 10, 'n_test': 50, 'n_envs': 10, 'max_steps': 150, 'n_obs_steps': 1, 'n_action_steps': 1, 'instruction_type': 'b', 'feedback_type': ['hp', 'hn', 'fp'], 'visual': False, 'discount': 0.99}, 'dataset': {'_target_': 'llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset', 'data_path': 'datasets/adroit-hand-relocate-v1-general-reward.pt', 'data_path2': 'datasets/adroit-hand-relocate-v1-reward.pt', 'horizon': 1, 'pad_before': 0, 'pad_after': 0, 'obs_eef_target': True, 'use_manual_normalizer': False, 'val_ratio': 0.1, 'dummy_normalizer': True}, 'instructor': {'_target_': 'llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor'}}} +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:init():669] starting backend +2026-04-10 14:18:55,235 INFO MainThread:234263 [wandb_init.py:init():673] sending inform_init request +2026-04-10 14:18:55,236 INFO MainThread:234263 [backend.py:_multiprocessing_setup():104] multiprocessing start_methods=fork,spawn,forkserver, using: spawn +2026-04-10 14:18:55,236 INFO MainThread:234263 [wandb_init.py:init():686] backend started and connected +2026-04-10 14:18:55,239 INFO MainThread:234263 [wandb_init.py:init():781] updated telemetry +2026-04-10 14:18:55,256 INFO MainThread:234263 [wandb_init.py:init():814] communicating run to backend with 90.0 second timeout +2026-04-10 14:18:56,297 INFO MainThread:234263 [wandb_init.py:init():867] starting run threads in backend +2026-04-10 14:18:56,455 INFO MainThread:234263 [wandb_run.py:_console_start():2451] atexit reg +2026-04-10 14:18:56,455 INFO MainThread:234263 [wandb_run.py:_redirect():2299] redirect: wrap_raw +2026-04-10 14:18:56,455 INFO MainThread:234263 [wandb_run.py:_redirect():2364] Wrapping output streams. +2026-04-10 14:18:56,455 INFO MainThread:234263 [wandb_run.py:_redirect():2389] Redirects installed. +2026-04-10 14:18:56,457 INFO MainThread:234263 [wandb_init.py:init():911] run started, returning control to user process +2026-04-10 14:18:56,457 INFO MainThread:234263 [wandb_run.py:_config_callback():1389] config_cb None None {'output_dir': '/tmp2/chyang/workspace/LLM-BC/data/outputs/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1'} +2026-04-10 14:22:11,169 WARNING MsgRouterThr:234263 [router.py:message_loop():75] message_loop has been closed diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/run-1qgdq8b7.wandb b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/run-1qgdq8b7.wandb new file mode 100644 index 0000000000000000000000000000000000000000..faf40f31e3653a5328f46070925643bab42762c8 --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/run-20260410_141855-1qgdq8b7/run-1qgdq8b7.wandb @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7158a5e9d7b738817bb3013be3b122de425becfb808850308e23a1230da88ef4 +size 28101580 diff --git a/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/wandb-resume.json b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/wandb-resume.json new file mode 100644 index 0000000000000000000000000000000000000000..7c04590e694739776063414cbfee5f202dd0eb3f --- /dev/null +++ b/2026.04.10/14.18.53_train_rbc_reward_model_adroit-hand-relocate-v1/wandb/wandb-resume.json @@ -0,0 +1 @@ +{"run_id": "1qgdq8b7"} \ No newline at end of file diff --git a/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml b/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..b35e693795765afcf8d1ce2baf706f0406104500 --- /dev/null +++ b/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml @@ -0,0 +1,163 @@ +name: train_llmbc_lowdim +_target_: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +model_name: ${llm.name} +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +llm_orig_expert_feedback: true +llm_do_sample: false +policy: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + task_id: ${task_name} + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + cfg: ${llm} + obs_dim: ${task.obs_dim} + action_dim: ${task.action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + loss_bc_weight: 1.0 + loss_llm_weight: 0.01 + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + normalize_llm_loss: true +dataloader: + batch_size: 32 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 32 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.001 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 43 + debug: false + resume: false + lr_scheduler: constant + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 4 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training-mix + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: adroit-hand-relocate-v1 + obs_dim: 39 + action_dim: 30 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + env_name: llf-adroit-adroit-hand-relocate-v1 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 150 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-mix(csp100).pt + data_path2: datasets/adroit-hand-relocate-v1.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.1 + dummy_normalizer: true + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor +llm: + name: HuggingFaceTB/SmolLM2-135M-Instruct + model_name: SmolLM2-135M-Instruct + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + use_quantization: false + use_joint_mlp_projector: true + llm_mode: ete-finetuned + finetune_mode: orig + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + max_length: 100 + lora_config: + r: 32 + lora_alpha: 64 + lora_dropout: 0.05 + bias: none + task_type: CAUSAL_LM + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + hydra: + job: + override_dirname: ${model_name} + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${model_name} diff --git a/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml b/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..e1debd61b6e2232fab8b4cd59d148ec10b3efce5 --- /dev/null +++ b/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml @@ -0,0 +1,155 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: + - training.seed=43 + job: + name: train + chdir: null + override_dirname: training.seed=43 + id: ??? + num: ??? + config_name: llmbc_adroit-hand-relocate-v1_mix(csp100).yaml + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/config/mix_data + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1 + choices: + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml b/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..ebf61aeca24956094799b4a1cab9370df040027f --- /dev/null +++ b/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml @@ -0,0 +1 @@ +- training.seed=43 diff --git a/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log b/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log new file mode 100644 index 0000000000000000000000000000000000000000..b611b4490af3a39ad614e5de196475206a3c4f33 --- /dev/null +++ b/2026.04.10/14.20.18_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log @@ -0,0 +1,6 @@ +[2026-04-10 14:20:19,500][numexpr.utils][INFO] - Note: detected 144 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable. +[2026-04-10 14:20:19,500][numexpr.utils][INFO] - Note: NumExpr detected 144 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 16. +[2026-04-10 14:20:19,501][numexpr.utils][INFO] - NumExpr defaulting to 16 threads. +[2026-04-10 14:20:23,212][datasets][INFO] - PyTorch version 2.2.2 available. +[2026-04-10 14:20:23,213][datasets][INFO] - TensorFlow version 2.15.1 available. +[2026-04-10 14:20:23,213][datasets][INFO] - JAX version 0.4.30 available. diff --git a/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml b/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1b5b5182460a2847e1d1b3692bb97c8ed77d42f0 --- /dev/null +++ b/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml @@ -0,0 +1,163 @@ +name: train_llmbc_lowdim +_target_: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +model_name: ${llm.name} +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +llm_orig_expert_feedback: true +llm_do_sample: false +policy: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + task_id: ${task_name} + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + cfg: ${llm} + obs_dim: ${task.obs_dim} + action_dim: ${task.action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + loss_bc_weight: 1.0 + loss_llm_weight: 0.01 + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + normalize_llm_loss: true +dataloader: + batch_size: 32 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 32 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.001 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: constant + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 4 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training-mix + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: adroit-hand-relocate-v1 + obs_dim: 39 + action_dim: 30 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + env_name: llf-adroit-adroit-hand-relocate-v1 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 150 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-mix(csp100).pt + data_path2: datasets/adroit-hand-relocate-v1.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.1 + dummy_normalizer: true + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor +llm: + name: HuggingFaceTB/SmolLM2-135M-Instruct + model_name: SmolLM2-135M-Instruct + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + use_quantization: false + use_joint_mlp_projector: true + llm_mode: ete-finetuned + finetune_mode: orig + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + max_length: 100 + lora_config: + r: 32 + lora_alpha: 64 + lora_dropout: 0.05 + bias: none + task_type: CAUSAL_LM + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + hydra: + job: + override_dirname: ${model_name} + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${model_name} diff --git a/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml b/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..939cd1bd92e5b3c61a3749967edd3c83899dfb4f --- /dev/null +++ b/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml @@ -0,0 +1,154 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: llmbc_adroit-hand-relocate-v1_mix(csp100).yaml + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/config/mix_data + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1 + choices: + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml b/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log b/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log new file mode 100644 index 0000000000000000000000000000000000000000..64b94df4a347abee1f32baf90141fddf7e3881b3 --- /dev/null +++ b/2026.04.10/14.20.19_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log @@ -0,0 +1,3 @@ +[2026-04-10 14:20:20,916][numexpr.utils][INFO] - Note: detected 144 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable. +[2026-04-10 14:20:20,916][numexpr.utils][INFO] - Note: NumExpr detected 144 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 16. +[2026-04-10 14:20:20,916][numexpr.utils][INFO] - NumExpr defaulting to 16 threads. diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..1b5b5182460a2847e1d1b3692bb97c8ed77d42f0 --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/config.yaml @@ -0,0 +1,163 @@ +name: train_llmbc_lowdim +_target_: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +obs_dim: ${task.obs_dim} +action_dim: ${task.action_dim} +task_name: ${task.name} +exp_name: default +model_name: ${llm.name} +horizon: 1 +n_obs_steps: 1 +n_action_steps: 1 +n_latency_steps: 0 +past_action_visible: false +llm_orig_expert_feedback: true +llm_do_sample: false +policy: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + input_size: ${eval:'${n_obs_steps}*${obs_dim}'} + hidden_size: + - 256 + - 256 + output_size: ${eval:'${n_action_steps}*${action_dim}'} + activation: relu + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + obs_dim: ${obs_dim} + action_dim: ${action_dim} + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + task_id: ${task_name} + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + cfg: ${llm} + obs_dim: ${task.obs_dim} + action_dim: ${task.action_dim} + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + loss_bc_weight: 1.0 + loss_llm_weight: 0.01 + horizon: ${horizon} + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + normalize_llm_loss: true +dataloader: + batch_size: 32 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +val_dataloader: + batch_size: 32 + num_workers: 0 + shuffle: true + pin_memory: false + persistent_workers: false +optimizer: + _target_: torch.optim.AdamW + lr: 0.001 + betas: + - 0.95 + - 0.999 + eps: 1.0e-08 + weight_decay: 1.0e-06 +training: + device: cuda:0 + seed: 42 + debug: false + resume: false + lr_scheduler: constant + lr_warmup_steps: 10 + num_epochs: 1001 + gradient_accumulate_every: 4 + grad_norm_clip: 0.5 + rollout_every: 5 + checkpoint_every: 5 + val_every: 1 + sample_every: 5 + sample_max_batch: 128 + max_train_steps: null + max_val_steps: null + tqdm_interval_sec: 1.0 +logging: + project: ${task.name}-training-mix + resume: true + mode: online + name: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} + tags: + - ${name} + - ${task_name} + - ${exp_name} + id: null + group: null +checkpoint: + topk: + monitor_key: test_success_rate + mode: max + k: 5 + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + save_last_ckpt: true + save_last_snapshot: false +multi_run: + run_dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + wandb_name_base: ${now:%Y.%m.%d-%H.%M.%S}_${name}_${task_name} +task: + name: adroit-hand-relocate-v1 + obs_dim: 39 + action_dim: 30 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + env_name: llf-adroit-adroit-hand-relocate-v1 + n_train: 10 + n_test: 50 + n_envs: 10 + max_steps: 150 + n_obs_steps: ${n_obs_steps} + n_action_steps: ${n_action_steps} + instruction_type: b + feedback_type: + - hp + - hn + - fp + visual: false + discount: 0.99 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-mix(csp100).pt + data_path2: datasets/adroit-hand-relocate-v1.pt + horizon: ${horizon} + pad_before: ${eval:'${n_obs_steps}-1'} + pad_after: ${eval:'${n_action_steps}-1'} + obs_eef_target: true + use_manual_normalizer: false + val_ratio: 0.1 + dummy_normalizer: true + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor +llm: + name: HuggingFaceTB/SmolLM2-135M-Instruct + model_name: SmolLM2-135M-Instruct + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + use_quantization: false + use_joint_mlp_projector: true + llm_mode: ete-finetuned + finetune_mode: orig + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + max_length: 100 + lora_config: + r: 32 + lora_alpha: 64 + lora_dropout: 0.05 + bias: none + task_type: CAUSAL_LM + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + hydra: + job: + override_dirname: ${model_name} + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${model_name} diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml new file mode 100644 index 0000000000000000000000000000000000000000..133404b3e60cd48537822b36b7c2cf9c3473a58a --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/hydra.yaml @@ -0,0 +1,154 @@ +hydra: + run: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + sweep: + dir: data/outputs/${now:%Y.%m.%d}/${now:%H.%M.%S}_${name}_${task_name} + subdir: ${hydra.job.num} + launcher: + _target_: hydra._internal.core_plugins.basic_launcher.BasicLauncher + sweeper: + _target_: hydra._internal.core_plugins.basic_sweeper.BasicSweeper + max_batch_size: null + params: null + help: + app_name: ${hydra.job.name} + header: '${hydra.help.app_name} is powered by Hydra. + + ' + footer: 'Powered by Hydra (https://hydra.cc) + + Use --hydra-help to view Hydra specific help + + ' + template: '${hydra.help.header} + + == Configuration groups == + + Compose your configuration from those groups (group=option) + + + $APP_CONFIG_GROUPS + + + == Config == + + Override anything in the config (foo.bar=value) + + + $CONFIG + + + ${hydra.help.footer} + + ' + hydra_help: + template: 'Hydra (${hydra.runtime.version}) + + See https://hydra.cc for more info. + + + == Flags == + + $FLAGS_HELP + + + == Configuration groups == + + Compose your configuration from those groups (For example, append hydra/job_logging=disabled + to command line) + + + $HYDRA_CONFIG_GROUPS + + + Use ''--cfg hydra'' to Show the Hydra config. + + ' + hydra_help: ??? + hydra_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][HYDRA] %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + root: + level: INFO + handlers: + - console + loggers: + logging_example: + level: DEBUG + disable_existing_loggers: false + job_logging: + version: 1 + formatters: + simple: + format: '[%(asctime)s][%(name)s][%(levelname)s] - %(message)s' + handlers: + console: + class: logging.StreamHandler + formatter: simple + stream: ext://sys.stdout + file: + class: logging.FileHandler + formatter: simple + filename: ${hydra.runtime.output_dir}/${hydra.job.name}.log + root: + level: INFO + handlers: + - console + - file + disable_existing_loggers: false + env: {} + mode: RUN + searchpath: [] + callbacks: {} + output_subdir: .hydra + overrides: + hydra: + - hydra.mode=RUN + task: [] + job: + name: train + chdir: null + override_dirname: '' + id: ??? + num: ??? + config_name: llmbc_adroit-hand-relocate-v1_mix(csp100).yaml + env_set: {} + env_copy: [] + config: + override_dirname: + kv_sep: '=' + item_sep: ',' + exclude_keys: [] + runtime: + version: 1.2.0 + version_base: '1.2' + cwd: /workspace/LLM-BC + config_sources: + - path: hydra.conf + schema: pkg + provider: hydra + - path: /workspace/LLM-BC/config/mix_data + schema: file + provider: main + - path: '' + schema: structured + provider: schema + output_dir: /workspace/LLM-BC/data/outputs/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1 + choices: + hydra/env: default + hydra/callbacks: null + hydra/job_logging: default + hydra/hydra_logging: default + hydra/hydra_help: default + hydra/help: default + hydra/sweeper: basic + hydra/launcher: basic + hydra/output: default + verbose: false diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml new file mode 100644 index 0000000000000000000000000000000000000000..fe51488c7066f6687ef680d6bfaa4f7768ef205c --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/.hydra/overrides.yaml @@ -0,0 +1 @@ +[] diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0115-test_success_rate=0.320.ckpt b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0115-test_success_rate=0.320.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..57567282fe2ce9cf3745395244ada0233eb3c74c --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0115-test_success_rate=0.320.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1d649ff2982ff96cb7901075a7ea2e41fb9f98bf23a6be825fa18bcc24b50481 +size 1036936 diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0210-test_success_rate=0.340.ckpt b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0210-test_success_rate=0.340.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..5d9852d717a09fbe26adf823e86ff2f52d7e71a3 --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0210-test_success_rate=0.340.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2394bf989b6800516b8ce6630e841bb6fd252a41c1d450fcfe21914b56fd92d6 +size 1036936 diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0285-test_success_rate=0.320.ckpt b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0285-test_success_rate=0.320.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..e0bce8b70a449836dfb872dad18cb48125a458c0 --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0285-test_success_rate=0.320.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:63f2a28bfa4589667bbd0328a3d368126576a3055152aeb5cd406ef1f28a12bb +size 1036936 diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0595-test_success_rate=0.280.ckpt b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0595-test_success_rate=0.280.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..be2ce57dc29debe35ce3d578f078d5320032a283 --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0595-test_success_rate=0.280.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8d3ebff43f0771c5e8ab662649a2ae72ca3bf960ea5996fced283bb11f5ce945 +size 1036936 diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0755-test_success_rate=0.300.ckpt b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0755-test_success_rate=0.300.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..d31ff73ec0ad110afe8de3d7e7d5bdab8847c083 --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/epoch=0755-test_success_rate=0.300.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4bd2b15f7b3c8d1db79315da8f7df4e59a8037b96615af4e47a317c94563ad9d +size 1036936 diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/latest.ckpt b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/latest.ckpt new file mode 100644 index 0000000000000000000000000000000000000000..8688224db5d968fa71f9828d3f9ed9598f7cabb9 --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/checkpoints/latest.ckpt @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:55340d364a595bff52d1a094e4cfd36d1a7fa4a4b938b8b7e4e8394ff97d262c +size 1036936 diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt new file mode 100644 index 0000000000000000000000000000000000000000..660eb22b6fa77f9e1c838dd7c4e372cf06e194a8 --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/logs.json.txt @@ -0,0 +1,36036 @@ +{"train_loss": 0.33417993783950806, "train_loss_bc": 0.33263349533081055, "train_loss_llm": 0.1546449065208435, "grad_norm": 0.07081812620162964, "global_step": 0, "epoch": 0, "lr": 0.001} +{"train_loss": 0.29666706919670105, "train_loss_bc": 0.29505738615989685, "train_loss_llm": 0.16096952557563782, "grad_norm": 0.06832075119018555, "global_step": 1, "epoch": 0, "lr": 0.001} +{"train_loss": 0.30729377269744873, "train_loss_bc": 0.3064597249031067, "train_loss_llm": 0.083404541015625, "grad_norm": 0.13414114713668823, "global_step": 2, "epoch": 0, "lr": 0.001} +{"train_loss": 0.30370891094207764, "train_loss_bc": 0.30268386006355286, "train_loss_llm": 0.10250619053840637, "grad_norm": 0.1992519050836563, "global_step": 3, "epoch": 0, "lr": 0.001} +{"train_loss": 0.30288198590278625, "train_loss_bc": 0.30164504051208496, "train_loss_llm": 0.12369495630264282, "grad_norm": 0.2660602033138275, "global_step": 4, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2631026804447174, "train_loss_bc": 0.2620241940021515, "train_loss_llm": 0.10784879326820374, "grad_norm": 0.06255818903446198, "global_step": 5, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2727557420730591, "train_loss_bc": 0.2717650234699249, "train_loss_llm": 0.09907281398773193, "grad_norm": 0.1258833259344101, "global_step": 6, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2919948101043701, "train_loss_bc": 0.2908846437931061, "train_loss_llm": 0.11101800203323364, "grad_norm": 0.19144868850708008, "global_step": 7, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2359585464000702, "train_loss_bc": 0.23494061827659607, "train_loss_llm": 0.10179319977760315, "grad_norm": 0.25360357761383057, "global_step": 8, "epoch": 0, "lr": 0.001} +{"train_loss": 0.22962844371795654, "train_loss_bc": 0.2286834865808487, "train_loss_llm": 0.09449610114097595, "grad_norm": 0.06701260060071945, "global_step": 9, "epoch": 0, "lr": 0.001} +{"train_loss": 0.25155526399612427, "train_loss_bc": 0.25079676508903503, "train_loss_llm": 0.07584938406944275, "grad_norm": 0.12355737388134003, "global_step": 10, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2393786758184433, "train_loss_bc": 0.23857155442237854, "train_loss_llm": 0.08071184158325195, "grad_norm": 0.18559664487838745, "global_step": 11, "epoch": 0, "lr": 0.001} +{"train_loss": 0.24244333803653717, "train_loss_bc": 0.24164727330207825, "train_loss_llm": 0.0796070396900177, "grad_norm": 0.2476871907711029, "global_step": 12, "epoch": 0, "lr": 0.001} +{"train_loss": 0.22287960350513458, "train_loss_bc": 0.2221919596195221, "train_loss_llm": 0.06876447796821594, "grad_norm": 0.05741574242711067, "global_step": 13, "epoch": 0, "lr": 0.001} +{"train_loss": 0.22776256501674652, "train_loss_bc": 0.22730913758277893, "train_loss_llm": 0.04534211754798889, "grad_norm": 0.11210285127162933, "global_step": 14, "epoch": 0, "lr": 0.001} +{"train_loss": 0.21483373641967773, "train_loss_bc": 0.21430127322673798, "train_loss_llm": 0.05324661731719971, "grad_norm": 0.16947202384471893, "global_step": 15, "epoch": 0, "lr": 0.001} +{"train_loss": 0.20982077717781067, "train_loss_bc": 0.20917512476444244, "train_loss_llm": 0.06456509232521057, "grad_norm": 0.22861050069332123, "global_step": 16, "epoch": 0, "lr": 0.001} +{"train_loss": 0.20063169300556183, "train_loss_bc": 0.19993287324905396, "train_loss_llm": 0.06988167762756348, "grad_norm": 0.0562383309006691, "global_step": 17, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2100711464881897, "train_loss_bc": 0.2095283567905426, "train_loss_llm": 0.054279059171676636, "grad_norm": 0.11120761185884476, "global_step": 18, "epoch": 0, "lr": 0.001} +{"train_loss": 0.18883870542049408, "train_loss_bc": 0.1882684826850891, "train_loss_llm": 0.05702179670333862, "grad_norm": 0.16150981187820435, "global_step": 19, "epoch": 0, "lr": 0.001} +{"train_loss": 0.2231723517179489, "train_loss_bc": 0.22266227006912231, "train_loss_llm": 0.05100753903388977, "grad_norm": 0.21718935668468475, "global_step": 20, "epoch": 0, "lr": 0.001} +{"train_loss": 0.17977574467658997, "train_loss_bc": 0.1794614940881729, "train_loss_llm": 0.03142547607421875, "grad_norm": 0.054592687636613846, "global_step": 21, "epoch": 0, "lr": 0.001} +{"train_loss": 0.18479956686496735, "train_loss_bc": 0.18419340252876282, "train_loss_llm": 0.06061708927154541, "grad_norm": 0.10479485988616943, "global_step": 22, "epoch": 0, "lr": 0.001} +{"train_loss": 0.1977708488702774, "train_loss_bc": 0.19731594622135162, "train_loss_llm": 0.045490801334381104, "grad_norm": 0.15253210067749023, "global_step": 23, "epoch": 0, "lr": 0.001} +{"train_loss": 0.16539078950881958, "train_loss_bc": 0.16495342552661896, "train_loss_llm": 0.043737053871154785, "grad_norm": 0.2066734880208969, "global_step": 24, "epoch": 0, "lr": 0.001} +{"train_loss": 0.1792425811290741, "train_loss_bc": 0.17882947623729706, "train_loss_llm": 0.041310638189315796, "grad_norm": 0.04754088819026947, "global_step": 25, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19725419580936432, "train_loss_bc": 0.19698888063430786, "train_loss_llm": 0.026531964540481567, "grad_norm": 0.09326529502868652, "global_step": 26, "epoch": 0, "lr": 0.001} +{"train_loss": 0.1754775494337082, "train_loss_bc": 0.17486830055713654, "train_loss_llm": 0.06092536449432373, "grad_norm": 0.1383977234363556, "global_step": 27, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19483931362628937, "train_loss_bc": 0.1944057196378708, "train_loss_llm": 0.043359577655792236, "grad_norm": 0.18667197227478027, "global_step": 28, "epoch": 0, "lr": 0.001} +{"train_loss": 0.19440998136997223, "train_loss_bc": 0.19396266341209412, "train_loss_llm": 0.04473203420639038, "grad_norm": 0.05193614959716797, "global_step": 29, "epoch": 0, "lr": 0.001} +{"train_loss": 0.1753966361284256, "train_loss_bc": 0.17480593919754028, "train_loss_llm": 0.05906960368156433, "grad_norm": 0.08610650897026062, "global_step": 30, "epoch": 0, "lr": 0.001} +{"train_loss": 0.16164803504943848, "train_loss_bc": 0.16104815900325775, "train_loss_llm": 0.05998712778091431, "grad_norm": 0.12085042148828506, "global_step": 31, "epoch": 0, "lr": 0.001} +{"train_loss": 0.14899751543998718, "train_loss_bc": 0.14848819375038147, "train_loss_llm": 0.050931453704833984, "grad_norm": 0.15778455138206482, "global_step": 32, "epoch": 0, "lr": 0.001} +{"train_loss": 0.14965128898620605, "train_loss_bc": 0.14899152517318726, "train_loss_llm": 0.06597697734832764, "grad_norm": 0.04335929825901985, "global_step": 33, "epoch": 0, "lr": 0.001} +{"train_loss": 0.1576658934354782, "train_loss_bc": 0.15677198767662048, "train_loss_llm": 0.08939129114151001, "grad_norm": 0.08594752848148346, "global_step": 34, "epoch": 0, "lr": 0.001} +{"train_loss": 0.21988501730892393, "train_loss_bc": 0.18346592783927917, "train_loss_llm": 0.05150008201599121, "grad_norm": 0.11817123740911484, "global_step": 35, "epoch": 0, "lr": 0.001, "train/cumulative_reward": -2.348049533105913, "train/mean_score": -0.026168878280877345, "train/success_rate": 0.0, "test/cumulative_reward": -2.113905910555222, "test/mean_score": -0.022711834880611004, "test/success_rate": 0.0, "val_loss": 0.15491078794002533, "train_action_mse_error": 0.16912131011486053} +{"train_loss": 0.15050576627254486, "train_loss_bc": 0.14998802542686462, "train_loss_llm": 0.05177411437034607, "grad_norm": 0.15454189479351044, "global_step": 36, "epoch": 1, "lr": 0.001} +{"train_loss": 0.14738981425762177, "train_loss_bc": 0.14614741504192352, "train_loss_llm": 0.12423941493034363, "grad_norm": 0.05890548974275589, "global_step": 37, "epoch": 1, "lr": 0.001} +{"train_loss": 0.13475607335567474, "train_loss_bc": 0.1336408406496048, "train_loss_llm": 0.1115226149559021, "grad_norm": 0.11099521070718765, "global_step": 38, "epoch": 1, "lr": 0.001} +{"train_loss": 0.18595534563064575, "train_loss_bc": 0.18508371710777283, "train_loss_llm": 0.08716347813606262, "grad_norm": 0.1560196429491043, "global_step": 39, "epoch": 1, "lr": 0.001} +{"train_loss": 0.1684442013502121, "train_loss_bc": 0.1677614003419876, "train_loss_llm": 0.06827977299690247, "grad_norm": 0.20123830437660217, "global_step": 40, "epoch": 1, "lr": 0.001} +{"train_loss": 0.1771121472120285, "train_loss_bc": 0.1762508749961853, "train_loss_llm": 0.08612769842147827, "grad_norm": 0.06296784430742264, "global_step": 41, "epoch": 1, "lr": 0.001} +{"train_loss": 0.15717315673828125, "train_loss_bc": 0.15649446845054626, "train_loss_llm": 0.06786859035491943, "grad_norm": 0.12341085076332092, "global_step": 42, "epoch": 1, "lr": 0.001} +{"train_loss": 0.14304466545581818, "train_loss_bc": 0.14216957986354828, "train_loss_llm": 0.08750823140144348, "grad_norm": 0.1897730976343155, "global_step": 43, "epoch": 1, "lr": 0.001} +{"train_loss": 0.16410033404827118, "train_loss_bc": 0.16288568079471588, "train_loss_llm": 0.12146461009979248, "grad_norm": 0.24817441403865814, "global_step": 44, "epoch": 1, "lr": 0.001} +{"train_loss": 0.14154581725597382, "train_loss_bc": 0.14054988324642181, "train_loss_llm": 0.09959360957145691, "grad_norm": 0.08523659408092499, "global_step": 45, "epoch": 1, "lr": 0.001} +{"train_loss": 0.149897500872612, "train_loss_bc": 0.1489792913198471, "train_loss_llm": 0.09182098507881165, "grad_norm": 0.132100448012352, "global_step": 46, "epoch": 1, "lr": 0.001} +{"train_loss": 0.15656661987304688, "train_loss_bc": 0.15570004284381866, "train_loss_llm": 0.08665776252746582, "grad_norm": 0.19892741739749908, "global_step": 47, "epoch": 1, "lr": 0.001} +{"train_loss": 0.15512318909168243, "train_loss_bc": 0.15390004217624664, "train_loss_llm": 0.12231525778770447, "grad_norm": 0.2637503445148468, "global_step": 48, "epoch": 1, "lr": 0.001} +{"train_loss": 0.1472008377313614, "train_loss_bc": 0.14605821669101715, "train_loss_llm": 0.11426159739494324, "grad_norm": 0.08122682571411133, "global_step": 49, "epoch": 1, "lr": 0.001} +{"train_loss": 0.12306812405586243, "train_loss_bc": 0.12234130501747131, "train_loss_llm": 0.07268175482749939, "grad_norm": 0.1563989818096161, "global_step": 50, "epoch": 1, "lr": 0.001} +{"train_loss": 0.15008535981178284, "train_loss_bc": 0.14926083385944366, "train_loss_llm": 0.08245334029197693, "grad_norm": 0.21952711045742035, "global_step": 51, "epoch": 1, "lr": 0.001} +{"train_loss": 0.1156650111079216, "train_loss_bc": 0.11488819867372513, "train_loss_llm": 0.0776812732219696, "grad_norm": 0.29608362913131714, "global_step": 52, "epoch": 1, "lr": 0.001} +{"train_loss": 0.11812815815210342, "train_loss_bc": 0.11733168363571167, "train_loss_llm": 0.07964709401130676, "grad_norm": 0.08096154779195786, "global_step": 53, "epoch": 1, "lr": 0.001} +{"train_loss": 0.13973397016525269, "train_loss_bc": 0.13921493291854858, "train_loss_llm": 0.051903486251831055, "grad_norm": 0.1329706311225891, "global_step": 54, "epoch": 1, "lr": 0.001} +{"train_loss": 0.1304982602596283, "train_loss_bc": 0.12970970571041107, "train_loss_llm": 0.07885581254959106, "grad_norm": 0.1988958716392517, "global_step": 55, "epoch": 1, "lr": 0.001} +{"train_loss": 0.11107300966978073, "train_loss_bc": 0.11007261276245117, "train_loss_llm": 0.10003992915153503, "grad_norm": 0.27357998490333557, "global_step": 56, "epoch": 1, "lr": 0.001} +{"train_loss": 0.13490548729896545, "train_loss_bc": 0.1344105303287506, "train_loss_llm": 0.049496233463287354, "grad_norm": 0.08058036118745804, "global_step": 57, "epoch": 1, "lr": 0.001} +{"train_loss": 0.10955533385276794, "train_loss_bc": 0.1090720072388649, "train_loss_llm": 0.04833272099494934, "grad_norm": 0.1352105736732483, "global_step": 58, "epoch": 1, "lr": 0.001} +{"train_loss": 0.11791649460792542, "train_loss_bc": 0.11735618114471436, "train_loss_llm": 0.05603170394897461, "grad_norm": 0.19313649833202362, "global_step": 59, "epoch": 1, "lr": 0.001} +{"train_loss": 0.11617564409971237, "train_loss_bc": 0.11578831821680069, "train_loss_llm": 0.03873258829116821, "grad_norm": 0.2501852810382843, "global_step": 60, "epoch": 1, "lr": 0.001} +{"train_loss": 0.11815374344587326, "train_loss_bc": 0.11767500638961792, "train_loss_llm": 0.04787352681159973, "grad_norm": 0.05164386332035065, "global_step": 61, "epoch": 1, "lr": 0.001} +{"train_loss": 0.10448193550109863, "train_loss_bc": 0.10409371554851532, "train_loss_llm": 0.038821905851364136, "grad_norm": 0.0968480110168457, "global_step": 62, "epoch": 1, "lr": 0.001} +{"train_loss": 0.11605336517095566, "train_loss_bc": 0.11589537560939789, "train_loss_llm": 0.015798598527908325, "grad_norm": 0.1500571221113205, "global_step": 63, "epoch": 1, "lr": 0.001} +{"train_loss": 0.10788318514823914, "train_loss_bc": 0.10758447647094727, "train_loss_llm": 0.0298711359500885, "grad_norm": 0.19071431457996368, "global_step": 64, "epoch": 1, "lr": 0.001} +{"train_loss": 0.11839849501848221, "train_loss_bc": 0.11820702999830246, "train_loss_llm": 0.01914629340171814, "grad_norm": 0.038348741829395294, "global_step": 65, "epoch": 1, "lr": 0.001} +{"train_loss": 0.11891594529151917, "train_loss_bc": 0.11870867013931274, "train_loss_llm": 0.02072736620903015, "grad_norm": 0.0766918808221817, "global_step": 66, "epoch": 1, "lr": 0.001} +{"train_loss": 0.10524646937847137, "train_loss_bc": 0.10497375577688217, "train_loss_llm": 0.027271121740341187, "grad_norm": 0.11380691826343536, "global_step": 67, "epoch": 1, "lr": 0.001} +{"train_loss": 0.0957607552409172, "train_loss_bc": 0.09563320130109787, "train_loss_llm": 0.01275557279586792, "grad_norm": 0.15387485921382904, "global_step": 68, "epoch": 1, "lr": 0.001} +{"train_loss": 0.09389801323413849, "train_loss_bc": 0.09374428540468216, "train_loss_llm": 0.015372663736343384, "grad_norm": 0.03617599979043007, "global_step": 69, "epoch": 1, "lr": 0.001} +{"train_loss": 0.09383583068847656, "train_loss_bc": 0.09380079060792923, "train_loss_llm": 0.003504246473312378, "grad_norm": 0.06231410428881645, "global_step": 70, "epoch": 1, "lr": 0.001} +{"train_loss": 0.13136776867840025, "train_loss_bc": 0.11087111383676529, "train_loss_llm": 0.01204991340637207, "grad_norm": 0.10716580599546432, "global_step": 71, "epoch": 1, "lr": 0.001, "val_loss": 0.09548087418079376} +{"train_loss": 0.08514580130577087, "train_loss_bc": 0.0850776731967926, "train_loss_llm": 0.006812483072280884, "grad_norm": 0.13125890493392944, "global_step": 72, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08539284020662308, "train_loss_bc": 0.08532287925481796, "train_loss_llm": 0.006996244192123413, "grad_norm": 0.03279044106602669, "global_step": 73, "epoch": 2, "lr": 0.001} +{"train_loss": 0.0953591987490654, "train_loss_bc": 0.09533640742301941, "train_loss_llm": 0.0022792816162109375, "grad_norm": 0.05935373529791832, "global_step": 74, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08262152969837189, "train_loss_bc": 0.08252564072608948, "train_loss_llm": 0.009588778018951416, "grad_norm": 0.09257499873638153, "global_step": 75, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08216217160224915, "train_loss_bc": 0.0821518525481224, "train_loss_llm": 0.001031637191772461, "grad_norm": 0.1183607280254364, "global_step": 76, "epoch": 2, "lr": 0.001} +{"train_loss": 0.09111765772104263, "train_loss_bc": 0.09107367694377899, "train_loss_llm": 0.004397839307785034, "grad_norm": 0.029349692165851593, "global_step": 77, "epoch": 2, "lr": 0.001} +{"train_loss": 0.0879141315817833, "train_loss_bc": 0.08786826580762863, "train_loss_llm": 0.004586279392242432, "grad_norm": 0.0579628124833107, "global_step": 78, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08750811964273453, "train_loss_bc": 0.08748722821474075, "train_loss_llm": 0.002088993787765503, "grad_norm": 0.08332771062850952, "global_step": 79, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08590230345726013, "train_loss_bc": 0.0858050063252449, "train_loss_llm": 0.00973007082939148, "grad_norm": 0.11693568527698517, "global_step": 80, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08403275907039642, "train_loss_bc": 0.08399887382984161, "train_loss_llm": 0.0033882856369018555, "grad_norm": 0.025199787691235542, "global_step": 81, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08204181492328644, "train_loss_bc": 0.08198491483926773, "train_loss_llm": 0.00568997859954834, "grad_norm": 0.04933592304587364, "global_step": 82, "epoch": 2, "lr": 0.001} +{"train_loss": 0.07893094420433044, "train_loss_bc": 0.07888577878475189, "train_loss_llm": 0.004516661167144775, "grad_norm": 0.07317051291465759, "global_step": 83, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08824880421161652, "train_loss_bc": 0.08824880421161652, "train_loss_llm": 0.0, "grad_norm": 0.09500908106565475, "global_step": 84, "epoch": 2, "lr": 0.001} +{"train_loss": 0.0822557657957077, "train_loss_bc": 0.08213935792446136, "train_loss_llm": 0.0116405189037323, "grad_norm": 0.02943960763514042, "global_step": 85, "epoch": 2, "lr": 0.001} +{"train_loss": 0.07250358909368515, "train_loss_bc": 0.07242656499147415, "train_loss_llm": 0.0077021121978759766, "grad_norm": 0.057275980710983276, "global_step": 86, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08161371946334839, "train_loss_bc": 0.08161371946334839, "train_loss_llm": 0.0, "grad_norm": 0.08636434376239777, "global_step": 87, "epoch": 2, "lr": 0.001} +{"train_loss": 0.07365914434194565, "train_loss_bc": 0.07364202290773392, "train_loss_llm": 0.001712203025817871, "grad_norm": 0.10604052990674973, "global_step": 88, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08253642171621323, "train_loss_bc": 0.08236472308635712, "train_loss_llm": 0.01716974377632141, "grad_norm": 0.02948630601167679, "global_step": 89, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08278562873601913, "train_loss_bc": 0.08277592808008194, "train_loss_llm": 0.0009698569774627686, "grad_norm": 0.056550245732069016, "global_step": 90, "epoch": 2, "lr": 0.001} +{"train_loss": 0.07539838552474976, "train_loss_bc": 0.07520857453346252, "train_loss_llm": 0.018981367349624634, "grad_norm": 0.08432463556528091, "global_step": 91, "epoch": 2, "lr": 0.001} +{"train_loss": 0.0861327052116394, "train_loss_bc": 0.0861327052116394, "train_loss_llm": 0.0, "grad_norm": 0.10834867507219315, "global_step": 92, "epoch": 2, "lr": 0.001} +{"train_loss": 0.07055738568305969, "train_loss_bc": 0.0705428272485733, "train_loss_llm": 0.0014560520648956299, "grad_norm": 0.036106325685977936, "global_step": 93, "epoch": 2, "lr": 0.001} +{"train_loss": 0.06933951377868652, "train_loss_bc": 0.06919165700674057, "train_loss_llm": 0.014785557985305786, "grad_norm": 0.0600740872323513, "global_step": 94, "epoch": 2, "lr": 0.001} +{"train_loss": 0.07866871356964111, "train_loss_bc": 0.07856092602014542, "train_loss_llm": 0.010778844356536865, "grad_norm": 0.08890736103057861, "global_step": 95, "epoch": 2, "lr": 0.001} +{"train_loss": 0.083490289747715, "train_loss_bc": 0.08334947377443314, "train_loss_llm": 0.01408165693283081, "grad_norm": 0.10839393734931946, "global_step": 96, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08444800227880478, "train_loss_bc": 0.08415356278419495, "train_loss_llm": 0.029444068670272827, "grad_norm": 0.03152940049767494, "global_step": 97, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08576589077711105, "train_loss_bc": 0.08543340861797333, "train_loss_llm": 0.03324797749519348, "grad_norm": 0.06256766617298126, "global_step": 98, "epoch": 2, "lr": 0.001} +{"train_loss": 0.07607192546129227, "train_loss_bc": 0.07601644098758698, "train_loss_llm": 0.005548804998397827, "grad_norm": 0.08933773636817932, "global_step": 99, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08090734481811523, "train_loss_bc": 0.08087822794914246, "train_loss_llm": 0.0029113590717315674, "grad_norm": 0.11964273452758789, "global_step": 100, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08601375669240952, "train_loss_bc": 0.08578149229288101, "train_loss_llm": 0.023226618766784668, "grad_norm": 0.03437696024775505, "global_step": 101, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08220518380403519, "train_loss_bc": 0.08192698657512665, "train_loss_llm": 0.027819663286209106, "grad_norm": 0.07013748586177826, "global_step": 102, "epoch": 2, "lr": 0.001} +{"train_loss": 0.0757322832942009, "train_loss_bc": 0.0757322832942009, "train_loss_llm": 0.0, "grad_norm": 0.09910586476325989, "global_step": 103, "epoch": 2, "lr": 0.001} +{"train_loss": 0.1025790125131607, "train_loss_bc": 0.1025419756770134, "train_loss_llm": 0.0037039518356323242, "grad_norm": 0.14036281406879425, "global_step": 104, "epoch": 2, "lr": 0.001} +{"train_loss": 0.09655417501926422, "train_loss_bc": 0.09619559347629547, "train_loss_llm": 0.03585779666900635, "grad_norm": 0.046981681138277054, "global_step": 105, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08999045938253403, "train_loss_bc": 0.08994846045970917, "train_loss_llm": 0.004199802875518799, "grad_norm": 0.08149876445531845, "global_step": 106, "epoch": 2, "lr": 0.001} +{"train_loss": 0.08311468900905715, "train_loss_bc": 0.07647249102592468, "train_loss_llm": 0.00689387321472168, "grad_norm": 0.1044260635972023, "global_step": 107, "epoch": 2, "lr": 0.001, "val_loss": 0.08063984662294388} +{"train_loss": 0.08324689418077469, "train_loss_bc": 0.08291232585906982, "train_loss_llm": 0.03345674276351929, "grad_norm": 0.13821442425251007, "global_step": 108, "epoch": 3, "lr": 0.001} +{"train_loss": 0.08320071548223495, "train_loss_bc": 0.0827709510922432, "train_loss_llm": 0.04297667741775513, "grad_norm": 0.03630191087722778, "global_step": 109, "epoch": 3, "lr": 0.001} +{"train_loss": 0.0767817273736, "train_loss_bc": 0.07669715583324432, "train_loss_llm": 0.008456975221633911, "grad_norm": 0.06889642030000687, "global_step": 110, "epoch": 3, "lr": 0.001} +{"train_loss": 0.09529038518667221, "train_loss_bc": 0.09519656747579575, "train_loss_llm": 0.009382098913192749, "grad_norm": 0.10792939364910126, "global_step": 111, "epoch": 3, "lr": 0.001} +{"train_loss": 0.0790187269449234, "train_loss_bc": 0.07891406863927841, "train_loss_llm": 0.010465681552886963, "grad_norm": 0.14150451123714447, "global_step": 112, "epoch": 3, "lr": 0.001} +{"train_loss": 0.0853617936372757, "train_loss_bc": 0.08507999032735825, "train_loss_llm": 0.028180211782455444, "grad_norm": 0.04749515280127525, "global_step": 113, "epoch": 3, "lr": 0.001} +{"train_loss": 0.09807360172271729, "train_loss_bc": 0.09803265333175659, "train_loss_llm": 0.004095196723937988, "grad_norm": 0.0820637121796608, "global_step": 114, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07623589783906937, "train_loss_bc": 0.07601924985647202, "train_loss_llm": 0.021664828062057495, "grad_norm": 0.10974918305873871, "global_step": 115, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07530578970909119, "train_loss_bc": 0.07497749477624893, "train_loss_llm": 0.03282913565635681, "grad_norm": 0.13644802570343018, "global_step": 116, "epoch": 3, "lr": 0.001} +{"train_loss": 0.08758726716041565, "train_loss_bc": 0.0873701423406601, "train_loss_llm": 0.02171224355697632, "grad_norm": 0.04311204329133034, "global_step": 117, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07680424302816391, "train_loss_bc": 0.07671515643596649, "train_loss_llm": 0.008908361196517944, "grad_norm": 0.06451256573200226, "global_step": 118, "epoch": 3, "lr": 0.001} +{"train_loss": 0.0818992480635643, "train_loss_bc": 0.0818992480635643, "train_loss_llm": 0.0, "grad_norm": 0.09640331566333771, "global_step": 119, "epoch": 3, "lr": 0.001} +{"train_loss": 0.08005398511886597, "train_loss_bc": 0.0799797922372818, "train_loss_llm": 0.007418990135192871, "grad_norm": 0.11996687203645706, "global_step": 120, "epoch": 3, "lr": 0.001} +{"train_loss": 0.0973653793334961, "train_loss_bc": 0.09725606441497803, "train_loss_llm": 0.01093176007270813, "grad_norm": 0.0357821099460125, "global_step": 121, "epoch": 3, "lr": 0.001} +{"train_loss": 0.08146407455205917, "train_loss_bc": 0.08144266158342361, "train_loss_llm": 0.002141326665878296, "grad_norm": 0.06469948589801788, "global_step": 122, "epoch": 3, "lr": 0.001} +{"train_loss": 0.0753559097647667, "train_loss_bc": 0.07514787465333939, "train_loss_llm": 0.0208035409450531, "grad_norm": 0.0940680280327797, "global_step": 123, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07908204942941666, "train_loss_bc": 0.07904765754938126, "train_loss_llm": 0.0034392476081848145, "grad_norm": 0.12202748656272888, "global_step": 124, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07586877793073654, "train_loss_bc": 0.07573097199201584, "train_loss_llm": 0.013780653476715088, "grad_norm": 0.03632582724094391, "global_step": 125, "epoch": 3, "lr": 0.001} +{"train_loss": 0.08391119539737701, "train_loss_bc": 0.08380527794361115, "train_loss_llm": 0.010591894388198853, "grad_norm": 0.06547248363494873, "global_step": 126, "epoch": 3, "lr": 0.001} +{"train_loss": 0.0703359842300415, "train_loss_bc": 0.07027570903301239, "train_loss_llm": 0.006027728319168091, "grad_norm": 0.08978860080242157, "global_step": 127, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07826254516839981, "train_loss_bc": 0.07822852581739426, "train_loss_llm": 0.003402113914489746, "grad_norm": 0.11869083344936371, "global_step": 128, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07143451273441315, "train_loss_bc": 0.07097014039754868, "train_loss_llm": 0.04643750190734863, "grad_norm": 0.0288070160895586, "global_step": 129, "epoch": 3, "lr": 0.001} +{"train_loss": 0.08295420557260513, "train_loss_bc": 0.08280608803033829, "train_loss_llm": 0.01481202244758606, "grad_norm": 0.057669151574373245, "global_step": 130, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07528524845838547, "train_loss_bc": 0.07523015141487122, "train_loss_llm": 0.00550997257232666, "grad_norm": 0.07884678989648819, "global_step": 131, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07807356864213943, "train_loss_bc": 0.07803419977426529, "train_loss_llm": 0.003936707973480225, "grad_norm": 0.10587422549724579, "global_step": 132, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07053627073764801, "train_loss_bc": 0.07048098742961884, "train_loss_llm": 0.005528479814529419, "grad_norm": 0.026347925886511803, "global_step": 133, "epoch": 3, "lr": 0.001} +{"train_loss": 0.09011418372392654, "train_loss_bc": 0.09007778763771057, "train_loss_llm": 0.0036392509937286377, "grad_norm": 0.054984934628009796, "global_step": 134, "epoch": 3, "lr": 0.001} +{"train_loss": 0.08149770647287369, "train_loss_bc": 0.08137228339910507, "train_loss_llm": 0.012542307376861572, "grad_norm": 0.08671777695417404, "global_step": 135, "epoch": 3, "lr": 0.001} +{"train_loss": 0.059055112302303314, "train_loss_bc": 0.05887514352798462, "train_loss_llm": 0.017996996641159058, "grad_norm": 0.11316715180873871, "global_step": 136, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07592678815126419, "train_loss_bc": 0.0759122222661972, "train_loss_llm": 0.0014563202857971191, "grad_norm": 0.02865324169397354, "global_step": 137, "epoch": 3, "lr": 0.001} +{"train_loss": 0.0743570476770401, "train_loss_bc": 0.07435060292482376, "train_loss_llm": 0.0006444156169891357, "grad_norm": 0.05004405602812767, "global_step": 138, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07794541865587234, "train_loss_bc": 0.07783754914999008, "train_loss_llm": 0.01078706979751587, "grad_norm": 0.08000926673412323, "global_step": 139, "epoch": 3, "lr": 0.001} +{"train_loss": 0.06558370590209961, "train_loss_bc": 0.06558370590209961, "train_loss_llm": 0.0, "grad_norm": 0.10063409060239792, "global_step": 140, "epoch": 3, "lr": 0.001} +{"train_loss": 0.06180437654256821, "train_loss_bc": 0.06176682189106941, "train_loss_llm": 0.0037554800510406494, "grad_norm": 0.03916173800826073, "global_step": 141, "epoch": 3, "lr": 0.001} +{"train_loss": 0.06927559524774551, "train_loss_bc": 0.06927559524774551, "train_loss_llm": 0.0, "grad_norm": 0.07217301428318024, "global_step": 142, "epoch": 3, "lr": 0.001} +{"train_loss": 0.07868659268650743, "train_loss_bc": 0.07836740463972092, "train_loss_llm": 0.0, "grad_norm": 0.10836897045373917, "global_step": 143, "epoch": 3, "lr": 0.001, "val_loss": 0.0693485289812088} +{"train_loss": 0.06904329359531403, "train_loss_bc": 0.06904329359531403, "train_loss_llm": 0.0, "grad_norm": 0.13099902868270874, "global_step": 144, "epoch": 4, "lr": 0.001} +{"train_loss": 0.055853359401226044, "train_loss_bc": 0.055808939039707184, "train_loss_llm": 0.004441946744918823, "grad_norm": 0.024607980623841286, "global_step": 145, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06356307864189148, "train_loss_bc": 0.06352220475673676, "train_loss_llm": 0.004087090492248535, "grad_norm": 0.04687747731804848, "global_step": 146, "epoch": 4, "lr": 0.001} +{"train_loss": 0.0725642666220665, "train_loss_bc": 0.07253964990377426, "train_loss_llm": 0.0024617016315460205, "grad_norm": 0.06468348950147629, "global_step": 147, "epoch": 4, "lr": 0.001} +{"train_loss": 0.07817354798316956, "train_loss_bc": 0.07817354798316956, "train_loss_llm": 0.0, "grad_norm": 0.08133935928344727, "global_step": 148, "epoch": 4, "lr": 0.001} +{"train_loss": 0.08228011429309845, "train_loss_bc": 0.08228011429309845, "train_loss_llm": 0.0, "grad_norm": 0.02611471898853779, "global_step": 149, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06008812040090561, "train_loss_bc": 0.06008812040090561, "train_loss_llm": 0.0, "grad_norm": 0.047335315495729446, "global_step": 150, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06612466275691986, "train_loss_bc": 0.0661148801445961, "train_loss_llm": 0.0009781122207641602, "grad_norm": 0.06231296434998512, "global_step": 151, "epoch": 4, "lr": 0.001} +{"train_loss": 0.0641748458147049, "train_loss_bc": 0.06414794921875, "train_loss_llm": 0.002689361572265625, "grad_norm": 0.08176790177822113, "global_step": 152, "epoch": 4, "lr": 0.001} +{"train_loss": 0.0644775778055191, "train_loss_bc": 0.0644775778055191, "train_loss_llm": 0.0, "grad_norm": 0.029367942363023758, "global_step": 153, "epoch": 4, "lr": 0.001} +{"train_loss": 0.060478705912828445, "train_loss_bc": 0.060478705912828445, "train_loss_llm": 0.0, "grad_norm": 0.04266729578375816, "global_step": 154, "epoch": 4, "lr": 0.001} +{"train_loss": 0.07013380527496338, "train_loss_bc": 0.07013101875782013, "train_loss_llm": 0.0002785325050354004, "grad_norm": 0.052063729614019394, "global_step": 155, "epoch": 4, "lr": 0.001} +{"train_loss": 0.07255759835243225, "train_loss_bc": 0.07255759835243225, "train_loss_llm": 0.0, "grad_norm": 0.07054711878299713, "global_step": 156, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06826630979776382, "train_loss_bc": 0.06826630979776382, "train_loss_llm": 0.0, "grad_norm": 0.02049613744020462, "global_step": 157, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06845840811729431, "train_loss_bc": 0.06845840811729431, "train_loss_llm": 0.0, "grad_norm": 0.031714294105768204, "global_step": 158, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06051059812307358, "train_loss_bc": 0.06051059812307358, "train_loss_llm": 0.0, "grad_norm": 0.04486490786075592, "global_step": 159, "epoch": 4, "lr": 0.001} +{"train_loss": 0.07100626081228256, "train_loss_bc": 0.07100626081228256, "train_loss_llm": 0.0, "grad_norm": 0.05970165878534317, "global_step": 160, "epoch": 4, "lr": 0.001} +{"train_loss": 0.05390162020921707, "train_loss_bc": 0.05390162020921707, "train_loss_llm": 0.0, "grad_norm": 0.02282865345478058, "global_step": 161, "epoch": 4, "lr": 0.001} +{"train_loss": 0.07422439754009247, "train_loss_bc": 0.07422439754009247, "train_loss_llm": 0.0, "grad_norm": 0.04533824697136879, "global_step": 162, "epoch": 4, "lr": 0.001} +{"train_loss": 0.054731108248233795, "train_loss_bc": 0.05468939244747162, "train_loss_llm": 0.004171609878540039, "grad_norm": 0.05688827857375145, "global_step": 163, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06418431550264359, "train_loss_bc": 0.06418431550264359, "train_loss_llm": 0.0, "grad_norm": 0.07180362939834595, "global_step": 164, "epoch": 4, "lr": 0.001} +{"train_loss": 0.08196824043989182, "train_loss_bc": 0.08190544694662094, "train_loss_llm": 0.006279438734054565, "grad_norm": 0.029046647250652313, "global_step": 165, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06280729919672012, "train_loss_bc": 0.06280729919672012, "train_loss_llm": 0.0, "grad_norm": 0.03585397079586983, "global_step": 166, "epoch": 4, "lr": 0.001} +{"train_loss": 0.07829765230417252, "train_loss_bc": 0.0782717913389206, "train_loss_llm": 0.002585768699645996, "grad_norm": 0.06358509510755539, "global_step": 167, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06841910630464554, "train_loss_bc": 0.06838169693946838, "train_loss_llm": 0.0037411749362945557, "grad_norm": 0.08515624701976776, "global_step": 168, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06326606124639511, "train_loss_bc": 0.06323078274726868, "train_loss_llm": 0.003527522087097168, "grad_norm": 0.021248389035463333, "global_step": 169, "epoch": 4, "lr": 0.001} +{"train_loss": 0.07268975675106049, "train_loss_bc": 0.07268132269382477, "train_loss_llm": 0.0008435249328613281, "grad_norm": 0.038149259984493256, "global_step": 170, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06499861180782318, "train_loss_bc": 0.06496705114841461, "train_loss_llm": 0.0031562745571136475, "grad_norm": 0.06371229887008667, "global_step": 171, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06585197895765305, "train_loss_bc": 0.06582131236791611, "train_loss_llm": 0.0030667483806610107, "grad_norm": 0.08498216420412064, "global_step": 172, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06832756102085114, "train_loss_bc": 0.06829891353845596, "train_loss_llm": 0.0028649568557739258, "grad_norm": 0.02621711976826191, "global_step": 173, "epoch": 4, "lr": 0.001} +{"train_loss": 0.07194164395332336, "train_loss_bc": 0.07188723236322403, "train_loss_llm": 0.005441188812255859, "grad_norm": 0.04285333678126335, "global_step": 174, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06771280616521835, "train_loss_bc": 0.06771280616521835, "train_loss_llm": 0.0, "grad_norm": 0.06276906281709671, "global_step": 175, "epoch": 4, "lr": 0.001} +{"train_loss": 0.07152236998081207, "train_loss_bc": 0.07149367779493332, "train_loss_llm": 0.002868860960006714, "grad_norm": 0.07894676923751831, "global_step": 176, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06383213400840759, "train_loss_bc": 0.06383129954338074, "train_loss_llm": 8.308887481689453e-05, "grad_norm": 0.02201147936284542, "global_step": 177, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06678248196840286, "train_loss_bc": 0.066782146692276, "train_loss_llm": 3.319978713989258e-05, "grad_norm": 0.034137457609176636, "global_step": 178, "epoch": 4, "lr": 0.001} +{"train_loss": 0.06792155539409982, "train_loss_bc": 0.08196229487657547, "train_loss_llm": 0.0, "grad_norm": 0.05289863795042038, "global_step": 179, "epoch": 4, "lr": 0.001, "val_loss": 0.062316738069057465} +{"train_loss": 0.06801294535398483, "train_loss_bc": 0.06793256849050522, "train_loss_llm": 0.008037745952606201, "grad_norm": 0.0705992728471756, "global_step": 180, "epoch": 5, "lr": 0.001} +{"train_loss": 0.07052955776453018, "train_loss_bc": 0.07042136788368225, "train_loss_llm": 0.010818779468536377, "grad_norm": 0.031079059466719627, "global_step": 181, "epoch": 5, "lr": 0.001} +{"train_loss": 0.057784851640462875, "train_loss_bc": 0.05772611126303673, "train_loss_llm": 0.005873918533325195, "grad_norm": 0.03912213444709778, "global_step": 182, "epoch": 5, "lr": 0.001} +{"train_loss": 0.054082613438367844, "train_loss_bc": 0.05407712236046791, "train_loss_llm": 0.0005489587783813477, "grad_norm": 0.060440145432949066, "global_step": 183, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06350745260715485, "train_loss_bc": 0.06349346786737442, "train_loss_llm": 0.001398235559463501, "grad_norm": 0.08212807774543762, "global_step": 184, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06412016600370407, "train_loss_bc": 0.0641188845038414, "train_loss_llm": 0.00012817978858947754, "grad_norm": 0.02098291926085949, "global_step": 185, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06847255676984787, "train_loss_bc": 0.06847255676984787, "train_loss_llm": 0.0, "grad_norm": 0.042683061212301254, "global_step": 186, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06922929733991623, "train_loss_bc": 0.06921839714050293, "train_loss_llm": 0.001089930534362793, "grad_norm": 0.05838584527373314, "global_step": 187, "epoch": 5, "lr": 0.001} +{"train_loss": 0.07055778056383133, "train_loss_bc": 0.07055778056383133, "train_loss_llm": 0.0, "grad_norm": 0.07204063981771469, "global_step": 188, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06067642942070961, "train_loss_bc": 0.06063413247466087, "train_loss_llm": 0.004229873418807983, "grad_norm": 0.02246907912194729, "global_step": 189, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06995216757059097, "train_loss_bc": 0.06995216757059097, "train_loss_llm": 0.0, "grad_norm": 0.03499627485871315, "global_step": 190, "epoch": 5, "lr": 0.001} +{"train_loss": 0.07212399691343307, "train_loss_bc": 0.07206197082996368, "train_loss_llm": 0.006202727556228638, "grad_norm": 0.0528438538312912, "global_step": 191, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06622835248708725, "train_loss_bc": 0.06621386110782623, "train_loss_llm": 0.0014493465423583984, "grad_norm": 0.0673261508345604, "global_step": 192, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06327985972166061, "train_loss_bc": 0.06324179470539093, "train_loss_llm": 0.003806203603744507, "grad_norm": 0.025711728259921074, "global_step": 193, "epoch": 5, "lr": 0.001} +{"train_loss": 0.0661410391330719, "train_loss_bc": 0.06609070301055908, "train_loss_llm": 0.005033761262893677, "grad_norm": 0.04590120539069176, "global_step": 194, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06577622890472412, "train_loss_bc": 0.06577622890472412, "train_loss_llm": 0.0, "grad_norm": 0.06124860420823097, "global_step": 195, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06616134196519852, "train_loss_bc": 0.0660867989063263, "train_loss_llm": 0.007454395294189453, "grad_norm": 0.0789732113480568, "global_step": 196, "epoch": 5, "lr": 0.001} +{"train_loss": 0.061861198395490646, "train_loss_bc": 0.06179310753941536, "train_loss_llm": 0.006809115409851074, "grad_norm": 0.023910928517580032, "global_step": 197, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06163787469267845, "train_loss_bc": 0.06160613149404526, "train_loss_llm": 0.0031743645668029785, "grad_norm": 0.03905171528458595, "global_step": 198, "epoch": 5, "lr": 0.001} +{"train_loss": 0.07977976649999619, "train_loss_bc": 0.07977976649999619, "train_loss_llm": 0.0, "grad_norm": 0.054405923932790756, "global_step": 199, "epoch": 5, "lr": 0.001} +{"train_loss": 0.07100566476583481, "train_loss_bc": 0.07100566476583481, "train_loss_llm": 0.0, "grad_norm": 0.06740519404411316, "global_step": 200, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06469764560461044, "train_loss_bc": 0.06463919579982758, "train_loss_llm": 0.005845129489898682, "grad_norm": 0.02045482210814953, "global_step": 201, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06952007114887238, "train_loss_bc": 0.06948735564947128, "train_loss_llm": 0.0032714903354644775, "grad_norm": 0.03488137573003769, "global_step": 202, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06347373127937317, "train_loss_bc": 0.06343258917331696, "train_loss_llm": 0.004114240407943726, "grad_norm": 0.05365242063999176, "global_step": 203, "epoch": 5, "lr": 0.001} +{"train_loss": 0.0686735287308693, "train_loss_bc": 0.06858106702566147, "train_loss_llm": 0.009245812892913818, "grad_norm": 0.06775783747434616, "global_step": 204, "epoch": 5, "lr": 0.001} +{"train_loss": 0.07626858353614807, "train_loss_bc": 0.07626858353614807, "train_loss_llm": 0.0, "grad_norm": 0.030766669660806656, "global_step": 205, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06918276846408844, "train_loss_bc": 0.06915315985679626, "train_loss_llm": 0.0029609501361846924, "grad_norm": 0.03930740803480148, "global_step": 206, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06445353478193283, "train_loss_bc": 0.06445353478193283, "train_loss_llm": 0.0, "grad_norm": 0.058900654315948486, "global_step": 207, "epoch": 5, "lr": 0.001} +{"train_loss": 0.07457854598760605, "train_loss_bc": 0.07455088198184967, "train_loss_llm": 0.002766162157058716, "grad_norm": 0.07643464207649231, "global_step": 208, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06813756376504898, "train_loss_bc": 0.0681132972240448, "train_loss_llm": 0.0024264156818389893, "grad_norm": 0.02389916405081749, "global_step": 209, "epoch": 5, "lr": 0.001} +{"train_loss": 0.0770978182554245, "train_loss_bc": 0.07709505409002304, "train_loss_llm": 0.0002765953540802002, "grad_norm": 0.03820963203907013, "global_step": 210, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06924589723348618, "train_loss_bc": 0.06921534985303879, "train_loss_llm": 0.003054410219192505, "grad_norm": 0.04439821094274521, "global_step": 211, "epoch": 5, "lr": 0.001} +{"train_loss": 0.05871967971324921, "train_loss_bc": 0.05869924649596214, "train_loss_llm": 0.0020431578159332275, "grad_norm": 0.056775014847517014, "global_step": 212, "epoch": 5, "lr": 0.001} +{"train_loss": 0.05677890405058861, "train_loss_bc": 0.05675162747502327, "train_loss_llm": 0.0027276277542114258, "grad_norm": 0.02315608784556389, "global_step": 213, "epoch": 5, "lr": 0.001} +{"train_loss": 0.06706459075212479, "train_loss_bc": 0.06704214960336685, "train_loss_llm": 0.0022438764572143555, "grad_norm": 0.03542454540729523, "global_step": 214, "epoch": 5, "lr": 0.001} +{"train_loss": 0.0666743643167946, "train_loss_bc": 0.061451710760593414, "train_loss_llm": 0.001139998435974121, "grad_norm": 0.05226783826947212, "global_step": 215, "epoch": 5, "lr": 0.001, "train/cumulative_reward": -1.9580580909723828, "train/mean_score": -0.012384377424153681, "train/success_rate": 0.0, "test/cumulative_reward": -1.9417149005733771, "test/mean_score": -0.014269949721550525, "test/success_rate": 0.0, "val_loss": 0.05880626663565636, "train_action_mse_error": 0.06415405869483948} +{"train_loss": 0.058765653520822525, "train_loss_bc": 0.05876460298895836, "train_loss_llm": 0.0001049339771270752, "grad_norm": 0.05917933210730553, "global_step": 216, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06422530114650726, "train_loss_bc": 0.06419514864683151, "train_loss_llm": 0.00301516056060791, "grad_norm": 0.02250281162559986, "global_step": 217, "epoch": 6, "lr": 0.001} +{"train_loss": 0.061993129551410675, "train_loss_bc": 0.06198911368846893, "train_loss_llm": 0.00040140748023986816, "grad_norm": 0.038895901292562485, "global_step": 218, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06263460218906403, "train_loss_bc": 0.06255735456943512, "train_loss_llm": 0.0077245235443115234, "grad_norm": 0.045062948018312454, "global_step": 219, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06155487149953842, "train_loss_bc": 0.06155487149953842, "train_loss_llm": 0.0, "grad_norm": 0.05501548573374748, "global_step": 220, "epoch": 6, "lr": 0.001} +{"train_loss": 0.07351231575012207, "train_loss_bc": 0.07345397770404816, "train_loss_llm": 0.005834072828292847, "grad_norm": 0.021739432588219643, "global_step": 221, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06309612095355988, "train_loss_bc": 0.06309612095355988, "train_loss_llm": 0.0, "grad_norm": 0.02797253057360649, "global_step": 222, "epoch": 6, "lr": 0.001} +{"train_loss": 0.05847320333123207, "train_loss_bc": 0.05844920501112938, "train_loss_llm": 0.002400010824203491, "grad_norm": 0.04440337046980858, "global_step": 223, "epoch": 6, "lr": 0.001} +{"train_loss": 0.057315584272146225, "train_loss_bc": 0.05730487406253815, "train_loss_llm": 0.0010709166526794434, "grad_norm": 0.05229783430695534, "global_step": 224, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06075450778007507, "train_loss_bc": 0.06072854995727539, "train_loss_llm": 0.002595961093902588, "grad_norm": 0.016385428607463837, "global_step": 225, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06199093163013458, "train_loss_bc": 0.06199093163013458, "train_loss_llm": 0.0, "grad_norm": 0.031170688569545746, "global_step": 226, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06460154801607132, "train_loss_bc": 0.0646013393998146, "train_loss_llm": 2.104043960571289e-05, "grad_norm": 0.04270561784505844, "global_step": 227, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06697934865951538, "train_loss_bc": 0.06695900857448578, "train_loss_llm": 0.0020338892936706543, "grad_norm": 0.049565404653549194, "global_step": 228, "epoch": 6, "lr": 0.001} +{"train_loss": 0.05702749267220497, "train_loss_bc": 0.05702749267220497, "train_loss_llm": 0.0, "grad_norm": 0.01768457517027855, "global_step": 229, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06891944259405136, "train_loss_bc": 0.06891944259405136, "train_loss_llm": 0.0, "grad_norm": 0.024752924218773842, "global_step": 230, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06263751536607742, "train_loss_bc": 0.06263751536607742, "train_loss_llm": 0.0, "grad_norm": 0.0330590158700943, "global_step": 231, "epoch": 6, "lr": 0.001} +{"train_loss": 0.05603894591331482, "train_loss_bc": 0.056025080382823944, "train_loss_llm": 0.0013867318630218506, "grad_norm": 0.04327157139778137, "global_step": 232, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06820101290941238, "train_loss_bc": 0.06819552183151245, "train_loss_llm": 0.00054892897605896, "grad_norm": 0.02295525185763836, "global_step": 233, "epoch": 6, "lr": 0.001} +{"train_loss": 0.0658290758728981, "train_loss_bc": 0.06582033634185791, "train_loss_llm": 0.0008738934993743896, "grad_norm": 0.03030378557741642, "global_step": 234, "epoch": 6, "lr": 0.001} +{"train_loss": 0.05934295430779457, "train_loss_bc": 0.059333398938179016, "train_loss_llm": 0.0009555220603942871, "grad_norm": 0.038951002061367035, "global_step": 235, "epoch": 6, "lr": 0.001} +{"train_loss": 0.05885326489806175, "train_loss_bc": 0.05885326489806175, "train_loss_llm": 0.0, "grad_norm": 0.04764864966273308, "global_step": 236, "epoch": 6, "lr": 0.001} +{"train_loss": 0.062336213886737823, "train_loss_bc": 0.062336213886737823, "train_loss_llm": 0.0, "grad_norm": 0.026546979323029518, "global_step": 237, "epoch": 6, "lr": 0.001} +{"train_loss": 0.0565670020878315, "train_loss_bc": 0.056554727256298065, "train_loss_llm": 0.0012274086475372314, "grad_norm": 0.03200583904981613, "global_step": 238, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06354999542236328, "train_loss_bc": 0.06354999542236328, "train_loss_llm": 0.0, "grad_norm": 0.04813380539417267, "global_step": 239, "epoch": 6, "lr": 0.001} +{"train_loss": 0.05655144527554512, "train_loss_bc": 0.05655144527554512, "train_loss_llm": 0.0, "grad_norm": 0.05529145896434784, "global_step": 240, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06758465617895126, "train_loss_bc": 0.06758465617895126, "train_loss_llm": 0.0, "grad_norm": 0.022493869066238403, "global_step": 241, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06684653460979462, "train_loss_bc": 0.06684653460979462, "train_loss_llm": 0.0, "grad_norm": 0.03450334444642067, "global_step": 242, "epoch": 6, "lr": 0.001} +{"train_loss": 0.052745889872312546, "train_loss_bc": 0.05272839218378067, "train_loss_llm": 0.0017496347427368164, "grad_norm": 0.05375980958342552, "global_step": 243, "epoch": 6, "lr": 0.001} +{"train_loss": 0.057369355112314224, "train_loss_bc": 0.057369355112314224, "train_loss_llm": 0.0, "grad_norm": 0.05921115726232529, "global_step": 244, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06332226097583771, "train_loss_bc": 0.06330795586109161, "train_loss_llm": 0.0014302730560302734, "grad_norm": 0.021531550213694572, "global_step": 245, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06385477632284164, "train_loss_bc": 0.06383110582828522, "train_loss_llm": 0.0023673176765441895, "grad_norm": 0.034382060170173645, "global_step": 246, "epoch": 6, "lr": 0.001} +{"train_loss": 0.05679604038596153, "train_loss_bc": 0.056760720908641815, "train_loss_llm": 0.0035318732261657715, "grad_norm": 0.04417864605784416, "global_step": 247, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06286464631557465, "train_loss_bc": 0.06284908950328827, "train_loss_llm": 0.001555800437927246, "grad_norm": 0.051117002964019775, "global_step": 248, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06051599979400635, "train_loss_bc": 0.06051599979400635, "train_loss_llm": 0.0, "grad_norm": 0.020172974094748497, "global_step": 249, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06357142329216003, "train_loss_bc": 0.06354116648435593, "train_loss_llm": 0.0030258595943450928, "grad_norm": 0.034920431673526764, "global_step": 250, "epoch": 6, "lr": 0.001} +{"train_loss": 0.06221125347332822, "train_loss_bc": 0.07232381403446198, "train_loss_llm": 0.005824655294418335, "grad_norm": 0.045654673129320145, "global_step": 251, "epoch": 6, "lr": 0.001, "val_loss": 0.05637954920530319} +{"train_loss": 0.05708528682589531, "train_loss_bc": 0.05708528682589531, "train_loss_llm": 0.0, "grad_norm": 0.057879019528627396, "global_step": 252, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05827203020453453, "train_loss_bc": 0.05822095274925232, "train_loss_llm": 0.005107671022415161, "grad_norm": 0.017780829221010208, "global_step": 253, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05852058157324791, "train_loss_bc": 0.05851900950074196, "train_loss_llm": 0.00015720725059509277, "grad_norm": 0.03089502826333046, "global_step": 254, "epoch": 7, "lr": 0.001} +{"train_loss": 0.062558613717556, "train_loss_bc": 0.06255161762237549, "train_loss_llm": 0.0006994307041168213, "grad_norm": 0.044130440801382065, "global_step": 255, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05638311058282852, "train_loss_bc": 0.05638311058282852, "train_loss_llm": 0.0, "grad_norm": 0.05925316363573074, "global_step": 256, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05490505322813988, "train_loss_bc": 0.05489257350564003, "train_loss_llm": 0.0012479424476623535, "grad_norm": 0.019745154306292534, "global_step": 257, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06285416334867477, "train_loss_bc": 0.06285416334867477, "train_loss_llm": 0.0, "grad_norm": 0.03177862986922264, "global_step": 258, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06263235956430435, "train_loss_bc": 0.06263036280870438, "train_loss_llm": 0.00019937753677368164, "grad_norm": 0.0391080267727375, "global_step": 259, "epoch": 7, "lr": 0.001} +{"train_loss": 0.048638250678777695, "train_loss_bc": 0.04860140383243561, "train_loss_llm": 0.0036846697330474854, "grad_norm": 0.04830625280737877, "global_step": 260, "epoch": 7, "lr": 0.001} +{"train_loss": 0.054129522293806076, "train_loss_bc": 0.05412166938185692, "train_loss_llm": 0.0007854700088500977, "grad_norm": 0.021874669939279556, "global_step": 261, "epoch": 7, "lr": 0.001} +{"train_loss": 0.051204614341259, "train_loss_bc": 0.05118630453944206, "train_loss_llm": 0.0018309056758880615, "grad_norm": 0.03291689604520798, "global_step": 262, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06591618806123734, "train_loss_bc": 0.06590061634778976, "train_loss_llm": 0.0015574097633361816, "grad_norm": 0.038737207651138306, "global_step": 263, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06017795950174332, "train_loss_bc": 0.06017795950174332, "train_loss_llm": 0.0, "grad_norm": 0.052480313926935196, "global_step": 264, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06373428553342819, "train_loss_bc": 0.06372243165969849, "train_loss_llm": 0.001185089349746704, "grad_norm": 0.021652525290846825, "global_step": 265, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05308135226368904, "train_loss_bc": 0.05306018143892288, "train_loss_llm": 0.002116978168487549, "grad_norm": 0.03165609762072563, "global_step": 266, "epoch": 7, "lr": 0.001} +{"train_loss": 0.057408519089221954, "train_loss_bc": 0.05735716223716736, "train_loss_llm": 0.005135595798492432, "grad_norm": 0.03975195437669754, "global_step": 267, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05622658133506775, "train_loss_bc": 0.05622658133506775, "train_loss_llm": 0.0, "grad_norm": 0.05523362010717392, "global_step": 268, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06915558874607086, "train_loss_bc": 0.06915353238582611, "train_loss_llm": 0.00020569562911987305, "grad_norm": 0.021225642412900925, "global_step": 269, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05268959701061249, "train_loss_bc": 0.052663739770650864, "train_loss_llm": 0.0025856494903564453, "grad_norm": 0.030418753623962402, "global_step": 270, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06136137247085571, "train_loss_bc": 0.06136137247085571, "train_loss_llm": 0.0, "grad_norm": 0.039983849972486496, "global_step": 271, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05798204243183136, "train_loss_bc": 0.05792815238237381, "train_loss_llm": 0.005389034748077393, "grad_norm": 0.049785736948251724, "global_step": 272, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06929613649845123, "train_loss_bc": 0.06928012520074844, "train_loss_llm": 0.0016014277935028076, "grad_norm": 0.022126348689198494, "global_step": 273, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06644885987043381, "train_loss_bc": 0.06644885987043381, "train_loss_llm": 0.0, "grad_norm": 0.034302107989788055, "global_step": 274, "epoch": 7, "lr": 0.001} +{"train_loss": 0.060943301767110825, "train_loss_bc": 0.06085958704352379, "train_loss_llm": 0.008371502161026001, "grad_norm": 0.04536191001534462, "global_step": 275, "epoch": 7, "lr": 0.001} +{"train_loss": 0.061716243624687195, "train_loss_bc": 0.061716243624687195, "train_loss_llm": 0.0, "grad_norm": 0.04844403266906738, "global_step": 276, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06168093532323837, "train_loss_bc": 0.06164824962615967, "train_loss_llm": 0.0032683908939361572, "grad_norm": 0.02023092657327652, "global_step": 277, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05636126920580864, "train_loss_bc": 0.05635423958301544, "train_loss_llm": 0.0007027983665466309, "grad_norm": 0.03568748012185097, "global_step": 278, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06374143064022064, "train_loss_bc": 0.06374143064022064, "train_loss_llm": 0.0, "grad_norm": 0.04037167876958847, "global_step": 279, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06419119238853455, "train_loss_bc": 0.06416424363851547, "train_loss_llm": 0.0026951730251312256, "grad_norm": 0.05832806974649429, "global_step": 280, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05350833386182785, "train_loss_bc": 0.05350833386182785, "train_loss_llm": 0.0, "grad_norm": 0.01812378130853176, "global_step": 281, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06371966749429703, "train_loss_bc": 0.06371848285198212, "train_loss_llm": 0.00011810660362243652, "grad_norm": 0.02694198116660118, "global_step": 282, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06105432286858559, "train_loss_bc": 0.06105432286858559, "train_loss_llm": 0.0, "grad_norm": 0.043402865529060364, "global_step": 283, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06045055761933327, "train_loss_bc": 0.06039811298251152, "train_loss_llm": 0.005244612693786621, "grad_norm": 0.052693210542201996, "global_step": 284, "epoch": 7, "lr": 0.001} +{"train_loss": 0.07064253836870193, "train_loss_bc": 0.07064253836870193, "train_loss_llm": 0.0, "grad_norm": 0.022619768977165222, "global_step": 285, "epoch": 7, "lr": 0.001} +{"train_loss": 0.05889289826154709, "train_loss_bc": 0.05889289826154709, "train_loss_llm": 0.0, "grad_norm": 0.02943265251815319, "global_step": 286, "epoch": 7, "lr": 0.001} +{"train_loss": 0.06001961614108748, "train_loss_bc": 0.06314142048358917, "train_loss_llm": 0.0, "grad_norm": 0.04291527718305588, "global_step": 287, "epoch": 7, "lr": 0.001, "val_loss": 0.05460534989833832} +{"train_loss": 0.06467384845018387, "train_loss_bc": 0.06463705003261566, "train_loss_llm": 0.0036799609661102295, "grad_norm": 0.05663304403424263, "global_step": 288, "epoch": 8, "lr": 0.001} +{"train_loss": 0.06569677591323853, "train_loss_bc": 0.0656815841794014, "train_loss_llm": 0.0015195012092590332, "grad_norm": 0.02027694694697857, "global_step": 289, "epoch": 8, "lr": 0.001} +{"train_loss": 0.05806124210357666, "train_loss_bc": 0.05801801010966301, "train_loss_llm": 0.004323214292526245, "grad_norm": 0.027804506942629814, "global_step": 290, "epoch": 8, "lr": 0.001} +{"train_loss": 0.047821350395679474, "train_loss_bc": 0.04776664078235626, "train_loss_llm": 0.005470842123031616, "grad_norm": 0.03705855831503868, "global_step": 291, "epoch": 8, "lr": 0.001} +{"train_loss": 0.05686245113611221, "train_loss_bc": 0.05685644969344139, "train_loss_llm": 0.0006002187728881836, "grad_norm": 0.040800899267196655, "global_step": 292, "epoch": 8, "lr": 0.001} +{"train_loss": 0.058828797191381454, "train_loss_bc": 0.05878676846623421, "train_loss_llm": 0.0042028725147247314, "grad_norm": 0.015014877542853355, "global_step": 293, "epoch": 8, "lr": 0.001} +{"train_loss": 0.06216036528348923, "train_loss_bc": 0.06216036528348923, "train_loss_llm": 0.0, "grad_norm": 0.024678325280547142, "global_step": 294, "epoch": 8, "lr": 0.001} +{"train_loss": 0.05833727493882179, "train_loss_bc": 0.05831889063119888, "train_loss_llm": 0.0018382668495178223, "grad_norm": 0.032366376370191574, "global_step": 295, "epoch": 8, "lr": 0.001} +{"train_loss": 0.05386917293071747, "train_loss_bc": 0.05386567488312721, "train_loss_llm": 0.0003498196601867676, "grad_norm": 0.03566116839647293, "global_step": 296, "epoch": 8, "lr": 0.001} +{"train_loss": 0.05904914066195488, "train_loss_bc": 0.05904439091682434, "train_loss_llm": 0.0004750490188598633, "grad_norm": 0.023196382448077202, "global_step": 297, "epoch": 8, "lr": 0.001} +{"train_loss": 0.0636909231543541, "train_loss_bc": 0.0636620745062828, "train_loss_llm": 0.0028845667839050293, "grad_norm": 0.032286446541547775, "global_step": 298, "epoch": 8, "lr": 0.001} +{"train_loss": 0.06350472569465637, "train_loss_bc": 0.06350472569465637, "train_loss_llm": 0.0, "grad_norm": 0.04246988892555237, "global_step": 299, "epoch": 8, "lr": 0.001} +{"train_loss": 0.052968207746744156, "train_loss_bc": 0.052958063781261444, "train_loss_llm": 0.0010142922401428223, "grad_norm": 0.049026377499103546, "global_step": 300, "epoch": 8, "lr": 0.001} +{"train_loss": 0.054505638778209686, "train_loss_bc": 0.054505638778209686, "train_loss_llm": 0.0, "grad_norm": 0.019438637420535088, "global_step": 301, "epoch": 8, "lr": 0.001} +{"train_loss": 0.05737936124205589, "train_loss_bc": 0.05737936124205589, "train_loss_llm": 0.0, "grad_norm": 0.025293387472629547, "global_step": 302, "epoch": 8, "lr": 0.001} +{"train_loss": 0.06798258423805237, "train_loss_bc": 0.06794892996549606, "train_loss_llm": 0.0033652186393737793, "grad_norm": 0.03362920507788658, "global_step": 303, "epoch": 8, "lr": 0.001} +{"train_loss": 0.0641881451010704, "train_loss_bc": 0.0641881451010704, "train_loss_llm": 0.0, "grad_norm": 0.04941610246896744, "global_step": 304, "epoch": 8, "lr": 0.001} +{"train_loss": 0.0542001947760582, "train_loss_bc": 0.0542001947760582, "train_loss_llm": 0.0, "grad_norm": 0.015820778906345367, "global_step": 305, "epoch": 8, "lr": 0.001} +{"train_loss": 0.049646347761154175, "train_loss_bc": 0.04961442947387695, "train_loss_llm": 0.0031917691230773926, "grad_norm": 0.027747323736548424, "global_step": 306, "epoch": 8, "lr": 0.001} +{"train_loss": 0.053040653467178345, "train_loss_bc": 0.05302075296640396, "train_loss_llm": 0.0019898712635040283, "grad_norm": 0.034892357885837555, "global_step": 307, "epoch": 8, "lr": 0.001} +{"train_loss": 0.06058806553483009, "train_loss_bc": 0.06058806553483009, "train_loss_llm": 0.0, "grad_norm": 0.04777761921286583, "global_step": 308, "epoch": 8, "lr": 0.001} +{"train_loss": 0.06651228666305542, "train_loss_bc": 0.06651228666305542, "train_loss_llm": 0.0, "grad_norm": 0.015666432678699493, "global_step": 309, "epoch": 8, "lr": 0.001} +{"train_loss": 0.05585971102118492, "train_loss_bc": 0.05581320449709892, "train_loss_llm": 0.004650771617889404, "grad_norm": 0.026844076812267303, "global_step": 310, "epoch": 8, "lr": 0.001} +{"train_loss": 0.06785425543785095, "train_loss_bc": 0.06784246116876602, "train_loss_llm": 0.0011792182922363281, "grad_norm": 0.04301225021481514, "global_step": 311, "epoch": 8, "lr": 0.001} +{"train_loss": 0.05522586777806282, "train_loss_bc": 0.05522586777806282, "train_loss_llm": 0.0, "grad_norm": 0.06125212460756302, "global_step": 312, "epoch": 8, "lr": 0.001} +{"train_loss": 0.059800293296575546, "train_loss_bc": 0.059800293296575546, "train_loss_llm": 0.0, "grad_norm": 0.02388831228017807, "global_step": 313, "epoch": 8, "lr": 0.001} +{"train_loss": 0.06424391269683838, "train_loss_bc": 0.06424258649349213, "train_loss_llm": 0.0001329183578491211, "grad_norm": 0.03767275810241699, "global_step": 314, "epoch": 8, "lr": 0.001} +{"train_loss": 0.061842046678066254, "train_loss_bc": 0.061842046678066254, "train_loss_llm": 0.0, "grad_norm": 0.039205826818943024, "global_step": 315, "epoch": 8, "lr": 0.001} +{"train_loss": 0.054501086473464966, "train_loss_bc": 0.054501086473464966, "train_loss_llm": 0.0, "grad_norm": 0.048118166625499725, "global_step": 316, "epoch": 8, "lr": 0.001} +{"train_loss": 0.057254429906606674, "train_loss_bc": 0.05721772834658623, "train_loss_llm": 0.003670036792755127, "grad_norm": 0.01605859585106373, "global_step": 317, "epoch": 8, "lr": 0.001} +{"train_loss": 0.05538579076528549, "train_loss_bc": 0.05537313595414162, "train_loss_llm": 0.001265406608581543, "grad_norm": 0.024524947628378868, "global_step": 318, "epoch": 8, "lr": 0.001} +{"train_loss": 0.045454658567905426, "train_loss_bc": 0.045449983328580856, "train_loss_llm": 0.0004673600196838379, "grad_norm": 0.033202070742845535, "global_step": 319, "epoch": 8, "lr": 0.001} +{"train_loss": 0.059891898185014725, "train_loss_bc": 0.059891898185014725, "train_loss_llm": 0.0, "grad_norm": 0.04195922240614891, "global_step": 320, "epoch": 8, "lr": 0.001} +{"train_loss": 0.0457860566675663, "train_loss_bc": 0.0457860566675663, "train_loss_llm": 0.0, "grad_norm": 0.016420990228652954, "global_step": 321, "epoch": 8, "lr": 0.001} +{"train_loss": 0.0522875115275383, "train_loss_bc": 0.0522875115275383, "train_loss_llm": 0.0, "grad_norm": 0.022013558074831963, "global_step": 322, "epoch": 8, "lr": 0.001} +{"train_loss": 0.05789630549649397, "train_loss_bc": 0.05530063807964325, "train_loss_llm": 0.0011288821697235107, "grad_norm": 0.02970026805996895, "global_step": 323, "epoch": 8, "lr": 0.001, "val_loss": 0.053425200283527374} +{"train_loss": 0.057547830045223236, "train_loss_bc": 0.05753973126411438, "train_loss_llm": 0.0008098483085632324, "grad_norm": 0.03992471471428871, "global_step": 324, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05335819348692894, "train_loss_bc": 0.05335819348692894, "train_loss_llm": 0.0, "grad_norm": 0.013436920940876007, "global_step": 325, "epoch": 9, "lr": 0.001} +{"train_loss": 0.04463016614317894, "train_loss_bc": 0.04462656378746033, "train_loss_llm": 0.0003603994846343994, "grad_norm": 0.022462734952569008, "global_step": 326, "epoch": 9, "lr": 0.001} +{"train_loss": 0.055975187569856644, "train_loss_bc": 0.055975187569856644, "train_loss_llm": 0.0, "grad_norm": 0.03418093919754028, "global_step": 327, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05004965886473656, "train_loss_bc": 0.05004965886473656, "train_loss_llm": 0.0, "grad_norm": 0.035535991191864014, "global_step": 328, "epoch": 9, "lr": 0.001} +{"train_loss": 0.06134308874607086, "train_loss_bc": 0.06132784113287926, "train_loss_llm": 0.0015248358249664307, "grad_norm": 0.015258165076375008, "global_step": 329, "epoch": 9, "lr": 0.001} +{"train_loss": 0.06538324058055878, "train_loss_bc": 0.06536760181188583, "train_loss_llm": 0.0015635192394256592, "grad_norm": 0.025148645043373108, "global_step": 330, "epoch": 9, "lr": 0.001} +{"train_loss": 0.06500710546970367, "train_loss_bc": 0.06500710546970367, "train_loss_llm": 0.0, "grad_norm": 0.03383837640285492, "global_step": 331, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05101141706109047, "train_loss_bc": 0.05101141706109047, "train_loss_llm": 0.0, "grad_norm": 0.037815481424331665, "global_step": 332, "epoch": 9, "lr": 0.001} +{"train_loss": 0.051677629351615906, "train_loss_bc": 0.051677629351615906, "train_loss_llm": 0.0, "grad_norm": 0.020448872819542885, "global_step": 333, "epoch": 9, "lr": 0.001} +{"train_loss": 0.061324264854192734, "train_loss_bc": 0.06131315603852272, "train_loss_llm": 0.0011109709739685059, "grad_norm": 0.030069401487708092, "global_step": 334, "epoch": 9, "lr": 0.001} +{"train_loss": 0.059066470712423325, "train_loss_bc": 0.05904271826148033, "train_loss_llm": 0.002375185489654541, "grad_norm": 0.041523344814777374, "global_step": 335, "epoch": 9, "lr": 0.001} +{"train_loss": 0.056401900947093964, "train_loss_bc": 0.056383438408374786, "train_loss_llm": 0.0018461346626281738, "grad_norm": 0.045276034623384476, "global_step": 336, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05358367785811424, "train_loss_bc": 0.05358367785811424, "train_loss_llm": 0.0, "grad_norm": 0.017353305593132973, "global_step": 337, "epoch": 9, "lr": 0.001} +{"train_loss": 0.06318695843219757, "train_loss_bc": 0.06314168870449066, "train_loss_llm": 0.004527151584625244, "grad_norm": 0.022925036028027534, "global_step": 338, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05784209072589874, "train_loss_bc": 0.05784209072589874, "train_loss_llm": 0.0, "grad_norm": 0.02772483043372631, "global_step": 339, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05361068621277809, "train_loss_bc": 0.053576353937387466, "train_loss_llm": 0.0034331977367401123, "grad_norm": 0.03447410836815834, "global_step": 340, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05958077311515808, "train_loss_bc": 0.059578292071819305, "train_loss_llm": 0.00024819374084472656, "grad_norm": 0.017147284001111984, "global_step": 341, "epoch": 9, "lr": 0.001} +{"train_loss": 0.06329365074634552, "train_loss_bc": 0.06327866017818451, "train_loss_llm": 0.0014994144439697266, "grad_norm": 0.0350462906062603, "global_step": 342, "epoch": 9, "lr": 0.001} +{"train_loss": 0.06110391020774841, "train_loss_bc": 0.06104971095919609, "train_loss_llm": 0.005419820547103882, "grad_norm": 0.05347765237092972, "global_step": 343, "epoch": 9, "lr": 0.001} +{"train_loss": 0.04854653403162956, "train_loss_bc": 0.04848003759980202, "train_loss_llm": 0.0066495537757873535, "grad_norm": 0.062384139746427536, "global_step": 344, "epoch": 9, "lr": 0.001} +{"train_loss": 0.053870633244514465, "train_loss_bc": 0.053870633244514465, "train_loss_llm": 0.0, "grad_norm": 0.0177169106900692, "global_step": 345, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05223672464489937, "train_loss_bc": 0.052227627485990524, "train_loss_llm": 0.0009096264839172363, "grad_norm": 0.03187365457415581, "global_step": 346, "epoch": 9, "lr": 0.001} +{"train_loss": 0.06435777992010117, "train_loss_bc": 0.06429703533649445, "train_loss_llm": 0.00607454776763916, "grad_norm": 0.037116020917892456, "global_step": 347, "epoch": 9, "lr": 0.001} +{"train_loss": 0.04686807841062546, "train_loss_bc": 0.046863995492458344, "train_loss_llm": 0.000408172607421875, "grad_norm": 0.0448499359190464, "global_step": 348, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05984696373343468, "train_loss_bc": 0.0597749724984169, "train_loss_llm": 0.007199078798294067, "grad_norm": 0.02443735860288143, "global_step": 349, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05236385762691498, "train_loss_bc": 0.05236385762691498, "train_loss_llm": 0.0, "grad_norm": 0.030488906428217888, "global_step": 350, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05381534993648529, "train_loss_bc": 0.05380324274301529, "train_loss_llm": 0.001210540533065796, "grad_norm": 0.03954392299056053, "global_step": 351, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05652456730604172, "train_loss_bc": 0.05644542723894119, "train_loss_llm": 0.007914125919342041, "grad_norm": 0.04093939810991287, "global_step": 352, "epoch": 9, "lr": 0.001} +{"train_loss": 0.06001013517379761, "train_loss_bc": 0.059998031705617905, "train_loss_llm": 0.0012105107307434082, "grad_norm": 0.014754610136151314, "global_step": 353, "epoch": 9, "lr": 0.001} +{"train_loss": 0.0579654835164547, "train_loss_bc": 0.05793457478284836, "train_loss_llm": 0.0030907094478607178, "grad_norm": 0.028260424733161926, "global_step": 354, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05850602313876152, "train_loss_bc": 0.05849442258477211, "train_loss_llm": 0.0011599957942962646, "grad_norm": 0.034066345542669296, "global_step": 355, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05412996560335159, "train_loss_bc": 0.05410563573241234, "train_loss_llm": 0.0024328231811523438, "grad_norm": 0.04671750217676163, "global_step": 356, "epoch": 9, "lr": 0.001} +{"train_loss": 0.062045320868492126, "train_loss_bc": 0.06201131269335747, "train_loss_llm": 0.003400951623916626, "grad_norm": 0.017467627301812172, "global_step": 357, "epoch": 9, "lr": 0.001} +{"train_loss": 0.05864517763257027, "train_loss_bc": 0.05864156410098076, "train_loss_llm": 0.00036153197288513184, "grad_norm": 0.025339029729366302, "global_step": 358, "epoch": 9, "lr": 0.001} +{"train_loss": 0.0567943048146036, "train_loss_bc": 0.059820253401994705, "train_loss_llm": 0.006422311067581177, "grad_norm": 0.03911619260907173, "global_step": 359, "epoch": 9, "lr": 0.001, "val_loss": 0.053440604358911514} +{"train_loss": 0.05869801342487335, "train_loss_bc": 0.05869801342487335, "train_loss_llm": 0.0, "grad_norm": 0.04870745167136192, "global_step": 360, "epoch": 10, "lr": 0.001} +{"train_loss": 0.06262849271297455, "train_loss_bc": 0.06257010996341705, "train_loss_llm": 0.005838543176651001, "grad_norm": 0.017750393599271774, "global_step": 361, "epoch": 10, "lr": 0.001} +{"train_loss": 0.06224217638373375, "train_loss_bc": 0.06224217638373375, "train_loss_llm": 0.0, "grad_norm": 0.0208909809589386, "global_step": 362, "epoch": 10, "lr": 0.001} +{"train_loss": 0.058426398783922195, "train_loss_bc": 0.058426398783922195, "train_loss_llm": 0.0, "grad_norm": 0.02868884988129139, "global_step": 363, "epoch": 10, "lr": 0.001} +{"train_loss": 0.058472611010074615, "train_loss_bc": 0.05843616649508476, "train_loss_llm": 0.003644615411758423, "grad_norm": 0.03889380767941475, "global_step": 364, "epoch": 10, "lr": 0.001} +{"train_loss": 0.0527503676712513, "train_loss_bc": 0.05273157358169556, "train_loss_llm": 0.0018794238567352295, "grad_norm": 0.015483994036912918, "global_step": 365, "epoch": 10, "lr": 0.001} +{"train_loss": 0.049168508499860764, "train_loss_bc": 0.04916166141629219, "train_loss_llm": 0.000684589147567749, "grad_norm": 0.022484108805656433, "global_step": 366, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05516976863145828, "train_loss_bc": 0.05511898174881935, "train_loss_llm": 0.005078613758087158, "grad_norm": 0.026479439809918404, "global_step": 367, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05757557973265648, "train_loss_bc": 0.05757557973265648, "train_loss_llm": 0.0, "grad_norm": 0.03473513945937157, "global_step": 368, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05143015831708908, "train_loss_bc": 0.05138213559985161, "train_loss_llm": 0.0048021674156188965, "grad_norm": 0.0169138852506876, "global_step": 369, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05075843259692192, "train_loss_bc": 0.050751883536577225, "train_loss_llm": 0.0006549358367919922, "grad_norm": 0.02391778863966465, "global_step": 370, "epoch": 10, "lr": 0.001} +{"train_loss": 0.04931732639670372, "train_loss_bc": 0.04928861930966377, "train_loss_llm": 0.0028708577156066895, "grad_norm": 0.033696260303258896, "global_step": 371, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05740174651145935, "train_loss_bc": 0.057383183389902115, "train_loss_llm": 0.001856386661529541, "grad_norm": 0.04491938650608063, "global_step": 372, "epoch": 10, "lr": 0.001} +{"train_loss": 0.06939499825239182, "train_loss_bc": 0.06939499825239182, "train_loss_llm": 0.0, "grad_norm": 0.016682324931025505, "global_step": 373, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05118647217750549, "train_loss_bc": 0.051136478781700134, "train_loss_llm": 0.004999220371246338, "grad_norm": 0.023986777290701866, "global_step": 374, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05367520824074745, "train_loss_bc": 0.05367520824074745, "train_loss_llm": 0.0, "grad_norm": 0.03106469288468361, "global_step": 375, "epoch": 10, "lr": 0.001} +{"train_loss": 0.057194385677576065, "train_loss_bc": 0.057194385677576065, "train_loss_llm": 0.0, "grad_norm": 0.04027155786752701, "global_step": 376, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05363523215055466, "train_loss_bc": 0.05363187938928604, "train_loss_llm": 0.0003350973129272461, "grad_norm": 0.016320589929819107, "global_step": 377, "epoch": 10, "lr": 0.001} +{"train_loss": 0.06129070371389389, "train_loss_bc": 0.061256881803274155, "train_loss_llm": 0.003382056951522827, "grad_norm": 0.027828503400087357, "global_step": 378, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05017833039164543, "train_loss_bc": 0.0501323901116848, "train_loss_llm": 0.004594206809997559, "grad_norm": 0.033429499715566635, "global_step": 379, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05279533192515373, "train_loss_bc": 0.05273439362645149, "train_loss_llm": 0.006093859672546387, "grad_norm": 0.042008254677057266, "global_step": 380, "epoch": 10, "lr": 0.001} +{"train_loss": 0.06119384244084358, "train_loss_bc": 0.06115071475505829, "train_loss_llm": 0.004312872886657715, "grad_norm": 0.02000274695456028, "global_step": 381, "epoch": 10, "lr": 0.001} +{"train_loss": 0.053416185081005096, "train_loss_bc": 0.053379397839307785, "train_loss_llm": 0.003678649663925171, "grad_norm": 0.029650453478097916, "global_step": 382, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05323241278529167, "train_loss_bc": 0.05323241278529167, "train_loss_llm": 0.0, "grad_norm": 0.03395348787307739, "global_step": 383, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05295678973197937, "train_loss_bc": 0.05295678973197937, "train_loss_llm": 0.0, "grad_norm": 0.03897923231124878, "global_step": 384, "epoch": 10, "lr": 0.001} +{"train_loss": 0.06903157383203506, "train_loss_bc": 0.06903157383203506, "train_loss_llm": 0.0, "grad_norm": 0.018818408250808716, "global_step": 385, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05933724716305733, "train_loss_bc": 0.05933724716305733, "train_loss_llm": 0.0, "grad_norm": 0.022152047604322433, "global_step": 386, "epoch": 10, "lr": 0.001} +{"train_loss": 0.0429045669734478, "train_loss_bc": 0.0428876169025898, "train_loss_llm": 0.0016951262950897217, "grad_norm": 0.03350174427032471, "global_step": 387, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05639615282416344, "train_loss_bc": 0.05639615282416344, "train_loss_llm": 0.0, "grad_norm": 0.03535538539290428, "global_step": 388, "epoch": 10, "lr": 0.001} +{"train_loss": 0.0532563291490078, "train_loss_bc": 0.053220659494400024, "train_loss_llm": 0.0035669803619384766, "grad_norm": 0.01629580743610859, "global_step": 389, "epoch": 10, "lr": 0.001} +{"train_loss": 0.054289232939481735, "train_loss_bc": 0.054289232939481735, "train_loss_llm": 0.0, "grad_norm": 0.022637126967310905, "global_step": 390, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05773071199655533, "train_loss_bc": 0.05771276354789734, "train_loss_llm": 0.0017948150634765625, "grad_norm": 0.027201339602470398, "global_step": 391, "epoch": 10, "lr": 0.001} +{"train_loss": 0.04997510090470314, "train_loss_bc": 0.049925658851861954, "train_loss_llm": 0.004944354295730591, "grad_norm": 0.03530190512537956, "global_step": 392, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05243561789393425, "train_loss_bc": 0.05243561789393425, "train_loss_llm": 0.0, "grad_norm": 0.017819685861468315, "global_step": 393, "epoch": 10, "lr": 0.001} +{"train_loss": 0.052044883370399475, "train_loss_bc": 0.05203702673316002, "train_loss_llm": 0.000785529613494873, "grad_norm": 0.032273851335048676, "global_step": 394, "epoch": 10, "lr": 0.001} +{"train_loss": 0.05562482318944401, "train_loss_bc": 0.06082981079816818, "train_loss_llm": 0.007293492555618286, "grad_norm": 0.03728695958852768, "global_step": 395, "epoch": 10, "lr": 0.001, "train/cumulative_reward": 3.9853673835137293, "train/mean_score": 0.16621109832563902, "train/success_rate": 0.0, "test/cumulative_reward": 3.9786811768755466, "test/mean_score": 0.37267120029250284, "test/success_rate": 0.0, "val_loss": 0.05296405404806137, "train_action_mse_error": 0.057585373520851135} +{"train_loss": 0.04645763710141182, "train_loss_bc": 0.046415627002716064, "train_loss_llm": 0.004201024770736694, "grad_norm": 0.0446164533495903, "global_step": 396, "epoch": 11, "lr": 0.001} +{"train_loss": 0.062142547219991684, "train_loss_bc": 0.062101781368255615, "train_loss_llm": 0.004076540470123291, "grad_norm": 0.016615889966487885, "global_step": 397, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05629865452647209, "train_loss_bc": 0.05629865452647209, "train_loss_llm": 0.0, "grad_norm": 0.024958457797765732, "global_step": 398, "epoch": 11, "lr": 0.001} +{"train_loss": 0.06380895525217056, "train_loss_bc": 0.06380895525217056, "train_loss_llm": 0.0, "grad_norm": 0.03061787039041519, "global_step": 399, "epoch": 11, "lr": 0.001} +{"train_loss": 0.055503543466329575, "train_loss_bc": 0.05546458810567856, "train_loss_llm": 0.0038954615592956543, "grad_norm": 0.03720258176326752, "global_step": 400, "epoch": 11, "lr": 0.001} +{"train_loss": 0.052608560770750046, "train_loss_bc": 0.05260771885514259, "train_loss_llm": 8.404254913330078e-05, "grad_norm": 0.017658505588769913, "global_step": 401, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05055428668856621, "train_loss_bc": 0.05055428668856621, "train_loss_llm": 0.0, "grad_norm": 0.02939501591026783, "global_step": 402, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05749828740954399, "train_loss_bc": 0.05749828740954399, "train_loss_llm": 0.0, "grad_norm": 0.038115598261356354, "global_step": 403, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05324948579072952, "train_loss_bc": 0.05324817821383476, "train_loss_llm": 0.00013071298599243164, "grad_norm": 0.04594617336988449, "global_step": 404, "epoch": 11, "lr": 0.001} +{"train_loss": 0.04921271279454231, "train_loss_bc": 0.04917967692017555, "train_loss_llm": 0.003303736448287964, "grad_norm": 0.018596990033984184, "global_step": 405, "epoch": 11, "lr": 0.001} +{"train_loss": 0.058138586580753326, "train_loss_bc": 0.05813402682542801, "train_loss_llm": 0.0004559755325317383, "grad_norm": 0.030905431136488914, "global_step": 406, "epoch": 11, "lr": 0.001} +{"train_loss": 0.04688408598303795, "train_loss_bc": 0.04685677960515022, "train_loss_llm": 0.002730637788772583, "grad_norm": 0.038212236016988754, "global_step": 407, "epoch": 11, "lr": 0.001} +{"train_loss": 0.04813240095973015, "train_loss_bc": 0.04810164123773575, "train_loss_llm": 0.00307580828666687, "grad_norm": 0.04153737425804138, "global_step": 408, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05837138742208481, "train_loss_bc": 0.05829392746090889, "train_loss_llm": 0.007746100425720215, "grad_norm": 0.01648503541946411, "global_step": 409, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05733641982078552, "train_loss_bc": 0.057298753410577774, "train_loss_llm": 0.0037667155265808105, "grad_norm": 0.027592450380325317, "global_step": 410, "epoch": 11, "lr": 0.001} +{"train_loss": 0.057003431022167206, "train_loss_bc": 0.05698999762535095, "train_loss_llm": 0.0013435184955596924, "grad_norm": 0.03243814408779144, "global_step": 411, "epoch": 11, "lr": 0.001} +{"train_loss": 0.04858312010765076, "train_loss_bc": 0.04854549095034599, "train_loss_llm": 0.0037629306316375732, "grad_norm": 0.03935850039124489, "global_step": 412, "epoch": 11, "lr": 0.001} +{"train_loss": 0.058557357639074326, "train_loss_bc": 0.05852885544300079, "train_loss_llm": 0.002850264310836792, "grad_norm": 0.01669768989086151, "global_step": 413, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05392623692750931, "train_loss_bc": 0.05390162393450737, "train_loss_llm": 0.002461254596710205, "grad_norm": 0.025255942717194557, "global_step": 414, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05184873566031456, "train_loss_bc": 0.05184467136859894, "train_loss_llm": 0.00040647387504577637, "grad_norm": 0.030571598559617996, "global_step": 415, "epoch": 11, "lr": 0.001} +{"train_loss": 0.06055910885334015, "train_loss_bc": 0.06050185114145279, "train_loss_llm": 0.00572580099105835, "grad_norm": 0.030812866985797882, "global_step": 416, "epoch": 11, "lr": 0.001} +{"train_loss": 0.049395401030778885, "train_loss_bc": 0.049347683787345886, "train_loss_llm": 0.00477185845375061, "grad_norm": 0.018191369250416756, "global_step": 417, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05984077975153923, "train_loss_bc": 0.05981845781207085, "train_loss_llm": 0.0022322535514831543, "grad_norm": 0.02345619909465313, "global_step": 418, "epoch": 11, "lr": 0.001} +{"train_loss": 0.06966403871774673, "train_loss_bc": 0.06966403871774673, "train_loss_llm": 0.0, "grad_norm": 0.03294474631547928, "global_step": 419, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05719407647848129, "train_loss_bc": 0.05719407647848129, "train_loss_llm": 0.0, "grad_norm": 0.04188127443194389, "global_step": 420, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05124368146061897, "train_loss_bc": 0.05123613402247429, "train_loss_llm": 0.0007547140121459961, "grad_norm": 0.016249150037765503, "global_step": 421, "epoch": 11, "lr": 0.001} +{"train_loss": 0.0562683641910553, "train_loss_bc": 0.05626777932047844, "train_loss_llm": 5.85019588470459e-05, "grad_norm": 0.02711968496441841, "global_step": 422, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05611473694443703, "train_loss_bc": 0.05609104782342911, "train_loss_llm": 0.0023688971996307373, "grad_norm": 0.03156057372689247, "global_step": 423, "epoch": 11, "lr": 0.001} +{"train_loss": 0.055130284279584885, "train_loss_bc": 0.055130284279584885, "train_loss_llm": 0.0, "grad_norm": 0.03497987240552902, "global_step": 424, "epoch": 11, "lr": 0.001} +{"train_loss": 0.054380856454372406, "train_loss_bc": 0.054380856454372406, "train_loss_llm": 0.0, "grad_norm": 0.01781068742275238, "global_step": 425, "epoch": 11, "lr": 0.001} +{"train_loss": 0.051953669637441635, "train_loss_bc": 0.05192581191658974, "train_loss_llm": 0.00278589129447937, "grad_norm": 0.02735717035830021, "global_step": 426, "epoch": 11, "lr": 0.001} +{"train_loss": 0.03859017789363861, "train_loss_bc": 0.03859017789363861, "train_loss_llm": 0.0, "grad_norm": 0.031819287687540054, "global_step": 427, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05885597690939903, "train_loss_bc": 0.058797020465135574, "train_loss_llm": 0.005895644426345825, "grad_norm": 0.04293760657310486, "global_step": 428, "epoch": 11, "lr": 0.001} +{"train_loss": 0.0537613108754158, "train_loss_bc": 0.05375077947974205, "train_loss_llm": 0.0010530650615692139, "grad_norm": 0.016843833029270172, "global_step": 429, "epoch": 11, "lr": 0.001} +{"train_loss": 0.05571020767092705, "train_loss_bc": 0.05571020767092705, "train_loss_llm": 0.0, "grad_norm": 0.026604635640978813, "global_step": 430, "epoch": 11, "lr": 0.001} +{"train_loss": 0.0545721718420585, "train_loss_bc": 0.04981908202171326, "train_loss_llm": 0.0, "grad_norm": 0.03574055805802345, "global_step": 431, "epoch": 11, "lr": 0.001, "val_loss": 0.05215999856591225} +{"train_loss": 0.05280902609229088, "train_loss_bc": 0.05280902609229088, "train_loss_llm": 0.0, "grad_norm": 0.04813385009765625, "global_step": 432, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05836417153477669, "train_loss_bc": 0.05833146721124649, "train_loss_llm": 0.003270477056503296, "grad_norm": 0.013335930183529854, "global_step": 433, "epoch": 12, "lr": 0.001} +{"train_loss": 0.044187311083078384, "train_loss_bc": 0.0441787987947464, "train_loss_llm": 0.0008511245250701904, "grad_norm": 0.01961112953722477, "global_step": 434, "epoch": 12, "lr": 0.001} +{"train_loss": 0.0504629872739315, "train_loss_bc": 0.05044454336166382, "train_loss_llm": 0.0018444657325744629, "grad_norm": 0.02695925533771515, "global_step": 435, "epoch": 12, "lr": 0.001} +{"train_loss": 0.051737334579229355, "train_loss_bc": 0.05171637982130051, "train_loss_llm": 0.002095431089401245, "grad_norm": 0.03660678490996361, "global_step": 436, "epoch": 12, "lr": 0.001} +{"train_loss": 0.050693005323410034, "train_loss_bc": 0.05068797618150711, "train_loss_llm": 0.000502854585647583, "grad_norm": 0.01615212671458721, "global_step": 437, "epoch": 12, "lr": 0.001} +{"train_loss": 0.059600502252578735, "train_loss_bc": 0.05958466976881027, "train_loss_llm": 0.0015834271907806396, "grad_norm": 0.0236215740442276, "global_step": 438, "epoch": 12, "lr": 0.001} +{"train_loss": 0.056866731494665146, "train_loss_bc": 0.056866731494665146, "train_loss_llm": 0.0, "grad_norm": 0.02839084342122078, "global_step": 439, "epoch": 12, "lr": 0.001} +{"train_loss": 0.0493704229593277, "train_loss_bc": 0.04930836334824562, "train_loss_llm": 0.006206095218658447, "grad_norm": 0.03254980966448784, "global_step": 440, "epoch": 12, "lr": 0.001} +{"train_loss": 0.04894590005278587, "train_loss_bc": 0.04890255630016327, "train_loss_llm": 0.004334449768066406, "grad_norm": 0.019673995673656464, "global_step": 441, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05008834972977638, "train_loss_bc": 0.05006527528166771, "train_loss_llm": 0.002307385206222534, "grad_norm": 0.03316047042608261, "global_step": 442, "epoch": 12, "lr": 0.001} +{"train_loss": 0.0630301758646965, "train_loss_bc": 0.0630134642124176, "train_loss_llm": 0.0016708672046661377, "grad_norm": 0.040266841650009155, "global_step": 443, "epoch": 12, "lr": 0.001} +{"train_loss": 0.049766357988119125, "train_loss_bc": 0.049766357988119125, "train_loss_llm": 0.0, "grad_norm": 0.041202954947948456, "global_step": 444, "epoch": 12, "lr": 0.001} +{"train_loss": 0.040168844163417816, "train_loss_bc": 0.040168844163417816, "train_loss_llm": 0.0, "grad_norm": 0.014036710374057293, "global_step": 445, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05169323831796646, "train_loss_bc": 0.051638163626194, "train_loss_llm": 0.005507469177246094, "grad_norm": 0.019264420494437218, "global_step": 446, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05390624329447746, "train_loss_bc": 0.0538765974342823, "train_loss_llm": 0.0029647648334503174, "grad_norm": 0.02587580867111683, "global_step": 447, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05942376330494881, "train_loss_bc": 0.05936948210000992, "train_loss_llm": 0.005428135395050049, "grad_norm": 0.03332855924963951, "global_step": 448, "epoch": 12, "lr": 0.001} +{"train_loss": 0.054479554295539856, "train_loss_bc": 0.054446641355752945, "train_loss_llm": 0.003291189670562744, "grad_norm": 0.017343338578939438, "global_step": 449, "epoch": 12, "lr": 0.001} +{"train_loss": 0.051081184297800064, "train_loss_bc": 0.051081184297800064, "train_loss_llm": 0.0, "grad_norm": 0.024140071123838425, "global_step": 450, "epoch": 12, "lr": 0.001} +{"train_loss": 0.06120739504694939, "train_loss_bc": 0.06120077148079872, "train_loss_llm": 0.0006622970104217529, "grad_norm": 0.03194906935095787, "global_step": 451, "epoch": 12, "lr": 0.001} +{"train_loss": 0.06321810930967331, "train_loss_bc": 0.06321810930967331, "train_loss_llm": 0.0, "grad_norm": 0.0410754457116127, "global_step": 452, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05495362728834152, "train_loss_bc": 0.054945208132267, "train_loss_llm": 0.0008419156074523926, "grad_norm": 0.014063679613173008, "global_step": 453, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05175914615392685, "train_loss_bc": 0.05175914615392685, "train_loss_llm": 0.0, "grad_norm": 0.024434024468064308, "global_step": 454, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05561878904700279, "train_loss_bc": 0.05561878904700279, "train_loss_llm": 0.0, "grad_norm": 0.03256193920969963, "global_step": 455, "epoch": 12, "lr": 0.001} +{"train_loss": 0.052835866808891296, "train_loss_bc": 0.05282028019428253, "train_loss_llm": 0.001558840274810791, "grad_norm": 0.03746951371431351, "global_step": 456, "epoch": 12, "lr": 0.001} +{"train_loss": 0.058785371482372284, "train_loss_bc": 0.058785371482372284, "train_loss_llm": 0.0, "grad_norm": 0.02148321643471718, "global_step": 457, "epoch": 12, "lr": 0.001} +{"train_loss": 0.055684782564640045, "train_loss_bc": 0.055684782564640045, "train_loss_llm": 0.0, "grad_norm": 0.023735731840133667, "global_step": 458, "epoch": 12, "lr": 0.001} +{"train_loss": 0.055896658450365067, "train_loss_bc": 0.05587804317474365, "train_loss_llm": 0.001861572265625, "grad_norm": 0.030038636177778244, "global_step": 459, "epoch": 12, "lr": 0.001} +{"train_loss": 0.051145270466804504, "train_loss_bc": 0.05113428086042404, "train_loss_llm": 0.0010990798473358154, "grad_norm": 0.037984203547239304, "global_step": 460, "epoch": 12, "lr": 0.001} +{"train_loss": 0.06100178137421608, "train_loss_bc": 0.06100178137421608, "train_loss_llm": 0.0, "grad_norm": 0.02004675567150116, "global_step": 461, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05094822123646736, "train_loss_bc": 0.05094200745224953, "train_loss_llm": 0.0006215572357177734, "grad_norm": 0.02435974031686783, "global_step": 462, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05187280476093292, "train_loss_bc": 0.05186161398887634, "train_loss_llm": 0.0011190176010131836, "grad_norm": 0.03117789514362812, "global_step": 463, "epoch": 12, "lr": 0.001} +{"train_loss": 0.050183724611997604, "train_loss_bc": 0.05017675831913948, "train_loss_llm": 0.0006966590881347656, "grad_norm": 0.03638967126607895, "global_step": 464, "epoch": 12, "lr": 0.001} +{"train_loss": 0.060749221593141556, "train_loss_bc": 0.06073024496436119, "train_loss_llm": 0.0018976926803588867, "grad_norm": 0.02011777088046074, "global_step": 465, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05168675631284714, "train_loss_bc": 0.051651980727910995, "train_loss_llm": 0.003477424383163452, "grad_norm": 0.024524949491024017, "global_step": 466, "epoch": 12, "lr": 0.001} +{"train_loss": 0.05368598115940889, "train_loss_bc": 0.048454415053129196, "train_loss_llm": 0.0018275976181030273, "grad_norm": 0.04044504836201668, "global_step": 467, "epoch": 12, "lr": 0.001, "val_loss": 0.052178822457790375} +{"train_loss": 0.04445702210068703, "train_loss_bc": 0.044452372938394547, "train_loss_llm": 0.00046506524085998535, "grad_norm": 0.045004840940237045, "global_step": 468, "epoch": 13, "lr": 0.001} +{"train_loss": 0.051017846912145615, "train_loss_bc": 0.05099028721451759, "train_loss_llm": 0.0027558207511901855, "grad_norm": 0.014950042590498924, "global_step": 469, "epoch": 13, "lr": 0.001} +{"train_loss": 0.046929869800806046, "train_loss_bc": 0.046929869800806046, "train_loss_llm": 0.0, "grad_norm": 0.024056123569607735, "global_step": 470, "epoch": 13, "lr": 0.001} +{"train_loss": 0.052403636276721954, "train_loss_bc": 0.05239332839846611, "train_loss_llm": 0.0010307729244232178, "grad_norm": 0.024113943800330162, "global_step": 471, "epoch": 13, "lr": 0.001} +{"train_loss": 0.04552203044295311, "train_loss_bc": 0.04550697281956673, "train_loss_llm": 0.001505821943283081, "grad_norm": 0.03198089078068733, "global_step": 472, "epoch": 13, "lr": 0.001} +{"train_loss": 0.054682061076164246, "train_loss_bc": 0.054658442735672, "train_loss_llm": 0.002361863851547241, "grad_norm": 0.01318492740392685, "global_step": 473, "epoch": 13, "lr": 0.001} +{"train_loss": 0.058780066668987274, "train_loss_bc": 0.05877823010087013, "train_loss_llm": 0.0001836717128753662, "grad_norm": 0.019709255546331406, "global_step": 474, "epoch": 13, "lr": 0.001} +{"train_loss": 0.055381014943122864, "train_loss_bc": 0.05534317344427109, "train_loss_llm": 0.0037842988967895508, "grad_norm": 0.023324191570281982, "global_step": 475, "epoch": 13, "lr": 0.001} +{"train_loss": 0.056389134377241135, "train_loss_bc": 0.056389134377241135, "train_loss_llm": 0.0, "grad_norm": 0.03256314620375633, "global_step": 476, "epoch": 13, "lr": 0.001} +{"train_loss": 0.06084250658750534, "train_loss_bc": 0.06084250658750534, "train_loss_llm": 0.0, "grad_norm": 0.018206272274255753, "global_step": 477, "epoch": 13, "lr": 0.001} +{"train_loss": 0.050132349133491516, "train_loss_bc": 0.05010014399886131, "train_loss_llm": 0.0032205581665039062, "grad_norm": 0.024690305814146996, "global_step": 478, "epoch": 13, "lr": 0.001} +{"train_loss": 0.0488346628844738, "train_loss_bc": 0.04880977049469948, "train_loss_llm": 0.0024891793727874756, "grad_norm": 0.02861434407532215, "global_step": 479, "epoch": 13, "lr": 0.001} +{"train_loss": 0.059336692094802856, "train_loss_bc": 0.059336692094802856, "train_loss_llm": 0.0, "grad_norm": 0.029947472736239433, "global_step": 480, "epoch": 13, "lr": 0.001} +{"train_loss": 0.044161394238471985, "train_loss_bc": 0.04414299875497818, "train_loss_llm": 0.0018394887447357178, "grad_norm": 0.015211371704936028, "global_step": 481, "epoch": 13, "lr": 0.001} +{"train_loss": 0.0614357627928257, "train_loss_bc": 0.06141532585024834, "train_loss_llm": 0.002043783664703369, "grad_norm": 0.020624304190278053, "global_step": 482, "epoch": 13, "lr": 0.001} +{"train_loss": 0.0502556748688221, "train_loss_bc": 0.0502556748688221, "train_loss_llm": 0.0, "grad_norm": 0.026455266401171684, "global_step": 483, "epoch": 13, "lr": 0.001} +{"train_loss": 0.0497225858271122, "train_loss_bc": 0.0497225858271122, "train_loss_llm": 0.0, "grad_norm": 0.040330346673727036, "global_step": 484, "epoch": 13, "lr": 0.001} +{"train_loss": 0.05206671357154846, "train_loss_bc": 0.05206484720110893, "train_loss_llm": 0.00018674135208129883, "grad_norm": 0.016031071543693542, "global_step": 485, "epoch": 13, "lr": 0.001} +{"train_loss": 0.05314910039305687, "train_loss_bc": 0.05314910039305687, "train_loss_llm": 0.0, "grad_norm": 0.025085868313908577, "global_step": 486, "epoch": 13, "lr": 0.001} +{"train_loss": 0.057773157954216, "train_loss_bc": 0.057741906493902206, "train_loss_llm": 0.0031250715255737305, "grad_norm": 0.032892853021621704, "global_step": 487, "epoch": 13, "lr": 0.001} +{"train_loss": 0.06304274499416351, "train_loss_bc": 0.06298824399709702, "train_loss_llm": 0.005449801683425903, "grad_norm": 0.040918879210948944, "global_step": 488, "epoch": 13, "lr": 0.001} +{"train_loss": 0.05439355969429016, "train_loss_bc": 0.05439355969429016, "train_loss_llm": 0.0, "grad_norm": 0.017553100362420082, "global_step": 489, "epoch": 13, "lr": 0.001} +{"train_loss": 0.053755346685647964, "train_loss_bc": 0.053744636476039886, "train_loss_llm": 0.0010710954666137695, "grad_norm": 0.029224401339888573, "global_step": 490, "epoch": 13, "lr": 0.001} +{"train_loss": 0.04818698763847351, "train_loss_bc": 0.048145536333322525, "train_loss_llm": 0.004145145416259766, "grad_norm": 0.04165388271212578, "global_step": 491, "epoch": 13, "lr": 0.001} +{"train_loss": 0.058611877262592316, "train_loss_bc": 0.058582331985235214, "train_loss_llm": 0.0029543936252593994, "grad_norm": 0.051526281982660294, "global_step": 492, "epoch": 13, "lr": 0.001} +{"train_loss": 0.050182782113552094, "train_loss_bc": 0.0501549057662487, "train_loss_llm": 0.002787768840789795, "grad_norm": 0.015268241986632347, "global_step": 493, "epoch": 13, "lr": 0.001} +{"train_loss": 0.04804367199540138, "train_loss_bc": 0.047987595200538635, "train_loss_llm": 0.005607575178146362, "grad_norm": 0.021117057651281357, "global_step": 494, "epoch": 13, "lr": 0.001} +{"train_loss": 0.04999828338623047, "train_loss_bc": 0.04999828338623047, "train_loss_llm": 0.0, "grad_norm": 0.02954690530896187, "global_step": 495, "epoch": 13, "lr": 0.001} +{"train_loss": 0.052381597459316254, "train_loss_bc": 0.052381597459316254, "train_loss_llm": 0.0, "grad_norm": 0.0330120325088501, "global_step": 496, "epoch": 13, "lr": 0.001} +{"train_loss": 0.05500134453177452, "train_loss_bc": 0.05500134453177452, "train_loss_llm": 0.0, "grad_norm": 0.01822754554450512, "global_step": 497, "epoch": 13, "lr": 0.001} +{"train_loss": 0.047444142401218414, "train_loss_bc": 0.047444142401218414, "train_loss_llm": 0.0, "grad_norm": 0.02251560427248478, "global_step": 498, "epoch": 13, "lr": 0.001} +{"train_loss": 0.06419092416763306, "train_loss_bc": 0.06419092416763306, "train_loss_llm": 0.0, "grad_norm": 0.03074686974287033, "global_step": 499, "epoch": 13, "lr": 0.001} +{"train_loss": 0.05511688068509102, "train_loss_bc": 0.05511369928717613, "train_loss_llm": 0.0003179609775543213, "grad_norm": 0.040387000888586044, "global_step": 500, "epoch": 13, "lr": 0.001} +{"train_loss": 0.048127494752407074, "train_loss_bc": 0.048127494752407074, "train_loss_llm": 0.0, "grad_norm": 0.01809745468199253, "global_step": 501, "epoch": 13, "lr": 0.001} +{"train_loss": 0.049206167459487915, "train_loss_bc": 0.0491974763572216, "train_loss_llm": 0.000869065523147583, "grad_norm": 0.02535359375178814, "global_step": 502, "epoch": 13, "lr": 0.001} +{"train_loss": 0.05313868406746122, "train_loss_bc": 0.06201482191681862, "train_loss_llm": 0.0020716190338134766, "grad_norm": 0.03314005210995674, "global_step": 503, "epoch": 13, "lr": 0.001, "val_loss": 0.05159345269203186} +{"train_loss": 0.05034827068448067, "train_loss_bc": 0.050286050885915756, "train_loss_llm": 0.006221950054168701, "grad_norm": 0.04226822033524513, "global_step": 504, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05038098618388176, "train_loss_bc": 0.050345901399850845, "train_loss_llm": 0.0035084784030914307, "grad_norm": 0.019015252590179443, "global_step": 505, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05936053767800331, "train_loss_bc": 0.05936053767800331, "train_loss_llm": 0.0, "grad_norm": 0.029409470036625862, "global_step": 506, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05375025421380997, "train_loss_bc": 0.053740985691547394, "train_loss_llm": 0.0009267628192901611, "grad_norm": 0.03164864704012871, "global_step": 507, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05083169415593147, "train_loss_bc": 0.050821658223867416, "train_loss_llm": 0.001003563404083252, "grad_norm": 0.03188463672995567, "global_step": 508, "epoch": 14, "lr": 0.001} +{"train_loss": 0.04929475858807564, "train_loss_bc": 0.04925401508808136, "train_loss_llm": 0.004074394702911377, "grad_norm": 0.01566682942211628, "global_step": 509, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05197260528802872, "train_loss_bc": 0.05193358659744263, "train_loss_llm": 0.003902047872543335, "grad_norm": 0.022770514711737633, "global_step": 510, "epoch": 14, "lr": 0.001} +{"train_loss": 0.06040853261947632, "train_loss_bc": 0.060396190732717514, "train_loss_llm": 0.0012340545654296875, "grad_norm": 0.029208026826381683, "global_step": 511, "epoch": 14, "lr": 0.001} +{"train_loss": 0.052191976457834244, "train_loss_bc": 0.052191976457834244, "train_loss_llm": 0.0, "grad_norm": 0.031674668192863464, "global_step": 512, "epoch": 14, "lr": 0.001} +{"train_loss": 0.044210270047187805, "train_loss_bc": 0.044210270047187805, "train_loss_llm": 0.0, "grad_norm": 0.013191834092140198, "global_step": 513, "epoch": 14, "lr": 0.001} +{"train_loss": 0.051013313233852386, "train_loss_bc": 0.050955988466739655, "train_loss_llm": 0.005732357501983643, "grad_norm": 0.02521658129990101, "global_step": 514, "epoch": 14, "lr": 0.001} +{"train_loss": 0.04978834092617035, "train_loss_bc": 0.04972515627741814, "train_loss_llm": 0.006318569183349609, "grad_norm": 0.02980412356555462, "global_step": 515, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05953711271286011, "train_loss_bc": 0.05953711271286011, "train_loss_llm": 0.0, "grad_norm": 0.03581071272492409, "global_step": 516, "epoch": 14, "lr": 0.001} +{"train_loss": 0.0564042292535305, "train_loss_bc": 0.0564042292535305, "train_loss_llm": 0.0, "grad_norm": 0.02044476382434368, "global_step": 517, "epoch": 14, "lr": 0.001} +{"train_loss": 0.0448627695441246, "train_loss_bc": 0.044838350266218185, "train_loss_llm": 0.0024420320987701416, "grad_norm": 0.029683392494916916, "global_step": 518, "epoch": 14, "lr": 0.001} +{"train_loss": 0.054022353142499924, "train_loss_bc": 0.054022353142499924, "train_loss_llm": 0.0, "grad_norm": 0.03970117121934891, "global_step": 519, "epoch": 14, "lr": 0.001} +{"train_loss": 0.04490988701581955, "train_loss_bc": 0.04490988701581955, "train_loss_llm": 0.0, "grad_norm": 0.04533626511693001, "global_step": 520, "epoch": 14, "lr": 0.001} +{"train_loss": 0.047309357672929764, "train_loss_bc": 0.04729754477739334, "train_loss_llm": 0.0011813044548034668, "grad_norm": 0.01781230978667736, "global_step": 521, "epoch": 14, "lr": 0.001} +{"train_loss": 0.04606635496020317, "train_loss_bc": 0.046046409755945206, "train_loss_llm": 0.001994520425796509, "grad_norm": 0.0248578991740942, "global_step": 522, "epoch": 14, "lr": 0.001} +{"train_loss": 0.051753196865320206, "train_loss_bc": 0.05169418081641197, "train_loss_llm": 0.005901545286178589, "grad_norm": 0.03608761355280876, "global_step": 523, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05325638875365257, "train_loss_bc": 0.05325638875365257, "train_loss_llm": 0.0, "grad_norm": 0.0419645719230175, "global_step": 524, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05125242844223976, "train_loss_bc": 0.051236581057310104, "train_loss_llm": 0.0015846192836761475, "grad_norm": 0.017101898789405823, "global_step": 525, "epoch": 14, "lr": 0.001} +{"train_loss": 0.06007711961865425, "train_loss_bc": 0.06007711961865425, "train_loss_llm": 0.0, "grad_norm": 0.027391953393816948, "global_step": 526, "epoch": 14, "lr": 0.001} +{"train_loss": 0.0491824708878994, "train_loss_bc": 0.04917849972844124, "train_loss_llm": 0.00039711594581604004, "grad_norm": 0.0354631170630455, "global_step": 527, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05752885714173317, "train_loss_bc": 0.05752885714173317, "train_loss_llm": 0.0, "grad_norm": 0.0412670262157917, "global_step": 528, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05337494611740112, "train_loss_bc": 0.053372059017419815, "train_loss_llm": 0.0002886950969696045, "grad_norm": 0.017207710072398186, "global_step": 529, "epoch": 14, "lr": 0.001} +{"train_loss": 0.06130996719002724, "train_loss_bc": 0.06130996719002724, "train_loss_llm": 0.0, "grad_norm": 0.025816528126597404, "global_step": 530, "epoch": 14, "lr": 0.001} +{"train_loss": 0.046012379229068756, "train_loss_bc": 0.046012379229068756, "train_loss_llm": 0.0, "grad_norm": 0.030166272073984146, "global_step": 531, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05282023549079895, "train_loss_bc": 0.052766650915145874, "train_loss_llm": 0.005358338356018066, "grad_norm": 0.03260180354118347, "global_step": 532, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05398540198802948, "train_loss_bc": 0.0539691224694252, "train_loss_llm": 0.0016280412673950195, "grad_norm": 0.01691851206123829, "global_step": 533, "epoch": 14, "lr": 0.001} +{"train_loss": 0.0579676516354084, "train_loss_bc": 0.0579676516354084, "train_loss_llm": 0.0, "grad_norm": 0.029219254851341248, "global_step": 534, "epoch": 14, "lr": 0.001} +{"train_loss": 0.05165757238864899, "train_loss_bc": 0.051651909947395325, "train_loss_llm": 0.0005660653114318848, "grad_norm": 0.04050663858652115, "global_step": 535, "epoch": 14, "lr": 0.001} +{"train_loss": 0.04743091017007828, "train_loss_bc": 0.047417204827070236, "train_loss_llm": 0.0013704895973205566, "grad_norm": 0.04710990563035011, "global_step": 536, "epoch": 14, "lr": 0.001} +{"train_loss": 0.049094267189502716, "train_loss_bc": 0.049094267189502716, "train_loss_llm": 0.0, "grad_norm": 0.015270213596522808, "global_step": 537, "epoch": 14, "lr": 0.001} +{"train_loss": 0.06036539375782013, "train_loss_bc": 0.06033368036150932, "train_loss_llm": 0.0031712353229522705, "grad_norm": 0.02436836250126362, "global_step": 538, "epoch": 14, "lr": 0.001} +{"train_loss": 0.0525870266266995, "train_loss_bc": 0.0593411922454834, "train_loss_llm": 0.005897581577301025, "grad_norm": 0.03258056566119194, "global_step": 539, "epoch": 14, "lr": 0.001, "val_loss": 0.05115365609526634} +{"train_loss": 0.05048980563879013, "train_loss_bc": 0.05048292130231857, "train_loss_llm": 0.0006883144378662109, "grad_norm": 0.041018519550561905, "global_step": 540, "epoch": 15, "lr": 0.001} +{"train_loss": 0.04722639173269272, "train_loss_bc": 0.04719390720129013, "train_loss_llm": 0.003248542547225952, "grad_norm": 0.0189408790320158, "global_step": 541, "epoch": 15, "lr": 0.001} +{"train_loss": 0.057610493153333664, "train_loss_bc": 0.057581935077905655, "train_loss_llm": 0.0028558075428009033, "grad_norm": 0.02304726280272007, "global_step": 542, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05489885061979294, "train_loss_bc": 0.05489885061979294, "train_loss_llm": 0.0, "grad_norm": 0.02883157506585121, "global_step": 543, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05475413054227829, "train_loss_bc": 0.054732922464609146, "train_loss_llm": 0.0021208226680755615, "grad_norm": 0.03400120511651039, "global_step": 544, "epoch": 15, "lr": 0.001} +{"train_loss": 0.058154474943876266, "train_loss_bc": 0.058154474943876266, "train_loss_llm": 0.0, "grad_norm": 0.016426362097263336, "global_step": 545, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05782931670546532, "train_loss_bc": 0.05776579678058624, "train_loss_llm": 0.006352126598358154, "grad_norm": 0.021602313965559006, "global_step": 546, "epoch": 15, "lr": 0.001} +{"train_loss": 0.06257612258195877, "train_loss_bc": 0.06252384185791016, "train_loss_llm": 0.0052280426025390625, "grad_norm": 0.02720576710999012, "global_step": 547, "epoch": 15, "lr": 0.001} +{"train_loss": 0.04554140195250511, "train_loss_bc": 0.04554140195250511, "train_loss_llm": 0.0, "grad_norm": 0.03461551293730736, "global_step": 548, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05132337287068367, "train_loss_bc": 0.05129734054207802, "train_loss_llm": 0.0026031434535980225, "grad_norm": 0.01497515570372343, "global_step": 549, "epoch": 15, "lr": 0.001} +{"train_loss": 0.04523294419050217, "train_loss_bc": 0.04523294419050217, "train_loss_llm": 0.0, "grad_norm": 0.021933255717158318, "global_step": 550, "epoch": 15, "lr": 0.001} +{"train_loss": 0.06022646650671959, "train_loss_bc": 0.06022234261035919, "train_loss_llm": 0.00041237473487854004, "grad_norm": 0.027644295245409012, "global_step": 551, "epoch": 15, "lr": 0.001} +{"train_loss": 0.0466025248169899, "train_loss_bc": 0.04659118130803108, "train_loss_llm": 0.0011343061923980713, "grad_norm": 0.02663995884358883, "global_step": 552, "epoch": 15, "lr": 0.001} +{"train_loss": 0.04733286052942276, "train_loss_bc": 0.0473298542201519, "train_loss_llm": 0.0003007650375366211, "grad_norm": 0.015919925644993782, "global_step": 553, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05245313420891762, "train_loss_bc": 0.052431367337703705, "train_loss_llm": 0.002176523208618164, "grad_norm": 0.027135642245411873, "global_step": 554, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05542059987783432, "train_loss_bc": 0.05542059987783432, "train_loss_llm": 0.0, "grad_norm": 0.03373271971940994, "global_step": 555, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05312800407409668, "train_loss_bc": 0.05311891436576843, "train_loss_llm": 0.0009090900421142578, "grad_norm": 0.03919215500354767, "global_step": 556, "epoch": 15, "lr": 0.001} +{"train_loss": 0.054356127977371216, "train_loss_bc": 0.054356127977371216, "train_loss_llm": 0.0, "grad_norm": 0.01656745746731758, "global_step": 557, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05112304538488388, "train_loss_bc": 0.05112304538488388, "train_loss_llm": 0.0, "grad_norm": 0.025869710370898247, "global_step": 558, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05761977285146713, "train_loss_bc": 0.05758382007479668, "train_loss_llm": 0.003595322370529175, "grad_norm": 0.027251675724983215, "global_step": 559, "epoch": 15, "lr": 0.001} +{"train_loss": 0.048789698630571365, "train_loss_bc": 0.048774220049381256, "train_loss_llm": 0.0015478730201721191, "grad_norm": 0.030383745208382607, "global_step": 560, "epoch": 15, "lr": 0.001} +{"train_loss": 0.053764067590236664, "train_loss_bc": 0.05375431478023529, "train_loss_llm": 0.0009753406047821045, "grad_norm": 0.015352161601185799, "global_step": 561, "epoch": 15, "lr": 0.001} +{"train_loss": 0.04712600260972977, "train_loss_bc": 0.04712600260972977, "train_loss_llm": 0.0, "grad_norm": 0.020156987011432648, "global_step": 562, "epoch": 15, "lr": 0.001} +{"train_loss": 0.04545322060585022, "train_loss_bc": 0.04542526602745056, "train_loss_llm": 0.0027953684329986572, "grad_norm": 0.028470072895288467, "global_step": 563, "epoch": 15, "lr": 0.001} +{"train_loss": 0.04696113243699074, "train_loss_bc": 0.04696113243699074, "train_loss_llm": 0.0, "grad_norm": 0.036071669310331345, "global_step": 564, "epoch": 15, "lr": 0.001} +{"train_loss": 0.0528281107544899, "train_loss_bc": 0.052787281572818756, "train_loss_llm": 0.004083067178726196, "grad_norm": 0.016400622203946114, "global_step": 565, "epoch": 15, "lr": 0.001} +{"train_loss": 0.04708882048726082, "train_loss_bc": 0.04707137867808342, "train_loss_llm": 0.001744002103805542, "grad_norm": 0.02616957202553749, "global_step": 566, "epoch": 15, "lr": 0.001} +{"train_loss": 0.04690772667527199, "train_loss_bc": 0.04690772667527199, "train_loss_llm": 0.0, "grad_norm": 0.032520975917577744, "global_step": 567, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05848614498972893, "train_loss_bc": 0.05848614498972893, "train_loss_llm": 0.0, "grad_norm": 0.04160594567656517, "global_step": 568, "epoch": 15, "lr": 0.001} +{"train_loss": 0.050793152302503586, "train_loss_bc": 0.050793152302503586, "train_loss_llm": 0.0, "grad_norm": 0.015688437968492508, "global_step": 569, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05229347199201584, "train_loss_bc": 0.05229347199201584, "train_loss_llm": 0.0, "grad_norm": 0.023834530264139175, "global_step": 570, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05684368312358856, "train_loss_bc": 0.05682097747921944, "train_loss_llm": 0.002270728349685669, "grad_norm": 0.037381742149591446, "global_step": 571, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05133889615535736, "train_loss_bc": 0.05133889615535736, "train_loss_llm": 0.0, "grad_norm": 0.043845970183610916, "global_step": 572, "epoch": 15, "lr": 0.001} +{"train_loss": 0.04262377694249153, "train_loss_bc": 0.04255824536085129, "train_loss_llm": 0.006553202867507935, "grad_norm": 0.016429517418146133, "global_step": 573, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05298413708806038, "train_loss_bc": 0.05298413708806038, "train_loss_llm": 0.0, "grad_norm": 0.023384328931570053, "global_step": 574, "epoch": 15, "lr": 0.001} +{"train_loss": 0.05192149358077182, "train_loss_bc": 0.05098063871264458, "train_loss_llm": 0.0011246800422668457, "grad_norm": 0.030310489237308502, "global_step": 575, "epoch": 15, "lr": 0.001, "train/cumulative_reward": 19.081103563840834, "train/mean_score": 1.1463044260526618, "train/success_rate": 0.0, "test/cumulative_reward": 3.2713016057705095, "test/mean_score": 0.1511373854410639, "test/success_rate": 0.0, "val_loss": 0.051882266998291016, "train_action_mse_error": 0.054335817694664} +{"train_loss": 0.0563444122672081, "train_loss_bc": 0.05629855394363403, "train_loss_llm": 0.004585772752761841, "grad_norm": 0.04124387353658676, "global_step": 576, "epoch": 16, "lr": 0.001} +{"train_loss": 0.06242736801505089, "train_loss_bc": 0.062412749975919724, "train_loss_llm": 0.0014619827270507812, "grad_norm": 0.015995070338249207, "global_step": 577, "epoch": 16, "lr": 0.001} +{"train_loss": 0.037941187620162964, "train_loss_bc": 0.037941187620162964, "train_loss_llm": 0.0, "grad_norm": 0.023910226300358772, "global_step": 578, "epoch": 16, "lr": 0.001} +{"train_loss": 0.05030936375260353, "train_loss_bc": 0.05030936375260353, "train_loss_llm": 0.0, "grad_norm": 0.02880840003490448, "global_step": 579, "epoch": 16, "lr": 0.001} +{"train_loss": 0.051820676773786545, "train_loss_bc": 0.051820676773786545, "train_loss_llm": 0.0, "grad_norm": 0.03600841388106346, "global_step": 580, "epoch": 16, "lr": 0.001} +{"train_loss": 0.0547250397503376, "train_loss_bc": 0.054694272577762604, "train_loss_llm": 0.00307658314704895, "grad_norm": 0.020963359624147415, "global_step": 581, "epoch": 16, "lr": 0.001} +{"train_loss": 0.04179403930902481, "train_loss_bc": 0.041778065264225006, "train_loss_llm": 0.001597285270690918, "grad_norm": 0.02523048408329487, "global_step": 582, "epoch": 16, "lr": 0.001} +{"train_loss": 0.04897158965468407, "train_loss_bc": 0.04896949231624603, "train_loss_llm": 0.00020968914031982422, "grad_norm": 0.030350107699632645, "global_step": 583, "epoch": 16, "lr": 0.001} +{"train_loss": 0.06327594816684723, "train_loss_bc": 0.063270203769207, "train_loss_llm": 0.0005741417407989502, "grad_norm": 0.0331851989030838, "global_step": 584, "epoch": 16, "lr": 0.001} +{"train_loss": 0.04931899905204773, "train_loss_bc": 0.04931899905204773, "train_loss_llm": 0.0, "grad_norm": 0.024780604988336563, "global_step": 585, "epoch": 16, "lr": 0.001} +{"train_loss": 0.048041269183158875, "train_loss_bc": 0.04802316054701805, "train_loss_llm": 0.0018109679222106934, "grad_norm": 0.0285800788551569, "global_step": 586, "epoch": 16, "lr": 0.001} +{"train_loss": 0.04849576577544212, "train_loss_bc": 0.04842748865485191, "train_loss_llm": 0.00682753324508667, "grad_norm": 0.03205893188714981, "global_step": 587, "epoch": 16, "lr": 0.001} +{"train_loss": 0.06232317164540291, "train_loss_bc": 0.06232317164540291, "train_loss_llm": 0.0, "grad_norm": 0.03875717893242836, "global_step": 588, "epoch": 16, "lr": 0.001} +{"train_loss": 0.06145372614264488, "train_loss_bc": 0.06143941357731819, "train_loss_llm": 0.001431196928024292, "grad_norm": 0.015646493062376976, "global_step": 589, "epoch": 16, "lr": 0.001} +{"train_loss": 0.04709344729781151, "train_loss_bc": 0.04707203060388565, "train_loss_llm": 0.0021416842937469482, "grad_norm": 0.023004237562417984, "global_step": 590, "epoch": 16, "lr": 0.001} +{"train_loss": 0.05085188150405884, "train_loss_bc": 0.050818946212530136, "train_loss_llm": 0.0032935738563537598, "grad_norm": 0.03129405528306961, "global_step": 591, "epoch": 16, "lr": 0.001} +{"train_loss": 0.045681606978178024, "train_loss_bc": 0.04568038880825043, "train_loss_llm": 0.00012180209159851074, "grad_norm": 0.038058385252952576, "global_step": 592, "epoch": 16, "lr": 0.001} +{"train_loss": 0.04896516352891922, "train_loss_bc": 0.04896516352891922, "train_loss_llm": 0.0, "grad_norm": 0.014746711589396, "global_step": 593, "epoch": 16, "lr": 0.001} +{"train_loss": 0.05163083225488663, "train_loss_bc": 0.05163083225488663, "train_loss_llm": 0.0, "grad_norm": 0.0240660198032856, "global_step": 594, "epoch": 16, "lr": 0.001} +{"train_loss": 0.05743316188454628, "train_loss_bc": 0.05739147588610649, "train_loss_llm": 0.004168689250946045, "grad_norm": 0.028939595445990562, "global_step": 595, "epoch": 16, "lr": 0.001} +{"train_loss": 0.053845446556806564, "train_loss_bc": 0.053845446556806564, "train_loss_llm": 0.0, "grad_norm": 0.036266688257455826, "global_step": 596, "epoch": 16, "lr": 0.001} +{"train_loss": 0.051406897604465485, "train_loss_bc": 0.05135446414351463, "train_loss_llm": 0.005243360996246338, "grad_norm": 0.014918236061930656, "global_step": 597, "epoch": 16, "lr": 0.001} +{"train_loss": 0.0470610074698925, "train_loss_bc": 0.0470610074698925, "train_loss_llm": 0.0, "grad_norm": 0.025482771918177605, "global_step": 598, "epoch": 16, "lr": 0.001} +{"train_loss": 0.05457746610045433, "train_loss_bc": 0.054511621594429016, "train_loss_llm": 0.006584346294403076, "grad_norm": 0.034947946667671204, "global_step": 599, "epoch": 16, "lr": 0.001} +{"train_loss": 0.048227425664663315, "train_loss_bc": 0.048227425664663315, "train_loss_llm": 0.0, "grad_norm": 0.03513480722904205, "global_step": 600, "epoch": 16, "lr": 0.001} +{"train_loss": 0.053029078990221024, "train_loss_bc": 0.053006283938884735, "train_loss_llm": 0.0022794604301452637, "grad_norm": 0.016650879755616188, "global_step": 601, "epoch": 16, "lr": 0.001} +{"train_loss": 0.04865154251456261, "train_loss_bc": 0.04865154251456261, "train_loss_llm": 0.0, "grad_norm": 0.022403568029403687, "global_step": 602, "epoch": 16, "lr": 0.001} +{"train_loss": 0.04890986531972885, "train_loss_bc": 0.04890986531972885, "train_loss_llm": 0.0, "grad_norm": 0.026411989703774452, "global_step": 603, "epoch": 16, "lr": 0.001} +{"train_loss": 0.042034246027469635, "train_loss_bc": 0.042014267295598984, "train_loss_llm": 0.001997828483581543, "grad_norm": 0.029500482603907585, "global_step": 604, "epoch": 16, "lr": 0.001} +{"train_loss": 0.05178942158818245, "train_loss_bc": 0.05178942158818245, "train_loss_llm": 0.0, "grad_norm": 0.01379373762756586, "global_step": 605, "epoch": 16, "lr": 0.001} +{"train_loss": 0.05058545619249344, "train_loss_bc": 0.050557345151901245, "train_loss_llm": 0.002811044454574585, "grad_norm": 0.021846286952495575, "global_step": 606, "epoch": 16, "lr": 0.001} +{"train_loss": 0.06482535600662231, "train_loss_bc": 0.06482535600662231, "train_loss_llm": 0.0, "grad_norm": 0.022563030943274498, "global_step": 607, "epoch": 16, "lr": 0.001} +{"train_loss": 0.04939040169119835, "train_loss_bc": 0.04935842752456665, "train_loss_llm": 0.003197312355041504, "grad_norm": 0.0287137720733881, "global_step": 608, "epoch": 16, "lr": 0.001} +{"train_loss": 0.052162159234285355, "train_loss_bc": 0.05216182395815849, "train_loss_llm": 3.36766242980957e-05, "grad_norm": 0.01836187206208706, "global_step": 609, "epoch": 16, "lr": 0.001} +{"train_loss": 0.055935442447662354, "train_loss_bc": 0.05593167617917061, "train_loss_llm": 0.00037670135498046875, "grad_norm": 0.030017629265785217, "global_step": 610, "epoch": 16, "lr": 0.001} +{"train_loss": 0.05126328166160318, "train_loss_bc": 0.03414827585220337, "train_loss_llm": 0.0, "grad_norm": 0.04132054001092911, "global_step": 611, "epoch": 16, "lr": 0.001, "val_loss": 0.051290109753608704} +{"train_loss": 0.047619543969631195, "train_loss_bc": 0.047619543969631195, "train_loss_llm": 0.0, "grad_norm": 0.0455770343542099, "global_step": 612, "epoch": 17, "lr": 0.001} +{"train_loss": 0.04009094834327698, "train_loss_bc": 0.04009094834327698, "train_loss_llm": 0.0, "grad_norm": 0.01674078404903412, "global_step": 613, "epoch": 17, "lr": 0.001} +{"train_loss": 0.051471270620822906, "train_loss_bc": 0.051451876759529114, "train_loss_llm": 0.0019392669200897217, "grad_norm": 0.026789769530296326, "global_step": 614, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05803677439689636, "train_loss_bc": 0.05803677439689636, "train_loss_llm": 0.0, "grad_norm": 0.03823623061180115, "global_step": 615, "epoch": 17, "lr": 0.001} +{"train_loss": 0.03337608650326729, "train_loss_bc": 0.03337608650326729, "train_loss_llm": 0.0, "grad_norm": 0.04117236286401749, "global_step": 616, "epoch": 17, "lr": 0.001} +{"train_loss": 0.053037021309137344, "train_loss_bc": 0.053015630692243576, "train_loss_llm": 0.0021390318870544434, "grad_norm": 0.013517972081899643, "global_step": 617, "epoch": 17, "lr": 0.001} +{"train_loss": 0.046254996210336685, "train_loss_bc": 0.046254996210336685, "train_loss_llm": 0.0, "grad_norm": 0.022960079833865166, "global_step": 618, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05057123303413391, "train_loss_bc": 0.05056285485625267, "train_loss_llm": 0.0008376538753509521, "grad_norm": 0.030713966116309166, "global_step": 619, "epoch": 17, "lr": 0.001} +{"train_loss": 0.04826745390892029, "train_loss_bc": 0.04826381057500839, "train_loss_llm": 0.0003643929958343506, "grad_norm": 0.039368994534015656, "global_step": 620, "epoch": 17, "lr": 0.001} +{"train_loss": 0.031418465077877045, "train_loss_bc": 0.031418465077877045, "train_loss_llm": 0.0, "grad_norm": 0.012361963279545307, "global_step": 621, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05032805725932121, "train_loss_bc": 0.05032197758555412, "train_loss_llm": 0.0006080269813537598, "grad_norm": 0.01914357766509056, "global_step": 622, "epoch": 17, "lr": 0.001} +{"train_loss": 0.058580294251441956, "train_loss_bc": 0.058555103838443756, "train_loss_llm": 0.0025191903114318848, "grad_norm": 0.028749585151672363, "global_step": 623, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05099530518054962, "train_loss_bc": 0.050947509706020355, "train_loss_llm": 0.00477936863899231, "grad_norm": 0.03293602541089058, "global_step": 624, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05868256837129593, "train_loss_bc": 0.05866006761789322, "train_loss_llm": 0.002249985933303833, "grad_norm": 0.02637968584895134, "global_step": 625, "epoch": 17, "lr": 0.001} +{"train_loss": 0.053940240293741226, "train_loss_bc": 0.053929418325424194, "train_loss_llm": 0.0010822713375091553, "grad_norm": 0.029845979064702988, "global_step": 626, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05141908675432205, "train_loss_bc": 0.05141908675432205, "train_loss_llm": 0.0, "grad_norm": 0.03350738435983658, "global_step": 627, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05475109815597534, "train_loss_bc": 0.05473634600639343, "train_loss_llm": 0.0014750957489013672, "grad_norm": 0.04101266339421272, "global_step": 628, "epoch": 17, "lr": 0.001} +{"train_loss": 0.06120467558503151, "train_loss_bc": 0.06116458773612976, "train_loss_llm": 0.004008769989013672, "grad_norm": 0.017614519223570824, "global_step": 629, "epoch": 17, "lr": 0.001} +{"train_loss": 0.04897141456604004, "train_loss_bc": 0.04894639179110527, "train_loss_llm": 0.002502143383026123, "grad_norm": 0.02429872937500477, "global_step": 630, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05780930817127228, "train_loss_bc": 0.05780930817127228, "train_loss_llm": 0.0, "grad_norm": 0.028441887348890305, "global_step": 631, "epoch": 17, "lr": 0.001} +{"train_loss": 0.0567997507750988, "train_loss_bc": 0.05676877871155739, "train_loss_llm": 0.0030971169471740723, "grad_norm": 0.02982722967863083, "global_step": 632, "epoch": 17, "lr": 0.001} +{"train_loss": 0.04981836676597595, "train_loss_bc": 0.04978708550333977, "train_loss_llm": 0.0031279921531677246, "grad_norm": 0.01569082774221897, "global_step": 633, "epoch": 17, "lr": 0.001} +{"train_loss": 0.04795822873711586, "train_loss_bc": 0.04795822873711586, "train_loss_llm": 0.0, "grad_norm": 0.026925260201096535, "global_step": 634, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05534755811095238, "train_loss_bc": 0.055330097675323486, "train_loss_llm": 0.0017460882663726807, "grad_norm": 0.03026188723742962, "global_step": 635, "epoch": 17, "lr": 0.001} +{"train_loss": 0.04625225067138672, "train_loss_bc": 0.04625225067138672, "train_loss_llm": 0.0, "grad_norm": 0.031072495505213737, "global_step": 636, "epoch": 17, "lr": 0.001} +{"train_loss": 0.058801013976335526, "train_loss_bc": 0.058784957975149155, "train_loss_llm": 0.0016056299209594727, "grad_norm": 0.017482023686170578, "global_step": 637, "epoch": 17, "lr": 0.001} +{"train_loss": 0.053577035665512085, "train_loss_bc": 0.05346505343914032, "train_loss_llm": 0.011198371648788452, "grad_norm": 0.03200952708721161, "global_step": 638, "epoch": 17, "lr": 0.001} +{"train_loss": 0.04757948964834213, "train_loss_bc": 0.04757948964834213, "train_loss_llm": 0.0, "grad_norm": 0.0403364822268486, "global_step": 639, "epoch": 17, "lr": 0.001} +{"train_loss": 0.044849008321762085, "train_loss_bc": 0.044849008321762085, "train_loss_llm": 0.0, "grad_norm": 0.04680727422237396, "global_step": 640, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05731630697846413, "train_loss_bc": 0.05731380730867386, "train_loss_llm": 0.0002499520778656006, "grad_norm": 0.01819225214421749, "global_step": 641, "epoch": 17, "lr": 0.001} +{"train_loss": 0.057383641600608826, "train_loss_bc": 0.057358480989933014, "train_loss_llm": 0.0025159120559692383, "grad_norm": 0.026906946673989296, "global_step": 642, "epoch": 17, "lr": 0.001} +{"train_loss": 0.04971516504883766, "train_loss_bc": 0.049687713384628296, "train_loss_llm": 0.0027452707290649414, "grad_norm": 0.03359764441847801, "global_step": 643, "epoch": 17, "lr": 0.001} +{"train_loss": 0.06062763184309006, "train_loss_bc": 0.060627520084381104, "train_loss_llm": 1.1265277862548828e-05, "grad_norm": 0.04067676514387131, "global_step": 644, "epoch": 17, "lr": 0.001} +{"train_loss": 0.04750637710094452, "train_loss_bc": 0.047479864209890366, "train_loss_llm": 0.0026511549949645996, "grad_norm": 0.018286719918251038, "global_step": 645, "epoch": 17, "lr": 0.001} +{"train_loss": 0.05109256133437157, "train_loss_bc": 0.05108914151787758, "train_loss_llm": 0.0003421008586883545, "grad_norm": 0.028180740773677826, "global_step": 646, "epoch": 17, "lr": 0.001} +{"train_loss": 0.0510195085985793, "train_loss_bc": 0.045252081006765366, "train_loss_llm": 0.0, "grad_norm": 0.030451584607362747, "global_step": 647, "epoch": 17, "lr": 0.001, "val_loss": 0.05144234001636505} +{"train_loss": 0.05220814794301987, "train_loss_bc": 0.052202969789505005, "train_loss_llm": 0.0005177557468414307, "grad_norm": 0.04127621278166771, "global_step": 648, "epoch": 18, "lr": 0.001} +{"train_loss": 0.04834222421050072, "train_loss_bc": 0.04829087480902672, "train_loss_llm": 0.0051351189613342285, "grad_norm": 0.014074042439460754, "global_step": 649, "epoch": 18, "lr": 0.001} +{"train_loss": 0.054230768233537674, "train_loss_bc": 0.054230768233537674, "train_loss_llm": 0.0, "grad_norm": 0.020522672683000565, "global_step": 650, "epoch": 18, "lr": 0.001} +{"train_loss": 0.053541503846645355, "train_loss_bc": 0.053541503846645355, "train_loss_llm": 0.0, "grad_norm": 0.02334488183259964, "global_step": 651, "epoch": 18, "lr": 0.001} +{"train_loss": 0.05077464133501053, "train_loss_bc": 0.05074348300695419, "train_loss_llm": 0.003116011619567871, "grad_norm": 0.031101010739803314, "global_step": 652, "epoch": 18, "lr": 0.001} +{"train_loss": 0.035239968448877335, "train_loss_bc": 0.03522695228457451, "train_loss_llm": 0.001301497220993042, "grad_norm": 0.01293313130736351, "global_step": 653, "epoch": 18, "lr": 0.001} +{"train_loss": 0.05153545364737511, "train_loss_bc": 0.0515223927795887, "train_loss_llm": 0.001306086778640747, "grad_norm": 0.022296274080872536, "global_step": 654, "epoch": 18, "lr": 0.001} +{"train_loss": 0.060597822070121765, "train_loss_bc": 0.060597822070121765, "train_loss_llm": 0.0, "grad_norm": 0.027369361370801926, "global_step": 655, "epoch": 18, "lr": 0.001} +{"train_loss": 0.05069009214639664, "train_loss_bc": 0.05068308115005493, "train_loss_llm": 0.0007011294364929199, "grad_norm": 0.03444215655326843, "global_step": 656, "epoch": 18, "lr": 0.001} +{"train_loss": 0.04377663880586624, "train_loss_bc": 0.04377663880586624, "train_loss_llm": 0.0, "grad_norm": 0.01612498238682747, "global_step": 657, "epoch": 18, "lr": 0.001} +{"train_loss": 0.0494549497961998, "train_loss_bc": 0.0494549497961998, "train_loss_llm": 0.0, "grad_norm": 0.02056102454662323, "global_step": 658, "epoch": 18, "lr": 0.001} +{"train_loss": 0.04897649958729744, "train_loss_bc": 0.04896834120154381, "train_loss_llm": 0.0008160173892974854, "grad_norm": 0.0273369662463665, "global_step": 659, "epoch": 18, "lr": 0.001} +{"train_loss": 0.049965713173151016, "train_loss_bc": 0.04993998631834984, "train_loss_llm": 0.002572566270828247, "grad_norm": 0.033008940517902374, "global_step": 660, "epoch": 18, "lr": 0.001} +{"train_loss": 0.04865358769893646, "train_loss_bc": 0.048614632338285446, "train_loss_llm": 0.003895401954650879, "grad_norm": 0.017710018903017044, "global_step": 661, "epoch": 18, "lr": 0.001} +{"train_loss": 0.05686411261558533, "train_loss_bc": 0.05686411261558533, "train_loss_llm": 0.0, "grad_norm": 0.02523397095501423, "global_step": 662, "epoch": 18, "lr": 0.001} +{"train_loss": 0.0572182759642601, "train_loss_bc": 0.057157061994075775, "train_loss_llm": 0.006121516227722168, "grad_norm": 0.041249483823776245, "global_step": 663, "epoch": 18, "lr": 0.001} +{"train_loss": 0.05156055465340614, "train_loss_bc": 0.051549360156059265, "train_loss_llm": 0.0011193156242370605, "grad_norm": 0.05342080816626549, "global_step": 664, "epoch": 18, "lr": 0.001} +{"train_loss": 0.053140368312597275, "train_loss_bc": 0.053133971989154816, "train_loss_llm": 0.0006397664546966553, "grad_norm": 0.022834496572613716, "global_step": 665, "epoch": 18, "lr": 0.001} +{"train_loss": 0.05967153608798981, "train_loss_bc": 0.05967153608798981, "train_loss_llm": 0.0, "grad_norm": 0.040665481239557266, "global_step": 666, "epoch": 18, "lr": 0.001} +{"train_loss": 0.054703474044799805, "train_loss_bc": 0.054695457220077515, "train_loss_llm": 0.0008017122745513916, "grad_norm": 0.04580199345946312, "global_step": 667, "epoch": 18, "lr": 0.001} +{"train_loss": 0.04912078380584717, "train_loss_bc": 0.04912078380584717, "train_loss_llm": 0.0, "grad_norm": 0.04990319162607193, "global_step": 668, "epoch": 18, "lr": 0.001} +{"train_loss": 0.03723442927002907, "train_loss_bc": 0.03722074255347252, "train_loss_llm": 0.0013686418533325195, "grad_norm": 0.01840214990079403, "global_step": 669, "epoch": 18, "lr": 0.001} +{"train_loss": 0.04565144330263138, "train_loss_bc": 0.045638229697942734, "train_loss_llm": 0.0013214051723480225, "grad_norm": 0.02900826185941696, "global_step": 670, "epoch": 18, "lr": 0.001} +{"train_loss": 0.05528491735458374, "train_loss_bc": 0.05528491735458374, "train_loss_llm": 0.0, "grad_norm": 0.035988252609968185, "global_step": 671, "epoch": 18, "lr": 0.001} +{"train_loss": 0.0558461993932724, "train_loss_bc": 0.055798862129449844, "train_loss_llm": 0.004733562469482422, "grad_norm": 0.049106888473033905, "global_step": 672, "epoch": 18, "lr": 0.001} +{"train_loss": 0.04174216836690903, "train_loss_bc": 0.04174216836690903, "train_loss_llm": 0.0, "grad_norm": 0.018229994922876358, "global_step": 673, "epoch": 18, "lr": 0.001} +{"train_loss": 0.055799636989831924, "train_loss_bc": 0.055799636989831924, "train_loss_llm": 0.0, "grad_norm": 0.023214517161250114, "global_step": 674, "epoch": 18, "lr": 0.001} +{"train_loss": 0.049108900129795074, "train_loss_bc": 0.049074653536081314, "train_loss_llm": 0.003424525260925293, "grad_norm": 0.031210921704769135, "global_step": 675, "epoch": 18, "lr": 0.001} +{"train_loss": 0.04972962290048599, "train_loss_bc": 0.04972962290048599, "train_loss_llm": 0.0, "grad_norm": 0.04130186513066292, "global_step": 676, "epoch": 18, "lr": 0.001} +{"train_loss": 0.0588698610663414, "train_loss_bc": 0.0588698610663414, "train_loss_llm": 0.0, "grad_norm": 0.012110721319913864, "global_step": 677, "epoch": 18, "lr": 0.001} +{"train_loss": 0.048662081360816956, "train_loss_bc": 0.04861828684806824, "train_loss_llm": 0.004379361867904663, "grad_norm": 0.01944759674370289, "global_step": 678, "epoch": 18, "lr": 0.001} +{"train_loss": 0.049000758677721024, "train_loss_bc": 0.049000758677721024, "train_loss_llm": 0.0, "grad_norm": 0.02338794246315956, "global_step": 679, "epoch": 18, "lr": 0.001} +{"train_loss": 0.055034831166267395, "train_loss_bc": 0.055034831166267395, "train_loss_llm": 0.0, "grad_norm": 0.029879098758101463, "global_step": 680, "epoch": 18, "lr": 0.001} +{"train_loss": 0.0453411303460598, "train_loss_bc": 0.0453411303460598, "train_loss_llm": 0.0, "grad_norm": 0.014201915822923183, "global_step": 681, "epoch": 18, "lr": 0.001} +{"train_loss": 0.04760649800300598, "train_loss_bc": 0.047581542283296585, "train_loss_llm": 0.00249558687210083, "grad_norm": 0.01735687255859375, "global_step": 682, "epoch": 18, "lr": 0.001} +{"train_loss": 0.05073710075683064, "train_loss_bc": 0.05131302773952484, "train_loss_llm": 0.004300415515899658, "grad_norm": 0.03311735391616821, "global_step": 683, "epoch": 18, "lr": 0.001, "val_loss": 0.05105552077293396} +{"train_loss": 0.04743706062436104, "train_loss_bc": 0.04743706062436104, "train_loss_llm": 0.0, "grad_norm": 0.0423407256603241, "global_step": 684, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05276281759142876, "train_loss_bc": 0.05274525657296181, "train_loss_llm": 0.0017561614513397217, "grad_norm": 0.018575124442577362, "global_step": 685, "epoch": 19, "lr": 0.001} +{"train_loss": 0.039615243673324585, "train_loss_bc": 0.03955269977450371, "train_loss_llm": 0.006254345178604126, "grad_norm": 0.022452544420957565, "global_step": 686, "epoch": 19, "lr": 0.001} +{"train_loss": 0.04864344373345375, "train_loss_bc": 0.048632893711328506, "train_loss_llm": 0.0010550320148468018, "grad_norm": 0.026649028062820435, "global_step": 687, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05607198178768158, "train_loss_bc": 0.05603886395692825, "train_loss_llm": 0.0033119022846221924, "grad_norm": 0.03231159597635269, "global_step": 688, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05250658467411995, "train_loss_bc": 0.05250658467411995, "train_loss_llm": 0.0, "grad_norm": 0.017713693901896477, "global_step": 689, "epoch": 19, "lr": 0.001} +{"train_loss": 0.043191250413656235, "train_loss_bc": 0.04318094253540039, "train_loss_llm": 0.0010308027267456055, "grad_norm": 0.024376478046178818, "global_step": 690, "epoch": 19, "lr": 0.001} +{"train_loss": 0.051261693239212036, "train_loss_bc": 0.051261693239212036, "train_loss_llm": 0.0, "grad_norm": 0.028475342318415642, "global_step": 691, "epoch": 19, "lr": 0.001} +{"train_loss": 0.049443863332271576, "train_loss_bc": 0.049430880695581436, "train_loss_llm": 0.0012984275817871094, "grad_norm": 0.03442278876900673, "global_step": 692, "epoch": 19, "lr": 0.001} +{"train_loss": 0.04934501647949219, "train_loss_bc": 0.04931028187274933, "train_loss_llm": 0.0034734904766082764, "grad_norm": 0.01698760874569416, "global_step": 693, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05154554173350334, "train_loss_bc": 0.05151667818427086, "train_loss_llm": 0.0028863847255706787, "grad_norm": 0.027630982920527458, "global_step": 694, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05578989163041115, "train_loss_bc": 0.05576951801776886, "train_loss_llm": 0.0020373165607452393, "grad_norm": 0.030540335923433304, "global_step": 695, "epoch": 19, "lr": 0.001} +{"train_loss": 0.053375497460365295, "train_loss_bc": 0.053375497460365295, "train_loss_llm": 0.0, "grad_norm": 0.042120933532714844, "global_step": 696, "epoch": 19, "lr": 0.001} +{"train_loss": 0.04457687586545944, "train_loss_bc": 0.04457642883062363, "train_loss_llm": 4.45246696472168e-05, "grad_norm": 0.01992553472518921, "global_step": 697, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05108528211712837, "train_loss_bc": 0.05106046423316002, "train_loss_llm": 0.002481818199157715, "grad_norm": 0.040038105100393295, "global_step": 698, "epoch": 19, "lr": 0.001} +{"train_loss": 0.040327537804841995, "train_loss_bc": 0.040327537804841995, "train_loss_llm": 0.0, "grad_norm": 0.04844240844249725, "global_step": 699, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05513250455260277, "train_loss_bc": 0.05509559065103531, "train_loss_llm": 0.003691375255584717, "grad_norm": 0.05025886371731758, "global_step": 700, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05428102985024452, "train_loss_bc": 0.05427304282784462, "train_loss_llm": 0.000798642635345459, "grad_norm": 0.01527281291782856, "global_step": 701, "epoch": 19, "lr": 0.001} +{"train_loss": 0.046345796436071396, "train_loss_bc": 0.04634209722280502, "train_loss_llm": 0.00036975741386413574, "grad_norm": 0.024050282314419746, "global_step": 702, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05565432831645012, "train_loss_bc": 0.05560808628797531, "train_loss_llm": 0.004624038934707642, "grad_norm": 0.03635271638631821, "global_step": 703, "epoch": 19, "lr": 0.001} +{"train_loss": 0.050447240471839905, "train_loss_bc": 0.05040210485458374, "train_loss_llm": 0.004513680934906006, "grad_norm": 0.05158925801515579, "global_step": 704, "epoch": 19, "lr": 0.001} +{"train_loss": 0.055678840726614, "train_loss_bc": 0.05567573383450508, "train_loss_llm": 0.00031051039695739746, "grad_norm": 0.015747472643852234, "global_step": 705, "epoch": 19, "lr": 0.001} +{"train_loss": 0.0465974323451519, "train_loss_bc": 0.0465974323451519, "train_loss_llm": 0.0, "grad_norm": 0.027714384719729424, "global_step": 706, "epoch": 19, "lr": 0.001} +{"train_loss": 0.04671137034893036, "train_loss_bc": 0.04670509696006775, "train_loss_llm": 0.0006271600723266602, "grad_norm": 0.035532787442207336, "global_step": 707, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05005701258778572, "train_loss_bc": 0.05005701258778572, "train_loss_llm": 0.0, "grad_norm": 0.039908431470394135, "global_step": 708, "epoch": 19, "lr": 0.001} +{"train_loss": 0.050601549446582794, "train_loss_bc": 0.05057302489876747, "train_loss_llm": 0.002852529287338257, "grad_norm": 0.018785636872053146, "global_step": 709, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05063823610544205, "train_loss_bc": 0.05060819908976555, "train_loss_llm": 0.0030036866664886475, "grad_norm": 0.025466270744800568, "global_step": 710, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05988112464547157, "train_loss_bc": 0.05985109508037567, "train_loss_llm": 0.003003031015396118, "grad_norm": 0.03515801951289177, "global_step": 711, "epoch": 19, "lr": 0.001} +{"train_loss": 0.05402381718158722, "train_loss_bc": 0.05402381718158722, "train_loss_llm": 0.0, "grad_norm": 0.04124542698264122, "global_step": 712, "epoch": 19, "lr": 0.001} +{"train_loss": 0.04784298688173294, "train_loss_bc": 0.04784298688173294, "train_loss_llm": 0.0, "grad_norm": 0.022494876757264137, "global_step": 713, "epoch": 19, "lr": 0.001} +{"train_loss": 0.0646209791302681, "train_loss_bc": 0.0646209791302681, "train_loss_llm": 0.0, "grad_norm": 0.039026252925395966, "global_step": 714, "epoch": 19, "lr": 0.001} +{"train_loss": 0.04871270805597305, "train_loss_bc": 0.04871270805597305, "train_loss_llm": 0.0, "grad_norm": 0.046258557587862015, "global_step": 715, "epoch": 19, "lr": 0.001} +{"train_loss": 0.047787223011255264, "train_loss_bc": 0.047783322632312775, "train_loss_llm": 0.0003899037837982178, "grad_norm": 0.053581055253744125, "global_step": 716, "epoch": 19, "lr": 0.001} +{"train_loss": 0.04508700966835022, "train_loss_bc": 0.045071229338645935, "train_loss_llm": 0.001578211784362793, "grad_norm": 0.018471498042345047, "global_step": 717, "epoch": 19, "lr": 0.001} +{"train_loss": 0.04889068007469177, "train_loss_bc": 0.04889068007469177, "train_loss_llm": 0.0, "grad_norm": 0.024011222645640373, "global_step": 718, "epoch": 19, "lr": 0.001} +{"train_loss": 0.050502286189132266, "train_loss_bc": 0.052059926092624664, "train_loss_llm": 0.0050925612449646, "grad_norm": 0.04296550527215004, "global_step": 719, "epoch": 19, "lr": 0.001, "val_loss": 0.05078086256980896} +{"train_loss": 0.04727741703391075, "train_loss_bc": 0.04727741703391075, "train_loss_llm": 0.0, "grad_norm": 0.0521802082657814, "global_step": 720, "epoch": 20, "lr": 0.001} +{"train_loss": 0.053961802273988724, "train_loss_bc": 0.0539347343146801, "train_loss_llm": 0.0027069151401519775, "grad_norm": 0.018421217799186707, "global_step": 721, "epoch": 20, "lr": 0.001} +{"train_loss": 0.053449276834726334, "train_loss_bc": 0.053449276834726334, "train_loss_llm": 0.0, "grad_norm": 0.030372019857168198, "global_step": 722, "epoch": 20, "lr": 0.001} +{"train_loss": 0.03835985064506531, "train_loss_bc": 0.03835985064506531, "train_loss_llm": 0.0, "grad_norm": 0.04151902347803116, "global_step": 723, "epoch": 20, "lr": 0.001} +{"train_loss": 0.05168316513299942, "train_loss_bc": 0.05165508762001991, "train_loss_llm": 0.002807915210723877, "grad_norm": 0.051895853132009506, "global_step": 724, "epoch": 20, "lr": 0.001} +{"train_loss": 0.059009965509176254, "train_loss_bc": 0.05899859964847565, "train_loss_llm": 0.0011364519596099854, "grad_norm": 0.018481118604540825, "global_step": 725, "epoch": 20, "lr": 0.001} +{"train_loss": 0.052566900849342346, "train_loss_bc": 0.05255122482776642, "train_loss_llm": 0.0015676021575927734, "grad_norm": 0.03081272728741169, "global_step": 726, "epoch": 20, "lr": 0.001} +{"train_loss": 0.058541446924209595, "train_loss_bc": 0.058541446924209595, "train_loss_llm": 0.0, "grad_norm": 0.04231453314423561, "global_step": 727, "epoch": 20, "lr": 0.001} +{"train_loss": 0.044715866446495056, "train_loss_bc": 0.044715866446495056, "train_loss_llm": 0.0, "grad_norm": 0.05018923431634903, "global_step": 728, "epoch": 20, "lr": 0.001} +{"train_loss": 0.03507907688617706, "train_loss_bc": 0.03507907688617706, "train_loss_llm": 0.0, "grad_norm": 0.015365309081971645, "global_step": 729, "epoch": 20, "lr": 0.001} +{"train_loss": 0.04693108797073364, "train_loss_bc": 0.046915389597415924, "train_loss_llm": 0.0015697777271270752, "grad_norm": 0.026871580630540848, "global_step": 730, "epoch": 20, "lr": 0.001} +{"train_loss": 0.05960036814212799, "train_loss_bc": 0.05960036814212799, "train_loss_llm": 0.0, "grad_norm": 0.04444550350308418, "global_step": 731, "epoch": 20, "lr": 0.001} +{"train_loss": 0.058223553001880646, "train_loss_bc": 0.058223553001880646, "train_loss_llm": 0.0, "grad_norm": 0.05719362944364548, "global_step": 732, "epoch": 20, "lr": 0.001} +{"train_loss": 0.06058802455663681, "train_loss_bc": 0.06058802455663681, "train_loss_llm": 0.0, "grad_norm": 0.018589021638035774, "global_step": 733, "epoch": 20, "lr": 0.001} +{"train_loss": 0.04004361853003502, "train_loss_bc": 0.04004361853003502, "train_loss_llm": 0.0, "grad_norm": 0.0270317941904068, "global_step": 734, "epoch": 20, "lr": 0.001} +{"train_loss": 0.04432979226112366, "train_loss_bc": 0.04432979226112366, "train_loss_llm": 0.0, "grad_norm": 0.036482322961091995, "global_step": 735, "epoch": 20, "lr": 0.001} +{"train_loss": 0.04751923680305481, "train_loss_bc": 0.04751923680305481, "train_loss_llm": 0.0, "grad_norm": 0.043674226850271225, "global_step": 736, "epoch": 20, "lr": 0.001} +{"train_loss": 0.05862828344106674, "train_loss_bc": 0.05859376862645149, "train_loss_llm": 0.003451317548751831, "grad_norm": 0.021330809220671654, "global_step": 737, "epoch": 20, "lr": 0.001} +{"train_loss": 0.05451679602265358, "train_loss_bc": 0.05446327105164528, "train_loss_llm": 0.005352646112442017, "grad_norm": 0.026447277516126633, "global_step": 738, "epoch": 20, "lr": 0.001} +{"train_loss": 0.043756067752838135, "train_loss_bc": 0.043756067752838135, "train_loss_llm": 0.0, "grad_norm": 0.03130350634455681, "global_step": 739, "epoch": 20, "lr": 0.001} +{"train_loss": 0.043546199798583984, "train_loss_bc": 0.04351331666111946, "train_loss_llm": 0.003288477659225464, "grad_norm": 0.035897497087717056, "global_step": 740, "epoch": 20, "lr": 0.001} +{"train_loss": 0.05233757197856903, "train_loss_bc": 0.05233757197856903, "train_loss_llm": 0.0, "grad_norm": 0.02033655345439911, "global_step": 741, "epoch": 20, "lr": 0.001} +{"train_loss": 0.050828371196985245, "train_loss_bc": 0.050828371196985245, "train_loss_llm": 0.0, "grad_norm": 0.02962089143693447, "global_step": 742, "epoch": 20, "lr": 0.001} +{"train_loss": 0.051931653171777725, "train_loss_bc": 0.051931653171777725, "train_loss_llm": 0.0, "grad_norm": 0.03652297705411911, "global_step": 743, "epoch": 20, "lr": 0.001} +{"train_loss": 0.05551057308912277, "train_loss_bc": 0.05549168586730957, "train_loss_llm": 0.001888871192932129, "grad_norm": 0.046598922461271286, "global_step": 744, "epoch": 20, "lr": 0.001} +{"train_loss": 0.054777421057224274, "train_loss_bc": 0.05477288365364075, "train_loss_llm": 0.00045362114906311035, "grad_norm": 0.01938752271234989, "global_step": 745, "epoch": 20, "lr": 0.001} +{"train_loss": 0.05517931655049324, "train_loss_bc": 0.055156346410512924, "train_loss_llm": 0.002296954393386841, "grad_norm": 0.02652035653591156, "global_step": 746, "epoch": 20, "lr": 0.001} +{"train_loss": 0.045617446303367615, "train_loss_bc": 0.045607566833496094, "train_loss_llm": 0.0009879469871520996, "grad_norm": 0.035323016345500946, "global_step": 747, "epoch": 20, "lr": 0.001} +{"train_loss": 0.04758624732494354, "train_loss_bc": 0.04756871983408928, "train_loss_llm": 0.0017528831958770752, "grad_norm": 0.041418224573135376, "global_step": 748, "epoch": 20, "lr": 0.001} +{"train_loss": 0.04340295493602753, "train_loss_bc": 0.043382614850997925, "train_loss_llm": 0.002033919095993042, "grad_norm": 0.020090162754058838, "global_step": 749, "epoch": 20, "lr": 0.001} +{"train_loss": 0.058057669550180435, "train_loss_bc": 0.05805625766515732, "train_loss_llm": 0.000141143798828125, "grad_norm": 0.03193703293800354, "global_step": 750, "epoch": 20, "lr": 0.001} +{"train_loss": 0.054995737969875336, "train_loss_bc": 0.054995737969875336, "train_loss_llm": 0.0, "grad_norm": 0.04145744815468788, "global_step": 751, "epoch": 20, "lr": 0.001} +{"train_loss": 0.045330945402383804, "train_loss_bc": 0.04531746730208397, "train_loss_llm": 0.0013478994369506836, "grad_norm": 0.04610714688897133, "global_step": 752, "epoch": 20, "lr": 0.001} +{"train_loss": 0.046208035200834274, "train_loss_bc": 0.04620012640953064, "train_loss_llm": 0.0007908642292022705, "grad_norm": 0.01754305697977543, "global_step": 753, "epoch": 20, "lr": 0.001} +{"train_loss": 0.04735882580280304, "train_loss_bc": 0.04733627662062645, "train_loss_llm": 0.0022547543048858643, "grad_norm": 0.04152127355337143, "global_step": 754, "epoch": 20, "lr": 0.001} +{"train_loss": 0.05028127320110798, "train_loss_bc": 0.04867526888847351, "train_loss_llm": 0.0, "grad_norm": 0.058106742799282074, "global_step": 755, "epoch": 20, "lr": 0.001, "train/cumulative_reward": 6.444797974990476, "train/mean_score": 0.15670879473863858, "train/success_rate": 0.0, "test/cumulative_reward": 2.1646561924610936, "test/mean_score": 0.08879064759974425, "test/success_rate": 0.0, "val_loss": 0.05128291994333267, "train_action_mse_error": 0.05328262597322464} +{"train_loss": 0.053375571966171265, "train_loss_bc": 0.053374841809272766, "train_loss_llm": 7.283687591552734e-05, "grad_norm": 0.07438715547323227, "global_step": 756, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05230309069156647, "train_loss_bc": 0.05230309069156647, "train_loss_llm": 0.0, "grad_norm": 0.012664960697293282, "global_step": 757, "epoch": 21, "lr": 0.001} +{"train_loss": 0.0532706156373024, "train_loss_bc": 0.05325702950358391, "train_loss_llm": 0.001358717679977417, "grad_norm": 0.022063998505473137, "global_step": 758, "epoch": 21, "lr": 0.001} +{"train_loss": 0.041102029383182526, "train_loss_bc": 0.04106865078210831, "train_loss_llm": 0.003337681293487549, "grad_norm": 0.029042916372418404, "global_step": 759, "epoch": 21, "lr": 0.001} +{"train_loss": 0.055152952671051025, "train_loss_bc": 0.05508790537714958, "train_loss_llm": 0.006504714488983154, "grad_norm": 0.036772191524505615, "global_step": 760, "epoch": 21, "lr": 0.001} +{"train_loss": 0.053856056183576584, "train_loss_bc": 0.053856056183576584, "train_loss_llm": 0.0, "grad_norm": 0.014340701512992382, "global_step": 761, "epoch": 21, "lr": 0.001} +{"train_loss": 0.06300155818462372, "train_loss_bc": 0.06300155818462372, "train_loss_llm": 0.0, "grad_norm": 0.023303600028157234, "global_step": 762, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05339192599058151, "train_loss_bc": 0.05339192599058151, "train_loss_llm": 0.0, "grad_norm": 0.03142198547720909, "global_step": 763, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05748283490538597, "train_loss_bc": 0.05748283490538597, "train_loss_llm": 0.0, "grad_norm": 0.037277910858392715, "global_step": 764, "epoch": 21, "lr": 0.001} +{"train_loss": 0.03459784761071205, "train_loss_bc": 0.03459374979138374, "train_loss_llm": 0.0004099607467651367, "grad_norm": 0.014509479515254498, "global_step": 765, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05299465358257294, "train_loss_bc": 0.05299465358257294, "train_loss_llm": 0.0, "grad_norm": 0.025753308087587357, "global_step": 766, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04889478161931038, "train_loss_bc": 0.04888642951846123, "train_loss_llm": 0.0008352398872375488, "grad_norm": 0.031376905739307404, "global_step": 767, "epoch": 21, "lr": 0.001} +{"train_loss": 0.054464858025312424, "train_loss_bc": 0.054433755576610565, "train_loss_llm": 0.003110349178314209, "grad_norm": 0.04240444302558899, "global_step": 768, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04679055139422417, "train_loss_bc": 0.04679055139422417, "train_loss_llm": 0.0, "grad_norm": 0.01996578648686409, "global_step": 769, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05174591764807701, "train_loss_bc": 0.0517251193523407, "train_loss_llm": 0.0020799338817596436, "grad_norm": 0.032904062420129776, "global_step": 770, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05187851935625076, "train_loss_bc": 0.05185827240347862, "train_loss_llm": 0.002024829387664795, "grad_norm": 0.04191536828875542, "global_step": 771, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04414725303649902, "train_loss_bc": 0.04412661865353584, "train_loss_llm": 0.0020636022090911865, "grad_norm": 0.04647541046142578, "global_step": 772, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05034801363945007, "train_loss_bc": 0.05034801363945007, "train_loss_llm": 0.0, "grad_norm": 0.017748922109603882, "global_step": 773, "epoch": 21, "lr": 0.001} +{"train_loss": 0.0461147204041481, "train_loss_bc": 0.0461147204041481, "train_loss_llm": 0.0, "grad_norm": 0.02580171450972557, "global_step": 774, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04724331200122833, "train_loss_bc": 0.04724331200122833, "train_loss_llm": 0.0, "grad_norm": 0.03241383284330368, "global_step": 775, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05981918424367905, "train_loss_bc": 0.05981918424367905, "train_loss_llm": 0.0, "grad_norm": 0.03816897049546242, "global_step": 776, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04966136813163757, "train_loss_bc": 0.04962208494544029, "train_loss_llm": 0.003928273916244507, "grad_norm": 0.020253514871001244, "global_step": 777, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04107540473341942, "train_loss_bc": 0.04106106981635094, "train_loss_llm": 0.0014335215091705322, "grad_norm": 0.02817891724407673, "global_step": 778, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04821338504552841, "train_loss_bc": 0.04814707860350609, "train_loss_llm": 0.0066304802894592285, "grad_norm": 0.03660280257463455, "global_step": 779, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04337964951992035, "train_loss_bc": 0.04337964951992035, "train_loss_llm": 0.0, "grad_norm": 0.04269000515341759, "global_step": 780, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04717150703072548, "train_loss_bc": 0.047142304480075836, "train_loss_llm": 0.0029203295707702637, "grad_norm": 0.019702810794115067, "global_step": 781, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04413171112537384, "train_loss_bc": 0.04409264773130417, "train_loss_llm": 0.0039062201976776123, "grad_norm": 0.025353699922561646, "global_step": 782, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05125952512025833, "train_loss_bc": 0.05125952512025833, "train_loss_llm": 0.0, "grad_norm": 0.02779560536146164, "global_step": 783, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04447702318429947, "train_loss_bc": 0.04447702318429947, "train_loss_llm": 0.0, "grad_norm": 0.03358397260308266, "global_step": 784, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05113409087061882, "train_loss_bc": 0.05113409087061882, "train_loss_llm": 0.0, "grad_norm": 0.020885752514004707, "global_step": 785, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04938241094350815, "train_loss_bc": 0.04936489462852478, "train_loss_llm": 0.001751720905303955, "grad_norm": 0.03104458563029766, "global_step": 786, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05167527496814728, "train_loss_bc": 0.05167527496814728, "train_loss_llm": 0.0, "grad_norm": 0.04034145548939705, "global_step": 787, "epoch": 21, "lr": 0.001} +{"train_loss": 0.05165277048945427, "train_loss_bc": 0.05164172127842903, "train_loss_llm": 0.0011050701141357422, "grad_norm": 0.04750646650791168, "global_step": 788, "epoch": 21, "lr": 0.001} +{"train_loss": 0.050731111317873, "train_loss_bc": 0.05063233524560928, "train_loss_llm": 0.009877502918243408, "grad_norm": 0.015589782036840916, "global_step": 789, "epoch": 21, "lr": 0.001} +{"train_loss": 0.043916553258895874, "train_loss_bc": 0.043916553258895874, "train_loss_llm": 0.0, "grad_norm": 0.02054024301469326, "global_step": 790, "epoch": 21, "lr": 0.001} +{"train_loss": 0.04994086941911115, "train_loss_bc": 0.05800977349281311, "train_loss_llm": 0.0023493170738220215, "grad_norm": 0.033059634268283844, "global_step": 791, "epoch": 21, "lr": 0.001, "val_loss": 0.05019237846136093} +{"train_loss": 0.04979483410716057, "train_loss_bc": 0.04979483410716057, "train_loss_llm": 0.0, "grad_norm": 0.04011155292391777, "global_step": 792, "epoch": 22, "lr": 0.001} +{"train_loss": 0.06195499375462532, "train_loss_bc": 0.06195499375462532, "train_loss_llm": 0.0, "grad_norm": 0.018142549321055412, "global_step": 793, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04426543042063713, "train_loss_bc": 0.04426543042063713, "train_loss_llm": 0.0, "grad_norm": 0.022067798301577568, "global_step": 794, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04315567761659622, "train_loss_bc": 0.043144453316926956, "train_loss_llm": 0.0011223554611206055, "grad_norm": 0.028066666796803474, "global_step": 795, "epoch": 22, "lr": 0.001} +{"train_loss": 0.045256357640028, "train_loss_bc": 0.04525411128997803, "train_loss_llm": 0.00022467970848083496, "grad_norm": 0.038602400571107864, "global_step": 796, "epoch": 22, "lr": 0.001} +{"train_loss": 0.046774912625551224, "train_loss_bc": 0.04676513001322746, "train_loss_llm": 0.000978231430053711, "grad_norm": 0.014859781600534916, "global_step": 797, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05116339027881622, "train_loss_bc": 0.05115661397576332, "train_loss_llm": 0.000677645206451416, "grad_norm": 0.02506769821047783, "global_step": 798, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04610384628176689, "train_loss_bc": 0.04610384628176689, "train_loss_llm": 0.0, "grad_norm": 0.036943066865205765, "global_step": 799, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05259731784462929, "train_loss_bc": 0.0525357685983181, "train_loss_llm": 0.006154805421829224, "grad_norm": 0.04468323290348053, "global_step": 800, "epoch": 22, "lr": 0.001} +{"train_loss": 0.052305348217487335, "train_loss_bc": 0.052305348217487335, "train_loss_llm": 0.0, "grad_norm": 0.01814235933125019, "global_step": 801, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05088639631867409, "train_loss_bc": 0.05087336152791977, "train_loss_llm": 0.0013033747673034668, "grad_norm": 0.024761274456977844, "global_step": 802, "epoch": 22, "lr": 0.001} +{"train_loss": 0.053861696273088455, "train_loss_bc": 0.053861089050769806, "train_loss_llm": 6.070733070373535e-05, "grad_norm": 0.03382722660899162, "global_step": 803, "epoch": 22, "lr": 0.001} +{"train_loss": 0.047744013369083405, "train_loss_bc": 0.047709930688142776, "train_loss_llm": 0.0034082233905792236, "grad_norm": 0.03851410746574402, "global_step": 804, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04691750556230545, "train_loss_bc": 0.04690517857670784, "train_loss_llm": 0.0012326836585998535, "grad_norm": 0.016136040911078453, "global_step": 805, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04921230673789978, "train_loss_bc": 0.04921230673789978, "train_loss_llm": 0.0, "grad_norm": 0.028065500780940056, "global_step": 806, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04892222583293915, "train_loss_bc": 0.04892222583293915, "train_loss_llm": 0.0, "grad_norm": 0.033971887081861496, "global_step": 807, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05220022797584534, "train_loss_bc": 0.052195947617292404, "train_loss_llm": 0.00042808055877685547, "grad_norm": 0.040407028049230576, "global_step": 808, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05344139784574509, "train_loss_bc": 0.05344139784574509, "train_loss_llm": 0.0, "grad_norm": 0.01732599548995495, "global_step": 809, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04566599801182747, "train_loss_bc": 0.045638397336006165, "train_loss_llm": 0.0027602016925811768, "grad_norm": 0.024914653971791267, "global_step": 810, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04628872498869896, "train_loss_bc": 0.04628872498869896, "train_loss_llm": 0.0, "grad_norm": 0.038876187056303024, "global_step": 811, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05127593129873276, "train_loss_bc": 0.05127593129873276, "train_loss_llm": 0.0, "grad_norm": 0.04695899039506912, "global_step": 812, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04821294546127319, "train_loss_bc": 0.04821294546127319, "train_loss_llm": 0.0, "grad_norm": 0.015584669075906277, "global_step": 813, "epoch": 22, "lr": 0.001} +{"train_loss": 0.03588409349322319, "train_loss_bc": 0.03588409349322319, "train_loss_llm": 0.0, "grad_norm": 0.022341102361679077, "global_step": 814, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05428696051239967, "train_loss_bc": 0.05422316491603851, "train_loss_llm": 0.006379634141921997, "grad_norm": 0.02859603241086006, "global_step": 815, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04854029417037964, "train_loss_bc": 0.04853744059801102, "train_loss_llm": 0.0002855062484741211, "grad_norm": 0.0361311212182045, "global_step": 816, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05243637412786484, "train_loss_bc": 0.052393630146980286, "train_loss_llm": 0.004274338483810425, "grad_norm": 0.023706110194325447, "global_step": 817, "epoch": 22, "lr": 0.001} +{"train_loss": 0.044743288308382034, "train_loss_bc": 0.044743288308382034, "train_loss_llm": 0.0, "grad_norm": 0.03371867910027504, "global_step": 818, "epoch": 22, "lr": 0.001} +{"train_loss": 0.050178784877061844, "train_loss_bc": 0.05017028748989105, "train_loss_llm": 0.0008498132228851318, "grad_norm": 0.04188179597258568, "global_step": 819, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04970981553196907, "train_loss_bc": 0.04970942437648773, "train_loss_llm": 3.916025161743164e-05, "grad_norm": 0.049317967146635056, "global_step": 820, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04968038573861122, "train_loss_bc": 0.04968038573861122, "train_loss_llm": 0.0, "grad_norm": 0.012373870238661766, "global_step": 821, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05323975160717964, "train_loss_bc": 0.053183939307928085, "train_loss_llm": 0.005581378936767578, "grad_norm": 0.019796719774603844, "global_step": 822, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05295531824231148, "train_loss_bc": 0.05295238643884659, "train_loss_llm": 0.00029328465461730957, "grad_norm": 0.022078534588217735, "global_step": 823, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05491567403078079, "train_loss_bc": 0.05491567403078079, "train_loss_llm": 0.0, "grad_norm": 0.0330771878361702, "global_step": 824, "epoch": 22, "lr": 0.001} +{"train_loss": 0.05756732076406479, "train_loss_bc": 0.05756732076406479, "train_loss_llm": 0.0, "grad_norm": 0.017991403117775917, "global_step": 825, "epoch": 22, "lr": 0.001} +{"train_loss": 0.04273246228694916, "train_loss_bc": 0.04273246228694916, "train_loss_llm": 0.0, "grad_norm": 0.022680731490254402, "global_step": 826, "epoch": 22, "lr": 0.001} +{"train_loss": 0.049530171478788056, "train_loss_bc": 0.04818955808877945, "train_loss_llm": 0.002461165189743042, "grad_norm": 0.03496195375919342, "global_step": 827, "epoch": 22, "lr": 0.001, "val_loss": 0.05035330355167389} +{"train_loss": 0.04460366070270538, "train_loss_bc": 0.04456324130296707, "train_loss_llm": 0.004041999578475952, "grad_norm": 0.03660867363214493, "global_step": 828, "epoch": 23, "lr": 0.001} +{"train_loss": 0.051311194896698, "train_loss_bc": 0.051311194896698, "train_loss_llm": 0.0, "grad_norm": 0.01971854455769062, "global_step": 829, "epoch": 23, "lr": 0.001} +{"train_loss": 0.051826462149620056, "train_loss_bc": 0.051826462149620056, "train_loss_llm": 0.0, "grad_norm": 0.02245698869228363, "global_step": 830, "epoch": 23, "lr": 0.001} +{"train_loss": 0.049295634031295776, "train_loss_bc": 0.049295634031295776, "train_loss_llm": 0.0, "grad_norm": 0.033903561532497406, "global_step": 831, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04411850869655609, "train_loss_bc": 0.04411850869655609, "train_loss_llm": 0.0, "grad_norm": 0.03914875164628029, "global_step": 832, "epoch": 23, "lr": 0.001} +{"train_loss": 0.05008081719279289, "train_loss_bc": 0.05006427317857742, "train_loss_llm": 0.0016543269157409668, "grad_norm": 0.018018536269664764, "global_step": 833, "epoch": 23, "lr": 0.001} +{"train_loss": 0.049353986978530884, "train_loss_bc": 0.049353986978530884, "train_loss_llm": 0.0, "grad_norm": 0.02626061998307705, "global_step": 834, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04578107222914696, "train_loss_bc": 0.045765653252601624, "train_loss_llm": 0.0015419423580169678, "grad_norm": 0.03397524356842041, "global_step": 835, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04660802707076073, "train_loss_bc": 0.04660802707076073, "train_loss_llm": 0.0, "grad_norm": 0.03946276381611824, "global_step": 836, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04279499128460884, "train_loss_bc": 0.04279499128460884, "train_loss_llm": 0.0, "grad_norm": 0.01769789680838585, "global_step": 837, "epoch": 23, "lr": 0.001} +{"train_loss": 0.054525211453437805, "train_loss_bc": 0.054494038224220276, "train_loss_llm": 0.003117382526397705, "grad_norm": 0.0252128466963768, "global_step": 838, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04811064526438713, "train_loss_bc": 0.048107586801052094, "train_loss_llm": 0.0003058910369873047, "grad_norm": 0.02672877162694931, "global_step": 839, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04582236707210541, "train_loss_bc": 0.04582236707210541, "train_loss_llm": 0.0, "grad_norm": 0.03438042104244232, "global_step": 840, "epoch": 23, "lr": 0.001} +{"train_loss": 0.0532093271613121, "train_loss_bc": 0.0532093271613121, "train_loss_llm": 0.0, "grad_norm": 0.015872009098529816, "global_step": 841, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04904027283191681, "train_loss_bc": 0.04904027283191681, "train_loss_llm": 0.0, "grad_norm": 0.023484153673052788, "global_step": 842, "epoch": 23, "lr": 0.001} +{"train_loss": 0.060753434896469116, "train_loss_bc": 0.060753434896469116, "train_loss_llm": 0.0, "grad_norm": 0.027259396389126778, "global_step": 843, "epoch": 23, "lr": 0.001} +{"train_loss": 0.048228226602077484, "train_loss_bc": 0.04815853014588356, "train_loss_llm": 0.0069696009159088135, "grad_norm": 0.033024292439222336, "global_step": 844, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04873128607869148, "train_loss_bc": 0.04873128607869148, "train_loss_llm": 0.0, "grad_norm": 0.01679421029984951, "global_step": 845, "epoch": 23, "lr": 0.001} +{"train_loss": 0.05163243040442467, "train_loss_bc": 0.05163243040442467, "train_loss_llm": 0.0, "grad_norm": 0.02644573524594307, "global_step": 846, "epoch": 23, "lr": 0.001} +{"train_loss": 0.06259528547525406, "train_loss_bc": 0.06258159130811691, "train_loss_llm": 0.0013691186904907227, "grad_norm": 0.034512877464294434, "global_step": 847, "epoch": 23, "lr": 0.001} +{"train_loss": 0.044710371643304825, "train_loss_bc": 0.044710371643304825, "train_loss_llm": 0.0, "grad_norm": 0.03573702648282051, "global_step": 848, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04807613790035248, "train_loss_bc": 0.04805627465248108, "train_loss_llm": 0.0019863545894622803, "grad_norm": 0.016008278355002403, "global_step": 849, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04282630234956741, "train_loss_bc": 0.04281451553106308, "train_loss_llm": 0.0011786222457885742, "grad_norm": 0.022659694775938988, "global_step": 850, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04581012949347496, "train_loss_bc": 0.04580414295196533, "train_loss_llm": 0.0005986690521240234, "grad_norm": 0.030652357265353203, "global_step": 851, "epoch": 23, "lr": 0.001} +{"train_loss": 0.05190637707710266, "train_loss_bc": 0.05190637707710266, "train_loss_llm": 0.0, "grad_norm": 0.04335864633321762, "global_step": 852, "epoch": 23, "lr": 0.001} +{"train_loss": 0.05486476421356201, "train_loss_bc": 0.05486476421356201, "train_loss_llm": 0.0, "grad_norm": 0.019894469529390335, "global_step": 853, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04437825456261635, "train_loss_bc": 0.044368911534547806, "train_loss_llm": 0.0009343624114990234, "grad_norm": 0.026792213320732117, "global_step": 854, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04163529723882675, "train_loss_bc": 0.04163529723882675, "train_loss_llm": 0.0, "grad_norm": 0.032907843589782715, "global_step": 855, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04713761433959007, "train_loss_bc": 0.04713761433959007, "train_loss_llm": 0.0, "grad_norm": 0.04113473370671272, "global_step": 856, "epoch": 23, "lr": 0.001} +{"train_loss": 0.0614420548081398, "train_loss_bc": 0.06137670576572418, "train_loss_llm": 0.006534993648529053, "grad_norm": 0.026619112119078636, "global_step": 857, "epoch": 23, "lr": 0.001} +{"train_loss": 0.045893654227256775, "train_loss_bc": 0.045893654227256775, "train_loss_llm": 0.0, "grad_norm": 0.032152727246284485, "global_step": 858, "epoch": 23, "lr": 0.001} +{"train_loss": 0.044976651668548584, "train_loss_bc": 0.04495980218052864, "train_loss_llm": 0.001685023307800293, "grad_norm": 0.03698039799928665, "global_step": 859, "epoch": 23, "lr": 0.001} +{"train_loss": 0.05215929076075554, "train_loss_bc": 0.052155401557683945, "train_loss_llm": 0.0003890693187713623, "grad_norm": 0.04617733135819435, "global_step": 860, "epoch": 23, "lr": 0.001} +{"train_loss": 0.05386245995759964, "train_loss_bc": 0.053846485912799835, "train_loss_llm": 0.001597374677658081, "grad_norm": 0.015137174166738987, "global_step": 861, "epoch": 23, "lr": 0.001} +{"train_loss": 0.044405095279216766, "train_loss_bc": 0.044405095279216766, "train_loss_llm": 0.0, "grad_norm": 0.024896200746297836, "global_step": 862, "epoch": 23, "lr": 0.001} +{"train_loss": 0.04926274582329723, "train_loss_bc": 0.050951551645994186, "train_loss_llm": 0.0, "grad_norm": 0.03769350051879883, "global_step": 863, "epoch": 23, "lr": 0.001, "val_loss": 0.0505957156419754} +{"train_loss": 0.04725523665547371, "train_loss_bc": 0.04724768176674843, "train_loss_llm": 0.000755608081817627, "grad_norm": 0.046824861317873, "global_step": 864, "epoch": 24, "lr": 0.001} +{"train_loss": 0.05117780715227127, "train_loss_bc": 0.051176972687244415, "train_loss_llm": 8.356571197509766e-05, "grad_norm": 0.014673389494419098, "global_step": 865, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04480809345841408, "train_loss_bc": 0.04480809345841408, "train_loss_llm": 0.0, "grad_norm": 0.023131713271141052, "global_step": 866, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04648824408650398, "train_loss_bc": 0.04647457227110863, "train_loss_llm": 0.0013671517372131348, "grad_norm": 0.03183066099882126, "global_step": 867, "epoch": 24, "lr": 0.001} +{"train_loss": 0.042386166751384735, "train_loss_bc": 0.04235827922821045, "train_loss_llm": 0.0027889013290405273, "grad_norm": 0.0405932292342186, "global_step": 868, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04251916706562042, "train_loss_bc": 0.0425022654235363, "train_loss_llm": 0.001690208911895752, "grad_norm": 0.013013768941164017, "global_step": 869, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04450254514813423, "train_loss_bc": 0.04448326677083969, "train_loss_llm": 0.0019279420375823975, "grad_norm": 0.02179584465920925, "global_step": 870, "epoch": 24, "lr": 0.001} +{"train_loss": 0.047964632511138916, "train_loss_bc": 0.047964632511138916, "train_loss_llm": 0.0, "grad_norm": 0.03176490589976311, "global_step": 871, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04895834997296333, "train_loss_bc": 0.04894381761550903, "train_loss_llm": 0.0014531314373016357, "grad_norm": 0.038439009338617325, "global_step": 872, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04030262678861618, "train_loss_bc": 0.04030262678861618, "train_loss_llm": 0.0, "grad_norm": 0.014268133789300919, "global_step": 873, "epoch": 24, "lr": 0.001} +{"train_loss": 0.05119864642620087, "train_loss_bc": 0.05118942633271217, "train_loss_llm": 0.0009219050407409668, "grad_norm": 0.02637104131281376, "global_step": 874, "epoch": 24, "lr": 0.001} +{"train_loss": 0.05133076384663582, "train_loss_bc": 0.05131332203745842, "train_loss_llm": 0.0017440319061279297, "grad_norm": 0.035343851894140244, "global_step": 875, "epoch": 24, "lr": 0.001} +{"train_loss": 0.050471000373363495, "train_loss_bc": 0.050471000373363495, "train_loss_llm": 0.0, "grad_norm": 0.0425790399312973, "global_step": 876, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04590185359120369, "train_loss_bc": 0.045880693942308426, "train_loss_llm": 0.0021161139011383057, "grad_norm": 0.01377555076032877, "global_step": 877, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04894576221704483, "train_loss_bc": 0.04894576221704483, "train_loss_llm": 0.0, "grad_norm": 0.02308770827949047, "global_step": 878, "epoch": 24, "lr": 0.001} +{"train_loss": 0.0519915372133255, "train_loss_bc": 0.05198349803686142, "train_loss_llm": 0.0008040666580200195, "grad_norm": 0.03270358219742775, "global_step": 879, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04987107962369919, "train_loss_bc": 0.049869295209646225, "train_loss_llm": 0.00017842650413513184, "grad_norm": 0.03809539973735809, "global_step": 880, "epoch": 24, "lr": 0.001} +{"train_loss": 0.05580127611756325, "train_loss_bc": 0.05580127611756325, "train_loss_llm": 0.0, "grad_norm": 0.015291284769773483, "global_step": 881, "epoch": 24, "lr": 0.001} +{"train_loss": 0.052149511873722076, "train_loss_bc": 0.052149511873722076, "train_loss_llm": 0.0, "grad_norm": 0.021411458030343056, "global_step": 882, "epoch": 24, "lr": 0.001} +{"train_loss": 0.051568832248449326, "train_loss_bc": 0.051568832248449326, "train_loss_llm": 0.0, "grad_norm": 0.02613506279885769, "global_step": 883, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04214125871658325, "train_loss_bc": 0.04214125871658325, "train_loss_llm": 0.0, "grad_norm": 0.03553682193160057, "global_step": 884, "epoch": 24, "lr": 0.001} +{"train_loss": 0.05264890938997269, "train_loss_bc": 0.052614543586969376, "train_loss_llm": 0.0034367144107818604, "grad_norm": 0.01555558294057846, "global_step": 885, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04596013203263283, "train_loss_bc": 0.045946154743433, "train_loss_llm": 0.0013976097106933594, "grad_norm": 0.022083593532443047, "global_step": 886, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04816940054297447, "train_loss_bc": 0.04816940054297447, "train_loss_llm": 0.0, "grad_norm": 0.025385480374097824, "global_step": 887, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04730420932173729, "train_loss_bc": 0.04730307683348656, "train_loss_llm": 0.0001131594181060791, "grad_norm": 0.029056807979941368, "global_step": 888, "epoch": 24, "lr": 0.001} +{"train_loss": 0.05637122318148613, "train_loss_bc": 0.05637122318148613, "train_loss_llm": 0.0, "grad_norm": 0.019321586936712265, "global_step": 889, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04821978509426117, "train_loss_bc": 0.04821978509426117, "train_loss_llm": 0.0, "grad_norm": 0.025429684668779373, "global_step": 890, "epoch": 24, "lr": 0.001} +{"train_loss": 0.0526760071516037, "train_loss_bc": 0.052671998739242554, "train_loss_llm": 0.0004010200500488281, "grad_norm": 0.03437910974025726, "global_step": 891, "epoch": 24, "lr": 0.001} +{"train_loss": 0.050731875002384186, "train_loss_bc": 0.050721656531095505, "train_loss_llm": 0.0010216832160949707, "grad_norm": 0.037277575582265854, "global_step": 892, "epoch": 24, "lr": 0.001} +{"train_loss": 0.062184397131204605, "train_loss_bc": 0.0621427521109581, "train_loss_llm": 0.004164636135101318, "grad_norm": 0.016711054369807243, "global_step": 893, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04724341258406639, "train_loss_bc": 0.04722535237669945, "train_loss_llm": 0.0018061697483062744, "grad_norm": 0.02179551310837269, "global_step": 894, "epoch": 24, "lr": 0.001} +{"train_loss": 0.046688321977853775, "train_loss_bc": 0.046688321977853775, "train_loss_llm": 0.0, "grad_norm": 0.02411830425262451, "global_step": 895, "epoch": 24, "lr": 0.001} +{"train_loss": 0.05086164176464081, "train_loss_bc": 0.050822824239730835, "train_loss_llm": 0.0038816332817077637, "grad_norm": 0.029214421287178993, "global_step": 896, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04729785770177841, "train_loss_bc": 0.047277167439460754, "train_loss_llm": 0.00206911563873291, "grad_norm": 0.01873028464615345, "global_step": 897, "epoch": 24, "lr": 0.001} +{"train_loss": 0.05040096491575241, "train_loss_bc": 0.05038192495703697, "train_loss_llm": 0.0019041001796722412, "grad_norm": 0.02662784792482853, "global_step": 898, "epoch": 24, "lr": 0.001} +{"train_loss": 0.04906793021493488, "train_loss_bc": 0.05195168778300285, "train_loss_llm": 0.0001271963119506836, "grad_norm": 0.03895556926727295, "global_step": 899, "epoch": 24, "lr": 0.001, "val_loss": 0.05064220726490021} +{"train_loss": 0.03952071815729141, "train_loss_bc": 0.03949859365820885, "train_loss_llm": 0.002212315797805786, "grad_norm": 0.04092773422598839, "global_step": 900, "epoch": 25, "lr": 0.001} +{"train_loss": 0.05028408765792847, "train_loss_bc": 0.05028408765792847, "train_loss_llm": 0.0, "grad_norm": 0.016718951985239983, "global_step": 901, "epoch": 25, "lr": 0.001} +{"train_loss": 0.03674760088324547, "train_loss_bc": 0.03674760088324547, "train_loss_llm": 0.0, "grad_norm": 0.022985177114605904, "global_step": 902, "epoch": 25, "lr": 0.001} +{"train_loss": 0.0613018199801445, "train_loss_bc": 0.0613018199801445, "train_loss_llm": 0.0, "grad_norm": 0.034707210958004, "global_step": 903, "epoch": 25, "lr": 0.001} +{"train_loss": 0.03816359117627144, "train_loss_bc": 0.038111262023448944, "train_loss_llm": 0.005233049392700195, "grad_norm": 0.03955276310443878, "global_step": 904, "epoch": 25, "lr": 0.001} +{"train_loss": 0.05325179919600487, "train_loss_bc": 0.053209029138088226, "train_loss_llm": 0.004276841878890991, "grad_norm": 0.016835715621709824, "global_step": 905, "epoch": 25, "lr": 0.001} +{"train_loss": 0.04710180312395096, "train_loss_bc": 0.047078851610422134, "train_loss_llm": 0.0022951364517211914, "grad_norm": 0.02211189456284046, "global_step": 906, "epoch": 25, "lr": 0.001} +{"train_loss": 0.05179201439023018, "train_loss_bc": 0.05178283154964447, "train_loss_llm": 0.0009181201457977295, "grad_norm": 0.02909340150654316, "global_step": 907, "epoch": 25, "lr": 0.001} +{"train_loss": 0.046132780611515045, "train_loss_bc": 0.046132780611515045, "train_loss_llm": 0.0, "grad_norm": 0.03294687718153, "global_step": 908, "epoch": 25, "lr": 0.001} +{"train_loss": 0.048101626336574554, "train_loss_bc": 0.048101626336574554, "train_loss_llm": 0.0, "grad_norm": 0.014489652588963509, "global_step": 909, "epoch": 25, "lr": 0.001} +{"train_loss": 0.050566475838422775, "train_loss_bc": 0.050566475838422775, "train_loss_llm": 0.0, "grad_norm": 0.0222152229398489, "global_step": 910, "epoch": 25, "lr": 0.001} +{"train_loss": 0.0479990653693676, "train_loss_bc": 0.0479990653693676, "train_loss_llm": 0.0, "grad_norm": 0.02732541784644127, "global_step": 911, "epoch": 25, "lr": 0.001} +{"train_loss": 0.06054874509572983, "train_loss_bc": 0.06054278463125229, "train_loss_llm": 0.0005960166454315186, "grad_norm": 0.031773097813129425, "global_step": 912, "epoch": 25, "lr": 0.001} +{"train_loss": 0.0482119545340538, "train_loss_bc": 0.048208072781562805, "train_loss_llm": 0.00038805603981018066, "grad_norm": 0.01712680608034134, "global_step": 913, "epoch": 25, "lr": 0.001} +{"train_loss": 0.051739275455474854, "train_loss_bc": 0.05169901251792908, "train_loss_llm": 0.004026204347610474, "grad_norm": 0.02740200236439705, "global_step": 914, "epoch": 25, "lr": 0.001} +{"train_loss": 0.04285833612084389, "train_loss_bc": 0.042837414890527725, "train_loss_llm": 0.002092301845550537, "grad_norm": 0.033032938838005066, "global_step": 915, "epoch": 25, "lr": 0.001} +{"train_loss": 0.05660518258810043, "train_loss_bc": 0.05659639462828636, "train_loss_llm": 0.0008788108825683594, "grad_norm": 0.03931451961398125, "global_step": 916, "epoch": 25, "lr": 0.001} +{"train_loss": 0.05267472565174103, "train_loss_bc": 0.05267472565174103, "train_loss_llm": 0.0, "grad_norm": 0.015172308310866356, "global_step": 917, "epoch": 25, "lr": 0.001} +{"train_loss": 0.04366433992981911, "train_loss_bc": 0.04365455359220505, "train_loss_llm": 0.0009784698486328125, "grad_norm": 0.021648459136486053, "global_step": 918, "epoch": 25, "lr": 0.001} +{"train_loss": 0.04855300858616829, "train_loss_bc": 0.04855300858616829, "train_loss_llm": 0.0, "grad_norm": 0.03123720921576023, "global_step": 919, "epoch": 25, "lr": 0.001} +{"train_loss": 0.04873697832226753, "train_loss_bc": 0.048736412078142166, "train_loss_llm": 5.671381950378418e-05, "grad_norm": 0.03200396150350571, "global_step": 920, "epoch": 25, "lr": 0.001} +{"train_loss": 0.044637229293584824, "train_loss_bc": 0.044637229293584824, "train_loss_llm": 0.0, "grad_norm": 0.020353084430098534, "global_step": 921, "epoch": 25, "lr": 0.001} +{"train_loss": 0.049342986196279526, "train_loss_bc": 0.04933478683233261, "train_loss_llm": 0.0008199214935302734, "grad_norm": 0.021113427355885506, "global_step": 922, "epoch": 25, "lr": 0.001} +{"train_loss": 0.0515783429145813, "train_loss_bc": 0.051569052040576935, "train_loss_llm": 0.0009289979934692383, "grad_norm": 0.030049698427319527, "global_step": 923, "epoch": 25, "lr": 0.001} +{"train_loss": 0.05395214632153511, "train_loss_bc": 0.0539386123418808, "train_loss_llm": 0.0013533234596252441, "grad_norm": 0.03838544338941574, "global_step": 924, "epoch": 25, "lr": 0.001} +{"train_loss": 0.04899697005748749, "train_loss_bc": 0.048979464918375015, "train_loss_llm": 0.0017504394054412842, "grad_norm": 0.01927538774907589, "global_step": 925, "epoch": 25, "lr": 0.001} +{"train_loss": 0.04453767463564873, "train_loss_bc": 0.04448657110333443, "train_loss_llm": 0.005110383033752441, "grad_norm": 0.029151281341910362, "global_step": 926, "epoch": 25, "lr": 0.001} +{"train_loss": 0.05136490613222122, "train_loss_bc": 0.05136490613222122, "train_loss_llm": 0.0, "grad_norm": 0.030423900112509727, "global_step": 927, "epoch": 25, "lr": 0.001} +{"train_loss": 0.05040331929922104, "train_loss_bc": 0.05036748945713043, "train_loss_llm": 0.0035828053951263428, "grad_norm": 0.04024991765618324, "global_step": 928, "epoch": 25, "lr": 0.001} +{"train_loss": 0.05615760758519173, "train_loss_bc": 0.05615760758519173, "train_loss_llm": 0.0, "grad_norm": 0.01979129947721958, "global_step": 929, "epoch": 25, "lr": 0.001} +{"train_loss": 0.04238392412662506, "train_loss_bc": 0.0423680804669857, "train_loss_llm": 0.0015842616558074951, "grad_norm": 0.028369301930069923, "global_step": 930, "epoch": 25, "lr": 0.001} +{"train_loss": 0.05250449851155281, "train_loss_bc": 0.05246259644627571, "train_loss_llm": 0.004190027713775635, "grad_norm": 0.03031531348824501, "global_step": 931, "epoch": 25, "lr": 0.001} +{"train_loss": 0.052071914076805115, "train_loss_bc": 0.052071914076805115, "train_loss_llm": 0.0, "grad_norm": 0.0381869375705719, "global_step": 932, "epoch": 25, "lr": 0.001} +{"train_loss": 0.04316573217511177, "train_loss_bc": 0.04316573217511177, "train_loss_llm": 0.0, "grad_norm": 0.016063164919614792, "global_step": 933, "epoch": 25, "lr": 0.001} +{"train_loss": 0.04521955922245979, "train_loss_bc": 0.04521955922245979, "train_loss_llm": 0.0, "grad_norm": 0.025710249319672585, "global_step": 934, "epoch": 25, "lr": 0.001} +{"train_loss": 0.048770459472305246, "train_loss_bc": 0.04484686627984047, "train_loss_llm": 0.001693636178970337, "grad_norm": 0.02978622168302536, "global_step": 935, "epoch": 25, "lr": 0.001, "train/cumulative_reward": 53.11306297376251, "train/mean_score": 4.264729734690755, "train/success_rate": 0.1, "test/cumulative_reward": 11.211574105796032, "test/mean_score": 0.8376864922572679, "test/success_rate": 0.0, "val_loss": 0.05056159570813179, "train_action_mse_error": 0.052687402814626694} +{"train_loss": 0.04457119107246399, "train_loss_bc": 0.04454405605792999, "train_loss_llm": 0.0027134716510772705, "grad_norm": 0.041795335710048676, "global_step": 936, "epoch": 26, "lr": 0.001} +{"train_loss": 0.038021743297576904, "train_loss_bc": 0.03802105039358139, "train_loss_llm": 6.914138793945312e-05, "grad_norm": 0.01891111023724079, "global_step": 937, "epoch": 26, "lr": 0.001} +{"train_loss": 0.056862011551856995, "train_loss_bc": 0.056813132017850876, "train_loss_llm": 0.004887998104095459, "grad_norm": 0.021269597113132477, "global_step": 938, "epoch": 26, "lr": 0.001} +{"train_loss": 0.05013372376561165, "train_loss_bc": 0.050050340592861176, "train_loss_llm": 0.008338391780853271, "grad_norm": 0.02854994870722294, "global_step": 939, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04099099338054657, "train_loss_bc": 0.04098862409591675, "train_loss_llm": 0.00023677945137023926, "grad_norm": 0.028515394777059555, "global_step": 940, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04277006909251213, "train_loss_bc": 0.042752642184495926, "train_loss_llm": 0.0017428696155548096, "grad_norm": 0.014695344492793083, "global_step": 941, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04784166067838669, "train_loss_bc": 0.04777375981211662, "train_loss_llm": 0.006790071725845337, "grad_norm": 0.020213700830936432, "global_step": 942, "epoch": 26, "lr": 0.001} +{"train_loss": 0.047046687453985214, "train_loss_bc": 0.047044605016708374, "train_loss_llm": 0.00020813941955566406, "grad_norm": 0.02685050293803215, "global_step": 943, "epoch": 26, "lr": 0.001} +{"train_loss": 0.05019254982471466, "train_loss_bc": 0.05019254982471466, "train_loss_llm": 0.0, "grad_norm": 0.034774988889694214, "global_step": 944, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04693128541111946, "train_loss_bc": 0.04693128541111946, "train_loss_llm": 0.0, "grad_norm": 0.016426824033260345, "global_step": 945, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04810761660337448, "train_loss_bc": 0.048096176236867905, "train_loss_llm": 0.0011439919471740723, "grad_norm": 0.024006279185414314, "global_step": 946, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04543061554431915, "train_loss_bc": 0.04543061554431915, "train_loss_llm": 0.0, "grad_norm": 0.02926630899310112, "global_step": 947, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04777391999959946, "train_loss_bc": 0.04777391999959946, "train_loss_llm": 0.0, "grad_norm": 0.03216264024376869, "global_step": 948, "epoch": 26, "lr": 0.001} +{"train_loss": 0.05582458898425102, "train_loss_bc": 0.05582166835665703, "train_loss_llm": 0.0002919137477874756, "grad_norm": 0.014520835131406784, "global_step": 949, "epoch": 26, "lr": 0.001} +{"train_loss": 0.056433022022247314, "train_loss_bc": 0.056433022022247314, "train_loss_llm": 0.0, "grad_norm": 0.020190464332699776, "global_step": 950, "epoch": 26, "lr": 0.001} +{"train_loss": 0.061148032546043396, "train_loss_bc": 0.061148032546043396, "train_loss_llm": 0.0, "grad_norm": 0.023076256737113, "global_step": 951, "epoch": 26, "lr": 0.001} +{"train_loss": 0.05384260043501854, "train_loss_bc": 0.05384260043501854, "train_loss_llm": 0.0, "grad_norm": 0.02942088432610035, "global_step": 952, "epoch": 26, "lr": 0.001} +{"train_loss": 0.048205018043518066, "train_loss_bc": 0.04818478599190712, "train_loss_llm": 0.0020232796669006348, "grad_norm": 0.015507985837757587, "global_step": 953, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04821446165442467, "train_loss_bc": 0.04821446165442467, "train_loss_llm": 0.0, "grad_norm": 0.02156274951994419, "global_step": 954, "epoch": 26, "lr": 0.001} +{"train_loss": 0.0398203544318676, "train_loss_bc": 0.03980693966150284, "train_loss_llm": 0.0013413727283477783, "grad_norm": 0.028656307607889175, "global_step": 955, "epoch": 26, "lr": 0.001} +{"train_loss": 0.044795650988817215, "train_loss_bc": 0.04478922858834267, "train_loss_llm": 0.0006423294544219971, "grad_norm": 0.033435698598623276, "global_step": 956, "epoch": 26, "lr": 0.001} +{"train_loss": 0.049311235547065735, "train_loss_bc": 0.049311235547065735, "train_loss_llm": 0.0, "grad_norm": 0.015998665243387222, "global_step": 957, "epoch": 26, "lr": 0.001} +{"train_loss": 0.041311830282211304, "train_loss_bc": 0.041311830282211304, "train_loss_llm": 0.0, "grad_norm": 0.024069322273135185, "global_step": 958, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04410766810178757, "train_loss_bc": 0.044091831892728806, "train_loss_llm": 0.0015836358070373535, "grad_norm": 0.02956698276102543, "global_step": 959, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04161030054092407, "train_loss_bc": 0.04161030054092407, "train_loss_llm": 0.0, "grad_norm": 0.034249648451805115, "global_step": 960, "epoch": 26, "lr": 0.001} +{"train_loss": 0.05514569580554962, "train_loss_bc": 0.05514569580554962, "train_loss_llm": 0.0, "grad_norm": 0.015897635370492935, "global_step": 961, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04894010350108147, "train_loss_bc": 0.04891383275389671, "train_loss_llm": 0.002627164125442505, "grad_norm": 0.02725263684988022, "global_step": 962, "epoch": 26, "lr": 0.001} +{"train_loss": 0.052372515201568604, "train_loss_bc": 0.052372515201568604, "train_loss_llm": 0.0, "grad_norm": 0.030278364196419716, "global_step": 963, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04757487401366234, "train_loss_bc": 0.047563254833221436, "train_loss_llm": 0.0011618435382843018, "grad_norm": 0.0344805084168911, "global_step": 964, "epoch": 26, "lr": 0.001} +{"train_loss": 0.048839498311281204, "train_loss_bc": 0.048839498311281204, "train_loss_llm": 0.0, "grad_norm": 0.01611444167792797, "global_step": 965, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04687213897705078, "train_loss_bc": 0.04687213897705078, "train_loss_llm": 0.0, "grad_norm": 0.026440897956490517, "global_step": 966, "epoch": 26, "lr": 0.001} +{"train_loss": 0.04769119992852211, "train_loss_bc": 0.0476817786693573, "train_loss_llm": 0.0009419918060302734, "grad_norm": 0.03299824893474579, "global_step": 967, "epoch": 26, "lr": 0.001} +{"train_loss": 0.0470191054046154, "train_loss_bc": 0.04700135067105293, "train_loss_llm": 0.0017753839492797852, "grad_norm": 0.04034806042909622, "global_step": 968, "epoch": 26, "lr": 0.001} +{"train_loss": 0.05470987781882286, "train_loss_bc": 0.05470987781882286, "train_loss_llm": 0.0, "grad_norm": 0.014960800297558308, "global_step": 969, "epoch": 26, "lr": 0.001} +{"train_loss": 0.05453360453248024, "train_loss_bc": 0.05453174561262131, "train_loss_llm": 0.00018593668937683105, "grad_norm": 0.026055393740534782, "global_step": 970, "epoch": 26, "lr": 0.001} +{"train_loss": 0.048631205016540155, "train_loss_bc": 0.05571891739964485, "train_loss_llm": 0.0007020235061645508, "grad_norm": 0.03399701416492462, "global_step": 971, "epoch": 26, "lr": 0.001, "val_loss": 0.05055787414312363} +{"train_loss": 0.05316400155425072, "train_loss_bc": 0.05311059579253197, "train_loss_llm": 0.005340754985809326, "grad_norm": 0.04251507669687271, "global_step": 972, "epoch": 27, "lr": 0.001} +{"train_loss": 0.042795199900865555, "train_loss_bc": 0.04279135912656784, "train_loss_llm": 0.0003839731216430664, "grad_norm": 0.012525739148259163, "global_step": 973, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04698263108730316, "train_loss_bc": 0.04698263108730316, "train_loss_llm": 0.0, "grad_norm": 0.0205389354377985, "global_step": 974, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04370833560824394, "train_loss_bc": 0.04370833560824394, "train_loss_llm": 0.0, "grad_norm": 0.02413162961602211, "global_step": 975, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04784888029098511, "train_loss_bc": 0.047833651304244995, "train_loss_llm": 0.0015228688716888428, "grad_norm": 0.025794558227062225, "global_step": 976, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04906504228711128, "train_loss_bc": 0.04906504228711128, "train_loss_llm": 0.0, "grad_norm": 0.019995663315057755, "global_step": 977, "epoch": 27, "lr": 0.001} +{"train_loss": 0.055950332432985306, "train_loss_bc": 0.055950332432985306, "train_loss_llm": 0.0, "grad_norm": 0.027601229026913643, "global_step": 978, "epoch": 27, "lr": 0.001} +{"train_loss": 0.05087876692414284, "train_loss_bc": 0.05087876692414284, "train_loss_llm": 0.0, "grad_norm": 0.033712226897478104, "global_step": 979, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04159146547317505, "train_loss_bc": 0.04159146547317505, "train_loss_llm": 0.0, "grad_norm": 0.04217935726046562, "global_step": 980, "epoch": 27, "lr": 0.001} +{"train_loss": 0.05091272294521332, "train_loss_bc": 0.050885751843452454, "train_loss_llm": 0.002697199583053589, "grad_norm": 0.015194246545433998, "global_step": 981, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04151535406708717, "train_loss_bc": 0.04150913655757904, "train_loss_llm": 0.0006216466426849365, "grad_norm": 0.025422079488635063, "global_step": 982, "epoch": 27, "lr": 0.001} +{"train_loss": 0.051205579191446304, "train_loss_bc": 0.05119580402970314, "train_loss_llm": 0.0009773671627044678, "grad_norm": 0.031910233199596405, "global_step": 983, "epoch": 27, "lr": 0.001} +{"train_loss": 0.042345087975263596, "train_loss_bc": 0.04233784228563309, "train_loss_llm": 0.000724494457244873, "grad_norm": 0.03824663907289505, "global_step": 984, "epoch": 27, "lr": 0.001} +{"train_loss": 0.049854423850774765, "train_loss_bc": 0.04982294514775276, "train_loss_llm": 0.003147721290588379, "grad_norm": 0.018419068306684494, "global_step": 985, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04888538271188736, "train_loss_bc": 0.04888538271188736, "train_loss_llm": 0.0, "grad_norm": 0.023876745253801346, "global_step": 986, "epoch": 27, "lr": 0.001} +{"train_loss": 0.046896833926439285, "train_loss_bc": 0.046896833926439285, "train_loss_llm": 0.0, "grad_norm": 0.029347794130444527, "global_step": 987, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04930288344621658, "train_loss_bc": 0.04930288344621658, "train_loss_llm": 0.0, "grad_norm": 0.032465606927871704, "global_step": 988, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04387583211064339, "train_loss_bc": 0.04386723414063454, "train_loss_llm": 0.0008598268032073975, "grad_norm": 0.014916733838617802, "global_step": 989, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04378510266542435, "train_loss_bc": 0.04377788305282593, "train_loss_llm": 0.0007219016551971436, "grad_norm": 0.01888609677553177, "global_step": 990, "epoch": 27, "lr": 0.001} +{"train_loss": 0.05898766964673996, "train_loss_bc": 0.05898766964673996, "train_loss_llm": 0.0, "grad_norm": 0.02611583098769188, "global_step": 991, "epoch": 27, "lr": 0.001} +{"train_loss": 0.05075905844569206, "train_loss_bc": 0.050743456929922104, "train_loss_llm": 0.0015599727630615234, "grad_norm": 0.030363846570253372, "global_step": 992, "epoch": 27, "lr": 0.001} +{"train_loss": 0.043745219707489014, "train_loss_bc": 0.04374070465564728, "train_loss_llm": 0.0004515647888183594, "grad_norm": 0.013555825687944889, "global_step": 993, "epoch": 27, "lr": 0.001} +{"train_loss": 0.048156265169382095, "train_loss_bc": 0.048156265169382095, "train_loss_llm": 0.0, "grad_norm": 0.01784861832857132, "global_step": 994, "epoch": 27, "lr": 0.001} +{"train_loss": 0.05600738152861595, "train_loss_bc": 0.05600738152861595, "train_loss_llm": 0.0, "grad_norm": 0.030461721122264862, "global_step": 995, "epoch": 27, "lr": 0.001} +{"train_loss": 0.05257892236113548, "train_loss_bc": 0.052570804953575134, "train_loss_llm": 0.0008116364479064941, "grad_norm": 0.03170604631304741, "global_step": 996, "epoch": 27, "lr": 0.001} +{"train_loss": 0.05244527757167816, "train_loss_bc": 0.052436307072639465, "train_loss_llm": 0.0008969604969024658, "grad_norm": 0.01929805427789688, "global_step": 997, "epoch": 27, "lr": 0.001} +{"train_loss": 0.050777334719896317, "train_loss_bc": 0.05074838176369667, "train_loss_llm": 0.002895176410675049, "grad_norm": 0.03396955877542496, "global_step": 998, "epoch": 27, "lr": 0.001} +{"train_loss": 0.053812719881534576, "train_loss_bc": 0.053812719881534576, "train_loss_llm": 0.0, "grad_norm": 0.03020925633609295, "global_step": 999, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04613744467496872, "train_loss_bc": 0.0461336188018322, "train_loss_llm": 0.00038248300552368164, "grad_norm": 0.03197542950510979, "global_step": 1000, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04632498696446419, "train_loss_bc": 0.046309228986501694, "train_loss_llm": 0.0015757083892822266, "grad_norm": 0.019021406769752502, "global_step": 1001, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04369017481803894, "train_loss_bc": 0.04367707297205925, "train_loss_llm": 0.0013102591037750244, "grad_norm": 0.02948182448744774, "global_step": 1002, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04247933626174927, "train_loss_bc": 0.042465005069971085, "train_loss_llm": 0.0014330744743347168, "grad_norm": 0.041027456521987915, "global_step": 1003, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04889171943068504, "train_loss_bc": 0.04887661337852478, "train_loss_llm": 0.0015107393264770508, "grad_norm": 0.05759160965681076, "global_step": 1004, "epoch": 27, "lr": 0.001} +{"train_loss": 0.05275999754667282, "train_loss_bc": 0.05275249108672142, "train_loss_llm": 0.0007507503032684326, "grad_norm": 0.016552355140447617, "global_step": 1005, "epoch": 27, "lr": 0.001} +{"train_loss": 0.04973578080534935, "train_loss_bc": 0.04965982586145401, "train_loss_llm": 0.00759539008140564, "grad_norm": 0.02333229035139084, "global_step": 1006, "epoch": 27, "lr": 0.001} +{"train_loss": 0.048359274967677064, "train_loss_bc": 0.04308075085282326, "train_loss_llm": 0.0, "grad_norm": 0.03463708609342575, "global_step": 1007, "epoch": 27, "lr": 0.001, "val_loss": 0.050547149032354355} +{"train_loss": 0.03975335881114006, "train_loss_bc": 0.03975335881114006, "train_loss_llm": 0.0, "grad_norm": 0.03993407264351845, "global_step": 1008, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04973059520125389, "train_loss_bc": 0.049727726727724075, "train_loss_llm": 0.0002866685390472412, "grad_norm": 0.022941293194890022, "global_step": 1009, "epoch": 28, "lr": 0.001} +{"train_loss": 0.039422716945409775, "train_loss_bc": 0.039395540952682495, "train_loss_llm": 0.0027174651622772217, "grad_norm": 0.02147454023361206, "global_step": 1010, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04655938595533371, "train_loss_bc": 0.04650871083140373, "train_loss_llm": 0.00506746768951416, "grad_norm": 0.028370346873998642, "global_step": 1011, "epoch": 28, "lr": 0.001} +{"train_loss": 0.057008739560842514, "train_loss_bc": 0.057008739560842514, "train_loss_llm": 0.0, "grad_norm": 0.03528646379709244, "global_step": 1012, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04893115535378456, "train_loss_bc": 0.04893115535378456, "train_loss_llm": 0.0, "grad_norm": 0.015434889122843742, "global_step": 1013, "epoch": 28, "lr": 0.001} +{"train_loss": 0.044782280921936035, "train_loss_bc": 0.04477835074067116, "train_loss_llm": 0.0003929734230041504, "grad_norm": 0.023832838982343674, "global_step": 1014, "epoch": 28, "lr": 0.001} +{"train_loss": 0.05171110853552818, "train_loss_bc": 0.05171110853552818, "train_loss_llm": 0.0, "grad_norm": 0.030032167211174965, "global_step": 1015, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04449166730046272, "train_loss_bc": 0.044489581137895584, "train_loss_llm": 0.00020870566368103027, "grad_norm": 0.03734353557229042, "global_step": 1016, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04896632954478264, "train_loss_bc": 0.04896632954478264, "train_loss_llm": 0.0, "grad_norm": 0.01598871313035488, "global_step": 1017, "epoch": 28, "lr": 0.001} +{"train_loss": 0.040149468928575516, "train_loss_bc": 0.04012428596615791, "train_loss_llm": 0.002518296241760254, "grad_norm": 0.023731231689453125, "global_step": 1018, "epoch": 28, "lr": 0.001} +{"train_loss": 0.05543471500277519, "train_loss_bc": 0.05543471500277519, "train_loss_llm": 0.0, "grad_norm": 0.027802346274256706, "global_step": 1019, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04902983456850052, "train_loss_bc": 0.04902983456850052, "train_loss_llm": 0.0, "grad_norm": 0.039355386048555374, "global_step": 1020, "epoch": 28, "lr": 0.001} +{"train_loss": 0.0445585772395134, "train_loss_bc": 0.04451419785618782, "train_loss_llm": 0.00443798303604126, "grad_norm": 0.013984203338623047, "global_step": 1021, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04075806215405464, "train_loss_bc": 0.04075664281845093, "train_loss_llm": 0.00014197826385498047, "grad_norm": 0.018953662365674973, "global_step": 1022, "epoch": 28, "lr": 0.001} +{"train_loss": 0.059405215084552765, "train_loss_bc": 0.059405215084552765, "train_loss_llm": 0.0, "grad_norm": 0.02581487037241459, "global_step": 1023, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04705927520990372, "train_loss_bc": 0.04705927520990372, "train_loss_llm": 0.0, "grad_norm": 0.032201845198869705, "global_step": 1024, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04355058819055557, "train_loss_bc": 0.04353147745132446, "train_loss_llm": 0.001911073923110962, "grad_norm": 0.015035771764814854, "global_step": 1025, "epoch": 28, "lr": 0.001} +{"train_loss": 0.05085901916027069, "train_loss_bc": 0.05085901916027069, "train_loss_llm": 0.0, "grad_norm": 0.02183963917195797, "global_step": 1026, "epoch": 28, "lr": 0.001} +{"train_loss": 0.05398085340857506, "train_loss_bc": 0.053950343281030655, "train_loss_llm": 0.0030509531497955322, "grad_norm": 0.026193393394351006, "global_step": 1027, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04196279123425484, "train_loss_bc": 0.04196279123425484, "train_loss_llm": 0.0, "grad_norm": 0.030476348474621773, "global_step": 1028, "epoch": 28, "lr": 0.001} +{"train_loss": 0.05810581520199776, "train_loss_bc": 0.05810581520199776, "train_loss_llm": 0.0, "grad_norm": 0.022100668400526047, "global_step": 1029, "epoch": 28, "lr": 0.001} +{"train_loss": 0.046485308557748795, "train_loss_bc": 0.046485308557748795, "train_loss_llm": 0.0, "grad_norm": 0.037729229778051376, "global_step": 1030, "epoch": 28, "lr": 0.001} +{"train_loss": 0.05242615565657616, "train_loss_bc": 0.052409734576940536, "train_loss_llm": 0.001641988754272461, "grad_norm": 0.04446086287498474, "global_step": 1031, "epoch": 28, "lr": 0.001} +{"train_loss": 0.049404632300138474, "train_loss_bc": 0.049404632300138474, "train_loss_llm": 0.0, "grad_norm": 0.04708107188344002, "global_step": 1032, "epoch": 28, "lr": 0.001} +{"train_loss": 0.047711219638586044, "train_loss_bc": 0.047680534422397614, "train_loss_llm": 0.003068476915359497, "grad_norm": 0.013491692021489143, "global_step": 1033, "epoch": 28, "lr": 0.001} +{"train_loss": 0.049615517258644104, "train_loss_bc": 0.049615517258644104, "train_loss_llm": 0.0, "grad_norm": 0.02372836321592331, "global_step": 1034, "epoch": 28, "lr": 0.001} +{"train_loss": 0.050846993923187256, "train_loss_bc": 0.050846993923187256, "train_loss_llm": 0.0, "grad_norm": 0.028722360730171204, "global_step": 1035, "epoch": 28, "lr": 0.001} +{"train_loss": 0.05233466997742653, "train_loss_bc": 0.05233466997742653, "train_loss_llm": 0.0, "grad_norm": 0.03830366209149361, "global_step": 1036, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04739023372530937, "train_loss_bc": 0.04739023372530937, "train_loss_llm": 0.0, "grad_norm": 0.01942819356918335, "global_step": 1037, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04609264060854912, "train_loss_bc": 0.046079400926828384, "train_loss_llm": 0.0013238787651062012, "grad_norm": 0.027965134009718895, "global_step": 1038, "epoch": 28, "lr": 0.001} +{"train_loss": 0.038386665284633636, "train_loss_bc": 0.038386665284633636, "train_loss_llm": 0.0, "grad_norm": 0.03475569933652878, "global_step": 1039, "epoch": 28, "lr": 0.001} +{"train_loss": 0.05057235807180405, "train_loss_bc": 0.05057235807180405, "train_loss_llm": 0.0, "grad_norm": 0.03520738333463669, "global_step": 1040, "epoch": 28, "lr": 0.001} +{"train_loss": 0.058592718094587326, "train_loss_bc": 0.058592718094587326, "train_loss_llm": 0.0, "grad_norm": 0.018018223345279694, "global_step": 1041, "epoch": 28, "lr": 0.001} +{"train_loss": 0.043081704527139664, "train_loss_bc": 0.043081704527139664, "train_loss_llm": 0.0, "grad_norm": 0.027376264333724976, "global_step": 1042, "epoch": 28, "lr": 0.001} +{"train_loss": 0.04828272956526942, "train_loss_bc": 0.04894886910915375, "train_loss_llm": 0.007702350616455078, "grad_norm": 0.04406509920954704, "global_step": 1043, "epoch": 28, "lr": 0.001, "val_loss": 0.05014680325984955} +{"train_loss": 0.04550962522625923, "train_loss_bc": 0.04550962522625923, "train_loss_llm": 0.0, "grad_norm": 0.04785943776369095, "global_step": 1044, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04196885600686073, "train_loss_bc": 0.041934557259082794, "train_loss_llm": 0.003429800271987915, "grad_norm": 0.014519184827804565, "global_step": 1045, "epoch": 29, "lr": 0.001} +{"train_loss": 0.055159490555524826, "train_loss_bc": 0.055159490555524826, "train_loss_llm": 0.0, "grad_norm": 0.017880389466881752, "global_step": 1046, "epoch": 29, "lr": 0.001} +{"train_loss": 0.050568629056215286, "train_loss_bc": 0.05054231360554695, "train_loss_llm": 0.00263136625289917, "grad_norm": 0.024878688156604767, "global_step": 1047, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04039904475212097, "train_loss_bc": 0.04035036265850067, "train_loss_llm": 0.004868239164352417, "grad_norm": 0.031141774728894234, "global_step": 1048, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04428543895483017, "train_loss_bc": 0.04428097605705261, "train_loss_llm": 0.0004463493824005127, "grad_norm": 0.01564970053732395, "global_step": 1049, "epoch": 29, "lr": 0.001} +{"train_loss": 0.049695320427417755, "train_loss_bc": 0.049695320427417755, "train_loss_llm": 0.0, "grad_norm": 0.030801638960838318, "global_step": 1050, "epoch": 29, "lr": 0.001} +{"train_loss": 0.05365500971674919, "train_loss_bc": 0.05365500971674919, "train_loss_llm": 0.0, "grad_norm": 0.03409305214881897, "global_step": 1051, "epoch": 29, "lr": 0.001} +{"train_loss": 0.05196087062358856, "train_loss_bc": 0.05193901062011719, "train_loss_llm": 0.0021861791610717773, "grad_norm": 0.0389067642390728, "global_step": 1052, "epoch": 29, "lr": 0.001} +{"train_loss": 0.0482519306242466, "train_loss_bc": 0.04825082793831825, "train_loss_llm": 0.00011023879051208496, "grad_norm": 0.018722116947174072, "global_step": 1053, "epoch": 29, "lr": 0.001} +{"train_loss": 0.053426556289196014, "train_loss_bc": 0.053426556289196014, "train_loss_llm": 0.0, "grad_norm": 0.025745602324604988, "global_step": 1054, "epoch": 29, "lr": 0.001} +{"train_loss": 0.05382269620895386, "train_loss_bc": 0.05380392447113991, "train_loss_llm": 0.0018770694732666016, "grad_norm": 0.028508121147751808, "global_step": 1055, "epoch": 29, "lr": 0.001} +{"train_loss": 0.05323948338627815, "train_loss_bc": 0.05323948338627815, "train_loss_llm": 0.0, "grad_norm": 0.03327299281954765, "global_step": 1056, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04601266235113144, "train_loss_bc": 0.04601266235113144, "train_loss_llm": 0.0, "grad_norm": 0.019950520247220993, "global_step": 1057, "epoch": 29, "lr": 0.001} +{"train_loss": 0.03858322277665138, "train_loss_bc": 0.03857710584998131, "train_loss_llm": 0.0006115734577178955, "grad_norm": 0.028467101976275444, "global_step": 1058, "epoch": 29, "lr": 0.001} +{"train_loss": 0.03843758627772331, "train_loss_bc": 0.03843387961387634, "train_loss_llm": 0.0003705322742462158, "grad_norm": 0.03370156139135361, "global_step": 1059, "epoch": 29, "lr": 0.001} +{"train_loss": 0.053523026406764984, "train_loss_bc": 0.05350584536790848, "train_loss_llm": 0.0017180442810058594, "grad_norm": 0.03279431536793709, "global_step": 1060, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04421687126159668, "train_loss_bc": 0.04421687126159668, "train_loss_llm": 0.0, "grad_norm": 0.016467615962028503, "global_step": 1061, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04725513979792595, "train_loss_bc": 0.04724029824137688, "train_loss_llm": 0.0014840364456176758, "grad_norm": 0.024948077276349068, "global_step": 1062, "epoch": 29, "lr": 0.001} +{"train_loss": 0.05803482234477997, "train_loss_bc": 0.058021754026412964, "train_loss_llm": 0.001306772232055664, "grad_norm": 0.032162342220544815, "global_step": 1063, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04164743795990944, "train_loss_bc": 0.04163352772593498, "train_loss_llm": 0.001391083002090454, "grad_norm": 0.04142409563064575, "global_step": 1064, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04885931313037872, "train_loss_bc": 0.04885931313037872, "train_loss_llm": 0.0, "grad_norm": 0.017388304695487022, "global_step": 1065, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04657549411058426, "train_loss_bc": 0.04657549411058426, "train_loss_llm": 0.0, "grad_norm": 0.02129034325480461, "global_step": 1066, "epoch": 29, "lr": 0.001} +{"train_loss": 0.050415292382240295, "train_loss_bc": 0.050387781113386154, "train_loss_llm": 0.0027510523796081543, "grad_norm": 0.02659629099071026, "global_step": 1067, "epoch": 29, "lr": 0.001} +{"train_loss": 0.05714399367570877, "train_loss_bc": 0.057110607624053955, "train_loss_llm": 0.0033385157585144043, "grad_norm": 0.030709320679306984, "global_step": 1068, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04908182844519615, "train_loss_bc": 0.04908182844519615, "train_loss_llm": 0.0, "grad_norm": 0.015563186258077621, "global_step": 1069, "epoch": 29, "lr": 0.001} +{"train_loss": 0.037166766822338104, "train_loss_bc": 0.037166766822338104, "train_loss_llm": 0.0, "grad_norm": 0.02461824379861355, "global_step": 1070, "epoch": 29, "lr": 0.001} +{"train_loss": 0.055672649294137955, "train_loss_bc": 0.055672649294137955, "train_loss_llm": 0.0, "grad_norm": 0.03356856480240822, "global_step": 1071, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04151465743780136, "train_loss_bc": 0.04151465743780136, "train_loss_llm": 0.0, "grad_norm": 0.03759726881980896, "global_step": 1072, "epoch": 29, "lr": 0.001} +{"train_loss": 0.055281974375247955, "train_loss_bc": 0.055281974375247955, "train_loss_llm": 0.0, "grad_norm": 0.017949631437659264, "global_step": 1073, "epoch": 29, "lr": 0.001} +{"train_loss": 0.057101186364889145, "train_loss_bc": 0.05703601986169815, "train_loss_llm": 0.006516635417938232, "grad_norm": 0.029285544529557228, "global_step": 1074, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04751468077301979, "train_loss_bc": 0.0474901907145977, "train_loss_llm": 0.0024489164352416992, "grad_norm": 0.03380931541323662, "global_step": 1075, "epoch": 29, "lr": 0.001} +{"train_loss": 0.03669052943587303, "train_loss_bc": 0.03669052943587303, "train_loss_llm": 0.0, "grad_norm": 0.03737975284457207, "global_step": 1076, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04975668713450432, "train_loss_bc": 0.04975668713450432, "train_loss_llm": 0.0, "grad_norm": 0.02024662494659424, "global_step": 1077, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04487217217683792, "train_loss_bc": 0.04487217217683792, "train_loss_llm": 0.0, "grad_norm": 0.027362573891878128, "global_step": 1078, "epoch": 29, "lr": 0.001} +{"train_loss": 0.04802583230452405, "train_loss_bc": 0.041629016399383545, "train_loss_llm": 0.0, "grad_norm": 0.03342261537909508, "global_step": 1079, "epoch": 29, "lr": 0.001, "val_loss": 0.05003791302442551} +{"train_loss": 0.04663439095020294, "train_loss_bc": 0.04663439095020294, "train_loss_llm": 0.0, "grad_norm": 0.040643174201250076, "global_step": 1080, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04669461026787758, "train_loss_bc": 0.04669461026787758, "train_loss_llm": 0.0, "grad_norm": 0.017504652962088585, "global_step": 1081, "epoch": 30, "lr": 0.001} +{"train_loss": 0.041483763605356216, "train_loss_bc": 0.041483763605356216, "train_loss_llm": 0.0, "grad_norm": 0.02651299722492695, "global_step": 1082, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04191115126013756, "train_loss_bc": 0.04191115126013756, "train_loss_llm": 0.0, "grad_norm": 0.03690016269683838, "global_step": 1083, "epoch": 30, "lr": 0.001} +{"train_loss": 0.05576413869857788, "train_loss_bc": 0.055730026215314865, "train_loss_llm": 0.0034112930297851562, "grad_norm": 0.04099569097161293, "global_step": 1084, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04617970064282417, "train_loss_bc": 0.04616495221853256, "train_loss_llm": 0.0014749765396118164, "grad_norm": 0.020555181428790092, "global_step": 1085, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04983530193567276, "train_loss_bc": 0.04983290284872055, "train_loss_llm": 0.00023987889289855957, "grad_norm": 0.024403216317296028, "global_step": 1086, "epoch": 30, "lr": 0.001} +{"train_loss": 0.048039041459560394, "train_loss_bc": 0.048039041459560394, "train_loss_llm": 0.0, "grad_norm": 0.03500664234161377, "global_step": 1087, "epoch": 30, "lr": 0.001} +{"train_loss": 0.05734354630112648, "train_loss_bc": 0.05734354630112648, "train_loss_llm": 0.0, "grad_norm": 0.04453141614794731, "global_step": 1088, "epoch": 30, "lr": 0.001} +{"train_loss": 0.050401341170072556, "train_loss_bc": 0.050351209938526154, "train_loss_llm": 0.005013018846511841, "grad_norm": 0.017837312072515488, "global_step": 1089, "epoch": 30, "lr": 0.001} +{"train_loss": 0.036788057535886765, "train_loss_bc": 0.036788057535886765, "train_loss_llm": 0.0, "grad_norm": 0.023793499916791916, "global_step": 1090, "epoch": 30, "lr": 0.001} +{"train_loss": 0.045246705412864685, "train_loss_bc": 0.045246705412864685, "train_loss_llm": 0.0, "grad_norm": 0.03244409337639809, "global_step": 1091, "epoch": 30, "lr": 0.001} +{"train_loss": 0.057592228055000305, "train_loss_bc": 0.057592228055000305, "train_loss_llm": 0.0, "grad_norm": 0.04119754210114479, "global_step": 1092, "epoch": 30, "lr": 0.001} +{"train_loss": 0.03831659257411957, "train_loss_bc": 0.03831659257411957, "train_loss_llm": 0.0, "grad_norm": 0.013706473633646965, "global_step": 1093, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04486655071377754, "train_loss_bc": 0.04486655071377754, "train_loss_llm": 0.0, "grad_norm": 0.020055735483765602, "global_step": 1094, "epoch": 30, "lr": 0.001} +{"train_loss": 0.054290443658828735, "train_loss_bc": 0.054290443658828735, "train_loss_llm": 0.0, "grad_norm": 0.022756271064281464, "global_step": 1095, "epoch": 30, "lr": 0.001} +{"train_loss": 0.057376962155103683, "train_loss_bc": 0.057376962155103683, "train_loss_llm": 0.0, "grad_norm": 0.027627184987068176, "global_step": 1096, "epoch": 30, "lr": 0.001} +{"train_loss": 0.045285001397132874, "train_loss_bc": 0.04526069760322571, "train_loss_llm": 0.0024304091930389404, "grad_norm": 0.014454932883381844, "global_step": 1097, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04833754524588585, "train_loss_bc": 0.04831342026591301, "train_loss_llm": 0.0024124979972839355, "grad_norm": 0.01914345845580101, "global_step": 1098, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04882405325770378, "train_loss_bc": 0.04880472272634506, "train_loss_llm": 0.0019330382347106934, "grad_norm": 0.02504720352590084, "global_step": 1099, "epoch": 30, "lr": 0.001} +{"train_loss": 0.049246691167354584, "train_loss_bc": 0.049246691167354584, "train_loss_llm": 0.0, "grad_norm": 0.029667766764760017, "global_step": 1100, "epoch": 30, "lr": 0.001} +{"train_loss": 0.05077924206852913, "train_loss_bc": 0.05073001980781555, "train_loss_llm": 0.004922330379486084, "grad_norm": 0.02354213036596775, "global_step": 1101, "epoch": 30, "lr": 0.001} +{"train_loss": 0.050599679350852966, "train_loss_bc": 0.050599679350852966, "train_loss_llm": 0.0, "grad_norm": 0.03016488067805767, "global_step": 1102, "epoch": 30, "lr": 0.001} +{"train_loss": 0.047513868659734726, "train_loss_bc": 0.047505319118499756, "train_loss_llm": 0.0008549392223358154, "grad_norm": 0.03213400021195412, "global_step": 1103, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04953571781516075, "train_loss_bc": 0.04952944070100784, "train_loss_llm": 0.0006275773048400879, "grad_norm": 0.03135112673044205, "global_step": 1104, "epoch": 30, "lr": 0.001} +{"train_loss": 0.047984778881073, "train_loss_bc": 0.047984778881073, "train_loss_llm": 0.0, "grad_norm": 0.016121255233883858, "global_step": 1105, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04712681099772453, "train_loss_bc": 0.04711643606424332, "train_loss_llm": 0.0010374188423156738, "grad_norm": 0.023895278573036194, "global_step": 1106, "epoch": 30, "lr": 0.001} +{"train_loss": 0.03772816061973572, "train_loss_bc": 0.03772406280040741, "train_loss_llm": 0.00040972232818603516, "grad_norm": 0.02795850671827793, "global_step": 1107, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04614332690834999, "train_loss_bc": 0.0461241640150547, "train_loss_llm": 0.0019162297248840332, "grad_norm": 0.03989759460091591, "global_step": 1108, "epoch": 30, "lr": 0.001} +{"train_loss": 0.054367464035749435, "train_loss_bc": 0.05435815081000328, "train_loss_llm": 0.0009312629699707031, "grad_norm": 0.01651105470955372, "global_step": 1109, "epoch": 30, "lr": 0.001} +{"train_loss": 0.05259837582707405, "train_loss_bc": 0.05259213224053383, "train_loss_llm": 0.0006245076656341553, "grad_norm": 0.024051522836089134, "global_step": 1110, "epoch": 30, "lr": 0.001} +{"train_loss": 0.042050380259752274, "train_loss_bc": 0.042050380259752274, "train_loss_llm": 0.0, "grad_norm": 0.032269686460494995, "global_step": 1111, "epoch": 30, "lr": 0.001} +{"train_loss": 0.046795085072517395, "train_loss_bc": 0.046795085072517395, "train_loss_llm": 0.0, "grad_norm": 0.03433522582054138, "global_step": 1112, "epoch": 30, "lr": 0.001} +{"train_loss": 0.048710890114307404, "train_loss_bc": 0.048710890114307404, "train_loss_llm": 0.0, "grad_norm": 0.013442844152450562, "global_step": 1113, "epoch": 30, "lr": 0.001} +{"train_loss": 0.04333048313856125, "train_loss_bc": 0.04333048313856125, "train_loss_llm": 0.0, "grad_norm": 0.028970375657081604, "global_step": 1114, "epoch": 30, "lr": 0.001} +{"train_loss": 0.048029058509402804, "train_loss_bc": 0.05332402512431145, "train_loss_llm": 0.0, "grad_norm": 0.04124332219362259, "global_step": 1115, "epoch": 30, "lr": 0.001, "train/cumulative_reward": 18.461640224960163, "train/mean_score": 3.34758933714849, "train/success_rate": 0.1, "test/cumulative_reward": 32.624661756613044, "test/mean_score": 2.9241424172262724, "test/success_rate": 0.04, "val_loss": 0.050925854593515396, "train_action_mse_error": 0.051819562911987305} +{"train_loss": 0.04820284992456436, "train_loss_bc": 0.048159196972846985, "train_loss_llm": 0.004365444183349609, "grad_norm": 0.04786582291126251, "global_step": 1116, "epoch": 31, "lr": 0.001} +{"train_loss": 0.05451276898384094, "train_loss_bc": 0.05449402332305908, "train_loss_llm": 0.0018745958805084229, "grad_norm": 0.013303055427968502, "global_step": 1117, "epoch": 31, "lr": 0.001} +{"train_loss": 0.051816411316394806, "train_loss_bc": 0.051816411316394806, "train_loss_llm": 0.0, "grad_norm": 0.022502388805150986, "global_step": 1118, "epoch": 31, "lr": 0.001} +{"train_loss": 0.034763090312480927, "train_loss_bc": 0.03475434333086014, "train_loss_llm": 0.0008746981620788574, "grad_norm": 0.028990378603339195, "global_step": 1119, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04283514246344566, "train_loss_bc": 0.04282544553279877, "train_loss_llm": 0.0009698569774627686, "grad_norm": 0.03401005268096924, "global_step": 1120, "epoch": 31, "lr": 0.001} +{"train_loss": 0.05628882721066475, "train_loss_bc": 0.05628882721066475, "train_loss_llm": 0.0, "grad_norm": 0.01556380558758974, "global_step": 1121, "epoch": 31, "lr": 0.001} +{"train_loss": 0.0476580448448658, "train_loss_bc": 0.04765409976243973, "train_loss_llm": 0.00039464235305786133, "grad_norm": 0.017285384237766266, "global_step": 1122, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04935827851295471, "train_loss_bc": 0.04935827851295471, "train_loss_llm": 0.0, "grad_norm": 0.025011692196130753, "global_step": 1123, "epoch": 31, "lr": 0.001} +{"train_loss": 0.053621985018253326, "train_loss_bc": 0.053621985018253326, "train_loss_llm": 0.0, "grad_norm": 0.03422284126281738, "global_step": 1124, "epoch": 31, "lr": 0.001} +{"train_loss": 0.055752042680978775, "train_loss_bc": 0.055752042680978775, "train_loss_llm": 0.0, "grad_norm": 0.019491886720061302, "global_step": 1125, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04442083463072777, "train_loss_bc": 0.044414497911930084, "train_loss_llm": 0.0006338357925415039, "grad_norm": 0.02016003616154194, "global_step": 1126, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04534314200282097, "train_loss_bc": 0.04534314200282097, "train_loss_llm": 0.0, "grad_norm": 0.02756091021001339, "global_step": 1127, "epoch": 31, "lr": 0.001} +{"train_loss": 0.05297340825200081, "train_loss_bc": 0.05292776599526405, "train_loss_llm": 0.0045642852783203125, "grad_norm": 0.03227188438177109, "global_step": 1128, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04579494521021843, "train_loss_bc": 0.04578348994255066, "train_loss_llm": 0.0011455416679382324, "grad_norm": 0.014233858324587345, "global_step": 1129, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04591844603419304, "train_loss_bc": 0.04591844603419304, "train_loss_llm": 0.0, "grad_norm": 0.01903476007282734, "global_step": 1130, "epoch": 31, "lr": 0.001} +{"train_loss": 0.05317706614732742, "train_loss_bc": 0.05316861718893051, "train_loss_llm": 0.0008449554443359375, "grad_norm": 0.024057885631918907, "global_step": 1131, "epoch": 31, "lr": 0.001} +{"train_loss": 0.042418550699949265, "train_loss_bc": 0.042418550699949265, "train_loss_llm": 0.0, "grad_norm": 0.030207689851522446, "global_step": 1132, "epoch": 31, "lr": 0.001} +{"train_loss": 0.047671400010585785, "train_loss_bc": 0.047671400010585785, "train_loss_llm": 0.0, "grad_norm": 0.013611352071166039, "global_step": 1133, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04770166799426079, "train_loss_bc": 0.04770166799426079, "train_loss_llm": 0.0, "grad_norm": 0.018786385655403137, "global_step": 1134, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04303646460175514, "train_loss_bc": 0.04301496595144272, "train_loss_llm": 0.002149909734725952, "grad_norm": 0.027570798993110657, "global_step": 1135, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04980296641588211, "train_loss_bc": 0.04980296641588211, "train_loss_llm": 0.0, "grad_norm": 0.03171715512871742, "global_step": 1136, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04887416586279869, "train_loss_bc": 0.048866480588912964, "train_loss_llm": 0.0007685720920562744, "grad_norm": 0.016292480751872063, "global_step": 1137, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04917452484369278, "train_loss_bc": 0.04917452484369278, "train_loss_llm": 0.0, "grad_norm": 0.026502562686800957, "global_step": 1138, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04675821587443352, "train_loss_bc": 0.04675821587443352, "train_loss_llm": 0.0, "grad_norm": 0.03595609962940216, "global_step": 1139, "epoch": 31, "lr": 0.001} +{"train_loss": 0.0402044802904129, "train_loss_bc": 0.0402044802904129, "train_loss_llm": 0.0, "grad_norm": 0.039982326328754425, "global_step": 1140, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04990680143237114, "train_loss_bc": 0.04989362508058548, "train_loss_llm": 0.0013175904750823975, "grad_norm": 0.016807615756988525, "global_step": 1141, "epoch": 31, "lr": 0.001} +{"train_loss": 0.039167001843452454, "train_loss_bc": 0.039167001843452454, "train_loss_llm": 0.0, "grad_norm": 0.028519490733742714, "global_step": 1142, "epoch": 31, "lr": 0.001} +{"train_loss": 0.048510484397411346, "train_loss_bc": 0.04847387596964836, "train_loss_llm": 0.0036609768867492676, "grad_norm": 0.03490380570292473, "global_step": 1143, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04788116365671158, "train_loss_bc": 0.04788116365671158, "train_loss_llm": 0.0, "grad_norm": 0.03925751894712448, "global_step": 1144, "epoch": 31, "lr": 0.001} +{"train_loss": 0.05492602661252022, "train_loss_bc": 0.05490479618310928, "train_loss_llm": 0.0021231770515441895, "grad_norm": 0.015950189903378487, "global_step": 1145, "epoch": 31, "lr": 0.001} +{"train_loss": 0.05256621912121773, "train_loss_bc": 0.05253000557422638, "train_loss_llm": 0.0036212801933288574, "grad_norm": 0.02255089022219181, "global_step": 1146, "epoch": 31, "lr": 0.001} +{"train_loss": 0.044818487018346786, "train_loss_bc": 0.044818487018346786, "train_loss_llm": 0.0, "grad_norm": 0.026723159477114677, "global_step": 1147, "epoch": 31, "lr": 0.001} +{"train_loss": 0.038181595504283905, "train_loss_bc": 0.03817029297351837, "train_loss_llm": 0.0011304020881652832, "grad_norm": 0.025827128440141678, "global_step": 1148, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04366295784711838, "train_loss_bc": 0.04366295784711838, "train_loss_llm": 0.0, "grad_norm": 0.018041010946035385, "global_step": 1149, "epoch": 31, "lr": 0.001} +{"train_loss": 0.05315008759498596, "train_loss_bc": 0.05315008759498596, "train_loss_llm": 0.0, "grad_norm": 0.02385101467370987, "global_step": 1150, "epoch": 31, "lr": 0.001} +{"train_loss": 0.04774955504884323, "train_loss_bc": 0.04813343659043312, "train_loss_llm": 0.0, "grad_norm": 0.03300077095627785, "global_step": 1151, "epoch": 31, "lr": 0.001, "val_loss": 0.050094928592443466} +{"train_loss": 0.049431703984737396, "train_loss_bc": 0.049429792910814285, "train_loss_llm": 0.00019112229347229004, "grad_norm": 0.03577649965882301, "global_step": 1152, "epoch": 32, "lr": 0.001} +{"train_loss": 0.05033516883850098, "train_loss_bc": 0.05030552297830582, "train_loss_llm": 0.0029645562171936035, "grad_norm": 0.0176937784999609, "global_step": 1153, "epoch": 32, "lr": 0.001} +{"train_loss": 0.046446774154901505, "train_loss_bc": 0.04641328006982803, "train_loss_llm": 0.003349393606185913, "grad_norm": 0.02696712128818035, "global_step": 1154, "epoch": 32, "lr": 0.001} +{"train_loss": 0.043193381279706955, "train_loss_bc": 0.043193381279706955, "train_loss_llm": 0.0, "grad_norm": 0.032602012157440186, "global_step": 1155, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04715902358293533, "train_loss_bc": 0.04713401943445206, "train_loss_llm": 0.0025002360343933105, "grad_norm": 0.03700768202543259, "global_step": 1156, "epoch": 32, "lr": 0.001} +{"train_loss": 0.03499487414956093, "train_loss_bc": 0.03499487414956093, "train_loss_llm": 0.0, "grad_norm": 0.013008981011807919, "global_step": 1157, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04809483140707016, "train_loss_bc": 0.04809483140707016, "train_loss_llm": 0.0, "grad_norm": 0.023596089333295822, "global_step": 1158, "epoch": 32, "lr": 0.001} +{"train_loss": 0.043885115534067154, "train_loss_bc": 0.043885115534067154, "train_loss_llm": 0.0, "grad_norm": 0.027927247807383537, "global_step": 1159, "epoch": 32, "lr": 0.001} +{"train_loss": 0.05000750720500946, "train_loss_bc": 0.04999576136469841, "train_loss_llm": 0.0011745095252990723, "grad_norm": 0.03488176688551903, "global_step": 1160, "epoch": 32, "lr": 0.001} +{"train_loss": 0.03813423961400986, "train_loss_bc": 0.03813423961400986, "train_loss_llm": 0.0, "grad_norm": 0.01268808078020811, "global_step": 1161, "epoch": 32, "lr": 0.001} +{"train_loss": 0.05065229535102844, "train_loss_bc": 0.05064844712615013, "train_loss_llm": 0.00038486719131469727, "grad_norm": 0.017958326265215874, "global_step": 1162, "epoch": 32, "lr": 0.001} +{"train_loss": 0.040484946221113205, "train_loss_bc": 0.040484946221113205, "train_loss_llm": 0.0, "grad_norm": 0.02323162741959095, "global_step": 1163, "epoch": 32, "lr": 0.001} +{"train_loss": 0.06502055376768112, "train_loss_bc": 0.06500788778066635, "train_loss_llm": 0.00126686692237854, "grad_norm": 0.033145032823085785, "global_step": 1164, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04464774578809738, "train_loss_bc": 0.04461788386106491, "train_loss_llm": 0.0029861629009246826, "grad_norm": 0.015418127179145813, "global_step": 1165, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04339118301868439, "train_loss_bc": 0.04339118301868439, "train_loss_llm": 0.0, "grad_norm": 0.022986285388469696, "global_step": 1166, "epoch": 32, "lr": 0.001} +{"train_loss": 0.044610362499952316, "train_loss_bc": 0.04461013525724411, "train_loss_llm": 2.282857894897461e-05, "grad_norm": 0.0341670997440815, "global_step": 1167, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04816560819745064, "train_loss_bc": 0.04816560819745064, "train_loss_llm": 0.0, "grad_norm": 0.039861198514699936, "global_step": 1168, "epoch": 32, "lr": 0.001} +{"train_loss": 0.05188387259840965, "train_loss_bc": 0.05188387259840965, "train_loss_llm": 0.0, "grad_norm": 0.015515961684286594, "global_step": 1169, "epoch": 32, "lr": 0.001} +{"train_loss": 0.0535384900867939, "train_loss_bc": 0.05350589379668236, "train_loss_llm": 0.0032597482204437256, "grad_norm": 0.02009403333067894, "global_step": 1170, "epoch": 32, "lr": 0.001} +{"train_loss": 0.03958161547780037, "train_loss_bc": 0.03956065699458122, "train_loss_llm": 0.0020958781242370605, "grad_norm": 0.025076519697904587, "global_step": 1171, "epoch": 32, "lr": 0.001} +{"train_loss": 0.049659762531518936, "train_loss_bc": 0.04961560666561127, "train_loss_llm": 0.0044155120849609375, "grad_norm": 0.030390160158276558, "global_step": 1172, "epoch": 32, "lr": 0.001} +{"train_loss": 0.049206215888261795, "train_loss_bc": 0.049206215888261795, "train_loss_llm": 0.0, "grad_norm": 0.015445873141288757, "global_step": 1173, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04531865194439888, "train_loss_bc": 0.04531865194439888, "train_loss_llm": 0.0, "grad_norm": 0.026779796928167343, "global_step": 1174, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04723430052399635, "train_loss_bc": 0.047213561832904816, "train_loss_llm": 0.002073943614959717, "grad_norm": 0.02825063467025757, "global_step": 1175, "epoch": 32, "lr": 0.001} +{"train_loss": 0.045495305210351944, "train_loss_bc": 0.04548049345612526, "train_loss_llm": 0.001481175422668457, "grad_norm": 0.03379213809967041, "global_step": 1176, "epoch": 32, "lr": 0.001} +{"train_loss": 0.054683130234479904, "train_loss_bc": 0.054670970886945724, "train_loss_llm": 0.0012160837650299072, "grad_norm": 0.020285500213503838, "global_step": 1177, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04848017171025276, "train_loss_bc": 0.04847469553351402, "train_loss_llm": 0.0005477070808410645, "grad_norm": 0.0279476810246706, "global_step": 1178, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04272771254181862, "train_loss_bc": 0.0427028052508831, "train_loss_llm": 0.0024908483028411865, "grad_norm": 0.032979026436805725, "global_step": 1179, "epoch": 32, "lr": 0.001} +{"train_loss": 0.0495537705719471, "train_loss_bc": 0.0495537705719471, "train_loss_llm": 0.0, "grad_norm": 0.041411109268665314, "global_step": 1180, "epoch": 32, "lr": 0.001} +{"train_loss": 0.057791538536548615, "train_loss_bc": 0.057791538536548615, "train_loss_llm": 0.0, "grad_norm": 0.01700224168598652, "global_step": 1181, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04977649077773094, "train_loss_bc": 0.04977649077773094, "train_loss_llm": 0.0, "grad_norm": 0.02445058897137642, "global_step": 1182, "epoch": 32, "lr": 0.001} +{"train_loss": 0.05157800763845444, "train_loss_bc": 0.05157800763845444, "train_loss_llm": 0.0, "grad_norm": 0.030077217146754265, "global_step": 1183, "epoch": 32, "lr": 0.001} +{"train_loss": 0.0437077097594738, "train_loss_bc": 0.04369911551475525, "train_loss_llm": 0.0008594095706939697, "grad_norm": 0.031966932117938995, "global_step": 1184, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04902270808815956, "train_loss_bc": 0.048990990966558456, "train_loss_llm": 0.003171563148498535, "grad_norm": 0.02198147587478161, "global_step": 1185, "epoch": 32, "lr": 0.001} +{"train_loss": 0.045006413012742996, "train_loss_bc": 0.045006413012742996, "train_loss_llm": 0.0, "grad_norm": 0.028499633073806763, "global_step": 1186, "epoch": 32, "lr": 0.001} +{"train_loss": 0.04765849819199906, "train_loss_bc": 0.05280456319451332, "train_loss_llm": 1.9162893295288086e-05, "grad_norm": 0.03831426799297333, "global_step": 1187, "epoch": 32, "lr": 0.001, "val_loss": 0.05070924758911133} +{"train_loss": 0.049435511231422424, "train_loss_bc": 0.049435511231422424, "train_loss_llm": 0.0, "grad_norm": 0.04656810313463211, "global_step": 1188, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04723481833934784, "train_loss_bc": 0.04723481833934784, "train_loss_llm": 0.0, "grad_norm": 0.01629256270825863, "global_step": 1189, "epoch": 33, "lr": 0.001} +{"train_loss": 0.054398588836193085, "train_loss_bc": 0.054398588836193085, "train_loss_llm": 0.0, "grad_norm": 0.02881050482392311, "global_step": 1190, "epoch": 33, "lr": 0.001} +{"train_loss": 0.055250514298677444, "train_loss_bc": 0.055250514298677444, "train_loss_llm": 0.0, "grad_norm": 0.03360729664564133, "global_step": 1191, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04673459380865097, "train_loss_bc": 0.04673437401652336, "train_loss_llm": 2.199411392211914e-05, "grad_norm": 0.046904340386390686, "global_step": 1192, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04889169707894325, "train_loss_bc": 0.04889169707894325, "train_loss_llm": 0.0, "grad_norm": 0.019200829789042473, "global_step": 1193, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04938744381070137, "train_loss_bc": 0.04937734827399254, "train_loss_llm": 0.0010096430778503418, "grad_norm": 0.026095708832144737, "global_step": 1194, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04990125447511673, "train_loss_bc": 0.04989038407802582, "train_loss_llm": 0.0010870695114135742, "grad_norm": 0.028156736865639687, "global_step": 1195, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04960701987147331, "train_loss_bc": 0.04960701987147331, "train_loss_llm": 0.0, "grad_norm": 0.02740592323243618, "global_step": 1196, "epoch": 33, "lr": 0.001} +{"train_loss": 0.05401245504617691, "train_loss_bc": 0.05399079620838165, "train_loss_llm": 0.0021659433841705322, "grad_norm": 0.01677640900015831, "global_step": 1197, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04711722955107689, "train_loss_bc": 0.04711722955107689, "train_loss_llm": 0.0, "grad_norm": 0.026959136128425598, "global_step": 1198, "epoch": 33, "lr": 0.001} +{"train_loss": 0.03903016820549965, "train_loss_bc": 0.03899907320737839, "train_loss_llm": 0.003109544515609741, "grad_norm": 0.03301740065217018, "global_step": 1199, "epoch": 33, "lr": 0.001} +{"train_loss": 0.043839845806360245, "train_loss_bc": 0.043839845806360245, "train_loss_llm": 0.0, "grad_norm": 0.03746049106121063, "global_step": 1200, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04382215440273285, "train_loss_bc": 0.04382215440273285, "train_loss_llm": 0.0, "grad_norm": 0.013424348086118698, "global_step": 1201, "epoch": 33, "lr": 0.001} +{"train_loss": 0.05293184518814087, "train_loss_bc": 0.05293184518814087, "train_loss_llm": 0.0, "grad_norm": 0.023237841203808784, "global_step": 1202, "epoch": 33, "lr": 0.001} +{"train_loss": 0.047574546188116074, "train_loss_bc": 0.047574546188116074, "train_loss_llm": 0.0, "grad_norm": 0.02953289821743965, "global_step": 1203, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04336278513073921, "train_loss_bc": 0.04336278513073921, "train_loss_llm": 0.0, "grad_norm": 0.03697141259908676, "global_step": 1204, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04626457020640373, "train_loss_bc": 0.04626457020640373, "train_loss_llm": 0.0, "grad_norm": 0.013468155637383461, "global_step": 1205, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04258732870221138, "train_loss_bc": 0.04258732870221138, "train_loss_llm": 0.0, "grad_norm": 0.023089108988642693, "global_step": 1206, "epoch": 33, "lr": 0.001} +{"train_loss": 0.049983277916908264, "train_loss_bc": 0.049983277916908264, "train_loss_llm": 0.0, "grad_norm": 0.026769418269395828, "global_step": 1207, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04601169377565384, "train_loss_bc": 0.0459686815738678, "train_loss_llm": 0.004301279783248901, "grad_norm": 0.03008059225976467, "global_step": 1208, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04764741286635399, "train_loss_bc": 0.04761124029755592, "train_loss_llm": 0.0036171674728393555, "grad_norm": 0.01827159896492958, "global_step": 1209, "epoch": 33, "lr": 0.001} +{"train_loss": 0.043256133794784546, "train_loss_bc": 0.04324126988649368, "train_loss_llm": 0.001486271619796753, "grad_norm": 0.02315588667988777, "global_step": 1210, "epoch": 33, "lr": 0.001} +{"train_loss": 0.049545835703611374, "train_loss_bc": 0.0495421439409256, "train_loss_llm": 0.0003693103790283203, "grad_norm": 0.027542755007743835, "global_step": 1211, "epoch": 33, "lr": 0.001} +{"train_loss": 0.054644159972667694, "train_loss_bc": 0.05456089973449707, "train_loss_llm": 0.008325934410095215, "grad_norm": 0.034849636256694794, "global_step": 1212, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04213346168398857, "train_loss_bc": 0.04212257266044617, "train_loss_llm": 0.0010887384414672852, "grad_norm": 0.015876280143857002, "global_step": 1213, "epoch": 33, "lr": 0.001} +{"train_loss": 0.052388280630111694, "train_loss_bc": 0.052388280630111694, "train_loss_llm": 0.0, "grad_norm": 0.02567141316831112, "global_step": 1214, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04732535034418106, "train_loss_bc": 0.04732535034418106, "train_loss_llm": 0.0, "grad_norm": 0.03244715929031372, "global_step": 1215, "epoch": 33, "lr": 0.001} +{"train_loss": 0.042477820068597794, "train_loss_bc": 0.04245671629905701, "train_loss_llm": 0.0021102726459503174, "grad_norm": 0.03925861045718193, "global_step": 1216, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04235663637518883, "train_loss_bc": 0.04235663637518883, "train_loss_llm": 0.0, "grad_norm": 0.018883325159549713, "global_step": 1217, "epoch": 33, "lr": 0.001} +{"train_loss": 0.05504509061574936, "train_loss_bc": 0.05504509061574936, "train_loss_llm": 0.0, "grad_norm": 0.029127521440386772, "global_step": 1218, "epoch": 33, "lr": 0.001} +{"train_loss": 0.042602285742759705, "train_loss_bc": 0.042602285742759705, "train_loss_llm": 0.0, "grad_norm": 0.03683101013302803, "global_step": 1219, "epoch": 33, "lr": 0.001} +{"train_loss": 0.049356892704963684, "train_loss_bc": 0.049356892704963684, "train_loss_llm": 0.0, "grad_norm": 0.040236663073301315, "global_step": 1220, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04633158817887306, "train_loss_bc": 0.04628420248627663, "train_loss_llm": 0.004738450050354004, "grad_norm": 0.018066680058836937, "global_step": 1221, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04110569506883621, "train_loss_bc": 0.04110569506883621, "train_loss_llm": 0.0, "grad_norm": 0.02436964213848114, "global_step": 1222, "epoch": 33, "lr": 0.001} +{"train_loss": 0.04758693960805734, "train_loss_bc": 0.04953384026885033, "train_loss_llm": 0.0, "grad_norm": 0.02707325853407383, "global_step": 1223, "epoch": 33, "lr": 0.001, "val_loss": 0.05034424364566803} +{"train_loss": 0.04607149958610535, "train_loss_bc": 0.046031560748815536, "train_loss_llm": 0.003993958234786987, "grad_norm": 0.03152443841099739, "global_step": 1224, "epoch": 34, "lr": 0.001} +{"train_loss": 0.05167057365179062, "train_loss_bc": 0.051617544144392014, "train_loss_llm": 0.005303084850311279, "grad_norm": 0.02117404341697693, "global_step": 1225, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04718542471528053, "train_loss_bc": 0.04718542471528053, "train_loss_llm": 0.0, "grad_norm": 0.022952815517783165, "global_step": 1226, "epoch": 34, "lr": 0.001} +{"train_loss": 0.03759662061929703, "train_loss_bc": 0.037589091807603836, "train_loss_llm": 0.0007530152797698975, "grad_norm": 0.028755871579051018, "global_step": 1227, "epoch": 34, "lr": 0.001} +{"train_loss": 0.05079271271824837, "train_loss_bc": 0.05079271271824837, "train_loss_llm": 0.0, "grad_norm": 0.036275625228881836, "global_step": 1228, "epoch": 34, "lr": 0.001} +{"train_loss": 0.044875696301460266, "train_loss_bc": 0.044858817011117935, "train_loss_llm": 0.0016878247261047363, "grad_norm": 0.018974632024765015, "global_step": 1229, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04668530821800232, "train_loss_bc": 0.04668530821800232, "train_loss_llm": 0.0, "grad_norm": 0.02592528983950615, "global_step": 1230, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04008181020617485, "train_loss_bc": 0.04008181020617485, "train_loss_llm": 0.0, "grad_norm": 0.03235633671283722, "global_step": 1231, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04255380481481552, "train_loss_bc": 0.042550064623355865, "train_loss_llm": 0.00037404894828796387, "grad_norm": 0.04615195095539093, "global_step": 1232, "epoch": 34, "lr": 0.001} +{"train_loss": 0.05178588256239891, "train_loss_bc": 0.05177411437034607, "train_loss_llm": 0.0011766552925109863, "grad_norm": 0.01688198931515217, "global_step": 1233, "epoch": 34, "lr": 0.001} +{"train_loss": 0.05261602997779846, "train_loss_bc": 0.052604515105485916, "train_loss_llm": 0.0011513233184814453, "grad_norm": 0.028134480118751526, "global_step": 1234, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04900102689862251, "train_loss_bc": 0.04900102689862251, "train_loss_llm": 0.0, "grad_norm": 0.031120682135224342, "global_step": 1235, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04755549132823944, "train_loss_bc": 0.04755549132823944, "train_loss_llm": 0.0, "grad_norm": 0.035662706941366196, "global_step": 1236, "epoch": 34, "lr": 0.001} +{"train_loss": 0.040255047380924225, "train_loss_bc": 0.04023228585720062, "train_loss_llm": 0.002276182174682617, "grad_norm": 0.0157964788377285, "global_step": 1237, "epoch": 34, "lr": 0.001} +{"train_loss": 0.05043601617217064, "train_loss_bc": 0.05043601617217064, "train_loss_llm": 0.0, "grad_norm": 0.020787978544831276, "global_step": 1238, "epoch": 34, "lr": 0.001} +{"train_loss": 0.044513899832963943, "train_loss_bc": 0.044498588889837265, "train_loss_llm": 0.0015310347080230713, "grad_norm": 0.02690398506820202, "global_step": 1239, "epoch": 34, "lr": 0.001} +{"train_loss": 0.044901493936777115, "train_loss_bc": 0.044901493936777115, "train_loss_llm": 0.0, "grad_norm": 0.03587207943201065, "global_step": 1240, "epoch": 34, "lr": 0.001} +{"train_loss": 0.05062795802950859, "train_loss_bc": 0.050611209124326706, "train_loss_llm": 0.001674741506576538, "grad_norm": 0.015417348593473434, "global_step": 1241, "epoch": 34, "lr": 0.001} +{"train_loss": 0.03986101225018501, "train_loss_bc": 0.03983865678310394, "train_loss_llm": 0.002235502004623413, "grad_norm": 0.020327217876911163, "global_step": 1242, "epoch": 34, "lr": 0.001} +{"train_loss": 0.041111674159765244, "train_loss_bc": 0.041111674159765244, "train_loss_llm": 0.0, "grad_norm": 0.028170278295874596, "global_step": 1243, "epoch": 34, "lr": 0.001} +{"train_loss": 0.05086616054177284, "train_loss_bc": 0.05086616054177284, "train_loss_llm": 0.0, "grad_norm": 0.032983869314193726, "global_step": 1244, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04256107658147812, "train_loss_bc": 0.04256107658147812, "train_loss_llm": 0.0, "grad_norm": 0.015526161529123783, "global_step": 1245, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04022059589624405, "train_loss_bc": 0.04022059589624405, "train_loss_llm": 0.0, "grad_norm": 0.02646041475236416, "global_step": 1246, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04630590230226517, "train_loss_bc": 0.04630590230226517, "train_loss_llm": 0.0, "grad_norm": 0.029543794691562653, "global_step": 1247, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04782918840646744, "train_loss_bc": 0.0478060282766819, "train_loss_llm": 0.002316117286682129, "grad_norm": 0.03271353244781494, "global_step": 1248, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04913654923439026, "train_loss_bc": 0.04911480471491814, "train_loss_llm": 0.0021743178367614746, "grad_norm": 0.013537431135773659, "global_step": 1249, "epoch": 34, "lr": 0.001} +{"train_loss": 0.05403504520654678, "train_loss_bc": 0.05403504520654678, "train_loss_llm": 0.0, "grad_norm": 0.024883585050702095, "global_step": 1250, "epoch": 34, "lr": 0.001} +{"train_loss": 0.05219041928648949, "train_loss_bc": 0.05219041928648949, "train_loss_llm": 0.0, "grad_norm": 0.033341195434331894, "global_step": 1251, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04678308591246605, "train_loss_bc": 0.046782977879047394, "train_loss_llm": 1.0848045349121094e-05, "grad_norm": 0.04206569865345955, "global_step": 1252, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04453403502702713, "train_loss_bc": 0.04453403502702713, "train_loss_llm": 0.0, "grad_norm": 0.016762923449277878, "global_step": 1253, "epoch": 34, "lr": 0.001} +{"train_loss": 0.0513390451669693, "train_loss_bc": 0.05132989585399628, "train_loss_llm": 0.0009147822856903076, "grad_norm": 0.02953050285577774, "global_step": 1254, "epoch": 34, "lr": 0.001} +{"train_loss": 0.054184865206480026, "train_loss_bc": 0.054184865206480026, "train_loss_llm": 0.0, "grad_norm": 0.03638848662376404, "global_step": 1255, "epoch": 34, "lr": 0.001} +{"train_loss": 0.0523054338991642, "train_loss_bc": 0.052297815680503845, "train_loss_llm": 0.0007616579532623291, "grad_norm": 0.04642517492175102, "global_step": 1256, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04645181819796562, "train_loss_bc": 0.04645181819796562, "train_loss_llm": 0.0, "grad_norm": 0.015534145757555962, "global_step": 1257, "epoch": 34, "lr": 0.001} +{"train_loss": 0.05509882792830467, "train_loss_bc": 0.05509882792830467, "train_loss_llm": 0.0, "grad_norm": 0.0221568550914526, "global_step": 1258, "epoch": 34, "lr": 0.001} +{"train_loss": 0.04757303134020832, "train_loss_bc": 0.058579567819833755, "train_loss_llm": 0.0028527677059173584, "grad_norm": 0.03543728590011597, "global_step": 1259, "epoch": 34, "lr": 0.001, "val_loss": 0.05037667602300644} +{"train_loss": 0.043742384761571884, "train_loss_bc": 0.04367896169424057, "train_loss_llm": 0.006342202425003052, "grad_norm": 0.049451928585767746, "global_step": 1260, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04391308128833771, "train_loss_bc": 0.04391308128833771, "train_loss_llm": 0.0, "grad_norm": 0.014343330636620522, "global_step": 1261, "epoch": 35, "lr": 0.001} +{"train_loss": 0.05120933800935745, "train_loss_bc": 0.05116173252463341, "train_loss_llm": 0.004760563373565674, "grad_norm": 0.02055826038122177, "global_step": 1262, "epoch": 35, "lr": 0.001} +{"train_loss": 0.0455450601875782, "train_loss_bc": 0.04553129896521568, "train_loss_llm": 0.0013760626316070557, "grad_norm": 0.025026148185133934, "global_step": 1263, "epoch": 35, "lr": 0.001} +{"train_loss": 0.049218062311410904, "train_loss_bc": 0.049218062311410904, "train_loss_llm": 0.0, "grad_norm": 0.03244687616825104, "global_step": 1264, "epoch": 35, "lr": 0.001} +{"train_loss": 0.050365909934043884, "train_loss_bc": 0.05036463961005211, "train_loss_llm": 0.00012695789337158203, "grad_norm": 0.016436008736491203, "global_step": 1265, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04471360146999359, "train_loss_bc": 0.04471360146999359, "train_loss_llm": 0.0, "grad_norm": 0.023391123861074448, "global_step": 1266, "epoch": 35, "lr": 0.001} +{"train_loss": 0.0427074059844017, "train_loss_bc": 0.0427074059844017, "train_loss_llm": 0.0, "grad_norm": 0.027523759752511978, "global_step": 1267, "epoch": 35, "lr": 0.001} +{"train_loss": 0.050700537860393524, "train_loss_bc": 0.050692327320575714, "train_loss_llm": 0.0008208751678466797, "grad_norm": 0.03390774130821228, "global_step": 1268, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04661014676094055, "train_loss_bc": 0.04661014676094055, "train_loss_llm": 0.0, "grad_norm": 0.016188079491257668, "global_step": 1269, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04461155831813812, "train_loss_bc": 0.04461155831813812, "train_loss_llm": 0.0, "grad_norm": 0.028591036796569824, "global_step": 1270, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04902986064553261, "train_loss_bc": 0.04902986064553261, "train_loss_llm": 0.0, "grad_norm": 0.033663466572761536, "global_step": 1271, "epoch": 35, "lr": 0.001} +{"train_loss": 0.048213791102170944, "train_loss_bc": 0.04817737266421318, "train_loss_llm": 0.0036419928073883057, "grad_norm": 0.04092693701386452, "global_step": 1272, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04970449209213257, "train_loss_bc": 0.04966046288609505, "train_loss_llm": 0.0044029951095581055, "grad_norm": 0.018493076786398888, "global_step": 1273, "epoch": 35, "lr": 0.001} +{"train_loss": 0.0458570159971714, "train_loss_bc": 0.0458570159971714, "train_loss_llm": 0.0, "grad_norm": 0.024586845189332962, "global_step": 1274, "epoch": 35, "lr": 0.001} +{"train_loss": 0.0545792393386364, "train_loss_bc": 0.0545792356133461, "train_loss_llm": 3.8743019104003906e-07, "grad_norm": 0.02968950942158699, "global_step": 1275, "epoch": 35, "lr": 0.001} +{"train_loss": 0.05101017281413078, "train_loss_bc": 0.05099603906273842, "train_loss_llm": 0.0014135539531707764, "grad_norm": 0.041729636490345, "global_step": 1276, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04406328499317169, "train_loss_bc": 0.04406328499317169, "train_loss_llm": 0.0, "grad_norm": 0.017021719366312027, "global_step": 1277, "epoch": 35, "lr": 0.001} +{"train_loss": 0.05208269879221916, "train_loss_bc": 0.05206278711557388, "train_loss_llm": 0.0019912123680114746, "grad_norm": 0.029873820021748543, "global_step": 1278, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04293225705623627, "train_loss_bc": 0.04293225705623627, "train_loss_llm": 0.0, "grad_norm": 0.04256017133593559, "global_step": 1279, "epoch": 35, "lr": 0.001} +{"train_loss": 0.046365536749362946, "train_loss_bc": 0.046365536749362946, "train_loss_llm": 0.0, "grad_norm": 0.04038463905453682, "global_step": 1280, "epoch": 35, "lr": 0.001} +{"train_loss": 0.0462646521627903, "train_loss_bc": 0.0462646521627903, "train_loss_llm": 0.0, "grad_norm": 0.020694047212600708, "global_step": 1281, "epoch": 35, "lr": 0.001} +{"train_loss": 0.05169288069009781, "train_loss_bc": 0.05166822671890259, "train_loss_llm": 0.002465277910232544, "grad_norm": 0.03144387900829315, "global_step": 1282, "epoch": 35, "lr": 0.001} +{"train_loss": 0.0477789044380188, "train_loss_bc": 0.04776603356003761, "train_loss_llm": 0.0012871026992797852, "grad_norm": 0.04634212702512741, "global_step": 1283, "epoch": 35, "lr": 0.001} +{"train_loss": 0.046151936054229736, "train_loss_bc": 0.046151936054229736, "train_loss_llm": 0.0, "grad_norm": 0.05259094014763832, "global_step": 1284, "epoch": 35, "lr": 0.001} +{"train_loss": 0.042351510375738144, "train_loss_bc": 0.042351510375738144, "train_loss_llm": 0.0, "grad_norm": 0.018075888976454735, "global_step": 1285, "epoch": 35, "lr": 0.001} +{"train_loss": 0.0413629487156868, "train_loss_bc": 0.04135442152619362, "train_loss_llm": 0.0008528530597686768, "grad_norm": 0.02038061060011387, "global_step": 1286, "epoch": 35, "lr": 0.001} +{"train_loss": 0.046181369572877884, "train_loss_bc": 0.046181369572877884, "train_loss_llm": 0.0, "grad_norm": 0.032150741666555405, "global_step": 1287, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04563850536942482, "train_loss_bc": 0.04559798166155815, "train_loss_llm": 0.0040524303913116455, "grad_norm": 0.04103843495249748, "global_step": 1288, "epoch": 35, "lr": 0.001} +{"train_loss": 0.05301029235124588, "train_loss_bc": 0.05301029235124588, "train_loss_llm": 0.0, "grad_norm": 0.018116610124707222, "global_step": 1289, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04423967003822327, "train_loss_bc": 0.044228047132492065, "train_loss_llm": 0.001162409782409668, "grad_norm": 0.024699216708540916, "global_step": 1290, "epoch": 35, "lr": 0.001} +{"train_loss": 0.0475001260638237, "train_loss_bc": 0.04749938100576401, "train_loss_llm": 7.444620132446289e-05, "grad_norm": 0.030278656631708145, "global_step": 1291, "epoch": 35, "lr": 0.001} +{"train_loss": 0.05346345901489258, "train_loss_bc": 0.05346345901489258, "train_loss_llm": 0.0, "grad_norm": 0.032095178961753845, "global_step": 1292, "epoch": 35, "lr": 0.001} +{"train_loss": 0.048645444214344025, "train_loss_bc": 0.048645444214344025, "train_loss_llm": 0.0, "grad_norm": 0.01581384614109993, "global_step": 1293, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04574903845787048, "train_loss_bc": 0.04574903845787048, "train_loss_llm": 0.0, "grad_norm": 0.020199157297611237, "global_step": 1294, "epoch": 35, "lr": 0.001} +{"train_loss": 0.04745121165696117, "train_loss_bc": 0.05103744566440582, "train_loss_llm": 0.0, "grad_norm": 0.03432321175932884, "global_step": 1295, "epoch": 35, "lr": 0.001, "train/cumulative_reward": 32.75222346113784, "train/mean_score": 10.52436221984428, "train/success_rate": 0.3, "test/cumulative_reward": 30.547554563165722, "test/mean_score": 1.5491390610198996, "test/success_rate": 0.0, "val_loss": 0.050608519464731216, "train_action_mse_error": 0.05103608965873718} +{"train_loss": 0.043580248951911926, "train_loss_bc": 0.043580248951911926, "train_loss_llm": 0.0, "grad_norm": 0.03935728222131729, "global_step": 1296, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04429435357451439, "train_loss_bc": 0.04427735134959221, "train_loss_llm": 0.0017001628875732422, "grad_norm": 0.01816481165587902, "global_step": 1297, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04643383249640465, "train_loss_bc": 0.04642988368868828, "train_loss_llm": 0.0003947317600250244, "grad_norm": 0.022485850378870964, "global_step": 1298, "epoch": 36, "lr": 0.001} +{"train_loss": 0.043822020292282104, "train_loss_bc": 0.04379923641681671, "train_loss_llm": 0.002278566360473633, "grad_norm": 0.027865035459399223, "global_step": 1299, "epoch": 36, "lr": 0.001} +{"train_loss": 0.05315715819597244, "train_loss_bc": 0.05315715819597244, "train_loss_llm": 0.0, "grad_norm": 0.039928875863552094, "global_step": 1300, "epoch": 36, "lr": 0.001} +{"train_loss": 0.0465884804725647, "train_loss_bc": 0.0465884804725647, "train_loss_llm": 0.0, "grad_norm": 0.019024791195988655, "global_step": 1301, "epoch": 36, "lr": 0.001} +{"train_loss": 0.05317152664065361, "train_loss_bc": 0.05317152664065361, "train_loss_llm": 0.0, "grad_norm": 0.03506194427609444, "global_step": 1302, "epoch": 36, "lr": 0.001} +{"train_loss": 0.049970027059316635, "train_loss_bc": 0.049970027059316635, "train_loss_llm": 0.0, "grad_norm": 0.04533182457089424, "global_step": 1303, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04396934062242508, "train_loss_bc": 0.04396934062242508, "train_loss_llm": 0.0, "grad_norm": 0.049774717539548874, "global_step": 1304, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04568171128630638, "train_loss_bc": 0.04568171128630638, "train_loss_llm": 0.0, "grad_norm": 0.017179058864712715, "global_step": 1305, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04369146376848221, "train_loss_bc": 0.04369146376848221, "train_loss_llm": 0.0, "grad_norm": 0.02369825541973114, "global_step": 1306, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04513164982199669, "train_loss_bc": 0.045127734541893005, "train_loss_llm": 0.0003915727138519287, "grad_norm": 0.02796451933681965, "global_step": 1307, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04769056662917137, "train_loss_bc": 0.047630783170461655, "train_loss_llm": 0.0059781670570373535, "grad_norm": 0.03382708877325058, "global_step": 1308, "epoch": 36, "lr": 0.001} +{"train_loss": 0.047500863671302795, "train_loss_bc": 0.047500863671302795, "train_loss_llm": 0.0, "grad_norm": 0.019296616315841675, "global_step": 1309, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04190481826663017, "train_loss_bc": 0.04186781495809555, "train_loss_llm": 0.00370025634765625, "grad_norm": 0.020313147455453873, "global_step": 1310, "epoch": 36, "lr": 0.001} +{"train_loss": 0.03593726456165314, "train_loss_bc": 0.03593726456165314, "train_loss_llm": 0.0, "grad_norm": 0.024454953148961067, "global_step": 1311, "epoch": 36, "lr": 0.001} +{"train_loss": 0.047391876578330994, "train_loss_bc": 0.047391876578330994, "train_loss_llm": 0.0, "grad_norm": 0.025836773216724396, "global_step": 1312, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04736999794840813, "train_loss_bc": 0.04736999794840813, "train_loss_llm": 0.0, "grad_norm": 0.016760770231485367, "global_step": 1313, "epoch": 36, "lr": 0.001} +{"train_loss": 0.05008057504892349, "train_loss_bc": 0.050050873309373856, "train_loss_llm": 0.0029700100421905518, "grad_norm": 0.029303662478923798, "global_step": 1314, "epoch": 36, "lr": 0.001} +{"train_loss": 0.03540069982409477, "train_loss_bc": 0.03539399430155754, "train_loss_llm": 0.0006703734397888184, "grad_norm": 0.029268376529216766, "global_step": 1315, "epoch": 36, "lr": 0.001} +{"train_loss": 0.05132593959569931, "train_loss_bc": 0.051296744495630264, "train_loss_llm": 0.002919524908065796, "grad_norm": 0.03571697324514389, "global_step": 1316, "epoch": 36, "lr": 0.001} +{"train_loss": 0.046282604336738586, "train_loss_bc": 0.046282604336738586, "train_loss_llm": 0.0, "grad_norm": 0.01773665100336075, "global_step": 1317, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04999876394867897, "train_loss_bc": 0.04999876394867897, "train_loss_llm": 0.0, "grad_norm": 0.024771200492978096, "global_step": 1318, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04818344488739967, "train_loss_bc": 0.04818344488739967, "train_loss_llm": 0.0, "grad_norm": 0.02963789738714695, "global_step": 1319, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04158679395914078, "train_loss_bc": 0.041548095643520355, "train_loss_llm": 0.00386965274810791, "grad_norm": 0.0381474606692791, "global_step": 1320, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04446921497583389, "train_loss_bc": 0.04446921497583389, "train_loss_llm": 0.0, "grad_norm": 0.013180010952055454, "global_step": 1321, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04325983300805092, "train_loss_bc": 0.043258924037218094, "train_loss_llm": 9.09566879272461e-05, "grad_norm": 0.02111760899424553, "global_step": 1322, "epoch": 36, "lr": 0.001} +{"train_loss": 0.051370248198509216, "train_loss_bc": 0.051370248198509216, "train_loss_llm": 0.0, "grad_norm": 0.032988958060741425, "global_step": 1323, "epoch": 36, "lr": 0.001} +{"train_loss": 0.051804594695568085, "train_loss_bc": 0.051794491708278656, "train_loss_llm": 0.001010119915008545, "grad_norm": 0.033708442002534866, "global_step": 1324, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04770497605204582, "train_loss_bc": 0.04763496667146683, "train_loss_llm": 0.007000952959060669, "grad_norm": 0.0138611551374197, "global_step": 1325, "epoch": 36, "lr": 0.001} +{"train_loss": 0.05306711792945862, "train_loss_bc": 0.05306164547801018, "train_loss_llm": 0.0005471110343933105, "grad_norm": 0.02331458032131195, "global_step": 1326, "epoch": 36, "lr": 0.001} +{"train_loss": 0.05682430416345596, "train_loss_bc": 0.05682430416345596, "train_loss_llm": 0.0, "grad_norm": 0.030236000195145607, "global_step": 1327, "epoch": 36, "lr": 0.001} +{"train_loss": 0.05125775560736656, "train_loss_bc": 0.05125775560736656, "train_loss_llm": 0.0, "grad_norm": 0.03601735085248947, "global_step": 1328, "epoch": 36, "lr": 0.001} +{"train_loss": 0.0558057464659214, "train_loss_bc": 0.0558057464659214, "train_loss_llm": 0.0, "grad_norm": 0.01972905360162258, "global_step": 1329, "epoch": 36, "lr": 0.001} +{"train_loss": 0.04747607558965683, "train_loss_bc": 0.04746929556131363, "train_loss_llm": 0.0006781518459320068, "grad_norm": 0.02525908127427101, "global_step": 1330, "epoch": 36, "lr": 0.001} +{"train_loss": 0.0472303226383196, "train_loss_bc": 0.04710572585463524, "train_loss_llm": 0.0, "grad_norm": 0.031388163566589355, "global_step": 1331, "epoch": 36, "lr": 0.001, "val_loss": 0.0502866730093956} +{"train_loss": 0.04521038755774498, "train_loss_bc": 0.04519153758883476, "train_loss_llm": 0.001885056495666504, "grad_norm": 0.03674406558275223, "global_step": 1332, "epoch": 37, "lr": 0.001} +{"train_loss": 0.03931448981165886, "train_loss_bc": 0.03930936008691788, "train_loss_llm": 0.0005129873752593994, "grad_norm": 0.015572626143693924, "global_step": 1333, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04331443831324577, "train_loss_bc": 0.0432700514793396, "train_loss_llm": 0.004438519477844238, "grad_norm": 0.02697146125137806, "global_step": 1334, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04085996374487877, "train_loss_bc": 0.04085996374487877, "train_loss_llm": 0.0, "grad_norm": 0.03187624737620354, "global_step": 1335, "epoch": 37, "lr": 0.001} +{"train_loss": 0.048428017646074295, "train_loss_bc": 0.048428017646074295, "train_loss_llm": 0.0, "grad_norm": 0.038712382316589355, "global_step": 1336, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04071462154388428, "train_loss_bc": 0.040652014315128326, "train_loss_llm": 0.006260901689529419, "grad_norm": 0.019397098571062088, "global_step": 1337, "epoch": 37, "lr": 0.001} +{"train_loss": 0.0492217093706131, "train_loss_bc": 0.049197059124708176, "train_loss_llm": 0.002465188503265381, "grad_norm": 0.029260460287332535, "global_step": 1338, "epoch": 37, "lr": 0.001} +{"train_loss": 0.05819283798336983, "train_loss_bc": 0.05819283798336983, "train_loss_llm": 0.0, "grad_norm": 0.03699789568781853, "global_step": 1339, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04757743701338768, "train_loss_bc": 0.04752960428595543, "train_loss_llm": 0.00478324294090271, "grad_norm": 0.03609336540102959, "global_step": 1340, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04663553833961487, "train_loss_bc": 0.04663551598787308, "train_loss_llm": 2.3543834686279297e-06, "grad_norm": 0.01905973255634308, "global_step": 1341, "epoch": 37, "lr": 0.001} +{"train_loss": 0.043317750096321106, "train_loss_bc": 0.043317750096321106, "train_loss_llm": 0.0, "grad_norm": 0.022064704447984695, "global_step": 1342, "epoch": 37, "lr": 0.001} +{"train_loss": 0.0527915395796299, "train_loss_bc": 0.05278491973876953, "train_loss_llm": 0.0006618797779083252, "grad_norm": 0.02929440140724182, "global_step": 1343, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04657953977584839, "train_loss_bc": 0.04657953977584839, "train_loss_llm": 0.0, "grad_norm": 0.03571082279086113, "global_step": 1344, "epoch": 37, "lr": 0.001} +{"train_loss": 0.0350298248231411, "train_loss_bc": 0.035023145377635956, "train_loss_llm": 0.0006680190563201904, "grad_norm": 0.02069534733891487, "global_step": 1345, "epoch": 37, "lr": 0.001} +{"train_loss": 0.052244093269109726, "train_loss_bc": 0.052244093269109726, "train_loss_llm": 0.0, "grad_norm": 0.029498262330889702, "global_step": 1346, "epoch": 37, "lr": 0.001} +{"train_loss": 0.05674360319972038, "train_loss_bc": 0.05674360319972038, "train_loss_llm": 0.0, "grad_norm": 0.04215968772768974, "global_step": 1347, "epoch": 37, "lr": 0.001} +{"train_loss": 0.049297407269477844, "train_loss_bc": 0.049297407269477844, "train_loss_llm": 0.0, "grad_norm": 0.05108087509870529, "global_step": 1348, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04420971870422363, "train_loss_bc": 0.04420945793390274, "train_loss_llm": 2.6166439056396484e-05, "grad_norm": 0.017365654930472374, "global_step": 1349, "epoch": 37, "lr": 0.001} +{"train_loss": 0.042743634432554245, "train_loss_bc": 0.04273061081767082, "train_loss_llm": 0.0013023018836975098, "grad_norm": 0.022762656211853027, "global_step": 1350, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04445384442806244, "train_loss_bc": 0.04445384442806244, "train_loss_llm": 0.0, "grad_norm": 0.030264027416706085, "global_step": 1351, "epoch": 37, "lr": 0.001} +{"train_loss": 0.05532702058553696, "train_loss_bc": 0.05530083552002907, "train_loss_llm": 0.002618342638015747, "grad_norm": 0.04092087596654892, "global_step": 1352, "epoch": 37, "lr": 0.001} +{"train_loss": 0.05215838924050331, "train_loss_bc": 0.05212916433811188, "train_loss_llm": 0.002922624349594116, "grad_norm": 0.01800466887652874, "global_step": 1353, "epoch": 37, "lr": 0.001} +{"train_loss": 0.05371667444705963, "train_loss_bc": 0.053705837577581406, "train_loss_llm": 0.0010838210582733154, "grad_norm": 0.024739844724535942, "global_step": 1354, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04376566782593727, "train_loss_bc": 0.04376354068517685, "train_loss_llm": 0.00021269917488098145, "grad_norm": 0.03343644365668297, "global_step": 1355, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04127930477261543, "train_loss_bc": 0.04127930477261543, "train_loss_llm": 0.0, "grad_norm": 0.03623735532164574, "global_step": 1356, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04175113886594772, "train_loss_bc": 0.04173082858324051, "train_loss_llm": 0.002030998468399048, "grad_norm": 0.017628654837608337, "global_step": 1357, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04384055361151695, "train_loss_bc": 0.04384055361151695, "train_loss_llm": 0.0, "grad_norm": 0.02440751902759075, "global_step": 1358, "epoch": 37, "lr": 0.001} +{"train_loss": 0.050613660365343094, "train_loss_bc": 0.050613660365343094, "train_loss_llm": 0.0, "grad_norm": 0.027872920036315918, "global_step": 1359, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04563511535525322, "train_loss_bc": 0.045629505068063736, "train_loss_llm": 0.0005610883235931396, "grad_norm": 0.03838987648487091, "global_step": 1360, "epoch": 37, "lr": 0.001} +{"train_loss": 0.03841589391231537, "train_loss_bc": 0.03841589391231537, "train_loss_llm": 0.0, "grad_norm": 0.017885722219944, "global_step": 1361, "epoch": 37, "lr": 0.001} +{"train_loss": 0.05008480325341225, "train_loss_bc": 0.050077348947525024, "train_loss_llm": 0.0007453858852386475, "grad_norm": 0.026985254138708115, "global_step": 1362, "epoch": 37, "lr": 0.001} +{"train_loss": 0.05253348499536514, "train_loss_bc": 0.05253348499536514, "train_loss_llm": 0.0, "grad_norm": 0.0376121960580349, "global_step": 1363, "epoch": 37, "lr": 0.001} +{"train_loss": 0.051123470067977905, "train_loss_bc": 0.051123470067977905, "train_loss_llm": 0.0, "grad_norm": 0.04537016525864601, "global_step": 1364, "epoch": 37, "lr": 0.001} +{"train_loss": 0.050918098539114, "train_loss_bc": 0.050918098539114, "train_loss_llm": 0.0, "grad_norm": 0.018764996901154518, "global_step": 1365, "epoch": 37, "lr": 0.001} +{"train_loss": 0.05347384512424469, "train_loss_bc": 0.05346658453345299, "train_loss_llm": 0.000726163387298584, "grad_norm": 0.03200354799628258, "global_step": 1366, "epoch": 37, "lr": 0.001} +{"train_loss": 0.04715001303702593, "train_loss_bc": 0.04588068649172783, "train_loss_llm": 0.0002263486385345459, "grad_norm": 0.041230421513319016, "global_step": 1367, "epoch": 37, "lr": 0.001, "val_loss": 0.0502697229385376} +{"train_loss": 0.04244047403335571, "train_loss_bc": 0.04244047403335571, "train_loss_llm": 0.0, "grad_norm": 0.04919158294796944, "global_step": 1368, "epoch": 38, "lr": 0.001} +{"train_loss": 0.041091568768024445, "train_loss_bc": 0.041091568768024445, "train_loss_llm": 0.0, "grad_norm": 0.01992052234709263, "global_step": 1369, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04854288324713707, "train_loss_bc": 0.0485273152589798, "train_loss_llm": 0.0015568137168884277, "grad_norm": 0.025245122611522675, "global_step": 1370, "epoch": 38, "lr": 0.001} +{"train_loss": 0.05012749135494232, "train_loss_bc": 0.05012749135494232, "train_loss_llm": 0.0, "grad_norm": 0.03480498865246773, "global_step": 1371, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04486217349767685, "train_loss_bc": 0.04486217349767685, "train_loss_llm": 0.0, "grad_norm": 0.04204843193292618, "global_step": 1372, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04724344611167908, "train_loss_bc": 0.04724344611167908, "train_loss_llm": 0.0, "grad_norm": 0.015184415504336357, "global_step": 1373, "epoch": 38, "lr": 0.001} +{"train_loss": 0.03658624738454819, "train_loss_bc": 0.03656216710805893, "train_loss_llm": 0.0024079978466033936, "grad_norm": 0.02289380133152008, "global_step": 1374, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04207007959485054, "train_loss_bc": 0.04205391928553581, "train_loss_llm": 0.0016161203384399414, "grad_norm": 0.027273064479231834, "global_step": 1375, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04942192882299423, "train_loss_bc": 0.04942192882299423, "train_loss_llm": 0.0, "grad_norm": 0.03536273166537285, "global_step": 1376, "epoch": 38, "lr": 0.001} +{"train_loss": 0.05016652122139931, "train_loss_bc": 0.05016652122139931, "train_loss_llm": 0.0, "grad_norm": 0.01844005659222603, "global_step": 1377, "epoch": 38, "lr": 0.001} +{"train_loss": 0.05057067051529884, "train_loss_bc": 0.05057067051529884, "train_loss_llm": 0.0, "grad_norm": 0.02612774446606636, "global_step": 1378, "epoch": 38, "lr": 0.001} +{"train_loss": 0.052553918212652206, "train_loss_bc": 0.052553918212652206, "train_loss_llm": 0.0, "grad_norm": 0.026687031611800194, "global_step": 1379, "epoch": 38, "lr": 0.001} +{"train_loss": 0.057282704859972, "train_loss_bc": 0.05724506825208664, "train_loss_llm": 0.0037637650966644287, "grad_norm": 0.034762561321258545, "global_step": 1380, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04439760744571686, "train_loss_bc": 0.04439760744571686, "train_loss_llm": 0.0, "grad_norm": 0.017688637599349022, "global_step": 1381, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04766768217086792, "train_loss_bc": 0.047657255083322525, "train_loss_llm": 0.0010425746440887451, "grad_norm": 0.02414894849061966, "global_step": 1382, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04652877151966095, "train_loss_bc": 0.04652877151966095, "train_loss_llm": 0.0, "grad_norm": 0.031180983409285545, "global_step": 1383, "epoch": 38, "lr": 0.001} +{"train_loss": 0.048207320272922516, "train_loss_bc": 0.04818210005760193, "train_loss_llm": 0.0025218427181243896, "grad_norm": 0.03808436915278435, "global_step": 1384, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04460446536540985, "train_loss_bc": 0.04459930583834648, "train_loss_llm": 0.0005158483982086182, "grad_norm": 0.01678847149014473, "global_step": 1385, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04724918678402901, "train_loss_bc": 0.04724918678402901, "train_loss_llm": 0.0, "grad_norm": 0.026341794058680534, "global_step": 1386, "epoch": 38, "lr": 0.001} +{"train_loss": 0.05793340131640434, "train_loss_bc": 0.057930607348680496, "train_loss_llm": 0.00027933716773986816, "grad_norm": 0.03844065219163895, "global_step": 1387, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04456479847431183, "train_loss_bc": 0.04456479847431183, "train_loss_llm": 0.0, "grad_norm": 0.04627750813961029, "global_step": 1388, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04995971545577049, "train_loss_bc": 0.04988731071352959, "train_loss_llm": 0.007240593433380127, "grad_norm": 0.015240451321005821, "global_step": 1389, "epoch": 38, "lr": 0.001} +{"train_loss": 0.0428941547870636, "train_loss_bc": 0.0428941547870636, "train_loss_llm": 0.0, "grad_norm": 0.02697155438363552, "global_step": 1390, "epoch": 38, "lr": 0.001} +{"train_loss": 0.040053293108940125, "train_loss_bc": 0.040053293108940125, "train_loss_llm": 0.0, "grad_norm": 0.03222375735640526, "global_step": 1391, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04345878213644028, "train_loss_bc": 0.04345878213644028, "train_loss_llm": 0.0, "grad_norm": 0.039102546870708466, "global_step": 1392, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04536832869052887, "train_loss_bc": 0.04535527154803276, "train_loss_llm": 0.0013058483600616455, "grad_norm": 0.017591578885912895, "global_step": 1393, "epoch": 38, "lr": 0.001} +{"train_loss": 0.05123180150985718, "train_loss_bc": 0.05123180150985718, "train_loss_llm": 0.0, "grad_norm": 0.02747233584523201, "global_step": 1394, "epoch": 38, "lr": 0.001} +{"train_loss": 0.054106779396533966, "train_loss_bc": 0.054106779396533966, "train_loss_llm": 0.0, "grad_norm": 0.034224532544612885, "global_step": 1395, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04729336500167847, "train_loss_bc": 0.047280773520469666, "train_loss_llm": 0.0012590885162353516, "grad_norm": 0.03837920352816582, "global_step": 1396, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04655611142516136, "train_loss_bc": 0.046556007117033005, "train_loss_llm": 1.055002212524414e-05, "grad_norm": 0.016320889815688133, "global_step": 1397, "epoch": 38, "lr": 0.001} +{"train_loss": 0.0408308319747448, "train_loss_bc": 0.0408308319747448, "train_loss_llm": 0.0, "grad_norm": 0.02243073470890522, "global_step": 1398, "epoch": 38, "lr": 0.001} +{"train_loss": 0.051545191556215286, "train_loss_bc": 0.051530227065086365, "train_loss_llm": 0.0014962852001190186, "grad_norm": 0.030773548409342766, "global_step": 1399, "epoch": 38, "lr": 0.001} +{"train_loss": 0.05208219215273857, "train_loss_bc": 0.05208219215273857, "train_loss_llm": 0.0, "grad_norm": 0.04093867167830467, "global_step": 1400, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04074743390083313, "train_loss_bc": 0.04074743390083313, "train_loss_llm": 0.0, "grad_norm": 0.019327078014612198, "global_step": 1401, "epoch": 38, "lr": 0.001} +{"train_loss": 0.04689998924732208, "train_loss_bc": 0.04689998924732208, "train_loss_llm": 0.0, "grad_norm": 0.02710980549454689, "global_step": 1402, "epoch": 38, "lr": 0.001} +{"train_loss": 0.046999121809171304, "train_loss_bc": 0.044808898121118546, "train_loss_llm": 0.0018177330493927002, "grad_norm": 0.03615553677082062, "global_step": 1403, "epoch": 38, "lr": 0.001, "val_loss": 0.05084298178553581} +{"train_loss": 0.04980425536632538, "train_loss_bc": 0.04978363588452339, "train_loss_llm": 0.002061814069747925, "grad_norm": 0.04315907880663872, "global_step": 1404, "epoch": 39, "lr": 0.001} +{"train_loss": 0.05049348250031471, "train_loss_bc": 0.050486937165260315, "train_loss_llm": 0.0006545782089233398, "grad_norm": 0.017581695690751076, "global_step": 1405, "epoch": 39, "lr": 0.001} +{"train_loss": 0.038669269531965256, "train_loss_bc": 0.038669269531965256, "train_loss_llm": 0.0, "grad_norm": 0.020974062383174896, "global_step": 1406, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04319428279995918, "train_loss_bc": 0.04319428279995918, "train_loss_llm": 0.0, "grad_norm": 0.023238321766257286, "global_step": 1407, "epoch": 39, "lr": 0.001} +{"train_loss": 0.042744822800159454, "train_loss_bc": 0.042744822800159454, "train_loss_llm": 0.0, "grad_norm": 0.0237446129322052, "global_step": 1408, "epoch": 39, "lr": 0.001} +{"train_loss": 0.05006524175405502, "train_loss_bc": 0.050062309950590134, "train_loss_llm": 0.0002930760383605957, "grad_norm": 0.01920393854379654, "global_step": 1409, "epoch": 39, "lr": 0.001} +{"train_loss": 0.05018489807844162, "train_loss_bc": 0.05018489807844162, "train_loss_llm": 0.0, "grad_norm": 0.02454458363354206, "global_step": 1410, "epoch": 39, "lr": 0.001} +{"train_loss": 0.0466647744178772, "train_loss_bc": 0.046663351356983185, "train_loss_llm": 0.00014218688011169434, "grad_norm": 0.029017023742198944, "global_step": 1411, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04728022590279579, "train_loss_bc": 0.04728022590279579, "train_loss_llm": 0.0, "grad_norm": 0.03257809951901436, "global_step": 1412, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04625202342867851, "train_loss_bc": 0.04625202342867851, "train_loss_llm": 0.0, "grad_norm": 0.015271654352545738, "global_step": 1413, "epoch": 39, "lr": 0.001} +{"train_loss": 0.05002465099096298, "train_loss_bc": 0.05001349374651909, "train_loss_llm": 0.001115560531616211, "grad_norm": 0.02417657896876335, "global_step": 1414, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04528886824846268, "train_loss_bc": 0.045226942747831345, "train_loss_llm": 0.006192445755004883, "grad_norm": 0.02817106433212757, "global_step": 1415, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04796909913420677, "train_loss_bc": 0.04796909913420677, "train_loss_llm": 0.0, "grad_norm": 0.03951895982027054, "global_step": 1416, "epoch": 39, "lr": 0.001} +{"train_loss": 0.041781194508075714, "train_loss_bc": 0.041781194508075714, "train_loss_llm": 0.0, "grad_norm": 0.020424149930477142, "global_step": 1417, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04889971762895584, "train_loss_bc": 0.04889971762895584, "train_loss_llm": 0.0, "grad_norm": 0.02799716405570507, "global_step": 1418, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04505154862999916, "train_loss_bc": 0.04497028514742851, "train_loss_llm": 0.00812646746635437, "grad_norm": 0.03228433430194855, "global_step": 1419, "epoch": 39, "lr": 0.001} +{"train_loss": 0.0530591644346714, "train_loss_bc": 0.05305662378668785, "train_loss_llm": 0.0002542436122894287, "grad_norm": 0.043299127370119095, "global_step": 1420, "epoch": 39, "lr": 0.001} +{"train_loss": 0.05713878571987152, "train_loss_bc": 0.0571303628385067, "train_loss_llm": 0.0008422434329986572, "grad_norm": 0.018489902839064598, "global_step": 1421, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04953456297516823, "train_loss_bc": 0.04953456297516823, "train_loss_llm": 0.0, "grad_norm": 0.027620544657111168, "global_step": 1422, "epoch": 39, "lr": 0.001} +{"train_loss": 0.05052788555622101, "train_loss_bc": 0.05052788555622101, "train_loss_llm": 0.0, "grad_norm": 0.04202811419963837, "global_step": 1423, "epoch": 39, "lr": 0.001} +{"train_loss": 0.03969178348779678, "train_loss_bc": 0.03966877609491348, "train_loss_llm": 0.0023006200790405273, "grad_norm": 0.04830225557088852, "global_step": 1424, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04772408679127693, "train_loss_bc": 0.047715168446302414, "train_loss_llm": 0.0008917450904846191, "grad_norm": 0.014558405615389347, "global_step": 1425, "epoch": 39, "lr": 0.001} +{"train_loss": 0.0482604056596756, "train_loss_bc": 0.04825467988848686, "train_loss_llm": 0.00057259202003479, "grad_norm": 0.02372012659907341, "global_step": 1426, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04865771159529686, "train_loss_bc": 0.04865771159529686, "train_loss_llm": 0.0, "grad_norm": 0.028544018045067787, "global_step": 1427, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04260401427745819, "train_loss_bc": 0.04259795695543289, "train_loss_llm": 0.0006057620048522949, "grad_norm": 0.03606318682432175, "global_step": 1428, "epoch": 39, "lr": 0.001} +{"train_loss": 0.041416991502046585, "train_loss_bc": 0.041416991502046585, "train_loss_llm": 0.0, "grad_norm": 0.013471786864101887, "global_step": 1429, "epoch": 39, "lr": 0.001} +{"train_loss": 0.05047963559627533, "train_loss_bc": 0.05047963559627533, "train_loss_llm": 0.0, "grad_norm": 0.02293265610933304, "global_step": 1430, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04189878702163696, "train_loss_bc": 0.04189878702163696, "train_loss_llm": 0.0, "grad_norm": 0.02468453347682953, "global_step": 1431, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04406728222966194, "train_loss_bc": 0.044056013226509094, "train_loss_llm": 0.0011269450187683105, "grad_norm": 0.02658364176750183, "global_step": 1432, "epoch": 39, "lr": 0.001} +{"train_loss": 0.053345561027526855, "train_loss_bc": 0.053345561027526855, "train_loss_llm": 0.0, "grad_norm": 0.016826143488287926, "global_step": 1433, "epoch": 39, "lr": 0.001} +{"train_loss": 0.05191250145435333, "train_loss_bc": 0.05189032107591629, "train_loss_llm": 0.0022180676460266113, "grad_norm": 0.02476758509874344, "global_step": 1434, "epoch": 39, "lr": 0.001} +{"train_loss": 0.042504243552684784, "train_loss_bc": 0.042504243552684784, "train_loss_llm": 0.0, "grad_norm": 0.02910856530070305, "global_step": 1435, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04915982484817505, "train_loss_bc": 0.049157071858644485, "train_loss_llm": 0.0002751648426055908, "grad_norm": 0.04004192724823952, "global_step": 1436, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04622619226574898, "train_loss_bc": 0.04622619226574898, "train_loss_llm": 0.0, "grad_norm": 0.012929413467645645, "global_step": 1437, "epoch": 39, "lr": 0.001} +{"train_loss": 0.043494705110788345, "train_loss_bc": 0.043494705110788345, "train_loss_llm": 0.0, "grad_norm": 0.020920734852552414, "global_step": 1438, "epoch": 39, "lr": 0.001} +{"train_loss": 0.04685928817424509, "train_loss_bc": 0.04085766151547432, "train_loss_llm": 0.00042301416397094727, "grad_norm": 0.027504147961735725, "global_step": 1439, "epoch": 39, "lr": 0.001, "val_loss": 0.05055329203605652} +{"train_loss": 0.04767455905675888, "train_loss_bc": 0.04767455905675888, "train_loss_llm": 0.0, "grad_norm": 0.03388664126396179, "global_step": 1440, "epoch": 40, "lr": 0.001} +{"train_loss": 0.0403122752904892, "train_loss_bc": 0.04030647501349449, "train_loss_llm": 0.0005799233913421631, "grad_norm": 0.01536238007247448, "global_step": 1441, "epoch": 40, "lr": 0.001} +{"train_loss": 0.053052883595228195, "train_loss_bc": 0.053048305213451385, "train_loss_llm": 0.0004579126834869385, "grad_norm": 0.0234227254986763, "global_step": 1442, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04254274442791939, "train_loss_bc": 0.04254274442791939, "train_loss_llm": 0.0, "grad_norm": 0.02675662189722061, "global_step": 1443, "epoch": 40, "lr": 0.001} +{"train_loss": 0.0383814312517643, "train_loss_bc": 0.03835073485970497, "train_loss_llm": 0.003069758415222168, "grad_norm": 0.03277597576379776, "global_step": 1444, "epoch": 40, "lr": 0.001} +{"train_loss": 0.050541214644908905, "train_loss_bc": 0.05053260922431946, "train_loss_llm": 0.0008605420589447021, "grad_norm": 0.015355453826487064, "global_step": 1445, "epoch": 40, "lr": 0.001} +{"train_loss": 0.047167934477329254, "train_loss_bc": 0.047167934477329254, "train_loss_llm": 0.0, "grad_norm": 0.023803384974598885, "global_step": 1446, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04473940655589104, "train_loss_bc": 0.04473940655589104, "train_loss_llm": 0.0, "grad_norm": 0.030709607526659966, "global_step": 1447, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04970595985651016, "train_loss_bc": 0.04970595985651016, "train_loss_llm": 0.0, "grad_norm": 0.0331789031624794, "global_step": 1448, "epoch": 40, "lr": 0.001} +{"train_loss": 0.05001675710082054, "train_loss_bc": 0.05001675710082054, "train_loss_llm": 0.0, "grad_norm": 0.015418507158756256, "global_step": 1449, "epoch": 40, "lr": 0.001} +{"train_loss": 0.05251052603125572, "train_loss_bc": 0.05251052603125572, "train_loss_llm": 0.0, "grad_norm": 0.021212227642536163, "global_step": 1450, "epoch": 40, "lr": 0.001} +{"train_loss": 0.058174341917037964, "train_loss_bc": 0.058174341917037964, "train_loss_llm": 0.0, "grad_norm": 0.023994551971554756, "global_step": 1451, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04747508093714714, "train_loss_bc": 0.047465384006500244, "train_loss_llm": 0.0009695291519165039, "grad_norm": 0.03120311163365841, "global_step": 1452, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04956626519560814, "train_loss_bc": 0.04956626519560814, "train_loss_llm": 0.0, "grad_norm": 0.014675351791083813, "global_step": 1453, "epoch": 40, "lr": 0.001} +{"train_loss": 0.05565503239631653, "train_loss_bc": 0.05565503239631653, "train_loss_llm": 0.0, "grad_norm": 0.024070249870419502, "global_step": 1454, "epoch": 40, "lr": 0.001} +{"train_loss": 0.036122776567935944, "train_loss_bc": 0.036122776567935944, "train_loss_llm": 0.0, "grad_norm": 0.027963418513536453, "global_step": 1455, "epoch": 40, "lr": 0.001} +{"train_loss": 0.044186294078826904, "train_loss_bc": 0.04415939003229141, "train_loss_llm": 0.0026903152465820312, "grad_norm": 0.03501899167895317, "global_step": 1456, "epoch": 40, "lr": 0.001} +{"train_loss": 0.03803262859582901, "train_loss_bc": 0.03803262859582901, "train_loss_llm": 0.0, "grad_norm": 0.013825592584908009, "global_step": 1457, "epoch": 40, "lr": 0.001} +{"train_loss": 0.05032623931765556, "train_loss_bc": 0.05031273141503334, "train_loss_llm": 0.0013507306575775146, "grad_norm": 0.02074298821389675, "global_step": 1458, "epoch": 40, "lr": 0.001} +{"train_loss": 0.06008363887667656, "train_loss_bc": 0.059978634119033813, "train_loss_llm": 0.010500341653823853, "grad_norm": 0.03056674264371395, "global_step": 1459, "epoch": 40, "lr": 0.001} +{"train_loss": 0.053287360817193985, "train_loss_bc": 0.053274378180503845, "train_loss_llm": 0.0012983381748199463, "grad_norm": 0.03597915545105934, "global_step": 1460, "epoch": 40, "lr": 0.001} +{"train_loss": 0.050416249781847, "train_loss_bc": 0.050404421985149384, "train_loss_llm": 0.0011827647686004639, "grad_norm": 0.015775538980960846, "global_step": 1461, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04293910413980484, "train_loss_bc": 0.04293910413980484, "train_loss_llm": 0.0, "grad_norm": 0.023746438324451447, "global_step": 1462, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04495770484209061, "train_loss_bc": 0.04494745284318924, "train_loss_llm": 0.0010251998901367188, "grad_norm": 0.03266507759690285, "global_step": 1463, "epoch": 40, "lr": 0.001} +{"train_loss": 0.05529068037867546, "train_loss_bc": 0.05529068037867546, "train_loss_llm": 0.0, "grad_norm": 0.03950796648859978, "global_step": 1464, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04524015262722969, "train_loss_bc": 0.04524015262722969, "train_loss_llm": 0.0, "grad_norm": 0.015959544107317924, "global_step": 1465, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04216967523097992, "train_loss_bc": 0.04216967523097992, "train_loss_llm": 0.0, "grad_norm": 0.022198669612407684, "global_step": 1466, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04048948734998703, "train_loss_bc": 0.040456607937812805, "train_loss_llm": 0.0032879114151000977, "grad_norm": 0.02528402768075466, "global_step": 1467, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04259157180786133, "train_loss_bc": 0.042575281113386154, "train_loss_llm": 0.0016289949417114258, "grad_norm": 0.031169166788458824, "global_step": 1468, "epoch": 40, "lr": 0.001} +{"train_loss": 0.040773674845695496, "train_loss_bc": 0.040773674845695496, "train_loss_llm": 0.0, "grad_norm": 0.015639513731002808, "global_step": 1469, "epoch": 40, "lr": 0.001} +{"train_loss": 0.041098304092884064, "train_loss_bc": 0.041098304092884064, "train_loss_llm": 0.0, "grad_norm": 0.020082982257008553, "global_step": 1470, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04308219999074936, "train_loss_bc": 0.04308219999074936, "train_loss_llm": 0.0, "grad_norm": 0.026095276698470116, "global_step": 1471, "epoch": 40, "lr": 0.001} +{"train_loss": 0.04793199151754379, "train_loss_bc": 0.04788350686430931, "train_loss_llm": 0.004848450422286987, "grad_norm": 0.03007105551660061, "global_step": 1472, "epoch": 40, "lr": 0.001} +{"train_loss": 0.048612356185913086, "train_loss_bc": 0.048590779304504395, "train_loss_llm": 0.0021576881408691406, "grad_norm": 0.014805245213210583, "global_step": 1473, "epoch": 40, "lr": 0.001} +{"train_loss": 0.040989942848682404, "train_loss_bc": 0.04098762199282646, "train_loss_llm": 0.00023216009140014648, "grad_norm": 0.022439150139689445, "global_step": 1474, "epoch": 40, "lr": 0.001} +{"train_loss": 0.046742786860300437, "train_loss_bc": 0.046601880341768265, "train_loss_llm": 0.0, "grad_norm": 0.029482489451766014, "global_step": 1475, "epoch": 40, "lr": 0.001, "train/cumulative_reward": 38.28101398150974, "train/mean_score": 3.5266478237161842, "train/success_rate": 0.0, "test/cumulative_reward": 43.69155865925497, "test/mean_score": 2.4395850761226754, "test/success_rate": 0.02, "val_loss": 0.050310149788856506, "train_action_mse_error": 0.05086097493767738} +{"train_loss": 0.04706098511815071, "train_loss_bc": 0.04703928902745247, "train_loss_llm": 0.002169787883758545, "grad_norm": 0.0322464182972908, "global_step": 1476, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04508756846189499, "train_loss_bc": 0.04508756846189499, "train_loss_llm": 0.0, "grad_norm": 0.012412609532475471, "global_step": 1477, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04736889898777008, "train_loss_bc": 0.04732612892985344, "train_loss_llm": 0.00427699089050293, "grad_norm": 0.018955839797854424, "global_step": 1478, "epoch": 41, "lr": 0.001} +{"train_loss": 0.05047200620174408, "train_loss_bc": 0.05045873671770096, "train_loss_llm": 0.0013267993927001953, "grad_norm": 0.02413235418498516, "global_step": 1479, "epoch": 41, "lr": 0.001} +{"train_loss": 0.05903634428977966, "train_loss_bc": 0.05903634428977966, "train_loss_llm": 0.0, "grad_norm": 0.02981051616370678, "global_step": 1480, "epoch": 41, "lr": 0.001} +{"train_loss": 0.051922913640737534, "train_loss_bc": 0.05191928520798683, "train_loss_llm": 0.00036266446113586426, "grad_norm": 0.016481701284646988, "global_step": 1481, "epoch": 41, "lr": 0.001} +{"train_loss": 0.03675449267029762, "train_loss_bc": 0.03675449267029762, "train_loss_llm": 0.0, "grad_norm": 0.01846720650792122, "global_step": 1482, "epoch": 41, "lr": 0.001} +{"train_loss": 0.049650970846414566, "train_loss_bc": 0.04960725083947182, "train_loss_llm": 0.004371911287307739, "grad_norm": 0.0290363896638155, "global_step": 1483, "epoch": 41, "lr": 0.001} +{"train_loss": 0.052622098475694656, "train_loss_bc": 0.052622098475694656, "train_loss_llm": 0.0, "grad_norm": 0.033252228051424026, "global_step": 1484, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04427379369735718, "train_loss_bc": 0.04427379369735718, "train_loss_llm": 0.0, "grad_norm": 0.013813563622534275, "global_step": 1485, "epoch": 41, "lr": 0.001} +{"train_loss": 0.046500127762556076, "train_loss_bc": 0.046497441828250885, "train_loss_llm": 0.0002684295177459717, "grad_norm": 0.0177664402872324, "global_step": 1486, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04614638537168503, "train_loss_bc": 0.04613501578569412, "train_loss_llm": 0.0011370480060577393, "grad_norm": 0.026051431894302368, "global_step": 1487, "epoch": 41, "lr": 0.001} +{"train_loss": 0.05051056295633316, "train_loss_bc": 0.05048292875289917, "train_loss_llm": 0.0027632713317871094, "grad_norm": 0.028878476470708847, "global_step": 1488, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04248008877038956, "train_loss_bc": 0.04246523603796959, "train_loss_llm": 0.001485377550125122, "grad_norm": 0.0164061076939106, "global_step": 1489, "epoch": 41, "lr": 0.001} +{"train_loss": 0.05434287711977959, "train_loss_bc": 0.0543331578373909, "train_loss_llm": 0.0009720027446746826, "grad_norm": 0.023814169690012932, "global_step": 1490, "epoch": 41, "lr": 0.001} +{"train_loss": 0.041836127638816833, "train_loss_bc": 0.04182765632867813, "train_loss_llm": 0.000847160816192627, "grad_norm": 0.029839707538485527, "global_step": 1491, "epoch": 41, "lr": 0.001} +{"train_loss": 0.05506439879536629, "train_loss_bc": 0.05504703149199486, "train_loss_llm": 0.0017366111278533936, "grad_norm": 0.04147283732891083, "global_step": 1492, "epoch": 41, "lr": 0.001} +{"train_loss": 0.034461770206689835, "train_loss_bc": 0.03443717211484909, "train_loss_llm": 0.0024599432945251465, "grad_norm": 0.012423933483660221, "global_step": 1493, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04456750676035881, "train_loss_bc": 0.04456371068954468, "train_loss_llm": 0.0003795027732849121, "grad_norm": 0.022020738571882248, "global_step": 1494, "epoch": 41, "lr": 0.001} +{"train_loss": 0.05422056093811989, "train_loss_bc": 0.05422056093811989, "train_loss_llm": 0.0, "grad_norm": 0.031884580850601196, "global_step": 1495, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04743028059601784, "train_loss_bc": 0.04743028059601784, "train_loss_llm": 0.0, "grad_norm": 0.03518747538328171, "global_step": 1496, "epoch": 41, "lr": 0.001} +{"train_loss": 0.0464932918548584, "train_loss_bc": 0.0464932918548584, "train_loss_llm": 0.0, "grad_norm": 0.01721429079771042, "global_step": 1497, "epoch": 41, "lr": 0.001} +{"train_loss": 0.05291832610964775, "train_loss_bc": 0.05291832610964775, "train_loss_llm": 0.0, "grad_norm": 0.027048548683524132, "global_step": 1498, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04567693546414375, "train_loss_bc": 0.04567693546414375, "train_loss_llm": 0.0, "grad_norm": 0.034344032406806946, "global_step": 1499, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04282158613204956, "train_loss_bc": 0.04282158613204956, "train_loss_llm": 0.0, "grad_norm": 0.0363403744995594, "global_step": 1500, "epoch": 41, "lr": 0.001} +{"train_loss": 0.046675439924001694, "train_loss_bc": 0.046675439924001694, "train_loss_llm": 0.0, "grad_norm": 0.016211234033107758, "global_step": 1501, "epoch": 41, "lr": 0.001} +{"train_loss": 0.038892604410648346, "train_loss_bc": 0.038892604410648346, "train_loss_llm": 0.0, "grad_norm": 0.023371633142232895, "global_step": 1502, "epoch": 41, "lr": 0.001} +{"train_loss": 0.043281182646751404, "train_loss_bc": 0.04327411204576492, "train_loss_llm": 0.0007071197032928467, "grad_norm": 0.028353659436106682, "global_step": 1503, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04887840151786804, "train_loss_bc": 0.04887840151786804, "train_loss_llm": 0.0, "grad_norm": 0.03227822110056877, "global_step": 1504, "epoch": 41, "lr": 0.001} +{"train_loss": 0.03896154463291168, "train_loss_bc": 0.03896154463291168, "train_loss_llm": 0.0, "grad_norm": 0.016129635274410248, "global_step": 1505, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04708915576338768, "train_loss_bc": 0.047087039798498154, "train_loss_llm": 0.0002117156982421875, "grad_norm": 0.02506934478878975, "global_step": 1506, "epoch": 41, "lr": 0.001} +{"train_loss": 0.03919804096221924, "train_loss_bc": 0.03919099271297455, "train_loss_llm": 0.0007048249244689941, "grad_norm": 0.03062884695827961, "global_step": 1507, "epoch": 41, "lr": 0.001} +{"train_loss": 0.046466656029224396, "train_loss_bc": 0.046466656029224396, "train_loss_llm": 0.0, "grad_norm": 0.0350615531206131, "global_step": 1508, "epoch": 41, "lr": 0.001} +{"train_loss": 0.049049653112888336, "train_loss_bc": 0.04903290793299675, "train_loss_llm": 0.0016744136810302734, "grad_norm": 0.01703922636806965, "global_step": 1509, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04547598585486412, "train_loss_bc": 0.04546928033232689, "train_loss_llm": 0.0006705224514007568, "grad_norm": 0.02486134134232998, "global_step": 1510, "epoch": 41, "lr": 0.001} +{"train_loss": 0.04658888218303522, "train_loss_bc": 0.04351019486784935, "train_loss_llm": 0.0, "grad_norm": 0.036168359220027924, "global_step": 1511, "epoch": 41, "lr": 0.001, "val_loss": 0.05066293478012085} +{"train_loss": 0.04358115792274475, "train_loss_bc": 0.04357365146279335, "train_loss_llm": 0.0007506608963012695, "grad_norm": 0.04235134646296501, "global_step": 1512, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04516556113958359, "train_loss_bc": 0.04514632001519203, "train_loss_llm": 0.001923978328704834, "grad_norm": 0.0137428417801857, "global_step": 1513, "epoch": 42, "lr": 0.001} +{"train_loss": 0.050273697823286057, "train_loss_bc": 0.050263576209545135, "train_loss_llm": 0.0010120868682861328, "grad_norm": 0.021347958594560623, "global_step": 1514, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04244089126586914, "train_loss_bc": 0.04244089126586914, "train_loss_llm": 0.0, "grad_norm": 0.02695123851299286, "global_step": 1515, "epoch": 42, "lr": 0.001} +{"train_loss": 0.05040017142891884, "train_loss_bc": 0.05040017142891884, "train_loss_llm": 0.0, "grad_norm": 0.03654365986585617, "global_step": 1516, "epoch": 42, "lr": 0.001} +{"train_loss": 0.039147038012742996, "train_loss_bc": 0.039147038012742996, "train_loss_llm": 0.0, "grad_norm": 0.014204381965100765, "global_step": 1517, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04824545234441757, "train_loss_bc": 0.04824545234441757, "train_loss_llm": 0.0, "grad_norm": 0.024401890113949776, "global_step": 1518, "epoch": 42, "lr": 0.001} +{"train_loss": 0.053365834057331085, "train_loss_bc": 0.053365834057331085, "train_loss_llm": 0.0, "grad_norm": 0.03362996503710747, "global_step": 1519, "epoch": 42, "lr": 0.001} +{"train_loss": 0.045938510447740555, "train_loss_bc": 0.04589341580867767, "train_loss_llm": 0.004509389400482178, "grad_norm": 0.039056532084941864, "global_step": 1520, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04739919304847717, "train_loss_bc": 0.04739919304847717, "train_loss_llm": 0.0, "grad_norm": 0.015916790813207626, "global_step": 1521, "epoch": 42, "lr": 0.001} +{"train_loss": 0.043523550033569336, "train_loss_bc": 0.0435229167342186, "train_loss_llm": 6.321072578430176e-05, "grad_norm": 0.022078195586800575, "global_step": 1522, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04987628757953644, "train_loss_bc": 0.04987628757953644, "train_loss_llm": 0.0, "grad_norm": 0.03008136712014675, "global_step": 1523, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04750123247504234, "train_loss_bc": 0.04750123247504234, "train_loss_llm": 0.0, "grad_norm": 0.03201981261372566, "global_step": 1524, "epoch": 42, "lr": 0.001} +{"train_loss": 0.05199010670185089, "train_loss_bc": 0.05197007954120636, "train_loss_llm": 0.002002894878387451, "grad_norm": 0.016632843762636185, "global_step": 1525, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04890182986855507, "train_loss_bc": 0.04885397478938103, "train_loss_llm": 0.004785358905792236, "grad_norm": 0.027793090790510178, "global_step": 1526, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04514879733324051, "train_loss_bc": 0.0451405867934227, "train_loss_llm": 0.0008211731910705566, "grad_norm": 0.035153500735759735, "global_step": 1527, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04778367653489113, "train_loss_bc": 0.04777878522872925, "train_loss_llm": 0.0004889965057373047, "grad_norm": 0.03472364693880081, "global_step": 1528, "epoch": 42, "lr": 0.001} +{"train_loss": 0.05658696964383125, "train_loss_bc": 0.05658696964383125, "train_loss_llm": 0.0, "grad_norm": 0.017681390047073364, "global_step": 1529, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04105188325047493, "train_loss_bc": 0.04105188325047493, "train_loss_llm": 0.0, "grad_norm": 0.02683509700000286, "global_step": 1530, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04496109113097191, "train_loss_bc": 0.04494880139827728, "train_loss_llm": 0.001229017972946167, "grad_norm": 0.033013347536325455, "global_step": 1531, "epoch": 42, "lr": 0.001} +{"train_loss": 0.051758404821157455, "train_loss_bc": 0.051758404821157455, "train_loss_llm": 0.0, "grad_norm": 0.036005448549985886, "global_step": 1532, "epoch": 42, "lr": 0.001} +{"train_loss": 0.044048700481653214, "train_loss_bc": 0.044048700481653214, "train_loss_llm": 0.0, "grad_norm": 0.016563137993216515, "global_step": 1533, "epoch": 42, "lr": 0.001} +{"train_loss": 0.05062796548008919, "train_loss_bc": 0.05061223730444908, "train_loss_llm": 0.0015726685523986816, "grad_norm": 0.021073050796985626, "global_step": 1534, "epoch": 42, "lr": 0.001} +{"train_loss": 0.041172657161951065, "train_loss_bc": 0.041172657161951065, "train_loss_llm": 0.0, "grad_norm": 0.027579771354794502, "global_step": 1535, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04365992918610573, "train_loss_bc": 0.04365354776382446, "train_loss_llm": 0.0006379783153533936, "grad_norm": 0.0293271504342556, "global_step": 1536, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04289160296320915, "train_loss_bc": 0.04289160296320915, "train_loss_llm": 0.0, "grad_norm": 0.016213586553931236, "global_step": 1537, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04392717033624649, "train_loss_bc": 0.04389965906739235, "train_loss_llm": 0.0027511417865753174, "grad_norm": 0.024687381461262703, "global_step": 1538, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04473365843296051, "train_loss_bc": 0.04473365843296051, "train_loss_llm": 0.0, "grad_norm": 0.033929500728845596, "global_step": 1539, "epoch": 42, "lr": 0.001} +{"train_loss": 0.0482657216489315, "train_loss_bc": 0.0482657216489315, "train_loss_llm": 0.0, "grad_norm": 0.04036315158009529, "global_step": 1540, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04689927399158478, "train_loss_bc": 0.04689927399158478, "train_loss_llm": 0.0, "grad_norm": 0.016748279333114624, "global_step": 1541, "epoch": 42, "lr": 0.001} +{"train_loss": 0.046998754143714905, "train_loss_bc": 0.046998754143714905, "train_loss_llm": 0.0, "grad_norm": 0.026588844135403633, "global_step": 1542, "epoch": 42, "lr": 0.001} +{"train_loss": 0.041142091155052185, "train_loss_bc": 0.041142091155052185, "train_loss_llm": 0.0, "grad_norm": 0.03124512918293476, "global_step": 1543, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04555654153227806, "train_loss_bc": 0.04555654153227806, "train_loss_llm": 0.0, "grad_norm": 0.03896665200591087, "global_step": 1544, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04768498241901398, "train_loss_bc": 0.04768498241901398, "train_loss_llm": 0.0, "grad_norm": 0.018835222348570824, "global_step": 1545, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04422607272863388, "train_loss_bc": 0.044189225882291794, "train_loss_llm": 0.003684699535369873, "grad_norm": 0.02429971471428871, "global_step": 1546, "epoch": 42, "lr": 0.001} +{"train_loss": 0.04658915764755673, "train_loss_bc": 0.05033321678638458, "train_loss_llm": 0.0, "grad_norm": 0.031875137239694595, "global_step": 1547, "epoch": 42, "lr": 0.001, "val_loss": 0.05053401738405228} +{"train_loss": 0.04084963724017143, "train_loss_bc": 0.04082781821489334, "train_loss_llm": 0.002181828022003174, "grad_norm": 0.03460141271352768, "global_step": 1548, "epoch": 43, "lr": 0.001} +{"train_loss": 0.049056585878133774, "train_loss_bc": 0.049043986946344376, "train_loss_llm": 0.001259922981262207, "grad_norm": 0.01838257722556591, "global_step": 1549, "epoch": 43, "lr": 0.001} +{"train_loss": 0.042116113007068634, "train_loss_bc": 0.042116113007068634, "train_loss_llm": 0.0, "grad_norm": 0.019057143479585648, "global_step": 1550, "epoch": 43, "lr": 0.001} +{"train_loss": 0.05035056173801422, "train_loss_bc": 0.05035056173801422, "train_loss_llm": 0.0, "grad_norm": 0.025746431201696396, "global_step": 1551, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04124265909194946, "train_loss_bc": 0.041200440376996994, "train_loss_llm": 0.0042220354080200195, "grad_norm": 0.02638545259833336, "global_step": 1552, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04439816251397133, "train_loss_bc": 0.044379204511642456, "train_loss_llm": 0.0018956959247589111, "grad_norm": 0.015112791210412979, "global_step": 1553, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04717452451586723, "train_loss_bc": 0.04717452451586723, "train_loss_llm": 0.0, "grad_norm": 0.02202260121703148, "global_step": 1554, "epoch": 43, "lr": 0.001} +{"train_loss": 0.048246968537569046, "train_loss_bc": 0.048246968537569046, "train_loss_llm": 0.0, "grad_norm": 0.021571706980466843, "global_step": 1555, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04414881393313408, "train_loss_bc": 0.04414881393313408, "train_loss_llm": 0.0, "grad_norm": 0.02786208875477314, "global_step": 1556, "epoch": 43, "lr": 0.001} +{"train_loss": 0.05407341569662094, "train_loss_bc": 0.05407341569662094, "train_loss_llm": 0.0, "grad_norm": 0.017461327835917473, "global_step": 1557, "epoch": 43, "lr": 0.001} +{"train_loss": 0.036867935210466385, "train_loss_bc": 0.036867935210466385, "train_loss_llm": 0.0, "grad_norm": 0.026865776628255844, "global_step": 1558, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04699115455150604, "train_loss_bc": 0.04698760807514191, "train_loss_llm": 0.00035458803176879883, "grad_norm": 0.03200768679380417, "global_step": 1559, "epoch": 43, "lr": 0.001} +{"train_loss": 0.046330489218235016, "train_loss_bc": 0.04630949720740318, "train_loss_llm": 0.002099156379699707, "grad_norm": 0.04037162661552429, "global_step": 1560, "epoch": 43, "lr": 0.001} +{"train_loss": 0.039317887276411057, "train_loss_bc": 0.039317887276411057, "train_loss_llm": 0.0, "grad_norm": 0.01550893485546112, "global_step": 1561, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04982151836156845, "train_loss_bc": 0.04982151836156845, "train_loss_llm": 0.0, "grad_norm": 0.021359503269195557, "global_step": 1562, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04577875882387161, "train_loss_bc": 0.04576355963945389, "train_loss_llm": 0.0015199482440948486, "grad_norm": 0.026319028809666634, "global_step": 1563, "epoch": 43, "lr": 0.001} +{"train_loss": 0.049196865409612656, "train_loss_bc": 0.049196865409612656, "train_loss_llm": 0.0, "grad_norm": 0.028855115175247192, "global_step": 1564, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04918188601732254, "train_loss_bc": 0.0491718091070652, "train_loss_llm": 0.0010077357292175293, "grad_norm": 0.017170852050185204, "global_step": 1565, "epoch": 43, "lr": 0.001} +{"train_loss": 0.049246665090322495, "train_loss_bc": 0.04921957477927208, "train_loss_llm": 0.002709031105041504, "grad_norm": 0.02655664272606373, "global_step": 1566, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04889804869890213, "train_loss_bc": 0.04889804869890213, "train_loss_llm": 0.0, "grad_norm": 0.03808387741446495, "global_step": 1567, "epoch": 43, "lr": 0.001} +{"train_loss": 0.053237900137901306, "train_loss_bc": 0.053237900137901306, "train_loss_llm": 0.0, "grad_norm": 0.04695797711610794, "global_step": 1568, "epoch": 43, "lr": 0.001} +{"train_loss": 0.044661745429039, "train_loss_bc": 0.044661745429039, "train_loss_llm": 0.0, "grad_norm": 0.016539866104722023, "global_step": 1569, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04559777304530144, "train_loss_bc": 0.04559777304530144, "train_loss_llm": 0.0, "grad_norm": 0.01906280219554901, "global_step": 1570, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04577186331152916, "train_loss_bc": 0.04577186331152916, "train_loss_llm": 0.0, "grad_norm": 0.025474857538938522, "global_step": 1571, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04662624001502991, "train_loss_bc": 0.04655994102358818, "train_loss_llm": 0.006629735231399536, "grad_norm": 0.03289027884602547, "global_step": 1572, "epoch": 43, "lr": 0.001} +{"train_loss": 0.05141797661781311, "train_loss_bc": 0.051362309604883194, "train_loss_llm": 0.0055665671825408936, "grad_norm": 0.015143911354243755, "global_step": 1573, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04238106682896614, "train_loss_bc": 0.04237974062561989, "train_loss_llm": 0.00013247132301330566, "grad_norm": 0.021025152876973152, "global_step": 1574, "epoch": 43, "lr": 0.001} +{"train_loss": 0.05447952821850777, "train_loss_bc": 0.05444924160838127, "train_loss_llm": 0.0030288398265838623, "grad_norm": 0.029320167377591133, "global_step": 1575, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04886353015899658, "train_loss_bc": 0.04885679855942726, "train_loss_llm": 0.0006733238697052002, "grad_norm": 0.039369694888591766, "global_step": 1576, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04758786782622337, "train_loss_bc": 0.04758685827255249, "train_loss_llm": 0.0001011192798614502, "grad_norm": 0.017854727804660797, "global_step": 1577, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04520286247134209, "train_loss_bc": 0.04520286247134209, "train_loss_llm": 0.0, "grad_norm": 0.024716902524232864, "global_step": 1578, "epoch": 43, "lr": 0.001} +{"train_loss": 0.05339928716421127, "train_loss_bc": 0.05339928716421127, "train_loss_llm": 0.0, "grad_norm": 0.03311137855052948, "global_step": 1579, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04954279586672783, "train_loss_bc": 0.04950903356075287, "train_loss_llm": 0.0033760666847229004, "grad_norm": 0.04317270591855049, "global_step": 1580, "epoch": 43, "lr": 0.001} +{"train_loss": 0.041515007615089417, "train_loss_bc": 0.041515007615089417, "train_loss_llm": 0.0, "grad_norm": 0.015187469311058521, "global_step": 1581, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04235651716589928, "train_loss_bc": 0.04235651716589928, "train_loss_llm": 0.0, "grad_norm": 0.025993291288614273, "global_step": 1582, "epoch": 43, "lr": 0.001} +{"train_loss": 0.04640608891430828, "train_loss_bc": 0.03468858823180199, "train_loss_llm": 0.0, "grad_norm": 0.032217543572187424, "global_step": 1583, "epoch": 43, "lr": 0.001, "val_loss": 0.050404101610183716} +{"train_loss": 0.04742329940199852, "train_loss_bc": 0.04742329940199852, "train_loss_llm": 0.0, "grad_norm": 0.03500228747725487, "global_step": 1584, "epoch": 44, "lr": 0.001} +{"train_loss": 0.040745723992586136, "train_loss_bc": 0.040745723992586136, "train_loss_llm": 0.0, "grad_norm": 0.01593197137117386, "global_step": 1585, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04379397630691528, "train_loss_bc": 0.043780531734228134, "train_loss_llm": 0.0013443231582641602, "grad_norm": 0.023646393790841103, "global_step": 1586, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04971931502223015, "train_loss_bc": 0.04969623312354088, "train_loss_llm": 0.0023081302642822266, "grad_norm": 0.029956234619021416, "global_step": 1587, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04034632071852684, "train_loss_bc": 0.04034632071852684, "train_loss_llm": 0.0, "grad_norm": 0.034168895334005356, "global_step": 1588, "epoch": 44, "lr": 0.001} +{"train_loss": 0.03694603964686394, "train_loss_bc": 0.036922354251146317, "train_loss_llm": 0.0023685097694396973, "grad_norm": 0.01672930084168911, "global_step": 1589, "epoch": 44, "lr": 0.001} +{"train_loss": 0.041979119181632996, "train_loss_bc": 0.04196402430534363, "train_loss_llm": 0.0015095174312591553, "grad_norm": 0.02024279534816742, "global_step": 1590, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04662061855196953, "train_loss_bc": 0.04660702124238014, "train_loss_llm": 0.0013595819473266602, "grad_norm": 0.024320067837834358, "global_step": 1591, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04572420194745064, "train_loss_bc": 0.04572420194745064, "train_loss_llm": 0.0, "grad_norm": 0.028284596279263496, "global_step": 1592, "epoch": 44, "lr": 0.001} +{"train_loss": 0.0469338484108448, "train_loss_bc": 0.04691055417060852, "train_loss_llm": 0.0023294389247894287, "grad_norm": 0.018032941967248917, "global_step": 1593, "epoch": 44, "lr": 0.001} +{"train_loss": 0.06094607710838318, "train_loss_bc": 0.06094607710838318, "train_loss_llm": 0.0, "grad_norm": 0.028394145891070366, "global_step": 1594, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04183550924062729, "train_loss_bc": 0.04183550924062729, "train_loss_llm": 0.0, "grad_norm": 0.035593289881944656, "global_step": 1595, "epoch": 44, "lr": 0.001} +{"train_loss": 0.0504908561706543, "train_loss_bc": 0.05048636719584465, "train_loss_llm": 0.0004490017890930176, "grad_norm": 0.03867786377668381, "global_step": 1596, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04161986708641052, "train_loss_bc": 0.04161986708641052, "train_loss_llm": 0.0, "grad_norm": 0.016084447503089905, "global_step": 1597, "epoch": 44, "lr": 0.001} +{"train_loss": 0.05317065864801407, "train_loss_bc": 0.05316983163356781, "train_loss_llm": 8.270144462585449e-05, "grad_norm": 0.01978413388133049, "global_step": 1598, "epoch": 44, "lr": 0.001} +{"train_loss": 0.051792919635772705, "train_loss_bc": 0.051792919635772705, "train_loss_llm": 0.0, "grad_norm": 0.031058378517627716, "global_step": 1599, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04926172271370888, "train_loss_bc": 0.049245353788137436, "train_loss_llm": 0.0016367137432098389, "grad_norm": 0.03560105711221695, "global_step": 1600, "epoch": 44, "lr": 0.001} +{"train_loss": 0.052413228899240494, "train_loss_bc": 0.052411869168281555, "train_loss_llm": 0.00013580918312072754, "grad_norm": 0.014471810311079025, "global_step": 1601, "epoch": 44, "lr": 0.001} +{"train_loss": 0.048439860343933105, "train_loss_bc": 0.04843245446681976, "train_loss_llm": 0.0007404088973999023, "grad_norm": 0.023175641894340515, "global_step": 1602, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04168562963604927, "train_loss_bc": 0.04166362062096596, "train_loss_llm": 0.002201080322265625, "grad_norm": 0.03010174073278904, "global_step": 1603, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04514434561133385, "train_loss_bc": 0.04514414072036743, "train_loss_llm": 2.0623207092285156e-05, "grad_norm": 0.04385870322585106, "global_step": 1604, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04205460101366043, "train_loss_bc": 0.04205460101366043, "train_loss_llm": 0.0, "grad_norm": 0.017333487048745155, "global_step": 1605, "epoch": 44, "lr": 0.001} +{"train_loss": 0.0373002327978611, "train_loss_bc": 0.037251994013786316, "train_loss_llm": 0.004823803901672363, "grad_norm": 0.0237493384629488, "global_step": 1606, "epoch": 44, "lr": 0.001} +{"train_loss": 0.049068599939346313, "train_loss_bc": 0.049068599939346313, "train_loss_llm": 0.0, "grad_norm": 0.025676194578409195, "global_step": 1607, "epoch": 44, "lr": 0.001} +{"train_loss": 0.051490139216184616, "train_loss_bc": 0.051490139216184616, "train_loss_llm": 0.0, "grad_norm": 0.03723141551017761, "global_step": 1608, "epoch": 44, "lr": 0.001} +{"train_loss": 0.0507148914039135, "train_loss_bc": 0.0507148914039135, "train_loss_llm": 0.0, "grad_norm": 0.017957353964447975, "global_step": 1609, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04726060479879379, "train_loss_bc": 0.04725787043571472, "train_loss_llm": 0.0002734363079071045, "grad_norm": 0.024860959500074387, "global_step": 1610, "epoch": 44, "lr": 0.001} +{"train_loss": 0.053055815398693085, "train_loss_bc": 0.053025614470243454, "train_loss_llm": 0.0030201971530914307, "grad_norm": 0.029098274186253548, "global_step": 1611, "epoch": 44, "lr": 0.001} +{"train_loss": 0.05042097717523575, "train_loss_bc": 0.05042097717523575, "train_loss_llm": 0.0, "grad_norm": 0.034066393971443176, "global_step": 1612, "epoch": 44, "lr": 0.001} +{"train_loss": 0.0378631092607975, "train_loss_bc": 0.03786000236868858, "train_loss_llm": 0.00031068921089172363, "grad_norm": 0.01561181340366602, "global_step": 1613, "epoch": 44, "lr": 0.001} +{"train_loss": 0.044737182557582855, "train_loss_bc": 0.044737182557582855, "train_loss_llm": 0.0, "grad_norm": 0.025404490530490875, "global_step": 1614, "epoch": 44, "lr": 0.001} +{"train_loss": 0.045889321714639664, "train_loss_bc": 0.045889321714639664, "train_loss_llm": 0.0, "grad_norm": 0.033092398196458817, "global_step": 1615, "epoch": 44, "lr": 0.001} +{"train_loss": 0.046525049954652786, "train_loss_bc": 0.046525049954652786, "train_loss_llm": 0.0, "grad_norm": 0.03978271409869194, "global_step": 1616, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04582052677869797, "train_loss_bc": 0.045805711299180984, "train_loss_llm": 0.001481562852859497, "grad_norm": 0.0142361456528306, "global_step": 1617, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04191301390528679, "train_loss_bc": 0.04191301390528679, "train_loss_llm": 0.0, "grad_norm": 0.025036418810486794, "global_step": 1618, "epoch": 44, "lr": 0.001} +{"train_loss": 0.04651551952378617, "train_loss_bc": 0.05341149866580963, "train_loss_llm": 0.0, "grad_norm": 0.03683309257030487, "global_step": 1619, "epoch": 44, "lr": 0.001, "val_loss": 0.050341930240392685} +{"train_loss": 0.04612666368484497, "train_loss_bc": 0.04612666368484497, "train_loss_llm": 0.0, "grad_norm": 0.04856239631772041, "global_step": 1620, "epoch": 45, "lr": 0.001} +{"train_loss": 0.041358932852745056, "train_loss_bc": 0.041358932852745056, "train_loss_llm": 0.0, "grad_norm": 0.013697952032089233, "global_step": 1621, "epoch": 45, "lr": 0.001} +{"train_loss": 0.049740247428417206, "train_loss_bc": 0.04973699524998665, "train_loss_llm": 0.00032520294189453125, "grad_norm": 0.025560973212122917, "global_step": 1622, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04382884129881859, "train_loss_bc": 0.04380359873175621, "train_loss_llm": 0.0025242269039154053, "grad_norm": 0.02928975783288479, "global_step": 1623, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04393193870782852, "train_loss_bc": 0.04390258342027664, "train_loss_llm": 0.0029355883598327637, "grad_norm": 0.0347905308008194, "global_step": 1624, "epoch": 45, "lr": 0.001} +{"train_loss": 0.03803667053580284, "train_loss_bc": 0.03800801560282707, "train_loss_llm": 0.0028655529022216797, "grad_norm": 0.014946612529456615, "global_step": 1625, "epoch": 45, "lr": 0.001} +{"train_loss": 0.05118396133184433, "train_loss_bc": 0.05118396133184433, "train_loss_llm": 0.0, "grad_norm": 0.02459106221795082, "global_step": 1626, "epoch": 45, "lr": 0.001} +{"train_loss": 0.05104384571313858, "train_loss_bc": 0.05101553723216057, "train_loss_llm": 0.0028307735919952393, "grad_norm": 0.03290838375687599, "global_step": 1627, "epoch": 45, "lr": 0.001} +{"train_loss": 0.048880379647016525, "train_loss_bc": 0.048880379647016525, "train_loss_llm": 0.0, "grad_norm": 0.03702418506145477, "global_step": 1628, "epoch": 45, "lr": 0.001} +{"train_loss": 0.0398690328001976, "train_loss_bc": 0.03985121101140976, "train_loss_llm": 0.0017821192741394043, "grad_norm": 0.015354029834270477, "global_step": 1629, "epoch": 45, "lr": 0.001} +{"train_loss": 0.049049925059080124, "train_loss_bc": 0.049049925059080124, "train_loss_llm": 0.0, "grad_norm": 0.026228222995996475, "global_step": 1630, "epoch": 45, "lr": 0.001} +{"train_loss": 0.05071636289358139, "train_loss_bc": 0.050672195851802826, "train_loss_llm": 0.004416525363922119, "grad_norm": 0.034874629229307175, "global_step": 1631, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04127512499690056, "train_loss_bc": 0.04127512499690056, "train_loss_llm": 0.0, "grad_norm": 0.04091547057032585, "global_step": 1632, "epoch": 45, "lr": 0.001} +{"train_loss": 0.041634026914834976, "train_loss_bc": 0.04159841313958168, "train_loss_llm": 0.003561258316040039, "grad_norm": 0.01169856358319521, "global_step": 1633, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04209311306476593, "train_loss_bc": 0.04209311306476593, "train_loss_llm": 0.0, "grad_norm": 0.018260236829519272, "global_step": 1634, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04377327486872673, "train_loss_bc": 0.04377327486872673, "train_loss_llm": 0.0, "grad_norm": 0.024565717205405235, "global_step": 1635, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04178953915834427, "train_loss_bc": 0.041780196130275726, "train_loss_llm": 0.0009342730045318604, "grad_norm": 0.03250898793339729, "global_step": 1636, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04432950168848038, "train_loss_bc": 0.04430577903985977, "train_loss_llm": 0.0023721158504486084, "grad_norm": 0.017678668722510338, "global_step": 1637, "epoch": 45, "lr": 0.001} +{"train_loss": 0.055313799530267715, "train_loss_bc": 0.05527554824948311, "train_loss_llm": 0.003824979066848755, "grad_norm": 0.02927376702427864, "global_step": 1638, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04615478590130806, "train_loss_bc": 0.04615478590130806, "train_loss_llm": 0.0, "grad_norm": 0.03872997313737869, "global_step": 1639, "epoch": 45, "lr": 0.001} +{"train_loss": 0.05375831574201584, "train_loss_bc": 0.05375831574201584, "train_loss_llm": 0.0, "grad_norm": 0.044701073318719864, "global_step": 1640, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04204875975847244, "train_loss_bc": 0.04204875975847244, "train_loss_llm": 0.0, "grad_norm": 0.014282656833529472, "global_step": 1641, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04577789455652237, "train_loss_bc": 0.04577789455652237, "train_loss_llm": 0.0, "grad_norm": 0.023601138964295387, "global_step": 1642, "epoch": 45, "lr": 0.001} +{"train_loss": 0.0512683168053627, "train_loss_bc": 0.0512683168053627, "train_loss_llm": 0.0, "grad_norm": 0.026475993916392326, "global_step": 1643, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04672059044241905, "train_loss_bc": 0.04672059044241905, "train_loss_llm": 0.0, "grad_norm": 0.030860301107168198, "global_step": 1644, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04769572615623474, "train_loss_bc": 0.04769572615623474, "train_loss_llm": 0.0, "grad_norm": 0.017193617299199104, "global_step": 1645, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04355431720614433, "train_loss_bc": 0.04355431720614433, "train_loss_llm": 0.0, "grad_norm": 0.02715473808348179, "global_step": 1646, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04150989279150963, "train_loss_bc": 0.04150989279150963, "train_loss_llm": 0.0, "grad_norm": 0.04026709496974945, "global_step": 1647, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04270375147461891, "train_loss_bc": 0.04270375147461891, "train_loss_llm": 0.0, "grad_norm": 0.045812852680683136, "global_step": 1648, "epoch": 45, "lr": 0.001} +{"train_loss": 0.0466286838054657, "train_loss_bc": 0.0466286838054657, "train_loss_llm": 0.0, "grad_norm": 0.017958147451281548, "global_step": 1649, "epoch": 45, "lr": 0.001} +{"train_loss": 0.047906871885061264, "train_loss_bc": 0.0478997677564621, "train_loss_llm": 0.0007103383541107178, "grad_norm": 0.024236679077148438, "global_step": 1650, "epoch": 45, "lr": 0.001} +{"train_loss": 0.05260820686817169, "train_loss_bc": 0.05260820686817169, "train_loss_llm": 0.0, "grad_norm": 0.03424287587404251, "global_step": 1651, "epoch": 45, "lr": 0.001} +{"train_loss": 0.043662719428539276, "train_loss_bc": 0.043622519820928574, "train_loss_llm": 0.00401991605758667, "grad_norm": 0.03759089857339859, "global_step": 1652, "epoch": 45, "lr": 0.001} +{"train_loss": 0.049741167575120926, "train_loss_bc": 0.049741167575120926, "train_loss_llm": 0.0, "grad_norm": 0.021001853048801422, "global_step": 1653, "epoch": 45, "lr": 0.001} +{"train_loss": 0.05615539476275444, "train_loss_bc": 0.05615539476275444, "train_loss_llm": 0.0, "grad_norm": 0.03368401899933815, "global_step": 1654, "epoch": 45, "lr": 0.001} +{"train_loss": 0.04644832035733594, "train_loss_bc": 0.050268955528736115, "train_loss_llm": 0.0, "grad_norm": 0.04374130442738533, "global_step": 1655, "epoch": 45, "lr": 0.001, "train/cumulative_reward": 52.00238173087284, "train/mean_score": 14.717598778929425, "train/success_rate": 0.4, "test/cumulative_reward": 80.9682383378526, "test/mean_score": 8.46112306296185, "test/success_rate": 0.14, "val_loss": 0.05063735693693161, "train_action_mse_error": 0.05047937482595444} +{"train_loss": 0.048116303980350494, "train_loss_bc": 0.048116303980350494, "train_loss_llm": 0.0, "grad_norm": 0.04591255635023117, "global_step": 1656, "epoch": 46, "lr": 0.001} +{"train_loss": 0.05016133934259415, "train_loss_bc": 0.05014012008905411, "train_loss_llm": 0.0021218955516815186, "grad_norm": 0.016325360164046288, "global_step": 1657, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04966329410672188, "train_loss_bc": 0.04966329410672188, "train_loss_llm": 0.0, "grad_norm": 0.025890490040183067, "global_step": 1658, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04691869765520096, "train_loss_bc": 0.04691869765520096, "train_loss_llm": 0.0, "grad_norm": 0.032473355531692505, "global_step": 1659, "epoch": 46, "lr": 0.001} +{"train_loss": 0.05006863549351692, "train_loss_bc": 0.05006863549351692, "train_loss_llm": 0.0, "grad_norm": 0.039341479539871216, "global_step": 1660, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04639425128698349, "train_loss_bc": 0.04639391601085663, "train_loss_llm": 3.349781036376953e-05, "grad_norm": 0.013933454640209675, "global_step": 1661, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04720262810587883, "train_loss_bc": 0.04720262810587883, "train_loss_llm": 0.0, "grad_norm": 0.02066778391599655, "global_step": 1662, "epoch": 46, "lr": 0.001} +{"train_loss": 0.037125855684280396, "train_loss_bc": 0.037125855684280396, "train_loss_llm": 0.0, "grad_norm": 0.02721899002790451, "global_step": 1663, "epoch": 46, "lr": 0.001} +{"train_loss": 0.05040402710437775, "train_loss_bc": 0.05040402710437775, "train_loss_llm": 0.0, "grad_norm": 0.030862750485539436, "global_step": 1664, "epoch": 46, "lr": 0.001} +{"train_loss": 0.048355475068092346, "train_loss_bc": 0.0483531728386879, "train_loss_llm": 0.0002302229404449463, "grad_norm": 0.014965006150305271, "global_step": 1665, "epoch": 46, "lr": 0.001} +{"train_loss": 0.0508732832968235, "train_loss_bc": 0.0508732832968235, "train_loss_llm": 0.0, "grad_norm": 0.021092398092150688, "global_step": 1666, "epoch": 46, "lr": 0.001} +{"train_loss": 0.040483083575963974, "train_loss_bc": 0.04041517153382301, "train_loss_llm": 0.00679132342338562, "grad_norm": 0.027887122705578804, "global_step": 1667, "epoch": 46, "lr": 0.001} +{"train_loss": 0.046491410583257675, "train_loss_bc": 0.04647579416632652, "train_loss_llm": 0.001561671495437622, "grad_norm": 0.03283992037177086, "global_step": 1668, "epoch": 46, "lr": 0.001} +{"train_loss": 0.041687700897455215, "train_loss_bc": 0.041687700897455215, "train_loss_llm": 0.0, "grad_norm": 0.018594205379486084, "global_step": 1669, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04003864899277687, "train_loss_bc": 0.0400235615670681, "train_loss_llm": 0.0015089213848114014, "grad_norm": 0.025307409465312958, "global_step": 1670, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04310936480760574, "train_loss_bc": 0.04310936480760574, "train_loss_llm": 0.0, "grad_norm": 0.029847191646695137, "global_step": 1671, "epoch": 46, "lr": 0.001} +{"train_loss": 0.05091540515422821, "train_loss_bc": 0.05089706927537918, "train_loss_llm": 0.0018334686756134033, "grad_norm": 0.03352600708603859, "global_step": 1672, "epoch": 46, "lr": 0.001} +{"train_loss": 0.0513051375746727, "train_loss_bc": 0.0513051375746727, "train_loss_llm": 0.0, "grad_norm": 0.017271393910050392, "global_step": 1673, "epoch": 46, "lr": 0.001} +{"train_loss": 0.049008678644895554, "train_loss_bc": 0.0490066297352314, "train_loss_llm": 0.00020486116409301758, "grad_norm": 0.029253419488668442, "global_step": 1674, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04798511043190956, "train_loss_bc": 0.04798511043190956, "train_loss_llm": 0.0, "grad_norm": 0.034928638488054276, "global_step": 1675, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04940048232674599, "train_loss_bc": 0.04940048232674599, "train_loss_llm": 0.0, "grad_norm": 0.037891898304224014, "global_step": 1676, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04432947561144829, "train_loss_bc": 0.04432947561144829, "train_loss_llm": 0.0, "grad_norm": 0.017979659140110016, "global_step": 1677, "epoch": 46, "lr": 0.001} +{"train_loss": 0.053718432784080505, "train_loss_bc": 0.053718432784080505, "train_loss_llm": 0.0, "grad_norm": 0.022961730137467384, "global_step": 1678, "epoch": 46, "lr": 0.001} +{"train_loss": 0.042693257331848145, "train_loss_bc": 0.042644232511520386, "train_loss_llm": 0.004902631044387817, "grad_norm": 0.033407602459192276, "global_step": 1679, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04920649901032448, "train_loss_bc": 0.04920649901032448, "train_loss_llm": 0.0, "grad_norm": 0.04195694625377655, "global_step": 1680, "epoch": 46, "lr": 0.001} +{"train_loss": 0.03875409811735153, "train_loss_bc": 0.03875409811735153, "train_loss_llm": 0.0, "grad_norm": 0.01596752181649208, "global_step": 1681, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04769361391663551, "train_loss_bc": 0.04769361391663551, "train_loss_llm": 0.0, "grad_norm": 0.022050946950912476, "global_step": 1682, "epoch": 46, "lr": 0.001} +{"train_loss": 0.03824050351977348, "train_loss_bc": 0.038211967796087265, "train_loss_llm": 0.0028534233570098877, "grad_norm": 0.02901061624288559, "global_step": 1683, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04201329126954079, "train_loss_bc": 0.04201329126954079, "train_loss_llm": 0.0, "grad_norm": 0.03366317227482796, "global_step": 1684, "epoch": 46, "lr": 0.001} +{"train_loss": 0.05157637223601341, "train_loss_bc": 0.05157637223601341, "train_loss_llm": 0.0, "grad_norm": 0.015820970758795738, "global_step": 1685, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04587411507964134, "train_loss_bc": 0.04586965590715408, "train_loss_llm": 0.00044599175453186035, "grad_norm": 0.032012272626161575, "global_step": 1686, "epoch": 46, "lr": 0.001} +{"train_loss": 0.0434044674038887, "train_loss_bc": 0.04339653253555298, "train_loss_llm": 0.0007935762405395508, "grad_norm": 0.04090011492371559, "global_step": 1687, "epoch": 46, "lr": 0.001} +{"train_loss": 0.052947062999010086, "train_loss_bc": 0.052947062999010086, "train_loss_llm": 0.0, "grad_norm": 0.045946408063173294, "global_step": 1688, "epoch": 46, "lr": 0.001} +{"train_loss": 0.03574591502547264, "train_loss_bc": 0.03574591502547264, "train_loss_llm": 0.0, "grad_norm": 0.014407656155526638, "global_step": 1689, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04744131490588188, "train_loss_bc": 0.0474325530230999, "train_loss_llm": 0.0008761286735534668, "grad_norm": 0.02510974369943142, "global_step": 1690, "epoch": 46, "lr": 0.001} +{"train_loss": 0.04618087907632192, "train_loss_bc": 0.0431535430252552, "train_loss_llm": 0.0010871589183807373, "grad_norm": 0.03210731968283653, "global_step": 1691, "epoch": 46, "lr": 0.001, "val_loss": 0.05045459792017937} +{"train_loss": 0.04349002614617348, "train_loss_bc": 0.043466199189424515, "train_loss_llm": 0.002382516860961914, "grad_norm": 0.037341438233852386, "global_step": 1692, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04430452361702919, "train_loss_bc": 0.04430452361702919, "train_loss_llm": 0.0, "grad_norm": 0.015304001048207283, "global_step": 1693, "epoch": 47, "lr": 0.001} +{"train_loss": 0.042296234518289566, "train_loss_bc": 0.04228799417614937, "train_loss_llm": 0.0008240044116973877, "grad_norm": 0.022152213379740715, "global_step": 1694, "epoch": 47, "lr": 0.001} +{"train_loss": 0.0513131320476532, "train_loss_bc": 0.0513131320476532, "train_loss_llm": 0.0, "grad_norm": 0.033675260841846466, "global_step": 1695, "epoch": 47, "lr": 0.001} +{"train_loss": 0.05087599903345108, "train_loss_bc": 0.05087599903345108, "train_loss_llm": 0.0, "grad_norm": 0.03816961869597435, "global_step": 1696, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04526497796177864, "train_loss_bc": 0.04526497796177864, "train_loss_llm": 0.0, "grad_norm": 0.016222145408391953, "global_step": 1697, "epoch": 47, "lr": 0.001} +{"train_loss": 0.039841216057538986, "train_loss_bc": 0.039841216057538986, "train_loss_llm": 0.0, "grad_norm": 0.023167487233877182, "global_step": 1698, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04298339784145355, "train_loss_bc": 0.0429515540599823, "train_loss_llm": 0.003184378147125244, "grad_norm": 0.03215877711772919, "global_step": 1699, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04558530077338219, "train_loss_bc": 0.04558530077338219, "train_loss_llm": 0.0, "grad_norm": 0.03492066264152527, "global_step": 1700, "epoch": 47, "lr": 0.001} +{"train_loss": 0.041753143072128296, "train_loss_bc": 0.041753143072128296, "train_loss_llm": 0.0, "grad_norm": 0.017284344881772995, "global_step": 1701, "epoch": 47, "lr": 0.001} +{"train_loss": 0.043462205678224564, "train_loss_bc": 0.043462205678224564, "train_loss_llm": 0.0, "grad_norm": 0.024204542860388756, "global_step": 1702, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04775499925017357, "train_loss_bc": 0.04775499925017357, "train_loss_llm": 0.0, "grad_norm": 0.032741643488407135, "global_step": 1703, "epoch": 47, "lr": 0.001} +{"train_loss": 0.037722136825323105, "train_loss_bc": 0.03771420940756798, "train_loss_llm": 0.0007925629615783691, "grad_norm": 0.03766074776649475, "global_step": 1704, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04313140735030174, "train_loss_bc": 0.04311464726924896, "train_loss_llm": 0.0016761720180511475, "grad_norm": 0.01569540426135063, "global_step": 1705, "epoch": 47, "lr": 0.001} +{"train_loss": 0.049703504890203476, "train_loss_bc": 0.049703504890203476, "train_loss_llm": 0.0, "grad_norm": 0.025831755250692368, "global_step": 1706, "epoch": 47, "lr": 0.001} +{"train_loss": 0.041616782546043396, "train_loss_bc": 0.041613683104515076, "train_loss_llm": 0.00030991435050964355, "grad_norm": 0.03344135358929634, "global_step": 1707, "epoch": 47, "lr": 0.001} +{"train_loss": 0.05262456834316254, "train_loss_bc": 0.05262456834316254, "train_loss_llm": 0.0, "grad_norm": 0.03753691539168358, "global_step": 1708, "epoch": 47, "lr": 0.001} +{"train_loss": 0.047756128013134, "train_loss_bc": 0.04775308072566986, "train_loss_llm": 0.00030490756034851074, "grad_norm": 0.01831776835024357, "global_step": 1709, "epoch": 47, "lr": 0.001} +{"train_loss": 0.0525185652077198, "train_loss_bc": 0.0524984709918499, "train_loss_llm": 0.002009272575378418, "grad_norm": 0.022354716435074806, "global_step": 1710, "epoch": 47, "lr": 0.001} +{"train_loss": 0.033403120934963226, "train_loss_bc": 0.03339946269989014, "train_loss_llm": 0.00036579370498657227, "grad_norm": 0.028731124475598335, "global_step": 1711, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04464718699455261, "train_loss_bc": 0.04464718699455261, "train_loss_llm": 0.0, "grad_norm": 0.03148572891950607, "global_step": 1712, "epoch": 47, "lr": 0.001} +{"train_loss": 0.044107913970947266, "train_loss_bc": 0.044107913970947266, "train_loss_llm": 0.0, "grad_norm": 0.015235393308103085, "global_step": 1713, "epoch": 47, "lr": 0.001} +{"train_loss": 0.043938200920820236, "train_loss_bc": 0.043938200920820236, "train_loss_llm": 0.0, "grad_norm": 0.020658330991864204, "global_step": 1714, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04422597587108612, "train_loss_bc": 0.04422440752387047, "train_loss_llm": 0.00015673041343688965, "grad_norm": 0.026715556159615517, "global_step": 1715, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04762646555900574, "train_loss_bc": 0.04761625453829765, "train_loss_llm": 0.0010211467742919922, "grad_norm": 0.030949262902140617, "global_step": 1716, "epoch": 47, "lr": 0.001} +{"train_loss": 0.045097578316926956, "train_loss_bc": 0.04502427950501442, "train_loss_llm": 0.007330000400543213, "grad_norm": 0.01366552896797657, "global_step": 1717, "epoch": 47, "lr": 0.001} +{"train_loss": 0.05274656042456627, "train_loss_bc": 0.05272623896598816, "train_loss_llm": 0.002032041549682617, "grad_norm": 0.02419009432196617, "global_step": 1718, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04504834860563278, "train_loss_bc": 0.04501791298389435, "train_loss_llm": 0.0030435919761657715, "grad_norm": 0.03255132958292961, "global_step": 1719, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04842415079474449, "train_loss_bc": 0.04842415079474449, "train_loss_llm": 0.0, "grad_norm": 0.03986498713493347, "global_step": 1720, "epoch": 47, "lr": 0.001} +{"train_loss": 0.056861769407987595, "train_loss_bc": 0.056861769407987595, "train_loss_llm": 0.0, "grad_norm": 0.016988754272460938, "global_step": 1721, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04754693806171417, "train_loss_bc": 0.04754693806171417, "train_loss_llm": 0.0, "grad_norm": 0.025103671476244926, "global_step": 1722, "epoch": 47, "lr": 0.001} +{"train_loss": 0.051483895629644394, "train_loss_bc": 0.051483895629644394, "train_loss_llm": 0.0, "grad_norm": 0.029401440173387527, "global_step": 1723, "epoch": 47, "lr": 0.001} +{"train_loss": 0.05084041506052017, "train_loss_bc": 0.050818152725696564, "train_loss_llm": 0.0022260546684265137, "grad_norm": 0.03633074834942818, "global_step": 1724, "epoch": 47, "lr": 0.001} +{"train_loss": 0.05012115463614464, "train_loss_bc": 0.05009494721889496, "train_loss_llm": 0.0026207566261291504, "grad_norm": 0.017964214086532593, "global_step": 1725, "epoch": 47, "lr": 0.001} +{"train_loss": 0.04620247706770897, "train_loss_bc": 0.04620247706770897, "train_loss_llm": 0.0, "grad_norm": 0.025304941460490227, "global_step": 1726, "epoch": 47, "lr": 0.001} +{"train_loss": 0.046186921290225454, "train_loss_bc": 0.0464087650179863, "train_loss_llm": 0.0, "grad_norm": 0.04199793189764023, "global_step": 1727, "epoch": 47, "lr": 0.001, "val_loss": 0.050545357167720795} +{"train_loss": 0.045990657061338425, "train_loss_bc": 0.045990657061338425, "train_loss_llm": 0.0, "grad_norm": 0.05129946395754814, "global_step": 1728, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04761047661304474, "train_loss_bc": 0.04758738353848457, "train_loss_llm": 0.0023092329502105713, "grad_norm": 0.018442587926983833, "global_step": 1729, "epoch": 48, "lr": 0.001} +{"train_loss": 0.05466753616929054, "train_loss_bc": 0.0546664334833622, "train_loss_llm": 0.00011008977890014648, "grad_norm": 0.02462266944348812, "global_step": 1730, "epoch": 48, "lr": 0.001} +{"train_loss": 0.040876518934965134, "train_loss_bc": 0.040876518934965134, "train_loss_llm": 0.0, "grad_norm": 0.026648910716176033, "global_step": 1731, "epoch": 48, "lr": 0.001} +{"train_loss": 0.0427311509847641, "train_loss_bc": 0.042725834995508194, "train_loss_llm": 0.0005315542221069336, "grad_norm": 0.02780134230852127, "global_step": 1732, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04647750407457352, "train_loss_bc": 0.04645371064543724, "train_loss_llm": 0.0023793280124664307, "grad_norm": 0.015873344615101814, "global_step": 1733, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04343515634536743, "train_loss_bc": 0.04343515634536743, "train_loss_llm": 0.0, "grad_norm": 0.02455298788845539, "global_step": 1734, "epoch": 48, "lr": 0.001} +{"train_loss": 0.03939182311296463, "train_loss_bc": 0.03939182311296463, "train_loss_llm": 0.0, "grad_norm": 0.03149306774139404, "global_step": 1735, "epoch": 48, "lr": 0.001} +{"train_loss": 0.05008814483880997, "train_loss_bc": 0.05008814483880997, "train_loss_llm": 0.0, "grad_norm": 0.03926997259259224, "global_step": 1736, "epoch": 48, "lr": 0.001} +{"train_loss": 0.05028524249792099, "train_loss_bc": 0.050259921699762344, "train_loss_llm": 0.0025319457054138184, "grad_norm": 0.017098505049943924, "global_step": 1737, "epoch": 48, "lr": 0.001} +{"train_loss": 0.05221434310078621, "train_loss_bc": 0.05216830223798752, "train_loss_llm": 0.004604071378707886, "grad_norm": 0.024675248190760612, "global_step": 1738, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04507818445563316, "train_loss_bc": 0.04507818445563316, "train_loss_llm": 0.0, "grad_norm": 0.026671383529901505, "global_step": 1739, "epoch": 48, "lr": 0.001} +{"train_loss": 0.05355638265609741, "train_loss_bc": 0.05355638265609741, "train_loss_llm": 0.0, "grad_norm": 0.031534627079963684, "global_step": 1740, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04196276143193245, "train_loss_bc": 0.041955336928367615, "train_loss_llm": 0.0007424056529998779, "grad_norm": 0.016990141943097115, "global_step": 1741, "epoch": 48, "lr": 0.001} +{"train_loss": 0.046242740005254745, "train_loss_bc": 0.0462162159383297, "train_loss_llm": 0.0026522576808929443, "grad_norm": 0.023656798526644707, "global_step": 1742, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04472813382744789, "train_loss_bc": 0.04472813382744789, "train_loss_llm": 0.0, "grad_norm": 0.027517160400748253, "global_step": 1743, "epoch": 48, "lr": 0.001} +{"train_loss": 0.0495380200445652, "train_loss_bc": 0.0495380200445652, "train_loss_llm": 0.0, "grad_norm": 0.035714030265808105, "global_step": 1744, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04501541703939438, "train_loss_bc": 0.04501541703939438, "train_loss_llm": 0.0, "grad_norm": 0.01655830256640911, "global_step": 1745, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04269210621714592, "train_loss_bc": 0.04269210621714592, "train_loss_llm": 0.0, "grad_norm": 0.022388340905308723, "global_step": 1746, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04898124933242798, "train_loss_bc": 0.04897705093026161, "train_loss_llm": 0.00041979551315307617, "grad_norm": 0.027181869372725487, "global_step": 1747, "epoch": 48, "lr": 0.001} +{"train_loss": 0.05191494897007942, "train_loss_bc": 0.05191494897007942, "train_loss_llm": 0.0, "grad_norm": 0.029472114518284798, "global_step": 1748, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04815545305609703, "train_loss_bc": 0.04815545305609703, "train_loss_llm": 0.0, "grad_norm": 0.014318936504423618, "global_step": 1749, "epoch": 48, "lr": 0.001} +{"train_loss": 0.03947754204273224, "train_loss_bc": 0.03947754204273224, "train_loss_llm": 0.0, "grad_norm": 0.023801403120160103, "global_step": 1750, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04951786994934082, "train_loss_bc": 0.04951322451233864, "train_loss_llm": 0.0004646778106689453, "grad_norm": 0.029929539188742638, "global_step": 1751, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04833966866135597, "train_loss_bc": 0.04833966866135597, "train_loss_llm": 0.0, "grad_norm": 0.038357239216566086, "global_step": 1752, "epoch": 48, "lr": 0.001} +{"train_loss": 0.040791045874357224, "train_loss_bc": 0.040791045874357224, "train_loss_llm": 0.0, "grad_norm": 0.015521937981247902, "global_step": 1753, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04273800924420357, "train_loss_bc": 0.04273800924420357, "train_loss_llm": 0.0, "grad_norm": 0.024158909916877747, "global_step": 1754, "epoch": 48, "lr": 0.001} +{"train_loss": 0.042907215654850006, "train_loss_bc": 0.042896371334791183, "train_loss_llm": 0.0010845363140106201, "grad_norm": 0.03167096525430679, "global_step": 1755, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04870276525616646, "train_loss_bc": 0.04870276525616646, "train_loss_llm": 0.0, "grad_norm": 0.03398461639881134, "global_step": 1756, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04499570652842522, "train_loss_bc": 0.04497581720352173, "train_loss_llm": 0.0019887685775756836, "grad_norm": 0.014126214198768139, "global_step": 1757, "epoch": 48, "lr": 0.001} +{"train_loss": 0.052383847534656525, "train_loss_bc": 0.052383847534656525, "train_loss_llm": 0.0, "grad_norm": 0.026003839448094368, "global_step": 1758, "epoch": 48, "lr": 0.001} +{"train_loss": 0.03910158947110176, "train_loss_bc": 0.03910158947110176, "train_loss_llm": 0.0, "grad_norm": 0.031874656677246094, "global_step": 1759, "epoch": 48, "lr": 0.001} +{"train_loss": 0.046807754784822464, "train_loss_bc": 0.04677269607782364, "train_loss_llm": 0.003506004810333252, "grad_norm": 0.03709651157259941, "global_step": 1760, "epoch": 48, "lr": 0.001} +{"train_loss": 0.03862236067652702, "train_loss_bc": 0.038615114986896515, "train_loss_llm": 0.00072440505027771, "grad_norm": 0.01351760420948267, "global_step": 1761, "epoch": 48, "lr": 0.001} +{"train_loss": 0.04945078119635582, "train_loss_bc": 0.049442682415246964, "train_loss_llm": 0.0008097589015960693, "grad_norm": 0.020294398069381714, "global_step": 1762, "epoch": 48, "lr": 0.001} +{"train_loss": 0.046048425862358675, "train_loss_bc": 0.04227322340011597, "train_loss_llm": 0.0, "grad_norm": 0.03111240826547146, "global_step": 1763, "epoch": 48, "lr": 0.001, "val_loss": 0.050580937415361404} +{"train_loss": 0.04685591906309128, "train_loss_bc": 0.04684862866997719, "train_loss_llm": 0.000728905200958252, "grad_norm": 0.03687826171517372, "global_step": 1764, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04332670941948891, "train_loss_bc": 0.04332670941948891, "train_loss_llm": 0.0, "grad_norm": 0.01861628144979477, "global_step": 1765, "epoch": 49, "lr": 0.001} +{"train_loss": 0.05741447955369949, "train_loss_bc": 0.05741447955369949, "train_loss_llm": 0.0, "grad_norm": 0.03102109394967556, "global_step": 1766, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04108578339219093, "train_loss_bc": 0.04106995463371277, "train_loss_llm": 0.0015829801559448242, "grad_norm": 0.03370469808578491, "global_step": 1767, "epoch": 49, "lr": 0.001} +{"train_loss": 0.042008817195892334, "train_loss_bc": 0.042008817195892334, "train_loss_llm": 0.0, "grad_norm": 0.03366491198539734, "global_step": 1768, "epoch": 49, "lr": 0.001} +{"train_loss": 0.050348419696092606, "train_loss_bc": 0.050348419696092606, "train_loss_llm": 0.0, "grad_norm": 0.017530187964439392, "global_step": 1769, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04099876433610916, "train_loss_bc": 0.04098973423242569, "train_loss_llm": 0.0009029507637023926, "grad_norm": 0.02484436146914959, "global_step": 1770, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04772709310054779, "train_loss_bc": 0.047703903168439865, "train_loss_llm": 0.0023188889026641846, "grad_norm": 0.02939377911388874, "global_step": 1771, "epoch": 49, "lr": 0.001} +{"train_loss": 0.03785848990082741, "train_loss_bc": 0.03785848990082741, "train_loss_llm": 0.0, "grad_norm": 0.03191345930099487, "global_step": 1772, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04958883300423622, "train_loss_bc": 0.04958883300423622, "train_loss_llm": 0.0, "grad_norm": 0.017480071634054184, "global_step": 1773, "epoch": 49, "lr": 0.001} +{"train_loss": 0.047949861735105515, "train_loss_bc": 0.047947704792022705, "train_loss_llm": 0.00021573901176452637, "grad_norm": 0.022757992148399353, "global_step": 1774, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04653047397732735, "train_loss_bc": 0.04650473967194557, "train_loss_llm": 0.002573549747467041, "grad_norm": 0.03171461820602417, "global_step": 1775, "epoch": 49, "lr": 0.001} +{"train_loss": 0.052286289632320404, "train_loss_bc": 0.05226821079850197, "train_loss_llm": 0.0018078982830047607, "grad_norm": 0.036071497946977615, "global_step": 1776, "epoch": 49, "lr": 0.001} +{"train_loss": 0.03738277032971382, "train_loss_bc": 0.03738277032971382, "train_loss_llm": 0.0, "grad_norm": 0.013734528794884682, "global_step": 1777, "epoch": 49, "lr": 0.001} +{"train_loss": 0.055672433227300644, "train_loss_bc": 0.055672433227300644, "train_loss_llm": 0.0, "grad_norm": 0.026995429769158363, "global_step": 1778, "epoch": 49, "lr": 0.001} +{"train_loss": 0.036671336740255356, "train_loss_bc": 0.036671336740255356, "train_loss_llm": 0.0, "grad_norm": 0.027519138529896736, "global_step": 1779, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04000215232372284, "train_loss_bc": 0.0400010384619236, "train_loss_llm": 0.00011131167411804199, "grad_norm": 0.029684120789170265, "global_step": 1780, "epoch": 49, "lr": 0.001} +{"train_loss": 0.0494096539914608, "train_loss_bc": 0.0494096539914608, "train_loss_llm": 0.0, "grad_norm": 0.01826779916882515, "global_step": 1781, "epoch": 49, "lr": 0.001} +{"train_loss": 0.046463120728731155, "train_loss_bc": 0.04644477739930153, "train_loss_llm": 0.0018343031406402588, "grad_norm": 0.027295157313346863, "global_step": 1782, "epoch": 49, "lr": 0.001} +{"train_loss": 0.05501978471875191, "train_loss_bc": 0.05501978471875191, "train_loss_llm": 0.0, "grad_norm": 0.030996985733509064, "global_step": 1783, "epoch": 49, "lr": 0.001} +{"train_loss": 0.049089618027210236, "train_loss_bc": 0.04906372353434563, "train_loss_llm": 0.0025895535945892334, "grad_norm": 0.03491901978850365, "global_step": 1784, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04776213318109512, "train_loss_bc": 0.04776213318109512, "train_loss_llm": 0.0, "grad_norm": 0.016754938289523125, "global_step": 1785, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04262251406908035, "train_loss_bc": 0.04261937737464905, "train_loss_llm": 0.0003136098384857178, "grad_norm": 0.0242401584982872, "global_step": 1786, "epoch": 49, "lr": 0.001} +{"train_loss": 0.045500241219997406, "train_loss_bc": 0.045500241219997406, "train_loss_llm": 0.0, "grad_norm": 0.027748897671699524, "global_step": 1787, "epoch": 49, "lr": 0.001} +{"train_loss": 0.05201949551701546, "train_loss_bc": 0.05200507864356041, "train_loss_llm": 0.0014416277408599854, "grad_norm": 0.034492023289203644, "global_step": 1788, "epoch": 49, "lr": 0.001} +{"train_loss": 0.044278211891651154, "train_loss_bc": 0.044253282248973846, "train_loss_llm": 0.002493053674697876, "grad_norm": 0.015787361189723015, "global_step": 1789, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04546931013464928, "train_loss_bc": 0.045461129397153854, "train_loss_llm": 0.0008181929588317871, "grad_norm": 0.027564499527215958, "global_step": 1790, "epoch": 49, "lr": 0.001} +{"train_loss": 0.03911424055695534, "train_loss_bc": 0.03911424055695534, "train_loss_llm": 0.0, "grad_norm": 0.03803243488073349, "global_step": 1791, "epoch": 49, "lr": 0.001} +{"train_loss": 0.0401918850839138, "train_loss_bc": 0.04016875475645065, "train_loss_llm": 0.002312898635864258, "grad_norm": 0.04022606834769249, "global_step": 1792, "epoch": 49, "lr": 0.001} +{"train_loss": 0.05345639958977699, "train_loss_bc": 0.05343898758292198, "train_loss_llm": 0.0017411112785339355, "grad_norm": 0.014393633231520653, "global_step": 1793, "epoch": 49, "lr": 0.001} +{"train_loss": 0.045466791838407516, "train_loss_bc": 0.045466791838407516, "train_loss_llm": 0.0, "grad_norm": 0.024248413741588593, "global_step": 1794, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04400797188282013, "train_loss_bc": 0.04400797188282013, "train_loss_llm": 0.0, "grad_norm": 0.03224683925509453, "global_step": 1795, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04901877045631409, "train_loss_bc": 0.04900055751204491, "train_loss_llm": 0.0018212199211120605, "grad_norm": 0.04051325097680092, "global_step": 1796, "epoch": 49, "lr": 0.001} +{"train_loss": 0.043366994708776474, "train_loss_bc": 0.04336327686905861, "train_loss_llm": 0.000371783971786499, "grad_norm": 0.01625644788146019, "global_step": 1797, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04788878187537193, "train_loss_bc": 0.04788878187537193, "train_loss_llm": 0.0, "grad_norm": 0.025720397010445595, "global_step": 1798, "epoch": 49, "lr": 0.001} +{"train_loss": 0.04599144351151255, "train_loss_bc": 0.041836559772491455, "train_loss_llm": 8.594989776611328e-05, "grad_norm": 0.038540318608284, "global_step": 1799, "epoch": 49, "lr": 0.001, "val_loss": 0.05073088780045509} +{"train_loss": 0.04350930452346802, "train_loss_bc": 0.04346403852105141, "train_loss_llm": 0.004526585340499878, "grad_norm": 0.047196175903081894, "global_step": 1800, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04783197119832039, "train_loss_bc": 0.04779364913702011, "train_loss_llm": 0.0038323700428009033, "grad_norm": 0.017301512882113457, "global_step": 1801, "epoch": 50, "lr": 0.001} +{"train_loss": 0.045875150710344315, "train_loss_bc": 0.045875150710344315, "train_loss_llm": 0.0, "grad_norm": 0.022304143756628036, "global_step": 1802, "epoch": 50, "lr": 0.001} +{"train_loss": 0.039816223084926605, "train_loss_bc": 0.039816223084926605, "train_loss_llm": 0.0, "grad_norm": 0.028914976865053177, "global_step": 1803, "epoch": 50, "lr": 0.001} +{"train_loss": 0.03933265805244446, "train_loss_bc": 0.03931459039449692, "train_loss_llm": 0.0018066167831420898, "grad_norm": 0.036133065819740295, "global_step": 1804, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04613093286752701, "train_loss_bc": 0.04613093286752701, "train_loss_llm": 0.0, "grad_norm": 0.013068152591586113, "global_step": 1805, "epoch": 50, "lr": 0.001} +{"train_loss": 0.03888023644685745, "train_loss_bc": 0.03887341916561127, "train_loss_llm": 0.0006817281246185303, "grad_norm": 0.02166866697371006, "global_step": 1806, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04124461114406586, "train_loss_bc": 0.04124461114406586, "train_loss_llm": 0.0, "grad_norm": 0.02972474694252014, "global_step": 1807, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04954564943909645, "train_loss_bc": 0.04954564943909645, "train_loss_llm": 0.0, "grad_norm": 0.035933420062065125, "global_step": 1808, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04187518730759621, "train_loss_bc": 0.0418723002076149, "train_loss_llm": 0.000288546085357666, "grad_norm": 0.012336519546806812, "global_step": 1809, "epoch": 50, "lr": 0.001} +{"train_loss": 0.05279142037034035, "train_loss_bc": 0.05279142037034035, "train_loss_llm": 0.0, "grad_norm": 0.01775292493402958, "global_step": 1810, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04645584896206856, "train_loss_bc": 0.04644522815942764, "train_loss_llm": 0.0010621845722198486, "grad_norm": 0.02295362576842308, "global_step": 1811, "epoch": 50, "lr": 0.001} +{"train_loss": 0.046780358999967575, "train_loss_bc": 0.046779658645391464, "train_loss_llm": 7.000565528869629e-05, "grad_norm": 0.02941950410604477, "global_step": 1812, "epoch": 50, "lr": 0.001} +{"train_loss": 0.03616621717810631, "train_loss_bc": 0.03615635260939598, "train_loss_llm": 0.000986337661743164, "grad_norm": 0.012793160043656826, "global_step": 1813, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04791032150387764, "train_loss_bc": 0.04787706956267357, "train_loss_llm": 0.003325134515762329, "grad_norm": 0.0202939435839653, "global_step": 1814, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04265587404370308, "train_loss_bc": 0.04265587404370308, "train_loss_llm": 0.0, "grad_norm": 0.025150207802653313, "global_step": 1815, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04853837937116623, "train_loss_bc": 0.048529528081417084, "train_loss_llm": 0.0008850991725921631, "grad_norm": 0.02995344065129757, "global_step": 1816, "epoch": 50, "lr": 0.001} +{"train_loss": 0.05010059103369713, "train_loss_bc": 0.05010059103369713, "train_loss_llm": 0.0, "grad_norm": 0.021401412785053253, "global_step": 1817, "epoch": 50, "lr": 0.001} +{"train_loss": 0.05072233825922012, "train_loss_bc": 0.05072233825922012, "train_loss_llm": 0.0, "grad_norm": 0.026747969910502434, "global_step": 1818, "epoch": 50, "lr": 0.001} +{"train_loss": 0.047223251312971115, "train_loss_bc": 0.047223251312971115, "train_loss_llm": 0.0, "grad_norm": 0.031930480152368546, "global_step": 1819, "epoch": 50, "lr": 0.001} +{"train_loss": 0.057329241186380386, "train_loss_bc": 0.057329241186380386, "train_loss_llm": 0.0, "grad_norm": 0.04132894054055214, "global_step": 1820, "epoch": 50, "lr": 0.001} +{"train_loss": 0.03965049237012863, "train_loss_bc": 0.03965049237012863, "train_loss_llm": 0.0, "grad_norm": 0.016927426680922508, "global_step": 1821, "epoch": 50, "lr": 0.001} +{"train_loss": 0.05113431066274643, "train_loss_bc": 0.05113431066274643, "train_loss_llm": 0.0, "grad_norm": 0.029796840623021126, "global_step": 1822, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04528377205133438, "train_loss_bc": 0.04527663812041283, "train_loss_llm": 0.0007132887840270996, "grad_norm": 0.041230637580156326, "global_step": 1823, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04912499338388443, "train_loss_bc": 0.04912499338388443, "train_loss_llm": 0.0, "grad_norm": 0.047584690153598785, "global_step": 1824, "epoch": 50, "lr": 0.001} +{"train_loss": 0.042477671056985855, "train_loss_bc": 0.04247429594397545, "train_loss_llm": 0.00033739209175109863, "grad_norm": 0.015994593501091003, "global_step": 1825, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04829515889286995, "train_loss_bc": 0.04829515889286995, "train_loss_llm": 0.0, "grad_norm": 0.020219597965478897, "global_step": 1826, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04279017820954323, "train_loss_bc": 0.04277185723185539, "train_loss_llm": 0.0018320083618164062, "grad_norm": 0.029111938551068306, "global_step": 1827, "epoch": 50, "lr": 0.001} +{"train_loss": 0.05047345533967018, "train_loss_bc": 0.050422754138708115, "train_loss_llm": 0.005070030689239502, "grad_norm": 0.03747190535068512, "global_step": 1828, "epoch": 50, "lr": 0.001} +{"train_loss": 0.05488530173897743, "train_loss_bc": 0.05488530173897743, "train_loss_llm": 0.0, "grad_norm": 0.019261177629232407, "global_step": 1829, "epoch": 50, "lr": 0.001} +{"train_loss": 0.03982245549559593, "train_loss_bc": 0.03981426730751991, "train_loss_llm": 0.0008188486099243164, "grad_norm": 0.026526039466261864, "global_step": 1830, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04399440437555313, "train_loss_bc": 0.04399440437555313, "train_loss_llm": 0.0, "grad_norm": 0.03312679007649422, "global_step": 1831, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04807329550385475, "train_loss_bc": 0.048066895455121994, "train_loss_llm": 0.0006399154663085938, "grad_norm": 0.04040776193141937, "global_step": 1832, "epoch": 50, "lr": 0.001} +{"train_loss": 0.046070802956819534, "train_loss_bc": 0.046070802956819534, "train_loss_llm": 0.0, "grad_norm": 0.01596589758992195, "global_step": 1833, "epoch": 50, "lr": 0.001} +{"train_loss": 0.04583664983510971, "train_loss_bc": 0.04583664983510971, "train_loss_llm": 0.0, "grad_norm": 0.02274254523217678, "global_step": 1834, "epoch": 50, "lr": 0.001} +{"train_loss": 0.045959285356932215, "train_loss_bc": 0.04590556398034096, "train_loss_llm": 0.0, "grad_norm": 0.029063338413834572, "global_step": 1835, "epoch": 50, "lr": 0.001, "train/cumulative_reward": 95.11066986649105, "train/mean_score": 5.540110964683924, "train/success_rate": 0.0, "test/cumulative_reward": 44.943042500190174, "test/mean_score": 5.7288809438250725, "test/success_rate": 0.1, "val_loss": 0.05064410716295242, "train_action_mse_error": 0.05012989044189453} +{"train_loss": 0.04552660137414932, "train_loss_bc": 0.04552660137414932, "train_loss_llm": 0.0, "grad_norm": 0.0329180471599102, "global_step": 1836, "epoch": 51, "lr": 0.001} +{"train_loss": 0.050573550164699554, "train_loss_bc": 0.050573550164699554, "train_loss_llm": 0.0, "grad_norm": 0.01858924888074398, "global_step": 1837, "epoch": 51, "lr": 0.001} +{"train_loss": 0.05083237960934639, "train_loss_bc": 0.05082681030035019, "train_loss_llm": 0.0005568861961364746, "grad_norm": 0.025599393993616104, "global_step": 1838, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04259316623210907, "train_loss_bc": 0.042570482939481735, "train_loss_llm": 0.0022684037685394287, "grad_norm": 0.02747948467731476, "global_step": 1839, "epoch": 51, "lr": 0.001} +{"train_loss": 0.049271952360868454, "train_loss_bc": 0.04926799237728119, "train_loss_llm": 0.0003961026668548584, "grad_norm": 0.031398139894008636, "global_step": 1840, "epoch": 51, "lr": 0.001} +{"train_loss": 0.05105794593691826, "train_loss_bc": 0.0510387122631073, "train_loss_llm": 0.001923292875289917, "grad_norm": 0.016841573640704155, "global_step": 1841, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04338734596967697, "train_loss_bc": 0.04338734596967697, "train_loss_llm": 0.0, "grad_norm": 0.024876676499843597, "global_step": 1842, "epoch": 51, "lr": 0.001} +{"train_loss": 0.03680393844842911, "train_loss_bc": 0.03680393844842911, "train_loss_llm": 0.0, "grad_norm": 0.02846301719546318, "global_step": 1843, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04456290602684021, "train_loss_bc": 0.04456290602684021, "train_loss_llm": 0.0, "grad_norm": 0.02938118577003479, "global_step": 1844, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04762301221489906, "train_loss_bc": 0.04762301221489906, "train_loss_llm": 0.0, "grad_norm": 0.015881473198533058, "global_step": 1845, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04758581891655922, "train_loss_bc": 0.04758581891655922, "train_loss_llm": 0.0, "grad_norm": 0.025866178795695305, "global_step": 1846, "epoch": 51, "lr": 0.001} +{"train_loss": 0.044697970151901245, "train_loss_bc": 0.044697970151901245, "train_loss_llm": 0.0, "grad_norm": 0.0319472998380661, "global_step": 1847, "epoch": 51, "lr": 0.001} +{"train_loss": 0.043292637914419174, "train_loss_bc": 0.04329149052500725, "train_loss_llm": 0.00011491775512695312, "grad_norm": 0.0364115908741951, "global_step": 1848, "epoch": 51, "lr": 0.001} +{"train_loss": 0.048773180693387985, "train_loss_bc": 0.04873325303196907, "train_loss_llm": 0.003992676734924316, "grad_norm": 0.016986163333058357, "global_step": 1849, "epoch": 51, "lr": 0.001} +{"train_loss": 0.0504501610994339, "train_loss_bc": 0.0504501610994339, "train_loss_llm": 0.0, "grad_norm": 0.02507304586470127, "global_step": 1850, "epoch": 51, "lr": 0.001} +{"train_loss": 0.045184724032878876, "train_loss_bc": 0.045184724032878876, "train_loss_llm": 0.0, "grad_norm": 0.029486626386642456, "global_step": 1851, "epoch": 51, "lr": 0.001} +{"train_loss": 0.043007008731365204, "train_loss_bc": 0.042982958257198334, "train_loss_llm": 0.0024051666259765625, "grad_norm": 0.03322337940335274, "global_step": 1852, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04102877527475357, "train_loss_bc": 0.04101927578449249, "train_loss_llm": 0.0009498000144958496, "grad_norm": 0.015325334854424, "global_step": 1853, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04424973949790001, "train_loss_bc": 0.04424918070435524, "train_loss_llm": 5.605816841125488e-05, "grad_norm": 0.023451831191778183, "global_step": 1854, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04917003959417343, "train_loss_bc": 0.049159709364175797, "train_loss_llm": 0.0010328888893127441, "grad_norm": 0.03009694442152977, "global_step": 1855, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04028749838471413, "train_loss_bc": 0.04027887061238289, "train_loss_llm": 0.0008627176284790039, "grad_norm": 0.034761302173137665, "global_step": 1856, "epoch": 51, "lr": 0.001} +{"train_loss": 0.043201956897974014, "train_loss_bc": 0.043201956897974014, "train_loss_llm": 0.0, "grad_norm": 0.016942022368311882, "global_step": 1857, "epoch": 51, "lr": 0.001} +{"train_loss": 0.050276026129722595, "train_loss_bc": 0.050276026129722595, "train_loss_llm": 0.0, "grad_norm": 0.022218020632863045, "global_step": 1858, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04500667005777359, "train_loss_bc": 0.044980887323617935, "train_loss_llm": 0.0025784075260162354, "grad_norm": 0.025501776486635208, "global_step": 1859, "epoch": 51, "lr": 0.001} +{"train_loss": 0.036901939660310745, "train_loss_bc": 0.03689185157418251, "train_loss_llm": 0.0010088086128234863, "grad_norm": 0.03041643463075161, "global_step": 1860, "epoch": 51, "lr": 0.001} +{"train_loss": 0.03910990059375763, "train_loss_bc": 0.03910990059375763, "train_loss_llm": 0.0, "grad_norm": 0.01560545526444912, "global_step": 1861, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04540069401264191, "train_loss_bc": 0.04540069401264191, "train_loss_llm": 0.0, "grad_norm": 0.02588796429336071, "global_step": 1862, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04507974162697792, "train_loss_bc": 0.04507974162697792, "train_loss_llm": 0.0, "grad_norm": 0.03186768665909767, "global_step": 1863, "epoch": 51, "lr": 0.001} +{"train_loss": 0.03839685395359993, "train_loss_bc": 0.03839685395359993, "train_loss_llm": 0.0, "grad_norm": 0.03399535268545151, "global_step": 1864, "epoch": 51, "lr": 0.001} +{"train_loss": 0.047220081090927124, "train_loss_bc": 0.0472073033452034, "train_loss_llm": 0.0012777447700500488, "grad_norm": 0.015879850834608078, "global_step": 1865, "epoch": 51, "lr": 0.001} +{"train_loss": 0.047925326973199844, "train_loss_bc": 0.0479230061173439, "train_loss_llm": 0.00023216009140014648, "grad_norm": 0.022245580330491066, "global_step": 1866, "epoch": 51, "lr": 0.001} +{"train_loss": 0.055861055850982666, "train_loss_bc": 0.055861055850982666, "train_loss_llm": 0.0, "grad_norm": 0.025905977934598923, "global_step": 1867, "epoch": 51, "lr": 0.001} +{"train_loss": 0.05200973525643349, "train_loss_bc": 0.05200973525643349, "train_loss_llm": 0.0, "grad_norm": 0.0339985154569149, "global_step": 1868, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04391428828239441, "train_loss_bc": 0.04388333857059479, "train_loss_llm": 0.0030950605869293213, "grad_norm": 0.016206808388233185, "global_step": 1869, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04874888435006142, "train_loss_bc": 0.04874888435006142, "train_loss_llm": 0.0, "grad_norm": 0.029742931947112083, "global_step": 1870, "epoch": 51, "lr": 0.001} +{"train_loss": 0.04587852654771672, "train_loss_bc": 0.05261344835162163, "train_loss_llm": 0.0, "grad_norm": 0.03667698800563812, "global_step": 1871, "epoch": 51, "lr": 0.001, "val_loss": 0.05082818493247032} +{"train_loss": 0.04076356068253517, "train_loss_bc": 0.04075722023844719, "train_loss_llm": 0.0006339848041534424, "grad_norm": 0.045282963663339615, "global_step": 1872, "epoch": 52, "lr": 0.001} +{"train_loss": 0.052381373941898346, "train_loss_bc": 0.05237933620810509, "train_loss_llm": 0.00020381808280944824, "grad_norm": 0.020492149516940117, "global_step": 1873, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04982118308544159, "train_loss_bc": 0.04982118308544159, "train_loss_llm": 0.0, "grad_norm": 0.02515985816717148, "global_step": 1874, "epoch": 52, "lr": 0.001} +{"train_loss": 0.05001227930188179, "train_loss_bc": 0.050010018050670624, "train_loss_llm": 0.00022619962692260742, "grad_norm": 0.03180456906557083, "global_step": 1875, "epoch": 52, "lr": 0.001} +{"train_loss": 0.05158573389053345, "train_loss_bc": 0.05158573389053345, "train_loss_llm": 0.0, "grad_norm": 0.033868562430143356, "global_step": 1876, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04744192212820053, "train_loss_bc": 0.04742371663451195, "train_loss_llm": 0.0018205642700195312, "grad_norm": 0.014761107973754406, "global_step": 1877, "epoch": 52, "lr": 0.001} +{"train_loss": 0.045996952801942825, "train_loss_bc": 0.045996952801942825, "train_loss_llm": 0.0, "grad_norm": 0.0292382650077343, "global_step": 1878, "epoch": 52, "lr": 0.001} +{"train_loss": 0.0434131845831871, "train_loss_bc": 0.043395284563302994, "train_loss_llm": 0.0017899572849273682, "grad_norm": 0.028553253039717674, "global_step": 1879, "epoch": 52, "lr": 0.001} +{"train_loss": 0.05296161398291588, "train_loss_bc": 0.05296161398291588, "train_loss_llm": 0.0, "grad_norm": 0.03352396935224533, "global_step": 1880, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04206753894686699, "train_loss_bc": 0.04206753894686699, "train_loss_llm": 0.0, "grad_norm": 0.016514236107468605, "global_step": 1881, "epoch": 52, "lr": 0.001} +{"train_loss": 0.038820404559373856, "train_loss_bc": 0.038820404559373856, "train_loss_llm": 0.0, "grad_norm": 0.020184751600027084, "global_step": 1882, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04988822340965271, "train_loss_bc": 0.049876295030117035, "train_loss_llm": 0.0011929571628570557, "grad_norm": 0.03271003067493439, "global_step": 1883, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04185551404953003, "train_loss_bc": 0.04185551404953003, "train_loss_llm": 0.0, "grad_norm": 0.03705708682537079, "global_step": 1884, "epoch": 52, "lr": 0.001} +{"train_loss": 0.047685399651527405, "train_loss_bc": 0.047685399651527405, "train_loss_llm": 0.0, "grad_norm": 0.019016820937395096, "global_step": 1885, "epoch": 52, "lr": 0.001} +{"train_loss": 0.038766272366046906, "train_loss_bc": 0.038738030940294266, "train_loss_llm": 0.0028241872787475586, "grad_norm": 0.026189344003796577, "global_step": 1886, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04036823660135269, "train_loss_bc": 0.04036823660135269, "train_loss_llm": 0.0, "grad_norm": 0.029707545414566994, "global_step": 1887, "epoch": 52, "lr": 0.001} +{"train_loss": 0.041206225752830505, "train_loss_bc": 0.04118334501981735, "train_loss_llm": 0.0022881627082824707, "grad_norm": 0.03904018923640251, "global_step": 1888, "epoch": 52, "lr": 0.001} +{"train_loss": 0.05100153759121895, "train_loss_bc": 0.05097969248890877, "train_loss_llm": 0.002184540033340454, "grad_norm": 0.016439305618405342, "global_step": 1889, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04210976883769035, "train_loss_bc": 0.042067982256412506, "train_loss_llm": 0.004178494215011597, "grad_norm": 0.024625727906823158, "global_step": 1890, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04525279626250267, "train_loss_bc": 0.04522122070193291, "train_loss_llm": 0.00315740704536438, "grad_norm": 0.033852025866508484, "global_step": 1891, "epoch": 52, "lr": 0.001} +{"train_loss": 0.042472999542951584, "train_loss_bc": 0.042472999542951584, "train_loss_llm": 0.0, "grad_norm": 0.03897897154092789, "global_step": 1892, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04007618874311447, "train_loss_bc": 0.040068261325359344, "train_loss_llm": 0.0007926523685455322, "grad_norm": 0.01602754183113575, "global_step": 1893, "epoch": 52, "lr": 0.001} +{"train_loss": 0.05190436169505119, "train_loss_bc": 0.05190436169505119, "train_loss_llm": 0.0, "grad_norm": 0.021713437512516975, "global_step": 1894, "epoch": 52, "lr": 0.001} +{"train_loss": 0.047982871532440186, "train_loss_bc": 0.047982871532440186, "train_loss_llm": 0.0, "grad_norm": 0.03103829361498356, "global_step": 1895, "epoch": 52, "lr": 0.001} +{"train_loss": 0.047103311866521835, "train_loss_bc": 0.04706460237503052, "train_loss_llm": 0.003871023654937744, "grad_norm": 0.036578137427568436, "global_step": 1896, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04614822939038277, "train_loss_bc": 0.04614822939038277, "train_loss_llm": 0.0, "grad_norm": 0.01694231852889061, "global_step": 1897, "epoch": 52, "lr": 0.001} +{"train_loss": 0.053010549396276474, "train_loss_bc": 0.053010549396276474, "train_loss_llm": 0.0, "grad_norm": 0.031156081706285477, "global_step": 1898, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04842127487063408, "train_loss_bc": 0.04842127487063408, "train_loss_llm": 0.0, "grad_norm": 0.03504963591694832, "global_step": 1899, "epoch": 52, "lr": 0.001} +{"train_loss": 0.036461155861616135, "train_loss_bc": 0.036461155861616135, "train_loss_llm": 0.0, "grad_norm": 0.04006277397274971, "global_step": 1900, "epoch": 52, "lr": 0.001} +{"train_loss": 0.05022235959768295, "train_loss_bc": 0.05022235959768295, "train_loss_llm": 0.0, "grad_norm": 0.014882110990583897, "global_step": 1901, "epoch": 52, "lr": 0.001} +{"train_loss": 0.048610053956508636, "train_loss_bc": 0.048610053956508636, "train_loss_llm": 0.0, "grad_norm": 0.02322045899927616, "global_step": 1902, "epoch": 52, "lr": 0.001} +{"train_loss": 0.04489056393504143, "train_loss_bc": 0.04484320804476738, "train_loss_llm": 0.004735678434371948, "grad_norm": 0.02708723023533821, "global_step": 1903, "epoch": 52, "lr": 0.001} +{"train_loss": 0.043677911162376404, "train_loss_bc": 0.043676599860191345, "train_loss_llm": 0.00013116002082824707, "grad_norm": 0.03326718881726265, "global_step": 1904, "epoch": 52, "lr": 0.001} +{"train_loss": 0.05077078193426132, "train_loss_bc": 0.05077078193426132, "train_loss_llm": 0.0, "grad_norm": 0.016710426658391953, "global_step": 1905, "epoch": 52, "lr": 0.001} +{"train_loss": 0.03822941333055496, "train_loss_bc": 0.03822941333055496, "train_loss_llm": 0.0, "grad_norm": 0.022137310355901718, "global_step": 1906, "epoch": 52, "lr": 0.001} +{"train_loss": 0.045780657376680106, "train_loss_bc": 0.04472191631793976, "train_loss_llm": 0.0, "grad_norm": 0.031198734417557716, "global_step": 1907, "epoch": 52, "lr": 0.001, "val_loss": 0.050646740943193436} +{"train_loss": 0.04979455843567848, "train_loss_bc": 0.04979455843567848, "train_loss_llm": 0.0, "grad_norm": 0.04353999346494675, "global_step": 1908, "epoch": 53, "lr": 0.001} +{"train_loss": 0.046166349202394485, "train_loss_bc": 0.046166349202394485, "train_loss_llm": 0.0, "grad_norm": 0.01384154986590147, "global_step": 1909, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04128407686948776, "train_loss_bc": 0.04128407686948776, "train_loss_llm": 0.0, "grad_norm": 0.025155294686555862, "global_step": 1910, "epoch": 53, "lr": 0.001} +{"train_loss": 0.050379540771245956, "train_loss_bc": 0.050367143005132675, "train_loss_llm": 0.0012398958206176758, "grad_norm": 0.03645912930369377, "global_step": 1911, "epoch": 53, "lr": 0.001} +{"train_loss": 0.0488274060189724, "train_loss_bc": 0.0488274060189724, "train_loss_llm": 0.0, "grad_norm": 0.03655533865094185, "global_step": 1912, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04860253259539604, "train_loss_bc": 0.0486023910343647, "train_loss_llm": 1.4036893844604492e-05, "grad_norm": 0.01403921376913786, "global_step": 1913, "epoch": 53, "lr": 0.001} +{"train_loss": 0.052166376262903214, "train_loss_bc": 0.052166376262903214, "train_loss_llm": 0.0, "grad_norm": 0.02439105324447155, "global_step": 1914, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04567331075668335, "train_loss_bc": 0.04567331075668335, "train_loss_llm": 0.0, "grad_norm": 0.03223823010921478, "global_step": 1915, "epoch": 53, "lr": 0.001} +{"train_loss": 0.05024338141083717, "train_loss_bc": 0.05024338141083717, "train_loss_llm": 0.0, "grad_norm": 0.035117797553539276, "global_step": 1916, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04525590315461159, "train_loss_bc": 0.04525590315461159, "train_loss_llm": 0.0, "grad_norm": 0.019179437309503555, "global_step": 1917, "epoch": 53, "lr": 0.001} +{"train_loss": 0.050828441977500916, "train_loss_bc": 0.05080845206975937, "train_loss_llm": 0.0019989311695098877, "grad_norm": 0.02511531487107277, "global_step": 1918, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04666353762149811, "train_loss_bc": 0.04666353762149811, "train_loss_llm": 0.0, "grad_norm": 0.03146660700440407, "global_step": 1919, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04161437600851059, "train_loss_bc": 0.041603777557611465, "train_loss_llm": 0.0010599493980407715, "grad_norm": 0.038792844861745834, "global_step": 1920, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04119200259447098, "train_loss_bc": 0.04119035229086876, "train_loss_llm": 0.00016495585441589355, "grad_norm": 0.01725730113685131, "global_step": 1921, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04366948828101158, "train_loss_bc": 0.04362548887729645, "train_loss_llm": 0.00439983606338501, "grad_norm": 0.028397077694535255, "global_step": 1922, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04136094078421593, "train_loss_bc": 0.04136094078421593, "train_loss_llm": 0.0, "grad_norm": 0.03250572457909584, "global_step": 1923, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04902864247560501, "train_loss_bc": 0.04902864247560501, "train_loss_llm": 0.0, "grad_norm": 0.0383693091571331, "global_step": 1924, "epoch": 53, "lr": 0.001} +{"train_loss": 0.0392850860953331, "train_loss_bc": 0.039280787110328674, "train_loss_llm": 0.0004298985004425049, "grad_norm": 0.013725859113037586, "global_step": 1925, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04202297702431679, "train_loss_bc": 0.04202297702431679, "train_loss_llm": 0.0, "grad_norm": 0.026687350124120712, "global_step": 1926, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04523796588182449, "train_loss_bc": 0.04523796588182449, "train_loss_llm": 0.0, "grad_norm": 0.037307195365428925, "global_step": 1927, "epoch": 53, "lr": 0.001} +{"train_loss": 0.048068344593048096, "train_loss_bc": 0.0480429083108902, "train_loss_llm": 0.002543509006500244, "grad_norm": 0.04159228131175041, "global_step": 1928, "epoch": 53, "lr": 0.001} +{"train_loss": 0.037870291620492935, "train_loss_bc": 0.037870291620492935, "train_loss_llm": 0.0, "grad_norm": 0.01730315387248993, "global_step": 1929, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04595400020480156, "train_loss_bc": 0.04594246298074722, "train_loss_llm": 0.0011537373065948486, "grad_norm": 0.027804788202047348, "global_step": 1930, "epoch": 53, "lr": 0.001} +{"train_loss": 0.05037441477179527, "train_loss_bc": 0.0503445602953434, "train_loss_llm": 0.0029854178428649902, "grad_norm": 0.03829585388302803, "global_step": 1931, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04674602672457695, "train_loss_bc": 0.04672710970044136, "train_loss_llm": 0.001891791820526123, "grad_norm": 0.04588988795876503, "global_step": 1932, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04533055052161217, "train_loss_bc": 0.04533055052161217, "train_loss_llm": 0.0, "grad_norm": 0.01688571646809578, "global_step": 1933, "epoch": 53, "lr": 0.001} +{"train_loss": 0.043979711830616, "train_loss_bc": 0.043979711830616, "train_loss_llm": 0.0, "grad_norm": 0.023955751210451126, "global_step": 1934, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04811966419219971, "train_loss_bc": 0.04811966419219971, "train_loss_llm": 0.0, "grad_norm": 0.03136301785707474, "global_step": 1935, "epoch": 53, "lr": 0.001} +{"train_loss": 0.03970440477132797, "train_loss_bc": 0.03969366103410721, "train_loss_llm": 0.0010745525360107422, "grad_norm": 0.04130249097943306, "global_step": 1936, "epoch": 53, "lr": 0.001} +{"train_loss": 0.040868863463401794, "train_loss_bc": 0.04086107760667801, "train_loss_llm": 0.00077858567237854, "grad_norm": 0.013744800351560116, "global_step": 1937, "epoch": 53, "lr": 0.001} +{"train_loss": 0.047815266996622086, "train_loss_bc": 0.047801073640584946, "train_loss_llm": 0.0014194250106811523, "grad_norm": 0.022015534341335297, "global_step": 1938, "epoch": 53, "lr": 0.001} +{"train_loss": 0.05485230311751366, "train_loss_bc": 0.05485230311751366, "train_loss_llm": 0.0, "grad_norm": 0.033319033682346344, "global_step": 1939, "epoch": 53, "lr": 0.001} +{"train_loss": 0.0412459559738636, "train_loss_bc": 0.0412459559738636, "train_loss_llm": 0.0, "grad_norm": 0.037026599049568176, "global_step": 1940, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04354230687022209, "train_loss_bc": 0.04354230687022209, "train_loss_llm": 0.0, "grad_norm": 0.014684691093862057, "global_step": 1941, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04289012774825096, "train_loss_bc": 0.04287932068109512, "train_loss_llm": 0.0010806620121002197, "grad_norm": 0.021132351830601692, "global_step": 1942, "epoch": 53, "lr": 0.001} +{"train_loss": 0.04591811489727762, "train_loss_bc": 0.056364625692367554, "train_loss_llm": 0.002838224172592163, "grad_norm": 0.03234334662556648, "global_step": 1943, "epoch": 53, "lr": 0.001, "val_loss": 0.05073698237538338} +{"train_loss": 0.04785817861557007, "train_loss_bc": 0.04785817861557007, "train_loss_llm": 0.0, "grad_norm": 0.04017387703061104, "global_step": 1944, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04686043784022331, "train_loss_bc": 0.04683084040880203, "train_loss_llm": 0.002959907054901123, "grad_norm": 0.015846582129597664, "global_step": 1945, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04383349418640137, "train_loss_bc": 0.04383349418640137, "train_loss_llm": 0.0, "grad_norm": 0.023002859205007553, "global_step": 1946, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04337475448846817, "train_loss_bc": 0.04337475448846817, "train_loss_llm": 0.0, "grad_norm": 0.026641344651579857, "global_step": 1947, "epoch": 54, "lr": 0.001} +{"train_loss": 0.045003972947597504, "train_loss_bc": 0.045003972947597504, "train_loss_llm": 0.0, "grad_norm": 0.031639184802770615, "global_step": 1948, "epoch": 54, "lr": 0.001} +{"train_loss": 0.03869583457708359, "train_loss_bc": 0.03869583457708359, "train_loss_llm": 0.0, "grad_norm": 0.01731000281870365, "global_step": 1949, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04971204698085785, "train_loss_bc": 0.04970230534672737, "train_loss_llm": 0.0009739995002746582, "grad_norm": 0.02949542924761772, "global_step": 1950, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04653669521212578, "train_loss_bc": 0.04653669521212578, "train_loss_llm": 0.0, "grad_norm": 0.039063163101673126, "global_step": 1951, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04994769021868706, "train_loss_bc": 0.04994652420282364, "train_loss_llm": 0.00011673569679260254, "grad_norm": 0.04840662330389023, "global_step": 1952, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04233360290527344, "train_loss_bc": 0.04232122376561165, "train_loss_llm": 0.0012380480766296387, "grad_norm": 0.019369009882211685, "global_step": 1953, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04199197515845299, "train_loss_bc": 0.04199197515845299, "train_loss_llm": 0.0, "grad_norm": 0.028479112312197685, "global_step": 1954, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04429543390870094, "train_loss_bc": 0.04429543390870094, "train_loss_llm": 0.0, "grad_norm": 0.03460143133997917, "global_step": 1955, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04687025398015976, "train_loss_bc": 0.04687025398015976, "train_loss_llm": 0.0, "grad_norm": 0.041985802352428436, "global_step": 1956, "epoch": 54, "lr": 0.001} +{"train_loss": 0.0418865941464901, "train_loss_bc": 0.0418865941464901, "train_loss_llm": 0.0, "grad_norm": 0.013607210479676723, "global_step": 1957, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04316474124789238, "train_loss_bc": 0.04316474124789238, "train_loss_llm": 0.0, "grad_norm": 0.018922120332717896, "global_step": 1958, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04508219286799431, "train_loss_bc": 0.04508219286799431, "train_loss_llm": 0.0, "grad_norm": 0.027718622237443924, "global_step": 1959, "epoch": 54, "lr": 0.001} +{"train_loss": 0.042374443262815475, "train_loss_bc": 0.04235462471842766, "train_loss_llm": 0.0019819438457489014, "grad_norm": 0.037017498165369034, "global_step": 1960, "epoch": 54, "lr": 0.001} +{"train_loss": 0.050812218338251114, "train_loss_bc": 0.050785910338163376, "train_loss_llm": 0.00263097882270813, "grad_norm": 0.017522485926747322, "global_step": 1961, "epoch": 54, "lr": 0.001} +{"train_loss": 0.0408669114112854, "train_loss_bc": 0.0408669114112854, "train_loss_llm": 0.0, "grad_norm": 0.023679956793785095, "global_step": 1962, "epoch": 54, "lr": 0.001} +{"train_loss": 0.0420977957546711, "train_loss_bc": 0.04208846390247345, "train_loss_llm": 0.0009330213069915771, "grad_norm": 0.024203673005104065, "global_step": 1963, "epoch": 54, "lr": 0.001} +{"train_loss": 0.0459524542093277, "train_loss_bc": 0.0459524542093277, "train_loss_llm": 0.0, "grad_norm": 0.02958843670785427, "global_step": 1964, "epoch": 54, "lr": 0.001} +{"train_loss": 0.049937307834625244, "train_loss_bc": 0.04993557557463646, "train_loss_llm": 0.00017324090003967285, "grad_norm": 0.013109882362186909, "global_step": 1965, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04714931175112724, "train_loss_bc": 0.04714931175112724, "train_loss_llm": 0.0, "grad_norm": 0.01751399226486683, "global_step": 1966, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04855729267001152, "train_loss_bc": 0.04854007810354233, "train_loss_llm": 0.001721501350402832, "grad_norm": 0.02131984941661358, "global_step": 1967, "epoch": 54, "lr": 0.001} +{"train_loss": 0.05102916061878204, "train_loss_bc": 0.05102916061878204, "train_loss_llm": 0.0, "grad_norm": 0.028773963451385498, "global_step": 1968, "epoch": 54, "lr": 0.001} +{"train_loss": 0.047970887273550034, "train_loss_bc": 0.047970887273550034, "train_loss_llm": 0.0, "grad_norm": 0.02017807401716709, "global_step": 1969, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04423220455646515, "train_loss_bc": 0.04423220455646515, "train_loss_llm": 0.0, "grad_norm": 0.02503066323697567, "global_step": 1970, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04288472235202789, "train_loss_bc": 0.04288472235202789, "train_loss_llm": 0.0, "grad_norm": 0.03022746741771698, "global_step": 1971, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04974538832902908, "train_loss_bc": 0.04974538832902908, "train_loss_llm": 0.0, "grad_norm": 0.040934257209300995, "global_step": 1972, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04205501824617386, "train_loss_bc": 0.041988272219896317, "train_loss_llm": 0.006674468517303467, "grad_norm": 0.015417531132698059, "global_step": 1973, "epoch": 54, "lr": 0.001} +{"train_loss": 0.05116861313581467, "train_loss_bc": 0.05115200951695442, "train_loss_llm": 0.0016603171825408936, "grad_norm": 0.024341927841305733, "global_step": 1974, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04942893609404564, "train_loss_bc": 0.04942893609404564, "train_loss_llm": 0.0, "grad_norm": 0.030256150290369987, "global_step": 1975, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04872226342558861, "train_loss_bc": 0.04870808124542236, "train_loss_llm": 0.0014181435108184814, "grad_norm": 0.038973450660705566, "global_step": 1976, "epoch": 54, "lr": 0.001} +{"train_loss": 0.04496383294463158, "train_loss_bc": 0.04491959884762764, "train_loss_llm": 0.004423558712005615, "grad_norm": 0.016306357458233833, "global_step": 1977, "epoch": 54, "lr": 0.001} +{"train_loss": 0.03993486613035202, "train_loss_bc": 0.03990291804075241, "train_loss_llm": 0.0031948983669281006, "grad_norm": 0.026735341176390648, "global_step": 1978, "epoch": 54, "lr": 0.001} +{"train_loss": 0.045755673510332905, "train_loss_bc": 0.04985556751489639, "train_loss_llm": 0.0017151832580566406, "grad_norm": 0.041912417858839035, "global_step": 1979, "epoch": 54, "lr": 0.001, "val_loss": 0.05087349936366081} +{"train_loss": 0.03611886873841286, "train_loss_bc": 0.03611886873841286, "train_loss_llm": 0.0, "grad_norm": 0.05027013272047043, "global_step": 1980, "epoch": 55, "lr": 0.001} +{"train_loss": 0.045823875814676285, "train_loss_bc": 0.045823875814676285, "train_loss_llm": 0.0, "grad_norm": 0.013653757981956005, "global_step": 1981, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04764299467206001, "train_loss_bc": 0.04764299467206001, "train_loss_llm": 0.0, "grad_norm": 0.018697720021009445, "global_step": 1982, "epoch": 55, "lr": 0.001} +{"train_loss": 0.043210916221141815, "train_loss_bc": 0.043210916221141815, "train_loss_llm": 0.0, "grad_norm": 0.025508733466267586, "global_step": 1983, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04957358539104462, "train_loss_bc": 0.04955543205142021, "train_loss_llm": 0.0018152594566345215, "grad_norm": 0.03126128762960434, "global_step": 1984, "epoch": 55, "lr": 0.001} +{"train_loss": 0.05027073994278908, "train_loss_bc": 0.05027073994278908, "train_loss_llm": 0.0, "grad_norm": 0.014239020645618439, "global_step": 1985, "epoch": 55, "lr": 0.001} +{"train_loss": 0.0492924302816391, "train_loss_bc": 0.0492924302816391, "train_loss_llm": 0.0, "grad_norm": 0.023133307695388794, "global_step": 1986, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04931193217635155, "train_loss_bc": 0.049279533326625824, "train_loss_llm": 0.003239929676055908, "grad_norm": 0.02951113134622574, "global_step": 1987, "epoch": 55, "lr": 0.001} +{"train_loss": 0.046783771365880966, "train_loss_bc": 0.04678359255194664, "train_loss_llm": 1.7762184143066406e-05, "grad_norm": 0.03119739703834057, "global_step": 1988, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04722768813371658, "train_loss_bc": 0.047222983092069626, "train_loss_llm": 0.0004706382751464844, "grad_norm": 0.017927788197994232, "global_step": 1989, "epoch": 55, "lr": 0.001} +{"train_loss": 0.03401566669344902, "train_loss_bc": 0.03401566669344902, "train_loss_llm": 0.0, "grad_norm": 0.021015189588069916, "global_step": 1990, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04302827641367912, "train_loss_bc": 0.04302827641367912, "train_loss_llm": 0.0, "grad_norm": 0.028143389150500298, "global_step": 1991, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04302344098687172, "train_loss_bc": 0.043021224439144135, "train_loss_llm": 0.00022149085998535156, "grad_norm": 0.03159896284341812, "global_step": 1992, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04195224866271019, "train_loss_bc": 0.04195183515548706, "train_loss_llm": 4.124641418457031e-05, "grad_norm": 0.014521926641464233, "global_step": 1993, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04762145131826401, "train_loss_bc": 0.04761870205402374, "train_loss_llm": 0.00027489662170410156, "grad_norm": 0.03301399573683739, "global_step": 1994, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04845016822218895, "train_loss_bc": 0.048447493463754654, "train_loss_llm": 0.0002675950527191162, "grad_norm": 0.04245426133275032, "global_step": 1995, "epoch": 55, "lr": 0.001} +{"train_loss": 0.045423708856105804, "train_loss_bc": 0.04540301486849785, "train_loss_llm": 0.0020694732666015625, "grad_norm": 0.04951196536421776, "global_step": 1996, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04841853305697441, "train_loss_bc": 0.04841853305697441, "train_loss_llm": 0.0, "grad_norm": 0.017816947773098946, "global_step": 1997, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04385979101061821, "train_loss_bc": 0.04385979101061821, "train_loss_llm": 0.0, "grad_norm": 0.024393968284130096, "global_step": 1998, "epoch": 55, "lr": 0.001} +{"train_loss": 0.046356845647096634, "train_loss_bc": 0.046356845647096634, "train_loss_llm": 0.0, "grad_norm": 0.03255070745944977, "global_step": 1999, "epoch": 55, "lr": 0.001} +{"train_loss": 0.0429229773581028, "train_loss_bc": 0.0429229773581028, "train_loss_llm": 0.0, "grad_norm": 0.041763052344322205, "global_step": 2000, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04565019533038139, "train_loss_bc": 0.04565019533038139, "train_loss_llm": 0.0, "grad_norm": 0.014708483591675758, "global_step": 2001, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04126700386404991, "train_loss_bc": 0.04126700386404991, "train_loss_llm": 0.0, "grad_norm": 0.025778844952583313, "global_step": 2002, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04552285373210907, "train_loss_bc": 0.04552067443728447, "train_loss_llm": 0.00021794438362121582, "grad_norm": 0.03247632458806038, "global_step": 2003, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04640756919980049, "train_loss_bc": 0.04640756919980049, "train_loss_llm": 0.0, "grad_norm": 0.036897655576467514, "global_step": 2004, "epoch": 55, "lr": 0.001} +{"train_loss": 0.05733611807227135, "train_loss_bc": 0.05733484402298927, "train_loss_llm": 0.0001272261142730713, "grad_norm": 0.014090325683355331, "global_step": 2005, "epoch": 55, "lr": 0.001} +{"train_loss": 0.042716916650533676, "train_loss_bc": 0.04270840063691139, "train_loss_llm": 0.0008515119552612305, "grad_norm": 0.0236407108604908, "global_step": 2006, "epoch": 55, "lr": 0.001} +{"train_loss": 0.046483274549245834, "train_loss_bc": 0.04647436365485191, "train_loss_llm": 0.0008912384510040283, "grad_norm": 0.027914736419916153, "global_step": 2007, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04070713743567467, "train_loss_bc": 0.04070713743567467, "train_loss_llm": 0.0, "grad_norm": 0.03241521120071411, "global_step": 2008, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04137859866023064, "train_loss_bc": 0.0413595475256443, "train_loss_llm": 0.0019050836563110352, "grad_norm": 0.017545588314533234, "global_step": 2009, "epoch": 55, "lr": 0.001} +{"train_loss": 0.05519542843103409, "train_loss_bc": 0.05519542843103409, "train_loss_llm": 0.0, "grad_norm": 0.02068227343261242, "global_step": 2010, "epoch": 55, "lr": 0.001} +{"train_loss": 0.03631731867790222, "train_loss_bc": 0.03631731867790222, "train_loss_llm": 0.0, "grad_norm": 0.028111785650253296, "global_step": 2011, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04735118895769119, "train_loss_bc": 0.047320738434791565, "train_loss_llm": 0.003044992685317993, "grad_norm": 0.03555553779006004, "global_step": 2012, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04687879607081413, "train_loss_bc": 0.04687879607081413, "train_loss_llm": 0.0, "grad_norm": 0.018030237406492233, "global_step": 2013, "epoch": 55, "lr": 0.001} +{"train_loss": 0.047998443245887756, "train_loss_bc": 0.047998443245887756, "train_loss_llm": 0.0, "grad_norm": 0.022474996745586395, "global_step": 2014, "epoch": 55, "lr": 0.001} +{"train_loss": 0.04567159619182348, "train_loss_bc": 0.052636709064245224, "train_loss_llm": 0.0, "grad_norm": 0.03801355138421059, "global_step": 2015, "epoch": 55, "lr": 0.001, "train/cumulative_reward": 69.6529534177109, "train/mean_score": 12.803204577850755, "train/success_rate": 0.3, "test/cumulative_reward": 68.49532786294148, "test/mean_score": 9.674767353794431, "test/success_rate": 0.22, "val_loss": 0.05053424835205078, "train_action_mse_error": 0.04947744309902191} +{"train_loss": 0.04841666668653488, "train_loss_bc": 0.04841182380914688, "train_loss_llm": 0.000484466552734375, "grad_norm": 0.04210978373885155, "global_step": 2016, "epoch": 56, "lr": 0.001} +{"train_loss": 0.047860417515039444, "train_loss_bc": 0.047860417515039444, "train_loss_llm": 0.0, "grad_norm": 0.018698349595069885, "global_step": 2017, "epoch": 56, "lr": 0.001} +{"train_loss": 0.041198525577783585, "train_loss_bc": 0.041198525577783585, "train_loss_llm": 0.0, "grad_norm": 0.02357310801744461, "global_step": 2018, "epoch": 56, "lr": 0.001} +{"train_loss": 0.03950457647442818, "train_loss_bc": 0.03950457647442818, "train_loss_llm": 0.0, "grad_norm": 0.02653251588344574, "global_step": 2019, "epoch": 56, "lr": 0.001} +{"train_loss": 0.040955785661935806, "train_loss_bc": 0.040955785661935806, "train_loss_llm": 0.0, "grad_norm": 0.0323246493935585, "global_step": 2020, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04634596034884453, "train_loss_bc": 0.04634596034884453, "train_loss_llm": 0.0, "grad_norm": 0.017931530252099037, "global_step": 2021, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04765898361802101, "train_loss_bc": 0.04765608534216881, "train_loss_llm": 0.0002899467945098877, "grad_norm": 0.030385412275791168, "global_step": 2022, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04696853086352348, "train_loss_bc": 0.04695696383714676, "train_loss_llm": 0.0011567771434783936, "grad_norm": 0.03942682966589928, "global_step": 2023, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04475958272814751, "train_loss_bc": 0.04475958272814751, "train_loss_llm": 0.0, "grad_norm": 0.04573652893304825, "global_step": 2024, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04245678335428238, "train_loss_bc": 0.042437221854925156, "train_loss_llm": 0.0019563138484954834, "grad_norm": 0.016475392505526543, "global_step": 2025, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04954029992222786, "train_loss_bc": 0.049507297575473785, "train_loss_llm": 0.003300398588180542, "grad_norm": 0.022599956020712852, "global_step": 2026, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04625660926103592, "train_loss_bc": 0.04625660926103592, "train_loss_llm": 0.0, "grad_norm": 0.031475380063056946, "global_step": 2027, "epoch": 56, "lr": 0.001} +{"train_loss": 0.043620042502880096, "train_loss_bc": 0.043604202568531036, "train_loss_llm": 0.0015841126441955566, "grad_norm": 0.038119494915008545, "global_step": 2028, "epoch": 56, "lr": 0.001} +{"train_loss": 0.05007598176598549, "train_loss_bc": 0.05007598176598549, "train_loss_llm": 0.0, "grad_norm": 0.015210511162877083, "global_step": 2029, "epoch": 56, "lr": 0.001} +{"train_loss": 0.03248945623636246, "train_loss_bc": 0.03248945623636246, "train_loss_llm": 0.0, "grad_norm": 0.02234974503517151, "global_step": 2030, "epoch": 56, "lr": 0.001} +{"train_loss": 0.05134667828679085, "train_loss_bc": 0.051340654492378235, "train_loss_llm": 0.0006022155284881592, "grad_norm": 0.028595294803380966, "global_step": 2031, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04220356047153473, "train_loss_bc": 0.04220317304134369, "train_loss_llm": 3.883242607116699e-05, "grad_norm": 0.0338345542550087, "global_step": 2032, "epoch": 56, "lr": 0.001} +{"train_loss": 0.046405937522649765, "train_loss_bc": 0.046405937522649765, "train_loss_llm": 0.0, "grad_norm": 0.02125152014195919, "global_step": 2033, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04115282744169235, "train_loss_bc": 0.04115282744169235, "train_loss_llm": 0.0, "grad_norm": 0.03352019935846329, "global_step": 2034, "epoch": 56, "lr": 0.001} +{"train_loss": 0.0419892892241478, "train_loss_bc": 0.04198259860277176, "train_loss_llm": 0.0006691813468933105, "grad_norm": 0.04487663507461548, "global_step": 2035, "epoch": 56, "lr": 0.001} +{"train_loss": 0.045793939381837845, "train_loss_bc": 0.045793939381837845, "train_loss_llm": 0.0, "grad_norm": 0.05767858773469925, "global_step": 2036, "epoch": 56, "lr": 0.001} +{"train_loss": 0.037415795028209686, "train_loss_bc": 0.037403129041194916, "train_loss_llm": 0.0012666881084442139, "grad_norm": 0.01489931158721447, "global_step": 2037, "epoch": 56, "lr": 0.001} +{"train_loss": 0.05156770721077919, "train_loss_bc": 0.05156770721077919, "train_loss_llm": 0.0, "grad_norm": 0.022686483338475227, "global_step": 2038, "epoch": 56, "lr": 0.001} +{"train_loss": 0.049092963337898254, "train_loss_bc": 0.049092963337898254, "train_loss_llm": 0.0, "grad_norm": 0.030229778960347176, "global_step": 2039, "epoch": 56, "lr": 0.001} +{"train_loss": 0.049233537167310715, "train_loss_bc": 0.049233537167310715, "train_loss_llm": 0.0, "grad_norm": 0.031336709856987, "global_step": 2040, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04435189440846443, "train_loss_bc": 0.04435189440846443, "train_loss_llm": 0.0, "grad_norm": 0.016722237691283226, "global_step": 2041, "epoch": 56, "lr": 0.001} +{"train_loss": 0.05144000053405762, "train_loss_bc": 0.05144000053405762, "train_loss_llm": 0.0, "grad_norm": 0.025388143956661224, "global_step": 2042, "epoch": 56, "lr": 0.001} +{"train_loss": 0.050109706819057465, "train_loss_bc": 0.050109706819057465, "train_loss_llm": 0.0, "grad_norm": 0.03251449018716812, "global_step": 2043, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04950874298810959, "train_loss_bc": 0.04949292913079262, "train_loss_llm": 0.0015812814235687256, "grad_norm": 0.03934869542717934, "global_step": 2044, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04591525346040726, "train_loss_bc": 0.04589880630373955, "train_loss_llm": 0.0016445517539978027, "grad_norm": 0.016673710197210312, "global_step": 2045, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04924793168902397, "train_loss_bc": 0.049221619963645935, "train_loss_llm": 0.0026311874389648438, "grad_norm": 0.021618226543068886, "global_step": 2046, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04516331106424332, "train_loss_bc": 0.04516331106424332, "train_loss_llm": 0.0, "grad_norm": 0.027900459244847298, "global_step": 2047, "epoch": 56, "lr": 0.001} +{"train_loss": 0.05207888036966324, "train_loss_bc": 0.05207888036966324, "train_loss_llm": 0.0, "grad_norm": 0.04010845348238945, "global_step": 2048, "epoch": 56, "lr": 0.001} +{"train_loss": 0.049111030995845795, "train_loss_bc": 0.04907295107841492, "train_loss_llm": 0.0038078725337982178, "grad_norm": 0.01452819723635912, "global_step": 2049, "epoch": 56, "lr": 0.001} +{"train_loss": 0.03795398771762848, "train_loss_bc": 0.0379476323723793, "train_loss_llm": 0.0006357133388519287, "grad_norm": 0.025183744728565216, "global_step": 2050, "epoch": 56, "lr": 0.001} +{"train_loss": 0.04557774526377519, "train_loss_bc": 0.041607651859521866, "train_loss_llm": 0.0, "grad_norm": 0.03509484976530075, "global_step": 2051, "epoch": 56, "lr": 0.001, "val_loss": 0.050699952989816666} +{"train_loss": 0.05060287564992905, "train_loss_bc": 0.05060287564992905, "train_loss_llm": 0.0, "grad_norm": 0.03652777150273323, "global_step": 2052, "epoch": 57, "lr": 0.001} +{"train_loss": 0.04002537205815315, "train_loss_bc": 0.0399978831410408, "train_loss_llm": 0.0027490556240081787, "grad_norm": 0.015766886994242668, "global_step": 2053, "epoch": 57, "lr": 0.001} +{"train_loss": 0.047763776034116745, "train_loss_bc": 0.04775933921337128, "train_loss_llm": 0.0004436671733856201, "grad_norm": 0.018438467755913734, "global_step": 2054, "epoch": 57, "lr": 0.001} +{"train_loss": 0.040775954723358154, "train_loss_bc": 0.040775954723358154, "train_loss_llm": 0.0, "grad_norm": 0.024120783433318138, "global_step": 2055, "epoch": 57, "lr": 0.001} +{"train_loss": 0.053721003234386444, "train_loss_bc": 0.05371219664812088, "train_loss_llm": 0.0008806586265563965, "grad_norm": 0.031767915934324265, "global_step": 2056, "epoch": 57, "lr": 0.001} +{"train_loss": 0.040583573281764984, "train_loss_bc": 0.04057534411549568, "train_loss_llm": 0.0008229315280914307, "grad_norm": 0.014790761284530163, "global_step": 2057, "epoch": 57, "lr": 0.001} +{"train_loss": 0.04397546499967575, "train_loss_bc": 0.04397546499967575, "train_loss_llm": 0.0, "grad_norm": 0.02005748450756073, "global_step": 2058, "epoch": 57, "lr": 0.001} +{"train_loss": 0.046285662800073624, "train_loss_bc": 0.046285662800073624, "train_loss_llm": 0.0, "grad_norm": 0.029075413942337036, "global_step": 2059, "epoch": 57, "lr": 0.001} +{"train_loss": 0.040520403534173965, "train_loss_bc": 0.040489256381988525, "train_loss_llm": 0.0031147301197052, "grad_norm": 0.040976639837026596, "global_step": 2060, "epoch": 57, "lr": 0.001} +{"train_loss": 0.04477702081203461, "train_loss_bc": 0.04477702081203461, "train_loss_llm": 0.0, "grad_norm": 0.016362037509679794, "global_step": 2061, "epoch": 57, "lr": 0.001} +{"train_loss": 0.05383666232228279, "train_loss_bc": 0.05383632332086563, "train_loss_llm": 3.3855438232421875e-05, "grad_norm": 0.02779228612780571, "global_step": 2062, "epoch": 57, "lr": 0.001} +{"train_loss": 0.05083543807268143, "train_loss_bc": 0.05083543807268143, "train_loss_llm": 0.0, "grad_norm": 0.03081860952079296, "global_step": 2063, "epoch": 57, "lr": 0.001} +{"train_loss": 0.049252402037382126, "train_loss_bc": 0.0492408350110054, "train_loss_llm": 0.001156628131866455, "grad_norm": 0.040869127959012985, "global_step": 2064, "epoch": 57, "lr": 0.001} +{"train_loss": 0.04103853926062584, "train_loss_bc": 0.04103853926062584, "train_loss_llm": 0.0, "grad_norm": 0.015420577488839626, "global_step": 2065, "epoch": 57, "lr": 0.001} +{"train_loss": 0.04321901872754097, "train_loss_bc": 0.04321901872754097, "train_loss_llm": 0.0, "grad_norm": 0.019646624103188515, "global_step": 2066, "epoch": 57, "lr": 0.001} +{"train_loss": 0.05010541155934334, "train_loss_bc": 0.05010541155934334, "train_loss_llm": 0.0, "grad_norm": 0.030183814465999603, "global_step": 2067, "epoch": 57, "lr": 0.001} +{"train_loss": 0.041231051087379456, "train_loss_bc": 0.041231051087379456, "train_loss_llm": 0.0, "grad_norm": 0.0336262546479702, "global_step": 2068, "epoch": 57, "lr": 0.001} +{"train_loss": 0.044222570955753326, "train_loss_bc": 0.044222570955753326, "train_loss_llm": 0.0, "grad_norm": 0.014812873676419258, "global_step": 2069, "epoch": 57, "lr": 0.001} +{"train_loss": 0.03839454799890518, "train_loss_bc": 0.03839454799890518, "train_loss_llm": 0.0, "grad_norm": 0.02115112543106079, "global_step": 2070, "epoch": 57, "lr": 0.001} +{"train_loss": 0.04014832526445389, "train_loss_bc": 0.04013688117265701, "train_loss_llm": 0.0011443793773651123, "grad_norm": 0.029559539631009102, "global_step": 2071, "epoch": 57, "lr": 0.001} +{"train_loss": 0.049761585891246796, "train_loss_bc": 0.049713898450136185, "train_loss_llm": 0.0047689080238342285, "grad_norm": 0.03208106756210327, "global_step": 2072, "epoch": 57, "lr": 0.001} +{"train_loss": 0.04775966331362724, "train_loss_bc": 0.04775966331362724, "train_loss_llm": 0.0, "grad_norm": 0.015009148977696896, "global_step": 2073, "epoch": 57, "lr": 0.001} +{"train_loss": 0.03898262232542038, "train_loss_bc": 0.03898262232542038, "train_loss_llm": 0.0, "grad_norm": 0.02580363117158413, "global_step": 2074, "epoch": 57, "lr": 0.001} +{"train_loss": 0.051227912306785583, "train_loss_bc": 0.05121168866753578, "train_loss_llm": 0.0016224980354309082, "grad_norm": 0.03668908029794693, "global_step": 2075, "epoch": 57, "lr": 0.001} +{"train_loss": 0.05381447449326515, "train_loss_bc": 0.05381447449326515, "train_loss_llm": 0.0, "grad_norm": 0.04385273903608322, "global_step": 2076, "epoch": 57, "lr": 0.001} +{"train_loss": 0.042280685156583786, "train_loss_bc": 0.04226988926529884, "train_loss_llm": 0.001079469919204712, "grad_norm": 0.02006165124475956, "global_step": 2077, "epoch": 57, "lr": 0.001} +{"train_loss": 0.04552638158202171, "train_loss_bc": 0.04552638158202171, "train_loss_llm": 0.0, "grad_norm": 0.036315642297267914, "global_step": 2078, "epoch": 57, "lr": 0.001} +{"train_loss": 0.046590641140937805, "train_loss_bc": 0.046585001051425934, "train_loss_llm": 0.0005640089511871338, "grad_norm": 0.05190960690379143, "global_step": 2079, "epoch": 57, "lr": 0.001} +{"train_loss": 0.04793432727456093, "train_loss_bc": 0.04792676866054535, "train_loss_llm": 0.0007559657096862793, "grad_norm": 0.0632161945104599, "global_step": 2080, "epoch": 57, "lr": 0.001} +{"train_loss": 0.042232267558574677, "train_loss_bc": 0.042232267558574677, "train_loss_llm": 0.0, "grad_norm": 0.014424431137740612, "global_step": 2081, "epoch": 57, "lr": 0.001} +{"train_loss": 0.050699539482593536, "train_loss_bc": 0.050699539482593536, "train_loss_llm": 0.0, "grad_norm": 0.02625381201505661, "global_step": 2082, "epoch": 57, "lr": 0.001} +{"train_loss": 0.046912796795368195, "train_loss_bc": 0.046888794749975204, "train_loss_llm": 0.0024000704288482666, "grad_norm": 0.0275142639875412, "global_step": 2083, "epoch": 57, "lr": 0.001} +{"train_loss": 0.049762606620788574, "train_loss_bc": 0.049762606620788574, "train_loss_llm": 0.0, "grad_norm": 0.029040319845080376, "global_step": 2084, "epoch": 57, "lr": 0.001} +{"train_loss": 0.03797420859336853, "train_loss_bc": 0.03797420859336853, "train_loss_llm": 0.0, "grad_norm": 0.016580762341618538, "global_step": 2085, "epoch": 57, "lr": 0.001} +{"train_loss": 0.03903656080365181, "train_loss_bc": 0.03903656080365181, "train_loss_llm": 0.0, "grad_norm": 0.02153075486421585, "global_step": 2086, "epoch": 57, "lr": 0.001} +{"train_loss": 0.04549497521171967, "train_loss_bc": 0.046002451330423355, "train_loss_llm": 0.0005307197570800781, "grad_norm": 0.030265459790825844, "global_step": 2087, "epoch": 57, "lr": 0.001, "val_loss": 0.05055279657244682} +{"train_loss": 0.03588956221938133, "train_loss_bc": 0.03588956221938133, "train_loss_llm": 0.0, "grad_norm": 0.03456328064203262, "global_step": 2088, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04226835072040558, "train_loss_bc": 0.04226835072040558, "train_loss_llm": 0.0, "grad_norm": 0.015916872769594193, "global_step": 2089, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04564562812447548, "train_loss_bc": 0.04564562812447548, "train_loss_llm": 0.0, "grad_norm": 0.025034401565790176, "global_step": 2090, "epoch": 58, "lr": 0.001} +{"train_loss": 0.0507822223007679, "train_loss_bc": 0.05074469372630119, "train_loss_llm": 0.0037528574466705322, "grad_norm": 0.03148740530014038, "global_step": 2091, "epoch": 58, "lr": 0.001} +{"train_loss": 0.037729233503341675, "train_loss_bc": 0.037729233503341675, "train_loss_llm": 0.0, "grad_norm": 0.034490760415792465, "global_step": 2092, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04553959518671036, "train_loss_bc": 0.04552505165338516, "train_loss_llm": 0.0014542937278747559, "grad_norm": 0.01613747328519821, "global_step": 2093, "epoch": 58, "lr": 0.001} +{"train_loss": 0.0495307631790638, "train_loss_bc": 0.04950931295752525, "train_loss_llm": 0.0021449923515319824, "grad_norm": 0.028099438175559044, "global_step": 2094, "epoch": 58, "lr": 0.001} +{"train_loss": 0.044758886098861694, "train_loss_bc": 0.044758886098861694, "train_loss_llm": 0.0, "grad_norm": 0.03588566184043884, "global_step": 2095, "epoch": 58, "lr": 0.001} +{"train_loss": 0.050322242081165314, "train_loss_bc": 0.050322242081165314, "train_loss_llm": 0.0, "grad_norm": 0.03846323490142822, "global_step": 2096, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04997435584664345, "train_loss_bc": 0.04997435584664345, "train_loss_llm": 0.0, "grad_norm": 0.01898326352238655, "global_step": 2097, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04381507635116577, "train_loss_bc": 0.04381507635116577, "train_loss_llm": 0.0, "grad_norm": 0.031470391899347305, "global_step": 2098, "epoch": 58, "lr": 0.001} +{"train_loss": 0.051107797771692276, "train_loss_bc": 0.05107211321592331, "train_loss_llm": 0.0035685598850250244, "grad_norm": 0.041852325201034546, "global_step": 2099, "epoch": 58, "lr": 0.001} +{"train_loss": 0.045534782111644745, "train_loss_bc": 0.04550322890281677, "train_loss_llm": 0.0031552910804748535, "grad_norm": 0.04998723044991493, "global_step": 2100, "epoch": 58, "lr": 0.001} +{"train_loss": 0.03710620850324631, "train_loss_bc": 0.03710530698299408, "train_loss_llm": 8.997321128845215e-05, "grad_norm": 0.016473881900310516, "global_step": 2101, "epoch": 58, "lr": 0.001} +{"train_loss": 0.0508580319583416, "train_loss_bc": 0.0508524514734745, "train_loss_llm": 0.0005582273006439209, "grad_norm": 0.029219456017017365, "global_step": 2102, "epoch": 58, "lr": 0.001} +{"train_loss": 0.050951141864061356, "train_loss_bc": 0.05092618614435196, "train_loss_llm": 0.002495676279067993, "grad_norm": 0.0489705391228199, "global_step": 2103, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04214896261692047, "train_loss_bc": 0.04214896261692047, "train_loss_llm": 0.0, "grad_norm": 0.05956723541021347, "global_step": 2104, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04591386392712593, "train_loss_bc": 0.04589167237281799, "train_loss_llm": 0.0022192299365997314, "grad_norm": 0.016360808163881302, "global_step": 2105, "epoch": 58, "lr": 0.001} +{"train_loss": 0.051184121519327164, "train_loss_bc": 0.051184121519327164, "train_loss_llm": 0.0, "grad_norm": 0.029457755386829376, "global_step": 2106, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04218747466802597, "train_loss_bc": 0.04218747466802597, "train_loss_llm": 0.0, "grad_norm": 0.04021252691745758, "global_step": 2107, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04494987428188324, "train_loss_bc": 0.04494987428188324, "train_loss_llm": 0.0, "grad_norm": 0.038324106484651566, "global_step": 2108, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04531480744481087, "train_loss_bc": 0.04531480744481087, "train_loss_llm": 0.0, "grad_norm": 0.014402591623365879, "global_step": 2109, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04555274173617363, "train_loss_bc": 0.04555274173617363, "train_loss_llm": 0.0, "grad_norm": 0.02261495590209961, "global_step": 2110, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04243561998009682, "train_loss_bc": 0.04239781200885773, "train_loss_llm": 0.003780931234359741, "grad_norm": 0.023629236966371536, "global_step": 2111, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04480605572462082, "train_loss_bc": 0.044799283146858215, "train_loss_llm": 0.0006773769855499268, "grad_norm": 0.029853418469429016, "global_step": 2112, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04700917750597, "train_loss_bc": 0.04700917750597, "train_loss_llm": 0.0, "grad_norm": 0.01996341161429882, "global_step": 2113, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04907907173037529, "train_loss_bc": 0.049075398594141006, "train_loss_llm": 0.0003674030303955078, "grad_norm": 0.031750768423080444, "global_step": 2114, "epoch": 58, "lr": 0.001} +{"train_loss": 0.034242164343595505, "train_loss_bc": 0.034242164343595505, "train_loss_llm": 0.0, "grad_norm": 0.0401189811527729, "global_step": 2115, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04335678368806839, "train_loss_bc": 0.04335678368806839, "train_loss_llm": 0.0, "grad_norm": 0.04371991753578186, "global_step": 2116, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04519977048039436, "train_loss_bc": 0.04519977048039436, "train_loss_llm": 0.0, "grad_norm": 0.021657658740878105, "global_step": 2117, "epoch": 58, "lr": 0.001} +{"train_loss": 0.047740232199430466, "train_loss_bc": 0.047730084508657455, "train_loss_llm": 0.001014798879623413, "grad_norm": 0.03643082082271576, "global_step": 2118, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04663308337330818, "train_loss_bc": 0.04663308337330818, "train_loss_llm": 0.0, "grad_norm": 0.04818258434534073, "global_step": 2119, "epoch": 58, "lr": 0.001} +{"train_loss": 0.05013422295451164, "train_loss_bc": 0.05013422295451164, "train_loss_llm": 0.0, "grad_norm": 0.05735187232494354, "global_step": 2120, "epoch": 58, "lr": 0.001} +{"train_loss": 0.050258368253707886, "train_loss_bc": 0.050258368253707886, "train_loss_llm": 0.0, "grad_norm": 0.01986224576830864, "global_step": 2121, "epoch": 58, "lr": 0.001} +{"train_loss": 0.043595410883426666, "train_loss_bc": 0.043595410883426666, "train_loss_llm": 0.0, "grad_norm": 0.03813163936138153, "global_step": 2122, "epoch": 58, "lr": 0.001} +{"train_loss": 0.04560526315536764, "train_loss_bc": 0.04823378846049309, "train_loss_llm": 0.0, "grad_norm": 0.04903056472539902, "global_step": 2123, "epoch": 58, "lr": 0.001, "val_loss": 0.050953324884176254} +{"train_loss": 0.03971267119050026, "train_loss_bc": 0.03971267119050026, "train_loss_llm": 0.0, "grad_norm": 0.06275077909231186, "global_step": 2124, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04472501203417778, "train_loss_bc": 0.04471452906727791, "train_loss_llm": 0.0010482966899871826, "grad_norm": 0.016407372429966927, "global_step": 2125, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04364771768450737, "train_loss_bc": 0.04364771768450737, "train_loss_llm": 0.0, "grad_norm": 0.021044505760073662, "global_step": 2126, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04313473403453827, "train_loss_bc": 0.043118853121995926, "train_loss_llm": 0.0015879273414611816, "grad_norm": 0.027153117582201958, "global_step": 2127, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04134605824947357, "train_loss_bc": 0.04131791368126869, "train_loss_llm": 0.0028145015239715576, "grad_norm": 0.03151678293943405, "global_step": 2128, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04853461682796478, "train_loss_bc": 0.04853461682796478, "train_loss_llm": 0.0, "grad_norm": 0.016805296763777733, "global_step": 2129, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04678579419851303, "train_loss_bc": 0.04676462709903717, "train_loss_llm": 0.0021167397499084473, "grad_norm": 0.023635756224393845, "global_step": 2130, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04295666143298149, "train_loss_bc": 0.0429471880197525, "train_loss_llm": 0.000947415828704834, "grad_norm": 0.03171030804514885, "global_step": 2131, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04653787985444069, "train_loss_bc": 0.04653787985444069, "train_loss_llm": 0.0, "grad_norm": 0.040104445070028305, "global_step": 2132, "epoch": 59, "lr": 0.001} +{"train_loss": 0.0550033338367939, "train_loss_bc": 0.054981786757707596, "train_loss_llm": 0.002154827117919922, "grad_norm": 0.01608707569539547, "global_step": 2133, "epoch": 59, "lr": 0.001} +{"train_loss": 0.0423889234662056, "train_loss_bc": 0.04235672205686569, "train_loss_llm": 0.0032201409339904785, "grad_norm": 0.027132434770464897, "global_step": 2134, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04734916612505913, "train_loss_bc": 0.04731477424502373, "train_loss_llm": 0.0034391582012176514, "grad_norm": 0.039702169597148895, "global_step": 2135, "epoch": 59, "lr": 0.001} +{"train_loss": 0.045414362102746964, "train_loss_bc": 0.045414362102746964, "train_loss_llm": 0.0, "grad_norm": 0.04797899350523949, "global_step": 2136, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04079076275229454, "train_loss_bc": 0.04079076275229454, "train_loss_llm": 0.0, "grad_norm": 0.01510319672524929, "global_step": 2137, "epoch": 59, "lr": 0.001} +{"train_loss": 0.049334004521369934, "train_loss_bc": 0.049334004521369934, "train_loss_llm": 0.0, "grad_norm": 0.025506265461444855, "global_step": 2138, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04740911349654198, "train_loss_bc": 0.04740911349654198, "train_loss_llm": 0.0, "grad_norm": 0.03730475902557373, "global_step": 2139, "epoch": 59, "lr": 0.001} +{"train_loss": 0.03986413776874542, "train_loss_bc": 0.03986413776874542, "train_loss_llm": 0.0, "grad_norm": 0.04445119574666023, "global_step": 2140, "epoch": 59, "lr": 0.001} +{"train_loss": 0.045228827744722366, "train_loss_bc": 0.045228827744722366, "train_loss_llm": 0.0, "grad_norm": 0.017759576439857483, "global_step": 2141, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04156125709414482, "train_loss_bc": 0.04156125709414482, "train_loss_llm": 0.0, "grad_norm": 0.026268761605024338, "global_step": 2142, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04647501930594444, "train_loss_bc": 0.046444330364465714, "train_loss_llm": 0.0030689537525177, "grad_norm": 0.03377841040492058, "global_step": 2143, "epoch": 59, "lr": 0.001} +{"train_loss": 0.05058307200670242, "train_loss_bc": 0.05058307200670242, "train_loss_llm": 0.0, "grad_norm": 0.03725048899650574, "global_step": 2144, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04857836663722992, "train_loss_bc": 0.04857836663722992, "train_loss_llm": 0.0, "grad_norm": 0.016730625182390213, "global_step": 2145, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04684538394212723, "train_loss_bc": 0.04684538394212723, "train_loss_llm": 0.0, "grad_norm": 0.02772475965321064, "global_step": 2146, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04429307207465172, "train_loss_bc": 0.04429307207465172, "train_loss_llm": 0.0, "grad_norm": 0.0358239971101284, "global_step": 2147, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04491807520389557, "train_loss_bc": 0.044891875237226486, "train_loss_llm": 0.002620100975036621, "grad_norm": 0.04381736367940903, "global_step": 2148, "epoch": 59, "lr": 0.001} +{"train_loss": 0.048510510474443436, "train_loss_bc": 0.04850027337670326, "train_loss_llm": 0.0010236799716949463, "grad_norm": 0.01849202811717987, "global_step": 2149, "epoch": 59, "lr": 0.001} +{"train_loss": 0.05388058349490166, "train_loss_bc": 0.05388058349490166, "train_loss_llm": 0.0, "grad_norm": 0.026210898533463478, "global_step": 2150, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04791119322180748, "train_loss_bc": 0.04791119322180748, "train_loss_llm": 0.0, "grad_norm": 0.031799595803022385, "global_step": 2151, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04017337039113045, "train_loss_bc": 0.040172286331653595, "train_loss_llm": 0.00010856986045837402, "grad_norm": 0.04083440825343132, "global_step": 2152, "epoch": 59, "lr": 0.001} +{"train_loss": 0.05438520014286041, "train_loss_bc": 0.05438520014286041, "train_loss_llm": 0.0, "grad_norm": 0.01940612867474556, "global_step": 2153, "epoch": 59, "lr": 0.001} +{"train_loss": 0.035332392901182175, "train_loss_bc": 0.03532503917813301, "train_loss_llm": 0.0007354915142059326, "grad_norm": 0.02901085652410984, "global_step": 2154, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04495500028133392, "train_loss_bc": 0.04495500028133392, "train_loss_llm": 0.0, "grad_norm": 0.03706982359290123, "global_step": 2155, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04544549807906151, "train_loss_bc": 0.04543520137667656, "train_loss_llm": 0.0010297298431396484, "grad_norm": 0.045195311307907104, "global_step": 2156, "epoch": 59, "lr": 0.001} +{"train_loss": 0.05162278935313225, "train_loss_bc": 0.05162278935313225, "train_loss_llm": 0.0, "grad_norm": 0.017061008140444756, "global_step": 2157, "epoch": 59, "lr": 0.001} +{"train_loss": 0.0361119769513607, "train_loss_bc": 0.036101333796978, "train_loss_llm": 0.0010643303394317627, "grad_norm": 0.02928754687309265, "global_step": 2158, "epoch": 59, "lr": 0.001} +{"train_loss": 0.04547004122287035, "train_loss_bc": 0.04517494514584541, "train_loss_llm": 0.0, "grad_norm": 0.030025841668248177, "global_step": 2159, "epoch": 59, "lr": 0.001, "val_loss": 0.05071645230054855} +{"train_loss": 0.0390535444021225, "train_loss_bc": 0.039045438170433044, "train_loss_llm": 0.000810772180557251, "grad_norm": 0.03513983637094498, "global_step": 2160, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04786930978298187, "train_loss_bc": 0.047847867012023926, "train_loss_llm": 0.002144366502761841, "grad_norm": 0.01774393767118454, "global_step": 2161, "epoch": 60, "lr": 0.001} +{"train_loss": 0.0435602180659771, "train_loss_bc": 0.043509725481271744, "train_loss_llm": 0.005049288272857666, "grad_norm": 0.03068171627819538, "global_step": 2162, "epoch": 60, "lr": 0.001} +{"train_loss": 0.03850962966680527, "train_loss_bc": 0.03849544748663902, "train_loss_llm": 0.0014181733131408691, "grad_norm": 0.03694119304418564, "global_step": 2163, "epoch": 60, "lr": 0.001} +{"train_loss": 0.042753782123327255, "train_loss_bc": 0.042753782123327255, "train_loss_llm": 0.0, "grad_norm": 0.04244448244571686, "global_step": 2164, "epoch": 60, "lr": 0.001} +{"train_loss": 0.05016350373625755, "train_loss_bc": 0.05016350373625755, "train_loss_llm": 0.0, "grad_norm": 0.017214538529515266, "global_step": 2165, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04190122336149216, "train_loss_bc": 0.04190122336149216, "train_loss_llm": 0.0, "grad_norm": 0.030905665829777718, "global_step": 2166, "epoch": 60, "lr": 0.001} +{"train_loss": 0.0458633154630661, "train_loss_bc": 0.0458633154630661, "train_loss_llm": 0.0, "grad_norm": 0.040200330317020416, "global_step": 2167, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04028274863958359, "train_loss_bc": 0.04028274863958359, "train_loss_llm": 0.0, "grad_norm": 0.056720003485679626, "global_step": 2168, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04682324826717377, "train_loss_bc": 0.04682324826717377, "train_loss_llm": 0.0, "grad_norm": 0.016704602167010307, "global_step": 2169, "epoch": 60, "lr": 0.001} +{"train_loss": 0.0468536838889122, "train_loss_bc": 0.0468536838889122, "train_loss_llm": 0.0, "grad_norm": 0.027647927403450012, "global_step": 2170, "epoch": 60, "lr": 0.001} +{"train_loss": 0.03833683207631111, "train_loss_bc": 0.03833683207631111, "train_loss_llm": 0.0, "grad_norm": 0.03362717106938362, "global_step": 2171, "epoch": 60, "lr": 0.001} +{"train_loss": 0.0485457144677639, "train_loss_bc": 0.0485457144677639, "train_loss_llm": 0.0, "grad_norm": 0.04275202378630638, "global_step": 2172, "epoch": 60, "lr": 0.001} +{"train_loss": 0.049834150820970535, "train_loss_bc": 0.049834150820970535, "train_loss_llm": 0.0, "grad_norm": 0.01613820344209671, "global_step": 2173, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04759204760193825, "train_loss_bc": 0.047579918056726456, "train_loss_llm": 0.0012128055095672607, "grad_norm": 0.02203342504799366, "global_step": 2174, "epoch": 60, "lr": 0.001} +{"train_loss": 0.044349901378154755, "train_loss_bc": 0.044349901378154755, "train_loss_llm": 0.0, "grad_norm": 0.028511954471468925, "global_step": 2175, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04297468811273575, "train_loss_bc": 0.04297468811273575, "train_loss_llm": 0.0, "grad_norm": 0.02925259992480278, "global_step": 2176, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04357149451971054, "train_loss_bc": 0.04357149451971054, "train_loss_llm": 0.0, "grad_norm": 0.012562175281345844, "global_step": 2177, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04749550670385361, "train_loss_bc": 0.04749550670385361, "train_loss_llm": 0.0, "grad_norm": 0.026333998888731003, "global_step": 2178, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04724336788058281, "train_loss_bc": 0.04724336788058281, "train_loss_llm": 0.0, "grad_norm": 0.030488699674606323, "global_step": 2179, "epoch": 60, "lr": 0.001} +{"train_loss": 0.05058322846889496, "train_loss_bc": 0.05057716742157936, "train_loss_llm": 0.0006059408187866211, "grad_norm": 0.04017098248004913, "global_step": 2180, "epoch": 60, "lr": 0.001} +{"train_loss": 0.03942357003688812, "train_loss_bc": 0.03940590098500252, "train_loss_llm": 0.0017669498920440674, "grad_norm": 0.015234542079269886, "global_step": 2181, "epoch": 60, "lr": 0.001} +{"train_loss": 0.05225016176700592, "train_loss_bc": 0.05223206430673599, "train_loss_llm": 0.0018098056316375732, "grad_norm": 0.026212472468614578, "global_step": 2182, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04470619186758995, "train_loss_bc": 0.044705912470817566, "train_loss_llm": 2.7835369110107422e-05, "grad_norm": 0.03256138041615486, "global_step": 2183, "epoch": 60, "lr": 0.001} +{"train_loss": 0.052641887217760086, "train_loss_bc": 0.052641887217760086, "train_loss_llm": 0.0, "grad_norm": 0.0334806852042675, "global_step": 2184, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04708521440625191, "train_loss_bc": 0.04708521440625191, "train_loss_llm": 0.0, "grad_norm": 0.01938379369676113, "global_step": 2185, "epoch": 60, "lr": 0.001} +{"train_loss": 0.047727808356285095, "train_loss_bc": 0.047718532383441925, "train_loss_llm": 0.000927656888961792, "grad_norm": 0.02575608342885971, "global_step": 2186, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04609500989317894, "train_loss_bc": 0.04609500989317894, "train_loss_llm": 0.0, "grad_norm": 0.04235449433326721, "global_step": 2187, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04609217494726181, "train_loss_bc": 0.046070411801338196, "train_loss_llm": 0.0021763741970062256, "grad_norm": 0.04650083929300308, "global_step": 2188, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04356422275304794, "train_loss_bc": 0.04355776682496071, "train_loss_llm": 0.0006455183029174805, "grad_norm": 0.015535175800323486, "global_step": 2189, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04316946119070053, "train_loss_bc": 0.04316946119070053, "train_loss_llm": 0.0, "grad_norm": 0.027513446286320686, "global_step": 2190, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04232224076986313, "train_loss_bc": 0.04232224076986313, "train_loss_llm": 0.0, "grad_norm": 0.039466727524995804, "global_step": 2191, "epoch": 60, "lr": 0.001} +{"train_loss": 0.043930184096097946, "train_loss_bc": 0.043930184096097946, "train_loss_llm": 0.0, "grad_norm": 0.04344657063484192, "global_step": 2192, "epoch": 60, "lr": 0.001} +{"train_loss": 0.04375186935067177, "train_loss_bc": 0.04375186935067177, "train_loss_llm": 0.0, "grad_norm": 0.018782472237944603, "global_step": 2193, "epoch": 60, "lr": 0.001} +{"train_loss": 0.05007369816303253, "train_loss_bc": 0.05007369816303253, "train_loss_llm": 0.0, "grad_norm": 0.02743283472955227, "global_step": 2194, "epoch": 60, "lr": 0.001} +{"train_loss": 0.045453929238849215, "train_loss_bc": 0.049346618354320526, "train_loss_llm": 0.0, "grad_norm": 0.03682214766740799, "global_step": 2195, "epoch": 60, "lr": 0.001, "train/cumulative_reward": 46.7060686459547, "train/mean_score": 7.621246964420176, "train/success_rate": 0.2, "test/cumulative_reward": 81.86032109930579, "test/mean_score": 6.587902389032052, "test/success_rate": 0.08, "val_loss": 0.05124939978122711, "train_action_mse_error": 0.04890860244631767} +{"train_loss": 0.03867773339152336, "train_loss_bc": 0.03867773339152336, "train_loss_llm": 0.0, "grad_norm": 0.041580986231565475, "global_step": 2196, "epoch": 61, "lr": 0.001} +{"train_loss": 0.050792280584573746, "train_loss_bc": 0.05076908692717552, "train_loss_llm": 0.0023192167282104492, "grad_norm": 0.015155038796365261, "global_step": 2197, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04740459471940994, "train_loss_bc": 0.047387540340423584, "train_loss_llm": 0.0017053782939910889, "grad_norm": 0.02331746369600296, "global_step": 2198, "epoch": 61, "lr": 0.001} +{"train_loss": 0.037329062819480896, "train_loss_bc": 0.037329062819480896, "train_loss_llm": 0.0, "grad_norm": 0.027288157492876053, "global_step": 2199, "epoch": 61, "lr": 0.001} +{"train_loss": 0.05717117339372635, "train_loss_bc": 0.05717117339372635, "train_loss_llm": 0.0, "grad_norm": 0.03560720756649971, "global_step": 2200, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04860275238752365, "train_loss_bc": 0.04859566688537598, "train_loss_llm": 0.0007085800170898438, "grad_norm": 0.018307937309145927, "global_step": 2201, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04796784743666649, "train_loss_bc": 0.04795261099934578, "train_loss_llm": 0.0015235543251037598, "grad_norm": 0.024815283715724945, "global_step": 2202, "epoch": 61, "lr": 0.001} +{"train_loss": 0.042824968695640564, "train_loss_bc": 0.04280341789126396, "train_loss_llm": 0.002155095338821411, "grad_norm": 0.02986331842839718, "global_step": 2203, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04399019479751587, "train_loss_bc": 0.04399019479751587, "train_loss_llm": 0.0, "grad_norm": 0.03363265469670296, "global_step": 2204, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04613328352570534, "train_loss_bc": 0.04613328352570534, "train_loss_llm": 0.0, "grad_norm": 0.014615556225180626, "global_step": 2205, "epoch": 61, "lr": 0.001} +{"train_loss": 0.05609097704291344, "train_loss_bc": 0.05608202889561653, "train_loss_llm": 0.0008948743343353271, "grad_norm": 0.0245396476238966, "global_step": 2206, "epoch": 61, "lr": 0.001} +{"train_loss": 0.03699437156319618, "train_loss_bc": 0.03699437156319618, "train_loss_llm": 0.0, "grad_norm": 0.03318902105093002, "global_step": 2207, "epoch": 61, "lr": 0.001} +{"train_loss": 0.03992731496691704, "train_loss_bc": 0.0399167500436306, "train_loss_llm": 0.0010566413402557373, "grad_norm": 0.043267570436000824, "global_step": 2208, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04753080755472183, "train_loss_bc": 0.04751624912023544, "train_loss_llm": 0.001455932855606079, "grad_norm": 0.01707116886973381, "global_step": 2209, "epoch": 61, "lr": 0.001} +{"train_loss": 0.042507801204919815, "train_loss_bc": 0.042507801204919815, "train_loss_llm": 0.0, "grad_norm": 0.02636059559881687, "global_step": 2210, "epoch": 61, "lr": 0.001} +{"train_loss": 0.045710157603025436, "train_loss_bc": 0.045689672231674194, "train_loss_llm": 0.0020485520362854004, "grad_norm": 0.030019525438547134, "global_step": 2211, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04450197517871857, "train_loss_bc": 0.04450197517871857, "train_loss_llm": 0.0, "grad_norm": 0.03631442412734032, "global_step": 2212, "epoch": 61, "lr": 0.001} +{"train_loss": 0.05270087346434593, "train_loss_bc": 0.05270087346434593, "train_loss_llm": 0.0, "grad_norm": 0.016244901344180107, "global_step": 2213, "epoch": 61, "lr": 0.001} +{"train_loss": 0.045673929154872894, "train_loss_bc": 0.045673929154872894, "train_loss_llm": 0.0, "grad_norm": 0.02476394549012184, "global_step": 2214, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04425322264432907, "train_loss_bc": 0.04425322264432907, "train_loss_llm": 0.0, "grad_norm": 0.03171110525727272, "global_step": 2215, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04012245684862137, "train_loss_bc": 0.04012245684862137, "train_loss_llm": 0.0, "grad_norm": 0.03812238574028015, "global_step": 2216, "epoch": 61, "lr": 0.001} +{"train_loss": 0.0387505404651165, "train_loss_bc": 0.0387505404651165, "train_loss_llm": 0.0, "grad_norm": 0.012577529065310955, "global_step": 2217, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04213463515043259, "train_loss_bc": 0.04213463515043259, "train_loss_llm": 0.0, "grad_norm": 0.018643684685230255, "global_step": 2218, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04729730263352394, "train_loss_bc": 0.04729730263352394, "train_loss_llm": 0.0, "grad_norm": 0.02877674251794815, "global_step": 2219, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04436720907688141, "train_loss_bc": 0.04436720907688141, "train_loss_llm": 0.0, "grad_norm": 0.0406700074672699, "global_step": 2220, "epoch": 61, "lr": 0.001} +{"train_loss": 0.042832911014556885, "train_loss_bc": 0.04282534494996071, "train_loss_llm": 0.0007565915584564209, "grad_norm": 0.014847216196358204, "global_step": 2221, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04735127463936806, "train_loss_bc": 0.047258783131837845, "train_loss_llm": 0.009249210357666016, "grad_norm": 0.02677248604595661, "global_step": 2222, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04458966478705406, "train_loss_bc": 0.04458966478705406, "train_loss_llm": 0.0, "grad_norm": 0.031512822955846786, "global_step": 2223, "epoch": 61, "lr": 0.001} +{"train_loss": 0.041669830679893494, "train_loss_bc": 0.041669830679893494, "train_loss_llm": 0.0, "grad_norm": 0.03972310572862625, "global_step": 2224, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04256374388933182, "train_loss_bc": 0.04256374388933182, "train_loss_llm": 0.0, "grad_norm": 0.015732141211628914, "global_step": 2225, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04736654832959175, "train_loss_bc": 0.04736654832959175, "train_loss_llm": 0.0, "grad_norm": 0.024508940055966377, "global_step": 2226, "epoch": 61, "lr": 0.001} +{"train_loss": 0.03604898974299431, "train_loss_bc": 0.036040302366018295, "train_loss_llm": 0.0008688271045684814, "grad_norm": 0.03180632367730141, "global_step": 2227, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04840434342622757, "train_loss_bc": 0.04840434342622757, "train_loss_llm": 0.0, "grad_norm": 0.032652851194143295, "global_step": 2228, "epoch": 61, "lr": 0.001} +{"train_loss": 0.052530527114868164, "train_loss_bc": 0.0525270476937294, "train_loss_llm": 0.000347822904586792, "grad_norm": 0.018833795562386513, "global_step": 2229, "epoch": 61, "lr": 0.001} +{"train_loss": 0.045790016651153564, "train_loss_bc": 0.045790016651153564, "train_loss_llm": 0.0, "grad_norm": 0.024322621524333954, "global_step": 2230, "epoch": 61, "lr": 0.001} +{"train_loss": 0.04535280975202719, "train_loss_bc": 0.05403111129999161, "train_loss_llm": 0.006472468376159668, "grad_norm": 0.0352947898209095, "global_step": 2231, "epoch": 61, "lr": 0.001, "val_loss": 0.0506034716963768} +{"train_loss": 0.03599466010928154, "train_loss_bc": 0.03597242385149002, "train_loss_llm": 0.0022235512733459473, "grad_norm": 0.04255323484539986, "global_step": 2232, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04565875604748726, "train_loss_bc": 0.04565875604748726, "train_loss_llm": 0.0, "grad_norm": 0.012964505702257156, "global_step": 2233, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04340394213795662, "train_loss_bc": 0.04340150207281113, "train_loss_llm": 0.0002440810203552246, "grad_norm": 0.021874457597732544, "global_step": 2234, "epoch": 62, "lr": 0.001} +{"train_loss": 0.040591735392808914, "train_loss_bc": 0.040591735392808914, "train_loss_llm": 0.0, "grad_norm": 0.0281640887260437, "global_step": 2235, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04404140263795853, "train_loss_bc": 0.04404140263795853, "train_loss_llm": 0.0, "grad_norm": 0.03667512163519859, "global_step": 2236, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04792796075344086, "train_loss_bc": 0.04792796075344086, "train_loss_llm": 0.0, "grad_norm": 0.01427640113979578, "global_step": 2237, "epoch": 62, "lr": 0.001} +{"train_loss": 0.03892459720373154, "train_loss_bc": 0.038917940109968185, "train_loss_llm": 0.0006656348705291748, "grad_norm": 0.018841151148080826, "global_step": 2238, "epoch": 62, "lr": 0.001} +{"train_loss": 0.047527752816677094, "train_loss_bc": 0.047527752816677094, "train_loss_llm": 0.0, "grad_norm": 0.03056022897362709, "global_step": 2239, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04656987264752388, "train_loss_bc": 0.04656987264752388, "train_loss_llm": 0.0, "grad_norm": 0.0363137423992157, "global_step": 2240, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04832207038998604, "train_loss_bc": 0.048321664333343506, "train_loss_llm": 4.06801700592041e-05, "grad_norm": 0.02202332578599453, "global_step": 2241, "epoch": 62, "lr": 0.001} +{"train_loss": 0.041998133063316345, "train_loss_bc": 0.041998133063316345, "train_loss_llm": 0.0, "grad_norm": 0.031121177598834038, "global_step": 2242, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04631128907203674, "train_loss_bc": 0.04631128907203674, "train_loss_llm": 0.0, "grad_norm": 0.0356995165348053, "global_step": 2243, "epoch": 62, "lr": 0.001} +{"train_loss": 0.03807923570275307, "train_loss_bc": 0.03807184845209122, "train_loss_llm": 0.0007388889789581299, "grad_norm": 0.04407022148370743, "global_step": 2244, "epoch": 62, "lr": 0.001} +{"train_loss": 0.0472123883664608, "train_loss_bc": 0.04718024656176567, "train_loss_llm": 0.003214210271835327, "grad_norm": 0.017737900838255882, "global_step": 2245, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04779309034347534, "train_loss_bc": 0.04779309034347534, "train_loss_llm": 0.0, "grad_norm": 0.02673458680510521, "global_step": 2246, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04659383371472359, "train_loss_bc": 0.04659383371472359, "train_loss_llm": 0.0, "grad_norm": 0.032426029443740845, "global_step": 2247, "epoch": 62, "lr": 0.001} +{"train_loss": 0.051064103841781616, "train_loss_bc": 0.051047418266534805, "train_loss_llm": 0.001668393611907959, "grad_norm": 0.04010750725865364, "global_step": 2248, "epoch": 62, "lr": 0.001} +{"train_loss": 0.040303152054548264, "train_loss_bc": 0.040271125733852386, "train_loss_llm": 0.0032025575637817383, "grad_norm": 0.01527405809611082, "global_step": 2249, "epoch": 62, "lr": 0.001} +{"train_loss": 0.040573231875896454, "train_loss_bc": 0.04054338485002518, "train_loss_llm": 0.002984851598739624, "grad_norm": 0.024109112098813057, "global_step": 2250, "epoch": 62, "lr": 0.001} +{"train_loss": 0.042476214468479156, "train_loss_bc": 0.042476214468479156, "train_loss_llm": 0.0, "grad_norm": 0.034029822796583176, "global_step": 2251, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04630238190293312, "train_loss_bc": 0.04630238190293312, "train_loss_llm": 0.0, "grad_norm": 0.03821173310279846, "global_step": 2252, "epoch": 62, "lr": 0.001} +{"train_loss": 0.0418848916888237, "train_loss_bc": 0.0418848916888237, "train_loss_llm": 0.0, "grad_norm": 0.014902028255164623, "global_step": 2253, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04057185724377632, "train_loss_bc": 0.04057185724377632, "train_loss_llm": 0.0, "grad_norm": 0.020415833219885826, "global_step": 2254, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04929189383983612, "train_loss_bc": 0.04926165193319321, "train_loss_llm": 0.0030240416526794434, "grad_norm": 0.020370477810502052, "global_step": 2255, "epoch": 62, "lr": 0.001} +{"train_loss": 0.047641459852457047, "train_loss_bc": 0.0476355142891407, "train_loss_llm": 0.00059470534324646, "grad_norm": 0.029347818344831467, "global_step": 2256, "epoch": 62, "lr": 0.001} +{"train_loss": 0.05066914111375809, "train_loss_bc": 0.05066914111375809, "train_loss_llm": 0.0, "grad_norm": 0.016861826181411743, "global_step": 2257, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04720235615968704, "train_loss_bc": 0.04720235615968704, "train_loss_llm": 0.0, "grad_norm": 0.02638622373342514, "global_step": 2258, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04343660920858383, "train_loss_bc": 0.04343660920858383, "train_loss_llm": 0.0, "grad_norm": 0.03512739762663841, "global_step": 2259, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04683000594377518, "train_loss_bc": 0.04680214822292328, "train_loss_llm": 0.002785921096801758, "grad_norm": 0.047938380390405655, "global_step": 2260, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04969439655542374, "train_loss_bc": 0.04969439655542374, "train_loss_llm": 0.0, "grad_norm": 0.02095196582376957, "global_step": 2261, "epoch": 62, "lr": 0.001} +{"train_loss": 0.04951462149620056, "train_loss_bc": 0.04951462149620056, "train_loss_llm": 0.0, "grad_norm": 0.032402317970991135, "global_step": 2262, "epoch": 62, "lr": 0.001} +{"train_loss": 0.049859967082738876, "train_loss_bc": 0.049853380769491196, "train_loss_llm": 0.0006587803363800049, "grad_norm": 0.04149309918284416, "global_step": 2263, "epoch": 62, "lr": 0.001} +{"train_loss": 0.03876999765634537, "train_loss_bc": 0.03876999765634537, "train_loss_llm": 0.0, "grad_norm": 0.04324797913432121, "global_step": 2264, "epoch": 62, "lr": 0.001} +{"train_loss": 0.0531301274895668, "train_loss_bc": 0.0531301274895668, "train_loss_llm": 0.0, "grad_norm": 0.020468665286898613, "global_step": 2265, "epoch": 62, "lr": 0.001} +{"train_loss": 0.044280968606472015, "train_loss_bc": 0.044280968606472015, "train_loss_llm": 0.0, "grad_norm": 0.02344564162194729, "global_step": 2266, "epoch": 62, "lr": 0.001} +{"train_loss": 0.045262205310993724, "train_loss_bc": 0.04899129271507263, "train_loss_llm": 0.0, "grad_norm": 0.029723362997174263, "global_step": 2267, "epoch": 62, "lr": 0.001, "val_loss": 0.05092225223779678} +{"train_loss": 0.049217093735933304, "train_loss_bc": 0.049217093735933304, "train_loss_llm": 0.0, "grad_norm": 0.037928011268377304, "global_step": 2268, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04821409285068512, "train_loss_bc": 0.0482034757733345, "train_loss_llm": 0.0010617375373840332, "grad_norm": 0.015084990300238132, "global_step": 2269, "epoch": 63, "lr": 0.001} +{"train_loss": 0.047517407685518265, "train_loss_bc": 0.04750436544418335, "train_loss_llm": 0.0013040602207183838, "grad_norm": 0.024894190952181816, "global_step": 2270, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04545603692531586, "train_loss_bc": 0.04545603692531586, "train_loss_llm": 0.0, "grad_norm": 0.03213087469339371, "global_step": 2271, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04962678998708725, "train_loss_bc": 0.04961202293634415, "train_loss_llm": 0.0014768540859222412, "grad_norm": 0.03664582595229149, "global_step": 2272, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04610303416848183, "train_loss_bc": 0.04610303416848183, "train_loss_llm": 0.0, "grad_norm": 0.014384722337126732, "global_step": 2273, "epoch": 63, "lr": 0.001} +{"train_loss": 0.03990856930613518, "train_loss_bc": 0.03990856930613518, "train_loss_llm": 0.0, "grad_norm": 0.022059233859181404, "global_step": 2274, "epoch": 63, "lr": 0.001} +{"train_loss": 0.044489771127700806, "train_loss_bc": 0.044489771127700806, "train_loss_llm": 0.0, "grad_norm": 0.027924593538045883, "global_step": 2275, "epoch": 63, "lr": 0.001} +{"train_loss": 0.03920114412903786, "train_loss_bc": 0.039190199226140976, "train_loss_llm": 0.001094430685043335, "grad_norm": 0.03527709096670151, "global_step": 2276, "epoch": 63, "lr": 0.001} +{"train_loss": 0.05469786748290062, "train_loss_bc": 0.05467597767710686, "train_loss_llm": 0.0021889805793762207, "grad_norm": 0.02155429869890213, "global_step": 2277, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04920731112360954, "train_loss_bc": 0.04920731112360954, "train_loss_llm": 0.0, "grad_norm": 0.03752056509256363, "global_step": 2278, "epoch": 63, "lr": 0.001} +{"train_loss": 0.03800250589847565, "train_loss_bc": 0.03800250589847565, "train_loss_llm": 0.0, "grad_norm": 0.04118826612830162, "global_step": 2279, "epoch": 63, "lr": 0.001} +{"train_loss": 0.0375075601041317, "train_loss_bc": 0.0375075601041317, "train_loss_llm": 0.0, "grad_norm": 0.03681108355522156, "global_step": 2280, "epoch": 63, "lr": 0.001} +{"train_loss": 0.043004926294088364, "train_loss_bc": 0.04300076514482498, "train_loss_llm": 0.00041624903678894043, "grad_norm": 0.020038792863488197, "global_step": 2281, "epoch": 63, "lr": 0.001} +{"train_loss": 0.0450894720852375, "train_loss_bc": 0.0450894720852375, "train_loss_llm": 0.0, "grad_norm": 0.02719101309776306, "global_step": 2282, "epoch": 63, "lr": 0.001} +{"train_loss": 0.0425359308719635, "train_loss_bc": 0.0425359308719635, "train_loss_llm": 0.0, "grad_norm": 0.03540336340665817, "global_step": 2283, "epoch": 63, "lr": 0.001} +{"train_loss": 0.03650218993425369, "train_loss_bc": 0.036500055342912674, "train_loss_llm": 0.00021329522132873535, "grad_norm": 0.03739197179675102, "global_step": 2284, "epoch": 63, "lr": 0.001} +{"train_loss": 0.03989453986287117, "train_loss_bc": 0.03989453986287117, "train_loss_llm": 0.0, "grad_norm": 0.019987229257822037, "global_step": 2285, "epoch": 63, "lr": 0.001} +{"train_loss": 0.049309320747852325, "train_loss_bc": 0.049295999109745026, "train_loss_llm": 0.0013319849967956543, "grad_norm": 0.02768006920814514, "global_step": 2286, "epoch": 63, "lr": 0.001} +{"train_loss": 0.057478148490190506, "train_loss_bc": 0.057478148490190506, "train_loss_llm": 0.0, "grad_norm": 0.03714057058095932, "global_step": 2287, "epoch": 63, "lr": 0.001} +{"train_loss": 0.0417337641119957, "train_loss_bc": 0.0417337641119957, "train_loss_llm": 0.0, "grad_norm": 0.043537016957998276, "global_step": 2288, "epoch": 63, "lr": 0.001} +{"train_loss": 0.03527016192674637, "train_loss_bc": 0.03527016192674637, "train_loss_llm": 0.0, "grad_norm": 0.013825763948261738, "global_step": 2289, "epoch": 63, "lr": 0.001} +{"train_loss": 0.039537206292152405, "train_loss_bc": 0.03952591121196747, "train_loss_llm": 0.0011295974254608154, "grad_norm": 0.016322892159223557, "global_step": 2290, "epoch": 63, "lr": 0.001} +{"train_loss": 0.040125735104084015, "train_loss_bc": 0.0401131734251976, "train_loss_llm": 0.0012561380863189697, "grad_norm": 0.0229828841984272, "global_step": 2291, "epoch": 63, "lr": 0.001} +{"train_loss": 0.037743084132671356, "train_loss_bc": 0.03771417215466499, "train_loss_llm": 0.0028912723064422607, "grad_norm": 0.03065524809062481, "global_step": 2292, "epoch": 63, "lr": 0.001} +{"train_loss": 0.048826221376657486, "train_loss_bc": 0.048826221376657486, "train_loss_llm": 0.0, "grad_norm": 0.013918997719883919, "global_step": 2293, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04895305261015892, "train_loss_bc": 0.04895305261015892, "train_loss_llm": 0.0, "grad_norm": 0.025384606793522835, "global_step": 2294, "epoch": 63, "lr": 0.001} +{"train_loss": 0.053944092243909836, "train_loss_bc": 0.053939394652843475, "train_loss_llm": 0.0004697144031524658, "grad_norm": 0.03130576014518738, "global_step": 2295, "epoch": 63, "lr": 0.001} +{"train_loss": 0.049133677035570145, "train_loss_bc": 0.049133677035570145, "train_loss_llm": 0.0, "grad_norm": 0.04122794792056084, "global_step": 2296, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04640349745750427, "train_loss_bc": 0.04640349745750427, "train_loss_llm": 0.0, "grad_norm": 0.015398940071463585, "global_step": 2297, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04988919198513031, "train_loss_bc": 0.04988919198513031, "train_loss_llm": 0.0, "grad_norm": 0.026890616863965988, "global_step": 2298, "epoch": 63, "lr": 0.001} +{"train_loss": 0.042935509234666824, "train_loss_bc": 0.042935509234666824, "train_loss_llm": 0.0, "grad_norm": 0.02909449301660061, "global_step": 2299, "epoch": 63, "lr": 0.001} +{"train_loss": 0.048686739057302475, "train_loss_bc": 0.048686739057302475, "train_loss_llm": 0.0, "grad_norm": 0.04613823816180229, "global_step": 2300, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04382026195526123, "train_loss_bc": 0.04381907731294632, "train_loss_llm": 0.00011831521987915039, "grad_norm": 0.020276226103305817, "global_step": 2301, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04921840876340866, "train_loss_bc": 0.04918495938181877, "train_loss_llm": 0.0033448636531829834, "grad_norm": 0.029595252126455307, "global_step": 2302, "epoch": 63, "lr": 0.001} +{"train_loss": 0.04529222297585673, "train_loss_bc": 0.051335711032152176, "train_loss_llm": 0.0, "grad_norm": 0.03501395136117935, "global_step": 2303, "epoch": 63, "lr": 0.001, "val_loss": 0.051196105778217316} +{"train_loss": 0.04183289408683777, "train_loss_bc": 0.04183289408683777, "train_loss_llm": 0.0, "grad_norm": 0.04020969569683075, "global_step": 2304, "epoch": 64, "lr": 0.001} +{"train_loss": 0.048052672296762466, "train_loss_bc": 0.048052672296762466, "train_loss_llm": 0.0, "grad_norm": 0.020378755405545235, "global_step": 2305, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04634398967027664, "train_loss_bc": 0.04633277282118797, "train_loss_llm": 0.0011217892169952393, "grad_norm": 0.028541773557662964, "global_step": 2306, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04780082404613495, "train_loss_bc": 0.04780082404613495, "train_loss_llm": 0.0, "grad_norm": 0.03168873488903046, "global_step": 2307, "epoch": 64, "lr": 0.001} +{"train_loss": 0.0382651649415493, "train_loss_bc": 0.0382651649415493, "train_loss_llm": 0.0, "grad_norm": 0.03591763600707054, "global_step": 2308, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04168485850095749, "train_loss_bc": 0.04168485850095749, "train_loss_llm": 0.0, "grad_norm": 0.015028527937829494, "global_step": 2309, "epoch": 64, "lr": 0.001} +{"train_loss": 0.048207689076662064, "train_loss_bc": 0.04817189648747444, "train_loss_llm": 0.0035794079303741455, "grad_norm": 0.022219890728592873, "global_step": 2310, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04333814978599548, "train_loss_bc": 0.04333814978599548, "train_loss_llm": 0.0, "grad_norm": 0.025323620066046715, "global_step": 2311, "epoch": 64, "lr": 0.001} +{"train_loss": 0.045190371572971344, "train_loss_bc": 0.045190371572971344, "train_loss_llm": 0.0, "grad_norm": 0.03015734814107418, "global_step": 2312, "epoch": 64, "lr": 0.001} +{"train_loss": 0.03951572626829147, "train_loss_bc": 0.03951572626829147, "train_loss_llm": 0.0, "grad_norm": 0.013716434128582478, "global_step": 2313, "epoch": 64, "lr": 0.001} +{"train_loss": 0.05089239031076431, "train_loss_bc": 0.05089239031076431, "train_loss_llm": 0.0, "grad_norm": 0.02450213022530079, "global_step": 2314, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04473663866519928, "train_loss_bc": 0.04473663866519928, "train_loss_llm": 0.0, "grad_norm": 0.036577336490154266, "global_step": 2315, "epoch": 64, "lr": 0.001} +{"train_loss": 0.036186136305332184, "train_loss_bc": 0.03618297725915909, "train_loss_llm": 0.0003160834312438965, "grad_norm": 0.049696773290634155, "global_step": 2316, "epoch": 64, "lr": 0.001} +{"train_loss": 0.0532563179731369, "train_loss_bc": 0.053242020308971405, "train_loss_llm": 0.0014298856258392334, "grad_norm": 0.01471036672592163, "global_step": 2317, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04702390730381012, "train_loss_bc": 0.04702390730381012, "train_loss_llm": 0.0, "grad_norm": 0.023424310609698296, "global_step": 2318, "epoch": 64, "lr": 0.001} +{"train_loss": 0.047319959849119186, "train_loss_bc": 0.04729080572724342, "train_loss_llm": 0.002915501594543457, "grad_norm": 0.030998263508081436, "global_step": 2319, "epoch": 64, "lr": 0.001} +{"train_loss": 0.03911814093589783, "train_loss_bc": 0.03911814093589783, "train_loss_llm": 0.0, "grad_norm": 0.039000216871500015, "global_step": 2320, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04581326246261597, "train_loss_bc": 0.04581326246261597, "train_loss_llm": 0.0, "grad_norm": 0.014439171180129051, "global_step": 2321, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04281417652964592, "train_loss_bc": 0.04281417652964592, "train_loss_llm": 0.0, "grad_norm": 0.02324214018881321, "global_step": 2322, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04619087278842926, "train_loss_bc": 0.04617617279291153, "train_loss_llm": 0.0014699697494506836, "grad_norm": 0.028105689212679863, "global_step": 2323, "epoch": 64, "lr": 0.001} +{"train_loss": 0.03947765752673149, "train_loss_bc": 0.039471253752708435, "train_loss_llm": 0.0006402730941772461, "grad_norm": 0.03631827235221863, "global_step": 2324, "epoch": 64, "lr": 0.001} +{"train_loss": 0.03473851457238197, "train_loss_bc": 0.034730929881334305, "train_loss_llm": 0.0007586181163787842, "grad_norm": 0.015550965443253517, "global_step": 2325, "epoch": 64, "lr": 0.001} +{"train_loss": 0.05138114094734192, "train_loss_bc": 0.05138114094734192, "train_loss_llm": 0.0, "grad_norm": 0.022342903539538383, "global_step": 2326, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04001065343618393, "train_loss_bc": 0.04001065343618393, "train_loss_llm": 0.0, "grad_norm": 0.030052833259105682, "global_step": 2327, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04893939569592476, "train_loss_bc": 0.04893939569592476, "train_loss_llm": 0.0, "grad_norm": 0.039888687431812286, "global_step": 2328, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04564603045582771, "train_loss_bc": 0.04561501368880272, "train_loss_llm": 0.0031017959117889404, "grad_norm": 0.01677366904914379, "global_step": 2329, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04043744504451752, "train_loss_bc": 0.04043744504451752, "train_loss_llm": 0.0, "grad_norm": 0.027414238080382347, "global_step": 2330, "epoch": 64, "lr": 0.001} +{"train_loss": 0.05210080370306969, "train_loss_bc": 0.05210080370306969, "train_loss_llm": 0.0, "grad_norm": 0.030640851706266403, "global_step": 2331, "epoch": 64, "lr": 0.001} +{"train_loss": 0.05310693010687828, "train_loss_bc": 0.05310693010687828, "train_loss_llm": 0.0, "grad_norm": 0.03693031519651413, "global_step": 2332, "epoch": 64, "lr": 0.001} +{"train_loss": 0.05074234679341316, "train_loss_bc": 0.05073577165603638, "train_loss_llm": 0.000657498836517334, "grad_norm": 0.016756920143961906, "global_step": 2333, "epoch": 64, "lr": 0.001} +{"train_loss": 0.03563366085290909, "train_loss_bc": 0.03562310338020325, "train_loss_llm": 0.0010555684566497803, "grad_norm": 0.018827959895133972, "global_step": 2334, "epoch": 64, "lr": 0.001} +{"train_loss": 0.0524262860417366, "train_loss_bc": 0.0524262860417366, "train_loss_llm": 0.0, "grad_norm": 0.026665255427360535, "global_step": 2335, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04841572046279907, "train_loss_bc": 0.04841572046279907, "train_loss_llm": 0.0, "grad_norm": 0.03284144401550293, "global_step": 2336, "epoch": 64, "lr": 0.001} +{"train_loss": 0.04359648376703262, "train_loss_bc": 0.043595314025878906, "train_loss_llm": 0.00011706352233886719, "grad_norm": 0.015264328569173813, "global_step": 2337, "epoch": 64, "lr": 0.001} +{"train_loss": 0.047210849821567535, "train_loss_bc": 0.0472075454890728, "train_loss_llm": 0.00033041834831237793, "grad_norm": 0.029873603954911232, "global_step": 2338, "epoch": 64, "lr": 0.001} +{"train_loss": 0.045127206895914346, "train_loss_bc": 0.047131385654211044, "train_loss_llm": 0.0, "grad_norm": 0.03969758003950119, "global_step": 2339, "epoch": 64, "lr": 0.001, "val_loss": 0.051083534955978394} +{"train_loss": 0.039538394659757614, "train_loss_bc": 0.03953465074300766, "train_loss_llm": 0.00037425756454467773, "grad_norm": 0.049703482538461685, "global_step": 2340, "epoch": 65, "lr": 0.001} +{"train_loss": 0.042424462735652924, "train_loss_bc": 0.04239813610911369, "train_loss_llm": 0.002632826566696167, "grad_norm": 0.0173091609030962, "global_step": 2341, "epoch": 65, "lr": 0.001} +{"train_loss": 0.049561757594347, "train_loss_bc": 0.049561757594347, "train_loss_llm": 0.0, "grad_norm": 0.023032227531075478, "global_step": 2342, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04710031673312187, "train_loss_bc": 0.04708922654390335, "train_loss_llm": 0.001109004020690918, "grad_norm": 0.027122002094984055, "global_step": 2343, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04695460945367813, "train_loss_bc": 0.04695460945367813, "train_loss_llm": 0.0, "grad_norm": 0.033155616372823715, "global_step": 2344, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04804280400276184, "train_loss_bc": 0.04804280400276184, "train_loss_llm": 0.0, "grad_norm": 0.015858329832553864, "global_step": 2345, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04447899013757706, "train_loss_bc": 0.04447077587246895, "train_loss_llm": 0.000821530818939209, "grad_norm": 0.022474316880106926, "global_step": 2346, "epoch": 65, "lr": 0.001} +{"train_loss": 0.0448850579559803, "train_loss_bc": 0.0448850579559803, "train_loss_llm": 0.0, "grad_norm": 0.0275716669857502, "global_step": 2347, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04199698939919472, "train_loss_bc": 0.041967928409576416, "train_loss_llm": 0.0029062628746032715, "grad_norm": 0.03064480610191822, "global_step": 2348, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04607374221086502, "train_loss_bc": 0.04607374221086502, "train_loss_llm": 0.0, "grad_norm": 0.01633249782025814, "global_step": 2349, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04245080426335335, "train_loss_bc": 0.04245080426335335, "train_loss_llm": 0.0, "grad_norm": 0.023933149874210358, "global_step": 2350, "epoch": 65, "lr": 0.001} +{"train_loss": 0.049117982387542725, "train_loss_bc": 0.049117982387542725, "train_loss_llm": 0.0, "grad_norm": 0.03425213322043419, "global_step": 2351, "epoch": 65, "lr": 0.001} +{"train_loss": 0.05413238704204559, "train_loss_bc": 0.05413238704204559, "train_loss_llm": 0.0, "grad_norm": 0.0448080375790596, "global_step": 2352, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04274584725499153, "train_loss_bc": 0.04273201897740364, "train_loss_llm": 0.0013827979564666748, "grad_norm": 0.014268525876104832, "global_step": 2353, "epoch": 65, "lr": 0.001} +{"train_loss": 0.049282509833574295, "train_loss_bc": 0.049244072288274765, "train_loss_llm": 0.0038438737392425537, "grad_norm": 0.023349465802311897, "global_step": 2354, "epoch": 65, "lr": 0.001} +{"train_loss": 0.038401879370212555, "train_loss_bc": 0.03838243708014488, "train_loss_llm": 0.0019441545009613037, "grad_norm": 0.02926037460565567, "global_step": 2355, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04793250188231468, "train_loss_bc": 0.04793250188231468, "train_loss_llm": 0.0, "grad_norm": 0.03769706189632416, "global_step": 2356, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04277483746409416, "train_loss_bc": 0.04277483746409416, "train_loss_llm": 0.0, "grad_norm": 0.013919185847043991, "global_step": 2357, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04283872991800308, "train_loss_bc": 0.042797934263944626, "train_loss_llm": 0.00407940149307251, "grad_norm": 0.02155987173318863, "global_step": 2358, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04838085174560547, "train_loss_bc": 0.04836275056004524, "train_loss_llm": 0.0018101334571838379, "grad_norm": 0.031188104301691055, "global_step": 2359, "epoch": 65, "lr": 0.001} +{"train_loss": 0.045537665486335754, "train_loss_bc": 0.04552312567830086, "train_loss_llm": 0.0014541149139404297, "grad_norm": 0.03314026817679405, "global_step": 2360, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04146246984601021, "train_loss_bc": 0.04145113006234169, "train_loss_llm": 0.0011338591575622559, "grad_norm": 0.020662812516093254, "global_step": 2361, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04229804873466492, "train_loss_bc": 0.04229804873466492, "train_loss_llm": 0.0, "grad_norm": 0.025098515674471855, "global_step": 2362, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04958927258849144, "train_loss_bc": 0.04958927258849144, "train_loss_llm": 0.0, "grad_norm": 0.03749150410294533, "global_step": 2363, "epoch": 65, "lr": 0.001} +{"train_loss": 0.03937486186623573, "train_loss_bc": 0.03937486186623573, "train_loss_llm": 0.0, "grad_norm": 0.04706092178821564, "global_step": 2364, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04084358736872673, "train_loss_bc": 0.040835585445165634, "train_loss_llm": 0.0008003115653991699, "grad_norm": 0.017762303352355957, "global_step": 2365, "epoch": 65, "lr": 0.001} +{"train_loss": 0.03950660303235054, "train_loss_bc": 0.03950660303235054, "train_loss_llm": 0.0, "grad_norm": 0.024793632328510284, "global_step": 2366, "epoch": 65, "lr": 0.001} +{"train_loss": 0.041189324110746384, "train_loss_bc": 0.041189324110746384, "train_loss_llm": 0.0, "grad_norm": 0.03147049993276596, "global_step": 2367, "epoch": 65, "lr": 0.001} +{"train_loss": 0.042110417038202286, "train_loss_bc": 0.042108554393053055, "train_loss_llm": 0.00018611550331115723, "grad_norm": 0.037027694284915924, "global_step": 2368, "epoch": 65, "lr": 0.001} +{"train_loss": 0.0531940720975399, "train_loss_bc": 0.0531940720975399, "train_loss_llm": 0.0, "grad_norm": 0.015799256041646004, "global_step": 2369, "epoch": 65, "lr": 0.001} +{"train_loss": 0.05201191455125809, "train_loss_bc": 0.05201191455125809, "train_loss_llm": 0.0, "grad_norm": 0.020345227792859077, "global_step": 2370, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04521757364273071, "train_loss_bc": 0.04521757364273071, "train_loss_llm": 0.0, "grad_norm": 0.0258063655346632, "global_step": 2371, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04524435102939606, "train_loss_bc": 0.04524435102939606, "train_loss_llm": 0.0, "grad_norm": 0.02994166500866413, "global_step": 2372, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04184618964791298, "train_loss_bc": 0.04184427857398987, "train_loss_llm": 0.00019127130508422852, "grad_norm": 0.019202755764126778, "global_step": 2373, "epoch": 65, "lr": 0.001} +{"train_loss": 0.04450365528464317, "train_loss_bc": 0.04450365528464317, "train_loss_llm": 0.0, "grad_norm": 0.02712935395538807, "global_step": 2374, "epoch": 65, "lr": 0.001} +{"train_loss": 0.045113543358941875, "train_loss_bc": 0.0510420985519886, "train_loss_llm": 0.0, "grad_norm": 0.03454617038369179, "global_step": 2375, "epoch": 65, "lr": 0.001, "train/cumulative_reward": 113.24767278583238, "train/mean_score": 8.627907290914795, "train/success_rate": 0.1, "test/cumulative_reward": 55.33696173789142, "test/mean_score": 8.04375435526705, "test/success_rate": 0.16, "val_loss": 0.050892289727926254, "train_action_mse_error": 0.04904784634709358} +{"train_loss": 0.04787829518318176, "train_loss_bc": 0.04787829518318176, "train_loss_llm": 0.0, "grad_norm": 0.04085121676325798, "global_step": 2376, "epoch": 66, "lr": 0.001} +{"train_loss": 0.042307619005441666, "train_loss_bc": 0.04229211062192917, "train_loss_llm": 0.001550823450088501, "grad_norm": 0.01409398764371872, "global_step": 2377, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04232712835073471, "train_loss_bc": 0.04232712835073471, "train_loss_llm": 0.0, "grad_norm": 0.022000038996338844, "global_step": 2378, "epoch": 66, "lr": 0.001} +{"train_loss": 0.042271580547094345, "train_loss_bc": 0.04227123782038689, "train_loss_llm": 3.4362077713012695e-05, "grad_norm": 0.024891283363103867, "global_step": 2379, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04490608721971512, "train_loss_bc": 0.04490608721971512, "train_loss_llm": 0.0, "grad_norm": 0.03029317781329155, "global_step": 2380, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04943333566188812, "train_loss_bc": 0.04943333566188812, "train_loss_llm": 0.0, "grad_norm": 0.015953416004776955, "global_step": 2381, "epoch": 66, "lr": 0.001} +{"train_loss": 0.040517762303352356, "train_loss_bc": 0.040517762303352356, "train_loss_llm": 0.0, "grad_norm": 0.021188678219914436, "global_step": 2382, "epoch": 66, "lr": 0.001} +{"train_loss": 0.037845537066459656, "train_loss_bc": 0.037845537066459656, "train_loss_llm": 0.0, "grad_norm": 0.029584182426333427, "global_step": 2383, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04332857206463814, "train_loss_bc": 0.04332857206463814, "train_loss_llm": 0.0, "grad_norm": 0.0361749604344368, "global_step": 2384, "epoch": 66, "lr": 0.001} +{"train_loss": 0.038089945912361145, "train_loss_bc": 0.038089945912361145, "train_loss_llm": 0.0, "grad_norm": 0.016290733590722084, "global_step": 2385, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04617864266037941, "train_loss_bc": 0.04615595564246178, "train_loss_llm": 0.0022686123847961426, "grad_norm": 0.023067424073815346, "global_step": 2386, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04419878497719765, "train_loss_bc": 0.04419878497719765, "train_loss_llm": 0.0, "grad_norm": 0.034279778599739075, "global_step": 2387, "epoch": 66, "lr": 0.001} +{"train_loss": 0.045538075268268585, "train_loss_bc": 0.045538075268268585, "train_loss_llm": 0.0, "grad_norm": 0.03947685286402702, "global_step": 2388, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04154014587402344, "train_loss_bc": 0.04151215776801109, "train_loss_llm": 0.002798736095428467, "grad_norm": 0.01567784510552883, "global_step": 2389, "epoch": 66, "lr": 0.001} +{"train_loss": 0.040135499089956284, "train_loss_bc": 0.040128443390131, "train_loss_llm": 0.0007056593894958496, "grad_norm": 0.0253938976675272, "global_step": 2390, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04839882254600525, "train_loss_bc": 0.04839882254600525, "train_loss_llm": 0.0, "grad_norm": 0.029832353815436363, "global_step": 2391, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04502127319574356, "train_loss_bc": 0.04502127319574356, "train_loss_llm": 0.0, "grad_norm": 0.03634783625602722, "global_step": 2392, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04628463089466095, "train_loss_bc": 0.04627305269241333, "train_loss_llm": 0.0011578202247619629, "grad_norm": 0.012943385168910027, "global_step": 2393, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04238329082727432, "train_loss_bc": 0.042370766401290894, "train_loss_llm": 0.0012522637844085693, "grad_norm": 0.01840064860880375, "global_step": 2394, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04842697083950043, "train_loss_bc": 0.04842697083950043, "train_loss_llm": 0.0, "grad_norm": 0.028801679611206055, "global_step": 2395, "epoch": 66, "lr": 0.001} +{"train_loss": 0.05213984102010727, "train_loss_bc": 0.05212923139333725, "train_loss_llm": 0.0010611116886138916, "grad_norm": 0.03930648788809776, "global_step": 2396, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04090768098831177, "train_loss_bc": 0.04090768098831177, "train_loss_llm": 0.0, "grad_norm": 0.015796354040503502, "global_step": 2397, "epoch": 66, "lr": 0.001} +{"train_loss": 0.0510127916932106, "train_loss_bc": 0.0510127916932106, "train_loss_llm": 0.0, "grad_norm": 0.02317660301923752, "global_step": 2398, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04592765122652054, "train_loss_bc": 0.04592054709792137, "train_loss_llm": 0.0007102787494659424, "grad_norm": 0.02679063007235527, "global_step": 2399, "epoch": 66, "lr": 0.001} +{"train_loss": 0.056159619241952896, "train_loss_bc": 0.056137073785066605, "train_loss_llm": 0.002254694700241089, "grad_norm": 0.03700893744826317, "global_step": 2400, "epoch": 66, "lr": 0.001} +{"train_loss": 0.041180361062288284, "train_loss_bc": 0.04116299748420715, "train_loss_llm": 0.0017364919185638428, "grad_norm": 0.013899576850235462, "global_step": 2401, "epoch": 66, "lr": 0.001} +{"train_loss": 0.045497749000787735, "train_loss_bc": 0.04549384489655495, "train_loss_llm": 0.0003904402256011963, "grad_norm": 0.01645510084927082, "global_step": 2402, "epoch": 66, "lr": 0.001} +{"train_loss": 0.050211068242788315, "train_loss_bc": 0.05014313384890556, "train_loss_llm": 0.006793379783630371, "grad_norm": 0.026525918394327164, "global_step": 2403, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04944685474038124, "train_loss_bc": 0.04944685474038124, "train_loss_llm": 0.0, "grad_norm": 0.03700828552246094, "global_step": 2404, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04374384507536888, "train_loss_bc": 0.04373113065958023, "train_loss_llm": 0.001271575689315796, "grad_norm": 0.016696466132998466, "global_step": 2405, "epoch": 66, "lr": 0.001} +{"train_loss": 0.037921350449323654, "train_loss_bc": 0.037921350449323654, "train_loss_llm": 0.0, "grad_norm": 0.02684692293405533, "global_step": 2406, "epoch": 66, "lr": 0.001} +{"train_loss": 0.045114390552043915, "train_loss_bc": 0.0450914166867733, "train_loss_llm": 0.0022972822189331055, "grad_norm": 0.03385165333747864, "global_step": 2407, "epoch": 66, "lr": 0.001} +{"train_loss": 0.045792777091264725, "train_loss_bc": 0.04578563570976257, "train_loss_llm": 0.0007142424583435059, "grad_norm": 0.04265203699469566, "global_step": 2408, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04290629178285599, "train_loss_bc": 0.04289694502949715, "train_loss_llm": 0.0009345412254333496, "grad_norm": 0.014475097879767418, "global_step": 2409, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04683158919215202, "train_loss_bc": 0.04683158919215202, "train_loss_llm": 0.0, "grad_norm": 0.020152874290943146, "global_step": 2410, "epoch": 66, "lr": 0.001} +{"train_loss": 0.04493848161978854, "train_loss_bc": 0.04597947746515274, "train_loss_llm": 0.0, "grad_norm": 0.026317240670323372, "global_step": 2411, "epoch": 66, "lr": 0.001, "val_loss": 0.0510203018784523} +{"train_loss": 0.040123358368873596, "train_loss_bc": 0.04009261354804039, "train_loss_llm": 0.00307464599609375, "grad_norm": 0.03222029656171799, "global_step": 2412, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04127030447125435, "train_loss_bc": 0.0412687249481678, "train_loss_llm": 0.00015798211097717285, "grad_norm": 0.012159515172243118, "global_step": 2413, "epoch": 67, "lr": 0.001} +{"train_loss": 0.042795971035957336, "train_loss_bc": 0.042795971035957336, "train_loss_llm": 0.0, "grad_norm": 0.019198790192604065, "global_step": 2414, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04406968876719475, "train_loss_bc": 0.04406968876719475, "train_loss_llm": 0.0, "grad_norm": 0.02651650831103325, "global_step": 2415, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04503992199897766, "train_loss_bc": 0.045029569417238235, "train_loss_llm": 0.001035153865814209, "grad_norm": 0.03243723511695862, "global_step": 2416, "epoch": 67, "lr": 0.001} +{"train_loss": 0.05028456449508667, "train_loss_bc": 0.05028456449508667, "train_loss_llm": 0.0, "grad_norm": 0.017505154013633728, "global_step": 2417, "epoch": 67, "lr": 0.001} +{"train_loss": 0.046723365783691406, "train_loss_bc": 0.046723365783691406, "train_loss_llm": 0.0, "grad_norm": 0.027853012084960938, "global_step": 2418, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04037805646657944, "train_loss_bc": 0.04033250734210014, "train_loss_llm": 0.0045549869537353516, "grad_norm": 0.035448893904685974, "global_step": 2419, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04821833223104477, "train_loss_bc": 0.04821833223104477, "train_loss_llm": 0.0, "grad_norm": 0.041844628751277924, "global_step": 2420, "epoch": 67, "lr": 0.001} +{"train_loss": 0.0469987578690052, "train_loss_bc": 0.0469987578690052, "train_loss_llm": 0.0, "grad_norm": 0.017666660249233246, "global_step": 2421, "epoch": 67, "lr": 0.001} +{"train_loss": 0.05454466864466667, "train_loss_bc": 0.05453334003686905, "train_loss_llm": 0.0011326968669891357, "grad_norm": 0.024175001308321953, "global_step": 2422, "epoch": 67, "lr": 0.001} +{"train_loss": 0.0396718829870224, "train_loss_bc": 0.0396718829870224, "train_loss_llm": 0.0, "grad_norm": 0.0314440093934536, "global_step": 2423, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04070442169904709, "train_loss_bc": 0.04070442169904709, "train_loss_llm": 0.0, "grad_norm": 0.038644060492515564, "global_step": 2424, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04750106483697891, "train_loss_bc": 0.04750064015388489, "train_loss_llm": 4.240870475769043e-05, "grad_norm": 0.014962424524128437, "global_step": 2425, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04396402835845947, "train_loss_bc": 0.04395933449268341, "train_loss_llm": 0.0004692375659942627, "grad_norm": 0.024245409294962883, "global_step": 2426, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04632410407066345, "train_loss_bc": 0.04632410407066345, "train_loss_llm": 0.0, "grad_norm": 0.0307086743414402, "global_step": 2427, "epoch": 67, "lr": 0.001} +{"train_loss": 0.03443215787410736, "train_loss_bc": 0.03443215787410736, "train_loss_llm": 0.0, "grad_norm": 0.03854997083544731, "global_step": 2428, "epoch": 67, "lr": 0.001} +{"train_loss": 0.03216017037630081, "train_loss_bc": 0.03216017037630081, "train_loss_llm": 0.0, "grad_norm": 0.018107131123542786, "global_step": 2429, "epoch": 67, "lr": 0.001} +{"train_loss": 0.045547518879175186, "train_loss_bc": 0.045547518879175186, "train_loss_llm": 0.0, "grad_norm": 0.025940630584955215, "global_step": 2430, "epoch": 67, "lr": 0.001} +{"train_loss": 0.049631182104349136, "train_loss_bc": 0.04962931573390961, "train_loss_llm": 0.00018650293350219727, "grad_norm": 0.024459408596158028, "global_step": 2431, "epoch": 67, "lr": 0.001} +{"train_loss": 0.05418757349252701, "train_loss_bc": 0.05418757349252701, "train_loss_llm": 0.0, "grad_norm": 0.029361439868807793, "global_step": 2432, "epoch": 67, "lr": 0.001} +{"train_loss": 0.046669911593198776, "train_loss_bc": 0.046669334173202515, "train_loss_llm": 5.78761100769043e-05, "grad_norm": 0.02250763028860092, "global_step": 2433, "epoch": 67, "lr": 0.001} +{"train_loss": 0.03514912351965904, "train_loss_bc": 0.03514912351965904, "train_loss_llm": 0.0, "grad_norm": 0.02880765125155449, "global_step": 2434, "epoch": 67, "lr": 0.001} +{"train_loss": 0.047970034182071686, "train_loss_bc": 0.04793602600693703, "train_loss_llm": 0.0034008026123046875, "grad_norm": 0.04404386878013611, "global_step": 2435, "epoch": 67, "lr": 0.001} +{"train_loss": 0.044870954006910324, "train_loss_bc": 0.04483124241232872, "train_loss_llm": 0.00397104024887085, "grad_norm": 0.05154049023985863, "global_step": 2436, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04664967209100723, "train_loss_bc": 0.04664967209100723, "train_loss_llm": 0.0, "grad_norm": 0.01596832275390625, "global_step": 2437, "epoch": 67, "lr": 0.001} +{"train_loss": 0.03876106068491936, "train_loss_bc": 0.038757357746362686, "train_loss_llm": 0.00037044286727905273, "grad_norm": 0.024599740281701088, "global_step": 2438, "epoch": 67, "lr": 0.001} +{"train_loss": 0.042635418474674225, "train_loss_bc": 0.042635418474674225, "train_loss_llm": 0.0, "grad_norm": 0.031112145632505417, "global_step": 2439, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04511800408363342, "train_loss_bc": 0.04511800408363342, "train_loss_llm": 0.0, "grad_norm": 0.036001626402139664, "global_step": 2440, "epoch": 67, "lr": 0.001} +{"train_loss": 0.053179364651441574, "train_loss_bc": 0.053179364651441574, "train_loss_llm": 0.0, "grad_norm": 0.017659012228250504, "global_step": 2441, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04222506657242775, "train_loss_bc": 0.04222506657242775, "train_loss_llm": 0.0, "grad_norm": 0.025038491934537888, "global_step": 2442, "epoch": 67, "lr": 0.001} +{"train_loss": 0.049433719366788864, "train_loss_bc": 0.049433719366788864, "train_loss_llm": 0.0, "grad_norm": 0.035754263401031494, "global_step": 2443, "epoch": 67, "lr": 0.001} +{"train_loss": 0.051050785928964615, "train_loss_bc": 0.051050785928964615, "train_loss_llm": 0.0, "grad_norm": 0.04049651324748993, "global_step": 2444, "epoch": 67, "lr": 0.001} +{"train_loss": 0.04736050218343735, "train_loss_bc": 0.04736050218343735, "train_loss_llm": 0.0, "grad_norm": 0.01533749233931303, "global_step": 2445, "epoch": 67, "lr": 0.001} +{"train_loss": 0.048100393265485764, "train_loss_bc": 0.048100393265485764, "train_loss_llm": 0.0, "grad_norm": 0.026570135727524757, "global_step": 2446, "epoch": 67, "lr": 0.001} +{"train_loss": 0.044895395222637385, "train_loss_bc": 0.042480893433094025, "train_loss_llm": 0.0008228123188018799, "grad_norm": 0.037905797362327576, "global_step": 2447, "epoch": 67, "lr": 0.001, "val_loss": 0.050741374492645264} +{"train_loss": 0.042243167757987976, "train_loss_bc": 0.042243167757987976, "train_loss_llm": 0.0, "grad_norm": 0.04343995824456215, "global_step": 2448, "epoch": 68, "lr": 0.001} +{"train_loss": 0.05287803336977959, "train_loss_bc": 0.05286087468266487, "train_loss_llm": 0.0017160475254058838, "grad_norm": 0.014685766771435738, "global_step": 2449, "epoch": 68, "lr": 0.001} +{"train_loss": 0.0525534451007843, "train_loss_bc": 0.0525534451007843, "train_loss_llm": 0.0, "grad_norm": 0.02358257956802845, "global_step": 2450, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04546733573079109, "train_loss_bc": 0.04546733573079109, "train_loss_llm": 0.0, "grad_norm": 0.03218553960323334, "global_step": 2451, "epoch": 68, "lr": 0.001} +{"train_loss": 0.042664702981710434, "train_loss_bc": 0.042655546218156815, "train_loss_llm": 0.0009155869483947754, "grad_norm": 0.03736536204814911, "global_step": 2452, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04563101753592491, "train_loss_bc": 0.0456036813557148, "train_loss_llm": 0.0027337968349456787, "grad_norm": 0.014801268465816975, "global_step": 2453, "epoch": 68, "lr": 0.001} +{"train_loss": 0.03619726747274399, "train_loss_bc": 0.03619726747274399, "train_loss_llm": 0.0, "grad_norm": 0.02546205185353756, "global_step": 2454, "epoch": 68, "lr": 0.001} +{"train_loss": 0.050701871514320374, "train_loss_bc": 0.050701871514320374, "train_loss_llm": 0.0, "grad_norm": 0.032251987606287, "global_step": 2455, "epoch": 68, "lr": 0.001} +{"train_loss": 0.05027095600962639, "train_loss_bc": 0.05027095600962639, "train_loss_llm": 0.0, "grad_norm": 0.04060821235179901, "global_step": 2456, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04557322710752487, "train_loss_bc": 0.04556942358613014, "train_loss_llm": 0.00038051605224609375, "grad_norm": 0.014797759242355824, "global_step": 2457, "epoch": 68, "lr": 0.001} +{"train_loss": 0.0425027497112751, "train_loss_bc": 0.0425027497112751, "train_loss_llm": 0.0, "grad_norm": 0.021205179393291473, "global_step": 2458, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04181097447872162, "train_loss_bc": 0.04179956391453743, "train_loss_llm": 0.0011410713195800781, "grad_norm": 0.02517588809132576, "global_step": 2459, "epoch": 68, "lr": 0.001} +{"train_loss": 0.03785380348563194, "train_loss_bc": 0.037818338721990585, "train_loss_llm": 0.003546416759490967, "grad_norm": 0.03162965178489685, "global_step": 2460, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04877718538045883, "train_loss_bc": 0.04875703901052475, "train_loss_llm": 0.00201454758644104, "grad_norm": 0.01613662764430046, "global_step": 2461, "epoch": 68, "lr": 0.001} +{"train_loss": 0.045689236372709274, "train_loss_bc": 0.045689236372709274, "train_loss_llm": 0.0, "grad_norm": 0.02339436300098896, "global_step": 2462, "epoch": 68, "lr": 0.001} +{"train_loss": 0.05577128380537033, "train_loss_bc": 0.05577128380537033, "train_loss_llm": 0.0, "grad_norm": 0.0312273558229208, "global_step": 2463, "epoch": 68, "lr": 0.001} +{"train_loss": 0.03881042078137398, "train_loss_bc": 0.03881042078137398, "train_loss_llm": 0.0, "grad_norm": 0.03887532651424408, "global_step": 2464, "epoch": 68, "lr": 0.001} +{"train_loss": 0.053028181195259094, "train_loss_bc": 0.053028181195259094, "train_loss_llm": 0.0, "grad_norm": 0.020037854090332985, "global_step": 2465, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04219616577029228, "train_loss_bc": 0.04219616577029228, "train_loss_llm": 0.0, "grad_norm": 0.02265722118318081, "global_step": 2466, "epoch": 68, "lr": 0.001} +{"train_loss": 0.049709949642419815, "train_loss_bc": 0.049709949642419815, "train_loss_llm": 0.0, "grad_norm": 0.03000759705901146, "global_step": 2467, "epoch": 68, "lr": 0.001} +{"train_loss": 0.041410207748413086, "train_loss_bc": 0.041386399418115616, "train_loss_llm": 0.002380967140197754, "grad_norm": 0.033008601516485214, "global_step": 2468, "epoch": 68, "lr": 0.001} +{"train_loss": 0.03381238877773285, "train_loss_bc": 0.03381238877773285, "train_loss_llm": 0.0, "grad_norm": 0.011767201125621796, "global_step": 2469, "epoch": 68, "lr": 0.001} +{"train_loss": 0.05176354944705963, "train_loss_bc": 0.05176354944705963, "train_loss_llm": 0.0, "grad_norm": 0.020559199154376984, "global_step": 2470, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04369203746318817, "train_loss_bc": 0.04368859902024269, "train_loss_llm": 0.00034368038177490234, "grad_norm": 0.03756524622440338, "global_step": 2471, "epoch": 68, "lr": 0.001} +{"train_loss": 0.05406004190444946, "train_loss_bc": 0.05406004190444946, "train_loss_llm": 0.0, "grad_norm": 0.045036762952804565, "global_step": 2472, "epoch": 68, "lr": 0.001} +{"train_loss": 0.0429299920797348, "train_loss_bc": 0.04292593523859978, "train_loss_llm": 0.0004056692123413086, "grad_norm": 0.013574928976595402, "global_step": 2473, "epoch": 68, "lr": 0.001} +{"train_loss": 0.0462663434445858, "train_loss_bc": 0.0462663434445858, "train_loss_llm": 0.0, "grad_norm": 0.028573796153068542, "global_step": 2474, "epoch": 68, "lr": 0.001} +{"train_loss": 0.039419468492269516, "train_loss_bc": 0.039419468492269516, "train_loss_llm": 0.0, "grad_norm": 0.031893566250801086, "global_step": 2475, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04196690395474434, "train_loss_bc": 0.04196690395474434, "train_loss_llm": 0.0, "grad_norm": 0.034837838262319565, "global_step": 2476, "epoch": 68, "lr": 0.001} +{"train_loss": 0.03924513980746269, "train_loss_bc": 0.03924513980746269, "train_loss_llm": 0.0, "grad_norm": 0.017533672973513603, "global_step": 2477, "epoch": 68, "lr": 0.001} +{"train_loss": 0.0395391471683979, "train_loss_bc": 0.039517175406217575, "train_loss_llm": 0.002197086811065674, "grad_norm": 0.02596898190677166, "global_step": 2478, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04060541093349457, "train_loss_bc": 0.04060541093349457, "train_loss_llm": 0.0, "grad_norm": 0.032208558171987534, "global_step": 2479, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04386063665151596, "train_loss_bc": 0.043846771121025085, "train_loss_llm": 0.0013867318630218506, "grad_norm": 0.037737857550382614, "global_step": 2480, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04872330650687218, "train_loss_bc": 0.04872330650687218, "train_loss_llm": 0.0, "grad_norm": 0.01700615882873535, "global_step": 2481, "epoch": 68, "lr": 0.001} +{"train_loss": 0.04302237927913666, "train_loss_bc": 0.0430038720369339, "train_loss_llm": 0.0018505454063415527, "grad_norm": 0.026073837652802467, "global_step": 2482, "epoch": 68, "lr": 0.001} +{"train_loss": 0.044799559439222016, "train_loss_bc": 0.04213522747159004, "train_loss_llm": 9.849667549133301e-05, "grad_norm": 0.03383619710803032, "global_step": 2483, "epoch": 68, "lr": 0.001, "val_loss": 0.05122559517621994} +{"train_loss": 0.043008167296648026, "train_loss_bc": 0.043002478778362274, "train_loss_llm": 0.0005689561367034912, "grad_norm": 0.04061013460159302, "global_step": 2484, "epoch": 69, "lr": 0.001} +{"train_loss": 0.03844291716814041, "train_loss_bc": 0.03844291716814041, "train_loss_llm": 0.0, "grad_norm": 0.010733437724411488, "global_step": 2485, "epoch": 69, "lr": 0.001} +{"train_loss": 0.045656394213438034, "train_loss_bc": 0.045628320425748825, "train_loss_llm": 0.0028072893619537354, "grad_norm": 0.014732862822711468, "global_step": 2486, "epoch": 69, "lr": 0.001} +{"train_loss": 0.045649684965610504, "train_loss_bc": 0.045623406767845154, "train_loss_llm": 0.0026277005672454834, "grad_norm": 0.021903609856963158, "global_step": 2487, "epoch": 69, "lr": 0.001} +{"train_loss": 0.03982250392436981, "train_loss_bc": 0.03980948403477669, "train_loss_llm": 0.0013021528720855713, "grad_norm": 0.028613785281777382, "global_step": 2488, "epoch": 69, "lr": 0.001} +{"train_loss": 0.05277527496218681, "train_loss_bc": 0.05277527496218681, "train_loss_llm": 0.0, "grad_norm": 0.019299685955047607, "global_step": 2489, "epoch": 69, "lr": 0.001} +{"train_loss": 0.045740995556116104, "train_loss_bc": 0.045740995556116104, "train_loss_llm": 0.0, "grad_norm": 0.025525644421577454, "global_step": 2490, "epoch": 69, "lr": 0.001} +{"train_loss": 0.042166370898485184, "train_loss_bc": 0.042166370898485184, "train_loss_llm": 0.0, "grad_norm": 0.03457290679216385, "global_step": 2491, "epoch": 69, "lr": 0.001} +{"train_loss": 0.040867116302251816, "train_loss_bc": 0.040867116302251816, "train_loss_llm": 0.0, "grad_norm": 0.040140118449926376, "global_step": 2492, "epoch": 69, "lr": 0.001} +{"train_loss": 0.048051971942186356, "train_loss_bc": 0.048034198582172394, "train_loss_llm": 0.0017774701118469238, "grad_norm": 0.019578319042921066, "global_step": 2493, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04398108646273613, "train_loss_bc": 0.04398108646273613, "train_loss_llm": 0.0, "grad_norm": 0.024785319343209267, "global_step": 2494, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04292283207178116, "train_loss_bc": 0.04292283207178116, "train_loss_llm": 0.0, "grad_norm": 0.03179861977696419, "global_step": 2495, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04501332715153694, "train_loss_bc": 0.04501332715153694, "train_loss_llm": 0.0, "grad_norm": 0.03487210348248482, "global_step": 2496, "epoch": 69, "lr": 0.001} +{"train_loss": 0.038208067417144775, "train_loss_bc": 0.03820786625146866, "train_loss_llm": 2.0056962966918945e-05, "grad_norm": 0.012385568581521511, "global_step": 2497, "epoch": 69, "lr": 0.001} +{"train_loss": 0.0490996316075325, "train_loss_bc": 0.04905635118484497, "train_loss_llm": 0.004327863454818726, "grad_norm": 0.02358839474618435, "global_step": 2498, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04163269326090813, "train_loss_bc": 0.04163269326090813, "train_loss_llm": 0.0, "grad_norm": 0.024662133306264877, "global_step": 2499, "epoch": 69, "lr": 0.001} +{"train_loss": 0.0481971800327301, "train_loss_bc": 0.0481971800327301, "train_loss_llm": 0.0, "grad_norm": 0.02681867778301239, "global_step": 2500, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04190031811594963, "train_loss_bc": 0.04190031811594963, "train_loss_llm": 0.0, "grad_norm": 0.014753171242773533, "global_step": 2501, "epoch": 69, "lr": 0.001} +{"train_loss": 0.03874758630990982, "train_loss_bc": 0.03872014209628105, "train_loss_llm": 0.0027444064617156982, "grad_norm": 0.020732173696160316, "global_step": 2502, "epoch": 69, "lr": 0.001} +{"train_loss": 0.0398164838552475, "train_loss_bc": 0.03979672119021416, "train_loss_llm": 0.001976221799850464, "grad_norm": 0.026904765516519547, "global_step": 2503, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04297072812914848, "train_loss_bc": 0.04297072812914848, "train_loss_llm": 0.0, "grad_norm": 0.03290332481265068, "global_step": 2504, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04416404291987419, "train_loss_bc": 0.044149260967969894, "train_loss_llm": 0.0014781653881072998, "grad_norm": 0.015393958427011967, "global_step": 2505, "epoch": 69, "lr": 0.001} +{"train_loss": 0.043218184262514114, "train_loss_bc": 0.043218184262514114, "train_loss_llm": 0.0, "grad_norm": 0.023931050673127174, "global_step": 2506, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04544456675648689, "train_loss_bc": 0.04544456675648689, "train_loss_llm": 0.0, "grad_norm": 0.03248270973563194, "global_step": 2507, "epoch": 69, "lr": 0.001} +{"train_loss": 0.05295107141137123, "train_loss_bc": 0.05295107141137123, "train_loss_llm": 0.0, "grad_norm": 0.03806426003575325, "global_step": 2508, "epoch": 69, "lr": 0.001} +{"train_loss": 0.05217549949884415, "train_loss_bc": 0.05217549949884415, "train_loss_llm": 0.0, "grad_norm": 0.018080445006489754, "global_step": 2509, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04495732858777046, "train_loss_bc": 0.04495732858777046, "train_loss_llm": 0.0, "grad_norm": 0.026332881301641464, "global_step": 2510, "epoch": 69, "lr": 0.001} +{"train_loss": 0.05030880123376846, "train_loss_bc": 0.05030880123376846, "train_loss_llm": 0.0, "grad_norm": 0.03874776512384415, "global_step": 2511, "epoch": 69, "lr": 0.001} +{"train_loss": 0.05517495051026344, "train_loss_bc": 0.05517495051026344, "train_loss_llm": 0.0, "grad_norm": 0.04654364287853241, "global_step": 2512, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04749232903122902, "train_loss_bc": 0.04747764766216278, "train_loss_llm": 0.0014681816101074219, "grad_norm": 0.019821863621473312, "global_step": 2513, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04145728424191475, "train_loss_bc": 0.04145728424191475, "train_loss_llm": 0.0, "grad_norm": 0.028649726882576942, "global_step": 2514, "epoch": 69, "lr": 0.001} +{"train_loss": 0.0420137420296669, "train_loss_bc": 0.0420137420296669, "train_loss_llm": 0.0, "grad_norm": 0.03173885866999626, "global_step": 2515, "epoch": 69, "lr": 0.001} +{"train_loss": 0.05065062269568443, "train_loss_bc": 0.05065062269568443, "train_loss_llm": 0.0, "grad_norm": 0.038722604513168335, "global_step": 2516, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04312529042363167, "train_loss_bc": 0.04312128946185112, "train_loss_llm": 0.0003999769687652588, "grad_norm": 0.016908112913370132, "global_step": 2517, "epoch": 69, "lr": 0.001} +{"train_loss": 0.04425741359591484, "train_loss_bc": 0.044247955083847046, "train_loss_llm": 0.000945746898651123, "grad_norm": 0.024196244776248932, "global_step": 2518, "epoch": 69, "lr": 0.001} +{"train_loss": 0.044751076131231256, "train_loss_bc": 0.0389762818813324, "train_loss_llm": 0.0, "grad_norm": 0.03646812215447426, "global_step": 2519, "epoch": 69, "lr": 0.001, "val_loss": 0.05072912573814392} +{"train_loss": 0.039736710488796234, "train_loss_bc": 0.039736710488796234, "train_loss_llm": 0.0, "grad_norm": 0.03622126579284668, "global_step": 2520, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04894719272851944, "train_loss_bc": 0.04894719272851944, "train_loss_llm": 0.0, "grad_norm": 0.018085593357682228, "global_step": 2521, "epoch": 70, "lr": 0.001} +{"train_loss": 0.03483328968286514, "train_loss_bc": 0.03483328968286514, "train_loss_llm": 0.0, "grad_norm": 0.022236144170165062, "global_step": 2522, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04511614888906479, "train_loss_bc": 0.04505588114261627, "train_loss_llm": 0.006026655435562134, "grad_norm": 0.023824652656912804, "global_step": 2523, "epoch": 70, "lr": 0.001} +{"train_loss": 0.043554943054914474, "train_loss_bc": 0.04355240240693092, "train_loss_llm": 0.00025394558906555176, "grad_norm": 0.02842804044485092, "global_step": 2524, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04333404451608658, "train_loss_bc": 0.04333404451608658, "train_loss_llm": 0.0, "grad_norm": 0.01832609996199608, "global_step": 2525, "epoch": 70, "lr": 0.001} +{"train_loss": 0.05328657850623131, "train_loss_bc": 0.05328657850623131, "train_loss_llm": 0.0, "grad_norm": 0.02213941514492035, "global_step": 2526, "epoch": 70, "lr": 0.001} +{"train_loss": 0.044829726219177246, "train_loss_bc": 0.04481440782546997, "train_loss_llm": 0.0015316903591156006, "grad_norm": 0.030061490833759308, "global_step": 2527, "epoch": 70, "lr": 0.001} +{"train_loss": 0.05370120331645012, "train_loss_bc": 0.05370120331645012, "train_loss_llm": 0.0, "grad_norm": 0.03309601545333862, "global_step": 2528, "epoch": 70, "lr": 0.001} +{"train_loss": 0.046063944697380066, "train_loss_bc": 0.046063944697380066, "train_loss_llm": 0.0, "grad_norm": 0.01749950647354126, "global_step": 2529, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04191108047962189, "train_loss_bc": 0.04191108047962189, "train_loss_llm": 0.0, "grad_norm": 0.022754937410354614, "global_step": 2530, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04669645428657532, "train_loss_bc": 0.04669645428657532, "train_loss_llm": 0.0, "grad_norm": 0.030259564518928528, "global_step": 2531, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04812072217464447, "train_loss_bc": 0.04812072217464447, "train_loss_llm": 0.0, "grad_norm": 0.03245817497372627, "global_step": 2532, "epoch": 70, "lr": 0.001} +{"train_loss": 0.049143947660923004, "train_loss_bc": 0.049143947660923004, "train_loss_llm": 0.0, "grad_norm": 0.018548861145973206, "global_step": 2533, "epoch": 70, "lr": 0.001} +{"train_loss": 0.040733516216278076, "train_loss_bc": 0.040733516216278076, "train_loss_llm": 0.0, "grad_norm": 0.023216666653752327, "global_step": 2534, "epoch": 70, "lr": 0.001} +{"train_loss": 0.038227684795856476, "train_loss_bc": 0.03822128102183342, "train_loss_llm": 0.00064048171043396, "grad_norm": 0.029852142557501793, "global_step": 2535, "epoch": 70, "lr": 0.001} +{"train_loss": 0.042815472930669785, "train_loss_bc": 0.04281032457947731, "train_loss_llm": 0.0005147457122802734, "grad_norm": 0.04090126231312752, "global_step": 2536, "epoch": 70, "lr": 0.001} +{"train_loss": 0.03605293482542038, "train_loss_bc": 0.03605293482542038, "train_loss_llm": 0.0, "grad_norm": 0.014948545955121517, "global_step": 2537, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04791275039315224, "train_loss_bc": 0.04791275039315224, "train_loss_llm": 0.0, "grad_norm": 0.026917247101664543, "global_step": 2538, "epoch": 70, "lr": 0.001} +{"train_loss": 0.03616369888186455, "train_loss_bc": 0.036124952137470245, "train_loss_llm": 0.00387457013130188, "grad_norm": 0.03359018638730049, "global_step": 2539, "epoch": 70, "lr": 0.001} +{"train_loss": 0.052472107112407684, "train_loss_bc": 0.05245671421289444, "train_loss_llm": 0.0015392899513244629, "grad_norm": 0.0407058522105217, "global_step": 2540, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04765995591878891, "train_loss_bc": 0.04765995591878891, "train_loss_llm": 0.0, "grad_norm": 0.014703931286931038, "global_step": 2541, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04303918406367302, "train_loss_bc": 0.04303918406367302, "train_loss_llm": 0.0, "grad_norm": 0.02386518567800522, "global_step": 2542, "epoch": 70, "lr": 0.001} +{"train_loss": 0.047225840389728546, "train_loss_bc": 0.047187287360429764, "train_loss_llm": 0.003855407238006592, "grad_norm": 0.02969440072774887, "global_step": 2543, "epoch": 70, "lr": 0.001} +{"train_loss": 0.049647942185401917, "train_loss_bc": 0.049647942185401917, "train_loss_llm": 0.0, "grad_norm": 0.041676461696624756, "global_step": 2544, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04127541929483414, "train_loss_bc": 0.04126448184251785, "train_loss_llm": 0.0010938048362731934, "grad_norm": 0.01251995749771595, "global_step": 2545, "epoch": 70, "lr": 0.001} +{"train_loss": 0.046645637601614, "train_loss_bc": 0.046634312719106674, "train_loss_llm": 0.001132667064666748, "grad_norm": 0.02241775579750538, "global_step": 2546, "epoch": 70, "lr": 0.001} +{"train_loss": 0.046592604368925095, "train_loss_bc": 0.046592604368925095, "train_loss_llm": 0.0, "grad_norm": 0.02983173355460167, "global_step": 2547, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04967579245567322, "train_loss_bc": 0.04966229200363159, "train_loss_llm": 0.0013498961925506592, "grad_norm": 0.03653613477945328, "global_step": 2548, "epoch": 70, "lr": 0.001} +{"train_loss": 0.05466800555586815, "train_loss_bc": 0.05466800555586815, "train_loss_llm": 0.0, "grad_norm": 0.018390899524092674, "global_step": 2549, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04120023921132088, "train_loss_bc": 0.04119167476892471, "train_loss_llm": 0.0008563399314880371, "grad_norm": 0.026719171553850174, "global_step": 2550, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04214964807033539, "train_loss_bc": 0.04214964807033539, "train_loss_llm": 0.0, "grad_norm": 0.035452473908662796, "global_step": 2551, "epoch": 70, "lr": 0.001} +{"train_loss": 0.043688029050827026, "train_loss_bc": 0.043688029050827026, "train_loss_llm": 0.0, "grad_norm": 0.045311130583286285, "global_step": 2552, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04817359149456024, "train_loss_bc": 0.048170238733291626, "train_loss_llm": 0.00033536553382873535, "grad_norm": 0.01682140678167343, "global_step": 2553, "epoch": 70, "lr": 0.001} +{"train_loss": 0.03628033027052879, "train_loss_bc": 0.03628033027052879, "train_loss_llm": 0.0, "grad_norm": 0.030303020030260086, "global_step": 2554, "epoch": 70, "lr": 0.001} +{"train_loss": 0.04486923447499672, "train_loss_bc": 0.043692585080862045, "train_loss_llm": 0.002348482608795166, "grad_norm": 0.04080379754304886, "global_step": 2555, "epoch": 70, "lr": 0.001, "train/cumulative_reward": 76.81180763816766, "train/mean_score": 14.646289626857842, "train/success_rate": 0.4, "test/cumulative_reward": 61.39101073771996, "test/mean_score": 8.415080901888958, "test/success_rate": 0.16, "val_loss": 0.05124451965093613, "train_action_mse_error": 0.04845694079995155} +{"train_loss": 0.043073832988739014, "train_loss_bc": 0.043073832988739014, "train_loss_llm": 0.0, "grad_norm": 0.046750690788030624, "global_step": 2556, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04597683995962143, "train_loss_bc": 0.04597683995962143, "train_loss_llm": 0.0, "grad_norm": 0.014849990606307983, "global_step": 2557, "epoch": 71, "lr": 0.001} +{"train_loss": 0.042848486453294754, "train_loss_bc": 0.042848486453294754, "train_loss_llm": 0.0, "grad_norm": 0.02134890854358673, "global_step": 2558, "epoch": 71, "lr": 0.001} +{"train_loss": 0.05240457504987717, "train_loss_bc": 0.052383724600076675, "train_loss_llm": 0.0020852088928222656, "grad_norm": 0.027780404314398766, "global_step": 2559, "epoch": 71, "lr": 0.001} +{"train_loss": 0.05254682898521423, "train_loss_bc": 0.05254682898521423, "train_loss_llm": 0.0, "grad_norm": 0.03734935075044632, "global_step": 2560, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04286760464310646, "train_loss_bc": 0.04285724461078644, "train_loss_llm": 0.001036137342453003, "grad_norm": 0.011834345757961273, "global_step": 2561, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04050637409090996, "train_loss_bc": 0.04050637409090996, "train_loss_llm": 0.0, "grad_norm": 0.017104197293519974, "global_step": 2562, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04861966520547867, "train_loss_bc": 0.04861009120941162, "train_loss_llm": 0.0009574294090270996, "grad_norm": 0.027756191790103912, "global_step": 2563, "epoch": 71, "lr": 0.001} +{"train_loss": 0.03888791427016258, "train_loss_bc": 0.03888791427016258, "train_loss_llm": 0.0, "grad_norm": 0.033150576055049896, "global_step": 2564, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04376436397433281, "train_loss_bc": 0.04376436397433281, "train_loss_llm": 0.0, "grad_norm": 0.017506666481494904, "global_step": 2565, "epoch": 71, "lr": 0.001} +{"train_loss": 0.03554869815707207, "train_loss_bc": 0.03554869815707207, "train_loss_llm": 0.0, "grad_norm": 0.022413086146116257, "global_step": 2566, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04429571330547333, "train_loss_bc": 0.04429571330547333, "train_loss_llm": 0.0, "grad_norm": 0.032635271549224854, "global_step": 2567, "epoch": 71, "lr": 0.001} +{"train_loss": 0.03737741336226463, "train_loss_bc": 0.03737741336226463, "train_loss_llm": 0.0, "grad_norm": 0.03973546624183655, "global_step": 2568, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04337168112397194, "train_loss_bc": 0.04337168112397194, "train_loss_llm": 0.0, "grad_norm": 0.01881745643913746, "global_step": 2569, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04265343397855759, "train_loss_bc": 0.04264043644070625, "train_loss_llm": 0.0012997090816497803, "grad_norm": 0.02461695857346058, "global_step": 2570, "epoch": 71, "lr": 0.001} +{"train_loss": 0.047357864677906036, "train_loss_bc": 0.04735647141933441, "train_loss_llm": 0.0001391768455505371, "grad_norm": 0.030799290165305138, "global_step": 2571, "epoch": 71, "lr": 0.001} +{"train_loss": 0.045908596366643906, "train_loss_bc": 0.045897550880908966, "train_loss_llm": 0.0011046826839447021, "grad_norm": 0.03969404846429825, "global_step": 2572, "epoch": 71, "lr": 0.001} +{"train_loss": 0.03464334085583687, "train_loss_bc": 0.03464334085583687, "train_loss_llm": 0.0, "grad_norm": 0.014861175790429115, "global_step": 2573, "epoch": 71, "lr": 0.001} +{"train_loss": 0.040385790169239044, "train_loss_bc": 0.04038080945611, "train_loss_llm": 0.0004981458187103271, "grad_norm": 0.027299750596284866, "global_step": 2574, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04268474131822586, "train_loss_bc": 0.042683422565460205, "train_loss_llm": 0.00013184547424316406, "grad_norm": 0.03370985761284828, "global_step": 2575, "epoch": 71, "lr": 0.001} +{"train_loss": 0.042422663420438766, "train_loss_bc": 0.042422663420438766, "train_loss_llm": 0.0, "grad_norm": 0.04463689401745796, "global_step": 2576, "epoch": 71, "lr": 0.001} +{"train_loss": 0.05489323288202286, "train_loss_bc": 0.05489323288202286, "train_loss_llm": 0.0, "grad_norm": 0.017918676137924194, "global_step": 2577, "epoch": 71, "lr": 0.001} +{"train_loss": 0.05431204289197922, "train_loss_bc": 0.054310187697410583, "train_loss_llm": 0.0001856386661529541, "grad_norm": 0.02910560742020607, "global_step": 2578, "epoch": 71, "lr": 0.001} +{"train_loss": 0.05354104936122894, "train_loss_bc": 0.05354104936122894, "train_loss_llm": 0.0, "grad_norm": 0.037312474101781845, "global_step": 2579, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04460759088397026, "train_loss_bc": 0.04459874704480171, "train_loss_llm": 0.000884324312210083, "grad_norm": 0.043986689299345016, "global_step": 2580, "epoch": 71, "lr": 0.001} +{"train_loss": 0.05233008414506912, "train_loss_bc": 0.05233001336455345, "train_loss_llm": 7.212162017822266e-06, "grad_norm": 0.02131030522286892, "global_step": 2581, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04611198231577873, "train_loss_bc": 0.04611198231577873, "train_loss_llm": 0.0, "grad_norm": 0.02684902586042881, "global_step": 2582, "epoch": 71, "lr": 0.001} +{"train_loss": 0.03473994880914688, "train_loss_bc": 0.03473994880914688, "train_loss_llm": 0.0, "grad_norm": 0.03272605687379837, "global_step": 2583, "epoch": 71, "lr": 0.001} +{"train_loss": 0.05379807949066162, "train_loss_bc": 0.05379807949066162, "train_loss_llm": 0.0, "grad_norm": 0.04209768399596214, "global_step": 2584, "epoch": 71, "lr": 0.001} +{"train_loss": 0.040358610451221466, "train_loss_bc": 0.040354009717702866, "train_loss_llm": 0.00046005845069885254, "grad_norm": 0.016704069450497627, "global_step": 2585, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04837924614548683, "train_loss_bc": 0.048355553299188614, "train_loss_llm": 0.002369225025177002, "grad_norm": 0.022243699058890343, "global_step": 2586, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04956534877419472, "train_loss_bc": 0.04956534877419472, "train_loss_llm": 0.0, "grad_norm": 0.03427362069487572, "global_step": 2587, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04824848473072052, "train_loss_bc": 0.04823458567261696, "train_loss_llm": 0.0013898611068725586, "grad_norm": 0.042770497500896454, "global_step": 2588, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04313937574625015, "train_loss_bc": 0.04312928766012192, "train_loss_llm": 0.0010088980197906494, "grad_norm": 0.016547800973057747, "global_step": 2589, "epoch": 71, "lr": 0.001} +{"train_loss": 0.033833473920822144, "train_loss_bc": 0.033833473920822144, "train_loss_llm": 0.0, "grad_norm": 0.024099895730614662, "global_step": 2590, "epoch": 71, "lr": 0.001} +{"train_loss": 0.04483680261505975, "train_loss_bc": 0.0481199212372303, "train_loss_llm": 0.0, "grad_norm": 0.03604962304234505, "global_step": 2591, "epoch": 71, "lr": 0.001, "val_loss": 0.0510522723197937} +{"train_loss": 0.04877364635467529, "train_loss_bc": 0.04877364635467529, "train_loss_llm": 0.0, "grad_norm": 0.0456467904150486, "global_step": 2592, "epoch": 72, "lr": 0.001} +{"train_loss": 0.035400163382291794, "train_loss_bc": 0.035400163382291794, "train_loss_llm": 0.0, "grad_norm": 0.014018561691045761, "global_step": 2593, "epoch": 72, "lr": 0.001} +{"train_loss": 0.045665085315704346, "train_loss_bc": 0.045665085315704346, "train_loss_llm": 0.0, "grad_norm": 0.023543275892734528, "global_step": 2594, "epoch": 72, "lr": 0.001} +{"train_loss": 0.049809493124485016, "train_loss_bc": 0.049794699996709824, "train_loss_llm": 0.0014792382717132568, "grad_norm": 0.032428741455078125, "global_step": 2595, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04146641492843628, "train_loss_bc": 0.04146641492843628, "train_loss_llm": 0.0, "grad_norm": 0.04241422936320305, "global_step": 2596, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04107875004410744, "train_loss_bc": 0.04107875004410744, "train_loss_llm": 0.0, "grad_norm": 0.016483746469020844, "global_step": 2597, "epoch": 72, "lr": 0.001} +{"train_loss": 0.049521129578351974, "train_loss_bc": 0.04950866103172302, "train_loss_llm": 0.0012466907501220703, "grad_norm": 0.019650183618068695, "global_step": 2598, "epoch": 72, "lr": 0.001} +{"train_loss": 0.049474086612463, "train_loss_bc": 0.049469806253910065, "train_loss_llm": 0.00042808055877685547, "grad_norm": 0.02683420665562153, "global_step": 2599, "epoch": 72, "lr": 0.001} +{"train_loss": 0.03900672495365143, "train_loss_bc": 0.03900672495365143, "train_loss_llm": 0.0, "grad_norm": 0.0330699197947979, "global_step": 2600, "epoch": 72, "lr": 0.001} +{"train_loss": 0.055101267993450165, "train_loss_bc": 0.055087875574827194, "train_loss_llm": 0.0013391673564910889, "grad_norm": 0.020068520680069923, "global_step": 2601, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04316268116235733, "train_loss_bc": 0.04316268116235733, "train_loss_llm": 0.0, "grad_norm": 0.026345040649175644, "global_step": 2602, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04626304656267166, "train_loss_bc": 0.04623287916183472, "train_loss_llm": 0.003016829490661621, "grad_norm": 0.031943537294864655, "global_step": 2603, "epoch": 72, "lr": 0.001} +{"train_loss": 0.037372469902038574, "train_loss_bc": 0.03736481815576553, "train_loss_llm": 0.0007653534412384033, "grad_norm": 0.04045034199953079, "global_step": 2604, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04178036376833916, "train_loss_bc": 0.041778672486543655, "train_loss_llm": 0.00016900897026062012, "grad_norm": 0.015309793874621391, "global_step": 2605, "epoch": 72, "lr": 0.001} +{"train_loss": 0.041833363473415375, "train_loss_bc": 0.041833363473415375, "train_loss_llm": 0.0, "grad_norm": 0.022845977917313576, "global_step": 2606, "epoch": 72, "lr": 0.001} +{"train_loss": 0.037126168608665466, "train_loss_bc": 0.037126168608665466, "train_loss_llm": 0.0, "grad_norm": 0.030907640233635902, "global_step": 2607, "epoch": 72, "lr": 0.001} +{"train_loss": 0.044492628425359726, "train_loss_bc": 0.044492628425359726, "train_loss_llm": 0.0, "grad_norm": 0.039237793534994125, "global_step": 2608, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04100034013390541, "train_loss_bc": 0.040989235043525696, "train_loss_llm": 0.0011105239391326904, "grad_norm": 0.016663264483213425, "global_step": 2609, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04553142562508583, "train_loss_bc": 0.04553011804819107, "train_loss_llm": 0.00013080239295959473, "grad_norm": 0.031075291335582733, "global_step": 2610, "epoch": 72, "lr": 0.001} +{"train_loss": 0.03811522573232651, "train_loss_bc": 0.03811522573232651, "train_loss_llm": 0.0, "grad_norm": 0.039481230080127716, "global_step": 2611, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04011690244078636, "train_loss_bc": 0.04011690244078636, "train_loss_llm": 0.0, "grad_norm": 0.04734951630234718, "global_step": 2612, "epoch": 72, "lr": 0.001} +{"train_loss": 0.044702354818582535, "train_loss_bc": 0.044702354818582535, "train_loss_llm": 0.0, "grad_norm": 0.021930864080786705, "global_step": 2613, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04739977791905403, "train_loss_bc": 0.04739977791905403, "train_loss_llm": 0.0, "grad_norm": 0.025519782677292824, "global_step": 2614, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04210976883769035, "train_loss_bc": 0.04210976883769035, "train_loss_llm": 0.0, "grad_norm": 0.0336165651679039, "global_step": 2615, "epoch": 72, "lr": 0.001} +{"train_loss": 0.05365815386176109, "train_loss_bc": 0.05365815386176109, "train_loss_llm": 0.0, "grad_norm": 0.04604552313685417, "global_step": 2616, "epoch": 72, "lr": 0.001} +{"train_loss": 0.047352150082588196, "train_loss_bc": 0.047352150082588196, "train_loss_llm": 0.0, "grad_norm": 0.018563056364655495, "global_step": 2617, "epoch": 72, "lr": 0.001} +{"train_loss": 0.05449007824063301, "train_loss_bc": 0.05448482930660248, "train_loss_llm": 0.0005247890949249268, "grad_norm": 0.026464449241757393, "global_step": 2618, "epoch": 72, "lr": 0.001} +{"train_loss": 0.044295746833086014, "train_loss_bc": 0.044242002069950104, "train_loss_llm": 0.00537458062171936, "grad_norm": 0.030561313033103943, "global_step": 2619, "epoch": 72, "lr": 0.001} +{"train_loss": 0.050257641822099686, "train_loss_bc": 0.050233278423547745, "train_loss_llm": 0.0024362802505493164, "grad_norm": 0.036978334188461304, "global_step": 2620, "epoch": 72, "lr": 0.001} +{"train_loss": 0.042290136218070984, "train_loss_bc": 0.042290136218070984, "train_loss_llm": 0.0, "grad_norm": 0.016863016411662102, "global_step": 2621, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04810747876763344, "train_loss_bc": 0.04810747876763344, "train_loss_llm": 0.0, "grad_norm": 0.024463584646582603, "global_step": 2622, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04402823746204376, "train_loss_bc": 0.04402823746204376, "train_loss_llm": 0.0, "grad_norm": 0.03209627792239189, "global_step": 2623, "epoch": 72, "lr": 0.001} +{"train_loss": 0.05402856320142746, "train_loss_bc": 0.05400542542338371, "train_loss_llm": 0.0023136138916015625, "grad_norm": 0.03732334449887276, "global_step": 2624, "epoch": 72, "lr": 0.001} +{"train_loss": 0.035839904099702835, "train_loss_bc": 0.035838790237903595, "train_loss_llm": 0.0001112818717956543, "grad_norm": 0.01734420470893383, "global_step": 2625, "epoch": 72, "lr": 0.001} +{"train_loss": 0.040718089789152145, "train_loss_bc": 0.04071107134222984, "train_loss_llm": 0.0007018744945526123, "grad_norm": 0.026624688878655434, "global_step": 2626, "epoch": 72, "lr": 0.001} +{"train_loss": 0.04475332724137439, "train_loss_bc": 0.049725160002708435, "train_loss_llm": 0.0025161802768707275, "grad_norm": 0.03759898245334625, "global_step": 2627, "epoch": 72, "lr": 0.001, "val_loss": 0.05124782398343086} +{"train_loss": 0.04330965504050255, "train_loss_bc": 0.04330606013536453, "train_loss_llm": 0.00035959482192993164, "grad_norm": 0.0485672801733017, "global_step": 2628, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04287645220756531, "train_loss_bc": 0.04287645220756531, "train_loss_llm": 0.0, "grad_norm": 0.015984227880835533, "global_step": 2629, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04585779830813408, "train_loss_bc": 0.04585779830813408, "train_loss_llm": 0.0, "grad_norm": 0.022519893944263458, "global_step": 2630, "epoch": 73, "lr": 0.001} +{"train_loss": 0.048242826014757156, "train_loss_bc": 0.048242826014757156, "train_loss_llm": 0.0, "grad_norm": 0.03051288053393364, "global_step": 2631, "epoch": 73, "lr": 0.001} +{"train_loss": 0.03754216060042381, "train_loss_bc": 0.03753389045596123, "train_loss_llm": 0.0008269250392913818, "grad_norm": 0.043234068900346756, "global_step": 2632, "epoch": 73, "lr": 0.001} +{"train_loss": 0.036715541034936905, "train_loss_bc": 0.036715541034936905, "train_loss_llm": 0.0, "grad_norm": 0.01490043941885233, "global_step": 2633, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04097224026918411, "train_loss_bc": 0.04096774384379387, "train_loss_llm": 0.0004495382308959961, "grad_norm": 0.02459564246237278, "global_step": 2634, "epoch": 73, "lr": 0.001} +{"train_loss": 0.042907439172267914, "train_loss_bc": 0.042907439172267914, "train_loss_llm": 0.0, "grad_norm": 0.042546115815639496, "global_step": 2635, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04404985159635544, "train_loss_bc": 0.04404985159635544, "train_loss_llm": 0.0, "grad_norm": 0.054088201373815536, "global_step": 2636, "epoch": 73, "lr": 0.001} +{"train_loss": 0.050802428275346756, "train_loss_bc": 0.050752658396959305, "train_loss_llm": 0.004976809024810791, "grad_norm": 0.020004961639642715, "global_step": 2637, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04075850918889046, "train_loss_bc": 0.04074589163064957, "train_loss_llm": 0.001261681318283081, "grad_norm": 0.026807822287082672, "global_step": 2638, "epoch": 73, "lr": 0.001} +{"train_loss": 0.035393133759498596, "train_loss_bc": 0.035393133759498596, "train_loss_llm": 0.0, "grad_norm": 0.028470022603869438, "global_step": 2639, "epoch": 73, "lr": 0.001} +{"train_loss": 0.048114243894815445, "train_loss_bc": 0.048114243894815445, "train_loss_llm": 0.0, "grad_norm": 0.03516388311982155, "global_step": 2640, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04361261799931526, "train_loss_bc": 0.04359659180045128, "train_loss_llm": 0.0016025304794311523, "grad_norm": 0.01387865748256445, "global_step": 2641, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04596593603491783, "train_loss_bc": 0.04596593603491783, "train_loss_llm": 0.0, "grad_norm": 0.02198830060660839, "global_step": 2642, "epoch": 73, "lr": 0.001} +{"train_loss": 0.044849906116724014, "train_loss_bc": 0.044849906116724014, "train_loss_llm": 0.0, "grad_norm": 0.02914717234671116, "global_step": 2643, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04069440811872482, "train_loss_bc": 0.040685124695301056, "train_loss_llm": 0.0009285211563110352, "grad_norm": 0.037406496703624725, "global_step": 2644, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04606294259428978, "train_loss_bc": 0.04606294259428978, "train_loss_llm": 0.0, "grad_norm": 0.01775132119655609, "global_step": 2645, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04730195552110672, "train_loss_bc": 0.04730195552110672, "train_loss_llm": 0.0, "grad_norm": 0.0245395228266716, "global_step": 2646, "epoch": 73, "lr": 0.001} +{"train_loss": 0.048655081540346146, "train_loss_bc": 0.048655081540346146, "train_loss_llm": 0.0, "grad_norm": 0.02931409701704979, "global_step": 2647, "epoch": 73, "lr": 0.001} +{"train_loss": 0.034922584891319275, "train_loss_bc": 0.034922584891319275, "train_loss_llm": 0.0, "grad_norm": 0.03136099874973297, "global_step": 2648, "epoch": 73, "lr": 0.001} +{"train_loss": 0.045929957181215286, "train_loss_bc": 0.04592462629079819, "train_loss_llm": 0.0005329251289367676, "grad_norm": 0.015544571913778782, "global_step": 2649, "epoch": 73, "lr": 0.001} +{"train_loss": 0.045912131667137146, "train_loss_bc": 0.045912131667137146, "train_loss_llm": 0.0, "grad_norm": 0.02344638668000698, "global_step": 2650, "epoch": 73, "lr": 0.001} +{"train_loss": 0.034700095653533936, "train_loss_bc": 0.034700095653533936, "train_loss_llm": 0.0, "grad_norm": 0.02732277475297451, "global_step": 2651, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04865216091275215, "train_loss_bc": 0.048625145107507706, "train_loss_llm": 0.00270158052444458, "grad_norm": 0.03700614720582962, "global_step": 2652, "epoch": 73, "lr": 0.001} +{"train_loss": 0.05122463032603264, "train_loss_bc": 0.05122463032603264, "train_loss_llm": 0.0, "grad_norm": 0.016288097947835922, "global_step": 2653, "epoch": 73, "lr": 0.001} +{"train_loss": 0.05523713305592537, "train_loss_bc": 0.05523713305592537, "train_loss_llm": 0.0, "grad_norm": 0.024494333192706108, "global_step": 2654, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04857240989804268, "train_loss_bc": 0.04856602102518082, "train_loss_llm": 0.000638812780380249, "grad_norm": 0.030762024223804474, "global_step": 2655, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04587608948349953, "train_loss_bc": 0.04584759473800659, "train_loss_llm": 0.0028495490550994873, "grad_norm": 0.03350818529725075, "global_step": 2656, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04943756386637688, "train_loss_bc": 0.049376290291547775, "train_loss_llm": 0.006127387285232544, "grad_norm": 0.015620469115674496, "global_step": 2657, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04930280148983002, "train_loss_bc": 0.04930280148983002, "train_loss_llm": 0.0, "grad_norm": 0.02304830029606819, "global_step": 2658, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04456011578440666, "train_loss_bc": 0.04456011578440666, "train_loss_llm": 0.0, "grad_norm": 0.02498375065624714, "global_step": 2659, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04124050214886665, "train_loss_bc": 0.04124050214886665, "train_loss_llm": 0.0, "grad_norm": 0.031087055802345276, "global_step": 2660, "epoch": 73, "lr": 0.001} +{"train_loss": 0.05040321126580238, "train_loss_bc": 0.05040321126580238, "train_loss_llm": 0.0, "grad_norm": 0.014177829958498478, "global_step": 2661, "epoch": 73, "lr": 0.001} +{"train_loss": 0.04492383077740669, "train_loss_bc": 0.04492383077740669, "train_loss_llm": 0.0, "grad_norm": 0.027513321489095688, "global_step": 2662, "epoch": 73, "lr": 0.001} +{"train_loss": 0.044614014215767384, "train_loss_bc": 0.04052416607737541, "train_loss_llm": 0.0, "grad_norm": 0.03209124132990837, "global_step": 2663, "epoch": 73, "lr": 0.001, "val_loss": 0.05152379721403122} +{"train_loss": 0.04167548194527626, "train_loss_bc": 0.04167548194527626, "train_loss_llm": 0.0, "grad_norm": 0.03347643464803696, "global_step": 2664, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04883013665676117, "train_loss_bc": 0.04883013665676117, "train_loss_llm": 0.0, "grad_norm": 0.012608213350176811, "global_step": 2665, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04620099067687988, "train_loss_bc": 0.0461924746632576, "train_loss_llm": 0.0008514225482940674, "grad_norm": 0.02055862359702587, "global_step": 2666, "epoch": 74, "lr": 0.001} +{"train_loss": 0.039632342755794525, "train_loss_bc": 0.03962220251560211, "train_loss_llm": 0.0010140538215637207, "grad_norm": 0.024485042318701744, "global_step": 2667, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04492706432938576, "train_loss_bc": 0.044900789856910706, "train_loss_llm": 0.0026274025440216064, "grad_norm": 0.03237288445234299, "global_step": 2668, "epoch": 74, "lr": 0.001} +{"train_loss": 0.042148519307374954, "train_loss_bc": 0.042148519307374954, "train_loss_llm": 0.0, "grad_norm": 0.013227079063653946, "global_step": 2669, "epoch": 74, "lr": 0.001} +{"train_loss": 0.037548504769802094, "train_loss_bc": 0.037548504769802094, "train_loss_llm": 0.0, "grad_norm": 0.025173025205731392, "global_step": 2670, "epoch": 74, "lr": 0.001} +{"train_loss": 0.038045912981033325, "train_loss_bc": 0.038045912981033325, "train_loss_llm": 0.0, "grad_norm": 0.03676807880401611, "global_step": 2671, "epoch": 74, "lr": 0.001} +{"train_loss": 0.0450197234749794, "train_loss_bc": 0.04499999061226845, "train_loss_llm": 0.0019731223583221436, "grad_norm": 0.041685234755277634, "global_step": 2672, "epoch": 74, "lr": 0.001} +{"train_loss": 0.038825083523988724, "train_loss_bc": 0.038825083523988724, "train_loss_llm": 0.0, "grad_norm": 0.015193923376500607, "global_step": 2673, "epoch": 74, "lr": 0.001} +{"train_loss": 0.048878252506256104, "train_loss_bc": 0.04886516183614731, "train_loss_llm": 0.001309126615524292, "grad_norm": 0.022884314879775047, "global_step": 2674, "epoch": 74, "lr": 0.001} +{"train_loss": 0.03531966358423233, "train_loss_bc": 0.03531966358423233, "train_loss_llm": 0.0, "grad_norm": 0.025804564356803894, "global_step": 2675, "epoch": 74, "lr": 0.001} +{"train_loss": 0.042104270309209824, "train_loss_bc": 0.042104270309209824, "train_loss_llm": 0.0, "grad_norm": 0.03710419312119484, "global_step": 2676, "epoch": 74, "lr": 0.001} +{"train_loss": 0.039544232189655304, "train_loss_bc": 0.03953483700752258, "train_loss_llm": 0.0009395778179168701, "grad_norm": 0.015532699413597584, "global_step": 2677, "epoch": 74, "lr": 0.001} +{"train_loss": 0.044812314212322235, "train_loss_bc": 0.044812314212322235, "train_loss_llm": 0.0, "grad_norm": 0.01997123844921589, "global_step": 2678, "epoch": 74, "lr": 0.001} +{"train_loss": 0.05230198800563812, "train_loss_bc": 0.05230198800563812, "train_loss_llm": 0.0, "grad_norm": 0.026144063100218773, "global_step": 2679, "epoch": 74, "lr": 0.001} +{"train_loss": 0.05102768912911415, "train_loss_bc": 0.05102768912911415, "train_loss_llm": 0.0, "grad_norm": 0.03575656935572624, "global_step": 2680, "epoch": 74, "lr": 0.001} +{"train_loss": 0.0427093431353569, "train_loss_bc": 0.0427093431353569, "train_loss_llm": 0.0, "grad_norm": 0.01322284247726202, "global_step": 2681, "epoch": 74, "lr": 0.001} +{"train_loss": 0.048664819449186325, "train_loss_bc": 0.048664819449186325, "train_loss_llm": 0.0, "grad_norm": 0.022877374663949013, "global_step": 2682, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04441190883517265, "train_loss_bc": 0.04441190883517265, "train_loss_llm": 0.0, "grad_norm": 0.026222744956612587, "global_step": 2683, "epoch": 74, "lr": 0.001} +{"train_loss": 0.05139927566051483, "train_loss_bc": 0.05139927566051483, "train_loss_llm": 0.0, "grad_norm": 0.033994030207395554, "global_step": 2684, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04588213190436363, "train_loss_bc": 0.045877229422330856, "train_loss_llm": 0.0004900693893432617, "grad_norm": 0.017677435651421547, "global_step": 2685, "epoch": 74, "lr": 0.001} +{"train_loss": 0.042989883571863174, "train_loss_bc": 0.042989883571863174, "train_loss_llm": 0.0, "grad_norm": 0.025569472461938858, "global_step": 2686, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04113740473985672, "train_loss_bc": 0.04113740473985672, "train_loss_llm": 0.0, "grad_norm": 0.03198247402906418, "global_step": 2687, "epoch": 74, "lr": 0.001} +{"train_loss": 0.03930426761507988, "train_loss_bc": 0.03930426761507988, "train_loss_llm": 0.0, "grad_norm": 0.03754457086324692, "global_step": 2688, "epoch": 74, "lr": 0.001} +{"train_loss": 0.047580089420080185, "train_loss_bc": 0.047580089420080185, "train_loss_llm": 0.0, "grad_norm": 0.019045036286115646, "global_step": 2689, "epoch": 74, "lr": 0.001} +{"train_loss": 0.05175429582595825, "train_loss_bc": 0.05175429582595825, "train_loss_llm": 0.0, "grad_norm": 0.027380630373954773, "global_step": 2690, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04416042938828468, "train_loss_bc": 0.04416042938828468, "train_loss_llm": 0.0, "grad_norm": 0.03304266184568405, "global_step": 2691, "epoch": 74, "lr": 0.001} +{"train_loss": 0.035409051924943924, "train_loss_bc": 0.035409051924943924, "train_loss_llm": 0.0, "grad_norm": 0.04441060498356819, "global_step": 2692, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04517636075615883, "train_loss_bc": 0.04517636075615883, "train_loss_llm": 0.0, "grad_norm": 0.016546789556741714, "global_step": 2693, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04724249988794327, "train_loss_bc": 0.04722106084227562, "train_loss_llm": 0.0021438300609588623, "grad_norm": 0.019939692690968513, "global_step": 2694, "epoch": 74, "lr": 0.001} +{"train_loss": 0.052780892699956894, "train_loss_bc": 0.052780892699956894, "train_loss_llm": 0.0, "grad_norm": 0.029535826295614243, "global_step": 2695, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04868302866816521, "train_loss_bc": 0.04865114390850067, "train_loss_llm": 0.0031883716583251953, "grad_norm": 0.033488333225250244, "global_step": 2696, "epoch": 74, "lr": 0.001} +{"train_loss": 0.03949551656842232, "train_loss_bc": 0.03949551656842232, "train_loss_llm": 0.0, "grad_norm": 0.016459215432405472, "global_step": 2697, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04861973971128464, "train_loss_bc": 0.048616230487823486, "train_loss_llm": 0.0003510415554046631, "grad_norm": 0.024243555963039398, "global_step": 2698, "epoch": 74, "lr": 0.001} +{"train_loss": 0.04464884609397915, "train_loss_bc": 0.05309005081653595, "train_loss_llm": 0.0025297999382019043, "grad_norm": 0.04038993641734123, "global_step": 2699, "epoch": 74, "lr": 0.001, "val_loss": 0.05089007690548897} +{"train_loss": 0.04568091034889221, "train_loss_bc": 0.04567917808890343, "train_loss_llm": 0.00017306208610534668, "grad_norm": 0.044383592903614044, "global_step": 2700, "epoch": 75, "lr": 0.001} +{"train_loss": 0.037194278091192245, "train_loss_bc": 0.037194278091192245, "train_loss_llm": 0.0, "grad_norm": 0.012543460354208946, "global_step": 2701, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04648526757955551, "train_loss_bc": 0.04648526757955551, "train_loss_llm": 0.0, "grad_norm": 0.019246840849518776, "global_step": 2702, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04092498496174812, "train_loss_bc": 0.04092498496174812, "train_loss_llm": 0.0, "grad_norm": 0.026573533192276955, "global_step": 2703, "epoch": 75, "lr": 0.001} +{"train_loss": 0.051685452461242676, "train_loss_bc": 0.051683373749256134, "train_loss_llm": 0.0002079606056213379, "grad_norm": 0.030895819887518883, "global_step": 2704, "epoch": 75, "lr": 0.001} +{"train_loss": 0.040145143866539, "train_loss_bc": 0.040145143866539, "train_loss_llm": 0.0, "grad_norm": 0.014312629587948322, "global_step": 2705, "epoch": 75, "lr": 0.001} +{"train_loss": 0.048881396651268005, "train_loss_bc": 0.048881396651268005, "train_loss_llm": 0.0, "grad_norm": 0.02281159535050392, "global_step": 2706, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04438408464193344, "train_loss_bc": 0.04438408464193344, "train_loss_llm": 0.0, "grad_norm": 0.037247683852910995, "global_step": 2707, "epoch": 75, "lr": 0.001} +{"train_loss": 0.0457908995449543, "train_loss_bc": 0.04576919227838516, "train_loss_llm": 0.0021705925464630127, "grad_norm": 0.05748850479722023, "global_step": 2708, "epoch": 75, "lr": 0.001} +{"train_loss": 0.041642263531684875, "train_loss_bc": 0.041642263531684875, "train_loss_llm": 0.0, "grad_norm": 0.015210818499326706, "global_step": 2709, "epoch": 75, "lr": 0.001} +{"train_loss": 0.039630256593227386, "train_loss_bc": 0.039630256593227386, "train_loss_llm": 0.0, "grad_norm": 0.023286473006010056, "global_step": 2710, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04598889872431755, "train_loss_bc": 0.04598889872431755, "train_loss_llm": 0.0, "grad_norm": 0.03751680999994278, "global_step": 2711, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04250890761613846, "train_loss_bc": 0.04250890761613846, "train_loss_llm": 0.0, "grad_norm": 0.04570833966135979, "global_step": 2712, "epoch": 75, "lr": 0.001} +{"train_loss": 0.047721993178129196, "train_loss_bc": 0.047721993178129196, "train_loss_llm": 0.0, "grad_norm": 0.01582174561917782, "global_step": 2713, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04869619756937027, "train_loss_bc": 0.04869619756937027, "train_loss_llm": 0.0, "grad_norm": 0.028636440634727478, "global_step": 2714, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04357902705669403, "train_loss_bc": 0.04357137903571129, "train_loss_llm": 0.0007649064064025879, "grad_norm": 0.029578328132629395, "global_step": 2715, "epoch": 75, "lr": 0.001} +{"train_loss": 0.048261698335409164, "train_loss_bc": 0.048261698335409164, "train_loss_llm": 0.0, "grad_norm": 0.04129338264465332, "global_step": 2716, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04031926020979881, "train_loss_bc": 0.04031926020979881, "train_loss_llm": 0.0, "grad_norm": 0.014902648515999317, "global_step": 2717, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04255461320281029, "train_loss_bc": 0.04253525286912918, "train_loss_llm": 0.0019359588623046875, "grad_norm": 0.022774916142225266, "global_step": 2718, "epoch": 75, "lr": 0.001} +{"train_loss": 0.044872939586639404, "train_loss_bc": 0.044872939586639404, "train_loss_llm": 0.0, "grad_norm": 0.027504784986376762, "global_step": 2719, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04217001423239708, "train_loss_bc": 0.042164407670497894, "train_loss_llm": 0.0005605816841125488, "grad_norm": 0.0337987020611763, "global_step": 2720, "epoch": 75, "lr": 0.001} +{"train_loss": 0.05052145570516586, "train_loss_bc": 0.05052145570516586, "train_loss_llm": 0.0, "grad_norm": 0.016141988337039948, "global_step": 2721, "epoch": 75, "lr": 0.001} +{"train_loss": 0.05520130321383476, "train_loss_bc": 0.05520130321383476, "train_loss_llm": 0.0, "grad_norm": 0.02507186122238636, "global_step": 2722, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04506690427660942, "train_loss_bc": 0.0450531467795372, "train_loss_llm": 0.0013758838176727295, "grad_norm": 0.039549097418785095, "global_step": 2723, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04981960728764534, "train_loss_bc": 0.04981960728764534, "train_loss_llm": 0.0, "grad_norm": 0.05389458313584328, "global_step": 2724, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04072900488972664, "train_loss_bc": 0.040710315108299255, "train_loss_llm": 0.0018689930438995361, "grad_norm": 0.018167542293667793, "global_step": 2725, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04384608194231987, "train_loss_bc": 0.04384608194231987, "train_loss_llm": 0.0, "grad_norm": 0.0300582367926836, "global_step": 2726, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04096429422497749, "train_loss_bc": 0.0409596748650074, "train_loss_llm": 0.0004621148109436035, "grad_norm": 0.03351237624883652, "global_step": 2727, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04577001929283142, "train_loss_bc": 0.04577001929283142, "train_loss_llm": 0.0, "grad_norm": 0.0402861051261425, "global_step": 2728, "epoch": 75, "lr": 0.001} +{"train_loss": 0.0464160330593586, "train_loss_bc": 0.04637943208217621, "train_loss_llm": 0.003660142421722412, "grad_norm": 0.01603560708463192, "global_step": 2729, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04090781509876251, "train_loss_bc": 0.04088638722896576, "train_loss_llm": 0.0021427571773529053, "grad_norm": 0.024301571771502495, "global_step": 2730, "epoch": 75, "lr": 0.001} +{"train_loss": 0.0449853390455246, "train_loss_bc": 0.0449853390455246, "train_loss_llm": 0.0, "grad_norm": 0.027917956933379173, "global_step": 2731, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04674161598086357, "train_loss_bc": 0.04674140363931656, "train_loss_llm": 2.136826515197754e-05, "grad_norm": 0.03278220817446709, "global_step": 2732, "epoch": 75, "lr": 0.001} +{"train_loss": 0.030316315591335297, "train_loss_bc": 0.030316315591335297, "train_loss_llm": 0.0, "grad_norm": 0.012777473777532578, "global_step": 2733, "epoch": 75, "lr": 0.001} +{"train_loss": 0.04926849901676178, "train_loss_bc": 0.04926849901676178, "train_loss_llm": 0.0, "grad_norm": 0.02363225817680359, "global_step": 2734, "epoch": 75, "lr": 0.001} +{"train_loss": 0.0445737327552504, "train_loss_bc": 0.04498160257935524, "train_loss_llm": 0.0, "grad_norm": 0.03334737569093704, "global_step": 2735, "epoch": 75, "lr": 0.001, "train/cumulative_reward": 69.22960148800499, "train/mean_score": 5.6203930404452125, "train/success_rate": 0.1, "test/cumulative_reward": 48.402232637943996, "test/mean_score": 7.169443441369145, "test/success_rate": 0.16, "val_loss": 0.05101681500673294, "train_action_mse_error": 0.04824916645884514} +{"train_loss": 0.04631893336772919, "train_loss_bc": 0.04631893336772919, "train_loss_llm": 0.0, "grad_norm": 0.03601447492837906, "global_step": 2736, "epoch": 76, "lr": 0.001} +{"train_loss": 0.045608267188072205, "train_loss_bc": 0.045608267188072205, "train_loss_llm": 0.0, "grad_norm": 0.01850942336022854, "global_step": 2737, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04367730766534805, "train_loss_bc": 0.04367730766534805, "train_loss_llm": 0.0, "grad_norm": 0.02521883323788643, "global_step": 2738, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04714164137840271, "train_loss_bc": 0.04713626950979233, "train_loss_llm": 0.0005372464656829834, "grad_norm": 0.03156472370028496, "global_step": 2739, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04806164279580116, "train_loss_bc": 0.04806164279580116, "train_loss_llm": 0.0, "grad_norm": 0.04083826020359993, "global_step": 2740, "epoch": 76, "lr": 0.001} +{"train_loss": 0.038323864340782166, "train_loss_bc": 0.038323864340782166, "train_loss_llm": 0.0, "grad_norm": 0.017892740666866302, "global_step": 2741, "epoch": 76, "lr": 0.001} +{"train_loss": 0.044149357825517654, "train_loss_bc": 0.04412062466144562, "train_loss_llm": 0.0028732717037200928, "grad_norm": 0.026478294283151627, "global_step": 2742, "epoch": 76, "lr": 0.001} +{"train_loss": 0.05475466698408127, "train_loss_bc": 0.054744262248277664, "train_loss_llm": 0.0010405480861663818, "grad_norm": 0.03126436099410057, "global_step": 2743, "epoch": 76, "lr": 0.001} +{"train_loss": 0.03933524340391159, "train_loss_bc": 0.03931739553809166, "train_loss_llm": 0.00178489089012146, "grad_norm": 0.03996488079428673, "global_step": 2744, "epoch": 76, "lr": 0.001} +{"train_loss": 0.042463842779397964, "train_loss_bc": 0.04243941977620125, "train_loss_llm": 0.002442300319671631, "grad_norm": 0.011994430795311928, "global_step": 2745, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04728125408291817, "train_loss_bc": 0.04728125408291817, "train_loss_llm": 0.0, "grad_norm": 0.021295050159096718, "global_step": 2746, "epoch": 76, "lr": 0.001} +{"train_loss": 0.043725043535232544, "train_loss_bc": 0.043688856065273285, "train_loss_llm": 0.0036187469959259033, "grad_norm": 0.027228299528360367, "global_step": 2747, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04091701656579971, "train_loss_bc": 0.04091701656579971, "train_loss_llm": 0.0, "grad_norm": 0.031098533421754837, "global_step": 2748, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04993542283773422, "train_loss_bc": 0.04993542283773422, "train_loss_llm": 0.0, "grad_norm": 0.02190546505153179, "global_step": 2749, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04247570037841797, "train_loss_bc": 0.0424686074256897, "train_loss_llm": 0.0007094740867614746, "grad_norm": 0.031174391508102417, "global_step": 2750, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04092691093683243, "train_loss_bc": 0.04092691093683243, "train_loss_llm": 0.0, "grad_norm": 0.032623257488012314, "global_step": 2751, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04161826893687248, "train_loss_bc": 0.04161826893687248, "train_loss_llm": 0.0, "grad_norm": 0.037772469222545624, "global_step": 2752, "epoch": 76, "lr": 0.001} +{"train_loss": 0.051018960773944855, "train_loss_bc": 0.051018960773944855, "train_loss_llm": 0.0, "grad_norm": 0.01526708621531725, "global_step": 2753, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04000157490372658, "train_loss_bc": 0.03998425602912903, "train_loss_llm": 0.0017319917678833008, "grad_norm": 0.022239115089178085, "global_step": 2754, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04394088685512543, "train_loss_bc": 0.04394088685512543, "train_loss_llm": 0.0, "grad_norm": 0.028737299144268036, "global_step": 2755, "epoch": 76, "lr": 0.001} +{"train_loss": 0.03755650669336319, "train_loss_bc": 0.03753617778420448, "train_loss_llm": 0.0020329654216766357, "grad_norm": 0.03194062411785126, "global_step": 2756, "epoch": 76, "lr": 0.001} +{"train_loss": 0.05279391258955002, "train_loss_bc": 0.052782390266656876, "train_loss_llm": 0.0011522769927978516, "grad_norm": 0.018672069534659386, "global_step": 2757, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04553402215242386, "train_loss_bc": 0.04552929475903511, "train_loss_llm": 0.00047278404235839844, "grad_norm": 0.02250104397535324, "global_step": 2758, "epoch": 76, "lr": 0.001} +{"train_loss": 0.03702007234096527, "train_loss_bc": 0.03702007234096527, "train_loss_llm": 0.0, "grad_norm": 0.02395973913371563, "global_step": 2759, "epoch": 76, "lr": 0.001} +{"train_loss": 0.0505625456571579, "train_loss_bc": 0.05054101347923279, "train_loss_llm": 0.0021531283855438232, "grad_norm": 0.03148879483342171, "global_step": 2760, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04629649594426155, "train_loss_bc": 0.04629649594426155, "train_loss_llm": 0.0, "grad_norm": 0.014879034832119942, "global_step": 2761, "epoch": 76, "lr": 0.001} +{"train_loss": 0.04423659294843674, "train_loss_bc": 0.04423659294843674, "train_loss_llm": 0.0, "grad_norm": 0.02747378684580326, "global_step": 2762, "epoch": 76, "lr": 0.001} +{"train_loss": 0.048936184495687485, "train_loss_bc": 0.04892173781991005, "train_loss_llm": 0.0014447569847106934, "grad_norm": 0.031821660697460175, "global_step": 2763, "epoch": 76, "lr": 0.001} +{"train_loss": 0.0411052405834198, "train_loss_bc": 0.0411052405834198, "train_loss_llm": 0.0, "grad_norm": 0.03855184465646744, "global_step": 2764, "epoch": 76, "lr": 0.001} +{"train_loss": 0.03961119055747986, "train_loss_bc": 0.03959958627820015, "train_loss_llm": 0.0011605322360992432, "grad_norm": 0.015178307890892029, "global_step": 2765, "epoch": 76, "lr": 0.001} +{"train_loss": 0.03760773688554764, "train_loss_bc": 0.03760773688554764, "train_loss_llm": 0.0, "grad_norm": 0.0240308977663517, "global_step": 2766, "epoch": 76, "lr": 0.001} +{"train_loss": 0.05367409437894821, "train_loss_bc": 0.05366576835513115, "train_loss_llm": 0.0008325576782226562, "grad_norm": 0.033760663121938705, "global_step": 2767, "epoch": 76, "lr": 0.001} +{"train_loss": 0.038819219917058945, "train_loss_bc": 0.038819219917058945, "train_loss_llm": 0.0, "grad_norm": 0.040629949420690536, "global_step": 2768, "epoch": 76, "lr": 0.001} +{"train_loss": 0.05652608349919319, "train_loss_bc": 0.05652608349919319, "train_loss_llm": 0.0, "grad_norm": 0.015550720505416393, "global_step": 2769, "epoch": 76, "lr": 0.001} +{"train_loss": 0.03480048477649689, "train_loss_bc": 0.03480048477649689, "train_loss_llm": 0.0, "grad_norm": 0.02454596571624279, "global_step": 2770, "epoch": 76, "lr": 0.001} +{"train_loss": 0.044373531722360186, "train_loss_bc": 0.04069095104932785, "train_loss_llm": 0.0, "grad_norm": 0.03340350091457367, "global_step": 2771, "epoch": 76, "lr": 0.001, "val_loss": 0.05141821503639221} +{"train_loss": 0.04894376918673515, "train_loss_bc": 0.048906147480010986, "train_loss_llm": 0.0037620067596435547, "grad_norm": 0.03930799663066864, "global_step": 2772, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04427054896950722, "train_loss_bc": 0.04427054896950722, "train_loss_llm": 0.0, "grad_norm": 0.015839824452996254, "global_step": 2773, "epoch": 77, "lr": 0.001} +{"train_loss": 0.037396449595689774, "train_loss_bc": 0.03739350661635399, "train_loss_llm": 0.00029417872428894043, "grad_norm": 0.021641701459884644, "global_step": 2774, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04502202197909355, "train_loss_bc": 0.04502202197909355, "train_loss_llm": 0.0, "grad_norm": 0.026376331225037575, "global_step": 2775, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04677559435367584, "train_loss_bc": 0.04677559435367584, "train_loss_llm": 0.0, "grad_norm": 0.02993614599108696, "global_step": 2776, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04919450730085373, "train_loss_bc": 0.049194201827049255, "train_loss_llm": 3.0517578125e-05, "grad_norm": 0.019509682431817055, "global_step": 2777, "epoch": 77, "lr": 0.001} +{"train_loss": 0.0428924560546875, "train_loss_bc": 0.0428924560546875, "train_loss_llm": 0.0, "grad_norm": 0.029721636325120926, "global_step": 2778, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04086018726229668, "train_loss_bc": 0.04086018726229668, "train_loss_llm": 0.0, "grad_norm": 0.03117138333618641, "global_step": 2779, "epoch": 77, "lr": 0.001} +{"train_loss": 0.05229095369577408, "train_loss_bc": 0.052275367081165314, "train_loss_llm": 0.0015586018562316895, "grad_norm": 0.035343244671821594, "global_step": 2780, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04704374074935913, "train_loss_bc": 0.04704374074935913, "train_loss_llm": 0.0, "grad_norm": 0.01498335786163807, "global_step": 2781, "epoch": 77, "lr": 0.001} +{"train_loss": 0.05107735097408295, "train_loss_bc": 0.05107735097408295, "train_loss_llm": 0.0, "grad_norm": 0.02043807879090309, "global_step": 2782, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04242538660764694, "train_loss_bc": 0.04242538660764694, "train_loss_llm": 0.0, "grad_norm": 0.026126578450202942, "global_step": 2783, "epoch": 77, "lr": 0.001} +{"train_loss": 0.03966602683067322, "train_loss_bc": 0.03965922072529793, "train_loss_llm": 0.0006807148456573486, "grad_norm": 0.0349113792181015, "global_step": 2784, "epoch": 77, "lr": 0.001} +{"train_loss": 0.0490633100271225, "train_loss_bc": 0.0490633100271225, "train_loss_llm": 0.0, "grad_norm": 0.01676318049430847, "global_step": 2785, "epoch": 77, "lr": 0.001} +{"train_loss": 0.03445064648985863, "train_loss_bc": 0.03445064648985863, "train_loss_llm": 0.0, "grad_norm": 0.02377401292324066, "global_step": 2786, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04890689253807068, "train_loss_bc": 0.048902347683906555, "train_loss_llm": 0.00045430660247802734, "grad_norm": 0.0315212644636631, "global_step": 2787, "epoch": 77, "lr": 0.001} +{"train_loss": 0.03815685585141182, "train_loss_bc": 0.03815685585141182, "train_loss_llm": 0.0, "grad_norm": 0.040125325322151184, "global_step": 2788, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04154466092586517, "train_loss_bc": 0.04154466092586517, "train_loss_llm": 0.0, "grad_norm": 0.014281372539699078, "global_step": 2789, "epoch": 77, "lr": 0.001} +{"train_loss": 0.044004958122968674, "train_loss_bc": 0.044004958122968674, "train_loss_llm": 0.0, "grad_norm": 0.029125545173883438, "global_step": 2790, "epoch": 77, "lr": 0.001} +{"train_loss": 0.05328908562660217, "train_loss_bc": 0.05328908562660217, "train_loss_llm": 0.0, "grad_norm": 0.035385873168706894, "global_step": 2791, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04302297532558441, "train_loss_bc": 0.04302297532558441, "train_loss_llm": 0.0, "grad_norm": 0.03719347342848778, "global_step": 2792, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04264817014336586, "train_loss_bc": 0.04264817014336586, "train_loss_llm": 0.0, "grad_norm": 0.016882682219147682, "global_step": 2793, "epoch": 77, "lr": 0.001} +{"train_loss": 0.05096707493066788, "train_loss_bc": 0.050966259092092514, "train_loss_llm": 8.162856101989746e-05, "grad_norm": 0.03076518513262272, "global_step": 2794, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04441288113594055, "train_loss_bc": 0.04441288113594055, "train_loss_llm": 0.0, "grad_norm": 0.03755616396665573, "global_step": 2795, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04149160161614418, "train_loss_bc": 0.041464656591415405, "train_loss_llm": 0.0026944875717163086, "grad_norm": 0.040999650955200195, "global_step": 2796, "epoch": 77, "lr": 0.001} +{"train_loss": 0.03892497345805168, "train_loss_bc": 0.03892497345805168, "train_loss_llm": 0.0, "grad_norm": 0.016666974872350693, "global_step": 2797, "epoch": 77, "lr": 0.001} +{"train_loss": 0.041175805032253265, "train_loss_bc": 0.04115765169262886, "train_loss_llm": 0.0018152892589569092, "grad_norm": 0.021222863346338272, "global_step": 2798, "epoch": 77, "lr": 0.001} +{"train_loss": 0.038373954594135284, "train_loss_bc": 0.038373954594135284, "train_loss_llm": 0.0, "grad_norm": 0.02689235284924507, "global_step": 2799, "epoch": 77, "lr": 0.001} +{"train_loss": 0.047306712716817856, "train_loss_bc": 0.047306712716817856, "train_loss_llm": 0.0, "grad_norm": 0.03496495261788368, "global_step": 2800, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04957159236073494, "train_loss_bc": 0.04955865442752838, "train_loss_llm": 0.0012936890125274658, "grad_norm": 0.022784292697906494, "global_step": 2801, "epoch": 77, "lr": 0.001} +{"train_loss": 0.042240582406520844, "train_loss_bc": 0.042240582406520844, "train_loss_llm": 0.0, "grad_norm": 0.027760231867432594, "global_step": 2802, "epoch": 77, "lr": 0.001} +{"train_loss": 0.040675558149814606, "train_loss_bc": 0.04063660651445389, "train_loss_llm": 0.003894984722137451, "grad_norm": 0.033966708928346634, "global_step": 2803, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04761868715286255, "train_loss_bc": 0.04761868715286255, "train_loss_llm": 0.0, "grad_norm": 0.04842734709382057, "global_step": 2804, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04977218434214592, "train_loss_bc": 0.04977218434214592, "train_loss_llm": 0.0, "grad_norm": 0.016755321994423866, "global_step": 2805, "epoch": 77, "lr": 0.001} +{"train_loss": 0.03677334636449814, "train_loss_bc": 0.036765359342098236, "train_loss_llm": 0.0007987320423126221, "grad_norm": 0.024658484384417534, "global_step": 2806, "epoch": 77, "lr": 0.001} +{"train_loss": 0.04441119502815935, "train_loss_bc": 0.04649528115987778, "train_loss_llm": 0.005623877048492432, "grad_norm": 0.03572145104408264, "global_step": 2807, "epoch": 77, "lr": 0.001, "val_loss": 0.0507601797580719} +{"train_loss": 0.04115278273820877, "train_loss_bc": 0.04115278273820877, "train_loss_llm": 0.0, "grad_norm": 0.039072275161743164, "global_step": 2808, "epoch": 78, "lr": 0.001} +{"train_loss": 0.044987477362155914, "train_loss_bc": 0.044987477362155914, "train_loss_llm": 0.0, "grad_norm": 0.01882181130349636, "global_step": 2809, "epoch": 78, "lr": 0.001} +{"train_loss": 0.0474560521543026, "train_loss_bc": 0.0474560521543026, "train_loss_llm": 0.0, "grad_norm": 0.027407875284552574, "global_step": 2810, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04946129769086838, "train_loss_bc": 0.04944905638694763, "train_loss_llm": 0.001224219799041748, "grad_norm": 0.03456101194024086, "global_step": 2811, "epoch": 78, "lr": 0.001} +{"train_loss": 0.043130144476890564, "train_loss_bc": 0.04309723526239395, "train_loss_llm": 0.003290802240371704, "grad_norm": 0.042995210736989975, "global_step": 2812, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04997754096984863, "train_loss_bc": 0.04997754096984863, "train_loss_llm": 0.0, "grad_norm": 0.015231778845191002, "global_step": 2813, "epoch": 78, "lr": 0.001} +{"train_loss": 0.03762619569897652, "train_loss_bc": 0.03762619569897652, "train_loss_llm": 0.0, "grad_norm": 0.02149513177573681, "global_step": 2814, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04439552500844002, "train_loss_bc": 0.04439552500844002, "train_loss_llm": 0.0, "grad_norm": 0.024116335436701775, "global_step": 2815, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04024709388613701, "train_loss_bc": 0.04022328183054924, "train_loss_llm": 0.002381354570388794, "grad_norm": 0.027550164610147476, "global_step": 2816, "epoch": 78, "lr": 0.001} +{"train_loss": 0.0499330498278141, "train_loss_bc": 0.0499260276556015, "train_loss_llm": 0.0007021427154541016, "grad_norm": 0.015270478092133999, "global_step": 2817, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04346229508519173, "train_loss_bc": 0.04346229508519173, "train_loss_llm": 0.0, "grad_norm": 0.023935100063681602, "global_step": 2818, "epoch": 78, "lr": 0.001} +{"train_loss": 0.03600367158651352, "train_loss_bc": 0.0359962098300457, "train_loss_llm": 0.000746309757232666, "grad_norm": 0.02472887933254242, "global_step": 2819, "epoch": 78, "lr": 0.001} +{"train_loss": 0.047370538115501404, "train_loss_bc": 0.04734541103243828, "train_loss_llm": 0.0025126636028289795, "grad_norm": 0.03381729498505592, "global_step": 2820, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04385837912559509, "train_loss_bc": 0.04385837912559509, "train_loss_llm": 0.0, "grad_norm": 0.01580747216939926, "global_step": 2821, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04551387205719948, "train_loss_bc": 0.04551387205719948, "train_loss_llm": 0.0, "grad_norm": 0.02042904496192932, "global_step": 2822, "epoch": 78, "lr": 0.001} +{"train_loss": 0.041311781853437424, "train_loss_bc": 0.041311781853437424, "train_loss_llm": 0.0, "grad_norm": 0.02882266417145729, "global_step": 2823, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04761519655585289, "train_loss_bc": 0.04761519655585289, "train_loss_llm": 0.0, "grad_norm": 0.03931219130754471, "global_step": 2824, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04477445036172867, "train_loss_bc": 0.04477445036172867, "train_loss_llm": 0.0, "grad_norm": 0.0135600371286273, "global_step": 2825, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04640039801597595, "train_loss_bc": 0.04640039801597595, "train_loss_llm": 0.0, "grad_norm": 0.025316838175058365, "global_step": 2826, "epoch": 78, "lr": 0.001} +{"train_loss": 0.038253456354141235, "train_loss_bc": 0.038253456354141235, "train_loss_llm": 0.0, "grad_norm": 0.035576846450567245, "global_step": 2827, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04929517209529877, "train_loss_bc": 0.0492904931306839, "train_loss_llm": 0.0004680454730987549, "grad_norm": 0.040630996227264404, "global_step": 2828, "epoch": 78, "lr": 0.001} +{"train_loss": 0.053799599409103394, "train_loss_bc": 0.053799599409103394, "train_loss_llm": 0.0, "grad_norm": 0.02025136910378933, "global_step": 2829, "epoch": 78, "lr": 0.001} +{"train_loss": 0.03597814962267876, "train_loss_bc": 0.03596382215619087, "train_loss_llm": 0.0014328360557556152, "grad_norm": 0.02959325537085533, "global_step": 2830, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04352779686450958, "train_loss_bc": 0.0434926375746727, "train_loss_llm": 0.0035160481929779053, "grad_norm": 0.0344148613512516, "global_step": 2831, "epoch": 78, "lr": 0.001} +{"train_loss": 0.05102848634123802, "train_loss_bc": 0.05100709944963455, "train_loss_llm": 0.002138793468475342, "grad_norm": 0.0396064929664135, "global_step": 2832, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04323280602693558, "train_loss_bc": 0.04323280602693558, "train_loss_llm": 0.0, "grad_norm": 0.01344539225101471, "global_step": 2833, "epoch": 78, "lr": 0.001} +{"train_loss": 0.048333410173654556, "train_loss_bc": 0.048333410173654556, "train_loss_llm": 0.0, "grad_norm": 0.018663441762328148, "global_step": 2834, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04498306289315224, "train_loss_bc": 0.04498306289315224, "train_loss_llm": 0.0, "grad_norm": 0.025252413004636765, "global_step": 2835, "epoch": 78, "lr": 0.001} +{"train_loss": 0.03956352919340134, "train_loss_bc": 0.03956352919340134, "train_loss_llm": 0.0, "grad_norm": 0.03179648891091347, "global_step": 2836, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04546564817428589, "train_loss_bc": 0.04546097666025162, "train_loss_llm": 0.0004672408103942871, "grad_norm": 0.019144747406244278, "global_step": 2837, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04288468509912491, "train_loss_bc": 0.04288468509912491, "train_loss_llm": 0.0, "grad_norm": 0.023634955286979675, "global_step": 2838, "epoch": 78, "lr": 0.001} +{"train_loss": 0.0436965636909008, "train_loss_bc": 0.04369077831506729, "train_loss_llm": 0.0005784928798675537, "grad_norm": 0.03277577459812164, "global_step": 2839, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04264317452907562, "train_loss_bc": 0.04264317452907562, "train_loss_llm": 0.0, "grad_norm": 0.03611712157726288, "global_step": 2840, "epoch": 78, "lr": 0.001} +{"train_loss": 0.04919208958745003, "train_loss_bc": 0.04919208958745003, "train_loss_llm": 0.0, "grad_norm": 0.015122472308576107, "global_step": 2841, "epoch": 78, "lr": 0.001} +{"train_loss": 0.03311686962842941, "train_loss_bc": 0.03311265632510185, "train_loss_llm": 0.00042119622230529785, "grad_norm": 0.020983682945370674, "global_step": 2842, "epoch": 78, "lr": 0.001} +{"train_loss": 0.044315386873980366, "train_loss_bc": 0.04567568749189377, "train_loss_llm": 0.0005999505519866943, "grad_norm": 0.0265561044216156, "global_step": 2843, "epoch": 78, "lr": 0.001, "val_loss": 0.05102349445223808} +{"train_loss": 0.04764814302325249, "train_loss_bc": 0.04763608053326607, "train_loss_llm": 0.0012061595916748047, "grad_norm": 0.03433286026120186, "global_step": 2844, "epoch": 79, "lr": 0.001} +{"train_loss": 0.0444827526807785, "train_loss_bc": 0.0444827526807785, "train_loss_llm": 0.0, "grad_norm": 0.015192100778222084, "global_step": 2845, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04701707139611244, "train_loss_bc": 0.04699759557843208, "train_loss_llm": 0.0019477307796478271, "grad_norm": 0.019219713285565376, "global_step": 2846, "epoch": 79, "lr": 0.001} +{"train_loss": 0.045304883271455765, "train_loss_bc": 0.045304883271455765, "train_loss_llm": 0.0, "grad_norm": 0.02602512203156948, "global_step": 2847, "epoch": 79, "lr": 0.001} +{"train_loss": 0.045558374375104904, "train_loss_bc": 0.04555805027484894, "train_loss_llm": 3.2275915145874023e-05, "grad_norm": 0.032927967607975006, "global_step": 2848, "epoch": 79, "lr": 0.001} +{"train_loss": 0.046211834996938705, "train_loss_bc": 0.046211834996938705, "train_loss_llm": 0.0, "grad_norm": 0.01590562053024769, "global_step": 2849, "epoch": 79, "lr": 0.001} +{"train_loss": 0.0369311161339283, "train_loss_bc": 0.03692160174250603, "train_loss_llm": 0.0009513497352600098, "grad_norm": 0.025183379650115967, "global_step": 2850, "epoch": 79, "lr": 0.001} +{"train_loss": 0.040588606148958206, "train_loss_bc": 0.04058465734124184, "train_loss_llm": 0.0003947019577026367, "grad_norm": 0.02864784188568592, "global_step": 2851, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04749738425016403, "train_loss_bc": 0.04748852550983429, "train_loss_llm": 0.0008857250213623047, "grad_norm": 0.03406054526567459, "global_step": 2852, "epoch": 79, "lr": 0.001} +{"train_loss": 0.043493352830410004, "train_loss_bc": 0.043493352830410004, "train_loss_llm": 0.0, "grad_norm": 0.022497354075312614, "global_step": 2853, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04234738647937775, "train_loss_bc": 0.04234255477786064, "train_loss_llm": 0.0004830658435821533, "grad_norm": 0.028636179864406586, "global_step": 2854, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04313686117529869, "train_loss_bc": 0.04313686117529869, "train_loss_llm": 0.0, "grad_norm": 0.04157601296901703, "global_step": 2855, "epoch": 79, "lr": 0.001} +{"train_loss": 0.05018167942762375, "train_loss_bc": 0.05018167942762375, "train_loss_llm": 0.0, "grad_norm": 0.04619023576378822, "global_step": 2856, "epoch": 79, "lr": 0.001} +{"train_loss": 0.0423353835940361, "train_loss_bc": 0.0423353835940361, "train_loss_llm": 0.0, "grad_norm": 0.013722283765673637, "global_step": 2857, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04206544905900955, "train_loss_bc": 0.04206544905900955, "train_loss_llm": 0.0, "grad_norm": 0.021091748028993607, "global_step": 2858, "epoch": 79, "lr": 0.001} +{"train_loss": 0.048121098428964615, "train_loss_bc": 0.048121098428964615, "train_loss_llm": 0.0, "grad_norm": 0.027337055653333664, "global_step": 2859, "epoch": 79, "lr": 0.001} +{"train_loss": 0.043544359505176544, "train_loss_bc": 0.0435430184006691, "train_loss_llm": 0.000134199857711792, "grad_norm": 0.03127053380012512, "global_step": 2860, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04233025014400482, "train_loss_bc": 0.04233025014400482, "train_loss_llm": 0.0, "grad_norm": 0.01632096990942955, "global_step": 2861, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04034782573580742, "train_loss_bc": 0.040327899158000946, "train_loss_llm": 0.0019927620887756348, "grad_norm": 0.02839031256735325, "global_step": 2862, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04448946565389633, "train_loss_bc": 0.04448946565389633, "train_loss_llm": 0.0, "grad_norm": 0.033803194761276245, "global_step": 2863, "epoch": 79, "lr": 0.001} +{"train_loss": 0.044374920427799225, "train_loss_bc": 0.044374920427799225, "train_loss_llm": 0.0, "grad_norm": 0.04567018151283264, "global_step": 2864, "epoch": 79, "lr": 0.001} +{"train_loss": 0.051005445420742035, "train_loss_bc": 0.05099346116185188, "train_loss_llm": 0.0011982917785644531, "grad_norm": 0.013975984416902065, "global_step": 2865, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04046042263507843, "train_loss_bc": 0.04046042263507843, "train_loss_llm": 0.0, "grad_norm": 0.01765494979918003, "global_step": 2866, "epoch": 79, "lr": 0.001} +{"train_loss": 0.044877927750349045, "train_loss_bc": 0.044877927750349045, "train_loss_llm": 0.0, "grad_norm": 0.025070222094655037, "global_step": 2867, "epoch": 79, "lr": 0.001} +{"train_loss": 0.049029815942049026, "train_loss_bc": 0.049029815942049026, "train_loss_llm": 0.0, "grad_norm": 0.03128732740879059, "global_step": 2868, "epoch": 79, "lr": 0.001} +{"train_loss": 0.047066476196050644, "train_loss_bc": 0.04704110696911812, "train_loss_llm": 0.002536773681640625, "grad_norm": 0.016484489664435387, "global_step": 2869, "epoch": 79, "lr": 0.001} +{"train_loss": 0.05082854628562927, "train_loss_bc": 0.05082854628562927, "train_loss_llm": 0.0, "grad_norm": 0.02293507568538189, "global_step": 2870, "epoch": 79, "lr": 0.001} +{"train_loss": 0.039013367146253586, "train_loss_bc": 0.038983967155218124, "train_loss_llm": 0.002939850091934204, "grad_norm": 0.025144577026367188, "global_step": 2871, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04144890233874321, "train_loss_bc": 0.04144890233874321, "train_loss_llm": 0.0, "grad_norm": 0.029589205980300903, "global_step": 2872, "epoch": 79, "lr": 0.001} +{"train_loss": 0.0400397889316082, "train_loss_bc": 0.0400397889316082, "train_loss_llm": 0.0, "grad_norm": 0.021613681688904762, "global_step": 2873, "epoch": 79, "lr": 0.001} +{"train_loss": 0.040372226387262344, "train_loss_bc": 0.040368035435676575, "train_loss_llm": 0.0004191398620605469, "grad_norm": 0.029362833127379417, "global_step": 2874, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04453003779053688, "train_loss_bc": 0.04453003779053688, "train_loss_llm": 0.0, "grad_norm": 0.03824489936232567, "global_step": 2875, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04413973540067673, "train_loss_bc": 0.044135287404060364, "train_loss_llm": 0.0004449188709259033, "grad_norm": 0.04049081355333328, "global_step": 2876, "epoch": 79, "lr": 0.001} +{"train_loss": 0.03986859321594238, "train_loss_bc": 0.03985266014933586, "train_loss_llm": 0.0015934407711029053, "grad_norm": 0.017127040773630142, "global_step": 2877, "epoch": 79, "lr": 0.001} +{"train_loss": 0.0464884415268898, "train_loss_bc": 0.04647008702158928, "train_loss_llm": 0.001835554838180542, "grad_norm": 0.02768201380968094, "global_step": 2878, "epoch": 79, "lr": 0.001} +{"train_loss": 0.04428629939340883, "train_loss_bc": 0.04711032658815384, "train_loss_llm": 0.001852661371231079, "grad_norm": 0.03024456277489662, "global_step": 2879, "epoch": 79, "lr": 0.001, "val_loss": 0.051261648535728455} +{"train_loss": 0.051305871456861496, "train_loss_bc": 0.05129726231098175, "train_loss_llm": 0.0008608996868133545, "grad_norm": 0.031033052131533623, "global_step": 2880, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04231036826968193, "train_loss_bc": 0.04227074980735779, "train_loss_llm": 0.00396198034286499, "grad_norm": 0.01527082547545433, "global_step": 2881, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04204583540558815, "train_loss_bc": 0.04204583540558815, "train_loss_llm": 0.0, "grad_norm": 0.022450456395745277, "global_step": 2882, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04884222894906998, "train_loss_bc": 0.04884222894906998, "train_loss_llm": 0.0, "grad_norm": 0.029664404690265656, "global_step": 2883, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04404548555612564, "train_loss_bc": 0.044042300432920456, "train_loss_llm": 0.0003185868263244629, "grad_norm": 0.04050128161907196, "global_step": 2884, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04167305678129196, "train_loss_bc": 0.041664496064186096, "train_loss_llm": 0.0008558928966522217, "grad_norm": 0.015211034566164017, "global_step": 2885, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04067032411694527, "train_loss_bc": 0.040651835501194, "train_loss_llm": 0.0018488168716430664, "grad_norm": 0.023448927327990532, "global_step": 2886, "epoch": 80, "lr": 0.001} +{"train_loss": 0.0446571409702301, "train_loss_bc": 0.0446571409702301, "train_loss_llm": 0.0, "grad_norm": 0.026321925222873688, "global_step": 2887, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04040144756436348, "train_loss_bc": 0.040381964296102524, "train_loss_llm": 0.0019482672214508057, "grad_norm": 0.037544865161180496, "global_step": 2888, "epoch": 80, "lr": 0.001} +{"train_loss": 0.0476604662835598, "train_loss_bc": 0.0476604662835598, "train_loss_llm": 0.0, "grad_norm": 0.01612587831914425, "global_step": 2889, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04447349160909653, "train_loss_bc": 0.04447349160909653, "train_loss_llm": 0.0, "grad_norm": 0.021808085963129997, "global_step": 2890, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04014484956860542, "train_loss_bc": 0.04014484956860542, "train_loss_llm": 0.0, "grad_norm": 0.02838823013007641, "global_step": 2891, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04589810594916344, "train_loss_bc": 0.04588085412979126, "train_loss_llm": 0.0017251372337341309, "grad_norm": 0.03341062739491463, "global_step": 2892, "epoch": 80, "lr": 0.001} +{"train_loss": 0.03923863545060158, "train_loss_bc": 0.03922831267118454, "train_loss_llm": 0.0010321438312530518, "grad_norm": 0.013161132112145424, "global_step": 2893, "epoch": 80, "lr": 0.001} +{"train_loss": 0.043974366039037704, "train_loss_bc": 0.043974366039037704, "train_loss_llm": 0.0, "grad_norm": 0.02117171697318554, "global_step": 2894, "epoch": 80, "lr": 0.001} +{"train_loss": 0.03914854675531387, "train_loss_bc": 0.03914854675531387, "train_loss_llm": 0.0, "grad_norm": 0.024208253249526024, "global_step": 2895, "epoch": 80, "lr": 0.001} +{"train_loss": 0.046782832592725754, "train_loss_bc": 0.046782832592725754, "train_loss_llm": 0.0, "grad_norm": 0.033613454550504684, "global_step": 2896, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04854618012905121, "train_loss_bc": 0.04851735755801201, "train_loss_llm": 0.002882152795791626, "grad_norm": 0.016696451231837273, "global_step": 2897, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04214593768119812, "train_loss_bc": 0.04214593768119812, "train_loss_llm": 0.0, "grad_norm": 0.021508295089006424, "global_step": 2898, "epoch": 80, "lr": 0.001} +{"train_loss": 0.042800817638635635, "train_loss_bc": 0.042800817638635635, "train_loss_llm": 0.0, "grad_norm": 0.025125235319137573, "global_step": 2899, "epoch": 80, "lr": 0.001} +{"train_loss": 0.045965444296598434, "train_loss_bc": 0.045965444296598434, "train_loss_llm": 0.0, "grad_norm": 0.03138145059347153, "global_step": 2900, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04829052463173866, "train_loss_bc": 0.04829052463173866, "train_loss_llm": 0.0, "grad_norm": 0.018759822472929955, "global_step": 2901, "epoch": 80, "lr": 0.001} +{"train_loss": 0.041945502161979675, "train_loss_bc": 0.041945502161979675, "train_loss_llm": 0.0, "grad_norm": 0.020554330199956894, "global_step": 2902, "epoch": 80, "lr": 0.001} +{"train_loss": 0.048074837774038315, "train_loss_bc": 0.04806877300143242, "train_loss_llm": 0.0006064176559448242, "grad_norm": 0.02611585706472397, "global_step": 2903, "epoch": 80, "lr": 0.001} +{"train_loss": 0.0405830480158329, "train_loss_bc": 0.04057556390762329, "train_loss_llm": 0.0007483959197998047, "grad_norm": 0.03399301692843437, "global_step": 2904, "epoch": 80, "lr": 0.001} +{"train_loss": 0.046351831406354904, "train_loss_bc": 0.046351831406354904, "train_loss_llm": 0.0, "grad_norm": 0.016493799164891243, "global_step": 2905, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04281684011220932, "train_loss_bc": 0.04281684011220932, "train_loss_llm": 0.0, "grad_norm": 0.02613510936498642, "global_step": 2906, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04577769711613655, "train_loss_bc": 0.04577769711613655, "train_loss_llm": 0.0, "grad_norm": 0.03477894887328148, "global_step": 2907, "epoch": 80, "lr": 0.001} +{"train_loss": 0.041722819209098816, "train_loss_bc": 0.041722819209098816, "train_loss_llm": 0.0, "grad_norm": 0.03895162045955658, "global_step": 2908, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04800742492079735, "train_loss_bc": 0.04800742492079735, "train_loss_llm": 0.0, "grad_norm": 0.016641443595290184, "global_step": 2909, "epoch": 80, "lr": 0.001} +{"train_loss": 0.039922453463077545, "train_loss_bc": 0.03991766646504402, "train_loss_llm": 0.0004786849021911621, "grad_norm": 0.022780826315283775, "global_step": 2910, "epoch": 80, "lr": 0.001} +{"train_loss": 0.041933584958314896, "train_loss_bc": 0.041933584958314896, "train_loss_llm": 0.0, "grad_norm": 0.02847127802670002, "global_step": 2911, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04902532324194908, "train_loss_bc": 0.04902532324194908, "train_loss_llm": 0.0, "grad_norm": 0.04188496991991997, "global_step": 2912, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04363405704498291, "train_loss_bc": 0.04363405704498291, "train_loss_llm": 0.0, "grad_norm": 0.01886347495019436, "global_step": 2913, "epoch": 80, "lr": 0.001} +{"train_loss": 0.048444584012031555, "train_loss_bc": 0.04841674864292145, "train_loss_llm": 0.0027833878993988037, "grad_norm": 0.02659761533141136, "global_step": 2914, "epoch": 80, "lr": 0.001} +{"train_loss": 0.04419068754133251, "train_loss_bc": 0.04160279035568237, "train_loss_llm": 0.0, "grad_norm": 0.03375176712870598, "global_step": 2915, "epoch": 80, "lr": 0.001, "train/cumulative_reward": 113.52550627112086, "train/mean_score": 16.829236455387917, "train/success_rate": 0.4, "test/cumulative_reward": 73.06897874733855, "test/mean_score": 10.882551824255621, "test/success_rate": 0.22, "val_loss": 0.05140473693609238, "train_action_mse_error": 0.04811590909957886} +{"train_loss": 0.042774345725774765, "train_loss_bc": 0.04277119040489197, "train_loss_llm": 0.0003155171871185303, "grad_norm": 0.0413215272128582, "global_step": 2916, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04654589667916298, "train_loss_bc": 0.04654589667916298, "train_loss_llm": 0.0, "grad_norm": 0.015991222113370895, "global_step": 2917, "epoch": 81, "lr": 0.001} +{"train_loss": 0.035956718027591705, "train_loss_bc": 0.035956718027591705, "train_loss_llm": 0.0, "grad_norm": 0.024802954867482185, "global_step": 2918, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04524044319987297, "train_loss_bc": 0.04521142691373825, "train_loss_llm": 0.0029015541076660156, "grad_norm": 0.029697027057409286, "global_step": 2919, "epoch": 81, "lr": 0.001} +{"train_loss": 0.039726268500089645, "train_loss_bc": 0.039726268500089645, "train_loss_llm": 0.0, "grad_norm": 0.03282442316412926, "global_step": 2920, "epoch": 81, "lr": 0.001} +{"train_loss": 0.040653906762599945, "train_loss_bc": 0.040653906762599945, "train_loss_llm": 0.0, "grad_norm": 0.016988690942525864, "global_step": 2921, "epoch": 81, "lr": 0.001} +{"train_loss": 0.05141989886760712, "train_loss_bc": 0.05139296501874924, "train_loss_llm": 0.0026934444904327393, "grad_norm": 0.01966380514204502, "global_step": 2922, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04156295210123062, "train_loss_bc": 0.041525907814502716, "train_loss_llm": 0.003704279661178589, "grad_norm": 0.02701956406235695, "global_step": 2923, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04148288443684578, "train_loss_bc": 0.04147114232182503, "train_loss_llm": 0.0011740922927856445, "grad_norm": 0.031364765018224716, "global_step": 2924, "epoch": 81, "lr": 0.001} +{"train_loss": 0.040908291935920715, "train_loss_bc": 0.040908291935920715, "train_loss_llm": 0.0, "grad_norm": 0.01596880331635475, "global_step": 2925, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04197937250137329, "train_loss_bc": 0.04196754842996597, "train_loss_llm": 0.0011823177337646484, "grad_norm": 0.022509420290589333, "global_step": 2926, "epoch": 81, "lr": 0.001} +{"train_loss": 0.045643020421266556, "train_loss_bc": 0.045643020421266556, "train_loss_llm": 0.0, "grad_norm": 0.03207788243889809, "global_step": 2927, "epoch": 81, "lr": 0.001} +{"train_loss": 0.042005810886621475, "train_loss_bc": 0.042005810886621475, "train_loss_llm": 0.0, "grad_norm": 0.030387459322810173, "global_step": 2928, "epoch": 81, "lr": 0.001} +{"train_loss": 0.047268521040678024, "train_loss_bc": 0.047268521040678024, "train_loss_llm": 0.0, "grad_norm": 0.017360297963023186, "global_step": 2929, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04038199782371521, "train_loss_bc": 0.04038199782371521, "train_loss_llm": 0.0, "grad_norm": 0.019710177555680275, "global_step": 2930, "epoch": 81, "lr": 0.001} +{"train_loss": 0.03611530736088753, "train_loss_bc": 0.03610444441437721, "train_loss_llm": 0.0010861456394195557, "grad_norm": 0.02873850427567959, "global_step": 2931, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04769258573651314, "train_loss_bc": 0.04769258573651314, "train_loss_llm": 0.0, "grad_norm": 0.03305879607796669, "global_step": 2932, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04493148997426033, "train_loss_bc": 0.04491780698299408, "train_loss_llm": 0.0013681650161743164, "grad_norm": 0.014111499302089214, "global_step": 2933, "epoch": 81, "lr": 0.001} +{"train_loss": 0.047208938747644424, "train_loss_bc": 0.047208938747644424, "train_loss_llm": 0.0, "grad_norm": 0.02239961363375187, "global_step": 2934, "epoch": 81, "lr": 0.001} +{"train_loss": 0.043780744075775146, "train_loss_bc": 0.043780744075775146, "train_loss_llm": 0.0, "grad_norm": 0.030350342392921448, "global_step": 2935, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04045694321393967, "train_loss_bc": 0.04044550284743309, "train_loss_llm": 0.0011439025402069092, "grad_norm": 0.0404142364859581, "global_step": 2936, "epoch": 81, "lr": 0.001} +{"train_loss": 0.043714605271816254, "train_loss_bc": 0.043714605271816254, "train_loss_llm": 0.0, "grad_norm": 0.01476973481476307, "global_step": 2937, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04580025002360344, "train_loss_bc": 0.04580025002360344, "train_loss_llm": 0.0, "grad_norm": 0.026912305504083633, "global_step": 2938, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04853786900639534, "train_loss_bc": 0.04853786900639534, "train_loss_llm": 0.0, "grad_norm": 0.03134138137102127, "global_step": 2939, "epoch": 81, "lr": 0.001} +{"train_loss": 0.041952986270189285, "train_loss_bc": 0.041952986270189285, "train_loss_llm": 0.0, "grad_norm": 0.039837174117565155, "global_step": 2940, "epoch": 81, "lr": 0.001} +{"train_loss": 0.046058475971221924, "train_loss_bc": 0.046058475971221924, "train_loss_llm": 0.0, "grad_norm": 0.01728517934679985, "global_step": 2941, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04636445268988609, "train_loss_bc": 0.04636445268988609, "train_loss_llm": 0.0, "grad_norm": 0.027971606701612473, "global_step": 2942, "epoch": 81, "lr": 0.001} +{"train_loss": 0.036156099289655685, "train_loss_bc": 0.036156099289655685, "train_loss_llm": 0.0, "grad_norm": 0.03707464784383774, "global_step": 2943, "epoch": 81, "lr": 0.001} +{"train_loss": 0.0429360531270504, "train_loss_bc": 0.042915478348731995, "train_loss_llm": 0.002057492733001709, "grad_norm": 0.0489969402551651, "global_step": 2944, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04770602658390999, "train_loss_bc": 0.04770602658390999, "train_loss_llm": 0.0, "grad_norm": 0.02453812025487423, "global_step": 2945, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04608302190899849, "train_loss_bc": 0.04608302190899849, "train_loss_llm": 0.0, "grad_norm": 0.041178833693265915, "global_step": 2946, "epoch": 81, "lr": 0.001} +{"train_loss": 0.05617593601346016, "train_loss_bc": 0.05617593601346016, "train_loss_llm": 0.0, "grad_norm": 0.04991702735424042, "global_step": 2947, "epoch": 81, "lr": 0.001} +{"train_loss": 0.049722760915756226, "train_loss_bc": 0.049722760915756226, "train_loss_llm": 0.0, "grad_norm": 0.05971205234527588, "global_step": 2948, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04450622946023941, "train_loss_bc": 0.04450622946023941, "train_loss_llm": 0.0, "grad_norm": 0.01404736191034317, "global_step": 2949, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04743747040629387, "train_loss_bc": 0.04742332920432091, "train_loss_llm": 0.0014140307903289795, "grad_norm": 0.01977439783513546, "global_step": 2950, "epoch": 81, "lr": 0.001} +{"train_loss": 0.04429072607308626, "train_loss_bc": 0.04757756367325783, "train_loss_llm": 0.0, "grad_norm": 0.030781235545873642, "global_step": 2951, "epoch": 81, "lr": 0.001, "val_loss": 0.05146773159503937} +{"train_loss": 0.04076742008328438, "train_loss_bc": 0.04076742008328438, "train_loss_llm": 0.0, "grad_norm": 0.0359371155500412, "global_step": 2952, "epoch": 82, "lr": 0.001} +{"train_loss": 0.055809419602155685, "train_loss_bc": 0.055809419602155685, "train_loss_llm": 0.0, "grad_norm": 0.014340470544993877, "global_step": 2953, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04811800643801689, "train_loss_bc": 0.04809897020459175, "train_loss_llm": 0.0019035041332244873, "grad_norm": 0.026600340381264687, "global_step": 2954, "epoch": 82, "lr": 0.001} +{"train_loss": 0.056171368807554245, "train_loss_bc": 0.056171368807554245, "train_loss_llm": 0.0, "grad_norm": 0.03395300358533859, "global_step": 2955, "epoch": 82, "lr": 0.001} +{"train_loss": 0.042876340448856354, "train_loss_bc": 0.042876340448856354, "train_loss_llm": 0.0, "grad_norm": 0.042080357670784, "global_step": 2956, "epoch": 82, "lr": 0.001} +{"train_loss": 0.048091962933540344, "train_loss_bc": 0.048091962933540344, "train_loss_llm": 0.0, "grad_norm": 0.023384563624858856, "global_step": 2957, "epoch": 82, "lr": 0.001} +{"train_loss": 0.03523746505379677, "train_loss_bc": 0.03523746505379677, "train_loss_llm": 0.0, "grad_norm": 0.03761795535683632, "global_step": 2958, "epoch": 82, "lr": 0.001} +{"train_loss": 0.050896741449832916, "train_loss_bc": 0.050896741449832916, "train_loss_llm": 0.0, "grad_norm": 0.05123346671462059, "global_step": 2959, "epoch": 82, "lr": 0.001} +{"train_loss": 0.046774838119745255, "train_loss_bc": 0.046772655099630356, "train_loss_llm": 0.00021842122077941895, "grad_norm": 0.059789374470710754, "global_step": 2960, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04528851807117462, "train_loss_bc": 0.04528851807117462, "train_loss_llm": 0.0, "grad_norm": 0.01486974861472845, "global_step": 2961, "epoch": 82, "lr": 0.001} +{"train_loss": 0.049088191241025925, "train_loss_bc": 0.049079086631536484, "train_loss_llm": 0.000910341739654541, "grad_norm": 0.03569437190890312, "global_step": 2962, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04333813115954399, "train_loss_bc": 0.043302271515131, "train_loss_llm": 0.0035857856273651123, "grad_norm": 0.046683430671691895, "global_step": 2963, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04473503679037094, "train_loss_bc": 0.04473503679037094, "train_loss_llm": 0.0, "grad_norm": 0.055138736963272095, "global_step": 2964, "epoch": 82, "lr": 0.001} +{"train_loss": 0.037358589470386505, "train_loss_bc": 0.037358589470386505, "train_loss_llm": 0.0, "grad_norm": 0.016668051481246948, "global_step": 2965, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04412304610013962, "train_loss_bc": 0.04411778599023819, "train_loss_llm": 0.0005259811878204346, "grad_norm": 0.02678106725215912, "global_step": 2966, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04222574084997177, "train_loss_bc": 0.042218178510665894, "train_loss_llm": 0.000756382942199707, "grad_norm": 0.033346183598041534, "global_step": 2967, "epoch": 82, "lr": 0.001} +{"train_loss": 0.03977746143937111, "train_loss_bc": 0.03976992890238762, "train_loss_llm": 0.0007532238960266113, "grad_norm": 0.04204863682389259, "global_step": 2968, "epoch": 82, "lr": 0.001} +{"train_loss": 0.046092454344034195, "train_loss_bc": 0.046092454344034195, "train_loss_llm": 0.0, "grad_norm": 0.017916066572070122, "global_step": 2969, "epoch": 82, "lr": 0.001} +{"train_loss": 0.044265225529670715, "train_loss_bc": 0.044265225529670715, "train_loss_llm": 0.0, "grad_norm": 0.025008639320731163, "global_step": 2970, "epoch": 82, "lr": 0.001} +{"train_loss": 0.0413028709590435, "train_loss_bc": 0.04128756374120712, "train_loss_llm": 0.0015305578708648682, "grad_norm": 0.031026119366288185, "global_step": 2971, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04405692592263222, "train_loss_bc": 0.04405692592263222, "train_loss_llm": 0.0, "grad_norm": 0.03560434281826019, "global_step": 2972, "epoch": 82, "lr": 0.001} +{"train_loss": 0.03976660594344139, "train_loss_bc": 0.0397246778011322, "train_loss_llm": 0.004192769527435303, "grad_norm": 0.017389167100191116, "global_step": 2973, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04444534704089165, "train_loss_bc": 0.04444534704089165, "train_loss_llm": 0.0, "grad_norm": 0.02789471670985222, "global_step": 2974, "epoch": 82, "lr": 0.001} +{"train_loss": 0.047907933592796326, "train_loss_bc": 0.04788558557629585, "train_loss_llm": 0.002234727144241333, "grad_norm": 0.03234708309173584, "global_step": 2975, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04973095655441284, "train_loss_bc": 0.04973095655441284, "train_loss_llm": 0.0, "grad_norm": 0.0437137633562088, "global_step": 2976, "epoch": 82, "lr": 0.001} +{"train_loss": 0.044925060123205185, "train_loss_bc": 0.044925060123205185, "train_loss_llm": 0.0, "grad_norm": 0.024454649537801743, "global_step": 2977, "epoch": 82, "lr": 0.001} +{"train_loss": 0.03967728093266487, "train_loss_bc": 0.03967184200882912, "train_loss_llm": 0.000543743371963501, "grad_norm": 0.04306589439511299, "global_step": 2978, "epoch": 82, "lr": 0.001} +{"train_loss": 0.041319359093904495, "train_loss_bc": 0.041319359093904495, "train_loss_llm": 0.0, "grad_norm": 0.04859476909041405, "global_step": 2979, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04672710970044136, "train_loss_bc": 0.04672710970044136, "train_loss_llm": 0.0, "grad_norm": 0.06644129008054733, "global_step": 2980, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04045573249459267, "train_loss_bc": 0.04045573249459267, "train_loss_llm": 0.0, "grad_norm": 0.023540589958429337, "global_step": 2981, "epoch": 82, "lr": 0.001} +{"train_loss": 0.043954066932201385, "train_loss_bc": 0.043954066932201385, "train_loss_llm": 0.0, "grad_norm": 0.03816019743680954, "global_step": 2982, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04438423365354538, "train_loss_bc": 0.04438423365354538, "train_loss_llm": 0.0, "grad_norm": 0.048883721232414246, "global_step": 2983, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04641890898346901, "train_loss_bc": 0.04641890898346901, "train_loss_llm": 0.0, "grad_norm": 0.06767250597476959, "global_step": 2984, "epoch": 82, "lr": 0.001} +{"train_loss": 0.040226422250270844, "train_loss_bc": 0.04022403806447983, "train_loss_llm": 0.00023853778839111328, "grad_norm": 0.014775276184082031, "global_step": 2985, "epoch": 82, "lr": 0.001} +{"train_loss": 0.041838280856609344, "train_loss_bc": 0.041823312640190125, "train_loss_llm": 0.0014968812465667725, "grad_norm": 0.026273615658283234, "global_step": 2986, "epoch": 82, "lr": 0.001} +{"train_loss": 0.04450354269809193, "train_loss_bc": 0.04394983500242233, "train_loss_llm": 0.00046509504318237305, "grad_norm": 0.03962681442499161, "global_step": 2987, "epoch": 82, "lr": 0.001, "val_loss": 0.05120836943387985} +{"train_loss": 0.045120298862457275, "train_loss_bc": 0.045120298862457275, "train_loss_llm": 0.0, "grad_norm": 0.0550791472196579, "global_step": 2988, "epoch": 83, "lr": 0.001} +{"train_loss": 0.0415402390062809, "train_loss_bc": 0.0415402390062809, "train_loss_llm": 0.0, "grad_norm": 0.012918436899781227, "global_step": 2989, "epoch": 83, "lr": 0.001} +{"train_loss": 0.038058679550886154, "train_loss_bc": 0.038058679550886154, "train_loss_llm": 0.0, "grad_norm": 0.01663358323276043, "global_step": 2990, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04140887409448624, "train_loss_bc": 0.04138368368148804, "train_loss_llm": 0.0025192201137542725, "grad_norm": 0.02448853850364685, "global_step": 2991, "epoch": 83, "lr": 0.001} +{"train_loss": 0.03990896791219711, "train_loss_bc": 0.03990896791219711, "train_loss_llm": 0.0, "grad_norm": 0.028098467737436295, "global_step": 2992, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04464537650346756, "train_loss_bc": 0.04462765157222748, "train_loss_llm": 0.0017723441123962402, "grad_norm": 0.018905945122241974, "global_step": 2993, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04227883368730545, "train_loss_bc": 0.04227188602089882, "train_loss_llm": 0.0006946027278900146, "grad_norm": 0.033477459102869034, "global_step": 2994, "epoch": 83, "lr": 0.001} +{"train_loss": 0.044846177101135254, "train_loss_bc": 0.044846177101135254, "train_loss_llm": 0.0, "grad_norm": 0.034727793186903, "global_step": 2995, "epoch": 83, "lr": 0.001} +{"train_loss": 0.048333410173654556, "train_loss_bc": 0.048333410173654556, "train_loss_llm": 0.0, "grad_norm": 0.044240932911634445, "global_step": 2996, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04025135561823845, "train_loss_bc": 0.040236491709947586, "train_loss_llm": 0.0014862418174743652, "grad_norm": 0.015357636846601963, "global_step": 2997, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04359579458832741, "train_loss_bc": 0.043560631573200226, "train_loss_llm": 0.0035164058208465576, "grad_norm": 0.02580878511071205, "global_step": 2998, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04692525416612625, "train_loss_bc": 0.04692525416612625, "train_loss_llm": 0.0, "grad_norm": 0.03596716374158859, "global_step": 2999, "epoch": 83, "lr": 0.001} +{"train_loss": 0.041314128786325455, "train_loss_bc": 0.04128814861178398, "train_loss_llm": 0.0025978386402130127, "grad_norm": 0.04398244619369507, "global_step": 3000, "epoch": 83, "lr": 0.001} +{"train_loss": 0.047902099788188934, "train_loss_bc": 0.04788565635681152, "train_loss_llm": 0.0016444623470306396, "grad_norm": 0.017336543649435043, "global_step": 3001, "epoch": 83, "lr": 0.001} +{"train_loss": 0.0548197440803051, "train_loss_bc": 0.0548197440803051, "train_loss_llm": 0.0, "grad_norm": 0.029354307800531387, "global_step": 3002, "epoch": 83, "lr": 0.001} +{"train_loss": 0.048954397439956665, "train_loss_bc": 0.04895029962062836, "train_loss_llm": 0.00040990114212036133, "grad_norm": 0.03886963054537773, "global_step": 3003, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04197583347558975, "train_loss_bc": 0.04194629564881325, "train_loss_llm": 0.0029537975788116455, "grad_norm": 0.044187769293785095, "global_step": 3004, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04475041478872299, "train_loss_bc": 0.04475041478872299, "train_loss_llm": 0.0, "grad_norm": 0.021741321310400963, "global_step": 3005, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04775445535778999, "train_loss_bc": 0.047746311873197556, "train_loss_llm": 0.0008144676685333252, "grad_norm": 0.0308124627918005, "global_step": 3006, "epoch": 83, "lr": 0.001} +{"train_loss": 0.045772213488817215, "train_loss_bc": 0.04575604945421219, "train_loss_llm": 0.0016162395477294922, "grad_norm": 0.04318026825785637, "global_step": 3007, "epoch": 83, "lr": 0.001} +{"train_loss": 0.048877131193876266, "train_loss_bc": 0.04886970669031143, "train_loss_llm": 0.0007423162460327148, "grad_norm": 0.050484925508499146, "global_step": 3008, "epoch": 83, "lr": 0.001} +{"train_loss": 0.0358663834631443, "train_loss_bc": 0.0358663834631443, "train_loss_llm": 0.0, "grad_norm": 0.01699690893292427, "global_step": 3009, "epoch": 83, "lr": 0.001} +{"train_loss": 0.047210633754730225, "train_loss_bc": 0.047210633754730225, "train_loss_llm": 0.0, "grad_norm": 0.030006781220436096, "global_step": 3010, "epoch": 83, "lr": 0.001} +{"train_loss": 0.039842139929533005, "train_loss_bc": 0.039842139929533005, "train_loss_llm": 0.0, "grad_norm": 0.03741362318396568, "global_step": 3011, "epoch": 83, "lr": 0.001} +{"train_loss": 0.050480134785175323, "train_loss_bc": 0.05045854672789574, "train_loss_llm": 0.00215873122215271, "grad_norm": 0.0458134226500988, "global_step": 3012, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04393956437706947, "train_loss_bc": 0.04393956437706947, "train_loss_llm": 0.0, "grad_norm": 0.01611081324517727, "global_step": 3013, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04354554042220116, "train_loss_bc": 0.04351848363876343, "train_loss_llm": 0.0027056336402893066, "grad_norm": 0.02413179539144039, "global_step": 3014, "epoch": 83, "lr": 0.001} +{"train_loss": 0.040447864681482315, "train_loss_bc": 0.040447864681482315, "train_loss_llm": 0.0, "grad_norm": 0.034104686230421066, "global_step": 3015, "epoch": 83, "lr": 0.001} +{"train_loss": 0.05185993015766144, "train_loss_bc": 0.051846206188201904, "train_loss_llm": 0.0013723671436309814, "grad_norm": 0.04630211740732193, "global_step": 3016, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04446817561984062, "train_loss_bc": 0.04446817561984062, "train_loss_llm": 0.0, "grad_norm": 0.020079778507351875, "global_step": 3017, "epoch": 83, "lr": 0.001} +{"train_loss": 0.03593101724982262, "train_loss_bc": 0.035913337022066116, "train_loss_llm": 0.0017681717872619629, "grad_norm": 0.026847142726182938, "global_step": 3018, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04389658570289612, "train_loss_bc": 0.04388118162751198, "train_loss_llm": 0.001540541648864746, "grad_norm": 0.04099256545305252, "global_step": 3019, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04527568444609642, "train_loss_bc": 0.04524582624435425, "train_loss_llm": 0.002985715866088867, "grad_norm": 0.04805842787027359, "global_step": 3020, "epoch": 83, "lr": 0.001} +{"train_loss": 0.041829511523246765, "train_loss_bc": 0.04180107265710831, "train_loss_llm": 0.00284382700920105, "grad_norm": 0.016715947538614273, "global_step": 3021, "epoch": 83, "lr": 0.001} +{"train_loss": 0.04860693961381912, "train_loss_bc": 0.04860693961381912, "train_loss_llm": 0.0, "grad_norm": 0.018539683893322945, "global_step": 3022, "epoch": 83, "lr": 0.001} +{"train_loss": 0.044259787847598396, "train_loss_bc": 0.04111857712268829, "train_loss_llm": 0.0, "grad_norm": 0.02893880568444729, "global_step": 3023, "epoch": 83, "lr": 0.001, "val_loss": 0.051583677530288696} +{"train_loss": 0.044496625661849976, "train_loss_bc": 0.044496625661849976, "train_loss_llm": 0.0, "grad_norm": 0.03595450893044472, "global_step": 3024, "epoch": 84, "lr": 0.001} +{"train_loss": 0.046632688492536545, "train_loss_bc": 0.046632688492536545, "train_loss_llm": 0.0, "grad_norm": 0.020328085869550705, "global_step": 3025, "epoch": 84, "lr": 0.001} +{"train_loss": 0.03810528293251991, "train_loss_bc": 0.03810528293251991, "train_loss_llm": 0.0, "grad_norm": 0.034772831946611404, "global_step": 3026, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04237736761569977, "train_loss_bc": 0.04237736761569977, "train_loss_llm": 0.0, "grad_norm": 0.045041222125291824, "global_step": 3027, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04022514075040817, "train_loss_bc": 0.04022514075040817, "train_loss_llm": 0.0, "grad_norm": 0.04535980522632599, "global_step": 3028, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04604191333055496, "train_loss_bc": 0.04604191333055496, "train_loss_llm": 0.0, "grad_norm": 0.01722869835793972, "global_step": 3029, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04746272414922714, "train_loss_bc": 0.047452062368392944, "train_loss_llm": 0.0010663270950317383, "grad_norm": 0.03097933530807495, "global_step": 3030, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04390351101756096, "train_loss_bc": 0.04390351101756096, "train_loss_llm": 0.0, "grad_norm": 0.04475235939025879, "global_step": 3031, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04409821704030037, "train_loss_bc": 0.04409821704030037, "train_loss_llm": 0.0, "grad_norm": 0.05075099319219589, "global_step": 3032, "epoch": 84, "lr": 0.001} +{"train_loss": 0.05305337533354759, "train_loss_bc": 0.05302874743938446, "train_loss_llm": 0.0024628937244415283, "grad_norm": 0.01606871746480465, "global_step": 3033, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04316967725753784, "train_loss_bc": 0.04316489398479462, "train_loss_llm": 0.00047838687896728516, "grad_norm": 0.023949110880494118, "global_step": 3034, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04412924498319626, "train_loss_bc": 0.04412463679909706, "train_loss_llm": 0.0004608333110809326, "grad_norm": 0.033762283623218536, "global_step": 3035, "epoch": 84, "lr": 0.001} +{"train_loss": 0.044310312718153, "train_loss_bc": 0.04430276155471802, "train_loss_llm": 0.000755012035369873, "grad_norm": 0.03874346613883972, "global_step": 3036, "epoch": 84, "lr": 0.001} +{"train_loss": 0.046337779611349106, "train_loss_bc": 0.046337779611349106, "train_loss_llm": 0.0, "grad_norm": 0.020155951380729675, "global_step": 3037, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04341314733028412, "train_loss_bc": 0.04340886324644089, "train_loss_llm": 0.0004284381866455078, "grad_norm": 0.03471503034234047, "global_step": 3038, "epoch": 84, "lr": 0.001} +{"train_loss": 0.046426087617874146, "train_loss_bc": 0.046426087617874146, "train_loss_llm": 0.0, "grad_norm": 0.046111736446619034, "global_step": 3039, "epoch": 84, "lr": 0.001} +{"train_loss": 0.0469241626560688, "train_loss_bc": 0.0469241626560688, "train_loss_llm": 0.0, "grad_norm": 0.05405931919813156, "global_step": 3040, "epoch": 84, "lr": 0.001} +{"train_loss": 0.048241741955280304, "train_loss_bc": 0.048241741955280304, "train_loss_llm": 0.0, "grad_norm": 0.017120910808444023, "global_step": 3041, "epoch": 84, "lr": 0.001} +{"train_loss": 0.03974476829171181, "train_loss_bc": 0.03974476829171181, "train_loss_llm": 0.0, "grad_norm": 0.023121554404497147, "global_step": 3042, "epoch": 84, "lr": 0.001} +{"train_loss": 0.042885493487119675, "train_loss_bc": 0.042885493487119675, "train_loss_llm": 0.0, "grad_norm": 0.026267822831869125, "global_step": 3043, "epoch": 84, "lr": 0.001} +{"train_loss": 0.03329874947667122, "train_loss_bc": 0.033295877277851105, "train_loss_llm": 0.00028714537620544434, "grad_norm": 0.031683143228292465, "global_step": 3044, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04145915433764458, "train_loss_bc": 0.04145915433764458, "train_loss_llm": 0.0, "grad_norm": 0.020568037405610085, "global_step": 3045, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04511117935180664, "train_loss_bc": 0.04510840028524399, "train_loss_llm": 0.0002778768539428711, "grad_norm": 0.03303340822458267, "global_step": 3046, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04384400323033333, "train_loss_bc": 0.04384400323033333, "train_loss_llm": 0.0, "grad_norm": 0.045544832944869995, "global_step": 3047, "epoch": 84, "lr": 0.001} +{"train_loss": 0.0466015450656414, "train_loss_bc": 0.0466015450656414, "train_loss_llm": 0.0, "grad_norm": 0.05752550810575485, "global_step": 3048, "epoch": 84, "lr": 0.001} +{"train_loss": 0.05090530961751938, "train_loss_bc": 0.050874173641204834, "train_loss_llm": 0.003113776445388794, "grad_norm": 0.018175577744841576, "global_step": 3049, "epoch": 84, "lr": 0.001} +{"train_loss": 0.0396478995680809, "train_loss_bc": 0.0396478995680809, "train_loss_llm": 0.0, "grad_norm": 0.028673648834228516, "global_step": 3050, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04744340106844902, "train_loss_bc": 0.04742279276251793, "train_loss_llm": 0.0020606517791748047, "grad_norm": 0.04098597168922424, "global_step": 3051, "epoch": 84, "lr": 0.001} +{"train_loss": 0.045564472675323486, "train_loss_bc": 0.045564472675323486, "train_loss_llm": 0.0, "grad_norm": 0.05740964785218239, "global_step": 3052, "epoch": 84, "lr": 0.001} +{"train_loss": 0.03785610571503639, "train_loss_bc": 0.037846408784389496, "train_loss_llm": 0.0009695589542388916, "grad_norm": 0.013790817931294441, "global_step": 3053, "epoch": 84, "lr": 0.001} +{"train_loss": 0.04770493879914284, "train_loss_bc": 0.04770493879914284, "train_loss_llm": 0.0, "grad_norm": 0.026234854012727737, "global_step": 3054, "epoch": 84, "lr": 0.001} +{"train_loss": 0.0411880686879158, "train_loss_bc": 0.0411880686879158, "train_loss_llm": 0.0, "grad_norm": 0.031792838126420975, "global_step": 3055, "epoch": 84, "lr": 0.001} +{"train_loss": 0.0464162603020668, "train_loss_bc": 0.0464162603020668, "train_loss_llm": 0.0, "grad_norm": 0.04112021252512932, "global_step": 3056, "epoch": 84, "lr": 0.001} +{"train_loss": 0.05095132067799568, "train_loss_bc": 0.050934676080942154, "train_loss_llm": 0.0016643106937408447, "grad_norm": 0.01596103608608246, "global_step": 3057, "epoch": 84, "lr": 0.001} +{"train_loss": 0.045821867883205414, "train_loss_bc": 0.045821867883205414, "train_loss_llm": 0.0, "grad_norm": 0.028643688187003136, "global_step": 3058, "epoch": 84, "lr": 0.001} +{"train_loss": 0.044284718214637704, "train_loss_bc": 0.038449496030807495, "train_loss_llm": 0.000682145357131958, "grad_norm": 0.04091833531856537, "global_step": 3059, "epoch": 84, "lr": 0.001, "val_loss": 0.051961153745651245} +{"train_loss": 0.05127841979265213, "train_loss_bc": 0.05127841979265213, "train_loss_llm": 0.0, "grad_norm": 0.047317784279584885, "global_step": 3060, "epoch": 85, "lr": 0.001} +{"train_loss": 0.05095984786748886, "train_loss_bc": 0.05093277990818024, "train_loss_llm": 0.002706676721572876, "grad_norm": 0.016161464154720306, "global_step": 3061, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04532486945390701, "train_loss_bc": 0.04532486945390701, "train_loss_llm": 0.0, "grad_norm": 0.02852131612598896, "global_step": 3062, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04089860990643501, "train_loss_bc": 0.04089860990643501, "train_loss_llm": 0.0, "grad_norm": 0.032010938972234726, "global_step": 3063, "epoch": 85, "lr": 0.001} +{"train_loss": 0.026734448969364166, "train_loss_bc": 0.026734448969364166, "train_loss_llm": 0.0, "grad_norm": 0.03693249821662903, "global_step": 3064, "epoch": 85, "lr": 0.001} +{"train_loss": 0.05137711018323898, "train_loss_bc": 0.05137711018323898, "train_loss_llm": 0.0, "grad_norm": 0.02409113384783268, "global_step": 3065, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04754634201526642, "train_loss_bc": 0.04754099249839783, "train_loss_llm": 0.0005348324775695801, "grad_norm": 0.03465010225772858, "global_step": 3066, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04006428271532059, "train_loss_bc": 0.04000965505838394, "train_loss_llm": 0.0054627954959869385, "grad_norm": 0.04057580605149269, "global_step": 3067, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04689362272620201, "train_loss_bc": 0.04689362272620201, "train_loss_llm": 0.0, "grad_norm": 0.05315140262246132, "global_step": 3068, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04343869909644127, "train_loss_bc": 0.04343869909644127, "train_loss_llm": 0.0, "grad_norm": 0.02085547149181366, "global_step": 3069, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04221782088279724, "train_loss_bc": 0.04221782088279724, "train_loss_llm": 0.0, "grad_norm": 0.028771447017788887, "global_step": 3070, "epoch": 85, "lr": 0.001} +{"train_loss": 0.05043397098779678, "train_loss_bc": 0.05041790381073952, "train_loss_llm": 0.0016068518161773682, "grad_norm": 0.037985190749168396, "global_step": 3071, "epoch": 85, "lr": 0.001} +{"train_loss": 0.037956636399030685, "train_loss_bc": 0.037956636399030685, "train_loss_llm": 0.0, "grad_norm": 0.0472811758518219, "global_step": 3072, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04370349273085594, "train_loss_bc": 0.04369843378663063, "train_loss_llm": 0.0005058944225311279, "grad_norm": 0.01583392545580864, "global_step": 3073, "epoch": 85, "lr": 0.001} +{"train_loss": 0.047802895307540894, "train_loss_bc": 0.047802895307540894, "train_loss_llm": 0.0, "grad_norm": 0.024256709963083267, "global_step": 3074, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04364544525742531, "train_loss_bc": 0.0436265729367733, "train_loss_llm": 0.0018871426582336426, "grad_norm": 0.03547317534685135, "global_step": 3075, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04312599077820778, "train_loss_bc": 0.04312599077820778, "train_loss_llm": 0.0, "grad_norm": 0.03758879005908966, "global_step": 3076, "epoch": 85, "lr": 0.001} +{"train_loss": 0.03849658742547035, "train_loss_bc": 0.03845227137207985, "train_loss_llm": 0.004431724548339844, "grad_norm": 0.015659037977457047, "global_step": 3077, "epoch": 85, "lr": 0.001} +{"train_loss": 0.043231941759586334, "train_loss_bc": 0.043231941759586334, "train_loss_llm": 0.0, "grad_norm": 0.024060118943452835, "global_step": 3078, "epoch": 85, "lr": 0.001} +{"train_loss": 0.05186709016561508, "train_loss_bc": 0.05186709016561508, "train_loss_llm": 0.0, "grad_norm": 0.030497586354613304, "global_step": 3079, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04517153277993202, "train_loss_bc": 0.04517153277993202, "train_loss_llm": 0.0, "grad_norm": 0.03584090992808342, "global_step": 3080, "epoch": 85, "lr": 0.001} +{"train_loss": 0.0421939343214035, "train_loss_bc": 0.042186688631772995, "train_loss_llm": 0.0007244646549224854, "grad_norm": 0.016960613429546356, "global_step": 3081, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04436907544732094, "train_loss_bc": 0.04436907544732094, "train_loss_llm": 0.0, "grad_norm": 0.029126325622200966, "global_step": 3082, "epoch": 85, "lr": 0.001} +{"train_loss": 0.046738095581531525, "train_loss_bc": 0.046738095581531525, "train_loss_llm": 0.0, "grad_norm": 0.03733077645301819, "global_step": 3083, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04388824850320816, "train_loss_bc": 0.04388824850320816, "train_loss_llm": 0.0, "grad_norm": 0.04541308805346489, "global_step": 3084, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04661724343895912, "train_loss_bc": 0.04661724343895912, "train_loss_llm": 0.0, "grad_norm": 0.015583977103233337, "global_step": 3085, "epoch": 85, "lr": 0.001} +{"train_loss": 0.042937636375427246, "train_loss_bc": 0.04293468967080116, "train_loss_llm": 0.00029477477073669434, "grad_norm": 0.02247512899339199, "global_step": 3086, "epoch": 85, "lr": 0.001} +{"train_loss": 0.044125307351350784, "train_loss_bc": 0.044125307351350784, "train_loss_llm": 0.0, "grad_norm": 0.03271356225013733, "global_step": 3087, "epoch": 85, "lr": 0.001} +{"train_loss": 0.040616221725940704, "train_loss_bc": 0.040598854422569275, "train_loss_llm": 0.0017368495464324951, "grad_norm": 0.04413486272096634, "global_step": 3088, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04575568437576294, "train_loss_bc": 0.04575568437576294, "train_loss_llm": 0.0, "grad_norm": 0.019834639504551888, "global_step": 3089, "epoch": 85, "lr": 0.001} +{"train_loss": 0.0432603694498539, "train_loss_bc": 0.043248601257801056, "train_loss_llm": 0.0011767446994781494, "grad_norm": 0.03619896620512009, "global_step": 3090, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04436970502138138, "train_loss_bc": 0.044333942234516144, "train_loss_llm": 0.003576129674911499, "grad_norm": 0.052793532609939575, "global_step": 3091, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04545911028981209, "train_loss_bc": 0.04545911028981209, "train_loss_llm": 0.0, "grad_norm": 0.06049240007996559, "global_step": 3092, "epoch": 85, "lr": 0.001} +{"train_loss": 0.03405366837978363, "train_loss_bc": 0.03405366837978363, "train_loss_llm": 0.0, "grad_norm": 0.013997387140989304, "global_step": 3093, "epoch": 85, "lr": 0.001} +{"train_loss": 0.047204770147800446, "train_loss_bc": 0.047187041491270065, "train_loss_llm": 0.0017730295658111572, "grad_norm": 0.022507939487695694, "global_step": 3094, "epoch": 85, "lr": 0.001} +{"train_loss": 0.04437116285165151, "train_loss_bc": 0.05360313504934311, "train_loss_llm": 0.0, "grad_norm": 0.04032173380255699, "global_step": 3095, "epoch": 85, "lr": 0.001, "train/cumulative_reward": 36.024915124381494, "train/mean_score": 8.532814529626592, "train/success_rate": 0.2, "test/cumulative_reward": 49.40442293237578, "test/mean_score": 5.879751103105491, "test/success_rate": 0.1, "val_loss": 0.051581837236881256, "train_action_mse_error": 0.04857942461967468} +{"train_loss": 0.03856071084737778, "train_loss_bc": 0.03856071084737778, "train_loss_llm": 0.0, "grad_norm": 0.05305011197924614, "global_step": 3096, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04648265987634659, "train_loss_bc": 0.04648265987634659, "train_loss_llm": 0.0, "grad_norm": 0.0172567181289196, "global_step": 3097, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04438840597867966, "train_loss_bc": 0.04438364878296852, "train_loss_llm": 0.00047576427459716797, "grad_norm": 0.02915823832154274, "global_step": 3098, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04655947536230087, "train_loss_bc": 0.04655947536230087, "train_loss_llm": 0.0, "grad_norm": 0.03783353418111801, "global_step": 3099, "epoch": 86, "lr": 0.001} +{"train_loss": 0.03980497643351555, "train_loss_bc": 0.03980497643351555, "train_loss_llm": 0.0, "grad_norm": 0.04458516836166382, "global_step": 3100, "epoch": 86, "lr": 0.001} +{"train_loss": 0.045408762991428375, "train_loss_bc": 0.045376524329185486, "train_loss_llm": 0.0032238364219665527, "grad_norm": 0.01596486195921898, "global_step": 3101, "epoch": 86, "lr": 0.001} +{"train_loss": 0.044189248234033585, "train_loss_bc": 0.044189248234033585, "train_loss_llm": 0.0, "grad_norm": 0.02645810879766941, "global_step": 3102, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04422200843691826, "train_loss_bc": 0.04419354721903801, "train_loss_llm": 0.002846240997314453, "grad_norm": 0.032217834144830704, "global_step": 3103, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04500873014330864, "train_loss_bc": 0.04498416185379028, "train_loss_llm": 0.002456963062286377, "grad_norm": 0.0422666072845459, "global_step": 3104, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04070578143000603, "train_loss_bc": 0.04069562256336212, "train_loss_llm": 0.0010158121585845947, "grad_norm": 0.01404505129903555, "global_step": 3105, "epoch": 86, "lr": 0.001} +{"train_loss": 0.03899778053164482, "train_loss_bc": 0.03899778053164482, "train_loss_llm": 0.0, "grad_norm": 0.01873946562409401, "global_step": 3106, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04787636920809746, "train_loss_bc": 0.04787636920809746, "train_loss_llm": 0.0, "grad_norm": 0.02590162307024002, "global_step": 3107, "epoch": 86, "lr": 0.001} +{"train_loss": 0.050297707319259644, "train_loss_bc": 0.050297707319259644, "train_loss_llm": 0.0, "grad_norm": 0.03518914803862572, "global_step": 3108, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04234180599451065, "train_loss_bc": 0.04234180599451065, "train_loss_llm": 0.0, "grad_norm": 0.016690919175744057, "global_step": 3109, "epoch": 86, "lr": 0.001} +{"train_loss": 0.043572939932346344, "train_loss_bc": 0.043572939932346344, "train_loss_llm": 0.0, "grad_norm": 0.026836620643734932, "global_step": 3110, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04394979029893875, "train_loss_bc": 0.04393157735466957, "train_loss_llm": 0.0018212497234344482, "grad_norm": 0.034620415419340134, "global_step": 3111, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04376275837421417, "train_loss_bc": 0.043759625405073166, "train_loss_llm": 0.0003133416175842285, "grad_norm": 0.04326538369059563, "global_step": 3112, "epoch": 86, "lr": 0.001} +{"train_loss": 0.03775575011968613, "train_loss_bc": 0.03775575011968613, "train_loss_llm": 0.0, "grad_norm": 0.016392791643738747, "global_step": 3113, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04843243956565857, "train_loss_bc": 0.04843243956565857, "train_loss_llm": 0.0, "grad_norm": 0.029168160632252693, "global_step": 3114, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04053004831075668, "train_loss_bc": 0.04053004831075668, "train_loss_llm": 0.0, "grad_norm": 0.03266545757651329, "global_step": 3115, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04240945726633072, "train_loss_bc": 0.04240945726633072, "train_loss_llm": 0.0, "grad_norm": 0.04573746398091316, "global_step": 3116, "epoch": 86, "lr": 0.001} +{"train_loss": 0.041138678789138794, "train_loss_bc": 0.04109631106257439, "train_loss_llm": 0.004236936569213867, "grad_norm": 0.016211841255426407, "global_step": 3117, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04865482822060585, "train_loss_bc": 0.04863269627094269, "train_loss_llm": 0.0022130608558654785, "grad_norm": 0.023813096806406975, "global_step": 3118, "epoch": 86, "lr": 0.001} +{"train_loss": 0.046379461884498596, "train_loss_bc": 0.046372849494218826, "train_loss_llm": 0.0006613731384277344, "grad_norm": 0.033329881727695465, "global_step": 3119, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04130803048610687, "train_loss_bc": 0.04130803048610687, "train_loss_llm": 0.0, "grad_norm": 0.04262867569923401, "global_step": 3120, "epoch": 86, "lr": 0.001} +{"train_loss": 0.03732111677527428, "train_loss_bc": 0.03732111677527428, "train_loss_llm": 0.0, "grad_norm": 0.018269838765263557, "global_step": 3121, "epoch": 86, "lr": 0.001} +{"train_loss": 0.0474449023604393, "train_loss_bc": 0.0474449023604393, "train_loss_llm": 0.0, "grad_norm": 0.02807873860001564, "global_step": 3122, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04286152869462967, "train_loss_bc": 0.04286152869462967, "train_loss_llm": 0.0, "grad_norm": 0.03582237660884857, "global_step": 3123, "epoch": 86, "lr": 0.001} +{"train_loss": 0.049060191959142685, "train_loss_bc": 0.049037277698516846, "train_loss_llm": 0.0022913217544555664, "grad_norm": 0.04319528117775917, "global_step": 3124, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04748689383268356, "train_loss_bc": 0.04747108370065689, "train_loss_llm": 0.0015808641910552979, "grad_norm": 0.02244466170668602, "global_step": 3125, "epoch": 86, "lr": 0.001} +{"train_loss": 0.05029573664069176, "train_loss_bc": 0.05029573664069176, "train_loss_llm": 0.0, "grad_norm": 0.029468879103660583, "global_step": 3126, "epoch": 86, "lr": 0.001} +{"train_loss": 0.05058126896619797, "train_loss_bc": 0.05058126896619797, "train_loss_llm": 0.0, "grad_norm": 0.04037652537226677, "global_step": 3127, "epoch": 86, "lr": 0.001} +{"train_loss": 0.0344269722700119, "train_loss_bc": 0.0344269722700119, "train_loss_llm": 0.0, "grad_norm": 0.04707759618759155, "global_step": 3128, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04210183769464493, "train_loss_bc": 0.04210183769464493, "train_loss_llm": 0.0, "grad_norm": 0.013498244807124138, "global_step": 3129, "epoch": 86, "lr": 0.001} +{"train_loss": 0.045196425169706345, "train_loss_bc": 0.045196425169706345, "train_loss_llm": 0.0, "grad_norm": 0.021744363009929657, "global_step": 3130, "epoch": 86, "lr": 0.001} +{"train_loss": 0.04418983620901903, "train_loss_bc": 0.05131862312555313, "train_loss_llm": 0.0, "grad_norm": 0.027652563527226448, "global_step": 3131, "epoch": 86, "lr": 0.001, "val_loss": 0.05142633989453316} +{"train_loss": 0.04235342517495155, "train_loss_bc": 0.04235270246863365, "train_loss_llm": 7.233023643493652e-05, "grad_norm": 0.030322924256324768, "global_step": 3132, "epoch": 87, "lr": 0.001} +{"train_loss": 0.047141462564468384, "train_loss_bc": 0.04713854193687439, "train_loss_llm": 0.00029224157333374023, "grad_norm": 0.017807932570576668, "global_step": 3133, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04590241238474846, "train_loss_bc": 0.04590241238474846, "train_loss_llm": 0.0, "grad_norm": 0.022219769656658173, "global_step": 3134, "epoch": 87, "lr": 0.001} +{"train_loss": 0.032362330704927444, "train_loss_bc": 0.032362330704927444, "train_loss_llm": 0.0, "grad_norm": 0.028353499248623848, "global_step": 3135, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04724745452404022, "train_loss_bc": 0.04724745452404022, "train_loss_llm": 0.0, "grad_norm": 0.03572341427206993, "global_step": 3136, "epoch": 87, "lr": 0.001} +{"train_loss": 0.035649772733449936, "train_loss_bc": 0.035649772733449936, "train_loss_llm": 0.0, "grad_norm": 0.014939394779503345, "global_step": 3137, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04303327947854996, "train_loss_bc": 0.04303327947854996, "train_loss_llm": 0.0, "grad_norm": 0.024929147213697433, "global_step": 3138, "epoch": 87, "lr": 0.001} +{"train_loss": 0.041126929223537445, "train_loss_bc": 0.041126929223537445, "train_loss_llm": 0.0, "grad_norm": 0.02961524948477745, "global_step": 3139, "epoch": 87, "lr": 0.001} +{"train_loss": 0.0438448004424572, "train_loss_bc": 0.0438448004424572, "train_loss_llm": 0.0, "grad_norm": 0.037022560834884644, "global_step": 3140, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04678276926279068, "train_loss_bc": 0.046782393008470535, "train_loss_llm": 3.764033317565918e-05, "grad_norm": 0.019119979813694954, "global_step": 3141, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04717250540852547, "train_loss_bc": 0.04714144393801689, "train_loss_llm": 0.0031061172485351562, "grad_norm": 0.024810267612338066, "global_step": 3142, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04165056720376015, "train_loss_bc": 0.04165056720376015, "train_loss_llm": 0.0, "grad_norm": 0.0317879244685173, "global_step": 3143, "epoch": 87, "lr": 0.001} +{"train_loss": 0.043726541101932526, "train_loss_bc": 0.043726541101932526, "train_loss_llm": 0.0, "grad_norm": 0.040008094161748886, "global_step": 3144, "epoch": 87, "lr": 0.001} +{"train_loss": 0.05282293260097504, "train_loss_bc": 0.05282293260097504, "train_loss_llm": 0.0, "grad_norm": 0.020106801763176918, "global_step": 3145, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04061135649681091, "train_loss_bc": 0.040597785264253616, "train_loss_llm": 0.0013571679592132568, "grad_norm": 0.03473525121808052, "global_step": 3146, "epoch": 87, "lr": 0.001} +{"train_loss": 0.03724348172545433, "train_loss_bc": 0.03724348172545433, "train_loss_llm": 0.0, "grad_norm": 0.04760151728987694, "global_step": 3147, "epoch": 87, "lr": 0.001} +{"train_loss": 0.05179162323474884, "train_loss_bc": 0.05179162323474884, "train_loss_llm": 0.0, "grad_norm": 0.05113570764660835, "global_step": 3148, "epoch": 87, "lr": 0.001} +{"train_loss": 0.03991508483886719, "train_loss_bc": 0.039905160665512085, "train_loss_llm": 0.000992417335510254, "grad_norm": 0.015736879780888557, "global_step": 3149, "epoch": 87, "lr": 0.001} +{"train_loss": 0.044874921441078186, "train_loss_bc": 0.04485531151294708, "train_loss_llm": 0.00196114182472229, "grad_norm": 0.025944909080863, "global_step": 3150, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04500100761651993, "train_loss_bc": 0.04500100761651993, "train_loss_llm": 0.0, "grad_norm": 0.030024826526641846, "global_step": 3151, "epoch": 87, "lr": 0.001} +{"train_loss": 0.03886094689369202, "train_loss_bc": 0.03886094689369202, "train_loss_llm": 0.0, "grad_norm": 0.0332178995013237, "global_step": 3152, "epoch": 87, "lr": 0.001} +{"train_loss": 0.03956861048936844, "train_loss_bc": 0.03956861048936844, "train_loss_llm": 0.0, "grad_norm": 0.014724912121891975, "global_step": 3153, "epoch": 87, "lr": 0.001} +{"train_loss": 0.042426858097314835, "train_loss_bc": 0.042418066412210464, "train_loss_llm": 0.000879138708114624, "grad_norm": 0.026593035086989403, "global_step": 3154, "epoch": 87, "lr": 0.001} +{"train_loss": 0.038765210658311844, "train_loss_bc": 0.038765210658311844, "train_loss_llm": 0.0, "grad_norm": 0.03234832361340523, "global_step": 3155, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04912695288658142, "train_loss_bc": 0.04912695288658142, "train_loss_llm": 0.0, "grad_norm": 0.041058532893657684, "global_step": 3156, "epoch": 87, "lr": 0.001} +{"train_loss": 0.045323822647333145, "train_loss_bc": 0.045323822647333145, "train_loss_llm": 0.0, "grad_norm": 0.013507875613868237, "global_step": 3157, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04896220564842224, "train_loss_bc": 0.04896220564842224, "train_loss_llm": 0.0, "grad_norm": 0.024045953527092934, "global_step": 3158, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04391719028353691, "train_loss_bc": 0.04391719028353691, "train_loss_llm": 0.0, "grad_norm": 0.030291887000203133, "global_step": 3159, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04538493603467941, "train_loss_bc": 0.045353565365076065, "train_loss_llm": 0.003136962652206421, "grad_norm": 0.03663527965545654, "global_step": 3160, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04778839275240898, "train_loss_bc": 0.04775068536400795, "train_loss_llm": 0.0037706494331359863, "grad_norm": 0.01805407553911209, "global_step": 3161, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04261081665754318, "train_loss_bc": 0.04261081665754318, "train_loss_llm": 0.0, "grad_norm": 0.025596147403120995, "global_step": 3162, "epoch": 87, "lr": 0.001} +{"train_loss": 0.03810183331370354, "train_loss_bc": 0.03810183331370354, "train_loss_llm": 0.0, "grad_norm": 0.031558480113744736, "global_step": 3163, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04354026913642883, "train_loss_bc": 0.04353952780365944, "train_loss_llm": 7.426738739013672e-05, "grad_norm": 0.0418064258992672, "global_step": 3164, "epoch": 87, "lr": 0.001} +{"train_loss": 0.05139788240194321, "train_loss_bc": 0.05139788240194321, "train_loss_llm": 0.0, "grad_norm": 0.017355065792798996, "global_step": 3165, "epoch": 87, "lr": 0.001} +{"train_loss": 0.045087192207574844, "train_loss_bc": 0.045087192207574844, "train_loss_llm": 0.0, "grad_norm": 0.02306104078888893, "global_step": 3166, "epoch": 87, "lr": 0.001} +{"train_loss": 0.04410576944549879, "train_loss_bc": 0.05668085813522339, "train_loss_llm": 0.0009062588214874268, "grad_norm": 0.03209676593542099, "global_step": 3167, "epoch": 87, "lr": 0.001, "val_loss": 0.05187171697616577} +{"train_loss": 0.04421492666006088, "train_loss_bc": 0.04419383406639099, "train_loss_llm": 0.0021091103553771973, "grad_norm": 0.03926721215248108, "global_step": 3168, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04054497182369232, "train_loss_bc": 0.04053923487663269, "train_loss_llm": 0.0005738735198974609, "grad_norm": 0.018800362944602966, "global_step": 3169, "epoch": 88, "lr": 0.001} +{"train_loss": 0.042710065841674805, "train_loss_bc": 0.04267719015479088, "train_loss_llm": 0.003287404775619507, "grad_norm": 0.02350303716957569, "global_step": 3170, "epoch": 88, "lr": 0.001} +{"train_loss": 0.034925248473882675, "train_loss_bc": 0.034925248473882675, "train_loss_llm": 0.0, "grad_norm": 0.028652679175138474, "global_step": 3171, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04987533017992973, "train_loss_bc": 0.04987533017992973, "train_loss_llm": 0.0, "grad_norm": 0.0354444682598114, "global_step": 3172, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04132695496082306, "train_loss_bc": 0.04132651537656784, "train_loss_llm": 4.413723945617676e-05, "grad_norm": 0.015287945978343487, "global_step": 3173, "epoch": 88, "lr": 0.001} +{"train_loss": 0.046163883060216904, "train_loss_bc": 0.04614178463816643, "train_loss_llm": 0.002209872007369995, "grad_norm": 0.02394910715520382, "global_step": 3174, "epoch": 88, "lr": 0.001} +{"train_loss": 0.046416983008384705, "train_loss_bc": 0.046416983008384705, "train_loss_llm": 0.0, "grad_norm": 0.027752872556447983, "global_step": 3175, "epoch": 88, "lr": 0.001} +{"train_loss": 0.047681208699941635, "train_loss_bc": 0.0476590171456337, "train_loss_llm": 0.0022193193435668945, "grad_norm": 0.03771352767944336, "global_step": 3176, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04535117745399475, "train_loss_bc": 0.04535117745399475, "train_loss_llm": 0.0, "grad_norm": 0.01661481335759163, "global_step": 3177, "epoch": 88, "lr": 0.001} +{"train_loss": 0.03862893581390381, "train_loss_bc": 0.038626234978437424, "train_loss_llm": 0.0002702474594116211, "grad_norm": 0.021895701065659523, "global_step": 3178, "epoch": 88, "lr": 0.001} +{"train_loss": 0.040881119668483734, "train_loss_bc": 0.040881119668483734, "train_loss_llm": 0.0, "grad_norm": 0.02809038758277893, "global_step": 3179, "epoch": 88, "lr": 0.001} +{"train_loss": 0.044903628528118134, "train_loss_bc": 0.044900573790073395, "train_loss_llm": 0.00030553340911865234, "grad_norm": 0.03329683095216751, "global_step": 3180, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04980241879820824, "train_loss_bc": 0.049788445234298706, "train_loss_llm": 0.0013972818851470947, "grad_norm": 0.015280606225132942, "global_step": 3181, "epoch": 88, "lr": 0.001} +{"train_loss": 0.051993437111377716, "train_loss_bc": 0.051993437111377716, "train_loss_llm": 0.0, "grad_norm": 0.02337060309946537, "global_step": 3182, "epoch": 88, "lr": 0.001} +{"train_loss": 0.047207579016685486, "train_loss_bc": 0.04718725010752678, "train_loss_llm": 0.002033025026321411, "grad_norm": 0.03343860059976578, "global_step": 3183, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04303983226418495, "train_loss_bc": 0.04303983226418495, "train_loss_llm": 0.0, "grad_norm": 0.038074105978012085, "global_step": 3184, "epoch": 88, "lr": 0.001} +{"train_loss": 0.03937729448080063, "train_loss_bc": 0.03937729448080063, "train_loss_llm": 0.0, "grad_norm": 0.018735364079475403, "global_step": 3185, "epoch": 88, "lr": 0.001} +{"train_loss": 0.03973713517189026, "train_loss_bc": 0.03973713517189026, "train_loss_llm": 0.0, "grad_norm": 0.02913665398955345, "global_step": 3186, "epoch": 88, "lr": 0.001} +{"train_loss": 0.041365694254636765, "train_loss_bc": 0.041365694254636765, "train_loss_llm": 0.0, "grad_norm": 0.044550564140081406, "global_step": 3187, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04366316273808479, "train_loss_bc": 0.04366316273808479, "train_loss_llm": 0.0, "grad_norm": 0.052189040929079056, "global_step": 3188, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04616163671016693, "train_loss_bc": 0.04616163671016693, "train_loss_llm": 0.0, "grad_norm": 0.019204549491405487, "global_step": 3189, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04888539016246796, "train_loss_bc": 0.04883768782019615, "train_loss_llm": 0.004770249128341675, "grad_norm": 0.02469545416533947, "global_step": 3190, "epoch": 88, "lr": 0.001} +{"train_loss": 0.043951235711574554, "train_loss_bc": 0.043951235711574554, "train_loss_llm": 0.0, "grad_norm": 0.03707895427942276, "global_step": 3191, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04510381445288658, "train_loss_bc": 0.04510381445288658, "train_loss_llm": 0.0, "grad_norm": 0.04019832983613014, "global_step": 3192, "epoch": 88, "lr": 0.001} +{"train_loss": 0.039416100829839706, "train_loss_bc": 0.039416100829839706, "train_loss_llm": 0.0, "grad_norm": 0.013869539834558964, "global_step": 3193, "epoch": 88, "lr": 0.001} +{"train_loss": 0.039412207901477814, "train_loss_bc": 0.039412207901477814, "train_loss_llm": 0.0, "grad_norm": 0.022843554615974426, "global_step": 3194, "epoch": 88, "lr": 0.001} +{"train_loss": 0.046885889023542404, "train_loss_bc": 0.046885889023542404, "train_loss_llm": 0.0, "grad_norm": 0.027153773233294487, "global_step": 3195, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04435478895902634, "train_loss_bc": 0.04435478895902634, "train_loss_llm": 0.0, "grad_norm": 0.03612688556313515, "global_step": 3196, "epoch": 88, "lr": 0.001} +{"train_loss": 0.03904198110103607, "train_loss_bc": 0.03904198110103607, "train_loss_llm": 0.0, "grad_norm": 0.017230449244379997, "global_step": 3197, "epoch": 88, "lr": 0.001} +{"train_loss": 0.05324244499206543, "train_loss_bc": 0.05322996899485588, "train_loss_llm": 0.0012476146221160889, "grad_norm": 0.025509417057037354, "global_step": 3198, "epoch": 88, "lr": 0.001} +{"train_loss": 0.03676581010222435, "train_loss_bc": 0.03673212602734566, "train_loss_llm": 0.0033683478832244873, "grad_norm": 0.03079182840883732, "global_step": 3199, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04744362831115723, "train_loss_bc": 0.04744362831115723, "train_loss_llm": 0.0, "grad_norm": 0.038557931780815125, "global_step": 3200, "epoch": 88, "lr": 0.001} +{"train_loss": 0.043574802577495575, "train_loss_bc": 0.043574802577495575, "train_loss_llm": 0.0, "grad_norm": 0.017881028354167938, "global_step": 3201, "epoch": 88, "lr": 0.001} +{"train_loss": 0.038496337831020355, "train_loss_bc": 0.03849535807967186, "train_loss_llm": 9.807944297790527e-05, "grad_norm": 0.02743835747241974, "global_step": 3202, "epoch": 88, "lr": 0.001} +{"train_loss": 0.04408757626596424, "train_loss_bc": 0.05460568889975548, "train_loss_llm": 0.0, "grad_norm": 0.0416991226375103, "global_step": 3203, "epoch": 88, "lr": 0.001, "val_loss": 0.051766689866781235} +{"train_loss": 0.04892358556389809, "train_loss_bc": 0.04892358556389809, "train_loss_llm": 0.0, "grad_norm": 0.05422481149435043, "global_step": 3204, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04773370921611786, "train_loss_bc": 0.04773370921611786, "train_loss_llm": 0.0, "grad_norm": 0.017552439123392105, "global_step": 3205, "epoch": 89, "lr": 0.001} +{"train_loss": 0.046568311750888824, "train_loss_bc": 0.046568311750888824, "train_loss_llm": 0.0, "grad_norm": 0.02277468703687191, "global_step": 3206, "epoch": 89, "lr": 0.001} +{"train_loss": 0.041446104645729065, "train_loss_bc": 0.04141657426953316, "train_loss_llm": 0.0029532015323638916, "grad_norm": 0.025145743042230606, "global_step": 3207, "epoch": 89, "lr": 0.001} +{"train_loss": 0.03699430450797081, "train_loss_bc": 0.03698793798685074, "train_loss_llm": 0.0006365478038787842, "grad_norm": 0.028949886560440063, "global_step": 3208, "epoch": 89, "lr": 0.001} +{"train_loss": 0.051979754120111465, "train_loss_bc": 0.05197972059249878, "train_loss_llm": 3.2782554626464844e-06, "grad_norm": 0.013957364484667778, "global_step": 3209, "epoch": 89, "lr": 0.001} +{"train_loss": 0.03999115154147148, "train_loss_bc": 0.039985038340091705, "train_loss_llm": 0.0006113052368164062, "grad_norm": 0.023502539843320847, "global_step": 3210, "epoch": 89, "lr": 0.001} +{"train_loss": 0.043436773121356964, "train_loss_bc": 0.043413713574409485, "train_loss_llm": 0.0023060142993927, "grad_norm": 0.03069991245865822, "global_step": 3211, "epoch": 89, "lr": 0.001} +{"train_loss": 0.046012211591005325, "train_loss_bc": 0.046012211591005325, "train_loss_llm": 0.0, "grad_norm": 0.03546600043773651, "global_step": 3212, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04779457673430443, "train_loss_bc": 0.04779457673430443, "train_loss_llm": 0.0, "grad_norm": 0.019274061545729637, "global_step": 3213, "epoch": 89, "lr": 0.001} +{"train_loss": 0.038236722350120544, "train_loss_bc": 0.038236722350120544, "train_loss_llm": 0.0, "grad_norm": 0.023728100582957268, "global_step": 3214, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04034090414643288, "train_loss_bc": 0.040327463299036026, "train_loss_llm": 0.0013442039489746094, "grad_norm": 0.03295794129371643, "global_step": 3215, "epoch": 89, "lr": 0.001} +{"train_loss": 0.045035433024168015, "train_loss_bc": 0.04503091424703598, "train_loss_llm": 0.00045183300971984863, "grad_norm": 0.04368233680725098, "global_step": 3216, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04210195317864418, "train_loss_bc": 0.042097121477127075, "train_loss_llm": 0.0004831552505493164, "grad_norm": 0.020525110885500908, "global_step": 3217, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04758474975824356, "train_loss_bc": 0.04758474975824356, "train_loss_llm": 0.0, "grad_norm": 0.028895804658532143, "global_step": 3218, "epoch": 89, "lr": 0.001} +{"train_loss": 0.043663010001182556, "train_loss_bc": 0.043663010001182556, "train_loss_llm": 0.0, "grad_norm": 0.035641785711050034, "global_step": 3219, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04167763516306877, "train_loss_bc": 0.04167763516306877, "train_loss_llm": 0.0, "grad_norm": 0.0427817702293396, "global_step": 3220, "epoch": 89, "lr": 0.001} +{"train_loss": 0.038901038467884064, "train_loss_bc": 0.038894765079021454, "train_loss_llm": 0.0006272196769714355, "grad_norm": 0.013856828212738037, "global_step": 3221, "epoch": 89, "lr": 0.001} +{"train_loss": 0.043458614498376846, "train_loss_bc": 0.043458614498376846, "train_loss_llm": 0.0, "grad_norm": 0.022620802745223045, "global_step": 3222, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04575671628117561, "train_loss_bc": 0.04575671628117561, "train_loss_llm": 0.0, "grad_norm": 0.02691301517188549, "global_step": 3223, "epoch": 89, "lr": 0.001} +{"train_loss": 0.044943906366825104, "train_loss_bc": 0.044943906366825104, "train_loss_llm": 0.0, "grad_norm": 0.034882768988609314, "global_step": 3224, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04624354466795921, "train_loss_bc": 0.04624304547905922, "train_loss_llm": 4.997849464416504e-05, "grad_norm": 0.01652810536324978, "global_step": 3225, "epoch": 89, "lr": 0.001} +{"train_loss": 0.03909619152545929, "train_loss_bc": 0.03909619152545929, "train_loss_llm": 0.0, "grad_norm": 0.021280862390995026, "global_step": 3226, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04719454050064087, "train_loss_bc": 0.04719454050064087, "train_loss_llm": 0.0, "grad_norm": 0.027918195351958275, "global_step": 3227, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04451959207653999, "train_loss_bc": 0.04451959207653999, "train_loss_llm": 0.0, "grad_norm": 0.03403020650148392, "global_step": 3228, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04323463886976242, "train_loss_bc": 0.04323463886976242, "train_loss_llm": 0.0, "grad_norm": 0.016979588195681572, "global_step": 3229, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04195122793316841, "train_loss_bc": 0.04193758964538574, "train_loss_llm": 0.0013639330863952637, "grad_norm": 0.021530693396925926, "global_step": 3230, "epoch": 89, "lr": 0.001} +{"train_loss": 0.040013864636421204, "train_loss_bc": 0.040013864636421204, "train_loss_llm": 0.0, "grad_norm": 0.027056070044636726, "global_step": 3231, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04856489971280098, "train_loss_bc": 0.04856489971280098, "train_loss_llm": 0.0, "grad_norm": 0.038166385143995285, "global_step": 3232, "epoch": 89, "lr": 0.001} +{"train_loss": 0.049825578927993774, "train_loss_bc": 0.049825578927993774, "train_loss_llm": 0.0, "grad_norm": 0.015682807192206383, "global_step": 3233, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04100761190056801, "train_loss_bc": 0.04092923179268837, "train_loss_llm": 0.007838010787963867, "grad_norm": 0.023143742233514786, "global_step": 3234, "epoch": 89, "lr": 0.001} +{"train_loss": 0.04160211607813835, "train_loss_bc": 0.04160211607813835, "train_loss_llm": 0.0, "grad_norm": 0.03157726302742958, "global_step": 3235, "epoch": 89, "lr": 0.001} +{"train_loss": 0.037299275398254395, "train_loss_bc": 0.03726721554994583, "train_loss_llm": 0.0032059848308563232, "grad_norm": 0.04066719114780426, "global_step": 3236, "epoch": 89, "lr": 0.001} +{"train_loss": 0.046262260526418686, "train_loss_bc": 0.046262260526418686, "train_loss_llm": 0.0, "grad_norm": 0.01601712591946125, "global_step": 3237, "epoch": 89, "lr": 0.001} +{"train_loss": 0.043705735355615616, "train_loss_bc": 0.043705735355615616, "train_loss_llm": 0.0, "grad_norm": 0.025072934105992317, "global_step": 3238, "epoch": 89, "lr": 0.001} +{"train_loss": 0.044068183853394456, "train_loss_bc": 0.05335237458348274, "train_loss_llm": 0.0, "grad_norm": 0.03436269238591194, "global_step": 3239, "epoch": 89, "lr": 0.001, "val_loss": 0.052040062844753265} +{"train_loss": 0.04248444363474846, "train_loss_bc": 0.04246286302804947, "train_loss_llm": 0.0021581947803497314, "grad_norm": 0.03343818336725235, "global_step": 3240, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04551731422543526, "train_loss_bc": 0.04551731422543526, "train_loss_llm": 0.0, "grad_norm": 0.01738055609166622, "global_step": 3241, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04912880063056946, "train_loss_bc": 0.04912880063056946, "train_loss_llm": 0.0, "grad_norm": 0.021398479118943214, "global_step": 3242, "epoch": 90, "lr": 0.001} +{"train_loss": 0.03539056330919266, "train_loss_bc": 0.03539056330919266, "train_loss_llm": 0.0, "grad_norm": 0.024771030992269516, "global_step": 3243, "epoch": 90, "lr": 0.001} +{"train_loss": 0.03816785663366318, "train_loss_bc": 0.03813716024160385, "train_loss_llm": 0.003069460391998291, "grad_norm": 0.029313281178474426, "global_step": 3244, "epoch": 90, "lr": 0.001} +{"train_loss": 0.05107036232948303, "train_loss_bc": 0.0510241761803627, "train_loss_llm": 0.004618704319000244, "grad_norm": 0.015859238803386688, "global_step": 3245, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04204866662621498, "train_loss_bc": 0.042038578540086746, "train_loss_llm": 0.001008838415145874, "grad_norm": 0.021333424374461174, "global_step": 3246, "epoch": 90, "lr": 0.001} +{"train_loss": 0.02807602472603321, "train_loss_bc": 0.02807602472603321, "train_loss_llm": 0.0, "grad_norm": 0.0236346535384655, "global_step": 3247, "epoch": 90, "lr": 0.001} +{"train_loss": 0.049660202115774155, "train_loss_bc": 0.049660202115774155, "train_loss_llm": 0.0, "grad_norm": 0.027496378868818283, "global_step": 3248, "epoch": 90, "lr": 0.001} +{"train_loss": 0.044401030987501144, "train_loss_bc": 0.044390663504600525, "train_loss_llm": 0.0010367631912231445, "grad_norm": 0.020050467923283577, "global_step": 3249, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04591476917266846, "train_loss_bc": 0.045899372547864914, "train_loss_llm": 0.0015395879745483398, "grad_norm": 0.02507150173187256, "global_step": 3250, "epoch": 90, "lr": 0.001} +{"train_loss": 0.0433359332382679, "train_loss_bc": 0.043303437530994415, "train_loss_llm": 0.0032495856285095215, "grad_norm": 0.03578516095876694, "global_step": 3251, "epoch": 90, "lr": 0.001} +{"train_loss": 0.03760021552443504, "train_loss_bc": 0.03760021552443504, "train_loss_llm": 0.0, "grad_norm": 0.03648388758301735, "global_step": 3252, "epoch": 90, "lr": 0.001} +{"train_loss": 0.05023692548274994, "train_loss_bc": 0.05021769553422928, "train_loss_llm": 0.0019229650497436523, "grad_norm": 0.020440220832824707, "global_step": 3253, "epoch": 90, "lr": 0.001} +{"train_loss": 0.03839634731411934, "train_loss_bc": 0.03839634731411934, "train_loss_llm": 0.0, "grad_norm": 0.031069733202457428, "global_step": 3254, "epoch": 90, "lr": 0.001} +{"train_loss": 0.049547839909791946, "train_loss_bc": 0.049547839909791946, "train_loss_llm": 0.0, "grad_norm": 0.03938641771674156, "global_step": 3255, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04596884921193123, "train_loss_bc": 0.04596884921193123, "train_loss_llm": 0.0, "grad_norm": 0.04379694163799286, "global_step": 3256, "epoch": 90, "lr": 0.001} +{"train_loss": 0.03746356815099716, "train_loss_bc": 0.03746356815099716, "train_loss_llm": 0.0, "grad_norm": 0.016271093860268593, "global_step": 3257, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04446335509419441, "train_loss_bc": 0.044439662247896194, "train_loss_llm": 0.002369105815887451, "grad_norm": 0.02629578858613968, "global_step": 3258, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04249744489789009, "train_loss_bc": 0.04247628152370453, "train_loss_llm": 0.002116471529006958, "grad_norm": 0.03227079659700394, "global_step": 3259, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04342542961239815, "train_loss_bc": 0.04342542961239815, "train_loss_llm": 0.0, "grad_norm": 0.04191796854138374, "global_step": 3260, "epoch": 90, "lr": 0.001} +{"train_loss": 0.03908878192305565, "train_loss_bc": 0.03908878192305565, "train_loss_llm": 0.0, "grad_norm": 0.016616802662611008, "global_step": 3261, "epoch": 90, "lr": 0.001} +{"train_loss": 0.045164912939071655, "train_loss_bc": 0.045164912939071655, "train_loss_llm": 0.0, "grad_norm": 0.025547141209244728, "global_step": 3262, "epoch": 90, "lr": 0.001} +{"train_loss": 0.047963228076696396, "train_loss_bc": 0.047963228076696396, "train_loss_llm": 0.0, "grad_norm": 0.032402507960796356, "global_step": 3263, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04456976801156998, "train_loss_bc": 0.04456976801156998, "train_loss_llm": 0.0, "grad_norm": 0.03622781112790108, "global_step": 3264, "epoch": 90, "lr": 0.001} +{"train_loss": 0.043020665645599365, "train_loss_bc": 0.043020665645599365, "train_loss_llm": 0.0, "grad_norm": 0.01640867441892624, "global_step": 3265, "epoch": 90, "lr": 0.001} +{"train_loss": 0.046260517090559006, "train_loss_bc": 0.046253420412540436, "train_loss_llm": 0.0007097125053405762, "grad_norm": 0.025067243725061417, "global_step": 3266, "epoch": 90, "lr": 0.001} +{"train_loss": 0.041305311024188995, "train_loss_bc": 0.041305311024188995, "train_loss_llm": 0.0, "grad_norm": 0.029988301917910576, "global_step": 3267, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04417756572365761, "train_loss_bc": 0.04417756572365761, "train_loss_llm": 0.0, "grad_norm": 0.03904498741030693, "global_step": 3268, "epoch": 90, "lr": 0.001} +{"train_loss": 0.045684605836868286, "train_loss_bc": 0.045684605836868286, "train_loss_llm": 0.0, "grad_norm": 0.019815659150481224, "global_step": 3269, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04797550290822983, "train_loss_bc": 0.04797550290822983, "train_loss_llm": 0.0, "grad_norm": 0.030884018167853355, "global_step": 3270, "epoch": 90, "lr": 0.001} +{"train_loss": 0.044572293758392334, "train_loss_bc": 0.044572293758392334, "train_loss_llm": 0.0, "grad_norm": 0.041108231991529465, "global_step": 3271, "epoch": 90, "lr": 0.001} +{"train_loss": 0.04086751863360405, "train_loss_bc": 0.04086700454354286, "train_loss_llm": 5.131959915161133e-05, "grad_norm": 0.0492253378033638, "global_step": 3272, "epoch": 90, "lr": 0.001} +{"train_loss": 0.05361608788371086, "train_loss_bc": 0.05361608788371086, "train_loss_llm": 0.0, "grad_norm": 0.02279573306441307, "global_step": 3273, "epoch": 90, "lr": 0.001} +{"train_loss": 0.03826341778039932, "train_loss_bc": 0.03826341778039932, "train_loss_llm": 0.0, "grad_norm": 0.027576994150877, "global_step": 3274, "epoch": 90, "lr": 0.001} +{"train_loss": 0.043851708320693836, "train_loss_bc": 0.05133537948131561, "train_loss_llm": 0.0, "grad_norm": 0.03341073915362358, "global_step": 3275, "epoch": 90, "lr": 0.001, "train/cumulative_reward": 76.73722599278639, "train/mean_score": 17.911822176355248, "train/success_rate": 0.5, "test/cumulative_reward": 88.5202471396179, "test/mean_score": 10.452128804511824, "test/success_rate": 0.22, "val_loss": 0.05171174183487892, "train_action_mse_error": 0.04776793718338013} +{"train_loss": 0.04086686298251152, "train_loss_bc": 0.04086686298251152, "train_loss_llm": 0.0, "grad_norm": 0.037900496274232864, "global_step": 3276, "epoch": 91, "lr": 0.001} +{"train_loss": 0.041229497641325, "train_loss_bc": 0.041229497641325, "train_loss_llm": 0.0, "grad_norm": 0.011958815157413483, "global_step": 3277, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04685387760400772, "train_loss_bc": 0.04681785777211189, "train_loss_llm": 0.003602057695388794, "grad_norm": 0.021347157657146454, "global_step": 3278, "epoch": 91, "lr": 0.001} +{"train_loss": 0.048354968428611755, "train_loss_bc": 0.048354968428611755, "train_loss_llm": 0.0, "grad_norm": 0.02566857822239399, "global_step": 3279, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04504326358437538, "train_loss_bc": 0.04504326358437538, "train_loss_llm": 0.0, "grad_norm": 0.03547355532646179, "global_step": 3280, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04553933069109917, "train_loss_bc": 0.04553933069109917, "train_loss_llm": 0.0, "grad_norm": 0.01689246855676174, "global_step": 3281, "epoch": 91, "lr": 0.001} +{"train_loss": 0.048782624304294586, "train_loss_bc": 0.04876090958714485, "train_loss_llm": 0.002171546220779419, "grad_norm": 0.02173621580004692, "global_step": 3282, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04939384385943413, "train_loss_bc": 0.04939384385943413, "train_loss_llm": 0.0, "grad_norm": 0.03188677877187729, "global_step": 3283, "epoch": 91, "lr": 0.001} +{"train_loss": 0.05032367259263992, "train_loss_bc": 0.05032367259263992, "train_loss_llm": 0.0, "grad_norm": 0.0378783717751503, "global_step": 3284, "epoch": 91, "lr": 0.001} +{"train_loss": 0.0429261215031147, "train_loss_bc": 0.04291923716664314, "train_loss_llm": 0.0006885826587677002, "grad_norm": 0.02032652497291565, "global_step": 3285, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04303070902824402, "train_loss_bc": 0.04303070902824402, "train_loss_llm": 0.0, "grad_norm": 0.023681284859776497, "global_step": 3286, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04055536165833473, "train_loss_bc": 0.04055536165833473, "train_loss_llm": 0.0, "grad_norm": 0.030120162293314934, "global_step": 3287, "epoch": 91, "lr": 0.001} +{"train_loss": 0.038921620696783066, "train_loss_bc": 0.038921620696783066, "train_loss_llm": 0.0, "grad_norm": 0.0396796315908432, "global_step": 3288, "epoch": 91, "lr": 0.001} +{"train_loss": 0.03621215745806694, "train_loss_bc": 0.03619077056646347, "train_loss_llm": 0.0021387338638305664, "grad_norm": 0.016112741082906723, "global_step": 3289, "epoch": 91, "lr": 0.001} +{"train_loss": 0.042058173567056656, "train_loss_bc": 0.04205084219574928, "train_loss_llm": 0.0007332563400268555, "grad_norm": 0.02462039329111576, "global_step": 3290, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04410644620656967, "train_loss_bc": 0.04410644620656967, "train_loss_llm": 0.0, "grad_norm": 0.030705170705914497, "global_step": 3291, "epoch": 91, "lr": 0.001} +{"train_loss": 0.041129689663648605, "train_loss_bc": 0.04112791270017624, "train_loss_llm": 0.00017768144607543945, "grad_norm": 0.03719671815633774, "global_step": 3292, "epoch": 91, "lr": 0.001} +{"train_loss": 0.05044427141547203, "train_loss_bc": 0.050437405705451965, "train_loss_llm": 0.0006864368915557861, "grad_norm": 0.021031243726611137, "global_step": 3293, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04202297329902649, "train_loss_bc": 0.04201199486851692, "train_loss_llm": 0.001097947359085083, "grad_norm": 0.031613387167453766, "global_step": 3294, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04661865532398224, "train_loss_bc": 0.04661865532398224, "train_loss_llm": 0.0, "grad_norm": 0.03744214028120041, "global_step": 3295, "epoch": 91, "lr": 0.001} +{"train_loss": 0.045199524611234665, "train_loss_bc": 0.04519374668598175, "train_loss_llm": 0.0005778968334197998, "grad_norm": 0.05126643553376198, "global_step": 3296, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04457860440015793, "train_loss_bc": 0.04457860440015793, "train_loss_llm": 0.0, "grad_norm": 0.016693275421857834, "global_step": 3297, "epoch": 91, "lr": 0.001} +{"train_loss": 0.03832584246993065, "train_loss_bc": 0.03832584246993065, "train_loss_llm": 0.0, "grad_norm": 0.02414299175143242, "global_step": 3298, "epoch": 91, "lr": 0.001} +{"train_loss": 0.038427237421274185, "train_loss_bc": 0.038427237421274185, "train_loss_llm": 0.0, "grad_norm": 0.02834210731089115, "global_step": 3299, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04477417469024658, "train_loss_bc": 0.04477417469024658, "train_loss_llm": 0.0, "grad_norm": 0.033203136175870895, "global_step": 3300, "epoch": 91, "lr": 0.001} +{"train_loss": 0.03781937062740326, "train_loss_bc": 0.03781937062740326, "train_loss_llm": 0.0, "grad_norm": 0.013539460487663746, "global_step": 3301, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04443111643195152, "train_loss_bc": 0.04438472166657448, "train_loss_llm": 0.004639416933059692, "grad_norm": 0.02721245400607586, "global_step": 3302, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04794830083847046, "train_loss_bc": 0.04792984575033188, "train_loss_llm": 0.0018453896045684814, "grad_norm": 0.03588408604264259, "global_step": 3303, "epoch": 91, "lr": 0.001} +{"train_loss": 0.05355125665664673, "train_loss_bc": 0.05355125665664673, "train_loss_llm": 0.0, "grad_norm": 0.050978176295757294, "global_step": 3304, "epoch": 91, "lr": 0.001} +{"train_loss": 0.036902714520692825, "train_loss_bc": 0.036902714520692825, "train_loss_llm": 0.0, "grad_norm": 0.012379121966660023, "global_step": 3305, "epoch": 91, "lr": 0.001} +{"train_loss": 0.044883206486701965, "train_loss_bc": 0.044883206486701965, "train_loss_llm": 0.0, "grad_norm": 0.0205093827098608, "global_step": 3306, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04424342140555382, "train_loss_bc": 0.04421631619334221, "train_loss_llm": 0.0027106404304504395, "grad_norm": 0.02981460466980934, "global_step": 3307, "epoch": 91, "lr": 0.001} +{"train_loss": 0.036632660776376724, "train_loss_bc": 0.036632660776376724, "train_loss_llm": 0.0, "grad_norm": 0.038541264832019806, "global_step": 3308, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04500529170036316, "train_loss_bc": 0.04500529170036316, "train_loss_llm": 0.0, "grad_norm": 0.019713083282113075, "global_step": 3309, "epoch": 91, "lr": 0.001} +{"train_loss": 0.045309003442525864, "train_loss_bc": 0.045309003442525864, "train_loss_llm": 0.0, "grad_norm": 0.03302471712231636, "global_step": 3310, "epoch": 91, "lr": 0.001} +{"train_loss": 0.04378089557091395, "train_loss_bc": 0.04363448545336723, "train_loss_llm": 0.0031906068325042725, "grad_norm": 0.0425100103020668, "global_step": 3311, "epoch": 91, "lr": 0.001, "val_loss": 0.051969412714242935} +{"train_loss": 0.045759495347738266, "train_loss_bc": 0.045759495347738266, "train_loss_llm": 0.0, "grad_norm": 0.044880714267492294, "global_step": 3312, "epoch": 92, "lr": 0.001} +{"train_loss": 0.045419927686452866, "train_loss_bc": 0.045419927686452866, "train_loss_llm": 0.0, "grad_norm": 0.01826317608356476, "global_step": 3313, "epoch": 92, "lr": 0.001} +{"train_loss": 0.03939174860715866, "train_loss_bc": 0.03936537355184555, "train_loss_llm": 0.002637326717376709, "grad_norm": 0.020642267540097237, "global_step": 3314, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04888809844851494, "train_loss_bc": 0.04888809844851494, "train_loss_llm": 0.0, "grad_norm": 0.028503671288490295, "global_step": 3315, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04780757799744606, "train_loss_bc": 0.04779583960771561, "train_loss_llm": 0.0011739134788513184, "grad_norm": 0.04024381935596466, "global_step": 3316, "epoch": 92, "lr": 0.001} +{"train_loss": 0.043036360293626785, "train_loss_bc": 0.043036360293626785, "train_loss_llm": 0.0, "grad_norm": 0.01346642803400755, "global_step": 3317, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04084881767630577, "train_loss_bc": 0.04084881767630577, "train_loss_llm": 0.0, "grad_norm": 0.02095482312142849, "global_step": 3318, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04655713960528374, "train_loss_bc": 0.04654879868030548, "train_loss_llm": 0.0008339881896972656, "grad_norm": 0.028728479519486427, "global_step": 3319, "epoch": 92, "lr": 0.001} +{"train_loss": 0.03707857057452202, "train_loss_bc": 0.03706192225217819, "train_loss_llm": 0.001664966344833374, "grad_norm": 0.032698098570108414, "global_step": 3320, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04260091856122017, "train_loss_bc": 0.042580366134643555, "train_loss_llm": 0.002055257558822632, "grad_norm": 0.01792248710989952, "global_step": 3321, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04042661562561989, "train_loss_bc": 0.04042661562561989, "train_loss_llm": 0.0, "grad_norm": 0.022116389125585556, "global_step": 3322, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04261893406510353, "train_loss_bc": 0.04261893406510353, "train_loss_llm": 0.0, "grad_norm": 0.024673016741871834, "global_step": 3323, "epoch": 92, "lr": 0.001} +{"train_loss": 0.03827054798603058, "train_loss_bc": 0.03827054798603058, "train_loss_llm": 0.0, "grad_norm": 0.030385978519916534, "global_step": 3324, "epoch": 92, "lr": 0.001} +{"train_loss": 0.03750685229897499, "train_loss_bc": 0.03750685229897499, "train_loss_llm": 0.0, "grad_norm": 0.013788574375212193, "global_step": 3325, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04124460741877556, "train_loss_bc": 0.041214946657419205, "train_loss_llm": 0.0029661357402801514, "grad_norm": 0.022311490029096603, "global_step": 3326, "epoch": 92, "lr": 0.001} +{"train_loss": 0.0390462726354599, "train_loss_bc": 0.0390462726354599, "train_loss_llm": 0.0, "grad_norm": 0.027529554441571236, "global_step": 3327, "epoch": 92, "lr": 0.001} +{"train_loss": 0.041500501334667206, "train_loss_bc": 0.04149986431002617, "train_loss_llm": 6.35683536529541e-05, "grad_norm": 0.03379372879862785, "global_step": 3328, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04510289803147316, "train_loss_bc": 0.0450771264731884, "train_loss_llm": 0.0025772154331207275, "grad_norm": 0.018627747893333435, "global_step": 3329, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04014011099934578, "train_loss_bc": 0.04014011099934578, "train_loss_llm": 0.0, "grad_norm": 0.028900736942887306, "global_step": 3330, "epoch": 92, "lr": 0.001} +{"train_loss": 0.03731480613350868, "train_loss_bc": 0.03731480613350868, "train_loss_llm": 0.0, "grad_norm": 0.03520243242383003, "global_step": 3331, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04453234747052193, "train_loss_bc": 0.04450812563300133, "train_loss_llm": 0.0024223029613494873, "grad_norm": 0.04617586359381676, "global_step": 3332, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04479197412729263, "train_loss_bc": 0.04479197412729263, "train_loss_llm": 0.0, "grad_norm": 0.01670125313103199, "global_step": 3333, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04249599948525429, "train_loss_bc": 0.04249599948525429, "train_loss_llm": 0.0, "grad_norm": 0.024568023160099983, "global_step": 3334, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04483317211270332, "train_loss_bc": 0.04483148828148842, "train_loss_llm": 0.00016823410987854004, "grad_norm": 0.027592835947871208, "global_step": 3335, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04537474736571312, "train_loss_bc": 0.04537474736571312, "train_loss_llm": 0.0, "grad_norm": 0.03003574162721634, "global_step": 3336, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04635969176888466, "train_loss_bc": 0.04635969176888466, "train_loss_llm": 0.0, "grad_norm": 0.017775095999240875, "global_step": 3337, "epoch": 92, "lr": 0.001} +{"train_loss": 0.0444849357008934, "train_loss_bc": 0.0444849357008934, "train_loss_llm": 0.0, "grad_norm": 0.02533232606947422, "global_step": 3338, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04919056221842766, "train_loss_bc": 0.04919056221842766, "train_loss_llm": 0.0, "grad_norm": 0.03146160766482353, "global_step": 3339, "epoch": 92, "lr": 0.001} +{"train_loss": 0.0435606986284256, "train_loss_bc": 0.0435606986284256, "train_loss_llm": 0.0, "grad_norm": 0.04081215709447861, "global_step": 3340, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04447907209396362, "train_loss_bc": 0.04447907209396362, "train_loss_llm": 0.0, "grad_norm": 0.018223775550723076, "global_step": 3341, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04787348210811615, "train_loss_bc": 0.04787348210811615, "train_loss_llm": 0.0, "grad_norm": 0.026484647765755653, "global_step": 3342, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04799149930477142, "train_loss_bc": 0.04799149930477142, "train_loss_llm": 0.0, "grad_norm": 0.03846510127186775, "global_step": 3343, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04535145312547684, "train_loss_bc": 0.04535145312547684, "train_loss_llm": 0.0, "grad_norm": 0.04851217195391655, "global_step": 3344, "epoch": 92, "lr": 0.001} +{"train_loss": 0.041710611432790756, "train_loss_bc": 0.04170343279838562, "train_loss_llm": 0.0007179975509643555, "grad_norm": 0.021054385229945183, "global_step": 3345, "epoch": 92, "lr": 0.001} +{"train_loss": 0.05333131551742554, "train_loss_bc": 0.05333131551742554, "train_loss_llm": 0.0, "grad_norm": 0.028560800477862358, "global_step": 3346, "epoch": 92, "lr": 0.001} +{"train_loss": 0.04380296398368147, "train_loss_bc": 0.0499642938375473, "train_loss_llm": 0.002004653215408325, "grad_norm": 0.0400812141597271, "global_step": 3347, "epoch": 92, "lr": 0.001, "val_loss": 0.05185282230377197} +{"train_loss": 0.04453419893980026, "train_loss_bc": 0.044501934200525284, "train_loss_llm": 0.0032266080379486084, "grad_norm": 0.0487784743309021, "global_step": 3348, "epoch": 93, "lr": 0.001} +{"train_loss": 0.044313207268714905, "train_loss_bc": 0.044307783246040344, "train_loss_llm": 0.0005424618721008301, "grad_norm": 0.01896895095705986, "global_step": 3349, "epoch": 93, "lr": 0.001} +{"train_loss": 0.05087806656956673, "train_loss_bc": 0.050858695060014725, "train_loss_llm": 0.0019372403621673584, "grad_norm": 0.023494675755500793, "global_step": 3350, "epoch": 93, "lr": 0.001} +{"train_loss": 0.05233563855290413, "train_loss_bc": 0.05233563855290413, "train_loss_llm": 0.0, "grad_norm": 0.031105857342481613, "global_step": 3351, "epoch": 93, "lr": 0.001} +{"train_loss": 0.047854531556367874, "train_loss_bc": 0.047854531556367874, "train_loss_llm": 0.0, "grad_norm": 0.03742583096027374, "global_step": 3352, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04740535467863083, "train_loss_bc": 0.047389328479766846, "train_loss_llm": 0.0016026198863983154, "grad_norm": 0.02780156210064888, "global_step": 3353, "epoch": 93, "lr": 0.001} +{"train_loss": 0.043886490166187286, "train_loss_bc": 0.043886490166187286, "train_loss_llm": 0.0, "grad_norm": 0.028198331594467163, "global_step": 3354, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04206601530313492, "train_loss_bc": 0.04205027595162392, "train_loss_llm": 0.0015738606452941895, "grad_norm": 0.030811233446002007, "global_step": 3355, "epoch": 93, "lr": 0.001} +{"train_loss": 0.03684890642762184, "train_loss_bc": 0.03684890642762184, "train_loss_llm": 0.0, "grad_norm": 0.03241106495261192, "global_step": 3356, "epoch": 93, "lr": 0.001} +{"train_loss": 0.043569210916757584, "train_loss_bc": 0.04354024678468704, "train_loss_llm": 0.002896547317504883, "grad_norm": 0.01637231558561325, "global_step": 3357, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04028606787323952, "train_loss_bc": 0.04028606787323952, "train_loss_llm": 0.0, "grad_norm": 0.02158956788480282, "global_step": 3358, "epoch": 93, "lr": 0.001} +{"train_loss": 0.03830869495868683, "train_loss_bc": 0.03830869495868683, "train_loss_llm": 0.0, "grad_norm": 0.02496403269469738, "global_step": 3359, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04493088275194168, "train_loss_bc": 0.04493088275194168, "train_loss_llm": 0.0, "grad_norm": 0.033248353749513626, "global_step": 3360, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04519174247980118, "train_loss_bc": 0.04519174247980118, "train_loss_llm": 0.0, "grad_norm": 0.01612250693142414, "global_step": 3361, "epoch": 93, "lr": 0.001} +{"train_loss": 0.03968461602926254, "train_loss_bc": 0.03968461602926254, "train_loss_llm": 0.0, "grad_norm": 0.024867186322808266, "global_step": 3362, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04367460310459137, "train_loss_bc": 0.043656978756189346, "train_loss_llm": 0.0017625987529754639, "grad_norm": 0.038720808923244476, "global_step": 3363, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04508958011865616, "train_loss_bc": 0.04508958011865616, "train_loss_llm": 0.0, "grad_norm": 0.0455620251595974, "global_step": 3364, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04624668508768082, "train_loss_bc": 0.04624668508768082, "train_loss_llm": 0.0, "grad_norm": 0.020973268896341324, "global_step": 3365, "epoch": 93, "lr": 0.001} +{"train_loss": 0.05106312781572342, "train_loss_bc": 0.05104026943445206, "train_loss_llm": 0.0022857189178466797, "grad_norm": 0.03233162313699722, "global_step": 3366, "epoch": 93, "lr": 0.001} +{"train_loss": 0.049757204949855804, "train_loss_bc": 0.049757204949855804, "train_loss_llm": 0.0, "grad_norm": 0.0394611656665802, "global_step": 3367, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04764122515916824, "train_loss_bc": 0.04764122515916824, "train_loss_llm": 0.0, "grad_norm": 0.04886847361922264, "global_step": 3368, "epoch": 93, "lr": 0.001} +{"train_loss": 0.042827848345041275, "train_loss_bc": 0.042827848345041275, "train_loss_llm": 0.0, "grad_norm": 0.014692727476358414, "global_step": 3369, "epoch": 93, "lr": 0.001} +{"train_loss": 0.046728167682886124, "train_loss_bc": 0.046728167682886124, "train_loss_llm": 0.0, "grad_norm": 0.025527924299240112, "global_step": 3370, "epoch": 93, "lr": 0.001} +{"train_loss": 0.03812132775783539, "train_loss_bc": 0.03812132775783539, "train_loss_llm": 0.0, "grad_norm": 0.033276211470365524, "global_step": 3371, "epoch": 93, "lr": 0.001} +{"train_loss": 0.041431277990341187, "train_loss_bc": 0.041431277990341187, "train_loss_llm": 0.0, "grad_norm": 0.0436868891119957, "global_step": 3372, "epoch": 93, "lr": 0.001} +{"train_loss": 0.03826255351305008, "train_loss_bc": 0.038260143250226974, "train_loss_llm": 0.000241011381149292, "grad_norm": 0.015922553837299347, "global_step": 3373, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04244747385382652, "train_loss_bc": 0.0424453504383564, "train_loss_llm": 0.0002123415470123291, "grad_norm": 0.023257635533809662, "global_step": 3374, "epoch": 93, "lr": 0.001} +{"train_loss": 0.03959258273243904, "train_loss_bc": 0.03959258273243904, "train_loss_llm": 0.0, "grad_norm": 0.03390192613005638, "global_step": 3375, "epoch": 93, "lr": 0.001} +{"train_loss": 0.04133426025509834, "train_loss_bc": 0.04133426025509834, "train_loss_llm": 0.0, "grad_norm": 0.044515229761600494, "global_step": 3376, "epoch": 93, "lr": 0.001} +{"train_loss": 0.044648781418800354, "train_loss_bc": 0.044648781418800354, "train_loss_llm": 0.0, "grad_norm": 0.01664925552904606, "global_step": 3377, "epoch": 93, "lr": 0.001} +{"train_loss": 0.039942871779203415, "train_loss_bc": 0.039942871779203415, "train_loss_llm": 0.0, "grad_norm": 0.026887288317084312, "global_step": 3378, "epoch": 93, "lr": 0.001} +{"train_loss": 0.037235427647829056, "train_loss_bc": 0.03722183033823967, "train_loss_llm": 0.0013598203659057617, "grad_norm": 0.032829802483320236, "global_step": 3379, "epoch": 93, "lr": 0.001} +{"train_loss": 0.045917775481939316, "train_loss_bc": 0.045917775481939316, "train_loss_llm": 0.0, "grad_norm": 0.038033176213502884, "global_step": 3380, "epoch": 93, "lr": 0.001} +{"train_loss": 0.043569259345531464, "train_loss_bc": 0.04355666786432266, "train_loss_llm": 0.0012590587139129639, "grad_norm": 0.015929438173770905, "global_step": 3381, "epoch": 93, "lr": 0.001} +{"train_loss": 0.03611594811081886, "train_loss_bc": 0.03611594811081886, "train_loss_llm": 0.0, "grad_norm": 0.02190607599914074, "global_step": 3382, "epoch": 93, "lr": 0.001} +{"train_loss": 0.043874955736100674, "train_loss_bc": 0.05575679987668991, "train_loss_llm": 0.0, "grad_norm": 0.034447237849235535, "global_step": 3383, "epoch": 93, "lr": 0.001, "val_loss": 0.0524655282497406} +{"train_loss": 0.04919594153761864, "train_loss_bc": 0.04919594153761864, "train_loss_llm": 0.0, "grad_norm": 0.04005519300699234, "global_step": 3384, "epoch": 94, "lr": 0.001} +{"train_loss": 0.05665470287203789, "train_loss_bc": 0.05665470287203789, "train_loss_llm": 0.0, "grad_norm": 0.01733401231467724, "global_step": 3385, "epoch": 94, "lr": 0.001} +{"train_loss": 0.03690309822559357, "train_loss_bc": 0.03690309822559357, "train_loss_llm": 0.0, "grad_norm": 0.02507326379418373, "global_step": 3386, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04164893925189972, "train_loss_bc": 0.04164893925189972, "train_loss_llm": 0.0, "grad_norm": 0.034299012273550034, "global_step": 3387, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04463977366685867, "train_loss_bc": 0.044635940343141556, "train_loss_llm": 0.0003834068775177002, "grad_norm": 0.03977643698453903, "global_step": 3388, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04519300162792206, "train_loss_bc": 0.04517844691872597, "train_loss_llm": 0.0014553964138031006, "grad_norm": 0.01507621631026268, "global_step": 3389, "epoch": 94, "lr": 0.001} +{"train_loss": 0.02971307933330536, "train_loss_bc": 0.029693450778722763, "train_loss_llm": 0.001962810754776001, "grad_norm": 0.02550712786614895, "global_step": 3390, "epoch": 94, "lr": 0.001} +{"train_loss": 0.039905861020088196, "train_loss_bc": 0.039905861020088196, "train_loss_llm": 0.0, "grad_norm": 0.026942476630210876, "global_step": 3391, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04050752893090248, "train_loss_bc": 0.04050752893090248, "train_loss_llm": 0.0, "grad_norm": 0.035206250846385956, "global_step": 3392, "epoch": 94, "lr": 0.001} +{"train_loss": 0.0375794880092144, "train_loss_bc": 0.037562523037195206, "train_loss_llm": 0.0016964077949523926, "grad_norm": 0.017522672191262245, "global_step": 3393, "epoch": 94, "lr": 0.001} +{"train_loss": 0.03926753252744675, "train_loss_bc": 0.03926753252744675, "train_loss_llm": 0.0, "grad_norm": 0.024579646065831184, "global_step": 3394, "epoch": 94, "lr": 0.001} +{"train_loss": 0.042725399136543274, "train_loss_bc": 0.042725399136543274, "train_loss_llm": 0.0, "grad_norm": 0.036882106214761734, "global_step": 3395, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04155351221561432, "train_loss_bc": 0.04154946282505989, "train_loss_llm": 0.0004049241542816162, "grad_norm": 0.041442304849624634, "global_step": 3396, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04112016037106514, "train_loss_bc": 0.04112016037106514, "train_loss_llm": 0.0, "grad_norm": 0.016297033056616783, "global_step": 3397, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04178997874259949, "train_loss_bc": 0.04178997874259949, "train_loss_llm": 0.0, "grad_norm": 0.022651072591543198, "global_step": 3398, "epoch": 94, "lr": 0.001} +{"train_loss": 0.0450969859957695, "train_loss_bc": 0.0450969859957695, "train_loss_llm": 0.0, "grad_norm": 0.029684120789170265, "global_step": 3399, "epoch": 94, "lr": 0.001} +{"train_loss": 0.046994648873806, "train_loss_bc": 0.046994648873806, "train_loss_llm": 0.0, "grad_norm": 0.03484473377466202, "global_step": 3400, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04543920233845711, "train_loss_bc": 0.04543920233845711, "train_loss_llm": 0.0, "grad_norm": 0.016543081030249596, "global_step": 3401, "epoch": 94, "lr": 0.001} +{"train_loss": 0.02685355581343174, "train_loss_bc": 0.02684955485165119, "train_loss_llm": 0.00040012598037719727, "grad_norm": 0.019049540162086487, "global_step": 3402, "epoch": 94, "lr": 0.001} +{"train_loss": 0.0469120629131794, "train_loss_bc": 0.04687383398413658, "train_loss_llm": 0.003822833299636841, "grad_norm": 0.02728167362511158, "global_step": 3403, "epoch": 94, "lr": 0.001} +{"train_loss": 0.046636320650577545, "train_loss_bc": 0.046636320650577545, "train_loss_llm": 0.0, "grad_norm": 0.030465001240372658, "global_step": 3404, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04488865286111832, "train_loss_bc": 0.04487667605280876, "train_loss_llm": 0.00119781494140625, "grad_norm": 0.023993292823433876, "global_step": 3405, "epoch": 94, "lr": 0.001} +{"train_loss": 0.043098609894514084, "train_loss_bc": 0.043098609894514084, "train_loss_llm": 0.0, "grad_norm": 0.03354596346616745, "global_step": 3406, "epoch": 94, "lr": 0.001} +{"train_loss": 0.042432866990566254, "train_loss_bc": 0.042432866990566254, "train_loss_llm": 0.0, "grad_norm": 0.035148266702890396, "global_step": 3407, "epoch": 94, "lr": 0.001} +{"train_loss": 0.052613575011491776, "train_loss_bc": 0.052585210651159286, "train_loss_llm": 0.002836257219314575, "grad_norm": 0.035899557173252106, "global_step": 3408, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04204237088561058, "train_loss_bc": 0.042033616453409195, "train_loss_llm": 0.0008754730224609375, "grad_norm": 0.015065648593008518, "global_step": 3409, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04496743902564049, "train_loss_bc": 0.044950131326913834, "train_loss_llm": 0.0017306506633758545, "grad_norm": 0.02267182432115078, "global_step": 3410, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04348248243331909, "train_loss_bc": 0.04348248243331909, "train_loss_llm": 0.0, "grad_norm": 0.028651664033532143, "global_step": 3411, "epoch": 94, "lr": 0.001} +{"train_loss": 0.05431694537401199, "train_loss_bc": 0.05431694537401199, "train_loss_llm": 0.0, "grad_norm": 0.0374194011092186, "global_step": 3412, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04004207253456116, "train_loss_bc": 0.04004207253456116, "train_loss_llm": 0.0, "grad_norm": 0.016259435564279556, "global_step": 3413, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04769548773765564, "train_loss_bc": 0.04768431559205055, "train_loss_llm": 0.0011171698570251465, "grad_norm": 0.022063836455345154, "global_step": 3414, "epoch": 94, "lr": 0.001} +{"train_loss": 0.048582494258880615, "train_loss_bc": 0.04856898635625839, "train_loss_llm": 0.0013506412506103516, "grad_norm": 0.03303150832653046, "global_step": 3415, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04972593113780022, "train_loss_bc": 0.04972593113780022, "train_loss_llm": 0.0, "grad_norm": 0.04260213300585747, "global_step": 3416, "epoch": 94, "lr": 0.001} +{"train_loss": 0.044421713799238205, "train_loss_bc": 0.0443982295691967, "train_loss_llm": 0.002348393201828003, "grad_norm": 0.016934039071202278, "global_step": 3417, "epoch": 94, "lr": 0.001} +{"train_loss": 0.03709806501865387, "train_loss_bc": 0.03709806501865387, "train_loss_llm": 0.0, "grad_norm": 0.02420736290514469, "global_step": 3418, "epoch": 94, "lr": 0.001} +{"train_loss": 0.04374361064078079, "train_loss_bc": 0.05301731452345848, "train_loss_llm": 0.001318603754043579, "grad_norm": 0.03059634007513523, "global_step": 3419, "epoch": 94, "lr": 0.001, "val_loss": 0.05196192115545273} +{"train_loss": 0.03984897583723068, "train_loss_bc": 0.03984897583723068, "train_loss_llm": 0.0, "grad_norm": 0.0352020300924778, "global_step": 3420, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04728877171874046, "train_loss_bc": 0.04728877171874046, "train_loss_llm": 0.0, "grad_norm": 0.01792948506772518, "global_step": 3421, "epoch": 95, "lr": 0.001} +{"train_loss": 0.03847980126738548, "train_loss_bc": 0.03847980126738548, "train_loss_llm": 0.0, "grad_norm": 0.027304228395223618, "global_step": 3422, "epoch": 95, "lr": 0.001} +{"train_loss": 0.037224575877189636, "train_loss_bc": 0.03720008209347725, "train_loss_llm": 0.002449333667755127, "grad_norm": 0.03513074293732643, "global_step": 3423, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04470003768801689, "train_loss_bc": 0.04467825964093208, "train_loss_llm": 0.002177983522415161, "grad_norm": 0.042876776307821274, "global_step": 3424, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04973173141479492, "train_loss_bc": 0.04972502589225769, "train_loss_llm": 0.0006706416606903076, "grad_norm": 0.01782151125371456, "global_step": 3425, "epoch": 95, "lr": 0.001} +{"train_loss": 0.0437336340546608, "train_loss_bc": 0.04372826963663101, "train_loss_llm": 0.0005363225936889648, "grad_norm": 0.025168556720018387, "global_step": 3426, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04787498712539673, "train_loss_bc": 0.04786950349807739, "train_loss_llm": 0.0005484521389007568, "grad_norm": 0.03086845390498638, "global_step": 3427, "epoch": 95, "lr": 0.001} +{"train_loss": 0.048211999237537384, "train_loss_bc": 0.04819544777274132, "train_loss_llm": 0.0016551017761230469, "grad_norm": 0.032880283892154694, "global_step": 3428, "epoch": 95, "lr": 0.001} +{"train_loss": 0.046065282076597214, "train_loss_bc": 0.04604317620396614, "train_loss_llm": 0.002210646867752075, "grad_norm": 0.020072396844625473, "global_step": 3429, "epoch": 95, "lr": 0.001} +{"train_loss": 0.03867670148611069, "train_loss_bc": 0.038664333522319794, "train_loss_llm": 0.0012368857860565186, "grad_norm": 0.02739163674414158, "global_step": 3430, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04271281883120537, "train_loss_bc": 0.04271281883120537, "train_loss_llm": 0.0, "grad_norm": 0.033071327954530716, "global_step": 3431, "epoch": 95, "lr": 0.001} +{"train_loss": 0.0487883985042572, "train_loss_bc": 0.0487883985042572, "train_loss_llm": 0.0, "grad_norm": 0.039945900440216064, "global_step": 3432, "epoch": 95, "lr": 0.001} +{"train_loss": 0.03952239826321602, "train_loss_bc": 0.03952239826321602, "train_loss_llm": 0.0, "grad_norm": 0.018503744155168533, "global_step": 3433, "epoch": 95, "lr": 0.001} +{"train_loss": 0.047950372099876404, "train_loss_bc": 0.047950372099876404, "train_loss_llm": 0.0, "grad_norm": 0.02741086110472679, "global_step": 3434, "epoch": 95, "lr": 0.001} +{"train_loss": 0.03392918407917023, "train_loss_bc": 0.03392918407917023, "train_loss_llm": 0.0, "grad_norm": 0.03630951792001724, "global_step": 3435, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04246994107961655, "train_loss_bc": 0.04245278984308243, "train_loss_llm": 0.0017152130603790283, "grad_norm": 0.041958048939704895, "global_step": 3436, "epoch": 95, "lr": 0.001} +{"train_loss": 0.047936685383319855, "train_loss_bc": 0.04793250188231468, "train_loss_llm": 0.0004185140132904053, "grad_norm": 0.016265016049146652, "global_step": 3437, "epoch": 95, "lr": 0.001} +{"train_loss": 0.041876643896102905, "train_loss_bc": 0.041876643896102905, "train_loss_llm": 0.0, "grad_norm": 0.025350896641612053, "global_step": 3438, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04553888365626335, "train_loss_bc": 0.04553888365626335, "train_loss_llm": 0.0, "grad_norm": 0.032146770507097244, "global_step": 3439, "epoch": 95, "lr": 0.001} +{"train_loss": 0.0479692779481411, "train_loss_bc": 0.04793218895792961, "train_loss_llm": 0.0037088990211486816, "grad_norm": 0.03525344654917717, "global_step": 3440, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04803580790758133, "train_loss_bc": 0.04803580790758133, "train_loss_llm": 0.0, "grad_norm": 0.02224009670317173, "global_step": 3441, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04969486594200134, "train_loss_bc": 0.04969486594200134, "train_loss_llm": 0.0, "grad_norm": 0.028482142835855484, "global_step": 3442, "epoch": 95, "lr": 0.001} +{"train_loss": 0.03842736408114433, "train_loss_bc": 0.03842736408114433, "train_loss_llm": 0.0, "grad_norm": 0.03756856545805931, "global_step": 3443, "epoch": 95, "lr": 0.001} +{"train_loss": 0.040286075323820114, "train_loss_bc": 0.040286075323820114, "train_loss_llm": 0.0, "grad_norm": 0.042935434728860855, "global_step": 3444, "epoch": 95, "lr": 0.001} +{"train_loss": 0.042342670261859894, "train_loss_bc": 0.04231719300150871, "train_loss_llm": 0.002547621726989746, "grad_norm": 0.016866419464349747, "global_step": 3445, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04242333024740219, "train_loss_bc": 0.04242333024740219, "train_loss_llm": 0.0, "grad_norm": 0.022152487188577652, "global_step": 3446, "epoch": 95, "lr": 0.001} +{"train_loss": 0.03869261220097542, "train_loss_bc": 0.03869261220097542, "train_loss_llm": 0.0, "grad_norm": 0.032155752182006836, "global_step": 3447, "epoch": 95, "lr": 0.001} +{"train_loss": 0.049139246344566345, "train_loss_bc": 0.04913716018199921, "train_loss_llm": 0.0002084672451019287, "grad_norm": 0.03639109060168266, "global_step": 3448, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04494590684771538, "train_loss_bc": 0.04494158551096916, "train_loss_llm": 0.00043201446533203125, "grad_norm": 0.01790975034236908, "global_step": 3449, "epoch": 95, "lr": 0.001} +{"train_loss": 0.03975585103034973, "train_loss_bc": 0.0397508479654789, "train_loss_llm": 0.0005002319812774658, "grad_norm": 0.023142455145716667, "global_step": 3450, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04912913963198662, "train_loss_bc": 0.04912913963198662, "train_loss_llm": 0.0, "grad_norm": 0.03357703238725662, "global_step": 3451, "epoch": 95, "lr": 0.001} +{"train_loss": 0.035549186170101166, "train_loss_bc": 0.03553753346204758, "train_loss_llm": 0.0011651217937469482, "grad_norm": 0.039630137383937836, "global_step": 3452, "epoch": 95, "lr": 0.001} +{"train_loss": 0.03777339681982994, "train_loss_bc": 0.03777339681982994, "train_loss_llm": 0.0, "grad_norm": 0.012823358178138733, "global_step": 3453, "epoch": 95, "lr": 0.001} +{"train_loss": 0.049002666026353836, "train_loss_bc": 0.049002666026353836, "train_loss_llm": 0.0, "grad_norm": 0.022920558229088783, "global_step": 3454, "epoch": 95, "lr": 0.001} +{"train_loss": 0.04357005055579874, "train_loss_bc": 0.04278524965047836, "train_loss_llm": 0.0007349252700805664, "grad_norm": 0.028013942763209343, "global_step": 3455, "epoch": 95, "lr": 0.001, "train/cumulative_reward": 83.2946369447931, "train/mean_score": 17.92371526995783, "train/success_rate": 0.4, "test/cumulative_reward": 81.40201277614119, "test/mean_score": 9.613018098603641, "test/success_rate": 0.2, "val_loss": 0.0523238480091095, "train_action_mse_error": 0.04772220924496651} +{"train_loss": 0.04040824621915817, "train_loss_bc": 0.04040824621915817, "train_loss_llm": 0.0, "grad_norm": 0.031838130205869675, "global_step": 3456, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04528944566845894, "train_loss_bc": 0.04528944566845894, "train_loss_llm": 0.0, "grad_norm": 0.018675843253731728, "global_step": 3457, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04818914830684662, "train_loss_bc": 0.04818914830684662, "train_loss_llm": 0.0, "grad_norm": 0.024910034611821175, "global_step": 3458, "epoch": 96, "lr": 0.001} +{"train_loss": 0.044748157262802124, "train_loss_bc": 0.044748157262802124, "train_loss_llm": 0.0, "grad_norm": 0.03739747032523155, "global_step": 3459, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04864045977592468, "train_loss_bc": 0.04864045977592468, "train_loss_llm": 0.0, "grad_norm": 0.037498656660318375, "global_step": 3460, "epoch": 96, "lr": 0.001} +{"train_loss": 0.05000951141119003, "train_loss_bc": 0.05000951141119003, "train_loss_llm": 0.0, "grad_norm": 0.015996800735592842, "global_step": 3461, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04665501415729523, "train_loss_bc": 0.04663683846592903, "train_loss_llm": 0.0018174052238464355, "grad_norm": 0.025948263704776764, "global_step": 3462, "epoch": 96, "lr": 0.001} +{"train_loss": 0.042635492980480194, "train_loss_bc": 0.042635492980480194, "train_loss_llm": 0.0, "grad_norm": 0.029889069497585297, "global_step": 3463, "epoch": 96, "lr": 0.001} +{"train_loss": 0.048998475074768066, "train_loss_bc": 0.048998475074768066, "train_loss_llm": 0.0, "grad_norm": 0.04618833214044571, "global_step": 3464, "epoch": 96, "lr": 0.001} +{"train_loss": 0.033728599548339844, "train_loss_bc": 0.033728599548339844, "train_loss_llm": 0.0, "grad_norm": 0.012796885333955288, "global_step": 3465, "epoch": 96, "lr": 0.001} +{"train_loss": 0.035974036902189255, "train_loss_bc": 0.035974036902189255, "train_loss_llm": 0.0, "grad_norm": 0.018759477883577347, "global_step": 3466, "epoch": 96, "lr": 0.001} +{"train_loss": 0.045874811708927155, "train_loss_bc": 0.045874811708927155, "train_loss_llm": 0.0, "grad_norm": 0.025206677615642548, "global_step": 3467, "epoch": 96, "lr": 0.001} +{"train_loss": 0.03808784484863281, "train_loss_bc": 0.03808179497718811, "train_loss_llm": 0.0006051063537597656, "grad_norm": 0.028707366436719894, "global_step": 3468, "epoch": 96, "lr": 0.001} +{"train_loss": 0.044385235756635666, "train_loss_bc": 0.044385235756635666, "train_loss_llm": 0.0, "grad_norm": 0.01937776245176792, "global_step": 3469, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04437302052974701, "train_loss_bc": 0.04437302052974701, "train_loss_llm": 0.0, "grad_norm": 0.024319496005773544, "global_step": 3470, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04670790955424309, "train_loss_bc": 0.04669463634490967, "train_loss_llm": 0.0013271868228912354, "grad_norm": 0.028187448158860207, "global_step": 3471, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04120425134897232, "train_loss_bc": 0.04120425134897232, "train_loss_llm": 0.0, "grad_norm": 0.038664303719997406, "global_step": 3472, "epoch": 96, "lr": 0.001} +{"train_loss": 0.046712834388017654, "train_loss_bc": 0.046712834388017654, "train_loss_llm": 0.0, "grad_norm": 0.016670595854520798, "global_step": 3473, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04427957907319069, "train_loss_bc": 0.04426255449652672, "train_loss_llm": 0.0017025172710418701, "grad_norm": 0.022312916815280914, "global_step": 3474, "epoch": 96, "lr": 0.001} +{"train_loss": 0.05052296817302704, "train_loss_bc": 0.05051184073090553, "train_loss_llm": 0.0011126697063446045, "grad_norm": 0.03436989709734917, "global_step": 3475, "epoch": 96, "lr": 0.001} +{"train_loss": 0.043814901262521744, "train_loss_bc": 0.043814901262521744, "train_loss_llm": 0.0, "grad_norm": 0.04283629357814789, "global_step": 3476, "epoch": 96, "lr": 0.001} +{"train_loss": 0.03294731676578522, "train_loss_bc": 0.03294731676578522, "train_loss_llm": 0.0, "grad_norm": 0.01833534799516201, "global_step": 3477, "epoch": 96, "lr": 0.001} +{"train_loss": 0.046562474220991135, "train_loss_bc": 0.046562474220991135, "train_loss_llm": 0.0, "grad_norm": 0.029020750895142555, "global_step": 3478, "epoch": 96, "lr": 0.001} +{"train_loss": 0.042964715510606766, "train_loss_bc": 0.04293540492653847, "train_loss_llm": 0.0029311180114746094, "grad_norm": 0.03785859793424606, "global_step": 3479, "epoch": 96, "lr": 0.001} +{"train_loss": 0.041473858058452606, "train_loss_bc": 0.041455842554569244, "train_loss_llm": 0.0018016397953033447, "grad_norm": 0.045305684208869934, "global_step": 3480, "epoch": 96, "lr": 0.001} +{"train_loss": 0.034490883350372314, "train_loss_bc": 0.034476783126592636, "train_loss_llm": 0.0014100968837738037, "grad_norm": 0.01453483197838068, "global_step": 3481, "epoch": 96, "lr": 0.001} +{"train_loss": 0.05413096770644188, "train_loss_bc": 0.05410224199295044, "train_loss_llm": 0.002872556447982788, "grad_norm": 0.026879118755459785, "global_step": 3482, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04192814230918884, "train_loss_bc": 0.04192814230918884, "train_loss_llm": 0.0, "grad_norm": 0.031093604862689972, "global_step": 3483, "epoch": 96, "lr": 0.001} +{"train_loss": 0.033975061029195786, "train_loss_bc": 0.03396916016936302, "train_loss_llm": 0.0005900263786315918, "grad_norm": 0.03469974920153618, "global_step": 3484, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04469494894146919, "train_loss_bc": 0.04469494894146919, "train_loss_llm": 0.0, "grad_norm": 0.014396379701793194, "global_step": 3485, "epoch": 96, "lr": 0.001} +{"train_loss": 0.042719196528196335, "train_loss_bc": 0.042719196528196335, "train_loss_llm": 0.0, "grad_norm": 0.023716824129223824, "global_step": 3486, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04089262709021568, "train_loss_bc": 0.04087986797094345, "train_loss_llm": 0.0012757480144500732, "grad_norm": 0.030829407274723053, "global_step": 3487, "epoch": 96, "lr": 0.001} +{"train_loss": 0.037488847970962524, "train_loss_bc": 0.037484560161828995, "train_loss_llm": 0.00042876601219177246, "grad_norm": 0.03744882717728615, "global_step": 3488, "epoch": 96, "lr": 0.001} +{"train_loss": 0.050530094653367996, "train_loss_bc": 0.050530094653367996, "train_loss_llm": 0.0, "grad_norm": 0.022891851142048836, "global_step": 3489, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04434986785054207, "train_loss_bc": 0.04433910548686981, "train_loss_llm": 0.0010762214660644531, "grad_norm": 0.03253205493092537, "global_step": 3490, "epoch": 96, "lr": 0.001} +{"train_loss": 0.04365524494399627, "train_loss_bc": 0.05119987204670906, "train_loss_llm": 0.0, "grad_norm": 0.04030413553118706, "global_step": 3491, "epoch": 96, "lr": 0.001, "val_loss": 0.05192633718252182} +{"train_loss": 0.03555664047598839, "train_loss_bc": 0.03555664047598839, "train_loss_llm": 0.0, "grad_norm": 0.05020453408360481, "global_step": 3492, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04865816608071327, "train_loss_bc": 0.048639025539159775, "train_loss_llm": 0.001913994550704956, "grad_norm": 0.01955007202923298, "global_step": 3493, "epoch": 97, "lr": 0.001} +{"train_loss": 0.039262138307094574, "train_loss_bc": 0.039262138307094574, "train_loss_llm": 0.0, "grad_norm": 0.024444295093417168, "global_step": 3494, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04352424293756485, "train_loss_bc": 0.04352424293756485, "train_loss_llm": 0.0, "grad_norm": 0.03720296174287796, "global_step": 3495, "epoch": 97, "lr": 0.001} +{"train_loss": 0.042921364307403564, "train_loss_bc": 0.042921364307403564, "train_loss_llm": 0.0, "grad_norm": 0.03920771926641464, "global_step": 3496, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04894515872001648, "train_loss_bc": 0.04894515872001648, "train_loss_llm": 0.0, "grad_norm": 0.020119840279221535, "global_step": 3497, "epoch": 97, "lr": 0.001} +{"train_loss": 0.043927572667598724, "train_loss_bc": 0.04390651732683182, "train_loss_llm": 0.0021053850650787354, "grad_norm": 0.027414001524448395, "global_step": 3498, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04560113698244095, "train_loss_bc": 0.04560113698244095, "train_loss_llm": 0.0, "grad_norm": 0.03550698235630989, "global_step": 3499, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04895704984664917, "train_loss_bc": 0.04894515126943588, "train_loss_llm": 0.0011900365352630615, "grad_norm": 0.04272053763270378, "global_step": 3500, "epoch": 97, "lr": 0.001} +{"train_loss": 0.050444360822439194, "train_loss_bc": 0.050444360822439194, "train_loss_llm": 0.0, "grad_norm": 0.016863111406564713, "global_step": 3501, "epoch": 97, "lr": 0.001} +{"train_loss": 0.0390147902071476, "train_loss_bc": 0.03900320082902908, "train_loss_llm": 0.0011588633060455322, "grad_norm": 0.029288453981280327, "global_step": 3502, "epoch": 97, "lr": 0.001} +{"train_loss": 0.03969411179423332, "train_loss_bc": 0.03969411179423332, "train_loss_llm": 0.0, "grad_norm": 0.03766332194209099, "global_step": 3503, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04016740247607231, "train_loss_bc": 0.04016740247607231, "train_loss_llm": 0.0, "grad_norm": 0.042477261275053024, "global_step": 3504, "epoch": 97, "lr": 0.001} +{"train_loss": 0.035432975739240646, "train_loss_bc": 0.03540930151939392, "train_loss_llm": 0.002367377281188965, "grad_norm": 0.013662378303706646, "global_step": 3505, "epoch": 97, "lr": 0.001} +{"train_loss": 0.05329277738928795, "train_loss_bc": 0.05329277738928795, "train_loss_llm": 0.0, "grad_norm": 0.02276095747947693, "global_step": 3506, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04876647889614105, "train_loss_bc": 0.04876647889614105, "train_loss_llm": 0.0, "grad_norm": 0.03139230236411095, "global_step": 3507, "epoch": 97, "lr": 0.001} +{"train_loss": 0.045401208102703094, "train_loss_bc": 0.04540056735277176, "train_loss_llm": 6.404519081115723e-05, "grad_norm": 0.034759778529405594, "global_step": 3508, "epoch": 97, "lr": 0.001} +{"train_loss": 0.0451233796775341, "train_loss_bc": 0.0451233796775341, "train_loss_llm": 0.0, "grad_norm": 0.01870730146765709, "global_step": 3509, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04397033527493477, "train_loss_bc": 0.04397033527493477, "train_loss_llm": 0.0, "grad_norm": 0.028439868241548538, "global_step": 3510, "epoch": 97, "lr": 0.001} +{"train_loss": 0.043244618922472, "train_loss_bc": 0.043241098523139954, "train_loss_llm": 0.0003522038459777832, "grad_norm": 0.0410827174782753, "global_step": 3511, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04888881370425224, "train_loss_bc": 0.04888881370425224, "train_loss_llm": 0.0, "grad_norm": 0.045462802052497864, "global_step": 3512, "epoch": 97, "lr": 0.001} +{"train_loss": 0.03988755866885185, "train_loss_bc": 0.03988755866885185, "train_loss_llm": 0.0, "grad_norm": 0.01783653162419796, "global_step": 3513, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04188248887658119, "train_loss_bc": 0.041880879551172256, "train_loss_llm": 0.00016075372695922852, "grad_norm": 0.023455403745174408, "global_step": 3514, "epoch": 97, "lr": 0.001} +{"train_loss": 0.03346984088420868, "train_loss_bc": 0.03346984088420868, "train_loss_llm": 0.0, "grad_norm": 0.031076887622475624, "global_step": 3515, "epoch": 97, "lr": 0.001} +{"train_loss": 0.03691604360938072, "train_loss_bc": 0.03691604360938072, "train_loss_llm": 0.0, "grad_norm": 0.03766537457704544, "global_step": 3516, "epoch": 97, "lr": 0.001} +{"train_loss": 0.05016227811574936, "train_loss_bc": 0.05016227811574936, "train_loss_llm": 0.0, "grad_norm": 0.018155064433813095, "global_step": 3517, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04963225871324539, "train_loss_bc": 0.04963119328022003, "train_loss_llm": 0.00010666251182556152, "grad_norm": 0.027170181274414062, "global_step": 3518, "epoch": 97, "lr": 0.001} +{"train_loss": 0.046355608850717545, "train_loss_bc": 0.046355608850717545, "train_loss_llm": 0.0, "grad_norm": 0.037783704698085785, "global_step": 3519, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04816338047385216, "train_loss_bc": 0.04816338047385216, "train_loss_llm": 0.0, "grad_norm": 0.051442552357912064, "global_step": 3520, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04161501303315163, "train_loss_bc": 0.04161501303315163, "train_loss_llm": 0.0, "grad_norm": 0.017592277377843857, "global_step": 3521, "epoch": 97, "lr": 0.001} +{"train_loss": 0.040426500141620636, "train_loss_bc": 0.040426500141620636, "train_loss_llm": 0.0, "grad_norm": 0.029847467318177223, "global_step": 3522, "epoch": 97, "lr": 0.001} +{"train_loss": 0.03658802807331085, "train_loss_bc": 0.03658802807331085, "train_loss_llm": 0.0, "grad_norm": 0.03538472205400467, "global_step": 3523, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04676280543208122, "train_loss_bc": 0.046760618686676025, "train_loss_llm": 0.0002187192440032959, "grad_norm": 0.04768427461385727, "global_step": 3524, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04375700280070305, "train_loss_bc": 0.04374207928776741, "train_loss_llm": 0.0014922022819519043, "grad_norm": 0.01918485201895237, "global_step": 3525, "epoch": 97, "lr": 0.001} +{"train_loss": 0.036676354706287384, "train_loss_bc": 0.03666836395859718, "train_loss_llm": 0.0007992386817932129, "grad_norm": 0.032215263694524765, "global_step": 3526, "epoch": 97, "lr": 0.001} +{"train_loss": 0.04367158065239588, "train_loss_bc": 0.04907023161649704, "train_loss_llm": 0.0016786754131317139, "grad_norm": 0.04054122045636177, "global_step": 3527, "epoch": 97, "lr": 0.001, "val_loss": 0.05189692601561546} +{"train_loss": 0.03623099997639656, "train_loss_bc": 0.03623099997639656, "train_loss_llm": 0.0, "grad_norm": 0.04604380577802658, "global_step": 3528, "epoch": 98, "lr": 0.001} +{"train_loss": 0.0400419645011425, "train_loss_bc": 0.0400419645011425, "train_loss_llm": 0.0, "grad_norm": 0.015440945513546467, "global_step": 3529, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04986948519945145, "train_loss_bc": 0.04986948519945145, "train_loss_llm": 0.0, "grad_norm": 0.023570548743009567, "global_step": 3530, "epoch": 98, "lr": 0.001} +{"train_loss": 0.0426238551735878, "train_loss_bc": 0.0426238551735878, "train_loss_llm": 0.0, "grad_norm": 0.02939540334045887, "global_step": 3531, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04454085975885391, "train_loss_bc": 0.04454085975885391, "train_loss_llm": 0.0, "grad_norm": 0.03654138371348381, "global_step": 3532, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04491870105266571, "train_loss_bc": 0.04490699991583824, "train_loss_llm": 0.0011700093746185303, "grad_norm": 0.021710051223635674, "global_step": 3533, "epoch": 98, "lr": 0.001} +{"train_loss": 0.042850688099861145, "train_loss_bc": 0.042850688099861145, "train_loss_llm": 0.0, "grad_norm": 0.024307742714881897, "global_step": 3534, "epoch": 98, "lr": 0.001} +{"train_loss": 0.042111676186323166, "train_loss_bc": 0.042111676186323166, "train_loss_llm": 0.0, "grad_norm": 0.029514174908399582, "global_step": 3535, "epoch": 98, "lr": 0.001} +{"train_loss": 0.05016731843352318, "train_loss_bc": 0.05016731843352318, "train_loss_llm": 0.0, "grad_norm": 0.034535396844148636, "global_step": 3536, "epoch": 98, "lr": 0.001} +{"train_loss": 0.03894347324967384, "train_loss_bc": 0.03891948238015175, "train_loss_llm": 0.002399057149887085, "grad_norm": 0.013014065101742744, "global_step": 3537, "epoch": 98, "lr": 0.001} +{"train_loss": 0.053788989782333374, "train_loss_bc": 0.053788989782333374, "train_loss_llm": 0.0, "grad_norm": 0.022835111245512962, "global_step": 3538, "epoch": 98, "lr": 0.001} +{"train_loss": 0.03822694718837738, "train_loss_bc": 0.03820718824863434, "train_loss_llm": 0.0019759833812713623, "grad_norm": 0.02977490797638893, "global_step": 3539, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04921962693333626, "train_loss_bc": 0.04917343333363533, "train_loss_llm": 0.004619181156158447, "grad_norm": 0.03706486150622368, "global_step": 3540, "epoch": 98, "lr": 0.001} +{"train_loss": 0.050841450691223145, "train_loss_bc": 0.050841450691223145, "train_loss_llm": 0.0, "grad_norm": 0.015466747805476189, "global_step": 3541, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04189984127879143, "train_loss_bc": 0.04189984127879143, "train_loss_llm": 0.0, "grad_norm": 0.027163850143551826, "global_step": 3542, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04509309306740761, "train_loss_bc": 0.04509309306740761, "train_loss_llm": 0.0, "grad_norm": 0.032556019723415375, "global_step": 3543, "epoch": 98, "lr": 0.001} +{"train_loss": 0.044647227972745895, "train_loss_bc": 0.044647227972745895, "train_loss_llm": 0.0, "grad_norm": 0.04019802436232567, "global_step": 3544, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04091191291809082, "train_loss_bc": 0.04091191291809082, "train_loss_llm": 0.0, "grad_norm": 0.016737276688218117, "global_step": 3545, "epoch": 98, "lr": 0.001} +{"train_loss": 0.040241654962301254, "train_loss_bc": 0.040241654962301254, "train_loss_llm": 0.0, "grad_norm": 0.025685634464025497, "global_step": 3546, "epoch": 98, "lr": 0.001} +{"train_loss": 0.043069787323474884, "train_loss_bc": 0.04306896775960922, "train_loss_llm": 8.20457935333252e-05, "grad_norm": 0.03859773650765419, "global_step": 3547, "epoch": 98, "lr": 0.001} +{"train_loss": 0.05057324096560478, "train_loss_bc": 0.05057324096560478, "train_loss_llm": 0.0, "grad_norm": 0.04394879192113876, "global_step": 3548, "epoch": 98, "lr": 0.001} +{"train_loss": 0.0400581881403923, "train_loss_bc": 0.0400581881403923, "train_loss_llm": 0.0, "grad_norm": 0.017841342836618423, "global_step": 3549, "epoch": 98, "lr": 0.001} +{"train_loss": 0.039245519787073135, "train_loss_bc": 0.039245519787073135, "train_loss_llm": 0.0, "grad_norm": 0.02990717440843582, "global_step": 3550, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04120589792728424, "train_loss_bc": 0.04120589792728424, "train_loss_llm": 0.0, "grad_norm": 0.03369414806365967, "global_step": 3551, "epoch": 98, "lr": 0.001} +{"train_loss": 0.045158952474594116, "train_loss_bc": 0.045158952474594116, "train_loss_llm": 0.0, "grad_norm": 0.03832812234759331, "global_step": 3552, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04505693539977074, "train_loss_bc": 0.045056555420160294, "train_loss_llm": 3.796815872192383e-05, "grad_norm": 0.016437841579318047, "global_step": 3553, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04244472458958626, "train_loss_bc": 0.04244472458958626, "train_loss_llm": 0.0, "grad_norm": 0.023612048476934433, "global_step": 3554, "epoch": 98, "lr": 0.001} +{"train_loss": 0.0407361276447773, "train_loss_bc": 0.0407361276447773, "train_loss_llm": 0.0, "grad_norm": 0.02712315134704113, "global_step": 3555, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04398385435342789, "train_loss_bc": 0.0439557284116745, "train_loss_llm": 0.002812594175338745, "grad_norm": 0.03451419249176979, "global_step": 3556, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04536081477999687, "train_loss_bc": 0.04535132646560669, "train_loss_llm": 0.0009487569332122803, "grad_norm": 0.020136281847953796, "global_step": 3557, "epoch": 98, "lr": 0.001} +{"train_loss": 0.03339197114109993, "train_loss_bc": 0.03339197114109993, "train_loss_llm": 0.0, "grad_norm": 0.02761225216090679, "global_step": 3558, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04818546772003174, "train_loss_bc": 0.04818546772003174, "train_loss_llm": 0.0, "grad_norm": 0.04218149557709694, "global_step": 3559, "epoch": 98, "lr": 0.001} +{"train_loss": 0.037987083196640015, "train_loss_bc": 0.037987083196640015, "train_loss_llm": 0.0, "grad_norm": 0.051442548632621765, "global_step": 3560, "epoch": 98, "lr": 0.001} +{"train_loss": 0.044551555067300797, "train_loss_bc": 0.044548649340867996, "train_loss_llm": 0.0002905726432800293, "grad_norm": 0.016926366835832596, "global_step": 3561, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04643234238028526, "train_loss_bc": 0.04643234238028526, "train_loss_llm": 0.0, "grad_norm": 0.0247463658452034, "global_step": 3562, "epoch": 98, "lr": 0.001} +{"train_loss": 0.04352273243582911, "train_loss_bc": 0.04215922951698303, "train_loss_llm": 0.004690974950790405, "grad_norm": 0.0337698869407177, "global_step": 3563, "epoch": 98, "lr": 0.001, "val_loss": 0.05221228674054146} +{"train_loss": 0.04263019561767578, "train_loss_bc": 0.04263019561767578, "train_loss_llm": 0.0, "grad_norm": 0.04246984422206879, "global_step": 3564, "epoch": 99, "lr": 0.001} +{"train_loss": 0.041378118097782135, "train_loss_bc": 0.041378118097782135, "train_loss_llm": 0.0, "grad_norm": 0.015816254541277885, "global_step": 3565, "epoch": 99, "lr": 0.001} +{"train_loss": 0.03921070694923401, "train_loss_bc": 0.03921070694923401, "train_loss_llm": 0.0, "grad_norm": 0.022156162187457085, "global_step": 3566, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04418399557471275, "train_loss_bc": 0.04418399557471275, "train_loss_llm": 0.0, "grad_norm": 0.03089514560997486, "global_step": 3567, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04577163606882095, "train_loss_bc": 0.04576600715517998, "train_loss_llm": 0.0005628764629364014, "grad_norm": 0.03660105913877487, "global_step": 3568, "epoch": 99, "lr": 0.001} +{"train_loss": 0.037517208606004715, "train_loss_bc": 0.037517208606004715, "train_loss_llm": 0.0, "grad_norm": 0.017010416835546494, "global_step": 3569, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04120907187461853, "train_loss_bc": 0.04120907187461853, "train_loss_llm": 0.0, "grad_norm": 0.024689538404345512, "global_step": 3570, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04735827073454857, "train_loss_bc": 0.04735827073454857, "train_loss_llm": 0.0, "grad_norm": 0.030059488490223885, "global_step": 3571, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04522578790783882, "train_loss_bc": 0.04522578790783882, "train_loss_llm": 0.0, "grad_norm": 0.0358254536986351, "global_step": 3572, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04508204385638237, "train_loss_bc": 0.04508204385638237, "train_loss_llm": 0.0, "grad_norm": 0.015586157329380512, "global_step": 3573, "epoch": 99, "lr": 0.001} +{"train_loss": 0.03388991951942444, "train_loss_bc": 0.03388119488954544, "train_loss_llm": 0.0008726418018341064, "grad_norm": 0.020141854882240295, "global_step": 3574, "epoch": 99, "lr": 0.001} +{"train_loss": 0.05056925490498543, "train_loss_bc": 0.05056925490498543, "train_loss_llm": 0.0, "grad_norm": 0.02851954847574234, "global_step": 3575, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04983582720160484, "train_loss_bc": 0.04983582720160484, "train_loss_llm": 0.0, "grad_norm": 0.03955363854765892, "global_step": 3576, "epoch": 99, "lr": 0.001} +{"train_loss": 0.045319247990846634, "train_loss_bc": 0.045319247990846634, "train_loss_llm": 0.0, "grad_norm": 0.019632574170827866, "global_step": 3577, "epoch": 99, "lr": 0.001} +{"train_loss": 0.046558067202568054, "train_loss_bc": 0.046558067202568054, "train_loss_llm": 0.0, "grad_norm": 0.03033805452287197, "global_step": 3578, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04262937977910042, "train_loss_bc": 0.0425923727452755, "train_loss_llm": 0.003700852394104004, "grad_norm": 0.03641781955957413, "global_step": 3579, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04701681062579155, "train_loss_bc": 0.04700075089931488, "train_loss_llm": 0.0016058683395385742, "grad_norm": 0.042740389704704285, "global_step": 3580, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04071293771266937, "train_loss_bc": 0.04070861637592316, "train_loss_llm": 0.00043210387229919434, "grad_norm": 0.014727246947586536, "global_step": 3581, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04374009743332863, "train_loss_bc": 0.04374009743332863, "train_loss_llm": 0.0, "grad_norm": 0.020862305536866188, "global_step": 3582, "epoch": 99, "lr": 0.001} +{"train_loss": 0.0422101765871048, "train_loss_bc": 0.0422101765871048, "train_loss_llm": 0.0, "grad_norm": 0.0294186994433403, "global_step": 3583, "epoch": 99, "lr": 0.001} +{"train_loss": 0.042463332414627075, "train_loss_bc": 0.04245409741997719, "train_loss_llm": 0.0009236633777618408, "grad_norm": 0.037159163504838943, "global_step": 3584, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04102989658713341, "train_loss_bc": 0.04102989658713341, "train_loss_llm": 0.0, "grad_norm": 0.01522754691541195, "global_step": 3585, "epoch": 99, "lr": 0.001} +{"train_loss": 0.05115065723657608, "train_loss_bc": 0.05115065723657608, "train_loss_llm": 0.0, "grad_norm": 0.02291853539645672, "global_step": 3586, "epoch": 99, "lr": 0.001} +{"train_loss": 0.03942231088876724, "train_loss_bc": 0.03942231088876724, "train_loss_llm": 0.0, "grad_norm": 0.022465461865067482, "global_step": 3587, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04263918846845627, "train_loss_bc": 0.042634330689907074, "train_loss_llm": 0.0004856288433074951, "grad_norm": 0.030980492010712624, "global_step": 3588, "epoch": 99, "lr": 0.001} +{"train_loss": 0.041796598583459854, "train_loss_bc": 0.041796598583459854, "train_loss_llm": 0.0, "grad_norm": 0.015176600776612759, "global_step": 3589, "epoch": 99, "lr": 0.001} +{"train_loss": 0.052432380616664886, "train_loss_bc": 0.052432380616664886, "train_loss_llm": 0.0, "grad_norm": 0.0212082676589489, "global_step": 3590, "epoch": 99, "lr": 0.001} +{"train_loss": 0.038782451301813126, "train_loss_bc": 0.038782451301813126, "train_loss_llm": 0.0, "grad_norm": 0.026148563250899315, "global_step": 3591, "epoch": 99, "lr": 0.001} +{"train_loss": 0.03765137493610382, "train_loss_bc": 0.03765137493610382, "train_loss_llm": 0.0, "grad_norm": 0.03117811679840088, "global_step": 3592, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04395940527319908, "train_loss_bc": 0.0439593642950058, "train_loss_llm": 4.202127456665039e-06, "grad_norm": 0.02118118293583393, "global_step": 3593, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04616924375295639, "train_loss_bc": 0.046145569533109665, "train_loss_llm": 0.0023674368858337402, "grad_norm": 0.0256236232817173, "global_step": 3594, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04156602919101715, "train_loss_bc": 0.04153488576412201, "train_loss_llm": 0.0031145215034484863, "grad_norm": 0.034589290618896484, "global_step": 3595, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04723616689443588, "train_loss_bc": 0.04723616689443588, "train_loss_llm": 0.0, "grad_norm": 0.04011299088597298, "global_step": 3596, "epoch": 99, "lr": 0.001} +{"train_loss": 0.04747803881764412, "train_loss_bc": 0.04747803881764412, "train_loss_llm": 0.0, "grad_norm": 0.02101733349263668, "global_step": 3597, "epoch": 99, "lr": 0.001} +{"train_loss": 0.033667922019958496, "train_loss_bc": 0.033667922019958496, "train_loss_llm": 0.0, "grad_norm": 0.023240836337208748, "global_step": 3598, "epoch": 99, "lr": 0.001} +{"train_loss": 0.0434728976753023, "train_loss_bc": 0.04548709839582443, "train_loss_llm": 0.004346698522567749, "grad_norm": 0.03490313142538071, "global_step": 3599, "epoch": 99, "lr": 0.001, "val_loss": 0.05234067142009735} +{"train_loss": 0.04328159615397453, "train_loss_bc": 0.04328159615397453, "train_loss_llm": 0.0, "grad_norm": 0.04432176426053047, "global_step": 3600, "epoch": 100, "lr": 0.001} +{"train_loss": 0.0496492013335228, "train_loss_bc": 0.0496492013335228, "train_loss_llm": 0.0, "grad_norm": 0.01594873145222664, "global_step": 3601, "epoch": 100, "lr": 0.001} +{"train_loss": 0.0482180193066597, "train_loss_bc": 0.04821132495999336, "train_loss_llm": 0.0006693005561828613, "grad_norm": 0.019043566659092903, "global_step": 3602, "epoch": 100, "lr": 0.001} +{"train_loss": 0.038028549402952194, "train_loss_bc": 0.038028549402952194, "train_loss_llm": 0.0, "grad_norm": 0.0250051561743021, "global_step": 3603, "epoch": 100, "lr": 0.001} +{"train_loss": 0.03942934423685074, "train_loss_bc": 0.039421770721673965, "train_loss_llm": 0.000757366418838501, "grad_norm": 0.039868198335170746, "global_step": 3604, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04365583136677742, "train_loss_bc": 0.043623823672533035, "train_loss_llm": 0.003200709819793701, "grad_norm": 0.017337366938591003, "global_step": 3605, "epoch": 100, "lr": 0.001} +{"train_loss": 0.042742423713207245, "train_loss_bc": 0.042742423713207245, "train_loss_llm": 0.0, "grad_norm": 0.02631284110248089, "global_step": 3606, "epoch": 100, "lr": 0.001} +{"train_loss": 0.049673739820718765, "train_loss_bc": 0.049665581434965134, "train_loss_llm": 0.00081595778465271, "grad_norm": 0.03258835896849632, "global_step": 3607, "epoch": 100, "lr": 0.001} +{"train_loss": 0.044089462608098984, "train_loss_bc": 0.044089462608098984, "train_loss_llm": 0.0, "grad_norm": 0.046311382204294205, "global_step": 3608, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04224530979990959, "train_loss_bc": 0.04224356263875961, "train_loss_llm": 0.0001748800277709961, "grad_norm": 0.014430779963731766, "global_step": 3609, "epoch": 100, "lr": 0.001} +{"train_loss": 0.03974653407931328, "train_loss_bc": 0.03974653407931328, "train_loss_llm": 0.0, "grad_norm": 0.02821001410484314, "global_step": 3610, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04111149534583092, "train_loss_bc": 0.04111149534583092, "train_loss_llm": 0.0, "grad_norm": 0.03414174169301987, "global_step": 3611, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04186062514781952, "train_loss_bc": 0.041846863925457, "train_loss_llm": 0.0013760030269622803, "grad_norm": 0.03668792545795441, "global_step": 3612, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04058394208550453, "train_loss_bc": 0.04058394208550453, "train_loss_llm": 0.0, "grad_norm": 0.01718899793922901, "global_step": 3613, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04572083428502083, "train_loss_bc": 0.04572083428502083, "train_loss_llm": 0.0, "grad_norm": 0.022012850269675255, "global_step": 3614, "epoch": 100, "lr": 0.001} +{"train_loss": 0.03759625181555748, "train_loss_bc": 0.03759625181555748, "train_loss_llm": 0.0, "grad_norm": 0.026234744116663933, "global_step": 3615, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04098791629076004, "train_loss_bc": 0.04096846655011177, "train_loss_llm": 0.0019449293613433838, "grad_norm": 0.030461637303233147, "global_step": 3616, "epoch": 100, "lr": 0.001} +{"train_loss": 0.046478040516376495, "train_loss_bc": 0.046473391354084015, "train_loss_llm": 0.00046503543853759766, "grad_norm": 0.01894763857126236, "global_step": 3617, "epoch": 100, "lr": 0.001} +{"train_loss": 0.041422877460718155, "train_loss_bc": 0.041422877460718155, "train_loss_llm": 0.0, "grad_norm": 0.024915551766753197, "global_step": 3618, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04101218655705452, "train_loss_bc": 0.04101218655705452, "train_loss_llm": 0.0, "grad_norm": 0.03305288404226303, "global_step": 3619, "epoch": 100, "lr": 0.001} +{"train_loss": 0.03430682420730591, "train_loss_bc": 0.03429608792066574, "train_loss_llm": 0.0010734498500823975, "grad_norm": 0.03655064478516579, "global_step": 3620, "epoch": 100, "lr": 0.001} +{"train_loss": 0.039284516125917435, "train_loss_bc": 0.039284516125917435, "train_loss_llm": 0.0, "grad_norm": 0.01814616471529007, "global_step": 3621, "epoch": 100, "lr": 0.001} +{"train_loss": 0.05094847455620766, "train_loss_bc": 0.05094847455620766, "train_loss_llm": 0.0, "grad_norm": 0.02427578531205654, "global_step": 3622, "epoch": 100, "lr": 0.001} +{"train_loss": 0.043691810220479965, "train_loss_bc": 0.04367366060614586, "train_loss_llm": 0.0018147826194763184, "grad_norm": 0.0348602719604969, "global_step": 3623, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04794631153345108, "train_loss_bc": 0.04794631153345108, "train_loss_llm": 0.0, "grad_norm": 0.03434653952717781, "global_step": 3624, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04744945839047432, "train_loss_bc": 0.047434426844120026, "train_loss_llm": 0.0015031099319458008, "grad_norm": 0.018710531294345856, "global_step": 3625, "epoch": 100, "lr": 0.001} +{"train_loss": 0.041653793305158615, "train_loss_bc": 0.041653793305158615, "train_loss_llm": 0.0, "grad_norm": 0.027508778497576714, "global_step": 3626, "epoch": 100, "lr": 0.001} +{"train_loss": 0.03625095263123512, "train_loss_bc": 0.03625095263123512, "train_loss_llm": 0.0, "grad_norm": 0.03210519254207611, "global_step": 3627, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04899558797478676, "train_loss_bc": 0.048992738127708435, "train_loss_llm": 0.0002848505973815918, "grad_norm": 0.03567664697766304, "global_step": 3628, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04230445995926857, "train_loss_bc": 0.04230445995926857, "train_loss_llm": 0.0, "grad_norm": 0.01909399777650833, "global_step": 3629, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04381543770432472, "train_loss_bc": 0.04381543770432472, "train_loss_llm": 0.0, "grad_norm": 0.030154291540384293, "global_step": 3630, "epoch": 100, "lr": 0.001} +{"train_loss": 0.05184042081236839, "train_loss_bc": 0.05184042081236839, "train_loss_llm": 0.0, "grad_norm": 0.03634858876466751, "global_step": 3631, "epoch": 100, "lr": 0.001} +{"train_loss": 0.05195433273911476, "train_loss_bc": 0.05194023624062538, "train_loss_llm": 0.0014095008373260498, "grad_norm": 0.04507946968078613, "global_step": 3632, "epoch": 100, "lr": 0.001} +{"train_loss": 0.04194352775812149, "train_loss_bc": 0.04193615913391113, "train_loss_llm": 0.0007367730140686035, "grad_norm": 0.01640338823199272, "global_step": 3633, "epoch": 100, "lr": 0.001} +{"train_loss": 0.03456619381904602, "train_loss_bc": 0.03454533591866493, "train_loss_llm": 0.0020858049392700195, "grad_norm": 0.02740512229502201, "global_step": 3634, "epoch": 100, "lr": 0.001} +{"train_loss": 0.043525228587289654, "train_loss_bc": 0.052421946078538895, "train_loss_llm": 0.0, "grad_norm": 0.035818278789520264, "global_step": 3635, "epoch": 100, "lr": 0.001, "train/cumulative_reward": 77.61771930238368, "train/mean_score": 8.619647028094237, "train/success_rate": 0.1, "test/cumulative_reward": 46.424044007001754, "test/mean_score": 7.9175701428665795, "test/success_rate": 0.2, "val_loss": 0.05225425958633423, "train_action_mse_error": 0.047739554196596146} +{"train_loss": 0.04529297724366188, "train_loss_bc": 0.04528454318642616, "train_loss_llm": 0.0008435845375061035, "grad_norm": 0.04278668761253357, "global_step": 3636, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04391525685787201, "train_loss_bc": 0.04391525685787201, "train_loss_llm": 0.0, "grad_norm": 0.01755460910499096, "global_step": 3637, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04355853050947189, "train_loss_bc": 0.04355853050947189, "train_loss_llm": 0.0, "grad_norm": 0.029083073139190674, "global_step": 3638, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04993027448654175, "train_loss_bc": 0.04993027448654175, "train_loss_llm": 0.0, "grad_norm": 0.03190770372748375, "global_step": 3639, "epoch": 101, "lr": 0.001} +{"train_loss": 0.045478321611881256, "train_loss_bc": 0.045478321611881256, "train_loss_llm": 0.0, "grad_norm": 0.036970846354961395, "global_step": 3640, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04384274035692215, "train_loss_bc": 0.04384274035692215, "train_loss_llm": 0.0, "grad_norm": 0.019551606848835945, "global_step": 3641, "epoch": 101, "lr": 0.001} +{"train_loss": 0.041593123227357864, "train_loss_bc": 0.041593123227357864, "train_loss_llm": 0.0, "grad_norm": 0.022230932489037514, "global_step": 3642, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04002438858151436, "train_loss_bc": 0.04002438858151436, "train_loss_llm": 0.0, "grad_norm": 0.024136604741215706, "global_step": 3643, "epoch": 101, "lr": 0.001} +{"train_loss": 0.03553309664130211, "train_loss_bc": 0.03553309664130211, "train_loss_llm": 0.0, "grad_norm": 0.02877221070230007, "global_step": 3644, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04071839153766632, "train_loss_bc": 0.04071839153766632, "train_loss_llm": 0.0, "grad_norm": 0.016480006277561188, "global_step": 3645, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04547509178519249, "train_loss_bc": 0.04547509178519249, "train_loss_llm": 0.0, "grad_norm": 0.028479192405939102, "global_step": 3646, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04040706530213356, "train_loss_bc": 0.040374141186475754, "train_loss_llm": 0.0032923519611358643, "grad_norm": 0.031598541885614395, "global_step": 3647, "epoch": 101, "lr": 0.001} +{"train_loss": 0.0393810048699379, "train_loss_bc": 0.03936639428138733, "train_loss_llm": 0.0014611780643463135, "grad_norm": 0.039408788084983826, "global_step": 3648, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04428689926862717, "train_loss_bc": 0.044280800968408585, "train_loss_llm": 0.0006097257137298584, "grad_norm": 0.017140381038188934, "global_step": 3649, "epoch": 101, "lr": 0.001} +{"train_loss": 0.03624851256608963, "train_loss_bc": 0.03624851256608963, "train_loss_llm": 0.0, "grad_norm": 0.023644376546144485, "global_step": 3650, "epoch": 101, "lr": 0.001} +{"train_loss": 0.03919830918312073, "train_loss_bc": 0.03919830918312073, "train_loss_llm": 0.0, "grad_norm": 0.03178669512271881, "global_step": 3651, "epoch": 101, "lr": 0.001} +{"train_loss": 0.0429871492087841, "train_loss_bc": 0.0429871492087841, "train_loss_llm": 0.0, "grad_norm": 0.03686146065592766, "global_step": 3652, "epoch": 101, "lr": 0.001} +{"train_loss": 0.040768351405858994, "train_loss_bc": 0.040760338306427, "train_loss_llm": 0.0008012354373931885, "grad_norm": 0.01526531670242548, "global_step": 3653, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04802896827459335, "train_loss_bc": 0.04802896827459335, "train_loss_llm": 0.0, "grad_norm": 0.026027308776974678, "global_step": 3654, "epoch": 101, "lr": 0.001} +{"train_loss": 0.048405349254608154, "train_loss_bc": 0.048405349254608154, "train_loss_llm": 0.0, "grad_norm": 0.027984462678432465, "global_step": 3655, "epoch": 101, "lr": 0.001} +{"train_loss": 0.0446646474301815, "train_loss_bc": 0.04465766251087189, "train_loss_llm": 0.0006984472274780273, "grad_norm": 0.035318586975336075, "global_step": 3656, "epoch": 101, "lr": 0.001} +{"train_loss": 0.039632830768823624, "train_loss_bc": 0.039632830768823624, "train_loss_llm": 0.0, "grad_norm": 0.02072836086153984, "global_step": 3657, "epoch": 101, "lr": 0.001} +{"train_loss": 0.050334811210632324, "train_loss_bc": 0.05033285170793533, "train_loss_llm": 0.0001958310604095459, "grad_norm": 0.0320260226726532, "global_step": 3658, "epoch": 101, "lr": 0.001} +{"train_loss": 0.03844945877790451, "train_loss_bc": 0.03843340277671814, "train_loss_llm": 0.001605689525604248, "grad_norm": 0.04208505526185036, "global_step": 3659, "epoch": 101, "lr": 0.001} +{"train_loss": 0.048356860876083374, "train_loss_bc": 0.0483088456094265, "train_loss_llm": 0.004801452159881592, "grad_norm": 0.0539330318570137, "global_step": 3660, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04164860397577286, "train_loss_bc": 0.04164065048098564, "train_loss_llm": 0.000795513391494751, "grad_norm": 0.021132638677954674, "global_step": 3661, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04861326888203621, "train_loss_bc": 0.04860098659992218, "train_loss_llm": 0.001228243112564087, "grad_norm": 0.031150296330451965, "global_step": 3662, "epoch": 101, "lr": 0.001} +{"train_loss": 0.044893451035022736, "train_loss_bc": 0.044893451035022736, "train_loss_llm": 0.0, "grad_norm": 0.0370820090174675, "global_step": 3663, "epoch": 101, "lr": 0.001} +{"train_loss": 0.045532386749982834, "train_loss_bc": 0.045532386749982834, "train_loss_llm": 0.0, "grad_norm": 0.04664352163672447, "global_step": 3664, "epoch": 101, "lr": 0.001} +{"train_loss": 0.040884923189878464, "train_loss_bc": 0.040872830897569656, "train_loss_llm": 0.0012094080448150635, "grad_norm": 0.014743193052709103, "global_step": 3665, "epoch": 101, "lr": 0.001} +{"train_loss": 0.0450374074280262, "train_loss_bc": 0.045022059231996536, "train_loss_llm": 0.001534968614578247, "grad_norm": 0.021854223683476448, "global_step": 3666, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04682302474975586, "train_loss_bc": 0.04682302474975586, "train_loss_llm": 0.0, "grad_norm": 0.031143538653850555, "global_step": 3667, "epoch": 101, "lr": 0.001} +{"train_loss": 0.038121312856674194, "train_loss_bc": 0.038121312856674194, "train_loss_llm": 0.0, "grad_norm": 0.03378927707672119, "global_step": 3668, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04231606051325798, "train_loss_bc": 0.04231606051325798, "train_loss_llm": 0.0, "grad_norm": 0.015215086750686169, "global_step": 3669, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04697565361857414, "train_loss_bc": 0.04697565361857414, "train_loss_llm": 0.0, "grad_norm": 0.024388372898101807, "global_step": 3670, "epoch": 101, "lr": 0.001} +{"train_loss": 0.04342758986685011, "train_loss_bc": 0.04603473097085953, "train_loss_llm": 0.0, "grad_norm": 0.035409215837717056, "global_step": 3671, "epoch": 101, "lr": 0.001, "val_loss": 0.05246094614267349} +{"train_loss": 0.03567935898900032, "train_loss_bc": 0.03567597642540932, "train_loss_llm": 0.0003380775451660156, "grad_norm": 0.03569682687520981, "global_step": 3672, "epoch": 102, "lr": 0.001} +{"train_loss": 0.043186336755752563, "train_loss_bc": 0.04317215457558632, "train_loss_llm": 0.0014180541038513184, "grad_norm": 0.013044139370322227, "global_step": 3673, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04437055438756943, "train_loss_bc": 0.04437055438756943, "train_loss_llm": 0.0, "grad_norm": 0.02460743859410286, "global_step": 3674, "epoch": 102, "lr": 0.001} +{"train_loss": 0.03276943787932396, "train_loss_bc": 0.03276943787932396, "train_loss_llm": 0.0, "grad_norm": 0.03429817408323288, "global_step": 3675, "epoch": 102, "lr": 0.001} +{"train_loss": 0.03785823658108711, "train_loss_bc": 0.03785758838057518, "train_loss_llm": 6.496906280517578e-05, "grad_norm": 0.0398547388613224, "global_step": 3676, "epoch": 102, "lr": 0.001} +{"train_loss": 0.043076638132333755, "train_loss_bc": 0.043076638132333755, "train_loss_llm": 0.0, "grad_norm": 0.016896365210413933, "global_step": 3677, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04411104694008827, "train_loss_bc": 0.04411104694008827, "train_loss_llm": 0.0, "grad_norm": 0.024611812084913254, "global_step": 3678, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04726878181099892, "train_loss_bc": 0.047219906002283096, "train_loss_llm": 0.0048875510692596436, "grad_norm": 0.03449726104736328, "global_step": 3679, "epoch": 102, "lr": 0.001} +{"train_loss": 0.046985674649477005, "train_loss_bc": 0.046985674649477005, "train_loss_llm": 0.0, "grad_norm": 0.042494017630815506, "global_step": 3680, "epoch": 102, "lr": 0.001} +{"train_loss": 0.05147934705018997, "train_loss_bc": 0.05147934705018997, "train_loss_llm": 0.0, "grad_norm": 0.01899832673370838, "global_step": 3681, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04300987720489502, "train_loss_bc": 0.04300987720489502, "train_loss_llm": 0.0, "grad_norm": 0.029875507578253746, "global_step": 3682, "epoch": 102, "lr": 0.001} +{"train_loss": 0.033537887036800385, "train_loss_bc": 0.033537887036800385, "train_loss_llm": 0.0, "grad_norm": 0.04303790256381035, "global_step": 3683, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04805692657828331, "train_loss_bc": 0.04805692657828331, "train_loss_llm": 0.0, "grad_norm": 0.055852774530649185, "global_step": 3684, "epoch": 102, "lr": 0.001} +{"train_loss": 0.030415818095207214, "train_loss_bc": 0.030415818095207214, "train_loss_llm": 0.0, "grad_norm": 0.014693010598421097, "global_step": 3685, "epoch": 102, "lr": 0.001} +{"train_loss": 0.03869649022817612, "train_loss_bc": 0.03869649022817612, "train_loss_llm": 0.0, "grad_norm": 0.025445709004998207, "global_step": 3686, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04385052248835564, "train_loss_bc": 0.04385052248835564, "train_loss_llm": 0.0, "grad_norm": 0.029914112761616707, "global_step": 3687, "epoch": 102, "lr": 0.001} +{"train_loss": 0.03719436377286911, "train_loss_bc": 0.03718266263604164, "train_loss_llm": 0.0011699795722961426, "grad_norm": 0.033984724432229996, "global_step": 3688, "epoch": 102, "lr": 0.001} +{"train_loss": 0.046596333384513855, "train_loss_bc": 0.046596333384513855, "train_loss_llm": 0.0, "grad_norm": 0.01901109516620636, "global_step": 3689, "epoch": 102, "lr": 0.001} +{"train_loss": 0.0499386191368103, "train_loss_bc": 0.04992520809173584, "train_loss_llm": 0.0013412535190582275, "grad_norm": 0.02572023496031761, "global_step": 3690, "epoch": 102, "lr": 0.001} +{"train_loss": 0.05443571135401726, "train_loss_bc": 0.05443571135401726, "train_loss_llm": 0.0, "grad_norm": 0.03203711658716202, "global_step": 3691, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04470430314540863, "train_loss_bc": 0.04470430314540863, "train_loss_llm": 0.0, "grad_norm": 0.043877195566892624, "global_step": 3692, "epoch": 102, "lr": 0.001} +{"train_loss": 0.040925282984972, "train_loss_bc": 0.04090867191553116, "train_loss_llm": 0.0016612410545349121, "grad_norm": 0.016421526670455933, "global_step": 3693, "epoch": 102, "lr": 0.001} +{"train_loss": 0.05097704380750656, "train_loss_bc": 0.05097382515668869, "train_loss_llm": 0.00032201409339904785, "grad_norm": 0.029165172949433327, "global_step": 3694, "epoch": 102, "lr": 0.001} +{"train_loss": 0.045267991721630096, "train_loss_bc": 0.04525753855705261, "train_loss_llm": 0.0010452866554260254, "grad_norm": 0.039892714470624924, "global_step": 3695, "epoch": 102, "lr": 0.001} +{"train_loss": 0.041530534625053406, "train_loss_bc": 0.041530534625053406, "train_loss_llm": 0.0, "grad_norm": 0.04497482255101204, "global_step": 3696, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04688708856701851, "train_loss_bc": 0.04688708856701851, "train_loss_llm": 0.0, "grad_norm": 0.02053004503250122, "global_step": 3697, "epoch": 102, "lr": 0.001} +{"train_loss": 0.041657593101263046, "train_loss_bc": 0.041657593101263046, "train_loss_llm": 0.0, "grad_norm": 0.02760651521384716, "global_step": 3698, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04733313247561455, "train_loss_bc": 0.047322195023298264, "train_loss_llm": 0.0010938942432403564, "grad_norm": 0.0362580306828022, "global_step": 3699, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04001963883638382, "train_loss_bc": 0.04001963883638382, "train_loss_llm": 0.0, "grad_norm": 0.04133591055870056, "global_step": 3700, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04169474542140961, "train_loss_bc": 0.04169474542140961, "train_loss_llm": 0.0, "grad_norm": 0.018095536157488823, "global_step": 3701, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04123353213071823, "train_loss_bc": 0.04123353213071823, "train_loss_llm": 0.0, "grad_norm": 0.02830939181149006, "global_step": 3702, "epoch": 102, "lr": 0.001} +{"train_loss": 0.046522919088602066, "train_loss_bc": 0.046515896916389465, "train_loss_llm": 0.0007021725177764893, "grad_norm": 0.03616984561085701, "global_step": 3703, "epoch": 102, "lr": 0.001} +{"train_loss": 0.050676584243774414, "train_loss_bc": 0.050676584243774414, "train_loss_llm": 0.0, "grad_norm": 0.047463659197092056, "global_step": 3704, "epoch": 102, "lr": 0.001} +{"train_loss": 0.042636383324861526, "train_loss_bc": 0.042636383324861526, "train_loss_llm": 0.0, "grad_norm": 0.017739443108439445, "global_step": 3705, "epoch": 102, "lr": 0.001} +{"train_loss": 0.041959263384342194, "train_loss_bc": 0.041959263384342194, "train_loss_llm": 0.0, "grad_norm": 0.02266904152929783, "global_step": 3706, "epoch": 102, "lr": 0.001} +{"train_loss": 0.04351439502918058, "train_loss_bc": 0.0499742217361927, "train_loss_llm": 0.0, "grad_norm": 0.03441030904650688, "global_step": 3707, "epoch": 102, "lr": 0.001, "val_loss": 0.05277702212333679} +{"train_loss": 0.03755665197968483, "train_loss_bc": 0.03755665197968483, "train_loss_llm": 0.0, "grad_norm": 0.03668302297592163, "global_step": 3708, "epoch": 103, "lr": 0.001} +{"train_loss": 0.03720421344041824, "train_loss_bc": 0.03720421344041824, "train_loss_llm": 0.0, "grad_norm": 0.016468478366732597, "global_step": 3709, "epoch": 103, "lr": 0.001} +{"train_loss": 0.041191570460796356, "train_loss_bc": 0.041191570460796356, "train_loss_llm": 0.0, "grad_norm": 0.021160904318094254, "global_step": 3710, "epoch": 103, "lr": 0.001} +{"train_loss": 0.04608257859945297, "train_loss_bc": 0.04608257859945297, "train_loss_llm": 0.0, "grad_norm": 0.029429348185658455, "global_step": 3711, "epoch": 103, "lr": 0.001} +{"train_loss": 0.046031732112169266, "train_loss_bc": 0.046031732112169266, "train_loss_llm": 0.0, "grad_norm": 0.03263259306550026, "global_step": 3712, "epoch": 103, "lr": 0.001} +{"train_loss": 0.0396101288497448, "train_loss_bc": 0.0396101288497448, "train_loss_llm": 0.0, "grad_norm": 0.011925159953534603, "global_step": 3713, "epoch": 103, "lr": 0.001} +{"train_loss": 0.03845958784222603, "train_loss_bc": 0.03845958784222603, "train_loss_llm": 0.0, "grad_norm": 0.02301017753779888, "global_step": 3714, "epoch": 103, "lr": 0.001} +{"train_loss": 0.04752301052212715, "train_loss_bc": 0.047507163137197495, "train_loss_llm": 0.0015846192836761475, "grad_norm": 0.02978799119591713, "global_step": 3715, "epoch": 103, "lr": 0.001} +{"train_loss": 0.04112543910741806, "train_loss_bc": 0.04112543910741806, "train_loss_llm": 0.0, "grad_norm": 0.03741900622844696, "global_step": 3716, "epoch": 103, "lr": 0.001} +{"train_loss": 0.0444706454873085, "train_loss_bc": 0.04445686936378479, "train_loss_llm": 0.001377701759338379, "grad_norm": 0.018259812146425247, "global_step": 3717, "epoch": 103, "lr": 0.001} +{"train_loss": 0.03784511238336563, "train_loss_bc": 0.03784511238336563, "train_loss_llm": 0.0, "grad_norm": 0.024378813803195953, "global_step": 3718, "epoch": 103, "lr": 0.001} +{"train_loss": 0.050084564834833145, "train_loss_bc": 0.05006010830402374, "train_loss_llm": 0.002445816993713379, "grad_norm": 0.03296191617846489, "global_step": 3719, "epoch": 103, "lr": 0.001} +{"train_loss": 0.043739113956689835, "train_loss_bc": 0.04371985048055649, "train_loss_llm": 0.0019261837005615234, "grad_norm": 0.04129248112440109, "global_step": 3720, "epoch": 103, "lr": 0.001} +{"train_loss": 0.049669720232486725, "train_loss_bc": 0.04966936260461807, "train_loss_llm": 3.5703182220458984e-05, "grad_norm": 0.01870935969054699, "global_step": 3721, "epoch": 103, "lr": 0.001} +{"train_loss": 0.05045560002326965, "train_loss_bc": 0.05045560002326965, "train_loss_llm": 0.0, "grad_norm": 0.03173983469605446, "global_step": 3722, "epoch": 103, "lr": 0.001} +{"train_loss": 0.0353395938873291, "train_loss_bc": 0.0353395938873291, "train_loss_llm": 0.0, "grad_norm": 0.04585690796375275, "global_step": 3723, "epoch": 103, "lr": 0.001} +{"train_loss": 0.045883968472480774, "train_loss_bc": 0.04586901143193245, "train_loss_llm": 0.00149574875831604, "grad_norm": 0.0523274727165699, "global_step": 3724, "epoch": 103, "lr": 0.001} +{"train_loss": 0.04130249097943306, "train_loss_bc": 0.04129711538553238, "train_loss_llm": 0.0005373954772949219, "grad_norm": 0.016153911128640175, "global_step": 3725, "epoch": 103, "lr": 0.001} +{"train_loss": 0.041866183280944824, "train_loss_bc": 0.04185071960091591, "train_loss_llm": 0.001546323299407959, "grad_norm": 0.021600505337119102, "global_step": 3726, "epoch": 103, "lr": 0.001} +{"train_loss": 0.047044601291418076, "train_loss_bc": 0.047044601291418076, "train_loss_llm": 0.0, "grad_norm": 0.02787576988339424, "global_step": 3727, "epoch": 103, "lr": 0.001} +{"train_loss": 0.0477619506418705, "train_loss_bc": 0.04771776869893074, "train_loss_llm": 0.004418015480041504, "grad_norm": 0.03506021574139595, "global_step": 3728, "epoch": 103, "lr": 0.001} +{"train_loss": 0.044762223958969116, "train_loss_bc": 0.04474106803536415, "train_loss_llm": 0.002115607261657715, "grad_norm": 0.018520338460803032, "global_step": 3729, "epoch": 103, "lr": 0.001} +{"train_loss": 0.04658821225166321, "train_loss_bc": 0.04657852277159691, "train_loss_llm": 0.0009690523147583008, "grad_norm": 0.03144335746765137, "global_step": 3730, "epoch": 103, "lr": 0.001} +{"train_loss": 0.03781135380268097, "train_loss_bc": 0.03781135380268097, "train_loss_llm": 0.0, "grad_norm": 0.036066215485334396, "global_step": 3731, "epoch": 103, "lr": 0.001} +{"train_loss": 0.047030191868543625, "train_loss_bc": 0.047030191868543625, "train_loss_llm": 0.0, "grad_norm": 0.04326988756656647, "global_step": 3732, "epoch": 103, "lr": 0.001} +{"train_loss": 0.04749036580324173, "train_loss_bc": 0.04748445749282837, "train_loss_llm": 0.0005908310413360596, "grad_norm": 0.018624143674969673, "global_step": 3733, "epoch": 103, "lr": 0.001} +{"train_loss": 0.04075181856751442, "train_loss_bc": 0.04075181856751442, "train_loss_llm": 0.0, "grad_norm": 0.024137161672115326, "global_step": 3734, "epoch": 103, "lr": 0.001} +{"train_loss": 0.03461287170648575, "train_loss_bc": 0.0345943458378315, "train_loss_llm": 0.0018525421619415283, "grad_norm": 0.036507755517959595, "global_step": 3735, "epoch": 103, "lr": 0.001} +{"train_loss": 0.049489330500364304, "train_loss_bc": 0.049489330500364304, "train_loss_llm": 0.0, "grad_norm": 0.04602736607193947, "global_step": 3736, "epoch": 103, "lr": 0.001} +{"train_loss": 0.04853140190243721, "train_loss_bc": 0.04853140190243721, "train_loss_llm": 0.0, "grad_norm": 0.017419643700122833, "global_step": 3737, "epoch": 103, "lr": 0.001} +{"train_loss": 0.04968453571200371, "train_loss_bc": 0.04968453571200371, "train_loss_llm": 0.0, "grad_norm": 0.02315651811659336, "global_step": 3738, "epoch": 103, "lr": 0.001} +{"train_loss": 0.043184708803892136, "train_loss_bc": 0.043174389749765396, "train_loss_llm": 0.001032024621963501, "grad_norm": 0.028010869398713112, "global_step": 3739, "epoch": 103, "lr": 0.001} +{"train_loss": 0.04665428027510643, "train_loss_bc": 0.04665428027510643, "train_loss_llm": 0.0, "grad_norm": 0.03758566081523895, "global_step": 3740, "epoch": 103, "lr": 0.001} +{"train_loss": 0.03889823332428932, "train_loss_bc": 0.03887159377336502, "train_loss_llm": 0.0026640594005584717, "grad_norm": 0.018261877819895744, "global_step": 3741, "epoch": 103, "lr": 0.001} +{"train_loss": 0.0381900779902935, "train_loss_bc": 0.0381900779902935, "train_loss_llm": 0.0, "grad_norm": 0.026018192991614342, "global_step": 3742, "epoch": 103, "lr": 0.001} +{"train_loss": 0.043354380772345595, "train_loss_bc": 0.036825861781835556, "train_loss_llm": 0.00037810206413269043, "grad_norm": 0.04094308987259865, "global_step": 3743, "epoch": 103, "lr": 0.001, "val_loss": 0.051820337772369385} +{"train_loss": 0.043181147426366806, "train_loss_bc": 0.04317089915275574, "train_loss_llm": 0.0010248124599456787, "grad_norm": 0.04794713109731674, "global_step": 3744, "epoch": 104, "lr": 0.001} +{"train_loss": 0.03837917372584343, "train_loss_bc": 0.03837917372584343, "train_loss_llm": 0.0, "grad_norm": 0.016785748302936554, "global_step": 3745, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04656059294939041, "train_loss_bc": 0.04656059294939041, "train_loss_llm": 0.0, "grad_norm": 0.024179449304938316, "global_step": 3746, "epoch": 104, "lr": 0.001} +{"train_loss": 0.03694874420762062, "train_loss_bc": 0.03694874420762062, "train_loss_llm": 0.0, "grad_norm": 0.02935965172946453, "global_step": 3747, "epoch": 104, "lr": 0.001} +{"train_loss": 0.03673359751701355, "train_loss_bc": 0.03673359751701355, "train_loss_llm": 0.0, "grad_norm": 0.03565803915262222, "global_step": 3748, "epoch": 104, "lr": 0.001} +{"train_loss": 0.050465766340494156, "train_loss_bc": 0.050465766340494156, "train_loss_llm": 0.0, "grad_norm": 0.02168673276901245, "global_step": 3749, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04550790414214134, "train_loss_bc": 0.04550790414214134, "train_loss_llm": 0.0, "grad_norm": 0.03303813561797142, "global_step": 3750, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04469095170497894, "train_loss_bc": 0.04469095170497894, "train_loss_llm": 0.0, "grad_norm": 0.03781777620315552, "global_step": 3751, "epoch": 104, "lr": 0.001} +{"train_loss": 0.048526957631111145, "train_loss_bc": 0.048516515642404556, "train_loss_llm": 0.0010443031787872314, "grad_norm": 0.04090277478098869, "global_step": 3752, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04060044884681702, "train_loss_bc": 0.04060044884681702, "train_loss_llm": 0.0, "grad_norm": 0.01965085044503212, "global_step": 3753, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04574434831738472, "train_loss_bc": 0.04574434831738472, "train_loss_llm": 0.0, "grad_norm": 0.0393935889005661, "global_step": 3754, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04126806557178497, "train_loss_bc": 0.04126806557178497, "train_loss_llm": 0.0, "grad_norm": 0.04586895555257797, "global_step": 3755, "epoch": 104, "lr": 0.001} +{"train_loss": 0.037727389484643936, "train_loss_bc": 0.037727389484643936, "train_loss_llm": 0.0, "grad_norm": 0.05823462828993797, "global_step": 3756, "epoch": 104, "lr": 0.001} +{"train_loss": 0.05091572552919388, "train_loss_bc": 0.05091572552919388, "train_loss_llm": 0.0, "grad_norm": 0.017879096791148186, "global_step": 3757, "epoch": 104, "lr": 0.001} +{"train_loss": 0.03754798322916031, "train_loss_bc": 0.03753961995244026, "train_loss_llm": 0.0008362829685211182, "grad_norm": 0.025868559256196022, "global_step": 3758, "epoch": 104, "lr": 0.001} +{"train_loss": 0.042217716574668884, "train_loss_bc": 0.042217716574668884, "train_loss_llm": 0.0, "grad_norm": 0.03386734053492546, "global_step": 3759, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04487546160817146, "train_loss_bc": 0.04487546160817146, "train_loss_llm": 0.0, "grad_norm": 0.040292758494615555, "global_step": 3760, "epoch": 104, "lr": 0.001} +{"train_loss": 0.03757663816213608, "train_loss_bc": 0.03757663816213608, "train_loss_llm": 0.0, "grad_norm": 0.01297824364155531, "global_step": 3761, "epoch": 104, "lr": 0.001} +{"train_loss": 0.047906503081321716, "train_loss_bc": 0.047906503081321716, "train_loss_llm": 0.0, "grad_norm": 0.0206855870783329, "global_step": 3762, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04828402027487755, "train_loss_bc": 0.04825185239315033, "train_loss_llm": 0.0032168328762054443, "grad_norm": 0.02963969297707081, "global_step": 3763, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04227456450462341, "train_loss_bc": 0.04227456450462341, "train_loss_llm": 0.0, "grad_norm": 0.0318385548889637, "global_step": 3764, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04435967653989792, "train_loss_bc": 0.044331103563308716, "train_loss_llm": 0.0028574764728546143, "grad_norm": 0.015538484789431095, "global_step": 3765, "epoch": 104, "lr": 0.001} +{"train_loss": 0.047504786401987076, "train_loss_bc": 0.047504786401987076, "train_loss_llm": 0.0, "grad_norm": 0.02324146218597889, "global_step": 3766, "epoch": 104, "lr": 0.001} +{"train_loss": 0.05105467885732651, "train_loss_bc": 0.05105467885732651, "train_loss_llm": 0.0, "grad_norm": 0.030501507222652435, "global_step": 3767, "epoch": 104, "lr": 0.001} +{"train_loss": 0.045710302889347076, "train_loss_bc": 0.045710302889347076, "train_loss_llm": 0.0, "grad_norm": 0.03569338843226433, "global_step": 3768, "epoch": 104, "lr": 0.001} +{"train_loss": 0.03806723281741142, "train_loss_bc": 0.03806548938155174, "train_loss_llm": 0.00017434358596801758, "grad_norm": 0.01750057376921177, "global_step": 3769, "epoch": 104, "lr": 0.001} +{"train_loss": 0.03976742550730705, "train_loss_bc": 0.03976742550730705, "train_loss_llm": 0.0, "grad_norm": 0.02961842715740204, "global_step": 3770, "epoch": 104, "lr": 0.001} +{"train_loss": 0.03635501116514206, "train_loss_bc": 0.03635188564658165, "train_loss_llm": 0.00031250715255737305, "grad_norm": 0.032518282532691956, "global_step": 3771, "epoch": 104, "lr": 0.001} +{"train_loss": 0.035658638924360275, "train_loss_bc": 0.035658638924360275, "train_loss_llm": 0.0, "grad_norm": 0.042396917939186096, "global_step": 3772, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04391679912805557, "train_loss_bc": 0.04391679912805557, "train_loss_llm": 0.0, "grad_norm": 0.015708139166235924, "global_step": 3773, "epoch": 104, "lr": 0.001} +{"train_loss": 0.048084307461977005, "train_loss_bc": 0.04806624352931976, "train_loss_llm": 0.0018063485622406006, "grad_norm": 0.025162577629089355, "global_step": 3774, "epoch": 104, "lr": 0.001} +{"train_loss": 0.05250454321503639, "train_loss_bc": 0.05250454321503639, "train_loss_llm": 0.0, "grad_norm": 0.036218371242284775, "global_step": 3775, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04266371577978134, "train_loss_bc": 0.04265287518501282, "train_loss_llm": 0.001083970069885254, "grad_norm": 0.04085152596235275, "global_step": 3776, "epoch": 104, "lr": 0.001} +{"train_loss": 0.042121030390262604, "train_loss_bc": 0.042121030390262604, "train_loss_llm": 0.0, "grad_norm": 0.018146540969610214, "global_step": 3777, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04179330915212631, "train_loss_bc": 0.04179330915212631, "train_loss_llm": 0.0, "grad_norm": 0.024054288864135742, "global_step": 3778, "epoch": 104, "lr": 0.001} +{"train_loss": 0.04324554300142659, "train_loss_bc": 0.03932863846421242, "train_loss_llm": 0.0015749633312225342, "grad_norm": 0.03992561250925064, "global_step": 3779, "epoch": 104, "lr": 0.001, "val_loss": 0.05283015966415405} +{"train_loss": 0.04916298761963844, "train_loss_bc": 0.04916298761963844, "train_loss_llm": 0.0, "grad_norm": 0.043547119945287704, "global_step": 3780, "epoch": 105, "lr": 0.001} +{"train_loss": 0.03619459643959999, "train_loss_bc": 0.03619459643959999, "train_loss_llm": 0.0, "grad_norm": 0.011871927417814732, "global_step": 3781, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04134245961904526, "train_loss_bc": 0.04134245961904526, "train_loss_llm": 0.0, "grad_norm": 0.017972810193896294, "global_step": 3782, "epoch": 105, "lr": 0.001} +{"train_loss": 0.037187520414590836, "train_loss_bc": 0.03718180954456329, "train_loss_llm": 0.0005710124969482422, "grad_norm": 0.023809175938367844, "global_step": 3783, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04257328063249588, "train_loss_bc": 0.04257328063249588, "train_loss_llm": 0.0, "grad_norm": 0.03052879497408867, "global_step": 3784, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04705078899860382, "train_loss_bc": 0.04704809561371803, "train_loss_llm": 0.00026917457580566406, "grad_norm": 0.016824278980493546, "global_step": 3785, "epoch": 105, "lr": 0.001} +{"train_loss": 0.039993759244680405, "train_loss_bc": 0.03998701274394989, "train_loss_llm": 0.0006745457649230957, "grad_norm": 0.022382834926247597, "global_step": 3786, "epoch": 105, "lr": 0.001} +{"train_loss": 0.043019652366638184, "train_loss_bc": 0.043019652366638184, "train_loss_llm": 0.0, "grad_norm": 0.03150646388530731, "global_step": 3787, "epoch": 105, "lr": 0.001} +{"train_loss": 0.044194526970386505, "train_loss_bc": 0.044194526970386505, "train_loss_llm": 0.0, "grad_norm": 0.04019920900464058, "global_step": 3788, "epoch": 105, "lr": 0.001} +{"train_loss": 0.042706869542598724, "train_loss_bc": 0.042706869542598724, "train_loss_llm": 0.0, "grad_norm": 0.01589266210794449, "global_step": 3789, "epoch": 105, "lr": 0.001} +{"train_loss": 0.040797680616378784, "train_loss_bc": 0.040797680616378784, "train_loss_llm": 0.0, "grad_norm": 0.022627979516983032, "global_step": 3790, "epoch": 105, "lr": 0.001} +{"train_loss": 0.03662848100066185, "train_loss_bc": 0.03661445900797844, "train_loss_llm": 0.0014022886753082275, "grad_norm": 0.030887935310602188, "global_step": 3791, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04952327534556389, "train_loss_bc": 0.04951446130871773, "train_loss_llm": 0.0008813440799713135, "grad_norm": 0.038451455533504486, "global_step": 3792, "epoch": 105, "lr": 0.001} +{"train_loss": 0.05289389565587044, "train_loss_bc": 0.05289389565587044, "train_loss_llm": 0.0, "grad_norm": 0.018890341743826866, "global_step": 3793, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04454268887639046, "train_loss_bc": 0.04452144354581833, "train_loss_llm": 0.002124607563018799, "grad_norm": 0.026209484785795212, "global_step": 3794, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04305591061711311, "train_loss_bc": 0.04305591061711311, "train_loss_llm": 0.0, "grad_norm": 0.03128515183925629, "global_step": 3795, "epoch": 105, "lr": 0.001} +{"train_loss": 0.051086023449897766, "train_loss_bc": 0.051086023449897766, "train_loss_llm": 0.0, "grad_norm": 0.03917817398905754, "global_step": 3796, "epoch": 105, "lr": 0.001} +{"train_loss": 0.03993690758943558, "train_loss_bc": 0.03993690758943558, "train_loss_llm": 0.0, "grad_norm": 0.014854001812636852, "global_step": 3797, "epoch": 105, "lr": 0.001} +{"train_loss": 0.05073162540793419, "train_loss_bc": 0.05070952698588371, "train_loss_llm": 0.002209961414337158, "grad_norm": 0.022877920418977737, "global_step": 3798, "epoch": 105, "lr": 0.001} +{"train_loss": 0.03832272067666054, "train_loss_bc": 0.03832272067666054, "train_loss_llm": 0.0, "grad_norm": 0.02856575697660446, "global_step": 3799, "epoch": 105, "lr": 0.001} +{"train_loss": 0.03879442438483238, "train_loss_bc": 0.038780201226472855, "train_loss_llm": 0.0014224052429199219, "grad_norm": 0.03806445747613907, "global_step": 3800, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04600848630070686, "train_loss_bc": 0.04600848630070686, "train_loss_llm": 0.0, "grad_norm": 0.01670762151479721, "global_step": 3801, "epoch": 105, "lr": 0.001} +{"train_loss": 0.047492239624261856, "train_loss_bc": 0.047492239624261856, "train_loss_llm": 0.0, "grad_norm": 0.029593566432595253, "global_step": 3802, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04107664152979851, "train_loss_bc": 0.04107664152979851, "train_loss_llm": 0.0, "grad_norm": 0.03788266330957413, "global_step": 3803, "epoch": 105, "lr": 0.001} +{"train_loss": 0.039826419204473495, "train_loss_bc": 0.039826419204473495, "train_loss_llm": 0.0, "grad_norm": 0.047951966524124146, "global_step": 3804, "epoch": 105, "lr": 0.001} +{"train_loss": 0.03364051133394241, "train_loss_bc": 0.03362572193145752, "train_loss_llm": 0.001478821039199829, "grad_norm": 0.014180254191160202, "global_step": 3805, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04172048345208168, "train_loss_bc": 0.04172048345208168, "train_loss_llm": 0.0, "grad_norm": 0.024017447605729103, "global_step": 3806, "epoch": 105, "lr": 0.001} +{"train_loss": 0.03942194953560829, "train_loss_bc": 0.03942194953560829, "train_loss_llm": 0.0, "grad_norm": 0.02730914019048214, "global_step": 3807, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04533456265926361, "train_loss_bc": 0.04533456265926361, "train_loss_llm": 0.0, "grad_norm": 0.03844013065099716, "global_step": 3808, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04707431420683861, "train_loss_bc": 0.04707431420683861, "train_loss_llm": 0.0, "grad_norm": 0.017380664125084877, "global_step": 3809, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04205702990293503, "train_loss_bc": 0.04205702990293503, "train_loss_llm": 0.0, "grad_norm": 0.024874364957213402, "global_step": 3810, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04128829017281532, "train_loss_bc": 0.04128829017281532, "train_loss_llm": 0.0, "grad_norm": 0.028341686353087425, "global_step": 3811, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04129622131586075, "train_loss_bc": 0.04127231985330582, "train_loss_llm": 0.0023902058601379395, "grad_norm": 0.03375409543514252, "global_step": 3812, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04798072949051857, "train_loss_bc": 0.047950077801942825, "train_loss_llm": 0.0030652284622192383, "grad_norm": 0.014150972478091717, "global_step": 3813, "epoch": 105, "lr": 0.001} +{"train_loss": 0.04513395205140114, "train_loss_bc": 0.04513395205140114, "train_loss_llm": 0.0, "grad_norm": 0.020310495048761368, "global_step": 3814, "epoch": 105, "lr": 0.001} +{"train_loss": 0.043196536497109465, "train_loss_bc": 0.04598340764641762, "train_loss_llm": 0.0, "grad_norm": 0.03913205489516258, "global_step": 3815, "epoch": 105, "lr": 0.001, "train/cumulative_reward": 51.29562099742022, "train/mean_score": 7.790897577664197, "train/success_rate": 0.1, "test/cumulative_reward": 70.68482341735348, "test/mean_score": 8.487984441194618, "test/success_rate": 0.18, "val_loss": 0.052347514778375626, "train_action_mse_error": 0.04690811038017273} +{"train_loss": 0.040524814277887344, "train_loss_bc": 0.040524814277887344, "train_loss_llm": 0.0, "grad_norm": 0.047040026634931564, "global_step": 3816, "epoch": 106, "lr": 0.001} +{"train_loss": 0.040813200175762177, "train_loss_bc": 0.04081035777926445, "train_loss_llm": 0.0002843141555786133, "grad_norm": 0.016273796558380127, "global_step": 3817, "epoch": 106, "lr": 0.001} +{"train_loss": 0.042053673416376114, "train_loss_bc": 0.042053673416376114, "train_loss_llm": 0.0, "grad_norm": 0.02562297321856022, "global_step": 3818, "epoch": 106, "lr": 0.001} +{"train_loss": 0.033061444759368896, "train_loss_bc": 0.033061444759368896, "train_loss_llm": 0.0, "grad_norm": 0.03169775754213333, "global_step": 3819, "epoch": 106, "lr": 0.001} +{"train_loss": 0.042626745998859406, "train_loss_bc": 0.0426173061132431, "train_loss_llm": 0.0009439289569854736, "grad_norm": 0.03374837711453438, "global_step": 3820, "epoch": 106, "lr": 0.001} +{"train_loss": 0.045669812709093094, "train_loss_bc": 0.045669812709093094, "train_loss_llm": 0.0, "grad_norm": 0.017616622149944305, "global_step": 3821, "epoch": 106, "lr": 0.001} +{"train_loss": 0.041577983647584915, "train_loss_bc": 0.041577983647584915, "train_loss_llm": 0.0, "grad_norm": 0.021790271624922752, "global_step": 3822, "epoch": 106, "lr": 0.001} +{"train_loss": 0.046684630215168, "train_loss_bc": 0.046684630215168, "train_loss_llm": 0.0, "grad_norm": 0.023517295718193054, "global_step": 3823, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04507916420698166, "train_loss_bc": 0.04506955295801163, "train_loss_llm": 0.0009609758853912354, "grad_norm": 0.027983440086245537, "global_step": 3824, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04316664859652519, "train_loss_bc": 0.04316664859652519, "train_loss_llm": 0.0, "grad_norm": 0.019571280106902122, "global_step": 3825, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04212191328406334, "train_loss_bc": 0.04211214557290077, "train_loss_llm": 0.000976651906967163, "grad_norm": 0.035572346299886703, "global_step": 3826, "epoch": 106, "lr": 0.001} +{"train_loss": 0.044021137058734894, "train_loss_bc": 0.043997328728437424, "train_loss_llm": 0.0023806989192962646, "grad_norm": 0.037068065255880356, "global_step": 3827, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04192918539047241, "train_loss_bc": 0.04192918539047241, "train_loss_llm": 0.0, "grad_norm": 0.0467788390815258, "global_step": 3828, "epoch": 106, "lr": 0.001} +{"train_loss": 0.041534774005413055, "train_loss_bc": 0.04151906818151474, "train_loss_llm": 0.0015707612037658691, "grad_norm": 0.01964130997657776, "global_step": 3829, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04487060382962227, "train_loss_bc": 0.04487060382962227, "train_loss_llm": 0.0, "grad_norm": 0.0302070714533329, "global_step": 3830, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04072000831365585, "train_loss_bc": 0.040699269622564316, "train_loss_llm": 0.0020737051963806152, "grad_norm": 0.038468457758426666, "global_step": 3831, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04666489362716675, "train_loss_bc": 0.04666489362716675, "train_loss_llm": 0.0, "grad_norm": 0.043021030724048615, "global_step": 3832, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04948636516928673, "train_loss_bc": 0.04947662353515625, "train_loss_llm": 0.0009742677211761475, "grad_norm": 0.014913856983184814, "global_step": 3833, "epoch": 106, "lr": 0.001} +{"train_loss": 0.049304377287626266, "train_loss_bc": 0.049302954226732254, "train_loss_llm": 0.00014230608940124512, "grad_norm": 0.02426052838563919, "global_step": 3834, "epoch": 106, "lr": 0.001} +{"train_loss": 0.03205910697579384, "train_loss_bc": 0.03204561769962311, "train_loss_llm": 0.0013487935066223145, "grad_norm": 0.027830922976136208, "global_step": 3835, "epoch": 106, "lr": 0.001} +{"train_loss": 0.044361669570207596, "train_loss_bc": 0.044361669570207596, "train_loss_llm": 0.0, "grad_norm": 0.031114520505070686, "global_step": 3836, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04480777308344841, "train_loss_bc": 0.04480777308344841, "train_loss_llm": 0.0, "grad_norm": 0.01550450548529625, "global_step": 3837, "epoch": 106, "lr": 0.001} +{"train_loss": 0.045941680669784546, "train_loss_bc": 0.045941680669784546, "train_loss_llm": 0.0, "grad_norm": 0.02023959532380104, "global_step": 3838, "epoch": 106, "lr": 0.001} +{"train_loss": 0.045184310525655746, "train_loss_bc": 0.04518264904618263, "train_loss_llm": 0.00016623735427856445, "grad_norm": 0.026937954127788544, "global_step": 3839, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04313577339053154, "train_loss_bc": 0.04313577339053154, "train_loss_llm": 0.0, "grad_norm": 0.0306532084941864, "global_step": 3840, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04909198358654976, "train_loss_bc": 0.04909198358654976, "train_loss_llm": 0.0, "grad_norm": 0.015197169966995716, "global_step": 3841, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04276586323976517, "train_loss_bc": 0.04276586323976517, "train_loss_llm": 0.0, "grad_norm": 0.026007769629359245, "global_step": 3842, "epoch": 106, "lr": 0.001} +{"train_loss": 0.03862586244940758, "train_loss_bc": 0.03859598934650421, "train_loss_llm": 0.0029873251914978027, "grad_norm": 0.02897951379418373, "global_step": 3843, "epoch": 106, "lr": 0.001} +{"train_loss": 0.0398404486477375, "train_loss_bc": 0.0398404486477375, "train_loss_llm": 0.0, "grad_norm": 0.031984761357307434, "global_step": 3844, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04051872715353966, "train_loss_bc": 0.04049999639391899, "train_loss_llm": 0.0018731653690338135, "grad_norm": 0.015050007961690426, "global_step": 3845, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04537323862314224, "train_loss_bc": 0.04537323862314224, "train_loss_llm": 0.0, "grad_norm": 0.02214268408715725, "global_step": 3846, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04956834390759468, "train_loss_bc": 0.04956834390759468, "train_loss_llm": 0.0, "grad_norm": 0.035173725336790085, "global_step": 3847, "epoch": 106, "lr": 0.001} +{"train_loss": 0.039965949952602386, "train_loss_bc": 0.039965949952602386, "train_loss_llm": 0.0, "grad_norm": 0.0439293310046196, "global_step": 3848, "epoch": 106, "lr": 0.001} +{"train_loss": 0.038467224687337875, "train_loss_bc": 0.038467224687337875, "train_loss_llm": 0.0, "grad_norm": 0.016418877989053726, "global_step": 3849, "epoch": 106, "lr": 0.001} +{"train_loss": 0.050764720886945724, "train_loss_bc": 0.05074826255440712, "train_loss_llm": 0.0016457736492156982, "grad_norm": 0.025547418743371964, "global_step": 3850, "epoch": 106, "lr": 0.001} +{"train_loss": 0.04306946125709348, "train_loss_bc": 0.03811655193567276, "train_loss_llm": 0.0, "grad_norm": 0.02819889783859253, "global_step": 3851, "epoch": 106, "lr": 0.001, "val_loss": 0.05226588621735573} +{"train_loss": 0.05204680562019348, "train_loss_bc": 0.05204680562019348, "train_loss_llm": 0.0, "grad_norm": 0.03753799572587013, "global_step": 3852, "epoch": 107, "lr": 0.001} +{"train_loss": 0.03996172174811363, "train_loss_bc": 0.03996172174811363, "train_loss_llm": 0.0, "grad_norm": 0.012813924811780453, "global_step": 3853, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04577365517616272, "train_loss_bc": 0.04577365517616272, "train_loss_llm": 0.0, "grad_norm": 0.021282687783241272, "global_step": 3854, "epoch": 107, "lr": 0.001} +{"train_loss": 0.05082795023918152, "train_loss_bc": 0.05081833526492119, "train_loss_llm": 0.0009613335132598877, "grad_norm": 0.026803432032465935, "global_step": 3855, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04495744779706001, "train_loss_bc": 0.04495744779706001, "train_loss_llm": 0.0, "grad_norm": 0.028550026938319206, "global_step": 3856, "epoch": 107, "lr": 0.001} +{"train_loss": 0.05013342201709747, "train_loss_bc": 0.05012597143650055, "train_loss_llm": 0.0007451772689819336, "grad_norm": 0.018486084416508675, "global_step": 3857, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04844405874609947, "train_loss_bc": 0.04844405874609947, "train_loss_llm": 0.0, "grad_norm": 0.024426156654953957, "global_step": 3858, "epoch": 107, "lr": 0.001} +{"train_loss": 0.044266823679208755, "train_loss_bc": 0.04426659643650055, "train_loss_llm": 2.2679567337036133e-05, "grad_norm": 0.03284265846014023, "global_step": 3859, "epoch": 107, "lr": 0.001} +{"train_loss": 0.044612061232328415, "train_loss_bc": 0.044612061232328415, "train_loss_llm": 0.0, "grad_norm": 0.041335199028253555, "global_step": 3860, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04411575198173523, "train_loss_bc": 0.04408489912748337, "train_loss_llm": 0.003085225820541382, "grad_norm": 0.016916651278734207, "global_step": 3861, "epoch": 107, "lr": 0.001} +{"train_loss": 0.040713559836149216, "train_loss_bc": 0.040690917521715164, "train_loss_llm": 0.0022642910480499268, "grad_norm": 0.024775661528110504, "global_step": 3862, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04011297598481178, "train_loss_bc": 0.04011297598481178, "train_loss_llm": 0.0, "grad_norm": 0.03105776011943817, "global_step": 3863, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04260743781924248, "train_loss_bc": 0.04260743781924248, "train_loss_llm": 0.0, "grad_norm": 0.0357745960354805, "global_step": 3864, "epoch": 107, "lr": 0.001} +{"train_loss": 0.039424795657396317, "train_loss_bc": 0.039424795657396317, "train_loss_llm": 0.0, "grad_norm": 0.014374740421772003, "global_step": 3865, "epoch": 107, "lr": 0.001} +{"train_loss": 0.03983299061655998, "train_loss_bc": 0.03983250632882118, "train_loss_llm": 4.842877388000488e-05, "grad_norm": 0.021658334881067276, "global_step": 3866, "epoch": 107, "lr": 0.001} +{"train_loss": 0.042941849678754807, "train_loss_bc": 0.042917024344205856, "train_loss_llm": 0.002482593059539795, "grad_norm": 0.029055839404463768, "global_step": 3867, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04024317488074303, "train_loss_bc": 0.04024317488074303, "train_loss_llm": 0.0, "grad_norm": 0.03674236685037613, "global_step": 3868, "epoch": 107, "lr": 0.001} +{"train_loss": 0.044035088270902634, "train_loss_bc": 0.044035088270902634, "train_loss_llm": 0.0, "grad_norm": 0.01971319690346718, "global_step": 3869, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04503269121050835, "train_loss_bc": 0.04502652585506439, "train_loss_llm": 0.0006165504455566406, "grad_norm": 0.032185960561037064, "global_step": 3870, "epoch": 107, "lr": 0.001} +{"train_loss": 0.035800013691186905, "train_loss_bc": 0.035800013691186905, "train_loss_llm": 0.0, "grad_norm": 0.03379787877202034, "global_step": 3871, "epoch": 107, "lr": 0.001} +{"train_loss": 0.048880260437726974, "train_loss_bc": 0.048880260437726974, "train_loss_llm": 0.0, "grad_norm": 0.04541365057229996, "global_step": 3872, "epoch": 107, "lr": 0.001} +{"train_loss": 0.039968039840459824, "train_loss_bc": 0.039968039840459824, "train_loss_llm": 0.0, "grad_norm": 0.018276982009410858, "global_step": 3873, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04421409219503403, "train_loss_bc": 0.04419683665037155, "train_loss_llm": 0.0017254948616027832, "grad_norm": 0.02858934924006462, "global_step": 3874, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04607713222503662, "train_loss_bc": 0.04603131487965584, "train_loss_llm": 0.004581660032272339, "grad_norm": 0.03241872414946556, "global_step": 3875, "epoch": 107, "lr": 0.001} +{"train_loss": 0.042994484305381775, "train_loss_bc": 0.042994484305381775, "train_loss_llm": 0.0, "grad_norm": 0.03423875570297241, "global_step": 3876, "epoch": 107, "lr": 0.001} +{"train_loss": 0.03927002102136612, "train_loss_bc": 0.03927002102136612, "train_loss_llm": 0.0, "grad_norm": 0.01636786013841629, "global_step": 3877, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04137760400772095, "train_loss_bc": 0.04137760400772095, "train_loss_llm": 0.0, "grad_norm": 0.020814113318920135, "global_step": 3878, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04404934495687485, "train_loss_bc": 0.04404789209365845, "train_loss_llm": 0.00014540553092956543, "grad_norm": 0.027178671211004257, "global_step": 3879, "epoch": 107, "lr": 0.001} +{"train_loss": 0.03629317507147789, "train_loss_bc": 0.03629317507147789, "train_loss_llm": 0.0, "grad_norm": 0.03213439881801605, "global_step": 3880, "epoch": 107, "lr": 0.001} +{"train_loss": 0.03452354297041893, "train_loss_bc": 0.03451632708311081, "train_loss_llm": 0.0007215142250061035, "grad_norm": 0.015033997595310211, "global_step": 3881, "epoch": 107, "lr": 0.001} +{"train_loss": 0.041173744946718216, "train_loss_bc": 0.041159432381391525, "train_loss_llm": 0.0014312565326690674, "grad_norm": 0.02495875209569931, "global_step": 3882, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04759574308991432, "train_loss_bc": 0.04759574308991432, "train_loss_llm": 0.0, "grad_norm": 0.03455948084592819, "global_step": 3883, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04326910898089409, "train_loss_bc": 0.04326700046658516, "train_loss_llm": 0.00021073222160339355, "grad_norm": 0.050792090594768524, "global_step": 3884, "epoch": 107, "lr": 0.001} +{"train_loss": 0.046810343861579895, "train_loss_bc": 0.0468096099793911, "train_loss_llm": 7.322430610656738e-05, "grad_norm": 0.025838486850261688, "global_step": 3885, "epoch": 107, "lr": 0.001} +{"train_loss": 0.04844529181718826, "train_loss_bc": 0.04844529181718826, "train_loss_llm": 0.0, "grad_norm": 0.03150773420929909, "global_step": 3886, "epoch": 107, "lr": 0.001} +{"train_loss": 0.043024725280702114, "train_loss_bc": 0.028061771765351295, "train_loss_llm": 0.00021320581436157227, "grad_norm": 0.041895218193531036, "global_step": 3887, "epoch": 107, "lr": 0.001, "val_loss": 0.05279489979147911} +{"train_loss": 0.041318971663713455, "train_loss_bc": 0.041318971663713455, "train_loss_llm": 0.0, "grad_norm": 0.050974633544683456, "global_step": 3888, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04915736988186836, "train_loss_bc": 0.04915736988186836, "train_loss_llm": 0.0, "grad_norm": 0.014119795523583889, "global_step": 3889, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04652716591954231, "train_loss_bc": 0.04652676358819008, "train_loss_llm": 4.011392593383789e-05, "grad_norm": 0.022161951288580894, "global_step": 3890, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04496676102280617, "train_loss_bc": 0.04493672028183937, "train_loss_llm": 0.0030039548873901367, "grad_norm": 0.027459362521767616, "global_step": 3891, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04194338992238045, "train_loss_bc": 0.04193470627069473, "train_loss_llm": 0.0008683502674102783, "grad_norm": 0.035455599427223206, "global_step": 3892, "epoch": 108, "lr": 0.001} +{"train_loss": 0.0456976592540741, "train_loss_bc": 0.0456976592540741, "train_loss_llm": 0.0, "grad_norm": 0.01608380116522312, "global_step": 3893, "epoch": 108, "lr": 0.001} +{"train_loss": 0.05039164796471596, "train_loss_bc": 0.05039164796471596, "train_loss_llm": 0.0, "grad_norm": 0.021005846560001373, "global_step": 3894, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04250454530119896, "train_loss_bc": 0.042483728379011154, "train_loss_llm": 0.0020817220211029053, "grad_norm": 0.0268485639244318, "global_step": 3895, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04402276128530502, "train_loss_bc": 0.04402276128530502, "train_loss_llm": 0.0, "grad_norm": 0.031826745718717575, "global_step": 3896, "epoch": 108, "lr": 0.001} +{"train_loss": 0.0358898900449276, "train_loss_bc": 0.0358898900449276, "train_loss_llm": 0.0, "grad_norm": 0.017475439235568047, "global_step": 3897, "epoch": 108, "lr": 0.001} +{"train_loss": 0.042441289871931076, "train_loss_bc": 0.042441289871931076, "train_loss_llm": 0.0, "grad_norm": 0.03157443925738335, "global_step": 3898, "epoch": 108, "lr": 0.001} +{"train_loss": 0.03140820562839508, "train_loss_bc": 0.03140820562839508, "train_loss_llm": 0.0, "grad_norm": 0.039539724588394165, "global_step": 3899, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04203624278306961, "train_loss_bc": 0.04203624278306961, "train_loss_llm": 0.0, "grad_norm": 0.04514347389340401, "global_step": 3900, "epoch": 108, "lr": 0.001} +{"train_loss": 0.040248140692710876, "train_loss_bc": 0.040248140692710876, "train_loss_llm": 0.0, "grad_norm": 0.017669426277279854, "global_step": 3901, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04704616963863373, "train_loss_bc": 0.04704616963863373, "train_loss_llm": 0.0, "grad_norm": 0.029406916350126266, "global_step": 3902, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04476936534047127, "train_loss_bc": 0.04476361721754074, "train_loss_llm": 0.000574946403503418, "grad_norm": 0.038207996636629105, "global_step": 3903, "epoch": 108, "lr": 0.001} +{"train_loss": 0.042796142399311066, "train_loss_bc": 0.042796142399311066, "train_loss_llm": 0.0, "grad_norm": 0.0477403961122036, "global_step": 3904, "epoch": 108, "lr": 0.001} +{"train_loss": 0.046113986521959305, "train_loss_bc": 0.046113986521959305, "train_loss_llm": 0.0, "grad_norm": 0.01807418465614319, "global_step": 3905, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04381154105067253, "train_loss_bc": 0.04379092529416084, "train_loss_llm": 0.0020614266395568848, "grad_norm": 0.024943850934505463, "global_step": 3906, "epoch": 108, "lr": 0.001} +{"train_loss": 0.03736083582043648, "train_loss_bc": 0.03736083582043648, "train_loss_llm": 0.0, "grad_norm": 0.03270530328154564, "global_step": 3907, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04598326236009598, "train_loss_bc": 0.04598326236009598, "train_loss_llm": 0.0, "grad_norm": 0.03856039419770241, "global_step": 3908, "epoch": 108, "lr": 0.001} +{"train_loss": 0.03751164302229881, "train_loss_bc": 0.03751164302229881, "train_loss_llm": 0.0, "grad_norm": 0.01801200397312641, "global_step": 3909, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04120076075196266, "train_loss_bc": 0.04120076075196266, "train_loss_llm": 0.0, "grad_norm": 0.02520793490111828, "global_step": 3910, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04335569962859154, "train_loss_bc": 0.04334825277328491, "train_loss_llm": 0.0007445216178894043, "grad_norm": 0.03344394639134407, "global_step": 3911, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04605266451835632, "train_loss_bc": 0.04605266451835632, "train_loss_llm": 0.0, "grad_norm": 0.040429938584566116, "global_step": 3912, "epoch": 108, "lr": 0.001} +{"train_loss": 0.044212456792593, "train_loss_bc": 0.04420841112732887, "train_loss_llm": 0.0004044473171234131, "grad_norm": 0.01620659977197647, "global_step": 3913, "epoch": 108, "lr": 0.001} +{"train_loss": 0.0355670191347599, "train_loss_bc": 0.0355575829744339, "train_loss_llm": 0.0009436309337615967, "grad_norm": 0.021510150283575058, "global_step": 3914, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04255341365933418, "train_loss_bc": 0.04253800958395004, "train_loss_llm": 0.0015405118465423584, "grad_norm": 0.027149556204676628, "global_step": 3915, "epoch": 108, "lr": 0.001} +{"train_loss": 0.046081189066171646, "train_loss_bc": 0.046081189066171646, "train_loss_llm": 0.0, "grad_norm": 0.03300010412931442, "global_step": 3916, "epoch": 108, "lr": 0.001} +{"train_loss": 0.0422617606818676, "train_loss_bc": 0.0422617606818676, "train_loss_llm": 0.0, "grad_norm": 0.01681138575077057, "global_step": 3917, "epoch": 108, "lr": 0.001} +{"train_loss": 0.045233823359012604, "train_loss_bc": 0.045233823359012604, "train_loss_llm": 0.0, "grad_norm": 0.02544134482741356, "global_step": 3918, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04709947854280472, "train_loss_bc": 0.04709947854280472, "train_loss_llm": 0.0, "grad_norm": 0.03493676707148552, "global_step": 3919, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04260779544711113, "train_loss_bc": 0.04260779544711113, "train_loss_llm": 0.0, "grad_norm": 0.045724671334028244, "global_step": 3920, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04412592574954033, "train_loss_bc": 0.04412592574954033, "train_loss_llm": 0.0, "grad_norm": 0.021885741502046585, "global_step": 3921, "epoch": 108, "lr": 0.001} +{"train_loss": 0.0464223213493824, "train_loss_bc": 0.046405043452978134, "train_loss_llm": 0.0017277896404266357, "grad_norm": 0.028482886031270027, "global_step": 3922, "epoch": 108, "lr": 0.001} +{"train_loss": 0.04304797719750139, "train_loss_bc": 0.039005883038043976, "train_loss_llm": 0.0, "grad_norm": 0.043218955397605896, "global_step": 3923, "epoch": 108, "lr": 0.001, "val_loss": 0.052569590508937836} +{"train_loss": 0.04320601001381874, "train_loss_bc": 0.04320601001381874, "train_loss_llm": 0.0, "grad_norm": 0.05215084180235863, "global_step": 3924, "epoch": 109, "lr": 0.001} +{"train_loss": 0.038356900215148926, "train_loss_bc": 0.03833194077014923, "train_loss_llm": 0.00249597430229187, "grad_norm": 0.02085486613214016, "global_step": 3925, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04149951785802841, "train_loss_bc": 0.04149605706334114, "train_loss_llm": 0.00034615397453308105, "grad_norm": 0.03172016516327858, "global_step": 3926, "epoch": 109, "lr": 0.001} +{"train_loss": 0.039762575179338455, "train_loss_bc": 0.039762575179338455, "train_loss_llm": 0.0, "grad_norm": 0.0435350202023983, "global_step": 3927, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04223954677581787, "train_loss_bc": 0.04223954677581787, "train_loss_llm": 0.0, "grad_norm": 0.049468182027339935, "global_step": 3928, "epoch": 109, "lr": 0.001} +{"train_loss": 0.03842613101005554, "train_loss_bc": 0.03842613101005554, "train_loss_llm": 0.0, "grad_norm": 0.017782190814614296, "global_step": 3929, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04110737517476082, "train_loss_bc": 0.04110737517476082, "train_loss_llm": 0.0, "grad_norm": 0.024274788796901703, "global_step": 3930, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04172004759311676, "train_loss_bc": 0.04171926900744438, "train_loss_llm": 7.793307304382324e-05, "grad_norm": 0.03039429523050785, "global_step": 3931, "epoch": 109, "lr": 0.001} +{"train_loss": 0.048451460897922516, "train_loss_bc": 0.048451460897922516, "train_loss_llm": 0.0, "grad_norm": 0.03910364955663681, "global_step": 3932, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04032111540436745, "train_loss_bc": 0.04032111540436745, "train_loss_llm": 0.0, "grad_norm": 0.014619586989283562, "global_step": 3933, "epoch": 109, "lr": 0.001} +{"train_loss": 0.03880675137042999, "train_loss_bc": 0.03880675137042999, "train_loss_llm": 0.0, "grad_norm": 0.02005680464208126, "global_step": 3934, "epoch": 109, "lr": 0.001} +{"train_loss": 0.043068237602710724, "train_loss_bc": 0.043068237602710724, "train_loss_llm": 0.0, "grad_norm": 0.03004043735563755, "global_step": 3935, "epoch": 109, "lr": 0.001} +{"train_loss": 0.048533856868743896, "train_loss_bc": 0.048533856868743896, "train_loss_llm": 0.0, "grad_norm": 0.0398482121527195, "global_step": 3936, "epoch": 109, "lr": 0.001} +{"train_loss": 0.044109586626291275, "train_loss_bc": 0.044109586626291275, "train_loss_llm": 0.0, "grad_norm": 0.018585335463285446, "global_step": 3937, "epoch": 109, "lr": 0.001} +{"train_loss": 0.050056107342243195, "train_loss_bc": 0.05004046484827995, "train_loss_llm": 0.0015641748905181885, "grad_norm": 0.029686249792575836, "global_step": 3938, "epoch": 109, "lr": 0.001} +{"train_loss": 0.040379077196121216, "train_loss_bc": 0.040365103632211685, "train_loss_llm": 0.001397460699081421, "grad_norm": 0.041464801877737045, "global_step": 3939, "epoch": 109, "lr": 0.001} +{"train_loss": 0.039263270795345306, "train_loss_bc": 0.03923362120985985, "train_loss_llm": 0.0029650628566741943, "grad_norm": 0.05175645276904106, "global_step": 3940, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04435114562511444, "train_loss_bc": 0.044349782168865204, "train_loss_llm": 0.00013643503189086914, "grad_norm": 0.019816912710666656, "global_step": 3941, "epoch": 109, "lr": 0.001} +{"train_loss": 0.050189655274152756, "train_loss_bc": 0.050189655274152756, "train_loss_llm": 0.0, "grad_norm": 0.03774230182170868, "global_step": 3942, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04323434829711914, "train_loss_bc": 0.0432303287088871, "train_loss_llm": 0.00040203332901000977, "grad_norm": 0.04945217818021774, "global_step": 3943, "epoch": 109, "lr": 0.001} +{"train_loss": 0.050814129412174225, "train_loss_bc": 0.050814129412174225, "train_loss_llm": 0.0, "grad_norm": 0.06624430418014526, "global_step": 3944, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04299597814679146, "train_loss_bc": 0.042986851185560226, "train_loss_llm": 0.0009128749370574951, "grad_norm": 0.015806015580892563, "global_step": 3945, "epoch": 109, "lr": 0.001} +{"train_loss": 0.0335787758231163, "train_loss_bc": 0.03357198089361191, "train_loss_llm": 0.0006794631481170654, "grad_norm": 0.021282624453306198, "global_step": 3946, "epoch": 109, "lr": 0.001} +{"train_loss": 0.041170384734869, "train_loss_bc": 0.04115888476371765, "train_loss_llm": 0.0011498630046844482, "grad_norm": 0.028093595057725906, "global_step": 3947, "epoch": 109, "lr": 0.001} +{"train_loss": 0.040233906358480453, "train_loss_bc": 0.040233906358480453, "train_loss_llm": 0.0, "grad_norm": 0.034613966941833496, "global_step": 3948, "epoch": 109, "lr": 0.001} +{"train_loss": 0.044487904757261276, "train_loss_bc": 0.044487904757261276, "train_loss_llm": 0.0, "grad_norm": 0.015597113408148289, "global_step": 3949, "epoch": 109, "lr": 0.001} +{"train_loss": 0.03973224386572838, "train_loss_bc": 0.03973224386572838, "train_loss_llm": 0.0, "grad_norm": 0.023828420788049698, "global_step": 3950, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04310900717973709, "train_loss_bc": 0.04310900717973709, "train_loss_llm": 0.0, "grad_norm": 0.033567678183317184, "global_step": 3951, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04459433630108833, "train_loss_bc": 0.04459433630108833, "train_loss_llm": 0.0, "grad_norm": 0.03974197804927826, "global_step": 3952, "epoch": 109, "lr": 0.001} +{"train_loss": 0.03836798295378685, "train_loss_bc": 0.03836798295378685, "train_loss_llm": 0.0, "grad_norm": 0.01775887981057167, "global_step": 3953, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04232745245099068, "train_loss_bc": 0.04232557862997055, "train_loss_llm": 0.00018745660781860352, "grad_norm": 0.024246947839856148, "global_step": 3954, "epoch": 109, "lr": 0.001} +{"train_loss": 0.0515255406498909, "train_loss_bc": 0.05147404223680496, "train_loss_llm": 0.005149662494659424, "grad_norm": 0.03674549609422684, "global_step": 3955, "epoch": 109, "lr": 0.001} +{"train_loss": 0.042610086500644684, "train_loss_bc": 0.04260476678609848, "train_loss_llm": 0.0005319118499755859, "grad_norm": 0.04889475926756859, "global_step": 3956, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04676702618598938, "train_loss_bc": 0.04676702618598938, "train_loss_llm": 0.0, "grad_norm": 0.0198757816106081, "global_step": 3957, "epoch": 109, "lr": 0.001} +{"train_loss": 0.04175089672207832, "train_loss_bc": 0.04175089672207832, "train_loss_llm": 0.0, "grad_norm": 0.03574874624609947, "global_step": 3958, "epoch": 109, "lr": 0.001} +{"train_loss": 0.043327747430238456, "train_loss_bc": 0.05865053832530975, "train_loss_llm": 0.0, "grad_norm": 0.04834631830453873, "global_step": 3959, "epoch": 109, "lr": 0.001, "val_loss": 0.05284992605447769} +{"train_loss": 0.04025883972644806, "train_loss_bc": 0.04025883972644806, "train_loss_llm": 0.0, "grad_norm": 0.05526787042617798, "global_step": 3960, "epoch": 110, "lr": 0.001} +{"train_loss": 0.03398989140987396, "train_loss_bc": 0.03398957848548889, "train_loss_llm": 3.126263618469238e-05, "grad_norm": 0.01845216378569603, "global_step": 3961, "epoch": 110, "lr": 0.001} +{"train_loss": 0.05226510018110275, "train_loss_bc": 0.05226510018110275, "train_loss_llm": 0.0, "grad_norm": 0.031320180743932724, "global_step": 3962, "epoch": 110, "lr": 0.001} +{"train_loss": 0.039498813450336456, "train_loss_bc": 0.039498813450336456, "train_loss_llm": 0.0, "grad_norm": 0.04067792370915413, "global_step": 3963, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04106684774160385, "train_loss_bc": 0.04106684774160385, "train_loss_llm": 0.0, "grad_norm": 0.041875507682561874, "global_step": 3964, "epoch": 110, "lr": 0.001} +{"train_loss": 0.0416993647813797, "train_loss_bc": 0.0416993647813797, "train_loss_llm": 0.0, "grad_norm": 0.018434196710586548, "global_step": 3965, "epoch": 110, "lr": 0.001} +{"train_loss": 0.036705296486616135, "train_loss_bc": 0.036705296486616135, "train_loss_llm": 0.0, "grad_norm": 0.021375861018896103, "global_step": 3966, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04789268597960472, "train_loss_bc": 0.047887906432151794, "train_loss_llm": 0.0004779696464538574, "grad_norm": 0.026764998212456703, "global_step": 3967, "epoch": 110, "lr": 0.001} +{"train_loss": 0.049351032823324203, "train_loss_bc": 0.04933594539761543, "train_loss_llm": 0.0015088915824890137, "grad_norm": 0.03319171816110611, "global_step": 3968, "epoch": 110, "lr": 0.001} +{"train_loss": 0.03859684616327286, "train_loss_bc": 0.03859419375658035, "train_loss_llm": 0.0002651214599609375, "grad_norm": 0.014680548571050167, "global_step": 3969, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04317796975374222, "train_loss_bc": 0.043172404170036316, "train_loss_llm": 0.0005567073822021484, "grad_norm": 0.02130252867937088, "global_step": 3970, "epoch": 110, "lr": 0.001} +{"train_loss": 0.035326018929481506, "train_loss_bc": 0.035326018929481506, "train_loss_llm": 0.0, "grad_norm": 0.027376240119338036, "global_step": 3971, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04340619593858719, "train_loss_bc": 0.043399378657341, "train_loss_llm": 0.0006816983222961426, "grad_norm": 0.03629273548722267, "global_step": 3972, "epoch": 110, "lr": 0.001} +{"train_loss": 0.046048376709222794, "train_loss_bc": 0.046048376709222794, "train_loss_llm": 0.0, "grad_norm": 0.013946869410574436, "global_step": 3973, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04357832670211792, "train_loss_bc": 0.04357696697115898, "train_loss_llm": 0.0001360476016998291, "grad_norm": 0.02943943440914154, "global_step": 3974, "epoch": 110, "lr": 0.001} +{"train_loss": 0.03222999721765518, "train_loss_bc": 0.03222540393471718, "train_loss_llm": 0.00045949220657348633, "grad_norm": 0.03500603511929512, "global_step": 3975, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04954284802079201, "train_loss_bc": 0.04951133951544762, "train_loss_llm": 0.0031507909297943115, "grad_norm": 0.04771733283996582, "global_step": 3976, "epoch": 110, "lr": 0.001} +{"train_loss": 0.05008001625537872, "train_loss_bc": 0.05008001625537872, "train_loss_llm": 0.0, "grad_norm": 0.016043556854128838, "global_step": 3977, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04330578073859215, "train_loss_bc": 0.04330578073859215, "train_loss_llm": 0.0, "grad_norm": 0.025125982239842415, "global_step": 3978, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04524034261703491, "train_loss_bc": 0.04524034261703491, "train_loss_llm": 0.0, "grad_norm": 0.04056916758418083, "global_step": 3979, "epoch": 110, "lr": 0.001} +{"train_loss": 0.046139463782310486, "train_loss_bc": 0.046139463782310486, "train_loss_llm": 0.0, "grad_norm": 0.0557936355471611, "global_step": 3980, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04215560853481293, "train_loss_bc": 0.04215560853481293, "train_loss_llm": 0.0, "grad_norm": 0.01826424151659012, "global_step": 3981, "epoch": 110, "lr": 0.001} +{"train_loss": 0.049326956272125244, "train_loss_bc": 0.049326956272125244, "train_loss_llm": 0.0, "grad_norm": 0.03292560949921608, "global_step": 3982, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04591920226812363, "train_loss_bc": 0.04591920226812363, "train_loss_llm": 0.0, "grad_norm": 0.04121571406722069, "global_step": 3983, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04545741155743599, "train_loss_bc": 0.04545705392956734, "train_loss_llm": 3.591179847717285e-05, "grad_norm": 0.05111268162727356, "global_step": 3984, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04229862615466118, "train_loss_bc": 0.04229862615466118, "train_loss_llm": 0.0, "grad_norm": 0.012351210229098797, "global_step": 3985, "epoch": 110, "lr": 0.001} +{"train_loss": 0.038130197674036026, "train_loss_bc": 0.038130197674036026, "train_loss_llm": 0.0, "grad_norm": 0.027280621230602264, "global_step": 3986, "epoch": 110, "lr": 0.001} +{"train_loss": 0.055396053940057755, "train_loss_bc": 0.055396053940057755, "train_loss_llm": 0.0, "grad_norm": 0.0328744538128376, "global_step": 3987, "epoch": 110, "lr": 0.001} +{"train_loss": 0.0460980124771595, "train_loss_bc": 0.04606190323829651, "train_loss_llm": 0.003610968589782715, "grad_norm": 0.04125737026333809, "global_step": 3988, "epoch": 110, "lr": 0.001} +{"train_loss": 0.033559322357177734, "train_loss_bc": 0.03353998064994812, "train_loss_llm": 0.0019342601299285889, "grad_norm": 0.015082485973834991, "global_step": 3989, "epoch": 110, "lr": 0.001} +{"train_loss": 0.043856896460056305, "train_loss_bc": 0.043856896460056305, "train_loss_llm": 0.0, "grad_norm": 0.023125924170017242, "global_step": 3990, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04025810584425926, "train_loss_bc": 0.04024609923362732, "train_loss_llm": 0.0012008249759674072, "grad_norm": 0.03033711388707161, "global_step": 3991, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04071621969342232, "train_loss_bc": 0.040679678320884705, "train_loss_llm": 0.00365409255027771, "grad_norm": 0.03759212791919708, "global_step": 3992, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04945272207260132, "train_loss_bc": 0.04945272207260132, "train_loss_llm": 0.0, "grad_norm": 0.017877111211419106, "global_step": 3993, "epoch": 110, "lr": 0.001} +{"train_loss": 0.038311950862407684, "train_loss_bc": 0.038311950862407684, "train_loss_llm": 0.0, "grad_norm": 0.02766272984445095, "global_step": 3994, "epoch": 110, "lr": 0.001} +{"train_loss": 0.04303650495906671, "train_loss_bc": 0.03897683694958687, "train_loss_llm": 0.0, "grad_norm": 0.03170117363333702, "global_step": 3995, "epoch": 110, "lr": 0.001, "train/cumulative_reward": 78.32963449052328, "train/mean_score": 6.590886171115082, "train/success_rate": 0.2, "test/cumulative_reward": 61.6395579346864, "test/mean_score": 7.691017621038127, "test/success_rate": 0.16, "val_loss": 0.05244708061218262, "train_action_mse_error": 0.046684857457876205} +{"train_loss": 0.03888067230582237, "train_loss_bc": 0.03887220472097397, "train_loss_llm": 0.0008469223976135254, "grad_norm": 0.04153477028012276, "global_step": 3996, "epoch": 111, "lr": 0.001} +{"train_loss": 0.03917215019464493, "train_loss_bc": 0.03916975110769272, "train_loss_llm": 0.00023981928825378418, "grad_norm": 0.01748225837945938, "global_step": 3997, "epoch": 111, "lr": 0.001} +{"train_loss": 0.050898052752017975, "train_loss_bc": 0.050882622599601746, "train_loss_llm": 0.001543194055557251, "grad_norm": 0.023301150649785995, "global_step": 3998, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04223273694515228, "train_loss_bc": 0.04223273694515228, "train_loss_llm": 0.0, "grad_norm": 0.033247169107198715, "global_step": 3999, "epoch": 111, "lr": 0.001} +{"train_loss": 0.045303426682949066, "train_loss_bc": 0.045303426682949066, "train_loss_llm": 0.0, "grad_norm": 0.03910250961780548, "global_step": 4000, "epoch": 111, "lr": 0.001} +{"train_loss": 0.045457128435373306, "train_loss_bc": 0.045457128435373306, "train_loss_llm": 0.0, "grad_norm": 0.01878213696181774, "global_step": 4001, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04182279482483864, "train_loss_bc": 0.04182279482483864, "train_loss_llm": 0.0, "grad_norm": 0.031760577112436295, "global_step": 4002, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04170665144920349, "train_loss_bc": 0.04170581325888634, "train_loss_llm": 8.377432823181152e-05, "grad_norm": 0.044165775179862976, "global_step": 4003, "epoch": 111, "lr": 0.001} +{"train_loss": 0.0415920726954937, "train_loss_bc": 0.04158986359834671, "train_loss_llm": 0.00022080540657043457, "grad_norm": 0.05291292816400528, "global_step": 4004, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04210083931684494, "train_loss_bc": 0.042055800557136536, "train_loss_llm": 0.004503846168518066, "grad_norm": 0.01992061547935009, "global_step": 4005, "epoch": 111, "lr": 0.001} +{"train_loss": 0.05088837817311287, "train_loss_bc": 0.05086704343557358, "train_loss_llm": 0.002133578062057495, "grad_norm": 0.032229866832494736, "global_step": 4006, "epoch": 111, "lr": 0.001} +{"train_loss": 0.05041899159550667, "train_loss_bc": 0.05041899159550667, "train_loss_llm": 0.0, "grad_norm": 0.039083145558834076, "global_step": 4007, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04650578647851944, "train_loss_bc": 0.04650578647851944, "train_loss_llm": 0.0, "grad_norm": 0.045237116515636444, "global_step": 4008, "epoch": 111, "lr": 0.001} +{"train_loss": 0.03819110617041588, "train_loss_bc": 0.03819110617041588, "train_loss_llm": 0.0, "grad_norm": 0.0168094951659441, "global_step": 4009, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04689569026231766, "train_loss_bc": 0.046881407499313354, "train_loss_llm": 0.0014284253120422363, "grad_norm": 0.025396864861249924, "global_step": 4010, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04526901617646217, "train_loss_bc": 0.04525764286518097, "train_loss_llm": 0.001137465238571167, "grad_norm": 0.03769722208380699, "global_step": 4011, "epoch": 111, "lr": 0.001} +{"train_loss": 0.044462900608778, "train_loss_bc": 0.044433463364839554, "train_loss_llm": 0.0029438436031341553, "grad_norm": 0.04977777972817421, "global_step": 4012, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04784996807575226, "train_loss_bc": 0.04784996807575226, "train_loss_llm": 0.0, "grad_norm": 0.019026849418878555, "global_step": 4013, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04192359745502472, "train_loss_bc": 0.04192359745502472, "train_loss_llm": 0.0, "grad_norm": 0.027814948931336403, "global_step": 4014, "epoch": 111, "lr": 0.001} +{"train_loss": 0.03914043307304382, "train_loss_bc": 0.03914043307304382, "train_loss_llm": 0.0, "grad_norm": 0.03396500274538994, "global_step": 4015, "epoch": 111, "lr": 0.001} +{"train_loss": 0.046991411596536636, "train_loss_bc": 0.046991411596536636, "train_loss_llm": 0.0, "grad_norm": 0.04591776430606842, "global_step": 4016, "epoch": 111, "lr": 0.001} +{"train_loss": 0.040556177496910095, "train_loss_bc": 0.04053304344415665, "train_loss_llm": 0.002313375473022461, "grad_norm": 0.018996838480234146, "global_step": 4017, "epoch": 111, "lr": 0.001} +{"train_loss": 0.03311265632510185, "train_loss_bc": 0.03309564292430878, "train_loss_llm": 0.001701265573501587, "grad_norm": 0.028626909479498863, "global_step": 4018, "epoch": 111, "lr": 0.001} +{"train_loss": 0.048719439655542374, "train_loss_bc": 0.04871807619929314, "train_loss_llm": 0.00013616681098937988, "grad_norm": 0.043182406574487686, "global_step": 4019, "epoch": 111, "lr": 0.001} +{"train_loss": 0.039809953421354294, "train_loss_bc": 0.039809953421354294, "train_loss_llm": 0.0, "grad_norm": 0.0481906533241272, "global_step": 4020, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04258818179368973, "train_loss_bc": 0.04258818179368973, "train_loss_llm": 0.0, "grad_norm": 0.022183142602443695, "global_step": 4021, "epoch": 111, "lr": 0.001} +{"train_loss": 0.03790997341275215, "train_loss_bc": 0.03790997341275215, "train_loss_llm": 0.0, "grad_norm": 0.026982659474015236, "global_step": 4022, "epoch": 111, "lr": 0.001} +{"train_loss": 0.03674250841140747, "train_loss_bc": 0.03673969581723213, "train_loss_llm": 0.00028121471405029297, "grad_norm": 0.03395334631204605, "global_step": 4023, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04150844365358353, "train_loss_bc": 0.041499242186546326, "train_loss_llm": 0.0009202361106872559, "grad_norm": 0.04648721590638161, "global_step": 4024, "epoch": 111, "lr": 0.001} +{"train_loss": 0.043715741485357285, "train_loss_bc": 0.043715741485357285, "train_loss_llm": 0.0, "grad_norm": 0.02017069421708584, "global_step": 4025, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04669354483485222, "train_loss_bc": 0.04669354483485222, "train_loss_llm": 0.0, "grad_norm": 0.030759045854210854, "global_step": 4026, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04278959333896637, "train_loss_bc": 0.04278959333896637, "train_loss_llm": 0.0, "grad_norm": 0.04098875820636749, "global_step": 4027, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04157640412449837, "train_loss_bc": 0.041539013385772705, "train_loss_llm": 0.0037389397621154785, "grad_norm": 0.044217415153980255, "global_step": 4028, "epoch": 111, "lr": 0.001} +{"train_loss": 0.0422012135386467, "train_loss_bc": 0.0422012135386467, "train_loss_llm": 0.0, "grad_norm": 0.016126031056046486, "global_step": 4029, "epoch": 111, "lr": 0.001} +{"train_loss": 0.042598381638526917, "train_loss_bc": 0.04259360209107399, "train_loss_llm": 0.0004779994487762451, "grad_norm": 0.025910114869475365, "global_step": 4030, "epoch": 111, "lr": 0.001} +{"train_loss": 0.04318777502824863, "train_loss_bc": 0.04653388261795044, "train_loss_llm": 0.0, "grad_norm": 0.03926270082592964, "global_step": 4031, "epoch": 111, "lr": 0.001, "val_loss": 0.052868202328681946} +{"train_loss": 0.040970370173454285, "train_loss_bc": 0.04096201807260513, "train_loss_llm": 0.0008350610733032227, "grad_norm": 0.05063166469335556, "global_step": 4032, "epoch": 112, "lr": 0.001} +{"train_loss": 0.03881194069981575, "train_loss_bc": 0.03881194069981575, "train_loss_llm": 0.0, "grad_norm": 0.01586749590933323, "global_step": 4033, "epoch": 112, "lr": 0.001} +{"train_loss": 0.04370477423071861, "train_loss_bc": 0.0436982586979866, "train_loss_llm": 0.0006515383720397949, "grad_norm": 0.027426978573203087, "global_step": 4034, "epoch": 112, "lr": 0.001} +{"train_loss": 0.049438852816820145, "train_loss_bc": 0.049438852816820145, "train_loss_llm": 0.0, "grad_norm": 0.038646478205919266, "global_step": 4035, "epoch": 112, "lr": 0.001} +{"train_loss": 0.05063356086611748, "train_loss_bc": 0.05063356086611748, "train_loss_llm": 0.0, "grad_norm": 0.044808004051446915, "global_step": 4036, "epoch": 112, "lr": 0.001} +{"train_loss": 0.037023525685071945, "train_loss_bc": 0.037013404071331024, "train_loss_llm": 0.0010120868682861328, "grad_norm": 0.018757445737719536, "global_step": 4037, "epoch": 112, "lr": 0.001} +{"train_loss": 0.039345093071460724, "train_loss_bc": 0.0393301397562027, "train_loss_llm": 0.001495361328125, "grad_norm": 0.026965226978063583, "global_step": 4038, "epoch": 112, "lr": 0.001} +{"train_loss": 0.038902487605810165, "train_loss_bc": 0.038902487605810165, "train_loss_llm": 0.0, "grad_norm": 0.03903285786509514, "global_step": 4039, "epoch": 112, "lr": 0.001} +{"train_loss": 0.053825873881578445, "train_loss_bc": 0.053825873881578445, "train_loss_llm": 0.0, "grad_norm": 0.048461299389600754, "global_step": 4040, "epoch": 112, "lr": 0.001} +{"train_loss": 0.0384419746696949, "train_loss_bc": 0.0384419746696949, "train_loss_llm": 0.0, "grad_norm": 0.01968166045844555, "global_step": 4041, "epoch": 112, "lr": 0.001} +{"train_loss": 0.037555668503046036, "train_loss_bc": 0.037521522492170334, "train_loss_llm": 0.0034146010875701904, "grad_norm": 0.029746314510703087, "global_step": 4042, "epoch": 112, "lr": 0.001} +{"train_loss": 0.039292771369218826, "train_loss_bc": 0.039292771369218826, "train_loss_llm": 0.0, "grad_norm": 0.03751266002655029, "global_step": 4043, "epoch": 112, "lr": 0.001} +{"train_loss": 0.048722486943006516, "train_loss_bc": 0.048722486943006516, "train_loss_llm": 0.0, "grad_norm": 0.05081583932042122, "global_step": 4044, "epoch": 112, "lr": 0.001} +{"train_loss": 0.052648816257715225, "train_loss_bc": 0.05262254178524017, "train_loss_llm": 0.0026274025440216064, "grad_norm": 0.017409037798643112, "global_step": 4045, "epoch": 112, "lr": 0.001} +{"train_loss": 0.050940364599227905, "train_loss_bc": 0.050940364599227905, "train_loss_llm": 0.0, "grad_norm": 0.032170914113521576, "global_step": 4046, "epoch": 112, "lr": 0.001} +{"train_loss": 0.038034744560718536, "train_loss_bc": 0.038034744560718536, "train_loss_llm": 0.0, "grad_norm": 0.039838433265686035, "global_step": 4047, "epoch": 112, "lr": 0.001} +{"train_loss": 0.04442182183265686, "train_loss_bc": 0.04442182183265686, "train_loss_llm": 0.0, "grad_norm": 0.047808028757572174, "global_step": 4048, "epoch": 112, "lr": 0.001} +{"train_loss": 0.04275158420205116, "train_loss_bc": 0.04271189868450165, "train_loss_llm": 0.003968596458435059, "grad_norm": 0.021081412211060524, "global_step": 4049, "epoch": 112, "lr": 0.001} +{"train_loss": 0.03983868286013603, "train_loss_bc": 0.03983868286013603, "train_loss_llm": 0.0, "grad_norm": 0.023118702694773674, "global_step": 4050, "epoch": 112, "lr": 0.001} +{"train_loss": 0.045024849474430084, "train_loss_bc": 0.045024849474430084, "train_loss_llm": 0.0, "grad_norm": 0.03074558638036251, "global_step": 4051, "epoch": 112, "lr": 0.001} +{"train_loss": 0.043763667345047, "train_loss_bc": 0.04373310133814812, "train_loss_llm": 0.003056764602661133, "grad_norm": 0.03846075385808945, "global_step": 4052, "epoch": 112, "lr": 0.001} +{"train_loss": 0.042527202516794205, "train_loss_bc": 0.042524706572294235, "train_loss_llm": 0.00024965405464172363, "grad_norm": 0.016105804592370987, "global_step": 4053, "epoch": 112, "lr": 0.001} +{"train_loss": 0.041701048612594604, "train_loss_bc": 0.041701048612594604, "train_loss_llm": 0.0, "grad_norm": 0.022240562364459038, "global_step": 4054, "epoch": 112, "lr": 0.001} +{"train_loss": 0.03690772503614426, "train_loss_bc": 0.03690772503614426, "train_loss_llm": 0.0, "grad_norm": 0.027812248095870018, "global_step": 4055, "epoch": 112, "lr": 0.001} +{"train_loss": 0.03313109278678894, "train_loss_bc": 0.03313109278678894, "train_loss_llm": 0.0, "grad_norm": 0.03598836064338684, "global_step": 4056, "epoch": 112, "lr": 0.001} +{"train_loss": 0.043675556778907776, "train_loss_bc": 0.04365547001361847, "train_loss_llm": 0.0020087361335754395, "grad_norm": 0.013558763079345226, "global_step": 4057, "epoch": 112, "lr": 0.001} +{"train_loss": 0.03396706283092499, "train_loss_bc": 0.03396706283092499, "train_loss_llm": 0.0, "grad_norm": 0.025519611313939095, "global_step": 4058, "epoch": 112, "lr": 0.001} +{"train_loss": 0.0480269193649292, "train_loss_bc": 0.0480269193649292, "train_loss_llm": 0.0, "grad_norm": 0.03180006891489029, "global_step": 4059, "epoch": 112, "lr": 0.001} +{"train_loss": 0.049299776554107666, "train_loss_bc": 0.049299776554107666, "train_loss_llm": 0.0, "grad_norm": 0.03829670324921608, "global_step": 4060, "epoch": 112, "lr": 0.001} +{"train_loss": 0.04212162271142006, "train_loss_bc": 0.04211321100592613, "train_loss_llm": 0.0008411407470703125, "grad_norm": 0.014034693129360676, "global_step": 4061, "epoch": 112, "lr": 0.001} +{"train_loss": 0.04401770606637001, "train_loss_bc": 0.04401770606637001, "train_loss_llm": 0.0, "grad_norm": 0.023331759497523308, "global_step": 4062, "epoch": 112, "lr": 0.001} +{"train_loss": 0.04466446116566658, "train_loss_bc": 0.04466446116566658, "train_loss_llm": 0.0, "grad_norm": 0.03916338086128235, "global_step": 4063, "epoch": 112, "lr": 0.001} +{"train_loss": 0.04304051026701927, "train_loss_bc": 0.04300596937537193, "train_loss_llm": 0.00345417857170105, "grad_norm": 0.039699215441942215, "global_step": 4064, "epoch": 112, "lr": 0.001} +{"train_loss": 0.04578104987740517, "train_loss_bc": 0.04578104987740517, "train_loss_llm": 0.0, "grad_norm": 0.013134375214576721, "global_step": 4065, "epoch": 112, "lr": 0.001} +{"train_loss": 0.04316624999046326, "train_loss_bc": 0.04316624999046326, "train_loss_llm": 0.0, "grad_norm": 0.020369600504636765, "global_step": 4066, "epoch": 112, "lr": 0.001} +{"train_loss": 0.04303704181479083, "train_loss_bc": 0.04321160912513733, "train_loss_llm": 0.0, "grad_norm": 0.035001225769519806, "global_step": 4067, "epoch": 112, "lr": 0.001, "val_loss": 0.05260670185089111} +{"train_loss": 0.042686108499765396, "train_loss_bc": 0.042686108499765396, "train_loss_llm": 0.0, "grad_norm": 0.04242979362607002, "global_step": 4068, "epoch": 113, "lr": 0.001} +{"train_loss": 0.05099574476480484, "train_loss_bc": 0.05099260434508324, "train_loss_llm": 0.0003139376640319824, "grad_norm": 0.01486548688262701, "global_step": 4069, "epoch": 113, "lr": 0.001} +{"train_loss": 0.05008453130722046, "train_loss_bc": 0.05008453130722046, "train_loss_llm": 0.0, "grad_norm": 0.02970331907272339, "global_step": 4070, "epoch": 113, "lr": 0.001} +{"train_loss": 0.0469062440097332, "train_loss_bc": 0.0469062440097332, "train_loss_llm": 0.0, "grad_norm": 0.043723735958337784, "global_step": 4071, "epoch": 113, "lr": 0.001} +{"train_loss": 0.031596455723047256, "train_loss_bc": 0.03156770393252373, "train_loss_llm": 0.0028753578662872314, "grad_norm": 0.05727871134877205, "global_step": 4072, "epoch": 113, "lr": 0.001} +{"train_loss": 0.047791119664907455, "train_loss_bc": 0.047790348529815674, "train_loss_llm": 7.69495964050293e-05, "grad_norm": 0.023426318541169167, "global_step": 4073, "epoch": 113, "lr": 0.001} +{"train_loss": 0.044408801943063736, "train_loss_bc": 0.044408801943063736, "train_loss_llm": 0.0, "grad_norm": 0.0317981094121933, "global_step": 4074, "epoch": 113, "lr": 0.001} +{"train_loss": 0.052780576050281525, "train_loss_bc": 0.052780576050281525, "train_loss_llm": 0.0, "grad_norm": 0.03924267739057541, "global_step": 4075, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04608538746833801, "train_loss_bc": 0.04606971889734268, "train_loss_llm": 0.0015667378902435303, "grad_norm": 0.042671166360378265, "global_step": 4076, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04312979802489281, "train_loss_bc": 0.04311773180961609, "train_loss_llm": 0.0012066066265106201, "grad_norm": 0.015155467204749584, "global_step": 4077, "epoch": 113, "lr": 0.001} +{"train_loss": 0.0409393273293972, "train_loss_bc": 0.040913328528404236, "train_loss_llm": 0.002599775791168213, "grad_norm": 0.025049036368727684, "global_step": 4078, "epoch": 113, "lr": 0.001} +{"train_loss": 0.03987857699394226, "train_loss_bc": 0.03987857699394226, "train_loss_llm": 0.0, "grad_norm": 0.033082444220781326, "global_step": 4079, "epoch": 113, "lr": 0.001} +{"train_loss": 0.03708191215991974, "train_loss_bc": 0.03708191215991974, "train_loss_llm": 0.0, "grad_norm": 0.04074067249894142, "global_step": 4080, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04008882865309715, "train_loss_bc": 0.04008882865309715, "train_loss_llm": 0.0, "grad_norm": 0.019087329506874084, "global_step": 4081, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04599117115139961, "train_loss_bc": 0.04598802328109741, "train_loss_llm": 0.0003146231174468994, "grad_norm": 0.027544382959604263, "global_step": 4082, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04929671809077263, "train_loss_bc": 0.04929671809077263, "train_loss_llm": 0.0, "grad_norm": 0.03923466056585312, "global_step": 4083, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04210079088807106, "train_loss_bc": 0.04210079088807106, "train_loss_llm": 0.0, "grad_norm": 0.04625118151307106, "global_step": 4084, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04176386073231697, "train_loss_bc": 0.04175955802202225, "train_loss_llm": 0.0004302561283111572, "grad_norm": 0.012687750160694122, "global_step": 4085, "epoch": 113, "lr": 0.001} +{"train_loss": 0.035201460123062134, "train_loss_bc": 0.03519844636321068, "train_loss_llm": 0.0003015100955963135, "grad_norm": 0.029708711430430412, "global_step": 4086, "epoch": 113, "lr": 0.001} +{"train_loss": 0.02903556264936924, "train_loss_bc": 0.02903556264936924, "train_loss_llm": 0.0, "grad_norm": 0.04090403392910957, "global_step": 4087, "epoch": 113, "lr": 0.001} +{"train_loss": 0.037956468760967255, "train_loss_bc": 0.03795384243130684, "train_loss_llm": 0.0002624690532684326, "grad_norm": 0.05050784721970558, "global_step": 4088, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04309932887554169, "train_loss_bc": 0.04309932887554169, "train_loss_llm": 0.0, "grad_norm": 0.022759579122066498, "global_step": 4089, "epoch": 113, "lr": 0.001} +{"train_loss": 0.048577550798654556, "train_loss_bc": 0.048562437295913696, "train_loss_llm": 0.0015111863613128662, "grad_norm": 0.03109951689839363, "global_step": 4090, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04279380291700363, "train_loss_bc": 0.04276864975690842, "train_loss_llm": 0.002515256404876709, "grad_norm": 0.04340806230902672, "global_step": 4091, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04480943828821182, "train_loss_bc": 0.04480943828821182, "train_loss_llm": 0.0, "grad_norm": 0.06026490405201912, "global_step": 4092, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04903431981801987, "train_loss_bc": 0.04903431981801987, "train_loss_llm": 0.0, "grad_norm": 0.016926772892475128, "global_step": 4093, "epoch": 113, "lr": 0.001} +{"train_loss": 0.047648996114730835, "train_loss_bc": 0.047648996114730835, "train_loss_llm": 0.0, "grad_norm": 0.02666466496884823, "global_step": 4094, "epoch": 113, "lr": 0.001} +{"train_loss": 0.040690839290618896, "train_loss_bc": 0.04067805036902428, "train_loss_llm": 0.0012790262699127197, "grad_norm": 0.033122215420007706, "global_step": 4095, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04537872597575188, "train_loss_bc": 0.0453646145761013, "train_loss_llm": 0.0014110207557678223, "grad_norm": 0.04258851334452629, "global_step": 4096, "epoch": 113, "lr": 0.001} +{"train_loss": 0.040817707777023315, "train_loss_bc": 0.040817707777023315, "train_loss_llm": 0.0, "grad_norm": 0.01926238089799881, "global_step": 4097, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04110633581876755, "train_loss_bc": 0.04110633581876755, "train_loss_llm": 0.0, "grad_norm": 0.029457509517669678, "global_step": 4098, "epoch": 113, "lr": 0.001} +{"train_loss": 0.044123683124780655, "train_loss_bc": 0.044123683124780655, "train_loss_llm": 0.0, "grad_norm": 0.03359245881438255, "global_step": 4099, "epoch": 113, "lr": 0.001} +{"train_loss": 0.04180126264691353, "train_loss_bc": 0.04180126264691353, "train_loss_llm": 0.0, "grad_norm": 0.04192481189966202, "global_step": 4100, "epoch": 113, "lr": 0.001} +{"train_loss": 0.038841504603624344, "train_loss_bc": 0.038841504603624344, "train_loss_llm": 0.0, "grad_norm": 0.018247632309794426, "global_step": 4101, "epoch": 113, "lr": 0.001} +{"train_loss": 0.038469500839710236, "train_loss_bc": 0.038462236523628235, "train_loss_llm": 0.0007264912128448486, "grad_norm": 0.026057636365294456, "global_step": 4102, "epoch": 113, "lr": 0.001} +{"train_loss": 0.043100449411819376, "train_loss_bc": 0.04862373694777489, "train_loss_llm": 0.0, "grad_norm": 0.033854831010103226, "global_step": 4103, "epoch": 113, "lr": 0.001, "val_loss": 0.05291718617081642} +{"train_loss": 0.04660717025399208, "train_loss_bc": 0.04660717025399208, "train_loss_llm": 0.0, "grad_norm": 0.04305068403482437, "global_step": 4104, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04481355473399162, "train_loss_bc": 0.04481121897697449, "train_loss_llm": 0.00023373961448669434, "grad_norm": 0.012953493744134903, "global_step": 4105, "epoch": 114, "lr": 0.001} +{"train_loss": 0.038928765803575516, "train_loss_bc": 0.038928765803575516, "train_loss_llm": 0.0, "grad_norm": 0.027443308383226395, "global_step": 4106, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04163217172026634, "train_loss_bc": 0.04163160175085068, "train_loss_llm": 5.7131052017211914e-05, "grad_norm": 0.03611597418785095, "global_step": 4107, "epoch": 114, "lr": 0.001} +{"train_loss": 0.044605471193790436, "train_loss_bc": 0.044605471193790436, "train_loss_llm": 0.0, "grad_norm": 0.048321887850761414, "global_step": 4108, "epoch": 114, "lr": 0.001} +{"train_loss": 0.044629886746406555, "train_loss_bc": 0.044629886746406555, "train_loss_llm": 0.0, "grad_norm": 0.020493071526288986, "global_step": 4109, "epoch": 114, "lr": 0.001} +{"train_loss": 0.043213460594415665, "train_loss_bc": 0.043201398104429245, "train_loss_llm": 0.0012060701847076416, "grad_norm": 0.031429264694452286, "global_step": 4110, "epoch": 114, "lr": 0.001} +{"train_loss": 0.05053891986608505, "train_loss_bc": 0.05053891986608505, "train_loss_llm": 0.0, "grad_norm": 0.04641611501574516, "global_step": 4111, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04065796360373497, "train_loss_bc": 0.04065796360373497, "train_loss_llm": 0.0, "grad_norm": 0.062444381415843964, "global_step": 4112, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04174615815281868, "train_loss_bc": 0.04174615815281868, "train_loss_llm": 0.0, "grad_norm": 0.018834877759218216, "global_step": 4113, "epoch": 114, "lr": 0.001} +{"train_loss": 0.049787331372499466, "train_loss_bc": 0.049787331372499466, "train_loss_llm": 0.0, "grad_norm": 0.026217477396130562, "global_step": 4114, "epoch": 114, "lr": 0.001} +{"train_loss": 0.036618754267692566, "train_loss_bc": 0.03659960627555847, "train_loss_llm": 0.0019149184226989746, "grad_norm": 0.03677837550640106, "global_step": 4115, "epoch": 114, "lr": 0.001} +{"train_loss": 0.040034737437963486, "train_loss_bc": 0.03999954089522362, "train_loss_llm": 0.0035195648670196533, "grad_norm": 0.047947078943252563, "global_step": 4116, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04543321207165718, "train_loss_bc": 0.045431848615407944, "train_loss_llm": 0.00013622641563415527, "grad_norm": 0.015311668626964092, "global_step": 4117, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04981957748532295, "train_loss_bc": 0.04980852082371712, "train_loss_llm": 0.0011057555675506592, "grad_norm": 0.023645911365747452, "global_step": 4118, "epoch": 114, "lr": 0.001} +{"train_loss": 0.045742690563201904, "train_loss_bc": 0.045742690563201904, "train_loss_llm": 0.0, "grad_norm": 0.02954276278614998, "global_step": 4119, "epoch": 114, "lr": 0.001} +{"train_loss": 0.03953086957335472, "train_loss_bc": 0.03951319679617882, "train_loss_llm": 0.0017673969268798828, "grad_norm": 0.03440217673778534, "global_step": 4120, "epoch": 114, "lr": 0.001} +{"train_loss": 0.03981741890311241, "train_loss_bc": 0.03981741890311241, "train_loss_llm": 0.0, "grad_norm": 0.021144241094589233, "global_step": 4121, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04019444063305855, "train_loss_bc": 0.04019444063305855, "train_loss_llm": 0.0, "grad_norm": 0.02983182854950428, "global_step": 4122, "epoch": 114, "lr": 0.001} +{"train_loss": 0.05068065598607063, "train_loss_bc": 0.05068065598607063, "train_loss_llm": 0.0, "grad_norm": 0.038128677755594254, "global_step": 4123, "epoch": 114, "lr": 0.001} +{"train_loss": 0.046521373093128204, "train_loss_bc": 0.046521373093128204, "train_loss_llm": 0.0, "grad_norm": 0.05290419980883598, "global_step": 4124, "epoch": 114, "lr": 0.001} +{"train_loss": 0.039053067564964294, "train_loss_bc": 0.039053067564964294, "train_loss_llm": 0.0, "grad_norm": 0.018524620682001114, "global_step": 4125, "epoch": 114, "lr": 0.001} +{"train_loss": 0.0451863631606102, "train_loss_bc": 0.0451863631606102, "train_loss_llm": 0.0, "grad_norm": 0.031357791274785995, "global_step": 4126, "epoch": 114, "lr": 0.001} +{"train_loss": 0.0380425862967968, "train_loss_bc": 0.0380425862967968, "train_loss_llm": 0.0, "grad_norm": 0.040114328265190125, "global_step": 4127, "epoch": 114, "lr": 0.001} +{"train_loss": 0.03636029735207558, "train_loss_bc": 0.03636029735207558, "train_loss_llm": 0.0, "grad_norm": 0.0486118383705616, "global_step": 4128, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04056096076965332, "train_loss_bc": 0.04056096076965332, "train_loss_llm": 0.0, "grad_norm": 0.02176983281970024, "global_step": 4129, "epoch": 114, "lr": 0.001} +{"train_loss": 0.03942963480949402, "train_loss_bc": 0.03942963480949402, "train_loss_llm": 0.0, "grad_norm": 0.03022858500480652, "global_step": 4130, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04363234341144562, "train_loss_bc": 0.04363234341144562, "train_loss_llm": 0.0, "grad_norm": 0.04845914989709854, "global_step": 4131, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04221313074231148, "train_loss_bc": 0.04221313074231148, "train_loss_llm": 0.0, "grad_norm": 0.058327507227659225, "global_step": 4132, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04282541573047638, "train_loss_bc": 0.04280287027359009, "train_loss_llm": 0.0022544264793395996, "grad_norm": 0.01603035442531109, "global_step": 4133, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04673292115330696, "train_loss_bc": 0.046711310744285583, "train_loss_llm": 0.002160876989364624, "grad_norm": 0.03242987394332886, "global_step": 4134, "epoch": 114, "lr": 0.001} +{"train_loss": 0.044886887073516846, "train_loss_bc": 0.04487531632184982, "train_loss_llm": 0.0011569857597351074, "grad_norm": 0.04221576079726219, "global_step": 4135, "epoch": 114, "lr": 0.001} +{"train_loss": 0.03851651772856712, "train_loss_bc": 0.038485679775476456, "train_loss_llm": 0.0030837953090667725, "grad_norm": 0.0455511137843132, "global_step": 4136, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04597359895706177, "train_loss_bc": 0.04597359895706177, "train_loss_llm": 0.0, "grad_norm": 0.017221862450242043, "global_step": 4137, "epoch": 114, "lr": 0.001} +{"train_loss": 0.044418081641197205, "train_loss_bc": 0.044418081641197205, "train_loss_llm": 0.0, "grad_norm": 0.02467343769967556, "global_step": 4138, "epoch": 114, "lr": 0.001} +{"train_loss": 0.04307395157714685, "train_loss_bc": 0.04127586632966995, "train_loss_llm": 0.0, "grad_norm": 0.033190567046403885, "global_step": 4139, "epoch": 114, "lr": 0.001, "val_loss": 0.052988745272159576} +{"train_loss": 0.039487324655056, "train_loss_bc": 0.039487324655056, "train_loss_llm": 0.0, "grad_norm": 0.04341457784175873, "global_step": 4140, "epoch": 115, "lr": 0.001} +{"train_loss": 0.04563141241669655, "train_loss_bc": 0.04563141241669655, "train_loss_llm": 0.0, "grad_norm": 0.01800115592777729, "global_step": 4141, "epoch": 115, "lr": 0.001} +{"train_loss": 0.04749276861548424, "train_loss_bc": 0.04749276861548424, "train_loss_llm": 0.0, "grad_norm": 0.026305722072720528, "global_step": 4142, "epoch": 115, "lr": 0.001} +{"train_loss": 0.03660844266414642, "train_loss_bc": 0.03660844266414642, "train_loss_llm": 0.0, "grad_norm": 0.030686182901263237, "global_step": 4143, "epoch": 115, "lr": 0.001} +{"train_loss": 0.04236946254968643, "train_loss_bc": 0.04236946254968643, "train_loss_llm": 0.0, "grad_norm": 0.037386056035757065, "global_step": 4144, "epoch": 115, "lr": 0.001} +{"train_loss": 0.04690029099583626, "train_loss_bc": 0.04688886180520058, "train_loss_llm": 0.0011428296566009521, "grad_norm": 0.01632075197994709, "global_step": 4145, "epoch": 115, "lr": 0.001} +{"train_loss": 0.04573706537485123, "train_loss_bc": 0.04573706537485123, "train_loss_llm": 0.0, "grad_norm": 0.0220839474350214, "global_step": 4146, "epoch": 115, "lr": 0.001} +{"train_loss": 0.039726030081510544, "train_loss_bc": 0.039726030081510544, "train_loss_llm": 0.0, "grad_norm": 0.03403062745928764, "global_step": 4147, "epoch": 115, "lr": 0.001} +{"train_loss": 0.035789959132671356, "train_loss_bc": 0.035789959132671356, "train_loss_llm": 0.0, "grad_norm": 0.04046124219894409, "global_step": 4148, "epoch": 115, "lr": 0.001} +{"train_loss": 0.041914455592632294, "train_loss_bc": 0.041914455592632294, "train_loss_llm": 0.0, "grad_norm": 0.016892973333597183, "global_step": 4149, "epoch": 115, "lr": 0.001} +{"train_loss": 0.03960384055972099, "train_loss_bc": 0.03960384055972099, "train_loss_llm": 0.0, "grad_norm": 0.024611234664916992, "global_step": 4150, "epoch": 115, "lr": 0.001} +{"train_loss": 0.04390544444322586, "train_loss_bc": 0.043900858610868454, "train_loss_llm": 0.0004584193229675293, "grad_norm": 0.03284890577197075, "global_step": 4151, "epoch": 115, "lr": 0.001} +{"train_loss": 0.046060360968112946, "train_loss_bc": 0.046060360968112946, "train_loss_llm": 0.0, "grad_norm": 0.03631843253970146, "global_step": 4152, "epoch": 115, "lr": 0.001} +{"train_loss": 0.048904675990343094, "train_loss_bc": 0.048904675990343094, "train_loss_llm": 0.0, "grad_norm": 0.01677805371582508, "global_step": 4153, "epoch": 115, "lr": 0.001} +{"train_loss": 0.048181239515542984, "train_loss_bc": 0.048181239515542984, "train_loss_llm": 0.0, "grad_norm": 0.021894557401537895, "global_step": 4154, "epoch": 115, "lr": 0.001} +{"train_loss": 0.044498227536678314, "train_loss_bc": 0.044497571885585785, "train_loss_llm": 6.556510925292969e-05, "grad_norm": 0.02490261197090149, "global_step": 4155, "epoch": 115, "lr": 0.001} +{"train_loss": 0.0423135906457901, "train_loss_bc": 0.04231330007314682, "train_loss_llm": 2.8908252716064453e-05, "grad_norm": 0.027850603684782982, "global_step": 4156, "epoch": 115, "lr": 0.001} +{"train_loss": 0.039534881711006165, "train_loss_bc": 0.03952949494123459, "train_loss_llm": 0.0005385875701904297, "grad_norm": 0.013078566640615463, "global_step": 4157, "epoch": 115, "lr": 0.001} +{"train_loss": 0.045684780925512314, "train_loss_bc": 0.04568133503198624, "train_loss_llm": 0.000344693660736084, "grad_norm": 0.02281155064702034, "global_step": 4158, "epoch": 115, "lr": 0.001} +{"train_loss": 0.0460934117436409, "train_loss_bc": 0.0460934117436409, "train_loss_llm": 0.0, "grad_norm": 0.029740463942289352, "global_step": 4159, "epoch": 115, "lr": 0.001} +{"train_loss": 0.0422646589577198, "train_loss_bc": 0.0422646589577198, "train_loss_llm": 0.0, "grad_norm": 0.033433109521865845, "global_step": 4160, "epoch": 115, "lr": 0.001} +{"train_loss": 0.045768991112709045, "train_loss_bc": 0.045768991112709045, "train_loss_llm": 0.0, "grad_norm": 0.01440398395061493, "global_step": 4161, "epoch": 115, "lr": 0.001} +{"train_loss": 0.047882575541734695, "train_loss_bc": 0.047861550003290176, "train_loss_llm": 0.002102375030517578, "grad_norm": 0.028199221938848495, "global_step": 4162, "epoch": 115, "lr": 0.001} +{"train_loss": 0.03848119452595711, "train_loss_bc": 0.03846414387226105, "train_loss_llm": 0.001705080270767212, "grad_norm": 0.03500369191169739, "global_step": 4163, "epoch": 115, "lr": 0.001} +{"train_loss": 0.03706679493188858, "train_loss_bc": 0.03706679493188858, "train_loss_llm": 0.0, "grad_norm": 0.041135817766189575, "global_step": 4164, "epoch": 115, "lr": 0.001} +{"train_loss": 0.044708073139190674, "train_loss_bc": 0.04470790550112724, "train_loss_llm": 1.665949821472168e-05, "grad_norm": 0.018888823688030243, "global_step": 4165, "epoch": 115, "lr": 0.001} +{"train_loss": 0.04003724828362465, "train_loss_bc": 0.040023401379585266, "train_loss_llm": 0.001384824514389038, "grad_norm": 0.03106316365301609, "global_step": 4166, "epoch": 115, "lr": 0.001} +{"train_loss": 0.04816829413175583, "train_loss_bc": 0.04814564064145088, "train_loss_llm": 0.002265363931655884, "grad_norm": 0.032705456018447876, "global_step": 4167, "epoch": 115, "lr": 0.001} +{"train_loss": 0.039231810718774796, "train_loss_bc": 0.03920823708176613, "train_loss_llm": 0.0023574531078338623, "grad_norm": 0.041205599904060364, "global_step": 4168, "epoch": 115, "lr": 0.001} +{"train_loss": 0.040076568722724915, "train_loss_bc": 0.040074389427900314, "train_loss_llm": 0.0002180933952331543, "grad_norm": 0.016541844233870506, "global_step": 4169, "epoch": 115, "lr": 0.001} +{"train_loss": 0.042957987636327744, "train_loss_bc": 0.04295659437775612, "train_loss_llm": 0.00013938546180725098, "grad_norm": 0.02607058733701706, "global_step": 4170, "epoch": 115, "lr": 0.001} +{"train_loss": 0.039915334433317184, "train_loss_bc": 0.039915334433317184, "train_loss_llm": 0.0, "grad_norm": 0.02938576228916645, "global_step": 4171, "epoch": 115, "lr": 0.001} +{"train_loss": 0.04951054975390434, "train_loss_bc": 0.04950476065278053, "train_loss_llm": 0.0005789101123809814, "grad_norm": 0.03917246311903, "global_step": 4172, "epoch": 115, "lr": 0.001} +{"train_loss": 0.0418231375515461, "train_loss_bc": 0.0418231375515461, "train_loss_llm": 0.0, "grad_norm": 0.01839897409081459, "global_step": 4173, "epoch": 115, "lr": 0.001} +{"train_loss": 0.038656026124954224, "train_loss_bc": 0.03865601867437363, "train_loss_llm": 5.960464477539062e-07, "grad_norm": 0.03132692724466324, "global_step": 4174, "epoch": 115, "lr": 0.001} +{"train_loss": 0.04277149515433444, "train_loss_bc": 0.03678691387176514, "train_loss_llm": 0.0, "grad_norm": 0.03729957342147827, "global_step": 4175, "epoch": 115, "lr": 0.001, "train/cumulative_reward": 139.0743750465274, "train/mean_score": 20.831960754668835, "train/success_rate": 0.6, "test/cumulative_reward": 74.45397913120728, "test/mean_score": 12.805033489098383, "test/success_rate": 0.32, "val_loss": 0.053090184926986694, "train_action_mse_error": 0.04618952050805092} +{"train_loss": 0.04400559514760971, "train_loss_bc": 0.04400559514760971, "train_loss_llm": 0.0, "grad_norm": 0.0461752824485302, "global_step": 4176, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04952830448746681, "train_loss_bc": 0.049522075802087784, "train_loss_llm": 0.0006230175495147705, "grad_norm": 0.016463467851281166, "global_step": 4177, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04013815149664879, "train_loss_bc": 0.04013815149664879, "train_loss_llm": 0.0, "grad_norm": 0.023688310757279396, "global_step": 4178, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04207095503807068, "train_loss_bc": 0.04205622151494026, "train_loss_llm": 0.0014733076095581055, "grad_norm": 0.029388869181275368, "global_step": 4179, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04140179604291916, "train_loss_bc": 0.04140179604291916, "train_loss_llm": 0.0, "grad_norm": 0.03250465914607048, "global_step": 4180, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04058947041630745, "train_loss_bc": 0.04058947041630745, "train_loss_llm": 0.0, "grad_norm": 0.017502371221780777, "global_step": 4181, "epoch": 116, "lr": 0.001} +{"train_loss": 0.037250131368637085, "train_loss_bc": 0.037250131368637085, "train_loss_llm": 0.0, "grad_norm": 0.023664623498916626, "global_step": 4182, "epoch": 116, "lr": 0.001} +{"train_loss": 0.039192769676446915, "train_loss_bc": 0.039190396666526794, "train_loss_llm": 0.00023734569549560547, "grad_norm": 0.03271229565143585, "global_step": 4183, "epoch": 116, "lr": 0.001} +{"train_loss": 0.047121431678533554, "train_loss_bc": 0.047121431678533554, "train_loss_llm": 0.0, "grad_norm": 0.03861517831683159, "global_step": 4184, "epoch": 116, "lr": 0.001} +{"train_loss": 0.03664564713835716, "train_loss_bc": 0.03664564713835716, "train_loss_llm": 0.0, "grad_norm": 0.01828497275710106, "global_step": 4185, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04459204524755478, "train_loss_bc": 0.04459204524755478, "train_loss_llm": 0.0, "grad_norm": 0.03140219673514366, "global_step": 4186, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04372931271791458, "train_loss_bc": 0.04369974881410599, "train_loss_llm": 0.0029565393924713135, "grad_norm": 0.03383356332778931, "global_step": 4187, "epoch": 116, "lr": 0.001} +{"train_loss": 0.041158758103847504, "train_loss_bc": 0.041148699820041656, "train_loss_llm": 0.0010057389736175537, "grad_norm": 0.04420579969882965, "global_step": 4188, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04174813628196716, "train_loss_bc": 0.04174239933490753, "train_loss_llm": 0.000573575496673584, "grad_norm": 0.01825500652194023, "global_step": 4189, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04390731826424599, "train_loss_bc": 0.04385778307914734, "train_loss_llm": 0.004953622817993164, "grad_norm": 0.026329021900892258, "global_step": 4190, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04594477638602257, "train_loss_bc": 0.045944079756736755, "train_loss_llm": 6.964802742004395e-05, "grad_norm": 0.030392585322260857, "global_step": 4191, "epoch": 116, "lr": 0.001} +{"train_loss": 0.045074671506881714, "train_loss_bc": 0.045074671506881714, "train_loss_llm": 0.0, "grad_norm": 0.0386289618909359, "global_step": 4192, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04021619260311127, "train_loss_bc": 0.0402057059109211, "train_loss_llm": 0.0010486841201782227, "grad_norm": 0.01630696840584278, "global_step": 4193, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04332441836595535, "train_loss_bc": 0.04332441836595535, "train_loss_llm": 0.0, "grad_norm": 0.028252357617020607, "global_step": 4194, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04366467893123627, "train_loss_bc": 0.04366467893123627, "train_loss_llm": 0.0, "grad_norm": 0.038835559040308, "global_step": 4195, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04042171686887741, "train_loss_bc": 0.04042171686887741, "train_loss_llm": 0.0, "grad_norm": 0.04651843011379242, "global_step": 4196, "epoch": 116, "lr": 0.001} +{"train_loss": 0.042103249579668045, "train_loss_bc": 0.042066894471645355, "train_loss_llm": 0.003635495901107788, "grad_norm": 0.01852310076355934, "global_step": 4197, "epoch": 116, "lr": 0.001} +{"train_loss": 0.05040688067674637, "train_loss_bc": 0.05040688067674637, "train_loss_llm": 0.0, "grad_norm": 0.02601209655404091, "global_step": 4198, "epoch": 116, "lr": 0.001} +{"train_loss": 0.05210333317518234, "train_loss_bc": 0.05210333317518234, "train_loss_llm": 0.0, "grad_norm": 0.04222506284713745, "global_step": 4199, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04284418746829033, "train_loss_bc": 0.04284418746829033, "train_loss_llm": 0.0, "grad_norm": 0.05401615798473358, "global_step": 4200, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04292115941643715, "train_loss_bc": 0.04292115941643715, "train_loss_llm": 0.0, "grad_norm": 0.01350471843034029, "global_step": 4201, "epoch": 116, "lr": 0.001} +{"train_loss": 0.0306090135127306, "train_loss_bc": 0.0306090135127306, "train_loss_llm": 0.0, "grad_norm": 0.017892755568027496, "global_step": 4202, "epoch": 116, "lr": 0.001} +{"train_loss": 0.03925119340419769, "train_loss_bc": 0.03925119340419769, "train_loss_llm": 0.0, "grad_norm": 0.026517214253544807, "global_step": 4203, "epoch": 116, "lr": 0.001} +{"train_loss": 0.039970725774765015, "train_loss_bc": 0.039970725774765015, "train_loss_llm": 0.0, "grad_norm": 0.033598851412534714, "global_step": 4204, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04484832286834717, "train_loss_bc": 0.04484832286834717, "train_loss_llm": 0.0, "grad_norm": 0.014590529724955559, "global_step": 4205, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04606122151017189, "train_loss_bc": 0.04606122151017189, "train_loss_llm": 0.0, "grad_norm": 0.021664638072252274, "global_step": 4206, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04238343611359596, "train_loss_bc": 0.04238343611359596, "train_loss_llm": 0.0, "grad_norm": 0.02838868647813797, "global_step": 4207, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04863881319761276, "train_loss_bc": 0.048618417233228683, "train_loss_llm": 0.0020395517349243164, "grad_norm": 0.03598669171333313, "global_step": 4208, "epoch": 116, "lr": 0.001} +{"train_loss": 0.03827314451336861, "train_loss_bc": 0.038269657641649246, "train_loss_llm": 0.00034883618354797363, "grad_norm": 0.015466476790606976, "global_step": 4209, "epoch": 116, "lr": 0.001} +{"train_loss": 0.04565398767590523, "train_loss_bc": 0.045620959252119064, "train_loss_llm": 0.0033028721809387207, "grad_norm": 0.029981479048728943, "global_step": 4210, "epoch": 116, "lr": 0.001} +{"train_loss": 0.042784911735604204, "train_loss_bc": 0.04246187582612038, "train_loss_llm": 0.0, "grad_norm": 0.03965155407786369, "global_step": 4211, "epoch": 116, "lr": 0.001, "val_loss": 0.05294780433177948} +{"train_loss": 0.048306092619895935, "train_loss_bc": 0.048306092619895935, "train_loss_llm": 0.0, "grad_norm": 0.050892043858766556, "global_step": 4212, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04485639929771423, "train_loss_bc": 0.04485639929771423, "train_loss_llm": 0.0, "grad_norm": 0.01511296071112156, "global_step": 4213, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04463162645697594, "train_loss_bc": 0.04463162645697594, "train_loss_llm": 0.0, "grad_norm": 0.020247304812073708, "global_step": 4214, "epoch": 117, "lr": 0.001} +{"train_loss": 0.0449567586183548, "train_loss_bc": 0.04493873566389084, "train_loss_llm": 0.0018023550510406494, "grad_norm": 0.0242227241396904, "global_step": 4215, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04016761854290962, "train_loss_bc": 0.04016761854290962, "train_loss_llm": 0.0, "grad_norm": 0.02757589891552925, "global_step": 4216, "epoch": 117, "lr": 0.001} +{"train_loss": 0.047006044536828995, "train_loss_bc": 0.047006044536828995, "train_loss_llm": 0.0, "grad_norm": 0.018072234466671944, "global_step": 4217, "epoch": 117, "lr": 0.001} +{"train_loss": 0.037736814469099045, "train_loss_bc": 0.037736814469099045, "train_loss_llm": 0.0, "grad_norm": 0.024409333243966103, "global_step": 4218, "epoch": 117, "lr": 0.001} +{"train_loss": 0.038473717868328094, "train_loss_bc": 0.038473717868328094, "train_loss_llm": 0.0, "grad_norm": 0.03798963129520416, "global_step": 4219, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04065987467765808, "train_loss_bc": 0.04065987467765808, "train_loss_llm": 0.0, "grad_norm": 0.0435788594186306, "global_step": 4220, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04404407739639282, "train_loss_bc": 0.04404407739639282, "train_loss_llm": 0.0, "grad_norm": 0.01636006310582161, "global_step": 4221, "epoch": 117, "lr": 0.001} +{"train_loss": 0.03828965872526169, "train_loss_bc": 0.03828457370400429, "train_loss_llm": 0.0005085468292236328, "grad_norm": 0.029853248968720436, "global_step": 4222, "epoch": 117, "lr": 0.001} +{"train_loss": 0.03902424871921539, "train_loss_bc": 0.038998834788799286, "train_loss_llm": 0.002541273832321167, "grad_norm": 0.04073900729417801, "global_step": 4223, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04430081322789192, "train_loss_bc": 0.04430081322789192, "train_loss_llm": 0.0, "grad_norm": 0.04703042283654213, "global_step": 4224, "epoch": 117, "lr": 0.001} +{"train_loss": 0.050668563693761826, "train_loss_bc": 0.050662677735090256, "train_loss_llm": 0.0005886554718017578, "grad_norm": 0.015162059105932713, "global_step": 4225, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04175284504890442, "train_loss_bc": 0.04175284504890442, "train_loss_llm": 0.0, "grad_norm": 0.022505590692162514, "global_step": 4226, "epoch": 117, "lr": 0.001} +{"train_loss": 0.039265282452106476, "train_loss_bc": 0.03925083577632904, "train_loss_llm": 0.0014445185661315918, "grad_norm": 0.035100772976875305, "global_step": 4227, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04244309663772583, "train_loss_bc": 0.04242774844169617, "train_loss_llm": 0.001534879207611084, "grad_norm": 0.039946187287569046, "global_step": 4228, "epoch": 117, "lr": 0.001} +{"train_loss": 0.043511781841516495, "train_loss_bc": 0.043511781841516495, "train_loss_llm": 0.0, "grad_norm": 0.01645454578101635, "global_step": 4229, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04047107324004173, "train_loss_bc": 0.04047107324004173, "train_loss_llm": 0.0, "grad_norm": 0.02248447760939598, "global_step": 4230, "epoch": 117, "lr": 0.001} +{"train_loss": 0.03956571966409683, "train_loss_bc": 0.03953412547707558, "train_loss_llm": 0.003159552812576294, "grad_norm": 0.03084101341664791, "global_step": 4231, "epoch": 117, "lr": 0.001} +{"train_loss": 0.040519557893276215, "train_loss_bc": 0.04051610827445984, "train_loss_llm": 0.00034493207931518555, "grad_norm": 0.04029131308197975, "global_step": 4232, "epoch": 117, "lr": 0.001} +{"train_loss": 0.038857001811265945, "train_loss_bc": 0.03885434940457344, "train_loss_llm": 0.0002650618553161621, "grad_norm": 0.016529487445950508, "global_step": 4233, "epoch": 117, "lr": 0.001} +{"train_loss": 0.044480517506599426, "train_loss_bc": 0.044480517506599426, "train_loss_llm": 0.0, "grad_norm": 0.02497413568198681, "global_step": 4234, "epoch": 117, "lr": 0.001} +{"train_loss": 0.050672609359025955, "train_loss_bc": 0.050672609359025955, "train_loss_llm": 0.0, "grad_norm": 0.03450774401426315, "global_step": 4235, "epoch": 117, "lr": 0.001} +{"train_loss": 0.047163862735033035, "train_loss_bc": 0.047163862735033035, "train_loss_llm": 0.0, "grad_norm": 0.0422234870493412, "global_step": 4236, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04294035583734512, "train_loss_bc": 0.04294035583734512, "train_loss_llm": 0.0, "grad_norm": 0.012517807073891163, "global_step": 4237, "epoch": 117, "lr": 0.001} +{"train_loss": 0.045790284872055054, "train_loss_bc": 0.045790284872055054, "train_loss_llm": 0.0, "grad_norm": 0.025280175730586052, "global_step": 4238, "epoch": 117, "lr": 0.001} +{"train_loss": 0.039859555661678314, "train_loss_bc": 0.0398445799946785, "train_loss_llm": 0.0014973878860473633, "grad_norm": 0.03493761271238327, "global_step": 4239, "epoch": 117, "lr": 0.001} +{"train_loss": 0.03686261922121048, "train_loss_bc": 0.0368601493537426, "train_loss_llm": 0.00024700164794921875, "grad_norm": 0.04292884096503258, "global_step": 4240, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04358203709125519, "train_loss_bc": 0.04358203709125519, "train_loss_llm": 0.0, "grad_norm": 0.021212350577116013, "global_step": 4241, "epoch": 117, "lr": 0.001} +{"train_loss": 0.048627372831106186, "train_loss_bc": 0.048627372831106186, "train_loss_llm": 0.0, "grad_norm": 0.031185904517769814, "global_step": 4242, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04197799041867256, "train_loss_bc": 0.04196326434612274, "train_loss_llm": 0.0014724433422088623, "grad_norm": 0.03753320500254631, "global_step": 4243, "epoch": 117, "lr": 0.001} +{"train_loss": 0.0405452586710453, "train_loss_bc": 0.0405452586710453, "train_loss_llm": 0.0, "grad_norm": 0.04350452125072479, "global_step": 4244, "epoch": 117, "lr": 0.001} +{"train_loss": 0.03788299486041069, "train_loss_bc": 0.03787897527217865, "train_loss_llm": 0.0004019439220428467, "grad_norm": 0.014403403736650944, "global_step": 4245, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04621311277151108, "train_loss_bc": 0.04621311277151108, "train_loss_llm": 0.0, "grad_norm": 0.02211752161383629, "global_step": 4246, "epoch": 117, "lr": 0.001} +{"train_loss": 0.04279754621287187, "train_loss_bc": 0.044570084661245346, "train_loss_llm": 0.0034340322017669678, "grad_norm": 0.03834362328052521, "global_step": 4247, "epoch": 117, "lr": 0.001, "val_loss": 0.053165219724178314} +{"train_loss": 0.04192205145955086, "train_loss_bc": 0.041907645761966705, "train_loss_llm": 0.0014406442642211914, "grad_norm": 0.04398214444518089, "global_step": 4248, "epoch": 118, "lr": 0.001} +{"train_loss": 0.050320446491241455, "train_loss_bc": 0.050320446491241455, "train_loss_llm": 0.0, "grad_norm": 0.017633341252803802, "global_step": 4249, "epoch": 118, "lr": 0.001} +{"train_loss": 0.042025353759527206, "train_loss_bc": 0.042025353759527206, "train_loss_llm": 0.0, "grad_norm": 0.027294401079416275, "global_step": 4250, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04260067269206047, "train_loss_bc": 0.04260067269206047, "train_loss_llm": 0.0, "grad_norm": 0.0347198061645031, "global_step": 4251, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04301445186138153, "train_loss_bc": 0.043000951409339905, "train_loss_llm": 0.0013501644134521484, "grad_norm": 0.04041200503706932, "global_step": 4252, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04623904824256897, "train_loss_bc": 0.04623904824256897, "train_loss_llm": 0.0, "grad_norm": 0.01677389070391655, "global_step": 4253, "epoch": 118, "lr": 0.001} +{"train_loss": 0.040898844599723816, "train_loss_bc": 0.04088148474693298, "train_loss_llm": 0.001736074686050415, "grad_norm": 0.026447415351867676, "global_step": 4254, "epoch": 118, "lr": 0.001} +{"train_loss": 0.042013972997665405, "train_loss_bc": 0.042013972997665405, "train_loss_llm": 0.0, "grad_norm": 0.03449448570609093, "global_step": 4255, "epoch": 118, "lr": 0.001} +{"train_loss": 0.039938658475875854, "train_loss_bc": 0.039938658475875854, "train_loss_llm": 0.0, "grad_norm": 0.04148133099079132, "global_step": 4256, "epoch": 118, "lr": 0.001} +{"train_loss": 0.03210945799946785, "train_loss_bc": 0.03210945799946785, "train_loss_llm": 0.0, "grad_norm": 0.016169143840670586, "global_step": 4257, "epoch": 118, "lr": 0.001} +{"train_loss": 0.03698626533150673, "train_loss_bc": 0.03698626533150673, "train_loss_llm": 0.0, "grad_norm": 0.023779915645718575, "global_step": 4258, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04909380152821541, "train_loss_bc": 0.04909380152821541, "train_loss_llm": 0.0, "grad_norm": 0.030955901369452477, "global_step": 4259, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04167548194527626, "train_loss_bc": 0.04165494441986084, "train_loss_llm": 0.002053886651992798, "grad_norm": 0.0351945124566555, "global_step": 4260, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04790107160806656, "train_loss_bc": 0.04790107160806656, "train_loss_llm": 0.0, "grad_norm": 0.015018499456346035, "global_step": 4261, "epoch": 118, "lr": 0.001} +{"train_loss": 0.046132612973451614, "train_loss_bc": 0.046132612973451614, "train_loss_llm": 0.0, "grad_norm": 0.019992703571915627, "global_step": 4262, "epoch": 118, "lr": 0.001} +{"train_loss": 0.03499578312039375, "train_loss_bc": 0.03498588874936104, "train_loss_llm": 0.000989466905593872, "grad_norm": 0.025088440626859665, "global_step": 4263, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04172380268573761, "train_loss_bc": 0.04171612113714218, "train_loss_llm": 0.0007680058479309082, "grad_norm": 0.03212058171629906, "global_step": 4264, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04727528244256973, "train_loss_bc": 0.0472535714507103, "train_loss_llm": 0.002171069383621216, "grad_norm": 0.017035190016031265, "global_step": 4265, "epoch": 118, "lr": 0.001} +{"train_loss": 0.044087961316108704, "train_loss_bc": 0.044087961316108704, "train_loss_llm": 0.0, "grad_norm": 0.026065973564982414, "global_step": 4266, "epoch": 118, "lr": 0.001} +{"train_loss": 0.0385901965200901, "train_loss_bc": 0.0385901965200901, "train_loss_llm": 0.0, "grad_norm": 0.024950746446847916, "global_step": 4267, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04396316409111023, "train_loss_bc": 0.043959055095911026, "train_loss_llm": 0.00041103363037109375, "grad_norm": 0.03264157474040985, "global_step": 4268, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04527769237756729, "train_loss_bc": 0.04527372866868973, "train_loss_llm": 0.0003962218761444092, "grad_norm": 0.02055397443473339, "global_step": 4269, "epoch": 118, "lr": 0.001} +{"train_loss": 0.03941113501787186, "train_loss_bc": 0.03939558565616608, "train_loss_llm": 0.0015550851821899414, "grad_norm": 0.028790973126888275, "global_step": 4270, "epoch": 118, "lr": 0.001} +{"train_loss": 0.047517430037260056, "train_loss_bc": 0.047517430037260056, "train_loss_llm": 0.0, "grad_norm": 0.03685731068253517, "global_step": 4271, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04722897708415985, "train_loss_bc": 0.047228798270225525, "train_loss_llm": 1.7940998077392578e-05, "grad_norm": 0.048500191420316696, "global_step": 4272, "epoch": 118, "lr": 0.001} +{"train_loss": 0.0381467267870903, "train_loss_bc": 0.0381467267870903, "train_loss_llm": 0.0, "grad_norm": 0.016898343339562416, "global_step": 4273, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04069552943110466, "train_loss_bc": 0.04069552943110466, "train_loss_llm": 0.0, "grad_norm": 0.022922489792108536, "global_step": 4274, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04555172100663185, "train_loss_bc": 0.04554678872227669, "train_loss_llm": 0.0004933178424835205, "grad_norm": 0.027938293293118477, "global_step": 4275, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04004158824682236, "train_loss_bc": 0.04004158824682236, "train_loss_llm": 0.0, "grad_norm": 0.03521398454904556, "global_step": 4276, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04469895735383034, "train_loss_bc": 0.04469895735383034, "train_loss_llm": 0.0, "grad_norm": 0.016205623745918274, "global_step": 4277, "epoch": 118, "lr": 0.001} +{"train_loss": 0.03760901466012001, "train_loss_bc": 0.03759832680225372, "train_loss_llm": 0.001068800687789917, "grad_norm": 0.02002228982746601, "global_step": 4278, "epoch": 118, "lr": 0.001} +{"train_loss": 0.038615643978118896, "train_loss_bc": 0.03860859572887421, "train_loss_llm": 0.0007046759128570557, "grad_norm": 0.024668265134096146, "global_step": 4279, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04547332227230072, "train_loss_bc": 0.04547332227230072, "train_loss_llm": 0.0, "grad_norm": 0.0324355885386467, "global_step": 4280, "epoch": 118, "lr": 0.001} +{"train_loss": 0.047451164573431015, "train_loss_bc": 0.047451164573431015, "train_loss_llm": 0.0, "grad_norm": 0.016441063955426216, "global_step": 4281, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04443097859621048, "train_loss_bc": 0.04443097859621048, "train_loss_llm": 0.0, "grad_norm": 0.024395840242505074, "global_step": 4282, "epoch": 118, "lr": 0.001} +{"train_loss": 0.04262196282959647, "train_loss_bc": 0.03871898353099823, "train_loss_llm": 0.0013415813446044922, "grad_norm": 0.033305779099464417, "global_step": 4283, "epoch": 118, "lr": 0.001, "val_loss": 0.053139954805374146} +{"train_loss": 0.05053334683179855, "train_loss_bc": 0.05053334683179855, "train_loss_llm": 0.0, "grad_norm": 0.03835475817322731, "global_step": 4284, "epoch": 119, "lr": 0.001} +{"train_loss": 0.03933378681540489, "train_loss_bc": 0.03931019455194473, "train_loss_llm": 0.0023592114448547363, "grad_norm": 0.01987660862505436, "global_step": 4285, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04477570578455925, "train_loss_bc": 0.04477570578455925, "train_loss_llm": 0.0, "grad_norm": 0.022052418440580368, "global_step": 4286, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04509642720222473, "train_loss_bc": 0.04509642720222473, "train_loss_llm": 0.0, "grad_norm": 0.022914256900548935, "global_step": 4287, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04706215858459473, "train_loss_bc": 0.04704827815294266, "train_loss_llm": 0.0013880729675292969, "grad_norm": 0.02985047921538353, "global_step": 4288, "epoch": 119, "lr": 0.001} +{"train_loss": 0.037427451461553574, "train_loss_bc": 0.037427451461553574, "train_loss_llm": 0.0, "grad_norm": 0.01747238077223301, "global_step": 4289, "epoch": 119, "lr": 0.001} +{"train_loss": 0.048016078770160675, "train_loss_bc": 0.04800501838326454, "train_loss_llm": 0.0011059939861297607, "grad_norm": 0.023453081026673317, "global_step": 4290, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04937239736318588, "train_loss_bc": 0.04937239736318588, "train_loss_llm": 0.0, "grad_norm": 0.029929356649518013, "global_step": 4291, "epoch": 119, "lr": 0.001} +{"train_loss": 0.03800639510154724, "train_loss_bc": 0.03800639510154724, "train_loss_llm": 0.0, "grad_norm": 0.0333123505115509, "global_step": 4292, "epoch": 119, "lr": 0.001} +{"train_loss": 0.038131337612867355, "train_loss_bc": 0.038131337612867355, "train_loss_llm": 0.0, "grad_norm": 0.012341105379164219, "global_step": 4293, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04303160309791565, "train_loss_bc": 0.04303160309791565, "train_loss_llm": 0.0, "grad_norm": 0.02056398242712021, "global_step": 4294, "epoch": 119, "lr": 0.001} +{"train_loss": 0.030810222029685974, "train_loss_bc": 0.03080764040350914, "train_loss_llm": 0.0002580881118774414, "grad_norm": 0.024056196212768555, "global_step": 4295, "epoch": 119, "lr": 0.001} +{"train_loss": 0.05022880434989929, "train_loss_bc": 0.05022880434989929, "train_loss_llm": 0.0, "grad_norm": 0.03044300340116024, "global_step": 4296, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04663976654410362, "train_loss_bc": 0.04663976654410362, "train_loss_llm": 0.0, "grad_norm": 0.01653948798775673, "global_step": 4297, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04306863993406296, "train_loss_bc": 0.04306863993406296, "train_loss_llm": 0.0, "grad_norm": 0.022623075172305107, "global_step": 4298, "epoch": 119, "lr": 0.001} +{"train_loss": 0.03872600570321083, "train_loss_bc": 0.03872350975871086, "train_loss_llm": 0.00024959444999694824, "grad_norm": 0.03233564645051956, "global_step": 4299, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04537570849061012, "train_loss_bc": 0.04537570849061012, "train_loss_llm": 0.0, "grad_norm": 0.035382840782403946, "global_step": 4300, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04415564239025116, "train_loss_bc": 0.0441543348133564, "train_loss_llm": 0.0001308917999267578, "grad_norm": 0.01472486648708582, "global_step": 4301, "epoch": 119, "lr": 0.001} +{"train_loss": 0.043432679027318954, "train_loss_bc": 0.04342074692249298, "train_loss_llm": 0.0011932849884033203, "grad_norm": 0.025704294443130493, "global_step": 4302, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04549819231033325, "train_loss_bc": 0.04549819231033325, "train_loss_llm": 0.0, "grad_norm": 0.032258473336696625, "global_step": 4303, "epoch": 119, "lr": 0.001} +{"train_loss": 0.03557170182466507, "train_loss_bc": 0.0355641171336174, "train_loss_llm": 0.0007582902908325195, "grad_norm": 0.03808397799730301, "global_step": 4304, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04437332972884178, "train_loss_bc": 0.04436738044023514, "train_loss_llm": 0.0005948841571807861, "grad_norm": 0.016610026359558105, "global_step": 4305, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04050629213452339, "train_loss_bc": 0.04050629213452339, "train_loss_llm": 0.0, "grad_norm": 0.02624325454235077, "global_step": 4306, "epoch": 119, "lr": 0.001} +{"train_loss": 0.040044769644737244, "train_loss_bc": 0.040044769644737244, "train_loss_llm": 0.0, "grad_norm": 0.03276931121945381, "global_step": 4307, "epoch": 119, "lr": 0.001} +{"train_loss": 0.046938709914684296, "train_loss_bc": 0.04692918062210083, "train_loss_llm": 0.0009530782699584961, "grad_norm": 0.03908475115895271, "global_step": 4308, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04308358207345009, "train_loss_bc": 0.04308358207345009, "train_loss_llm": 0.0, "grad_norm": 0.015216222032904625, "global_step": 4309, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04849494993686676, "train_loss_bc": 0.048455383628606796, "train_loss_llm": 0.0039566755294799805, "grad_norm": 0.02212783880531788, "global_step": 4310, "epoch": 119, "lr": 0.001} +{"train_loss": 0.033410392701625824, "train_loss_bc": 0.03340534493327141, "train_loss_llm": 0.0005047619342803955, "grad_norm": 0.02506110444664955, "global_step": 4311, "epoch": 119, "lr": 0.001} +{"train_loss": 0.049500543624162674, "train_loss_bc": 0.049500543624162674, "train_loss_llm": 0.0, "grad_norm": 0.03241222724318504, "global_step": 4312, "epoch": 119, "lr": 0.001} +{"train_loss": 0.03984591364860535, "train_loss_bc": 0.039840251207351685, "train_loss_llm": 0.000566333532333374, "grad_norm": 0.015944533050060272, "global_step": 4313, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04545222222805023, "train_loss_bc": 0.045436423271894455, "train_loss_llm": 0.0015800297260284424, "grad_norm": 0.022690756246447563, "global_step": 4314, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04021691530942917, "train_loss_bc": 0.04021691530942917, "train_loss_llm": 0.0, "grad_norm": 0.03160018101334572, "global_step": 4315, "epoch": 119, "lr": 0.001} +{"train_loss": 0.043091241270303726, "train_loss_bc": 0.04308145493268967, "train_loss_llm": 0.0009786486625671387, "grad_norm": 0.04058164730668068, "global_step": 4316, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04023903235793114, "train_loss_bc": 0.04023692384362221, "train_loss_llm": 0.00021085143089294434, "grad_norm": 0.01285514421761036, "global_step": 4317, "epoch": 119, "lr": 0.001} +{"train_loss": 0.04023886099457741, "train_loss_bc": 0.04023886099457741, "train_loss_llm": 0.0, "grad_norm": 0.022598063573241234, "global_step": 4318, "epoch": 119, "lr": 0.001} +{"train_loss": 0.042474878744946584, "train_loss_bc": 0.029364828020334244, "train_loss_llm": 0.0, "grad_norm": 0.029535330832004547, "global_step": 4319, "epoch": 119, "lr": 0.001, "val_loss": 0.05306572467088699} +{"train_loss": 0.045926284044981, "train_loss_bc": 0.045926284044981, "train_loss_llm": 0.0, "grad_norm": 0.03693505749106407, "global_step": 4320, "epoch": 120, "lr": 0.001} +{"train_loss": 0.051364388316869736, "train_loss_bc": 0.05134260654449463, "train_loss_llm": 0.0021781325340270996, "grad_norm": 0.020428862422704697, "global_step": 4321, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04613792151212692, "train_loss_bc": 0.046129900962114334, "train_loss_llm": 0.0008020699024200439, "grad_norm": 0.028948472812771797, "global_step": 4322, "epoch": 120, "lr": 0.001} +{"train_loss": 0.049327027052640915, "train_loss_bc": 0.049327027052640915, "train_loss_llm": 0.0, "grad_norm": 0.03641236573457718, "global_step": 4323, "epoch": 120, "lr": 0.001} +{"train_loss": 0.043530840426683426, "train_loss_bc": 0.043530840426683426, "train_loss_llm": 0.0, "grad_norm": 0.044013869017362595, "global_step": 4324, "epoch": 120, "lr": 0.001} +{"train_loss": 0.039958350360393524, "train_loss_bc": 0.039958350360393524, "train_loss_llm": 0.0, "grad_norm": 0.014921213500201702, "global_step": 4325, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04791981354355812, "train_loss_bc": 0.04791624844074249, "train_loss_llm": 0.00035646557807922363, "grad_norm": 0.025927972048521042, "global_step": 4326, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04369078204035759, "train_loss_bc": 0.043686944991350174, "train_loss_llm": 0.000383526086807251, "grad_norm": 0.029073014855384827, "global_step": 4327, "epoch": 120, "lr": 0.001} +{"train_loss": 0.044488534331321716, "train_loss_bc": 0.044488534331321716, "train_loss_llm": 0.0, "grad_norm": 0.032727353274822235, "global_step": 4328, "epoch": 120, "lr": 0.001} +{"train_loss": 0.0478179045021534, "train_loss_bc": 0.04779382050037384, "train_loss_llm": 0.002408355474472046, "grad_norm": 0.016751056537032127, "global_step": 4329, "epoch": 120, "lr": 0.001} +{"train_loss": 0.0402190126478672, "train_loss_bc": 0.0402190126478672, "train_loss_llm": 0.0, "grad_norm": 0.025347871705889702, "global_step": 4330, "epoch": 120, "lr": 0.001} +{"train_loss": 0.034062907099723816, "train_loss_bc": 0.034062907099723816, "train_loss_llm": 0.0, "grad_norm": 0.033450618386268616, "global_step": 4331, "epoch": 120, "lr": 0.001} +{"train_loss": 0.03707917034626007, "train_loss_bc": 0.03707917034626007, "train_loss_llm": 0.0, "grad_norm": 0.03798463195562363, "global_step": 4332, "epoch": 120, "lr": 0.001} +{"train_loss": 0.035266175866127014, "train_loss_bc": 0.035261258482933044, "train_loss_llm": 0.0004917681217193604, "grad_norm": 0.012373335659503937, "global_step": 4333, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04109610244631767, "train_loss_bc": 0.04109610244631767, "train_loss_llm": 0.0, "grad_norm": 0.022542236372828484, "global_step": 4334, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04431338980793953, "train_loss_bc": 0.04431338980793953, "train_loss_llm": 0.0, "grad_norm": 0.029003234580159187, "global_step": 4335, "epoch": 120, "lr": 0.001} +{"train_loss": 0.03504886478185654, "train_loss_bc": 0.03504886478185654, "train_loss_llm": 0.0, "grad_norm": 0.03248398005962372, "global_step": 4336, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04933221638202667, "train_loss_bc": 0.04933221638202667, "train_loss_llm": 0.0, "grad_norm": 0.017563102766871452, "global_step": 4337, "epoch": 120, "lr": 0.001} +{"train_loss": 0.03777486830949783, "train_loss_bc": 0.03777486830949783, "train_loss_llm": 0.0, "grad_norm": 0.023711854591965675, "global_step": 4338, "epoch": 120, "lr": 0.001} +{"train_loss": 0.041765857487916946, "train_loss_bc": 0.041765857487916946, "train_loss_llm": 0.0, "grad_norm": 0.029124442487955093, "global_step": 4339, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04062918201088905, "train_loss_bc": 0.040605317801237106, "train_loss_llm": 0.0023862719535827637, "grad_norm": 0.03693445771932602, "global_step": 4340, "epoch": 120, "lr": 0.001} +{"train_loss": 0.050560805946588516, "train_loss_bc": 0.05052107945084572, "train_loss_llm": 0.003972470760345459, "grad_norm": 0.017966890707612038, "global_step": 4341, "epoch": 120, "lr": 0.001} +{"train_loss": 0.040430326014757156, "train_loss_bc": 0.040430326014757156, "train_loss_llm": 0.0, "grad_norm": 0.02914779633283615, "global_step": 4342, "epoch": 120, "lr": 0.001} +{"train_loss": 0.043696820735931396, "train_loss_bc": 0.043681107461452484, "train_loss_llm": 0.00157126784324646, "grad_norm": 0.03458361327648163, "global_step": 4343, "epoch": 120, "lr": 0.001} +{"train_loss": 0.0331304557621479, "train_loss_bc": 0.0331304557621479, "train_loss_llm": 0.0, "grad_norm": 0.041928187012672424, "global_step": 4344, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04592622444033623, "train_loss_bc": 0.04591050371527672, "train_loss_llm": 0.0015720725059509277, "grad_norm": 0.018930954858660698, "global_step": 4345, "epoch": 120, "lr": 0.001} +{"train_loss": 0.03870026767253876, "train_loss_bc": 0.03870026767253876, "train_loss_llm": 0.0, "grad_norm": 0.030743762850761414, "global_step": 4346, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04056240990757942, "train_loss_bc": 0.04053683578968048, "train_loss_llm": 0.0025572776794433594, "grad_norm": 0.04281030222773552, "global_step": 4347, "epoch": 120, "lr": 0.001} +{"train_loss": 0.051191408187150955, "train_loss_bc": 0.05117429420351982, "train_loss_llm": 0.0017113089561462402, "grad_norm": 0.04792085289955139, "global_step": 4348, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04429306834936142, "train_loss_bc": 0.04428522661328316, "train_loss_llm": 0.0007843077182769775, "grad_norm": 0.015854058787226677, "global_step": 4349, "epoch": 120, "lr": 0.001} +{"train_loss": 0.040326077491045, "train_loss_bc": 0.04032108187675476, "train_loss_llm": 0.000499725341796875, "grad_norm": 0.027987733483314514, "global_step": 4350, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04100678488612175, "train_loss_bc": 0.04100678488612175, "train_loss_llm": 0.0, "grad_norm": 0.041192349046468735, "global_step": 4351, "epoch": 120, "lr": 0.001} +{"train_loss": 0.03999575600028038, "train_loss_bc": 0.03999575600028038, "train_loss_llm": 0.0, "grad_norm": 0.04948185384273529, "global_step": 4352, "epoch": 120, "lr": 0.001} +{"train_loss": 0.041205573827028275, "train_loss_bc": 0.041205573827028275, "train_loss_llm": 0.0, "grad_norm": 0.016009947285056114, "global_step": 4353, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04263186827301979, "train_loss_bc": 0.04263186827301979, "train_loss_llm": 0.0, "grad_norm": 0.022775301709771156, "global_step": 4354, "epoch": 120, "lr": 0.001} +{"train_loss": 0.04268018094201883, "train_loss_bc": 0.04607907310128212, "train_loss_llm": 0.0, "grad_norm": 0.03342786803841591, "global_step": 4355, "epoch": 120, "lr": 0.001, "train/cumulative_reward": 90.82590855699102, "train/mean_score": 20.841713462257495, "train/success_rate": 0.5, "test/cumulative_reward": 48.94262070847685, "test/mean_score": 8.942265598496924, "test/success_rate": 0.2, "val_loss": 0.053251877427101135, "train_action_mse_error": 0.04673623666167259} +{"train_loss": 0.03355066478252411, "train_loss_bc": 0.03354354947805405, "train_loss_llm": 0.0007114112377166748, "grad_norm": 0.04018879309296608, "global_step": 4356, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04728184640407562, "train_loss_bc": 0.04728184640407562, "train_loss_llm": 0.0, "grad_norm": 0.014867505058646202, "global_step": 4357, "epoch": 121, "lr": 0.001} +{"train_loss": 0.045366499572992325, "train_loss_bc": 0.04534696415066719, "train_loss_llm": 0.0019535720348358154, "grad_norm": 0.028690751641988754, "global_step": 4358, "epoch": 121, "lr": 0.001} +{"train_loss": 0.044891249388456345, "train_loss_bc": 0.044891249388456345, "train_loss_llm": 0.0, "grad_norm": 0.04175904765725136, "global_step": 4359, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04458927363157272, "train_loss_bc": 0.044570233672857285, "train_loss_llm": 0.001903921365737915, "grad_norm": 0.052171822637319565, "global_step": 4360, "epoch": 121, "lr": 0.001} +{"train_loss": 0.03793812915682793, "train_loss_bc": 0.03793812915682793, "train_loss_llm": 0.0, "grad_norm": 0.017275257036089897, "global_step": 4361, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04009469226002693, "train_loss_bc": 0.04008283093571663, "train_loss_llm": 0.0011862516403198242, "grad_norm": 0.030952660366892815, "global_step": 4362, "epoch": 121, "lr": 0.001} +{"train_loss": 0.03359362483024597, "train_loss_bc": 0.03359362483024597, "train_loss_llm": 0.0, "grad_norm": 0.04242919757962227, "global_step": 4363, "epoch": 121, "lr": 0.001} +{"train_loss": 0.03299226239323616, "train_loss_bc": 0.03297388553619385, "train_loss_llm": 0.0018375217914581299, "grad_norm": 0.05633723363280296, "global_step": 4364, "epoch": 121, "lr": 0.001} +{"train_loss": 0.047196097671985626, "train_loss_bc": 0.04716185852885246, "train_loss_llm": 0.00342404842376709, "grad_norm": 0.0214141346514225, "global_step": 4365, "epoch": 121, "lr": 0.001} +{"train_loss": 0.044807422906160355, "train_loss_bc": 0.044807422906160355, "train_loss_llm": 0.0, "grad_norm": 0.027719074860215187, "global_step": 4366, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04200687259435654, "train_loss_bc": 0.04198490455746651, "train_loss_llm": 0.0021968185901641846, "grad_norm": 0.03467293828725815, "global_step": 4367, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04410301148891449, "train_loss_bc": 0.04409318417310715, "train_loss_llm": 0.000982731580734253, "grad_norm": 0.04183071851730347, "global_step": 4368, "epoch": 121, "lr": 0.001} +{"train_loss": 0.041894346475601196, "train_loss_bc": 0.04187890887260437, "train_loss_llm": 0.001543670892715454, "grad_norm": 0.01967259868979454, "global_step": 4369, "epoch": 121, "lr": 0.001} +{"train_loss": 0.046873316168785095, "train_loss_bc": 0.046873316168785095, "train_loss_llm": 0.0, "grad_norm": 0.023366034030914307, "global_step": 4370, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04078412055969238, "train_loss_bc": 0.0407659150660038, "train_loss_llm": 0.0018205046653747559, "grad_norm": 0.036942604929208755, "global_step": 4371, "epoch": 121, "lr": 0.001} +{"train_loss": 0.03720150887966156, "train_loss_bc": 0.03720150887966156, "train_loss_llm": 0.0, "grad_norm": 0.046905335038900375, "global_step": 4372, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04535980150103569, "train_loss_bc": 0.04535980150103569, "train_loss_llm": 0.0, "grad_norm": 0.02079235576093197, "global_step": 4373, "epoch": 121, "lr": 0.001} +{"train_loss": 0.042265672236680984, "train_loss_bc": 0.042265672236680984, "train_loss_llm": 0.0, "grad_norm": 0.027934813871979713, "global_step": 4374, "epoch": 121, "lr": 0.001} +{"train_loss": 0.038800839334726334, "train_loss_bc": 0.0387984961271286, "train_loss_llm": 0.00023439526557922363, "grad_norm": 0.03398409113287926, "global_step": 4375, "epoch": 121, "lr": 0.001} +{"train_loss": 0.048015888780355453, "train_loss_bc": 0.047989342361688614, "train_loss_llm": 0.002654731273651123, "grad_norm": 0.04191165789961815, "global_step": 4376, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04445074498653412, "train_loss_bc": 0.04445074498653412, "train_loss_llm": 0.0, "grad_norm": 0.023420704528689384, "global_step": 4377, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04248303920030594, "train_loss_bc": 0.04248303920030594, "train_loss_llm": 0.0, "grad_norm": 0.03126769885420799, "global_step": 4378, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04667697101831436, "train_loss_bc": 0.04666262865066528, "train_loss_llm": 0.001434415578842163, "grad_norm": 0.03583179786801338, "global_step": 4379, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04342438280582428, "train_loss_bc": 0.04342438280582428, "train_loss_llm": 0.0, "grad_norm": 0.04904086887836456, "global_step": 4380, "epoch": 121, "lr": 0.001} +{"train_loss": 0.043380483984947205, "train_loss_bc": 0.043380483984947205, "train_loss_llm": 0.0, "grad_norm": 0.014371532015502453, "global_step": 4381, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04024377837777138, "train_loss_bc": 0.04024377837777138, "train_loss_llm": 0.0, "grad_norm": 0.023937208577990532, "global_step": 4382, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04396767169237137, "train_loss_bc": 0.04396283999085426, "train_loss_llm": 0.0004831254482269287, "grad_norm": 0.033097416162490845, "global_step": 4383, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04718456789851189, "train_loss_bc": 0.04717687889933586, "train_loss_llm": 0.0007688701152801514, "grad_norm": 0.04104107990860939, "global_step": 4384, "epoch": 121, "lr": 0.001} +{"train_loss": 0.041550349444150925, "train_loss_bc": 0.041550349444150925, "train_loss_llm": 0.0, "grad_norm": 0.016443056985735893, "global_step": 4385, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04811186343431473, "train_loss_bc": 0.04811186343431473, "train_loss_llm": 0.0, "grad_norm": 0.022060727700591087, "global_step": 4386, "epoch": 121, "lr": 0.001} +{"train_loss": 0.048873450607061386, "train_loss_bc": 0.048873450607061386, "train_loss_llm": 0.0, "grad_norm": 0.03105357475578785, "global_step": 4387, "epoch": 121, "lr": 0.001} +{"train_loss": 0.038525208830833435, "train_loss_bc": 0.038525208830833435, "train_loss_llm": 0.0, "grad_norm": 0.034052759408950806, "global_step": 4388, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04119878262281418, "train_loss_bc": 0.04119878262281418, "train_loss_llm": 0.0, "grad_norm": 0.016611842438578606, "global_step": 4389, "epoch": 121, "lr": 0.001} +{"train_loss": 0.04453826695680618, "train_loss_bc": 0.04453504458069801, "train_loss_llm": 0.0003223121166229248, "grad_norm": 0.024241607636213303, "global_step": 4390, "epoch": 121, "lr": 0.001} +{"train_loss": 0.042722467022637524, "train_loss_bc": 0.043792109936475754, "train_loss_llm": 0.0, "grad_norm": 0.04441177845001221, "global_step": 4391, "epoch": 121, "lr": 0.001, "val_loss": 0.052669458091259} +{"train_loss": 0.03417322412133217, "train_loss_bc": 0.034168072044849396, "train_loss_llm": 0.0005150437355041504, "grad_norm": 0.0504477322101593, "global_step": 4392, "epoch": 122, "lr": 0.001} +{"train_loss": 0.047956157475709915, "train_loss_bc": 0.047955840826034546, "train_loss_llm": 3.167986869812012e-05, "grad_norm": 0.014427568763494492, "global_step": 4393, "epoch": 122, "lr": 0.001} +{"train_loss": 0.046833328902721405, "train_loss_bc": 0.046833187341690063, "train_loss_llm": 1.4275312423706055e-05, "grad_norm": 0.021582357585430145, "global_step": 4394, "epoch": 122, "lr": 0.001} +{"train_loss": 0.041945680975914, "train_loss_bc": 0.041945680975914, "train_loss_llm": 0.0, "grad_norm": 0.023531513288617134, "global_step": 4395, "epoch": 122, "lr": 0.001} +{"train_loss": 0.043445151299238205, "train_loss_bc": 0.043445151299238205, "train_loss_llm": 0.0, "grad_norm": 0.02648129127919674, "global_step": 4396, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04371795058250427, "train_loss_bc": 0.043683603405952454, "train_loss_llm": 0.0034348368644714355, "grad_norm": 0.018099961802363396, "global_step": 4397, "epoch": 122, "lr": 0.001} +{"train_loss": 0.042916253209114075, "train_loss_bc": 0.042916253209114075, "train_loss_llm": 0.0, "grad_norm": 0.022544514387845993, "global_step": 4398, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04099980369210243, "train_loss_bc": 0.04099980369210243, "train_loss_llm": 0.0, "grad_norm": 0.02717871591448784, "global_step": 4399, "epoch": 122, "lr": 0.001} +{"train_loss": 0.03775401413440704, "train_loss_bc": 0.03775401413440704, "train_loss_llm": 0.0, "grad_norm": 0.034378182142972946, "global_step": 4400, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04448975622653961, "train_loss_bc": 0.04448975622653961, "train_loss_llm": 0.0, "grad_norm": 0.02001611888408661, "global_step": 4401, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04144851863384247, "train_loss_bc": 0.04144851863384247, "train_loss_llm": 0.0, "grad_norm": 0.026995887979865074, "global_step": 4402, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04628315195441246, "train_loss_bc": 0.04628315195441246, "train_loss_llm": 0.0, "grad_norm": 0.038975030183792114, "global_step": 4403, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04443857818841934, "train_loss_bc": 0.04443857818841934, "train_loss_llm": 0.0, "grad_norm": 0.04969995468854904, "global_step": 4404, "epoch": 122, "lr": 0.001} +{"train_loss": 0.03957487642765045, "train_loss_bc": 0.03957487642765045, "train_loss_llm": 0.0, "grad_norm": 0.0156212467700243, "global_step": 4405, "epoch": 122, "lr": 0.001} +{"train_loss": 0.042259737849235535, "train_loss_bc": 0.042259737849235535, "train_loss_llm": 0.0, "grad_norm": 0.023995859548449516, "global_step": 4406, "epoch": 122, "lr": 0.001} +{"train_loss": 0.043599121272563934, "train_loss_bc": 0.04357912018895149, "train_loss_llm": 0.0020002424716949463, "grad_norm": 0.033089954406023026, "global_step": 4407, "epoch": 122, "lr": 0.001} +{"train_loss": 0.03463265299797058, "train_loss_bc": 0.03463265299797058, "train_loss_llm": 0.0, "grad_norm": 0.03726861625909805, "global_step": 4408, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04381343349814415, "train_loss_bc": 0.04381343349814415, "train_loss_llm": 0.0, "grad_norm": 0.01383980829268694, "global_step": 4409, "epoch": 122, "lr": 0.001} +{"train_loss": 0.035629794001579285, "train_loss_bc": 0.035629794001579285, "train_loss_llm": 0.0, "grad_norm": 0.02619427815079689, "global_step": 4410, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04196877032518387, "train_loss_bc": 0.04196877032518387, "train_loss_llm": 0.0, "grad_norm": 0.03616401180624962, "global_step": 4411, "epoch": 122, "lr": 0.001} +{"train_loss": 0.0430397093296051, "train_loss_bc": 0.04301237314939499, "train_loss_llm": 0.0027336180210113525, "grad_norm": 0.04168367758393288, "global_step": 4412, "epoch": 122, "lr": 0.001} +{"train_loss": 0.039198849350214005, "train_loss_bc": 0.039175860583782196, "train_loss_llm": 0.0022988319396972656, "grad_norm": 0.016861407086253166, "global_step": 4413, "epoch": 122, "lr": 0.001} +{"train_loss": 0.0453205332159996, "train_loss_bc": 0.0453205332159996, "train_loss_llm": 0.0, "grad_norm": 0.024065805599093437, "global_step": 4414, "epoch": 122, "lr": 0.001} +{"train_loss": 0.03989259526133537, "train_loss_bc": 0.03989259526133537, "train_loss_llm": 0.0, "grad_norm": 0.028733991086483, "global_step": 4415, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04916650056838989, "train_loss_bc": 0.04916650056838989, "train_loss_llm": 0.0, "grad_norm": 0.0301275085657835, "global_step": 4416, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04051802307367325, "train_loss_bc": 0.04051802307367325, "train_loss_llm": 0.0, "grad_norm": 0.0174137894064188, "global_step": 4417, "epoch": 122, "lr": 0.001} +{"train_loss": 0.0420854277908802, "train_loss_bc": 0.0420854277908802, "train_loss_llm": 0.0, "grad_norm": 0.02706051804125309, "global_step": 4418, "epoch": 122, "lr": 0.001} +{"train_loss": 0.03610260412096977, "train_loss_bc": 0.03610260412096977, "train_loss_llm": 0.0, "grad_norm": 0.03194810450077057, "global_step": 4419, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04124470055103302, "train_loss_bc": 0.04124470055103302, "train_loss_llm": 0.0, "grad_norm": 0.041413672268390656, "global_step": 4420, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04543629288673401, "train_loss_bc": 0.04543629288673401, "train_loss_llm": 0.0, "grad_norm": 0.019812583923339844, "global_step": 4421, "epoch": 122, "lr": 0.001} +{"train_loss": 0.03826424106955528, "train_loss_bc": 0.038244184106588364, "train_loss_llm": 0.002005666494369507, "grad_norm": 0.032292529940605164, "global_step": 4422, "epoch": 122, "lr": 0.001} +{"train_loss": 0.051135480403900146, "train_loss_bc": 0.051135480403900146, "train_loss_llm": 0.0, "grad_norm": 0.04257798567414284, "global_step": 4423, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04889819025993347, "train_loss_bc": 0.04888484627008438, "train_loss_llm": 0.0013343095779418945, "grad_norm": 0.05045982077717781, "global_step": 4424, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04864006116986275, "train_loss_bc": 0.04863078147172928, "train_loss_llm": 0.0009279549121856689, "grad_norm": 0.01774577610194683, "global_step": 4425, "epoch": 122, "lr": 0.001} +{"train_loss": 0.04393976926803589, "train_loss_bc": 0.04393976926803589, "train_loss_llm": 0.0, "grad_norm": 0.0313648022711277, "global_step": 4426, "epoch": 122, "lr": 0.001} +{"train_loss": 0.042650434395505324, "train_loss_bc": 0.0446416474878788, "train_loss_llm": 0.00110548734664917, "grad_norm": 0.050923902541399, "global_step": 4427, "epoch": 122, "lr": 0.001, "val_loss": 0.05323038995265961} +{"train_loss": 0.04443889111280441, "train_loss_bc": 0.044430721551179886, "train_loss_llm": 0.000817030668258667, "grad_norm": 0.05404903367161751, "global_step": 4428, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04418323561549187, "train_loss_bc": 0.04417850077152252, "train_loss_llm": 0.0004735291004180908, "grad_norm": 0.012833474203944206, "global_step": 4429, "epoch": 123, "lr": 0.001} +{"train_loss": 0.039577048271894455, "train_loss_bc": 0.03955664113163948, "train_loss_llm": 0.0020407438278198242, "grad_norm": 0.021274203434586525, "global_step": 4430, "epoch": 123, "lr": 0.001} +{"train_loss": 0.03903636708855629, "train_loss_bc": 0.039033740758895874, "train_loss_llm": 0.0002625882625579834, "grad_norm": 0.029002489522099495, "global_step": 4431, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04823578521609306, "train_loss_bc": 0.048218198120594025, "train_loss_llm": 0.0017585456371307373, "grad_norm": 0.035909634083509445, "global_step": 4432, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04538154974579811, "train_loss_bc": 0.04536273330450058, "train_loss_llm": 0.0018814802169799805, "grad_norm": 0.01796593703329563, "global_step": 4433, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04559183865785599, "train_loss_bc": 0.04559095948934555, "train_loss_llm": 8.806586265563965e-05, "grad_norm": 0.026467347517609596, "global_step": 4434, "epoch": 123, "lr": 0.001} +{"train_loss": 0.040777064859867096, "train_loss_bc": 0.040777064859867096, "train_loss_llm": 0.0, "grad_norm": 0.0319979302585125, "global_step": 4435, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04299400374293327, "train_loss_bc": 0.04299400374293327, "train_loss_llm": 0.0, "grad_norm": 0.04272459074854851, "global_step": 4436, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04435526579618454, "train_loss_bc": 0.044339586049318314, "train_loss_llm": 0.0015681087970733643, "grad_norm": 0.020481515675783157, "global_step": 4437, "epoch": 123, "lr": 0.001} +{"train_loss": 0.03809183090925217, "train_loss_bc": 0.03809183090925217, "train_loss_llm": 0.0, "grad_norm": 0.027330515906214714, "global_step": 4438, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04602944478392601, "train_loss_bc": 0.04602944478392601, "train_loss_llm": 0.0, "grad_norm": 0.041932009160518646, "global_step": 4439, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04289115592837334, "train_loss_bc": 0.04287911579012871, "train_loss_llm": 0.0012038946151733398, "grad_norm": 0.048733096569776535, "global_step": 4440, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04027635604143143, "train_loss_bc": 0.04027635604143143, "train_loss_llm": 0.0, "grad_norm": 0.013930337503552437, "global_step": 4441, "epoch": 123, "lr": 0.001} +{"train_loss": 0.044673629105091095, "train_loss_bc": 0.04467122629284859, "train_loss_llm": 0.0002402961254119873, "grad_norm": 0.025264915078878403, "global_step": 4442, "epoch": 123, "lr": 0.001} +{"train_loss": 0.05332095921039581, "train_loss_bc": 0.05332095921039581, "train_loss_llm": 0.0, "grad_norm": 0.03602393716573715, "global_step": 4443, "epoch": 123, "lr": 0.001} +{"train_loss": 0.046453241258859634, "train_loss_bc": 0.046453241258859634, "train_loss_llm": 0.0, "grad_norm": 0.043547362089157104, "global_step": 4444, "epoch": 123, "lr": 0.001} +{"train_loss": 0.03932805359363556, "train_loss_bc": 0.03932805359363556, "train_loss_llm": 0.0, "grad_norm": 0.017432058230042458, "global_step": 4445, "epoch": 123, "lr": 0.001} +{"train_loss": 0.040547505021095276, "train_loss_bc": 0.040547505021095276, "train_loss_llm": 0.0, "grad_norm": 0.023661818355321884, "global_step": 4446, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04098276048898697, "train_loss_bc": 0.040977124124765396, "train_loss_llm": 0.0005638003349304199, "grad_norm": 0.02848817966878414, "global_step": 4447, "epoch": 123, "lr": 0.001} +{"train_loss": 0.040789276361465454, "train_loss_bc": 0.040776629000902176, "train_loss_llm": 0.0012647509574890137, "grad_norm": 0.036799415946006775, "global_step": 4448, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04229436069726944, "train_loss_bc": 0.04229436069726944, "train_loss_llm": 0.0, "grad_norm": 0.014440467581152916, "global_step": 4449, "epoch": 123, "lr": 0.001} +{"train_loss": 0.0347176119685173, "train_loss_bc": 0.03470298647880554, "train_loss_llm": 0.0014625191688537598, "grad_norm": 0.021135179325938225, "global_step": 4450, "epoch": 123, "lr": 0.001} +{"train_loss": 0.042313240468502045, "train_loss_bc": 0.042313240468502045, "train_loss_llm": 0.0, "grad_norm": 0.02983093447983265, "global_step": 4451, "epoch": 123, "lr": 0.001} +{"train_loss": 0.03404109925031662, "train_loss_bc": 0.03404109925031662, "train_loss_llm": 0.0, "grad_norm": 0.03845812752842903, "global_step": 4452, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04081906005740166, "train_loss_bc": 0.04081906005740166, "train_loss_llm": 0.0, "grad_norm": 0.014312717132270336, "global_step": 4453, "epoch": 123, "lr": 0.001} +{"train_loss": 0.03599255159497261, "train_loss_bc": 0.03599255159497261, "train_loss_llm": 0.0, "grad_norm": 0.021482545882463455, "global_step": 4454, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04624653980135918, "train_loss_bc": 0.04624653980135918, "train_loss_llm": 0.0, "grad_norm": 0.029436685144901276, "global_step": 4455, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04851843789219856, "train_loss_bc": 0.04851843789219856, "train_loss_llm": 0.0, "grad_norm": 0.03927300497889519, "global_step": 4456, "epoch": 123, "lr": 0.001} +{"train_loss": 0.041566915810108185, "train_loss_bc": 0.041566915810108185, "train_loss_llm": 0.0, "grad_norm": 0.016332615166902542, "global_step": 4457, "epoch": 123, "lr": 0.001} +{"train_loss": 0.046722572296857834, "train_loss_bc": 0.046722572296857834, "train_loss_llm": 0.0, "grad_norm": 0.026235610246658325, "global_step": 4458, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04410278797149658, "train_loss_bc": 0.04410278797149658, "train_loss_llm": 0.0, "grad_norm": 0.03855544328689575, "global_step": 4459, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04568380489945412, "train_loss_bc": 0.04568380489945412, "train_loss_llm": 0.0, "grad_norm": 0.04644370824098587, "global_step": 4460, "epoch": 123, "lr": 0.001} +{"train_loss": 0.0357867106795311, "train_loss_bc": 0.0357867106795311, "train_loss_llm": 0.0, "grad_norm": 0.018069522455334663, "global_step": 4461, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04011033475399017, "train_loss_bc": 0.04009881615638733, "train_loss_llm": 0.0011518299579620361, "grad_norm": 0.020810892805457115, "global_step": 4462, "epoch": 123, "lr": 0.001} +{"train_loss": 0.04258157012777196, "train_loss_bc": 0.04604913294315338, "train_loss_llm": 0.0016101598739624023, "grad_norm": 0.03775925561785698, "global_step": 4463, "epoch": 123, "lr": 0.001, "val_loss": 0.05360199138522148} +{"train_loss": 0.047626059502363205, "train_loss_bc": 0.047626059502363205, "train_loss_llm": 0.0, "grad_norm": 0.045522332191467285, "global_step": 4464, "epoch": 124, "lr": 0.001} +{"train_loss": 0.043143130838871, "train_loss_bc": 0.043143130838871, "train_loss_llm": 0.0, "grad_norm": 0.016598576679825783, "global_step": 4465, "epoch": 124, "lr": 0.001} +{"train_loss": 0.05099828541278839, "train_loss_bc": 0.05099828541278839, "train_loss_llm": 0.0, "grad_norm": 0.0327388197183609, "global_step": 4466, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04210010543465614, "train_loss_bc": 0.04210010543465614, "train_loss_llm": 0.0, "grad_norm": 0.04535200819373131, "global_step": 4467, "epoch": 124, "lr": 0.001} +{"train_loss": 0.0420185923576355, "train_loss_bc": 0.0420185923576355, "train_loss_llm": 0.0, "grad_norm": 0.04962563514709473, "global_step": 4468, "epoch": 124, "lr": 0.001} +{"train_loss": 0.03458248823881149, "train_loss_bc": 0.03458248823881149, "train_loss_llm": 0.0, "grad_norm": 0.02130875550210476, "global_step": 4469, "epoch": 124, "lr": 0.001} +{"train_loss": 0.038455210626125336, "train_loss_bc": 0.038450583815574646, "train_loss_llm": 0.00046256184577941895, "grad_norm": 0.02752509154379368, "global_step": 4470, "epoch": 124, "lr": 0.001} +{"train_loss": 0.042355380952358246, "train_loss_bc": 0.042355380952358246, "train_loss_llm": 0.0, "grad_norm": 0.03466033563017845, "global_step": 4471, "epoch": 124, "lr": 0.001} +{"train_loss": 0.036675017327070236, "train_loss_bc": 0.036675017327070236, "train_loss_llm": 0.0, "grad_norm": 0.03899339959025383, "global_step": 4472, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04116092249751091, "train_loss_bc": 0.04116092249751091, "train_loss_llm": 0.0, "grad_norm": 0.01917792297899723, "global_step": 4473, "epoch": 124, "lr": 0.001} +{"train_loss": 0.0407114252448082, "train_loss_bc": 0.04069754481315613, "train_loss_llm": 0.001387953758239746, "grad_norm": 0.02846270427107811, "global_step": 4474, "epoch": 124, "lr": 0.001} +{"train_loss": 0.0457785539329052, "train_loss_bc": 0.045744750648736954, "train_loss_llm": 0.003380388021469116, "grad_norm": 0.033259011805057526, "global_step": 4475, "epoch": 124, "lr": 0.001} +{"train_loss": 0.028342634439468384, "train_loss_bc": 0.028342634439468384, "train_loss_llm": 0.0, "grad_norm": 0.040189098566770554, "global_step": 4476, "epoch": 124, "lr": 0.001} +{"train_loss": 0.03882117569446564, "train_loss_bc": 0.03882117569446564, "train_loss_llm": 0.0, "grad_norm": 0.018739929422736168, "global_step": 4477, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04222194850444794, "train_loss_bc": 0.04222194850444794, "train_loss_llm": 0.0, "grad_norm": 0.0275086872279644, "global_step": 4478, "epoch": 124, "lr": 0.001} +{"train_loss": 0.0459737628698349, "train_loss_bc": 0.04593305662274361, "train_loss_llm": 0.00407060980796814, "grad_norm": 0.04153282195329666, "global_step": 4479, "epoch": 124, "lr": 0.001} +{"train_loss": 0.046348828822374344, "train_loss_bc": 0.046348828822374344, "train_loss_llm": 0.0, "grad_norm": 0.0555475652217865, "global_step": 4480, "epoch": 124, "lr": 0.001} +{"train_loss": 0.03385802358388901, "train_loss_bc": 0.03385802358388901, "train_loss_llm": 0.0, "grad_norm": 0.012071189470589161, "global_step": 4481, "epoch": 124, "lr": 0.001} +{"train_loss": 0.033754102885723114, "train_loss_bc": 0.033754102885723114, "train_loss_llm": 0.0, "grad_norm": 0.02492057904601097, "global_step": 4482, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04349736496806145, "train_loss_bc": 0.043489813804626465, "train_loss_llm": 0.0007551610469818115, "grad_norm": 0.02360094152390957, "global_step": 4483, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04648531600832939, "train_loss_bc": 0.04648531600832939, "train_loss_llm": 0.0, "grad_norm": 0.029953161254525185, "global_step": 4484, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04171022027730942, "train_loss_bc": 0.04171022027730942, "train_loss_llm": 0.0, "grad_norm": 0.019201774150133133, "global_step": 4485, "epoch": 124, "lr": 0.001} +{"train_loss": 0.042906712740659714, "train_loss_bc": 0.04289783164858818, "train_loss_llm": 0.0008881986141204834, "grad_norm": 0.027058567851781845, "global_step": 4486, "epoch": 124, "lr": 0.001} +{"train_loss": 0.05117112770676613, "train_loss_bc": 0.051150061190128326, "train_loss_llm": 0.002106785774230957, "grad_norm": 0.038284026086330414, "global_step": 4487, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04779769480228424, "train_loss_bc": 0.04779769480228424, "train_loss_llm": 0.0, "grad_norm": 0.04030540585517883, "global_step": 4488, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04712187871336937, "train_loss_bc": 0.04712187871336937, "train_loss_llm": 0.0, "grad_norm": 0.015766534954309464, "global_step": 4489, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04553723707795143, "train_loss_bc": 0.04553723707795143, "train_loss_llm": 0.0, "grad_norm": 0.0228828564286232, "global_step": 4490, "epoch": 124, "lr": 0.001} +{"train_loss": 0.03843894228339195, "train_loss_bc": 0.03843851760029793, "train_loss_llm": 4.2557716369628906e-05, "grad_norm": 0.030047327280044556, "global_step": 4491, "epoch": 124, "lr": 0.001} +{"train_loss": 0.03886635601520538, "train_loss_bc": 0.03886635601520538, "train_loss_llm": 0.0, "grad_norm": 0.0324905589222908, "global_step": 4492, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04742877930402756, "train_loss_bc": 0.04741846024990082, "train_loss_llm": 0.001031726598739624, "grad_norm": 0.01888391189277172, "global_step": 4493, "epoch": 124, "lr": 0.001} +{"train_loss": 0.0400555245578289, "train_loss_bc": 0.0400555245578289, "train_loss_llm": 0.0, "grad_norm": 0.026399552822113037, "global_step": 4494, "epoch": 124, "lr": 0.001} +{"train_loss": 0.05278531461954117, "train_loss_bc": 0.05276426300406456, "train_loss_llm": 0.0021050572395324707, "grad_norm": 0.035723377019166946, "global_step": 4495, "epoch": 124, "lr": 0.001} +{"train_loss": 0.0454365611076355, "train_loss_bc": 0.04542508348822594, "train_loss_llm": 0.0011476874351501465, "grad_norm": 0.042104292660951614, "global_step": 4496, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04461148753762245, "train_loss_bc": 0.04461148753762245, "train_loss_llm": 0.0, "grad_norm": 0.02113841101527214, "global_step": 4497, "epoch": 124, "lr": 0.001} +{"train_loss": 0.041517965495586395, "train_loss_bc": 0.0415058508515358, "train_loss_llm": 0.0012116432189941406, "grad_norm": 0.02947542816400528, "global_step": 4498, "epoch": 124, "lr": 0.001} +{"train_loss": 0.04252375258753697, "train_loss_bc": 0.04048561677336693, "train_loss_llm": 0.00753435492515564, "grad_norm": 0.03317204490303993, "global_step": 4499, "epoch": 124, "lr": 0.001, "val_loss": 0.053308941423892975} +{"train_loss": 0.03274207189679146, "train_loss_bc": 0.032727401703596115, "train_loss_llm": 0.0014670491218566895, "grad_norm": 0.03700496256351471, "global_step": 4500, "epoch": 125, "lr": 0.001} +{"train_loss": 0.046517424285411835, "train_loss_bc": 0.046510886400938034, "train_loss_llm": 0.0006538927555084229, "grad_norm": 0.0157129168510437, "global_step": 4501, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04051194712519646, "train_loss_bc": 0.04050087183713913, "train_loss_llm": 0.0011074841022491455, "grad_norm": 0.028208555653691292, "global_step": 4502, "epoch": 125, "lr": 0.001} +{"train_loss": 0.0437374971807003, "train_loss_bc": 0.0437374971807003, "train_loss_llm": 0.0, "grad_norm": 0.0384022556245327, "global_step": 4503, "epoch": 125, "lr": 0.001} +{"train_loss": 0.03976628929376602, "train_loss_bc": 0.03975655138492584, "train_loss_llm": 0.0009737908840179443, "grad_norm": 0.051987405866384506, "global_step": 4504, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04520765319466591, "train_loss_bc": 0.04520765319466591, "train_loss_llm": 0.0, "grad_norm": 0.01626329869031906, "global_step": 4505, "epoch": 125, "lr": 0.001} +{"train_loss": 0.03510698303580284, "train_loss_bc": 0.03510698303580284, "train_loss_llm": 0.0, "grad_norm": 0.02188226394355297, "global_step": 4506, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04404699057340622, "train_loss_bc": 0.04403936117887497, "train_loss_llm": 0.0007631182670593262, "grad_norm": 0.031345952302217484, "global_step": 4507, "epoch": 125, "lr": 0.001} +{"train_loss": 0.051530539989471436, "train_loss_bc": 0.05151095613837242, "train_loss_llm": 0.0019582808017730713, "grad_norm": 0.032514143735170364, "global_step": 4508, "epoch": 125, "lr": 0.001} +{"train_loss": 0.03856716305017471, "train_loss_bc": 0.03856716305017471, "train_loss_llm": 0.0, "grad_norm": 0.015003844164311886, "global_step": 4509, "epoch": 125, "lr": 0.001} +{"train_loss": 0.03625991940498352, "train_loss_bc": 0.036251094192266464, "train_loss_llm": 0.0008824169635772705, "grad_norm": 0.0197158120572567, "global_step": 4510, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04889010265469551, "train_loss_bc": 0.04889010265469551, "train_loss_llm": 0.0, "grad_norm": 0.02849346026778221, "global_step": 4511, "epoch": 125, "lr": 0.001} +{"train_loss": 0.03945225104689598, "train_loss_bc": 0.03944438323378563, "train_loss_llm": 0.0007868707180023193, "grad_norm": 0.033163655549287796, "global_step": 4512, "epoch": 125, "lr": 0.001} +{"train_loss": 0.05223526805639267, "train_loss_bc": 0.05221717059612274, "train_loss_llm": 0.0018097460269927979, "grad_norm": 0.021459074690937996, "global_step": 4513, "epoch": 125, "lr": 0.001} +{"train_loss": 0.045045651495456696, "train_loss_bc": 0.045045651495456696, "train_loss_llm": 0.0, "grad_norm": 0.03457010164856911, "global_step": 4514, "epoch": 125, "lr": 0.001} +{"train_loss": 0.039052415639162064, "train_loss_bc": 0.039052415639162064, "train_loss_llm": 0.0, "grad_norm": 0.04474782943725586, "global_step": 4515, "epoch": 125, "lr": 0.001} +{"train_loss": 0.0383397601544857, "train_loss_bc": 0.0383397601544857, "train_loss_llm": 0.0, "grad_norm": 0.047654714435338974, "global_step": 4516, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04135683551430702, "train_loss_bc": 0.04135683551430702, "train_loss_llm": 0.0, "grad_norm": 0.016451207920908928, "global_step": 4517, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04399346560239792, "train_loss_bc": 0.04399346560239792, "train_loss_llm": 0.0, "grad_norm": 0.024627385661005974, "global_step": 4518, "epoch": 125, "lr": 0.001} +{"train_loss": 0.03870420157909393, "train_loss_bc": 0.03869103640317917, "train_loss_llm": 0.0013164281845092773, "grad_norm": 0.031413719058036804, "global_step": 4519, "epoch": 125, "lr": 0.001} +{"train_loss": 0.043553393334150314, "train_loss_bc": 0.043553393334150314, "train_loss_llm": 0.0, "grad_norm": 0.03297622874379158, "global_step": 4520, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04310188442468643, "train_loss_bc": 0.04310188442468643, "train_loss_llm": 0.0, "grad_norm": 0.016816318035125732, "global_step": 4521, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04294238239526749, "train_loss_bc": 0.04294238239526749, "train_loss_llm": 0.0, "grad_norm": 0.02619161456823349, "global_step": 4522, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04864398017525673, "train_loss_bc": 0.04864398017525673, "train_loss_llm": 0.0, "grad_norm": 0.038925718516111374, "global_step": 4523, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04392067715525627, "train_loss_bc": 0.04392067715525627, "train_loss_llm": 0.0, "grad_norm": 0.055088892579078674, "global_step": 4524, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04318331182003021, "train_loss_bc": 0.04318331182003021, "train_loss_llm": 0.0, "grad_norm": 0.019998205825686455, "global_step": 4525, "epoch": 125, "lr": 0.001} +{"train_loss": 0.046649470925331116, "train_loss_bc": 0.046637583523988724, "train_loss_llm": 0.0011887550354003906, "grad_norm": 0.025125646963715553, "global_step": 4526, "epoch": 125, "lr": 0.001} +{"train_loss": 0.050256483256816864, "train_loss_bc": 0.050256483256816864, "train_loss_llm": 0.0, "grad_norm": 0.03527326136827469, "global_step": 4527, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04097924754023552, "train_loss_bc": 0.04097924754023552, "train_loss_llm": 0.0, "grad_norm": 0.042739104479551315, "global_step": 4528, "epoch": 125, "lr": 0.001} +{"train_loss": 0.03603443503379822, "train_loss_bc": 0.03603443503379822, "train_loss_llm": 0.0, "grad_norm": 0.017206532880663872, "global_step": 4529, "epoch": 125, "lr": 0.001} +{"train_loss": 0.040193092077970505, "train_loss_bc": 0.040193092077970505, "train_loss_llm": 0.0, "grad_norm": 0.02552414871752262, "global_step": 4530, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04028100520372391, "train_loss_bc": 0.04028100520372391, "train_loss_llm": 0.0, "grad_norm": 0.027067754417657852, "global_step": 4531, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04731077700853348, "train_loss_bc": 0.04731077700853348, "train_loss_llm": 0.0, "grad_norm": 0.03315797448158264, "global_step": 4532, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04565749317407608, "train_loss_bc": 0.04563251510262489, "train_loss_llm": 0.002497941255569458, "grad_norm": 0.021596938371658325, "global_step": 4533, "epoch": 125, "lr": 0.001} +{"train_loss": 0.03531428799033165, "train_loss_bc": 0.0352899506688118, "train_loss_llm": 0.002433687448501587, "grad_norm": 0.03278475999832153, "global_step": 4534, "epoch": 125, "lr": 0.001} +{"train_loss": 0.04241889218489329, "train_loss_bc": 0.037993766367435455, "train_loss_llm": 0.0, "grad_norm": 0.03864940628409386, "global_step": 4535, "epoch": 125, "lr": 0.001, "train/cumulative_reward": 89.64988624658079, "train/mean_score": 11.732428079885796, "train/success_rate": 0.2, "test/cumulative_reward": 67.06044903559415, "test/mean_score": 7.815681351928244, "test/success_rate": 0.14, "val_loss": 0.05313679948449135, "train_action_mse_error": 0.046573106199502945} +{"train_loss": 0.040891218930482864, "train_loss_bc": 0.040891218930482864, "train_loss_llm": 0.0, "grad_norm": 0.044556569308042526, "global_step": 4536, "epoch": 126, "lr": 0.001} +{"train_loss": 0.037240270525217056, "train_loss_bc": 0.037240270525217056, "train_loss_llm": 0.0, "grad_norm": 0.01738937944173813, "global_step": 4537, "epoch": 126, "lr": 0.001} +{"train_loss": 0.038355909287929535, "train_loss_bc": 0.038355909287929535, "train_loss_llm": 0.0, "grad_norm": 0.02546827681362629, "global_step": 4538, "epoch": 126, "lr": 0.001} +{"train_loss": 0.0369148775935173, "train_loss_bc": 0.03691461309790611, "train_loss_llm": 2.6613473892211914e-05, "grad_norm": 0.030668288469314575, "global_step": 4539, "epoch": 126, "lr": 0.001} +{"train_loss": 0.035249944776296616, "train_loss_bc": 0.03524898365139961, "train_loss_llm": 9.605288505554199e-05, "grad_norm": 0.039277609437704086, "global_step": 4540, "epoch": 126, "lr": 0.001} +{"train_loss": 0.03451751172542572, "train_loss_bc": 0.03451751172542572, "train_loss_llm": 0.0, "grad_norm": 0.02079443633556366, "global_step": 4541, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04668967425823212, "train_loss_bc": 0.04668779671192169, "train_loss_llm": 0.00018769502639770508, "grad_norm": 0.03068705089390278, "global_step": 4542, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04041840881109238, "train_loss_bc": 0.0404122993350029, "train_loss_llm": 0.0006110072135925293, "grad_norm": 0.03357025608420372, "global_step": 4543, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04357330873608589, "train_loss_bc": 0.04357330873608589, "train_loss_llm": 0.0, "grad_norm": 0.037928543984889984, "global_step": 4544, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04542776942253113, "train_loss_bc": 0.04542776942253113, "train_loss_llm": 0.0, "grad_norm": 0.01522436086088419, "global_step": 4545, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04716605693101883, "train_loss_bc": 0.04716605693101883, "train_loss_llm": 0.0, "grad_norm": 0.021287458017468452, "global_step": 4546, "epoch": 126, "lr": 0.001} +{"train_loss": 0.050110891461372375, "train_loss_bc": 0.050110891461372375, "train_loss_llm": 0.0, "grad_norm": 0.029832709580659866, "global_step": 4547, "epoch": 126, "lr": 0.001} +{"train_loss": 0.03722643107175827, "train_loss_bc": 0.03722643107175827, "train_loss_llm": 0.0, "grad_norm": 0.03245047107338905, "global_step": 4548, "epoch": 126, "lr": 0.001} +{"train_loss": 0.044295892119407654, "train_loss_bc": 0.044295892119407654, "train_loss_llm": 0.0, "grad_norm": 0.019542071968317032, "global_step": 4549, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04585132747888565, "train_loss_bc": 0.04585132747888565, "train_loss_llm": 0.0, "grad_norm": 0.02195708639919758, "global_step": 4550, "epoch": 126, "lr": 0.001} +{"train_loss": 0.0422038696706295, "train_loss_bc": 0.0422038696706295, "train_loss_llm": 0.0, "grad_norm": 0.030253618955612183, "global_step": 4551, "epoch": 126, "lr": 0.001} +{"train_loss": 0.041256628930568695, "train_loss_bc": 0.041256628930568695, "train_loss_llm": 0.0, "grad_norm": 0.037348806858062744, "global_step": 4552, "epoch": 126, "lr": 0.001} +{"train_loss": 0.037777435034513474, "train_loss_bc": 0.037769027054309845, "train_loss_llm": 0.0008407533168792725, "grad_norm": 0.014630909077823162, "global_step": 4553, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04135080426931381, "train_loss_bc": 0.04135080426931381, "train_loss_llm": 0.0, "grad_norm": 0.0224978756159544, "global_step": 4554, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04585879668593407, "train_loss_bc": 0.04585879668593407, "train_loss_llm": 0.0, "grad_norm": 0.033097971230745316, "global_step": 4555, "epoch": 126, "lr": 0.001} +{"train_loss": 0.039563581347465515, "train_loss_bc": 0.039549898356199265, "train_loss_llm": 0.0013684630393981934, "grad_norm": 0.03761924058198929, "global_step": 4556, "epoch": 126, "lr": 0.001} +{"train_loss": 0.049222495406866074, "train_loss_bc": 0.049222495406866074, "train_loss_llm": 0.0, "grad_norm": 0.015471571125090122, "global_step": 4557, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04820626229047775, "train_loss_bc": 0.048181768506765366, "train_loss_llm": 0.002449333667755127, "grad_norm": 0.021693924441933632, "global_step": 4558, "epoch": 126, "lr": 0.001} +{"train_loss": 0.03825880587100983, "train_loss_bc": 0.03825880587100983, "train_loss_llm": 0.0, "grad_norm": 0.028104538097977638, "global_step": 4559, "epoch": 126, "lr": 0.001} +{"train_loss": 0.046463459730148315, "train_loss_bc": 0.046463459730148315, "train_loss_llm": 0.0, "grad_norm": 0.03699006512761116, "global_step": 4560, "epoch": 126, "lr": 0.001} +{"train_loss": 0.050441622734069824, "train_loss_bc": 0.05043411627411842, "train_loss_llm": 0.0007505118846893311, "grad_norm": 0.01920618675649166, "global_step": 4561, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04828089475631714, "train_loss_bc": 0.04828089475631714, "train_loss_llm": 0.0, "grad_norm": 0.026380470022559166, "global_step": 4562, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04069479927420616, "train_loss_bc": 0.04069479927420616, "train_loss_llm": 0.0, "grad_norm": 0.037862978875637054, "global_step": 4563, "epoch": 126, "lr": 0.001} +{"train_loss": 0.03466754034161568, "train_loss_bc": 0.03466728702187538, "train_loss_llm": 2.5331974029541016e-05, "grad_norm": 0.05010785534977913, "global_step": 4564, "epoch": 126, "lr": 0.001} +{"train_loss": 0.05149739235639572, "train_loss_bc": 0.05149739235639572, "train_loss_llm": 0.0, "grad_norm": 0.01578422822058201, "global_step": 4565, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04248444363474846, "train_loss_bc": 0.04248392954468727, "train_loss_llm": 5.131959915161133e-05, "grad_norm": 0.02162790298461914, "global_step": 4566, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04673865810036659, "train_loss_bc": 0.04673865810036659, "train_loss_llm": 0.0, "grad_norm": 0.030061250552535057, "global_step": 4567, "epoch": 126, "lr": 0.001} +{"train_loss": 0.03569938987493515, "train_loss_bc": 0.03569938987493515, "train_loss_llm": 0.0, "grad_norm": 0.040814854204654694, "global_step": 4568, "epoch": 126, "lr": 0.001} +{"train_loss": 0.04282943531870842, "train_loss_bc": 0.04280611127614975, "train_loss_llm": 0.0023325085639953613, "grad_norm": 0.015630898997187614, "global_step": 4569, "epoch": 126, "lr": 0.001} +{"train_loss": 0.03962527960538864, "train_loss_bc": 0.039624620229005814, "train_loss_llm": 6.601214408874512e-05, "grad_norm": 0.022289639338850975, "global_step": 4570, "epoch": 126, "lr": 0.001} +{"train_loss": 0.042409039516415864, "train_loss_bc": 0.03967433422803879, "train_loss_llm": 0.0, "grad_norm": 0.03384361416101456, "global_step": 4571, "epoch": 126, "lr": 0.001, "val_loss": 0.05354590713977814} +{"train_loss": 0.04397188127040863, "train_loss_bc": 0.04397188127040863, "train_loss_llm": 0.0, "grad_norm": 0.03508906811475754, "global_step": 4572, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04103580862283707, "train_loss_bc": 0.04103138670325279, "train_loss_llm": 0.00044220685958862305, "grad_norm": 0.014863030053675175, "global_step": 4573, "epoch": 127, "lr": 0.001} +{"train_loss": 0.041069064289331436, "train_loss_bc": 0.041069064289331436, "train_loss_llm": 0.0, "grad_norm": 0.025038480758666992, "global_step": 4574, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04325583949685097, "train_loss_bc": 0.04325583949685097, "train_loss_llm": 0.0, "grad_norm": 0.03355557098984718, "global_step": 4575, "epoch": 127, "lr": 0.001} +{"train_loss": 0.041051480919122696, "train_loss_bc": 0.041051480919122696, "train_loss_llm": 0.0, "grad_norm": 0.03962704911828041, "global_step": 4576, "epoch": 127, "lr": 0.001} +{"train_loss": 0.036231741309165955, "train_loss_bc": 0.036231741309165955, "train_loss_llm": 0.0, "grad_norm": 0.014159709215164185, "global_step": 4577, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04233649745583534, "train_loss_bc": 0.0423324778676033, "train_loss_llm": 0.00040200352668762207, "grad_norm": 0.021412111818790436, "global_step": 4578, "epoch": 127, "lr": 0.001} +{"train_loss": 0.03835858404636383, "train_loss_bc": 0.03835764154791832, "train_loss_llm": 9.40859317779541e-05, "grad_norm": 0.02713734097778797, "global_step": 4579, "epoch": 127, "lr": 0.001} +{"train_loss": 0.05194511264562607, "train_loss_bc": 0.05192512646317482, "train_loss_llm": 0.0019987821578979492, "grad_norm": 0.03512195870280266, "global_step": 4580, "epoch": 127, "lr": 0.001} +{"train_loss": 0.03672238439321518, "train_loss_bc": 0.03672238439321518, "train_loss_llm": 0.0, "grad_norm": 0.015010087750852108, "global_step": 4581, "epoch": 127, "lr": 0.001} +{"train_loss": 0.0393146350979805, "train_loss_bc": 0.0393146350979805, "train_loss_llm": 0.0, "grad_norm": 0.02346901036798954, "global_step": 4582, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04496132209897041, "train_loss_bc": 0.04496132209897041, "train_loss_llm": 0.0, "grad_norm": 0.02711673267185688, "global_step": 4583, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04355090111494064, "train_loss_bc": 0.04355090111494064, "train_loss_llm": 0.0, "grad_norm": 0.030355796217918396, "global_step": 4584, "epoch": 127, "lr": 0.001} +{"train_loss": 0.038588348776102066, "train_loss_bc": 0.03857872635126114, "train_loss_llm": 0.0009621977806091309, "grad_norm": 0.01489215437322855, "global_step": 4585, "epoch": 127, "lr": 0.001} +{"train_loss": 0.03784779831767082, "train_loss_bc": 0.03784779831767082, "train_loss_llm": 0.0, "grad_norm": 0.02347482182085514, "global_step": 4586, "epoch": 127, "lr": 0.001} +{"train_loss": 0.044104546308517456, "train_loss_bc": 0.04408508539199829, "train_loss_llm": 0.00194627046585083, "grad_norm": 0.0364864245057106, "global_step": 4587, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04784639552235603, "train_loss_bc": 0.04782326519489288, "train_loss_llm": 0.002313077449798584, "grad_norm": 0.04118508845567703, "global_step": 4588, "epoch": 127, "lr": 0.001} +{"train_loss": 0.046349991112947464, "train_loss_bc": 0.04633026197552681, "train_loss_llm": 0.0019729435443878174, "grad_norm": 0.016457444056868553, "global_step": 4589, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04020208865404129, "train_loss_bc": 0.04020208865404129, "train_loss_llm": 0.0, "grad_norm": 0.028216250240802765, "global_step": 4590, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04232679307460785, "train_loss_bc": 0.04232679307460785, "train_loss_llm": 0.0, "grad_norm": 0.040899816900491714, "global_step": 4591, "epoch": 127, "lr": 0.001} +{"train_loss": 0.043043989688158035, "train_loss_bc": 0.043043989688158035, "train_loss_llm": 0.0, "grad_norm": 0.04201732203364372, "global_step": 4592, "epoch": 127, "lr": 0.001} +{"train_loss": 0.03991912677884102, "train_loss_bc": 0.03990572318434715, "train_loss_llm": 0.001340419054031372, "grad_norm": 0.017688704654574394, "global_step": 4593, "epoch": 127, "lr": 0.001} +{"train_loss": 0.05221010372042656, "train_loss_bc": 0.05221010372042656, "train_loss_llm": 0.0, "grad_norm": 0.02937515638768673, "global_step": 4594, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04182293266057968, "train_loss_bc": 0.04182293266057968, "train_loss_llm": 0.0, "grad_norm": 0.038456372916698456, "global_step": 4595, "epoch": 127, "lr": 0.001} +{"train_loss": 0.047395192086696625, "train_loss_bc": 0.047395192086696625, "train_loss_llm": 0.0, "grad_norm": 0.04200904816389084, "global_step": 4596, "epoch": 127, "lr": 0.001} +{"train_loss": 0.043017350137233734, "train_loss_bc": 0.043002113699913025, "train_loss_llm": 0.0015234947204589844, "grad_norm": 0.020628873258829117, "global_step": 4597, "epoch": 127, "lr": 0.001} +{"train_loss": 0.03894342854619026, "train_loss_bc": 0.03894342854619026, "train_loss_llm": 0.0, "grad_norm": 0.022520678117871284, "global_step": 4598, "epoch": 127, "lr": 0.001} +{"train_loss": 0.03513379022479057, "train_loss_bc": 0.03513379022479057, "train_loss_llm": 0.0, "grad_norm": 0.03256620094180107, "global_step": 4599, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04349212720990181, "train_loss_bc": 0.04349212720990181, "train_loss_llm": 0.0, "grad_norm": 0.040027204900979996, "global_step": 4600, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04640696570277214, "train_loss_bc": 0.04640696570277214, "train_loss_llm": 0.0, "grad_norm": 0.017905013635754585, "global_step": 4601, "epoch": 127, "lr": 0.001} +{"train_loss": 0.03576810657978058, "train_loss_bc": 0.03576810657978058, "train_loss_llm": 0.0, "grad_norm": 0.022477027028799057, "global_step": 4602, "epoch": 127, "lr": 0.001} +{"train_loss": 0.03411903232336044, "train_loss_bc": 0.03409820422530174, "train_loss_llm": 0.002082645893096924, "grad_norm": 0.032884784042835236, "global_step": 4603, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04461151733994484, "train_loss_bc": 0.04461151733994484, "train_loss_llm": 0.0, "grad_norm": 0.044181421399116516, "global_step": 4604, "epoch": 127, "lr": 0.001} +{"train_loss": 0.044366225600242615, "train_loss_bc": 0.0443660132586956, "train_loss_llm": 2.1278858184814453e-05, "grad_norm": 0.019422614946961403, "global_step": 4605, "epoch": 127, "lr": 0.001} +{"train_loss": 0.05143764242529869, "train_loss_bc": 0.05143764242529869, "train_loss_llm": 0.0, "grad_norm": 0.027336250990629196, "global_step": 4606, "epoch": 127, "lr": 0.001} +{"train_loss": 0.04237944229195515, "train_loss_bc": 0.04286514222621918, "train_loss_llm": 0.0036035478115081787, "grad_norm": 0.033998358994722366, "global_step": 4607, "epoch": 127, "lr": 0.001, "val_loss": 0.05350422114133835} +{"train_loss": 0.046204473823308945, "train_loss_bc": 0.046204473823308945, "train_loss_llm": 0.0, "grad_norm": 0.04494677111506462, "global_step": 4608, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04639570415019989, "train_loss_bc": 0.04639570415019989, "train_loss_llm": 0.0, "grad_norm": 0.019357910379767418, "global_step": 4609, "epoch": 128, "lr": 0.001} +{"train_loss": 0.0378321148455143, "train_loss_bc": 0.037827298045158386, "train_loss_llm": 0.0004817843437194824, "grad_norm": 0.024054188281297684, "global_step": 4610, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04107329249382019, "train_loss_bc": 0.04107329249382019, "train_loss_llm": 0.0, "grad_norm": 0.030195869505405426, "global_step": 4611, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04533559828996658, "train_loss_bc": 0.04533559828996658, "train_loss_llm": 0.0, "grad_norm": 0.0330558679997921, "global_step": 4612, "epoch": 128, "lr": 0.001} +{"train_loss": 0.03539417311549187, "train_loss_bc": 0.03539417311549187, "train_loss_llm": 0.0, "grad_norm": 0.018234796822071075, "global_step": 4613, "epoch": 128, "lr": 0.001} +{"train_loss": 0.03948827460408211, "train_loss_bc": 0.03947422653436661, "train_loss_llm": 0.0014049410820007324, "grad_norm": 0.026354501023888588, "global_step": 4614, "epoch": 128, "lr": 0.001} +{"train_loss": 0.03584758937358856, "train_loss_bc": 0.03580975905060768, "train_loss_llm": 0.0037830471992492676, "grad_norm": 0.036050453782081604, "global_step": 4615, "epoch": 128, "lr": 0.001} +{"train_loss": 0.0461919941008091, "train_loss_bc": 0.0461919941008091, "train_loss_llm": 0.0, "grad_norm": 0.04105105623602867, "global_step": 4616, "epoch": 128, "lr": 0.001} +{"train_loss": 0.03641262277960777, "train_loss_bc": 0.03641262277960777, "train_loss_llm": 0.0, "grad_norm": 0.01666765660047531, "global_step": 4617, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04842841997742653, "train_loss_bc": 0.04842841997742653, "train_loss_llm": 0.0, "grad_norm": 0.02638765051960945, "global_step": 4618, "epoch": 128, "lr": 0.001} +{"train_loss": 0.0409538671374321, "train_loss_bc": 0.040950119495391846, "train_loss_llm": 0.00037467479705810547, "grad_norm": 0.03358721360564232, "global_step": 4619, "epoch": 128, "lr": 0.001} +{"train_loss": 0.046559687703847885, "train_loss_bc": 0.046559687703847885, "train_loss_llm": 0.0, "grad_norm": 0.03833835572004318, "global_step": 4620, "epoch": 128, "lr": 0.001} +{"train_loss": 0.039053402841091156, "train_loss_bc": 0.039053402841091156, "train_loss_llm": 0.0, "grad_norm": 0.01870274357497692, "global_step": 4621, "epoch": 128, "lr": 0.001} +{"train_loss": 0.0401713065803051, "train_loss_bc": 0.0401713065803051, "train_loss_llm": 0.0, "grad_norm": 0.02378656342625618, "global_step": 4622, "epoch": 128, "lr": 0.001} +{"train_loss": 0.05003543943166733, "train_loss_bc": 0.05003543943166733, "train_loss_llm": 0.0, "grad_norm": 0.03426514193415642, "global_step": 4623, "epoch": 128, "lr": 0.001} +{"train_loss": 0.0363147035241127, "train_loss_bc": 0.0363009013235569, "train_loss_llm": 0.0013802647590637207, "grad_norm": 0.04329507797956467, "global_step": 4624, "epoch": 128, "lr": 0.001} +{"train_loss": 0.030913783237338066, "train_loss_bc": 0.030913783237338066, "train_loss_llm": 0.0, "grad_norm": 0.01623551733791828, "global_step": 4625, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04231785982847214, "train_loss_bc": 0.04231785982847214, "train_loss_llm": 0.0, "grad_norm": 0.022943904623389244, "global_step": 4626, "epoch": 128, "lr": 0.001} +{"train_loss": 0.0434124581515789, "train_loss_bc": 0.0434124581515789, "train_loss_llm": 0.0, "grad_norm": 0.035280656069517136, "global_step": 4627, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04895436391234398, "train_loss_bc": 0.04895436391234398, "train_loss_llm": 0.0, "grad_norm": 0.03912436589598656, "global_step": 4628, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04269573837518692, "train_loss_bc": 0.04269340634346008, "train_loss_llm": 0.0002333223819732666, "grad_norm": 0.013251248747110367, "global_step": 4629, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04260624572634697, "train_loss_bc": 0.0425712987780571, "train_loss_llm": 0.003494828939437866, "grad_norm": 0.02312399074435234, "global_step": 4630, "epoch": 128, "lr": 0.001} +{"train_loss": 0.0410817489027977, "train_loss_bc": 0.04101823270320892, "train_loss_llm": 0.00635179877281189, "grad_norm": 0.026626000180840492, "global_step": 4631, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04791216924786568, "train_loss_bc": 0.04791216924786568, "train_loss_llm": 0.0, "grad_norm": 0.03202914074063301, "global_step": 4632, "epoch": 128, "lr": 0.001} +{"train_loss": 0.039540477097034454, "train_loss_bc": 0.03954008221626282, "train_loss_llm": 3.954768180847168e-05, "grad_norm": 0.018340634182095528, "global_step": 4633, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04068966209888458, "train_loss_bc": 0.04068966209888458, "train_loss_llm": 0.0, "grad_norm": 0.02472550980746746, "global_step": 4634, "epoch": 128, "lr": 0.001} +{"train_loss": 0.047657620161771774, "train_loss_bc": 0.047657620161771774, "train_loss_llm": 0.0, "grad_norm": 0.036321889609098434, "global_step": 4635, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04433530569076538, "train_loss_bc": 0.04433530569076538, "train_loss_llm": 0.0, "grad_norm": 0.046804070472717285, "global_step": 4636, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04427719488739967, "train_loss_bc": 0.04427637159824371, "train_loss_llm": 8.216500282287598e-05, "grad_norm": 0.01923583261668682, "global_step": 4637, "epoch": 128, "lr": 0.001} +{"train_loss": 0.039838965982198715, "train_loss_bc": 0.039838965982198715, "train_loss_llm": 0.0, "grad_norm": 0.03249398246407509, "global_step": 4638, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04306430742144585, "train_loss_bc": 0.04304680973291397, "train_loss_llm": 0.001749873161315918, "grad_norm": 0.044527098536491394, "global_step": 4639, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04659125208854675, "train_loss_bc": 0.04659125208854675, "train_loss_llm": 0.0, "grad_norm": 0.05164884030818939, "global_step": 4640, "epoch": 128, "lr": 0.001} +{"train_loss": 0.041622959077358246, "train_loss_bc": 0.041622959077358246, "train_loss_llm": 0.0, "grad_norm": 0.015645496547222137, "global_step": 4641, "epoch": 128, "lr": 0.001} +{"train_loss": 0.051272496581077576, "train_loss_bc": 0.051249586045742035, "train_loss_llm": 0.0022909939289093018, "grad_norm": 0.028616685420274734, "global_step": 4642, "epoch": 128, "lr": 0.001} +{"train_loss": 0.04235402706803547, "train_loss_bc": 0.03424052149057388, "train_loss_llm": 0.0027179718017578125, "grad_norm": 0.03435545042157173, "global_step": 4643, "epoch": 128, "lr": 0.001, "val_loss": 0.0535024031996727} +{"train_loss": 0.04387835040688515, "train_loss_bc": 0.04387835040688515, "train_loss_llm": 0.0, "grad_norm": 0.03418917581439018, "global_step": 4644, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04296640306711197, "train_loss_bc": 0.04296640306711197, "train_loss_llm": 0.0, "grad_norm": 0.01487769465893507, "global_step": 4645, "epoch": 129, "lr": 0.001} +{"train_loss": 0.05137401819229126, "train_loss_bc": 0.05137401819229126, "train_loss_llm": 0.0, "grad_norm": 0.024463610723614693, "global_step": 4646, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04103158786892891, "train_loss_bc": 0.04103049635887146, "train_loss_llm": 0.00010925531387329102, "grad_norm": 0.03269601985812187, "global_step": 4647, "epoch": 129, "lr": 0.001} +{"train_loss": 0.042447853833436966, "train_loss_bc": 0.042439136654138565, "train_loss_llm": 0.0008716881275177002, "grad_norm": 0.03872285410761833, "global_step": 4648, "epoch": 129, "lr": 0.001} +{"train_loss": 0.0443311482667923, "train_loss_bc": 0.0443311482667923, "train_loss_llm": 0.0, "grad_norm": 0.01727069355547428, "global_step": 4649, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04205978289246559, "train_loss_bc": 0.04205978289246559, "train_loss_llm": 0.0, "grad_norm": 0.026097573339939117, "global_step": 4650, "epoch": 129, "lr": 0.001} +{"train_loss": 0.042009223252534866, "train_loss_bc": 0.042002443224191666, "train_loss_llm": 0.0006779134273529053, "grad_norm": 0.032195404171943665, "global_step": 4651, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04302121326327324, "train_loss_bc": 0.04302121326327324, "train_loss_llm": 0.0, "grad_norm": 0.040215786546468735, "global_step": 4652, "epoch": 129, "lr": 0.001} +{"train_loss": 0.040717706084251404, "train_loss_bc": 0.040717706084251404, "train_loss_llm": 0.0, "grad_norm": 0.0182160884141922, "global_step": 4653, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04403107613325119, "train_loss_bc": 0.04403107613325119, "train_loss_llm": 0.0, "grad_norm": 0.02160411886870861, "global_step": 4654, "epoch": 129, "lr": 0.001} +{"train_loss": 0.047666870057582855, "train_loss_bc": 0.047666870057582855, "train_loss_llm": 0.0, "grad_norm": 0.029466357082128525, "global_step": 4655, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04098823294043541, "train_loss_bc": 0.04098823294043541, "train_loss_llm": 0.0, "grad_norm": 0.03183355927467346, "global_step": 4656, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04694322124123573, "train_loss_bc": 0.04694322124123573, "train_loss_llm": 0.0, "grad_norm": 0.016943035647273064, "global_step": 4657, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04399540647864342, "train_loss_bc": 0.04399540647864342, "train_loss_llm": 0.0, "grad_norm": 0.025763029232621193, "global_step": 4658, "epoch": 129, "lr": 0.001} +{"train_loss": 0.03803929686546326, "train_loss_bc": 0.03803929686546326, "train_loss_llm": 0.0, "grad_norm": 0.029978550970554352, "global_step": 4659, "epoch": 129, "lr": 0.001} +{"train_loss": 0.038478441536426544, "train_loss_bc": 0.038478441536426544, "train_loss_llm": 0.0, "grad_norm": 0.03746979311108589, "global_step": 4660, "epoch": 129, "lr": 0.001} +{"train_loss": 0.039157941937446594, "train_loss_bc": 0.03915765509009361, "train_loss_llm": 2.8640031814575195e-05, "grad_norm": 0.018116487190127373, "global_step": 4661, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04742223024368286, "train_loss_bc": 0.04742223024368286, "train_loss_llm": 0.0, "grad_norm": 0.0324539951980114, "global_step": 4662, "epoch": 129, "lr": 0.001} +{"train_loss": 0.0480496808886528, "train_loss_bc": 0.048042576760053635, "train_loss_llm": 0.0007105767726898193, "grad_norm": 0.03934584558010101, "global_step": 4663, "epoch": 129, "lr": 0.001} +{"train_loss": 0.03934089466929436, "train_loss_bc": 0.03934089466929436, "train_loss_llm": 0.0, "grad_norm": 0.05005452781915665, "global_step": 4664, "epoch": 129, "lr": 0.001} +{"train_loss": 0.03922830522060394, "train_loss_bc": 0.0392245352268219, "train_loss_llm": 0.0003770887851715088, "grad_norm": 0.019957492128014565, "global_step": 4665, "epoch": 129, "lr": 0.001} +{"train_loss": 0.042240045964717865, "train_loss_bc": 0.042240045964717865, "train_loss_llm": 0.0, "grad_norm": 0.029847685247659683, "global_step": 4666, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04161603003740311, "train_loss_bc": 0.04161603003740311, "train_loss_llm": 0.0, "grad_norm": 0.04515431076288223, "global_step": 4667, "epoch": 129, "lr": 0.001} +{"train_loss": 0.035244397819042206, "train_loss_bc": 0.035244397819042206, "train_loss_llm": 0.0, "grad_norm": 0.05177358537912369, "global_step": 4668, "epoch": 129, "lr": 0.001} +{"train_loss": 0.03825756162405014, "train_loss_bc": 0.03825202211737633, "train_loss_llm": 0.0005539357662200928, "grad_norm": 0.017123069614171982, "global_step": 4669, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04167494922876358, "train_loss_bc": 0.041651401668787, "train_loss_llm": 0.0023546218872070312, "grad_norm": 0.02381017617881298, "global_step": 4670, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04109599441289902, "train_loss_bc": 0.04109599441289902, "train_loss_llm": 0.0, "grad_norm": 0.03149591013789177, "global_step": 4671, "epoch": 129, "lr": 0.001} +{"train_loss": 0.0426630973815918, "train_loss_bc": 0.0426630973815918, "train_loss_llm": 0.0, "grad_norm": 0.03479016199707985, "global_step": 4672, "epoch": 129, "lr": 0.001} +{"train_loss": 0.043051548302173615, "train_loss_bc": 0.043051548302173615, "train_loss_llm": 0.0, "grad_norm": 0.014740625396370888, "global_step": 4673, "epoch": 129, "lr": 0.001} +{"train_loss": 0.03799060359597206, "train_loss_bc": 0.03799060359597206, "train_loss_llm": 0.0, "grad_norm": 0.020355308428406715, "global_step": 4674, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04118339344859123, "train_loss_bc": 0.04116280749440193, "train_loss_llm": 0.002058565616607666, "grad_norm": 0.028634151443839073, "global_step": 4675, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04904944822192192, "train_loss_bc": 0.04904944822192192, "train_loss_llm": 0.0, "grad_norm": 0.032355330884456635, "global_step": 4676, "epoch": 129, "lr": 0.001} +{"train_loss": 0.03805375099182129, "train_loss_bc": 0.038052719086408615, "train_loss_llm": 0.00010323524475097656, "grad_norm": 0.012909432873129845, "global_step": 4677, "epoch": 129, "lr": 0.001} +{"train_loss": 0.03857139125466347, "train_loss_bc": 0.03857139125466347, "train_loss_llm": 0.0, "grad_norm": 0.018831990659236908, "global_step": 4678, "epoch": 129, "lr": 0.001} +{"train_loss": 0.04238847674181064, "train_loss_bc": 0.04811406508088112, "train_loss_llm": 0.0, "grad_norm": 0.030836019665002823, "global_step": 4679, "epoch": 129, "lr": 0.001, "val_loss": 0.053884029388427734} +{"train_loss": 0.04599433019757271, "train_loss_bc": 0.04597684368491173, "train_loss_llm": 0.0017484724521636963, "grad_norm": 0.03847787529230118, "global_step": 4680, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04490166902542114, "train_loss_bc": 0.044900085777044296, "train_loss_llm": 0.0001582801342010498, "grad_norm": 0.01655707135796547, "global_step": 4681, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04335906729102135, "train_loss_bc": 0.04335906729102135, "train_loss_llm": 0.0, "grad_norm": 0.0282431747764349, "global_step": 4682, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04088970273733139, "train_loss_bc": 0.04088970273733139, "train_loss_llm": 0.0, "grad_norm": 0.029902193695306778, "global_step": 4683, "epoch": 130, "lr": 0.001} +{"train_loss": 0.05095637962222099, "train_loss_bc": 0.05095637962222099, "train_loss_llm": 0.0, "grad_norm": 0.03302527219057083, "global_step": 4684, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04381565377116203, "train_loss_bc": 0.04381565377116203, "train_loss_llm": 0.0, "grad_norm": 0.021667076274752617, "global_step": 4685, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04433143138885498, "train_loss_bc": 0.04433143138885498, "train_loss_llm": 0.0, "grad_norm": 0.03305865079164505, "global_step": 4686, "epoch": 130, "lr": 0.001} +{"train_loss": 0.042683087289333344, "train_loss_bc": 0.042683087289333344, "train_loss_llm": 0.0, "grad_norm": 0.044904034584760666, "global_step": 4687, "epoch": 130, "lr": 0.001} +{"train_loss": 0.03369151055812836, "train_loss_bc": 0.03369151055812836, "train_loss_llm": 0.0, "grad_norm": 0.04888635128736496, "global_step": 4688, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04061058163642883, "train_loss_bc": 0.04059065505862236, "train_loss_llm": 0.0019924938678741455, "grad_norm": 0.01436829287558794, "global_step": 4689, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04174510017037392, "train_loss_bc": 0.04173856973648071, "train_loss_llm": 0.000652998685836792, "grad_norm": 0.020392170175909996, "global_step": 4690, "epoch": 130, "lr": 0.001} +{"train_loss": 0.035451773554086685, "train_loss_bc": 0.03542197868227959, "train_loss_llm": 0.0029796063899993896, "grad_norm": 0.02827823907136917, "global_step": 4691, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04337293282151222, "train_loss_bc": 0.04336084797978401, "train_loss_llm": 0.001208484172821045, "grad_norm": 0.04002148285508156, "global_step": 4692, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04201177507638931, "train_loss_bc": 0.04201177507638931, "train_loss_llm": 0.0, "grad_norm": 0.01286376640200615, "global_step": 4693, "epoch": 130, "lr": 0.001} +{"train_loss": 0.03960344195365906, "train_loss_bc": 0.03960344195365906, "train_loss_llm": 0.0, "grad_norm": 0.021339161321520805, "global_step": 4694, "epoch": 130, "lr": 0.001} +{"train_loss": 0.047492191195487976, "train_loss_bc": 0.047492191195487976, "train_loss_llm": 0.0, "grad_norm": 0.02955036237835884, "global_step": 4695, "epoch": 130, "lr": 0.001} +{"train_loss": 0.046055298298597336, "train_loss_bc": 0.0460076704621315, "train_loss_llm": 0.004762768745422363, "grad_norm": 0.03789496421813965, "global_step": 4696, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04663249850273132, "train_loss_bc": 0.04663249850273132, "train_loss_llm": 0.0, "grad_norm": 0.017728112637996674, "global_step": 4697, "epoch": 130, "lr": 0.001} +{"train_loss": 0.045200467109680176, "train_loss_bc": 0.045197226107120514, "train_loss_llm": 0.0003241896629333496, "grad_norm": 0.02675829827785492, "global_step": 4698, "epoch": 130, "lr": 0.001} +{"train_loss": 0.0359995998442173, "train_loss_bc": 0.0359995998442173, "train_loss_llm": 0.0, "grad_norm": 0.031642694026231766, "global_step": 4699, "epoch": 130, "lr": 0.001} +{"train_loss": 0.03434281423687935, "train_loss_bc": 0.03434281423687935, "train_loss_llm": 0.0, "grad_norm": 0.03530648723244667, "global_step": 4700, "epoch": 130, "lr": 0.001} +{"train_loss": 0.035911284387111664, "train_loss_bc": 0.035911284387111664, "train_loss_llm": 0.0, "grad_norm": 0.012757272459566593, "global_step": 4701, "epoch": 130, "lr": 0.001} +{"train_loss": 0.042703479528427124, "train_loss_bc": 0.042703479528427124, "train_loss_llm": 0.0, "grad_norm": 0.025893883779644966, "global_step": 4702, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04252686724066734, "train_loss_bc": 0.04252686724066734, "train_loss_llm": 0.0, "grad_norm": 0.03749909624457359, "global_step": 4703, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04814538732171059, "train_loss_bc": 0.04813409969210625, "train_loss_llm": 0.001128852367401123, "grad_norm": 0.03743167221546173, "global_step": 4704, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04637642949819565, "train_loss_bc": 0.04637642949819565, "train_loss_llm": 0.0, "grad_norm": 0.023844104260206223, "global_step": 4705, "epoch": 130, "lr": 0.001} +{"train_loss": 0.03988330066204071, "train_loss_bc": 0.03988330066204071, "train_loss_llm": 0.0, "grad_norm": 0.036924123764038086, "global_step": 4706, "epoch": 130, "lr": 0.001} +{"train_loss": 0.042284950613975525, "train_loss_bc": 0.04226812720298767, "train_loss_llm": 0.0016823410987854004, "grad_norm": 0.047541674226522446, "global_step": 4707, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04381066933274269, "train_loss_bc": 0.04381066933274269, "train_loss_llm": 0.0, "grad_norm": 0.05174463987350464, "global_step": 4708, "epoch": 130, "lr": 0.001} +{"train_loss": 0.03833760321140289, "train_loss_bc": 0.03833760321140289, "train_loss_llm": 0.0, "grad_norm": 0.014280828647315502, "global_step": 4709, "epoch": 130, "lr": 0.001} +{"train_loss": 0.03799334540963173, "train_loss_bc": 0.0379931777715683, "train_loss_llm": 1.6748905181884766e-05, "grad_norm": 0.022946810349822044, "global_step": 4710, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04571264237165451, "train_loss_bc": 0.04568961635231972, "train_loss_llm": 0.002302706241607666, "grad_norm": 0.026805372908711433, "global_step": 4711, "epoch": 130, "lr": 0.001} +{"train_loss": 0.03803656995296478, "train_loss_bc": 0.03803656995296478, "train_loss_llm": 0.0, "grad_norm": 0.029973361641168594, "global_step": 4712, "epoch": 130, "lr": 0.001} +{"train_loss": 0.03855511546134949, "train_loss_bc": 0.03855511546134949, "train_loss_llm": 0.0, "grad_norm": 0.014242194592952728, "global_step": 4713, "epoch": 130, "lr": 0.001} +{"train_loss": 0.045747946947813034, "train_loss_bc": 0.045731354504823685, "train_loss_llm": 0.0016591846942901611, "grad_norm": 0.02259625308215618, "global_step": 4714, "epoch": 130, "lr": 0.001} +{"train_loss": 0.04232187672621674, "train_loss_bc": 0.04842066392302513, "train_loss_llm": 0.0, "grad_norm": 0.03848910704255104, "global_step": 4715, "epoch": 130, "lr": 0.001, "train/cumulative_reward": 107.66797754147395, "train/mean_score": 16.928336775258224, "train/success_rate": 0.4, "test/cumulative_reward": 57.975256780826804, "test/mean_score": 7.024548404135221, "test/success_rate": 0.14, "val_loss": 0.05386624485254288, "train_action_mse_error": 0.046387095004320145} +{"train_loss": 0.04410311579704285, "train_loss_bc": 0.04410311579704285, "train_loss_llm": 0.0, "grad_norm": 0.04524390399456024, "global_step": 4716, "epoch": 131, "lr": 0.001} +{"train_loss": 0.034867752343416214, "train_loss_bc": 0.03485194966197014, "train_loss_llm": 0.0015803873538970947, "grad_norm": 0.016263650730252266, "global_step": 4717, "epoch": 131, "lr": 0.001} +{"train_loss": 0.0422184094786644, "train_loss_bc": 0.0422184094786644, "train_loss_llm": 0.0, "grad_norm": 0.02999747171998024, "global_step": 4718, "epoch": 131, "lr": 0.001} +{"train_loss": 0.046122752130031586, "train_loss_bc": 0.046122752130031586, "train_loss_llm": 0.0, "grad_norm": 0.04213310778141022, "global_step": 4719, "epoch": 131, "lr": 0.001} +{"train_loss": 0.037072084844112396, "train_loss_bc": 0.037072084844112396, "train_loss_llm": 0.0, "grad_norm": 0.05504976958036423, "global_step": 4720, "epoch": 131, "lr": 0.001} +{"train_loss": 0.03798561170697212, "train_loss_bc": 0.03796249255537987, "train_loss_llm": 0.0023119449615478516, "grad_norm": 0.01652909629046917, "global_step": 4721, "epoch": 131, "lr": 0.001} +{"train_loss": 0.050043266266584396, "train_loss_bc": 0.050043266266584396, "train_loss_llm": 0.0, "grad_norm": 0.02739555947482586, "global_step": 4722, "epoch": 131, "lr": 0.001} +{"train_loss": 0.056306373327970505, "train_loss_bc": 0.056306373327970505, "train_loss_llm": 0.0, "grad_norm": 0.034375302493572235, "global_step": 4723, "epoch": 131, "lr": 0.001} +{"train_loss": 0.03219061717391014, "train_loss_bc": 0.032188672572374344, "train_loss_llm": 0.00019440054893493652, "grad_norm": 0.04033851996064186, "global_step": 4724, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04631694778800011, "train_loss_bc": 0.046271275728940964, "train_loss_llm": 0.004567176103591919, "grad_norm": 0.020999539643526077, "global_step": 4725, "epoch": 131, "lr": 0.001} +{"train_loss": 0.042954303324222565, "train_loss_bc": 0.042954303324222565, "train_loss_llm": 0.0, "grad_norm": 0.026783743873238564, "global_step": 4726, "epoch": 131, "lr": 0.001} +{"train_loss": 0.043521057814359665, "train_loss_bc": 0.043521057814359665, "train_loss_llm": 0.0, "grad_norm": 0.039927005767822266, "global_step": 4727, "epoch": 131, "lr": 0.001} +{"train_loss": 0.0495147779583931, "train_loss_bc": 0.0495147779583931, "train_loss_llm": 0.0, "grad_norm": 0.05629156902432442, "global_step": 4728, "epoch": 131, "lr": 0.001} +{"train_loss": 0.036926690489053726, "train_loss_bc": 0.03690215200185776, "train_loss_llm": 0.0024538636207580566, "grad_norm": 0.01572496071457863, "global_step": 4729, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04155053570866585, "train_loss_bc": 0.041529521346092224, "train_loss_llm": 0.002101391553878784, "grad_norm": 0.026728397235274315, "global_step": 4730, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04394719749689102, "train_loss_bc": 0.04394719749689102, "train_loss_llm": 0.0, "grad_norm": 0.03423149511218071, "global_step": 4731, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04077550396323204, "train_loss_bc": 0.04077550396323204, "train_loss_llm": 0.0, "grad_norm": 0.04329261928796768, "global_step": 4732, "epoch": 131, "lr": 0.001} +{"train_loss": 0.03536589816212654, "train_loss_bc": 0.03536107763648033, "train_loss_llm": 0.0004819035530090332, "grad_norm": 0.016595561057329178, "global_step": 4733, "epoch": 131, "lr": 0.001} +{"train_loss": 0.03877333179116249, "train_loss_bc": 0.03877333179116249, "train_loss_llm": 0.0, "grad_norm": 0.023608682677149773, "global_step": 4734, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04598423093557358, "train_loss_bc": 0.04598423093557358, "train_loss_llm": 0.0, "grad_norm": 0.02907257154583931, "global_step": 4735, "epoch": 131, "lr": 0.001} +{"train_loss": 0.043672580271959305, "train_loss_bc": 0.043672580271959305, "train_loss_llm": 0.0, "grad_norm": 0.04333029314875603, "global_step": 4736, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04027298837900162, "train_loss_bc": 0.04026990756392479, "train_loss_llm": 0.0003081858158111572, "grad_norm": 0.022027885541319847, "global_step": 4737, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04226433113217354, "train_loss_bc": 0.04226433113217354, "train_loss_llm": 0.0, "grad_norm": 0.0327000766992569, "global_step": 4738, "epoch": 131, "lr": 0.001} +{"train_loss": 0.03196942061185837, "train_loss_bc": 0.031952183693647385, "train_loss_llm": 0.0017237365245819092, "grad_norm": 0.04072004184126854, "global_step": 4739, "epoch": 131, "lr": 0.001} +{"train_loss": 0.044384412467479706, "train_loss_bc": 0.04438250511884689, "train_loss_llm": 0.00019073486328125, "grad_norm": 0.04357342794537544, "global_step": 4740, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04718778282403946, "train_loss_bc": 0.04718778282403946, "train_loss_llm": 0.0, "grad_norm": 0.01994328387081623, "global_step": 4741, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04330245777964592, "train_loss_bc": 0.04330245777964592, "train_loss_llm": 0.0, "grad_norm": 0.03089766576886177, "global_step": 4742, "epoch": 131, "lr": 0.001} +{"train_loss": 0.03974112123250961, "train_loss_bc": 0.03974112123250961, "train_loss_llm": 0.0, "grad_norm": 0.03933783620595932, "global_step": 4743, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04068273305892944, "train_loss_bc": 0.04068273305892944, "train_loss_llm": 0.0, "grad_norm": 0.0467124804854393, "global_step": 4744, "epoch": 131, "lr": 0.001} +{"train_loss": 0.033559903502464294, "train_loss_bc": 0.033559903502464294, "train_loss_llm": 0.0, "grad_norm": 0.018048850819468498, "global_step": 4745, "epoch": 131, "lr": 0.001} +{"train_loss": 0.051637906581163406, "train_loss_bc": 0.051637906581163406, "train_loss_llm": 0.0, "grad_norm": 0.03293982148170471, "global_step": 4746, "epoch": 131, "lr": 0.001} +{"train_loss": 0.044061511754989624, "train_loss_bc": 0.044061511754989624, "train_loss_llm": 0.0, "grad_norm": 0.04671795666217804, "global_step": 4747, "epoch": 131, "lr": 0.001} +{"train_loss": 0.052110396325588226, "train_loss_bc": 0.052109405398368835, "train_loss_llm": 9.92119312286377e-05, "grad_norm": 0.059840667992830276, "global_step": 4748, "epoch": 131, "lr": 0.001} +{"train_loss": 0.0499391183257103, "train_loss_bc": 0.0499391183257103, "train_loss_llm": 0.0, "grad_norm": 0.021111682057380676, "global_step": 4749, "epoch": 131, "lr": 0.001} +{"train_loss": 0.03657681494951248, "train_loss_bc": 0.03657681494951248, "train_loss_llm": 0.0, "grad_norm": 0.03212784603238106, "global_step": 4750, "epoch": 131, "lr": 0.001} +{"train_loss": 0.04237356078293589, "train_loss_bc": 0.03754425048828125, "train_loss_llm": 0.0, "grad_norm": 0.03997427970170975, "global_step": 4751, "epoch": 131, "lr": 0.001, "val_loss": 0.053418729454278946} +{"train_loss": 0.03977339342236519, "train_loss_bc": 0.03977339342236519, "train_loss_llm": 0.0, "grad_norm": 0.04912272095680237, "global_step": 4752, "epoch": 132, "lr": 0.001} +{"train_loss": 0.044004227966070175, "train_loss_bc": 0.044004227966070175, "train_loss_llm": 0.0, "grad_norm": 0.015847953036427498, "global_step": 4753, "epoch": 132, "lr": 0.001} +{"train_loss": 0.039096709340810776, "train_loss_bc": 0.03908238559961319, "train_loss_llm": 0.0014324188232421875, "grad_norm": 0.024046022444963455, "global_step": 4754, "epoch": 132, "lr": 0.001} +{"train_loss": 0.040888797491788864, "train_loss_bc": 0.040888797491788864, "train_loss_llm": 0.0, "grad_norm": 0.03363054245710373, "global_step": 4755, "epoch": 132, "lr": 0.001} +{"train_loss": 0.04661405831575394, "train_loss_bc": 0.04660433903336525, "train_loss_llm": 0.0009718537330627441, "grad_norm": 0.04050742834806442, "global_step": 4756, "epoch": 132, "lr": 0.001} +{"train_loss": 0.036895204335451126, "train_loss_bc": 0.03686322271823883, "train_loss_llm": 0.0031981170177459717, "grad_norm": 0.011617901735007763, "global_step": 4757, "epoch": 132, "lr": 0.001} +{"train_loss": 0.03653314337134361, "train_loss_bc": 0.03652592748403549, "train_loss_llm": 0.0007215142250061035, "grad_norm": 0.021873436868190765, "global_step": 4758, "epoch": 132, "lr": 0.001} +{"train_loss": 0.04535619542002678, "train_loss_bc": 0.04535619542002678, "train_loss_llm": 0.0, "grad_norm": 0.027237076312303543, "global_step": 4759, "epoch": 132, "lr": 0.001} +{"train_loss": 0.05261009931564331, "train_loss_bc": 0.05261009931564331, "train_loss_llm": 0.0, "grad_norm": 0.033473215997219086, "global_step": 4760, "epoch": 132, "lr": 0.001} +{"train_loss": 0.0459708534181118, "train_loss_bc": 0.04594998434185982, "train_loss_llm": 0.0020869076251983643, "grad_norm": 0.02105153165757656, "global_step": 4761, "epoch": 132, "lr": 0.001} +{"train_loss": 0.041385479271411896, "train_loss_bc": 0.041385479271411896, "train_loss_llm": 0.0, "grad_norm": 0.027166614308953285, "global_step": 4762, "epoch": 132, "lr": 0.001} +{"train_loss": 0.03788058087229729, "train_loss_bc": 0.03788058087229729, "train_loss_llm": 0.0, "grad_norm": 0.03566780686378479, "global_step": 4763, "epoch": 132, "lr": 0.001} +{"train_loss": 0.044116612523794174, "train_loss_bc": 0.04408276453614235, "train_loss_llm": 0.0033847689628601074, "grad_norm": 0.04692293331027031, "global_step": 4764, "epoch": 132, "lr": 0.001} +{"train_loss": 0.046393703669309616, "train_loss_bc": 0.04636675491929054, "train_loss_llm": 0.0026949644088745117, "grad_norm": 0.021527888253331184, "global_step": 4765, "epoch": 132, "lr": 0.001} +{"train_loss": 0.03704842925071716, "train_loss_bc": 0.03702240064740181, "train_loss_llm": 0.002602994441986084, "grad_norm": 0.0337977409362793, "global_step": 4766, "epoch": 132, "lr": 0.001} +{"train_loss": 0.045664772391319275, "train_loss_bc": 0.045635584741830826, "train_loss_llm": 0.002918720245361328, "grad_norm": 0.04796893149614334, "global_step": 4767, "epoch": 132, "lr": 0.001} +{"train_loss": 0.03774375468492508, "train_loss_bc": 0.03774375468492508, "train_loss_llm": 0.0, "grad_norm": 0.05998982861638069, "global_step": 4768, "epoch": 132, "lr": 0.001} +{"train_loss": 0.03774940222501755, "train_loss_bc": 0.03774940222501755, "train_loss_llm": 0.0, "grad_norm": 0.015980085358023643, "global_step": 4769, "epoch": 132, "lr": 0.001} +{"train_loss": 0.03989990055561066, "train_loss_bc": 0.03989990055561066, "train_loss_llm": 0.0, "grad_norm": 0.023764269426465034, "global_step": 4770, "epoch": 132, "lr": 0.001} +{"train_loss": 0.038252007216215134, "train_loss_bc": 0.038252007216215134, "train_loss_llm": 0.0, "grad_norm": 0.02923973649740219, "global_step": 4771, "epoch": 132, "lr": 0.001} +{"train_loss": 0.041997652500867844, "train_loss_bc": 0.041997652500867844, "train_loss_llm": 0.0, "grad_norm": 0.04224168881773949, "global_step": 4772, "epoch": 132, "lr": 0.001} +{"train_loss": 0.04868236556649208, "train_loss_bc": 0.04868236556649208, "train_loss_llm": 0.0, "grad_norm": 0.018941467627882957, "global_step": 4773, "epoch": 132, "lr": 0.001} +{"train_loss": 0.04252854362130165, "train_loss_bc": 0.04252854362130165, "train_loss_llm": 0.0, "grad_norm": 0.027398118749260902, "global_step": 4774, "epoch": 132, "lr": 0.001} +{"train_loss": 0.04158063977956772, "train_loss_bc": 0.04158063977956772, "train_loss_llm": 0.0, "grad_norm": 0.034899916499853134, "global_step": 4775, "epoch": 132, "lr": 0.001} +{"train_loss": 0.038056571036577225, "train_loss_bc": 0.038056571036577225, "train_loss_llm": 0.0, "grad_norm": 0.04075532406568527, "global_step": 4776, "epoch": 132, "lr": 0.001} +{"train_loss": 0.048981595784425735, "train_loss_bc": 0.048981595784425735, "train_loss_llm": 0.0, "grad_norm": 0.017359217628836632, "global_step": 4777, "epoch": 132, "lr": 0.001} +{"train_loss": 0.038280196487903595, "train_loss_bc": 0.038280196487903595, "train_loss_llm": 0.0, "grad_norm": 0.027807606384158134, "global_step": 4778, "epoch": 132, "lr": 0.001} +{"train_loss": 0.04407418519258499, "train_loss_bc": 0.04407418519258499, "train_loss_llm": 0.0, "grad_norm": 0.03260030597448349, "global_step": 4779, "epoch": 132, "lr": 0.001} +{"train_loss": 0.04249602556228638, "train_loss_bc": 0.04249602556228638, "train_loss_llm": 0.0, "grad_norm": 0.03717101365327835, "global_step": 4780, "epoch": 132, "lr": 0.001} +{"train_loss": 0.037211865186691284, "train_loss_bc": 0.037197645753622055, "train_loss_llm": 0.0014220476150512695, "grad_norm": 0.023593949154019356, "global_step": 4781, "epoch": 132, "lr": 0.001} +{"train_loss": 0.04089714214205742, "train_loss_bc": 0.04089714214205742, "train_loss_llm": 0.0, "grad_norm": 0.037484683096408844, "global_step": 4782, "epoch": 132, "lr": 0.001} +{"train_loss": 0.0411517396569252, "train_loss_bc": 0.0411517396569252, "train_loss_llm": 0.0, "grad_norm": 0.05537077784538269, "global_step": 4783, "epoch": 132, "lr": 0.001} +{"train_loss": 0.04462882876396179, "train_loss_bc": 0.04462882876396179, "train_loss_llm": 0.0, "grad_norm": 0.06901954859495163, "global_step": 4784, "epoch": 132, "lr": 0.001} +{"train_loss": 0.048804644495248795, "train_loss_bc": 0.048804644495248795, "train_loss_llm": 0.0, "grad_norm": 0.015891244634985924, "global_step": 4785, "epoch": 132, "lr": 0.001} +{"train_loss": 0.04018431529402733, "train_loss_bc": 0.04018431529402733, "train_loss_llm": 0.0, "grad_norm": 0.022243037819862366, "global_step": 4786, "epoch": 132, "lr": 0.001} +{"train_loss": 0.042415705819924675, "train_loss_bc": 0.05350649729371071, "train_loss_llm": 0.002527594566345215, "grad_norm": 0.045148346573114395, "global_step": 4787, "epoch": 132, "lr": 0.001, "val_loss": 0.05398620665073395} +{"train_loss": 0.04660608991980553, "train_loss_bc": 0.04660608991980553, "train_loss_llm": 0.0, "grad_norm": 0.06101014092564583, "global_step": 4788, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04242578148841858, "train_loss_bc": 0.04242578148841858, "train_loss_llm": 0.0, "grad_norm": 0.01586984284222126, "global_step": 4789, "epoch": 133, "lr": 0.001} +{"train_loss": 0.042176589369773865, "train_loss_bc": 0.042176589369773865, "train_loss_llm": 0.0, "grad_norm": 0.021831726655364037, "global_step": 4790, "epoch": 133, "lr": 0.001} +{"train_loss": 0.038238853216171265, "train_loss_bc": 0.038238853216171265, "train_loss_llm": 0.0, "grad_norm": 0.031776268035173416, "global_step": 4791, "epoch": 133, "lr": 0.001} +{"train_loss": 0.041094839572906494, "train_loss_bc": 0.041094839572906494, "train_loss_llm": 0.0, "grad_norm": 0.039862845093011856, "global_step": 4792, "epoch": 133, "lr": 0.001} +{"train_loss": 0.03955811262130737, "train_loss_bc": 0.03954000771045685, "train_loss_llm": 0.0018105506896972656, "grad_norm": 0.016216881573200226, "global_step": 4793, "epoch": 133, "lr": 0.001} +{"train_loss": 0.039194945245981216, "train_loss_bc": 0.03917771950364113, "train_loss_llm": 0.0017224252223968506, "grad_norm": 0.02075686864554882, "global_step": 4794, "epoch": 133, "lr": 0.001} +{"train_loss": 0.03687885031104088, "train_loss_bc": 0.03686957433819771, "train_loss_llm": 0.0009274482727050781, "grad_norm": 0.027707429602742195, "global_step": 4795, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04047328978776932, "train_loss_bc": 0.040453046560287476, "train_loss_llm": 0.0020245015621185303, "grad_norm": 0.03433386608958244, "global_step": 4796, "epoch": 133, "lr": 0.001} +{"train_loss": 0.03328780457377434, "train_loss_bc": 0.03328780457377434, "train_loss_llm": 0.0, "grad_norm": 0.014644862152636051, "global_step": 4797, "epoch": 133, "lr": 0.001} +{"train_loss": 0.0435066744685173, "train_loss_bc": 0.04349658638238907, "train_loss_llm": 0.0010087788105010986, "grad_norm": 0.02555975876748562, "global_step": 4798, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04246201738715172, "train_loss_bc": 0.04246201738715172, "train_loss_llm": 0.0, "grad_norm": 0.03670712187886238, "global_step": 4799, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04253930598497391, "train_loss_bc": 0.04253900796175003, "train_loss_llm": 2.9653310775756836e-05, "grad_norm": 0.05168510228395462, "global_step": 4800, "epoch": 133, "lr": 0.001} +{"train_loss": 0.0497693307697773, "train_loss_bc": 0.049747344106435776, "train_loss_llm": 0.0021986961364746094, "grad_norm": 0.022135455161333084, "global_step": 4801, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04532109200954437, "train_loss_bc": 0.04529181867837906, "train_loss_llm": 0.0029271841049194336, "grad_norm": 0.03747488930821419, "global_step": 4802, "epoch": 133, "lr": 0.001} +{"train_loss": 0.047478120774030685, "train_loss_bc": 0.047478120774030685, "train_loss_llm": 0.0, "grad_norm": 0.049104440957307816, "global_step": 4803, "epoch": 133, "lr": 0.001} +{"train_loss": 0.037871357053518295, "train_loss_bc": 0.037871357053518295, "train_loss_llm": 0.0, "grad_norm": 0.06208197399973869, "global_step": 4804, "epoch": 133, "lr": 0.001} +{"train_loss": 0.03952775150537491, "train_loss_bc": 0.03952775150537491, "train_loss_llm": 0.0, "grad_norm": 0.01776496134698391, "global_step": 4805, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04120827093720436, "train_loss_bc": 0.04120827093720436, "train_loss_llm": 0.0, "grad_norm": 0.027989724650979042, "global_step": 4806, "epoch": 133, "lr": 0.001} +{"train_loss": 0.0402202345430851, "train_loss_bc": 0.0402202345430851, "train_loss_llm": 0.0, "grad_norm": 0.03887644782662392, "global_step": 4807, "epoch": 133, "lr": 0.001} +{"train_loss": 0.051319003105163574, "train_loss_bc": 0.051314741373062134, "train_loss_llm": 0.00042629241943359375, "grad_norm": 0.05007965862751007, "global_step": 4808, "epoch": 133, "lr": 0.001} +{"train_loss": 0.035432662814855576, "train_loss_bc": 0.03541296347975731, "train_loss_llm": 0.0019700229167938232, "grad_norm": 0.016125980764627457, "global_step": 4809, "epoch": 133, "lr": 0.001} +{"train_loss": 0.044747449457645416, "train_loss_bc": 0.044747449457645416, "train_loss_llm": 0.0, "grad_norm": 0.025319047272205353, "global_step": 4810, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04598366841673851, "train_loss_bc": 0.04598366841673851, "train_loss_llm": 0.0, "grad_norm": 0.03672007471323013, "global_step": 4811, "epoch": 133, "lr": 0.001} +{"train_loss": 0.039852313697338104, "train_loss_bc": 0.03984900563955307, "train_loss_llm": 0.00033080577850341797, "grad_norm": 0.04393424466252327, "global_step": 4812, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04093398153781891, "train_loss_bc": 0.04093398153781891, "train_loss_llm": 0.0, "grad_norm": 0.014734968543052673, "global_step": 4813, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04309190437197685, "train_loss_bc": 0.04309190437197685, "train_loss_llm": 0.0, "grad_norm": 0.021277330815792084, "global_step": 4814, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04422855004668236, "train_loss_bc": 0.04420987144112587, "train_loss_llm": 0.0018679797649383545, "grad_norm": 0.026920141652226448, "global_step": 4815, "epoch": 133, "lr": 0.001} +{"train_loss": 0.044128403067588806, "train_loss_bc": 0.044128403067588806, "train_loss_llm": 0.0, "grad_norm": 0.03271867334842682, "global_step": 4816, "epoch": 133, "lr": 0.001} +{"train_loss": 0.0472123920917511, "train_loss_bc": 0.0472123920917511, "train_loss_llm": 0.0, "grad_norm": 0.01583452709019184, "global_step": 4817, "epoch": 133, "lr": 0.001} +{"train_loss": 0.03880569711327553, "train_loss_bc": 0.0388001911342144, "train_loss_llm": 0.0005506575107574463, "grad_norm": 0.025934385135769844, "global_step": 4818, "epoch": 133, "lr": 0.001} +{"train_loss": 0.0363132618367672, "train_loss_bc": 0.03631127253174782, "train_loss_llm": 0.0001990199089050293, "grad_norm": 0.031390078365802765, "global_step": 4819, "epoch": 133, "lr": 0.001} +{"train_loss": 0.047260310500860214, "train_loss_bc": 0.047247134149074554, "train_loss_llm": 0.0013176500797271729, "grad_norm": 0.03906102851033211, "global_step": 4820, "epoch": 133, "lr": 0.001} +{"train_loss": 0.049881137907505035, "train_loss_bc": 0.049881137907505035, "train_loss_llm": 0.0, "grad_norm": 0.02117207646369934, "global_step": 4821, "epoch": 133, "lr": 0.001} +{"train_loss": 0.04056520015001297, "train_loss_bc": 0.04056520015001297, "train_loss_llm": 0.0, "grad_norm": 0.02660023234784603, "global_step": 4822, "epoch": 133, "lr": 0.001} +{"train_loss": 0.042235778748161264, "train_loss_bc": 0.04089278727769852, "train_loss_llm": 0.0, "grad_norm": 0.039312973618507385, "global_step": 4823, "epoch": 133, "lr": 0.001, "val_loss": 0.05397462472319603} +{"train_loss": 0.041819263249635696, "train_loss_bc": 0.041819263249635696, "train_loss_llm": 0.0, "grad_norm": 0.04555642604827881, "global_step": 4824, "epoch": 134, "lr": 0.001} +{"train_loss": 0.03821168094873428, "train_loss_bc": 0.03821168094873428, "train_loss_llm": 0.0, "grad_norm": 0.022776754572987556, "global_step": 4825, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04221517965197563, "train_loss_bc": 0.042193394154310226, "train_loss_llm": 0.0021784603595733643, "grad_norm": 0.032666951417922974, "global_step": 4826, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04504236951470375, "train_loss_bc": 0.04504118487238884, "train_loss_llm": 0.00011840462684631348, "grad_norm": 0.04096236079931259, "global_step": 4827, "epoch": 134, "lr": 0.001} +{"train_loss": 0.03880050778388977, "train_loss_bc": 0.03878749907016754, "train_loss_llm": 0.0013009607791900635, "grad_norm": 0.048940565437078476, "global_step": 4828, "epoch": 134, "lr": 0.001} +{"train_loss": 0.042275574058294296, "train_loss_bc": 0.04226631671190262, "train_loss_llm": 0.0009258687496185303, "grad_norm": 0.017941735684871674, "global_step": 4829, "epoch": 134, "lr": 0.001} +{"train_loss": 0.036857083439826965, "train_loss_bc": 0.036857083439826965, "train_loss_llm": 0.0, "grad_norm": 0.028113994747400284, "global_step": 4830, "epoch": 134, "lr": 0.001} +{"train_loss": 0.03611583262681961, "train_loss_bc": 0.03611583262681961, "train_loss_llm": 0.0, "grad_norm": 0.03232845291495323, "global_step": 4831, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04163368046283722, "train_loss_bc": 0.04163368046283722, "train_loss_llm": 0.0, "grad_norm": 0.036810390651226044, "global_step": 4832, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04490210488438606, "train_loss_bc": 0.04490210488438606, "train_loss_llm": 0.0, "grad_norm": 0.02356118895113468, "global_step": 4833, "epoch": 134, "lr": 0.001} +{"train_loss": 0.03622663393616676, "train_loss_bc": 0.03622663393616676, "train_loss_llm": 0.0, "grad_norm": 0.028505291789770126, "global_step": 4834, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04299731180071831, "train_loss_bc": 0.042992204427719116, "train_loss_llm": 0.0005108118057250977, "grad_norm": 0.03582830727100372, "global_step": 4835, "epoch": 134, "lr": 0.001} +{"train_loss": 0.0459958016872406, "train_loss_bc": 0.04598981887102127, "train_loss_llm": 0.0005984008312225342, "grad_norm": 0.045516110956668854, "global_step": 4836, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04266129806637764, "train_loss_bc": 0.042642541229724884, "train_loss_llm": 0.001875549554824829, "grad_norm": 0.022893210873007774, "global_step": 4837, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04259948804974556, "train_loss_bc": 0.04259948804974556, "train_loss_llm": 0.0, "grad_norm": 0.03648190200328827, "global_step": 4838, "epoch": 134, "lr": 0.001} +{"train_loss": 0.047738298773765564, "train_loss_bc": 0.047738298773765564, "train_loss_llm": 0.0, "grad_norm": 0.043830227106809616, "global_step": 4839, "epoch": 134, "lr": 0.001} +{"train_loss": 0.044086601585149765, "train_loss_bc": 0.044086601585149765, "train_loss_llm": 0.0, "grad_norm": 0.05564005300402641, "global_step": 4840, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04202580451965332, "train_loss_bc": 0.04202580451965332, "train_loss_llm": 0.0, "grad_norm": 0.019903339445590973, "global_step": 4841, "epoch": 134, "lr": 0.001} +{"train_loss": 0.03660988807678223, "train_loss_bc": 0.03659607470035553, "train_loss_llm": 0.0013813674449920654, "grad_norm": 0.03544008731842041, "global_step": 4842, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04461142420768738, "train_loss_bc": 0.04459606856107712, "train_loss_llm": 0.0015354752540588379, "grad_norm": 0.04324803128838539, "global_step": 4843, "epoch": 134, "lr": 0.001} +{"train_loss": 0.0519084669649601, "train_loss_bc": 0.0519084669649601, "train_loss_llm": 0.0, "grad_norm": 0.060313042253255844, "global_step": 4844, "epoch": 134, "lr": 0.001} +{"train_loss": 0.048049647361040115, "train_loss_bc": 0.048049647361040115, "train_loss_llm": 0.0, "grad_norm": 0.017821744084358215, "global_step": 4845, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04079796373844147, "train_loss_bc": 0.04079796373844147, "train_loss_llm": 0.0, "grad_norm": 0.03184765577316284, "global_step": 4846, "epoch": 134, "lr": 0.001} +{"train_loss": 0.038492053747177124, "train_loss_bc": 0.038492053747177124, "train_loss_llm": 0.0, "grad_norm": 0.04235536605119705, "global_step": 4847, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04868142306804657, "train_loss_bc": 0.04868142306804657, "train_loss_llm": 0.0, "grad_norm": 0.04445573315024376, "global_step": 4848, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04160032793879509, "train_loss_bc": 0.04159620776772499, "train_loss_llm": 0.0004118680953979492, "grad_norm": 0.01876244693994522, "global_step": 4849, "epoch": 134, "lr": 0.001} +{"train_loss": 0.03600946441292763, "train_loss_bc": 0.03600946441292763, "train_loss_llm": 0.0, "grad_norm": 0.0339154452085495, "global_step": 4850, "epoch": 134, "lr": 0.001} +{"train_loss": 0.03369348123669624, "train_loss_bc": 0.033670078963041306, "train_loss_llm": 0.0023403167724609375, "grad_norm": 0.042250242084264755, "global_step": 4851, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04139711335301399, "train_loss_bc": 0.04139711335301399, "train_loss_llm": 0.0, "grad_norm": 0.054982393980026245, "global_step": 4852, "epoch": 134, "lr": 0.001} +{"train_loss": 0.042178038507699966, "train_loss_bc": 0.042178038507699966, "train_loss_llm": 0.0, "grad_norm": 0.01694353111088276, "global_step": 4853, "epoch": 134, "lr": 0.001} +{"train_loss": 0.04906651750206947, "train_loss_bc": 0.04906651750206947, "train_loss_llm": 0.0, "grad_norm": 0.027017243206501007, "global_step": 4854, "epoch": 134, "lr": 0.001} +{"train_loss": 0.035996850579977036, "train_loss_bc": 0.035954929888248444, "train_loss_llm": 0.004192084074020386, "grad_norm": 0.03280578926205635, "global_step": 4855, "epoch": 134, "lr": 0.001} +{"train_loss": 0.048377301543951035, "train_loss_bc": 0.048377301543951035, "train_loss_llm": 0.0, "grad_norm": 0.039532121270895004, "global_step": 4856, "epoch": 134, "lr": 0.001} +{"train_loss": 0.051885105669498444, "train_loss_bc": 0.051885105669498444, "train_loss_llm": 0.0, "grad_norm": 0.020677175372838974, "global_step": 4857, "epoch": 134, "lr": 0.001} +{"train_loss": 0.039336275309324265, "train_loss_bc": 0.039336275309324265, "train_loss_llm": 0.0, "grad_norm": 0.02715027704834938, "global_step": 4858, "epoch": 134, "lr": 0.001} +{"train_loss": 0.0423856678729256, "train_loss_bc": 0.04496258497238159, "train_loss_llm": 0.002560138702392578, "grad_norm": 0.044446639716625214, "global_step": 4859, "epoch": 134, "lr": 0.001, "val_loss": 0.05384622886776924} +{"train_loss": 0.03791913762688637, "train_loss_bc": 0.03791913762688637, "train_loss_llm": 0.0, "grad_norm": 0.04651585593819618, "global_step": 4860, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04437336325645447, "train_loss_bc": 0.04437336325645447, "train_loss_llm": 0.0, "grad_norm": 0.017620956525206566, "global_step": 4861, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04549171403050423, "train_loss_bc": 0.04549171403050423, "train_loss_llm": 0.0, "grad_norm": 0.030963188037276268, "global_step": 4862, "epoch": 135, "lr": 0.001} +{"train_loss": 0.038260411471128464, "train_loss_bc": 0.038260411471128464, "train_loss_llm": 0.0, "grad_norm": 0.044947028160095215, "global_step": 4863, "epoch": 135, "lr": 0.001} +{"train_loss": 0.03825307637453079, "train_loss_bc": 0.03822806850075722, "train_loss_llm": 0.0025008320808410645, "grad_norm": 0.051394060254096985, "global_step": 4864, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04062473028898239, "train_loss_bc": 0.04060850292444229, "train_loss_llm": 0.0016228854656219482, "grad_norm": 0.013631441630423069, "global_step": 4865, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04336300492286682, "train_loss_bc": 0.04336300492286682, "train_loss_llm": 0.0, "grad_norm": 0.03534717112779617, "global_step": 4866, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04429686442017555, "train_loss_bc": 0.04429686442017555, "train_loss_llm": 0.0, "grad_norm": 0.038910746574401855, "global_step": 4867, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04234624281525612, "train_loss_bc": 0.04234624281525612, "train_loss_llm": 0.0, "grad_norm": 0.04257570579648018, "global_step": 4868, "epoch": 135, "lr": 0.001} +{"train_loss": 0.03919323533773422, "train_loss_bc": 0.03919323533773422, "train_loss_llm": 0.0, "grad_norm": 0.01664663851261139, "global_step": 4869, "epoch": 135, "lr": 0.001} +{"train_loss": 0.03852804750204086, "train_loss_bc": 0.03852804750204086, "train_loss_llm": 0.0, "grad_norm": 0.028483258560299873, "global_step": 4870, "epoch": 135, "lr": 0.001} +{"train_loss": 0.042480722069740295, "train_loss_bc": 0.042480722069740295, "train_loss_llm": 0.0, "grad_norm": 0.03773382306098938, "global_step": 4871, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04259935021400452, "train_loss_bc": 0.04259935021400452, "train_loss_llm": 0.0, "grad_norm": 0.047290995717048645, "global_step": 4872, "epoch": 135, "lr": 0.001} +{"train_loss": 0.0393294021487236, "train_loss_bc": 0.0393294021487236, "train_loss_llm": 0.0, "grad_norm": 0.01378833968192339, "global_step": 4873, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04342634230852127, "train_loss_bc": 0.0433993861079216, "train_loss_llm": 0.0026955008506774902, "grad_norm": 0.024500172585248947, "global_step": 4874, "epoch": 135, "lr": 0.001} +{"train_loss": 0.043767038732767105, "train_loss_bc": 0.043767038732767105, "train_loss_llm": 0.0, "grad_norm": 0.028496814891695976, "global_step": 4875, "epoch": 135, "lr": 0.001} +{"train_loss": 0.046708691865205765, "train_loss_bc": 0.046708691865205765, "train_loss_llm": 0.0, "grad_norm": 0.031717389822006226, "global_step": 4876, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04545838385820389, "train_loss_bc": 0.04545838385820389, "train_loss_llm": 0.0, "grad_norm": 0.017746252939105034, "global_step": 4877, "epoch": 135, "lr": 0.001} +{"train_loss": 0.046815741807222366, "train_loss_bc": 0.04680497571825981, "train_loss_llm": 0.00107651948928833, "grad_norm": 0.02746456116437912, "global_step": 4878, "epoch": 135, "lr": 0.001} +{"train_loss": 0.05133388563990593, "train_loss_bc": 0.0513315424323082, "train_loss_llm": 0.00023439526557922363, "grad_norm": 0.036214012652635574, "global_step": 4879, "epoch": 135, "lr": 0.001} +{"train_loss": 0.038237668573856354, "train_loss_bc": 0.0382164902985096, "train_loss_llm": 0.002117842435836792, "grad_norm": 0.03939772769808769, "global_step": 4880, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04522780328989029, "train_loss_bc": 0.04519465193152428, "train_loss_llm": 0.003315180540084839, "grad_norm": 0.020461304113268852, "global_step": 4881, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04000620171427727, "train_loss_bc": 0.04000620171427727, "train_loss_llm": 0.0, "grad_norm": 0.02808961272239685, "global_step": 4882, "epoch": 135, "lr": 0.001} +{"train_loss": 0.0424828827381134, "train_loss_bc": 0.0424828827381134, "train_loss_llm": 0.0, "grad_norm": 0.036124080419540405, "global_step": 4883, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04227638989686966, "train_loss_bc": 0.042252521961927414, "train_loss_llm": 0.0023868680000305176, "grad_norm": 0.04224780201911926, "global_step": 4884, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04600776731967926, "train_loss_bc": 0.04600776731967926, "train_loss_llm": 0.0, "grad_norm": 0.019065923988819122, "global_step": 4885, "epoch": 135, "lr": 0.001} +{"train_loss": 0.038579534739255905, "train_loss_bc": 0.0385519377887249, "train_loss_llm": 0.0027598142623901367, "grad_norm": 0.026648560538887978, "global_step": 4886, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04004921391606331, "train_loss_bc": 0.04004921391606331, "train_loss_llm": 0.0, "grad_norm": 0.031402893364429474, "global_step": 4887, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04730752855539322, "train_loss_bc": 0.04730752855539322, "train_loss_llm": 0.0, "grad_norm": 0.04287858307361603, "global_step": 4888, "epoch": 135, "lr": 0.001} +{"train_loss": 0.040006790310144424, "train_loss_bc": 0.040006790310144424, "train_loss_llm": 0.0, "grad_norm": 0.014525780454277992, "global_step": 4889, "epoch": 135, "lr": 0.001} +{"train_loss": 0.03080061264336109, "train_loss_bc": 0.03080061264336109, "train_loss_llm": 0.0, "grad_norm": 0.02480914071202278, "global_step": 4890, "epoch": 135, "lr": 0.001} +{"train_loss": 0.03931914269924164, "train_loss_bc": 0.039305854588747025, "train_loss_llm": 0.0013287067413330078, "grad_norm": 0.03521513193845749, "global_step": 4891, "epoch": 135, "lr": 0.001} +{"train_loss": 0.038149379193782806, "train_loss_bc": 0.038149379193782806, "train_loss_llm": 0.0, "grad_norm": 0.042760834097862244, "global_step": 4892, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04558156430721283, "train_loss_bc": 0.04558156430721283, "train_loss_llm": 0.0, "grad_norm": 0.013802440837025642, "global_step": 4893, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04557109996676445, "train_loss_bc": 0.04557109996676445, "train_loss_llm": 0.0, "grad_norm": 0.024133628234267235, "global_step": 4894, "epoch": 135, "lr": 0.001} +{"train_loss": 0.04224132668847839, "train_loss_bc": 0.04651479423046112, "train_loss_llm": 0.0, "grad_norm": 0.035704802721738815, "global_step": 4895, "epoch": 135, "lr": 0.001, "train/cumulative_reward": 137.31436962973615, "train/mean_score": 14.836225035794797, "train/success_rate": 0.3, "test/cumulative_reward": 65.06327684210498, "test/mean_score": 10.591181821510629, "test/success_rate": 0.22, "val_loss": 0.05400477349758148, "train_action_mse_error": 0.04653368890285492} +{"train_loss": 0.04066487401723862, "train_loss_bc": 0.04066487401723862, "train_loss_llm": 0.0, "grad_norm": 0.03977113217115402, "global_step": 4896, "epoch": 136, "lr": 0.001} +{"train_loss": 0.03534136712551117, "train_loss_bc": 0.03534136712551117, "train_loss_llm": 0.0, "grad_norm": 0.013753490522503853, "global_step": 4897, "epoch": 136, "lr": 0.001} +{"train_loss": 0.0441322922706604, "train_loss_bc": 0.04410634934902191, "train_loss_llm": 0.0025944411754608154, "grad_norm": 0.028756825253367424, "global_step": 4898, "epoch": 136, "lr": 0.001} +{"train_loss": 0.040725529193878174, "train_loss_bc": 0.04072289541363716, "train_loss_llm": 0.00026345252990722656, "grad_norm": 0.036002017557621, "global_step": 4899, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04408898204565048, "train_loss_bc": 0.04408898204565048, "train_loss_llm": 0.0, "grad_norm": 0.03815367817878723, "global_step": 4900, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04241282492876053, "train_loss_bc": 0.04241282492876053, "train_loss_llm": 0.0, "grad_norm": 0.016797127202153206, "global_step": 4901, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04645983874797821, "train_loss_bc": 0.046455685049295425, "train_loss_llm": 0.0004152655601501465, "grad_norm": 0.026400303468108177, "global_step": 4902, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04607147350907326, "train_loss_bc": 0.046063169836997986, "train_loss_llm": 0.0008304417133331299, "grad_norm": 0.035312335938215256, "global_step": 4903, "epoch": 136, "lr": 0.001} +{"train_loss": 0.03973040357232094, "train_loss_bc": 0.03973040357232094, "train_loss_llm": 0.0, "grad_norm": 0.046948131173849106, "global_step": 4904, "epoch": 136, "lr": 0.001} +{"train_loss": 0.0512012243270874, "train_loss_bc": 0.0512012243270874, "train_loss_llm": 0.0, "grad_norm": 0.018027780577540398, "global_step": 4905, "epoch": 136, "lr": 0.001} +{"train_loss": 0.031865496188402176, "train_loss_bc": 0.031865496188402176, "train_loss_llm": 0.0, "grad_norm": 0.024749403819441795, "global_step": 4906, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04640347510576248, "train_loss_bc": 0.04640062153339386, "train_loss_llm": 0.0002852976322174072, "grad_norm": 0.03572773560881615, "global_step": 4907, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04029548913240433, "train_loss_bc": 0.040292397141456604, "train_loss_llm": 0.00030937790870666504, "grad_norm": 0.04762808606028557, "global_step": 4908, "epoch": 136, "lr": 0.001} +{"train_loss": 0.03385801985859871, "train_loss_bc": 0.03385801985859871, "train_loss_llm": 0.0, "grad_norm": 0.012874472886323929, "global_step": 4909, "epoch": 136, "lr": 0.001} +{"train_loss": 0.040255170315504074, "train_loss_bc": 0.040255170315504074, "train_loss_llm": 0.0, "grad_norm": 0.022537317126989365, "global_step": 4910, "epoch": 136, "lr": 0.001} +{"train_loss": 0.044175922870635986, "train_loss_bc": 0.044175922870635986, "train_loss_llm": 0.0, "grad_norm": 0.02523351088166237, "global_step": 4911, "epoch": 136, "lr": 0.001} +{"train_loss": 0.03439265117049217, "train_loss_bc": 0.03439265117049217, "train_loss_llm": 0.0, "grad_norm": 0.03469971939921379, "global_step": 4912, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04354117065668106, "train_loss_bc": 0.04354117065668106, "train_loss_llm": 0.0, "grad_norm": 0.016143005341291428, "global_step": 4913, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04211458936333656, "train_loss_bc": 0.04210353270173073, "train_loss_llm": 0.0011058151721954346, "grad_norm": 0.029362795874476433, "global_step": 4914, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04198770597577095, "train_loss_bc": 0.04198770597577095, "train_loss_llm": 0.0, "grad_norm": 0.032756660133600235, "global_step": 4915, "epoch": 136, "lr": 0.001} +{"train_loss": 0.045647330582141876, "train_loss_bc": 0.045647330582141876, "train_loss_llm": 0.0, "grad_norm": 0.044805288314819336, "global_step": 4916, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04168251156806946, "train_loss_bc": 0.04168251156806946, "train_loss_llm": 0.0, "grad_norm": 0.020520778372883797, "global_step": 4917, "epoch": 136, "lr": 0.001} +{"train_loss": 0.03511543944478035, "train_loss_bc": 0.03511543944478035, "train_loss_llm": 0.0, "grad_norm": 0.028625955805182457, "global_step": 4918, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04025544226169586, "train_loss_bc": 0.04023800417780876, "train_loss_llm": 0.0017438828945159912, "grad_norm": 0.0345136895775795, "global_step": 4919, "epoch": 136, "lr": 0.001} +{"train_loss": 0.033116623759269714, "train_loss_bc": 0.03309950232505798, "train_loss_llm": 0.0017122924327850342, "grad_norm": 0.03867078572511673, "global_step": 4920, "epoch": 136, "lr": 0.001} +{"train_loss": 0.048049259930849075, "train_loss_bc": 0.048049259930849075, "train_loss_llm": 0.0, "grad_norm": 0.023025471717119217, "global_step": 4921, "epoch": 136, "lr": 0.001} +{"train_loss": 0.03810688480734825, "train_loss_bc": 0.03810688480734825, "train_loss_llm": 0.0, "grad_norm": 0.032706860452890396, "global_step": 4922, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04083079844713211, "train_loss_bc": 0.04083079844713211, "train_loss_llm": 0.0, "grad_norm": 0.042729467153549194, "global_step": 4923, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04639146849513054, "train_loss_bc": 0.04639146849513054, "train_loss_llm": 0.0, "grad_norm": 0.04379984736442566, "global_step": 4924, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04078986868262291, "train_loss_bc": 0.040787480771541595, "train_loss_llm": 0.00023865699768066406, "grad_norm": 0.015399343334138393, "global_step": 4925, "epoch": 136, "lr": 0.001} +{"train_loss": 0.05433791130781174, "train_loss_bc": 0.05433791130781174, "train_loss_llm": 0.0, "grad_norm": 0.025845220312476158, "global_step": 4926, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04507201537489891, "train_loss_bc": 0.04505741968750954, "train_loss_llm": 0.0014595985412597656, "grad_norm": 0.03787579387426376, "global_step": 4927, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04503246396780014, "train_loss_bc": 0.04503246396780014, "train_loss_llm": 0.0, "grad_norm": 0.051320336759090424, "global_step": 4928, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04320959746837616, "train_loss_bc": 0.04320959746837616, "train_loss_llm": 0.0, "grad_norm": 0.017187630757689476, "global_step": 4929, "epoch": 136, "lr": 0.001} +{"train_loss": 0.044726401567459106, "train_loss_bc": 0.044726401567459106, "train_loss_llm": 0.0, "grad_norm": 0.028593599796295166, "global_step": 4930, "epoch": 136, "lr": 0.001} +{"train_loss": 0.04227275639358494, "train_loss_bc": 0.049736712127923965, "train_loss_llm": 0.0, "grad_norm": 0.04935965687036514, "global_step": 4931, "epoch": 136, "lr": 0.001, "val_loss": 0.05454900115728378} +{"train_loss": 0.04588910937309265, "train_loss_bc": 0.045882657170295715, "train_loss_llm": 0.0006453394889831543, "grad_norm": 0.05518345162272453, "global_step": 4932, "epoch": 137, "lr": 0.001} +{"train_loss": 0.045041002333164215, "train_loss_bc": 0.045041002333164215, "train_loss_llm": 0.0, "grad_norm": 0.016464348882436752, "global_step": 4933, "epoch": 137, "lr": 0.001} +{"train_loss": 0.037056971341371536, "train_loss_bc": 0.0370476171374321, "train_loss_llm": 0.0009352564811706543, "grad_norm": 0.02168256975710392, "global_step": 4934, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04192308709025383, "train_loss_bc": 0.04192308709025383, "train_loss_llm": 0.0, "grad_norm": 0.03300921246409416, "global_step": 4935, "epoch": 137, "lr": 0.001} +{"train_loss": 0.03716480731964111, "train_loss_bc": 0.03715475648641586, "train_loss_llm": 0.0010052025318145752, "grad_norm": 0.038528624922037125, "global_step": 4936, "epoch": 137, "lr": 0.001} +{"train_loss": 0.02928759530186653, "train_loss_bc": 0.02928759530186653, "train_loss_llm": 0.0, "grad_norm": 0.013171552680432796, "global_step": 4937, "epoch": 137, "lr": 0.001} +{"train_loss": 0.043922312557697296, "train_loss_bc": 0.043922312557697296, "train_loss_llm": 0.0, "grad_norm": 0.023510169237852097, "global_step": 4938, "epoch": 137, "lr": 0.001} +{"train_loss": 0.044085998088121414, "train_loss_bc": 0.044081661850214005, "train_loss_llm": 0.0004337728023529053, "grad_norm": 0.02983706444501877, "global_step": 4939, "epoch": 137, "lr": 0.001} +{"train_loss": 0.040586743503808975, "train_loss_bc": 0.04056263342499733, "train_loss_llm": 0.0024109482765197754, "grad_norm": 0.037649258971214294, "global_step": 4940, "epoch": 137, "lr": 0.001} +{"train_loss": 0.042421989142894745, "train_loss_bc": 0.04240958020091057, "train_loss_llm": 0.0012409389019012451, "grad_norm": 0.016905061900615692, "global_step": 4941, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04001985490322113, "train_loss_bc": 0.04001985490322113, "train_loss_llm": 0.0, "grad_norm": 0.02547677420079708, "global_step": 4942, "epoch": 137, "lr": 0.001} +{"train_loss": 0.042233970016241074, "train_loss_bc": 0.042233970016241074, "train_loss_llm": 0.0, "grad_norm": 0.031916964799165726, "global_step": 4943, "epoch": 137, "lr": 0.001} +{"train_loss": 0.03823889419436455, "train_loss_bc": 0.03823889419436455, "train_loss_llm": 0.0, "grad_norm": 0.035266630351543427, "global_step": 4944, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04559759423136711, "train_loss_bc": 0.04559759423136711, "train_loss_llm": 0.0, "grad_norm": 0.017609911039471626, "global_step": 4945, "epoch": 137, "lr": 0.001} +{"train_loss": 0.03979840502142906, "train_loss_bc": 0.03979840502142906, "train_loss_llm": 0.0, "grad_norm": 0.028476430103182793, "global_step": 4946, "epoch": 137, "lr": 0.001} +{"train_loss": 0.049197956919670105, "train_loss_bc": 0.049197956919670105, "train_loss_llm": 0.0, "grad_norm": 0.0433877594769001, "global_step": 4947, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04176733270287514, "train_loss_bc": 0.04176733270287514, "train_loss_llm": 0.0, "grad_norm": 0.04394330084323883, "global_step": 4948, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04479433223605156, "train_loss_bc": 0.04479433223605156, "train_loss_llm": 0.0, "grad_norm": 0.015344089828431606, "global_step": 4949, "epoch": 137, "lr": 0.001} +{"train_loss": 0.05090198665857315, "train_loss_bc": 0.05090198665857315, "train_loss_llm": 0.0, "grad_norm": 0.027490749955177307, "global_step": 4950, "epoch": 137, "lr": 0.001} +{"train_loss": 0.040055837482213974, "train_loss_bc": 0.040055837482213974, "train_loss_llm": 0.0, "grad_norm": 0.04063643142580986, "global_step": 4951, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04164246469736099, "train_loss_bc": 0.04164246469736099, "train_loss_llm": 0.0, "grad_norm": 0.047010838985443115, "global_step": 4952, "epoch": 137, "lr": 0.001} +{"train_loss": 0.041585732251405716, "train_loss_bc": 0.041585732251405716, "train_loss_llm": 0.0, "grad_norm": 0.015186117962002754, "global_step": 4953, "epoch": 137, "lr": 0.001} +{"train_loss": 0.041459228843450546, "train_loss_bc": 0.041459228843450546, "train_loss_llm": 0.0, "grad_norm": 0.02293974719941616, "global_step": 4954, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04496980831027031, "train_loss_bc": 0.04495422914624214, "train_loss_llm": 0.001557856798171997, "grad_norm": 0.03366188704967499, "global_step": 4955, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04262161627411842, "train_loss_bc": 0.04262161627411842, "train_loss_llm": 0.0, "grad_norm": 0.04095352068543434, "global_step": 4956, "epoch": 137, "lr": 0.001} +{"train_loss": 0.034921761602163315, "train_loss_bc": 0.034919373691082, "train_loss_llm": 0.00023883581161499023, "grad_norm": 0.015272674150764942, "global_step": 4957, "epoch": 137, "lr": 0.001} +{"train_loss": 0.03258374705910683, "train_loss_bc": 0.03258374705910683, "train_loss_llm": 0.0, "grad_norm": 0.022881440818309784, "global_step": 4958, "epoch": 137, "lr": 0.001} +{"train_loss": 0.044514305889606476, "train_loss_bc": 0.044514305889606476, "train_loss_llm": 0.0, "grad_norm": 0.032254401594400406, "global_step": 4959, "epoch": 137, "lr": 0.001} +{"train_loss": 0.038650646805763245, "train_loss_bc": 0.03864804655313492, "train_loss_llm": 0.0002599358558654785, "grad_norm": 0.03682166337966919, "global_step": 4960, "epoch": 137, "lr": 0.001} +{"train_loss": 0.040032994002103806, "train_loss_bc": 0.040032994002103806, "train_loss_llm": 0.0, "grad_norm": 0.015542657114565372, "global_step": 4961, "epoch": 137, "lr": 0.001} +{"train_loss": 0.053175508975982666, "train_loss_bc": 0.05317520350217819, "train_loss_llm": 3.0517578125e-05, "grad_norm": 0.0222475603222847, "global_step": 4962, "epoch": 137, "lr": 0.001} +{"train_loss": 0.03730912134051323, "train_loss_bc": 0.03730912134051323, "train_loss_llm": 0.0, "grad_norm": 0.02805158868432045, "global_step": 4963, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04427376762032509, "train_loss_bc": 0.0442635752260685, "train_loss_llm": 0.001019209623336792, "grad_norm": 0.033477216958999634, "global_step": 4964, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04956218600273132, "train_loss_bc": 0.04956218600273132, "train_loss_llm": 0.0, "grad_norm": 0.02117052674293518, "global_step": 4965, "epoch": 137, "lr": 0.001} +{"train_loss": 0.03816816955804825, "train_loss_bc": 0.03816816955804825, "train_loss_llm": 0.0, "grad_norm": 0.028307896107435226, "global_step": 4966, "epoch": 137, "lr": 0.001} +{"train_loss": 0.04223466985341576, "train_loss_bc": 0.054948482662439346, "train_loss_llm": 0.004279226064682007, "grad_norm": 0.04349646344780922, "global_step": 4967, "epoch": 137, "lr": 0.001, "val_loss": 0.05483902990818024} +{"train_loss": 0.04293210059404373, "train_loss_bc": 0.04293210059404373, "train_loss_llm": 0.0, "grad_norm": 0.052184514701366425, "global_step": 4968, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04468343034386635, "train_loss_bc": 0.04468343034386635, "train_loss_llm": 0.0, "grad_norm": 0.018858101218938828, "global_step": 4969, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04370861127972603, "train_loss_bc": 0.04370861127972603, "train_loss_llm": 0.0, "grad_norm": 0.029765307903289795, "global_step": 4970, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04651765152812004, "train_loss_bc": 0.04651765152812004, "train_loss_llm": 0.0, "grad_norm": 0.04323326796293259, "global_step": 4971, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04236382246017456, "train_loss_bc": 0.04236382246017456, "train_loss_llm": 0.0, "grad_norm": 0.0563509576022625, "global_step": 4972, "epoch": 138, "lr": 0.001} +{"train_loss": 0.035951271653175354, "train_loss_bc": 0.03594733774662018, "train_loss_llm": 0.0003934800624847412, "grad_norm": 0.017491202801465988, "global_step": 4973, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04624708741903305, "train_loss_bc": 0.04624708741903305, "train_loss_llm": 0.0, "grad_norm": 0.029811304062604904, "global_step": 4974, "epoch": 138, "lr": 0.001} +{"train_loss": 0.042397741228342056, "train_loss_bc": 0.042397741228342056, "train_loss_llm": 0.0, "grad_norm": 0.03988996893167496, "global_step": 4975, "epoch": 138, "lr": 0.001} +{"train_loss": 0.03487768769264221, "train_loss_bc": 0.03487768769264221, "train_loss_llm": 0.0, "grad_norm": 0.04411992430686951, "global_step": 4976, "epoch": 138, "lr": 0.001} +{"train_loss": 0.03738722205162048, "train_loss_bc": 0.03737730532884598, "train_loss_llm": 0.0009917020797729492, "grad_norm": 0.012571718543767929, "global_step": 4977, "epoch": 138, "lr": 0.001} +{"train_loss": 0.0440717488527298, "train_loss_bc": 0.0440717488527298, "train_loss_llm": 0.0, "grad_norm": 0.022613465785980225, "global_step": 4978, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04350975900888443, "train_loss_bc": 0.04350975900888443, "train_loss_llm": 0.0, "grad_norm": 0.023222675547003746, "global_step": 4979, "epoch": 138, "lr": 0.001} +{"train_loss": 0.03788549825549126, "train_loss_bc": 0.03788549825549126, "train_loss_llm": 0.0, "grad_norm": 0.027102963998913765, "global_step": 4980, "epoch": 138, "lr": 0.001} +{"train_loss": 0.040361981838941574, "train_loss_bc": 0.040361981838941574, "train_loss_llm": 0.0, "grad_norm": 0.013902995735406876, "global_step": 4981, "epoch": 138, "lr": 0.001} +{"train_loss": 0.05359094589948654, "train_loss_bc": 0.05359094589948654, "train_loss_llm": 0.0, "grad_norm": 0.023455511778593063, "global_step": 4982, "epoch": 138, "lr": 0.001} +{"train_loss": 0.03594285622239113, "train_loss_bc": 0.03594285622239113, "train_loss_llm": 0.0, "grad_norm": 0.028235744684934616, "global_step": 4983, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04573351889848709, "train_loss_bc": 0.04573351889848709, "train_loss_llm": 0.0, "grad_norm": 0.03001556172966957, "global_step": 4984, "epoch": 138, "lr": 0.001} +{"train_loss": 0.044372543692588806, "train_loss_bc": 0.04435189813375473, "train_loss_llm": 0.002064615488052368, "grad_norm": 0.016467861831188202, "global_step": 4985, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04376814514398575, "train_loss_bc": 0.04376814514398575, "train_loss_llm": 0.0, "grad_norm": 0.02825237438082695, "global_step": 4986, "epoch": 138, "lr": 0.001} +{"train_loss": 0.041595060378313065, "train_loss_bc": 0.04157343879342079, "train_loss_llm": 0.002162039279937744, "grad_norm": 0.039215635508298874, "global_step": 4987, "epoch": 138, "lr": 0.001} +{"train_loss": 0.03662508726119995, "train_loss_bc": 0.03661758452653885, "train_loss_llm": 0.0007500946521759033, "grad_norm": 0.05213243514299393, "global_step": 4988, "epoch": 138, "lr": 0.001} +{"train_loss": 0.03832806274294853, "train_loss_bc": 0.03831220790743828, "train_loss_llm": 0.0015856027603149414, "grad_norm": 0.015848983079195023, "global_step": 4989, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04947742074728012, "train_loss_bc": 0.04947742074728012, "train_loss_llm": 0.0, "grad_norm": 0.0289341788738966, "global_step": 4990, "epoch": 138, "lr": 0.001} +{"train_loss": 0.036131564527750015, "train_loss_bc": 0.036097366362810135, "train_loss_llm": 0.0034198760986328125, "grad_norm": 0.03551112860441208, "global_step": 4991, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04321250319480896, "train_loss_bc": 0.04321250319480896, "train_loss_llm": 0.0, "grad_norm": 0.04088679328560829, "global_step": 4992, "epoch": 138, "lr": 0.001} +{"train_loss": 0.046180397272109985, "train_loss_bc": 0.046137381345033646, "train_loss_llm": 0.0043014585971832275, "grad_norm": 0.020650450140237808, "global_step": 4993, "epoch": 138, "lr": 0.001} +{"train_loss": 0.03916363790631294, "train_loss_bc": 0.039148252457380295, "train_loss_llm": 0.0015383660793304443, "grad_norm": 0.026007741689682007, "global_step": 4994, "epoch": 138, "lr": 0.001} +{"train_loss": 0.047048769891262054, "train_loss_bc": 0.04704342782497406, "train_loss_llm": 0.0005340576171875, "grad_norm": 0.031774166971445084, "global_step": 4995, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04002777487039566, "train_loss_bc": 0.04002641886472702, "train_loss_llm": 0.00013571977615356445, "grad_norm": 0.04315311461687088, "global_step": 4996, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04217444732785225, "train_loss_bc": 0.042170844972133636, "train_loss_llm": 0.00036028027534484863, "grad_norm": 0.017745180055499077, "global_step": 4997, "epoch": 138, "lr": 0.001} +{"train_loss": 0.047011591494083405, "train_loss_bc": 0.04699802026152611, "train_loss_llm": 0.0013570785522460938, "grad_norm": 0.028842510655522346, "global_step": 4998, "epoch": 138, "lr": 0.001} +{"train_loss": 0.03881092369556427, "train_loss_bc": 0.03881092369556427, "train_loss_llm": 0.0, "grad_norm": 0.03653961420059204, "global_step": 4999, "epoch": 138, "lr": 0.001} +{"train_loss": 0.039094582200050354, "train_loss_bc": 0.039094582200050354, "train_loss_llm": 0.0, "grad_norm": 0.042517244815826416, "global_step": 5000, "epoch": 138, "lr": 0.001} +{"train_loss": 0.043902937322854996, "train_loss_bc": 0.043902937322854996, "train_loss_llm": 0.0, "grad_norm": 0.020293748006224632, "global_step": 5001, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04281899333000183, "train_loss_bc": 0.04281899333000183, "train_loss_llm": 0.0, "grad_norm": 0.026644300669431686, "global_step": 5002, "epoch": 138, "lr": 0.001} +{"train_loss": 0.04205466392967436, "train_loss_bc": 0.036048319190740585, "train_loss_llm": 0.0012205839157104492, "grad_norm": 0.0379447415471077, "global_step": 5003, "epoch": 138, "lr": 0.001, "val_loss": 0.05428987741470337} +{"train_loss": 0.049384504556655884, "train_loss_bc": 0.049384504556655884, "train_loss_llm": 0.0, "grad_norm": 0.04169343039393425, "global_step": 5004, "epoch": 139, "lr": 0.001} +{"train_loss": 0.03911938518285751, "train_loss_bc": 0.03911938518285751, "train_loss_llm": 0.0, "grad_norm": 0.015016019344329834, "global_step": 5005, "epoch": 139, "lr": 0.001} +{"train_loss": 0.04423169791698456, "train_loss_bc": 0.04423169791698456, "train_loss_llm": 0.0, "grad_norm": 0.022435052320361137, "global_step": 5006, "epoch": 139, "lr": 0.001} +{"train_loss": 0.04368869587779045, "train_loss_bc": 0.04368869587779045, "train_loss_llm": 0.0, "grad_norm": 0.02943236008286476, "global_step": 5007, "epoch": 139, "lr": 0.001} +{"train_loss": 0.045054078102111816, "train_loss_bc": 0.045054078102111816, "train_loss_llm": 0.0, "grad_norm": 0.03993551805615425, "global_step": 5008, "epoch": 139, "lr": 0.001} +{"train_loss": 0.04109390825033188, "train_loss_bc": 0.04109390825033188, "train_loss_llm": 0.0, "grad_norm": 0.016175268217921257, "global_step": 5009, "epoch": 139, "lr": 0.001} +{"train_loss": 0.03575582802295685, "train_loss_bc": 0.03573788329958916, "train_loss_llm": 0.0017945468425750732, "grad_norm": 0.022705188021063805, "global_step": 5010, "epoch": 139, "lr": 0.001} +{"train_loss": 0.041396114975214005, "train_loss_bc": 0.04139343276619911, "train_loss_llm": 0.0002681314945220947, "grad_norm": 0.03221619501709938, "global_step": 5011, "epoch": 139, "lr": 0.001} +{"train_loss": 0.046504683792591095, "train_loss_bc": 0.046504683792591095, "train_loss_llm": 0.0, "grad_norm": 0.04165255278348923, "global_step": 5012, "epoch": 139, "lr": 0.001} +{"train_loss": 0.04142862558364868, "train_loss_bc": 0.04142862558364868, "train_loss_llm": 0.0, "grad_norm": 0.01546111423522234, "global_step": 5013, "epoch": 139, "lr": 0.001} +{"train_loss": 0.046048857271671295, "train_loss_bc": 0.046048857271671295, "train_loss_llm": 0.0, "grad_norm": 0.02011697366833687, "global_step": 5014, "epoch": 139, "lr": 0.001} +{"train_loss": 0.03505324572324753, "train_loss_bc": 0.03505324572324753, "train_loss_llm": 0.0, "grad_norm": 0.028971677646040916, "global_step": 5015, "epoch": 139, "lr": 0.001} +{"train_loss": 0.0378088653087616, "train_loss_bc": 0.0378088653087616, "train_loss_llm": 0.0, "grad_norm": 0.03675451874732971, "global_step": 5016, "epoch": 139, "lr": 0.001} +{"train_loss": 0.04396775737404823, "train_loss_bc": 0.04396775737404823, "train_loss_llm": 0.0, "grad_norm": 0.01717899926006794, "global_step": 5017, "epoch": 139, "lr": 0.001} +{"train_loss": 0.048097580671310425, "train_loss_bc": 0.048097580671310425, "train_loss_llm": 0.0, "grad_norm": 0.031299252063035965, "global_step": 5018, "epoch": 139, "lr": 0.001} +{"train_loss": 0.03797852620482445, "train_loss_bc": 0.03795618563890457, "train_loss_llm": 0.0022340118885040283, "grad_norm": 0.0399184376001358, "global_step": 5019, "epoch": 139, "lr": 0.001} +{"train_loss": 0.041442714631557465, "train_loss_bc": 0.041442714631557465, "train_loss_llm": 0.0, "grad_norm": 0.04665206000208855, "global_step": 5020, "epoch": 139, "lr": 0.001} +{"train_loss": 0.036041684448719025, "train_loss_bc": 0.036041684448719025, "train_loss_llm": 0.0, "grad_norm": 0.018490120768547058, "global_step": 5021, "epoch": 139, "lr": 0.001} +{"train_loss": 0.047924723476171494, "train_loss_bc": 0.047924723476171494, "train_loss_llm": 0.0, "grad_norm": 0.028477156534790993, "global_step": 5022, "epoch": 139, "lr": 0.001} +{"train_loss": 0.03952155262231827, "train_loss_bc": 0.03952155262231827, "train_loss_llm": 0.0, "grad_norm": 0.037590932101011276, "global_step": 5023, "epoch": 139, "lr": 0.001} +{"train_loss": 0.045943595468997955, "train_loss_bc": 0.045940086245536804, "train_loss_llm": 0.0003509521484375, "grad_norm": 0.04842688515782356, "global_step": 5024, "epoch": 139, "lr": 0.001} +{"train_loss": 0.03259100019931793, "train_loss_bc": 0.03259100019931793, "train_loss_llm": 0.0, "grad_norm": 0.015553917735815048, "global_step": 5025, "epoch": 139, "lr": 0.001} +{"train_loss": 0.04615086317062378, "train_loss_bc": 0.04615086317062378, "train_loss_llm": 0.0, "grad_norm": 0.022725163027644157, "global_step": 5026, "epoch": 139, "lr": 0.001} +{"train_loss": 0.045482754707336426, "train_loss_bc": 0.0454791821539402, "train_loss_llm": 0.0003573894500732422, "grad_norm": 0.02938220463693142, "global_step": 5027, "epoch": 139, "lr": 0.001} +{"train_loss": 0.03651294857263565, "train_loss_bc": 0.03651294857263565, "train_loss_llm": 0.0, "grad_norm": 0.0338955782353878, "global_step": 5028, "epoch": 139, "lr": 0.001} +{"train_loss": 0.05013100430369377, "train_loss_bc": 0.05013100430369377, "train_loss_llm": 0.0, "grad_norm": 0.01829075627028942, "global_step": 5029, "epoch": 139, "lr": 0.001} +{"train_loss": 0.03773155435919762, "train_loss_bc": 0.037718161940574646, "train_loss_llm": 0.0013391375541687012, "grad_norm": 0.02702573500573635, "global_step": 5030, "epoch": 139, "lr": 0.001} +{"train_loss": 0.043931253254413605, "train_loss_bc": 0.043931253254413605, "train_loss_llm": 0.0, "grad_norm": 0.041224341839551926, "global_step": 5031, "epoch": 139, "lr": 0.001} +{"train_loss": 0.041085220873355865, "train_loss_bc": 0.041085220873355865, "train_loss_llm": 0.0, "grad_norm": 0.04599495604634285, "global_step": 5032, "epoch": 139, "lr": 0.001} +{"train_loss": 0.04099392890930176, "train_loss_bc": 0.04098808765411377, "train_loss_llm": 0.0005840659141540527, "grad_norm": 0.023078851401805878, "global_step": 5033, "epoch": 139, "lr": 0.001} +{"train_loss": 0.04413183405995369, "train_loss_bc": 0.04412314295768738, "train_loss_llm": 0.0008691251277923584, "grad_norm": 0.034428488463163376, "global_step": 5034, "epoch": 139, "lr": 0.001} +{"train_loss": 0.04355565086007118, "train_loss_bc": 0.04355565086007118, "train_loss_llm": 0.0, "grad_norm": 0.041254013776779175, "global_step": 5035, "epoch": 139, "lr": 0.001} +{"train_loss": 0.03412732481956482, "train_loss_bc": 0.03412732481956482, "train_loss_llm": 0.0, "grad_norm": 0.04548873379826546, "global_step": 5036, "epoch": 139, "lr": 0.001} +{"train_loss": 0.03627368062734604, "train_loss_bc": 0.03627368062734604, "train_loss_llm": 0.0, "grad_norm": 0.017992569133639336, "global_step": 5037, "epoch": 139, "lr": 0.001} +{"train_loss": 0.04634176194667816, "train_loss_bc": 0.04633919522166252, "train_loss_llm": 0.00025659799575805664, "grad_norm": 0.031091488897800446, "global_step": 5038, "epoch": 139, "lr": 0.001} +{"train_loss": 0.042153415684070855, "train_loss_bc": 0.05097464844584465, "train_loss_llm": 0.0020909905433654785, "grad_norm": 0.04215525463223457, "global_step": 5039, "epoch": 139, "lr": 0.001, "val_loss": 0.054245807230472565} +{"train_loss": 0.04671018570661545, "train_loss_bc": 0.04667465388774872, "train_loss_llm": 0.0035532116889953613, "grad_norm": 0.05385627597570419, "global_step": 5040, "epoch": 140, "lr": 0.001} +{"train_loss": 0.03280404210090637, "train_loss_bc": 0.03280404210090637, "train_loss_llm": 0.0, "grad_norm": 0.013993735425174236, "global_step": 5041, "epoch": 140, "lr": 0.001} +{"train_loss": 0.04746600612998009, "train_loss_bc": 0.047461770474910736, "train_loss_llm": 0.0004235208034515381, "grad_norm": 0.021148787811398506, "global_step": 5042, "epoch": 140, "lr": 0.001} +{"train_loss": 0.049782827496528625, "train_loss_bc": 0.049782827496528625, "train_loss_llm": 0.0, "grad_norm": 0.023544955998659134, "global_step": 5043, "epoch": 140, "lr": 0.001} +{"train_loss": 0.03605090081691742, "train_loss_bc": 0.036012571305036545, "train_loss_llm": 0.0038328468799591064, "grad_norm": 0.0335734598338604, "global_step": 5044, "epoch": 140, "lr": 0.001} +{"train_loss": 0.03621259331703186, "train_loss_bc": 0.03621259331703186, "train_loss_llm": 0.0, "grad_norm": 0.01772419363260269, "global_step": 5045, "epoch": 140, "lr": 0.001} +{"train_loss": 0.03544861450791359, "train_loss_bc": 0.035410311073064804, "train_loss_llm": 0.003830432891845703, "grad_norm": 0.02798025868833065, "global_step": 5046, "epoch": 140, "lr": 0.001} +{"train_loss": 0.045450106263160706, "train_loss_bc": 0.045450106263160706, "train_loss_llm": 0.0, "grad_norm": 0.03847837075591087, "global_step": 5047, "epoch": 140, "lr": 0.001} +{"train_loss": 0.04543842747807503, "train_loss_bc": 0.04543842747807503, "train_loss_llm": 0.0, "grad_norm": 0.054775383323431015, "global_step": 5048, "epoch": 140, "lr": 0.001} +{"train_loss": 0.04420103505253792, "train_loss_bc": 0.04420103505253792, "train_loss_llm": 0.0, "grad_norm": 0.01827862113714218, "global_step": 5049, "epoch": 140, "lr": 0.001} +{"train_loss": 0.0447046160697937, "train_loss_bc": 0.04470377042889595, "train_loss_llm": 8.448958396911621e-05, "grad_norm": 0.029161248356103897, "global_step": 5050, "epoch": 140, "lr": 0.001} +{"train_loss": 0.04605667665600777, "train_loss_bc": 0.04605667665600777, "train_loss_llm": 0.0, "grad_norm": 0.03682180866599083, "global_step": 5051, "epoch": 140, "lr": 0.001} +{"train_loss": 0.037852730602025986, "train_loss_bc": 0.0378386527299881, "train_loss_llm": 0.0014079511165618896, "grad_norm": 0.041551049798727036, "global_step": 5052, "epoch": 140, "lr": 0.001} +{"train_loss": 0.04532685875892639, "train_loss_bc": 0.04529789462685585, "train_loss_llm": 0.0028964877128601074, "grad_norm": 0.018723908811807632, "global_step": 5053, "epoch": 140, "lr": 0.001} +{"train_loss": 0.04581848159432411, "train_loss_bc": 0.04581848159432411, "train_loss_llm": 0.0, "grad_norm": 0.03405614569783211, "global_step": 5054, "epoch": 140, "lr": 0.001} +{"train_loss": 0.04215502738952637, "train_loss_bc": 0.04215502738952637, "train_loss_llm": 0.0, "grad_norm": 0.042549505829811096, "global_step": 5055, "epoch": 140, "lr": 0.001} +{"train_loss": 0.03947058320045471, "train_loss_bc": 0.0394645631313324, "train_loss_llm": 0.0006019175052642822, "grad_norm": 0.05893109366297722, "global_step": 5056, "epoch": 140, "lr": 0.001} +{"train_loss": 0.038011882454156876, "train_loss_bc": 0.038011882454156876, "train_loss_llm": 0.0, "grad_norm": 0.016964640468358994, "global_step": 5057, "epoch": 140, "lr": 0.001} +{"train_loss": 0.03958529978990555, "train_loss_bc": 0.03958529978990555, "train_loss_llm": 0.0, "grad_norm": 0.02224559150636196, "global_step": 5058, "epoch": 140, "lr": 0.001} +{"train_loss": 0.039471518248319626, "train_loss_bc": 0.039471518248319626, "train_loss_llm": 0.0, "grad_norm": 0.027144795283675194, "global_step": 5059, "epoch": 140, "lr": 0.001} +{"train_loss": 0.039781175553798676, "train_loss_bc": 0.039781175553798676, "train_loss_llm": 0.0, "grad_norm": 0.0355711355805397, "global_step": 5060, "epoch": 140, "lr": 0.001} +{"train_loss": 0.03769562393426895, "train_loss_bc": 0.03769562393426895, "train_loss_llm": 0.0, "grad_norm": 0.01830134727060795, "global_step": 5061, "epoch": 140, "lr": 0.001} +{"train_loss": 0.0429837629199028, "train_loss_bc": 0.0429837629199028, "train_loss_llm": 0.0, "grad_norm": 0.029223961755633354, "global_step": 5062, "epoch": 140, "lr": 0.001} +{"train_loss": 0.051720138639211655, "train_loss_bc": 0.051720138639211655, "train_loss_llm": 0.0, "grad_norm": 0.04271742329001427, "global_step": 5063, "epoch": 140, "lr": 0.001} +{"train_loss": 0.04840870201587677, "train_loss_bc": 0.04840870201587677, "train_loss_llm": 0.0, "grad_norm": 0.05308151617646217, "global_step": 5064, "epoch": 140, "lr": 0.001} +{"train_loss": 0.052862606942653656, "train_loss_bc": 0.052862606942653656, "train_loss_llm": 0.0, "grad_norm": 0.02223953790962696, "global_step": 5065, "epoch": 140, "lr": 0.001} +{"train_loss": 0.046279437839984894, "train_loss_bc": 0.046266525983810425, "train_loss_llm": 0.0012912750244140625, "grad_norm": 0.038928378373384476, "global_step": 5066, "epoch": 140, "lr": 0.001} +{"train_loss": 0.0360860750079155, "train_loss_bc": 0.0360860750079155, "train_loss_llm": 0.0, "grad_norm": 0.04158827289938927, "global_step": 5067, "epoch": 140, "lr": 0.001} +{"train_loss": 0.038727082312107086, "train_loss_bc": 0.03872251883149147, "train_loss_llm": 0.00045618414878845215, "grad_norm": 0.047519609332084656, "global_step": 5068, "epoch": 140, "lr": 0.001} +{"train_loss": 0.04157060757279396, "train_loss_bc": 0.04157060757279396, "train_loss_llm": 0.0, "grad_norm": 0.022681426256895065, "global_step": 5069, "epoch": 140, "lr": 0.001} +{"train_loss": 0.033615291118621826, "train_loss_bc": 0.033604696393013, "train_loss_llm": 0.0010595917701721191, "grad_norm": 0.03318061679601669, "global_step": 5070, "epoch": 140, "lr": 0.001} +{"train_loss": 0.035545822232961655, "train_loss_bc": 0.035545822232961655, "train_loss_llm": 0.0, "grad_norm": 0.04203350469470024, "global_step": 5071, "epoch": 140, "lr": 0.001} +{"train_loss": 0.05339561030268669, "train_loss_bc": 0.05339561030268669, "train_loss_llm": 0.0, "grad_norm": 0.05072483792901039, "global_step": 5072, "epoch": 140, "lr": 0.001} +{"train_loss": 0.042547088116407394, "train_loss_bc": 0.04253462329506874, "train_loss_llm": 0.0012466609477996826, "grad_norm": 0.015666205435991287, "global_step": 5073, "epoch": 140, "lr": 0.001} +{"train_loss": 0.039018429815769196, "train_loss_bc": 0.03901459276676178, "train_loss_llm": 0.00038370490074157715, "grad_norm": 0.02456197328865528, "global_step": 5074, "epoch": 140, "lr": 0.001} +{"train_loss": 0.041978264744910926, "train_loss_bc": 0.032961662858724594, "train_loss_llm": 0.0, "grad_norm": 0.03184359893202782, "global_step": 5075, "epoch": 140, "lr": 0.001, "train/cumulative_reward": 60.79109288487825, "train/mean_score": 14.798615763005113, "train/success_rate": 0.3, "test/cumulative_reward": 63.59596096618287, "test/mean_score": 6.09177279369078, "test/success_rate": 0.1, "val_loss": 0.05405490845441818, "train_action_mse_error": 0.045724838972091675} +{"train_loss": 0.04132295772433281, "train_loss_bc": 0.04132295772433281, "train_loss_llm": 0.0, "grad_norm": 0.041008222848176956, "global_step": 5076, "epoch": 141, "lr": 0.001} +{"train_loss": 0.05014340206980705, "train_loss_bc": 0.05014340206980705, "train_loss_llm": 0.0, "grad_norm": 0.020840777084231377, "global_step": 5077, "epoch": 141, "lr": 0.001} +{"train_loss": 0.043638527393341064, "train_loss_bc": 0.043638527393341064, "train_loss_llm": 0.0, "grad_norm": 0.03255776688456535, "global_step": 5078, "epoch": 141, "lr": 0.001} +{"train_loss": 0.05026368796825409, "train_loss_bc": 0.05026368796825409, "train_loss_llm": 0.0, "grad_norm": 0.032264482229948044, "global_step": 5079, "epoch": 141, "lr": 0.001} +{"train_loss": 0.0440424308180809, "train_loss_bc": 0.0440424308180809, "train_loss_llm": 0.0, "grad_norm": 0.034423597157001495, "global_step": 5080, "epoch": 141, "lr": 0.001} +{"train_loss": 0.039819106459617615, "train_loss_bc": 0.039819106459617615, "train_loss_llm": 0.0, "grad_norm": 0.015216914005577564, "global_step": 5081, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04035937041044235, "train_loss_bc": 0.04035937041044235, "train_loss_llm": 0.0, "grad_norm": 0.02683245576918125, "global_step": 5082, "epoch": 141, "lr": 0.001} +{"train_loss": 0.037712257355451584, "train_loss_bc": 0.037712257355451584, "train_loss_llm": 0.0, "grad_norm": 0.037015654146671295, "global_step": 5083, "epoch": 141, "lr": 0.001} +{"train_loss": 0.036870572715997696, "train_loss_bc": 0.036870572715997696, "train_loss_llm": 0.0, "grad_norm": 0.04654422774910927, "global_step": 5084, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04843008518218994, "train_loss_bc": 0.04843008518218994, "train_loss_llm": 0.0, "grad_norm": 0.01935688406229019, "global_step": 5085, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04213887080550194, "train_loss_bc": 0.04212414100766182, "train_loss_llm": 0.0014729499816894531, "grad_norm": 0.030321350321173668, "global_step": 5086, "epoch": 141, "lr": 0.001} +{"train_loss": 0.0409037321805954, "train_loss_bc": 0.0409037321805954, "train_loss_llm": 0.0, "grad_norm": 0.03964317962527275, "global_step": 5087, "epoch": 141, "lr": 0.001} +{"train_loss": 0.03674694895744324, "train_loss_bc": 0.03674478828907013, "train_loss_llm": 0.0002160966396331787, "grad_norm": 0.052896156907081604, "global_step": 5088, "epoch": 141, "lr": 0.001} +{"train_loss": 0.03861866891384125, "train_loss_bc": 0.03861038014292717, "train_loss_llm": 0.0008289515972137451, "grad_norm": 0.017540212720632553, "global_step": 5089, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04185426980257034, "train_loss_bc": 0.04184654727578163, "train_loss_llm": 0.000772327184677124, "grad_norm": 0.025283019989728928, "global_step": 5090, "epoch": 141, "lr": 0.001} +{"train_loss": 0.03711817413568497, "train_loss_bc": 0.03711817413568497, "train_loss_llm": 0.0, "grad_norm": 0.032331135123968124, "global_step": 5091, "epoch": 141, "lr": 0.001} +{"train_loss": 0.032290056347846985, "train_loss_bc": 0.032290056347846985, "train_loss_llm": 0.0, "grad_norm": 0.03640775382518768, "global_step": 5092, "epoch": 141, "lr": 0.001} +{"train_loss": 0.045436110347509384, "train_loss_bc": 0.045436110347509384, "train_loss_llm": 0.0, "grad_norm": 0.017257947474718094, "global_step": 5093, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04774457961320877, "train_loss_bc": 0.04772435128688812, "train_loss_llm": 0.0020227134227752686, "grad_norm": 0.021452724933624268, "global_step": 5094, "epoch": 141, "lr": 0.001} +{"train_loss": 0.03998468816280365, "train_loss_bc": 0.03998468816280365, "train_loss_llm": 0.0, "grad_norm": 0.034633148461580276, "global_step": 5095, "epoch": 141, "lr": 0.001} +{"train_loss": 0.030323069542646408, "train_loss_bc": 0.030323069542646408, "train_loss_llm": 0.0, "grad_norm": 0.03986506536602974, "global_step": 5096, "epoch": 141, "lr": 0.001} +{"train_loss": 0.03843950480222702, "train_loss_bc": 0.03843950480222702, "train_loss_llm": 0.0, "grad_norm": 0.015565000474452972, "global_step": 5097, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04330383986234665, "train_loss_bc": 0.04327964037656784, "train_loss_llm": 0.0024199187755584717, "grad_norm": 0.020230190828442574, "global_step": 5098, "epoch": 141, "lr": 0.001} +{"train_loss": 0.040620122104883194, "train_loss_bc": 0.040620122104883194, "train_loss_llm": 0.0, "grad_norm": 0.0279321800917387, "global_step": 5099, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04197142645716667, "train_loss_bc": 0.04194937273859978, "train_loss_llm": 0.00220528244972229, "grad_norm": 0.0342395044863224, "global_step": 5100, "epoch": 141, "lr": 0.001} +{"train_loss": 0.03855085372924805, "train_loss_bc": 0.03852805495262146, "train_loss_llm": 0.0022800564765930176, "grad_norm": 0.018907321617007256, "global_step": 5101, "epoch": 141, "lr": 0.001} +{"train_loss": 0.0432230606675148, "train_loss_bc": 0.04322062060236931, "train_loss_llm": 0.000244140625, "grad_norm": 0.021863875910639763, "global_step": 5102, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04792577773332596, "train_loss_bc": 0.04792577773332596, "train_loss_llm": 0.0, "grad_norm": 0.03237708657979965, "global_step": 5103, "epoch": 141, "lr": 0.001} +{"train_loss": 0.042064882814884186, "train_loss_bc": 0.042064882814884186, "train_loss_llm": 0.0, "grad_norm": 0.03570815175771713, "global_step": 5104, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04024142771959305, "train_loss_bc": 0.04024142771959305, "train_loss_llm": 0.0, "grad_norm": 0.01720980368554592, "global_step": 5105, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04437025636434555, "train_loss_bc": 0.04437025636434555, "train_loss_llm": 0.0, "grad_norm": 0.026992317289114, "global_step": 5106, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04756079241633415, "train_loss_bc": 0.04756079241633415, "train_loss_llm": 0.0, "grad_norm": 0.03558095917105675, "global_step": 5107, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04397658258676529, "train_loss_bc": 0.04397658258676529, "train_loss_llm": 0.0, "grad_norm": 0.040801163762807846, "global_step": 5108, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04477902874350548, "train_loss_bc": 0.04477079212665558, "train_loss_llm": 0.0008236467838287354, "grad_norm": 0.022008340805768967, "global_step": 5109, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04476260393857956, "train_loss_bc": 0.04476260393857956, "train_loss_llm": 0.0, "grad_norm": 0.02849891409277916, "global_step": 5110, "epoch": 141, "lr": 0.001} +{"train_loss": 0.04192079676108228, "train_loss_bc": 0.04158896952867508, "train_loss_llm": 0.0007988214492797852, "grad_norm": 0.03946750611066818, "global_step": 5111, "epoch": 141, "lr": 0.001, "val_loss": 0.053923770785331726} +{"train_loss": 0.03336380422115326, "train_loss_bc": 0.03336380422115326, "train_loss_llm": 0.0, "grad_norm": 0.04677797853946686, "global_step": 5112, "epoch": 142, "lr": 0.001} +{"train_loss": 0.03865429386496544, "train_loss_bc": 0.03865429386496544, "train_loss_llm": 0.0, "grad_norm": 0.017302880063652992, "global_step": 5113, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04488832876086235, "train_loss_bc": 0.04488832876086235, "train_loss_llm": 0.0, "grad_norm": 0.029203787446022034, "global_step": 5114, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04684862866997719, "train_loss_bc": 0.04684862866997719, "train_loss_llm": 0.0, "grad_norm": 0.03514522314071655, "global_step": 5115, "epoch": 142, "lr": 0.001} +{"train_loss": 0.03867421671748161, "train_loss_bc": 0.038660887628793716, "train_loss_llm": 0.0013328492641448975, "grad_norm": 0.040926896035671234, "global_step": 5116, "epoch": 142, "lr": 0.001} +{"train_loss": 0.043390095233917236, "train_loss_bc": 0.043374598026275635, "train_loss_llm": 0.00154954195022583, "grad_norm": 0.014210028573870659, "global_step": 5117, "epoch": 142, "lr": 0.001} +{"train_loss": 0.049644071608781815, "train_loss_bc": 0.04962192475795746, "train_loss_llm": 0.0022145211696624756, "grad_norm": 0.02589704655110836, "global_step": 5118, "epoch": 142, "lr": 0.001} +{"train_loss": 0.03552894666790962, "train_loss_bc": 0.03552894666790962, "train_loss_llm": 0.0, "grad_norm": 0.03438357263803482, "global_step": 5119, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04433588683605194, "train_loss_bc": 0.0443294532597065, "train_loss_llm": 0.0006434023380279541, "grad_norm": 0.04430494084954262, "global_step": 5120, "epoch": 142, "lr": 0.001} +{"train_loss": 0.03589218482375145, "train_loss_bc": 0.03589218482375145, "train_loss_llm": 0.0, "grad_norm": 0.016365019604563713, "global_step": 5121, "epoch": 142, "lr": 0.001} +{"train_loss": 0.0411037802696228, "train_loss_bc": 0.0411037802696228, "train_loss_llm": 0.0, "grad_norm": 0.0295402854681015, "global_step": 5122, "epoch": 142, "lr": 0.001} +{"train_loss": 0.03712017089128494, "train_loss_bc": 0.037109777331352234, "train_loss_llm": 0.0010393261909484863, "grad_norm": 0.03786240518093109, "global_step": 5123, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04144231602549553, "train_loss_bc": 0.04144231602549553, "train_loss_llm": 0.0, "grad_norm": 0.04338562488555908, "global_step": 5124, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04165871441364288, "train_loss_bc": 0.04165871441364288, "train_loss_llm": 0.0, "grad_norm": 0.015231695957481861, "global_step": 5125, "epoch": 142, "lr": 0.001} +{"train_loss": 0.040493130683898926, "train_loss_bc": 0.04047271981835365, "train_loss_llm": 0.0020410120487213135, "grad_norm": 0.02381179668009281, "global_step": 5126, "epoch": 142, "lr": 0.001} +{"train_loss": 0.0367414616048336, "train_loss_bc": 0.03672614321112633, "train_loss_llm": 0.0015320181846618652, "grad_norm": 0.025830183178186417, "global_step": 5127, "epoch": 142, "lr": 0.001} +{"train_loss": 0.0439620278775692, "train_loss_bc": 0.0439620278775692, "train_loss_llm": 0.0, "grad_norm": 0.029862400144338608, "global_step": 5128, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04139290377497673, "train_loss_bc": 0.04138210043311119, "train_loss_llm": 0.0010801851749420166, "grad_norm": 0.01585032045841217, "global_step": 5129, "epoch": 142, "lr": 0.001} +{"train_loss": 0.0436328686773777, "train_loss_bc": 0.0436328686773777, "train_loss_llm": 0.0, "grad_norm": 0.023702850565314293, "global_step": 5130, "epoch": 142, "lr": 0.001} +{"train_loss": 0.030728910118341446, "train_loss_bc": 0.030728910118341446, "train_loss_llm": 0.0, "grad_norm": 0.026042666286230087, "global_step": 5131, "epoch": 142, "lr": 0.001} +{"train_loss": 0.05162651836872101, "train_loss_bc": 0.05162651836872101, "train_loss_llm": 0.0, "grad_norm": 0.03732878342270851, "global_step": 5132, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04095141962170601, "train_loss_bc": 0.04095141962170601, "train_loss_llm": 0.0, "grad_norm": 0.02366878092288971, "global_step": 5133, "epoch": 142, "lr": 0.001} +{"train_loss": 0.05197836086153984, "train_loss_bc": 0.05197836086153984, "train_loss_llm": 0.0, "grad_norm": 0.028668934479355812, "global_step": 5134, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04104466736316681, "train_loss_bc": 0.04103318229317665, "train_loss_llm": 0.001148611307144165, "grad_norm": 0.0343019999563694, "global_step": 5135, "epoch": 142, "lr": 0.001} +{"train_loss": 0.03657713159918785, "train_loss_bc": 0.03657713159918785, "train_loss_llm": 0.0, "grad_norm": 0.03908301889896393, "global_step": 5136, "epoch": 142, "lr": 0.001} +{"train_loss": 0.042748238891363144, "train_loss_bc": 0.042748238891363144, "train_loss_llm": 0.0, "grad_norm": 0.014145907014608383, "global_step": 5137, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04360015690326691, "train_loss_bc": 0.04360015690326691, "train_loss_llm": 0.0, "grad_norm": 0.024468792602419853, "global_step": 5138, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04225466400384903, "train_loss_bc": 0.04225233569741249, "train_loss_llm": 0.0002326667308807373, "grad_norm": 0.03231650963425636, "global_step": 5139, "epoch": 142, "lr": 0.001} +{"train_loss": 0.05198054015636444, "train_loss_bc": 0.05198054015636444, "train_loss_llm": 0.0, "grad_norm": 0.03887857869267464, "global_step": 5140, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04088998958468437, "train_loss_bc": 0.04088998958468437, "train_loss_llm": 0.0, "grad_norm": 0.01536914985626936, "global_step": 5141, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04044333100318909, "train_loss_bc": 0.04044333100318909, "train_loss_llm": 0.0, "grad_norm": 0.020389018580317497, "global_step": 5142, "epoch": 142, "lr": 0.001} +{"train_loss": 0.03988559916615486, "train_loss_bc": 0.03988559916615486, "train_loss_llm": 0.0, "grad_norm": 0.02644549310207367, "global_step": 5143, "epoch": 142, "lr": 0.001} +{"train_loss": 0.0404052697122097, "train_loss_bc": 0.0404052697122097, "train_loss_llm": 0.0, "grad_norm": 0.03220544382929802, "global_step": 5144, "epoch": 142, "lr": 0.001} +{"train_loss": 0.042647842317819595, "train_loss_bc": 0.042647842317819595, "train_loss_llm": 0.0, "grad_norm": 0.01688028872013092, "global_step": 5145, "epoch": 142, "lr": 0.001} +{"train_loss": 0.046599891036748886, "train_loss_bc": 0.0465974360704422, "train_loss_llm": 0.0002455711364746094, "grad_norm": 0.025567136704921722, "global_step": 5146, "epoch": 142, "lr": 0.001} +{"train_loss": 0.04190339158392615, "train_loss_bc": 0.04536235332489014, "train_loss_llm": 0.002938181161880493, "grad_norm": 0.03544178232550621, "global_step": 5147, "epoch": 142, "lr": 0.001, "val_loss": 0.05474523454904556} +{"train_loss": 0.03924761340022087, "train_loss_bc": 0.03924761340022087, "train_loss_llm": 0.0, "grad_norm": 0.03907737135887146, "global_step": 5148, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04216865077614784, "train_loss_bc": 0.04216031730175018, "train_loss_llm": 0.0008332133293151855, "grad_norm": 0.014933335594832897, "global_step": 5149, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04570245370268822, "train_loss_bc": 0.04568217694759369, "train_loss_llm": 0.0020277798175811768, "grad_norm": 0.021702218800783157, "global_step": 5150, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04081973060965538, "train_loss_bc": 0.04081973060965538, "train_loss_llm": 0.0, "grad_norm": 0.02789395861327648, "global_step": 5151, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04621313512325287, "train_loss_bc": 0.04620786011219025, "train_loss_llm": 0.0005276799201965332, "grad_norm": 0.032189421355724335, "global_step": 5152, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04145500063896179, "train_loss_bc": 0.04145500063896179, "train_loss_llm": 0.0, "grad_norm": 0.02178793214261532, "global_step": 5153, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04448641091585159, "train_loss_bc": 0.04448641091585159, "train_loss_llm": 0.0, "grad_norm": 0.03136403486132622, "global_step": 5154, "epoch": 143, "lr": 0.001} +{"train_loss": 0.03893667086958885, "train_loss_bc": 0.03893667086958885, "train_loss_llm": 0.0, "grad_norm": 0.03664621710777283, "global_step": 5155, "epoch": 143, "lr": 0.001} +{"train_loss": 0.036961063742637634, "train_loss_bc": 0.036961063742637634, "train_loss_llm": 0.0, "grad_norm": 0.03416535630822182, "global_step": 5156, "epoch": 143, "lr": 0.001} +{"train_loss": 0.03467004373669624, "train_loss_bc": 0.03467004373669624, "train_loss_llm": 0.0, "grad_norm": 0.013288543559610844, "global_step": 5157, "epoch": 143, "lr": 0.001} +{"train_loss": 0.03939402848482132, "train_loss_bc": 0.03939402848482132, "train_loss_llm": 0.0, "grad_norm": 0.023695195093750954, "global_step": 5158, "epoch": 143, "lr": 0.001} +{"train_loss": 0.043967533856630325, "train_loss_bc": 0.043967533856630325, "train_loss_llm": 0.0, "grad_norm": 0.03070288896560669, "global_step": 5159, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04125501587986946, "train_loss_bc": 0.04125501587986946, "train_loss_llm": 0.0, "grad_norm": 0.03600125014781952, "global_step": 5160, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04366154968738556, "train_loss_bc": 0.04366154968738556, "train_loss_llm": 0.0, "grad_norm": 0.016670579090714455, "global_step": 5161, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04558843746781349, "train_loss_bc": 0.04552706703543663, "train_loss_llm": 0.006136953830718994, "grad_norm": 0.03338538482785225, "global_step": 5162, "epoch": 143, "lr": 0.001} +{"train_loss": 0.045525193214416504, "train_loss_bc": 0.045525193214416504, "train_loss_llm": 0.0, "grad_norm": 0.04615648835897446, "global_step": 5163, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04033234715461731, "train_loss_bc": 0.04033234715461731, "train_loss_llm": 0.0, "grad_norm": 0.05554055795073509, "global_step": 5164, "epoch": 143, "lr": 0.001} +{"train_loss": 0.039078567177057266, "train_loss_bc": 0.03903212025761604, "train_loss_llm": 0.0046447813510894775, "grad_norm": 0.015629883855581284, "global_step": 5165, "epoch": 143, "lr": 0.001} +{"train_loss": 0.038670700043439865, "train_loss_bc": 0.03864942863583565, "train_loss_llm": 0.0021272897720336914, "grad_norm": 0.02300814352929592, "global_step": 5166, "epoch": 143, "lr": 0.001} +{"train_loss": 0.03699703514575958, "train_loss_bc": 0.03699703514575958, "train_loss_llm": 0.0, "grad_norm": 0.029454756528139114, "global_step": 5167, "epoch": 143, "lr": 0.001} +{"train_loss": 0.03901412710547447, "train_loss_bc": 0.03901412710547447, "train_loss_llm": 0.0, "grad_norm": 0.031067634001374245, "global_step": 5168, "epoch": 143, "lr": 0.001} +{"train_loss": 0.044915810227394104, "train_loss_bc": 0.04490695148706436, "train_loss_llm": 0.0008858740329742432, "grad_norm": 0.01390068233013153, "global_step": 5169, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04501059278845787, "train_loss_bc": 0.04501059278845787, "train_loss_llm": 0.0, "grad_norm": 0.022331157699227333, "global_step": 5170, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04498239979147911, "train_loss_bc": 0.04498239979147911, "train_loss_llm": 0.0, "grad_norm": 0.02566051296889782, "global_step": 5171, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04089374095201492, "train_loss_bc": 0.04089374095201492, "train_loss_llm": 0.0, "grad_norm": 0.03041161224246025, "global_step": 5172, "epoch": 143, "lr": 0.001} +{"train_loss": 0.048991087824106216, "train_loss_bc": 0.048991087824106216, "train_loss_llm": 0.0, "grad_norm": 0.020623572170734406, "global_step": 5173, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04801034927368164, "train_loss_bc": 0.047976020723581314, "train_loss_llm": 0.003432750701904297, "grad_norm": 0.02898213267326355, "global_step": 5174, "epoch": 143, "lr": 0.001} +{"train_loss": 0.03930842503905296, "train_loss_bc": 0.03930842503905296, "train_loss_llm": 0.0, "grad_norm": 0.035585857927799225, "global_step": 5175, "epoch": 143, "lr": 0.001} +{"train_loss": 0.044665005058050156, "train_loss_bc": 0.04466162994503975, "train_loss_llm": 0.0003375113010406494, "grad_norm": 0.04006601497530937, "global_step": 5176, "epoch": 143, "lr": 0.001} +{"train_loss": 0.03668300434947014, "train_loss_bc": 0.036672595888376236, "train_loss_llm": 0.0010409057140350342, "grad_norm": 0.015855591744184494, "global_step": 5177, "epoch": 143, "lr": 0.001} +{"train_loss": 0.042003996670246124, "train_loss_bc": 0.041994620114564896, "train_loss_llm": 0.0009376704692840576, "grad_norm": 0.023650482296943665, "global_step": 5178, "epoch": 143, "lr": 0.001} +{"train_loss": 0.03783189132809639, "train_loss_bc": 0.03783189132809639, "train_loss_llm": 0.0, "grad_norm": 0.025628868490457535, "global_step": 5179, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04595992714166641, "train_loss_bc": 0.04595992714166641, "train_loss_llm": 0.0, "grad_norm": 0.03334999457001686, "global_step": 5180, "epoch": 143, "lr": 0.001} +{"train_loss": 0.04375737905502319, "train_loss_bc": 0.04375737905502319, "train_loss_llm": 0.0, "grad_norm": 0.013243639841675758, "global_step": 5181, "epoch": 143, "lr": 0.001} +{"train_loss": 0.037588778883218765, "train_loss_bc": 0.03758751228451729, "train_loss_llm": 0.00012668967247009277, "grad_norm": 0.02403670735657215, "global_step": 5182, "epoch": 143, "lr": 0.001} +{"train_loss": 0.041743186509443655, "train_loss_bc": 0.037995848804712296, "train_loss_llm": 0.0011166632175445557, "grad_norm": 0.031465403735637665, "global_step": 5183, "epoch": 143, "lr": 0.001, "val_loss": 0.0538703128695488} +{"train_loss": 0.044645097106695175, "train_loss_bc": 0.044645097106695175, "train_loss_llm": 0.0, "grad_norm": 0.03901515156030655, "global_step": 5184, "epoch": 144, "lr": 0.001} +{"train_loss": 0.03918951377272606, "train_loss_bc": 0.03918951377272606, "train_loss_llm": 0.0, "grad_norm": 0.017389601096510887, "global_step": 5185, "epoch": 144, "lr": 0.001} +{"train_loss": 0.03616328909993172, "train_loss_bc": 0.03615168482065201, "train_loss_llm": 0.001160264015197754, "grad_norm": 0.028375595808029175, "global_step": 5186, "epoch": 144, "lr": 0.001} +{"train_loss": 0.03301346302032471, "train_loss_bc": 0.033000148832798004, "train_loss_llm": 0.001331418752670288, "grad_norm": 0.04117211699485779, "global_step": 5187, "epoch": 144, "lr": 0.001} +{"train_loss": 0.039090659469366074, "train_loss_bc": 0.039090659469366074, "train_loss_llm": 0.0, "grad_norm": 0.055861249566078186, "global_step": 5188, "epoch": 144, "lr": 0.001} +{"train_loss": 0.046935755759477615, "train_loss_bc": 0.04692407324910164, "train_loss_llm": 0.0011681616306304932, "grad_norm": 0.01948472112417221, "global_step": 5189, "epoch": 144, "lr": 0.001} +{"train_loss": 0.04446745663881302, "train_loss_bc": 0.04446745663881302, "train_loss_llm": 0.0, "grad_norm": 0.02047394961118698, "global_step": 5190, "epoch": 144, "lr": 0.001} +{"train_loss": 0.03784649074077606, "train_loss_bc": 0.03784649074077606, "train_loss_llm": 0.0, "grad_norm": 0.023709218949079514, "global_step": 5191, "epoch": 144, "lr": 0.001} +{"train_loss": 0.051451049745082855, "train_loss_bc": 0.051451049745082855, "train_loss_llm": 0.0, "grad_norm": 0.03210405632853508, "global_step": 5192, "epoch": 144, "lr": 0.001} +{"train_loss": 0.03978131711483002, "train_loss_bc": 0.039747316390275955, "train_loss_llm": 0.0033999383449554443, "grad_norm": 0.014355644583702087, "global_step": 5193, "epoch": 144, "lr": 0.001} +{"train_loss": 0.04307035356760025, "train_loss_bc": 0.04307035356760025, "train_loss_llm": 0.0, "grad_norm": 0.02054426074028015, "global_step": 5194, "epoch": 144, "lr": 0.001} +{"train_loss": 0.033264804631471634, "train_loss_bc": 0.033251725137233734, "train_loss_llm": 0.0013078153133392334, "grad_norm": 0.02807980217039585, "global_step": 5195, "epoch": 144, "lr": 0.001} +{"train_loss": 0.038867488503456116, "train_loss_bc": 0.03885701671242714, "train_loss_llm": 0.0010471045970916748, "grad_norm": 0.037113018333911896, "global_step": 5196, "epoch": 144, "lr": 0.001} +{"train_loss": 0.044265128672122955, "train_loss_bc": 0.044265128672122955, "train_loss_llm": 0.0, "grad_norm": 0.023670712485909462, "global_step": 5197, "epoch": 144, "lr": 0.001} +{"train_loss": 0.04282982647418976, "train_loss_bc": 0.04282982647418976, "train_loss_llm": 0.0, "grad_norm": 0.030923306941986084, "global_step": 5198, "epoch": 144, "lr": 0.001} +{"train_loss": 0.03976315259933472, "train_loss_bc": 0.039758700877428055, "train_loss_llm": 0.00044530630111694336, "grad_norm": 0.04026099666953087, "global_step": 5199, "epoch": 144, "lr": 0.001} +{"train_loss": 0.03948784992098808, "train_loss_bc": 0.03948784992098808, "train_loss_llm": 0.0, "grad_norm": 0.04749102145433426, "global_step": 5200, "epoch": 144, "lr": 0.001} +{"train_loss": 0.040038350969552994, "train_loss_bc": 0.040038350969552994, "train_loss_llm": 0.0, "grad_norm": 0.020177975296974182, "global_step": 5201, "epoch": 144, "lr": 0.001} +{"train_loss": 0.04133522883057594, "train_loss_bc": 0.04133522883057594, "train_loss_llm": 0.0, "grad_norm": 0.03330046683549881, "global_step": 5202, "epoch": 144, "lr": 0.001} +{"train_loss": 0.045694172382354736, "train_loss_bc": 0.045694172382354736, "train_loss_llm": 0.0, "grad_norm": 0.04079682007431984, "global_step": 5203, "epoch": 144, "lr": 0.001} +{"train_loss": 0.03970241919159889, "train_loss_bc": 0.03970241919159889, "train_loss_llm": 0.0, "grad_norm": 0.053781263530254364, "global_step": 5204, "epoch": 144, "lr": 0.001} +{"train_loss": 0.03716616705060005, "train_loss_bc": 0.03716616705060005, "train_loss_llm": 0.0, "grad_norm": 0.014952164143323898, "global_step": 5205, "epoch": 144, "lr": 0.001} +{"train_loss": 0.042943719774484634, "train_loss_bc": 0.04293868690729141, "train_loss_llm": 0.0005033314228057861, "grad_norm": 0.022125229239463806, "global_step": 5206, "epoch": 144, "lr": 0.001} +{"train_loss": 0.049228351563215256, "train_loss_bc": 0.04921194538474083, "train_loss_llm": 0.0016405284404754639, "grad_norm": 0.03157360106706619, "global_step": 5207, "epoch": 144, "lr": 0.001} +{"train_loss": 0.04074327275156975, "train_loss_bc": 0.04074327275156975, "train_loss_llm": 0.0, "grad_norm": 0.035584382712841034, "global_step": 5208, "epoch": 144, "lr": 0.001} +{"train_loss": 0.04665675386786461, "train_loss_bc": 0.04665675386786461, "train_loss_llm": 0.0, "grad_norm": 0.01782514527440071, "global_step": 5209, "epoch": 144, "lr": 0.001} +{"train_loss": 0.042625393718481064, "train_loss_bc": 0.042625393718481064, "train_loss_llm": 0.0, "grad_norm": 0.026007696986198425, "global_step": 5210, "epoch": 144, "lr": 0.001} +{"train_loss": 0.03814096748828888, "train_loss_bc": 0.03814096748828888, "train_loss_llm": 0.0, "grad_norm": 0.03394964337348938, "global_step": 5211, "epoch": 144, "lr": 0.001} +{"train_loss": 0.04309554025530815, "train_loss_bc": 0.04309000447392464, "train_loss_llm": 0.000553429126739502, "grad_norm": 0.03921923413872719, "global_step": 5212, "epoch": 144, "lr": 0.001} +{"train_loss": 0.051540594547986984, "train_loss_bc": 0.051540594547986984, "train_loss_llm": 0.0, "grad_norm": 0.01635970175266266, "global_step": 5213, "epoch": 144, "lr": 0.001} +{"train_loss": 0.032910268753767014, "train_loss_bc": 0.032910268753767014, "train_loss_llm": 0.0, "grad_norm": 0.022658292204141617, "global_step": 5214, "epoch": 144, "lr": 0.001} +{"train_loss": 0.04186169430613518, "train_loss_bc": 0.041854336857795715, "train_loss_llm": 0.0007356107234954834, "grad_norm": 0.032826978713274, "global_step": 5215, "epoch": 144, "lr": 0.001} +{"train_loss": 0.044359780848026276, "train_loss_bc": 0.044359780848026276, "train_loss_llm": 0.0, "grad_norm": 0.043785762041807175, "global_step": 5216, "epoch": 144, "lr": 0.001} +{"train_loss": 0.04034570977091789, "train_loss_bc": 0.04034570977091789, "train_loss_llm": 0.0, "grad_norm": 0.0178530253469944, "global_step": 5217, "epoch": 144, "lr": 0.001} +{"train_loss": 0.04890943318605423, "train_loss_bc": 0.04889142885804176, "train_loss_llm": 0.0018005967140197754, "grad_norm": 0.03287983313202858, "global_step": 5218, "epoch": 144, "lr": 0.001} +{"train_loss": 0.041880924358136125, "train_loss_bc": 0.04628276079893112, "train_loss_llm": 0.0, "grad_norm": 0.039788149297237396, "global_step": 5219, "epoch": 144, "lr": 0.001, "val_loss": 0.054466456174850464} +{"train_loss": 0.03839943930506706, "train_loss_bc": 0.03839943930506706, "train_loss_llm": 0.0, "grad_norm": 0.049807943403720856, "global_step": 5220, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04593854025006294, "train_loss_bc": 0.04593854025006294, "train_loss_llm": 0.0, "grad_norm": 0.024201441556215286, "global_step": 5221, "epoch": 145, "lr": 0.001} +{"train_loss": 0.038197942078113556, "train_loss_bc": 0.038197942078113556, "train_loss_llm": 0.0, "grad_norm": 0.03372630476951599, "global_step": 5222, "epoch": 145, "lr": 0.001} +{"train_loss": 0.033441103994846344, "train_loss_bc": 0.033441103994846344, "train_loss_llm": 0.0, "grad_norm": 0.042079176753759384, "global_step": 5223, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04271288961172104, "train_loss_bc": 0.04270723834633827, "train_loss_llm": 0.0005652308464050293, "grad_norm": 0.044920664280653, "global_step": 5224, "epoch": 145, "lr": 0.001} +{"train_loss": 0.03990902379155159, "train_loss_bc": 0.03990902379155159, "train_loss_llm": 0.0, "grad_norm": 0.016396746039390564, "global_step": 5225, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04118569940328598, "train_loss_bc": 0.04118569940328598, "train_loss_llm": 0.0, "grad_norm": 0.02056352235376835, "global_step": 5226, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04204131290316582, "train_loss_bc": 0.04204131290316582, "train_loss_llm": 0.0, "grad_norm": 0.025191448628902435, "global_step": 5227, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04317338019609451, "train_loss_bc": 0.043141234666109085, "train_loss_llm": 0.0032146573066711426, "grad_norm": 0.031012991443276405, "global_step": 5228, "epoch": 145, "lr": 0.001} +{"train_loss": 0.036350276321172714, "train_loss_bc": 0.036350276321172714, "train_loss_llm": 0.0, "grad_norm": 0.013734552077949047, "global_step": 5229, "epoch": 145, "lr": 0.001} +{"train_loss": 0.03932938724756241, "train_loss_bc": 0.03932938724756241, "train_loss_llm": 0.0, "grad_norm": 0.02219422720372677, "global_step": 5230, "epoch": 145, "lr": 0.001} +{"train_loss": 0.046312231570482254, "train_loss_bc": 0.046312231570482254, "train_loss_llm": 0.0, "grad_norm": 0.03392810374498367, "global_step": 5231, "epoch": 145, "lr": 0.001} +{"train_loss": 0.03804359957575798, "train_loss_bc": 0.03804359957575798, "train_loss_llm": 0.0, "grad_norm": 0.03562764823436737, "global_step": 5232, "epoch": 145, "lr": 0.001} +{"train_loss": 0.045652229338884354, "train_loss_bc": 0.045652229338884354, "train_loss_llm": 0.0, "grad_norm": 0.015075797215104103, "global_step": 5233, "epoch": 145, "lr": 0.001} +{"train_loss": 0.036848776042461395, "train_loss_bc": 0.03682243078947067, "train_loss_llm": 0.0026344358921051025, "grad_norm": 0.021441513672471046, "global_step": 5234, "epoch": 145, "lr": 0.001} +{"train_loss": 0.039701979607343674, "train_loss_bc": 0.039692603051662445, "train_loss_llm": 0.0009375214576721191, "grad_norm": 0.028750861063599586, "global_step": 5235, "epoch": 145, "lr": 0.001} +{"train_loss": 0.041228387504816055, "train_loss_bc": 0.041217297315597534, "train_loss_llm": 0.0011089444160461426, "grad_norm": 0.03607209026813507, "global_step": 5236, "epoch": 145, "lr": 0.001} +{"train_loss": 0.0409705676138401, "train_loss_bc": 0.0409705676138401, "train_loss_llm": 0.0, "grad_norm": 0.013155621476471424, "global_step": 5237, "epoch": 145, "lr": 0.001} +{"train_loss": 0.043513454496860504, "train_loss_bc": 0.043513454496860504, "train_loss_llm": 0.0, "grad_norm": 0.01938900724053383, "global_step": 5238, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04151082783937454, "train_loss_bc": 0.04149320349097252, "train_loss_llm": 0.0017622709274291992, "grad_norm": 0.020351070910692215, "global_step": 5239, "epoch": 145, "lr": 0.001} +{"train_loss": 0.040263380855321884, "train_loss_bc": 0.040263380855321884, "train_loss_llm": 0.0, "grad_norm": 0.025563165545463562, "global_step": 5240, "epoch": 145, "lr": 0.001} +{"train_loss": 0.042681772261857986, "train_loss_bc": 0.04268098622560501, "train_loss_llm": 7.849931716918945e-05, "grad_norm": 0.015914088115096092, "global_step": 5241, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04247121140360832, "train_loss_bc": 0.04247121140360832, "train_loss_llm": 0.0, "grad_norm": 0.027495726943016052, "global_step": 5242, "epoch": 145, "lr": 0.001} +{"train_loss": 0.03611484542489052, "train_loss_bc": 0.03609307110309601, "train_loss_llm": 0.002177596092224121, "grad_norm": 0.03455303981900215, "global_step": 5243, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04648798331618309, "train_loss_bc": 0.0464838407933712, "train_loss_llm": 0.00041431188583374023, "grad_norm": 0.04475727677345276, "global_step": 5244, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04364236444234848, "train_loss_bc": 0.04364236444234848, "train_loss_llm": 0.0, "grad_norm": 0.020760327577590942, "global_step": 5245, "epoch": 145, "lr": 0.001} +{"train_loss": 0.044933296740055084, "train_loss_bc": 0.04491741210222244, "train_loss_llm": 0.0015886127948760986, "grad_norm": 0.02957651950418949, "global_step": 5246, "epoch": 145, "lr": 0.001} +{"train_loss": 0.0401853583753109, "train_loss_bc": 0.0401853583753109, "train_loss_llm": 0.0, "grad_norm": 0.037908971309661865, "global_step": 5247, "epoch": 145, "lr": 0.001} +{"train_loss": 0.045691005885601044, "train_loss_bc": 0.04568988457322121, "train_loss_llm": 0.00011199712753295898, "grad_norm": 0.053864818066358566, "global_step": 5248, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04379967972636223, "train_loss_bc": 0.04379967972636223, "train_loss_llm": 0.0, "grad_norm": 0.013324864208698273, "global_step": 5249, "epoch": 145, "lr": 0.001} +{"train_loss": 0.040337663143873215, "train_loss_bc": 0.040337663143873215, "train_loss_llm": 0.0, "grad_norm": 0.022274846211075783, "global_step": 5250, "epoch": 145, "lr": 0.001} +{"train_loss": 0.05278271064162254, "train_loss_bc": 0.052758537232875824, "train_loss_llm": 0.0024175047874450684, "grad_norm": 0.03142401576042175, "global_step": 5251, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04071027785539627, "train_loss_bc": 0.04071027785539627, "train_loss_llm": 0.0, "grad_norm": 0.04404762014746666, "global_step": 5252, "epoch": 145, "lr": 0.001} +{"train_loss": 0.03639332950115204, "train_loss_bc": 0.03639332950115204, "train_loss_llm": 0.0, "grad_norm": 0.015240252017974854, "global_step": 5253, "epoch": 145, "lr": 0.001} +{"train_loss": 0.049447737634181976, "train_loss_bc": 0.04940875619649887, "train_loss_llm": 0.003898322582244873, "grad_norm": 0.02483856864273548, "global_step": 5254, "epoch": 145, "lr": 0.001} +{"train_loss": 0.04184546238846249, "train_loss_bc": 0.04603298008441925, "train_loss_llm": 0.0, "grad_norm": 0.03194453567266464, "global_step": 5255, "epoch": 145, "lr": 0.001, "train/cumulative_reward": 114.96970492172584, "train/mean_score": 17.846216913187032, "train/success_rate": 0.4, "test/cumulative_reward": 84.36341301299832, "test/mean_score": 11.423747207371719, "test/success_rate": 0.24, "val_loss": 0.05406264588236809, "train_action_mse_error": 0.045783981680870056} +{"train_loss": 0.03846363723278046, "train_loss_bc": 0.03845195472240448, "train_loss_llm": 0.0011682510375976562, "grad_norm": 0.034179311245679855, "global_step": 5256, "epoch": 146, "lr": 0.001} +{"train_loss": 0.043222978711128235, "train_loss_bc": 0.043222978711128235, "train_loss_llm": 0.0, "grad_norm": 0.013934282585978508, "global_step": 5257, "epoch": 146, "lr": 0.001} +{"train_loss": 0.0421149767935276, "train_loss_bc": 0.04210267961025238, "train_loss_llm": 0.0012296736240386963, "grad_norm": 0.023612530902028084, "global_step": 5258, "epoch": 146, "lr": 0.001} +{"train_loss": 0.03946053609251976, "train_loss_bc": 0.03944080322980881, "train_loss_llm": 0.0019731223583221436, "grad_norm": 0.028131524100899696, "global_step": 5259, "epoch": 146, "lr": 0.001} +{"train_loss": 0.0417163111269474, "train_loss_bc": 0.0417163111269474, "train_loss_llm": 0.0, "grad_norm": 0.03323085606098175, "global_step": 5260, "epoch": 146, "lr": 0.001} +{"train_loss": 0.0366097055375576, "train_loss_bc": 0.0366097055375576, "train_loss_llm": 0.0, "grad_norm": 0.014237516559660435, "global_step": 5261, "epoch": 146, "lr": 0.001} +{"train_loss": 0.05448427051305771, "train_loss_bc": 0.054464828222990036, "train_loss_llm": 0.0019440948963165283, "grad_norm": 0.029973722994327545, "global_step": 5262, "epoch": 146, "lr": 0.001} +{"train_loss": 0.04194425418972969, "train_loss_bc": 0.0419357530772686, "train_loss_llm": 0.0008500218391418457, "grad_norm": 0.034692563116550446, "global_step": 5263, "epoch": 146, "lr": 0.001} +{"train_loss": 0.04012151062488556, "train_loss_bc": 0.04009811952710152, "train_loss_llm": 0.0023391544818878174, "grad_norm": 0.046212051063776016, "global_step": 5264, "epoch": 146, "lr": 0.001} +{"train_loss": 0.041938234120607376, "train_loss_bc": 0.041938234120607376, "train_loss_llm": 0.0, "grad_norm": 0.015187995508313179, "global_step": 5265, "epoch": 146, "lr": 0.001} +{"train_loss": 0.044233281165361404, "train_loss_bc": 0.044211119413375854, "train_loss_llm": 0.002216249704360962, "grad_norm": 0.022923074662685394, "global_step": 5266, "epoch": 146, "lr": 0.001} +{"train_loss": 0.03869129717350006, "train_loss_bc": 0.03869129717350006, "train_loss_llm": 0.0, "grad_norm": 0.03055020049214363, "global_step": 5267, "epoch": 146, "lr": 0.001} +{"train_loss": 0.04891881346702576, "train_loss_bc": 0.04891881346702576, "train_loss_llm": 0.0, "grad_norm": 0.03816016763448715, "global_step": 5268, "epoch": 146, "lr": 0.001} +{"train_loss": 0.04169211536645889, "train_loss_bc": 0.04169211536645889, "train_loss_llm": 0.0, "grad_norm": 0.02238301746547222, "global_step": 5269, "epoch": 146, "lr": 0.001} +{"train_loss": 0.03738737478852272, "train_loss_bc": 0.03738737478852272, "train_loss_llm": 0.0, "grad_norm": 0.02804429642856121, "global_step": 5270, "epoch": 146, "lr": 0.001} +{"train_loss": 0.04331035166978836, "train_loss_bc": 0.04331035166978836, "train_loss_llm": 0.0, "grad_norm": 0.03599052131175995, "global_step": 5271, "epoch": 146, "lr": 0.001} +{"train_loss": 0.039508093148469925, "train_loss_bc": 0.039508093148469925, "train_loss_llm": 0.0, "grad_norm": 0.04639333859086037, "global_step": 5272, "epoch": 146, "lr": 0.001} +{"train_loss": 0.047974295914173126, "train_loss_bc": 0.047974295914173126, "train_loss_llm": 0.0, "grad_norm": 0.015290778130292892, "global_step": 5273, "epoch": 146, "lr": 0.001} +{"train_loss": 0.039158012717962265, "train_loss_bc": 0.039158012717962265, "train_loss_llm": 0.0, "grad_norm": 0.024154068902134895, "global_step": 5274, "epoch": 146, "lr": 0.001} +{"train_loss": 0.045338522642850876, "train_loss_bc": 0.04533367231488228, "train_loss_llm": 0.0004852116107940674, "grad_norm": 0.02925642766058445, "global_step": 5275, "epoch": 146, "lr": 0.001} +{"train_loss": 0.04337114095687866, "train_loss_bc": 0.04337114095687866, "train_loss_llm": 0.0, "grad_norm": 0.03265441209077835, "global_step": 5276, "epoch": 146, "lr": 0.001} +{"train_loss": 0.036565911024808884, "train_loss_bc": 0.036565911024808884, "train_loss_llm": 0.0, "grad_norm": 0.013830260373651981, "global_step": 5277, "epoch": 146, "lr": 0.001} +{"train_loss": 0.0422627255320549, "train_loss_bc": 0.0422627255320549, "train_loss_llm": 0.0, "grad_norm": 0.019704699516296387, "global_step": 5278, "epoch": 146, "lr": 0.001} +{"train_loss": 0.045230455696582794, "train_loss_bc": 0.045230455696582794, "train_loss_llm": 0.0, "grad_norm": 0.02900855988264084, "global_step": 5279, "epoch": 146, "lr": 0.001} +{"train_loss": 0.046956222504377365, "train_loss_bc": 0.04692909121513367, "train_loss_llm": 0.002713024616241455, "grad_norm": 0.03461600840091705, "global_step": 5280, "epoch": 146, "lr": 0.001} +{"train_loss": 0.03810272365808487, "train_loss_bc": 0.03810272365808487, "train_loss_llm": 0.0, "grad_norm": 0.013972454704344273, "global_step": 5281, "epoch": 146, "lr": 0.001} +{"train_loss": 0.05242585018277168, "train_loss_bc": 0.052425771951675415, "train_loss_llm": 7.867813110351562e-06, "grad_norm": 0.019877739250659943, "global_step": 5282, "epoch": 146, "lr": 0.001} +{"train_loss": 0.045663706958293915, "train_loss_bc": 0.0456533282995224, "train_loss_llm": 0.001037687063217163, "grad_norm": 0.025220686569809914, "global_step": 5283, "epoch": 146, "lr": 0.001} +{"train_loss": 0.04145956411957741, "train_loss_bc": 0.04145956411957741, "train_loss_llm": 0.0, "grad_norm": 0.031210564076900482, "global_step": 5284, "epoch": 146, "lr": 0.001} +{"train_loss": 0.03644426167011261, "train_loss_bc": 0.03644426167011261, "train_loss_llm": 0.0, "grad_norm": 0.013392799533903599, "global_step": 5285, "epoch": 146, "lr": 0.001} +{"train_loss": 0.03504369407892227, "train_loss_bc": 0.035035356879234314, "train_loss_llm": 0.0008337795734405518, "grad_norm": 0.01835809089243412, "global_step": 5286, "epoch": 146, "lr": 0.001} +{"train_loss": 0.04325591027736664, "train_loss_bc": 0.04324272647500038, "train_loss_llm": 0.0013184547424316406, "grad_norm": 0.02398764155805111, "global_step": 5287, "epoch": 146, "lr": 0.001} +{"train_loss": 0.03205295279622078, "train_loss_bc": 0.03205295279622078, "train_loss_llm": 0.0, "grad_norm": 0.029491130262613297, "global_step": 5288, "epoch": 146, "lr": 0.001} +{"train_loss": 0.03447799012064934, "train_loss_bc": 0.03447799012064934, "train_loss_llm": 0.0, "grad_norm": 0.017213501036167145, "global_step": 5289, "epoch": 146, "lr": 0.001} +{"train_loss": 0.03925709053874016, "train_loss_bc": 0.03925709053874016, "train_loss_llm": 0.0, "grad_norm": 0.021121500059962273, "global_step": 5290, "epoch": 146, "lr": 0.001} +{"train_loss": 0.04162698383960459, "train_loss_bc": 0.03971264511346817, "train_loss_llm": 0.0, "grad_norm": 0.03388381749391556, "global_step": 5291, "epoch": 146, "lr": 0.001, "val_loss": 0.05430597811937332} +{"train_loss": 0.03643396124243736, "train_loss_bc": 0.03643396124243736, "train_loss_llm": 0.0, "grad_norm": 0.040479935705661774, "global_step": 5292, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04441065341234207, "train_loss_bc": 0.04441065341234207, "train_loss_llm": 0.0, "grad_norm": 0.020827043801546097, "global_step": 5293, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04626849666237831, "train_loss_bc": 0.04626849666237831, "train_loss_llm": 0.0, "grad_norm": 0.027223654091358185, "global_step": 5294, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04074616730213165, "train_loss_bc": 0.04074616730213165, "train_loss_llm": 0.0, "grad_norm": 0.029190875589847565, "global_step": 5295, "epoch": 147, "lr": 0.001} +{"train_loss": 0.038011614233255386, "train_loss_bc": 0.038011614233255386, "train_loss_llm": 0.0, "grad_norm": 0.0347612090408802, "global_step": 5296, "epoch": 147, "lr": 0.001} +{"train_loss": 0.03717488795518875, "train_loss_bc": 0.03717488795518875, "train_loss_llm": 0.0, "grad_norm": 0.015703709796071053, "global_step": 5297, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04460636153817177, "train_loss_bc": 0.044597599655389786, "train_loss_llm": 0.0008761584758758545, "grad_norm": 0.021296381950378418, "global_step": 5298, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04984060302376747, "train_loss_bc": 0.04984060302376747, "train_loss_llm": 0.0, "grad_norm": 0.028928780928254128, "global_step": 5299, "epoch": 147, "lr": 0.001} +{"train_loss": 0.037276677787303925, "train_loss_bc": 0.037276677787303925, "train_loss_llm": 0.0, "grad_norm": 0.041403040289878845, "global_step": 5300, "epoch": 147, "lr": 0.001} +{"train_loss": 0.040452741086483, "train_loss_bc": 0.040452741086483, "train_loss_llm": 0.0, "grad_norm": 0.01544478815048933, "global_step": 5301, "epoch": 147, "lr": 0.001} +{"train_loss": 0.040697209537029266, "train_loss_bc": 0.040697209537029266, "train_loss_llm": 0.0, "grad_norm": 0.02281833253800869, "global_step": 5302, "epoch": 147, "lr": 0.001} +{"train_loss": 0.038536302745342255, "train_loss_bc": 0.038536302745342255, "train_loss_llm": 0.0, "grad_norm": 0.029108773916959763, "global_step": 5303, "epoch": 147, "lr": 0.001} +{"train_loss": 0.035772405564785004, "train_loss_bc": 0.035772405564785004, "train_loss_llm": 0.0, "grad_norm": 0.03410525619983673, "global_step": 5304, "epoch": 147, "lr": 0.001} +{"train_loss": 0.038411565124988556, "train_loss_bc": 0.03840186074376106, "train_loss_llm": 0.0009703636169433594, "grad_norm": 0.018873311579227448, "global_step": 5305, "epoch": 147, "lr": 0.001} +{"train_loss": 0.043409790843725204, "train_loss_bc": 0.043406400829553604, "train_loss_llm": 0.000338822603225708, "grad_norm": 0.031690485775470734, "global_step": 5306, "epoch": 147, "lr": 0.001} +{"train_loss": 0.05066264048218727, "train_loss_bc": 0.05066264048218727, "train_loss_llm": 0.0, "grad_norm": 0.038276202976703644, "global_step": 5307, "epoch": 147, "lr": 0.001} +{"train_loss": 0.03945760056376457, "train_loss_bc": 0.03945760056376457, "train_loss_llm": 0.0, "grad_norm": 0.043056000024080276, "global_step": 5308, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04162340611219406, "train_loss_bc": 0.04162340611219406, "train_loss_llm": 0.0, "grad_norm": 0.019967587664723396, "global_step": 5309, "epoch": 147, "lr": 0.001} +{"train_loss": 0.03973317891359329, "train_loss_bc": 0.03971438854932785, "train_loss_llm": 0.0018789172172546387, "grad_norm": 0.026082463562488556, "global_step": 5310, "epoch": 147, "lr": 0.001} +{"train_loss": 0.046689558774232864, "train_loss_bc": 0.04668387770652771, "train_loss_llm": 0.0005679428577423096, "grad_norm": 0.037409354001283646, "global_step": 5311, "epoch": 147, "lr": 0.001} +{"train_loss": 0.03506440669298172, "train_loss_bc": 0.03506440669298172, "train_loss_llm": 0.0, "grad_norm": 0.048456478863954544, "global_step": 5312, "epoch": 147, "lr": 0.001} +{"train_loss": 0.037639930844306946, "train_loss_bc": 0.03763556852936745, "train_loss_llm": 0.0004361271858215332, "grad_norm": 0.013966875150799751, "global_step": 5313, "epoch": 147, "lr": 0.001} +{"train_loss": 0.045359332114458084, "train_loss_bc": 0.045359332114458084, "train_loss_llm": 0.0, "grad_norm": 0.0212722010910511, "global_step": 5314, "epoch": 147, "lr": 0.001} +{"train_loss": 0.03982062637805939, "train_loss_bc": 0.03981664404273033, "train_loss_llm": 0.00039821863174438477, "grad_norm": 0.030852533876895905, "global_step": 5315, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04562777653336525, "train_loss_bc": 0.04562777653336525, "train_loss_llm": 0.0, "grad_norm": 0.03738082945346832, "global_step": 5316, "epoch": 147, "lr": 0.001} +{"train_loss": 0.03865828737616539, "train_loss_bc": 0.03865828737616539, "train_loss_llm": 0.0, "grad_norm": 0.01924794539809227, "global_step": 5317, "epoch": 147, "lr": 0.001} +{"train_loss": 0.049858029931783676, "train_loss_bc": 0.04983106628060341, "train_loss_llm": 0.0026962459087371826, "grad_norm": 0.02304256707429886, "global_step": 5318, "epoch": 147, "lr": 0.001} +{"train_loss": 0.048684485256671906, "train_loss_bc": 0.04866114631295204, "train_loss_llm": 0.0023339688777923584, "grad_norm": 0.028016673400998116, "global_step": 5319, "epoch": 147, "lr": 0.001} +{"train_loss": 0.041102126240730286, "train_loss_bc": 0.041102126240730286, "train_loss_llm": 0.0, "grad_norm": 0.034480832517147064, "global_step": 5320, "epoch": 147, "lr": 0.001} +{"train_loss": 0.03618920221924782, "train_loss_bc": 0.03618920221924782, "train_loss_llm": 0.0, "grad_norm": 0.014545056968927383, "global_step": 5321, "epoch": 147, "lr": 0.001} +{"train_loss": 0.03968038782477379, "train_loss_bc": 0.03968038782477379, "train_loss_llm": 0.0, "grad_norm": 0.026891998946666718, "global_step": 5322, "epoch": 147, "lr": 0.001} +{"train_loss": 0.040759678930044174, "train_loss_bc": 0.04074906185269356, "train_loss_llm": 0.0010616481304168701, "grad_norm": 0.031195661053061485, "global_step": 5323, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04565899819135666, "train_loss_bc": 0.04565740004181862, "train_loss_llm": 0.00015971064567565918, "grad_norm": 0.03785731643438339, "global_step": 5324, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04047752544283867, "train_loss_bc": 0.04047638177871704, "train_loss_llm": 0.00011429190635681152, "grad_norm": 0.015349763445556164, "global_step": 5325, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04981693997979164, "train_loss_bc": 0.04981693997979164, "train_loss_llm": 0.0, "grad_norm": 0.022153904661536217, "global_step": 5326, "epoch": 147, "lr": 0.001} +{"train_loss": 0.04160455190059212, "train_loss_bc": 0.033142220228910446, "train_loss_llm": 0.0008090734481811523, "grad_norm": 0.03026687726378441, "global_step": 5327, "epoch": 147, "lr": 0.001, "val_loss": 0.05426332727074623} +{"train_loss": 0.04691512882709503, "train_loss_bc": 0.04691512882709503, "train_loss_llm": 0.0, "grad_norm": 0.03659641370177269, "global_step": 5328, "epoch": 148, "lr": 0.001} +{"train_loss": 0.041583333164453506, "train_loss_bc": 0.04157472029328346, "train_loss_llm": 0.0008611977100372314, "grad_norm": 0.01817931979894638, "global_step": 5329, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04201239347457886, "train_loss_bc": 0.042010702192783356, "train_loss_llm": 0.0001690387725830078, "grad_norm": 0.028263846412301064, "global_step": 5330, "epoch": 148, "lr": 0.001} +{"train_loss": 0.03896871209144592, "train_loss_bc": 0.038968347012996674, "train_loss_llm": 3.635883331298828e-05, "grad_norm": 0.03984593227505684, "global_step": 5331, "epoch": 148, "lr": 0.001} +{"train_loss": 0.047991350293159485, "train_loss_bc": 0.047991350293159485, "train_loss_llm": 0.0, "grad_norm": 0.05571853369474411, "global_step": 5332, "epoch": 148, "lr": 0.001} +{"train_loss": 0.036516331136226654, "train_loss_bc": 0.036516331136226654, "train_loss_llm": 0.0, "grad_norm": 0.020222511142492294, "global_step": 5333, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04574257507920265, "train_loss_bc": 0.04574257507920265, "train_loss_llm": 0.0, "grad_norm": 0.030445130541920662, "global_step": 5334, "epoch": 148, "lr": 0.001} +{"train_loss": 0.044547513127326965, "train_loss_bc": 0.044547513127326965, "train_loss_llm": 0.0, "grad_norm": 0.04856206476688385, "global_step": 5335, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04661991074681282, "train_loss_bc": 0.046604085713624954, "train_loss_llm": 0.001582324504852295, "grad_norm": 0.06338169425725937, "global_step": 5336, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04543574899435043, "train_loss_bc": 0.04543574899435043, "train_loss_llm": 0.0, "grad_norm": 0.017272386699914932, "global_step": 5337, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04266775771975517, "train_loss_bc": 0.04263759404420853, "train_loss_llm": 0.0030164718627929688, "grad_norm": 0.02851831540465355, "global_step": 5338, "epoch": 148, "lr": 0.001} +{"train_loss": 0.03615376725792885, "train_loss_bc": 0.03615376725792885, "train_loss_llm": 0.0, "grad_norm": 0.03374647721648216, "global_step": 5339, "epoch": 148, "lr": 0.001} +{"train_loss": 0.03977418318390846, "train_loss_bc": 0.03977418318390846, "train_loss_llm": 0.0, "grad_norm": 0.03858186677098274, "global_step": 5340, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04044058173894882, "train_loss_bc": 0.04042751342058182, "train_loss_llm": 0.0013068020343780518, "grad_norm": 0.01347984466701746, "global_step": 5341, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04382790997624397, "train_loss_bc": 0.04382790997624397, "train_loss_llm": 0.0, "grad_norm": 0.021668128669261932, "global_step": 5342, "epoch": 148, "lr": 0.001} +{"train_loss": 0.03775374963879585, "train_loss_bc": 0.03775374963879585, "train_loss_llm": 0.0, "grad_norm": 0.0282866433262825, "global_step": 5343, "epoch": 148, "lr": 0.001} +{"train_loss": 0.042873483151197433, "train_loss_bc": 0.042873483151197433, "train_loss_llm": 0.0, "grad_norm": 0.03333515673875809, "global_step": 5344, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04057828709483147, "train_loss_bc": 0.04057828709483147, "train_loss_llm": 0.0, "grad_norm": 0.017221668735146523, "global_step": 5345, "epoch": 148, "lr": 0.001} +{"train_loss": 0.045997705310583115, "train_loss_bc": 0.045997705310583115, "train_loss_llm": 0.0, "grad_norm": 0.026247719302773476, "global_step": 5346, "epoch": 148, "lr": 0.001} +{"train_loss": 0.039978645741939545, "train_loss_bc": 0.03997132182121277, "train_loss_llm": 0.0007322430610656738, "grad_norm": 0.03136847913265228, "global_step": 5347, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04176650941371918, "train_loss_bc": 0.04176650941371918, "train_loss_llm": 0.0, "grad_norm": 0.03458095341920853, "global_step": 5348, "epoch": 148, "lr": 0.001} +{"train_loss": 0.03785201907157898, "train_loss_bc": 0.03785201907157898, "train_loss_llm": 0.0, "grad_norm": 0.016930606216192245, "global_step": 5349, "epoch": 148, "lr": 0.001} +{"train_loss": 0.039347320795059204, "train_loss_bc": 0.039347320795059204, "train_loss_llm": 0.0, "grad_norm": 0.028134869411587715, "global_step": 5350, "epoch": 148, "lr": 0.001} +{"train_loss": 0.052475448697805405, "train_loss_bc": 0.052475448697805405, "train_loss_llm": 0.0, "grad_norm": 0.03192887827754021, "global_step": 5351, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04451967403292656, "train_loss_bc": 0.04451967403292656, "train_loss_llm": 0.0, "grad_norm": 0.04330195114016533, "global_step": 5352, "epoch": 148, "lr": 0.001} +{"train_loss": 0.040281154215335846, "train_loss_bc": 0.040281154215335846, "train_loss_llm": 0.0, "grad_norm": 0.01779683120548725, "global_step": 5353, "epoch": 148, "lr": 0.001} +{"train_loss": 0.0456426665186882, "train_loss_bc": 0.04562968760728836, "train_loss_llm": 0.0012980103492736816, "grad_norm": 0.028805788606405258, "global_step": 5354, "epoch": 148, "lr": 0.001} +{"train_loss": 0.0350269116461277, "train_loss_bc": 0.035001497715711594, "train_loss_llm": 0.002541571855545044, "grad_norm": 0.03652605786919594, "global_step": 5355, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04439802095293999, "train_loss_bc": 0.04436744377017021, "train_loss_llm": 0.003057539463043213, "grad_norm": 0.0552215576171875, "global_step": 5356, "epoch": 148, "lr": 0.001} +{"train_loss": 0.045708268880844116, "train_loss_bc": 0.045708268880844116, "train_loss_llm": 0.0, "grad_norm": 0.016608592122793198, "global_step": 5357, "epoch": 148, "lr": 0.001} +{"train_loss": 0.03491770476102829, "train_loss_bc": 0.034891288727521896, "train_loss_llm": 0.0026417672634124756, "grad_norm": 0.026426812633872032, "global_step": 5358, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04090293124318123, "train_loss_bc": 0.04087258502840996, "train_loss_llm": 0.0030347704887390137, "grad_norm": 0.03545352444052696, "global_step": 5359, "epoch": 148, "lr": 0.001} +{"train_loss": 0.038501136004924774, "train_loss_bc": 0.0384867861866951, "train_loss_llm": 0.0014350712299346924, "grad_norm": 0.04080771654844284, "global_step": 5360, "epoch": 148, "lr": 0.001} +{"train_loss": 0.03699062019586563, "train_loss_bc": 0.03699062019586563, "train_loss_llm": 0.0, "grad_norm": 0.017295395955443382, "global_step": 5361, "epoch": 148, "lr": 0.001} +{"train_loss": 0.038694869726896286, "train_loss_bc": 0.038694869726896286, "train_loss_llm": 0.0, "grad_norm": 0.023764772340655327, "global_step": 5362, "epoch": 148, "lr": 0.001} +{"train_loss": 0.04172585149192148, "train_loss_bc": 0.0387263298034668, "train_loss_llm": 0.0, "grad_norm": 0.032366275787353516, "global_step": 5363, "epoch": 148, "lr": 0.001, "val_loss": 0.054966092109680176} +{"train_loss": 0.03940524533390999, "train_loss_bc": 0.03940524533390999, "train_loss_llm": 0.0, "grad_norm": 0.03845192492008209, "global_step": 5364, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04532924294471741, "train_loss_bc": 0.04532924294471741, "train_loss_llm": 0.0, "grad_norm": 0.018724050372838974, "global_step": 5365, "epoch": 149, "lr": 0.001} +{"train_loss": 0.031643081456422806, "train_loss_bc": 0.03163876757025719, "train_loss_llm": 0.00043144822120666504, "grad_norm": 0.025748074054718018, "global_step": 5366, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04011552408337593, "train_loss_bc": 0.04011552408337593, "train_loss_llm": 0.0, "grad_norm": 0.033826589584350586, "global_step": 5367, "epoch": 149, "lr": 0.001} +{"train_loss": 0.03547409176826477, "train_loss_bc": 0.035471487790346146, "train_loss_llm": 0.00026023387908935547, "grad_norm": 0.03742821142077446, "global_step": 5368, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04345959052443504, "train_loss_bc": 0.04343947768211365, "train_loss_llm": 0.002011120319366455, "grad_norm": 0.017119573429226875, "global_step": 5369, "epoch": 149, "lr": 0.001} +{"train_loss": 0.0479901023209095, "train_loss_bc": 0.04798765853047371, "train_loss_llm": 0.00024446845054626465, "grad_norm": 0.026317482814192772, "global_step": 5370, "epoch": 149, "lr": 0.001} +{"train_loss": 0.040580376982688904, "train_loss_bc": 0.040580376982688904, "train_loss_llm": 0.0, "grad_norm": 0.03461047634482384, "global_step": 5371, "epoch": 149, "lr": 0.001} +{"train_loss": 0.0492643378674984, "train_loss_bc": 0.0492643378674984, "train_loss_llm": 0.0, "grad_norm": 0.044509995728731155, "global_step": 5372, "epoch": 149, "lr": 0.001} +{"train_loss": 0.042028896510601044, "train_loss_bc": 0.04199938103556633, "train_loss_llm": 0.00295141339302063, "grad_norm": 0.018502334132790565, "global_step": 5373, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04192765802145004, "train_loss_bc": 0.04192765802145004, "train_loss_llm": 0.0, "grad_norm": 0.025378189980983734, "global_step": 5374, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04297139495611191, "train_loss_bc": 0.04297139495611191, "train_loss_llm": 0.0, "grad_norm": 0.029941828921437263, "global_step": 5375, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04783892631530762, "train_loss_bc": 0.04783892631530762, "train_loss_llm": 0.0, "grad_norm": 0.03819608688354492, "global_step": 5376, "epoch": 149, "lr": 0.001} +{"train_loss": 0.05013098567724228, "train_loss_bc": 0.050118863582611084, "train_loss_llm": 0.0012120604515075684, "grad_norm": 0.017376312986016273, "global_step": 5377, "epoch": 149, "lr": 0.001} +{"train_loss": 0.041937511414289474, "train_loss_bc": 0.041937511414289474, "train_loss_llm": 0.0, "grad_norm": 0.023562295362353325, "global_step": 5378, "epoch": 149, "lr": 0.001} +{"train_loss": 0.03700793534517288, "train_loss_bc": 0.03700482100248337, "train_loss_llm": 0.00031131505966186523, "grad_norm": 0.025234701111912727, "global_step": 5379, "epoch": 149, "lr": 0.001} +{"train_loss": 0.049197398126125336, "train_loss_bc": 0.049197398126125336, "train_loss_llm": 0.0, "grad_norm": 0.030162932351231575, "global_step": 5380, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04104934632778168, "train_loss_bc": 0.04104934632778168, "train_loss_llm": 0.0, "grad_norm": 0.01707167737185955, "global_step": 5381, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04313592240214348, "train_loss_bc": 0.04313592240214348, "train_loss_llm": 0.0, "grad_norm": 0.022822974249720573, "global_step": 5382, "epoch": 149, "lr": 0.001} +{"train_loss": 0.0348278284072876, "train_loss_bc": 0.0348278284072876, "train_loss_llm": 0.0, "grad_norm": 0.02942253090441227, "global_step": 5383, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04528093710541725, "train_loss_bc": 0.04528093710541725, "train_loss_llm": 0.0, "grad_norm": 0.03138628974556923, "global_step": 5384, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04274780675768852, "train_loss_bc": 0.04274780675768852, "train_loss_llm": 0.0, "grad_norm": 0.01756015419960022, "global_step": 5385, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04220273718237877, "train_loss_bc": 0.04220273718237877, "train_loss_llm": 0.0, "grad_norm": 0.024397175759077072, "global_step": 5386, "epoch": 149, "lr": 0.001} +{"train_loss": 0.03759870305657387, "train_loss_bc": 0.037584926933050156, "train_loss_llm": 0.0013775229454040527, "grad_norm": 0.030084796249866486, "global_step": 5387, "epoch": 149, "lr": 0.001} +{"train_loss": 0.03640352562069893, "train_loss_bc": 0.03640352562069893, "train_loss_llm": 0.0, "grad_norm": 0.03712148219347, "global_step": 5388, "epoch": 149, "lr": 0.001} +{"train_loss": 0.043984655290842056, "train_loss_bc": 0.043984655290842056, "train_loss_llm": 0.0, "grad_norm": 0.020495228469371796, "global_step": 5389, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04185350984334946, "train_loss_bc": 0.04185350984334946, "train_loss_llm": 0.0, "grad_norm": 0.03291250392794609, "global_step": 5390, "epoch": 149, "lr": 0.001} +{"train_loss": 0.048869989812374115, "train_loss_bc": 0.048869989812374115, "train_loss_llm": 0.0, "grad_norm": 0.04324832186102867, "global_step": 5391, "epoch": 149, "lr": 0.001} +{"train_loss": 0.036477476358413696, "train_loss_bc": 0.03644976392388344, "train_loss_llm": 0.0027710795402526855, "grad_norm": 0.05508818104863167, "global_step": 5392, "epoch": 149, "lr": 0.001} +{"train_loss": 0.033483728766441345, "train_loss_bc": 0.033483728766441345, "train_loss_llm": 0.0, "grad_norm": 0.016491781920194626, "global_step": 5393, "epoch": 149, "lr": 0.001} +{"train_loss": 0.03944515436887741, "train_loss_bc": 0.03944515436887741, "train_loss_llm": 0.0, "grad_norm": 0.026537366211414337, "global_step": 5394, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04488128796219826, "train_loss_bc": 0.04488128796219826, "train_loss_llm": 0.0, "grad_norm": 0.03766448423266411, "global_step": 5395, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04109944775700569, "train_loss_bc": 0.04106435924768448, "train_loss_llm": 0.0035087168216705322, "grad_norm": 0.05227658897638321, "global_step": 5396, "epoch": 149, "lr": 0.001} +{"train_loss": 0.03305402398109436, "train_loss_bc": 0.03305402398109436, "train_loss_llm": 0.0, "grad_norm": 0.016353057697415352, "global_step": 5397, "epoch": 149, "lr": 0.001} +{"train_loss": 0.043501682579517365, "train_loss_bc": 0.04349195212125778, "train_loss_llm": 0.000973045825958252, "grad_norm": 0.02732963301241398, "global_step": 5398, "epoch": 149, "lr": 0.001} +{"train_loss": 0.04166558364199267, "train_loss_bc": 0.04376094788312912, "train_loss_llm": 0.0, "grad_norm": 0.032701075077056885, "global_step": 5399, "epoch": 149, "lr": 0.001, "val_loss": 0.05502087250351906} +{"train_loss": 0.04750322178006172, "train_loss_bc": 0.04750322178006172, "train_loss_llm": 0.0, "grad_norm": 0.03743534907698631, "global_step": 5400, "epoch": 150, "lr": 0.001} +{"train_loss": 0.031410131603479385, "train_loss_bc": 0.03140142932534218, "train_loss_llm": 0.0008703470230102539, "grad_norm": 0.016022853553295135, "global_step": 5401, "epoch": 150, "lr": 0.001} +{"train_loss": 0.03938555717468262, "train_loss_bc": 0.039382725954055786, "train_loss_llm": 0.00028321146965026855, "grad_norm": 0.020640747621655464, "global_step": 5402, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04360944777727127, "train_loss_bc": 0.043601591140031815, "train_loss_llm": 0.0007856488227844238, "grad_norm": 0.023924406617879868, "global_step": 5403, "epoch": 150, "lr": 0.001} +{"train_loss": 0.042927335947752, "train_loss_bc": 0.04290212318301201, "train_loss_llm": 0.0025212764739990234, "grad_norm": 0.030231263488531113, "global_step": 5404, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04082723706960678, "train_loss_bc": 0.04082723706960678, "train_loss_llm": 0.0, "grad_norm": 0.016780409961938858, "global_step": 5405, "epoch": 150, "lr": 0.001} +{"train_loss": 0.03580651804804802, "train_loss_bc": 0.03580651804804802, "train_loss_llm": 0.0, "grad_norm": 0.02547585591673851, "global_step": 5406, "epoch": 150, "lr": 0.001} +{"train_loss": 0.03925158083438873, "train_loss_bc": 0.03925158083438873, "train_loss_llm": 0.0, "grad_norm": 0.029429864138364792, "global_step": 5407, "epoch": 150, "lr": 0.001} +{"train_loss": 0.044816624373197556, "train_loss_bc": 0.044816624373197556, "train_loss_llm": 0.0, "grad_norm": 0.037680044770240784, "global_step": 5408, "epoch": 150, "lr": 0.001} +{"train_loss": 0.041034676134586334, "train_loss_bc": 0.041034676134586334, "train_loss_llm": 0.0, "grad_norm": 0.02067958563566208, "global_step": 5409, "epoch": 150, "lr": 0.001} +{"train_loss": 0.040333379060029984, "train_loss_bc": 0.040333379060029984, "train_loss_llm": 0.0, "grad_norm": 0.029827186837792397, "global_step": 5410, "epoch": 150, "lr": 0.001} +{"train_loss": 0.0344596728682518, "train_loss_bc": 0.0344596728682518, "train_loss_llm": 0.0, "grad_norm": 0.04172155633568764, "global_step": 5411, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04384053871035576, "train_loss_bc": 0.04384053871035576, "train_loss_llm": 0.0, "grad_norm": 0.062270890921354294, "global_step": 5412, "epoch": 150, "lr": 0.001} +{"train_loss": 0.03757945075631142, "train_loss_bc": 0.03757945075631142, "train_loss_llm": 0.0, "grad_norm": 0.017704464495182037, "global_step": 5413, "epoch": 150, "lr": 0.001} +{"train_loss": 0.040018919855356216, "train_loss_bc": 0.039986398071050644, "train_loss_llm": 0.003251999616622925, "grad_norm": 0.02627521939575672, "global_step": 5414, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04391138255596161, "train_loss_bc": 0.04391138255596161, "train_loss_llm": 0.0, "grad_norm": 0.034047212451696396, "global_step": 5415, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04994026944041252, "train_loss_bc": 0.04994026944041252, "train_loss_llm": 0.0, "grad_norm": 0.039043981581926346, "global_step": 5416, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04156632721424103, "train_loss_bc": 0.04156632721424103, "train_loss_llm": 0.0, "grad_norm": 0.01668829470872879, "global_step": 5417, "epoch": 150, "lr": 0.001} +{"train_loss": 0.03643601015210152, "train_loss_bc": 0.03643601015210152, "train_loss_llm": 0.0, "grad_norm": 0.02561955526471138, "global_step": 5418, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04310305044054985, "train_loss_bc": 0.04310305044054985, "train_loss_llm": 0.0, "grad_norm": 0.03613648563623428, "global_step": 5419, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04103899374604225, "train_loss_bc": 0.04103899374604225, "train_loss_llm": 0.0, "grad_norm": 0.04268612712621689, "global_step": 5420, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04307962581515312, "train_loss_bc": 0.04307962581515312, "train_loss_llm": 0.0, "grad_norm": 0.015606638044118881, "global_step": 5421, "epoch": 150, "lr": 0.001} +{"train_loss": 0.041455961763858795, "train_loss_bc": 0.041455961763858795, "train_loss_llm": 0.0, "grad_norm": 0.030968112871050835, "global_step": 5422, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04232216998934746, "train_loss_bc": 0.04232216998934746, "train_loss_llm": 0.0, "grad_norm": 0.04144046828150749, "global_step": 5423, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04625121131539345, "train_loss_bc": 0.04625121131539345, "train_loss_llm": 0.0, "grad_norm": 0.053711362183094025, "global_step": 5424, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04136049747467041, "train_loss_bc": 0.041348040103912354, "train_loss_llm": 0.0012456178665161133, "grad_norm": 0.015519878827035427, "global_step": 5425, "epoch": 150, "lr": 0.001} +{"train_loss": 0.042411115020513535, "train_loss_bc": 0.042411115020513535, "train_loss_llm": 0.0, "grad_norm": 0.025803836062550545, "global_step": 5426, "epoch": 150, "lr": 0.001} +{"train_loss": 0.039236463606357574, "train_loss_bc": 0.039193037897348404, "train_loss_llm": 0.004342436790466309, "grad_norm": 0.03332105651497841, "global_step": 5427, "epoch": 150, "lr": 0.001} +{"train_loss": 0.044662971049547195, "train_loss_bc": 0.044662971049547195, "train_loss_llm": 0.0, "grad_norm": 0.03728608414530754, "global_step": 5428, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04143426567316055, "train_loss_bc": 0.04143426567316055, "train_loss_llm": 0.0, "grad_norm": 0.020373040810227394, "global_step": 5429, "epoch": 150, "lr": 0.001} +{"train_loss": 0.05205430090427399, "train_loss_bc": 0.05205430090427399, "train_loss_llm": 0.0, "grad_norm": 0.029618706554174423, "global_step": 5430, "epoch": 150, "lr": 0.001} +{"train_loss": 0.02948693186044693, "train_loss_bc": 0.02948693186044693, "train_loss_llm": 0.0, "grad_norm": 0.0420825257897377, "global_step": 5431, "epoch": 150, "lr": 0.001} +{"train_loss": 0.04854288324713707, "train_loss_bc": 0.04851815477013588, "train_loss_llm": 0.002472788095474243, "grad_norm": 0.04981840401887894, "global_step": 5432, "epoch": 150, "lr": 0.001} +{"train_loss": 0.05057297646999359, "train_loss_bc": 0.05057297646999359, "train_loss_llm": 0.0, "grad_norm": 0.020940259099006653, "global_step": 5433, "epoch": 150, "lr": 0.001} +{"train_loss": 0.0352231040596962, "train_loss_bc": 0.0352231040596962, "train_loss_llm": 0.0, "grad_norm": 0.030592793598771095, "global_step": 5434, "epoch": 150, "lr": 0.001} +{"train_loss": 0.041731316389309034, "train_loss_bc": 0.04543258622288704, "train_loss_llm": 0.0, "grad_norm": 0.043833669275045395, "global_step": 5435, "epoch": 150, "lr": 0.001, "train/cumulative_reward": 73.68621244333812, "train/mean_score": 17.898610362249553, "train/success_rate": 0.5, "test/cumulative_reward": 47.825374010079585, "test/mean_score": 10.32191562187444, "test/success_rate": 0.28, "val_loss": 0.05469541996717453, "train_action_mse_error": 0.04522599279880524} +{"train_loss": 0.04056497663259506, "train_loss_bc": 0.04056497663259506, "train_loss_llm": 0.0, "grad_norm": 0.04373058304190636, "global_step": 5436, "epoch": 151, "lr": 0.001} +{"train_loss": 0.03971599042415619, "train_loss_bc": 0.03971599042415619, "train_loss_llm": 0.0, "grad_norm": 0.016483549028635025, "global_step": 5437, "epoch": 151, "lr": 0.001} +{"train_loss": 0.038786306977272034, "train_loss_bc": 0.038753289729356766, "train_loss_llm": 0.00330159068107605, "grad_norm": 0.02522354945540428, "global_step": 5438, "epoch": 151, "lr": 0.001} +{"train_loss": 0.048577506095170975, "train_loss_bc": 0.048577506095170975, "train_loss_llm": 0.0, "grad_norm": 0.0362737774848938, "global_step": 5439, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04704241454601288, "train_loss_bc": 0.04701985418796539, "train_loss_llm": 0.002255946397781372, "grad_norm": 0.04691747948527336, "global_step": 5440, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04245426133275032, "train_loss_bc": 0.0424351766705513, "train_loss_llm": 0.0019085407257080078, "grad_norm": 0.022950738668441772, "global_step": 5441, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04477306455373764, "train_loss_bc": 0.04477306455373764, "train_loss_llm": 0.0, "grad_norm": 0.03524152934551239, "global_step": 5442, "epoch": 151, "lr": 0.001} +{"train_loss": 0.03404967486858368, "train_loss_bc": 0.03404967486858368, "train_loss_llm": 0.0, "grad_norm": 0.04554006829857826, "global_step": 5443, "epoch": 151, "lr": 0.001} +{"train_loss": 0.045356251299381256, "train_loss_bc": 0.045356251299381256, "train_loss_llm": 0.0, "grad_norm": 0.05390723794698715, "global_step": 5444, "epoch": 151, "lr": 0.001} +{"train_loss": 0.03568817302584648, "train_loss_bc": 0.035680871456861496, "train_loss_llm": 0.0007302761077880859, "grad_norm": 0.018145838752388954, "global_step": 5445, "epoch": 151, "lr": 0.001} +{"train_loss": 0.043510597199201584, "train_loss_bc": 0.043510597199201584, "train_loss_llm": 0.0, "grad_norm": 0.03106297180056572, "global_step": 5446, "epoch": 151, "lr": 0.001} +{"train_loss": 0.03881299868226051, "train_loss_bc": 0.03881299868226051, "train_loss_llm": 0.0, "grad_norm": 0.03999389708042145, "global_step": 5447, "epoch": 151, "lr": 0.001} +{"train_loss": 0.041843120008707047, "train_loss_bc": 0.041819121688604355, "train_loss_llm": 0.0023996829986572266, "grad_norm": 0.046724818646907806, "global_step": 5448, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04239051789045334, "train_loss_bc": 0.04239051789045334, "train_loss_llm": 0.0, "grad_norm": 0.01672941818833351, "global_step": 5449, "epoch": 151, "lr": 0.001} +{"train_loss": 0.047543853521347046, "train_loss_bc": 0.047535642981529236, "train_loss_llm": 0.000821232795715332, "grad_norm": 0.02944178320467472, "global_step": 5450, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04451478645205498, "train_loss_bc": 0.04451478645205498, "train_loss_llm": 0.0, "grad_norm": 0.03637406975030899, "global_step": 5451, "epoch": 151, "lr": 0.001} +{"train_loss": 0.03677194193005562, "train_loss_bc": 0.03677194193005562, "train_loss_llm": 0.0, "grad_norm": 0.04677785187959671, "global_step": 5452, "epoch": 151, "lr": 0.001} +{"train_loss": 0.027782732620835304, "train_loss_bc": 0.027764244005084038, "train_loss_llm": 0.0018487870693206787, "grad_norm": 0.013822242617607117, "global_step": 5453, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04274023324251175, "train_loss_bc": 0.04273578152060509, "train_loss_llm": 0.0004451274871826172, "grad_norm": 0.02460196428000927, "global_step": 5454, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04502188786864281, "train_loss_bc": 0.04500206932425499, "train_loss_llm": 0.0019820034503936768, "grad_norm": 0.03641539812088013, "global_step": 5455, "epoch": 151, "lr": 0.001} +{"train_loss": 0.03869018331170082, "train_loss_bc": 0.03869018331170082, "train_loss_llm": 0.0, "grad_norm": 0.033753085881471634, "global_step": 5456, "epoch": 151, "lr": 0.001} +{"train_loss": 0.040319494903087616, "train_loss_bc": 0.040294066071510315, "train_loss_llm": 0.002542853355407715, "grad_norm": 0.015373551286756992, "global_step": 5457, "epoch": 151, "lr": 0.001} +{"train_loss": 0.048845142126083374, "train_loss_bc": 0.048845142126083374, "train_loss_llm": 0.0, "grad_norm": 0.02309649996459484, "global_step": 5458, "epoch": 151, "lr": 0.001} +{"train_loss": 0.042384736239910126, "train_loss_bc": 0.042384736239910126, "train_loss_llm": 0.0, "grad_norm": 0.026484612375497818, "global_step": 5459, "epoch": 151, "lr": 0.001} +{"train_loss": 0.03630165755748749, "train_loss_bc": 0.03630165755748749, "train_loss_llm": 0.0, "grad_norm": 0.030972668901085854, "global_step": 5460, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04466784745454788, "train_loss_bc": 0.04466784745454788, "train_loss_llm": 0.0, "grad_norm": 0.019359763711690903, "global_step": 5461, "epoch": 151, "lr": 0.001} +{"train_loss": 0.036139823496341705, "train_loss_bc": 0.036139823496341705, "train_loss_llm": 0.0, "grad_norm": 0.02564396522939205, "global_step": 5462, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04781589284539223, "train_loss_bc": 0.04781589284539223, "train_loss_llm": 0.0, "grad_norm": 0.03647059202194214, "global_step": 5463, "epoch": 151, "lr": 0.001} +{"train_loss": 0.044882431626319885, "train_loss_bc": 0.044882431626319885, "train_loss_llm": 0.0, "grad_norm": 0.03859543427824974, "global_step": 5464, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04030315950512886, "train_loss_bc": 0.04030315950512886, "train_loss_llm": 0.0, "grad_norm": 0.014241211116313934, "global_step": 5465, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04080665484070778, "train_loss_bc": 0.04080665484070778, "train_loss_llm": 0.0, "grad_norm": 0.028295787051320076, "global_step": 5466, "epoch": 151, "lr": 0.001} +{"train_loss": 0.042105771601200104, "train_loss_bc": 0.04208208993077278, "train_loss_llm": 0.002368152141571045, "grad_norm": 0.04037909209728241, "global_step": 5467, "epoch": 151, "lr": 0.001} +{"train_loss": 0.03999268263578415, "train_loss_bc": 0.03999268263578415, "train_loss_llm": 0.0, "grad_norm": 0.048869021236896515, "global_step": 5468, "epoch": 151, "lr": 0.001} +{"train_loss": 0.041802361607551575, "train_loss_bc": 0.041802361607551575, "train_loss_llm": 0.0, "grad_norm": 0.019308611750602722, "global_step": 5469, "epoch": 151, "lr": 0.001} +{"train_loss": 0.043803129345178604, "train_loss_bc": 0.043803129345178604, "train_loss_llm": 0.0, "grad_norm": 0.0299240630120039, "global_step": 5470, "epoch": 151, "lr": 0.001} +{"train_loss": 0.04161759864331947, "train_loss_bc": 0.04143129289150238, "train_loss_llm": 0.0, "grad_norm": 0.04509349167346954, "global_step": 5471, "epoch": 151, "lr": 0.001, "val_loss": 0.054878007620573044} +{"train_loss": 0.04176166281104088, "train_loss_bc": 0.041758809238672256, "train_loss_llm": 0.0002853870391845703, "grad_norm": 0.05274474620819092, "global_step": 5472, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04458451271057129, "train_loss_bc": 0.04456569626927376, "train_loss_llm": 0.0018815994262695312, "grad_norm": 0.01371475588530302, "global_step": 5473, "epoch": 152, "lr": 0.001} +{"train_loss": 0.040147360414266586, "train_loss_bc": 0.040147360414266586, "train_loss_llm": 0.0, "grad_norm": 0.024541568011045456, "global_step": 5474, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04634258523583412, "train_loss_bc": 0.04634258523583412, "train_loss_llm": 0.0, "grad_norm": 0.03214050084352493, "global_step": 5475, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04238094016909599, "train_loss_bc": 0.04236797243356705, "train_loss_llm": 0.00129660964012146, "grad_norm": 0.0435129776597023, "global_step": 5476, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04106156527996063, "train_loss_bc": 0.04106156527996063, "train_loss_llm": 0.0, "grad_norm": 0.019505193457007408, "global_step": 5477, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04051006957888603, "train_loss_bc": 0.04051006957888603, "train_loss_llm": 0.0, "grad_norm": 0.024105891585350037, "global_step": 5478, "epoch": 152, "lr": 0.001} +{"train_loss": 0.0429806262254715, "train_loss_bc": 0.0429806262254715, "train_loss_llm": 0.0, "grad_norm": 0.03145318105816841, "global_step": 5479, "epoch": 152, "lr": 0.001} +{"train_loss": 0.042910829186439514, "train_loss_bc": 0.04290705919265747, "train_loss_llm": 0.0003771185874938965, "grad_norm": 0.03808668628334999, "global_step": 5480, "epoch": 152, "lr": 0.001} +{"train_loss": 0.037986207753419876, "train_loss_bc": 0.037986207753419876, "train_loss_llm": 0.0, "grad_norm": 0.01652880385518074, "global_step": 5481, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04231555014848709, "train_loss_bc": 0.04231555014848709, "train_loss_llm": 0.0, "grad_norm": 0.027620986104011536, "global_step": 5482, "epoch": 152, "lr": 0.001} +{"train_loss": 0.038390178233385086, "train_loss_bc": 0.038390152156353, "train_loss_llm": 2.652406692504883e-06, "grad_norm": 0.031631387770175934, "global_step": 5483, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04800640046596527, "train_loss_bc": 0.04800088703632355, "train_loss_llm": 0.0005514621734619141, "grad_norm": 0.04147664085030556, "global_step": 5484, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04851094260811806, "train_loss_bc": 0.04851094260811806, "train_loss_llm": 0.0, "grad_norm": 0.019233105704188347, "global_step": 5485, "epoch": 152, "lr": 0.001} +{"train_loss": 0.03722558915615082, "train_loss_bc": 0.03722558915615082, "train_loss_llm": 0.0, "grad_norm": 0.03259163722395897, "global_step": 5486, "epoch": 152, "lr": 0.001} +{"train_loss": 0.03460225090384483, "train_loss_bc": 0.03460225090384483, "train_loss_llm": 0.0, "grad_norm": 0.040854185819625854, "global_step": 5487, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04329322651028633, "train_loss_bc": 0.04329322651028633, "train_loss_llm": 0.0, "grad_norm": 0.04895840585231781, "global_step": 5488, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04901546984910965, "train_loss_bc": 0.04900399595499039, "train_loss_llm": 0.0011473298072814941, "grad_norm": 0.02205655723810196, "global_step": 5489, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04838309437036514, "train_loss_bc": 0.04838309437036514, "train_loss_llm": 0.0, "grad_norm": 0.032939109951257706, "global_step": 5490, "epoch": 152, "lr": 0.001} +{"train_loss": 0.038480695337057114, "train_loss_bc": 0.038480695337057114, "train_loss_llm": 0.0, "grad_norm": 0.04778825119137764, "global_step": 5491, "epoch": 152, "lr": 0.001} +{"train_loss": 0.042263779789209366, "train_loss_bc": 0.04226304218173027, "train_loss_llm": 7.37905502319336e-05, "grad_norm": 0.060305405408144, "global_step": 5492, "epoch": 152, "lr": 0.001} +{"train_loss": 0.03895867243409157, "train_loss_bc": 0.03895867243409157, "train_loss_llm": 0.0, "grad_norm": 0.0157770998775959, "global_step": 5493, "epoch": 152, "lr": 0.001} +{"train_loss": 0.03919312357902527, "train_loss_bc": 0.03919312357902527, "train_loss_llm": 0.0, "grad_norm": 0.02036488801240921, "global_step": 5494, "epoch": 152, "lr": 0.001} +{"train_loss": 0.045342233031988144, "train_loss_bc": 0.045342233031988144, "train_loss_llm": 0.0, "grad_norm": 0.025679942220449448, "global_step": 5495, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04002256318926811, "train_loss_bc": 0.04002256318926811, "train_loss_llm": 0.0, "grad_norm": 0.030113454908132553, "global_step": 5496, "epoch": 152, "lr": 0.001} +{"train_loss": 0.03204914927482605, "train_loss_bc": 0.03204914927482605, "train_loss_llm": 0.0, "grad_norm": 0.01833174005150795, "global_step": 5497, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04403878375887871, "train_loss_bc": 0.04403878375887871, "train_loss_llm": 0.0, "grad_norm": 0.02866472490131855, "global_step": 5498, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04019571468234062, "train_loss_bc": 0.04019571468234062, "train_loss_llm": 0.0, "grad_norm": 0.0381024032831192, "global_step": 5499, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04972076788544655, "train_loss_bc": 0.04972076788544655, "train_loss_llm": 0.0, "grad_norm": 0.056479983031749725, "global_step": 5500, "epoch": 152, "lr": 0.001} +{"train_loss": 0.03481058403849602, "train_loss_bc": 0.03481058403849602, "train_loss_llm": 0.0, "grad_norm": 0.018260588869452477, "global_step": 5501, "epoch": 152, "lr": 0.001} +{"train_loss": 0.0441308319568634, "train_loss_bc": 0.04412010312080383, "train_loss_llm": 0.0010730624198913574, "grad_norm": 0.02804875187575817, "global_step": 5502, "epoch": 152, "lr": 0.001} +{"train_loss": 0.044194627553224564, "train_loss_bc": 0.04418224096298218, "train_loss_llm": 0.0012385547161102295, "grad_norm": 0.03636592999100685, "global_step": 5503, "epoch": 152, "lr": 0.001} +{"train_loss": 0.035793449729681015, "train_loss_bc": 0.03579090163111687, "train_loss_llm": 0.000254899263381958, "grad_norm": 0.04437071830034256, "global_step": 5504, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04669513925909996, "train_loss_bc": 0.04669513925909996, "train_loss_llm": 0.0, "grad_norm": 0.01820523291826248, "global_step": 5505, "epoch": 152, "lr": 0.001} +{"train_loss": 0.0359673835337162, "train_loss_bc": 0.0359673835337162, "train_loss_llm": 0.0, "grad_norm": 0.020797811448574066, "global_step": 5506, "epoch": 152, "lr": 0.001} +{"train_loss": 0.04152105583084954, "train_loss_bc": 0.03248485550284386, "train_loss_llm": 0.0006592273712158203, "grad_norm": 0.029328063130378723, "global_step": 5507, "epoch": 152, "lr": 0.001, "val_loss": 0.054745107889175415} +{"train_loss": 0.04016236588358879, "train_loss_bc": 0.04016236588358879, "train_loss_llm": 0.0, "grad_norm": 0.03388599678874016, "global_step": 5508, "epoch": 153, "lr": 0.001} +{"train_loss": 0.048397962003946304, "train_loss_bc": 0.048369575291872025, "train_loss_llm": 0.0028384923934936523, "grad_norm": 0.0193189550191164, "global_step": 5509, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04532667249441147, "train_loss_bc": 0.04532667249441147, "train_loss_llm": 0.0, "grad_norm": 0.023188471794128418, "global_step": 5510, "epoch": 153, "lr": 0.001} +{"train_loss": 0.0411730632185936, "train_loss_bc": 0.0411730632185936, "train_loss_llm": 0.0, "grad_norm": 0.03513707220554352, "global_step": 5511, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04780209809541702, "train_loss_bc": 0.04780209809541702, "train_loss_llm": 0.0, "grad_norm": 0.03871948644518852, "global_step": 5512, "epoch": 153, "lr": 0.001} +{"train_loss": 0.03884885460138321, "train_loss_bc": 0.03884885460138321, "train_loss_llm": 0.0, "grad_norm": 0.016508523374795914, "global_step": 5513, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04040927067399025, "train_loss_bc": 0.04040927067399025, "train_loss_llm": 0.0, "grad_norm": 0.03171422332525253, "global_step": 5514, "epoch": 153, "lr": 0.001} +{"train_loss": 0.0443895049393177, "train_loss_bc": 0.0443895049393177, "train_loss_llm": 0.0, "grad_norm": 0.03699969872832298, "global_step": 5515, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04212915152311325, "train_loss_bc": 0.04212915152311325, "train_loss_llm": 0.0, "grad_norm": 0.03657186031341553, "global_step": 5516, "epoch": 153, "lr": 0.001} +{"train_loss": 0.044476669281721115, "train_loss_bc": 0.044476669281721115, "train_loss_llm": 0.0, "grad_norm": 0.017690517008304596, "global_step": 5517, "epoch": 153, "lr": 0.001} +{"train_loss": 0.037426289170980453, "train_loss_bc": 0.037426289170980453, "train_loss_llm": 0.0, "grad_norm": 0.032242514193058014, "global_step": 5518, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04405573010444641, "train_loss_bc": 0.04404069855809212, "train_loss_llm": 0.0015032291412353516, "grad_norm": 0.04314970597624779, "global_step": 5519, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04445032402873039, "train_loss_bc": 0.04445032402873039, "train_loss_llm": 0.0, "grad_norm": 0.043819814920425415, "global_step": 5520, "epoch": 153, "lr": 0.001} +{"train_loss": 0.043324124068021774, "train_loss_bc": 0.043324124068021774, "train_loss_llm": 0.0, "grad_norm": 0.018367452546954155, "global_step": 5521, "epoch": 153, "lr": 0.001} +{"train_loss": 0.047905758023262024, "train_loss_bc": 0.047905758023262024, "train_loss_llm": 0.0, "grad_norm": 0.02541753649711609, "global_step": 5522, "epoch": 153, "lr": 0.001} +{"train_loss": 0.034373000264167786, "train_loss_bc": 0.034366361796855927, "train_loss_llm": 0.0006639659404754639, "grad_norm": 0.038388386368751526, "global_step": 5523, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04065938666462898, "train_loss_bc": 0.04065938666462898, "train_loss_llm": 0.0, "grad_norm": 0.045176029205322266, "global_step": 5524, "epoch": 153, "lr": 0.001} +{"train_loss": 0.037614330649375916, "train_loss_bc": 0.037599317729473114, "train_loss_llm": 0.0015012621879577637, "grad_norm": 0.01554409135133028, "global_step": 5525, "epoch": 153, "lr": 0.001} +{"train_loss": 0.03725733980536461, "train_loss_bc": 0.037244394421577454, "train_loss_llm": 0.0012946724891662598, "grad_norm": 0.027599086984992027, "global_step": 5526, "epoch": 153, "lr": 0.001} +{"train_loss": 0.036036085337400436, "train_loss_bc": 0.03603334724903107, "train_loss_llm": 0.0002739429473876953, "grad_norm": 0.036682963371276855, "global_step": 5527, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04137951880693436, "train_loss_bc": 0.04137614369392395, "train_loss_llm": 0.0003375113010406494, "grad_norm": 0.0444549061357975, "global_step": 5528, "epoch": 153, "lr": 0.001} +{"train_loss": 0.03921116143465042, "train_loss_bc": 0.03921116143465042, "train_loss_llm": 0.0, "grad_norm": 0.014444402419030666, "global_step": 5529, "epoch": 153, "lr": 0.001} +{"train_loss": 0.042952269315719604, "train_loss_bc": 0.042944878339767456, "train_loss_llm": 0.0007391870021820068, "grad_norm": 0.022657260298728943, "global_step": 5530, "epoch": 153, "lr": 0.001} +{"train_loss": 0.03948687016963959, "train_loss_bc": 0.03948687016963959, "train_loss_llm": 0.0, "grad_norm": 0.029980409890413284, "global_step": 5531, "epoch": 153, "lr": 0.001} +{"train_loss": 0.03949357569217682, "train_loss_bc": 0.039490584284067154, "train_loss_llm": 0.00029906630516052246, "grad_norm": 0.038743115961551666, "global_step": 5532, "epoch": 153, "lr": 0.001} +{"train_loss": 0.036741770803928375, "train_loss_bc": 0.036741770803928375, "train_loss_llm": 0.0, "grad_norm": 0.015497318468987942, "global_step": 5533, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04045700654387474, "train_loss_bc": 0.04045700654387474, "train_loss_llm": 0.0, "grad_norm": 0.021499693393707275, "global_step": 5534, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04761675372719765, "train_loss_bc": 0.04761675372719765, "train_loss_llm": 0.0, "grad_norm": 0.032240044325590134, "global_step": 5535, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04186577349901199, "train_loss_bc": 0.04186577349901199, "train_loss_llm": 0.0, "grad_norm": 0.038023028522729874, "global_step": 5536, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04557998850941658, "train_loss_bc": 0.0455789640545845, "train_loss_llm": 0.00010246038436889648, "grad_norm": 0.01868627406656742, "global_step": 5537, "epoch": 153, "lr": 0.001} +{"train_loss": 0.040494710206985474, "train_loss_bc": 0.040487512946128845, "train_loss_llm": 0.0007198154926300049, "grad_norm": 0.02462262287735939, "global_step": 5538, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04104125499725342, "train_loss_bc": 0.04101106896996498, "train_loss_llm": 0.003018587827682495, "grad_norm": 0.028907237574458122, "global_step": 5539, "epoch": 153, "lr": 0.001} +{"train_loss": 0.04796502739191055, "train_loss_bc": 0.04796502739191055, "train_loss_llm": 0.0, "grad_norm": 0.03189592435956001, "global_step": 5540, "epoch": 153, "lr": 0.001} +{"train_loss": 0.032509297132492065, "train_loss_bc": 0.03248993679881096, "train_loss_llm": 0.0019359290599822998, "grad_norm": 0.01718986965715885, "global_step": 5541, "epoch": 153, "lr": 0.001} +{"train_loss": 0.03821524232625961, "train_loss_bc": 0.03821524232625961, "train_loss_llm": 0.0, "grad_norm": 0.02374989353120327, "global_step": 5542, "epoch": 153, "lr": 0.001} +{"train_loss": 0.041532740307350956, "train_loss_bc": 0.04395044967532158, "train_loss_llm": 0.0, "grad_norm": 0.038120221346616745, "global_step": 5543, "epoch": 153, "lr": 0.001, "val_loss": 0.05484993755817413} +{"train_loss": 0.03721514344215393, "train_loss_bc": 0.03721514344215393, "train_loss_llm": 0.0, "grad_norm": 0.041965339332818985, "global_step": 5544, "epoch": 154, "lr": 0.001} +{"train_loss": 0.044757992029190063, "train_loss_bc": 0.044743482023477554, "train_loss_llm": 0.0014510154724121094, "grad_norm": 0.018782777711749077, "global_step": 5545, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04503006488084793, "train_loss_bc": 0.04503006488084793, "train_loss_llm": 0.0, "grad_norm": 0.021710652858018875, "global_step": 5546, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04112815111875534, "train_loss_bc": 0.04112815111875534, "train_loss_llm": 0.0, "grad_norm": 0.028523730114102364, "global_step": 5547, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04326656088232994, "train_loss_bc": 0.04324987530708313, "train_loss_llm": 0.001668691635131836, "grad_norm": 0.03954552859067917, "global_step": 5548, "epoch": 154, "lr": 0.001} +{"train_loss": 0.03675958514213562, "train_loss_bc": 0.03675958514213562, "train_loss_llm": 0.0, "grad_norm": 0.015609251335263252, "global_step": 5549, "epoch": 154, "lr": 0.001} +{"train_loss": 0.044759850949048996, "train_loss_bc": 0.044759850949048996, "train_loss_llm": 0.0, "grad_norm": 0.022699758410453796, "global_step": 5550, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04072408378124237, "train_loss_bc": 0.04072408378124237, "train_loss_llm": 0.0, "grad_norm": 0.029108962044119835, "global_step": 5551, "epoch": 154, "lr": 0.001} +{"train_loss": 0.03845009580254555, "train_loss_bc": 0.03845009580254555, "train_loss_llm": 0.0, "grad_norm": 0.035081248730421066, "global_step": 5552, "epoch": 154, "lr": 0.001} +{"train_loss": 0.041725896298885345, "train_loss_bc": 0.04169556125998497, "train_loss_llm": 0.0030333399772644043, "grad_norm": 0.01608698070049286, "global_step": 5553, "epoch": 154, "lr": 0.001} +{"train_loss": 0.046740494668483734, "train_loss_bc": 0.04673707112669945, "train_loss_llm": 0.00034242868423461914, "grad_norm": 0.020701227709650993, "global_step": 5554, "epoch": 154, "lr": 0.001} +{"train_loss": 0.03415244445204735, "train_loss_bc": 0.034127429127693176, "train_loss_llm": 0.0025014877319335938, "grad_norm": 0.023131458088755608, "global_step": 5555, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04313237965106964, "train_loss_bc": 0.04313237965106964, "train_loss_llm": 0.0, "grad_norm": 0.030084848403930664, "global_step": 5556, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04361987113952637, "train_loss_bc": 0.0436183400452137, "train_loss_llm": 0.0001532435417175293, "grad_norm": 0.015588052570819855, "global_step": 5557, "epoch": 154, "lr": 0.001} +{"train_loss": 0.03863861784338951, "train_loss_bc": 0.03863861784338951, "train_loss_llm": 0.0, "grad_norm": 0.02076873928308487, "global_step": 5558, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04806021228432655, "train_loss_bc": 0.04806021228432655, "train_loss_llm": 0.0, "grad_norm": 0.02823699824512005, "global_step": 5559, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04178963601589203, "train_loss_bc": 0.04178963601589203, "train_loss_llm": 0.0, "grad_norm": 0.035893965512514114, "global_step": 5560, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04400206729769707, "train_loss_bc": 0.04400206729769707, "train_loss_llm": 0.0, "grad_norm": 0.020838826894760132, "global_step": 5561, "epoch": 154, "lr": 0.001} +{"train_loss": 0.0384305976331234, "train_loss_bc": 0.0384305976331234, "train_loss_llm": 0.0, "grad_norm": 0.02851485274732113, "global_step": 5562, "epoch": 154, "lr": 0.001} +{"train_loss": 0.03986106440424919, "train_loss_bc": 0.03986106440424919, "train_loss_llm": 0.0, "grad_norm": 0.03548602759838104, "global_step": 5563, "epoch": 154, "lr": 0.001} +{"train_loss": 0.03737834095954895, "train_loss_bc": 0.03737031668424606, "train_loss_llm": 0.0008022785186767578, "grad_norm": 0.03710047900676727, "global_step": 5564, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04369457811117172, "train_loss_bc": 0.04369457811117172, "train_loss_llm": 0.0, "grad_norm": 0.02053472399711609, "global_step": 5565, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04135926067829132, "train_loss_bc": 0.04134327545762062, "train_loss_llm": 0.0015985369682312012, "grad_norm": 0.0306258462369442, "global_step": 5566, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04264524206519127, "train_loss_bc": 0.04264524206519127, "train_loss_llm": 0.0, "grad_norm": 0.049645792692899704, "global_step": 5567, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04410061612725258, "train_loss_bc": 0.04410061612725258, "train_loss_llm": 0.0, "grad_norm": 0.058619871735572815, "global_step": 5568, "epoch": 154, "lr": 0.001} +{"train_loss": 0.03970048204064369, "train_loss_bc": 0.03970048204064369, "train_loss_llm": 0.0, "grad_norm": 0.024191154167056084, "global_step": 5569, "epoch": 154, "lr": 0.001} +{"train_loss": 0.039079442620277405, "train_loss_bc": 0.039079442620277405, "train_loss_llm": 0.0, "grad_norm": 0.029320182278752327, "global_step": 5570, "epoch": 154, "lr": 0.001} +{"train_loss": 0.03980682045221329, "train_loss_bc": 0.03980682045221329, "train_loss_llm": 0.0, "grad_norm": 0.04295899718999863, "global_step": 5571, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04267364740371704, "train_loss_bc": 0.04267364740371704, "train_loss_llm": 0.0, "grad_norm": 0.052735716104507446, "global_step": 5572, "epoch": 154, "lr": 0.001} +{"train_loss": 0.033237434923648834, "train_loss_bc": 0.033237434923648834, "train_loss_llm": 0.0, "grad_norm": 0.016013499349355698, "global_step": 5573, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04105941951274872, "train_loss_bc": 0.04105941951274872, "train_loss_llm": 0.0, "grad_norm": 0.020682739093899727, "global_step": 5574, "epoch": 154, "lr": 0.001} +{"train_loss": 0.045356784015893936, "train_loss_bc": 0.045356784015893936, "train_loss_llm": 0.0, "grad_norm": 0.030174657702445984, "global_step": 5575, "epoch": 154, "lr": 0.001} +{"train_loss": 0.0426761656999588, "train_loss_bc": 0.0426761656999588, "train_loss_llm": 0.0, "grad_norm": 0.037817150354385376, "global_step": 5576, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04108899459242821, "train_loss_bc": 0.04108456149697304, "train_loss_llm": 0.00044342875480651855, "grad_norm": 0.017758836969733238, "global_step": 5577, "epoch": 154, "lr": 0.001} +{"train_loss": 0.05127573013305664, "train_loss_bc": 0.051236219704151154, "train_loss_llm": 0.003950893878936768, "grad_norm": 0.030960069969296455, "global_step": 5578, "epoch": 154, "lr": 0.001} +{"train_loss": 0.04146832714064254, "train_loss_bc": 0.03548083454370499, "train_loss_llm": 0.00011739134788513184, "grad_norm": 0.042537230998277664, "global_step": 5579, "epoch": 154, "lr": 0.001, "val_loss": 0.05477292463183403} +{"train_loss": 0.042821355164051056, "train_loss_bc": 0.04281826317310333, "train_loss_llm": 0.0003090500831604004, "grad_norm": 0.04595423489809036, "global_step": 5580, "epoch": 155, "lr": 0.001} +{"train_loss": 0.040407806634902954, "train_loss_bc": 0.040404822677373886, "train_loss_llm": 0.00029844045639038086, "grad_norm": 0.018412986770272255, "global_step": 5581, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04113871231675148, "train_loss_bc": 0.041127853095531464, "train_loss_llm": 0.0010860264301300049, "grad_norm": 0.025245195254683495, "global_step": 5582, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04039151966571808, "train_loss_bc": 0.04037068039178848, "train_loss_llm": 0.00208398699760437, "grad_norm": 0.03488907217979431, "global_step": 5583, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04072834178805351, "train_loss_bc": 0.0407194048166275, "train_loss_llm": 0.0008936822414398193, "grad_norm": 0.03567370027303696, "global_step": 5584, "epoch": 155, "lr": 0.001} +{"train_loss": 0.040442030876874924, "train_loss_bc": 0.04043874144554138, "train_loss_llm": 0.00032892823219299316, "grad_norm": 0.019728271290659904, "global_step": 5585, "epoch": 155, "lr": 0.001} +{"train_loss": 0.039569925516843796, "train_loss_bc": 0.039569925516843796, "train_loss_llm": 0.0, "grad_norm": 0.034380100667476654, "global_step": 5586, "epoch": 155, "lr": 0.001} +{"train_loss": 0.03456277772784233, "train_loss_bc": 0.03456277772784233, "train_loss_llm": 0.0, "grad_norm": 0.03754255548119545, "global_step": 5587, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04308154806494713, "train_loss_bc": 0.043069206178188324, "train_loss_llm": 0.0012343525886535645, "grad_norm": 0.04412112012505531, "global_step": 5588, "epoch": 155, "lr": 0.001} +{"train_loss": 0.043405693024396896, "train_loss_bc": 0.043405693024396896, "train_loss_llm": 0.0, "grad_norm": 0.021635213866829872, "global_step": 5589, "epoch": 155, "lr": 0.001} +{"train_loss": 0.0464082807302475, "train_loss_bc": 0.0464082807302475, "train_loss_llm": 0.0, "grad_norm": 0.027109608054161072, "global_step": 5590, "epoch": 155, "lr": 0.001} +{"train_loss": 0.03908136487007141, "train_loss_bc": 0.03908136487007141, "train_loss_llm": 0.0, "grad_norm": 0.03818555921316147, "global_step": 5591, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04688294976949692, "train_loss_bc": 0.04688294976949692, "train_loss_llm": 0.0, "grad_norm": 0.04359365254640579, "global_step": 5592, "epoch": 155, "lr": 0.001} +{"train_loss": 0.03693506494164467, "train_loss_bc": 0.03692404553294182, "train_loss_llm": 0.001102060079574585, "grad_norm": 0.01417606882750988, "global_step": 5593, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04858558252453804, "train_loss_bc": 0.04858558252453804, "train_loss_llm": 0.0, "grad_norm": 0.02623612806200981, "global_step": 5594, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04165732115507126, "train_loss_bc": 0.04165732115507126, "train_loss_llm": 0.0, "grad_norm": 0.03961827605962753, "global_step": 5595, "epoch": 155, "lr": 0.001} +{"train_loss": 0.046144820749759674, "train_loss_bc": 0.04614043980836868, "train_loss_llm": 0.0004379451274871826, "grad_norm": 0.04667295515537262, "global_step": 5596, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04645393416285515, "train_loss_bc": 0.04645300284028053, "train_loss_llm": 9.295344352722168e-05, "grad_norm": 0.01895991526544094, "global_step": 5597, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04653404653072357, "train_loss_bc": 0.04653404653072357, "train_loss_llm": 0.0, "grad_norm": 0.03688573092222214, "global_step": 5598, "epoch": 155, "lr": 0.001} +{"train_loss": 0.03750932961702347, "train_loss_bc": 0.03750932961702347, "train_loss_llm": 0.0, "grad_norm": 0.04418172687292099, "global_step": 5599, "epoch": 155, "lr": 0.001} +{"train_loss": 0.03298609331250191, "train_loss_bc": 0.03298609331250191, "train_loss_llm": 0.0, "grad_norm": 0.046161096543073654, "global_step": 5600, "epoch": 155, "lr": 0.001} +{"train_loss": 0.03915051370859146, "train_loss_bc": 0.039143018424510956, "train_loss_llm": 0.0007496774196624756, "grad_norm": 0.018093153834342957, "global_step": 5601, "epoch": 155, "lr": 0.001} +{"train_loss": 0.0446036122739315, "train_loss_bc": 0.0446036122739315, "train_loss_llm": 0.0, "grad_norm": 0.030319761484861374, "global_step": 5602, "epoch": 155, "lr": 0.001} +{"train_loss": 0.03526751697063446, "train_loss_bc": 0.03526751697063446, "train_loss_llm": 0.0, "grad_norm": 0.04065103083848953, "global_step": 5603, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04166955500841141, "train_loss_bc": 0.04166955500841141, "train_loss_llm": 0.0, "grad_norm": 0.045372091233730316, "global_step": 5604, "epoch": 155, "lr": 0.001} +{"train_loss": 0.03908500447869301, "train_loss_bc": 0.03908500447869301, "train_loss_llm": 0.0, "grad_norm": 0.020925723016262054, "global_step": 5605, "epoch": 155, "lr": 0.001} +{"train_loss": 0.03929771110415459, "train_loss_bc": 0.03929771110415459, "train_loss_llm": 0.0, "grad_norm": 0.0357656329870224, "global_step": 5606, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04094897583127022, "train_loss_bc": 0.04094897583127022, "train_loss_llm": 0.0, "grad_norm": 0.04121578484773636, "global_step": 5607, "epoch": 155, "lr": 0.001} +{"train_loss": 0.042227212339639664, "train_loss_bc": 0.042227212339639664, "train_loss_llm": 0.0, "grad_norm": 0.05422902852296829, "global_step": 5608, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04005032405257225, "train_loss_bc": 0.04005032405257225, "train_loss_llm": 0.0, "grad_norm": 0.01851530186831951, "global_step": 5609, "epoch": 155, "lr": 0.001} +{"train_loss": 0.043093472719192505, "train_loss_bc": 0.043093472719192505, "train_loss_llm": 0.0, "grad_norm": 0.02457463927567005, "global_step": 5610, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04382704943418503, "train_loss_bc": 0.04382704943418503, "train_loss_llm": 0.0, "grad_norm": 0.032505929470062256, "global_step": 5611, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04110059514641762, "train_loss_bc": 0.04110059514641762, "train_loss_llm": 0.0, "grad_norm": 0.039271049201488495, "global_step": 5612, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04350954294204712, "train_loss_bc": 0.04350954294204712, "train_loss_llm": 0.0, "grad_norm": 0.014361284673213959, "global_step": 5613, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04322358965873718, "train_loss_bc": 0.04322358965873718, "train_loss_llm": 0.0, "grad_norm": 0.024625319987535477, "global_step": 5614, "epoch": 155, "lr": 0.001} +{"train_loss": 0.04149254918512371, "train_loss_bc": 0.040948595851659775, "train_loss_llm": 0.0, "grad_norm": 0.03454785421490669, "global_step": 5615, "epoch": 155, "lr": 0.001, "train/cumulative_reward": 103.3456450646951, "train/mean_score": 17.83799680538559, "train/success_rate": 0.4, "test/cumulative_reward": 75.52691737246916, "test/mean_score": 9.56224216473018, "test/success_rate": 0.22, "val_loss": 0.05486937612295151, "train_action_mse_error": 0.04480975866317749} +{"train_loss": 0.04067908599972725, "train_loss_bc": 0.04065949469804764, "train_loss_llm": 0.001958996057510376, "grad_norm": 0.0412910021841526, "global_step": 5616, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04141395539045334, "train_loss_bc": 0.04137679561972618, "train_loss_llm": 0.0037158429622650146, "grad_norm": 0.017969490960240364, "global_step": 5617, "epoch": 156, "lr": 0.001} +{"train_loss": 0.03793974593281746, "train_loss_bc": 0.0379263237118721, "train_loss_llm": 0.001342177391052246, "grad_norm": 0.024033909663558006, "global_step": 5618, "epoch": 156, "lr": 0.001} +{"train_loss": 0.039803631603717804, "train_loss_bc": 0.03979189321398735, "train_loss_llm": 0.0011737942695617676, "grad_norm": 0.03221359848976135, "global_step": 5619, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04031813517212868, "train_loss_bc": 0.040282443165779114, "train_loss_llm": 0.0035690665245056152, "grad_norm": 0.0369388721883297, "global_step": 5620, "epoch": 156, "lr": 0.001} +{"train_loss": 0.03903277963399887, "train_loss_bc": 0.03900884464383125, "train_loss_llm": 0.0023934543132781982, "grad_norm": 0.020652692764997482, "global_step": 5621, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04340536892414093, "train_loss_bc": 0.04340536892414093, "train_loss_llm": 0.0, "grad_norm": 0.0332329198718071, "global_step": 5622, "epoch": 156, "lr": 0.001} +{"train_loss": 0.048554107546806335, "train_loss_bc": 0.048554107546806335, "train_loss_llm": 0.0, "grad_norm": 0.04676690697669983, "global_step": 5623, "epoch": 156, "lr": 0.001} +{"train_loss": 0.042695552110672, "train_loss_bc": 0.042686767876148224, "train_loss_llm": 0.0008783340454101562, "grad_norm": 0.04900338500738144, "global_step": 5624, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04289460554718971, "train_loss_bc": 0.04289460554718971, "train_loss_llm": 0.0, "grad_norm": 0.01931147277355194, "global_step": 5625, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04289240390062332, "train_loss_bc": 0.04289240390062332, "train_loss_llm": 0.0, "grad_norm": 0.03267841786146164, "global_step": 5626, "epoch": 156, "lr": 0.001} +{"train_loss": 0.03660695627331734, "train_loss_bc": 0.036577172577381134, "train_loss_llm": 0.002978384494781494, "grad_norm": 0.03955288603901863, "global_step": 5627, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04270433634519577, "train_loss_bc": 0.04270433634519577, "train_loss_llm": 0.0, "grad_norm": 0.04897719621658325, "global_step": 5628, "epoch": 156, "lr": 0.001} +{"train_loss": 0.043149080127477646, "train_loss_bc": 0.043149080127477646, "train_loss_llm": 0.0, "grad_norm": 0.01638335920870304, "global_step": 5629, "epoch": 156, "lr": 0.001} +{"train_loss": 0.051239289343357086, "train_loss_bc": 0.051239289343357086, "train_loss_llm": 0.0, "grad_norm": 0.02291991375386715, "global_step": 5630, "epoch": 156, "lr": 0.001} +{"train_loss": 0.03978472575545311, "train_loss_bc": 0.03978472575545311, "train_loss_llm": 0.0, "grad_norm": 0.03234259784221649, "global_step": 5631, "epoch": 156, "lr": 0.001} +{"train_loss": 0.03704338148236275, "train_loss_bc": 0.03704338148236275, "train_loss_llm": 0.0, "grad_norm": 0.03544201701879501, "global_step": 5632, "epoch": 156, "lr": 0.001} +{"train_loss": 0.03858798369765282, "train_loss_bc": 0.03856319561600685, "train_loss_llm": 0.0024789273738861084, "grad_norm": 0.011674546636641026, "global_step": 5633, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04407653585076332, "train_loss_bc": 0.04407653585076332, "train_loss_llm": 0.0, "grad_norm": 0.021623870357871056, "global_step": 5634, "epoch": 156, "lr": 0.001} +{"train_loss": 0.042078591883182526, "train_loss_bc": 0.042078591883182526, "train_loss_llm": 0.0, "grad_norm": 0.029915697872638702, "global_step": 5635, "epoch": 156, "lr": 0.001} +{"train_loss": 0.0402999147772789, "train_loss_bc": 0.0402999147772789, "train_loss_llm": 0.0, "grad_norm": 0.03460586816072464, "global_step": 5636, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04018421471118927, "train_loss_bc": 0.04018421471118927, "train_loss_llm": 0.0, "grad_norm": 0.01837357133626938, "global_step": 5637, "epoch": 156, "lr": 0.001} +{"train_loss": 0.034553781151771545, "train_loss_bc": 0.034553781151771545, "train_loss_llm": 0.0, "grad_norm": 0.03135836496949196, "global_step": 5638, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04013419523835182, "train_loss_bc": 0.04013274610042572, "train_loss_llm": 0.0001450181007385254, "grad_norm": 0.042988698929548264, "global_step": 5639, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04628756269812584, "train_loss_bc": 0.04628756269812584, "train_loss_llm": 0.0, "grad_norm": 0.041331946849823, "global_step": 5640, "epoch": 156, "lr": 0.001} +{"train_loss": 0.045603733509778976, "train_loss_bc": 0.045603733509778976, "train_loss_llm": 0.0, "grad_norm": 0.02136281505227089, "global_step": 5641, "epoch": 156, "lr": 0.001} +{"train_loss": 0.041599027812480927, "train_loss_bc": 0.041599027812480927, "train_loss_llm": 0.0, "grad_norm": 0.022580541670322418, "global_step": 5642, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04804915189743042, "train_loss_bc": 0.04804915189743042, "train_loss_llm": 0.0, "grad_norm": 0.030673230066895485, "global_step": 5643, "epoch": 156, "lr": 0.001} +{"train_loss": 0.043022144585847855, "train_loss_bc": 0.043022144585847855, "train_loss_llm": 0.0, "grad_norm": 0.036646001040935516, "global_step": 5644, "epoch": 156, "lr": 0.001} +{"train_loss": 0.036395177245140076, "train_loss_bc": 0.036395177245140076, "train_loss_llm": 0.0, "grad_norm": 0.012420852668583393, "global_step": 5645, "epoch": 156, "lr": 0.001} +{"train_loss": 0.03741516172885895, "train_loss_bc": 0.037391532212495804, "train_loss_llm": 0.0023629069328308105, "grad_norm": 0.020451825112104416, "global_step": 5646, "epoch": 156, "lr": 0.001} +{"train_loss": 0.03373453766107559, "train_loss_bc": 0.03373453766107559, "train_loss_llm": 0.0, "grad_norm": 0.021545253694057465, "global_step": 5647, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04122314602136612, "train_loss_bc": 0.04122314602136612, "train_loss_llm": 0.0, "grad_norm": 0.02637961134314537, "global_step": 5648, "epoch": 156, "lr": 0.001} +{"train_loss": 0.035926785320043564, "train_loss_bc": 0.035926785320043564, "train_loss_llm": 0.0, "grad_norm": 0.014977570623159409, "global_step": 5649, "epoch": 156, "lr": 0.001} +{"train_loss": 0.04264703765511513, "train_loss_bc": 0.042610447853803635, "train_loss_llm": 0.003658860921859741, "grad_norm": 0.02438041754066944, "global_step": 5650, "epoch": 156, "lr": 0.001} +{"train_loss": 0.041468496744831405, "train_loss_bc": 0.05088447779417038, "train_loss_llm": 0.0005579888820648193, "grad_norm": 0.033703628927469254, "global_step": 5651, "epoch": 156, "lr": 0.001, "val_loss": 0.055279508233070374} +{"train_loss": 0.027012474834918976, "train_loss_bc": 0.027012474834918976, "train_loss_llm": 0.0, "grad_norm": 0.036898307502269745, "global_step": 5652, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04247428476810455, "train_loss_bc": 0.04247428476810455, "train_loss_llm": 0.0, "grad_norm": 0.022986626252532005, "global_step": 5653, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04514424130320549, "train_loss_bc": 0.04514424130320549, "train_loss_llm": 0.0, "grad_norm": 0.03402745723724365, "global_step": 5654, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04518304392695427, "train_loss_bc": 0.04517989233136177, "train_loss_llm": 0.0003151893615722656, "grad_norm": 0.035528235137462616, "global_step": 5655, "epoch": 157, "lr": 0.001} +{"train_loss": 0.03998669236898422, "train_loss_bc": 0.039986684918403625, "train_loss_llm": 6.258487701416016e-07, "grad_norm": 0.037682417780160904, "global_step": 5656, "epoch": 157, "lr": 0.001} +{"train_loss": 0.03963523358106613, "train_loss_bc": 0.039631735533475876, "train_loss_llm": 0.00034990906715393066, "grad_norm": 0.016114244237542152, "global_step": 5657, "epoch": 157, "lr": 0.001} +{"train_loss": 0.047592196613550186, "train_loss_bc": 0.047592196613550186, "train_loss_llm": 0.0, "grad_norm": 0.020901178941130638, "global_step": 5658, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04898236319422722, "train_loss_bc": 0.04898236319422722, "train_loss_llm": 0.0, "grad_norm": 0.031843654811382294, "global_step": 5659, "epoch": 157, "lr": 0.001} +{"train_loss": 0.03855818882584572, "train_loss_bc": 0.03855818882584572, "train_loss_llm": 0.0, "grad_norm": 0.036494042724370956, "global_step": 5660, "epoch": 157, "lr": 0.001} +{"train_loss": 0.0422200933098793, "train_loss_bc": 0.04218273237347603, "train_loss_llm": 0.003736168146133423, "grad_norm": 0.017316946759819984, "global_step": 5661, "epoch": 157, "lr": 0.001} +{"train_loss": 0.03306169807910919, "train_loss_bc": 0.03306169807910919, "train_loss_llm": 0.0, "grad_norm": 0.021333297714591026, "global_step": 5662, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04061656817793846, "train_loss_bc": 0.04059835150837898, "train_loss_llm": 0.0018217861652374268, "grad_norm": 0.026451872661709785, "global_step": 5663, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04466457664966583, "train_loss_bc": 0.04466457664966583, "train_loss_llm": 0.0, "grad_norm": 0.027735821902751923, "global_step": 5664, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04485279321670532, "train_loss_bc": 0.04485279321670532, "train_loss_llm": 0.0, "grad_norm": 0.012915244325995445, "global_step": 5665, "epoch": 157, "lr": 0.001} +{"train_loss": 0.03875439986586571, "train_loss_bc": 0.03875439986586571, "train_loss_llm": 0.0, "grad_norm": 0.023872442543506622, "global_step": 5666, "epoch": 157, "lr": 0.001} +{"train_loss": 0.03586846962571144, "train_loss_bc": 0.03586846962571144, "train_loss_llm": 0.0, "grad_norm": 0.02830170840024948, "global_step": 5667, "epoch": 157, "lr": 0.001} +{"train_loss": 0.044914864003658295, "train_loss_bc": 0.044914864003658295, "train_loss_llm": 0.0, "grad_norm": 0.03484789654612541, "global_step": 5668, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04166314750909805, "train_loss_bc": 0.04166314750909805, "train_loss_llm": 0.0, "grad_norm": 0.018374789506196976, "global_step": 5669, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04243812337517738, "train_loss_bc": 0.04243812337517738, "train_loss_llm": 0.0, "grad_norm": 0.02269311621785164, "global_step": 5670, "epoch": 157, "lr": 0.001} +{"train_loss": 0.03608538582921028, "train_loss_bc": 0.03606528788805008, "train_loss_llm": 0.0020098090171813965, "grad_norm": 0.030322572216391563, "global_step": 5671, "epoch": 157, "lr": 0.001} +{"train_loss": 0.039929624646902084, "train_loss_bc": 0.039929624646902084, "train_loss_llm": 0.0, "grad_norm": 0.03326283022761345, "global_step": 5672, "epoch": 157, "lr": 0.001} +{"train_loss": 0.0402715764939785, "train_loss_bc": 0.0402715764939785, "train_loss_llm": 0.0, "grad_norm": 0.014908784069120884, "global_step": 5673, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04346048831939697, "train_loss_bc": 0.04345550388097763, "train_loss_llm": 0.0004982948303222656, "grad_norm": 0.020368024706840515, "global_step": 5674, "epoch": 157, "lr": 0.001} +{"train_loss": 0.040897805243730545, "train_loss_bc": 0.040897805243730545, "train_loss_llm": 0.0, "grad_norm": 0.02914486452937126, "global_step": 5675, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04565642029047012, "train_loss_bc": 0.045637521892786026, "train_loss_llm": 0.0018897652626037598, "grad_norm": 0.03280598670244217, "global_step": 5676, "epoch": 157, "lr": 0.001} +{"train_loss": 0.037183914333581924, "train_loss_bc": 0.03717586398124695, "train_loss_llm": 0.000804901123046875, "grad_norm": 0.013703878968954086, "global_step": 5677, "epoch": 157, "lr": 0.001} +{"train_loss": 0.049902066588401794, "train_loss_bc": 0.049902066588401794, "train_loss_llm": 0.0, "grad_norm": 0.02627088502049446, "global_step": 5678, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04452427476644516, "train_loss_bc": 0.04452427476644516, "train_loss_llm": 0.0, "grad_norm": 0.03319389373064041, "global_step": 5679, "epoch": 157, "lr": 0.001} +{"train_loss": 0.03878520801663399, "train_loss_bc": 0.03877617418766022, "train_loss_llm": 0.0009034574031829834, "grad_norm": 0.03791092336177826, "global_step": 5680, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04111052677035332, "train_loss_bc": 0.04111052677035332, "train_loss_llm": 0.0, "grad_norm": 0.021223559975624084, "global_step": 5681, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04095182567834854, "train_loss_bc": 0.04095182567834854, "train_loss_llm": 0.0, "grad_norm": 0.022241173312067986, "global_step": 5682, "epoch": 157, "lr": 0.001} +{"train_loss": 0.043144334107637405, "train_loss_bc": 0.0431210994720459, "train_loss_llm": 0.0023235082626342773, "grad_norm": 0.03169870376586914, "global_step": 5683, "epoch": 157, "lr": 0.001} +{"train_loss": 0.03702918440103531, "train_loss_bc": 0.03701869398355484, "train_loss_llm": 0.0010489225387573242, "grad_norm": 0.03368775174021721, "global_step": 5684, "epoch": 157, "lr": 0.001} +{"train_loss": 0.0367230549454689, "train_loss_bc": 0.03671999275684357, "train_loss_llm": 0.00030604004859924316, "grad_norm": 0.01663612760603428, "global_step": 5685, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04569249227643013, "train_loss_bc": 0.04569249227643013, "train_loss_llm": 0.0, "grad_norm": 0.025933993980288506, "global_step": 5686, "epoch": 157, "lr": 0.001} +{"train_loss": 0.04123168821550078, "train_loss_bc": 0.039369139820337296, "train_loss_llm": 0.0, "grad_norm": 0.03174793720245361, "global_step": 5687, "epoch": 157, "lr": 0.001, "val_loss": 0.05497407913208008} +{"train_loss": 0.04403432831168175, "train_loss_bc": 0.04402540624141693, "train_loss_llm": 0.0008923113346099854, "grad_norm": 0.040415652096271515, "global_step": 5688, "epoch": 158, "lr": 0.001} +{"train_loss": 0.0420372411608696, "train_loss_bc": 0.0420372411608696, "train_loss_llm": 0.0, "grad_norm": 0.013999685645103455, "global_step": 5689, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04668349027633667, "train_loss_bc": 0.04668349027633667, "train_loss_llm": 0.0, "grad_norm": 0.02044036239385605, "global_step": 5690, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03513132780790329, "train_loss_bc": 0.03513132780790329, "train_loss_llm": 0.0, "grad_norm": 0.02730056457221508, "global_step": 5691, "epoch": 158, "lr": 0.001} +{"train_loss": 0.046705324202775955, "train_loss_bc": 0.046705324202775955, "train_loss_llm": 0.0, "grad_norm": 0.0330541767179966, "global_step": 5692, "epoch": 158, "lr": 0.001} +{"train_loss": 0.042945362627506256, "train_loss_bc": 0.042945362627506256, "train_loss_llm": 0.0, "grad_norm": 0.017553213983774185, "global_step": 5693, "epoch": 158, "lr": 0.001} +{"train_loss": 0.044070012867450714, "train_loss_bc": 0.04406635835766792, "train_loss_llm": 0.00036537647247314453, "grad_norm": 0.027801545336842537, "global_step": 5694, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03909124806523323, "train_loss_bc": 0.03909124806523323, "train_loss_llm": 0.0, "grad_norm": 0.0308280847966671, "global_step": 5695, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03965192660689354, "train_loss_bc": 0.03965192660689354, "train_loss_llm": 0.0, "grad_norm": 0.04247685521841049, "global_step": 5696, "epoch": 158, "lr": 0.001} +{"train_loss": 0.040030211210250854, "train_loss_bc": 0.04002542793750763, "train_loss_llm": 0.00047841668128967285, "grad_norm": 0.017897391691803932, "global_step": 5697, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03359849750995636, "train_loss_bc": 0.03359849750995636, "train_loss_llm": 0.0, "grad_norm": 0.03176053240895271, "global_step": 5698, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04504898563027382, "train_loss_bc": 0.04504898563027382, "train_loss_llm": 0.0, "grad_norm": 0.036902353167533875, "global_step": 5699, "epoch": 158, "lr": 0.001} +{"train_loss": 0.035903457552194595, "train_loss_bc": 0.035903457552194595, "train_loss_llm": 0.0, "grad_norm": 0.03844819962978363, "global_step": 5700, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04634849727153778, "train_loss_bc": 0.04634082689881325, "train_loss_llm": 0.0007669031620025635, "grad_norm": 0.016201533377170563, "global_step": 5701, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04700778052210808, "train_loss_bc": 0.04700778052210808, "train_loss_llm": 0.0, "grad_norm": 0.025277385488152504, "global_step": 5702, "epoch": 158, "lr": 0.001} +{"train_loss": 0.034637849777936935, "train_loss_bc": 0.034637849777936935, "train_loss_llm": 0.0, "grad_norm": 0.031946487724781036, "global_step": 5703, "epoch": 158, "lr": 0.001} +{"train_loss": 0.036136772483587265, "train_loss_bc": 0.036136772483587265, "train_loss_llm": 0.0, "grad_norm": 0.03448387607932091, "global_step": 5704, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04421045258641243, "train_loss_bc": 0.044206760823726654, "train_loss_llm": 0.00036901235580444336, "grad_norm": 0.015862874686717987, "global_step": 5705, "epoch": 158, "lr": 0.001} +{"train_loss": 0.048199862241744995, "train_loss_bc": 0.048175957053899765, "train_loss_llm": 0.002390623092651367, "grad_norm": 0.024314580485224724, "global_step": 5706, "epoch": 158, "lr": 0.001} +{"train_loss": 0.054927218705415726, "train_loss_bc": 0.054927218705415726, "train_loss_llm": 0.0, "grad_norm": 0.03061453439295292, "global_step": 5707, "epoch": 158, "lr": 0.001} +{"train_loss": 0.0368444100022316, "train_loss_bc": 0.0368148498237133, "train_loss_llm": 0.002956092357635498, "grad_norm": 0.03928513452410698, "global_step": 5708, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03540453314781189, "train_loss_bc": 0.03540179133415222, "train_loss_llm": 0.0002740025520324707, "grad_norm": 0.013095044530928135, "global_step": 5709, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04425160959362984, "train_loss_bc": 0.04425160959362984, "train_loss_llm": 0.0, "grad_norm": 0.02275337278842926, "global_step": 5710, "epoch": 158, "lr": 0.001} +{"train_loss": 0.044956404715776443, "train_loss_bc": 0.044955264776945114, "train_loss_llm": 0.00011411309242248535, "grad_norm": 0.029490409418940544, "global_step": 5711, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03354375436902046, "train_loss_bc": 0.03354375436902046, "train_loss_llm": 0.0, "grad_norm": 0.03468925505876541, "global_step": 5712, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03860922530293465, "train_loss_bc": 0.03860880061984062, "train_loss_llm": 4.2319297790527344e-05, "grad_norm": 0.01607842929661274, "global_step": 5713, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04510222747921944, "train_loss_bc": 0.04506662115454674, "train_loss_llm": 0.0035604536533355713, "grad_norm": 0.02646971307694912, "global_step": 5714, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03189893066883087, "train_loss_bc": 0.03188728913664818, "train_loss_llm": 0.0011642277240753174, "grad_norm": 0.03566386178135872, "global_step": 5715, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03990411013364792, "train_loss_bc": 0.03989173099398613, "train_loss_llm": 0.0012380480766296387, "grad_norm": 0.04658845439553261, "global_step": 5716, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04149407893419266, "train_loss_bc": 0.041467342525720596, "train_loss_llm": 0.0026735663414001465, "grad_norm": 0.01668100617825985, "global_step": 5717, "epoch": 158, "lr": 0.001} +{"train_loss": 0.038350850343704224, "train_loss_bc": 0.038350850343704224, "train_loss_llm": 0.0, "grad_norm": 0.02870943211019039, "global_step": 5718, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04837380349636078, "train_loss_bc": 0.04837380349636078, "train_loss_llm": 0.0, "grad_norm": 0.043813902884721756, "global_step": 5719, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03810508921742439, "train_loss_bc": 0.03810508921742439, "train_loss_llm": 0.0, "grad_norm": 0.049899082630872726, "global_step": 5720, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04546351730823517, "train_loss_bc": 0.04546351730823517, "train_loss_llm": 0.0, "grad_norm": 0.01742934063076973, "global_step": 5721, "epoch": 158, "lr": 0.001} +{"train_loss": 0.03964920714497566, "train_loss_bc": 0.039640285074710846, "train_loss_llm": 0.0008921027183532715, "grad_norm": 0.019477741792798042, "global_step": 5722, "epoch": 158, "lr": 0.001} +{"train_loss": 0.04125693502525488, "train_loss_bc": 0.036898061633110046, "train_loss_llm": 0.0, "grad_norm": 0.028352314606308937, "global_step": 5723, "epoch": 158, "lr": 0.001, "val_loss": 0.05447666347026825} +{"train_loss": 0.04350573569536209, "train_loss_bc": 0.04350167512893677, "train_loss_llm": 0.00040593743324279785, "grad_norm": 0.03430233150720596, "global_step": 5724, "epoch": 159, "lr": 0.001} +{"train_loss": 0.0470627062022686, "train_loss_bc": 0.0470627062022686, "train_loss_llm": 0.0, "grad_norm": 0.014573606662452221, "global_step": 5725, "epoch": 159, "lr": 0.001} +{"train_loss": 0.04470255598425865, "train_loss_bc": 0.04470255598425865, "train_loss_llm": 0.0, "grad_norm": 0.025024183094501495, "global_step": 5726, "epoch": 159, "lr": 0.001} +{"train_loss": 0.04370265454053879, "train_loss_bc": 0.043679382652044296, "train_loss_llm": 0.0023272931575775146, "grad_norm": 0.031270746141672134, "global_step": 5727, "epoch": 159, "lr": 0.001} +{"train_loss": 0.047823742032051086, "train_loss_bc": 0.04780999571084976, "train_loss_llm": 0.0013746023178100586, "grad_norm": 0.036228202283382416, "global_step": 5728, "epoch": 159, "lr": 0.001} +{"train_loss": 0.03950916975736618, "train_loss_bc": 0.039490412920713425, "train_loss_llm": 0.001875549554824829, "grad_norm": 0.017632095143198967, "global_step": 5729, "epoch": 159, "lr": 0.001} +{"train_loss": 0.03198036924004555, "train_loss_bc": 0.03198036924004555, "train_loss_llm": 0.0, "grad_norm": 0.0292071346193552, "global_step": 5730, "epoch": 159, "lr": 0.001} +{"train_loss": 0.04296976327896118, "train_loss_bc": 0.04296852648258209, "train_loss_llm": 0.00012379884719848633, "grad_norm": 0.03991337865591049, "global_step": 5731, "epoch": 159, "lr": 0.001} +{"train_loss": 0.042209167033433914, "train_loss_bc": 0.042209167033433914, "train_loss_llm": 0.0, "grad_norm": 0.048749327659606934, "global_step": 5732, "epoch": 159, "lr": 0.001} +{"train_loss": 0.04494727775454521, "train_loss_bc": 0.04494727775454521, "train_loss_llm": 0.0, "grad_norm": 0.014562993310391903, "global_step": 5733, "epoch": 159, "lr": 0.001} +{"train_loss": 0.038868166506290436, "train_loss_bc": 0.038868166506290436, "train_loss_llm": 0.0, "grad_norm": 0.0201396681368351, "global_step": 5734, "epoch": 159, "lr": 0.001} +{"train_loss": 0.03363584354519844, "train_loss_bc": 0.03363584354519844, "train_loss_llm": 0.0, "grad_norm": 0.022288566455245018, "global_step": 5735, "epoch": 159, "lr": 0.001} +{"train_loss": 0.04608564451336861, "train_loss_bc": 0.04608564451336861, "train_loss_llm": 0.0, "grad_norm": 0.03133155405521393, "global_step": 5736, "epoch": 159, "lr": 0.001} +{"train_loss": 0.040184203535318375, "train_loss_bc": 0.040184203535318375, "train_loss_llm": 0.0, "grad_norm": 0.01576324924826622, "global_step": 5737, "epoch": 159, "lr": 0.001} +{"train_loss": 0.047510869801044464, "train_loss_bc": 0.04750097170472145, "train_loss_llm": 0.0009897351264953613, "grad_norm": 0.02195245772600174, "global_step": 5738, "epoch": 159, "lr": 0.001} +{"train_loss": 0.047258760780096054, "train_loss_bc": 0.047258760780096054, "train_loss_llm": 0.0, "grad_norm": 0.026344602927565575, "global_step": 5739, "epoch": 159, "lr": 0.001} +{"train_loss": 0.03954143449664116, "train_loss_bc": 0.03954143449664116, "train_loss_llm": 0.0, "grad_norm": 0.027674544602632523, "global_step": 5740, "epoch": 159, "lr": 0.001} +{"train_loss": 0.0447772853076458, "train_loss_bc": 0.0447772853076458, "train_loss_llm": 0.0, "grad_norm": 0.01901586540043354, "global_step": 5741, "epoch": 159, "lr": 0.001} +{"train_loss": 0.046534158289432526, "train_loss_bc": 0.046534158289432526, "train_loss_llm": 0.0, "grad_norm": 0.02682286873459816, "global_step": 5742, "epoch": 159, "lr": 0.001} +{"train_loss": 0.028629140928387642, "train_loss_bc": 0.02861945331096649, "train_loss_llm": 0.0009687542915344238, "grad_norm": 0.032326530665159225, "global_step": 5743, "epoch": 159, "lr": 0.001} +{"train_loss": 0.042919959872961044, "train_loss_bc": 0.042919959872961044, "train_loss_llm": 0.0, "grad_norm": 0.03667588531970978, "global_step": 5744, "epoch": 159, "lr": 0.001} +{"train_loss": 0.027254628017544746, "train_loss_bc": 0.02724861353635788, "train_loss_llm": 0.0006014108657836914, "grad_norm": 0.013800048269331455, "global_step": 5745, "epoch": 159, "lr": 0.001} +{"train_loss": 0.034818876534700394, "train_loss_bc": 0.034818876534700394, "train_loss_llm": 0.0, "grad_norm": 0.01905958727002144, "global_step": 5746, "epoch": 159, "lr": 0.001} +{"train_loss": 0.03485027700662613, "train_loss_bc": 0.03484956547617912, "train_loss_llm": 7.107853889465332e-05, "grad_norm": 0.02879876084625721, "global_step": 5747, "epoch": 159, "lr": 0.001} +{"train_loss": 0.03354339301586151, "train_loss_bc": 0.03354339301586151, "train_loss_llm": 0.0, "grad_norm": 0.03109007142484188, "global_step": 5748, "epoch": 159, "lr": 0.001} +{"train_loss": 0.04576674848794937, "train_loss_bc": 0.04575568810105324, "train_loss_llm": 0.0011059343814849854, "grad_norm": 0.02022136002779007, "global_step": 5749, "epoch": 159, "lr": 0.001} +{"train_loss": 0.03684740141034126, "train_loss_bc": 0.03684740141034126, "train_loss_llm": 0.0, "grad_norm": 0.03113478235900402, "global_step": 5750, "epoch": 159, "lr": 0.001} +{"train_loss": 0.04644576832652092, "train_loss_bc": 0.04644576832652092, "train_loss_llm": 0.0, "grad_norm": 0.03734978660941124, "global_step": 5751, "epoch": 159, "lr": 0.001} +{"train_loss": 0.043999992311000824, "train_loss_bc": 0.043999992311000824, "train_loss_llm": 0.0, "grad_norm": 0.043185044080019, "global_step": 5752, "epoch": 159, "lr": 0.001} +{"train_loss": 0.04044673591852188, "train_loss_bc": 0.04044673591852188, "train_loss_llm": 0.0, "grad_norm": 0.020431378856301308, "global_step": 5753, "epoch": 159, "lr": 0.001} +{"train_loss": 0.048402562737464905, "train_loss_bc": 0.04840102046728134, "train_loss_llm": 0.00015440583229064941, "grad_norm": 0.03198070079088211, "global_step": 5754, "epoch": 159, "lr": 0.001} +{"train_loss": 0.04470721259713173, "train_loss_bc": 0.04470721259713173, "train_loss_llm": 0.0, "grad_norm": 0.041653554886579514, "global_step": 5755, "epoch": 159, "lr": 0.001} +{"train_loss": 0.04356703162193298, "train_loss_bc": 0.04356703162193298, "train_loss_llm": 0.0, "grad_norm": 0.05099476873874664, "global_step": 5756, "epoch": 159, "lr": 0.001} +{"train_loss": 0.041050054132938385, "train_loss_bc": 0.041050054132938385, "train_loss_llm": 0.0, "grad_norm": 0.0176316536962986, "global_step": 5757, "epoch": 159, "lr": 0.001} +{"train_loss": 0.038364898413419724, "train_loss_bc": 0.038364898413419724, "train_loss_llm": 0.0, "grad_norm": 0.02242191508412361, "global_step": 5758, "epoch": 159, "lr": 0.001} +{"train_loss": 0.041271907173924975, "train_loss_bc": 0.04136446863412857, "train_loss_llm": 0.0, "grad_norm": 0.03254769742488861, "global_step": 5759, "epoch": 159, "lr": 0.001, "val_loss": 0.05480441451072693} +{"train_loss": 0.036185916513204575, "train_loss_bc": 0.036185916513204575, "train_loss_llm": 0.0, "grad_norm": 0.034558914601802826, "global_step": 5760, "epoch": 160, "lr": 0.001} +{"train_loss": 0.045486822724342346, "train_loss_bc": 0.045486822724342346, "train_loss_llm": 0.0, "grad_norm": 0.018550870940089226, "global_step": 5761, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04209640249609947, "train_loss_bc": 0.042083099484443665, "train_loss_llm": 0.0013304054737091064, "grad_norm": 0.029819853603839874, "global_step": 5762, "epoch": 160, "lr": 0.001} +{"train_loss": 0.034412216395139694, "train_loss_bc": 0.034412216395139694, "train_loss_llm": 0.0, "grad_norm": 0.034940220415592194, "global_step": 5763, "epoch": 160, "lr": 0.001} +{"train_loss": 0.03798684850335121, "train_loss_bc": 0.03798684850335121, "train_loss_llm": 0.0, "grad_norm": 0.038493502885103226, "global_step": 5764, "epoch": 160, "lr": 0.001} +{"train_loss": 0.045273441821336746, "train_loss_bc": 0.045273441821336746, "train_loss_llm": 0.0, "grad_norm": 0.016016269102692604, "global_step": 5765, "epoch": 160, "lr": 0.001} +{"train_loss": 0.042634516954422, "train_loss_bc": 0.04262763634324074, "train_loss_llm": 0.0006879568099975586, "grad_norm": 0.02761509083211422, "global_step": 5766, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04049943760037422, "train_loss_bc": 0.04049943760037422, "train_loss_llm": 0.0, "grad_norm": 0.029272522777318954, "global_step": 5767, "epoch": 160, "lr": 0.001} +{"train_loss": 0.042479533702135086, "train_loss_bc": 0.042479533702135086, "train_loss_llm": 0.0, "grad_norm": 0.03442442789673805, "global_step": 5768, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04097992926836014, "train_loss_bc": 0.04097992926836014, "train_loss_llm": 0.0, "grad_norm": 0.012618010863661766, "global_step": 5769, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04876261577010155, "train_loss_bc": 0.04874492064118385, "train_loss_llm": 0.0017695426940917969, "grad_norm": 0.026539107784628868, "global_step": 5770, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04126071184873581, "train_loss_bc": 0.04125043749809265, "train_loss_llm": 0.0010272562503814697, "grad_norm": 0.03778375685214996, "global_step": 5771, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04940475523471832, "train_loss_bc": 0.049392204731702805, "train_loss_llm": 0.0012550652027130127, "grad_norm": 0.04610665887594223, "global_step": 5772, "epoch": 160, "lr": 0.001} +{"train_loss": 0.03760223835706711, "train_loss_bc": 0.03759602829813957, "train_loss_llm": 0.0006211698055267334, "grad_norm": 0.016483135521411896, "global_step": 5773, "epoch": 160, "lr": 0.001} +{"train_loss": 0.030039312317967415, "train_loss_bc": 0.030013399198651314, "train_loss_llm": 0.002591252326965332, "grad_norm": 0.024528473615646362, "global_step": 5774, "epoch": 160, "lr": 0.001} +{"train_loss": 0.038683369755744934, "train_loss_bc": 0.038683369755744934, "train_loss_llm": 0.0, "grad_norm": 0.028780613094568253, "global_step": 5775, "epoch": 160, "lr": 0.001} +{"train_loss": 0.0450420156121254, "train_loss_bc": 0.04502953588962555, "train_loss_llm": 0.0012480616569519043, "grad_norm": 0.03597071021795273, "global_step": 5776, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04606669396162033, "train_loss_bc": 0.04606509581208229, "train_loss_llm": 0.00015982985496520996, "grad_norm": 0.018017329275608063, "global_step": 5777, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04324568063020706, "train_loss_bc": 0.04324568063020706, "train_loss_llm": 0.0, "grad_norm": 0.02543867751955986, "global_step": 5778, "epoch": 160, "lr": 0.001} +{"train_loss": 0.03815062344074249, "train_loss_bc": 0.03815062344074249, "train_loss_llm": 0.0, "grad_norm": 0.036704786121845245, "global_step": 5779, "epoch": 160, "lr": 0.001} +{"train_loss": 0.05148186534643173, "train_loss_bc": 0.05147330462932587, "train_loss_llm": 0.0008561611175537109, "grad_norm": 0.04170569032430649, "global_step": 5780, "epoch": 160, "lr": 0.001} +{"train_loss": 0.040194325149059296, "train_loss_bc": 0.040194325149059296, "train_loss_llm": 0.0, "grad_norm": 0.017817778512835503, "global_step": 5781, "epoch": 160, "lr": 0.001} +{"train_loss": 0.032136816531419754, "train_loss_bc": 0.032135553658008575, "train_loss_llm": 0.00012615323066711426, "grad_norm": 0.025854719802737236, "global_step": 5782, "epoch": 160, "lr": 0.001} +{"train_loss": 0.03629140183329582, "train_loss_bc": 0.03628728911280632, "train_loss_llm": 0.00041112303733825684, "grad_norm": 0.03512720391154289, "global_step": 5783, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04469422623515129, "train_loss_bc": 0.04469422623515129, "train_loss_llm": 0.0, "grad_norm": 0.04638032242655754, "global_step": 5784, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04040598124265671, "train_loss_bc": 0.04040598124265671, "train_loss_llm": 0.0, "grad_norm": 0.01948367804288864, "global_step": 5785, "epoch": 160, "lr": 0.001} +{"train_loss": 0.036092549562454224, "train_loss_bc": 0.036092549562454224, "train_loss_llm": 0.0, "grad_norm": 0.028130363672971725, "global_step": 5786, "epoch": 160, "lr": 0.001} +{"train_loss": 0.03682136908173561, "train_loss_bc": 0.03682136908173561, "train_loss_llm": 0.0, "grad_norm": 0.03768692910671234, "global_step": 5787, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04391004517674446, "train_loss_bc": 0.04391004517674446, "train_loss_llm": 0.0, "grad_norm": 0.04123607277870178, "global_step": 5788, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04929785802960396, "train_loss_bc": 0.04929785802960396, "train_loss_llm": 0.0, "grad_norm": 0.018201807513833046, "global_step": 5789, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04018962383270264, "train_loss_bc": 0.04017604514956474, "train_loss_llm": 0.0013580024242401123, "grad_norm": 0.030472904443740845, "global_step": 5790, "epoch": 160, "lr": 0.001} +{"train_loss": 0.03919769451022148, "train_loss_bc": 0.03919769451022148, "train_loss_llm": 0.0, "grad_norm": 0.04375641793012619, "global_step": 5791, "epoch": 160, "lr": 0.001} +{"train_loss": 0.05009682476520538, "train_loss_bc": 0.05009682476520538, "train_loss_llm": 0.0, "grad_norm": 0.044692859053611755, "global_step": 5792, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04011860489845276, "train_loss_bc": 0.04011860489845276, "train_loss_llm": 0.0, "grad_norm": 0.021518178284168243, "global_step": 5793, "epoch": 160, "lr": 0.001} +{"train_loss": 0.03727658838033676, "train_loss_bc": 0.03727658838033676, "train_loss_llm": 0.0, "grad_norm": 0.02815934270620346, "global_step": 5794, "epoch": 160, "lr": 0.001} +{"train_loss": 0.04131815235854851, "train_loss_bc": 0.04295462742447853, "train_loss_llm": 0.0, "grad_norm": 0.04226170480251312, "global_step": 5795, "epoch": 160, "lr": 0.001, "train/cumulative_reward": 107.93985837503558, "train/mean_score": 16.831262558069835, "train/success_rate": 0.4, "test/cumulative_reward": 47.13257772571367, "test/mean_score": 8.186770543693031, "test/success_rate": 0.18, "val_loss": 0.05496034398674965, "train_action_mse_error": 0.04548760876059532} +{"train_loss": 0.04134516417980194, "train_loss_bc": 0.04134516417980194, "train_loss_llm": 0.0, "grad_norm": 0.053450338542461395, "global_step": 5796, "epoch": 161, "lr": 0.001} +{"train_loss": 0.04120698198676109, "train_loss_bc": 0.04119190573692322, "train_loss_llm": 0.0015076398849487305, "grad_norm": 0.015706978738307953, "global_step": 5797, "epoch": 161, "lr": 0.001} +{"train_loss": 0.05461089313030243, "train_loss_bc": 0.05461089313030243, "train_loss_llm": 0.0, "grad_norm": 0.02832491137087345, "global_step": 5798, "epoch": 161, "lr": 0.001} +{"train_loss": 0.04810931161046028, "train_loss_bc": 0.04810931161046028, "train_loss_llm": 0.0, "grad_norm": 0.0357356034219265, "global_step": 5799, "epoch": 161, "lr": 0.001} +{"train_loss": 0.03755037486553192, "train_loss_bc": 0.03755037486553192, "train_loss_llm": 0.0, "grad_norm": 0.04168951138854027, "global_step": 5800, "epoch": 161, "lr": 0.001} +{"train_loss": 0.05036444962024689, "train_loss_bc": 0.050354260951280594, "train_loss_llm": 0.0010188519954681396, "grad_norm": 0.020298052579164505, "global_step": 5801, "epoch": 161, "lr": 0.001} +{"train_loss": 0.041505176573991776, "train_loss_bc": 0.0414743609726429, "train_loss_llm": 0.0030816495418548584, "grad_norm": 0.031267184764146805, "global_step": 5802, "epoch": 161, "lr": 0.001} +{"train_loss": 0.04068602994084358, "train_loss_bc": 0.04067576676607132, "train_loss_llm": 0.0010262727737426758, "grad_norm": 0.04033923149108887, "global_step": 5803, "epoch": 161, "lr": 0.001} +{"train_loss": 0.044656116515398026, "train_loss_bc": 0.044656116515398026, "train_loss_llm": 0.0, "grad_norm": 0.04895525425672531, "global_step": 5804, "epoch": 161, "lr": 0.001} +{"train_loss": 0.04866619035601616, "train_loss_bc": 0.04866619035601616, "train_loss_llm": 0.0, "grad_norm": 0.017780840396881104, "global_step": 5805, "epoch": 161, "lr": 0.001} +{"train_loss": 0.03632045537233353, "train_loss_bc": 0.03632045537233353, "train_loss_llm": 0.0, "grad_norm": 0.03401040658354759, "global_step": 5806, "epoch": 161, "lr": 0.001} +{"train_loss": 0.040382448583841324, "train_loss_bc": 0.04035886377096176, "train_loss_llm": 0.002358347177505493, "grad_norm": 0.045842282474040985, "global_step": 5807, "epoch": 161, "lr": 0.001} +{"train_loss": 0.03977888450026512, "train_loss_bc": 0.03975759819149971, "train_loss_llm": 0.002128779888153076, "grad_norm": 0.061121147125959396, "global_step": 5808, "epoch": 161, "lr": 0.001} +{"train_loss": 0.041907671838998795, "train_loss_bc": 0.041907671838998795, "train_loss_llm": 0.0, "grad_norm": 0.01668541692197323, "global_step": 5809, "epoch": 161, "lr": 0.001} +{"train_loss": 0.03906378522515297, "train_loss_bc": 0.03906378522515297, "train_loss_llm": 0.0, "grad_norm": 0.025267038494348526, "global_step": 5810, "epoch": 161, "lr": 0.001} +{"train_loss": 0.03962056711316109, "train_loss_bc": 0.03962056711316109, "train_loss_llm": 0.0, "grad_norm": 0.03620132431387901, "global_step": 5811, "epoch": 161, "lr": 0.001} +{"train_loss": 0.037858325988054276, "train_loss_bc": 0.03785106912255287, "train_loss_llm": 0.000725865364074707, "grad_norm": 0.048667062073946, "global_step": 5812, "epoch": 161, "lr": 0.001} +{"train_loss": 0.052040599286556244, "train_loss_bc": 0.05202259495854378, "train_loss_llm": 0.0018002986907958984, "grad_norm": 0.02065396122634411, "global_step": 5813, "epoch": 161, "lr": 0.001} +{"train_loss": 0.046156108379364014, "train_loss_bc": 0.046156108379364014, "train_loss_llm": 0.0, "grad_norm": 0.029251057654619217, "global_step": 5814, "epoch": 161, "lr": 0.001} +{"train_loss": 0.03835030272603035, "train_loss_bc": 0.03835030272603035, "train_loss_llm": 0.0, "grad_norm": 0.041289638727903366, "global_step": 5815, "epoch": 161, "lr": 0.001} +{"train_loss": 0.04592476040124893, "train_loss_bc": 0.045910805463790894, "train_loss_llm": 0.00139540433883667, "grad_norm": 0.04957748204469681, "global_step": 5816, "epoch": 161, "lr": 0.001} +{"train_loss": 0.044896651059389114, "train_loss_bc": 0.044896651059389114, "train_loss_llm": 0.0, "grad_norm": 0.018948420882225037, "global_step": 5817, "epoch": 161, "lr": 0.001} +{"train_loss": 0.037803079932928085, "train_loss_bc": 0.037803079932928085, "train_loss_llm": 0.0, "grad_norm": 0.025196291506290436, "global_step": 5818, "epoch": 161, "lr": 0.001} +{"train_loss": 0.02883051335811615, "train_loss_bc": 0.02883051335811615, "train_loss_llm": 0.0, "grad_norm": 0.035990986973047256, "global_step": 5819, "epoch": 161, "lr": 0.001} +{"train_loss": 0.039437245577573776, "train_loss_bc": 0.039437245577573776, "train_loss_llm": 0.0, "grad_norm": 0.039827700704336166, "global_step": 5820, "epoch": 161, "lr": 0.001} +{"train_loss": 0.044586699455976486, "train_loss_bc": 0.044586699455976486, "train_loss_llm": 0.0, "grad_norm": 0.01704169623553753, "global_step": 5821, "epoch": 161, "lr": 0.001} +{"train_loss": 0.0425541065633297, "train_loss_bc": 0.0425541065633297, "train_loss_llm": 0.0, "grad_norm": 0.0249690730124712, "global_step": 5822, "epoch": 161, "lr": 0.001} +{"train_loss": 0.036231879144907, "train_loss_bc": 0.036231879144907, "train_loss_llm": 0.0, "grad_norm": 0.03555644303560257, "global_step": 5823, "epoch": 161, "lr": 0.001} +{"train_loss": 0.039315663278102875, "train_loss_bc": 0.039315663278102875, "train_loss_llm": 0.0, "grad_norm": 0.04459776729345322, "global_step": 5824, "epoch": 161, "lr": 0.001} +{"train_loss": 0.03532049432396889, "train_loss_bc": 0.035302117466926575, "train_loss_llm": 0.0018375813961029053, "grad_norm": 0.01754365675151348, "global_step": 5825, "epoch": 161, "lr": 0.001} +{"train_loss": 0.04656103625893593, "train_loss_bc": 0.04656103625893593, "train_loss_llm": 0.0, "grad_norm": 0.026225732639431953, "global_step": 5826, "epoch": 161, "lr": 0.001} +{"train_loss": 0.03310760110616684, "train_loss_bc": 0.03310760110616684, "train_loss_llm": 0.0, "grad_norm": 0.03498494252562523, "global_step": 5827, "epoch": 161, "lr": 0.001} +{"train_loss": 0.0335271842777729, "train_loss_bc": 0.0335271842777729, "train_loss_llm": 0.0, "grad_norm": 0.042120758444070816, "global_step": 5828, "epoch": 161, "lr": 0.001} +{"train_loss": 0.04418787732720375, "train_loss_bc": 0.044154971837997437, "train_loss_llm": 0.0032903850078582764, "grad_norm": 0.02085384912788868, "global_step": 5829, "epoch": 161, "lr": 0.001} +{"train_loss": 0.03801528736948967, "train_loss_bc": 0.03801528736948967, "train_loss_llm": 0.0, "grad_norm": 0.0243390966206789, "global_step": 5830, "epoch": 161, "lr": 0.001} +{"train_loss": 0.04149212232894368, "train_loss_bc": 0.04323648661375046, "train_loss_llm": 0.0, "grad_norm": 0.035017888993024826, "global_step": 5831, "epoch": 161, "lr": 0.001, "val_loss": 0.05547206848859787} +{"train_loss": 0.036746349185705185, "train_loss_bc": 0.036746349185705185, "train_loss_llm": 0.0, "grad_norm": 0.04715438559651375, "global_step": 5832, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04023626074194908, "train_loss_bc": 0.04023626074194908, "train_loss_llm": 0.0, "grad_norm": 0.01788557507097721, "global_step": 5833, "epoch": 162, "lr": 0.001} +{"train_loss": 0.0359557680785656, "train_loss_bc": 0.03595312312245369, "train_loss_llm": 0.0002644956111907959, "grad_norm": 0.021988360211253166, "global_step": 5834, "epoch": 162, "lr": 0.001} +{"train_loss": 0.030107304453849792, "train_loss_bc": 0.030107304453849792, "train_loss_llm": 0.0, "grad_norm": 0.026771774515509605, "global_step": 5835, "epoch": 162, "lr": 0.001} +{"train_loss": 0.03481198474764824, "train_loss_bc": 0.03480635955929756, "train_loss_llm": 0.0005624294281005859, "grad_norm": 0.03345948085188866, "global_step": 5836, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04247427359223366, "train_loss_bc": 0.04247427359223366, "train_loss_llm": 0.0, "grad_norm": 0.015630334615707397, "global_step": 5837, "epoch": 162, "lr": 0.001} +{"train_loss": 0.035972658544778824, "train_loss_bc": 0.0359468087553978, "train_loss_llm": 0.0025850534439086914, "grad_norm": 0.020133478567004204, "global_step": 5838, "epoch": 162, "lr": 0.001} +{"train_loss": 0.041087280958890915, "train_loss_bc": 0.041087280958890915, "train_loss_llm": 0.0, "grad_norm": 0.027266526594758034, "global_step": 5839, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04295589402318001, "train_loss_bc": 0.04295589402318001, "train_loss_llm": 0.0, "grad_norm": 0.0386081263422966, "global_step": 5840, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04327774420380592, "train_loss_bc": 0.04327774420380592, "train_loss_llm": 0.0, "grad_norm": 0.014697955921292305, "global_step": 5841, "epoch": 162, "lr": 0.001} +{"train_loss": 0.044408299028873444, "train_loss_bc": 0.044384799897670746, "train_loss_llm": 0.002349853515625, "grad_norm": 0.022105418145656586, "global_step": 5842, "epoch": 162, "lr": 0.001} +{"train_loss": 0.0442584790289402, "train_loss_bc": 0.0442584790289402, "train_loss_llm": 0.0, "grad_norm": 0.035737283527851105, "global_step": 5843, "epoch": 162, "lr": 0.001} +{"train_loss": 0.03734877333045006, "train_loss_bc": 0.03734877333045006, "train_loss_llm": 0.0, "grad_norm": 0.04306081682443619, "global_step": 5844, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04055001959204674, "train_loss_bc": 0.04055001959204674, "train_loss_llm": 0.0, "grad_norm": 0.02102101966738701, "global_step": 5845, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04317902773618698, "train_loss_bc": 0.04317902773618698, "train_loss_llm": 0.0, "grad_norm": 0.035951972007751465, "global_step": 5846, "epoch": 162, "lr": 0.001} +{"train_loss": 0.0419519767165184, "train_loss_bc": 0.0419519767165184, "train_loss_llm": 0.0, "grad_norm": 0.04293464124202728, "global_step": 5847, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04127933830022812, "train_loss_bc": 0.04127933830022812, "train_loss_llm": 0.0, "grad_norm": 0.0543261356651783, "global_step": 5848, "epoch": 162, "lr": 0.001} +{"train_loss": 0.03823314607143402, "train_loss_bc": 0.03823314607143402, "train_loss_llm": 0.0, "grad_norm": 0.01578628644347191, "global_step": 5849, "epoch": 162, "lr": 0.001} +{"train_loss": 0.03849959746003151, "train_loss_bc": 0.03849959746003151, "train_loss_llm": 0.0, "grad_norm": 0.029535256326198578, "global_step": 5850, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04963664710521698, "train_loss_bc": 0.049627162516117096, "train_loss_llm": 0.0009483695030212402, "grad_norm": 0.036741673946380615, "global_step": 5851, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04821885749697685, "train_loss_bc": 0.04821885749697685, "train_loss_llm": 0.0, "grad_norm": 0.05020672455430031, "global_step": 5852, "epoch": 162, "lr": 0.001} +{"train_loss": 0.044041652232408524, "train_loss_bc": 0.044041652232408524, "train_loss_llm": 0.0, "grad_norm": 0.017307715490460396, "global_step": 5853, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04706777632236481, "train_loss_bc": 0.04706777632236481, "train_loss_llm": 0.0, "grad_norm": 0.026567941531538963, "global_step": 5854, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04069900885224342, "train_loss_bc": 0.04069900885224342, "train_loss_llm": 0.0, "grad_norm": 0.03518584370613098, "global_step": 5855, "epoch": 162, "lr": 0.001} +{"train_loss": 0.03755083307623863, "train_loss_bc": 0.03754756972193718, "train_loss_llm": 0.00032651424407958984, "grad_norm": 0.04010176286101341, "global_step": 5856, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04206421971321106, "train_loss_bc": 0.0420544371008873, "train_loss_llm": 0.0009782612323760986, "grad_norm": 0.016293363645672798, "global_step": 5857, "epoch": 162, "lr": 0.001} +{"train_loss": 0.03944804146885872, "train_loss_bc": 0.03944804146885872, "train_loss_llm": 0.0, "grad_norm": 0.028161559253931046, "global_step": 5858, "epoch": 162, "lr": 0.001} +{"train_loss": 0.03805626556277275, "train_loss_bc": 0.03805626556277275, "train_loss_llm": 0.0, "grad_norm": 0.036990948021411896, "global_step": 5859, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04473245516419411, "train_loss_bc": 0.04473245516419411, "train_loss_llm": 0.0, "grad_norm": 0.046683669090270996, "global_step": 5860, "epoch": 162, "lr": 0.001} +{"train_loss": 0.0431617833673954, "train_loss_bc": 0.0431617833673954, "train_loss_llm": 0.0, "grad_norm": 0.016018804162740707, "global_step": 5861, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04012284800410271, "train_loss_bc": 0.04008176550269127, "train_loss_llm": 0.004108071327209473, "grad_norm": 0.027770888060331345, "global_step": 5862, "epoch": 162, "lr": 0.001} +{"train_loss": 0.0487077459692955, "train_loss_bc": 0.0487077459692955, "train_loss_llm": 0.0, "grad_norm": 0.03771167993545532, "global_step": 5863, "epoch": 162, "lr": 0.001} +{"train_loss": 0.0486283078789711, "train_loss_bc": 0.0486283078789711, "train_loss_llm": 0.0, "grad_norm": 0.050888244062662125, "global_step": 5864, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04122980311512947, "train_loss_bc": 0.04122810810804367, "train_loss_llm": 0.00016945600509643555, "grad_norm": 0.01668822392821312, "global_step": 5865, "epoch": 162, "lr": 0.001} +{"train_loss": 0.045033425092697144, "train_loss_bc": 0.045033425092697144, "train_loss_llm": 0.0, "grad_norm": 0.027594001963734627, "global_step": 5866, "epoch": 162, "lr": 0.001} +{"train_loss": 0.04132613560391797, "train_loss_bc": 0.03600703552365303, "train_loss_llm": 0.0, "grad_norm": 0.0425833985209465, "global_step": 5867, "epoch": 162, "lr": 0.001, "val_loss": 0.05533710494637489} +{"train_loss": 0.041490938514471054, "train_loss_bc": 0.041490938514471054, "train_loss_llm": 0.0, "grad_norm": 0.062479812651872635, "global_step": 5868, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04112742096185684, "train_loss_bc": 0.04112742096185684, "train_loss_llm": 0.0, "grad_norm": 0.018607528880238533, "global_step": 5869, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04192349687218666, "train_loss_bc": 0.04192349687218666, "train_loss_llm": 0.0, "grad_norm": 0.0272897370159626, "global_step": 5870, "epoch": 163, "lr": 0.001} +{"train_loss": 0.03449412062764168, "train_loss_bc": 0.03449412062764168, "train_loss_llm": 0.0, "grad_norm": 0.030776076018810272, "global_step": 5871, "epoch": 163, "lr": 0.001} +{"train_loss": 0.046701814979314804, "train_loss_bc": 0.04668653383851051, "train_loss_llm": 0.0015279948711395264, "grad_norm": 0.03970862925052643, "global_step": 5872, "epoch": 163, "lr": 0.001} +{"train_loss": 0.03730624169111252, "train_loss_bc": 0.03730316087603569, "train_loss_llm": 0.00030803680419921875, "grad_norm": 0.01974700577557087, "global_step": 5873, "epoch": 163, "lr": 0.001} +{"train_loss": 0.045979831367731094, "train_loss_bc": 0.04596678540110588, "train_loss_llm": 0.00130462646484375, "grad_norm": 0.02824164554476738, "global_step": 5874, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04190286621451378, "train_loss_bc": 0.04190286621451378, "train_loss_llm": 0.0, "grad_norm": 0.038920808583498, "global_step": 5875, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04129986837506294, "train_loss_bc": 0.041282664984464645, "train_loss_llm": 0.001720428466796875, "grad_norm": 0.042313020676374435, "global_step": 5876, "epoch": 163, "lr": 0.001} +{"train_loss": 0.03653700277209282, "train_loss_bc": 0.03653700277209282, "train_loss_llm": 0.0, "grad_norm": 0.016473829746246338, "global_step": 5877, "epoch": 163, "lr": 0.001} +{"train_loss": 0.047002602368593216, "train_loss_bc": 0.047002602368593216, "train_loss_llm": 0.0, "grad_norm": 0.02967521734535694, "global_step": 5878, "epoch": 163, "lr": 0.001} +{"train_loss": 0.03997569531202316, "train_loss_bc": 0.03996600583195686, "train_loss_llm": 0.0009689927101135254, "grad_norm": 0.04022657871246338, "global_step": 5879, "epoch": 163, "lr": 0.001} +{"train_loss": 0.03764685243368149, "train_loss_bc": 0.03763969987630844, "train_loss_llm": 0.0007150769233703613, "grad_norm": 0.04953659698367119, "global_step": 5880, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04359439015388489, "train_loss_bc": 0.043592192232608795, "train_loss_llm": 0.00021988153457641602, "grad_norm": 0.015031184069812298, "global_step": 5881, "epoch": 163, "lr": 0.001} +{"train_loss": 0.03574607893824577, "train_loss_bc": 0.03572371229529381, "train_loss_llm": 0.002236664295196533, "grad_norm": 0.02451261319220066, "global_step": 5882, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04167957231402397, "train_loss_bc": 0.04167957231402397, "train_loss_llm": 0.0, "grad_norm": 0.028057623654603958, "global_step": 5883, "epoch": 163, "lr": 0.001} +{"train_loss": 0.039541181176900864, "train_loss_bc": 0.03953530266880989, "train_loss_llm": 0.0005878210067749023, "grad_norm": 0.04044761136174202, "global_step": 5884, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04137297719717026, "train_loss_bc": 0.04137297719717026, "train_loss_llm": 0.0, "grad_norm": 0.01880456879734993, "global_step": 5885, "epoch": 163, "lr": 0.001} +{"train_loss": 0.049421217292547226, "train_loss_bc": 0.0494142547249794, "train_loss_llm": 0.0006963908672332764, "grad_norm": 0.041245605796575546, "global_step": 5886, "epoch": 163, "lr": 0.001} +{"train_loss": 0.03604384511709213, "train_loss_bc": 0.03604384511709213, "train_loss_llm": 0.0, "grad_norm": 0.04571370780467987, "global_step": 5887, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04176567122340202, "train_loss_bc": 0.04176567122340202, "train_loss_llm": 0.0, "grad_norm": 0.05451396480202675, "global_step": 5888, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04070534557104111, "train_loss_bc": 0.04069516807794571, "train_loss_llm": 0.0010177195072174072, "grad_norm": 0.022582348436117172, "global_step": 5889, "epoch": 163, "lr": 0.001} +{"train_loss": 0.03537518531084061, "train_loss_bc": 0.03537518531084061, "train_loss_llm": 0.0, "grad_norm": 0.04002382233738899, "global_step": 5890, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04443163052201271, "train_loss_bc": 0.044425930827856064, "train_loss_llm": 0.0005698800086975098, "grad_norm": 0.04614005982875824, "global_step": 5891, "epoch": 163, "lr": 0.001} +{"train_loss": 0.040547698736190796, "train_loss_bc": 0.0405266098678112, "train_loss_llm": 0.0021088123321533203, "grad_norm": 0.058702729642391205, "global_step": 5892, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04423382505774498, "train_loss_bc": 0.04423382505774498, "train_loss_llm": 0.0, "grad_norm": 0.020660171285271645, "global_step": 5893, "epoch": 163, "lr": 0.001} +{"train_loss": 0.047298576682806015, "train_loss_bc": 0.047250211238861084, "train_loss_llm": 0.004836440086364746, "grad_norm": 0.026100602000951767, "global_step": 5894, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04795597493648529, "train_loss_bc": 0.04795597493648529, "train_loss_llm": 0.0, "grad_norm": 0.030320817604660988, "global_step": 5895, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04207807406783104, "train_loss_bc": 0.042067814618349075, "train_loss_llm": 0.0010258257389068604, "grad_norm": 0.034977298229932785, "global_step": 5896, "epoch": 163, "lr": 0.001} +{"train_loss": 0.041718579828739166, "train_loss_bc": 0.041718579828739166, "train_loss_llm": 0.0, "grad_norm": 0.01633562706410885, "global_step": 5897, "epoch": 163, "lr": 0.001} +{"train_loss": 0.045052092522382736, "train_loss_bc": 0.045052092522382736, "train_loss_llm": 0.0, "grad_norm": 0.0301350150257349, "global_step": 5898, "epoch": 163, "lr": 0.001} +{"train_loss": 0.036417439579963684, "train_loss_bc": 0.036402322351932526, "train_loss_llm": 0.0015117526054382324, "grad_norm": 0.04168165102601051, "global_step": 5899, "epoch": 163, "lr": 0.001} +{"train_loss": 0.044825587421655655, "train_loss_bc": 0.044825587421655655, "train_loss_llm": 0.0, "grad_norm": 0.05094734951853752, "global_step": 5900, "epoch": 163, "lr": 0.001} +{"train_loss": 0.038347743451595306, "train_loss_bc": 0.038347743451595306, "train_loss_llm": 0.0, "grad_norm": 0.01933056302368641, "global_step": 5901, "epoch": 163, "lr": 0.001} +{"train_loss": 0.041637122631073, "train_loss_bc": 0.041637122631073, "train_loss_llm": 0.0, "grad_norm": 0.022228607907891273, "global_step": 5902, "epoch": 163, "lr": 0.001} +{"train_loss": 0.04128640310631858, "train_loss_bc": 0.03313194960355759, "train_loss_llm": 0.0, "grad_norm": 0.0288790762424469, "global_step": 5903, "epoch": 163, "lr": 0.001, "val_loss": 0.05572982877492905} +{"train_loss": 0.05080820247530937, "train_loss_bc": 0.05077071115374565, "train_loss_llm": 0.003749281167984009, "grad_norm": 0.03500792756676674, "global_step": 5904, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04260878264904022, "train_loss_bc": 0.04260878264904022, "train_loss_llm": 0.0, "grad_norm": 0.01942741870880127, "global_step": 5905, "epoch": 164, "lr": 0.001} +{"train_loss": 0.03732817992568016, "train_loss_bc": 0.03732817992568016, "train_loss_llm": 0.0, "grad_norm": 0.03489677235484123, "global_step": 5906, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04344358667731285, "train_loss_bc": 0.04344358667731285, "train_loss_llm": 0.0, "grad_norm": 0.0467473529279232, "global_step": 5907, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04264167323708534, "train_loss_bc": 0.04264167323708534, "train_loss_llm": 0.0, "grad_norm": 0.06094028055667877, "global_step": 5908, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04404420405626297, "train_loss_bc": 0.04402932524681091, "train_loss_llm": 0.0014879703521728516, "grad_norm": 0.01444198191165924, "global_step": 5909, "epoch": 164, "lr": 0.001} +{"train_loss": 0.043112002313137054, "train_loss_bc": 0.043112002313137054, "train_loss_llm": 0.0, "grad_norm": 0.023638421669602394, "global_step": 5910, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04616574943065643, "train_loss_bc": 0.04616574943065643, "train_loss_llm": 0.0, "grad_norm": 0.03817038610577583, "global_step": 5911, "epoch": 164, "lr": 0.001} +{"train_loss": 0.03734302893280983, "train_loss_bc": 0.03733693063259125, "train_loss_llm": 0.0006099939346313477, "grad_norm": 0.051283951848745346, "global_step": 5912, "epoch": 164, "lr": 0.001} +{"train_loss": 0.044337645173072815, "train_loss_bc": 0.04430045187473297, "train_loss_llm": 0.003719240427017212, "grad_norm": 0.019102083519101143, "global_step": 5913, "epoch": 164, "lr": 0.001} +{"train_loss": 0.03308229520916939, "train_loss_bc": 0.03308229520916939, "train_loss_llm": 0.0, "grad_norm": 0.026218654587864876, "global_step": 5914, "epoch": 164, "lr": 0.001} +{"train_loss": 0.03670387342572212, "train_loss_bc": 0.03670387342572212, "train_loss_llm": 0.0, "grad_norm": 0.029763929545879364, "global_step": 5915, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04131637513637543, "train_loss_bc": 0.04131637513637543, "train_loss_llm": 0.0, "grad_norm": 0.03399128466844559, "global_step": 5916, "epoch": 164, "lr": 0.001} +{"train_loss": 0.03983594849705696, "train_loss_bc": 0.03983088955283165, "train_loss_llm": 0.000505983829498291, "grad_norm": 0.019755616784095764, "global_step": 5917, "epoch": 164, "lr": 0.001} +{"train_loss": 0.041827041655778885, "train_loss_bc": 0.041827041655778885, "train_loss_llm": 0.0, "grad_norm": 0.03176337853074074, "global_step": 5918, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04260179400444031, "train_loss_bc": 0.04260179400444031, "train_loss_llm": 0.0, "grad_norm": 0.039083026349544525, "global_step": 5919, "epoch": 164, "lr": 0.001} +{"train_loss": 0.040884990245103836, "train_loss_bc": 0.040876127779483795, "train_loss_llm": 0.000886082649230957, "grad_norm": 0.0459258109331131, "global_step": 5920, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04294948652386665, "train_loss_bc": 0.04294948652386665, "train_loss_llm": 0.0, "grad_norm": 0.022053979337215424, "global_step": 5921, "epoch": 164, "lr": 0.001} +{"train_loss": 0.03666740283370018, "train_loss_bc": 0.03666740283370018, "train_loss_llm": 0.0, "grad_norm": 0.02856740541756153, "global_step": 5922, "epoch": 164, "lr": 0.001} +{"train_loss": 0.041143886744976044, "train_loss_bc": 0.04114386439323425, "train_loss_llm": 2.294778823852539e-06, "grad_norm": 0.04119967296719551, "global_step": 5923, "epoch": 164, "lr": 0.001} +{"train_loss": 0.03157466650009155, "train_loss_bc": 0.03157466650009155, "train_loss_llm": 0.0, "grad_norm": 0.0497935451567173, "global_step": 5924, "epoch": 164, "lr": 0.001} +{"train_loss": 0.042447544634342194, "train_loss_bc": 0.042447544634342194, "train_loss_llm": 0.0, "grad_norm": 0.022936290130019188, "global_step": 5925, "epoch": 164, "lr": 0.001} +{"train_loss": 0.03867166489362717, "train_loss_bc": 0.03867022320628166, "train_loss_llm": 0.00014400482177734375, "grad_norm": 0.034507010132074356, "global_step": 5926, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04407985880970955, "train_loss_bc": 0.044065285474061966, "train_loss_llm": 0.0014574527740478516, "grad_norm": 0.04521980509161949, "global_step": 5927, "epoch": 164, "lr": 0.001} +{"train_loss": 0.042101435363292694, "train_loss_bc": 0.042071081697940826, "train_loss_llm": 0.0030353963375091553, "grad_norm": 0.057436682283878326, "global_step": 5928, "epoch": 164, "lr": 0.001} +{"train_loss": 0.03766504302620888, "train_loss_bc": 0.03766504302620888, "train_loss_llm": 0.0, "grad_norm": 0.014276550151407719, "global_step": 5929, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04049690067768097, "train_loss_bc": 0.04049690067768097, "train_loss_llm": 0.0, "grad_norm": 0.024388844147324562, "global_step": 5930, "epoch": 164, "lr": 0.001} +{"train_loss": 0.045651160180568695, "train_loss_bc": 0.045651160180568695, "train_loss_llm": 0.0, "grad_norm": 0.03632093966007233, "global_step": 5931, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04550299048423767, "train_loss_bc": 0.04550299048423767, "train_loss_llm": 0.0, "grad_norm": 0.0482630729675293, "global_step": 5932, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04264187067747116, "train_loss_bc": 0.04264147952198982, "train_loss_llm": 3.927946090698242e-05, "grad_norm": 0.019079072400927544, "global_step": 5933, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04164398834109306, "train_loss_bc": 0.04164398834109306, "train_loss_llm": 0.0, "grad_norm": 0.026606984436511993, "global_step": 5934, "epoch": 164, "lr": 0.001} +{"train_loss": 0.03363055735826492, "train_loss_bc": 0.03363055735826492, "train_loss_llm": 0.0, "grad_norm": 0.03542669489979744, "global_step": 5935, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04553982987999916, "train_loss_bc": 0.04552791267633438, "train_loss_llm": 0.001191645860671997, "grad_norm": 0.04111602157354355, "global_step": 5936, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04162272810935974, "train_loss_bc": 0.04162272810935974, "train_loss_llm": 0.0, "grad_norm": 0.016131501644849777, "global_step": 5937, "epoch": 164, "lr": 0.001} +{"train_loss": 0.045731738209724426, "train_loss_bc": 0.045731738209724426, "train_loss_llm": 0.0, "grad_norm": 0.028475558385252953, "global_step": 5938, "epoch": 164, "lr": 0.001} +{"train_loss": 0.04153634742316273, "train_loss_bc": 0.04746038094162941, "train_loss_llm": 0.0, "grad_norm": 0.03945918008685112, "global_step": 5939, "epoch": 164, "lr": 0.001, "val_loss": 0.055280718952417374} +{"train_loss": 0.04009193554520607, "train_loss_bc": 0.04009193554520607, "train_loss_llm": 0.0, "grad_norm": 0.0499153696000576, "global_step": 5940, "epoch": 165, "lr": 0.001} +{"train_loss": 0.035555727779865265, "train_loss_bc": 0.035555727779865265, "train_loss_llm": 0.0, "grad_norm": 0.01603706367313862, "global_step": 5941, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04474964365363121, "train_loss_bc": 0.04474964365363121, "train_loss_llm": 0.0, "grad_norm": 0.02535211481153965, "global_step": 5942, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04420144110918045, "train_loss_bc": 0.04420144110918045, "train_loss_llm": 0.0, "grad_norm": 0.0335705392062664, "global_step": 5943, "epoch": 165, "lr": 0.001} +{"train_loss": 0.044943615794181824, "train_loss_bc": 0.04492868110537529, "train_loss_llm": 0.0014934837818145752, "grad_norm": 0.048583656549453735, "global_step": 5944, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04066106304526329, "train_loss_bc": 0.04066106304526329, "train_loss_llm": 0.0, "grad_norm": 0.023001128807663918, "global_step": 5945, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04295571520924568, "train_loss_bc": 0.04295571520924568, "train_loss_llm": 0.0, "grad_norm": 0.03496314212679863, "global_step": 5946, "epoch": 165, "lr": 0.001} +{"train_loss": 0.03471614420413971, "train_loss_bc": 0.03471614420413971, "train_loss_llm": 0.0, "grad_norm": 0.04661405831575394, "global_step": 5947, "epoch": 165, "lr": 0.001} +{"train_loss": 0.035948771983385086, "train_loss_bc": 0.035948771983385086, "train_loss_llm": 0.0, "grad_norm": 0.06299809366464615, "global_step": 5948, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04071702063083649, "train_loss_bc": 0.04069474712014198, "train_loss_llm": 0.002227216958999634, "grad_norm": 0.021580249071121216, "global_step": 5949, "epoch": 165, "lr": 0.001} +{"train_loss": 0.043916184455156326, "train_loss_bc": 0.043916184455156326, "train_loss_llm": 0.0, "grad_norm": 0.039575546979904175, "global_step": 5950, "epoch": 165, "lr": 0.001} +{"train_loss": 0.041852157562971115, "train_loss_bc": 0.041852157562971115, "train_loss_llm": 0.0, "grad_norm": 0.04667848348617554, "global_step": 5951, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04558190330862999, "train_loss_bc": 0.04556046798825264, "train_loss_llm": 0.0021434128284454346, "grad_norm": 0.057471729815006256, "global_step": 5952, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04701175168156624, "train_loss_bc": 0.04701175168156624, "train_loss_llm": 0.0, "grad_norm": 0.019039873033761978, "global_step": 5953, "epoch": 165, "lr": 0.001} +{"train_loss": 0.03935224935412407, "train_loss_bc": 0.03935224935412407, "train_loss_llm": 0.0, "grad_norm": 0.02891378104686737, "global_step": 5954, "epoch": 165, "lr": 0.001} +{"train_loss": 0.043352674692869186, "train_loss_bc": 0.04334626346826553, "train_loss_llm": 0.000641167163848877, "grad_norm": 0.03927403688430786, "global_step": 5955, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04226972162723541, "train_loss_bc": 0.04226972162723541, "train_loss_llm": 0.0, "grad_norm": 0.04408726096153259, "global_step": 5956, "epoch": 165, "lr": 0.001} +{"train_loss": 0.043744221329689026, "train_loss_bc": 0.043744221329689026, "train_loss_llm": 0.0, "grad_norm": 0.018144391477108, "global_step": 5957, "epoch": 165, "lr": 0.001} +{"train_loss": 0.03777432069182396, "train_loss_bc": 0.03777432069182396, "train_loss_llm": 0.0, "grad_norm": 0.02702212519943714, "global_step": 5958, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04642916098237038, "train_loss_bc": 0.04642916098237038, "train_loss_llm": 0.0, "grad_norm": 0.039763882756233215, "global_step": 5959, "epoch": 165, "lr": 0.001} +{"train_loss": 0.045471273362636566, "train_loss_bc": 0.045471273362636566, "train_loss_llm": 0.0, "grad_norm": 0.05081550031900406, "global_step": 5960, "epoch": 165, "lr": 0.001} +{"train_loss": 0.03882751613855362, "train_loss_bc": 0.03881708160042763, "train_loss_llm": 0.0010433793067932129, "grad_norm": 0.021527526900172234, "global_step": 5961, "epoch": 165, "lr": 0.001} +{"train_loss": 0.03126503527164459, "train_loss_bc": 0.031264644116163254, "train_loss_llm": 3.9130449295043945e-05, "grad_norm": 0.027740707620978355, "global_step": 5962, "epoch": 165, "lr": 0.001} +{"train_loss": 0.03786921128630638, "train_loss_bc": 0.03785810247063637, "train_loss_llm": 0.0011108815670013428, "grad_norm": 0.040434058755636215, "global_step": 5963, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04076527804136276, "train_loss_bc": 0.04075890779495239, "train_loss_llm": 0.0006369650363922119, "grad_norm": 0.05432182177901268, "global_step": 5964, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04437245801091194, "train_loss_bc": 0.04437245801091194, "train_loss_llm": 0.0, "grad_norm": 0.015852611511945724, "global_step": 5965, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04269938915967941, "train_loss_bc": 0.04269938915967941, "train_loss_llm": 0.0, "grad_norm": 0.026401737704873085, "global_step": 5966, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04437058046460152, "train_loss_bc": 0.044327687472105026, "train_loss_llm": 0.004289388656616211, "grad_norm": 0.03725593164563179, "global_step": 5967, "epoch": 165, "lr": 0.001} +{"train_loss": 0.03853854537010193, "train_loss_bc": 0.03851824626326561, "train_loss_llm": 0.002029985189437866, "grad_norm": 0.05190473794937134, "global_step": 5968, "epoch": 165, "lr": 0.001} +{"train_loss": 0.037315141409635544, "train_loss_bc": 0.037315141409635544, "train_loss_llm": 0.0, "grad_norm": 0.012378561310470104, "global_step": 5969, "epoch": 165, "lr": 0.001} +{"train_loss": 0.038004033267498016, "train_loss_bc": 0.038004033267498016, "train_loss_llm": 0.0, "grad_norm": 0.0223099272698164, "global_step": 5970, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04445981606841087, "train_loss_bc": 0.0444241501390934, "train_loss_llm": 0.0035665035247802734, "grad_norm": 0.033284611999988556, "global_step": 5971, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04310440272092819, "train_loss_bc": 0.04310440272092819, "train_loss_llm": 0.0, "grad_norm": 0.039541058242321014, "global_step": 5972, "epoch": 165, "lr": 0.001} +{"train_loss": 0.043170057237148285, "train_loss_bc": 0.043170057237148285, "train_loss_llm": 0.0, "grad_norm": 0.020906060934066772, "global_step": 5973, "epoch": 165, "lr": 0.001} +{"train_loss": 0.03648294135928154, "train_loss_bc": 0.03648294135928154, "train_loss_llm": 0.0, "grad_norm": 0.02915949560701847, "global_step": 5974, "epoch": 165, "lr": 0.001} +{"train_loss": 0.04146113174243106, "train_loss_bc": 0.05004069581627846, "train_loss_llm": 0.0018941760063171387, "grad_norm": 0.036933667957782745, "global_step": 5975, "epoch": 165, "lr": 0.001, "train/cumulative_reward": 85.55775043078434, "train/mean_score": 15.832011789554707, "train/success_rate": 0.3, "test/cumulative_reward": 42.08793555630357, "test/mean_score": 9.386631542111191, "test/success_rate": 0.22, "val_loss": 0.05544717237353325, "train_action_mse_error": 0.04494204744696617} +{"train_loss": 0.03614998236298561, "train_loss_bc": 0.03614184260368347, "train_loss_llm": 0.0008141398429870605, "grad_norm": 0.03878145292401314, "global_step": 5976, "epoch": 166, "lr": 0.001} +{"train_loss": 0.03530122712254524, "train_loss_bc": 0.03530122712254524, "train_loss_llm": 0.0, "grad_norm": 0.015574350021779537, "global_step": 5977, "epoch": 166, "lr": 0.001} +{"train_loss": 0.0388171412050724, "train_loss_bc": 0.0388171412050724, "train_loss_llm": 0.0, "grad_norm": 0.02416364476084709, "global_step": 5978, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04226154088973999, "train_loss_bc": 0.04226154088973999, "train_loss_llm": 0.0, "grad_norm": 0.03482597693800926, "global_step": 5979, "epoch": 166, "lr": 0.001} +{"train_loss": 0.044248711317777634, "train_loss_bc": 0.044248711317777634, "train_loss_llm": 0.0, "grad_norm": 0.042629700154066086, "global_step": 5980, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04416380077600479, "train_loss_bc": 0.04416380077600479, "train_loss_llm": 0.0, "grad_norm": 0.021508485078811646, "global_step": 5981, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04567514359951019, "train_loss_bc": 0.04567514359951019, "train_loss_llm": 0.0, "grad_norm": 0.03675292432308197, "global_step": 5982, "epoch": 166, "lr": 0.001} +{"train_loss": 0.037191130220890045, "train_loss_bc": 0.0371909961104393, "train_loss_llm": 1.3560056686401367e-05, "grad_norm": 0.04408849775791168, "global_step": 5983, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04974451661109924, "train_loss_bc": 0.04974451661109924, "train_loss_llm": 0.0, "grad_norm": 0.05096451938152313, "global_step": 5984, "epoch": 166, "lr": 0.001} +{"train_loss": 0.03812071681022644, "train_loss_bc": 0.03810204938054085, "train_loss_llm": 0.0018668770790100098, "grad_norm": 0.016969196498394012, "global_step": 5985, "epoch": 166, "lr": 0.001} +{"train_loss": 0.03204990178346634, "train_loss_bc": 0.03204990178346634, "train_loss_llm": 0.0, "grad_norm": 0.024755053222179413, "global_step": 5986, "epoch": 166, "lr": 0.001} +{"train_loss": 0.041134461760520935, "train_loss_bc": 0.041134461760520935, "train_loss_llm": 0.0, "grad_norm": 0.03436741232872009, "global_step": 5987, "epoch": 166, "lr": 0.001} +{"train_loss": 0.045975543558597565, "train_loss_bc": 0.045975543558597565, "train_loss_llm": 0.0, "grad_norm": 0.042887941002845764, "global_step": 5988, "epoch": 166, "lr": 0.001} +{"train_loss": 0.03738540783524513, "train_loss_bc": 0.03737742453813553, "train_loss_llm": 0.0007984340190887451, "grad_norm": 0.016679799184203148, "global_step": 5989, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04377000406384468, "train_loss_bc": 0.04377000406384468, "train_loss_llm": 0.0, "grad_norm": 0.02621326595544815, "global_step": 5990, "epoch": 166, "lr": 0.001} +{"train_loss": 0.03846075013279915, "train_loss_bc": 0.03846075013279915, "train_loss_llm": 0.0, "grad_norm": 0.03489437699317932, "global_step": 5991, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04015779122710228, "train_loss_bc": 0.04015779122710228, "train_loss_llm": 0.0, "grad_norm": 0.0466839037835598, "global_step": 5992, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04978703334927559, "train_loss_bc": 0.04977673292160034, "train_loss_llm": 0.0010299384593963623, "grad_norm": 0.02031862363219261, "global_step": 5993, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04048408567905426, "train_loss_bc": 0.040478501468896866, "train_loss_llm": 0.0005584657192230225, "grad_norm": 0.03133592754602432, "global_step": 5994, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04191247746348381, "train_loss_bc": 0.04191247746348381, "train_loss_llm": 0.0, "grad_norm": 0.03910217806696892, "global_step": 5995, "epoch": 166, "lr": 0.001} +{"train_loss": 0.0391499325633049, "train_loss_bc": 0.0391499325633049, "train_loss_llm": 0.0, "grad_norm": 0.04461236670613289, "global_step": 5996, "epoch": 166, "lr": 0.001} +{"train_loss": 0.045118559151887894, "train_loss_bc": 0.04511358588933945, "train_loss_llm": 0.0004974603652954102, "grad_norm": 0.018762610852718353, "global_step": 5997, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04456626623868942, "train_loss_bc": 0.04455069079995155, "train_loss_llm": 0.001557469367980957, "grad_norm": 0.023936398327350616, "global_step": 5998, "epoch": 166, "lr": 0.001} +{"train_loss": 0.0428200326859951, "train_loss_bc": 0.042811572551727295, "train_loss_llm": 0.0008460879325866699, "grad_norm": 0.030832378193736076, "global_step": 5999, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04466581717133522, "train_loss_bc": 0.04466581717133522, "train_loss_llm": 0.0, "grad_norm": 0.036317504942417145, "global_step": 6000, "epoch": 166, "lr": 0.001} +{"train_loss": 0.0450112447142601, "train_loss_bc": 0.0450112447142601, "train_loss_llm": 0.0, "grad_norm": 0.017026733607053757, "global_step": 6001, "epoch": 166, "lr": 0.001} +{"train_loss": 0.043455418199300766, "train_loss_bc": 0.0434463731944561, "train_loss_llm": 0.0009044110774993896, "grad_norm": 0.02239573746919632, "global_step": 6002, "epoch": 166, "lr": 0.001} +{"train_loss": 0.03300575166940689, "train_loss_bc": 0.03299245238304138, "train_loss_llm": 0.0013298094272613525, "grad_norm": 0.02903350628912449, "global_step": 6003, "epoch": 166, "lr": 0.001} +{"train_loss": 0.0381341278553009, "train_loss_bc": 0.0381341278553009, "train_loss_llm": 0.0, "grad_norm": 0.03878801316022873, "global_step": 6004, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04128926247358322, "train_loss_bc": 0.04128926247358322, "train_loss_llm": 0.0, "grad_norm": 0.018202314153313637, "global_step": 6005, "epoch": 166, "lr": 0.001} +{"train_loss": 0.038931623101234436, "train_loss_bc": 0.038931623101234436, "train_loss_llm": 0.0, "grad_norm": 0.030813150107860565, "global_step": 6006, "epoch": 166, "lr": 0.001} +{"train_loss": 0.03848273307085037, "train_loss_bc": 0.03848273307085037, "train_loss_llm": 0.0, "grad_norm": 0.0373665876686573, "global_step": 6007, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04731590300798416, "train_loss_bc": 0.04731590300798416, "train_loss_llm": 0.0, "grad_norm": 0.050419170409440994, "global_step": 6008, "epoch": 166, "lr": 0.001} +{"train_loss": 0.03457748889923096, "train_loss_bc": 0.03457748889923096, "train_loss_llm": 0.0, "grad_norm": 0.01583842560648918, "global_step": 6009, "epoch": 166, "lr": 0.001} +{"train_loss": 0.03981653228402138, "train_loss_bc": 0.03981577232480049, "train_loss_llm": 7.608532905578613e-05, "grad_norm": 0.021333206444978714, "global_step": 6010, "epoch": 166, "lr": 0.001} +{"train_loss": 0.04129435080620977, "train_loss_bc": 0.04746456816792488, "train_loss_llm": 0.0, "grad_norm": 0.030723027884960175, "global_step": 6011, "epoch": 166, "lr": 0.001, "val_loss": 0.055471234023571014} +{"train_loss": 0.04801616817712784, "train_loss_bc": 0.04801616817712784, "train_loss_llm": 0.0, "grad_norm": 0.038314588367938995, "global_step": 6012, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04670984670519829, "train_loss_bc": 0.04670661315321922, "train_loss_llm": 0.0003234744071960449, "grad_norm": 0.018727604299783707, "global_step": 6013, "epoch": 167, "lr": 0.001} +{"train_loss": 0.043912578374147415, "train_loss_bc": 0.043904416263103485, "train_loss_llm": 0.0008162856101989746, "grad_norm": 0.02559063583612442, "global_step": 6014, "epoch": 167, "lr": 0.001} +{"train_loss": 0.044060274958610535, "train_loss_bc": 0.04405693709850311, "train_loss_llm": 0.0003338754177093506, "grad_norm": 0.03197087347507477, "global_step": 6015, "epoch": 167, "lr": 0.001} +{"train_loss": 0.03878794610500336, "train_loss_bc": 0.03878794610500336, "train_loss_llm": 0.0, "grad_norm": 0.03905881568789482, "global_step": 6016, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04520762339234352, "train_loss_bc": 0.04520762339234352, "train_loss_llm": 0.0, "grad_norm": 0.018336810171604156, "global_step": 6017, "epoch": 167, "lr": 0.001} +{"train_loss": 0.03410020098090172, "train_loss_bc": 0.03409915789961815, "train_loss_llm": 0.00010442733764648438, "grad_norm": 0.01930549368262291, "global_step": 6018, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04211974889039993, "train_loss_bc": 0.04211974889039993, "train_loss_llm": 0.0, "grad_norm": 0.029541565105319023, "global_step": 6019, "epoch": 167, "lr": 0.001} +{"train_loss": 0.046981438994407654, "train_loss_bc": 0.04696568846702576, "train_loss_llm": 0.0015751123428344727, "grad_norm": 0.04162178933620453, "global_step": 6020, "epoch": 167, "lr": 0.001} +{"train_loss": 0.03916473686695099, "train_loss_bc": 0.03916473686695099, "train_loss_llm": 0.0, "grad_norm": 0.014481362886726856, "global_step": 6021, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04358970373868942, "train_loss_bc": 0.04358970373868942, "train_loss_llm": 0.0, "grad_norm": 0.0245825108140707, "global_step": 6022, "epoch": 167, "lr": 0.001} +{"train_loss": 0.02950269728899002, "train_loss_bc": 0.02946588210761547, "train_loss_llm": 0.0036816000938415527, "grad_norm": 0.02974226139485836, "global_step": 6023, "epoch": 167, "lr": 0.001} +{"train_loss": 0.043816033750772476, "train_loss_bc": 0.043809495866298676, "train_loss_llm": 0.0006537735462188721, "grad_norm": 0.03522428497672081, "global_step": 6024, "epoch": 167, "lr": 0.001} +{"train_loss": 0.039623551070690155, "train_loss_bc": 0.039623551070690155, "train_loss_llm": 0.0, "grad_norm": 0.016484400257468224, "global_step": 6025, "epoch": 167, "lr": 0.001} +{"train_loss": 0.03519584238529205, "train_loss_bc": 0.035192638635635376, "train_loss_llm": 0.0003205239772796631, "grad_norm": 0.023639971390366554, "global_step": 6026, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04441887512803078, "train_loss_bc": 0.044408299028873444, "train_loss_llm": 0.0010575056076049805, "grad_norm": 0.03407133370637894, "global_step": 6027, "epoch": 167, "lr": 0.001} +{"train_loss": 0.042456336319446564, "train_loss_bc": 0.042456336319446564, "train_loss_llm": 0.0, "grad_norm": 0.042760804295539856, "global_step": 6028, "epoch": 167, "lr": 0.001} +{"train_loss": 0.051957301795482635, "train_loss_bc": 0.051957301795482635, "train_loss_llm": 0.0, "grad_norm": 0.02050621621310711, "global_step": 6029, "epoch": 167, "lr": 0.001} +{"train_loss": 0.032359909266233444, "train_loss_bc": 0.0323423407971859, "train_loss_llm": 0.0017569363117218018, "grad_norm": 0.028234606608748436, "global_step": 6030, "epoch": 167, "lr": 0.001} +{"train_loss": 0.03573865815997124, "train_loss_bc": 0.03573865815997124, "train_loss_llm": 0.0, "grad_norm": 0.03733742982149124, "global_step": 6031, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04382546991109848, "train_loss_bc": 0.04382546991109848, "train_loss_llm": 0.0, "grad_norm": 0.04890693351626396, "global_step": 6032, "epoch": 167, "lr": 0.001} +{"train_loss": 0.041316550225019455, "train_loss_bc": 0.041316550225019455, "train_loss_llm": 0.0, "grad_norm": 0.016978975385427475, "global_step": 6033, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04257507994771004, "train_loss_bc": 0.04257507994771004, "train_loss_llm": 0.0, "grad_norm": 0.02548138052225113, "global_step": 6034, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04239482805132866, "train_loss_bc": 0.04239482805132866, "train_loss_llm": 0.0, "grad_norm": 0.033537112176418304, "global_step": 6035, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04448799416422844, "train_loss_bc": 0.04448799416422844, "train_loss_llm": 0.0, "grad_norm": 0.04068583995103836, "global_step": 6036, "epoch": 167, "lr": 0.001} +{"train_loss": 0.0363667868077755, "train_loss_bc": 0.03635723888874054, "train_loss_llm": 0.0009546577930450439, "grad_norm": 0.015489109791815281, "global_step": 6037, "epoch": 167, "lr": 0.001} +{"train_loss": 0.03620786592364311, "train_loss_bc": 0.03620786592364311, "train_loss_llm": 0.0, "grad_norm": 0.02423439733684063, "global_step": 6038, "epoch": 167, "lr": 0.001} +{"train_loss": 0.044600676745176315, "train_loss_bc": 0.044600676745176315, "train_loss_llm": 0.0, "grad_norm": 0.030283775180578232, "global_step": 6039, "epoch": 167, "lr": 0.001} +{"train_loss": 0.0350455678999424, "train_loss_bc": 0.03500581532716751, "train_loss_llm": 0.003975242376327515, "grad_norm": 0.0365392304956913, "global_step": 6040, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04677267372608185, "train_loss_bc": 0.046772632747888565, "train_loss_llm": 4.082918167114258e-06, "grad_norm": 0.016565851867198944, "global_step": 6041, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04262750223278999, "train_loss_bc": 0.04260629415512085, "train_loss_llm": 0.002120763063430786, "grad_norm": 0.024268535897135735, "global_step": 6042, "epoch": 167, "lr": 0.001} +{"train_loss": 0.034908585250377655, "train_loss_bc": 0.034908585250377655, "train_loss_llm": 0.0, "grad_norm": 0.029583405703306198, "global_step": 6043, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04059500992298126, "train_loss_bc": 0.04059500992298126, "train_loss_llm": 0.0, "grad_norm": 0.03823854774236679, "global_step": 6044, "epoch": 167, "lr": 0.001} +{"train_loss": 0.03929206728935242, "train_loss_bc": 0.039288319647312164, "train_loss_llm": 0.00037467479705810547, "grad_norm": 0.01648200862109661, "global_step": 6045, "epoch": 167, "lr": 0.001} +{"train_loss": 0.03905436024069786, "train_loss_bc": 0.03905436024069786, "train_loss_llm": 0.0, "grad_norm": 0.02856544405221939, "global_step": 6046, "epoch": 167, "lr": 0.001} +{"train_loss": 0.04113833295802275, "train_loss_bc": 0.04318949580192566, "train_loss_llm": 0.0, "grad_norm": 0.03625580295920372, "global_step": 6047, "epoch": 167, "lr": 0.001, "val_loss": 0.055775079876184464} +{"train_loss": 0.03017476201057434, "train_loss_bc": 0.03017476201057434, "train_loss_llm": 0.0, "grad_norm": 0.04745505750179291, "global_step": 6048, "epoch": 168, "lr": 0.001} +{"train_loss": 0.04678152874112129, "train_loss_bc": 0.0467788390815258, "train_loss_llm": 0.0002690553665161133, "grad_norm": 0.014705546200275421, "global_step": 6049, "epoch": 168, "lr": 0.001} +{"train_loss": 0.03936032950878143, "train_loss_bc": 0.03936032950878143, "train_loss_llm": 0.0, "grad_norm": 0.02605374902486801, "global_step": 6050, "epoch": 168, "lr": 0.001} +{"train_loss": 0.04629208520054817, "train_loss_bc": 0.04629208520054817, "train_loss_llm": 0.0, "grad_norm": 0.034000203013420105, "global_step": 6051, "epoch": 168, "lr": 0.001} +{"train_loss": 0.04151143878698349, "train_loss_bc": 0.04150069132447243, "train_loss_llm": 0.001074671745300293, "grad_norm": 0.04185493290424347, "global_step": 6052, "epoch": 168, "lr": 0.001} +{"train_loss": 0.040641605854034424, "train_loss_bc": 0.040641605854034424, "train_loss_llm": 0.0, "grad_norm": 0.02096107229590416, "global_step": 6053, "epoch": 168, "lr": 0.001} +{"train_loss": 0.0436011403799057, "train_loss_bc": 0.043596051633358, "train_loss_llm": 0.0005090534687042236, "grad_norm": 0.028346797451376915, "global_step": 6054, "epoch": 168, "lr": 0.001} +{"train_loss": 0.03702712804079056, "train_loss_bc": 0.03702712804079056, "train_loss_llm": 0.0, "grad_norm": 0.032475464046001434, "global_step": 6055, "epoch": 168, "lr": 0.001} +{"train_loss": 0.044106606394052505, "train_loss_bc": 0.044106606394052505, "train_loss_llm": 0.0, "grad_norm": 0.043391790241003036, "global_step": 6056, "epoch": 168, "lr": 0.001} +{"train_loss": 0.039029840379953384, "train_loss_bc": 0.03902522847056389, "train_loss_llm": 0.00046133995056152344, "grad_norm": 0.017305247485637665, "global_step": 6057, "epoch": 168, "lr": 0.001} +{"train_loss": 0.031020479276776314, "train_loss_bc": 0.031020479276776314, "train_loss_llm": 0.0, "grad_norm": 0.02380501851439476, "global_step": 6058, "epoch": 168, "lr": 0.001} +{"train_loss": 0.039417047053575516, "train_loss_bc": 0.03941365331411362, "train_loss_llm": 0.00033923983573913574, "grad_norm": 0.04002959653735161, "global_step": 6059, "epoch": 168, "lr": 0.001} +{"train_loss": 0.037832606583833694, "train_loss_bc": 0.037832606583833694, "train_loss_llm": 0.0, "grad_norm": 0.03674286603927612, "global_step": 6060, "epoch": 168, "lr": 0.001} +{"train_loss": 0.040087487548589706, "train_loss_bc": 0.04007855802774429, "train_loss_llm": 0.0008930563926696777, "grad_norm": 0.020504243671894073, "global_step": 6061, "epoch": 168, "lr": 0.001} +{"train_loss": 0.04952819272875786, "train_loss_bc": 0.04952819272875786, "train_loss_llm": 0.0, "grad_norm": 0.03515469282865524, "global_step": 6062, "epoch": 168, "lr": 0.001} +{"train_loss": 0.042949482798576355, "train_loss_bc": 0.04294195398688316, "train_loss_llm": 0.0007528960704803467, "grad_norm": 0.0396345816552639, "global_step": 6063, "epoch": 168, "lr": 0.001} +{"train_loss": 0.03904848173260689, "train_loss_bc": 0.03904848173260689, "train_loss_llm": 0.0, "grad_norm": 0.053950197994709015, "global_step": 6064, "epoch": 168, "lr": 0.001} +{"train_loss": 0.043622151017189026, "train_loss_bc": 0.04357461258769035, "train_loss_llm": 0.00475388765335083, "grad_norm": 0.01691490411758423, "global_step": 6065, "epoch": 168, "lr": 0.001} +{"train_loss": 0.03731391951441765, "train_loss_bc": 0.03731391951441765, "train_loss_llm": 0.0, "grad_norm": 0.02837425097823143, "global_step": 6066, "epoch": 168, "lr": 0.001} +{"train_loss": 0.035611528903245926, "train_loss_bc": 0.035611528903245926, "train_loss_llm": 0.0, "grad_norm": 0.03969801589846611, "global_step": 6067, "epoch": 168, "lr": 0.001} +{"train_loss": 0.038311079144477844, "train_loss_bc": 0.038311079144477844, "train_loss_llm": 0.0, "grad_norm": 0.04461033642292023, "global_step": 6068, "epoch": 168, "lr": 0.001} +{"train_loss": 0.039309170097112656, "train_loss_bc": 0.039309170097112656, "train_loss_llm": 0.0, "grad_norm": 0.017190173268318176, "global_step": 6069, "epoch": 168, "lr": 0.001} +{"train_loss": 0.04191037267446518, "train_loss_bc": 0.04191037267446518, "train_loss_llm": 0.0, "grad_norm": 0.02617420256137848, "global_step": 6070, "epoch": 168, "lr": 0.001} +{"train_loss": 0.046194612979888916, "train_loss_bc": 0.046194612979888916, "train_loss_llm": 0.0, "grad_norm": 0.03335433453321457, "global_step": 6071, "epoch": 168, "lr": 0.001} +{"train_loss": 0.03844486176967621, "train_loss_bc": 0.03844486176967621, "train_loss_llm": 0.0, "grad_norm": 0.04044965282082558, "global_step": 6072, "epoch": 168, "lr": 0.001} +{"train_loss": 0.03469737619161606, "train_loss_bc": 0.03469737619161606, "train_loss_llm": 0.0, "grad_norm": 0.022770829498767853, "global_step": 6073, "epoch": 168, "lr": 0.001} +{"train_loss": 0.04152723401784897, "train_loss_bc": 0.04152723401784897, "train_loss_llm": 0.0, "grad_norm": 0.032396864145994186, "global_step": 6074, "epoch": 168, "lr": 0.001} +{"train_loss": 0.048047907650470734, "train_loss_bc": 0.048047907650470734, "train_loss_llm": 0.0, "grad_norm": 0.03743424639105797, "global_step": 6075, "epoch": 168, "lr": 0.001} +{"train_loss": 0.042255647480487823, "train_loss_bc": 0.04225081205368042, "train_loss_llm": 0.00048342347145080566, "grad_norm": 0.043984171003103256, "global_step": 6076, "epoch": 168, "lr": 0.001} +{"train_loss": 0.04874540492892265, "train_loss_bc": 0.04874540492892265, "train_loss_llm": 0.0, "grad_norm": 0.020233063027262688, "global_step": 6077, "epoch": 168, "lr": 0.001} +{"train_loss": 0.048422373831272125, "train_loss_bc": 0.048422373831272125, "train_loss_llm": 0.0, "grad_norm": 0.035899143666028976, "global_step": 6078, "epoch": 168, "lr": 0.001} +{"train_loss": 0.03962995857000351, "train_loss_bc": 0.03958931937813759, "train_loss_llm": 0.004063963890075684, "grad_norm": 0.04558172821998596, "global_step": 6079, "epoch": 168, "lr": 0.001} +{"train_loss": 0.03839842602610588, "train_loss_bc": 0.03838539496064186, "train_loss_llm": 0.0013032853603363037, "grad_norm": 0.05360187217593193, "global_step": 6080, "epoch": 168, "lr": 0.001} +{"train_loss": 0.044408489018678665, "train_loss_bc": 0.04440772905945778, "train_loss_llm": 7.599592208862305e-05, "grad_norm": 0.016930071637034416, "global_step": 6081, "epoch": 168, "lr": 0.001} +{"train_loss": 0.04932769387960434, "train_loss_bc": 0.04932769387960434, "train_loss_llm": 0.0, "grad_norm": 0.028669284656643867, "global_step": 6082, "epoch": 168, "lr": 0.001} +{"train_loss": 0.0412390864868131, "train_loss_bc": 0.04001866281032562, "train_loss_llm": 0.0, "grad_norm": 0.04124843329191208, "global_step": 6083, "epoch": 168, "lr": 0.001, "val_loss": 0.055080167949199677} +{"train_loss": 0.04342693090438843, "train_loss_bc": 0.04342693090438843, "train_loss_llm": 0.0, "grad_norm": 0.04865172505378723, "global_step": 6084, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04024358093738556, "train_loss_bc": 0.040232058614492416, "train_loss_llm": 0.0011520683765411377, "grad_norm": 0.01591934822499752, "global_step": 6085, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04843100160360336, "train_loss_bc": 0.04843100160360336, "train_loss_llm": 0.0, "grad_norm": 0.029791073873639107, "global_step": 6086, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04504230618476868, "train_loss_bc": 0.045015670359134674, "train_loss_llm": 0.0026637017726898193, "grad_norm": 0.04190465807914734, "global_step": 6087, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04046287760138512, "train_loss_bc": 0.04046287760138512, "train_loss_llm": 0.0, "grad_norm": 0.05138895660638809, "global_step": 6088, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04622773826122284, "train_loss_bc": 0.04621211439371109, "train_loss_llm": 0.0015622973442077637, "grad_norm": 0.020815888419747353, "global_step": 6089, "epoch": 169, "lr": 0.001} +{"train_loss": 0.03879209980368614, "train_loss_bc": 0.03879209980368614, "train_loss_llm": 0.0, "grad_norm": 0.0335407592356205, "global_step": 6090, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04157627746462822, "train_loss_bc": 0.04157627746462822, "train_loss_llm": 0.0, "grad_norm": 0.044552527368068695, "global_step": 6091, "epoch": 169, "lr": 0.001} +{"train_loss": 0.03583315759897232, "train_loss_bc": 0.03583315759897232, "train_loss_llm": 0.0, "grad_norm": 0.04924428090453148, "global_step": 6092, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04428062587976456, "train_loss_bc": 0.04425494000315666, "train_loss_llm": 0.0025687217712402344, "grad_norm": 0.018306540325284004, "global_step": 6093, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04871668666601181, "train_loss_bc": 0.04871668666601181, "train_loss_llm": 0.0, "grad_norm": 0.02649933099746704, "global_step": 6094, "epoch": 169, "lr": 0.001} +{"train_loss": 0.043385930359363556, "train_loss_bc": 0.043385930359363556, "train_loss_llm": 0.0, "grad_norm": 0.031536079943180084, "global_step": 6095, "epoch": 169, "lr": 0.001} +{"train_loss": 0.03479473665356636, "train_loss_bc": 0.03479473665356636, "train_loss_llm": 0.0, "grad_norm": 0.03902444243431091, "global_step": 6096, "epoch": 169, "lr": 0.001} +{"train_loss": 0.0442107617855072, "train_loss_bc": 0.0442107617855072, "train_loss_llm": 0.0, "grad_norm": 0.020078256726264954, "global_step": 6097, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04032553359866142, "train_loss_bc": 0.04032553359866142, "train_loss_llm": 0.0, "grad_norm": 0.02739788591861725, "global_step": 6098, "epoch": 169, "lr": 0.001} +{"train_loss": 0.040229883044958115, "train_loss_bc": 0.040229883044958115, "train_loss_llm": 0.0, "grad_norm": 0.03700225055217743, "global_step": 6099, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04488469287753105, "train_loss_bc": 0.04488469287753105, "train_loss_llm": 0.0, "grad_norm": 0.04548338055610657, "global_step": 6100, "epoch": 169, "lr": 0.001} +{"train_loss": 0.036888256669044495, "train_loss_bc": 0.036888256669044495, "train_loss_llm": 0.0, "grad_norm": 0.017872169613838196, "global_step": 6101, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04127287492156029, "train_loss_bc": 0.04126809537410736, "train_loss_llm": 0.0004780590534210205, "grad_norm": 0.02528718300163746, "global_step": 6102, "epoch": 169, "lr": 0.001} +{"train_loss": 0.038858987390995026, "train_loss_bc": 0.038858987390995026, "train_loss_llm": 0.0, "grad_norm": 0.031066840514540672, "global_step": 6103, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04106207937002182, "train_loss_bc": 0.04106207937002182, "train_loss_llm": 0.0, "grad_norm": 0.038342639803886414, "global_step": 6104, "epoch": 169, "lr": 0.001} +{"train_loss": 0.03971007838845253, "train_loss_bc": 0.03971007838845253, "train_loss_llm": 0.0, "grad_norm": 0.02068733423948288, "global_step": 6105, "epoch": 169, "lr": 0.001} +{"train_loss": 0.03845546394586563, "train_loss_bc": 0.03843848407268524, "train_loss_llm": 0.001697927713394165, "grad_norm": 0.03083501197397709, "global_step": 6106, "epoch": 169, "lr": 0.001} +{"train_loss": 0.03796606883406639, "train_loss_bc": 0.03796606883406639, "train_loss_llm": 0.0, "grad_norm": 0.03822395205497742, "global_step": 6107, "epoch": 169, "lr": 0.001} +{"train_loss": 0.0418265163898468, "train_loss_bc": 0.0418265163898468, "train_loss_llm": 0.0, "grad_norm": 0.04396357387304306, "global_step": 6108, "epoch": 169, "lr": 0.001} +{"train_loss": 0.03640299662947655, "train_loss_bc": 0.03640299662947655, "train_loss_llm": 0.0, "grad_norm": 0.014971436001360416, "global_step": 6109, "epoch": 169, "lr": 0.001} +{"train_loss": 0.03715378791093826, "train_loss_bc": 0.03715378791093826, "train_loss_llm": 0.0, "grad_norm": 0.02326277084648609, "global_step": 6110, "epoch": 169, "lr": 0.001} +{"train_loss": 0.040717460215091705, "train_loss_bc": 0.040717460215091705, "train_loss_llm": 0.0, "grad_norm": 0.02865239419043064, "global_step": 6111, "epoch": 169, "lr": 0.001} +{"train_loss": 0.0409657247364521, "train_loss_bc": 0.0409657247364521, "train_loss_llm": 0.0, "grad_norm": 0.037729859352111816, "global_step": 6112, "epoch": 169, "lr": 0.001} +{"train_loss": 0.0456281416118145, "train_loss_bc": 0.0456281416118145, "train_loss_llm": 0.0, "grad_norm": 0.015886005014181137, "global_step": 6113, "epoch": 169, "lr": 0.001} +{"train_loss": 0.03480535000562668, "train_loss_bc": 0.03480535000562668, "train_loss_llm": 0.0, "grad_norm": 0.02377508021891117, "global_step": 6114, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04280301183462143, "train_loss_bc": 0.042779892683029175, "train_loss_llm": 0.002312004566192627, "grad_norm": 0.03232283145189285, "global_step": 6115, "epoch": 169, "lr": 0.001} +{"train_loss": 0.0400574691593647, "train_loss_bc": 0.0400574691593647, "train_loss_llm": 0.0, "grad_norm": 0.03971158340573311, "global_step": 6116, "epoch": 169, "lr": 0.001} +{"train_loss": 0.04420045018196106, "train_loss_bc": 0.04420045018196106, "train_loss_llm": 0.0, "grad_norm": 0.015746859833598137, "global_step": 6117, "epoch": 169, "lr": 0.001} +{"train_loss": 0.037865299731492996, "train_loss_bc": 0.037865299731492996, "train_loss_llm": 0.0, "grad_norm": 0.027244526892900467, "global_step": 6118, "epoch": 169, "lr": 0.001} +{"train_loss": 0.041206273457242384, "train_loss_bc": 0.04592100530862808, "train_loss_llm": 0.0, "grad_norm": 0.03640814870595932, "global_step": 6119, "epoch": 169, "lr": 0.001, "val_loss": 0.0561300665140152} +{"train_loss": 0.04076254740357399, "train_loss_bc": 0.04076254740357399, "train_loss_llm": 0.0, "grad_norm": 0.039767298847436905, "global_step": 6120, "epoch": 170, "lr": 0.001} +{"train_loss": 0.0423431321978569, "train_loss_bc": 0.0423431321978569, "train_loss_llm": 0.0, "grad_norm": 0.016331909224390984, "global_step": 6121, "epoch": 170, "lr": 0.001} +{"train_loss": 0.046001892536878586, "train_loss_bc": 0.046001892536878586, "train_loss_llm": 0.0, "grad_norm": 0.026752730831503868, "global_step": 6122, "epoch": 170, "lr": 0.001} +{"train_loss": 0.0474093072116375, "train_loss_bc": 0.0474093072116375, "train_loss_llm": 0.0, "grad_norm": 0.037103213369846344, "global_step": 6123, "epoch": 170, "lr": 0.001} +{"train_loss": 0.038829486817121506, "train_loss_bc": 0.038829486817121506, "train_loss_llm": 0.0, "grad_norm": 0.04044252634048462, "global_step": 6124, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03895823284983635, "train_loss_bc": 0.03895823284983635, "train_loss_llm": 0.0, "grad_norm": 0.015405595302581787, "global_step": 6125, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03582194820046425, "train_loss_bc": 0.03582194820046425, "train_loss_llm": 0.0, "grad_norm": 0.021328581497073174, "global_step": 6126, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04861680418252945, "train_loss_bc": 0.04861680418252945, "train_loss_llm": 0.0, "grad_norm": 0.03296119347214699, "global_step": 6127, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04555991291999817, "train_loss_bc": 0.045539095997810364, "train_loss_llm": 0.0020817816257476807, "grad_norm": 0.03488999232649803, "global_step": 6128, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04354119673371315, "train_loss_bc": 0.04354119673371315, "train_loss_llm": 0.0, "grad_norm": 0.01953285187482834, "global_step": 6129, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04068154841661453, "train_loss_bc": 0.04068154841661453, "train_loss_llm": 0.0, "grad_norm": 0.030051227658987045, "global_step": 6130, "epoch": 170, "lr": 0.001} +{"train_loss": 0.0431278795003891, "train_loss_bc": 0.0431278795003891, "train_loss_llm": 0.0, "grad_norm": 0.03340190649032593, "global_step": 6131, "epoch": 170, "lr": 0.001} +{"train_loss": 0.0418451651930809, "train_loss_bc": 0.0418272502720356, "train_loss_llm": 0.0017915070056915283, "grad_norm": 0.03691967576742172, "global_step": 6132, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04568057879805565, "train_loss_bc": 0.04566933214664459, "train_loss_llm": 0.0011247992515563965, "grad_norm": 0.02196691744029522, "global_step": 6133, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03726673126220703, "train_loss_bc": 0.03725913539528847, "train_loss_llm": 0.0007597506046295166, "grad_norm": 0.0310837272554636, "global_step": 6134, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04132122918963432, "train_loss_bc": 0.04132122918963432, "train_loss_llm": 0.0, "grad_norm": 0.0401814803481102, "global_step": 6135, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04670223966240883, "train_loss_bc": 0.04670223966240883, "train_loss_llm": 0.0, "grad_norm": 0.04669179022312164, "global_step": 6136, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04086734727025032, "train_loss_bc": 0.04086734727025032, "train_loss_llm": 0.0, "grad_norm": 0.017241165041923523, "global_step": 6137, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04065893217921257, "train_loss_bc": 0.04065893217921257, "train_loss_llm": 0.0, "grad_norm": 0.027316367253661156, "global_step": 6138, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03817087411880493, "train_loss_bc": 0.038161762058734894, "train_loss_llm": 0.0009113848209381104, "grad_norm": 0.0312996506690979, "global_step": 6139, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03858408331871033, "train_loss_bc": 0.03858408331871033, "train_loss_llm": 0.0, "grad_norm": 0.040979158133268356, "global_step": 6140, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04069123789668083, "train_loss_bc": 0.040675461292266846, "train_loss_llm": 0.001577615737915039, "grad_norm": 0.017094556242227554, "global_step": 6141, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03208599612116814, "train_loss_bc": 0.032069552689790726, "train_loss_llm": 0.0016442835330963135, "grad_norm": 0.022895824164152145, "global_step": 6142, "epoch": 170, "lr": 0.001} +{"train_loss": 0.037829216569662094, "train_loss_bc": 0.037829216569662094, "train_loss_llm": 0.0, "grad_norm": 0.03579967841506004, "global_step": 6143, "epoch": 170, "lr": 0.001} +{"train_loss": 0.046851277351379395, "train_loss_bc": 0.046851277351379395, "train_loss_llm": 0.0, "grad_norm": 0.04910014569759369, "global_step": 6144, "epoch": 170, "lr": 0.001} +{"train_loss": 0.039583995938301086, "train_loss_bc": 0.039583995938301086, "train_loss_llm": 0.0, "grad_norm": 0.017093433067202568, "global_step": 6145, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03892865404486656, "train_loss_bc": 0.03892865404486656, "train_loss_llm": 0.0, "grad_norm": 0.024064386263489723, "global_step": 6146, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03864708170294762, "train_loss_bc": 0.038642264902591705, "train_loss_llm": 0.00048163533210754395, "grad_norm": 0.031740110367536545, "global_step": 6147, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04191052168607712, "train_loss_bc": 0.04188891127705574, "train_loss_llm": 0.00216105580329895, "grad_norm": 0.03943142667412758, "global_step": 6148, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03969248756766319, "train_loss_bc": 0.03969248756766319, "train_loss_llm": 0.0, "grad_norm": 0.018004365265369415, "global_step": 6149, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03650384396314621, "train_loss_bc": 0.03649507090449333, "train_loss_llm": 0.0008774101734161377, "grad_norm": 0.027246393263339996, "global_step": 6150, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03809840977191925, "train_loss_bc": 0.03809596598148346, "train_loss_llm": 0.00024431943893432617, "grad_norm": 0.02899385429918766, "global_step": 6151, "epoch": 170, "lr": 0.001} +{"train_loss": 0.0355711430311203, "train_loss_bc": 0.035554077476263046, "train_loss_llm": 0.0017065703868865967, "grad_norm": 0.038096874952316284, "global_step": 6152, "epoch": 170, "lr": 0.001} +{"train_loss": 0.03840911388397217, "train_loss_bc": 0.03839836269617081, "train_loss_llm": 0.0010751783847808838, "grad_norm": 0.01728987507522106, "global_step": 6153, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04962226748466492, "train_loss_bc": 0.04962226748466492, "train_loss_llm": 0.0, "grad_norm": 0.027847180142998695, "global_step": 6154, "epoch": 170, "lr": 0.001} +{"train_loss": 0.04104616379158364, "train_loss_bc": 0.04048549756407738, "train_loss_llm": 8.314847946166992e-06, "grad_norm": 0.03981921076774597, "global_step": 6155, "epoch": 170, "lr": 0.001, "train/cumulative_reward": 44.71616846517617, "train/mean_score": 3.5155925384620317, "train/success_rate": 0.0, "test/cumulative_reward": 16.195520770061474, "test/mean_score": 3.735434460435414, "test/success_rate": 0.08, "val_loss": 0.05533483996987343, "train_action_mse_error": 0.04503560811281204} +{"train_loss": 0.03322510048747063, "train_loss_bc": 0.03322510048747063, "train_loss_llm": 0.0, "grad_norm": 0.049907270818948746, "global_step": 6156, "epoch": 171, "lr": 0.001} +{"train_loss": 0.044363923370838165, "train_loss_bc": 0.04436304792761803, "train_loss_llm": 8.767843246459961e-05, "grad_norm": 0.020313667133450508, "global_step": 6157, "epoch": 171, "lr": 0.001} +{"train_loss": 0.0380091592669487, "train_loss_bc": 0.03800464794039726, "train_loss_llm": 0.00045099854469299316, "grad_norm": 0.03117256797850132, "global_step": 6158, "epoch": 171, "lr": 0.001} +{"train_loss": 0.03815310448408127, "train_loss_bc": 0.038143057376146317, "train_loss_llm": 0.0010045766830444336, "grad_norm": 0.03869993984699249, "global_step": 6159, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04107959568500519, "train_loss_bc": 0.04107959568500519, "train_loss_llm": 0.0, "grad_norm": 0.048848625272512436, "global_step": 6160, "epoch": 171, "lr": 0.001} +{"train_loss": 0.0411541685461998, "train_loss_bc": 0.041141580790281296, "train_loss_llm": 0.0012587010860443115, "grad_norm": 0.016980361193418503, "global_step": 6161, "epoch": 171, "lr": 0.001} +{"train_loss": 0.043098341673612595, "train_loss_bc": 0.04308250918984413, "train_loss_llm": 0.0015831589698791504, "grad_norm": 0.0305457916110754, "global_step": 6162, "epoch": 171, "lr": 0.001} +{"train_loss": 0.042184967547655106, "train_loss_bc": 0.04217342659831047, "train_loss_llm": 0.0011542141437530518, "grad_norm": 0.03780243173241615, "global_step": 6163, "epoch": 171, "lr": 0.001} +{"train_loss": 0.03658894822001457, "train_loss_bc": 0.03658894822001457, "train_loss_llm": 0.0, "grad_norm": 0.043076079338788986, "global_step": 6164, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04577431082725525, "train_loss_bc": 0.04577431082725525, "train_loss_llm": 0.0, "grad_norm": 0.017377464100718498, "global_step": 6165, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04173142835497856, "train_loss_bc": 0.04173142835497856, "train_loss_llm": 0.0, "grad_norm": 0.024268021807074547, "global_step": 6166, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04208565875887871, "train_loss_bc": 0.04208565875887871, "train_loss_llm": 0.0, "grad_norm": 0.03451833501458168, "global_step": 6167, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04509728029370308, "train_loss_bc": 0.04509728029370308, "train_loss_llm": 0.0, "grad_norm": 0.04052881896495819, "global_step": 6168, "epoch": 171, "lr": 0.001} +{"train_loss": 0.043667301535606384, "train_loss_bc": 0.043667301535606384, "train_loss_llm": 0.0, "grad_norm": 0.02048368565738201, "global_step": 6169, "epoch": 171, "lr": 0.001} +{"train_loss": 0.03941746801137924, "train_loss_bc": 0.03939761593937874, "train_loss_llm": 0.001985311508178711, "grad_norm": 0.02918245829641819, "global_step": 6170, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04687843471765518, "train_loss_bc": 0.04687843471765518, "train_loss_llm": 0.0, "grad_norm": 0.040473852306604385, "global_step": 6171, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04180983826518059, "train_loss_bc": 0.04180796444416046, "train_loss_llm": 0.00018730759620666504, "grad_norm": 0.04161917790770531, "global_step": 6172, "epoch": 171, "lr": 0.001} +{"train_loss": 0.040339287370443344, "train_loss_bc": 0.040339287370443344, "train_loss_llm": 0.0, "grad_norm": 0.017271501943469048, "global_step": 6173, "epoch": 171, "lr": 0.001} +{"train_loss": 0.03479515388607979, "train_loss_bc": 0.03479515388607979, "train_loss_llm": 0.0, "grad_norm": 0.023797741159796715, "global_step": 6174, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04122118279337883, "train_loss_bc": 0.04122118279337883, "train_loss_llm": 0.0, "grad_norm": 0.03178047016263008, "global_step": 6175, "epoch": 171, "lr": 0.001} +{"train_loss": 0.038904596120119095, "train_loss_bc": 0.038904596120119095, "train_loss_llm": 0.0, "grad_norm": 0.038530487567186356, "global_step": 6176, "epoch": 171, "lr": 0.001} +{"train_loss": 0.03635336086153984, "train_loss_bc": 0.03633221983909607, "train_loss_llm": 0.00211411714553833, "grad_norm": 0.015349116176366806, "global_step": 6177, "epoch": 171, "lr": 0.001} +{"train_loss": 0.041936129331588745, "train_loss_bc": 0.041936129331588745, "train_loss_llm": 0.0, "grad_norm": 0.024164697155356407, "global_step": 6178, "epoch": 171, "lr": 0.001} +{"train_loss": 0.0410614088177681, "train_loss_bc": 0.0410614088177681, "train_loss_llm": 0.0, "grad_norm": 0.03447285667061806, "global_step": 6179, "epoch": 171, "lr": 0.001} +{"train_loss": 0.033468589186668396, "train_loss_bc": 0.033468589186668396, "train_loss_llm": 0.0, "grad_norm": 0.039900459349155426, "global_step": 6180, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04291535168886185, "train_loss_bc": 0.04291535168886185, "train_loss_llm": 0.0, "grad_norm": 0.021304693073034286, "global_step": 6181, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04530755430459976, "train_loss_bc": 0.04530755430459976, "train_loss_llm": 0.0, "grad_norm": 0.03272763639688492, "global_step": 6182, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04230411723256111, "train_loss_bc": 0.04230411723256111, "train_loss_llm": 0.0, "grad_norm": 0.04253808781504631, "global_step": 6183, "epoch": 171, "lr": 0.001} +{"train_loss": 0.049341727048158646, "train_loss_bc": 0.049341727048158646, "train_loss_llm": 0.0, "grad_norm": 0.05457061529159546, "global_step": 6184, "epoch": 171, "lr": 0.001} +{"train_loss": 0.03406025096774101, "train_loss_bc": 0.03406025096774101, "train_loss_llm": 0.0, "grad_norm": 0.017127953469753265, "global_step": 6185, "epoch": 171, "lr": 0.001} +{"train_loss": 0.03199837729334831, "train_loss_bc": 0.03199837729334831, "train_loss_llm": 0.0, "grad_norm": 0.025852223858237267, "global_step": 6186, "epoch": 171, "lr": 0.001} +{"train_loss": 0.03765734285116196, "train_loss_bc": 0.03765515610575676, "train_loss_llm": 0.00021851062774658203, "grad_norm": 0.037178125232458115, "global_step": 6187, "epoch": 171, "lr": 0.001} +{"train_loss": 0.040372334420681, "train_loss_bc": 0.040372334420681, "train_loss_llm": 0.0, "grad_norm": 0.04498238116502762, "global_step": 6188, "epoch": 171, "lr": 0.001} +{"train_loss": 0.05159150809049606, "train_loss_bc": 0.05159150809049606, "train_loss_llm": 0.0, "grad_norm": 0.02210705168545246, "global_step": 6189, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04463736340403557, "train_loss_bc": 0.04463736340403557, "train_loss_llm": 0.0, "grad_norm": 0.02709266170859337, "global_step": 6190, "epoch": 171, "lr": 0.001} +{"train_loss": 0.04118489163617293, "train_loss_bc": 0.050018686801195145, "train_loss_llm": 0.005274653434753418, "grad_norm": 0.03421410173177719, "global_step": 6191, "epoch": 171, "lr": 0.001, "val_loss": 0.05560765787959099} +{"train_loss": 0.04455597326159477, "train_loss_bc": 0.04455597326159477, "train_loss_llm": 0.0, "grad_norm": 0.04333992674946785, "global_step": 6192, "epoch": 172, "lr": 0.001} +{"train_loss": 0.0415547713637352, "train_loss_bc": 0.0415547713637352, "train_loss_llm": 0.0, "grad_norm": 0.015097731724381447, "global_step": 6193, "epoch": 172, "lr": 0.001} +{"train_loss": 0.043488163501024246, "train_loss_bc": 0.043488163501024246, "train_loss_llm": 0.0, "grad_norm": 0.024341825395822525, "global_step": 6194, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04167886823415756, "train_loss_bc": 0.04166455566883087, "train_loss_llm": 0.0014310777187347412, "grad_norm": 0.030513333156704903, "global_step": 6195, "epoch": 172, "lr": 0.001} +{"train_loss": 0.03877817466855049, "train_loss_bc": 0.03877817466855049, "train_loss_llm": 0.0, "grad_norm": 0.03965965658426285, "global_step": 6196, "epoch": 172, "lr": 0.001} +{"train_loss": 0.03657246753573418, "train_loss_bc": 0.03657246753573418, "train_loss_llm": 0.0, "grad_norm": 0.016133546829223633, "global_step": 6197, "epoch": 172, "lr": 0.001} +{"train_loss": 0.03925342112779617, "train_loss_bc": 0.03925342112779617, "train_loss_llm": 0.0, "grad_norm": 0.02471998520195484, "global_step": 6198, "epoch": 172, "lr": 0.001} +{"train_loss": 0.040837571024894714, "train_loss_bc": 0.04083595424890518, "train_loss_llm": 0.00016158819198608398, "grad_norm": 0.029943780973553658, "global_step": 6199, "epoch": 172, "lr": 0.001} +{"train_loss": 0.040212441235780716, "train_loss_bc": 0.040212441235780716, "train_loss_llm": 0.0, "grad_norm": 0.03204703703522682, "global_step": 6200, "epoch": 172, "lr": 0.001} +{"train_loss": 0.03900173306465149, "train_loss_bc": 0.03900016099214554, "train_loss_llm": 0.00015723705291748047, "grad_norm": 0.014401663094758987, "global_step": 6201, "epoch": 172, "lr": 0.001} +{"train_loss": 0.03724253550171852, "train_loss_bc": 0.03724218159914017, "train_loss_llm": 3.522634506225586e-05, "grad_norm": 0.02676425687968731, "global_step": 6202, "epoch": 172, "lr": 0.001} +{"train_loss": 0.03855610638856888, "train_loss_bc": 0.03855610638856888, "train_loss_llm": 0.0, "grad_norm": 0.031790897250175476, "global_step": 6203, "epoch": 172, "lr": 0.001} +{"train_loss": 0.034056562930345535, "train_loss_bc": 0.03403913602232933, "train_loss_llm": 0.001742631196975708, "grad_norm": 0.040113646537065506, "global_step": 6204, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04119880497455597, "train_loss_bc": 0.04119880497455597, "train_loss_llm": 0.0, "grad_norm": 0.017779836431145668, "global_step": 6205, "epoch": 172, "lr": 0.001} +{"train_loss": 0.0461004376411438, "train_loss_bc": 0.0461004376411438, "train_loss_llm": 0.0, "grad_norm": 0.03240131214261055, "global_step": 6206, "epoch": 172, "lr": 0.001} +{"train_loss": 0.046402741223573685, "train_loss_bc": 0.046402741223573685, "train_loss_llm": 0.0, "grad_norm": 0.03517303615808487, "global_step": 6207, "epoch": 172, "lr": 0.001} +{"train_loss": 0.042523447424173355, "train_loss_bc": 0.042523447424173355, "train_loss_llm": 0.0, "grad_norm": 0.044777873903512955, "global_step": 6208, "epoch": 172, "lr": 0.001} +{"train_loss": 0.042094867676496506, "train_loss_bc": 0.042094867676496506, "train_loss_llm": 0.0, "grad_norm": 0.020038267597556114, "global_step": 6209, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04458160325884819, "train_loss_bc": 0.04458029568195343, "train_loss_llm": 0.00013074278831481934, "grad_norm": 0.026535753160715103, "global_step": 6210, "epoch": 172, "lr": 0.001} +{"train_loss": 0.042450957000255585, "train_loss_bc": 0.042450957000255585, "train_loss_llm": 0.0, "grad_norm": 0.03824484720826149, "global_step": 6211, "epoch": 172, "lr": 0.001} +{"train_loss": 0.03245842456817627, "train_loss_bc": 0.03243200480937958, "train_loss_llm": 0.002641916275024414, "grad_norm": 0.04304453730583191, "global_step": 6212, "epoch": 172, "lr": 0.001} +{"train_loss": 0.03697441518306732, "train_loss_bc": 0.03697441518306732, "train_loss_llm": 0.0, "grad_norm": 0.01944490149617195, "global_step": 6213, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04485026374459267, "train_loss_bc": 0.04484805092215538, "train_loss_llm": 0.00022110342979431152, "grad_norm": 0.030821317806839943, "global_step": 6214, "epoch": 172, "lr": 0.001} +{"train_loss": 0.03716833516955376, "train_loss_bc": 0.03716833516955376, "train_loss_llm": 0.0, "grad_norm": 0.040367595851421356, "global_step": 6215, "epoch": 172, "lr": 0.001} +{"train_loss": 0.035242438316345215, "train_loss_bc": 0.035237982869148254, "train_loss_llm": 0.0004455149173736572, "grad_norm": 0.047473762184381485, "global_step": 6216, "epoch": 172, "lr": 0.001} +{"train_loss": 0.035627152770757675, "train_loss_bc": 0.03562648221850395, "train_loss_llm": 6.693601608276367e-05, "grad_norm": 0.013989838771522045, "global_step": 6217, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04300693795084953, "train_loss_bc": 0.04300693795084953, "train_loss_llm": 0.0, "grad_norm": 0.02787637524306774, "global_step": 6218, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04510807245969772, "train_loss_bc": 0.04508748650550842, "train_loss_llm": 0.002058684825897217, "grad_norm": 0.034698598086833954, "global_step": 6219, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04286428540945053, "train_loss_bc": 0.04284202307462692, "train_loss_llm": 0.0022261738777160645, "grad_norm": 0.042468633502721786, "global_step": 6220, "epoch": 172, "lr": 0.001} +{"train_loss": 0.043314531445503235, "train_loss_bc": 0.043314531445503235, "train_loss_llm": 0.0, "grad_norm": 0.021313073113560677, "global_step": 6221, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04623996838927269, "train_loss_bc": 0.04623996838927269, "train_loss_llm": 0.0, "grad_norm": 0.03400658816099167, "global_step": 6222, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04675193130970001, "train_loss_bc": 0.04675193130970001, "train_loss_llm": 0.0, "grad_norm": 0.041473183780908585, "global_step": 6223, "epoch": 172, "lr": 0.001} +{"train_loss": 0.034689199179410934, "train_loss_bc": 0.034689199179410934, "train_loss_llm": 0.0, "grad_norm": 0.04842911660671234, "global_step": 6224, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04449981823563576, "train_loss_bc": 0.04447677358984947, "train_loss_llm": 0.002304583787918091, "grad_norm": 0.020739348605275154, "global_step": 6225, "epoch": 172, "lr": 0.001} +{"train_loss": 0.041345685720443726, "train_loss_bc": 0.041345685720443726, "train_loss_llm": 0.0, "grad_norm": 0.03442323952913284, "global_step": 6226, "epoch": 172, "lr": 0.001} +{"train_loss": 0.04115755690468682, "train_loss_bc": 0.050388939678668976, "train_loss_llm": 0.0, "grad_norm": 0.06006899103522301, "global_step": 6227, "epoch": 172, "lr": 0.001, "val_loss": 0.056365467607975006} +{"train_loss": 0.04413598030805588, "train_loss_bc": 0.04412974417209625, "train_loss_llm": 0.0006234645843505859, "grad_norm": 0.06999144703149796, "global_step": 6228, "epoch": 173, "lr": 0.001} +{"train_loss": 0.04191027954220772, "train_loss_bc": 0.04189267009496689, "train_loss_llm": 0.0017610490322113037, "grad_norm": 0.01818322204053402, "global_step": 6229, "epoch": 173, "lr": 0.001} +{"train_loss": 0.043975863605737686, "train_loss_bc": 0.04397166520357132, "train_loss_llm": 0.0004196763038635254, "grad_norm": 0.024546047672629356, "global_step": 6230, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03739019110798836, "train_loss_bc": 0.03739019110798836, "train_loss_llm": 0.0, "grad_norm": 0.039151012897491455, "global_step": 6231, "epoch": 173, "lr": 0.001} +{"train_loss": 0.041722118854522705, "train_loss_bc": 0.04171956703066826, "train_loss_llm": 0.00025525689125061035, "grad_norm": 0.050618406385183334, "global_step": 6232, "epoch": 173, "lr": 0.001} +{"train_loss": 0.04641325771808624, "train_loss_bc": 0.04641325771808624, "train_loss_llm": 0.0, "grad_norm": 0.016889560967683792, "global_step": 6233, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03643254563212395, "train_loss_bc": 0.03643254563212395, "train_loss_llm": 0.0, "grad_norm": 0.03152335807681084, "global_step": 6234, "epoch": 173, "lr": 0.001} +{"train_loss": 0.0449705645442009, "train_loss_bc": 0.0449705645442009, "train_loss_llm": 0.0, "grad_norm": 0.04179280251264572, "global_step": 6235, "epoch": 173, "lr": 0.001} +{"train_loss": 0.04158446937799454, "train_loss_bc": 0.04158446937799454, "train_loss_llm": 0.0, "grad_norm": 0.05789954960346222, "global_step": 6236, "epoch": 173, "lr": 0.001} +{"train_loss": 0.044699616730213165, "train_loss_bc": 0.04468728229403496, "train_loss_llm": 0.0012334883213043213, "grad_norm": 0.023617157712578773, "global_step": 6237, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03951570391654968, "train_loss_bc": 0.03951213136315346, "train_loss_llm": 0.00035712122917175293, "grad_norm": 0.03378415107727051, "global_step": 6238, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03458819165825844, "train_loss_bc": 0.03457729518413544, "train_loss_llm": 0.001089543104171753, "grad_norm": 0.03934358432888985, "global_step": 6239, "epoch": 173, "lr": 0.001} +{"train_loss": 0.045776352286338806, "train_loss_bc": 0.045776352286338806, "train_loss_llm": 0.0, "grad_norm": 0.046797819435596466, "global_step": 6240, "epoch": 173, "lr": 0.001} +{"train_loss": 0.04096495732665062, "train_loss_bc": 0.04092954099178314, "train_loss_llm": 0.003541707992553711, "grad_norm": 0.018118850886821747, "global_step": 6241, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03969288244843483, "train_loss_bc": 0.03969288244843483, "train_loss_llm": 0.0, "grad_norm": 0.02969040535390377, "global_step": 6242, "epoch": 173, "lr": 0.001} +{"train_loss": 0.04283248633146286, "train_loss_bc": 0.04283248633146286, "train_loss_llm": 0.0, "grad_norm": 0.03939191997051239, "global_step": 6243, "epoch": 173, "lr": 0.001} +{"train_loss": 0.04458301141858101, "train_loss_bc": 0.04458301141858101, "train_loss_llm": 0.0, "grad_norm": 0.0445195809006691, "global_step": 6244, "epoch": 173, "lr": 0.001} +{"train_loss": 0.039202168583869934, "train_loss_bc": 0.039202168583869934, "train_loss_llm": 0.0, "grad_norm": 0.02024434506893158, "global_step": 6245, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03900916501879692, "train_loss_bc": 0.03900698944926262, "train_loss_llm": 0.0002174079418182373, "grad_norm": 0.033940207213163376, "global_step": 6246, "epoch": 173, "lr": 0.001} +{"train_loss": 0.038248803466558456, "train_loss_bc": 0.0382479764521122, "train_loss_llm": 8.252263069152832e-05, "grad_norm": 0.04355768859386444, "global_step": 6247, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03337273746728897, "train_loss_bc": 0.03337273746728897, "train_loss_llm": 0.0, "grad_norm": 0.057221364229917526, "global_step": 6248, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03759034723043442, "train_loss_bc": 0.03759034723043442, "train_loss_llm": 0.0, "grad_norm": 0.024629617109894753, "global_step": 6249, "epoch": 173, "lr": 0.001} +{"train_loss": 0.042734917253255844, "train_loss_bc": 0.042693667113780975, "train_loss_llm": 0.004124850034713745, "grad_norm": 0.03413134813308716, "global_step": 6250, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03637758642435074, "train_loss_bc": 0.03637256100773811, "train_loss_llm": 0.0005024075508117676, "grad_norm": 0.0454891100525856, "global_step": 6251, "epoch": 173, "lr": 0.001} +{"train_loss": 0.045019421726465225, "train_loss_bc": 0.045019421726465225, "train_loss_llm": 0.0, "grad_norm": 0.06187540292739868, "global_step": 6252, "epoch": 173, "lr": 0.001} +{"train_loss": 0.04370041936635971, "train_loss_bc": 0.04370041936635971, "train_loss_llm": 0.0, "grad_norm": 0.018184427171945572, "global_step": 6253, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03899705782532692, "train_loss_bc": 0.03899705782532692, "train_loss_llm": 0.0, "grad_norm": 0.027707556262612343, "global_step": 6254, "epoch": 173, "lr": 0.001} +{"train_loss": 0.045709770172834396, "train_loss_bc": 0.045709770172834396, "train_loss_llm": 0.0, "grad_norm": 0.03214990347623825, "global_step": 6255, "epoch": 173, "lr": 0.001} +{"train_loss": 0.04834543541073799, "train_loss_bc": 0.04834543541073799, "train_loss_llm": 0.0, "grad_norm": 0.047323551028966904, "global_step": 6256, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03668439760804176, "train_loss_bc": 0.03668439760804176, "train_loss_llm": 0.0, "grad_norm": 0.02530549466609955, "global_step": 6257, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03226287290453911, "train_loss_bc": 0.03226105123758316, "train_loss_llm": 0.00018230080604553223, "grad_norm": 0.03662134334445, "global_step": 6258, "epoch": 173, "lr": 0.001} +{"train_loss": 0.039549440145492554, "train_loss_bc": 0.039549440145492554, "train_loss_llm": 0.0, "grad_norm": 0.04672308638691902, "global_step": 6259, "epoch": 173, "lr": 0.001} +{"train_loss": 0.051648497581481934, "train_loss_bc": 0.0516403466463089, "train_loss_llm": 0.0008152127265930176, "grad_norm": 0.05456504225730896, "global_step": 6260, "epoch": 173, "lr": 0.001} +{"train_loss": 0.04401756078004837, "train_loss_bc": 0.04401756078004837, "train_loss_llm": 0.0, "grad_norm": 0.020932486280798912, "global_step": 6261, "epoch": 173, "lr": 0.001} +{"train_loss": 0.03848082572221756, "train_loss_bc": 0.03848082572221756, "train_loss_llm": 0.0, "grad_norm": 0.032373618334531784, "global_step": 6262, "epoch": 173, "lr": 0.001} +{"train_loss": 0.041195678214232125, "train_loss_bc": 0.04090451821684837, "train_loss_llm": 0.0, "grad_norm": 0.049000512808561325, "global_step": 6263, "epoch": 173, "lr": 0.001, "val_loss": 0.05595952644944191} +{"train_loss": 0.039063069969415665, "train_loss_bc": 0.039063069969415665, "train_loss_llm": 0.0, "grad_norm": 0.06390279531478882, "global_step": 6264, "epoch": 174, "lr": 0.001} +{"train_loss": 0.036116886883974075, "train_loss_bc": 0.036116886883974075, "train_loss_llm": 0.0, "grad_norm": 0.025403348729014397, "global_step": 6265, "epoch": 174, "lr": 0.001} +{"train_loss": 0.03542587161064148, "train_loss_bc": 0.03542587161064148, "train_loss_llm": 0.0, "grad_norm": 0.03598068654537201, "global_step": 6266, "epoch": 174, "lr": 0.001} +{"train_loss": 0.03760308772325516, "train_loss_bc": 0.03759482875466347, "train_loss_llm": 0.0008258223533630371, "grad_norm": 0.04302164167165756, "global_step": 6267, "epoch": 174, "lr": 0.001} +{"train_loss": 0.04353495314717293, "train_loss_bc": 0.04353495314717293, "train_loss_llm": 0.0, "grad_norm": 0.05839965119957924, "global_step": 6268, "epoch": 174, "lr": 0.001} +{"train_loss": 0.045204490423202515, "train_loss_bc": 0.045204490423202515, "train_loss_llm": 0.0, "grad_norm": 0.016895940527319908, "global_step": 6269, "epoch": 174, "lr": 0.001} +{"train_loss": 0.042278189212083817, "train_loss_bc": 0.042278189212083817, "train_loss_llm": 0.0, "grad_norm": 0.03229421377182007, "global_step": 6270, "epoch": 174, "lr": 0.001} +{"train_loss": 0.04451068490743637, "train_loss_bc": 0.04451068490743637, "train_loss_llm": 0.0, "grad_norm": 0.04319250211119652, "global_step": 6271, "epoch": 174, "lr": 0.001} +{"train_loss": 0.038803014904260635, "train_loss_bc": 0.03879363089799881, "train_loss_llm": 0.0009385347366333008, "grad_norm": 0.05701923742890358, "global_step": 6272, "epoch": 174, "lr": 0.001} +{"train_loss": 0.03472333028912544, "train_loss_bc": 0.03472333028912544, "train_loss_llm": 0.0, "grad_norm": 0.014905774965882301, "global_step": 6273, "epoch": 174, "lr": 0.001} +{"train_loss": 0.03928038850426674, "train_loss_bc": 0.03928038850426674, "train_loss_llm": 0.0, "grad_norm": 0.024412043392658234, "global_step": 6274, "epoch": 174, "lr": 0.001} +{"train_loss": 0.047449253499507904, "train_loss_bc": 0.047449253499507904, "train_loss_llm": 0.0, "grad_norm": 0.03636210039258003, "global_step": 6275, "epoch": 174, "lr": 0.001} +{"train_loss": 0.04020644351840019, "train_loss_bc": 0.04019877687096596, "train_loss_llm": 0.0007666349411010742, "grad_norm": 0.044433336704969406, "global_step": 6276, "epoch": 174, "lr": 0.001} +{"train_loss": 0.0432964563369751, "train_loss_bc": 0.0432964563369751, "train_loss_llm": 0.0, "grad_norm": 0.016133498400449753, "global_step": 6277, "epoch": 174, "lr": 0.001} +{"train_loss": 0.0422026664018631, "train_loss_bc": 0.04220201447606087, "train_loss_llm": 6.532669067382812e-05, "grad_norm": 0.022895706817507744, "global_step": 6278, "epoch": 174, "lr": 0.001} +{"train_loss": 0.03838596120476723, "train_loss_bc": 0.03838596120476723, "train_loss_llm": 0.0, "grad_norm": 0.034009356051683426, "global_step": 6279, "epoch": 174, "lr": 0.001} +{"train_loss": 0.040707655251026154, "train_loss_bc": 0.040707655251026154, "train_loss_llm": 0.0, "grad_norm": 0.041525594890117645, "global_step": 6280, "epoch": 174, "lr": 0.001} +{"train_loss": 0.0421011745929718, "train_loss_bc": 0.0421011745929718, "train_loss_llm": 0.0, "grad_norm": 0.021234942600131035, "global_step": 6281, "epoch": 174, "lr": 0.001} +{"train_loss": 0.04439403489232063, "train_loss_bc": 0.04439403489232063, "train_loss_llm": 0.0, "grad_norm": 0.04542059078812599, "global_step": 6282, "epoch": 174, "lr": 0.001} +{"train_loss": 0.04708391800522804, "train_loss_bc": 0.04703424498438835, "train_loss_llm": 0.0049672722816467285, "grad_norm": 0.05801933631300926, "global_step": 6283, "epoch": 174, "lr": 0.001} +{"train_loss": 0.042295124381780624, "train_loss_bc": 0.042295124381780624, "train_loss_llm": 0.0, "grad_norm": 0.06345950812101364, "global_step": 6284, "epoch": 174, "lr": 0.001} +{"train_loss": 0.03722730278968811, "train_loss_bc": 0.03722730278968811, "train_loss_llm": 0.0, "grad_norm": 0.017846370115876198, "global_step": 6285, "epoch": 174, "lr": 0.001} +{"train_loss": 0.03986538574099541, "train_loss_bc": 0.039861131459474564, "train_loss_llm": 0.00042557716369628906, "grad_norm": 0.028869235888123512, "global_step": 6286, "epoch": 174, "lr": 0.001} +{"train_loss": 0.0413186177611351, "train_loss_bc": 0.0413186177611351, "train_loss_llm": 0.0, "grad_norm": 0.04136000573635101, "global_step": 6287, "epoch": 174, "lr": 0.001} +{"train_loss": 0.04558868706226349, "train_loss_bc": 0.04558868706226349, "train_loss_llm": 0.0, "grad_norm": 0.05710936710238457, "global_step": 6288, "epoch": 174, "lr": 0.001} +{"train_loss": 0.04129255935549736, "train_loss_bc": 0.04129255935549736, "train_loss_llm": 0.0, "grad_norm": 0.01633899100124836, "global_step": 6289, "epoch": 174, "lr": 0.001} +{"train_loss": 0.041290003806352615, "train_loss_bc": 0.041256822645664215, "train_loss_llm": 0.003318220376968384, "grad_norm": 0.02628948539495468, "global_step": 6290, "epoch": 174, "lr": 0.001} +{"train_loss": 0.0376829095184803, "train_loss_bc": 0.0376829095184803, "train_loss_llm": 0.0, "grad_norm": 0.03601992130279541, "global_step": 6291, "epoch": 174, "lr": 0.001} +{"train_loss": 0.03502397984266281, "train_loss_bc": 0.03502397984266281, "train_loss_llm": 0.0, "grad_norm": 0.042256880551576614, "global_step": 6292, "epoch": 174, "lr": 0.001} +{"train_loss": 0.04320095479488373, "train_loss_bc": 0.04320095479488373, "train_loss_llm": 0.0, "grad_norm": 0.015743780881166458, "global_step": 6293, "epoch": 174, "lr": 0.001} +{"train_loss": 0.03388768434524536, "train_loss_bc": 0.03387833759188652, "train_loss_llm": 0.0009346604347229004, "grad_norm": 0.024136556312441826, "global_step": 6294, "epoch": 174, "lr": 0.001} +{"train_loss": 0.047015246003866196, "train_loss_bc": 0.04699595272541046, "train_loss_llm": 0.0019292235374450684, "grad_norm": 0.038460589945316315, "global_step": 6295, "epoch": 174, "lr": 0.001} +{"train_loss": 0.04438893124461174, "train_loss_bc": 0.04438893124461174, "train_loss_llm": 0.0, "grad_norm": 0.04721234366297722, "global_step": 6296, "epoch": 174, "lr": 0.001} +{"train_loss": 0.03903532400727272, "train_loss_bc": 0.03903532400727272, "train_loss_llm": 0.0, "grad_norm": 0.019947437569499016, "global_step": 6297, "epoch": 174, "lr": 0.001} +{"train_loss": 0.044219061732292175, "train_loss_bc": 0.044219061732292175, "train_loss_llm": 0.0, "grad_norm": 0.03082745335996151, "global_step": 6298, "epoch": 174, "lr": 0.001} +{"train_loss": 0.04114322209109863, "train_loss_bc": 0.04545269161462784, "train_loss_llm": 0.0, "grad_norm": 0.042982008308172226, "global_step": 6299, "epoch": 174, "lr": 0.001, "val_loss": 0.05599256977438927} +{"train_loss": 0.04136823117733002, "train_loss_bc": 0.04136823117733002, "train_loss_llm": 0.0, "grad_norm": 0.05115765333175659, "global_step": 6300, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04879423975944519, "train_loss_bc": 0.04879423975944519, "train_loss_llm": 0.0, "grad_norm": 0.018920153379440308, "global_step": 6301, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04370582848787308, "train_loss_bc": 0.043705713003873825, "train_loss_llm": 1.1593103408813477e-05, "grad_norm": 0.025851214304566383, "global_step": 6302, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04177671670913696, "train_loss_bc": 0.04177671670913696, "train_loss_llm": 0.0, "grad_norm": 0.03558709844946861, "global_step": 6303, "epoch": 175, "lr": 0.001} +{"train_loss": 0.03690853714942932, "train_loss_bc": 0.03690163046121597, "train_loss_llm": 0.0006905496120452881, "grad_norm": 0.04270034283399582, "global_step": 6304, "epoch": 175, "lr": 0.001} +{"train_loss": 0.03344126045703888, "train_loss_bc": 0.03344126045703888, "train_loss_llm": 0.0, "grad_norm": 0.013100804761052132, "global_step": 6305, "epoch": 175, "lr": 0.001} +{"train_loss": 0.03977569192647934, "train_loss_bc": 0.03977569192647934, "train_loss_llm": 0.0, "grad_norm": 0.02414601854979992, "global_step": 6306, "epoch": 175, "lr": 0.001} +{"train_loss": 0.044235002249479294, "train_loss_bc": 0.044235002249479294, "train_loss_llm": 0.0, "grad_norm": 0.03310248628258705, "global_step": 6307, "epoch": 175, "lr": 0.001} +{"train_loss": 0.03972641006112099, "train_loss_bc": 0.03972641006112099, "train_loss_llm": 0.0, "grad_norm": 0.04066985473036766, "global_step": 6308, "epoch": 175, "lr": 0.001} +{"train_loss": 0.045774247497320175, "train_loss_bc": 0.045746464282274246, "train_loss_llm": 0.0027782320976257324, "grad_norm": 0.019462615251541138, "global_step": 6309, "epoch": 175, "lr": 0.001} +{"train_loss": 0.03630400076508522, "train_loss_bc": 0.03628666326403618, "train_loss_llm": 0.0017337501049041748, "grad_norm": 0.029474327340722084, "global_step": 6310, "epoch": 175, "lr": 0.001} +{"train_loss": 0.03753680735826492, "train_loss_bc": 0.03753680735826492, "train_loss_llm": 0.0, "grad_norm": 0.04163328558206558, "global_step": 6311, "epoch": 175, "lr": 0.001} +{"train_loss": 0.0348571240901947, "train_loss_bc": 0.03485553711652756, "train_loss_llm": 0.00015875697135925293, "grad_norm": 0.048024982213974, "global_step": 6312, "epoch": 175, "lr": 0.001} +{"train_loss": 0.045266080647706985, "train_loss_bc": 0.045266080647706985, "train_loss_llm": 0.0, "grad_norm": 0.018942300230264664, "global_step": 6313, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04060247913002968, "train_loss_bc": 0.04060247913002968, "train_loss_llm": 0.0, "grad_norm": 0.02858462743461132, "global_step": 6314, "epoch": 175, "lr": 0.001} +{"train_loss": 0.038313716650009155, "train_loss_bc": 0.038313716650009155, "train_loss_llm": 0.0, "grad_norm": 0.03296126052737236, "global_step": 6315, "epoch": 175, "lr": 0.001} +{"train_loss": 0.036725182086229324, "train_loss_bc": 0.03671158105134964, "train_loss_llm": 0.0013599395751953125, "grad_norm": 0.03928158059716225, "global_step": 6316, "epoch": 175, "lr": 0.001} +{"train_loss": 0.037949539721012115, "train_loss_bc": 0.037949539721012115, "train_loss_llm": 0.0, "grad_norm": 0.019511450082063675, "global_step": 6317, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04317200183868408, "train_loss_bc": 0.04314379021525383, "train_loss_llm": 0.0028211772441864014, "grad_norm": 0.02363249845802784, "global_step": 6318, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04271138831973076, "train_loss_bc": 0.04269083961844444, "train_loss_llm": 0.0020547211170196533, "grad_norm": 0.030430998653173447, "global_step": 6319, "epoch": 175, "lr": 0.001} +{"train_loss": 0.044392768293619156, "train_loss_bc": 0.044368844479322433, "train_loss_llm": 0.0023924410343170166, "grad_norm": 0.04010271281003952, "global_step": 6320, "epoch": 175, "lr": 0.001} +{"train_loss": 0.037361107766628265, "train_loss_bc": 0.037361107766628265, "train_loss_llm": 0.0, "grad_norm": 0.017515240237116814, "global_step": 6321, "epoch": 175, "lr": 0.001} +{"train_loss": 0.035293251276016235, "train_loss_bc": 0.035293251276016235, "train_loss_llm": 0.0, "grad_norm": 0.02901466190814972, "global_step": 6322, "epoch": 175, "lr": 0.001} +{"train_loss": 0.05057675018906593, "train_loss_bc": 0.05057675018906593, "train_loss_llm": 0.0, "grad_norm": 0.035591501742601395, "global_step": 6323, "epoch": 175, "lr": 0.001} +{"train_loss": 0.042332716286182404, "train_loss_bc": 0.042332716286182404, "train_loss_llm": 0.0, "grad_norm": 0.04966217279434204, "global_step": 6324, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04128507152199745, "train_loss_bc": 0.04126829653978348, "train_loss_llm": 0.0016776025295257568, "grad_norm": 0.02022496983408928, "global_step": 6325, "epoch": 175, "lr": 0.001} +{"train_loss": 0.03623393550515175, "train_loss_bc": 0.03623393550515175, "train_loss_llm": 0.0, "grad_norm": 0.025203747674822807, "global_step": 6326, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04094937816262245, "train_loss_bc": 0.040931586176157, "train_loss_llm": 0.0017791688442230225, "grad_norm": 0.027357010170817375, "global_step": 6327, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04161893576383591, "train_loss_bc": 0.04161893576383591, "train_loss_llm": 0.0, "grad_norm": 0.0357075110077858, "global_step": 6328, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04329635947942734, "train_loss_bc": 0.04329635947942734, "train_loss_llm": 0.0, "grad_norm": 0.021744145080447197, "global_step": 6329, "epoch": 175, "lr": 0.001} +{"train_loss": 0.044620294123888016, "train_loss_bc": 0.044620294123888016, "train_loss_llm": 0.0, "grad_norm": 0.031499676406383514, "global_step": 6330, "epoch": 175, "lr": 0.001} +{"train_loss": 0.034077875316143036, "train_loss_bc": 0.034077875316143036, "train_loss_llm": 0.0, "grad_norm": 0.036649513989686966, "global_step": 6331, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04332417994737625, "train_loss_bc": 0.04332312196493149, "train_loss_llm": 0.00010594725608825684, "grad_norm": 0.043984733521938324, "global_step": 6332, "epoch": 175, "lr": 0.001} +{"train_loss": 0.04086827114224434, "train_loss_bc": 0.04086827114224434, "train_loss_llm": 0.0, "grad_norm": 0.013931600376963615, "global_step": 6333, "epoch": 175, "lr": 0.001} +{"train_loss": 0.03801202028989792, "train_loss_bc": 0.03801202028989792, "train_loss_llm": 0.0, "grad_norm": 0.023150626569986343, "global_step": 6334, "epoch": 175, "lr": 0.001} +{"train_loss": 0.041008655809693866, "train_loss_bc": 0.05312420800328255, "train_loss_llm": 0.0, "grad_norm": 0.03618369624018669, "global_step": 6335, "epoch": 175, "lr": 0.001, "train/cumulative_reward": 92.50907590460164, "train/mean_score": 22.831663596509195, "train/success_rate": 0.7, "test/cumulative_reward": 72.85052252400351, "test/mean_score": 9.797489473475036, "test/success_rate": 0.22, "val_loss": 0.05590689182281494, "train_action_mse_error": 0.04443741217255592} +{"train_loss": 0.04307756572961807, "train_loss_bc": 0.04307756572961807, "train_loss_llm": 0.0, "grad_norm": 0.042764171957969666, "global_step": 6336, "epoch": 176, "lr": 0.001} +{"train_loss": 0.042480140924453735, "train_loss_bc": 0.04247862845659256, "train_loss_llm": 0.0001512765884399414, "grad_norm": 0.01370450109243393, "global_step": 6337, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03388560563325882, "train_loss_bc": 0.03388560563325882, "train_loss_llm": 0.0, "grad_norm": 0.020334666594862938, "global_step": 6338, "epoch": 176, "lr": 0.001} +{"train_loss": 0.042782045900821686, "train_loss_bc": 0.042782045900821686, "train_loss_llm": 0.0, "grad_norm": 0.024145925417542458, "global_step": 6339, "epoch": 176, "lr": 0.001} +{"train_loss": 0.044185392558574677, "train_loss_bc": 0.044185392558574677, "train_loss_llm": 0.0, "grad_norm": 0.0276348739862442, "global_step": 6340, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04758264124393463, "train_loss_bc": 0.04758264124393463, "train_loss_llm": 0.0, "grad_norm": 0.018691936507821083, "global_step": 6341, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04603424295783043, "train_loss_bc": 0.04603424295783043, "train_loss_llm": 0.0, "grad_norm": 0.02556959167122841, "global_step": 6342, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03930157050490379, "train_loss_bc": 0.03930157050490379, "train_loss_llm": 0.0, "grad_norm": 0.032993514090776443, "global_step": 6343, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04790673404932022, "train_loss_bc": 0.04790673404932022, "train_loss_llm": 0.0, "grad_norm": 0.04434286057949066, "global_step": 6344, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03591616824269295, "train_loss_bc": 0.03591432422399521, "train_loss_llm": 0.0001842975616455078, "grad_norm": 0.015409011393785477, "global_step": 6345, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04680408537387848, "train_loss_bc": 0.04678104445338249, "train_loss_llm": 0.0023040473461151123, "grad_norm": 0.03286517783999443, "global_step": 6346, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03704793006181717, "train_loss_bc": 0.03704793006181717, "train_loss_llm": 0.0, "grad_norm": 0.043402932584285736, "global_step": 6347, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03371541574597359, "train_loss_bc": 0.03370913863182068, "train_loss_llm": 0.0006276071071624756, "grad_norm": 0.04936860501766205, "global_step": 6348, "epoch": 176, "lr": 0.001} +{"train_loss": 0.037319060415029526, "train_loss_bc": 0.03730025514960289, "train_loss_llm": 0.001880556344985962, "grad_norm": 0.017451105639338493, "global_step": 6349, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03604070097208023, "train_loss_bc": 0.03603897988796234, "train_loss_llm": 0.0001722574234008789, "grad_norm": 0.02447422966361046, "global_step": 6350, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04344599321484566, "train_loss_bc": 0.04344599321484566, "train_loss_llm": 0.0, "grad_norm": 0.031048011034727097, "global_step": 6351, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04367420822381973, "train_loss_bc": 0.04367420822381973, "train_loss_llm": 0.0, "grad_norm": 0.04025184363126755, "global_step": 6352, "epoch": 176, "lr": 0.001} +{"train_loss": 0.041369739919900894, "train_loss_bc": 0.041369739919900894, "train_loss_llm": 0.0, "grad_norm": 0.016501164063811302, "global_step": 6353, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04488113522529602, "train_loss_bc": 0.04488113522529602, "train_loss_llm": 0.0, "grad_norm": 0.025452587753534317, "global_step": 6354, "epoch": 176, "lr": 0.001} +{"train_loss": 0.041744131594896317, "train_loss_bc": 0.041744131594896317, "train_loss_llm": 0.0, "grad_norm": 0.036333270370960236, "global_step": 6355, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04077896103262901, "train_loss_bc": 0.04077896103262901, "train_loss_llm": 0.0, "grad_norm": 0.041262298822402954, "global_step": 6356, "epoch": 176, "lr": 0.001} +{"train_loss": 0.039682790637016296, "train_loss_bc": 0.039682790637016296, "train_loss_llm": 0.0, "grad_norm": 0.019528675824403763, "global_step": 6357, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03816532343626022, "train_loss_bc": 0.03815365955233574, "train_loss_llm": 0.0011662840843200684, "grad_norm": 0.03659527748823166, "global_step": 6358, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04488000273704529, "train_loss_bc": 0.04486200213432312, "train_loss_llm": 0.00180014967918396, "grad_norm": 0.04202040284872055, "global_step": 6359, "epoch": 176, "lr": 0.001} +{"train_loss": 0.040824804455041885, "train_loss_bc": 0.040824804455041885, "train_loss_llm": 0.0, "grad_norm": 0.04523087665438652, "global_step": 6360, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04343707114458084, "train_loss_bc": 0.04343707114458084, "train_loss_llm": 0.0, "grad_norm": 0.01716725155711174, "global_step": 6361, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03524356707930565, "train_loss_bc": 0.03524356707930565, "train_loss_llm": 0.0, "grad_norm": 0.02426394820213318, "global_step": 6362, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03891327977180481, "train_loss_bc": 0.038900915533304214, "train_loss_llm": 0.0012364685535430908, "grad_norm": 0.035591546446084976, "global_step": 6363, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04647205024957657, "train_loss_bc": 0.04647205024957657, "train_loss_llm": 0.0, "grad_norm": 0.0445585697889328, "global_step": 6364, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03152887523174286, "train_loss_bc": 0.03152887523174286, "train_loss_llm": 0.0, "grad_norm": 0.014127020724117756, "global_step": 6365, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03960355743765831, "train_loss_bc": 0.03960355743765831, "train_loss_llm": 0.0, "grad_norm": 0.025714633986353874, "global_step": 6366, "epoch": 176, "lr": 0.001} +{"train_loss": 0.03633498772978783, "train_loss_bc": 0.03630639240145683, "train_loss_llm": 0.00285947322845459, "grad_norm": 0.03755704686045647, "global_step": 6367, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04222548007965088, "train_loss_bc": 0.04221102595329285, "train_loss_llm": 0.0014454424381256104, "grad_norm": 0.04758651554584503, "global_step": 6368, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04375088959932327, "train_loss_bc": 0.04375088959932327, "train_loss_llm": 0.0, "grad_norm": 0.01712394319474697, "global_step": 6369, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04149073362350464, "train_loss_bc": 0.041480470448732376, "train_loss_llm": 0.0010261833667755127, "grad_norm": 0.029047707095742226, "global_step": 6370, "epoch": 176, "lr": 0.001} +{"train_loss": 0.04087921904606952, "train_loss_bc": 0.039118122309446335, "train_loss_llm": 0.0006879866123199463, "grad_norm": 0.04232918471097946, "global_step": 6371, "epoch": 176, "lr": 0.001, "val_loss": 0.05591424182057381} +{"train_loss": 0.04249883070588112, "train_loss_bc": 0.04249883070588112, "train_loss_llm": 0.0, "grad_norm": 0.04908333346247673, "global_step": 6372, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04155143350362778, "train_loss_bc": 0.04154946655035019, "train_loss_llm": 0.00019678473472595215, "grad_norm": 0.017602380365133286, "global_step": 6373, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04451591148972511, "train_loss_bc": 0.04451591148972511, "train_loss_llm": 0.0, "grad_norm": 0.022037072107195854, "global_step": 6374, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04211635887622833, "train_loss_bc": 0.04211316257715225, "train_loss_llm": 0.000319749116897583, "grad_norm": 0.027643723413348198, "global_step": 6375, "epoch": 177, "lr": 0.001} +{"train_loss": 0.03653177246451378, "train_loss_bc": 0.03653177246451378, "train_loss_llm": 0.0, "grad_norm": 0.03125034645199776, "global_step": 6376, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04732167348265648, "train_loss_bc": 0.04732167348265648, "train_loss_llm": 0.0, "grad_norm": 0.016802066937088966, "global_step": 6377, "epoch": 177, "lr": 0.001} +{"train_loss": 0.0430452786386013, "train_loss_bc": 0.04303140193223953, "train_loss_llm": 0.0013875067234039307, "grad_norm": 0.029365580528974533, "global_step": 6378, "epoch": 177, "lr": 0.001} +{"train_loss": 0.041856925934553146, "train_loss_bc": 0.041856925934553146, "train_loss_llm": 0.0, "grad_norm": 0.03452511504292488, "global_step": 6379, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04685213789343834, "train_loss_bc": 0.04684626683592796, "train_loss_llm": 0.0005872249603271484, "grad_norm": 0.046680957078933716, "global_step": 6380, "epoch": 177, "lr": 0.001} +{"train_loss": 0.033397261053323746, "train_loss_bc": 0.033393748104572296, "train_loss_llm": 0.00035122036933898926, "grad_norm": 0.014954440295696259, "global_step": 6381, "epoch": 177, "lr": 0.001} +{"train_loss": 0.039874691516160965, "train_loss_bc": 0.039874691516160965, "train_loss_llm": 0.0, "grad_norm": 0.023282162845134735, "global_step": 6382, "epoch": 177, "lr": 0.001} +{"train_loss": 0.040144238620996475, "train_loss_bc": 0.040144238620996475, "train_loss_llm": 0.0, "grad_norm": 0.033091988414525986, "global_step": 6383, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04078029841184616, "train_loss_bc": 0.04078029841184616, "train_loss_llm": 0.0, "grad_norm": 0.0388985238969326, "global_step": 6384, "epoch": 177, "lr": 0.001} +{"train_loss": 0.03475000336766243, "train_loss_bc": 0.03475000336766243, "train_loss_llm": 0.0, "grad_norm": 0.019886096939444542, "global_step": 6385, "epoch": 177, "lr": 0.001} +{"train_loss": 0.03621288388967514, "train_loss_bc": 0.03620556741952896, "train_loss_llm": 0.000731736421585083, "grad_norm": 0.023511918261647224, "global_step": 6386, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04501573368906975, "train_loss_bc": 0.04501573368906975, "train_loss_llm": 0.0, "grad_norm": 0.03899062052369118, "global_step": 6387, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04246806353330612, "train_loss_bc": 0.04246806353330612, "train_loss_llm": 0.0, "grad_norm": 0.04989621043205261, "global_step": 6388, "epoch": 177, "lr": 0.001} +{"train_loss": 0.034695759415626526, "train_loss_bc": 0.034695759415626526, "train_loss_llm": 0.0, "grad_norm": 0.01719432696700096, "global_step": 6389, "epoch": 177, "lr": 0.001} +{"train_loss": 0.042602088302373886, "train_loss_bc": 0.042602088302373886, "train_loss_llm": 0.0, "grad_norm": 0.034291259944438934, "global_step": 6390, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04072955623269081, "train_loss_bc": 0.04072955623269081, "train_loss_llm": 0.0, "grad_norm": 0.04057806730270386, "global_step": 6391, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04555100202560425, "train_loss_bc": 0.04555100202560425, "train_loss_llm": 0.0, "grad_norm": 0.05059373751282692, "global_step": 6392, "epoch": 177, "lr": 0.001} +{"train_loss": 0.03774159774184227, "train_loss_bc": 0.03774159774184227, "train_loss_llm": 0.0, "grad_norm": 0.020883910357952118, "global_step": 6393, "epoch": 177, "lr": 0.001} +{"train_loss": 0.047338757663965225, "train_loss_bc": 0.047338757663965225, "train_loss_llm": 0.0, "grad_norm": 0.03287948668003082, "global_step": 6394, "epoch": 177, "lr": 0.001} +{"train_loss": 0.03433012589812279, "train_loss_bc": 0.03433012589812279, "train_loss_llm": 0.0, "grad_norm": 0.040353938937187195, "global_step": 6395, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04323134571313858, "train_loss_bc": 0.04323134571313858, "train_loss_llm": 0.0, "grad_norm": 0.048298731446266174, "global_step": 6396, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04022437706589699, "train_loss_bc": 0.040223218500614166, "train_loss_llm": 0.00011596083641052246, "grad_norm": 0.015963153913617134, "global_step": 6397, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04355034604668617, "train_loss_bc": 0.04354856535792351, "train_loss_llm": 0.0001780688762664795, "grad_norm": 0.02748561091721058, "global_step": 6398, "epoch": 177, "lr": 0.001} +{"train_loss": 0.041793953627347946, "train_loss_bc": 0.04177642613649368, "train_loss_llm": 0.0017527639865875244, "grad_norm": 0.03073973022401333, "global_step": 6399, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04211173579096794, "train_loss_bc": 0.04210561513900757, "train_loss_llm": 0.0006121397018432617, "grad_norm": 0.03670228645205498, "global_step": 6400, "epoch": 177, "lr": 0.001} +{"train_loss": 0.03971840441226959, "train_loss_bc": 0.039673544466495514, "train_loss_llm": 0.004486143589019775, "grad_norm": 0.0176740400493145, "global_step": 6401, "epoch": 177, "lr": 0.001} +{"train_loss": 0.036646634340286255, "train_loss_bc": 0.036646634340286255, "train_loss_llm": 0.0, "grad_norm": 0.030848205089569092, "global_step": 6402, "epoch": 177, "lr": 0.001} +{"train_loss": 0.03753169625997543, "train_loss_bc": 0.037511564791202545, "train_loss_llm": 0.002013266086578369, "grad_norm": 0.046881720423698425, "global_step": 6403, "epoch": 177, "lr": 0.001} +{"train_loss": 0.038777854293584824, "train_loss_bc": 0.038777854293584824, "train_loss_llm": 0.0, "grad_norm": 0.05428367480635643, "global_step": 6404, "epoch": 177, "lr": 0.001} +{"train_loss": 0.04473462700843811, "train_loss_bc": 0.04473172128200531, "train_loss_llm": 0.0002904832363128662, "grad_norm": 0.017857812345027924, "global_step": 6405, "epoch": 177, "lr": 0.001} +{"train_loss": 0.042188260704278946, "train_loss_bc": 0.042188260704278946, "train_loss_llm": 0.0, "grad_norm": 0.028861364349722862, "global_step": 6406, "epoch": 177, "lr": 0.001} +{"train_loss": 0.040885039812160864, "train_loss_bc": 0.039429813623428345, "train_loss_llm": 0.0, "grad_norm": 0.03796200826764107, "global_step": 6407, "epoch": 177, "lr": 0.001, "val_loss": 0.05608204007148743} +{"train_loss": 0.04031350463628769, "train_loss_bc": 0.04029842093586922, "train_loss_llm": 0.0015082359313964844, "grad_norm": 0.05364825576543808, "global_step": 6408, "epoch": 178, "lr": 0.001} +{"train_loss": 0.037812087684869766, "train_loss_bc": 0.03780205547809601, "train_loss_llm": 0.0010030567646026611, "grad_norm": 0.022333161905407906, "global_step": 6409, "epoch": 178, "lr": 0.001} +{"train_loss": 0.0347842238843441, "train_loss_bc": 0.0347842238843441, "train_loss_llm": 0.0, "grad_norm": 0.02670487016439438, "global_step": 6410, "epoch": 178, "lr": 0.001} +{"train_loss": 0.04307068884372711, "train_loss_bc": 0.04306139051914215, "train_loss_llm": 0.0009300112724304199, "grad_norm": 0.029649054631590843, "global_step": 6411, "epoch": 178, "lr": 0.001} +{"train_loss": 0.040390126407146454, "train_loss_bc": 0.040390126407146454, "train_loss_llm": 0.0, "grad_norm": 0.03161296993494034, "global_step": 6412, "epoch": 178, "lr": 0.001} +{"train_loss": 0.04244505614042282, "train_loss_bc": 0.04244505614042282, "train_loss_llm": 0.0, "grad_norm": 0.019327132031321526, "global_step": 6413, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03658225014805794, "train_loss_bc": 0.03658225014805794, "train_loss_llm": 0.0, "grad_norm": 0.026921693235635757, "global_step": 6414, "epoch": 178, "lr": 0.001} +{"train_loss": 0.04451264068484306, "train_loss_bc": 0.04451201111078262, "train_loss_llm": 6.29127025604248e-05, "grad_norm": 0.02543572336435318, "global_step": 6415, "epoch": 178, "lr": 0.001} +{"train_loss": 0.044746994972229004, "train_loss_bc": 0.044746994972229004, "train_loss_llm": 0.0, "grad_norm": 0.03862662985920906, "global_step": 6416, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03753324970602989, "train_loss_bc": 0.03753324970602989, "train_loss_llm": 0.0, "grad_norm": 0.014813791029155254, "global_step": 6417, "epoch": 178, "lr": 0.001} +{"train_loss": 0.05560959130525589, "train_loss_bc": 0.05560185760259628, "train_loss_llm": 0.000773310661315918, "grad_norm": 0.023055117577314377, "global_step": 6418, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03825919330120087, "train_loss_bc": 0.038231946527957916, "train_loss_llm": 0.0027247369289398193, "grad_norm": 0.03428034484386444, "global_step": 6419, "epoch": 178, "lr": 0.001} +{"train_loss": 0.04074433445930481, "train_loss_bc": 0.04073290154337883, "train_loss_llm": 0.0011433660984039307, "grad_norm": 0.04120897874236107, "global_step": 6420, "epoch": 178, "lr": 0.001} +{"train_loss": 0.043880511075258255, "train_loss_bc": 0.04387155920267105, "train_loss_llm": 0.0008952617645263672, "grad_norm": 0.024025745689868927, "global_step": 6421, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03988831117749214, "train_loss_bc": 0.03988831117749214, "train_loss_llm": 0.0, "grad_norm": 0.0340287908911705, "global_step": 6422, "epoch": 178, "lr": 0.001} +{"train_loss": 0.040271930396556854, "train_loss_bc": 0.040271930396556854, "train_loss_llm": 0.0, "grad_norm": 0.03956589475274086, "global_step": 6423, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03993658348917961, "train_loss_bc": 0.03993658348917961, "train_loss_llm": 0.0, "grad_norm": 0.045775823295116425, "global_step": 6424, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03981224074959755, "train_loss_bc": 0.03981224074959755, "train_loss_llm": 0.0, "grad_norm": 0.018050121143460274, "global_step": 6425, "epoch": 178, "lr": 0.001} +{"train_loss": 0.038822900503873825, "train_loss_bc": 0.03881952911615372, "train_loss_llm": 0.00033718347549438477, "grad_norm": 0.02834421768784523, "global_step": 6426, "epoch": 178, "lr": 0.001} +{"train_loss": 0.04365804046392441, "train_loss_bc": 0.04365804046392441, "train_loss_llm": 0.0, "grad_norm": 0.03270222246646881, "global_step": 6427, "epoch": 178, "lr": 0.001} +{"train_loss": 0.041054751724004745, "train_loss_bc": 0.041054751724004745, "train_loss_llm": 0.0, "grad_norm": 0.037568774074316025, "global_step": 6428, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03573053702712059, "train_loss_bc": 0.03573053702712059, "train_loss_llm": 0.0, "grad_norm": 0.01663080044090748, "global_step": 6429, "epoch": 178, "lr": 0.001} +{"train_loss": 0.04050068184733391, "train_loss_bc": 0.04050068184733391, "train_loss_llm": 0.0, "grad_norm": 0.022787349298596382, "global_step": 6430, "epoch": 178, "lr": 0.001} +{"train_loss": 0.0374746173620224, "train_loss_bc": 0.0374746173620224, "train_loss_llm": 0.0, "grad_norm": 0.03280255198478699, "global_step": 6431, "epoch": 178, "lr": 0.001} +{"train_loss": 0.04526618868112564, "train_loss_bc": 0.04526618868112564, "train_loss_llm": 0.0, "grad_norm": 0.044388093054294586, "global_step": 6432, "epoch": 178, "lr": 0.001} +{"train_loss": 0.041413530707359314, "train_loss_bc": 0.041413530707359314, "train_loss_llm": 0.0, "grad_norm": 0.01901310123503208, "global_step": 6433, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03899222984910011, "train_loss_bc": 0.03897833079099655, "train_loss_llm": 0.0013900399208068848, "grad_norm": 0.02722070924937725, "global_step": 6434, "epoch": 178, "lr": 0.001} +{"train_loss": 0.035592932254076004, "train_loss_bc": 0.035581059753894806, "train_loss_llm": 0.0011871755123138428, "grad_norm": 0.0364610031247139, "global_step": 6435, "epoch": 178, "lr": 0.001} +{"train_loss": 0.04278517886996269, "train_loss_bc": 0.04278517886996269, "train_loss_llm": 0.0, "grad_norm": 0.04822710528969765, "global_step": 6436, "epoch": 178, "lr": 0.001} +{"train_loss": 0.039168551564216614, "train_loss_bc": 0.039168551564216614, "train_loss_llm": 0.0, "grad_norm": 0.02110421285033226, "global_step": 6437, "epoch": 178, "lr": 0.001} +{"train_loss": 0.047033701092004776, "train_loss_bc": 0.047033701092004776, "train_loss_llm": 0.0, "grad_norm": 0.03442681208252907, "global_step": 6438, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03592542186379433, "train_loss_bc": 0.03592542186379433, "train_loss_llm": 0.0, "grad_norm": 0.03930453583598137, "global_step": 6439, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03884533792734146, "train_loss_bc": 0.03884533792734146, "train_loss_llm": 0.0, "grad_norm": 0.051289599388837814, "global_step": 6440, "epoch": 178, "lr": 0.001} +{"train_loss": 0.03952351585030556, "train_loss_bc": 0.03952351585030556, "train_loss_llm": 0.0, "grad_norm": 0.016792116686701775, "global_step": 6441, "epoch": 178, "lr": 0.001} +{"train_loss": 0.04084715247154236, "train_loss_bc": 0.04084715247154236, "train_loss_llm": 0.0, "grad_norm": 0.0330958291888237, "global_step": 6442, "epoch": 178, "lr": 0.001} +{"train_loss": 0.04086756271620592, "train_loss_bc": 0.04799192398786545, "train_loss_llm": 0.001154482364654541, "grad_norm": 0.05383070930838585, "global_step": 6443, "epoch": 178, "lr": 0.001, "val_loss": 0.05590449646115303} +{"train_loss": 0.042075108736753464, "train_loss_bc": 0.042074766010046005, "train_loss_llm": 3.409385681152344e-05, "grad_norm": 0.054634515196084976, "global_step": 6444, "epoch": 179, "lr": 0.001} +{"train_loss": 0.03921234980225563, "train_loss_bc": 0.03921234980225563, "train_loss_llm": 0.0, "grad_norm": 0.017909280955791473, "global_step": 6445, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04268151894211769, "train_loss_bc": 0.042678963392972946, "train_loss_llm": 0.0002556145191192627, "grad_norm": 0.025306083261966705, "global_step": 6446, "epoch": 179, "lr": 0.001} +{"train_loss": 0.044289395213127136, "train_loss_bc": 0.04427929222583771, "train_loss_llm": 0.0010103285312652588, "grad_norm": 0.0330079160630703, "global_step": 6447, "epoch": 179, "lr": 0.001} +{"train_loss": 0.045477185398340225, "train_loss_bc": 0.045477185398340225, "train_loss_llm": 0.0, "grad_norm": 0.03735274448990822, "global_step": 6448, "epoch": 179, "lr": 0.001} +{"train_loss": 0.042942114174366, "train_loss_bc": 0.042942114174366, "train_loss_llm": 0.0, "grad_norm": 0.020139098167419434, "global_step": 6449, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04141319915652275, "train_loss_bc": 0.04141319915652275, "train_loss_llm": 0.0, "grad_norm": 0.026708072051405907, "global_step": 6450, "epoch": 179, "lr": 0.001} +{"train_loss": 0.032840751111507416, "train_loss_bc": 0.032840751111507416, "train_loss_llm": 0.0, "grad_norm": 0.028860870748758316, "global_step": 6451, "epoch": 179, "lr": 0.001} +{"train_loss": 0.044083740562200546, "train_loss_bc": 0.044083740562200546, "train_loss_llm": 0.0, "grad_norm": 0.03859084099531174, "global_step": 6452, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04730325937271118, "train_loss_bc": 0.04730325937271118, "train_loss_llm": 0.0, "grad_norm": 0.020692525431513786, "global_step": 6453, "epoch": 179, "lr": 0.001} +{"train_loss": 0.03403923660516739, "train_loss_bc": 0.03403923660516739, "train_loss_llm": 0.0, "grad_norm": 0.036199890077114105, "global_step": 6454, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04040484502911568, "train_loss_bc": 0.04040484502911568, "train_loss_llm": 0.0, "grad_norm": 0.05540372058749199, "global_step": 6455, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04550281539559364, "train_loss_bc": 0.04550281539559364, "train_loss_llm": 0.0, "grad_norm": 0.0675988718867302, "global_step": 6456, "epoch": 179, "lr": 0.001} +{"train_loss": 0.03302699327468872, "train_loss_bc": 0.03302699327468872, "train_loss_llm": 0.0, "grad_norm": 0.017093785107135773, "global_step": 6457, "epoch": 179, "lr": 0.001} +{"train_loss": 0.045866403728723526, "train_loss_bc": 0.045866403728723526, "train_loss_llm": 0.0, "grad_norm": 0.028397658839821815, "global_step": 6458, "epoch": 179, "lr": 0.001} +{"train_loss": 0.042812105268239975, "train_loss_bc": 0.04280632734298706, "train_loss_llm": 0.0005779564380645752, "grad_norm": 0.03911596164107323, "global_step": 6459, "epoch": 179, "lr": 0.001} +{"train_loss": 0.03519102558493614, "train_loss_bc": 0.035189662128686905, "train_loss_llm": 0.00013622641563415527, "grad_norm": 0.051561009138822556, "global_step": 6460, "epoch": 179, "lr": 0.001} +{"train_loss": 0.0436084121465683, "train_loss_bc": 0.0436084121465683, "train_loss_llm": 0.0, "grad_norm": 0.01767384074628353, "global_step": 6461, "epoch": 179, "lr": 0.001} +{"train_loss": 0.044312406331300735, "train_loss_bc": 0.04430224373936653, "train_loss_llm": 0.001016080379486084, "grad_norm": 0.023272419348359108, "global_step": 6462, "epoch": 179, "lr": 0.001} +{"train_loss": 0.03901752829551697, "train_loss_bc": 0.03901752829551697, "train_loss_llm": 0.0, "grad_norm": 0.026112025603652, "global_step": 6463, "epoch": 179, "lr": 0.001} +{"train_loss": 0.041241660714149475, "train_loss_bc": 0.041241660714149475, "train_loss_llm": 0.0, "grad_norm": 0.033983130007982254, "global_step": 6464, "epoch": 179, "lr": 0.001} +{"train_loss": 0.02863933891057968, "train_loss_bc": 0.028630217537283897, "train_loss_llm": 0.0009122192859649658, "grad_norm": 0.014501192606985569, "global_step": 6465, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04204380884766579, "train_loss_bc": 0.04204380884766579, "train_loss_llm": 0.0, "grad_norm": 0.026019275188446045, "global_step": 6466, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04057197645306587, "train_loss_bc": 0.04057197645306587, "train_loss_llm": 0.0, "grad_norm": 0.0397416353225708, "global_step": 6467, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04153495654463768, "train_loss_bc": 0.04152810573577881, "train_loss_llm": 0.0006850063800811768, "grad_norm": 0.04947548732161522, "global_step": 6468, "epoch": 179, "lr": 0.001} +{"train_loss": 0.0418529286980629, "train_loss_bc": 0.0418529286980629, "train_loss_llm": 0.0, "grad_norm": 0.01544627733528614, "global_step": 6469, "epoch": 179, "lr": 0.001} +{"train_loss": 0.03935284540057182, "train_loss_bc": 0.0393475703895092, "train_loss_llm": 0.0005274116992950439, "grad_norm": 0.02410614676773548, "global_step": 6470, "epoch": 179, "lr": 0.001} +{"train_loss": 0.05056333169341087, "train_loss_bc": 0.05056333169341087, "train_loss_llm": 0.0, "grad_norm": 0.03419480472803116, "global_step": 6471, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04550902917981148, "train_loss_bc": 0.04550902917981148, "train_loss_llm": 0.0, "grad_norm": 0.04760803282260895, "global_step": 6472, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04010657221078873, "train_loss_bc": 0.040068794041872025, "train_loss_llm": 0.0037779510021209717, "grad_norm": 0.017857996746897697, "global_step": 6473, "epoch": 179, "lr": 0.001} +{"train_loss": 0.0335419625043869, "train_loss_bc": 0.0335419625043869, "train_loss_llm": 0.0, "grad_norm": 0.027843061834573746, "global_step": 6474, "epoch": 179, "lr": 0.001} +{"train_loss": 0.03984997794032097, "train_loss_bc": 0.03980540484189987, "train_loss_llm": 0.004457443952560425, "grad_norm": 0.045117199420928955, "global_step": 6475, "epoch": 179, "lr": 0.001} +{"train_loss": 0.03954891487956047, "train_loss_bc": 0.03954891487956047, "train_loss_llm": 0.0, "grad_norm": 0.05502188578248024, "global_step": 6476, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04390862584114075, "train_loss_bc": 0.043907538056373596, "train_loss_llm": 0.00010889768600463867, "grad_norm": 0.02222565934062004, "global_step": 6477, "epoch": 179, "lr": 0.001} +{"train_loss": 0.0433807298541069, "train_loss_bc": 0.0433807298541069, "train_loss_llm": 0.0, "grad_norm": 0.034172698855400085, "global_step": 6478, "epoch": 179, "lr": 0.001} +{"train_loss": 0.04083333692202965, "train_loss_bc": 0.03223419561982155, "train_loss_llm": 0.0018880665302276611, "grad_norm": 0.04261983558535576, "global_step": 6479, "epoch": 179, "lr": 0.001, "val_loss": 0.05630793049931526} +{"train_loss": 0.043072231113910675, "train_loss_bc": 0.04304618388414383, "train_loss_llm": 0.002604663372039795, "grad_norm": 0.048226598650217056, "global_step": 6480, "epoch": 180, "lr": 0.001} +{"train_loss": 0.03811459615826607, "train_loss_bc": 0.03808920457959175, "train_loss_llm": 0.0025391578674316406, "grad_norm": 0.015343514271080494, "global_step": 6481, "epoch": 180, "lr": 0.001} +{"train_loss": 0.040438659489154816, "train_loss_bc": 0.04042695462703705, "train_loss_llm": 0.0011706054210662842, "grad_norm": 0.03777293115854263, "global_step": 6482, "epoch": 180, "lr": 0.001} +{"train_loss": 0.041226040571928024, "train_loss_bc": 0.041226040571928024, "train_loss_llm": 0.0, "grad_norm": 0.04152665287256241, "global_step": 6483, "epoch": 180, "lr": 0.001} +{"train_loss": 0.03229285404086113, "train_loss_bc": 0.03229285404086113, "train_loss_llm": 0.0, "grad_norm": 0.04713129997253418, "global_step": 6484, "epoch": 180, "lr": 0.001} +{"train_loss": 0.03865624591708183, "train_loss_bc": 0.03865624591708183, "train_loss_llm": 0.0, "grad_norm": 0.016079379245638847, "global_step": 6485, "epoch": 180, "lr": 0.001} +{"train_loss": 0.0343809574842453, "train_loss_bc": 0.034379325807094574, "train_loss_llm": 0.00016301870346069336, "grad_norm": 0.021225403994321823, "global_step": 6486, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04351968318223953, "train_loss_bc": 0.043507084250450134, "train_loss_llm": 0.0012597143650054932, "grad_norm": 0.027920138090848923, "global_step": 6487, "epoch": 180, "lr": 0.001} +{"train_loss": 0.0371752567589283, "train_loss_bc": 0.037171777337789536, "train_loss_llm": 0.00034800171852111816, "grad_norm": 0.03440837934613228, "global_step": 6488, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04217100143432617, "train_loss_bc": 0.04215240851044655, "train_loss_llm": 0.0018594563007354736, "grad_norm": 0.02165420912206173, "global_step": 6489, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04796665534377098, "train_loss_bc": 0.047949451953172684, "train_loss_llm": 0.0017202794551849365, "grad_norm": 0.032722216099500656, "global_step": 6490, "epoch": 180, "lr": 0.001} +{"train_loss": 0.039321769028902054, "train_loss_bc": 0.039321769028902054, "train_loss_llm": 0.0, "grad_norm": 0.04441079497337341, "global_step": 6491, "epoch": 180, "lr": 0.001} +{"train_loss": 0.038088563829660416, "train_loss_bc": 0.0380883626639843, "train_loss_llm": 2.0295381546020508e-05, "grad_norm": 0.05623449385166168, "global_step": 6492, "epoch": 180, "lr": 0.001} +{"train_loss": 0.03804412856698036, "train_loss_bc": 0.03804412856698036, "train_loss_llm": 0.0, "grad_norm": 0.02225680463016033, "global_step": 6493, "epoch": 180, "lr": 0.001} +{"train_loss": 0.044910747557878494, "train_loss_bc": 0.044910747557878494, "train_loss_llm": 0.0, "grad_norm": 0.02417328767478466, "global_step": 6494, "epoch": 180, "lr": 0.001} +{"train_loss": 0.03764871135354042, "train_loss_bc": 0.03764871135354042, "train_loss_llm": 0.0, "grad_norm": 0.03668878227472305, "global_step": 6495, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04349079355597496, "train_loss_bc": 0.043487630784511566, "train_loss_llm": 0.00031641125679016113, "grad_norm": 0.04544752091169357, "global_step": 6496, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04229322075843811, "train_loss_bc": 0.04228362441062927, "train_loss_llm": 0.0009596943855285645, "grad_norm": 0.017454229295253754, "global_step": 6497, "epoch": 180, "lr": 0.001} +{"train_loss": 0.039618294686079025, "train_loss_bc": 0.039618294686079025, "train_loss_llm": 0.0, "grad_norm": 0.024473024532198906, "global_step": 6498, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04173259809613228, "train_loss_bc": 0.04173259809613228, "train_loss_llm": 0.0, "grad_norm": 0.03269263356924057, "global_step": 6499, "epoch": 180, "lr": 0.001} +{"train_loss": 0.044643547385931015, "train_loss_bc": 0.044643547385931015, "train_loss_llm": 0.0, "grad_norm": 0.043543532490730286, "global_step": 6500, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04365997388958931, "train_loss_bc": 0.04365997388958931, "train_loss_llm": 0.0, "grad_norm": 0.01690789684653282, "global_step": 6501, "epoch": 180, "lr": 0.001} +{"train_loss": 0.03340493515133858, "train_loss_bc": 0.03340493515133858, "train_loss_llm": 0.0, "grad_norm": 0.030939409509301186, "global_step": 6502, "epoch": 180, "lr": 0.001} +{"train_loss": 0.043381016701459885, "train_loss_bc": 0.043381016701459885, "train_loss_llm": 0.0, "grad_norm": 0.0524430014193058, "global_step": 6503, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04103562608361244, "train_loss_bc": 0.041026849299669266, "train_loss_llm": 0.0008777976036071777, "grad_norm": 0.06405273079872131, "global_step": 6504, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04354916512966156, "train_loss_bc": 0.04354424774646759, "train_loss_llm": 0.0004919171333312988, "grad_norm": 0.015132563188672066, "global_step": 6505, "epoch": 180, "lr": 0.001} +{"train_loss": 0.03825219348073006, "train_loss_bc": 0.03825219348073006, "train_loss_llm": 0.0, "grad_norm": 0.021522678434848785, "global_step": 6506, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04038175940513611, "train_loss_bc": 0.04038175940513611, "train_loss_llm": 0.0, "grad_norm": 0.026802988722920418, "global_step": 6507, "epoch": 180, "lr": 0.001} +{"train_loss": 0.046683695167303085, "train_loss_bc": 0.046683695167303085, "train_loss_llm": 0.0, "grad_norm": 0.030265288427472115, "global_step": 6508, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04092446342110634, "train_loss_bc": 0.04092446342110634, "train_loss_llm": 0.0, "grad_norm": 0.0185259822756052, "global_step": 6509, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04454592242836952, "train_loss_bc": 0.04454592242836952, "train_loss_llm": 0.0, "grad_norm": 0.02461892180144787, "global_step": 6510, "epoch": 180, "lr": 0.001} +{"train_loss": 0.04307324066758156, "train_loss_bc": 0.04307324066758156, "train_loss_llm": 0.0, "grad_norm": 0.030330469831824303, "global_step": 6511, "epoch": 180, "lr": 0.001} +{"train_loss": 0.049540188163518906, "train_loss_bc": 0.049540188163518906, "train_loss_llm": 0.0, "grad_norm": 0.03896487504243851, "global_step": 6512, "epoch": 180, "lr": 0.001} +{"train_loss": 0.03741283714771271, "train_loss_bc": 0.03741283714771271, "train_loss_llm": 0.0, "grad_norm": 0.01734604313969612, "global_step": 6513, "epoch": 180, "lr": 0.001} +{"train_loss": 0.03182333707809448, "train_loss_bc": 0.03182333707809448, "train_loss_llm": 0.0, "grad_norm": 0.026264198124408722, "global_step": 6514, "epoch": 180, "lr": 0.001} +{"train_loss": 0.040829051906863846, "train_loss_bc": 0.04337095841765404, "train_loss_llm": 0.0, "grad_norm": 0.0317334309220314, "global_step": 6515, "epoch": 180, "lr": 0.001, "train/cumulative_reward": 88.37278171411218, "train/mean_score": 8.721687998101014, "train/success_rate": 0.1, "test/cumulative_reward": 24.335943132714203, "test/mean_score": 1.9723964311193471, "test/success_rate": 0.04, "val_loss": 0.05581141635775566, "train_action_mse_error": 0.04439149796962738} +{"train_loss": 0.03489359840750694, "train_loss_bc": 0.03489359840750694, "train_loss_llm": 0.0, "grad_norm": 0.038902077823877335, "global_step": 6516, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03933090344071388, "train_loss_bc": 0.03933090344071388, "train_loss_llm": 0.0, "grad_norm": 0.017550261691212654, "global_step": 6517, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03572192043066025, "train_loss_bc": 0.03572123125195503, "train_loss_llm": 6.875395774841309e-05, "grad_norm": 0.02790638990700245, "global_step": 6518, "epoch": 181, "lr": 0.001} +{"train_loss": 0.038344837725162506, "train_loss_bc": 0.03833171725273132, "train_loss_llm": 0.0013122260570526123, "grad_norm": 0.033413078635931015, "global_step": 6519, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04604487493634224, "train_loss_bc": 0.04604487493634224, "train_loss_llm": 0.0, "grad_norm": 0.04227505996823311, "global_step": 6520, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04097457230091095, "train_loss_bc": 0.04097457230091095, "train_loss_llm": 0.0, "grad_norm": 0.02535337395966053, "global_step": 6521, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03759582340717316, "train_loss_bc": 0.03755270689725876, "train_loss_llm": 0.004311710596084595, "grad_norm": 0.039385318756103516, "global_step": 6522, "epoch": 181, "lr": 0.001} +{"train_loss": 0.048291221261024475, "train_loss_bc": 0.048291221261024475, "train_loss_llm": 0.0, "grad_norm": 0.04886878654360771, "global_step": 6523, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03302359953522682, "train_loss_bc": 0.03302359953522682, "train_loss_llm": 0.0, "grad_norm": 0.06367630511522293, "global_step": 6524, "epoch": 181, "lr": 0.001} +{"train_loss": 0.0468008778989315, "train_loss_bc": 0.0468008778989315, "train_loss_llm": 0.0, "grad_norm": 0.017093824222683907, "global_step": 6525, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04660739377140999, "train_loss_bc": 0.04660739377140999, "train_loss_llm": 0.0, "grad_norm": 0.026464561000466347, "global_step": 6526, "epoch": 181, "lr": 0.001} +{"train_loss": 0.039800167083740234, "train_loss_bc": 0.03979211300611496, "train_loss_llm": 0.0008055269718170166, "grad_norm": 0.03506515920162201, "global_step": 6527, "epoch": 181, "lr": 0.001} +{"train_loss": 0.040027834475040436, "train_loss_bc": 0.040027834475040436, "train_loss_llm": 0.0, "grad_norm": 0.043606873601675034, "global_step": 6528, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04266209155321121, "train_loss_bc": 0.04266209155321121, "train_loss_llm": 0.0, "grad_norm": 0.01908661238849163, "global_step": 6529, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03484325483441353, "train_loss_bc": 0.03484325483441353, "train_loss_llm": 0.0, "grad_norm": 0.025397317484021187, "global_step": 6530, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03690407797694206, "train_loss_bc": 0.03690407797694206, "train_loss_llm": 0.0, "grad_norm": 0.03447302430868149, "global_step": 6531, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03694695234298706, "train_loss_bc": 0.036927707493305206, "train_loss_llm": 0.0019245743751525879, "grad_norm": 0.042440321296453476, "global_step": 6532, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03991953656077385, "train_loss_bc": 0.03991953656077385, "train_loss_llm": 0.0, "grad_norm": 0.019313551485538483, "global_step": 6533, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04112863540649414, "train_loss_bc": 0.041120272129774094, "train_loss_llm": 0.0008362233638763428, "grad_norm": 0.02940610609948635, "global_step": 6534, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04710432142019272, "train_loss_bc": 0.047073349356651306, "train_loss_llm": 0.0030972063541412354, "grad_norm": 0.042357467114925385, "global_step": 6535, "epoch": 181, "lr": 0.001} +{"train_loss": 0.0337374284863472, "train_loss_bc": 0.0337374284863472, "train_loss_llm": 0.0, "grad_norm": 0.05071302130818367, "global_step": 6536, "epoch": 181, "lr": 0.001} +{"train_loss": 0.040012460201978683, "train_loss_bc": 0.040012460201978683, "train_loss_llm": 0.0, "grad_norm": 0.021977173164486885, "global_step": 6537, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04766368493437767, "train_loss_bc": 0.04765775427222252, "train_loss_llm": 0.0005929768085479736, "grad_norm": 0.030752189457416534, "global_step": 6538, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04414929822087288, "train_loss_bc": 0.04414929822087288, "train_loss_llm": 0.0, "grad_norm": 0.03902614489197731, "global_step": 6539, "epoch": 181, "lr": 0.001} +{"train_loss": 0.044132620096206665, "train_loss_bc": 0.044132620096206665, "train_loss_llm": 0.0, "grad_norm": 0.042269185185432434, "global_step": 6540, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03654511272907257, "train_loss_bc": 0.03654511272907257, "train_loss_llm": 0.0, "grad_norm": 0.013451112434267998, "global_step": 6541, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03783966228365898, "train_loss_bc": 0.03783966228365898, "train_loss_llm": 0.0, "grad_norm": 0.022757722064852715, "global_step": 6542, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04171326383948326, "train_loss_bc": 0.04171326383948326, "train_loss_llm": 0.0, "grad_norm": 0.03541344404220581, "global_step": 6543, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04917632043361664, "train_loss_bc": 0.04917632043361664, "train_loss_llm": 0.0, "grad_norm": 0.049838997423648834, "global_step": 6544, "epoch": 181, "lr": 0.001} +{"train_loss": 0.03849659487605095, "train_loss_bc": 0.03849324584007263, "train_loss_llm": 0.00033485889434814453, "grad_norm": 0.017955927178263664, "global_step": 6545, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04605388641357422, "train_loss_bc": 0.04604322463274002, "train_loss_llm": 0.0010663270950317383, "grad_norm": 0.02683919109404087, "global_step": 6546, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04394451528787613, "train_loss_bc": 0.04394451528787613, "train_loss_llm": 0.0, "grad_norm": 0.037349406629800797, "global_step": 6547, "epoch": 181, "lr": 0.001} +{"train_loss": 0.039158474653959274, "train_loss_bc": 0.039158474653959274, "train_loss_llm": 0.0, "grad_norm": 0.0439232774078846, "global_step": 6548, "epoch": 181, "lr": 0.001} +{"train_loss": 0.037939175963401794, "train_loss_bc": 0.037939175963401794, "train_loss_llm": 0.0, "grad_norm": 0.019496330991387367, "global_step": 6549, "epoch": 181, "lr": 0.001} +{"train_loss": 0.04078996554017067, "train_loss_bc": 0.0407884381711483, "train_loss_llm": 0.00015279650688171387, "grad_norm": 0.03035042993724346, "global_step": 6550, "epoch": 181, "lr": 0.001} +{"train_loss": 0.040885466461380325, "train_loss_bc": 0.04353126510977745, "train_loss_llm": 0.002656877040863037, "grad_norm": 0.042262282222509384, "global_step": 6551, "epoch": 181, "lr": 0.001, "val_loss": 0.056275684386491776} +{"train_loss": 0.039158500730991364, "train_loss_bc": 0.039158500730991364, "train_loss_llm": 0.0, "grad_norm": 0.04835343733429909, "global_step": 6552, "epoch": 182, "lr": 0.001} +{"train_loss": 0.0367237888276577, "train_loss_bc": 0.036723438650369644, "train_loss_llm": 3.4868717193603516e-05, "grad_norm": 0.014519639313220978, "global_step": 6553, "epoch": 182, "lr": 0.001} +{"train_loss": 0.043513841927051544, "train_loss_bc": 0.043513841927051544, "train_loss_llm": 0.0, "grad_norm": 0.029496558010578156, "global_step": 6554, "epoch": 182, "lr": 0.001} +{"train_loss": 0.039035189896821976, "train_loss_bc": 0.039035189896821976, "train_loss_llm": 0.0, "grad_norm": 0.037819746881723404, "global_step": 6555, "epoch": 182, "lr": 0.001} +{"train_loss": 0.037497490644454956, "train_loss_bc": 0.037497490644454956, "train_loss_llm": 0.0, "grad_norm": 0.04138989374041557, "global_step": 6556, "epoch": 182, "lr": 0.001} +{"train_loss": 0.040187131613492966, "train_loss_bc": 0.040187131613492966, "train_loss_llm": 0.0, "grad_norm": 0.01532797608524561, "global_step": 6557, "epoch": 182, "lr": 0.001} +{"train_loss": 0.039859820157289505, "train_loss_bc": 0.039859820157289505, "train_loss_llm": 0.0, "grad_norm": 0.02500392310321331, "global_step": 6558, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04881040006875992, "train_loss_bc": 0.04881040006875992, "train_loss_llm": 0.0, "grad_norm": 0.03511624038219452, "global_step": 6559, "epoch": 182, "lr": 0.001} +{"train_loss": 0.03939014673233032, "train_loss_bc": 0.03937789797782898, "train_loss_llm": 0.0012250542640686035, "grad_norm": 0.04223515838384628, "global_step": 6560, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04317115247249603, "train_loss_bc": 0.04317115247249603, "train_loss_llm": 0.0, "grad_norm": 0.01681261695921421, "global_step": 6561, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04074600711464882, "train_loss_bc": 0.040743496268987656, "train_loss_llm": 0.0002512335777282715, "grad_norm": 0.022858066484332085, "global_step": 6562, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04148911312222481, "train_loss_bc": 0.04148911312222481, "train_loss_llm": 0.0, "grad_norm": 0.027812428772449493, "global_step": 6563, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04834188148379326, "train_loss_bc": 0.04834188148379326, "train_loss_llm": 0.0, "grad_norm": 0.03666352853178978, "global_step": 6564, "epoch": 182, "lr": 0.001} +{"train_loss": 0.036295585334300995, "train_loss_bc": 0.03628815338015556, "train_loss_llm": 0.0007430613040924072, "grad_norm": 0.013233328238129616, "global_step": 6565, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04197569191455841, "train_loss_bc": 0.04197050631046295, "train_loss_llm": 0.00051841139793396, "grad_norm": 0.02818029187619686, "global_step": 6566, "epoch": 182, "lr": 0.001} +{"train_loss": 0.040004972368478775, "train_loss_bc": 0.040004972368478775, "train_loss_llm": 0.0, "grad_norm": 0.034647271037101746, "global_step": 6567, "epoch": 182, "lr": 0.001} +{"train_loss": 0.032342176884412766, "train_loss_bc": 0.032342176884412766, "train_loss_llm": 0.0, "grad_norm": 0.040863797068595886, "global_step": 6568, "epoch": 182, "lr": 0.001} +{"train_loss": 0.0411735437810421, "train_loss_bc": 0.04117104038596153, "train_loss_llm": 0.0002504289150238037, "grad_norm": 0.018828261643648148, "global_step": 6569, "epoch": 182, "lr": 0.001} +{"train_loss": 0.03933160379528999, "train_loss_bc": 0.03933160379528999, "train_loss_llm": 0.0, "grad_norm": 0.028872432187199593, "global_step": 6570, "epoch": 182, "lr": 0.001} +{"train_loss": 0.03265592083334923, "train_loss_bc": 0.03265592083334923, "train_loss_llm": 0.0, "grad_norm": 0.033259328454732895, "global_step": 6571, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04224012792110443, "train_loss_bc": 0.04224012792110443, "train_loss_llm": 0.0, "grad_norm": 0.042442210018634796, "global_step": 6572, "epoch": 182, "lr": 0.001} +{"train_loss": 0.045078933238983154, "train_loss_bc": 0.045078933238983154, "train_loss_llm": 0.0, "grad_norm": 0.021742068231105804, "global_step": 6573, "epoch": 182, "lr": 0.001} +{"train_loss": 0.0474201999604702, "train_loss_bc": 0.0474201999604702, "train_loss_llm": 0.0, "grad_norm": 0.02799917757511139, "global_step": 6574, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04395051300525665, "train_loss_bc": 0.043929219245910645, "train_loss_llm": 0.002129495143890381, "grad_norm": 0.033656854182481766, "global_step": 6575, "epoch": 182, "lr": 0.001} +{"train_loss": 0.03987523913383484, "train_loss_bc": 0.03987523913383484, "train_loss_llm": 0.0, "grad_norm": 0.0404583178460598, "global_step": 6576, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04362931475043297, "train_loss_bc": 0.04362931475043297, "train_loss_llm": 0.0, "grad_norm": 0.016396580263972282, "global_step": 6577, "epoch": 182, "lr": 0.001} +{"train_loss": 0.040329013019800186, "train_loss_bc": 0.040329013019800186, "train_loss_llm": 0.0, "grad_norm": 0.02366996929049492, "global_step": 6578, "epoch": 182, "lr": 0.001} +{"train_loss": 0.03681470453739166, "train_loss_bc": 0.03681470453739166, "train_loss_llm": 0.0, "grad_norm": 0.02854456752538681, "global_step": 6579, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04850470647215843, "train_loss_bc": 0.04850440099835396, "train_loss_llm": 3.072619438171387e-05, "grad_norm": 0.04301915317773819, "global_step": 6580, "epoch": 182, "lr": 0.001} +{"train_loss": 0.038159701973199844, "train_loss_bc": 0.0381312221288681, "train_loss_llm": 0.002848029136657715, "grad_norm": 0.021523529663681984, "global_step": 6581, "epoch": 182, "lr": 0.001} +{"train_loss": 0.03650479018688202, "train_loss_bc": 0.036497559398412704, "train_loss_llm": 0.0007232427597045898, "grad_norm": 0.024180887266993523, "global_step": 6582, "epoch": 182, "lr": 0.001} +{"train_loss": 0.038630735129117966, "train_loss_bc": 0.038630735129117966, "train_loss_llm": 0.0, "grad_norm": 0.032137781381607056, "global_step": 6583, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04014267772436142, "train_loss_bc": 0.04014267772436142, "train_loss_llm": 0.0, "grad_norm": 0.03978481516242027, "global_step": 6584, "epoch": 182, "lr": 0.001} +{"train_loss": 0.039950452744960785, "train_loss_bc": 0.03994162008166313, "train_loss_llm": 0.0008831024169921875, "grad_norm": 0.01771235279738903, "global_step": 6585, "epoch": 182, "lr": 0.001} +{"train_loss": 0.04245762154459953, "train_loss_bc": 0.04245762154459953, "train_loss_llm": 0.0, "grad_norm": 0.027458691969513893, "global_step": 6586, "epoch": 182, "lr": 0.001} +{"train_loss": 0.0406522068919407, "train_loss_bc": 0.03806515783071518, "train_loss_llm": 0.0021600723266601562, "grad_norm": 0.03888969495892525, "global_step": 6587, "epoch": 182, "lr": 0.001, "val_loss": 0.05588382109999657} +{"train_loss": 0.046344585716724396, "train_loss_bc": 0.046344585716724396, "train_loss_llm": 0.0, "grad_norm": 0.045477915555238724, "global_step": 6588, "epoch": 183, "lr": 0.001} +{"train_loss": 0.03378206491470337, "train_loss_bc": 0.03378206491470337, "train_loss_llm": 0.0, "grad_norm": 0.0173909030854702, "global_step": 6589, "epoch": 183, "lr": 0.001} +{"train_loss": 0.044647566974163055, "train_loss_bc": 0.044647566974163055, "train_loss_llm": 0.0, "grad_norm": 0.028469501063227654, "global_step": 6590, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04494518041610718, "train_loss_bc": 0.04494518041610718, "train_loss_llm": 0.0, "grad_norm": 0.03618498519062996, "global_step": 6591, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04118252918124199, "train_loss_bc": 0.041176892817020416, "train_loss_llm": 0.0005637407302856445, "grad_norm": 0.04452222213149071, "global_step": 6592, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04111909866333008, "train_loss_bc": 0.04111909866333008, "train_loss_llm": 0.0, "grad_norm": 0.016738805919885635, "global_step": 6593, "epoch": 183, "lr": 0.001} +{"train_loss": 0.046848732978105545, "train_loss_bc": 0.04683356359601021, "train_loss_llm": 0.0015170276165008545, "grad_norm": 0.03006664849817753, "global_step": 6594, "epoch": 183, "lr": 0.001} +{"train_loss": 0.038150765001773834, "train_loss_bc": 0.03814588487148285, "train_loss_llm": 0.00048801302909851074, "grad_norm": 0.04181448742747307, "global_step": 6595, "epoch": 183, "lr": 0.001} +{"train_loss": 0.049473073333501816, "train_loss_bc": 0.049473073333501816, "train_loss_llm": 0.0, "grad_norm": 0.04559706151485443, "global_step": 6596, "epoch": 183, "lr": 0.001} +{"train_loss": 0.03602144494652748, "train_loss_bc": 0.03602144494652748, "train_loss_llm": 0.0, "grad_norm": 0.017933517694473267, "global_step": 6597, "epoch": 183, "lr": 0.001} +{"train_loss": 0.038101762533187866, "train_loss_bc": 0.038101762533187866, "train_loss_llm": 0.0, "grad_norm": 0.030435068532824516, "global_step": 6598, "epoch": 183, "lr": 0.001} +{"train_loss": 0.035745762288570404, "train_loss_bc": 0.03572409972548485, "train_loss_llm": 0.0021662116050720215, "grad_norm": 0.03719716519117355, "global_step": 6599, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04204071685671806, "train_loss_bc": 0.04204071685671806, "train_loss_llm": 0.0, "grad_norm": 0.04617266356945038, "global_step": 6600, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04434742033481598, "train_loss_bc": 0.04434742033481598, "train_loss_llm": 0.0, "grad_norm": 0.01689021661877632, "global_step": 6601, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04251381754875183, "train_loss_bc": 0.04251018911600113, "train_loss_llm": 0.0003628730773925781, "grad_norm": 0.030304234474897385, "global_step": 6602, "epoch": 183, "lr": 0.001} +{"train_loss": 0.03250707685947418, "train_loss_bc": 0.03250707685947418, "train_loss_llm": 0.0, "grad_norm": 0.03516995534300804, "global_step": 6603, "epoch": 183, "lr": 0.001} +{"train_loss": 0.03771113231778145, "train_loss_bc": 0.03770438954234123, "train_loss_llm": 0.0006744563579559326, "grad_norm": 0.04461779445409775, "global_step": 6604, "epoch": 183, "lr": 0.001} +{"train_loss": 0.039587538689374924, "train_loss_bc": 0.039587538689374924, "train_loss_llm": 0.0, "grad_norm": 0.016880037263035774, "global_step": 6605, "epoch": 183, "lr": 0.001} +{"train_loss": 0.03887082636356354, "train_loss_bc": 0.038860078901052475, "train_loss_llm": 0.0010746121406555176, "grad_norm": 0.028315985575318336, "global_step": 6606, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04363521561026573, "train_loss_bc": 0.04363521561026573, "train_loss_llm": 0.0, "grad_norm": 0.031124800443649292, "global_step": 6607, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04350542277097702, "train_loss_bc": 0.04350542277097702, "train_loss_llm": 0.0, "grad_norm": 0.03762277588248253, "global_step": 6608, "epoch": 183, "lr": 0.001} +{"train_loss": 0.03944387659430504, "train_loss_bc": 0.039443522691726685, "train_loss_llm": 3.522634506225586e-05, "grad_norm": 0.015150058083236217, "global_step": 6609, "epoch": 183, "lr": 0.001} +{"train_loss": 0.03365818411111832, "train_loss_bc": 0.03365818411111832, "train_loss_llm": 0.0, "grad_norm": 0.01887792907655239, "global_step": 6610, "epoch": 183, "lr": 0.001} +{"train_loss": 0.03631393238902092, "train_loss_bc": 0.036289282143116, "train_loss_llm": 0.002464860677719116, "grad_norm": 0.030768554657697678, "global_step": 6611, "epoch": 183, "lr": 0.001} +{"train_loss": 0.03992654010653496, "train_loss_bc": 0.03992654010653496, "train_loss_llm": 0.0, "grad_norm": 0.03472457081079483, "global_step": 6612, "epoch": 183, "lr": 0.001} +{"train_loss": 0.038348618894815445, "train_loss_bc": 0.03834830969572067, "train_loss_llm": 3.102421760559082e-05, "grad_norm": 0.01755957491695881, "global_step": 6613, "epoch": 183, "lr": 0.001} +{"train_loss": 0.039352480322122574, "train_loss_bc": 0.03935064747929573, "train_loss_llm": 0.00018325448036193848, "grad_norm": 0.03136163949966431, "global_step": 6614, "epoch": 183, "lr": 0.001} +{"train_loss": 0.0332522951066494, "train_loss_bc": 0.0332522951066494, "train_loss_llm": 0.0, "grad_norm": 0.03308745101094246, "global_step": 6615, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04265354201197624, "train_loss_bc": 0.04265354201197624, "train_loss_llm": 0.0, "grad_norm": 0.04388211667537689, "global_step": 6616, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04309037700295448, "train_loss_bc": 0.04309037700295448, "train_loss_llm": 0.0, "grad_norm": 0.02280859835445881, "global_step": 6617, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04402819648385048, "train_loss_bc": 0.04402819648385048, "train_loss_llm": 0.0, "grad_norm": 0.033664003014564514, "global_step": 6618, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04089374840259552, "train_loss_bc": 0.04089374840259552, "train_loss_llm": 0.0, "grad_norm": 0.04120791703462601, "global_step": 6619, "epoch": 183, "lr": 0.001} +{"train_loss": 0.046496715396642685, "train_loss_bc": 0.046496715396642685, "train_loss_llm": 0.0, "grad_norm": 0.0580282025039196, "global_step": 6620, "epoch": 183, "lr": 0.001} +{"train_loss": 0.046377357095479965, "train_loss_bc": 0.046377357095479965, "train_loss_llm": 0.0, "grad_norm": 0.022896891459822655, "global_step": 6621, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04614537954330444, "train_loss_bc": 0.04614537954330444, "train_loss_llm": 0.0, "grad_norm": 0.03053193911910057, "global_step": 6622, "epoch": 183, "lr": 0.001} +{"train_loss": 0.04066946295400461, "train_loss_bc": 0.033037688583135605, "train_loss_llm": 0.0, "grad_norm": 0.04155058041214943, "global_step": 6623, "epoch": 183, "lr": 0.001, "val_loss": 0.056319817900657654} +{"train_loss": 0.047459449619054794, "train_loss_bc": 0.047459449619054794, "train_loss_llm": 0.0, "grad_norm": 0.045458853244781494, "global_step": 6624, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03409272059798241, "train_loss_bc": 0.03409272059798241, "train_loss_llm": 0.0, "grad_norm": 0.01918158493936062, "global_step": 6625, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03722001239657402, "train_loss_bc": 0.03722001239657402, "train_loss_llm": 0.0, "grad_norm": 0.030764475464820862, "global_step": 6626, "epoch": 184, "lr": 0.001} +{"train_loss": 0.044063761830329895, "train_loss_bc": 0.044063761830329895, "train_loss_llm": 0.0, "grad_norm": 0.03832711651921272, "global_step": 6627, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03641493618488312, "train_loss_bc": 0.03641493618488312, "train_loss_llm": 0.0, "grad_norm": 0.04386643320322037, "global_step": 6628, "epoch": 184, "lr": 0.001} +{"train_loss": 0.04251892492175102, "train_loss_bc": 0.04250677675008774, "train_loss_llm": 0.0012146830558776855, "grad_norm": 0.019450310617685318, "global_step": 6629, "epoch": 184, "lr": 0.001} +{"train_loss": 0.04024546965956688, "train_loss_bc": 0.04024546965956688, "train_loss_llm": 0.0, "grad_norm": 0.023244278505444527, "global_step": 6630, "epoch": 184, "lr": 0.001} +{"train_loss": 0.0453479029238224, "train_loss_bc": 0.0453479029238224, "train_loss_llm": 0.0, "grad_norm": 0.03433946147561073, "global_step": 6631, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03963505104184151, "train_loss_bc": 0.03963505104184151, "train_loss_llm": 0.0, "grad_norm": 0.03942469134926796, "global_step": 6632, "epoch": 184, "lr": 0.001} +{"train_loss": 0.040356073528528214, "train_loss_bc": 0.04035567864775658, "train_loss_llm": 3.9517879486083984e-05, "grad_norm": 0.01948598586022854, "global_step": 6633, "epoch": 184, "lr": 0.001} +{"train_loss": 0.0418853797018528, "train_loss_bc": 0.04187685251235962, "train_loss_llm": 0.0008527040481567383, "grad_norm": 0.024497177451848984, "global_step": 6634, "epoch": 184, "lr": 0.001} +{"train_loss": 0.036700792610645294, "train_loss_bc": 0.036696843802928925, "train_loss_llm": 0.0003948211669921875, "grad_norm": 0.03208310902118683, "global_step": 6635, "epoch": 184, "lr": 0.001} +{"train_loss": 0.036662571132183075, "train_loss_bc": 0.036662571132183075, "train_loss_llm": 0.0, "grad_norm": 0.037200815975666046, "global_step": 6636, "epoch": 184, "lr": 0.001} +{"train_loss": 0.04509945213794708, "train_loss_bc": 0.04509945213794708, "train_loss_llm": 0.0, "grad_norm": 0.01847037300467491, "global_step": 6637, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03991454839706421, "train_loss_bc": 0.039913758635520935, "train_loss_llm": 7.894635200500488e-05, "grad_norm": 0.028420163318514824, "global_step": 6638, "epoch": 184, "lr": 0.001} +{"train_loss": 0.042581573128700256, "train_loss_bc": 0.04253910109400749, "train_loss_llm": 0.004247158765792847, "grad_norm": 0.0379943773150444, "global_step": 6639, "epoch": 184, "lr": 0.001} +{"train_loss": 0.04220309853553772, "train_loss_bc": 0.04220309853553772, "train_loss_llm": 0.0, "grad_norm": 0.04461411014199257, "global_step": 6640, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03684763237833977, "train_loss_bc": 0.03684763237833977, "train_loss_llm": 0.0, "grad_norm": 0.01685549132525921, "global_step": 6641, "epoch": 184, "lr": 0.001} +{"train_loss": 0.04627980291843414, "train_loss_bc": 0.046269189566373825, "train_loss_llm": 0.0010613203048706055, "grad_norm": 0.02667686715722084, "global_step": 6642, "epoch": 184, "lr": 0.001} +{"train_loss": 0.047998953610658646, "train_loss_bc": 0.047998953610658646, "train_loss_llm": 0.0, "grad_norm": 0.037972357124090195, "global_step": 6643, "epoch": 184, "lr": 0.001} +{"train_loss": 0.036806605756282806, "train_loss_bc": 0.036806605756282806, "train_loss_llm": 0.0, "grad_norm": 0.03843477740883827, "global_step": 6644, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03185848519206047, "train_loss_bc": 0.03185848519206047, "train_loss_llm": 0.0, "grad_norm": 0.01737399213016033, "global_step": 6645, "epoch": 184, "lr": 0.001} +{"train_loss": 0.040756236761808395, "train_loss_bc": 0.04073674976825714, "train_loss_llm": 0.001948535442352295, "grad_norm": 0.027533814311027527, "global_step": 6646, "epoch": 184, "lr": 0.001} +{"train_loss": 0.046115998178720474, "train_loss_bc": 0.046115998178720474, "train_loss_llm": 0.0, "grad_norm": 0.0434228852391243, "global_step": 6647, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03288169577717781, "train_loss_bc": 0.03288169577717781, "train_loss_llm": 0.0, "grad_norm": 0.05100847780704498, "global_step": 6648, "epoch": 184, "lr": 0.001} +{"train_loss": 0.04525648057460785, "train_loss_bc": 0.04525648057460785, "train_loss_llm": 0.0, "grad_norm": 0.019202226772904396, "global_step": 6649, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03346870467066765, "train_loss_bc": 0.033460188657045364, "train_loss_llm": 0.0008515715599060059, "grad_norm": 0.029285335913300514, "global_step": 6650, "epoch": 184, "lr": 0.001} +{"train_loss": 0.049769725650548935, "train_loss_bc": 0.04974718391895294, "train_loss_llm": 0.0022540688514709473, "grad_norm": 0.034104470163583755, "global_step": 6651, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03524529188871384, "train_loss_bc": 0.035229943692684174, "train_loss_llm": 0.0015349090099334717, "grad_norm": 0.034912921488285065, "global_step": 6652, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03827419504523277, "train_loss_bc": 0.03827419504523277, "train_loss_llm": 0.0, "grad_norm": 0.019539261236786842, "global_step": 6653, "epoch": 184, "lr": 0.001} +{"train_loss": 0.045936644077301025, "train_loss_bc": 0.045936644077301025, "train_loss_llm": 0.0, "grad_norm": 0.031660426408052444, "global_step": 6654, "epoch": 184, "lr": 0.001} +{"train_loss": 0.0417792983353138, "train_loss_bc": 0.0417792983353138, "train_loss_llm": 0.0, "grad_norm": 0.041674066334962845, "global_step": 6655, "epoch": 184, "lr": 0.001} +{"train_loss": 0.03773370757699013, "train_loss_bc": 0.03773370757699013, "train_loss_llm": 0.0, "grad_norm": 0.045883659273386, "global_step": 6656, "epoch": 184, "lr": 0.001} +{"train_loss": 0.04178678244352341, "train_loss_bc": 0.04178678244352341, "train_loss_llm": 0.0, "grad_norm": 0.018535960465669632, "global_step": 6657, "epoch": 184, "lr": 0.001} +{"train_loss": 0.04582953080534935, "train_loss_bc": 0.04582953080534935, "train_loss_llm": 0.0, "grad_norm": 0.026368174701929092, "global_step": 6658, "epoch": 184, "lr": 0.001} +{"train_loss": 0.0406604620317618, "train_loss_bc": 0.03674914315342903, "train_loss_llm": 0.0, "grad_norm": 0.039349429309368134, "global_step": 6659, "epoch": 184, "lr": 0.001, "val_loss": 0.05627773329615593} +{"train_loss": 0.04343775287270546, "train_loss_bc": 0.04343775287270546, "train_loss_llm": 0.0, "grad_norm": 0.047136493027210236, "global_step": 6660, "epoch": 185, "lr": 0.001} +{"train_loss": 0.03579934313893318, "train_loss_bc": 0.03577195852994919, "train_loss_llm": 0.0027385950088500977, "grad_norm": 0.017908161506056786, "global_step": 6661, "epoch": 185, "lr": 0.001} +{"train_loss": 0.039999376982450485, "train_loss_bc": 0.039999376982450485, "train_loss_llm": 0.0, "grad_norm": 0.023502402007579803, "global_step": 6662, "epoch": 185, "lr": 0.001} +{"train_loss": 0.045128557831048965, "train_loss_bc": 0.04510241001844406, "train_loss_llm": 0.002614647150039673, "grad_norm": 0.029180577024817467, "global_step": 6663, "epoch": 185, "lr": 0.001} +{"train_loss": 0.03833293169736862, "train_loss_bc": 0.03833293169736862, "train_loss_llm": 0.0, "grad_norm": 0.03637116774916649, "global_step": 6664, "epoch": 185, "lr": 0.001} +{"train_loss": 0.03191745653748512, "train_loss_bc": 0.031912971287965775, "train_loss_llm": 0.00044867396354675293, "grad_norm": 0.01698196679353714, "global_step": 6665, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04492974653840065, "train_loss_bc": 0.04492974653840065, "train_loss_llm": 0.0, "grad_norm": 0.024522501975297928, "global_step": 6666, "epoch": 185, "lr": 0.001} +{"train_loss": 0.041748661547899246, "train_loss_bc": 0.041748661547899246, "train_loss_llm": 0.0, "grad_norm": 0.03207278624176979, "global_step": 6667, "epoch": 185, "lr": 0.001} +{"train_loss": 0.03190755099058151, "train_loss_bc": 0.03190755099058151, "train_loss_llm": 0.0, "grad_norm": 0.04338478296995163, "global_step": 6668, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04338834062218666, "train_loss_bc": 0.04338834062218666, "train_loss_llm": 0.0, "grad_norm": 0.0185705553740263, "global_step": 6669, "epoch": 185, "lr": 0.001} +{"train_loss": 0.03578946366906166, "train_loss_bc": 0.03578946366906166, "train_loss_llm": 0.0, "grad_norm": 0.02620193175971508, "global_step": 6670, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04047706350684166, "train_loss_bc": 0.04047706350684166, "train_loss_llm": 0.0, "grad_norm": 0.03296481445431709, "global_step": 6671, "epoch": 185, "lr": 0.001} +{"train_loss": 0.03849801421165466, "train_loss_bc": 0.03849801421165466, "train_loss_llm": 0.0, "grad_norm": 0.040399398654699326, "global_step": 6672, "epoch": 185, "lr": 0.001} +{"train_loss": 0.03953922539949417, "train_loss_bc": 0.03953922539949417, "train_loss_llm": 0.0, "grad_norm": 0.01923244260251522, "global_step": 6673, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04627217352390289, "train_loss_bc": 0.04625673592090607, "train_loss_llm": 0.001543670892715454, "grad_norm": 0.028944194316864014, "global_step": 6674, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04169012978672981, "train_loss_bc": 0.041674524545669556, "train_loss_llm": 0.0015603899955749512, "grad_norm": 0.03681761398911476, "global_step": 6675, "epoch": 185, "lr": 0.001} +{"train_loss": 0.03989022597670555, "train_loss_bc": 0.03989022597670555, "train_loss_llm": 0.0, "grad_norm": 0.04555675387382507, "global_step": 6676, "epoch": 185, "lr": 0.001} +{"train_loss": 0.037162505090236664, "train_loss_bc": 0.037162505090236664, "train_loss_llm": 0.0, "grad_norm": 0.014391329139471054, "global_step": 6677, "epoch": 185, "lr": 0.001} +{"train_loss": 0.037945009768009186, "train_loss_bc": 0.03793003782629967, "train_loss_llm": 0.0014971494674682617, "grad_norm": 0.02220328524708748, "global_step": 6678, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04368813335895538, "train_loss_bc": 0.04368813335895538, "train_loss_llm": 0.0, "grad_norm": 0.03354109823703766, "global_step": 6679, "epoch": 185, "lr": 0.001} +{"train_loss": 0.046537846326828, "train_loss_bc": 0.046537846326828, "train_loss_llm": 0.0, "grad_norm": 0.04846116155385971, "global_step": 6680, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04302455857396126, "train_loss_bc": 0.04302005097270012, "train_loss_llm": 0.0004508793354034424, "grad_norm": 0.018328996375203133, "global_step": 6681, "epoch": 185, "lr": 0.001} +{"train_loss": 0.036347854882478714, "train_loss_bc": 0.036337681114673615, "train_loss_llm": 0.0010173022747039795, "grad_norm": 0.03296177461743355, "global_step": 6682, "epoch": 185, "lr": 0.001} +{"train_loss": 0.043072659522295, "train_loss_bc": 0.043072659522295, "train_loss_llm": 0.0, "grad_norm": 0.039803966879844666, "global_step": 6683, "epoch": 185, "lr": 0.001} +{"train_loss": 0.041104357689619064, "train_loss_bc": 0.041104357689619064, "train_loss_llm": 0.0, "grad_norm": 0.04915814474225044, "global_step": 6684, "epoch": 185, "lr": 0.001} +{"train_loss": 0.036688365042209625, "train_loss_bc": 0.036674849689006805, "train_loss_llm": 0.0013516545295715332, "grad_norm": 0.018248431384563446, "global_step": 6685, "epoch": 185, "lr": 0.001} +{"train_loss": 0.0453379787504673, "train_loss_bc": 0.04530609771609306, "train_loss_llm": 0.003188192844390869, "grad_norm": 0.03567218780517578, "global_step": 6686, "epoch": 185, "lr": 0.001} +{"train_loss": 0.038790229707956314, "train_loss_bc": 0.038782037794589996, "train_loss_llm": 0.0008191764354705811, "grad_norm": 0.049520209431648254, "global_step": 6687, "epoch": 185, "lr": 0.001} +{"train_loss": 0.044613346457481384, "train_loss_bc": 0.044613346457481384, "train_loss_llm": 0.0, "grad_norm": 0.06238723173737526, "global_step": 6688, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04145754128694534, "train_loss_bc": 0.04145754128694534, "train_loss_llm": 0.0, "grad_norm": 0.016596609726548195, "global_step": 6689, "epoch": 185, "lr": 0.001} +{"train_loss": 0.043515127152204514, "train_loss_bc": 0.043515127152204514, "train_loss_llm": 0.0, "grad_norm": 0.025209546089172363, "global_step": 6690, "epoch": 185, "lr": 0.001} +{"train_loss": 0.03213752061128616, "train_loss_bc": 0.03212955594062805, "train_loss_llm": 0.0007962882518768311, "grad_norm": 0.03193555027246475, "global_step": 6691, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04405940696597099, "train_loss_bc": 0.04403526708483696, "train_loss_llm": 0.0024141669273376465, "grad_norm": 0.04148193448781967, "global_step": 6692, "epoch": 185, "lr": 0.001} +{"train_loss": 0.0439024455845356, "train_loss_bc": 0.04390237480401993, "train_loss_llm": 6.9141387939453125e-06, "grad_norm": 0.01618824154138565, "global_step": 6693, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04693600535392761, "train_loss_bc": 0.04693600535392761, "train_loss_llm": 0.0, "grad_norm": 0.025886455550789833, "global_step": 6694, "epoch": 185, "lr": 0.001} +{"train_loss": 0.04076511785387993, "train_loss_bc": 0.042477335780858994, "train_loss_llm": 0.0, "grad_norm": 0.03740642964839935, "global_step": 6695, "epoch": 185, "lr": 0.001, "train/cumulative_reward": 102.41673793008599, "train/mean_score": 8.641448382848589, "train/success_rate": 0.1, "test/cumulative_reward": 43.19374776219842, "test/mean_score": 4.100338032496184, "test/success_rate": 0.06, "val_loss": 0.055725593119859695, "train_action_mse_error": 0.044208742678165436} +{"train_loss": 0.03738298639655113, "train_loss_bc": 0.03737045079469681, "train_loss_llm": 0.0012533962726593018, "grad_norm": 0.043103188276290894, "global_step": 6696, "epoch": 186, "lr": 0.001} +{"train_loss": 0.036930304020643234, "train_loss_bc": 0.036930304020643234, "train_loss_llm": 0.0, "grad_norm": 0.01789652556180954, "global_step": 6697, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04337267950177193, "train_loss_bc": 0.04337267950177193, "train_loss_llm": 0.0, "grad_norm": 0.027960369363427162, "global_step": 6698, "epoch": 186, "lr": 0.001} +{"train_loss": 0.03827810287475586, "train_loss_bc": 0.03826974332332611, "train_loss_llm": 0.0008359849452972412, "grad_norm": 0.034243691712617874, "global_step": 6699, "epoch": 186, "lr": 0.001} +{"train_loss": 0.039706211537122726, "train_loss_bc": 0.039686936885118484, "train_loss_llm": 0.0019273459911346436, "grad_norm": 0.039931051433086395, "global_step": 6700, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04054233431816101, "train_loss_bc": 0.04054233431816101, "train_loss_llm": 0.0, "grad_norm": 0.013406866230070591, "global_step": 6701, "epoch": 186, "lr": 0.001} +{"train_loss": 0.027331961318850517, "train_loss_bc": 0.027315126731991768, "train_loss_llm": 0.0016834735870361328, "grad_norm": 0.018026091158390045, "global_step": 6702, "epoch": 186, "lr": 0.001} +{"train_loss": 0.03975125402212143, "train_loss_bc": 0.03973159193992615, "train_loss_llm": 0.0019660592079162598, "grad_norm": 0.028312282636761665, "global_step": 6703, "epoch": 186, "lr": 0.001} +{"train_loss": 0.045476626604795456, "train_loss_bc": 0.045476626604795456, "train_loss_llm": 0.0, "grad_norm": 0.04162956774234772, "global_step": 6704, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04579182714223862, "train_loss_bc": 0.04579182714223862, "train_loss_llm": 0.0, "grad_norm": 0.017371729016304016, "global_step": 6705, "epoch": 186, "lr": 0.001} +{"train_loss": 0.03443505987524986, "train_loss_bc": 0.03443505987524986, "train_loss_llm": 0.0, "grad_norm": 0.026570210233330727, "global_step": 6706, "epoch": 186, "lr": 0.001} +{"train_loss": 0.03092251531779766, "train_loss_bc": 0.03092251531779766, "train_loss_llm": 0.0, "grad_norm": 0.0308015588670969, "global_step": 6707, "epoch": 186, "lr": 0.001} +{"train_loss": 0.03815913945436478, "train_loss_bc": 0.03813878074288368, "train_loss_llm": 0.00203588604927063, "grad_norm": 0.0450279600918293, "global_step": 6708, "epoch": 186, "lr": 0.001} +{"train_loss": 0.03642835468053818, "train_loss_bc": 0.03642835468053818, "train_loss_llm": 0.0, "grad_norm": 0.017336856573820114, "global_step": 6709, "epoch": 186, "lr": 0.001} +{"train_loss": 0.0431516095995903, "train_loss_bc": 0.0431516095995903, "train_loss_llm": 0.0, "grad_norm": 0.03259757161140442, "global_step": 6710, "epoch": 186, "lr": 0.001} +{"train_loss": 0.03986735641956329, "train_loss_bc": 0.039854489266872406, "train_loss_llm": 0.0012865662574768066, "grad_norm": 0.045322712510824203, "global_step": 6711, "epoch": 186, "lr": 0.001} +{"train_loss": 0.03930488973855972, "train_loss_bc": 0.03930192068219185, "train_loss_llm": 0.000296860933303833, "grad_norm": 0.05232788994908333, "global_step": 6712, "epoch": 186, "lr": 0.001} +{"train_loss": 0.041971512138843536, "train_loss_bc": 0.041971512138843536, "train_loss_llm": 0.0, "grad_norm": 0.02040104754269123, "global_step": 6713, "epoch": 186, "lr": 0.001} +{"train_loss": 0.034668393433094025, "train_loss_bc": 0.03464933857321739, "train_loss_llm": 0.0019055306911468506, "grad_norm": 0.028345290571451187, "global_step": 6714, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04496990144252777, "train_loss_bc": 0.04496990144252777, "train_loss_llm": 0.0, "grad_norm": 0.03441685065627098, "global_step": 6715, "epoch": 186, "lr": 0.001} +{"train_loss": 0.041776955127716064, "train_loss_bc": 0.04177505522966385, "train_loss_llm": 0.00018981099128723145, "grad_norm": 0.037447504699230194, "global_step": 6716, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04997081309556961, "train_loss_bc": 0.04997081309556961, "train_loss_llm": 0.0, "grad_norm": 0.01871039718389511, "global_step": 6717, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04048321396112442, "train_loss_bc": 0.04048321396112442, "train_loss_llm": 0.0, "grad_norm": 0.0233579333871603, "global_step": 6718, "epoch": 186, "lr": 0.001} +{"train_loss": 0.044751811772584915, "train_loss_bc": 0.044751811772584915, "train_loss_llm": 0.0, "grad_norm": 0.03227255493402481, "global_step": 6719, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04165019094944, "train_loss_bc": 0.04165019094944, "train_loss_llm": 0.0, "grad_norm": 0.04123213514685631, "global_step": 6720, "epoch": 186, "lr": 0.001} +{"train_loss": 0.037370000034570694, "train_loss_bc": 0.037370000034570694, "train_loss_llm": 0.0, "grad_norm": 0.01876079849898815, "global_step": 6721, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04615698754787445, "train_loss_bc": 0.04615698754787445, "train_loss_llm": 0.0, "grad_norm": 0.03037486970424652, "global_step": 6722, "epoch": 186, "lr": 0.001} +{"train_loss": 0.040557291358709335, "train_loss_bc": 0.040557291358709335, "train_loss_llm": 0.0, "grad_norm": 0.03729179501533508, "global_step": 6723, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04510774835944176, "train_loss_bc": 0.04510774835944176, "train_loss_llm": 0.0, "grad_norm": 0.04350483417510986, "global_step": 6724, "epoch": 186, "lr": 0.001} +{"train_loss": 0.035433314740657806, "train_loss_bc": 0.035433314740657806, "train_loss_llm": 0.0, "grad_norm": 0.015525582246482372, "global_step": 6725, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04152042418718338, "train_loss_bc": 0.04152042418718338, "train_loss_llm": 0.0, "grad_norm": 0.02585628628730774, "global_step": 6726, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04487881809473038, "train_loss_bc": 0.04487881809473038, "train_loss_llm": 0.0, "grad_norm": 0.037433624267578125, "global_step": 6727, "epoch": 186, "lr": 0.001} +{"train_loss": 0.043712615966796875, "train_loss_bc": 0.04370449483394623, "train_loss_llm": 0.0008122026920318604, "grad_norm": 0.04858699068427086, "global_step": 6728, "epoch": 186, "lr": 0.001} +{"train_loss": 0.043936122208833694, "train_loss_bc": 0.04390396922826767, "train_loss_llm": 0.003215283155441284, "grad_norm": 0.019544193521142006, "global_step": 6729, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04510289430618286, "train_loss_bc": 0.04510289430618286, "train_loss_llm": 0.0, "grad_norm": 0.02826380915939808, "global_step": 6730, "epoch": 186, "lr": 0.001} +{"train_loss": 0.04067463376041916, "train_loss_bc": 0.043434593826532364, "train_loss_llm": 0.0, "grad_norm": 0.036839231848716736, "global_step": 6731, "epoch": 186, "lr": 0.001, "val_loss": 0.056102506816387177} +{"train_loss": 0.042209967970848083, "train_loss_bc": 0.042209967970848083, "train_loss_llm": 0.0, "grad_norm": 0.04488847032189369, "global_step": 6732, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04224979877471924, "train_loss_bc": 0.04224979877471924, "train_loss_llm": 0.0, "grad_norm": 0.017064422369003296, "global_step": 6733, "epoch": 187, "lr": 0.001} +{"train_loss": 0.03860695660114288, "train_loss_bc": 0.03860695660114288, "train_loss_llm": 0.0, "grad_norm": 0.023539328947663307, "global_step": 6734, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04019540920853615, "train_loss_bc": 0.04019540920853615, "train_loss_llm": 0.0, "grad_norm": 0.03165982663631439, "global_step": 6735, "epoch": 187, "lr": 0.001} +{"train_loss": 0.032739486545324326, "train_loss_bc": 0.032739486545324326, "train_loss_llm": 0.0, "grad_norm": 0.03327551856637001, "global_step": 6736, "epoch": 187, "lr": 0.001} +{"train_loss": 0.041024576872587204, "train_loss_bc": 0.04100408777594566, "train_loss_llm": 0.002049088478088379, "grad_norm": 0.014346366748213768, "global_step": 6737, "epoch": 187, "lr": 0.001} +{"train_loss": 0.042094483971595764, "train_loss_bc": 0.042094483971595764, "train_loss_llm": 0.0, "grad_norm": 0.02392396703362465, "global_step": 6738, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04446341097354889, "train_loss_bc": 0.0444597490131855, "train_loss_llm": 0.0003663301467895508, "grad_norm": 0.0305540282279253, "global_step": 6739, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04035473242402077, "train_loss_bc": 0.04035473242402077, "train_loss_llm": 0.0, "grad_norm": 0.03646377474069595, "global_step": 6740, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04165639355778694, "train_loss_bc": 0.04165639355778694, "train_loss_llm": 0.0, "grad_norm": 0.01704062707722187, "global_step": 6741, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04126211628317833, "train_loss_bc": 0.04125889390707016, "train_loss_llm": 0.0003222525119781494, "grad_norm": 0.033594902604818344, "global_step": 6742, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04049823060631752, "train_loss_bc": 0.04049823060631752, "train_loss_llm": 0.0, "grad_norm": 0.037385858595371246, "global_step": 6743, "epoch": 187, "lr": 0.001} +{"train_loss": 0.040766000747680664, "train_loss_bc": 0.040763262659311295, "train_loss_llm": 0.00027376413345336914, "grad_norm": 0.04482274875044823, "global_step": 6744, "epoch": 187, "lr": 0.001} +{"train_loss": 0.03846994787454605, "train_loss_bc": 0.038449857383966446, "train_loss_llm": 0.002009183168411255, "grad_norm": 0.017047768458724022, "global_step": 6745, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04106550291180611, "train_loss_bc": 0.04106290638446808, "train_loss_llm": 0.00025957822799682617, "grad_norm": 0.029349030926823616, "global_step": 6746, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04667423665523529, "train_loss_bc": 0.04665270075201988, "train_loss_llm": 0.002153754234313965, "grad_norm": 0.047912612557411194, "global_step": 6747, "epoch": 187, "lr": 0.001} +{"train_loss": 0.042097948491573334, "train_loss_bc": 0.042097948491573334, "train_loss_llm": 0.0, "grad_norm": 0.053197890520095825, "global_step": 6748, "epoch": 187, "lr": 0.001} +{"train_loss": 0.038288481533527374, "train_loss_bc": 0.038288481533527374, "train_loss_llm": 0.0, "grad_norm": 0.017828641459345818, "global_step": 6749, "epoch": 187, "lr": 0.001} +{"train_loss": 0.043176718056201935, "train_loss_bc": 0.043176718056201935, "train_loss_llm": 0.0, "grad_norm": 0.02692980132997036, "global_step": 6750, "epoch": 187, "lr": 0.001} +{"train_loss": 0.03877655416727066, "train_loss_bc": 0.03877655416727066, "train_loss_llm": 0.0, "grad_norm": 0.033874306827783585, "global_step": 6751, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04399751126766205, "train_loss_bc": 0.04399751126766205, "train_loss_llm": 0.0, "grad_norm": 0.0452924445271492, "global_step": 6752, "epoch": 187, "lr": 0.001} +{"train_loss": 0.0398239940404892, "train_loss_bc": 0.0398239940404892, "train_loss_llm": 0.0, "grad_norm": 0.017529116943478584, "global_step": 6753, "epoch": 187, "lr": 0.001} +{"train_loss": 0.03908844664692879, "train_loss_bc": 0.03908844664692879, "train_loss_llm": 0.0, "grad_norm": 0.02845729887485504, "global_step": 6754, "epoch": 187, "lr": 0.001} +{"train_loss": 0.048285987228155136, "train_loss_bc": 0.048272907733917236, "train_loss_llm": 0.0013081133365631104, "grad_norm": 0.03776361048221588, "global_step": 6755, "epoch": 187, "lr": 0.001} +{"train_loss": 0.033053670078516006, "train_loss_bc": 0.033053670078516006, "train_loss_llm": 0.0, "grad_norm": 0.04183417186141014, "global_step": 6756, "epoch": 187, "lr": 0.001} +{"train_loss": 0.039640914648771286, "train_loss_bc": 0.039640191942453384, "train_loss_llm": 7.24494457244873e-05, "grad_norm": 0.02162949927151203, "global_step": 6757, "epoch": 187, "lr": 0.001} +{"train_loss": 0.03780858591198921, "train_loss_bc": 0.03780858591198921, "train_loss_llm": 0.0, "grad_norm": 0.029184818267822266, "global_step": 6758, "epoch": 187, "lr": 0.001} +{"train_loss": 0.034637488424777985, "train_loss_bc": 0.034637488424777985, "train_loss_llm": 0.0, "grad_norm": 0.03733527660369873, "global_step": 6759, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04760259389877319, "train_loss_bc": 0.04760259389877319, "train_loss_llm": 0.0, "grad_norm": 0.043054841458797455, "global_step": 6760, "epoch": 187, "lr": 0.001} +{"train_loss": 0.0410599410533905, "train_loss_bc": 0.0410599410533905, "train_loss_llm": 0.0, "grad_norm": 0.01546377781778574, "global_step": 6761, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04043690115213394, "train_loss_bc": 0.04043690115213394, "train_loss_llm": 0.0, "grad_norm": 0.024233194068074226, "global_step": 6762, "epoch": 187, "lr": 0.001} +{"train_loss": 0.03960280492901802, "train_loss_bc": 0.03960280492901802, "train_loss_llm": 0.0, "grad_norm": 0.029334068298339844, "global_step": 6763, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04210991412401199, "train_loss_bc": 0.04210800677537918, "train_loss_llm": 0.00019082427024841309, "grad_norm": 0.04262465983629227, "global_step": 6764, "epoch": 187, "lr": 0.001} +{"train_loss": 0.03891842067241669, "train_loss_bc": 0.03889826312661171, "train_loss_llm": 0.0020157694816589355, "grad_norm": 0.01692272163927555, "global_step": 6765, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04090110957622528, "train_loss_bc": 0.040900979191064835, "train_loss_llm": 1.3083219528198242e-05, "grad_norm": 0.021758662536740303, "global_step": 6766, "epoch": 187, "lr": 0.001} +{"train_loss": 0.04055518936365843, "train_loss_bc": 0.03634757921099663, "train_loss_llm": 0.0, "grad_norm": 0.04001300409436226, "global_step": 6767, "epoch": 187, "lr": 0.001, "val_loss": 0.056043725460767746} +{"train_loss": 0.037555743008852005, "train_loss_bc": 0.037555743008852005, "train_loss_llm": 0.0, "grad_norm": 0.05217968299984932, "global_step": 6768, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03768101707100868, "train_loss_bc": 0.03766706585884094, "train_loss_llm": 0.0013952851295471191, "grad_norm": 0.016212627291679382, "global_step": 6769, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03515671193599701, "train_loss_bc": 0.03515671193599701, "train_loss_llm": 0.0, "grad_norm": 0.022511541843414307, "global_step": 6770, "epoch": 188, "lr": 0.001} +{"train_loss": 0.039551690220832825, "train_loss_bc": 0.03953166306018829, "train_loss_llm": 0.0020028352737426758, "grad_norm": 0.025601457804441452, "global_step": 6771, "epoch": 188, "lr": 0.001} +{"train_loss": 0.040028490126132965, "train_loss_bc": 0.040028490126132965, "train_loss_llm": 0.0, "grad_norm": 0.03845847398042679, "global_step": 6772, "epoch": 188, "lr": 0.001} +{"train_loss": 0.04272580146789551, "train_loss_bc": 0.04270413890480995, "train_loss_llm": 0.0021662116050720215, "grad_norm": 0.01743398979306221, "global_step": 6773, "epoch": 188, "lr": 0.001} +{"train_loss": 0.04337302967905998, "train_loss_bc": 0.04337302967905998, "train_loss_llm": 0.0, "grad_norm": 0.027960360050201416, "global_step": 6774, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03490397334098816, "train_loss_bc": 0.03489844128489494, "train_loss_llm": 0.0005530714988708496, "grad_norm": 0.03356300666928291, "global_step": 6775, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03880039602518082, "train_loss_bc": 0.03880039602518082, "train_loss_llm": 0.0, "grad_norm": 0.034926656633615494, "global_step": 6776, "epoch": 188, "lr": 0.001} +{"train_loss": 0.04034925252199173, "train_loss_bc": 0.04033432900905609, "train_loss_llm": 0.001492321491241455, "grad_norm": 0.015590588562190533, "global_step": 6777, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03842206299304962, "train_loss_bc": 0.03842206299304962, "train_loss_llm": 0.0, "grad_norm": 0.025267647579312325, "global_step": 6778, "epoch": 188, "lr": 0.001} +{"train_loss": 0.045850787311792374, "train_loss_bc": 0.04584313556551933, "train_loss_llm": 0.0007652044296264648, "grad_norm": 0.033336516469717026, "global_step": 6779, "epoch": 188, "lr": 0.001} +{"train_loss": 0.040592748671770096, "train_loss_bc": 0.040592748671770096, "train_loss_llm": 0.0, "grad_norm": 0.042690154165029526, "global_step": 6780, "epoch": 188, "lr": 0.001} +{"train_loss": 0.04407382011413574, "train_loss_bc": 0.04406675323843956, "train_loss_llm": 0.0007066726684570312, "grad_norm": 0.01615219935774803, "global_step": 6781, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03488527610898018, "train_loss_bc": 0.034880951046943665, "train_loss_llm": 0.0004324615001678467, "grad_norm": 0.023337440565228462, "global_step": 6782, "epoch": 188, "lr": 0.001} +{"train_loss": 0.049582891166210175, "train_loss_bc": 0.04955179989337921, "train_loss_llm": 0.003108978271484375, "grad_norm": 0.034709639847278595, "global_step": 6783, "epoch": 188, "lr": 0.001} +{"train_loss": 0.036707211285829544, "train_loss_bc": 0.036705877631902695, "train_loss_llm": 0.00013333559036254883, "grad_norm": 0.04434090852737427, "global_step": 6784, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03765002265572548, "train_loss_bc": 0.037647563964128494, "train_loss_llm": 0.00024580955505371094, "grad_norm": 0.016765577718615532, "global_step": 6785, "epoch": 188, "lr": 0.001} +{"train_loss": 0.039959829300642014, "train_loss_bc": 0.039959829300642014, "train_loss_llm": 0.0, "grad_norm": 0.023065531626343727, "global_step": 6786, "epoch": 188, "lr": 0.001} +{"train_loss": 0.035066407173871994, "train_loss_bc": 0.035066407173871994, "train_loss_llm": 0.0, "grad_norm": 0.025405721738934517, "global_step": 6787, "epoch": 188, "lr": 0.001} +{"train_loss": 0.034975890070199966, "train_loss_bc": 0.0349610336124897, "train_loss_llm": 0.0014857649803161621, "grad_norm": 0.03127339109778404, "global_step": 6788, "epoch": 188, "lr": 0.001} +{"train_loss": 0.04342016577720642, "train_loss_bc": 0.04342016577720642, "train_loss_llm": 0.0, "grad_norm": 0.020371994003653526, "global_step": 6789, "epoch": 188, "lr": 0.001} +{"train_loss": 0.04507683217525482, "train_loss_bc": 0.04507683217525482, "train_loss_llm": 0.0, "grad_norm": 0.03182075545191765, "global_step": 6790, "epoch": 188, "lr": 0.001} +{"train_loss": 0.043052829802036285, "train_loss_bc": 0.04302561283111572, "train_loss_llm": 0.002721846103668213, "grad_norm": 0.034906234592199326, "global_step": 6791, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03970590606331825, "train_loss_bc": 0.03970590606331825, "train_loss_llm": 0.0, "grad_norm": 0.04085925966501236, "global_step": 6792, "epoch": 188, "lr": 0.001} +{"train_loss": 0.044279199093580246, "train_loss_bc": 0.044279199093580246, "train_loss_llm": 0.0, "grad_norm": 0.017917916178703308, "global_step": 6793, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03729253262281418, "train_loss_bc": 0.03729253262281418, "train_loss_llm": 0.0, "grad_norm": 0.023950422182679176, "global_step": 6794, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03759301081299782, "train_loss_bc": 0.03759301081299782, "train_loss_llm": 0.0, "grad_norm": 0.03109929896891117, "global_step": 6795, "epoch": 188, "lr": 0.001} +{"train_loss": 0.042898960411548615, "train_loss_bc": 0.042898960411548615, "train_loss_llm": 0.0, "grad_norm": 0.03653472661972046, "global_step": 6796, "epoch": 188, "lr": 0.001} +{"train_loss": 0.039496175944805145, "train_loss_bc": 0.039496175944805145, "train_loss_llm": 0.0, "grad_norm": 0.017720334231853485, "global_step": 6797, "epoch": 188, "lr": 0.001} +{"train_loss": 0.04537096619606018, "train_loss_bc": 0.04537096619606018, "train_loss_llm": 0.0, "grad_norm": 0.03153149038553238, "global_step": 6798, "epoch": 188, "lr": 0.001} +{"train_loss": 0.03670911118388176, "train_loss_bc": 0.03670911118388176, "train_loss_llm": 0.0, "grad_norm": 0.03761887177824974, "global_step": 6799, "epoch": 188, "lr": 0.001} +{"train_loss": 0.04796241596341133, "train_loss_bc": 0.04792913421988487, "train_loss_llm": 0.003328263759613037, "grad_norm": 0.0509522445499897, "global_step": 6800, "epoch": 188, "lr": 0.001} +{"train_loss": 0.04208746179938316, "train_loss_bc": 0.04207668453454971, "train_loss_llm": 0.0010776817798614502, "grad_norm": 0.016976743936538696, "global_step": 6801, "epoch": 188, "lr": 0.001} +{"train_loss": 0.047689974308013916, "train_loss_bc": 0.047689974308013916, "train_loss_llm": 0.0, "grad_norm": 0.027782650664448738, "global_step": 6802, "epoch": 188, "lr": 0.001} +{"train_loss": 0.040461129819353424, "train_loss_bc": 0.03605902940034866, "train_loss_llm": 0.0013048350811004639, "grad_norm": 0.03677147626876831, "global_step": 6803, "epoch": 188, "lr": 0.001, "val_loss": 0.05646225064992905} +{"train_loss": 0.04380335658788681, "train_loss_bc": 0.04379187151789665, "train_loss_llm": 0.0011484026908874512, "grad_norm": 0.0439206063747406, "global_step": 6804, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03832774981856346, "train_loss_bc": 0.03832774981856346, "train_loss_llm": 0.0, "grad_norm": 0.015169532969594002, "global_step": 6805, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04204603657126427, "train_loss_bc": 0.04204603657126427, "train_loss_llm": 0.0, "grad_norm": 0.025004349648952484, "global_step": 6806, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03780088946223259, "train_loss_bc": 0.03780088946223259, "train_loss_llm": 0.0, "grad_norm": 0.028246859088540077, "global_step": 6807, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03668428584933281, "train_loss_bc": 0.03668428584933281, "train_loss_llm": 0.0, "grad_norm": 0.038899920880794525, "global_step": 6808, "epoch": 189, "lr": 0.001} +{"train_loss": 0.042476437985897064, "train_loss_bc": 0.0424388088285923, "train_loss_llm": 0.0037629902362823486, "grad_norm": 0.0187075212597847, "global_step": 6809, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04285840317606926, "train_loss_bc": 0.04285840317606926, "train_loss_llm": 0.0, "grad_norm": 0.02467924728989601, "global_step": 6810, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04191267862915993, "train_loss_bc": 0.04191267862915993, "train_loss_llm": 0.0, "grad_norm": 0.02887853980064392, "global_step": 6811, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03436959534883499, "train_loss_bc": 0.03436959534883499, "train_loss_llm": 0.0, "grad_norm": 0.03502003848552704, "global_step": 6812, "epoch": 189, "lr": 0.001} +{"train_loss": 0.047267209738492966, "train_loss_bc": 0.047267209738492966, "train_loss_llm": 0.0, "grad_norm": 0.01611912250518799, "global_step": 6813, "epoch": 189, "lr": 0.001} +{"train_loss": 0.0444784089922905, "train_loss_bc": 0.0444784089922905, "train_loss_llm": 0.0, "grad_norm": 0.025416335090994835, "global_step": 6814, "epoch": 189, "lr": 0.001} +{"train_loss": 0.033099811524152756, "train_loss_bc": 0.033099811524152756, "train_loss_llm": 0.0, "grad_norm": 0.034491948783397675, "global_step": 6815, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04304388538002968, "train_loss_bc": 0.04304388538002968, "train_loss_llm": 0.0, "grad_norm": 0.04414790868759155, "global_step": 6816, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04457564279437065, "train_loss_bc": 0.04454812780022621, "train_loss_llm": 0.002751678228378296, "grad_norm": 0.02246139384806156, "global_step": 6817, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04425738751888275, "train_loss_bc": 0.04425738751888275, "train_loss_llm": 0.0, "grad_norm": 0.02602296508848667, "global_step": 6818, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04564565047621727, "train_loss_bc": 0.04564407467842102, "train_loss_llm": 0.0001577138900756836, "grad_norm": 0.029018668457865715, "global_step": 6819, "epoch": 189, "lr": 0.001} +{"train_loss": 0.0425742007791996, "train_loss_bc": 0.0425742007791996, "train_loss_llm": 0.0, "grad_norm": 0.036422714591026306, "global_step": 6820, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04475614055991173, "train_loss_bc": 0.04475614055991173, "train_loss_llm": 0.0, "grad_norm": 0.018071575090289116, "global_step": 6821, "epoch": 189, "lr": 0.001} +{"train_loss": 0.038122840225696564, "train_loss_bc": 0.038122840225696564, "train_loss_llm": 0.0, "grad_norm": 0.03004376031458378, "global_step": 6822, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03909293934702873, "train_loss_bc": 0.03909293934702873, "train_loss_llm": 0.0, "grad_norm": 0.03462483733892441, "global_step": 6823, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03817252814769745, "train_loss_bc": 0.03817252814769745, "train_loss_llm": 0.0, "grad_norm": 0.04534310847520828, "global_step": 6824, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03445350006222725, "train_loss_bc": 0.03445350006222725, "train_loss_llm": 0.0, "grad_norm": 0.015487764030694962, "global_step": 6825, "epoch": 189, "lr": 0.001} +{"train_loss": 0.027751190587878227, "train_loss_bc": 0.027751190587878227, "train_loss_llm": 0.0, "grad_norm": 0.02399837225675583, "global_step": 6826, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04665450006723404, "train_loss_bc": 0.04665203392505646, "train_loss_llm": 0.0002465546131134033, "grad_norm": 0.03511759266257286, "global_step": 6827, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04075538367033005, "train_loss_bc": 0.04075538367033005, "train_loss_llm": 0.0, "grad_norm": 0.046348974108695984, "global_step": 6828, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03895597159862518, "train_loss_bc": 0.038944683969020844, "train_loss_llm": 0.0011285841464996338, "grad_norm": 0.019191693514585495, "global_step": 6829, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04406163468956947, "train_loss_bc": 0.04405789077281952, "train_loss_llm": 0.00037428736686706543, "grad_norm": 0.02897251397371292, "global_step": 6830, "epoch": 189, "lr": 0.001} +{"train_loss": 0.038481704890728, "train_loss_bc": 0.038480356335639954, "train_loss_llm": 0.0001347661018371582, "grad_norm": 0.037599314004182816, "global_step": 6831, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03312310203909874, "train_loss_bc": 0.03312310203909874, "train_loss_llm": 0.0, "grad_norm": 0.04093340039253235, "global_step": 6832, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04349685460329056, "train_loss_bc": 0.04349229484796524, "train_loss_llm": 0.00045615434646606445, "grad_norm": 0.01919073611497879, "global_step": 6833, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03980609029531479, "train_loss_bc": 0.03980609029531479, "train_loss_llm": 0.0, "grad_norm": 0.027166523039340973, "global_step": 6834, "epoch": 189, "lr": 0.001} +{"train_loss": 0.03259092941880226, "train_loss_bc": 0.03259092941880226, "train_loss_llm": 0.0, "grad_norm": 0.037833135575056076, "global_step": 6835, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04330982267856598, "train_loss_bc": 0.04328800365328789, "train_loss_llm": 0.0021820068359375, "grad_norm": 0.045536577701568604, "global_step": 6836, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04291585832834244, "train_loss_bc": 0.04285483807325363, "train_loss_llm": 0.006101936101913452, "grad_norm": 0.016399193555116653, "global_step": 6837, "epoch": 189, "lr": 0.001} +{"train_loss": 0.04522005841135979, "train_loss_bc": 0.04522005841135979, "train_loss_llm": 0.0, "grad_norm": 0.03257179260253906, "global_step": 6838, "epoch": 189, "lr": 0.001} +{"train_loss": 0.040446579094148345, "train_loss_bc": 0.03911614790558815, "train_loss_llm": 0.0018020868301391602, "grad_norm": 0.03865044191479683, "global_step": 6839, "epoch": 189, "lr": 0.001, "val_loss": 0.05655180662870407} +{"train_loss": 0.03886796534061432, "train_loss_bc": 0.03886796534061432, "train_loss_llm": 0.0, "grad_norm": 0.043786853551864624, "global_step": 6840, "epoch": 190, "lr": 0.001} +{"train_loss": 0.03355931490659714, "train_loss_bc": 0.03355931490659714, "train_loss_llm": 0.0, "grad_norm": 0.01629871316254139, "global_step": 6841, "epoch": 190, "lr": 0.001} +{"train_loss": 0.03873750567436218, "train_loss_bc": 0.03873750567436218, "train_loss_llm": 0.0, "grad_norm": 0.0206290390342474, "global_step": 6842, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04426656290888786, "train_loss_bc": 0.04426656290888786, "train_loss_llm": 0.0, "grad_norm": 0.02735903300344944, "global_step": 6843, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04534004256129265, "train_loss_bc": 0.04534004256129265, "train_loss_llm": 0.0, "grad_norm": 0.03620299696922302, "global_step": 6844, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04215282201766968, "train_loss_bc": 0.04215282201766968, "train_loss_llm": 0.0, "grad_norm": 0.0176708921790123, "global_step": 6845, "epoch": 190, "lr": 0.001} +{"train_loss": 0.0451534204185009, "train_loss_bc": 0.045138169080019, "train_loss_llm": 0.0015252530574798584, "grad_norm": 0.03077550046145916, "global_step": 6846, "epoch": 190, "lr": 0.001} +{"train_loss": 0.029933739453554153, "train_loss_bc": 0.029933739453554153, "train_loss_llm": 0.0, "grad_norm": 0.0377204567193985, "global_step": 6847, "epoch": 190, "lr": 0.001} +{"train_loss": 0.0314062274992466, "train_loss_bc": 0.03140220791101456, "train_loss_llm": 0.00040209293365478516, "grad_norm": 0.04293255880475044, "global_step": 6848, "epoch": 190, "lr": 0.001} +{"train_loss": 0.03724851831793785, "train_loss_bc": 0.03724564239382744, "train_loss_llm": 0.00028759241104125977, "grad_norm": 0.018196608871221542, "global_step": 6849, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04801035299897194, "train_loss_bc": 0.04801035299897194, "train_loss_llm": 0.0, "grad_norm": 0.027753766626119614, "global_step": 6850, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04096699506044388, "train_loss_bc": 0.04095263034105301, "train_loss_llm": 0.0014365315437316895, "grad_norm": 0.038252588361501694, "global_step": 6851, "epoch": 190, "lr": 0.001} +{"train_loss": 0.041428882628679276, "train_loss_bc": 0.041428882628679276, "train_loss_llm": 0.0, "grad_norm": 0.04950202628970146, "global_step": 6852, "epoch": 190, "lr": 0.001} +{"train_loss": 0.041309699416160583, "train_loss_bc": 0.041309699416160583, "train_loss_llm": 0.0, "grad_norm": 0.021645912900567055, "global_step": 6853, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04050600901246071, "train_loss_bc": 0.04050350561738014, "train_loss_llm": 0.0002505183219909668, "grad_norm": 0.03276956453919411, "global_step": 6854, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04423157498240471, "train_loss_bc": 0.044227756559848785, "train_loss_llm": 0.00038176774978637695, "grad_norm": 0.03839195519685745, "global_step": 6855, "epoch": 190, "lr": 0.001} +{"train_loss": 0.040785130113363266, "train_loss_bc": 0.04078150913119316, "train_loss_llm": 0.00036212801933288574, "grad_norm": 0.047099243849515915, "global_step": 6856, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04662651568651199, "train_loss_bc": 0.04660196602344513, "train_loss_llm": 0.0024549663066864014, "grad_norm": 0.027237458154559135, "global_step": 6857, "epoch": 190, "lr": 0.001} +{"train_loss": 0.035016678273677826, "train_loss_bc": 0.03499334678053856, "train_loss_llm": 0.0023331046104431152, "grad_norm": 0.037727728486061096, "global_step": 6858, "epoch": 190, "lr": 0.001} +{"train_loss": 0.028925539925694466, "train_loss_bc": 0.028925539925694466, "train_loss_llm": 0.0, "grad_norm": 0.04414794221520424, "global_step": 6859, "epoch": 190, "lr": 0.001} +{"train_loss": 0.045629750937223434, "train_loss_bc": 0.045629750937223434, "train_loss_llm": 0.0, "grad_norm": 0.04908008873462677, "global_step": 6860, "epoch": 190, "lr": 0.001} +{"train_loss": 0.03908507525920868, "train_loss_bc": 0.03908507525920868, "train_loss_llm": 0.0, "grad_norm": 0.01764414831995964, "global_step": 6861, "epoch": 190, "lr": 0.001} +{"train_loss": 0.041671864688396454, "train_loss_bc": 0.041671864688396454, "train_loss_llm": 0.0, "grad_norm": 0.02702569216489792, "global_step": 6862, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04022364318370819, "train_loss_bc": 0.04022364318370819, "train_loss_llm": 0.0, "grad_norm": 0.040594059973955154, "global_step": 6863, "epoch": 190, "lr": 0.001} +{"train_loss": 0.041066888719797134, "train_loss_bc": 0.04105082526803017, "train_loss_llm": 0.0016061663627624512, "grad_norm": 0.050537969917058945, "global_step": 6864, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04057706147432327, "train_loss_bc": 0.04057346284389496, "train_loss_llm": 0.00036004185676574707, "grad_norm": 0.017788724973797798, "global_step": 6865, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04036515951156616, "train_loss_bc": 0.04032537713646889, "train_loss_llm": 0.003978163003921509, "grad_norm": 0.026731429621577263, "global_step": 6866, "epoch": 190, "lr": 0.001} +{"train_loss": 0.047116681933403015, "train_loss_bc": 0.04711637645959854, "train_loss_llm": 3.0606985092163086e-05, "grad_norm": 0.03404228389263153, "global_step": 6867, "epoch": 190, "lr": 0.001} +{"train_loss": 0.045013509690761566, "train_loss_bc": 0.045013509690761566, "train_loss_llm": 0.0, "grad_norm": 0.04240860790014267, "global_step": 6868, "epoch": 190, "lr": 0.001} +{"train_loss": 0.035339586436748505, "train_loss_bc": 0.035339586436748505, "train_loss_llm": 0.0, "grad_norm": 0.018749263137578964, "global_step": 6869, "epoch": 190, "lr": 0.001} +{"train_loss": 0.039525438100099564, "train_loss_bc": 0.039525438100099564, "train_loss_llm": 0.0, "grad_norm": 0.031819362193346024, "global_step": 6870, "epoch": 190, "lr": 0.001} +{"train_loss": 0.042653389275074005, "train_loss_bc": 0.042653389275074005, "train_loss_llm": 0.0, "grad_norm": 0.04422740638256073, "global_step": 6871, "epoch": 190, "lr": 0.001} +{"train_loss": 0.03969801217317581, "train_loss_bc": 0.0396869033575058, "train_loss_llm": 0.001111060380935669, "grad_norm": 0.05819671228528023, "global_step": 6872, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04568399488925934, "train_loss_bc": 0.04568399488925934, "train_loss_llm": 0.0, "grad_norm": 0.020078586414456367, "global_step": 6873, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04211360961198807, "train_loss_bc": 0.04209014028310776, "train_loss_llm": 0.002347022294998169, "grad_norm": 0.03392469510436058, "global_step": 6874, "epoch": 190, "lr": 0.001} +{"train_loss": 0.04056839639734891, "train_loss_bc": 0.040225107222795486, "train_loss_llm": 0.0, "grad_norm": 0.04168900102376938, "global_step": 6875, "epoch": 190, "lr": 0.001, "train/cumulative_reward": 63.8907331866428, "train/mean_score": 23.842535999623504, "train/success_rate": 0.7, "test/cumulative_reward": 50.769418689729775, "test/mean_score": 9.854288549574722, "test/success_rate": 0.26, "val_loss": 0.056237172335386276, "train_action_mse_error": 0.04454750940203667} +{"train_loss": 0.039699528366327286, "train_loss_bc": 0.03968726098537445, "train_loss_llm": 0.0012267827987670898, "grad_norm": 0.04968688637018204, "global_step": 6876, "epoch": 191, "lr": 0.001} +{"train_loss": 0.034996625036001205, "train_loss_bc": 0.03499419242143631, "train_loss_llm": 0.00024324655532836914, "grad_norm": 0.01910238340497017, "global_step": 6877, "epoch": 191, "lr": 0.001} +{"train_loss": 0.045994434505701065, "train_loss_bc": 0.045981116592884064, "train_loss_llm": 0.0013318061828613281, "grad_norm": 0.038781698793172836, "global_step": 6878, "epoch": 191, "lr": 0.001} +{"train_loss": 0.045074786990880966, "train_loss_bc": 0.045074786990880966, "train_loss_llm": 0.0, "grad_norm": 0.05194588750600815, "global_step": 6879, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04267134144902229, "train_loss_bc": 0.042664531618356705, "train_loss_llm": 0.0006810128688812256, "grad_norm": 0.05920000374317169, "global_step": 6880, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04059629514813423, "train_loss_bc": 0.04059629514813423, "train_loss_llm": 0.0, "grad_norm": 0.023562069982290268, "global_step": 6881, "epoch": 191, "lr": 0.001} +{"train_loss": 0.03389586880803108, "train_loss_bc": 0.03389586880803108, "train_loss_llm": 0.0, "grad_norm": 0.03122766688466072, "global_step": 6882, "epoch": 191, "lr": 0.001} +{"train_loss": 0.045601315796375275, "train_loss_bc": 0.04559604823589325, "train_loss_llm": 0.0005269348621368408, "grad_norm": 0.03710532188415527, "global_step": 6883, "epoch": 191, "lr": 0.001} +{"train_loss": 0.03782399743795395, "train_loss_bc": 0.03782399743795395, "train_loss_llm": 0.0, "grad_norm": 0.04273264482617378, "global_step": 6884, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04057424142956734, "train_loss_bc": 0.04057424142956734, "train_loss_llm": 0.0, "grad_norm": 0.018680129200220108, "global_step": 6885, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04104182496666908, "train_loss_bc": 0.04104182496666908, "train_loss_llm": 0.0, "grad_norm": 0.03588030859827995, "global_step": 6886, "epoch": 191, "lr": 0.001} +{"train_loss": 0.03777425363659859, "train_loss_bc": 0.03777425363659859, "train_loss_llm": 0.0, "grad_norm": 0.04959743097424507, "global_step": 6887, "epoch": 191, "lr": 0.001} +{"train_loss": 0.034831441938877106, "train_loss_bc": 0.034831441938877106, "train_loss_llm": 0.0, "grad_norm": 0.05513075739145279, "global_step": 6888, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04218025505542755, "train_loss_bc": 0.04218025505542755, "train_loss_llm": 0.0, "grad_norm": 0.021076690405607224, "global_step": 6889, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04287566989660263, "train_loss_bc": 0.04285719618201256, "train_loss_llm": 0.0018474459648132324, "grad_norm": 0.03133794292807579, "global_step": 6890, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04148818179965019, "train_loss_bc": 0.04148818179965019, "train_loss_llm": 0.0, "grad_norm": 0.038933005183935165, "global_step": 6891, "epoch": 191, "lr": 0.001} +{"train_loss": 0.042377717792987823, "train_loss_bc": 0.042377717792987823, "train_loss_llm": 0.0, "grad_norm": 0.042799174785614014, "global_step": 6892, "epoch": 191, "lr": 0.001} +{"train_loss": 0.040853481739759445, "train_loss_bc": 0.0408448725938797, "train_loss_llm": 0.0008610188961029053, "grad_norm": 0.016807569190859795, "global_step": 6893, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04527495056390762, "train_loss_bc": 0.04527495056390762, "train_loss_llm": 0.0, "grad_norm": 0.028078526258468628, "global_step": 6894, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04344567283987999, "train_loss_bc": 0.04344567283987999, "train_loss_llm": 0.0, "grad_norm": 0.040704745799303055, "global_step": 6895, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04115607962012291, "train_loss_bc": 0.04115607962012291, "train_loss_llm": 0.0, "grad_norm": 0.05102096498012543, "global_step": 6896, "epoch": 191, "lr": 0.001} +{"train_loss": 0.035069700330495834, "train_loss_bc": 0.035069700330495834, "train_loss_llm": 0.0, "grad_norm": 0.02504204399883747, "global_step": 6897, "epoch": 191, "lr": 0.001} +{"train_loss": 0.0369107648730278, "train_loss_bc": 0.0369107648730278, "train_loss_llm": 0.0, "grad_norm": 0.030399447306990623, "global_step": 6898, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04259003326296806, "train_loss_bc": 0.042588457465171814, "train_loss_llm": 0.0001577436923980713, "grad_norm": 0.04642649367451668, "global_step": 6899, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04390140250325203, "train_loss_bc": 0.04390140250325203, "train_loss_llm": 0.0, "grad_norm": 0.06814456731081009, "global_step": 6900, "epoch": 191, "lr": 0.001} +{"train_loss": 0.042522773146629333, "train_loss_bc": 0.042522773146629333, "train_loss_llm": 0.0, "grad_norm": 0.01708086021244526, "global_step": 6901, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04330480471253395, "train_loss_bc": 0.043283384293317795, "train_loss_llm": 0.0021421611309051514, "grad_norm": 0.03160100802779198, "global_step": 6902, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04580874741077423, "train_loss_bc": 0.04577508941292763, "train_loss_llm": 0.003365755081176758, "grad_norm": 0.04378800839185715, "global_step": 6903, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04597991704940796, "train_loss_bc": 0.04597991704940796, "train_loss_llm": 0.0, "grad_norm": 0.054118335247039795, "global_step": 6904, "epoch": 191, "lr": 0.001} +{"train_loss": 0.035834480077028275, "train_loss_bc": 0.035834480077028275, "train_loss_llm": 0.0, "grad_norm": 0.016270587220788002, "global_step": 6905, "epoch": 191, "lr": 0.001} +{"train_loss": 0.039357203990221024, "train_loss_bc": 0.039357203990221024, "train_loss_llm": 0.0, "grad_norm": 0.02153760939836502, "global_step": 6906, "epoch": 191, "lr": 0.001} +{"train_loss": 0.038572728633880615, "train_loss_bc": 0.038560010492801666, "train_loss_llm": 0.0012719929218292236, "grad_norm": 0.028739679604768753, "global_step": 6907, "epoch": 191, "lr": 0.001} +{"train_loss": 0.036389268934726715, "train_loss_bc": 0.03638274222612381, "train_loss_llm": 0.000652700662612915, "grad_norm": 0.040489841252565384, "global_step": 6908, "epoch": 191, "lr": 0.001} +{"train_loss": 0.03959983214735985, "train_loss_bc": 0.03959983214735985, "train_loss_llm": 0.0, "grad_norm": 0.015196289867162704, "global_step": 6909, "epoch": 191, "lr": 0.001} +{"train_loss": 0.03857143223285675, "train_loss_bc": 0.03857143223285675, "train_loss_llm": 0.0, "grad_norm": 0.024767836555838585, "global_step": 6910, "epoch": 191, "lr": 0.001} +{"train_loss": 0.04057046574436956, "train_loss_bc": 0.03588343784213066, "train_loss_llm": 0.0012275874614715576, "grad_norm": 0.03927771747112274, "global_step": 6911, "epoch": 191, "lr": 0.001, "val_loss": 0.056434839963912964} +{"train_loss": 0.04002387076616287, "train_loss_bc": 0.04002387076616287, "train_loss_llm": 0.0, "grad_norm": 0.051134996116161346, "global_step": 6912, "epoch": 192, "lr": 0.001} +{"train_loss": 0.043421048671007156, "train_loss_bc": 0.043421048671007156, "train_loss_llm": 0.0, "grad_norm": 0.01943526417016983, "global_step": 6913, "epoch": 192, "lr": 0.001} +{"train_loss": 0.03693843260407448, "train_loss_bc": 0.03693264350295067, "train_loss_llm": 0.0005788803100585938, "grad_norm": 0.03268302604556084, "global_step": 6914, "epoch": 192, "lr": 0.001} +{"train_loss": 0.03718896955251694, "train_loss_bc": 0.03718896955251694, "train_loss_llm": 0.0, "grad_norm": 0.03933345153927803, "global_step": 6915, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04634023457765579, "train_loss_bc": 0.04634023457765579, "train_loss_llm": 0.0, "grad_norm": 0.04894987493753433, "global_step": 6916, "epoch": 192, "lr": 0.001} +{"train_loss": 0.037016741931438446, "train_loss_bc": 0.037016741931438446, "train_loss_llm": 0.0, "grad_norm": 0.01847141981124878, "global_step": 6917, "epoch": 192, "lr": 0.001} +{"train_loss": 0.03692489117383957, "train_loss_bc": 0.03692489117383957, "train_loss_llm": 0.0, "grad_norm": 0.028158262372016907, "global_step": 6918, "epoch": 192, "lr": 0.001} +{"train_loss": 0.03882252424955368, "train_loss_bc": 0.03882252424955368, "train_loss_llm": 0.0, "grad_norm": 0.03509366512298584, "global_step": 6919, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04733187332749367, "train_loss_bc": 0.04733187332749367, "train_loss_llm": 0.0, "grad_norm": 0.04655144363641739, "global_step": 6920, "epoch": 192, "lr": 0.001} +{"train_loss": 0.043380215764045715, "train_loss_bc": 0.043380215764045715, "train_loss_llm": 0.0, "grad_norm": 0.01671484485268593, "global_step": 6921, "epoch": 192, "lr": 0.001} +{"train_loss": 0.038132939487695694, "train_loss_bc": 0.038132939487695694, "train_loss_llm": 0.0, "grad_norm": 0.02765515446662903, "global_step": 6922, "epoch": 192, "lr": 0.001} +{"train_loss": 0.047230180352926254, "train_loss_bc": 0.047216661274433136, "train_loss_llm": 0.0013520419597625732, "grad_norm": 0.0374717079102993, "global_step": 6923, "epoch": 192, "lr": 0.001} +{"train_loss": 0.0357886366546154, "train_loss_bc": 0.03578170761466026, "train_loss_llm": 0.000692903995513916, "grad_norm": 0.040022991597652435, "global_step": 6924, "epoch": 192, "lr": 0.001} +{"train_loss": 0.040738821029663086, "train_loss_bc": 0.040738821029663086, "train_loss_llm": 0.0, "grad_norm": 0.017929252237081528, "global_step": 6925, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04237472265958786, "train_loss_bc": 0.04236966371536255, "train_loss_llm": 0.0005057156085968018, "grad_norm": 0.030500179156661034, "global_step": 6926, "epoch": 192, "lr": 0.001} +{"train_loss": 0.03638578951358795, "train_loss_bc": 0.03638578951358795, "train_loss_llm": 0.0, "grad_norm": 0.03498408570885658, "global_step": 6927, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04242868348956108, "train_loss_bc": 0.04242868348956108, "train_loss_llm": 0.0, "grad_norm": 0.038333356380462646, "global_step": 6928, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04566410556435585, "train_loss_bc": 0.04566410556435585, "train_loss_llm": 0.0, "grad_norm": 0.016315964981913567, "global_step": 6929, "epoch": 192, "lr": 0.001} +{"train_loss": 0.03611062839627266, "train_loss_bc": 0.03607059270143509, "train_loss_llm": 0.00400349497795105, "grad_norm": 0.03050614893436432, "global_step": 6930, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04497714340686798, "train_loss_bc": 0.04497714340686798, "train_loss_llm": 0.0, "grad_norm": 0.03422974795103073, "global_step": 6931, "epoch": 192, "lr": 0.001} +{"train_loss": 0.02983936481177807, "train_loss_bc": 0.029828010126948357, "train_loss_llm": 0.0011355280876159668, "grad_norm": 0.03759972006082535, "global_step": 6932, "epoch": 192, "lr": 0.001} +{"train_loss": 0.039071958512067795, "train_loss_bc": 0.039071958512067795, "train_loss_llm": 0.0, "grad_norm": 0.017462339252233505, "global_step": 6933, "epoch": 192, "lr": 0.001} +{"train_loss": 0.040021054446697235, "train_loss_bc": 0.040021054446697235, "train_loss_llm": 0.0, "grad_norm": 0.02640787698328495, "global_step": 6934, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04297640547156334, "train_loss_bc": 0.04296570271253586, "train_loss_llm": 0.0010703206062316895, "grad_norm": 0.03672228381037712, "global_step": 6935, "epoch": 192, "lr": 0.001} +{"train_loss": 0.03880639001727104, "train_loss_bc": 0.0387921966612339, "train_loss_llm": 0.0014193952083587646, "grad_norm": 0.048344552516937256, "global_step": 6936, "epoch": 192, "lr": 0.001} +{"train_loss": 0.041874587535858154, "train_loss_bc": 0.041874587535858154, "train_loss_llm": 0.0, "grad_norm": 0.01681005209684372, "global_step": 6937, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04277591034770012, "train_loss_bc": 0.04277591034770012, "train_loss_llm": 0.0, "grad_norm": 0.02699943818151951, "global_step": 6938, "epoch": 192, "lr": 0.001} +{"train_loss": 0.03746366873383522, "train_loss_bc": 0.03746366873383522, "train_loss_llm": 0.0, "grad_norm": 0.037652645260095596, "global_step": 6939, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04387710988521576, "train_loss_bc": 0.04387710988521576, "train_loss_llm": 0.0, "grad_norm": 0.048107556998729706, "global_step": 6940, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04640241712331772, "train_loss_bc": 0.046385928988456726, "train_loss_llm": 0.0016489028930664062, "grad_norm": 0.017062043771147728, "global_step": 6941, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04199974238872528, "train_loss_bc": 0.04198490083217621, "train_loss_llm": 0.0014841854572296143, "grad_norm": 0.03063056245446205, "global_step": 6942, "epoch": 192, "lr": 0.001} +{"train_loss": 0.0433659628033638, "train_loss_bc": 0.04335368052124977, "train_loss_llm": 0.0012281835079193115, "grad_norm": 0.0399523489177227, "global_step": 6943, "epoch": 192, "lr": 0.001} +{"train_loss": 0.03283856436610222, "train_loss_bc": 0.03283664211630821, "train_loss_llm": 0.00019237399101257324, "grad_norm": 0.05348927155137062, "global_step": 6944, "epoch": 192, "lr": 0.001} +{"train_loss": 0.03204108402132988, "train_loss_bc": 0.03204108402132988, "train_loss_llm": 0.0, "grad_norm": 0.014196332544088364, "global_step": 6945, "epoch": 192, "lr": 0.001} +{"train_loss": 0.041806988418102264, "train_loss_bc": 0.041806988418102264, "train_loss_llm": 0.0, "grad_norm": 0.029214711859822273, "global_step": 6946, "epoch": 192, "lr": 0.001} +{"train_loss": 0.04050366740880741, "train_loss_bc": 0.045750364661216736, "train_loss_llm": 0.0, "grad_norm": 0.040375903248786926, "global_step": 6947, "epoch": 192, "lr": 0.001, "val_loss": 0.05684489384293556} +{"train_loss": 0.04432433843612671, "train_loss_bc": 0.044304825365543365, "train_loss_llm": 0.0019513368606567383, "grad_norm": 0.04235202074050903, "global_step": 6948, "epoch": 193, "lr": 0.001} +{"train_loss": 0.040872346609830856, "train_loss_bc": 0.040840692818164825, "train_loss_llm": 0.003165215253829956, "grad_norm": 0.015764852985739708, "global_step": 6949, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03379300609230995, "train_loss_bc": 0.03379300609230995, "train_loss_llm": 0.0, "grad_norm": 0.023696167394518852, "global_step": 6950, "epoch": 193, "lr": 0.001} +{"train_loss": 0.032841891050338745, "train_loss_bc": 0.032841891050338745, "train_loss_llm": 0.0, "grad_norm": 0.035379890352487564, "global_step": 6951, "epoch": 193, "lr": 0.001} +{"train_loss": 0.044783685356378555, "train_loss_bc": 0.044783685356378555, "train_loss_llm": 0.0, "grad_norm": 0.046897709369659424, "global_step": 6952, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04502781480550766, "train_loss_bc": 0.04502300173044205, "train_loss_llm": 0.0004814267158508301, "grad_norm": 0.01754598505795002, "global_step": 6953, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03980429470539093, "train_loss_bc": 0.03979698196053505, "train_loss_llm": 0.0007313191890716553, "grad_norm": 0.02980818599462509, "global_step": 6954, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04128469526767731, "train_loss_bc": 0.04128469526767731, "train_loss_llm": 0.0, "grad_norm": 0.045823048800230026, "global_step": 6955, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04071081057190895, "train_loss_bc": 0.04071081057190895, "train_loss_llm": 0.0, "grad_norm": 0.056972574442625046, "global_step": 6956, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03685649484395981, "train_loss_bc": 0.03685649484395981, "train_loss_llm": 0.0, "grad_norm": 0.01593567617237568, "global_step": 6957, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03959021717309952, "train_loss_bc": 0.03959021717309952, "train_loss_llm": 0.0, "grad_norm": 0.02560662478208542, "global_step": 6958, "epoch": 193, "lr": 0.001} +{"train_loss": 0.039196163415908813, "train_loss_bc": 0.039196163415908813, "train_loss_llm": 0.0, "grad_norm": 0.032666951417922974, "global_step": 6959, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04098286107182503, "train_loss_bc": 0.04098286107182503, "train_loss_llm": 0.0, "grad_norm": 0.0422215573489666, "global_step": 6960, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03219892084598541, "train_loss_bc": 0.032187119126319885, "train_loss_llm": 0.001180201768875122, "grad_norm": 0.016729477792978287, "global_step": 6961, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03981791436672211, "train_loss_bc": 0.03981791436672211, "train_loss_llm": 0.0, "grad_norm": 0.02125799097120762, "global_step": 6962, "epoch": 193, "lr": 0.001} +{"train_loss": 0.0376603864133358, "train_loss_bc": 0.0376603864133358, "train_loss_llm": 0.0, "grad_norm": 0.027908088639378548, "global_step": 6963, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04274618253111839, "train_loss_bc": 0.0427299439907074, "train_loss_llm": 0.0016239583492279053, "grad_norm": 0.03262690454721451, "global_step": 6964, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04174415394663811, "train_loss_bc": 0.04174415394663811, "train_loss_llm": 0.0, "grad_norm": 0.02098998799920082, "global_step": 6965, "epoch": 193, "lr": 0.001} +{"train_loss": 0.043650973588228226, "train_loss_bc": 0.043645575642585754, "train_loss_llm": 0.0005398392677307129, "grad_norm": 0.03091510571539402, "global_step": 6966, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03277371823787689, "train_loss_bc": 0.032768744975328445, "train_loss_llm": 0.0004971921443939209, "grad_norm": 0.035526376217603683, "global_step": 6967, "epoch": 193, "lr": 0.001} +{"train_loss": 0.042157143354415894, "train_loss_bc": 0.042151182889938354, "train_loss_llm": 0.0005961060523986816, "grad_norm": 0.037776313722133636, "global_step": 6968, "epoch": 193, "lr": 0.001} +{"train_loss": 0.045944973826408386, "train_loss_bc": 0.04592564329504967, "train_loss_llm": 0.001932978630065918, "grad_norm": 0.020796554163098335, "global_step": 6969, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03897048532962799, "train_loss_bc": 0.03897048532962799, "train_loss_llm": 0.0, "grad_norm": 0.02703041210770607, "global_step": 6970, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04850442707538605, "train_loss_bc": 0.04850442707538605, "train_loss_llm": 0.0, "grad_norm": 0.037752993404865265, "global_step": 6971, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04057397320866585, "train_loss_bc": 0.04057397320866585, "train_loss_llm": 0.0, "grad_norm": 0.04729819670319557, "global_step": 6972, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03822677209973335, "train_loss_bc": 0.03820597007870674, "train_loss_llm": 0.0020802319049835205, "grad_norm": 0.019646920263767242, "global_step": 6973, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04539410024881363, "train_loss_bc": 0.04539398476481438, "train_loss_llm": 1.1593103408813477e-05, "grad_norm": 0.02744274213910103, "global_step": 6974, "epoch": 193, "lr": 0.001} +{"train_loss": 0.043899573385715485, "train_loss_bc": 0.04389525577425957, "train_loss_llm": 0.0004316568374633789, "grad_norm": 0.03771168738603592, "global_step": 6975, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03953685984015465, "train_loss_bc": 0.03953685984015465, "train_loss_llm": 0.0, "grad_norm": 0.04974112659692764, "global_step": 6976, "epoch": 193, "lr": 0.001} +{"train_loss": 0.039581187069416046, "train_loss_bc": 0.039581187069416046, "train_loss_llm": 0.0, "grad_norm": 0.015896640717983246, "global_step": 6977, "epoch": 193, "lr": 0.001} +{"train_loss": 0.038761723786592484, "train_loss_bc": 0.038761723786592484, "train_loss_llm": 0.0, "grad_norm": 0.024215929210186005, "global_step": 6978, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04231009632349014, "train_loss_bc": 0.04231009632349014, "train_loss_llm": 0.0, "grad_norm": 0.037077806890010834, "global_step": 6979, "epoch": 193, "lr": 0.001} +{"train_loss": 0.0392906479537487, "train_loss_bc": 0.039290279150009155, "train_loss_llm": 3.6716461181640625e-05, "grad_norm": 0.04276837036013603, "global_step": 6980, "epoch": 193, "lr": 0.001} +{"train_loss": 0.04232150316238403, "train_loss_bc": 0.04232150316238403, "train_loss_llm": 0.0, "grad_norm": 0.018169233575463295, "global_step": 6981, "epoch": 193, "lr": 0.001} +{"train_loss": 0.03900553658604622, "train_loss_bc": 0.03900553658604622, "train_loss_llm": 0.0, "grad_norm": 0.028849279507994652, "global_step": 6982, "epoch": 193, "lr": 0.001} +{"train_loss": 0.040529838452736534, "train_loss_bc": 0.043917518109083176, "train_loss_llm": 0.0016792714595794678, "grad_norm": 0.038046058267354965, "global_step": 6983, "epoch": 193, "lr": 0.001, "val_loss": 0.05663488060235977} +{"train_loss": 0.037980545312166214, "train_loss_bc": 0.037974871695041656, "train_loss_llm": 0.0005674064159393311, "grad_norm": 0.044059738516807556, "global_step": 6984, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04173482209444046, "train_loss_bc": 0.04173482209444046, "train_loss_llm": 0.0, "grad_norm": 0.01879741996526718, "global_step": 6985, "epoch": 194, "lr": 0.001} +{"train_loss": 0.03649327531456947, "train_loss_bc": 0.036489084362983704, "train_loss_llm": 0.0004190206527709961, "grad_norm": 0.027675319463014603, "global_step": 6986, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04359006881713867, "train_loss_bc": 0.04359006881713867, "train_loss_llm": 0.0, "grad_norm": 0.02993507869541645, "global_step": 6987, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04370880499482155, "train_loss_bc": 0.04370880499482155, "train_loss_llm": 0.0, "grad_norm": 0.04091210290789604, "global_step": 6988, "epoch": 194, "lr": 0.001} +{"train_loss": 0.03299539163708687, "train_loss_bc": 0.03298725187778473, "train_loss_llm": 0.0008139312267303467, "grad_norm": 0.017214400693774223, "global_step": 6989, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04210800305008888, "train_loss_bc": 0.04210800305008888, "train_loss_llm": 0.0, "grad_norm": 0.026601053774356842, "global_step": 6990, "epoch": 194, "lr": 0.001} +{"train_loss": 0.0373353585600853, "train_loss_bc": 0.0373353585600853, "train_loss_llm": 0.0, "grad_norm": 0.03527708724141121, "global_step": 6991, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04647767171263695, "train_loss_bc": 0.04647604376077652, "train_loss_llm": 0.0001627504825592041, "grad_norm": 0.04484320804476738, "global_step": 6992, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04976143315434456, "train_loss_bc": 0.049759212881326675, "train_loss_llm": 0.00022208690643310547, "grad_norm": 0.02608892321586609, "global_step": 6993, "epoch": 194, "lr": 0.001} +{"train_loss": 0.03913004323840141, "train_loss_bc": 0.03911319747567177, "train_loss_llm": 0.0016845762729644775, "grad_norm": 0.03588171675801277, "global_step": 6994, "epoch": 194, "lr": 0.001} +{"train_loss": 0.047098953276872635, "train_loss_bc": 0.047098953276872635, "train_loss_llm": 0.0, "grad_norm": 0.04274163767695427, "global_step": 6995, "epoch": 194, "lr": 0.001} +{"train_loss": 0.040392082184553146, "train_loss_bc": 0.040392082184553146, "train_loss_llm": 0.0, "grad_norm": 0.053147248923778534, "global_step": 6996, "epoch": 194, "lr": 0.001} +{"train_loss": 0.038655467331409454, "train_loss_bc": 0.038655467331409454, "train_loss_llm": 0.0, "grad_norm": 0.01861886866390705, "global_step": 6997, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04541690647602081, "train_loss_bc": 0.04541134834289551, "train_loss_llm": 0.0005557537078857422, "grad_norm": 0.02961818128824234, "global_step": 6998, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04275703430175781, "train_loss_bc": 0.04275703430175781, "train_loss_llm": 0.0, "grad_norm": 0.03981741517782211, "global_step": 6999, "epoch": 194, "lr": 0.001} +{"train_loss": 0.0390588603913784, "train_loss_bc": 0.0390588603913784, "train_loss_llm": 0.0, "grad_norm": 0.048255804926157, "global_step": 7000, "epoch": 194, "lr": 0.001} +{"train_loss": 0.03181488439440727, "train_loss_bc": 0.03181488439440727, "train_loss_llm": 0.0, "grad_norm": 0.015631053596735, "global_step": 7001, "epoch": 194, "lr": 0.001} +{"train_loss": 0.039292946457862854, "train_loss_bc": 0.039292946457862854, "train_loss_llm": 0.0, "grad_norm": 0.025052927434444427, "global_step": 7002, "epoch": 194, "lr": 0.001} +{"train_loss": 0.0338224396109581, "train_loss_bc": 0.0338224396109581, "train_loss_llm": 0.0, "grad_norm": 0.03082120604813099, "global_step": 7003, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04070590063929558, "train_loss_bc": 0.04070590063929558, "train_loss_llm": 0.0, "grad_norm": 0.041262514889240265, "global_step": 7004, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04096077010035515, "train_loss_bc": 0.04094262421131134, "train_loss_llm": 0.0018146038055419922, "grad_norm": 0.018229283392429352, "global_step": 7005, "epoch": 194, "lr": 0.001} +{"train_loss": 0.041363395750522614, "train_loss_bc": 0.041363395750522614, "train_loss_llm": 0.0, "grad_norm": 0.02597189135849476, "global_step": 7006, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04461381584405899, "train_loss_bc": 0.04461381584405899, "train_loss_llm": 0.0, "grad_norm": 0.031947601586580276, "global_step": 7007, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04582637548446655, "train_loss_bc": 0.0458206832408905, "train_loss_llm": 0.0005692839622497559, "grad_norm": 0.0379304476082325, "global_step": 7008, "epoch": 194, "lr": 0.001} +{"train_loss": 0.040034446865320206, "train_loss_bc": 0.0400281585752964, "train_loss_llm": 0.0006288886070251465, "grad_norm": 0.020119832828640938, "global_step": 7009, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04225246608257294, "train_loss_bc": 0.04225246608257294, "train_loss_llm": 0.0, "grad_norm": 0.03292865306138992, "global_step": 7010, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04306795075535774, "train_loss_bc": 0.04306795075535774, "train_loss_llm": 0.0, "grad_norm": 0.04347760230302811, "global_step": 7011, "epoch": 194, "lr": 0.001} +{"train_loss": 0.038067057728767395, "train_loss_bc": 0.038067057728767395, "train_loss_llm": 0.0, "grad_norm": 0.0509735569357872, "global_step": 7012, "epoch": 194, "lr": 0.001} +{"train_loss": 0.03532802313566208, "train_loss_bc": 0.03532802313566208, "train_loss_llm": 0.0, "grad_norm": 0.016583433374762535, "global_step": 7013, "epoch": 194, "lr": 0.001} +{"train_loss": 0.03737084940075874, "train_loss_bc": 0.03736748918890953, "train_loss_llm": 0.00033611059188842773, "grad_norm": 0.022524667903780937, "global_step": 7014, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04793737828731537, "train_loss_bc": 0.04793737828731537, "train_loss_llm": 0.0, "grad_norm": 0.03593086823821068, "global_step": 7015, "epoch": 194, "lr": 0.001} +{"train_loss": 0.03383206948637962, "train_loss_bc": 0.03383206948637962, "train_loss_llm": 0.0, "grad_norm": 0.044091977179050446, "global_step": 7016, "epoch": 194, "lr": 0.001} +{"train_loss": 0.03626016154885292, "train_loss_bc": 0.036259882152080536, "train_loss_llm": 2.8014183044433594e-05, "grad_norm": 0.017080985009670258, "global_step": 7017, "epoch": 194, "lr": 0.001} +{"train_loss": 0.034323036670684814, "train_loss_bc": 0.034323036670684814, "train_loss_llm": 0.0, "grad_norm": 0.028847597539424896, "global_step": 7018, "epoch": 194, "lr": 0.001} +{"train_loss": 0.04044158560120397, "train_loss_bc": 0.04432179406285286, "train_loss_llm": 0.0006597340106964111, "grad_norm": 0.03827611729502678, "global_step": 7019, "epoch": 194, "lr": 0.001, "val_loss": 0.056767016649246216} +{"train_loss": 0.03855518624186516, "train_loss_bc": 0.03855518624186516, "train_loss_llm": 0.0, "grad_norm": 0.04537173733115196, "global_step": 7020, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04425332322716713, "train_loss_bc": 0.044250331819057465, "train_loss_llm": 0.00029924511909484863, "grad_norm": 0.021884692832827568, "global_step": 7021, "epoch": 195, "lr": 0.001} +{"train_loss": 0.037524644285440445, "train_loss_bc": 0.03751896321773529, "train_loss_llm": 0.000568002462387085, "grad_norm": 0.028783421963453293, "global_step": 7022, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04289932921528816, "train_loss_bc": 0.04289932921528816, "train_loss_llm": 0.0, "grad_norm": 0.03868845850229263, "global_step": 7023, "epoch": 195, "lr": 0.001} +{"train_loss": 0.035495828837156296, "train_loss_bc": 0.03547694906592369, "train_loss_llm": 0.0018880069255828857, "grad_norm": 0.04740065336227417, "global_step": 7024, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04530195891857147, "train_loss_bc": 0.045285165309906006, "train_loss_llm": 0.0016793310642242432, "grad_norm": 0.017803339287638664, "global_step": 7025, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04089679941534996, "train_loss_bc": 0.04089679941534996, "train_loss_llm": 0.0, "grad_norm": 0.030713863670825958, "global_step": 7026, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04114744812250137, "train_loss_bc": 0.04114744812250137, "train_loss_llm": 0.0, "grad_norm": 0.04034087061882019, "global_step": 7027, "epoch": 195, "lr": 0.001} +{"train_loss": 0.037813328206539154, "train_loss_bc": 0.037799566984176636, "train_loss_llm": 0.0013760030269622803, "grad_norm": 0.04831705987453461, "global_step": 7028, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03865030035376549, "train_loss_bc": 0.03865030035376549, "train_loss_llm": 0.0, "grad_norm": 0.01933826319873333, "global_step": 7029, "epoch": 195, "lr": 0.001} +{"train_loss": 0.037968121469020844, "train_loss_bc": 0.03792824223637581, "train_loss_llm": 0.0039878785610198975, "grad_norm": 0.02902044914662838, "global_step": 7030, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04540511593222618, "train_loss_bc": 0.04540511593222618, "train_loss_llm": 0.0, "grad_norm": 0.04041428491473198, "global_step": 7031, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03987470641732216, "train_loss_bc": 0.03987470641732216, "train_loss_llm": 0.0, "grad_norm": 0.05011645331978798, "global_step": 7032, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03793909773230553, "train_loss_bc": 0.03793909773230553, "train_loss_llm": 0.0, "grad_norm": 0.015225021168589592, "global_step": 7033, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04090547934174538, "train_loss_bc": 0.04090547934174538, "train_loss_llm": 0.0, "grad_norm": 0.02500777691602707, "global_step": 7034, "epoch": 195, "lr": 0.001} +{"train_loss": 0.037797968834638596, "train_loss_bc": 0.03777478635311127, "train_loss_llm": 0.0023182332515716553, "grad_norm": 0.03394981473684311, "global_step": 7035, "epoch": 195, "lr": 0.001} +{"train_loss": 0.043046895414590836, "train_loss_bc": 0.04302557185292244, "train_loss_llm": 0.0021323561668395996, "grad_norm": 0.041398677974939346, "global_step": 7036, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04483920335769653, "train_loss_bc": 0.04483920335769653, "train_loss_llm": 0.0, "grad_norm": 0.020597200840711594, "global_step": 7037, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04033612459897995, "train_loss_bc": 0.04033612459897995, "train_loss_llm": 0.0, "grad_norm": 0.022903427481651306, "global_step": 7038, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04552245885133743, "train_loss_bc": 0.045509085059165955, "train_loss_llm": 0.0013372600078582764, "grad_norm": 0.03329477086663246, "global_step": 7039, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03781605884432793, "train_loss_bc": 0.03781605884432793, "train_loss_llm": 0.0, "grad_norm": 0.038592319935560226, "global_step": 7040, "epoch": 195, "lr": 0.001} +{"train_loss": 0.040020667016506195, "train_loss_bc": 0.04001429304480553, "train_loss_llm": 0.0006374716758728027, "grad_norm": 0.018879517912864685, "global_step": 7041, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03949812427163124, "train_loss_bc": 0.03949672728776932, "train_loss_llm": 0.0001398324966430664, "grad_norm": 0.02918378636240959, "global_step": 7042, "epoch": 195, "lr": 0.001} +{"train_loss": 0.042937666177749634, "train_loss_bc": 0.042930543422698975, "train_loss_llm": 0.0007123649120330811, "grad_norm": 0.03557221218943596, "global_step": 7043, "epoch": 195, "lr": 0.001} +{"train_loss": 0.0379231721162796, "train_loss_bc": 0.0379231721162796, "train_loss_llm": 0.0, "grad_norm": 0.04273424297571182, "global_step": 7044, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03901408240199089, "train_loss_bc": 0.03898875042796135, "train_loss_llm": 0.0025331079959869385, "grad_norm": 0.02081599459052086, "global_step": 7045, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03777465969324112, "train_loss_bc": 0.03777465969324112, "train_loss_llm": 0.0, "grad_norm": 0.03357299789786339, "global_step": 7046, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03244844451546669, "train_loss_bc": 0.03244844451546669, "train_loss_llm": 0.0, "grad_norm": 0.03693527355790138, "global_step": 7047, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03733794018626213, "train_loss_bc": 0.03733794018626213, "train_loss_llm": 0.0, "grad_norm": 0.04954928532242775, "global_step": 7048, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03932581841945648, "train_loss_bc": 0.03932581841945648, "train_loss_llm": 0.0, "grad_norm": 0.01871747151017189, "global_step": 7049, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04193791002035141, "train_loss_bc": 0.04193791002035141, "train_loss_llm": 0.0, "grad_norm": 0.02944278158247471, "global_step": 7050, "epoch": 195, "lr": 0.001} +{"train_loss": 0.043404802680015564, "train_loss_bc": 0.043404802680015564, "train_loss_llm": 0.0, "grad_norm": 0.038666535168886185, "global_step": 7051, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04276297613978386, "train_loss_bc": 0.04273032397031784, "train_loss_llm": 0.0032651424407958984, "grad_norm": 0.043472979217767715, "global_step": 7052, "epoch": 195, "lr": 0.001} +{"train_loss": 0.03808179870247841, "train_loss_bc": 0.03808179870247841, "train_loss_llm": 0.0, "grad_norm": 0.01603369228541851, "global_step": 7053, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04221858084201813, "train_loss_bc": 0.04221858084201813, "train_loss_llm": 0.0, "grad_norm": 0.024997888132929802, "global_step": 7054, "epoch": 195, "lr": 0.001} +{"train_loss": 0.04045430881281694, "train_loss_bc": 0.04767909646034241, "train_loss_llm": 0.0, "grad_norm": 0.03505534678697586, "global_step": 7055, "epoch": 195, "lr": 0.001, "train/cumulative_reward": 105.75317994548182, "train/mean_score": 15.910791077310702, "train/success_rate": 0.4, "test/cumulative_reward": 56.53814617716713, "test/mean_score": 9.78148102072187, "test/success_rate": 0.22, "val_loss": 0.05722943693399429, "train_action_mse_error": 0.04399535059928894} +{"train_loss": 0.04134134575724602, "train_loss_bc": 0.04134134575724602, "train_loss_llm": 0.0, "grad_norm": 0.044339824467897415, "global_step": 7056, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04249647632241249, "train_loss_bc": 0.04249647632241249, "train_loss_llm": 0.0, "grad_norm": 0.01579165831208229, "global_step": 7057, "epoch": 196, "lr": 0.001} +{"train_loss": 0.03617562726140022, "train_loss_bc": 0.03617130219936371, "train_loss_llm": 0.00043261051177978516, "grad_norm": 0.02128332294523716, "global_step": 7058, "epoch": 196, "lr": 0.001} +{"train_loss": 0.03786493092775345, "train_loss_bc": 0.03786493092775345, "train_loss_llm": 0.0, "grad_norm": 0.031229058280587196, "global_step": 7059, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04342390596866608, "train_loss_bc": 0.04342390596866608, "train_loss_llm": 0.0, "grad_norm": 0.040256693959236145, "global_step": 7060, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04664251208305359, "train_loss_bc": 0.04664251208305359, "train_loss_llm": 0.0, "grad_norm": 0.022121014073491096, "global_step": 7061, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04159417375922203, "train_loss_bc": 0.04159417375922203, "train_loss_llm": 0.0, "grad_norm": 0.03211390972137451, "global_step": 7062, "epoch": 196, "lr": 0.001} +{"train_loss": 0.032050278037786484, "train_loss_bc": 0.032050278037786484, "train_loss_llm": 0.0, "grad_norm": 0.03852568939328194, "global_step": 7063, "epoch": 196, "lr": 0.001} +{"train_loss": 0.03445970639586449, "train_loss_bc": 0.03445970639586449, "train_loss_llm": 0.0, "grad_norm": 0.04656355828046799, "global_step": 7064, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04160157963633537, "train_loss_bc": 0.04160157963633537, "train_loss_llm": 0.0, "grad_norm": 0.017514515668153763, "global_step": 7065, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04168145731091499, "train_loss_bc": 0.04168145731091499, "train_loss_llm": 0.0, "grad_norm": 0.027395738288760185, "global_step": 7066, "epoch": 196, "lr": 0.001} +{"train_loss": 0.03557174280285835, "train_loss_bc": 0.03557174280285835, "train_loss_llm": 0.0, "grad_norm": 0.038098469376564026, "global_step": 7067, "epoch": 196, "lr": 0.001} +{"train_loss": 0.040718309581279755, "train_loss_bc": 0.040718309581279755, "train_loss_llm": 0.0, "grad_norm": 0.04929149150848389, "global_step": 7068, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04213149845600128, "train_loss_bc": 0.04213149845600128, "train_loss_llm": 0.0, "grad_norm": 0.019005559384822845, "global_step": 7069, "epoch": 196, "lr": 0.001} +{"train_loss": 0.03687869757413864, "train_loss_bc": 0.036875735968351364, "train_loss_llm": 0.0002962052822113037, "grad_norm": 0.02833009697496891, "global_step": 7070, "epoch": 196, "lr": 0.001} +{"train_loss": 0.03875970095396042, "train_loss_bc": 0.03875662013888359, "train_loss_llm": 0.00030812621116638184, "grad_norm": 0.034462571144104004, "global_step": 7071, "epoch": 196, "lr": 0.001} +{"train_loss": 0.03994521498680115, "train_loss_bc": 0.03994521498680115, "train_loss_llm": 0.0, "grad_norm": 0.04023345187306404, "global_step": 7072, "epoch": 196, "lr": 0.001} +{"train_loss": 0.045181553810834885, "train_loss_bc": 0.04518147557973862, "train_loss_llm": 7.927417755126953e-06, "grad_norm": 0.017162146046757698, "global_step": 7073, "epoch": 196, "lr": 0.001} +{"train_loss": 0.037306323647499084, "train_loss_bc": 0.037306323647499084, "train_loss_llm": 0.0, "grad_norm": 0.024272363632917404, "global_step": 7074, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04010266810655594, "train_loss_bc": 0.040093645453453064, "train_loss_llm": 0.0009023845195770264, "grad_norm": 0.03454700857400894, "global_step": 7075, "epoch": 196, "lr": 0.001} +{"train_loss": 0.0453149639070034, "train_loss_bc": 0.0453149639070034, "train_loss_llm": 0.0, "grad_norm": 0.03640253469347954, "global_step": 7076, "epoch": 196, "lr": 0.001} +{"train_loss": 0.035228364169597626, "train_loss_bc": 0.035228364169597626, "train_loss_llm": 0.0, "grad_norm": 0.01942569762468338, "global_step": 7077, "epoch": 196, "lr": 0.001} +{"train_loss": 0.03636251762509346, "train_loss_bc": 0.03635310381650925, "train_loss_llm": 0.0009414255619049072, "grad_norm": 0.023218249902129173, "global_step": 7078, "epoch": 196, "lr": 0.001} +{"train_loss": 0.036555640399456024, "train_loss_bc": 0.03653673455119133, "train_loss_llm": 0.0018906891345977783, "grad_norm": 0.03217719867825508, "global_step": 7079, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04096466302871704, "train_loss_bc": 0.0409526526927948, "train_loss_llm": 0.001200854778289795, "grad_norm": 0.03802620247006416, "global_step": 7080, "epoch": 196, "lr": 0.001} +{"train_loss": 0.0425824411213398, "train_loss_bc": 0.0425824411213398, "train_loss_llm": 0.0, "grad_norm": 0.01681230217218399, "global_step": 7081, "epoch": 196, "lr": 0.001} +{"train_loss": 0.047943685203790665, "train_loss_bc": 0.047930989414453506, "train_loss_llm": 0.001269608736038208, "grad_norm": 0.022032521665096283, "global_step": 7082, "epoch": 196, "lr": 0.001} +{"train_loss": 0.042848508805036545, "train_loss_bc": 0.042830102145671844, "train_loss_llm": 0.0018407702445983887, "grad_norm": 0.032123032957315445, "global_step": 7083, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04105711355805397, "train_loss_bc": 0.04105711355805397, "train_loss_llm": 0.0, "grad_norm": 0.03724779188632965, "global_step": 7084, "epoch": 196, "lr": 0.001} +{"train_loss": 0.03957502171397209, "train_loss_bc": 0.039554230868816376, "train_loss_llm": 0.0020789802074432373, "grad_norm": 0.019032299518585205, "global_step": 7085, "epoch": 196, "lr": 0.001} +{"train_loss": 0.0443546287715435, "train_loss_bc": 0.04434706270694733, "train_loss_llm": 0.0007567703723907471, "grad_norm": 0.02867325395345688, "global_step": 7086, "epoch": 196, "lr": 0.001} +{"train_loss": 0.038633354008197784, "train_loss_bc": 0.038605574518442154, "train_loss_llm": 0.002777993679046631, "grad_norm": 0.031131211668252945, "global_step": 7087, "epoch": 196, "lr": 0.001} +{"train_loss": 0.03373606875538826, "train_loss_bc": 0.03372015804052353, "train_loss_llm": 0.0015912353992462158, "grad_norm": 0.03870353847742081, "global_step": 7088, "epoch": 196, "lr": 0.001} +{"train_loss": 0.042257972061634064, "train_loss_bc": 0.042257972061634064, "train_loss_llm": 0.0, "grad_norm": 0.017132915556430817, "global_step": 7089, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04870810732245445, "train_loss_bc": 0.04870810732245445, "train_loss_llm": 0.0, "grad_norm": 0.034050505608320236, "global_step": 7090, "epoch": 196, "lr": 0.001} +{"train_loss": 0.04029796510520908, "train_loss_bc": 0.03867598995566368, "train_loss_llm": 0.0, "grad_norm": 0.04553045332431793, "global_step": 7091, "epoch": 196, "lr": 0.001, "val_loss": 0.05652988702058792} +{"train_loss": 0.04358161985874176, "train_loss_bc": 0.04354570806026459, "train_loss_llm": 0.0035912394523620605, "grad_norm": 0.04619782790541649, "global_step": 7092, "epoch": 197, "lr": 0.001} +{"train_loss": 0.03294840082526207, "train_loss_bc": 0.03294840082526207, "train_loss_llm": 0.0, "grad_norm": 0.012246066704392433, "global_step": 7093, "epoch": 197, "lr": 0.001} +{"train_loss": 0.03728942945599556, "train_loss_bc": 0.03728942945599556, "train_loss_llm": 0.0, "grad_norm": 0.025296062231063843, "global_step": 7094, "epoch": 197, "lr": 0.001} +{"train_loss": 0.03832460567355156, "train_loss_bc": 0.03832460567355156, "train_loss_llm": 0.0, "grad_norm": 0.0307548139244318, "global_step": 7095, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04255971685051918, "train_loss_bc": 0.04254435375332832, "train_loss_llm": 0.0015364289283752441, "grad_norm": 0.034370094537734985, "global_step": 7096, "epoch": 197, "lr": 0.001} +{"train_loss": 0.036603715270757675, "train_loss_bc": 0.036603715270757675, "train_loss_llm": 0.0, "grad_norm": 0.018006669357419014, "global_step": 7097, "epoch": 197, "lr": 0.001} +{"train_loss": 0.03865471109747887, "train_loss_bc": 0.03864545747637749, "train_loss_llm": 0.0009254515171051025, "grad_norm": 0.023750977590680122, "global_step": 7098, "epoch": 197, "lr": 0.001} +{"train_loss": 0.038334522396326065, "train_loss_bc": 0.038334522396326065, "train_loss_llm": 0.0, "grad_norm": 0.02652977965772152, "global_step": 7099, "epoch": 197, "lr": 0.001} +{"train_loss": 0.03800075501203537, "train_loss_bc": 0.03800075501203537, "train_loss_llm": 0.0, "grad_norm": 0.03478604555130005, "global_step": 7100, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04566611722111702, "train_loss_bc": 0.04566611722111702, "train_loss_llm": 0.0, "grad_norm": 0.0179729163646698, "global_step": 7101, "epoch": 197, "lr": 0.001} +{"train_loss": 0.037641193717718124, "train_loss_bc": 0.037641193717718124, "train_loss_llm": 0.0, "grad_norm": 0.02223648689687252, "global_step": 7102, "epoch": 197, "lr": 0.001} +{"train_loss": 0.03679247200489044, "train_loss_bc": 0.03679247200489044, "train_loss_llm": 0.0, "grad_norm": 0.03349614143371582, "global_step": 7103, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04426545277237892, "train_loss_bc": 0.04426545277237892, "train_loss_llm": 0.0, "grad_norm": 0.045384008437395096, "global_step": 7104, "epoch": 197, "lr": 0.001} +{"train_loss": 0.040897417813539505, "train_loss_bc": 0.040897417813539505, "train_loss_llm": 0.0, "grad_norm": 0.01745785027742386, "global_step": 7105, "epoch": 197, "lr": 0.001} +{"train_loss": 0.037500716745853424, "train_loss_bc": 0.037500716745853424, "train_loss_llm": 0.0, "grad_norm": 0.03376243636012077, "global_step": 7106, "epoch": 197, "lr": 0.001} +{"train_loss": 0.043843433260917664, "train_loss_bc": 0.043843433260917664, "train_loss_llm": 0.0, "grad_norm": 0.046860042959451675, "global_step": 7107, "epoch": 197, "lr": 0.001} +{"train_loss": 0.0433436818420887, "train_loss_bc": 0.0433436818420887, "train_loss_llm": 0.0, "grad_norm": 0.05948125943541527, "global_step": 7108, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04197796434164047, "train_loss_bc": 0.04197796434164047, "train_loss_llm": 0.0, "grad_norm": 0.02442142181098461, "global_step": 7109, "epoch": 197, "lr": 0.001} +{"train_loss": 0.033570148050785065, "train_loss_bc": 0.033570148050785065, "train_loss_llm": 0.0, "grad_norm": 0.034013040363788605, "global_step": 7110, "epoch": 197, "lr": 0.001} +{"train_loss": 0.03600035235285759, "train_loss_bc": 0.03600035235285759, "train_loss_llm": 0.0, "grad_norm": 0.044531259685754776, "global_step": 7111, "epoch": 197, "lr": 0.001} +{"train_loss": 0.03386867046356201, "train_loss_bc": 0.03386867046356201, "train_loss_llm": 0.0, "grad_norm": 0.05380069091916084, "global_step": 7112, "epoch": 197, "lr": 0.001} +{"train_loss": 0.039893534034490585, "train_loss_bc": 0.039893534034490585, "train_loss_llm": 0.0, "grad_norm": 0.015484276227653027, "global_step": 7113, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04783450439572334, "train_loss_bc": 0.04783450439572334, "train_loss_llm": 0.0, "grad_norm": 0.02380635403096676, "global_step": 7114, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04178938269615173, "train_loss_bc": 0.04178938269615173, "train_loss_llm": 0.0, "grad_norm": 0.02823013998568058, "global_step": 7115, "epoch": 197, "lr": 0.001} +{"train_loss": 0.037208326160907745, "train_loss_bc": 0.037186361849308014, "train_loss_llm": 0.0021962523460388184, "grad_norm": 0.03317839279770851, "global_step": 7116, "epoch": 197, "lr": 0.001} +{"train_loss": 0.043210987001657486, "train_loss_bc": 0.043210987001657486, "train_loss_llm": 0.0, "grad_norm": 0.018559399992227554, "global_step": 7117, "epoch": 197, "lr": 0.001} +{"train_loss": 0.029191425070166588, "train_loss_bc": 0.02918807603418827, "train_loss_llm": 0.0003348886966705322, "grad_norm": 0.028808100149035454, "global_step": 7118, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04026968032121658, "train_loss_bc": 0.04026716202497482, "train_loss_llm": 0.0002518296241760254, "grad_norm": 0.03482635319232941, "global_step": 7119, "epoch": 197, "lr": 0.001} +{"train_loss": 0.048155952244997025, "train_loss_bc": 0.048155952244997025, "train_loss_llm": 0.0, "grad_norm": 0.051791563630104065, "global_step": 7120, "epoch": 197, "lr": 0.001} +{"train_loss": 0.03665516525506973, "train_loss_bc": 0.03665516525506973, "train_loss_llm": 0.0, "grad_norm": 0.020719405263662338, "global_step": 7121, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04997364804148674, "train_loss_bc": 0.049973610788583755, "train_loss_llm": 3.6656856536865234e-06, "grad_norm": 0.03353375941514969, "global_step": 7122, "epoch": 197, "lr": 0.001} +{"train_loss": 0.0391920730471611, "train_loss_bc": 0.0391920730471611, "train_loss_llm": 0.0, "grad_norm": 0.035164978355169296, "global_step": 7123, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04584156721830368, "train_loss_bc": 0.045836031436920166, "train_loss_llm": 0.0005533993244171143, "grad_norm": 0.04090528562664986, "global_step": 7124, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04606618359684944, "train_loss_bc": 0.04606618359684944, "train_loss_llm": 0.0, "grad_norm": 0.0309041328728199, "global_step": 7125, "epoch": 197, "lr": 0.001} +{"train_loss": 0.0388919897377491, "train_loss_bc": 0.03888153284788132, "train_loss_llm": 0.001045525074005127, "grad_norm": 0.04229645058512688, "global_step": 7126, "epoch": 197, "lr": 0.001} +{"train_loss": 0.04039104308726059, "train_loss_bc": 0.048238035291433334, "train_loss_llm": 0.0, "grad_norm": 0.05435442551970482, "global_step": 7127, "epoch": 197, "lr": 0.001, "val_loss": 0.05704252049326897} +{"train_loss": 0.04130340740084648, "train_loss_bc": 0.04129435494542122, "train_loss_llm": 0.0009053349494934082, "grad_norm": 0.06879955530166626, "global_step": 7128, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04626563563942909, "train_loss_bc": 0.04626563563942909, "train_loss_llm": 0.0, "grad_norm": 0.03126941993832588, "global_step": 7129, "epoch": 198, "lr": 0.001} +{"train_loss": 0.042200859636068344, "train_loss_bc": 0.04219348728656769, "train_loss_llm": 0.0007373988628387451, "grad_norm": 0.04414007440209389, "global_step": 7130, "epoch": 198, "lr": 0.001} +{"train_loss": 0.03636167570948601, "train_loss_bc": 0.036346156150102615, "train_loss_llm": 0.0015521049499511719, "grad_norm": 0.05988990515470505, "global_step": 7131, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04453834891319275, "train_loss_bc": 0.04453834891319275, "train_loss_llm": 0.0, "grad_norm": 0.07015945762395859, "global_step": 7132, "epoch": 198, "lr": 0.001} +{"train_loss": 0.03868747502565384, "train_loss_bc": 0.03868042305111885, "train_loss_llm": 0.0007051229476928711, "grad_norm": 0.019325239583849907, "global_step": 7133, "epoch": 198, "lr": 0.001} +{"train_loss": 0.03641849756240845, "train_loss_bc": 0.036415569484233856, "train_loss_llm": 0.00029268860816955566, "grad_norm": 0.02471361868083477, "global_step": 7134, "epoch": 198, "lr": 0.001} +{"train_loss": 0.039489589631557465, "train_loss_bc": 0.03946063667535782, "train_loss_llm": 0.002895176410675049, "grad_norm": 0.03139835223555565, "global_step": 7135, "epoch": 198, "lr": 0.001} +{"train_loss": 0.038684651255607605, "train_loss_bc": 0.038684651255607605, "train_loss_llm": 0.0, "grad_norm": 0.04357873648405075, "global_step": 7136, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04176530987024307, "train_loss_bc": 0.04176530987024307, "train_loss_llm": 0.0, "grad_norm": 0.016294850036501884, "global_step": 7137, "epoch": 198, "lr": 0.001} +{"train_loss": 0.042640481144189835, "train_loss_bc": 0.04261904954910278, "train_loss_llm": 0.0021431148052215576, "grad_norm": 0.021675990894436836, "global_step": 7138, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04411472752690315, "train_loss_bc": 0.04411472752690315, "train_loss_llm": 0.0, "grad_norm": 0.02959195338189602, "global_step": 7139, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04121674224734306, "train_loss_bc": 0.041189175099134445, "train_loss_llm": 0.0027566850185394287, "grad_norm": 0.035157691687345505, "global_step": 7140, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04516330733895302, "train_loss_bc": 0.04516330733895302, "train_loss_llm": 0.0, "grad_norm": 0.021481744945049286, "global_step": 7141, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04910644143819809, "train_loss_bc": 0.049080707132816315, "train_loss_llm": 0.0025735199451446533, "grad_norm": 0.035505179315805435, "global_step": 7142, "epoch": 198, "lr": 0.001} +{"train_loss": 0.038847099989652634, "train_loss_bc": 0.038847099989652634, "train_loss_llm": 0.0, "grad_norm": 0.056832607835531235, "global_step": 7143, "epoch": 198, "lr": 0.001} +{"train_loss": 0.041640833020210266, "train_loss_bc": 0.041640833020210266, "train_loss_llm": 0.0, "grad_norm": 0.06504946947097778, "global_step": 7144, "epoch": 198, "lr": 0.001} +{"train_loss": 0.042953941971063614, "train_loss_bc": 0.04294035583734512, "train_loss_llm": 0.001358628273010254, "grad_norm": 0.0156211843714118, "global_step": 7145, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04481806978583336, "train_loss_bc": 0.04481806978583336, "train_loss_llm": 0.0, "grad_norm": 0.03705829009413719, "global_step": 7146, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04579845815896988, "train_loss_bc": 0.04579845815896988, "train_loss_llm": 0.0, "grad_norm": 0.05610792338848114, "global_step": 7147, "epoch": 198, "lr": 0.001} +{"train_loss": 0.045507822185754776, "train_loss_bc": 0.045507822185754776, "train_loss_llm": 0.0, "grad_norm": 0.07198753207921982, "global_step": 7148, "epoch": 198, "lr": 0.001} +{"train_loss": 0.038919560611248016, "train_loss_bc": 0.038919560611248016, "train_loss_llm": 0.0, "grad_norm": 0.016952354460954666, "global_step": 7149, "epoch": 198, "lr": 0.001} +{"train_loss": 0.03370824083685875, "train_loss_bc": 0.03370824083685875, "train_loss_llm": 0.0, "grad_norm": 0.024207495152950287, "global_step": 7150, "epoch": 198, "lr": 0.001} +{"train_loss": 0.03525587171316147, "train_loss_bc": 0.035255756229162216, "train_loss_llm": 1.1473894119262695e-05, "grad_norm": 0.033487170934677124, "global_step": 7151, "epoch": 198, "lr": 0.001} +{"train_loss": 0.03573790192604065, "train_loss_bc": 0.03572995588183403, "train_loss_llm": 0.0007945597171783447, "grad_norm": 0.04254622757434845, "global_step": 7152, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04030429944396019, "train_loss_bc": 0.04030429944396019, "train_loss_llm": 0.0, "grad_norm": 0.0213382076472044, "global_step": 7153, "epoch": 198, "lr": 0.001} +{"train_loss": 0.03693663701415062, "train_loss_bc": 0.03693663701415062, "train_loss_llm": 0.0, "grad_norm": 0.02492489479482174, "global_step": 7154, "epoch": 198, "lr": 0.001} +{"train_loss": 0.0374099537730217, "train_loss_bc": 0.0374099537730217, "train_loss_llm": 0.0, "grad_norm": 0.03455699607729912, "global_step": 7155, "epoch": 198, "lr": 0.001} +{"train_loss": 0.03655446693301201, "train_loss_bc": 0.03655446693301201, "train_loss_llm": 0.0, "grad_norm": 0.038584910333156586, "global_step": 7156, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04045035317540169, "train_loss_bc": 0.04045035317540169, "train_loss_llm": 0.0, "grad_norm": 0.018021564930677414, "global_step": 7157, "epoch": 198, "lr": 0.001} +{"train_loss": 0.036046333611011505, "train_loss_bc": 0.036046333611011505, "train_loss_llm": 0.0, "grad_norm": 0.03083670325577259, "global_step": 7158, "epoch": 198, "lr": 0.001} +{"train_loss": 0.03712012991309166, "train_loss_bc": 0.03712012991309166, "train_loss_llm": 0.0, "grad_norm": 0.0430852510035038, "global_step": 7159, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04427914321422577, "train_loss_bc": 0.04426674172282219, "train_loss_llm": 0.0012401044368743896, "grad_norm": 0.04687163606286049, "global_step": 7160, "epoch": 198, "lr": 0.001} +{"train_loss": 0.041101664304733276, "train_loss_bc": 0.041101664304733276, "train_loss_llm": 0.0, "grad_norm": 0.019868971779942513, "global_step": 7161, "epoch": 198, "lr": 0.001} +{"train_loss": 0.03350794315338135, "train_loss_bc": 0.03350794315338135, "train_loss_llm": 0.0, "grad_norm": 0.033010125160217285, "global_step": 7162, "epoch": 198, "lr": 0.001} +{"train_loss": 0.04044730982018842, "train_loss_bc": 0.04123671352863312, "train_loss_llm": 0.001056671142578125, "grad_norm": 0.056501731276512146, "global_step": 7163, "epoch": 198, "lr": 0.001, "val_loss": 0.05762762576341629} +{"train_loss": 0.04532965272665024, "train_loss_bc": 0.04532965272665024, "train_loss_llm": 0.0, "grad_norm": 0.07224918156862259, "global_step": 7164, "epoch": 199, "lr": 0.001} +{"train_loss": 0.038163259625434875, "train_loss_bc": 0.038163259625434875, "train_loss_llm": 0.0, "grad_norm": 0.022837338969111443, "global_step": 7165, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04413231462240219, "train_loss_bc": 0.04413231462240219, "train_loss_llm": 0.0, "grad_norm": 0.042196765542030334, "global_step": 7166, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03205999359488487, "train_loss_bc": 0.03205999359488487, "train_loss_llm": 0.0, "grad_norm": 0.04547756165266037, "global_step": 7167, "epoch": 199, "lr": 0.001} +{"train_loss": 0.042807113379240036, "train_loss_bc": 0.042807113379240036, "train_loss_llm": 0.0, "grad_norm": 0.055730368942022324, "global_step": 7168, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03470412641763687, "train_loss_bc": 0.03467265143990517, "train_loss_llm": 0.003147602081298828, "grad_norm": 0.019610190764069557, "global_step": 7169, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03848469629883766, "train_loss_bc": 0.03848469629883766, "train_loss_llm": 0.0, "grad_norm": 0.038536056876182556, "global_step": 7170, "epoch": 199, "lr": 0.001} +{"train_loss": 0.038236912339925766, "train_loss_bc": 0.038236912339925766, "train_loss_llm": 0.0, "grad_norm": 0.048870038241147995, "global_step": 7171, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04036511480808258, "train_loss_bc": 0.04034441336989403, "train_loss_llm": 0.002070128917694092, "grad_norm": 0.05966677889227867, "global_step": 7172, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03691514953970909, "train_loss_bc": 0.036904845386743546, "train_loss_llm": 0.0010304152965545654, "grad_norm": 0.01650381274521351, "global_step": 7173, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03844592720270157, "train_loss_bc": 0.03844592720270157, "train_loss_llm": 0.0, "grad_norm": 0.024334128946065903, "global_step": 7174, "epoch": 199, "lr": 0.001} +{"train_loss": 0.0407789871096611, "train_loss_bc": 0.0407789871096611, "train_loss_llm": 0.0, "grad_norm": 0.03312760964035988, "global_step": 7175, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04162466898560524, "train_loss_bc": 0.04162466898560524, "train_loss_llm": 0.0, "grad_norm": 0.04065579175949097, "global_step": 7176, "epoch": 199, "lr": 0.001} +{"train_loss": 0.0381137989461422, "train_loss_bc": 0.0381137989461422, "train_loss_llm": 0.0, "grad_norm": 0.01979796588420868, "global_step": 7177, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03921058401465416, "train_loss_bc": 0.039178237318992615, "train_loss_llm": 0.003234684467315674, "grad_norm": 0.03336184471845627, "global_step": 7178, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04195162281394005, "train_loss_bc": 0.04195162281394005, "train_loss_llm": 0.0, "grad_norm": 0.041788499802351, "global_step": 7179, "epoch": 199, "lr": 0.001} +{"train_loss": 0.044526707381010056, "train_loss_bc": 0.044526707381010056, "train_loss_llm": 0.0, "grad_norm": 0.05002077668905258, "global_step": 7180, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04648508131504059, "train_loss_bc": 0.046475816518068314, "train_loss_llm": 0.0009264051914215088, "grad_norm": 0.026195669546723366, "global_step": 7181, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04405124485492706, "train_loss_bc": 0.04405124485492706, "train_loss_llm": 0.0, "grad_norm": 0.044029220938682556, "global_step": 7182, "epoch": 199, "lr": 0.001} +{"train_loss": 0.041095469146966934, "train_loss_bc": 0.041095469146966934, "train_loss_llm": 0.0, "grad_norm": 0.05712651461362839, "global_step": 7183, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03985197842121124, "train_loss_bc": 0.03981325030326843, "train_loss_llm": 0.003872990608215332, "grad_norm": 0.06960316002368927, "global_step": 7184, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03952714428305626, "train_loss_bc": 0.03951047733426094, "train_loss_llm": 0.0016666650772094727, "grad_norm": 0.019049441441893578, "global_step": 7185, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04538936913013458, "train_loss_bc": 0.04538936913013458, "train_loss_llm": 0.0, "grad_norm": 0.02786254696547985, "global_step": 7186, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03208567202091217, "train_loss_bc": 0.03208567202091217, "train_loss_llm": 0.0, "grad_norm": 0.03532201424241066, "global_step": 7187, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04258567839860916, "train_loss_bc": 0.04258567839860916, "train_loss_llm": 0.0, "grad_norm": 0.04827041178941727, "global_step": 7188, "epoch": 199, "lr": 0.001} +{"train_loss": 0.048887766897678375, "train_loss_bc": 0.048887766897678375, "train_loss_llm": 0.0, "grad_norm": 0.02517676167190075, "global_step": 7189, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04289933294057846, "train_loss_bc": 0.04289933294057846, "train_loss_llm": 0.0, "grad_norm": 0.03583526983857155, "global_step": 7190, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04756264016032219, "train_loss_bc": 0.04756264016032219, "train_loss_llm": 0.0, "grad_norm": 0.048426553606987, "global_step": 7191, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03899158909916878, "train_loss_bc": 0.03899158909916878, "train_loss_llm": 0.0, "grad_norm": 0.05948896333575249, "global_step": 7192, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04165657237172127, "train_loss_bc": 0.04165657237172127, "train_loss_llm": 0.0, "grad_norm": 0.02298053354024887, "global_step": 7193, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03984466940164566, "train_loss_bc": 0.03982557728886604, "train_loss_llm": 0.0019093751907348633, "grad_norm": 0.03366468846797943, "global_step": 7194, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03859219700098038, "train_loss_bc": 0.03859219700098038, "train_loss_llm": 0.0, "grad_norm": 0.038504667580127716, "global_step": 7195, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03968549519777298, "train_loss_bc": 0.039682600647211075, "train_loss_llm": 0.0002893507480621338, "grad_norm": 0.05067979171872139, "global_step": 7196, "epoch": 199, "lr": 0.001} +{"train_loss": 0.040095143020153046, "train_loss_bc": 0.040095143020153046, "train_loss_llm": 0.0, "grad_norm": 0.02023707702755928, "global_step": 7197, "epoch": 199, "lr": 0.001} +{"train_loss": 0.03662461042404175, "train_loss_bc": 0.03662431985139847, "train_loss_llm": 2.917647361755371e-05, "grad_norm": 0.03130517899990082, "global_step": 7198, "epoch": 199, "lr": 0.001} +{"train_loss": 0.04045427042163081, "train_loss_bc": 0.03459145128726959, "train_loss_llm": 0.0, "grad_norm": 0.044230867177248, "global_step": 7199, "epoch": 199, "lr": 0.001, "val_loss": 0.05677307769656181} +{"train_loss": 0.03998261317610741, "train_loss_bc": 0.03998261317610741, "train_loss_llm": 0.0, "grad_norm": 0.0556609183549881, "global_step": 7200, "epoch": 200, "lr": 0.001} +{"train_loss": 0.034400906413793564, "train_loss_bc": 0.034400906413793564, "train_loss_llm": 0.0, "grad_norm": 0.021960338577628136, "global_step": 7201, "epoch": 200, "lr": 0.001} +{"train_loss": 0.03385328873991966, "train_loss_bc": 0.03385328873991966, "train_loss_llm": 0.0, "grad_norm": 0.02636641263961792, "global_step": 7202, "epoch": 200, "lr": 0.001} +{"train_loss": 0.0450889989733696, "train_loss_bc": 0.0450889989733696, "train_loss_llm": 0.0, "grad_norm": 0.03542070463299751, "global_step": 7203, "epoch": 200, "lr": 0.001} +{"train_loss": 0.040792278945446014, "train_loss_bc": 0.04078848659992218, "train_loss_llm": 0.00037923455238342285, "grad_norm": 0.04454713687300682, "global_step": 7204, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04226143658161163, "train_loss_bc": 0.04225804656744003, "train_loss_llm": 0.000338822603225708, "grad_norm": 0.02026999182999134, "global_step": 7205, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04145916551351547, "train_loss_bc": 0.04143298789858818, "train_loss_llm": 0.002617865800857544, "grad_norm": 0.03607342392206192, "global_step": 7206, "epoch": 200, "lr": 0.001} +{"train_loss": 0.03806759789586067, "train_loss_bc": 0.0380483902990818, "train_loss_llm": 0.0019205808639526367, "grad_norm": 0.05118710547685623, "global_step": 7207, "epoch": 200, "lr": 0.001} +{"train_loss": 0.036607492715120316, "train_loss_bc": 0.036607492715120316, "train_loss_llm": 0.0, "grad_norm": 0.06233706325292587, "global_step": 7208, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04429412633180618, "train_loss_bc": 0.0442863367497921, "train_loss_llm": 0.0007789433002471924, "grad_norm": 0.026091856881976128, "global_step": 7209, "epoch": 200, "lr": 0.001} +{"train_loss": 0.036188699305057526, "train_loss_bc": 0.03616288676857948, "train_loss_llm": 0.002581268548965454, "grad_norm": 0.03488614782691002, "global_step": 7210, "epoch": 200, "lr": 0.001} +{"train_loss": 0.037170786410570145, "train_loss_bc": 0.037170786410570145, "train_loss_llm": 0.0, "grad_norm": 0.04387897253036499, "global_step": 7211, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04149837791919708, "train_loss_bc": 0.041492462158203125, "train_loss_llm": 0.0005916953086853027, "grad_norm": 0.059004660695791245, "global_step": 7212, "epoch": 200, "lr": 0.001} +{"train_loss": 0.036446087062358856, "train_loss_bc": 0.036446087062358856, "train_loss_llm": 0.0, "grad_norm": 0.016880609095096588, "global_step": 7213, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04367358982563019, "train_loss_bc": 0.04366564005613327, "train_loss_llm": 0.0007949769496917725, "grad_norm": 0.03005360998213291, "global_step": 7214, "epoch": 200, "lr": 0.001} +{"train_loss": 0.03537715598940849, "train_loss_bc": 0.03537715598940849, "train_loss_llm": 0.0, "grad_norm": 0.03628641366958618, "global_step": 7215, "epoch": 200, "lr": 0.001} +{"train_loss": 0.03253469988703728, "train_loss_bc": 0.03253469988703728, "train_loss_llm": 0.0, "grad_norm": 0.04620593786239624, "global_step": 7216, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04015989974141121, "train_loss_bc": 0.04015989974141121, "train_loss_llm": 0.0, "grad_norm": 0.01622970774769783, "global_step": 7217, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04019683226943016, "train_loss_bc": 0.04019683226943016, "train_loss_llm": 0.0, "grad_norm": 0.02803700603544712, "global_step": 7218, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04548601433634758, "train_loss_bc": 0.04548601433634758, "train_loss_llm": 0.0, "grad_norm": 0.04024747759103775, "global_step": 7219, "epoch": 200, "lr": 0.001} +{"train_loss": 0.042155005037784576, "train_loss_bc": 0.042155005037784576, "train_loss_llm": 0.0, "grad_norm": 0.05834280326962471, "global_step": 7220, "epoch": 200, "lr": 0.001} +{"train_loss": 0.041383370757102966, "train_loss_bc": 0.041383370757102966, "train_loss_llm": 0.0, "grad_norm": 0.018460990861058235, "global_step": 7221, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04001295939087868, "train_loss_bc": 0.04000368341803551, "train_loss_llm": 0.000927656888961792, "grad_norm": 0.027955126017332077, "global_step": 7222, "epoch": 200, "lr": 0.001} +{"train_loss": 0.03899550065398216, "train_loss_bc": 0.03899550065398216, "train_loss_llm": 0.0, "grad_norm": 0.042025599628686905, "global_step": 7223, "epoch": 200, "lr": 0.001} +{"train_loss": 0.05037161707878113, "train_loss_bc": 0.05037161707878113, "train_loss_llm": 0.0, "grad_norm": 0.05184342339634895, "global_step": 7224, "epoch": 200, "lr": 0.001} +{"train_loss": 0.042690347880125046, "train_loss_bc": 0.04268748313188553, "train_loss_llm": 0.00028645992279052734, "grad_norm": 0.02643507346510887, "global_step": 7225, "epoch": 200, "lr": 0.001} +{"train_loss": 0.042219314724206924, "train_loss_bc": 0.042219314724206924, "train_loss_llm": 0.0, "grad_norm": 0.03892546519637108, "global_step": 7226, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04350518062710762, "train_loss_bc": 0.04350518062710762, "train_loss_llm": 0.0, "grad_norm": 0.04746481776237488, "global_step": 7227, "epoch": 200, "lr": 0.001} +{"train_loss": 0.03425911068916321, "train_loss_bc": 0.034254126250743866, "train_loss_llm": 0.0004984736442565918, "grad_norm": 0.054076336324214935, "global_step": 7228, "epoch": 200, "lr": 0.001} +{"train_loss": 0.039633527398109436, "train_loss_bc": 0.039625320583581924, "train_loss_llm": 0.0008208155632019043, "grad_norm": 0.018556643277406693, "global_step": 7229, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04229239374399185, "train_loss_bc": 0.04228825122117996, "train_loss_llm": 0.0004143714904785156, "grad_norm": 0.026580508798360825, "global_step": 7230, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04156950116157532, "train_loss_bc": 0.04156950116157532, "train_loss_llm": 0.0, "grad_norm": 0.031549714505672455, "global_step": 7231, "epoch": 200, "lr": 0.001} +{"train_loss": 0.0436931848526001, "train_loss_bc": 0.0436931848526001, "train_loss_llm": 0.0, "grad_norm": 0.03886936977505684, "global_step": 7232, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04407590255141258, "train_loss_bc": 0.044070079922676086, "train_loss_llm": 0.0005822181701660156, "grad_norm": 0.02426993101835251, "global_step": 7233, "epoch": 200, "lr": 0.001} +{"train_loss": 0.03884735703468323, "train_loss_bc": 0.03884353116154671, "train_loss_llm": 0.0003826618194580078, "grad_norm": 0.034878961741924286, "global_step": 7234, "epoch": 200, "lr": 0.001} +{"train_loss": 0.04050714522600174, "train_loss_bc": 0.04698367044329643, "train_loss_llm": 0.002923518419265747, "grad_norm": 0.04369828850030899, "global_step": 7235, "epoch": 200, "lr": 0.001, "train/cumulative_reward": 96.30496534850451, "train/mean_score": 18.827955019464127, "train/success_rate": 0.5, "test/cumulative_reward": 81.01871743002927, "test/mean_score": 9.575314594890084, "test/success_rate": 0.2, "val_loss": 0.05625741183757782, "train_action_mse_error": 0.04420056939125061} +{"train_loss": 0.043207984417676926, "train_loss_bc": 0.043207984417676926, "train_loss_llm": 0.0, "grad_norm": 0.05392531305551529, "global_step": 7236, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04144178330898285, "train_loss_bc": 0.04144178330898285, "train_loss_llm": 0.0, "grad_norm": 0.020617404952645302, "global_step": 7237, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04261413961648941, "train_loss_bc": 0.042593806982040405, "train_loss_llm": 0.002033144235610962, "grad_norm": 0.03445876017212868, "global_step": 7238, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04637664556503296, "train_loss_bc": 0.04637664556503296, "train_loss_llm": 0.0, "grad_norm": 0.04685945808887482, "global_step": 7239, "epoch": 201, "lr": 0.001} +{"train_loss": 0.049385279417037964, "train_loss_bc": 0.049385279417037964, "train_loss_llm": 0.0, "grad_norm": 0.049260661005973816, "global_step": 7240, "epoch": 201, "lr": 0.001} +{"train_loss": 0.048764169216156006, "train_loss_bc": 0.048764169216156006, "train_loss_llm": 0.0, "grad_norm": 0.018678506836295128, "global_step": 7241, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03801756352186203, "train_loss_bc": 0.03801581263542175, "train_loss_llm": 0.00017502903938293457, "grad_norm": 0.026138605549931526, "global_step": 7242, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04226897284388542, "train_loss_bc": 0.04226897284388542, "train_loss_llm": 0.0, "grad_norm": 0.03157095983624458, "global_step": 7243, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03337174281477928, "train_loss_bc": 0.03337174281477928, "train_loss_llm": 0.0, "grad_norm": 0.033169426023960114, "global_step": 7244, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03179098665714264, "train_loss_bc": 0.03179098665714264, "train_loss_llm": 0.0, "grad_norm": 0.017656002193689346, "global_step": 7245, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04075300693511963, "train_loss_bc": 0.040750447660684586, "train_loss_llm": 0.00025576353073120117, "grad_norm": 0.023441866040229797, "global_step": 7246, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03665696829557419, "train_loss_bc": 0.03661951422691345, "train_loss_llm": 0.0037452876567840576, "grad_norm": 0.036663979291915894, "global_step": 7247, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03137556463479996, "train_loss_bc": 0.03137233108282089, "train_loss_llm": 0.00032329559326171875, "grad_norm": 0.04443345218896866, "global_step": 7248, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04410681128501892, "train_loss_bc": 0.04410681128501892, "train_loss_llm": 0.0, "grad_norm": 0.017124619334936142, "global_step": 7249, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03922566771507263, "train_loss_bc": 0.03922566771507263, "train_loss_llm": 0.0, "grad_norm": 0.02382451854646206, "global_step": 7250, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03984007611870766, "train_loss_bc": 0.03984007611870766, "train_loss_llm": 0.0, "grad_norm": 0.03157179802656174, "global_step": 7251, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03596353903412819, "train_loss_bc": 0.03596353903412819, "train_loss_llm": 0.0, "grad_norm": 0.041063837707042694, "global_step": 7252, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04055273160338402, "train_loss_bc": 0.04052110016345978, "train_loss_llm": 0.003163278102874756, "grad_norm": 0.01789211481809616, "global_step": 7253, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03600480407476425, "train_loss_bc": 0.03600480407476425, "train_loss_llm": 0.0, "grad_norm": 0.024322403594851494, "global_step": 7254, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03996173292398453, "train_loss_bc": 0.03996173292398453, "train_loss_llm": 0.0, "grad_norm": 0.0356442853808403, "global_step": 7255, "epoch": 201, "lr": 0.001} +{"train_loss": 0.049454670399427414, "train_loss_bc": 0.04945313557982445, "train_loss_llm": 0.00015342235565185547, "grad_norm": 0.04978342726826668, "global_step": 7256, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04685026407241821, "train_loss_bc": 0.04685026407241821, "train_loss_llm": 0.0, "grad_norm": 0.015798626467585564, "global_step": 7257, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03178775683045387, "train_loss_bc": 0.03178775683045387, "train_loss_llm": 0.0, "grad_norm": 0.023121966049075127, "global_step": 7258, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03810586407780647, "train_loss_bc": 0.03810586407780647, "train_loss_llm": 0.0, "grad_norm": 0.028789201751351357, "global_step": 7259, "epoch": 201, "lr": 0.001} +{"train_loss": 0.037451229989528656, "train_loss_bc": 0.037451229989528656, "train_loss_llm": 0.0, "grad_norm": 0.03532271087169647, "global_step": 7260, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04912839084863663, "train_loss_bc": 0.04912839084863663, "train_loss_llm": 0.0, "grad_norm": 0.02587747946381569, "global_step": 7261, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04178398847579956, "train_loss_bc": 0.04178398847579956, "train_loss_llm": 0.0, "grad_norm": 0.03314763307571411, "global_step": 7262, "epoch": 201, "lr": 0.001} +{"train_loss": 0.0337640643119812, "train_loss_bc": 0.03376318886876106, "train_loss_llm": 8.749961853027344e-05, "grad_norm": 0.04143204912543297, "global_step": 7263, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03995796665549278, "train_loss_bc": 0.0399523489177227, "train_loss_llm": 0.0005616247653961182, "grad_norm": 0.052549418061971664, "global_step": 7264, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03481459617614746, "train_loss_bc": 0.03481459617614746, "train_loss_llm": 0.0, "grad_norm": 0.016668127849698067, "global_step": 7265, "epoch": 201, "lr": 0.001} +{"train_loss": 0.043953828513622284, "train_loss_bc": 0.04392983391880989, "train_loss_llm": 0.002399533987045288, "grad_norm": 0.02557525970041752, "global_step": 7266, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04398540034890175, "train_loss_bc": 0.04397955536842346, "train_loss_llm": 0.0005846321582794189, "grad_norm": 0.03590180352330208, "global_step": 7267, "epoch": 201, "lr": 0.001} +{"train_loss": 0.032549478113651276, "train_loss_bc": 0.03254574164748192, "train_loss_llm": 0.00037357211112976074, "grad_norm": 0.04349052906036377, "global_step": 7268, "epoch": 201, "lr": 0.001} +{"train_loss": 0.03884682059288025, "train_loss_bc": 0.03884682059288025, "train_loss_llm": 0.0, "grad_norm": 0.014808353036642075, "global_step": 7269, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04159407690167427, "train_loss_bc": 0.04159407690167427, "train_loss_llm": 0.0, "grad_norm": 0.023933766409754753, "global_step": 7270, "epoch": 201, "lr": 0.001} +{"train_loss": 0.04026136868115929, "train_loss_bc": 0.04370070621371269, "train_loss_llm": 0.0, "grad_norm": 0.03157579526305199, "global_step": 7271, "epoch": 201, "lr": 0.001, "val_loss": 0.05714799836277962} +{"train_loss": 0.04176042973995209, "train_loss_bc": 0.04174751788377762, "train_loss_llm": 0.0012912452220916748, "grad_norm": 0.03945545479655266, "global_step": 7272, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04877213388681412, "train_loss_bc": 0.04877138137817383, "train_loss_llm": 7.510185241699219e-05, "grad_norm": 0.020217718556523323, "global_step": 7273, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04099135473370552, "train_loss_bc": 0.04099135473370552, "train_loss_llm": 0.0, "grad_norm": 0.029259290546178818, "global_step": 7274, "epoch": 202, "lr": 0.001} +{"train_loss": 0.032351307570934296, "train_loss_bc": 0.03233548253774643, "train_loss_llm": 0.0015824437141418457, "grad_norm": 0.03785349801182747, "global_step": 7275, "epoch": 202, "lr": 0.001} +{"train_loss": 0.043757759034633636, "train_loss_bc": 0.043757759034633636, "train_loss_llm": 0.0, "grad_norm": 0.047264426946640015, "global_step": 7276, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04303186759352684, "train_loss_bc": 0.04302314668893814, "train_loss_llm": 0.000872194766998291, "grad_norm": 0.021636852994561195, "global_step": 7277, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04206167906522751, "train_loss_bc": 0.04206167906522751, "train_loss_llm": 0.0, "grad_norm": 0.04027580842375755, "global_step": 7278, "epoch": 202, "lr": 0.001} +{"train_loss": 0.03956810012459755, "train_loss_bc": 0.03956810012459755, "train_loss_llm": 0.0, "grad_norm": 0.05845058336853981, "global_step": 7279, "epoch": 202, "lr": 0.001} +{"train_loss": 0.03681052476167679, "train_loss_bc": 0.03681052476167679, "train_loss_llm": 0.0, "grad_norm": 0.06024685502052307, "global_step": 7280, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04329565167427063, "train_loss_bc": 0.04329565167427063, "train_loss_llm": 0.0, "grad_norm": 0.014424603432416916, "global_step": 7281, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04009770601987839, "train_loss_bc": 0.04009770601987839, "train_loss_llm": 0.0, "grad_norm": 0.024870259687304497, "global_step": 7282, "epoch": 202, "lr": 0.001} +{"train_loss": 0.046153921633958817, "train_loss_bc": 0.046153921633958817, "train_loss_llm": 0.0, "grad_norm": 0.0340573824942112, "global_step": 7283, "epoch": 202, "lr": 0.001} +{"train_loss": 0.0404663123190403, "train_loss_bc": 0.040459051728248596, "train_loss_llm": 0.0007259845733642578, "grad_norm": 0.03877774253487587, "global_step": 7284, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04260582476854324, "train_loss_bc": 0.04260582476854324, "train_loss_llm": 0.0, "grad_norm": 0.01686841994524002, "global_step": 7285, "epoch": 202, "lr": 0.001} +{"train_loss": 0.041670892387628555, "train_loss_bc": 0.041670892387628555, "train_loss_llm": 0.0, "grad_norm": 0.022618256509304047, "global_step": 7286, "epoch": 202, "lr": 0.001} +{"train_loss": 0.037673044949769974, "train_loss_bc": 0.037673044949769974, "train_loss_llm": 0.0, "grad_norm": 0.028599213808774948, "global_step": 7287, "epoch": 202, "lr": 0.001} +{"train_loss": 0.03904922679066658, "train_loss_bc": 0.039008207619190216, "train_loss_llm": 0.004101872444152832, "grad_norm": 0.03786960616707802, "global_step": 7288, "epoch": 202, "lr": 0.001} +{"train_loss": 0.03637465462088585, "train_loss_bc": 0.03637465462088585, "train_loss_llm": 0.0, "grad_norm": 0.018959086388349533, "global_step": 7289, "epoch": 202, "lr": 0.001} +{"train_loss": 0.03486935421824455, "train_loss_bc": 0.034856878221035004, "train_loss_llm": 0.0012475550174713135, "grad_norm": 0.025577668100595474, "global_step": 7290, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04037950932979584, "train_loss_bc": 0.04037950932979584, "train_loss_llm": 0.0, "grad_norm": 0.029231801629066467, "global_step": 7291, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04040093719959259, "train_loss_bc": 0.04040093719959259, "train_loss_llm": 0.0, "grad_norm": 0.034122344106435776, "global_step": 7292, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04368342459201813, "train_loss_bc": 0.04366130381822586, "train_loss_llm": 0.0022122561931610107, "grad_norm": 0.017963256686925888, "global_step": 7293, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04070854187011719, "train_loss_bc": 0.04070854187011719, "train_loss_llm": 0.0, "grad_norm": 0.030919702723622322, "global_step": 7294, "epoch": 202, "lr": 0.001} +{"train_loss": 0.03689555451273918, "train_loss_bc": 0.03689555451273918, "train_loss_llm": 0.0, "grad_norm": 0.04706863686442375, "global_step": 7295, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04051671549677849, "train_loss_bc": 0.04051671549677849, "train_loss_llm": 0.0, "grad_norm": 0.057550735771656036, "global_step": 7296, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04617053642868996, "train_loss_bc": 0.04616538807749748, "train_loss_llm": 0.0005147457122802734, "grad_norm": 0.021883253008127213, "global_step": 7297, "epoch": 202, "lr": 0.001} +{"train_loss": 0.03523949533700943, "train_loss_bc": 0.03523949533700943, "train_loss_llm": 0.0, "grad_norm": 0.031034765765070915, "global_step": 7298, "epoch": 202, "lr": 0.001} +{"train_loss": 0.045196667313575745, "train_loss_bc": 0.04519208148121834, "train_loss_llm": 0.00045859813690185547, "grad_norm": 0.04259398952126503, "global_step": 7299, "epoch": 202, "lr": 0.001} +{"train_loss": 0.043224234133958817, "train_loss_bc": 0.043224234133958817, "train_loss_llm": 0.0, "grad_norm": 0.05588095635175705, "global_step": 7300, "epoch": 202, "lr": 0.001} +{"train_loss": 0.035401441156864166, "train_loss_bc": 0.035401441156864166, "train_loss_llm": 0.0, "grad_norm": 0.019741596654057503, "global_step": 7301, "epoch": 202, "lr": 0.001} +{"train_loss": 0.035987451672554016, "train_loss_bc": 0.03598658740520477, "train_loss_llm": 8.624792098999023e-05, "grad_norm": 0.027426913380622864, "global_step": 7302, "epoch": 202, "lr": 0.001} +{"train_loss": 0.0417279489338398, "train_loss_bc": 0.0417279489338398, "train_loss_llm": 0.0, "grad_norm": 0.03288622573018074, "global_step": 7303, "epoch": 202, "lr": 0.001} +{"train_loss": 0.037667304277420044, "train_loss_bc": 0.03766358643770218, "train_loss_llm": 0.00037175416946411133, "grad_norm": 0.03998304158449173, "global_step": 7304, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04312130808830261, "train_loss_bc": 0.04312130808830261, "train_loss_llm": 0.0, "grad_norm": 0.01656954549252987, "global_step": 7305, "epoch": 202, "lr": 0.001} +{"train_loss": 0.03405630588531494, "train_loss_bc": 0.03405630588531494, "train_loss_llm": 0.0, "grad_norm": 0.023400796577334404, "global_step": 7306, "epoch": 202, "lr": 0.001} +{"train_loss": 0.04028799612489012, "train_loss_bc": 0.03862873837351799, "train_loss_llm": 0.0, "grad_norm": 0.03393524885177612, "global_step": 7307, "epoch": 202, "lr": 0.001, "val_loss": 0.056851617991924286} +{"train_loss": 0.029942773282527924, "train_loss_bc": 0.029942773282527924, "train_loss_llm": 0.0, "grad_norm": 0.040081385523080826, "global_step": 7308, "epoch": 203, "lr": 0.001} +{"train_loss": 0.0394267700612545, "train_loss_bc": 0.03942533954977989, "train_loss_llm": 0.00014290213584899902, "grad_norm": 0.018770843744277954, "global_step": 7309, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04004605486989021, "train_loss_bc": 0.04003768414258957, "train_loss_llm": 0.0008371472358703613, "grad_norm": 0.030605871230363846, "global_step": 7310, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03490588814020157, "train_loss_bc": 0.03490588814020157, "train_loss_llm": 0.0, "grad_norm": 0.03319452330470085, "global_step": 7311, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04723378270864487, "train_loss_bc": 0.04723378270864487, "train_loss_llm": 0.0, "grad_norm": 0.04646896943449974, "global_step": 7312, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04065695405006409, "train_loss_bc": 0.04065695405006409, "train_loss_llm": 0.0, "grad_norm": 0.017062416300177574, "global_step": 7313, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04603046178817749, "train_loss_bc": 0.04603046178817749, "train_loss_llm": 0.0, "grad_norm": 0.03018689900636673, "global_step": 7314, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03418971225619316, "train_loss_bc": 0.03416505828499794, "train_loss_llm": 0.002465486526489258, "grad_norm": 0.03484029695391655, "global_step": 7315, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04913895204663277, "train_loss_bc": 0.04913895204663277, "train_loss_llm": 0.0, "grad_norm": 0.04510396718978882, "global_step": 7316, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04268467426300049, "train_loss_bc": 0.04268467426300049, "train_loss_llm": 0.0, "grad_norm": 0.01534661278128624, "global_step": 7317, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03713180497288704, "train_loss_bc": 0.03713180497288704, "train_loss_llm": 0.0, "grad_norm": 0.028555618599057198, "global_step": 7318, "epoch": 203, "lr": 0.001} +{"train_loss": 0.037194471806287766, "train_loss_bc": 0.037194471806287766, "train_loss_llm": 0.0, "grad_norm": 0.03994929790496826, "global_step": 7319, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04305221140384674, "train_loss_bc": 0.043047502636909485, "train_loss_llm": 0.0004710555076599121, "grad_norm": 0.05356786027550697, "global_step": 7320, "epoch": 203, "lr": 0.001} +{"train_loss": 0.042337916791439056, "train_loss_bc": 0.042337916791439056, "train_loss_llm": 0.0, "grad_norm": 0.014937030151486397, "global_step": 7321, "epoch": 203, "lr": 0.001} +{"train_loss": 0.038872070610523224, "train_loss_bc": 0.03882638365030289, "train_loss_llm": 0.004568815231323242, "grad_norm": 0.02499840408563614, "global_step": 7322, "epoch": 203, "lr": 0.001} +{"train_loss": 0.05426011234521866, "train_loss_bc": 0.05426011234521866, "train_loss_llm": 0.0, "grad_norm": 0.04123743996024132, "global_step": 7323, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04065323993563652, "train_loss_bc": 0.0406351275742054, "train_loss_llm": 0.0018113255500793457, "grad_norm": 0.06010306999087334, "global_step": 7324, "epoch": 203, "lr": 0.001} +{"train_loss": 0.035642094910144806, "train_loss_bc": 0.035642094910144806, "train_loss_llm": 0.0, "grad_norm": 0.016778314486145973, "global_step": 7325, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04396013543009758, "train_loss_bc": 0.04396013543009758, "train_loss_llm": 0.0, "grad_norm": 0.024078350514173508, "global_step": 7326, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03231848403811455, "train_loss_bc": 0.032307207584381104, "train_loss_llm": 0.0011276304721832275, "grad_norm": 0.03032158873975277, "global_step": 7327, "epoch": 203, "lr": 0.001} +{"train_loss": 0.049398183822631836, "train_loss_bc": 0.049398183822631836, "train_loss_llm": 0.0, "grad_norm": 0.03920402377843857, "global_step": 7328, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04379081726074219, "train_loss_bc": 0.04377748444676399, "train_loss_llm": 0.0013331770896911621, "grad_norm": 0.018559709191322327, "global_step": 7329, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03966011479496956, "train_loss_bc": 0.03966011479496956, "train_loss_llm": 0.0, "grad_norm": 0.02797897346317768, "global_step": 7330, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03874245285987854, "train_loss_bc": 0.03874245285987854, "train_loss_llm": 0.0, "grad_norm": 0.033687133342027664, "global_step": 7331, "epoch": 203, "lr": 0.001} +{"train_loss": 0.040379129350185394, "train_loss_bc": 0.040379129350185394, "train_loss_llm": 0.0, "grad_norm": 0.04157179594039917, "global_step": 7332, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03805408626794815, "train_loss_bc": 0.038046058267354965, "train_loss_llm": 0.0008029341697692871, "grad_norm": 0.018385697156190872, "global_step": 7333, "epoch": 203, "lr": 0.001} +{"train_loss": 0.035888366401195526, "train_loss_bc": 0.035888366401195526, "train_loss_llm": 0.0, "grad_norm": 0.028994165360927582, "global_step": 7334, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03896581754088402, "train_loss_bc": 0.03896581754088402, "train_loss_llm": 0.0, "grad_norm": 0.036930911242961884, "global_step": 7335, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03830062970519066, "train_loss_bc": 0.03830062970519066, "train_loss_llm": 0.0, "grad_norm": 0.045539550483226776, "global_step": 7336, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03705562278628349, "train_loss_bc": 0.03705562278628349, "train_loss_llm": 0.0, "grad_norm": 0.017377952113747597, "global_step": 7337, "epoch": 203, "lr": 0.001} +{"train_loss": 0.035232316702604294, "train_loss_bc": 0.035232316702604294, "train_loss_llm": 0.0, "grad_norm": 0.024242112413048744, "global_step": 7338, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04388614371418953, "train_loss_bc": 0.04388614371418953, "train_loss_llm": 0.0, "grad_norm": 0.03749658167362213, "global_step": 7339, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03846011310815811, "train_loss_bc": 0.03846011310815811, "train_loss_llm": 0.0, "grad_norm": 0.050112035125494, "global_step": 7340, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04255702719092369, "train_loss_bc": 0.04255371168255806, "train_loss_llm": 0.0003313720226287842, "grad_norm": 0.02194673754274845, "global_step": 7341, "epoch": 203, "lr": 0.001} +{"train_loss": 0.03588222339749336, "train_loss_bc": 0.03588222339749336, "train_loss_llm": 0.0, "grad_norm": 0.031485043466091156, "global_step": 7342, "epoch": 203, "lr": 0.001} +{"train_loss": 0.04034239974700742, "train_loss_bc": 0.04638829082250595, "train_loss_llm": 0.00064888596534729, "grad_norm": 0.04888230562210083, "global_step": 7343, "epoch": 203, "lr": 0.001, "val_loss": 0.057050034403800964} +{"train_loss": 0.03468422219157219, "train_loss_bc": 0.03468422219157219, "train_loss_llm": 0.0, "grad_norm": 0.05027490481734276, "global_step": 7344, "epoch": 204, "lr": 0.001} +{"train_loss": 0.0378524512052536, "train_loss_bc": 0.03783169016242027, "train_loss_llm": 0.00207597017288208, "grad_norm": 0.017148535698652267, "global_step": 7345, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04334908351302147, "train_loss_bc": 0.043327104300260544, "train_loss_llm": 0.0021979808807373047, "grad_norm": 0.028666231781244278, "global_step": 7346, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04327280819416046, "train_loss_bc": 0.04327280819416046, "train_loss_llm": 0.0, "grad_norm": 0.04228900745511055, "global_step": 7347, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03950561210513115, "train_loss_bc": 0.0395006500184536, "train_loss_llm": 0.000496208667755127, "grad_norm": 0.05231582000851631, "global_step": 7348, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03783966973423958, "train_loss_bc": 0.03783966973423958, "train_loss_llm": 0.0, "grad_norm": 0.02000720240175724, "global_step": 7349, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04062995687127113, "train_loss_bc": 0.04062369838356972, "train_loss_llm": 0.0006258189678192139, "grad_norm": 0.027616111561655998, "global_step": 7350, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04804735630750656, "train_loss_bc": 0.04801574721932411, "train_loss_llm": 0.0031610727310180664, "grad_norm": 0.03473619371652603, "global_step": 7351, "epoch": 204, "lr": 0.001} +{"train_loss": 0.041242312639951706, "train_loss_bc": 0.041242312639951706, "train_loss_llm": 0.0, "grad_norm": 0.04294237494468689, "global_step": 7352, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03806779161095619, "train_loss_bc": 0.0380486398935318, "train_loss_llm": 0.0019153356552124023, "grad_norm": 0.020152827724814415, "global_step": 7353, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03624524176120758, "train_loss_bc": 0.03624169901013374, "train_loss_llm": 0.0003541111946105957, "grad_norm": 0.029692964628338814, "global_step": 7354, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03800487145781517, "train_loss_bc": 0.038001615554094315, "train_loss_llm": 0.0003254115581512451, "grad_norm": 0.036677855998277664, "global_step": 7355, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03881479427218437, "train_loss_bc": 0.03881479427218437, "train_loss_llm": 0.0, "grad_norm": 0.045795030891895294, "global_step": 7356, "epoch": 204, "lr": 0.001} +{"train_loss": 0.041449934244155884, "train_loss_bc": 0.041449934244155884, "train_loss_llm": 0.0, "grad_norm": 0.019676607102155685, "global_step": 7357, "epoch": 204, "lr": 0.001} +{"train_loss": 0.039718933403491974, "train_loss_bc": 0.039713483303785324, "train_loss_llm": 0.0005448758602142334, "grad_norm": 0.029234779998660088, "global_step": 7358, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03635132685303688, "train_loss_bc": 0.03635132685303688, "train_loss_llm": 0.0, "grad_norm": 0.03440636768937111, "global_step": 7359, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04025657847523689, "train_loss_bc": 0.040254153311252594, "train_loss_llm": 0.00024241209030151367, "grad_norm": 0.038565993309020996, "global_step": 7360, "epoch": 204, "lr": 0.001} +{"train_loss": 0.041845306754112244, "train_loss_bc": 0.041845306754112244, "train_loss_llm": 0.0, "grad_norm": 0.02232929691672325, "global_step": 7361, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03359672427177429, "train_loss_bc": 0.033595480024814606, "train_loss_llm": 0.00012436509132385254, "grad_norm": 0.029812190681695938, "global_step": 7362, "epoch": 204, "lr": 0.001} +{"train_loss": 0.05024369806051254, "train_loss_bc": 0.05024369806051254, "train_loss_llm": 0.0, "grad_norm": 0.04019689932465553, "global_step": 7363, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04093051329255104, "train_loss_bc": 0.04093051329255104, "train_loss_llm": 0.0, "grad_norm": 0.046464912593364716, "global_step": 7364, "epoch": 204, "lr": 0.001} +{"train_loss": 0.033281344920396805, "train_loss_bc": 0.033281344920396805, "train_loss_llm": 0.0, "grad_norm": 0.019985634833574295, "global_step": 7365, "epoch": 204, "lr": 0.001} +{"train_loss": 0.0445549413561821, "train_loss_bc": 0.0445549413561821, "train_loss_llm": 0.0, "grad_norm": 0.033180851489305496, "global_step": 7366, "epoch": 204, "lr": 0.001} +{"train_loss": 0.045201994478702545, "train_loss_bc": 0.045201994478702545, "train_loss_llm": 0.0, "grad_norm": 0.05142767354846001, "global_step": 7367, "epoch": 204, "lr": 0.001} +{"train_loss": 0.0394948273897171, "train_loss_bc": 0.0394948273897171, "train_loss_llm": 0.0, "grad_norm": 0.06438250839710236, "global_step": 7368, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03835119679570198, "train_loss_bc": 0.03835119679570198, "train_loss_llm": 0.0, "grad_norm": 0.020995570346713066, "global_step": 7369, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03573048487305641, "train_loss_bc": 0.03573048487305641, "train_loss_llm": 0.0, "grad_norm": 0.031247060745954514, "global_step": 7370, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04245135560631752, "train_loss_bc": 0.04245135560631752, "train_loss_llm": 0.0, "grad_norm": 0.041871458292007446, "global_step": 7371, "epoch": 204, "lr": 0.001} +{"train_loss": 0.035126592963933945, "train_loss_bc": 0.035126592963933945, "train_loss_llm": 0.0, "grad_norm": 0.04844464734196663, "global_step": 7372, "epoch": 204, "lr": 0.001} +{"train_loss": 0.03934936225414276, "train_loss_bc": 0.03934875875711441, "train_loss_llm": 6.026029586791992e-05, "grad_norm": 0.020608708262443542, "global_step": 7373, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04298445209860802, "train_loss_bc": 0.04298445209860802, "train_loss_llm": 0.0, "grad_norm": 0.024842828512191772, "global_step": 7374, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04392678663134575, "train_loss_bc": 0.04392533004283905, "train_loss_llm": 0.0001455545425415039, "grad_norm": 0.031408004462718964, "global_step": 7375, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04975469782948494, "train_loss_bc": 0.04975469782948494, "train_loss_llm": 0.0, "grad_norm": 0.03772994130849838, "global_step": 7376, "epoch": 204, "lr": 0.001} +{"train_loss": 0.02835596352815628, "train_loss_bc": 0.02834644541144371, "train_loss_llm": 0.0009517967700958252, "grad_norm": 0.01590656489133835, "global_step": 7377, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04567654803395271, "train_loss_bc": 0.04567654803395271, "train_loss_llm": 0.0, "grad_norm": 0.035290833562612534, "global_step": 7378, "epoch": 204, "lr": 0.001} +{"train_loss": 0.04021579223788447, "train_loss_bc": 0.04155547916889191, "train_loss_llm": 0.0023304522037506104, "grad_norm": 0.04815947264432907, "global_step": 7379, "epoch": 204, "lr": 0.001, "val_loss": 0.05759727209806442} +{"train_loss": 0.03412022441625595, "train_loss_bc": 0.0341070294380188, "train_loss_llm": 0.0013193190097808838, "grad_norm": 0.05887022987008095, "global_step": 7380, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03999071940779686, "train_loss_bc": 0.03999071940779686, "train_loss_llm": 0.0, "grad_norm": 0.020497821271419525, "global_step": 7381, "epoch": 205, "lr": 0.001} +{"train_loss": 0.04148007184267044, "train_loss_bc": 0.041477665305137634, "train_loss_llm": 0.00024056434631347656, "grad_norm": 0.03451189398765564, "global_step": 7382, "epoch": 205, "lr": 0.001} +{"train_loss": 0.0333707258105278, "train_loss_bc": 0.0333707258105278, "train_loss_llm": 0.0, "grad_norm": 0.04263022169470787, "global_step": 7383, "epoch": 205, "lr": 0.001} +{"train_loss": 0.039607055485248566, "train_loss_bc": 0.039607055485248566, "train_loss_llm": 0.0, "grad_norm": 0.05350415036082268, "global_step": 7384, "epoch": 205, "lr": 0.001} +{"train_loss": 0.031260132789611816, "train_loss_bc": 0.031260132789611816, "train_loss_llm": 0.0, "grad_norm": 0.017384864389896393, "global_step": 7385, "epoch": 205, "lr": 0.001} +{"train_loss": 0.035443857312202454, "train_loss_bc": 0.035443857312202454, "train_loss_llm": 0.0, "grad_norm": 0.030212227255105972, "global_step": 7386, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03733645752072334, "train_loss_bc": 0.03732021898031235, "train_loss_llm": 0.001623690128326416, "grad_norm": 0.03839710354804993, "global_step": 7387, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03871486708521843, "train_loss_bc": 0.03871486708521843, "train_loss_llm": 0.0, "grad_norm": 0.05573926866054535, "global_step": 7388, "epoch": 205, "lr": 0.001} +{"train_loss": 0.038479212671518326, "train_loss_bc": 0.038479212671518326, "train_loss_llm": 0.0, "grad_norm": 0.019454071298241615, "global_step": 7389, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03651012107729912, "train_loss_bc": 0.03651012107729912, "train_loss_llm": 0.0, "grad_norm": 0.03014407679438591, "global_step": 7390, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03845701739192009, "train_loss_bc": 0.03845701739192009, "train_loss_llm": 0.0, "grad_norm": 0.03514620289206505, "global_step": 7391, "epoch": 205, "lr": 0.001} +{"train_loss": 0.04206053167581558, "train_loss_bc": 0.04205657169222832, "train_loss_llm": 0.0003961324691772461, "grad_norm": 0.03971552848815918, "global_step": 7392, "epoch": 205, "lr": 0.001} +{"train_loss": 0.04333779588341713, "train_loss_bc": 0.04333554580807686, "train_loss_llm": 0.0002250373363494873, "grad_norm": 0.019660498946905136, "global_step": 7393, "epoch": 205, "lr": 0.001} +{"train_loss": 0.032034650444984436, "train_loss_bc": 0.032034650444984436, "train_loss_llm": 0.0, "grad_norm": 0.024691393598914146, "global_step": 7394, "epoch": 205, "lr": 0.001} +{"train_loss": 0.042045243084430695, "train_loss_bc": 0.042045243084430695, "train_loss_llm": 0.0, "grad_norm": 0.031123846769332886, "global_step": 7395, "epoch": 205, "lr": 0.001} +{"train_loss": 0.052758388221263885, "train_loss_bc": 0.05275776982307434, "train_loss_llm": 6.198883056640625e-05, "grad_norm": 0.0396500788629055, "global_step": 7396, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03924296796321869, "train_loss_bc": 0.03924071416258812, "train_loss_llm": 0.00022521615028381348, "grad_norm": 0.014940287917852402, "global_step": 7397, "epoch": 205, "lr": 0.001} +{"train_loss": 0.045626960694789886, "train_loss_bc": 0.045626960694789886, "train_loss_llm": 0.0, "grad_norm": 0.03247656300663948, "global_step": 7398, "epoch": 205, "lr": 0.001} +{"train_loss": 0.05027130991220474, "train_loss_bc": 0.05027130991220474, "train_loss_llm": 0.0, "grad_norm": 0.03693487122654915, "global_step": 7399, "epoch": 205, "lr": 0.001} +{"train_loss": 0.044069670140743256, "train_loss_bc": 0.04406772553920746, "train_loss_llm": 0.00019443035125732422, "grad_norm": 0.0439094640314579, "global_step": 7400, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03881441056728363, "train_loss_bc": 0.03881441056728363, "train_loss_llm": 0.0, "grad_norm": 0.023661598563194275, "global_step": 7401, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03745738044381142, "train_loss_bc": 0.03745738044381142, "train_loss_llm": 0.0, "grad_norm": 0.037082552909851074, "global_step": 7402, "epoch": 205, "lr": 0.001} +{"train_loss": 0.04312365502119064, "train_loss_bc": 0.04312192648649216, "train_loss_llm": 0.00017267465591430664, "grad_norm": 0.05273441597819328, "global_step": 7403, "epoch": 205, "lr": 0.001} +{"train_loss": 0.0434771291911602, "train_loss_bc": 0.0434771291911602, "train_loss_llm": 0.0, "grad_norm": 0.0634562224149704, "global_step": 7404, "epoch": 205, "lr": 0.001} +{"train_loss": 0.039412721991539, "train_loss_bc": 0.039412721991539, "train_loss_llm": 0.0, "grad_norm": 0.01978696510195732, "global_step": 7405, "epoch": 205, "lr": 0.001} +{"train_loss": 0.04389984533190727, "train_loss_bc": 0.04389984533190727, "train_loss_llm": 0.0, "grad_norm": 0.03939230367541313, "global_step": 7406, "epoch": 205, "lr": 0.001} +{"train_loss": 0.0423591211438179, "train_loss_bc": 0.0423591211438179, "train_loss_llm": 0.0, "grad_norm": 0.06154691427946091, "global_step": 7407, "epoch": 205, "lr": 0.001} +{"train_loss": 0.04343322664499283, "train_loss_bc": 0.04343322664499283, "train_loss_llm": 0.0, "grad_norm": 0.07728524506092072, "global_step": 7408, "epoch": 205, "lr": 0.001} +{"train_loss": 0.05062435194849968, "train_loss_bc": 0.05062435194849968, "train_loss_llm": 0.0, "grad_norm": 0.018861209973692894, "global_step": 7409, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03984716534614563, "train_loss_bc": 0.03984139859676361, "train_loss_llm": 0.0005767643451690674, "grad_norm": 0.02361227385699749, "global_step": 7410, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03505586460232735, "train_loss_bc": 0.03505586460232735, "train_loss_llm": 0.0, "grad_norm": 0.03380116447806358, "global_step": 7411, "epoch": 205, "lr": 0.001} +{"train_loss": 0.038237497210502625, "train_loss_bc": 0.038237497210502625, "train_loss_llm": 0.0, "grad_norm": 0.04291665554046631, "global_step": 7412, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03676818683743477, "train_loss_bc": 0.03676818683743477, "train_loss_llm": 0.0, "grad_norm": 0.020217081531882286, "global_step": 7413, "epoch": 205, "lr": 0.001} +{"train_loss": 0.03992703929543495, "train_loss_bc": 0.03992703929543495, "train_loss_llm": 0.0, "grad_norm": 0.02684071660041809, "global_step": 7414, "epoch": 205, "lr": 0.001} +{"train_loss": 0.04013735200795862, "train_loss_bc": 0.03628415986895561, "train_loss_llm": 0.0004935264587402344, "grad_norm": 0.037262964993715286, "global_step": 7415, "epoch": 205, "lr": 0.001, "train/cumulative_reward": 102.66334003979975, "train/mean_score": 8.803248607760583, "train/success_rate": 0.0, "test/cumulative_reward": 34.496578318521095, "test/mean_score": 3.6007179907948723, "test/success_rate": 0.06, "val_loss": 0.056949831545352936, "train_action_mse_error": 0.04394083470106125} +{"train_loss": 0.047457505017519, "train_loss_bc": 0.047457505017519, "train_loss_llm": 0.0, "grad_norm": 0.04432610794901848, "global_step": 7416, "epoch": 206, "lr": 0.001} +{"train_loss": 0.040214113891124725, "train_loss_bc": 0.04016740620136261, "train_loss_llm": 0.004670888185501099, "grad_norm": 0.01863248273730278, "global_step": 7417, "epoch": 206, "lr": 0.001} +{"train_loss": 0.04167632386088371, "train_loss_bc": 0.041651204228401184, "train_loss_llm": 0.002511918544769287, "grad_norm": 0.030962631106376648, "global_step": 7418, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03549269214272499, "train_loss_bc": 0.03549269214272499, "train_loss_llm": 0.0, "grad_norm": 0.03453076630830765, "global_step": 7419, "epoch": 206, "lr": 0.001} +{"train_loss": 0.044724974781274796, "train_loss_bc": 0.044724974781274796, "train_loss_llm": 0.0, "grad_norm": 0.04019197076559067, "global_step": 7420, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03438040614128113, "train_loss_bc": 0.03438040614128113, "train_loss_llm": 0.0, "grad_norm": 0.018558451905846596, "global_step": 7421, "epoch": 206, "lr": 0.001} +{"train_loss": 0.037254832684993744, "train_loss_bc": 0.037218205630779266, "train_loss_llm": 0.0036626458168029785, "grad_norm": 0.029651377350091934, "global_step": 7422, "epoch": 206, "lr": 0.001} +{"train_loss": 0.047204528003931046, "train_loss_bc": 0.047204528003931046, "train_loss_llm": 0.0, "grad_norm": 0.03333348408341408, "global_step": 7423, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03663956746459007, "train_loss_bc": 0.036637332290410995, "train_loss_llm": 0.00022360682487487793, "grad_norm": 0.042770396918058395, "global_step": 7424, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03884861245751381, "train_loss_bc": 0.03884861245751381, "train_loss_llm": 0.0, "grad_norm": 0.01876659505069256, "global_step": 7425, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03495408594608307, "train_loss_bc": 0.03495170548558235, "train_loss_llm": 0.00023791193962097168, "grad_norm": 0.03247840330004692, "global_step": 7426, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03740251809358597, "train_loss_bc": 0.03740251809358597, "train_loss_llm": 0.0, "grad_norm": 0.03758475184440613, "global_step": 7427, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03674871474504471, "train_loss_bc": 0.03674871474504471, "train_loss_llm": 0.0, "grad_norm": 0.04499709978699684, "global_step": 7428, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03619114309549332, "train_loss_bc": 0.03619114309549332, "train_loss_llm": 0.0, "grad_norm": 0.021273143589496613, "global_step": 7429, "epoch": 206, "lr": 0.001} +{"train_loss": 0.035809218883514404, "train_loss_bc": 0.035798996686935425, "train_loss_llm": 0.0010222792625427246, "grad_norm": 0.03236696869134903, "global_step": 7430, "epoch": 206, "lr": 0.001} +{"train_loss": 0.04145690053701401, "train_loss_bc": 0.04145690053701401, "train_loss_llm": 0.0, "grad_norm": 0.042058732360601425, "global_step": 7431, "epoch": 206, "lr": 0.001} +{"train_loss": 0.04454628750681877, "train_loss_bc": 0.04453321546316147, "train_loss_llm": 0.0013070404529571533, "grad_norm": 0.05479756370186806, "global_step": 7432, "epoch": 206, "lr": 0.001} +{"train_loss": 0.043499190360307693, "train_loss_bc": 0.043485864996910095, "train_loss_llm": 0.001332610845565796, "grad_norm": 0.019288845360279083, "global_step": 7433, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03790957108139992, "train_loss_bc": 0.03790957108139992, "train_loss_llm": 0.0, "grad_norm": 0.03287835419178009, "global_step": 7434, "epoch": 206, "lr": 0.001} +{"train_loss": 0.0350290983915329, "train_loss_bc": 0.03501439839601517, "train_loss_llm": 0.001470029354095459, "grad_norm": 0.038259632885456085, "global_step": 7435, "epoch": 206, "lr": 0.001} +{"train_loss": 0.040451619774103165, "train_loss_bc": 0.040451619774103165, "train_loss_llm": 0.0, "grad_norm": 0.04830680042505264, "global_step": 7436, "epoch": 206, "lr": 0.001} +{"train_loss": 0.04598332569003105, "train_loss_bc": 0.04598332569003105, "train_loss_llm": 0.0, "grad_norm": 0.014353920705616474, "global_step": 7437, "epoch": 206, "lr": 0.001} +{"train_loss": 0.041189197450876236, "train_loss_bc": 0.04118618369102478, "train_loss_llm": 0.0003012716770172119, "grad_norm": 0.020882487297058105, "global_step": 7438, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03280450403690338, "train_loss_bc": 0.03280450403690338, "train_loss_llm": 0.0, "grad_norm": 0.029981952160596848, "global_step": 7439, "epoch": 206, "lr": 0.001} +{"train_loss": 0.04598558321595192, "train_loss_bc": 0.04598558321595192, "train_loss_llm": 0.0, "grad_norm": 0.0378924161195755, "global_step": 7440, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03727911785244942, "train_loss_bc": 0.03727911785244942, "train_loss_llm": 0.0, "grad_norm": 0.02026776783168316, "global_step": 7441, "epoch": 206, "lr": 0.001} +{"train_loss": 0.036512766033411026, "train_loss_bc": 0.036507513374090195, "train_loss_llm": 0.0005251169204711914, "grad_norm": 0.027058787643909454, "global_step": 7442, "epoch": 206, "lr": 0.001} +{"train_loss": 0.04012354463338852, "train_loss_bc": 0.04012354463338852, "train_loss_llm": 0.0, "grad_norm": 0.03985228016972542, "global_step": 7443, "epoch": 206, "lr": 0.001} +{"train_loss": 0.03584052994847298, "train_loss_bc": 0.03584052994847298, "train_loss_llm": 0.0, "grad_norm": 0.04785354435443878, "global_step": 7444, "epoch": 206, "lr": 0.001} +{"train_loss": 0.0418725423514843, "train_loss_bc": 0.0418725423514843, "train_loss_llm": 0.0, "grad_norm": 0.018439508974552155, "global_step": 7445, "epoch": 206, "lr": 0.001} +{"train_loss": 0.04199853166937828, "train_loss_bc": 0.04199853166937828, "train_loss_llm": 0.0, "grad_norm": 0.03342868760228157, "global_step": 7446, "epoch": 206, "lr": 0.001} +{"train_loss": 0.043148674070835114, "train_loss_bc": 0.04314697906374931, "train_loss_llm": 0.00016948580741882324, "grad_norm": 0.04846984148025513, "global_step": 7447, "epoch": 206, "lr": 0.001} +{"train_loss": 0.04284701496362686, "train_loss_bc": 0.04284701496362686, "train_loss_llm": 0.0, "grad_norm": 0.05635557323694229, "global_step": 7448, "epoch": 206, "lr": 0.001} +{"train_loss": 0.04463953524827957, "train_loss_bc": 0.04463953524827957, "train_loss_llm": 0.0, "grad_norm": 0.024264218285679817, "global_step": 7449, "epoch": 206, "lr": 0.001} +{"train_loss": 0.04319409653544426, "train_loss_bc": 0.04318510740995407, "train_loss_llm": 0.0008987486362457275, "grad_norm": 0.03919926658272743, "global_step": 7450, "epoch": 206, "lr": 0.001} +{"train_loss": 0.040211773580975003, "train_loss_bc": 0.046312980353832245, "train_loss_llm": 0.0, "grad_norm": 0.05270444601774216, "global_step": 7451, "epoch": 206, "lr": 0.001, "val_loss": 0.057131849229335785} +{"train_loss": 0.031221505254507065, "train_loss_bc": 0.031221505254507065, "train_loss_llm": 0.0, "grad_norm": 0.06509962677955627, "global_step": 7452, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04072744399309158, "train_loss_bc": 0.04070433974266052, "train_loss_llm": 0.002310246229171753, "grad_norm": 0.01855825074017048, "global_step": 7453, "epoch": 207, "lr": 0.001} +{"train_loss": 0.041176240891218185, "train_loss_bc": 0.041176240891218185, "train_loss_llm": 0.0, "grad_norm": 0.03294065222144127, "global_step": 7454, "epoch": 207, "lr": 0.001} +{"train_loss": 0.03593146428465843, "train_loss_bc": 0.03593146428465843, "train_loss_llm": 0.0, "grad_norm": 0.043807730078697205, "global_step": 7455, "epoch": 207, "lr": 0.001} +{"train_loss": 0.03722317889332771, "train_loss_bc": 0.03721872344613075, "train_loss_llm": 0.00044545531272888184, "grad_norm": 0.050778187811374664, "global_step": 7456, "epoch": 207, "lr": 0.001} +{"train_loss": 0.037741489708423615, "train_loss_bc": 0.03773489221930504, "train_loss_llm": 0.0006598532199859619, "grad_norm": 0.014993925578892231, "global_step": 7457, "epoch": 207, "lr": 0.001} +{"train_loss": 0.0370575450360775, "train_loss_bc": 0.037052735686302185, "train_loss_llm": 0.00048103928565979004, "grad_norm": 0.02020486816763878, "global_step": 7458, "epoch": 207, "lr": 0.001} +{"train_loss": 0.03826406970620155, "train_loss_bc": 0.038263749331235886, "train_loss_llm": 3.2007694244384766e-05, "grad_norm": 0.02721492201089859, "global_step": 7459, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04057835042476654, "train_loss_bc": 0.04057835042476654, "train_loss_llm": 0.0, "grad_norm": 0.03090890496969223, "global_step": 7460, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04310435801744461, "train_loss_bc": 0.04310435801744461, "train_loss_llm": 0.0, "grad_norm": 0.017499445006251335, "global_step": 7461, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04603330418467522, "train_loss_bc": 0.04603330418467522, "train_loss_llm": 0.0, "grad_norm": 0.024403933435678482, "global_step": 7462, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04008958488702774, "train_loss_bc": 0.04008958488702774, "train_loss_llm": 0.0, "grad_norm": 0.03279868885874748, "global_step": 7463, "epoch": 207, "lr": 0.001} +{"train_loss": 0.03831722214818001, "train_loss_bc": 0.03831619769334793, "train_loss_llm": 0.00010246038436889648, "grad_norm": 0.04183221608400345, "global_step": 7464, "epoch": 207, "lr": 0.001} +{"train_loss": 0.044344257563352585, "train_loss_bc": 0.04433923959732056, "train_loss_llm": 0.0005016624927520752, "grad_norm": 0.02437420003116131, "global_step": 7465, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04392343759536743, "train_loss_bc": 0.04389388859272003, "train_loss_llm": 0.002954930067062378, "grad_norm": 0.02784258872270584, "global_step": 7466, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04553446173667908, "train_loss_bc": 0.04553446173667908, "train_loss_llm": 0.0, "grad_norm": 0.03550659865140915, "global_step": 7467, "epoch": 207, "lr": 0.001} +{"train_loss": 0.046023204922676086, "train_loss_bc": 0.046023204922676086, "train_loss_llm": 0.0, "grad_norm": 0.044134341180324554, "global_step": 7468, "epoch": 207, "lr": 0.001} +{"train_loss": 0.03349996730685234, "train_loss_bc": 0.03349996730685234, "train_loss_llm": 0.0, "grad_norm": 0.022864272817969322, "global_step": 7469, "epoch": 207, "lr": 0.001} +{"train_loss": 0.03749443218111992, "train_loss_bc": 0.03749443218111992, "train_loss_llm": 0.0, "grad_norm": 0.028152301907539368, "global_step": 7470, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04728075489401817, "train_loss_bc": 0.04727289080619812, "train_loss_llm": 0.0007863044738769531, "grad_norm": 0.04328108951449394, "global_step": 7471, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04329827055335045, "train_loss_bc": 0.04327026382088661, "train_loss_llm": 0.002800583839416504, "grad_norm": 0.06292366236448288, "global_step": 7472, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04625597223639488, "train_loss_bc": 0.04625597223639488, "train_loss_llm": 0.0, "grad_norm": 0.020380567759275436, "global_step": 7473, "epoch": 207, "lr": 0.001} +{"train_loss": 0.040683161467313766, "train_loss_bc": 0.040683161467313766, "train_loss_llm": 0.0, "grad_norm": 0.03103051893413067, "global_step": 7474, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04734627157449722, "train_loss_bc": 0.04731948673725128, "train_loss_llm": 0.0026783645153045654, "grad_norm": 0.04070631042122841, "global_step": 7475, "epoch": 207, "lr": 0.001} +{"train_loss": 0.0385652519762516, "train_loss_bc": 0.03854505345225334, "train_loss_llm": 0.0020197629928588867, "grad_norm": 0.04917532950639725, "global_step": 7476, "epoch": 207, "lr": 0.001} +{"train_loss": 0.040139563381671906, "train_loss_bc": 0.040139563381671906, "train_loss_llm": 0.0, "grad_norm": 0.0160540584474802, "global_step": 7477, "epoch": 207, "lr": 0.001} +{"train_loss": 0.03964458033442497, "train_loss_bc": 0.03964458033442497, "train_loss_llm": 0.0, "grad_norm": 0.02615443244576454, "global_step": 7478, "epoch": 207, "lr": 0.001} +{"train_loss": 0.034058745950460434, "train_loss_bc": 0.034058745950460434, "train_loss_llm": 0.0, "grad_norm": 0.03245590254664421, "global_step": 7479, "epoch": 207, "lr": 0.001} +{"train_loss": 0.02868114970624447, "train_loss_bc": 0.02868114970624447, "train_loss_llm": 0.0, "grad_norm": 0.037298694252967834, "global_step": 7480, "epoch": 207, "lr": 0.001} +{"train_loss": 0.046994633972644806, "train_loss_bc": 0.046994633972644806, "train_loss_llm": 0.0, "grad_norm": 0.019612116739153862, "global_step": 7481, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04062585532665253, "train_loss_bc": 0.04062585532665253, "train_loss_llm": 0.0, "grad_norm": 0.026866288855671883, "global_step": 7482, "epoch": 207, "lr": 0.001} +{"train_loss": 0.0405978187918663, "train_loss_bc": 0.0405978187918663, "train_loss_llm": 0.0, "grad_norm": 0.03733902424573898, "global_step": 7483, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04295815899968147, "train_loss_bc": 0.04295815899968147, "train_loss_llm": 0.0, "grad_norm": 0.04360504448413849, "global_step": 7484, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04030648246407509, "train_loss_bc": 0.04030648246407509, "train_loss_llm": 0.0, "grad_norm": 0.022398075088858604, "global_step": 7485, "epoch": 207, "lr": 0.001} +{"train_loss": 0.027699774131178856, "train_loss_bc": 0.027699774131178856, "train_loss_llm": 0.0, "grad_norm": 0.034154221415519714, "global_step": 7486, "epoch": 207, "lr": 0.001} +{"train_loss": 0.04011206463393238, "train_loss_bc": 0.04060324653983116, "train_loss_llm": 0.0009116232395172119, "grad_norm": 0.058660827577114105, "global_step": 7487, "epoch": 207, "lr": 0.001, "val_loss": 0.05692596733570099} +{"train_loss": 0.04081688076257706, "train_loss_bc": 0.04081688076257706, "train_loss_llm": 0.0, "grad_norm": 0.06584766507148743, "global_step": 7488, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03304317221045494, "train_loss_bc": 0.033028848469257355, "train_loss_llm": 0.0014322400093078613, "grad_norm": 0.019926482811570168, "global_step": 7489, "epoch": 208, "lr": 0.001} +{"train_loss": 0.0348392091691494, "train_loss_bc": 0.0348392091691494, "train_loss_llm": 0.0, "grad_norm": 0.0323345884680748, "global_step": 7490, "epoch": 208, "lr": 0.001} +{"train_loss": 0.044249776750802994, "train_loss_bc": 0.044249776750802994, "train_loss_llm": 0.0, "grad_norm": 0.042436614632606506, "global_step": 7491, "epoch": 208, "lr": 0.001} +{"train_loss": 0.04989907890558243, "train_loss_bc": 0.04989907890558243, "train_loss_llm": 0.0, "grad_norm": 0.05003044754266739, "global_step": 7492, "epoch": 208, "lr": 0.001} +{"train_loss": 0.031922124326229095, "train_loss_bc": 0.03190520405769348, "train_loss_llm": 0.0016921460628509521, "grad_norm": 0.017222875729203224, "global_step": 7493, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03979337960481644, "train_loss_bc": 0.03979337960481644, "train_loss_llm": 0.0, "grad_norm": 0.027507955208420753, "global_step": 7494, "epoch": 208, "lr": 0.001} +{"train_loss": 0.04149596020579338, "train_loss_bc": 0.04149596020579338, "train_loss_llm": 0.0, "grad_norm": 0.033209845423698425, "global_step": 7495, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03708531707525253, "train_loss_bc": 0.03708531707525253, "train_loss_llm": 0.0, "grad_norm": 0.04152756556868553, "global_step": 7496, "epoch": 208, "lr": 0.001} +{"train_loss": 0.04129745066165924, "train_loss_bc": 0.04129745066165924, "train_loss_llm": 0.0, "grad_norm": 0.01627676747739315, "global_step": 7497, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03829037398099899, "train_loss_bc": 0.03828056901693344, "train_loss_llm": 0.0009803175926208496, "grad_norm": 0.02472403086721897, "global_step": 7498, "epoch": 208, "lr": 0.001} +{"train_loss": 0.04354994371533394, "train_loss_bc": 0.04354994371533394, "train_loss_llm": 0.0, "grad_norm": 0.03984209522604942, "global_step": 7499, "epoch": 208, "lr": 0.001} +{"train_loss": 0.041009340435266495, "train_loss_bc": 0.041009340435266495, "train_loss_llm": 0.0, "grad_norm": 0.04421408474445343, "global_step": 7500, "epoch": 208, "lr": 0.001} +{"train_loss": 0.04234405234456062, "train_loss_bc": 0.04233362525701523, "train_loss_llm": 0.0010428428649902344, "grad_norm": 0.022723102942109108, "global_step": 7501, "epoch": 208, "lr": 0.001} +{"train_loss": 0.04152090102434158, "train_loss_bc": 0.04152090102434158, "train_loss_llm": 0.0, "grad_norm": 0.03493702784180641, "global_step": 7502, "epoch": 208, "lr": 0.001} +{"train_loss": 0.038643695414066315, "train_loss_bc": 0.03862999379634857, "train_loss_llm": 0.001370251178741455, "grad_norm": 0.03729652985930443, "global_step": 7503, "epoch": 208, "lr": 0.001} +{"train_loss": 0.04340774565935135, "train_loss_bc": 0.04340774565935135, "train_loss_llm": 0.0, "grad_norm": 0.0432114452123642, "global_step": 7504, "epoch": 208, "lr": 0.001} +{"train_loss": 0.033886563032865524, "train_loss_bc": 0.03388611599802971, "train_loss_llm": 4.461407661437988e-05, "grad_norm": 0.02001974545419216, "global_step": 7505, "epoch": 208, "lr": 0.001} +{"train_loss": 0.048997849225997925, "train_loss_bc": 0.048997849225997925, "train_loss_llm": 0.0, "grad_norm": 0.02609667368233204, "global_step": 7506, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03821750357747078, "train_loss_bc": 0.03821750357747078, "train_loss_llm": 0.0, "grad_norm": 0.03897467628121376, "global_step": 7507, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03917748108506203, "train_loss_bc": 0.03917748108506203, "train_loss_llm": 0.0, "grad_norm": 0.05081064626574516, "global_step": 7508, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03446352109313011, "train_loss_bc": 0.03446352109313011, "train_loss_llm": 0.0, "grad_norm": 0.020922189578413963, "global_step": 7509, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03872457891702652, "train_loss_bc": 0.03871762380003929, "train_loss_llm": 0.0006956756114959717, "grad_norm": 0.030235763639211655, "global_step": 7510, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03649933636188507, "train_loss_bc": 0.03649933636188507, "train_loss_llm": 0.0, "grad_norm": 0.042502760887145996, "global_step": 7511, "epoch": 208, "lr": 0.001} +{"train_loss": 0.04323871433734894, "train_loss_bc": 0.04323871433734894, "train_loss_llm": 0.0, "grad_norm": 0.056920815259218216, "global_step": 7512, "epoch": 208, "lr": 0.001} +{"train_loss": 0.031164897605776787, "train_loss_bc": 0.031164897605776787, "train_loss_llm": 0.0, "grad_norm": 0.017688419669866562, "global_step": 7513, "epoch": 208, "lr": 0.001} +{"train_loss": 0.045288510620594025, "train_loss_bc": 0.045276615768671036, "train_loss_llm": 0.0011895298957824707, "grad_norm": 0.030207958072423935, "global_step": 7514, "epoch": 208, "lr": 0.001} +{"train_loss": 0.040436629205942154, "train_loss_bc": 0.040436629205942154, "train_loss_llm": 0.0, "grad_norm": 0.047909703105688095, "global_step": 7515, "epoch": 208, "lr": 0.001} +{"train_loss": 0.037611279636621475, "train_loss_bc": 0.037611279636621475, "train_loss_llm": 0.0, "grad_norm": 0.06543150544166565, "global_step": 7516, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03699290007352829, "train_loss_bc": 0.03696583956480026, "train_loss_llm": 0.0027061402797698975, "grad_norm": 0.017725201323628426, "global_step": 7517, "epoch": 208, "lr": 0.001} +{"train_loss": 0.04640555754303932, "train_loss_bc": 0.04640555754303932, "train_loss_llm": 0.0, "grad_norm": 0.027164362370967865, "global_step": 7518, "epoch": 208, "lr": 0.001} +{"train_loss": 0.041844528168439865, "train_loss_bc": 0.04182658717036247, "train_loss_llm": 0.0017939209938049316, "grad_norm": 0.040176279842853546, "global_step": 7519, "epoch": 208, "lr": 0.001} +{"train_loss": 0.03794698789715767, "train_loss_bc": 0.03794698789715767, "train_loss_llm": 0.0, "grad_norm": 0.04974693804979324, "global_step": 7520, "epoch": 208, "lr": 0.001} +{"train_loss": 0.034090448170900345, "train_loss_bc": 0.034090448170900345, "train_loss_llm": 0.0, "grad_norm": 0.017648285254836082, "global_step": 7521, "epoch": 208, "lr": 0.001} +{"train_loss": 0.05010545998811722, "train_loss_bc": 0.050097014755010605, "train_loss_llm": 0.0008446276187896729, "grad_norm": 0.027937371283769608, "global_step": 7522, "epoch": 208, "lr": 0.001} +{"train_loss": 0.04023823840543628, "train_loss_bc": 0.05027543380856514, "train_loss_llm": 0.0, "grad_norm": 0.045297395437955856, "global_step": 7523, "epoch": 208, "lr": 0.001, "val_loss": 0.057847652584314346} +{"train_loss": 0.038128431886434555, "train_loss_bc": 0.038128431886434555, "train_loss_llm": 0.0, "grad_norm": 0.05203519016504288, "global_step": 7524, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04309939220547676, "train_loss_bc": 0.04305436089634895, "train_loss_llm": 0.0045029520988464355, "grad_norm": 0.022053172811865807, "global_step": 7525, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04136896878480911, "train_loss_bc": 0.04136896878480911, "train_loss_llm": 0.0, "grad_norm": 0.03435414656996727, "global_step": 7526, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04144160822033882, "train_loss_bc": 0.04144160822033882, "train_loss_llm": 0.0, "grad_norm": 0.04389609396457672, "global_step": 7527, "epoch": 209, "lr": 0.001} +{"train_loss": 0.03959117829799652, "train_loss_bc": 0.03958084061741829, "train_loss_llm": 0.001033782958984375, "grad_norm": 0.05435013771057129, "global_step": 7528, "epoch": 209, "lr": 0.001} +{"train_loss": 0.035016801208257675, "train_loss_bc": 0.035016801208257675, "train_loss_llm": 0.0, "grad_norm": 0.01632845401763916, "global_step": 7529, "epoch": 209, "lr": 0.001} +{"train_loss": 0.039571553468704224, "train_loss_bc": 0.03957127034664154, "train_loss_llm": 2.816319465637207e-05, "grad_norm": 0.03252895921468735, "global_step": 7530, "epoch": 209, "lr": 0.001} +{"train_loss": 0.03533943369984627, "train_loss_bc": 0.03533943369984627, "train_loss_llm": 0.0, "grad_norm": 0.04845836013555527, "global_step": 7531, "epoch": 209, "lr": 0.001} +{"train_loss": 0.03771933168172836, "train_loss_bc": 0.03771933168172836, "train_loss_llm": 0.0, "grad_norm": 0.06396330893039703, "global_step": 7532, "epoch": 209, "lr": 0.001} +{"train_loss": 0.042716529220342636, "train_loss_bc": 0.042716193944215775, "train_loss_llm": 3.364682197570801e-05, "grad_norm": 0.019649213179945946, "global_step": 7533, "epoch": 209, "lr": 0.001} +{"train_loss": 0.041052013635635376, "train_loss_bc": 0.04104965925216675, "train_loss_llm": 0.00023543834686279297, "grad_norm": 0.030690902844071388, "global_step": 7534, "epoch": 209, "lr": 0.001} +{"train_loss": 0.046616796404123306, "train_loss_bc": 0.046616796404123306, "train_loss_llm": 0.0, "grad_norm": 0.03858937695622444, "global_step": 7535, "epoch": 209, "lr": 0.001} +{"train_loss": 0.03606375679373741, "train_loss_bc": 0.03605853021144867, "train_loss_llm": 0.0005225539207458496, "grad_norm": 0.0473225861787796, "global_step": 7536, "epoch": 209, "lr": 0.001} +{"train_loss": 0.036550674587488174, "train_loss_bc": 0.036550674587488174, "train_loss_llm": 0.0, "grad_norm": 0.017132936045527458, "global_step": 7537, "epoch": 209, "lr": 0.001} +{"train_loss": 0.03906077891588211, "train_loss_bc": 0.039044689387083054, "train_loss_llm": 0.0016089975833892822, "grad_norm": 0.02739785984158516, "global_step": 7538, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04381521791219711, "train_loss_bc": 0.04381521791219711, "train_loss_llm": 0.0, "grad_norm": 0.03867771476507187, "global_step": 7539, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04026264697313309, "train_loss_bc": 0.04026264697313309, "train_loss_llm": 0.0, "grad_norm": 0.04819135740399361, "global_step": 7540, "epoch": 209, "lr": 0.001} +{"train_loss": 0.03815304487943649, "train_loss_bc": 0.03815304487943649, "train_loss_llm": 0.0, "grad_norm": 0.023296816274523735, "global_step": 7541, "epoch": 209, "lr": 0.001} +{"train_loss": 0.0472525954246521, "train_loss_bc": 0.0472525954246521, "train_loss_llm": 0.0, "grad_norm": 0.03656080737709999, "global_step": 7542, "epoch": 209, "lr": 0.001} +{"train_loss": 0.044255416840314865, "train_loss_bc": 0.044255416840314865, "train_loss_llm": 0.0, "grad_norm": 0.053094036877155304, "global_step": 7543, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04456837475299835, "train_loss_bc": 0.04456837475299835, "train_loss_llm": 0.0, "grad_norm": 0.05788661167025566, "global_step": 7544, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04638348147273064, "train_loss_bc": 0.04636877775192261, "train_loss_llm": 0.0014705359935760498, "grad_norm": 0.033415358513593674, "global_step": 7545, "epoch": 209, "lr": 0.001} +{"train_loss": 0.03683402016758919, "train_loss_bc": 0.03683402016758919, "train_loss_llm": 0.0, "grad_norm": 0.053442440927028656, "global_step": 7546, "epoch": 209, "lr": 0.001} +{"train_loss": 0.037711817771196365, "train_loss_bc": 0.037702832370996475, "train_loss_llm": 0.0008987188339233398, "grad_norm": 0.07124766707420349, "global_step": 7547, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04015721008181572, "train_loss_bc": 0.04013211280107498, "train_loss_llm": 0.0025098323822021484, "grad_norm": 0.08290428668260574, "global_step": 7548, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04145146533846855, "train_loss_bc": 0.04145146533846855, "train_loss_llm": 0.0, "grad_norm": 0.02552960254251957, "global_step": 7549, "epoch": 209, "lr": 0.001} +{"train_loss": 0.037430260330438614, "train_loss_bc": 0.03742676600813866, "train_loss_llm": 0.00034949183464050293, "grad_norm": 0.034267622977495193, "global_step": 7550, "epoch": 209, "lr": 0.001} +{"train_loss": 0.033671632409095764, "train_loss_bc": 0.03366198390722275, "train_loss_llm": 0.0009649097919464111, "grad_norm": 0.05004039779305458, "global_step": 7551, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04955748841166496, "train_loss_bc": 0.04955748841166496, "train_loss_llm": 0.0, "grad_norm": 0.06891791522502899, "global_step": 7552, "epoch": 209, "lr": 0.001} +{"train_loss": 0.03675185889005661, "train_loss_bc": 0.036743421107530594, "train_loss_llm": 0.000843733549118042, "grad_norm": 0.019743090495467186, "global_step": 7553, "epoch": 209, "lr": 0.001} +{"train_loss": 0.036821503192186356, "train_loss_bc": 0.036821503192186356, "train_loss_llm": 0.0, "grad_norm": 0.025542324408888817, "global_step": 7554, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04627786576747894, "train_loss_bc": 0.04626655951142311, "train_loss_llm": 0.0011304914951324463, "grad_norm": 0.03979936242103577, "global_step": 7555, "epoch": 209, "lr": 0.001} +{"train_loss": 0.038847386837005615, "train_loss_bc": 0.03884512931108475, "train_loss_llm": 0.0002256929874420166, "grad_norm": 0.05116055905818939, "global_step": 7556, "epoch": 209, "lr": 0.001} +{"train_loss": 0.03338580206036568, "train_loss_bc": 0.03338580206036568, "train_loss_llm": 0.0, "grad_norm": 0.016123397275805473, "global_step": 7557, "epoch": 209, "lr": 0.001} +{"train_loss": 0.03895216062664986, "train_loss_bc": 0.03895216062664986, "train_loss_llm": 0.0, "grad_norm": 0.024262497201561928, "global_step": 7558, "epoch": 209, "lr": 0.001} +{"train_loss": 0.04031921178102493, "train_loss_bc": 0.0456131249666214, "train_loss_llm": 0.0, "grad_norm": 0.03265666961669922, "global_step": 7559, "epoch": 209, "lr": 0.001, "val_loss": 0.057358670979738235} +{"train_loss": 0.03983953222632408, "train_loss_bc": 0.03983953222632408, "train_loss_llm": 0.0, "grad_norm": 0.039547067135572433, "global_step": 7560, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04008382186293602, "train_loss_bc": 0.04008382186293602, "train_loss_llm": 0.0, "grad_norm": 0.027683870866894722, "global_step": 7561, "epoch": 210, "lr": 0.001} +{"train_loss": 0.039129387587308884, "train_loss_bc": 0.039129387587308884, "train_loss_llm": 0.0, "grad_norm": 0.03672211989760399, "global_step": 7562, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04231446236371994, "train_loss_bc": 0.04231446236371994, "train_loss_llm": 0.0, "grad_norm": 0.05085662007331848, "global_step": 7563, "epoch": 210, "lr": 0.001} +{"train_loss": 0.044189196079969406, "train_loss_bc": 0.04418789595365524, "train_loss_llm": 0.00013014674186706543, "grad_norm": 0.058399077504873276, "global_step": 7564, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04400835186243057, "train_loss_bc": 0.04400835186243057, "train_loss_llm": 0.0, "grad_norm": 0.02053884044289589, "global_step": 7565, "epoch": 210, "lr": 0.001} +{"train_loss": 0.037913013249635696, "train_loss_bc": 0.037913013249635696, "train_loss_llm": 0.0, "grad_norm": 0.033988676965236664, "global_step": 7566, "epoch": 210, "lr": 0.001} +{"train_loss": 0.037241555750370026, "train_loss_bc": 0.037241555750370026, "train_loss_llm": 0.0, "grad_norm": 0.04707854613661766, "global_step": 7567, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03771218657493591, "train_loss_bc": 0.03771218657493591, "train_loss_llm": 0.0, "grad_norm": 0.05745479092001915, "global_step": 7568, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03773997724056244, "train_loss_bc": 0.03773997724056244, "train_loss_llm": 0.0, "grad_norm": 0.020793283358216286, "global_step": 7569, "epoch": 210, "lr": 0.001} +{"train_loss": 0.047514330595731735, "train_loss_bc": 0.04750027880072594, "train_loss_llm": 0.0014051198959350586, "grad_norm": 0.038538988679647446, "global_step": 7570, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04145295172929764, "train_loss_bc": 0.04145295172929764, "train_loss_llm": 0.0, "grad_norm": 0.04506545141339302, "global_step": 7571, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04482145234942436, "train_loss_bc": 0.04482145234942436, "train_loss_llm": 0.0, "grad_norm": 0.05446790158748627, "global_step": 7572, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04396034777164459, "train_loss_bc": 0.04396034777164459, "train_loss_llm": 0.0, "grad_norm": 0.02211909182369709, "global_step": 7573, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03425874561071396, "train_loss_bc": 0.03425850346684456, "train_loss_llm": 2.4080276489257812e-05, "grad_norm": 0.030004607513546944, "global_step": 7574, "epoch": 210, "lr": 0.001} +{"train_loss": 0.042550161480903625, "train_loss_bc": 0.042550161480903625, "train_loss_llm": 0.0, "grad_norm": 0.03511317819356918, "global_step": 7575, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04061245173215866, "train_loss_bc": 0.04061245173215866, "train_loss_llm": 0.0, "grad_norm": 0.03795882314443588, "global_step": 7576, "epoch": 210, "lr": 0.001} +{"train_loss": 0.038029320538043976, "train_loss_bc": 0.038029320538043976, "train_loss_llm": 0.0, "grad_norm": 0.02005610428750515, "global_step": 7577, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03781239315867424, "train_loss_bc": 0.03781239315867424, "train_loss_llm": 0.0, "grad_norm": 0.026640409603714943, "global_step": 7578, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04346362501382828, "train_loss_bc": 0.043452776968479156, "train_loss_llm": 0.0010846257209777832, "grad_norm": 0.03692111000418663, "global_step": 7579, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03468985855579376, "train_loss_bc": 0.034682512283325195, "train_loss_llm": 0.0007344484329223633, "grad_norm": 0.04391384497284889, "global_step": 7580, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03736100345849991, "train_loss_bc": 0.03736100345849991, "train_loss_llm": 0.0, "grad_norm": 0.01633988320827484, "global_step": 7581, "epoch": 210, "lr": 0.001} +{"train_loss": 0.042107462882995605, "train_loss_bc": 0.042107462882995605, "train_loss_llm": 0.0, "grad_norm": 0.025768114253878593, "global_step": 7582, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04310166835784912, "train_loss_bc": 0.043084681034088135, "train_loss_llm": 0.001698613166809082, "grad_norm": 0.03667212277650833, "global_step": 7583, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03694895654916763, "train_loss_bc": 0.03693555295467377, "train_loss_llm": 0.0013405084609985352, "grad_norm": 0.04294319823384285, "global_step": 7584, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03951107710599899, "train_loss_bc": 0.03951107710599899, "train_loss_llm": 0.0, "grad_norm": 0.017994234338402748, "global_step": 7585, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03928425535559654, "train_loss_bc": 0.03928055986762047, "train_loss_llm": 0.00036966800689697266, "grad_norm": 0.028172321617603302, "global_step": 7586, "epoch": 210, "lr": 0.001} +{"train_loss": 0.0378711074590683, "train_loss_bc": 0.0378711074590683, "train_loss_llm": 0.0, "grad_norm": 0.04420378804206848, "global_step": 7587, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04325498640537262, "train_loss_bc": 0.04325498640537262, "train_loss_llm": 0.0, "grad_norm": 0.06180925667285919, "global_step": 7588, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03781814128160477, "train_loss_bc": 0.03781814128160477, "train_loss_llm": 0.0, "grad_norm": 0.02007608860731125, "global_step": 7589, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04659637436270714, "train_loss_bc": 0.046587176620960236, "train_loss_llm": 0.0009198188781738281, "grad_norm": 0.04347357526421547, "global_step": 7590, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04048912599682808, "train_loss_bc": 0.04048912599682808, "train_loss_llm": 0.0, "grad_norm": 0.056637365370988846, "global_step": 7591, "epoch": 210, "lr": 0.001} +{"train_loss": 0.042524129152297974, "train_loss_bc": 0.042524129152297974, "train_loss_llm": 0.0, "grad_norm": 0.0674271509051323, "global_step": 7592, "epoch": 210, "lr": 0.001} +{"train_loss": 0.028528237715363503, "train_loss_bc": 0.02852698229253292, "train_loss_llm": 0.00012546777725219727, "grad_norm": 0.01990814507007599, "global_step": 7593, "epoch": 210, "lr": 0.001} +{"train_loss": 0.03894876688718796, "train_loss_bc": 0.03894876688718796, "train_loss_llm": 0.0, "grad_norm": 0.03569898381829262, "global_step": 7594, "epoch": 210, "lr": 0.001} +{"train_loss": 0.04011207617198428, "train_loss_bc": 0.04035232588648796, "train_loss_llm": 0.0, "grad_norm": 0.04366125538945198, "global_step": 7595, "epoch": 210, "lr": 0.001, "train/cumulative_reward": 159.2893538275827, "train/mean_score": 14.838609859158362, "train/success_rate": 0.3, "test/cumulative_reward": 46.03879572349963, "test/mean_score": 11.874281241204708, "test/success_rate": 0.34, "val_loss": 0.05751851946115494, "train_action_mse_error": 0.04386414214968681} +{"train_loss": 0.04149344190955162, "train_loss_bc": 0.04149344190955162, "train_loss_llm": 0.0, "grad_norm": 0.04968441650271416, "global_step": 7596, "epoch": 211, "lr": 0.001} +{"train_loss": 0.0389673076570034, "train_loss_bc": 0.0389673076570034, "train_loss_llm": 0.0, "grad_norm": 0.015004188753664494, "global_step": 7597, "epoch": 211, "lr": 0.001} +{"train_loss": 0.046202465891838074, "train_loss_bc": 0.04620150476694107, "train_loss_llm": 9.617209434509277e-05, "grad_norm": 0.02745833992958069, "global_step": 7598, "epoch": 211, "lr": 0.001} +{"train_loss": 0.0383249931037426, "train_loss_bc": 0.0383249931037426, "train_loss_llm": 0.0, "grad_norm": 0.03290431946516037, "global_step": 7599, "epoch": 211, "lr": 0.001} +{"train_loss": 0.036347053945064545, "train_loss_bc": 0.036319274455308914, "train_loss_llm": 0.0027781128883361816, "grad_norm": 0.038798391819000244, "global_step": 7600, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04786912351846695, "train_loss_bc": 0.04786912351846695, "train_loss_llm": 0.0, "grad_norm": 0.018631212413311005, "global_step": 7601, "epoch": 211, "lr": 0.001} +{"train_loss": 0.03868351876735687, "train_loss_bc": 0.03866470232605934, "train_loss_llm": 0.001881718635559082, "grad_norm": 0.0249678585678339, "global_step": 7602, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04048744961619377, "train_loss_bc": 0.0404827706515789, "train_loss_llm": 0.0004678666591644287, "grad_norm": 0.029781898483633995, "global_step": 7603, "epoch": 211, "lr": 0.001} +{"train_loss": 0.037287984043359756, "train_loss_bc": 0.03728092834353447, "train_loss_llm": 0.0007055699825286865, "grad_norm": 0.04006791114807129, "global_step": 7604, "epoch": 211, "lr": 0.001} +{"train_loss": 0.044647518545389175, "train_loss_bc": 0.044647518545389175, "train_loss_llm": 0.0, "grad_norm": 0.022119363769888878, "global_step": 7605, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04558606445789337, "train_loss_bc": 0.04558606445789337, "train_loss_llm": 0.0, "grad_norm": 0.039640720933675766, "global_step": 7606, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04018981382250786, "train_loss_bc": 0.04018981382250786, "train_loss_llm": 0.0, "grad_norm": 0.047046683728694916, "global_step": 7607, "epoch": 211, "lr": 0.001} +{"train_loss": 0.03923802077770233, "train_loss_bc": 0.03923802077770233, "train_loss_llm": 0.0, "grad_norm": 0.058653391897678375, "global_step": 7608, "epoch": 211, "lr": 0.001} +{"train_loss": 0.035823527723550797, "train_loss_bc": 0.035820189863443375, "train_loss_llm": 0.00033360719680786133, "grad_norm": 0.02459115721285343, "global_step": 7609, "epoch": 211, "lr": 0.001} +{"train_loss": 0.0365656353533268, "train_loss_bc": 0.0365656353533268, "train_loss_llm": 0.0, "grad_norm": 0.035775985568761826, "global_step": 7610, "epoch": 211, "lr": 0.001} +{"train_loss": 0.03385869413614273, "train_loss_bc": 0.03385869413614273, "train_loss_llm": 0.0, "grad_norm": 0.040864963084459305, "global_step": 7611, "epoch": 211, "lr": 0.001} +{"train_loss": 0.031956810504198074, "train_loss_bc": 0.031956810504198074, "train_loss_llm": 0.0, "grad_norm": 0.053229521960020065, "global_step": 7612, "epoch": 211, "lr": 0.001} +{"train_loss": 0.0372144915163517, "train_loss_bc": 0.0372144915163517, "train_loss_llm": 0.0, "grad_norm": 0.0154231246560812, "global_step": 7613, "epoch": 211, "lr": 0.001} +{"train_loss": 0.0332128144800663, "train_loss_bc": 0.0332128144800663, "train_loss_llm": 0.0, "grad_norm": 0.025335580110549927, "global_step": 7614, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04472813010215759, "train_loss_bc": 0.04472813010215759, "train_loss_llm": 0.0, "grad_norm": 0.03976558893918991, "global_step": 7615, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04399046301841736, "train_loss_bc": 0.04399046301841736, "train_loss_llm": 0.0, "grad_norm": 0.04969741776585579, "global_step": 7616, "epoch": 211, "lr": 0.001} +{"train_loss": 0.040802665054798126, "train_loss_bc": 0.040802665054798126, "train_loss_llm": 0.0, "grad_norm": 0.016130518168210983, "global_step": 7617, "epoch": 211, "lr": 0.001} +{"train_loss": 0.037368468940258026, "train_loss_bc": 0.037368468940258026, "train_loss_llm": 0.0, "grad_norm": 0.02217482216656208, "global_step": 7618, "epoch": 211, "lr": 0.001} +{"train_loss": 0.037477582693099976, "train_loss_bc": 0.037477582693099976, "train_loss_llm": 0.0, "grad_norm": 0.026363899931311607, "global_step": 7619, "epoch": 211, "lr": 0.001} +{"train_loss": 0.03898296132683754, "train_loss_bc": 0.03898296132683754, "train_loss_llm": 0.0, "grad_norm": 0.03323712572455406, "global_step": 7620, "epoch": 211, "lr": 0.001} +{"train_loss": 0.042605750262737274, "train_loss_bc": 0.042605750262737274, "train_loss_llm": 0.0, "grad_norm": 0.02228573150932789, "global_step": 7621, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04403619468212128, "train_loss_bc": 0.04403619468212128, "train_loss_llm": 0.0, "grad_norm": 0.03193158656358719, "global_step": 7622, "epoch": 211, "lr": 0.001} +{"train_loss": 0.034068960696458817, "train_loss_bc": 0.03405419737100601, "train_loss_llm": 0.0014764666557312012, "grad_norm": 0.04060351476073265, "global_step": 7623, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04147036373615265, "train_loss_bc": 0.04145592451095581, "train_loss_llm": 0.001444011926651001, "grad_norm": 0.046198051422834396, "global_step": 7624, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04310102388262749, "train_loss_bc": 0.04308844730257988, "train_loss_llm": 0.001257777214050293, "grad_norm": 0.02299400418996811, "global_step": 7625, "epoch": 211, "lr": 0.001} +{"train_loss": 0.046028606593608856, "train_loss_bc": 0.04602038487792015, "train_loss_llm": 0.0008220970630645752, "grad_norm": 0.03411140292882919, "global_step": 7626, "epoch": 211, "lr": 0.001} +{"train_loss": 0.0415094792842865, "train_loss_bc": 0.041502002626657486, "train_loss_llm": 0.0007477700710296631, "grad_norm": 0.04806972295045853, "global_step": 7627, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04102218523621559, "train_loss_bc": 0.04101935401558876, "train_loss_llm": 0.0002829432487487793, "grad_norm": 0.05550926923751831, "global_step": 7628, "epoch": 211, "lr": 0.001} +{"train_loss": 0.04229684919118881, "train_loss_bc": 0.04229553043842316, "train_loss_llm": 0.00013178586959838867, "grad_norm": 0.020821968093514442, "global_step": 7629, "epoch": 211, "lr": 0.001} +{"train_loss": 0.0389321893453598, "train_loss_bc": 0.038923218846321106, "train_loss_llm": 0.0008971691131591797, "grad_norm": 0.03593995049595833, "global_step": 7630, "epoch": 211, "lr": 0.001} +{"train_loss": 0.039946305048134595, "train_loss_bc": 0.0356883779168129, "train_loss_llm": 0.0, "grad_norm": 0.04961445555090904, "global_step": 7631, "epoch": 211, "lr": 0.001, "val_loss": 0.05717919394373894} +{"train_loss": 0.042510081082582474, "train_loss_bc": 0.042510081082582474, "train_loss_llm": 0.0, "grad_norm": 0.06189544498920441, "global_step": 7632, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04694501683115959, "train_loss_bc": 0.0469435416162014, "train_loss_llm": 0.0001475214958190918, "grad_norm": 0.026955466717481613, "global_step": 7633, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04659604653716087, "train_loss_bc": 0.04659604653716087, "train_loss_llm": 0.0, "grad_norm": 0.03975658491253853, "global_step": 7634, "epoch": 212, "lr": 0.001} +{"train_loss": 0.035359956324100494, "train_loss_bc": 0.035359956324100494, "train_loss_llm": 0.0, "grad_norm": 0.04571599140763283, "global_step": 7635, "epoch": 212, "lr": 0.001} +{"train_loss": 0.036018092185258865, "train_loss_bc": 0.036018092185258865, "train_loss_llm": 0.0, "grad_norm": 0.05156479403376579, "global_step": 7636, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03613411635160446, "train_loss_bc": 0.03613411635160446, "train_loss_llm": 0.0, "grad_norm": 0.014358781278133392, "global_step": 7637, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03601294383406639, "train_loss_bc": 0.036011289805173874, "train_loss_llm": 0.0001653730869293213, "grad_norm": 0.020820287987589836, "global_step": 7638, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04249020293354988, "train_loss_bc": 0.042485445737838745, "train_loss_llm": 0.0004757046699523926, "grad_norm": 0.02893921732902527, "global_step": 7639, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04484500363469124, "train_loss_bc": 0.04484500363469124, "train_loss_llm": 0.0, "grad_norm": 0.03647661581635475, "global_step": 7640, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04731842875480652, "train_loss_bc": 0.047318387776613235, "train_loss_llm": 4.023313522338867e-06, "grad_norm": 0.023929474875330925, "global_step": 7641, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03820767626166344, "train_loss_bc": 0.03820767626166344, "train_loss_llm": 0.0, "grad_norm": 0.03178592398762703, "global_step": 7642, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03578938543796539, "train_loss_bc": 0.035788822919130325, "train_loss_llm": 5.6415796279907227e-05, "grad_norm": 0.043969716876745224, "global_step": 7643, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03367941081523895, "train_loss_bc": 0.03367673605680466, "train_loss_llm": 0.00026744604110717773, "grad_norm": 0.05746583640575409, "global_step": 7644, "epoch": 212, "lr": 0.001} +{"train_loss": 0.040034905076026917, "train_loss_bc": 0.040034905076026917, "train_loss_llm": 0.0, "grad_norm": 0.021492810919880867, "global_step": 7645, "epoch": 212, "lr": 0.001} +{"train_loss": 0.045638397336006165, "train_loss_bc": 0.04561273381114006, "train_loss_llm": 0.0025664567947387695, "grad_norm": 0.03629012778401375, "global_step": 7646, "epoch": 212, "lr": 0.001} +{"train_loss": 0.043470676988363266, "train_loss_bc": 0.04347045719623566, "train_loss_llm": 2.193450927734375e-05, "grad_norm": 0.05516236647963524, "global_step": 7647, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04600934684276581, "train_loss_bc": 0.04600934684276581, "train_loss_llm": 0.0, "grad_norm": 0.07537012547254562, "global_step": 7648, "epoch": 212, "lr": 0.001} +{"train_loss": 0.0364101268351078, "train_loss_bc": 0.0364101268351078, "train_loss_llm": 0.0, "grad_norm": 0.021977892145514488, "global_step": 7649, "epoch": 212, "lr": 0.001} +{"train_loss": 0.040751975029706955, "train_loss_bc": 0.040751975029706955, "train_loss_llm": 0.0, "grad_norm": 0.03389449045062065, "global_step": 7650, "epoch": 212, "lr": 0.001} +{"train_loss": 0.046248048543930054, "train_loss_bc": 0.046248048543930054, "train_loss_llm": 0.0, "grad_norm": 0.056357380002737045, "global_step": 7651, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04277803376317024, "train_loss_bc": 0.04277803376317024, "train_loss_llm": 0.0, "grad_norm": 0.07382043451070786, "global_step": 7652, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04191369190812111, "train_loss_bc": 0.041875213384628296, "train_loss_llm": 0.003847867250442505, "grad_norm": 0.016410036012530327, "global_step": 7653, "epoch": 212, "lr": 0.001} +{"train_loss": 0.040040355175733566, "train_loss_bc": 0.040040355175733566, "train_loss_llm": 0.0, "grad_norm": 0.025178350508213043, "global_step": 7654, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03419674560427666, "train_loss_bc": 0.03419674560427666, "train_loss_llm": 0.0, "grad_norm": 0.02970830164849758, "global_step": 7655, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03601875528693199, "train_loss_bc": 0.03601875528693199, "train_loss_llm": 0.0, "grad_norm": 0.04068044573068619, "global_step": 7656, "epoch": 212, "lr": 0.001} +{"train_loss": 0.041056156158447266, "train_loss_bc": 0.041056156158447266, "train_loss_llm": 0.0, "grad_norm": 0.0224355086684227, "global_step": 7657, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04082339629530907, "train_loss_bc": 0.04080212488770485, "train_loss_llm": 0.0021269619464874268, "grad_norm": 0.034514255821704865, "global_step": 7658, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04068027436733246, "train_loss_bc": 0.04066545516252518, "train_loss_llm": 0.0014819502830505371, "grad_norm": 0.045731671154499054, "global_step": 7659, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03406798094511032, "train_loss_bc": 0.034040842205286026, "train_loss_llm": 0.0027137398719787598, "grad_norm": 0.05589371547102928, "global_step": 7660, "epoch": 212, "lr": 0.001} +{"train_loss": 0.04595702514052391, "train_loss_bc": 0.04595010355114937, "train_loss_llm": 0.0006921887397766113, "grad_norm": 0.01558404229581356, "global_step": 7661, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03801469877362251, "train_loss_bc": 0.03801469877362251, "train_loss_llm": 0.0, "grad_norm": 0.022237028926610947, "global_step": 7662, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03854171559214592, "train_loss_bc": 0.03853486850857735, "train_loss_llm": 0.0006847083568572998, "grad_norm": 0.04047178849577904, "global_step": 7663, "epoch": 212, "lr": 0.001} +{"train_loss": 0.043276261538267136, "train_loss_bc": 0.04324536770582199, "train_loss_llm": 0.003089427947998047, "grad_norm": 0.050456274300813675, "global_step": 7664, "epoch": 212, "lr": 0.001} +{"train_loss": 0.03228598088026047, "train_loss_bc": 0.03228598088026047, "train_loss_llm": 0.0, "grad_norm": 0.023591984063386917, "global_step": 7665, "epoch": 212, "lr": 0.001} +{"train_loss": 0.038590241223573685, "train_loss_bc": 0.038590241223573685, "train_loss_llm": 0.0, "grad_norm": 0.03302059695124626, "global_step": 7666, "epoch": 212, "lr": 0.001} +{"train_loss": 0.040001496051748596, "train_loss_bc": 0.03134270757436752, "train_loss_llm": 0.0, "grad_norm": 0.05149851366877556, "global_step": 7667, "epoch": 212, "lr": 0.001, "val_loss": 0.05754467099905014} +{"train_loss": 0.044985655695199966, "train_loss_bc": 0.04495420679450035, "train_loss_llm": 0.0031447112560272217, "grad_norm": 0.0538540855050087, "global_step": 7668, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03331487253308296, "train_loss_bc": 0.03330221399664879, "train_loss_llm": 0.001265883445739746, "grad_norm": 0.017374379560351372, "global_step": 7669, "epoch": 213, "lr": 0.001} +{"train_loss": 0.04322546347975731, "train_loss_bc": 0.0432233102619648, "train_loss_llm": 0.00021535158157348633, "grad_norm": 0.030101042240858078, "global_step": 7670, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03641173243522644, "train_loss_bc": 0.03641173243522644, "train_loss_llm": 0.0, "grad_norm": 0.03113081119954586, "global_step": 7671, "epoch": 213, "lr": 0.001} +{"train_loss": 0.042316559702157974, "train_loss_bc": 0.042316559702157974, "train_loss_llm": 0.0, "grad_norm": 0.040864940732717514, "global_step": 7672, "epoch": 213, "lr": 0.001} +{"train_loss": 0.05020337551832199, "train_loss_bc": 0.05020337551832199, "train_loss_llm": 0.0, "grad_norm": 0.020530181005597115, "global_step": 7673, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03800758719444275, "train_loss_bc": 0.038004543632268906, "train_loss_llm": 0.0003044605255126953, "grad_norm": 0.026176417246460915, "global_step": 7674, "epoch": 213, "lr": 0.001} +{"train_loss": 0.04554484039545059, "train_loss_bc": 0.045540839433670044, "train_loss_llm": 0.00040012598037719727, "grad_norm": 0.033003244549036026, "global_step": 7675, "epoch": 213, "lr": 0.001} +{"train_loss": 0.0421360619366169, "train_loss_bc": 0.0421360619366169, "train_loss_llm": 0.0, "grad_norm": 0.03853560611605644, "global_step": 7676, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03823588043451309, "train_loss_bc": 0.03823474049568176, "train_loss_llm": 0.00011414289474487305, "grad_norm": 0.016289975494146347, "global_step": 7677, "epoch": 213, "lr": 0.001} +{"train_loss": 0.039475612342357635, "train_loss_bc": 0.039475612342357635, "train_loss_llm": 0.0, "grad_norm": 0.025524018332362175, "global_step": 7678, "epoch": 213, "lr": 0.001} +{"train_loss": 0.04305393993854523, "train_loss_bc": 0.04305393993854523, "train_loss_llm": 0.0, "grad_norm": 0.04160439968109131, "global_step": 7679, "epoch": 213, "lr": 0.001} +{"train_loss": 0.04020320624113083, "train_loss_bc": 0.04020320624113083, "train_loss_llm": 0.0, "grad_norm": 0.04968881979584694, "global_step": 7680, "epoch": 213, "lr": 0.001} +{"train_loss": 0.049683474004268646, "train_loss_bc": 0.049683474004268646, "train_loss_llm": 0.0, "grad_norm": 0.02171955443918705, "global_step": 7681, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03961699828505516, "train_loss_bc": 0.03961699828505516, "train_loss_llm": 0.0, "grad_norm": 0.030934955924749374, "global_step": 7682, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03631552681326866, "train_loss_bc": 0.03631552681326866, "train_loss_llm": 0.0, "grad_norm": 0.04147731140255928, "global_step": 7683, "epoch": 213, "lr": 0.001} +{"train_loss": 0.04408333823084831, "train_loss_bc": 0.044074300676584244, "train_loss_llm": 0.0009037256240844727, "grad_norm": 0.048343878239393234, "global_step": 7684, "epoch": 213, "lr": 0.001} +{"train_loss": 0.04742715507745743, "train_loss_bc": 0.04742715507745743, "train_loss_llm": 0.0, "grad_norm": 0.02369324490427971, "global_step": 7685, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03918324410915375, "train_loss_bc": 0.03918324410915375, "train_loss_llm": 0.0, "grad_norm": 0.03673822432756424, "global_step": 7686, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03275633603334427, "train_loss_bc": 0.03275633603334427, "train_loss_llm": 0.0, "grad_norm": 0.042404718697071075, "global_step": 7687, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03947685286402702, "train_loss_bc": 0.03947685286402702, "train_loss_llm": 0.0, "grad_norm": 0.04986041784286499, "global_step": 7688, "epoch": 213, "lr": 0.001} +{"train_loss": 0.0373213104903698, "train_loss_bc": 0.0373213104903698, "train_loss_llm": 0.0, "grad_norm": 0.017738277092576027, "global_step": 7689, "epoch": 213, "lr": 0.001} +{"train_loss": 0.04177096486091614, "train_loss_bc": 0.04177096486091614, "train_loss_llm": 0.0, "grad_norm": 0.02570166066288948, "global_step": 7690, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03402596712112427, "train_loss_bc": 0.03402596712112427, "train_loss_llm": 0.0, "grad_norm": 0.031926508992910385, "global_step": 7691, "epoch": 213, "lr": 0.001} +{"train_loss": 0.04540430009365082, "train_loss_bc": 0.04540430009365082, "train_loss_llm": 0.0, "grad_norm": 0.04165475815534592, "global_step": 7692, "epoch": 213, "lr": 0.001} +{"train_loss": 0.04675920307636261, "train_loss_bc": 0.04675920307636261, "train_loss_llm": 0.0, "grad_norm": 0.01849469169974327, "global_step": 7693, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03130275756120682, "train_loss_bc": 0.03127764165401459, "train_loss_llm": 0.0025115907192230225, "grad_norm": 0.02531757764518261, "global_step": 7694, "epoch": 213, "lr": 0.001} +{"train_loss": 0.02947334200143814, "train_loss_bc": 0.02947334200143814, "train_loss_llm": 0.0, "grad_norm": 0.03460254892706871, "global_step": 7695, "epoch": 213, "lr": 0.001} +{"train_loss": 0.038629867136478424, "train_loss_bc": 0.038629867136478424, "train_loss_llm": 0.0, "grad_norm": 0.04425109177827835, "global_step": 7696, "epoch": 213, "lr": 0.001} +{"train_loss": 0.033653683960437775, "train_loss_bc": 0.03365332633256912, "train_loss_llm": 3.591179847717285e-05, "grad_norm": 0.017431197687983513, "global_step": 7697, "epoch": 213, "lr": 0.001} +{"train_loss": 0.043137598782777786, "train_loss_bc": 0.043137598782777786, "train_loss_llm": 0.0, "grad_norm": 0.02753840945661068, "global_step": 7698, "epoch": 213, "lr": 0.001} +{"train_loss": 0.04333070293068886, "train_loss_bc": 0.04333070293068886, "train_loss_llm": 0.0, "grad_norm": 0.03711910545825958, "global_step": 7699, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03518936038017273, "train_loss_bc": 0.03518936038017273, "train_loss_llm": 0.0, "grad_norm": 0.04616581276059151, "global_step": 7700, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03503865748643875, "train_loss_bc": 0.03503865748643875, "train_loss_llm": 0.0, "grad_norm": 0.01606941409409046, "global_step": 7701, "epoch": 213, "lr": 0.001} +{"train_loss": 0.040696945041418076, "train_loss_bc": 0.040696945041418076, "train_loss_llm": 0.0, "grad_norm": 0.02597617544233799, "global_step": 7702, "epoch": 213, "lr": 0.001} +{"train_loss": 0.03977608075365424, "train_loss_bc": 0.030543703585863113, "train_loss_llm": 0.0002828836441040039, "grad_norm": 0.034464675933122635, "global_step": 7703, "epoch": 213, "lr": 0.001, "val_loss": 0.05690193921327591} +{"train_loss": 0.035511888563632965, "train_loss_bc": 0.03548875078558922, "train_loss_llm": 0.002313673496246338, "grad_norm": 0.037984784692525864, "global_step": 7704, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04133279621601105, "train_loss_bc": 0.04133279621601105, "train_loss_llm": 0.0, "grad_norm": 0.017358170822262764, "global_step": 7705, "epoch": 214, "lr": 0.001} +{"train_loss": 0.03773627430200577, "train_loss_bc": 0.03773627430200577, "train_loss_llm": 0.0, "grad_norm": 0.02704521454870701, "global_step": 7706, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04084037244319916, "train_loss_bc": 0.04083959385752678, "train_loss_llm": 7.775425910949707e-05, "grad_norm": 0.032249316573143005, "global_step": 7707, "epoch": 214, "lr": 0.001} +{"train_loss": 0.044454775750637054, "train_loss_bc": 0.044454775750637054, "train_loss_llm": 0.0, "grad_norm": 0.0413653738796711, "global_step": 7708, "epoch": 214, "lr": 0.001} +{"train_loss": 0.039588168263435364, "train_loss_bc": 0.039588168263435364, "train_loss_llm": 0.0, "grad_norm": 0.020234422758221626, "global_step": 7709, "epoch": 214, "lr": 0.001} +{"train_loss": 0.040661610662937164, "train_loss_bc": 0.04066098853945732, "train_loss_llm": 6.237626075744629e-05, "grad_norm": 0.028889235109090805, "global_step": 7710, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04331502690911293, "train_loss_bc": 0.043295927345752716, "train_loss_llm": 0.0019100010395050049, "grad_norm": 0.04436106234788895, "global_step": 7711, "epoch": 214, "lr": 0.001} +{"train_loss": 0.041983552277088165, "train_loss_bc": 0.04196834936738014, "train_loss_llm": 0.0015202760696411133, "grad_norm": 0.05629735067486763, "global_step": 7712, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04252436384558678, "train_loss_bc": 0.04252436384558678, "train_loss_llm": 0.0, "grad_norm": 0.02140241116285324, "global_step": 7713, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04519880190491676, "train_loss_bc": 0.04519880190491676, "train_loss_llm": 0.0, "grad_norm": 0.03830486536026001, "global_step": 7714, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04225034639239311, "train_loss_bc": 0.04225034639239311, "train_loss_llm": 0.0, "grad_norm": 0.0547771193087101, "global_step": 7715, "epoch": 214, "lr": 0.001} +{"train_loss": 0.038334306329488754, "train_loss_bc": 0.038325436413288116, "train_loss_llm": 0.0008869767189025879, "grad_norm": 0.057136449962854385, "global_step": 7716, "epoch": 214, "lr": 0.001} +{"train_loss": 0.041883327066898346, "train_loss_bc": 0.041883327066898346, "train_loss_llm": 0.0, "grad_norm": 0.018490493297576904, "global_step": 7717, "epoch": 214, "lr": 0.001} +{"train_loss": 0.03915277495980263, "train_loss_bc": 0.03915277495980263, "train_loss_llm": 0.0, "grad_norm": 0.03143376111984253, "global_step": 7718, "epoch": 214, "lr": 0.001} +{"train_loss": 0.03639807179570198, "train_loss_bc": 0.03639807179570198, "train_loss_llm": 0.0, "grad_norm": 0.04035136103630066, "global_step": 7719, "epoch": 214, "lr": 0.001} +{"train_loss": 0.03674633800983429, "train_loss_bc": 0.03674633800983429, "train_loss_llm": 0.0, "grad_norm": 0.05149884894490242, "global_step": 7720, "epoch": 214, "lr": 0.001} +{"train_loss": 0.03805939480662346, "train_loss_bc": 0.03805939480662346, "train_loss_llm": 0.0, "grad_norm": 0.016526076942682266, "global_step": 7721, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04144859313964844, "train_loss_bc": 0.0414416566491127, "train_loss_llm": 0.0006937980651855469, "grad_norm": 0.023066122084856033, "global_step": 7722, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04139450192451477, "train_loss_bc": 0.04136250913143158, "train_loss_llm": 0.003199160099029541, "grad_norm": 0.029205944389104843, "global_step": 7723, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04591239616274834, "train_loss_bc": 0.04591239616274834, "train_loss_llm": 0.0, "grad_norm": 0.042303916066884995, "global_step": 7724, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04307207092642784, "train_loss_bc": 0.04305989295244217, "train_loss_llm": 0.0012176334857940674, "grad_norm": 0.01641337387263775, "global_step": 7725, "epoch": 214, "lr": 0.001} +{"train_loss": 0.042141593992710114, "train_loss_bc": 0.04212789982557297, "train_loss_llm": 0.0013692677021026611, "grad_norm": 0.028336314484477043, "global_step": 7726, "epoch": 214, "lr": 0.001} +{"train_loss": 0.038714971393346786, "train_loss_bc": 0.038714971393346786, "train_loss_llm": 0.0, "grad_norm": 0.03731453791260719, "global_step": 7727, "epoch": 214, "lr": 0.001} +{"train_loss": 0.041293419897556305, "train_loss_bc": 0.041293419897556305, "train_loss_llm": 0.0, "grad_norm": 0.03773845359683037, "global_step": 7728, "epoch": 214, "lr": 0.001} +{"train_loss": 0.0348953939974308, "train_loss_bc": 0.0348953939974308, "train_loss_llm": 0.0, "grad_norm": 0.017652379348874092, "global_step": 7729, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04197271168231964, "train_loss_bc": 0.04197271168231964, "train_loss_llm": 0.0, "grad_norm": 0.026015201583504677, "global_step": 7730, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04395701363682747, "train_loss_bc": 0.04395701363682747, "train_loss_llm": 0.0, "grad_norm": 0.03432966768741608, "global_step": 7731, "epoch": 214, "lr": 0.001} +{"train_loss": 0.03773238882422447, "train_loss_bc": 0.03773238882422447, "train_loss_llm": 0.0, "grad_norm": 0.03790535405278206, "global_step": 7732, "epoch": 214, "lr": 0.001} +{"train_loss": 0.035152703523635864, "train_loss_bc": 0.03513430804014206, "train_loss_llm": 0.00183945894241333, "grad_norm": 0.015148039907217026, "global_step": 7733, "epoch": 214, "lr": 0.001} +{"train_loss": 0.03107871487736702, "train_loss_bc": 0.03107871487736702, "train_loss_llm": 0.0, "grad_norm": 0.020293010398745537, "global_step": 7734, "epoch": 214, "lr": 0.001} +{"train_loss": 0.03468623384833336, "train_loss_bc": 0.03468623384833336, "train_loss_llm": 0.0, "grad_norm": 0.024948259815573692, "global_step": 7735, "epoch": 214, "lr": 0.001} +{"train_loss": 0.03882651403546333, "train_loss_bc": 0.03882651403546333, "train_loss_llm": 0.0, "grad_norm": 0.0355909988284111, "global_step": 7736, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04084012657403946, "train_loss_bc": 0.04084012657403946, "train_loss_llm": 0.0, "grad_norm": 0.02103046141564846, "global_step": 7737, "epoch": 214, "lr": 0.001} +{"train_loss": 0.04233725368976593, "train_loss_bc": 0.04233725368976593, "train_loss_llm": 0.0, "grad_norm": 0.028420496731996536, "global_step": 7738, "epoch": 214, "lr": 0.001} +{"train_loss": 0.03975504491892126, "train_loss_bc": 0.02975282445549965, "train_loss_llm": 0.0, "grad_norm": 0.039256200194358826, "global_step": 7739, "epoch": 214, "lr": 0.001, "val_loss": 0.05755995959043503} +{"train_loss": 0.04014414921402931, "train_loss_bc": 0.04014414921402931, "train_loss_llm": 0.0, "grad_norm": 0.046604666858911514, "global_step": 7740, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03982115164399147, "train_loss_bc": 0.03982115164399147, "train_loss_llm": 0.0, "grad_norm": 0.018431544303894043, "global_step": 7741, "epoch": 215, "lr": 0.001} +{"train_loss": 0.04220554232597351, "train_loss_bc": 0.04220220446586609, "train_loss_llm": 0.0003338456153869629, "grad_norm": 0.023803651332855225, "global_step": 7742, "epoch": 215, "lr": 0.001} +{"train_loss": 0.0330888107419014, "train_loss_bc": 0.0330888107419014, "train_loss_llm": 0.0, "grad_norm": 0.031076353043317795, "global_step": 7743, "epoch": 215, "lr": 0.001} +{"train_loss": 0.05268999934196472, "train_loss_bc": 0.05268999934196472, "train_loss_llm": 0.0, "grad_norm": 0.04149375855922699, "global_step": 7744, "epoch": 215, "lr": 0.001} +{"train_loss": 0.047568969428539276, "train_loss_bc": 0.04756471514701843, "train_loss_llm": 0.0004254281520843506, "grad_norm": 0.018702834844589233, "global_step": 7745, "epoch": 215, "lr": 0.001} +{"train_loss": 0.037758659571409225, "train_loss_bc": 0.03775482624769211, "train_loss_llm": 0.0003832876682281494, "grad_norm": 0.02922365441918373, "global_step": 7746, "epoch": 215, "lr": 0.001} +{"train_loss": 0.041976198554039, "train_loss_bc": 0.041976198554039, "train_loss_llm": 0.0, "grad_norm": 0.03827739506959915, "global_step": 7747, "epoch": 215, "lr": 0.001} +{"train_loss": 0.04304680600762367, "train_loss_bc": 0.04303082078695297, "train_loss_llm": 0.001598358154296875, "grad_norm": 0.0457017607986927, "global_step": 7748, "epoch": 215, "lr": 0.001} +{"train_loss": 0.043799035251140594, "train_loss_bc": 0.04378701001405716, "train_loss_llm": 0.0012025535106658936, "grad_norm": 0.019929824396967888, "global_step": 7749, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03344891220331192, "train_loss_bc": 0.03344891220331192, "train_loss_llm": 0.0, "grad_norm": 0.028737977147102356, "global_step": 7750, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03859381005167961, "train_loss_bc": 0.03859381005167961, "train_loss_llm": 0.0, "grad_norm": 0.043327510356903076, "global_step": 7751, "epoch": 215, "lr": 0.001} +{"train_loss": 0.038186412304639816, "train_loss_bc": 0.03817329183220863, "train_loss_llm": 0.001312166452407837, "grad_norm": 0.043601468205451965, "global_step": 7752, "epoch": 215, "lr": 0.001} +{"train_loss": 0.034653861075639725, "train_loss_bc": 0.034653861075639725, "train_loss_llm": 0.0, "grad_norm": 0.019645636901259422, "global_step": 7753, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03756294026970863, "train_loss_bc": 0.03752691298723221, "train_loss_llm": 0.0036027133464813232, "grad_norm": 0.029050836339592934, "global_step": 7754, "epoch": 215, "lr": 0.001} +{"train_loss": 0.038544803857803345, "train_loss_bc": 0.038544803857803345, "train_loss_llm": 0.0, "grad_norm": 0.04185350611805916, "global_step": 7755, "epoch": 215, "lr": 0.001} +{"train_loss": 0.04601643234491348, "train_loss_bc": 0.04601643234491348, "train_loss_llm": 0.0, "grad_norm": 0.05352608487010002, "global_step": 7756, "epoch": 215, "lr": 0.001} +{"train_loss": 0.037769708782434464, "train_loss_bc": 0.037769708782434464, "train_loss_llm": 0.0, "grad_norm": 0.01755097322165966, "global_step": 7757, "epoch": 215, "lr": 0.001} +{"train_loss": 0.035213273018598557, "train_loss_bc": 0.03520875424146652, "train_loss_llm": 0.00045171380043029785, "grad_norm": 0.027705611661076546, "global_step": 7758, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03608998656272888, "train_loss_bc": 0.03608998656272888, "train_loss_llm": 0.0, "grad_norm": 0.03655500337481499, "global_step": 7759, "epoch": 215, "lr": 0.001} +{"train_loss": 0.041285041719675064, "train_loss_bc": 0.041285041719675064, "train_loss_llm": 0.0, "grad_norm": 0.050295185297727585, "global_step": 7760, "epoch": 215, "lr": 0.001} +{"train_loss": 0.04057561233639717, "train_loss_bc": 0.040559906512498856, "train_loss_llm": 0.0015707015991210938, "grad_norm": 0.014799053780734539, "global_step": 7761, "epoch": 215, "lr": 0.001} +{"train_loss": 0.0402381531894207, "train_loss_bc": 0.0402381531894207, "train_loss_llm": 0.0, "grad_norm": 0.02756061591207981, "global_step": 7762, "epoch": 215, "lr": 0.001} +{"train_loss": 0.04186852276325226, "train_loss_bc": 0.04186852276325226, "train_loss_llm": 0.0, "grad_norm": 0.03551826253533363, "global_step": 7763, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03569183871150017, "train_loss_bc": 0.03569183871150017, "train_loss_llm": 0.0, "grad_norm": 0.04161466285586357, "global_step": 7764, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03889993205666542, "train_loss_bc": 0.038895849138498306, "train_loss_llm": 0.00040847063064575195, "grad_norm": 0.025342660024762154, "global_step": 7765, "epoch": 215, "lr": 0.001} +{"train_loss": 0.04552155360579491, "train_loss_bc": 0.04552155360579491, "train_loss_llm": 0.0, "grad_norm": 0.04770774394273758, "global_step": 7766, "epoch": 215, "lr": 0.001} +{"train_loss": 0.04043058305978775, "train_loss_bc": 0.04037942737340927, "train_loss_llm": 0.005115538835525513, "grad_norm": 0.053968749940395355, "global_step": 7767, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03223862498998642, "train_loss_bc": 0.03223862498998642, "train_loss_llm": 0.0, "grad_norm": 0.06440500169992447, "global_step": 7768, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03587900102138519, "train_loss_bc": 0.03587900102138519, "train_loss_llm": 0.0, "grad_norm": 0.01608560048043728, "global_step": 7769, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03946932405233383, "train_loss_bc": 0.03946932405233383, "train_loss_llm": 0.0, "grad_norm": 0.027902496978640556, "global_step": 7770, "epoch": 215, "lr": 0.001} +{"train_loss": 0.04029424488544464, "train_loss_bc": 0.040277473628520966, "train_loss_llm": 0.0016770660877227783, "grad_norm": 0.040608976036310196, "global_step": 7771, "epoch": 215, "lr": 0.001} +{"train_loss": 0.0405777171254158, "train_loss_bc": 0.0405777171254158, "train_loss_llm": 0.0, "grad_norm": 0.04804607108235359, "global_step": 7772, "epoch": 215, "lr": 0.001} +{"train_loss": 0.04251283407211304, "train_loss_bc": 0.04251283407211304, "train_loss_llm": 0.0, "grad_norm": 0.026135003194212914, "global_step": 7773, "epoch": 215, "lr": 0.001} +{"train_loss": 0.042069245129823685, "train_loss_bc": 0.042069245129823685, "train_loss_llm": 0.0, "grad_norm": 0.03864482417702675, "global_step": 7774, "epoch": 215, "lr": 0.001} +{"train_loss": 0.03991562728252676, "train_loss_bc": 0.04123089089989662, "train_loss_llm": 0.0, "grad_norm": 0.04610869288444519, "global_step": 7775, "epoch": 215, "lr": 0.001, "train/cumulative_reward": 136.729405990664, "train/mean_score": 14.724049651724064, "train/success_rate": 0.3, "test/cumulative_reward": 24.80086414184908, "test/mean_score": 3.4390112028839552, "test/success_rate": 0.08, "val_loss": 0.057594362646341324, "train_action_mse_error": 0.043427158147096634} +{"train_loss": 0.03833267465233803, "train_loss_bc": 0.03832952678203583, "train_loss_llm": 0.00031495094299316406, "grad_norm": 0.05774204060435295, "global_step": 7776, "epoch": 216, "lr": 0.001} +{"train_loss": 0.038557395339012146, "train_loss_bc": 0.038557395339012146, "train_loss_llm": 0.0, "grad_norm": 0.01660989411175251, "global_step": 7777, "epoch": 216, "lr": 0.001} +{"train_loss": 0.0403369776904583, "train_loss_bc": 0.0403369776904583, "train_loss_llm": 0.0, "grad_norm": 0.023008152842521667, "global_step": 7778, "epoch": 216, "lr": 0.001} +{"train_loss": 0.04010888189077377, "train_loss_bc": 0.040094826370477676, "train_loss_llm": 0.0014055967330932617, "grad_norm": 0.02898365445435047, "global_step": 7779, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03807205706834793, "train_loss_bc": 0.03807205706834793, "train_loss_llm": 0.0, "grad_norm": 0.031212249770760536, "global_step": 7780, "epoch": 216, "lr": 0.001} +{"train_loss": 0.036894068121910095, "train_loss_bc": 0.036894068121910095, "train_loss_llm": 0.0, "grad_norm": 0.02265729010105133, "global_step": 7781, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03894762694835663, "train_loss_bc": 0.03894762694835663, "train_loss_llm": 0.0, "grad_norm": 0.03230417147278786, "global_step": 7782, "epoch": 216, "lr": 0.001} +{"train_loss": 0.034326743334531784, "train_loss_bc": 0.03432435542345047, "train_loss_llm": 0.00023889541625976562, "grad_norm": 0.029525119811296463, "global_step": 7783, "epoch": 216, "lr": 0.001} +{"train_loss": 0.038562167435884476, "train_loss_bc": 0.038558147847652435, "train_loss_llm": 0.0004018247127532959, "grad_norm": 0.038731131702661514, "global_step": 7784, "epoch": 216, "lr": 0.001} +{"train_loss": 0.038975559175014496, "train_loss_bc": 0.038975559175014496, "train_loss_llm": 0.0, "grad_norm": 0.016592923551797867, "global_step": 7785, "epoch": 216, "lr": 0.001} +{"train_loss": 0.037301257252693176, "train_loss_bc": 0.037297289818525314, "train_loss_llm": 0.0003968477249145508, "grad_norm": 0.025790784507989883, "global_step": 7786, "epoch": 216, "lr": 0.001} +{"train_loss": 0.04762483015656471, "train_loss_bc": 0.04762483015656471, "train_loss_llm": 0.0, "grad_norm": 0.03398189693689346, "global_step": 7787, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03304505720734596, "train_loss_bc": 0.03304505720734596, "train_loss_llm": 0.0, "grad_norm": 0.03710422292351723, "global_step": 7788, "epoch": 216, "lr": 0.001} +{"train_loss": 0.04843627288937569, "train_loss_bc": 0.04843627288937569, "train_loss_llm": 0.0, "grad_norm": 0.023986393585801125, "global_step": 7789, "epoch": 216, "lr": 0.001} +{"train_loss": 0.037007447332143784, "train_loss_bc": 0.03698010370135307, "train_loss_llm": 0.002734541893005371, "grad_norm": 0.025796152651309967, "global_step": 7790, "epoch": 216, "lr": 0.001} +{"train_loss": 0.04096087068319321, "train_loss_bc": 0.04096087068319321, "train_loss_llm": 0.0, "grad_norm": 0.03725934028625488, "global_step": 7791, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03754492849111557, "train_loss_bc": 0.03754492849111557, "train_loss_llm": 0.0, "grad_norm": 0.04734482988715172, "global_step": 7792, "epoch": 216, "lr": 0.001} +{"train_loss": 0.035833876579999924, "train_loss_bc": 0.035833876579999924, "train_loss_llm": 0.0, "grad_norm": 0.022343020886182785, "global_step": 7793, "epoch": 216, "lr": 0.001} +{"train_loss": 0.04593440145254135, "train_loss_bc": 0.045914795249700546, "train_loss_llm": 0.00196075439453125, "grad_norm": 0.031231233850121498, "global_step": 7794, "epoch": 216, "lr": 0.001} +{"train_loss": 0.032830219715833664, "train_loss_bc": 0.032830219715833664, "train_loss_llm": 0.0, "grad_norm": 0.03637941554188728, "global_step": 7795, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03452496975660324, "train_loss_bc": 0.03452496975660324, "train_loss_llm": 0.0, "grad_norm": 0.045351944863796234, "global_step": 7796, "epoch": 216, "lr": 0.001} +{"train_loss": 0.044772326946258545, "train_loss_bc": 0.044752154499292374, "train_loss_llm": 0.0020172297954559326, "grad_norm": 0.015570652671158314, "global_step": 7797, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03643495589494705, "train_loss_bc": 0.03643004223704338, "train_loss_llm": 0.0004913806915283203, "grad_norm": 0.0232843104749918, "global_step": 7798, "epoch": 216, "lr": 0.001} +{"train_loss": 0.04097072035074234, "train_loss_bc": 0.04097072035074234, "train_loss_llm": 0.0, "grad_norm": 0.030520860105752945, "global_step": 7799, "epoch": 216, "lr": 0.001} +{"train_loss": 0.039918191730976105, "train_loss_bc": 0.039897285401821136, "train_loss_llm": 0.0020906031131744385, "grad_norm": 0.03414575383067131, "global_step": 7800, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03994704410433769, "train_loss_bc": 0.03994704410433769, "train_loss_llm": 0.0, "grad_norm": 0.017313674092292786, "global_step": 7801, "epoch": 216, "lr": 0.001} +{"train_loss": 0.04544999450445175, "train_loss_bc": 0.04541756585240364, "train_loss_llm": 0.00324288010597229, "grad_norm": 0.03327355906367302, "global_step": 7802, "epoch": 216, "lr": 0.001} +{"train_loss": 0.04874637350440025, "train_loss_bc": 0.04874637350440025, "train_loss_llm": 0.0, "grad_norm": 0.042184989899396896, "global_step": 7803, "epoch": 216, "lr": 0.001} +{"train_loss": 0.038295093923807144, "train_loss_bc": 0.038288913667201996, "train_loss_llm": 0.0006181895732879639, "grad_norm": 0.047629214823246, "global_step": 7804, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03810258209705353, "train_loss_bc": 0.03810258209705353, "train_loss_llm": 0.0, "grad_norm": 0.01985510252416134, "global_step": 7805, "epoch": 216, "lr": 0.001} +{"train_loss": 0.04541261866688728, "train_loss_bc": 0.04541261866688728, "train_loss_llm": 0.0, "grad_norm": 0.026737790554761887, "global_step": 7806, "epoch": 216, "lr": 0.001} +{"train_loss": 0.04417908191680908, "train_loss_bc": 0.04417001083493233, "train_loss_llm": 0.0009071230888366699, "grad_norm": 0.034529052674770355, "global_step": 7807, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03995838761329651, "train_loss_bc": 0.03995838761329651, "train_loss_llm": 0.0, "grad_norm": 0.04426059499382973, "global_step": 7808, "epoch": 216, "lr": 0.001} +{"train_loss": 0.035607822239398956, "train_loss_bc": 0.03559928014874458, "train_loss_llm": 0.00085410475730896, "grad_norm": 0.017289455980062485, "global_step": 7809, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03687208518385887, "train_loss_bc": 0.03687208518385887, "train_loss_llm": 0.0, "grad_norm": 0.027083998546004295, "global_step": 7810, "epoch": 216, "lr": 0.001} +{"train_loss": 0.03978477946172158, "train_loss_bc": 0.04341920092701912, "train_loss_llm": 0.0007298588752746582, "grad_norm": 0.03749139606952667, "global_step": 7811, "epoch": 216, "lr": 0.001, "val_loss": 0.057224929332733154} +{"train_loss": 0.045971084386110306, "train_loss_bc": 0.045971084386110306, "train_loss_llm": 0.0, "grad_norm": 0.04606590420007706, "global_step": 7812, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03709701821208, "train_loss_bc": 0.037095073610544205, "train_loss_llm": 0.000194549560546875, "grad_norm": 0.018113603815436363, "global_step": 7813, "epoch": 217, "lr": 0.001} +{"train_loss": 0.04577987268567085, "train_loss_bc": 0.04577987268567085, "train_loss_llm": 0.0, "grad_norm": 0.0309455506503582, "global_step": 7814, "epoch": 217, "lr": 0.001} +{"train_loss": 0.04513474553823471, "train_loss_bc": 0.04513474553823471, "train_loss_llm": 0.0, "grad_norm": 0.0402325876057148, "global_step": 7815, "epoch": 217, "lr": 0.001} +{"train_loss": 0.033432770520448685, "train_loss_bc": 0.033432770520448685, "train_loss_llm": 0.0, "grad_norm": 0.03958075866103172, "global_step": 7816, "epoch": 217, "lr": 0.001} +{"train_loss": 0.04120207205414772, "train_loss_bc": 0.04120174050331116, "train_loss_llm": 3.331899642944336e-05, "grad_norm": 0.016403302550315857, "global_step": 7817, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03969668224453926, "train_loss_bc": 0.0396948903799057, "train_loss_llm": 0.0001792311668395996, "grad_norm": 0.019538942724466324, "global_step": 7818, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03320274502038956, "train_loss_bc": 0.03320274502038956, "train_loss_llm": 0.0, "grad_norm": 0.025535350665450096, "global_step": 7819, "epoch": 217, "lr": 0.001} +{"train_loss": 0.039355382323265076, "train_loss_bc": 0.039355382323265076, "train_loss_llm": 0.0, "grad_norm": 0.033042360097169876, "global_step": 7820, "epoch": 217, "lr": 0.001} +{"train_loss": 0.04156040772795677, "train_loss_bc": 0.041552942246198654, "train_loss_llm": 0.0007463693618774414, "grad_norm": 0.021730761975049973, "global_step": 7821, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03536658361554146, "train_loss_bc": 0.03536658361554146, "train_loss_llm": 0.0, "grad_norm": 0.032491303980350494, "global_step": 7822, "epoch": 217, "lr": 0.001} +{"train_loss": 0.044537078589200974, "train_loss_bc": 0.044537078589200974, "train_loss_llm": 0.0, "grad_norm": 0.04558840021491051, "global_step": 7823, "epoch": 217, "lr": 0.001} +{"train_loss": 0.039542581886053085, "train_loss_bc": 0.039540283381938934, "train_loss_llm": 0.00022977590560913086, "grad_norm": 0.05220969766378403, "global_step": 7824, "epoch": 217, "lr": 0.001} +{"train_loss": 0.035906657576560974, "train_loss_bc": 0.035906657576560974, "train_loss_llm": 0.0, "grad_norm": 0.01696353778243065, "global_step": 7825, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03412245586514473, "train_loss_bc": 0.03412245586514473, "train_loss_llm": 0.0, "grad_norm": 0.028928358107805252, "global_step": 7826, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03634623810648918, "train_loss_bc": 0.036341238766908646, "train_loss_llm": 0.0004999339580535889, "grad_norm": 0.03302714601159096, "global_step": 7827, "epoch": 217, "lr": 0.001} +{"train_loss": 0.038973983377218246, "train_loss_bc": 0.038973983377218246, "train_loss_llm": 0.0, "grad_norm": 0.03431037440896034, "global_step": 7828, "epoch": 217, "lr": 0.001} +{"train_loss": 0.045614130795001984, "train_loss_bc": 0.045614130795001984, "train_loss_llm": 0.0, "grad_norm": 0.020643487572669983, "global_step": 7829, "epoch": 217, "lr": 0.001} +{"train_loss": 0.04363294318318367, "train_loss_bc": 0.04361489787697792, "train_loss_llm": 0.0018044114112854004, "grad_norm": 0.0289024468511343, "global_step": 7830, "epoch": 217, "lr": 0.001} +{"train_loss": 0.041914284229278564, "train_loss_bc": 0.041914284229278564, "train_loss_llm": 0.0, "grad_norm": 0.03264594450592995, "global_step": 7831, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03956535831093788, "train_loss_bc": 0.03956535831093788, "train_loss_llm": 0.0, "grad_norm": 0.03798757866024971, "global_step": 7832, "epoch": 217, "lr": 0.001} +{"train_loss": 0.042864348739385605, "train_loss_bc": 0.042864348739385605, "train_loss_llm": 0.0, "grad_norm": 0.01983366720378399, "global_step": 7833, "epoch": 217, "lr": 0.001} +{"train_loss": 0.041890859603881836, "train_loss_bc": 0.041890859603881836, "train_loss_llm": 0.0, "grad_norm": 0.02778005413711071, "global_step": 7834, "epoch": 217, "lr": 0.001} +{"train_loss": 0.0360175222158432, "train_loss_bc": 0.0360175222158432, "train_loss_llm": 0.0, "grad_norm": 0.03241686150431633, "global_step": 7835, "epoch": 217, "lr": 0.001} +{"train_loss": 0.04187469184398651, "train_loss_bc": 0.04186844080686569, "train_loss_llm": 0.000624924898147583, "grad_norm": 0.03578567132353783, "global_step": 7836, "epoch": 217, "lr": 0.001} +{"train_loss": 0.04293283820152283, "train_loss_bc": 0.04293283820152283, "train_loss_llm": 0.0, "grad_norm": 0.019956568256020546, "global_step": 7837, "epoch": 217, "lr": 0.001} +{"train_loss": 0.039043404161930084, "train_loss_bc": 0.039043404161930084, "train_loss_llm": 0.0, "grad_norm": 0.028049228712916374, "global_step": 7838, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03285025805234909, "train_loss_bc": 0.03285025805234909, "train_loss_llm": 0.0, "grad_norm": 0.035176608711481094, "global_step": 7839, "epoch": 217, "lr": 0.001} +{"train_loss": 0.039353907108306885, "train_loss_bc": 0.03932737559080124, "train_loss_llm": 0.0026533305644989014, "grad_norm": 0.0434795543551445, "global_step": 7840, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03810412436723709, "train_loss_bc": 0.03810412436723709, "train_loss_llm": 0.0, "grad_norm": 0.019771503284573555, "global_step": 7841, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03816505894064903, "train_loss_bc": 0.03816505894064903, "train_loss_llm": 0.0, "grad_norm": 0.030074577778577805, "global_step": 7842, "epoch": 217, "lr": 0.001} +{"train_loss": 0.038792386651039124, "train_loss_bc": 0.038792386651039124, "train_loss_llm": 0.0, "grad_norm": 0.04007619619369507, "global_step": 7843, "epoch": 217, "lr": 0.001} +{"train_loss": 0.03930816426873207, "train_loss_bc": 0.03930816426873207, "train_loss_llm": 0.0, "grad_norm": 0.04444195330142975, "global_step": 7844, "epoch": 217, "lr": 0.001} +{"train_loss": 0.04030481353402138, "train_loss_bc": 0.04030481353402138, "train_loss_llm": 0.0, "grad_norm": 0.018070336431264877, "global_step": 7845, "epoch": 217, "lr": 0.001} +{"train_loss": 0.04010424390435219, "train_loss_bc": 0.040097810328006744, "train_loss_llm": 0.0006434023380279541, "grad_norm": 0.02873663790524006, "global_step": 7846, "epoch": 217, "lr": 0.001} +{"train_loss": 0.039721471154027514, "train_loss_bc": 0.040393341332674026, "train_loss_llm": 0.001792311668395996, "grad_norm": 0.04361072927713394, "global_step": 7847, "epoch": 217, "lr": 0.001, "val_loss": 0.05788816511631012} +{"train_loss": 0.03599496930837631, "train_loss_bc": 0.03599496930837631, "train_loss_llm": 0.0, "grad_norm": 0.0543120801448822, "global_step": 7848, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04561780393123627, "train_loss_bc": 0.04561780393123627, "train_loss_llm": 0.0, "grad_norm": 0.018309103325009346, "global_step": 7849, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04516357183456421, "train_loss_bc": 0.04516357183456421, "train_loss_llm": 0.0, "grad_norm": 0.025275452062487602, "global_step": 7850, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03774980828166008, "train_loss_bc": 0.03774980828166008, "train_loss_llm": 0.0, "grad_norm": 0.03159086033701897, "global_step": 7851, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03701559081673622, "train_loss_bc": 0.037006739526987076, "train_loss_llm": 0.0008849501609802246, "grad_norm": 0.03345101326704025, "global_step": 7852, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04510509967803955, "train_loss_bc": 0.045095786452293396, "train_loss_llm": 0.00093117356300354, "grad_norm": 0.019895819947123528, "global_step": 7853, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03842918947339058, "train_loss_bc": 0.03842918947339058, "train_loss_llm": 0.0, "grad_norm": 0.029566127806901932, "global_step": 7854, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04423721507191658, "train_loss_bc": 0.04423721507191658, "train_loss_llm": 0.0, "grad_norm": 0.03441714122891426, "global_step": 7855, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04149563983082771, "train_loss_bc": 0.04149563983082771, "train_loss_llm": 0.0, "grad_norm": 0.03604791685938835, "global_step": 7856, "epoch": 218, "lr": 0.001} +{"train_loss": 0.038965001702308655, "train_loss_bc": 0.038965001702308655, "train_loss_llm": 0.0, "grad_norm": 0.014534237794578075, "global_step": 7857, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03846064954996109, "train_loss_bc": 0.03846064954996109, "train_loss_llm": 0.0, "grad_norm": 0.025404104962944984, "global_step": 7858, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04224173724651337, "train_loss_bc": 0.04224173724651337, "train_loss_llm": 0.0, "grad_norm": 0.039195191115140915, "global_step": 7859, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03067586198449135, "train_loss_bc": 0.03067586198449135, "train_loss_llm": 0.0, "grad_norm": 0.044745247811079025, "global_step": 7860, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03558290749788284, "train_loss_bc": 0.03557852283120155, "train_loss_llm": 0.00043833255767822266, "grad_norm": 0.017487907782197, "global_step": 7861, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03736128285527229, "train_loss_bc": 0.03736128285527229, "train_loss_llm": 0.0, "grad_norm": 0.03203631192445755, "global_step": 7862, "epoch": 218, "lr": 0.001} +{"train_loss": 0.036432117223739624, "train_loss_bc": 0.03642842918634415, "train_loss_llm": 0.00036895275115966797, "grad_norm": 0.04273094981908798, "global_step": 7863, "epoch": 218, "lr": 0.001} +{"train_loss": 0.040890712291002274, "train_loss_bc": 0.040890712291002274, "train_loss_llm": 0.0, "grad_norm": 0.04974169284105301, "global_step": 7864, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04296954721212387, "train_loss_bc": 0.04296954721212387, "train_loss_llm": 0.0, "grad_norm": 0.018740728497505188, "global_step": 7865, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04210720956325531, "train_loss_bc": 0.04210720956325531, "train_loss_llm": 0.0, "grad_norm": 0.035219427198171616, "global_step": 7866, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03846064954996109, "train_loss_bc": 0.038459084928035736, "train_loss_llm": 0.00015628337860107422, "grad_norm": 0.04413256421685219, "global_step": 7867, "epoch": 218, "lr": 0.001} +{"train_loss": 0.0420202799141407, "train_loss_bc": 0.0420202799141407, "train_loss_llm": 0.0, "grad_norm": 0.04846183583140373, "global_step": 7868, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03873918578028679, "train_loss_bc": 0.03873918578028679, "train_loss_llm": 0.0, "grad_norm": 0.018262658268213272, "global_step": 7869, "epoch": 218, "lr": 0.001} +{"train_loss": 0.043555159121751785, "train_loss_bc": 0.043555159121751785, "train_loss_llm": 0.0, "grad_norm": 0.02963143028318882, "global_step": 7870, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04291064664721489, "train_loss_bc": 0.04289920628070831, "train_loss_llm": 0.0011441707611083984, "grad_norm": 0.029728902503848076, "global_step": 7871, "epoch": 218, "lr": 0.001} +{"train_loss": 0.034350212663412094, "train_loss_bc": 0.034350212663412094, "train_loss_llm": 0.0, "grad_norm": 0.03691938519477844, "global_step": 7872, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04324415698647499, "train_loss_bc": 0.04324415698647499, "train_loss_llm": 0.0, "grad_norm": 0.016628121957182884, "global_step": 7873, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03297136351466179, "train_loss_bc": 0.03297136351466179, "train_loss_llm": 0.0, "grad_norm": 0.02636459469795227, "global_step": 7874, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03711916133761406, "train_loss_bc": 0.03711916133761406, "train_loss_llm": 0.0, "grad_norm": 0.02965359389781952, "global_step": 7875, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03611364960670471, "train_loss_bc": 0.03608320653438568, "train_loss_llm": 0.003044217824935913, "grad_norm": 0.03661332279443741, "global_step": 7876, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03967498242855072, "train_loss_bc": 0.03967498242855072, "train_loss_llm": 0.0, "grad_norm": 0.020128337666392326, "global_step": 7877, "epoch": 218, "lr": 0.001} +{"train_loss": 0.037804096937179565, "train_loss_bc": 0.037804096937179565, "train_loss_llm": 0.0, "grad_norm": 0.03285156190395355, "global_step": 7878, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04124879837036133, "train_loss_bc": 0.04122951254248619, "train_loss_llm": 0.0019285976886749268, "grad_norm": 0.03357655182480812, "global_step": 7879, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04458415508270264, "train_loss_bc": 0.04457034170627594, "train_loss_llm": 0.001381516456604004, "grad_norm": 0.03718319907784462, "global_step": 7880, "epoch": 218, "lr": 0.001} +{"train_loss": 0.04221910983324051, "train_loss_bc": 0.04221910983324051, "train_loss_llm": 0.0, "grad_norm": 0.01750560663640499, "global_step": 7881, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03972695395350456, "train_loss_bc": 0.03972608596086502, "train_loss_llm": 8.675456047058105e-05, "grad_norm": 0.022492272779345512, "global_step": 7882, "epoch": 218, "lr": 0.001} +{"train_loss": 0.03958562813285324, "train_loss_bc": 0.03384413570165634, "train_loss_llm": 0.0, "grad_norm": 0.028795139864087105, "global_step": 7883, "epoch": 218, "lr": 0.001, "val_loss": 0.057661816477775574} +{"train_loss": 0.04105120524764061, "train_loss_bc": 0.041031692177057266, "train_loss_llm": 0.0019511282444000244, "grad_norm": 0.034909412264823914, "global_step": 7884, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03403804078698158, "train_loss_bc": 0.03402896970510483, "train_loss_llm": 0.0009071230888366699, "grad_norm": 0.019346779212355614, "global_step": 7885, "epoch": 219, "lr": 0.001} +{"train_loss": 0.041715823113918304, "train_loss_bc": 0.041715823113918304, "train_loss_llm": 0.0, "grad_norm": 0.027220331132411957, "global_step": 7886, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03298070281744003, "train_loss_bc": 0.032971080392599106, "train_loss_llm": 0.0009622573852539062, "grad_norm": 0.03193418309092522, "global_step": 7887, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03896858170628548, "train_loss_bc": 0.03896447271108627, "train_loss_llm": 0.0004108548164367676, "grad_norm": 0.03915576636791229, "global_step": 7888, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03940702602267265, "train_loss_bc": 0.03940702602267265, "train_loss_llm": 0.0, "grad_norm": 0.016853701323270798, "global_step": 7889, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03963198885321617, "train_loss_bc": 0.03962814062833786, "train_loss_llm": 0.00038483738899230957, "grad_norm": 0.029593195766210556, "global_step": 7890, "epoch": 219, "lr": 0.001} +{"train_loss": 0.04500087723135948, "train_loss_bc": 0.04499547928571701, "train_loss_llm": 0.000539734959602356, "grad_norm": 0.041066281497478485, "global_step": 7891, "epoch": 219, "lr": 0.001} +{"train_loss": 0.04523370414972305, "train_loss_bc": 0.04521334543824196, "train_loss_llm": 0.002035766839981079, "grad_norm": 0.045527826994657516, "global_step": 7892, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03800036385655403, "train_loss_bc": 0.03800036385655403, "train_loss_llm": 0.0, "grad_norm": 0.01780323125422001, "global_step": 7893, "epoch": 219, "lr": 0.001} +{"train_loss": 0.04473204165697098, "train_loss_bc": 0.04473204165697098, "train_loss_llm": 0.0, "grad_norm": 0.02949041686952114, "global_step": 7894, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03649000823497772, "train_loss_bc": 0.036483004689216614, "train_loss_llm": 0.0007002055644989014, "grad_norm": 0.0343492291867733, "global_step": 7895, "epoch": 219, "lr": 0.001} +{"train_loss": 0.041335977613925934, "train_loss_bc": 0.041335977613925934, "train_loss_llm": 0.0, "grad_norm": 0.03524494543671608, "global_step": 7896, "epoch": 219, "lr": 0.001} +{"train_loss": 0.04017700254917145, "train_loss_bc": 0.04017700254917145, "train_loss_llm": 0.0, "grad_norm": 0.020912950858473778, "global_step": 7897, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03554631769657135, "train_loss_bc": 0.03554631769657135, "train_loss_llm": 0.0, "grad_norm": 0.034326549619436264, "global_step": 7898, "epoch": 219, "lr": 0.001} +{"train_loss": 0.04372577741742134, "train_loss_bc": 0.04372577741742134, "train_loss_llm": 0.0, "grad_norm": 0.03749021142721176, "global_step": 7899, "epoch": 219, "lr": 0.001} +{"train_loss": 0.034550152719020844, "train_loss_bc": 0.034550152719020844, "train_loss_llm": 0.0, "grad_norm": 0.04176856577396393, "global_step": 7900, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03861090913414955, "train_loss_bc": 0.03860437497496605, "train_loss_llm": 0.0006534159183502197, "grad_norm": 0.017505519092082977, "global_step": 7901, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03464352712035179, "train_loss_bc": 0.03464352712035179, "train_loss_llm": 0.0, "grad_norm": 0.023654915392398834, "global_step": 7902, "epoch": 219, "lr": 0.001} +{"train_loss": 0.04373439401388168, "train_loss_bc": 0.043719332665205, "train_loss_llm": 0.0015061795711517334, "grad_norm": 0.02471170388162136, "global_step": 7903, "epoch": 219, "lr": 0.001} +{"train_loss": 0.04002058878540993, "train_loss_bc": 0.04002058878540993, "train_loss_llm": 0.0, "grad_norm": 0.031194310635328293, "global_step": 7904, "epoch": 219, "lr": 0.001} +{"train_loss": 0.042529333382844925, "train_loss_bc": 0.042528893798589706, "train_loss_llm": 4.407763481140137e-05, "grad_norm": 0.01983770914375782, "global_step": 7905, "epoch": 219, "lr": 0.001} +{"train_loss": 0.044627074152231216, "train_loss_bc": 0.044627074152231216, "train_loss_llm": 0.0, "grad_norm": 0.027786094695329666, "global_step": 7906, "epoch": 219, "lr": 0.001} +{"train_loss": 0.032234951853752136, "train_loss_bc": 0.03221917152404785, "train_loss_llm": 0.0015781521797180176, "grad_norm": 0.03374164178967476, "global_step": 7907, "epoch": 219, "lr": 0.001} +{"train_loss": 0.04522688686847687, "train_loss_bc": 0.0451958104968071, "train_loss_llm": 0.003107815980911255, "grad_norm": 0.03882870823144913, "global_step": 7908, "epoch": 219, "lr": 0.001} +{"train_loss": 0.036607902497053146, "train_loss_bc": 0.036600980907678604, "train_loss_llm": 0.0006920099258422852, "grad_norm": 0.018547406420111656, "global_step": 7909, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03537838160991669, "train_loss_bc": 0.03537838160991669, "train_loss_llm": 0.0, "grad_norm": 0.029055260121822357, "global_step": 7910, "epoch": 219, "lr": 0.001} +{"train_loss": 0.046500351279973984, "train_loss_bc": 0.046500351279973984, "train_loss_llm": 0.0, "grad_norm": 0.03921667858958244, "global_step": 7911, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03327660262584686, "train_loss_bc": 0.03327660262584686, "train_loss_llm": 0.0, "grad_norm": 0.04504946991801262, "global_step": 7912, "epoch": 219, "lr": 0.001} +{"train_loss": 0.043631523847579956, "train_loss_bc": 0.043631523847579956, "train_loss_llm": 0.0, "grad_norm": 0.02003139816224575, "global_step": 7913, "epoch": 219, "lr": 0.001} +{"train_loss": 0.0377078615128994, "train_loss_bc": 0.0377078615128994, "train_loss_llm": 0.0, "grad_norm": 0.03084533102810383, "global_step": 7914, "epoch": 219, "lr": 0.001} +{"train_loss": 0.03434900566935539, "train_loss_bc": 0.03434900566935539, "train_loss_llm": 0.0, "grad_norm": 0.04196975380182266, "global_step": 7915, "epoch": 219, "lr": 0.001} +{"train_loss": 0.0430910661816597, "train_loss_bc": 0.04308874532580376, "train_loss_llm": 0.0002319812774658203, "grad_norm": 0.053621478378772736, "global_step": 7916, "epoch": 219, "lr": 0.001} +{"train_loss": 0.04397648945450783, "train_loss_bc": 0.04396454989910126, "train_loss_llm": 0.0011940300464630127, "grad_norm": 0.02070852927863598, "global_step": 7917, "epoch": 219, "lr": 0.001} +{"train_loss": 0.038480594754219055, "train_loss_bc": 0.038480594754219055, "train_loss_llm": 0.0, "grad_norm": 0.03087969869375229, "global_step": 7918, "epoch": 219, "lr": 0.001} +{"train_loss": 0.039619733579456806, "train_loss_bc": 0.0390973724424839, "train_loss_llm": 0.0, "grad_norm": 0.04068157076835632, "global_step": 7919, "epoch": 219, "lr": 0.001, "val_loss": 0.05778014659881592} +{"train_loss": 0.04428592696785927, "train_loss_bc": 0.04428592696785927, "train_loss_llm": 0.0, "grad_norm": 0.04955926537513733, "global_step": 7920, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04596986621618271, "train_loss_bc": 0.04596836119890213, "train_loss_llm": 0.00015050172805786133, "grad_norm": 0.017869779840111732, "global_step": 7921, "epoch": 220, "lr": 0.001} +{"train_loss": 0.037835463881492615, "train_loss_bc": 0.037835463881492615, "train_loss_llm": 0.0, "grad_norm": 0.026086270809173584, "global_step": 7922, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03614778071641922, "train_loss_bc": 0.03614778071641922, "train_loss_llm": 0.0, "grad_norm": 0.03150017932057381, "global_step": 7923, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04029332846403122, "train_loss_bc": 0.04029332846403122, "train_loss_llm": 0.0, "grad_norm": 0.042926106601953506, "global_step": 7924, "epoch": 220, "lr": 0.001} +{"train_loss": 0.044527292251586914, "train_loss_bc": 0.04452193155884743, "train_loss_llm": 0.000536113977432251, "grad_norm": 0.021542660892009735, "global_step": 7925, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03101530484855175, "train_loss_bc": 0.030993983149528503, "train_loss_llm": 0.0021320879459381104, "grad_norm": 0.029956093057990074, "global_step": 7926, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03236633539199829, "train_loss_bc": 0.03236633539199829, "train_loss_llm": 0.0, "grad_norm": 0.03140370547771454, "global_step": 7927, "epoch": 220, "lr": 0.001} +{"train_loss": 0.02745961956679821, "train_loss_bc": 0.02745961956679821, "train_loss_llm": 0.0, "grad_norm": 0.03358350321650505, "global_step": 7928, "epoch": 220, "lr": 0.001} +{"train_loss": 0.0501876138150692, "train_loss_bc": 0.0501876138150692, "train_loss_llm": 0.0, "grad_norm": 0.022717060521245003, "global_step": 7929, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04039615020155907, "train_loss_bc": 0.04039615020155907, "train_loss_llm": 0.0, "grad_norm": 0.03358214721083641, "global_step": 7930, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04312410578131676, "train_loss_bc": 0.043116308748722076, "train_loss_llm": 0.0007798671722412109, "grad_norm": 0.03821645304560661, "global_step": 7931, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03434530273079872, "train_loss_bc": 0.034337982535362244, "train_loss_llm": 0.0007320642471313477, "grad_norm": 0.042100872844457626, "global_step": 7932, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04134591296315193, "train_loss_bc": 0.04134591296315193, "train_loss_llm": 0.0, "grad_norm": 0.01709062233567238, "global_step": 7933, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04090379923582077, "train_loss_bc": 0.04090379923582077, "train_loss_llm": 0.0, "grad_norm": 0.024669397622346878, "global_step": 7934, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03995708003640175, "train_loss_bc": 0.03995661064982414, "train_loss_llm": 4.711747169494629e-05, "grad_norm": 0.03323191776871681, "global_step": 7935, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03487289696931839, "train_loss_bc": 0.034872326999902725, "train_loss_llm": 5.683302879333496e-05, "grad_norm": 0.04130597040057182, "global_step": 7936, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04104791581630707, "train_loss_bc": 0.04104791581630707, "train_loss_llm": 0.0, "grad_norm": 0.01776723563671112, "global_step": 7937, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04129017889499664, "train_loss_bc": 0.041283443570137024, "train_loss_llm": 0.0006736814975738525, "grad_norm": 0.0311085507273674, "global_step": 7938, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04403086379170418, "train_loss_bc": 0.04403086379170418, "train_loss_llm": 0.0, "grad_norm": 0.03553830087184906, "global_step": 7939, "epoch": 220, "lr": 0.001} +{"train_loss": 0.043461356312036514, "train_loss_bc": 0.043456997722387314, "train_loss_llm": 0.00043579936027526855, "grad_norm": 0.050818588584661484, "global_step": 7940, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04400220885872841, "train_loss_bc": 0.04400220885872841, "train_loss_llm": 0.0, "grad_norm": 0.021196207031607628, "global_step": 7941, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03842056170105934, "train_loss_bc": 0.03842056170105934, "train_loss_llm": 0.0, "grad_norm": 0.026666291058063507, "global_step": 7942, "epoch": 220, "lr": 0.001} +{"train_loss": 0.0430988110601902, "train_loss_bc": 0.04308729246258736, "train_loss_llm": 0.0011518001556396484, "grad_norm": 0.03261696174740791, "global_step": 7943, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03831270709633827, "train_loss_bc": 0.03831270709633827, "train_loss_llm": 0.0, "grad_norm": 0.0374293215572834, "global_step": 7944, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03469410166144371, "train_loss_bc": 0.03467677906155586, "train_loss_llm": 0.0017322301864624023, "grad_norm": 0.01700378768146038, "global_step": 7945, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03763457015156746, "train_loss_bc": 0.03763457015156746, "train_loss_llm": 0.0, "grad_norm": 0.02810089662671089, "global_step": 7946, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03859417140483856, "train_loss_bc": 0.03859417140483856, "train_loss_llm": 0.0, "grad_norm": 0.038967251777648926, "global_step": 7947, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04064653441309929, "train_loss_bc": 0.04062828794121742, "train_loss_llm": 0.0018247365951538086, "grad_norm": 0.048479825258255005, "global_step": 7948, "epoch": 220, "lr": 0.001} +{"train_loss": 0.040846507996320724, "train_loss_bc": 0.040846507996320724, "train_loss_llm": 0.0, "grad_norm": 0.015098853968083858, "global_step": 7949, "epoch": 220, "lr": 0.001} +{"train_loss": 0.043112773448228836, "train_loss_bc": 0.043112773448228836, "train_loss_llm": 0.0, "grad_norm": 0.028685765340924263, "global_step": 7950, "epoch": 220, "lr": 0.001} +{"train_loss": 0.04536228999495506, "train_loss_bc": 0.04536228999495506, "train_loss_llm": 0.0, "grad_norm": 0.03934088721871376, "global_step": 7951, "epoch": 220, "lr": 0.001} +{"train_loss": 0.025807229802012444, "train_loss_bc": 0.025806307792663574, "train_loss_llm": 9.226799011230469e-05, "grad_norm": 0.05371832102537155, "global_step": 7952, "epoch": 220, "lr": 0.001} +{"train_loss": 0.045555468648672104, "train_loss_bc": 0.045555468648672104, "train_loss_llm": 0.0, "grad_norm": 0.02175143174827099, "global_step": 7953, "epoch": 220, "lr": 0.001} +{"train_loss": 0.03263711556792259, "train_loss_bc": 0.03263413906097412, "train_loss_llm": 0.00029775500297546387, "grad_norm": 0.033493317663669586, "global_step": 7954, "epoch": 220, "lr": 0.001} +{"train_loss": 0.039731289415309824, "train_loss_bc": 0.046737272292375565, "train_loss_llm": 0.0, "grad_norm": 0.051469236612319946, "global_step": 7955, "epoch": 220, "lr": 0.001, "train/cumulative_reward": 67.01159691381527, "train/mean_score": 11.719807441630087, "train/success_rate": 0.1, "test/cumulative_reward": 60.25579794854288, "test/mean_score": 6.265182767680898, "test/success_rate": 0.12, "val_loss": 0.057857923209667206, "train_action_mse_error": 0.043364446610212326} +{"train_loss": 0.03283338621258736, "train_loss_bc": 0.03283338621258736, "train_loss_llm": 0.0, "grad_norm": 0.056909188628196716, "global_step": 7956, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03299473598599434, "train_loss_bc": 0.03299473598599434, "train_loss_llm": 0.0, "grad_norm": 0.014281855896115303, "global_step": 7957, "epoch": 221, "lr": 0.001} +{"train_loss": 0.033897630870342255, "train_loss_bc": 0.03389560803771019, "train_loss_llm": 0.0002022087574005127, "grad_norm": 0.024455735459923744, "global_step": 7958, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03774731233716011, "train_loss_bc": 0.03773679584264755, "train_loss_llm": 0.0010518133640289307, "grad_norm": 0.030666857957839966, "global_step": 7959, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03276138752698898, "train_loss_bc": 0.03276138752698898, "train_loss_llm": 0.0, "grad_norm": 0.032186709344387054, "global_step": 7960, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03875065594911575, "train_loss_bc": 0.03875065594911575, "train_loss_llm": 0.0, "grad_norm": 0.02174420841038227, "global_step": 7961, "epoch": 221, "lr": 0.001} +{"train_loss": 0.04075099155306816, "train_loss_bc": 0.04074955731630325, "train_loss_llm": 0.00014340877532958984, "grad_norm": 0.037878893315792084, "global_step": 7962, "epoch": 221, "lr": 0.001} +{"train_loss": 0.041613008826971054, "train_loss_bc": 0.041613008826971054, "train_loss_llm": 0.0, "grad_norm": 0.0413711816072464, "global_step": 7963, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03397172689437866, "train_loss_bc": 0.03394487127661705, "train_loss_llm": 0.0026853978633880615, "grad_norm": 0.04452580213546753, "global_step": 7964, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03666811063885689, "train_loss_bc": 0.03665516898036003, "train_loss_llm": 0.0012941956520080566, "grad_norm": 0.017804531380534172, "global_step": 7965, "epoch": 221, "lr": 0.001} +{"train_loss": 0.038448069244623184, "train_loss_bc": 0.038448069244623184, "train_loss_llm": 0.0, "grad_norm": 0.02755841612815857, "global_step": 7966, "epoch": 221, "lr": 0.001} +{"train_loss": 0.043113600462675095, "train_loss_bc": 0.043094098567962646, "train_loss_llm": 0.001950085163116455, "grad_norm": 0.03765472397208214, "global_step": 7967, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03765180706977844, "train_loss_bc": 0.03765180706977844, "train_loss_llm": 0.0, "grad_norm": 0.04777122661471367, "global_step": 7968, "epoch": 221, "lr": 0.001} +{"train_loss": 0.044928647577762604, "train_loss_bc": 0.044928647577762604, "train_loss_llm": 0.0, "grad_norm": 0.022061895579099655, "global_step": 7969, "epoch": 221, "lr": 0.001} +{"train_loss": 0.04695515334606171, "train_loss_bc": 0.04695515334606171, "train_loss_llm": 0.0, "grad_norm": 0.03670026361942291, "global_step": 7970, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03717229142785072, "train_loss_bc": 0.03717229142785072, "train_loss_llm": 0.0, "grad_norm": 0.03791854530572891, "global_step": 7971, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03905772417783737, "train_loss_bc": 0.03905772417783737, "train_loss_llm": 0.0, "grad_norm": 0.04796874523162842, "global_step": 7972, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03581295162439346, "train_loss_bc": 0.03580533713102341, "train_loss_llm": 0.0007613599300384521, "grad_norm": 0.016731688752770424, "global_step": 7973, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03506609797477722, "train_loss_bc": 0.03506609797477722, "train_loss_llm": 0.0, "grad_norm": 0.028225542977452278, "global_step": 7974, "epoch": 221, "lr": 0.001} +{"train_loss": 0.04127918928861618, "train_loss_bc": 0.04126695543527603, "train_loss_llm": 0.0012233555316925049, "grad_norm": 0.036617811769247055, "global_step": 7975, "epoch": 221, "lr": 0.001} +{"train_loss": 0.04030836746096611, "train_loss_bc": 0.040304478257894516, "train_loss_llm": 0.00038883090019226074, "grad_norm": 0.04526003077626228, "global_step": 7976, "epoch": 221, "lr": 0.001} +{"train_loss": 0.04645270109176636, "train_loss_bc": 0.04645270109176636, "train_loss_llm": 0.0, "grad_norm": 0.019674500450491905, "global_step": 7977, "epoch": 221, "lr": 0.001} +{"train_loss": 0.04259985312819481, "train_loss_bc": 0.04258901625871658, "train_loss_llm": 0.0010837018489837646, "grad_norm": 0.028416410088539124, "global_step": 7978, "epoch": 221, "lr": 0.001} +{"train_loss": 0.042427048087120056, "train_loss_bc": 0.042427048087120056, "train_loss_llm": 0.0, "grad_norm": 0.04157017543911934, "global_step": 7979, "epoch": 221, "lr": 0.001} +{"train_loss": 0.04992004856467247, "train_loss_bc": 0.04992004856467247, "train_loss_llm": 0.0, "grad_norm": 0.05306273326277733, "global_step": 7980, "epoch": 221, "lr": 0.001} +{"train_loss": 0.04520345479249954, "train_loss_bc": 0.045171961188316345, "train_loss_llm": 0.0031493008136749268, "grad_norm": 0.020596299320459366, "global_step": 7981, "epoch": 221, "lr": 0.001} +{"train_loss": 0.041157182306051254, "train_loss_bc": 0.04114434868097305, "train_loss_llm": 0.0012832283973693848, "grad_norm": 0.03092748299241066, "global_step": 7982, "epoch": 221, "lr": 0.001} +{"train_loss": 0.04076731204986572, "train_loss_bc": 0.040751248598098755, "train_loss_llm": 0.0016063451766967773, "grad_norm": 0.038198743015527725, "global_step": 7983, "epoch": 221, "lr": 0.001} +{"train_loss": 0.04602396860718727, "train_loss_bc": 0.04602207615971565, "train_loss_llm": 0.00018912553787231445, "grad_norm": 0.04823368787765503, "global_step": 7984, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03481524810194969, "train_loss_bc": 0.03481524810194969, "train_loss_llm": 0.0, "grad_norm": 0.015536339022219181, "global_step": 7985, "epoch": 221, "lr": 0.001} +{"train_loss": 0.041740406304597855, "train_loss_bc": 0.04173918813467026, "train_loss_llm": 0.00012168288230895996, "grad_norm": 0.023424427956342697, "global_step": 7986, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03282253444194794, "train_loss_bc": 0.03282253444194794, "train_loss_llm": 0.0, "grad_norm": 0.029275700449943542, "global_step": 7987, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03587803989648819, "train_loss_bc": 0.03587803989648819, "train_loss_llm": 0.0, "grad_norm": 0.035194553434848785, "global_step": 7988, "epoch": 221, "lr": 0.001} +{"train_loss": 0.044797711074352264, "train_loss_bc": 0.044797711074352264, "train_loss_llm": 0.0, "grad_norm": 0.019128840416669846, "global_step": 7989, "epoch": 221, "lr": 0.001} +{"train_loss": 0.044063471257686615, "train_loss_bc": 0.044063471257686615, "train_loss_llm": 0.0, "grad_norm": 0.0299516674131155, "global_step": 7990, "epoch": 221, "lr": 0.001} +{"train_loss": 0.03954174266093307, "train_loss_bc": 0.033050909638404846, "train_loss_llm": 0.0, "grad_norm": 0.04053344205021858, "global_step": 7991, "epoch": 221, "lr": 0.001, "val_loss": 0.05774042755365372} +{"train_loss": 0.04311971738934517, "train_loss_bc": 0.04311971738934517, "train_loss_llm": 0.0, "grad_norm": 0.04995046183466911, "global_step": 7992, "epoch": 222, "lr": 0.001} +{"train_loss": 0.040144409984350204, "train_loss_bc": 0.04013149067759514, "train_loss_llm": 0.0012920498847961426, "grad_norm": 0.018259353935718536, "global_step": 7993, "epoch": 222, "lr": 0.001} +{"train_loss": 0.036411628127098083, "train_loss_bc": 0.036411628127098083, "train_loss_llm": 0.0, "grad_norm": 0.025371422991156578, "global_step": 7994, "epoch": 222, "lr": 0.001} +{"train_loss": 0.0415985994040966, "train_loss_bc": 0.0415985994040966, "train_loss_llm": 0.0, "grad_norm": 0.033181894570589066, "global_step": 7995, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04242100194096565, "train_loss_bc": 0.04242100194096565, "train_loss_llm": 0.0, "grad_norm": 0.03944484889507294, "global_step": 7996, "epoch": 222, "lr": 0.001} +{"train_loss": 0.037030499428510666, "train_loss_bc": 0.037019841372966766, "train_loss_llm": 0.0010659098625183105, "grad_norm": 0.01737244240939617, "global_step": 7997, "epoch": 222, "lr": 0.001} +{"train_loss": 0.037025418132543564, "train_loss_bc": 0.037025418132543564, "train_loss_llm": 0.0, "grad_norm": 0.03013451397418976, "global_step": 7998, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04100218042731285, "train_loss_bc": 0.04100218042731285, "train_loss_llm": 0.0, "grad_norm": 0.03734336420893669, "global_step": 7999, "epoch": 222, "lr": 0.001} +{"train_loss": 0.034573692828416824, "train_loss_bc": 0.03456290438771248, "train_loss_llm": 0.001078784465789795, "grad_norm": 0.04274198040366173, "global_step": 8000, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04373438283801079, "train_loss_bc": 0.0437261164188385, "train_loss_llm": 0.0008265376091003418, "grad_norm": 0.01971260830760002, "global_step": 8001, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04251645505428314, "train_loss_bc": 0.04251645505428314, "train_loss_llm": 0.0, "grad_norm": 0.030595924705266953, "global_step": 8002, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04041970893740654, "train_loss_bc": 0.04041970893740654, "train_loss_llm": 0.0, "grad_norm": 0.037315841764211655, "global_step": 8003, "epoch": 222, "lr": 0.001} +{"train_loss": 0.041442401707172394, "train_loss_bc": 0.041442401707172394, "train_loss_llm": 0.0, "grad_norm": 0.04971987009048462, "global_step": 8004, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04462601989507675, "train_loss_bc": 0.04460585489869118, "train_loss_llm": 0.0020163655281066895, "grad_norm": 0.024848870933055878, "global_step": 8005, "epoch": 222, "lr": 0.001} +{"train_loss": 0.038046929985284805, "train_loss_bc": 0.038046929985284805, "train_loss_llm": 0.0, "grad_norm": 0.043516211211681366, "global_step": 8006, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04196174442768097, "train_loss_bc": 0.04196174442768097, "train_loss_llm": 0.0, "grad_norm": 0.05305451154708862, "global_step": 8007, "epoch": 222, "lr": 0.001} +{"train_loss": 0.03790811076760292, "train_loss_bc": 0.03790811076760292, "train_loss_llm": 0.0, "grad_norm": 0.05881696939468384, "global_step": 8008, "epoch": 222, "lr": 0.001} +{"train_loss": 0.03310731053352356, "train_loss_bc": 0.03310002014040947, "train_loss_llm": 0.000728994607925415, "grad_norm": 0.019158294424414635, "global_step": 8009, "epoch": 222, "lr": 0.001} +{"train_loss": 0.041514359414577484, "train_loss_bc": 0.041514359414577484, "train_loss_llm": 0.0, "grad_norm": 0.03055439703166485, "global_step": 8010, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04212113469839096, "train_loss_bc": 0.04212113469839096, "train_loss_llm": 0.0, "grad_norm": 0.04334321245551109, "global_step": 8011, "epoch": 222, "lr": 0.001} +{"train_loss": 0.041869599372148514, "train_loss_bc": 0.04185247793793678, "train_loss_llm": 0.0017121434211730957, "grad_norm": 0.047314032912254333, "global_step": 8012, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04087311029434204, "train_loss_bc": 0.04087311029434204, "train_loss_llm": 0.0, "grad_norm": 0.01734931208193302, "global_step": 8013, "epoch": 222, "lr": 0.001} +{"train_loss": 0.03701689466834068, "train_loss_bc": 0.03701689466834068, "train_loss_llm": 0.0, "grad_norm": 0.027959927916526794, "global_step": 8014, "epoch": 222, "lr": 0.001} +{"train_loss": 0.03254520893096924, "train_loss_bc": 0.03254520893096924, "train_loss_llm": 0.0, "grad_norm": 0.03804754093289375, "global_step": 8015, "epoch": 222, "lr": 0.001} +{"train_loss": 0.048447202891111374, "train_loss_bc": 0.048447202891111374, "train_loss_llm": 0.0, "grad_norm": 0.047343578189611435, "global_step": 8016, "epoch": 222, "lr": 0.001} +{"train_loss": 0.03673463687300682, "train_loss_bc": 0.03672751784324646, "train_loss_llm": 0.0007118880748748779, "grad_norm": 0.019761189818382263, "global_step": 8017, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04075750708580017, "train_loss_bc": 0.04075750708580017, "train_loss_llm": 0.0, "grad_norm": 0.035578008741140366, "global_step": 8018, "epoch": 222, "lr": 0.001} +{"train_loss": 0.03447382524609566, "train_loss_bc": 0.03446386381983757, "train_loss_llm": 0.0009961128234863281, "grad_norm": 0.052185140550136566, "global_step": 8019, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04284481704235077, "train_loss_bc": 0.04284481704235077, "train_loss_llm": 0.0, "grad_norm": 0.07104062288999557, "global_step": 8020, "epoch": 222, "lr": 0.001} +{"train_loss": 0.039911169558763504, "train_loss_bc": 0.039911169558763504, "train_loss_llm": 0.0, "grad_norm": 0.016036290675401688, "global_step": 8021, "epoch": 222, "lr": 0.001} +{"train_loss": 0.036807361990213394, "train_loss_bc": 0.036807361990213394, "train_loss_llm": 0.0, "grad_norm": 0.024675292894244194, "global_step": 8022, "epoch": 222, "lr": 0.001} +{"train_loss": 0.0393223874270916, "train_loss_bc": 0.03929614648222923, "train_loss_llm": 0.002623945474624634, "grad_norm": 0.03246280550956726, "global_step": 8023, "epoch": 222, "lr": 0.001} +{"train_loss": 0.037280406802892685, "train_loss_bc": 0.03727630153298378, "train_loss_llm": 0.00041049718856811523, "grad_norm": 0.040287066251039505, "global_step": 8024, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04098762944340706, "train_loss_bc": 0.04097265750169754, "train_loss_llm": 0.0014971494674682617, "grad_norm": 0.023244749754667282, "global_step": 8025, "epoch": 222, "lr": 0.001} +{"train_loss": 0.04325089976191521, "train_loss_bc": 0.04325089976191521, "train_loss_llm": 0.0, "grad_norm": 0.03729718178510666, "global_step": 8026, "epoch": 222, "lr": 0.001} +{"train_loss": 0.039722076203260154, "train_loss_bc": 0.03614024445414543, "train_loss_llm": 0.0006133913993835449, "grad_norm": 0.055006109178066254, "global_step": 8027, "epoch": 222, "lr": 0.001, "val_loss": 0.058531053364276886} +{"train_loss": 0.0326676182448864, "train_loss_bc": 0.0326676182448864, "train_loss_llm": 0.0, "grad_norm": 0.06533357501029968, "global_step": 8028, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03136660158634186, "train_loss_bc": 0.03136255219578743, "train_loss_llm": 0.000405043363571167, "grad_norm": 0.017865601927042007, "global_step": 8029, "epoch": 223, "lr": 0.001} +{"train_loss": 0.04027723893523216, "train_loss_bc": 0.04027723893523216, "train_loss_llm": 0.0, "grad_norm": 0.030139297246932983, "global_step": 8030, "epoch": 223, "lr": 0.001} +{"train_loss": 0.041463807225227356, "train_loss_bc": 0.041463807225227356, "train_loss_llm": 0.0, "grad_norm": 0.036115117371082306, "global_step": 8031, "epoch": 223, "lr": 0.001} +{"train_loss": 0.030238881707191467, "train_loss_bc": 0.030238881707191467, "train_loss_llm": 0.0, "grad_norm": 0.04593496024608612, "global_step": 8032, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03825727105140686, "train_loss_bc": 0.038205698132514954, "train_loss_llm": 0.005157381296157837, "grad_norm": 0.018232669681310654, "global_step": 8033, "epoch": 223, "lr": 0.001} +{"train_loss": 0.0382586345076561, "train_loss_bc": 0.0382586345076561, "train_loss_llm": 0.0, "grad_norm": 0.023323554545640945, "global_step": 8034, "epoch": 223, "lr": 0.001} +{"train_loss": 0.039581991732120514, "train_loss_bc": 0.039581991732120514, "train_loss_llm": 0.0, "grad_norm": 0.034002017229795456, "global_step": 8035, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03689279779791832, "train_loss_bc": 0.03689279779791832, "train_loss_llm": 0.0, "grad_norm": 0.04178424924612045, "global_step": 8036, "epoch": 223, "lr": 0.001} +{"train_loss": 0.044186096638441086, "train_loss_bc": 0.044186096638441086, "train_loss_llm": 0.0, "grad_norm": 0.01833307556807995, "global_step": 8037, "epoch": 223, "lr": 0.001} +{"train_loss": 0.04356763884425163, "train_loss_bc": 0.04356763884425163, "train_loss_llm": 0.0, "grad_norm": 0.02595711313188076, "global_step": 8038, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03137689456343651, "train_loss_bc": 0.03137689456343651, "train_loss_llm": 0.0, "grad_norm": 0.031668540090322495, "global_step": 8039, "epoch": 223, "lr": 0.001} +{"train_loss": 0.053924016654491425, "train_loss_bc": 0.053924016654491425, "train_loss_llm": 0.0, "grad_norm": 0.037715766578912735, "global_step": 8040, "epoch": 223, "lr": 0.001} +{"train_loss": 0.04040081053972244, "train_loss_bc": 0.04040081053972244, "train_loss_llm": 0.0, "grad_norm": 0.01720377802848816, "global_step": 8041, "epoch": 223, "lr": 0.001} +{"train_loss": 0.040557682514190674, "train_loss_bc": 0.04052431881427765, "train_loss_llm": 0.003336399793624878, "grad_norm": 0.03213689103722572, "global_step": 8042, "epoch": 223, "lr": 0.001} +{"train_loss": 0.0316474474966526, "train_loss_bc": 0.0316426083445549, "train_loss_llm": 0.00048407912254333496, "grad_norm": 0.03301509842276573, "global_step": 8043, "epoch": 223, "lr": 0.001} +{"train_loss": 0.046521104872226715, "train_loss_bc": 0.046521104872226715, "train_loss_llm": 0.0, "grad_norm": 0.04400121420621872, "global_step": 8044, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03859589993953705, "train_loss_bc": 0.03859391435980797, "train_loss_llm": 0.00019866228103637695, "grad_norm": 0.017839111387729645, "global_step": 8045, "epoch": 223, "lr": 0.001} +{"train_loss": 0.040760587900877, "train_loss_bc": 0.040760587900877, "train_loss_llm": 0.0, "grad_norm": 0.03114372491836548, "global_step": 8046, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03296167775988579, "train_loss_bc": 0.03296167775988579, "train_loss_llm": 0.0, "grad_norm": 0.0489032119512558, "global_step": 8047, "epoch": 223, "lr": 0.001} +{"train_loss": 0.04704304784536362, "train_loss_bc": 0.04704304784536362, "train_loss_llm": 0.0, "grad_norm": 0.06203818693757057, "global_step": 8048, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03823743388056755, "train_loss_bc": 0.03822851926088333, "train_loss_llm": 0.0008915960788726807, "grad_norm": 0.022232746705412865, "global_step": 8049, "epoch": 223, "lr": 0.001} +{"train_loss": 0.04191148653626442, "train_loss_bc": 0.04191148653626442, "train_loss_llm": 0.0, "grad_norm": 0.02552991732954979, "global_step": 8050, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03847706690430641, "train_loss_bc": 0.03847706690430641, "train_loss_llm": 0.0, "grad_norm": 0.03283621370792389, "global_step": 8051, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03772853687405586, "train_loss_bc": 0.03772560879588127, "train_loss_llm": 0.00029283761978149414, "grad_norm": 0.041931599378585815, "global_step": 8052, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03897756710648537, "train_loss_bc": 0.03897756710648537, "train_loss_llm": 0.0, "grad_norm": 0.016965078189969063, "global_step": 8053, "epoch": 223, "lr": 0.001} +{"train_loss": 0.04487277567386627, "train_loss_bc": 0.04487277567386627, "train_loss_llm": 0.0, "grad_norm": 0.02630903571844101, "global_step": 8054, "epoch": 223, "lr": 0.001} +{"train_loss": 0.046713490039110184, "train_loss_bc": 0.046713490039110184, "train_loss_llm": 0.0, "grad_norm": 0.039658427238464355, "global_step": 8055, "epoch": 223, "lr": 0.001} +{"train_loss": 0.037524256855249405, "train_loss_bc": 0.037524256855249405, "train_loss_llm": 0.0, "grad_norm": 0.04973215237259865, "global_step": 8056, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03319404274225235, "train_loss_bc": 0.03319404274225235, "train_loss_llm": 0.0, "grad_norm": 0.019512610509991646, "global_step": 8057, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03856198489665985, "train_loss_bc": 0.038554999977350235, "train_loss_llm": 0.000698387622833252, "grad_norm": 0.03170435503125191, "global_step": 8058, "epoch": 223, "lr": 0.001} +{"train_loss": 0.04176141694188118, "train_loss_bc": 0.041746221482753754, "train_loss_llm": 0.0015193819999694824, "grad_norm": 0.03458490967750549, "global_step": 8059, "epoch": 223, "lr": 0.001} +{"train_loss": 0.04147685691714287, "train_loss_bc": 0.04147685691714287, "train_loss_llm": 0.0, "grad_norm": 0.04364645481109619, "global_step": 8060, "epoch": 223, "lr": 0.001} +{"train_loss": 0.0495796799659729, "train_loss_bc": 0.0495796799659729, "train_loss_llm": 0.0, "grad_norm": 0.020081516355276108, "global_step": 8061, "epoch": 223, "lr": 0.001} +{"train_loss": 0.0403231680393219, "train_loss_bc": 0.04031809791922569, "train_loss_llm": 0.0005068778991699219, "grad_norm": 0.03126966580748558, "global_step": 8062, "epoch": 223, "lr": 0.001} +{"train_loss": 0.03967943156345023, "train_loss_bc": 0.038572024554014206, "train_loss_llm": 0.0, "grad_norm": 0.03608231991529465, "global_step": 8063, "epoch": 223, "lr": 0.001, "val_loss": 0.05771442875266075} +{"train_loss": 0.04118739068508148, "train_loss_bc": 0.04118739068508148, "train_loss_llm": 0.0, "grad_norm": 0.04511011019349098, "global_step": 8064, "epoch": 224, "lr": 0.001} +{"train_loss": 0.04348646476864815, "train_loss_bc": 0.04348646476864815, "train_loss_llm": 0.0, "grad_norm": 0.02057759463787079, "global_step": 8065, "epoch": 224, "lr": 0.001} +{"train_loss": 0.036103226244449615, "train_loss_bc": 0.036103226244449615, "train_loss_llm": 0.0, "grad_norm": 0.02790205366909504, "global_step": 8066, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03518976643681526, "train_loss_bc": 0.035187747329473495, "train_loss_llm": 0.00020179152488708496, "grad_norm": 0.03730742633342743, "global_step": 8067, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03325333818793297, "train_loss_bc": 0.03325333818793297, "train_loss_llm": 0.0, "grad_norm": 0.04258832708001137, "global_step": 8068, "epoch": 224, "lr": 0.001} +{"train_loss": 0.033143844455480576, "train_loss_bc": 0.033143483102321625, "train_loss_llm": 3.603100776672363e-05, "grad_norm": 0.02026684209704399, "global_step": 8069, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03399061784148216, "train_loss_bc": 0.03398185968399048, "train_loss_llm": 0.0008757412433624268, "grad_norm": 0.03093402273952961, "global_step": 8070, "epoch": 224, "lr": 0.001} +{"train_loss": 0.038692764937877655, "train_loss_bc": 0.038688305765390396, "train_loss_llm": 0.0004457831382751465, "grad_norm": 0.0441434308886528, "global_step": 8071, "epoch": 224, "lr": 0.001} +{"train_loss": 0.041896723210811615, "train_loss_bc": 0.041896723210811615, "train_loss_llm": 0.0, "grad_norm": 0.061143483966588974, "global_step": 8072, "epoch": 224, "lr": 0.001} +{"train_loss": 0.033904530107975006, "train_loss_bc": 0.0338917076587677, "train_loss_llm": 0.0012820661067962646, "grad_norm": 0.018855754286050797, "global_step": 8073, "epoch": 224, "lr": 0.001} +{"train_loss": 0.04205235093832016, "train_loss_bc": 0.04205235093832016, "train_loss_llm": 0.0, "grad_norm": 0.03092399798333645, "global_step": 8074, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03778404742479324, "train_loss_bc": 0.03778404742479324, "train_loss_llm": 0.0, "grad_norm": 0.0401439368724823, "global_step": 8075, "epoch": 224, "lr": 0.001} +{"train_loss": 0.039416342973709106, "train_loss_bc": 0.03940997272729874, "train_loss_llm": 0.0006371736526489258, "grad_norm": 0.0485614538192749, "global_step": 8076, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03844067454338074, "train_loss_bc": 0.03844067454338074, "train_loss_llm": 0.0, "grad_norm": 0.018663747236132622, "global_step": 8077, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03925289586186409, "train_loss_bc": 0.03925289586186409, "train_loss_llm": 0.0, "grad_norm": 0.023989709094166756, "global_step": 8078, "epoch": 224, "lr": 0.001} +{"train_loss": 0.039107371121644974, "train_loss_bc": 0.039084985852241516, "train_loss_llm": 0.002238541841506958, "grad_norm": 0.03558897599577904, "global_step": 8079, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03935848921537399, "train_loss_bc": 0.03935848921537399, "train_loss_llm": 0.0, "grad_norm": 0.043357331305742264, "global_step": 8080, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03722758591175079, "train_loss_bc": 0.03722758591175079, "train_loss_llm": 0.0, "grad_norm": 0.021791528910398483, "global_step": 8081, "epoch": 224, "lr": 0.001} +{"train_loss": 0.04413264989852905, "train_loss_bc": 0.04413104057312012, "train_loss_llm": 0.00016108155250549316, "grad_norm": 0.03543667867779732, "global_step": 8082, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03743567690253258, "train_loss_bc": 0.03743567690253258, "train_loss_llm": 0.0, "grad_norm": 0.04479746147990227, "global_step": 8083, "epoch": 224, "lr": 0.001} +{"train_loss": 0.04217446222901344, "train_loss_bc": 0.04217446222901344, "train_loss_llm": 0.0, "grad_norm": 0.04873719811439514, "global_step": 8084, "epoch": 224, "lr": 0.001} +{"train_loss": 0.04396383836865425, "train_loss_bc": 0.04396383836865425, "train_loss_llm": 0.0, "grad_norm": 0.020695760846138, "global_step": 8085, "epoch": 224, "lr": 0.001} +{"train_loss": 0.04227118566632271, "train_loss_bc": 0.04227118566632271, "train_loss_llm": 0.0, "grad_norm": 0.030132072046399117, "global_step": 8086, "epoch": 224, "lr": 0.001} +{"train_loss": 0.038803331553936005, "train_loss_bc": 0.038803331553936005, "train_loss_llm": 0.0, "grad_norm": 0.03803602606058121, "global_step": 8087, "epoch": 224, "lr": 0.001} +{"train_loss": 0.05360962450504303, "train_loss_bc": 0.05360962450504303, "train_loss_llm": 0.0, "grad_norm": 0.047147367149591446, "global_step": 8088, "epoch": 224, "lr": 0.001} +{"train_loss": 0.04526059702038765, "train_loss_bc": 0.04526059702038765, "train_loss_llm": 0.0, "grad_norm": 0.021990491077303886, "global_step": 8089, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03765552118420601, "train_loss_bc": 0.03764292970299721, "train_loss_llm": 0.0012591779232025146, "grad_norm": 0.03101467527449131, "global_step": 8090, "epoch": 224, "lr": 0.001} +{"train_loss": 0.039756786078214645, "train_loss_bc": 0.039756786078214645, "train_loss_llm": 0.0, "grad_norm": 0.04369830712676048, "global_step": 8091, "epoch": 224, "lr": 0.001} +{"train_loss": 0.0445321649312973, "train_loss_bc": 0.04452997446060181, "train_loss_llm": 0.00021907687187194824, "grad_norm": 0.0557524599134922, "global_step": 8092, "epoch": 224, "lr": 0.001} +{"train_loss": 0.04340312257409096, "train_loss_bc": 0.04340312257409096, "train_loss_llm": 0.0, "grad_norm": 0.022370832040905952, "global_step": 8093, "epoch": 224, "lr": 0.001} +{"train_loss": 0.04338468238711357, "train_loss_bc": 0.04338468238711357, "train_loss_llm": 0.0, "grad_norm": 0.041374024003744125, "global_step": 8094, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03903070464730263, "train_loss_bc": 0.03903070464730263, "train_loss_llm": 0.0, "grad_norm": 0.05299907922744751, "global_step": 8095, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03291529417037964, "train_loss_bc": 0.03291529417037964, "train_loss_llm": 0.0, "grad_norm": 0.07008994370698929, "global_step": 8096, "epoch": 224, "lr": 0.001} +{"train_loss": 0.039453908801078796, "train_loss_bc": 0.039453908801078796, "train_loss_llm": 0.0, "grad_norm": 0.022579334676265717, "global_step": 8097, "epoch": 224, "lr": 0.001} +{"train_loss": 0.036045003682374954, "train_loss_bc": 0.036045003682374954, "train_loss_llm": 0.0, "grad_norm": 0.02978796325623989, "global_step": 8098, "epoch": 224, "lr": 0.001} +{"train_loss": 0.03985663689672947, "train_loss_bc": 0.04752178117632866, "train_loss_llm": 1.6689300537109375e-05, "grad_norm": 0.0395050086081028, "global_step": 8099, "epoch": 224, "lr": 0.001, "val_loss": 0.05901641398668289} +{"train_loss": 0.03547985106706619, "train_loss_bc": 0.03547985106706619, "train_loss_llm": 0.0, "grad_norm": 0.04574389010667801, "global_step": 8100, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03705224767327309, "train_loss_bc": 0.03705224767327309, "train_loss_llm": 0.0, "grad_norm": 0.017995256930589676, "global_step": 8101, "epoch": 225, "lr": 0.001} +{"train_loss": 0.042530957609415054, "train_loss_bc": 0.042530957609415054, "train_loss_llm": 0.0, "grad_norm": 0.030410286039114, "global_step": 8102, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03894105553627014, "train_loss_bc": 0.03893301635980606, "train_loss_llm": 0.0008037984371185303, "grad_norm": 0.037741828709840775, "global_step": 8103, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04438469558954239, "train_loss_bc": 0.04438469558954239, "train_loss_llm": 0.0, "grad_norm": 0.04719571769237518, "global_step": 8104, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04301578924059868, "train_loss_bc": 0.04301578924059868, "train_loss_llm": 0.0, "grad_norm": 0.023436041548848152, "global_step": 8105, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03535137698054314, "train_loss_bc": 0.03535137698054314, "train_loss_llm": 0.0, "grad_norm": 0.03295641019940376, "global_step": 8106, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03295985236763954, "train_loss_bc": 0.03295985236763954, "train_loss_llm": 0.0, "grad_norm": 0.038082342594861984, "global_step": 8107, "epoch": 225, "lr": 0.001} +{"train_loss": 0.037635378539562225, "train_loss_bc": 0.037635378539562225, "train_loss_llm": 0.0, "grad_norm": 0.045067280530929565, "global_step": 8108, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04386107623577118, "train_loss_bc": 0.0438527874648571, "train_loss_llm": 0.0008290410041809082, "grad_norm": 0.01659376360476017, "global_step": 8109, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04649529606103897, "train_loss_bc": 0.04649529606103897, "train_loss_llm": 0.0, "grad_norm": 0.025545261800289154, "global_step": 8110, "epoch": 225, "lr": 0.001} +{"train_loss": 0.046477869153022766, "train_loss_bc": 0.046477869153022766, "train_loss_llm": 0.0, "grad_norm": 0.0392187274992466, "global_step": 8111, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03507312387228012, "train_loss_bc": 0.03507312387228012, "train_loss_llm": 0.0, "grad_norm": 0.046872660517692566, "global_step": 8112, "epoch": 225, "lr": 0.001} +{"train_loss": 0.039877887815237045, "train_loss_bc": 0.039877887815237045, "train_loss_llm": 0.0, "grad_norm": 0.019280914217233658, "global_step": 8113, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03788505867123604, "train_loss_bc": 0.03786957636475563, "train_loss_llm": 0.001548081636428833, "grad_norm": 0.03369327262043953, "global_step": 8114, "epoch": 225, "lr": 0.001} +{"train_loss": 0.033765338361263275, "train_loss_bc": 0.03375718742609024, "train_loss_llm": 0.0008151233196258545, "grad_norm": 0.05009148642420769, "global_step": 8115, "epoch": 225, "lr": 0.001} +{"train_loss": 0.036370716989040375, "train_loss_bc": 0.03635028004646301, "train_loss_llm": 0.0020435750484466553, "grad_norm": 0.06835963577032089, "global_step": 8116, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04224710538983345, "train_loss_bc": 0.042246703058481216, "train_loss_llm": 4.029273986816406e-05, "grad_norm": 0.018267497420310974, "global_step": 8117, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04336196184158325, "train_loss_bc": 0.04334862157702446, "train_loss_llm": 0.0013339817523956299, "grad_norm": 0.026094892993569374, "global_step": 8118, "epoch": 225, "lr": 0.001} +{"train_loss": 0.040269188582897186, "train_loss_bc": 0.040269188582897186, "train_loss_llm": 0.0, "grad_norm": 0.03507896512746811, "global_step": 8119, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03924991935491562, "train_loss_bc": 0.03924991935491562, "train_loss_llm": 0.0, "grad_norm": 0.04413232579827309, "global_step": 8120, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04924963414669037, "train_loss_bc": 0.049221526831388474, "train_loss_llm": 0.002810865640640259, "grad_norm": 0.0242164209485054, "global_step": 8121, "epoch": 225, "lr": 0.001} +{"train_loss": 0.038320526480674744, "train_loss_bc": 0.038320526480674744, "train_loss_llm": 0.0, "grad_norm": 0.033902354538440704, "global_step": 8122, "epoch": 225, "lr": 0.001} +{"train_loss": 0.046068474650382996, "train_loss_bc": 0.046068474650382996, "train_loss_llm": 0.0, "grad_norm": 0.042436812072992325, "global_step": 8123, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04102898761630058, "train_loss_bc": 0.04102898761630058, "train_loss_llm": 0.0, "grad_norm": 0.05409559980034828, "global_step": 8124, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03676446154713631, "train_loss_bc": 0.03675653412938118, "train_loss_llm": 0.000792771577835083, "grad_norm": 0.018832391127943993, "global_step": 8125, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04072262719273567, "train_loss_bc": 0.04072262719273567, "train_loss_llm": 0.0, "grad_norm": 0.030580362305045128, "global_step": 8126, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03797418624162674, "train_loss_bc": 0.037972137331962585, "train_loss_llm": 0.00020495057106018066, "grad_norm": 0.041236504912376404, "global_step": 8127, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04158160462975502, "train_loss_bc": 0.04158160462975502, "train_loss_llm": 0.0, "grad_norm": 0.05018584057688713, "global_step": 8128, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04092545807361603, "train_loss_bc": 0.04092545807361603, "train_loss_llm": 0.0, "grad_norm": 0.019207678735256195, "global_step": 8129, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04017442837357521, "train_loss_bc": 0.040160320699214935, "train_loss_llm": 0.001410841941833496, "grad_norm": 0.0272049643099308, "global_step": 8130, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03196995332837105, "train_loss_bc": 0.03196749836206436, "train_loss_llm": 0.0002455413341522217, "grad_norm": 0.03725847601890564, "global_step": 8131, "epoch": 225, "lr": 0.001} +{"train_loss": 0.033146318048238754, "train_loss_bc": 0.03314404934644699, "train_loss_llm": 0.00022682547569274902, "grad_norm": 0.0486937090754509, "global_step": 8132, "epoch": 225, "lr": 0.001} +{"train_loss": 0.04169592261314392, "train_loss_bc": 0.04169592261314392, "train_loss_llm": 0.0, "grad_norm": 0.021405845880508423, "global_step": 8133, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03789500892162323, "train_loss_bc": 0.03787371516227722, "train_loss_llm": 0.002129197120666504, "grad_norm": 0.0358416922390461, "global_step": 8134, "epoch": 225, "lr": 0.001} +{"train_loss": 0.03981537309785684, "train_loss_bc": 0.04355009272694588, "train_loss_llm": 0.0, "grad_norm": 0.050763681530952454, "global_step": 8135, "epoch": 225, "lr": 0.001, "train/cumulative_reward": 174.05974115412968, "train/mean_score": 18.93781270364185, "train/success_rate": 0.4, "test/cumulative_reward": 65.21131775133267, "test/mean_score": 10.097471417147021, "test/success_rate": 0.22, "val_loss": 0.05767197161912918, "train_action_mse_error": 0.04314345121383667} +{"train_loss": 0.03812705725431442, "train_loss_bc": 0.03809874877333641, "train_loss_llm": 0.0028308331966400146, "grad_norm": 0.06392721831798553, "global_step": 8136, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04205360263586044, "train_loss_bc": 0.04205360263586044, "train_loss_llm": 0.0, "grad_norm": 0.02281152829527855, "global_step": 8137, "epoch": 226, "lr": 0.001} +{"train_loss": 0.03939144313335419, "train_loss_bc": 0.03938610851764679, "train_loss_llm": 0.0005333125591278076, "grad_norm": 0.032825786620378494, "global_step": 8138, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04091862961649895, "train_loss_bc": 0.04091862961649895, "train_loss_llm": 0.0, "grad_norm": 0.0444793738424778, "global_step": 8139, "epoch": 226, "lr": 0.001} +{"train_loss": 0.040536969900131226, "train_loss_bc": 0.040536969900131226, "train_loss_llm": 0.0, "grad_norm": 0.056165147572755814, "global_step": 8140, "epoch": 226, "lr": 0.001} +{"train_loss": 0.042316872626543045, "train_loss_bc": 0.042316872626543045, "train_loss_llm": 0.0, "grad_norm": 0.021509550511837006, "global_step": 8141, "epoch": 226, "lr": 0.001} +{"train_loss": 0.037580255419015884, "train_loss_bc": 0.03755757585167885, "train_loss_llm": 0.002267986536026001, "grad_norm": 0.030840544030070305, "global_step": 8142, "epoch": 226, "lr": 0.001} +{"train_loss": 0.03813813626766205, "train_loss_bc": 0.03813813626766205, "train_loss_llm": 0.0, "grad_norm": 0.04147212207317352, "global_step": 8143, "epoch": 226, "lr": 0.001} +{"train_loss": 0.03971711918711662, "train_loss_bc": 0.03971711918711662, "train_loss_llm": 0.0, "grad_norm": 0.05932368338108063, "global_step": 8144, "epoch": 226, "lr": 0.001} +{"train_loss": 0.039323125034570694, "train_loss_bc": 0.039323125034570694, "train_loss_llm": 0.0, "grad_norm": 0.027936069294810295, "global_step": 8145, "epoch": 226, "lr": 0.001} +{"train_loss": 0.03662026301026344, "train_loss_bc": 0.03662026301026344, "train_loss_llm": 0.0, "grad_norm": 0.046860240399837494, "global_step": 8146, "epoch": 226, "lr": 0.001} +{"train_loss": 0.03416023403406143, "train_loss_bc": 0.03415531665086746, "train_loss_llm": 0.0004919171333312988, "grad_norm": 0.05731064826250076, "global_step": 8147, "epoch": 226, "lr": 0.001} +{"train_loss": 0.041700951755046844, "train_loss_bc": 0.041700951755046844, "train_loss_llm": 0.0, "grad_norm": 0.07295016944408417, "global_step": 8148, "epoch": 226, "lr": 0.001} +{"train_loss": 0.039394672960042953, "train_loss_bc": 0.039394672960042953, "train_loss_llm": 0.0, "grad_norm": 0.02285049296915531, "global_step": 8149, "epoch": 226, "lr": 0.001} +{"train_loss": 0.033320773392915726, "train_loss_bc": 0.03331965208053589, "train_loss_llm": 0.00011226534843444824, "grad_norm": 0.045035094022750854, "global_step": 8150, "epoch": 226, "lr": 0.001} +{"train_loss": 0.040942028164863586, "train_loss_bc": 0.040942028164863586, "train_loss_llm": 0.0, "grad_norm": 0.0685778260231018, "global_step": 8151, "epoch": 226, "lr": 0.001} +{"train_loss": 0.0374179445207119, "train_loss_bc": 0.0374179445207119, "train_loss_llm": 0.0, "grad_norm": 0.0779423639178276, "global_step": 8152, "epoch": 226, "lr": 0.001} +{"train_loss": 0.042274359613657, "train_loss_bc": 0.042274359613657, "train_loss_llm": 0.0, "grad_norm": 0.023251302540302277, "global_step": 8153, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04522499442100525, "train_loss_bc": 0.04522499442100525, "train_loss_llm": 0.0, "grad_norm": 0.032547615468502045, "global_step": 8154, "epoch": 226, "lr": 0.001} +{"train_loss": 0.0445994958281517, "train_loss_bc": 0.0445994958281517, "train_loss_llm": 0.0, "grad_norm": 0.053454913198947906, "global_step": 8155, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04125340282917023, "train_loss_bc": 0.04125340282917023, "train_loss_llm": 0.0, "grad_norm": 0.07737113535404205, "global_step": 8156, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04042193666100502, "train_loss_bc": 0.04042193666100502, "train_loss_llm": 0.0, "grad_norm": 0.019076326861977577, "global_step": 8157, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04242474585771561, "train_loss_bc": 0.04242474585771561, "train_loss_llm": 0.0, "grad_norm": 0.039539121091365814, "global_step": 8158, "epoch": 226, "lr": 0.001} +{"train_loss": 0.0454462394118309, "train_loss_bc": 0.0454462394118309, "train_loss_llm": 0.0, "grad_norm": 0.05157314985990524, "global_step": 8159, "epoch": 226, "lr": 0.001} +{"train_loss": 0.040112171322107315, "train_loss_bc": 0.040112171322107315, "train_loss_llm": 0.0, "grad_norm": 0.06704163551330566, "global_step": 8160, "epoch": 226, "lr": 0.001} +{"train_loss": 0.041038405150175095, "train_loss_bc": 0.041038405150175095, "train_loss_llm": 0.0, "grad_norm": 0.017177188768982887, "global_step": 8161, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04267435520887375, "train_loss_bc": 0.04267430678009987, "train_loss_llm": 4.678964614868164e-06, "grad_norm": 0.02656659111380577, "global_step": 8162, "epoch": 226, "lr": 0.001} +{"train_loss": 0.032513733953237534, "train_loss_bc": 0.032513733953237534, "train_loss_llm": 0.0, "grad_norm": 0.03866996988654137, "global_step": 8163, "epoch": 226, "lr": 0.001} +{"train_loss": 0.041971355676651, "train_loss_bc": 0.041935402899980545, "train_loss_llm": 0.0035953521728515625, "grad_norm": 0.04662550985813141, "global_step": 8164, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04916679486632347, "train_loss_bc": 0.0491529256105423, "train_loss_llm": 0.0013867616653442383, "grad_norm": 0.024936003610491753, "global_step": 8165, "epoch": 226, "lr": 0.001} +{"train_loss": 0.03455915302038193, "train_loss_bc": 0.03455203399062157, "train_loss_llm": 0.0007117688655853271, "grad_norm": 0.03194236010313034, "global_step": 8166, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04046665504574776, "train_loss_bc": 0.04045331850647926, "train_loss_llm": 0.001333773136138916, "grad_norm": 0.047825608402490616, "global_step": 8167, "epoch": 226, "lr": 0.001} +{"train_loss": 0.03645201772451401, "train_loss_bc": 0.036423977464437485, "train_loss_llm": 0.0028040409088134766, "grad_norm": 0.07417180389165878, "global_step": 8168, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04252777248620987, "train_loss_bc": 0.04252777248620987, "train_loss_llm": 0.0, "grad_norm": 0.027617881074547768, "global_step": 8169, "epoch": 226, "lr": 0.001} +{"train_loss": 0.03715761750936508, "train_loss_bc": 0.03715761750936508, "train_loss_llm": 0.0, "grad_norm": 0.03614773973822594, "global_step": 8170, "epoch": 226, "lr": 0.001} +{"train_loss": 0.04008312492320935, "train_loss_bc": 0.04103940725326538, "train_loss_llm": 0.0007803142070770264, "grad_norm": 0.051033373922109604, "global_step": 8171, "epoch": 226, "lr": 0.001, "val_loss": 0.058712586760520935} +{"train_loss": 0.03930731862783432, "train_loss_bc": 0.03928963840007782, "train_loss_llm": 0.001767873764038086, "grad_norm": 0.07147020846605301, "global_step": 8172, "epoch": 227, "lr": 0.001} +{"train_loss": 0.038103945553302765, "train_loss_bc": 0.038103945553302765, "train_loss_llm": 0.0, "grad_norm": 0.017650146037340164, "global_step": 8173, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03638448938727379, "train_loss_bc": 0.03634899482131004, "train_loss_llm": 0.003549337387084961, "grad_norm": 0.03850540891289711, "global_step": 8174, "epoch": 227, "lr": 0.001} +{"train_loss": 0.045865777879953384, "train_loss_bc": 0.045865777879953384, "train_loss_llm": 0.0, "grad_norm": 0.055324405431747437, "global_step": 8175, "epoch": 227, "lr": 0.001} +{"train_loss": 0.04565010964870453, "train_loss_bc": 0.04565010964870453, "train_loss_llm": 0.0, "grad_norm": 0.06756559759378433, "global_step": 8176, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03757546469569206, "train_loss_bc": 0.03757546469569206, "train_loss_llm": 0.0, "grad_norm": 0.019251514226198196, "global_step": 8177, "epoch": 227, "lr": 0.001} +{"train_loss": 0.04533960297703743, "train_loss_bc": 0.04533599689602852, "train_loss_llm": 0.00036078691482543945, "grad_norm": 0.03507431223988533, "global_step": 8178, "epoch": 227, "lr": 0.001} +{"train_loss": 0.04467558115720749, "train_loss_bc": 0.04460453242063522, "train_loss_llm": 0.007104963064193726, "grad_norm": 0.05134804919362068, "global_step": 8179, "epoch": 227, "lr": 0.001} +{"train_loss": 0.041377101093530655, "train_loss_bc": 0.041377101093530655, "train_loss_llm": 0.0, "grad_norm": 0.07061908394098282, "global_step": 8180, "epoch": 227, "lr": 0.001} +{"train_loss": 0.037518396973609924, "train_loss_bc": 0.03748767077922821, "train_loss_llm": 0.0030724704265594482, "grad_norm": 0.017107577994465828, "global_step": 8181, "epoch": 227, "lr": 0.001} +{"train_loss": 0.029651900753378868, "train_loss_bc": 0.029651900753378868, "train_loss_llm": 0.0, "grad_norm": 0.024499014019966125, "global_step": 8182, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03651633486151695, "train_loss_bc": 0.03651633486151695, "train_loss_llm": 0.0, "grad_norm": 0.02808559127151966, "global_step": 8183, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03564672917127609, "train_loss_bc": 0.03564672917127609, "train_loss_llm": 0.0, "grad_norm": 0.03330400213599205, "global_step": 8184, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03599371016025543, "train_loss_bc": 0.03598000109195709, "train_loss_llm": 0.0013709068298339844, "grad_norm": 0.018186314031481743, "global_step": 8185, "epoch": 227, "lr": 0.001} +{"train_loss": 0.0441829152405262, "train_loss_bc": 0.0441829152405262, "train_loss_llm": 0.0, "grad_norm": 0.03214085474610329, "global_step": 8186, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03742220252752304, "train_loss_bc": 0.03742220252752304, "train_loss_llm": 0.0, "grad_norm": 0.040547970682382584, "global_step": 8187, "epoch": 227, "lr": 0.001} +{"train_loss": 0.034677743911743164, "train_loss_bc": 0.034677743911743164, "train_loss_llm": 0.0, "grad_norm": 0.04662741720676422, "global_step": 8188, "epoch": 227, "lr": 0.001} +{"train_loss": 0.044384606182575226, "train_loss_bc": 0.044384606182575226, "train_loss_llm": 0.0, "grad_norm": 0.024599678814411163, "global_step": 8189, "epoch": 227, "lr": 0.001} +{"train_loss": 0.0418066643178463, "train_loss_bc": 0.0418066643178463, "train_loss_llm": 0.0, "grad_norm": 0.04684661328792572, "global_step": 8190, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03206947445869446, "train_loss_bc": 0.03206705302000046, "train_loss_llm": 0.00024199485778808594, "grad_norm": 0.06566029787063599, "global_step": 8191, "epoch": 227, "lr": 0.001} +{"train_loss": 0.04617541655898094, "train_loss_bc": 0.04617541655898094, "train_loss_llm": 0.0, "grad_norm": 0.08327507972717285, "global_step": 8192, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03442218899726868, "train_loss_bc": 0.03442218899726868, "train_loss_llm": 0.0, "grad_norm": 0.01831013895571232, "global_step": 8193, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03757213428616524, "train_loss_bc": 0.03757213428616524, "train_loss_llm": 0.0, "grad_norm": 0.03150462731719017, "global_step": 8194, "epoch": 227, "lr": 0.001} +{"train_loss": 0.044636450707912445, "train_loss_bc": 0.044636450707912445, "train_loss_llm": 0.0, "grad_norm": 0.04695141687989235, "global_step": 8195, "epoch": 227, "lr": 0.001} +{"train_loss": 0.043492551892995834, "train_loss_bc": 0.043488774448633194, "train_loss_llm": 0.0003777742385864258, "grad_norm": 0.071098692715168, "global_step": 8196, "epoch": 227, "lr": 0.001} +{"train_loss": 0.040706198662519455, "train_loss_bc": 0.040706198662519455, "train_loss_llm": 0.0, "grad_norm": 0.028794344514608383, "global_step": 8197, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03483312577009201, "train_loss_bc": 0.03483312577009201, "train_loss_llm": 0.0, "grad_norm": 0.0412743054330349, "global_step": 8198, "epoch": 227, "lr": 0.001} +{"train_loss": 0.04077117517590523, "train_loss_bc": 0.04076571762561798, "train_loss_llm": 0.0005456209182739258, "grad_norm": 0.05997936800122261, "global_step": 8199, "epoch": 227, "lr": 0.001} +{"train_loss": 0.042171504348516464, "train_loss_bc": 0.042171504348516464, "train_loss_llm": 0.0, "grad_norm": 0.07658828794956207, "global_step": 8200, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03712380304932594, "train_loss_bc": 0.03712380304932594, "train_loss_llm": 0.0, "grad_norm": 0.023069340735673904, "global_step": 8201, "epoch": 227, "lr": 0.001} +{"train_loss": 0.039584703743457794, "train_loss_bc": 0.03958117216825485, "train_loss_llm": 0.00035318732261657715, "grad_norm": 0.03136071562767029, "global_step": 8202, "epoch": 227, "lr": 0.001} +{"train_loss": 0.04507974907755852, "train_loss_bc": 0.04507974907755852, "train_loss_llm": 0.0, "grad_norm": 0.04280179366469383, "global_step": 8203, "epoch": 227, "lr": 0.001} +{"train_loss": 0.03689011186361313, "train_loss_bc": 0.03689011186361313, "train_loss_llm": 0.0, "grad_norm": 0.05043447017669678, "global_step": 8204, "epoch": 227, "lr": 0.001} +{"train_loss": 0.04396653547883034, "train_loss_bc": 0.04396653547883034, "train_loss_llm": 0.0, "grad_norm": 0.019658366218209267, "global_step": 8205, "epoch": 227, "lr": 0.001} +{"train_loss": 0.046231459826231, "train_loss_bc": 0.046231459826231, "train_loss_llm": 0.0, "grad_norm": 0.03472979739308357, "global_step": 8206, "epoch": 227, "lr": 0.001} +{"train_loss": 0.040025378939592175, "train_loss_bc": 0.04310646280646324, "train_loss_llm": 0.0, "grad_norm": 0.04200915992259979, "global_step": 8207, "epoch": 227, "lr": 0.001, "val_loss": 0.05820652097463608} +{"train_loss": 0.0361073836684227, "train_loss_bc": 0.0361073836684227, "train_loss_llm": 0.0, "grad_norm": 0.05122436210513115, "global_step": 8208, "epoch": 228, "lr": 0.001} +{"train_loss": 0.0340893492102623, "train_loss_bc": 0.0340893492102623, "train_loss_llm": 0.0, "grad_norm": 0.017703913152217865, "global_step": 8209, "epoch": 228, "lr": 0.001} +{"train_loss": 0.04036413133144379, "train_loss_bc": 0.04036413133144379, "train_loss_llm": 0.0, "grad_norm": 0.03976687416434288, "global_step": 8210, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03799819573760033, "train_loss_bc": 0.03799322620034218, "train_loss_llm": 0.0004968941211700439, "grad_norm": 0.056786153465509415, "global_step": 8211, "epoch": 228, "lr": 0.001} +{"train_loss": 0.048534099012613297, "train_loss_bc": 0.04852363467216492, "train_loss_llm": 0.0010464489459991455, "grad_norm": 0.07871539145708084, "global_step": 8212, "epoch": 228, "lr": 0.001} +{"train_loss": 0.030282143503427505, "train_loss_bc": 0.030253605917096138, "train_loss_llm": 0.0028537511825561523, "grad_norm": 0.017721563577651978, "global_step": 8213, "epoch": 228, "lr": 0.001} +{"train_loss": 0.045468468219041824, "train_loss_bc": 0.045468468219041824, "train_loss_llm": 0.0, "grad_norm": 0.03836888447403908, "global_step": 8214, "epoch": 228, "lr": 0.001} +{"train_loss": 0.036533698439598083, "train_loss_bc": 0.036533698439598083, "train_loss_llm": 0.0, "grad_norm": 0.04592839628458023, "global_step": 8215, "epoch": 228, "lr": 0.001} +{"train_loss": 0.039163924753665924, "train_loss_bc": 0.039163924753665924, "train_loss_llm": 0.0, "grad_norm": 0.060528527945280075, "global_step": 8216, "epoch": 228, "lr": 0.001} +{"train_loss": 0.029146650806069374, "train_loss_bc": 0.02913050353527069, "train_loss_llm": 0.0016148090362548828, "grad_norm": 0.02213410474359989, "global_step": 8217, "epoch": 228, "lr": 0.001} +{"train_loss": 0.04621856287121773, "train_loss_bc": 0.04621856287121773, "train_loss_llm": 0.0, "grad_norm": 0.033943116664886475, "global_step": 8218, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03691107779741287, "train_loss_bc": 0.03691107779741287, "train_loss_llm": 0.0, "grad_norm": 0.04655062034726143, "global_step": 8219, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03729823976755142, "train_loss_bc": 0.03727063909173012, "train_loss_llm": 0.0027601420879364014, "grad_norm": 0.05341978371143341, "global_step": 8220, "epoch": 228, "lr": 0.001} +{"train_loss": 0.042743369936943054, "train_loss_bc": 0.04273100942373276, "train_loss_llm": 0.0012360215187072754, "grad_norm": 0.02074645832180977, "global_step": 8221, "epoch": 228, "lr": 0.001} +{"train_loss": 0.0417785681784153, "train_loss_bc": 0.0417616181075573, "train_loss_llm": 0.0016951262950897217, "grad_norm": 0.03238707780838013, "global_step": 8222, "epoch": 228, "lr": 0.001} +{"train_loss": 0.04430164396762848, "train_loss_bc": 0.04430164396762848, "train_loss_llm": 0.0, "grad_norm": 0.04093689098954201, "global_step": 8223, "epoch": 228, "lr": 0.001} +{"train_loss": 0.04297191649675369, "train_loss_bc": 0.04297191649675369, "train_loss_llm": 0.0, "grad_norm": 0.0482589527964592, "global_step": 8224, "epoch": 228, "lr": 0.001} +{"train_loss": 0.042914409190416336, "train_loss_bc": 0.042914409190416336, "train_loss_llm": 0.0, "grad_norm": 0.024803616106510162, "global_step": 8225, "epoch": 228, "lr": 0.001} +{"train_loss": 0.036956898868083954, "train_loss_bc": 0.03692790865898132, "train_loss_llm": 0.0028989017009735107, "grad_norm": 0.03990554064512253, "global_step": 8226, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03507980704307556, "train_loss_bc": 0.03507980704307556, "train_loss_llm": 0.0, "grad_norm": 0.05302223563194275, "global_step": 8227, "epoch": 228, "lr": 0.001} +{"train_loss": 0.04180563613772392, "train_loss_bc": 0.04180563613772392, "train_loss_llm": 0.0, "grad_norm": 0.07476751506328583, "global_step": 8228, "epoch": 228, "lr": 0.001} +{"train_loss": 0.043063946068286896, "train_loss_bc": 0.043063946068286896, "train_loss_llm": 0.0, "grad_norm": 0.025257648900151253, "global_step": 8229, "epoch": 228, "lr": 0.001} +{"train_loss": 0.04139188304543495, "train_loss_bc": 0.04139188304543495, "train_loss_llm": 0.0, "grad_norm": 0.04118096083402634, "global_step": 8230, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03983217105269432, "train_loss_bc": 0.03983217105269432, "train_loss_llm": 0.0, "grad_norm": 0.05293959751725197, "global_step": 8231, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03928234055638313, "train_loss_bc": 0.03928234055638313, "train_loss_llm": 0.0, "grad_norm": 0.07279898971319199, "global_step": 8232, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03702552989125252, "train_loss_bc": 0.03702368214726448, "train_loss_llm": 0.00018471479415893555, "grad_norm": 0.021743884310126305, "global_step": 8233, "epoch": 228, "lr": 0.001} +{"train_loss": 0.04747645929455757, "train_loss_bc": 0.047458548098802567, "train_loss_llm": 0.0017910301685333252, "grad_norm": 0.031096767634153366, "global_step": 8234, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03586087375879288, "train_loss_bc": 0.03586087375879288, "train_loss_llm": 0.0, "grad_norm": 0.042482372373342514, "global_step": 8235, "epoch": 228, "lr": 0.001} +{"train_loss": 0.039484404027462006, "train_loss_bc": 0.03947487100958824, "train_loss_llm": 0.0009532570838928223, "grad_norm": 0.05399550870060921, "global_step": 8236, "epoch": 228, "lr": 0.001} +{"train_loss": 0.04161594808101654, "train_loss_bc": 0.04161594808101654, "train_loss_llm": 0.0, "grad_norm": 0.01962040178477764, "global_step": 8237, "epoch": 228, "lr": 0.001} +{"train_loss": 0.041639991104602814, "train_loss_bc": 0.04163803532719612, "train_loss_llm": 0.00019571185111999512, "grad_norm": 0.030315525829792023, "global_step": 8238, "epoch": 228, "lr": 0.001} +{"train_loss": 0.04860818013548851, "train_loss_bc": 0.04860818013548851, "train_loss_llm": 0.0, "grad_norm": 0.03752756863832474, "global_step": 8239, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03944670781493187, "train_loss_bc": 0.03944670781493187, "train_loss_llm": 0.0, "grad_norm": 0.05114490166306496, "global_step": 8240, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03807505965232849, "train_loss_bc": 0.03807505965232849, "train_loss_llm": 0.0, "grad_norm": 0.023172961547970772, "global_step": 8241, "epoch": 228, "lr": 0.001} +{"train_loss": 0.03530946001410484, "train_loss_bc": 0.03530946001410484, "train_loss_llm": 0.0, "grad_norm": 0.03468500077724457, "global_step": 8242, "epoch": 228, "lr": 0.001} +{"train_loss": 0.039899431427733764, "train_loss_bc": 0.04158756881952286, "train_loss_llm": 0.0010832250118255615, "grad_norm": 0.04265445098280907, "global_step": 8243, "epoch": 228, "lr": 0.001, "val_loss": 0.057738132774829865} +{"train_loss": 0.037157438695430756, "train_loss_bc": 0.037157438695430756, "train_loss_llm": 0.0, "grad_norm": 0.050660256296396255, "global_step": 8244, "epoch": 229, "lr": 0.001} +{"train_loss": 0.04363837093114853, "train_loss_bc": 0.04363008216023445, "train_loss_llm": 0.0008288323879241943, "grad_norm": 0.02229509875178337, "global_step": 8245, "epoch": 229, "lr": 0.001} +{"train_loss": 0.04254646971821785, "train_loss_bc": 0.04254646971821785, "train_loss_llm": 0.0, "grad_norm": 0.031918805092573166, "global_step": 8246, "epoch": 229, "lr": 0.001} +{"train_loss": 0.046412136405706406, "train_loss_bc": 0.046364959329366684, "train_loss_llm": 0.0047178566455841064, "grad_norm": 0.03645139932632446, "global_step": 8247, "epoch": 229, "lr": 0.001} +{"train_loss": 0.034689974039793015, "train_loss_bc": 0.03467915579676628, "train_loss_llm": 0.0010818839073181152, "grad_norm": 0.050433531403541565, "global_step": 8248, "epoch": 229, "lr": 0.001} +{"train_loss": 0.041540369391441345, "train_loss_bc": 0.04152340441942215, "train_loss_llm": 0.0016965866088867188, "grad_norm": 0.01945425197482109, "global_step": 8249, "epoch": 229, "lr": 0.001} +{"train_loss": 0.03275603428483009, "train_loss_bc": 0.032754361629486084, "train_loss_llm": 0.00016710162162780762, "grad_norm": 0.0342620424926281, "global_step": 8250, "epoch": 229, "lr": 0.001} +{"train_loss": 0.039057496935129166, "train_loss_bc": 0.039044126868247986, "train_loss_llm": 0.0013369917869567871, "grad_norm": 0.05195041000843048, "global_step": 8251, "epoch": 229, "lr": 0.001} +{"train_loss": 0.03729095309972763, "train_loss_bc": 0.0372888520359993, "train_loss_llm": 0.00021028518676757812, "grad_norm": 0.0619710348546505, "global_step": 8252, "epoch": 229, "lr": 0.001} +{"train_loss": 0.042240291833877563, "train_loss_bc": 0.042240291833877563, "train_loss_llm": 0.0, "grad_norm": 0.02790171280503273, "global_step": 8253, "epoch": 229, "lr": 0.001} +{"train_loss": 0.03760489076375961, "train_loss_bc": 0.03760489076375961, "train_loss_llm": 0.0, "grad_norm": 0.0416184663772583, "global_step": 8254, "epoch": 229, "lr": 0.001} +{"train_loss": 0.04470525309443474, "train_loss_bc": 0.04470525309443474, "train_loss_llm": 0.0, "grad_norm": 0.06029190495610237, "global_step": 8255, "epoch": 229, "lr": 0.001} +{"train_loss": 0.0331168994307518, "train_loss_bc": 0.03311355412006378, "train_loss_llm": 0.0003343820571899414, "grad_norm": 0.06506113708019257, "global_step": 8256, "epoch": 229, "lr": 0.001} +{"train_loss": 0.039931513369083405, "train_loss_bc": 0.039931513369083405, "train_loss_llm": 0.0, "grad_norm": 0.018909865990281105, "global_step": 8257, "epoch": 229, "lr": 0.001} +{"train_loss": 0.042220521718263626, "train_loss_bc": 0.0422186404466629, "train_loss_llm": 0.0001881718635559082, "grad_norm": 0.022307850420475006, "global_step": 8258, "epoch": 229, "lr": 0.001} +{"train_loss": 0.042660877108573914, "train_loss_bc": 0.042660877108573914, "train_loss_llm": 0.0, "grad_norm": 0.03331654891371727, "global_step": 8259, "epoch": 229, "lr": 0.001} +{"train_loss": 0.042252663522958755, "train_loss_bc": 0.042252663522958755, "train_loss_llm": 0.0, "grad_norm": 0.03948865830898285, "global_step": 8260, "epoch": 229, "lr": 0.001} +{"train_loss": 0.0332123339176178, "train_loss_bc": 0.0332123339176178, "train_loss_llm": 0.0, "grad_norm": 0.019906051456928253, "global_step": 8261, "epoch": 229, "lr": 0.001} +{"train_loss": 0.03912407532334328, "train_loss_bc": 0.03912407532334328, "train_loss_llm": 0.0, "grad_norm": 0.03099140338599682, "global_step": 8262, "epoch": 229, "lr": 0.001} +{"train_loss": 0.04113909974694252, "train_loss_bc": 0.04113909974694252, "train_loss_llm": 0.0, "grad_norm": 0.04707913473248482, "global_step": 8263, "epoch": 229, "lr": 0.001} +{"train_loss": 0.04616696387529373, "train_loss_bc": 0.04616696387529373, "train_loss_llm": 0.0, "grad_norm": 0.05680074915289879, "global_step": 8264, "epoch": 229, "lr": 0.001} +{"train_loss": 0.045570679008960724, "train_loss_bc": 0.045570679008960724, "train_loss_llm": 0.0, "grad_norm": 0.021430741995573044, "global_step": 8265, "epoch": 229, "lr": 0.001} +{"train_loss": 0.040058813989162445, "train_loss_bc": 0.040058813989162445, "train_loss_llm": 0.0, "grad_norm": 0.03179642930626869, "global_step": 8266, "epoch": 229, "lr": 0.001} +{"train_loss": 0.04416663944721222, "train_loss_bc": 0.04416663944721222, "train_loss_llm": 0.0, "grad_norm": 0.04236213117837906, "global_step": 8267, "epoch": 229, "lr": 0.001} +{"train_loss": 0.030933164060115814, "train_loss_bc": 0.030933164060115814, "train_loss_llm": 0.0, "grad_norm": 0.052376311272382736, "global_step": 8268, "epoch": 229, "lr": 0.001} +{"train_loss": 0.04311636835336685, "train_loss_bc": 0.04310750216245651, "train_loss_llm": 0.0008865296840667725, "grad_norm": 0.01620236039161682, "global_step": 8269, "epoch": 229, "lr": 0.001} +{"train_loss": 0.03993602469563484, "train_loss_bc": 0.03993602469563484, "train_loss_llm": 0.0, "grad_norm": 0.02600063942372799, "global_step": 8270, "epoch": 229, "lr": 0.001} +{"train_loss": 0.037218157202005386, "train_loss_bc": 0.037218157202005386, "train_loss_llm": 0.0, "grad_norm": 0.03812912479043007, "global_step": 8271, "epoch": 229, "lr": 0.001} +{"train_loss": 0.04005010798573494, "train_loss_bc": 0.04005010798573494, "train_loss_llm": 0.0, "grad_norm": 0.04708999767899513, "global_step": 8272, "epoch": 229, "lr": 0.001} +{"train_loss": 0.040142498910427094, "train_loss_bc": 0.040142498910427094, "train_loss_llm": 0.0, "grad_norm": 0.018847523257136345, "global_step": 8273, "epoch": 229, "lr": 0.001} +{"train_loss": 0.038831863552331924, "train_loss_bc": 0.038831863552331924, "train_loss_llm": 0.0, "grad_norm": 0.0326274111866951, "global_step": 8274, "epoch": 229, "lr": 0.001} +{"train_loss": 0.036936648190021515, "train_loss_bc": 0.036936648190021515, "train_loss_llm": 0.0, "grad_norm": 0.036111436784267426, "global_step": 8275, "epoch": 229, "lr": 0.001} +{"train_loss": 0.03626110777258873, "train_loss_bc": 0.03626110777258873, "train_loss_llm": 0.0, "grad_norm": 0.04834061115980148, "global_step": 8276, "epoch": 229, "lr": 0.001} +{"train_loss": 0.041350431740283966, "train_loss_bc": 0.041350431740283966, "train_loss_llm": 0.0, "grad_norm": 0.01670997031033039, "global_step": 8277, "epoch": 229, "lr": 0.001} +{"train_loss": 0.04030534625053406, "train_loss_bc": 0.04029681161046028, "train_loss_llm": 0.0008533298969268799, "grad_norm": 0.034418314695358276, "global_step": 8278, "epoch": 229, "lr": 0.001} +{"train_loss": 0.03968091422898902, "train_loss_bc": 0.03417043387889862, "train_loss_llm": 0.0, "grad_norm": 0.05322686955332756, "global_step": 8279, "epoch": 229, "lr": 0.001, "val_loss": 0.058874040842056274} +{"train_loss": 0.042957257479429245, "train_loss_bc": 0.042957257479429245, "train_loss_llm": 0.0, "grad_norm": 0.06130773574113846, "global_step": 8280, "epoch": 230, "lr": 0.001} +{"train_loss": 0.0491364486515522, "train_loss_bc": 0.0491364486515522, "train_loss_llm": 0.0, "grad_norm": 0.028175901621580124, "global_step": 8281, "epoch": 230, "lr": 0.001} +{"train_loss": 0.04218002036213875, "train_loss_bc": 0.04218002036213875, "train_loss_llm": 0.0, "grad_norm": 0.03718540072441101, "global_step": 8282, "epoch": 230, "lr": 0.001} +{"train_loss": 0.03740565478801727, "train_loss_bc": 0.03740565478801727, "train_loss_llm": 0.0, "grad_norm": 0.0466456264257431, "global_step": 8283, "epoch": 230, "lr": 0.001} +{"train_loss": 0.04256877303123474, "train_loss_bc": 0.04256877303123474, "train_loss_llm": 0.0, "grad_norm": 0.0583287812769413, "global_step": 8284, "epoch": 230, "lr": 0.001} +{"train_loss": 0.0342799611389637, "train_loss_bc": 0.0342799611389637, "train_loss_llm": 0.0, "grad_norm": 0.020430846139788628, "global_step": 8285, "epoch": 230, "lr": 0.001} +{"train_loss": 0.04150848835706711, "train_loss_bc": 0.04150848835706711, "train_loss_llm": 0.0, "grad_norm": 0.02823820523917675, "global_step": 8286, "epoch": 230, "lr": 0.001} +{"train_loss": 0.03546392172574997, "train_loss_bc": 0.03546392172574997, "train_loss_llm": 0.0, "grad_norm": 0.03233408182859421, "global_step": 8287, "epoch": 230, "lr": 0.001} +{"train_loss": 0.03943672776222229, "train_loss_bc": 0.03941969946026802, "train_loss_llm": 0.0017028748989105225, "grad_norm": 0.04216959327459335, "global_step": 8288, "epoch": 230, "lr": 0.001} +{"train_loss": 0.031165478751063347, "train_loss_bc": 0.031165478751063347, "train_loss_llm": 0.0, "grad_norm": 0.016562875360250473, "global_step": 8289, "epoch": 230, "lr": 0.001} +{"train_loss": 0.038046035915613174, "train_loss_bc": 0.0380355529487133, "train_loss_llm": 0.0010484158992767334, "grad_norm": 0.03045729734003544, "global_step": 8290, "epoch": 230, "lr": 0.001} +{"train_loss": 0.03558206930756569, "train_loss_bc": 0.03557012975215912, "train_loss_llm": 0.0011940598487854004, "grad_norm": 0.038562119007110596, "global_step": 8291, "epoch": 230, "lr": 0.001} +{"train_loss": 0.029896628111600876, "train_loss_bc": 0.029893845319747925, "train_loss_llm": 0.00027826428413391113, "grad_norm": 0.04290526360273361, "global_step": 8292, "epoch": 230, "lr": 0.001} +{"train_loss": 0.0431925505399704, "train_loss_bc": 0.04315390810370445, "train_loss_llm": 0.003864198923110962, "grad_norm": 0.0179967749863863, "global_step": 8293, "epoch": 230, "lr": 0.001} +{"train_loss": 0.04147884622216225, "train_loss_bc": 0.04147425666451454, "train_loss_llm": 0.00045880675315856934, "grad_norm": 0.03390500321984291, "global_step": 8294, "epoch": 230, "lr": 0.001} +{"train_loss": 0.040561091154813766, "train_loss_bc": 0.040561091154813766, "train_loss_llm": 0.0, "grad_norm": 0.04364030063152313, "global_step": 8295, "epoch": 230, "lr": 0.001} +{"train_loss": 0.037553999572992325, "train_loss_bc": 0.037540603429079056, "train_loss_llm": 0.0013397932052612305, "grad_norm": 0.05519745498895645, "global_step": 8296, "epoch": 230, "lr": 0.001} +{"train_loss": 0.043001651763916016, "train_loss_bc": 0.042994383722543716, "train_loss_llm": 0.0007266998291015625, "grad_norm": 0.026171723380684853, "global_step": 8297, "epoch": 230, "lr": 0.001} +{"train_loss": 0.04894325137138367, "train_loss_bc": 0.04894325137138367, "train_loss_llm": 0.0, "grad_norm": 0.040618300437927246, "global_step": 8298, "epoch": 230, "lr": 0.001} +{"train_loss": 0.039847519248723984, "train_loss_bc": 0.0398324653506279, "train_loss_llm": 0.0015054941177368164, "grad_norm": 0.06025933474302292, "global_step": 8299, "epoch": 230, "lr": 0.001} +{"train_loss": 0.04273693636059761, "train_loss_bc": 0.042735543102025986, "train_loss_llm": 0.00013914704322814941, "grad_norm": 0.07052884995937347, "global_step": 8300, "epoch": 230, "lr": 0.001} +{"train_loss": 0.036837562918663025, "train_loss_bc": 0.036837562918663025, "train_loss_llm": 0.0, "grad_norm": 0.019600726664066315, "global_step": 8301, "epoch": 230, "lr": 0.001} +{"train_loss": 0.04343544691801071, "train_loss_bc": 0.04343544691801071, "train_loss_llm": 0.0, "grad_norm": 0.03262715041637421, "global_step": 8302, "epoch": 230, "lr": 0.001} +{"train_loss": 0.035443197935819626, "train_loss_bc": 0.035435039550065994, "train_loss_llm": 0.000815659761428833, "grad_norm": 0.04473424702882767, "global_step": 8303, "epoch": 230, "lr": 0.001} +{"train_loss": 0.039755381643772125, "train_loss_bc": 0.03975536674261093, "train_loss_llm": 1.4901161193847656e-06, "grad_norm": 0.06143173202872276, "global_step": 8304, "epoch": 230, "lr": 0.001} +{"train_loss": 0.03819501772522926, "train_loss_bc": 0.03819501772522926, "train_loss_llm": 0.0, "grad_norm": 0.015217945910990238, "global_step": 8305, "epoch": 230, "lr": 0.001} +{"train_loss": 0.040171679109334946, "train_loss_bc": 0.040171679109334946, "train_loss_llm": 0.0, "grad_norm": 0.0224357470870018, "global_step": 8306, "epoch": 230, "lr": 0.001} +{"train_loss": 0.03534438833594322, "train_loss_bc": 0.03534211590886116, "train_loss_llm": 0.00022724270820617676, "grad_norm": 0.029666971415281296, "global_step": 8307, "epoch": 230, "lr": 0.001} +{"train_loss": 0.03717901557683945, "train_loss_bc": 0.03717901557683945, "train_loss_llm": 0.0, "grad_norm": 0.03628561645746231, "global_step": 8308, "epoch": 230, "lr": 0.001} +{"train_loss": 0.04375390708446503, "train_loss_bc": 0.04375390708446503, "train_loss_llm": 0.0, "grad_norm": 0.018742026761174202, "global_step": 8309, "epoch": 230, "lr": 0.001} +{"train_loss": 0.031983841210603714, "train_loss_bc": 0.03198149427771568, "train_loss_llm": 0.00023472309112548828, "grad_norm": 0.027710625901818275, "global_step": 8310, "epoch": 230, "lr": 0.001} +{"train_loss": 0.040638938546180725, "train_loss_bc": 0.040638938546180725, "train_loss_llm": 0.0, "grad_norm": 0.03874817118048668, "global_step": 8311, "epoch": 230, "lr": 0.001} +{"train_loss": 0.045305222272872925, "train_loss_bc": 0.045305222272872925, "train_loss_llm": 0.0, "grad_norm": 0.04906568303704262, "global_step": 8312, "epoch": 230, "lr": 0.001} +{"train_loss": 0.04529831185936928, "train_loss_bc": 0.045286525040864944, "train_loss_llm": 0.0011785626411437988, "grad_norm": 0.021085664629936218, "global_step": 8313, "epoch": 230, "lr": 0.001} +{"train_loss": 0.04158975929021835, "train_loss_bc": 0.04158975929021835, "train_loss_llm": 0.0, "grad_norm": 0.03936510160565376, "global_step": 8314, "epoch": 230, "lr": 0.001} +{"train_loss": 0.03960215947073367, "train_loss_bc": 0.033802758902311325, "train_loss_llm": 0.0, "grad_norm": 0.047230206429958344, "global_step": 8315, "epoch": 230, "lr": 0.001, "train/cumulative_reward": 131.00341944515102, "train/mean_score": 21.84701374289448, "train/success_rate": 0.6, "test/cumulative_reward": 53.89232144397643, "test/mean_score": 10.158822369774963, "test/success_rate": 0.24, "val_loss": 0.05789553001523018, "train_action_mse_error": 0.04360238090157509} +{"train_loss": 0.04210822656750679, "train_loss_bc": 0.04210822656750679, "train_loss_llm": 0.0, "grad_norm": 0.062046002596616745, "global_step": 8316, "epoch": 231, "lr": 0.001} +{"train_loss": 0.036159761250019073, "train_loss_bc": 0.036159761250019073, "train_loss_llm": 0.0, "grad_norm": 0.01835767738521099, "global_step": 8317, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03168777748942375, "train_loss_bc": 0.03168777748942375, "train_loss_llm": 0.0, "grad_norm": 0.03129462152719498, "global_step": 8318, "epoch": 231, "lr": 0.001} +{"train_loss": 0.038651835173368454, "train_loss_bc": 0.038651835173368454, "train_loss_llm": 0.0, "grad_norm": 0.03900729492306709, "global_step": 8319, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04361811652779579, "train_loss_bc": 0.04360763356089592, "train_loss_llm": 0.0010481476783752441, "grad_norm": 0.05762872472405434, "global_step": 8320, "epoch": 231, "lr": 0.001} +{"train_loss": 0.036707039922475815, "train_loss_bc": 0.036707039922475815, "train_loss_llm": 0.0, "grad_norm": 0.026344606652855873, "global_step": 8321, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04028235003352165, "train_loss_bc": 0.04026596620678902, "train_loss_llm": 0.0016383230686187744, "grad_norm": 0.041334204375743866, "global_step": 8322, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04036327078938484, "train_loss_bc": 0.04036327078938484, "train_loss_llm": 0.0, "grad_norm": 0.058242067694664, "global_step": 8323, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04100484028458595, "train_loss_bc": 0.04100484028458595, "train_loss_llm": 0.0, "grad_norm": 0.06971006840467453, "global_step": 8324, "epoch": 231, "lr": 0.001} +{"train_loss": 0.029952911660075188, "train_loss_bc": 0.02994352951645851, "train_loss_llm": 0.0009382963180541992, "grad_norm": 0.01338705513626337, "global_step": 8325, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04176166281104088, "train_loss_bc": 0.04176166281104088, "train_loss_llm": 0.0, "grad_norm": 0.022993123158812523, "global_step": 8326, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04290549084544182, "train_loss_bc": 0.04290549084544182, "train_loss_llm": 0.0, "grad_norm": 0.0319276861846447, "global_step": 8327, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03946324810385704, "train_loss_bc": 0.03945353254675865, "train_loss_llm": 0.0009714365005493164, "grad_norm": 0.038266297429800034, "global_step": 8328, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04034816101193428, "train_loss_bc": 0.04034816101193428, "train_loss_llm": 0.0, "grad_norm": 0.015269633382558823, "global_step": 8329, "epoch": 231, "lr": 0.001} +{"train_loss": 0.05063994973897934, "train_loss_bc": 0.05063994973897934, "train_loss_llm": 0.0, "grad_norm": 0.024504654109477997, "global_step": 8330, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03682708740234375, "train_loss_bc": 0.03682275488972664, "train_loss_llm": 0.00043338537216186523, "grad_norm": 0.03277644142508507, "global_step": 8331, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03345384821295738, "train_loss_bc": 0.03345384821295738, "train_loss_llm": 0.0, "grad_norm": 0.040978532284498215, "global_step": 8332, "epoch": 231, "lr": 0.001} +{"train_loss": 0.042942073196172714, "train_loss_bc": 0.042942073196172714, "train_loss_llm": 0.0, "grad_norm": 0.023312347009778023, "global_step": 8333, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03640035539865494, "train_loss_bc": 0.03640035539865494, "train_loss_llm": 0.0, "grad_norm": 0.04202532395720482, "global_step": 8334, "epoch": 231, "lr": 0.001} +{"train_loss": 0.0434369295835495, "train_loss_bc": 0.04341966658830643, "train_loss_llm": 0.0017262697219848633, "grad_norm": 0.04742991179227829, "global_step": 8335, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03834548220038414, "train_loss_bc": 0.03834548220038414, "train_loss_llm": 0.0, "grad_norm": 0.06636390835046768, "global_step": 8336, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04223281517624855, "train_loss_bc": 0.042226843535900116, "train_loss_llm": 0.0005971193313598633, "grad_norm": 0.026338081806898117, "global_step": 8337, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03734714537858963, "train_loss_bc": 0.037346251308918, "train_loss_llm": 8.955597877502441e-05, "grad_norm": 0.03614576905965805, "global_step": 8338, "epoch": 231, "lr": 0.001} +{"train_loss": 0.040776923298835754, "train_loss_bc": 0.040762193500995636, "train_loss_llm": 0.001473158597946167, "grad_norm": 0.04791892692446709, "global_step": 8339, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04322793707251549, "train_loss_bc": 0.04322793707251549, "train_loss_llm": 0.0, "grad_norm": 0.05600649490952492, "global_step": 8340, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04466904327273369, "train_loss_bc": 0.04466904327273369, "train_loss_llm": 0.0, "grad_norm": 0.023698246106505394, "global_step": 8341, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03737939894199371, "train_loss_bc": 0.03737235441803932, "train_loss_llm": 0.0007044672966003418, "grad_norm": 0.042023785412311554, "global_step": 8342, "epoch": 231, "lr": 0.001} +{"train_loss": 0.039706043899059296, "train_loss_bc": 0.039706043899059296, "train_loss_llm": 0.0, "grad_norm": 0.047433484345674515, "global_step": 8343, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03565843775868416, "train_loss_bc": 0.03565232828259468, "train_loss_llm": 0.000611037015914917, "grad_norm": 0.05550107732415199, "global_step": 8344, "epoch": 231, "lr": 0.001} +{"train_loss": 0.04433942958712578, "train_loss_bc": 0.04432439059019089, "train_loss_llm": 0.0015038549900054932, "grad_norm": 0.019908467307686806, "global_step": 8345, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03198406100273132, "train_loss_bc": 0.03198406100273132, "train_loss_llm": 0.0, "grad_norm": 0.029714973643422127, "global_step": 8346, "epoch": 231, "lr": 0.001} +{"train_loss": 0.0452413335442543, "train_loss_bc": 0.04523221403360367, "train_loss_llm": 0.0009119212627410889, "grad_norm": 0.03871585428714752, "global_step": 8347, "epoch": 231, "lr": 0.001} +{"train_loss": 0.0388987697660923, "train_loss_bc": 0.0388987697660923, "train_loss_llm": 0.0, "grad_norm": 0.047573357820510864, "global_step": 8348, "epoch": 231, "lr": 0.001} +{"train_loss": 0.03979356586933136, "train_loss_bc": 0.03979356586933136, "train_loss_llm": 0.0, "grad_norm": 0.017928028479218483, "global_step": 8349, "epoch": 231, "lr": 0.001} +{"train_loss": 0.0409865565598011, "train_loss_bc": 0.0409865565598011, "train_loss_llm": 0.0, "grad_norm": 0.028370391577482224, "global_step": 8350, "epoch": 231, "lr": 0.001} +{"train_loss": 0.0396467338109182, "train_loss_bc": 0.03798053786158562, "train_loss_llm": 0.0, "grad_norm": 0.040588025003671646, "global_step": 8351, "epoch": 231, "lr": 0.001, "val_loss": 0.05901341885328293} +{"train_loss": 0.03893169015645981, "train_loss_bc": 0.03893169015645981, "train_loss_llm": 0.0, "grad_norm": 0.04434828832745552, "global_step": 8352, "epoch": 232, "lr": 0.001} +{"train_loss": 0.04512672498822212, "train_loss_bc": 0.04512672498822212, "train_loss_llm": 0.0, "grad_norm": 0.017676733434200287, "global_step": 8353, "epoch": 232, "lr": 0.001} +{"train_loss": 0.0364508181810379, "train_loss_bc": 0.0364508181810379, "train_loss_llm": 0.0, "grad_norm": 0.03120126575231552, "global_step": 8354, "epoch": 232, "lr": 0.001} +{"train_loss": 0.04213492572307587, "train_loss_bc": 0.0421263724565506, "train_loss_llm": 0.0008553564548492432, "grad_norm": 0.045438822358846664, "global_step": 8355, "epoch": 232, "lr": 0.001} +{"train_loss": 0.033862583339214325, "train_loss_bc": 0.033862583339214325, "train_loss_llm": 0.0, "grad_norm": 0.05914550647139549, "global_step": 8356, "epoch": 232, "lr": 0.001} +{"train_loss": 0.0437183640897274, "train_loss_bc": 0.0437183640897274, "train_loss_llm": 0.0, "grad_norm": 0.01822846569120884, "global_step": 8357, "epoch": 232, "lr": 0.001} +{"train_loss": 0.0412195585668087, "train_loss_bc": 0.041217394173145294, "train_loss_llm": 0.00021654367446899414, "grad_norm": 0.02752545103430748, "global_step": 8358, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03776826336979866, "train_loss_bc": 0.03776826336979866, "train_loss_llm": 0.0, "grad_norm": 0.04326291382312775, "global_step": 8359, "epoch": 232, "lr": 0.001} +{"train_loss": 0.04253723844885826, "train_loss_bc": 0.04253723844885826, "train_loss_llm": 0.0, "grad_norm": 0.04833139106631279, "global_step": 8360, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03917277231812477, "train_loss_bc": 0.03917277231812477, "train_loss_llm": 0.0, "grad_norm": 0.01914362609386444, "global_step": 8361, "epoch": 232, "lr": 0.001} +{"train_loss": 0.038540177047252655, "train_loss_bc": 0.038540177047252655, "train_loss_llm": 0.0, "grad_norm": 0.030650120228528976, "global_step": 8362, "epoch": 232, "lr": 0.001} +{"train_loss": 0.044241368770599365, "train_loss_bc": 0.044241368770599365, "train_loss_llm": 0.0, "grad_norm": 0.045830853283405304, "global_step": 8363, "epoch": 232, "lr": 0.001} +{"train_loss": 0.035893768072128296, "train_loss_bc": 0.035893768072128296, "train_loss_llm": 0.0, "grad_norm": 0.05536813288927078, "global_step": 8364, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03633667528629303, "train_loss_bc": 0.036331526935100555, "train_loss_llm": 0.0005146563053131104, "grad_norm": 0.01850665546953678, "global_step": 8365, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03932330012321472, "train_loss_bc": 0.03932330012321472, "train_loss_llm": 0.0, "grad_norm": 0.03385670855641365, "global_step": 8366, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03541606292128563, "train_loss_bc": 0.035393983125686646, "train_loss_llm": 0.002207815647125244, "grad_norm": 0.04389113932847977, "global_step": 8367, "epoch": 232, "lr": 0.001} +{"train_loss": 0.04087543487548828, "train_loss_bc": 0.04087543487548828, "train_loss_llm": 0.0, "grad_norm": 0.04766111448407173, "global_step": 8368, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03731062635779381, "train_loss_bc": 0.03731062635779381, "train_loss_llm": 0.0, "grad_norm": 0.01619507186114788, "global_step": 8369, "epoch": 232, "lr": 0.001} +{"train_loss": 0.04135475680232048, "train_loss_bc": 0.041322845965623856, "train_loss_llm": 0.003190934658050537, "grad_norm": 0.026131775230169296, "global_step": 8370, "epoch": 232, "lr": 0.001} +{"train_loss": 0.035692647099494934, "train_loss_bc": 0.035692647099494934, "train_loss_llm": 0.0, "grad_norm": 0.04020732641220093, "global_step": 8371, "epoch": 232, "lr": 0.001} +{"train_loss": 0.041496261954307556, "train_loss_bc": 0.041496261954307556, "train_loss_llm": 0.0, "grad_norm": 0.05224795266985893, "global_step": 8372, "epoch": 232, "lr": 0.001} +{"train_loss": 0.0458064004778862, "train_loss_bc": 0.04579204320907593, "train_loss_llm": 0.0014358460903167725, "grad_norm": 0.01946455053985119, "global_step": 8373, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03926899656653404, "train_loss_bc": 0.03926899656653404, "train_loss_llm": 0.0, "grad_norm": 0.030516626313328743, "global_step": 8374, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03683864325284958, "train_loss_bc": 0.03681883588433266, "train_loss_llm": 0.001980692148208618, "grad_norm": 0.040633492171764374, "global_step": 8375, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03116687387228012, "train_loss_bc": 0.03116687387228012, "train_loss_llm": 0.0, "grad_norm": 0.05043364688754082, "global_step": 8376, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03711185231804848, "train_loss_bc": 0.03708552569150925, "train_loss_llm": 0.0026325881481170654, "grad_norm": 0.020257212221622467, "global_step": 8377, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03791911527514458, "train_loss_bc": 0.03791532665491104, "train_loss_llm": 0.0003787577152252197, "grad_norm": 0.029545830562710762, "global_step": 8378, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03840155899524689, "train_loss_bc": 0.03840155899524689, "train_loss_llm": 0.0, "grad_norm": 0.039711594581604004, "global_step": 8379, "epoch": 232, "lr": 0.001} +{"train_loss": 0.04413105919957161, "train_loss_bc": 0.04413105919957161, "train_loss_llm": 0.0, "grad_norm": 0.05192155763506889, "global_step": 8380, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03374740481376648, "train_loss_bc": 0.03374740481376648, "train_loss_llm": 0.0, "grad_norm": 0.01902640610933304, "global_step": 8381, "epoch": 232, "lr": 0.001} +{"train_loss": 0.046266622841358185, "train_loss_bc": 0.046266622841358185, "train_loss_llm": 0.0, "grad_norm": 0.03755416348576546, "global_step": 8382, "epoch": 232, "lr": 0.001} +{"train_loss": 0.044293589890003204, "train_loss_bc": 0.044293589890003204, "train_loss_llm": 0.0, "grad_norm": 0.055615056306123734, "global_step": 8383, "epoch": 232, "lr": 0.001} +{"train_loss": 0.04494960606098175, "train_loss_bc": 0.04494960606098175, "train_loss_llm": 0.0, "grad_norm": 0.07088804990053177, "global_step": 8384, "epoch": 232, "lr": 0.001} +{"train_loss": 0.04202838987112045, "train_loss_bc": 0.04200073331594467, "train_loss_llm": 0.002765655517578125, "grad_norm": 0.022096268832683563, "global_step": 8385, "epoch": 232, "lr": 0.001} +{"train_loss": 0.040525197982788086, "train_loss_bc": 0.04052230715751648, "train_loss_llm": 0.00028908252716064453, "grad_norm": 0.04377412423491478, "global_step": 8386, "epoch": 232, "lr": 0.001} +{"train_loss": 0.03954422639475928, "train_loss_bc": 0.033732820302248, "train_loss_llm": 0.0, "grad_norm": 0.044949937611818314, "global_step": 8387, "epoch": 232, "lr": 0.001, "val_loss": 0.05807391181588173} +{"train_loss": 0.031260330229997635, "train_loss_bc": 0.03125438094139099, "train_loss_llm": 0.000594794750213623, "grad_norm": 0.0535803884267807, "global_step": 8388, "epoch": 233, "lr": 0.001} +{"train_loss": 0.03474985808134079, "train_loss_bc": 0.034747056663036346, "train_loss_llm": 0.00027999281883239746, "grad_norm": 0.018510257825255394, "global_step": 8389, "epoch": 233, "lr": 0.001} +{"train_loss": 0.03905387967824936, "train_loss_bc": 0.03905387967824936, "train_loss_llm": 0.0, "grad_norm": 0.030587751418352127, "global_step": 8390, "epoch": 233, "lr": 0.001} +{"train_loss": 0.0380687490105629, "train_loss_bc": 0.03806238994002342, "train_loss_llm": 0.0006359219551086426, "grad_norm": 0.043430451303720474, "global_step": 8391, "epoch": 233, "lr": 0.001} +{"train_loss": 0.04215226322412491, "train_loss_bc": 0.042142607271671295, "train_loss_llm": 0.0009656548500061035, "grad_norm": 0.0487024150788784, "global_step": 8392, "epoch": 233, "lr": 0.001} +{"train_loss": 0.039226289838552475, "train_loss_bc": 0.039226289838552475, "train_loss_llm": 0.0, "grad_norm": 0.019240187481045723, "global_step": 8393, "epoch": 233, "lr": 0.001} +{"train_loss": 0.043104756623506546, "train_loss_bc": 0.043104756623506546, "train_loss_llm": 0.0, "grad_norm": 0.027252282947301865, "global_step": 8394, "epoch": 233, "lr": 0.001} +{"train_loss": 0.03159923851490021, "train_loss_bc": 0.031595613807439804, "train_loss_llm": 0.0003623068332672119, "grad_norm": 0.03470727801322937, "global_step": 8395, "epoch": 233, "lr": 0.001} +{"train_loss": 0.039083633571863174, "train_loss_bc": 0.039083633571863174, "train_loss_llm": 0.0, "grad_norm": 0.03713676705956459, "global_step": 8396, "epoch": 233, "lr": 0.001} +{"train_loss": 0.031137552112340927, "train_loss_bc": 0.031137552112340927, "train_loss_llm": 0.0, "grad_norm": 0.019971206784248352, "global_step": 8397, "epoch": 233, "lr": 0.001} +{"train_loss": 0.04670627415180206, "train_loss_bc": 0.04670627415180206, "train_loss_llm": 0.0, "grad_norm": 0.0375029481947422, "global_step": 8398, "epoch": 233, "lr": 0.001} +{"train_loss": 0.04650752991437912, "train_loss_bc": 0.04649453982710838, "train_loss_llm": 0.0012988448143005371, "grad_norm": 0.047014616429805756, "global_step": 8399, "epoch": 233, "lr": 0.001} +{"train_loss": 0.03849438577890396, "train_loss_bc": 0.03848676756024361, "train_loss_llm": 0.0007616579532623291, "grad_norm": 0.05578159540891647, "global_step": 8400, "epoch": 233, "lr": 0.001} +{"train_loss": 0.034321367740631104, "train_loss_bc": 0.034321367740631104, "train_loss_llm": 0.0, "grad_norm": 0.021016022190451622, "global_step": 8401, "epoch": 233, "lr": 0.001} +{"train_loss": 0.0404958538711071, "train_loss_bc": 0.0404958538711071, "train_loss_llm": 0.0, "grad_norm": 0.037091612815856934, "global_step": 8402, "epoch": 233, "lr": 0.001} +{"train_loss": 0.0404994897544384, "train_loss_bc": 0.04049736261367798, "train_loss_llm": 0.00021287798881530762, "grad_norm": 0.051946163177490234, "global_step": 8403, "epoch": 233, "lr": 0.001} +{"train_loss": 0.044218510389328, "train_loss_bc": 0.044218510389328, "train_loss_llm": 0.0, "grad_norm": 0.06929309666156769, "global_step": 8404, "epoch": 233, "lr": 0.001} +{"train_loss": 0.04002890735864639, "train_loss_bc": 0.04002890735864639, "train_loss_llm": 0.0, "grad_norm": 0.0194109957665205, "global_step": 8405, "epoch": 233, "lr": 0.001} +{"train_loss": 0.045672863721847534, "train_loss_bc": 0.045672863721847534, "train_loss_llm": 0.0, "grad_norm": 0.033012934029102325, "global_step": 8406, "epoch": 233, "lr": 0.001} +{"train_loss": 0.03827366605401039, "train_loss_bc": 0.03827366605401039, "train_loss_llm": 0.0, "grad_norm": 0.04479570314288139, "global_step": 8407, "epoch": 233, "lr": 0.001} +{"train_loss": 0.04324844479560852, "train_loss_bc": 0.04324844479560852, "train_loss_llm": 0.0, "grad_norm": 0.056190114468336105, "global_step": 8408, "epoch": 233, "lr": 0.001} +{"train_loss": 0.04237963259220123, "train_loss_bc": 0.04237963259220123, "train_loss_llm": 0.0, "grad_norm": 0.01695455051958561, "global_step": 8409, "epoch": 233, "lr": 0.001} +{"train_loss": 0.041632894426584244, "train_loss_bc": 0.041632894426584244, "train_loss_llm": 0.0, "grad_norm": 0.029255056753754616, "global_step": 8410, "epoch": 233, "lr": 0.001} +{"train_loss": 0.038460247218608856, "train_loss_bc": 0.038460247218608856, "train_loss_llm": 0.0, "grad_norm": 0.038610950112342834, "global_step": 8411, "epoch": 233, "lr": 0.001} +{"train_loss": 0.03805074468255043, "train_loss_bc": 0.038041308522224426, "train_loss_llm": 0.0009436607360839844, "grad_norm": 0.04246817156672478, "global_step": 8412, "epoch": 233, "lr": 0.001} +{"train_loss": 0.04454147070646286, "train_loss_bc": 0.04454147070646286, "train_loss_llm": 0.0, "grad_norm": 0.023504581302404404, "global_step": 8413, "epoch": 233, "lr": 0.001} +{"train_loss": 0.03065214306116104, "train_loss_bc": 0.03065214306116104, "train_loss_llm": 0.0, "grad_norm": 0.027777766808867455, "global_step": 8414, "epoch": 233, "lr": 0.001} +{"train_loss": 0.032769858837127686, "train_loss_bc": 0.032769858837127686, "train_loss_llm": 0.0, "grad_norm": 0.03288991004228592, "global_step": 8415, "epoch": 233, "lr": 0.001} +{"train_loss": 0.04133438318967819, "train_loss_bc": 0.04133438318967819, "train_loss_llm": 0.0, "grad_norm": 0.04280320927500725, "global_step": 8416, "epoch": 233, "lr": 0.001} +{"train_loss": 0.04189566895365715, "train_loss_bc": 0.04189566895365715, "train_loss_llm": 0.0, "grad_norm": 0.02171969786286354, "global_step": 8417, "epoch": 233, "lr": 0.001} +{"train_loss": 0.042186833918094635, "train_loss_bc": 0.042186833918094635, "train_loss_llm": 0.0, "grad_norm": 0.03482668101787567, "global_step": 8418, "epoch": 233, "lr": 0.001} +{"train_loss": 0.03497283160686493, "train_loss_bc": 0.03497283160686493, "train_loss_llm": 0.0, "grad_norm": 0.0425427220761776, "global_step": 8419, "epoch": 233, "lr": 0.001} +{"train_loss": 0.03862227126955986, "train_loss_bc": 0.03862227126955986, "train_loss_llm": 0.0, "grad_norm": 0.04904956370592117, "global_step": 8420, "epoch": 233, "lr": 0.001} +{"train_loss": 0.04248468577861786, "train_loss_bc": 0.042463675141334534, "train_loss_llm": 0.0021008849143981934, "grad_norm": 0.022085973992943764, "global_step": 8421, "epoch": 233, "lr": 0.001} +{"train_loss": 0.03965618088841438, "train_loss_bc": 0.03965618088841438, "train_loss_llm": 0.0, "grad_norm": 0.036007799208164215, "global_step": 8422, "epoch": 233, "lr": 0.001} +{"train_loss": 0.039609293142954506, "train_loss_bc": 0.048390861600637436, "train_loss_llm": 0.0, "grad_norm": 0.05309362709522247, "global_step": 8423, "epoch": 233, "lr": 0.001, "val_loss": 0.05869554728269577} +{"train_loss": 0.045726705342531204, "train_loss_bc": 0.04571564495563507, "train_loss_llm": 0.0011060535907745361, "grad_norm": 0.06283669918775558, "global_step": 8424, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03542384132742882, "train_loss_bc": 0.03540773317217827, "train_loss_llm": 0.0016107559204101562, "grad_norm": 0.018645066767930984, "global_step": 8425, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04263510927557945, "train_loss_bc": 0.04261530935764313, "train_loss_llm": 0.0019798576831817627, "grad_norm": 0.03064681775867939, "global_step": 8426, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03740512579679489, "train_loss_bc": 0.03740512579679489, "train_loss_llm": 0.0, "grad_norm": 0.041303593665361404, "global_step": 8427, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04347178712487221, "train_loss_bc": 0.04347178712487221, "train_loss_llm": 0.0, "grad_norm": 0.053746096789836884, "global_step": 8428, "epoch": 234, "lr": 0.001} +{"train_loss": 0.030837906524538994, "train_loss_bc": 0.030837906524538994, "train_loss_llm": 0.0, "grad_norm": 0.01841653883457184, "global_step": 8429, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04292292892932892, "train_loss_bc": 0.04292292892932892, "train_loss_llm": 0.0, "grad_norm": 0.039231885224580765, "global_step": 8430, "epoch": 234, "lr": 0.001} +{"train_loss": 0.05242178589105606, "train_loss_bc": 0.05242178589105606, "train_loss_llm": 0.0, "grad_norm": 0.05927840992808342, "global_step": 8431, "epoch": 234, "lr": 0.001} +{"train_loss": 0.046216510236263275, "train_loss_bc": 0.046216510236263275, "train_loss_llm": 0.0, "grad_norm": 0.06905079632997513, "global_step": 8432, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03273788094520569, "train_loss_bc": 0.03273788094520569, "train_loss_llm": 0.0, "grad_norm": 0.018431968986988068, "global_step": 8433, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04044230654835701, "train_loss_bc": 0.040419768542051315, "train_loss_llm": 0.002253979444503784, "grad_norm": 0.02880910597741604, "global_step": 8434, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04162972420454025, "train_loss_bc": 0.04162972420454025, "train_loss_llm": 0.0, "grad_norm": 0.04311497509479523, "global_step": 8435, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03726305812597275, "train_loss_bc": 0.03726305812597275, "train_loss_llm": 0.0, "grad_norm": 0.04842006042599678, "global_step": 8436, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03249156102538109, "train_loss_bc": 0.03249156102538109, "train_loss_llm": 0.0, "grad_norm": 0.017728570848703384, "global_step": 8437, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04235893487930298, "train_loss_bc": 0.04235893487930298, "train_loss_llm": 0.0, "grad_norm": 0.032571855932474136, "global_step": 8438, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03426223620772362, "train_loss_bc": 0.03426223620772362, "train_loss_llm": 0.0, "grad_norm": 0.0409802682697773, "global_step": 8439, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04401201382279396, "train_loss_bc": 0.04400742053985596, "train_loss_llm": 0.0004591941833496094, "grad_norm": 0.05213076248764992, "global_step": 8440, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04576096683740616, "train_loss_bc": 0.04576096683740616, "train_loss_llm": 0.0, "grad_norm": 0.018134305253624916, "global_step": 8441, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03394210338592529, "train_loss_bc": 0.03393368422985077, "train_loss_llm": 0.0008418262004852295, "grad_norm": 0.03038051724433899, "global_step": 8442, "epoch": 234, "lr": 0.001} +{"train_loss": 0.033264778554439545, "train_loss_bc": 0.033246129751205444, "train_loss_llm": 0.0018649697303771973, "grad_norm": 0.03826716169714928, "global_step": 8443, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03641049936413765, "train_loss_bc": 0.03639565780758858, "train_loss_llm": 0.0014841258525848389, "grad_norm": 0.04994187504053116, "global_step": 8444, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04099708050489426, "train_loss_bc": 0.04099708050489426, "train_loss_llm": 0.0, "grad_norm": 0.021811235696077347, "global_step": 8445, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04012307897210121, "train_loss_bc": 0.04009552672505379, "train_loss_llm": 0.0027552545070648193, "grad_norm": 0.028868645429611206, "global_step": 8446, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03815405070781708, "train_loss_bc": 0.03815405070781708, "train_loss_llm": 0.0, "grad_norm": 0.03937240317463875, "global_step": 8447, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04405180364847183, "train_loss_bc": 0.04405180364847183, "train_loss_llm": 0.0, "grad_norm": 0.05801457166671753, "global_step": 8448, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03993421047925949, "train_loss_bc": 0.03993421047925949, "train_loss_llm": 0.0, "grad_norm": 0.020770283415913582, "global_step": 8449, "epoch": 234, "lr": 0.001} +{"train_loss": 0.040660709142684937, "train_loss_bc": 0.040660709142684937, "train_loss_llm": 0.0, "grad_norm": 0.035084109753370285, "global_step": 8450, "epoch": 234, "lr": 0.001} +{"train_loss": 0.030624959617853165, "train_loss_bc": 0.030624959617853165, "train_loss_llm": 0.0, "grad_norm": 0.04432431235909462, "global_step": 8451, "epoch": 234, "lr": 0.001} +{"train_loss": 0.0398385152220726, "train_loss_bc": 0.03983001038432121, "train_loss_llm": 0.0008504390716552734, "grad_norm": 0.05454724654555321, "global_step": 8452, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03959924355149269, "train_loss_bc": 0.03959168866276741, "train_loss_llm": 0.0007555782794952393, "grad_norm": 0.0167720764875412, "global_step": 8453, "epoch": 234, "lr": 0.001} +{"train_loss": 0.04574533924460411, "train_loss_bc": 0.04574533924460411, "train_loss_llm": 0.0, "grad_norm": 0.026166612282395363, "global_step": 8454, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03796403482556343, "train_loss_bc": 0.03796403482556343, "train_loss_llm": 0.0, "grad_norm": 0.032647162675857544, "global_step": 8455, "epoch": 234, "lr": 0.001} +{"train_loss": 0.0414653904736042, "train_loss_bc": 0.0414653904736042, "train_loss_llm": 0.0, "grad_norm": 0.04244813695549965, "global_step": 8456, "epoch": 234, "lr": 0.001} +{"train_loss": 0.039344605058431625, "train_loss_bc": 0.03932883217930794, "train_loss_llm": 0.0015771687030792236, "grad_norm": 0.020646875724196434, "global_step": 8457, "epoch": 234, "lr": 0.001} +{"train_loss": 0.03701294958591461, "train_loss_bc": 0.03701294958591461, "train_loss_llm": 0.0, "grad_norm": 0.027002600952982903, "global_step": 8458, "epoch": 234, "lr": 0.001} +{"train_loss": 0.039485964458435774, "train_loss_bc": 0.034340985119342804, "train_loss_llm": 0.0, "grad_norm": 0.03899392858147621, "global_step": 8459, "epoch": 234, "lr": 0.001, "val_loss": 0.05821651592850685} +{"train_loss": 0.037996888160705566, "train_loss_bc": 0.037976495921611786, "train_loss_llm": 0.0020392239093780518, "grad_norm": 0.05115976184606552, "global_step": 8460, "epoch": 235, "lr": 0.001} +{"train_loss": 0.038986172527074814, "train_loss_bc": 0.038951750844717026, "train_loss_llm": 0.003442138433456421, "grad_norm": 0.016751110553741455, "global_step": 8461, "epoch": 235, "lr": 0.001} +{"train_loss": 0.04029250517487526, "train_loss_bc": 0.04029250517487526, "train_loss_llm": 0.0, "grad_norm": 0.024126263335347176, "global_step": 8462, "epoch": 235, "lr": 0.001} +{"train_loss": 0.0444759801030159, "train_loss_bc": 0.0444759801030159, "train_loss_llm": 0.0, "grad_norm": 0.029909605160355568, "global_step": 8463, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03642997518181801, "train_loss_bc": 0.03642997518181801, "train_loss_llm": 0.0, "grad_norm": 0.03931469842791557, "global_step": 8464, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03567414730787277, "train_loss_bc": 0.03567228466272354, "train_loss_llm": 0.00018644332885742188, "grad_norm": 0.01607239432632923, "global_step": 8465, "epoch": 235, "lr": 0.001} +{"train_loss": 0.0483240932226181, "train_loss_bc": 0.0483240932226181, "train_loss_llm": 0.0, "grad_norm": 0.02582145296037197, "global_step": 8466, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03610556945204735, "train_loss_bc": 0.03610556945204735, "train_loss_llm": 0.0, "grad_norm": 0.03559373319149017, "global_step": 8467, "epoch": 235, "lr": 0.001} +{"train_loss": 0.031272005289793015, "train_loss_bc": 0.031272005289793015, "train_loss_llm": 0.0, "grad_norm": 0.04318137466907501, "global_step": 8468, "epoch": 235, "lr": 0.001} +{"train_loss": 0.04135693237185478, "train_loss_bc": 0.04132365435361862, "train_loss_llm": 0.003327667713165283, "grad_norm": 0.0263249222189188, "global_step": 8469, "epoch": 235, "lr": 0.001} +{"train_loss": 0.04276289418339729, "train_loss_bc": 0.04276289418339729, "train_loss_llm": 0.0, "grad_norm": 0.03763516619801521, "global_step": 8470, "epoch": 235, "lr": 0.001} +{"train_loss": 0.040732018649578094, "train_loss_bc": 0.040732018649578094, "train_loss_llm": 0.0, "grad_norm": 0.047756679356098175, "global_step": 8471, "epoch": 235, "lr": 0.001} +{"train_loss": 0.04306017607450485, "train_loss_bc": 0.04304716736078262, "train_loss_llm": 0.001300901174545288, "grad_norm": 0.05965407192707062, "global_step": 8472, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03739956021308899, "train_loss_bc": 0.03739916533231735, "train_loss_llm": 3.9517879486083984e-05, "grad_norm": 0.01959368959069252, "global_step": 8473, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03213799372315407, "train_loss_bc": 0.03212982416152954, "train_loss_llm": 0.0008169412612915039, "grad_norm": 0.024055371060967445, "global_step": 8474, "epoch": 235, "lr": 0.001} +{"train_loss": 0.04569452628493309, "train_loss_bc": 0.04569452628493309, "train_loss_llm": 0.0, "grad_norm": 0.03878380358219147, "global_step": 8475, "epoch": 235, "lr": 0.001} +{"train_loss": 0.04109310358762741, "train_loss_bc": 0.04109310358762741, "train_loss_llm": 0.0, "grad_norm": 0.05288378521800041, "global_step": 8476, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03200903907418251, "train_loss_bc": 0.03200903907418251, "train_loss_llm": 0.0, "grad_norm": 0.017611362040042877, "global_step": 8477, "epoch": 235, "lr": 0.001} +{"train_loss": 0.044478483498096466, "train_loss_bc": 0.044477496296167374, "train_loss_llm": 9.861588478088379e-05, "grad_norm": 0.026399243623018265, "global_step": 8478, "epoch": 235, "lr": 0.001} +{"train_loss": 0.038332466036081314, "train_loss_bc": 0.038332466036081314, "train_loss_llm": 0.0, "grad_norm": 0.031158391386270523, "global_step": 8479, "epoch": 235, "lr": 0.001} +{"train_loss": 0.043044883757829666, "train_loss_bc": 0.043044883757829666, "train_loss_llm": 0.0, "grad_norm": 0.04265293851494789, "global_step": 8480, "epoch": 235, "lr": 0.001} +{"train_loss": 0.0369611494243145, "train_loss_bc": 0.0369611494243145, "train_loss_llm": 0.0, "grad_norm": 0.02136426977813244, "global_step": 8481, "epoch": 235, "lr": 0.001} +{"train_loss": 0.039006464183330536, "train_loss_bc": 0.039006464183330536, "train_loss_llm": 0.0, "grad_norm": 0.029552534222602844, "global_step": 8482, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03848958760499954, "train_loss_bc": 0.03848640248179436, "train_loss_llm": 0.0003184974193572998, "grad_norm": 0.03653121367096901, "global_step": 8483, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03950754925608635, "train_loss_bc": 0.039501871913671494, "train_loss_llm": 0.000567704439163208, "grad_norm": 0.04391758516430855, "global_step": 8484, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03587762266397476, "train_loss_bc": 0.035868316888809204, "train_loss_llm": 0.0009304583072662354, "grad_norm": 0.018619703128933907, "global_step": 8485, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03379964828491211, "train_loss_bc": 0.03377748653292656, "train_loss_llm": 0.0022162199020385742, "grad_norm": 0.028965914621949196, "global_step": 8486, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03456006571650505, "train_loss_bc": 0.03453287482261658, "train_loss_llm": 0.0027191638946533203, "grad_norm": 0.03852177783846855, "global_step": 8487, "epoch": 235, "lr": 0.001} +{"train_loss": 0.037970393896102905, "train_loss_bc": 0.03795887902379036, "train_loss_llm": 0.0011513829231262207, "grad_norm": 0.0478929802775383, "global_step": 8488, "epoch": 235, "lr": 0.001} +{"train_loss": 0.041822630912065506, "train_loss_bc": 0.041816696524620056, "train_loss_llm": 0.0005934536457061768, "grad_norm": 0.018410872668027878, "global_step": 8489, "epoch": 235, "lr": 0.001} +{"train_loss": 0.0506698414683342, "train_loss_bc": 0.0506698414683342, "train_loss_llm": 0.0, "grad_norm": 0.02876889333128929, "global_step": 8490, "epoch": 235, "lr": 0.001} +{"train_loss": 0.046086203306913376, "train_loss_bc": 0.046086203306913376, "train_loss_llm": 0.0, "grad_norm": 0.03792698681354523, "global_step": 8491, "epoch": 235, "lr": 0.001} +{"train_loss": 0.039048515260219574, "train_loss_bc": 0.039048515260219574, "train_loss_llm": 0.0, "grad_norm": 0.04670821130275726, "global_step": 8492, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03827565535902977, "train_loss_bc": 0.03827565535902977, "train_loss_llm": 0.0, "grad_norm": 0.021138926967978477, "global_step": 8493, "epoch": 235, "lr": 0.001} +{"train_loss": 0.03750806674361229, "train_loss_bc": 0.03750806674361229, "train_loss_llm": 0.0, "grad_norm": 0.03495151177048683, "global_step": 8494, "epoch": 235, "lr": 0.001} +{"train_loss": 0.0393958216946986, "train_loss_bc": 0.03700677305459976, "train_loss_llm": 0.0, "grad_norm": 0.04984011873602867, "global_step": 8495, "epoch": 235, "lr": 0.001, "train/cumulative_reward": 50.846326133477504, "train/mean_score": 10.609623778724544, "train/success_rate": 0.3, "test/cumulative_reward": 42.259867670310285, "test/mean_score": 4.1600175363807885, "test/success_rate": 0.06, "val_loss": 0.05910596251487732, "train_action_mse_error": 0.04325147345662117} +{"train_loss": 0.03728847578167915, "train_loss_bc": 0.03728456050157547, "train_loss_llm": 0.0003916025161743164, "grad_norm": 0.06072640419006348, "global_step": 8496, "epoch": 236, "lr": 0.001} +{"train_loss": 0.04320476949214935, "train_loss_bc": 0.04320476949214935, "train_loss_llm": 0.0, "grad_norm": 0.018358929082751274, "global_step": 8497, "epoch": 236, "lr": 0.001} +{"train_loss": 0.04454467073082924, "train_loss_bc": 0.044534701853990555, "train_loss_llm": 0.0009969770908355713, "grad_norm": 0.030171237885951996, "global_step": 8498, "epoch": 236, "lr": 0.001} +{"train_loss": 0.047610633075237274, "train_loss_bc": 0.04760512709617615, "train_loss_llm": 0.0005507469177246094, "grad_norm": 0.03781357780098915, "global_step": 8499, "epoch": 236, "lr": 0.001} +{"train_loss": 0.041858214884996414, "train_loss_bc": 0.041858214884996414, "train_loss_llm": 0.0, "grad_norm": 0.04888596385717392, "global_step": 8500, "epoch": 236, "lr": 0.001} +{"train_loss": 0.0407019779086113, "train_loss_bc": 0.0407019779086113, "train_loss_llm": 0.0, "grad_norm": 0.01703529804944992, "global_step": 8501, "epoch": 236, "lr": 0.001} +{"train_loss": 0.04008035361766815, "train_loss_bc": 0.04008035361766815, "train_loss_llm": 0.0, "grad_norm": 0.02872232161462307, "global_step": 8502, "epoch": 236, "lr": 0.001} +{"train_loss": 0.046796832233667374, "train_loss_bc": 0.046796832233667374, "train_loss_llm": 0.0, "grad_norm": 0.031845662742853165, "global_step": 8503, "epoch": 236, "lr": 0.001} +{"train_loss": 0.034104179590940475, "train_loss_bc": 0.034104179590940475, "train_loss_llm": 0.0, "grad_norm": 0.041169174015522, "global_step": 8504, "epoch": 236, "lr": 0.001} +{"train_loss": 0.03676043823361397, "train_loss_bc": 0.03674895688891411, "train_loss_llm": 0.0011480152606964111, "grad_norm": 0.019163763150572777, "global_step": 8505, "epoch": 236, "lr": 0.001} +{"train_loss": 0.03727804124355316, "train_loss_bc": 0.03727804124355316, "train_loss_llm": 0.0, "grad_norm": 0.028913216665387154, "global_step": 8506, "epoch": 236, "lr": 0.001} +{"train_loss": 0.04064393788576126, "train_loss_bc": 0.04064393788576126, "train_loss_llm": 0.0, "grad_norm": 0.03260425478219986, "global_step": 8507, "epoch": 236, "lr": 0.001} +{"train_loss": 0.04368370771408081, "train_loss_bc": 0.04368370771408081, "train_loss_llm": 0.0, "grad_norm": 0.03670460358262062, "global_step": 8508, "epoch": 236, "lr": 0.001} +{"train_loss": 0.03605816140770912, "train_loss_bc": 0.03602832183241844, "train_loss_llm": 0.002984076738357544, "grad_norm": 0.014605528675019741, "global_step": 8509, "epoch": 236, "lr": 0.001} +{"train_loss": 0.03784559667110443, "train_loss_bc": 0.03784559667110443, "train_loss_llm": 0.0, "grad_norm": 0.023181350901722908, "global_step": 8510, "epoch": 236, "lr": 0.001} +{"train_loss": 0.040912073105573654, "train_loss_bc": 0.040912073105573654, "train_loss_llm": 0.0, "grad_norm": 0.031474076211452484, "global_step": 8511, "epoch": 236, "lr": 0.001} +{"train_loss": 0.04355202615261078, "train_loss_bc": 0.04354500025510788, "train_loss_llm": 0.0007025301456451416, "grad_norm": 0.041161246597766876, "global_step": 8512, "epoch": 236, "lr": 0.001} +{"train_loss": 0.0326554961502552, "train_loss_bc": 0.0326554961502552, "train_loss_llm": 0.0, "grad_norm": 0.017611142247915268, "global_step": 8513, "epoch": 236, "lr": 0.001} +{"train_loss": 0.039598025381565094, "train_loss_bc": 0.039598025381565094, "train_loss_llm": 0.0, "grad_norm": 0.02970157563686371, "global_step": 8514, "epoch": 236, "lr": 0.001} +{"train_loss": 0.0372193269431591, "train_loss_bc": 0.0372193269431591, "train_loss_llm": 0.0, "grad_norm": 0.036622948944568634, "global_step": 8515, "epoch": 236, "lr": 0.001} +{"train_loss": 0.0368245504796505, "train_loss_bc": 0.0368245504796505, "train_loss_llm": 0.0, "grad_norm": 0.04122282937169075, "global_step": 8516, "epoch": 236, "lr": 0.001} +{"train_loss": 0.03196835145354271, "train_loss_bc": 0.03194282576441765, "train_loss_llm": 0.0025523900985717773, "grad_norm": 0.016591668128967285, "global_step": 8517, "epoch": 236, "lr": 0.001} +{"train_loss": 0.039609991014003754, "train_loss_bc": 0.039609991014003754, "train_loss_llm": 0.0, "grad_norm": 0.027020061388611794, "global_step": 8518, "epoch": 236, "lr": 0.001} +{"train_loss": 0.039313048124313354, "train_loss_bc": 0.039313048124313354, "train_loss_llm": 0.0, "grad_norm": 0.03345433250069618, "global_step": 8519, "epoch": 236, "lr": 0.001} +{"train_loss": 0.0343078151345253, "train_loss_bc": 0.0343078151345253, "train_loss_llm": 0.0, "grad_norm": 0.04415975883603096, "global_step": 8520, "epoch": 236, "lr": 0.001} +{"train_loss": 0.03959963843226433, "train_loss_bc": 0.03959663212299347, "train_loss_llm": 0.0003007054328918457, "grad_norm": 0.0224442221224308, "global_step": 8521, "epoch": 236, "lr": 0.001} +{"train_loss": 0.04502503201365471, "train_loss_bc": 0.04502503201365471, "train_loss_llm": 0.0, "grad_norm": 0.039433833211660385, "global_step": 8522, "epoch": 236, "lr": 0.001} +{"train_loss": 0.0397069975733757, "train_loss_bc": 0.0397069975733757, "train_loss_llm": 0.0, "grad_norm": 0.049342408776283264, "global_step": 8523, "epoch": 236, "lr": 0.001} +{"train_loss": 0.039720114320516586, "train_loss_bc": 0.039720114320516586, "train_loss_llm": 0.0, "grad_norm": 0.05313117429614067, "global_step": 8524, "epoch": 236, "lr": 0.001} +{"train_loss": 0.03883751481771469, "train_loss_bc": 0.03883751481771469, "train_loss_llm": 0.0, "grad_norm": 0.016191985458135605, "global_step": 8525, "epoch": 236, "lr": 0.001} +{"train_loss": 0.040466759353876114, "train_loss_bc": 0.04046067222952843, "train_loss_llm": 0.0006086528301239014, "grad_norm": 0.029157791286706924, "global_step": 8526, "epoch": 236, "lr": 0.001} +{"train_loss": 0.03892794996500015, "train_loss_bc": 0.03892112150788307, "train_loss_llm": 0.0006828606128692627, "grad_norm": 0.03553258627653122, "global_step": 8527, "epoch": 236, "lr": 0.001} +{"train_loss": 0.0391496866941452, "train_loss_bc": 0.0391496866941452, "train_loss_llm": 0.0, "grad_norm": 0.038484714925289154, "global_step": 8528, "epoch": 236, "lr": 0.001} +{"train_loss": 0.03310120105743408, "train_loss_bc": 0.03309990093111992, "train_loss_llm": 0.00013017654418945312, "grad_norm": 0.01382533274590969, "global_step": 8529, "epoch": 236, "lr": 0.001} +{"train_loss": 0.03380165249109268, "train_loss_bc": 0.033774614334106445, "train_loss_llm": 0.0027039945125579834, "grad_norm": 0.02325563319027424, "global_step": 8530, "epoch": 236, "lr": 0.001} +{"train_loss": 0.039296389660901494, "train_loss_bc": 0.04190640151500702, "train_loss_llm": 0.0006386637687683105, "grad_norm": 0.040371011942625046, "global_step": 8531, "epoch": 236, "lr": 0.001, "val_loss": 0.05851246416568756} +{"train_loss": 0.04223453253507614, "train_loss_bc": 0.04223126918077469, "train_loss_llm": 0.00032642483711242676, "grad_norm": 0.05048903822898865, "global_step": 8532, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04209514707326889, "train_loss_bc": 0.04209514707326889, "train_loss_llm": 0.0, "grad_norm": 0.021643972024321556, "global_step": 8533, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03957675024867058, "train_loss_bc": 0.03957675024867058, "train_loss_llm": 0.0, "grad_norm": 0.030574368312954903, "global_step": 8534, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04059350863099098, "train_loss_bc": 0.04059350863099098, "train_loss_llm": 0.0, "grad_norm": 0.03562292829155922, "global_step": 8535, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04072044417262077, "train_loss_bc": 0.040716834366321564, "train_loss_llm": 0.000361025333404541, "grad_norm": 0.04072235897183418, "global_step": 8536, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04485544189810753, "train_loss_bc": 0.04483269527554512, "train_loss_llm": 0.0022745132446289062, "grad_norm": 0.017921799793839455, "global_step": 8537, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04270034655928612, "train_loss_bc": 0.042680758982896805, "train_loss_llm": 0.0019587576389312744, "grad_norm": 0.02691585384309292, "global_step": 8538, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04624215140938759, "train_loss_bc": 0.04624215140938759, "train_loss_llm": 0.0, "grad_norm": 0.035781823098659515, "global_step": 8539, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03370670974254608, "train_loss_bc": 0.033701032400131226, "train_loss_llm": 0.0005678534507751465, "grad_norm": 0.04285888746380806, "global_step": 8540, "epoch": 237, "lr": 0.001} +{"train_loss": 0.031591132283210754, "train_loss_bc": 0.03158308565616608, "train_loss_llm": 0.0008047223091125488, "grad_norm": 0.019030451774597168, "global_step": 8541, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03051738627254963, "train_loss_bc": 0.030491096898913383, "train_loss_llm": 0.0026289522647857666, "grad_norm": 0.03331390395760536, "global_step": 8542, "epoch": 237, "lr": 0.001} +{"train_loss": 0.033993788063526154, "train_loss_bc": 0.033993788063526154, "train_loss_llm": 0.0, "grad_norm": 0.041760072112083435, "global_step": 8543, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03564165160059929, "train_loss_bc": 0.03564165160059929, "train_loss_llm": 0.0, "grad_norm": 0.04892771691083908, "global_step": 8544, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04265560954809189, "train_loss_bc": 0.042651090770959854, "train_loss_llm": 0.0004520416259765625, "grad_norm": 0.019344091415405273, "global_step": 8545, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04014337435364723, "train_loss_bc": 0.04014337435364723, "train_loss_llm": 0.0, "grad_norm": 0.02839300036430359, "global_step": 8546, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04113905131816864, "train_loss_bc": 0.041123244911432266, "train_loss_llm": 0.0015808045864105225, "grad_norm": 0.04289049655199051, "global_step": 8547, "epoch": 237, "lr": 0.001} +{"train_loss": 0.039956849068403244, "train_loss_bc": 0.039956849068403244, "train_loss_llm": 0.0, "grad_norm": 0.05552966147661209, "global_step": 8548, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03248629719018936, "train_loss_bc": 0.03248629719018936, "train_loss_llm": 0.0, "grad_norm": 0.015932362526655197, "global_step": 8549, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03267020732164383, "train_loss_bc": 0.03267020732164383, "train_loss_llm": 0.0, "grad_norm": 0.024497030302882195, "global_step": 8550, "epoch": 237, "lr": 0.001} +{"train_loss": 0.0400497131049633, "train_loss_bc": 0.0400497131049633, "train_loss_llm": 0.0, "grad_norm": 0.033785659819841385, "global_step": 8551, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03651484102010727, "train_loss_bc": 0.03651484102010727, "train_loss_llm": 0.0, "grad_norm": 0.04631159082055092, "global_step": 8552, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04171965643763542, "train_loss_bc": 0.041706256568431854, "train_loss_llm": 0.0013401210308074951, "grad_norm": 0.01910092867910862, "global_step": 8553, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03967118635773659, "train_loss_bc": 0.03967118635773659, "train_loss_llm": 0.0, "grad_norm": 0.025852810591459274, "global_step": 8554, "epoch": 237, "lr": 0.001} +{"train_loss": 0.040349386632442474, "train_loss_bc": 0.040349386632442474, "train_loss_llm": 0.0, "grad_norm": 0.029632173478603363, "global_step": 8555, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03672110289335251, "train_loss_bc": 0.036718305200338364, "train_loss_llm": 0.00027993321418762207, "grad_norm": 0.034926339983940125, "global_step": 8556, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03975991904735565, "train_loss_bc": 0.03975991904735565, "train_loss_llm": 0.0, "grad_norm": 0.02441316284239292, "global_step": 8557, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04263307526707649, "train_loss_bc": 0.04263307526707649, "train_loss_llm": 0.0, "grad_norm": 0.03183240815997124, "global_step": 8558, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04277917742729187, "train_loss_bc": 0.04277917742729187, "train_loss_llm": 0.0, "grad_norm": 0.038585323840379715, "global_step": 8559, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04261338338255882, "train_loss_bc": 0.042606472969055176, "train_loss_llm": 0.0006911754608154297, "grad_norm": 0.04574637487530708, "global_step": 8560, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03647835552692413, "train_loss_bc": 0.03647835552692413, "train_loss_llm": 0.0, "grad_norm": 0.018485255539417267, "global_step": 8561, "epoch": 237, "lr": 0.001} +{"train_loss": 0.041809577494859695, "train_loss_bc": 0.041809577494859695, "train_loss_llm": 0.0, "grad_norm": 0.03149540349841118, "global_step": 8562, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04160452261567116, "train_loss_bc": 0.04160452261567116, "train_loss_llm": 0.0, "grad_norm": 0.038780439645051956, "global_step": 8563, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04153335839509964, "train_loss_bc": 0.04153335839509964, "train_loss_llm": 0.0, "grad_norm": 0.048921361565589905, "global_step": 8564, "epoch": 237, "lr": 0.001} +{"train_loss": 0.037135880440473557, "train_loss_bc": 0.037135880440473557, "train_loss_llm": 0.0, "grad_norm": 0.018344957381486893, "global_step": 8565, "epoch": 237, "lr": 0.001} +{"train_loss": 0.04361707717180252, "train_loss_bc": 0.04361707717180252, "train_loss_llm": 0.0, "grad_norm": 0.024559352546930313, "global_step": 8566, "epoch": 237, "lr": 0.001} +{"train_loss": 0.03921073235364424, "train_loss_bc": 0.03307577222585678, "train_loss_llm": 0.0, "grad_norm": 0.035400498658418655, "global_step": 8567, "epoch": 237, "lr": 0.001, "val_loss": 0.05823799967765808} +{"train_loss": 0.03017377108335495, "train_loss_bc": 0.03017377108335495, "train_loss_llm": 0.0, "grad_norm": 0.03933810442686081, "global_step": 8568, "epoch": 238, "lr": 0.001} +{"train_loss": 0.03840605169534683, "train_loss_bc": 0.03840605169534683, "train_loss_llm": 0.0, "grad_norm": 0.02105804719030857, "global_step": 8569, "epoch": 238, "lr": 0.001} +{"train_loss": 0.044721685349941254, "train_loss_bc": 0.044721685349941254, "train_loss_llm": 0.0, "grad_norm": 0.03181040659546852, "global_step": 8570, "epoch": 238, "lr": 0.001} +{"train_loss": 0.03305218368768692, "train_loss_bc": 0.03305218368768692, "train_loss_llm": 0.0, "grad_norm": 0.04051993042230606, "global_step": 8571, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04120407998561859, "train_loss_bc": 0.04120407998561859, "train_loss_llm": 0.0, "grad_norm": 0.045113224536180496, "global_step": 8572, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04135216027498245, "train_loss_bc": 0.04135216027498245, "train_loss_llm": 0.0, "grad_norm": 0.02279786951839924, "global_step": 8573, "epoch": 238, "lr": 0.001} +{"train_loss": 0.040703434497117996, "train_loss_bc": 0.040703434497117996, "train_loss_llm": 0.0, "grad_norm": 0.035192012786865234, "global_step": 8574, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04641464725136757, "train_loss_bc": 0.04641464725136757, "train_loss_llm": 0.0, "grad_norm": 0.04924057796597481, "global_step": 8575, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04619312286376953, "train_loss_bc": 0.04616086557507515, "train_loss_llm": 0.0032258927822113037, "grad_norm": 0.061274077743291855, "global_step": 8576, "epoch": 238, "lr": 0.001} +{"train_loss": 0.0330079011619091, "train_loss_bc": 0.03300323709845543, "train_loss_llm": 0.00046640634536743164, "grad_norm": 0.01828923635184765, "global_step": 8577, "epoch": 238, "lr": 0.001} +{"train_loss": 0.033900510519742966, "train_loss_bc": 0.033900510519742966, "train_loss_llm": 0.0, "grad_norm": 0.03183285892009735, "global_step": 8578, "epoch": 238, "lr": 0.001} +{"train_loss": 0.042931828647851944, "train_loss_bc": 0.042931828647851944, "train_loss_llm": 0.0, "grad_norm": 0.04199511185288429, "global_step": 8579, "epoch": 238, "lr": 0.001} +{"train_loss": 0.038462281227111816, "train_loss_bc": 0.038450345396995544, "train_loss_llm": 0.0011936426162719727, "grad_norm": 0.05240541696548462, "global_step": 8580, "epoch": 238, "lr": 0.001} +{"train_loss": 0.0450025349855423, "train_loss_bc": 0.0450025349855423, "train_loss_llm": 0.0, "grad_norm": 0.021580707281827927, "global_step": 8581, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04104677215218544, "train_loss_bc": 0.04104677215218544, "train_loss_llm": 0.0, "grad_norm": 0.033844172954559326, "global_step": 8582, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04564293473958969, "train_loss_bc": 0.04564293473958969, "train_loss_llm": 0.0, "grad_norm": 0.042449984699487686, "global_step": 8583, "epoch": 238, "lr": 0.001} +{"train_loss": 0.045237332582473755, "train_loss_bc": 0.045237332582473755, "train_loss_llm": 0.0, "grad_norm": 0.05771719664335251, "global_step": 8584, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04302648454904556, "train_loss_bc": 0.04302648454904556, "train_loss_llm": 0.0, "grad_norm": 0.021243978291749954, "global_step": 8585, "epoch": 238, "lr": 0.001} +{"train_loss": 0.03563203290104866, "train_loss_bc": 0.03563203290104866, "train_loss_llm": 0.0, "grad_norm": 0.030421946197748184, "global_step": 8586, "epoch": 238, "lr": 0.001} +{"train_loss": 0.040353164076805115, "train_loss_bc": 0.040353164076805115, "train_loss_llm": 0.0, "grad_norm": 0.04164351895451546, "global_step": 8587, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04261869192123413, "train_loss_bc": 0.04261869192123413, "train_loss_llm": 0.0, "grad_norm": 0.04697573930025101, "global_step": 8588, "epoch": 238, "lr": 0.001} +{"train_loss": 0.0338214747607708, "train_loss_bc": 0.0338214747607708, "train_loss_llm": 0.0, "grad_norm": 0.01427699439227581, "global_step": 8589, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04254212975502014, "train_loss_bc": 0.042526040226221085, "train_loss_llm": 0.0016088783740997314, "grad_norm": 0.02668851800262928, "global_step": 8590, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04317844659090042, "train_loss_bc": 0.04317844659090042, "train_loss_llm": 0.0, "grad_norm": 0.03299010172486305, "global_step": 8591, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04427534341812134, "train_loss_bc": 0.04427534341812134, "train_loss_llm": 0.0, "grad_norm": 0.04804743453860283, "global_step": 8592, "epoch": 238, "lr": 0.001} +{"train_loss": 0.0396897979080677, "train_loss_bc": 0.039681095629930496, "train_loss_llm": 0.0008700788021087646, "grad_norm": 0.02025875262916088, "global_step": 8593, "epoch": 238, "lr": 0.001} +{"train_loss": 0.038990650326013565, "train_loss_bc": 0.038990650326013565, "train_loss_llm": 0.0, "grad_norm": 0.027301331982016563, "global_step": 8594, "epoch": 238, "lr": 0.001} +{"train_loss": 0.04077736660838127, "train_loss_bc": 0.04077736660838127, "train_loss_llm": 0.0, "grad_norm": 0.03716301918029785, "global_step": 8595, "epoch": 238, "lr": 0.001} +{"train_loss": 0.03825414925813675, "train_loss_bc": 0.03821737319231033, "train_loss_llm": 0.0036777257919311523, "grad_norm": 0.04505088925361633, "global_step": 8596, "epoch": 238, "lr": 0.001} +{"train_loss": 0.03919978439807892, "train_loss_bc": 0.03918631747364998, "train_loss_llm": 0.0013467073440551758, "grad_norm": 0.02107176184654236, "global_step": 8597, "epoch": 238, "lr": 0.001} +{"train_loss": 0.03954608365893364, "train_loss_bc": 0.039527665823698044, "train_loss_llm": 0.0018416047096252441, "grad_norm": 0.03039691038429737, "global_step": 8598, "epoch": 238, "lr": 0.001} +{"train_loss": 0.03245090693235397, "train_loss_bc": 0.03245090693235397, "train_loss_llm": 0.0, "grad_norm": 0.040639642626047134, "global_step": 8599, "epoch": 238, "lr": 0.001} +{"train_loss": 0.033332280814647675, "train_loss_bc": 0.03331785649061203, "train_loss_llm": 0.0014425814151763916, "grad_norm": 0.049021508544683456, "global_step": 8600, "epoch": 238, "lr": 0.001} +{"train_loss": 0.03440815955400467, "train_loss_bc": 0.034406598657369614, "train_loss_llm": 0.00015607476234436035, "grad_norm": 0.02072170563042164, "global_step": 8601, "epoch": 238, "lr": 0.001} +{"train_loss": 0.036834653466939926, "train_loss_bc": 0.036834653466939926, "train_loss_llm": 0.0, "grad_norm": 0.02573939971625805, "global_step": 8602, "epoch": 238, "lr": 0.001} +{"train_loss": 0.03933776237277521, "train_loss_bc": 0.029774611815810204, "train_loss_llm": 0.0, "grad_norm": 0.04088079184293747, "global_step": 8603, "epoch": 238, "lr": 0.001, "val_loss": 0.05972941964864731} +{"train_loss": 0.03270356357097626, "train_loss_bc": 0.03270356357097626, "train_loss_llm": 0.0, "grad_norm": 0.04908515885472298, "global_step": 8604, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03996323049068451, "train_loss_bc": 0.03996323049068451, "train_loss_llm": 0.0, "grad_norm": 0.021714121103286743, "global_step": 8605, "epoch": 239, "lr": 0.001} +{"train_loss": 0.04094930738210678, "train_loss_bc": 0.04094930738210678, "train_loss_llm": 0.0, "grad_norm": 0.03132464364171028, "global_step": 8606, "epoch": 239, "lr": 0.001} +{"train_loss": 0.043048009276390076, "train_loss_bc": 0.043048009276390076, "train_loss_llm": 0.0, "grad_norm": 0.04876849800348282, "global_step": 8607, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03567088767886162, "train_loss_bc": 0.03567088767886162, "train_loss_llm": 0.0, "grad_norm": 0.05514657497406006, "global_step": 8608, "epoch": 239, "lr": 0.001} +{"train_loss": 0.04464619979262352, "train_loss_bc": 0.044643428176641464, "train_loss_llm": 0.0002773404121398926, "grad_norm": 0.01719849742949009, "global_step": 8609, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03823619708418846, "train_loss_bc": 0.03823619708418846, "train_loss_llm": 0.0, "grad_norm": 0.022585278376936913, "global_step": 8610, "epoch": 239, "lr": 0.001} +{"train_loss": 0.040455907583236694, "train_loss_bc": 0.04045458137989044, "train_loss_llm": 0.00013276934623718262, "grad_norm": 0.032289374619722366, "global_step": 8611, "epoch": 239, "lr": 0.001} +{"train_loss": 0.034922681748867035, "train_loss_bc": 0.03491029143333435, "train_loss_llm": 0.001239180564880371, "grad_norm": 0.040366604924201965, "global_step": 8612, "epoch": 239, "lr": 0.001} +{"train_loss": 0.043829090893268585, "train_loss_bc": 0.043829090893268585, "train_loss_llm": 0.0, "grad_norm": 0.019068362191319466, "global_step": 8613, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03650076314806938, "train_loss_bc": 0.036492690443992615, "train_loss_llm": 0.0008071362972259521, "grad_norm": 0.027316078543663025, "global_step": 8614, "epoch": 239, "lr": 0.001} +{"train_loss": 0.04560190811753273, "train_loss_bc": 0.04560190811753273, "train_loss_llm": 0.0, "grad_norm": 0.04026082158088684, "global_step": 8615, "epoch": 239, "lr": 0.001} +{"train_loss": 0.04040353372693062, "train_loss_bc": 0.04040353372693062, "train_loss_llm": 0.0, "grad_norm": 0.048771537840366364, "global_step": 8616, "epoch": 239, "lr": 0.001} +{"train_loss": 0.037396322935819626, "train_loss_bc": 0.037396203726530075, "train_loss_llm": 1.189112663269043e-05, "grad_norm": 0.025562217459082603, "global_step": 8617, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03692776337265968, "train_loss_bc": 0.036915119737386703, "train_loss_llm": 0.0012643635272979736, "grad_norm": 0.033394526690244675, "global_step": 8618, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03658200427889824, "train_loss_bc": 0.03658200427889824, "train_loss_llm": 0.0, "grad_norm": 0.038649849593639374, "global_step": 8619, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03481528162956238, "train_loss_bc": 0.03481528162956238, "train_loss_llm": 0.0, "grad_norm": 0.04540308564901352, "global_step": 8620, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03765008971095085, "train_loss_bc": 0.03765008971095085, "train_loss_llm": 0.0, "grad_norm": 0.021672405302524567, "global_step": 8621, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03510899096727371, "train_loss_bc": 0.03510503098368645, "train_loss_llm": 0.0003959536552429199, "grad_norm": 0.03236149996519089, "global_step": 8622, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03768787533044815, "train_loss_bc": 0.037681322544813156, "train_loss_llm": 0.0006553828716278076, "grad_norm": 0.039182182401418686, "global_step": 8623, "epoch": 239, "lr": 0.001} +{"train_loss": 0.04090315103530884, "train_loss_bc": 0.040897976607084274, "train_loss_llm": 0.0005175471305847168, "grad_norm": 0.05435577780008316, "global_step": 8624, "epoch": 239, "lr": 0.001} +{"train_loss": 0.040729958564043045, "train_loss_bc": 0.040729958564043045, "train_loss_llm": 0.0, "grad_norm": 0.021156352013349533, "global_step": 8625, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03573836758732796, "train_loss_bc": 0.03573836758732796, "train_loss_llm": 0.0, "grad_norm": 0.03721175342798233, "global_step": 8626, "epoch": 239, "lr": 0.001} +{"train_loss": 0.035885289311409, "train_loss_bc": 0.035885289311409, "train_loss_llm": 0.0, "grad_norm": 0.04282931983470917, "global_step": 8627, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03979804739356041, "train_loss_bc": 0.03979409113526344, "train_loss_llm": 0.0003956258296966553, "grad_norm": 0.053189583122730255, "global_step": 8628, "epoch": 239, "lr": 0.001} +{"train_loss": 0.04784867912530899, "train_loss_bc": 0.04784867912530899, "train_loss_llm": 0.0, "grad_norm": 0.02058473229408264, "global_step": 8629, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03942525386810303, "train_loss_bc": 0.039411887526512146, "train_loss_llm": 0.0013367533683776855, "grad_norm": 0.03130244091153145, "global_step": 8630, "epoch": 239, "lr": 0.001} +{"train_loss": 0.04017910733819008, "train_loss_bc": 0.04017910733819008, "train_loss_llm": 0.0, "grad_norm": 0.03835075721144676, "global_step": 8631, "epoch": 239, "lr": 0.001} +{"train_loss": 0.040485721081495285, "train_loss_bc": 0.04047447070479393, "train_loss_llm": 0.001125037670135498, "grad_norm": 0.04936717450618744, "global_step": 8632, "epoch": 239, "lr": 0.001} +{"train_loss": 0.030503325164318085, "train_loss_bc": 0.030503325164318085, "train_loss_llm": 0.0, "grad_norm": 0.01786879450082779, "global_step": 8633, "epoch": 239, "lr": 0.001} +{"train_loss": 0.04458295553922653, "train_loss_bc": 0.04458295553922653, "train_loss_llm": 0.0, "grad_norm": 0.030544426292181015, "global_step": 8634, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03752904385328293, "train_loss_bc": 0.037518154829740524, "train_loss_llm": 0.0010890364646911621, "grad_norm": 0.037836238741874695, "global_step": 8635, "epoch": 239, "lr": 0.001} +{"train_loss": 0.04505032300949097, "train_loss_bc": 0.04505032300949097, "train_loss_llm": 0.0, "grad_norm": 0.04879511147737503, "global_step": 8636, "epoch": 239, "lr": 0.001} +{"train_loss": 0.045356281101703644, "train_loss_bc": 0.04533200338482857, "train_loss_llm": 0.002427905797958374, "grad_norm": 0.027005212381482124, "global_step": 8637, "epoch": 239, "lr": 0.001} +{"train_loss": 0.04105202108621597, "train_loss_bc": 0.041034940630197525, "train_loss_llm": 0.0017081201076507568, "grad_norm": 0.03684863820672035, "global_step": 8638, "epoch": 239, "lr": 0.001} +{"train_loss": 0.03952653457721075, "train_loss_bc": 0.04478811100125313, "train_loss_llm": 0.0, "grad_norm": 0.05063450336456299, "global_step": 8639, "epoch": 239, "lr": 0.001, "val_loss": 0.058227792382240295} +{"train_loss": 0.04006759449839592, "train_loss_bc": 0.04005011171102524, "train_loss_llm": 0.0017483532428741455, "grad_norm": 0.057825133204460144, "global_step": 8640, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03662308678030968, "train_loss_bc": 0.03660891205072403, "train_loss_llm": 0.0014175176620483398, "grad_norm": 0.018074743449687958, "global_step": 8641, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03516851365566254, "train_loss_bc": 0.035168424248695374, "train_loss_llm": 8.970499038696289e-06, "grad_norm": 0.026183372363448143, "global_step": 8642, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03698332607746124, "train_loss_bc": 0.03698332607746124, "train_loss_llm": 0.0, "grad_norm": 0.03690927103161812, "global_step": 8643, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03785703331232071, "train_loss_bc": 0.03785325586795807, "train_loss_llm": 0.0003777146339416504, "grad_norm": 0.045001350343227386, "global_step": 8644, "epoch": 240, "lr": 0.001} +{"train_loss": 0.04467633366584778, "train_loss_bc": 0.04467633366584778, "train_loss_llm": 0.0, "grad_norm": 0.024268830195069313, "global_step": 8645, "epoch": 240, "lr": 0.001} +{"train_loss": 0.04820229858160019, "train_loss_bc": 0.04820229858160019, "train_loss_llm": 0.0, "grad_norm": 0.03794938325881958, "global_step": 8646, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03549366816878319, "train_loss_bc": 0.03549366816878319, "train_loss_llm": 0.0, "grad_norm": 0.041252247989177704, "global_step": 8647, "epoch": 240, "lr": 0.001} +{"train_loss": 0.04724220186471939, "train_loss_bc": 0.04724220186471939, "train_loss_llm": 0.0, "grad_norm": 0.04986400529742241, "global_step": 8648, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03307674825191498, "train_loss_bc": 0.03307674825191498, "train_loss_llm": 0.0, "grad_norm": 0.019211843609809875, "global_step": 8649, "epoch": 240, "lr": 0.001} +{"train_loss": 0.05003105849027634, "train_loss_bc": 0.05003105849027634, "train_loss_llm": 0.0, "grad_norm": 0.03185976296663284, "global_step": 8650, "epoch": 240, "lr": 0.001} +{"train_loss": 0.04499739408493042, "train_loss_bc": 0.04499739408493042, "train_loss_llm": 0.0, "grad_norm": 0.040971625596284866, "global_step": 8651, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03532572090625763, "train_loss_bc": 0.03532572090625763, "train_loss_llm": 0.0, "grad_norm": 0.051169246435165405, "global_step": 8652, "epoch": 240, "lr": 0.001} +{"train_loss": 0.04404516890645027, "train_loss_bc": 0.04404516890645027, "train_loss_llm": 0.0, "grad_norm": 0.018535004928708076, "global_step": 8653, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03897721692919731, "train_loss_bc": 0.03897721692919731, "train_loss_llm": 0.0, "grad_norm": 0.031124824658036232, "global_step": 8654, "epoch": 240, "lr": 0.001} +{"train_loss": 0.04177713766694069, "train_loss_bc": 0.04177407920360565, "train_loss_llm": 0.0003058016300201416, "grad_norm": 0.04491462931036949, "global_step": 8655, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03936462476849556, "train_loss_bc": 0.03936462476849556, "train_loss_llm": 0.0, "grad_norm": 0.06012758985161781, "global_step": 8656, "epoch": 240, "lr": 0.001} +{"train_loss": 0.044349100440740585, "train_loss_bc": 0.04433203861117363, "train_loss_llm": 0.0017060637474060059, "grad_norm": 0.016222836449742317, "global_step": 8657, "epoch": 240, "lr": 0.001} +{"train_loss": 0.041621431708335876, "train_loss_bc": 0.04161117970943451, "train_loss_llm": 0.0010251402854919434, "grad_norm": 0.031632326543331146, "global_step": 8658, "epoch": 240, "lr": 0.001} +{"train_loss": 0.034706756472587585, "train_loss_bc": 0.034706756472587585, "train_loss_llm": 0.0, "grad_norm": 0.037891585379838943, "global_step": 8659, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03732244297862053, "train_loss_bc": 0.03732244297862053, "train_loss_llm": 0.0, "grad_norm": 0.04312556982040405, "global_step": 8660, "epoch": 240, "lr": 0.001} +{"train_loss": 0.038888853043317795, "train_loss_bc": 0.038888853043317795, "train_loss_llm": 0.0, "grad_norm": 0.01948016881942749, "global_step": 8661, "epoch": 240, "lr": 0.001} +{"train_loss": 0.04353154078125954, "train_loss_bc": 0.04353154078125954, "train_loss_llm": 0.0, "grad_norm": 0.03101726621389389, "global_step": 8662, "epoch": 240, "lr": 0.001} +{"train_loss": 0.034293822944164276, "train_loss_bc": 0.03429169952869415, "train_loss_llm": 0.0002122819423675537, "grad_norm": 0.03894268721342087, "global_step": 8663, "epoch": 240, "lr": 0.001} +{"train_loss": 0.0442056842148304, "train_loss_bc": 0.0442056842148304, "train_loss_llm": 0.0, "grad_norm": 0.049435585737228394, "global_step": 8664, "epoch": 240, "lr": 0.001} +{"train_loss": 0.045382916927337646, "train_loss_bc": 0.045382916927337646, "train_loss_llm": 0.0, "grad_norm": 0.019977740943431854, "global_step": 8665, "epoch": 240, "lr": 0.001} +{"train_loss": 0.035621363669633865, "train_loss_bc": 0.0356052964925766, "train_loss_llm": 0.0016067326068878174, "grad_norm": 0.03190941736102104, "global_step": 8666, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03841780871152878, "train_loss_bc": 0.03841356560587883, "train_loss_llm": 0.0004241466522216797, "grad_norm": 0.0471508763730526, "global_step": 8667, "epoch": 240, "lr": 0.001} +{"train_loss": 0.036916334182024, "train_loss_bc": 0.03689514845609665, "train_loss_llm": 0.002118438482284546, "grad_norm": 0.05507476255297661, "global_step": 8668, "epoch": 240, "lr": 0.001} +{"train_loss": 0.04349730163812637, "train_loss_bc": 0.04349730163812637, "train_loss_llm": 0.0, "grad_norm": 0.020808205008506775, "global_step": 8669, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03213895857334137, "train_loss_bc": 0.03213895857334137, "train_loss_llm": 0.0, "grad_norm": 0.033151257783174515, "global_step": 8670, "epoch": 240, "lr": 0.001} +{"train_loss": 0.029561612755060196, "train_loss_bc": 0.029561612755060196, "train_loss_llm": 0.0, "grad_norm": 0.039230361580848694, "global_step": 8671, "epoch": 240, "lr": 0.001} +{"train_loss": 0.037694741040468216, "train_loss_bc": 0.037694741040468216, "train_loss_llm": 0.0, "grad_norm": 0.05165260657668114, "global_step": 8672, "epoch": 240, "lr": 0.001} +{"train_loss": 0.02869814820587635, "train_loss_bc": 0.02867140993475914, "train_loss_llm": 0.002673894166946411, "grad_norm": 0.017773350700736046, "global_step": 8673, "epoch": 240, "lr": 0.001} +{"train_loss": 0.039880331605672836, "train_loss_bc": 0.03986474499106407, "train_loss_llm": 0.001558750867843628, "grad_norm": 0.0281340591609478, "global_step": 8674, "epoch": 240, "lr": 0.001} +{"train_loss": 0.03937972404269709, "train_loss_bc": 0.041028622537851334, "train_loss_llm": 0.0003167092800140381, "grad_norm": 0.04599083214998245, "global_step": 8675, "epoch": 240, "lr": 0.001, "train/cumulative_reward": 102.99088571886759, "train/mean_score": 16.713215984525988, "train/success_rate": 0.5, "test/cumulative_reward": 40.26658296297705, "test/mean_score": 3.9793780392082656, "test/success_rate": 0.06, "val_loss": 0.0589715912938118, "train_action_mse_error": 0.04293088987469673} +{"train_loss": 0.034032732248306274, "train_loss_bc": 0.03400279954075813, "train_loss_llm": 0.0029934048652648926, "grad_norm": 0.051635708659887314, "global_step": 8676, "epoch": 241, "lr": 0.001} +{"train_loss": 0.04215187206864357, "train_loss_bc": 0.042140550911426544, "train_loss_llm": 0.001131981611251831, "grad_norm": 0.02453809231519699, "global_step": 8677, "epoch": 241, "lr": 0.001} +{"train_loss": 0.04294184595346451, "train_loss_bc": 0.04294184595346451, "train_loss_llm": 0.0, "grad_norm": 0.03143645077943802, "global_step": 8678, "epoch": 241, "lr": 0.001} +{"train_loss": 0.04519663751125336, "train_loss_bc": 0.04518110305070877, "train_loss_llm": 0.0015535354614257812, "grad_norm": 0.040854573249816895, "global_step": 8679, "epoch": 241, "lr": 0.001} +{"train_loss": 0.041001006960868835, "train_loss_bc": 0.041001006960868835, "train_loss_llm": 0.0, "grad_norm": 0.04803788661956787, "global_step": 8680, "epoch": 241, "lr": 0.001} +{"train_loss": 0.031153256073594093, "train_loss_bc": 0.031153256073594093, "train_loss_llm": 0.0, "grad_norm": 0.016047358512878418, "global_step": 8681, "epoch": 241, "lr": 0.001} +{"train_loss": 0.04135864973068237, "train_loss_bc": 0.04135864973068237, "train_loss_llm": 0.0, "grad_norm": 0.027172906324267387, "global_step": 8682, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03828743100166321, "train_loss_bc": 0.03828743100166321, "train_loss_llm": 0.0, "grad_norm": 0.03760068118572235, "global_step": 8683, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03279707953333855, "train_loss_bc": 0.03279707953333855, "train_loss_llm": 0.0, "grad_norm": 0.04208526015281677, "global_step": 8684, "epoch": 241, "lr": 0.001} +{"train_loss": 0.0430879145860672, "train_loss_bc": 0.0430879145860672, "train_loss_llm": 0.0, "grad_norm": 0.01709946244955063, "global_step": 8685, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03968915343284607, "train_loss_bc": 0.03967387601733208, "train_loss_llm": 0.0015278458595275879, "grad_norm": 0.026959937065839767, "global_step": 8686, "epoch": 241, "lr": 0.001} +{"train_loss": 0.04230288416147232, "train_loss_bc": 0.04230288416147232, "train_loss_llm": 0.0, "grad_norm": 0.0365871787071228, "global_step": 8687, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03382299840450287, "train_loss_bc": 0.03382299840450287, "train_loss_llm": 0.0, "grad_norm": 0.04823331534862518, "global_step": 8688, "epoch": 241, "lr": 0.001} +{"train_loss": 0.038764405995607376, "train_loss_bc": 0.038764405995607376, "train_loss_llm": 0.0, "grad_norm": 0.02534140646457672, "global_step": 8689, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03615124151110649, "train_loss_bc": 0.03615124151110649, "train_loss_llm": 0.0, "grad_norm": 0.040689364075660706, "global_step": 8690, "epoch": 241, "lr": 0.001} +{"train_loss": 0.0351356603205204, "train_loss_bc": 0.0351356603205204, "train_loss_llm": 0.0, "grad_norm": 0.047522421926259995, "global_step": 8691, "epoch": 241, "lr": 0.001} +{"train_loss": 0.04602358117699623, "train_loss_bc": 0.04602358117699623, "train_loss_llm": 0.0, "grad_norm": 0.053763680160045624, "global_step": 8692, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03758486360311508, "train_loss_bc": 0.03757855296134949, "train_loss_llm": 0.0006311535835266113, "grad_norm": 0.017465753480792046, "global_step": 8693, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03919031471014023, "train_loss_bc": 0.0391881950199604, "train_loss_llm": 0.00021183490753173828, "grad_norm": 0.025710957124829292, "global_step": 8694, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03754691407084465, "train_loss_bc": 0.03754691407084465, "train_loss_llm": 0.0, "grad_norm": 0.03549501672387123, "global_step": 8695, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03695985674858093, "train_loss_bc": 0.036947835236787796, "train_loss_llm": 0.0012022554874420166, "grad_norm": 0.047332506626844406, "global_step": 8696, "epoch": 241, "lr": 0.001} +{"train_loss": 0.04512311890721321, "train_loss_bc": 0.04512311890721321, "train_loss_llm": 0.0, "grad_norm": 0.018341680988669395, "global_step": 8697, "epoch": 241, "lr": 0.001} +{"train_loss": 0.04447484388947487, "train_loss_bc": 0.04447484388947487, "train_loss_llm": 0.0, "grad_norm": 0.035266101360321045, "global_step": 8698, "epoch": 241, "lr": 0.001} +{"train_loss": 0.042571060359478, "train_loss_bc": 0.042571060359478, "train_loss_llm": 0.0, "grad_norm": 0.04425016790628433, "global_step": 8699, "epoch": 241, "lr": 0.001} +{"train_loss": 0.042437758296728134, "train_loss_bc": 0.042437758296728134, "train_loss_llm": 0.0, "grad_norm": 0.051341600716114044, "global_step": 8700, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03412376716732979, "train_loss_bc": 0.03412376716732979, "train_loss_llm": 0.0, "grad_norm": 0.019203314557671547, "global_step": 8701, "epoch": 241, "lr": 0.001} +{"train_loss": 0.035409849137067795, "train_loss_bc": 0.035409849137067795, "train_loss_llm": 0.0, "grad_norm": 0.033527906984090805, "global_step": 8702, "epoch": 241, "lr": 0.001} +{"train_loss": 0.040677495300769806, "train_loss_bc": 0.040677495300769806, "train_loss_llm": 0.0, "grad_norm": 0.036899227648973465, "global_step": 8703, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03294725716114044, "train_loss_bc": 0.03294725716114044, "train_loss_llm": 0.0, "grad_norm": 0.04206254705786705, "global_step": 8704, "epoch": 241, "lr": 0.001} +{"train_loss": 0.0368976853787899, "train_loss_bc": 0.0368976853787899, "train_loss_llm": 0.0, "grad_norm": 0.023666847497224808, "global_step": 8705, "epoch": 241, "lr": 0.001} +{"train_loss": 0.04087907075881958, "train_loss_bc": 0.04087907075881958, "train_loss_llm": 0.0, "grad_norm": 0.04271983727812767, "global_step": 8706, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03554292023181915, "train_loss_bc": 0.03554292023181915, "train_loss_llm": 0.0, "grad_norm": 0.05371581390500069, "global_step": 8707, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03863772004842758, "train_loss_bc": 0.03863772004842758, "train_loss_llm": 0.0, "grad_norm": 0.05670662224292755, "global_step": 8708, "epoch": 241, "lr": 0.001} +{"train_loss": 0.04703881964087486, "train_loss_bc": 0.04703853279352188, "train_loss_llm": 2.8789043426513672e-05, "grad_norm": 0.022325636819005013, "global_step": 8709, "epoch": 241, "lr": 0.001} +{"train_loss": 0.035424768924713135, "train_loss_bc": 0.0354134663939476, "train_loss_llm": 0.0011301636695861816, "grad_norm": 0.02500752918422222, "global_step": 8710, "epoch": 241, "lr": 0.001} +{"train_loss": 0.03935098311760359, "train_loss_bc": 0.04926895722746849, "train_loss_llm": 0.0, "grad_norm": 0.03665458783507347, "global_step": 8711, "epoch": 241, "lr": 0.001, "val_loss": 0.058772195130586624} +{"train_loss": 0.03933463990688324, "train_loss_bc": 0.03933463990688324, "train_loss_llm": 0.0, "grad_norm": 0.046794239431619644, "global_step": 8712, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04073558747768402, "train_loss_bc": 0.04073558747768402, "train_loss_llm": 0.0, "grad_norm": 0.03021221235394478, "global_step": 8713, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04182363674044609, "train_loss_bc": 0.04182363674044609, "train_loss_llm": 0.0, "grad_norm": 0.03872379660606384, "global_step": 8714, "epoch": 242, "lr": 0.001} +{"train_loss": 0.040985140949487686, "train_loss_bc": 0.040985140949487686, "train_loss_llm": 0.0, "grad_norm": 0.05477548763155937, "global_step": 8715, "epoch": 242, "lr": 0.001} +{"train_loss": 0.038035523146390915, "train_loss_bc": 0.0380169078707695, "train_loss_llm": 0.0018616914749145508, "grad_norm": 0.06508781015872955, "global_step": 8716, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04756855592131615, "train_loss_bc": 0.04756855592131615, "train_loss_llm": 0.0, "grad_norm": 0.02284441515803337, "global_step": 8717, "epoch": 242, "lr": 0.001} +{"train_loss": 0.040379803627729416, "train_loss_bc": 0.040379803627729416, "train_loss_llm": 0.0, "grad_norm": 0.031627412885427475, "global_step": 8718, "epoch": 242, "lr": 0.001} +{"train_loss": 0.0385303795337677, "train_loss_bc": 0.0385303795337677, "train_loss_llm": 0.0, "grad_norm": 0.042466625571250916, "global_step": 8719, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03958794102072716, "train_loss_bc": 0.03958794102072716, "train_loss_llm": 0.0, "grad_norm": 0.05666140466928482, "global_step": 8720, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03705422580242157, "train_loss_bc": 0.03705422580242157, "train_loss_llm": 0.0, "grad_norm": 0.018494468182325363, "global_step": 8721, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03845961019396782, "train_loss_bc": 0.03845703974366188, "train_loss_llm": 0.000256955623626709, "grad_norm": 0.027720170095562935, "global_step": 8722, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03434716910123825, "train_loss_bc": 0.03434716910123825, "train_loss_llm": 0.0, "grad_norm": 0.03453986719250679, "global_step": 8723, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03352928161621094, "train_loss_bc": 0.03352928161621094, "train_loss_llm": 0.0, "grad_norm": 0.03875836730003357, "global_step": 8724, "epoch": 242, "lr": 0.001} +{"train_loss": 0.042573533952236176, "train_loss_bc": 0.042573533952236176, "train_loss_llm": 0.0, "grad_norm": 0.021654943004250526, "global_step": 8725, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04405814781785011, "train_loss_bc": 0.04405814781785011, "train_loss_llm": 0.0, "grad_norm": 0.036707568913698196, "global_step": 8726, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04009811207652092, "train_loss_bc": 0.040070127695798874, "train_loss_llm": 0.0027982890605926514, "grad_norm": 0.0461876355111599, "global_step": 8727, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04544704034924507, "train_loss_bc": 0.045437250286340714, "train_loss_llm": 0.0009791851043701172, "grad_norm": 0.046727463603019714, "global_step": 8728, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04376488924026489, "train_loss_bc": 0.04376201704144478, "train_loss_llm": 0.00028705596923828125, "grad_norm": 0.020724231377243996, "global_step": 8729, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03963795676827431, "train_loss_bc": 0.03960936889052391, "train_loss_llm": 0.0028589069843292236, "grad_norm": 0.028058378025889397, "global_step": 8730, "epoch": 242, "lr": 0.001} +{"train_loss": 0.028290312737226486, "train_loss_bc": 0.028273561969399452, "train_loss_llm": 0.0016751587390899658, "grad_norm": 0.03994978591799736, "global_step": 8731, "epoch": 242, "lr": 0.001} +{"train_loss": 0.043432269245386124, "train_loss_bc": 0.04342181608080864, "train_loss_llm": 0.0010454654693603516, "grad_norm": 0.05098189413547516, "global_step": 8732, "epoch": 242, "lr": 0.001} +{"train_loss": 0.035921674221754074, "train_loss_bc": 0.0359160378575325, "train_loss_llm": 0.0005636513233184814, "grad_norm": 0.01900099404156208, "global_step": 8733, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03527633473277092, "train_loss_bc": 0.03525223210453987, "train_loss_llm": 0.002410322427749634, "grad_norm": 0.031522735953330994, "global_step": 8734, "epoch": 242, "lr": 0.001} +{"train_loss": 0.024469174444675446, "train_loss_bc": 0.024469174444675446, "train_loss_llm": 0.0, "grad_norm": 0.04049833491444588, "global_step": 8735, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03772590681910515, "train_loss_bc": 0.03772590681910515, "train_loss_llm": 0.0, "grad_norm": 0.049917008727788925, "global_step": 8736, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03770552948117256, "train_loss_bc": 0.03770552948117256, "train_loss_llm": 0.0, "grad_norm": 0.020652692764997482, "global_step": 8737, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03756751865148544, "train_loss_bc": 0.03755343332886696, "train_loss_llm": 0.001408606767654419, "grad_norm": 0.03039204142987728, "global_step": 8738, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04851855710148811, "train_loss_bc": 0.04851855710148811, "train_loss_llm": 0.0, "grad_norm": 0.051262419670820236, "global_step": 8739, "epoch": 242, "lr": 0.001} +{"train_loss": 0.0375046469271183, "train_loss_bc": 0.03749074786901474, "train_loss_llm": 0.0013899803161621094, "grad_norm": 0.06576427072286606, "global_step": 8740, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04589686542749405, "train_loss_bc": 0.04589686542749405, "train_loss_llm": 0.0, "grad_norm": 0.01811624877154827, "global_step": 8741, "epoch": 242, "lr": 0.001} +{"train_loss": 0.042160313576459885, "train_loss_bc": 0.042160313576459885, "train_loss_llm": 0.0, "grad_norm": 0.028857814148068428, "global_step": 8742, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04352206736803055, "train_loss_bc": 0.04352206736803055, "train_loss_llm": 0.0, "grad_norm": 0.038617122918367386, "global_step": 8743, "epoch": 242, "lr": 0.001} +{"train_loss": 0.04119701683521271, "train_loss_bc": 0.04119701683521271, "train_loss_llm": 0.0, "grad_norm": 0.04571735858917236, "global_step": 8744, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03442050889134407, "train_loss_bc": 0.034414634108543396, "train_loss_llm": 0.0005875825881958008, "grad_norm": 0.017808178439736366, "global_step": 8745, "epoch": 242, "lr": 0.001} +{"train_loss": 0.03760073333978653, "train_loss_bc": 0.03760073333978653, "train_loss_llm": 0.0, "grad_norm": 0.028242656961083412, "global_step": 8746, "epoch": 242, "lr": 0.001} +{"train_loss": 0.039289585418171354, "train_loss_bc": 0.037264470010995865, "train_loss_llm": 0.0, "grad_norm": 0.041478272527456284, "global_step": 8747, "epoch": 242, "lr": 0.001, "val_loss": 0.05937730148434639} +{"train_loss": 0.038092829287052155, "train_loss_bc": 0.038092829287052155, "train_loss_llm": 0.0, "grad_norm": 0.047354187816381454, "global_step": 8748, "epoch": 243, "lr": 0.001} +{"train_loss": 0.031093301251530647, "train_loss_bc": 0.031093301251530647, "train_loss_llm": 0.0, "grad_norm": 0.020238181576132774, "global_step": 8749, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03925328701734543, "train_loss_bc": 0.03925328701734543, "train_loss_llm": 0.0, "grad_norm": 0.03346368297934532, "global_step": 8750, "epoch": 243, "lr": 0.001} +{"train_loss": 0.049554821103811264, "train_loss_bc": 0.04954291507601738, "train_loss_llm": 0.0011906027793884277, "grad_norm": 0.04986845701932907, "global_step": 8751, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03896236792206764, "train_loss_bc": 0.03896236792206764, "train_loss_llm": 0.0, "grad_norm": 0.056365277618169785, "global_step": 8752, "epoch": 243, "lr": 0.001} +{"train_loss": 0.04038485884666443, "train_loss_bc": 0.04038485884666443, "train_loss_llm": 0.0, "grad_norm": 0.01789587177336216, "global_step": 8753, "epoch": 243, "lr": 0.001} +{"train_loss": 0.04622861370444298, "train_loss_bc": 0.04622861370444298, "train_loss_llm": 0.0, "grad_norm": 0.024478724226355553, "global_step": 8754, "epoch": 243, "lr": 0.001} +{"train_loss": 0.02914612367749214, "train_loss_bc": 0.02913946658372879, "train_loss_llm": 0.0006656944751739502, "grad_norm": 0.02656519040465355, "global_step": 8755, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03429259732365608, "train_loss_bc": 0.03429259732365608, "train_loss_llm": 0.0, "grad_norm": 0.03575553372502327, "global_step": 8756, "epoch": 243, "lr": 0.001} +{"train_loss": 0.04860757291316986, "train_loss_bc": 0.04860757291316986, "train_loss_llm": 0.0, "grad_norm": 0.021208791062235832, "global_step": 8757, "epoch": 243, "lr": 0.001} +{"train_loss": 0.035799045115709305, "train_loss_bc": 0.035799045115709305, "train_loss_llm": 0.0, "grad_norm": 0.028722696006298065, "global_step": 8758, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03168253228068352, "train_loss_bc": 0.03168253228068352, "train_loss_llm": 0.0, "grad_norm": 0.034536153078079224, "global_step": 8759, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03330247476696968, "train_loss_bc": 0.03330044820904732, "train_loss_llm": 0.00020247697830200195, "grad_norm": 0.04457671195268631, "global_step": 8760, "epoch": 243, "lr": 0.001} +{"train_loss": 0.041132912039756775, "train_loss_bc": 0.041132912039756775, "train_loss_llm": 0.0, "grad_norm": 0.018733618780970573, "global_step": 8761, "epoch": 243, "lr": 0.001} +{"train_loss": 0.034718457609415054, "train_loss_bc": 0.034710656851530075, "train_loss_llm": 0.0007798969745635986, "grad_norm": 0.02684933692216873, "global_step": 8762, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03574391454458237, "train_loss_bc": 0.03574391454458237, "train_loss_llm": 0.0, "grad_norm": 0.032894961535930634, "global_step": 8763, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03651486709713936, "train_loss_bc": 0.03651411458849907, "train_loss_llm": 7.513165473937988e-05, "grad_norm": 0.04780541732907295, "global_step": 8764, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03933839872479439, "train_loss_bc": 0.03933839872479439, "train_loss_llm": 0.0, "grad_norm": 0.019231313839554787, "global_step": 8765, "epoch": 243, "lr": 0.001} +{"train_loss": 0.04750458151102066, "train_loss_bc": 0.04747128114104271, "train_loss_llm": 0.0033299624919891357, "grad_norm": 0.029626116156578064, "global_step": 8766, "epoch": 243, "lr": 0.001} +{"train_loss": 0.04416608065366745, "train_loss_bc": 0.04416608065366745, "train_loss_llm": 0.0, "grad_norm": 0.04102593660354614, "global_step": 8767, "epoch": 243, "lr": 0.001} +{"train_loss": 0.0425080880522728, "train_loss_bc": 0.0425080880522728, "train_loss_llm": 0.0, "grad_norm": 0.04427560418844223, "global_step": 8768, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03872444108128548, "train_loss_bc": 0.03872444108128548, "train_loss_llm": 0.0, "grad_norm": 0.01986715942621231, "global_step": 8769, "epoch": 243, "lr": 0.001} +{"train_loss": 0.035900238901376724, "train_loss_bc": 0.035900238901376724, "train_loss_llm": 0.0, "grad_norm": 0.029817303642630577, "global_step": 8770, "epoch": 243, "lr": 0.001} +{"train_loss": 0.037105005234479904, "train_loss_bc": 0.037105005234479904, "train_loss_llm": 0.0, "grad_norm": 0.039604563266038895, "global_step": 8771, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03583355247974396, "train_loss_bc": 0.03583355247974396, "train_loss_llm": 0.0, "grad_norm": 0.04984224587678909, "global_step": 8772, "epoch": 243, "lr": 0.001} +{"train_loss": 0.04221615195274353, "train_loss_bc": 0.042193297296762466, "train_loss_llm": 0.0022854506969451904, "grad_norm": 0.020507704466581345, "global_step": 8773, "epoch": 243, "lr": 0.001} +{"train_loss": 0.04048102721571922, "train_loss_bc": 0.04047122597694397, "train_loss_llm": 0.0009800493717193604, "grad_norm": 0.028446899726986885, "global_step": 8774, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03843643143773079, "train_loss_bc": 0.038413986563682556, "train_loss_llm": 0.0022444725036621094, "grad_norm": 0.03933967649936676, "global_step": 8775, "epoch": 243, "lr": 0.001} +{"train_loss": 0.037757135927677155, "train_loss_bc": 0.03775297850370407, "train_loss_llm": 0.0004158914089202881, "grad_norm": 0.05267104133963585, "global_step": 8776, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03819826617836952, "train_loss_bc": 0.03819826617836952, "train_loss_llm": 0.0, "grad_norm": 0.017450187355279922, "global_step": 8777, "epoch": 243, "lr": 0.001} +{"train_loss": 0.047736067324876785, "train_loss_bc": 0.04772122949361801, "train_loss_llm": 0.0014836490154266357, "grad_norm": 0.027558542788028717, "global_step": 8778, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03936268016695976, "train_loss_bc": 0.03936268016695976, "train_loss_llm": 0.0, "grad_norm": 0.040318552404642105, "global_step": 8779, "epoch": 243, "lr": 0.001} +{"train_loss": 0.04026084020733833, "train_loss_bc": 0.04026084020733833, "train_loss_llm": 0.0, "grad_norm": 0.04948728531599045, "global_step": 8780, "epoch": 243, "lr": 0.001} +{"train_loss": 0.0375007688999176, "train_loss_bc": 0.037493184208869934, "train_loss_llm": 0.0007585585117340088, "grad_norm": 0.01754244975745678, "global_step": 8781, "epoch": 243, "lr": 0.001} +{"train_loss": 0.04077388346195221, "train_loss_bc": 0.040770210325717926, "train_loss_llm": 0.00036719441413879395, "grad_norm": 0.027403494343161583, "global_step": 8782, "epoch": 243, "lr": 0.001} +{"train_loss": 0.03927627583551738, "train_loss_bc": 0.045631714165210724, "train_loss_llm": 0.0, "grad_norm": 0.039765164256095886, "global_step": 8783, "epoch": 243, "lr": 0.001, "val_loss": 0.05852419510483742} +{"train_loss": 0.042849648743867874, "train_loss_bc": 0.042849648743867874, "train_loss_llm": 0.0, "grad_norm": 0.04686989262700081, "global_step": 8784, "epoch": 244, "lr": 0.001} +{"train_loss": 0.04341687634587288, "train_loss_bc": 0.04341687634587288, "train_loss_llm": 0.0, "grad_norm": 0.01848205365240574, "global_step": 8785, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03890945389866829, "train_loss_bc": 0.03890945389866829, "train_loss_llm": 0.0, "grad_norm": 0.027417467907071114, "global_step": 8786, "epoch": 244, "lr": 0.001} +{"train_loss": 0.04231462627649307, "train_loss_bc": 0.04229584336280823, "train_loss_llm": 0.0018781721591949463, "grad_norm": 0.034522198140621185, "global_step": 8787, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03670326620340347, "train_loss_bc": 0.03670326620340347, "train_loss_llm": 0.0, "grad_norm": 0.04092033579945564, "global_step": 8788, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03413122519850731, "train_loss_bc": 0.03413122519850731, "train_loss_llm": 0.0, "grad_norm": 0.01745135895907879, "global_step": 8789, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03138617426156998, "train_loss_bc": 0.03138617426156998, "train_loss_llm": 0.0, "grad_norm": 0.02574106678366661, "global_step": 8790, "epoch": 244, "lr": 0.001} +{"train_loss": 0.04193528741598129, "train_loss_bc": 0.04193528741598129, "train_loss_llm": 0.0, "grad_norm": 0.040896251797676086, "global_step": 8791, "epoch": 244, "lr": 0.001} +{"train_loss": 0.0382196418941021, "train_loss_bc": 0.0382196418941021, "train_loss_llm": 0.0, "grad_norm": 0.05047227442264557, "global_step": 8792, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03767617791891098, "train_loss_bc": 0.037663597613573074, "train_loss_llm": 0.00125807523727417, "grad_norm": 0.021161360666155815, "global_step": 8793, "epoch": 244, "lr": 0.001} +{"train_loss": 0.040579598397016525, "train_loss_bc": 0.040579598397016525, "train_loss_llm": 0.0, "grad_norm": 0.03299681469798088, "global_step": 8794, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03985902667045593, "train_loss_bc": 0.039855532348155975, "train_loss_llm": 0.00034925341606140137, "grad_norm": 0.04616576060652733, "global_step": 8795, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03920162096619606, "train_loss_bc": 0.03920162096619606, "train_loss_llm": 0.0, "grad_norm": 0.05723513662815094, "global_step": 8796, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03975541889667511, "train_loss_bc": 0.03975541889667511, "train_loss_llm": 0.0, "grad_norm": 0.022271163761615753, "global_step": 8797, "epoch": 244, "lr": 0.001} +{"train_loss": 0.04160907864570618, "train_loss_bc": 0.04160907864570618, "train_loss_llm": 0.0, "grad_norm": 0.029024558141827583, "global_step": 8798, "epoch": 244, "lr": 0.001} +{"train_loss": 0.036357391625642776, "train_loss_bc": 0.036357391625642776, "train_loss_llm": 0.0, "grad_norm": 0.03392307460308075, "global_step": 8799, "epoch": 244, "lr": 0.001} +{"train_loss": 0.036779262125492096, "train_loss_bc": 0.036779262125492096, "train_loss_llm": 0.0, "grad_norm": 0.03744526579976082, "global_step": 8800, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03857241943478584, "train_loss_bc": 0.03857241943478584, "train_loss_llm": 0.0, "grad_norm": 0.021360479295253754, "global_step": 8801, "epoch": 244, "lr": 0.001} +{"train_loss": 0.032643236219882965, "train_loss_bc": 0.032643236219882965, "train_loss_llm": 0.0, "grad_norm": 0.02874758280813694, "global_step": 8802, "epoch": 244, "lr": 0.001} +{"train_loss": 0.044529907405376434, "train_loss_bc": 0.044529907405376434, "train_loss_llm": 0.0, "grad_norm": 0.036138180643320084, "global_step": 8803, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03409937024116516, "train_loss_bc": 0.03409937024116516, "train_loss_llm": 0.0, "grad_norm": 0.04541052132844925, "global_step": 8804, "epoch": 244, "lr": 0.001} +{"train_loss": 0.042286381125450134, "train_loss_bc": 0.04227007180452347, "train_loss_llm": 0.0016309022903442383, "grad_norm": 0.020207975059747696, "global_step": 8805, "epoch": 244, "lr": 0.001} +{"train_loss": 0.045590370893478394, "train_loss_bc": 0.045590370893478394, "train_loss_llm": 0.0, "grad_norm": 0.029371974989771843, "global_step": 8806, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03984261304140091, "train_loss_bc": 0.03984261304140091, "train_loss_llm": 0.0, "grad_norm": 0.04797513410449028, "global_step": 8807, "epoch": 244, "lr": 0.001} +{"train_loss": 0.037409912794828415, "train_loss_bc": 0.03740016371011734, "train_loss_llm": 0.0009748637676239014, "grad_norm": 0.05790001153945923, "global_step": 8808, "epoch": 244, "lr": 0.001} +{"train_loss": 0.04284457117319107, "train_loss_bc": 0.04284156486392021, "train_loss_llm": 0.0003007054328918457, "grad_norm": 0.02036936953663826, "global_step": 8809, "epoch": 244, "lr": 0.001} +{"train_loss": 0.038193438202142715, "train_loss_bc": 0.03817973658442497, "train_loss_llm": 0.0013703405857086182, "grad_norm": 0.0297031681984663, "global_step": 8810, "epoch": 244, "lr": 0.001} +{"train_loss": 0.041303616017103195, "train_loss_bc": 0.041303616017103195, "train_loss_llm": 0.0, "grad_norm": 0.03741173818707466, "global_step": 8811, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03617488592863083, "train_loss_bc": 0.03617488592863083, "train_loss_llm": 0.0, "grad_norm": 0.04063989967107773, "global_step": 8812, "epoch": 244, "lr": 0.001} +{"train_loss": 0.04190738871693611, "train_loss_bc": 0.04189155250787735, "train_loss_llm": 0.0015835464000701904, "grad_norm": 0.023579057306051254, "global_step": 8813, "epoch": 244, "lr": 0.001} +{"train_loss": 0.04019307345151901, "train_loss_bc": 0.04017512872815132, "train_loss_llm": 0.0017944872379302979, "grad_norm": 0.02922850474715233, "global_step": 8814, "epoch": 244, "lr": 0.001} +{"train_loss": 0.03603879734873772, "train_loss_bc": 0.03603506460785866, "train_loss_llm": 0.00037339329719543457, "grad_norm": 0.03853113204240799, "global_step": 8815, "epoch": 244, "lr": 0.001} +{"train_loss": 0.038883138447999954, "train_loss_bc": 0.03888027369976044, "train_loss_llm": 0.00028651952743530273, "grad_norm": 0.0459691621363163, "global_step": 8816, "epoch": 244, "lr": 0.001} +{"train_loss": 0.04335039108991623, "train_loss_bc": 0.04335039108991623, "train_loss_llm": 0.0, "grad_norm": 0.025153787806630135, "global_step": 8817, "epoch": 244, "lr": 0.001} +{"train_loss": 0.029539206996560097, "train_loss_bc": 0.029539206996560097, "train_loss_llm": 0.0, "grad_norm": 0.03158382698893547, "global_step": 8818, "epoch": 244, "lr": 0.001} +{"train_loss": 0.039185112187018, "train_loss_bc": 0.045530643314123154, "train_loss_llm": 0.004690110683441162, "grad_norm": 0.041765883564949036, "global_step": 8819, "epoch": 244, "lr": 0.001, "val_loss": 0.05901318043470383} +{"train_loss": 0.047609854489564896, "train_loss_bc": 0.047609854489564896, "train_loss_llm": 0.0, "grad_norm": 0.05301866680383682, "global_step": 8820, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03637378662824631, "train_loss_bc": 0.03637378662824631, "train_loss_llm": 0.0, "grad_norm": 0.018116051331162453, "global_step": 8821, "epoch": 245, "lr": 0.001} +{"train_loss": 0.040419772267341614, "train_loss_bc": 0.040419772267341614, "train_loss_llm": 0.0, "grad_norm": 0.028669679537415504, "global_step": 8822, "epoch": 245, "lr": 0.001} +{"train_loss": 0.040777839720249176, "train_loss_bc": 0.040777839720249176, "train_loss_llm": 0.0, "grad_norm": 0.04112815856933594, "global_step": 8823, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03668471425771713, "train_loss_bc": 0.03668471425771713, "train_loss_llm": 0.0, "grad_norm": 0.045665185898542404, "global_step": 8824, "epoch": 245, "lr": 0.001} +{"train_loss": 0.032228387892246246, "train_loss_bc": 0.032228387892246246, "train_loss_llm": 0.0, "grad_norm": 0.014948844909667969, "global_step": 8825, "epoch": 245, "lr": 0.001} +{"train_loss": 0.04349042847752571, "train_loss_bc": 0.04349042847752571, "train_loss_llm": 0.0, "grad_norm": 0.029286164790391922, "global_step": 8826, "epoch": 245, "lr": 0.001} +{"train_loss": 0.038390256464481354, "train_loss_bc": 0.038390256464481354, "train_loss_llm": 0.0, "grad_norm": 0.03652452677488327, "global_step": 8827, "epoch": 245, "lr": 0.001} +{"train_loss": 0.04163786396384239, "train_loss_bc": 0.04160938039422035, "train_loss_llm": 0.0028483569622039795, "grad_norm": 0.04654785990715027, "global_step": 8828, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03747310861945152, "train_loss_bc": 0.03746921569108963, "train_loss_llm": 0.0003891289234161377, "grad_norm": 0.0185984019190073, "global_step": 8829, "epoch": 245, "lr": 0.001} +{"train_loss": 0.04425593838095665, "train_loss_bc": 0.04425593838095665, "train_loss_llm": 0.0, "grad_norm": 0.031602758914232254, "global_step": 8830, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03905076906085014, "train_loss_bc": 0.03905076906085014, "train_loss_llm": 0.0, "grad_norm": 0.03940145671367645, "global_step": 8831, "epoch": 245, "lr": 0.001} +{"train_loss": 0.043390847742557526, "train_loss_bc": 0.043390847742557526, "train_loss_llm": 0.0, "grad_norm": 0.047970738261938095, "global_step": 8832, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03740780055522919, "train_loss_bc": 0.037398602813482285, "train_loss_llm": 0.0009198188781738281, "grad_norm": 0.021921880543231964, "global_step": 8833, "epoch": 245, "lr": 0.001} +{"train_loss": 0.04087289050221443, "train_loss_bc": 0.04087289050221443, "train_loss_llm": 0.0, "grad_norm": 0.030975546687841415, "global_step": 8834, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03515971824526787, "train_loss_bc": 0.03515971824526787, "train_loss_llm": 0.0, "grad_norm": 0.04305748641490936, "global_step": 8835, "epoch": 245, "lr": 0.001} +{"train_loss": 0.049325115978717804, "train_loss_bc": 0.04932078719139099, "train_loss_llm": 0.0004329979419708252, "grad_norm": 0.05066804587841034, "global_step": 8836, "epoch": 245, "lr": 0.001} +{"train_loss": 0.030314432457089424, "train_loss_bc": 0.030314328148961067, "train_loss_llm": 1.043081283569336e-05, "grad_norm": 0.017963407561182976, "global_step": 8837, "epoch": 245, "lr": 0.001} +{"train_loss": 0.029802003875374794, "train_loss_bc": 0.029802003875374794, "train_loss_llm": 0.0, "grad_norm": 0.029127083718776703, "global_step": 8838, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03577473387122154, "train_loss_bc": 0.03577473387122154, "train_loss_llm": 0.0, "grad_norm": 0.043323542922735214, "global_step": 8839, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03815283626317978, "train_loss_bc": 0.03815140202641487, "train_loss_llm": 0.00014337897300720215, "grad_norm": 0.05226203799247742, "global_step": 8840, "epoch": 245, "lr": 0.001} +{"train_loss": 0.04342961311340332, "train_loss_bc": 0.04342961311340332, "train_loss_llm": 0.0, "grad_norm": 0.02245274744927883, "global_step": 8841, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03495272621512413, "train_loss_bc": 0.034930191934108734, "train_loss_llm": 0.002253413200378418, "grad_norm": 0.03461332246661186, "global_step": 8842, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03459171578288078, "train_loss_bc": 0.03457694128155708, "train_loss_llm": 0.0014773905277252197, "grad_norm": 0.04975689575076103, "global_step": 8843, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03472152352333069, "train_loss_bc": 0.03472152352333069, "train_loss_llm": 0.0, "grad_norm": 0.05562875419855118, "global_step": 8844, "epoch": 245, "lr": 0.001} +{"train_loss": 0.039735663682222366, "train_loss_bc": 0.039735663682222366, "train_loss_llm": 0.0, "grad_norm": 0.014811797067523003, "global_step": 8845, "epoch": 245, "lr": 0.001} +{"train_loss": 0.0446966178715229, "train_loss_bc": 0.044695738703012466, "train_loss_llm": 8.7738037109375e-05, "grad_norm": 0.02160950005054474, "global_step": 8846, "epoch": 245, "lr": 0.001} +{"train_loss": 0.04235357046127319, "train_loss_bc": 0.04235357046127319, "train_loss_llm": 0.0, "grad_norm": 0.029679274186491966, "global_step": 8847, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03805459290742874, "train_loss_bc": 0.03805459290742874, "train_loss_llm": 0.0, "grad_norm": 0.036949776113033295, "global_step": 8848, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03761247918009758, "train_loss_bc": 0.03761247918009758, "train_loss_llm": 0.0, "grad_norm": 0.018742859363555908, "global_step": 8849, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03516927734017372, "train_loss_bc": 0.03515467047691345, "train_loss_llm": 0.0014607608318328857, "grad_norm": 0.027129242196679115, "global_step": 8850, "epoch": 245, "lr": 0.001} +{"train_loss": 0.04601474478840828, "train_loss_bc": 0.04597749933600426, "train_loss_llm": 0.0037244856357574463, "grad_norm": 0.03677019104361534, "global_step": 8851, "epoch": 245, "lr": 0.001} +{"train_loss": 0.04013732075691223, "train_loss_bc": 0.04013076052069664, "train_loss_llm": 0.0006560981273651123, "grad_norm": 0.045410942286252975, "global_step": 8852, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03846893459558487, "train_loss_bc": 0.03846893459558487, "train_loss_llm": 0.0, "grad_norm": 0.01858695223927498, "global_step": 8853, "epoch": 245, "lr": 0.001} +{"train_loss": 0.04074796289205551, "train_loss_bc": 0.04074796289205551, "train_loss_llm": 0.0, "grad_norm": 0.02458239160478115, "global_step": 8854, "epoch": 245, "lr": 0.001} +{"train_loss": 0.03919851779937744, "train_loss_bc": 0.04186679795384407, "train_loss_llm": 0.0, "grad_norm": 0.03177333250641823, "global_step": 8855, "epoch": 245, "lr": 0.001, "train/cumulative_reward": 119.02553367319665, "train/mean_score": 10.71286753462143, "train/success_rate": 0.2, "test/cumulative_reward": 26.749044080616013, "test/mean_score": 5.908074514257436, "test/success_rate": 0.14, "val_loss": 0.05964906886219978, "train_action_mse_error": 0.04299076646566391} +{"train_loss": 0.04535723105072975, "train_loss_bc": 0.04535111039876938, "train_loss_llm": 0.0006119310855865479, "grad_norm": 0.041125908493995667, "global_step": 8856, "epoch": 246, "lr": 0.001} +{"train_loss": 0.042898450046777725, "train_loss_bc": 0.042883507907390594, "train_loss_llm": 0.001494288444519043, "grad_norm": 0.024024048820137978, "global_step": 8857, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03958005830645561, "train_loss_bc": 0.03957062214612961, "train_loss_llm": 0.000943601131439209, "grad_norm": 0.02979719266295433, "global_step": 8858, "epoch": 246, "lr": 0.001} +{"train_loss": 0.04328388720750809, "train_loss_bc": 0.04328388720750809, "train_loss_llm": 0.0, "grad_norm": 0.04607022926211357, "global_step": 8859, "epoch": 246, "lr": 0.001} +{"train_loss": 0.04466971009969711, "train_loss_bc": 0.04466971009969711, "train_loss_llm": 0.0, "grad_norm": 0.059231121093034744, "global_step": 8860, "epoch": 246, "lr": 0.001} +{"train_loss": 0.031300533562898636, "train_loss_bc": 0.031300533562898636, "train_loss_llm": 0.0, "grad_norm": 0.018604492768645287, "global_step": 8861, "epoch": 246, "lr": 0.001} +{"train_loss": 0.037153393030166626, "train_loss_bc": 0.037153393030166626, "train_loss_llm": 0.0, "grad_norm": 0.027834244072437286, "global_step": 8862, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03462744876742363, "train_loss_bc": 0.03462744876742363, "train_loss_llm": 0.0, "grad_norm": 0.03543166443705559, "global_step": 8863, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03886349871754646, "train_loss_bc": 0.03886349871754646, "train_loss_llm": 0.0, "grad_norm": 0.04823237657546997, "global_step": 8864, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03412703797221184, "train_loss_bc": 0.03412703797221184, "train_loss_llm": 0.0, "grad_norm": 0.019715361297130585, "global_step": 8865, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03314940631389618, "train_loss_bc": 0.03314940631389618, "train_loss_llm": 0.0, "grad_norm": 0.0278631579130888, "global_step": 8866, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03999750316143036, "train_loss_bc": 0.03999750316143036, "train_loss_llm": 0.0, "grad_norm": 0.04152548313140869, "global_step": 8867, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03858419507741928, "train_loss_bc": 0.03858419507741928, "train_loss_llm": 0.0, "grad_norm": 0.04825393855571747, "global_step": 8868, "epoch": 246, "lr": 0.001} +{"train_loss": 0.051107678562402725, "train_loss_bc": 0.051107678562402725, "train_loss_llm": 0.0, "grad_norm": 0.01945432648062706, "global_step": 8869, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03753463551402092, "train_loss_bc": 0.03752441704273224, "train_loss_llm": 0.0010216832160949707, "grad_norm": 0.03021715022623539, "global_step": 8870, "epoch": 246, "lr": 0.001} +{"train_loss": 0.032785896211862564, "train_loss_bc": 0.032763149589300156, "train_loss_llm": 0.0022745728492736816, "grad_norm": 0.037004657089710236, "global_step": 8871, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03849637508392334, "train_loss_bc": 0.038466718047857285, "train_loss_llm": 0.002965778112411499, "grad_norm": 0.04459408298134804, "global_step": 8872, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03747621551156044, "train_loss_bc": 0.03747621551156044, "train_loss_llm": 0.0, "grad_norm": 0.018118157982826233, "global_step": 8873, "epoch": 246, "lr": 0.001} +{"train_loss": 0.04041241854429245, "train_loss_bc": 0.04041241854429245, "train_loss_llm": 0.0, "grad_norm": 0.03349357098340988, "global_step": 8874, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03453337773680687, "train_loss_bc": 0.034500766545534134, "train_loss_llm": 0.0032611191272735596, "grad_norm": 0.042406149208545685, "global_step": 8875, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03484977409243584, "train_loss_bc": 0.034836869686841965, "train_loss_llm": 0.0012902915477752686, "grad_norm": 0.05435096099972725, "global_step": 8876, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03733406960964203, "train_loss_bc": 0.037333302199840546, "train_loss_llm": 7.671117782592773e-05, "grad_norm": 0.021340612322092056, "global_step": 8877, "epoch": 246, "lr": 0.001} +{"train_loss": 0.04258299246430397, "train_loss_bc": 0.04258299246430397, "train_loss_llm": 0.0, "grad_norm": 0.031468313187360764, "global_step": 8878, "epoch": 246, "lr": 0.001} +{"train_loss": 0.034484781324863434, "train_loss_bc": 0.03447695076465607, "train_loss_llm": 0.0007829070091247559, "grad_norm": 0.04153946042060852, "global_step": 8879, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03923269733786583, "train_loss_bc": 0.03923269733786583, "train_loss_llm": 0.0, "grad_norm": 0.051319997757673264, "global_step": 8880, "epoch": 246, "lr": 0.001} +{"train_loss": 0.04285665974020958, "train_loss_bc": 0.04285665974020958, "train_loss_llm": 0.0, "grad_norm": 0.017836786806583405, "global_step": 8881, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03967498242855072, "train_loss_bc": 0.03967498242855072, "train_loss_llm": 0.0, "grad_norm": 0.025950873270630836, "global_step": 8882, "epoch": 246, "lr": 0.001} +{"train_loss": 0.04024813324213028, "train_loss_bc": 0.04024813324213028, "train_loss_llm": 0.0, "grad_norm": 0.04337180033326149, "global_step": 8883, "epoch": 246, "lr": 0.001} +{"train_loss": 0.038565561175346375, "train_loss_bc": 0.03856207802891731, "train_loss_llm": 0.0003483891487121582, "grad_norm": 0.04920397698879242, "global_step": 8884, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03667528182268143, "train_loss_bc": 0.03667528182268143, "train_loss_llm": 0.0, "grad_norm": 0.024775998666882515, "global_step": 8885, "epoch": 246, "lr": 0.001} +{"train_loss": 0.03876679763197899, "train_loss_bc": 0.03876679763197899, "train_loss_llm": 0.0, "grad_norm": 0.03552908077836037, "global_step": 8886, "epoch": 246, "lr": 0.001} +{"train_loss": 0.04249567911028862, "train_loss_bc": 0.042487647384405136, "train_loss_llm": 0.000803142786026001, "grad_norm": 0.04054075852036476, "global_step": 8887, "epoch": 246, "lr": 0.001} +{"train_loss": 0.041208475828170776, "train_loss_bc": 0.041208475828170776, "train_loss_llm": 0.0, "grad_norm": 0.0529300794005394, "global_step": 8888, "epoch": 246, "lr": 0.001} +{"train_loss": 0.04100813344120979, "train_loss_bc": 0.04100813344120979, "train_loss_llm": 0.0, "grad_norm": 0.01981675624847412, "global_step": 8889, "epoch": 246, "lr": 0.001} +{"train_loss": 0.04163374751806259, "train_loss_bc": 0.04163374751806259, "train_loss_llm": 0.0, "grad_norm": 0.03036852739751339, "global_step": 8890, "epoch": 246, "lr": 0.001} +{"train_loss": 0.039229785195655294, "train_loss_bc": 0.04471312835812569, "train_loss_llm": 0.0002422630786895752, "grad_norm": 0.04216421768069267, "global_step": 8891, "epoch": 246, "lr": 0.001, "val_loss": 0.059324510395526886} +{"train_loss": 0.03178667277097702, "train_loss_bc": 0.03178667277097702, "train_loss_llm": 0.0, "grad_norm": 0.05640346556901932, "global_step": 8892, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04390542954206467, "train_loss_bc": 0.04390079155564308, "train_loss_llm": 0.00046393275260925293, "grad_norm": 0.020082345232367516, "global_step": 8893, "epoch": 247, "lr": 0.001} +{"train_loss": 0.044585444033145905, "train_loss_bc": 0.044585444033145905, "train_loss_llm": 0.0, "grad_norm": 0.03290944918990135, "global_step": 8894, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04163222759962082, "train_loss_bc": 0.04160178825259209, "train_loss_llm": 0.0030440688133239746, "grad_norm": 0.04327348992228508, "global_step": 8895, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04363522306084633, "train_loss_bc": 0.043630823493003845, "train_loss_llm": 0.000440061092376709, "grad_norm": 0.05508748069405556, "global_step": 8896, "epoch": 247, "lr": 0.001} +{"train_loss": 0.039659954607486725, "train_loss_bc": 0.039659954607486725, "train_loss_llm": 0.0, "grad_norm": 0.02061617188155651, "global_step": 8897, "epoch": 247, "lr": 0.001} +{"train_loss": 0.039582427591085434, "train_loss_bc": 0.039582427591085434, "train_loss_llm": 0.0, "grad_norm": 0.02982429228723049, "global_step": 8898, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03498247638344765, "train_loss_bc": 0.03498247638344765, "train_loss_llm": 0.0, "grad_norm": 0.040531814098358154, "global_step": 8899, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03990907222032547, "train_loss_bc": 0.03989310935139656, "train_loss_llm": 0.0015963315963745117, "grad_norm": 0.050009775906801224, "global_step": 8900, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03538687899708748, "train_loss_bc": 0.03538687899708748, "train_loss_llm": 0.0, "grad_norm": 0.021190159022808075, "global_step": 8901, "epoch": 247, "lr": 0.001} +{"train_loss": 0.047700416296720505, "train_loss_bc": 0.047697003930807114, "train_loss_llm": 0.00034108757972717285, "grad_norm": 0.03765013441443443, "global_step": 8902, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03218147158622742, "train_loss_bc": 0.03218147158622742, "train_loss_llm": 0.0, "grad_norm": 0.04695165902376175, "global_step": 8903, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03999707102775574, "train_loss_bc": 0.03999707102775574, "train_loss_llm": 0.0, "grad_norm": 0.05207040160894394, "global_step": 8904, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04042195528745651, "train_loss_bc": 0.04042195528745651, "train_loss_llm": 0.0, "grad_norm": 0.022081580013036728, "global_step": 8905, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03795066848397255, "train_loss_bc": 0.03795066848397255, "train_loss_llm": 0.0, "grad_norm": 0.031405605375766754, "global_step": 8906, "epoch": 247, "lr": 0.001} +{"train_loss": 0.039636678993701935, "train_loss_bc": 0.039623573422431946, "train_loss_llm": 0.001310586929321289, "grad_norm": 0.04362988471984863, "global_step": 8907, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04651001840829849, "train_loss_bc": 0.04651001840829849, "train_loss_llm": 0.0, "grad_norm": 0.05661256983876228, "global_step": 8908, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04057369381189346, "train_loss_bc": 0.04057369381189346, "train_loss_llm": 0.0, "grad_norm": 0.021329645067453384, "global_step": 8909, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03280831500887871, "train_loss_bc": 0.03280463069677353, "train_loss_llm": 0.00036838650703430176, "grad_norm": 0.03189370036125183, "global_step": 8910, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03526204451918602, "train_loss_bc": 0.03526204451918602, "train_loss_llm": 0.0, "grad_norm": 0.044948846101760864, "global_step": 8911, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04102807119488716, "train_loss_bc": 0.04102807119488716, "train_loss_llm": 0.0, "grad_norm": 0.050972335040569305, "global_step": 8912, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03172781318426132, "train_loss_bc": 0.03172781318426132, "train_loss_llm": 0.0, "grad_norm": 0.022251661866903305, "global_step": 8913, "epoch": 247, "lr": 0.001} +{"train_loss": 0.039552606642246246, "train_loss_bc": 0.03955135494470596, "train_loss_llm": 0.00012502074241638184, "grad_norm": 0.04425693675875664, "global_step": 8914, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04048113152384758, "train_loss_bc": 0.04048113152384758, "train_loss_llm": 0.0, "grad_norm": 0.06383707374334335, "global_step": 8915, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03643210604786873, "train_loss_bc": 0.03642962500452995, "train_loss_llm": 0.0002479851245880127, "grad_norm": 0.07853240519762039, "global_step": 8916, "epoch": 247, "lr": 0.001} +{"train_loss": 0.0405791774392128, "train_loss_bc": 0.040568627417087555, "train_loss_llm": 0.0010548532009124756, "grad_norm": 0.026758091524243355, "global_step": 8917, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03949737921357155, "train_loss_bc": 0.03948045149445534, "train_loss_llm": 0.0016926229000091553, "grad_norm": 0.039933834224939346, "global_step": 8918, "epoch": 247, "lr": 0.001} +{"train_loss": 0.039280276745557785, "train_loss_bc": 0.039280276745557785, "train_loss_llm": 0.0, "grad_norm": 0.04803188145160675, "global_step": 8919, "epoch": 247, "lr": 0.001} +{"train_loss": 0.041794732213020325, "train_loss_bc": 0.041794732213020325, "train_loss_llm": 0.0, "grad_norm": 0.059468790888786316, "global_step": 8920, "epoch": 247, "lr": 0.001} +{"train_loss": 0.040603090077638626, "train_loss_bc": 0.040603090077638626, "train_loss_llm": 0.0, "grad_norm": 0.02039046585559845, "global_step": 8921, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04050179570913315, "train_loss_bc": 0.040501680225133896, "train_loss_llm": 1.1414289474487305e-05, "grad_norm": 0.027932271361351013, "global_step": 8922, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04272449016571045, "train_loss_bc": 0.04272449016571045, "train_loss_llm": 0.0, "grad_norm": 0.04350564628839493, "global_step": 8923, "epoch": 247, "lr": 0.001} +{"train_loss": 0.04539153724908829, "train_loss_bc": 0.04536200314760208, "train_loss_llm": 0.002953261137008667, "grad_norm": 0.05640446022152901, "global_step": 8924, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03813541680574417, "train_loss_bc": 0.03813541680574417, "train_loss_llm": 0.0, "grad_norm": 0.018766021355986595, "global_step": 8925, "epoch": 247, "lr": 0.001} +{"train_loss": 0.029026899486780167, "train_loss_bc": 0.029021741822361946, "train_loss_llm": 0.0005157589912414551, "grad_norm": 0.02930179052054882, "global_step": 8926, "epoch": 247, "lr": 0.001} +{"train_loss": 0.03920145384553406, "train_loss_bc": 0.03638438507914543, "train_loss_llm": 0.0003288388252258301, "grad_norm": 0.036068033427000046, "global_step": 8927, "epoch": 247, "lr": 0.001, "val_loss": 0.05890766531229019} +{"train_loss": 0.044674646109342575, "train_loss_bc": 0.044674646109342575, "train_loss_llm": 0.0, "grad_norm": 0.03913986310362816, "global_step": 8928, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04034673422574997, "train_loss_bc": 0.04033615067601204, "train_loss_llm": 0.0010582506656646729, "grad_norm": 0.01583140529692173, "global_step": 8929, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03047681786119938, "train_loss_bc": 0.03047681786119938, "train_loss_llm": 0.0, "grad_norm": 0.02382393181324005, "global_step": 8930, "epoch": 248, "lr": 0.001} +{"train_loss": 0.042470164597034454, "train_loss_bc": 0.042460501194000244, "train_loss_llm": 0.0009662806987762451, "grad_norm": 0.03101445734500885, "global_step": 8931, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03838515281677246, "train_loss_bc": 0.03838149085640907, "train_loss_llm": 0.0003660917282104492, "grad_norm": 0.033640723675489426, "global_step": 8932, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04296514391899109, "train_loss_bc": 0.04295862466096878, "train_loss_llm": 0.0006518959999084473, "grad_norm": 0.02097068913280964, "global_step": 8933, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03388112038373947, "train_loss_bc": 0.03388112038373947, "train_loss_llm": 0.0, "grad_norm": 0.030557449907064438, "global_step": 8934, "epoch": 248, "lr": 0.001} +{"train_loss": 0.040077321231365204, "train_loss_bc": 0.040077321231365204, "train_loss_llm": 0.0, "grad_norm": 0.04157271981239319, "global_step": 8935, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04368142411112785, "train_loss_bc": 0.043680742383003235, "train_loss_llm": 6.809830665588379e-05, "grad_norm": 0.05366886779665947, "global_step": 8936, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04147019609808922, "train_loss_bc": 0.04147019609808922, "train_loss_llm": 0.0, "grad_norm": 0.020207885652780533, "global_step": 8937, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03047981858253479, "train_loss_bc": 0.03047981858253479, "train_loss_llm": 0.0, "grad_norm": 0.04045712575316429, "global_step": 8938, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04737575352191925, "train_loss_bc": 0.04737575352191925, "train_loss_llm": 0.0, "grad_norm": 0.056346725672483444, "global_step": 8939, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03857181966304779, "train_loss_bc": 0.03857181966304779, "train_loss_llm": 0.0, "grad_norm": 0.063094861805439, "global_step": 8940, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03583985194563866, "train_loss_bc": 0.03583985194563866, "train_loss_llm": 0.0, "grad_norm": 0.01927904598414898, "global_step": 8941, "epoch": 248, "lr": 0.001} +{"train_loss": 0.035973675549030304, "train_loss_bc": 0.035973675549030304, "train_loss_llm": 0.0, "grad_norm": 0.03169029951095581, "global_step": 8942, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04092588275671005, "train_loss_bc": 0.04092588275671005, "train_loss_llm": 0.0, "grad_norm": 0.052268821746110916, "global_step": 8943, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03740038722753525, "train_loss_bc": 0.03738408908247948, "train_loss_llm": 0.0016299784183502197, "grad_norm": 0.06999941915273666, "global_step": 8944, "epoch": 248, "lr": 0.001} +{"train_loss": 0.028021948412060738, "train_loss_bc": 0.028014851734042168, "train_loss_llm": 0.0007097125053405762, "grad_norm": 0.017563076689839363, "global_step": 8945, "epoch": 248, "lr": 0.001} +{"train_loss": 0.039898715913295746, "train_loss_bc": 0.039898715913295746, "train_loss_llm": 0.0, "grad_norm": 0.03044721856713295, "global_step": 8946, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04609161242842674, "train_loss_bc": 0.046079304069280624, "train_loss_llm": 0.0012309253215789795, "grad_norm": 0.04456476494669914, "global_step": 8947, "epoch": 248, "lr": 0.001} +{"train_loss": 0.040327075868844986, "train_loss_bc": 0.040327075868844986, "train_loss_llm": 0.0, "grad_norm": 0.04990632086992264, "global_step": 8948, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03756062686443329, "train_loss_bc": 0.03755834326148033, "train_loss_llm": 0.00022837519645690918, "grad_norm": 0.01908193714916706, "global_step": 8949, "epoch": 248, "lr": 0.001} +{"train_loss": 0.037300895899534225, "train_loss_bc": 0.037300895899534225, "train_loss_llm": 0.0, "grad_norm": 0.028685761615633965, "global_step": 8950, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03607284277677536, "train_loss_bc": 0.03607284277677536, "train_loss_llm": 0.0, "grad_norm": 0.03852414712309837, "global_step": 8951, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03884395584464073, "train_loss_bc": 0.03884395584464073, "train_loss_llm": 0.0, "grad_norm": 0.04675978794693947, "global_step": 8952, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04118890315294266, "train_loss_bc": 0.04118890315294266, "train_loss_llm": 0.0, "grad_norm": 0.01907327026128769, "global_step": 8953, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04368411749601364, "train_loss_bc": 0.04368411749601364, "train_loss_llm": 0.0, "grad_norm": 0.030037371441721916, "global_step": 8954, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03843379765748978, "train_loss_bc": 0.038427919149398804, "train_loss_llm": 0.0005877315998077393, "grad_norm": 0.04487626999616623, "global_step": 8955, "epoch": 248, "lr": 0.001} +{"train_loss": 0.0462544709444046, "train_loss_bc": 0.0462544709444046, "train_loss_llm": 0.0, "grad_norm": 0.0550527460873127, "global_step": 8956, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04254831746220589, "train_loss_bc": 0.04254831746220589, "train_loss_llm": 0.0, "grad_norm": 0.020240604877471924, "global_step": 8957, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03734219819307327, "train_loss_bc": 0.03732701390981674, "train_loss_llm": 0.0015183985233306885, "grad_norm": 0.033543143421411514, "global_step": 8958, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03701965510845184, "train_loss_bc": 0.03701888769865036, "train_loss_llm": 7.659196853637695e-05, "grad_norm": 0.04681505262851715, "global_step": 8959, "epoch": 248, "lr": 0.001} +{"train_loss": 0.04791579023003578, "train_loss_bc": 0.04790692403912544, "train_loss_llm": 0.0008867383003234863, "grad_norm": 0.06657224148511887, "global_step": 8960, "epoch": 248, "lr": 0.001} +{"train_loss": 0.035429153591394424, "train_loss_bc": 0.035429153591394424, "train_loss_llm": 0.0, "grad_norm": 0.020078537985682487, "global_step": 8961, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03763397037982941, "train_loss_bc": 0.03762734308838844, "train_loss_llm": 0.0006628930568695068, "grad_norm": 0.04077233001589775, "global_step": 8962, "epoch": 248, "lr": 0.001} +{"train_loss": 0.03916812129318714, "train_loss_bc": 0.03348840773105621, "train_loss_llm": 0.0, "grad_norm": 0.05452360585331917, "global_step": 8963, "epoch": 248, "lr": 0.001, "val_loss": 0.05957198888063431} +{"train_loss": 0.03130974620580673, "train_loss_bc": 0.03130974620580673, "train_loss_llm": 0.0, "grad_norm": 0.06509999930858612, "global_step": 8964, "epoch": 249, "lr": 0.001} +{"train_loss": 0.04659652337431908, "train_loss_bc": 0.04659652337431908, "train_loss_llm": 0.0, "grad_norm": 0.02577420324087143, "global_step": 8965, "epoch": 249, "lr": 0.001} +{"train_loss": 0.0344797745347023, "train_loss_bc": 0.0344797745347023, "train_loss_llm": 0.0, "grad_norm": 0.03504867106676102, "global_step": 8966, "epoch": 249, "lr": 0.001} +{"train_loss": 0.039978642016649246, "train_loss_bc": 0.03997235745191574, "train_loss_llm": 0.0006284415721893311, "grad_norm": 0.046627290546894073, "global_step": 8967, "epoch": 249, "lr": 0.001} +{"train_loss": 0.04004008322954178, "train_loss_bc": 0.040025122463703156, "train_loss_llm": 0.0014959871768951416, "grad_norm": 0.05812178924679756, "global_step": 8968, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03794538229703903, "train_loss_bc": 0.03794538229703903, "train_loss_llm": 0.0, "grad_norm": 0.02028649114072323, "global_step": 8969, "epoch": 249, "lr": 0.001} +{"train_loss": 0.033874813467264175, "train_loss_bc": 0.03385855630040169, "train_loss_llm": 0.00162583589553833, "grad_norm": 0.030054083094000816, "global_step": 8970, "epoch": 249, "lr": 0.001} +{"train_loss": 0.037979159504175186, "train_loss_bc": 0.037979159504175186, "train_loss_llm": 0.0, "grad_norm": 0.03602849319577217, "global_step": 8971, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03287958353757858, "train_loss_bc": 0.03287958353757858, "train_loss_llm": 0.0, "grad_norm": 0.042557500302791595, "global_step": 8972, "epoch": 249, "lr": 0.001} +{"train_loss": 0.036836449056863785, "train_loss_bc": 0.03681729733943939, "train_loss_llm": 0.001915067434310913, "grad_norm": 0.01870538294315338, "global_step": 8973, "epoch": 249, "lr": 0.001} +{"train_loss": 0.038002122193574905, "train_loss_bc": 0.038002122193574905, "train_loss_llm": 0.0, "grad_norm": 0.028721703216433525, "global_step": 8974, "epoch": 249, "lr": 0.001} +{"train_loss": 0.036500483751297, "train_loss_bc": 0.036500483751297, "train_loss_llm": 0.0, "grad_norm": 0.039040692150592804, "global_step": 8975, "epoch": 249, "lr": 0.001} +{"train_loss": 0.04246916621923447, "train_loss_bc": 0.04246916621923447, "train_loss_llm": 0.0, "grad_norm": 0.042441319674253464, "global_step": 8976, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03811691701412201, "train_loss_bc": 0.03811199963092804, "train_loss_llm": 0.0004916489124298096, "grad_norm": 0.01652636006474495, "global_step": 8977, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03716432303190231, "train_loss_bc": 0.03715202957391739, "train_loss_llm": 0.0012294352054595947, "grad_norm": 0.0324806272983551, "global_step": 8978, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03693784773349762, "train_loss_bc": 0.03693784773349762, "train_loss_llm": 0.0, "grad_norm": 0.04247184097766876, "global_step": 8979, "epoch": 249, "lr": 0.001} +{"train_loss": 0.038809336721897125, "train_loss_bc": 0.038809336721897125, "train_loss_llm": 0.0, "grad_norm": 0.055818796157836914, "global_step": 8980, "epoch": 249, "lr": 0.001} +{"train_loss": 0.04218437150120735, "train_loss_bc": 0.04218437150120735, "train_loss_llm": 0.0, "grad_norm": 0.018611986190080643, "global_step": 8981, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03518351912498474, "train_loss_bc": 0.03518351912498474, "train_loss_llm": 0.0, "grad_norm": 0.02724667824804783, "global_step": 8982, "epoch": 249, "lr": 0.001} +{"train_loss": 0.04549507424235344, "train_loss_bc": 0.04549507424235344, "train_loss_llm": 0.0, "grad_norm": 0.045874446630477905, "global_step": 8983, "epoch": 249, "lr": 0.001} +{"train_loss": 0.04426230490207672, "train_loss_bc": 0.04426230490207672, "train_loss_llm": 0.0, "grad_norm": 0.060079023241996765, "global_step": 8984, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03352101147174835, "train_loss_bc": 0.03351040557026863, "train_loss_llm": 0.0010606944561004639, "grad_norm": 0.019774314016103745, "global_step": 8985, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03678686171770096, "train_loss_bc": 0.03678686171770096, "train_loss_llm": 0.0, "grad_norm": 0.027335014194250107, "global_step": 8986, "epoch": 249, "lr": 0.001} +{"train_loss": 0.0442802757024765, "train_loss_bc": 0.0442802757024765, "train_loss_llm": 0.0, "grad_norm": 0.04702761396765709, "global_step": 8987, "epoch": 249, "lr": 0.001} +{"train_loss": 0.045332442969083786, "train_loss_bc": 0.045332442969083786, "train_loss_llm": 0.0, "grad_norm": 0.05736057460308075, "global_step": 8988, "epoch": 249, "lr": 0.001} +{"train_loss": 0.04090830683708191, "train_loss_bc": 0.04090830683708191, "train_loss_llm": 0.0, "grad_norm": 0.022207466885447502, "global_step": 8989, "epoch": 249, "lr": 0.001} +{"train_loss": 0.04025696963071823, "train_loss_bc": 0.04025696963071823, "train_loss_llm": 0.0, "grad_norm": 0.025747086852788925, "global_step": 8990, "epoch": 249, "lr": 0.001} +{"train_loss": 0.036788150668144226, "train_loss_bc": 0.03676842898130417, "train_loss_llm": 0.0019721388816833496, "grad_norm": 0.03633604943752289, "global_step": 8991, "epoch": 249, "lr": 0.001} +{"train_loss": 0.043317653238773346, "train_loss_bc": 0.043273571878671646, "train_loss_llm": 0.00440824031829834, "grad_norm": 0.046491190791130066, "global_step": 8992, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03863149508833885, "train_loss_bc": 0.03863149508833885, "train_loss_llm": 0.0, "grad_norm": 0.020915716886520386, "global_step": 8993, "epoch": 249, "lr": 0.001} +{"train_loss": 0.04102136194705963, "train_loss_bc": 0.04102136194705963, "train_loss_llm": 0.0, "grad_norm": 0.03374708071351051, "global_step": 8994, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03274029120802879, "train_loss_bc": 0.03274029120802879, "train_loss_llm": 0.0, "grad_norm": 0.04732068255543709, "global_step": 8995, "epoch": 249, "lr": 0.001} +{"train_loss": 0.0434613972902298, "train_loss_bc": 0.043436210602521896, "train_loss_llm": 0.0025185346603393555, "grad_norm": 0.060833100229501724, "global_step": 8996, "epoch": 249, "lr": 0.001} +{"train_loss": 0.0463884063065052, "train_loss_bc": 0.0463884063065052, "train_loss_llm": 0.0, "grad_norm": 0.022984815761446953, "global_step": 8997, "epoch": 249, "lr": 0.001} +{"train_loss": 0.04048401489853859, "train_loss_bc": 0.04048241674900055, "train_loss_llm": 0.00015977025032043457, "grad_norm": 0.03127864748239517, "global_step": 8998, "epoch": 249, "lr": 0.001} +{"train_loss": 0.03921077317661709, "train_loss_bc": 0.04062357172369957, "train_loss_llm": 0.0, "grad_norm": 0.0473625473678112, "global_step": 8999, "epoch": 249, "lr": 0.001, "val_loss": 0.05935797095298767} +{"train_loss": 0.03671608865261078, "train_loss_bc": 0.036697350442409515, "train_loss_llm": 0.001873999834060669, "grad_norm": 0.06075941026210785, "global_step": 9000, "epoch": 250, "lr": 0.001} +{"train_loss": 0.039477501064538956, "train_loss_bc": 0.039477501064538956, "train_loss_llm": 0.0, "grad_norm": 0.022664621472358704, "global_step": 9001, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04026508703827858, "train_loss_bc": 0.04026508703827858, "train_loss_llm": 0.0, "grad_norm": 0.037536125630140305, "global_step": 9002, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04097496718168259, "train_loss_bc": 0.04096893593668938, "train_loss_llm": 0.000603020191192627, "grad_norm": 0.038727860897779465, "global_step": 9003, "epoch": 250, "lr": 0.001} +{"train_loss": 0.040604814887046814, "train_loss_bc": 0.040604814887046814, "train_loss_llm": 0.0, "grad_norm": 0.040945686399936676, "global_step": 9004, "epoch": 250, "lr": 0.001} +{"train_loss": 0.03385504335165024, "train_loss_bc": 0.03385504335165024, "train_loss_llm": 0.0, "grad_norm": 0.0159572996199131, "global_step": 9005, "epoch": 250, "lr": 0.001} +{"train_loss": 0.038137033581733704, "train_loss_bc": 0.038137033581733704, "train_loss_llm": 0.0, "grad_norm": 0.020164143294095993, "global_step": 9006, "epoch": 250, "lr": 0.001} +{"train_loss": 0.0412435345351696, "train_loss_bc": 0.0412435345351696, "train_loss_llm": 0.0, "grad_norm": 0.027422824874520302, "global_step": 9007, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04310283809900284, "train_loss_bc": 0.04310283809900284, "train_loss_llm": 0.0, "grad_norm": 0.03782263025641441, "global_step": 9008, "epoch": 250, "lr": 0.001} +{"train_loss": 0.043152257800102234, "train_loss_bc": 0.04312833026051521, "train_loss_llm": 0.002392798662185669, "grad_norm": 0.022291012108325958, "global_step": 9009, "epoch": 250, "lr": 0.001} +{"train_loss": 0.031139515340328217, "train_loss_bc": 0.031139515340328217, "train_loss_llm": 0.0, "grad_norm": 0.025343379005789757, "global_step": 9010, "epoch": 250, "lr": 0.001} +{"train_loss": 0.037802889943122864, "train_loss_bc": 0.03778675198554993, "train_loss_llm": 0.0016139745712280273, "grad_norm": 0.032926902174949646, "global_step": 9011, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04118215665221214, "train_loss_bc": 0.04117861017584801, "train_loss_llm": 0.0003548264503479004, "grad_norm": 0.04150465503334999, "global_step": 9012, "epoch": 250, "lr": 0.001} +{"train_loss": 0.038585808128118515, "train_loss_bc": 0.038585808128118515, "train_loss_llm": 0.0, "grad_norm": 0.0195549875497818, "global_step": 9013, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04217705503106117, "train_loss_bc": 0.04215773940086365, "train_loss_llm": 0.001931607723236084, "grad_norm": 0.03299766033887863, "global_step": 9014, "epoch": 250, "lr": 0.001} +{"train_loss": 0.03768157958984375, "train_loss_bc": 0.037675462663173676, "train_loss_llm": 0.0006116926670074463, "grad_norm": 0.045933909714221954, "global_step": 9015, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04305373877286911, "train_loss_bc": 0.04302959516644478, "train_loss_llm": 0.002414524555206299, "grad_norm": 0.062185924500226974, "global_step": 9016, "epoch": 250, "lr": 0.001} +{"train_loss": 0.03527327999472618, "train_loss_bc": 0.03527327999472618, "train_loss_llm": 0.0, "grad_norm": 0.02322138287127018, "global_step": 9017, "epoch": 250, "lr": 0.001} +{"train_loss": 0.037840161472558975, "train_loss_bc": 0.03783776983618736, "train_loss_llm": 0.00023925304412841797, "grad_norm": 0.03532850369811058, "global_step": 9018, "epoch": 250, "lr": 0.001} +{"train_loss": 0.041675642132759094, "train_loss_bc": 0.041675642132759094, "train_loss_llm": 0.0, "grad_norm": 0.053969915956258774, "global_step": 9019, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04273511841893196, "train_loss_bc": 0.04273511841893196, "train_loss_llm": 0.0, "grad_norm": 0.07047469913959503, "global_step": 9020, "epoch": 250, "lr": 0.001} +{"train_loss": 0.03512994572520256, "train_loss_bc": 0.03512994572520256, "train_loss_llm": 0.0, "grad_norm": 0.031045738607645035, "global_step": 9021, "epoch": 250, "lr": 0.001} +{"train_loss": 0.041895247995853424, "train_loss_bc": 0.041895247995853424, "train_loss_llm": 0.0, "grad_norm": 0.041690893471241, "global_step": 9022, "epoch": 250, "lr": 0.001} +{"train_loss": 0.03854217752814293, "train_loss_bc": 0.03854217752814293, "train_loss_llm": 0.0, "grad_norm": 0.04963892698287964, "global_step": 9023, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04263525828719139, "train_loss_bc": 0.04263510927557945, "train_loss_llm": 1.4930963516235352e-05, "grad_norm": 0.06466805934906006, "global_step": 9024, "epoch": 250, "lr": 0.001} +{"train_loss": 0.0406964085996151, "train_loss_bc": 0.0406964085996151, "train_loss_llm": 0.0, "grad_norm": 0.0187001321464777, "global_step": 9025, "epoch": 250, "lr": 0.001} +{"train_loss": 0.034502193331718445, "train_loss_bc": 0.034493573009967804, "train_loss_llm": 0.0008622109889984131, "grad_norm": 0.025182723999023438, "global_step": 9026, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04373713582754135, "train_loss_bc": 0.04373713582754135, "train_loss_llm": 0.0, "grad_norm": 0.03948508948087692, "global_step": 9027, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04078902676701546, "train_loss_bc": 0.04078902676701546, "train_loss_llm": 0.0, "grad_norm": 0.046942662447690964, "global_step": 9028, "epoch": 250, "lr": 0.001} +{"train_loss": 0.04198218882083893, "train_loss_bc": 0.04195759817957878, "train_loss_llm": 0.0024589598178863525, "grad_norm": 0.0202637892216444, "global_step": 9029, "epoch": 250, "lr": 0.001} +{"train_loss": 0.0408148430287838, "train_loss_bc": 0.0408148430287838, "train_loss_llm": 0.0, "grad_norm": 0.027147067710757256, "global_step": 9030, "epoch": 250, "lr": 0.001} +{"train_loss": 0.03503194823861122, "train_loss_bc": 0.03503194823861122, "train_loss_llm": 0.0, "grad_norm": 0.03460079804062843, "global_step": 9031, "epoch": 250, "lr": 0.001} +{"train_loss": 0.030024854466319084, "train_loss_bc": 0.030024854466319084, "train_loss_llm": 0.0, "grad_norm": 0.04239269345998764, "global_step": 9032, "epoch": 250, "lr": 0.001} +{"train_loss": 0.03465424105525017, "train_loss_bc": 0.034647103399038315, "train_loss_llm": 0.0007139444351196289, "grad_norm": 0.017534075304865837, "global_step": 9033, "epoch": 250, "lr": 0.001} +{"train_loss": 0.042981602251529694, "train_loss_bc": 0.042971424758434296, "train_loss_llm": 0.001017838716506958, "grad_norm": 0.029345154762268066, "global_step": 9034, "epoch": 250, "lr": 0.001} +{"train_loss": 0.03913331399154332, "train_loss_bc": 0.03870612010359764, "train_loss_llm": 0.0, "grad_norm": 0.03926673159003258, "global_step": 9035, "epoch": 250, "lr": 0.001, "train/cumulative_reward": 48.16726264977267, "train/mean_score": 10.804775141737831, "train/success_rate": 0.3, "test/cumulative_reward": 23.19284290133586, "test/mean_score": 3.2326432496334525, "test/success_rate": 0.06, "val_loss": 0.05960003286600113, "train_action_mse_error": 0.042803648859262466} +{"train_loss": 0.0450219064950943, "train_loss_bc": 0.045019835233688354, "train_loss_llm": 0.00020697712898254395, "grad_norm": 0.04998737946152687, "global_step": 9036, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03952130675315857, "train_loss_bc": 0.03952130675315857, "train_loss_llm": 0.0, "grad_norm": 0.020481472834944725, "global_step": 9037, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03907380253076553, "train_loss_bc": 0.03907380253076553, "train_loss_llm": 0.0, "grad_norm": 0.032585080713033676, "global_step": 9038, "epoch": 251, "lr": 0.001} +{"train_loss": 0.034976594150066376, "train_loss_bc": 0.034976594150066376, "train_loss_llm": 0.0, "grad_norm": 0.04580414667725563, "global_step": 9039, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03993070870637894, "train_loss_bc": 0.03993070870637894, "train_loss_llm": 0.0, "grad_norm": 0.054815806448459625, "global_step": 9040, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03787430748343468, "train_loss_bc": 0.03787430748343468, "train_loss_llm": 0.0, "grad_norm": 0.023826204240322113, "global_step": 9041, "epoch": 251, "lr": 0.001} +{"train_loss": 0.04349305480718613, "train_loss_bc": 0.043459467589855194, "train_loss_llm": 0.0033588707447052, "grad_norm": 0.045620232820510864, "global_step": 9042, "epoch": 251, "lr": 0.001} +{"train_loss": 0.043679166585206985, "train_loss_bc": 0.043679166585206985, "train_loss_llm": 0.0, "grad_norm": 0.04991329833865166, "global_step": 9043, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03687543421983719, "train_loss_bc": 0.036856431514024734, "train_loss_llm": 0.0019001960754394531, "grad_norm": 0.060062654316425323, "global_step": 9044, "epoch": 251, "lr": 0.001} +{"train_loss": 0.040447212755680084, "train_loss_bc": 0.040447212755680084, "train_loss_llm": 0.0, "grad_norm": 0.023314066231250763, "global_step": 9045, "epoch": 251, "lr": 0.001} +{"train_loss": 0.034735556691884995, "train_loss_bc": 0.034735556691884995, "train_loss_llm": 0.0, "grad_norm": 0.031119097024202347, "global_step": 9046, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03503528609871864, "train_loss_bc": 0.03503528609871864, "train_loss_llm": 0.0, "grad_norm": 0.04410184547305107, "global_step": 9047, "epoch": 251, "lr": 0.001} +{"train_loss": 0.0454595685005188, "train_loss_bc": 0.0454595685005188, "train_loss_llm": 0.0, "grad_norm": 0.05486837029457092, "global_step": 9048, "epoch": 251, "lr": 0.001} +{"train_loss": 0.04399175941944122, "train_loss_bc": 0.04399175941944122, "train_loss_llm": 0.0, "grad_norm": 0.024987246841192245, "global_step": 9049, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03835058957338333, "train_loss_bc": 0.03835058957338333, "train_loss_llm": 0.0, "grad_norm": 0.034598976373672485, "global_step": 9050, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03451751172542572, "train_loss_bc": 0.03451751172542572, "train_loss_llm": 0.0, "grad_norm": 0.04501572996377945, "global_step": 9051, "epoch": 251, "lr": 0.001} +{"train_loss": 0.04609369859099388, "train_loss_bc": 0.04609369859099388, "train_loss_llm": 0.0, "grad_norm": 0.052483174949884415, "global_step": 9052, "epoch": 251, "lr": 0.001} +{"train_loss": 0.040493156760931015, "train_loss_bc": 0.04045531153678894, "train_loss_llm": 0.0037845969200134277, "grad_norm": 0.02058706060051918, "global_step": 9053, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03722875937819481, "train_loss_bc": 0.03722875937819481, "train_loss_llm": 0.0, "grad_norm": 0.02526922896504402, "global_step": 9054, "epoch": 251, "lr": 0.001} +{"train_loss": 0.041011933237314224, "train_loss_bc": 0.041008979082107544, "train_loss_llm": 0.0002955496311187744, "grad_norm": 0.03398895263671875, "global_step": 9055, "epoch": 251, "lr": 0.001} +{"train_loss": 0.034369129687547684, "train_loss_bc": 0.034369129687547684, "train_loss_llm": 0.0, "grad_norm": 0.04719621315598488, "global_step": 9056, "epoch": 251, "lr": 0.001} +{"train_loss": 0.04694950208067894, "train_loss_bc": 0.04694950208067894, "train_loss_llm": 0.0, "grad_norm": 0.023561159148812294, "global_step": 9057, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03916056826710701, "train_loss_bc": 0.03916056826710701, "train_loss_llm": 0.0, "grad_norm": 0.02952229231595993, "global_step": 9058, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03635347634553909, "train_loss_bc": 0.03635300695896149, "train_loss_llm": 4.7087669372558594e-05, "grad_norm": 0.037928804755210876, "global_step": 9059, "epoch": 251, "lr": 0.001} +{"train_loss": 0.04096386581659317, "train_loss_bc": 0.040962543338537216, "train_loss_llm": 0.00013235211372375488, "grad_norm": 0.04692210257053375, "global_step": 9060, "epoch": 251, "lr": 0.001} +{"train_loss": 0.025275973603129387, "train_loss_bc": 0.025269031524658203, "train_loss_llm": 0.0006941556930541992, "grad_norm": 0.017244746908545494, "global_step": 9061, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03863275423645973, "train_loss_bc": 0.03863275423645973, "train_loss_llm": 0.0, "grad_norm": 0.03074360080063343, "global_step": 9062, "epoch": 251, "lr": 0.001} +{"train_loss": 0.04031660780310631, "train_loss_bc": 0.040316350758075714, "train_loss_llm": 2.5779008865356445e-05, "grad_norm": 0.04404592514038086, "global_step": 9063, "epoch": 251, "lr": 0.001} +{"train_loss": 0.037652142345905304, "train_loss_bc": 0.037648290395736694, "train_loss_llm": 0.0003853440284729004, "grad_norm": 0.05494160205125809, "global_step": 9064, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03774561360478401, "train_loss_bc": 0.037736840546131134, "train_loss_llm": 0.0008772611618041992, "grad_norm": 0.02629886195063591, "global_step": 9065, "epoch": 251, "lr": 0.001} +{"train_loss": 0.031994570046663284, "train_loss_bc": 0.031994570046663284, "train_loss_llm": 0.0, "grad_norm": 0.03974009305238724, "global_step": 9066, "epoch": 251, "lr": 0.001} +{"train_loss": 0.04031023383140564, "train_loss_bc": 0.04030634090304375, "train_loss_llm": 0.00038930773735046387, "grad_norm": 0.05504951998591423, "global_step": 9067, "epoch": 251, "lr": 0.001} +{"train_loss": 0.04065949097275734, "train_loss_bc": 0.0406561903655529, "train_loss_llm": 0.0003298819065093994, "grad_norm": 0.06247348338365555, "global_step": 9068, "epoch": 251, "lr": 0.001} +{"train_loss": 0.032671261578798294, "train_loss_bc": 0.03266304358839989, "train_loss_llm": 0.0008217990398406982, "grad_norm": 0.0171609316021204, "global_step": 9069, "epoch": 251, "lr": 0.001} +{"train_loss": 0.047851867973804474, "train_loss_bc": 0.047851867973804474, "train_loss_llm": 0.0, "grad_norm": 0.031288716942071915, "global_step": 9070, "epoch": 251, "lr": 0.001} +{"train_loss": 0.03918735936491026, "train_loss_bc": 0.04202656447887421, "train_loss_llm": 0.0, "grad_norm": 0.04212743416428566, "global_step": 9071, "epoch": 251, "lr": 0.001, "val_loss": 0.05984362214803696} +{"train_loss": 0.036879099905490875, "train_loss_bc": 0.0368630513548851, "train_loss_llm": 0.001604914665222168, "grad_norm": 0.04922650754451752, "global_step": 9072, "epoch": 252, "lr": 0.001} +{"train_loss": 0.033975858241319656, "train_loss_bc": 0.033975858241319656, "train_loss_llm": 0.0, "grad_norm": 0.021044107154011726, "global_step": 9073, "epoch": 252, "lr": 0.001} +{"train_loss": 0.039761971682310104, "train_loss_bc": 0.039761971682310104, "train_loss_llm": 0.0, "grad_norm": 0.03353956714272499, "global_step": 9074, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03892481327056885, "train_loss_bc": 0.03887980431318283, "train_loss_llm": 0.004500865936279297, "grad_norm": 0.047743335366249084, "global_step": 9075, "epoch": 252, "lr": 0.001} +{"train_loss": 0.04283257946372032, "train_loss_bc": 0.04282134771347046, "train_loss_llm": 0.0011230409145355225, "grad_norm": 0.06288927793502808, "global_step": 9076, "epoch": 252, "lr": 0.001} +{"train_loss": 0.04546160623431206, "train_loss_bc": 0.04546160623431206, "train_loss_llm": 0.0, "grad_norm": 0.017741074785590172, "global_step": 9077, "epoch": 252, "lr": 0.001} +{"train_loss": 0.04726407676935196, "train_loss_bc": 0.04726407676935196, "train_loss_llm": 0.0, "grad_norm": 0.025987939909100533, "global_step": 9078, "epoch": 252, "lr": 0.001} +{"train_loss": 0.04477185383439064, "train_loss_bc": 0.04477185383439064, "train_loss_llm": 0.0, "grad_norm": 0.032625965774059296, "global_step": 9079, "epoch": 252, "lr": 0.001} +{"train_loss": 0.038631416857242584, "train_loss_bc": 0.03862417861819267, "train_loss_llm": 0.0007239878177642822, "grad_norm": 0.039860211312770844, "global_step": 9080, "epoch": 252, "lr": 0.001} +{"train_loss": 0.037638790905475616, "train_loss_bc": 0.03762242943048477, "train_loss_llm": 0.0016362667083740234, "grad_norm": 0.02120562642812729, "global_step": 9081, "epoch": 252, "lr": 0.001} +{"train_loss": 0.046604037284851074, "train_loss_bc": 0.04660094156861305, "train_loss_llm": 0.0003095269203186035, "grad_norm": 0.03193648159503937, "global_step": 9082, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03543602675199509, "train_loss_bc": 0.03543602675199509, "train_loss_llm": 0.0, "grad_norm": 0.03762428089976311, "global_step": 9083, "epoch": 252, "lr": 0.001} +{"train_loss": 0.04059789702296257, "train_loss_bc": 0.04059789702296257, "train_loss_llm": 0.0, "grad_norm": 0.04731195420026779, "global_step": 9084, "epoch": 252, "lr": 0.001} +{"train_loss": 0.035916853696107864, "train_loss_bc": 0.035916853696107864, "train_loss_llm": 0.0, "grad_norm": 0.01801532693207264, "global_step": 9085, "epoch": 252, "lr": 0.001} +{"train_loss": 0.031741663813591, "train_loss_bc": 0.031741663813591, "train_loss_llm": 0.0, "grad_norm": 0.02705669030547142, "global_step": 9086, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03829096630215645, "train_loss_bc": 0.03827415406703949, "train_loss_llm": 0.0016813576221466064, "grad_norm": 0.03910830244421959, "global_step": 9087, "epoch": 252, "lr": 0.001} +{"train_loss": 0.04410518333315849, "train_loss_bc": 0.04410518333315849, "train_loss_llm": 0.0, "grad_norm": 0.0511728972196579, "global_step": 9088, "epoch": 252, "lr": 0.001} +{"train_loss": 0.04584972560405731, "train_loss_bc": 0.045842960476875305, "train_loss_llm": 0.0006765127182006836, "grad_norm": 0.022894835099577904, "global_step": 9089, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03878680616617203, "train_loss_bc": 0.03878680616617203, "train_loss_llm": 0.0, "grad_norm": 0.028903065249323845, "global_step": 9090, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03271143510937691, "train_loss_bc": 0.032701294869184494, "train_loss_llm": 0.001014113426208496, "grad_norm": 0.03582506626844406, "global_step": 9091, "epoch": 252, "lr": 0.001} +{"train_loss": 0.035139601677656174, "train_loss_bc": 0.03512463718652725, "train_loss_llm": 0.0014965832233428955, "grad_norm": 0.04528112709522247, "global_step": 9092, "epoch": 252, "lr": 0.001} +{"train_loss": 0.043061334639787674, "train_loss_bc": 0.043061334639787674, "train_loss_llm": 0.0, "grad_norm": 0.02340647578239441, "global_step": 9093, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03933010995388031, "train_loss_bc": 0.03933010995388031, "train_loss_llm": 0.0, "grad_norm": 0.029886320233345032, "global_step": 9094, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03909141570329666, "train_loss_bc": 0.03909141570329666, "train_loss_llm": 0.0, "grad_norm": 0.04899508133530617, "global_step": 9095, "epoch": 252, "lr": 0.001} +{"train_loss": 0.04898136854171753, "train_loss_bc": 0.04898136854171753, "train_loss_llm": 0.0, "grad_norm": 0.0681583359837532, "global_step": 9096, "epoch": 252, "lr": 0.001} +{"train_loss": 0.02973572351038456, "train_loss_bc": 0.02973572351038456, "train_loss_llm": 0.0, "grad_norm": 0.013355549424886703, "global_step": 9097, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03684322163462639, "train_loss_bc": 0.03684322163462639, "train_loss_llm": 0.0, "grad_norm": 0.02517920546233654, "global_step": 9098, "epoch": 252, "lr": 0.001} +{"train_loss": 0.039942577481269836, "train_loss_bc": 0.039942577481269836, "train_loss_llm": 0.0, "grad_norm": 0.03090553916990757, "global_step": 9099, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03789250925183296, "train_loss_bc": 0.03786319121718407, "train_loss_llm": 0.0029318630695343018, "grad_norm": 0.04704757779836655, "global_step": 9100, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03642256557941437, "train_loss_bc": 0.03642256557941437, "train_loss_llm": 0.0, "grad_norm": 0.016598587855696678, "global_step": 9101, "epoch": 252, "lr": 0.001} +{"train_loss": 0.04197464883327484, "train_loss_bc": 0.04197464883327484, "train_loss_llm": 0.0, "grad_norm": 0.02541985549032688, "global_step": 9102, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03194573521614075, "train_loss_bc": 0.03194573521614075, "train_loss_llm": 0.0, "grad_norm": 0.03054395690560341, "global_step": 9103, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03699135780334473, "train_loss_bc": 0.03696553781628609, "train_loss_llm": 0.002582043409347534, "grad_norm": 0.03965708613395691, "global_step": 9104, "epoch": 252, "lr": 0.001} +{"train_loss": 0.04043283686041832, "train_loss_bc": 0.04043283686041832, "train_loss_llm": 0.0, "grad_norm": 0.020009338855743408, "global_step": 9105, "epoch": 252, "lr": 0.001} +{"train_loss": 0.033412788063287735, "train_loss_bc": 0.033412788063287735, "train_loss_llm": 0.0, "grad_norm": 0.02966373786330223, "global_step": 9106, "epoch": 252, "lr": 0.001} +{"train_loss": 0.03904447161282102, "train_loss_bc": 0.038220521062612534, "train_loss_llm": 0.0, "grad_norm": 0.04097149148583412, "global_step": 9107, "epoch": 252, "lr": 0.001, "val_loss": 0.05944647639989853} +{"train_loss": 0.04206652566790581, "train_loss_bc": 0.04206652566790581, "train_loss_llm": 0.0, "grad_norm": 0.05370141565799713, "global_step": 9108, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04016704484820366, "train_loss_bc": 0.04016612470149994, "train_loss_llm": 9.205937385559082e-05, "grad_norm": 0.016838546842336655, "global_step": 9109, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03481883555650711, "train_loss_bc": 0.03481883555650711, "train_loss_llm": 0.0, "grad_norm": 0.025491787120699883, "global_step": 9110, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04181907698512077, "train_loss_bc": 0.041812360286712646, "train_loss_llm": 0.0006716251373291016, "grad_norm": 0.031901586800813675, "global_step": 9111, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03197275102138519, "train_loss_bc": 0.03197275102138519, "train_loss_llm": 0.0, "grad_norm": 0.03697165474295616, "global_step": 9112, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04181176424026489, "train_loss_bc": 0.04181176424026489, "train_loss_llm": 0.0, "grad_norm": 0.022937960922718048, "global_step": 9113, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04500790312886238, "train_loss_bc": 0.044973790645599365, "train_loss_llm": 0.0034110844135284424, "grad_norm": 0.04181342199444771, "global_step": 9114, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03134419769048691, "train_loss_bc": 0.03134419769048691, "train_loss_llm": 0.0, "grad_norm": 0.0480358749628067, "global_step": 9115, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04006931558251381, "train_loss_bc": 0.04006543383002281, "train_loss_llm": 0.00038802623748779297, "grad_norm": 0.056667882949113846, "global_step": 9116, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03720637783408165, "train_loss_bc": 0.037197012454271317, "train_loss_llm": 0.0009365677833557129, "grad_norm": 0.01979459822177887, "global_step": 9117, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04213482886552811, "train_loss_bc": 0.04213482886552811, "train_loss_llm": 0.0, "grad_norm": 0.02695365995168686, "global_step": 9118, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03820253908634186, "train_loss_bc": 0.038180138915777206, "train_loss_llm": 0.002240121364593506, "grad_norm": 0.03620649874210358, "global_step": 9119, "epoch": 253, "lr": 0.001} +{"train_loss": 0.037056561559438705, "train_loss_bc": 0.037056561559438705, "train_loss_llm": 0.0, "grad_norm": 0.044988952577114105, "global_step": 9120, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04215988889336586, "train_loss_bc": 0.04215235263109207, "train_loss_llm": 0.000753551721572876, "grad_norm": 0.019949113950133324, "global_step": 9121, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03849373012781143, "train_loss_bc": 0.03849373012781143, "train_loss_llm": 0.0, "grad_norm": 0.03731360286474228, "global_step": 9122, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04041515663266182, "train_loss_bc": 0.04041515663266182, "train_loss_llm": 0.0, "grad_norm": 0.04598705843091011, "global_step": 9123, "epoch": 253, "lr": 0.001} +{"train_loss": 0.031388022005558014, "train_loss_bc": 0.031388022005558014, "train_loss_llm": 0.0, "grad_norm": 0.05527481436729431, "global_step": 9124, "epoch": 253, "lr": 0.001} +{"train_loss": 0.040208302438259125, "train_loss_bc": 0.040208302438259125, "train_loss_llm": 0.0, "grad_norm": 0.015121590346097946, "global_step": 9125, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04822883754968643, "train_loss_bc": 0.04822883754968643, "train_loss_llm": 0.0, "grad_norm": 0.02885710634291172, "global_step": 9126, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03862244635820389, "train_loss_bc": 0.03862244635820389, "train_loss_llm": 0.0, "grad_norm": 0.04167591780424118, "global_step": 9127, "epoch": 253, "lr": 0.001} +{"train_loss": 0.037033505737781525, "train_loss_bc": 0.03702922910451889, "train_loss_llm": 0.00042754411697387695, "grad_norm": 0.05195743963122368, "global_step": 9128, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04471945762634277, "train_loss_bc": 0.04471945762634277, "train_loss_llm": 0.0, "grad_norm": 0.018992021679878235, "global_step": 9129, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04149612784385681, "train_loss_bc": 0.04149612784385681, "train_loss_llm": 0.0, "grad_norm": 0.02512049674987793, "global_step": 9130, "epoch": 253, "lr": 0.001} +{"train_loss": 0.041058167815208435, "train_loss_bc": 0.041058167815208435, "train_loss_llm": 0.0, "grad_norm": 0.030287988483905792, "global_step": 9131, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03577009215950966, "train_loss_bc": 0.03577009215950966, "train_loss_llm": 0.0, "grad_norm": 0.035916056483983994, "global_step": 9132, "epoch": 253, "lr": 0.001} +{"train_loss": 0.0334661602973938, "train_loss_bc": 0.03344090282917023, "train_loss_llm": 0.0025257766246795654, "grad_norm": 0.016858479008078575, "global_step": 9133, "epoch": 253, "lr": 0.001} +{"train_loss": 0.030930062755942345, "train_loss_bc": 0.030929218977689743, "train_loss_llm": 8.431077003479004e-05, "grad_norm": 0.022105827927589417, "global_step": 9134, "epoch": 253, "lr": 0.001} +{"train_loss": 0.045131538063287735, "train_loss_bc": 0.045131538063287735, "train_loss_llm": 0.0, "grad_norm": 0.03429684042930603, "global_step": 9135, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03806843236088753, "train_loss_bc": 0.03806843236088753, "train_loss_llm": 0.0, "grad_norm": 0.042656488716602325, "global_step": 9136, "epoch": 253, "lr": 0.001} +{"train_loss": 0.040848638862371445, "train_loss_bc": 0.040848638862371445, "train_loss_llm": 0.0, "grad_norm": 0.015399458818137646, "global_step": 9137, "epoch": 253, "lr": 0.001} +{"train_loss": 0.030680568888783455, "train_loss_bc": 0.03067353181540966, "train_loss_llm": 0.0007036328315734863, "grad_norm": 0.024154057726264, "global_step": 9138, "epoch": 253, "lr": 0.001} +{"train_loss": 0.034537073224782944, "train_loss_bc": 0.034537073224782944, "train_loss_llm": 0.0, "grad_norm": 0.02936672233045101, "global_step": 9139, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03779163211584091, "train_loss_bc": 0.03779163211584091, "train_loss_llm": 0.0, "grad_norm": 0.035766202956438065, "global_step": 9140, "epoch": 253, "lr": 0.001} +{"train_loss": 0.04129265248775482, "train_loss_bc": 0.04129265248775482, "train_loss_llm": 0.0, "grad_norm": 0.020902009680867195, "global_step": 9141, "epoch": 253, "lr": 0.001} +{"train_loss": 0.042852215468883514, "train_loss_bc": 0.042852215468883514, "train_loss_llm": 0.0, "grad_norm": 0.029104314744472504, "global_step": 9142, "epoch": 253, "lr": 0.001} +{"train_loss": 0.03891194529003567, "train_loss_bc": 0.041959598660469055, "train_loss_llm": 0.0, "grad_norm": 0.04110570624470711, "global_step": 9143, "epoch": 253, "lr": 0.001, "val_loss": 0.059631649404764175} +{"train_loss": 0.03997860103845596, "train_loss_bc": 0.039968091994524, "train_loss_llm": 0.0010508894920349121, "grad_norm": 0.052030012011528015, "global_step": 9144, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03758160397410393, "train_loss_bc": 0.03758160397410393, "train_loss_llm": 0.0, "grad_norm": 0.01820170320570469, "global_step": 9145, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03646932542324066, "train_loss_bc": 0.03646932542324066, "train_loss_llm": 0.0, "grad_norm": 0.023954354226589203, "global_step": 9146, "epoch": 254, "lr": 0.001} +{"train_loss": 0.042388271540403366, "train_loss_bc": 0.042388271540403366, "train_loss_llm": 0.0, "grad_norm": 0.030542593449354172, "global_step": 9147, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03672317788004875, "train_loss_bc": 0.03672317788004875, "train_loss_llm": 0.0, "grad_norm": 0.034427106380462646, "global_step": 9148, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03645459935069084, "train_loss_bc": 0.03645459935069084, "train_loss_llm": 0.0, "grad_norm": 0.01947062276303768, "global_step": 9149, "epoch": 254, "lr": 0.001} +{"train_loss": 0.04269900918006897, "train_loss_bc": 0.04269900918006897, "train_loss_llm": 0.0, "grad_norm": 0.036139436066150665, "global_step": 9150, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03323647007346153, "train_loss_bc": 0.03323647007346153, "train_loss_llm": 0.0, "grad_norm": 0.04428514465689659, "global_step": 9151, "epoch": 254, "lr": 0.001} +{"train_loss": 0.042687736451625824, "train_loss_bc": 0.042687736451625824, "train_loss_llm": 0.0, "grad_norm": 0.06330829858779907, "global_step": 9152, "epoch": 254, "lr": 0.001} +{"train_loss": 0.041371073573827744, "train_loss_bc": 0.041371073573827744, "train_loss_llm": 0.0, "grad_norm": 0.018808988854289055, "global_step": 9153, "epoch": 254, "lr": 0.001} +{"train_loss": 0.04025350138545036, "train_loss_bc": 0.04023601859807968, "train_loss_llm": 0.0017481446266174316, "grad_norm": 0.030242910608649254, "global_step": 9154, "epoch": 254, "lr": 0.001} +{"train_loss": 0.0322054848074913, "train_loss_bc": 0.0322054848074913, "train_loss_llm": 0.0, "grad_norm": 0.0378427729010582, "global_step": 9155, "epoch": 254, "lr": 0.001} +{"train_loss": 0.04052208364009857, "train_loss_bc": 0.04052208364009857, "train_loss_llm": 0.0, "grad_norm": 0.0410178117454052, "global_step": 9156, "epoch": 254, "lr": 0.001} +{"train_loss": 0.040961854159832, "train_loss_bc": 0.040961854159832, "train_loss_llm": 0.0, "grad_norm": 0.02271018922328949, "global_step": 9157, "epoch": 254, "lr": 0.001} +{"train_loss": 0.0390666127204895, "train_loss_bc": 0.03905438631772995, "train_loss_llm": 0.0012226402759552002, "grad_norm": 0.03389449417591095, "global_step": 9158, "epoch": 254, "lr": 0.001} +{"train_loss": 0.04127968102693558, "train_loss_bc": 0.04127638414502144, "train_loss_llm": 0.00032967329025268555, "grad_norm": 0.04698546230792999, "global_step": 9159, "epoch": 254, "lr": 0.001} +{"train_loss": 0.047769542783498764, "train_loss_bc": 0.047769542783498764, "train_loss_llm": 0.0, "grad_norm": 0.058015480637550354, "global_step": 9160, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03412590175867081, "train_loss_bc": 0.03412590175867081, "train_loss_llm": 0.0, "grad_norm": 0.017934506759047508, "global_step": 9161, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03382515907287598, "train_loss_bc": 0.033797211945056915, "train_loss_llm": 0.0027947723865509033, "grad_norm": 0.0244873259216547, "global_step": 9162, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03871184214949608, "train_loss_bc": 0.03871184214949608, "train_loss_llm": 0.0, "grad_norm": 0.03077796660363674, "global_step": 9163, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03499070182442665, "train_loss_bc": 0.03497947379946709, "train_loss_llm": 0.0011227130889892578, "grad_norm": 0.03437907621264458, "global_step": 9164, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03760223463177681, "train_loss_bc": 0.03759036213159561, "train_loss_llm": 0.0011870861053466797, "grad_norm": 0.0225759856402874, "global_step": 9165, "epoch": 254, "lr": 0.001} +{"train_loss": 0.043747544288635254, "train_loss_bc": 0.043747544288635254, "train_loss_llm": 0.0, "grad_norm": 0.029599705711007118, "global_step": 9166, "epoch": 254, "lr": 0.001} +{"train_loss": 0.038408368825912476, "train_loss_bc": 0.038408368825912476, "train_loss_llm": 0.0, "grad_norm": 0.036620792001485825, "global_step": 9167, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03555558994412422, "train_loss_bc": 0.035540007054805756, "train_loss_llm": 0.0015582442283630371, "grad_norm": 0.04672916606068611, "global_step": 9168, "epoch": 254, "lr": 0.001} +{"train_loss": 0.041713178157806396, "train_loss_bc": 0.04171271249651909, "train_loss_llm": 4.673004150390625e-05, "grad_norm": 0.02184504084289074, "global_step": 9169, "epoch": 254, "lr": 0.001} +{"train_loss": 0.04628317430615425, "train_loss_bc": 0.04626448079943657, "train_loss_llm": 0.001869291067123413, "grad_norm": 0.03369884938001633, "global_step": 9170, "epoch": 254, "lr": 0.001} +{"train_loss": 0.04042026028037071, "train_loss_bc": 0.04042026028037071, "train_loss_llm": 0.0, "grad_norm": 0.04703133553266525, "global_step": 9171, "epoch": 254, "lr": 0.001} +{"train_loss": 0.04004888981580734, "train_loss_bc": 0.04004888981580734, "train_loss_llm": 0.0, "grad_norm": 0.06223863363265991, "global_step": 9172, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03246519714593887, "train_loss_bc": 0.03246519714593887, "train_loss_llm": 0.0, "grad_norm": 0.014289375394582748, "global_step": 9173, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03292710334062576, "train_loss_bc": 0.03292710334062576, "train_loss_llm": 0.0, "grad_norm": 0.02816457487642765, "global_step": 9174, "epoch": 254, "lr": 0.001} +{"train_loss": 0.04274551570415497, "train_loss_bc": 0.04273591563105583, "train_loss_llm": 0.0009598433971405029, "grad_norm": 0.03503899276256561, "global_step": 9175, "epoch": 254, "lr": 0.001} +{"train_loss": 0.039802275598049164, "train_loss_bc": 0.039802275598049164, "train_loss_llm": 0.0, "grad_norm": 0.04749332740902901, "global_step": 9176, "epoch": 254, "lr": 0.001} +{"train_loss": 0.046435195952653885, "train_loss_bc": 0.046435195952653885, "train_loss_llm": 0.0, "grad_norm": 0.02202153392136097, "global_step": 9177, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03712795674800873, "train_loss_bc": 0.03712795674800873, "train_loss_llm": 0.0, "grad_norm": 0.03989860787987709, "global_step": 9178, "epoch": 254, "lr": 0.001} +{"train_loss": 0.03888131512535943, "train_loss_bc": 0.031142625957727432, "train_loss_llm": 0.0, "grad_norm": 0.04963603615760803, "global_step": 9179, "epoch": 254, "lr": 0.001, "val_loss": 0.06057186424732208} +{"train_loss": 0.0331462025642395, "train_loss_bc": 0.03313780575990677, "train_loss_llm": 0.00083962082862854, "grad_norm": 0.06278339773416519, "global_step": 9180, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03890406712889671, "train_loss_bc": 0.03890406712889671, "train_loss_llm": 0.0, "grad_norm": 0.016628514975309372, "global_step": 9181, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03570787236094475, "train_loss_bc": 0.03570787236094475, "train_loss_llm": 0.0, "grad_norm": 0.026059651747345924, "global_step": 9182, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03198372945189476, "train_loss_bc": 0.03198372945189476, "train_loss_llm": 0.0, "grad_norm": 0.03240383788943291, "global_step": 9183, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03633095696568489, "train_loss_bc": 0.03633095696568489, "train_loss_llm": 0.0, "grad_norm": 0.0362653024494648, "global_step": 9184, "epoch": 255, "lr": 0.001} +{"train_loss": 0.04330237954854965, "train_loss_bc": 0.04330237954854965, "train_loss_llm": 0.0, "grad_norm": 0.020274568349123, "global_step": 9185, "epoch": 255, "lr": 0.001} +{"train_loss": 0.037522271275520325, "train_loss_bc": 0.037522271275520325, "train_loss_llm": 0.0, "grad_norm": 0.03149723261594772, "global_step": 9186, "epoch": 255, "lr": 0.001} +{"train_loss": 0.038762468844652176, "train_loss_bc": 0.038762468844652176, "train_loss_llm": 0.0, "grad_norm": 0.033663615584373474, "global_step": 9187, "epoch": 255, "lr": 0.001} +{"train_loss": 0.041597478091716766, "train_loss_bc": 0.041597478091716766, "train_loss_llm": 0.0, "grad_norm": 0.037451427429914474, "global_step": 9188, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03776220604777336, "train_loss_bc": 0.03775551915168762, "train_loss_llm": 0.0006687343120574951, "grad_norm": 0.019329898059368134, "global_step": 9189, "epoch": 255, "lr": 0.001} +{"train_loss": 0.0371578074991703, "train_loss_bc": 0.0371578074991703, "train_loss_llm": 0.0, "grad_norm": 0.02826899290084839, "global_step": 9190, "epoch": 255, "lr": 0.001} +{"train_loss": 0.0375833734869957, "train_loss_bc": 0.0375833734869957, "train_loss_llm": 0.0, "grad_norm": 0.035320721566677094, "global_step": 9191, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03962530568242073, "train_loss_bc": 0.03962530568242073, "train_loss_llm": 0.0, "grad_norm": 0.04824678227305412, "global_step": 9192, "epoch": 255, "lr": 0.001} +{"train_loss": 0.044563185423612595, "train_loss_bc": 0.044563185423612595, "train_loss_llm": 0.0, "grad_norm": 0.022861646488308907, "global_step": 9193, "epoch": 255, "lr": 0.001} +{"train_loss": 0.036473412066698074, "train_loss_bc": 0.03641805052757263, "train_loss_llm": 0.005536079406738281, "grad_norm": 0.03314530476927757, "global_step": 9194, "epoch": 255, "lr": 0.001} +{"train_loss": 0.039634451270103455, "train_loss_bc": 0.039634451270103455, "train_loss_llm": 0.0, "grad_norm": 0.041153136640787125, "global_step": 9195, "epoch": 255, "lr": 0.001} +{"train_loss": 0.034739475697278976, "train_loss_bc": 0.034730251878499985, "train_loss_llm": 0.0009223520755767822, "grad_norm": 0.05051543563604355, "global_step": 9196, "epoch": 255, "lr": 0.001} +{"train_loss": 0.036972880363464355, "train_loss_bc": 0.0369599424302578, "train_loss_llm": 0.0012938380241394043, "grad_norm": 0.02115091308951378, "global_step": 9197, "epoch": 255, "lr": 0.001} +{"train_loss": 0.040559470653533936, "train_loss_bc": 0.040559470653533936, "train_loss_llm": 0.0, "grad_norm": 0.031030524522066116, "global_step": 9198, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03849295154213905, "train_loss_bc": 0.03849295154213905, "train_loss_llm": 0.0, "grad_norm": 0.046327654272317886, "global_step": 9199, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03807518631219864, "train_loss_bc": 0.03807518631219864, "train_loss_llm": 0.0, "grad_norm": 0.05662053823471069, "global_step": 9200, "epoch": 255, "lr": 0.001} +{"train_loss": 0.035908330231904984, "train_loss_bc": 0.035908330231904984, "train_loss_llm": 0.0, "grad_norm": 0.023060206323862076, "global_step": 9201, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03917919844388962, "train_loss_bc": 0.03917919844388962, "train_loss_llm": 0.0, "grad_norm": 0.030938416719436646, "global_step": 9202, "epoch": 255, "lr": 0.001} +{"train_loss": 0.035577431321144104, "train_loss_bc": 0.03553037717938423, "train_loss_llm": 0.004705280065536499, "grad_norm": 0.041697606444358826, "global_step": 9203, "epoch": 255, "lr": 0.001} +{"train_loss": 0.040346574038267136, "train_loss_bc": 0.04034044221043587, "train_loss_llm": 0.0006133615970611572, "grad_norm": 0.0498582124710083, "global_step": 9204, "epoch": 255, "lr": 0.001} +{"train_loss": 0.0446375273168087, "train_loss_bc": 0.044617582112550735, "train_loss_llm": 0.0019946396350860596, "grad_norm": 0.01949005201458931, "global_step": 9205, "epoch": 255, "lr": 0.001} +{"train_loss": 0.04228130355477333, "train_loss_bc": 0.04228130355477333, "train_loss_llm": 0.0, "grad_norm": 0.03929510712623596, "global_step": 9206, "epoch": 255, "lr": 0.001} +{"train_loss": 0.04101886600255966, "train_loss_bc": 0.04101886600255966, "train_loss_llm": 0.0, "grad_norm": 0.050945404917001724, "global_step": 9207, "epoch": 255, "lr": 0.001} +{"train_loss": 0.034900564700365067, "train_loss_bc": 0.03489336743950844, "train_loss_llm": 0.0007196664810180664, "grad_norm": 0.05982551723718643, "global_step": 9208, "epoch": 255, "lr": 0.001} +{"train_loss": 0.048934146761894226, "train_loss_bc": 0.048934146761894226, "train_loss_llm": 0.0, "grad_norm": 0.023028075695037842, "global_step": 9209, "epoch": 255, "lr": 0.001} +{"train_loss": 0.0411546416580677, "train_loss_bc": 0.0411546416580677, "train_loss_llm": 0.0, "grad_norm": 0.032010894268751144, "global_step": 9210, "epoch": 255, "lr": 0.001} +{"train_loss": 0.042544715106487274, "train_loss_bc": 0.04252086579799652, "train_loss_llm": 0.0023851096630096436, "grad_norm": 0.03946579992771149, "global_step": 9211, "epoch": 255, "lr": 0.001} +{"train_loss": 0.039592135697603226, "train_loss_bc": 0.03958424553275108, "train_loss_llm": 0.0007888674736022949, "grad_norm": 0.052256226539611816, "global_step": 9212, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03973791003227234, "train_loss_bc": 0.03973791003227234, "train_loss_llm": 0.0, "grad_norm": 0.01957698166370392, "global_step": 9213, "epoch": 255, "lr": 0.001} +{"train_loss": 0.037470169365406036, "train_loss_bc": 0.037470169365406036, "train_loss_llm": 0.0, "grad_norm": 0.033013682812452316, "global_step": 9214, "epoch": 255, "lr": 0.001} +{"train_loss": 0.03894707632975446, "train_loss_bc": 0.039914101362228394, "train_loss_llm": 0.0, "grad_norm": 0.04710569977760315, "global_step": 9215, "epoch": 255, "lr": 0.001, "train/cumulative_reward": 121.20122861659566, "train/mean_score": 18.914315188421092, "train/success_rate": 0.5, "test/cumulative_reward": 79.90443951424692, "test/mean_score": 9.94375207319128, "test/success_rate": 0.22, "val_loss": 0.05928013101220131, "train_action_mse_error": 0.0428902730345726} +{"train_loss": 0.03846405819058418, "train_loss_bc": 0.03845694661140442, "train_loss_llm": 0.0007113218307495117, "grad_norm": 0.06435124576091766, "global_step": 9216, "epoch": 256, "lr": 0.001} +{"train_loss": 0.039439570158720016, "train_loss_bc": 0.039439570158720016, "train_loss_llm": 0.0, "grad_norm": 0.021659286692738533, "global_step": 9217, "epoch": 256, "lr": 0.001} +{"train_loss": 0.032767970114946365, "train_loss_bc": 0.032767970114946365, "train_loss_llm": 0.0, "grad_norm": 0.03037095069885254, "global_step": 9218, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03284286707639694, "train_loss_bc": 0.03284286707639694, "train_loss_llm": 0.0, "grad_norm": 0.0433429479598999, "global_step": 9219, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03757496550679207, "train_loss_bc": 0.03757496550679207, "train_loss_llm": 0.0, "grad_norm": 0.0486418642103672, "global_step": 9220, "epoch": 256, "lr": 0.001} +{"train_loss": 0.04000188037753105, "train_loss_bc": 0.04000188037753105, "train_loss_llm": 0.0, "grad_norm": 0.018458640202879906, "global_step": 9221, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03931642696261406, "train_loss_bc": 0.03931642696261406, "train_loss_llm": 0.0, "grad_norm": 0.0265323668718338, "global_step": 9222, "epoch": 256, "lr": 0.001} +{"train_loss": 0.04102588817477226, "train_loss_bc": 0.04102588817477226, "train_loss_llm": 0.0, "grad_norm": 0.0442671924829483, "global_step": 9223, "epoch": 256, "lr": 0.001} +{"train_loss": 0.04192141816020012, "train_loss_bc": 0.04192141816020012, "train_loss_llm": 0.0, "grad_norm": 0.05475636571645737, "global_step": 9224, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03638062998652458, "train_loss_bc": 0.03637794032692909, "train_loss_llm": 0.0002690255641937256, "grad_norm": 0.020026816055178642, "global_step": 9225, "epoch": 256, "lr": 0.001} +{"train_loss": 0.038522329181432724, "train_loss_bc": 0.0384967215359211, "train_loss_llm": 0.0025607645511627197, "grad_norm": 0.03246118873357773, "global_step": 9226, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03476529195904732, "train_loss_bc": 0.03474150225520134, "train_loss_llm": 0.0023789703845977783, "grad_norm": 0.044174812734127045, "global_step": 9227, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03384685143828392, "train_loss_bc": 0.03383694589138031, "train_loss_llm": 0.000990539789199829, "grad_norm": 0.046776629984378815, "global_step": 9228, "epoch": 256, "lr": 0.001} +{"train_loss": 0.041802771389484406, "train_loss_bc": 0.041802771389484406, "train_loss_llm": 0.0, "grad_norm": 0.024909839034080505, "global_step": 9229, "epoch": 256, "lr": 0.001} +{"train_loss": 0.04166187345981598, "train_loss_bc": 0.04166187345981598, "train_loss_llm": 0.0, "grad_norm": 0.04538453742861748, "global_step": 9230, "epoch": 256, "lr": 0.001} +{"train_loss": 0.04371567443013191, "train_loss_bc": 0.04371567443013191, "train_loss_llm": 0.0, "grad_norm": 0.0530981682240963, "global_step": 9231, "epoch": 256, "lr": 0.001} +{"train_loss": 0.04770439863204956, "train_loss_bc": 0.04770439863204956, "train_loss_llm": 0.0, "grad_norm": 0.06493991613388062, "global_step": 9232, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03901578485965729, "train_loss_bc": 0.03901578485965729, "train_loss_llm": 0.0, "grad_norm": 0.021944325417280197, "global_step": 9233, "epoch": 256, "lr": 0.001} +{"train_loss": 0.039159100502729416, "train_loss_bc": 0.039159100502729416, "train_loss_llm": 0.0, "grad_norm": 0.03201112896203995, "global_step": 9234, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03330729529261589, "train_loss_bc": 0.03330729529261589, "train_loss_llm": 0.0, "grad_norm": 0.04349752515554428, "global_step": 9235, "epoch": 256, "lr": 0.001} +{"train_loss": 0.04607009515166283, "train_loss_bc": 0.04607009515166283, "train_loss_llm": 0.0, "grad_norm": 0.051538821309804916, "global_step": 9236, "epoch": 256, "lr": 0.001} +{"train_loss": 0.0358525849878788, "train_loss_bc": 0.0358525849878788, "train_loss_llm": 0.0, "grad_norm": 0.020833510905504227, "global_step": 9237, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03831075504422188, "train_loss_bc": 0.038308724761009216, "train_loss_llm": 0.00020295381546020508, "grad_norm": 0.03133084625005722, "global_step": 9238, "epoch": 256, "lr": 0.001} +{"train_loss": 0.040232982486486435, "train_loss_bc": 0.040232982486486435, "train_loss_llm": 0.0, "grad_norm": 0.04011938348412514, "global_step": 9239, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03530210629105568, "train_loss_bc": 0.03529907017946243, "train_loss_llm": 0.00030353665351867676, "grad_norm": 0.04844794422388077, "global_step": 9240, "epoch": 256, "lr": 0.001} +{"train_loss": 0.0360523946583271, "train_loss_bc": 0.0360523946583271, "train_loss_llm": 0.0, "grad_norm": 0.01740608550608158, "global_step": 9241, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03763803839683533, "train_loss_bc": 0.03763589262962341, "train_loss_llm": 0.00021463632583618164, "grad_norm": 0.025828251615166664, "global_step": 9242, "epoch": 256, "lr": 0.001} +{"train_loss": 0.04337335750460625, "train_loss_bc": 0.04337335750460625, "train_loss_llm": 0.0, "grad_norm": 0.0389871820807457, "global_step": 9243, "epoch": 256, "lr": 0.001} +{"train_loss": 0.045254942029714584, "train_loss_bc": 0.04525031894445419, "train_loss_llm": 0.0004622936248779297, "grad_norm": 0.0559808723628521, "global_step": 9244, "epoch": 256, "lr": 0.001} +{"train_loss": 0.034909576177597046, "train_loss_bc": 0.034909576177597046, "train_loss_llm": 0.0, "grad_norm": 0.01525818184018135, "global_step": 9245, "epoch": 256, "lr": 0.001} +{"train_loss": 0.042343929409980774, "train_loss_bc": 0.042343929409980774, "train_loss_llm": 0.0, "grad_norm": 0.031357377767562866, "global_step": 9246, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03991546854376793, "train_loss_bc": 0.03991546854376793, "train_loss_llm": 0.0, "grad_norm": 0.04472579434514046, "global_step": 9247, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03951588645577431, "train_loss_bc": 0.03951588645577431, "train_loss_llm": 0.0, "grad_norm": 0.05408377945423126, "global_step": 9248, "epoch": 256, "lr": 0.001} +{"train_loss": 0.038447897881269455, "train_loss_bc": 0.038447897881269455, "train_loss_llm": 0.0, "grad_norm": 0.014624199829995632, "global_step": 9249, "epoch": 256, "lr": 0.001} +{"train_loss": 0.0387401357293129, "train_loss_bc": 0.03873147442936897, "train_loss_llm": 0.0008659660816192627, "grad_norm": 0.027194514870643616, "global_step": 9250, "epoch": 256, "lr": 0.001} +{"train_loss": 0.03898341032779879, "train_loss_bc": 0.038205575197935104, "train_loss_llm": 0.0, "grad_norm": 0.041574906557798386, "global_step": 9251, "epoch": 256, "lr": 0.001, "val_loss": 0.05979544296860695} +{"train_loss": 0.029876237735152245, "train_loss_bc": 0.029876237735152245, "train_loss_llm": 0.0, "grad_norm": 0.048666175454854965, "global_step": 9252, "epoch": 257, "lr": 0.001} +{"train_loss": 0.033426642417907715, "train_loss_bc": 0.033426642417907715, "train_loss_llm": 0.0, "grad_norm": 0.018173901364207268, "global_step": 9253, "epoch": 257, "lr": 0.001} +{"train_loss": 0.03616834804415703, "train_loss_bc": 0.03616834804415703, "train_loss_llm": 0.0, "grad_norm": 0.03210592642426491, "global_step": 9254, "epoch": 257, "lr": 0.001} +{"train_loss": 0.04273119941353798, "train_loss_bc": 0.04273119941353798, "train_loss_llm": 0.0, "grad_norm": 0.040168244391679764, "global_step": 9255, "epoch": 257, "lr": 0.001} +{"train_loss": 0.038034602999687195, "train_loss_bc": 0.038034602999687195, "train_loss_llm": 0.0, "grad_norm": 0.04297484830021858, "global_step": 9256, "epoch": 257, "lr": 0.001} +{"train_loss": 0.04637523368000984, "train_loss_bc": 0.046366941183805466, "train_loss_llm": 0.0008292198181152344, "grad_norm": 0.02942613698542118, "global_step": 9257, "epoch": 257, "lr": 0.001} +{"train_loss": 0.0374564491212368, "train_loss_bc": 0.0374564491212368, "train_loss_llm": 0.0, "grad_norm": 0.0390227846801281, "global_step": 9258, "epoch": 257, "lr": 0.001} +{"train_loss": 0.03122212179005146, "train_loss_bc": 0.03122212179005146, "train_loss_llm": 0.0, "grad_norm": 0.045605629682540894, "global_step": 9259, "epoch": 257, "lr": 0.001} +{"train_loss": 0.0463545024394989, "train_loss_bc": 0.04633089900016785, "train_loss_llm": 0.0023602843284606934, "grad_norm": 0.06022294983267784, "global_step": 9260, "epoch": 257, "lr": 0.001} +{"train_loss": 0.04008099436759949, "train_loss_bc": 0.04008099436759949, "train_loss_llm": 0.0, "grad_norm": 0.026183996349573135, "global_step": 9261, "epoch": 257, "lr": 0.001} +{"train_loss": 0.0395619161427021, "train_loss_bc": 0.0395619161427021, "train_loss_llm": 0.0, "grad_norm": 0.040449611842632294, "global_step": 9262, "epoch": 257, "lr": 0.001} +{"train_loss": 0.03363143652677536, "train_loss_bc": 0.03363143652677536, "train_loss_llm": 0.0, "grad_norm": 0.0537848174571991, "global_step": 9263, "epoch": 257, "lr": 0.001} +{"train_loss": 0.03711482137441635, "train_loss_bc": 0.037108249962329865, "train_loss_llm": 0.0006573200225830078, "grad_norm": 0.0635189637541771, "global_step": 9264, "epoch": 257, "lr": 0.001} +{"train_loss": 0.044727303087711334, "train_loss_bc": 0.044727303087711334, "train_loss_llm": 0.0, "grad_norm": 0.018290113657712936, "global_step": 9265, "epoch": 257, "lr": 0.001} +{"train_loss": 0.037413910031318665, "train_loss_bc": 0.03741388022899628, "train_loss_llm": 2.8312206268310547e-06, "grad_norm": 0.024639375507831573, "global_step": 9266, "epoch": 257, "lr": 0.001} +{"train_loss": 0.0437152124941349, "train_loss_bc": 0.043688442558050156, "train_loss_llm": 0.0026769638061523438, "grad_norm": 0.02902640588581562, "global_step": 9267, "epoch": 257, "lr": 0.001} +{"train_loss": 0.04383351281285286, "train_loss_bc": 0.04383351281285286, "train_loss_llm": 0.0, "grad_norm": 0.037695929408073425, "global_step": 9268, "epoch": 257, "lr": 0.001} +{"train_loss": 0.038780417293310165, "train_loss_bc": 0.038780417293310165, "train_loss_llm": 0.0, "grad_norm": 0.02130609191954136, "global_step": 9269, "epoch": 257, "lr": 0.001} +{"train_loss": 0.040251556783914566, "train_loss_bc": 0.0402354970574379, "train_loss_llm": 0.0016061365604400635, "grad_norm": 0.029999064281582832, "global_step": 9270, "epoch": 257, "lr": 0.001} +{"train_loss": 0.04164806380867958, "train_loss_bc": 0.041642796248197556, "train_loss_llm": 0.0005269348621368408, "grad_norm": 0.04260560870170593, "global_step": 9271, "epoch": 257, "lr": 0.001} +{"train_loss": 0.047332558780908585, "train_loss_bc": 0.04732804372906685, "train_loss_llm": 0.0004515349864959717, "grad_norm": 0.06165270507335663, "global_step": 9272, "epoch": 257, "lr": 0.001} +{"train_loss": 0.03595134615898132, "train_loss_bc": 0.03591343015432358, "train_loss_llm": 0.003791600465774536, "grad_norm": 0.023198001086711884, "global_step": 9273, "epoch": 257, "lr": 0.001} +{"train_loss": 0.04278218001127243, "train_loss_bc": 0.04276913404464722, "train_loss_llm": 0.0013047754764556885, "grad_norm": 0.036150477826595306, "global_step": 9274, "epoch": 257, "lr": 0.001} +{"train_loss": 0.04134291037917137, "train_loss_bc": 0.04133966192603111, "train_loss_llm": 0.00032469630241394043, "grad_norm": 0.04359429329633713, "global_step": 9275, "epoch": 257, "lr": 0.001} +{"train_loss": 0.04447684809565544, "train_loss_bc": 0.04447684809565544, "train_loss_llm": 0.0, "grad_norm": 0.05696606636047363, "global_step": 9276, "epoch": 257, "lr": 0.001} +{"train_loss": 0.034363582730293274, "train_loss_bc": 0.034363582730293274, "train_loss_llm": 0.0, "grad_norm": 0.02242497354745865, "global_step": 9277, "epoch": 257, "lr": 0.001} +{"train_loss": 0.03959633782505989, "train_loss_bc": 0.039596185088157654, "train_loss_llm": 1.537799835205078e-05, "grad_norm": 0.030658261850476265, "global_step": 9278, "epoch": 257, "lr": 0.001} +{"train_loss": 0.03639468923211098, "train_loss_bc": 0.03639468923211098, "train_loss_llm": 0.0, "grad_norm": 0.041136227548122406, "global_step": 9279, "epoch": 257, "lr": 0.001} +{"train_loss": 0.036177024245262146, "train_loss_bc": 0.036177024245262146, "train_loss_llm": 0.0, "grad_norm": 0.046685345470905304, "global_step": 9280, "epoch": 257, "lr": 0.001} +{"train_loss": 0.03233803063631058, "train_loss_bc": 0.03233803063631058, "train_loss_llm": 0.0, "grad_norm": 0.01810941845178604, "global_step": 9281, "epoch": 257, "lr": 0.001} +{"train_loss": 0.040442269295454025, "train_loss_bc": 0.040439423173666, "train_loss_llm": 0.0002847611904144287, "grad_norm": 0.026486165821552277, "global_step": 9282, "epoch": 257, "lr": 0.001} +{"train_loss": 0.036997489631175995, "train_loss_bc": 0.036997489631175995, "train_loss_llm": 0.0, "grad_norm": 0.032954152673482895, "global_step": 9283, "epoch": 257, "lr": 0.001} +{"train_loss": 0.04348904639482498, "train_loss_bc": 0.04348904639482498, "train_loss_llm": 0.0, "grad_norm": 0.040996894240379333, "global_step": 9284, "epoch": 257, "lr": 0.001} +{"train_loss": 0.03812023997306824, "train_loss_bc": 0.03811969235539436, "train_loss_llm": 5.492568016052246e-05, "grad_norm": 0.017788134515285492, "global_step": 9285, "epoch": 257, "lr": 0.001} +{"train_loss": 0.039430294185876846, "train_loss_bc": 0.039430294185876846, "train_loss_llm": 0.0, "grad_norm": 0.03473052382469177, "global_step": 9286, "epoch": 257, "lr": 0.001} +{"train_loss": 0.039010525991519295, "train_loss_bc": 0.03273322805762291, "train_loss_llm": 0.0006377995014190674, "grad_norm": 0.04535987228155136, "global_step": 9287, "epoch": 257, "lr": 0.001, "val_loss": 0.06024746596813202} +{"train_loss": 0.040223319083452225, "train_loss_bc": 0.04019872844219208, "train_loss_llm": 0.002459019422531128, "grad_norm": 0.05428425595164299, "global_step": 9288, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03384644538164139, "train_loss_bc": 0.03384644538164139, "train_loss_llm": 0.0, "grad_norm": 0.02148464322090149, "global_step": 9289, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04060700908303261, "train_loss_bc": 0.04060593247413635, "train_loss_llm": 0.00010770559310913086, "grad_norm": 0.03469480574131012, "global_step": 9290, "epoch": 258, "lr": 0.001} +{"train_loss": 0.044206298887729645, "train_loss_bc": 0.044206298887729645, "train_loss_llm": 0.0, "grad_norm": 0.04416099190711975, "global_step": 9291, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04000994935631752, "train_loss_bc": 0.04000994935631752, "train_loss_llm": 0.0, "grad_norm": 0.054585907608270645, "global_step": 9292, "epoch": 258, "lr": 0.001} +{"train_loss": 0.037687789648771286, "train_loss_bc": 0.03767746314406395, "train_loss_llm": 0.001032799482345581, "grad_norm": 0.01876249723136425, "global_step": 9293, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03457987308502197, "train_loss_bc": 0.034559570252895355, "train_loss_llm": 0.002030402421951294, "grad_norm": 0.03130104020237923, "global_step": 9294, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03632919490337372, "train_loss_bc": 0.036324575543403625, "train_loss_llm": 0.00046199560165405273, "grad_norm": 0.04248354211449623, "global_step": 9295, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04553063213825226, "train_loss_bc": 0.04553063213825226, "train_loss_llm": 0.0, "grad_norm": 0.05388571694493294, "global_step": 9296, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04294891655445099, "train_loss_bc": 0.04294891655445099, "train_loss_llm": 0.0, "grad_norm": 0.022013980895280838, "global_step": 9297, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03596563637256622, "train_loss_bc": 0.03596388176083565, "train_loss_llm": 0.00017547607421875, "grad_norm": 0.03287479281425476, "global_step": 9298, "epoch": 258, "lr": 0.001} +{"train_loss": 0.0412726104259491, "train_loss_bc": 0.04125888645648956, "train_loss_llm": 0.0013722777366638184, "grad_norm": 0.038766589015722275, "global_step": 9299, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03492226451635361, "train_loss_bc": 0.03492226451635361, "train_loss_llm": 0.0, "grad_norm": 0.05273648723959923, "global_step": 9300, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04580598697066307, "train_loss_bc": 0.04580598697066307, "train_loss_llm": 0.0, "grad_norm": 0.025733480229973793, "global_step": 9301, "epoch": 258, "lr": 0.001} +{"train_loss": 0.039872437715530396, "train_loss_bc": 0.039872437715530396, "train_loss_llm": 0.0, "grad_norm": 0.0388830341398716, "global_step": 9302, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03897598013281822, "train_loss_bc": 0.03897598013281822, "train_loss_llm": 0.0, "grad_norm": 0.04898282140493393, "global_step": 9303, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03219185769557953, "train_loss_bc": 0.03217969834804535, "train_loss_llm": 0.0012159347534179688, "grad_norm": 0.06825866550207138, "global_step": 9304, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04510697349905968, "train_loss_bc": 0.04510697349905968, "train_loss_llm": 0.0, "grad_norm": 0.023039720952510834, "global_step": 9305, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04411792755126953, "train_loss_bc": 0.04411792755126953, "train_loss_llm": 0.0, "grad_norm": 0.03244509920477867, "global_step": 9306, "epoch": 258, "lr": 0.001} +{"train_loss": 0.036180224269628525, "train_loss_bc": 0.03615284338593483, "train_loss_llm": 0.002738058567047119, "grad_norm": 0.03837842121720314, "global_step": 9307, "epoch": 258, "lr": 0.001} +{"train_loss": 0.038252413272857666, "train_loss_bc": 0.03823923319578171, "train_loss_llm": 0.0013179183006286621, "grad_norm": 0.04449529945850372, "global_step": 9308, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04512767121195793, "train_loss_bc": 0.04511615261435509, "train_loss_llm": 0.0011518001556396484, "grad_norm": 0.01793348789215088, "global_step": 9309, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04032830893993378, "train_loss_bc": 0.04032493010163307, "train_loss_llm": 0.00033780932426452637, "grad_norm": 0.028928101062774658, "global_step": 9310, "epoch": 258, "lr": 0.001} +{"train_loss": 0.030802706256508827, "train_loss_bc": 0.030802706256508827, "train_loss_llm": 0.0, "grad_norm": 0.04123597964644432, "global_step": 9311, "epoch": 258, "lr": 0.001} +{"train_loss": 0.0374828465282917, "train_loss_bc": 0.0374828465282917, "train_loss_llm": 0.0, "grad_norm": 0.04677888751029968, "global_step": 9312, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03956175595521927, "train_loss_bc": 0.03956175595521927, "train_loss_llm": 0.0, "grad_norm": 0.022081146016716957, "global_step": 9313, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03777959942817688, "train_loss_bc": 0.03777771815657616, "train_loss_llm": 0.0001881718635559082, "grad_norm": 0.03913301229476929, "global_step": 9314, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03240828961133957, "train_loss_bc": 0.03239025920629501, "train_loss_llm": 0.0018029212951660156, "grad_norm": 0.046122707426548004, "global_step": 9315, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04228426516056061, "train_loss_bc": 0.04228426516056061, "train_loss_llm": 0.0, "grad_norm": 0.05963022634387016, "global_step": 9316, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03300977870821953, "train_loss_bc": 0.03300543129444122, "train_loss_llm": 0.00043466687202453613, "grad_norm": 0.017742648720741272, "global_step": 9317, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03571709245443344, "train_loss_bc": 0.03571709245443344, "train_loss_llm": 0.0, "grad_norm": 0.02721296437084675, "global_step": 9318, "epoch": 258, "lr": 0.001} +{"train_loss": 0.04231075569987297, "train_loss_bc": 0.04231075569987297, "train_loss_llm": 0.0, "grad_norm": 0.03548949211835861, "global_step": 9319, "epoch": 258, "lr": 0.001} +{"train_loss": 0.043689701706171036, "train_loss_bc": 0.043689701706171036, "train_loss_llm": 0.0, "grad_norm": 0.046517882496118546, "global_step": 9320, "epoch": 258, "lr": 0.001} +{"train_loss": 0.040842436254024506, "train_loss_bc": 0.040842436254024506, "train_loss_llm": 0.0, "grad_norm": 0.02430902048945427, "global_step": 9321, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03902328386902809, "train_loss_bc": 0.03902328386902809, "train_loss_llm": 0.0, "grad_norm": 0.0329863503575325, "global_step": 9322, "epoch": 258, "lr": 0.001} +{"train_loss": 0.03905203136511975, "train_loss_bc": 0.03685872256755829, "train_loss_llm": 0.0012174248695373535, "grad_norm": 0.04039071500301361, "global_step": 9323, "epoch": 258, "lr": 0.001, "val_loss": 0.059120066463947296} +{"train_loss": 0.040786013007164, "train_loss_bc": 0.040786013007164, "train_loss_llm": 0.0, "grad_norm": 0.04756453260779381, "global_step": 9324, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03384145721793175, "train_loss_bc": 0.03383000195026398, "train_loss_llm": 0.0011453628540039062, "grad_norm": 0.021042145788669586, "global_step": 9325, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03840988129377365, "train_loss_bc": 0.03840988129377365, "train_loss_llm": 0.0, "grad_norm": 0.03205648064613342, "global_step": 9326, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03332625702023506, "train_loss_bc": 0.03330537676811218, "train_loss_llm": 0.002087891101837158, "grad_norm": 0.04274599999189377, "global_step": 9327, "epoch": 259, "lr": 0.001} +{"train_loss": 0.041988473385572433, "train_loss_bc": 0.04198646917939186, "train_loss_llm": 0.00020036101341247559, "grad_norm": 0.05254501849412918, "global_step": 9328, "epoch": 259, "lr": 0.001} +{"train_loss": 0.037571463733911514, "train_loss_bc": 0.037571463733911514, "train_loss_llm": 0.0, "grad_norm": 0.016547972336411476, "global_step": 9329, "epoch": 259, "lr": 0.001} +{"train_loss": 0.04543218016624451, "train_loss_bc": 0.04543218016624451, "train_loss_llm": 0.0, "grad_norm": 0.026289768517017365, "global_step": 9330, "epoch": 259, "lr": 0.001} +{"train_loss": 0.035936467349529266, "train_loss_bc": 0.035936467349529266, "train_loss_llm": 0.0, "grad_norm": 0.0344981849193573, "global_step": 9331, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03135433793067932, "train_loss_bc": 0.03135433793067932, "train_loss_llm": 0.0, "grad_norm": 0.04480164498090744, "global_step": 9332, "epoch": 259, "lr": 0.001} +{"train_loss": 0.044854938983917236, "train_loss_bc": 0.044854938983917236, "train_loss_llm": 0.0, "grad_norm": 0.019709186628460884, "global_step": 9333, "epoch": 259, "lr": 0.001} +{"train_loss": 0.04206586256623268, "train_loss_bc": 0.04206586256623268, "train_loss_llm": 0.0, "grad_norm": 0.03233497217297554, "global_step": 9334, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03273623436689377, "train_loss_bc": 0.03273623436689377, "train_loss_llm": 0.0, "grad_norm": 0.040834538638591766, "global_step": 9335, "epoch": 259, "lr": 0.001} +{"train_loss": 0.027936527505517006, "train_loss_bc": 0.027904802933335304, "train_loss_llm": 0.0031724870204925537, "grad_norm": 0.047960422933101654, "global_step": 9336, "epoch": 259, "lr": 0.001} +{"train_loss": 0.034255195409059525, "train_loss_bc": 0.03423778712749481, "train_loss_llm": 0.001740962266921997, "grad_norm": 0.02015022560954094, "global_step": 9337, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03469538688659668, "train_loss_bc": 0.03469538688659668, "train_loss_llm": 0.0, "grad_norm": 0.026343723759055138, "global_step": 9338, "epoch": 259, "lr": 0.001} +{"train_loss": 0.038637954741716385, "train_loss_bc": 0.038637954741716385, "train_loss_llm": 0.0, "grad_norm": 0.03904947638511658, "global_step": 9339, "epoch": 259, "lr": 0.001} +{"train_loss": 0.039001479744911194, "train_loss_bc": 0.03898365795612335, "train_loss_llm": 0.0017820894718170166, "grad_norm": 0.05036504566669464, "global_step": 9340, "epoch": 259, "lr": 0.001} +{"train_loss": 0.04500455781817436, "train_loss_bc": 0.044999897480010986, "train_loss_llm": 0.0004659891128540039, "grad_norm": 0.025401799008250237, "global_step": 9341, "epoch": 259, "lr": 0.001} +{"train_loss": 0.030136654153466225, "train_loss_bc": 0.030135558918118477, "train_loss_llm": 0.00010949373245239258, "grad_norm": 0.033025775104761124, "global_step": 9342, "epoch": 259, "lr": 0.001} +{"train_loss": 0.04600812494754791, "train_loss_bc": 0.04600812494754791, "train_loss_llm": 0.0, "grad_norm": 0.04444010928273201, "global_step": 9343, "epoch": 259, "lr": 0.001} +{"train_loss": 0.04254854843020439, "train_loss_bc": 0.04254854843020439, "train_loss_llm": 0.0, "grad_norm": 0.05662589520215988, "global_step": 9344, "epoch": 259, "lr": 0.001} +{"train_loss": 0.037616901099681854, "train_loss_bc": 0.037616901099681854, "train_loss_llm": 0.0, "grad_norm": 0.020794592797756195, "global_step": 9345, "epoch": 259, "lr": 0.001} +{"train_loss": 0.040743254125118256, "train_loss_bc": 0.040743254125118256, "train_loss_llm": 0.0, "grad_norm": 0.036020051687955856, "global_step": 9346, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03658716753125191, "train_loss_bc": 0.036575064063072205, "train_loss_llm": 0.0012104809284210205, "grad_norm": 0.04404658079147339, "global_step": 9347, "epoch": 259, "lr": 0.001} +{"train_loss": 0.045104045420885086, "train_loss_bc": 0.04510367289185524, "train_loss_llm": 3.7223100662231445e-05, "grad_norm": 0.05182149261236191, "global_step": 9348, "epoch": 259, "lr": 0.001} +{"train_loss": 0.04276302829384804, "train_loss_bc": 0.04276302829384804, "train_loss_llm": 0.0, "grad_norm": 0.02237437106668949, "global_step": 9349, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03875184431672096, "train_loss_bc": 0.03875184431672096, "train_loss_llm": 0.0, "grad_norm": 0.033619243651628494, "global_step": 9350, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03801603987812996, "train_loss_bc": 0.038015931844711304, "train_loss_llm": 1.0967254638671875e-05, "grad_norm": 0.03939559683203697, "global_step": 9351, "epoch": 259, "lr": 0.001} +{"train_loss": 0.036517027765512466, "train_loss_bc": 0.036517027765512466, "train_loss_llm": 0.0, "grad_norm": 0.04281972348690033, "global_step": 9352, "epoch": 259, "lr": 0.001} +{"train_loss": 0.04315785691142082, "train_loss_bc": 0.04315785691142082, "train_loss_llm": 0.0, "grad_norm": 0.020416712388396263, "global_step": 9353, "epoch": 259, "lr": 0.001} +{"train_loss": 0.0432528592646122, "train_loss_bc": 0.0432528592646122, "train_loss_llm": 0.0, "grad_norm": 0.029922286048531532, "global_step": 9354, "epoch": 259, "lr": 0.001} +{"train_loss": 0.039508868008852005, "train_loss_bc": 0.039508868008852005, "train_loss_llm": 0.0, "grad_norm": 0.035914868116378784, "global_step": 9355, "epoch": 259, "lr": 0.001} +{"train_loss": 0.035205692052841187, "train_loss_bc": 0.035205692052841187, "train_loss_llm": 0.0, "grad_norm": 0.04058130085468292, "global_step": 9356, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03844689205288887, "train_loss_bc": 0.03844081610441208, "train_loss_llm": 0.0006077587604522705, "grad_norm": 0.019159747287631035, "global_step": 9357, "epoch": 259, "lr": 0.001} +{"train_loss": 0.04817100614309311, "train_loss_bc": 0.04817100614309311, "train_loss_llm": 0.0, "grad_norm": 0.03030768595635891, "global_step": 9358, "epoch": 259, "lr": 0.001} +{"train_loss": 0.03895262349396944, "train_loss_bc": 0.04192396625876427, "train_loss_llm": 0.0, "grad_norm": 0.04064113274216652, "global_step": 9359, "epoch": 259, "lr": 0.001, "val_loss": 0.06007970869541168} +{"train_loss": 0.04012475162744522, "train_loss_bc": 0.04012475162744522, "train_loss_llm": 0.0, "grad_norm": 0.04506010562181473, "global_step": 9360, "epoch": 260, "lr": 0.001} +{"train_loss": 0.03664834052324295, "train_loss_bc": 0.03664741665124893, "train_loss_llm": 9.232759475708008e-05, "grad_norm": 0.016427936032414436, "global_step": 9361, "epoch": 260, "lr": 0.001} +{"train_loss": 0.04122637212276459, "train_loss_bc": 0.04122637212276459, "train_loss_llm": 0.0, "grad_norm": 0.026271358132362366, "global_step": 9362, "epoch": 260, "lr": 0.001} +{"train_loss": 0.028382018208503723, "train_loss_bc": 0.028373733162879944, "train_loss_llm": 0.000828474760055542, "grad_norm": 0.03740070387721062, "global_step": 9363, "epoch": 260, "lr": 0.001} +{"train_loss": 0.032989345490932465, "train_loss_bc": 0.032989345490932465, "train_loss_llm": 0.0, "grad_norm": 0.044514067471027374, "global_step": 9364, "epoch": 260, "lr": 0.001} +{"train_loss": 0.036797139793634415, "train_loss_bc": 0.036797139793634415, "train_loss_llm": 0.0, "grad_norm": 0.018597060814499855, "global_step": 9365, "epoch": 260, "lr": 0.001} +{"train_loss": 0.046274635940790176, "train_loss_bc": 0.046274635940790176, "train_loss_llm": 0.0, "grad_norm": 0.0323430635035038, "global_step": 9366, "epoch": 260, "lr": 0.001} +{"train_loss": 0.036835215985774994, "train_loss_bc": 0.036835215985774994, "train_loss_llm": 0.0, "grad_norm": 0.04624991863965988, "global_step": 9367, "epoch": 260, "lr": 0.001} +{"train_loss": 0.037881601601839066, "train_loss_bc": 0.037881601601839066, "train_loss_llm": 0.0, "grad_norm": 0.05498143285512924, "global_step": 9368, "epoch": 260, "lr": 0.001} +{"train_loss": 0.0321340374648571, "train_loss_bc": 0.0321340374648571, "train_loss_llm": 0.0, "grad_norm": 0.019134070724248886, "global_step": 9369, "epoch": 260, "lr": 0.001} +{"train_loss": 0.038880180567502975, "train_loss_bc": 0.03883327171206474, "train_loss_llm": 0.004690736532211304, "grad_norm": 0.0278128981590271, "global_step": 9370, "epoch": 260, "lr": 0.001} +{"train_loss": 0.04633614793419838, "train_loss_bc": 0.04633614793419838, "train_loss_llm": 0.0, "grad_norm": 0.03962161764502525, "global_step": 9371, "epoch": 260, "lr": 0.001} +{"train_loss": 0.036924898624420166, "train_loss_bc": 0.03692332282662392, "train_loss_llm": 0.00015756487846374512, "grad_norm": 0.04712236300110817, "global_step": 9372, "epoch": 260, "lr": 0.001} +{"train_loss": 0.034078970551490784, "train_loss_bc": 0.034078970551490784, "train_loss_llm": 0.0, "grad_norm": 0.017684409394860268, "global_step": 9373, "epoch": 260, "lr": 0.001} +{"train_loss": 0.03610115498304367, "train_loss_bc": 0.036097973585128784, "train_loss_llm": 0.0003179609775543213, "grad_norm": 0.03436867892742157, "global_step": 9374, "epoch": 260, "lr": 0.001} +{"train_loss": 0.03489421308040619, "train_loss_bc": 0.03489421308040619, "train_loss_llm": 0.0, "grad_norm": 0.04309907183051109, "global_step": 9375, "epoch": 260, "lr": 0.001} +{"train_loss": 0.0342208556830883, "train_loss_bc": 0.03421896696090698, "train_loss_llm": 0.0001888573169708252, "grad_norm": 0.04882125183939934, "global_step": 9376, "epoch": 260, "lr": 0.001} +{"train_loss": 0.039006397128105164, "train_loss_bc": 0.03900481387972832, "train_loss_llm": 0.0001583099365234375, "grad_norm": 0.01938808336853981, "global_step": 9377, "epoch": 260, "lr": 0.001} +{"train_loss": 0.04579922929406166, "train_loss_bc": 0.04579922929406166, "train_loss_llm": 0.0, "grad_norm": 0.03248082101345062, "global_step": 9378, "epoch": 260, "lr": 0.001} +{"train_loss": 0.03382141515612602, "train_loss_bc": 0.03382141515612602, "train_loss_llm": 0.0, "grad_norm": 0.038510508835315704, "global_step": 9379, "epoch": 260, "lr": 0.001} +{"train_loss": 0.04166445881128311, "train_loss_bc": 0.04165610671043396, "train_loss_llm": 0.0008350908756256104, "grad_norm": 0.044841911643743515, "global_step": 9380, "epoch": 260, "lr": 0.001} +{"train_loss": 0.03592177480459213, "train_loss_bc": 0.03591068089008331, "train_loss_llm": 0.0011092722415924072, "grad_norm": 0.01718217320740223, "global_step": 9381, "epoch": 260, "lr": 0.001} +{"train_loss": 0.04227527976036072, "train_loss_bc": 0.04227527976036072, "train_loss_llm": 0.0, "grad_norm": 0.029893679544329643, "global_step": 9382, "epoch": 260, "lr": 0.001} +{"train_loss": 0.03802686184644699, "train_loss_bc": 0.03802686184644699, "train_loss_llm": 0.0, "grad_norm": 0.03950788080692291, "global_step": 9383, "epoch": 260, "lr": 0.001} +{"train_loss": 0.04130600392818451, "train_loss_bc": 0.04130600392818451, "train_loss_llm": 0.0, "grad_norm": 0.048173170536756516, "global_step": 9384, "epoch": 260, "lr": 0.001} +{"train_loss": 0.04459644481539726, "train_loss_bc": 0.0445956252515316, "train_loss_llm": 8.213520050048828e-05, "grad_norm": 0.023471886292099953, "global_step": 9385, "epoch": 260, "lr": 0.001} +{"train_loss": 0.04534120112657547, "train_loss_bc": 0.04534120112657547, "train_loss_llm": 0.0, "grad_norm": 0.0390508696436882, "global_step": 9386, "epoch": 260, "lr": 0.001} +{"train_loss": 0.03897172957658768, "train_loss_bc": 0.03897172957658768, "train_loss_llm": 0.0, "grad_norm": 0.0508817583322525, "global_step": 9387, "epoch": 260, "lr": 0.001} +{"train_loss": 0.04303765669465065, "train_loss_bc": 0.04303765669465065, "train_loss_llm": 0.0, "grad_norm": 0.06456306576728821, "global_step": 9388, "epoch": 260, "lr": 0.001} +{"train_loss": 0.040966615080833435, "train_loss_bc": 0.040966615080833435, "train_loss_llm": 0.0, "grad_norm": 0.025042150169610977, "global_step": 9389, "epoch": 260, "lr": 0.001} +{"train_loss": 0.038757964968681335, "train_loss_bc": 0.038757964968681335, "train_loss_llm": 0.0, "grad_norm": 0.03286198899149895, "global_step": 9390, "epoch": 260, "lr": 0.001} +{"train_loss": 0.0423867329955101, "train_loss_bc": 0.0423867329955101, "train_loss_llm": 0.0, "grad_norm": 0.04521007090806961, "global_step": 9391, "epoch": 260, "lr": 0.001} +{"train_loss": 0.03199345991015434, "train_loss_bc": 0.031985942274332047, "train_loss_llm": 0.0007519423961639404, "grad_norm": 0.06002860143780708, "global_step": 9392, "epoch": 260, "lr": 0.001} +{"train_loss": 0.04586957395076752, "train_loss_bc": 0.04586957395076752, "train_loss_llm": 0.0, "grad_norm": 0.024388398975133896, "global_step": 9393, "epoch": 260, "lr": 0.001} +{"train_loss": 0.03828122094273567, "train_loss_bc": 0.03828122094273567, "train_loss_llm": 0.0, "grad_norm": 0.030401039868593216, "global_step": 9394, "epoch": 260, "lr": 0.001} +{"train_loss": 0.039006156743400626, "train_loss_bc": 0.045467741787433624, "train_loss_llm": 0.0, "grad_norm": 0.039910513907670975, "global_step": 9395, "epoch": 260, "lr": 0.001, "train/cumulative_reward": 132.95916147354973, "train/mean_score": 14.826377168970737, "train/success_rate": 0.3, "test/cumulative_reward": 87.7943507605394, "test/mean_score": 10.190824870670308, "test/success_rate": 0.16, "val_loss": 0.059952739626169205, "train_action_mse_error": 0.04272562265396118} +{"train_loss": 0.04247391223907471, "train_loss_bc": 0.04247391223907471, "train_loss_llm": 0.0, "grad_norm": 0.046169914305210114, "global_step": 9396, "epoch": 261, "lr": 0.001} +{"train_loss": 0.038448773324489594, "train_loss_bc": 0.038448773324489594, "train_loss_llm": 0.0, "grad_norm": 0.016705958172678947, "global_step": 9397, "epoch": 261, "lr": 0.001} +{"train_loss": 0.04256918653845787, "train_loss_bc": 0.04256918653845787, "train_loss_llm": 0.0, "grad_norm": 0.02363215573132038, "global_step": 9398, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03982314467430115, "train_loss_bc": 0.039818137884140015, "train_loss_llm": 0.0005005598068237305, "grad_norm": 0.03688845783472061, "global_step": 9399, "epoch": 261, "lr": 0.001} +{"train_loss": 0.04579577594995499, "train_loss_bc": 0.04579577594995499, "train_loss_llm": 0.0, "grad_norm": 0.05215498432517052, "global_step": 9400, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03701939061284065, "train_loss_bc": 0.03699715808033943, "train_loss_llm": 0.002223074436187744, "grad_norm": 0.018806183710694313, "global_step": 9401, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03820030763745308, "train_loss_bc": 0.03820030763745308, "train_loss_llm": 0.0, "grad_norm": 0.0284804068505764, "global_step": 9402, "epoch": 261, "lr": 0.001} +{"train_loss": 0.035770233720541, "train_loss_bc": 0.035770233720541, "train_loss_llm": 0.0, "grad_norm": 0.03670363500714302, "global_step": 9403, "epoch": 261, "lr": 0.001} +{"train_loss": 0.042505450546741486, "train_loss_bc": 0.042486026883125305, "train_loss_llm": 0.0019425153732299805, "grad_norm": 0.04948760196566582, "global_step": 9404, "epoch": 261, "lr": 0.001} +{"train_loss": 0.033656708896160126, "train_loss_bc": 0.033656708896160126, "train_loss_llm": 0.0, "grad_norm": 0.01653338596224785, "global_step": 9405, "epoch": 261, "lr": 0.001} +{"train_loss": 0.035217057913541794, "train_loss_bc": 0.035217057913541794, "train_loss_llm": 0.0, "grad_norm": 0.020354000851511955, "global_step": 9406, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03738276660442352, "train_loss_bc": 0.037357717752456665, "train_loss_llm": 0.002505004405975342, "grad_norm": 0.027323301881551743, "global_step": 9407, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03868739679455757, "train_loss_bc": 0.03868739679455757, "train_loss_llm": 0.0, "grad_norm": 0.0339692085981369, "global_step": 9408, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03362399339675903, "train_loss_bc": 0.03362296521663666, "train_loss_llm": 0.00010278820991516113, "grad_norm": 0.015286723151803017, "global_step": 9409, "epoch": 261, "lr": 0.001} +{"train_loss": 0.044437553733587265, "train_loss_bc": 0.044437553733587265, "train_loss_llm": 0.0, "grad_norm": 0.027627218514680862, "global_step": 9410, "epoch": 261, "lr": 0.001} +{"train_loss": 0.037291862070560455, "train_loss_bc": 0.037291862070560455, "train_loss_llm": 0.0, "grad_norm": 0.040676653385162354, "global_step": 9411, "epoch": 261, "lr": 0.001} +{"train_loss": 0.042676959186792374, "train_loss_bc": 0.042676959186792374, "train_loss_llm": 0.0, "grad_norm": 0.0498451367020607, "global_step": 9412, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03983144089579582, "train_loss_bc": 0.03983144089579582, "train_loss_llm": 0.0, "grad_norm": 0.028148695826530457, "global_step": 9413, "epoch": 261, "lr": 0.001} +{"train_loss": 0.036525558680295944, "train_loss_bc": 0.03651326522231102, "train_loss_llm": 0.0012292861938476562, "grad_norm": 0.02944992296397686, "global_step": 9414, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03215388208627701, "train_loss_bc": 0.03215388208627701, "train_loss_llm": 0.0, "grad_norm": 0.037542421370744705, "global_step": 9415, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03737279772758484, "train_loss_bc": 0.03737279772758484, "train_loss_llm": 0.0, "grad_norm": 0.04334608465433121, "global_step": 9416, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03746431693434715, "train_loss_bc": 0.03746431693434715, "train_loss_llm": 0.0, "grad_norm": 0.02439849078655243, "global_step": 9417, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03144793584942818, "train_loss_bc": 0.03144793584942818, "train_loss_llm": 0.0, "grad_norm": 0.03454919904470444, "global_step": 9418, "epoch": 261, "lr": 0.001} +{"train_loss": 0.04017087072134018, "train_loss_bc": 0.04017087072134018, "train_loss_llm": 0.0, "grad_norm": 0.04113176092505455, "global_step": 9419, "epoch": 261, "lr": 0.001} +{"train_loss": 0.038407906889915466, "train_loss_bc": 0.038407906889915466, "train_loss_llm": 0.0, "grad_norm": 0.05671308934688568, "global_step": 9420, "epoch": 261, "lr": 0.001} +{"train_loss": 0.0468643419444561, "train_loss_bc": 0.0468643419444561, "train_loss_llm": 0.0, "grad_norm": 0.017810901626944542, "global_step": 9421, "epoch": 261, "lr": 0.001} +{"train_loss": 0.04523060470819473, "train_loss_bc": 0.04523060470819473, "train_loss_llm": 0.0, "grad_norm": 0.03264031186699867, "global_step": 9422, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03405757248401642, "train_loss_bc": 0.03405757248401642, "train_loss_llm": 0.0, "grad_norm": 0.03782421350479126, "global_step": 9423, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03759625926613808, "train_loss_bc": 0.03759625926613808, "train_loss_llm": 0.0, "grad_norm": 0.0482894666492939, "global_step": 9424, "epoch": 261, "lr": 0.001} +{"train_loss": 0.0372646264731884, "train_loss_bc": 0.0372646264731884, "train_loss_llm": 0.0, "grad_norm": 0.018400119617581367, "global_step": 9425, "epoch": 261, "lr": 0.001} +{"train_loss": 0.04249239340424538, "train_loss_bc": 0.04249239340424538, "train_loss_llm": 0.0, "grad_norm": 0.02883395552635193, "global_step": 9426, "epoch": 261, "lr": 0.001} +{"train_loss": 0.038919251412153244, "train_loss_bc": 0.038919251412153244, "train_loss_llm": 0.0, "grad_norm": 0.03478822857141495, "global_step": 9427, "epoch": 261, "lr": 0.001} +{"train_loss": 0.04047466069459915, "train_loss_bc": 0.040465135127305984, "train_loss_llm": 0.0009525418281555176, "grad_norm": 0.04359956830739975, "global_step": 9428, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03345871716737747, "train_loss_bc": 0.03345780074596405, "train_loss_llm": 9.170174598693848e-05, "grad_norm": 0.019056566059589386, "global_step": 9429, "epoch": 261, "lr": 0.001} +{"train_loss": 0.034558914601802826, "train_loss_bc": 0.034558914601802826, "train_loss_llm": 0.0, "grad_norm": 0.031012754887342453, "global_step": 9430, "epoch": 261, "lr": 0.001} +{"train_loss": 0.03878556533406178, "train_loss_bc": 0.046407826244831085, "train_loss_llm": 0.0, "grad_norm": 0.043530892580747604, "global_step": 9431, "epoch": 261, "lr": 0.001, "val_loss": 0.06001100689172745} +{"train_loss": 0.03661404177546501, "train_loss_bc": 0.03661404177546501, "train_loss_llm": 0.0, "grad_norm": 0.04727688431739807, "global_step": 9432, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03572061285376549, "train_loss_bc": 0.03572061285376549, "train_loss_llm": 0.0, "grad_norm": 0.019967136904597282, "global_step": 9433, "epoch": 262, "lr": 0.001} +{"train_loss": 0.04505174607038498, "train_loss_bc": 0.04505174607038498, "train_loss_llm": 0.0, "grad_norm": 0.02855100855231285, "global_step": 9434, "epoch": 262, "lr": 0.001} +{"train_loss": 0.034488268196582794, "train_loss_bc": 0.034488268196582794, "train_loss_llm": 0.0, "grad_norm": 0.04138046130537987, "global_step": 9435, "epoch": 262, "lr": 0.001} +{"train_loss": 0.037764132022857666, "train_loss_bc": 0.037763167172670364, "train_loss_llm": 9.647011756896973e-05, "grad_norm": 0.053087685257196426, "global_step": 9436, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03611214458942413, "train_loss_bc": 0.03611214458942413, "train_loss_llm": 0.0, "grad_norm": 0.017273424193263054, "global_step": 9437, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03914637491106987, "train_loss_bc": 0.03914637491106987, "train_loss_llm": 0.0, "grad_norm": 0.025394603610038757, "global_step": 9438, "epoch": 262, "lr": 0.001} +{"train_loss": 0.05015161260962486, "train_loss_bc": 0.050132691860198975, "train_loss_llm": 0.0018921494483947754, "grad_norm": 0.034574493765830994, "global_step": 9439, "epoch": 262, "lr": 0.001} +{"train_loss": 0.0402912013232708, "train_loss_bc": 0.0402657613158226, "train_loss_llm": 0.0025440752506256104, "grad_norm": 0.040177468210458755, "global_step": 9440, "epoch": 262, "lr": 0.001} +{"train_loss": 0.036561738699674606, "train_loss_bc": 0.036554593592882156, "train_loss_llm": 0.0007145404815673828, "grad_norm": 0.02114398032426834, "global_step": 9441, "epoch": 262, "lr": 0.001} +{"train_loss": 0.045547567307949066, "train_loss_bc": 0.045547567307949066, "train_loss_llm": 0.0, "grad_norm": 0.037213798612356186, "global_step": 9442, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03612494841217995, "train_loss_bc": 0.03612494841217995, "train_loss_llm": 0.0, "grad_norm": 0.04558150842785835, "global_step": 9443, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03288067877292633, "train_loss_bc": 0.03286247327923775, "train_loss_llm": 0.0018205344676971436, "grad_norm": 0.054743826389312744, "global_step": 9444, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03730001673102379, "train_loss_bc": 0.037293482571840286, "train_loss_llm": 0.0006534159183502197, "grad_norm": 0.01994585432112217, "global_step": 9445, "epoch": 262, "lr": 0.001} +{"train_loss": 0.035697970539331436, "train_loss_bc": 0.035647016018629074, "train_loss_llm": 0.005095303058624268, "grad_norm": 0.028045838698744774, "global_step": 9446, "epoch": 262, "lr": 0.001} +{"train_loss": 0.042986441403627396, "train_loss_bc": 0.042986441403627396, "train_loss_llm": 0.0, "grad_norm": 0.035678308457136154, "global_step": 9447, "epoch": 262, "lr": 0.001} +{"train_loss": 0.0347403958439827, "train_loss_bc": 0.0347403958439827, "train_loss_llm": 0.0, "grad_norm": 0.03930545598268509, "global_step": 9448, "epoch": 262, "lr": 0.001} +{"train_loss": 0.04587360471487045, "train_loss_bc": 0.04586825519800186, "train_loss_llm": 0.0005348324775695801, "grad_norm": 0.020805243402719498, "global_step": 9449, "epoch": 262, "lr": 0.001} +{"train_loss": 0.0360635407269001, "train_loss_bc": 0.03604971989989281, "train_loss_llm": 0.0013821721076965332, "grad_norm": 0.025553598999977112, "global_step": 9450, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03359963372349739, "train_loss_bc": 0.03359963372349739, "train_loss_llm": 0.0, "grad_norm": 0.030570870265364647, "global_step": 9451, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03953493759036064, "train_loss_bc": 0.03953493759036064, "train_loss_llm": 0.0, "grad_norm": 0.03770391643047333, "global_step": 9452, "epoch": 262, "lr": 0.001} +{"train_loss": 0.024911079555749893, "train_loss_bc": 0.02490929327905178, "train_loss_llm": 0.0001786649227142334, "grad_norm": 0.017343543469905853, "global_step": 9453, "epoch": 262, "lr": 0.001} +{"train_loss": 0.037491340190172195, "train_loss_bc": 0.037491340190172195, "train_loss_llm": 0.0, "grad_norm": 0.023032722994685173, "global_step": 9454, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03336929902434349, "train_loss_bc": 0.03336929902434349, "train_loss_llm": 0.0, "grad_norm": 0.029161697253584862, "global_step": 9455, "epoch": 262, "lr": 0.001} +{"train_loss": 0.0313042588531971, "train_loss_bc": 0.03129658102989197, "train_loss_llm": 0.0007679462432861328, "grad_norm": 0.03572418540716171, "global_step": 9456, "epoch": 262, "lr": 0.001} +{"train_loss": 0.046779680997133255, "train_loss_bc": 0.046779680997133255, "train_loss_llm": 0.0, "grad_norm": 0.01806531846523285, "global_step": 9457, "epoch": 262, "lr": 0.001} +{"train_loss": 0.043199338018894196, "train_loss_bc": 0.0431840755045414, "train_loss_llm": 0.0015263557434082031, "grad_norm": 0.026652362197637558, "global_step": 9458, "epoch": 262, "lr": 0.001} +{"train_loss": 0.04515523836016655, "train_loss_bc": 0.04515523836016655, "train_loss_llm": 0.0, "grad_norm": 0.033608946949243546, "global_step": 9459, "epoch": 262, "lr": 0.001} +{"train_loss": 0.037911687046289444, "train_loss_bc": 0.037911687046289444, "train_loss_llm": 0.0, "grad_norm": 0.04632602632045746, "global_step": 9460, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03931855410337448, "train_loss_bc": 0.03931855410337448, "train_loss_llm": 0.0, "grad_norm": 0.019204579293727875, "global_step": 9461, "epoch": 262, "lr": 0.001} +{"train_loss": 0.04420122131705284, "train_loss_bc": 0.04420122131705284, "train_loss_llm": 0.0, "grad_norm": 0.029798587784171104, "global_step": 9462, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03518364578485489, "train_loss_bc": 0.03518364578485489, "train_loss_llm": 0.0, "grad_norm": 0.03626362979412079, "global_step": 9463, "epoch": 262, "lr": 0.001} +{"train_loss": 0.043366704136133194, "train_loss_bc": 0.043352868407964706, "train_loss_llm": 0.001383453607559204, "grad_norm": 0.04236459732055664, "global_step": 9464, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03760373964905739, "train_loss_bc": 0.03760373964905739, "train_loss_llm": 0.0, "grad_norm": 0.019672973081469536, "global_step": 9465, "epoch": 262, "lr": 0.001} +{"train_loss": 0.044348299503326416, "train_loss_bc": 0.044348299503326416, "train_loss_llm": 0.0, "grad_norm": 0.03430119901895523, "global_step": 9466, "epoch": 262, "lr": 0.001} +{"train_loss": 0.03866936959740189, "train_loss_bc": 0.03568236157298088, "train_loss_llm": 0.001924753189086914, "grad_norm": 0.03850114345550537, "global_step": 9467, "epoch": 262, "lr": 0.001, "val_loss": 0.059295982122421265} +{"train_loss": 0.039751969277858734, "train_loss_bc": 0.039751969277858734, "train_loss_llm": 0.0, "grad_norm": 0.04919128492474556, "global_step": 9468, "epoch": 263, "lr": 0.001} +{"train_loss": 0.034917932003736496, "train_loss_bc": 0.034917932003736496, "train_loss_llm": 0.0, "grad_norm": 0.022512223571538925, "global_step": 9469, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03243391215801239, "train_loss_bc": 0.03243391215801239, "train_loss_llm": 0.0, "grad_norm": 0.028299076482653618, "global_step": 9470, "epoch": 263, "lr": 0.001} +{"train_loss": 0.04428860545158386, "train_loss_bc": 0.04428860545158386, "train_loss_llm": 0.0, "grad_norm": 0.04243634268641472, "global_step": 9471, "epoch": 263, "lr": 0.001} +{"train_loss": 0.039786890149116516, "train_loss_bc": 0.03976196423172951, "train_loss_llm": 0.002492755651473999, "grad_norm": 0.04746158793568611, "global_step": 9472, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03378972411155701, "train_loss_bc": 0.03378972411155701, "train_loss_llm": 0.0, "grad_norm": 0.018343841657042503, "global_step": 9473, "epoch": 263, "lr": 0.001} +{"train_loss": 0.039683155715465546, "train_loss_bc": 0.039683155715465546, "train_loss_llm": 0.0, "grad_norm": 0.030717045068740845, "global_step": 9474, "epoch": 263, "lr": 0.001} +{"train_loss": 0.037933237850666046, "train_loss_bc": 0.037933237850666046, "train_loss_llm": 0.0, "grad_norm": 0.04613213613629341, "global_step": 9475, "epoch": 263, "lr": 0.001} +{"train_loss": 0.039323821663856506, "train_loss_bc": 0.03932245075702667, "train_loss_llm": 0.00013691186904907227, "grad_norm": 0.05232081562280655, "global_step": 9476, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03982362896203995, "train_loss_bc": 0.03982362896203995, "train_loss_llm": 0.0, "grad_norm": 0.018084336072206497, "global_step": 9477, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03887057676911354, "train_loss_bc": 0.03887057676911354, "train_loss_llm": 0.0, "grad_norm": 0.026012301445007324, "global_step": 9478, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03182327374815941, "train_loss_bc": 0.031818900257349014, "train_loss_llm": 0.00043720006942749023, "grad_norm": 0.03228321298956871, "global_step": 9479, "epoch": 263, "lr": 0.001} +{"train_loss": 0.040144868195056915, "train_loss_bc": 0.040144868195056915, "train_loss_llm": 0.0, "grad_norm": 0.04384803771972656, "global_step": 9480, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03911224380135536, "train_loss_bc": 0.03911224380135536, "train_loss_llm": 0.0, "grad_norm": 0.019512223079800606, "global_step": 9481, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03990810737013817, "train_loss_bc": 0.03990810737013817, "train_loss_llm": 0.0, "grad_norm": 0.035684000700712204, "global_step": 9482, "epoch": 263, "lr": 0.001} +{"train_loss": 0.042523376643657684, "train_loss_bc": 0.04252150282263756, "train_loss_llm": 0.00018733739852905273, "grad_norm": 0.04594971612095833, "global_step": 9483, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03555426001548767, "train_loss_bc": 0.03555426001548767, "train_loss_llm": 0.0, "grad_norm": 0.058664388954639435, "global_step": 9484, "epoch": 263, "lr": 0.001} +{"train_loss": 0.045803435146808624, "train_loss_bc": 0.045803435146808624, "train_loss_llm": 0.0, "grad_norm": 0.017386050894856453, "global_step": 9485, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03827427700161934, "train_loss_bc": 0.03825993835926056, "train_loss_llm": 0.001433730125427246, "grad_norm": 0.02839159220457077, "global_step": 9486, "epoch": 263, "lr": 0.001} +{"train_loss": 0.035526491701602936, "train_loss_bc": 0.035514943301677704, "train_loss_llm": 0.0011546611785888672, "grad_norm": 0.03362547606229782, "global_step": 9487, "epoch": 263, "lr": 0.001} +{"train_loss": 0.039387594908475876, "train_loss_bc": 0.039387594908475876, "train_loss_llm": 0.0, "grad_norm": 0.03819112852215767, "global_step": 9488, "epoch": 263, "lr": 0.001} +{"train_loss": 0.04018646106123924, "train_loss_bc": 0.04018646106123924, "train_loss_llm": 0.0, "grad_norm": 0.024216484278440475, "global_step": 9489, "epoch": 263, "lr": 0.001} +{"train_loss": 0.0394083596765995, "train_loss_bc": 0.0394083596765995, "train_loss_llm": 0.0, "grad_norm": 0.038529377430677414, "global_step": 9490, "epoch": 263, "lr": 0.001} +{"train_loss": 0.04525810107588768, "train_loss_bc": 0.04525810107588768, "train_loss_llm": 0.0, "grad_norm": 0.052723195403814316, "global_step": 9491, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03442268446087837, "train_loss_bc": 0.03442268446087837, "train_loss_llm": 0.0, "grad_norm": 0.06089324876666069, "global_step": 9492, "epoch": 263, "lr": 0.001} +{"train_loss": 0.042198702692985535, "train_loss_bc": 0.042198702692985535, "train_loss_llm": 0.0, "grad_norm": 0.02101425640285015, "global_step": 9493, "epoch": 263, "lr": 0.001} +{"train_loss": 0.036785997450351715, "train_loss_bc": 0.036785997450351715, "train_loss_llm": 0.0, "grad_norm": 0.03336675092577934, "global_step": 9494, "epoch": 263, "lr": 0.001} +{"train_loss": 0.036994364112615585, "train_loss_bc": 0.036994364112615585, "train_loss_llm": 0.0, "grad_norm": 0.04320495203137398, "global_step": 9495, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03371758759021759, "train_loss_bc": 0.03371758759021759, "train_loss_llm": 0.0, "grad_norm": 0.048133350908756256, "global_step": 9496, "epoch": 263, "lr": 0.001} +{"train_loss": 0.04077007994055748, "train_loss_bc": 0.04077007994055748, "train_loss_llm": 0.0, "grad_norm": 0.020022768527269363, "global_step": 9497, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03727482631802559, "train_loss_bc": 0.03727482631802559, "train_loss_llm": 0.0, "grad_norm": 0.032651860266923904, "global_step": 9498, "epoch": 263, "lr": 0.001} +{"train_loss": 0.037169646471738815, "train_loss_bc": 0.037169646471738815, "train_loss_llm": 0.0, "grad_norm": 0.03561202436685562, "global_step": 9499, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03902173787355423, "train_loss_bc": 0.03902173787355423, "train_loss_llm": 0.0, "grad_norm": 0.04981616139411926, "global_step": 9500, "epoch": 263, "lr": 0.001} +{"train_loss": 0.039767421782016754, "train_loss_bc": 0.039757270365953445, "train_loss_llm": 0.0010150671005249023, "grad_norm": 0.01948908530175686, "global_step": 9501, "epoch": 263, "lr": 0.001} +{"train_loss": 0.04101911559700966, "train_loss_bc": 0.040989603847265244, "train_loss_llm": 0.0029510855674743652, "grad_norm": 0.03183240815997124, "global_step": 9502, "epoch": 263, "lr": 0.001} +{"train_loss": 0.03878135989523596, "train_loss_bc": 0.04347248747944832, "train_loss_llm": 0.0, "grad_norm": 0.04671335592865944, "global_step": 9503, "epoch": 263, "lr": 0.001, "val_loss": 0.059964947402477264} +{"train_loss": 0.03898553177714348, "train_loss_bc": 0.038981568068265915, "train_loss_llm": 0.00039634108543395996, "grad_norm": 0.05449434369802475, "global_step": 9504, "epoch": 264, "lr": 0.001} +{"train_loss": 0.032192155718803406, "train_loss_bc": 0.03216493874788284, "train_loss_llm": 0.0027217864990234375, "grad_norm": 0.019015921279788017, "global_step": 9505, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03539062663912773, "train_loss_bc": 0.03539062663912773, "train_loss_llm": 0.0, "grad_norm": 0.03146325796842575, "global_step": 9506, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03788677975535393, "train_loss_bc": 0.03788677975535393, "train_loss_llm": 0.0, "grad_norm": 0.04196760058403015, "global_step": 9507, "epoch": 264, "lr": 0.001} +{"train_loss": 0.028430534526705742, "train_loss_bc": 0.028411442413926125, "train_loss_llm": 0.0019092559814453125, "grad_norm": 0.04725485295057297, "global_step": 9508, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03939512372016907, "train_loss_bc": 0.03939512372016907, "train_loss_llm": 0.0, "grad_norm": 0.02212810516357422, "global_step": 9509, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03455523028969765, "train_loss_bc": 0.03455523028969765, "train_loss_llm": 0.0, "grad_norm": 0.03158341348171234, "global_step": 9510, "epoch": 264, "lr": 0.001} +{"train_loss": 0.042276781052351, "train_loss_bc": 0.042276781052351, "train_loss_llm": 0.0, "grad_norm": 0.04098254442214966, "global_step": 9511, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03923429921269417, "train_loss_bc": 0.03923429921269417, "train_loss_llm": 0.0, "grad_norm": 0.05394723266363144, "global_step": 9512, "epoch": 264, "lr": 0.001} +{"train_loss": 0.038050923496484756, "train_loss_bc": 0.038050923496484756, "train_loss_llm": 0.0, "grad_norm": 0.016719220206141472, "global_step": 9513, "epoch": 264, "lr": 0.001} +{"train_loss": 0.030719071626663208, "train_loss_bc": 0.03071616217494011, "train_loss_llm": 0.00029098987579345703, "grad_norm": 0.02329971082508564, "global_step": 9514, "epoch": 264, "lr": 0.001} +{"train_loss": 0.031942348927259445, "train_loss_bc": 0.031942348927259445, "train_loss_llm": 0.0, "grad_norm": 0.03213535621762276, "global_step": 9515, "epoch": 264, "lr": 0.001} +{"train_loss": 0.040706273168325424, "train_loss_bc": 0.04069504514336586, "train_loss_llm": 0.0011226534843444824, "grad_norm": 0.04092030972242355, "global_step": 9516, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03922915458679199, "train_loss_bc": 0.03922012820839882, "train_loss_llm": 0.0009025633335113525, "grad_norm": 0.020310869440436363, "global_step": 9517, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03741614893078804, "train_loss_bc": 0.03741614893078804, "train_loss_llm": 0.0, "grad_norm": 0.02345106191933155, "global_step": 9518, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03664907440543175, "train_loss_bc": 0.036630112677812576, "train_loss_llm": 0.0018962323665618896, "grad_norm": 0.028921598568558693, "global_step": 9519, "epoch": 264, "lr": 0.001} +{"train_loss": 0.043305281549692154, "train_loss_bc": 0.043305281549692154, "train_loss_llm": 0.0, "grad_norm": 0.03858635947108269, "global_step": 9520, "epoch": 264, "lr": 0.001} +{"train_loss": 0.0398753397166729, "train_loss_bc": 0.0398692786693573, "train_loss_llm": 0.0006059408187866211, "grad_norm": 0.016748452559113503, "global_step": 9521, "epoch": 264, "lr": 0.001} +{"train_loss": 0.04071185365319252, "train_loss_bc": 0.04068216681480408, "train_loss_llm": 0.0029688477516174316, "grad_norm": 0.022768966853618622, "global_step": 9522, "epoch": 264, "lr": 0.001} +{"train_loss": 0.041058413684368134, "train_loss_bc": 0.041058413684368134, "train_loss_llm": 0.0, "grad_norm": 0.03058556094765663, "global_step": 9523, "epoch": 264, "lr": 0.001} +{"train_loss": 0.0440404899418354, "train_loss_bc": 0.0440404899418354, "train_loss_llm": 0.0, "grad_norm": 0.042824454605579376, "global_step": 9524, "epoch": 264, "lr": 0.001} +{"train_loss": 0.036147166043519974, "train_loss_bc": 0.03613612800836563, "train_loss_llm": 0.0011038482189178467, "grad_norm": 0.016517361626029015, "global_step": 9525, "epoch": 264, "lr": 0.001} +{"train_loss": 0.032353904098272324, "train_loss_bc": 0.032353904098272324, "train_loss_llm": 0.0, "grad_norm": 0.025501634925603867, "global_step": 9526, "epoch": 264, "lr": 0.001} +{"train_loss": 0.04490851238369942, "train_loss_bc": 0.04490851238369942, "train_loss_llm": 0.0, "grad_norm": 0.03593682125210762, "global_step": 9527, "epoch": 264, "lr": 0.001} +{"train_loss": 0.040573179721832275, "train_loss_bc": 0.040573179721832275, "train_loss_llm": 0.0, "grad_norm": 0.05151812732219696, "global_step": 9528, "epoch": 264, "lr": 0.001} +{"train_loss": 0.034831222146749496, "train_loss_bc": 0.03481677547097206, "train_loss_llm": 0.0014445185661315918, "grad_norm": 0.019220024347305298, "global_step": 9529, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03904649615287781, "train_loss_bc": 0.03904649615287781, "train_loss_llm": 0.0, "grad_norm": 0.03435742110013962, "global_step": 9530, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03610696271061897, "train_loss_bc": 0.03610696271061897, "train_loss_llm": 0.0, "grad_norm": 0.042891666293144226, "global_step": 9531, "epoch": 264, "lr": 0.001} +{"train_loss": 0.0408870093524456, "train_loss_bc": 0.0408870093524456, "train_loss_llm": 0.0, "grad_norm": 0.04900975897908211, "global_step": 9532, "epoch": 264, "lr": 0.001} +{"train_loss": 0.04418065398931503, "train_loss_bc": 0.044172078371047974, "train_loss_llm": 0.0008574426174163818, "grad_norm": 0.01921544410288334, "global_step": 9533, "epoch": 264, "lr": 0.001} +{"train_loss": 0.038611866533756256, "train_loss_bc": 0.038611866533756256, "train_loss_llm": 0.0, "grad_norm": 0.029428288340568542, "global_step": 9534, "epoch": 264, "lr": 0.001} +{"train_loss": 0.038779087364673615, "train_loss_bc": 0.038779087364673615, "train_loss_llm": 0.0, "grad_norm": 0.03566677123308182, "global_step": 9535, "epoch": 264, "lr": 0.001} +{"train_loss": 0.04455302655696869, "train_loss_bc": 0.04455302655696869, "train_loss_llm": 0.0, "grad_norm": 0.049865517765283585, "global_step": 9536, "epoch": 264, "lr": 0.001} +{"train_loss": 0.04188942909240723, "train_loss_bc": 0.04188942909240723, "train_loss_llm": 0.0, "grad_norm": 0.01916988007724285, "global_step": 9537, "epoch": 264, "lr": 0.001} +{"train_loss": 0.04746824502944946, "train_loss_bc": 0.04746824502944946, "train_loss_llm": 0.0, "grad_norm": 0.037314124405384064, "global_step": 9538, "epoch": 264, "lr": 0.001} +{"train_loss": 0.03870798393877016, "train_loss_bc": 0.041099224239587784, "train_loss_llm": 0.0, "grad_norm": 0.04664790257811546, "global_step": 9539, "epoch": 264, "lr": 0.001, "val_loss": 0.06012294068932533} +{"train_loss": 0.04155513271689415, "train_loss_bc": 0.04155513271689415, "train_loss_llm": 0.0, "grad_norm": 0.057134468108415604, "global_step": 9540, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03439878672361374, "train_loss_bc": 0.03439878672361374, "train_loss_llm": 0.0, "grad_norm": 0.019889909774065018, "global_step": 9541, "epoch": 265, "lr": 0.001} +{"train_loss": 0.04118810221552849, "train_loss_bc": 0.041185181587934494, "train_loss_llm": 0.0002919435501098633, "grad_norm": 0.03628045693039894, "global_step": 9542, "epoch": 265, "lr": 0.001} +{"train_loss": 0.037901975214481354, "train_loss_bc": 0.037901975214481354, "train_loss_llm": 0.0, "grad_norm": 0.04999736696481705, "global_step": 9543, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03971711918711662, "train_loss_bc": 0.03971711918711662, "train_loss_llm": 0.0, "grad_norm": 0.05973896384239197, "global_step": 9544, "epoch": 265, "lr": 0.001} +{"train_loss": 0.036956895142793655, "train_loss_bc": 0.036951933056116104, "train_loss_llm": 0.0004960596561431885, "grad_norm": 0.0208860132843256, "global_step": 9545, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03592934086918831, "train_loss_bc": 0.03592934086918831, "train_loss_llm": 0.0, "grad_norm": 0.02949276939034462, "global_step": 9546, "epoch": 265, "lr": 0.001} +{"train_loss": 0.04361063614487648, "train_loss_bc": 0.04361063614487648, "train_loss_llm": 0.0, "grad_norm": 0.037333156913518906, "global_step": 9547, "epoch": 265, "lr": 0.001} +{"train_loss": 0.033923279494047165, "train_loss_bc": 0.033923279494047165, "train_loss_llm": 0.0, "grad_norm": 0.0485338419675827, "global_step": 9548, "epoch": 265, "lr": 0.001} +{"train_loss": 0.039583731442689896, "train_loss_bc": 0.039583731442689896, "train_loss_llm": 0.0, "grad_norm": 0.018821028992533684, "global_step": 9549, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03193294629454613, "train_loss_bc": 0.03192860633134842, "train_loss_llm": 0.0004340708255767822, "grad_norm": 0.030856886878609657, "global_step": 9550, "epoch": 265, "lr": 0.001} +{"train_loss": 0.04307641088962555, "train_loss_bc": 0.04307641088962555, "train_loss_llm": 0.0, "grad_norm": 0.05693428963422775, "global_step": 9551, "epoch": 265, "lr": 0.001} +{"train_loss": 0.039289701730012894, "train_loss_bc": 0.03928703814744949, "train_loss_llm": 0.000266343355178833, "grad_norm": 0.06651327759027481, "global_step": 9552, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03621537983417511, "train_loss_bc": 0.03621537983417511, "train_loss_llm": 0.0, "grad_norm": 0.019707689061760902, "global_step": 9553, "epoch": 265, "lr": 0.001} +{"train_loss": 0.04029035568237305, "train_loss_bc": 0.04028887301683426, "train_loss_llm": 0.00014832615852355957, "grad_norm": 0.03032342717051506, "global_step": 9554, "epoch": 265, "lr": 0.001} +{"train_loss": 0.035184942185878754, "train_loss_bc": 0.03514904901385307, "train_loss_llm": 0.003589421510696411, "grad_norm": 0.04211793467402458, "global_step": 9555, "epoch": 265, "lr": 0.001} +{"train_loss": 0.039239685982465744, "train_loss_bc": 0.039239685982465744, "train_loss_llm": 0.0, "grad_norm": 0.046219710260629654, "global_step": 9556, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03033086657524109, "train_loss_bc": 0.03033086657524109, "train_loss_llm": 0.0, "grad_norm": 0.01762331835925579, "global_step": 9557, "epoch": 265, "lr": 0.001} +{"train_loss": 0.038120467215776443, "train_loss_bc": 0.038120467215776443, "train_loss_llm": 0.0, "grad_norm": 0.0302982646971941, "global_step": 9558, "epoch": 265, "lr": 0.001} +{"train_loss": 0.039624277502298355, "train_loss_bc": 0.03960680589079857, "train_loss_llm": 0.0017469823360443115, "grad_norm": 0.04238681122660637, "global_step": 9559, "epoch": 265, "lr": 0.001} +{"train_loss": 0.04123712703585625, "train_loss_bc": 0.04123712703585625, "train_loss_llm": 0.0, "grad_norm": 0.0454142726957798, "global_step": 9560, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03898954764008522, "train_loss_bc": 0.03898954764008522, "train_loss_llm": 0.0, "grad_norm": 0.02226325124502182, "global_step": 9561, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03419461101293564, "train_loss_bc": 0.03417836129665375, "train_loss_llm": 0.0016250312328338623, "grad_norm": 0.030469682067632675, "global_step": 9562, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03556939959526062, "train_loss_bc": 0.03556939959526062, "train_loss_llm": 0.0, "grad_norm": 0.035899125039577484, "global_step": 9563, "epoch": 265, "lr": 0.001} +{"train_loss": 0.04498846083879471, "train_loss_bc": 0.04498201608657837, "train_loss_llm": 0.0006444752216339111, "grad_norm": 0.05114378407597542, "global_step": 9564, "epoch": 265, "lr": 0.001} +{"train_loss": 0.04209307208657265, "train_loss_bc": 0.042058318853378296, "train_loss_llm": 0.003475487232208252, "grad_norm": 0.020636847242712975, "global_step": 9565, "epoch": 265, "lr": 0.001} +{"train_loss": 0.043574556708335876, "train_loss_bc": 0.04355105757713318, "train_loss_llm": 0.0023499727249145508, "grad_norm": 0.03673127293586731, "global_step": 9566, "epoch": 265, "lr": 0.001} +{"train_loss": 0.04257439821958542, "train_loss_bc": 0.042535338550806046, "train_loss_llm": 0.00390586256980896, "grad_norm": 0.05122683197259903, "global_step": 9567, "epoch": 265, "lr": 0.001} +{"train_loss": 0.04315033555030823, "train_loss_bc": 0.04315033555030823, "train_loss_llm": 0.0, "grad_norm": 0.06109481677412987, "global_step": 9568, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03301827237010002, "train_loss_bc": 0.03301827237010002, "train_loss_llm": 0.0, "grad_norm": 0.021509597077965736, "global_step": 9569, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03554536774754524, "train_loss_bc": 0.03554536774754524, "train_loss_llm": 0.0, "grad_norm": 0.03672447428107262, "global_step": 9570, "epoch": 265, "lr": 0.001} +{"train_loss": 0.041775189340114594, "train_loss_bc": 0.041775189340114594, "train_loss_llm": 0.0, "grad_norm": 0.05397217348217964, "global_step": 9571, "epoch": 265, "lr": 0.001} +{"train_loss": 0.04661832004785538, "train_loss_bc": 0.04661832004785538, "train_loss_llm": 0.0, "grad_norm": 0.07064110040664673, "global_step": 9572, "epoch": 265, "lr": 0.001} +{"train_loss": 0.036504000425338745, "train_loss_bc": 0.036504000425338745, "train_loss_llm": 0.0, "grad_norm": 0.026570487767457962, "global_step": 9573, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03925073519349098, "train_loss_bc": 0.03925073519349098, "train_loss_llm": 0.0, "grad_norm": 0.039387401193380356, "global_step": 9574, "epoch": 265, "lr": 0.001} +{"train_loss": 0.03877756610098812, "train_loss_bc": 0.0388389527797699, "train_loss_llm": 0.0, "grad_norm": 0.05211712792515755, "global_step": 9575, "epoch": 265, "lr": 0.001, "train/cumulative_reward": 82.20143153937043, "train/mean_score": 7.798291561879327, "train/success_rate": 0.1, "test/cumulative_reward": 30.796015953081582, "test/mean_score": 6.497132952259791, "test/success_rate": 0.14, "val_loss": 0.060227543115615845, "train_action_mse_error": 0.04224005714058876} +{"train_loss": 0.045124489814043045, "train_loss_bc": 0.045124489814043045, "train_loss_llm": 0.0, "grad_norm": 0.06347721815109253, "global_step": 9576, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03590608760714531, "train_loss_bc": 0.03590608760714531, "train_loss_llm": 0.0, "grad_norm": 0.01608320325613022, "global_step": 9577, "epoch": 266, "lr": 0.001} +{"train_loss": 0.0367034375667572, "train_loss_bc": 0.03669767454266548, "train_loss_llm": 0.0005764663219451904, "grad_norm": 0.02476535364985466, "global_step": 9578, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03721283748745918, "train_loss_bc": 0.03721283748745918, "train_loss_llm": 0.0, "grad_norm": 0.03049028292298317, "global_step": 9579, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03419874608516693, "train_loss_bc": 0.03419874608516693, "train_loss_llm": 0.0, "grad_norm": 0.036171942949295044, "global_step": 9580, "epoch": 266, "lr": 0.001} +{"train_loss": 0.042919255793094635, "train_loss_bc": 0.042919255793094635, "train_loss_llm": 0.0, "grad_norm": 0.022210611030459404, "global_step": 9581, "epoch": 266, "lr": 0.001} +{"train_loss": 0.047177232801914215, "train_loss_bc": 0.04717369005084038, "train_loss_llm": 0.00035443902015686035, "grad_norm": 0.02918325923383236, "global_step": 9582, "epoch": 266, "lr": 0.001} +{"train_loss": 0.033575624227523804, "train_loss_bc": 0.033575624227523804, "train_loss_llm": 0.0, "grad_norm": 0.03671025484800339, "global_step": 9583, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03692978248000145, "train_loss_bc": 0.03692978248000145, "train_loss_llm": 0.0, "grad_norm": 0.04723643511533737, "global_step": 9584, "epoch": 266, "lr": 0.001} +{"train_loss": 0.04074021428823471, "train_loss_bc": 0.04074021428823471, "train_loss_llm": 0.0, "grad_norm": 0.020236657932400703, "global_step": 9585, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03560103476047516, "train_loss_bc": 0.03560103476047516, "train_loss_llm": 0.0, "grad_norm": 0.025544753298163414, "global_step": 9586, "epoch": 266, "lr": 0.001} +{"train_loss": 0.0462225079536438, "train_loss_bc": 0.0462225079536438, "train_loss_llm": 0.0, "grad_norm": 0.033720046281814575, "global_step": 9587, "epoch": 266, "lr": 0.001} +{"train_loss": 0.04309774935245514, "train_loss_bc": 0.04309774935245514, "train_loss_llm": 0.0, "grad_norm": 0.03913570195436478, "global_step": 9588, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03476881980895996, "train_loss_bc": 0.03476881980895996, "train_loss_llm": 0.0, "grad_norm": 0.02111484855413437, "global_step": 9589, "epoch": 266, "lr": 0.001} +{"train_loss": 0.037909552454948425, "train_loss_bc": 0.03787371516227722, "train_loss_llm": 0.0035836994647979736, "grad_norm": 0.028244150802493095, "global_step": 9590, "epoch": 266, "lr": 0.001} +{"train_loss": 0.038227494806051254, "train_loss_bc": 0.038227494806051254, "train_loss_llm": 0.0, "grad_norm": 0.03706182539463043, "global_step": 9591, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03646852448582649, "train_loss_bc": 0.03646852448582649, "train_loss_llm": 0.0, "grad_norm": 0.04414781928062439, "global_step": 9592, "epoch": 266, "lr": 0.001} +{"train_loss": 0.04054294526576996, "train_loss_bc": 0.04054294526576996, "train_loss_llm": 0.0, "grad_norm": 0.02222134731709957, "global_step": 9593, "epoch": 266, "lr": 0.001} +{"train_loss": 0.04302624240517616, "train_loss_bc": 0.04302624240517616, "train_loss_llm": 0.0, "grad_norm": 0.0362011194229126, "global_step": 9594, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03401920944452286, "train_loss_bc": 0.03401920944452286, "train_loss_llm": 0.0, "grad_norm": 0.0453459769487381, "global_step": 9595, "epoch": 266, "lr": 0.001} +{"train_loss": 0.04069283977150917, "train_loss_bc": 0.04069283977150917, "train_loss_llm": 0.0, "grad_norm": 0.05473120138049126, "global_step": 9596, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03440608084201813, "train_loss_bc": 0.034405965358018875, "train_loss_llm": 1.1533498764038086e-05, "grad_norm": 0.018646838143467903, "global_step": 9597, "epoch": 266, "lr": 0.001} +{"train_loss": 0.04821640998125076, "train_loss_bc": 0.04821640998125076, "train_loss_llm": 0.0, "grad_norm": 0.025871973484754562, "global_step": 9598, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03123316541314125, "train_loss_bc": 0.031188078224658966, "train_loss_llm": 0.0045087337493896484, "grad_norm": 0.03518014773726463, "global_step": 9599, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03914906457066536, "train_loss_bc": 0.03914906457066536, "train_loss_llm": 0.0, "grad_norm": 0.04601162672042847, "global_step": 9600, "epoch": 266, "lr": 0.001} +{"train_loss": 0.04369482398033142, "train_loss_bc": 0.04369482398033142, "train_loss_llm": 0.0, "grad_norm": 0.020788727328181267, "global_step": 9601, "epoch": 266, "lr": 0.001} +{"train_loss": 0.036475393921136856, "train_loss_bc": 0.036475393921136856, "train_loss_llm": 0.0, "grad_norm": 0.04045538976788521, "global_step": 9602, "epoch": 266, "lr": 0.001} +{"train_loss": 0.039770495146512985, "train_loss_bc": 0.039770495146512985, "train_loss_llm": 0.0, "grad_norm": 0.0490887388586998, "global_step": 9603, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03634141385555267, "train_loss_bc": 0.03634141385555267, "train_loss_llm": 0.0, "grad_norm": 0.06194432079792023, "global_step": 9604, "epoch": 266, "lr": 0.001} +{"train_loss": 0.035901378840208054, "train_loss_bc": 0.03588424250483513, "train_loss_llm": 0.0017137229442596436, "grad_norm": 0.028107095509767532, "global_step": 9605, "epoch": 266, "lr": 0.001} +{"train_loss": 0.04467738792300224, "train_loss_bc": 0.04467738792300224, "train_loss_llm": 0.0, "grad_norm": 0.033358216285705566, "global_step": 9606, "epoch": 266, "lr": 0.001} +{"train_loss": 0.030336223542690277, "train_loss_bc": 0.030336223542690277, "train_loss_llm": 0.0, "grad_norm": 0.03797464072704315, "global_step": 9607, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03677752986550331, "train_loss_bc": 0.036768391728401184, "train_loss_llm": 0.0009138584136962891, "grad_norm": 0.04358229041099548, "global_step": 9608, "epoch": 266, "lr": 0.001} +{"train_loss": 0.0395515039563179, "train_loss_bc": 0.039541978389024734, "train_loss_llm": 0.0009525120258331299, "grad_norm": 0.019130626693367958, "global_step": 9609, "epoch": 266, "lr": 0.001} +{"train_loss": 0.03379271924495697, "train_loss_bc": 0.03379271924495697, "train_loss_llm": 0.0, "grad_norm": 0.0286040510982275, "global_step": 9610, "epoch": 266, "lr": 0.001} +{"train_loss": 0.038581225710610546, "train_loss_bc": 0.037535905838012695, "train_loss_llm": 0.0, "grad_norm": 0.03642894700169563, "global_step": 9611, "epoch": 266, "lr": 0.001, "val_loss": 0.0601242370903492} +{"train_loss": 0.03732597082853317, "train_loss_bc": 0.037297580391168594, "train_loss_llm": 0.002839028835296631, "grad_norm": 0.04271198436617851, "global_step": 9612, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03802365064620972, "train_loss_bc": 0.03801374509930611, "train_loss_llm": 0.0009907186031341553, "grad_norm": 0.018958423286676407, "global_step": 9613, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03880097717046738, "train_loss_bc": 0.038777004927396774, "train_loss_llm": 0.0023971498012542725, "grad_norm": 0.031083330512046814, "global_step": 9614, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03193395212292671, "train_loss_bc": 0.03193395212292671, "train_loss_llm": 0.0, "grad_norm": 0.03456719592213631, "global_step": 9615, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03886773809790611, "train_loss_bc": 0.03886773809790611, "train_loss_llm": 0.0, "grad_norm": 0.03993208333849907, "global_step": 9616, "epoch": 267, "lr": 0.001} +{"train_loss": 0.041885633021593094, "train_loss_bc": 0.041885633021593094, "train_loss_llm": 0.0, "grad_norm": 0.022766754031181335, "global_step": 9617, "epoch": 267, "lr": 0.001} +{"train_loss": 0.039461951702833176, "train_loss_bc": 0.03945237770676613, "train_loss_llm": 0.0009575486183166504, "grad_norm": 0.03530595451593399, "global_step": 9618, "epoch": 267, "lr": 0.001} +{"train_loss": 0.038652293384075165, "train_loss_bc": 0.03864256292581558, "train_loss_llm": 0.0009731650352478027, "grad_norm": 0.0366433784365654, "global_step": 9619, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03440967574715614, "train_loss_bc": 0.03440967574715614, "train_loss_llm": 0.0, "grad_norm": 0.04515233263373375, "global_step": 9620, "epoch": 267, "lr": 0.001} +{"train_loss": 0.04242737591266632, "train_loss_bc": 0.04242737591266632, "train_loss_llm": 0.0, "grad_norm": 0.02217438444495201, "global_step": 9621, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03834909945726395, "train_loss_bc": 0.03834909945726395, "train_loss_llm": 0.0, "grad_norm": 0.038323745131492615, "global_step": 9622, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03860216215252876, "train_loss_bc": 0.03860216215252876, "train_loss_llm": 0.0, "grad_norm": 0.05087592452764511, "global_step": 9623, "epoch": 267, "lr": 0.001} +{"train_loss": 0.035166703164577484, "train_loss_bc": 0.035166703164577484, "train_loss_llm": 0.0, "grad_norm": 0.05758742243051529, "global_step": 9624, "epoch": 267, "lr": 0.001} +{"train_loss": 0.04106968641281128, "train_loss_bc": 0.04106968641281128, "train_loss_llm": 0.0, "grad_norm": 0.021000413224101067, "global_step": 9625, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03352907672524452, "train_loss_bc": 0.03352229669690132, "train_loss_llm": 0.0006778538227081299, "grad_norm": 0.02938949689269066, "global_step": 9626, "epoch": 267, "lr": 0.001} +{"train_loss": 0.04968688637018204, "train_loss_bc": 0.04967015981674194, "train_loss_llm": 0.0016724765300750732, "grad_norm": 0.032322511076927185, "global_step": 9627, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03308805823326111, "train_loss_bc": 0.03304868936538696, "train_loss_llm": 0.003936856985092163, "grad_norm": 0.037927720695734024, "global_step": 9628, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03829857334494591, "train_loss_bc": 0.03829857334494591, "train_loss_llm": 0.0, "grad_norm": 0.019967883825302124, "global_step": 9629, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03655972704291344, "train_loss_bc": 0.03655972704291344, "train_loss_llm": 0.0, "grad_norm": 0.03029806539416313, "global_step": 9630, "epoch": 267, "lr": 0.001} +{"train_loss": 0.031228680163621902, "train_loss_bc": 0.031207386404275894, "train_loss_llm": 0.0021293163299560547, "grad_norm": 0.034235015511512756, "global_step": 9631, "epoch": 267, "lr": 0.001} +{"train_loss": 0.04133399575948715, "train_loss_bc": 0.04133399575948715, "train_loss_llm": 0.0, "grad_norm": 0.04176466166973114, "global_step": 9632, "epoch": 267, "lr": 0.001} +{"train_loss": 0.04713262990117073, "train_loss_bc": 0.04713262990117073, "train_loss_llm": 0.0, "grad_norm": 0.021106991916894913, "global_step": 9633, "epoch": 267, "lr": 0.001} +{"train_loss": 0.037239912897348404, "train_loss_bc": 0.037239912897348404, "train_loss_llm": 0.0, "grad_norm": 0.03376840054988861, "global_step": 9634, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03510310873389244, "train_loss_bc": 0.03510310873389244, "train_loss_llm": 0.0, "grad_norm": 0.04331544041633606, "global_step": 9635, "epoch": 267, "lr": 0.001} +{"train_loss": 0.037279292941093445, "train_loss_bc": 0.037279292941093445, "train_loss_llm": 0.0, "grad_norm": 0.054350148886442184, "global_step": 9636, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03294855356216431, "train_loss_bc": 0.03294855356216431, "train_loss_llm": 0.0, "grad_norm": 0.017983950674533844, "global_step": 9637, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03769833222031593, "train_loss_bc": 0.03769833222031593, "train_loss_llm": 0.0, "grad_norm": 0.03040333278477192, "global_step": 9638, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03963712975382805, "train_loss_bc": 0.03963712975382805, "train_loss_llm": 0.0, "grad_norm": 0.04157480597496033, "global_step": 9639, "epoch": 267, "lr": 0.001} +{"train_loss": 0.0380263589322567, "train_loss_bc": 0.03801340609788895, "train_loss_llm": 0.001295328140258789, "grad_norm": 0.050215400755405426, "global_step": 9640, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03522925451397896, "train_loss_bc": 0.035227179527282715, "train_loss_llm": 0.00020745396614074707, "grad_norm": 0.019561903551220894, "global_step": 9641, "epoch": 267, "lr": 0.001} +{"train_loss": 0.04133648797869682, "train_loss_bc": 0.04133648797869682, "train_loss_llm": 0.0, "grad_norm": 0.0255351010710001, "global_step": 9642, "epoch": 267, "lr": 0.001} +{"train_loss": 0.0392291322350502, "train_loss_bc": 0.0392291322350502, "train_loss_llm": 0.0, "grad_norm": 0.03306596726179123, "global_step": 9643, "epoch": 267, "lr": 0.001} +{"train_loss": 0.03715375438332558, "train_loss_bc": 0.03715375438332558, "train_loss_llm": 0.0, "grad_norm": 0.04322982206940651, "global_step": 9644, "epoch": 267, "lr": 0.001} +{"train_loss": 0.043523259460926056, "train_loss_bc": 0.043523259460926056, "train_loss_llm": 0.0, "grad_norm": 0.022121217101812363, "global_step": 9645, "epoch": 267, "lr": 0.001} +{"train_loss": 0.04547712951898575, "train_loss_bc": 0.04547712951898575, "train_loss_llm": 0.0, "grad_norm": 0.03382208198308945, "global_step": 9646, "epoch": 267, "lr": 0.001} +{"train_loss": 0.0386274094796843, "train_loss_bc": 0.044870536774396896, "train_loss_llm": 0.0, "grad_norm": 0.045763012021780014, "global_step": 9647, "epoch": 267, "lr": 0.001, "val_loss": 0.06008096784353256} +{"train_loss": 0.037083473056554794, "train_loss_bc": 0.037083473056554794, "train_loss_llm": 0.0, "grad_norm": 0.059652041643857956, "global_step": 9648, "epoch": 268, "lr": 0.001} +{"train_loss": 0.04332160949707031, "train_loss_bc": 0.04332093521952629, "train_loss_llm": 6.73830509185791e-05, "grad_norm": 0.020648052915930748, "global_step": 9649, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03945654630661011, "train_loss_bc": 0.03943539410829544, "train_loss_llm": 0.002115398645401001, "grad_norm": 0.02984943799674511, "global_step": 9650, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03267800435423851, "train_loss_bc": 0.03267800435423851, "train_loss_llm": 0.0, "grad_norm": 0.0389309860765934, "global_step": 9651, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03965819627046585, "train_loss_bc": 0.03965819627046585, "train_loss_llm": 0.0, "grad_norm": 0.04627427086234093, "global_step": 9652, "epoch": 268, "lr": 0.001} +{"train_loss": 0.041644707322120667, "train_loss_bc": 0.041644707322120667, "train_loss_llm": 0.0, "grad_norm": 0.019243644550442696, "global_step": 9653, "epoch": 268, "lr": 0.001} +{"train_loss": 0.04307997599244118, "train_loss_bc": 0.04307997599244118, "train_loss_llm": 0.0, "grad_norm": 0.03152649849653244, "global_step": 9654, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03814303129911423, "train_loss_bc": 0.03814303129911423, "train_loss_llm": 0.0, "grad_norm": 0.03618202358484268, "global_step": 9655, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03488847613334656, "train_loss_bc": 0.03488847613334656, "train_loss_llm": 0.0, "grad_norm": 0.04711387678980827, "global_step": 9656, "epoch": 268, "lr": 0.001} +{"train_loss": 0.039435483515262604, "train_loss_bc": 0.039435483515262604, "train_loss_llm": 0.0, "grad_norm": 0.019075652584433556, "global_step": 9657, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03665275499224663, "train_loss_bc": 0.03664516285061836, "train_loss_llm": 0.000759124755859375, "grad_norm": 0.03072429448366165, "global_step": 9658, "epoch": 268, "lr": 0.001} +{"train_loss": 0.0393693782389164, "train_loss_bc": 0.0393693782389164, "train_loss_llm": 0.0, "grad_norm": 0.04318951815366745, "global_step": 9659, "epoch": 268, "lr": 0.001} +{"train_loss": 0.037963952869176865, "train_loss_bc": 0.037959855049848557, "train_loss_llm": 0.000409930944442749, "grad_norm": 0.04882073029875755, "global_step": 9660, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03871193155646324, "train_loss_bc": 0.03870484232902527, "train_loss_llm": 0.0007088184356689453, "grad_norm": 0.019195951521396637, "global_step": 9661, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03442925959825516, "train_loss_bc": 0.03442925959825516, "train_loss_llm": 0.0, "grad_norm": 0.03326556831598282, "global_step": 9662, "epoch": 268, "lr": 0.001} +{"train_loss": 0.037712279707193375, "train_loss_bc": 0.03770943731069565, "train_loss_llm": 0.0002842545509338379, "grad_norm": 0.047023266553878784, "global_step": 9663, "epoch": 268, "lr": 0.001} +{"train_loss": 0.033715590834617615, "train_loss_bc": 0.03370797634124756, "train_loss_llm": 0.0007612705230712891, "grad_norm": 0.06375738233327866, "global_step": 9664, "epoch": 268, "lr": 0.001} +{"train_loss": 0.04228620603680611, "train_loss_bc": 0.04228620603680611, "train_loss_llm": 0.0, "grad_norm": 0.022070957347750664, "global_step": 9665, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03943978622555733, "train_loss_bc": 0.03943978622555733, "train_loss_llm": 0.0, "grad_norm": 0.02546156942844391, "global_step": 9666, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03459326550364494, "train_loss_bc": 0.03459326550364494, "train_loss_llm": 0.0, "grad_norm": 0.03412160649895668, "global_step": 9667, "epoch": 268, "lr": 0.001} +{"train_loss": 0.04147028550505638, "train_loss_bc": 0.041462671011686325, "train_loss_llm": 0.0007612705230712891, "grad_norm": 0.046480730175971985, "global_step": 9668, "epoch": 268, "lr": 0.001} +{"train_loss": 0.035244643688201904, "train_loss_bc": 0.035244643688201904, "train_loss_llm": 0.0, "grad_norm": 0.020207803696393967, "global_step": 9669, "epoch": 268, "lr": 0.001} +{"train_loss": 0.045485079288482666, "train_loss_bc": 0.045485079288482666, "train_loss_llm": 0.0, "grad_norm": 0.03384680673480034, "global_step": 9670, "epoch": 268, "lr": 0.001} +{"train_loss": 0.04008559137582779, "train_loss_bc": 0.04008559137582779, "train_loss_llm": 0.0, "grad_norm": 0.036713626235723495, "global_step": 9671, "epoch": 268, "lr": 0.001} +{"train_loss": 0.041390445083379745, "train_loss_bc": 0.041390445083379745, "train_loss_llm": 0.0, "grad_norm": 0.03893858939409256, "global_step": 9672, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03586136922240257, "train_loss_bc": 0.035860802978277206, "train_loss_llm": 5.6624412536621094e-05, "grad_norm": 0.018382955342531204, "global_step": 9673, "epoch": 268, "lr": 0.001} +{"train_loss": 0.039928190410137177, "train_loss_bc": 0.039919618517160416, "train_loss_llm": 0.0008572638034820557, "grad_norm": 0.02974868193268776, "global_step": 9674, "epoch": 268, "lr": 0.001} +{"train_loss": 0.04047201946377754, "train_loss_bc": 0.04047201946377754, "train_loss_llm": 0.0, "grad_norm": 0.030995603650808334, "global_step": 9675, "epoch": 268, "lr": 0.001} +{"train_loss": 0.04175559803843498, "train_loss_bc": 0.04175559803843498, "train_loss_llm": 0.0, "grad_norm": 0.03681882098317146, "global_step": 9676, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03622252494096756, "train_loss_bc": 0.03622252494096756, "train_loss_llm": 0.0, "grad_norm": 0.01819782517850399, "global_step": 9677, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03652742877602577, "train_loss_bc": 0.03652742877602577, "train_loss_llm": 0.0, "grad_norm": 0.03351958468556404, "global_step": 9678, "epoch": 268, "lr": 0.001} +{"train_loss": 0.034351449459791183, "train_loss_bc": 0.034351449459791183, "train_loss_llm": 0.0, "grad_norm": 0.03927505761384964, "global_step": 9679, "epoch": 268, "lr": 0.001} +{"train_loss": 0.0390499122440815, "train_loss_bc": 0.039035823196172714, "train_loss_llm": 0.0014089345932006836, "grad_norm": 0.04998581483960152, "global_step": 9680, "epoch": 268, "lr": 0.001} +{"train_loss": 0.03517526760697365, "train_loss_bc": 0.035162195563316345, "train_loss_llm": 0.001307070255279541, "grad_norm": 0.017714371904730797, "global_step": 9681, "epoch": 268, "lr": 0.001} +{"train_loss": 0.039127167314291, "train_loss_bc": 0.03912167623639107, "train_loss_llm": 0.0005491971969604492, "grad_norm": 0.028674453496932983, "global_step": 9682, "epoch": 268, "lr": 0.001} +{"train_loss": 0.038478967940641776, "train_loss_bc": 0.03879531845450401, "train_loss_llm": 0.0036644339561462402, "grad_norm": 0.03865805268287659, "global_step": 9683, "epoch": 268, "lr": 0.001, "val_loss": 0.06051690876483917} +{"train_loss": 0.027527429163455963, "train_loss_bc": 0.027527429163455963, "train_loss_llm": 0.0, "grad_norm": 0.04779980331659317, "global_step": 9684, "epoch": 269, "lr": 0.001} +{"train_loss": 0.04179185628890991, "train_loss_bc": 0.04179185628890991, "train_loss_llm": 0.0, "grad_norm": 0.02137887105345726, "global_step": 9685, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03941540792584419, "train_loss_bc": 0.03941540792584419, "train_loss_llm": 0.0, "grad_norm": 0.0295823123306036, "global_step": 9686, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03383292630314827, "train_loss_bc": 0.03383292630314827, "train_loss_llm": 0.0, "grad_norm": 0.0389518141746521, "global_step": 9687, "epoch": 269, "lr": 0.001} +{"train_loss": 0.04007334262132645, "train_loss_bc": 0.04007334262132645, "train_loss_llm": 0.0, "grad_norm": 0.04774602875113487, "global_step": 9688, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03124411590397358, "train_loss_bc": 0.03124411590397358, "train_loss_llm": 0.0, "grad_norm": 0.0185419749468565, "global_step": 9689, "epoch": 269, "lr": 0.001} +{"train_loss": 0.038978952914476395, "train_loss_bc": 0.03897586837410927, "train_loss_llm": 0.00030857324600219727, "grad_norm": 0.03247968852519989, "global_step": 9690, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03972243517637253, "train_loss_bc": 0.03971896320581436, "train_loss_llm": 0.0003470778465270996, "grad_norm": 0.03910769149661064, "global_step": 9691, "epoch": 269, "lr": 0.001} +{"train_loss": 0.04081030562520027, "train_loss_bc": 0.04081030562520027, "train_loss_llm": 0.0, "grad_norm": 0.04851452261209488, "global_step": 9692, "epoch": 269, "lr": 0.001} +{"train_loss": 0.030700145289301872, "train_loss_bc": 0.030700145289301872, "train_loss_llm": 0.0, "grad_norm": 0.020847458392381668, "global_step": 9693, "epoch": 269, "lr": 0.001} +{"train_loss": 0.047514934092760086, "train_loss_bc": 0.047514934092760086, "train_loss_llm": 0.0, "grad_norm": 0.031607646495103836, "global_step": 9694, "epoch": 269, "lr": 0.001} +{"train_loss": 0.04497493803501129, "train_loss_bc": 0.04497130215167999, "train_loss_llm": 0.0003637075424194336, "grad_norm": 0.03817380219697952, "global_step": 9695, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03661596402525902, "train_loss_bc": 0.03661596402525902, "train_loss_llm": 0.0, "grad_norm": 0.05449822545051575, "global_step": 9696, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03671926632523537, "train_loss_bc": 0.03671926632523537, "train_loss_llm": 0.0, "grad_norm": 0.02133728750050068, "global_step": 9697, "epoch": 269, "lr": 0.001} +{"train_loss": 0.04024370759725571, "train_loss_bc": 0.04024370759725571, "train_loss_llm": 0.0, "grad_norm": 0.028987765312194824, "global_step": 9698, "epoch": 269, "lr": 0.001} +{"train_loss": 0.045732587575912476, "train_loss_bc": 0.045732587575912476, "train_loss_llm": 0.0, "grad_norm": 0.043837372213602066, "global_step": 9699, "epoch": 269, "lr": 0.001} +{"train_loss": 0.043195709586143494, "train_loss_bc": 0.043195709586143494, "train_loss_llm": 0.0, "grad_norm": 0.05074246600270271, "global_step": 9700, "epoch": 269, "lr": 0.001} +{"train_loss": 0.039126161485910416, "train_loss_bc": 0.039126161485910416, "train_loss_llm": 0.0, "grad_norm": 0.01809583604335785, "global_step": 9701, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03993872180581093, "train_loss_bc": 0.03993872180581093, "train_loss_llm": 0.0, "grad_norm": 0.03173484280705452, "global_step": 9702, "epoch": 269, "lr": 0.001} +{"train_loss": 0.038360755890607834, "train_loss_bc": 0.038360755890607834, "train_loss_llm": 0.0, "grad_norm": 0.04326234385371208, "global_step": 9703, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03611152246594429, "train_loss_bc": 0.03611152246594429, "train_loss_llm": 0.0, "grad_norm": 0.04832719266414642, "global_step": 9704, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03637749329209328, "train_loss_bc": 0.03635803237557411, "train_loss_llm": 0.0019460320472717285, "grad_norm": 0.02768419124186039, "global_step": 9705, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03462410718202591, "train_loss_bc": 0.03462410718202591, "train_loss_llm": 0.0, "grad_norm": 0.03850845992565155, "global_step": 9706, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03780601918697357, "train_loss_bc": 0.03780601918697357, "train_loss_llm": 0.0, "grad_norm": 0.05394158512353897, "global_step": 9707, "epoch": 269, "lr": 0.001} +{"train_loss": 0.043302085250616074, "train_loss_bc": 0.043302085250616074, "train_loss_llm": 0.0, "grad_norm": 0.0633326843380928, "global_step": 9708, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03898705542087555, "train_loss_bc": 0.038971491158008575, "train_loss_llm": 0.0015563666820526123, "grad_norm": 0.018033061176538467, "global_step": 9709, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03690777346491814, "train_loss_bc": 0.03690777346491814, "train_loss_llm": 0.0, "grad_norm": 0.026371769607067108, "global_step": 9710, "epoch": 269, "lr": 0.001} +{"train_loss": 0.036980219185352325, "train_loss_bc": 0.036980219185352325, "train_loss_llm": 0.0, "grad_norm": 0.03491722047328949, "global_step": 9711, "epoch": 269, "lr": 0.001} +{"train_loss": 0.036673206835985184, "train_loss_bc": 0.036673206835985184, "train_loss_llm": 0.0, "grad_norm": 0.039441823959350586, "global_step": 9712, "epoch": 269, "lr": 0.001} +{"train_loss": 0.04925400763750076, "train_loss_bc": 0.04925400763750076, "train_loss_llm": 0.0, "grad_norm": 0.020519452169537544, "global_step": 9713, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03693205490708351, "train_loss_bc": 0.03693017363548279, "train_loss_llm": 0.00018802285194396973, "grad_norm": 0.03139905259013176, "global_step": 9714, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03497568517923355, "train_loss_bc": 0.0349711999297142, "train_loss_llm": 0.00044846534729003906, "grad_norm": 0.03550480306148529, "global_step": 9715, "epoch": 269, "lr": 0.001} +{"train_loss": 0.0439663901925087, "train_loss_bc": 0.0439663901925087, "train_loss_llm": 0.0, "grad_norm": 0.04340618848800659, "global_step": 9716, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03768419474363327, "train_loss_bc": 0.03766204044222832, "train_loss_llm": 0.002215564250946045, "grad_norm": 0.01885693334043026, "global_step": 9717, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03408774733543396, "train_loss_bc": 0.034083809703588486, "train_loss_llm": 0.0003936588764190674, "grad_norm": 0.024268027395009995, "global_step": 9718, "epoch": 269, "lr": 0.001} +{"train_loss": 0.03848279660774602, "train_loss_bc": 0.03419145196676254, "train_loss_llm": 0.0, "grad_norm": 0.03403666615486145, "global_step": 9719, "epoch": 269, "lr": 0.001, "val_loss": 0.06004183739423752} +{"train_loss": 0.04362403601408005, "train_loss_bc": 0.043611571192741394, "train_loss_llm": 0.0012465715408325195, "grad_norm": 0.043780870735645294, "global_step": 9720, "epoch": 270, "lr": 0.001} +{"train_loss": 0.0353962704539299, "train_loss_bc": 0.03539081662893295, "train_loss_llm": 0.0005455315113067627, "grad_norm": 0.01493159681558609, "global_step": 9721, "epoch": 270, "lr": 0.001} +{"train_loss": 0.04022439941763878, "train_loss_bc": 0.04022439941763878, "train_loss_llm": 0.0, "grad_norm": 0.023725833743810654, "global_step": 9722, "epoch": 270, "lr": 0.001} +{"train_loss": 0.041857603937387466, "train_loss_bc": 0.04183068498969078, "train_loss_llm": 0.0026920437812805176, "grad_norm": 0.029124891385436058, "global_step": 9723, "epoch": 270, "lr": 0.001} +{"train_loss": 0.04278157651424408, "train_loss_bc": 0.04277530685067177, "train_loss_llm": 0.0006270706653594971, "grad_norm": 0.04527350515127182, "global_step": 9724, "epoch": 270, "lr": 0.001} +{"train_loss": 0.04037277027964592, "train_loss_bc": 0.04036779701709747, "train_loss_llm": 0.000497281551361084, "grad_norm": 0.022953268140554428, "global_step": 9725, "epoch": 270, "lr": 0.001} +{"train_loss": 0.04272778332233429, "train_loss_bc": 0.042721472680568695, "train_loss_llm": 0.0006309151649475098, "grad_norm": 0.034202806651592255, "global_step": 9726, "epoch": 270, "lr": 0.001} +{"train_loss": 0.0378190353512764, "train_loss_bc": 0.0378190353512764, "train_loss_llm": 0.0, "grad_norm": 0.04535849392414093, "global_step": 9727, "epoch": 270, "lr": 0.001} +{"train_loss": 0.033483829349279404, "train_loss_bc": 0.033483829349279404, "train_loss_llm": 0.0, "grad_norm": 0.0562145859003067, "global_step": 9728, "epoch": 270, "lr": 0.001} +{"train_loss": 0.031609926372766495, "train_loss_bc": 0.031609926372766495, "train_loss_llm": 0.0, "grad_norm": 0.02011393941938877, "global_step": 9729, "epoch": 270, "lr": 0.001} +{"train_loss": 0.03494231402873993, "train_loss_bc": 0.03494231402873993, "train_loss_llm": 0.0, "grad_norm": 0.027361273765563965, "global_step": 9730, "epoch": 270, "lr": 0.001} +{"train_loss": 0.038224030286073685, "train_loss_bc": 0.038224030286073685, "train_loss_llm": 0.0, "grad_norm": 0.03844720125198364, "global_step": 9731, "epoch": 270, "lr": 0.001} +{"train_loss": 0.040249016135931015, "train_loss_bc": 0.040249016135931015, "train_loss_llm": 0.0, "grad_norm": 0.04791713133454323, "global_step": 9732, "epoch": 270, "lr": 0.001} +{"train_loss": 0.03871208056807518, "train_loss_bc": 0.03871208056807518, "train_loss_llm": 0.0, "grad_norm": 0.01929280161857605, "global_step": 9733, "epoch": 270, "lr": 0.001} +{"train_loss": 0.0360114611685276, "train_loss_bc": 0.0360114611685276, "train_loss_llm": 0.0, "grad_norm": 0.03145361691713333, "global_step": 9734, "epoch": 270, "lr": 0.001} +{"train_loss": 0.0414452962577343, "train_loss_bc": 0.0414452962577343, "train_loss_llm": 0.0, "grad_norm": 0.040834829211235046, "global_step": 9735, "epoch": 270, "lr": 0.001} +{"train_loss": 0.03431226685643196, "train_loss_bc": 0.03431226685643196, "train_loss_llm": 0.0, "grad_norm": 0.03978646919131279, "global_step": 9736, "epoch": 270, "lr": 0.001} +{"train_loss": 0.03741154447197914, "train_loss_bc": 0.03741154447197914, "train_loss_llm": 0.0, "grad_norm": 0.019812902435660362, "global_step": 9737, "epoch": 270, "lr": 0.001} +{"train_loss": 0.04155092313885689, "train_loss_bc": 0.04155092313885689, "train_loss_llm": 0.0, "grad_norm": 0.03011447936296463, "global_step": 9738, "epoch": 270, "lr": 0.001} +{"train_loss": 0.04006529226899147, "train_loss_bc": 0.04006529226899147, "train_loss_llm": 0.0, "grad_norm": 0.04689248278737068, "global_step": 9739, "epoch": 270, "lr": 0.001} +{"train_loss": 0.034379489719867706, "train_loss_bc": 0.03437720239162445, "train_loss_llm": 0.00022861361503601074, "grad_norm": 0.048781733959913254, "global_step": 9740, "epoch": 270, "lr": 0.001} +{"train_loss": 0.033454202115535736, "train_loss_bc": 0.033454202115535736, "train_loss_llm": 0.0, "grad_norm": 0.016404759138822556, "global_step": 9741, "epoch": 270, "lr": 0.001} +{"train_loss": 0.041301429271698, "train_loss_bc": 0.041301429271698, "train_loss_llm": 0.0, "grad_norm": 0.026042615994811058, "global_step": 9742, "epoch": 270, "lr": 0.001} +{"train_loss": 0.03840643912553787, "train_loss_bc": 0.03840643912553787, "train_loss_llm": 0.0, "grad_norm": 0.038737937808036804, "global_step": 9743, "epoch": 270, "lr": 0.001} +{"train_loss": 0.03701302036643028, "train_loss_bc": 0.03701302036643028, "train_loss_llm": 0.0, "grad_norm": 0.04451090469956398, "global_step": 9744, "epoch": 270, "lr": 0.001} +{"train_loss": 0.04102478176355362, "train_loss_bc": 0.04102478176355362, "train_loss_llm": 0.0, "grad_norm": 0.017977332696318626, "global_step": 9745, "epoch": 270, "lr": 0.001} +{"train_loss": 0.04217618703842163, "train_loss_bc": 0.04217618703842163, "train_loss_llm": 0.0, "grad_norm": 0.02860116772353649, "global_step": 9746, "epoch": 270, "lr": 0.001} +{"train_loss": 0.032237883657217026, "train_loss_bc": 0.03223670274019241, "train_loss_llm": 0.00011807680130004883, "grad_norm": 0.04035456106066704, "global_step": 9747, "epoch": 270, "lr": 0.001} +{"train_loss": 0.028658343479037285, "train_loss_bc": 0.028658343479037285, "train_loss_llm": 0.0, "grad_norm": 0.046138375997543335, "global_step": 9748, "epoch": 270, "lr": 0.001} +{"train_loss": 0.04393457621335983, "train_loss_bc": 0.04393457621335983, "train_loss_llm": 0.0, "grad_norm": 0.019954420626163483, "global_step": 9749, "epoch": 270, "lr": 0.001} +{"train_loss": 0.034907449036836624, "train_loss_bc": 0.03488955646753311, "train_loss_llm": 0.0017893612384796143, "grad_norm": 0.028034307062625885, "global_step": 9750, "epoch": 270, "lr": 0.001} +{"train_loss": 0.037978142499923706, "train_loss_bc": 0.0379771888256073, "train_loss_llm": 9.542703628540039e-05, "grad_norm": 0.03722327575087547, "global_step": 9751, "epoch": 270, "lr": 0.001} +{"train_loss": 0.04118400812149048, "train_loss_bc": 0.04118400812149048, "train_loss_llm": 0.0, "grad_norm": 0.045834366232156754, "global_step": 9752, "epoch": 270, "lr": 0.001} +{"train_loss": 0.05227390304207802, "train_loss_bc": 0.05225450545549393, "train_loss_llm": 0.0019397437572479248, "grad_norm": 0.024583466351032257, "global_step": 9753, "epoch": 270, "lr": 0.001} +{"train_loss": 0.040948331356048584, "train_loss_bc": 0.040948331356048584, "train_loss_llm": 0.0, "grad_norm": 0.033657923340797424, "global_step": 9754, "epoch": 270, "lr": 0.001} +{"train_loss": 0.0384298682005869, "train_loss_bc": 0.030775612220168114, "train_loss_llm": 0.0, "grad_norm": 0.04085512086749077, "global_step": 9755, "epoch": 270, "lr": 0.001, "train/cumulative_reward": 105.98102877395743, "train/mean_score": 8.718878567831283, "train/success_rate": 0.2, "test/cumulative_reward": 29.40348922548489, "test/mean_score": 6.706744768434716, "test/success_rate": 0.16, "val_loss": 0.06041323021054268, "train_action_mse_error": 0.0418558195233345} +{"train_loss": 0.03842378407716751, "train_loss_bc": 0.03841351717710495, "train_loss_llm": 0.0010266304016113281, "grad_norm": 0.04397232085466385, "global_step": 9756, "epoch": 271, "lr": 0.001} +{"train_loss": 0.042367007583379745, "train_loss_bc": 0.04236464202404022, "train_loss_llm": 0.0002364516258239746, "grad_norm": 0.017620088532567024, "global_step": 9757, "epoch": 271, "lr": 0.001} +{"train_loss": 0.04100188612937927, "train_loss_bc": 0.04100188612937927, "train_loss_llm": 0.0, "grad_norm": 0.0276477187871933, "global_step": 9758, "epoch": 271, "lr": 0.001} +{"train_loss": 0.037004899233579636, "train_loss_bc": 0.036991436034440994, "train_loss_llm": 0.001346290111541748, "grad_norm": 0.045533161610364914, "global_step": 9759, "epoch": 271, "lr": 0.001} +{"train_loss": 0.03786776214838028, "train_loss_bc": 0.03786419332027435, "train_loss_llm": 0.00035697221755981445, "grad_norm": 0.06203814223408699, "global_step": 9760, "epoch": 271, "lr": 0.001} +{"train_loss": 0.034970786422491074, "train_loss_bc": 0.034970786422491074, "train_loss_llm": 0.0, "grad_norm": 0.019738659262657166, "global_step": 9761, "epoch": 271, "lr": 0.001} +{"train_loss": 0.04288017749786377, "train_loss_bc": 0.04286295175552368, "train_loss_llm": 0.0017227530479431152, "grad_norm": 0.03389693796634674, "global_step": 9762, "epoch": 271, "lr": 0.001} +{"train_loss": 0.04051867499947548, "train_loss_bc": 0.0405174121260643, "train_loss_llm": 0.00012627243995666504, "grad_norm": 0.04880586639046669, "global_step": 9763, "epoch": 271, "lr": 0.001} +{"train_loss": 0.04376405104994774, "train_loss_bc": 0.04375029355287552, "train_loss_llm": 0.0013759136199951172, "grad_norm": 0.059668757021427155, "global_step": 9764, "epoch": 271, "lr": 0.001} +{"train_loss": 0.033634986728429794, "train_loss_bc": 0.033634986728429794, "train_loss_llm": 0.0, "grad_norm": 0.016253674402832985, "global_step": 9765, "epoch": 271, "lr": 0.001} +{"train_loss": 0.03911127522587776, "train_loss_bc": 0.03911127522587776, "train_loss_llm": 0.0, "grad_norm": 0.03032028116285801, "global_step": 9766, "epoch": 271, "lr": 0.001} +{"train_loss": 0.03841874375939369, "train_loss_bc": 0.03841874375939369, "train_loss_llm": 0.0, "grad_norm": 0.043106239289045334, "global_step": 9767, "epoch": 271, "lr": 0.001} +{"train_loss": 0.04624912887811661, "train_loss_bc": 0.04624912887811661, "train_loss_llm": 0.0, "grad_norm": 0.05513940379023552, "global_step": 9768, "epoch": 271, "lr": 0.001} +{"train_loss": 0.040090564638376236, "train_loss_bc": 0.040090564638376236, "train_loss_llm": 0.0, "grad_norm": 0.020955221727490425, "global_step": 9769, "epoch": 271, "lr": 0.001} +{"train_loss": 0.03069302998483181, "train_loss_bc": 0.03069302998483181, "train_loss_llm": 0.0, "grad_norm": 0.02598019316792488, "global_step": 9770, "epoch": 271, "lr": 0.001} +{"train_loss": 0.033517759293317795, "train_loss_bc": 0.03351373225450516, "train_loss_llm": 0.00040262937545776367, "grad_norm": 0.03337249532341957, "global_step": 9771, "epoch": 271, "lr": 0.001} +{"train_loss": 0.03785938769578934, "train_loss_bc": 0.03785938769578934, "train_loss_llm": 0.0, "grad_norm": 0.045260291546583176, "global_step": 9772, "epoch": 271, "lr": 0.001} +{"train_loss": 0.041065387427806854, "train_loss_bc": 0.04104555770754814, "train_loss_llm": 0.0019829869270324707, "grad_norm": 0.020159434527158737, "global_step": 9773, "epoch": 271, "lr": 0.001} +{"train_loss": 0.034989502280950546, "train_loss_bc": 0.034989502280950546, "train_loss_llm": 0.0, "grad_norm": 0.029581159353256226, "global_step": 9774, "epoch": 271, "lr": 0.001} +{"train_loss": 0.038206104189157486, "train_loss_bc": 0.038206104189157486, "train_loss_llm": 0.0, "grad_norm": 0.03536452725529671, "global_step": 9775, "epoch": 271, "lr": 0.001} +{"train_loss": 0.033558353781700134, "train_loss_bc": 0.033558353781700134, "train_loss_llm": 0.0, "grad_norm": 0.043197713792324066, "global_step": 9776, "epoch": 271, "lr": 0.001} +{"train_loss": 0.034397806972265244, "train_loss_bc": 0.034397806972265244, "train_loss_llm": 0.0, "grad_norm": 0.02020970918238163, "global_step": 9777, "epoch": 271, "lr": 0.001} +{"train_loss": 0.03259667381644249, "train_loss_bc": 0.03259503096342087, "train_loss_llm": 0.00016430020332336426, "grad_norm": 0.029339581727981567, "global_step": 9778, "epoch": 271, "lr": 0.001} +{"train_loss": 0.0448632575571537, "train_loss_bc": 0.0448632575571537, "train_loss_llm": 0.0, "grad_norm": 0.04001501947641373, "global_step": 9779, "epoch": 271, "lr": 0.001} +{"train_loss": 0.04369177669286728, "train_loss_bc": 0.04369177669286728, "train_loss_llm": 0.0, "grad_norm": 0.054006900638341904, "global_step": 9780, "epoch": 271, "lr": 0.001} +{"train_loss": 0.04520436376333237, "train_loss_bc": 0.04520436376333237, "train_loss_llm": 0.0, "grad_norm": 0.02578836679458618, "global_step": 9781, "epoch": 271, "lr": 0.001} +{"train_loss": 0.0376497320830822, "train_loss_bc": 0.0376497320830822, "train_loss_llm": 0.0, "grad_norm": 0.03289220109581947, "global_step": 9782, "epoch": 271, "lr": 0.001} +{"train_loss": 0.030352013185620308, "train_loss_bc": 0.030352013185620308, "train_loss_llm": 0.0, "grad_norm": 0.047882284969091415, "global_step": 9783, "epoch": 271, "lr": 0.001} +{"train_loss": 0.046233441680669785, "train_loss_bc": 0.046233441680669785, "train_loss_llm": 0.0, "grad_norm": 0.06931207329034805, "global_step": 9784, "epoch": 271, "lr": 0.001} +{"train_loss": 0.04435766860842705, "train_loss_bc": 0.04435766860842705, "train_loss_llm": 0.0, "grad_norm": 0.025105224922299385, "global_step": 9785, "epoch": 271, "lr": 0.001} +{"train_loss": 0.043977633118629456, "train_loss_bc": 0.043977633118629456, "train_loss_llm": 0.0, "grad_norm": 0.033622197806835175, "global_step": 9786, "epoch": 271, "lr": 0.001} +{"train_loss": 0.03866391256451607, "train_loss_bc": 0.03866391256451607, "train_loss_llm": 0.0, "grad_norm": 0.039704933762550354, "global_step": 9787, "epoch": 271, "lr": 0.001} +{"train_loss": 0.040365345776081085, "train_loss_bc": 0.040343910455703735, "train_loss_llm": 0.002143681049346924, "grad_norm": 0.06014251336455345, "global_step": 9788, "epoch": 271, "lr": 0.001} +{"train_loss": 0.03603476658463478, "train_loss_bc": 0.03603376820683479, "train_loss_llm": 9.998679161071777e-05, "grad_norm": 0.01701611652970314, "global_step": 9789, "epoch": 271, "lr": 0.001} +{"train_loss": 0.03475970774888992, "train_loss_bc": 0.03475970774888992, "train_loss_llm": 0.0, "grad_norm": 0.031063657253980637, "global_step": 9790, "epoch": 271, "lr": 0.001} +{"train_loss": 0.03846141245837013, "train_loss_bc": 0.02526623010635376, "train_loss_llm": 0.00032651424407958984, "grad_norm": 0.04077654704451561, "global_step": 9791, "epoch": 271, "lr": 0.001, "val_loss": 0.06035655736923218} +{"train_loss": 0.03969137743115425, "train_loss_bc": 0.03969137743115425, "train_loss_llm": 0.0, "grad_norm": 0.05019466206431389, "global_step": 9792, "epoch": 272, "lr": 0.001} +{"train_loss": 0.027910543605685234, "train_loss_bc": 0.02790042757987976, "train_loss_llm": 0.0010116398334503174, "grad_norm": 0.014664806425571442, "global_step": 9793, "epoch": 272, "lr": 0.001} +{"train_loss": 0.041362665593624115, "train_loss_bc": 0.041362665593624115, "train_loss_llm": 0.0, "grad_norm": 0.027170367538928986, "global_step": 9794, "epoch": 272, "lr": 0.001} +{"train_loss": 0.035910893231630325, "train_loss_bc": 0.035899944603443146, "train_loss_llm": 0.0010947585105895996, "grad_norm": 0.043325137346982956, "global_step": 9795, "epoch": 272, "lr": 0.001} +{"train_loss": 0.04027911275625229, "train_loss_bc": 0.040258608758449554, "train_loss_llm": 0.0020504891872406006, "grad_norm": 0.0546778179705143, "global_step": 9796, "epoch": 272, "lr": 0.001} +{"train_loss": 0.042662471532821655, "train_loss_bc": 0.042655784636735916, "train_loss_llm": 0.0006687939167022705, "grad_norm": 0.030866535380482674, "global_step": 9797, "epoch": 272, "lr": 0.001} +{"train_loss": 0.038315728306770325, "train_loss_bc": 0.038315728306770325, "train_loss_llm": 0.0, "grad_norm": 0.04706647992134094, "global_step": 9798, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03501025214791298, "train_loss_bc": 0.03501025214791298, "train_loss_llm": 0.0, "grad_norm": 0.053203098475933075, "global_step": 9799, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03550441190600395, "train_loss_bc": 0.03549637272953987, "train_loss_llm": 0.0008039772510528564, "grad_norm": 0.07542534917593002, "global_step": 9800, "epoch": 272, "lr": 0.001} +{"train_loss": 0.0427609346807003, "train_loss_bc": 0.0427609346807003, "train_loss_llm": 0.0, "grad_norm": 0.017112301662564278, "global_step": 9801, "epoch": 272, "lr": 0.001} +{"train_loss": 0.039418067783117294, "train_loss_bc": 0.039418067783117294, "train_loss_llm": 0.0, "grad_norm": 0.03089393489062786, "global_step": 9802, "epoch": 272, "lr": 0.001} +{"train_loss": 0.04339706152677536, "train_loss_bc": 0.04339706152677536, "train_loss_llm": 0.0, "grad_norm": 0.04426230862736702, "global_step": 9803, "epoch": 272, "lr": 0.001} +{"train_loss": 0.04218022897839546, "train_loss_bc": 0.04218022897839546, "train_loss_llm": 0.0, "grad_norm": 0.050221435725688934, "global_step": 9804, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03585609048604965, "train_loss_bc": 0.03585609048604965, "train_loss_llm": 0.0, "grad_norm": 0.01905178278684616, "global_step": 9805, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03729486092925072, "train_loss_bc": 0.03729486092925072, "train_loss_llm": 0.0, "grad_norm": 0.028642872348427773, "global_step": 9806, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03519907221198082, "train_loss_bc": 0.03519907221198082, "train_loss_llm": 0.0, "grad_norm": 0.040473032742738724, "global_step": 9807, "epoch": 272, "lr": 0.001} +{"train_loss": 0.029964005574584007, "train_loss_bc": 0.029937144368886948, "train_loss_llm": 0.0026862025260925293, "grad_norm": 0.04823951795697212, "global_step": 9808, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03562335669994354, "train_loss_bc": 0.03562335669994354, "train_loss_llm": 0.0, "grad_norm": 0.016520392149686813, "global_step": 9809, "epoch": 272, "lr": 0.001} +{"train_loss": 0.041435226798057556, "train_loss_bc": 0.041435226798057556, "train_loss_llm": 0.0, "grad_norm": 0.03145971894264221, "global_step": 9810, "epoch": 272, "lr": 0.001} +{"train_loss": 0.033978771418333054, "train_loss_bc": 0.03395354002714157, "train_loss_llm": 0.0025231540203094482, "grad_norm": 0.03987026959657669, "global_step": 9811, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03594667837023735, "train_loss_bc": 0.03594326227903366, "train_loss_llm": 0.00034174323081970215, "grad_norm": 0.04490043967962265, "global_step": 9812, "epoch": 272, "lr": 0.001} +{"train_loss": 0.0444616973400116, "train_loss_bc": 0.0444616973400116, "train_loss_llm": 0.0, "grad_norm": 0.020132865756750107, "global_step": 9813, "epoch": 272, "lr": 0.001} +{"train_loss": 0.04503484070301056, "train_loss_bc": 0.04501382261514664, "train_loss_llm": 0.002101719379425049, "grad_norm": 0.03565453365445137, "global_step": 9814, "epoch": 272, "lr": 0.001} +{"train_loss": 0.04179989546537399, "train_loss_bc": 0.04179989546537399, "train_loss_llm": 0.0, "grad_norm": 0.04359978064894676, "global_step": 9815, "epoch": 272, "lr": 0.001} +{"train_loss": 0.044577568769454956, "train_loss_bc": 0.04457169771194458, "train_loss_llm": 0.00058707594871521, "grad_norm": 0.048444874584674835, "global_step": 9816, "epoch": 272, "lr": 0.001} +{"train_loss": 0.043356675654649734, "train_loss_bc": 0.043356675654649734, "train_loss_llm": 0.0, "grad_norm": 0.019501976668834686, "global_step": 9817, "epoch": 272, "lr": 0.001} +{"train_loss": 0.04077965021133423, "train_loss_bc": 0.04077965021133423, "train_loss_llm": 0.0, "grad_norm": 0.03173185884952545, "global_step": 9818, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03551563620567322, "train_loss_bc": 0.03551563620567322, "train_loss_llm": 0.0, "grad_norm": 0.040845151990652084, "global_step": 9819, "epoch": 272, "lr": 0.001} +{"train_loss": 0.036705367267131805, "train_loss_bc": 0.036705367267131805, "train_loss_llm": 0.0, "grad_norm": 0.04435030743479729, "global_step": 9820, "epoch": 272, "lr": 0.001} +{"train_loss": 0.035281650722026825, "train_loss_bc": 0.035281650722026825, "train_loss_llm": 0.0, "grad_norm": 0.015344622544944286, "global_step": 9821, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03672792389988899, "train_loss_bc": 0.03672792389988899, "train_loss_llm": 0.0, "grad_norm": 0.026056818664073944, "global_step": 9822, "epoch": 272, "lr": 0.001} +{"train_loss": 0.044315896928310394, "train_loss_bc": 0.044315896928310394, "train_loss_llm": 0.0, "grad_norm": 0.03757821395993233, "global_step": 9823, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03910687193274498, "train_loss_bc": 0.03910687193274498, "train_loss_llm": 0.0, "grad_norm": 0.04687639698386192, "global_step": 9824, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03614911064505577, "train_loss_bc": 0.03612880781292915, "train_loss_llm": 0.0020302534103393555, "grad_norm": 0.015818225219845772, "global_step": 9825, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03788626566529274, "train_loss_bc": 0.03788051754236221, "train_loss_llm": 0.0005747973918914795, "grad_norm": 0.029199685901403427, "global_step": 9826, "epoch": 272, "lr": 0.001} +{"train_loss": 0.03852404995510975, "train_loss_bc": 0.03546493500471115, "train_loss_llm": 0.0, "grad_norm": 0.041071515530347824, "global_step": 9827, "epoch": 272, "lr": 0.001, "val_loss": 0.06050950661301613} +{"train_loss": 0.03580860793590546, "train_loss_bc": 0.03580164164304733, "train_loss_llm": 0.000696718692779541, "grad_norm": 0.05040764808654785, "global_step": 9828, "epoch": 273, "lr": 0.001} +{"train_loss": 0.04553523287177086, "train_loss_bc": 0.04553523287177086, "train_loss_llm": 0.0, "grad_norm": 0.027017388492822647, "global_step": 9829, "epoch": 273, "lr": 0.001} +{"train_loss": 0.04277012124657631, "train_loss_bc": 0.04277012124657631, "train_loss_llm": 0.0, "grad_norm": 0.03469597548246384, "global_step": 9830, "epoch": 273, "lr": 0.001} +{"train_loss": 0.04293863847851753, "train_loss_bc": 0.04293863847851753, "train_loss_llm": 0.0, "grad_norm": 0.04326815903186798, "global_step": 9831, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03066018596291542, "train_loss_bc": 0.03066018596291542, "train_loss_llm": 0.0, "grad_norm": 0.04501005634665489, "global_step": 9832, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03830283507704735, "train_loss_bc": 0.03830283507704735, "train_loss_llm": 0.0, "grad_norm": 0.017897121608257294, "global_step": 9833, "epoch": 273, "lr": 0.001} +{"train_loss": 0.0364156998693943, "train_loss_bc": 0.0364156998693943, "train_loss_llm": 0.0, "grad_norm": 0.026764387264847755, "global_step": 9834, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03987077251076698, "train_loss_bc": 0.03987077251076698, "train_loss_llm": 0.0, "grad_norm": 0.04371389001607895, "global_step": 9835, "epoch": 273, "lr": 0.001} +{"train_loss": 0.04382040724158287, "train_loss_bc": 0.04382040724158287, "train_loss_llm": 0.0, "grad_norm": 0.053053952753543854, "global_step": 9836, "epoch": 273, "lr": 0.001} +{"train_loss": 0.044815972447395325, "train_loss_bc": 0.044815972447395325, "train_loss_llm": 0.0, "grad_norm": 0.018726106733083725, "global_step": 9837, "epoch": 273, "lr": 0.001} +{"train_loss": 0.034660160541534424, "train_loss_bc": 0.034660160541534424, "train_loss_llm": 0.0, "grad_norm": 0.02705889753997326, "global_step": 9838, "epoch": 273, "lr": 0.001} +{"train_loss": 0.04012755677103996, "train_loss_bc": 0.04012755677103996, "train_loss_llm": 0.0, "grad_norm": 0.03993324935436249, "global_step": 9839, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03944928199052811, "train_loss_bc": 0.03944387659430504, "train_loss_llm": 0.0005405843257904053, "grad_norm": 0.05457969382405281, "global_step": 9840, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03386857360601425, "train_loss_bc": 0.03386642783880234, "train_loss_llm": 0.00021457672119140625, "grad_norm": 0.018891414627432823, "global_step": 9841, "epoch": 273, "lr": 0.001} +{"train_loss": 0.0404234305024147, "train_loss_bc": 0.0404234305024147, "train_loss_llm": 0.0, "grad_norm": 0.03076552413403988, "global_step": 9842, "epoch": 273, "lr": 0.001} +{"train_loss": 0.036285243928432465, "train_loss_bc": 0.036285243928432465, "train_loss_llm": 0.0, "grad_norm": 0.03729100525379181, "global_step": 9843, "epoch": 273, "lr": 0.001} +{"train_loss": 0.04026522859930992, "train_loss_bc": 0.04026522859930992, "train_loss_llm": 0.0, "grad_norm": 0.04478831589221954, "global_step": 9844, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03432605415582657, "train_loss_bc": 0.03432605415582657, "train_loss_llm": 0.0, "grad_norm": 0.017345258966088295, "global_step": 9845, "epoch": 273, "lr": 0.001} +{"train_loss": 0.041934721171855927, "train_loss_bc": 0.04192974790930748, "train_loss_llm": 0.0004973113536834717, "grad_norm": 0.03159422427415848, "global_step": 9846, "epoch": 273, "lr": 0.001} +{"train_loss": 0.031075607985258102, "train_loss_bc": 0.03106488659977913, "train_loss_llm": 0.001072227954864502, "grad_norm": 0.033989906311035156, "global_step": 9847, "epoch": 273, "lr": 0.001} +{"train_loss": 0.039479341357946396, "train_loss_bc": 0.03946434333920479, "train_loss_llm": 0.0014996826648712158, "grad_norm": 0.03891782462596893, "global_step": 9848, "epoch": 273, "lr": 0.001} +{"train_loss": 0.0393657311797142, "train_loss_bc": 0.0393657311797142, "train_loss_llm": 0.0, "grad_norm": 0.018537767231464386, "global_step": 9849, "epoch": 273, "lr": 0.001} +{"train_loss": 0.0374099425971508, "train_loss_bc": 0.0374099425971508, "train_loss_llm": 0.0, "grad_norm": 0.03105100430548191, "global_step": 9850, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03551440313458443, "train_loss_bc": 0.03551440313458443, "train_loss_llm": 0.0, "grad_norm": 0.03693855181336403, "global_step": 9851, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03460472449660301, "train_loss_bc": 0.03460472449660301, "train_loss_llm": 0.0, "grad_norm": 0.04336284473538399, "global_step": 9852, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03945127874612808, "train_loss_bc": 0.03942371904850006, "train_loss_llm": 0.0027558505535125732, "grad_norm": 0.01692911982536316, "global_step": 9853, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03345277160406113, "train_loss_bc": 0.03345277160406113, "train_loss_llm": 0.0, "grad_norm": 0.02848534844815731, "global_step": 9854, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03893033787608147, "train_loss_bc": 0.038922421634197235, "train_loss_llm": 0.0007917881011962891, "grad_norm": 0.03498665615916252, "global_step": 9855, "epoch": 273, "lr": 0.001} +{"train_loss": 0.039993446320295334, "train_loss_bc": 0.03997205197811127, "train_loss_llm": 0.002139478921890259, "grad_norm": 0.04457725211977959, "global_step": 9856, "epoch": 273, "lr": 0.001} +{"train_loss": 0.040124185383319855, "train_loss_bc": 0.040124185383319855, "train_loss_llm": 0.0, "grad_norm": 0.021354790776968002, "global_step": 9857, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03464711830019951, "train_loss_bc": 0.03464711830019951, "train_loss_llm": 0.0, "grad_norm": 0.029976043850183487, "global_step": 9858, "epoch": 273, "lr": 0.001} +{"train_loss": 0.04056635871529579, "train_loss_bc": 0.0405542328953743, "train_loss_llm": 0.0012127459049224854, "grad_norm": 0.039624180644750595, "global_step": 9859, "epoch": 273, "lr": 0.001} +{"train_loss": 0.03194466233253479, "train_loss_bc": 0.03194466233253479, "train_loss_llm": 0.0, "grad_norm": 0.04438154399394989, "global_step": 9860, "epoch": 273, "lr": 0.001} +{"train_loss": 0.04124021530151367, "train_loss_bc": 0.04124021530151367, "train_loss_llm": 0.0, "grad_norm": 0.019506391137838364, "global_step": 9861, "epoch": 273, "lr": 0.001} +{"train_loss": 0.04340022802352905, "train_loss_bc": 0.04337858408689499, "train_loss_llm": 0.002164393663406372, "grad_norm": 0.027188491076231003, "global_step": 9862, "epoch": 273, "lr": 0.001} +{"train_loss": 0.038356836471292705, "train_loss_bc": 0.03736703470349312, "train_loss_llm": 0.0, "grad_norm": 0.03920837491750717, "global_step": 9863, "epoch": 273, "lr": 0.001, "val_loss": 0.060776591300964355} +{"train_loss": 0.04292626306414604, "train_loss_bc": 0.042914580553770065, "train_loss_llm": 0.001168370246887207, "grad_norm": 0.05254754424095154, "global_step": 9864, "epoch": 274, "lr": 0.001} +{"train_loss": 0.0371873676776886, "train_loss_bc": 0.0371873676776886, "train_loss_llm": 0.0, "grad_norm": 0.024195939302444458, "global_step": 9865, "epoch": 274, "lr": 0.001} +{"train_loss": 0.045830387622117996, "train_loss_bc": 0.045830387622117996, "train_loss_llm": 0.0, "grad_norm": 0.033971406519412994, "global_step": 9866, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03526909276843071, "train_loss_bc": 0.03526909276843071, "train_loss_llm": 0.0, "grad_norm": 0.04748213663697243, "global_step": 9867, "epoch": 274, "lr": 0.001} +{"train_loss": 0.043366141617298126, "train_loss_bc": 0.043366141617298126, "train_loss_llm": 0.0, "grad_norm": 0.058281783014535904, "global_step": 9868, "epoch": 274, "lr": 0.001} +{"train_loss": 0.045147866010665894, "train_loss_bc": 0.045147866010665894, "train_loss_llm": 0.0, "grad_norm": 0.023593783378601074, "global_step": 9869, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03134766221046448, "train_loss_bc": 0.03134766221046448, "train_loss_llm": 0.0, "grad_norm": 0.0312346164137125, "global_step": 9870, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03829715773463249, "train_loss_bc": 0.03828601539134979, "train_loss_llm": 0.00111427903175354, "grad_norm": 0.03725519776344299, "global_step": 9871, "epoch": 274, "lr": 0.001} +{"train_loss": 0.04378056898713112, "train_loss_bc": 0.04378056898713112, "train_loss_llm": 0.0, "grad_norm": 0.04322148486971855, "global_step": 9872, "epoch": 274, "lr": 0.001} +{"train_loss": 0.04110309109091759, "train_loss_bc": 0.04110309109091759, "train_loss_llm": 0.0, "grad_norm": 0.025620274245738983, "global_step": 9873, "epoch": 274, "lr": 0.001} +{"train_loss": 0.038198020309209824, "train_loss_bc": 0.038198020309209824, "train_loss_llm": 0.0, "grad_norm": 0.03568415343761444, "global_step": 9874, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03354327008128166, "train_loss_bc": 0.03354327008128166, "train_loss_llm": 0.0, "grad_norm": 0.0445510670542717, "global_step": 9875, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03438828885555267, "train_loss_bc": 0.034387748688459396, "train_loss_llm": 5.3942203521728516e-05, "grad_norm": 0.04767633602023125, "global_step": 9876, "epoch": 274, "lr": 0.001} +{"train_loss": 0.038852907717227936, "train_loss_bc": 0.038852907717227936, "train_loss_llm": 0.0, "grad_norm": 0.018398290500044823, "global_step": 9877, "epoch": 274, "lr": 0.001} +{"train_loss": 0.035524219274520874, "train_loss_bc": 0.035524219274520874, "train_loss_llm": 0.0, "grad_norm": 0.026283985003829002, "global_step": 9878, "epoch": 274, "lr": 0.001} +{"train_loss": 0.030543284490704536, "train_loss_bc": 0.030543284490704536, "train_loss_llm": 0.0, "grad_norm": 0.03317561000585556, "global_step": 9879, "epoch": 274, "lr": 0.001} +{"train_loss": 0.031043685972690582, "train_loss_bc": 0.031031429767608643, "train_loss_llm": 0.0012257099151611328, "grad_norm": 0.04318416491150856, "global_step": 9880, "epoch": 274, "lr": 0.001} +{"train_loss": 0.034416694194078445, "train_loss_bc": 0.034416694194078445, "train_loss_llm": 0.0, "grad_norm": 0.02440444380044937, "global_step": 9881, "epoch": 274, "lr": 0.001} +{"train_loss": 0.04309871420264244, "train_loss_bc": 0.04309871420264244, "train_loss_llm": 0.0, "grad_norm": 0.04053174704313278, "global_step": 9882, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03566278517246246, "train_loss_bc": 0.03566278517246246, "train_loss_llm": 0.0, "grad_norm": 0.05531533062458038, "global_step": 9883, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03575311601161957, "train_loss_bc": 0.03572199493646622, "train_loss_llm": 0.00311201810836792, "grad_norm": 0.06069597601890564, "global_step": 9884, "epoch": 274, "lr": 0.001} +{"train_loss": 0.039535798132419586, "train_loss_bc": 0.039535798132419586, "train_loss_llm": 0.0, "grad_norm": 0.017635146155953407, "global_step": 9885, "epoch": 274, "lr": 0.001} +{"train_loss": 0.040137309581041336, "train_loss_bc": 0.040137309581041336, "train_loss_llm": 0.0, "grad_norm": 0.032718226313591, "global_step": 9886, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03827337548136711, "train_loss_bc": 0.03827337548136711, "train_loss_llm": 0.0, "grad_norm": 0.04233254864811897, "global_step": 9887, "epoch": 274, "lr": 0.001} +{"train_loss": 0.04323605075478554, "train_loss_bc": 0.04323605075478554, "train_loss_llm": 0.0, "grad_norm": 0.05125121772289276, "global_step": 9888, "epoch": 274, "lr": 0.001} +{"train_loss": 0.04376336187124252, "train_loss_bc": 0.04376336187124252, "train_loss_llm": 0.0, "grad_norm": 0.018003789708018303, "global_step": 9889, "epoch": 274, "lr": 0.001} +{"train_loss": 0.04622747376561165, "train_loss_bc": 0.046213969588279724, "train_loss_llm": 0.0013504624366760254, "grad_norm": 0.02960890717804432, "global_step": 9890, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03529373183846474, "train_loss_bc": 0.03529373183846474, "train_loss_llm": 0.0, "grad_norm": 0.039614658802747726, "global_step": 9891, "epoch": 274, "lr": 0.001} +{"train_loss": 0.028264835476875305, "train_loss_bc": 0.02825777232646942, "train_loss_llm": 0.0007063746452331543, "grad_norm": 0.04893317446112633, "global_step": 9892, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03294328600168228, "train_loss_bc": 0.03294328600168228, "train_loss_llm": 0.0, "grad_norm": 0.018978573381900787, "global_step": 9893, "epoch": 274, "lr": 0.001} +{"train_loss": 0.04027071222662926, "train_loss_bc": 0.04027071222662926, "train_loss_llm": 0.0, "grad_norm": 0.02770826406776905, "global_step": 9894, "epoch": 274, "lr": 0.001} +{"train_loss": 0.04533705115318298, "train_loss_bc": 0.04533402994275093, "train_loss_llm": 0.00030228495597839355, "grad_norm": 0.04345077648758888, "global_step": 9895, "epoch": 274, "lr": 0.001} +{"train_loss": 0.04530993103981018, "train_loss_bc": 0.04530496895313263, "train_loss_llm": 0.000496208667755127, "grad_norm": 0.050579220056533813, "global_step": 9896, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03726937994360924, "train_loss_bc": 0.03726937994360924, "train_loss_llm": 0.0, "grad_norm": 0.01912865974009037, "global_step": 9897, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03402552753686905, "train_loss_bc": 0.03402552753686905, "train_loss_llm": 0.0, "grad_norm": 0.028633393347263336, "global_step": 9898, "epoch": 274, "lr": 0.001} +{"train_loss": 0.03855137407986654, "train_loss_bc": 0.04268105700612068, "train_loss_llm": 0.0, "grad_norm": 0.05149852856993675, "global_step": 9899, "epoch": 274, "lr": 0.001, "val_loss": 0.06082521378993988} +{"train_loss": 0.03548222407698631, "train_loss_bc": 0.03546016290783882, "train_loss_llm": 0.0022059381008148193, "grad_norm": 0.05566577613353729, "global_step": 9900, "epoch": 275, "lr": 0.001} +{"train_loss": 0.0392029844224453, "train_loss_bc": 0.0392029844224453, "train_loss_llm": 0.0, "grad_norm": 0.017985882237553596, "global_step": 9901, "epoch": 275, "lr": 0.001} +{"train_loss": 0.034593742340803146, "train_loss_bc": 0.034593742340803146, "train_loss_llm": 0.0, "grad_norm": 0.02986053191125393, "global_step": 9902, "epoch": 275, "lr": 0.001} +{"train_loss": 0.04445035010576248, "train_loss_bc": 0.04445035010576248, "train_loss_llm": 0.0, "grad_norm": 0.04709925875067711, "global_step": 9903, "epoch": 275, "lr": 0.001} +{"train_loss": 0.04284243658185005, "train_loss_bc": 0.04284243658185005, "train_loss_llm": 0.0, "grad_norm": 0.06342513859272003, "global_step": 9904, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03580809012055397, "train_loss_bc": 0.03580809012055397, "train_loss_llm": 0.0, "grad_norm": 0.018509063869714737, "global_step": 9905, "epoch": 275, "lr": 0.001} +{"train_loss": 0.039688024669885635, "train_loss_bc": 0.039688024669885635, "train_loss_llm": 0.0, "grad_norm": 0.026521241292357445, "global_step": 9906, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03890840709209442, "train_loss_bc": 0.03890840709209442, "train_loss_llm": 0.0, "grad_norm": 0.035649918019771576, "global_step": 9907, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03857565298676491, "train_loss_bc": 0.03857025131583214, "train_loss_llm": 0.0005402565002441406, "grad_norm": 0.0392535999417305, "global_step": 9908, "epoch": 275, "lr": 0.001} +{"train_loss": 0.027429720386862755, "train_loss_bc": 0.027429720386862755, "train_loss_llm": 0.0, "grad_norm": 0.015744652599096298, "global_step": 9909, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03464195877313614, "train_loss_bc": 0.0346328504383564, "train_loss_llm": 0.0009109079837799072, "grad_norm": 0.022295396775007248, "global_step": 9910, "epoch": 275, "lr": 0.001} +{"train_loss": 0.040725257247686386, "train_loss_bc": 0.040725257247686386, "train_loss_llm": 0.0, "grad_norm": 0.032599784433841705, "global_step": 9911, "epoch": 275, "lr": 0.001} +{"train_loss": 0.035453423857688904, "train_loss_bc": 0.035453423857688904, "train_loss_llm": 0.0, "grad_norm": 0.04419620707631111, "global_step": 9912, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03545748069882393, "train_loss_bc": 0.03545505553483963, "train_loss_llm": 0.00024244189262390137, "grad_norm": 0.018656650558114052, "global_step": 9913, "epoch": 275, "lr": 0.001} +{"train_loss": 0.04293528199195862, "train_loss_bc": 0.04289766028523445, "train_loss_llm": 0.0037622153759002686, "grad_norm": 0.03990441560745239, "global_step": 9914, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03970978036522865, "train_loss_bc": 0.03970978036522865, "train_loss_llm": 0.0, "grad_norm": 0.04733927175402641, "global_step": 9915, "epoch": 275, "lr": 0.001} +{"train_loss": 0.043613288551568985, "train_loss_bc": 0.043608155101537704, "train_loss_llm": 0.000513225793838501, "grad_norm": 0.05273142084479332, "global_step": 9916, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03326168283820152, "train_loss_bc": 0.03326168283820152, "train_loss_llm": 0.0, "grad_norm": 0.015680482611060143, "global_step": 9917, "epoch": 275, "lr": 0.001} +{"train_loss": 0.035832758992910385, "train_loss_bc": 0.035832758992910385, "train_loss_llm": 0.0, "grad_norm": 0.03480864688754082, "global_step": 9918, "epoch": 275, "lr": 0.001} +{"train_loss": 0.043799061328172684, "train_loss_bc": 0.043799061328172684, "train_loss_llm": 0.0, "grad_norm": 0.048282552510499954, "global_step": 9919, "epoch": 275, "lr": 0.001} +{"train_loss": 0.041647762060165405, "train_loss_bc": 0.041635941714048386, "train_loss_llm": 0.001181960105895996, "grad_norm": 0.059564217925071716, "global_step": 9920, "epoch": 275, "lr": 0.001} +{"train_loss": 0.0396982878446579, "train_loss_bc": 0.0396982878446579, "train_loss_llm": 0.0, "grad_norm": 0.020376356318593025, "global_step": 9921, "epoch": 275, "lr": 0.001} +{"train_loss": 0.04527302458882332, "train_loss_bc": 0.04527302458882332, "train_loss_llm": 0.0, "grad_norm": 0.03618459776043892, "global_step": 9922, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03856663033366203, "train_loss_bc": 0.03856663033366203, "train_loss_llm": 0.0, "grad_norm": 0.04812421649694443, "global_step": 9923, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03983079642057419, "train_loss_bc": 0.03983079642057419, "train_loss_llm": 0.0, "grad_norm": 0.0627957358956337, "global_step": 9924, "epoch": 275, "lr": 0.001} +{"train_loss": 0.04416428133845329, "train_loss_bc": 0.04416428133845329, "train_loss_llm": 0.0, "grad_norm": 0.022061094641685486, "global_step": 9925, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03521173819899559, "train_loss_bc": 0.03520093113183975, "train_loss_llm": 0.0010807216167449951, "grad_norm": 0.02996152825653553, "global_step": 9926, "epoch": 275, "lr": 0.001} +{"train_loss": 0.04385748133063316, "train_loss_bc": 0.04385748133063316, "train_loss_llm": 0.0, "grad_norm": 0.040970753878355026, "global_step": 9927, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03137907758355141, "train_loss_bc": 0.03137907758355141, "train_loss_llm": 0.0, "grad_norm": 0.05385425686836243, "global_step": 9928, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03362177312374115, "train_loss_bc": 0.03362177312374115, "train_loss_llm": 0.0, "grad_norm": 0.017161233350634575, "global_step": 9929, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03760362043976784, "train_loss_bc": 0.03760362043976784, "train_loss_llm": 0.0, "grad_norm": 0.026895197108387947, "global_step": 9930, "epoch": 275, "lr": 0.001} +{"train_loss": 0.03755031153559685, "train_loss_bc": 0.03755031153559685, "train_loss_llm": 0.0, "grad_norm": 0.03470414876937866, "global_step": 9931, "epoch": 275, "lr": 0.001} +{"train_loss": 0.04067479446530342, "train_loss_bc": 0.04067354276776314, "train_loss_llm": 0.0001252293586730957, "grad_norm": 0.04373910278081894, "global_step": 9932, "epoch": 275, "lr": 0.001} +{"train_loss": 0.04111384600400925, "train_loss_bc": 0.041112542152404785, "train_loss_llm": 0.00013047456741333008, "grad_norm": 0.025648074224591255, "global_step": 9933, "epoch": 275, "lr": 0.001} +{"train_loss": 0.036683786660432816, "train_loss_bc": 0.036683786660432816, "train_loss_llm": 0.0, "grad_norm": 0.038775596767663956, "global_step": 9934, "epoch": 275, "lr": 0.001} +{"train_loss": 0.038430648493684, "train_loss_bc": 0.03420369327068329, "train_loss_llm": 0.0010831356048583984, "grad_norm": 0.04399817809462547, "global_step": 9935, "epoch": 275, "lr": 0.001, "train/cumulative_reward": 128.53102553302125, "train/mean_score": 17.91642705963276, "train/success_rate": 0.4, "test/cumulative_reward": 72.83245220385028, "test/mean_score": 9.708775130796, "test/success_rate": 0.2, "val_loss": 0.060288239270448685, "train_action_mse_error": 0.04182874411344528} +{"train_loss": 0.0382872074842453, "train_loss_bc": 0.0382872074842453, "train_loss_llm": 0.0, "grad_norm": 0.0499672032892704, "global_step": 9936, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03518199920654297, "train_loss_bc": 0.03518199920654297, "train_loss_llm": 0.0, "grad_norm": 0.01749321073293686, "global_step": 9937, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03829948976635933, "train_loss_bc": 0.03827304393053055, "train_loss_llm": 0.002644747495651245, "grad_norm": 0.02724589966237545, "global_step": 9938, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03394393250346184, "train_loss_bc": 0.03394393250346184, "train_loss_llm": 0.0, "grad_norm": 0.03782539814710617, "global_step": 9939, "epoch": 276, "lr": 0.001} +{"train_loss": 0.04054756835103035, "train_loss_bc": 0.04054756835103035, "train_loss_llm": 0.0, "grad_norm": 0.05650603771209717, "global_step": 9940, "epoch": 276, "lr": 0.001} +{"train_loss": 0.028268735855817795, "train_loss_bc": 0.028254251927137375, "train_loss_llm": 0.0014483332633972168, "grad_norm": 0.01829027757048607, "global_step": 9941, "epoch": 276, "lr": 0.001} +{"train_loss": 0.0372416116297245, "train_loss_bc": 0.0372416116297245, "train_loss_llm": 0.0, "grad_norm": 0.03440474718809128, "global_step": 9942, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03223668411374092, "train_loss_bc": 0.03223235160112381, "train_loss_llm": 0.00043332576751708984, "grad_norm": 0.039713822305202484, "global_step": 9943, "epoch": 276, "lr": 0.001} +{"train_loss": 0.04078759625554085, "train_loss_bc": 0.04078759625554085, "train_loss_llm": 0.0, "grad_norm": 0.05307196080684662, "global_step": 9944, "epoch": 276, "lr": 0.001} +{"train_loss": 0.027262398973107338, "train_loss_bc": 0.027262398973107338, "train_loss_llm": 0.0, "grad_norm": 0.015371985733509064, "global_step": 9945, "epoch": 276, "lr": 0.001} +{"train_loss": 0.04541054368019104, "train_loss_bc": 0.04541054368019104, "train_loss_llm": 0.0, "grad_norm": 0.033963415771722794, "global_step": 9946, "epoch": 276, "lr": 0.001} +{"train_loss": 0.039290640503168106, "train_loss_bc": 0.039290640503168106, "train_loss_llm": 0.0, "grad_norm": 0.04287222400307655, "global_step": 9947, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03928890451788902, "train_loss_bc": 0.03926273435354233, "train_loss_llm": 0.0026171505451202393, "grad_norm": 0.04501822590827942, "global_step": 9948, "epoch": 276, "lr": 0.001} +{"train_loss": 0.042015254497528076, "train_loss_bc": 0.042015254497528076, "train_loss_llm": 0.0, "grad_norm": 0.019320113584399223, "global_step": 9949, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03888436406850815, "train_loss_bc": 0.038867443799972534, "train_loss_llm": 0.001692056655883789, "grad_norm": 0.029776332899928093, "global_step": 9950, "epoch": 276, "lr": 0.001} +{"train_loss": 0.04000236093997955, "train_loss_bc": 0.04000236093997955, "train_loss_llm": 0.0, "grad_norm": 0.037231046706438065, "global_step": 9951, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03875749558210373, "train_loss_bc": 0.03874475508928299, "train_loss_llm": 0.00127410888671875, "grad_norm": 0.045305754989385605, "global_step": 9952, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03868306055665016, "train_loss_bc": 0.03868306055665016, "train_loss_llm": 0.0, "grad_norm": 0.01941370591521263, "global_step": 9953, "epoch": 276, "lr": 0.001} +{"train_loss": 0.041952747851610184, "train_loss_bc": 0.041952747851610184, "train_loss_llm": 0.0, "grad_norm": 0.033918172121047974, "global_step": 9954, "epoch": 276, "lr": 0.001} +{"train_loss": 0.044471658766269684, "train_loss_bc": 0.04445594549179077, "train_loss_llm": 0.0015713274478912354, "grad_norm": 0.04202153906226158, "global_step": 9955, "epoch": 276, "lr": 0.001} +{"train_loss": 0.04683502018451691, "train_loss_bc": 0.04683502018451691, "train_loss_llm": 0.0, "grad_norm": 0.05212761461734772, "global_step": 9956, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03882303833961487, "train_loss_bc": 0.0388195775449276, "train_loss_llm": 0.0003460049629211426, "grad_norm": 0.018429912626743317, "global_step": 9957, "epoch": 276, "lr": 0.001} +{"train_loss": 0.040361735969781876, "train_loss_bc": 0.040361735969781876, "train_loss_llm": 0.0, "grad_norm": 0.030298979952931404, "global_step": 9958, "epoch": 276, "lr": 0.001} +{"train_loss": 0.04357924684882164, "train_loss_bc": 0.04350744187831879, "train_loss_llm": 0.007180660963058472, "grad_norm": 0.04193735122680664, "global_step": 9959, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03807083144783974, "train_loss_bc": 0.03807083144783974, "train_loss_llm": 0.0, "grad_norm": 0.0502738319337368, "global_step": 9960, "epoch": 276, "lr": 0.001} +{"train_loss": 0.039665088057518005, "train_loss_bc": 0.039665088057518005, "train_loss_llm": 0.0, "grad_norm": 0.018659302964806557, "global_step": 9961, "epoch": 276, "lr": 0.001} +{"train_loss": 0.04325596243143082, "train_loss_bc": 0.04325596243143082, "train_loss_llm": 0.0, "grad_norm": 0.03412993624806404, "global_step": 9962, "epoch": 276, "lr": 0.001} +{"train_loss": 0.026745229959487915, "train_loss_bc": 0.026745229959487915, "train_loss_llm": 0.0, "grad_norm": 0.042273491621017456, "global_step": 9963, "epoch": 276, "lr": 0.001} +{"train_loss": 0.04097553342580795, "train_loss_bc": 0.04097019135951996, "train_loss_llm": 0.0005341768264770508, "grad_norm": 0.05342083051800728, "global_step": 9964, "epoch": 276, "lr": 0.001} +{"train_loss": 0.04278762638568878, "train_loss_bc": 0.04278762638568878, "train_loss_llm": 0.0, "grad_norm": 0.024048779159784317, "global_step": 9965, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03897175192832947, "train_loss_bc": 0.03897175192832947, "train_loss_llm": 0.0, "grad_norm": 0.034990422427654266, "global_step": 9966, "epoch": 276, "lr": 0.001} +{"train_loss": 0.0325719453394413, "train_loss_bc": 0.0325719453394413, "train_loss_llm": 0.0, "grad_norm": 0.04562431573867798, "global_step": 9967, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03869489207863808, "train_loss_bc": 0.03869489207863808, "train_loss_llm": 0.0, "grad_norm": 0.05200276896357536, "global_step": 9968, "epoch": 276, "lr": 0.001} +{"train_loss": 0.038967546075582504, "train_loss_bc": 0.038967546075582504, "train_loss_llm": 0.0, "grad_norm": 0.020008165389299393, "global_step": 9969, "epoch": 276, "lr": 0.001} +{"train_loss": 0.03596527874469757, "train_loss_bc": 0.03596527874469757, "train_loss_llm": 0.0, "grad_norm": 0.0326089970767498, "global_step": 9970, "epoch": 276, "lr": 0.001} +{"train_loss": 0.038435465190559626, "train_loss_bc": 0.03659176453948021, "train_loss_llm": 0.0, "grad_norm": 0.04622555524110794, "global_step": 9971, "epoch": 276, "lr": 0.001, "val_loss": 0.06067158654332161} +{"train_loss": 0.03643618896603584, "train_loss_bc": 0.03643618896603584, "train_loss_llm": 0.0, "grad_norm": 0.053663965314626694, "global_step": 9972, "epoch": 277, "lr": 0.001} +{"train_loss": 0.032155949622392654, "train_loss_bc": 0.03213635832071304, "train_loss_llm": 0.00195920467376709, "grad_norm": 0.01914825849235058, "global_step": 9973, "epoch": 277, "lr": 0.001} +{"train_loss": 0.039898380637168884, "train_loss_bc": 0.039898380637168884, "train_loss_llm": 0.0, "grad_norm": 0.03196931630373001, "global_step": 9974, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03824355825781822, "train_loss_bc": 0.03824355825781822, "train_loss_llm": 0.0, "grad_norm": 0.04060246795415878, "global_step": 9975, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03714955598115921, "train_loss_bc": 0.03714955598115921, "train_loss_llm": 0.0, "grad_norm": 0.04665946215391159, "global_step": 9976, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03684161230921745, "train_loss_bc": 0.03684161230921745, "train_loss_llm": 0.0, "grad_norm": 0.021450139582157135, "global_step": 9977, "epoch": 277, "lr": 0.001} +{"train_loss": 0.037701528519392014, "train_loss_bc": 0.037701528519392014, "train_loss_llm": 0.0, "grad_norm": 0.029229318723082542, "global_step": 9978, "epoch": 277, "lr": 0.001} +{"train_loss": 0.05088729411363602, "train_loss_bc": 0.05088729411363602, "train_loss_llm": 0.0, "grad_norm": 0.047281887382268906, "global_step": 9979, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03663202002644539, "train_loss_bc": 0.03663202002644539, "train_loss_llm": 0.0, "grad_norm": 0.056514810770750046, "global_step": 9980, "epoch": 277, "lr": 0.001} +{"train_loss": 0.0337214358150959, "train_loss_bc": 0.03371363505721092, "train_loss_llm": 0.0007800757884979248, "grad_norm": 0.020461436361074448, "global_step": 9981, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03455986827611923, "train_loss_bc": 0.03455986827611923, "train_loss_llm": 0.0, "grad_norm": 0.03075644001364708, "global_step": 9982, "epoch": 277, "lr": 0.001} +{"train_loss": 0.04024666175246239, "train_loss_bc": 0.04024666175246239, "train_loss_llm": 0.0, "grad_norm": 0.04335881397128105, "global_step": 9983, "epoch": 277, "lr": 0.001} +{"train_loss": 0.041128646582365036, "train_loss_bc": 0.041128646582365036, "train_loss_llm": 0.0, "grad_norm": 0.04720589891076088, "global_step": 9984, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03744043782353401, "train_loss_bc": 0.03744043782353401, "train_loss_llm": 0.0, "grad_norm": 0.017872385680675507, "global_step": 9985, "epoch": 277, "lr": 0.001} +{"train_loss": 0.034840915352106094, "train_loss_bc": 0.034825343638658524, "train_loss_llm": 0.00155717134475708, "grad_norm": 0.026636967435479164, "global_step": 9986, "epoch": 277, "lr": 0.001} +{"train_loss": 0.034216493368148804, "train_loss_bc": 0.034216493368148804, "train_loss_llm": 0.0, "grad_norm": 0.037721652537584305, "global_step": 9987, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03883182257413864, "train_loss_bc": 0.03883182257413864, "train_loss_llm": 0.0, "grad_norm": 0.04332022741436958, "global_step": 9988, "epoch": 277, "lr": 0.001} +{"train_loss": 0.033693354576826096, "train_loss_bc": 0.033693354576826096, "train_loss_llm": 0.0, "grad_norm": 0.019118113443255424, "global_step": 9989, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03990405052900314, "train_loss_bc": 0.03988545015454292, "train_loss_llm": 0.0018599331378936768, "grad_norm": 0.03083202615380287, "global_step": 9990, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03208853304386139, "train_loss_bc": 0.03208853304386139, "train_loss_llm": 0.0, "grad_norm": 0.04138179123401642, "global_step": 9991, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03557654842734337, "train_loss_bc": 0.03555329889059067, "train_loss_llm": 0.0023249387741088867, "grad_norm": 0.05198951065540314, "global_step": 9992, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03826853260397911, "train_loss_bc": 0.03826358914375305, "train_loss_llm": 0.0004945099353790283, "grad_norm": 0.02586047723889351, "global_step": 9993, "epoch": 277, "lr": 0.001} +{"train_loss": 0.042775023728609085, "train_loss_bc": 0.042775023728609085, "train_loss_llm": 0.0, "grad_norm": 0.036196909844875336, "global_step": 9994, "epoch": 277, "lr": 0.001} +{"train_loss": 0.039469752460718155, "train_loss_bc": 0.039469752460718155, "train_loss_llm": 0.0, "grad_norm": 0.03782008960843086, "global_step": 9995, "epoch": 277, "lr": 0.001} +{"train_loss": 0.033531319350004196, "train_loss_bc": 0.033531319350004196, "train_loss_llm": 0.0, "grad_norm": 0.04624587669968605, "global_step": 9996, "epoch": 277, "lr": 0.001} +{"train_loss": 0.04537029191851616, "train_loss_bc": 0.04537029191851616, "train_loss_llm": 0.0, "grad_norm": 0.023127995431423187, "global_step": 9997, "epoch": 277, "lr": 0.001} +{"train_loss": 0.04297564551234245, "train_loss_bc": 0.04297564551234245, "train_loss_llm": 0.0, "grad_norm": 0.031985118985176086, "global_step": 9998, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03198065608739853, "train_loss_bc": 0.03198065608739853, "train_loss_llm": 0.0, "grad_norm": 0.04199480637907982, "global_step": 9999, "epoch": 277, "lr": 0.001} +{"train_loss": 0.03428278863430023, "train_loss_bc": 0.03427280858159065, "train_loss_llm": 0.0009981095790863037, "grad_norm": 0.053274355828762054, "global_step": 10000, "epoch": 277, "lr": 0.001} +{"train_loss": 0.046505752950906754, "train_loss_bc": 0.046503063291311264, "train_loss_llm": 0.000268787145614624, "grad_norm": 0.02360066957771778, "global_step": 10001, "epoch": 277, "lr": 0.001} +{"train_loss": 0.04091981053352356, "train_loss_bc": 0.040907979011535645, "train_loss_llm": 0.00118333101272583, "grad_norm": 0.02818378433585167, "global_step": 10002, "epoch": 277, "lr": 0.001} +{"train_loss": 0.04046960175037384, "train_loss_bc": 0.04046960175037384, "train_loss_llm": 0.0, "grad_norm": 0.03963272273540497, "global_step": 10003, "epoch": 277, "lr": 0.001} +{"train_loss": 0.0363939069211483, "train_loss_bc": 0.03639364615082741, "train_loss_llm": 2.5898218154907227e-05, "grad_norm": 0.049016959965229034, "global_step": 10004, "epoch": 277, "lr": 0.001} +{"train_loss": 0.039571356028318405, "train_loss_bc": 0.039571356028318405, "train_loss_llm": 0.0, "grad_norm": 0.024201184511184692, "global_step": 10005, "epoch": 277, "lr": 0.001} +{"train_loss": 0.04660489782691002, "train_loss_bc": 0.046593453735113144, "train_loss_llm": 0.0011443793773651123, "grad_norm": 0.028549762442708015, "global_step": 10006, "epoch": 277, "lr": 0.001} +{"train_loss": 0.038329238485958844, "train_loss_bc": 0.038533855229616165, "train_loss_llm": 0.00045365095138549805, "grad_norm": 0.04062514379620552, "global_step": 10007, "epoch": 277, "lr": 0.001, "val_loss": 0.0604177862405777} +{"train_loss": 0.040602706372737885, "train_loss_bc": 0.040602706372737885, "train_loss_llm": 0.0, "grad_norm": 0.046352338045835495, "global_step": 10008, "epoch": 278, "lr": 0.001} +{"train_loss": 0.036139555275440216, "train_loss_bc": 0.036139555275440216, "train_loss_llm": 0.0, "grad_norm": 0.020930560305714607, "global_step": 10009, "epoch": 278, "lr": 0.001} +{"train_loss": 0.043340135365724564, "train_loss_bc": 0.043340135365724564, "train_loss_llm": 0.0, "grad_norm": 0.03286105766892433, "global_step": 10010, "epoch": 278, "lr": 0.001} +{"train_loss": 0.03571554645895958, "train_loss_bc": 0.0356961190700531, "train_loss_llm": 0.001942664384841919, "grad_norm": 0.0371299684047699, "global_step": 10011, "epoch": 278, "lr": 0.001} +{"train_loss": 0.03709885850548744, "train_loss_bc": 0.03709885850548744, "train_loss_llm": 0.0, "grad_norm": 0.039490651339292526, "global_step": 10012, "epoch": 278, "lr": 0.001} +{"train_loss": 0.04224978759884834, "train_loss_bc": 0.04224978759884834, "train_loss_llm": 0.0, "grad_norm": 0.02122044935822487, "global_step": 10013, "epoch": 278, "lr": 0.001} +{"train_loss": 0.04543014243245125, "train_loss_bc": 0.04543014243245125, "train_loss_llm": 0.0, "grad_norm": 0.029121462255716324, "global_step": 10014, "epoch": 278, "lr": 0.001} +{"train_loss": 0.0399121418595314, "train_loss_bc": 0.03989582508802414, "train_loss_llm": 0.0016315877437591553, "grad_norm": 0.04008755087852478, "global_step": 10015, "epoch": 278, "lr": 0.001} +{"train_loss": 0.04188273102045059, "train_loss_bc": 0.04188273102045059, "train_loss_llm": 0.0, "grad_norm": 0.046387024223804474, "global_step": 10016, "epoch": 278, "lr": 0.001} +{"train_loss": 0.037528060376644135, "train_loss_bc": 0.037528060376644135, "train_loss_llm": 0.0, "grad_norm": 0.02069963328540325, "global_step": 10017, "epoch": 278, "lr": 0.001} +{"train_loss": 0.03609369322657585, "train_loss_bc": 0.03609369322657585, "train_loss_llm": 0.0, "grad_norm": 0.031020117923617363, "global_step": 10018, "epoch": 278, "lr": 0.001} +{"train_loss": 0.03941987454891205, "train_loss_bc": 0.03941987454891205, "train_loss_llm": 0.0, "grad_norm": 0.039043884724378586, "global_step": 10019, "epoch": 278, "lr": 0.001} +{"train_loss": 0.034125473350286484, "train_loss_bc": 0.034060511738061905, "train_loss_llm": 0.0064961910247802734, "grad_norm": 0.048529528081417084, "global_step": 10020, "epoch": 278, "lr": 0.001} +{"train_loss": 0.04511420801281929, "train_loss_bc": 0.04508942738175392, "train_loss_llm": 0.0024781227111816406, "grad_norm": 0.021639565005898476, "global_step": 10021, "epoch": 278, "lr": 0.001} +{"train_loss": 0.03347443789243698, "train_loss_bc": 0.03347443789243698, "train_loss_llm": 0.0, "grad_norm": 0.03495633974671364, "global_step": 10022, "epoch": 278, "lr": 0.001} +{"train_loss": 0.038337644189596176, "train_loss_bc": 0.038337644189596176, "train_loss_llm": 0.0, "grad_norm": 0.04590833932161331, "global_step": 10023, "epoch": 278, "lr": 0.001} +{"train_loss": 0.03731174394488335, "train_loss_bc": 0.03728730231523514, "train_loss_llm": 0.0024439990520477295, "grad_norm": 0.060264769941568375, "global_step": 10024, "epoch": 278, "lr": 0.001} +{"train_loss": 0.04082444682717323, "train_loss_bc": 0.04082444682717323, "train_loss_llm": 0.0, "grad_norm": 0.020514555275440216, "global_step": 10025, "epoch": 278, "lr": 0.001} +{"train_loss": 0.04088803008198738, "train_loss_bc": 0.04087408632040024, "train_loss_llm": 0.0013944506645202637, "grad_norm": 0.03433995693922043, "global_step": 10026, "epoch": 278, "lr": 0.001} +{"train_loss": 0.03773035854101181, "train_loss_bc": 0.03773035854101181, "train_loss_llm": 0.0, "grad_norm": 0.042282722890377045, "global_step": 10027, "epoch": 278, "lr": 0.001} +{"train_loss": 0.04100877791643143, "train_loss_bc": 0.04097531735897064, "train_loss_llm": 0.003346174955368042, "grad_norm": 0.05617803707718849, "global_step": 10028, "epoch": 278, "lr": 0.001} +{"train_loss": 0.038213811814785004, "train_loss_bc": 0.038213811814785004, "train_loss_llm": 0.0, "grad_norm": 0.017898311838507652, "global_step": 10029, "epoch": 278, "lr": 0.001} +{"train_loss": 0.0299962367862463, "train_loss_bc": 0.029982799664139748, "train_loss_llm": 0.0013437867164611816, "grad_norm": 0.02699527144432068, "global_step": 10030, "epoch": 278, "lr": 0.001} +{"train_loss": 0.031338054686784744, "train_loss_bc": 0.031338054686784744, "train_loss_llm": 0.0, "grad_norm": 0.03492244333028793, "global_step": 10031, "epoch": 278, "lr": 0.001} +{"train_loss": 0.036869727075099945, "train_loss_bc": 0.036869727075099945, "train_loss_llm": 0.0, "grad_norm": 0.0445585697889328, "global_step": 10032, "epoch": 278, "lr": 0.001} +{"train_loss": 0.042691487818956375, "train_loss_bc": 0.042691487818956375, "train_loss_llm": 0.0, "grad_norm": 0.024841926991939545, "global_step": 10033, "epoch": 278, "lr": 0.001} +{"train_loss": 0.035964880138635635, "train_loss_bc": 0.035964880138635635, "train_loss_llm": 0.0, "grad_norm": 0.03132955729961395, "global_step": 10034, "epoch": 278, "lr": 0.001} +{"train_loss": 0.04044628515839577, "train_loss_bc": 0.04044628515839577, "train_loss_llm": 0.0, "grad_norm": 0.040884703397750854, "global_step": 10035, "epoch": 278, "lr": 0.001} +{"train_loss": 0.036401934921741486, "train_loss_bc": 0.0363922156393528, "train_loss_llm": 0.0009717941284179688, "grad_norm": 0.05286913737654686, "global_step": 10036, "epoch": 278, "lr": 0.001} +{"train_loss": 0.0361747182905674, "train_loss_bc": 0.0361747182905674, "train_loss_llm": 0.0, "grad_norm": 0.020841049030423164, "global_step": 10037, "epoch": 278, "lr": 0.001} +{"train_loss": 0.03903159871697426, "train_loss_bc": 0.03901305049657822, "train_loss_llm": 0.0018548667430877686, "grad_norm": 0.027669858187437057, "global_step": 10038, "epoch": 278, "lr": 0.001} +{"train_loss": 0.03813519328832626, "train_loss_bc": 0.038120582699775696, "train_loss_llm": 0.0014608800411224365, "grad_norm": 0.03517945855855942, "global_step": 10039, "epoch": 278, "lr": 0.001} +{"train_loss": 0.04042957350611687, "train_loss_bc": 0.04042058810591698, "train_loss_llm": 0.000898510217666626, "grad_norm": 0.043293461203575134, "global_step": 10040, "epoch": 278, "lr": 0.001} +{"train_loss": 0.035086117684841156, "train_loss_bc": 0.035086117684841156, "train_loss_llm": 0.0, "grad_norm": 0.01917232759296894, "global_step": 10041, "epoch": 278, "lr": 0.001} +{"train_loss": 0.037459596991539, "train_loss_bc": 0.037459596991539, "train_loss_llm": 0.0, "grad_norm": 0.026639921590685844, "global_step": 10042, "epoch": 278, "lr": 0.001} +{"train_loss": 0.03830952047266894, "train_loss_bc": 0.036675166338682175, "train_loss_llm": 0.0, "grad_norm": 0.04283441975712776, "global_step": 10043, "epoch": 278, "lr": 0.001, "val_loss": 0.06170473247766495} +{"train_loss": 0.037526555359363556, "train_loss_bc": 0.037526555359363556, "train_loss_llm": 0.0, "grad_norm": 0.053028129041194916, "global_step": 10044, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03931550309062004, "train_loss_bc": 0.03931550309062004, "train_loss_llm": 0.0, "grad_norm": 0.024627210572361946, "global_step": 10045, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03452692925930023, "train_loss_bc": 0.03452692925930023, "train_loss_llm": 0.0, "grad_norm": 0.038308512419462204, "global_step": 10046, "epoch": 279, "lr": 0.001} +{"train_loss": 0.032901108264923096, "train_loss_bc": 0.032901108264923096, "train_loss_llm": 0.0, "grad_norm": 0.04428188502788544, "global_step": 10047, "epoch": 279, "lr": 0.001} +{"train_loss": 0.04799662530422211, "train_loss_bc": 0.04799662530422211, "train_loss_llm": 0.0, "grad_norm": 0.05864846706390381, "global_step": 10048, "epoch": 279, "lr": 0.001} +{"train_loss": 0.0347929410636425, "train_loss_bc": 0.0347929410636425, "train_loss_llm": 0.0, "grad_norm": 0.020857078954577446, "global_step": 10049, "epoch": 279, "lr": 0.001} +{"train_loss": 0.0379326269030571, "train_loss_bc": 0.0379326269030571, "train_loss_llm": 0.0, "grad_norm": 0.03314512223005295, "global_step": 10050, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03508194535970688, "train_loss_bc": 0.03508194535970688, "train_loss_llm": 0.0, "grad_norm": 0.044575951993465424, "global_step": 10051, "epoch": 279, "lr": 0.001} +{"train_loss": 0.04349078610539436, "train_loss_bc": 0.04349078610539436, "train_loss_llm": 0.0, "grad_norm": 0.049813419580459595, "global_step": 10052, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03588198870420456, "train_loss_bc": 0.035872381180524826, "train_loss_llm": 0.0009607970714569092, "grad_norm": 0.018478304147720337, "global_step": 10053, "epoch": 279, "lr": 0.001} +{"train_loss": 0.04083041474223137, "train_loss_bc": 0.04083041474223137, "train_loss_llm": 0.0, "grad_norm": 0.03223196417093277, "global_step": 10054, "epoch": 279, "lr": 0.001} +{"train_loss": 0.041108254343271255, "train_loss_bc": 0.041108254343271255, "train_loss_llm": 0.0, "grad_norm": 0.0512135811150074, "global_step": 10055, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03516484797000885, "train_loss_bc": 0.03512831777334213, "train_loss_llm": 0.0036530494689941406, "grad_norm": 0.06400763243436813, "global_step": 10056, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03557153791189194, "train_loss_bc": 0.03557153791189194, "train_loss_llm": 0.0, "grad_norm": 0.01596270315349102, "global_step": 10057, "epoch": 279, "lr": 0.001} +{"train_loss": 0.04656793922185898, "train_loss_bc": 0.04656793922185898, "train_loss_llm": 0.0, "grad_norm": 0.029274938628077507, "global_step": 10058, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03873462229967117, "train_loss_bc": 0.03869587928056717, "train_loss_llm": 0.003874450922012329, "grad_norm": 0.03690582141280174, "global_step": 10059, "epoch": 279, "lr": 0.001} +{"train_loss": 0.036375876516103745, "train_loss_bc": 0.03636163845658302, "train_loss_llm": 0.0014237463474273682, "grad_norm": 0.04290083423256874, "global_step": 10060, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03606782853603363, "train_loss_bc": 0.03605179116129875, "train_loss_llm": 0.0016035735607147217, "grad_norm": 0.021036844700574875, "global_step": 10061, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03778545930981636, "train_loss_bc": 0.03778545930981636, "train_loss_llm": 0.0, "grad_norm": 0.030613822862505913, "global_step": 10062, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03138386830687523, "train_loss_bc": 0.03138386830687523, "train_loss_llm": 0.0, "grad_norm": 0.03898550197482109, "global_step": 10063, "epoch": 279, "lr": 0.001} +{"train_loss": 0.027900278568267822, "train_loss_bc": 0.027900278568267822, "train_loss_llm": 0.0, "grad_norm": 0.04645371437072754, "global_step": 10064, "epoch": 279, "lr": 0.001} +{"train_loss": 0.04154398292303085, "train_loss_bc": 0.04154398292303085, "train_loss_llm": 0.0, "grad_norm": 0.018015509471297264, "global_step": 10065, "epoch": 279, "lr": 0.001} +{"train_loss": 0.039102569222450256, "train_loss_bc": 0.039102569222450256, "train_loss_llm": 0.0, "grad_norm": 0.02459653839468956, "global_step": 10066, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03930726647377014, "train_loss_bc": 0.03930726647377014, "train_loss_llm": 0.0, "grad_norm": 0.036522261798381805, "global_step": 10067, "epoch": 279, "lr": 0.001} +{"train_loss": 0.04298480227589607, "train_loss_bc": 0.04298480227589607, "train_loss_llm": 0.0, "grad_norm": 0.04634904861450195, "global_step": 10068, "epoch": 279, "lr": 0.001} +{"train_loss": 0.042838335037231445, "train_loss_bc": 0.042831797152757645, "train_loss_llm": 0.0006536245346069336, "grad_norm": 0.023568591102957726, "global_step": 10069, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03775179758667946, "train_loss_bc": 0.03775179758667946, "train_loss_llm": 0.0, "grad_norm": 0.03744287043809891, "global_step": 10070, "epoch": 279, "lr": 0.001} +{"train_loss": 0.04754548892378807, "train_loss_bc": 0.04754548892378807, "train_loss_llm": 0.0, "grad_norm": 0.055126041173934937, "global_step": 10071, "epoch": 279, "lr": 0.001} +{"train_loss": 0.04064159467816353, "train_loss_bc": 0.04063734412193298, "train_loss_llm": 0.00042510032653808594, "grad_norm": 0.06987406313419342, "global_step": 10072, "epoch": 279, "lr": 0.001} +{"train_loss": 0.04211704060435295, "train_loss_bc": 0.042104773223400116, "train_loss_llm": 0.0012269020080566406, "grad_norm": 0.023866431787610054, "global_step": 10073, "epoch": 279, "lr": 0.001} +{"train_loss": 0.04392818734049797, "train_loss_bc": 0.04392818734049797, "train_loss_llm": 0.0, "grad_norm": 0.03786192461848259, "global_step": 10074, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03946366533637047, "train_loss_bc": 0.03946366533637047, "train_loss_llm": 0.0, "grad_norm": 0.043642040342092514, "global_step": 10075, "epoch": 279, "lr": 0.001} +{"train_loss": 0.036564458161592484, "train_loss_bc": 0.036564458161592484, "train_loss_llm": 0.0, "grad_norm": 0.048802535980939865, "global_step": 10076, "epoch": 279, "lr": 0.001} +{"train_loss": 0.0335155613720417, "train_loss_bc": 0.0335051566362381, "train_loss_llm": 0.0010405778884887695, "grad_norm": 0.023876484483480453, "global_step": 10077, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03607989102602005, "train_loss_bc": 0.03607989102602005, "train_loss_llm": 0.0, "grad_norm": 0.039010170847177505, "global_step": 10078, "epoch": 279, "lr": 0.001} +{"train_loss": 0.03826607918987671, "train_loss_bc": 0.027256272733211517, "train_loss_llm": 0.0, "grad_norm": 0.053943660110235214, "global_step": 10079, "epoch": 279, "lr": 0.001, "val_loss": 0.060786016285419464} +{"train_loss": 0.03741289675235748, "train_loss_bc": 0.037407729774713516, "train_loss_llm": 0.0005168616771697998, "grad_norm": 0.06636690348386765, "global_step": 10080, "epoch": 280, "lr": 0.001} +{"train_loss": 0.04337181895971298, "train_loss_bc": 0.04337181895971298, "train_loss_llm": 0.0, "grad_norm": 0.020319515839219093, "global_step": 10081, "epoch": 280, "lr": 0.001} +{"train_loss": 0.039640456438064575, "train_loss_bc": 0.039640456438064575, "train_loss_llm": 0.0, "grad_norm": 0.02961345575749874, "global_step": 10082, "epoch": 280, "lr": 0.001} +{"train_loss": 0.036683980375528336, "train_loss_bc": 0.03667493909597397, "train_loss_llm": 0.0009042024612426758, "grad_norm": 0.04016094282269478, "global_step": 10083, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03556584194302559, "train_loss_bc": 0.03556584194302559, "train_loss_llm": 0.0, "grad_norm": 0.04772597923874855, "global_step": 10084, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03698249161243439, "train_loss_bc": 0.03698249161243439, "train_loss_llm": 0.0, "grad_norm": 0.01818995550274849, "global_step": 10085, "epoch": 280, "lr": 0.001} +{"train_loss": 0.04483264684677124, "train_loss_bc": 0.04483264684677124, "train_loss_llm": 0.0, "grad_norm": 0.03199329972267151, "global_step": 10086, "epoch": 280, "lr": 0.001} +{"train_loss": 0.0384574830532074, "train_loss_bc": 0.0384574830532074, "train_loss_llm": 0.0, "grad_norm": 0.04145505279302597, "global_step": 10087, "epoch": 280, "lr": 0.001} +{"train_loss": 0.035228028893470764, "train_loss_bc": 0.035228028893470764, "train_loss_llm": 0.0, "grad_norm": 0.04994574561715126, "global_step": 10088, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03496303781867027, "train_loss_bc": 0.03496303781867027, "train_loss_llm": 0.0, "grad_norm": 0.02608533762395382, "global_step": 10089, "epoch": 280, "lr": 0.001} +{"train_loss": 0.042439866811037064, "train_loss_bc": 0.042439866811037064, "train_loss_llm": 0.0, "grad_norm": 0.04020121321082115, "global_step": 10090, "epoch": 280, "lr": 0.001} +{"train_loss": 0.04273582994937897, "train_loss_bc": 0.04273286089301109, "train_loss_llm": 0.0002968907356262207, "grad_norm": 0.04719673842191696, "global_step": 10091, "epoch": 280, "lr": 0.001} +{"train_loss": 0.040146391838788986, "train_loss_bc": 0.040146391838788986, "train_loss_llm": 0.0, "grad_norm": 0.06061483919620514, "global_step": 10092, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03561360388994217, "train_loss_bc": 0.035594671964645386, "train_loss_llm": 0.0018932819366455078, "grad_norm": 0.024223150685429573, "global_step": 10093, "epoch": 280, "lr": 0.001} +{"train_loss": 0.04298708587884903, "train_loss_bc": 0.04298708587884903, "train_loss_llm": 0.0, "grad_norm": 0.03818748891353607, "global_step": 10094, "epoch": 280, "lr": 0.001} +{"train_loss": 0.038010574877262115, "train_loss_bc": 0.038010574877262115, "train_loss_llm": 0.0, "grad_norm": 0.05025949701666832, "global_step": 10095, "epoch": 280, "lr": 0.001} +{"train_loss": 0.032062314450740814, "train_loss_bc": 0.032062314450740814, "train_loss_llm": 0.0, "grad_norm": 0.06175387278199196, "global_step": 10096, "epoch": 280, "lr": 0.001} +{"train_loss": 0.039107006043195724, "train_loss_bc": 0.039107006043195724, "train_loss_llm": 0.0, "grad_norm": 0.019838403910398483, "global_step": 10097, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03273382410407066, "train_loss_bc": 0.03273382410407066, "train_loss_llm": 0.0, "grad_norm": 0.03928668424487114, "global_step": 10098, "epoch": 280, "lr": 0.001} +{"train_loss": 0.04221445694565773, "train_loss_bc": 0.04218974709510803, "train_loss_llm": 0.002471059560775757, "grad_norm": 0.04859159141778946, "global_step": 10099, "epoch": 280, "lr": 0.001} +{"train_loss": 0.04610162600874901, "train_loss_bc": 0.04610162600874901, "train_loss_llm": 0.0, "grad_norm": 0.07165119051933289, "global_step": 10100, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03478499501943588, "train_loss_bc": 0.03478499501943588, "train_loss_llm": 0.0, "grad_norm": 0.022422676905989647, "global_step": 10101, "epoch": 280, "lr": 0.001} +{"train_loss": 0.036580637097358704, "train_loss_bc": 0.036580637097358704, "train_loss_llm": 0.0, "grad_norm": 0.031562965363264084, "global_step": 10102, "epoch": 280, "lr": 0.001} +{"train_loss": 0.044806577265262604, "train_loss_bc": 0.044806577265262604, "train_loss_llm": 0.0, "grad_norm": 0.04680295288562775, "global_step": 10103, "epoch": 280, "lr": 0.001} +{"train_loss": 0.044877830892801285, "train_loss_bc": 0.044873613864183426, "train_loss_llm": 0.0004215836524963379, "grad_norm": 0.04961895942687988, "global_step": 10104, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03485431894659996, "train_loss_bc": 0.03484770655632019, "train_loss_llm": 0.0006612241268157959, "grad_norm": 0.018238091841340065, "global_step": 10105, "epoch": 280, "lr": 0.001} +{"train_loss": 0.036790881305933, "train_loss_bc": 0.03678470104932785, "train_loss_llm": 0.0006179511547088623, "grad_norm": 0.029778821393847466, "global_step": 10106, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03832786902785301, "train_loss_bc": 0.03832786902785301, "train_loss_llm": 0.0, "grad_norm": 0.03986864164471626, "global_step": 10107, "epoch": 280, "lr": 0.001} +{"train_loss": 0.041152939200401306, "train_loss_bc": 0.041152939200401306, "train_loss_llm": 0.0, "grad_norm": 0.05303424224257469, "global_step": 10108, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03661065548658371, "train_loss_bc": 0.03661065548658371, "train_loss_llm": 0.0, "grad_norm": 0.0205631572753191, "global_step": 10109, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03722359240055084, "train_loss_bc": 0.03722359240055084, "train_loss_llm": 0.0, "grad_norm": 0.03173281252384186, "global_step": 10110, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03689790889620781, "train_loss_bc": 0.03689790889620781, "train_loss_llm": 0.0, "grad_norm": 0.04057082161307335, "global_step": 10111, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03630947321653366, "train_loss_bc": 0.036302901804447174, "train_loss_llm": 0.0006569623947143555, "grad_norm": 0.05148019269108772, "global_step": 10112, "epoch": 280, "lr": 0.001} +{"train_loss": 0.033224284648895264, "train_loss_bc": 0.033224284648895264, "train_loss_llm": 0.0, "grad_norm": 0.021263137459754944, "global_step": 10113, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03679686784744263, "train_loss_bc": 0.03679129481315613, "train_loss_llm": 0.00055733323097229, "grad_norm": 0.03828844428062439, "global_step": 10114, "epoch": 280, "lr": 0.001} +{"train_loss": 0.03837515920814541, "train_loss_bc": 0.03497563675045967, "train_loss_llm": 0.0, "grad_norm": 0.051607098430395126, "global_step": 10115, "epoch": 280, "lr": 0.001, "train/cumulative_reward": 49.59161978819959, "train/mean_score": 11.642712851086788, "train/success_rate": 0.3, "test/cumulative_reward": 40.01517095968125, "test/mean_score": 4.385352683576585, "test/success_rate": 0.08, "val_loss": 0.06084895133972168, "train_action_mse_error": 0.04203413426876068} +{"train_loss": 0.04034072533249855, "train_loss_bc": 0.04033922404050827, "train_loss_llm": 0.00015020370483398438, "grad_norm": 0.06119073927402496, "global_step": 10116, "epoch": 281, "lr": 0.001} +{"train_loss": 0.041814591735601425, "train_loss_bc": 0.041814591735601425, "train_loss_llm": 0.0, "grad_norm": 0.023604046553373337, "global_step": 10117, "epoch": 281, "lr": 0.001} +{"train_loss": 0.0359046496450901, "train_loss_bc": 0.0359046496450901, "train_loss_llm": 0.0, "grad_norm": 0.030488228425383568, "global_step": 10118, "epoch": 281, "lr": 0.001} +{"train_loss": 0.042185164988040924, "train_loss_bc": 0.04217147082090378, "train_loss_llm": 0.0013695955276489258, "grad_norm": 0.037172652781009674, "global_step": 10119, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03749943524599075, "train_loss_bc": 0.03749943524599075, "train_loss_llm": 0.0, "grad_norm": 0.04762400686740875, "global_step": 10120, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03263552114367485, "train_loss_bc": 0.0326203852891922, "train_loss_llm": 0.001513451337814331, "grad_norm": 0.021032104268670082, "global_step": 10121, "epoch": 281, "lr": 0.001} +{"train_loss": 0.0393436998128891, "train_loss_bc": 0.03932753950357437, "train_loss_llm": 0.0016158819198608398, "grad_norm": 0.03245626762509346, "global_step": 10122, "epoch": 281, "lr": 0.001} +{"train_loss": 0.046606723219156265, "train_loss_bc": 0.04658101871609688, "train_loss_llm": 0.002570509910583496, "grad_norm": 0.05374622344970703, "global_step": 10123, "epoch": 281, "lr": 0.001} +{"train_loss": 0.039151452481746674, "train_loss_bc": 0.039151452481746674, "train_loss_llm": 0.0, "grad_norm": 0.07200033962726593, "global_step": 10124, "epoch": 281, "lr": 0.001} +{"train_loss": 0.040071696043014526, "train_loss_bc": 0.040071696043014526, "train_loss_llm": 0.0, "grad_norm": 0.02118707075715065, "global_step": 10125, "epoch": 281, "lr": 0.001} +{"train_loss": 0.036007221788167953, "train_loss_bc": 0.036007221788167953, "train_loss_llm": 0.0, "grad_norm": 0.0383528396487236, "global_step": 10126, "epoch": 281, "lr": 0.001} +{"train_loss": 0.04253803938627243, "train_loss_bc": 0.04253803938627243, "train_loss_llm": 0.0, "grad_norm": 0.04905802011489868, "global_step": 10127, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03806992247700691, "train_loss_bc": 0.03806332126259804, "train_loss_llm": 0.0006603002548217773, "grad_norm": 0.06123321130871773, "global_step": 10128, "epoch": 281, "lr": 0.001} +{"train_loss": 0.04150274023413658, "train_loss_bc": 0.04150274023413658, "train_loss_llm": 0.0, "grad_norm": 0.0197139959782362, "global_step": 10129, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03200788050889969, "train_loss_bc": 0.03200788050889969, "train_loss_llm": 0.0, "grad_norm": 0.027066221460700035, "global_step": 10130, "epoch": 281, "lr": 0.001} +{"train_loss": 0.04395320639014244, "train_loss_bc": 0.04395320639014244, "train_loss_llm": 0.0, "grad_norm": 0.03975547477602959, "global_step": 10131, "epoch": 281, "lr": 0.001} +{"train_loss": 0.046087443828582764, "train_loss_bc": 0.046085234731435776, "train_loss_llm": 0.00022074580192565918, "grad_norm": 0.048387326300144196, "global_step": 10132, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03815576061606407, "train_loss_bc": 0.03815576061606407, "train_loss_llm": 0.0, "grad_norm": 0.0201471708714962, "global_step": 10133, "epoch": 281, "lr": 0.001} +{"train_loss": 0.0411166287958622, "train_loss_bc": 0.0411166287958622, "train_loss_llm": 0.0, "grad_norm": 0.03261418640613556, "global_step": 10134, "epoch": 281, "lr": 0.001} +{"train_loss": 0.035908397287130356, "train_loss_bc": 0.035908397287130356, "train_loss_llm": 0.0, "grad_norm": 0.041743747889995575, "global_step": 10135, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03708997741341591, "train_loss_bc": 0.03708997741341591, "train_loss_llm": 0.0, "grad_norm": 0.05128318443894386, "global_step": 10136, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03453211858868599, "train_loss_bc": 0.03453211858868599, "train_loss_llm": 0.0, "grad_norm": 0.02079361490905285, "global_step": 10137, "epoch": 281, "lr": 0.001} +{"train_loss": 0.04048949480056763, "train_loss_bc": 0.04048949480056763, "train_loss_llm": 0.0, "grad_norm": 0.04101435840129852, "global_step": 10138, "epoch": 281, "lr": 0.001} +{"train_loss": 0.035614609718322754, "train_loss_bc": 0.035614609718322754, "train_loss_llm": 0.0, "grad_norm": 0.05207367613911629, "global_step": 10139, "epoch": 281, "lr": 0.001} +{"train_loss": 0.042690303176641464, "train_loss_bc": 0.042690303176641464, "train_loss_llm": 0.0, "grad_norm": 0.06692416965961456, "global_step": 10140, "epoch": 281, "lr": 0.001} +{"train_loss": 0.0345960333943367, "train_loss_bc": 0.034587036818265915, "train_loss_llm": 0.0008997023105621338, "grad_norm": 0.019550517201423645, "global_step": 10141, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03764330595731735, "train_loss_bc": 0.03764330595731735, "train_loss_llm": 0.0, "grad_norm": 0.03521331772208214, "global_step": 10142, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03257071599364281, "train_loss_bc": 0.032557424157857895, "train_loss_llm": 0.0013293027877807617, "grad_norm": 0.044625937938690186, "global_step": 10143, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03809366747736931, "train_loss_bc": 0.03809366747736931, "train_loss_llm": 0.0, "grad_norm": 0.062794528901577, "global_step": 10144, "epoch": 281, "lr": 0.001} +{"train_loss": 0.0349910631775856, "train_loss_bc": 0.03498561307787895, "train_loss_llm": 0.0005451738834381104, "grad_norm": 0.017982419580221176, "global_step": 10145, "epoch": 281, "lr": 0.001} +{"train_loss": 0.04082033038139343, "train_loss_bc": 0.04082033038139343, "train_loss_llm": 0.0, "grad_norm": 0.030001822859048843, "global_step": 10146, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03886103630065918, "train_loss_bc": 0.0388476587831974, "train_loss_llm": 0.0013376474380493164, "grad_norm": 0.040147993713617325, "global_step": 10147, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03936009481549263, "train_loss_bc": 0.03936009481549263, "train_loss_llm": 0.0, "grad_norm": 0.04908570647239685, "global_step": 10148, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03505542501807213, "train_loss_bc": 0.03505542501807213, "train_loss_llm": 0.0, "grad_norm": 0.019371744245290756, "global_step": 10149, "epoch": 281, "lr": 0.001} +{"train_loss": 0.0299686286598444, "train_loss_bc": 0.0299686286598444, "train_loss_llm": 0.0, "grad_norm": 0.033228252083063126, "global_step": 10150, "epoch": 281, "lr": 0.001} +{"train_loss": 0.03835526527836919, "train_loss_bc": 0.03750300034880638, "train_loss_llm": 0.000884324312210083, "grad_norm": 0.04332057759165764, "global_step": 10151, "epoch": 281, "lr": 0.001, "val_loss": 0.06113985553383827} +{"train_loss": 0.03530476242303848, "train_loss_bc": 0.03530476242303848, "train_loss_llm": 0.0, "grad_norm": 0.04906848073005676, "global_step": 10152, "epoch": 282, "lr": 0.001} +{"train_loss": 0.0407782606780529, "train_loss_bc": 0.0407782606780529, "train_loss_llm": 0.0, "grad_norm": 0.02068346180021763, "global_step": 10153, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03359063342213631, "train_loss_bc": 0.0335661880671978, "train_loss_llm": 0.0024443864822387695, "grad_norm": 0.026911910623311996, "global_step": 10154, "epoch": 282, "lr": 0.001} +{"train_loss": 0.04189888760447502, "train_loss_bc": 0.04189888760447502, "train_loss_llm": 0.0, "grad_norm": 0.036335453391075134, "global_step": 10155, "epoch": 282, "lr": 0.001} +{"train_loss": 0.044160276651382446, "train_loss_bc": 0.044160276651382446, "train_loss_llm": 0.0, "grad_norm": 0.04655458778142929, "global_step": 10156, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03219304606318474, "train_loss_bc": 0.03219304606318474, "train_loss_llm": 0.0, "grad_norm": 0.020950639620423317, "global_step": 10157, "epoch": 282, "lr": 0.001} +{"train_loss": 0.040019743144512177, "train_loss_bc": 0.04001060128211975, "train_loss_llm": 0.0009143352508544922, "grad_norm": 0.033594708889722824, "global_step": 10158, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03582135960459709, "train_loss_bc": 0.03582135960459709, "train_loss_llm": 0.0, "grad_norm": 0.03776673600077629, "global_step": 10159, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03472664952278137, "train_loss_bc": 0.03472664952278137, "train_loss_llm": 0.0, "grad_norm": 0.04490033537149429, "global_step": 10160, "epoch": 282, "lr": 0.001} +{"train_loss": 0.032028928399086, "train_loss_bc": 0.03202052414417267, "train_loss_llm": 0.0008402764797210693, "grad_norm": 0.022909728810191154, "global_step": 10161, "epoch": 282, "lr": 0.001} +{"train_loss": 0.033000051975250244, "train_loss_bc": 0.033000051975250244, "train_loss_llm": 0.0, "grad_norm": 0.03652885556221008, "global_step": 10162, "epoch": 282, "lr": 0.001} +{"train_loss": 0.04068729281425476, "train_loss_bc": 0.04068729281425476, "train_loss_llm": 0.0, "grad_norm": 0.05034118518233299, "global_step": 10163, "epoch": 282, "lr": 0.001} +{"train_loss": 0.047554075717926025, "train_loss_bc": 0.047554075717926025, "train_loss_llm": 0.0, "grad_norm": 0.05958300456404686, "global_step": 10164, "epoch": 282, "lr": 0.001} +{"train_loss": 0.0330275297164917, "train_loss_bc": 0.0330275297164917, "train_loss_llm": 0.0, "grad_norm": 0.017484666779637337, "global_step": 10165, "epoch": 282, "lr": 0.001} +{"train_loss": 0.045161694288253784, "train_loss_bc": 0.045161694288253784, "train_loss_llm": 0.0, "grad_norm": 0.032822925597429276, "global_step": 10166, "epoch": 282, "lr": 0.001} +{"train_loss": 0.0379832461476326, "train_loss_bc": 0.0379832461476326, "train_loss_llm": 0.0, "grad_norm": 0.04178628697991371, "global_step": 10167, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03781034052371979, "train_loss_bc": 0.03781034052371979, "train_loss_llm": 0.0, "grad_norm": 0.05582260340452194, "global_step": 10168, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03838032856583595, "train_loss_bc": 0.03838032856583595, "train_loss_llm": 0.0, "grad_norm": 0.019896572455763817, "global_step": 10169, "epoch": 282, "lr": 0.001} +{"train_loss": 0.0397307351231575, "train_loss_bc": 0.0397307351231575, "train_loss_llm": 0.0, "grad_norm": 0.029576893895864487, "global_step": 10170, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03398177772760391, "train_loss_bc": 0.03398177772760391, "train_loss_llm": 0.0, "grad_norm": 0.03924219682812691, "global_step": 10171, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03517058119177818, "train_loss_bc": 0.03517058119177818, "train_loss_llm": 0.0, "grad_norm": 0.050299033522605896, "global_step": 10172, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03841965273022652, "train_loss_bc": 0.03841965273022652, "train_loss_llm": 0.0, "grad_norm": 0.016747931018471718, "global_step": 10173, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03337074816226959, "train_loss_bc": 0.03337074816226959, "train_loss_llm": 0.0, "grad_norm": 0.02518448606133461, "global_step": 10174, "epoch": 282, "lr": 0.001} +{"train_loss": 0.040309805423021317, "train_loss_bc": 0.040309805423021317, "train_loss_llm": 0.0, "grad_norm": 0.030344834551215172, "global_step": 10175, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03649651259183884, "train_loss_bc": 0.03649238124489784, "train_loss_llm": 0.0004130899906158447, "grad_norm": 0.03682113066315651, "global_step": 10176, "epoch": 282, "lr": 0.001} +{"train_loss": 0.037897221744060516, "train_loss_bc": 0.037897221744060516, "train_loss_llm": 0.0, "grad_norm": 0.016736602410674095, "global_step": 10177, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03782873973250389, "train_loss_bc": 0.03782873973250389, "train_loss_llm": 0.0, "grad_norm": 0.030942823737859726, "global_step": 10178, "epoch": 282, "lr": 0.001} +{"train_loss": 0.041146114468574524, "train_loss_bc": 0.041146114468574524, "train_loss_llm": 0.0, "grad_norm": 0.040053997188806534, "global_step": 10179, "epoch": 282, "lr": 0.001} +{"train_loss": 0.043051883578300476, "train_loss_bc": 0.043051883578300476, "train_loss_llm": 0.0, "grad_norm": 0.055374689400196075, "global_step": 10180, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03732505068182945, "train_loss_bc": 0.0373058095574379, "train_loss_llm": 0.0019242167472839355, "grad_norm": 0.01778688281774521, "global_step": 10181, "epoch": 282, "lr": 0.001} +{"train_loss": 0.029035963118076324, "train_loss_bc": 0.029023436829447746, "train_loss_llm": 0.0012525618076324463, "grad_norm": 0.02739083021879196, "global_step": 10182, "epoch": 282, "lr": 0.001} +{"train_loss": 0.04103832319378853, "train_loss_bc": 0.04103832319378853, "train_loss_llm": 0.0, "grad_norm": 0.03521762043237686, "global_step": 10183, "epoch": 282, "lr": 0.001} +{"train_loss": 0.04002906754612923, "train_loss_bc": 0.0400225892663002, "train_loss_llm": 0.0006479322910308838, "grad_norm": 0.04456581175327301, "global_step": 10184, "epoch": 282, "lr": 0.001} +{"train_loss": 0.04321565851569176, "train_loss_bc": 0.04321565851569176, "train_loss_llm": 0.0, "grad_norm": 0.026617154479026794, "global_step": 10185, "epoch": 282, "lr": 0.001} +{"train_loss": 0.04651417210698128, "train_loss_bc": 0.04651417210698128, "train_loss_llm": 0.0, "grad_norm": 0.037998948246240616, "global_step": 10186, "epoch": 282, "lr": 0.001} +{"train_loss": 0.03830305135084523, "train_loss_bc": 0.040193621069192886, "train_loss_llm": 0.002711355686187744, "grad_norm": 0.05583111196756363, "global_step": 10187, "epoch": 282, "lr": 0.001, "val_loss": 0.06127336248755455} +{"train_loss": 0.03935973718762398, "train_loss_bc": 0.03935973718762398, "train_loss_llm": 0.0, "grad_norm": 0.06394290924072266, "global_step": 10188, "epoch": 283, "lr": 0.001} +{"train_loss": 0.0367530919611454, "train_loss_bc": 0.0367530919611454, "train_loss_llm": 0.0, "grad_norm": 0.01811358705163002, "global_step": 10189, "epoch": 283, "lr": 0.001} +{"train_loss": 0.033926598727703094, "train_loss_bc": 0.033926598727703094, "train_loss_llm": 0.0, "grad_norm": 0.028913702815771103, "global_step": 10190, "epoch": 283, "lr": 0.001} +{"train_loss": 0.037513669580221176, "train_loss_bc": 0.037513669580221176, "train_loss_llm": 0.0, "grad_norm": 0.03852860629558563, "global_step": 10191, "epoch": 283, "lr": 0.001} +{"train_loss": 0.038905881345272064, "train_loss_bc": 0.038905881345272064, "train_loss_llm": 0.0, "grad_norm": 0.04695688560605049, "global_step": 10192, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03716227039694786, "train_loss_bc": 0.037158772349357605, "train_loss_llm": 0.00034993886947631836, "grad_norm": 0.020916534587740898, "global_step": 10193, "epoch": 283, "lr": 0.001} +{"train_loss": 0.033393844962120056, "train_loss_bc": 0.033393844962120056, "train_loss_llm": 0.0, "grad_norm": 0.028958573937416077, "global_step": 10194, "epoch": 283, "lr": 0.001} +{"train_loss": 0.038856979459524155, "train_loss_bc": 0.038856979459524155, "train_loss_llm": 0.0, "grad_norm": 0.04005841538310051, "global_step": 10195, "epoch": 283, "lr": 0.001} +{"train_loss": 0.04153161123394966, "train_loss_bc": 0.04153161123394966, "train_loss_llm": 0.0, "grad_norm": 0.05115535110235214, "global_step": 10196, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03754538670182228, "train_loss_bc": 0.03754538670182228, "train_loss_llm": 0.0, "grad_norm": 0.01834660768508911, "global_step": 10197, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03812996298074722, "train_loss_bc": 0.03812996298074722, "train_loss_llm": 0.0, "grad_norm": 0.030076254159212112, "global_step": 10198, "epoch": 283, "lr": 0.001} +{"train_loss": 0.0416380912065506, "train_loss_bc": 0.04162978380918503, "train_loss_llm": 0.000830918550491333, "grad_norm": 0.035201117396354675, "global_step": 10199, "epoch": 283, "lr": 0.001} +{"train_loss": 0.04451354593038559, "train_loss_bc": 0.04451354593038559, "train_loss_llm": 0.0, "grad_norm": 0.04178495705127716, "global_step": 10200, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03309987857937813, "train_loss_bc": 0.03309987857937813, "train_loss_llm": 0.0, "grad_norm": 0.01827712543308735, "global_step": 10201, "epoch": 283, "lr": 0.001} +{"train_loss": 0.027376459911465645, "train_loss_bc": 0.027376459911465645, "train_loss_llm": 0.0, "grad_norm": 0.02787186950445175, "global_step": 10202, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03868544101715088, "train_loss_bc": 0.03868088871240616, "train_loss_llm": 0.0004550516605377197, "grad_norm": 0.03578396886587143, "global_step": 10203, "epoch": 283, "lr": 0.001} +{"train_loss": 0.0382261797785759, "train_loss_bc": 0.0382261797785759, "train_loss_llm": 0.0, "grad_norm": 0.03979197517037392, "global_step": 10204, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03458539396524429, "train_loss_bc": 0.03457247465848923, "train_loss_llm": 0.001291811466217041, "grad_norm": 0.022451922297477722, "global_step": 10205, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03759720176458359, "train_loss_bc": 0.03759720176458359, "train_loss_llm": 0.0, "grad_norm": 0.04222943261265755, "global_step": 10206, "epoch": 283, "lr": 0.001} +{"train_loss": 0.045367855578660965, "train_loss_bc": 0.045367855578660965, "train_loss_llm": 0.0, "grad_norm": 0.048355475068092346, "global_step": 10207, "epoch": 283, "lr": 0.001} +{"train_loss": 0.038487233221530914, "train_loss_bc": 0.038471631705760956, "train_loss_llm": 0.0015600025653839111, "grad_norm": 0.060257524251937866, "global_step": 10208, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03612503036856651, "train_loss_bc": 0.036123957484960556, "train_loss_llm": 0.00010719895362854004, "grad_norm": 0.01480127964168787, "global_step": 10209, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03999224677681923, "train_loss_bc": 0.03999224677681923, "train_loss_llm": 0.0, "grad_norm": 0.027256010100245476, "global_step": 10210, "epoch": 283, "lr": 0.001} +{"train_loss": 0.04532605782151222, "train_loss_bc": 0.045299094170331955, "train_loss_llm": 0.002696424722671509, "grad_norm": 0.03869837149977684, "global_step": 10211, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03956575691699982, "train_loss_bc": 0.0395553782582283, "train_loss_llm": 0.0010378360748291016, "grad_norm": 0.05446932464838028, "global_step": 10212, "epoch": 283, "lr": 0.001} +{"train_loss": 0.043536216020584106, "train_loss_bc": 0.04352225363254547, "train_loss_llm": 0.0013961195945739746, "grad_norm": 0.020996231585741043, "global_step": 10213, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03843391686677933, "train_loss_bc": 0.03841744363307953, "train_loss_llm": 0.0016473233699798584, "grad_norm": 0.02984904870390892, "global_step": 10214, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03665243834257126, "train_loss_bc": 0.03665243834257126, "train_loss_llm": 0.0, "grad_norm": 0.03959166631102562, "global_step": 10215, "epoch": 283, "lr": 0.001} +{"train_loss": 0.0394434817135334, "train_loss_bc": 0.039422135800123215, "train_loss_llm": 0.002134561538696289, "grad_norm": 0.04821646586060524, "global_step": 10216, "epoch": 283, "lr": 0.001} +{"train_loss": 0.040598608553409576, "train_loss_bc": 0.040598608553409576, "train_loss_llm": 0.0, "grad_norm": 0.019220946356654167, "global_step": 10217, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03540494665503502, "train_loss_bc": 0.03540494665503502, "train_loss_llm": 0.0, "grad_norm": 0.031198393553495407, "global_step": 10218, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03833469748497009, "train_loss_bc": 0.03833469748497009, "train_loss_llm": 0.0, "grad_norm": 0.03915034607052803, "global_step": 10219, "epoch": 283, "lr": 0.001} +{"train_loss": 0.04134289547801018, "train_loss_bc": 0.04133862257003784, "train_loss_llm": 0.00042745471000671387, "grad_norm": 0.05052630603313446, "global_step": 10220, "epoch": 283, "lr": 0.001} +{"train_loss": 0.03850750997662544, "train_loss_bc": 0.03850750997662544, "train_loss_llm": 0.0, "grad_norm": 0.02259615994989872, "global_step": 10221, "epoch": 283, "lr": 0.001} +{"train_loss": 0.034985415637493134, "train_loss_bc": 0.034961920231580734, "train_loss_llm": 0.00234946608543396, "grad_norm": 0.03210250288248062, "global_step": 10222, "epoch": 283, "lr": 0.001} +{"train_loss": 0.038168089309086405, "train_loss_bc": 0.03314568102359772, "train_loss_llm": 0.0, "grad_norm": 0.03842918202280998, "global_step": 10223, "epoch": 283, "lr": 0.001, "val_loss": 0.061892878264188766} +{"train_loss": 0.033842749893665314, "train_loss_bc": 0.033842749893665314, "train_loss_llm": 0.0, "grad_norm": 0.04814935103058815, "global_step": 10224, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03813274949789047, "train_loss_bc": 0.03813274949789047, "train_loss_llm": 0.0, "grad_norm": 0.024934718385338783, "global_step": 10225, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03760218992829323, "train_loss_bc": 0.03760218992829323, "train_loss_llm": 0.0, "grad_norm": 0.04164021089673042, "global_step": 10226, "epoch": 284, "lr": 0.001} +{"train_loss": 0.045642171055078506, "train_loss_bc": 0.045642171055078506, "train_loss_llm": 0.0, "grad_norm": 0.06065923720598221, "global_step": 10227, "epoch": 284, "lr": 0.001} +{"train_loss": 0.0383744090795517, "train_loss_bc": 0.0383477509021759, "train_loss_llm": 0.002665966749191284, "grad_norm": 0.0690874308347702, "global_step": 10228, "epoch": 284, "lr": 0.001} +{"train_loss": 0.02984452061355114, "train_loss_bc": 0.02984452061355114, "train_loss_llm": 0.0, "grad_norm": 0.02061295323073864, "global_step": 10229, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03836182504892349, "train_loss_bc": 0.03836182504892349, "train_loss_llm": 0.0, "grad_norm": 0.035456977784633636, "global_step": 10230, "epoch": 284, "lr": 0.001} +{"train_loss": 0.04637998715043068, "train_loss_bc": 0.046362776309251785, "train_loss_llm": 0.0017209053039550781, "grad_norm": 0.05080825835466385, "global_step": 10231, "epoch": 284, "lr": 0.001} +{"train_loss": 0.04729960858821869, "train_loss_bc": 0.04729960858821869, "train_loss_llm": 0.0, "grad_norm": 0.059063006192445755, "global_step": 10232, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03573219105601311, "train_loss_bc": 0.03573219105601311, "train_loss_llm": 0.0, "grad_norm": 0.022124044597148895, "global_step": 10233, "epoch": 284, "lr": 0.001} +{"train_loss": 0.038042180240154266, "train_loss_bc": 0.03800375759601593, "train_loss_llm": 0.0038421452045440674, "grad_norm": 0.03439972922205925, "global_step": 10234, "epoch": 284, "lr": 0.001} +{"train_loss": 0.038499489426612854, "train_loss_bc": 0.038499489426612854, "train_loss_llm": 0.0, "grad_norm": 0.05088364705443382, "global_step": 10235, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03504431992769241, "train_loss_bc": 0.03504431992769241, "train_loss_llm": 0.0, "grad_norm": 0.062020860612392426, "global_step": 10236, "epoch": 284, "lr": 0.001} +{"train_loss": 0.04189791530370712, "train_loss_bc": 0.04189791530370712, "train_loss_llm": 0.0, "grad_norm": 0.022871743887662888, "global_step": 10237, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03352999687194824, "train_loss_bc": 0.03352999687194824, "train_loss_llm": 0.0, "grad_norm": 0.03240250423550606, "global_step": 10238, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03757539764046669, "train_loss_bc": 0.03757539764046669, "train_loss_llm": 0.0, "grad_norm": 0.041493114084005356, "global_step": 10239, "epoch": 284, "lr": 0.001} +{"train_loss": 0.0404677577316761, "train_loss_bc": 0.0404677577316761, "train_loss_llm": 0.0, "grad_norm": 0.05386119708418846, "global_step": 10240, "epoch": 284, "lr": 0.001} +{"train_loss": 0.036446910351514816, "train_loss_bc": 0.036445725709199905, "train_loss_llm": 0.00011831521987915039, "grad_norm": 0.019424403086304665, "global_step": 10241, "epoch": 284, "lr": 0.001} +{"train_loss": 0.0393490344285965, "train_loss_bc": 0.0393490344285965, "train_loss_llm": 0.0, "grad_norm": 0.031006954610347748, "global_step": 10242, "epoch": 284, "lr": 0.001} +{"train_loss": 0.04629902169108391, "train_loss_bc": 0.04629902169108391, "train_loss_llm": 0.0, "grad_norm": 0.045858532190322876, "global_step": 10243, "epoch": 284, "lr": 0.001} +{"train_loss": 0.040720585733652115, "train_loss_bc": 0.040720585733652115, "train_loss_llm": 0.0, "grad_norm": 0.06369441002607346, "global_step": 10244, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03895263746380806, "train_loss_bc": 0.03895263746380806, "train_loss_llm": 0.0, "grad_norm": 0.024268627166748047, "global_step": 10245, "epoch": 284, "lr": 0.001} +{"train_loss": 0.04383954033255577, "train_loss_bc": 0.04383954033255577, "train_loss_llm": 0.0, "grad_norm": 0.05306869372725487, "global_step": 10246, "epoch": 284, "lr": 0.001} +{"train_loss": 0.04250253736972809, "train_loss_bc": 0.04243536293506622, "train_loss_llm": 0.0067174434661865234, "grad_norm": 0.061867449432611465, "global_step": 10247, "epoch": 284, "lr": 0.001} +{"train_loss": 0.037279363721609116, "train_loss_bc": 0.037279363721609116, "train_loss_llm": 0.0, "grad_norm": 0.07183635234832764, "global_step": 10248, "epoch": 284, "lr": 0.001} +{"train_loss": 0.036009281873703, "train_loss_bc": 0.035999931395053864, "train_loss_llm": 0.000934988260269165, "grad_norm": 0.02317829430103302, "global_step": 10249, "epoch": 284, "lr": 0.001} +{"train_loss": 0.04059448465704918, "train_loss_bc": 0.04059448465704918, "train_loss_llm": 0.0, "grad_norm": 0.03138093650341034, "global_step": 10250, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03813222423195839, "train_loss_bc": 0.03813222423195839, "train_loss_llm": 0.0, "grad_norm": 0.049294307827949524, "global_step": 10251, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03186944127082825, "train_loss_bc": 0.03186944127082825, "train_loss_llm": 0.0, "grad_norm": 0.06427035480737686, "global_step": 10252, "epoch": 284, "lr": 0.001} +{"train_loss": 0.035075537860393524, "train_loss_bc": 0.035075537860393524, "train_loss_llm": 0.0, "grad_norm": 0.017835788428783417, "global_step": 10253, "epoch": 284, "lr": 0.001} +{"train_loss": 0.04425305873155594, "train_loss_bc": 0.04425305873155594, "train_loss_llm": 0.0, "grad_norm": 0.030954936519265175, "global_step": 10254, "epoch": 284, "lr": 0.001} +{"train_loss": 0.02790265716612339, "train_loss_bc": 0.02790265716612339, "train_loss_llm": 0.0, "grad_norm": 0.03621501103043556, "global_step": 10255, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03992835059762001, "train_loss_bc": 0.03992835059762001, "train_loss_llm": 0.0, "grad_norm": 0.04547217860817909, "global_step": 10256, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03416363522410393, "train_loss_bc": 0.03416363522410393, "train_loss_llm": 0.0, "grad_norm": 0.01903049647808075, "global_step": 10257, "epoch": 284, "lr": 0.001} +{"train_loss": 0.038437336683273315, "train_loss_bc": 0.038437336683273315, "train_loss_llm": 0.0, "grad_norm": 0.04625867307186127, "global_step": 10258, "epoch": 284, "lr": 0.001} +{"train_loss": 0.03848045878112316, "train_loss_bc": 0.037271417677402496, "train_loss_llm": 0.0, "grad_norm": 0.06200473755598068, "global_step": 10259, "epoch": 284, "lr": 0.001, "val_loss": 0.06099690869450569} +{"train_loss": 0.042765527963638306, "train_loss_bc": 0.042765527963638306, "train_loss_llm": 0.0, "grad_norm": 0.07193245738744736, "global_step": 10260, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03616907820105553, "train_loss_bc": 0.03615358844399452, "train_loss_llm": 0.0015491247177124023, "grad_norm": 0.02086777053773403, "global_step": 10261, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03518161177635193, "train_loss_bc": 0.03518078848719597, "train_loss_llm": 8.234381675720215e-05, "grad_norm": 0.032649245113134384, "global_step": 10262, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03757616505026817, "train_loss_bc": 0.03756153956055641, "train_loss_llm": 0.0014626681804656982, "grad_norm": 0.055415187031030655, "global_step": 10263, "epoch": 285, "lr": 0.001} +{"train_loss": 0.035667531192302704, "train_loss_bc": 0.035667531192302704, "train_loss_llm": 0.0, "grad_norm": 0.0696072056889534, "global_step": 10264, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03873217850923538, "train_loss_bc": 0.03873217850923538, "train_loss_llm": 0.0, "grad_norm": 0.019257741048932076, "global_step": 10265, "epoch": 285, "lr": 0.001} +{"train_loss": 0.04460888355970383, "train_loss_bc": 0.04459065943956375, "train_loss_llm": 0.001822441816329956, "grad_norm": 0.028422754257917404, "global_step": 10266, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03886156156659126, "train_loss_bc": 0.03886156156659126, "train_loss_llm": 0.0, "grad_norm": 0.04208016023039818, "global_step": 10267, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03312632068991661, "train_loss_bc": 0.03312632068991661, "train_loss_llm": 0.0, "grad_norm": 0.05283808335661888, "global_step": 10268, "epoch": 285, "lr": 0.001} +{"train_loss": 0.04255164787173271, "train_loss_bc": 0.04255164787173271, "train_loss_llm": 0.0, "grad_norm": 0.023288799449801445, "global_step": 10269, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03827651962637901, "train_loss_bc": 0.03827651962637901, "train_loss_llm": 0.0, "grad_norm": 0.03528076037764549, "global_step": 10270, "epoch": 285, "lr": 0.001} +{"train_loss": 0.04099233075976372, "train_loss_bc": 0.04097039997577667, "train_loss_llm": 0.0021930038928985596, "grad_norm": 0.0452260747551918, "global_step": 10271, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03978747874498367, "train_loss_bc": 0.03978747874498367, "train_loss_llm": 0.0, "grad_norm": 0.059481918811798096, "global_step": 10272, "epoch": 285, "lr": 0.001} +{"train_loss": 0.039678558707237244, "train_loss_bc": 0.03967272490262985, "train_loss_llm": 0.000583350658416748, "grad_norm": 0.022152604535222054, "global_step": 10273, "epoch": 285, "lr": 0.001} +{"train_loss": 0.04146207869052887, "train_loss_bc": 0.04146207869052887, "train_loss_llm": 0.0, "grad_norm": 0.04366132989525795, "global_step": 10274, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03981499746441841, "train_loss_bc": 0.03981499746441841, "train_loss_llm": 0.0, "grad_norm": 0.05309554934501648, "global_step": 10275, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03668202832341194, "train_loss_bc": 0.03666181489825249, "train_loss_llm": 0.0020215213298797607, "grad_norm": 0.06559779495000839, "global_step": 10276, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03958868607878685, "train_loss_bc": 0.039584334939718246, "train_loss_llm": 0.00043511390686035156, "grad_norm": 0.024062301963567734, "global_step": 10277, "epoch": 285, "lr": 0.001} +{"train_loss": 0.042422983795404434, "train_loss_bc": 0.042422983795404434, "train_loss_llm": 0.0, "grad_norm": 0.031980883330106735, "global_step": 10278, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03467918932437897, "train_loss_bc": 0.03467370569705963, "train_loss_llm": 0.0005483329296112061, "grad_norm": 0.050590045750141144, "global_step": 10279, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03966823220252991, "train_loss_bc": 0.03964083269238472, "train_loss_llm": 0.0027398765087127686, "grad_norm": 0.06514063477516174, "global_step": 10280, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03465071693062782, "train_loss_bc": 0.034644778817892075, "train_loss_llm": 0.0005939602851867676, "grad_norm": 0.01981324702501297, "global_step": 10281, "epoch": 285, "lr": 0.001} +{"train_loss": 0.043115582317113876, "train_loss_bc": 0.043115582317113876, "train_loss_llm": 0.0, "grad_norm": 0.032958753407001495, "global_step": 10282, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03665275499224663, "train_loss_bc": 0.036641743034124374, "train_loss_llm": 0.0011011064052581787, "grad_norm": 0.0467798225581646, "global_step": 10283, "epoch": 285, "lr": 0.001} +{"train_loss": 0.0434470996260643, "train_loss_bc": 0.0434470996260643, "train_loss_llm": 0.0, "grad_norm": 0.06304235011339188, "global_step": 10284, "epoch": 285, "lr": 0.001} +{"train_loss": 0.040243811905384064, "train_loss_bc": 0.04022626578807831, "train_loss_llm": 0.0017545819282531738, "grad_norm": 0.026950128376483917, "global_step": 10285, "epoch": 285, "lr": 0.001} +{"train_loss": 0.04004531353712082, "train_loss_bc": 0.04004531353712082, "train_loss_llm": 0.0, "grad_norm": 0.05087072774767876, "global_step": 10286, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03767358511686325, "train_loss_bc": 0.03767358511686325, "train_loss_llm": 0.0, "grad_norm": 0.06020065397024155, "global_step": 10287, "epoch": 285, "lr": 0.001} +{"train_loss": 0.0389627069234848, "train_loss_bc": 0.0389627069234848, "train_loss_llm": 0.0, "grad_norm": 0.07787618786096573, "global_step": 10288, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03801756352186203, "train_loss_bc": 0.03801756352186203, "train_loss_llm": 0.0, "grad_norm": 0.01880061626434326, "global_step": 10289, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03300929814577103, "train_loss_bc": 0.03300929814577103, "train_loss_llm": 0.0, "grad_norm": 0.026523040607571602, "global_step": 10290, "epoch": 285, "lr": 0.001} +{"train_loss": 0.042094551026821136, "train_loss_bc": 0.042094551026821136, "train_loss_llm": 0.0, "grad_norm": 0.03450794517993927, "global_step": 10291, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03716938570141792, "train_loss_bc": 0.03714686259627342, "train_loss_llm": 0.00225222110748291, "grad_norm": 0.044294390827417374, "global_step": 10292, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03041069395840168, "train_loss_bc": 0.03040323033928871, "train_loss_llm": 0.000746309757232666, "grad_norm": 0.01879667490720749, "global_step": 10293, "epoch": 285, "lr": 0.001} +{"train_loss": 0.033309753984212875, "train_loss_bc": 0.033309753984212875, "train_loss_llm": 0.0, "grad_norm": 0.030168304219841957, "global_step": 10294, "epoch": 285, "lr": 0.001} +{"train_loss": 0.03835189533937308, "train_loss_bc": 0.03357182443141937, "train_loss_llm": 0.0, "grad_norm": 0.048234887421131134, "global_step": 10295, "epoch": 285, "lr": 0.001, "train/cumulative_reward": 130.2417031825275, "train/mean_score": 16.923915998982444, "train/success_rate": 0.4, "test/cumulative_reward": 71.84086501496725, "test/mean_score": 12.749624437102517, "test/success_rate": 0.32, "val_loss": 0.0613553449511528, "train_action_mse_error": 0.041590675711631775} +{"train_loss": 0.037933893501758575, "train_loss_bc": 0.037933893501758575, "train_loss_llm": 0.0, "grad_norm": 0.052437715232372284, "global_step": 10296, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03908737748861313, "train_loss_bc": 0.03908737748861313, "train_loss_llm": 0.0, "grad_norm": 0.03008432500064373, "global_step": 10297, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03938601538538933, "train_loss_bc": 0.03938601538538933, "train_loss_llm": 0.0, "grad_norm": 0.05289863795042038, "global_step": 10298, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03789984807372093, "train_loss_bc": 0.03789922222495079, "train_loss_llm": 6.246566772460938e-05, "grad_norm": 0.05391295254230499, "global_step": 10299, "epoch": 286, "lr": 0.001} +{"train_loss": 0.04002378508448601, "train_loss_bc": 0.04002378508448601, "train_loss_llm": 0.0, "grad_norm": 0.06106649339199066, "global_step": 10300, "epoch": 286, "lr": 0.001} +{"train_loss": 0.034255679696798325, "train_loss_bc": 0.03425220772624016, "train_loss_llm": 0.0003471076488494873, "grad_norm": 0.025998780503869057, "global_step": 10301, "epoch": 286, "lr": 0.001} +{"train_loss": 0.0449833907186985, "train_loss_bc": 0.0449833907186985, "train_loss_llm": 0.0, "grad_norm": 0.047391101717948914, "global_step": 10302, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03850908204913139, "train_loss_bc": 0.038502857089042664, "train_loss_llm": 0.0006226301193237305, "grad_norm": 0.059955157339572906, "global_step": 10303, "epoch": 286, "lr": 0.001} +{"train_loss": 0.0376468263566494, "train_loss_bc": 0.037626247853040695, "train_loss_llm": 0.002057880163192749, "grad_norm": 0.07698016613721848, "global_step": 10304, "epoch": 286, "lr": 0.001} +{"train_loss": 0.037542589008808136, "train_loss_bc": 0.03753086179494858, "train_loss_llm": 0.0011726021766662598, "grad_norm": 0.02212236262857914, "global_step": 10305, "epoch": 286, "lr": 0.001} +{"train_loss": 0.040705397725105286, "train_loss_bc": 0.040705397725105286, "train_loss_llm": 0.0, "grad_norm": 0.04203168302774429, "global_step": 10306, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03965546935796738, "train_loss_bc": 0.03963741660118103, "train_loss_llm": 0.001805335283279419, "grad_norm": 0.052349455654621124, "global_step": 10307, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03878576308488846, "train_loss_bc": 0.03878270089626312, "train_loss_llm": 0.00030624866485595703, "grad_norm": 0.0750947818160057, "global_step": 10308, "epoch": 286, "lr": 0.001} +{"train_loss": 0.041085392236709595, "train_loss_bc": 0.041085392236709595, "train_loss_llm": 0.0, "grad_norm": 0.021867474541068077, "global_step": 10309, "epoch": 286, "lr": 0.001} +{"train_loss": 0.045479487627744675, "train_loss_bc": 0.045479487627744675, "train_loss_llm": 0.0, "grad_norm": 0.035844456404447556, "global_step": 10310, "epoch": 286, "lr": 0.001} +{"train_loss": 0.039033930748701096, "train_loss_bc": 0.039033930748701096, "train_loss_llm": 0.0, "grad_norm": 0.04642066732048988, "global_step": 10311, "epoch": 286, "lr": 0.001} +{"train_loss": 0.04284319281578064, "train_loss_bc": 0.04284319281578064, "train_loss_llm": 0.0, "grad_norm": 0.05189317837357521, "global_step": 10312, "epoch": 286, "lr": 0.001} +{"train_loss": 0.0388895682990551, "train_loss_bc": 0.0388895682990551, "train_loss_llm": 0.0, "grad_norm": 0.01915912702679634, "global_step": 10313, "epoch": 286, "lr": 0.001} +{"train_loss": 0.036630161106586456, "train_loss_bc": 0.036630161106586456, "train_loss_llm": 0.0, "grad_norm": 0.037178464233875275, "global_step": 10314, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03822191059589386, "train_loss_bc": 0.03822191059589386, "train_loss_llm": 0.0, "grad_norm": 0.051022008061409, "global_step": 10315, "epoch": 286, "lr": 0.001} +{"train_loss": 0.034905754029750824, "train_loss_bc": 0.034905754029750824, "train_loss_llm": 0.0, "grad_norm": 0.06439375132322311, "global_step": 10316, "epoch": 286, "lr": 0.001} +{"train_loss": 0.030833881348371506, "train_loss_bc": 0.030833881348371506, "train_loss_llm": 0.0, "grad_norm": 0.0171798188239336, "global_step": 10317, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03910898417234421, "train_loss_bc": 0.03910898417234421, "train_loss_llm": 0.0, "grad_norm": 0.029559774324297905, "global_step": 10318, "epoch": 286, "lr": 0.001} +{"train_loss": 0.032511305063962936, "train_loss_bc": 0.03248988091945648, "train_loss_llm": 0.0021422505378723145, "grad_norm": 0.04205760732293129, "global_step": 10319, "epoch": 286, "lr": 0.001} +{"train_loss": 0.04051591083407402, "train_loss_bc": 0.04051591083407402, "train_loss_llm": 0.0, "grad_norm": 0.06280913203954697, "global_step": 10320, "epoch": 286, "lr": 0.001} +{"train_loss": 0.037243809551000595, "train_loss_bc": 0.037243809551000595, "train_loss_llm": 0.0, "grad_norm": 0.026097362861037254, "global_step": 10321, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03755667433142662, "train_loss_bc": 0.03755667433142662, "train_loss_llm": 0.0, "grad_norm": 0.03565830737352371, "global_step": 10322, "epoch": 286, "lr": 0.001} +{"train_loss": 0.035202059894800186, "train_loss_bc": 0.035202059894800186, "train_loss_llm": 0.0, "grad_norm": 0.04620562493801117, "global_step": 10323, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03534749895334244, "train_loss_bc": 0.035332873463630676, "train_loss_llm": 0.0014625191688537598, "grad_norm": 0.06102510914206505, "global_step": 10324, "epoch": 286, "lr": 0.001} +{"train_loss": 0.036270879209041595, "train_loss_bc": 0.036270879209041595, "train_loss_llm": 0.0, "grad_norm": 0.024184027686715126, "global_step": 10325, "epoch": 286, "lr": 0.001} +{"train_loss": 0.038276489824056625, "train_loss_bc": 0.038276489824056625, "train_loss_llm": 0.0, "grad_norm": 0.03466358408331871, "global_step": 10326, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03844064101576805, "train_loss_bc": 0.03844064101576805, "train_loss_llm": 0.0, "grad_norm": 0.05146604776382446, "global_step": 10327, "epoch": 286, "lr": 0.001} +{"train_loss": 0.03945441544055939, "train_loss_bc": 0.03945441544055939, "train_loss_llm": 0.0, "grad_norm": 0.06980506330728531, "global_step": 10328, "epoch": 286, "lr": 0.001} +{"train_loss": 0.04020347073674202, "train_loss_bc": 0.04018981754779816, "train_loss_llm": 0.0013654232025146484, "grad_norm": 0.02673318050801754, "global_step": 10329, "epoch": 286, "lr": 0.001} +{"train_loss": 0.04040262848138809, "train_loss_bc": 0.04040262848138809, "train_loss_llm": 0.0, "grad_norm": 0.03584444895386696, "global_step": 10330, "epoch": 286, "lr": 0.001} +{"train_loss": 0.038482843794756465, "train_loss_bc": 0.040509212762117386, "train_loss_llm": 0.0, "grad_norm": 0.05175398290157318, "global_step": 10331, "epoch": 286, "lr": 0.001, "val_loss": 0.06200845539569855} +{"train_loss": 0.03983068838715553, "train_loss_bc": 0.03981611877679825, "train_loss_llm": 0.0014568567276000977, "grad_norm": 0.06149666756391525, "global_step": 10332, "epoch": 287, "lr": 0.001} +{"train_loss": 0.04216625168919563, "train_loss_bc": 0.04216625168919563, "train_loss_llm": 0.0, "grad_norm": 0.020691778510808945, "global_step": 10333, "epoch": 287, "lr": 0.001} +{"train_loss": 0.034863390028476715, "train_loss_bc": 0.034863390028476715, "train_loss_llm": 0.0, "grad_norm": 0.03462342917919159, "global_step": 10334, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03725992515683174, "train_loss_bc": 0.03725992515683174, "train_loss_llm": 0.0, "grad_norm": 0.04727904126048088, "global_step": 10335, "epoch": 287, "lr": 0.001} +{"train_loss": 0.04669886827468872, "train_loss_bc": 0.04669886827468872, "train_loss_llm": 0.0, "grad_norm": 0.062003567814826965, "global_step": 10336, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03840741887688637, "train_loss_bc": 0.03840457648038864, "train_loss_llm": 0.0002841651439666748, "grad_norm": 0.024425534531474113, "global_step": 10337, "epoch": 287, "lr": 0.001} +{"train_loss": 0.04720480740070343, "train_loss_bc": 0.04720480740070343, "train_loss_llm": 0.0, "grad_norm": 0.033665962517261505, "global_step": 10338, "epoch": 287, "lr": 0.001} +{"train_loss": 0.040623560547828674, "train_loss_bc": 0.040623560547828674, "train_loss_llm": 0.0, "grad_norm": 0.045975882560014725, "global_step": 10339, "epoch": 287, "lr": 0.001} +{"train_loss": 0.0391094945371151, "train_loss_bc": 0.039091404527425766, "train_loss_llm": 0.0018088817596435547, "grad_norm": 0.05549092963337898, "global_step": 10340, "epoch": 287, "lr": 0.001} +{"train_loss": 0.0402013324201107, "train_loss_bc": 0.0402013324201107, "train_loss_llm": 0.0, "grad_norm": 0.019136223942041397, "global_step": 10341, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03981390967965126, "train_loss_bc": 0.03981390967965126, "train_loss_llm": 0.0, "grad_norm": 0.03590073063969612, "global_step": 10342, "epoch": 287, "lr": 0.001} +{"train_loss": 0.035999845713377, "train_loss_bc": 0.035999845713377, "train_loss_llm": 0.0, "grad_norm": 0.046749427914619446, "global_step": 10343, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03469979763031006, "train_loss_bc": 0.03469979763031006, "train_loss_llm": 0.0, "grad_norm": 0.05707219988107681, "global_step": 10344, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03981072083115578, "train_loss_bc": 0.03981072083115578, "train_loss_llm": 0.0, "grad_norm": 0.023397136479616165, "global_step": 10345, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03516748175024986, "train_loss_bc": 0.03516748175024986, "train_loss_llm": 0.0, "grad_norm": 0.03551013395190239, "global_step": 10346, "epoch": 287, "lr": 0.001} +{"train_loss": 0.04077226668596268, "train_loss_bc": 0.04077226668596268, "train_loss_llm": 0.0, "grad_norm": 0.04645896703004837, "global_step": 10347, "epoch": 287, "lr": 0.001} +{"train_loss": 0.04037361964583397, "train_loss_bc": 0.04037361964583397, "train_loss_llm": 0.0, "grad_norm": 0.058276090770959854, "global_step": 10348, "epoch": 287, "lr": 0.001} +{"train_loss": 0.0411117821931839, "train_loss_bc": 0.04110019654035568, "train_loss_llm": 0.0011586248874664307, "grad_norm": 0.02355663850903511, "global_step": 10349, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03433719277381897, "train_loss_bc": 0.03432052209973335, "train_loss_llm": 0.0016669631004333496, "grad_norm": 0.03462861850857735, "global_step": 10350, "epoch": 287, "lr": 0.001} +{"train_loss": 0.04142818599939346, "train_loss_bc": 0.04142818599939346, "train_loss_llm": 0.0, "grad_norm": 0.03955245763063431, "global_step": 10351, "epoch": 287, "lr": 0.001} +{"train_loss": 0.041219841688871384, "train_loss_bc": 0.041219841688871384, "train_loss_llm": 0.0, "grad_norm": 0.049497090280056, "global_step": 10352, "epoch": 287, "lr": 0.001} +{"train_loss": 0.04143335670232773, "train_loss_bc": 0.04143335670232773, "train_loss_llm": 0.0, "grad_norm": 0.019131161272525787, "global_step": 10353, "epoch": 287, "lr": 0.001} +{"train_loss": 0.02979225292801857, "train_loss_bc": 0.02979225292801857, "train_loss_llm": 0.0, "grad_norm": 0.0241193026304245, "global_step": 10354, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03157509118318558, "train_loss_bc": 0.03155776858329773, "train_loss_llm": 0.0017323493957519531, "grad_norm": 0.032145969569683075, "global_step": 10355, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03649047017097473, "train_loss_bc": 0.03648838400840759, "train_loss_llm": 0.00020879507064819336, "grad_norm": 0.04102695360779762, "global_step": 10356, "epoch": 287, "lr": 0.001} +{"train_loss": 0.0432329997420311, "train_loss_bc": 0.0432329997420311, "train_loss_llm": 0.0, "grad_norm": 0.024293655529618263, "global_step": 10357, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03361039236187935, "train_loss_bc": 0.03360389918088913, "train_loss_llm": 0.0006492137908935547, "grad_norm": 0.03323421999812126, "global_step": 10358, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03876524791121483, "train_loss_bc": 0.03876524791121483, "train_loss_llm": 0.0, "grad_norm": 0.046732112765312195, "global_step": 10359, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03303039073944092, "train_loss_bc": 0.03303039073944092, "train_loss_llm": 0.0, "grad_norm": 0.05625026300549507, "global_step": 10360, "epoch": 287, "lr": 0.001} +{"train_loss": 0.040028732270002365, "train_loss_bc": 0.03999679908156395, "train_loss_llm": 0.003193289041519165, "grad_norm": 0.026003485545516014, "global_step": 10361, "epoch": 287, "lr": 0.001} +{"train_loss": 0.035247933119535446, "train_loss_bc": 0.03523510321974754, "train_loss_llm": 0.0012829601764678955, "grad_norm": 0.04272032529115677, "global_step": 10362, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03685914725065231, "train_loss_bc": 0.03685914725065231, "train_loss_llm": 0.0, "grad_norm": 0.05431892350316048, "global_step": 10363, "epoch": 287, "lr": 0.001} +{"train_loss": 0.033804841339588165, "train_loss_bc": 0.033804841339588165, "train_loss_llm": 0.0, "grad_norm": 0.06435390561819077, "global_step": 10364, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03573517128825188, "train_loss_bc": 0.03573517128825188, "train_loss_llm": 0.0, "grad_norm": 0.023170026019215584, "global_step": 10365, "epoch": 287, "lr": 0.001} +{"train_loss": 0.04176909849047661, "train_loss_bc": 0.04176909849047661, "train_loss_llm": 0.0, "grad_norm": 0.03170778974890709, "global_step": 10366, "epoch": 287, "lr": 0.001} +{"train_loss": 0.03830937503112687, "train_loss_bc": 0.03666199371218681, "train_loss_llm": 0.0, "grad_norm": 0.04472104460000992, "global_step": 10367, "epoch": 287, "lr": 0.001, "val_loss": 0.06130574643611908} +{"train_loss": 0.03655554726719856, "train_loss_bc": 0.03655554726719856, "train_loss_llm": 0.0, "grad_norm": 0.05140363797545433, "global_step": 10368, "epoch": 288, "lr": 0.001} +{"train_loss": 0.038079891353845596, "train_loss_bc": 0.038079891353845596, "train_loss_llm": 0.0, "grad_norm": 0.022787969559431076, "global_step": 10369, "epoch": 288, "lr": 0.001} +{"train_loss": 0.04102252423763275, "train_loss_bc": 0.04102252423763275, "train_loss_llm": 0.0, "grad_norm": 0.03042091429233551, "global_step": 10370, "epoch": 288, "lr": 0.001} +{"train_loss": 0.03976893052458763, "train_loss_bc": 0.03976893052458763, "train_loss_llm": 0.0, "grad_norm": 0.041045088320970535, "global_step": 10371, "epoch": 288, "lr": 0.001} +{"train_loss": 0.0453728549182415, "train_loss_bc": 0.0453728549182415, "train_loss_llm": 0.0, "grad_norm": 0.056434646248817444, "global_step": 10372, "epoch": 288, "lr": 0.001} +{"train_loss": 0.03755994141101837, "train_loss_bc": 0.03753315657377243, "train_loss_llm": 0.0026785433292388916, "grad_norm": 0.021115412935614586, "global_step": 10373, "epoch": 288, "lr": 0.001} +{"train_loss": 0.030727075412869453, "train_loss_bc": 0.030727075412869453, "train_loss_llm": 0.0, "grad_norm": 0.0329306460916996, "global_step": 10374, "epoch": 288, "lr": 0.001} +{"train_loss": 0.04078928008675575, "train_loss_bc": 0.04078928008675575, "train_loss_llm": 0.0, "grad_norm": 0.04866738244891167, "global_step": 10375, "epoch": 288, "lr": 0.001} +{"train_loss": 0.03675182908773422, "train_loss_bc": 0.03675182908773422, "train_loss_llm": 0.0, "grad_norm": 0.05715803802013397, "global_step": 10376, "epoch": 288, "lr": 0.001} +{"train_loss": 0.03746654838323593, "train_loss_bc": 0.03746654838323593, "train_loss_llm": 0.0, "grad_norm": 0.020247604697942734, "global_step": 10377, "epoch": 288, "lr": 0.001} +{"train_loss": 0.042758334428071976, "train_loss_bc": 0.042755335569381714, "train_loss_llm": 0.00029981136322021484, "grad_norm": 0.03315727785229683, "global_step": 10378, "epoch": 288, "lr": 0.001} +{"train_loss": 0.04194951802492142, "train_loss_bc": 0.04192253574728966, "train_loss_llm": 0.0026981234550476074, "grad_norm": 0.0464017391204834, "global_step": 10379, "epoch": 288, "lr": 0.001} +{"train_loss": 0.035738032311201096, "train_loss_bc": 0.035738032311201096, "train_loss_llm": 0.0, "grad_norm": 0.05482713133096695, "global_step": 10380, "epoch": 288, "lr": 0.001} +{"train_loss": 0.035767171531915665, "train_loss_bc": 0.035767171531915665, "train_loss_llm": 0.0, "grad_norm": 0.023748261854052544, "global_step": 10381, "epoch": 288, "lr": 0.001} +{"train_loss": 0.047195181250572205, "train_loss_bc": 0.047195181250572205, "train_loss_llm": 0.0, "grad_norm": 0.037630725651979446, "global_step": 10382, "epoch": 288, "lr": 0.001} +{"train_loss": 0.032235562801361084, "train_loss_bc": 0.032235562801361084, "train_loss_llm": 0.0, "grad_norm": 0.045236241072416306, "global_step": 10383, "epoch": 288, "lr": 0.001} +{"train_loss": 0.03522650897502899, "train_loss_bc": 0.03522650897502899, "train_loss_llm": 0.0, "grad_norm": 0.05551588535308838, "global_step": 10384, "epoch": 288, "lr": 0.001} +{"train_loss": 0.032101280987262726, "train_loss_bc": 0.032101280987262726, "train_loss_llm": 0.0, "grad_norm": 0.017515890300273895, "global_step": 10385, "epoch": 288, "lr": 0.001} +{"train_loss": 0.035017818212509155, "train_loss_bc": 0.035017818212509155, "train_loss_llm": 0.0, "grad_norm": 0.026468321681022644, "global_step": 10386, "epoch": 288, "lr": 0.001} +{"train_loss": 0.035395726561546326, "train_loss_bc": 0.035395726561546326, "train_loss_llm": 0.0, "grad_norm": 0.0401601567864418, "global_step": 10387, "epoch": 288, "lr": 0.001} +{"train_loss": 0.0359443798661232, "train_loss_bc": 0.0359443798661232, "train_loss_llm": 0.0, "grad_norm": 0.0393650084733963, "global_step": 10388, "epoch": 288, "lr": 0.001} +{"train_loss": 0.03555753454566002, "train_loss_bc": 0.03555753454566002, "train_loss_llm": 0.0, "grad_norm": 0.01818639598786831, "global_step": 10389, "epoch": 288, "lr": 0.001} +{"train_loss": 0.03787117823958397, "train_loss_bc": 0.03784654289484024, "train_loss_llm": 0.002463698387145996, "grad_norm": 0.029498357325792313, "global_step": 10390, "epoch": 288, "lr": 0.001} +{"train_loss": 0.04245144501328468, "train_loss_bc": 0.042424071580171585, "train_loss_llm": 0.0027372539043426514, "grad_norm": 0.03997839242219925, "global_step": 10391, "epoch": 288, "lr": 0.001} +{"train_loss": 0.04348449781537056, "train_loss_bc": 0.04348449781537056, "train_loss_llm": 0.0, "grad_norm": 0.05290931463241577, "global_step": 10392, "epoch": 288, "lr": 0.001} +{"train_loss": 0.03959382697939873, "train_loss_bc": 0.03959382697939873, "train_loss_llm": 0.0, "grad_norm": 0.02074737474322319, "global_step": 10393, "epoch": 288, "lr": 0.001} +{"train_loss": 0.04235508665442467, "train_loss_bc": 0.04235508665442467, "train_loss_llm": 0.0, "grad_norm": 0.04088550806045532, "global_step": 10394, "epoch": 288, "lr": 0.001} +{"train_loss": 0.04072679206728935, "train_loss_bc": 0.04072679206728935, "train_loss_llm": 0.0, "grad_norm": 0.05422143638134003, "global_step": 10395, "epoch": 288, "lr": 0.001} +{"train_loss": 0.0385601744055748, "train_loss_bc": 0.0385601744055748, "train_loss_llm": 0.0, "grad_norm": 0.06964659690856934, "global_step": 10396, "epoch": 288, "lr": 0.001} +{"train_loss": 0.030326543375849724, "train_loss_bc": 0.03030884452164173, "train_loss_llm": 0.0017698407173156738, "grad_norm": 0.01850971393287182, "global_step": 10397, "epoch": 288, "lr": 0.001} +{"train_loss": 0.037793464958667755, "train_loss_bc": 0.037793464958667755, "train_loss_llm": 0.0, "grad_norm": 0.028904112055897713, "global_step": 10398, "epoch": 288, "lr": 0.001} +{"train_loss": 0.035811711102724075, "train_loss_bc": 0.035811711102724075, "train_loss_llm": 0.0, "grad_norm": 0.04829370230436325, "global_step": 10399, "epoch": 288, "lr": 0.001} +{"train_loss": 0.03702523559331894, "train_loss_bc": 0.03702523559331894, "train_loss_llm": 0.0, "grad_norm": 0.05845591425895691, "global_step": 10400, "epoch": 288, "lr": 0.001} +{"train_loss": 0.04129967465996742, "train_loss_bc": 0.04129967465996742, "train_loss_llm": 0.0, "grad_norm": 0.02014395408332348, "global_step": 10401, "epoch": 288, "lr": 0.001} +{"train_loss": 0.04013296589255333, "train_loss_bc": 0.04013296589255333, "train_loss_llm": 0.0, "grad_norm": 0.031331777572631836, "global_step": 10402, "epoch": 288, "lr": 0.001} +{"train_loss": 0.03833032678812742, "train_loss_bc": 0.04345681890845299, "train_loss_llm": 0.002087712287902832, "grad_norm": 0.04774883762001991, "global_step": 10403, "epoch": 288, "lr": 0.001, "val_loss": 0.06125672906637192} +{"train_loss": 0.036358319222927094, "train_loss_bc": 0.036358319222927094, "train_loss_llm": 0.0, "grad_norm": 0.06422743201255798, "global_step": 10404, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03451275825500488, "train_loss_bc": 0.03451275825500488, "train_loss_llm": 0.0, "grad_norm": 0.01870560087263584, "global_step": 10405, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03622152656316757, "train_loss_bc": 0.03622152656316757, "train_loss_llm": 0.0, "grad_norm": 0.02656572498381138, "global_step": 10406, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03544104844331741, "train_loss_bc": 0.03544104844331741, "train_loss_llm": 0.0, "grad_norm": 0.0291900672018528, "global_step": 10407, "epoch": 289, "lr": 0.001} +{"train_loss": 0.042559556663036346, "train_loss_bc": 0.042559556663036346, "train_loss_llm": 0.0, "grad_norm": 0.035284094512462616, "global_step": 10408, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03803543373942375, "train_loss_bc": 0.03803543373942375, "train_loss_llm": 0.0, "grad_norm": 0.025819377973675728, "global_step": 10409, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03893711045384407, "train_loss_bc": 0.03893711045384407, "train_loss_llm": 0.0, "grad_norm": 0.03833988308906555, "global_step": 10410, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03973599523305893, "train_loss_bc": 0.03973599523305893, "train_loss_llm": 0.0, "grad_norm": 0.05394795909523964, "global_step": 10411, "epoch": 289, "lr": 0.001} +{"train_loss": 0.036313582211732864, "train_loss_bc": 0.036313582211732864, "train_loss_llm": 0.0, "grad_norm": 0.0601985789835453, "global_step": 10412, "epoch": 289, "lr": 0.001} +{"train_loss": 0.0332183800637722, "train_loss_bc": 0.0332183800637722, "train_loss_llm": 0.0, "grad_norm": 0.022670011967420578, "global_step": 10413, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03638650104403496, "train_loss_bc": 0.03638647496700287, "train_loss_llm": 2.6226043701171875e-06, "grad_norm": 0.03324083611369133, "global_step": 10414, "epoch": 289, "lr": 0.001} +{"train_loss": 0.034879256039857864, "train_loss_bc": 0.034879256039857864, "train_loss_llm": 0.0, "grad_norm": 0.039611849933862686, "global_step": 10415, "epoch": 289, "lr": 0.001} +{"train_loss": 0.04145108163356781, "train_loss_bc": 0.04145108163356781, "train_loss_llm": 0.0, "grad_norm": 0.04450472071766853, "global_step": 10416, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03495251387357712, "train_loss_bc": 0.03495251387357712, "train_loss_llm": 0.0, "grad_norm": 0.019101876765489578, "global_step": 10417, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03021513856947422, "train_loss_bc": 0.03021513856947422, "train_loss_llm": 0.0, "grad_norm": 0.02663208730518818, "global_step": 10418, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03696826845407486, "train_loss_bc": 0.03696826845407486, "train_loss_llm": 0.0, "grad_norm": 0.03763740509748459, "global_step": 10419, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03962346911430359, "train_loss_bc": 0.03962346911430359, "train_loss_llm": 0.0, "grad_norm": 0.0484296940267086, "global_step": 10420, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03800303861498833, "train_loss_bc": 0.037992801517248154, "train_loss_llm": 0.0010235905647277832, "grad_norm": 0.018105313181877136, "global_step": 10421, "epoch": 289, "lr": 0.001} +{"train_loss": 0.0384732186794281, "train_loss_bc": 0.0384732186794281, "train_loss_llm": 0.0, "grad_norm": 0.02622300013899803, "global_step": 10422, "epoch": 289, "lr": 0.001} +{"train_loss": 0.04037703946232796, "train_loss_bc": 0.04037703946232796, "train_loss_llm": 0.0, "grad_norm": 0.03797469288110733, "global_step": 10423, "epoch": 289, "lr": 0.001} +{"train_loss": 0.042930975556373596, "train_loss_bc": 0.042930975556373596, "train_loss_llm": 0.0, "grad_norm": 0.0436117984354496, "global_step": 10424, "epoch": 289, "lr": 0.001} +{"train_loss": 0.04396124929189682, "train_loss_bc": 0.04396124929189682, "train_loss_llm": 0.0, "grad_norm": 0.02219424583017826, "global_step": 10425, "epoch": 289, "lr": 0.001} +{"train_loss": 0.04799755662679672, "train_loss_bc": 0.04799755662679672, "train_loss_llm": 0.0, "grad_norm": 0.03802845627069473, "global_step": 10426, "epoch": 289, "lr": 0.001} +{"train_loss": 0.043669819831848145, "train_loss_bc": 0.043669819831848145, "train_loss_llm": 0.0, "grad_norm": 0.04829314351081848, "global_step": 10427, "epoch": 289, "lr": 0.001} +{"train_loss": 0.035376694053411484, "train_loss_bc": 0.035376694053411484, "train_loss_llm": 0.0, "grad_norm": 0.04965818300843239, "global_step": 10428, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03906062990427017, "train_loss_bc": 0.03906062990427017, "train_loss_llm": 0.0, "grad_norm": 0.01760043017566204, "global_step": 10429, "epoch": 289, "lr": 0.001} +{"train_loss": 0.04101049154996872, "train_loss_bc": 0.04101049154996872, "train_loss_llm": 0.0, "grad_norm": 0.027686437591910362, "global_step": 10430, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03703966736793518, "train_loss_bc": 0.0370376892387867, "train_loss_llm": 0.0001977682113647461, "grad_norm": 0.03735193610191345, "global_step": 10431, "epoch": 289, "lr": 0.001} +{"train_loss": 0.030759911984205246, "train_loss_bc": 0.030759911984205246, "train_loss_llm": 0.0, "grad_norm": 0.05131317302584648, "global_step": 10432, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03248320892453194, "train_loss_bc": 0.03248320892453194, "train_loss_llm": 0.0, "grad_norm": 0.022136317566037178, "global_step": 10433, "epoch": 289, "lr": 0.001} +{"train_loss": 0.035015471279621124, "train_loss_bc": 0.03500230610370636, "train_loss_llm": 0.0013165771961212158, "grad_norm": 0.02686229534447193, "global_step": 10434, "epoch": 289, "lr": 0.001} +{"train_loss": 0.040126726031303406, "train_loss_bc": 0.040118854492902756, "train_loss_llm": 0.0007870197296142578, "grad_norm": 0.03940552473068237, "global_step": 10435, "epoch": 289, "lr": 0.001} +{"train_loss": 0.041593037545681, "train_loss_bc": 0.041593037545681, "train_loss_llm": 0.0, "grad_norm": 0.046963389962911606, "global_step": 10436, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03530211001634598, "train_loss_bc": 0.03530211001634598, "train_loss_llm": 0.0, "grad_norm": 0.018359344452619553, "global_step": 10437, "epoch": 289, "lr": 0.001} +{"train_loss": 0.044238049536943436, "train_loss_bc": 0.04420962557196617, "train_loss_llm": 0.0028423964977264404, "grad_norm": 0.029121553525328636, "global_step": 10438, "epoch": 289, "lr": 0.001} +{"train_loss": 0.03803310082811448, "train_loss_bc": 0.035957660526037216, "train_loss_llm": 0.0005134046077728271, "grad_norm": 0.04253467172384262, "global_step": 10439, "epoch": 289, "lr": 0.001, "val_loss": 0.06065049022436142} +{"train_loss": 0.04015401005744934, "train_loss_bc": 0.04015401005744934, "train_loss_llm": 0.0, "grad_norm": 0.04931100830435753, "global_step": 10440, "epoch": 290, "lr": 0.001} +{"train_loss": 0.035297077149152756, "train_loss_bc": 0.035297077149152756, "train_loss_llm": 0.0, "grad_norm": 0.01558397151529789, "global_step": 10441, "epoch": 290, "lr": 0.001} +{"train_loss": 0.028973538428544998, "train_loss_bc": 0.02895798161625862, "train_loss_llm": 0.00155562162399292, "grad_norm": 0.023876868188381195, "global_step": 10442, "epoch": 290, "lr": 0.001} +{"train_loss": 0.039843734353780746, "train_loss_bc": 0.039843734353780746, "train_loss_llm": 0.0, "grad_norm": 0.03267398104071617, "global_step": 10443, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03882799297571182, "train_loss_bc": 0.038812439888715744, "train_loss_llm": 0.0015551745891571045, "grad_norm": 0.03807349130511284, "global_step": 10444, "epoch": 290, "lr": 0.001} +{"train_loss": 0.042799536138772964, "train_loss_bc": 0.042799536138772964, "train_loss_llm": 0.0, "grad_norm": 0.022742265835404396, "global_step": 10445, "epoch": 290, "lr": 0.001} +{"train_loss": 0.043782856315374374, "train_loss_bc": 0.043782856315374374, "train_loss_llm": 0.0, "grad_norm": 0.04446133226156235, "global_step": 10446, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03900780901312828, "train_loss_bc": 0.03900780901312828, "train_loss_llm": 0.0, "grad_norm": 0.052665825933218, "global_step": 10447, "epoch": 290, "lr": 0.001} +{"train_loss": 0.036662254482507706, "train_loss_bc": 0.036662254482507706, "train_loss_llm": 0.0, "grad_norm": 0.06224777176976204, "global_step": 10448, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03975904360413551, "train_loss_bc": 0.039756420999765396, "train_loss_llm": 0.00026226043701171875, "grad_norm": 0.018157999962568283, "global_step": 10449, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03487549349665642, "train_loss_bc": 0.034856971353292465, "train_loss_llm": 0.0018523931503295898, "grad_norm": 0.026735380291938782, "global_step": 10450, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03640742227435112, "train_loss_bc": 0.03640742227435112, "train_loss_llm": 0.0, "grad_norm": 0.0374910905957222, "global_step": 10451, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03938498720526695, "train_loss_bc": 0.03938498720526695, "train_loss_llm": 0.0, "grad_norm": 0.05288510024547577, "global_step": 10452, "epoch": 290, "lr": 0.001} +{"train_loss": 0.04926538094878197, "train_loss_bc": 0.04926538094878197, "train_loss_llm": 0.0, "grad_norm": 0.025501253083348274, "global_step": 10453, "epoch": 290, "lr": 0.001} +{"train_loss": 0.035365957766771317, "train_loss_bc": 0.035365957766771317, "train_loss_llm": 0.0, "grad_norm": 0.04347914829850197, "global_step": 10454, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03857872262597084, "train_loss_bc": 0.03852048143744469, "train_loss_llm": 0.005824148654937744, "grad_norm": 0.05218684673309326, "global_step": 10455, "epoch": 290, "lr": 0.001} +{"train_loss": 0.042184725403785706, "train_loss_bc": 0.042184725403785706, "train_loss_llm": 0.0, "grad_norm": 0.061802640557289124, "global_step": 10456, "epoch": 290, "lr": 0.001} +{"train_loss": 0.037675920873880386, "train_loss_bc": 0.037675920873880386, "train_loss_llm": 0.0, "grad_norm": 0.01797039434313774, "global_step": 10457, "epoch": 290, "lr": 0.001} +{"train_loss": 0.04081398993730545, "train_loss_bc": 0.04081207886338234, "train_loss_llm": 0.00019115209579467773, "grad_norm": 0.032847821712493896, "global_step": 10458, "epoch": 290, "lr": 0.001} +{"train_loss": 0.036976803094148636, "train_loss_bc": 0.036976803094148636, "train_loss_llm": 0.0, "grad_norm": 0.04126022383570671, "global_step": 10459, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03713977709412575, "train_loss_bc": 0.03713977709412575, "train_loss_llm": 0.0, "grad_norm": 0.048891037702560425, "global_step": 10460, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03759707883000374, "train_loss_bc": 0.03759707883000374, "train_loss_llm": 0.0, "grad_norm": 0.021912824362516403, "global_step": 10461, "epoch": 290, "lr": 0.001} +{"train_loss": 0.044191937893629074, "train_loss_bc": 0.044191937893629074, "train_loss_llm": 0.0, "grad_norm": 0.04015026614069939, "global_step": 10462, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03565558046102524, "train_loss_bc": 0.03565558046102524, "train_loss_llm": 0.0, "grad_norm": 0.04683658853173256, "global_step": 10463, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03897795081138611, "train_loss_bc": 0.038949593901634216, "train_loss_llm": 0.002835780382156372, "grad_norm": 0.06178560480475426, "global_step": 10464, "epoch": 290, "lr": 0.001} +{"train_loss": 0.041051845997571945, "train_loss_bc": 0.041051845997571945, "train_loss_llm": 0.0, "grad_norm": 0.01946515403687954, "global_step": 10465, "epoch": 290, "lr": 0.001} +{"train_loss": 0.034801483154296875, "train_loss_bc": 0.03479670733213425, "train_loss_llm": 0.0004776120185852051, "grad_norm": 0.033150069415569305, "global_step": 10466, "epoch": 290, "lr": 0.001} +{"train_loss": 0.031302593648433685, "train_loss_bc": 0.031302593648433685, "train_loss_llm": 0.0, "grad_norm": 0.03887898847460747, "global_step": 10467, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03542338311672211, "train_loss_bc": 0.03542338311672211, "train_loss_llm": 0.0, "grad_norm": 0.04700455069541931, "global_step": 10468, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03302748501300812, "train_loss_bc": 0.03302748501300812, "train_loss_llm": 0.0, "grad_norm": 0.02098672091960907, "global_step": 10469, "epoch": 290, "lr": 0.001} +{"train_loss": 0.040613383054733276, "train_loss_bc": 0.040613383054733276, "train_loss_llm": 0.0, "grad_norm": 0.034178610891103745, "global_step": 10470, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03010708838701248, "train_loss_bc": 0.030101343989372253, "train_loss_llm": 0.0005744397640228271, "grad_norm": 0.0464390367269516, "global_step": 10471, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03502514585852623, "train_loss_bc": 0.03502514585852623, "train_loss_llm": 0.0, "grad_norm": 0.051404792815446854, "global_step": 10472, "epoch": 290, "lr": 0.001} +{"train_loss": 0.04720715060830116, "train_loss_bc": 0.04719184339046478, "train_loss_llm": 0.0015306174755096436, "grad_norm": 0.030001293867826462, "global_step": 10473, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03693229332566261, "train_loss_bc": 0.03693229332566261, "train_loss_llm": 0.0, "grad_norm": 0.032686732709407806, "global_step": 10474, "epoch": 290, "lr": 0.001} +{"train_loss": 0.03800426061368651, "train_loss_bc": 0.032455481588840485, "train_loss_llm": 0.0006467700004577637, "grad_norm": 0.036501720547676086, "global_step": 10475, "epoch": 290, "lr": 0.001, "train/cumulative_reward": 89.96386465761552, "train/mean_score": 11.728085310550542, "train/success_rate": 0.2, "test/cumulative_reward": 54.63706770763289, "test/mean_score": 9.5169692580971, "test/success_rate": 0.22, "val_loss": 0.0618865080177784, "train_action_mse_error": 0.041719794273376465} +{"train_loss": 0.043637391179800034, "train_loss_bc": 0.04361945390701294, "train_loss_llm": 0.0017936527729034424, "grad_norm": 0.042726803570985794, "global_step": 10476, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03536810353398323, "train_loss_bc": 0.03535667434334755, "train_loss_llm": 0.0011428296566009521, "grad_norm": 0.02148202806711197, "global_step": 10477, "epoch": 291, "lr": 0.001} +{"train_loss": 0.038134053349494934, "train_loss_bc": 0.038134053349494934, "train_loss_llm": 0.0, "grad_norm": 0.030592959374189377, "global_step": 10478, "epoch": 291, "lr": 0.001} +{"train_loss": 0.039538703858852386, "train_loss_bc": 0.039538703858852386, "train_loss_llm": 0.0, "grad_norm": 0.05221804976463318, "global_step": 10479, "epoch": 291, "lr": 0.001} +{"train_loss": 0.032420478761196136, "train_loss_bc": 0.032416582107543945, "train_loss_llm": 0.0003895759582519531, "grad_norm": 0.0573236458003521, "global_step": 10480, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03760667145252228, "train_loss_bc": 0.03757450729608536, "train_loss_llm": 0.003216564655303955, "grad_norm": 0.01753147505223751, "global_step": 10481, "epoch": 291, "lr": 0.001} +{"train_loss": 0.036432333290576935, "train_loss_bc": 0.0364285409450531, "train_loss_llm": 0.000379413366317749, "grad_norm": 0.026857146993279457, "global_step": 10482, "epoch": 291, "lr": 0.001} +{"train_loss": 0.038131531327962875, "train_loss_bc": 0.038131531327962875, "train_loss_llm": 0.0, "grad_norm": 0.030933599919080734, "global_step": 10483, "epoch": 291, "lr": 0.001} +{"train_loss": 0.04323341324925423, "train_loss_bc": 0.04323341324925423, "train_loss_llm": 0.0, "grad_norm": 0.0345364548265934, "global_step": 10484, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03572281822562218, "train_loss_bc": 0.03572281822562218, "train_loss_llm": 0.0, "grad_norm": 0.02459646947681904, "global_step": 10485, "epoch": 291, "lr": 0.001} +{"train_loss": 0.04263630881905556, "train_loss_bc": 0.04263630881905556, "train_loss_llm": 0.0, "grad_norm": 0.03802650794386864, "global_step": 10486, "epoch": 291, "lr": 0.001} +{"train_loss": 0.034185007214546204, "train_loss_bc": 0.03417817875742912, "train_loss_llm": 0.0006830096244812012, "grad_norm": 0.04916777089238167, "global_step": 10487, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03835035115480423, "train_loss_bc": 0.03835035115480423, "train_loss_llm": 0.0, "grad_norm": 0.0546395368874073, "global_step": 10488, "epoch": 291, "lr": 0.001} +{"train_loss": 0.037483979016542435, "train_loss_bc": 0.037483979016542435, "train_loss_llm": 0.0, "grad_norm": 0.020249363034963608, "global_step": 10489, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03658610209822655, "train_loss_bc": 0.03658610209822655, "train_loss_llm": 0.0, "grad_norm": 0.028189493343234062, "global_step": 10490, "epoch": 291, "lr": 0.001} +{"train_loss": 0.039508506655693054, "train_loss_bc": 0.039508506655693054, "train_loss_llm": 0.0, "grad_norm": 0.04418520629405975, "global_step": 10491, "epoch": 291, "lr": 0.001} +{"train_loss": 0.04038110747933388, "train_loss_bc": 0.040365371853113174, "train_loss_llm": 0.0015737414360046387, "grad_norm": 0.05723458528518677, "global_step": 10492, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03486166149377823, "train_loss_bc": 0.03484886512160301, "train_loss_llm": 0.0012796223163604736, "grad_norm": 0.02001018635928631, "global_step": 10493, "epoch": 291, "lr": 0.001} +{"train_loss": 0.04278693348169327, "train_loss_bc": 0.04276396334171295, "train_loss_llm": 0.0022968649864196777, "grad_norm": 0.04425477609038353, "global_step": 10494, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03638691455125809, "train_loss_bc": 0.03638691455125809, "train_loss_llm": 0.0, "grad_norm": 0.06023336946964264, "global_step": 10495, "epoch": 291, "lr": 0.001} +{"train_loss": 0.041531581431627274, "train_loss_bc": 0.041531581431627274, "train_loss_llm": 0.0, "grad_norm": 0.07164575159549713, "global_step": 10496, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03417634218931198, "train_loss_bc": 0.03417343646287918, "train_loss_llm": 0.0002903938293457031, "grad_norm": 0.02029421180486679, "global_step": 10497, "epoch": 291, "lr": 0.001} +{"train_loss": 0.0407242588698864, "train_loss_bc": 0.0407242588698864, "train_loss_llm": 0.0, "grad_norm": 0.029759792611002922, "global_step": 10498, "epoch": 291, "lr": 0.001} +{"train_loss": 0.034366361796855927, "train_loss_bc": 0.034366361796855927, "train_loss_llm": 0.0, "grad_norm": 0.03830694779753685, "global_step": 10499, "epoch": 291, "lr": 0.001} +{"train_loss": 0.035367514938116074, "train_loss_bc": 0.03536435589194298, "train_loss_llm": 0.0003159046173095703, "grad_norm": 0.048185285180807114, "global_step": 10500, "epoch": 291, "lr": 0.001} +{"train_loss": 0.038577593863010406, "train_loss_bc": 0.038577593863010406, "train_loss_llm": 0.0, "grad_norm": 0.020174788311123848, "global_step": 10501, "epoch": 291, "lr": 0.001} +{"train_loss": 0.038359157741069794, "train_loss_bc": 0.038359157741069794, "train_loss_llm": 0.0, "grad_norm": 0.031288545578718185, "global_step": 10502, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03702649101614952, "train_loss_bc": 0.03702649101614952, "train_loss_llm": 0.0, "grad_norm": 0.04052143543958664, "global_step": 10503, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03683362156152725, "train_loss_bc": 0.03683362156152725, "train_loss_llm": 0.0, "grad_norm": 0.05027033016085625, "global_step": 10504, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03954955190420151, "train_loss_bc": 0.03954955190420151, "train_loss_llm": 0.0, "grad_norm": 0.017792558297514915, "global_step": 10505, "epoch": 291, "lr": 0.001} +{"train_loss": 0.042985156178474426, "train_loss_bc": 0.042985156178474426, "train_loss_llm": 0.0, "grad_norm": 0.03192055597901344, "global_step": 10506, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03127375245094299, "train_loss_bc": 0.031251437962055206, "train_loss_llm": 0.002231299877166748, "grad_norm": 0.039631325751543045, "global_step": 10507, "epoch": 291, "lr": 0.001} +{"train_loss": 0.04023205488920212, "train_loss_bc": 0.0402117520570755, "train_loss_llm": 0.002030402421951294, "grad_norm": 0.05195501074194908, "global_step": 10508, "epoch": 291, "lr": 0.001} +{"train_loss": 0.035464003682136536, "train_loss_bc": 0.035456620156764984, "train_loss_llm": 0.0007382333278656006, "grad_norm": 0.022776946425437927, "global_step": 10509, "epoch": 291, "lr": 0.001} +{"train_loss": 0.03900367394089699, "train_loss_bc": 0.03900367394089699, "train_loss_llm": 0.0, "grad_norm": 0.037855442613363266, "global_step": 10510, "epoch": 291, "lr": 0.001} +{"train_loss": 0.0380990712179078, "train_loss_bc": 0.04270307719707489, "train_loss_llm": 0.0, "grad_norm": 0.06422290951013565, "global_step": 10511, "epoch": 291, "lr": 0.001, "val_loss": 0.06155664473772049} +{"train_loss": 0.03465661779046059, "train_loss_bc": 0.03465661779046059, "train_loss_llm": 0.0, "grad_norm": 0.07432679831981659, "global_step": 10512, "epoch": 292, "lr": 0.001} +{"train_loss": 0.039764247834682465, "train_loss_bc": 0.03975565731525421, "train_loss_llm": 0.0008591711521148682, "grad_norm": 0.019938426092267036, "global_step": 10513, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03511321172118187, "train_loss_bc": 0.03508087992668152, "train_loss_llm": 0.0032332539558410645, "grad_norm": 0.029423212632536888, "global_step": 10514, "epoch": 292, "lr": 0.001} +{"train_loss": 0.04072902724146843, "train_loss_bc": 0.04072435200214386, "train_loss_llm": 0.00046747922897338867, "grad_norm": 0.037854284048080444, "global_step": 10515, "epoch": 292, "lr": 0.001} +{"train_loss": 0.0391099713742733, "train_loss_bc": 0.0391099713742733, "train_loss_llm": 0.0, "grad_norm": 0.04640159383416176, "global_step": 10516, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03836562857031822, "train_loss_bc": 0.03836562857031822, "train_loss_llm": 0.0, "grad_norm": 0.02009398117661476, "global_step": 10517, "epoch": 292, "lr": 0.001} +{"train_loss": 0.036772288382053375, "train_loss_bc": 0.036772288382053375, "train_loss_llm": 0.0, "grad_norm": 0.02853953279554844, "global_step": 10518, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03683464601635933, "train_loss_bc": 0.03683464601635933, "train_loss_llm": 0.0, "grad_norm": 0.03930428996682167, "global_step": 10519, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03643282130360603, "train_loss_bc": 0.03643282130360603, "train_loss_llm": 0.0, "grad_norm": 0.04675036296248436, "global_step": 10520, "epoch": 292, "lr": 0.001} +{"train_loss": 0.04342779889702797, "train_loss_bc": 0.04342179000377655, "train_loss_llm": 0.0006007254123687744, "grad_norm": 0.026816124096512794, "global_step": 10521, "epoch": 292, "lr": 0.001} +{"train_loss": 0.036763329058885574, "train_loss_bc": 0.036753345280885696, "train_loss_llm": 0.0009983181953430176, "grad_norm": 0.035664137452840805, "global_step": 10522, "epoch": 292, "lr": 0.001} +{"train_loss": 0.04122527316212654, "train_loss_bc": 0.041218239814043045, "train_loss_llm": 0.000703275203704834, "grad_norm": 0.04542415216565132, "global_step": 10523, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03307313472032547, "train_loss_bc": 0.03307313472032547, "train_loss_llm": 0.0, "grad_norm": 0.05040019005537033, "global_step": 10524, "epoch": 292, "lr": 0.001} +{"train_loss": 0.039443716406822205, "train_loss_bc": 0.039435237646102905, "train_loss_llm": 0.0008480250835418701, "grad_norm": 0.023262616246938705, "global_step": 10525, "epoch": 292, "lr": 0.001} +{"train_loss": 0.04527057334780693, "train_loss_bc": 0.04527057334780693, "train_loss_llm": 0.0, "grad_norm": 0.03600054979324341, "global_step": 10526, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03584582358598709, "train_loss_bc": 0.03584582358598709, "train_loss_llm": 0.0, "grad_norm": 0.04354216158390045, "global_step": 10527, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03223891928792, "train_loss_bc": 0.03223891928792, "train_loss_llm": 0.0, "grad_norm": 0.049299899488687515, "global_step": 10528, "epoch": 292, "lr": 0.001} +{"train_loss": 0.0418509915471077, "train_loss_bc": 0.04183884710073471, "train_loss_llm": 0.0012145936489105225, "grad_norm": 0.02352858893573284, "global_step": 10529, "epoch": 292, "lr": 0.001} +{"train_loss": 0.041469551622867584, "train_loss_bc": 0.041469551622867584, "train_loss_llm": 0.0, "grad_norm": 0.039217863231897354, "global_step": 10530, "epoch": 292, "lr": 0.001} +{"train_loss": 0.039250340312719345, "train_loss_bc": 0.039250340312719345, "train_loss_llm": 0.0, "grad_norm": 0.04554617404937744, "global_step": 10531, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03374640271067619, "train_loss_bc": 0.03374640271067619, "train_loss_llm": 0.0, "grad_norm": 0.051468972116708755, "global_step": 10532, "epoch": 292, "lr": 0.001} +{"train_loss": 0.04029243811964989, "train_loss_bc": 0.04029243811964989, "train_loss_llm": 0.0, "grad_norm": 0.02483157254755497, "global_step": 10533, "epoch": 292, "lr": 0.001} +{"train_loss": 0.0426759272813797, "train_loss_bc": 0.0426759272813797, "train_loss_llm": 0.0, "grad_norm": 0.04031047224998474, "global_step": 10534, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03337554261088371, "train_loss_bc": 0.03337554261088371, "train_loss_llm": 0.0, "grad_norm": 0.060759931802749634, "global_step": 10535, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03634519875049591, "train_loss_bc": 0.03634519875049591, "train_loss_llm": 0.0, "grad_norm": 0.07287126034498215, "global_step": 10536, "epoch": 292, "lr": 0.001} +{"train_loss": 0.042618993669748306, "train_loss_bc": 0.042618993669748306, "train_loss_llm": 0.0, "grad_norm": 0.0302384402602911, "global_step": 10537, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03576800599694252, "train_loss_bc": 0.03576800599694252, "train_loss_llm": 0.0, "grad_norm": 0.03886266052722931, "global_step": 10538, "epoch": 292, "lr": 0.001} +{"train_loss": 0.04720127210021019, "train_loss_bc": 0.047192785888910294, "train_loss_llm": 0.0008485615253448486, "grad_norm": 0.04630134254693985, "global_step": 10539, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03448938578367233, "train_loss_bc": 0.03448938578367233, "train_loss_llm": 0.0, "grad_norm": 0.059862978756427765, "global_step": 10540, "epoch": 292, "lr": 0.001} +{"train_loss": 0.039904214441776276, "train_loss_bc": 0.039880577474832535, "train_loss_llm": 0.00236356258392334, "grad_norm": 0.020312979817390442, "global_step": 10541, "epoch": 292, "lr": 0.001} +{"train_loss": 0.030543651431798935, "train_loss_bc": 0.030543651431798935, "train_loss_llm": 0.0, "grad_norm": 0.03474809229373932, "global_step": 10542, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03153614327311516, "train_loss_bc": 0.03153614327311516, "train_loss_llm": 0.0, "grad_norm": 0.04020129144191742, "global_step": 10543, "epoch": 292, "lr": 0.001} +{"train_loss": 0.039355985820293427, "train_loss_bc": 0.039355985820293427, "train_loss_llm": 0.0, "grad_norm": 0.04983550310134888, "global_step": 10544, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03904971480369568, "train_loss_bc": 0.03904971480369568, "train_loss_llm": 0.0, "grad_norm": 0.015088500455021858, "global_step": 10545, "epoch": 292, "lr": 0.001} +{"train_loss": 0.03490453585982323, "train_loss_bc": 0.03490453585982323, "train_loss_llm": 0.0, "grad_norm": 0.026259569451212883, "global_step": 10546, "epoch": 292, "lr": 0.001} +{"train_loss": 0.0380370214374529, "train_loss_bc": 0.0398874469101429, "train_loss_llm": 0.0, "grad_norm": 0.03656374290585518, "global_step": 10547, "epoch": 292, "lr": 0.001, "val_loss": 0.06175753474235535} +{"train_loss": 0.03911248967051506, "train_loss_bc": 0.03911248967051506, "train_loss_llm": 0.0, "grad_norm": 0.047035425901412964, "global_step": 10548, "epoch": 293, "lr": 0.001} +{"train_loss": 0.042705826461315155, "train_loss_bc": 0.042705826461315155, "train_loss_llm": 0.0, "grad_norm": 0.025336842983961105, "global_step": 10549, "epoch": 293, "lr": 0.001} +{"train_loss": 0.032298196107149124, "train_loss_bc": 0.032293640077114105, "train_loss_llm": 0.00045567750930786133, "grad_norm": 0.036561764776706696, "global_step": 10550, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03359123319387436, "train_loss_bc": 0.03359123319387436, "train_loss_llm": 0.0, "grad_norm": 0.05795282498002052, "global_step": 10551, "epoch": 293, "lr": 0.001} +{"train_loss": 0.039712827652692795, "train_loss_bc": 0.039712827652692795, "train_loss_llm": 0.0, "grad_norm": 0.07324539870023727, "global_step": 10552, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03023901954293251, "train_loss_bc": 0.03023901954293251, "train_loss_llm": 0.0, "grad_norm": 0.017607713118195534, "global_step": 10553, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03417053818702698, "train_loss_bc": 0.034141503274440765, "train_loss_llm": 0.0029033422470092773, "grad_norm": 0.027919014915823936, "global_step": 10554, "epoch": 293, "lr": 0.001} +{"train_loss": 0.024045530706644058, "train_loss_bc": 0.024020448327064514, "train_loss_llm": 0.002508193254470825, "grad_norm": 0.0374540351331234, "global_step": 10555, "epoch": 293, "lr": 0.001} +{"train_loss": 0.037624504417181015, "train_loss_bc": 0.037624504417181015, "train_loss_llm": 0.0, "grad_norm": 0.058574751019477844, "global_step": 10556, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03868982568383217, "train_loss_bc": 0.03868672251701355, "train_loss_llm": 0.0003101825714111328, "grad_norm": 0.025199417024850845, "global_step": 10557, "epoch": 293, "lr": 0.001} +{"train_loss": 0.04209882393479347, "train_loss_bc": 0.04209882393479347, "train_loss_llm": 0.0, "grad_norm": 0.036501046270132065, "global_step": 10558, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03796388581395149, "train_loss_bc": 0.03796388581395149, "train_loss_llm": 0.0, "grad_norm": 0.04731952026486397, "global_step": 10559, "epoch": 293, "lr": 0.001} +{"train_loss": 0.038984451442956924, "train_loss_bc": 0.038984451442956924, "train_loss_llm": 0.0, "grad_norm": 0.057998012751340866, "global_step": 10560, "epoch": 293, "lr": 0.001} +{"train_loss": 0.043572135269641876, "train_loss_bc": 0.043572135269641876, "train_loss_llm": 0.0, "grad_norm": 0.027621468529105186, "global_step": 10561, "epoch": 293, "lr": 0.001} +{"train_loss": 0.04352978244423866, "train_loss_bc": 0.04352978244423866, "train_loss_llm": 0.0, "grad_norm": 0.031739793717861176, "global_step": 10562, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03820228949189186, "train_loss_bc": 0.038188450038433075, "train_loss_llm": 0.001383960247039795, "grad_norm": 0.038505490869283676, "global_step": 10563, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03327939659357071, "train_loss_bc": 0.03327939659357071, "train_loss_llm": 0.0, "grad_norm": 0.04869715869426727, "global_step": 10564, "epoch": 293, "lr": 0.001} +{"train_loss": 0.04593655467033386, "train_loss_bc": 0.04593655467033386, "train_loss_llm": 0.0, "grad_norm": 0.024766815826296806, "global_step": 10565, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03915407881140709, "train_loss_bc": 0.039145149290561676, "train_loss_llm": 0.0008928775787353516, "grad_norm": 0.03217155486345291, "global_step": 10566, "epoch": 293, "lr": 0.001} +{"train_loss": 0.028040068224072456, "train_loss_bc": 0.028040068224072456, "train_loss_llm": 0.0, "grad_norm": 0.040746185928583145, "global_step": 10567, "epoch": 293, "lr": 0.001} +{"train_loss": 0.0425475537776947, "train_loss_bc": 0.0425475537776947, "train_loss_llm": 0.0, "grad_norm": 0.050743911415338516, "global_step": 10568, "epoch": 293, "lr": 0.001} +{"train_loss": 0.042710911482572556, "train_loss_bc": 0.042710911482572556, "train_loss_llm": 0.0, "grad_norm": 0.01901429332792759, "global_step": 10569, "epoch": 293, "lr": 0.001} +{"train_loss": 0.0375131294131279, "train_loss_bc": 0.0375131294131279, "train_loss_llm": 0.0, "grad_norm": 0.03193043917417526, "global_step": 10570, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03997156769037247, "train_loss_bc": 0.03997156769037247, "train_loss_llm": 0.0, "grad_norm": 0.03919442743062973, "global_step": 10571, "epoch": 293, "lr": 0.001} +{"train_loss": 0.036045823246240616, "train_loss_bc": 0.036045823246240616, "train_loss_llm": 0.0, "grad_norm": 0.04904678836464882, "global_step": 10572, "epoch": 293, "lr": 0.001} +{"train_loss": 0.044415395706892014, "train_loss_bc": 0.044410210102796555, "train_loss_llm": 0.000518500804901123, "grad_norm": 0.020193977281451225, "global_step": 10573, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03829871490597725, "train_loss_bc": 0.03829871490597725, "train_loss_llm": 0.0, "grad_norm": 0.03668985515832901, "global_step": 10574, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03553406894207001, "train_loss_bc": 0.03553406894207001, "train_loss_llm": 0.0, "grad_norm": 0.052333854138851166, "global_step": 10575, "epoch": 293, "lr": 0.001} +{"train_loss": 0.04204782098531723, "train_loss_bc": 0.04204782098531723, "train_loss_llm": 0.0, "grad_norm": 0.06470439583063126, "global_step": 10576, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03800445422530174, "train_loss_bc": 0.03800445422530174, "train_loss_llm": 0.0, "grad_norm": 0.01785428822040558, "global_step": 10577, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03531985357403755, "train_loss_bc": 0.03531578555703163, "train_loss_llm": 0.0004069805145263672, "grad_norm": 0.027156058698892593, "global_step": 10578, "epoch": 293, "lr": 0.001} +{"train_loss": 0.033887460827827454, "train_loss_bc": 0.03388744592666626, "train_loss_llm": 1.4603137969970703e-06, "grad_norm": 0.03586190938949585, "global_step": 10579, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03939208388328552, "train_loss_bc": 0.03939208388328552, "train_loss_llm": 0.0, "grad_norm": 0.04151761904358864, "global_step": 10580, "epoch": 293, "lr": 0.001} +{"train_loss": 0.04381746053695679, "train_loss_bc": 0.04381746053695679, "train_loss_llm": 0.0, "grad_norm": 0.02500508911907673, "global_step": 10581, "epoch": 293, "lr": 0.001} +{"train_loss": 0.037186529487371445, "train_loss_bc": 0.0371575802564621, "train_loss_llm": 0.0028948187828063965, "grad_norm": 0.034269146621227264, "global_step": 10582, "epoch": 293, "lr": 0.001} +{"train_loss": 0.03816902363258931, "train_loss_bc": 0.04443627968430519, "train_loss_llm": 0.0004284381866455078, "grad_norm": 0.04913056269288063, "global_step": 10583, "epoch": 293, "lr": 0.001, "val_loss": 0.0612405464053154} +{"train_loss": 0.03454696387052536, "train_loss_bc": 0.03454696387052536, "train_loss_llm": 0.0, "grad_norm": 0.05865471810102463, "global_step": 10584, "epoch": 294, "lr": 0.001} +{"train_loss": 0.04393716901540756, "train_loss_bc": 0.04393716901540756, "train_loss_llm": 0.0, "grad_norm": 0.019535431638360023, "global_step": 10585, "epoch": 294, "lr": 0.001} +{"train_loss": 0.042463887482881546, "train_loss_bc": 0.04244202375411987, "train_loss_llm": 0.002186387777328491, "grad_norm": 0.033876705914735794, "global_step": 10586, "epoch": 294, "lr": 0.001} +{"train_loss": 0.04035269096493721, "train_loss_bc": 0.04035269096493721, "train_loss_llm": 0.0, "grad_norm": 0.03885822370648384, "global_step": 10587, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03448549285531044, "train_loss_bc": 0.03448549285531044, "train_loss_llm": 0.0, "grad_norm": 0.03960287570953369, "global_step": 10588, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03323467820882797, "train_loss_bc": 0.03323467820882797, "train_loss_llm": 0.0, "grad_norm": 0.018205003812909126, "global_step": 10589, "epoch": 294, "lr": 0.001} +{"train_loss": 0.043389055877923965, "train_loss_bc": 0.04337938874959946, "train_loss_llm": 0.0009668171405792236, "grad_norm": 0.029429959133267403, "global_step": 10590, "epoch": 294, "lr": 0.001} +{"train_loss": 0.033264681696891785, "train_loss_bc": 0.033264681696891785, "train_loss_llm": 0.0, "grad_norm": 0.039788469672203064, "global_step": 10591, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03803686797618866, "train_loss_bc": 0.0380040667951107, "train_loss_llm": 0.00328025221824646, "grad_norm": 0.047365497797727585, "global_step": 10592, "epoch": 294, "lr": 0.001} +{"train_loss": 0.036462411284446716, "train_loss_bc": 0.036462411284446716, "train_loss_llm": 0.0, "grad_norm": 0.017102565616369247, "global_step": 10593, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03472237288951874, "train_loss_bc": 0.03472237288951874, "train_loss_llm": 0.0, "grad_norm": 0.027340199798345566, "global_step": 10594, "epoch": 294, "lr": 0.001} +{"train_loss": 0.031776025891304016, "train_loss_bc": 0.03175998479127884, "train_loss_llm": 0.0016040802001953125, "grad_norm": 0.03571214899420738, "global_step": 10595, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03774081915616989, "train_loss_bc": 0.03774081915616989, "train_loss_llm": 0.0, "grad_norm": 0.045243341475725174, "global_step": 10596, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03849949315190315, "train_loss_bc": 0.0384950265288353, "train_loss_llm": 0.0004467964172363281, "grad_norm": 0.022867275401949883, "global_step": 10597, "epoch": 294, "lr": 0.001} +{"train_loss": 0.040297362953424454, "train_loss_bc": 0.040297362953424454, "train_loss_llm": 0.0, "grad_norm": 0.04689624533057213, "global_step": 10598, "epoch": 294, "lr": 0.001} +{"train_loss": 0.0341024324297905, "train_loss_bc": 0.0341024324297905, "train_loss_llm": 0.0, "grad_norm": 0.05202178657054901, "global_step": 10599, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03157051280140877, "train_loss_bc": 0.03157051280140877, "train_loss_llm": 0.0, "grad_norm": 0.06650755554437637, "global_step": 10600, "epoch": 294, "lr": 0.001} +{"train_loss": 0.04217769578099251, "train_loss_bc": 0.04217769578099251, "train_loss_llm": 0.0, "grad_norm": 0.026325762271881104, "global_step": 10601, "epoch": 294, "lr": 0.001} +{"train_loss": 0.04145800694823265, "train_loss_bc": 0.04145761579275131, "train_loss_llm": 3.9011240005493164e-05, "grad_norm": 0.04685286805033684, "global_step": 10602, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03764566406607628, "train_loss_bc": 0.03763949126005173, "train_loss_llm": 0.0006173253059387207, "grad_norm": 0.051040761172771454, "global_step": 10603, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03847358003258705, "train_loss_bc": 0.03846989572048187, "train_loss_llm": 0.00036838650703430176, "grad_norm": 0.06031140685081482, "global_step": 10604, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03960062563419342, "train_loss_bc": 0.0395965538918972, "train_loss_llm": 0.00040709972381591797, "grad_norm": 0.020541269332170486, "global_step": 10605, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03500279039144516, "train_loss_bc": 0.03500279039144516, "train_loss_llm": 0.0, "grad_norm": 0.029674218967556953, "global_step": 10606, "epoch": 294, "lr": 0.001} +{"train_loss": 0.0369308665394783, "train_loss_bc": 0.03691843897104263, "train_loss_llm": 0.001242905855178833, "grad_norm": 0.040748342871665955, "global_step": 10607, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03315964713692665, "train_loss_bc": 0.03315964713692665, "train_loss_llm": 0.0, "grad_norm": 0.04835653677582741, "global_step": 10608, "epoch": 294, "lr": 0.001} +{"train_loss": 0.040952444076538086, "train_loss_bc": 0.040952444076538086, "train_loss_llm": 0.0, "grad_norm": 0.02283901534974575, "global_step": 10609, "epoch": 294, "lr": 0.001} +{"train_loss": 0.038193609565496445, "train_loss_bc": 0.038193609565496445, "train_loss_llm": 0.0, "grad_norm": 0.036507345736026764, "global_step": 10610, "epoch": 294, "lr": 0.001} +{"train_loss": 0.05174007639288902, "train_loss_bc": 0.05174007639288902, "train_loss_llm": 0.0, "grad_norm": 0.046354446560144424, "global_step": 10611, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03409641981124878, "train_loss_bc": 0.03409641981124878, "train_loss_llm": 0.0, "grad_norm": 0.054413147270679474, "global_step": 10612, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03880993649363518, "train_loss_bc": 0.03880993649363518, "train_loss_llm": 0.0, "grad_norm": 0.021092593669891357, "global_step": 10613, "epoch": 294, "lr": 0.001} +{"train_loss": 0.042161498218774796, "train_loss_bc": 0.042161498218774796, "train_loss_llm": 0.0, "grad_norm": 0.040004096925258636, "global_step": 10614, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03355565294623375, "train_loss_bc": 0.03355565294623375, "train_loss_llm": 0.0, "grad_norm": 0.05495607107877731, "global_step": 10615, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03705335408449173, "train_loss_bc": 0.03704516217112541, "train_loss_llm": 0.0008191168308258057, "grad_norm": 0.06447553634643555, "global_step": 10616, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03735034167766571, "train_loss_bc": 0.03735034167766571, "train_loss_llm": 0.0, "grad_norm": 0.02285817824304104, "global_step": 10617, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03912730515003204, "train_loss_bc": 0.03912730515003204, "train_loss_llm": 0.0, "grad_norm": 0.03423862159252167, "global_step": 10618, "epoch": 294, "lr": 0.001} +{"train_loss": 0.03805090507699384, "train_loss_bc": 0.04346014931797981, "train_loss_llm": 0.0, "grad_norm": 0.048547226935625076, "global_step": 10619, "epoch": 294, "lr": 0.001, "val_loss": 0.06215281039476395} +{"train_loss": 0.04080163687467575, "train_loss_bc": 0.04080163687467575, "train_loss_llm": 0.0, "grad_norm": 0.05929434671998024, "global_step": 10620, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03799201920628548, "train_loss_bc": 0.03799201920628548, "train_loss_llm": 0.0, "grad_norm": 0.0196546483784914, "global_step": 10621, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03634827584028244, "train_loss_bc": 0.03633790835738182, "train_loss_llm": 0.0010366439819335938, "grad_norm": 0.026926258578896523, "global_step": 10622, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03419427573680878, "train_loss_bc": 0.03419427573680878, "train_loss_llm": 0.0, "grad_norm": 0.03487103059887886, "global_step": 10623, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03606337681412697, "train_loss_bc": 0.03606337681412697, "train_loss_llm": 0.0, "grad_norm": 0.041703201830387115, "global_step": 10624, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03976035863161087, "train_loss_bc": 0.03975066542625427, "train_loss_llm": 0.0009692907333374023, "grad_norm": 0.02681787498295307, "global_step": 10625, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03870908543467522, "train_loss_bc": 0.03870908543467522, "train_loss_llm": 0.0, "grad_norm": 0.03649475425481796, "global_step": 10626, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03988383337855339, "train_loss_bc": 0.03988383337855339, "train_loss_llm": 0.0, "grad_norm": 0.0464341938495636, "global_step": 10627, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03507629409432411, "train_loss_bc": 0.03507629409432411, "train_loss_llm": 0.0, "grad_norm": 0.05368758365511894, "global_step": 10628, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03351951017975807, "train_loss_bc": 0.033511798828840256, "train_loss_llm": 0.0007710456848144531, "grad_norm": 0.022821780294179916, "global_step": 10629, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03573334962129593, "train_loss_bc": 0.035732708871364594, "train_loss_llm": 6.42538070678711e-05, "grad_norm": 0.03127806633710861, "global_step": 10630, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03301874175667763, "train_loss_bc": 0.03301874175667763, "train_loss_llm": 0.0, "grad_norm": 0.03906247764825821, "global_step": 10631, "epoch": 295, "lr": 0.001} +{"train_loss": 0.0330459401011467, "train_loss_bc": 0.0330459401011467, "train_loss_llm": 0.0, "grad_norm": 0.054597947746515274, "global_step": 10632, "epoch": 295, "lr": 0.001} +{"train_loss": 0.04114292562007904, "train_loss_bc": 0.04114292562007904, "train_loss_llm": 0.0, "grad_norm": 0.02346232160925865, "global_step": 10633, "epoch": 295, "lr": 0.001} +{"train_loss": 0.038428716361522675, "train_loss_bc": 0.038377635180950165, "train_loss_llm": 0.005107969045639038, "grad_norm": 0.03540021553635597, "global_step": 10634, "epoch": 295, "lr": 0.001} +{"train_loss": 0.039111439138650894, "train_loss_bc": 0.039111439138650894, "train_loss_llm": 0.0, "grad_norm": 0.043376374989748, "global_step": 10635, "epoch": 295, "lr": 0.001} +{"train_loss": 0.04180673882365227, "train_loss_bc": 0.04180673882365227, "train_loss_llm": 0.0, "grad_norm": 0.05262821540236473, "global_step": 10636, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03078284114599228, "train_loss_bc": 0.030749782919883728, "train_loss_llm": 0.0033059120178222656, "grad_norm": 0.018119771033525467, "global_step": 10637, "epoch": 295, "lr": 0.001} +{"train_loss": 0.031691841781139374, "train_loss_bc": 0.031686101108789444, "train_loss_llm": 0.0005740821361541748, "grad_norm": 0.030595283955335617, "global_step": 10638, "epoch": 295, "lr": 0.001} +{"train_loss": 0.039543267339468, "train_loss_bc": 0.03952779993414879, "train_loss_llm": 0.0015468001365661621, "grad_norm": 0.04088994115591049, "global_step": 10639, "epoch": 295, "lr": 0.001} +{"train_loss": 0.041338153183460236, "train_loss_bc": 0.041338153183460236, "train_loss_llm": 0.0, "grad_norm": 0.051604725420475006, "global_step": 10640, "epoch": 295, "lr": 0.001} +{"train_loss": 0.0412532277405262, "train_loss_bc": 0.0412532277405262, "train_loss_llm": 0.0, "grad_norm": 0.025092026218771935, "global_step": 10641, "epoch": 295, "lr": 0.001} +{"train_loss": 0.039319220930337906, "train_loss_bc": 0.039319220930337906, "train_loss_llm": 0.0, "grad_norm": 0.03368331119418144, "global_step": 10642, "epoch": 295, "lr": 0.001} +{"train_loss": 0.037046752870082855, "train_loss_bc": 0.037046752870082855, "train_loss_llm": 0.0, "grad_norm": 0.04384995251893997, "global_step": 10643, "epoch": 295, "lr": 0.001} +{"train_loss": 0.051594432443380356, "train_loss_bc": 0.051594432443380356, "train_loss_llm": 0.0, "grad_norm": 0.05899563431739807, "global_step": 10644, "epoch": 295, "lr": 0.001} +{"train_loss": 0.04276694357395172, "train_loss_bc": 0.04276694357395172, "train_loss_llm": 0.0, "grad_norm": 0.0196987371891737, "global_step": 10645, "epoch": 295, "lr": 0.001} +{"train_loss": 0.028702033683657646, "train_loss_bc": 0.028702033683657646, "train_loss_llm": 0.0, "grad_norm": 0.031596776098012924, "global_step": 10646, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03792121633887291, "train_loss_bc": 0.03792121633887291, "train_loss_llm": 0.0, "grad_norm": 0.036727484315633774, "global_step": 10647, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03779644891619682, "train_loss_bc": 0.03779644891619682, "train_loss_llm": 0.0, "grad_norm": 0.047388531267642975, "global_step": 10648, "epoch": 295, "lr": 0.001} +{"train_loss": 0.035961396992206573, "train_loss_bc": 0.03596004471182823, "train_loss_llm": 0.00013533234596252441, "grad_norm": 0.023022588342428207, "global_step": 10649, "epoch": 295, "lr": 0.001} +{"train_loss": 0.046865951269865036, "train_loss_bc": 0.046865951269865036, "train_loss_llm": 0.0, "grad_norm": 0.03425350785255432, "global_step": 10650, "epoch": 295, "lr": 0.001} +{"train_loss": 0.05065607652068138, "train_loss_bc": 0.05065607652068138, "train_loss_llm": 0.0, "grad_norm": 0.047695573419332504, "global_step": 10651, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03340671956539154, "train_loss_bc": 0.03340671956539154, "train_loss_llm": 0.0, "grad_norm": 0.05817381292581558, "global_step": 10652, "epoch": 295, "lr": 0.001} +{"train_loss": 0.034551121294498444, "train_loss_bc": 0.034551121294498444, "train_loss_llm": 0.0, "grad_norm": 0.019498972222208977, "global_step": 10653, "epoch": 295, "lr": 0.001} +{"train_loss": 0.03888421133160591, "train_loss_bc": 0.03888421133160591, "train_loss_llm": 0.0, "grad_norm": 0.03258794546127319, "global_step": 10654, "epoch": 295, "lr": 0.001} +{"train_loss": 0.037954388238075704, "train_loss_bc": 0.03163960203528404, "train_loss_llm": 0.0, "grad_norm": 0.053285252302885056, "global_step": 10655, "epoch": 295, "lr": 0.001, "train/cumulative_reward": 46.82476090421344, "train/mean_score": 9.6283704211647, "train/success_rate": 0.2, "test/cumulative_reward": 35.02530734770227, "test/mean_score": 3.991289910721619, "test/success_rate": 0.08, "val_loss": 0.06245996057987213, "train_action_mse_error": 0.041842810809612274} +{"train_loss": 0.04491223767399788, "train_loss_bc": 0.04491223767399788, "train_loss_llm": 0.0, "grad_norm": 0.07063144445419312, "global_step": 10656, "epoch": 296, "lr": 0.001} +{"train_loss": 0.039618734270334244, "train_loss_bc": 0.039605818688869476, "train_loss_llm": 0.0012916922569274902, "grad_norm": 0.020133014768362045, "global_step": 10657, "epoch": 296, "lr": 0.001} +{"train_loss": 0.040226977318525314, "train_loss_bc": 0.040226977318525314, "train_loss_llm": 0.0, "grad_norm": 0.027260735630989075, "global_step": 10658, "epoch": 296, "lr": 0.001} +{"train_loss": 0.04052180051803589, "train_loss_bc": 0.04052180051803589, "train_loss_llm": 0.0, "grad_norm": 0.039087437093257904, "global_step": 10659, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03580136224627495, "train_loss_bc": 0.03580136224627495, "train_loss_llm": 0.0, "grad_norm": 0.04052969813346863, "global_step": 10660, "epoch": 296, "lr": 0.001} +{"train_loss": 0.04566951096057892, "train_loss_bc": 0.04566951096057892, "train_loss_llm": 0.0, "grad_norm": 0.02213376946747303, "global_step": 10661, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03653580695390701, "train_loss_bc": 0.03653580695390701, "train_loss_llm": 0.0, "grad_norm": 0.03741215169429779, "global_step": 10662, "epoch": 296, "lr": 0.001} +{"train_loss": 0.0354684554040432, "train_loss_bc": 0.0354684554040432, "train_loss_llm": 0.0, "grad_norm": 0.043882086873054504, "global_step": 10663, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03906162083148956, "train_loss_bc": 0.03906162083148956, "train_loss_llm": 0.0, "grad_norm": 0.047425273805856705, "global_step": 10664, "epoch": 296, "lr": 0.001} +{"train_loss": 0.02993939258158207, "train_loss_bc": 0.02993939258158207, "train_loss_llm": 0.0, "grad_norm": 0.018145401030778885, "global_step": 10665, "epoch": 296, "lr": 0.001} +{"train_loss": 0.041835684329271317, "train_loss_bc": 0.041835684329271317, "train_loss_llm": 0.0, "grad_norm": 0.029429351910948753, "global_step": 10666, "epoch": 296, "lr": 0.001} +{"train_loss": 0.04115966334939003, "train_loss_bc": 0.0411330983042717, "train_loss_llm": 0.0026563405990600586, "grad_norm": 0.049533553421497345, "global_step": 10667, "epoch": 296, "lr": 0.001} +{"train_loss": 0.034654662013053894, "train_loss_bc": 0.03465097025036812, "train_loss_llm": 0.00036910176277160645, "grad_norm": 0.053344085812568665, "global_step": 10668, "epoch": 296, "lr": 0.001} +{"train_loss": 0.038299765437841415, "train_loss_bc": 0.03829248994588852, "train_loss_llm": 0.0007275044918060303, "grad_norm": 0.020734891295433044, "global_step": 10669, "epoch": 296, "lr": 0.001} +{"train_loss": 0.040988754481077194, "train_loss_bc": 0.04098356142640114, "train_loss_llm": 0.0005193650722503662, "grad_norm": 0.03586909547448158, "global_step": 10670, "epoch": 296, "lr": 0.001} +{"train_loss": 0.037576328963041306, "train_loss_bc": 0.037576328963041306, "train_loss_llm": 0.0, "grad_norm": 0.04235072806477547, "global_step": 10671, "epoch": 296, "lr": 0.001} +{"train_loss": 0.040756016969680786, "train_loss_bc": 0.040756016969680786, "train_loss_llm": 0.0, "grad_norm": 0.05932490900158882, "global_step": 10672, "epoch": 296, "lr": 0.001} +{"train_loss": 0.0366809107363224, "train_loss_bc": 0.03667094185948372, "train_loss_llm": 0.0009968280792236328, "grad_norm": 0.019125469028949738, "global_step": 10673, "epoch": 296, "lr": 0.001} +{"train_loss": 0.044448766857385635, "train_loss_bc": 0.044448766857385635, "train_loss_llm": 0.0, "grad_norm": 0.030531074851751328, "global_step": 10674, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03337261825799942, "train_loss_bc": 0.033361297100782394, "train_loss_llm": 0.001132190227508545, "grad_norm": 0.04302207753062248, "global_step": 10675, "epoch": 296, "lr": 0.001} +{"train_loss": 0.037463556975126266, "train_loss_bc": 0.037463556975126266, "train_loss_llm": 0.0, "grad_norm": 0.054832588881254196, "global_step": 10676, "epoch": 296, "lr": 0.001} +{"train_loss": 0.030785376206040382, "train_loss_bc": 0.030785376206040382, "train_loss_llm": 0.0, "grad_norm": 0.020247476175427437, "global_step": 10677, "epoch": 296, "lr": 0.001} +{"train_loss": 0.044116947799921036, "train_loss_bc": 0.044114869087934494, "train_loss_llm": 0.00020769238471984863, "grad_norm": 0.033263809978961945, "global_step": 10678, "epoch": 296, "lr": 0.001} +{"train_loss": 0.039146050810813904, "train_loss_bc": 0.039146050810813904, "train_loss_llm": 0.0, "grad_norm": 0.04656071588397026, "global_step": 10679, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03229638561606407, "train_loss_bc": 0.03229481354355812, "train_loss_llm": 0.00015717744827270508, "grad_norm": 0.059737373143434525, "global_step": 10680, "epoch": 296, "lr": 0.001} +{"train_loss": 0.033446457237005234, "train_loss_bc": 0.03343269228935242, "train_loss_llm": 0.0013765394687652588, "grad_norm": 0.01960201933979988, "global_step": 10681, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03855461627244949, "train_loss_bc": 0.03855461627244949, "train_loss_llm": 0.0, "grad_norm": 0.03400116413831711, "global_step": 10682, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03077377751469612, "train_loss_bc": 0.030772201716899872, "train_loss_llm": 0.00015753507614135742, "grad_norm": 0.04534480720758438, "global_step": 10683, "epoch": 296, "lr": 0.001} +{"train_loss": 0.04085972160100937, "train_loss_bc": 0.04085972160100937, "train_loss_llm": 0.0, "grad_norm": 0.05622788518667221, "global_step": 10684, "epoch": 296, "lr": 0.001} +{"train_loss": 0.04279845952987671, "train_loss_bc": 0.04279845952987671, "train_loss_llm": 0.0, "grad_norm": 0.027817413210868835, "global_step": 10685, "epoch": 296, "lr": 0.001} +{"train_loss": 0.027549361810088158, "train_loss_bc": 0.02754157781600952, "train_loss_llm": 0.0007784068584442139, "grad_norm": 0.03472505137324333, "global_step": 10686, "epoch": 296, "lr": 0.001} +{"train_loss": 0.04441646859049797, "train_loss_bc": 0.04441646859049797, "train_loss_llm": 0.0, "grad_norm": 0.04355667531490326, "global_step": 10687, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03736279532313347, "train_loss_bc": 0.03736279532313347, "train_loss_llm": 0.0, "grad_norm": 0.056074872612953186, "global_step": 10688, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03951720520853996, "train_loss_bc": 0.03951720520853996, "train_loss_llm": 0.0, "grad_norm": 0.02173009142279625, "global_step": 10689, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03475810959935188, "train_loss_bc": 0.03475810959935188, "train_loss_llm": 0.0, "grad_norm": 0.03233851492404938, "global_step": 10690, "epoch": 296, "lr": 0.001} +{"train_loss": 0.03804694762867358, "train_loss_bc": 0.038315754383802414, "train_loss_llm": 0.0, "grad_norm": 0.0447973757982254, "global_step": 10691, "epoch": 296, "lr": 0.001, "val_loss": 0.061710041016340256} +{"train_loss": 0.038062337785959244, "train_loss_bc": 0.038062337785959244, "train_loss_llm": 0.0, "grad_norm": 0.05744126811623573, "global_step": 10692, "epoch": 297, "lr": 0.001} +{"train_loss": 0.04005592316389084, "train_loss_bc": 0.04005592316389084, "train_loss_llm": 0.0, "grad_norm": 0.020389502868056297, "global_step": 10693, "epoch": 297, "lr": 0.001} +{"train_loss": 0.043198101222515106, "train_loss_bc": 0.043198101222515106, "train_loss_llm": 0.0, "grad_norm": 0.029782189056277275, "global_step": 10694, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03254024311900139, "train_loss_bc": 0.03254024311900139, "train_loss_llm": 0.0, "grad_norm": 0.035753991454839706, "global_step": 10695, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03387296572327614, "train_loss_bc": 0.03386137634515762, "train_loss_llm": 0.0011588335037231445, "grad_norm": 0.04424627497792244, "global_step": 10696, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03782012313604355, "train_loss_bc": 0.03780730068683624, "train_loss_llm": 0.0012822449207305908, "grad_norm": 0.016355346888303757, "global_step": 10697, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03964477404952049, "train_loss_bc": 0.03964477404952049, "train_loss_llm": 0.0, "grad_norm": 0.023683886975049973, "global_step": 10698, "epoch": 297, "lr": 0.001} +{"train_loss": 0.0416409857571125, "train_loss_bc": 0.041639115661382675, "train_loss_llm": 0.00018715858459472656, "grad_norm": 0.03411254659295082, "global_step": 10699, "epoch": 297, "lr": 0.001} +{"train_loss": 0.033065930008888245, "train_loss_bc": 0.033065930008888245, "train_loss_llm": 0.0, "grad_norm": 0.04869049787521362, "global_step": 10700, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03581753373146057, "train_loss_bc": 0.035806622356176376, "train_loss_llm": 0.0010912418365478516, "grad_norm": 0.020782126113772392, "global_step": 10701, "epoch": 297, "lr": 0.001} +{"train_loss": 0.036978136748075485, "train_loss_bc": 0.036924317479133606, "train_loss_llm": 0.0053820013999938965, "grad_norm": 0.03033612295985222, "global_step": 10702, "epoch": 297, "lr": 0.001} +{"train_loss": 0.028130995109677315, "train_loss_bc": 0.028130995109677315, "train_loss_llm": 0.0, "grad_norm": 0.03769366070628166, "global_step": 10703, "epoch": 297, "lr": 0.001} +{"train_loss": 0.04066124185919762, "train_loss_bc": 0.04066124185919762, "train_loss_llm": 0.0, "grad_norm": 0.0519273616373539, "global_step": 10704, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03766988590359688, "train_loss_bc": 0.03764762729406357, "train_loss_llm": 0.0022258758544921875, "grad_norm": 0.018022431060671806, "global_step": 10705, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03981418535113335, "train_loss_bc": 0.03981418535113335, "train_loss_llm": 0.0, "grad_norm": 0.03175828605890274, "global_step": 10706, "epoch": 297, "lr": 0.001} +{"train_loss": 0.033498167991638184, "train_loss_bc": 0.033498167991638184, "train_loss_llm": 0.0, "grad_norm": 0.03407851979136467, "global_step": 10707, "epoch": 297, "lr": 0.001} +{"train_loss": 0.0406450517475605, "train_loss_bc": 0.040623608976602554, "train_loss_llm": 0.002144336700439453, "grad_norm": 0.048569321632385254, "global_step": 10708, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03638765960931778, "train_loss_bc": 0.03638765960931778, "train_loss_llm": 0.0, "grad_norm": 0.01845420151948929, "global_step": 10709, "epoch": 297, "lr": 0.001} +{"train_loss": 0.04509753733873367, "train_loss_bc": 0.04509753733873367, "train_loss_llm": 0.0, "grad_norm": 0.02560591511428356, "global_step": 10710, "epoch": 297, "lr": 0.001} +{"train_loss": 0.034634049981832504, "train_loss_bc": 0.034634049981832504, "train_loss_llm": 0.0, "grad_norm": 0.04161318019032478, "global_step": 10711, "epoch": 297, "lr": 0.001} +{"train_loss": 0.04275038093328476, "train_loss_bc": 0.04275038093328476, "train_loss_llm": 0.0, "grad_norm": 0.05192119628190994, "global_step": 10712, "epoch": 297, "lr": 0.001} +{"train_loss": 0.02718493342399597, "train_loss_bc": 0.02718493342399597, "train_loss_llm": 0.0, "grad_norm": 0.015378558076918125, "global_step": 10713, "epoch": 297, "lr": 0.001} +{"train_loss": 0.030233876779675484, "train_loss_bc": 0.030233876779675484, "train_loss_llm": 0.0, "grad_norm": 0.0330548994243145, "global_step": 10714, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03505437821149826, "train_loss_bc": 0.03505437821149826, "train_loss_llm": 0.0, "grad_norm": 0.03871060535311699, "global_step": 10715, "epoch": 297, "lr": 0.001} +{"train_loss": 0.040443532168865204, "train_loss_bc": 0.040443532168865204, "train_loss_llm": 0.0, "grad_norm": 0.05534517765045166, "global_step": 10716, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03950580582022667, "train_loss_bc": 0.03950580582022667, "train_loss_llm": 0.0, "grad_norm": 0.02264809049665928, "global_step": 10717, "epoch": 297, "lr": 0.001} +{"train_loss": 0.04371247813105583, "train_loss_bc": 0.04371247813105583, "train_loss_llm": 0.0, "grad_norm": 0.038653962314128876, "global_step": 10718, "epoch": 297, "lr": 0.001} +{"train_loss": 0.042051348835229874, "train_loss_bc": 0.042051348835229874, "train_loss_llm": 0.0, "grad_norm": 0.048487599939107895, "global_step": 10719, "epoch": 297, "lr": 0.001} +{"train_loss": 0.0351293608546257, "train_loss_bc": 0.0351293608546257, "train_loss_llm": 0.0, "grad_norm": 0.05718081817030907, "global_step": 10720, "epoch": 297, "lr": 0.001} +{"train_loss": 0.041958630084991455, "train_loss_bc": 0.041958630084991455, "train_loss_llm": 0.0, "grad_norm": 0.020380984991788864, "global_step": 10721, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03567374870181084, "train_loss_bc": 0.03567374870181084, "train_loss_llm": 0.0, "grad_norm": 0.03671947121620178, "global_step": 10722, "epoch": 297, "lr": 0.001} +{"train_loss": 0.042293209582567215, "train_loss_bc": 0.042293209582567215, "train_loss_llm": 0.0, "grad_norm": 0.044496357440948486, "global_step": 10723, "epoch": 297, "lr": 0.001} +{"train_loss": 0.041672296822071075, "train_loss_bc": 0.041672296822071075, "train_loss_llm": 0.0, "grad_norm": 0.049491386860609055, "global_step": 10724, "epoch": 297, "lr": 0.001} +{"train_loss": 0.0425429530441761, "train_loss_bc": 0.042540598660707474, "train_loss_llm": 0.00023546814918518066, "grad_norm": 0.022656336426734924, "global_step": 10725, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03901940956711769, "train_loss_bc": 0.03901940956711769, "train_loss_llm": 0.0, "grad_norm": 0.03170502185821533, "global_step": 10726, "epoch": 297, "lr": 0.001} +{"train_loss": 0.03799104090366098, "train_loss_bc": 0.03920239955186844, "train_loss_llm": 0.0012901723384857178, "grad_norm": 0.049378737807273865, "global_step": 10727, "epoch": 297, "lr": 0.001, "val_loss": 0.06198284775018692} +{"train_loss": 0.031051699072122574, "train_loss_bc": 0.031051699072122574, "train_loss_llm": 0.0, "grad_norm": 0.0616900660097599, "global_step": 10728, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03438396751880646, "train_loss_bc": 0.034368645399808884, "train_loss_llm": 0.0015323758125305176, "grad_norm": 0.022061090916395187, "global_step": 10729, "epoch": 298, "lr": 0.001} +{"train_loss": 0.034640468657016754, "train_loss_bc": 0.034640468657016754, "train_loss_llm": 0.0, "grad_norm": 0.02901608683168888, "global_step": 10730, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03779046609997749, "train_loss_bc": 0.03779046609997749, "train_loss_llm": 0.0, "grad_norm": 0.040837910026311874, "global_step": 10731, "epoch": 298, "lr": 0.001} +{"train_loss": 0.037443608045578, "train_loss_bc": 0.037443608045578, "train_loss_llm": 0.0, "grad_norm": 0.050690747797489166, "global_step": 10732, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03652169927954674, "train_loss_bc": 0.03652169927954674, "train_loss_llm": 0.0, "grad_norm": 0.02393120527267456, "global_step": 10733, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03850363567471504, "train_loss_bc": 0.03850363567471504, "train_loss_llm": 0.0, "grad_norm": 0.035766083747148514, "global_step": 10734, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03860144689679146, "train_loss_bc": 0.03859155252575874, "train_loss_llm": 0.000989288091659546, "grad_norm": 0.043512649834156036, "global_step": 10735, "epoch": 298, "lr": 0.001} +{"train_loss": 0.04372190684080124, "train_loss_bc": 0.04372190684080124, "train_loss_llm": 0.0, "grad_norm": 0.05534107983112335, "global_step": 10736, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03589291870594025, "train_loss_bc": 0.03589291870594025, "train_loss_llm": 0.0, "grad_norm": 0.020178204402327538, "global_step": 10737, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03825084865093231, "train_loss_bc": 0.03825084865093231, "train_loss_llm": 0.0, "grad_norm": 0.037367384880781174, "global_step": 10738, "epoch": 298, "lr": 0.001} +{"train_loss": 0.044016048312187195, "train_loss_bc": 0.044016048312187195, "train_loss_llm": 0.0, "grad_norm": 0.0540430024266243, "global_step": 10739, "epoch": 298, "lr": 0.001} +{"train_loss": 0.037865713238716125, "train_loss_bc": 0.037865713238716125, "train_loss_llm": 0.0, "grad_norm": 0.06539331376552582, "global_step": 10740, "epoch": 298, "lr": 0.001} +{"train_loss": 0.037892527878284454, "train_loss_bc": 0.037892527878284454, "train_loss_llm": 0.0, "grad_norm": 0.018034519627690315, "global_step": 10741, "epoch": 298, "lr": 0.001} +{"train_loss": 0.031870387494564056, "train_loss_bc": 0.031870387494564056, "train_loss_llm": 0.0, "grad_norm": 0.028590889647603035, "global_step": 10742, "epoch": 298, "lr": 0.001} +{"train_loss": 0.0407748706638813, "train_loss_bc": 0.0407748706638813, "train_loss_llm": 0.0, "grad_norm": 0.03990812599658966, "global_step": 10743, "epoch": 298, "lr": 0.001} +{"train_loss": 0.0381767675280571, "train_loss_bc": 0.03815833106637001, "train_loss_llm": 0.0018438100814819336, "grad_norm": 0.05295516178011894, "global_step": 10744, "epoch": 298, "lr": 0.001} +{"train_loss": 0.04195932671427727, "train_loss_bc": 0.041946444660425186, "train_loss_llm": 0.0012882351875305176, "grad_norm": 0.019980909302830696, "global_step": 10745, "epoch": 298, "lr": 0.001} +{"train_loss": 0.04056891053915024, "train_loss_bc": 0.04056891053915024, "train_loss_llm": 0.0, "grad_norm": 0.031870171427726746, "global_step": 10746, "epoch": 298, "lr": 0.001} +{"train_loss": 0.04019869491457939, "train_loss_bc": 0.04019790515303612, "train_loss_llm": 7.900595664978027e-05, "grad_norm": 0.03891034424304962, "global_step": 10747, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03743860870599747, "train_loss_bc": 0.03743860870599747, "train_loss_llm": 0.0, "grad_norm": 0.04467710480093956, "global_step": 10748, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03942490369081497, "train_loss_bc": 0.03942490369081497, "train_loss_llm": 0.0, "grad_norm": 0.02250584401190281, "global_step": 10749, "epoch": 298, "lr": 0.001} +{"train_loss": 0.0419577918946743, "train_loss_bc": 0.04195359721779823, "train_loss_llm": 0.0004194676876068115, "grad_norm": 0.035326045006513596, "global_step": 10750, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03467566519975662, "train_loss_bc": 0.03467134013772011, "train_loss_llm": 0.00043258070945739746, "grad_norm": 0.04220057651400566, "global_step": 10751, "epoch": 298, "lr": 0.001} +{"train_loss": 0.0322716161608696, "train_loss_bc": 0.0322716161608696, "train_loss_llm": 0.0, "grad_norm": 0.054324839264154434, "global_step": 10752, "epoch": 298, "lr": 0.001} +{"train_loss": 0.035711146891117096, "train_loss_bc": 0.035711146891117096, "train_loss_llm": 0.0, "grad_norm": 0.024676019325852394, "global_step": 10753, "epoch": 298, "lr": 0.001} +{"train_loss": 0.033879004418849945, "train_loss_bc": 0.033873431384563446, "train_loss_llm": 0.000557243824005127, "grad_norm": 0.03840416669845581, "global_step": 10754, "epoch": 298, "lr": 0.001} +{"train_loss": 0.04593558609485626, "train_loss_bc": 0.04593360796570778, "train_loss_llm": 0.00019788742065429688, "grad_norm": 0.046918004751205444, "global_step": 10755, "epoch": 298, "lr": 0.001} +{"train_loss": 0.04353290796279907, "train_loss_bc": 0.04353290796279907, "train_loss_llm": 0.0, "grad_norm": 0.05939564108848572, "global_step": 10756, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03988533839583397, "train_loss_bc": 0.03988533839583397, "train_loss_llm": 0.0, "grad_norm": 0.020921198651194572, "global_step": 10757, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03761789947748184, "train_loss_bc": 0.03761789947748184, "train_loss_llm": 0.0, "grad_norm": 0.04106169193983078, "global_step": 10758, "epoch": 298, "lr": 0.001} +{"train_loss": 0.04049782082438469, "train_loss_bc": 0.04049782082438469, "train_loss_llm": 0.0, "grad_norm": 0.05905183404684067, "global_step": 10759, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03191717714071274, "train_loss_bc": 0.03191717714071274, "train_loss_llm": 0.0, "grad_norm": 0.06366067379713058, "global_step": 10760, "epoch": 298, "lr": 0.001} +{"train_loss": 0.031674981117248535, "train_loss_bc": 0.031674981117248535, "train_loss_llm": 0.0, "grad_norm": 0.024262672290205956, "global_step": 10761, "epoch": 298, "lr": 0.001} +{"train_loss": 0.0415990948677063, "train_loss_bc": 0.04159565269947052, "train_loss_llm": 0.0003440976142883301, "grad_norm": 0.0354543961584568, "global_step": 10762, "epoch": 298, "lr": 0.001} +{"train_loss": 0.03810113399393029, "train_loss_bc": 0.04344910383224487, "train_loss_llm": 0.004626601934432983, "grad_norm": 0.04982266575098038, "global_step": 10763, "epoch": 298, "lr": 0.001, "val_loss": 0.061825625598430634} +{"train_loss": 0.04255690798163414, "train_loss_bc": 0.04255690798163414, "train_loss_llm": 0.0, "grad_norm": 0.06662946194410324, "global_step": 10764, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03496496379375458, "train_loss_bc": 0.03494781255722046, "train_loss_llm": 0.0017151236534118652, "grad_norm": 0.01764516532421112, "global_step": 10765, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03487229347229004, "train_loss_bc": 0.03487229347229004, "train_loss_llm": 0.0, "grad_norm": 0.0231992956250906, "global_step": 10766, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03297978639602661, "train_loss_bc": 0.03297232463955879, "train_loss_llm": 0.0007462203502655029, "grad_norm": 0.02902134321630001, "global_step": 10767, "epoch": 299, "lr": 0.001} +{"train_loss": 0.04010969400405884, "train_loss_bc": 0.04010786488652229, "train_loss_llm": 0.00018292665481567383, "grad_norm": 0.040430568158626556, "global_step": 10768, "epoch": 299, "lr": 0.001} +{"train_loss": 0.035719793289899826, "train_loss_bc": 0.035719793289899826, "train_loss_llm": 0.0, "grad_norm": 0.016103411093354225, "global_step": 10769, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03158756345510483, "train_loss_bc": 0.03158756345510483, "train_loss_llm": 0.0, "grad_norm": 0.02300191856920719, "global_step": 10770, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03448387235403061, "train_loss_bc": 0.03448387235403061, "train_loss_llm": 0.0, "grad_norm": 0.032172881066799164, "global_step": 10771, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03679428622126579, "train_loss_bc": 0.03679428622126579, "train_loss_llm": 0.0, "grad_norm": 0.03749033436179161, "global_step": 10772, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03702440857887268, "train_loss_bc": 0.03702440857887268, "train_loss_llm": 0.0, "grad_norm": 0.03060660883784294, "global_step": 10773, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03590376675128937, "train_loss_bc": 0.03590376675128937, "train_loss_llm": 0.0, "grad_norm": 0.04464155435562134, "global_step": 10774, "epoch": 299, "lr": 0.001} +{"train_loss": 0.0395188145339489, "train_loss_bc": 0.0395188145339489, "train_loss_llm": 0.0, "grad_norm": 0.04549329727888107, "global_step": 10775, "epoch": 299, "lr": 0.001} +{"train_loss": 0.041926778852939606, "train_loss_bc": 0.041926778852939606, "train_loss_llm": 0.0, "grad_norm": 0.051059313118457794, "global_step": 10776, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03892572969198227, "train_loss_bc": 0.03892572969198227, "train_loss_llm": 0.0, "grad_norm": 0.030678031966090202, "global_step": 10777, "epoch": 299, "lr": 0.001} +{"train_loss": 0.04279562085866928, "train_loss_bc": 0.04279562085866928, "train_loss_llm": 0.0, "grad_norm": 0.05117471143603325, "global_step": 10778, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03544514626264572, "train_loss_bc": 0.03544514626264572, "train_loss_llm": 0.0, "grad_norm": 0.0677742138504982, "global_step": 10779, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03639328107237816, "train_loss_bc": 0.03637158125638962, "train_loss_llm": 0.0021701455116271973, "grad_norm": 0.08033596724271774, "global_step": 10780, "epoch": 299, "lr": 0.001} +{"train_loss": 0.04547959566116333, "train_loss_bc": 0.04547959566116333, "train_loss_llm": 0.0, "grad_norm": 0.02606778033077717, "global_step": 10781, "epoch": 299, "lr": 0.001} +{"train_loss": 0.04235485568642616, "train_loss_bc": 0.04235485568642616, "train_loss_llm": 0.0, "grad_norm": 0.03531214967370033, "global_step": 10782, "epoch": 299, "lr": 0.001} +{"train_loss": 0.034803591668605804, "train_loss_bc": 0.034803591668605804, "train_loss_llm": 0.0, "grad_norm": 0.046231091022491455, "global_step": 10783, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03368854522705078, "train_loss_bc": 0.03368854522705078, "train_loss_llm": 0.0, "grad_norm": 0.05870870128273964, "global_step": 10784, "epoch": 299, "lr": 0.001} +{"train_loss": 0.045819204300642014, "train_loss_bc": 0.045819204300642014, "train_loss_llm": 0.0, "grad_norm": 0.020117951557040215, "global_step": 10785, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03161603957414627, "train_loss_bc": 0.03161603957414627, "train_loss_llm": 0.0, "grad_norm": 0.02556796930730343, "global_step": 10786, "epoch": 299, "lr": 0.001} +{"train_loss": 0.04328106343746185, "train_loss_bc": 0.04328106343746185, "train_loss_llm": 0.0, "grad_norm": 0.03956015780568123, "global_step": 10787, "epoch": 299, "lr": 0.001} +{"train_loss": 0.035267915576696396, "train_loss_bc": 0.035267915576696396, "train_loss_llm": 0.0, "grad_norm": 0.05059688910841942, "global_step": 10788, "epoch": 299, "lr": 0.001} +{"train_loss": 0.038035012781620026, "train_loss_bc": 0.03801603615283966, "train_loss_llm": 0.0018978416919708252, "grad_norm": 0.01981671340763569, "global_step": 10789, "epoch": 299, "lr": 0.001} +{"train_loss": 0.031323328614234924, "train_loss_bc": 0.031323328614234924, "train_loss_llm": 0.0, "grad_norm": 0.02790580317378044, "global_step": 10790, "epoch": 299, "lr": 0.001} +{"train_loss": 0.04273625835776329, "train_loss_bc": 0.042720288038253784, "train_loss_llm": 0.0015971660614013672, "grad_norm": 0.03821217641234398, "global_step": 10791, "epoch": 299, "lr": 0.001} +{"train_loss": 0.0410335399210453, "train_loss_bc": 0.04102855920791626, "train_loss_llm": 0.0004981756210327148, "grad_norm": 0.05227937549352646, "global_step": 10792, "epoch": 299, "lr": 0.001} +{"train_loss": 0.0376807264983654, "train_loss_bc": 0.03766830265522003, "train_loss_llm": 0.0012424290180206299, "grad_norm": 0.02322583645582199, "global_step": 10793, "epoch": 299, "lr": 0.001} +{"train_loss": 0.033612705767154694, "train_loss_bc": 0.033612705767154694, "train_loss_llm": 0.0, "grad_norm": 0.03322263062000275, "global_step": 10794, "epoch": 299, "lr": 0.001} +{"train_loss": 0.04494621977210045, "train_loss_bc": 0.04494621977210045, "train_loss_llm": 0.0, "grad_norm": 0.0414285808801651, "global_step": 10795, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03865116834640503, "train_loss_bc": 0.038640160113573074, "train_loss_llm": 0.0011008977890014648, "grad_norm": 0.05024915933609009, "global_step": 10796, "epoch": 299, "lr": 0.001} +{"train_loss": 0.03279927745461464, "train_loss_bc": 0.03279927745461464, "train_loss_llm": 0.0, "grad_norm": 0.018793998286128044, "global_step": 10797, "epoch": 299, "lr": 0.001} +{"train_loss": 0.04136588051915169, "train_loss_bc": 0.04136588051915169, "train_loss_llm": 0.0, "grad_norm": 0.028141004964709282, "global_step": 10798, "epoch": 299, "lr": 0.001} +{"train_loss": 0.0378334993082616, "train_loss_bc": 0.03550833836197853, "train_loss_llm": 0.0, "grad_norm": 0.040213171392679214, "global_step": 10799, "epoch": 299, "lr": 0.001, "val_loss": 0.06164964288473129} +{"train_loss": 0.03587550297379494, "train_loss_bc": 0.03583922237157822, "train_loss_llm": 0.003628075122833252, "grad_norm": 0.05123736709356308, "global_step": 10800, "epoch": 300, "lr": 0.001} +{"train_loss": 0.038762398064136505, "train_loss_bc": 0.038762398064136505, "train_loss_llm": 0.0, "grad_norm": 0.021264053881168365, "global_step": 10801, "epoch": 300, "lr": 0.001} +{"train_loss": 0.046115487813949585, "train_loss_bc": 0.046104345470666885, "train_loss_llm": 0.001114189624786377, "grad_norm": 0.03374699130654335, "global_step": 10802, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03662831336259842, "train_loss_bc": 0.03662141412496567, "train_loss_llm": 0.0006899535655975342, "grad_norm": 0.04140710085630417, "global_step": 10803, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03408970311284065, "train_loss_bc": 0.03408970311284065, "train_loss_llm": 0.0, "grad_norm": 0.05058339610695839, "global_step": 10804, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03999276086688042, "train_loss_bc": 0.03999276086688042, "train_loss_llm": 0.0, "grad_norm": 0.014925004914402962, "global_step": 10805, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03755025193095207, "train_loss_bc": 0.03755025193095207, "train_loss_llm": 0.0, "grad_norm": 0.027333436533808708, "global_step": 10806, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03771026059985161, "train_loss_bc": 0.03770555183291435, "train_loss_llm": 0.00047081708908081055, "grad_norm": 0.03584718331694603, "global_step": 10807, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03357713669538498, "train_loss_bc": 0.03357713669538498, "train_loss_llm": 0.0, "grad_norm": 0.044167786836624146, "global_step": 10808, "epoch": 300, "lr": 0.001} +{"train_loss": 0.028935197740793228, "train_loss_bc": 0.02892112545669079, "train_loss_llm": 0.0014072656631469727, "grad_norm": 0.0190604105591774, "global_step": 10809, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03900742903351784, "train_loss_bc": 0.03900742903351784, "train_loss_llm": 0.0, "grad_norm": 0.025593886151909828, "global_step": 10810, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03064117208123207, "train_loss_bc": 0.03064117208123207, "train_loss_llm": 0.0, "grad_norm": 0.0301196351647377, "global_step": 10811, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03567996993660927, "train_loss_bc": 0.03567996993660927, "train_loss_llm": 0.0, "grad_norm": 0.03386826813220978, "global_step": 10812, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03840789198875427, "train_loss_bc": 0.03838009759783745, "train_loss_llm": 0.0027796030044555664, "grad_norm": 0.02353578805923462, "global_step": 10813, "epoch": 300, "lr": 0.001} +{"train_loss": 0.04051980376243591, "train_loss_bc": 0.04051980376243591, "train_loss_llm": 0.0, "grad_norm": 0.02950972504913807, "global_step": 10814, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03319430723786354, "train_loss_bc": 0.033185139298439026, "train_loss_llm": 0.0009167194366455078, "grad_norm": 0.037532538175582886, "global_step": 10815, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03646097332239151, "train_loss_bc": 0.03646097332239151, "train_loss_llm": 0.0, "grad_norm": 0.04504703730344772, "global_step": 10816, "epoch": 300, "lr": 0.001} +{"train_loss": 0.035513680428266525, "train_loss_bc": 0.03550966456532478, "train_loss_llm": 0.0004017353057861328, "grad_norm": 0.018833547830581665, "global_step": 10817, "epoch": 300, "lr": 0.001} +{"train_loss": 0.037037648260593414, "train_loss_bc": 0.037037648260593414, "train_loss_llm": 0.0, "grad_norm": 0.03055604360997677, "global_step": 10818, "epoch": 300, "lr": 0.001} +{"train_loss": 0.035346776247024536, "train_loss_bc": 0.035346776247024536, "train_loss_llm": 0.0, "grad_norm": 0.03765982389450073, "global_step": 10819, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03392855077981949, "train_loss_bc": 0.03392855077981949, "train_loss_llm": 0.0, "grad_norm": 0.0412367582321167, "global_step": 10820, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03991355746984482, "train_loss_bc": 0.03991355746984482, "train_loss_llm": 0.0, "grad_norm": 0.017338844016194344, "global_step": 10821, "epoch": 300, "lr": 0.001} +{"train_loss": 0.04111804440617561, "train_loss_bc": 0.04111804440617561, "train_loss_llm": 0.0, "grad_norm": 0.03328128904104233, "global_step": 10822, "epoch": 300, "lr": 0.001} +{"train_loss": 0.041497744619846344, "train_loss_bc": 0.041497744619846344, "train_loss_llm": 0.0, "grad_norm": 0.041726257652044296, "global_step": 10823, "epoch": 300, "lr": 0.001} +{"train_loss": 0.036412544548511505, "train_loss_bc": 0.036412544548511505, "train_loss_llm": 0.0, "grad_norm": 0.05087984725832939, "global_step": 10824, "epoch": 300, "lr": 0.001} +{"train_loss": 0.04131234809756279, "train_loss_bc": 0.04131234809756279, "train_loss_llm": 0.0, "grad_norm": 0.019849911332130432, "global_step": 10825, "epoch": 300, "lr": 0.001} +{"train_loss": 0.042960040271282196, "train_loss_bc": 0.042938657104969025, "train_loss_llm": 0.0021383166313171387, "grad_norm": 0.027586044743657112, "global_step": 10826, "epoch": 300, "lr": 0.001} +{"train_loss": 0.034804727882146835, "train_loss_bc": 0.034804727882146835, "train_loss_llm": 0.0, "grad_norm": 0.0358128622174263, "global_step": 10827, "epoch": 300, "lr": 0.001} +{"train_loss": 0.041535522788763046, "train_loss_bc": 0.04153463616967201, "train_loss_llm": 8.851289749145508e-05, "grad_norm": 0.04800564423203468, "global_step": 10828, "epoch": 300, "lr": 0.001} +{"train_loss": 0.038469988852739334, "train_loss_bc": 0.038469988852739334, "train_loss_llm": 0.0, "grad_norm": 0.023215988650918007, "global_step": 10829, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03651890903711319, "train_loss_bc": 0.03651890903711319, "train_loss_llm": 0.0, "grad_norm": 0.033392976969480515, "global_step": 10830, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03673993796110153, "train_loss_bc": 0.03673993796110153, "train_loss_llm": 0.0, "grad_norm": 0.03792088106274605, "global_step": 10831, "epoch": 300, "lr": 0.001} +{"train_loss": 0.042698029428720474, "train_loss_bc": 0.042698029428720474, "train_loss_llm": 0.0, "grad_norm": 0.04333191365003586, "global_step": 10832, "epoch": 300, "lr": 0.001} +{"train_loss": 0.040716927498579025, "train_loss_bc": 0.040716927498579025, "train_loss_llm": 0.0, "grad_norm": 0.024608472362160683, "global_step": 10833, "epoch": 300, "lr": 0.001} +{"train_loss": 0.043711308389902115, "train_loss_bc": 0.043711308389902115, "train_loss_llm": 0.0, "grad_norm": 0.03758790343999863, "global_step": 10834, "epoch": 300, "lr": 0.001} +{"train_loss": 0.03777456304265393, "train_loss_bc": 0.036499422043561935, "train_loss_llm": 0.0, "grad_norm": 0.056083615869283676, "global_step": 10835, "epoch": 300, "lr": 0.001, "train/cumulative_reward": 75.99691157497327, "train/mean_score": 8.608884520276577, "train/success_rate": 0.2, "test/cumulative_reward": 21.42186489928742, "test/mean_score": 3.507710328296241, "test/success_rate": 0.06, "val_loss": 0.061654478311538696, "train_action_mse_error": 0.041076041758060455} +{"train_loss": 0.04196744039654732, "train_loss_bc": 0.04196744039654732, "train_loss_llm": 0.0, "grad_norm": 0.06662260740995407, "global_step": 10836, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03114953078329563, "train_loss_bc": 0.031129956245422363, "train_loss_llm": 0.0019574761390686035, "grad_norm": 0.017342638224363327, "global_step": 10837, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03963089361786842, "train_loss_bc": 0.03961937502026558, "train_loss_llm": 0.0011517703533172607, "grad_norm": 0.027157332748174667, "global_step": 10838, "epoch": 301, "lr": 0.001} +{"train_loss": 0.038435641676187515, "train_loss_bc": 0.03843512386083603, "train_loss_llm": 5.1647424697875977e-05, "grad_norm": 0.03697620704770088, "global_step": 10839, "epoch": 301, "lr": 0.001} +{"train_loss": 0.04333573207259178, "train_loss_bc": 0.0433209091424942, "train_loss_llm": 0.001482158899307251, "grad_norm": 0.042486634105443954, "global_step": 10840, "epoch": 301, "lr": 0.001} +{"train_loss": 0.04004960134625435, "train_loss_bc": 0.04004960134625435, "train_loss_llm": 0.0, "grad_norm": 0.021792268380522728, "global_step": 10841, "epoch": 301, "lr": 0.001} +{"train_loss": 0.037907786667346954, "train_loss_bc": 0.037907786667346954, "train_loss_llm": 0.0, "grad_norm": 0.032360535115003586, "global_step": 10842, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03863748908042908, "train_loss_bc": 0.03863377869129181, "train_loss_llm": 0.00037106871604919434, "grad_norm": 0.04082728922367096, "global_step": 10843, "epoch": 301, "lr": 0.001} +{"train_loss": 0.0366898849606514, "train_loss_bc": 0.0366898849606514, "train_loss_llm": 0.0, "grad_norm": 0.04498763382434845, "global_step": 10844, "epoch": 301, "lr": 0.001} +{"train_loss": 0.036021847277879715, "train_loss_bc": 0.036021847277879715, "train_loss_llm": 0.0, "grad_norm": 0.020334886386990547, "global_step": 10845, "epoch": 301, "lr": 0.001} +{"train_loss": 0.04080381989479065, "train_loss_bc": 0.04079300910234451, "train_loss_llm": 0.0010809004306793213, "grad_norm": 0.033575210720300674, "global_step": 10846, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03768333047628403, "train_loss_bc": 0.03768333047628403, "train_loss_llm": 0.0, "grad_norm": 0.046213798224925995, "global_step": 10847, "epoch": 301, "lr": 0.001} +{"train_loss": 0.028039945289492607, "train_loss_bc": 0.02803962491452694, "train_loss_llm": 3.209710121154785e-05, "grad_norm": 0.047907326370477676, "global_step": 10848, "epoch": 301, "lr": 0.001} +{"train_loss": 0.04145095497369766, "train_loss_bc": 0.04145095497369766, "train_loss_llm": 0.0, "grad_norm": 0.02251490205526352, "global_step": 10849, "epoch": 301, "lr": 0.001} +{"train_loss": 0.039369989186525345, "train_loss_bc": 0.039354532957077026, "train_loss_llm": 0.0015457868576049805, "grad_norm": 0.03410026431083679, "global_step": 10850, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03920654579997063, "train_loss_bc": 0.03920654579997063, "train_loss_llm": 0.0, "grad_norm": 0.04514014720916748, "global_step": 10851, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03914753720164299, "train_loss_bc": 0.03914753720164299, "train_loss_llm": 0.0, "grad_norm": 0.05166729539632797, "global_step": 10852, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03499531000852585, "train_loss_bc": 0.03499256819486618, "train_loss_llm": 0.00027436017990112305, "grad_norm": 0.021242331713438034, "global_step": 10853, "epoch": 301, "lr": 0.001} +{"train_loss": 0.04022209718823433, "train_loss_bc": 0.04022209718823433, "train_loss_llm": 0.0, "grad_norm": 0.0276923980563879, "global_step": 10854, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03528175130486488, "train_loss_bc": 0.03528175130486488, "train_loss_llm": 0.0, "grad_norm": 0.035812947899103165, "global_step": 10855, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03073178231716156, "train_loss_bc": 0.03073178231716156, "train_loss_llm": 0.0, "grad_norm": 0.040436241775751114, "global_step": 10856, "epoch": 301, "lr": 0.001} +{"train_loss": 0.04019305855035782, "train_loss_bc": 0.04019305855035782, "train_loss_llm": 0.0, "grad_norm": 0.02179737575352192, "global_step": 10857, "epoch": 301, "lr": 0.001} +{"train_loss": 0.0398549884557724, "train_loss_bc": 0.0398549884557724, "train_loss_llm": 0.0, "grad_norm": 0.029505504295229912, "global_step": 10858, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03787295147776604, "train_loss_bc": 0.03786209970712662, "train_loss_llm": 0.0010851621627807617, "grad_norm": 0.038220055401325226, "global_step": 10859, "epoch": 301, "lr": 0.001} +{"train_loss": 0.04036564379930496, "train_loss_bc": 0.04036564379930496, "train_loss_llm": 0.0, "grad_norm": 0.0439264290034771, "global_step": 10860, "epoch": 301, "lr": 0.001} +{"train_loss": 0.04105931520462036, "train_loss_bc": 0.04105931520462036, "train_loss_llm": 0.0, "grad_norm": 0.025598619133234024, "global_step": 10861, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03885383903980255, "train_loss_bc": 0.03885383903980255, "train_loss_llm": 0.0, "grad_norm": 0.0331856906414032, "global_step": 10862, "epoch": 301, "lr": 0.001} +{"train_loss": 0.029733801260590553, "train_loss_bc": 0.029733801260590553, "train_loss_llm": 0.0, "grad_norm": 0.040261611342430115, "global_step": 10863, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03732844069600105, "train_loss_bc": 0.03731021657586098, "train_loss_llm": 0.0018225014209747314, "grad_norm": 0.04895525425672531, "global_step": 10864, "epoch": 301, "lr": 0.001} +{"train_loss": 0.028550714254379272, "train_loss_bc": 0.028543490916490555, "train_loss_llm": 0.000722348690032959, "grad_norm": 0.015042237006127834, "global_step": 10865, "epoch": 301, "lr": 0.001} +{"train_loss": 0.039541736245155334, "train_loss_bc": 0.03953178972005844, "train_loss_llm": 0.0009948015213012695, "grad_norm": 0.028656184673309326, "global_step": 10866, "epoch": 301, "lr": 0.001} +{"train_loss": 0.04095049947500229, "train_loss_bc": 0.04095049947500229, "train_loss_llm": 0.0, "grad_norm": 0.04105313867330551, "global_step": 10867, "epoch": 301, "lr": 0.001} +{"train_loss": 0.039908092468976974, "train_loss_bc": 0.039908092468976974, "train_loss_llm": 0.0, "grad_norm": 0.04640975594520569, "global_step": 10868, "epoch": 301, "lr": 0.001} +{"train_loss": 0.04229313135147095, "train_loss_bc": 0.04229313135147095, "train_loss_llm": 0.0, "grad_norm": 0.02533613331615925, "global_step": 10869, "epoch": 301, "lr": 0.001} +{"train_loss": 0.03535401076078415, "train_loss_bc": 0.035343240946531296, "train_loss_llm": 0.0010771453380584717, "grad_norm": 0.035643622279167175, "global_step": 10870, "epoch": 301, "lr": 0.001} +{"train_loss": 0.037793256911552615, "train_loss_bc": 0.03793811425566673, "train_loss_llm": 0.0, "grad_norm": 0.041518233716487885, "global_step": 10871, "epoch": 301, "lr": 0.001, "val_loss": 0.06198950111865997} +{"train_loss": 0.037674739956855774, "train_loss_bc": 0.037674739956855774, "train_loss_llm": 0.0, "grad_norm": 0.05069112032651901, "global_step": 10872, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03210598602890968, "train_loss_bc": 0.03210598602890968, "train_loss_llm": 0.0, "grad_norm": 0.0168763380497694, "global_step": 10873, "epoch": 302, "lr": 0.001} +{"train_loss": 0.043410010635852814, "train_loss_bc": 0.043410010635852814, "train_loss_llm": 0.0, "grad_norm": 0.026834819465875626, "global_step": 10874, "epoch": 302, "lr": 0.001} +{"train_loss": 0.04383492469787598, "train_loss_bc": 0.04383231699466705, "train_loss_llm": 0.000260770320892334, "grad_norm": 0.03306342288851738, "global_step": 10875, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03562504053115845, "train_loss_bc": 0.03562504053115845, "train_loss_llm": 0.0, "grad_norm": 0.03897903859615326, "global_step": 10876, "epoch": 302, "lr": 0.001} +{"train_loss": 0.04510137438774109, "train_loss_bc": 0.04510137438774109, "train_loss_llm": 0.0, "grad_norm": 0.02459539659321308, "global_step": 10877, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03069019503891468, "train_loss_bc": 0.03069019503891468, "train_loss_llm": 0.0, "grad_norm": 0.030768442898988724, "global_step": 10878, "epoch": 302, "lr": 0.001} +{"train_loss": 0.032270871102809906, "train_loss_bc": 0.03226267173886299, "train_loss_llm": 0.000819772481918335, "grad_norm": 0.03978114202618599, "global_step": 10879, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03973362594842911, "train_loss_bc": 0.03973362594842911, "train_loss_llm": 0.0, "grad_norm": 0.04160292074084282, "global_step": 10880, "epoch": 302, "lr": 0.001} +{"train_loss": 0.032237373292446136, "train_loss_bc": 0.03223259747028351, "train_loss_llm": 0.0004774630069732666, "grad_norm": 0.015701893717050552, "global_step": 10881, "epoch": 302, "lr": 0.001} +{"train_loss": 0.045282527804374695, "train_loss_bc": 0.045282527804374695, "train_loss_llm": 0.0, "grad_norm": 0.023398302495479584, "global_step": 10882, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03838165104389191, "train_loss_bc": 0.03837337717413902, "train_loss_llm": 0.000827491283416748, "grad_norm": 0.030957337468862534, "global_step": 10883, "epoch": 302, "lr": 0.001} +{"train_loss": 0.04026412591338158, "train_loss_bc": 0.04026412591338158, "train_loss_llm": 0.0, "grad_norm": 0.043286193162202835, "global_step": 10884, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03487883880734444, "train_loss_bc": 0.03487883880734444, "train_loss_llm": 0.0, "grad_norm": 0.019574986770749092, "global_step": 10885, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03174400329589844, "train_loss_bc": 0.031720712780952454, "train_loss_llm": 0.0023291707038879395, "grad_norm": 0.031309712678194046, "global_step": 10886, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03910526633262634, "train_loss_bc": 0.03908683732151985, "train_loss_llm": 0.0018430650234222412, "grad_norm": 0.04412897303700447, "global_step": 10887, "epoch": 302, "lr": 0.001} +{"train_loss": 0.036212269216775894, "train_loss_bc": 0.036209434270858765, "train_loss_llm": 0.0002835392951965332, "grad_norm": 0.052470240741968155, "global_step": 10888, "epoch": 302, "lr": 0.001} +{"train_loss": 0.036106038838624954, "train_loss_bc": 0.03609022498130798, "train_loss_llm": 0.0015813708305358887, "grad_norm": 0.01728696934878826, "global_step": 10889, "epoch": 302, "lr": 0.001} +{"train_loss": 0.033932678401470184, "train_loss_bc": 0.033932678401470184, "train_loss_llm": 0.0, "grad_norm": 0.02657085657119751, "global_step": 10890, "epoch": 302, "lr": 0.001} +{"train_loss": 0.036687642335891724, "train_loss_bc": 0.03667595610022545, "train_loss_llm": 0.0011686086654663086, "grad_norm": 0.03384610265493393, "global_step": 10891, "epoch": 302, "lr": 0.001} +{"train_loss": 0.0340905599296093, "train_loss_bc": 0.0340905599296093, "train_loss_llm": 0.0, "grad_norm": 0.049626126885414124, "global_step": 10892, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03920536860823631, "train_loss_bc": 0.03920536860823631, "train_loss_llm": 0.0, "grad_norm": 0.018911607563495636, "global_step": 10893, "epoch": 302, "lr": 0.001} +{"train_loss": 0.04686911031603813, "train_loss_bc": 0.04686911031603813, "train_loss_llm": 0.0, "grad_norm": 0.025311700999736786, "global_step": 10894, "epoch": 302, "lr": 0.001} +{"train_loss": 0.035927511751651764, "train_loss_bc": 0.03592490404844284, "train_loss_llm": 0.0002606511116027832, "grad_norm": 0.03078562021255493, "global_step": 10895, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03900471329689026, "train_loss_bc": 0.03899442404508591, "train_loss_llm": 0.0010290145874023438, "grad_norm": 0.04152572154998779, "global_step": 10896, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03818738833069801, "train_loss_bc": 0.03818738833069801, "train_loss_llm": 0.0, "grad_norm": 0.01830708608031273, "global_step": 10897, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03720473870635033, "train_loss_bc": 0.037198297679424286, "train_loss_llm": 0.0006439387798309326, "grad_norm": 0.03080473467707634, "global_step": 10898, "epoch": 302, "lr": 0.001} +{"train_loss": 0.0439191572368145, "train_loss_bc": 0.0439191572368145, "train_loss_llm": 0.0, "grad_norm": 0.0359053835272789, "global_step": 10899, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03158628195524216, "train_loss_bc": 0.03158628195524216, "train_loss_llm": 0.0, "grad_norm": 0.040524616837501526, "global_step": 10900, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03376832231879234, "train_loss_bc": 0.03376832231879234, "train_loss_llm": 0.0, "grad_norm": 0.0171437356621027, "global_step": 10901, "epoch": 302, "lr": 0.001} +{"train_loss": 0.030624710023403168, "train_loss_bc": 0.030624710023403168, "train_loss_llm": 0.0, "grad_norm": 0.035028763115406036, "global_step": 10902, "epoch": 302, "lr": 0.001} +{"train_loss": 0.043575312942266464, "train_loss_bc": 0.043575312942266464, "train_loss_llm": 0.0, "grad_norm": 0.04578527435660362, "global_step": 10903, "epoch": 302, "lr": 0.001} +{"train_loss": 0.04393045976758003, "train_loss_bc": 0.04393045976758003, "train_loss_llm": 0.0, "grad_norm": 0.04923835024237633, "global_step": 10904, "epoch": 302, "lr": 0.001} +{"train_loss": 0.03121121972799301, "train_loss_bc": 0.03121121972799301, "train_loss_llm": 0.0, "grad_norm": 0.02105735056102276, "global_step": 10905, "epoch": 302, "lr": 0.001} +{"train_loss": 0.042645011097192764, "train_loss_bc": 0.042645011097192764, "train_loss_llm": 0.0, "grad_norm": 0.03947047144174576, "global_step": 10906, "epoch": 302, "lr": 0.001} +{"train_loss": 0.037735178083595305, "train_loss_bc": 0.041437361389398575, "train_loss_llm": 0.0, "grad_norm": 0.05404401198029518, "global_step": 10907, "epoch": 302, "lr": 0.001, "val_loss": 0.06243519112467766} +{"train_loss": 0.04026607424020767, "train_loss_bc": 0.04026399552822113, "train_loss_llm": 0.0002078413963317871, "grad_norm": 0.06628218293190002, "global_step": 10908, "epoch": 303, "lr": 0.001} +{"train_loss": 0.034180331975221634, "train_loss_bc": 0.034180331975221634, "train_loss_llm": 0.0, "grad_norm": 0.025297092273831367, "global_step": 10909, "epoch": 303, "lr": 0.001} +{"train_loss": 0.039479393512010574, "train_loss_bc": 0.039479393512010574, "train_loss_llm": 0.0, "grad_norm": 0.03463749587535858, "global_step": 10910, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03695949912071228, "train_loss_bc": 0.03695949912071228, "train_loss_llm": 0.0, "grad_norm": 0.0430750735104084, "global_step": 10911, "epoch": 303, "lr": 0.001} +{"train_loss": 0.038961779326200485, "train_loss_bc": 0.03894178941845894, "train_loss_llm": 0.001999109983444214, "grad_norm": 0.05642619729042053, "global_step": 10912, "epoch": 303, "lr": 0.001} +{"train_loss": 0.04250110313296318, "train_loss_bc": 0.04250110313296318, "train_loss_llm": 0.0, "grad_norm": 0.020367518067359924, "global_step": 10913, "epoch": 303, "lr": 0.001} +{"train_loss": 0.04037849232554436, "train_loss_bc": 0.04037849232554436, "train_loss_llm": 0.0, "grad_norm": 0.032510995864868164, "global_step": 10914, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03713762015104294, "train_loss_bc": 0.03713276609778404, "train_loss_llm": 0.00048539042472839355, "grad_norm": 0.037389859557151794, "global_step": 10915, "epoch": 303, "lr": 0.001} +{"train_loss": 0.037628211081027985, "train_loss_bc": 0.037628211081027985, "train_loss_llm": 0.0, "grad_norm": 0.04800577089190483, "global_step": 10916, "epoch": 303, "lr": 0.001} +{"train_loss": 0.035647083073854446, "train_loss_bc": 0.035647083073854446, "train_loss_llm": 0.0, "grad_norm": 0.02112746611237526, "global_step": 10917, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03253395855426788, "train_loss_bc": 0.03253395855426788, "train_loss_llm": 0.0, "grad_norm": 0.03127149119973183, "global_step": 10918, "epoch": 303, "lr": 0.001} +{"train_loss": 0.04311622306704521, "train_loss_bc": 0.04311622306704521, "train_loss_llm": 0.0, "grad_norm": 0.03741191700100899, "global_step": 10919, "epoch": 303, "lr": 0.001} +{"train_loss": 0.033698227256536484, "train_loss_bc": 0.033698227256536484, "train_loss_llm": 0.0, "grad_norm": 0.04598333314061165, "global_step": 10920, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03825896233320236, "train_loss_bc": 0.03823287412524223, "train_loss_llm": 0.0026087164878845215, "grad_norm": 0.018634561449289322, "global_step": 10921, "epoch": 303, "lr": 0.001} +{"train_loss": 0.04117554798722267, "train_loss_bc": 0.04116526618599892, "train_loss_llm": 0.0010280907154083252, "grad_norm": 0.02807021141052246, "global_step": 10922, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03621130436658859, "train_loss_bc": 0.03621130436658859, "train_loss_llm": 0.0, "grad_norm": 0.03511748090386391, "global_step": 10923, "epoch": 303, "lr": 0.001} +{"train_loss": 0.0357910580933094, "train_loss_bc": 0.0357910580933094, "train_loss_llm": 0.0, "grad_norm": 0.04437631368637085, "global_step": 10924, "epoch": 303, "lr": 0.001} +{"train_loss": 0.04573369398713112, "train_loss_bc": 0.045711908489465714, "train_loss_llm": 0.0021784603595733643, "grad_norm": 0.025716371834278107, "global_step": 10925, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03874073177576065, "train_loss_bc": 0.03874073177576065, "train_loss_llm": 0.0, "grad_norm": 0.031079694628715515, "global_step": 10926, "epoch": 303, "lr": 0.001} +{"train_loss": 0.033301256597042084, "train_loss_bc": 0.033301256597042084, "train_loss_llm": 0.0, "grad_norm": 0.03863922506570816, "global_step": 10927, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03725951537489891, "train_loss_bc": 0.037256430834531784, "train_loss_llm": 0.000308305025100708, "grad_norm": 0.0466364324092865, "global_step": 10928, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03215918689966202, "train_loss_bc": 0.03215918689966202, "train_loss_llm": 0.0, "grad_norm": 0.025400307029485703, "global_step": 10929, "epoch": 303, "lr": 0.001} +{"train_loss": 0.038090791553258896, "train_loss_bc": 0.0380697101354599, "train_loss_llm": 0.0021080970764160156, "grad_norm": 0.03906043991446495, "global_step": 10930, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03142305463552475, "train_loss_bc": 0.03142305463552475, "train_loss_llm": 0.0, "grad_norm": 0.04999546334147453, "global_step": 10931, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03717827424407005, "train_loss_bc": 0.037174005061388016, "train_loss_llm": 0.00042697787284851074, "grad_norm": 0.059680912643671036, "global_step": 10932, "epoch": 303, "lr": 0.001} +{"train_loss": 0.04219399020075798, "train_loss_bc": 0.04218636453151703, "train_loss_llm": 0.0007624924182891846, "grad_norm": 0.024778936058282852, "global_step": 10933, "epoch": 303, "lr": 0.001} +{"train_loss": 0.04727834463119507, "train_loss_bc": 0.04727834463119507, "train_loss_llm": 0.0, "grad_norm": 0.04193815216422081, "global_step": 10934, "epoch": 303, "lr": 0.001} +{"train_loss": 0.035987790673971176, "train_loss_bc": 0.035987790673971176, "train_loss_llm": 0.0, "grad_norm": 0.05530305206775665, "global_step": 10935, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03314468264579773, "train_loss_bc": 0.03313404321670532, "train_loss_llm": 0.0010638535022735596, "grad_norm": 0.06253602355718613, "global_step": 10936, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03757454454898834, "train_loss_bc": 0.03756987303495407, "train_loss_llm": 0.00046712160110473633, "grad_norm": 0.01963384822010994, "global_step": 10937, "epoch": 303, "lr": 0.001} +{"train_loss": 0.0325281023979187, "train_loss_bc": 0.0325281023979187, "train_loss_llm": 0.0, "grad_norm": 0.028829285874962807, "global_step": 10938, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03966011852025986, "train_loss_bc": 0.03965887427330017, "train_loss_llm": 0.00012430548667907715, "grad_norm": 0.03907035291194916, "global_step": 10939, "epoch": 303, "lr": 0.001} +{"train_loss": 0.0392303429543972, "train_loss_bc": 0.039226993918418884, "train_loss_llm": 0.0003349184989929199, "grad_norm": 0.04773929715156555, "global_step": 10940, "epoch": 303, "lr": 0.001} +{"train_loss": 0.04250301048159599, "train_loss_bc": 0.04250301048159599, "train_loss_llm": 0.0, "grad_norm": 0.026286697015166283, "global_step": 10941, "epoch": 303, "lr": 0.001} +{"train_loss": 0.03845859691500664, "train_loss_bc": 0.03845859691500664, "train_loss_llm": 0.0, "grad_norm": 0.03963787853717804, "global_step": 10942, "epoch": 303, "lr": 0.001} +{"train_loss": 0.037619960132158466, "train_loss_bc": 0.027943195775151253, "train_loss_llm": 0.0004470348358154297, "grad_norm": 0.04944193363189697, "global_step": 10943, "epoch": 303, "lr": 0.001, "val_loss": 0.06189139559864998} +{"train_loss": 0.03973715007305145, "train_loss_bc": 0.0397198460996151, "train_loss_llm": 0.0017303526401519775, "grad_norm": 0.05767365172505379, "global_step": 10944, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03668494150042534, "train_loss_bc": 0.036679238080978394, "train_loss_llm": 0.0005703270435333252, "grad_norm": 0.017456095665693283, "global_step": 10945, "epoch": 304, "lr": 0.001} +{"train_loss": 0.037648722529411316, "train_loss_bc": 0.03764845058321953, "train_loss_llm": 2.709031105041504e-05, "grad_norm": 0.03289075568318367, "global_step": 10946, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03758065402507782, "train_loss_bc": 0.03758065402507782, "train_loss_llm": 0.0, "grad_norm": 0.0371004156768322, "global_step": 10947, "epoch": 304, "lr": 0.001} +{"train_loss": 0.04266703873872757, "train_loss_bc": 0.04266703873872757, "train_loss_llm": 0.0, "grad_norm": 0.04740675538778305, "global_step": 10948, "epoch": 304, "lr": 0.001} +{"train_loss": 0.04017892852425575, "train_loss_bc": 0.04017892852425575, "train_loss_llm": 0.0, "grad_norm": 0.021205857396125793, "global_step": 10949, "epoch": 304, "lr": 0.001} +{"train_loss": 0.04440654441714287, "train_loss_bc": 0.04437936469912529, "train_loss_llm": 0.0027180612087249756, "grad_norm": 0.029306916519999504, "global_step": 10950, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03707580268383026, "train_loss_bc": 0.037051647901535034, "train_loss_llm": 0.0024153590202331543, "grad_norm": 0.04489511996507645, "global_step": 10951, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03627627715468407, "train_loss_bc": 0.03627627715468407, "train_loss_llm": 0.0, "grad_norm": 0.05634855479001999, "global_step": 10952, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03458232432603836, "train_loss_bc": 0.03457978740334511, "train_loss_llm": 0.0002536475658416748, "grad_norm": 0.027009213343262672, "global_step": 10953, "epoch": 304, "lr": 0.001} +{"train_loss": 0.034791748970746994, "train_loss_bc": 0.034789618104696274, "train_loss_llm": 0.00021320581436157227, "grad_norm": 0.031678300350904465, "global_step": 10954, "epoch": 304, "lr": 0.001} +{"train_loss": 0.04727524146437645, "train_loss_bc": 0.04727524146437645, "train_loss_llm": 0.0, "grad_norm": 0.047533974051475525, "global_step": 10955, "epoch": 304, "lr": 0.001} +{"train_loss": 0.04030347988009453, "train_loss_bc": 0.04030347988009453, "train_loss_llm": 0.0, "grad_norm": 0.05804338678717613, "global_step": 10956, "epoch": 304, "lr": 0.001} +{"train_loss": 0.0404973030090332, "train_loss_bc": 0.0404910184442997, "train_loss_llm": 0.0006285905838012695, "grad_norm": 0.018393350765109062, "global_step": 10957, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03574231639504433, "train_loss_bc": 0.03574231639504433, "train_loss_llm": 0.0, "grad_norm": 0.027978243306279182, "global_step": 10958, "epoch": 304, "lr": 0.001} +{"train_loss": 0.04113222658634186, "train_loss_bc": 0.04110940918326378, "train_loss_llm": 0.0022816061973571777, "grad_norm": 0.041172172874212265, "global_step": 10959, "epoch": 304, "lr": 0.001} +{"train_loss": 0.034698739647865295, "train_loss_bc": 0.034698739647865295, "train_loss_llm": 0.0, "grad_norm": 0.046028558164834976, "global_step": 10960, "epoch": 304, "lr": 0.001} +{"train_loss": 0.04338608682155609, "train_loss_bc": 0.043383583426475525, "train_loss_llm": 0.00025027990341186523, "grad_norm": 0.02521084062755108, "global_step": 10961, "epoch": 304, "lr": 0.001} +{"train_loss": 0.04365159198641777, "train_loss_bc": 0.04365159198641777, "train_loss_llm": 0.0, "grad_norm": 0.03730785474181175, "global_step": 10962, "epoch": 304, "lr": 0.001} +{"train_loss": 0.0391281396150589, "train_loss_bc": 0.0391281396150589, "train_loss_llm": 0.0, "grad_norm": 0.05031178519129753, "global_step": 10963, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03202728182077408, "train_loss_bc": 0.03202728182077408, "train_loss_llm": 0.0, "grad_norm": 0.0580952949821949, "global_step": 10964, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03584228828549385, "train_loss_bc": 0.03584228828549385, "train_loss_llm": 0.0, "grad_norm": 0.02414078637957573, "global_step": 10965, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03615530580282211, "train_loss_bc": 0.03615530580282211, "train_loss_llm": 0.0, "grad_norm": 0.03546975180506706, "global_step": 10966, "epoch": 304, "lr": 0.001} +{"train_loss": 0.038129772990942, "train_loss_bc": 0.03812680020928383, "train_loss_llm": 0.0002973973751068115, "grad_norm": 0.046316128224134445, "global_step": 10967, "epoch": 304, "lr": 0.001} +{"train_loss": 0.030745603144168854, "train_loss_bc": 0.030745238065719604, "train_loss_llm": 3.647804260253906e-05, "grad_norm": 0.05809476971626282, "global_step": 10968, "epoch": 304, "lr": 0.001} +{"train_loss": 0.04087427631020546, "train_loss_bc": 0.04087427631020546, "train_loss_llm": 0.0, "grad_norm": 0.025853728875517845, "global_step": 10969, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03148290142416954, "train_loss_bc": 0.03148290142416954, "train_loss_llm": 0.0, "grad_norm": 0.0354953333735466, "global_step": 10970, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03782716393470764, "train_loss_bc": 0.03782716393470764, "train_loss_llm": 0.0, "grad_norm": 0.048261094838380814, "global_step": 10971, "epoch": 304, "lr": 0.001} +{"train_loss": 0.046029429882764816, "train_loss_bc": 0.046029429882764816, "train_loss_llm": 0.0, "grad_norm": 0.05783505365252495, "global_step": 10972, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03203587234020233, "train_loss_bc": 0.03203587234020233, "train_loss_llm": 0.0, "grad_norm": 0.02128874510526657, "global_step": 10973, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03539887070655823, "train_loss_bc": 0.03539887070655823, "train_loss_llm": 0.0, "grad_norm": 0.034527067095041275, "global_step": 10974, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03522687032818794, "train_loss_bc": 0.03522687032818794, "train_loss_llm": 0.0, "grad_norm": 0.043876197189092636, "global_step": 10975, "epoch": 304, "lr": 0.001} +{"train_loss": 0.029097849503159523, "train_loss_bc": 0.029097849503159523, "train_loss_llm": 0.0, "grad_norm": 0.051596466451883316, "global_step": 10976, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03549258038401604, "train_loss_bc": 0.03549258038401604, "train_loss_llm": 0.0, "grad_norm": 0.01939217932522297, "global_step": 10977, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03669215738773346, "train_loss_bc": 0.03669215738773346, "train_loss_llm": 0.0, "grad_norm": 0.025093775242567062, "global_step": 10978, "epoch": 304, "lr": 0.001} +{"train_loss": 0.03768165952836474, "train_loss_bc": 0.036022529006004333, "train_loss_llm": 0.001373291015625, "grad_norm": 0.03669264167547226, "global_step": 10979, "epoch": 304, "lr": 0.001, "val_loss": 0.06303625553846359} +{"train_loss": 0.03861698508262634, "train_loss_bc": 0.03861698508262634, "train_loss_llm": 0.0, "grad_norm": 0.044034987688064575, "global_step": 10980, "epoch": 305, "lr": 0.001} +{"train_loss": 0.04190889000892639, "train_loss_bc": 0.041888527572155, "train_loss_llm": 0.0020362138748168945, "grad_norm": 0.020994586870074272, "global_step": 10981, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03848417103290558, "train_loss_bc": 0.03848417103290558, "train_loss_llm": 0.0, "grad_norm": 0.04124720022082329, "global_step": 10982, "epoch": 305, "lr": 0.001} +{"train_loss": 0.0386381559073925, "train_loss_bc": 0.0386381559073925, "train_loss_llm": 0.0, "grad_norm": 0.05322949215769768, "global_step": 10983, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03343424201011658, "train_loss_bc": 0.033413808792829514, "train_loss_llm": 0.002043426036834717, "grad_norm": 0.0637015849351883, "global_step": 10984, "epoch": 305, "lr": 0.001} +{"train_loss": 0.04715504124760628, "train_loss_bc": 0.04715504124760628, "train_loss_llm": 0.0, "grad_norm": 0.023917989805340767, "global_step": 10985, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03728795424103737, "train_loss_bc": 0.03728795424103737, "train_loss_llm": 0.0, "grad_norm": 0.03796226903796196, "global_step": 10986, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03390957787632942, "train_loss_bc": 0.03390957787632942, "train_loss_llm": 0.0, "grad_norm": 0.04495497792959213, "global_step": 10987, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03782579302787781, "train_loss_bc": 0.03782579302787781, "train_loss_llm": 0.0, "grad_norm": 0.05524611100554466, "global_step": 10988, "epoch": 305, "lr": 0.001} +{"train_loss": 0.034447167068719864, "train_loss_bc": 0.034447167068719864, "train_loss_llm": 0.0, "grad_norm": 0.021834291517734528, "global_step": 10989, "epoch": 305, "lr": 0.001} +{"train_loss": 0.030277535319328308, "train_loss_bc": 0.030270574614405632, "train_loss_llm": 0.000696033239364624, "grad_norm": 0.033358559012413025, "global_step": 10990, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03617914021015167, "train_loss_bc": 0.03617914021015167, "train_loss_llm": 0.0, "grad_norm": 0.03955375403165817, "global_step": 10991, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03879297897219658, "train_loss_bc": 0.03879297897219658, "train_loss_llm": 0.0, "grad_norm": 0.047377295792102814, "global_step": 10992, "epoch": 305, "lr": 0.001} +{"train_loss": 0.046336766332387924, "train_loss_bc": 0.046336766332387924, "train_loss_llm": 0.0, "grad_norm": 0.019459018483757973, "global_step": 10993, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03357737511396408, "train_loss_bc": 0.03357737511396408, "train_loss_llm": 0.0, "grad_norm": 0.031388573348522186, "global_step": 10994, "epoch": 305, "lr": 0.001} +{"train_loss": 0.038562312722206116, "train_loss_bc": 0.038549069315195084, "train_loss_llm": 0.0013245046138763428, "grad_norm": 0.03504636883735657, "global_step": 10995, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03564043343067169, "train_loss_bc": 0.03563418239355087, "train_loss_llm": 0.000624924898147583, "grad_norm": 0.04047275334596634, "global_step": 10996, "epoch": 305, "lr": 0.001} +{"train_loss": 0.034128472208976746, "train_loss_bc": 0.034128472208976746, "train_loss_llm": 0.0, "grad_norm": 0.02033015340566635, "global_step": 10997, "epoch": 305, "lr": 0.001} +{"train_loss": 0.038818590342998505, "train_loss_bc": 0.038818590342998505, "train_loss_llm": 0.0, "grad_norm": 0.03209322690963745, "global_step": 10998, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03295870125293732, "train_loss_bc": 0.03295870125293732, "train_loss_llm": 0.0, "grad_norm": 0.033916544169187546, "global_step": 10999, "epoch": 305, "lr": 0.001} +{"train_loss": 0.04031241312623024, "train_loss_bc": 0.04031241312623024, "train_loss_llm": 0.0, "grad_norm": 0.05016469210386276, "global_step": 11000, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03679872304201126, "train_loss_bc": 0.036771588027477264, "train_loss_llm": 0.002713322639465332, "grad_norm": 0.022335179150104523, "global_step": 11001, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03898061439394951, "train_loss_bc": 0.03898061439394951, "train_loss_llm": 0.0, "grad_norm": 0.038269780576229095, "global_step": 11002, "epoch": 305, "lr": 0.001} +{"train_loss": 0.029895544052124023, "train_loss_bc": 0.02989152818918228, "train_loss_llm": 0.0004016458988189697, "grad_norm": 0.04752569645643234, "global_step": 11003, "epoch": 305, "lr": 0.001} +{"train_loss": 0.04003838449716568, "train_loss_bc": 0.0400373600423336, "train_loss_llm": 0.00010246038436889648, "grad_norm": 0.05267205834388733, "global_step": 11004, "epoch": 305, "lr": 0.001} +{"train_loss": 0.04137043282389641, "train_loss_bc": 0.04137043282389641, "train_loss_llm": 0.0, "grad_norm": 0.024267494678497314, "global_step": 11005, "epoch": 305, "lr": 0.001} +{"train_loss": 0.0414888970553875, "train_loss_bc": 0.0414888970553875, "train_loss_llm": 0.0, "grad_norm": 0.035004060715436935, "global_step": 11006, "epoch": 305, "lr": 0.001} +{"train_loss": 0.04108237475156784, "train_loss_bc": 0.04108237475156784, "train_loss_llm": 0.0, "grad_norm": 0.04588913172483444, "global_step": 11007, "epoch": 305, "lr": 0.001} +{"train_loss": 0.044937580823898315, "train_loss_bc": 0.044937580823898315, "train_loss_llm": 0.0, "grad_norm": 0.055492714047431946, "global_step": 11008, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03125733882188797, "train_loss_bc": 0.03125733882188797, "train_loss_llm": 0.0, "grad_norm": 0.01888883113861084, "global_step": 11009, "epoch": 305, "lr": 0.001} +{"train_loss": 0.029934270307421684, "train_loss_bc": 0.029934270307421684, "train_loss_llm": 0.0, "grad_norm": 0.028143953531980515, "global_step": 11010, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03811264410614967, "train_loss_bc": 0.038112107664346695, "train_loss_llm": 5.379319190979004e-05, "grad_norm": 0.04056506231427193, "global_step": 11011, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03545469790697098, "train_loss_bc": 0.03545469790697098, "train_loss_llm": 0.0, "grad_norm": 0.0480801947414875, "global_step": 11012, "epoch": 305, "lr": 0.001} +{"train_loss": 0.04199293628334999, "train_loss_bc": 0.04199293628334999, "train_loss_llm": 0.0, "grad_norm": 0.025092551484704018, "global_step": 11013, "epoch": 305, "lr": 0.001} +{"train_loss": 0.039853498339653015, "train_loss_bc": 0.039853498339653015, "train_loss_llm": 0.0, "grad_norm": 0.044882383197546005, "global_step": 11014, "epoch": 305, "lr": 0.001} +{"train_loss": 0.03772014922772845, "train_loss_bc": 0.03943473845720291, "train_loss_llm": 0.0, "grad_norm": 0.06005994230508804, "global_step": 11015, "epoch": 305, "lr": 0.001, "train/cumulative_reward": 113.2237973030451, "train/mean_score": 17.90093118256371, "train/success_rate": 0.5, "test/cumulative_reward": 41.94300288342143, "test/mean_score": 8.537133884752608, "test/success_rate": 0.2, "val_loss": 0.06187316030263901, "train_action_mse_error": 0.04099377617239952} +{"train_loss": 0.03186039999127388, "train_loss_bc": 0.03186039999127388, "train_loss_llm": 0.0, "grad_norm": 0.06848081201314926, "global_step": 11016, "epoch": 306, "lr": 0.001} +{"train_loss": 0.034433864057064056, "train_loss_bc": 0.034433864057064056, "train_loss_llm": 0.0, "grad_norm": 0.018572304397821426, "global_step": 11017, "epoch": 306, "lr": 0.001} +{"train_loss": 0.04273500293493271, "train_loss_bc": 0.04273500293493271, "train_loss_llm": 0.0, "grad_norm": 0.02857718989253044, "global_step": 11018, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03372765704989433, "train_loss_bc": 0.03372765704989433, "train_loss_llm": 0.0, "grad_norm": 0.0300735030323267, "global_step": 11019, "epoch": 306, "lr": 0.001} +{"train_loss": 0.042340487241744995, "train_loss_bc": 0.042340487241744995, "train_loss_llm": 0.0, "grad_norm": 0.04121923819184303, "global_step": 11020, "epoch": 306, "lr": 0.001} +{"train_loss": 0.031019264832139015, "train_loss_bc": 0.031019264832139015, "train_loss_llm": 0.0, "grad_norm": 0.018231870606541634, "global_step": 11021, "epoch": 306, "lr": 0.001} +{"train_loss": 0.040280040353536606, "train_loss_bc": 0.040280040353536606, "train_loss_llm": 0.0, "grad_norm": 0.028555629774928093, "global_step": 11022, "epoch": 306, "lr": 0.001} +{"train_loss": 0.031044134870171547, "train_loss_bc": 0.031044134870171547, "train_loss_llm": 0.0, "grad_norm": 0.04178500920534134, "global_step": 11023, "epoch": 306, "lr": 0.001} +{"train_loss": 0.039152733981609344, "train_loss_bc": 0.039152733981609344, "train_loss_llm": 0.0, "grad_norm": 0.050552576780319214, "global_step": 11024, "epoch": 306, "lr": 0.001} +{"train_loss": 0.043262191116809845, "train_loss_bc": 0.043262191116809845, "train_loss_llm": 0.0, "grad_norm": 0.023471329361200333, "global_step": 11025, "epoch": 306, "lr": 0.001} +{"train_loss": 0.047159429639577866, "train_loss_bc": 0.047159429639577866, "train_loss_llm": 0.0, "grad_norm": 0.036358386278152466, "global_step": 11026, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03218849003314972, "train_loss_bc": 0.03218849003314972, "train_loss_llm": 0.0, "grad_norm": 0.04324565827846527, "global_step": 11027, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03197769448161125, "train_loss_bc": 0.03197769448161125, "train_loss_llm": 0.0, "grad_norm": 0.05519901588559151, "global_step": 11028, "epoch": 306, "lr": 0.001} +{"train_loss": 0.038054853677749634, "train_loss_bc": 0.038054853677749634, "train_loss_llm": 0.0, "grad_norm": 0.02393735572695732, "global_step": 11029, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03507615253329277, "train_loss_bc": 0.035062819719314575, "train_loss_llm": 0.0013333559036254883, "grad_norm": 0.03420644253492355, "global_step": 11030, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03719015419483185, "train_loss_bc": 0.037184733897447586, "train_loss_llm": 0.0005418956279754639, "grad_norm": 0.041305821388959885, "global_step": 11031, "epoch": 306, "lr": 0.001} +{"train_loss": 0.0324116051197052, "train_loss_bc": 0.03240863233804703, "train_loss_llm": 0.00029730796813964844, "grad_norm": 0.047209639102220535, "global_step": 11032, "epoch": 306, "lr": 0.001} +{"train_loss": 0.037112124264240265, "train_loss_bc": 0.037090953439474106, "train_loss_llm": 0.0021169185638427734, "grad_norm": 0.020770160481333733, "global_step": 11033, "epoch": 306, "lr": 0.001} +{"train_loss": 0.041752398014068604, "train_loss_bc": 0.041752398014068604, "train_loss_llm": 0.0, "grad_norm": 0.03158428892493248, "global_step": 11034, "epoch": 306, "lr": 0.001} +{"train_loss": 0.037342485040426254, "train_loss_bc": 0.037333037704229355, "train_loss_llm": 0.0009447038173675537, "grad_norm": 0.038812924176454544, "global_step": 11035, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03571869805455208, "train_loss_bc": 0.03571869805455208, "train_loss_llm": 0.0, "grad_norm": 0.04451879486441612, "global_step": 11036, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03503870218992233, "train_loss_bc": 0.03503870218992233, "train_loss_llm": 0.0, "grad_norm": 0.02438165433704853, "global_step": 11037, "epoch": 306, "lr": 0.001} +{"train_loss": 0.0407753586769104, "train_loss_bc": 0.0407753586769104, "train_loss_llm": 0.0, "grad_norm": 0.03124411031603813, "global_step": 11038, "epoch": 306, "lr": 0.001} +{"train_loss": 0.04240737482905388, "train_loss_bc": 0.04240249842405319, "train_loss_llm": 0.0004877746105194092, "grad_norm": 0.03525656834244728, "global_step": 11039, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03813871368765831, "train_loss_bc": 0.038132987916469574, "train_loss_llm": 0.000572502613067627, "grad_norm": 0.04463862627744675, "global_step": 11040, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03776627779006958, "train_loss_bc": 0.037739288061857224, "train_loss_llm": 0.002698928117752075, "grad_norm": 0.024469634518027306, "global_step": 11041, "epoch": 306, "lr": 0.001} +{"train_loss": 0.028544284403324127, "train_loss_bc": 0.028544284403324127, "train_loss_llm": 0.0, "grad_norm": 0.03648349270224571, "global_step": 11042, "epoch": 306, "lr": 0.001} +{"train_loss": 0.04436146840453148, "train_loss_bc": 0.04434804245829582, "train_loss_llm": 0.0013426542282104492, "grad_norm": 0.05367371439933777, "global_step": 11043, "epoch": 306, "lr": 0.001} +{"train_loss": 0.037810634821653366, "train_loss_bc": 0.037810634821653366, "train_loss_llm": 0.0, "grad_norm": 0.07003241777420044, "global_step": 11044, "epoch": 306, "lr": 0.001} +{"train_loss": 0.04300053045153618, "train_loss_bc": 0.0429844968020916, "train_loss_llm": 0.001603543758392334, "grad_norm": 0.02418266236782074, "global_step": 11045, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03704395517706871, "train_loss_bc": 0.0370243601500988, "train_loss_llm": 0.0019595324993133545, "grad_norm": 0.038927704095840454, "global_step": 11046, "epoch": 306, "lr": 0.001} +{"train_loss": 0.04440589249134064, "train_loss_bc": 0.04440589249134064, "train_loss_llm": 0.0, "grad_norm": 0.045369334518909454, "global_step": 11047, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03679205849766731, "train_loss_bc": 0.03679205849766731, "train_loss_llm": 0.0, "grad_norm": 0.05317007750272751, "global_step": 11048, "epoch": 306, "lr": 0.001} +{"train_loss": 0.0369381457567215, "train_loss_bc": 0.0369381457567215, "train_loss_llm": 0.0, "grad_norm": 0.01898212358355522, "global_step": 11049, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03950006514787674, "train_loss_bc": 0.03949262201786041, "train_loss_llm": 0.0007444322109222412, "grad_norm": 0.03125385567545891, "global_step": 11050, "epoch": 306, "lr": 0.001} +{"train_loss": 0.03767553137408362, "train_loss_bc": 0.03795580565929413, "train_loss_llm": 0.0, "grad_norm": 0.043347276747226715, "global_step": 11051, "epoch": 306, "lr": 0.001, "val_loss": 0.06301118433475494} +{"train_loss": 0.03201346471905708, "train_loss_bc": 0.03201105445623398, "train_loss_llm": 0.0002409219741821289, "grad_norm": 0.050274960696697235, "global_step": 11052, "epoch": 307, "lr": 0.001} +{"train_loss": 0.036993227899074554, "train_loss_bc": 0.036993227899074554, "train_loss_llm": 0.0, "grad_norm": 0.018485529348254204, "global_step": 11053, "epoch": 307, "lr": 0.001} +{"train_loss": 0.027388807386159897, "train_loss_bc": 0.027384918183088303, "train_loss_llm": 0.0003889799118041992, "grad_norm": 0.024917714297771454, "global_step": 11054, "epoch": 307, "lr": 0.001} +{"train_loss": 0.031239336356520653, "train_loss_bc": 0.031239336356520653, "train_loss_llm": 0.0, "grad_norm": 0.03006376139819622, "global_step": 11055, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03544885292649269, "train_loss_bc": 0.03544406592845917, "train_loss_llm": 0.0004788041114807129, "grad_norm": 0.033598318696022034, "global_step": 11056, "epoch": 307, "lr": 0.001} +{"train_loss": 0.038252394646406174, "train_loss_bc": 0.038252394646406174, "train_loss_llm": 0.0, "grad_norm": 0.017683468759059906, "global_step": 11057, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03054998070001602, "train_loss_bc": 0.03054998070001602, "train_loss_llm": 0.0, "grad_norm": 0.029305094853043556, "global_step": 11058, "epoch": 307, "lr": 0.001} +{"train_loss": 0.036342963576316833, "train_loss_bc": 0.03633204847574234, "train_loss_llm": 0.0010915100574493408, "grad_norm": 0.03541833162307739, "global_step": 11059, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03970499336719513, "train_loss_bc": 0.03970499336719513, "train_loss_llm": 0.0, "grad_norm": 0.04231013357639313, "global_step": 11060, "epoch": 307, "lr": 0.001} +{"train_loss": 0.034711405634880066, "train_loss_bc": 0.034711405634880066, "train_loss_llm": 0.0, "grad_norm": 0.018816111609339714, "global_step": 11061, "epoch": 307, "lr": 0.001} +{"train_loss": 0.042786527425050735, "train_loss_bc": 0.04277877137064934, "train_loss_llm": 0.000775754451751709, "grad_norm": 0.04437319189310074, "global_step": 11062, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03789365664124489, "train_loss_bc": 0.03789365664124489, "train_loss_llm": 0.0, "grad_norm": 0.05801163986325264, "global_step": 11063, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04202549532055855, "train_loss_bc": 0.04202549532055855, "train_loss_llm": 0.0, "grad_norm": 0.07232996076345444, "global_step": 11064, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03680504113435745, "train_loss_bc": 0.03680504113435745, "train_loss_llm": 0.0, "grad_norm": 0.022462649270892143, "global_step": 11065, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04167991131544113, "train_loss_bc": 0.04167991131544113, "train_loss_llm": 0.0, "grad_norm": 0.03934739902615547, "global_step": 11066, "epoch": 307, "lr": 0.001} +{"train_loss": 0.030611667782068253, "train_loss_bc": 0.030611667782068253, "train_loss_llm": 0.0, "grad_norm": 0.050407763570547104, "global_step": 11067, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04472147300839424, "train_loss_bc": 0.04471278563141823, "train_loss_llm": 0.0008687674999237061, "grad_norm": 0.07165850698947906, "global_step": 11068, "epoch": 307, "lr": 0.001} +{"train_loss": 0.034180156886577606, "train_loss_bc": 0.034180156886577606, "train_loss_llm": 0.0, "grad_norm": 0.020116109400987625, "global_step": 11069, "epoch": 307, "lr": 0.001} +{"train_loss": 0.032705843448638916, "train_loss_bc": 0.032705843448638916, "train_loss_llm": 0.0, "grad_norm": 0.03533204272389412, "global_step": 11070, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04060763865709305, "train_loss_bc": 0.04060763865709305, "train_loss_llm": 0.0, "grad_norm": 0.05453319475054741, "global_step": 11071, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04096875339746475, "train_loss_bc": 0.04096875339746475, "train_loss_llm": 0.0, "grad_norm": 0.06626420468091965, "global_step": 11072, "epoch": 307, "lr": 0.001} +{"train_loss": 0.040168311446905136, "train_loss_bc": 0.040165308862924576, "train_loss_llm": 0.0003001987934112549, "grad_norm": 0.02327047847211361, "global_step": 11073, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03888411819934845, "train_loss_bc": 0.03888411819934845, "train_loss_llm": 0.0, "grad_norm": 0.029948383569717407, "global_step": 11074, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04565685987472534, "train_loss_bc": 0.04565685987472534, "train_loss_llm": 0.0, "grad_norm": 0.04089716449379921, "global_step": 11075, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04854882135987282, "train_loss_bc": 0.04854882135987282, "train_loss_llm": 0.0, "grad_norm": 0.04721074923872948, "global_step": 11076, "epoch": 307, "lr": 0.001} +{"train_loss": 0.033517640084028244, "train_loss_bc": 0.033517640084028244, "train_loss_llm": 0.0, "grad_norm": 0.022100867703557014, "global_step": 11077, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03345145657658577, "train_loss_bc": 0.03345145657658577, "train_loss_llm": 0.0, "grad_norm": 0.03332478553056717, "global_step": 11078, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03809313103556633, "train_loss_bc": 0.038088005036115646, "train_loss_llm": 0.0005125999450683594, "grad_norm": 0.04028195142745972, "global_step": 11079, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03459012508392334, "train_loss_bc": 0.034555546939373016, "train_loss_llm": 0.003457963466644287, "grad_norm": 0.04739637300372124, "global_step": 11080, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04247039556503296, "train_loss_bc": 0.04247039556503296, "train_loss_llm": 0.0, "grad_norm": 0.02152044139802456, "global_step": 11081, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04412363842129707, "train_loss_bc": 0.04412363842129707, "train_loss_llm": 0.0, "grad_norm": 0.031008973717689514, "global_step": 11082, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03455734997987747, "train_loss_bc": 0.03453989699482918, "train_loss_llm": 0.001745462417602539, "grad_norm": 0.041253894567489624, "global_step": 11083, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03440362960100174, "train_loss_bc": 0.03440362960100174, "train_loss_llm": 0.0, "grad_norm": 0.0530475452542305, "global_step": 11084, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04498216509819031, "train_loss_bc": 0.04498216509819031, "train_loss_llm": 0.0, "grad_norm": 0.027198653668165207, "global_step": 11085, "epoch": 307, "lr": 0.001} +{"train_loss": 0.04285962134599686, "train_loss_bc": 0.042832233011722565, "train_loss_llm": 0.002738773822784424, "grad_norm": 0.043763820081949234, "global_step": 11086, "epoch": 307, "lr": 0.001} +{"train_loss": 0.03782491825727953, "train_loss_bc": 0.037758201360702515, "train_loss_llm": 0.0, "grad_norm": 0.05550404638051987, "global_step": 11087, "epoch": 307, "lr": 0.001, "val_loss": 0.06228669732809067} +{"train_loss": 0.04167252779006958, "train_loss_bc": 0.04167252779006958, "train_loss_llm": 0.0, "grad_norm": 0.06337733566761017, "global_step": 11088, "epoch": 308, "lr": 0.001} +{"train_loss": 0.036013610661029816, "train_loss_bc": 0.036006905138492584, "train_loss_llm": 0.0006705522537231445, "grad_norm": 0.023298438638448715, "global_step": 11089, "epoch": 308, "lr": 0.001} +{"train_loss": 0.04127509519457817, "train_loss_bc": 0.04127509519457817, "train_loss_llm": 0.0, "grad_norm": 0.036861903965473175, "global_step": 11090, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03523426875472069, "train_loss_bc": 0.035206086933612823, "train_loss_llm": 0.0028182566165924072, "grad_norm": 0.05055483803153038, "global_step": 11091, "epoch": 308, "lr": 0.001} +{"train_loss": 0.037179090082645416, "train_loss_bc": 0.037179090082645416, "train_loss_llm": 0.0, "grad_norm": 0.06588388234376907, "global_step": 11092, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03859433904290199, "train_loss_bc": 0.03859433904290199, "train_loss_llm": 0.0, "grad_norm": 0.023138973861932755, "global_step": 11093, "epoch": 308, "lr": 0.001} +{"train_loss": 0.0317571684718132, "train_loss_bc": 0.0317571684718132, "train_loss_llm": 0.0, "grad_norm": 0.03556946665048599, "global_step": 11094, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03770918771624565, "train_loss_bc": 0.037687573581933975, "train_loss_llm": 0.0021612942218780518, "grad_norm": 0.048869963735342026, "global_step": 11095, "epoch": 308, "lr": 0.001} +{"train_loss": 0.037831224501132965, "train_loss_bc": 0.037831224501132965, "train_loss_llm": 0.0, "grad_norm": 0.0668143555521965, "global_step": 11096, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03777725249528885, "train_loss_bc": 0.03776757791638374, "train_loss_llm": 0.0009672939777374268, "grad_norm": 0.020359190180897713, "global_step": 11097, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03629932552576065, "train_loss_bc": 0.03629932552576065, "train_loss_llm": 0.0, "grad_norm": 0.03380607068538666, "global_step": 11098, "epoch": 308, "lr": 0.001} +{"train_loss": 0.04003629833459854, "train_loss_bc": 0.04003120958805084, "train_loss_llm": 0.0005089342594146729, "grad_norm": 0.04000652953982353, "global_step": 11099, "epoch": 308, "lr": 0.001} +{"train_loss": 0.037212006747722626, "train_loss_bc": 0.037212006747722626, "train_loss_llm": 0.0, "grad_norm": 0.04957626760005951, "global_step": 11100, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03971695527434349, "train_loss_bc": 0.03971071168780327, "train_loss_llm": 0.000624239444732666, "grad_norm": 0.02182319201529026, "global_step": 11101, "epoch": 308, "lr": 0.001} +{"train_loss": 0.034459520131349564, "train_loss_bc": 0.034444715827703476, "train_loss_llm": 0.0014803409576416016, "grad_norm": 0.03188329190015793, "global_step": 11102, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03338327258825302, "train_loss_bc": 0.03338070958852768, "train_loss_llm": 0.00025641918182373047, "grad_norm": 0.04054617881774902, "global_step": 11103, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03987947851419449, "train_loss_bc": 0.03987342491745949, "train_loss_llm": 0.0006053149700164795, "grad_norm": 0.05510108172893524, "global_step": 11104, "epoch": 308, "lr": 0.001} +{"train_loss": 0.0298955999314785, "train_loss_bc": 0.0298955999314785, "train_loss_llm": 0.0, "grad_norm": 0.023653607815504074, "global_step": 11105, "epoch": 308, "lr": 0.001} +{"train_loss": 0.04049614816904068, "train_loss_bc": 0.04049614816904068, "train_loss_llm": 0.0, "grad_norm": 0.03917526453733444, "global_step": 11106, "epoch": 308, "lr": 0.001} +{"train_loss": 0.04311176389455795, "train_loss_bc": 0.04311176389455795, "train_loss_llm": 0.0, "grad_norm": 0.05066147819161415, "global_step": 11107, "epoch": 308, "lr": 0.001} +{"train_loss": 0.033063165843486786, "train_loss_bc": 0.033063165843486786, "train_loss_llm": 0.0, "grad_norm": 0.056049928069114685, "global_step": 11108, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03771214187145233, "train_loss_bc": 0.03770260512828827, "train_loss_llm": 0.0009537935256958008, "grad_norm": 0.024655643850564957, "global_step": 11109, "epoch": 308, "lr": 0.001} +{"train_loss": 0.041701316833496094, "train_loss_bc": 0.041701316833496094, "train_loss_llm": 0.0, "grad_norm": 0.04282616451382637, "global_step": 11110, "epoch": 308, "lr": 0.001} +{"train_loss": 0.04125090688467026, "train_loss_bc": 0.04125090688467026, "train_loss_llm": 0.0, "grad_norm": 0.058800481259822845, "global_step": 11111, "epoch": 308, "lr": 0.001} +{"train_loss": 0.04417238384485245, "train_loss_bc": 0.04417238384485245, "train_loss_llm": 0.0, "grad_norm": 0.07320784032344818, "global_step": 11112, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03524181991815567, "train_loss_bc": 0.03524181991815567, "train_loss_llm": 0.0, "grad_norm": 0.017332356423139572, "global_step": 11113, "epoch": 308, "lr": 0.001} +{"train_loss": 0.04072905704379082, "train_loss_bc": 0.04072905704379082, "train_loss_llm": 0.0, "grad_norm": 0.038426902145147324, "global_step": 11114, "epoch": 308, "lr": 0.001} +{"train_loss": 0.04266604408621788, "train_loss_bc": 0.04266604408621788, "train_loss_llm": 0.0, "grad_norm": 0.054684702306985855, "global_step": 11115, "epoch": 308, "lr": 0.001} +{"train_loss": 0.04283861443400383, "train_loss_bc": 0.04283861443400383, "train_loss_llm": 0.0, "grad_norm": 0.06977247446775436, "global_step": 11116, "epoch": 308, "lr": 0.001} +{"train_loss": 0.0352950245141983, "train_loss_bc": 0.0352950245141983, "train_loss_llm": 0.0, "grad_norm": 0.021748913452029228, "global_step": 11117, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03645486384630203, "train_loss_bc": 0.03645486384630203, "train_loss_llm": 0.0, "grad_norm": 0.030729983001947403, "global_step": 11118, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03456657752394676, "train_loss_bc": 0.03456578776240349, "train_loss_llm": 7.912516593933105e-05, "grad_norm": 0.04496607184410095, "global_step": 11119, "epoch": 308, "lr": 0.001} +{"train_loss": 0.04738784581422806, "train_loss_bc": 0.04738784581422806, "train_loss_llm": 0.0, "grad_norm": 0.05927782133221626, "global_step": 11120, "epoch": 308, "lr": 0.001} +{"train_loss": 0.03559219092130661, "train_loss_bc": 0.03559219092130661, "train_loss_llm": 0.0, "grad_norm": 0.01738668605685234, "global_step": 11121, "epoch": 308, "lr": 0.001} +{"train_loss": 0.031054871156811714, "train_loss_bc": 0.03102317824959755, "train_loss_llm": 0.003169327974319458, "grad_norm": 0.028675513342022896, "global_step": 11122, "epoch": 308, "lr": 0.001} +{"train_loss": 0.0379458152068158, "train_loss_bc": 0.03678838908672333, "train_loss_llm": 0.0, "grad_norm": 0.03611227869987488, "global_step": 11123, "epoch": 308, "lr": 0.001, "val_loss": 0.06255915760993958} +{"train_loss": 0.04078865796327591, "train_loss_bc": 0.04078865796327591, "train_loss_llm": 0.0, "grad_norm": 0.041342489421367645, "global_step": 11124, "epoch": 309, "lr": 0.001} +{"train_loss": 0.02786487527191639, "train_loss_bc": 0.02786487527191639, "train_loss_llm": 0.0, "grad_norm": 0.015708932653069496, "global_step": 11125, "epoch": 309, "lr": 0.001} +{"train_loss": 0.029719585552811623, "train_loss_bc": 0.029719585552811623, "train_loss_llm": 0.0, "grad_norm": 0.024357765913009644, "global_step": 11126, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03200668469071388, "train_loss_bc": 0.03200517222285271, "train_loss_llm": 0.0001512467861175537, "grad_norm": 0.03426460921764374, "global_step": 11127, "epoch": 309, "lr": 0.001} +{"train_loss": 0.04060510918498039, "train_loss_bc": 0.04060510918498039, "train_loss_llm": 0.0, "grad_norm": 0.041836097836494446, "global_step": 11128, "epoch": 309, "lr": 0.001} +{"train_loss": 0.041638705879449844, "train_loss_bc": 0.041638705879449844, "train_loss_llm": 0.0, "grad_norm": 0.017474116757512093, "global_step": 11129, "epoch": 309, "lr": 0.001} +{"train_loss": 0.035155244171619415, "train_loss_bc": 0.035155244171619415, "train_loss_llm": 0.0, "grad_norm": 0.028478115797042847, "global_step": 11130, "epoch": 309, "lr": 0.001} +{"train_loss": 0.036696139723062515, "train_loss_bc": 0.03669167682528496, "train_loss_llm": 0.0004461705684661865, "grad_norm": 0.03953556343913078, "global_step": 11131, "epoch": 309, "lr": 0.001} +{"train_loss": 0.042377036064863205, "train_loss_bc": 0.042377036064863205, "train_loss_llm": 0.0, "grad_norm": 0.05151662230491638, "global_step": 11132, "epoch": 309, "lr": 0.001} +{"train_loss": 0.0408855602145195, "train_loss_bc": 0.0408855602145195, "train_loss_llm": 0.0, "grad_norm": 0.02831587940454483, "global_step": 11133, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03461942821741104, "train_loss_bc": 0.03461942821741104, "train_loss_llm": 0.0, "grad_norm": 0.04498206079006195, "global_step": 11134, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03676402196288109, "train_loss_bc": 0.03675887733697891, "train_loss_llm": 0.000514298677444458, "grad_norm": 0.05604543536901474, "global_step": 11135, "epoch": 309, "lr": 0.001} +{"train_loss": 0.037571631371974945, "train_loss_bc": 0.037571631371974945, "train_loss_llm": 0.0, "grad_norm": 0.06416386365890503, "global_step": 11136, "epoch": 309, "lr": 0.001} +{"train_loss": 0.04146425798535347, "train_loss_bc": 0.04146425798535347, "train_loss_llm": 0.0, "grad_norm": 0.02802668884396553, "global_step": 11137, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03690234199166298, "train_loss_bc": 0.03690234199166298, "train_loss_llm": 0.0, "grad_norm": 0.04440655931830406, "global_step": 11138, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03595945984125137, "train_loss_bc": 0.03595945984125137, "train_loss_llm": 0.0, "grad_norm": 0.05575171113014221, "global_step": 11139, "epoch": 309, "lr": 0.001} +{"train_loss": 0.036790408194065094, "train_loss_bc": 0.036790408194065094, "train_loss_llm": 0.0, "grad_norm": 0.06605199724435806, "global_step": 11140, "epoch": 309, "lr": 0.001} +{"train_loss": 0.035550449043512344, "train_loss_bc": 0.035550449043512344, "train_loss_llm": 0.0, "grad_norm": 0.021679352968931198, "global_step": 11141, "epoch": 309, "lr": 0.001} +{"train_loss": 0.038883425295352936, "train_loss_bc": 0.0388764813542366, "train_loss_llm": 0.0006943047046661377, "grad_norm": 0.03391826152801514, "global_step": 11142, "epoch": 309, "lr": 0.001} +{"train_loss": 0.042526815086603165, "train_loss_bc": 0.042526740580797195, "train_loss_llm": 7.450580596923828e-06, "grad_norm": 0.05138684809207916, "global_step": 11143, "epoch": 309, "lr": 0.001} +{"train_loss": 0.04343818500638008, "train_loss_bc": 0.04343818500638008, "train_loss_llm": 0.0, "grad_norm": 0.06759779155254364, "global_step": 11144, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03279048204421997, "train_loss_bc": 0.03278512880206108, "train_loss_llm": 0.0005351901054382324, "grad_norm": 0.02324373833835125, "global_step": 11145, "epoch": 309, "lr": 0.001} +{"train_loss": 0.042046405375003815, "train_loss_bc": 0.042046405375003815, "train_loss_llm": 0.0, "grad_norm": 0.03268629312515259, "global_step": 11146, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03978757932782173, "train_loss_bc": 0.03978757932782173, "train_loss_llm": 0.0, "grad_norm": 0.03821258246898651, "global_step": 11147, "epoch": 309, "lr": 0.001} +{"train_loss": 0.030395064502954483, "train_loss_bc": 0.030395064502954483, "train_loss_llm": 0.0, "grad_norm": 0.042798347771167755, "global_step": 11148, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03913290798664093, "train_loss_bc": 0.03913290798664093, "train_loss_llm": 0.0, "grad_norm": 0.018219416961073875, "global_step": 11149, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03756292164325714, "train_loss_bc": 0.03756292164325714, "train_loss_llm": 0.0, "grad_norm": 0.029330043122172356, "global_step": 11150, "epoch": 309, "lr": 0.001} +{"train_loss": 0.04356035962700844, "train_loss_bc": 0.04356035962700844, "train_loss_llm": 0.0, "grad_norm": 0.03754306212067604, "global_step": 11151, "epoch": 309, "lr": 0.001} +{"train_loss": 0.04167013242840767, "train_loss_bc": 0.04166082292795181, "train_loss_llm": 0.0009308755397796631, "grad_norm": 0.05441560223698616, "global_step": 11152, "epoch": 309, "lr": 0.001} +{"train_loss": 0.04091997444629669, "train_loss_bc": 0.04091997444629669, "train_loss_llm": 0.0, "grad_norm": 0.022180762141942978, "global_step": 11153, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03364573419094086, "train_loss_bc": 0.03364105895161629, "train_loss_llm": 0.0004674196243286133, "grad_norm": 0.0313684456050396, "global_step": 11154, "epoch": 309, "lr": 0.001} +{"train_loss": 0.04621163383126259, "train_loss_bc": 0.04619969055056572, "train_loss_llm": 0.0011942684650421143, "grad_norm": 0.04927322268486023, "global_step": 11155, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03833070769906044, "train_loss_bc": 0.03833070769906044, "train_loss_llm": 0.0, "grad_norm": 0.061800871044397354, "global_step": 11156, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03612639009952545, "train_loss_bc": 0.03611459210515022, "train_loss_llm": 0.0011796653270721436, "grad_norm": 0.018794061616063118, "global_step": 11157, "epoch": 309, "lr": 0.001} +{"train_loss": 0.02933972328901291, "train_loss_bc": 0.029320452362298965, "train_loss_llm": 0.0019270479679107666, "grad_norm": 0.02965782955288887, "global_step": 11158, "epoch": 309, "lr": 0.001} +{"train_loss": 0.03781667838080062, "train_loss_bc": 0.04166548699140549, "train_loss_llm": 0.0006894767284393311, "grad_norm": 0.04121682420372963, "global_step": 11159, "epoch": 309, "lr": 0.001, "val_loss": 0.06241975724697113} +{"train_loss": 0.031247520819306374, "train_loss_bc": 0.031247520819306374, "train_loss_llm": 0.0, "grad_norm": 0.0548868291079998, "global_step": 11160, "epoch": 310, "lr": 0.001} +{"train_loss": 0.050113745033741, "train_loss_bc": 0.050113745033741, "train_loss_llm": 0.0, "grad_norm": 0.03313314914703369, "global_step": 11161, "epoch": 310, "lr": 0.001} +{"train_loss": 0.04690057411789894, "train_loss_bc": 0.04687100648880005, "train_loss_llm": 0.002956897020339966, "grad_norm": 0.04404078796505928, "global_step": 11162, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03828611597418785, "train_loss_bc": 0.03828611597418785, "train_loss_llm": 0.0, "grad_norm": 0.05171922221779823, "global_step": 11163, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03792046010494232, "train_loss_bc": 0.03791888803243637, "train_loss_llm": 0.0001571178436279297, "grad_norm": 0.06397230178117752, "global_step": 11164, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03573426976799965, "train_loss_bc": 0.035727035254240036, "train_loss_llm": 0.0007232725620269775, "grad_norm": 0.02085317298769951, "global_step": 11165, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03687049821019173, "train_loss_bc": 0.03686787188053131, "train_loss_llm": 0.0002626776695251465, "grad_norm": 0.038295358419418335, "global_step": 11166, "epoch": 310, "lr": 0.001} +{"train_loss": 0.038966622203588486, "train_loss_bc": 0.03896569088101387, "train_loss_llm": 9.328126907348633e-05, "grad_norm": 0.049040015786886215, "global_step": 11167, "epoch": 310, "lr": 0.001} +{"train_loss": 0.0339251346886158, "train_loss_bc": 0.0339251346886158, "train_loss_llm": 0.0, "grad_norm": 0.05911587178707123, "global_step": 11168, "epoch": 310, "lr": 0.001} +{"train_loss": 0.02986833266913891, "train_loss_bc": 0.02986833266913891, "train_loss_llm": 0.0, "grad_norm": 0.013087301515042782, "global_step": 11169, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03826580196619034, "train_loss_bc": 0.03826580196619034, "train_loss_llm": 0.0, "grad_norm": 0.025652946904301643, "global_step": 11170, "epoch": 310, "lr": 0.001} +{"train_loss": 0.037807125598192215, "train_loss_bc": 0.037807125598192215, "train_loss_llm": 0.0, "grad_norm": 0.037541888654232025, "global_step": 11171, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03369482606649399, "train_loss_bc": 0.03369482606649399, "train_loss_llm": 0.0, "grad_norm": 0.042642414569854736, "global_step": 11172, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03881935030221939, "train_loss_bc": 0.03881935030221939, "train_loss_llm": 0.0, "grad_norm": 0.0209275521337986, "global_step": 11173, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03990761935710907, "train_loss_bc": 0.03990761935710907, "train_loss_llm": 0.0, "grad_norm": 0.034895431250333786, "global_step": 11174, "epoch": 310, "lr": 0.001} +{"train_loss": 0.039516761898994446, "train_loss_bc": 0.03950349614024162, "train_loss_llm": 0.001326441764831543, "grad_norm": 0.04055001586675644, "global_step": 11175, "epoch": 310, "lr": 0.001} +{"train_loss": 0.039578359574079514, "train_loss_bc": 0.039552222937345505, "train_loss_llm": 0.0026138126850128174, "grad_norm": 0.045832935720682144, "global_step": 11176, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03530963510274887, "train_loss_bc": 0.035307127982378006, "train_loss_llm": 0.00025078654289245605, "grad_norm": 0.018915297463536263, "global_step": 11177, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03970802202820778, "train_loss_bc": 0.03970802202820778, "train_loss_llm": 0.0, "grad_norm": 0.03172391280531883, "global_step": 11178, "epoch": 310, "lr": 0.001} +{"train_loss": 0.036285024136304855, "train_loss_bc": 0.03628254309296608, "train_loss_llm": 0.0002479851245880127, "grad_norm": 0.03770409896969795, "global_step": 11179, "epoch": 310, "lr": 0.001} +{"train_loss": 0.022648267447948456, "train_loss_bc": 0.02262740582227707, "train_loss_llm": 0.002086132764816284, "grad_norm": 0.04098197817802429, "global_step": 11180, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03756985813379288, "train_loss_bc": 0.037565797567367554, "train_loss_llm": 0.00040608644485473633, "grad_norm": 0.022397339344024658, "global_step": 11181, "epoch": 310, "lr": 0.001} +{"train_loss": 0.039809130132198334, "train_loss_bc": 0.03980446234345436, "train_loss_llm": 0.0004667043685913086, "grad_norm": 0.0407063364982605, "global_step": 11182, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03942772373557091, "train_loss_bc": 0.03942772373557091, "train_loss_llm": 0.0, "grad_norm": 0.05075082927942276, "global_step": 11183, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03918841481208801, "train_loss_bc": 0.03918275609612465, "train_loss_llm": 0.0005659162998199463, "grad_norm": 0.06567665189504623, "global_step": 11184, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03580082580447197, "train_loss_bc": 0.03580082580447197, "train_loss_llm": 0.0, "grad_norm": 0.021390458568930626, "global_step": 11185, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03791821375489235, "train_loss_bc": 0.03791821375489235, "train_loss_llm": 0.0, "grad_norm": 0.03224875032901764, "global_step": 11186, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03320373594760895, "train_loss_bc": 0.033199720084667206, "train_loss_llm": 0.00040140748023986816, "grad_norm": 0.04541635140776634, "global_step": 11187, "epoch": 310, "lr": 0.001} +{"train_loss": 0.038684338331222534, "train_loss_bc": 0.03867250308394432, "train_loss_llm": 0.0011834800243377686, "grad_norm": 0.06160813197493553, "global_step": 11188, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03615012764930725, "train_loss_bc": 0.03615012764930725, "train_loss_llm": 0.0, "grad_norm": 0.020944029092788696, "global_step": 11189, "epoch": 310, "lr": 0.001} +{"train_loss": 0.043312814086675644, "train_loss_bc": 0.043312814086675644, "train_loss_llm": 0.0, "grad_norm": 0.041554294526576996, "global_step": 11190, "epoch": 310, "lr": 0.001} +{"train_loss": 0.037588439881801605, "train_loss_bc": 0.03758201748132706, "train_loss_llm": 0.0006424188613891602, "grad_norm": 0.05435526743531227, "global_step": 11191, "epoch": 310, "lr": 0.001} +{"train_loss": 0.0384710393846035, "train_loss_bc": 0.0384710393846035, "train_loss_llm": 0.0, "grad_norm": 0.06856632977724075, "global_step": 11192, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03444584459066391, "train_loss_bc": 0.03444584459066391, "train_loss_llm": 0.0, "grad_norm": 0.019523847848176956, "global_step": 11193, "epoch": 310, "lr": 0.001} +{"train_loss": 0.04556037858128548, "train_loss_bc": 0.04556037858128548, "train_loss_llm": 0.0, "grad_norm": 0.03252890333533287, "global_step": 11194, "epoch": 310, "lr": 0.001} +{"train_loss": 0.03780731724368201, "train_loss_bc": 0.042558468878269196, "train_loss_llm": 0.0, "grad_norm": 0.0468989796936512, "global_step": 11195, "epoch": 310, "lr": 0.001, "train/cumulative_reward": 57.24609969311602, "train/mean_score": 14.87609436904051, "train/success_rate": 0.4, "test/cumulative_reward": 44.91603311379672, "test/mean_score": 7.674661826415937, "test/success_rate": 0.2, "val_loss": 0.062411367893218994, "train_action_mse_error": 0.04055476188659668} +{"train_loss": 0.03535541146993637, "train_loss_bc": 0.035342250019311905, "train_loss_llm": 0.0013160407543182373, "grad_norm": 0.05151123180985451, "global_step": 11196, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03542912378907204, "train_loss_bc": 0.035408053547143936, "train_loss_llm": 0.0021071135997772217, "grad_norm": 0.018975479528307915, "global_step": 11197, "epoch": 311, "lr": 0.001} +{"train_loss": 0.033307235687971115, "train_loss_bc": 0.033307235687971115, "train_loss_llm": 0.0, "grad_norm": 0.03144710883498192, "global_step": 11198, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03676493093371391, "train_loss_bc": 0.03676493093371391, "train_loss_llm": 0.0, "grad_norm": 0.040832456201314926, "global_step": 11199, "epoch": 311, "lr": 0.001} +{"train_loss": 0.039870575070381165, "train_loss_bc": 0.039870575070381165, "train_loss_llm": 0.0, "grad_norm": 0.04925541579723358, "global_step": 11200, "epoch": 311, "lr": 0.001} +{"train_loss": 0.04001157730817795, "train_loss_bc": 0.04001157730817795, "train_loss_llm": 0.0, "grad_norm": 0.02332965098321438, "global_step": 11201, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03592071682214737, "train_loss_bc": 0.03592071682214737, "train_loss_llm": 0.0, "grad_norm": 0.03518150374293327, "global_step": 11202, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03897492215037346, "train_loss_bc": 0.03897492215037346, "train_loss_llm": 0.0, "grad_norm": 0.03863208368420601, "global_step": 11203, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03675959259271622, "train_loss_bc": 0.03675959259271622, "train_loss_llm": 0.0, "grad_norm": 0.0439167283475399, "global_step": 11204, "epoch": 311, "lr": 0.001} +{"train_loss": 0.0405929796397686, "train_loss_bc": 0.04057985916733742, "train_loss_llm": 0.0013121068477630615, "grad_norm": 0.020527657121419907, "global_step": 11205, "epoch": 311, "lr": 0.001} +{"train_loss": 0.040106967091560364, "train_loss_bc": 0.040106967091560364, "train_loss_llm": 0.0, "grad_norm": 0.03715011477470398, "global_step": 11206, "epoch": 311, "lr": 0.001} +{"train_loss": 0.043823204934597015, "train_loss_bc": 0.043823204934597015, "train_loss_llm": 0.0, "grad_norm": 0.04788517579436302, "global_step": 11207, "epoch": 311, "lr": 0.001} +{"train_loss": 0.0391300767660141, "train_loss_bc": 0.039122533053159714, "train_loss_llm": 0.0007544755935668945, "grad_norm": 0.05845160409808159, "global_step": 11208, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03938770666718483, "train_loss_bc": 0.03938770666718483, "train_loss_llm": 0.0, "grad_norm": 0.021037766709923744, "global_step": 11209, "epoch": 311, "lr": 0.001} +{"train_loss": 0.04247065633535385, "train_loss_bc": 0.04247065633535385, "train_loss_llm": 0.0, "grad_norm": 0.030460452660918236, "global_step": 11210, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03987007215619087, "train_loss_bc": 0.03987007215619087, "train_loss_llm": 0.0, "grad_norm": 0.045664072036743164, "global_step": 11211, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03806237876415253, "train_loss_bc": 0.03806237876415253, "train_loss_llm": 0.0, "grad_norm": 0.052455853670835495, "global_step": 11212, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03826337680220604, "train_loss_bc": 0.03826337680220604, "train_loss_llm": 0.0, "grad_norm": 0.019411036744713783, "global_step": 11213, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03260434418916702, "train_loss_bc": 0.03260434418916702, "train_loss_llm": 0.0, "grad_norm": 0.032236505299806595, "global_step": 11214, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03503570705652237, "train_loss_bc": 0.03503570705652237, "train_loss_llm": 0.0, "grad_norm": 0.043574463576078415, "global_step": 11215, "epoch": 311, "lr": 0.001} +{"train_loss": 0.039930932223796844, "train_loss_bc": 0.039930932223796844, "train_loss_llm": 0.0, "grad_norm": 0.057817719876766205, "global_step": 11216, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03985264152288437, "train_loss_bc": 0.03984680399298668, "train_loss_llm": 0.0005837976932525635, "grad_norm": 0.022724488750100136, "global_step": 11217, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03918914124369621, "train_loss_bc": 0.03918914124369621, "train_loss_llm": 0.0, "grad_norm": 0.029041528701782227, "global_step": 11218, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03729905188083649, "train_loss_bc": 0.03728749230504036, "train_loss_llm": 0.001155942678451538, "grad_norm": 0.03660472482442856, "global_step": 11219, "epoch": 311, "lr": 0.001} +{"train_loss": 0.031044675037264824, "train_loss_bc": 0.03104342333972454, "train_loss_llm": 0.00012513995170593262, "grad_norm": 0.04961729794740677, "global_step": 11220, "epoch": 311, "lr": 0.001} +{"train_loss": 0.037232719361782074, "train_loss_bc": 0.037232719361782074, "train_loss_llm": 0.0, "grad_norm": 0.019312318414449692, "global_step": 11221, "epoch": 311, "lr": 0.001} +{"train_loss": 0.04441748186945915, "train_loss_bc": 0.04441748186945915, "train_loss_llm": 0.0, "grad_norm": 0.03666408732533455, "global_step": 11222, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03656230866909027, "train_loss_bc": 0.03656230866909027, "train_loss_llm": 0.0, "grad_norm": 0.051911626011133194, "global_step": 11223, "epoch": 311, "lr": 0.001} +{"train_loss": 0.030814219266176224, "train_loss_bc": 0.030814219266176224, "train_loss_llm": 0.0, "grad_norm": 0.06898055970668793, "global_step": 11224, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03304928541183472, "train_loss_bc": 0.03304928541183472, "train_loss_llm": 0.0, "grad_norm": 0.021870635449886322, "global_step": 11225, "epoch": 311, "lr": 0.001} +{"train_loss": 0.032876912504434586, "train_loss_bc": 0.03286660462617874, "train_loss_llm": 0.0010307133197784424, "grad_norm": 0.03658408299088478, "global_step": 11226, "epoch": 311, "lr": 0.001} +{"train_loss": 0.038237281143665314, "train_loss_bc": 0.038237281143665314, "train_loss_llm": 0.0, "grad_norm": 0.04653700441122055, "global_step": 11227, "epoch": 311, "lr": 0.001} +{"train_loss": 0.035051919519901276, "train_loss_bc": 0.035051919519901276, "train_loss_llm": 0.0, "grad_norm": 0.05779612436890602, "global_step": 11228, "epoch": 311, "lr": 0.001} +{"train_loss": 0.038956571370363235, "train_loss_bc": 0.038956571370363235, "train_loss_llm": 0.0, "grad_norm": 0.017748335376381874, "global_step": 11229, "epoch": 311, "lr": 0.001} +{"train_loss": 0.039922162890434265, "train_loss_bc": 0.039922162890434265, "train_loss_llm": 0.0, "grad_norm": 0.02641390450298786, "global_step": 11230, "epoch": 311, "lr": 0.001} +{"train_loss": 0.03771919607081347, "train_loss_bc": 0.0417121984064579, "train_loss_llm": 0.0, "grad_norm": 0.04611331596970558, "global_step": 11231, "epoch": 311, "lr": 0.001, "val_loss": 0.0630471333861351} +{"train_loss": 0.03719433769583702, "train_loss_bc": 0.03719433769583702, "train_loss_llm": 0.0, "grad_norm": 0.05109329894185066, "global_step": 11232, "epoch": 312, "lr": 0.001} +{"train_loss": 0.042331479489803314, "train_loss_bc": 0.042331479489803314, "train_loss_llm": 0.0, "grad_norm": 0.019937964156270027, "global_step": 11233, "epoch": 312, "lr": 0.001} +{"train_loss": 0.038371492177248, "train_loss_bc": 0.038371492177248, "train_loss_llm": 0.0, "grad_norm": 0.03588842228055, "global_step": 11234, "epoch": 312, "lr": 0.001} +{"train_loss": 0.04139472171664238, "train_loss_bc": 0.04139472171664238, "train_loss_llm": 0.0, "grad_norm": 0.05362621322274208, "global_step": 11235, "epoch": 312, "lr": 0.001} +{"train_loss": 0.037989601492881775, "train_loss_bc": 0.03796955198049545, "train_loss_llm": 0.0020051300525665283, "grad_norm": 0.0697198212146759, "global_step": 11236, "epoch": 312, "lr": 0.001} +{"train_loss": 0.04124388098716736, "train_loss_bc": 0.04124388098716736, "train_loss_llm": 0.0, "grad_norm": 0.021156219765543938, "global_step": 11237, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03335092216730118, "train_loss_bc": 0.03335092216730118, "train_loss_llm": 0.0, "grad_norm": 0.03050830215215683, "global_step": 11238, "epoch": 312, "lr": 0.001} +{"train_loss": 0.0334678515791893, "train_loss_bc": 0.0334678515791893, "train_loss_llm": 0.0, "grad_norm": 0.03649506717920303, "global_step": 11239, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03788322955369949, "train_loss_bc": 0.03788322955369949, "train_loss_llm": 0.0, "grad_norm": 0.04106775298714638, "global_step": 11240, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03839202597737312, "train_loss_bc": 0.03839202597737312, "train_loss_llm": 0.0, "grad_norm": 0.02129879780113697, "global_step": 11241, "epoch": 312, "lr": 0.001} +{"train_loss": 0.04165630415081978, "train_loss_bc": 0.04165630415081978, "train_loss_llm": 0.0, "grad_norm": 0.030033603310585022, "global_step": 11242, "epoch": 312, "lr": 0.001} +{"train_loss": 0.0292836744338274, "train_loss_bc": 0.029277324676513672, "train_loss_llm": 0.0006349384784698486, "grad_norm": 0.035574451088905334, "global_step": 11243, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03658680245280266, "train_loss_bc": 0.03658680245280266, "train_loss_llm": 0.0, "grad_norm": 0.04258907958865166, "global_step": 11244, "epoch": 312, "lr": 0.001} +{"train_loss": 0.037089601159095764, "train_loss_bc": 0.03707398101687431, "train_loss_llm": 0.0015619993209838867, "grad_norm": 0.02035989984869957, "global_step": 11245, "epoch": 312, "lr": 0.001} +{"train_loss": 0.045908208936452866, "train_loss_bc": 0.045908208936452866, "train_loss_llm": 0.0, "grad_norm": 0.027647893875837326, "global_step": 11246, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03432071581482887, "train_loss_bc": 0.034309618175029755, "train_loss_llm": 0.0011097490787506104, "grad_norm": 0.03639160841703415, "global_step": 11247, "epoch": 312, "lr": 0.001} +{"train_loss": 0.0384240560233593, "train_loss_bc": 0.03841651231050491, "train_loss_llm": 0.0007544755935668945, "grad_norm": 0.04591454938054085, "global_step": 11248, "epoch": 312, "lr": 0.001} +{"train_loss": 0.0336851142346859, "train_loss_bc": 0.03367524966597557, "train_loss_llm": 0.0009863078594207764, "grad_norm": 0.01869022659957409, "global_step": 11249, "epoch": 312, "lr": 0.001} +{"train_loss": 0.031808894127607346, "train_loss_bc": 0.031808894127607346, "train_loss_llm": 0.0, "grad_norm": 0.03086056374013424, "global_step": 11250, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03223392739892006, "train_loss_bc": 0.03223392739892006, "train_loss_llm": 0.0, "grad_norm": 0.04012514278292656, "global_step": 11251, "epoch": 312, "lr": 0.001} +{"train_loss": 0.036646101623773575, "train_loss_bc": 0.036646101623773575, "train_loss_llm": 0.0, "grad_norm": 0.0446009486913681, "global_step": 11252, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03218555450439453, "train_loss_bc": 0.032161012291908264, "train_loss_llm": 0.002454042434692383, "grad_norm": 0.020562414079904556, "global_step": 11253, "epoch": 312, "lr": 0.001} +{"train_loss": 0.034575674682855606, "train_loss_bc": 0.034575674682855606, "train_loss_llm": 0.0, "grad_norm": 0.03632250055670738, "global_step": 11254, "epoch": 312, "lr": 0.001} +{"train_loss": 0.04018237814307213, "train_loss_bc": 0.040182068943977356, "train_loss_llm": 3.081560134887695e-05, "grad_norm": 0.0540376752614975, "global_step": 11255, "epoch": 312, "lr": 0.001} +{"train_loss": 0.04025861620903015, "train_loss_bc": 0.04025861620903015, "train_loss_llm": 0.0, "grad_norm": 0.07195486128330231, "global_step": 11256, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03761560469865799, "train_loss_bc": 0.03759711608290672, "train_loss_llm": 0.0018489360809326172, "grad_norm": 0.019193511456251144, "global_step": 11257, "epoch": 312, "lr": 0.001} +{"train_loss": 0.036162182688713074, "train_loss_bc": 0.036162182688713074, "train_loss_llm": 0.0, "grad_norm": 0.031117558479309082, "global_step": 11258, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03810855746269226, "train_loss_bc": 0.03810855746269226, "train_loss_llm": 0.0, "grad_norm": 0.04421099275350571, "global_step": 11259, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03795412927865982, "train_loss_bc": 0.03795412927865982, "train_loss_llm": 0.0, "grad_norm": 0.05188478156924248, "global_step": 11260, "epoch": 312, "lr": 0.001} +{"train_loss": 0.04429060220718384, "train_loss_bc": 0.04429060220718384, "train_loss_llm": 0.0, "grad_norm": 0.020559947937726974, "global_step": 11261, "epoch": 312, "lr": 0.001} +{"train_loss": 0.04587918519973755, "train_loss_bc": 0.04587918519973755, "train_loss_llm": 0.0, "grad_norm": 0.03652831166982651, "global_step": 11262, "epoch": 312, "lr": 0.001} +{"train_loss": 0.04194341599941254, "train_loss_bc": 0.04194341599941254, "train_loss_llm": 0.0, "grad_norm": 0.04610038176178932, "global_step": 11263, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03478856384754181, "train_loss_bc": 0.03478009253740311, "train_loss_llm": 0.0008469820022583008, "grad_norm": 0.0667157769203186, "global_step": 11264, "epoch": 312, "lr": 0.001} +{"train_loss": 0.04104389622807503, "train_loss_bc": 0.04102780297398567, "train_loss_llm": 0.0016091763973236084, "grad_norm": 0.024803493171930313, "global_step": 11265, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03155330568552017, "train_loss_bc": 0.03152575343847275, "train_loss_llm": 0.002755194902420044, "grad_norm": 0.03491250425577164, "global_step": 11266, "epoch": 312, "lr": 0.001} +{"train_loss": 0.03762477817427781, "train_loss_bc": 0.03868245705962181, "train_loss_llm": 0.0004947483539581299, "grad_norm": 0.0455542728304863, "global_step": 11267, "epoch": 312, "lr": 0.001, "val_loss": 0.062415529042482376} +{"train_loss": 0.03950360789895058, "train_loss_bc": 0.03950360789895058, "train_loss_llm": 0.0, "grad_norm": 0.05870409682393074, "global_step": 11268, "epoch": 313, "lr": 0.001} +{"train_loss": 0.0458611398935318, "train_loss_bc": 0.0458611398935318, "train_loss_llm": 0.0, "grad_norm": 0.025316227227449417, "global_step": 11269, "epoch": 313, "lr": 0.001} +{"train_loss": 0.039258915930986404, "train_loss_bc": 0.03925264999270439, "train_loss_llm": 0.0006266534328460693, "grad_norm": 0.034460365772247314, "global_step": 11270, "epoch": 313, "lr": 0.001} +{"train_loss": 0.04271014407277107, "train_loss_bc": 0.04270713031291962, "train_loss_llm": 0.0003013312816619873, "grad_norm": 0.03859498351812363, "global_step": 11271, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03865443915128708, "train_loss_bc": 0.03865443915128708, "train_loss_llm": 0.0, "grad_norm": 0.05410385876893997, "global_step": 11272, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03407231718301773, "train_loss_bc": 0.03404470533132553, "train_loss_llm": 0.0027611851692199707, "grad_norm": 0.01868145726621151, "global_step": 11273, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03733007237315178, "train_loss_bc": 0.03733007237315178, "train_loss_llm": 0.0, "grad_norm": 0.03300320729613304, "global_step": 11274, "epoch": 313, "lr": 0.001} +{"train_loss": 0.041827909648418427, "train_loss_bc": 0.041827909648418427, "train_loss_llm": 0.0, "grad_norm": 0.049180302768945694, "global_step": 11275, "epoch": 313, "lr": 0.001} +{"train_loss": 0.036541152745485306, "train_loss_bc": 0.036541152745485306, "train_loss_llm": 0.0, "grad_norm": 0.0577729232609272, "global_step": 11276, "epoch": 313, "lr": 0.001} +{"train_loss": 0.030495967715978622, "train_loss_bc": 0.030493054538965225, "train_loss_llm": 0.0002913177013397217, "grad_norm": 0.022902904078364372, "global_step": 11277, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03308049589395523, "train_loss_bc": 0.03308049589395523, "train_loss_llm": 0.0, "grad_norm": 0.042117103934288025, "global_step": 11278, "epoch": 313, "lr": 0.001} +{"train_loss": 0.04296629875898361, "train_loss_bc": 0.04296629875898361, "train_loss_llm": 0.0, "grad_norm": 0.06448758393526077, "global_step": 11279, "epoch": 313, "lr": 0.001} +{"train_loss": 0.04294198006391525, "train_loss_bc": 0.04294198006391525, "train_loss_llm": 0.0, "grad_norm": 0.08121837675571442, "global_step": 11280, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03800122067332268, "train_loss_bc": 0.03800122067332268, "train_loss_llm": 0.0, "grad_norm": 0.018838362768292427, "global_step": 11281, "epoch": 313, "lr": 0.001} +{"train_loss": 0.04145980253815651, "train_loss_bc": 0.04145980253815651, "train_loss_llm": 0.0, "grad_norm": 0.03432577848434448, "global_step": 11282, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03781156614422798, "train_loss_bc": 0.037811558693647385, "train_loss_llm": 6.854534149169922e-07, "grad_norm": 0.03810051828622818, "global_step": 11283, "epoch": 313, "lr": 0.001} +{"train_loss": 0.039056044071912766, "train_loss_bc": 0.039056044071912766, "train_loss_llm": 0.0, "grad_norm": 0.05331779271364212, "global_step": 11284, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03298380970954895, "train_loss_bc": 0.03298380970954895, "train_loss_llm": 0.0, "grad_norm": 0.021361475810408592, "global_step": 11285, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03521839529275894, "train_loss_bc": 0.035217009484767914, "train_loss_llm": 0.0001385509967803955, "grad_norm": 0.03200288861989975, "global_step": 11286, "epoch": 313, "lr": 0.001} +{"train_loss": 0.04079120233654976, "train_loss_bc": 0.04079120233654976, "train_loss_llm": 0.0, "grad_norm": 0.03948932886123657, "global_step": 11287, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03989073261618614, "train_loss_bc": 0.03989073261618614, "train_loss_llm": 0.0, "grad_norm": 0.05031319707632065, "global_step": 11288, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03179696574807167, "train_loss_bc": 0.03179696574807167, "train_loss_llm": 0.0, "grad_norm": 0.020824439823627472, "global_step": 11289, "epoch": 313, "lr": 0.001} +{"train_loss": 0.037236135452985764, "train_loss_bc": 0.037236135452985764, "train_loss_llm": 0.0, "grad_norm": 0.030727094039320946, "global_step": 11290, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03576101362705231, "train_loss_bc": 0.035748448222875595, "train_loss_llm": 0.0012564659118652344, "grad_norm": 0.041358545422554016, "global_step": 11291, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03202042728662491, "train_loss_bc": 0.03202042728662491, "train_loss_llm": 0.0, "grad_norm": 0.042514920234680176, "global_step": 11292, "epoch": 313, "lr": 0.001} +{"train_loss": 0.039425838738679886, "train_loss_bc": 0.039411380887031555, "train_loss_llm": 0.001445680856704712, "grad_norm": 0.019681621342897415, "global_step": 11293, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03841128572821617, "train_loss_bc": 0.03841128572821617, "train_loss_llm": 0.0, "grad_norm": 0.037618014961481094, "global_step": 11294, "epoch": 313, "lr": 0.001} +{"train_loss": 0.031981002539396286, "train_loss_bc": 0.03196472302079201, "train_loss_llm": 0.0016279816627502441, "grad_norm": 0.045919034630060196, "global_step": 11295, "epoch": 313, "lr": 0.001} +{"train_loss": 0.031122881919145584, "train_loss_bc": 0.031117958948016167, "train_loss_llm": 0.0004922449588775635, "grad_norm": 0.05722536891698837, "global_step": 11296, "epoch": 313, "lr": 0.001} +{"train_loss": 0.042224422097206116, "train_loss_bc": 0.0422147773206234, "train_loss_llm": 0.0009644627571105957, "grad_norm": 0.02574939839541912, "global_step": 11297, "epoch": 313, "lr": 0.001} +{"train_loss": 0.039033789187669754, "train_loss_bc": 0.039033789187669754, "train_loss_llm": 0.0, "grad_norm": 0.041991252452135086, "global_step": 11298, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03799586370587349, "train_loss_bc": 0.03799565136432648, "train_loss_llm": 2.1249055862426758e-05, "grad_norm": 0.05285540968179703, "global_step": 11299, "epoch": 313, "lr": 0.001} +{"train_loss": 0.04011373966932297, "train_loss_bc": 0.04008760303258896, "train_loss_llm": 0.0026137828826904297, "grad_norm": 0.07272686809301376, "global_step": 11300, "epoch": 313, "lr": 0.001} +{"train_loss": 0.038717981427907944, "train_loss_bc": 0.038717981427907944, "train_loss_llm": 0.0, "grad_norm": 0.024932190775871277, "global_step": 11301, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03772268444299698, "train_loss_bc": 0.03772268444299698, "train_loss_llm": 0.0, "grad_norm": 0.03267662972211838, "global_step": 11302, "epoch": 313, "lr": 0.001} +{"train_loss": 0.03760653594508767, "train_loss_bc": 0.02981404773890972, "train_loss_llm": 0.0, "grad_norm": 0.04792211949825287, "global_step": 11303, "epoch": 313, "lr": 0.001, "val_loss": 0.0625157505273819} +{"train_loss": 0.03783659636974335, "train_loss_bc": 0.03783659636974335, "train_loss_llm": 0.0, "grad_norm": 0.06281495839357376, "global_step": 11304, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03528599441051483, "train_loss_bc": 0.03528599441051483, "train_loss_llm": 0.0, "grad_norm": 0.019358502700924873, "global_step": 11305, "epoch": 314, "lr": 0.001} +{"train_loss": 0.031230177730321884, "train_loss_bc": 0.031230177730321884, "train_loss_llm": 0.0, "grad_norm": 0.028295427560806274, "global_step": 11306, "epoch": 314, "lr": 0.001} +{"train_loss": 0.04077646881341934, "train_loss_bc": 0.04077646881341934, "train_loss_llm": 0.0, "grad_norm": 0.0424547977745533, "global_step": 11307, "epoch": 314, "lr": 0.001} +{"train_loss": 0.04007052630186081, "train_loss_bc": 0.04006640240550041, "train_loss_llm": 0.00041225552558898926, "grad_norm": 0.05423789471387863, "global_step": 11308, "epoch": 314, "lr": 0.001} +{"train_loss": 0.04040909558534622, "train_loss_bc": 0.04040802642703056, "train_loss_llm": 0.0001068115234375, "grad_norm": 0.021992266178131104, "global_step": 11309, "epoch": 314, "lr": 0.001} +{"train_loss": 0.037160810083150864, "train_loss_bc": 0.037153080105781555, "train_loss_llm": 0.0007729828357696533, "grad_norm": 0.03439247980713844, "global_step": 11310, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03921643644571304, "train_loss_bc": 0.03921643644571304, "train_loss_llm": 0.0, "grad_norm": 0.04428677260875702, "global_step": 11311, "epoch": 314, "lr": 0.001} +{"train_loss": 0.04234078899025917, "train_loss_bc": 0.04234078899025917, "train_loss_llm": 0.0, "grad_norm": 0.059071242809295654, "global_step": 11312, "epoch": 314, "lr": 0.001} +{"train_loss": 0.04116593673825264, "train_loss_bc": 0.04116593673825264, "train_loss_llm": 0.0, "grad_norm": 0.03310719132423401, "global_step": 11313, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03776305913925171, "train_loss_bc": 0.03774723410606384, "train_loss_llm": 0.0015823543071746826, "grad_norm": 0.04546981677412987, "global_step": 11314, "epoch": 314, "lr": 0.001} +{"train_loss": 0.044235896319150925, "train_loss_bc": 0.044235773384571075, "train_loss_llm": 1.2218952178955078e-05, "grad_norm": 0.06010279059410095, "global_step": 11315, "epoch": 314, "lr": 0.001} +{"train_loss": 0.04414596036076546, "train_loss_bc": 0.04414596036076546, "train_loss_llm": 0.0, "grad_norm": 0.07301577180624008, "global_step": 11316, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03663294389843941, "train_loss_bc": 0.03663294389843941, "train_loss_llm": 0.0, "grad_norm": 0.021432340145111084, "global_step": 11317, "epoch": 314, "lr": 0.001} +{"train_loss": 0.039206285029649734, "train_loss_bc": 0.03920046612620354, "train_loss_llm": 0.0005818307399749756, "grad_norm": 0.037455976009368896, "global_step": 11318, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03594452515244484, "train_loss_bc": 0.035934027284383774, "train_loss_llm": 0.0010496079921722412, "grad_norm": 0.049531448632478714, "global_step": 11319, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03245959430932999, "train_loss_bc": 0.03245959430932999, "train_loss_llm": 0.0, "grad_norm": 0.058559369295835495, "global_step": 11320, "epoch": 314, "lr": 0.001} +{"train_loss": 0.039809245616197586, "train_loss_bc": 0.039809245616197586, "train_loss_llm": 0.0, "grad_norm": 0.020056815817952156, "global_step": 11321, "epoch": 314, "lr": 0.001} +{"train_loss": 0.036584410816431046, "train_loss_bc": 0.036584410816431046, "train_loss_llm": 0.0, "grad_norm": 0.031688667833805084, "global_step": 11322, "epoch": 314, "lr": 0.001} +{"train_loss": 0.0374305285513401, "train_loss_bc": 0.0374305285513401, "train_loss_llm": 0.0, "grad_norm": 0.0452994704246521, "global_step": 11323, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03837943822145462, "train_loss_bc": 0.03837943822145462, "train_loss_llm": 0.0, "grad_norm": 0.058733414858579636, "global_step": 11324, "epoch": 314, "lr": 0.001} +{"train_loss": 0.04049938917160034, "train_loss_bc": 0.04049938917160034, "train_loss_llm": 0.0, "grad_norm": 0.01891792006790638, "global_step": 11325, "epoch": 314, "lr": 0.001} +{"train_loss": 0.0377500094473362, "train_loss_bc": 0.0377500094473362, "train_loss_llm": 0.0, "grad_norm": 0.025438295677304268, "global_step": 11326, "epoch": 314, "lr": 0.001} +{"train_loss": 0.04237539693713188, "train_loss_bc": 0.0423751026391983, "train_loss_llm": 2.9593706130981445e-05, "grad_norm": 0.036318618804216385, "global_step": 11327, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03327121213078499, "train_loss_bc": 0.03327121213078499, "train_loss_llm": 0.0, "grad_norm": 0.0388239286839962, "global_step": 11328, "epoch": 314, "lr": 0.001} +{"train_loss": 0.0364496186375618, "train_loss_bc": 0.036449264734983444, "train_loss_llm": 3.555417060852051e-05, "grad_norm": 0.019834600389003754, "global_step": 11329, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03527780622243881, "train_loss_bc": 0.035272616893053055, "train_loss_llm": 0.0005187690258026123, "grad_norm": 0.027307800948619843, "global_step": 11330, "epoch": 314, "lr": 0.001} +{"train_loss": 0.031279489398002625, "train_loss_bc": 0.031279489398002625, "train_loss_llm": 0.0, "grad_norm": 0.03530516102910042, "global_step": 11331, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03442808985710144, "train_loss_bc": 0.03442808985710144, "train_loss_llm": 0.0, "grad_norm": 0.04794223606586456, "global_step": 11332, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03168638423085213, "train_loss_bc": 0.03168638423085213, "train_loss_llm": 0.0, "grad_norm": 0.026596857234835625, "global_step": 11333, "epoch": 314, "lr": 0.001} +{"train_loss": 0.0388437882065773, "train_loss_bc": 0.0388437882065773, "train_loss_llm": 0.0, "grad_norm": 0.03625287488102913, "global_step": 11334, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03644363954663277, "train_loss_bc": 0.03644363954663277, "train_loss_llm": 0.0, "grad_norm": 0.050876542925834656, "global_step": 11335, "epoch": 314, "lr": 0.001} +{"train_loss": 0.033304788172245026, "train_loss_bc": 0.033295098692178726, "train_loss_llm": 0.0009690523147583008, "grad_norm": 0.05675259977579117, "global_step": 11336, "epoch": 314, "lr": 0.001} +{"train_loss": 0.040507640689611435, "train_loss_bc": 0.04050615057349205, "train_loss_llm": 0.0001488327980041504, "grad_norm": 0.022241229191422462, "global_step": 11337, "epoch": 314, "lr": 0.001} +{"train_loss": 0.04032805562019348, "train_loss_bc": 0.04032805562019348, "train_loss_llm": 0.0, "grad_norm": 0.03651755303144455, "global_step": 11338, "epoch": 314, "lr": 0.001} +{"train_loss": 0.03757868127690421, "train_loss_bc": 0.032292697578668594, "train_loss_llm": 0.0009800493717193604, "grad_norm": 0.04994633421301842, "global_step": 11339, "epoch": 314, "lr": 0.001, "val_loss": 0.06334971636533737} +{"train_loss": 0.03805767372250557, "train_loss_bc": 0.03805727884173393, "train_loss_llm": 3.936886787414551e-05, "grad_norm": 0.06579883396625519, "global_step": 11340, "epoch": 315, "lr": 0.001} +{"train_loss": 0.035286832600831985, "train_loss_bc": 0.035286832600831985, "train_loss_llm": 0.0, "grad_norm": 0.02173888124525547, "global_step": 11341, "epoch": 315, "lr": 0.001} +{"train_loss": 0.0331563875079155, "train_loss_bc": 0.0331563875079155, "train_loss_llm": 0.0, "grad_norm": 0.03417286276817322, "global_step": 11342, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03827940300107002, "train_loss_bc": 0.03827940300107002, "train_loss_llm": 0.0, "grad_norm": 0.051184143871068954, "global_step": 11343, "epoch": 315, "lr": 0.001} +{"train_loss": 0.032995738089084625, "train_loss_bc": 0.032995738089084625, "train_loss_llm": 0.0, "grad_norm": 0.060008905827999115, "global_step": 11344, "epoch": 315, "lr": 0.001} +{"train_loss": 0.02976389043033123, "train_loss_bc": 0.02976389043033123, "train_loss_llm": 0.0, "grad_norm": 0.01913829706609249, "global_step": 11345, "epoch": 315, "lr": 0.001} +{"train_loss": 0.04071105644106865, "train_loss_bc": 0.04071105644106865, "train_loss_llm": 0.0, "grad_norm": 0.03250928968191147, "global_step": 11346, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03922514244914055, "train_loss_bc": 0.03922514244914055, "train_loss_llm": 0.0, "grad_norm": 0.03912593051791191, "global_step": 11347, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03964710608124733, "train_loss_bc": 0.03964710608124733, "train_loss_llm": 0.0, "grad_norm": 0.04850618168711662, "global_step": 11348, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03899481147527695, "train_loss_bc": 0.03899481147527695, "train_loss_llm": 0.0, "grad_norm": 0.021503988653421402, "global_step": 11349, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03304139897227287, "train_loss_bc": 0.03304139897227287, "train_loss_llm": 0.0, "grad_norm": 0.030027687549591064, "global_step": 11350, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03601667657494545, "train_loss_bc": 0.03601667657494545, "train_loss_llm": 0.0, "grad_norm": 0.040354952216148376, "global_step": 11351, "epoch": 315, "lr": 0.001} +{"train_loss": 0.038742225617170334, "train_loss_bc": 0.038742225617170334, "train_loss_llm": 0.0, "grad_norm": 0.045411884784698486, "global_step": 11352, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03686608374118805, "train_loss_bc": 0.03686608374118805, "train_loss_llm": 0.0, "grad_norm": 0.022054750472307205, "global_step": 11353, "epoch": 315, "lr": 0.001} +{"train_loss": 0.0408179871737957, "train_loss_bc": 0.040803682059049606, "train_loss_llm": 0.0014303326606750488, "grad_norm": 0.03484748303890228, "global_step": 11354, "epoch": 315, "lr": 0.001} +{"train_loss": 0.04528827965259552, "train_loss_bc": 0.04528827965259552, "train_loss_llm": 0.0, "grad_norm": 0.04646120220422745, "global_step": 11355, "epoch": 315, "lr": 0.001} +{"train_loss": 0.039720483124256134, "train_loss_bc": 0.03970589116215706, "train_loss_llm": 0.0014590620994567871, "grad_norm": 0.054128680378198624, "global_step": 11356, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03813299909234047, "train_loss_bc": 0.03813299909234047, "train_loss_llm": 0.0, "grad_norm": 0.021928422152996063, "global_step": 11357, "epoch": 315, "lr": 0.001} +{"train_loss": 0.04963213950395584, "train_loss_bc": 0.04963213950395584, "train_loss_llm": 0.0, "grad_norm": 0.04367680475115776, "global_step": 11358, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03281943500041962, "train_loss_bc": 0.03281943500041962, "train_loss_llm": 0.0, "grad_norm": 0.06440695375204086, "global_step": 11359, "epoch": 315, "lr": 0.001} +{"train_loss": 0.0392487570643425, "train_loss_bc": 0.03923493251204491, "train_loss_llm": 0.001382291316986084, "grad_norm": 0.07464352250099182, "global_step": 11360, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03853807970881462, "train_loss_bc": 0.03853807970881462, "train_loss_llm": 0.0, "grad_norm": 0.024210389703512192, "global_step": 11361, "epoch": 315, "lr": 0.001} +{"train_loss": 0.04230959340929985, "train_loss_bc": 0.04230959340929985, "train_loss_llm": 0.0, "grad_norm": 0.033459682017564774, "global_step": 11362, "epoch": 315, "lr": 0.001} +{"train_loss": 0.037107039242982864, "train_loss_bc": 0.037100061774253845, "train_loss_llm": 0.0006976127624511719, "grad_norm": 0.04972030967473984, "global_step": 11363, "epoch": 315, "lr": 0.001} +{"train_loss": 0.036414872854948044, "train_loss_bc": 0.03640289232134819, "train_loss_llm": 0.001197904348373413, "grad_norm": 0.06517994403839111, "global_step": 11364, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03418230265378952, "train_loss_bc": 0.034169718623161316, "train_loss_llm": 0.0012582242488861084, "grad_norm": 0.019093727692961693, "global_step": 11365, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03841696307063103, "train_loss_bc": 0.03841696307063103, "train_loss_llm": 0.0, "grad_norm": 0.0289897657930851, "global_step": 11366, "epoch": 315, "lr": 0.001} +{"train_loss": 0.038229815661907196, "train_loss_bc": 0.03819480165839195, "train_loss_llm": 0.0035012364387512207, "grad_norm": 0.037768762558698654, "global_step": 11367, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03443378955125809, "train_loss_bc": 0.03443378955125809, "train_loss_llm": 0.0, "grad_norm": 0.046094540506601334, "global_step": 11368, "epoch": 315, "lr": 0.001} +{"train_loss": 0.04029959812760353, "train_loss_bc": 0.04028673470020294, "train_loss_llm": 0.0012862682342529297, "grad_norm": 0.02244037576019764, "global_step": 11369, "epoch": 315, "lr": 0.001} +{"train_loss": 0.0321686714887619, "train_loss_bc": 0.0321686714887619, "train_loss_llm": 0.0, "grad_norm": 0.03758113458752632, "global_step": 11370, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03447310999035835, "train_loss_bc": 0.03447310999035835, "train_loss_llm": 0.0, "grad_norm": 0.050302885472774506, "global_step": 11371, "epoch": 315, "lr": 0.001} +{"train_loss": 0.032886553555727005, "train_loss_bc": 0.032886553555727005, "train_loss_llm": 0.0, "grad_norm": 0.0642407238483429, "global_step": 11372, "epoch": 315, "lr": 0.001} +{"train_loss": 0.04089741036295891, "train_loss_bc": 0.040868520736694336, "train_loss_llm": 0.002889007329940796, "grad_norm": 0.02883293479681015, "global_step": 11373, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03682003915309906, "train_loss_bc": 0.03682003915309906, "train_loss_llm": 0.0, "grad_norm": 0.036660198122262955, "global_step": 11374, "epoch": 315, "lr": 0.001} +{"train_loss": 0.03769905953150657, "train_loss_bc": 0.043543796986341476, "train_loss_llm": 0.0, "grad_norm": 0.05599961429834366, "global_step": 11375, "epoch": 315, "lr": 0.001, "train/cumulative_reward": 98.2769303216319, "train/mean_score": 16.92034678403718, "train/success_rate": 0.4, "test/cumulative_reward": 52.11901716239887, "test/mean_score": 9.131073770146523, "test/success_rate": 0.22, "val_loss": 0.06324749439954758, "train_action_mse_error": 0.04057474806904793} +{"train_loss": 0.035405367612838745, "train_loss_bc": 0.035403523594141006, "train_loss_llm": 0.00018456578254699707, "grad_norm": 0.05580393970012665, "global_step": 11376, "epoch": 316, "lr": 0.001} +{"train_loss": 0.0392976775765419, "train_loss_bc": 0.03929729387164116, "train_loss_llm": 3.8236379623413086e-05, "grad_norm": 0.025694115087389946, "global_step": 11377, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03379441425204277, "train_loss_bc": 0.03379441425204277, "train_loss_llm": 0.0, "grad_norm": 0.0403197705745697, "global_step": 11378, "epoch": 316, "lr": 0.001} +{"train_loss": 0.04022977501153946, "train_loss_bc": 0.04022977501153946, "train_loss_llm": 0.0, "grad_norm": 0.058074429631233215, "global_step": 11379, "epoch": 316, "lr": 0.001} +{"train_loss": 0.038525573909282684, "train_loss_bc": 0.038525573909282684, "train_loss_llm": 0.0, "grad_norm": 0.06815952062606812, "global_step": 11380, "epoch": 316, "lr": 0.001} +{"train_loss": 0.044335946440696716, "train_loss_bc": 0.044335946440696716, "train_loss_llm": 0.0, "grad_norm": 0.02162599377334118, "global_step": 11381, "epoch": 316, "lr": 0.001} +{"train_loss": 0.031694523990154266, "train_loss_bc": 0.031694523990154266, "train_loss_llm": 0.0, "grad_norm": 0.03812817856669426, "global_step": 11382, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03094363398849964, "train_loss_bc": 0.03094363398849964, "train_loss_llm": 0.0, "grad_norm": 0.05594475194811821, "global_step": 11383, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03211404010653496, "train_loss_bc": 0.03211404010653496, "train_loss_llm": 0.0, "grad_norm": 0.06287931650876999, "global_step": 11384, "epoch": 316, "lr": 0.001} +{"train_loss": 0.036029934883117676, "train_loss_bc": 0.03602544963359833, "train_loss_llm": 0.00044858455657958984, "grad_norm": 0.019959434866905212, "global_step": 11385, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03291263058781624, "train_loss_bc": 0.03289588913321495, "train_loss_llm": 0.001674264669418335, "grad_norm": 0.026098256930708885, "global_step": 11386, "epoch": 316, "lr": 0.001} +{"train_loss": 0.0291296374052763, "train_loss_bc": 0.0291296374052763, "train_loss_llm": 0.0, "grad_norm": 0.0348581038415432, "global_step": 11387, "epoch": 316, "lr": 0.001} +{"train_loss": 0.04375821352005005, "train_loss_bc": 0.04375821352005005, "train_loss_llm": 0.0, "grad_norm": 0.04763644188642502, "global_step": 11388, "epoch": 316, "lr": 0.001} +{"train_loss": 0.04209321364760399, "train_loss_bc": 0.04206900671124458, "train_loss_llm": 0.0024206936359405518, "grad_norm": 0.02690321020781994, "global_step": 11389, "epoch": 316, "lr": 0.001} +{"train_loss": 0.04254631698131561, "train_loss_bc": 0.04254631698131561, "train_loss_llm": 0.0, "grad_norm": 0.033245787024497986, "global_step": 11390, "epoch": 316, "lr": 0.001} +{"train_loss": 0.042637936770915985, "train_loss_bc": 0.042637936770915985, "train_loss_llm": 0.0, "grad_norm": 0.04050357639789581, "global_step": 11391, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03621480613946915, "train_loss_bc": 0.03621480613946915, "train_loss_llm": 0.0, "grad_norm": 0.050295036286115646, "global_step": 11392, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03997023403644562, "train_loss_bc": 0.03996461257338524, "train_loss_llm": 0.0005622208118438721, "grad_norm": 0.026522548869252205, "global_step": 11393, "epoch": 316, "lr": 0.001} +{"train_loss": 0.041745059192180634, "train_loss_bc": 0.041745059192180634, "train_loss_llm": 0.0, "grad_norm": 0.04396800696849823, "global_step": 11394, "epoch": 316, "lr": 0.001} +{"train_loss": 0.04006219655275345, "train_loss_bc": 0.04006219655275345, "train_loss_llm": 0.0, "grad_norm": 0.0495397225022316, "global_step": 11395, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03637298196554184, "train_loss_bc": 0.03637298196554184, "train_loss_llm": 0.0, "grad_norm": 0.06445992738008499, "global_step": 11396, "epoch": 316, "lr": 0.001} +{"train_loss": 0.036493171006441116, "train_loss_bc": 0.036493171006441116, "train_loss_llm": 0.0, "grad_norm": 0.01915688067674637, "global_step": 11397, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03892366215586662, "train_loss_bc": 0.03892366215586662, "train_loss_llm": 0.0, "grad_norm": 0.03008873201906681, "global_step": 11398, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03600628674030304, "train_loss_bc": 0.03600628674030304, "train_loss_llm": 0.0, "grad_norm": 0.04095527529716492, "global_step": 11399, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03987549617886543, "train_loss_bc": 0.03986803814768791, "train_loss_llm": 0.0007458627223968506, "grad_norm": 0.054989010095596313, "global_step": 11400, "epoch": 316, "lr": 0.001} +{"train_loss": 0.0323248915374279, "train_loss_bc": 0.0323248915374279, "train_loss_llm": 0.0, "grad_norm": 0.018437882885336876, "global_step": 11401, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03678663820028305, "train_loss_bc": 0.03678663820028305, "train_loss_llm": 0.0, "grad_norm": 0.028517279773950577, "global_step": 11402, "epoch": 316, "lr": 0.001} +{"train_loss": 0.04149220138788223, "train_loss_bc": 0.04149220138788223, "train_loss_llm": 0.0, "grad_norm": 0.04078254476189613, "global_step": 11403, "epoch": 316, "lr": 0.001} +{"train_loss": 0.035900551825761795, "train_loss_bc": 0.035889215767383575, "train_loss_llm": 0.0011337101459503174, "grad_norm": 0.04937480017542839, "global_step": 11404, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03922848403453827, "train_loss_bc": 0.03922848403453827, "train_loss_llm": 0.0, "grad_norm": 0.02303905040025711, "global_step": 11405, "epoch": 316, "lr": 0.001} +{"train_loss": 0.0332629531621933, "train_loss_bc": 0.0332629531621933, "train_loss_llm": 0.0, "grad_norm": 0.03381303697824478, "global_step": 11406, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03839687258005142, "train_loss_bc": 0.03838079050183296, "train_loss_llm": 0.0016081035137176514, "grad_norm": 0.046991050243377686, "global_step": 11407, "epoch": 316, "lr": 0.001} +{"train_loss": 0.038082219660282135, "train_loss_bc": 0.038066424429416656, "train_loss_llm": 0.0015796124935150146, "grad_norm": 0.0598531998693943, "global_step": 11408, "epoch": 316, "lr": 0.001} +{"train_loss": 0.03773966431617737, "train_loss_bc": 0.037729475647211075, "train_loss_llm": 0.0010189712047576904, "grad_norm": 0.020518068224191666, "global_step": 11409, "epoch": 316, "lr": 0.001} +{"train_loss": 0.04170016944408417, "train_loss_bc": 0.04167521372437477, "train_loss_llm": 0.002495408058166504, "grad_norm": 0.03034817986190319, "global_step": 11410, "epoch": 316, "lr": 0.001} +{"train_loss": 0.037570913736191064, "train_loss_bc": 0.03652551770210266, "train_loss_llm": 0.0, "grad_norm": 0.04629585146903992, "global_step": 11411, "epoch": 316, "lr": 0.001, "val_loss": 0.06199461594223976} +{"train_loss": 0.03937117010354996, "train_loss_bc": 0.039367202669382095, "train_loss_llm": 0.00039690732955932617, "grad_norm": 0.06499544531106949, "global_step": 11412, "epoch": 317, "lr": 0.001} +{"train_loss": 0.04020841419696808, "train_loss_bc": 0.04020841419696808, "train_loss_llm": 0.0, "grad_norm": 0.024322230368852615, "global_step": 11413, "epoch": 317, "lr": 0.001} +{"train_loss": 0.02491871453821659, "train_loss_bc": 0.02491666190326214, "train_loss_llm": 0.00020518898963928223, "grad_norm": 0.027603957802057266, "global_step": 11414, "epoch": 317, "lr": 0.001} +{"train_loss": 0.039179328829050064, "train_loss_bc": 0.039179328829050064, "train_loss_llm": 0.0, "grad_norm": 0.033664584159851074, "global_step": 11415, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03645091876387596, "train_loss_bc": 0.03645091876387596, "train_loss_llm": 0.0, "grad_norm": 0.03970248997211456, "global_step": 11416, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03441540151834488, "train_loss_bc": 0.03441540151834488, "train_loss_llm": 0.0, "grad_norm": 0.019537759944796562, "global_step": 11417, "epoch": 317, "lr": 0.001} +{"train_loss": 0.04267318546772003, "train_loss_bc": 0.04267318546772003, "train_loss_llm": 0.0, "grad_norm": 0.02769518829882145, "global_step": 11418, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03764600306749344, "train_loss_bc": 0.03760877996683121, "train_loss_llm": 0.0037224888801574707, "grad_norm": 0.044067878276109695, "global_step": 11419, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03223823010921478, "train_loss_bc": 0.03223823010921478, "train_loss_llm": 0.0, "grad_norm": 0.05875113606452942, "global_step": 11420, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03714682534337044, "train_loss_bc": 0.03714682534337044, "train_loss_llm": 0.0, "grad_norm": 0.023413721472024918, "global_step": 11421, "epoch": 317, "lr": 0.001} +{"train_loss": 0.04001866281032562, "train_loss_bc": 0.04001866281032562, "train_loss_llm": 0.0, "grad_norm": 0.035314906388521194, "global_step": 11422, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03174975514411926, "train_loss_bc": 0.03174975514411926, "train_loss_llm": 0.0, "grad_norm": 0.04131441190838814, "global_step": 11423, "epoch": 317, "lr": 0.001} +{"train_loss": 0.042886827141046524, "train_loss_bc": 0.042886827141046524, "train_loss_llm": 0.0, "grad_norm": 0.05584828555583954, "global_step": 11424, "epoch": 317, "lr": 0.001} +{"train_loss": 0.04005458950996399, "train_loss_bc": 0.0400397889316082, "train_loss_llm": 0.001480013132095337, "grad_norm": 0.02176728844642639, "global_step": 11425, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03673676401376724, "train_loss_bc": 0.03673676401376724, "train_loss_llm": 0.0, "grad_norm": 0.029907332733273506, "global_step": 11426, "epoch": 317, "lr": 0.001} +{"train_loss": 0.041898153722286224, "train_loss_bc": 0.041881874203681946, "train_loss_llm": 0.0016280412673950195, "grad_norm": 0.038795843720436096, "global_step": 11427, "epoch": 317, "lr": 0.001} +{"train_loss": 0.036309193819761276, "train_loss_bc": 0.036309193819761276, "train_loss_llm": 0.0, "grad_norm": 0.04776230454444885, "global_step": 11428, "epoch": 317, "lr": 0.001} +{"train_loss": 0.042411986738443375, "train_loss_bc": 0.042411986738443375, "train_loss_llm": 0.0, "grad_norm": 0.03214731439948082, "global_step": 11429, "epoch": 317, "lr": 0.001} +{"train_loss": 0.036734819412231445, "train_loss_bc": 0.036733660846948624, "train_loss_llm": 0.00011590123176574707, "grad_norm": 0.04283362999558449, "global_step": 11430, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03538765385746956, "train_loss_bc": 0.03537795692682266, "train_loss_llm": 0.0009696781635284424, "grad_norm": 0.051282498985528946, "global_step": 11431, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03755423426628113, "train_loss_bc": 0.03755423426628113, "train_loss_llm": 0.0, "grad_norm": 0.05959899723529816, "global_step": 11432, "epoch": 317, "lr": 0.001} +{"train_loss": 0.040942881256341934, "train_loss_bc": 0.040942881256341934, "train_loss_llm": 0.0, "grad_norm": 0.026236509904265404, "global_step": 11433, "epoch": 317, "lr": 0.001} +{"train_loss": 0.044261299073696136, "train_loss_bc": 0.044261299073696136, "train_loss_llm": 0.0, "grad_norm": 0.039278775453567505, "global_step": 11434, "epoch": 317, "lr": 0.001} +{"train_loss": 0.036883193999528885, "train_loss_bc": 0.036883193999528885, "train_loss_llm": 0.0, "grad_norm": 0.05176088586449623, "global_step": 11435, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03589149937033653, "train_loss_bc": 0.035890452563762665, "train_loss_llm": 0.00010463595390319824, "grad_norm": 0.0635034367442131, "global_step": 11436, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03438775986433029, "train_loss_bc": 0.0343734510242939, "train_loss_llm": 0.0014309287071228027, "grad_norm": 0.02545315958559513, "global_step": 11437, "epoch": 317, "lr": 0.001} +{"train_loss": 0.038396816700696945, "train_loss_bc": 0.038396816700696945, "train_loss_llm": 0.0, "grad_norm": 0.036820877343416214, "global_step": 11438, "epoch": 317, "lr": 0.001} +{"train_loss": 0.032589808106422424, "train_loss_bc": 0.032589808106422424, "train_loss_llm": 0.0, "grad_norm": 0.05505014955997467, "global_step": 11439, "epoch": 317, "lr": 0.001} +{"train_loss": 0.036218903958797455, "train_loss_bc": 0.036194976419210434, "train_loss_llm": 0.002392590045928955, "grad_norm": 0.0745992511510849, "global_step": 11440, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03739691153168678, "train_loss_bc": 0.03739691153168678, "train_loss_llm": 0.0, "grad_norm": 0.02890719100832939, "global_step": 11441, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03638168051838875, "train_loss_bc": 0.03638168051838875, "train_loss_llm": 0.0, "grad_norm": 0.04226426035165787, "global_step": 11442, "epoch": 317, "lr": 0.001} +{"train_loss": 0.037941060960292816, "train_loss_bc": 0.037941060960292816, "train_loss_llm": 0.0, "grad_norm": 0.05379128456115723, "global_step": 11443, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03984343633055687, "train_loss_bc": 0.039831891655921936, "train_loss_llm": 0.0011543631553649902, "grad_norm": 0.06503409147262573, "global_step": 11444, "epoch": 317, "lr": 0.001} +{"train_loss": 0.04043681547045708, "train_loss_bc": 0.04043681547045708, "train_loss_llm": 0.0, "grad_norm": 0.020744893699884415, "global_step": 11445, "epoch": 317, "lr": 0.001} +{"train_loss": 0.03320790082216263, "train_loss_bc": 0.03320790082216263, "train_loss_llm": 0.0, "grad_norm": 0.028877418488264084, "global_step": 11446, "epoch": 317, "lr": 0.001} +{"train_loss": 0.037602819657574095, "train_loss_bc": 0.04293070733547211, "train_loss_llm": 0.0, "grad_norm": 0.04187358170747757, "global_step": 11447, "epoch": 317, "lr": 0.001, "val_loss": 0.06376300007104874} +{"train_loss": 0.036031533032655716, "train_loss_bc": 0.036031533032655716, "train_loss_llm": 0.0, "grad_norm": 0.05008547753095627, "global_step": 11448, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03618169203400612, "train_loss_bc": 0.03617880493402481, "train_loss_llm": 0.0002886056900024414, "grad_norm": 0.019261520355939865, "global_step": 11449, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03657136112451553, "train_loss_bc": 0.03657136112451553, "train_loss_llm": 0.0, "grad_norm": 0.026568450033664703, "global_step": 11450, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03376661613583565, "train_loss_bc": 0.03376661613583565, "train_loss_llm": 0.0, "grad_norm": 0.038185004144907, "global_step": 11451, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03856014460325241, "train_loss_bc": 0.03854502737522125, "train_loss_llm": 0.001511693000793457, "grad_norm": 0.047173332422971725, "global_step": 11452, "epoch": 318, "lr": 0.001} +{"train_loss": 0.043594278395175934, "train_loss_bc": 0.0435522198677063, "train_loss_llm": 0.004205673933029175, "grad_norm": 0.02904387377202511, "global_step": 11453, "epoch": 318, "lr": 0.001} +{"train_loss": 0.030756687745451927, "train_loss_bc": 0.030756687745451927, "train_loss_llm": 0.0, "grad_norm": 0.04093937203288078, "global_step": 11454, "epoch": 318, "lr": 0.001} +{"train_loss": 0.04000221937894821, "train_loss_bc": 0.03999371826648712, "train_loss_llm": 0.0008501410484313965, "grad_norm": 0.04431489482522011, "global_step": 11455, "epoch": 318, "lr": 0.001} +{"train_loss": 0.028337698429822922, "train_loss_bc": 0.028337698429822922, "train_loss_llm": 0.0, "grad_norm": 0.05220978707075119, "global_step": 11456, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03586138039827347, "train_loss_bc": 0.03585915267467499, "train_loss_llm": 0.00022280216217041016, "grad_norm": 0.029269464313983917, "global_step": 11457, "epoch": 318, "lr": 0.001} +{"train_loss": 0.04532970115542412, "train_loss_bc": 0.04532970115542412, "train_loss_llm": 0.0, "grad_norm": 0.04507042467594147, "global_step": 11458, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03508954867720604, "train_loss_bc": 0.03508954867720604, "train_loss_llm": 0.0, "grad_norm": 0.057318542152643204, "global_step": 11459, "epoch": 318, "lr": 0.001} +{"train_loss": 0.04385434836149216, "train_loss_bc": 0.04385434836149216, "train_loss_llm": 0.0, "grad_norm": 0.07826610654592514, "global_step": 11460, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03536784648895264, "train_loss_bc": 0.03536784648895264, "train_loss_llm": 0.0, "grad_norm": 0.02020321413874626, "global_step": 11461, "epoch": 318, "lr": 0.001} +{"train_loss": 0.032109584659338, "train_loss_bc": 0.03210483491420746, "train_loss_llm": 0.00047510862350463867, "grad_norm": 0.033135268837213516, "global_step": 11462, "epoch": 318, "lr": 0.001} +{"train_loss": 0.035427749156951904, "train_loss_bc": 0.035427749156951904, "train_loss_llm": 0.0, "grad_norm": 0.03866977244615555, "global_step": 11463, "epoch": 318, "lr": 0.001} +{"train_loss": 0.0362318716943264, "train_loss_bc": 0.03623141720890999, "train_loss_llm": 4.556775093078613e-05, "grad_norm": 0.0503438338637352, "global_step": 11464, "epoch": 318, "lr": 0.001} +{"train_loss": 0.033756762742996216, "train_loss_bc": 0.033756762742996216, "train_loss_llm": 0.0, "grad_norm": 0.019111579284071922, "global_step": 11465, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03820990025997162, "train_loss_bc": 0.03820990025997162, "train_loss_llm": 0.0, "grad_norm": 0.028931118547916412, "global_step": 11466, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03735821321606636, "train_loss_bc": 0.03735821321606636, "train_loss_llm": 0.0, "grad_norm": 0.0382436104118824, "global_step": 11467, "epoch": 318, "lr": 0.001} +{"train_loss": 0.037884075194597244, "train_loss_bc": 0.037883784621953964, "train_loss_llm": 2.893805503845215e-05, "grad_norm": 0.043627720326185226, "global_step": 11468, "epoch": 318, "lr": 0.001} +{"train_loss": 0.040921248495578766, "train_loss_bc": 0.040921248495578766, "train_loss_llm": 0.0, "grad_norm": 0.02221117727458477, "global_step": 11469, "epoch": 318, "lr": 0.001} +{"train_loss": 0.04024868831038475, "train_loss_bc": 0.04024868831038475, "train_loss_llm": 0.0, "grad_norm": 0.03165292739868164, "global_step": 11470, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03945503756403923, "train_loss_bc": 0.03945503756403923, "train_loss_llm": 0.0, "grad_norm": 0.04051237180829048, "global_step": 11471, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03911764174699783, "train_loss_bc": 0.03911764174699783, "train_loss_llm": 0.0, "grad_norm": 0.049488335847854614, "global_step": 11472, "epoch": 318, "lr": 0.001} +{"train_loss": 0.035372037440538406, "train_loss_bc": 0.035371433943510056, "train_loss_llm": 6.031990051269531e-05, "grad_norm": 0.02288970537483692, "global_step": 11473, "epoch": 318, "lr": 0.001} +{"train_loss": 0.04438859224319458, "train_loss_bc": 0.04438859224319458, "train_loss_llm": 0.0, "grad_norm": 0.043159838765859604, "global_step": 11474, "epoch": 318, "lr": 0.001} +{"train_loss": 0.047397058457136154, "train_loss_bc": 0.047390151768922806, "train_loss_llm": 0.000690758228302002, "grad_norm": 0.05115441977977753, "global_step": 11475, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03298819810152054, "train_loss_bc": 0.03298819810152054, "train_loss_llm": 0.0, "grad_norm": 0.06366945803165436, "global_step": 11476, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03879935294389725, "train_loss_bc": 0.03879935294389725, "train_loss_llm": 0.0, "grad_norm": 0.02458007074892521, "global_step": 11477, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03660333529114723, "train_loss_bc": 0.03659995272755623, "train_loss_llm": 0.00033843517303466797, "grad_norm": 0.03875608742237091, "global_step": 11478, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03650854155421257, "train_loss_bc": 0.03648637980222702, "train_loss_llm": 0.0022161006927490234, "grad_norm": 0.05244218185544014, "global_step": 11479, "epoch": 318, "lr": 0.001} +{"train_loss": 0.038139745593070984, "train_loss_bc": 0.038139745593070984, "train_loss_llm": 0.0, "grad_norm": 0.06193673610687256, "global_step": 11480, "epoch": 318, "lr": 0.001} +{"train_loss": 0.036594536155462265, "train_loss_bc": 0.036594536155462265, "train_loss_llm": 0.0, "grad_norm": 0.019508957862854004, "global_step": 11481, "epoch": 318, "lr": 0.001} +{"train_loss": 0.034494370222091675, "train_loss_bc": 0.034490007907152176, "train_loss_llm": 0.0004363059997558594, "grad_norm": 0.02940685860812664, "global_step": 11482, "epoch": 318, "lr": 0.001} +{"train_loss": 0.03763364633131358, "train_loss_bc": 0.043499670922756195, "train_loss_llm": 0.0, "grad_norm": 0.044693224132061005, "global_step": 11483, "epoch": 318, "lr": 0.001, "val_loss": 0.06338344514369965} +{"train_loss": 0.03004857897758484, "train_loss_bc": 0.03004857897758484, "train_loss_llm": 0.0, "grad_norm": 0.04677646607160568, "global_step": 11484, "epoch": 319, "lr": 0.001} +{"train_loss": 0.043538980185985565, "train_loss_bc": 0.0435367226600647, "train_loss_llm": 0.00022593140602111816, "grad_norm": 0.02405877225100994, "global_step": 11485, "epoch": 319, "lr": 0.001} +{"train_loss": 0.04121294990181923, "train_loss_bc": 0.04121294990181923, "train_loss_llm": 0.0, "grad_norm": 0.03414499759674072, "global_step": 11486, "epoch": 319, "lr": 0.001} +{"train_loss": 0.04047266021370888, "train_loss_bc": 0.04047266021370888, "train_loss_llm": 0.0, "grad_norm": 0.044055208563804626, "global_step": 11487, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03773678094148636, "train_loss_bc": 0.03773678094148636, "train_loss_llm": 0.0, "grad_norm": 0.05126291885972023, "global_step": 11488, "epoch": 319, "lr": 0.001} +{"train_loss": 0.038395099341869354, "train_loss_bc": 0.038395099341869354, "train_loss_llm": 0.0, "grad_norm": 0.020347164943814278, "global_step": 11489, "epoch": 319, "lr": 0.001} +{"train_loss": 0.04293785244226456, "train_loss_bc": 0.04293785244226456, "train_loss_llm": 0.0, "grad_norm": 0.03635852783918381, "global_step": 11490, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03744303062558174, "train_loss_bc": 0.03741014376282692, "train_loss_llm": 0.003288567066192627, "grad_norm": 0.045079831033945084, "global_step": 11491, "epoch": 319, "lr": 0.001} +{"train_loss": 0.028733588755130768, "train_loss_bc": 0.02872398868203163, "train_loss_llm": 0.0009599626064300537, "grad_norm": 0.057881634682416916, "global_step": 11492, "epoch": 319, "lr": 0.001} +{"train_loss": 0.036494191735982895, "train_loss_bc": 0.036494191735982895, "train_loss_llm": 0.0, "grad_norm": 0.02037912793457508, "global_step": 11493, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03390219807624817, "train_loss_bc": 0.03389943763613701, "train_loss_llm": 0.00027620792388916016, "grad_norm": 0.03078942745923996, "global_step": 11494, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03263173997402191, "train_loss_bc": 0.03263173997402191, "train_loss_llm": 0.0, "grad_norm": 0.039223216474056244, "global_step": 11495, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03533657640218735, "train_loss_bc": 0.03533657640218735, "train_loss_llm": 0.0, "grad_norm": 0.05208464711904526, "global_step": 11496, "epoch": 319, "lr": 0.001} +{"train_loss": 0.029129136353731155, "train_loss_bc": 0.029129136353731155, "train_loss_llm": 0.0, "grad_norm": 0.021584276109933853, "global_step": 11497, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03682861104607582, "train_loss_bc": 0.036808665841817856, "train_loss_llm": 0.0019944310188293457, "grad_norm": 0.03869309276342392, "global_step": 11498, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03795018047094345, "train_loss_bc": 0.03791438043117523, "train_loss_llm": 0.00358012318611145, "grad_norm": 0.05529090762138367, "global_step": 11499, "epoch": 319, "lr": 0.001} +{"train_loss": 0.035164058208465576, "train_loss_bc": 0.035164058208465576, "train_loss_llm": 0.0, "grad_norm": 0.058396466076374054, "global_step": 11500, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03882284462451935, "train_loss_bc": 0.03882284462451935, "train_loss_llm": 0.0, "grad_norm": 0.018408117815852165, "global_step": 11501, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03118561953306198, "train_loss_bc": 0.03118561953306198, "train_loss_llm": 0.0, "grad_norm": 0.025227617472410202, "global_step": 11502, "epoch": 319, "lr": 0.001} +{"train_loss": 0.04572100192308426, "train_loss_bc": 0.04572100192308426, "train_loss_llm": 0.0, "grad_norm": 0.03272370994091034, "global_step": 11503, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03377633914351463, "train_loss_bc": 0.03377633914351463, "train_loss_llm": 0.0, "grad_norm": 0.039265841245651245, "global_step": 11504, "epoch": 319, "lr": 0.001} +{"train_loss": 0.04092098027467728, "train_loss_bc": 0.04092098027467728, "train_loss_llm": 0.0, "grad_norm": 0.021229268983006477, "global_step": 11505, "epoch": 319, "lr": 0.001} +{"train_loss": 0.032162416726350784, "train_loss_bc": 0.032141052186489105, "train_loss_llm": 0.002136319875717163, "grad_norm": 0.02824927680194378, "global_step": 11506, "epoch": 319, "lr": 0.001} +{"train_loss": 0.04042510315775871, "train_loss_bc": 0.04042510315775871, "train_loss_llm": 0.0, "grad_norm": 0.03889003396034241, "global_step": 11507, "epoch": 319, "lr": 0.001} +{"train_loss": 0.04197509214282036, "train_loss_bc": 0.04197509214282036, "train_loss_llm": 0.0, "grad_norm": 0.04681616649031639, "global_step": 11508, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03553984314203262, "train_loss_bc": 0.03553984314203262, "train_loss_llm": 0.0, "grad_norm": 0.019894344732165337, "global_step": 11509, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03755795210599899, "train_loss_bc": 0.03755367547273636, "train_loss_llm": 0.00042757391929626465, "grad_norm": 0.03606867417693138, "global_step": 11510, "epoch": 319, "lr": 0.001} +{"train_loss": 0.039107415825128555, "train_loss_bc": 0.039107415825128555, "train_loss_llm": 0.0, "grad_norm": 0.04762640967965126, "global_step": 11511, "epoch": 319, "lr": 0.001} +{"train_loss": 0.04292237386107445, "train_loss_bc": 0.04291911423206329, "train_loss_llm": 0.0003260672092437744, "grad_norm": 0.06033117696642876, "global_step": 11512, "epoch": 319, "lr": 0.001} +{"train_loss": 0.0416443794965744, "train_loss_bc": 0.0416443794965744, "train_loss_llm": 0.0, "grad_norm": 0.020575206726789474, "global_step": 11513, "epoch": 319, "lr": 0.001} +{"train_loss": 0.036014705896377563, "train_loss_bc": 0.03601391986012459, "train_loss_llm": 7.852911949157715e-05, "grad_norm": 0.032015908509492874, "global_step": 11514, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03828657418489456, "train_loss_bc": 0.03828657418489456, "train_loss_llm": 0.0, "grad_norm": 0.041598498821258545, "global_step": 11515, "epoch": 319, "lr": 0.001} +{"train_loss": 0.04205648601055145, "train_loss_bc": 0.04205648601055145, "train_loss_llm": 0.0, "grad_norm": 0.04536770284175873, "global_step": 11516, "epoch": 319, "lr": 0.001} +{"train_loss": 0.04124383255839348, "train_loss_bc": 0.04120437428355217, "train_loss_llm": 0.003945797681808472, "grad_norm": 0.021899661049246788, "global_step": 11517, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03666510432958603, "train_loss_bc": 0.03666510432958603, "train_loss_llm": 0.0, "grad_norm": 0.031733885407447815, "global_step": 11518, "epoch": 319, "lr": 0.001} +{"train_loss": 0.03730603095350994, "train_loss_bc": 0.029032835736870766, "train_loss_llm": 0.0, "grad_norm": 0.04901999607682228, "global_step": 11519, "epoch": 319, "lr": 0.001, "val_loss": 0.06269975751638412} +{"train_loss": 0.033724214881658554, "train_loss_bc": 0.033721860498189926, "train_loss_llm": 0.00023540854454040527, "grad_norm": 0.05481841787695885, "global_step": 11520, "epoch": 320, "lr": 0.001} +{"train_loss": 0.036166924983263016, "train_loss_bc": 0.03616407886147499, "train_loss_llm": 0.00028458237648010254, "grad_norm": 0.021887123584747314, "global_step": 11521, "epoch": 320, "lr": 0.001} +{"train_loss": 0.042970381677150726, "train_loss_bc": 0.042970381677150726, "train_loss_llm": 0.0, "grad_norm": 0.036043282598257065, "global_step": 11522, "epoch": 320, "lr": 0.001} +{"train_loss": 0.036359693855047226, "train_loss_bc": 0.036359693855047226, "train_loss_llm": 0.0, "grad_norm": 0.0381925143301487, "global_step": 11523, "epoch": 320, "lr": 0.001} +{"train_loss": 0.031068170443177223, "train_loss_bc": 0.031068170443177223, "train_loss_llm": 0.0, "grad_norm": 0.04542835056781769, "global_step": 11524, "epoch": 320, "lr": 0.001} +{"train_loss": 0.03815791755914688, "train_loss_bc": 0.03815791755914688, "train_loss_llm": 0.0, "grad_norm": 0.022040553390979767, "global_step": 11525, "epoch": 320, "lr": 0.001} +{"train_loss": 0.04706553742289543, "train_loss_bc": 0.04706553742289543, "train_loss_llm": 0.0, "grad_norm": 0.03307194262742996, "global_step": 11526, "epoch": 320, "lr": 0.001} +{"train_loss": 0.0380641408264637, "train_loss_bc": 0.03805745020508766, "train_loss_llm": 0.0006691217422485352, "grad_norm": 0.04357719421386719, "global_step": 11527, "epoch": 320, "lr": 0.001} +{"train_loss": 0.0364089161157608, "train_loss_bc": 0.0364089161157608, "train_loss_llm": 0.0, "grad_norm": 0.052123669534921646, "global_step": 11528, "epoch": 320, "lr": 0.001} +{"train_loss": 0.041076015681028366, "train_loss_bc": 0.0410618893802166, "train_loss_llm": 0.0014126002788543701, "grad_norm": 0.020289679989218712, "global_step": 11529, "epoch": 320, "lr": 0.001} +{"train_loss": 0.03874104097485542, "train_loss_bc": 0.03874104097485542, "train_loss_llm": 0.0, "grad_norm": 0.029751278460025787, "global_step": 11530, "epoch": 320, "lr": 0.001} +{"train_loss": 0.03223680332303047, "train_loss_bc": 0.032222308218479156, "train_loss_llm": 0.0014494359493255615, "grad_norm": 0.0375882051885128, "global_step": 11531, "epoch": 320, "lr": 0.001} +{"train_loss": 0.03579207882285118, "train_loss_bc": 0.03579207882285118, "train_loss_llm": 0.0, "grad_norm": 0.04756719991564751, "global_step": 11532, "epoch": 320, "lr": 0.001} +{"train_loss": 0.04195038601756096, "train_loss_bc": 0.04195038601756096, "train_loss_llm": 0.0, "grad_norm": 0.028072897344827652, "global_step": 11533, "epoch": 320, "lr": 0.001} +{"train_loss": 0.04069375991821289, "train_loss_bc": 0.04069375991821289, "train_loss_llm": 0.0, "grad_norm": 0.03526921942830086, "global_step": 11534, "epoch": 320, "lr": 0.001} +{"train_loss": 0.03182046487927437, "train_loss_bc": 0.031817734241485596, "train_loss_llm": 0.00027301907539367676, "grad_norm": 0.03632260113954544, "global_step": 11535, "epoch": 320, "lr": 0.001} +{"train_loss": 0.03813713788986206, "train_loss_bc": 0.03813435509800911, "train_loss_llm": 0.0002784132957458496, "grad_norm": 0.045187294483184814, "global_step": 11536, "epoch": 320, "lr": 0.001} +{"train_loss": 0.035829320549964905, "train_loss_bc": 0.035829320549964905, "train_loss_llm": 0.0, "grad_norm": 0.020561009645462036, "global_step": 11537, "epoch": 320, "lr": 0.001} +{"train_loss": 0.040327493101358414, "train_loss_bc": 0.040327493101358414, "train_loss_llm": 0.0, "grad_norm": 0.02339177392423153, "global_step": 11538, "epoch": 320, "lr": 0.001} +{"train_loss": 0.0402703583240509, "train_loss_bc": 0.0402703583240509, "train_loss_llm": 0.0, "grad_norm": 0.038642872124910355, "global_step": 11539, "epoch": 320, "lr": 0.001} +{"train_loss": 0.02780902199447155, "train_loss_bc": 0.02780902199447155, "train_loss_llm": 0.0, "grad_norm": 0.046991750597953796, "global_step": 11540, "epoch": 320, "lr": 0.001} +{"train_loss": 0.04275292158126831, "train_loss_bc": 0.04275292158126831, "train_loss_llm": 0.0, "grad_norm": 0.025873860344290733, "global_step": 11541, "epoch": 320, "lr": 0.001} +{"train_loss": 0.031946901232004166, "train_loss_bc": 0.03194236382842064, "train_loss_llm": 0.0004538297653198242, "grad_norm": 0.037059683352708817, "global_step": 11542, "epoch": 320, "lr": 0.001} +{"train_loss": 0.035757988691329956, "train_loss_bc": 0.035757988691329956, "train_loss_llm": 0.0, "grad_norm": 0.05137696862220764, "global_step": 11543, "epoch": 320, "lr": 0.001} +{"train_loss": 0.039849523454904556, "train_loss_bc": 0.0398268885910511, "train_loss_llm": 0.0022635161876678467, "grad_norm": 0.06898454576730728, "global_step": 11544, "epoch": 320, "lr": 0.001} +{"train_loss": 0.029113952070474625, "train_loss_bc": 0.029113952070474625, "train_loss_llm": 0.0, "grad_norm": 0.02133800834417343, "global_step": 11545, "epoch": 320, "lr": 0.001} +{"train_loss": 0.03949182108044624, "train_loss_bc": 0.03949182108044624, "train_loss_llm": 0.0, "grad_norm": 0.03160984441637993, "global_step": 11546, "epoch": 320, "lr": 0.001} +{"train_loss": 0.039152614772319794, "train_loss_bc": 0.039152614772319794, "train_loss_llm": 0.0, "grad_norm": 0.0444711297750473, "global_step": 11547, "epoch": 320, "lr": 0.001} +{"train_loss": 0.035273369401693344, "train_loss_bc": 0.035273369401693344, "train_loss_llm": 0.0, "grad_norm": 0.061743758618831635, "global_step": 11548, "epoch": 320, "lr": 0.001} +{"train_loss": 0.03772639483213425, "train_loss_bc": 0.03772639483213425, "train_loss_llm": 0.0, "grad_norm": 0.020846335217356682, "global_step": 11549, "epoch": 320, "lr": 0.001} +{"train_loss": 0.04189732298254967, "train_loss_bc": 0.04189732298254967, "train_loss_llm": 0.0, "grad_norm": 0.03328248858451843, "global_step": 11550, "epoch": 320, "lr": 0.001} +{"train_loss": 0.04387938231229782, "train_loss_bc": 0.04387691244482994, "train_loss_llm": 0.0002468526363372803, "grad_norm": 0.043867290019989014, "global_step": 11551, "epoch": 320, "lr": 0.001} +{"train_loss": 0.037990670651197433, "train_loss_bc": 0.037952739745378494, "train_loss_llm": 0.0037929415702819824, "grad_norm": 0.052592091262340546, "global_step": 11552, "epoch": 320, "lr": 0.001} +{"train_loss": 0.031117314472794533, "train_loss_bc": 0.031117314472794533, "train_loss_llm": 0.0, "grad_norm": 0.017706168815493584, "global_step": 11553, "epoch": 320, "lr": 0.001} +{"train_loss": 0.03526967763900757, "train_loss_bc": 0.03526967763900757, "train_loss_llm": 0.0, "grad_norm": 0.02458537369966507, "global_step": 11554, "epoch": 320, "lr": 0.001} +{"train_loss": 0.03745107047466768, "train_loss_bc": 0.04234890267252922, "train_loss_llm": 0.0, "grad_norm": 0.03766241669654846, "global_step": 11555, "epoch": 320, "lr": 0.001, "train/cumulative_reward": 97.98003414343907, "train/mean_score": 16.830157163615397, "train/success_rate": 0.4, "test/cumulative_reward": 40.196056213705106, "test/mean_score": 9.483668219244556, "test/success_rate": 0.22, "val_loss": 0.06268910318613052, "train_action_mse_error": 0.040456581860780716} +{"train_loss": 0.03553323820233345, "train_loss_bc": 0.03551958501338959, "train_loss_llm": 0.0013651549816131592, "grad_norm": 0.045624881982803345, "global_step": 11556, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03414691612124443, "train_loss_bc": 0.03414691612124443, "train_loss_llm": 0.0, "grad_norm": 0.016504688188433647, "global_step": 11557, "epoch": 321, "lr": 0.001} +{"train_loss": 0.029462244361639023, "train_loss_bc": 0.029462244361639023, "train_loss_llm": 0.0, "grad_norm": 0.026306649670004845, "global_step": 11558, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03822215646505356, "train_loss_bc": 0.03822215646505356, "train_loss_llm": 0.0, "grad_norm": 0.03217365965247154, "global_step": 11559, "epoch": 321, "lr": 0.001} +{"train_loss": 0.04150937497615814, "train_loss_bc": 0.04150937497615814, "train_loss_llm": 0.0, "grad_norm": 0.04003334790468216, "global_step": 11560, "epoch": 321, "lr": 0.001} +{"train_loss": 0.035453274846076965, "train_loss_bc": 0.035453274846076965, "train_loss_llm": 0.0, "grad_norm": 0.01985728181898594, "global_step": 11561, "epoch": 321, "lr": 0.001} +{"train_loss": 0.036471977829933167, "train_loss_bc": 0.036471977829933167, "train_loss_llm": 0.0, "grad_norm": 0.030299361795186996, "global_step": 11562, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03383807837963104, "train_loss_bc": 0.03383807837963104, "train_loss_llm": 0.0, "grad_norm": 0.03677767142653465, "global_step": 11563, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03782276436686516, "train_loss_bc": 0.03782276436686516, "train_loss_llm": 0.0, "grad_norm": 0.04630376398563385, "global_step": 11564, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03702394291758537, "train_loss_bc": 0.037018779665231705, "train_loss_llm": 0.0005165040493011475, "grad_norm": 0.025831058621406555, "global_step": 11565, "epoch": 321, "lr": 0.001} +{"train_loss": 0.036559537053108215, "train_loss_bc": 0.036559537053108215, "train_loss_llm": 0.0, "grad_norm": 0.04589664191007614, "global_step": 11566, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03352142497897148, "train_loss_bc": 0.03352142497897148, "train_loss_llm": 0.0, "grad_norm": 0.06439477205276489, "global_step": 11567, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03198326751589775, "train_loss_bc": 0.03198326751589775, "train_loss_llm": 0.0, "grad_norm": 0.0798143818974495, "global_step": 11568, "epoch": 321, "lr": 0.001} +{"train_loss": 0.033599548041820526, "train_loss_bc": 0.033599548041820526, "train_loss_llm": 0.0, "grad_norm": 0.01895982399582863, "global_step": 11569, "epoch": 321, "lr": 0.001} +{"train_loss": 0.031488385051488876, "train_loss_bc": 0.031450603157281876, "train_loss_llm": 0.00377810001373291, "grad_norm": 0.032475851476192474, "global_step": 11570, "epoch": 321, "lr": 0.001} +{"train_loss": 0.04228894039988518, "train_loss_bc": 0.04228894039988518, "train_loss_llm": 0.0, "grad_norm": 0.04953952506184578, "global_step": 11571, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03681296855211258, "train_loss_bc": 0.03680181875824928, "train_loss_llm": 0.001114875078201294, "grad_norm": 0.050650473684072495, "global_step": 11572, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03680020198225975, "train_loss_bc": 0.03680020198225975, "train_loss_llm": 0.0, "grad_norm": 0.02013365738093853, "global_step": 11573, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03619442135095596, "train_loss_bc": 0.03619442135095596, "train_loss_llm": 0.0, "grad_norm": 0.031202854588627815, "global_step": 11574, "epoch": 321, "lr": 0.001} +{"train_loss": 0.039074286818504333, "train_loss_bc": 0.039074286818504333, "train_loss_llm": 0.0, "grad_norm": 0.04114123806357384, "global_step": 11575, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03950716555118561, "train_loss_bc": 0.03950716555118561, "train_loss_llm": 0.0, "grad_norm": 0.052754633128643036, "global_step": 11576, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03466596454381943, "train_loss_bc": 0.03466596454381943, "train_loss_llm": 0.0, "grad_norm": 0.020951248705387115, "global_step": 11577, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03886036202311516, "train_loss_bc": 0.03886036202311516, "train_loss_llm": 0.0, "grad_norm": 0.03865503519773483, "global_step": 11578, "epoch": 321, "lr": 0.001} +{"train_loss": 0.046458471566438675, "train_loss_bc": 0.046458471566438675, "train_loss_llm": 0.0, "grad_norm": 0.0485050305724144, "global_step": 11579, "epoch": 321, "lr": 0.001} +{"train_loss": 0.040264345705509186, "train_loss_bc": 0.04024919867515564, "train_loss_llm": 0.0015147924423217773, "grad_norm": 0.06148441508412361, "global_step": 11580, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03459233045578003, "train_loss_bc": 0.03459233045578003, "train_loss_llm": 0.0, "grad_norm": 0.022440997883677483, "global_step": 11581, "epoch": 321, "lr": 0.001} +{"train_loss": 0.04009174183011055, "train_loss_bc": 0.04008401557803154, "train_loss_llm": 0.000772625207901001, "grad_norm": 0.03779448941349983, "global_step": 11582, "epoch": 321, "lr": 0.001} +{"train_loss": 0.04351845756173134, "train_loss_bc": 0.04351845756173134, "train_loss_llm": 0.0, "grad_norm": 0.052081093192100525, "global_step": 11583, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03481312096118927, "train_loss_bc": 0.03481312096118927, "train_loss_llm": 0.0, "grad_norm": 0.059127192944288254, "global_step": 11584, "epoch": 321, "lr": 0.001} +{"train_loss": 0.0432356521487236, "train_loss_bc": 0.0432356521487236, "train_loss_llm": 0.0, "grad_norm": 0.01916597969830036, "global_step": 11585, "epoch": 321, "lr": 0.001} +{"train_loss": 0.044337306171655655, "train_loss_bc": 0.044337306171655655, "train_loss_llm": 0.0, "grad_norm": 0.02646879106760025, "global_step": 11586, "epoch": 321, "lr": 0.001} +{"train_loss": 0.035728778690099716, "train_loss_bc": 0.035718049854040146, "train_loss_llm": 0.0010728538036346436, "grad_norm": 0.0367245189845562, "global_step": 11587, "epoch": 321, "lr": 0.001} +{"train_loss": 0.0395924337208271, "train_loss_bc": 0.0395924337208271, "train_loss_llm": 0.0, "grad_norm": 0.04909144341945648, "global_step": 11588, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03591537848114967, "train_loss_bc": 0.03591537848114967, "train_loss_llm": 0.0, "grad_norm": 0.021698622032999992, "global_step": 11589, "epoch": 321, "lr": 0.001} +{"train_loss": 0.046843454241752625, "train_loss_bc": 0.046843454241752625, "train_loss_llm": 0.0, "grad_norm": 0.04899655282497406, "global_step": 11590, "epoch": 321, "lr": 0.001} +{"train_loss": 0.03735484855456485, "train_loss_bc": 0.02854263409972191, "train_loss_llm": 0.0, "grad_norm": 0.060963280498981476, "global_step": 11591, "epoch": 321, "lr": 0.001, "val_loss": 0.06343242526054382} +{"train_loss": 0.04230576753616333, "train_loss_bc": 0.04230576753616333, "train_loss_llm": 0.0, "grad_norm": 0.08084096759557724, "global_step": 11592, "epoch": 322, "lr": 0.001} +{"train_loss": 0.034246522933244705, "train_loss_bc": 0.034246522933244705, "train_loss_llm": 0.0, "grad_norm": 0.018547769635915756, "global_step": 11593, "epoch": 322, "lr": 0.001} +{"train_loss": 0.0381118543446064, "train_loss_bc": 0.0381118543446064, "train_loss_llm": 0.0, "grad_norm": 0.027817167341709137, "global_step": 11594, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03409311920404434, "train_loss_bc": 0.03408895060420036, "train_loss_llm": 0.00041672587394714355, "grad_norm": 0.04005890339612961, "global_step": 11595, "epoch": 322, "lr": 0.001} +{"train_loss": 0.034949444234371185, "train_loss_bc": 0.03494526445865631, "train_loss_llm": 0.00041788816452026367, "grad_norm": 0.04521116986870766, "global_step": 11596, "epoch": 322, "lr": 0.001} +{"train_loss": 0.036437127739191055, "train_loss_bc": 0.036437127739191055, "train_loss_llm": 0.0, "grad_norm": 0.017205417156219482, "global_step": 11597, "epoch": 322, "lr": 0.001} +{"train_loss": 0.038211386650800705, "train_loss_bc": 0.038211386650800705, "train_loss_llm": 0.0, "grad_norm": 0.03433440625667572, "global_step": 11598, "epoch": 322, "lr": 0.001} +{"train_loss": 0.035389646887779236, "train_loss_bc": 0.035389646887779236, "train_loss_llm": 0.0, "grad_norm": 0.04466156288981438, "global_step": 11599, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03994830697774887, "train_loss_bc": 0.03994830697774887, "train_loss_llm": 0.0, "grad_norm": 0.04897984862327576, "global_step": 11600, "epoch": 322, "lr": 0.001} +{"train_loss": 0.04230385273694992, "train_loss_bc": 0.04230385273694992, "train_loss_llm": 0.0, "grad_norm": 0.02248777449131012, "global_step": 11601, "epoch": 322, "lr": 0.001} +{"train_loss": 0.0359591543674469, "train_loss_bc": 0.0359351821243763, "train_loss_llm": 0.00239717960357666, "grad_norm": 0.029760532081127167, "global_step": 11602, "epoch": 322, "lr": 0.001} +{"train_loss": 0.042321305721998215, "train_loss_bc": 0.04230186715722084, "train_loss_llm": 0.0019439160823822021, "grad_norm": 0.04118926450610161, "global_step": 11603, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03177151083946228, "train_loss_bc": 0.031764015555381775, "train_loss_llm": 0.0007496476173400879, "grad_norm": 0.049772921949625015, "global_step": 11604, "epoch": 322, "lr": 0.001} +{"train_loss": 0.04230732470750809, "train_loss_bc": 0.04230732470750809, "train_loss_llm": 0.0, "grad_norm": 0.02737630158662796, "global_step": 11605, "epoch": 322, "lr": 0.001} +{"train_loss": 0.036625202745199203, "train_loss_bc": 0.03661526367068291, "train_loss_llm": 0.000993967056274414, "grad_norm": 0.0486300103366375, "global_step": 11606, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03762011602520943, "train_loss_bc": 0.03762011602520943, "train_loss_llm": 0.0, "grad_norm": 0.06388087570667267, "global_step": 11607, "epoch": 322, "lr": 0.001} +{"train_loss": 0.031621940433979034, "train_loss_bc": 0.031620170921087265, "train_loss_llm": 0.00017687678337097168, "grad_norm": 0.07107588648796082, "global_step": 11608, "epoch": 322, "lr": 0.001} +{"train_loss": 0.041920892894268036, "train_loss_bc": 0.041920892894268036, "train_loss_llm": 0.0, "grad_norm": 0.02427000366151333, "global_step": 11609, "epoch": 322, "lr": 0.001} +{"train_loss": 0.0394447036087513, "train_loss_bc": 0.0394447036087513, "train_loss_llm": 0.0, "grad_norm": 0.032865751534700394, "global_step": 11610, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03872188925743103, "train_loss_bc": 0.03872188925743103, "train_loss_llm": 0.0, "grad_norm": 0.046102311462163925, "global_step": 11611, "epoch": 322, "lr": 0.001} +{"train_loss": 0.033975500613451004, "train_loss_bc": 0.033975500613451004, "train_loss_llm": 0.0, "grad_norm": 0.05621680989861488, "global_step": 11612, "epoch": 322, "lr": 0.001} +{"train_loss": 0.039951007813215256, "train_loss_bc": 0.039951007813215256, "train_loss_llm": 0.0, "grad_norm": 0.023375840857625008, "global_step": 11613, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03430578112602234, "train_loss_bc": 0.03430578112602234, "train_loss_llm": 0.0, "grad_norm": 0.03357807919383049, "global_step": 11614, "epoch": 322, "lr": 0.001} +{"train_loss": 0.031023897230625153, "train_loss_bc": 0.031023897230625153, "train_loss_llm": 0.0, "grad_norm": 0.043773800134658813, "global_step": 11615, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03900308161973953, "train_loss_bc": 0.03900308161973953, "train_loss_llm": 0.0, "grad_norm": 0.05317315086722374, "global_step": 11616, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03762434050440788, "train_loss_bc": 0.037604231387376785, "train_loss_llm": 0.002010732889175415, "grad_norm": 0.021956810727715492, "global_step": 11617, "epoch": 322, "lr": 0.001} +{"train_loss": 0.04597802460193634, "train_loss_bc": 0.04597802460193634, "train_loss_llm": 0.0, "grad_norm": 0.04028766602277756, "global_step": 11618, "epoch": 322, "lr": 0.001} +{"train_loss": 0.035063162446022034, "train_loss_bc": 0.03506019711494446, "train_loss_llm": 0.00029671192169189453, "grad_norm": 0.05107327923178673, "global_step": 11619, "epoch": 322, "lr": 0.001} +{"train_loss": 0.04388977959752083, "train_loss_bc": 0.04388977959752083, "train_loss_llm": 0.0, "grad_norm": 0.06307344138622284, "global_step": 11620, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03530719131231308, "train_loss_bc": 0.03530719131231308, "train_loss_llm": 0.0, "grad_norm": 0.021106846630573273, "global_step": 11621, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03737321496009827, "train_loss_bc": 0.03737321496009827, "train_loss_llm": 0.0, "grad_norm": 0.03069642186164856, "global_step": 11622, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03324487432837486, "train_loss_bc": 0.03323320299386978, "train_loss_llm": 0.0011671483516693115, "grad_norm": 0.041707076132297516, "global_step": 11623, "epoch": 322, "lr": 0.001} +{"train_loss": 0.0400330051779747, "train_loss_bc": 0.0400330051779747, "train_loss_llm": 0.0, "grad_norm": 0.05072339251637459, "global_step": 11624, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03390241041779518, "train_loss_bc": 0.03389737010002136, "train_loss_llm": 0.0005041658878326416, "grad_norm": 0.02013033628463745, "global_step": 11625, "epoch": 322, "lr": 0.001} +{"train_loss": 0.03771928697824478, "train_loss_bc": 0.03771928697824478, "train_loss_llm": 0.0, "grad_norm": 0.04233977198600769, "global_step": 11626, "epoch": 322, "lr": 0.001} +{"train_loss": 0.037510418333113194, "train_loss_bc": 0.037669431418180466, "train_loss_llm": 0.0, "grad_norm": 0.05964232236146927, "global_step": 11627, "epoch": 322, "lr": 0.001, "val_loss": 0.06433834880590439} +{"train_loss": 0.04164135828614235, "train_loss_bc": 0.04164135828614235, "train_loss_llm": 0.0, "grad_norm": 0.07816524803638458, "global_step": 11628, "epoch": 323, "lr": 0.001} +{"train_loss": 0.029217958450317383, "train_loss_bc": 0.02921588160097599, "train_loss_llm": 0.00020772218704223633, "grad_norm": 0.0188293494284153, "global_step": 11629, "epoch": 323, "lr": 0.001} +{"train_loss": 0.038891080766916275, "train_loss_bc": 0.038891080766916275, "train_loss_llm": 0.0, "grad_norm": 0.032729621976614, "global_step": 11630, "epoch": 323, "lr": 0.001} +{"train_loss": 0.042192984372377396, "train_loss_bc": 0.04217412695288658, "train_loss_llm": 0.0018858611583709717, "grad_norm": 0.04522500932216644, "global_step": 11631, "epoch": 323, "lr": 0.001} +{"train_loss": 0.035120345652103424, "train_loss_bc": 0.035120345652103424, "train_loss_llm": 0.0, "grad_norm": 0.05911753326654434, "global_step": 11632, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03656487539410591, "train_loss_bc": 0.036555107682943344, "train_loss_llm": 0.0009767115116119385, "grad_norm": 0.020155848935246468, "global_step": 11633, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03401617705821991, "train_loss_bc": 0.034009870141744614, "train_loss_llm": 0.0006306171417236328, "grad_norm": 0.023915432393550873, "global_step": 11634, "epoch": 323, "lr": 0.001} +{"train_loss": 0.04159923642873764, "train_loss_bc": 0.04159800335764885, "train_loss_llm": 0.0001232624053955078, "grad_norm": 0.0339212529361248, "global_step": 11635, "epoch": 323, "lr": 0.001} +{"train_loss": 0.038983169943094254, "train_loss_bc": 0.038983169943094254, "train_loss_llm": 0.0, "grad_norm": 0.0413428395986557, "global_step": 11636, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03890574350953102, "train_loss_bc": 0.03890574350953102, "train_loss_llm": 0.0, "grad_norm": 0.024267444387078285, "global_step": 11637, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03871096298098564, "train_loss_bc": 0.03871096298098564, "train_loss_llm": 0.0, "grad_norm": 0.039565883576869965, "global_step": 11638, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03364766761660576, "train_loss_bc": 0.03364766761660576, "train_loss_llm": 0.0, "grad_norm": 0.054495494812726974, "global_step": 11639, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03384881466627121, "train_loss_bc": 0.03384881466627121, "train_loss_llm": 0.0, "grad_norm": 0.06550712883472443, "global_step": 11640, "epoch": 323, "lr": 0.001} +{"train_loss": 0.028181958943605423, "train_loss_bc": 0.028175801038742065, "train_loss_llm": 0.0006158053874969482, "grad_norm": 0.0168023519217968, "global_step": 11641, "epoch": 323, "lr": 0.001} +{"train_loss": 0.04044557735323906, "train_loss_bc": 0.04044557735323906, "train_loss_llm": 0.0, "grad_norm": 0.02786978706717491, "global_step": 11642, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03276079520583153, "train_loss_bc": 0.03274928778409958, "train_loss_llm": 0.0011507272720336914, "grad_norm": 0.03749535605311394, "global_step": 11643, "epoch": 323, "lr": 0.001} +{"train_loss": 0.04265303537249565, "train_loss_bc": 0.04265303537249565, "train_loss_llm": 0.0, "grad_norm": 0.048683080822229385, "global_step": 11644, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03761932626366615, "train_loss_bc": 0.03759874030947685, "train_loss_llm": 0.002058565616607666, "grad_norm": 0.023252267390489578, "global_step": 11645, "epoch": 323, "lr": 0.001} +{"train_loss": 0.04618346318602562, "train_loss_bc": 0.04616179317235947, "train_loss_llm": 0.00216713547706604, "grad_norm": 0.03289664164185524, "global_step": 11646, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03379791975021362, "train_loss_bc": 0.03378475829958916, "train_loss_llm": 0.0013162791728973389, "grad_norm": 0.04697858914732933, "global_step": 11647, "epoch": 323, "lr": 0.001} +{"train_loss": 0.0376964770257473, "train_loss_bc": 0.0376964770257473, "train_loss_llm": 0.0, "grad_norm": 0.0618513785302639, "global_step": 11648, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03448185697197914, "train_loss_bc": 0.03448185697197914, "train_loss_llm": 0.0, "grad_norm": 0.02240406721830368, "global_step": 11649, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03754887357354164, "train_loss_bc": 0.0375431664288044, "train_loss_llm": 0.0005705654621124268, "grad_norm": 0.032102297991514206, "global_step": 11650, "epoch": 323, "lr": 0.001} +{"train_loss": 0.040663301944732666, "train_loss_bc": 0.040663301944732666, "train_loss_llm": 0.0, "grad_norm": 0.04077474772930145, "global_step": 11651, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03455556556582451, "train_loss_bc": 0.03455556556582451, "train_loss_llm": 0.0, "grad_norm": 0.05187100172042847, "global_step": 11652, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03146400302648544, "train_loss_bc": 0.03146400302648544, "train_loss_llm": 0.0, "grad_norm": 0.017966944724321365, "global_step": 11653, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03314900025725365, "train_loss_bc": 0.03314900025725365, "train_loss_llm": 0.0, "grad_norm": 0.028811858966946602, "global_step": 11654, "epoch": 323, "lr": 0.001} +{"train_loss": 0.04593394696712494, "train_loss_bc": 0.04593394696712494, "train_loss_llm": 0.0, "grad_norm": 0.04187554866075516, "global_step": 11655, "epoch": 323, "lr": 0.001} +{"train_loss": 0.04024634510278702, "train_loss_bc": 0.04024634510278702, "train_loss_llm": 0.0, "grad_norm": 0.05689743533730507, "global_step": 11656, "epoch": 323, "lr": 0.001} +{"train_loss": 0.035161927342414856, "train_loss_bc": 0.035161927342414856, "train_loss_llm": 0.0, "grad_norm": 0.02385692484676838, "global_step": 11657, "epoch": 323, "lr": 0.001} +{"train_loss": 0.04337500035762787, "train_loss_bc": 0.043356675654649734, "train_loss_llm": 0.0018323063850402832, "grad_norm": 0.033313240855932236, "global_step": 11658, "epoch": 323, "lr": 0.001} +{"train_loss": 0.034285884350538254, "train_loss_bc": 0.034285884350538254, "train_loss_llm": 0.0, "grad_norm": 0.04301795735955238, "global_step": 11659, "epoch": 323, "lr": 0.001} +{"train_loss": 0.04055742919445038, "train_loss_bc": 0.04054972901940346, "train_loss_llm": 0.0007699429988861084, "grad_norm": 0.05267376825213432, "global_step": 11660, "epoch": 323, "lr": 0.001} +{"train_loss": 0.04106166213750839, "train_loss_bc": 0.041050221771001816, "train_loss_llm": 0.0011439621448516846, "grad_norm": 0.02556980773806572, "global_step": 11661, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03709058836102486, "train_loss_bc": 0.03709058836102486, "train_loss_llm": 0.0, "grad_norm": 0.032768409699201584, "global_step": 11662, "epoch": 323, "lr": 0.001} +{"train_loss": 0.03736340906471014, "train_loss_bc": 0.03282841295003891, "train_loss_llm": 0.0, "grad_norm": 0.0503111258149147, "global_step": 11663, "epoch": 323, "lr": 0.001, "val_loss": 0.06297864019870758} +{"train_loss": 0.034595273435115814, "train_loss_bc": 0.034588854759931564, "train_loss_llm": 0.0006417632102966309, "grad_norm": 0.05699067935347557, "global_step": 11664, "epoch": 324, "lr": 0.001} +{"train_loss": 0.044064510613679886, "train_loss_bc": 0.044064510613679886, "train_loss_llm": 0.0, "grad_norm": 0.01982964761555195, "global_step": 11665, "epoch": 324, "lr": 0.001} +{"train_loss": 0.04209050536155701, "train_loss_bc": 0.04209050536155701, "train_loss_llm": 0.0, "grad_norm": 0.028491422533988953, "global_step": 11666, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03258020058274269, "train_loss_bc": 0.032555438578128815, "train_loss_llm": 0.002476334571838379, "grad_norm": 0.03582267090678215, "global_step": 11667, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03855465352535248, "train_loss_bc": 0.03855465352535248, "train_loss_llm": 0.0, "grad_norm": 0.0428210012614727, "global_step": 11668, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03354400396347046, "train_loss_bc": 0.03354400396347046, "train_loss_llm": 0.0, "grad_norm": 0.019378457218408585, "global_step": 11669, "epoch": 324, "lr": 0.001} +{"train_loss": 0.037782520055770874, "train_loss_bc": 0.037782520055770874, "train_loss_llm": 0.0, "grad_norm": 0.03309513255953789, "global_step": 11670, "epoch": 324, "lr": 0.001} +{"train_loss": 0.038060255348682404, "train_loss_bc": 0.038060255348682404, "train_loss_llm": 0.0, "grad_norm": 0.03725031390786171, "global_step": 11671, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03278907760977745, "train_loss_bc": 0.03277985006570816, "train_loss_llm": 0.000922858715057373, "grad_norm": 0.04036669433116913, "global_step": 11672, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03897593915462494, "train_loss_bc": 0.03897443041205406, "train_loss_llm": 0.00015082955360412598, "grad_norm": 0.016535624861717224, "global_step": 11673, "epoch": 324, "lr": 0.001} +{"train_loss": 0.04233304783701897, "train_loss_bc": 0.04232882708311081, "train_loss_llm": 0.00042191147804260254, "grad_norm": 0.028589170426130295, "global_step": 11674, "epoch": 324, "lr": 0.001} +{"train_loss": 0.031648773699998856, "train_loss_bc": 0.031648773699998856, "train_loss_llm": 0.0, "grad_norm": 0.034626416862010956, "global_step": 11675, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03556869179010391, "train_loss_bc": 0.03556869179010391, "train_loss_llm": 0.0, "grad_norm": 0.044378116726875305, "global_step": 11676, "epoch": 324, "lr": 0.001} +{"train_loss": 0.035044752061367035, "train_loss_bc": 0.035044752061367035, "train_loss_llm": 0.0, "grad_norm": 0.022278064861893654, "global_step": 11677, "epoch": 324, "lr": 0.001} +{"train_loss": 0.0378984659910202, "train_loss_bc": 0.0378984659910202, "train_loss_llm": 0.0, "grad_norm": 0.034786008298397064, "global_step": 11678, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03268111124634743, "train_loss_bc": 0.032655782997608185, "train_loss_llm": 0.002532780170440674, "grad_norm": 0.042699817568063736, "global_step": 11679, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03332678973674774, "train_loss_bc": 0.033321745693683624, "train_loss_llm": 0.0005044341087341309, "grad_norm": 0.0507255420088768, "global_step": 11680, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03668878972530365, "train_loss_bc": 0.03668878972530365, "train_loss_llm": 0.0, "grad_norm": 0.021354809403419495, "global_step": 11681, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03890993446111679, "train_loss_bc": 0.03890993446111679, "train_loss_llm": 0.0, "grad_norm": 0.03225792199373245, "global_step": 11682, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03791562467813492, "train_loss_bc": 0.03790959715843201, "train_loss_llm": 0.0006026029586791992, "grad_norm": 0.046797946095466614, "global_step": 11683, "epoch": 324, "lr": 0.001} +{"train_loss": 0.04146350547671318, "train_loss_bc": 0.04146350547671318, "train_loss_llm": 0.0, "grad_norm": 0.05978098511695862, "global_step": 11684, "epoch": 324, "lr": 0.001} +{"train_loss": 0.04051760211586952, "train_loss_bc": 0.04051760211586952, "train_loss_llm": 0.0, "grad_norm": 0.016811702400445938, "global_step": 11685, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03712642565369606, "train_loss_bc": 0.03712642565369606, "train_loss_llm": 0.0, "grad_norm": 0.025491224601864815, "global_step": 11686, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03953436389565468, "train_loss_bc": 0.03952956572175026, "train_loss_llm": 0.00047972798347473145, "grad_norm": 0.03336687758564949, "global_step": 11687, "epoch": 324, "lr": 0.001} +{"train_loss": 0.0347885861992836, "train_loss_bc": 0.0347885861992836, "train_loss_llm": 0.0, "grad_norm": 0.0435778833925724, "global_step": 11688, "epoch": 324, "lr": 0.001} +{"train_loss": 0.032602354884147644, "train_loss_bc": 0.032602354884147644, "train_loss_llm": 0.0, "grad_norm": 0.018621211871504784, "global_step": 11689, "epoch": 324, "lr": 0.001} +{"train_loss": 0.0395001582801342, "train_loss_bc": 0.0395001582801342, "train_loss_llm": 0.0, "grad_norm": 0.029199007898569107, "global_step": 11690, "epoch": 324, "lr": 0.001} +{"train_loss": 0.036140456795692444, "train_loss_bc": 0.036140456795692444, "train_loss_llm": 0.0, "grad_norm": 0.03819321095943451, "global_step": 11691, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03623583912849426, "train_loss_bc": 0.03622747212648392, "train_loss_llm": 0.0008365213871002197, "grad_norm": 0.04656009376049042, "global_step": 11692, "epoch": 324, "lr": 0.001} +{"train_loss": 0.041275687515735626, "train_loss_bc": 0.041275687515735626, "train_loss_llm": 0.0, "grad_norm": 0.024887315928936005, "global_step": 11693, "epoch": 324, "lr": 0.001} +{"train_loss": 0.04265781119465828, "train_loss_bc": 0.04265781119465828, "train_loss_llm": 0.0, "grad_norm": 0.03504079952836037, "global_step": 11694, "epoch": 324, "lr": 0.001} +{"train_loss": 0.037229735404253006, "train_loss_bc": 0.0372106097638607, "train_loss_llm": 0.0019125044345855713, "grad_norm": 0.035970915108919144, "global_step": 11695, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03486625477671623, "train_loss_bc": 0.03486625477671623, "train_loss_llm": 0.0, "grad_norm": 0.050696004182100296, "global_step": 11696, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03346940875053406, "train_loss_bc": 0.03346940875053406, "train_loss_llm": 0.0, "grad_norm": 0.020588157698512077, "global_step": 11697, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03591133654117584, "train_loss_bc": 0.03591133654117584, "train_loss_llm": 0.0, "grad_norm": 0.0341489240527153, "global_step": 11698, "epoch": 324, "lr": 0.001} +{"train_loss": 0.03738987259566784, "train_loss_bc": 0.04762907698750496, "train_loss_llm": 0.0033888518810272217, "grad_norm": 0.04988398775458336, "global_step": 11699, "epoch": 324, "lr": 0.001, "val_loss": 0.06310998648405075} +{"train_loss": 0.03516294062137604, "train_loss_bc": 0.03515557199716568, "train_loss_llm": 0.0007369518280029297, "grad_norm": 0.05911216139793396, "global_step": 11700, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03195442631840706, "train_loss_bc": 0.03194613382220268, "train_loss_llm": 0.0008292198181152344, "grad_norm": 0.018266385421156883, "global_step": 11701, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03344842046499252, "train_loss_bc": 0.03344842046499252, "train_loss_llm": 0.0, "grad_norm": 0.033952198922634125, "global_step": 11702, "epoch": 325, "lr": 0.001} +{"train_loss": 0.04357103258371353, "train_loss_bc": 0.04357103258371353, "train_loss_llm": 0.0, "grad_norm": 0.04513933137059212, "global_step": 11703, "epoch": 325, "lr": 0.001} +{"train_loss": 0.038526393473148346, "train_loss_bc": 0.03852579742670059, "train_loss_llm": 5.957484245300293e-05, "grad_norm": 0.056216299533843994, "global_step": 11704, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03543023392558098, "train_loss_bc": 0.03543023392558098, "train_loss_llm": 0.0, "grad_norm": 0.01994016021490097, "global_step": 11705, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03793859854340553, "train_loss_bc": 0.03793859854340553, "train_loss_llm": 0.0, "grad_norm": 0.02822123095393181, "global_step": 11706, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03561166301369667, "train_loss_bc": 0.03559472784399986, "train_loss_llm": 0.0016934573650360107, "grad_norm": 0.034720513969659805, "global_step": 11707, "epoch": 325, "lr": 0.001} +{"train_loss": 0.042585182934999466, "train_loss_bc": 0.042585182934999466, "train_loss_llm": 0.0, "grad_norm": 0.04397570714354515, "global_step": 11708, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03788210079073906, "train_loss_bc": 0.03788210079073906, "train_loss_llm": 0.0, "grad_norm": 0.01584850251674652, "global_step": 11709, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03969992324709892, "train_loss_bc": 0.03967730328440666, "train_loss_llm": 0.0022621750831604004, "grad_norm": 0.02368205599486828, "global_step": 11710, "epoch": 325, "lr": 0.001} +{"train_loss": 0.036917995661497116, "train_loss_bc": 0.036917995661497116, "train_loss_llm": 0.0, "grad_norm": 0.03723819553852081, "global_step": 11711, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03843791037797928, "train_loss_bc": 0.03843791037797928, "train_loss_llm": 0.0, "grad_norm": 0.04335948824882507, "global_step": 11712, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03818317502737045, "train_loss_bc": 0.03817574307322502, "train_loss_llm": 0.0007433593273162842, "grad_norm": 0.022670850157737732, "global_step": 11713, "epoch": 325, "lr": 0.001} +{"train_loss": 0.0328722782433033, "train_loss_bc": 0.0328722782433033, "train_loss_llm": 0.0, "grad_norm": 0.029133925214409828, "global_step": 11714, "epoch": 325, "lr": 0.001} +{"train_loss": 0.04067935422062874, "train_loss_bc": 0.04067124053835869, "train_loss_llm": 0.0008113086223602295, "grad_norm": 0.04493206366896629, "global_step": 11715, "epoch": 325, "lr": 0.001} +{"train_loss": 0.043606407940387726, "train_loss_bc": 0.043602317571640015, "train_loss_llm": 0.0004089474678039551, "grad_norm": 0.05542167276144028, "global_step": 11716, "epoch": 325, "lr": 0.001} +{"train_loss": 0.0307689867913723, "train_loss_bc": 0.030759243294596672, "train_loss_llm": 0.0009742975234985352, "grad_norm": 0.021424926817417145, "global_step": 11717, "epoch": 325, "lr": 0.001} +{"train_loss": 0.04167706519365311, "train_loss_bc": 0.04167706519365311, "train_loss_llm": 0.0, "grad_norm": 0.0356595516204834, "global_step": 11718, "epoch": 325, "lr": 0.001} +{"train_loss": 0.0353909470140934, "train_loss_bc": 0.03538822382688522, "train_loss_llm": 0.0002721846103668213, "grad_norm": 0.044827885925769806, "global_step": 11719, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03325022757053375, "train_loss_bc": 0.03325022757053375, "train_loss_llm": 0.0, "grad_norm": 0.05581064522266388, "global_step": 11720, "epoch": 325, "lr": 0.001} +{"train_loss": 0.035450588911771774, "train_loss_bc": 0.03544696420431137, "train_loss_llm": 0.0003623366355895996, "grad_norm": 0.022943036630749702, "global_step": 11721, "epoch": 325, "lr": 0.001} +{"train_loss": 0.04004104807972908, "train_loss_bc": 0.04004104807972908, "train_loss_llm": 0.0, "grad_norm": 0.030754584819078445, "global_step": 11722, "epoch": 325, "lr": 0.001} +{"train_loss": 0.044818103313446045, "train_loss_bc": 0.04478227347135544, "train_loss_llm": 0.0035830438137054443, "grad_norm": 0.03887144848704338, "global_step": 11723, "epoch": 325, "lr": 0.001} +{"train_loss": 0.025848712772130966, "train_loss_bc": 0.025848712772130966, "train_loss_llm": 0.0, "grad_norm": 0.04637763649225235, "global_step": 11724, "epoch": 325, "lr": 0.001} +{"train_loss": 0.036033887416124344, "train_loss_bc": 0.036033887416124344, "train_loss_llm": 0.0, "grad_norm": 0.020455142483115196, "global_step": 11725, "epoch": 325, "lr": 0.001} +{"train_loss": 0.04028240218758583, "train_loss_bc": 0.04028240218758583, "train_loss_llm": 0.0, "grad_norm": 0.036031559109687805, "global_step": 11726, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03101370297372341, "train_loss_bc": 0.03101370297372341, "train_loss_llm": 0.0, "grad_norm": 0.03757248818874359, "global_step": 11727, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03846210241317749, "train_loss_bc": 0.03844457492232323, "train_loss_llm": 0.0017529129981994629, "grad_norm": 0.047676343470811844, "global_step": 11728, "epoch": 325, "lr": 0.001} +{"train_loss": 0.0367935448884964, "train_loss_bc": 0.0367935448884964, "train_loss_llm": 0.0, "grad_norm": 0.021312350407242775, "global_step": 11729, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03631490096449852, "train_loss_bc": 0.03631490096449852, "train_loss_llm": 0.0, "grad_norm": 0.03365037217736244, "global_step": 11730, "epoch": 325, "lr": 0.001} +{"train_loss": 0.04433903098106384, "train_loss_bc": 0.04433903098106384, "train_loss_llm": 0.0, "grad_norm": 0.049081966280937195, "global_step": 11731, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03728385642170906, "train_loss_bc": 0.03728385642170906, "train_loss_llm": 0.0, "grad_norm": 0.056904278695583344, "global_step": 11732, "epoch": 325, "lr": 0.001} +{"train_loss": 0.03987296670675278, "train_loss_bc": 0.03987296670675278, "train_loss_llm": 0.0, "grad_norm": 0.02734094113111496, "global_step": 11733, "epoch": 325, "lr": 0.001} +{"train_loss": 0.035208821296691895, "train_loss_bc": 0.035208821296691895, "train_loss_llm": 0.0, "grad_norm": 0.03829742223024368, "global_step": 11734, "epoch": 325, "lr": 0.001} +{"train_loss": 0.037292291772448354, "train_loss_bc": 0.037163570523262024, "train_loss_llm": 0.0, "grad_norm": 0.04924331232905388, "global_step": 11735, "epoch": 325, "lr": 0.001, "train/cumulative_reward": 78.83949724063336, "train/mean_score": 12.633810682744098, "train/success_rate": 0.3, "test/cumulative_reward": 41.683822830371106, "test/mean_score": 7.499623058205985, "test/success_rate": 0.18, "val_loss": 0.06362195312976837, "train_action_mse_error": 0.040683645755052567} +{"train_loss": 0.03133007138967514, "train_loss_bc": 0.03133007138967514, "train_loss_llm": 0.0, "grad_norm": 0.05060510337352753, "global_step": 11736, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03107193298637867, "train_loss_bc": 0.03107193298637867, "train_loss_llm": 0.0, "grad_norm": 0.020078659057617188, "global_step": 11737, "epoch": 326, "lr": 0.001} +{"train_loss": 0.0421685129404068, "train_loss_bc": 0.04215634986758232, "train_loss_llm": 0.001216292381286621, "grad_norm": 0.030108919367194176, "global_step": 11738, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03250371664762497, "train_loss_bc": 0.03250371664762497, "train_loss_llm": 0.0, "grad_norm": 0.03474009409546852, "global_step": 11739, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03638964518904686, "train_loss_bc": 0.03638964518904686, "train_loss_llm": 0.0, "grad_norm": 0.0431179404258728, "global_step": 11740, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03361940383911133, "train_loss_bc": 0.03361940383911133, "train_loss_llm": 0.0, "grad_norm": 0.01596830040216446, "global_step": 11741, "epoch": 326, "lr": 0.001} +{"train_loss": 0.034670066088438034, "train_loss_bc": 0.03465387225151062, "train_loss_llm": 0.0016192793846130371, "grad_norm": 0.022461486980319023, "global_step": 11742, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03880997374653816, "train_loss_bc": 0.03880997374653816, "train_loss_llm": 0.0, "grad_norm": 0.03777821362018585, "global_step": 11743, "epoch": 326, "lr": 0.001} +{"train_loss": 0.037181321531534195, "train_loss_bc": 0.037181321531534195, "train_loss_llm": 0.0, "grad_norm": 0.04678894206881523, "global_step": 11744, "epoch": 326, "lr": 0.001} +{"train_loss": 0.034099508076906204, "train_loss_bc": 0.034099508076906204, "train_loss_llm": 0.0, "grad_norm": 0.01846248470246792, "global_step": 11745, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03837081044912338, "train_loss_bc": 0.03836531192064285, "train_loss_llm": 0.0005499422550201416, "grad_norm": 0.033115942031145096, "global_step": 11746, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03342973440885544, "train_loss_bc": 0.03342973440885544, "train_loss_llm": 0.0, "grad_norm": 0.035632867366075516, "global_step": 11747, "epoch": 326, "lr": 0.001} +{"train_loss": 0.038612715899944305, "train_loss_bc": 0.038612715899944305, "train_loss_llm": 0.0, "grad_norm": 0.059965815395116806, "global_step": 11748, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03581087663769722, "train_loss_bc": 0.03581087663769722, "train_loss_llm": 0.0, "grad_norm": 0.017805906012654305, "global_step": 11749, "epoch": 326, "lr": 0.001} +{"train_loss": 0.037325166165828705, "train_loss_bc": 0.03729701787233353, "train_loss_llm": 0.00281485915184021, "grad_norm": 0.028774255886673927, "global_step": 11750, "epoch": 326, "lr": 0.001} +{"train_loss": 0.04031234234571457, "train_loss_bc": 0.04031234234571457, "train_loss_llm": 0.0, "grad_norm": 0.03523840010166168, "global_step": 11751, "epoch": 326, "lr": 0.001} +{"train_loss": 0.0370715893805027, "train_loss_bc": 0.0370715893805027, "train_loss_llm": 0.0, "grad_norm": 0.042307354509830475, "global_step": 11752, "epoch": 326, "lr": 0.001} +{"train_loss": 0.04191169515252113, "train_loss_bc": 0.04191169515252113, "train_loss_llm": 0.0, "grad_norm": 0.02282981388270855, "global_step": 11753, "epoch": 326, "lr": 0.001} +{"train_loss": 0.040528059005737305, "train_loss_bc": 0.040528059005737305, "train_loss_llm": 0.0, "grad_norm": 0.04093475639820099, "global_step": 11754, "epoch": 326, "lr": 0.001} +{"train_loss": 0.04788151755928993, "train_loss_bc": 0.04788151755928993, "train_loss_llm": 0.0, "grad_norm": 0.060248441994190216, "global_step": 11755, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03281495347619057, "train_loss_bc": 0.03277792036533356, "train_loss_llm": 0.003703176975250244, "grad_norm": 0.07077665627002716, "global_step": 11756, "epoch": 326, "lr": 0.001} +{"train_loss": 0.035650309175252914, "train_loss_bc": 0.035650309175252914, "train_loss_llm": 0.0, "grad_norm": 0.019197115674614906, "global_step": 11757, "epoch": 326, "lr": 0.001} +{"train_loss": 0.035604994744062424, "train_loss_bc": 0.035604994744062424, "train_loss_llm": 0.0, "grad_norm": 0.035569727420806885, "global_step": 11758, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03374016284942627, "train_loss_bc": 0.03374016284942627, "train_loss_llm": 0.0, "grad_norm": 0.04614442586898804, "global_step": 11759, "epoch": 326, "lr": 0.001} +{"train_loss": 0.050679657608270645, "train_loss_bc": 0.050679657608270645, "train_loss_llm": 0.0, "grad_norm": 0.058603767305612564, "global_step": 11760, "epoch": 326, "lr": 0.001} +{"train_loss": 0.043096236884593964, "train_loss_bc": 0.043096236884593964, "train_loss_llm": 0.0, "grad_norm": 0.021028848364949226, "global_step": 11761, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03458884358406067, "train_loss_bc": 0.03458530828356743, "train_loss_llm": 0.000353395938873291, "grad_norm": 0.03523770719766617, "global_step": 11762, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03674555942416191, "train_loss_bc": 0.03674555942416191, "train_loss_llm": 0.0, "grad_norm": 0.050686608999967575, "global_step": 11763, "epoch": 326, "lr": 0.001} +{"train_loss": 0.04141131415963173, "train_loss_bc": 0.04141131415963173, "train_loss_llm": 0.0, "grad_norm": 0.06384016573429108, "global_step": 11764, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03884527459740639, "train_loss_bc": 0.03884527459740639, "train_loss_llm": 0.0, "grad_norm": 0.022370368242263794, "global_step": 11765, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03560628741979599, "train_loss_bc": 0.03560442477464676, "train_loss_llm": 0.0001862943172454834, "grad_norm": 0.028728609904646873, "global_step": 11766, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03938175365328789, "train_loss_bc": 0.03938175365328789, "train_loss_llm": 0.0, "grad_norm": 0.034731727093458176, "global_step": 11767, "epoch": 326, "lr": 0.001} +{"train_loss": 0.04062888026237488, "train_loss_bc": 0.04062888026237488, "train_loss_llm": 0.0, "grad_norm": 0.04448502138257027, "global_step": 11768, "epoch": 326, "lr": 0.001} +{"train_loss": 0.033886756747961044, "train_loss_bc": 0.033886756747961044, "train_loss_llm": 0.0, "grad_norm": 0.01744147762656212, "global_step": 11769, "epoch": 326, "lr": 0.001} +{"train_loss": 0.034140199422836304, "train_loss_bc": 0.034140199422836304, "train_loss_llm": 0.0, "grad_norm": 0.030286654829978943, "global_step": 11770, "epoch": 326, "lr": 0.001} +{"train_loss": 0.03713569821168979, "train_loss_bc": 0.026965292170643806, "train_loss_llm": 0.0, "grad_norm": 0.038532890379428864, "global_step": 11771, "epoch": 326, "lr": 0.001, "val_loss": 0.06332141906023026} +{"train_loss": 0.039404530078172684, "train_loss_bc": 0.039404530078172684, "train_loss_llm": 0.0, "grad_norm": 0.04710766300559044, "global_step": 11772, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03400280699133873, "train_loss_bc": 0.03400280699133873, "train_loss_llm": 0.0, "grad_norm": 0.01811894401907921, "global_step": 11773, "epoch": 327, "lr": 0.001} +{"train_loss": 0.039240606129169464, "train_loss_bc": 0.039240606129169464, "train_loss_llm": 0.0, "grad_norm": 0.030401675030589104, "global_step": 11774, "epoch": 327, "lr": 0.001} +{"train_loss": 0.041241519153118134, "train_loss_bc": 0.04123800992965698, "train_loss_llm": 0.0003511011600494385, "grad_norm": 0.037119120359420776, "global_step": 11775, "epoch": 327, "lr": 0.001} +{"train_loss": 0.037295594811439514, "train_loss_bc": 0.037295594811439514, "train_loss_llm": 0.0, "grad_norm": 0.042807769030332565, "global_step": 11776, "epoch": 327, "lr": 0.001} +{"train_loss": 0.032723333686590195, "train_loss_bc": 0.032718461006879807, "train_loss_llm": 0.00048726797103881836, "grad_norm": 0.022379247471690178, "global_step": 11777, "epoch": 327, "lr": 0.001} +{"train_loss": 0.034762442111968994, "train_loss_bc": 0.034762442111968994, "train_loss_llm": 0.0, "grad_norm": 0.036524977535009384, "global_step": 11778, "epoch": 327, "lr": 0.001} +{"train_loss": 0.032887522131204605, "train_loss_bc": 0.032887522131204605, "train_loss_llm": 0.0, "grad_norm": 0.0519140250980854, "global_step": 11779, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03809709846973419, "train_loss_bc": 0.03809709846973419, "train_loss_llm": 0.0, "grad_norm": 0.05894586816430092, "global_step": 11780, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03861255198717117, "train_loss_bc": 0.038612376898527145, "train_loss_llm": 1.7434358596801758e-05, "grad_norm": 0.024213450029492378, "global_step": 11781, "epoch": 327, "lr": 0.001} +{"train_loss": 0.038357362151145935, "train_loss_bc": 0.038357362151145935, "train_loss_llm": 0.0, "grad_norm": 0.036400772631168365, "global_step": 11782, "epoch": 327, "lr": 0.001} +{"train_loss": 0.037288665771484375, "train_loss_bc": 0.03727667033672333, "train_loss_llm": 0.0011997222900390625, "grad_norm": 0.042537614703178406, "global_step": 11783, "epoch": 327, "lr": 0.001} +{"train_loss": 0.042947717010974884, "train_loss_bc": 0.042947717010974884, "train_loss_llm": 0.0, "grad_norm": 0.05510108917951584, "global_step": 11784, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03570423275232315, "train_loss_bc": 0.03570423275232315, "train_loss_llm": 0.0, "grad_norm": 0.018394771963357925, "global_step": 11785, "epoch": 327, "lr": 0.001} +{"train_loss": 0.04912251979112625, "train_loss_bc": 0.04912251979112625, "train_loss_llm": 0.0, "grad_norm": 0.02993677742779255, "global_step": 11786, "epoch": 327, "lr": 0.001} +{"train_loss": 0.040000513195991516, "train_loss_bc": 0.040000513195991516, "train_loss_llm": 0.0, "grad_norm": 0.04231768846511841, "global_step": 11787, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03522243723273277, "train_loss_bc": 0.03522243723273277, "train_loss_llm": 0.0, "grad_norm": 0.05478567257523537, "global_step": 11788, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03589382767677307, "train_loss_bc": 0.03589382767677307, "train_loss_llm": 0.0, "grad_norm": 0.021045878529548645, "global_step": 11789, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03354768827557564, "train_loss_bc": 0.03354768827557564, "train_loss_llm": 0.0, "grad_norm": 0.0382545031607151, "global_step": 11790, "epoch": 327, "lr": 0.001} +{"train_loss": 0.04007044434547424, "train_loss_bc": 0.04007044434547424, "train_loss_llm": 0.0, "grad_norm": 0.05200841277837753, "global_step": 11791, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03637993335723877, "train_loss_bc": 0.03636965900659561, "train_loss_llm": 0.0010272860527038574, "grad_norm": 0.05980052053928375, "global_step": 11792, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03990473970770836, "train_loss_bc": 0.03990473970770836, "train_loss_llm": 0.0, "grad_norm": 0.022573722526431084, "global_step": 11793, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03591221943497658, "train_loss_bc": 0.03589535504579544, "train_loss_llm": 0.0016865432262420654, "grad_norm": 0.03045165352523327, "global_step": 11794, "epoch": 327, "lr": 0.001} +{"train_loss": 0.038310304284095764, "train_loss_bc": 0.03828929737210274, "train_loss_llm": 0.002100735902786255, "grad_norm": 0.04827718436717987, "global_step": 11795, "epoch": 327, "lr": 0.001} +{"train_loss": 0.042322590947151184, "train_loss_bc": 0.042322590947151184, "train_loss_llm": 0.0, "grad_norm": 0.05536019802093506, "global_step": 11796, "epoch": 327, "lr": 0.001} +{"train_loss": 0.031104013323783875, "train_loss_bc": 0.031104013323783875, "train_loss_llm": 0.0, "grad_norm": 0.019735164940357208, "global_step": 11797, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03819536417722702, "train_loss_bc": 0.03819536417722702, "train_loss_llm": 0.0, "grad_norm": 0.028070759028196335, "global_step": 11798, "epoch": 327, "lr": 0.001} +{"train_loss": 0.029916414991021156, "train_loss_bc": 0.029916414991021156, "train_loss_llm": 0.0, "grad_norm": 0.03740474954247475, "global_step": 11799, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03506655991077423, "train_loss_bc": 0.03506156802177429, "train_loss_llm": 0.000499039888381958, "grad_norm": 0.04826157167553902, "global_step": 11800, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03643844649195671, "train_loss_bc": 0.03640787675976753, "train_loss_llm": 0.0030570626258850098, "grad_norm": 0.01958545483648777, "global_step": 11801, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03763006627559662, "train_loss_bc": 0.03763006627559662, "train_loss_llm": 0.0, "grad_norm": 0.026133691892027855, "global_step": 11802, "epoch": 327, "lr": 0.001} +{"train_loss": 0.037681400775909424, "train_loss_bc": 0.037675950676202774, "train_loss_llm": 0.0005449056625366211, "grad_norm": 0.03273428604006767, "global_step": 11803, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03601936995983124, "train_loss_bc": 0.03601936995983124, "train_loss_llm": 0.0, "grad_norm": 0.04169825091958046, "global_step": 11804, "epoch": 327, "lr": 0.001} +{"train_loss": 0.035809628665447235, "train_loss_bc": 0.035809628665447235, "train_loss_llm": 0.0, "grad_norm": 0.02425391785800457, "global_step": 11805, "epoch": 327, "lr": 0.001} +{"train_loss": 0.03693177178502083, "train_loss_bc": 0.03693177178502083, "train_loss_llm": 0.0, "grad_norm": 0.034161221235990524, "global_step": 11806, "epoch": 327, "lr": 0.001} +{"train_loss": 0.037300512867255345, "train_loss_bc": 0.03877222537994385, "train_loss_llm": 0.0, "grad_norm": 0.044674959033727646, "global_step": 11807, "epoch": 327, "lr": 0.001, "val_loss": 0.06279570609331131} +{"train_loss": 0.037782277911901474, "train_loss_bc": 0.037782277911901474, "train_loss_llm": 0.0, "grad_norm": 0.05075676739215851, "global_step": 11808, "epoch": 328, "lr": 0.001} +{"train_loss": 0.02894764021039009, "train_loss_bc": 0.02894764021039009, "train_loss_llm": 0.0, "grad_norm": 0.01900053210556507, "global_step": 11809, "epoch": 328, "lr": 0.001} +{"train_loss": 0.036832116544246674, "train_loss_bc": 0.036826636642217636, "train_loss_llm": 0.0005480051040649414, "grad_norm": 0.031900521367788315, "global_step": 11810, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03829347342252731, "train_loss_bc": 0.03829347342252731, "train_loss_llm": 0.0, "grad_norm": 0.035656463354825974, "global_step": 11811, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03633423149585724, "train_loss_bc": 0.03633423149585724, "train_loss_llm": 0.0, "grad_norm": 0.03797478973865509, "global_step": 11812, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03778406232595444, "train_loss_bc": 0.03777984902262688, "train_loss_llm": 0.00042125582695007324, "grad_norm": 0.023686833679676056, "global_step": 11813, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03464767336845398, "train_loss_bc": 0.03464767336845398, "train_loss_llm": 0.0, "grad_norm": 0.03379322588443756, "global_step": 11814, "epoch": 328, "lr": 0.001} +{"train_loss": 0.0381796695291996, "train_loss_bc": 0.0381796695291996, "train_loss_llm": 0.0, "grad_norm": 0.0461636446416378, "global_step": 11815, "epoch": 328, "lr": 0.001} +{"train_loss": 0.038514696061611176, "train_loss_bc": 0.038514696061611176, "train_loss_llm": 0.0, "grad_norm": 0.055587396025657654, "global_step": 11816, "epoch": 328, "lr": 0.001} +{"train_loss": 0.036652080714702606, "train_loss_bc": 0.03664638102054596, "train_loss_llm": 0.0005699396133422852, "grad_norm": 0.020770801231265068, "global_step": 11817, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03685731068253517, "train_loss_bc": 0.03682010993361473, "train_loss_llm": 0.0037199556827545166, "grad_norm": 0.029667604714632034, "global_step": 11818, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03181081637740135, "train_loss_bc": 0.03181081637740135, "train_loss_llm": 0.0, "grad_norm": 0.03932956978678703, "global_step": 11819, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03528369963169098, "train_loss_bc": 0.03528127074241638, "train_loss_llm": 0.0002429187297821045, "grad_norm": 0.051754504442214966, "global_step": 11820, "epoch": 328, "lr": 0.001} +{"train_loss": 0.036942195147275925, "train_loss_bc": 0.036942195147275925, "train_loss_llm": 0.0, "grad_norm": 0.02069128304719925, "global_step": 11821, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03440326452255249, "train_loss_bc": 0.03440001606941223, "train_loss_llm": 0.0003250241279602051, "grad_norm": 0.027680428698658943, "global_step": 11822, "epoch": 328, "lr": 0.001} +{"train_loss": 0.039612527936697006, "train_loss_bc": 0.0396038219332695, "train_loss_llm": 0.0008706450462341309, "grad_norm": 0.03622708469629288, "global_step": 11823, "epoch": 328, "lr": 0.001} +{"train_loss": 0.040035516023635864, "train_loss_bc": 0.040035516023635864, "train_loss_llm": 0.0, "grad_norm": 0.044971320778131485, "global_step": 11824, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03716149926185608, "train_loss_bc": 0.03716149926185608, "train_loss_llm": 0.0, "grad_norm": 0.01831478253006935, "global_step": 11825, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03553870692849159, "train_loss_bc": 0.03553870692849159, "train_loss_llm": 0.0, "grad_norm": 0.02834043651819229, "global_step": 11826, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03547792136669159, "train_loss_bc": 0.03547792136669159, "train_loss_llm": 0.0, "grad_norm": 0.034690480679273605, "global_step": 11827, "epoch": 328, "lr": 0.001} +{"train_loss": 0.04748474061489105, "train_loss_bc": 0.04748474061489105, "train_loss_llm": 0.0, "grad_norm": 0.046569228172302246, "global_step": 11828, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03857594355940819, "train_loss_bc": 0.03857594355940819, "train_loss_llm": 0.0, "grad_norm": 0.02045166678726673, "global_step": 11829, "epoch": 328, "lr": 0.001} +{"train_loss": 0.04541376233100891, "train_loss_bc": 0.04541376233100891, "train_loss_llm": 0.0, "grad_norm": 0.03354393318295479, "global_step": 11830, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03466985374689102, "train_loss_bc": 0.03466985374689102, "train_loss_llm": 0.0, "grad_norm": 0.03771171346306801, "global_step": 11831, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03394819423556328, "train_loss_bc": 0.03394819423556328, "train_loss_llm": 0.0, "grad_norm": 0.04232725873589516, "global_step": 11832, "epoch": 328, "lr": 0.001} +{"train_loss": 0.04509800300002098, "train_loss_bc": 0.045092109590768814, "train_loss_llm": 0.000589519739151001, "grad_norm": 0.02546241320669651, "global_step": 11833, "epoch": 328, "lr": 0.001} +{"train_loss": 0.029344720765948296, "train_loss_bc": 0.029344720765948296, "train_loss_llm": 0.0, "grad_norm": 0.03323781117796898, "global_step": 11834, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03896272927522659, "train_loss_bc": 0.03896272927522659, "train_loss_llm": 0.0, "grad_norm": 0.05360054969787598, "global_step": 11835, "epoch": 328, "lr": 0.001} +{"train_loss": 0.034649528563022614, "train_loss_bc": 0.034649528563022614, "train_loss_llm": 0.0, "grad_norm": 0.061787690967321396, "global_step": 11836, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03563797473907471, "train_loss_bc": 0.035629190504550934, "train_loss_llm": 0.0008783340454101562, "grad_norm": 0.0302660521119833, "global_step": 11837, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03873378783464432, "train_loss_bc": 0.03873378783464432, "train_loss_llm": 0.0, "grad_norm": 0.04200620949268341, "global_step": 11838, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03688168153166771, "train_loss_bc": 0.03688168153166771, "train_loss_llm": 0.0, "grad_norm": 0.05178738012909889, "global_step": 11839, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03425716981291771, "train_loss_bc": 0.03423714637756348, "train_loss_llm": 0.0020023584365844727, "grad_norm": 0.07051143795251846, "global_step": 11840, "epoch": 328, "lr": 0.001} +{"train_loss": 0.04537886008620262, "train_loss_bc": 0.04537886008620262, "train_loss_llm": 0.0, "grad_norm": 0.03635687381029129, "global_step": 11841, "epoch": 328, "lr": 0.001} +{"train_loss": 0.04030406102538109, "train_loss_bc": 0.04030406102538109, "train_loss_llm": 0.0, "grad_norm": 0.05662962421774864, "global_step": 11842, "epoch": 328, "lr": 0.001} +{"train_loss": 0.03716537951388293, "train_loss_bc": 0.0315212719142437, "train_loss_llm": 0.0, "grad_norm": 0.0726785883307457, "global_step": 11843, "epoch": 328, "lr": 0.001, "val_loss": 0.06406322121620178} +{"train_loss": 0.03586946427822113, "train_loss_bc": 0.03586946427822113, "train_loss_llm": 0.0, "grad_norm": 0.08319813758134842, "global_step": 11844, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03731143847107887, "train_loss_bc": 0.03731143847107887, "train_loss_llm": 0.0, "grad_norm": 0.022775055840611458, "global_step": 11845, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03496617078781128, "train_loss_bc": 0.034951046109199524, "train_loss_llm": 0.0015126168727874756, "grad_norm": 0.0297901201993227, "global_step": 11846, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03546133637428284, "train_loss_bc": 0.03544074296951294, "train_loss_llm": 0.002059370279312134, "grad_norm": 0.03840476647019386, "global_step": 11847, "epoch": 329, "lr": 0.001} +{"train_loss": 0.030197307467460632, "train_loss_bc": 0.030197307467460632, "train_loss_llm": 0.0, "grad_norm": 0.04747170954942703, "global_step": 11848, "epoch": 329, "lr": 0.001} +{"train_loss": 0.035048097372055054, "train_loss_bc": 0.03501763567328453, "train_loss_llm": 0.0030460357666015625, "grad_norm": 0.02420768514275551, "global_step": 11849, "epoch": 329, "lr": 0.001} +{"train_loss": 0.043452367186546326, "train_loss_bc": 0.043452367186546326, "train_loss_llm": 0.0, "grad_norm": 0.035651303827762604, "global_step": 11850, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03966628015041351, "train_loss_bc": 0.03966568782925606, "train_loss_llm": 5.918741226196289e-05, "grad_norm": 0.05039365962147713, "global_step": 11851, "epoch": 329, "lr": 0.001} +{"train_loss": 0.037145379930734634, "train_loss_bc": 0.03713871166110039, "train_loss_llm": 0.0006670057773590088, "grad_norm": 0.06419090181589127, "global_step": 11852, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03705674037337303, "train_loss_bc": 0.03705674037337303, "train_loss_llm": 0.0, "grad_norm": 0.020699111744761467, "global_step": 11853, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03822411969304085, "train_loss_bc": 0.03822411969304085, "train_loss_llm": 0.0, "grad_norm": 0.037444088608026505, "global_step": 11854, "epoch": 329, "lr": 0.001} +{"train_loss": 0.034615494310855865, "train_loss_bc": 0.034615494310855865, "train_loss_llm": 0.0, "grad_norm": 0.04871062934398651, "global_step": 11855, "epoch": 329, "lr": 0.001} +{"train_loss": 0.029432782903313637, "train_loss_bc": 0.029432782903313637, "train_loss_llm": 0.0, "grad_norm": 0.06645482778549194, "global_step": 11856, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03551837056875229, "train_loss_bc": 0.03551837056875229, "train_loss_llm": 0.0, "grad_norm": 0.029880190268158913, "global_step": 11857, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03508557006716728, "train_loss_bc": 0.03508557006716728, "train_loss_llm": 0.0, "grad_norm": 0.04670824110507965, "global_step": 11858, "epoch": 329, "lr": 0.001} +{"train_loss": 0.04174365475773811, "train_loss_bc": 0.04174365475773811, "train_loss_llm": 0.0, "grad_norm": 0.06504791229963303, "global_step": 11859, "epoch": 329, "lr": 0.001} +{"train_loss": 0.04327525570988655, "train_loss_bc": 0.04327525570988655, "train_loss_llm": 0.0, "grad_norm": 0.08252400904893875, "global_step": 11860, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03981812298297882, "train_loss_bc": 0.03981190174818039, "train_loss_llm": 0.0006219446659088135, "grad_norm": 0.01784519851207733, "global_step": 11861, "epoch": 329, "lr": 0.001} +{"train_loss": 0.04178527370095253, "train_loss_bc": 0.04178527370095253, "train_loss_llm": 0.0, "grad_norm": 0.03663400560617447, "global_step": 11862, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03858881816267967, "train_loss_bc": 0.03858881816267967, "train_loss_llm": 0.0, "grad_norm": 0.04645386338233948, "global_step": 11863, "epoch": 329, "lr": 0.001} +{"train_loss": 0.039395771920681, "train_loss_bc": 0.039395771920681, "train_loss_llm": 0.0, "grad_norm": 0.059263624250888824, "global_step": 11864, "epoch": 329, "lr": 0.001} +{"train_loss": 0.040334805846214294, "train_loss_bc": 0.04032887890934944, "train_loss_llm": 0.0005927979946136475, "grad_norm": 0.0230292659252882, "global_step": 11865, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03059425950050354, "train_loss_bc": 0.03059185855090618, "train_loss_llm": 0.00024005770683288574, "grad_norm": 0.033423442393541336, "global_step": 11866, "epoch": 329, "lr": 0.001} +{"train_loss": 0.041184984147548676, "train_loss_bc": 0.041184984147548676, "train_loss_llm": 0.0, "grad_norm": 0.03830217942595482, "global_step": 11867, "epoch": 329, "lr": 0.001} +{"train_loss": 0.036588944494724274, "train_loss_bc": 0.036555975675582886, "train_loss_llm": 0.0032970011234283447, "grad_norm": 0.05085909739136696, "global_step": 11868, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03929374739527702, "train_loss_bc": 0.03929374739527702, "train_loss_llm": 0.0, "grad_norm": 0.026862334460020065, "global_step": 11869, "epoch": 329, "lr": 0.001} +{"train_loss": 0.04437081515789032, "train_loss_bc": 0.04437081515789032, "train_loss_llm": 0.0, "grad_norm": 0.039803389459848404, "global_step": 11870, "epoch": 329, "lr": 0.001} +{"train_loss": 0.04043637961149216, "train_loss_bc": 0.04042932391166687, "train_loss_llm": 0.0007057487964630127, "grad_norm": 0.053756099194288254, "global_step": 11871, "epoch": 329, "lr": 0.001} +{"train_loss": 0.034051667898893356, "train_loss_bc": 0.034051667898893356, "train_loss_llm": 0.0, "grad_norm": 0.06133546307682991, "global_step": 11872, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03922601789236069, "train_loss_bc": 0.0392066165804863, "train_loss_llm": 0.0019401609897613525, "grad_norm": 0.02532605640590191, "global_step": 11873, "epoch": 329, "lr": 0.001} +{"train_loss": 0.036669474095106125, "train_loss_bc": 0.036660246551036835, "train_loss_llm": 0.0009226500988006592, "grad_norm": 0.03231487795710564, "global_step": 11874, "epoch": 329, "lr": 0.001} +{"train_loss": 0.02859099581837654, "train_loss_bc": 0.02857804484665394, "train_loss_llm": 0.0012951791286468506, "grad_norm": 0.03874362260103226, "global_step": 11875, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03460182994604111, "train_loss_bc": 0.034596338868141174, "train_loss_llm": 0.0005491971969604492, "grad_norm": 0.05226775258779526, "global_step": 11876, "epoch": 329, "lr": 0.001} +{"train_loss": 0.04228011891245842, "train_loss_bc": 0.04228011891245842, "train_loss_llm": 0.0, "grad_norm": 0.02436113730072975, "global_step": 11877, "epoch": 329, "lr": 0.001} +{"train_loss": 0.03651506453752518, "train_loss_bc": 0.03651506453752518, "train_loss_llm": 0.0, "grad_norm": 0.04185321927070618, "global_step": 11878, "epoch": 329, "lr": 0.001} +{"train_loss": 0.037419828948461346, "train_loss_bc": 0.03871691972017288, "train_loss_llm": 0.0, "grad_norm": 0.05766782909631729, "global_step": 11879, "epoch": 329, "lr": 0.001, "val_loss": 0.06405472010374069} +{"train_loss": 0.03533032163977623, "train_loss_bc": 0.03533032163977623, "train_loss_llm": 0.0, "grad_norm": 0.0737902894616127, "global_step": 11880, "epoch": 330, "lr": 0.001} +{"train_loss": 0.039947036653757095, "train_loss_bc": 0.03993714228272438, "train_loss_llm": 0.0009895563125610352, "grad_norm": 0.023741072043776512, "global_step": 11881, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03989469259977341, "train_loss_bc": 0.03989469259977341, "train_loss_llm": 0.0, "grad_norm": 0.03584824129939079, "global_step": 11882, "epoch": 330, "lr": 0.001} +{"train_loss": 0.042273785918951035, "train_loss_bc": 0.042273785918951035, "train_loss_llm": 0.0, "grad_norm": 0.04746563732624054, "global_step": 11883, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03855045139789581, "train_loss_bc": 0.03855045139789581, "train_loss_llm": 0.0, "grad_norm": 0.06047642603516579, "global_step": 11884, "epoch": 330, "lr": 0.001} +{"train_loss": 0.030688472092151642, "train_loss_bc": 0.030681300908327103, "train_loss_llm": 0.0007170438766479492, "grad_norm": 0.024642925709486008, "global_step": 11885, "epoch": 330, "lr": 0.001} +{"train_loss": 0.040414948016405106, "train_loss_bc": 0.040410738438367844, "train_loss_llm": 0.00042110681533813477, "grad_norm": 0.03961564227938652, "global_step": 11886, "epoch": 330, "lr": 0.001} +{"train_loss": 0.034268975257873535, "train_loss_bc": 0.03425939753651619, "train_loss_llm": 0.0009576082229614258, "grad_norm": 0.05316372960805893, "global_step": 11887, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03588767722249031, "train_loss_bc": 0.03587544709444046, "train_loss_llm": 0.0012230873107910156, "grad_norm": 0.06419889628887177, "global_step": 11888, "epoch": 330, "lr": 0.001} +{"train_loss": 0.032946325838565826, "train_loss_bc": 0.032946325838565826, "train_loss_llm": 0.0, "grad_norm": 0.02772464044392109, "global_step": 11889, "epoch": 330, "lr": 0.001} +{"train_loss": 0.035131361335515976, "train_loss_bc": 0.035131361335515976, "train_loss_llm": 0.0, "grad_norm": 0.037049807608127594, "global_step": 11890, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03363262116909027, "train_loss_bc": 0.033629801124334335, "train_loss_llm": 0.0002821683883666992, "grad_norm": 0.04690730199217796, "global_step": 11891, "epoch": 330, "lr": 0.001} +{"train_loss": 0.035090669989585876, "train_loss_bc": 0.035090669989585876, "train_loss_llm": 0.0, "grad_norm": 0.06187558174133301, "global_step": 11892, "epoch": 330, "lr": 0.001} +{"train_loss": 0.041218411177396774, "train_loss_bc": 0.041215699166059494, "train_loss_llm": 0.00027129054069519043, "grad_norm": 0.0319630429148674, "global_step": 11893, "epoch": 330, "lr": 0.001} +{"train_loss": 0.04173342511057854, "train_loss_bc": 0.04173342511057854, "train_loss_llm": 0.0, "grad_norm": 0.04436691105365753, "global_step": 11894, "epoch": 330, "lr": 0.001} +{"train_loss": 0.038541145622730255, "train_loss_bc": 0.038541145622730255, "train_loss_llm": 0.0, "grad_norm": 0.05496687814593315, "global_step": 11895, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03708036243915558, "train_loss_bc": 0.03708036243915558, "train_loss_llm": 0.0, "grad_norm": 0.06616221368312836, "global_step": 11896, "epoch": 330, "lr": 0.001} +{"train_loss": 0.037519246339797974, "train_loss_bc": 0.037504587322473526, "train_loss_llm": 0.0014660358428955078, "grad_norm": 0.023469073697924614, "global_step": 11897, "epoch": 330, "lr": 0.001} +{"train_loss": 0.04217446595430374, "train_loss_bc": 0.04217446595430374, "train_loss_llm": 0.0, "grad_norm": 0.038115281611680984, "global_step": 11898, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03497839346528053, "train_loss_bc": 0.03495641425251961, "train_loss_llm": 0.0021978020668029785, "grad_norm": 0.042522016912698746, "global_step": 11899, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03281053900718689, "train_loss_bc": 0.03281053900718689, "train_loss_llm": 0.0, "grad_norm": 0.05407067760825157, "global_step": 11900, "epoch": 330, "lr": 0.001} +{"train_loss": 0.04248046875, "train_loss_bc": 0.04248046875, "train_loss_llm": 0.0, "grad_norm": 0.026940301060676575, "global_step": 11901, "epoch": 330, "lr": 0.001} +{"train_loss": 0.029743673279881477, "train_loss_bc": 0.029743673279881477, "train_loss_llm": 0.0, "grad_norm": 0.04263351112604141, "global_step": 11902, "epoch": 330, "lr": 0.001} +{"train_loss": 0.033631693571805954, "train_loss_bc": 0.033631693571805954, "train_loss_llm": 0.0, "grad_norm": 0.05536556616425514, "global_step": 11903, "epoch": 330, "lr": 0.001} +{"train_loss": 0.034405309706926346, "train_loss_bc": 0.034393973648548126, "train_loss_llm": 0.0011337101459503174, "grad_norm": 0.07011045515537262, "global_step": 11904, "epoch": 330, "lr": 0.001} +{"train_loss": 0.041464898735284805, "train_loss_bc": 0.04145023226737976, "train_loss_llm": 0.0014665722846984863, "grad_norm": 0.026175858452916145, "global_step": 11905, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03687186911702156, "train_loss_bc": 0.03687186911702156, "train_loss_llm": 0.0, "grad_norm": 0.04309418052434921, "global_step": 11906, "epoch": 330, "lr": 0.001} +{"train_loss": 0.04026027023792267, "train_loss_bc": 0.04026027023792267, "train_loss_llm": 0.0, "grad_norm": 0.06306314468383789, "global_step": 11907, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03264424577355385, "train_loss_bc": 0.03263168781995773, "train_loss_llm": 0.0012559592723846436, "grad_norm": 0.07258961349725723, "global_step": 11908, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03987576439976692, "train_loss_bc": 0.03987377509474754, "train_loss_llm": 0.00019878149032592773, "grad_norm": 0.021207241341471672, "global_step": 11909, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03975645825266838, "train_loss_bc": 0.03975645825266838, "train_loss_llm": 0.0, "grad_norm": 0.033185478299856186, "global_step": 11910, "epoch": 330, "lr": 0.001} +{"train_loss": 0.041346475481987, "train_loss_bc": 0.04132824018597603, "train_loss_llm": 0.0018233954906463623, "grad_norm": 0.04442017152905464, "global_step": 11911, "epoch": 330, "lr": 0.001} +{"train_loss": 0.04308882728219032, "train_loss_bc": 0.04308882728219032, "train_loss_llm": 0.0, "grad_norm": 0.05398403853178024, "global_step": 11912, "epoch": 330, "lr": 0.001} +{"train_loss": 0.04119754210114479, "train_loss_bc": 0.04119754210114479, "train_loss_llm": 0.0, "grad_norm": 0.0287691168487072, "global_step": 11913, "epoch": 330, "lr": 0.001} +{"train_loss": 0.03143798187375069, "train_loss_bc": 0.03143798187375069, "train_loss_llm": 0.0, "grad_norm": 0.04021989554166794, "global_step": 11914, "epoch": 330, "lr": 0.001} +{"train_loss": 0.037477921746257276, "train_loss_bc": 0.040886376053094864, "train_loss_llm": 0.0, "grad_norm": 0.07014529407024384, "global_step": 11915, "epoch": 330, "lr": 0.001, "train/cumulative_reward": 24.22105194521371, "train/mean_score": 6.516947774564182, "train/success_rate": 0.2, "test/cumulative_reward": 18.399709360836948, "test/mean_score": 2.1149350295448706, "test/success_rate": 0.04, "val_loss": 0.06382101774215698, "train_action_mse_error": 0.04084814712405205} +{"train_loss": 0.03330299258232117, "train_loss_bc": 0.03330299258232117, "train_loss_llm": 0.0, "grad_norm": 0.08005093783140182, "global_step": 11916, "epoch": 331, "lr": 0.001} +{"train_loss": 0.04724553972482681, "train_loss_bc": 0.04723857343196869, "train_loss_llm": 0.0006964802742004395, "grad_norm": 0.02955983206629753, "global_step": 11917, "epoch": 331, "lr": 0.001} +{"train_loss": 0.033234186470508575, "train_loss_bc": 0.03320933133363724, "train_loss_llm": 0.00248563289642334, "grad_norm": 0.04359850287437439, "global_step": 11918, "epoch": 331, "lr": 0.001} +{"train_loss": 0.036492180079221725, "train_loss_bc": 0.036492180079221725, "train_loss_llm": 0.0, "grad_norm": 0.06127668917179108, "global_step": 11919, "epoch": 331, "lr": 0.001} +{"train_loss": 0.033721309155225754, "train_loss_bc": 0.033721309155225754, "train_loss_llm": 0.0, "grad_norm": 0.07848720997571945, "global_step": 11920, "epoch": 331, "lr": 0.001} +{"train_loss": 0.04413276165723801, "train_loss_bc": 0.0441303513944149, "train_loss_llm": 0.0002408921718597412, "grad_norm": 0.03446916490793228, "global_step": 11921, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03488721698522568, "train_loss_bc": 0.03488721698522568, "train_loss_llm": 0.0, "grad_norm": 0.04680866375565529, "global_step": 11922, "epoch": 331, "lr": 0.001} +{"train_loss": 0.038831792771816254, "train_loss_bc": 0.038831792771816254, "train_loss_llm": 0.0, "grad_norm": 0.055651016533374786, "global_step": 11923, "epoch": 331, "lr": 0.001} +{"train_loss": 0.029145073145627975, "train_loss_bc": 0.029145073145627975, "train_loss_llm": 0.0, "grad_norm": 0.07066777348518372, "global_step": 11924, "epoch": 331, "lr": 0.001} +{"train_loss": 0.035386934876441956, "train_loss_bc": 0.035386934876441956, "train_loss_llm": 0.0, "grad_norm": 0.020836109295487404, "global_step": 11925, "epoch": 331, "lr": 0.001} +{"train_loss": 0.04388345405459404, "train_loss_bc": 0.04388345405459404, "train_loss_llm": 0.0, "grad_norm": 0.034956417977809906, "global_step": 11926, "epoch": 331, "lr": 0.001} +{"train_loss": 0.04179016873240471, "train_loss_bc": 0.04179016873240471, "train_loss_llm": 0.0, "grad_norm": 0.04402773827314377, "global_step": 11927, "epoch": 331, "lr": 0.001} +{"train_loss": 0.041403453797101974, "train_loss_bc": 0.041403453797101974, "train_loss_llm": 0.0, "grad_norm": 0.05509907379746437, "global_step": 11928, "epoch": 331, "lr": 0.001} +{"train_loss": 0.0362677276134491, "train_loss_bc": 0.0362677276134491, "train_loss_llm": 0.0, "grad_norm": 0.016970178112387657, "global_step": 11929, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03812715783715248, "train_loss_bc": 0.03807148337364197, "train_loss_llm": 0.0055675506591796875, "grad_norm": 0.030406704172492027, "global_step": 11930, "epoch": 331, "lr": 0.001} +{"train_loss": 0.040600746870040894, "train_loss_bc": 0.040600746870040894, "train_loss_llm": 0.0, "grad_norm": 0.039481524378061295, "global_step": 11931, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03385259956121445, "train_loss_bc": 0.03385259956121445, "train_loss_llm": 0.0, "grad_norm": 0.05102324113249779, "global_step": 11932, "epoch": 331, "lr": 0.001} +{"train_loss": 0.04217378795146942, "train_loss_bc": 0.04217378795146942, "train_loss_llm": 0.0, "grad_norm": 0.028662024065852165, "global_step": 11933, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03482144698500633, "train_loss_bc": 0.03482144698500633, "train_loss_llm": 0.0, "grad_norm": 0.04640002176165581, "global_step": 11934, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03702019155025482, "train_loss_bc": 0.03702019155025482, "train_loss_llm": 0.0, "grad_norm": 0.06083790957927704, "global_step": 11935, "epoch": 331, "lr": 0.001} +{"train_loss": 0.042323462665081024, "train_loss_bc": 0.042323462665081024, "train_loss_llm": 0.0, "grad_norm": 0.0787888616323471, "global_step": 11936, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03800927847623825, "train_loss_bc": 0.03800927847623825, "train_loss_llm": 0.0, "grad_norm": 0.024098819121718407, "global_step": 11937, "epoch": 331, "lr": 0.001} +{"train_loss": 0.04366388916969299, "train_loss_bc": 0.04366388916969299, "train_loss_llm": 0.0, "grad_norm": 0.04028037562966347, "global_step": 11938, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03941114991903305, "train_loss_bc": 0.03941114991903305, "train_loss_llm": 0.0, "grad_norm": 0.06244722753763199, "global_step": 11939, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03786126524209976, "train_loss_bc": 0.03786126524209976, "train_loss_llm": 0.0, "grad_norm": 0.07601261138916016, "global_step": 11940, "epoch": 331, "lr": 0.001} +{"train_loss": 0.035765036940574646, "train_loss_bc": 0.035765036940574646, "train_loss_llm": 0.0, "grad_norm": 0.022483250126242638, "global_step": 11941, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03562237694859505, "train_loss_bc": 0.03562237694859505, "train_loss_llm": 0.0, "grad_norm": 0.04048436880111694, "global_step": 11942, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03226109966635704, "train_loss_bc": 0.032259732484817505, "train_loss_llm": 0.0001367628574371338, "grad_norm": 0.055653657764196396, "global_step": 11943, "epoch": 331, "lr": 0.001} +{"train_loss": 0.033782921731472015, "train_loss_bc": 0.033782921731472015, "train_loss_llm": 0.0, "grad_norm": 0.07781368494033813, "global_step": 11944, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03151444345712662, "train_loss_bc": 0.03151444345712662, "train_loss_llm": 0.0, "grad_norm": 0.020080285146832466, "global_step": 11945, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03748996555805206, "train_loss_bc": 0.03748996555805206, "train_loss_llm": 0.0, "grad_norm": 0.030484633520245552, "global_step": 11946, "epoch": 331, "lr": 0.001} +{"train_loss": 0.04079904407262802, "train_loss_bc": 0.04079288989305496, "train_loss_llm": 0.0006153881549835205, "grad_norm": 0.04421110451221466, "global_step": 11947, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03485458716750145, "train_loss_bc": 0.03485458716750145, "train_loss_llm": 0.0, "grad_norm": 0.05854307860136032, "global_step": 11948, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03881073370575905, "train_loss_bc": 0.038775354623794556, "train_loss_llm": 0.0035380125045776367, "grad_norm": 0.02031046710908413, "global_step": 11949, "epoch": 331, "lr": 0.001} +{"train_loss": 0.037245362997055054, "train_loss_bc": 0.037245362997055054, "train_loss_llm": 0.0, "grad_norm": 0.029543189331889153, "global_step": 11950, "epoch": 331, "lr": 0.001} +{"train_loss": 0.03744136800782548, "train_loss_bc": 0.032146938145160675, "train_loss_llm": 0.0006969869136810303, "grad_norm": 0.03635592386126518, "global_step": 11951, "epoch": 331, "lr": 0.001, "val_loss": 0.06354948878288269} +{"train_loss": 0.03650028258562088, "train_loss_bc": 0.03650028258562088, "train_loss_llm": 0.0, "grad_norm": 0.04273826256394386, "global_step": 11952, "epoch": 332, "lr": 0.001} +{"train_loss": 0.033418577164411545, "train_loss_bc": 0.033418577164411545, "train_loss_llm": 0.0, "grad_norm": 0.02111007273197174, "global_step": 11953, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03913060203194618, "train_loss_bc": 0.03913060203194618, "train_loss_llm": 0.0, "grad_norm": 0.03159545361995697, "global_step": 11954, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03516091778874397, "train_loss_bc": 0.03516091778874397, "train_loss_llm": 0.0, "grad_norm": 0.041002146899700165, "global_step": 11955, "epoch": 332, "lr": 0.001} +{"train_loss": 0.0328991562128067, "train_loss_bc": 0.0328991562128067, "train_loss_llm": 0.0, "grad_norm": 0.043315477669239044, "global_step": 11956, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03705828636884689, "train_loss_bc": 0.037045110017061234, "train_loss_llm": 0.0013177990913391113, "grad_norm": 0.023536162450909615, "global_step": 11957, "epoch": 332, "lr": 0.001} +{"train_loss": 0.042737945914268494, "train_loss_bc": 0.042737945914268494, "train_loss_llm": 0.0, "grad_norm": 0.046229105442762375, "global_step": 11958, "epoch": 332, "lr": 0.001} +{"train_loss": 0.039473600685596466, "train_loss_bc": 0.039473600685596466, "train_loss_llm": 0.0, "grad_norm": 0.07005389779806137, "global_step": 11959, "epoch": 332, "lr": 0.001} +{"train_loss": 0.0347956158220768, "train_loss_bc": 0.03478209301829338, "train_loss_llm": 0.0013522803783416748, "grad_norm": 0.08909353613853455, "global_step": 11960, "epoch": 332, "lr": 0.001} +{"train_loss": 0.039640698581933975, "train_loss_bc": 0.039639443159103394, "train_loss_llm": 0.00012570619583129883, "grad_norm": 0.023190593346953392, "global_step": 11961, "epoch": 332, "lr": 0.001} +{"train_loss": 0.041475992649793625, "train_loss_bc": 0.04147127643227577, "train_loss_llm": 0.00047156214714050293, "grad_norm": 0.041232503950595856, "global_step": 11962, "epoch": 332, "lr": 0.001} +{"train_loss": 0.029322229325771332, "train_loss_bc": 0.029290180653333664, "train_loss_llm": 0.003204941749572754, "grad_norm": 0.05242129787802696, "global_step": 11963, "epoch": 332, "lr": 0.001} +{"train_loss": 0.036426838487386703, "train_loss_bc": 0.036426838487386703, "train_loss_llm": 0.0, "grad_norm": 0.07064492255449295, "global_step": 11964, "epoch": 332, "lr": 0.001} +{"train_loss": 0.045590560883283615, "train_loss_bc": 0.045579712837934494, "train_loss_llm": 0.0010846257209777832, "grad_norm": 0.025836337357759476, "global_step": 11965, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03708513826131821, "train_loss_bc": 0.03708513826131821, "train_loss_llm": 0.0, "grad_norm": 0.03877617046236992, "global_step": 11966, "epoch": 332, "lr": 0.001} +{"train_loss": 0.033221352845430374, "train_loss_bc": 0.03321770206093788, "train_loss_llm": 0.0003650784492492676, "grad_norm": 0.05257224291563034, "global_step": 11967, "epoch": 332, "lr": 0.001} +{"train_loss": 0.030674872919917107, "train_loss_bc": 0.030674872919917107, "train_loss_llm": 0.0, "grad_norm": 0.063630111515522, "global_step": 11968, "epoch": 332, "lr": 0.001} +{"train_loss": 0.034479618072509766, "train_loss_bc": 0.034479618072509766, "train_loss_llm": 0.0, "grad_norm": 0.025263218209147453, "global_step": 11969, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03969477117061615, "train_loss_bc": 0.03969477117061615, "train_loss_llm": 0.0, "grad_norm": 0.02894897386431694, "global_step": 11970, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03800845518708229, "train_loss_bc": 0.03800845518708229, "train_loss_llm": 0.0, "grad_norm": 0.038477156311273575, "global_step": 11971, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03768562525510788, "train_loss_bc": 0.03768562525510788, "train_loss_llm": 0.0, "grad_norm": 0.04528046026825905, "global_step": 11972, "epoch": 332, "lr": 0.001} +{"train_loss": 0.04607875645160675, "train_loss_bc": 0.04607875645160675, "train_loss_llm": 0.0, "grad_norm": 0.025257857516407967, "global_step": 11973, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03839392960071564, "train_loss_bc": 0.03839392960071564, "train_loss_llm": 0.0, "grad_norm": 0.03924410790205002, "global_step": 11974, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03465903922915459, "train_loss_bc": 0.03465903922915459, "train_loss_llm": 0.0, "grad_norm": 0.05062359943985939, "global_step": 11975, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03473612293601036, "train_loss_bc": 0.03473612293601036, "train_loss_llm": 0.0, "grad_norm": 0.06611239165067673, "global_step": 11976, "epoch": 332, "lr": 0.001} +{"train_loss": 0.04305688664317131, "train_loss_bc": 0.04305688664317131, "train_loss_llm": 0.0, "grad_norm": 0.028568008914589882, "global_step": 11977, "epoch": 332, "lr": 0.001} +{"train_loss": 0.035360123962163925, "train_loss_bc": 0.03535469248890877, "train_loss_llm": 0.0005431175231933594, "grad_norm": 0.04628525301814079, "global_step": 11978, "epoch": 332, "lr": 0.001} +{"train_loss": 0.04222883656620979, "train_loss_bc": 0.04222883656620979, "train_loss_llm": 0.0, "grad_norm": 0.052929025143384933, "global_step": 11979, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03325731307268143, "train_loss_bc": 0.03322752192616463, "train_loss_llm": 0.002979069948196411, "grad_norm": 0.061612360179424286, "global_step": 11980, "epoch": 332, "lr": 0.001} +{"train_loss": 0.030851269140839577, "train_loss_bc": 0.030851269140839577, "train_loss_llm": 0.0, "grad_norm": 0.01959364488720894, "global_step": 11981, "epoch": 332, "lr": 0.001} +{"train_loss": 0.039233703166246414, "train_loss_bc": 0.03922269493341446, "train_loss_llm": 0.001100778579711914, "grad_norm": 0.0414930060505867, "global_step": 11982, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03783706575632095, "train_loss_bc": 0.03783706575632095, "train_loss_llm": 0.0, "grad_norm": 0.05976063013076782, "global_step": 11983, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03822995349764824, "train_loss_bc": 0.03822995349764824, "train_loss_llm": 0.0, "grad_norm": 0.08464889973402023, "global_step": 11984, "epoch": 332, "lr": 0.001} +{"train_loss": 0.04479610174894333, "train_loss_bc": 0.0447842963039875, "train_loss_llm": 0.0011806786060333252, "grad_norm": 0.02018597722053528, "global_step": 11985, "epoch": 332, "lr": 0.001} +{"train_loss": 0.03983544558286667, "train_loss_bc": 0.03983544558286667, "train_loss_llm": 0.0, "grad_norm": 0.029244428500533104, "global_step": 11986, "epoch": 332, "lr": 0.001} +{"train_loss": 0.037428241326577134, "train_loss_bc": 0.03438100218772888, "train_loss_llm": 0.0, "grad_norm": 0.04039255529642105, "global_step": 11987, "epoch": 332, "lr": 0.001, "val_loss": 0.06462034583091736} +{"train_loss": 0.041562266647815704, "train_loss_bc": 0.041562266647815704, "train_loss_llm": 0.0, "grad_norm": 0.05423648655414581, "global_step": 11988, "epoch": 333, "lr": 0.001} +{"train_loss": 0.04240712895989418, "train_loss_bc": 0.04239984601736069, "train_loss_llm": 0.0007281303405761719, "grad_norm": 0.021055446937680244, "global_step": 11989, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03683791309595108, "train_loss_bc": 0.03683791309595108, "train_loss_llm": 0.0, "grad_norm": 0.02569737657904625, "global_step": 11990, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03609555587172508, "train_loss_bc": 0.03609555587172508, "train_loss_llm": 0.0, "grad_norm": 0.036869414150714874, "global_step": 11991, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03454838693141937, "train_loss_bc": 0.03454838693141937, "train_loss_llm": 0.0, "grad_norm": 0.046537384390830994, "global_step": 11992, "epoch": 333, "lr": 0.001} +{"train_loss": 0.031135691329836845, "train_loss_bc": 0.03113078884780407, "train_loss_llm": 0.0004901885986328125, "grad_norm": 0.020544160157442093, "global_step": 11993, "epoch": 333, "lr": 0.001} +{"train_loss": 0.04231686145067215, "train_loss_bc": 0.04231686145067215, "train_loss_llm": 0.0, "grad_norm": 0.03928756341338158, "global_step": 11994, "epoch": 333, "lr": 0.001} +{"train_loss": 0.032261431217193604, "train_loss_bc": 0.032261431217193604, "train_loss_llm": 0.0, "grad_norm": 0.051309142261743546, "global_step": 11995, "epoch": 333, "lr": 0.001} +{"train_loss": 0.029488617554306984, "train_loss_bc": 0.029488617554306984, "train_loss_llm": 0.0, "grad_norm": 0.0655323937535286, "global_step": 11996, "epoch": 333, "lr": 0.001} +{"train_loss": 0.030503619462251663, "train_loss_bc": 0.030503619462251663, "train_loss_llm": 0.0, "grad_norm": 0.017842190340161324, "global_step": 11997, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03475947305560112, "train_loss_bc": 0.03474379703402519, "train_loss_llm": 0.0015676915645599365, "grad_norm": 0.03159699589014053, "global_step": 11998, "epoch": 333, "lr": 0.001} +{"train_loss": 0.04192616418004036, "train_loss_bc": 0.04192616418004036, "train_loss_llm": 0.0, "grad_norm": 0.0388147234916687, "global_step": 11999, "epoch": 333, "lr": 0.001} +{"train_loss": 0.04385482519865036, "train_loss_bc": 0.043849341571331024, "train_loss_llm": 0.000548243522644043, "grad_norm": 0.055900052189826965, "global_step": 12000, "epoch": 333, "lr": 0.001} +{"train_loss": 0.04197130352258682, "train_loss_bc": 0.04197130352258682, "train_loss_llm": 0.0, "grad_norm": 0.022694673389196396, "global_step": 12001, "epoch": 333, "lr": 0.001} +{"train_loss": 0.038710325956344604, "train_loss_bc": 0.038709525018930435, "train_loss_llm": 8.025765419006348e-05, "grad_norm": 0.03333049267530441, "global_step": 12002, "epoch": 333, "lr": 0.001} +{"train_loss": 0.0382855050265789, "train_loss_bc": 0.0382855050265789, "train_loss_llm": 0.0, "grad_norm": 0.04755565524101257, "global_step": 12003, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03789239004254341, "train_loss_bc": 0.03789239004254341, "train_loss_llm": 0.0, "grad_norm": 0.06021227687597275, "global_step": 12004, "epoch": 333, "lr": 0.001} +{"train_loss": 0.029909281060099602, "train_loss_bc": 0.029905587434768677, "train_loss_llm": 0.000369340181350708, "grad_norm": 0.020532682538032532, "global_step": 12005, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03718568757176399, "train_loss_bc": 0.03716503456234932, "train_loss_llm": 0.0020654797554016113, "grad_norm": 0.03323446959257126, "global_step": 12006, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03684188798069954, "train_loss_bc": 0.03684188798069954, "train_loss_llm": 0.0, "grad_norm": 0.04288465529680252, "global_step": 12007, "epoch": 333, "lr": 0.001} +{"train_loss": 0.0314437635242939, "train_loss_bc": 0.0314437635242939, "train_loss_llm": 0.0, "grad_norm": 0.049140866845846176, "global_step": 12008, "epoch": 333, "lr": 0.001} +{"train_loss": 0.04280763119459152, "train_loss_bc": 0.04280763119459152, "train_loss_llm": 0.0, "grad_norm": 0.023957744240760803, "global_step": 12009, "epoch": 333, "lr": 0.001} +{"train_loss": 0.038164809346199036, "train_loss_bc": 0.03815511614084244, "train_loss_llm": 0.0009694397449493408, "grad_norm": 0.03726537898182869, "global_step": 12010, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03981794789433479, "train_loss_bc": 0.03981794789433479, "train_loss_llm": 0.0, "grad_norm": 0.052762582898139954, "global_step": 12011, "epoch": 333, "lr": 0.001} +{"train_loss": 0.040030527859926224, "train_loss_bc": 0.040030527859926224, "train_loss_llm": 0.0, "grad_norm": 0.06264901161193848, "global_step": 12012, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03536192700266838, "train_loss_bc": 0.03535714000463486, "train_loss_llm": 0.0004787445068359375, "grad_norm": 0.024402543902397156, "global_step": 12013, "epoch": 333, "lr": 0.001} +{"train_loss": 0.04331512376666069, "train_loss_bc": 0.043302252888679504, "train_loss_llm": 0.0012870430946350098, "grad_norm": 0.04021630808711052, "global_step": 12014, "epoch": 333, "lr": 0.001} +{"train_loss": 0.026094943284988403, "train_loss_bc": 0.026094943284988403, "train_loss_llm": 0.0, "grad_norm": 0.050566304475069046, "global_step": 12015, "epoch": 333, "lr": 0.001} +{"train_loss": 0.039053842425346375, "train_loss_bc": 0.039053842425346375, "train_loss_llm": 0.0, "grad_norm": 0.06652741879224777, "global_step": 12016, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03191692382097244, "train_loss_bc": 0.03191692382097244, "train_loss_llm": 0.0, "grad_norm": 0.022263886407017708, "global_step": 12017, "epoch": 333, "lr": 0.001} +{"train_loss": 0.04159170389175415, "train_loss_bc": 0.04159170389175415, "train_loss_llm": 0.0, "grad_norm": 0.0373905748128891, "global_step": 12018, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03866030275821686, "train_loss_bc": 0.0386468842625618, "train_loss_llm": 0.0013418495655059814, "grad_norm": 0.0526999793946743, "global_step": 12019, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03712988272309303, "train_loss_bc": 0.03712988272309303, "train_loss_llm": 0.0, "grad_norm": 0.06442664563655853, "global_step": 12020, "epoch": 333, "lr": 0.001} +{"train_loss": 0.03758217766880989, "train_loss_bc": 0.03758217766880989, "train_loss_llm": 0.0, "grad_norm": 0.019090648740530014, "global_step": 12021, "epoch": 333, "lr": 0.001} +{"train_loss": 0.04080681502819061, "train_loss_bc": 0.04078637808561325, "train_loss_llm": 0.00204351544380188, "grad_norm": 0.03324943408370018, "global_step": 12022, "epoch": 333, "lr": 0.001} +{"train_loss": 0.037254172635989055, "train_loss_bc": 0.03887757658958435, "train_loss_llm": 0.0, "grad_norm": 0.04939057305455208, "global_step": 12023, "epoch": 333, "lr": 0.001, "val_loss": 0.06427209824323654} +{"train_loss": 0.033568259328603745, "train_loss_bc": 0.033568259328603745, "train_loss_llm": 0.0, "grad_norm": 0.059348803013563156, "global_step": 12024, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03263069689273834, "train_loss_bc": 0.03263069689273834, "train_loss_llm": 0.0, "grad_norm": 0.014714138582348824, "global_step": 12025, "epoch": 334, "lr": 0.001} +{"train_loss": 0.037049319595098495, "train_loss_bc": 0.03704756125807762, "train_loss_llm": 0.00017598271369934082, "grad_norm": 0.02427085116505623, "global_step": 12026, "epoch": 334, "lr": 0.001} +{"train_loss": 0.040109798312187195, "train_loss_bc": 0.040091004222631454, "train_loss_llm": 0.0018792450428009033, "grad_norm": 0.028764592483639717, "global_step": 12027, "epoch": 334, "lr": 0.001} +{"train_loss": 0.033280033618211746, "train_loss_bc": 0.033280033618211746, "train_loss_llm": 0.0, "grad_norm": 0.030663784593343735, "global_step": 12028, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03673970326781273, "train_loss_bc": 0.03673190623521805, "train_loss_llm": 0.0007795393466949463, "grad_norm": 0.019458871334791183, "global_step": 12029, "epoch": 334, "lr": 0.001} +{"train_loss": 0.046676523983478546, "train_loss_bc": 0.046676523983478546, "train_loss_llm": 0.0, "grad_norm": 0.035046856850385666, "global_step": 12030, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03298788145184517, "train_loss_bc": 0.03298189118504524, "train_loss_llm": 0.0005988478660583496, "grad_norm": 0.04302789270877838, "global_step": 12031, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03391632065176964, "train_loss_bc": 0.03391632065176964, "train_loss_llm": 0.0, "grad_norm": 0.05403268709778786, "global_step": 12032, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03292297199368477, "train_loss_bc": 0.032916225492954254, "train_loss_llm": 0.0006747245788574219, "grad_norm": 0.023673562332987785, "global_step": 12033, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03379826620221138, "train_loss_bc": 0.033795066177845, "train_loss_llm": 0.00032001733779907227, "grad_norm": 0.037210606038570404, "global_step": 12034, "epoch": 334, "lr": 0.001} +{"train_loss": 0.04576205834746361, "train_loss_bc": 0.04576205834746361, "train_loss_llm": 0.0, "grad_norm": 0.051123443990945816, "global_step": 12035, "epoch": 334, "lr": 0.001} +{"train_loss": 0.042152464389801025, "train_loss_bc": 0.042152464389801025, "train_loss_llm": 0.0, "grad_norm": 0.06388843804597855, "global_step": 12036, "epoch": 334, "lr": 0.001} +{"train_loss": 0.033811699599027634, "train_loss_bc": 0.033808596432209015, "train_loss_llm": 0.0003103911876678467, "grad_norm": 0.0254445131868124, "global_step": 12037, "epoch": 334, "lr": 0.001} +{"train_loss": 0.041272107511758804, "train_loss_bc": 0.041250817477703094, "train_loss_llm": 0.002129107713699341, "grad_norm": 0.0447169728577137, "global_step": 12038, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03879241272807121, "train_loss_bc": 0.038785919547080994, "train_loss_llm": 0.0006492137908935547, "grad_norm": 0.06208466738462448, "global_step": 12039, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03355446830391884, "train_loss_bc": 0.03354792669415474, "train_loss_llm": 0.0006543397903442383, "grad_norm": 0.07271789759397507, "global_step": 12040, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03555424138903618, "train_loss_bc": 0.03555424138903618, "train_loss_llm": 0.0, "grad_norm": 0.026633046567440033, "global_step": 12041, "epoch": 334, "lr": 0.001} +{"train_loss": 0.04063985124230385, "train_loss_bc": 0.04063931480050087, "train_loss_llm": 5.361437797546387e-05, "grad_norm": 0.04066488891839981, "global_step": 12042, "epoch": 334, "lr": 0.001} +{"train_loss": 0.04162812605500221, "train_loss_bc": 0.04162812605500221, "train_loss_llm": 0.0, "grad_norm": 0.055419258773326874, "global_step": 12043, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03850443288683891, "train_loss_bc": 0.03850443288683891, "train_loss_llm": 0.0, "grad_norm": 0.07115264981985092, "global_step": 12044, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03252318128943443, "train_loss_bc": 0.03252318128943443, "train_loss_llm": 0.0, "grad_norm": 0.01835445500910282, "global_step": 12045, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03625432774424553, "train_loss_bc": 0.03625432774424553, "train_loss_llm": 0.0, "grad_norm": 0.02908657118678093, "global_step": 12046, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03964115306735039, "train_loss_bc": 0.03964115306735039, "train_loss_llm": 0.0, "grad_norm": 0.03643038496375084, "global_step": 12047, "epoch": 334, "lr": 0.001} +{"train_loss": 0.0369110181927681, "train_loss_bc": 0.0368943065404892, "train_loss_llm": 0.0016711056232452393, "grad_norm": 0.04916200786828995, "global_step": 12048, "epoch": 334, "lr": 0.001} +{"train_loss": 0.033402591943740845, "train_loss_bc": 0.03339691460132599, "train_loss_llm": 0.0005678236484527588, "grad_norm": 0.016532061621546745, "global_step": 12049, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03819205239415169, "train_loss_bc": 0.03819205239415169, "train_loss_llm": 0.0, "grad_norm": 0.025835847482085228, "global_step": 12050, "epoch": 334, "lr": 0.001} +{"train_loss": 0.032540060579776764, "train_loss_bc": 0.032540060579776764, "train_loss_llm": 0.0, "grad_norm": 0.0373091995716095, "global_step": 12051, "epoch": 334, "lr": 0.001} +{"train_loss": 0.038819268345832825, "train_loss_bc": 0.038819268345832825, "train_loss_llm": 0.0, "grad_norm": 0.05317378044128418, "global_step": 12052, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03802420571446419, "train_loss_bc": 0.03802420571446419, "train_loss_llm": 0.0, "grad_norm": 0.02125166542828083, "global_step": 12053, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03785373270511627, "train_loss_bc": 0.03785373270511627, "train_loss_llm": 0.0, "grad_norm": 0.03720694035291672, "global_step": 12054, "epoch": 334, "lr": 0.001} +{"train_loss": 0.043639056384563446, "train_loss_bc": 0.04363831877708435, "train_loss_llm": 7.382035255432129e-05, "grad_norm": 0.0509629026055336, "global_step": 12055, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03486980125308037, "train_loss_bc": 0.03486980125308037, "train_loss_llm": 0.0, "grad_norm": 0.061525214463472366, "global_step": 12056, "epoch": 334, "lr": 0.001} +{"train_loss": 0.03619064390659332, "train_loss_bc": 0.0361785963177681, "train_loss_llm": 0.0012045800685882568, "grad_norm": 0.02076294645667076, "global_step": 12057, "epoch": 334, "lr": 0.001} +{"train_loss": 0.031483300030231476, "train_loss_bc": 0.031483300030231476, "train_loss_llm": 0.0, "grad_norm": 0.031224045902490616, "global_step": 12058, "epoch": 334, "lr": 0.001} +{"train_loss": 0.037322200938231416, "train_loss_bc": 0.04790320247411728, "train_loss_llm": 0.0, "grad_norm": 0.05002390593290329, "global_step": 12059, "epoch": 334, "lr": 0.001, "val_loss": 0.06401088088750839} +{"train_loss": 0.03257012367248535, "train_loss_bc": 0.032555706799030304, "train_loss_llm": 0.0014416277408599854, "grad_norm": 0.061434272676706314, "global_step": 12060, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03536234423518181, "train_loss_bc": 0.03536234423518181, "train_loss_llm": 0.0, "grad_norm": 0.018816666677594185, "global_step": 12061, "epoch": 335, "lr": 0.001} +{"train_loss": 0.031160345301032066, "train_loss_bc": 0.031153259798884392, "train_loss_llm": 0.0007084906101226807, "grad_norm": 0.031205320730805397, "global_step": 12062, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03423145040869713, "train_loss_bc": 0.03423145040869713, "train_loss_llm": 0.0, "grad_norm": 0.047059353440999985, "global_step": 12063, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03589635342359543, "train_loss_bc": 0.03589635342359543, "train_loss_llm": 0.0, "grad_norm": 0.057228174060583115, "global_step": 12064, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03839348629117012, "train_loss_bc": 0.03839348629117012, "train_loss_llm": 0.0, "grad_norm": 0.023875921964645386, "global_step": 12065, "epoch": 335, "lr": 0.001} +{"train_loss": 0.04243460297584534, "train_loss_bc": 0.042433321475982666, "train_loss_llm": 0.00012820959091186523, "grad_norm": 0.034544628113508224, "global_step": 12066, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03920312225818634, "train_loss_bc": 0.03920312225818634, "train_loss_llm": 0.0, "grad_norm": 0.0487031526863575, "global_step": 12067, "epoch": 335, "lr": 0.001} +{"train_loss": 0.04467402771115303, "train_loss_bc": 0.04467402771115303, "train_loss_llm": 0.0, "grad_norm": 0.06647409498691559, "global_step": 12068, "epoch": 335, "lr": 0.001} +{"train_loss": 0.04063108190894127, "train_loss_bc": 0.04063108190894127, "train_loss_llm": 0.0, "grad_norm": 0.019744515419006348, "global_step": 12069, "epoch": 335, "lr": 0.001} +{"train_loss": 0.034795716404914856, "train_loss_bc": 0.034795716404914856, "train_loss_llm": 0.0, "grad_norm": 0.03289765119552612, "global_step": 12070, "epoch": 335, "lr": 0.001} +{"train_loss": 0.039030760526657104, "train_loss_bc": 0.039030760526657104, "train_loss_llm": 0.0, "grad_norm": 0.03775913268327713, "global_step": 12071, "epoch": 335, "lr": 0.001} +{"train_loss": 0.04018550366163254, "train_loss_bc": 0.04018089920282364, "train_loss_llm": 0.0004603564739227295, "grad_norm": 0.0515129379928112, "global_step": 12072, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03591923788189888, "train_loss_bc": 0.03591923788189888, "train_loss_llm": 0.0, "grad_norm": 0.021432606503367424, "global_step": 12073, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03524109721183777, "train_loss_bc": 0.03524109721183777, "train_loss_llm": 0.0, "grad_norm": 0.03351961076259613, "global_step": 12074, "epoch": 335, "lr": 0.001} +{"train_loss": 0.04094434529542923, "train_loss_bc": 0.04094434529542923, "train_loss_llm": 0.0, "grad_norm": 0.038281895220279694, "global_step": 12075, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03535611182451248, "train_loss_bc": 0.03535611182451248, "train_loss_llm": 0.0, "grad_norm": 0.044473834335803986, "global_step": 12076, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03443261608481407, "train_loss_bc": 0.03443261608481407, "train_loss_llm": 0.0, "grad_norm": 0.019668089225888252, "global_step": 12077, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03134443610906601, "train_loss_bc": 0.03134443610906601, "train_loss_llm": 0.0, "grad_norm": 0.028297575190663338, "global_step": 12078, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03195236250758171, "train_loss_bc": 0.03195236250758171, "train_loss_llm": 0.0, "grad_norm": 0.03812664374709129, "global_step": 12079, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03489336371421814, "train_loss_bc": 0.03489336371421814, "train_loss_llm": 0.0, "grad_norm": 0.0466955304145813, "global_step": 12080, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03765815123915672, "train_loss_bc": 0.037637583911418915, "train_loss_llm": 0.002056896686553955, "grad_norm": 0.026268064975738525, "global_step": 12081, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03784095495939255, "train_loss_bc": 0.03784095495939255, "train_loss_llm": 0.0, "grad_norm": 0.034313540905714035, "global_step": 12082, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03991108760237694, "train_loss_bc": 0.03991014137864113, "train_loss_llm": 9.447336196899414e-05, "grad_norm": 0.045902661979198456, "global_step": 12083, "epoch": 335, "lr": 0.001} +{"train_loss": 0.04107559844851494, "train_loss_bc": 0.04107559844851494, "train_loss_llm": 0.0, "grad_norm": 0.06233104318380356, "global_step": 12084, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03316637873649597, "train_loss_bc": 0.03316637873649597, "train_loss_llm": 0.0, "grad_norm": 0.022709380835294724, "global_step": 12085, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03419315814971924, "train_loss_bc": 0.03419315814971924, "train_loss_llm": 0.0, "grad_norm": 0.03750531002879143, "global_step": 12086, "epoch": 335, "lr": 0.001} +{"train_loss": 0.042642995715141296, "train_loss_bc": 0.0426144152879715, "train_loss_llm": 0.0028582215309143066, "grad_norm": 0.05544021353125572, "global_step": 12087, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03981468454003334, "train_loss_bc": 0.03981468454003334, "train_loss_llm": 0.0, "grad_norm": 0.06815100461244583, "global_step": 12088, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03977236896753311, "train_loss_bc": 0.03977236896753311, "train_loss_llm": 0.0, "grad_norm": 0.022720040753483772, "global_step": 12089, "epoch": 335, "lr": 0.001} +{"train_loss": 0.0359521321952343, "train_loss_bc": 0.0359521321952343, "train_loss_llm": 0.0, "grad_norm": 0.03536463528871536, "global_step": 12090, "epoch": 335, "lr": 0.001} +{"train_loss": 0.030881939455866814, "train_loss_bc": 0.030881939455866814, "train_loss_llm": 0.0, "grad_norm": 0.04396998882293701, "global_step": 12091, "epoch": 335, "lr": 0.001} +{"train_loss": 0.038126323372125626, "train_loss_bc": 0.03811955824494362, "train_loss_llm": 0.0006764233112335205, "grad_norm": 0.05345126613974571, "global_step": 12092, "epoch": 335, "lr": 0.001} +{"train_loss": 0.04137745127081871, "train_loss_bc": 0.04137745127081871, "train_loss_llm": 0.0, "grad_norm": 0.017485473304986954, "global_step": 12093, "epoch": 335, "lr": 0.001} +{"train_loss": 0.04072365537285805, "train_loss_bc": 0.04067298397421837, "train_loss_llm": 0.00506708025932312, "grad_norm": 0.03316156938672066, "global_step": 12094, "epoch": 335, "lr": 0.001} +{"train_loss": 0.03724981223543485, "train_loss_bc": 0.03920387104153633, "train_loss_llm": 0.0, "grad_norm": 0.0428154319524765, "global_step": 12095, "epoch": 335, "lr": 0.001, "train/cumulative_reward": 115.28173793508374, "train/mean_score": 13.901746864433983, "train/success_rate": 0.3, "test/cumulative_reward": 67.93997216763432, "test/mean_score": 8.770467483750526, "test/success_rate": 0.16, "val_loss": 0.06481275707483292, "train_action_mse_error": 0.04032565653324127} +{"train_loss": 0.03923080489039421, "train_loss_bc": 0.03923080489039421, "train_loss_llm": 0.0, "grad_norm": 0.057003892958164215, "global_step": 12096, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03392631188035011, "train_loss_bc": 0.033915262669324875, "train_loss_llm": 0.0011050105094909668, "grad_norm": 0.026226239278912544, "global_step": 12097, "epoch": 336, "lr": 0.001} +{"train_loss": 0.04191406071186066, "train_loss_bc": 0.04191406071186066, "train_loss_llm": 0.0, "grad_norm": 0.035305388271808624, "global_step": 12098, "epoch": 336, "lr": 0.001} +{"train_loss": 0.037367552518844604, "train_loss_bc": 0.037367552518844604, "train_loss_llm": 0.0, "grad_norm": 0.048496149480342865, "global_step": 12099, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03238935396075249, "train_loss_bc": 0.03238935396075249, "train_loss_llm": 0.0, "grad_norm": 0.06085347756743431, "global_step": 12100, "epoch": 336, "lr": 0.001} +{"train_loss": 0.04190373420715332, "train_loss_bc": 0.04190373420715332, "train_loss_llm": 0.0, "grad_norm": 0.027083758264780045, "global_step": 12101, "epoch": 336, "lr": 0.001} +{"train_loss": 0.04563119634985924, "train_loss_bc": 0.045625824481248856, "train_loss_llm": 0.0005370080471038818, "grad_norm": 0.04870542138814926, "global_step": 12102, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03917349502444267, "train_loss_bc": 0.03917349502444267, "train_loss_llm": 0.0, "grad_norm": 0.06011905148625374, "global_step": 12103, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03441648930311203, "train_loss_bc": 0.03441648930311203, "train_loss_llm": 0.0, "grad_norm": 0.06779193133115768, "global_step": 12104, "epoch": 336, "lr": 0.001} +{"train_loss": 0.032679397612810135, "train_loss_bc": 0.032679397612810135, "train_loss_llm": 0.0, "grad_norm": 0.024869875982403755, "global_step": 12105, "epoch": 336, "lr": 0.001} +{"train_loss": 0.040106020867824554, "train_loss_bc": 0.040106020867824554, "train_loss_llm": 0.0, "grad_norm": 0.031385719776153564, "global_step": 12106, "epoch": 336, "lr": 0.001} +{"train_loss": 0.029825467616319656, "train_loss_bc": 0.02980312705039978, "train_loss_llm": 0.0022339820861816406, "grad_norm": 0.04133420065045357, "global_step": 12107, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03786669671535492, "train_loss_bc": 0.03786669671535492, "train_loss_llm": 0.0, "grad_norm": 0.0549386702477932, "global_step": 12108, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03879876807332039, "train_loss_bc": 0.0387846902012825, "train_loss_llm": 0.001407712697982788, "grad_norm": 0.026049233973026276, "global_step": 12109, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03977832943201065, "train_loss_bc": 0.03977832943201065, "train_loss_llm": 0.0, "grad_norm": 0.03269874304533005, "global_step": 12110, "epoch": 336, "lr": 0.001} +{"train_loss": 0.0418604239821434, "train_loss_bc": 0.0418604239821434, "train_loss_llm": 0.0, "grad_norm": 0.050139181315898895, "global_step": 12111, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03675702586770058, "train_loss_bc": 0.036751583218574524, "train_loss_llm": 0.0005441904067993164, "grad_norm": 0.06259730458259583, "global_step": 12112, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03402277082204819, "train_loss_bc": 0.03402277082204819, "train_loss_llm": 0.0, "grad_norm": 0.021909059956669807, "global_step": 12113, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03483661264181137, "train_loss_bc": 0.03483661264181137, "train_loss_llm": 0.0, "grad_norm": 0.030412262305617332, "global_step": 12114, "epoch": 336, "lr": 0.001} +{"train_loss": 0.04037543758749962, "train_loss_bc": 0.04035445302724838, "train_loss_llm": 0.0020985007286071777, "grad_norm": 0.03965805098414421, "global_step": 12115, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03649679571390152, "train_loss_bc": 0.03649679571390152, "train_loss_llm": 0.0, "grad_norm": 0.04816584289073944, "global_step": 12116, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03468678891658783, "train_loss_bc": 0.03468577191233635, "train_loss_llm": 0.00010162591934204102, "grad_norm": 0.02159268595278263, "global_step": 12117, "epoch": 336, "lr": 0.001} +{"train_loss": 0.048244014382362366, "train_loss_bc": 0.048244014382362366, "train_loss_llm": 0.0, "grad_norm": 0.040048085153102875, "global_step": 12118, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03570778667926788, "train_loss_bc": 0.03569380193948746, "train_loss_llm": 0.0013982951641082764, "grad_norm": 0.05074289068579674, "global_step": 12119, "epoch": 336, "lr": 0.001} +{"train_loss": 0.042785368859767914, "train_loss_bc": 0.042785368859767914, "train_loss_llm": 0.0, "grad_norm": 0.05745408684015274, "global_step": 12120, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03470589593052864, "train_loss_bc": 0.0347016341984272, "train_loss_llm": 0.0004259943962097168, "grad_norm": 0.016125809401273727, "global_step": 12121, "epoch": 336, "lr": 0.001} +{"train_loss": 0.02843073382973671, "train_loss_bc": 0.028425930067896843, "train_loss_llm": 0.00048029422760009766, "grad_norm": 0.025324903428554535, "global_step": 12122, "epoch": 336, "lr": 0.001} +{"train_loss": 0.04005351662635803, "train_loss_bc": 0.04004961624741554, "train_loss_llm": 0.00039002299308776855, "grad_norm": 0.04292004927992821, "global_step": 12123, "epoch": 336, "lr": 0.001} +{"train_loss": 0.04039028286933899, "train_loss_bc": 0.04036000743508339, "train_loss_llm": 0.0030275285243988037, "grad_norm": 0.05333990231156349, "global_step": 12124, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03406213968992233, "train_loss_bc": 0.03403757885098457, "train_loss_llm": 0.002456158399581909, "grad_norm": 0.01950874738395214, "global_step": 12125, "epoch": 336, "lr": 0.001} +{"train_loss": 0.028282297775149345, "train_loss_bc": 0.028274355456233025, "train_loss_llm": 0.0007942616939544678, "grad_norm": 0.02993016503751278, "global_step": 12126, "epoch": 336, "lr": 0.001} +{"train_loss": 0.031932245939970016, "train_loss_bc": 0.03192656487226486, "train_loss_llm": 0.0005682706832885742, "grad_norm": 0.0410928800702095, "global_step": 12127, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03507787734270096, "train_loss_bc": 0.03507787734270096, "train_loss_llm": 0.0, "grad_norm": 0.05460149794816971, "global_step": 12128, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03796713799238205, "train_loss_bc": 0.03796713799238205, "train_loss_llm": 0.0, "grad_norm": 0.023678475990891457, "global_step": 12129, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03896978497505188, "train_loss_bc": 0.03896978497505188, "train_loss_llm": 0.0, "grad_norm": 0.04244144633412361, "global_step": 12130, "epoch": 336, "lr": 0.001} +{"train_loss": 0.03741357987746596, "train_loss_bc": 0.04622074216604233, "train_loss_llm": 0.0015514194965362549, "grad_norm": 0.06860966980457306, "global_step": 12131, "epoch": 336, "lr": 0.001, "val_loss": 0.06349104642868042} +{"train_loss": 0.04396367445588112, "train_loss_bc": 0.043931689113378525, "train_loss_llm": 0.0031987130641937256, "grad_norm": 0.09007865935564041, "global_step": 12132, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03551708534359932, "train_loss_bc": 0.03551708534359932, "train_loss_llm": 0.0, "grad_norm": 0.020002251490950584, "global_step": 12133, "epoch": 337, "lr": 0.001} +{"train_loss": 0.04187148064374924, "train_loss_bc": 0.04187148064374924, "train_loss_llm": 0.0, "grad_norm": 0.025498339906334877, "global_step": 12134, "epoch": 337, "lr": 0.001} +{"train_loss": 0.031159840524196625, "train_loss_bc": 0.031159840524196625, "train_loss_llm": 0.0, "grad_norm": 0.03274843096733093, "global_step": 12135, "epoch": 337, "lr": 0.001} +{"train_loss": 0.044570162892341614, "train_loss_bc": 0.04456941783428192, "train_loss_llm": 7.456541061401367e-05, "grad_norm": 0.03821840509772301, "global_step": 12136, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03351897373795509, "train_loss_bc": 0.033498071134090424, "train_loss_llm": 0.002090364694595337, "grad_norm": 0.02368140034377575, "global_step": 12137, "epoch": 337, "lr": 0.001} +{"train_loss": 0.04182244464755058, "train_loss_bc": 0.04182244464755058, "train_loss_llm": 0.0, "grad_norm": 0.03734922036528587, "global_step": 12138, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03867089003324509, "train_loss_bc": 0.03867089003324509, "train_loss_llm": 0.0, "grad_norm": 0.04651062935590744, "global_step": 12139, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03374366834759712, "train_loss_bc": 0.03373267129063606, "train_loss_llm": 0.0010995566844940186, "grad_norm": 0.05869809165596962, "global_step": 12140, "epoch": 337, "lr": 0.001} +{"train_loss": 0.04449952766299248, "train_loss_bc": 0.044498588889837265, "train_loss_llm": 9.381771087646484e-05, "grad_norm": 0.019758962094783783, "global_step": 12141, "epoch": 337, "lr": 0.001} +{"train_loss": 0.032195355743169785, "train_loss_bc": 0.032195281237363815, "train_loss_llm": 7.569789886474609e-06, "grad_norm": 0.028087276965379715, "global_step": 12142, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03398624435067177, "train_loss_bc": 0.03398402780294418, "train_loss_llm": 0.00022149085998535156, "grad_norm": 0.04161117970943451, "global_step": 12143, "epoch": 337, "lr": 0.001} +{"train_loss": 0.040348153561353683, "train_loss_bc": 0.04034222662448883, "train_loss_llm": 0.0005925893783569336, "grad_norm": 0.047320373356342316, "global_step": 12144, "epoch": 337, "lr": 0.001} +{"train_loss": 0.030164187774062157, "train_loss_bc": 0.030164187774062157, "train_loss_llm": 0.0, "grad_norm": 0.01820860058069229, "global_step": 12145, "epoch": 337, "lr": 0.001} +{"train_loss": 0.040928494185209274, "train_loss_bc": 0.040928494185209274, "train_loss_llm": 0.0, "grad_norm": 0.023865506052970886, "global_step": 12146, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03185209631919861, "train_loss_bc": 0.03185209631919861, "train_loss_llm": 0.0, "grad_norm": 0.034415103495121, "global_step": 12147, "epoch": 337, "lr": 0.001} +{"train_loss": 0.035339467227458954, "train_loss_bc": 0.035339467227458954, "train_loss_llm": 0.0, "grad_norm": 0.044188037514686584, "global_step": 12148, "epoch": 337, "lr": 0.001} +{"train_loss": 0.04200728237628937, "train_loss_bc": 0.04200728237628937, "train_loss_llm": 0.0, "grad_norm": 0.022241707891225815, "global_step": 12149, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03428792208433151, "train_loss_bc": 0.03428792208433151, "train_loss_llm": 0.0, "grad_norm": 0.031558386981487274, "global_step": 12150, "epoch": 337, "lr": 0.001} +{"train_loss": 0.041226692497730255, "train_loss_bc": 0.041218578815460205, "train_loss_llm": 0.0008113682270050049, "grad_norm": 0.04650796204805374, "global_step": 12151, "epoch": 337, "lr": 0.001} +{"train_loss": 0.034116100519895554, "train_loss_bc": 0.034111764281988144, "train_loss_llm": 0.0004337728023529053, "grad_norm": 0.06431742757558823, "global_step": 12152, "epoch": 337, "lr": 0.001} +{"train_loss": 0.027565568685531616, "train_loss_bc": 0.02755911834537983, "train_loss_llm": 0.0006450414657592773, "grad_norm": 0.021079065278172493, "global_step": 12153, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03735659271478653, "train_loss_bc": 0.03735659271478653, "train_loss_llm": 0.0, "grad_norm": 0.03386218845844269, "global_step": 12154, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03443051874637604, "train_loss_bc": 0.03441693261265755, "train_loss_llm": 0.0013585686683654785, "grad_norm": 0.044460833072662354, "global_step": 12155, "epoch": 337, "lr": 0.001} +{"train_loss": 0.04414304718375206, "train_loss_bc": 0.04414304718375206, "train_loss_llm": 0.0, "grad_norm": 0.06283428519964218, "global_step": 12156, "epoch": 337, "lr": 0.001} +{"train_loss": 0.04114757850766182, "train_loss_bc": 0.04114757850766182, "train_loss_llm": 0.0, "grad_norm": 0.02794448472559452, "global_step": 12157, "epoch": 337, "lr": 0.001} +{"train_loss": 0.0396527536213398, "train_loss_bc": 0.03965261951088905, "train_loss_llm": 1.3589859008789062e-05, "grad_norm": 0.04350727051496506, "global_step": 12158, "epoch": 337, "lr": 0.001} +{"train_loss": 0.0329829677939415, "train_loss_bc": 0.03298012912273407, "train_loss_llm": 0.0002836883068084717, "grad_norm": 0.058329418301582336, "global_step": 12159, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03511163592338562, "train_loss_bc": 0.03511163592338562, "train_loss_llm": 0.0, "grad_norm": 0.06903205066919327, "global_step": 12160, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03670626878738403, "train_loss_bc": 0.036689262837171555, "train_loss_llm": 0.0017005503177642822, "grad_norm": 0.0213968213647604, "global_step": 12161, "epoch": 337, "lr": 0.001} +{"train_loss": 0.032998960465192795, "train_loss_bc": 0.032998960465192795, "train_loss_llm": 0.0, "grad_norm": 0.029730139300227165, "global_step": 12162, "epoch": 337, "lr": 0.001} +{"train_loss": 0.035475049167871475, "train_loss_bc": 0.035475049167871475, "train_loss_llm": 0.0, "grad_norm": 0.03816989064216614, "global_step": 12163, "epoch": 337, "lr": 0.001} +{"train_loss": 0.04020104557275772, "train_loss_bc": 0.04020104557275772, "train_loss_llm": 0.0, "grad_norm": 0.05309692397713661, "global_step": 12164, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03941797465085983, "train_loss_bc": 0.03939887881278992, "train_loss_llm": 0.001909494400024414, "grad_norm": 0.02566450834274292, "global_step": 12165, "epoch": 337, "lr": 0.001} +{"train_loss": 0.03647904843091965, "train_loss_bc": 0.03647904843091965, "train_loss_llm": 0.0, "grad_norm": 0.039856940507888794, "global_step": 12166, "epoch": 337, "lr": 0.001} +{"train_loss": 0.037083300606658064, "train_loss_bc": 0.03554006665945053, "train_loss_llm": 0.0, "grad_norm": 0.05633123591542244, "global_step": 12167, "epoch": 337, "lr": 0.001, "val_loss": 0.06419676542282104} +{"train_loss": 0.032356951385736465, "train_loss_bc": 0.032356951385736465, "train_loss_llm": 0.0, "grad_norm": 0.0633028894662857, "global_step": 12168, "epoch": 338, "lr": 0.001} +{"train_loss": 0.042988523840904236, "train_loss_bc": 0.042988523840904236, "train_loss_llm": 0.0, "grad_norm": 0.022816743701696396, "global_step": 12169, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03684978187084198, "train_loss_bc": 0.03684978187084198, "train_loss_llm": 0.0, "grad_norm": 0.03710297495126724, "global_step": 12170, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03360694274306297, "train_loss_bc": 0.03360694274306297, "train_loss_llm": 0.0, "grad_norm": 0.04071013629436493, "global_step": 12171, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03434668853878975, "train_loss_bc": 0.03434668853878975, "train_loss_llm": 0.0, "grad_norm": 0.046025779098272324, "global_step": 12172, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03673416003584862, "train_loss_bc": 0.03673416003584862, "train_loss_llm": 0.0, "grad_norm": 0.018862124532461166, "global_step": 12173, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03911281004548073, "train_loss_bc": 0.03911281004548073, "train_loss_llm": 0.0, "grad_norm": 0.03105313703417778, "global_step": 12174, "epoch": 338, "lr": 0.001} +{"train_loss": 0.0335083082318306, "train_loss_bc": 0.03349078819155693, "train_loss_llm": 0.001752018928527832, "grad_norm": 0.04152602329850197, "global_step": 12175, "epoch": 338, "lr": 0.001} +{"train_loss": 0.04205949604511261, "train_loss_bc": 0.04205949604511261, "train_loss_llm": 0.0, "grad_norm": 0.0501079224050045, "global_step": 12176, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03312535956501961, "train_loss_bc": 0.03312535956501961, "train_loss_llm": 0.0, "grad_norm": 0.024797821417450905, "global_step": 12177, "epoch": 338, "lr": 0.001} +{"train_loss": 0.037596456706523895, "train_loss_bc": 0.03759421780705452, "train_loss_llm": 0.00022390484809875488, "grad_norm": 0.03898335620760918, "global_step": 12178, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03679961338639259, "train_loss_bc": 0.03679961338639259, "train_loss_llm": 0.0, "grad_norm": 0.05874032527208328, "global_step": 12179, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03685912489891052, "train_loss_bc": 0.03685912489891052, "train_loss_llm": 0.0, "grad_norm": 0.07111196219921112, "global_step": 12180, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03758355975151062, "train_loss_bc": 0.03758355975151062, "train_loss_llm": 0.0, "grad_norm": 0.019067546352744102, "global_step": 12181, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03543343394994736, "train_loss_bc": 0.03543343394994736, "train_loss_llm": 0.0, "grad_norm": 0.02829406037926674, "global_step": 12182, "epoch": 338, "lr": 0.001} +{"train_loss": 0.040698204189538956, "train_loss_bc": 0.0406913198530674, "train_loss_llm": 0.0006882548332214355, "grad_norm": 0.037485331296920776, "global_step": 12183, "epoch": 338, "lr": 0.001} +{"train_loss": 0.039391692727804184, "train_loss_bc": 0.039391692727804184, "train_loss_llm": 0.0, "grad_norm": 0.052607521414756775, "global_step": 12184, "epoch": 338, "lr": 0.001} +{"train_loss": 0.04215655475854874, "train_loss_bc": 0.04215583950281143, "train_loss_llm": 7.143616676330566e-05, "grad_norm": 0.023369429633021355, "global_step": 12185, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03124506212770939, "train_loss_bc": 0.031236812472343445, "train_loss_llm": 0.0008250176906585693, "grad_norm": 0.03448723256587982, "global_step": 12186, "epoch": 338, "lr": 0.001} +{"train_loss": 0.04483429715037346, "train_loss_bc": 0.04483429715037346, "train_loss_llm": 0.0, "grad_norm": 0.04279753193259239, "global_step": 12187, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03788858279585838, "train_loss_bc": 0.03788594901561737, "train_loss_llm": 0.0002632439136505127, "grad_norm": 0.053569257259368896, "global_step": 12188, "epoch": 338, "lr": 0.001} +{"train_loss": 0.033807992935180664, "train_loss_bc": 0.033807992935180664, "train_loss_llm": 0.0, "grad_norm": 0.01867859996855259, "global_step": 12189, "epoch": 338, "lr": 0.001} +{"train_loss": 0.04414281249046326, "train_loss_bc": 0.04414281249046326, "train_loss_llm": 0.0, "grad_norm": 0.034713175147771835, "global_step": 12190, "epoch": 338, "lr": 0.001} +{"train_loss": 0.035417743027210236, "train_loss_bc": 0.035417743027210236, "train_loss_llm": 0.0, "grad_norm": 0.03980425000190735, "global_step": 12191, "epoch": 338, "lr": 0.001} +{"train_loss": 0.039091870188713074, "train_loss_bc": 0.039091870188713074, "train_loss_llm": 0.0, "grad_norm": 0.044577669352293015, "global_step": 12192, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03926656022667885, "train_loss_bc": 0.03922409564256668, "train_loss_llm": 0.00424647331237793, "grad_norm": 0.02083885669708252, "global_step": 12193, "epoch": 338, "lr": 0.001} +{"train_loss": 0.033281631767749786, "train_loss_bc": 0.033281631767749786, "train_loss_llm": 0.0, "grad_norm": 0.03216259554028511, "global_step": 12194, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03303810581564903, "train_loss_bc": 0.03303810581564903, "train_loss_llm": 0.0, "grad_norm": 0.03978096321225166, "global_step": 12195, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03678895905613899, "train_loss_bc": 0.03678895905613899, "train_loss_llm": 0.0, "grad_norm": 0.05052661895751953, "global_step": 12196, "epoch": 338, "lr": 0.001} +{"train_loss": 0.04227287694811821, "train_loss_bc": 0.04226217418909073, "train_loss_llm": 0.0010703504085540771, "grad_norm": 0.02493097260594368, "global_step": 12197, "epoch": 338, "lr": 0.001} +{"train_loss": 0.038389723747968674, "train_loss_bc": 0.03838057070970535, "train_loss_llm": 0.0009154677391052246, "grad_norm": 0.03777572140097618, "global_step": 12198, "epoch": 338, "lr": 0.001} +{"train_loss": 0.037397369742393494, "train_loss_bc": 0.037397369742393494, "train_loss_llm": 0.0, "grad_norm": 0.047657012939453125, "global_step": 12199, "epoch": 338, "lr": 0.001} +{"train_loss": 0.030530119314789772, "train_loss_bc": 0.030515313148498535, "train_loss_llm": 0.0014806389808654785, "grad_norm": 0.05854698643088341, "global_step": 12200, "epoch": 338, "lr": 0.001} +{"train_loss": 0.029490670189261436, "train_loss_bc": 0.029490670189261436, "train_loss_llm": 0.0, "grad_norm": 0.016837812960147858, "global_step": 12201, "epoch": 338, "lr": 0.001} +{"train_loss": 0.034804537892341614, "train_loss_bc": 0.034804537892341614, "train_loss_llm": 0.0, "grad_norm": 0.021510528400540352, "global_step": 12202, "epoch": 338, "lr": 0.001} +{"train_loss": 0.03698948724195361, "train_loss_bc": 0.03870421275496483, "train_loss_llm": 0.002075105905532837, "grad_norm": 0.041002750396728516, "global_step": 12203, "epoch": 338, "lr": 0.001, "val_loss": 0.06419920921325684} +{"train_loss": 0.03426935896277428, "train_loss_bc": 0.03426935896277428, "train_loss_llm": 0.0, "grad_norm": 0.05108777433633804, "global_step": 12204, "epoch": 339, "lr": 0.001} +{"train_loss": 0.035144705325365067, "train_loss_bc": 0.035144705325365067, "train_loss_llm": 0.0, "grad_norm": 0.02020488865673542, "global_step": 12205, "epoch": 339, "lr": 0.001} +{"train_loss": 0.04104156047105789, "train_loss_bc": 0.04104156047105789, "train_loss_llm": 0.0, "grad_norm": 0.03387605771422386, "global_step": 12206, "epoch": 339, "lr": 0.001} +{"train_loss": 0.04336801916360855, "train_loss_bc": 0.043361373245716095, "train_loss_llm": 0.000664442777633667, "grad_norm": 0.0524296835064888, "global_step": 12207, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03861355409026146, "train_loss_bc": 0.03860227018594742, "train_loss_llm": 0.0011283159255981445, "grad_norm": 0.0645957961678505, "global_step": 12208, "epoch": 339, "lr": 0.001} +{"train_loss": 0.031178729608654976, "train_loss_bc": 0.031178729608654976, "train_loss_llm": 0.0, "grad_norm": 0.019453011453151703, "global_step": 12209, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03660277649760246, "train_loss_bc": 0.03660277649760246, "train_loss_llm": 0.0, "grad_norm": 0.03337754309177399, "global_step": 12210, "epoch": 339, "lr": 0.001} +{"train_loss": 0.038779281079769135, "train_loss_bc": 0.0387762151658535, "train_loss_llm": 0.0003065764904022217, "grad_norm": 0.04235713556408882, "global_step": 12211, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03678269684314728, "train_loss_bc": 0.03678269684314728, "train_loss_llm": 0.0, "grad_norm": 0.057445235550403595, "global_step": 12212, "epoch": 339, "lr": 0.001} +{"train_loss": 0.04103662073612213, "train_loss_bc": 0.04103662073612213, "train_loss_llm": 0.0, "grad_norm": 0.02558158151805401, "global_step": 12213, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03539600223302841, "train_loss_bc": 0.03539600223302841, "train_loss_llm": 0.0, "grad_norm": 0.039254240691661835, "global_step": 12214, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03582701459527016, "train_loss_bc": 0.03581314533948898, "train_loss_llm": 0.0013869702816009521, "grad_norm": 0.05029832199215889, "global_step": 12215, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03384458273649216, "train_loss_bc": 0.03384458273649216, "train_loss_llm": 0.0, "grad_norm": 0.056000709533691406, "global_step": 12216, "epoch": 339, "lr": 0.001} +{"train_loss": 0.04209720343351364, "train_loss_bc": 0.04209720343351364, "train_loss_llm": 0.0, "grad_norm": 0.025239955633878708, "global_step": 12217, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03469594568014145, "train_loss_bc": 0.03469594568014145, "train_loss_llm": 0.0, "grad_norm": 0.03272292762994766, "global_step": 12218, "epoch": 339, "lr": 0.001} +{"train_loss": 0.04052925109863281, "train_loss_bc": 0.04052530601620674, "train_loss_llm": 0.00039458274841308594, "grad_norm": 0.04309241101145744, "global_step": 12219, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03475932031869888, "train_loss_bc": 0.03475932031869888, "train_loss_llm": 0.0, "grad_norm": 0.04902203381061554, "global_step": 12220, "epoch": 339, "lr": 0.001} +{"train_loss": 0.033021263778209686, "train_loss_bc": 0.033021263778209686, "train_loss_llm": 0.0, "grad_norm": 0.01692986860871315, "global_step": 12221, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03665531799197197, "train_loss_bc": 0.03665531799197197, "train_loss_llm": 0.0, "grad_norm": 0.03202133625745773, "global_step": 12222, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03857315704226494, "train_loss_bc": 0.03857315704226494, "train_loss_llm": 0.0, "grad_norm": 0.04188067466020584, "global_step": 12223, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03817851096391678, "train_loss_bc": 0.03817851096391678, "train_loss_llm": 0.0, "grad_norm": 0.057383276522159576, "global_step": 12224, "epoch": 339, "lr": 0.001} +{"train_loss": 0.034315548837184906, "train_loss_bc": 0.034315548837184906, "train_loss_llm": 0.0, "grad_norm": 0.02108726091682911, "global_step": 12225, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03931717947125435, "train_loss_bc": 0.03930973261594772, "train_loss_llm": 0.0007447600364685059, "grad_norm": 0.04028046503663063, "global_step": 12226, "epoch": 339, "lr": 0.001} +{"train_loss": 0.028030818328261375, "train_loss_bc": 0.028030818328261375, "train_loss_llm": 0.0, "grad_norm": 0.04854976385831833, "global_step": 12227, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03151785954833031, "train_loss_bc": 0.03151785954833031, "train_loss_llm": 0.0, "grad_norm": 0.055114515125751495, "global_step": 12228, "epoch": 339, "lr": 0.001} +{"train_loss": 0.037545010447502136, "train_loss_bc": 0.037545010447502136, "train_loss_llm": 0.0, "grad_norm": 0.022205481305718422, "global_step": 12229, "epoch": 339, "lr": 0.001} +{"train_loss": 0.04309443011879921, "train_loss_bc": 0.04309443011879921, "train_loss_llm": 0.0, "grad_norm": 0.039798762649297714, "global_step": 12230, "epoch": 339, "lr": 0.001} +{"train_loss": 0.040055304765701294, "train_loss_bc": 0.040055304765701294, "train_loss_llm": 0.0, "grad_norm": 0.05682497099041939, "global_step": 12231, "epoch": 339, "lr": 0.001} +{"train_loss": 0.038024935871362686, "train_loss_bc": 0.038024935871362686, "train_loss_llm": 0.0, "grad_norm": 0.07240966707468033, "global_step": 12232, "epoch": 339, "lr": 0.001} +{"train_loss": 0.04161791503429413, "train_loss_bc": 0.04159800708293915, "train_loss_llm": 0.0019908547401428223, "grad_norm": 0.02081579715013504, "global_step": 12233, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03950592130422592, "train_loss_bc": 0.03950592130422592, "train_loss_llm": 0.0, "grad_norm": 0.029183709993958473, "global_step": 12234, "epoch": 339, "lr": 0.001} +{"train_loss": 0.028910880908370018, "train_loss_bc": 0.02890743874013424, "train_loss_llm": 0.0003441274166107178, "grad_norm": 0.03342786058783531, "global_step": 12235, "epoch": 339, "lr": 0.001} +{"train_loss": 0.03673110157251358, "train_loss_bc": 0.03673110157251358, "train_loss_llm": 0.0, "grad_norm": 0.04413104057312012, "global_step": 12236, "epoch": 339, "lr": 0.001} +{"train_loss": 0.038067858666181564, "train_loss_bc": 0.038067858666181564, "train_loss_llm": 0.0, "grad_norm": 0.02031862363219261, "global_step": 12237, "epoch": 339, "lr": 0.001} +{"train_loss": 0.04329703375697136, "train_loss_bc": 0.04329703375697136, "train_loss_llm": 0.0, "grad_norm": 0.03497491031885147, "global_step": 12238, "epoch": 339, "lr": 0.001} +{"train_loss": 0.036959819288717374, "train_loss_bc": 0.030109979212284088, "train_loss_llm": 0.0016843676567077637, "grad_norm": 0.04278317838907242, "global_step": 12239, "epoch": 339, "lr": 0.001, "val_loss": 0.06353724002838135} +{"train_loss": 0.029185622930526733, "train_loss_bc": 0.029185622930526733, "train_loss_llm": 0.0, "grad_norm": 0.04956667497754097, "global_step": 12240, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03331734985113144, "train_loss_bc": 0.03331734985113144, "train_loss_llm": 0.0, "grad_norm": 0.01499564666301012, "global_step": 12241, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03575994446873665, "train_loss_bc": 0.03575994446873665, "train_loss_llm": 0.0, "grad_norm": 0.025547701865434647, "global_step": 12242, "epoch": 340, "lr": 0.001} +{"train_loss": 0.032531339675188065, "train_loss_bc": 0.032531339675188065, "train_loss_llm": 0.0, "grad_norm": 0.03398991376161575, "global_step": 12243, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03982473909854889, "train_loss_bc": 0.03981827199459076, "train_loss_llm": 0.0006467103958129883, "grad_norm": 0.04292415454983711, "global_step": 12244, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03840605542063713, "train_loss_bc": 0.03840605542063713, "train_loss_llm": 0.0, "grad_norm": 0.021094053983688354, "global_step": 12245, "epoch": 340, "lr": 0.001} +{"train_loss": 0.040395352989435196, "train_loss_bc": 0.04037974402308464, "train_loss_llm": 0.0015610158443450928, "grad_norm": 0.0302402526140213, "global_step": 12246, "epoch": 340, "lr": 0.001} +{"train_loss": 0.038909874856472015, "train_loss_bc": 0.038909874856472015, "train_loss_llm": 0.0, "grad_norm": 0.041689105331897736, "global_step": 12247, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03549505025148392, "train_loss_bc": 0.03549505025148392, "train_loss_llm": 0.0, "grad_norm": 0.04540041461586952, "global_step": 12248, "epoch": 340, "lr": 0.001} +{"train_loss": 0.036700110882520676, "train_loss_bc": 0.03669877350330353, "train_loss_llm": 0.00013366341590881348, "grad_norm": 0.018849363550543785, "global_step": 12249, "epoch": 340, "lr": 0.001} +{"train_loss": 0.0421118326485157, "train_loss_bc": 0.0421118326485157, "train_loss_llm": 0.0, "grad_norm": 0.02465996891260147, "global_step": 12250, "epoch": 340, "lr": 0.001} +{"train_loss": 0.036043353378772736, "train_loss_bc": 0.03604086861014366, "train_loss_llm": 0.00024831295013427734, "grad_norm": 0.0348249226808548, "global_step": 12251, "epoch": 340, "lr": 0.001} +{"train_loss": 0.04539415240287781, "train_loss_bc": 0.04539415240287781, "train_loss_llm": 0.0, "grad_norm": 0.048152823001146317, "global_step": 12252, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03655436262488365, "train_loss_bc": 0.036542076617479324, "train_loss_llm": 0.001228630542755127, "grad_norm": 0.025039298459887505, "global_step": 12253, "epoch": 340, "lr": 0.001} +{"train_loss": 0.039923351258039474, "train_loss_bc": 0.039923351258039474, "train_loss_llm": 0.0, "grad_norm": 0.03397548571228981, "global_step": 12254, "epoch": 340, "lr": 0.001} +{"train_loss": 0.0366414412856102, "train_loss_bc": 0.0366414412856102, "train_loss_llm": 0.0, "grad_norm": 0.040708210319280624, "global_step": 12255, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03417468070983887, "train_loss_bc": 0.03416677564382553, "train_loss_llm": 0.0007906556129455566, "grad_norm": 0.04942093417048454, "global_step": 12256, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03311789780855179, "train_loss_bc": 0.03310878202319145, "train_loss_llm": 0.0009115934371948242, "grad_norm": 0.021374406293034554, "global_step": 12257, "epoch": 340, "lr": 0.001} +{"train_loss": 0.036074697971343994, "train_loss_bc": 0.03606196492910385, "train_loss_llm": 0.0012731552124023438, "grad_norm": 0.02637139894068241, "global_step": 12258, "epoch": 340, "lr": 0.001} +{"train_loss": 0.035961754620075226, "train_loss_bc": 0.0359477698802948, "train_loss_llm": 0.001398622989654541, "grad_norm": 0.03707542270421982, "global_step": 12259, "epoch": 340, "lr": 0.001} +{"train_loss": 0.037065289914608, "train_loss_bc": 0.037065289914608, "train_loss_llm": 0.0, "grad_norm": 0.04624783247709274, "global_step": 12260, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03464150428771973, "train_loss_bc": 0.03464150428771973, "train_loss_llm": 0.0, "grad_norm": 0.021956531330943108, "global_step": 12261, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03573840856552124, "train_loss_bc": 0.03572287783026695, "train_loss_llm": 0.001552969217300415, "grad_norm": 0.032034385949373245, "global_step": 12262, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03570050373673439, "train_loss_bc": 0.03567250072956085, "train_loss_llm": 0.0028001368045806885, "grad_norm": 0.046778976917266846, "global_step": 12263, "epoch": 340, "lr": 0.001} +{"train_loss": 0.04122011736035347, "train_loss_bc": 0.04122011736035347, "train_loss_llm": 0.0, "grad_norm": 0.06645487993955612, "global_step": 12264, "epoch": 340, "lr": 0.001} +{"train_loss": 0.042823199182748795, "train_loss_bc": 0.042823199182748795, "train_loss_llm": 0.0, "grad_norm": 0.03668411821126938, "global_step": 12265, "epoch": 340, "lr": 0.001} +{"train_loss": 0.04049069434404373, "train_loss_bc": 0.04049069434404373, "train_loss_llm": 0.0, "grad_norm": 0.04776177927851677, "global_step": 12266, "epoch": 340, "lr": 0.001} +{"train_loss": 0.031066613271832466, "train_loss_bc": 0.031060129404067993, "train_loss_llm": 0.0006483495235443115, "grad_norm": 0.05533578619360924, "global_step": 12267, "epoch": 340, "lr": 0.001} +{"train_loss": 0.033499449491500854, "train_loss_bc": 0.033499449491500854, "train_loss_llm": 0.0, "grad_norm": 0.07372208684682846, "global_step": 12268, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03429754078388214, "train_loss_bc": 0.03429754078388214, "train_loss_llm": 0.0, "grad_norm": 0.020055953413248062, "global_step": 12269, "epoch": 340, "lr": 0.001} +{"train_loss": 0.042338356375694275, "train_loss_bc": 0.042338356375694275, "train_loss_llm": 0.0, "grad_norm": 0.028884282335639, "global_step": 12270, "epoch": 340, "lr": 0.001} +{"train_loss": 0.038285065442323685, "train_loss_bc": 0.038280654698610306, "train_loss_llm": 0.00044098496437072754, "grad_norm": 0.03767336159944534, "global_step": 12271, "epoch": 340, "lr": 0.001} +{"train_loss": 0.03267373517155647, "train_loss_bc": 0.03266600891947746, "train_loss_llm": 0.000772625207901001, "grad_norm": 0.04444724693894386, "global_step": 12272, "epoch": 340, "lr": 0.001} +{"train_loss": 0.036254990845918655, "train_loss_bc": 0.036254990845918655, "train_loss_llm": 0.0, "grad_norm": 0.021622877568006516, "global_step": 12273, "epoch": 340, "lr": 0.001} +{"train_loss": 0.0348987840116024, "train_loss_bc": 0.0348987840116024, "train_loss_llm": 0.0, "grad_norm": 0.034579820930957794, "global_step": 12274, "epoch": 340, "lr": 0.001} +{"train_loss": 0.037036394959108696, "train_loss_bc": 0.04579300060868263, "train_loss_llm": 0.0, "grad_norm": 0.05406416207551956, "global_step": 12275, "epoch": 340, "lr": 0.001, "train/cumulative_reward": 68.22244929848198, "train/mean_score": 9.62801956622241, "train/success_rate": 0.1, "test/cumulative_reward": 44.99568970679255, "test/mean_score": 7.4287150456311295, "test/success_rate": 0.16, "val_loss": 0.06477555632591248, "train_action_mse_error": 0.03987779840826988} +{"train_loss": 0.035187553614377975, "train_loss_bc": 0.035187553614377975, "train_loss_llm": 0.0, "grad_norm": 0.05958953872323036, "global_step": 12276, "epoch": 341, "lr": 0.001} +{"train_loss": 0.040059156715869904, "train_loss_bc": 0.040059156715869904, "train_loss_llm": 0.0, "grad_norm": 0.025221645832061768, "global_step": 12277, "epoch": 341, "lr": 0.001} +{"train_loss": 0.034527093172073364, "train_loss_bc": 0.034527093172073364, "train_loss_llm": 0.0, "grad_norm": 0.033315371721982956, "global_step": 12278, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03914449363946915, "train_loss_bc": 0.03914039209485054, "train_loss_llm": 0.0004100501537322998, "grad_norm": 0.04994133859872818, "global_step": 12279, "epoch": 341, "lr": 0.001} +{"train_loss": 0.041220877319574356, "train_loss_bc": 0.041220877319574356, "train_loss_llm": 0.0, "grad_norm": 0.06649084389209747, "global_step": 12280, "epoch": 341, "lr": 0.001} +{"train_loss": 0.042165808379650116, "train_loss_bc": 0.042165808379650116, "train_loss_llm": 0.0, "grad_norm": 0.02339821495115757, "global_step": 12281, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03681602701544762, "train_loss_bc": 0.03681602701544762, "train_loss_llm": 0.0, "grad_norm": 0.03930709883570671, "global_step": 12282, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03578386828303337, "train_loss_bc": 0.03578386828303337, "train_loss_llm": 0.0, "grad_norm": 0.05447018891572952, "global_step": 12283, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03308228775858879, "train_loss_bc": 0.03308124840259552, "train_loss_llm": 0.00010386109352111816, "grad_norm": 0.07352815568447113, "global_step": 12284, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03485807403922081, "train_loss_bc": 0.03485807403922081, "train_loss_llm": 0.0, "grad_norm": 0.01792055554687977, "global_step": 12285, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03820560500025749, "train_loss_bc": 0.038185589015483856, "train_loss_llm": 0.0020016729831695557, "grad_norm": 0.032916903495788574, "global_step": 12286, "epoch": 341, "lr": 0.001} +{"train_loss": 0.033853545784950256, "train_loss_bc": 0.033853545784950256, "train_loss_llm": 0.0, "grad_norm": 0.04472600668668747, "global_step": 12287, "epoch": 341, "lr": 0.001} +{"train_loss": 0.039766598492860794, "train_loss_bc": 0.039766598492860794, "train_loss_llm": 0.0, "grad_norm": 0.05800590664148331, "global_step": 12288, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03876819461584091, "train_loss_bc": 0.03876686468720436, "train_loss_llm": 0.00013309717178344727, "grad_norm": 0.0245677437633276, "global_step": 12289, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03727192431688309, "train_loss_bc": 0.03727192431688309, "train_loss_llm": 0.0, "grad_norm": 0.03370009362697601, "global_step": 12290, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03168622776865959, "train_loss_bc": 0.03168622776865959, "train_loss_llm": 0.0, "grad_norm": 0.044899825006723404, "global_step": 12291, "epoch": 341, "lr": 0.001} +{"train_loss": 0.034173887223005295, "train_loss_bc": 0.034173887223005295, "train_loss_llm": 0.0, "grad_norm": 0.05405130609869957, "global_step": 12292, "epoch": 341, "lr": 0.001} +{"train_loss": 0.037559349089860916, "train_loss_bc": 0.037538912147283554, "train_loss_llm": 0.0020437538623809814, "grad_norm": 0.029252968728542328, "global_step": 12293, "epoch": 341, "lr": 0.001} +{"train_loss": 0.032705120742321014, "train_loss_bc": 0.032705120742321014, "train_loss_llm": 0.0, "grad_norm": 0.04100693017244339, "global_step": 12294, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03709886223077774, "train_loss_bc": 0.03709886223077774, "train_loss_llm": 0.0, "grad_norm": 0.04386982321739197, "global_step": 12295, "epoch": 341, "lr": 0.001} +{"train_loss": 0.04436039179563522, "train_loss_bc": 0.04436039179563522, "train_loss_llm": 0.0, "grad_norm": 0.05437437817454338, "global_step": 12296, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03663424029946327, "train_loss_bc": 0.03663424029946327, "train_loss_llm": 0.0, "grad_norm": 0.0218502264469862, "global_step": 12297, "epoch": 341, "lr": 0.001} +{"train_loss": 0.028491079807281494, "train_loss_bc": 0.028491079807281494, "train_loss_llm": 0.0, "grad_norm": 0.03306077793240547, "global_step": 12298, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03592453896999359, "train_loss_bc": 0.03592453896999359, "train_loss_llm": 0.0, "grad_norm": 0.04815312474966049, "global_step": 12299, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03911434859037399, "train_loss_bc": 0.03909650444984436, "train_loss_llm": 0.0017843246459960938, "grad_norm": 0.06446796655654907, "global_step": 12300, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03581944480538368, "train_loss_bc": 0.03580779582262039, "train_loss_llm": 0.0011650621891021729, "grad_norm": 0.021143605932593346, "global_step": 12301, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03554622083902359, "train_loss_bc": 0.03554622083902359, "train_loss_llm": 0.0, "grad_norm": 0.03273520991206169, "global_step": 12302, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03865252807736397, "train_loss_bc": 0.03865252807736397, "train_loss_llm": 0.0, "grad_norm": 0.04551439359784126, "global_step": 12303, "epoch": 341, "lr": 0.001} +{"train_loss": 0.0435967855155468, "train_loss_bc": 0.0435967855155468, "train_loss_llm": 0.0, "grad_norm": 0.05832602083683014, "global_step": 12304, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03985178843140602, "train_loss_bc": 0.03985178843140602, "train_loss_llm": 0.0, "grad_norm": 0.021420584991574287, "global_step": 12305, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03653590753674507, "train_loss_bc": 0.03653590753674507, "train_loss_llm": 0.0, "grad_norm": 0.02835681661963463, "global_step": 12306, "epoch": 341, "lr": 0.001} +{"train_loss": 0.039405398070812225, "train_loss_bc": 0.03939927741885185, "train_loss_llm": 0.0006119906902313232, "grad_norm": 0.039719682186841965, "global_step": 12307, "epoch": 341, "lr": 0.001} +{"train_loss": 0.033459220081567764, "train_loss_bc": 0.033459220081567764, "train_loss_llm": 0.0, "grad_norm": 0.049564626067876816, "global_step": 12308, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03902799263596535, "train_loss_bc": 0.03902799263596535, "train_loss_llm": 0.0, "grad_norm": 0.025756895542144775, "global_step": 12309, "epoch": 341, "lr": 0.001} +{"train_loss": 0.04370657727122307, "train_loss_bc": 0.043688610196113586, "train_loss_llm": 0.001796722412109375, "grad_norm": 0.035882990807294846, "global_step": 12310, "epoch": 341, "lr": 0.001} +{"train_loss": 0.03693801268107361, "train_loss_bc": 0.025707438588142395, "train_loss_llm": 0.0, "grad_norm": 0.043313153088092804, "global_step": 12311, "epoch": 341, "lr": 0.001, "val_loss": 0.06378844380378723} +{"train_loss": 0.036940280348062515, "train_loss_bc": 0.036940280348062515, "train_loss_llm": 0.0, "grad_norm": 0.05380025878548622, "global_step": 12312, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03249119967222214, "train_loss_bc": 0.03249119967222214, "train_loss_llm": 0.0, "grad_norm": 0.0196242518723011, "global_step": 12313, "epoch": 342, "lr": 0.001} +{"train_loss": 0.04008810594677925, "train_loss_bc": 0.04008810594677925, "train_loss_llm": 0.0, "grad_norm": 0.03474833816289902, "global_step": 12314, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03667477145791054, "train_loss_bc": 0.03667477145791054, "train_loss_llm": 0.0, "grad_norm": 0.04495367780327797, "global_step": 12315, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03798577934503555, "train_loss_bc": 0.03798577934503555, "train_loss_llm": 0.0, "grad_norm": 0.04846934229135513, "global_step": 12316, "epoch": 342, "lr": 0.001} +{"train_loss": 0.033562298864126205, "train_loss_bc": 0.03355489298701286, "train_loss_llm": 0.0007405579090118408, "grad_norm": 0.020168190822005272, "global_step": 12317, "epoch": 342, "lr": 0.001} +{"train_loss": 0.033461809158325195, "train_loss_bc": 0.033461809158325195, "train_loss_llm": 0.0, "grad_norm": 0.03348071128129959, "global_step": 12318, "epoch": 342, "lr": 0.001} +{"train_loss": 0.035884417593479156, "train_loss_bc": 0.03588160127401352, "train_loss_llm": 0.0002814829349517822, "grad_norm": 0.04011737182736397, "global_step": 12319, "epoch": 342, "lr": 0.001} +{"train_loss": 0.043553631752729416, "train_loss_bc": 0.043553631752729416, "train_loss_llm": 0.0, "grad_norm": 0.04552144557237625, "global_step": 12320, "epoch": 342, "lr": 0.001} +{"train_loss": 0.04308051988482475, "train_loss_bc": 0.04308051988482475, "train_loss_llm": 0.0, "grad_norm": 0.030349599197506905, "global_step": 12321, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03413661569356918, "train_loss_bc": 0.03413661569356918, "train_loss_llm": 0.0, "grad_norm": 0.03640463948249817, "global_step": 12322, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03174830973148346, "train_loss_bc": 0.03174400329589844, "train_loss_llm": 0.0004305839538574219, "grad_norm": 0.03559240326285362, "global_step": 12323, "epoch": 342, "lr": 0.001} +{"train_loss": 0.034060437232255936, "train_loss_bc": 0.034060437232255936, "train_loss_llm": 0.0, "grad_norm": 0.038331545889377594, "global_step": 12324, "epoch": 342, "lr": 0.001} +{"train_loss": 0.037470221519470215, "train_loss_bc": 0.037466686218976974, "train_loss_llm": 0.00035369396209716797, "grad_norm": 0.023689644411206245, "global_step": 12325, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03590100258588791, "train_loss_bc": 0.03590100258588791, "train_loss_llm": 0.0, "grad_norm": 0.0336824469268322, "global_step": 12326, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03738908842206001, "train_loss_bc": 0.03738908842206001, "train_loss_llm": 0.0, "grad_norm": 0.04413895308971405, "global_step": 12327, "epoch": 342, "lr": 0.001} +{"train_loss": 0.044309720396995544, "train_loss_bc": 0.044309720396995544, "train_loss_llm": 0.0, "grad_norm": 0.048601239919662476, "global_step": 12328, "epoch": 342, "lr": 0.001} +{"train_loss": 0.030127519741654396, "train_loss_bc": 0.030127519741654396, "train_loss_llm": 0.0, "grad_norm": 0.016243960708379745, "global_step": 12329, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03632303699851036, "train_loss_bc": 0.03631903603672981, "train_loss_llm": 0.00040012598037719727, "grad_norm": 0.02906615659594536, "global_step": 12330, "epoch": 342, "lr": 0.001} +{"train_loss": 0.0413171648979187, "train_loss_bc": 0.0413171648979187, "train_loss_llm": 0.0, "grad_norm": 0.0443008728325367, "global_step": 12331, "epoch": 342, "lr": 0.001} +{"train_loss": 0.040420882403850555, "train_loss_bc": 0.040420882403850555, "train_loss_llm": 0.0, "grad_norm": 0.04887552931904793, "global_step": 12332, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03197764232754707, "train_loss_bc": 0.03197764232754707, "train_loss_llm": 0.0, "grad_norm": 0.018681427463889122, "global_step": 12333, "epoch": 342, "lr": 0.001} +{"train_loss": 0.042588163167238235, "train_loss_bc": 0.042588163167238235, "train_loss_llm": 0.0, "grad_norm": 0.032018549740314484, "global_step": 12334, "epoch": 342, "lr": 0.001} +{"train_loss": 0.027981380000710487, "train_loss_bc": 0.027981380000710487, "train_loss_llm": 0.0, "grad_norm": 0.040413688868284225, "global_step": 12335, "epoch": 342, "lr": 0.001} +{"train_loss": 0.0363970510661602, "train_loss_bc": 0.03639066591858864, "train_loss_llm": 0.0006385445594787598, "grad_norm": 0.05278567597270012, "global_step": 12336, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03493407741189003, "train_loss_bc": 0.03493407741189003, "train_loss_llm": 0.0, "grad_norm": 0.026897508651018143, "global_step": 12337, "epoch": 342, "lr": 0.001} +{"train_loss": 0.038465466350317, "train_loss_bc": 0.038459960371255875, "train_loss_llm": 0.0005507171154022217, "grad_norm": 0.03860274329781532, "global_step": 12338, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03421127423644066, "train_loss_bc": 0.03421127423644066, "train_loss_llm": 0.0, "grad_norm": 0.050728581845760345, "global_step": 12339, "epoch": 342, "lr": 0.001} +{"train_loss": 0.037861570715904236, "train_loss_bc": 0.037861570715904236, "train_loss_llm": 0.0, "grad_norm": 0.05324476957321167, "global_step": 12340, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03966131433844566, "train_loss_bc": 0.03966131433844566, "train_loss_llm": 0.0, "grad_norm": 0.028845544904470444, "global_step": 12341, "epoch": 342, "lr": 0.001} +{"train_loss": 0.039167191833257675, "train_loss_bc": 0.039167191833257675, "train_loss_llm": 0.0, "grad_norm": 0.040750160813331604, "global_step": 12342, "epoch": 342, "lr": 0.001} +{"train_loss": 0.036797329783439636, "train_loss_bc": 0.036797329783439636, "train_loss_llm": 0.0, "grad_norm": 0.05274615436792374, "global_step": 12343, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03639670088887215, "train_loss_bc": 0.03639670088887215, "train_loss_llm": 0.0, "grad_norm": 0.06981558352708817, "global_step": 12344, "epoch": 342, "lr": 0.001} +{"train_loss": 0.038879603147506714, "train_loss_bc": 0.038860220462083817, "train_loss_llm": 0.0019381940364837646, "grad_norm": 0.021839266642928123, "global_step": 12345, "epoch": 342, "lr": 0.001} +{"train_loss": 0.04046246409416199, "train_loss_bc": 0.04044000431895256, "train_loss_llm": 0.0022461116313934326, "grad_norm": 0.029701801016926765, "global_step": 12346, "epoch": 342, "lr": 0.001} +{"train_loss": 0.03675902888385786, "train_loss_bc": 0.030616996809840202, "train_loss_llm": 0.0, "grad_norm": 0.04279586300253868, "global_step": 12347, "epoch": 342, "lr": 0.001, "val_loss": 0.06467388570308685} +{"train_loss": 0.03491638973355293, "train_loss_bc": 0.034915726631879807, "train_loss_llm": 6.625056266784668e-05, "grad_norm": 0.053386084735393524, "global_step": 12348, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03074725717306137, "train_loss_bc": 0.03074725717306137, "train_loss_llm": 0.0, "grad_norm": 0.020884761586785316, "global_step": 12349, "epoch": 343, "lr": 0.001} +{"train_loss": 0.035271938890218735, "train_loss_bc": 0.035271938890218735, "train_loss_llm": 0.0, "grad_norm": 0.03252587094902992, "global_step": 12350, "epoch": 343, "lr": 0.001} +{"train_loss": 0.04609069600701332, "train_loss_bc": 0.04609069600701332, "train_loss_llm": 0.0, "grad_norm": 0.05234436318278313, "global_step": 12351, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03324455767869949, "train_loss_bc": 0.03324455767869949, "train_loss_llm": 0.0, "grad_norm": 0.06542988866567612, "global_step": 12352, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03921959549188614, "train_loss_bc": 0.03921959549188614, "train_loss_llm": 0.0, "grad_norm": 0.021575400605797768, "global_step": 12353, "epoch": 343, "lr": 0.001} +{"train_loss": 0.038614556193351746, "train_loss_bc": 0.03859926760196686, "train_loss_llm": 0.0015287995338439941, "grad_norm": 0.03749573230743408, "global_step": 12354, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03637833893299103, "train_loss_bc": 0.03637833893299103, "train_loss_llm": 0.0, "grad_norm": 0.04324350878596306, "global_step": 12355, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03463786840438843, "train_loss_bc": 0.03463786840438843, "train_loss_llm": 0.0, "grad_norm": 0.0504305437207222, "global_step": 12356, "epoch": 343, "lr": 0.001} +{"train_loss": 0.031478121876716614, "train_loss_bc": 0.031478121876716614, "train_loss_llm": 0.0, "grad_norm": 0.017470935359597206, "global_step": 12357, "epoch": 343, "lr": 0.001} +{"train_loss": 0.04157997667789459, "train_loss_bc": 0.04157194495201111, "train_loss_llm": 0.0008032321929931641, "grad_norm": 0.030182039365172386, "global_step": 12358, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03448548540472984, "train_loss_bc": 0.03446951135993004, "train_loss_llm": 0.0015975236892700195, "grad_norm": 0.03836015984416008, "global_step": 12359, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03604232519865036, "train_loss_bc": 0.03604232519865036, "train_loss_llm": 0.0, "grad_norm": 0.05157805606722832, "global_step": 12360, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03554869815707207, "train_loss_bc": 0.03554869815707207, "train_loss_llm": 0.0, "grad_norm": 0.02048570103943348, "global_step": 12361, "epoch": 343, "lr": 0.001} +{"train_loss": 0.035384006798267365, "train_loss_bc": 0.035364724695682526, "train_loss_llm": 0.0019280612468719482, "grad_norm": 0.0235952939838171, "global_step": 12362, "epoch": 343, "lr": 0.001} +{"train_loss": 0.029725637286901474, "train_loss_bc": 0.029708411544561386, "train_loss_llm": 0.001722574234008789, "grad_norm": 0.029666321352124214, "global_step": 12363, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03630457818508148, "train_loss_bc": 0.03630457818508148, "train_loss_llm": 0.0, "grad_norm": 0.037360209971666336, "global_step": 12364, "epoch": 343, "lr": 0.001} +{"train_loss": 0.04033597186207771, "train_loss_bc": 0.04033597186207771, "train_loss_llm": 0.0, "grad_norm": 0.023979678750038147, "global_step": 12365, "epoch": 343, "lr": 0.001} +{"train_loss": 0.031059250235557556, "train_loss_bc": 0.031059250235557556, "train_loss_llm": 0.0, "grad_norm": 0.03461712971329689, "global_step": 12366, "epoch": 343, "lr": 0.001} +{"train_loss": 0.039186861366033554, "train_loss_bc": 0.03917564079165459, "train_loss_llm": 0.001122117042541504, "grad_norm": 0.03630572929978371, "global_step": 12367, "epoch": 343, "lr": 0.001} +{"train_loss": 0.0407848060131073, "train_loss_bc": 0.0407848060131073, "train_loss_llm": 0.0, "grad_norm": 0.043322522193193436, "global_step": 12368, "epoch": 343, "lr": 0.001} +{"train_loss": 0.04335068166255951, "train_loss_bc": 0.04335068166255951, "train_loss_llm": 0.0, "grad_norm": 0.0258509311825037, "global_step": 12369, "epoch": 343, "lr": 0.001} +{"train_loss": 0.033294253051280975, "train_loss_bc": 0.033294253051280975, "train_loss_llm": 0.0, "grad_norm": 0.03205449879169464, "global_step": 12370, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03191511332988739, "train_loss_bc": 0.03191511332988739, "train_loss_llm": 0.0, "grad_norm": 0.04509831219911575, "global_step": 12371, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03455732762813568, "train_loss_bc": 0.03455732762813568, "train_loss_llm": 0.0, "grad_norm": 0.05740032717585564, "global_step": 12372, "epoch": 343, "lr": 0.001} +{"train_loss": 0.033487897366285324, "train_loss_bc": 0.033487897366285324, "train_loss_llm": 0.0, "grad_norm": 0.022448023781180382, "global_step": 12373, "epoch": 343, "lr": 0.001} +{"train_loss": 0.04483398050069809, "train_loss_bc": 0.04483398050069809, "train_loss_llm": 0.0, "grad_norm": 0.03273557871580124, "global_step": 12374, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03689662739634514, "train_loss_bc": 0.03689662739634514, "train_loss_llm": 0.0, "grad_norm": 0.05356257036328316, "global_step": 12375, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03469729423522949, "train_loss_bc": 0.03469729423522949, "train_loss_llm": 0.0, "grad_norm": 0.07496944814920425, "global_step": 12376, "epoch": 343, "lr": 0.001} +{"train_loss": 0.04598681628704071, "train_loss_bc": 0.04598681628704071, "train_loss_llm": 0.0, "grad_norm": 0.02294427715241909, "global_step": 12377, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03603507578372955, "train_loss_bc": 0.03603507578372955, "train_loss_llm": 0.0, "grad_norm": 0.030189121142029762, "global_step": 12378, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03345773369073868, "train_loss_bc": 0.033432625234127045, "train_loss_llm": 0.002510666847229004, "grad_norm": 0.03673236817121506, "global_step": 12379, "epoch": 343, "lr": 0.001} +{"train_loss": 0.03729058429598808, "train_loss_bc": 0.03729058429598808, "train_loss_llm": 0.0, "grad_norm": 0.04516937956213951, "global_step": 12380, "epoch": 343, "lr": 0.001} +{"train_loss": 0.04274121671915054, "train_loss_bc": 0.04274121671915054, "train_loss_llm": 0.0, "grad_norm": 0.020904406905174255, "global_step": 12381, "epoch": 343, "lr": 0.001} +{"train_loss": 0.04018065705895424, "train_loss_bc": 0.040161408483982086, "train_loss_llm": 0.0019248723983764648, "grad_norm": 0.03215973451733589, "global_step": 12382, "epoch": 343, "lr": 0.001} +{"train_loss": 0.036846325215366155, "train_loss_bc": 0.03670553117990494, "train_loss_llm": 0.0, "grad_norm": 0.03980037942528725, "global_step": 12383, "epoch": 343, "lr": 0.001, "val_loss": 0.06437113881111145} +{"train_loss": 0.04172464460134506, "train_loss_bc": 0.04172464460134506, "train_loss_llm": 0.0, "grad_norm": 0.043848276138305664, "global_step": 12384, "epoch": 344, "lr": 0.001} +{"train_loss": 0.033740267157554626, "train_loss_bc": 0.03373466432094574, "train_loss_llm": 0.0005601048469543457, "grad_norm": 0.02380823902785778, "global_step": 12385, "epoch": 344, "lr": 0.001} +{"train_loss": 0.0373099185526371, "train_loss_bc": 0.0373099185526371, "train_loss_llm": 0.0, "grad_norm": 0.03085467405617237, "global_step": 12386, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03981601446866989, "train_loss_bc": 0.03980780392885208, "train_loss_llm": 0.0008211135864257812, "grad_norm": 0.047343235462903976, "global_step": 12387, "epoch": 344, "lr": 0.001} +{"train_loss": 0.0363336019217968, "train_loss_bc": 0.0363336019217968, "train_loss_llm": 0.0, "grad_norm": 0.050007980316877365, "global_step": 12388, "epoch": 344, "lr": 0.001} +{"train_loss": 0.043358612805604935, "train_loss_bc": 0.043358612805604935, "train_loss_llm": 0.0, "grad_norm": 0.027997709810733795, "global_step": 12389, "epoch": 344, "lr": 0.001} +{"train_loss": 0.035501137375831604, "train_loss_bc": 0.035501137375831604, "train_loss_llm": 0.0, "grad_norm": 0.04062250629067421, "global_step": 12390, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03573812544345856, "train_loss_bc": 0.03572302684187889, "train_loss_llm": 0.0015097558498382568, "grad_norm": 0.0549156479537487, "global_step": 12391, "epoch": 344, "lr": 0.001} +{"train_loss": 0.022564908489584923, "train_loss_bc": 0.022562477737665176, "train_loss_llm": 0.00024315714836120605, "grad_norm": 0.06679077446460724, "global_step": 12392, "epoch": 344, "lr": 0.001} +{"train_loss": 0.0425679087638855, "train_loss_bc": 0.0425679087638855, "train_loss_llm": 0.0, "grad_norm": 0.028336888179183006, "global_step": 12393, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03896786645054817, "train_loss_bc": 0.03896786645054817, "train_loss_llm": 0.0, "grad_norm": 0.03684186190366745, "global_step": 12394, "epoch": 344, "lr": 0.001} +{"train_loss": 0.035689085721969604, "train_loss_bc": 0.03568807616829872, "train_loss_llm": 0.0001011192798614502, "grad_norm": 0.039318863302469254, "global_step": 12395, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03291735053062439, "train_loss_bc": 0.03291735053062439, "train_loss_llm": 0.0, "grad_norm": 0.04676355421543121, "global_step": 12396, "epoch": 344, "lr": 0.001} +{"train_loss": 0.033624425530433655, "train_loss_bc": 0.033624425530433655, "train_loss_llm": 0.0, "grad_norm": 0.016118114814162254, "global_step": 12397, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03523055464029312, "train_loss_bc": 0.03523055464029312, "train_loss_llm": 0.0, "grad_norm": 0.02616680972278118, "global_step": 12398, "epoch": 344, "lr": 0.001} +{"train_loss": 0.035416509956121445, "train_loss_bc": 0.035416509956121445, "train_loss_llm": 0.0, "grad_norm": 0.03552109748125076, "global_step": 12399, "epoch": 344, "lr": 0.001} +{"train_loss": 0.039731111377477646, "train_loss_bc": 0.039731111377477646, "train_loss_llm": 0.0, "grad_norm": 0.04615287855267525, "global_step": 12400, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03581014648079872, "train_loss_bc": 0.03581014648079872, "train_loss_llm": 0.0, "grad_norm": 0.02173631079494953, "global_step": 12401, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03515952453017235, "train_loss_bc": 0.03515952453017235, "train_loss_llm": 0.0, "grad_norm": 0.03416439890861511, "global_step": 12402, "epoch": 344, "lr": 0.001} +{"train_loss": 0.0390513576567173, "train_loss_bc": 0.0390513576567173, "train_loss_llm": 0.0, "grad_norm": 0.04470748454332352, "global_step": 12403, "epoch": 344, "lr": 0.001} +{"train_loss": 0.04030371084809303, "train_loss_bc": 0.04030371084809303, "train_loss_llm": 0.0, "grad_norm": 0.055500857532024384, "global_step": 12404, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03667536750435829, "train_loss_bc": 0.03667536750435829, "train_loss_llm": 0.0, "grad_norm": 0.020089678466320038, "global_step": 12405, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03908463567495346, "train_loss_bc": 0.039081983268260956, "train_loss_llm": 0.0002652108669281006, "grad_norm": 0.035192880779504776, "global_step": 12406, "epoch": 344, "lr": 0.001} +{"train_loss": 0.035623714327812195, "train_loss_bc": 0.03562337905168533, "train_loss_llm": 3.36766242980957e-05, "grad_norm": 0.04367706924676895, "global_step": 12407, "epoch": 344, "lr": 0.001} +{"train_loss": 0.037116922438144684, "train_loss_bc": 0.037116922438144684, "train_loss_llm": 0.0, "grad_norm": 0.05259348079562187, "global_step": 12408, "epoch": 344, "lr": 0.001} +{"train_loss": 0.04631112515926361, "train_loss_bc": 0.04631112515926361, "train_loss_llm": 0.0, "grad_norm": 0.023659761995077133, "global_step": 12409, "epoch": 344, "lr": 0.001} +{"train_loss": 0.04143395274877548, "train_loss_bc": 0.0414116233587265, "train_loss_llm": 0.0022329092025756836, "grad_norm": 0.03146078810095787, "global_step": 12410, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03671664744615555, "train_loss_bc": 0.03671664744615555, "train_loss_llm": 0.0, "grad_norm": 0.03989745303988457, "global_step": 12411, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03521544858813286, "train_loss_bc": 0.035198330879211426, "train_loss_llm": 0.001711726188659668, "grad_norm": 0.04414527863264084, "global_step": 12412, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03652165085077286, "train_loss_bc": 0.03652165085077286, "train_loss_llm": 0.0, "grad_norm": 0.020523566752672195, "global_step": 12413, "epoch": 344, "lr": 0.001} +{"train_loss": 0.033491313457489014, "train_loss_bc": 0.033491313457489014, "train_loss_llm": 0.0, "grad_norm": 0.02578340470790863, "global_step": 12414, "epoch": 344, "lr": 0.001} +{"train_loss": 0.033323630690574646, "train_loss_bc": 0.03331664577126503, "train_loss_llm": 0.0006986558437347412, "grad_norm": 0.035927366465330124, "global_step": 12415, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03798101097345352, "train_loss_bc": 0.03798101097345352, "train_loss_llm": 0.0, "grad_norm": 0.04970017075538635, "global_step": 12416, "epoch": 344, "lr": 0.001} +{"train_loss": 0.02663695439696312, "train_loss_bc": 0.02663695439696312, "train_loss_llm": 0.0, "grad_norm": 0.018585098907351494, "global_step": 12417, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03959095850586891, "train_loss_bc": 0.03959095850586891, "train_loss_llm": 0.0, "grad_norm": 0.032731056213378906, "global_step": 12418, "epoch": 344, "lr": 0.001} +{"train_loss": 0.03686705816330181, "train_loss_bc": 0.04091827943921089, "train_loss_llm": 0.0015697181224822998, "grad_norm": 0.04904615506529808, "global_step": 12419, "epoch": 344, "lr": 0.001, "val_loss": 0.06347434222698212} +{"train_loss": 0.033216316252946854, "train_loss_bc": 0.033216316252946854, "train_loss_llm": 0.0, "grad_norm": 0.052984416484832764, "global_step": 12420, "epoch": 345, "lr": 0.001} +{"train_loss": 0.04358154907822609, "train_loss_bc": 0.043573617935180664, "train_loss_llm": 0.0007929503917694092, "grad_norm": 0.021713197231292725, "global_step": 12421, "epoch": 345, "lr": 0.001} +{"train_loss": 0.04010774940252304, "train_loss_bc": 0.040105730295181274, "train_loss_llm": 0.00020182132720947266, "grad_norm": 0.030755000188946724, "global_step": 12422, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03192462399601936, "train_loss_bc": 0.03192462399601936, "train_loss_llm": 0.0, "grad_norm": 0.04016190394759178, "global_step": 12423, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03414897248148918, "train_loss_bc": 0.03414897248148918, "train_loss_llm": 0.0, "grad_norm": 0.05096041411161423, "global_step": 12424, "epoch": 345, "lr": 0.001} +{"train_loss": 0.029062695801258087, "train_loss_bc": 0.029062695801258087, "train_loss_llm": 0.0, "grad_norm": 0.01776949129998684, "global_step": 12425, "epoch": 345, "lr": 0.001} +{"train_loss": 0.02843090519309044, "train_loss_bc": 0.02842264249920845, "train_loss_llm": 0.0008262693881988525, "grad_norm": 0.026466311886906624, "global_step": 12426, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03907380625605583, "train_loss_bc": 0.03907380625605583, "train_loss_llm": 0.0, "grad_norm": 0.037595804780721664, "global_step": 12427, "epoch": 345, "lr": 0.001} +{"train_loss": 0.038205333054065704, "train_loss_bc": 0.038188133388757706, "train_loss_llm": 0.0017201006412506104, "grad_norm": 0.047289103269577026, "global_step": 12428, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03718702122569084, "train_loss_bc": 0.03718702122569084, "train_loss_llm": 0.0, "grad_norm": 0.02156747505068779, "global_step": 12429, "epoch": 345, "lr": 0.001} +{"train_loss": 0.04662734270095825, "train_loss_bc": 0.04662099853157997, "train_loss_llm": 0.0006342828273773193, "grad_norm": 0.03146738559007645, "global_step": 12430, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03444475680589676, "train_loss_bc": 0.03444475680589676, "train_loss_llm": 0.0, "grad_norm": 0.041520148515701294, "global_step": 12431, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03514588624238968, "train_loss_bc": 0.03514588624238968, "train_loss_llm": 0.0, "grad_norm": 0.04818645492196083, "global_step": 12432, "epoch": 345, "lr": 0.001} +{"train_loss": 0.040162693709135056, "train_loss_bc": 0.040162693709135056, "train_loss_llm": 0.0, "grad_norm": 0.022337501868605614, "global_step": 12433, "epoch": 345, "lr": 0.001} +{"train_loss": 0.04195253923535347, "train_loss_bc": 0.04195253923535347, "train_loss_llm": 0.0, "grad_norm": 0.03580756485462189, "global_step": 12434, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03338070586323738, "train_loss_bc": 0.03336884453892708, "train_loss_llm": 0.0011861920356750488, "grad_norm": 0.042995210736989975, "global_step": 12435, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03610352799296379, "train_loss_bc": 0.03610352799296379, "train_loss_llm": 0.0, "grad_norm": 0.05083819851279259, "global_step": 12436, "epoch": 345, "lr": 0.001} +{"train_loss": 0.038114286959171295, "train_loss_bc": 0.038114286959171295, "train_loss_llm": 0.0, "grad_norm": 0.02047545090317726, "global_step": 12437, "epoch": 345, "lr": 0.001} +{"train_loss": 0.04085223749279976, "train_loss_bc": 0.0408424474298954, "train_loss_llm": 0.0009790658950805664, "grad_norm": 0.03166567534208298, "global_step": 12438, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03648877516388893, "train_loss_bc": 0.03648877516388893, "train_loss_llm": 0.0, "grad_norm": 0.04454291611909866, "global_step": 12439, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03804830089211464, "train_loss_bc": 0.03804830089211464, "train_loss_llm": 0.0, "grad_norm": 0.05182039365172386, "global_step": 12440, "epoch": 345, "lr": 0.001} +{"train_loss": 0.043441418558359146, "train_loss_bc": 0.043441418558359146, "train_loss_llm": 0.0, "grad_norm": 0.02381879650056362, "global_step": 12441, "epoch": 345, "lr": 0.001} +{"train_loss": 0.033573370426893234, "train_loss_bc": 0.033573370426893234, "train_loss_llm": 0.0, "grad_norm": 0.03402312844991684, "global_step": 12442, "epoch": 345, "lr": 0.001} +{"train_loss": 0.038594987243413925, "train_loss_bc": 0.0385906919836998, "train_loss_llm": 0.00042948126792907715, "grad_norm": 0.04398121312260628, "global_step": 12443, "epoch": 345, "lr": 0.001} +{"train_loss": 0.035653769969940186, "train_loss_bc": 0.035653769969940186, "train_loss_llm": 0.0, "grad_norm": 0.05081721767783165, "global_step": 12444, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03358228877186775, "train_loss_bc": 0.03358228877186775, "train_loss_llm": 0.0, "grad_norm": 0.026115648448467255, "global_step": 12445, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03866984695196152, "train_loss_bc": 0.03865795582532883, "train_loss_llm": 0.0011890828609466553, "grad_norm": 0.03500843420624733, "global_step": 12446, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03927622735500336, "train_loss_bc": 0.03927622735500336, "train_loss_llm": 0.0, "grad_norm": 0.04607081413269043, "global_step": 12447, "epoch": 345, "lr": 0.001} +{"train_loss": 0.035046130418777466, "train_loss_bc": 0.03503704071044922, "train_loss_llm": 0.0009087920188903809, "grad_norm": 0.05301786586642265, "global_step": 12448, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03570494055747986, "train_loss_bc": 0.03570494055747986, "train_loss_llm": 0.0, "grad_norm": 0.020772181451320648, "global_step": 12449, "epoch": 345, "lr": 0.001} +{"train_loss": 0.028257835656404495, "train_loss_bc": 0.02825060859322548, "train_loss_llm": 0.0007226169109344482, "grad_norm": 0.03831449896097183, "global_step": 12450, "epoch": 345, "lr": 0.001} +{"train_loss": 0.04443519562482834, "train_loss_bc": 0.04443519562482834, "train_loss_llm": 0.0, "grad_norm": 0.05378399416804314, "global_step": 12451, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03738929331302643, "train_loss_bc": 0.03735840693116188, "train_loss_llm": 0.0030885636806488037, "grad_norm": 0.0669936016201973, "global_step": 12452, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03474348038434982, "train_loss_bc": 0.03474348038434982, "train_loss_llm": 0.0, "grad_norm": 0.02380949631333351, "global_step": 12453, "epoch": 345, "lr": 0.001} +{"train_loss": 0.032912831753492355, "train_loss_bc": 0.032912831753492355, "train_loss_llm": 0.0, "grad_norm": 0.03763594105839729, "global_step": 12454, "epoch": 345, "lr": 0.001} +{"train_loss": 0.03684459088577165, "train_loss_bc": 0.03886362910270691, "train_loss_llm": 0.0, "grad_norm": 0.05086548626422882, "global_step": 12455, "epoch": 345, "lr": 0.001, "train/cumulative_reward": 95.60709892360269, "train/mean_score": 18.832987647919317, "train/success_rate": 0.5, "test/cumulative_reward": 35.720740439626766, "test/mean_score": 9.16210579296852, "test/success_rate": 0.22, "val_loss": 0.06514256447553635, "train_action_mse_error": 0.039470039308071136} +{"train_loss": 0.028509285300970078, "train_loss_bc": 0.028509285300970078, "train_loss_llm": 0.0, "grad_norm": 0.05693376436829567, "global_step": 12456, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03835714980959892, "train_loss_bc": 0.03835714980959892, "train_loss_llm": 0.0, "grad_norm": 0.02115786075592041, "global_step": 12457, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03711262717843056, "train_loss_bc": 0.037095896899700165, "train_loss_llm": 0.0016728639602661133, "grad_norm": 0.03506801649928093, "global_step": 12458, "epoch": 346, "lr": 0.001} +{"train_loss": 0.044438447803258896, "train_loss_bc": 0.044437531381845474, "train_loss_llm": 9.15229320526123e-05, "grad_norm": 0.04026925191283226, "global_step": 12459, "epoch": 346, "lr": 0.001} +{"train_loss": 0.033694084733724594, "train_loss_bc": 0.033694084733724594, "train_loss_llm": 0.0, "grad_norm": 0.0478486567735672, "global_step": 12460, "epoch": 346, "lr": 0.001} +{"train_loss": 0.038914769887924194, "train_loss_bc": 0.038914769887924194, "train_loss_llm": 0.0, "grad_norm": 0.02119440957903862, "global_step": 12461, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03744605556130409, "train_loss_bc": 0.03744440898299217, "train_loss_llm": 0.0001647472381591797, "grad_norm": 0.030335113406181335, "global_step": 12462, "epoch": 346, "lr": 0.001} +{"train_loss": 0.04198858514428139, "train_loss_bc": 0.04198858514428139, "train_loss_llm": 0.0, "grad_norm": 0.04877137765288353, "global_step": 12463, "epoch": 346, "lr": 0.001} +{"train_loss": 0.034528814256191254, "train_loss_bc": 0.034528814256191254, "train_loss_llm": 0.0, "grad_norm": 0.05614571273326874, "global_step": 12464, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03203655034303665, "train_loss_bc": 0.03203655034303665, "train_loss_llm": 0.0, "grad_norm": 0.02568751387298107, "global_step": 12465, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03710511326789856, "train_loss_bc": 0.03710360825061798, "train_loss_llm": 0.0001506805419921875, "grad_norm": 0.03522466868162155, "global_step": 12466, "epoch": 346, "lr": 0.001} +{"train_loss": 0.039297107607126236, "train_loss_bc": 0.03929450735449791, "train_loss_llm": 0.0002598762512207031, "grad_norm": 0.04637141153216362, "global_step": 12467, "epoch": 346, "lr": 0.001} +{"train_loss": 0.0316733755171299, "train_loss_bc": 0.031657516956329346, "train_loss_llm": 0.001585930585861206, "grad_norm": 0.06128842756152153, "global_step": 12468, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03808606043457985, "train_loss_bc": 0.038072437047958374, "train_loss_llm": 0.001362442970275879, "grad_norm": 0.019859373569488525, "global_step": 12469, "epoch": 346, "lr": 0.001} +{"train_loss": 0.026681028306484222, "train_loss_bc": 0.026681028306484222, "train_loss_llm": 0.0, "grad_norm": 0.02608872391283512, "global_step": 12470, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03689393773674965, "train_loss_bc": 0.03689393773674965, "train_loss_llm": 0.0, "grad_norm": 0.03804819658398628, "global_step": 12471, "epoch": 346, "lr": 0.001} +{"train_loss": 0.035381659865379333, "train_loss_bc": 0.03537879139184952, "train_loss_llm": 0.00028702616691589355, "grad_norm": 0.045719437301158905, "global_step": 12472, "epoch": 346, "lr": 0.001} +{"train_loss": 0.05151446908712387, "train_loss_bc": 0.05151446908712387, "train_loss_llm": 0.0, "grad_norm": 0.02696504257619381, "global_step": 12473, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03969863802194595, "train_loss_bc": 0.0396907813847065, "train_loss_llm": 0.0007854998111724854, "grad_norm": 0.04172468185424805, "global_step": 12474, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03823664039373398, "train_loss_bc": 0.03823664039373398, "train_loss_llm": 0.0, "grad_norm": 0.05691448971629143, "global_step": 12475, "epoch": 346, "lr": 0.001} +{"train_loss": 0.0371125154197216, "train_loss_bc": 0.0371125154197216, "train_loss_llm": 0.0, "grad_norm": 0.06483783572912216, "global_step": 12476, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03727208077907562, "train_loss_bc": 0.03727208077907562, "train_loss_llm": 0.0, "grad_norm": 0.01886269822716713, "global_step": 12477, "epoch": 346, "lr": 0.001} +{"train_loss": 0.038909897208213806, "train_loss_bc": 0.038909897208213806, "train_loss_llm": 0.0, "grad_norm": 0.03081723302602768, "global_step": 12478, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03456293046474457, "train_loss_bc": 0.03456293046474457, "train_loss_llm": 0.0, "grad_norm": 0.04453141242265701, "global_step": 12479, "epoch": 346, "lr": 0.001} +{"train_loss": 0.040148571133613586, "train_loss_bc": 0.040148571133613586, "train_loss_llm": 0.0, "grad_norm": 0.06072822958230972, "global_step": 12480, "epoch": 346, "lr": 0.001} +{"train_loss": 0.042793843895196915, "train_loss_bc": 0.042793843895196915, "train_loss_llm": 0.0, "grad_norm": 0.02769235149025917, "global_step": 12481, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03588379546999931, "train_loss_bc": 0.03588379546999931, "train_loss_llm": 0.0, "grad_norm": 0.04173123091459274, "global_step": 12482, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03145809471607208, "train_loss_bc": 0.03145809471607208, "train_loss_llm": 0.0, "grad_norm": 0.05194331333041191, "global_step": 12483, "epoch": 346, "lr": 0.001} +{"train_loss": 0.035707831382751465, "train_loss_bc": 0.035707831382751465, "train_loss_llm": 0.0, "grad_norm": 0.06353852897882462, "global_step": 12484, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03271923586726189, "train_loss_bc": 0.03271923586726189, "train_loss_llm": 0.0, "grad_norm": 0.020451514050364494, "global_step": 12485, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03681260719895363, "train_loss_bc": 0.03681260719895363, "train_loss_llm": 0.0, "grad_norm": 0.03375248238444328, "global_step": 12486, "epoch": 346, "lr": 0.001} +{"train_loss": 0.04128897935152054, "train_loss_bc": 0.04128897935152054, "train_loss_llm": 0.0, "grad_norm": 0.04977114498615265, "global_step": 12487, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03773508965969086, "train_loss_bc": 0.03773508965969086, "train_loss_llm": 0.0, "grad_norm": 0.05710649490356445, "global_step": 12488, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03397572040557861, "train_loss_bc": 0.03397068381309509, "train_loss_llm": 0.0005037784576416016, "grad_norm": 0.01905418187379837, "global_step": 12489, "epoch": 346, "lr": 0.001} +{"train_loss": 0.03242902457714081, "train_loss_bc": 0.03242902457714081, "train_loss_llm": 0.0, "grad_norm": 0.02947215363383293, "global_step": 12490, "epoch": 346, "lr": 0.001} +{"train_loss": 0.036940420460369855, "train_loss_bc": 0.0394367054104805, "train_loss_llm": 0.00138130784034729, "grad_norm": 0.04830918088555336, "global_step": 12491, "epoch": 346, "lr": 0.001, "val_loss": 0.06478435546159744} +{"train_loss": 0.03343084082007408, "train_loss_bc": 0.033412981778383255, "train_loss_llm": 0.00178605318069458, "grad_norm": 0.0628993809223175, "global_step": 12492, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03452146053314209, "train_loss_bc": 0.03452146053314209, "train_loss_llm": 0.0, "grad_norm": 0.021292071789503098, "global_step": 12493, "epoch": 347, "lr": 0.001} +{"train_loss": 0.037977177649736404, "train_loss_bc": 0.037977177649736404, "train_loss_llm": 0.0, "grad_norm": 0.0327078215777874, "global_step": 12494, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03723183274269104, "train_loss_bc": 0.03721178323030472, "train_loss_llm": 0.0020049214363098145, "grad_norm": 0.03735603392124176, "global_step": 12495, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03797052800655365, "train_loss_bc": 0.03797052800655365, "train_loss_llm": 0.0, "grad_norm": 0.0451151505112648, "global_step": 12496, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03576399385929108, "train_loss_bc": 0.035750728100538254, "train_loss_llm": 0.0013266503810882568, "grad_norm": 0.02285733073949814, "global_step": 12497, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03238609433174133, "train_loss_bc": 0.03238609433174133, "train_loss_llm": 0.0, "grad_norm": 0.02811519242823124, "global_step": 12498, "epoch": 347, "lr": 0.001} +{"train_loss": 0.035520389676094055, "train_loss_bc": 0.03550994396209717, "train_loss_llm": 0.0010445117950439453, "grad_norm": 0.031619541347026825, "global_step": 12499, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03779180347919464, "train_loss_bc": 0.03778960555791855, "train_loss_llm": 0.0002199709415435791, "grad_norm": 0.03605953976511955, "global_step": 12500, "epoch": 347, "lr": 0.001} +{"train_loss": 0.037021566182374954, "train_loss_bc": 0.037021566182374954, "train_loss_llm": 0.0, "grad_norm": 0.02680697664618492, "global_step": 12501, "epoch": 347, "lr": 0.001} +{"train_loss": 0.04049311950802803, "train_loss_bc": 0.040481336414813995, "train_loss_llm": 0.0011782348155975342, "grad_norm": 0.04859193414449692, "global_step": 12502, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03482881933450699, "train_loss_bc": 0.03482881933450699, "train_loss_llm": 0.0, "grad_norm": 0.062292277812957764, "global_step": 12503, "epoch": 347, "lr": 0.001} +{"train_loss": 0.04023532196879387, "train_loss_bc": 0.04023532196879387, "train_loss_llm": 0.0, "grad_norm": 0.07012446969747543, "global_step": 12504, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03802883252501488, "train_loss_bc": 0.03802883252501488, "train_loss_llm": 0.0, "grad_norm": 0.02243305929005146, "global_step": 12505, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03297839313745499, "train_loss_bc": 0.03297839313745499, "train_loss_llm": 0.0, "grad_norm": 0.03231201693415642, "global_step": 12506, "epoch": 347, "lr": 0.001} +{"train_loss": 0.039612844586372375, "train_loss_bc": 0.039612844586372375, "train_loss_llm": 0.0, "grad_norm": 0.0476391464471817, "global_step": 12507, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03674204275012016, "train_loss_bc": 0.03674204275012016, "train_loss_llm": 0.0, "grad_norm": 0.060386113822460175, "global_step": 12508, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03445514291524887, "train_loss_bc": 0.03445514291524887, "train_loss_llm": 0.0, "grad_norm": 0.021886654198169708, "global_step": 12509, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03904648870229721, "train_loss_bc": 0.03904648870229721, "train_loss_llm": 0.0, "grad_norm": 0.039323609322309494, "global_step": 12510, "epoch": 347, "lr": 0.001} +{"train_loss": 0.0309006217867136, "train_loss_bc": 0.0309006217867136, "train_loss_llm": 0.0, "grad_norm": 0.04812385141849518, "global_step": 12511, "epoch": 347, "lr": 0.001} +{"train_loss": 0.036911092698574066, "train_loss_bc": 0.036911092698574066, "train_loss_llm": 0.0, "grad_norm": 0.05981787294149399, "global_step": 12512, "epoch": 347, "lr": 0.001} +{"train_loss": 0.04014858230948448, "train_loss_bc": 0.04014858230948448, "train_loss_llm": 0.0, "grad_norm": 0.023559050634503365, "global_step": 12513, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03318240866065025, "train_loss_bc": 0.03318240866065025, "train_loss_llm": 0.0, "grad_norm": 0.03475941717624664, "global_step": 12514, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03709430247545242, "train_loss_bc": 0.03709430247545242, "train_loss_llm": 0.0, "grad_norm": 0.03921632096171379, "global_step": 12515, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03951408341526985, "train_loss_bc": 0.03951408341526985, "train_loss_llm": 0.0, "grad_norm": 0.05728583037853241, "global_step": 12516, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03770424798130989, "train_loss_bc": 0.03770424798130989, "train_loss_llm": 0.0, "grad_norm": 0.021214399486780167, "global_step": 12517, "epoch": 347, "lr": 0.001} +{"train_loss": 0.036408472806215286, "train_loss_bc": 0.036408472806215286, "train_loss_llm": 0.0, "grad_norm": 0.030398450791835785, "global_step": 12518, "epoch": 347, "lr": 0.001} +{"train_loss": 0.030868366360664368, "train_loss_bc": 0.030868366360664368, "train_loss_llm": 0.0, "grad_norm": 0.04124760627746582, "global_step": 12519, "epoch": 347, "lr": 0.001} +{"train_loss": 0.034216150641441345, "train_loss_bc": 0.034216150641441345, "train_loss_llm": 0.0, "grad_norm": 0.05169454962015152, "global_step": 12520, "epoch": 347, "lr": 0.001} +{"train_loss": 0.047089189291000366, "train_loss_bc": 0.047089189291000366, "train_loss_llm": 0.0, "grad_norm": 0.02203815057873726, "global_step": 12521, "epoch": 347, "lr": 0.001} +{"train_loss": 0.04021185263991356, "train_loss_bc": 0.04021185263991356, "train_loss_llm": 0.0, "grad_norm": 0.030453752726316452, "global_step": 12522, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03523581847548485, "train_loss_bc": 0.03523581847548485, "train_loss_llm": 0.0, "grad_norm": 0.03483296185731888, "global_step": 12523, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03352326899766922, "train_loss_bc": 0.03352326899766922, "train_loss_llm": 0.0, "grad_norm": 0.04855187609791756, "global_step": 12524, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03825678303837776, "train_loss_bc": 0.03824475780129433, "train_loss_llm": 0.0012024641036987305, "grad_norm": 0.025556281208992004, "global_step": 12525, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03827223181724548, "train_loss_bc": 0.03827223181724548, "train_loss_llm": 0.0, "grad_norm": 0.039525676518678665, "global_step": 12526, "epoch": 347, "lr": 0.001} +{"train_loss": 0.03682376335685452, "train_loss_bc": 0.040081314742565155, "train_loss_llm": 0.0, "grad_norm": 0.049929503351449966, "global_step": 12527, "epoch": 347, "lr": 0.001, "val_loss": 0.06385974586009979} +{"train_loss": 0.044246699661016464, "train_loss_bc": 0.044246699661016464, "train_loss_llm": 0.0, "grad_norm": 0.05906348675489426, "global_step": 12528, "epoch": 348, "lr": 0.001} +{"train_loss": 0.033046044409275055, "train_loss_bc": 0.033046044409275055, "train_loss_llm": 0.0, "grad_norm": 0.019345473498106003, "global_step": 12529, "epoch": 348, "lr": 0.001} +{"train_loss": 0.041041478514671326, "train_loss_bc": 0.041041478514671326, "train_loss_llm": 0.0, "grad_norm": 0.027376744896173477, "global_step": 12530, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03588752821087837, "train_loss_bc": 0.03588752821087837, "train_loss_llm": 0.0, "grad_norm": 0.0395171232521534, "global_step": 12531, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03768836706876755, "train_loss_bc": 0.03768836706876755, "train_loss_llm": 0.0, "grad_norm": 0.04886407032608986, "global_step": 12532, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03871535137295723, "train_loss_bc": 0.03869549557566643, "train_loss_llm": 0.001985520124435425, "grad_norm": 0.0250412467867136, "global_step": 12533, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03642217814922333, "train_loss_bc": 0.03642217814922333, "train_loss_llm": 0.0, "grad_norm": 0.036997586488723755, "global_step": 12534, "epoch": 348, "lr": 0.001} +{"train_loss": 0.027958080172538757, "train_loss_bc": 0.0279417484998703, "train_loss_llm": 0.0016331672668457031, "grad_norm": 0.047650180757045746, "global_step": 12535, "epoch": 348, "lr": 0.001} +{"train_loss": 0.039676543325185776, "train_loss_bc": 0.0396723598241806, "train_loss_llm": 0.0004183053970336914, "grad_norm": 0.07586779445409775, "global_step": 12536, "epoch": 348, "lr": 0.001} +{"train_loss": 0.035653240978717804, "train_loss_bc": 0.035653240978717804, "train_loss_llm": 0.0, "grad_norm": 0.027234701439738274, "global_step": 12537, "epoch": 348, "lr": 0.001} +{"train_loss": 0.0382184274494648, "train_loss_bc": 0.0382184274494648, "train_loss_llm": 0.0, "grad_norm": 0.03978671133518219, "global_step": 12538, "epoch": 348, "lr": 0.001} +{"train_loss": 0.037855952978134155, "train_loss_bc": 0.037855952978134155, "train_loss_llm": 0.0, "grad_norm": 0.051714979112148285, "global_step": 12539, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03527790680527687, "train_loss_bc": 0.03525526821613312, "train_loss_llm": 0.0022637546062469482, "grad_norm": 0.06212653964757919, "global_step": 12540, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03985823690891266, "train_loss_bc": 0.03985823690891266, "train_loss_llm": 0.0, "grad_norm": 0.022191528230905533, "global_step": 12541, "epoch": 348, "lr": 0.001} +{"train_loss": 0.02925175428390503, "train_loss_bc": 0.02925175428390503, "train_loss_llm": 0.0, "grad_norm": 0.02999381348490715, "global_step": 12542, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03654741495847702, "train_loss_bc": 0.03654741495847702, "train_loss_llm": 0.0, "grad_norm": 0.04163700342178345, "global_step": 12543, "epoch": 348, "lr": 0.001} +{"train_loss": 0.040055908262729645, "train_loss_bc": 0.040055908262729645, "train_loss_llm": 0.0, "grad_norm": 0.046776868402957916, "global_step": 12544, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03748423606157303, "train_loss_bc": 0.03748423606157303, "train_loss_llm": 0.0, "grad_norm": 0.024108022451400757, "global_step": 12545, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03808870166540146, "train_loss_bc": 0.03808870166540146, "train_loss_llm": 0.0, "grad_norm": 0.03109539858996868, "global_step": 12546, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03486606106162071, "train_loss_bc": 0.03486606106162071, "train_loss_llm": 0.0, "grad_norm": 0.03822784498333931, "global_step": 12547, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03861172869801521, "train_loss_bc": 0.03859800845384598, "train_loss_llm": 0.0013720393180847168, "grad_norm": 0.05397818610072136, "global_step": 12548, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03523048758506775, "train_loss_bc": 0.03523048758506775, "train_loss_llm": 0.0, "grad_norm": 0.0256639514118433, "global_step": 12549, "epoch": 348, "lr": 0.001} +{"train_loss": 0.032214801758527756, "train_loss_bc": 0.03220503032207489, "train_loss_llm": 0.0009772777557373047, "grad_norm": 0.03409775719046593, "global_step": 12550, "epoch": 348, "lr": 0.001} +{"train_loss": 0.04461253061890602, "train_loss_bc": 0.04461253061890602, "train_loss_llm": 0.0, "grad_norm": 0.0481640100479126, "global_step": 12551, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03463881462812424, "train_loss_bc": 0.03463881462812424, "train_loss_llm": 0.0, "grad_norm": 0.05994540452957153, "global_step": 12552, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03927122801542282, "train_loss_bc": 0.03927122801542282, "train_loss_llm": 0.0, "grad_norm": 0.024006998166441917, "global_step": 12553, "epoch": 348, "lr": 0.001} +{"train_loss": 0.04103458300232887, "train_loss_bc": 0.04103458300232887, "train_loss_llm": 0.0, "grad_norm": 0.038993142545223236, "global_step": 12554, "epoch": 348, "lr": 0.001} +{"train_loss": 0.039192043244838715, "train_loss_bc": 0.039192043244838715, "train_loss_llm": 0.0, "grad_norm": 0.04976505786180496, "global_step": 12555, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03749334067106247, "train_loss_bc": 0.03749334067106247, "train_loss_llm": 0.0, "grad_norm": 0.061912890523672104, "global_step": 12556, "epoch": 348, "lr": 0.001} +{"train_loss": 0.033624738454818726, "train_loss_bc": 0.033624738454818726, "train_loss_llm": 0.0, "grad_norm": 0.0238812118768692, "global_step": 12557, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03446293994784355, "train_loss_bc": 0.0344596728682518, "train_loss_llm": 0.0003267824649810791, "grad_norm": 0.0311738308519125, "global_step": 12558, "epoch": 348, "lr": 0.001} +{"train_loss": 0.036362964659929276, "train_loss_bc": 0.036362964659929276, "train_loss_llm": 0.0, "grad_norm": 0.041652992367744446, "global_step": 12559, "epoch": 348, "lr": 0.001} +{"train_loss": 0.0322243869304657, "train_loss_bc": 0.03222038969397545, "train_loss_llm": 0.00039967894554138184, "grad_norm": 0.05034903436899185, "global_step": 12560, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03587185963988304, "train_loss_bc": 0.0358678363263607, "train_loss_llm": 0.000402301549911499, "grad_norm": 0.01930087059736252, "global_step": 12561, "epoch": 348, "lr": 0.001} +{"train_loss": 0.03514466434717178, "train_loss_bc": 0.03514466434717178, "train_loss_llm": 0.0, "grad_norm": 0.028961585834622383, "global_step": 12562, "epoch": 348, "lr": 0.001} +{"train_loss": 0.036900336750679545, "train_loss_bc": 0.04058089852333069, "train_loss_llm": 0.0, "grad_norm": 0.04303791746497154, "global_step": 12563, "epoch": 348, "lr": 0.001, "val_loss": 0.06532365828752518} +{"train_loss": 0.03936849534511566, "train_loss_bc": 0.03936849534511566, "train_loss_llm": 0.0, "grad_norm": 0.048863496631383896, "global_step": 12564, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03468349575996399, "train_loss_bc": 0.03467883542180061, "train_loss_llm": 0.0004661083221435547, "grad_norm": 0.022681720554828644, "global_step": 12565, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03571697324514389, "train_loss_bc": 0.03571697324514389, "train_loss_llm": 0.0, "grad_norm": 0.030421441420912743, "global_step": 12566, "epoch": 349, "lr": 0.001} +{"train_loss": 0.04031335562467575, "train_loss_bc": 0.04031335562467575, "train_loss_llm": 0.0, "grad_norm": 0.03914444521069527, "global_step": 12567, "epoch": 349, "lr": 0.001} +{"train_loss": 0.037752002477645874, "train_loss_bc": 0.03773986175656319, "train_loss_llm": 0.0012141764163970947, "grad_norm": 0.04839106276631355, "global_step": 12568, "epoch": 349, "lr": 0.001} +{"train_loss": 0.033128462731838226, "train_loss_bc": 0.033128462731838226, "train_loss_llm": 0.0, "grad_norm": 0.019158829003572464, "global_step": 12569, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03978687897324562, "train_loss_bc": 0.0397687703371048, "train_loss_llm": 0.0018107295036315918, "grad_norm": 0.030234502628445625, "global_step": 12570, "epoch": 349, "lr": 0.001} +{"train_loss": 0.030520055443048477, "train_loss_bc": 0.030520055443048477, "train_loss_llm": 0.0, "grad_norm": 0.03720709681510925, "global_step": 12571, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03667868673801422, "train_loss_bc": 0.03665824979543686, "train_loss_llm": 0.002043604850769043, "grad_norm": 0.04790249466896057, "global_step": 12572, "epoch": 349, "lr": 0.001} +{"train_loss": 0.036335527896881104, "train_loss_bc": 0.03633176535367966, "train_loss_llm": 0.00037619471549987793, "grad_norm": 0.022245561704039574, "global_step": 12573, "epoch": 349, "lr": 0.001} +{"train_loss": 0.04168875515460968, "train_loss_bc": 0.04168875515460968, "train_loss_llm": 0.0, "grad_norm": 0.0448969341814518, "global_step": 12574, "epoch": 349, "lr": 0.001} +{"train_loss": 0.035143136978149414, "train_loss_bc": 0.035137273371219635, "train_loss_llm": 0.0005864799022674561, "grad_norm": 0.05742854252457619, "global_step": 12575, "epoch": 349, "lr": 0.001} +{"train_loss": 0.039631105959415436, "train_loss_bc": 0.039631105959415436, "train_loss_llm": 0.0, "grad_norm": 0.07435180991888046, "global_step": 12576, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03965624421834946, "train_loss_bc": 0.03965624421834946, "train_loss_llm": 0.0, "grad_norm": 0.02098521590232849, "global_step": 12577, "epoch": 349, "lr": 0.001} +{"train_loss": 0.0336931049823761, "train_loss_bc": 0.033692240715026855, "train_loss_llm": 8.64565372467041e-05, "grad_norm": 0.036968499422073364, "global_step": 12578, "epoch": 349, "lr": 0.001} +{"train_loss": 0.035392843186855316, "train_loss_bc": 0.035388410091400146, "train_loss_llm": 0.00044333934783935547, "grad_norm": 0.045066241174936295, "global_step": 12579, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03451578691601753, "train_loss_bc": 0.03451578691601753, "train_loss_llm": 0.0, "grad_norm": 0.057784706354141235, "global_step": 12580, "epoch": 349, "lr": 0.001} +{"train_loss": 0.037319574505090714, "train_loss_bc": 0.037319574505090714, "train_loss_llm": 0.0, "grad_norm": 0.021315498277544975, "global_step": 12581, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03869212418794632, "train_loss_bc": 0.038685426115989685, "train_loss_llm": 0.0006698966026306152, "grad_norm": 0.02936573140323162, "global_step": 12582, "epoch": 349, "lr": 0.001} +{"train_loss": 0.038596007972955704, "train_loss_bc": 0.03859451785683632, "train_loss_llm": 0.0001488327980041504, "grad_norm": 0.03880138322710991, "global_step": 12583, "epoch": 349, "lr": 0.001} +{"train_loss": 0.034215833991765976, "train_loss_bc": 0.034215833991765976, "train_loss_llm": 0.0, "grad_norm": 0.051122166216373444, "global_step": 12584, "epoch": 349, "lr": 0.001} +{"train_loss": 0.04593053087592125, "train_loss_bc": 0.04593053087592125, "train_loss_llm": 0.0, "grad_norm": 0.033100105822086334, "global_step": 12585, "epoch": 349, "lr": 0.001} +{"train_loss": 0.029446899890899658, "train_loss_bc": 0.029434233903884888, "train_loss_llm": 0.0012666285037994385, "grad_norm": 0.0430593341588974, "global_step": 12586, "epoch": 349, "lr": 0.001} +{"train_loss": 0.034718990325927734, "train_loss_bc": 0.03470471873879433, "train_loss_llm": 0.0014270246028900146, "grad_norm": 0.05273417755961418, "global_step": 12587, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03376439958810806, "train_loss_bc": 0.03375282883644104, "train_loss_llm": 0.0011571943759918213, "grad_norm": 0.06967516988515854, "global_step": 12588, "epoch": 349, "lr": 0.001} +{"train_loss": 0.04303906485438347, "train_loss_bc": 0.04303906485438347, "train_loss_llm": 0.0, "grad_norm": 0.021103795617818832, "global_step": 12589, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03210543841123581, "train_loss_bc": 0.03210543841123581, "train_loss_llm": 0.0, "grad_norm": 0.028834201395511627, "global_step": 12590, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03723224624991417, "train_loss_bc": 0.03723224624991417, "train_loss_llm": 0.0, "grad_norm": 0.039486635476350784, "global_step": 12591, "epoch": 349, "lr": 0.001} +{"train_loss": 0.04017060622572899, "train_loss_bc": 0.04017060622572899, "train_loss_llm": 0.0, "grad_norm": 0.04697832837700844, "global_step": 12592, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03472229465842247, "train_loss_bc": 0.03472229465842247, "train_loss_llm": 0.0, "grad_norm": 0.019526474177837372, "global_step": 12593, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03413112461566925, "train_loss_bc": 0.03413112461566925, "train_loss_llm": 0.0, "grad_norm": 0.0377492755651474, "global_step": 12594, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03703360632061958, "train_loss_bc": 0.03703360632061958, "train_loss_llm": 0.0, "grad_norm": 0.04921269416809082, "global_step": 12595, "epoch": 349, "lr": 0.001} +{"train_loss": 0.034738410264253616, "train_loss_bc": 0.034738410264253616, "train_loss_llm": 0.0, "grad_norm": 0.062310971319675446, "global_step": 12596, "epoch": 349, "lr": 0.001} +{"train_loss": 0.04194429889321327, "train_loss_bc": 0.04194429889321327, "train_loss_llm": 0.0, "grad_norm": 0.028215115889906883, "global_step": 12597, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03173966333270073, "train_loss_bc": 0.03172712400555611, "train_loss_llm": 0.0012538135051727295, "grad_norm": 0.04063909128308296, "global_step": 12598, "epoch": 349, "lr": 0.001} +{"train_loss": 0.03678591683920887, "train_loss_bc": 0.04074697941541672, "train_loss_llm": 0.0, "grad_norm": 0.04588974639773369, "global_step": 12599, "epoch": 349, "lr": 0.001, "val_loss": 0.06415325403213501} +{"train_loss": 0.03998995199799538, "train_loss_bc": 0.0399806946516037, "train_loss_llm": 0.0009256899356842041, "grad_norm": 0.05795939266681671, "global_step": 12600, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03164668753743172, "train_loss_bc": 0.031646039336919785, "train_loss_llm": 6.470084190368652e-05, "grad_norm": 0.017652414739131927, "global_step": 12601, "epoch": 350, "lr": 0.001} +{"train_loss": 0.035039544105529785, "train_loss_bc": 0.035039544105529785, "train_loss_llm": 0.0, "grad_norm": 0.0260942243039608, "global_step": 12602, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03373768553137779, "train_loss_bc": 0.03373768553137779, "train_loss_llm": 0.0, "grad_norm": 0.035567980259656906, "global_step": 12603, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03443034365773201, "train_loss_bc": 0.034410510212183, "train_loss_llm": 0.001983463764190674, "grad_norm": 0.04624518007040024, "global_step": 12604, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03914433345198631, "train_loss_bc": 0.03914433345198631, "train_loss_llm": 0.0, "grad_norm": 0.03077307529747486, "global_step": 12605, "epoch": 350, "lr": 0.001} +{"train_loss": 0.038246385753154755, "train_loss_bc": 0.038246385753154755, "train_loss_llm": 0.0, "grad_norm": 0.04157974198460579, "global_step": 12606, "epoch": 350, "lr": 0.001} +{"train_loss": 0.040673308074474335, "train_loss_bc": 0.04065833240747452, "train_loss_llm": 0.001497715711593628, "grad_norm": 0.0488964207470417, "global_step": 12607, "epoch": 350, "lr": 0.001} +{"train_loss": 0.0337807759642601, "train_loss_bc": 0.0337807759642601, "train_loss_llm": 0.0, "grad_norm": 0.05630340054631233, "global_step": 12608, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03807186707854271, "train_loss_bc": 0.03807186707854271, "train_loss_llm": 0.0, "grad_norm": 0.023100778460502625, "global_step": 12609, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03864484652876854, "train_loss_bc": 0.03864484652876854, "train_loss_llm": 0.0, "grad_norm": 0.0451006256043911, "global_step": 12610, "epoch": 350, "lr": 0.001} +{"train_loss": 0.041385941207408905, "train_loss_bc": 0.041385941207408905, "train_loss_llm": 0.0, "grad_norm": 0.05402308702468872, "global_step": 12611, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03464432805776596, "train_loss_bc": 0.03464432805776596, "train_loss_llm": 0.0, "grad_norm": 0.07975590974092484, "global_step": 12612, "epoch": 350, "lr": 0.001} +{"train_loss": 0.034078534692525864, "train_loss_bc": 0.034078534692525864, "train_loss_llm": 0.0, "grad_norm": 0.02312341518700123, "global_step": 12613, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03774573281407356, "train_loss_bc": 0.03774573281407356, "train_loss_llm": 0.0, "grad_norm": 0.036014530807733536, "global_step": 12614, "epoch": 350, "lr": 0.001} +{"train_loss": 0.040012575685977936, "train_loss_bc": 0.040012575685977936, "train_loss_llm": 0.0, "grad_norm": 0.05478869751095772, "global_step": 12615, "epoch": 350, "lr": 0.001} +{"train_loss": 0.04105639085173607, "train_loss_bc": 0.041042789816856384, "train_loss_llm": 0.0013599991798400879, "grad_norm": 0.061448246240615845, "global_step": 12616, "epoch": 350, "lr": 0.001} +{"train_loss": 0.04055272787809372, "train_loss_bc": 0.040543485432863235, "train_loss_llm": 0.0009242892265319824, "grad_norm": 0.025823939591646194, "global_step": 12617, "epoch": 350, "lr": 0.001} +{"train_loss": 0.035513948649168015, "train_loss_bc": 0.035513948649168015, "train_loss_llm": 0.0, "grad_norm": 0.03599366545677185, "global_step": 12618, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03937647119164467, "train_loss_bc": 0.03937434405088425, "train_loss_llm": 0.00021272897720336914, "grad_norm": 0.045022472739219666, "global_step": 12619, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03641000762581825, "train_loss_bc": 0.03640102595090866, "train_loss_llm": 0.0008980631828308105, "grad_norm": 0.05009278282523155, "global_step": 12620, "epoch": 350, "lr": 0.001} +{"train_loss": 0.0347096286714077, "train_loss_bc": 0.0347096286714077, "train_loss_llm": 0.0, "grad_norm": 0.019276263192296028, "global_step": 12621, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03424323722720146, "train_loss_bc": 0.03424323722720146, "train_loss_llm": 0.0, "grad_norm": 0.029787493869662285, "global_step": 12622, "epoch": 350, "lr": 0.001} +{"train_loss": 0.032158952206373215, "train_loss_bc": 0.032158952206373215, "train_loss_llm": 0.0, "grad_norm": 0.038290128111839294, "global_step": 12623, "epoch": 350, "lr": 0.001} +{"train_loss": 0.04349065572023392, "train_loss_bc": 0.04349065572023392, "train_loss_llm": 0.0, "grad_norm": 0.05852486193180084, "global_step": 12624, "epoch": 350, "lr": 0.001} +{"train_loss": 0.04075352102518082, "train_loss_bc": 0.04075168818235397, "train_loss_llm": 0.00018310546875, "grad_norm": 0.027666080743074417, "global_step": 12625, "epoch": 350, "lr": 0.001} +{"train_loss": 0.031069906428456306, "train_loss_bc": 0.031069906428456306, "train_loss_llm": 0.0, "grad_norm": 0.04106006398797035, "global_step": 12626, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03247436508536339, "train_loss_bc": 0.03247436508536339, "train_loss_llm": 0.0, "grad_norm": 0.05233670398592949, "global_step": 12627, "epoch": 350, "lr": 0.001} +{"train_loss": 0.0322682224214077, "train_loss_bc": 0.03226392716169357, "train_loss_llm": 0.00042951107025146484, "grad_norm": 0.06734556704759598, "global_step": 12628, "epoch": 350, "lr": 0.001} +{"train_loss": 0.04049002006649971, "train_loss_bc": 0.04049002006649971, "train_loss_llm": 0.0, "grad_norm": 0.01980435661971569, "global_step": 12629, "epoch": 350, "lr": 0.001} +{"train_loss": 0.037887509912252426, "train_loss_bc": 0.03786274790763855, "train_loss_llm": 0.0024763643741607666, "grad_norm": 0.03615481033921242, "global_step": 12630, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03141218423843384, "train_loss_bc": 0.03141218423843384, "train_loss_llm": 0.0, "grad_norm": 0.03800944611430168, "global_step": 12631, "epoch": 350, "lr": 0.001} +{"train_loss": 0.030638758093118668, "train_loss_bc": 0.030638758093118668, "train_loss_llm": 0.0, "grad_norm": 0.04513736069202423, "global_step": 12632, "epoch": 350, "lr": 0.001} +{"train_loss": 0.036667875945568085, "train_loss_bc": 0.036667875945568085, "train_loss_llm": 0.0, "grad_norm": 0.02551034837961197, "global_step": 12633, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03613543510437012, "train_loss_bc": 0.03613048046827316, "train_loss_llm": 0.0004954934120178223, "grad_norm": 0.035158153623342514, "global_step": 12634, "epoch": 350, "lr": 0.001} +{"train_loss": 0.03693573016466366, "train_loss_bc": 0.051082544028759, "train_loss_llm": 0.002108246088027954, "grad_norm": 0.04604269936680794, "global_step": 12635, "epoch": 350, "lr": 0.001, "train/cumulative_reward": 145.00127272947125, "train/mean_score": 13.825689263992519, "train/success_rate": 0.3, "test/cumulative_reward": 55.76369001626209, "test/mean_score": 9.570516105310826, "test/success_rate": 0.2, "val_loss": 0.06473103910684586, "train_action_mse_error": 0.039698194712400436} +{"train_loss": 0.032645124942064285, "train_loss_bc": 0.032645124942064285, "train_loss_llm": 0.0, "grad_norm": 0.056602418422698975, "global_step": 12636, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03467768058180809, "train_loss_bc": 0.034660179167985916, "train_loss_llm": 0.0017502009868621826, "grad_norm": 0.018908221274614334, "global_step": 12637, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03851887211203575, "train_loss_bc": 0.038512371480464935, "train_loss_llm": 0.0006498992443084717, "grad_norm": 0.027430914342403412, "global_step": 12638, "epoch": 351, "lr": 0.001} +{"train_loss": 0.0354413241147995, "train_loss_bc": 0.0354413241147995, "train_loss_llm": 0.0, "grad_norm": 0.043172530829906464, "global_step": 12639, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03444572165608406, "train_loss_bc": 0.03444572165608406, "train_loss_llm": 0.0, "grad_norm": 0.05162815377116203, "global_step": 12640, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03624698147177696, "train_loss_bc": 0.0362432487308979, "train_loss_llm": 0.0003733634948730469, "grad_norm": 0.018335288390517235, "global_step": 12641, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03418423980474472, "train_loss_bc": 0.03418423980474472, "train_loss_llm": 0.0, "grad_norm": 0.02656595967710018, "global_step": 12642, "epoch": 351, "lr": 0.001} +{"train_loss": 0.030700204893946648, "train_loss_bc": 0.03069533407688141, "train_loss_llm": 0.0004870593547821045, "grad_norm": 0.03569788113236427, "global_step": 12643, "epoch": 351, "lr": 0.001} +{"train_loss": 0.036456912755966187, "train_loss_bc": 0.036456912755966187, "train_loss_llm": 0.0, "grad_norm": 0.039907023310661316, "global_step": 12644, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03983773663640022, "train_loss_bc": 0.03982029855251312, "train_loss_llm": 0.0017437636852264404, "grad_norm": 0.026703128591179848, "global_step": 12645, "epoch": 351, "lr": 0.001} +{"train_loss": 0.042599763721227646, "train_loss_bc": 0.042599763721227646, "train_loss_llm": 0.0, "grad_norm": 0.035997845232486725, "global_step": 12646, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03503188490867615, "train_loss_bc": 0.03503188490867615, "train_loss_llm": 0.0, "grad_norm": 0.04343599081039429, "global_step": 12647, "epoch": 351, "lr": 0.001} +{"train_loss": 0.04783002659678459, "train_loss_bc": 0.04783002659678459, "train_loss_llm": 0.0, "grad_norm": 0.05904800444841385, "global_step": 12648, "epoch": 351, "lr": 0.001} +{"train_loss": 0.030288606882095337, "train_loss_bc": 0.030281003564596176, "train_loss_llm": 0.0007604062557220459, "grad_norm": 0.02059628628194332, "global_step": 12649, "epoch": 351, "lr": 0.001} +{"train_loss": 0.040225014090538025, "train_loss_bc": 0.040225014090538025, "train_loss_llm": 0.0, "grad_norm": 0.02960648015141487, "global_step": 12650, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03522104024887085, "train_loss_bc": 0.0352102667093277, "train_loss_llm": 0.0010773837566375732, "grad_norm": 0.04294730722904205, "global_step": 12651, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03668391332030296, "train_loss_bc": 0.03668391332030296, "train_loss_llm": 0.0, "grad_norm": 0.04773535951972008, "global_step": 12652, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03765464946627617, "train_loss_bc": 0.03765464946627617, "train_loss_llm": 0.0, "grad_norm": 0.023959433659911156, "global_step": 12653, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03722051903605461, "train_loss_bc": 0.03722051903605461, "train_loss_llm": 0.0, "grad_norm": 0.03610324114561081, "global_step": 12654, "epoch": 351, "lr": 0.001} +{"train_loss": 0.031881239265203476, "train_loss_bc": 0.03187316283583641, "train_loss_llm": 0.0008075237274169922, "grad_norm": 0.04472623020410538, "global_step": 12655, "epoch": 351, "lr": 0.001} +{"train_loss": 0.0366179496049881, "train_loss_bc": 0.0366179496049881, "train_loss_llm": 0.0, "grad_norm": 0.0631292536854744, "global_step": 12656, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03187340870499611, "train_loss_bc": 0.03187340870499611, "train_loss_llm": 0.0, "grad_norm": 0.024134116247296333, "global_step": 12657, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03723068907856941, "train_loss_bc": 0.03723068907856941, "train_loss_llm": 0.0, "grad_norm": 0.03333362564444542, "global_step": 12658, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03594464436173439, "train_loss_bc": 0.03593345731496811, "train_loss_llm": 0.0011185407638549805, "grad_norm": 0.03889913484454155, "global_step": 12659, "epoch": 351, "lr": 0.001} +{"train_loss": 0.04475705698132515, "train_loss_bc": 0.04475705698132515, "train_loss_llm": 0.0, "grad_norm": 0.05206979066133499, "global_step": 12660, "epoch": 351, "lr": 0.001} +{"train_loss": 0.041497621685266495, "train_loss_bc": 0.041497621685266495, "train_loss_llm": 0.0, "grad_norm": 0.02514549158513546, "global_step": 12661, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03414057567715645, "train_loss_bc": 0.03414057567715645, "train_loss_llm": 0.0, "grad_norm": 0.03306857496500015, "global_step": 12662, "epoch": 351, "lr": 0.001} +{"train_loss": 0.038756005465984344, "train_loss_bc": 0.038756005465984344, "train_loss_llm": 0.0, "grad_norm": 0.04288362339138985, "global_step": 12663, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03796164691448212, "train_loss_bc": 0.03796164691448212, "train_loss_llm": 0.0, "grad_norm": 0.05720878764986992, "global_step": 12664, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03672570735216141, "train_loss_bc": 0.03672570735216141, "train_loss_llm": 0.0, "grad_norm": 0.022740285843610764, "global_step": 12665, "epoch": 351, "lr": 0.001} +{"train_loss": 0.0277046300470829, "train_loss_bc": 0.0277046300470829, "train_loss_llm": 0.0, "grad_norm": 0.030758261680603027, "global_step": 12666, "epoch": 351, "lr": 0.001} +{"train_loss": 0.030732287093997, "train_loss_bc": 0.030727200210094452, "train_loss_llm": 0.0005086958408355713, "grad_norm": 0.03813783451914787, "global_step": 12667, "epoch": 351, "lr": 0.001} +{"train_loss": 0.03526616096496582, "train_loss_bc": 0.03526616096496582, "train_loss_llm": 0.0, "grad_norm": 0.044519100338220596, "global_step": 12668, "epoch": 351, "lr": 0.001} +{"train_loss": 0.04512713849544525, "train_loss_bc": 0.04512713849544525, "train_loss_llm": 0.0, "grad_norm": 0.0249045230448246, "global_step": 12669, "epoch": 351, "lr": 0.001} +{"train_loss": 0.04063791036605835, "train_loss_bc": 0.04063791036605835, "train_loss_llm": 0.0, "grad_norm": 0.03931061923503876, "global_step": 12670, "epoch": 351, "lr": 0.001} +{"train_loss": 0.0367825191674961, "train_loss_bc": 0.041371602565050125, "train_loss_llm": 0.0034199059009552, "grad_norm": 0.06178441643714905, "global_step": 12671, "epoch": 351, "lr": 0.001, "val_loss": 0.06390565633773804} +{"train_loss": 0.03851008415222168, "train_loss_bc": 0.03850158676505089, "train_loss_llm": 0.0008496344089508057, "grad_norm": 0.06353320181369781, "global_step": 12672, "epoch": 352, "lr": 0.001} +{"train_loss": 0.033441077917814255, "train_loss_bc": 0.033441077917814255, "train_loss_llm": 0.0, "grad_norm": 0.020635321736335754, "global_step": 12673, "epoch": 352, "lr": 0.001} +{"train_loss": 0.04143775627017021, "train_loss_bc": 0.04143775627017021, "train_loss_llm": 0.0, "grad_norm": 0.03566825017333031, "global_step": 12674, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03279978036880493, "train_loss_bc": 0.03277626261115074, "train_loss_llm": 0.002351701259613037, "grad_norm": 0.04074300453066826, "global_step": 12675, "epoch": 352, "lr": 0.001} +{"train_loss": 0.033412981778383255, "train_loss_bc": 0.033412981778383255, "train_loss_llm": 0.0, "grad_norm": 0.06016644090414047, "global_step": 12676, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03854427486658096, "train_loss_bc": 0.03854427486658096, "train_loss_llm": 0.0, "grad_norm": 0.027306905016303062, "global_step": 12677, "epoch": 352, "lr": 0.001} +{"train_loss": 0.04098491743206978, "train_loss_bc": 0.04098491743206978, "train_loss_llm": 0.0, "grad_norm": 0.04733414575457573, "global_step": 12678, "epoch": 352, "lr": 0.001} +{"train_loss": 0.033870894461870193, "train_loss_bc": 0.033870894461870193, "train_loss_llm": 0.0, "grad_norm": 0.057131700217723846, "global_step": 12679, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03437797725200653, "train_loss_bc": 0.034367866814136505, "train_loss_llm": 0.0010109245777130127, "grad_norm": 0.06781575083732605, "global_step": 12680, "epoch": 352, "lr": 0.001} +{"train_loss": 0.04195212572813034, "train_loss_bc": 0.041944291442632675, "train_loss_llm": 0.0007834434509277344, "grad_norm": 0.020750122144818306, "global_step": 12681, "epoch": 352, "lr": 0.001} +{"train_loss": 0.033236291259527206, "train_loss_bc": 0.033236291259527206, "train_loss_llm": 0.0, "grad_norm": 0.02905368246138096, "global_step": 12682, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03895462304353714, "train_loss_bc": 0.038922395557165146, "train_loss_llm": 0.0032228827476501465, "grad_norm": 0.043752606958150864, "global_step": 12683, "epoch": 352, "lr": 0.001} +{"train_loss": 0.036151450127363205, "train_loss_bc": 0.036151450127363205, "train_loss_llm": 0.0, "grad_norm": 0.06188134476542473, "global_step": 12684, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03199802339076996, "train_loss_bc": 0.03198765590786934, "train_loss_llm": 0.0010367035865783691, "grad_norm": 0.02052011527121067, "global_step": 12685, "epoch": 352, "lr": 0.001} +{"train_loss": 0.030468108132481575, "train_loss_bc": 0.030468108132481575, "train_loss_llm": 0.0, "grad_norm": 0.03367749601602554, "global_step": 12686, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03530528396368027, "train_loss_bc": 0.035299379378557205, "train_loss_llm": 0.0005904734134674072, "grad_norm": 0.04815366491675377, "global_step": 12687, "epoch": 352, "lr": 0.001} +{"train_loss": 0.0404236875474453, "train_loss_bc": 0.0404236875474453, "train_loss_llm": 0.0, "grad_norm": 0.06338837742805481, "global_step": 12688, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03801020234823227, "train_loss_bc": 0.03801020234823227, "train_loss_llm": 0.0, "grad_norm": 0.01952989771962166, "global_step": 12689, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03273870423436165, "train_loss_bc": 0.03273870423436165, "train_loss_llm": 0.0, "grad_norm": 0.030816838145256042, "global_step": 12690, "epoch": 352, "lr": 0.001} +{"train_loss": 0.042008936405181885, "train_loss_bc": 0.042008936405181885, "train_loss_llm": 0.0, "grad_norm": 0.03997168317437172, "global_step": 12691, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03928349167108536, "train_loss_bc": 0.03928349167108536, "train_loss_llm": 0.0, "grad_norm": 0.04700198397040367, "global_step": 12692, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03858986869454384, "train_loss_bc": 0.03858986869454384, "train_loss_llm": 0.0, "grad_norm": 0.020893633365631104, "global_step": 12693, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03894718363881111, "train_loss_bc": 0.03894718363881111, "train_loss_llm": 0.0, "grad_norm": 0.03223711997270584, "global_step": 12694, "epoch": 352, "lr": 0.001} +{"train_loss": 0.037659212946891785, "train_loss_bc": 0.037653740495443344, "train_loss_llm": 0.0005471408367156982, "grad_norm": 0.04844636097550392, "global_step": 12695, "epoch": 352, "lr": 0.001} +{"train_loss": 0.0409889817237854, "train_loss_bc": 0.04097571596503258, "train_loss_llm": 0.00132673978805542, "grad_norm": 0.06535828113555908, "global_step": 12696, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03794488683342934, "train_loss_bc": 0.03793569654226303, "train_loss_llm": 0.0009190738201141357, "grad_norm": 0.021053433418273926, "global_step": 12697, "epoch": 352, "lr": 0.001} +{"train_loss": 0.038245849311351776, "train_loss_bc": 0.038245849311351776, "train_loss_llm": 0.0, "grad_norm": 0.0323016531765461, "global_step": 12698, "epoch": 352, "lr": 0.001} +{"train_loss": 0.02824532613158226, "train_loss_bc": 0.02824532613158226, "train_loss_llm": 0.0, "grad_norm": 0.03793015331029892, "global_step": 12699, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03767465427517891, "train_loss_bc": 0.03767465427517891, "train_loss_llm": 0.0, "grad_norm": 0.045601293444633484, "global_step": 12700, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03251440450549126, "train_loss_bc": 0.03251440450549126, "train_loss_llm": 0.0, "grad_norm": 0.02265242300927639, "global_step": 12701, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03838428854942322, "train_loss_bc": 0.03838428854942322, "train_loss_llm": 0.0, "grad_norm": 0.04192746803164482, "global_step": 12702, "epoch": 352, "lr": 0.001} +{"train_loss": 0.04234365001320839, "train_loss_bc": 0.04234365001320839, "train_loss_llm": 0.0, "grad_norm": 0.06014968454837799, "global_step": 12703, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03647235408425331, "train_loss_bc": 0.0364711619913578, "train_loss_llm": 0.00011932849884033203, "grad_norm": 0.08112068474292755, "global_step": 12704, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03634602203965187, "train_loss_bc": 0.03634602203965187, "train_loss_llm": 0.0, "grad_norm": 0.030970074236392975, "global_step": 12705, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03702186048030853, "train_loss_bc": 0.03700416907668114, "train_loss_llm": 0.0017691850662231445, "grad_norm": 0.04647289961576462, "global_step": 12706, "epoch": 352, "lr": 0.001} +{"train_loss": 0.03698641387745738, "train_loss_bc": 0.04220576211810112, "train_loss_llm": 0.0015920400619506836, "grad_norm": 0.06368803977966309, "global_step": 12707, "epoch": 352, "lr": 0.001, "val_loss": 0.06629825383424759} +{"train_loss": 0.03688129782676697, "train_loss_bc": 0.03688129782676697, "train_loss_llm": 0.0, "grad_norm": 0.0702466145157814, "global_step": 12708, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03996068239212036, "train_loss_bc": 0.03995329886674881, "train_loss_llm": 0.0007382035255432129, "grad_norm": 0.021919935941696167, "global_step": 12709, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03453197702765465, "train_loss_bc": 0.03453197702765465, "train_loss_llm": 0.0, "grad_norm": 0.02885841578245163, "global_step": 12710, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03154116868972778, "train_loss_bc": 0.03154116868972778, "train_loss_llm": 0.0, "grad_norm": 0.037645675241947174, "global_step": 12711, "epoch": 353, "lr": 0.001} +{"train_loss": 0.030518269166350365, "train_loss_bc": 0.030507860705256462, "train_loss_llm": 0.0010408759117126465, "grad_norm": 0.04513878747820854, "global_step": 12712, "epoch": 353, "lr": 0.001} +{"train_loss": 0.02810649946331978, "train_loss_bc": 0.02810649946331978, "train_loss_llm": 0.0, "grad_norm": 0.017815740779042244, "global_step": 12713, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03328986465930939, "train_loss_bc": 0.033287737518548965, "train_loss_llm": 0.00021281838417053223, "grad_norm": 0.035846516489982605, "global_step": 12714, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03805200755596161, "train_loss_bc": 0.03805200755596161, "train_loss_llm": 0.0, "grad_norm": 0.05028405413031578, "global_step": 12715, "epoch": 353, "lr": 0.001} +{"train_loss": 0.04059990495443344, "train_loss_bc": 0.04059990495443344, "train_loss_llm": 0.0, "grad_norm": 0.05588417127728462, "global_step": 12716, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03244078531861305, "train_loss_bc": 0.03241404891014099, "train_loss_llm": 0.002673596143722534, "grad_norm": 0.027072424069046974, "global_step": 12717, "epoch": 353, "lr": 0.001} +{"train_loss": 0.039219051599502563, "train_loss_bc": 0.03921523690223694, "train_loss_llm": 0.0003815889358520508, "grad_norm": 0.04689420014619827, "global_step": 12718, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03811246529221535, "train_loss_bc": 0.03810583055019379, "train_loss_llm": 0.0006633400917053223, "grad_norm": 0.06172600015997887, "global_step": 12719, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03332727029919624, "train_loss_bc": 0.03330901265144348, "train_loss_llm": 0.0018257200717926025, "grad_norm": 0.06770262867212296, "global_step": 12720, "epoch": 353, "lr": 0.001} +{"train_loss": 0.037589121609926224, "train_loss_bc": 0.03757635876536369, "train_loss_llm": 0.0012764036655426025, "grad_norm": 0.02832062356173992, "global_step": 12721, "epoch": 353, "lr": 0.001} +{"train_loss": 0.0338851660490036, "train_loss_bc": 0.0338851660490036, "train_loss_llm": 0.0, "grad_norm": 0.04405224323272705, "global_step": 12722, "epoch": 353, "lr": 0.001} +{"train_loss": 0.034206222742795944, "train_loss_bc": 0.034206222742795944, "train_loss_llm": 0.0, "grad_norm": 0.055273860692977905, "global_step": 12723, "epoch": 353, "lr": 0.001} +{"train_loss": 0.036283623427152634, "train_loss_bc": 0.036283623427152634, "train_loss_llm": 0.0, "grad_norm": 0.07378821074962616, "global_step": 12724, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03171370178461075, "train_loss_bc": 0.03170483931899071, "train_loss_llm": 0.0008862614631652832, "grad_norm": 0.02859354391694069, "global_step": 12725, "epoch": 353, "lr": 0.001} +{"train_loss": 0.038664642721414566, "train_loss_bc": 0.03863532841205597, "train_loss_llm": 0.0029313862323760986, "grad_norm": 0.05123911425471306, "global_step": 12726, "epoch": 353, "lr": 0.001} +{"train_loss": 0.04051623493432999, "train_loss_bc": 0.04051623493432999, "train_loss_llm": 0.0, "grad_norm": 0.061311788856983185, "global_step": 12727, "epoch": 353, "lr": 0.001} +{"train_loss": 0.04570763185620308, "train_loss_bc": 0.04570763185620308, "train_loss_llm": 0.0, "grad_norm": 0.07805879414081573, "global_step": 12728, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03509046137332916, "train_loss_bc": 0.035090427845716476, "train_loss_llm": 3.3974647521972656e-06, "grad_norm": 0.018177036195993423, "global_step": 12729, "epoch": 353, "lr": 0.001} +{"train_loss": 0.04100218415260315, "train_loss_bc": 0.04100218415260315, "train_loss_llm": 0.0, "grad_norm": 0.03068043291568756, "global_step": 12730, "epoch": 353, "lr": 0.001} +{"train_loss": 0.04055557772517204, "train_loss_bc": 0.04055557772517204, "train_loss_llm": 0.0, "grad_norm": 0.04308988153934479, "global_step": 12731, "epoch": 353, "lr": 0.001} +{"train_loss": 0.035775136202573776, "train_loss_bc": 0.035775136202573776, "train_loss_llm": 0.0, "grad_norm": 0.05539282411336899, "global_step": 12732, "epoch": 353, "lr": 0.001} +{"train_loss": 0.029641345143318176, "train_loss_bc": 0.02963925153017044, "train_loss_llm": 0.00020945072174072266, "grad_norm": 0.023768914863467216, "global_step": 12733, "epoch": 353, "lr": 0.001} +{"train_loss": 0.0367107018828392, "train_loss_bc": 0.0367107018828392, "train_loss_llm": 0.0, "grad_norm": 0.04609634727239609, "global_step": 12734, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03338605538010597, "train_loss_bc": 0.033383578062057495, "train_loss_llm": 0.0002478361129760742, "grad_norm": 0.05844045430421829, "global_step": 12735, "epoch": 353, "lr": 0.001} +{"train_loss": 0.04149641469120979, "train_loss_bc": 0.04149641469120979, "train_loss_llm": 0.0, "grad_norm": 0.0795125663280487, "global_step": 12736, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03853931650519371, "train_loss_bc": 0.03853749483823776, "train_loss_llm": 0.00018221139907836914, "grad_norm": 0.02303418703377247, "global_step": 12737, "epoch": 353, "lr": 0.001} +{"train_loss": 0.029465196654200554, "train_loss_bc": 0.029465196654200554, "train_loss_llm": 0.0, "grad_norm": 0.029626591131091118, "global_step": 12738, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03967355191707611, "train_loss_bc": 0.03964873030781746, "train_loss_llm": 0.0024823248386383057, "grad_norm": 0.04319320246577263, "global_step": 12739, "epoch": 353, "lr": 0.001} +{"train_loss": 0.046333909034729004, "train_loss_bc": 0.046333909034729004, "train_loss_llm": 0.0, "grad_norm": 0.05353947728872299, "global_step": 12740, "epoch": 353, "lr": 0.001} +{"train_loss": 0.04217607527971268, "train_loss_bc": 0.04217607527971268, "train_loss_llm": 0.0, "grad_norm": 0.023250408470630646, "global_step": 12741, "epoch": 353, "lr": 0.001} +{"train_loss": 0.04463740065693855, "train_loss_bc": 0.04463440924882889, "train_loss_llm": 0.00029900670051574707, "grad_norm": 0.0412820428609848, "global_step": 12742, "epoch": 353, "lr": 0.001} +{"train_loss": 0.03696360294189718, "train_loss_bc": 0.04105878993868828, "train_loss_llm": 0.0, "grad_norm": 0.061530232429504395, "global_step": 12743, "epoch": 353, "lr": 0.001, "val_loss": 0.06577970832586288} +{"train_loss": 0.03335139900445938, "train_loss_bc": 0.03335139900445938, "train_loss_llm": 0.0, "grad_norm": 0.07357969880104065, "global_step": 12744, "epoch": 354, "lr": 0.001} +{"train_loss": 0.04068920761346817, "train_loss_bc": 0.04068920761346817, "train_loss_llm": 0.0, "grad_norm": 0.026632806286215782, "global_step": 12745, "epoch": 354, "lr": 0.001} +{"train_loss": 0.043922774493694305, "train_loss_bc": 0.043922774493694305, "train_loss_llm": 0.0, "grad_norm": 0.04740919917821884, "global_step": 12746, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03714735433459282, "train_loss_bc": 0.03714735433459282, "train_loss_llm": 0.0, "grad_norm": 0.06525898724794388, "global_step": 12747, "epoch": 354, "lr": 0.001} +{"train_loss": 0.037099581211805344, "train_loss_bc": 0.03709927573800087, "train_loss_llm": 3.0487775802612305e-05, "grad_norm": 0.07877247780561447, "global_step": 12748, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03795560076832771, "train_loss_bc": 0.03795560076832771, "train_loss_llm": 0.0, "grad_norm": 0.028009982779622078, "global_step": 12749, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03681764379143715, "train_loss_bc": 0.03680508956313133, "train_loss_llm": 0.001255333423614502, "grad_norm": 0.04680268466472626, "global_step": 12750, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03433826193213463, "train_loss_bc": 0.03433811664581299, "train_loss_llm": 1.4483928680419922e-05, "grad_norm": 0.056986019015312195, "global_step": 12751, "epoch": 354, "lr": 0.001} +{"train_loss": 0.035704176872968674, "train_loss_bc": 0.035704176872968674, "train_loss_llm": 0.0, "grad_norm": 0.06681883335113525, "global_step": 12752, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03629429638385773, "train_loss_bc": 0.03629075735807419, "train_loss_llm": 0.0003540515899658203, "grad_norm": 0.023191599175333977, "global_step": 12753, "epoch": 354, "lr": 0.001} +{"train_loss": 0.034458670765161514, "train_loss_bc": 0.03445588797330856, "train_loss_llm": 0.00027823448181152344, "grad_norm": 0.03604287654161453, "global_step": 12754, "epoch": 354, "lr": 0.001} +{"train_loss": 0.031575243920087814, "train_loss_bc": 0.031575243920087814, "train_loss_llm": 0.0, "grad_norm": 0.043911583721637726, "global_step": 12755, "epoch": 354, "lr": 0.001} +{"train_loss": 0.028208384290337563, "train_loss_bc": 0.028208384290337563, "train_loss_llm": 0.0, "grad_norm": 0.05161396041512489, "global_step": 12756, "epoch": 354, "lr": 0.001} +{"train_loss": 0.038921698927879333, "train_loss_bc": 0.038921698927879333, "train_loss_llm": 0.0, "grad_norm": 0.03600706905126572, "global_step": 12757, "epoch": 354, "lr": 0.001} +{"train_loss": 0.0345589816570282, "train_loss_bc": 0.0345589816570282, "train_loss_llm": 0.0, "grad_norm": 0.05332782864570618, "global_step": 12758, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03532426059246063, "train_loss_bc": 0.03532426059246063, "train_loss_llm": 0.0, "grad_norm": 0.0676712617278099, "global_step": 12759, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03934668004512787, "train_loss_bc": 0.03934388607740402, "train_loss_llm": 0.0002792477607727051, "grad_norm": 0.08394715189933777, "global_step": 12760, "epoch": 354, "lr": 0.001} +{"train_loss": 0.036532118916511536, "train_loss_bc": 0.036532118916511536, "train_loss_llm": 0.0, "grad_norm": 0.025271452963352203, "global_step": 12761, "epoch": 354, "lr": 0.001} +{"train_loss": 0.037191882729530334, "train_loss_bc": 0.037191882729530334, "train_loss_llm": 0.0, "grad_norm": 0.03414624556899071, "global_step": 12762, "epoch": 354, "lr": 0.001} +{"train_loss": 0.041739705950021744, "train_loss_bc": 0.041739705950021744, "train_loss_llm": 0.0, "grad_norm": 0.04894647374749184, "global_step": 12763, "epoch": 354, "lr": 0.001} +{"train_loss": 0.032813217490911484, "train_loss_bc": 0.032813217490911484, "train_loss_llm": 0.0, "grad_norm": 0.06381475925445557, "global_step": 12764, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03862748295068741, "train_loss_bc": 0.03860282152891159, "train_loss_llm": 0.0024662017822265625, "grad_norm": 0.024155067279934883, "global_step": 12765, "epoch": 354, "lr": 0.001} +{"train_loss": 0.04192544147372246, "train_loss_bc": 0.04192544147372246, "train_loss_llm": 0.0, "grad_norm": 0.037712424993515015, "global_step": 12766, "epoch": 354, "lr": 0.001} +{"train_loss": 0.041392065584659576, "train_loss_bc": 0.041392065584659576, "train_loss_llm": 0.0, "grad_norm": 0.057458922266960144, "global_step": 12767, "epoch": 354, "lr": 0.001} +{"train_loss": 0.035974305123090744, "train_loss_bc": 0.035957761108875275, "train_loss_llm": 0.001654505729675293, "grad_norm": 0.07280313968658447, "global_step": 12768, "epoch": 354, "lr": 0.001} +{"train_loss": 0.04039626568555832, "train_loss_bc": 0.04039626568555832, "train_loss_llm": 0.0, "grad_norm": 0.024135416373610497, "global_step": 12769, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03658639267086983, "train_loss_bc": 0.03658639267086983, "train_loss_llm": 0.0, "grad_norm": 0.0358845517039299, "global_step": 12770, "epoch": 354, "lr": 0.001} +{"train_loss": 0.04016055166721344, "train_loss_bc": 0.04014419764280319, "train_loss_llm": 0.0016353726387023926, "grad_norm": 0.054422777146101, "global_step": 12771, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03630821406841278, "train_loss_bc": 0.03630821406841278, "train_loss_llm": 0.0, "grad_norm": 0.06324872374534607, "global_step": 12772, "epoch": 354, "lr": 0.001} +{"train_loss": 0.039431460201740265, "train_loss_bc": 0.039431460201740265, "train_loss_llm": 0.0, "grad_norm": 0.032367043197155, "global_step": 12773, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03902294486761093, "train_loss_bc": 0.03900960460305214, "train_loss_llm": 0.0013339817523956299, "grad_norm": 0.04393688961863518, "global_step": 12774, "epoch": 354, "lr": 0.001} +{"train_loss": 0.029970616102218628, "train_loss_bc": 0.029967905953526497, "train_loss_llm": 0.00027102231979370117, "grad_norm": 0.05490150302648544, "global_step": 12775, "epoch": 354, "lr": 0.001} +{"train_loss": 0.03786175698041916, "train_loss_bc": 0.03786175698041916, "train_loss_llm": 0.0, "grad_norm": 0.07210854440927505, "global_step": 12776, "epoch": 354, "lr": 0.001} +{"train_loss": 0.039532288908958435, "train_loss_bc": 0.039532288908958435, "train_loss_llm": 0.0, "grad_norm": 0.029186084866523743, "global_step": 12777, "epoch": 354, "lr": 0.001} +{"train_loss": 0.039725057780742645, "train_loss_bc": 0.039725057780742645, "train_loss_llm": 0.0, "grad_norm": 0.0434439480304718, "global_step": 12778, "epoch": 354, "lr": 0.001} +{"train_loss": 0.037065012432220906, "train_loss_bc": 0.0334317646920681, "train_loss_llm": 0.0002697408199310303, "grad_norm": 0.06198854744434357, "global_step": 12779, "epoch": 354, "lr": 0.001, "val_loss": 0.06522160023450851} +{"train_loss": 0.038765911012887955, "train_loss_bc": 0.038765911012887955, "train_loss_llm": 0.0, "grad_norm": 0.07774647325277328, "global_step": 12780, "epoch": 355, "lr": 0.001} +{"train_loss": 0.0343761146068573, "train_loss_bc": 0.0343761146068573, "train_loss_llm": 0.0, "grad_norm": 0.02285575307905674, "global_step": 12781, "epoch": 355, "lr": 0.001} +{"train_loss": 0.030123993754386902, "train_loss_bc": 0.030123993754386902, "train_loss_llm": 0.0, "grad_norm": 0.04026167467236519, "global_step": 12782, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03670593351125717, "train_loss_bc": 0.03670593351125717, "train_loss_llm": 0.0, "grad_norm": 0.05842503160238266, "global_step": 12783, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03738557547330856, "train_loss_bc": 0.03738557547330856, "train_loss_llm": 0.0, "grad_norm": 0.07339813560247421, "global_step": 12784, "epoch": 355, "lr": 0.001} +{"train_loss": 0.038386303931474686, "train_loss_bc": 0.038386303931474686, "train_loss_llm": 0.0, "grad_norm": 0.03125359117984772, "global_step": 12785, "epoch": 355, "lr": 0.001} +{"train_loss": 0.039134759455919266, "train_loss_bc": 0.039134759455919266, "train_loss_llm": 0.0, "grad_norm": 0.049271341413259506, "global_step": 12786, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03212475776672363, "train_loss_bc": 0.03212475776672363, "train_loss_llm": 0.0, "grad_norm": 0.06781894713640213, "global_step": 12787, "epoch": 355, "lr": 0.001} +{"train_loss": 0.034307438880205154, "train_loss_bc": 0.034307438880205154, "train_loss_llm": 0.0, "grad_norm": 0.07835213094949722, "global_step": 12788, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03258540853857994, "train_loss_bc": 0.03258540853857994, "train_loss_llm": 0.0, "grad_norm": 0.021221792325377464, "global_step": 12789, "epoch": 355, "lr": 0.001} +{"train_loss": 0.044047851115465164, "train_loss_bc": 0.044047851115465164, "train_loss_llm": 0.0, "grad_norm": 0.03978719934821129, "global_step": 12790, "epoch": 355, "lr": 0.001} +{"train_loss": 0.031895995140075684, "train_loss_bc": 0.031872011721134186, "train_loss_llm": 0.0023982226848602295, "grad_norm": 0.050365008413791656, "global_step": 12791, "epoch": 355, "lr": 0.001} +{"train_loss": 0.029855817556381226, "train_loss_bc": 0.029855817556381226, "train_loss_llm": 0.0, "grad_norm": 0.06913238018751144, "global_step": 12792, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03271154686808586, "train_loss_bc": 0.03271154686808586, "train_loss_llm": 0.0, "grad_norm": 0.02253907360136509, "global_step": 12793, "epoch": 355, "lr": 0.001} +{"train_loss": 0.033216893672943115, "train_loss_bc": 0.033216893672943115, "train_loss_llm": 0.0, "grad_norm": 0.030123498290777206, "global_step": 12794, "epoch": 355, "lr": 0.001} +{"train_loss": 0.0433434396982193, "train_loss_bc": 0.0433434396982193, "train_loss_llm": 0.0, "grad_norm": 0.042867615818977356, "global_step": 12795, "epoch": 355, "lr": 0.001} +{"train_loss": 0.036015741527080536, "train_loss_bc": 0.03600994125008583, "train_loss_llm": 0.0005799233913421631, "grad_norm": 0.05602136254310608, "global_step": 12796, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03857676684856415, "train_loss_bc": 0.03857676684856415, "train_loss_llm": 0.0, "grad_norm": 0.02169390767812729, "global_step": 12797, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03866003826260567, "train_loss_bc": 0.03866003826260567, "train_loss_llm": 0.0, "grad_norm": 0.03748831897974014, "global_step": 12798, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03423897176980972, "train_loss_bc": 0.03423897176980972, "train_loss_llm": 0.0, "grad_norm": 0.04938314110040665, "global_step": 12799, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03576616197824478, "train_loss_bc": 0.03576616197824478, "train_loss_llm": 0.0, "grad_norm": 0.058905743062496185, "global_step": 12800, "epoch": 355, "lr": 0.001} +{"train_loss": 0.0367489829659462, "train_loss_bc": 0.0367489829659462, "train_loss_llm": 0.0, "grad_norm": 0.02981634810566902, "global_step": 12801, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03583328053355217, "train_loss_bc": 0.03583328053355217, "train_loss_llm": 0.0, "grad_norm": 0.05393742397427559, "global_step": 12802, "epoch": 355, "lr": 0.001} +{"train_loss": 0.043594155460596085, "train_loss_bc": 0.043594155460596085, "train_loss_llm": 0.0, "grad_norm": 0.07513459026813507, "global_step": 12803, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03681734576821327, "train_loss_bc": 0.03681088238954544, "train_loss_llm": 0.000646442174911499, "grad_norm": 0.09150717407464981, "global_step": 12804, "epoch": 355, "lr": 0.001} +{"train_loss": 0.0432044081389904, "train_loss_bc": 0.04319657012820244, "train_loss_llm": 0.0007837414741516113, "grad_norm": 0.02564522810280323, "global_step": 12805, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03252742439508438, "train_loss_bc": 0.032514408230781555, "train_loss_llm": 0.0013016164302825928, "grad_norm": 0.038146063685417175, "global_step": 12806, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03869418427348137, "train_loss_bc": 0.03869418427348137, "train_loss_llm": 0.0, "grad_norm": 0.0513756237924099, "global_step": 12807, "epoch": 355, "lr": 0.001} +{"train_loss": 0.04396099969744682, "train_loss_bc": 0.04395059868693352, "train_loss_llm": 0.001039952039718628, "grad_norm": 0.07539696246385574, "global_step": 12808, "epoch": 355, "lr": 0.001} +{"train_loss": 0.041019707918167114, "train_loss_bc": 0.04099724441766739, "train_loss_llm": 0.002246350049972534, "grad_norm": 0.026379037648439407, "global_step": 12809, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03996994346380234, "train_loss_bc": 0.03996994346380234, "train_loss_llm": 0.0, "grad_norm": 0.049487072974443436, "global_step": 12810, "epoch": 355, "lr": 0.001} +{"train_loss": 0.040551021695137024, "train_loss_bc": 0.040518347173929214, "train_loss_llm": 0.0032674670219421387, "grad_norm": 0.062420159578323364, "global_step": 12811, "epoch": 355, "lr": 0.001} +{"train_loss": 0.0400492399930954, "train_loss_bc": 0.040033478289842606, "train_loss_llm": 0.0015760362148284912, "grad_norm": 0.08548081666231155, "global_step": 12812, "epoch": 355, "lr": 0.001} +{"train_loss": 0.034189071506261826, "train_loss_bc": 0.034189071506261826, "train_loss_llm": 0.0, "grad_norm": 0.024691592901945114, "global_step": 12813, "epoch": 355, "lr": 0.001} +{"train_loss": 0.036174699664115906, "train_loss_bc": 0.036174699664115906, "train_loss_llm": 0.0, "grad_norm": 0.03940577432513237, "global_step": 12814, "epoch": 355, "lr": 0.001} +{"train_loss": 0.03715793343467845, "train_loss_bc": 0.04212571680545807, "train_loss_llm": 0.0, "grad_norm": 0.06069783493876457, "global_step": 12815, "epoch": 355, "lr": 0.001, "train/cumulative_reward": 98.90841688029442, "train/mean_score": 12.814397332457915, "train/success_rate": 0.3, "test/cumulative_reward": 34.53472467108313, "test/mean_score": 8.296276085971119, "test/success_rate": 0.18, "val_loss": 0.06509746611118317, "train_action_mse_error": 0.03980850428342819} +{"train_loss": 0.037501465529203415, "train_loss_bc": 0.037501465529203415, "train_loss_llm": 0.0, "grad_norm": 0.06930209696292877, "global_step": 12816, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03694121167063713, "train_loss_bc": 0.03694121167063713, "train_loss_llm": 0.0, "grad_norm": 0.03177996352314949, "global_step": 12817, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03681584820151329, "train_loss_bc": 0.03681584820151329, "train_loss_llm": 0.0, "grad_norm": 0.04566974192857742, "global_step": 12818, "epoch": 356, "lr": 0.001} +{"train_loss": 0.036437347531318665, "train_loss_bc": 0.036437347531318665, "train_loss_llm": 0.0, "grad_norm": 0.06294072419404984, "global_step": 12819, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03704164922237396, "train_loss_bc": 0.03704164922237396, "train_loss_llm": 0.0, "grad_norm": 0.06702551245689392, "global_step": 12820, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03822460398077965, "train_loss_bc": 0.03821844980120659, "train_loss_llm": 0.0006155669689178467, "grad_norm": 0.02456299215555191, "global_step": 12821, "epoch": 356, "lr": 0.001} +{"train_loss": 0.033114150166511536, "train_loss_bc": 0.033114150166511536, "train_loss_llm": 0.0, "grad_norm": 0.030811086297035217, "global_step": 12822, "epoch": 356, "lr": 0.001} +{"train_loss": 0.035232286900281906, "train_loss_bc": 0.035232286900281906, "train_loss_llm": 0.0, "grad_norm": 0.04629702493548393, "global_step": 12823, "epoch": 356, "lr": 0.001} +{"train_loss": 0.0413459837436676, "train_loss_bc": 0.041329219937324524, "train_loss_llm": 0.0016763508319854736, "grad_norm": 0.05880940333008766, "global_step": 12824, "epoch": 356, "lr": 0.001} +{"train_loss": 0.037535473704338074, "train_loss_bc": 0.037535473704338074, "train_loss_llm": 0.0, "grad_norm": 0.02267642877995968, "global_step": 12825, "epoch": 356, "lr": 0.001} +{"train_loss": 0.0349857397377491, "train_loss_bc": 0.0349857397377491, "train_loss_llm": 0.0, "grad_norm": 0.040381405502557755, "global_step": 12826, "epoch": 356, "lr": 0.001} +{"train_loss": 0.038257431238889694, "train_loss_bc": 0.038257431238889694, "train_loss_llm": 0.0, "grad_norm": 0.05319708585739136, "global_step": 12827, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03424959257245064, "train_loss_bc": 0.03424959257245064, "train_loss_llm": 0.0, "grad_norm": 0.06372654438018799, "global_step": 12828, "epoch": 356, "lr": 0.001} +{"train_loss": 0.04103780910372734, "train_loss_bc": 0.04103754460811615, "train_loss_llm": 2.6464462280273438e-05, "grad_norm": 0.02313815802335739, "global_step": 12829, "epoch": 356, "lr": 0.001} +{"train_loss": 0.04233997315168381, "train_loss_bc": 0.04233997315168381, "train_loss_llm": 0.0, "grad_norm": 0.0359196662902832, "global_step": 12830, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03123474307358265, "train_loss_bc": 0.03123474307358265, "train_loss_llm": 0.0, "grad_norm": 0.054405637085437775, "global_step": 12831, "epoch": 356, "lr": 0.001} +{"train_loss": 0.029505234211683273, "train_loss_bc": 0.029505234211683273, "train_loss_llm": 0.0, "grad_norm": 0.06810939311981201, "global_step": 12832, "epoch": 356, "lr": 0.001} +{"train_loss": 0.04073292016983032, "train_loss_bc": 0.04073000326752663, "train_loss_llm": 0.00029155611991882324, "grad_norm": 0.02789418213069439, "global_step": 12833, "epoch": 356, "lr": 0.001} +{"train_loss": 0.039063312113285065, "train_loss_bc": 0.039063312113285065, "train_loss_llm": 0.0, "grad_norm": 0.043193280696868896, "global_step": 12834, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03191892430186272, "train_loss_bc": 0.03191892430186272, "train_loss_llm": 0.0, "grad_norm": 0.057508453726768494, "global_step": 12835, "epoch": 356, "lr": 0.001} +{"train_loss": 0.04070190712809563, "train_loss_bc": 0.04070190712809563, "train_loss_llm": 0.0, "grad_norm": 0.06993849575519562, "global_step": 12836, "epoch": 356, "lr": 0.001} +{"train_loss": 0.04245812073349953, "train_loss_bc": 0.04245234653353691, "train_loss_llm": 0.0005773603916168213, "grad_norm": 0.025246750563383102, "global_step": 12837, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03576336055994034, "train_loss_bc": 0.03576336055994034, "train_loss_llm": 0.0, "grad_norm": 0.037379540503025055, "global_step": 12838, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03135213255882263, "train_loss_bc": 0.03135213255882263, "train_loss_llm": 0.0, "grad_norm": 0.04935399442911148, "global_step": 12839, "epoch": 356, "lr": 0.001} +{"train_loss": 0.037949059158563614, "train_loss_bc": 0.0379437692463398, "train_loss_llm": 0.0005290806293487549, "grad_norm": 0.05497107282280922, "global_step": 12840, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03862188756465912, "train_loss_bc": 0.03862188756465912, "train_loss_llm": 0.0, "grad_norm": 0.025291860103607178, "global_step": 12841, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03550732508301735, "train_loss_bc": 0.03549203276634216, "train_loss_llm": 0.0015293657779693604, "grad_norm": 0.04072181135416031, "global_step": 12842, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03175799921154976, "train_loss_bc": 0.03174085542559624, "train_loss_llm": 0.001714557409286499, "grad_norm": 0.05601077154278755, "global_step": 12843, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03099062852561474, "train_loss_bc": 0.0309757012873888, "train_loss_llm": 0.0014926493167877197, "grad_norm": 0.0716073289513588, "global_step": 12844, "epoch": 356, "lr": 0.001} +{"train_loss": 0.04143719747662544, "train_loss_bc": 0.04143719747662544, "train_loss_llm": 0.0, "grad_norm": 0.025640996173024178, "global_step": 12845, "epoch": 356, "lr": 0.001} +{"train_loss": 0.0352339968085289, "train_loss_bc": 0.03523070365190506, "train_loss_llm": 0.0003293156623840332, "grad_norm": 0.038959089666604996, "global_step": 12846, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03653409332036972, "train_loss_bc": 0.03653409332036972, "train_loss_llm": 0.0, "grad_norm": 0.05952581763267517, "global_step": 12847, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03738008067011833, "train_loss_bc": 0.03738008067011833, "train_loss_llm": 0.0, "grad_norm": 0.07215642184019089, "global_step": 12848, "epoch": 356, "lr": 0.001} +{"train_loss": 0.04110841453075409, "train_loss_bc": 0.04110841453075409, "train_loss_llm": 0.0, "grad_norm": 0.019228577613830566, "global_step": 12849, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03801334276795387, "train_loss_bc": 0.03801334276795387, "train_loss_llm": 0.0, "grad_norm": 0.03341839835047722, "global_step": 12850, "epoch": 356, "lr": 0.001} +{"train_loss": 0.03701307303789589, "train_loss_bc": 0.044145382940769196, "train_loss_llm": 0.0, "grad_norm": 0.054049983620643616, "global_step": 12851, "epoch": 356, "lr": 0.001, "val_loss": 0.06519252806901932} +{"train_loss": 0.0291743166744709, "train_loss_bc": 0.029166925698518753, "train_loss_llm": 0.0007390975952148438, "grad_norm": 0.06692169606685638, "global_step": 12852, "epoch": 357, "lr": 0.001} +{"train_loss": 0.04182363301515579, "train_loss_bc": 0.04180409014225006, "train_loss_llm": 0.001954406499862671, "grad_norm": 0.027928905561566353, "global_step": 12853, "epoch": 357, "lr": 0.001} +{"train_loss": 0.033918268978595734, "train_loss_bc": 0.03390178829431534, "train_loss_llm": 0.0016478896141052246, "grad_norm": 0.040151145309209824, "global_step": 12854, "epoch": 357, "lr": 0.001} +{"train_loss": 0.040913403034210205, "train_loss_bc": 0.040913403034210205, "train_loss_llm": 0.0, "grad_norm": 0.05093413591384888, "global_step": 12855, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03873589262366295, "train_loss_bc": 0.03873589262366295, "train_loss_llm": 0.0, "grad_norm": 0.06392337381839752, "global_step": 12856, "epoch": 357, "lr": 0.001} +{"train_loss": 0.04174800217151642, "train_loss_bc": 0.04173848405480385, "train_loss_llm": 0.0009516775608062744, "grad_norm": 0.023207198828458786, "global_step": 12857, "epoch": 357, "lr": 0.001} +{"train_loss": 0.0318644717335701, "train_loss_bc": 0.03185689076781273, "train_loss_llm": 0.0007579326629638672, "grad_norm": 0.037267301231622696, "global_step": 12858, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03512784466147423, "train_loss_bc": 0.03512784466147423, "train_loss_llm": 0.0, "grad_norm": 0.0472136065363884, "global_step": 12859, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03781447559595108, "train_loss_bc": 0.03781447559595108, "train_loss_llm": 0.0, "grad_norm": 0.061905231326818466, "global_step": 12860, "epoch": 357, "lr": 0.001} +{"train_loss": 0.0408613383769989, "train_loss_bc": 0.0408613383769989, "train_loss_llm": 0.0, "grad_norm": 0.026630358770489693, "global_step": 12861, "epoch": 357, "lr": 0.001} +{"train_loss": 0.035421352833509445, "train_loss_bc": 0.03540972247719765, "train_loss_llm": 0.0011632144451141357, "grad_norm": 0.03784029185771942, "global_step": 12862, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03848674148321152, "train_loss_bc": 0.03848674148321152, "train_loss_llm": 0.0, "grad_norm": 0.05698889493942261, "global_step": 12863, "epoch": 357, "lr": 0.001} +{"train_loss": 0.039167407900094986, "train_loss_bc": 0.039167407900094986, "train_loss_llm": 0.0, "grad_norm": 0.07204336673021317, "global_step": 12864, "epoch": 357, "lr": 0.001} +{"train_loss": 0.0337093286216259, "train_loss_bc": 0.0337093286216259, "train_loss_llm": 0.0, "grad_norm": 0.025096600875258446, "global_step": 12865, "epoch": 357, "lr": 0.001} +{"train_loss": 0.036752697080373764, "train_loss_bc": 0.036752697080373764, "train_loss_llm": 0.0, "grad_norm": 0.04082586243748665, "global_step": 12866, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03616029769182205, "train_loss_bc": 0.03616029769182205, "train_loss_llm": 0.0, "grad_norm": 0.05809410661458969, "global_step": 12867, "epoch": 357, "lr": 0.001} +{"train_loss": 0.030738528817892075, "train_loss_bc": 0.030738528817892075, "train_loss_llm": 0.0, "grad_norm": 0.0725727453827858, "global_step": 12868, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03348956257104874, "train_loss_bc": 0.03348956257104874, "train_loss_llm": 0.0, "grad_norm": 0.02208876796066761, "global_step": 12869, "epoch": 357, "lr": 0.001} +{"train_loss": 0.0349857397377491, "train_loss_bc": 0.03496244177222252, "train_loss_llm": 0.0023299753665924072, "grad_norm": 0.03931347653269768, "global_step": 12870, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03893851861357689, "train_loss_bc": 0.03893851861357689, "train_loss_llm": 0.0, "grad_norm": 0.053727082908153534, "global_step": 12871, "epoch": 357, "lr": 0.001} +{"train_loss": 0.041096970438957214, "train_loss_bc": 0.04109308868646622, "train_loss_llm": 0.0003883540630340576, "grad_norm": 0.0719810426235199, "global_step": 12872, "epoch": 357, "lr": 0.001} +{"train_loss": 0.038834866136312485, "train_loss_bc": 0.038834866136312485, "train_loss_llm": 0.0, "grad_norm": 0.022072087973356247, "global_step": 12873, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03852253407239914, "train_loss_bc": 0.03852253407239914, "train_loss_llm": 0.0, "grad_norm": 0.03191493824124336, "global_step": 12874, "epoch": 357, "lr": 0.001} +{"train_loss": 0.040215522050857544, "train_loss_bc": 0.040215522050857544, "train_loss_llm": 0.0, "grad_norm": 0.045712437480688095, "global_step": 12875, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03464015945792198, "train_loss_bc": 0.03464015945792198, "train_loss_llm": 0.0, "grad_norm": 0.057295121252536774, "global_step": 12876, "epoch": 357, "lr": 0.001} +{"train_loss": 0.040687400847673416, "train_loss_bc": 0.040687400847673416, "train_loss_llm": 0.0, "grad_norm": 0.026020821183919907, "global_step": 12877, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03538151830434799, "train_loss_bc": 0.03538151830434799, "train_loss_llm": 0.0, "grad_norm": 0.05164472758769989, "global_step": 12878, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03640666976571083, "train_loss_bc": 0.03639805689454079, "train_loss_llm": 0.0008611083030700684, "grad_norm": 0.07238030433654785, "global_step": 12879, "epoch": 357, "lr": 0.001} +{"train_loss": 0.035625237971544266, "train_loss_bc": 0.035625237971544266, "train_loss_llm": 0.0, "grad_norm": 0.08437875658273697, "global_step": 12880, "epoch": 357, "lr": 0.001} +{"train_loss": 0.031021814793348312, "train_loss_bc": 0.031021814793348312, "train_loss_llm": 0.0, "grad_norm": 0.023697616532444954, "global_step": 12881, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03613809496164322, "train_loss_bc": 0.036130357533693314, "train_loss_llm": 0.0007735788822174072, "grad_norm": 0.04557693377137184, "global_step": 12882, "epoch": 357, "lr": 0.001} +{"train_loss": 0.041621219366788864, "train_loss_bc": 0.0416153259575367, "train_loss_llm": 0.0005894601345062256, "grad_norm": 0.06885260343551636, "global_step": 12883, "epoch": 357, "lr": 0.001} +{"train_loss": 0.04231227934360504, "train_loss_bc": 0.04231227934360504, "train_loss_llm": 0.0, "grad_norm": 0.09012037515640259, "global_step": 12884, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03811699524521828, "train_loss_bc": 0.03811303898692131, "train_loss_llm": 0.00039571523666381836, "grad_norm": 0.03150395676493645, "global_step": 12885, "epoch": 357, "lr": 0.001} +{"train_loss": 0.041381970047950745, "train_loss_bc": 0.041381970047950745, "train_loss_llm": 0.0, "grad_norm": 0.04763809219002724, "global_step": 12886, "epoch": 357, "lr": 0.001} +{"train_loss": 0.03708805785410934, "train_loss_bc": 0.03333500772714615, "train_loss_llm": 0.0, "grad_norm": 0.06238890439271927, "global_step": 12887, "epoch": 357, "lr": 0.001, "val_loss": 0.06535094231367111} +{"train_loss": 0.04083928093314171, "train_loss_bc": 0.04082570597529411, "train_loss_llm": 0.0013573765754699707, "grad_norm": 0.07665058225393295, "global_step": 12888, "epoch": 358, "lr": 0.001} +{"train_loss": 0.04235998913645744, "train_loss_bc": 0.04235998913645744, "train_loss_llm": 0.0, "grad_norm": 0.02796722576022148, "global_step": 12889, "epoch": 358, "lr": 0.001} +{"train_loss": 0.032585494220256805, "train_loss_bc": 0.032585494220256805, "train_loss_llm": 0.0, "grad_norm": 0.034867603331804276, "global_step": 12890, "epoch": 358, "lr": 0.001} +{"train_loss": 0.029969168826937675, "train_loss_bc": 0.029969168826937675, "train_loss_llm": 0.0, "grad_norm": 0.039673689752817154, "global_step": 12891, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03230802342295647, "train_loss_bc": 0.03229174762964249, "train_loss_llm": 0.0016276538372039795, "grad_norm": 0.04520508646965027, "global_step": 12892, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03305595740675926, "train_loss_bc": 0.03305595740675926, "train_loss_llm": 0.0, "grad_norm": 0.02146364375948906, "global_step": 12893, "epoch": 358, "lr": 0.001} +{"train_loss": 0.036944951862096786, "train_loss_bc": 0.03692731261253357, "train_loss_llm": 0.0017639994621276855, "grad_norm": 0.031135793775320053, "global_step": 12894, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03222456946969032, "train_loss_bc": 0.03221455216407776, "train_loss_llm": 0.0010018348693847656, "grad_norm": 0.03774615004658699, "global_step": 12895, "epoch": 358, "lr": 0.001} +{"train_loss": 0.029590528458356857, "train_loss_bc": 0.029590528458356857, "train_loss_llm": 0.0, "grad_norm": 0.04460926353931427, "global_step": 12896, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03947017341852188, "train_loss_bc": 0.03947017341852188, "train_loss_llm": 0.0, "grad_norm": 0.027239341288805008, "global_step": 12897, "epoch": 358, "lr": 0.001} +{"train_loss": 0.033924274146556854, "train_loss_bc": 0.033924274146556854, "train_loss_llm": 0.0, "grad_norm": 0.033195700496435165, "global_step": 12898, "epoch": 358, "lr": 0.001} +{"train_loss": 0.04302363097667694, "train_loss_bc": 0.04302363097667694, "train_loss_llm": 0.0, "grad_norm": 0.04357320815324783, "global_step": 12899, "epoch": 358, "lr": 0.001} +{"train_loss": 0.038302935659885406, "train_loss_bc": 0.038302935659885406, "train_loss_llm": 0.0, "grad_norm": 0.05126819759607315, "global_step": 12900, "epoch": 358, "lr": 0.001} +{"train_loss": 0.0386073961853981, "train_loss_bc": 0.0386073961853981, "train_loss_llm": 0.0, "grad_norm": 0.024819694459438324, "global_step": 12901, "epoch": 358, "lr": 0.001} +{"train_loss": 0.040031738579273224, "train_loss_bc": 0.040031738579273224, "train_loss_llm": 0.0, "grad_norm": 0.03755135461688042, "global_step": 12902, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03441818431019783, "train_loss_bc": 0.03441818431019783, "train_loss_llm": 0.0, "grad_norm": 0.04696943610906601, "global_step": 12903, "epoch": 358, "lr": 0.001} +{"train_loss": 0.038718290627002716, "train_loss_bc": 0.038718290627002716, "train_loss_llm": 0.0, "grad_norm": 0.06680682301521301, "global_step": 12904, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03952523320913315, "train_loss_bc": 0.03952523320913315, "train_loss_llm": 0.0, "grad_norm": 0.024660127237439156, "global_step": 12905, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03806473687291145, "train_loss_bc": 0.03806473687291145, "train_loss_llm": 0.0, "grad_norm": 0.03348942846059799, "global_step": 12906, "epoch": 358, "lr": 0.001} +{"train_loss": 0.046775106340646744, "train_loss_bc": 0.046757422387599945, "train_loss_llm": 0.0017683207988739014, "grad_norm": 0.04542500898241997, "global_step": 12907, "epoch": 358, "lr": 0.001} +{"train_loss": 0.036266013979911804, "train_loss_bc": 0.036266013979911804, "train_loss_llm": 0.0, "grad_norm": 0.058205895125865936, "global_step": 12908, "epoch": 358, "lr": 0.001} +{"train_loss": 0.04431770741939545, "train_loss_bc": 0.04431770741939545, "train_loss_llm": 0.0, "grad_norm": 0.02896295301616192, "global_step": 12909, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03488108888268471, "train_loss_bc": 0.03488108888268471, "train_loss_llm": 0.0, "grad_norm": 0.04132557287812233, "global_step": 12910, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03682912141084671, "train_loss_bc": 0.036825891584157944, "train_loss_llm": 0.0003229975700378418, "grad_norm": 0.05759570002555847, "global_step": 12911, "epoch": 358, "lr": 0.001} +{"train_loss": 0.04245883226394653, "train_loss_bc": 0.04245883226394653, "train_loss_llm": 0.0, "grad_norm": 0.07944026589393616, "global_step": 12912, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03538820147514343, "train_loss_bc": 0.03537331521511078, "train_loss_llm": 0.001488715410232544, "grad_norm": 0.024318689480423927, "global_step": 12913, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03627369552850723, "train_loss_bc": 0.03627369552850723, "train_loss_llm": 0.0, "grad_norm": 0.03864292800426483, "global_step": 12914, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03638622909784317, "train_loss_bc": 0.03637513518333435, "train_loss_llm": 0.0011094212532043457, "grad_norm": 0.04676525294780731, "global_step": 12915, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03704246133565903, "train_loss_bc": 0.03704246133565903, "train_loss_llm": 0.0, "grad_norm": 0.04847957566380501, "global_step": 12916, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03710459917783737, "train_loss_bc": 0.03710459917783737, "train_loss_llm": 0.0, "grad_norm": 0.02129063755273819, "global_step": 12917, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03235279396176338, "train_loss_bc": 0.032342322170734406, "train_loss_llm": 0.0010471642017364502, "grad_norm": 0.036026738584041595, "global_step": 12918, "epoch": 358, "lr": 0.001} +{"train_loss": 0.039761144667863846, "train_loss_bc": 0.03975600376725197, "train_loss_llm": 0.0005142092704772949, "grad_norm": 0.04256376251578331, "global_step": 12919, "epoch": 358, "lr": 0.001} +{"train_loss": 0.03895494341850281, "train_loss_bc": 0.03895494341850281, "train_loss_llm": 0.0, "grad_norm": 0.05167354643344879, "global_step": 12920, "epoch": 358, "lr": 0.001} +{"train_loss": 0.0326526015996933, "train_loss_bc": 0.0326526015996933, "train_loss_llm": 0.0, "grad_norm": 0.019768865779042244, "global_step": 12921, "epoch": 358, "lr": 0.001} +{"train_loss": 0.028984442353248596, "train_loss_bc": 0.028966018930077553, "train_loss_llm": 0.0018423199653625488, "grad_norm": 0.03399524837732315, "global_step": 12922, "epoch": 358, "lr": 0.001} +{"train_loss": 0.036653850227594376, "train_loss_bc": 0.029145067557692528, "train_loss_llm": 0.0, "grad_norm": 0.04658123105764389, "global_step": 12923, "epoch": 358, "lr": 0.001, "val_loss": 0.06576821953058243} +{"train_loss": 0.03077351115643978, "train_loss_bc": 0.030773399397730827, "train_loss_llm": 1.1116266250610352e-05, "grad_norm": 0.05510472133755684, "global_step": 12924, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03854959085583687, "train_loss_bc": 0.03854959085583687, "train_loss_llm": 0.0, "grad_norm": 0.02939535677433014, "global_step": 12925, "epoch": 359, "lr": 0.001} +{"train_loss": 0.036332521587610245, "train_loss_bc": 0.03631429001688957, "train_loss_llm": 0.0018229782581329346, "grad_norm": 0.04664681479334831, "global_step": 12926, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03445906192064285, "train_loss_bc": 0.03445906192064285, "train_loss_llm": 0.0, "grad_norm": 0.06370323896408081, "global_step": 12927, "epoch": 359, "lr": 0.001} +{"train_loss": 0.04106305539608002, "train_loss_bc": 0.04106305539608002, "train_loss_llm": 0.0, "grad_norm": 0.0757194384932518, "global_step": 12928, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03991956636309624, "train_loss_bc": 0.03991956636309624, "train_loss_llm": 0.0, "grad_norm": 0.027375170961022377, "global_step": 12929, "epoch": 359, "lr": 0.001} +{"train_loss": 0.045787688344717026, "train_loss_bc": 0.045787688344717026, "train_loss_llm": 0.0, "grad_norm": 0.04034766927361488, "global_step": 12930, "epoch": 359, "lr": 0.001} +{"train_loss": 0.030920332297682762, "train_loss_bc": 0.030920332297682762, "train_loss_llm": 0.0, "grad_norm": 0.051353417336940765, "global_step": 12931, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03643772006034851, "train_loss_bc": 0.03643772006034851, "train_loss_llm": 0.0, "grad_norm": 0.06482899188995361, "global_step": 12932, "epoch": 359, "lr": 0.001} +{"train_loss": 0.034206606447696686, "train_loss_bc": 0.034206606447696686, "train_loss_llm": 0.0, "grad_norm": 0.021334799006581306, "global_step": 12933, "epoch": 359, "lr": 0.001} +{"train_loss": 0.035854458808898926, "train_loss_bc": 0.03583803027868271, "train_loss_llm": 0.0016429126262664795, "grad_norm": 0.032758571207523346, "global_step": 12934, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03733310103416443, "train_loss_bc": 0.037317678332328796, "train_loss_llm": 0.0015421509742736816, "grad_norm": 0.040738753974437714, "global_step": 12935, "epoch": 359, "lr": 0.001} +{"train_loss": 0.04170724004507065, "train_loss_bc": 0.04170724004507065, "train_loss_llm": 0.0, "grad_norm": 0.04946703463792801, "global_step": 12936, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03820890933275223, "train_loss_bc": 0.03820890933275223, "train_loss_llm": 0.0, "grad_norm": 0.022453177720308304, "global_step": 12937, "epoch": 359, "lr": 0.001} +{"train_loss": 0.0456966832280159, "train_loss_bc": 0.0456966832280159, "train_loss_llm": 0.0, "grad_norm": 0.041801732033491135, "global_step": 12938, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03197992965579033, "train_loss_bc": 0.03197949007153511, "train_loss_llm": 4.407763481140137e-05, "grad_norm": 0.055172909051179886, "global_step": 12939, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03367338702082634, "train_loss_bc": 0.03366856649518013, "train_loss_llm": 0.0004819929599761963, "grad_norm": 0.07183782011270523, "global_step": 12940, "epoch": 359, "lr": 0.001} +{"train_loss": 0.04239046201109886, "train_loss_bc": 0.04235968738794327, "train_loss_llm": 0.0030775368213653564, "grad_norm": 0.020361579954624176, "global_step": 12941, "epoch": 359, "lr": 0.001} +{"train_loss": 0.0385521836578846, "train_loss_bc": 0.0385521836578846, "train_loss_llm": 0.0, "grad_norm": 0.027410848066210747, "global_step": 12942, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03380493074655533, "train_loss_bc": 0.03380493074655533, "train_loss_llm": 0.0, "grad_norm": 0.040335047990083694, "global_step": 12943, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03745193034410477, "train_loss_bc": 0.03745193034410477, "train_loss_llm": 0.0, "grad_norm": 0.05331595242023468, "global_step": 12944, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03407572954893112, "train_loss_bc": 0.03407572954893112, "train_loss_llm": 0.0, "grad_norm": 0.019862258806824684, "global_step": 12945, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03653765842318535, "train_loss_bc": 0.036513496190309525, "train_loss_llm": 0.0024161934852600098, "grad_norm": 0.03674719110131264, "global_step": 12946, "epoch": 359, "lr": 0.001} +{"train_loss": 0.041006676852703094, "train_loss_bc": 0.041006676852703094, "train_loss_llm": 0.0, "grad_norm": 0.04774270951747894, "global_step": 12947, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03727748617529869, "train_loss_bc": 0.03727748617529869, "train_loss_llm": 0.0, "grad_norm": 0.06325280666351318, "global_step": 12948, "epoch": 359, "lr": 0.001} +{"train_loss": 0.038013603538274765, "train_loss_bc": 0.038013603538274765, "train_loss_llm": 0.0, "grad_norm": 0.027606336399912834, "global_step": 12949, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03354671224951744, "train_loss_bc": 0.03352916240692139, "train_loss_llm": 0.0017549097537994385, "grad_norm": 0.027241209521889687, "global_step": 12950, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03766642510890961, "train_loss_bc": 0.03766642510890961, "train_loss_llm": 0.0, "grad_norm": 0.040012575685977936, "global_step": 12951, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03706907480955124, "train_loss_bc": 0.03706907480955124, "train_loss_llm": 0.0, "grad_norm": 0.04972565919160843, "global_step": 12952, "epoch": 359, "lr": 0.001} +{"train_loss": 0.031413111835718155, "train_loss_bc": 0.031413111835718155, "train_loss_llm": 0.0, "grad_norm": 0.023254578933119774, "global_step": 12953, "epoch": 359, "lr": 0.001} +{"train_loss": 0.032597173005342484, "train_loss_bc": 0.03258459270000458, "train_loss_llm": 0.0012579560279846191, "grad_norm": 0.0344531275331974, "global_step": 12954, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03181707486510277, "train_loss_bc": 0.03180694207549095, "train_loss_llm": 0.0010131299495697021, "grad_norm": 0.04019924998283386, "global_step": 12955, "epoch": 359, "lr": 0.001} +{"train_loss": 0.04089128226041794, "train_loss_bc": 0.04089128226041794, "train_loss_llm": 0.0, "grad_norm": 0.054940901696681976, "global_step": 12956, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03416137024760246, "train_loss_bc": 0.03416137024760246, "train_loss_llm": 0.0, "grad_norm": 0.021351298317313194, "global_step": 12957, "epoch": 359, "lr": 0.001} +{"train_loss": 0.03661420941352844, "train_loss_bc": 0.03661302849650383, "train_loss_llm": 0.00011792778968811035, "grad_norm": 0.032963111996650696, "global_step": 12958, "epoch": 359, "lr": 0.001} +{"train_loss": 0.0366972450994783, "train_loss_bc": 0.03331036493182182, "train_loss_llm": 0.0, "grad_norm": 0.04490870609879494, "global_step": 12959, "epoch": 359, "lr": 0.001, "val_loss": 0.06529611349105835} +{"train_loss": 0.035140953958034515, "train_loss_bc": 0.035140953958034515, "train_loss_llm": 0.0, "grad_norm": 0.05511409789323807, "global_step": 12960, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03151404485106468, "train_loss_bc": 0.03151404485106468, "train_loss_llm": 0.0, "grad_norm": 0.022256338968873024, "global_step": 12961, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03812091052532196, "train_loss_bc": 0.03812091052532196, "train_loss_llm": 0.0, "grad_norm": 0.027308566495776176, "global_step": 12962, "epoch": 360, "lr": 0.001} +{"train_loss": 0.037148091942071915, "train_loss_bc": 0.037148091942071915, "train_loss_llm": 0.0, "grad_norm": 0.03873958811163902, "global_step": 12963, "epoch": 360, "lr": 0.001} +{"train_loss": 0.030228685587644577, "train_loss_bc": 0.030228685587644577, "train_loss_llm": 0.0, "grad_norm": 0.050112850964069366, "global_step": 12964, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03395936265587807, "train_loss_bc": 0.03394320234656334, "train_loss_llm": 0.001616060733795166, "grad_norm": 0.026629479601979256, "global_step": 12965, "epoch": 360, "lr": 0.001} +{"train_loss": 0.034030430018901825, "train_loss_bc": 0.034030430018901825, "train_loss_llm": 0.0, "grad_norm": 0.04265153408050537, "global_step": 12966, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03214574605226517, "train_loss_bc": 0.03213805332779884, "train_loss_llm": 0.0007692277431488037, "grad_norm": 0.05066128447651863, "global_step": 12967, "epoch": 360, "lr": 0.001} +{"train_loss": 0.038657259196043015, "train_loss_bc": 0.038657259196043015, "train_loss_llm": 0.0, "grad_norm": 0.06342899054288864, "global_step": 12968, "epoch": 360, "lr": 0.001} +{"train_loss": 0.042145032435655594, "train_loss_bc": 0.04214317724108696, "train_loss_llm": 0.0001856982707977295, "grad_norm": 0.021466271951794624, "global_step": 12969, "epoch": 360, "lr": 0.001} +{"train_loss": 0.032602425664663315, "train_loss_bc": 0.032602425664663315, "train_loss_llm": 0.0, "grad_norm": 0.03444571793079376, "global_step": 12970, "epoch": 360, "lr": 0.001} +{"train_loss": 0.04044947773218155, "train_loss_bc": 0.04044947773218155, "train_loss_llm": 0.0, "grad_norm": 0.0469248965382576, "global_step": 12971, "epoch": 360, "lr": 0.001} +{"train_loss": 0.04522404819726944, "train_loss_bc": 0.04522404819726944, "train_loss_llm": 0.0, "grad_norm": 0.06879042834043503, "global_step": 12972, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03702889382839203, "train_loss_bc": 0.03702889382839203, "train_loss_llm": 0.0, "grad_norm": 0.022289494052529335, "global_step": 12973, "epoch": 360, "lr": 0.001} +{"train_loss": 0.031470440328121185, "train_loss_bc": 0.031470440328121185, "train_loss_llm": 0.0, "grad_norm": 0.03264819085597992, "global_step": 12974, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03760064020752907, "train_loss_bc": 0.03760064020752907, "train_loss_llm": 0.0, "grad_norm": 0.048219356685876846, "global_step": 12975, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03960203751921654, "train_loss_bc": 0.039599813520908356, "train_loss_llm": 0.00022226572036743164, "grad_norm": 0.059289541095495224, "global_step": 12976, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03605891391634941, "train_loss_bc": 0.03605206310749054, "train_loss_llm": 0.0006850361824035645, "grad_norm": 0.024519506841897964, "global_step": 12977, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03880815953016281, "train_loss_bc": 0.03880815953016281, "train_loss_llm": 0.0, "grad_norm": 0.03251970186829567, "global_step": 12978, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03074982762336731, "train_loss_bc": 0.03074982762336731, "train_loss_llm": 0.0, "grad_norm": 0.05329383164644241, "global_step": 12979, "epoch": 360, "lr": 0.001} +{"train_loss": 0.038721635937690735, "train_loss_bc": 0.038721635937690735, "train_loss_llm": 0.0, "grad_norm": 0.058274660259485245, "global_step": 12980, "epoch": 360, "lr": 0.001} +{"train_loss": 0.04283633828163147, "train_loss_bc": 0.042833976447582245, "train_loss_llm": 0.00023615360260009766, "grad_norm": 0.02686823345720768, "global_step": 12981, "epoch": 360, "lr": 0.001} +{"train_loss": 0.033333953469991684, "train_loss_bc": 0.033333953469991684, "train_loss_llm": 0.0, "grad_norm": 0.04522727429866791, "global_step": 12982, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03216160088777542, "train_loss_bc": 0.03216160088777542, "train_loss_llm": 0.0, "grad_norm": 0.0528462715446949, "global_step": 12983, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03323648124933243, "train_loss_bc": 0.03323648124933243, "train_loss_llm": 0.0, "grad_norm": 0.05750834941864014, "global_step": 12984, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03773469105362892, "train_loss_bc": 0.037731580436229706, "train_loss_llm": 0.00031110644340515137, "grad_norm": 0.020805420354008675, "global_step": 12985, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03904861956834793, "train_loss_bc": 0.03904861956834793, "train_loss_llm": 0.0, "grad_norm": 0.02994566783308983, "global_step": 12986, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03819594532251358, "train_loss_bc": 0.03819594532251358, "train_loss_llm": 0.0, "grad_norm": 0.03882499411702156, "global_step": 12987, "epoch": 360, "lr": 0.001} +{"train_loss": 0.041638173162937164, "train_loss_bc": 0.04162568598985672, "train_loss_llm": 0.0012487471103668213, "grad_norm": 0.0530998557806015, "global_step": 12988, "epoch": 360, "lr": 0.001} +{"train_loss": 0.040510568767786026, "train_loss_bc": 0.040510568767786026, "train_loss_llm": 0.0, "grad_norm": 0.028279295191168785, "global_step": 12989, "epoch": 360, "lr": 0.001} +{"train_loss": 0.042813628911972046, "train_loss_bc": 0.042813628911972046, "train_loss_llm": 0.0, "grad_norm": 0.04737093299627304, "global_step": 12990, "epoch": 360, "lr": 0.001} +{"train_loss": 0.04429236426949501, "train_loss_bc": 0.04429236426949501, "train_loss_llm": 0.0, "grad_norm": 0.05861123278737068, "global_step": 12991, "epoch": 360, "lr": 0.001} +{"train_loss": 0.033802468329668045, "train_loss_bc": 0.033802468329668045, "train_loss_llm": 0.0, "grad_norm": 0.06194010749459267, "global_step": 12992, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03476458415389061, "train_loss_bc": 0.03476458415389061, "train_loss_llm": 0.0, "grad_norm": 0.022416796535253525, "global_step": 12993, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03380905091762543, "train_loss_bc": 0.03380905091762543, "train_loss_llm": 0.0, "grad_norm": 0.02655281126499176, "global_step": 12994, "epoch": 360, "lr": 0.001} +{"train_loss": 0.03653992748715811, "train_loss_bc": 0.02584248222410679, "train_loss_llm": 0.0009421408176422119, "grad_norm": 0.03837144374847412, "global_step": 12995, "epoch": 360, "lr": 0.001, "train/cumulative_reward": 112.79056190757508, "train/mean_score": 10.812740260822586, "train/success_rate": 0.2, "test/cumulative_reward": 29.939211030757352, "test/mean_score": 8.464264972821464, "test/success_rate": 0.2, "val_loss": 0.06533404439687729, "train_action_mse_error": 0.040035706013441086} +{"train_loss": 0.03577075153589249, "train_loss_bc": 0.03577075153589249, "train_loss_llm": 0.0, "grad_norm": 0.0505472756922245, "global_step": 12996, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03466547653079033, "train_loss_bc": 0.03466547653079033, "train_loss_llm": 0.0, "grad_norm": 0.020912976935505867, "global_step": 12997, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03850334510207176, "train_loss_bc": 0.03850334510207176, "train_loss_llm": 0.0, "grad_norm": 0.029164453968405724, "global_step": 12998, "epoch": 361, "lr": 0.001} +{"train_loss": 0.041168924421072006, "train_loss_bc": 0.041163112968206406, "train_loss_llm": 0.0005810856819152832, "grad_norm": 0.04070889204740524, "global_step": 12999, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03710264712572098, "train_loss_bc": 0.03710264712572098, "train_loss_llm": 0.0, "grad_norm": 0.046557582914829254, "global_step": 13000, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03721030056476593, "train_loss_bc": 0.03721030056476593, "train_loss_llm": 0.0, "grad_norm": 0.019604289904236794, "global_step": 13001, "epoch": 361, "lr": 0.001} +{"train_loss": 0.035146310925483704, "train_loss_bc": 0.035146310925483704, "train_loss_llm": 0.0, "grad_norm": 0.03330081328749657, "global_step": 13002, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03708968311548233, "train_loss_bc": 0.0370883084833622, "train_loss_llm": 0.0001373291015625, "grad_norm": 0.039123859256505966, "global_step": 13003, "epoch": 361, "lr": 0.001} +{"train_loss": 0.029214829206466675, "train_loss_bc": 0.029214829206466675, "train_loss_llm": 0.0, "grad_norm": 0.042564909905195236, "global_step": 13004, "epoch": 361, "lr": 0.001} +{"train_loss": 0.029675275087356567, "train_loss_bc": 0.029675275087356567, "train_loss_llm": 0.0, "grad_norm": 0.018351972103118896, "global_step": 13005, "epoch": 361, "lr": 0.001} +{"train_loss": 0.040319837629795074, "train_loss_bc": 0.040319837629795074, "train_loss_llm": 0.0, "grad_norm": 0.04477296769618988, "global_step": 13006, "epoch": 361, "lr": 0.001} +{"train_loss": 0.04074274003505707, "train_loss_bc": 0.04074274003505707, "train_loss_llm": 0.0, "grad_norm": 0.06952646374702454, "global_step": 13007, "epoch": 361, "lr": 0.001} +{"train_loss": 0.030310707166790962, "train_loss_bc": 0.0302958432585001, "train_loss_llm": 0.001486450433731079, "grad_norm": 0.08435576409101486, "global_step": 13008, "epoch": 361, "lr": 0.001} +{"train_loss": 0.04143424704670906, "train_loss_bc": 0.04143424704670906, "train_loss_llm": 0.0, "grad_norm": 0.026180369779467583, "global_step": 13009, "epoch": 361, "lr": 0.001} +{"train_loss": 0.0394582562148571, "train_loss_bc": 0.0394582562148571, "train_loss_llm": 0.0, "grad_norm": 0.04647301509976387, "global_step": 13010, "epoch": 361, "lr": 0.001} +{"train_loss": 0.033920831978321075, "train_loss_bc": 0.033920831978321075, "train_loss_llm": 0.0, "grad_norm": 0.05701342225074768, "global_step": 13011, "epoch": 361, "lr": 0.001} +{"train_loss": 0.04310230165719986, "train_loss_bc": 0.04310230165719986, "train_loss_llm": 0.0, "grad_norm": 0.06803244352340698, "global_step": 13012, "epoch": 361, "lr": 0.001} +{"train_loss": 0.043288618326187134, "train_loss_bc": 0.043288618326187134, "train_loss_llm": 0.0, "grad_norm": 0.023825107142329216, "global_step": 13013, "epoch": 361, "lr": 0.001} +{"train_loss": 0.02903483249247074, "train_loss_bc": 0.029033252969384193, "train_loss_llm": 0.00015798211097717285, "grad_norm": 0.03377167508006096, "global_step": 13014, "epoch": 361, "lr": 0.001} +{"train_loss": 0.043887149542570114, "train_loss_bc": 0.043887149542570114, "train_loss_llm": 0.0, "grad_norm": 0.04069836065173149, "global_step": 13015, "epoch": 361, "lr": 0.001} +{"train_loss": 0.04043438285589218, "train_loss_bc": 0.04043438285589218, "train_loss_llm": 0.0, "grad_norm": 0.04805215075612068, "global_step": 13016, "epoch": 361, "lr": 0.001} +{"train_loss": 0.037682320922613144, "train_loss_bc": 0.037682320922613144, "train_loss_llm": 0.0, "grad_norm": 0.01873793825507164, "global_step": 13017, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03534279391169548, "train_loss_bc": 0.03534279391169548, "train_loss_llm": 0.0, "grad_norm": 0.03502544388175011, "global_step": 13018, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03240228816866875, "train_loss_bc": 0.03240228816866875, "train_loss_llm": 0.0, "grad_norm": 0.045615073293447495, "global_step": 13019, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03711310029029846, "train_loss_bc": 0.03711310029029846, "train_loss_llm": 0.0, "grad_norm": 0.055140383541584015, "global_step": 13020, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03940696641802788, "train_loss_bc": 0.03939536586403847, "train_loss_llm": 0.0011601448059082031, "grad_norm": 0.02484869956970215, "global_step": 13021, "epoch": 361, "lr": 0.001} +{"train_loss": 0.034805651754140854, "train_loss_bc": 0.034805651754140854, "train_loss_llm": 0.0, "grad_norm": 0.03638987988233566, "global_step": 13022, "epoch": 361, "lr": 0.001} +{"train_loss": 0.030914977192878723, "train_loss_bc": 0.03090856410562992, "train_loss_llm": 0.0006413459777832031, "grad_norm": 0.047795090824365616, "global_step": 13023, "epoch": 361, "lr": 0.001} +{"train_loss": 0.0333324559032917, "train_loss_bc": 0.033331651240587234, "train_loss_llm": 8.028745651245117e-05, "grad_norm": 0.055660393089056015, "global_step": 13024, "epoch": 361, "lr": 0.001} +{"train_loss": 0.029865305870771408, "train_loss_bc": 0.02986001782119274, "train_loss_llm": 0.0005287230014801025, "grad_norm": 0.017925715073943138, "global_step": 13025, "epoch": 361, "lr": 0.001} +{"train_loss": 0.0396718792617321, "train_loss_bc": 0.0396718792617321, "train_loss_llm": 0.0, "grad_norm": 0.026902543380856514, "global_step": 13026, "epoch": 361, "lr": 0.001} +{"train_loss": 0.032418183982372284, "train_loss_bc": 0.03240765631198883, "train_loss_llm": 0.0010525882244110107, "grad_norm": 0.037279654294252396, "global_step": 13027, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03719590604305267, "train_loss_bc": 0.03719590604305267, "train_loss_llm": 0.0, "grad_norm": 0.04649260267615318, "global_step": 13028, "epoch": 361, "lr": 0.001} +{"train_loss": 0.041597168892621994, "train_loss_bc": 0.041597168892621994, "train_loss_llm": 0.0, "grad_norm": 0.028377369046211243, "global_step": 13029, "epoch": 361, "lr": 0.001} +{"train_loss": 0.037521738559007645, "train_loss_bc": 0.037521738559007645, "train_loss_llm": 0.0, "grad_norm": 0.03505503013730049, "global_step": 13030, "epoch": 361, "lr": 0.001} +{"train_loss": 0.03662276247309314, "train_loss_bc": 0.037666402757167816, "train_loss_llm": 8.589029312133789e-05, "grad_norm": 0.044580038636922836, "global_step": 13031, "epoch": 361, "lr": 0.001, "val_loss": 0.0656890869140625} +{"train_loss": 0.03584405407309532, "train_loss_bc": 0.03584043309092522, "train_loss_llm": 0.00036221742630004883, "grad_norm": 0.05045591667294502, "global_step": 13032, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03315959498286247, "train_loss_bc": 0.03315959498286247, "train_loss_llm": 0.0, "grad_norm": 0.026676254346966743, "global_step": 13033, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03840814530849457, "train_loss_bc": 0.03840814530849457, "train_loss_llm": 0.0, "grad_norm": 0.04290367290377617, "global_step": 13034, "epoch": 362, "lr": 0.001} +{"train_loss": 0.033162496984004974, "train_loss_bc": 0.033162496984004974, "train_loss_llm": 0.0, "grad_norm": 0.051580775529146194, "global_step": 13035, "epoch": 362, "lr": 0.001} +{"train_loss": 0.038136761635541916, "train_loss_bc": 0.038136761635541916, "train_loss_llm": 0.0, "grad_norm": 0.05702917277812958, "global_step": 13036, "epoch": 362, "lr": 0.001} +{"train_loss": 0.04192918539047241, "train_loss_bc": 0.04192918539047241, "train_loss_llm": 0.0, "grad_norm": 0.02846047282218933, "global_step": 13037, "epoch": 362, "lr": 0.001} +{"train_loss": 0.02955128811299801, "train_loss_bc": 0.029551055282354355, "train_loss_llm": 2.327561378479004e-05, "grad_norm": 0.03309919685125351, "global_step": 13038, "epoch": 362, "lr": 0.001} +{"train_loss": 0.040498554706573486, "train_loss_bc": 0.04047971963882446, "train_loss_llm": 0.0018835365772247314, "grad_norm": 0.05466599389910698, "global_step": 13039, "epoch": 362, "lr": 0.001} +{"train_loss": 0.0349602997303009, "train_loss_bc": 0.0349602997303009, "train_loss_llm": 0.0, "grad_norm": 0.06277473270893097, "global_step": 13040, "epoch": 362, "lr": 0.001} +{"train_loss": 0.039191875606775284, "train_loss_bc": 0.039191875606775284, "train_loss_llm": 0.0, "grad_norm": 0.0219391118735075, "global_step": 13041, "epoch": 362, "lr": 0.001} +{"train_loss": 0.04096145182847977, "train_loss_bc": 0.04096145182847977, "train_loss_llm": 0.0, "grad_norm": 0.03013569675385952, "global_step": 13042, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03393624722957611, "train_loss_bc": 0.03393624722957611, "train_loss_llm": 0.0, "grad_norm": 0.04236063361167908, "global_step": 13043, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03956609591841698, "train_loss_bc": 0.03956609591841698, "train_loss_llm": 0.0, "grad_norm": 0.04868541285395622, "global_step": 13044, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03393162786960602, "train_loss_bc": 0.03391857445240021, "train_loss_llm": 0.0013054311275482178, "grad_norm": 0.01959267072379589, "global_step": 13045, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03528967499732971, "train_loss_bc": 0.035285983234643936, "train_loss_llm": 0.0003692209720611572, "grad_norm": 0.031623031944036484, "global_step": 13046, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03919993340969086, "train_loss_bc": 0.039195042103528976, "train_loss_llm": 0.0004892647266387939, "grad_norm": 0.04409211128950119, "global_step": 13047, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03208091855049133, "train_loss_bc": 0.03208091855049133, "train_loss_llm": 0.0, "grad_norm": 0.04891417548060417, "global_step": 13048, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03621376305818558, "train_loss_bc": 0.03621376305818558, "train_loss_llm": 0.0, "grad_norm": 0.021288003772497177, "global_step": 13049, "epoch": 362, "lr": 0.001} +{"train_loss": 0.0301688089966774, "train_loss_bc": 0.0301688089966774, "train_loss_llm": 0.0, "grad_norm": 0.03332728147506714, "global_step": 13050, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03846588358283043, "train_loss_bc": 0.038427986204624176, "train_loss_llm": 0.0037899017333984375, "grad_norm": 0.04134565591812134, "global_step": 13051, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03158416599035263, "train_loss_bc": 0.03158416599035263, "train_loss_llm": 0.0, "grad_norm": 0.05146019533276558, "global_step": 13052, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03048294596374035, "train_loss_bc": 0.03048294596374035, "train_loss_llm": 0.0, "grad_norm": 0.020134437829256058, "global_step": 13053, "epoch": 362, "lr": 0.001} +{"train_loss": 0.033027008175849915, "train_loss_bc": 0.033027008175849915, "train_loss_llm": 0.0, "grad_norm": 0.03093944862484932, "global_step": 13054, "epoch": 362, "lr": 0.001} +{"train_loss": 0.044362910091876984, "train_loss_bc": 0.044362910091876984, "train_loss_llm": 0.0, "grad_norm": 0.04011564329266548, "global_step": 13055, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03261812776327133, "train_loss_bc": 0.03259444609284401, "train_loss_llm": 0.0023681223392486572, "grad_norm": 0.04955342039465904, "global_step": 13056, "epoch": 362, "lr": 0.001} +{"train_loss": 0.039489392191171646, "train_loss_bc": 0.039489392191171646, "train_loss_llm": 0.0, "grad_norm": 0.02270199917256832, "global_step": 13057, "epoch": 362, "lr": 0.001} +{"train_loss": 0.04036131873726845, "train_loss_bc": 0.04036131873726845, "train_loss_llm": 0.0, "grad_norm": 0.04413839802145958, "global_step": 13058, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03722192719578743, "train_loss_bc": 0.03722192719578743, "train_loss_llm": 0.0, "grad_norm": 0.061876315623521805, "global_step": 13059, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03480497747659683, "train_loss_bc": 0.03480497747659683, "train_loss_llm": 0.0, "grad_norm": 0.07148298621177673, "global_step": 13060, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03882332146167755, "train_loss_bc": 0.038809921592473984, "train_loss_llm": 0.0013400614261627197, "grad_norm": 0.02319416217505932, "global_step": 13061, "epoch": 362, "lr": 0.001} +{"train_loss": 0.037608552724123, "train_loss_bc": 0.037608552724123, "train_loss_llm": 0.0, "grad_norm": 0.03886787220835686, "global_step": 13062, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03385625407099724, "train_loss_bc": 0.033848296850919724, "train_loss_llm": 0.0007955431938171387, "grad_norm": 0.04570728912949562, "global_step": 13063, "epoch": 362, "lr": 0.001} +{"train_loss": 0.03249131515622139, "train_loss_bc": 0.03249131515622139, "train_loss_llm": 0.0, "grad_norm": 0.05982012301683426, "global_step": 13064, "epoch": 362, "lr": 0.001} +{"train_loss": 0.04281913489103317, "train_loss_bc": 0.04281913489103317, "train_loss_llm": 0.0, "grad_norm": 0.029360797256231308, "global_step": 13065, "epoch": 362, "lr": 0.001} +{"train_loss": 0.046288665384054184, "train_loss_bc": 0.046288665384054184, "train_loss_llm": 0.0, "grad_norm": 0.038538314402103424, "global_step": 13066, "epoch": 362, "lr": 0.001} +{"train_loss": 0.036493710656132966, "train_loss_bc": 0.03327690437436104, "train_loss_llm": 0.0, "grad_norm": 0.046754736453294754, "global_step": 13067, "epoch": 362, "lr": 0.001, "val_loss": 0.06589287519454956} +{"train_loss": 0.036283839493989944, "train_loss_bc": 0.036283839493989944, "train_loss_llm": 0.0, "grad_norm": 0.053941622376441956, "global_step": 13068, "epoch": 363, "lr": 0.001} +{"train_loss": 0.030168646946549416, "train_loss_bc": 0.030168646946549416, "train_loss_llm": 0.0, "grad_norm": 0.019597943872213364, "global_step": 13069, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03044958785176277, "train_loss_bc": 0.03044958785176277, "train_loss_llm": 0.0, "grad_norm": 0.03769772872328758, "global_step": 13070, "epoch": 363, "lr": 0.001} +{"train_loss": 0.032389212399721146, "train_loss_bc": 0.0323801189661026, "train_loss_llm": 0.000909268856048584, "grad_norm": 0.04659445956349373, "global_step": 13071, "epoch": 363, "lr": 0.001} +{"train_loss": 0.0397537499666214, "train_loss_bc": 0.0397537499666214, "train_loss_llm": 0.0, "grad_norm": 0.05812389776110649, "global_step": 13072, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03803064301609993, "train_loss_bc": 0.03803064301609993, "train_loss_llm": 0.0, "grad_norm": 0.025493158027529716, "global_step": 13073, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03191526606678963, "train_loss_bc": 0.0318780280649662, "train_loss_llm": 0.003723651170730591, "grad_norm": 0.036421388387680054, "global_step": 13074, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03752041608095169, "train_loss_bc": 0.03752041608095169, "train_loss_llm": 0.0, "grad_norm": 0.05000626668334007, "global_step": 13075, "epoch": 363, "lr": 0.001} +{"train_loss": 0.04028702154755592, "train_loss_bc": 0.04028702154755592, "train_loss_llm": 0.0, "grad_norm": 0.06437357515096664, "global_step": 13076, "epoch": 363, "lr": 0.001} +{"train_loss": 0.041820354759693146, "train_loss_bc": 0.04181723669171333, "train_loss_llm": 0.0003116428852081299, "grad_norm": 0.02484363317489624, "global_step": 13077, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03299522027373314, "train_loss_bc": 0.03299522027373314, "train_loss_llm": 0.0, "grad_norm": 0.02998073771595955, "global_step": 13078, "epoch": 363, "lr": 0.001} +{"train_loss": 0.038641106337308884, "train_loss_bc": 0.038641106337308884, "train_loss_llm": 0.0, "grad_norm": 0.034117307513952255, "global_step": 13079, "epoch": 363, "lr": 0.001} +{"train_loss": 0.04363980516791344, "train_loss_bc": 0.04363980516791344, "train_loss_llm": 0.0, "grad_norm": 0.04187152534723282, "global_step": 13080, "epoch": 363, "lr": 0.001} +{"train_loss": 0.041353799402713776, "train_loss_bc": 0.041353799402713776, "train_loss_llm": 0.0, "grad_norm": 0.027013404294848442, "global_step": 13081, "epoch": 363, "lr": 0.001} +{"train_loss": 0.034762486815452576, "train_loss_bc": 0.034762486815452576, "train_loss_llm": 0.0, "grad_norm": 0.03475760295987129, "global_step": 13082, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03913373127579689, "train_loss_bc": 0.03913373127579689, "train_loss_llm": 0.0, "grad_norm": 0.04477829858660698, "global_step": 13083, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03803454339504242, "train_loss_bc": 0.03803454339504242, "train_loss_llm": 0.0, "grad_norm": 0.05553567409515381, "global_step": 13084, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03689856454730034, "train_loss_bc": 0.036891061812639236, "train_loss_llm": 0.0007500946521759033, "grad_norm": 0.025049159303307533, "global_step": 13085, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03629397973418236, "train_loss_bc": 0.03629397973418236, "train_loss_llm": 0.0, "grad_norm": 0.03872894123196602, "global_step": 13086, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03230559825897217, "train_loss_bc": 0.03230278566479683, "train_loss_llm": 0.0002810955047607422, "grad_norm": 0.04502414911985397, "global_step": 13087, "epoch": 363, "lr": 0.001} +{"train_loss": 0.036842625588178635, "train_loss_bc": 0.036842625588178635, "train_loss_llm": 0.0, "grad_norm": 0.056407444179058075, "global_step": 13088, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03268022462725639, "train_loss_bc": 0.032676052302122116, "train_loss_llm": 0.00041735172271728516, "grad_norm": 0.021869195625185966, "global_step": 13089, "epoch": 363, "lr": 0.001} +{"train_loss": 0.039647605270147324, "train_loss_bc": 0.039647605270147324, "train_loss_llm": 0.0, "grad_norm": 0.03585938736796379, "global_step": 13090, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03622286766767502, "train_loss_bc": 0.03622286766767502, "train_loss_llm": 0.0, "grad_norm": 0.04726908355951309, "global_step": 13091, "epoch": 363, "lr": 0.001} +{"train_loss": 0.034633662551641464, "train_loss_bc": 0.034633662551641464, "train_loss_llm": 0.0, "grad_norm": 0.061284370720386505, "global_step": 13092, "epoch": 363, "lr": 0.001} +{"train_loss": 0.04253838211297989, "train_loss_bc": 0.04253838211297989, "train_loss_llm": 0.0, "grad_norm": 0.020476924255490303, "global_step": 13093, "epoch": 363, "lr": 0.001} +{"train_loss": 0.035185132175683975, "train_loss_bc": 0.035185132175683975, "train_loss_llm": 0.0, "grad_norm": 0.03211408108472824, "global_step": 13094, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03504887595772743, "train_loss_bc": 0.03503824770450592, "train_loss_llm": 0.0010628700256347656, "grad_norm": 0.04485874995589256, "global_step": 13095, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03606759384274483, "train_loss_bc": 0.03606759384274483, "train_loss_llm": 0.0, "grad_norm": 0.05548626929521561, "global_step": 13096, "epoch": 363, "lr": 0.001} +{"train_loss": 0.035236943513154984, "train_loss_bc": 0.03522185608744621, "train_loss_llm": 0.0015086531639099121, "grad_norm": 0.025940874591469765, "global_step": 13097, "epoch": 363, "lr": 0.001} +{"train_loss": 0.032834507524967194, "train_loss_bc": 0.032813746482133865, "train_loss_llm": 0.00207597017288208, "grad_norm": 0.03712392970919609, "global_step": 13098, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03677855804562569, "train_loss_bc": 0.036743905395269394, "train_loss_llm": 0.0034653842449188232, "grad_norm": 0.053349509835243225, "global_step": 13099, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03307350352406502, "train_loss_bc": 0.03307350352406502, "train_loss_llm": 0.0, "grad_norm": 0.061413198709487915, "global_step": 13100, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03145362064242363, "train_loss_bc": 0.03145362064242363, "train_loss_llm": 0.0, "grad_norm": 0.017846567556262016, "global_step": 13101, "epoch": 363, "lr": 0.001} +{"train_loss": 0.042965345084667206, "train_loss_bc": 0.042965345084667206, "train_loss_llm": 0.0, "grad_norm": 0.036610256880521774, "global_step": 13102, "epoch": 363, "lr": 0.001} +{"train_loss": 0.03661458809963531, "train_loss_bc": 0.044238150119781494, "train_loss_llm": 0.0, "grad_norm": 0.05081186071038246, "global_step": 13103, "epoch": 363, "lr": 0.001, "val_loss": 0.06589247286319733} +{"train_loss": 0.03131667152047157, "train_loss_bc": 0.03131667152047157, "train_loss_llm": 0.0, "grad_norm": 0.06150103360414505, "global_step": 13104, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03637920320034027, "train_loss_bc": 0.03637920320034027, "train_loss_llm": 0.0, "grad_norm": 0.02532508596777916, "global_step": 13105, "epoch": 364, "lr": 0.001} +{"train_loss": 0.038305286318063736, "train_loss_bc": 0.038305286318063736, "train_loss_llm": 0.0, "grad_norm": 0.03427029028534889, "global_step": 13106, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03441657871007919, "train_loss_bc": 0.03441657871007919, "train_loss_llm": 0.0, "grad_norm": 0.044195506721735, "global_step": 13107, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03332124650478363, "train_loss_bc": 0.03332124650478363, "train_loss_llm": 0.0, "grad_norm": 0.04714279621839523, "global_step": 13108, "epoch": 364, "lr": 0.001} +{"train_loss": 0.040614694356918335, "train_loss_bc": 0.040614694356918335, "train_loss_llm": 0.0, "grad_norm": 0.024657901376485825, "global_step": 13109, "epoch": 364, "lr": 0.001} +{"train_loss": 0.036652822047472, "train_loss_bc": 0.036652822047472, "train_loss_llm": 0.0, "grad_norm": 0.040781762450933456, "global_step": 13110, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03829450160264969, "train_loss_bc": 0.03829450160264969, "train_loss_llm": 0.0, "grad_norm": 0.051047224551439285, "global_step": 13111, "epoch": 364, "lr": 0.001} +{"train_loss": 0.033425651490688324, "train_loss_bc": 0.033425651490688324, "train_loss_llm": 0.0, "grad_norm": 0.05941880866885185, "global_step": 13112, "epoch": 364, "lr": 0.001} +{"train_loss": 0.030618971213698387, "train_loss_bc": 0.030618971213698387, "train_loss_llm": 0.0, "grad_norm": 0.021540170535445213, "global_step": 13113, "epoch": 364, "lr": 0.001} +{"train_loss": 0.031581878662109375, "train_loss_bc": 0.031581878662109375, "train_loss_llm": 0.0, "grad_norm": 0.03544376790523529, "global_step": 13114, "epoch": 364, "lr": 0.001} +{"train_loss": 0.037703730165958405, "train_loss_bc": 0.037703730165958405, "train_loss_llm": 0.0, "grad_norm": 0.05315098911523819, "global_step": 13115, "epoch": 364, "lr": 0.001} +{"train_loss": 0.039689578115940094, "train_loss_bc": 0.039689578115940094, "train_loss_llm": 0.0, "grad_norm": 0.06711231172084808, "global_step": 13116, "epoch": 364, "lr": 0.001} +{"train_loss": 0.0368831641972065, "train_loss_bc": 0.0368831641972065, "train_loss_llm": 0.0, "grad_norm": 0.023379284888505936, "global_step": 13117, "epoch": 364, "lr": 0.001} +{"train_loss": 0.037569426000118256, "train_loss_bc": 0.037569426000118256, "train_loss_llm": 0.0, "grad_norm": 0.037173621356487274, "global_step": 13118, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03785770386457443, "train_loss_bc": 0.037857405841350555, "train_loss_llm": 2.9742717742919922e-05, "grad_norm": 0.05076569318771362, "global_step": 13119, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03574107214808464, "train_loss_bc": 0.03574061021208763, "train_loss_llm": 4.622340202331543e-05, "grad_norm": 0.05961528792977333, "global_step": 13120, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03414272516965866, "train_loss_bc": 0.03414272516965866, "train_loss_llm": 0.0, "grad_norm": 0.023796321824193, "global_step": 13121, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03526143357157707, "train_loss_bc": 0.03526143357157707, "train_loss_llm": 0.0, "grad_norm": 0.03819485008716583, "global_step": 13122, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03725343197584152, "train_loss_bc": 0.03725343197584152, "train_loss_llm": 0.0, "grad_norm": 0.051437657326459885, "global_step": 13123, "epoch": 364, "lr": 0.001} +{"train_loss": 0.030205929651856422, "train_loss_bc": 0.03019976057112217, "train_loss_llm": 0.0006168484687805176, "grad_norm": 0.06132598966360092, "global_step": 13124, "epoch": 364, "lr": 0.001} +{"train_loss": 0.038424279540777206, "train_loss_bc": 0.03840332105755806, "train_loss_llm": 0.0020957589149475098, "grad_norm": 0.029188843443989754, "global_step": 13125, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03272898495197296, "train_loss_bc": 0.03272898495197296, "train_loss_llm": 0.0, "grad_norm": 0.039095520973205566, "global_step": 13126, "epoch": 364, "lr": 0.001} +{"train_loss": 0.039576444774866104, "train_loss_bc": 0.039576444774866104, "train_loss_llm": 0.0, "grad_norm": 0.04352054372429848, "global_step": 13127, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03711896017193794, "train_loss_bc": 0.03711896017193794, "train_loss_llm": 0.0, "grad_norm": 0.05868088826537132, "global_step": 13128, "epoch": 364, "lr": 0.001} +{"train_loss": 0.04327317327260971, "train_loss_bc": 0.04327317327260971, "train_loss_llm": 0.0, "grad_norm": 0.02425529435276985, "global_step": 13129, "epoch": 364, "lr": 0.001} +{"train_loss": 0.04199742153286934, "train_loss_bc": 0.04199742153286934, "train_loss_llm": 0.0, "grad_norm": 0.03957775980234146, "global_step": 13130, "epoch": 364, "lr": 0.001} +{"train_loss": 0.040738269686698914, "train_loss_bc": 0.040738269686698914, "train_loss_llm": 0.0, "grad_norm": 0.054328273981809616, "global_step": 13131, "epoch": 364, "lr": 0.001} +{"train_loss": 0.040277499705553055, "train_loss_bc": 0.040277499705553055, "train_loss_llm": 0.0, "grad_norm": 0.06724280118942261, "global_step": 13132, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03551654890179634, "train_loss_bc": 0.03551654890179634, "train_loss_llm": 0.0, "grad_norm": 0.0279363002628088, "global_step": 13133, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03157482296228409, "train_loss_bc": 0.031557850539684296, "train_loss_llm": 0.0016970932483673096, "grad_norm": 0.03937048092484474, "global_step": 13134, "epoch": 364, "lr": 0.001} +{"train_loss": 0.0419452004134655, "train_loss_bc": 0.0419452004134655, "train_loss_llm": 0.0, "grad_norm": 0.06056930869817734, "global_step": 13135, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03199798986315727, "train_loss_bc": 0.03199798986315727, "train_loss_llm": 0.0, "grad_norm": 0.0645650178194046, "global_step": 13136, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03444958105683327, "train_loss_bc": 0.034412965178489685, "train_loss_llm": 0.0036616623401641846, "grad_norm": 0.020731570199131966, "global_step": 13137, "epoch": 364, "lr": 0.001} +{"train_loss": 0.03767694532871246, "train_loss_bc": 0.037676747888326645, "train_loss_llm": 1.963973045349121e-05, "grad_norm": 0.034948721528053284, "global_step": 13138, "epoch": 364, "lr": 0.001} +{"train_loss": 0.0366027502136098, "train_loss_bc": 0.04316718503832817, "train_loss_llm": 0.0, "grad_norm": 0.03775559738278389, "global_step": 13139, "epoch": 364, "lr": 0.001, "val_loss": 0.06678547710180283} +{"train_loss": 0.03663620352745056, "train_loss_bc": 0.03662244230508804, "train_loss_llm": 0.0013762712478637695, "grad_norm": 0.04239198938012123, "global_step": 13140, "epoch": 365, "lr": 0.001} +{"train_loss": 0.038208913058042526, "train_loss_bc": 0.038208913058042526, "train_loss_llm": 0.0, "grad_norm": 0.025229966267943382, "global_step": 13141, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03947697952389717, "train_loss_bc": 0.03947453573346138, "train_loss_llm": 0.0002442896366119385, "grad_norm": 0.04109674692153931, "global_step": 13142, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03742524981498718, "train_loss_bc": 0.03742524981498718, "train_loss_llm": 0.0, "grad_norm": 0.0630890280008316, "global_step": 13143, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03227466344833374, "train_loss_bc": 0.032271843403577805, "train_loss_llm": 0.00028207898139953613, "grad_norm": 0.08630437403917313, "global_step": 13144, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03262590989470482, "train_loss_bc": 0.03261839225888252, "train_loss_llm": 0.000751882791519165, "grad_norm": 0.020079918205738068, "global_step": 13145, "epoch": 365, "lr": 0.001} +{"train_loss": 0.035703353583812714, "train_loss_bc": 0.03570149093866348, "train_loss_llm": 0.0001862943172454834, "grad_norm": 0.03607668727636337, "global_step": 13146, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03300640359520912, "train_loss_bc": 0.03298075124621391, "train_loss_llm": 0.002565234899520874, "grad_norm": 0.05907055735588074, "global_step": 13147, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03942941501736641, "train_loss_bc": 0.03942941501736641, "train_loss_llm": 0.0, "grad_norm": 0.07690297067165375, "global_step": 13148, "epoch": 365, "lr": 0.001} +{"train_loss": 0.030015481635928154, "train_loss_bc": 0.030015481635928154, "train_loss_llm": 0.0, "grad_norm": 0.03804267570376396, "global_step": 13149, "epoch": 365, "lr": 0.001} +{"train_loss": 0.037275299429893494, "train_loss_bc": 0.03727490082383156, "train_loss_llm": 3.984570503234863e-05, "grad_norm": 0.05091352015733719, "global_step": 13150, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03147662803530693, "train_loss_bc": 0.03147662803530693, "train_loss_llm": 0.0, "grad_norm": 0.06417656689882278, "global_step": 13151, "epoch": 365, "lr": 0.001} +{"train_loss": 0.029999209567904472, "train_loss_bc": 0.029999209567904472, "train_loss_llm": 0.0, "grad_norm": 0.07439979910850525, "global_step": 13152, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03566597029566765, "train_loss_bc": 0.035656288266181946, "train_loss_llm": 0.0009682178497314453, "grad_norm": 0.02285136841237545, "global_step": 13153, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03395264223217964, "train_loss_bc": 0.03395264223217964, "train_loss_llm": 0.0, "grad_norm": 0.033975712954998016, "global_step": 13154, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03298160061240196, "train_loss_bc": 0.03298160061240196, "train_loss_llm": 0.0, "grad_norm": 0.042183659970760345, "global_step": 13155, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03200936317443848, "train_loss_bc": 0.03200936317443848, "train_loss_llm": 0.0, "grad_norm": 0.05333617329597473, "global_step": 13156, "epoch": 365, "lr": 0.001} +{"train_loss": 0.04168277978897095, "train_loss_bc": 0.04168277978897095, "train_loss_llm": 0.0, "grad_norm": 0.023364458233118057, "global_step": 13157, "epoch": 365, "lr": 0.001} +{"train_loss": 0.032078880816698074, "train_loss_bc": 0.032078880816698074, "train_loss_llm": 0.0, "grad_norm": 0.03194596990942955, "global_step": 13158, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03910020366311073, "train_loss_bc": 0.03910020366311073, "train_loss_llm": 0.0, "grad_norm": 0.03865085914731026, "global_step": 13159, "epoch": 365, "lr": 0.001} +{"train_loss": 0.039479274302721024, "train_loss_bc": 0.039479274302721024, "train_loss_llm": 0.0, "grad_norm": 0.04589680954813957, "global_step": 13160, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03548239916563034, "train_loss_bc": 0.03548239916563034, "train_loss_llm": 0.0, "grad_norm": 0.024258388206362724, "global_step": 13161, "epoch": 365, "lr": 0.001} +{"train_loss": 0.04316270723938942, "train_loss_bc": 0.04316270723938942, "train_loss_llm": 0.0, "grad_norm": 0.0430673286318779, "global_step": 13162, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03443462401628494, "train_loss_bc": 0.03443461284041405, "train_loss_llm": 1.2814998626708984e-06, "grad_norm": 0.04690966755151749, "global_step": 13163, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03815550357103348, "train_loss_bc": 0.03815550357103348, "train_loss_llm": 0.0, "grad_norm": 0.04734406620264053, "global_step": 13164, "epoch": 365, "lr": 0.001} +{"train_loss": 0.045713428407907486, "train_loss_bc": 0.045713428407907486, "train_loss_llm": 0.0, "grad_norm": 0.02147788368165493, "global_step": 13165, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03899325057864189, "train_loss_bc": 0.03899325057864189, "train_loss_llm": 0.0, "grad_norm": 0.03296812251210213, "global_step": 13166, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03742991387844086, "train_loss_bc": 0.03741460666060448, "train_loss_llm": 0.0015305578708648682, "grad_norm": 0.04258742928504944, "global_step": 13167, "epoch": 365, "lr": 0.001} +{"train_loss": 0.04039572924375534, "train_loss_bc": 0.04039141163229942, "train_loss_llm": 0.0004316568374633789, "grad_norm": 0.051231496036052704, "global_step": 13168, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03971530497074127, "train_loss_bc": 0.03971530497074127, "train_loss_llm": 0.0, "grad_norm": 0.03336084261536598, "global_step": 13169, "epoch": 365, "lr": 0.001} +{"train_loss": 0.036080896854400635, "train_loss_bc": 0.036076538264751434, "train_loss_llm": 0.0004356801509857178, "grad_norm": 0.05198473855853081, "global_step": 13170, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03941631689667702, "train_loss_bc": 0.03941631689667702, "train_loss_llm": 0.0, "grad_norm": 0.06822985410690308, "global_step": 13171, "epoch": 365, "lr": 0.001} +{"train_loss": 0.04126697778701782, "train_loss_bc": 0.04126697778701782, "train_loss_llm": 0.0, "grad_norm": 0.08645161986351013, "global_step": 13172, "epoch": 365, "lr": 0.001} +{"train_loss": 0.036954112350940704, "train_loss_bc": 0.036942604929208755, "train_loss_llm": 0.0011506080627441406, "grad_norm": 0.025066301226615906, "global_step": 13173, "epoch": 365, "lr": 0.001} +{"train_loss": 0.034634288400411606, "train_loss_bc": 0.03463327884674072, "train_loss_llm": 0.00010082125663757324, "grad_norm": 0.03902003541588783, "global_step": 13174, "epoch": 365, "lr": 0.001} +{"train_loss": 0.03665387713246875, "train_loss_bc": 0.03719589486718178, "train_loss_llm": 0.00038042664527893066, "grad_norm": 0.04762634262442589, "global_step": 13175, "epoch": 365, "lr": 0.001, "train/cumulative_reward": 54.390127627027404, "train/mean_score": 11.810132526228085, "train/success_rate": 0.3, "test/cumulative_reward": 30.867210683492758, "test/mean_score": 6.892511110234187, "test/success_rate": 0.16, "val_loss": 0.06521791219711304, "train_action_mse_error": 0.039430972188711166} +{"train_loss": 0.04430186003446579, "train_loss_bc": 0.04430186003446579, "train_loss_llm": 0.0, "grad_norm": 0.06765148788690567, "global_step": 13176, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03420546278357506, "train_loss_bc": 0.03420546278357506, "train_loss_llm": 0.0, "grad_norm": 0.027276694774627686, "global_step": 13177, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03824024647474289, "train_loss_bc": 0.03823528811335564, "train_loss_llm": 0.0004957318305969238, "grad_norm": 0.04795728251338005, "global_step": 13178, "epoch": 366, "lr": 0.001} +{"train_loss": 0.0350838266313076, "train_loss_bc": 0.0350838266313076, "train_loss_llm": 0.0, "grad_norm": 0.05432308092713356, "global_step": 13179, "epoch": 366, "lr": 0.001} +{"train_loss": 0.031048396602272987, "train_loss_bc": 0.031048046424984932, "train_loss_llm": 3.501772880554199e-05, "grad_norm": 0.06238744780421257, "global_step": 13180, "epoch": 366, "lr": 0.001} +{"train_loss": 0.037649966776371, "train_loss_bc": 0.037649966776371, "train_loss_llm": 0.0, "grad_norm": 0.025913767516613007, "global_step": 13181, "epoch": 366, "lr": 0.001} +{"train_loss": 0.036470163613557816, "train_loss_bc": 0.036470163613557816, "train_loss_llm": 0.0, "grad_norm": 0.03489420935511589, "global_step": 13182, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03830534964799881, "train_loss_bc": 0.03830534964799881, "train_loss_llm": 0.0, "grad_norm": 0.04967144504189491, "global_step": 13183, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03685552626848221, "train_loss_bc": 0.03685552626848221, "train_loss_llm": 0.0, "grad_norm": 0.06040983647108078, "global_step": 13184, "epoch": 366, "lr": 0.001} +{"train_loss": 0.039961423724889755, "train_loss_bc": 0.039948903024196625, "train_loss_llm": 0.0012520253658294678, "grad_norm": 0.019652485847473145, "global_step": 13185, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03853931277990341, "train_loss_bc": 0.03853555768728256, "train_loss_llm": 0.00037541985511779785, "grad_norm": 0.032138656824827194, "global_step": 13186, "epoch": 366, "lr": 0.001} +{"train_loss": 0.027840059250593185, "train_loss_bc": 0.02782941237092018, "train_loss_llm": 0.0010646581649780273, "grad_norm": 0.03868468478322029, "global_step": 13187, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03049364686012268, "train_loss_bc": 0.03049364686012268, "train_loss_llm": 0.0, "grad_norm": 0.04531889036297798, "global_step": 13188, "epoch": 366, "lr": 0.001} +{"train_loss": 0.035616789013147354, "train_loss_bc": 0.035616789013147354, "train_loss_llm": 0.0, "grad_norm": 0.022785138338804245, "global_step": 13189, "epoch": 366, "lr": 0.001} +{"train_loss": 0.038638852536678314, "train_loss_bc": 0.038637641817331314, "train_loss_llm": 0.00012120604515075684, "grad_norm": 0.041607245802879333, "global_step": 13190, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03271820768713951, "train_loss_bc": 0.03271820768713951, "train_loss_llm": 0.0, "grad_norm": 0.04707830026745796, "global_step": 13191, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03129700943827629, "train_loss_bc": 0.03129700943827629, "train_loss_llm": 0.0, "grad_norm": 0.059221506118774414, "global_step": 13192, "epoch": 366, "lr": 0.001} +{"train_loss": 0.04349388927221298, "train_loss_bc": 0.04349388927221298, "train_loss_llm": 0.0, "grad_norm": 0.026374826207756996, "global_step": 13193, "epoch": 366, "lr": 0.001} +{"train_loss": 0.04442456364631653, "train_loss_bc": 0.044387757778167725, "train_loss_llm": 0.0036807358264923096, "grad_norm": 0.041757259517908096, "global_step": 13194, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03814477100968361, "train_loss_bc": 0.03814477100968361, "train_loss_llm": 0.0, "grad_norm": 0.055201541632413864, "global_step": 13195, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03295128047466278, "train_loss_bc": 0.03295128047466278, "train_loss_llm": 0.0, "grad_norm": 0.06283079832792282, "global_step": 13196, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03708402439951897, "train_loss_bc": 0.03708402439951897, "train_loss_llm": 0.0, "grad_norm": 0.025308912619948387, "global_step": 13197, "epoch": 366, "lr": 0.001} +{"train_loss": 0.04292752593755722, "train_loss_bc": 0.04292009025812149, "train_loss_llm": 0.0007434487342834473, "grad_norm": 0.044757258147001266, "global_step": 13198, "epoch": 366, "lr": 0.001} +{"train_loss": 0.04507235065102577, "train_loss_bc": 0.04507235065102577, "train_loss_llm": 0.0, "grad_norm": 0.056610770523548126, "global_step": 13199, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03529313579201698, "train_loss_bc": 0.035290829837322235, "train_loss_llm": 0.00023049116134643555, "grad_norm": 0.07307510077953339, "global_step": 13200, "epoch": 366, "lr": 0.001} +{"train_loss": 0.041184283792972565, "train_loss_bc": 0.041184283792972565, "train_loss_llm": 0.0, "grad_norm": 0.02589835226535797, "global_step": 13201, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03912173584103584, "train_loss_bc": 0.03910652920603752, "train_loss_llm": 0.0015207529067993164, "grad_norm": 0.038628432899713516, "global_step": 13202, "epoch": 366, "lr": 0.001} +{"train_loss": 0.029987607151269913, "train_loss_bc": 0.029987607151269913, "train_loss_llm": 0.0, "grad_norm": 0.044072866439819336, "global_step": 13203, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03337974101305008, "train_loss_bc": 0.033369287848472595, "train_loss_llm": 0.0010453462600708008, "grad_norm": 0.05286828801035881, "global_step": 13204, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03629155084490776, "train_loss_bc": 0.03629155084490776, "train_loss_llm": 0.0, "grad_norm": 0.017503095790743828, "global_step": 13205, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03415853530168533, "train_loss_bc": 0.034141041338443756, "train_loss_llm": 0.0017493963241577148, "grad_norm": 0.03225450962781906, "global_step": 13206, "epoch": 366, "lr": 0.001} +{"train_loss": 0.029909735545516014, "train_loss_bc": 0.029909735545516014, "train_loss_llm": 0.0, "grad_norm": 0.04053788259625435, "global_step": 13207, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03573944792151451, "train_loss_bc": 0.03573944792151451, "train_loss_llm": 0.0, "grad_norm": 0.04994124919176102, "global_step": 13208, "epoch": 366, "lr": 0.001} +{"train_loss": 0.0359492301940918, "train_loss_bc": 0.035929299890995026, "train_loss_llm": 0.0019929707050323486, "grad_norm": 0.0238376185297966, "global_step": 13209, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03757426142692566, "train_loss_bc": 0.03757426142692566, "train_loss_llm": 0.0, "grad_norm": 0.04128863662481308, "global_step": 13210, "epoch": 366, "lr": 0.001} +{"train_loss": 0.03649924364354876, "train_loss_bc": 0.034019000828266144, "train_loss_llm": 0.0, "grad_norm": 0.051955390721559525, "global_step": 13211, "epoch": 366, "lr": 0.001, "val_loss": 0.06536148488521576} +{"train_loss": 0.0316678024828434, "train_loss_bc": 0.0316678024828434, "train_loss_llm": 0.0, "grad_norm": 0.056990571320056915, "global_step": 13212, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03526834025979042, "train_loss_bc": 0.03525050729513168, "train_loss_llm": 0.0017832815647125244, "grad_norm": 0.022659290581941605, "global_step": 13213, "epoch": 367, "lr": 0.001} +{"train_loss": 0.039314500987529755, "train_loss_bc": 0.03931426256895065, "train_loss_llm": 2.3663043975830078e-05, "grad_norm": 0.04334045946598053, "global_step": 13214, "epoch": 367, "lr": 0.001} +{"train_loss": 0.031344398856163025, "train_loss_bc": 0.031344398856163025, "train_loss_llm": 0.0, "grad_norm": 0.05917395278811455, "global_step": 13215, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03225857764482498, "train_loss_bc": 0.032256051898002625, "train_loss_llm": 0.0002525746822357178, "grad_norm": 0.06154836341738701, "global_step": 13216, "epoch": 367, "lr": 0.001} +{"train_loss": 0.034389376640319824, "train_loss_bc": 0.034389376640319824, "train_loss_llm": 0.0, "grad_norm": 0.027930010110139847, "global_step": 13217, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03889157250523567, "train_loss_bc": 0.03889157250523567, "train_loss_llm": 0.0, "grad_norm": 0.0450015515089035, "global_step": 13218, "epoch": 367, "lr": 0.001} +{"train_loss": 0.030335767194628716, "train_loss_bc": 0.030335767194628716, "train_loss_llm": 0.0, "grad_norm": 0.0508158877491951, "global_step": 13219, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03302988409996033, "train_loss_bc": 0.03302988409996033, "train_loss_llm": 0.0, "grad_norm": 0.06696559488773346, "global_step": 13220, "epoch": 367, "lr": 0.001} +{"train_loss": 0.030788004398345947, "train_loss_bc": 0.030788004398345947, "train_loss_llm": 0.0, "grad_norm": 0.02587948366999626, "global_step": 13221, "epoch": 367, "lr": 0.001} +{"train_loss": 0.038995083421468735, "train_loss_bc": 0.038995083421468735, "train_loss_llm": 0.0, "grad_norm": 0.037816908210515976, "global_step": 13222, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03737308830022812, "train_loss_bc": 0.03736814484000206, "train_loss_llm": 0.0004941821098327637, "grad_norm": 0.04918888211250305, "global_step": 13223, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03680184483528137, "train_loss_bc": 0.03679816052317619, "train_loss_llm": 0.00036829710006713867, "grad_norm": 0.05969720333814621, "global_step": 13224, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03963490575551987, "train_loss_bc": 0.03963490575551987, "train_loss_llm": 0.0, "grad_norm": 0.02731654979288578, "global_step": 13225, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03228619694709778, "train_loss_bc": 0.03227148950099945, "train_loss_llm": 0.001470804214477539, "grad_norm": 0.03916867822408676, "global_step": 13226, "epoch": 367, "lr": 0.001} +{"train_loss": 0.038065068423748016, "train_loss_bc": 0.03806016966700554, "train_loss_llm": 0.0004897415637969971, "grad_norm": 0.04709698259830475, "global_step": 13227, "epoch": 367, "lr": 0.001} +{"train_loss": 0.0438346266746521, "train_loss_bc": 0.0438346266746521, "train_loss_llm": 0.0, "grad_norm": 0.06213955953717232, "global_step": 13228, "epoch": 367, "lr": 0.001} +{"train_loss": 0.040156010538339615, "train_loss_bc": 0.040156010538339615, "train_loss_llm": 0.0, "grad_norm": 0.022586051374673843, "global_step": 13229, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03748433664441109, "train_loss_bc": 0.03748433664441109, "train_loss_llm": 0.0, "grad_norm": 0.032489534467458725, "global_step": 13230, "epoch": 367, "lr": 0.001} +{"train_loss": 0.036872778087854385, "train_loss_bc": 0.036872778087854385, "train_loss_llm": 0.0, "grad_norm": 0.04063000530004501, "global_step": 13231, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03457184135913849, "train_loss_bc": 0.0345686599612236, "train_loss_llm": 0.000317990779876709, "grad_norm": 0.04720703139901161, "global_step": 13232, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03937298804521561, "train_loss_bc": 0.03937298804521561, "train_loss_llm": 0.0, "grad_norm": 0.019223297014832497, "global_step": 13233, "epoch": 367, "lr": 0.001} +{"train_loss": 0.041594721376895905, "train_loss_bc": 0.04158739373087883, "train_loss_llm": 0.0007326900959014893, "grad_norm": 0.03507315739989281, "global_step": 13234, "epoch": 367, "lr": 0.001} +{"train_loss": 0.034551262855529785, "train_loss_bc": 0.034551262855529785, "train_loss_llm": 0.0, "grad_norm": 0.04862610623240471, "global_step": 13235, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03476054221391678, "train_loss_bc": 0.03475746139883995, "train_loss_llm": 0.00030797719955444336, "grad_norm": 0.058556605130434036, "global_step": 13236, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03166072815656662, "train_loss_bc": 0.03166072815656662, "train_loss_llm": 0.0, "grad_norm": 0.0195918045938015, "global_step": 13237, "epoch": 367, "lr": 0.001} +{"train_loss": 0.04160681366920471, "train_loss_bc": 0.04160492494702339, "train_loss_llm": 0.0001887679100036621, "grad_norm": 0.03073597140610218, "global_step": 13238, "epoch": 367, "lr": 0.001} +{"train_loss": 0.040638457983732224, "train_loss_bc": 0.04063211753964424, "train_loss_llm": 0.0006340444087982178, "grad_norm": 0.0431862473487854, "global_step": 13239, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03059466928243637, "train_loss_bc": 0.03059466928243637, "train_loss_llm": 0.0, "grad_norm": 0.05203688144683838, "global_step": 13240, "epoch": 367, "lr": 0.001} +{"train_loss": 0.039883654564619064, "train_loss_bc": 0.039883654564619064, "train_loss_llm": 0.0, "grad_norm": 0.026296891272068024, "global_step": 13241, "epoch": 367, "lr": 0.001} +{"train_loss": 0.04240880906581879, "train_loss_bc": 0.04240880906581879, "train_loss_llm": 0.0, "grad_norm": 0.04007076099514961, "global_step": 13242, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03335456922650337, "train_loss_bc": 0.03335456922650337, "train_loss_llm": 0.0, "grad_norm": 0.053524330258369446, "global_step": 13243, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03989188000559807, "train_loss_bc": 0.03989105299115181, "train_loss_llm": 8.26120376586914e-05, "grad_norm": 0.07314221560955048, "global_step": 13244, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03821451961994171, "train_loss_bc": 0.03821451961994171, "train_loss_llm": 0.0, "grad_norm": 0.025996297597885132, "global_step": 13245, "epoch": 367, "lr": 0.001} +{"train_loss": 0.037613458931446075, "train_loss_bc": 0.037613458931446075, "train_loss_llm": 0.0, "grad_norm": 0.04376503825187683, "global_step": 13246, "epoch": 367, "lr": 0.001} +{"train_loss": 0.03644695458933711, "train_loss_bc": 0.03257928416132927, "train_loss_llm": 0.0, "grad_norm": 0.05500180274248123, "global_step": 13247, "epoch": 367, "lr": 0.001, "val_loss": 0.06634170562028885} +{"train_loss": 0.029431229457259178, "train_loss_bc": 0.02942916750907898, "train_loss_llm": 0.00020611286163330078, "grad_norm": 0.05522695928812027, "global_step": 13248, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03297935053706169, "train_loss_bc": 0.03297935053706169, "train_loss_llm": 0.0, "grad_norm": 0.022388223558664322, "global_step": 13249, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03430335596203804, "train_loss_bc": 0.03430335596203804, "train_loss_llm": 0.0, "grad_norm": 0.030653782188892365, "global_step": 13250, "epoch": 368, "lr": 0.001} +{"train_loss": 0.035254351794719696, "train_loss_bc": 0.035254351794719696, "train_loss_llm": 0.0, "grad_norm": 0.040835924446582794, "global_step": 13251, "epoch": 368, "lr": 0.001} +{"train_loss": 0.034837979823350906, "train_loss_bc": 0.034837979823350906, "train_loss_llm": 0.0, "grad_norm": 0.043442703783512115, "global_step": 13252, "epoch": 368, "lr": 0.001} +{"train_loss": 0.044420190155506134, "train_loss_bc": 0.04440741613507271, "train_loss_llm": 0.001277536153793335, "grad_norm": 0.029171114787459373, "global_step": 13253, "epoch": 368, "lr": 0.001} +{"train_loss": 0.034714944660663605, "train_loss_bc": 0.034701038151979446, "train_loss_llm": 0.0013905465602874756, "grad_norm": 0.03561631962656975, "global_step": 13254, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03857385367155075, "train_loss_bc": 0.038554877042770386, "train_loss_llm": 0.0018977522850036621, "grad_norm": 0.04719143360853195, "global_step": 13255, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03125770762562752, "train_loss_bc": 0.03121207281947136, "train_loss_llm": 0.004563331604003906, "grad_norm": 0.05718729645013809, "global_step": 13256, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03948305547237396, "train_loss_bc": 0.03948305547237396, "train_loss_llm": 0.0, "grad_norm": 0.02385166473686695, "global_step": 13257, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03680123761296272, "train_loss_bc": 0.036776501685380936, "train_loss_llm": 0.0024735629558563232, "grad_norm": 0.03246583417057991, "global_step": 13258, "epoch": 368, "lr": 0.001} +{"train_loss": 0.04041263461112976, "train_loss_bc": 0.040406495332717896, "train_loss_llm": 0.0006138980388641357, "grad_norm": 0.03752144053578377, "global_step": 13259, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03650728613138199, "train_loss_bc": 0.03650728613138199, "train_loss_llm": 0.0, "grad_norm": 0.050175491720438004, "global_step": 13260, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03739112243056297, "train_loss_bc": 0.03739112243056297, "train_loss_llm": 0.0, "grad_norm": 0.021946223452687263, "global_step": 13261, "epoch": 368, "lr": 0.001} +{"train_loss": 0.037873707711696625, "train_loss_bc": 0.03787190839648247, "train_loss_llm": 0.0001799166202545166, "grad_norm": 0.03288399055600166, "global_step": 13262, "epoch": 368, "lr": 0.001} +{"train_loss": 0.036547016352415085, "train_loss_bc": 0.036533888429403305, "train_loss_llm": 0.0013128221035003662, "grad_norm": 0.04918408393859863, "global_step": 13263, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03768811374902725, "train_loss_bc": 0.03768356144428253, "train_loss_llm": 0.0004552006721496582, "grad_norm": 0.07548096776008606, "global_step": 13264, "epoch": 368, "lr": 0.001} +{"train_loss": 0.041306719183921814, "train_loss_bc": 0.041306719183921814, "train_loss_llm": 0.0, "grad_norm": 0.022795913740992546, "global_step": 13265, "epoch": 368, "lr": 0.001} +{"train_loss": 0.04035042226314545, "train_loss_bc": 0.04035042226314545, "train_loss_llm": 0.0, "grad_norm": 0.03601422905921936, "global_step": 13266, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03839949890971184, "train_loss_bc": 0.03839949890971184, "train_loss_llm": 0.0, "grad_norm": 0.045764680951833725, "global_step": 13267, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03401900455355644, "train_loss_bc": 0.03401900455355644, "train_loss_llm": 0.0, "grad_norm": 0.05776119604706764, "global_step": 13268, "epoch": 368, "lr": 0.001} +{"train_loss": 0.030938830226659775, "train_loss_bc": 0.030938830226659775, "train_loss_llm": 0.0, "grad_norm": 0.018136385828256607, "global_step": 13269, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03806322067975998, "train_loss_bc": 0.03806322067975998, "train_loss_llm": 0.0, "grad_norm": 0.028594734147191048, "global_step": 13270, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03795348107814789, "train_loss_bc": 0.03795348107814789, "train_loss_llm": 0.0, "grad_norm": 0.0389997623860836, "global_step": 13271, "epoch": 368, "lr": 0.001} +{"train_loss": 0.036159999668598175, "train_loss_bc": 0.036159999668598175, "train_loss_llm": 0.0, "grad_norm": 0.047492749989032745, "global_step": 13272, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03377452865242958, "train_loss_bc": 0.03377452865242958, "train_loss_llm": 0.0, "grad_norm": 0.024138029664754868, "global_step": 13273, "epoch": 368, "lr": 0.001} +{"train_loss": 0.0342855378985405, "train_loss_bc": 0.0342855378985405, "train_loss_llm": 0.0, "grad_norm": 0.036524102091789246, "global_step": 13274, "epoch": 368, "lr": 0.001} +{"train_loss": 0.0445488803088665, "train_loss_bc": 0.0445488803088665, "train_loss_llm": 0.0, "grad_norm": 0.05101219564676285, "global_step": 13275, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03728129342198372, "train_loss_bc": 0.03728129342198372, "train_loss_llm": 0.0, "grad_norm": 0.06169653311371803, "global_step": 13276, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03363718464970589, "train_loss_bc": 0.03363718464970589, "train_loss_llm": 0.0, "grad_norm": 0.02271660417318344, "global_step": 13277, "epoch": 368, "lr": 0.001} +{"train_loss": 0.037900663912296295, "train_loss_bc": 0.037900663912296295, "train_loss_llm": 0.0, "grad_norm": 0.04087807238101959, "global_step": 13278, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03962825983762741, "train_loss_bc": 0.03962825983762741, "train_loss_llm": 0.0, "grad_norm": 0.054442938417196274, "global_step": 13279, "epoch": 368, "lr": 0.001} +{"train_loss": 0.030242204666137695, "train_loss_bc": 0.030242204666137695, "train_loss_llm": 0.0, "grad_norm": 0.06453493982553482, "global_step": 13280, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03511714190244675, "train_loss_bc": 0.03511714190244675, "train_loss_llm": 0.0, "grad_norm": 0.020893188193440437, "global_step": 13281, "epoch": 368, "lr": 0.001} +{"train_loss": 0.039130404591560364, "train_loss_bc": 0.039130404591560364, "train_loss_llm": 0.0, "grad_norm": 0.03254593908786774, "global_step": 13282, "epoch": 368, "lr": 0.001} +{"train_loss": 0.03630122889040245, "train_loss_bc": 0.02562982589006424, "train_loss_llm": 0.0, "grad_norm": 0.03907150775194168, "global_step": 13283, "epoch": 368, "lr": 0.001, "val_loss": 0.06647486984729767} +{"train_loss": 0.03321412578225136, "train_loss_bc": 0.03321412578225136, "train_loss_llm": 0.0, "grad_norm": 0.04981385916471481, "global_step": 13284, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03578828275203705, "train_loss_bc": 0.035781122744083405, "train_loss_llm": 0.0007161498069763184, "grad_norm": 0.020868830382823944, "global_step": 13285, "epoch": 369, "lr": 0.001} +{"train_loss": 0.04027198255062103, "train_loss_bc": 0.04027082771062851, "train_loss_llm": 0.00011533498764038086, "grad_norm": 0.03138912096619606, "global_step": 13286, "epoch": 369, "lr": 0.001} +{"train_loss": 0.037006113678216934, "train_loss_bc": 0.03700452297925949, "train_loss_llm": 0.00015908479690551758, "grad_norm": 0.03698459267616272, "global_step": 13287, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03759302943944931, "train_loss_bc": 0.03759302943944931, "train_loss_llm": 0.0, "grad_norm": 0.04226921871304512, "global_step": 13288, "epoch": 369, "lr": 0.001} +{"train_loss": 0.041606009006500244, "train_loss_bc": 0.041606009006500244, "train_loss_llm": 0.0, "grad_norm": 0.01832614839076996, "global_step": 13289, "epoch": 369, "lr": 0.001} +{"train_loss": 0.029234101995825768, "train_loss_bc": 0.029234101995825768, "train_loss_llm": 0.0, "grad_norm": 0.02675098367035389, "global_step": 13290, "epoch": 369, "lr": 0.001} +{"train_loss": 0.031948838382959366, "train_loss_bc": 0.031948838382959366, "train_loss_llm": 0.0, "grad_norm": 0.033610302954912186, "global_step": 13291, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03625640645623207, "train_loss_bc": 0.03624853119254112, "train_loss_llm": 0.0007875263690948486, "grad_norm": 0.04287402704358101, "global_step": 13292, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03360762819647789, "train_loss_bc": 0.033607397228479385, "train_loss_llm": 2.3037195205688477e-05, "grad_norm": 0.028614958748221397, "global_step": 13293, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03232841566205025, "train_loss_bc": 0.03232841566205025, "train_loss_llm": 0.0, "grad_norm": 0.03791254758834839, "global_step": 13294, "epoch": 369, "lr": 0.001} +{"train_loss": 0.030343009158968925, "train_loss_bc": 0.030343009158968925, "train_loss_llm": 0.0, "grad_norm": 0.049099717289209366, "global_step": 13295, "epoch": 369, "lr": 0.001} +{"train_loss": 0.0324295312166214, "train_loss_bc": 0.0324150025844574, "train_loss_llm": 0.001452714204788208, "grad_norm": 0.05348207801580429, "global_step": 13296, "epoch": 369, "lr": 0.001} +{"train_loss": 0.043381739407777786, "train_loss_bc": 0.043375153094530106, "train_loss_llm": 0.0006587207317352295, "grad_norm": 0.024840010330080986, "global_step": 13297, "epoch": 369, "lr": 0.001} +{"train_loss": 0.031883206218481064, "train_loss_bc": 0.03187112137675285, "train_loss_llm": 0.001208662986755371, "grad_norm": 0.033962432295084, "global_step": 13298, "epoch": 369, "lr": 0.001} +{"train_loss": 0.034527938812971115, "train_loss_bc": 0.034527938812971115, "train_loss_llm": 0.0, "grad_norm": 0.03899717703461647, "global_step": 13299, "epoch": 369, "lr": 0.001} +{"train_loss": 0.04042595997452736, "train_loss_bc": 0.04042595997452736, "train_loss_llm": 0.0, "grad_norm": 0.049763672053813934, "global_step": 13300, "epoch": 369, "lr": 0.001} +{"train_loss": 0.037244535982608795, "train_loss_bc": 0.037244535982608795, "train_loss_llm": 0.0, "grad_norm": 0.023107178509235382, "global_step": 13301, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03796227648854256, "train_loss_bc": 0.037943366914987564, "train_loss_llm": 0.0018909871578216553, "grad_norm": 0.033726856112480164, "global_step": 13302, "epoch": 369, "lr": 0.001} +{"train_loss": 0.04079653322696686, "train_loss_bc": 0.04079653322696686, "train_loss_llm": 0.0, "grad_norm": 0.0468447282910347, "global_step": 13303, "epoch": 369, "lr": 0.001} +{"train_loss": 0.04524774104356766, "train_loss_bc": 0.04524774104356766, "train_loss_llm": 0.0, "grad_norm": 0.06106108799576759, "global_step": 13304, "epoch": 369, "lr": 0.001} +{"train_loss": 0.036145035177469254, "train_loss_bc": 0.036145035177469254, "train_loss_llm": 0.0, "grad_norm": 0.021866992115974426, "global_step": 13305, "epoch": 369, "lr": 0.001} +{"train_loss": 0.033109307289123535, "train_loss_bc": 0.033109307289123535, "train_loss_llm": 0.0, "grad_norm": 0.029992571100592613, "global_step": 13306, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03948932886123657, "train_loss_bc": 0.039489056915044785, "train_loss_llm": 2.714991569519043e-05, "grad_norm": 0.04373748600482941, "global_step": 13307, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03938831761479378, "train_loss_bc": 0.03938831761479378, "train_loss_llm": 0.0, "grad_norm": 0.05895828828215599, "global_step": 13308, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03786427900195122, "train_loss_bc": 0.03784814849495888, "train_loss_llm": 0.0016130805015563965, "grad_norm": 0.02184629812836647, "global_step": 13309, "epoch": 369, "lr": 0.001} +{"train_loss": 0.035727810114622116, "train_loss_bc": 0.035727810114622116, "train_loss_llm": 0.0, "grad_norm": 0.03093542344868183, "global_step": 13310, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03998200222849846, "train_loss_bc": 0.039975978434085846, "train_loss_llm": 0.0006024539470672607, "grad_norm": 0.04880143702030182, "global_step": 13311, "epoch": 369, "lr": 0.001} +{"train_loss": 0.0382964052259922, "train_loss_bc": 0.0382964052259922, "train_loss_llm": 0.0, "grad_norm": 0.05962812155485153, "global_step": 13312, "epoch": 369, "lr": 0.001} +{"train_loss": 0.0391782745718956, "train_loss_bc": 0.0391782745718956, "train_loss_llm": 0.0, "grad_norm": 0.020251214504241943, "global_step": 13313, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03303203359246254, "train_loss_bc": 0.03303203359246254, "train_loss_llm": 0.0, "grad_norm": 0.029808225110173225, "global_step": 13314, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03522327542304993, "train_loss_bc": 0.03520816192030907, "train_loss_llm": 0.0015113353729248047, "grad_norm": 0.038088854402303696, "global_step": 13315, "epoch": 369, "lr": 0.001} +{"train_loss": 0.02794702723622322, "train_loss_bc": 0.027935709804296494, "train_loss_llm": 0.0011317729949951172, "grad_norm": 0.040974464267492294, "global_step": 13316, "epoch": 369, "lr": 0.001} +{"train_loss": 0.040073223412036896, "train_loss_bc": 0.040073223412036896, "train_loss_llm": 0.0, "grad_norm": 0.02316608466207981, "global_step": 13317, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03516431525349617, "train_loss_bc": 0.03516431525349617, "train_loss_llm": 0.0, "grad_norm": 0.035106759518384933, "global_step": 13318, "epoch": 369, "lr": 0.001} +{"train_loss": 0.03627700296541055, "train_loss_bc": 0.032246991991996765, "train_loss_llm": 0.0007072985172271729, "grad_norm": 0.04633285477757454, "global_step": 13319, "epoch": 369, "lr": 0.001, "val_loss": 0.06608586758375168} +{"train_loss": 0.035857148468494415, "train_loss_bc": 0.035855185240507126, "train_loss_llm": 0.00019630789756774902, "grad_norm": 0.058890778571367264, "global_step": 13320, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03739999234676361, "train_loss_bc": 0.03739999234676361, "train_loss_llm": 0.0, "grad_norm": 0.022950390353798866, "global_step": 13321, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03634963184595108, "train_loss_bc": 0.03634963184595108, "train_loss_llm": 0.0, "grad_norm": 0.035843055695295334, "global_step": 13322, "epoch": 370, "lr": 0.001} +{"train_loss": 0.04540352150797844, "train_loss_bc": 0.04540352150797844, "train_loss_llm": 0.0, "grad_norm": 0.05478167161345482, "global_step": 13323, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03811255469918251, "train_loss_bc": 0.03811255469918251, "train_loss_llm": 0.0, "grad_norm": 0.06931277364492416, "global_step": 13324, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03737214207649231, "train_loss_bc": 0.03737214207649231, "train_loss_llm": 0.0, "grad_norm": 0.02799937129020691, "global_step": 13325, "epoch": 370, "lr": 0.001} +{"train_loss": 0.04031791910529137, "train_loss_bc": 0.04031791910529137, "train_loss_llm": 0.0, "grad_norm": 0.040187716484069824, "global_step": 13326, "epoch": 370, "lr": 0.001} +{"train_loss": 0.034114498645067215, "train_loss_bc": 0.034114498645067215, "train_loss_llm": 0.0, "grad_norm": 0.049817077815532684, "global_step": 13327, "epoch": 370, "lr": 0.001} +{"train_loss": 0.031939562410116196, "train_loss_bc": 0.031912870705127716, "train_loss_llm": 0.00266912579536438, "grad_norm": 0.061467599123716354, "global_step": 13328, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03476636856794357, "train_loss_bc": 0.03474711254239082, "train_loss_llm": 0.001925438642501831, "grad_norm": 0.02570161782205105, "global_step": 13329, "epoch": 370, "lr": 0.001} +{"train_loss": 0.036812759935855865, "train_loss_bc": 0.036812759935855865, "train_loss_llm": 0.0, "grad_norm": 0.03646429628133774, "global_step": 13330, "epoch": 370, "lr": 0.001} +{"train_loss": 0.036187998950481415, "train_loss_bc": 0.036187998950481415, "train_loss_llm": 0.0, "grad_norm": 0.04860003665089607, "global_step": 13331, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03426133096218109, "train_loss_bc": 0.03426133096218109, "train_loss_llm": 0.0, "grad_norm": 0.06091452017426491, "global_step": 13332, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03858388960361481, "train_loss_bc": 0.03857138752937317, "train_loss_llm": 0.0012501776218414307, "grad_norm": 0.01928667537868023, "global_step": 13333, "epoch": 370, "lr": 0.001} +{"train_loss": 0.031384821981191635, "train_loss_bc": 0.03138164058327675, "train_loss_llm": 0.0003180205821990967, "grad_norm": 0.0353301502764225, "global_step": 13334, "epoch": 370, "lr": 0.001} +{"train_loss": 0.035917092114686966, "train_loss_bc": 0.035917092114686966, "train_loss_llm": 0.0, "grad_norm": 0.046073827892541885, "global_step": 13335, "epoch": 370, "lr": 0.001} +{"train_loss": 0.029357198625802994, "train_loss_bc": 0.029356537386775017, "train_loss_llm": 6.619095802307129e-05, "grad_norm": 0.05464228242635727, "global_step": 13336, "epoch": 370, "lr": 0.001} +{"train_loss": 0.04105795547366142, "train_loss_bc": 0.04105795547366142, "train_loss_llm": 0.0, "grad_norm": 0.02801477536559105, "global_step": 13337, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03717765212059021, "train_loss_bc": 0.03717765212059021, "train_loss_llm": 0.0, "grad_norm": 0.035311561077833176, "global_step": 13338, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03758297115564346, "train_loss_bc": 0.03758297115564346, "train_loss_llm": 0.0, "grad_norm": 0.04807677119970322, "global_step": 13339, "epoch": 370, "lr": 0.001} +{"train_loss": 0.030462300404906273, "train_loss_bc": 0.030452443286776543, "train_loss_llm": 0.0009857118129730225, "grad_norm": 0.053876329213380814, "global_step": 13340, "epoch": 370, "lr": 0.001} +{"train_loss": 0.04320848360657692, "train_loss_bc": 0.04320848360657692, "train_loss_llm": 0.0, "grad_norm": 0.026165753602981567, "global_step": 13341, "epoch": 370, "lr": 0.001} +{"train_loss": 0.027966715395450592, "train_loss_bc": 0.027966715395450592, "train_loss_llm": 0.0, "grad_norm": 0.033485330641269684, "global_step": 13342, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03758946433663368, "train_loss_bc": 0.03757919371128082, "train_loss_llm": 0.0010269880294799805, "grad_norm": 0.03800947591662407, "global_step": 13343, "epoch": 370, "lr": 0.001} +{"train_loss": 0.033699240535497665, "train_loss_bc": 0.033699240535497665, "train_loss_llm": 0.0, "grad_norm": 0.046130578964948654, "global_step": 13344, "epoch": 370, "lr": 0.001} +{"train_loss": 0.038294002413749695, "train_loss_bc": 0.038294002413749695, "train_loss_llm": 0.0, "grad_norm": 0.030798956751823425, "global_step": 13345, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03630998358130455, "train_loss_bc": 0.03630998358130455, "train_loss_llm": 0.0, "grad_norm": 0.04592226445674896, "global_step": 13346, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03495289757847786, "train_loss_bc": 0.03495289757847786, "train_loss_llm": 0.0, "grad_norm": 0.05957846716046333, "global_step": 13347, "epoch": 370, "lr": 0.001} +{"train_loss": 0.038676805794239044, "train_loss_bc": 0.0386657640337944, "train_loss_llm": 0.0011040270328521729, "grad_norm": 0.0712139904499054, "global_step": 13348, "epoch": 370, "lr": 0.001} +{"train_loss": 0.04333571717143059, "train_loss_bc": 0.04333571717143059, "train_loss_llm": 0.0, "grad_norm": 0.0199716929346323, "global_step": 13349, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03322075679898262, "train_loss_bc": 0.03322075679898262, "train_loss_llm": 0.0, "grad_norm": 0.032449256628751755, "global_step": 13350, "epoch": 370, "lr": 0.001} +{"train_loss": 0.04149872437119484, "train_loss_bc": 0.04148741066455841, "train_loss_llm": 0.0011314451694488525, "grad_norm": 0.04180579259991646, "global_step": 13351, "epoch": 370, "lr": 0.001} +{"train_loss": 0.030950646847486496, "train_loss_bc": 0.030950646847486496, "train_loss_llm": 0.0, "grad_norm": 0.05256848782300949, "global_step": 13352, "epoch": 370, "lr": 0.001} +{"train_loss": 0.04134230688214302, "train_loss_bc": 0.04133154824376106, "train_loss_llm": 0.00107574462890625, "grad_norm": 0.02324351668357849, "global_step": 13353, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03876405581831932, "train_loss_bc": 0.03876405581831932, "train_loss_llm": 0.0, "grad_norm": 0.04303920641541481, "global_step": 13354, "epoch": 370, "lr": 0.001} +{"train_loss": 0.03637773062412938, "train_loss_bc": 0.029369190335273743, "train_loss_llm": 0.0, "grad_norm": 0.04893623664975166, "global_step": 13355, "epoch": 370, "lr": 0.001, "train/cumulative_reward": 166.2298783577216, "train/mean_score": 14.835944934279974, "train/success_rate": 0.3, "test/cumulative_reward": 43.30460930316028, "test/mean_score": 8.939921504598168, "test/success_rate": 0.24, "val_loss": 0.06636902689933777, "train_action_mse_error": 0.03956200182437897} +{"train_loss": 0.03293909505009651, "train_loss_bc": 0.03293909505009651, "train_loss_llm": 0.0, "grad_norm": 0.060557086020708084, "global_step": 13356, "epoch": 371, "lr": 0.001} +{"train_loss": 0.04298650845885277, "train_loss_bc": 0.042984314262866974, "train_loss_llm": 0.0002194046974182129, "grad_norm": 0.02759200520813465, "global_step": 13357, "epoch": 371, "lr": 0.001} +{"train_loss": 0.040422286838293076, "train_loss_bc": 0.04038354009389877, "train_loss_llm": 0.00387457013130188, "grad_norm": 0.032701410353183746, "global_step": 13358, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03652312234044075, "train_loss_bc": 0.03652312234044075, "train_loss_llm": 0.0, "grad_norm": 0.03753245621919632, "global_step": 13359, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03914889693260193, "train_loss_bc": 0.03914889693260193, "train_loss_llm": 0.0, "grad_norm": 0.041774872690439224, "global_step": 13360, "epoch": 371, "lr": 0.001} +{"train_loss": 0.04186772555112839, "train_loss_bc": 0.04186772555112839, "train_loss_llm": 0.0, "grad_norm": 0.03123530000448227, "global_step": 13361, "epoch": 371, "lr": 0.001} +{"train_loss": 0.038094718009233475, "train_loss_bc": 0.038094718009233475, "train_loss_llm": 0.0, "grad_norm": 0.03840384632349014, "global_step": 13362, "epoch": 371, "lr": 0.001} +{"train_loss": 0.0422527976334095, "train_loss_bc": 0.0422527976334095, "train_loss_llm": 0.0, "grad_norm": 0.05082005262374878, "global_step": 13363, "epoch": 371, "lr": 0.001} +{"train_loss": 0.0342629998922348, "train_loss_bc": 0.034254465252161026, "train_loss_llm": 0.0008534789085388184, "grad_norm": 0.058863770216703415, "global_step": 13364, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03336397558450699, "train_loss_bc": 0.03336329013109207, "train_loss_llm": 6.851553916931152e-05, "grad_norm": 0.016601335257291794, "global_step": 13365, "epoch": 371, "lr": 0.001} +{"train_loss": 0.028047388419508934, "train_loss_bc": 0.028047388419508934, "train_loss_llm": 0.0, "grad_norm": 0.02309798076748848, "global_step": 13366, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03071305714547634, "train_loss_bc": 0.03069622628390789, "train_loss_llm": 0.0016829967498779297, "grad_norm": 0.0333084911108017, "global_step": 13367, "epoch": 371, "lr": 0.001} +{"train_loss": 0.034790974110364914, "train_loss_bc": 0.034790974110364914, "train_loss_llm": 0.0, "grad_norm": 0.041590943932533264, "global_step": 13368, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03304425626993179, "train_loss_bc": 0.03304425626993179, "train_loss_llm": 0.0, "grad_norm": 0.02423342689871788, "global_step": 13369, "epoch": 371, "lr": 0.001} +{"train_loss": 0.030933786183595657, "train_loss_bc": 0.030933786183595657, "train_loss_llm": 0.0, "grad_norm": 0.03413816913962364, "global_step": 13370, "epoch": 371, "lr": 0.001} +{"train_loss": 0.043022118508815765, "train_loss_bc": 0.043022118508815765, "train_loss_llm": 0.0, "grad_norm": 0.04235850274562836, "global_step": 13371, "epoch": 371, "lr": 0.001} +{"train_loss": 0.0353158637881279, "train_loss_bc": 0.0353158637881279, "train_loss_llm": 0.0, "grad_norm": 0.05287123844027519, "global_step": 13372, "epoch": 371, "lr": 0.001} +{"train_loss": 0.039361659437417984, "train_loss_bc": 0.039361659437417984, "train_loss_llm": 0.0, "grad_norm": 0.02545779198408127, "global_step": 13373, "epoch": 371, "lr": 0.001} +{"train_loss": 0.035894107073545456, "train_loss_bc": 0.035894107073545456, "train_loss_llm": 0.0, "grad_norm": 0.03485594317317009, "global_step": 13374, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03323499485850334, "train_loss_bc": 0.03323499485850334, "train_loss_llm": 0.0, "grad_norm": 0.05316213518381119, "global_step": 13375, "epoch": 371, "lr": 0.001} +{"train_loss": 0.0400797463953495, "train_loss_bc": 0.0400797463953495, "train_loss_llm": 0.0, "grad_norm": 0.06514229625463486, "global_step": 13376, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03668169304728508, "train_loss_bc": 0.03668169304728508, "train_loss_llm": 0.0, "grad_norm": 0.027246855199337006, "global_step": 13377, "epoch": 371, "lr": 0.001} +{"train_loss": 0.0415118932723999, "train_loss_bc": 0.0414961613714695, "train_loss_llm": 0.0015732049942016602, "grad_norm": 0.03470984846353531, "global_step": 13378, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03609175235033035, "train_loss_bc": 0.03609175235033035, "train_loss_llm": 0.0, "grad_norm": 0.04501786828041077, "global_step": 13379, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03562764823436737, "train_loss_bc": 0.03562270849943161, "train_loss_llm": 0.0004940032958984375, "grad_norm": 0.06430429965257645, "global_step": 13380, "epoch": 371, "lr": 0.001} +{"train_loss": 0.036642491817474365, "train_loss_bc": 0.03662842884659767, "train_loss_llm": 0.0014061331748962402, "grad_norm": 0.02230641432106495, "global_step": 13381, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03331105783581734, "train_loss_bc": 0.03331105783581734, "train_loss_llm": 0.0, "grad_norm": 0.03523797541856766, "global_step": 13382, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03880326822400093, "train_loss_bc": 0.03880326822400093, "train_loss_llm": 0.0, "grad_norm": 0.05118110775947571, "global_step": 13383, "epoch": 371, "lr": 0.001} +{"train_loss": 0.036823730915784836, "train_loss_bc": 0.036823730915784836, "train_loss_llm": 0.0, "grad_norm": 0.06012414023280144, "global_step": 13384, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03668831288814545, "train_loss_bc": 0.03668831288814545, "train_loss_llm": 0.0, "grad_norm": 0.020893681794404984, "global_step": 13385, "epoch": 371, "lr": 0.001} +{"train_loss": 0.030678339302539825, "train_loss_bc": 0.03066815249621868, "train_loss_llm": 0.0010187029838562012, "grad_norm": 0.03268835321068764, "global_step": 13386, "epoch": 371, "lr": 0.001} +{"train_loss": 0.037151940166950226, "train_loss_bc": 0.037151940166950226, "train_loss_llm": 0.0, "grad_norm": 0.04089511185884476, "global_step": 13387, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03957012668251991, "train_loss_bc": 0.03957012668251991, "train_loss_llm": 0.0, "grad_norm": 0.04831716790795326, "global_step": 13388, "epoch": 371, "lr": 0.001} +{"train_loss": 0.03229545056819916, "train_loss_bc": 0.03229545056819916, "train_loss_llm": 0.0, "grad_norm": 0.022340944036841393, "global_step": 13389, "epoch": 371, "lr": 0.001} +{"train_loss": 0.029741810634732246, "train_loss_bc": 0.02972652204334736, "train_loss_llm": 0.001528918743133545, "grad_norm": 0.030689619481563568, "global_step": 13390, "epoch": 371, "lr": 0.001} +{"train_loss": 0.036371948973586164, "train_loss_bc": 0.04141569137573242, "train_loss_llm": 0.005487650632858276, "grad_norm": 0.037186358124017715, "global_step": 13391, "epoch": 371, "lr": 0.001, "val_loss": 0.06604669988155365} +{"train_loss": 0.031972486525774, "train_loss_bc": 0.03197198361158371, "train_loss_llm": 5.015730857849121e-05, "grad_norm": 0.04897347837686539, "global_step": 13392, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03514668345451355, "train_loss_bc": 0.03512579947710037, "train_loss_llm": 0.002088487148284912, "grad_norm": 0.02225610241293907, "global_step": 13393, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03460269793868065, "train_loss_bc": 0.03460269793868065, "train_loss_llm": 0.0, "grad_norm": 0.0330791100859642, "global_step": 13394, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03190501034259796, "train_loss_bc": 0.031876206398010254, "train_loss_llm": 0.0028805136680603027, "grad_norm": 0.04562893882393837, "global_step": 13395, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03463011980056763, "train_loss_bc": 0.03461798280477524, "train_loss_llm": 0.0012135803699493408, "grad_norm": 0.06072067841887474, "global_step": 13396, "epoch": 372, "lr": 0.001} +{"train_loss": 0.030220042914152145, "train_loss_bc": 0.030220042914152145, "train_loss_llm": 0.0, "grad_norm": 0.017971286550164223, "global_step": 13397, "epoch": 372, "lr": 0.001} +{"train_loss": 0.037872858345508575, "train_loss_bc": 0.037866633385419846, "train_loss_llm": 0.000622481107711792, "grad_norm": 0.03527435287833214, "global_step": 13398, "epoch": 372, "lr": 0.001} +{"train_loss": 0.035659004002809525, "train_loss_bc": 0.035659004002809525, "train_loss_llm": 0.0, "grad_norm": 0.04556640610098839, "global_step": 13399, "epoch": 372, "lr": 0.001} +{"train_loss": 0.040374621748924255, "train_loss_bc": 0.040374621748924255, "train_loss_llm": 0.0, "grad_norm": 0.05280057340860367, "global_step": 13400, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03219502046704292, "train_loss_bc": 0.03219502046704292, "train_loss_llm": 0.0, "grad_norm": 0.01947879232466221, "global_step": 13401, "epoch": 372, "lr": 0.001} +{"train_loss": 0.034569066017866135, "train_loss_bc": 0.034544363617897034, "train_loss_llm": 0.002470344305038452, "grad_norm": 0.03249070793390274, "global_step": 13402, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03465169295668602, "train_loss_bc": 0.03464990109205246, "train_loss_llm": 0.0001793205738067627, "grad_norm": 0.04127487167716026, "global_step": 13403, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03611421957612038, "train_loss_bc": 0.03611421957612038, "train_loss_llm": 0.0, "grad_norm": 0.05206478014588356, "global_step": 13404, "epoch": 372, "lr": 0.001} +{"train_loss": 0.035701267421245575, "train_loss_bc": 0.035701267421245575, "train_loss_llm": 0.0, "grad_norm": 0.023096460849046707, "global_step": 13405, "epoch": 372, "lr": 0.001} +{"train_loss": 0.04260695353150368, "train_loss_bc": 0.04260695353150368, "train_loss_llm": 0.0, "grad_norm": 0.0314948670566082, "global_step": 13406, "epoch": 372, "lr": 0.001} +{"train_loss": 0.036970458924770355, "train_loss_bc": 0.036970458924770355, "train_loss_llm": 0.0, "grad_norm": 0.03710982948541641, "global_step": 13407, "epoch": 372, "lr": 0.001} +{"train_loss": 0.033651407808065414, "train_loss_bc": 0.033651407808065414, "train_loss_llm": 0.0, "grad_norm": 0.04444494843482971, "global_step": 13408, "epoch": 372, "lr": 0.001} +{"train_loss": 0.028879785910248756, "train_loss_bc": 0.028853008523583412, "train_loss_llm": 0.0026776492595672607, "grad_norm": 0.017132511362433434, "global_step": 13409, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03871472552418709, "train_loss_bc": 0.03871472552418709, "train_loss_llm": 0.0, "grad_norm": 0.0276417788118124, "global_step": 13410, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03566135838627815, "train_loss_bc": 0.03566135838627815, "train_loss_llm": 0.0, "grad_norm": 0.03642209991812706, "global_step": 13411, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03556940704584122, "train_loss_bc": 0.03555738553404808, "train_loss_llm": 0.0012021064758300781, "grad_norm": 0.04686468839645386, "global_step": 13412, "epoch": 372, "lr": 0.001} +{"train_loss": 0.039327189326286316, "train_loss_bc": 0.039327189326286316, "train_loss_llm": 0.0, "grad_norm": 0.02222047559916973, "global_step": 13413, "epoch": 372, "lr": 0.001} +{"train_loss": 0.038922786712646484, "train_loss_bc": 0.038912639021873474, "train_loss_llm": 0.0010147392749786377, "grad_norm": 0.03515014797449112, "global_step": 13414, "epoch": 372, "lr": 0.001} +{"train_loss": 0.0378543883562088, "train_loss_bc": 0.0378543883562088, "train_loss_llm": 0.0, "grad_norm": 0.04843800514936447, "global_step": 13415, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03993786498904228, "train_loss_bc": 0.03993786498904228, "train_loss_llm": 0.0, "grad_norm": 0.060551490634679794, "global_step": 13416, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03700008615851402, "train_loss_bc": 0.03700008615851402, "train_loss_llm": 0.0, "grad_norm": 0.024293340742588043, "global_step": 13417, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03063921071588993, "train_loss_bc": 0.03063921071588993, "train_loss_llm": 0.0, "grad_norm": 0.031804002821445465, "global_step": 13418, "epoch": 372, "lr": 0.001} +{"train_loss": 0.042299896478652954, "train_loss_bc": 0.042299896478652954, "train_loss_llm": 0.0, "grad_norm": 0.03975382074713707, "global_step": 13419, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03615895286202431, "train_loss_bc": 0.03615895286202431, "train_loss_llm": 0.0, "grad_norm": 0.0521065779030323, "global_step": 13420, "epoch": 372, "lr": 0.001} +{"train_loss": 0.035896074026823044, "train_loss_bc": 0.035886574536561966, "train_loss_llm": 0.0009500682353973389, "grad_norm": 0.022193023934960365, "global_step": 13421, "epoch": 372, "lr": 0.001} +{"train_loss": 0.044924888759851456, "train_loss_bc": 0.044924888759851456, "train_loss_llm": 0.0, "grad_norm": 0.031229756772518158, "global_step": 13422, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03724246099591255, "train_loss_bc": 0.03724246099591255, "train_loss_llm": 0.0, "grad_norm": 0.04459047317504883, "global_step": 13423, "epoch": 372, "lr": 0.001} +{"train_loss": 0.038295526057481766, "train_loss_bc": 0.038295526057481766, "train_loss_llm": 0.0, "grad_norm": 0.05619920417666435, "global_step": 13424, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03582173213362694, "train_loss_bc": 0.03582173213362694, "train_loss_llm": 0.0, "grad_norm": 0.023476364091038704, "global_step": 13425, "epoch": 372, "lr": 0.001} +{"train_loss": 0.03458782285451889, "train_loss_bc": 0.03458782285451889, "train_loss_llm": 0.0, "grad_norm": 0.035930030047893524, "global_step": 13426, "epoch": 372, "lr": 0.001} +{"train_loss": 0.036365592748754554, "train_loss_bc": 0.04258356988430023, "train_loss_llm": 0.0, "grad_norm": 0.05141552910208702, "global_step": 13427, "epoch": 372, "lr": 0.001, "val_loss": 0.06660562753677368} +{"train_loss": 0.04040630906820297, "train_loss_bc": 0.04040630906820297, "train_loss_llm": 0.0, "grad_norm": 0.05337907373905182, "global_step": 13428, "epoch": 373, "lr": 0.001} +{"train_loss": 0.030937500298023224, "train_loss_bc": 0.030930623412132263, "train_loss_llm": 0.0006876289844512939, "grad_norm": 0.02007978782057762, "global_step": 13429, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03709115460515022, "train_loss_bc": 0.03709115460515022, "train_loss_llm": 0.0, "grad_norm": 0.030545635148882866, "global_step": 13430, "epoch": 373, "lr": 0.001} +{"train_loss": 0.04305613413453102, "train_loss_bc": 0.04305613413453102, "train_loss_llm": 0.0, "grad_norm": 0.046751365065574646, "global_step": 13431, "epoch": 373, "lr": 0.001} +{"train_loss": 0.036619361490011215, "train_loss_bc": 0.036619361490011215, "train_loss_llm": 0.0, "grad_norm": 0.056346215307712555, "global_step": 13432, "epoch": 373, "lr": 0.001} +{"train_loss": 0.029597172513604164, "train_loss_bc": 0.029597172513604164, "train_loss_llm": 0.0, "grad_norm": 0.01983923465013504, "global_step": 13433, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03347212076187134, "train_loss_bc": 0.03347212076187134, "train_loss_llm": 0.0, "grad_norm": 0.030515065416693687, "global_step": 13434, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03691890090703964, "train_loss_bc": 0.03691890090703964, "train_loss_llm": 0.0, "grad_norm": 0.039708465337753296, "global_step": 13435, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03478069230914116, "train_loss_bc": 0.03478069230914116, "train_loss_llm": 0.0, "grad_norm": 0.04838130250573158, "global_step": 13436, "epoch": 373, "lr": 0.001} +{"train_loss": 0.030934728682041168, "train_loss_bc": 0.030934728682041168, "train_loss_llm": 0.0, "grad_norm": 0.01757904328405857, "global_step": 13437, "epoch": 373, "lr": 0.001} +{"train_loss": 0.038881342858076096, "train_loss_bc": 0.038881342858076096, "train_loss_llm": 0.0, "grad_norm": 0.033902812749147415, "global_step": 13438, "epoch": 373, "lr": 0.001} +{"train_loss": 0.04082838445901871, "train_loss_bc": 0.04082838445901871, "train_loss_llm": 0.0, "grad_norm": 0.04119810834527016, "global_step": 13439, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03514263033866882, "train_loss_bc": 0.03514263033866882, "train_loss_llm": 0.0, "grad_norm": 0.04634929820895195, "global_step": 13440, "epoch": 373, "lr": 0.001} +{"train_loss": 0.040822260081768036, "train_loss_bc": 0.04081854224205017, "train_loss_llm": 0.000371783971786499, "grad_norm": 0.024566849693655968, "global_step": 13441, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03491898998618126, "train_loss_bc": 0.03491898998618126, "train_loss_llm": 0.0, "grad_norm": 0.035885486751794815, "global_step": 13442, "epoch": 373, "lr": 0.001} +{"train_loss": 0.030650772154331207, "train_loss_bc": 0.030650772154331207, "train_loss_llm": 0.0, "grad_norm": 0.04599710926413536, "global_step": 13443, "epoch": 373, "lr": 0.001} +{"train_loss": 0.037450190633535385, "train_loss_bc": 0.037450190633535385, "train_loss_llm": 0.0, "grad_norm": 0.05389908701181412, "global_step": 13444, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03656955808401108, "train_loss_bc": 0.03656955808401108, "train_loss_llm": 0.0, "grad_norm": 0.02371896617114544, "global_step": 13445, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03931618109345436, "train_loss_bc": 0.03931618109345436, "train_loss_llm": 0.0, "grad_norm": 0.03845314309000969, "global_step": 13446, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03415520489215851, "train_loss_bc": 0.03415520489215851, "train_loss_llm": 0.0, "grad_norm": 0.05441289767622948, "global_step": 13447, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03842660039663315, "train_loss_bc": 0.03842660039663315, "train_loss_llm": 0.0, "grad_norm": 0.07107731699943542, "global_step": 13448, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03594011813402176, "train_loss_bc": 0.03594011813402176, "train_loss_llm": 0.0, "grad_norm": 0.025436194613575935, "global_step": 13449, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03649812191724777, "train_loss_bc": 0.03649481013417244, "train_loss_llm": 0.0003312826156616211, "grad_norm": 0.036776892840862274, "global_step": 13450, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03139795362949371, "train_loss_bc": 0.03139795362949371, "train_loss_llm": 0.0, "grad_norm": 0.05709817633032799, "global_step": 13451, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03352566435933113, "train_loss_bc": 0.03351550176739693, "train_loss_llm": 0.0010162293910980225, "grad_norm": 0.06672364473342896, "global_step": 13452, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03393551707267761, "train_loss_bc": 0.03393551707267761, "train_loss_llm": 0.0, "grad_norm": 0.019832594320178032, "global_step": 13453, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03887547180056572, "train_loss_bc": 0.03887547180056572, "train_loss_llm": 0.0, "grad_norm": 0.03805527836084366, "global_step": 13454, "epoch": 373, "lr": 0.001} +{"train_loss": 0.0296537633985281, "train_loss_bc": 0.0296537633985281, "train_loss_llm": 0.0, "grad_norm": 0.043881386518478394, "global_step": 13455, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03694482892751694, "train_loss_bc": 0.03694482892751694, "train_loss_llm": 0.0, "grad_norm": 0.056622229516506195, "global_step": 13456, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03909767046570778, "train_loss_bc": 0.03909767046570778, "train_loss_llm": 0.0, "grad_norm": 0.020453542470932007, "global_step": 13457, "epoch": 373, "lr": 0.001} +{"train_loss": 0.04209395498037338, "train_loss_bc": 0.04209395498037338, "train_loss_llm": 0.0, "grad_norm": 0.030233846977353096, "global_step": 13458, "epoch": 373, "lr": 0.001} +{"train_loss": 0.032191503793001175, "train_loss_bc": 0.032191503793001175, "train_loss_llm": 0.0, "grad_norm": 0.03927808627486229, "global_step": 13459, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03877643868327141, "train_loss_bc": 0.03877643868327141, "train_loss_llm": 0.0, "grad_norm": 0.04571046680212021, "global_step": 13460, "epoch": 373, "lr": 0.001} +{"train_loss": 0.04101308062672615, "train_loss_bc": 0.0409875214099884, "train_loss_llm": 0.0025557875633239746, "grad_norm": 0.027986211702227592, "global_step": 13461, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03722294047474861, "train_loss_bc": 0.03722294047474861, "train_loss_llm": 0.0, "grad_norm": 0.044774267822504044, "global_step": 13462, "epoch": 373, "lr": 0.001} +{"train_loss": 0.03637354707138406, "train_loss_bc": 0.04126652330160141, "train_loss_llm": 0.0037952065467834473, "grad_norm": 0.07060708105564117, "global_step": 13463, "epoch": 373, "lr": 0.001, "val_loss": 0.06654664874076843} +{"train_loss": 0.034578245133161545, "train_loss_bc": 0.034578245133161545, "train_loss_llm": 0.0, "grad_norm": 0.07638037949800491, "global_step": 13464, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03887864947319031, "train_loss_bc": 0.03887864947319031, "train_loss_llm": 0.0, "grad_norm": 0.02670590579509735, "global_step": 13465, "epoch": 374, "lr": 0.001} +{"train_loss": 0.04088734835386276, "train_loss_bc": 0.04088734835386276, "train_loss_llm": 0.0, "grad_norm": 0.03953269124031067, "global_step": 13466, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03730108216404915, "train_loss_bc": 0.03729601949453354, "train_loss_llm": 0.0005061924457550049, "grad_norm": 0.051247455179691315, "global_step": 13467, "epoch": 374, "lr": 0.001} +{"train_loss": 0.04193359613418579, "train_loss_bc": 0.04193359613418579, "train_loss_llm": 0.0, "grad_norm": 0.0601516030728817, "global_step": 13468, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03350212052464485, "train_loss_bc": 0.03349900245666504, "train_loss_llm": 0.00031182169914245605, "grad_norm": 0.02303997054696083, "global_step": 13469, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03704727068543434, "train_loss_bc": 0.03704727068543434, "train_loss_llm": 0.0, "grad_norm": 0.03925466910004616, "global_step": 13470, "epoch": 374, "lr": 0.001} +{"train_loss": 0.027770688757300377, "train_loss_bc": 0.027770688757300377, "train_loss_llm": 0.0, "grad_norm": 0.047798462212085724, "global_step": 13471, "epoch": 374, "lr": 0.001} +{"train_loss": 0.02984006144106388, "train_loss_bc": 0.02983359806239605, "train_loss_llm": 0.0006463527679443359, "grad_norm": 0.0645514652132988, "global_step": 13472, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03566673770546913, "train_loss_bc": 0.03566673770546913, "train_loss_llm": 0.0, "grad_norm": 0.03138499706983566, "global_step": 13473, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03576604649424553, "train_loss_bc": 0.03576604649424553, "train_loss_llm": 0.0, "grad_norm": 0.049043070524930954, "global_step": 13474, "epoch": 374, "lr": 0.001} +{"train_loss": 0.027990981936454773, "train_loss_bc": 0.027990981936454773, "train_loss_llm": 0.0, "grad_norm": 0.06702817231416702, "global_step": 13475, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03822394832968712, "train_loss_bc": 0.03822394832968712, "train_loss_llm": 0.0, "grad_norm": 0.08172901719808578, "global_step": 13476, "epoch": 374, "lr": 0.001} +{"train_loss": 0.04044051468372345, "train_loss_bc": 0.04044051468372345, "train_loss_llm": 0.0, "grad_norm": 0.029477249830961227, "global_step": 13477, "epoch": 374, "lr": 0.001} +{"train_loss": 0.029503390192985535, "train_loss_bc": 0.029503390192985535, "train_loss_llm": 0.0, "grad_norm": 0.03767773509025574, "global_step": 13478, "epoch": 374, "lr": 0.001} +{"train_loss": 0.040819451212882996, "train_loss_bc": 0.04081921651959419, "train_loss_llm": 2.3573637008666992e-05, "grad_norm": 0.054587990045547485, "global_step": 13479, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03366592898964882, "train_loss_bc": 0.03366592898964882, "train_loss_llm": 0.0, "grad_norm": 0.07497745752334595, "global_step": 13480, "epoch": 374, "lr": 0.001} +{"train_loss": 0.034541282802820206, "train_loss_bc": 0.03453466296195984, "train_loss_llm": 0.0006619095802307129, "grad_norm": 0.027383625507354736, "global_step": 13481, "epoch": 374, "lr": 0.001} +{"train_loss": 0.04178023710846901, "train_loss_bc": 0.041778359562158585, "train_loss_llm": 0.00018772482872009277, "grad_norm": 0.04334350675344467, "global_step": 13482, "epoch": 374, "lr": 0.001} +{"train_loss": 0.04176681116223335, "train_loss_bc": 0.04176681116223335, "train_loss_llm": 0.0, "grad_norm": 0.05344131588935852, "global_step": 13483, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03907780721783638, "train_loss_bc": 0.03907780721783638, "train_loss_llm": 0.0, "grad_norm": 0.07534778863191605, "global_step": 13484, "epoch": 374, "lr": 0.001} +{"train_loss": 0.037884604185819626, "train_loss_bc": 0.03787940368056297, "train_loss_llm": 0.0005200207233428955, "grad_norm": 0.02249741367995739, "global_step": 13485, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03404634818434715, "train_loss_bc": 0.03404634818434715, "train_loss_llm": 0.0, "grad_norm": 0.029485439881682396, "global_step": 13486, "epoch": 374, "lr": 0.001} +{"train_loss": 0.036071985960006714, "train_loss_bc": 0.036071985960006714, "train_loss_llm": 0.0, "grad_norm": 0.04136856645345688, "global_step": 13487, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03369896113872528, "train_loss_bc": 0.03369896113872528, "train_loss_llm": 0.0, "grad_norm": 0.050666313618421555, "global_step": 13488, "epoch": 374, "lr": 0.001} +{"train_loss": 0.035909201949834824, "train_loss_bc": 0.03590494021773338, "train_loss_llm": 0.00042635202407836914, "grad_norm": 0.026410449296236038, "global_step": 13489, "epoch": 374, "lr": 0.001} +{"train_loss": 0.038887012749910355, "train_loss_bc": 0.038887012749910355, "train_loss_llm": 0.0, "grad_norm": 0.052352625876665115, "global_step": 13490, "epoch": 374, "lr": 0.001} +{"train_loss": 0.0356961190700531, "train_loss_bc": 0.0356961190700531, "train_loss_llm": 0.0, "grad_norm": 0.06371384114027023, "global_step": 13491, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03981004282832146, "train_loss_bc": 0.03981004282832146, "train_loss_llm": 0.0, "grad_norm": 0.08857133984565735, "global_step": 13492, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03996198996901512, "train_loss_bc": 0.039938461035490036, "train_loss_llm": 0.0023529231548309326, "grad_norm": 0.024446137249469757, "global_step": 13493, "epoch": 374, "lr": 0.001} +{"train_loss": 0.042605459690093994, "train_loss_bc": 0.042605459690093994, "train_loss_llm": 0.0, "grad_norm": 0.04177090525627136, "global_step": 13494, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03849436715245247, "train_loss_bc": 0.03849436715245247, "train_loss_llm": 0.0, "grad_norm": 0.054793309420347214, "global_step": 13495, "epoch": 374, "lr": 0.001} +{"train_loss": 0.037840794771909714, "train_loss_bc": 0.037840794771909714, "train_loss_llm": 0.0, "grad_norm": 0.06457232683897018, "global_step": 13496, "epoch": 374, "lr": 0.001} +{"train_loss": 0.03666310757398605, "train_loss_bc": 0.03666310757398605, "train_loss_llm": 0.0, "grad_norm": 0.02622234635055065, "global_step": 13497, "epoch": 374, "lr": 0.001} +{"train_loss": 0.036275915801525116, "train_loss_bc": 0.03625316172838211, "train_loss_llm": 0.002275317907333374, "grad_norm": 0.04817532002925873, "global_step": 13498, "epoch": 374, "lr": 0.001} +{"train_loss": 0.036713488296502166, "train_loss_bc": 0.03685499355196953, "train_loss_llm": 0.0002473592758178711, "grad_norm": 0.06591130793094635, "global_step": 13499, "epoch": 374, "lr": 0.001, "val_loss": 0.06779924035072327} +{"train_loss": 0.041755568236112595, "train_loss_bc": 0.041746024042367935, "train_loss_llm": 0.0009544193744659424, "grad_norm": 0.07649201154708862, "global_step": 13500, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03289966657757759, "train_loss_bc": 0.03289966657757759, "train_loss_llm": 0.0, "grad_norm": 0.030553456395864487, "global_step": 13501, "epoch": 375, "lr": 0.001} +{"train_loss": 0.04028020054101944, "train_loss_bc": 0.04028020054101944, "train_loss_llm": 0.0, "grad_norm": 0.04120458662509918, "global_step": 13502, "epoch": 375, "lr": 0.001} +{"train_loss": 0.032745640724897385, "train_loss_bc": 0.032745640724897385, "train_loss_llm": 0.0, "grad_norm": 0.052238862961530685, "global_step": 13503, "epoch": 375, "lr": 0.001} +{"train_loss": 0.042914096266031265, "train_loss_bc": 0.042914096266031265, "train_loss_llm": 0.0, "grad_norm": 0.06102362275123596, "global_step": 13504, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03628694266080856, "train_loss_bc": 0.03628694266080856, "train_loss_llm": 0.0, "grad_norm": 0.023700328543782234, "global_step": 13505, "epoch": 375, "lr": 0.001} +{"train_loss": 0.0398620069026947, "train_loss_bc": 0.0398620069026947, "train_loss_llm": 0.0, "grad_norm": 0.04638127610087395, "global_step": 13506, "epoch": 375, "lr": 0.001} +{"train_loss": 0.04136804863810539, "train_loss_bc": 0.04136804863810539, "train_loss_llm": 0.0, "grad_norm": 0.04836477339267731, "global_step": 13507, "epoch": 375, "lr": 0.001} +{"train_loss": 0.034385595470666885, "train_loss_bc": 0.034385595470666885, "train_loss_llm": 0.0, "grad_norm": 0.05507810786366463, "global_step": 13508, "epoch": 375, "lr": 0.001} +{"train_loss": 0.036178506910800934, "train_loss_bc": 0.036178506910800934, "train_loss_llm": 0.0, "grad_norm": 0.0218808576464653, "global_step": 13509, "epoch": 375, "lr": 0.001} +{"train_loss": 0.039515137672424316, "train_loss_bc": 0.03951004147529602, "train_loss_llm": 0.0005094408988952637, "grad_norm": 0.034681566059589386, "global_step": 13510, "epoch": 375, "lr": 0.001} +{"train_loss": 0.035576920956373215, "train_loss_bc": 0.03557058051228523, "train_loss_llm": 0.0006340742111206055, "grad_norm": 0.04860972613096237, "global_step": 13511, "epoch": 375, "lr": 0.001} +{"train_loss": 0.029367070645093918, "train_loss_bc": 0.029367070645093918, "train_loss_llm": 0.0, "grad_norm": 0.05997499078512192, "global_step": 13512, "epoch": 375, "lr": 0.001} +{"train_loss": 0.033436331897974014, "train_loss_bc": 0.033436331897974014, "train_loss_llm": 0.0, "grad_norm": 0.026011614128947258, "global_step": 13513, "epoch": 375, "lr": 0.001} +{"train_loss": 0.035528138279914856, "train_loss_bc": 0.035528138279914856, "train_loss_llm": 0.0, "grad_norm": 0.04274258390069008, "global_step": 13514, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03918777406215668, "train_loss_bc": 0.03918777406215668, "train_loss_llm": 0.0, "grad_norm": 0.06501608341932297, "global_step": 13515, "epoch": 375, "lr": 0.001} +{"train_loss": 0.038092758506536484, "train_loss_bc": 0.0380815826356411, "train_loss_llm": 0.0011177361011505127, "grad_norm": 0.08545427769422531, "global_step": 13516, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03380395099520683, "train_loss_bc": 0.03380395099520683, "train_loss_llm": 0.0, "grad_norm": 0.019485723227262497, "global_step": 13517, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03775646165013313, "train_loss_bc": 0.03775646165013313, "train_loss_llm": 0.0, "grad_norm": 0.040620457381010056, "global_step": 13518, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03420000895857811, "train_loss_bc": 0.03420000895857811, "train_loss_llm": 0.0, "grad_norm": 0.059102922677993774, "global_step": 13519, "epoch": 375, "lr": 0.001} +{"train_loss": 0.035218797624111176, "train_loss_bc": 0.035218797624111176, "train_loss_llm": 0.0, "grad_norm": 0.07491561025381088, "global_step": 13520, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03442393243312836, "train_loss_bc": 0.03442393243312836, "train_loss_llm": 0.0, "grad_norm": 0.030887752771377563, "global_step": 13521, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03714245185256004, "train_loss_bc": 0.03714245185256004, "train_loss_llm": 0.0, "grad_norm": 0.046367332339286804, "global_step": 13522, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03915819153189659, "train_loss_bc": 0.039152901619672775, "train_loss_llm": 0.0005288422107696533, "grad_norm": 0.06419363617897034, "global_step": 13523, "epoch": 375, "lr": 0.001} +{"train_loss": 0.036080826073884964, "train_loss_bc": 0.03606033697724342, "train_loss_llm": 0.002049088478088379, "grad_norm": 0.0785081684589386, "global_step": 13524, "epoch": 375, "lr": 0.001} +{"train_loss": 0.033579271286726, "train_loss_bc": 0.033579271286726, "train_loss_llm": 0.0, "grad_norm": 0.027156373485922813, "global_step": 13525, "epoch": 375, "lr": 0.001} +{"train_loss": 0.04093397408723831, "train_loss_bc": 0.04093397408723831, "train_loss_llm": 0.0, "grad_norm": 0.04494930058717728, "global_step": 13526, "epoch": 375, "lr": 0.001} +{"train_loss": 0.039500873535871506, "train_loss_bc": 0.039500873535871506, "train_loss_llm": 0.0, "grad_norm": 0.05595856159925461, "global_step": 13527, "epoch": 375, "lr": 0.001} +{"train_loss": 0.0417453795671463, "train_loss_bc": 0.0417453795671463, "train_loss_llm": 0.0, "grad_norm": 0.0709085464477539, "global_step": 13528, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03103686310350895, "train_loss_bc": 0.03103686310350895, "train_loss_llm": 0.0, "grad_norm": 0.021239563822746277, "global_step": 13529, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03775680810213089, "train_loss_bc": 0.03775680810213089, "train_loss_llm": 0.0, "grad_norm": 0.03655471280217171, "global_step": 13530, "epoch": 375, "lr": 0.001} +{"train_loss": 0.036775145679712296, "train_loss_bc": 0.03677166625857353, "train_loss_llm": 0.0003477931022644043, "grad_norm": 0.04156019538640976, "global_step": 13531, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03787415847182274, "train_loss_bc": 0.03787415847182274, "train_loss_llm": 0.0, "grad_norm": 0.057951413094997406, "global_step": 13532, "epoch": 375, "lr": 0.001} +{"train_loss": 0.028115110471844673, "train_loss_bc": 0.028115110471844673, "train_loss_llm": 0.0, "grad_norm": 0.020017942413687706, "global_step": 13533, "epoch": 375, "lr": 0.001} +{"train_loss": 0.037215251475572586, "train_loss_bc": 0.037215251475572586, "train_loss_llm": 0.0, "grad_norm": 0.0351078137755394, "global_step": 13534, "epoch": 375, "lr": 0.001} +{"train_loss": 0.03653718303475115, "train_loss_bc": 0.033640727400779724, "train_loss_llm": 0.0, "grad_norm": 0.05373867228627205, "global_step": 13535, "epoch": 375, "lr": 0.001, "train/cumulative_reward": 146.06036952873853, "train/mean_score": 17.836891802862404, "train/success_rate": 0.4, "test/cumulative_reward": 45.05581291232455, "test/mean_score": 6.2369300173947035, "test/success_rate": 0.12, "val_loss": 0.06580083072185516, "train_action_mse_error": 0.03912007063627243} +{"train_loss": 0.03496482968330383, "train_loss_bc": 0.03496482968330383, "train_loss_llm": 0.0, "grad_norm": 0.068567655980587, "global_step": 13536, "epoch": 376, "lr": 0.001} +{"train_loss": 0.032230447977781296, "train_loss_bc": 0.0322275273501873, "train_loss_llm": 0.00029206275939941406, "grad_norm": 0.020716704428195953, "global_step": 13537, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03845865651965141, "train_loss_bc": 0.038432225584983826, "train_loss_llm": 0.002643078565597534, "grad_norm": 0.03459475189447403, "global_step": 13538, "epoch": 376, "lr": 0.001} +{"train_loss": 0.04216210916638374, "train_loss_bc": 0.04216210916638374, "train_loss_llm": 0.0, "grad_norm": 0.059482641518116, "global_step": 13539, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03266932815313339, "train_loss_bc": 0.03266746550798416, "train_loss_llm": 0.0001862943172454834, "grad_norm": 0.0706108957529068, "global_step": 13540, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03683385252952576, "train_loss_bc": 0.036828186362981796, "train_loss_llm": 0.000566631555557251, "grad_norm": 0.019289612770080566, "global_step": 13541, "epoch": 376, "lr": 0.001} +{"train_loss": 0.0392586886882782, "train_loss_bc": 0.03925824537873268, "train_loss_llm": 4.416704177856445e-05, "grad_norm": 0.033440541476011276, "global_step": 13542, "epoch": 376, "lr": 0.001} +{"train_loss": 0.034839797765016556, "train_loss_bc": 0.034839797765016556, "train_loss_llm": 0.0, "grad_norm": 0.04210265353322029, "global_step": 13543, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03729189559817314, "train_loss_bc": 0.03729189559817314, "train_loss_llm": 0.0, "grad_norm": 0.05198119580745697, "global_step": 13544, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03873853012919426, "train_loss_bc": 0.03873853012919426, "train_loss_llm": 0.0, "grad_norm": 0.024844858795404434, "global_step": 13545, "epoch": 376, "lr": 0.001} +{"train_loss": 0.039947494864463806, "train_loss_bc": 0.039947494864463806, "train_loss_llm": 0.0, "grad_norm": 0.042922601103782654, "global_step": 13546, "epoch": 376, "lr": 0.001} +{"train_loss": 0.038133155554533005, "train_loss_bc": 0.038133155554533005, "train_loss_llm": 0.0, "grad_norm": 0.048915617167949677, "global_step": 13547, "epoch": 376, "lr": 0.001} +{"train_loss": 0.038575898855924606, "train_loss_bc": 0.03856230899691582, "train_loss_llm": 0.0013589560985565186, "grad_norm": 0.05693387612700462, "global_step": 13548, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03717599809169769, "train_loss_bc": 0.03717599809169769, "train_loss_llm": 0.0, "grad_norm": 0.02581040933728218, "global_step": 13549, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03828516602516174, "train_loss_bc": 0.038272030651569366, "train_loss_llm": 0.0013135671615600586, "grad_norm": 0.04738481715321541, "global_step": 13550, "epoch": 376, "lr": 0.001} +{"train_loss": 0.031383730471134186, "train_loss_bc": 0.031383730471134186, "train_loss_llm": 0.0, "grad_norm": 0.061410803347826004, "global_step": 13551, "epoch": 376, "lr": 0.001} +{"train_loss": 0.04116756096482277, "train_loss_bc": 0.04116756096482277, "train_loss_llm": 0.0, "grad_norm": 0.07609609514474869, "global_step": 13552, "epoch": 376, "lr": 0.001} +{"train_loss": 0.04332461580634117, "train_loss_bc": 0.04330473393201828, "train_loss_llm": 0.0019882619380950928, "grad_norm": 0.02548365481197834, "global_step": 13553, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03749658912420273, "train_loss_bc": 0.03749658912420273, "train_loss_llm": 0.0, "grad_norm": 0.04703449830412865, "global_step": 13554, "epoch": 376, "lr": 0.001} +{"train_loss": 0.036983903497457504, "train_loss_bc": 0.036983903497457504, "train_loss_llm": 0.0, "grad_norm": 0.054328352212905884, "global_step": 13555, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03252499923110008, "train_loss_bc": 0.03252499923110008, "train_loss_llm": 0.0, "grad_norm": 0.07252953946590424, "global_step": 13556, "epoch": 376, "lr": 0.001} +{"train_loss": 0.033062126487493515, "train_loss_bc": 0.03306165710091591, "train_loss_llm": 4.70578670501709e-05, "grad_norm": 0.032662734389305115, "global_step": 13557, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03026510961353779, "train_loss_bc": 0.03026510961353779, "train_loss_llm": 0.0, "grad_norm": 0.04223431646823883, "global_step": 13558, "epoch": 376, "lr": 0.001} +{"train_loss": 0.0480090007185936, "train_loss_bc": 0.0480090007185936, "train_loss_llm": 0.0, "grad_norm": 0.06582615524530411, "global_step": 13559, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03305063396692276, "train_loss_bc": 0.03304479271173477, "train_loss_llm": 0.000584036111831665, "grad_norm": 0.06842243671417236, "global_step": 13560, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03463413566350937, "train_loss_bc": 0.03463413566350937, "train_loss_llm": 0.0, "grad_norm": 0.024196164682507515, "global_step": 13561, "epoch": 376, "lr": 0.001} +{"train_loss": 0.037527453154325485, "train_loss_bc": 0.037527453154325485, "train_loss_llm": 0.0, "grad_norm": 0.0393158458173275, "global_step": 13562, "epoch": 376, "lr": 0.001} +{"train_loss": 0.039438147097826004, "train_loss_bc": 0.039438147097826004, "train_loss_llm": 0.0, "grad_norm": 0.04909266158938408, "global_step": 13563, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03493981435894966, "train_loss_bc": 0.03493981435894966, "train_loss_llm": 0.0, "grad_norm": 0.06178790330886841, "global_step": 13564, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03478823974728584, "train_loss_bc": 0.03478823974728584, "train_loss_llm": 0.0, "grad_norm": 0.02425118163228035, "global_step": 13565, "epoch": 376, "lr": 0.001} +{"train_loss": 0.030321354046463966, "train_loss_bc": 0.03031374327838421, "train_loss_llm": 0.0007610023021697998, "grad_norm": 0.030757607892155647, "global_step": 13566, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03285026177763939, "train_loss_bc": 0.03285026177763939, "train_loss_llm": 0.0, "grad_norm": 0.04435277357697487, "global_step": 13567, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03965868428349495, "train_loss_bc": 0.039646368473768234, "train_loss_llm": 0.001231551170349121, "grad_norm": 0.058958157896995544, "global_step": 13568, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03299224004149437, "train_loss_bc": 0.03296252712607384, "train_loss_llm": 0.0029713213443756104, "grad_norm": 0.021922389045357704, "global_step": 13569, "epoch": 376, "lr": 0.001} +{"train_loss": 0.04040847346186638, "train_loss_bc": 0.04039562866091728, "train_loss_llm": 0.0012843608856201172, "grad_norm": 0.04069417715072632, "global_step": 13570, "epoch": 376, "lr": 0.001} +{"train_loss": 0.03658313718107012, "train_loss_bc": 0.03260001540184021, "train_loss_llm": 0.0, "grad_norm": 0.058308981359004974, "global_step": 13571, "epoch": 376, "lr": 0.001, "val_loss": 0.06753100454807281} +{"train_loss": 0.04309553653001785, "train_loss_bc": 0.04309553653001785, "train_loss_llm": 0.0, "grad_norm": 0.07512136548757553, "global_step": 13572, "epoch": 377, "lr": 0.001} +{"train_loss": 0.029924165457487106, "train_loss_bc": 0.029922431334853172, "train_loss_llm": 0.00017341971397399902, "grad_norm": 0.016540443524718285, "global_step": 13573, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03051532804965973, "train_loss_bc": 0.030505582690238953, "train_loss_llm": 0.0009744465351104736, "grad_norm": 0.03740235045552254, "global_step": 13574, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03980260714888573, "train_loss_bc": 0.03979496285319328, "train_loss_llm": 0.0007643699645996094, "grad_norm": 0.04991181939840317, "global_step": 13575, "epoch": 377, "lr": 0.001} +{"train_loss": 0.038810163736343384, "train_loss_bc": 0.03879990056157112, "train_loss_llm": 0.001026451587677002, "grad_norm": 0.06275288760662079, "global_step": 13576, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03923936188220978, "train_loss_bc": 0.039219047874212265, "train_loss_llm": 0.0020315051078796387, "grad_norm": 0.02445199154317379, "global_step": 13577, "epoch": 377, "lr": 0.001} +{"train_loss": 0.035564228892326355, "train_loss_bc": 0.035564228892326355, "train_loss_llm": 0.0, "grad_norm": 0.040694043040275574, "global_step": 13578, "epoch": 377, "lr": 0.001} +{"train_loss": 0.037288814783096313, "train_loss_bc": 0.03728878125548363, "train_loss_llm": 3.427267074584961e-06, "grad_norm": 0.053882207721471786, "global_step": 13579, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03677384927868843, "train_loss_bc": 0.03677384927868843, "train_loss_llm": 0.0, "grad_norm": 0.06256061792373657, "global_step": 13580, "epoch": 377, "lr": 0.001} +{"train_loss": 0.036501847207546234, "train_loss_bc": 0.03649027273058891, "train_loss_llm": 0.001157522201538086, "grad_norm": 0.02703959494829178, "global_step": 13581, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03867644444108009, "train_loss_bc": 0.03867644444108009, "train_loss_llm": 0.0, "grad_norm": 0.04151635989546776, "global_step": 13582, "epoch": 377, "lr": 0.001} +{"train_loss": 0.036177609115839005, "train_loss_bc": 0.03617667406797409, "train_loss_llm": 9.363889694213867e-05, "grad_norm": 0.05105368047952652, "global_step": 13583, "epoch": 377, "lr": 0.001} +{"train_loss": 0.04241054132580757, "train_loss_bc": 0.04240649938583374, "train_loss_llm": 0.0004042983055114746, "grad_norm": 0.06895148754119873, "global_step": 13584, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03764858841896057, "train_loss_bc": 0.03764858841896057, "train_loss_llm": 0.0, "grad_norm": 0.020313028246164322, "global_step": 13585, "epoch": 377, "lr": 0.001} +{"train_loss": 0.04155903309583664, "train_loss_bc": 0.041550494730472565, "train_loss_llm": 0.0008540153503417969, "grad_norm": 0.043803099542856216, "global_step": 13586, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03921298310160637, "train_loss_bc": 0.03921298310160637, "train_loss_llm": 0.0, "grad_norm": 0.04776845499873161, "global_step": 13587, "epoch": 377, "lr": 0.001} +{"train_loss": 0.036095961928367615, "train_loss_bc": 0.03608621284365654, "train_loss_llm": 0.0009748339653015137, "grad_norm": 0.061074305325746536, "global_step": 13588, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03376108035445213, "train_loss_bc": 0.03376108035445213, "train_loss_llm": 0.0, "grad_norm": 0.02234290912747383, "global_step": 13589, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03578326478600502, "train_loss_bc": 0.03578326478600502, "train_loss_llm": 0.0, "grad_norm": 0.036410462111234665, "global_step": 13590, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03962397947907448, "train_loss_bc": 0.03962397947907448, "train_loss_llm": 0.0, "grad_norm": 0.047751981765031815, "global_step": 13591, "epoch": 377, "lr": 0.001} +{"train_loss": 0.04179657623171806, "train_loss_bc": 0.04179593175649643, "train_loss_llm": 6.437301635742188e-05, "grad_norm": 0.059152308851480484, "global_step": 13592, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03255903348326683, "train_loss_bc": 0.03255903348326683, "train_loss_llm": 0.0, "grad_norm": 0.02886693738400936, "global_step": 13593, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03559102490544319, "train_loss_bc": 0.03559102490544319, "train_loss_llm": 0.0, "grad_norm": 0.03626687452197075, "global_step": 13594, "epoch": 377, "lr": 0.001} +{"train_loss": 0.034958045929670334, "train_loss_bc": 0.034958045929670334, "train_loss_llm": 0.0, "grad_norm": 0.04875932261347771, "global_step": 13595, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03213891014456749, "train_loss_bc": 0.03213891014456749, "train_loss_llm": 0.0, "grad_norm": 0.06775865703821182, "global_step": 13596, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03644021600484848, "train_loss_bc": 0.03644021600484848, "train_loss_llm": 0.0, "grad_norm": 0.023186122998595238, "global_step": 13597, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03680489584803581, "train_loss_bc": 0.03680489584803581, "train_loss_llm": 0.0, "grad_norm": 0.036984413862228394, "global_step": 13598, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03487076982855797, "train_loss_bc": 0.03487076982855797, "train_loss_llm": 0.0, "grad_norm": 0.057611577212810516, "global_step": 13599, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03274398669600487, "train_loss_bc": 0.03274398669600487, "train_loss_llm": 0.0, "grad_norm": 0.07293584197759628, "global_step": 13600, "epoch": 377, "lr": 0.001} +{"train_loss": 0.032293856143951416, "train_loss_bc": 0.032293856143951416, "train_loss_llm": 0.0, "grad_norm": 0.018945803865790367, "global_step": 13601, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03347432613372803, "train_loss_bc": 0.033457983285188675, "train_loss_llm": 0.0016342997550964355, "grad_norm": 0.036153316497802734, "global_step": 13602, "epoch": 377, "lr": 0.001} +{"train_loss": 0.04737556353211403, "train_loss_bc": 0.047365542501211166, "train_loss_llm": 0.0010021328926086426, "grad_norm": 0.0427577868103981, "global_step": 13603, "epoch": 377, "lr": 0.001} +{"train_loss": 0.036255884915590286, "train_loss_bc": 0.036255884915590286, "train_loss_llm": 0.0, "grad_norm": 0.04878069460391998, "global_step": 13604, "epoch": 377, "lr": 0.001} +{"train_loss": 0.032532256096601486, "train_loss_bc": 0.032532256096601486, "train_loss_llm": 0.0, "grad_norm": 0.02209308184683323, "global_step": 13605, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03368595615029335, "train_loss_bc": 0.03368595615029335, "train_loss_llm": 0.0, "grad_norm": 0.026568103581666946, "global_step": 13606, "epoch": 377, "lr": 0.001} +{"train_loss": 0.03638098450998465, "train_loss_bc": 0.027728751301765442, "train_loss_llm": 0.0, "grad_norm": 0.03715236857533455, "global_step": 13607, "epoch": 377, "lr": 0.001, "val_loss": 0.06609530001878738} +{"train_loss": 0.03747820481657982, "train_loss_bc": 0.03747820481657982, "train_loss_llm": 0.0, "grad_norm": 0.04887068644165993, "global_step": 13608, "epoch": 378, "lr": 0.001} +{"train_loss": 0.030601847916841507, "train_loss_bc": 0.030601847916841507, "train_loss_llm": 0.0, "grad_norm": 0.023780720308423042, "global_step": 13609, "epoch": 378, "lr": 0.001} +{"train_loss": 0.040153078734874725, "train_loss_bc": 0.040153078734874725, "train_loss_llm": 0.0, "grad_norm": 0.041496165096759796, "global_step": 13610, "epoch": 378, "lr": 0.001} +{"train_loss": 0.034784454852342606, "train_loss_bc": 0.034784454852342606, "train_loss_llm": 0.0, "grad_norm": 0.05489453673362732, "global_step": 13611, "epoch": 378, "lr": 0.001} +{"train_loss": 0.037823352962732315, "train_loss_bc": 0.037823352962732315, "train_loss_llm": 0.0, "grad_norm": 0.06451196223497391, "global_step": 13612, "epoch": 378, "lr": 0.001} +{"train_loss": 0.040139682590961456, "train_loss_bc": 0.040134742856025696, "train_loss_llm": 0.0004938244819641113, "grad_norm": 0.027287179604172707, "global_step": 13613, "epoch": 378, "lr": 0.001} +{"train_loss": 0.046170834451913834, "train_loss_bc": 0.04615268483757973, "train_loss_llm": 0.0018149912357330322, "grad_norm": 0.04658481478691101, "global_step": 13614, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03441065177321434, "train_loss_bc": 0.03438954055309296, "train_loss_llm": 0.0021109580993652344, "grad_norm": 0.06663063913583755, "global_step": 13615, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03648858144879341, "train_loss_bc": 0.03648858144879341, "train_loss_llm": 0.0, "grad_norm": 0.07757934182882309, "global_step": 13616, "epoch": 378, "lr": 0.001} +{"train_loss": 0.036074358969926834, "train_loss_bc": 0.036074358969926834, "train_loss_llm": 0.0, "grad_norm": 0.021720532327890396, "global_step": 13617, "epoch": 378, "lr": 0.001} +{"train_loss": 0.04206688702106476, "train_loss_bc": 0.042064376175403595, "train_loss_llm": 0.00025093555450439453, "grad_norm": 0.037002917379140854, "global_step": 13618, "epoch": 378, "lr": 0.001} +{"train_loss": 0.030973264947533607, "train_loss_bc": 0.030962323769927025, "train_loss_llm": 0.0010941922664642334, "grad_norm": 0.050119657069444656, "global_step": 13619, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03535279259085655, "train_loss_bc": 0.03535279259085655, "train_loss_llm": 0.0, "grad_norm": 0.07212543487548828, "global_step": 13620, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03996332734823227, "train_loss_bc": 0.03996332734823227, "train_loss_llm": 0.0, "grad_norm": 0.0231899656355381, "global_step": 13621, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03741050511598587, "train_loss_bc": 0.03741050511598587, "train_loss_llm": 0.0, "grad_norm": 0.03583132103085518, "global_step": 13622, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03841560706496239, "train_loss_bc": 0.03841560706496239, "train_loss_llm": 0.0, "grad_norm": 0.04857682064175606, "global_step": 13623, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03727949783205986, "train_loss_bc": 0.03727949783205986, "train_loss_llm": 0.0, "grad_norm": 0.055171720683574677, "global_step": 13624, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03683244436979294, "train_loss_bc": 0.036829765886068344, "train_loss_llm": 0.0002677738666534424, "grad_norm": 0.021594125777482986, "global_step": 13625, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03955131396651268, "train_loss_bc": 0.03955131396651268, "train_loss_llm": 0.0, "grad_norm": 0.03877907246351242, "global_step": 13626, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03462721034884453, "train_loss_bc": 0.03462721034884453, "train_loss_llm": 0.0, "grad_norm": 0.04906029254198074, "global_step": 13627, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03856942802667618, "train_loss_bc": 0.03856942802667618, "train_loss_llm": 0.0, "grad_norm": 0.05840115249156952, "global_step": 13628, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03583478927612305, "train_loss_bc": 0.03582330048084259, "train_loss_llm": 0.0011489689350128174, "grad_norm": 0.02549164742231369, "global_step": 13629, "epoch": 378, "lr": 0.001} +{"train_loss": 0.04136468097567558, "train_loss_bc": 0.04136468097567558, "train_loss_llm": 0.0, "grad_norm": 0.041768621653318405, "global_step": 13630, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03634020686149597, "train_loss_bc": 0.03634020686149597, "train_loss_llm": 0.0, "grad_norm": 0.0522988885641098, "global_step": 13631, "epoch": 378, "lr": 0.001} +{"train_loss": 0.037008948624134064, "train_loss_bc": 0.037008948624134064, "train_loss_llm": 0.0, "grad_norm": 0.0576217882335186, "global_step": 13632, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03200451657176018, "train_loss_bc": 0.03200451657176018, "train_loss_llm": 0.0, "grad_norm": 0.024307645857334137, "global_step": 13633, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03473711758852005, "train_loss_bc": 0.03473711758852005, "train_loss_llm": 0.0, "grad_norm": 0.03990265354514122, "global_step": 13634, "epoch": 378, "lr": 0.001} +{"train_loss": 0.037899449467659, "train_loss_bc": 0.03788601607084274, "train_loss_llm": 0.0013431906700134277, "grad_norm": 0.05431520566344261, "global_step": 13635, "epoch": 378, "lr": 0.001} +{"train_loss": 0.0367656834423542, "train_loss_bc": 0.036760054528713226, "train_loss_llm": 0.0005628466606140137, "grad_norm": 0.07197781652212143, "global_step": 13636, "epoch": 378, "lr": 0.001} +{"train_loss": 0.035168007016181946, "train_loss_bc": 0.035168007016181946, "train_loss_llm": 0.0, "grad_norm": 0.025458624586462975, "global_step": 13637, "epoch": 378, "lr": 0.001} +{"train_loss": 0.030846906825900078, "train_loss_bc": 0.030846906825900078, "train_loss_llm": 0.0, "grad_norm": 0.0379331149160862, "global_step": 13638, "epoch": 378, "lr": 0.001} +{"train_loss": 0.034481167793273926, "train_loss_bc": 0.034481167793273926, "train_loss_llm": 0.0, "grad_norm": 0.04828651621937752, "global_step": 13639, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03428510203957558, "train_loss_bc": 0.03428510203957558, "train_loss_llm": 0.0, "grad_norm": 0.06350620836019516, "global_step": 13640, "epoch": 378, "lr": 0.001} +{"train_loss": 0.030897479504346848, "train_loss_bc": 0.030897479504346848, "train_loss_llm": 0.0, "grad_norm": 0.022620052099227905, "global_step": 13641, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03765887767076492, "train_loss_bc": 0.03765887767076492, "train_loss_llm": 0.0, "grad_norm": 0.03483518213033676, "global_step": 13642, "epoch": 378, "lr": 0.001} +{"train_loss": 0.03655359635336532, "train_loss_bc": 0.03546920791268349, "train_loss_llm": 0.0, "grad_norm": 0.04046535864472389, "global_step": 13643, "epoch": 378, "lr": 0.001, "val_loss": 0.06742581725120544} +{"train_loss": 0.03436809405684471, "train_loss_bc": 0.03436809405684471, "train_loss_llm": 0.0, "grad_norm": 0.04840052127838135, "global_step": 13644, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03580217808485031, "train_loss_bc": 0.03580217808485031, "train_loss_llm": 0.0, "grad_norm": 0.020894106477499008, "global_step": 13645, "epoch": 379, "lr": 0.001} +{"train_loss": 0.04137536883354187, "train_loss_bc": 0.04137536883354187, "train_loss_llm": 0.0, "grad_norm": 0.030159393325448036, "global_step": 13646, "epoch": 379, "lr": 0.001} +{"train_loss": 0.036564506590366364, "train_loss_bc": 0.036564506590366364, "train_loss_llm": 0.0, "grad_norm": 0.04414084926247597, "global_step": 13647, "epoch": 379, "lr": 0.001} +{"train_loss": 0.0348932221531868, "train_loss_bc": 0.03488084301352501, "train_loss_llm": 0.0012380480766296387, "grad_norm": 0.05235786363482475, "global_step": 13648, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03742922097444534, "train_loss_bc": 0.03742922097444534, "train_loss_llm": 0.0, "grad_norm": 0.025217751041054726, "global_step": 13649, "epoch": 379, "lr": 0.001} +{"train_loss": 0.036784227937459946, "train_loss_bc": 0.036784227937459946, "train_loss_llm": 0.0, "grad_norm": 0.03343835100531578, "global_step": 13650, "epoch": 379, "lr": 0.001} +{"train_loss": 0.0341210700571537, "train_loss_bc": 0.034105319529771805, "train_loss_llm": 0.001575171947479248, "grad_norm": 0.041962966322898865, "global_step": 13651, "epoch": 379, "lr": 0.001} +{"train_loss": 0.02873765304684639, "train_loss_bc": 0.02873765304684639, "train_loss_llm": 0.0, "grad_norm": 0.05031491816043854, "global_step": 13652, "epoch": 379, "lr": 0.001} +{"train_loss": 0.039621926844120026, "train_loss_bc": 0.039621926844120026, "train_loss_llm": 0.0, "grad_norm": 0.02810426987707615, "global_step": 13653, "epoch": 379, "lr": 0.001} +{"train_loss": 0.035071056336164474, "train_loss_bc": 0.035071056336164474, "train_loss_llm": 0.0, "grad_norm": 0.0413723848760128, "global_step": 13654, "epoch": 379, "lr": 0.001} +{"train_loss": 0.04206009954214096, "train_loss_bc": 0.04206009954214096, "train_loss_llm": 0.0, "grad_norm": 0.0622384138405323, "global_step": 13655, "epoch": 379, "lr": 0.001} +{"train_loss": 0.034988317638635635, "train_loss_bc": 0.034988317638635635, "train_loss_llm": 0.0, "grad_norm": 0.07732430100440979, "global_step": 13656, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03303159400820732, "train_loss_bc": 0.03303159400820732, "train_loss_llm": 0.0, "grad_norm": 0.021294711157679558, "global_step": 13657, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03638680279254913, "train_loss_bc": 0.03638680279254913, "train_loss_llm": 0.0, "grad_norm": 0.035607367753982544, "global_step": 13658, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03862212598323822, "train_loss_bc": 0.03862212598323822, "train_loss_llm": 0.0, "grad_norm": 0.05448177084326744, "global_step": 13659, "epoch": 379, "lr": 0.001} +{"train_loss": 0.036463670432567596, "train_loss_bc": 0.03645964711904526, "train_loss_llm": 0.0004025101661682129, "grad_norm": 0.0639311820268631, "global_step": 13660, "epoch": 379, "lr": 0.001} +{"train_loss": 0.034497570246458054, "train_loss_bc": 0.034497570246458054, "train_loss_llm": 0.0, "grad_norm": 0.02713579684495926, "global_step": 13661, "epoch": 379, "lr": 0.001} +{"train_loss": 0.04358329996466637, "train_loss_bc": 0.04358329996466637, "train_loss_llm": 0.0, "grad_norm": 0.04293670505285263, "global_step": 13662, "epoch": 379, "lr": 0.001} +{"train_loss": 0.039604783058166504, "train_loss_bc": 0.03959433361887932, "train_loss_llm": 0.0010449886322021484, "grad_norm": 0.057808976620435715, "global_step": 13663, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03241251781582832, "train_loss_bc": 0.03241251781582832, "train_loss_llm": 0.0, "grad_norm": 0.0715218335390091, "global_step": 13664, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03503306955099106, "train_loss_bc": 0.03503306955099106, "train_loss_llm": 0.0, "grad_norm": 0.024074038490653038, "global_step": 13665, "epoch": 379, "lr": 0.001} +{"train_loss": 0.04138690605759621, "train_loss_bc": 0.04138404503464699, "train_loss_llm": 0.0002860724925994873, "grad_norm": 0.03499157354235649, "global_step": 13666, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03612479567527771, "train_loss_bc": 0.03612479567527771, "train_loss_llm": 0.0, "grad_norm": 0.050028905272483826, "global_step": 13667, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03603246808052063, "train_loss_bc": 0.03603246808052063, "train_loss_llm": 0.0, "grad_norm": 0.06926044821739197, "global_step": 13668, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03176964074373245, "train_loss_bc": 0.03176901116967201, "train_loss_llm": 6.306171417236328e-05, "grad_norm": 0.01886138878762722, "global_step": 13669, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03994448110461235, "train_loss_bc": 0.03992687165737152, "train_loss_llm": 0.0017609000205993652, "grad_norm": 0.03686204180121422, "global_step": 13670, "epoch": 379, "lr": 0.001} +{"train_loss": 0.033277951180934906, "train_loss_bc": 0.03325384110212326, "train_loss_llm": 0.002410888671875, "grad_norm": 0.04418220371007919, "global_step": 13671, "epoch": 379, "lr": 0.001} +{"train_loss": 0.032447248697280884, "train_loss_bc": 0.032447248697280884, "train_loss_llm": 0.0, "grad_norm": 0.0606970451772213, "global_step": 13672, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03918367996811867, "train_loss_bc": 0.03918082267045975, "train_loss_llm": 0.00028565526008605957, "grad_norm": 0.020919911563396454, "global_step": 13673, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03411517292261124, "train_loss_bc": 0.03411517292261124, "train_loss_llm": 0.0, "grad_norm": 0.03312158212065697, "global_step": 13674, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03403657302260399, "train_loss_bc": 0.03403657302260399, "train_loss_llm": 0.0, "grad_norm": 0.04534966126084328, "global_step": 13675, "epoch": 379, "lr": 0.001} +{"train_loss": 0.031097277998924255, "train_loss_bc": 0.031097277998924255, "train_loss_llm": 0.0, "grad_norm": 0.053491637110710144, "global_step": 13676, "epoch": 379, "lr": 0.001} +{"train_loss": 0.04114141687750816, "train_loss_bc": 0.04113556072115898, "train_loss_llm": 0.0005857348442077637, "grad_norm": 0.027278002351522446, "global_step": 13677, "epoch": 379, "lr": 0.001} +{"train_loss": 0.041792139410972595, "train_loss_bc": 0.041792139410972595, "train_loss_llm": 0.0, "grad_norm": 0.04764461889863014, "global_step": 13678, "epoch": 379, "lr": 0.001} +{"train_loss": 0.03641961980611086, "train_loss_bc": 0.03730418533086777, "train_loss_llm": 0.0, "grad_norm": 0.05855987221002579, "global_step": 13679, "epoch": 379, "lr": 0.001, "val_loss": 0.0659337267279625} +{"train_loss": 0.03381099924445152, "train_loss_bc": 0.03381099924445152, "train_loss_llm": 0.0, "grad_norm": 0.07269023358821869, "global_step": 13680, "epoch": 380, "lr": 0.001} +{"train_loss": 0.04095645248889923, "train_loss_bc": 0.040935687720775604, "train_loss_llm": 0.002076387405395508, "grad_norm": 0.02527535706758499, "global_step": 13681, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03471669927239418, "train_loss_bc": 0.03471669927239418, "train_loss_llm": 0.0, "grad_norm": 0.036218006163835526, "global_step": 13682, "epoch": 380, "lr": 0.001} +{"train_loss": 0.02956593595445156, "train_loss_bc": 0.02956593595445156, "train_loss_llm": 0.0, "grad_norm": 0.0490838885307312, "global_step": 13683, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03454320505261421, "train_loss_bc": 0.03454320505261421, "train_loss_llm": 0.0, "grad_norm": 0.05053475871682167, "global_step": 13684, "epoch": 380, "lr": 0.001} +{"train_loss": 0.039233796298503876, "train_loss_bc": 0.039219122380018234, "train_loss_llm": 0.0014674961566925049, "grad_norm": 0.024072853848338127, "global_step": 13685, "epoch": 380, "lr": 0.001} +{"train_loss": 0.04015600308775902, "train_loss_bc": 0.04015600308775902, "train_loss_llm": 0.0, "grad_norm": 0.036728233098983765, "global_step": 13686, "epoch": 380, "lr": 0.001} +{"train_loss": 0.031290203332901, "train_loss_bc": 0.03128216788172722, "train_loss_llm": 0.0008035898208618164, "grad_norm": 0.04601329192519188, "global_step": 13687, "epoch": 380, "lr": 0.001} +{"train_loss": 0.0389324352145195, "train_loss_bc": 0.03892626240849495, "train_loss_llm": 0.0006174445152282715, "grad_norm": 0.05593602731823921, "global_step": 13688, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03456490486860275, "train_loss_bc": 0.0345437154173851, "train_loss_llm": 0.002119004726409912, "grad_norm": 0.030901839956641197, "global_step": 13689, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03750702738761902, "train_loss_bc": 0.037498265504837036, "train_loss_llm": 0.0008763372898101807, "grad_norm": 0.04538940638303757, "global_step": 13690, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03196241706609726, "train_loss_bc": 0.03195472061634064, "train_loss_llm": 0.0007698237895965576, "grad_norm": 0.06276633590459824, "global_step": 13691, "epoch": 380, "lr": 0.001} +{"train_loss": 0.0388229675590992, "train_loss_bc": 0.03881067410111427, "train_loss_llm": 0.0012293756008148193, "grad_norm": 0.0747465118765831, "global_step": 13692, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03700222074985504, "train_loss_bc": 0.03700222074985504, "train_loss_llm": 0.0, "grad_norm": 0.023946119472384453, "global_step": 13693, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03522831201553345, "train_loss_bc": 0.03522831201553345, "train_loss_llm": 0.0, "grad_norm": 0.044147949665784836, "global_step": 13694, "epoch": 380, "lr": 0.001} +{"train_loss": 0.0393231101334095, "train_loss_bc": 0.0393231101334095, "train_loss_llm": 0.0, "grad_norm": 0.06691624969244003, "global_step": 13695, "epoch": 380, "lr": 0.001} +{"train_loss": 0.0391753688454628, "train_loss_bc": 0.0391753688454628, "train_loss_llm": 0.0, "grad_norm": 0.08226115256547928, "global_step": 13696, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03558463230729103, "train_loss_bc": 0.03558306023478508, "train_loss_llm": 0.00015738606452941895, "grad_norm": 0.020103562623262405, "global_step": 13697, "epoch": 380, "lr": 0.001} +{"train_loss": 0.040024932473897934, "train_loss_bc": 0.040024932473897934, "train_loss_llm": 0.0, "grad_norm": 0.03815460205078125, "global_step": 13698, "epoch": 380, "lr": 0.001} +{"train_loss": 0.034364838153123856, "train_loss_bc": 0.034364838153123856, "train_loss_llm": 0.0, "grad_norm": 0.04666886478662491, "global_step": 13699, "epoch": 380, "lr": 0.001} +{"train_loss": 0.04163243621587753, "train_loss_bc": 0.04163243621587753, "train_loss_llm": 0.0, "grad_norm": 0.07268484681844711, "global_step": 13700, "epoch": 380, "lr": 0.001} +{"train_loss": 0.022596413269639015, "train_loss_bc": 0.022595131769776344, "train_loss_llm": 0.00012814998626708984, "grad_norm": 0.01989501714706421, "global_step": 13701, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03697260096669197, "train_loss_bc": 0.03697260096669197, "train_loss_llm": 0.0, "grad_norm": 0.03354405611753464, "global_step": 13702, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03331562876701355, "train_loss_bc": 0.03331562876701355, "train_loss_llm": 0.0, "grad_norm": 0.038199979811906815, "global_step": 13703, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03221147879958153, "train_loss_bc": 0.032204724848270416, "train_loss_llm": 0.0006752312183380127, "grad_norm": 0.045998815447092056, "global_step": 13704, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03700847923755646, "train_loss_bc": 0.03700847923755646, "train_loss_llm": 0.0, "grad_norm": 0.02718770131468773, "global_step": 13705, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03720661252737045, "train_loss_bc": 0.03720661252737045, "train_loss_llm": 0.0, "grad_norm": 0.04552357271313667, "global_step": 13706, "epoch": 380, "lr": 0.001} +{"train_loss": 0.028556272387504578, "train_loss_bc": 0.028542017564177513, "train_loss_llm": 0.0014255642890930176, "grad_norm": 0.053669970482587814, "global_step": 13707, "epoch": 380, "lr": 0.001} +{"train_loss": 0.036003969609737396, "train_loss_bc": 0.036003969609737396, "train_loss_llm": 0.0, "grad_norm": 0.07339940965175629, "global_step": 13708, "epoch": 380, "lr": 0.001} +{"train_loss": 0.04239796847105026, "train_loss_bc": 0.04234982654452324, "train_loss_llm": 0.004814058542251587, "grad_norm": 0.02433336153626442, "global_step": 13709, "epoch": 380, "lr": 0.001} +{"train_loss": 0.04011903330683708, "train_loss_bc": 0.04010043665766716, "train_loss_llm": 0.0018595457077026367, "grad_norm": 0.03488615155220032, "global_step": 13710, "epoch": 380, "lr": 0.001} +{"train_loss": 0.04554635286331177, "train_loss_bc": 0.04554635286331177, "train_loss_llm": 0.0, "grad_norm": 0.049601875245571136, "global_step": 13711, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03976430743932724, "train_loss_bc": 0.03976430743932724, "train_loss_llm": 0.0, "grad_norm": 0.06468706578016281, "global_step": 13712, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03790178522467613, "train_loss_bc": 0.03790178522467613, "train_loss_llm": 0.0, "grad_norm": 0.03147096931934357, "global_step": 13713, "epoch": 380, "lr": 0.001} +{"train_loss": 0.03859112784266472, "train_loss_bc": 0.03859112784266472, "train_loss_llm": 0.0, "grad_norm": 0.0408288799226284, "global_step": 13714, "epoch": 380, "lr": 0.001} +{"train_loss": 0.0363812441420224, "train_loss_bc": 0.0331338606774807, "train_loss_llm": 0.0, "grad_norm": 0.05484545975923538, "global_step": 13715, "epoch": 380, "lr": 0.001, "train/cumulative_reward": 74.72278297935694, "train/mean_score": 8.793395653123435, "train/success_rate": 0.1, "test/cumulative_reward": 39.60694822013922, "test/mean_score": 8.919740429104143, "test/success_rate": 0.2, "val_loss": 0.06628889590501785, "train_action_mse_error": 0.03938654810190201} +{"train_loss": 0.0373857207596302, "train_loss_bc": 0.0373857207596302, "train_loss_llm": 0.0, "grad_norm": 0.06636892259120941, "global_step": 13716, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03950868546962738, "train_loss_bc": 0.03950868546962738, "train_loss_llm": 0.0, "grad_norm": 0.024275122210383415, "global_step": 13717, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03143765777349472, "train_loss_bc": 0.03143348544836044, "train_loss_llm": 0.000417172908782959, "grad_norm": 0.03702882304787636, "global_step": 13718, "epoch": 381, "lr": 0.001} +{"train_loss": 0.027342556044459343, "train_loss_bc": 0.027342556044459343, "train_loss_llm": 0.0, "grad_norm": 0.05403263494372368, "global_step": 13719, "epoch": 381, "lr": 0.001} +{"train_loss": 0.038935188204050064, "train_loss_bc": 0.038935188204050064, "train_loss_llm": 0.0, "grad_norm": 0.07768912613391876, "global_step": 13720, "epoch": 381, "lr": 0.001} +{"train_loss": 0.039894234389066696, "train_loss_bc": 0.039894234389066696, "train_loss_llm": 0.0, "grad_norm": 0.024719897657632828, "global_step": 13721, "epoch": 381, "lr": 0.001} +{"train_loss": 0.0352117158472538, "train_loss_bc": 0.0352117158472538, "train_loss_llm": 0.0, "grad_norm": 0.037080083042383194, "global_step": 13722, "epoch": 381, "lr": 0.001} +{"train_loss": 0.04334867373108864, "train_loss_bc": 0.04334867373108864, "train_loss_llm": 0.0, "grad_norm": 0.05368833988904953, "global_step": 13723, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03768101707100868, "train_loss_bc": 0.03768101707100868, "train_loss_llm": 0.0, "grad_norm": 0.07193455845117569, "global_step": 13724, "epoch": 381, "lr": 0.001} +{"train_loss": 0.042570240795612335, "train_loss_bc": 0.042570240795612335, "train_loss_llm": 0.0, "grad_norm": 0.02500893361866474, "global_step": 13725, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03598489612340927, "train_loss_bc": 0.03598489612340927, "train_loss_llm": 0.0, "grad_norm": 0.038224052637815475, "global_step": 13726, "epoch": 381, "lr": 0.001} +{"train_loss": 0.033460818231105804, "train_loss_bc": 0.033460818231105804, "train_loss_llm": 0.0, "grad_norm": 0.05390698462724686, "global_step": 13727, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03329901397228241, "train_loss_bc": 0.033288002014160156, "train_loss_llm": 0.0011012256145477295, "grad_norm": 0.06486358493566513, "global_step": 13728, "epoch": 381, "lr": 0.001} +{"train_loss": 0.037791475653648376, "train_loss_bc": 0.037791475653648376, "train_loss_llm": 0.0, "grad_norm": 0.02770106866955757, "global_step": 13729, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03540009632706642, "train_loss_bc": 0.035394877195358276, "train_loss_llm": 0.0005218088626861572, "grad_norm": 0.04543512314558029, "global_step": 13730, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03345944359898567, "train_loss_bc": 0.03345556557178497, "train_loss_llm": 0.00038763880729675293, "grad_norm": 0.06451180577278137, "global_step": 13731, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03230791538953781, "train_loss_bc": 0.03230791538953781, "train_loss_llm": 0.0, "grad_norm": 0.07021699845790863, "global_step": 13732, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03853749856352806, "train_loss_bc": 0.03853749856352806, "train_loss_llm": 0.0, "grad_norm": 0.023128895089030266, "global_step": 13733, "epoch": 381, "lr": 0.001} +{"train_loss": 0.035483382642269135, "train_loss_bc": 0.035483382642269135, "train_loss_llm": 0.0, "grad_norm": 0.03982848301529884, "global_step": 13734, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03715340420603752, "train_loss_bc": 0.03715340420603752, "train_loss_llm": 0.0, "grad_norm": 0.05703233554959297, "global_step": 13735, "epoch": 381, "lr": 0.001} +{"train_loss": 0.032605595886707306, "train_loss_bc": 0.0325959138572216, "train_loss_llm": 0.0009681880474090576, "grad_norm": 0.06203075125813484, "global_step": 13736, "epoch": 381, "lr": 0.001} +{"train_loss": 0.032202187925577164, "train_loss_bc": 0.03220200166106224, "train_loss_llm": 1.8775463104248047e-05, "grad_norm": 0.024816201999783516, "global_step": 13737, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03475048393011093, "train_loss_bc": 0.034744225442409515, "train_loss_llm": 0.00062599778175354, "grad_norm": 0.04021334648132324, "global_step": 13738, "epoch": 381, "lr": 0.001} +{"train_loss": 0.035532236099243164, "train_loss_bc": 0.035532236099243164, "train_loss_llm": 0.0, "grad_norm": 0.05662064626812935, "global_step": 13739, "epoch": 381, "lr": 0.001} +{"train_loss": 0.035647839307785034, "train_loss_bc": 0.035647839307785034, "train_loss_llm": 0.0, "grad_norm": 0.07229214161634445, "global_step": 13740, "epoch": 381, "lr": 0.001} +{"train_loss": 0.032914046198129654, "train_loss_bc": 0.032914046198129654, "train_loss_llm": 0.0, "grad_norm": 0.027033524587750435, "global_step": 13741, "epoch": 381, "lr": 0.001} +{"train_loss": 0.033604301512241364, "train_loss_bc": 0.03358267992734909, "train_loss_llm": 0.002162069082260132, "grad_norm": 0.04357421025633812, "global_step": 13742, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03865709528326988, "train_loss_bc": 0.03865709528326988, "train_loss_llm": 0.0, "grad_norm": 0.059300441294908524, "global_step": 13743, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03895465284585953, "train_loss_bc": 0.03895465284585953, "train_loss_llm": 0.0, "grad_norm": 0.07806237787008286, "global_step": 13744, "epoch": 381, "lr": 0.001} +{"train_loss": 0.0415165014564991, "train_loss_bc": 0.0415165014564991, "train_loss_llm": 0.0, "grad_norm": 0.02819759026169777, "global_step": 13745, "epoch": 381, "lr": 0.001} +{"train_loss": 0.04074004665017128, "train_loss_bc": 0.04074004665017128, "train_loss_llm": 0.0, "grad_norm": 0.04369954764842987, "global_step": 13746, "epoch": 381, "lr": 0.001} +{"train_loss": 0.042223334312438965, "train_loss_bc": 0.042219944298267365, "train_loss_llm": 0.00033912062644958496, "grad_norm": 0.06185464560985565, "global_step": 13747, "epoch": 381, "lr": 0.001} +{"train_loss": 0.0361064113676548, "train_loss_bc": 0.0361064113676548, "train_loss_llm": 0.0, "grad_norm": 0.0817626342177391, "global_step": 13748, "epoch": 381, "lr": 0.001} +{"train_loss": 0.03615709766745567, "train_loss_bc": 0.0361514575779438, "train_loss_llm": 0.0005640983581542969, "grad_norm": 0.018783748149871826, "global_step": 13749, "epoch": 381, "lr": 0.001} +{"train_loss": 0.034379832446575165, "train_loss_bc": 0.034379832446575165, "train_loss_llm": 0.0, "grad_norm": 0.029793303459882736, "global_step": 13750, "epoch": 381, "lr": 0.001} +{"train_loss": 0.036521577121069036, "train_loss_bc": 0.04259147867560387, "train_loss_llm": 0.0, "grad_norm": 0.049422767013311386, "global_step": 13751, "epoch": 381, "lr": 0.001, "val_loss": 0.06718482822179794} +{"train_loss": 0.031942978501319885, "train_loss_bc": 0.031942978501319885, "train_loss_llm": 0.0, "grad_norm": 0.06247595697641373, "global_step": 13752, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03635505214333534, "train_loss_bc": 0.03635505214333534, "train_loss_llm": 0.0, "grad_norm": 0.01993604376912117, "global_step": 13753, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03201648220419884, "train_loss_bc": 0.032010044902563095, "train_loss_llm": 0.0006435513496398926, "grad_norm": 0.0374470129609108, "global_step": 13754, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03924848139286041, "train_loss_bc": 0.03924848139286041, "train_loss_llm": 0.0, "grad_norm": 0.048718590289354324, "global_step": 13755, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03868792951107025, "train_loss_bc": 0.03868180140852928, "train_loss_llm": 0.000612795352935791, "grad_norm": 0.05009423568844795, "global_step": 13756, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03791249915957451, "train_loss_bc": 0.03791249915957451, "train_loss_llm": 0.0, "grad_norm": 0.026173358783125877, "global_step": 13757, "epoch": 382, "lr": 0.001} +{"train_loss": 0.029210751876235008, "train_loss_bc": 0.029210569337010384, "train_loss_llm": 1.8209218978881836e-05, "grad_norm": 0.03939947858452797, "global_step": 13758, "epoch": 382, "lr": 0.001} +{"train_loss": 0.039907101541757584, "train_loss_bc": 0.039907101541757584, "train_loss_llm": 0.0, "grad_norm": 0.05181806907057762, "global_step": 13759, "epoch": 382, "lr": 0.001} +{"train_loss": 0.04558293893933296, "train_loss_bc": 0.04558293893933296, "train_loss_llm": 0.0, "grad_norm": 0.0718517154455185, "global_step": 13760, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03475990518927574, "train_loss_bc": 0.03475990518927574, "train_loss_llm": 0.0, "grad_norm": 0.02412060648202896, "global_step": 13761, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03801925852894783, "train_loss_bc": 0.03801925852894783, "train_loss_llm": 0.0, "grad_norm": 0.04402034729719162, "global_step": 13762, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03373444080352783, "train_loss_bc": 0.03373444080352783, "train_loss_llm": 0.0, "grad_norm": 0.05873347073793411, "global_step": 13763, "epoch": 382, "lr": 0.001} +{"train_loss": 0.0365016870200634, "train_loss_bc": 0.0365016870200634, "train_loss_llm": 0.0, "grad_norm": 0.07053747773170471, "global_step": 13764, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03640275076031685, "train_loss_bc": 0.03640275076031685, "train_loss_llm": 0.0, "grad_norm": 0.03154426068067551, "global_step": 13765, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03830903396010399, "train_loss_bc": 0.038302790373563766, "train_loss_llm": 0.0006242692470550537, "grad_norm": 0.051503799855709076, "global_step": 13766, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03171191364526749, "train_loss_bc": 0.031709469854831696, "train_loss_llm": 0.0002442300319671631, "grad_norm": 0.07068753242492676, "global_step": 13767, "epoch": 382, "lr": 0.001} +{"train_loss": 0.0303034745156765, "train_loss_bc": 0.0303034745156765, "train_loss_llm": 0.0, "grad_norm": 0.08283088356256485, "global_step": 13768, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03772389516234398, "train_loss_bc": 0.03772389516234398, "train_loss_llm": 0.0, "grad_norm": 0.029308229684829712, "global_step": 13769, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03998875990509987, "train_loss_bc": 0.03998875990509987, "train_loss_llm": 0.0, "grad_norm": 0.049505241215229034, "global_step": 13770, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03211279958486557, "train_loss_bc": 0.03211279958486557, "train_loss_llm": 0.0, "grad_norm": 0.06550897657871246, "global_step": 13771, "epoch": 382, "lr": 0.001} +{"train_loss": 0.038843825459480286, "train_loss_bc": 0.038843825459480286, "train_loss_llm": 0.0, "grad_norm": 0.0846777930855751, "global_step": 13772, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03001827374100685, "train_loss_bc": 0.03001827374100685, "train_loss_llm": 0.0, "grad_norm": 0.02204268053174019, "global_step": 13773, "epoch": 382, "lr": 0.001} +{"train_loss": 0.032655805349349976, "train_loss_bc": 0.032655805349349976, "train_loss_llm": 0.0, "grad_norm": 0.033084575086832047, "global_step": 13774, "epoch": 382, "lr": 0.001} +{"train_loss": 0.040654126554727554, "train_loss_bc": 0.0406423881649971, "train_loss_llm": 0.0011736750602722168, "grad_norm": 0.045440804213285446, "global_step": 13775, "epoch": 382, "lr": 0.001} +{"train_loss": 0.034685567021369934, "train_loss_bc": 0.034685567021369934, "train_loss_llm": 0.0, "grad_norm": 0.05570561811327934, "global_step": 13776, "epoch": 382, "lr": 0.001} +{"train_loss": 0.041197072714567184, "train_loss_bc": 0.041157614439725876, "train_loss_llm": 0.00394594669342041, "grad_norm": 0.02630787156522274, "global_step": 13777, "epoch": 382, "lr": 0.001} +{"train_loss": 0.041980233043432236, "train_loss_bc": 0.04197732359170914, "train_loss_llm": 0.00029090046882629395, "grad_norm": 0.04120704159140587, "global_step": 13778, "epoch": 382, "lr": 0.001} +{"train_loss": 0.031720638275146484, "train_loss_bc": 0.031720638275146484, "train_loss_llm": 0.0, "grad_norm": 0.05570314824581146, "global_step": 13779, "epoch": 382, "lr": 0.001} +{"train_loss": 0.0448467843234539, "train_loss_bc": 0.0448467843234539, "train_loss_llm": 0.0, "grad_norm": 0.07698637992143631, "global_step": 13780, "epoch": 382, "lr": 0.001} +{"train_loss": 0.0317581407725811, "train_loss_bc": 0.0317581407725811, "train_loss_llm": 0.0, "grad_norm": 0.022268304601311684, "global_step": 13781, "epoch": 382, "lr": 0.001} +{"train_loss": 0.03136351332068443, "train_loss_bc": 0.03136351332068443, "train_loss_llm": 0.0, "grad_norm": 0.041319627314805984, "global_step": 13782, "epoch": 382, "lr": 0.001} +{"train_loss": 0.04126550257205963, "train_loss_bc": 0.041255977004766464, "train_loss_llm": 0.0009524822235107422, "grad_norm": 0.06794589757919312, "global_step": 13783, "epoch": 382, "lr": 0.001} +{"train_loss": 0.036976952105760574, "train_loss_bc": 0.03696279600262642, "train_loss_llm": 0.0014154314994812012, "grad_norm": 0.08234915137290955, "global_step": 13784, "epoch": 382, "lr": 0.001} +{"train_loss": 0.039801936596632004, "train_loss_bc": 0.039789680391550064, "train_loss_llm": 0.0012255609035491943, "grad_norm": 0.026446858420968056, "global_step": 13785, "epoch": 382, "lr": 0.001} +{"train_loss": 0.0375237874686718, "train_loss_bc": 0.0375237874686718, "train_loss_llm": 0.0, "grad_norm": 0.046696387231349945, "global_step": 13786, "epoch": 382, "lr": 0.001} +{"train_loss": 0.036520983558148146, "train_loss_bc": 0.03903111442923546, "train_loss_llm": 0.0, "grad_norm": 0.06575310230255127, "global_step": 13787, "epoch": 382, "lr": 0.001, "val_loss": 0.06706291437149048} +{"train_loss": 0.033931393176317215, "train_loss_bc": 0.03391857445240021, "train_loss_llm": 0.0012818872928619385, "grad_norm": 0.09726323187351227, "global_step": 13788, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03537629917263985, "train_loss_bc": 0.03537629917263985, "train_loss_llm": 0.0, "grad_norm": 0.02137346938252449, "global_step": 13789, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03523056209087372, "train_loss_bc": 0.03523056209087372, "train_loss_llm": 0.0, "grad_norm": 0.03684711828827858, "global_step": 13790, "epoch": 383, "lr": 0.001} +{"train_loss": 0.035622406750917435, "train_loss_bc": 0.035622406750917435, "train_loss_llm": 0.0, "grad_norm": 0.05346476286649704, "global_step": 13791, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03637305274605751, "train_loss_bc": 0.03637305274605751, "train_loss_llm": 0.0, "grad_norm": 0.06188201904296875, "global_step": 13792, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03970644995570183, "train_loss_bc": 0.039672624319791794, "train_loss_llm": 0.003382742404937744, "grad_norm": 0.02034938335418701, "global_step": 13793, "epoch": 383, "lr": 0.001} +{"train_loss": 0.04016731679439545, "train_loss_bc": 0.04016731679439545, "train_loss_llm": 0.0, "grad_norm": 0.03698104992508888, "global_step": 13794, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03518145531415939, "train_loss_bc": 0.035167478024959564, "train_loss_llm": 0.001397639513015747, "grad_norm": 0.04934309422969818, "global_step": 13795, "epoch": 383, "lr": 0.001} +{"train_loss": 0.028562411665916443, "train_loss_bc": 0.028561804443597794, "train_loss_llm": 6.0677528381347656e-05, "grad_norm": 0.061503130942583084, "global_step": 13796, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03226318582892418, "train_loss_bc": 0.03226318582892418, "train_loss_llm": 0.0, "grad_norm": 0.01938105747103691, "global_step": 13797, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03895871713757515, "train_loss_bc": 0.03895871713757515, "train_loss_llm": 0.0, "grad_norm": 0.03398342430591583, "global_step": 13798, "epoch": 383, "lr": 0.001} +{"train_loss": 0.0350944958627224, "train_loss_bc": 0.0350944958627224, "train_loss_llm": 0.0, "grad_norm": 0.04834771156311035, "global_step": 13799, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03757495433092117, "train_loss_bc": 0.03757495433092117, "train_loss_llm": 0.0, "grad_norm": 0.06413983553647995, "global_step": 13800, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03969305753707886, "train_loss_bc": 0.0396815724670887, "train_loss_llm": 0.0011484324932098389, "grad_norm": 0.02790992334485054, "global_step": 13801, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03329026326537132, "train_loss_bc": 0.03329026326537132, "train_loss_llm": 0.0, "grad_norm": 0.047849662601947784, "global_step": 13802, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03560648858547211, "train_loss_bc": 0.03560648858547211, "train_loss_llm": 0.0, "grad_norm": 0.06529625505208969, "global_step": 13803, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03418662026524544, "train_loss_bc": 0.03418662026524544, "train_loss_llm": 0.0, "grad_norm": 0.07912429422140121, "global_step": 13804, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03687606751918793, "train_loss_bc": 0.03687606751918793, "train_loss_llm": 0.0, "grad_norm": 0.030811071395874023, "global_step": 13805, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03557007759809494, "train_loss_bc": 0.03557007759809494, "train_loss_llm": 0.0, "grad_norm": 0.05269870534539223, "global_step": 13806, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03406986966729164, "train_loss_bc": 0.034042563289403915, "train_loss_llm": 0.0027307868003845215, "grad_norm": 0.071827232837677, "global_step": 13807, "epoch": 383, "lr": 0.001} +{"train_loss": 0.035139407962560654, "train_loss_bc": 0.035139407962560654, "train_loss_llm": 0.0, "grad_norm": 0.08983572572469711, "global_step": 13808, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03690771013498306, "train_loss_bc": 0.03690771013498306, "train_loss_llm": 0.0, "grad_norm": 0.03675004839897156, "global_step": 13809, "epoch": 383, "lr": 0.001} +{"train_loss": 0.035879965871572495, "train_loss_bc": 0.035879965871572495, "train_loss_llm": 0.0, "grad_norm": 0.06123574078083038, "global_step": 13810, "epoch": 383, "lr": 0.001} +{"train_loss": 0.04073872044682503, "train_loss_bc": 0.04073872044682503, "train_loss_llm": 0.0, "grad_norm": 0.07621109485626221, "global_step": 13811, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03957292437553406, "train_loss_bc": 0.03957292437553406, "train_loss_llm": 0.0, "grad_norm": 0.0872437134385109, "global_step": 13812, "epoch": 383, "lr": 0.001} +{"train_loss": 0.04157290980219841, "train_loss_bc": 0.04157290980219841, "train_loss_llm": 0.0, "grad_norm": 0.026042653247714043, "global_step": 13813, "epoch": 383, "lr": 0.001} +{"train_loss": 0.044287554919719696, "train_loss_bc": 0.04427691176533699, "train_loss_llm": 0.001064300537109375, "grad_norm": 0.04153776913881302, "global_step": 13814, "epoch": 383, "lr": 0.001} +{"train_loss": 0.028110243380069733, "train_loss_bc": 0.028110243380069733, "train_loss_llm": 0.0, "grad_norm": 0.05534381419420242, "global_step": 13815, "epoch": 383, "lr": 0.001} +{"train_loss": 0.04095560684800148, "train_loss_bc": 0.04095560684800148, "train_loss_llm": 0.0, "grad_norm": 0.06485134363174438, "global_step": 13816, "epoch": 383, "lr": 0.001} +{"train_loss": 0.04046877473592758, "train_loss_bc": 0.04046877473592758, "train_loss_llm": 0.0, "grad_norm": 0.023768413811922073, "global_step": 13817, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03593087196350098, "train_loss_bc": 0.03590654581785202, "train_loss_llm": 0.002432495355606079, "grad_norm": 0.034135572612285614, "global_step": 13818, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03787755221128464, "train_loss_bc": 0.03787755221128464, "train_loss_llm": 0.0, "grad_norm": 0.04896462336182594, "global_step": 13819, "epoch": 383, "lr": 0.001} +{"train_loss": 0.041001513600349426, "train_loss_bc": 0.041001513600349426, "train_loss_llm": 0.0, "grad_norm": 0.0686364695429802, "global_step": 13820, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03629554435610771, "train_loss_bc": 0.03629554435610771, "train_loss_llm": 0.0, "grad_norm": 0.021588053554296494, "global_step": 13821, "epoch": 383, "lr": 0.001} +{"train_loss": 0.035166651010513306, "train_loss_bc": 0.0351325124502182, "train_loss_llm": 0.003413766622543335, "grad_norm": 0.041698891669511795, "global_step": 13822, "epoch": 383, "lr": 0.001} +{"train_loss": 0.03651512031339937, "train_loss_bc": 0.031296972185373306, "train_loss_llm": 0.0006263554096221924, "grad_norm": 0.06800328195095062, "global_step": 13823, "epoch": 383, "lr": 0.001, "val_loss": 0.0670425295829773} +{"train_loss": 0.045193180441856384, "train_loss_bc": 0.0451701320707798, "train_loss_llm": 0.002304971218109131, "grad_norm": 0.08938014507293701, "global_step": 13824, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03285467252135277, "train_loss_bc": 0.03285467252135277, "train_loss_llm": 0.0, "grad_norm": 0.03987417370080948, "global_step": 13825, "epoch": 384, "lr": 0.001} +{"train_loss": 0.0347127690911293, "train_loss_bc": 0.0347127690911293, "train_loss_llm": 0.0, "grad_norm": 0.059536684304475784, "global_step": 13826, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03768913447856903, "train_loss_bc": 0.03767528012394905, "train_loss_llm": 0.0013852715492248535, "grad_norm": 0.07725318521261215, "global_step": 13827, "epoch": 384, "lr": 0.001} +{"train_loss": 0.038242630660533905, "train_loss_bc": 0.038242630660533905, "train_loss_llm": 0.0, "grad_norm": 0.08656536042690277, "global_step": 13828, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03254908323287964, "train_loss_bc": 0.03254908323287964, "train_loss_llm": 0.0, "grad_norm": 0.031909942626953125, "global_step": 13829, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03698575869202614, "train_loss_bc": 0.03698517009615898, "train_loss_llm": 5.900859832763672e-05, "grad_norm": 0.04978810250759125, "global_step": 13830, "epoch": 384, "lr": 0.001} +{"train_loss": 0.034724511206150055, "train_loss_bc": 0.034724511206150055, "train_loss_llm": 0.0, "grad_norm": 0.06251023709774017, "global_step": 13831, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03328549489378929, "train_loss_bc": 0.03328549489378929, "train_loss_llm": 0.0, "grad_norm": 0.07829748094081879, "global_step": 13832, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03446135297417641, "train_loss_bc": 0.03446135297417641, "train_loss_llm": 0.0, "grad_norm": 0.029232122004032135, "global_step": 13833, "epoch": 384, "lr": 0.001} +{"train_loss": 0.037801943719387054, "train_loss_bc": 0.037795525044202805, "train_loss_llm": 0.0006418526172637939, "grad_norm": 0.045333899557590485, "global_step": 13834, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03717030957341194, "train_loss_bc": 0.03717030957341194, "train_loss_llm": 0.0, "grad_norm": 0.061861712485551834, "global_step": 13835, "epoch": 384, "lr": 0.001} +{"train_loss": 0.043058130890131, "train_loss_bc": 0.043058130890131, "train_loss_llm": 0.0, "grad_norm": 0.07624223083257675, "global_step": 13836, "epoch": 384, "lr": 0.001} +{"train_loss": 0.035652920603752136, "train_loss_bc": 0.03562184423208237, "train_loss_llm": 0.003107607364654541, "grad_norm": 0.02766035683453083, "global_step": 13837, "epoch": 384, "lr": 0.001} +{"train_loss": 0.034895677119493484, "train_loss_bc": 0.034895677119493484, "train_loss_llm": 0.0, "grad_norm": 0.04299857094883919, "global_step": 13838, "epoch": 384, "lr": 0.001} +{"train_loss": 0.033053647726774216, "train_loss_bc": 0.033053647726774216, "train_loss_llm": 0.0, "grad_norm": 0.05012807622551918, "global_step": 13839, "epoch": 384, "lr": 0.001} +{"train_loss": 0.028247112408280373, "train_loss_bc": 0.028247112408280373, "train_loss_llm": 0.0, "grad_norm": 0.06803884357213974, "global_step": 13840, "epoch": 384, "lr": 0.001} +{"train_loss": 0.037658099085092545, "train_loss_bc": 0.037658099085092545, "train_loss_llm": 0.0, "grad_norm": 0.019987598061561584, "global_step": 13841, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03416362777352333, "train_loss_bc": 0.03416362777352333, "train_loss_llm": 0.0, "grad_norm": 0.03610233590006828, "global_step": 13842, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03755046799778938, "train_loss_bc": 0.03755046799778938, "train_loss_llm": 0.0, "grad_norm": 0.05024154484272003, "global_step": 13843, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03715389221906662, "train_loss_bc": 0.03715389221906662, "train_loss_llm": 0.0, "grad_norm": 0.05897282063961029, "global_step": 13844, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03491107374429703, "train_loss_bc": 0.03491107374429703, "train_loss_llm": 0.0, "grad_norm": 0.022687861695885658, "global_step": 13845, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03544994816184044, "train_loss_bc": 0.03544994816184044, "train_loss_llm": 0.0, "grad_norm": 0.037729792296886444, "global_step": 13846, "epoch": 384, "lr": 0.001} +{"train_loss": 0.035038918256759644, "train_loss_bc": 0.035037606954574585, "train_loss_llm": 0.00013107061386108398, "grad_norm": 0.0583667978644371, "global_step": 13847, "epoch": 384, "lr": 0.001} +{"train_loss": 0.035183560103178024, "train_loss_bc": 0.03517621010541916, "train_loss_llm": 0.000734865665435791, "grad_norm": 0.07932891696691513, "global_step": 13848, "epoch": 384, "lr": 0.001} +{"train_loss": 0.04026317968964577, "train_loss_bc": 0.04026317968964577, "train_loss_llm": 0.0, "grad_norm": 0.0256461463868618, "global_step": 13849, "epoch": 384, "lr": 0.001} +{"train_loss": 0.04389878362417221, "train_loss_bc": 0.04389878362417221, "train_loss_llm": 0.0, "grad_norm": 0.04129903018474579, "global_step": 13850, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03540476784110069, "train_loss_bc": 0.03540476784110069, "train_loss_llm": 0.0, "grad_norm": 0.058872442692518234, "global_step": 13851, "epoch": 384, "lr": 0.001} +{"train_loss": 0.030864082276821136, "train_loss_bc": 0.030864082276821136, "train_loss_llm": 0.0, "grad_norm": 0.07703517377376556, "global_step": 13852, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03877449408173561, "train_loss_bc": 0.038774363696575165, "train_loss_llm": 1.2993812561035156e-05, "grad_norm": 0.033067718148231506, "global_step": 13853, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03661275655031204, "train_loss_bc": 0.03661203756928444, "train_loss_llm": 7.173418998718262e-05, "grad_norm": 0.06268860399723053, "global_step": 13854, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03600160777568817, "train_loss_bc": 0.03600160777568817, "train_loss_llm": 0.0, "grad_norm": 0.08506845682859421, "global_step": 13855, "epoch": 384, "lr": 0.001} +{"train_loss": 0.0360746905207634, "train_loss_bc": 0.0360746905207634, "train_loss_llm": 0.0, "grad_norm": 0.09195054322481155, "global_step": 13856, "epoch": 384, "lr": 0.001} +{"train_loss": 0.037317413836717606, "train_loss_bc": 0.037317413836717606, "train_loss_llm": 0.0, "grad_norm": 0.018728021532297134, "global_step": 13857, "epoch": 384, "lr": 0.001} +{"train_loss": 0.042976222932338715, "train_loss_bc": 0.042976222932338715, "train_loss_llm": 0.0, "grad_norm": 0.02920682355761528, "global_step": 13858, "epoch": 384, "lr": 0.001} +{"train_loss": 0.03660142354460226, "train_loss_bc": 0.04175810143351555, "train_loss_llm": 0.0017229914665222168, "grad_norm": 0.04242700710892677, "global_step": 13859, "epoch": 384, "lr": 0.001, "val_loss": 0.0674169510602951} +{"train_loss": 0.037148043513298035, "train_loss_bc": 0.037148043513298035, "train_loss_llm": 0.0, "grad_norm": 0.05306875705718994, "global_step": 13860, "epoch": 385, "lr": 0.001} +{"train_loss": 0.036498185247182846, "train_loss_bc": 0.036498185247182846, "train_loss_llm": 0.0, "grad_norm": 0.02084825374186039, "global_step": 13861, "epoch": 385, "lr": 0.001} +{"train_loss": 0.037470582872629166, "train_loss_bc": 0.037470582872629166, "train_loss_llm": 0.0, "grad_norm": 0.03595222160220146, "global_step": 13862, "epoch": 385, "lr": 0.001} +{"train_loss": 0.04175540804862976, "train_loss_bc": 0.04175540804862976, "train_loss_llm": 0.0, "grad_norm": 0.04679242521524429, "global_step": 13863, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03609625995159149, "train_loss_bc": 0.03609625995159149, "train_loss_llm": 0.0, "grad_norm": 0.05894159898161888, "global_step": 13864, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03751484304666519, "train_loss_bc": 0.03751484304666519, "train_loss_llm": 0.0, "grad_norm": 0.030574411153793335, "global_step": 13865, "epoch": 385, "lr": 0.001} +{"train_loss": 0.0316992923617363, "train_loss_bc": 0.031695086508989334, "train_loss_llm": 0.00042045116424560547, "grad_norm": 0.04765546694397926, "global_step": 13866, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03748016431927681, "train_loss_bc": 0.03748016431927681, "train_loss_llm": 0.0, "grad_norm": 0.05754326283931732, "global_step": 13867, "epoch": 385, "lr": 0.001} +{"train_loss": 0.02953309379518032, "train_loss_bc": 0.02953309379518032, "train_loss_llm": 0.0, "grad_norm": 0.06695865839719772, "global_step": 13868, "epoch": 385, "lr": 0.001} +{"train_loss": 0.04021724686026573, "train_loss_bc": 0.04021724686026573, "train_loss_llm": 0.0, "grad_norm": 0.028596870601177216, "global_step": 13869, "epoch": 385, "lr": 0.001} +{"train_loss": 0.0336715430021286, "train_loss_bc": 0.0336715430021286, "train_loss_llm": 0.0, "grad_norm": 0.05277353152632713, "global_step": 13870, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03602723032236099, "train_loss_bc": 0.03602723032236099, "train_loss_llm": 0.0, "grad_norm": 0.07419247180223465, "global_step": 13871, "epoch": 385, "lr": 0.001} +{"train_loss": 0.04023774713277817, "train_loss_bc": 0.04023774713277817, "train_loss_llm": 0.0, "grad_norm": 0.0986129641532898, "global_step": 13872, "epoch": 385, "lr": 0.001} +{"train_loss": 0.04493064805865288, "train_loss_bc": 0.04493064805865288, "train_loss_llm": 0.0, "grad_norm": 0.027521144598722458, "global_step": 13873, "epoch": 385, "lr": 0.001} +{"train_loss": 0.036949705332517624, "train_loss_bc": 0.036893829703330994, "train_loss_llm": 0.005587488412857056, "grad_norm": 0.04514465853571892, "global_step": 13874, "epoch": 385, "lr": 0.001} +{"train_loss": 0.034617144614458084, "train_loss_bc": 0.034611474722623825, "train_loss_llm": 0.0005671679973602295, "grad_norm": 0.05408521741628647, "global_step": 13875, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03946000337600708, "train_loss_bc": 0.03944994881749153, "train_loss_llm": 0.0010055899620056152, "grad_norm": 0.06715324521064758, "global_step": 13876, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03699271008372307, "train_loss_bc": 0.03696165978908539, "train_loss_llm": 0.0031050145626068115, "grad_norm": 0.02218194119632244, "global_step": 13877, "epoch": 385, "lr": 0.001} +{"train_loss": 0.030849402770400047, "train_loss_bc": 0.030847733840346336, "train_loss_llm": 0.00016692280769348145, "grad_norm": 0.02746826969087124, "global_step": 13878, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03794018551707268, "train_loss_bc": 0.03794018551707268, "train_loss_llm": 0.0, "grad_norm": 0.041091155260801315, "global_step": 13879, "epoch": 385, "lr": 0.001} +{"train_loss": 0.034768782556056976, "train_loss_bc": 0.034768782556056976, "train_loss_llm": 0.0, "grad_norm": 0.04652208834886551, "global_step": 13880, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03203996270895004, "train_loss_bc": 0.03203996270895004, "train_loss_llm": 0.0, "grad_norm": 0.024306461215019226, "global_step": 13881, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03372008353471756, "train_loss_bc": 0.03370919078588486, "train_loss_llm": 0.0010891258716583252, "grad_norm": 0.03530080243945122, "global_step": 13882, "epoch": 385, "lr": 0.001} +{"train_loss": 0.030323224142193794, "train_loss_bc": 0.030323224142193794, "train_loss_llm": 0.0, "grad_norm": 0.044694721698760986, "global_step": 13883, "epoch": 385, "lr": 0.001} +{"train_loss": 0.0371028296649456, "train_loss_bc": 0.0371028296649456, "train_loss_llm": 0.0, "grad_norm": 0.050253015011548996, "global_step": 13884, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03859788924455643, "train_loss_bc": 0.03859788924455643, "train_loss_llm": 0.0, "grad_norm": 0.025899091735482216, "global_step": 13885, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03291129320859909, "train_loss_bc": 0.03291129320859909, "train_loss_llm": 0.0, "grad_norm": 0.036418307572603226, "global_step": 13886, "epoch": 385, "lr": 0.001} +{"train_loss": 0.04701356217265129, "train_loss_bc": 0.04697434604167938, "train_loss_llm": 0.003921717405319214, "grad_norm": 0.050661999732255936, "global_step": 13887, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03264157101511955, "train_loss_bc": 0.032632142305374146, "train_loss_llm": 0.0009427368640899658, "grad_norm": 0.07241217792034149, "global_step": 13888, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03683279827237129, "train_loss_bc": 0.0368315689265728, "train_loss_llm": 0.00012284517288208008, "grad_norm": 0.021741660311818123, "global_step": 13889, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03558764606714249, "train_loss_bc": 0.03558764606714249, "train_loss_llm": 0.0, "grad_norm": 0.03577132523059845, "global_step": 13890, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03358836844563484, "train_loss_bc": 0.03358836844563484, "train_loss_llm": 0.0, "grad_norm": 0.04726514592766762, "global_step": 13891, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03653279319405556, "train_loss_bc": 0.03653279319405556, "train_loss_llm": 0.0, "grad_norm": 0.051539573818445206, "global_step": 13892, "epoch": 385, "lr": 0.001} +{"train_loss": 0.0369550883769989, "train_loss_bc": 0.036942970007658005, "train_loss_llm": 0.0012117326259613037, "grad_norm": 0.026496034115552902, "global_step": 13893, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03003929741680622, "train_loss_bc": 0.03003929741680622, "train_loss_llm": 0.0, "grad_norm": 0.0353194922208786, "global_step": 13894, "epoch": 385, "lr": 0.001} +{"train_loss": 0.03622397221624851, "train_loss_bc": 0.038318369537591934, "train_loss_llm": 0.0, "grad_norm": 0.05809992179274559, "global_step": 13895, "epoch": 385, "lr": 0.001, "train/cumulative_reward": 124.20032995235275, "train/mean_score": 5.720854999592328, "train/success_rate": 0.0, "test/cumulative_reward": 31.894790878193703, "test/mean_score": 6.808361845959379, "test/success_rate": 0.16, "val_loss": 0.06708501279354095, "train_action_mse_error": 0.039113350212574005} +{"train_loss": 0.031044134870171547, "train_loss_bc": 0.031044134870171547, "train_loss_llm": 0.0, "grad_norm": 0.07512589544057846, "global_step": 13896, "epoch": 386, "lr": 0.001} +{"train_loss": 0.031010545790195465, "train_loss_bc": 0.031010545790195465, "train_loss_llm": 0.0, "grad_norm": 0.023982709273695946, "global_step": 13897, "epoch": 386, "lr": 0.001} +{"train_loss": 0.02980036474764347, "train_loss_bc": 0.02980036474764347, "train_loss_llm": 0.0, "grad_norm": 0.028768600896000862, "global_step": 13898, "epoch": 386, "lr": 0.001} +{"train_loss": 0.040651045739650726, "train_loss_bc": 0.040651045739650726, "train_loss_llm": 0.0, "grad_norm": 0.03342380002140999, "global_step": 13899, "epoch": 386, "lr": 0.001} +{"train_loss": 0.034247986972332, "train_loss_bc": 0.034247986972332, "train_loss_llm": 0.0, "grad_norm": 0.04087695851922035, "global_step": 13900, "epoch": 386, "lr": 0.001} +{"train_loss": 0.037594448775053024, "train_loss_bc": 0.037594448775053024, "train_loss_llm": 0.0, "grad_norm": 0.023342380300164223, "global_step": 13901, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03614398464560509, "train_loss_bc": 0.03613849729299545, "train_loss_llm": 0.0005488693714141846, "grad_norm": 0.03534413129091263, "global_step": 13902, "epoch": 386, "lr": 0.001} +{"train_loss": 0.027229515835642815, "train_loss_bc": 0.027219658717513084, "train_loss_llm": 0.000985652208328247, "grad_norm": 0.041713129729032516, "global_step": 13903, "epoch": 386, "lr": 0.001} +{"train_loss": 0.04135558381676674, "train_loss_bc": 0.041352778673172, "train_loss_llm": 0.0002804100513458252, "grad_norm": 0.040264081209897995, "global_step": 13904, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03606080636382103, "train_loss_bc": 0.0360579751431942, "train_loss_llm": 0.0002830326557159424, "grad_norm": 0.018040934577584267, "global_step": 13905, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03393499553203583, "train_loss_bc": 0.03392903879284859, "train_loss_llm": 0.0005955994129180908, "grad_norm": 0.029170840978622437, "global_step": 13906, "epoch": 386, "lr": 0.001} +{"train_loss": 0.033445075154304504, "train_loss_bc": 0.033444710075855255, "train_loss_llm": 3.656744956970215e-05, "grad_norm": 0.0414218008518219, "global_step": 13907, "epoch": 386, "lr": 0.001} +{"train_loss": 0.04264504462480545, "train_loss_bc": 0.04264504462480545, "train_loss_llm": 0.0, "grad_norm": 0.0592041015625, "global_step": 13908, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03448205068707466, "train_loss_bc": 0.03448205068707466, "train_loss_llm": 0.0, "grad_norm": 0.023040572181344032, "global_step": 13909, "epoch": 386, "lr": 0.001} +{"train_loss": 0.032869603484869, "train_loss_bc": 0.032869603484869, "train_loss_llm": 0.0, "grad_norm": 0.031352680176496506, "global_step": 13910, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03271634504199028, "train_loss_bc": 0.03271634504199028, "train_loss_llm": 0.0, "grad_norm": 0.04419088363647461, "global_step": 13911, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03776092827320099, "train_loss_bc": 0.0377432145178318, "train_loss_llm": 0.001771479845046997, "grad_norm": 0.05675074830651283, "global_step": 13912, "epoch": 386, "lr": 0.001} +{"train_loss": 0.042178407311439514, "train_loss_bc": 0.042178407311439514, "train_loss_llm": 0.0, "grad_norm": 0.027265232056379318, "global_step": 13913, "epoch": 386, "lr": 0.001} +{"train_loss": 0.040055807679891586, "train_loss_bc": 0.04004788026213646, "train_loss_llm": 0.0007928013801574707, "grad_norm": 0.03938591480255127, "global_step": 13914, "epoch": 386, "lr": 0.001} +{"train_loss": 0.031841594725847244, "train_loss_bc": 0.031841594725847244, "train_loss_llm": 0.0, "grad_norm": 0.04717381298542023, "global_step": 13915, "epoch": 386, "lr": 0.001} +{"train_loss": 0.041041530668735504, "train_loss_bc": 0.041041530668735504, "train_loss_llm": 0.0, "grad_norm": 0.052400507032871246, "global_step": 13916, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03977138176560402, "train_loss_bc": 0.039752375334501266, "train_loss_llm": 0.0019008219242095947, "grad_norm": 0.023099344223737717, "global_step": 13917, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03490850701928139, "train_loss_bc": 0.034906309098005295, "train_loss_llm": 0.00021979212760925293, "grad_norm": 0.03132083639502525, "global_step": 13918, "epoch": 386, "lr": 0.001} +{"train_loss": 0.034976981580257416, "train_loss_bc": 0.034968066960573196, "train_loss_llm": 0.0008913278579711914, "grad_norm": 0.03897416219115257, "global_step": 13919, "epoch": 386, "lr": 0.001} +{"train_loss": 0.038679856806993484, "train_loss_bc": 0.038679856806993484, "train_loss_llm": 0.0, "grad_norm": 0.04382779821753502, "global_step": 13920, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03743855655193329, "train_loss_bc": 0.03743855655193329, "train_loss_llm": 0.0, "grad_norm": 0.019928252324461937, "global_step": 13921, "epoch": 386, "lr": 0.001} +{"train_loss": 0.031236443668603897, "train_loss_bc": 0.031236443668603897, "train_loss_llm": 0.0, "grad_norm": 0.0262403953820467, "global_step": 13922, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03624962642788887, "train_loss_bc": 0.03624962642788887, "train_loss_llm": 0.0, "grad_norm": 0.034768104553222656, "global_step": 13923, "epoch": 386, "lr": 0.001} +{"train_loss": 0.037395596504211426, "train_loss_bc": 0.037382882088422775, "train_loss_llm": 0.0012713372707366943, "grad_norm": 0.047626037150621414, "global_step": 13924, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03202549368143082, "train_loss_bc": 0.03202549368143082, "train_loss_llm": 0.0, "grad_norm": 0.020678387954831123, "global_step": 13925, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03296176344156265, "train_loss_bc": 0.03296176344156265, "train_loss_llm": 0.0, "grad_norm": 0.027249636128544807, "global_step": 13926, "epoch": 386, "lr": 0.001} +{"train_loss": 0.04296017438173294, "train_loss_bc": 0.04296017438173294, "train_loss_llm": 0.0, "grad_norm": 0.03957853838801384, "global_step": 13927, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03866306319832802, "train_loss_bc": 0.03866306319832802, "train_loss_llm": 0.0, "grad_norm": 0.043936461210250854, "global_step": 13928, "epoch": 386, "lr": 0.001} +{"train_loss": 0.03884284570813179, "train_loss_bc": 0.03884284570813179, "train_loss_llm": 0.0, "grad_norm": 0.022659510374069214, "global_step": 13929, "epoch": 386, "lr": 0.001} +{"train_loss": 0.032513801008462906, "train_loss_bc": 0.032513801008462906, "train_loss_llm": 0.0, "grad_norm": 0.03335843235254288, "global_step": 13930, "epoch": 386, "lr": 0.001} +{"train_loss": 0.035990010688288346, "train_loss_bc": 0.03888682276010513, "train_loss_llm": 0.001967102289199829, "grad_norm": 0.0502542145550251, "global_step": 13931, "epoch": 386, "lr": 0.001, "val_loss": 0.06664080917835236} +{"train_loss": 0.028125230222940445, "train_loss_bc": 0.028125230222940445, "train_loss_llm": 0.0, "grad_norm": 0.058273036032915115, "global_step": 13932, "epoch": 387, "lr": 0.001} +{"train_loss": 0.035033177584409714, "train_loss_bc": 0.03502713143825531, "train_loss_llm": 0.0006047189235687256, "grad_norm": 0.024455714970827103, "global_step": 13933, "epoch": 387, "lr": 0.001} +{"train_loss": 0.02842697873711586, "train_loss_bc": 0.02842697873711586, "train_loss_llm": 0.0, "grad_norm": 0.029576720669865608, "global_step": 13934, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03399045392870903, "train_loss_bc": 0.03399045392870903, "train_loss_llm": 0.0, "grad_norm": 0.04017769172787666, "global_step": 13935, "epoch": 387, "lr": 0.001} +{"train_loss": 0.04000469669699669, "train_loss_bc": 0.04000469669699669, "train_loss_llm": 0.0, "grad_norm": 0.04854864627122879, "global_step": 13936, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03302086889743805, "train_loss_bc": 0.03302086889743805, "train_loss_llm": 0.0, "grad_norm": 0.019006792455911636, "global_step": 13937, "epoch": 387, "lr": 0.001} +{"train_loss": 0.04297202453017235, "train_loss_bc": 0.04296885430812836, "train_loss_llm": 0.0003171265125274658, "grad_norm": 0.030919069424271584, "global_step": 13938, "epoch": 387, "lr": 0.001} +{"train_loss": 0.033164091408252716, "train_loss_bc": 0.033164091408252716, "train_loss_llm": 0.0, "grad_norm": 0.04110098257660866, "global_step": 13939, "epoch": 387, "lr": 0.001} +{"train_loss": 0.038211431354284286, "train_loss_bc": 0.038211431354284286, "train_loss_llm": 0.0, "grad_norm": 0.05314501374959946, "global_step": 13940, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03345879539847374, "train_loss_bc": 0.03344978019595146, "train_loss_llm": 0.000901341438293457, "grad_norm": 0.0202966146171093, "global_step": 13941, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03614267706871033, "train_loss_bc": 0.03614267706871033, "train_loss_llm": 0.0, "grad_norm": 0.030513154342770576, "global_step": 13942, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03323744982481003, "train_loss_bc": 0.03323744982481003, "train_loss_llm": 0.0, "grad_norm": 0.045318763703107834, "global_step": 13943, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03717748820781708, "train_loss_bc": 0.03716442734003067, "train_loss_llm": 0.001305997371673584, "grad_norm": 0.054349884390830994, "global_step": 13944, "epoch": 387, "lr": 0.001} +{"train_loss": 0.038081686943769455, "train_loss_bc": 0.038081686943769455, "train_loss_llm": 0.0, "grad_norm": 0.025833647698163986, "global_step": 13945, "epoch": 387, "lr": 0.001} +{"train_loss": 0.0360918864607811, "train_loss_bc": 0.03608384355902672, "train_loss_llm": 0.0008043646812438965, "grad_norm": 0.03933141753077507, "global_step": 13946, "epoch": 387, "lr": 0.001} +{"train_loss": 0.029377343133091927, "train_loss_bc": 0.029372351244091988, "train_loss_llm": 0.0004992187023162842, "grad_norm": 0.05571847781538963, "global_step": 13947, "epoch": 387, "lr": 0.001} +{"train_loss": 0.04294930770993233, "train_loss_bc": 0.04294930770993233, "train_loss_llm": 0.0, "grad_norm": 0.05891256406903267, "global_step": 13948, "epoch": 387, "lr": 0.001} +{"train_loss": 0.037655308842659, "train_loss_bc": 0.037655308842659, "train_loss_llm": 0.0, "grad_norm": 0.024886993691325188, "global_step": 13949, "epoch": 387, "lr": 0.001} +{"train_loss": 0.02929365448653698, "train_loss_bc": 0.029292985796928406, "train_loss_llm": 6.687641143798828e-05, "grad_norm": 0.03263626620173454, "global_step": 13950, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03164001554250717, "train_loss_bc": 0.03164001554250717, "train_loss_llm": 0.0, "grad_norm": 0.03541722893714905, "global_step": 13951, "epoch": 387, "lr": 0.001} +{"train_loss": 0.045752812176942825, "train_loss_bc": 0.045752812176942825, "train_loss_llm": 0.0, "grad_norm": 0.04489967226982117, "global_step": 13952, "epoch": 387, "lr": 0.001} +{"train_loss": 0.025601614266633987, "train_loss_bc": 0.025598766282200813, "train_loss_llm": 0.0002848803997039795, "grad_norm": 0.020364530384540558, "global_step": 13953, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03490371257066727, "train_loss_bc": 0.03490371257066727, "train_loss_llm": 0.0, "grad_norm": 0.0309164896607399, "global_step": 13954, "epoch": 387, "lr": 0.001} +{"train_loss": 0.0356459803879261, "train_loss_bc": 0.0356459803879261, "train_loss_llm": 0.0, "grad_norm": 0.04468180239200592, "global_step": 13955, "epoch": 387, "lr": 0.001} +{"train_loss": 0.039241570979356766, "train_loss_bc": 0.039241570979356766, "train_loss_llm": 0.0, "grad_norm": 0.056614916771650314, "global_step": 13956, "epoch": 387, "lr": 0.001} +{"train_loss": 0.038534119725227356, "train_loss_bc": 0.038534119725227356, "train_loss_llm": 0.0, "grad_norm": 0.025613082572817802, "global_step": 13957, "epoch": 387, "lr": 0.001} +{"train_loss": 0.0360129289329052, "train_loss_bc": 0.03600470349192619, "train_loss_llm": 0.0008227229118347168, "grad_norm": 0.03322402387857437, "global_step": 13958, "epoch": 387, "lr": 0.001} +{"train_loss": 0.037931229919195175, "train_loss_bc": 0.03793108090758324, "train_loss_llm": 1.5020370483398438e-05, "grad_norm": 0.0451505109667778, "global_step": 13959, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03412274643778801, "train_loss_bc": 0.034098684787750244, "train_loss_llm": 0.0024063289165496826, "grad_norm": 0.06120944023132324, "global_step": 13960, "epoch": 387, "lr": 0.001} +{"train_loss": 0.037380095571279526, "train_loss_bc": 0.037380095571279526, "train_loss_llm": 0.0, "grad_norm": 0.02446773461997509, "global_step": 13961, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03548957780003548, "train_loss_bc": 0.03548813611268997, "train_loss_llm": 0.00014421343803405762, "grad_norm": 0.03585764020681381, "global_step": 13962, "epoch": 387, "lr": 0.001} +{"train_loss": 0.04141533002257347, "train_loss_bc": 0.04140816628932953, "train_loss_llm": 0.0007162094116210938, "grad_norm": 0.047764893621206284, "global_step": 13963, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03963102772831917, "train_loss_bc": 0.039609093219041824, "train_loss_llm": 0.002193450927734375, "grad_norm": 0.05931994691491127, "global_step": 13964, "epoch": 387, "lr": 0.001} +{"train_loss": 0.043223146349191666, "train_loss_bc": 0.0432160310447216, "train_loss_llm": 0.0007114708423614502, "grad_norm": 0.02804315648972988, "global_step": 13965, "epoch": 387, "lr": 0.001} +{"train_loss": 0.0371113084256649, "train_loss_bc": 0.0371113084256649, "train_loss_llm": 0.0, "grad_norm": 0.0426279678940773, "global_step": 13966, "epoch": 387, "lr": 0.001} +{"train_loss": 0.03599758270300097, "train_loss_bc": 0.03585828095674515, "train_loss_llm": 0.00029286742210388184, "grad_norm": 0.055117957293987274, "global_step": 13967, "epoch": 387, "lr": 0.001, "val_loss": 0.06753240525722504} +{"train_loss": 0.033451493829488754, "train_loss_bc": 0.033451493829488754, "train_loss_llm": 0.0, "grad_norm": 0.06429486721754074, "global_step": 13968, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03026519902050495, "train_loss_bc": 0.03026519902050495, "train_loss_llm": 0.0, "grad_norm": 0.024229872971773148, "global_step": 13969, "epoch": 388, "lr": 0.001} +{"train_loss": 0.0343613363802433, "train_loss_bc": 0.03434639051556587, "train_loss_llm": 0.0014946162700653076, "grad_norm": 0.029283126816153526, "global_step": 13970, "epoch": 388, "lr": 0.001} +{"train_loss": 0.030319759622216225, "train_loss_bc": 0.03027646616101265, "train_loss_llm": 0.004329383373260498, "grad_norm": 0.037732865661382675, "global_step": 13971, "epoch": 388, "lr": 0.001} +{"train_loss": 0.039423275738954544, "train_loss_bc": 0.039423275738954544, "train_loss_llm": 0.0, "grad_norm": 0.04646411910653114, "global_step": 13972, "epoch": 388, "lr": 0.001} +{"train_loss": 0.035670649260282516, "train_loss_bc": 0.035670649260282516, "train_loss_llm": 0.0, "grad_norm": 0.023485654965043068, "global_step": 13973, "epoch": 388, "lr": 0.001} +{"train_loss": 0.0340895913541317, "train_loss_bc": 0.0340895913541317, "train_loss_llm": 0.0, "grad_norm": 0.03981054574251175, "global_step": 13974, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03206927329301834, "train_loss_bc": 0.03206927329301834, "train_loss_llm": 0.0, "grad_norm": 0.04947619140148163, "global_step": 13975, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03905902057886124, "train_loss_bc": 0.03905902057886124, "train_loss_llm": 0.0, "grad_norm": 0.06588707864284515, "global_step": 13976, "epoch": 388, "lr": 0.001} +{"train_loss": 0.04029207304120064, "train_loss_bc": 0.04029207304120064, "train_loss_llm": 0.0, "grad_norm": 0.028505420312285423, "global_step": 13977, "epoch": 388, "lr": 0.001} +{"train_loss": 0.032908860594034195, "train_loss_bc": 0.03290756791830063, "train_loss_llm": 0.00012934207916259766, "grad_norm": 0.036601364612579346, "global_step": 13978, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03539041057229042, "train_loss_bc": 0.03539041057229042, "train_loss_llm": 0.0, "grad_norm": 0.04502890631556511, "global_step": 13979, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03927676007151604, "train_loss_bc": 0.03927676007151604, "train_loss_llm": 0.0, "grad_norm": 0.06064898148179054, "global_step": 13980, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03935505077242851, "train_loss_bc": 0.03935505077242851, "train_loss_llm": 0.0, "grad_norm": 0.026159677654504776, "global_step": 13981, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03995714336633682, "train_loss_bc": 0.03995714336633682, "train_loss_llm": 0.0, "grad_norm": 0.03764672204852104, "global_step": 13982, "epoch": 388, "lr": 0.001} +{"train_loss": 0.04240288585424423, "train_loss_bc": 0.04240288585424423, "train_loss_llm": 0.0, "grad_norm": 0.05089753493666649, "global_step": 13983, "epoch": 388, "lr": 0.001} +{"train_loss": 0.0306377075612545, "train_loss_bc": 0.0306377075612545, "train_loss_llm": 0.0, "grad_norm": 0.05562048405408859, "global_step": 13984, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03597971796989441, "train_loss_bc": 0.03597971796989441, "train_loss_llm": 0.0, "grad_norm": 0.020194754004478455, "global_step": 13985, "epoch": 388, "lr": 0.001} +{"train_loss": 0.040186554193496704, "train_loss_bc": 0.040186554193496704, "train_loss_llm": 0.0, "grad_norm": 0.031684793531894684, "global_step": 13986, "epoch": 388, "lr": 0.001} +{"train_loss": 0.0327487587928772, "train_loss_bc": 0.032747067511081696, "train_loss_llm": 0.0001690983772277832, "grad_norm": 0.03788398951292038, "global_step": 13987, "epoch": 388, "lr": 0.001} +{"train_loss": 0.031670648604631424, "train_loss_bc": 0.031669169664382935, "train_loss_llm": 0.00014784932136535645, "grad_norm": 0.04384160414338112, "global_step": 13988, "epoch": 388, "lr": 0.001} +{"train_loss": 0.04373452439904213, "train_loss_bc": 0.04373452439904213, "train_loss_llm": 0.0, "grad_norm": 0.029002048075199127, "global_step": 13989, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03895390033721924, "train_loss_bc": 0.038944561034440994, "train_loss_llm": 0.0009340345859527588, "grad_norm": 0.04176340252161026, "global_step": 13990, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03615792095661163, "train_loss_bc": 0.03615792095661163, "train_loss_llm": 0.0, "grad_norm": 0.04826615750789642, "global_step": 13991, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03668993338942528, "train_loss_bc": 0.03668993338942528, "train_loss_llm": 0.0, "grad_norm": 0.055413585156202316, "global_step": 13992, "epoch": 388, "lr": 0.001} +{"train_loss": 0.039800215512514114, "train_loss_bc": 0.039800215512514114, "train_loss_llm": 0.0, "grad_norm": 0.023981807753443718, "global_step": 13993, "epoch": 388, "lr": 0.001} +{"train_loss": 0.029039990156888962, "train_loss_bc": 0.029039990156888962, "train_loss_llm": 0.0, "grad_norm": 0.034181565046310425, "global_step": 13994, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03028712049126625, "train_loss_bc": 0.03026474267244339, "train_loss_llm": 0.0022377967834472656, "grad_norm": 0.04507805407047272, "global_step": 13995, "epoch": 388, "lr": 0.001} +{"train_loss": 0.03831808641552925, "train_loss_bc": 0.03831133246421814, "train_loss_llm": 0.0006752908229827881, "grad_norm": 0.05848630890250206, "global_step": 13996, "epoch": 388, "lr": 0.001} +{"train_loss": 0.036153316497802734, "train_loss_bc": 0.03612310439348221, "train_loss_llm": 0.0030211210250854492, "grad_norm": 0.023793471977114677, "global_step": 13997, "epoch": 388, "lr": 0.001} +{"train_loss": 0.032373812049627304, "train_loss_bc": 0.032373812049627304, "train_loss_llm": 0.0, "grad_norm": 0.035914964973926544, "global_step": 13998, "epoch": 388, "lr": 0.001} +{"train_loss": 0.029971225187182426, "train_loss_bc": 0.029971225187182426, "train_loss_llm": 0.0, "grad_norm": 0.04485606774687767, "global_step": 13999, "epoch": 388, "lr": 0.001} +{"train_loss": 0.040554121136665344, "train_loss_bc": 0.040554121136665344, "train_loss_llm": 0.0, "grad_norm": 0.06156584993004799, "global_step": 14000, "epoch": 388, "lr": 0.001} +{"train_loss": 0.0359133817255497, "train_loss_bc": 0.0359133817255497, "train_loss_llm": 0.0, "grad_norm": 0.019399942830204964, "global_step": 14001, "epoch": 388, "lr": 0.001} +{"train_loss": 0.04445290192961693, "train_loss_bc": 0.04444947466254234, "train_loss_llm": 0.0003427863121032715, "grad_norm": 0.03823614865541458, "global_step": 14002, "epoch": 388, "lr": 0.001} +{"train_loss": 0.036031986535009414, "train_loss_bc": 0.03523489460349083, "train_loss_llm": 0.0, "grad_norm": 0.05930827185511589, "global_step": 14003, "epoch": 388, "lr": 0.001, "val_loss": 0.06753188371658325} +{"train_loss": 0.029128775000572205, "train_loss_bc": 0.029128775000572205, "train_loss_llm": 0.0, "grad_norm": 0.07629860192537308, "global_step": 14004, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03699592873454094, "train_loss_bc": 0.03699592873454094, "train_loss_llm": 0.0, "grad_norm": 0.023820213973522186, "global_step": 14005, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03646175190806389, "train_loss_bc": 0.03646175190806389, "train_loss_llm": 0.0, "grad_norm": 0.04145669937133789, "global_step": 14006, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03670043870806694, "train_loss_bc": 0.03670043870806694, "train_loss_llm": 0.0, "grad_norm": 0.052462585270404816, "global_step": 14007, "epoch": 389, "lr": 0.001} +{"train_loss": 0.034903742372989655, "train_loss_bc": 0.034903742372989655, "train_loss_llm": 0.0, "grad_norm": 0.05946392938494682, "global_step": 14008, "epoch": 389, "lr": 0.001} +{"train_loss": 0.033516474068164825, "train_loss_bc": 0.033516474068164825, "train_loss_llm": 0.0, "grad_norm": 0.02637224644422531, "global_step": 14009, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03841095045208931, "train_loss_bc": 0.03841095045208931, "train_loss_llm": 0.0, "grad_norm": 0.04593124985694885, "global_step": 14010, "epoch": 389, "lr": 0.001} +{"train_loss": 0.04402397200465202, "train_loss_bc": 0.04401712492108345, "train_loss_llm": 0.0006845295429229736, "grad_norm": 0.059081319719552994, "global_step": 14011, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03516141697764397, "train_loss_bc": 0.03516141697764397, "train_loss_llm": 0.0, "grad_norm": 0.061855122447013855, "global_step": 14012, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03498541936278343, "train_loss_bc": 0.034974705427885056, "train_loss_llm": 0.0010712742805480957, "grad_norm": 0.023565534502267838, "global_step": 14013, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03659702464938164, "train_loss_bc": 0.03659702464938164, "train_loss_llm": 0.0, "grad_norm": 0.037548989057540894, "global_step": 14014, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03733369708061218, "train_loss_bc": 0.03733369708061218, "train_loss_llm": 0.0, "grad_norm": 0.0516735278069973, "global_step": 14015, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03887426480650902, "train_loss_bc": 0.03887426480650902, "train_loss_llm": 0.0, "grad_norm": 0.07038243860006332, "global_step": 14016, "epoch": 389, "lr": 0.001} +{"train_loss": 0.035997338593006134, "train_loss_bc": 0.035997338593006134, "train_loss_llm": 0.0, "grad_norm": 0.0282229445874691, "global_step": 14017, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03045804053544998, "train_loss_bc": 0.03045804053544998, "train_loss_llm": 0.0, "grad_norm": 0.04331187903881073, "global_step": 14018, "epoch": 389, "lr": 0.001} +{"train_loss": 0.0342201367020607, "train_loss_bc": 0.0342201367020607, "train_loss_llm": 0.0, "grad_norm": 0.05398625135421753, "global_step": 14019, "epoch": 389, "lr": 0.001} +{"train_loss": 0.036274783313274384, "train_loss_bc": 0.036274783313274384, "train_loss_llm": 0.0, "grad_norm": 0.07458552718162537, "global_step": 14020, "epoch": 389, "lr": 0.001} +{"train_loss": 0.039426758885383606, "train_loss_bc": 0.039419740438461304, "train_loss_llm": 0.0007017850875854492, "grad_norm": 0.02612612210214138, "global_step": 14021, "epoch": 389, "lr": 0.001} +{"train_loss": 0.04008219391107559, "train_loss_bc": 0.04008219391107559, "train_loss_llm": 0.0, "grad_norm": 0.0395100861787796, "global_step": 14022, "epoch": 389, "lr": 0.001} +{"train_loss": 0.045814547687768936, "train_loss_bc": 0.045814547687768936, "train_loss_llm": 0.0, "grad_norm": 0.05089794471859932, "global_step": 14023, "epoch": 389, "lr": 0.001} +{"train_loss": 0.026195427402853966, "train_loss_bc": 0.026195427402853966, "train_loss_llm": 0.0, "grad_norm": 0.061680231243371964, "global_step": 14024, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03321297466754913, "train_loss_bc": 0.03321297466754913, "train_loss_llm": 0.0, "grad_norm": 0.021463332697749138, "global_step": 14025, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03659351170063019, "train_loss_bc": 0.03659351170063019, "train_loss_llm": 0.0, "grad_norm": 0.035349082201719284, "global_step": 14026, "epoch": 389, "lr": 0.001} +{"train_loss": 0.040224187076091766, "train_loss_bc": 0.040224187076091766, "train_loss_llm": 0.0, "grad_norm": 0.045969098806381226, "global_step": 14027, "epoch": 389, "lr": 0.001} +{"train_loss": 0.039201900362968445, "train_loss_bc": 0.039194926619529724, "train_loss_llm": 0.0006975531578063965, "grad_norm": 0.04981454461812973, "global_step": 14028, "epoch": 389, "lr": 0.001} +{"train_loss": 0.04035786911845207, "train_loss_bc": 0.04035786911845207, "train_loss_llm": 0.0, "grad_norm": 0.022597117349505424, "global_step": 14029, "epoch": 389, "lr": 0.001} +{"train_loss": 0.02978990226984024, "train_loss_bc": 0.02978990226984024, "train_loss_llm": 0.0, "grad_norm": 0.04198354110121727, "global_step": 14030, "epoch": 389, "lr": 0.001} +{"train_loss": 0.03251660615205765, "train_loss_bc": 0.03251660615205765, "train_loss_llm": 0.0, "grad_norm": 0.055897992104291916, "global_step": 14031, "epoch": 389, "lr": 0.001} +{"train_loss": 0.034202951937913895, "train_loss_bc": 0.034202951937913895, "train_loss_llm": 0.0, "grad_norm": 0.07344352453947067, "global_step": 14032, "epoch": 389, "lr": 0.001} +{"train_loss": 0.0374094620347023, "train_loss_bc": 0.03737406060099602, "train_loss_llm": 0.0035402774810791016, "grad_norm": 0.027650538831949234, "global_step": 14033, "epoch": 389, "lr": 0.001} +{"train_loss": 0.032304760068655014, "train_loss_bc": 0.03229478746652603, "train_loss_llm": 0.000997096300125122, "grad_norm": 0.03918878734111786, "global_step": 14034, "epoch": 389, "lr": 0.001} +{"train_loss": 0.036922287195920944, "train_loss_bc": 0.03691282868385315, "train_loss_llm": 0.0009458363056182861, "grad_norm": 0.0535912923514843, "global_step": 14035, "epoch": 389, "lr": 0.001} +{"train_loss": 0.04331820085644722, "train_loss_bc": 0.04331820085644722, "train_loss_llm": 0.0, "grad_norm": 0.06998935341835022, "global_step": 14036, "epoch": 389, "lr": 0.001} +{"train_loss": 0.02935640700161457, "train_loss_bc": 0.029338175430893898, "train_loss_llm": 0.0018231570720672607, "grad_norm": 0.02130301482975483, "global_step": 14037, "epoch": 389, "lr": 0.001} +{"train_loss": 0.037739239633083344, "train_loss_bc": 0.03771523758769035, "train_loss_llm": 0.0024002790451049805, "grad_norm": 0.046738989651203156, "global_step": 14038, "epoch": 389, "lr": 0.001} +{"train_loss": 0.036131315450701446, "train_loss_bc": 0.0359976589679718, "train_loss_llm": 0.001635521650314331, "grad_norm": 0.05986829847097397, "global_step": 14039, "epoch": 389, "lr": 0.001, "val_loss": 0.06735086441040039} +{"train_loss": 0.03247019276022911, "train_loss_bc": 0.032455381006002426, "train_loss_llm": 0.0014812648296356201, "grad_norm": 0.0709744319319725, "global_step": 14040, "epoch": 390, "lr": 0.001} +{"train_loss": 0.034587692469358444, "train_loss_bc": 0.034587692469358444, "train_loss_llm": 0.0, "grad_norm": 0.024446453899145126, "global_step": 14041, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03886718675494194, "train_loss_bc": 0.03886718675494194, "train_loss_llm": 0.0, "grad_norm": 0.03905057534575462, "global_step": 14042, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03535781800746918, "train_loss_bc": 0.03535781800746918, "train_loss_llm": 0.0, "grad_norm": 0.045949049293994904, "global_step": 14043, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03626568987965584, "train_loss_bc": 0.03626295551657677, "train_loss_llm": 0.000273287296295166, "grad_norm": 0.0622539222240448, "global_step": 14044, "epoch": 390, "lr": 0.001} +{"train_loss": 0.04031644016504288, "train_loss_bc": 0.04031644016504288, "train_loss_llm": 0.0, "grad_norm": 0.0218362919986248, "global_step": 14045, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03374520689249039, "train_loss_bc": 0.03374520689249039, "train_loss_llm": 0.0, "grad_norm": 0.031076651066541672, "global_step": 14046, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03858007863163948, "train_loss_bc": 0.03854065015912056, "train_loss_llm": 0.003942698240280151, "grad_norm": 0.044776126742362976, "global_step": 14047, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03841987997293472, "train_loss_bc": 0.03841987997293472, "train_loss_llm": 0.0, "grad_norm": 0.05725827440619469, "global_step": 14048, "epoch": 390, "lr": 0.001} +{"train_loss": 0.030987676233053207, "train_loss_bc": 0.03097778931260109, "train_loss_llm": 0.000988781452178955, "grad_norm": 0.0205716323107481, "global_step": 14049, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03445030376315117, "train_loss_bc": 0.03445030376315117, "train_loss_llm": 0.0, "grad_norm": 0.03574356436729431, "global_step": 14050, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03483787551522255, "train_loss_bc": 0.03483787551522255, "train_loss_llm": 0.0, "grad_norm": 0.04611964896321297, "global_step": 14051, "epoch": 390, "lr": 0.001} +{"train_loss": 0.023361044004559517, "train_loss_bc": 0.023361044004559517, "train_loss_llm": 0.0, "grad_norm": 0.045212119817733765, "global_step": 14052, "epoch": 390, "lr": 0.001} +{"train_loss": 0.04030441492795944, "train_loss_bc": 0.04030441492795944, "train_loss_llm": 0.0, "grad_norm": 0.025520917028188705, "global_step": 14053, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03424910083413124, "train_loss_bc": 0.03424910083413124, "train_loss_llm": 0.0, "grad_norm": 0.04972521588206291, "global_step": 14054, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03945031389594078, "train_loss_bc": 0.03945031389594078, "train_loss_llm": 0.0, "grad_norm": 0.051142435520887375, "global_step": 14055, "epoch": 390, "lr": 0.001} +{"train_loss": 0.041099026799201965, "train_loss_bc": 0.0410938523709774, "train_loss_llm": 0.0005175173282623291, "grad_norm": 0.07112954556941986, "global_step": 14056, "epoch": 390, "lr": 0.001} +{"train_loss": 0.033160023391246796, "train_loss_bc": 0.033160023391246796, "train_loss_llm": 0.0, "grad_norm": 0.020039444789290428, "global_step": 14057, "epoch": 390, "lr": 0.001} +{"train_loss": 0.04098976030945778, "train_loss_bc": 0.04098976030945778, "train_loss_llm": 0.0, "grad_norm": 0.033568888902664185, "global_step": 14058, "epoch": 390, "lr": 0.001} +{"train_loss": 0.040959376841783524, "train_loss_bc": 0.040959376841783524, "train_loss_llm": 0.0, "grad_norm": 0.046872541308403015, "global_step": 14059, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03977248817682266, "train_loss_bc": 0.03977248817682266, "train_loss_llm": 0.0, "grad_norm": 0.059219762682914734, "global_step": 14060, "epoch": 390, "lr": 0.001} +{"train_loss": 0.040156830102205276, "train_loss_bc": 0.040156830102205276, "train_loss_llm": 0.0, "grad_norm": 0.024656863883137703, "global_step": 14061, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03461631014943123, "train_loss_bc": 0.03461631014943123, "train_loss_llm": 0.0, "grad_norm": 0.03355308622121811, "global_step": 14062, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03174290060997009, "train_loss_bc": 0.03174290060997009, "train_loss_llm": 0.0, "grad_norm": 0.04066077247262001, "global_step": 14063, "epoch": 390, "lr": 0.001} +{"train_loss": 0.040144436061382294, "train_loss_bc": 0.040144436061382294, "train_loss_llm": 0.0, "grad_norm": 0.05239439755678177, "global_step": 14064, "epoch": 390, "lr": 0.001} +{"train_loss": 0.02901034615933895, "train_loss_bc": 0.02901034615933895, "train_loss_llm": 0.0, "grad_norm": 0.020997190847992897, "global_step": 14065, "epoch": 390, "lr": 0.001} +{"train_loss": 0.041376881301403046, "train_loss_bc": 0.041376881301403046, "train_loss_llm": 0.0, "grad_norm": 0.034801509231328964, "global_step": 14066, "epoch": 390, "lr": 0.001} +{"train_loss": 0.0365775004029274, "train_loss_bc": 0.03657504916191101, "train_loss_llm": 0.00024503469467163086, "grad_norm": 0.04475361853837967, "global_step": 14067, "epoch": 390, "lr": 0.001} +{"train_loss": 0.028715182095766068, "train_loss_bc": 0.028715182095766068, "train_loss_llm": 0.0, "grad_norm": 0.053009096533060074, "global_step": 14068, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03721287474036217, "train_loss_bc": 0.03721287474036217, "train_loss_llm": 0.0, "grad_norm": 0.020325442776083946, "global_step": 14069, "epoch": 390, "lr": 0.001} +{"train_loss": 0.04162852466106415, "train_loss_bc": 0.04162852466106415, "train_loss_llm": 0.0, "grad_norm": 0.03403999283909798, "global_step": 14070, "epoch": 390, "lr": 0.001} +{"train_loss": 0.028700271621346474, "train_loss_bc": 0.028697676956653595, "train_loss_llm": 0.0002594590187072754, "grad_norm": 0.04371793568134308, "global_step": 14071, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03534507006406784, "train_loss_bc": 0.03533283993601799, "train_loss_llm": 0.0012230277061462402, "grad_norm": 0.05249599739909172, "global_step": 14072, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03340393677353859, "train_loss_bc": 0.03340163454413414, "train_loss_llm": 0.00023025274276733398, "grad_norm": 0.023491069674491882, "global_step": 14073, "epoch": 390, "lr": 0.001} +{"train_loss": 0.0315060131251812, "train_loss_bc": 0.0315060131251812, "train_loss_llm": 0.0, "grad_norm": 0.04049772396683693, "global_step": 14074, "epoch": 390, "lr": 0.001} +{"train_loss": 0.03610634871034159, "train_loss_bc": 0.047459885478019714, "train_loss_llm": 0.0, "grad_norm": 0.05678470805287361, "global_step": 14075, "epoch": 390, "lr": 0.001, "train/cumulative_reward": 154.81584726927267, "train/mean_score": 15.83943751823133, "train/success_rate": 0.3, "test/cumulative_reward": 52.76287007129664, "test/mean_score": 7.098310146514043, "test/success_rate": 0.16, "val_loss": 0.06693093478679657, "train_action_mse_error": 0.038929130882024765} +{"train_loss": 0.03698434308171272, "train_loss_bc": 0.03698151186108589, "train_loss_llm": 0.000282973051071167, "grad_norm": 0.06351286172866821, "global_step": 14076, "epoch": 391, "lr": 0.001} +{"train_loss": 0.035988010466098785, "train_loss_bc": 0.035988010466098785, "train_loss_llm": 0.0, "grad_norm": 0.02612059749662876, "global_step": 14077, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03629688918590546, "train_loss_bc": 0.03629688918590546, "train_loss_llm": 0.0, "grad_norm": 0.031180450692772865, "global_step": 14078, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03615931421518326, "train_loss_bc": 0.03615931421518326, "train_loss_llm": 0.0, "grad_norm": 0.03798152878880501, "global_step": 14079, "epoch": 391, "lr": 0.001} +{"train_loss": 0.038647398352622986, "train_loss_bc": 0.038647398352622986, "train_loss_llm": 0.0, "grad_norm": 0.05539016053080559, "global_step": 14080, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03043520636856556, "train_loss_bc": 0.03043520636856556, "train_loss_llm": 0.0, "grad_norm": 0.02328137494623661, "global_step": 14081, "epoch": 391, "lr": 0.001} +{"train_loss": 0.04143545776605606, "train_loss_bc": 0.04143545776605606, "train_loss_llm": 0.0, "grad_norm": 0.03819447383284569, "global_step": 14082, "epoch": 391, "lr": 0.001} +{"train_loss": 0.04378305748105049, "train_loss_bc": 0.04378305748105049, "train_loss_llm": 0.0, "grad_norm": 0.045316629111766815, "global_step": 14083, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03682754933834076, "train_loss_bc": 0.036817047744989395, "train_loss_llm": 0.0010502338409423828, "grad_norm": 0.05911215767264366, "global_step": 14084, "epoch": 391, "lr": 0.001} +{"train_loss": 0.036589138209819794, "train_loss_bc": 0.036589138209819794, "train_loss_llm": 0.0, "grad_norm": 0.023855626583099365, "global_step": 14085, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03798900917172432, "train_loss_bc": 0.03798900917172432, "train_loss_llm": 0.0, "grad_norm": 0.04235217720270157, "global_step": 14086, "epoch": 391, "lr": 0.001} +{"train_loss": 0.034205514937639236, "train_loss_bc": 0.034205514937639236, "train_loss_llm": 0.0, "grad_norm": 0.048031773418188095, "global_step": 14087, "epoch": 391, "lr": 0.001} +{"train_loss": 0.036621399223804474, "train_loss_bc": 0.036609359085559845, "train_loss_llm": 0.0012039542198181152, "grad_norm": 0.056983832269907, "global_step": 14088, "epoch": 391, "lr": 0.001} +{"train_loss": 0.0321599505841732, "train_loss_bc": 0.0321599505841732, "train_loss_llm": 0.0, "grad_norm": 0.023952152580022812, "global_step": 14089, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03303701430559158, "train_loss_bc": 0.03303701430559158, "train_loss_llm": 0.0, "grad_norm": 0.04039280489087105, "global_step": 14090, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03737969696521759, "train_loss_bc": 0.03737969696521759, "train_loss_llm": 0.0, "grad_norm": 0.06858640164136887, "global_step": 14091, "epoch": 391, "lr": 0.001} +{"train_loss": 0.038654670119285583, "train_loss_bc": 0.038654670119285583, "train_loss_llm": 0.0, "grad_norm": 0.09074997156858444, "global_step": 14092, "epoch": 391, "lr": 0.001} +{"train_loss": 0.02900674007833004, "train_loss_bc": 0.02900674007833004, "train_loss_llm": 0.0, "grad_norm": 0.022383734583854675, "global_step": 14093, "epoch": 391, "lr": 0.001} +{"train_loss": 0.04111460968852043, "train_loss_bc": 0.04111460968852043, "train_loss_llm": 0.0, "grad_norm": 0.04022178053855896, "global_step": 14094, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03816448897123337, "train_loss_bc": 0.03815867751836777, "train_loss_llm": 0.0005809664726257324, "grad_norm": 0.046360649168491364, "global_step": 14095, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03645838424563408, "train_loss_bc": 0.03645838424563408, "train_loss_llm": 0.0, "grad_norm": 0.046912696212530136, "global_step": 14096, "epoch": 391, "lr": 0.001} +{"train_loss": 0.037348393350839615, "train_loss_bc": 0.037348393350839615, "train_loss_llm": 0.0, "grad_norm": 0.02602311037480831, "global_step": 14097, "epoch": 391, "lr": 0.001} +{"train_loss": 0.035802166908979416, "train_loss_bc": 0.035802166908979416, "train_loss_llm": 0.0, "grad_norm": 0.033519573509693146, "global_step": 14098, "epoch": 391, "lr": 0.001} +{"train_loss": 0.036648962646722794, "train_loss_bc": 0.03663955256342888, "train_loss_llm": 0.0009410381317138672, "grad_norm": 0.04736305773258209, "global_step": 14099, "epoch": 391, "lr": 0.001} +{"train_loss": 0.0354226678609848, "train_loss_bc": 0.0354226678609848, "train_loss_llm": 0.0, "grad_norm": 0.06037749722599983, "global_step": 14100, "epoch": 391, "lr": 0.001} +{"train_loss": 0.035925574600696564, "train_loss_bc": 0.03592078760266304, "train_loss_llm": 0.00047853589057922363, "grad_norm": 0.018394354730844498, "global_step": 14101, "epoch": 391, "lr": 0.001} +{"train_loss": 0.039778150618076324, "train_loss_bc": 0.039778150618076324, "train_loss_llm": 0.0, "grad_norm": 0.032768987119197845, "global_step": 14102, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03684309124946594, "train_loss_bc": 0.03681405633687973, "train_loss_llm": 0.002903491258621216, "grad_norm": 0.04523108899593353, "global_step": 14103, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03175162523984909, "train_loss_bc": 0.031747005879879, "train_loss_llm": 0.00046181678771972656, "grad_norm": 0.053983673453330994, "global_step": 14104, "epoch": 391, "lr": 0.001} +{"train_loss": 0.029442349448800087, "train_loss_bc": 0.029426222667098045, "train_loss_llm": 0.0016127228736877441, "grad_norm": 0.021213574334979057, "global_step": 14105, "epoch": 391, "lr": 0.001} +{"train_loss": 0.036435406655073166, "train_loss_bc": 0.036435406655073166, "train_loss_llm": 0.0, "grad_norm": 0.04074481129646301, "global_step": 14106, "epoch": 391, "lr": 0.001} +{"train_loss": 0.029026668518781662, "train_loss_bc": 0.029025094583630562, "train_loss_llm": 0.00015741586685180664, "grad_norm": 0.05425624921917915, "global_step": 14107, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03849467635154724, "train_loss_bc": 0.03849467635154724, "train_loss_llm": 0.0, "grad_norm": 0.0701846182346344, "global_step": 14108, "epoch": 391, "lr": 0.001} +{"train_loss": 0.0334537997841835, "train_loss_bc": 0.0334537997841835, "train_loss_llm": 0.0, "grad_norm": 0.02551552653312683, "global_step": 14109, "epoch": 391, "lr": 0.001} +{"train_loss": 0.037852946668863297, "train_loss_bc": 0.037852946668863297, "train_loss_llm": 0.0, "grad_norm": 0.030508285388350487, "global_step": 14110, "epoch": 391, "lr": 0.001} +{"train_loss": 0.03595437166384525, "train_loss_bc": 0.032193057239055634, "train_loss_llm": 0.0, "grad_norm": 0.04397689551115036, "global_step": 14111, "epoch": 391, "lr": 0.001, "val_loss": 0.06715408712625504} +{"train_loss": 0.032286327332258224, "train_loss_bc": 0.032276105135679245, "train_loss_llm": 0.0010222792625427246, "grad_norm": 0.05834507197141647, "global_step": 14112, "epoch": 392, "lr": 0.001} +{"train_loss": 0.031532254070043564, "train_loss_bc": 0.031532254070043564, "train_loss_llm": 0.0, "grad_norm": 0.025869008153676987, "global_step": 14113, "epoch": 392, "lr": 0.001} +{"train_loss": 0.0415729396045208, "train_loss_bc": 0.0415729396045208, "train_loss_llm": 0.0, "grad_norm": 0.0470159687101841, "global_step": 14114, "epoch": 392, "lr": 0.001} +{"train_loss": 0.043332185596227646, "train_loss_bc": 0.043332185596227646, "train_loss_llm": 0.0, "grad_norm": 0.07168775796890259, "global_step": 14115, "epoch": 392, "lr": 0.001} +{"train_loss": 0.040373411029577255, "train_loss_bc": 0.040353886783123016, "train_loss_llm": 0.001952439546585083, "grad_norm": 0.08559999614953995, "global_step": 14116, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03564350679516792, "train_loss_bc": 0.03564350679516792, "train_loss_llm": 0.0, "grad_norm": 0.01891062594950199, "global_step": 14117, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03452387824654579, "train_loss_bc": 0.03452387824654579, "train_loss_llm": 0.0, "grad_norm": 0.03361388295888901, "global_step": 14118, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03851303458213806, "train_loss_bc": 0.03851146250963211, "train_loss_llm": 0.00015714764595031738, "grad_norm": 0.04313008859753609, "global_step": 14119, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03295169398188591, "train_loss_bc": 0.032946761697530746, "train_loss_llm": 0.0004931092262268066, "grad_norm": 0.05292380228638649, "global_step": 14120, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03582461178302765, "train_loss_bc": 0.03582461178302765, "train_loss_llm": 0.0, "grad_norm": 0.019257526844739914, "global_step": 14121, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03864506259560585, "train_loss_bc": 0.03864459693431854, "train_loss_llm": 4.6640634536743164e-05, "grad_norm": 0.032312437891960144, "global_step": 14122, "epoch": 392, "lr": 0.001} +{"train_loss": 0.028650393709540367, "train_loss_bc": 0.02863016538321972, "train_loss_llm": 0.0020228922367095947, "grad_norm": 0.03979116678237915, "global_step": 14123, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03577305004000664, "train_loss_bc": 0.0357547253370285, "train_loss_llm": 0.001832515001296997, "grad_norm": 0.04597285017371178, "global_step": 14124, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03358234837651253, "train_loss_bc": 0.03358234837651253, "train_loss_llm": 0.0, "grad_norm": 0.022657491266727448, "global_step": 14125, "epoch": 392, "lr": 0.001} +{"train_loss": 0.0329936146736145, "train_loss_bc": 0.03298689052462578, "train_loss_llm": 0.0006723701953887939, "grad_norm": 0.03576548397541046, "global_step": 14126, "epoch": 392, "lr": 0.001} +{"train_loss": 0.035036537796258926, "train_loss_bc": 0.03503319248557091, "train_loss_llm": 0.00033465027809143066, "grad_norm": 0.052782200276851654, "global_step": 14127, "epoch": 392, "lr": 0.001} +{"train_loss": 0.036228619515895844, "train_loss_bc": 0.036228619515895844, "train_loss_llm": 0.0, "grad_norm": 0.06709539145231247, "global_step": 14128, "epoch": 392, "lr": 0.001} +{"train_loss": 0.038654543459415436, "train_loss_bc": 0.038654543459415436, "train_loss_llm": 0.0, "grad_norm": 0.021334635093808174, "global_step": 14129, "epoch": 392, "lr": 0.001} +{"train_loss": 0.043675534427165985, "train_loss_bc": 0.043675534427165985, "train_loss_llm": 0.0, "grad_norm": 0.03634485974907875, "global_step": 14130, "epoch": 392, "lr": 0.001} +{"train_loss": 0.031034069135785103, "train_loss_bc": 0.031034069135785103, "train_loss_llm": 0.0, "grad_norm": 0.04706530272960663, "global_step": 14131, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03728359192609787, "train_loss_bc": 0.037276316434144974, "train_loss_llm": 0.0007276237010955811, "grad_norm": 0.06630494445562363, "global_step": 14132, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03525116667151451, "train_loss_bc": 0.03525116667151451, "train_loss_llm": 0.0, "grad_norm": 0.03581807389855385, "global_step": 14133, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03415316343307495, "train_loss_bc": 0.03415316343307495, "train_loss_llm": 0.0, "grad_norm": 0.04204512760043144, "global_step": 14134, "epoch": 392, "lr": 0.001} +{"train_loss": 0.02980734407901764, "train_loss_bc": 0.02980649098753929, "train_loss_llm": 8.52346420288086e-05, "grad_norm": 0.05887168273329735, "global_step": 14135, "epoch": 392, "lr": 0.001} +{"train_loss": 0.042469240725040436, "train_loss_bc": 0.042469240725040436, "train_loss_llm": 0.0, "grad_norm": 0.07058539986610413, "global_step": 14136, "epoch": 392, "lr": 0.001} +{"train_loss": 0.035821348428726196, "train_loss_bc": 0.035821348428726196, "train_loss_llm": 0.0, "grad_norm": 0.026714934036135674, "global_step": 14137, "epoch": 392, "lr": 0.001} +{"train_loss": 0.04429623484611511, "train_loss_bc": 0.04429398104548454, "train_loss_llm": 0.00022539496421813965, "grad_norm": 0.047138821333646774, "global_step": 14138, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03631727024912834, "train_loss_bc": 0.03631727024912834, "train_loss_llm": 0.0, "grad_norm": 0.06268689036369324, "global_step": 14139, "epoch": 392, "lr": 0.001} +{"train_loss": 0.034291334450244904, "train_loss_bc": 0.034291334450244904, "train_loss_llm": 0.0, "grad_norm": 0.07598895579576492, "global_step": 14140, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03722623363137245, "train_loss_bc": 0.03722050040960312, "train_loss_llm": 0.0005733072757720947, "grad_norm": 0.020792193710803986, "global_step": 14141, "epoch": 392, "lr": 0.001} +{"train_loss": 0.039411939680576324, "train_loss_bc": 0.03940584510564804, "train_loss_llm": 0.0006094276905059814, "grad_norm": 0.028402438387274742, "global_step": 14142, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03993166983127594, "train_loss_bc": 0.03993166983127594, "train_loss_llm": 0.0, "grad_norm": 0.03965560719370842, "global_step": 14143, "epoch": 392, "lr": 0.001} +{"train_loss": 0.027903037145733833, "train_loss_bc": 0.027903037145733833, "train_loss_llm": 0.0, "grad_norm": 0.045345697551965714, "global_step": 14144, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03730311617255211, "train_loss_bc": 0.03730311617255211, "train_loss_llm": 0.0, "grad_norm": 0.025870900601148605, "global_step": 14145, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03539463132619858, "train_loss_bc": 0.03539463132619858, "train_loss_llm": 0.0, "grad_norm": 0.03534127026796341, "global_step": 14146, "epoch": 392, "lr": 0.001} +{"train_loss": 0.03592916034782926, "train_loss_bc": 0.02576083317399025, "train_loss_llm": 0.0, "grad_norm": 0.042079150676727295, "global_step": 14147, "epoch": 392, "lr": 0.001, "val_loss": 0.06822194159030914} +{"train_loss": 0.03837265819311142, "train_loss_bc": 0.03837265819311142, "train_loss_llm": 0.0, "grad_norm": 0.053823161870241165, "global_step": 14148, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03961919620633125, "train_loss_bc": 0.039615027606487274, "train_loss_llm": 0.0004169046878814697, "grad_norm": 0.029547609388828278, "global_step": 14149, "epoch": 393, "lr": 0.001} +{"train_loss": 0.037238337099552155, "train_loss_bc": 0.037238337099552155, "train_loss_llm": 0.0, "grad_norm": 0.04239242896437645, "global_step": 14150, "epoch": 393, "lr": 0.001} +{"train_loss": 0.035661082714796066, "train_loss_bc": 0.03565320372581482, "train_loss_llm": 0.0007877945899963379, "grad_norm": 0.05989518389105797, "global_step": 14151, "epoch": 393, "lr": 0.001} +{"train_loss": 0.041387442499399185, "train_loss_bc": 0.041387442499399185, "train_loss_llm": 0.0, "grad_norm": 0.08066477626562119, "global_step": 14152, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03276913985610008, "train_loss_bc": 0.032769035547971725, "train_loss_llm": 1.0311603546142578e-05, "grad_norm": 0.02393244579434395, "global_step": 14153, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03618992865085602, "train_loss_bc": 0.03618992865085602, "train_loss_llm": 0.0, "grad_norm": 0.04771007224917412, "global_step": 14154, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03596949577331543, "train_loss_bc": 0.03596256673336029, "train_loss_llm": 0.0006930828094482422, "grad_norm": 0.0616021603345871, "global_step": 14155, "epoch": 393, "lr": 0.001} +{"train_loss": 0.029429234564304352, "train_loss_bc": 0.029425377026200294, "train_loss_llm": 0.0003857612609863281, "grad_norm": 0.07501422613859177, "global_step": 14156, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03544600307941437, "train_loss_bc": 0.03544600307941437, "train_loss_llm": 0.0, "grad_norm": 0.021923605352640152, "global_step": 14157, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03477480635046959, "train_loss_bc": 0.034756820648908615, "train_loss_llm": 0.0017986595630645752, "grad_norm": 0.03615438565611839, "global_step": 14158, "epoch": 393, "lr": 0.001} +{"train_loss": 0.031092923134565353, "train_loss_bc": 0.031092923134565353, "train_loss_llm": 0.0, "grad_norm": 0.04055297002196312, "global_step": 14159, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03617800399661064, "train_loss_bc": 0.03617800399661064, "train_loss_llm": 0.0, "grad_norm": 0.050168026238679886, "global_step": 14160, "epoch": 393, "lr": 0.001} +{"train_loss": 0.043670956045389175, "train_loss_bc": 0.043670956045389175, "train_loss_llm": 0.0, "grad_norm": 0.02791246399283409, "global_step": 14161, "epoch": 393, "lr": 0.001} +{"train_loss": 0.04485808312892914, "train_loss_bc": 0.04485808312892914, "train_loss_llm": 0.0, "grad_norm": 0.03498423472046852, "global_step": 14162, "epoch": 393, "lr": 0.001} +{"train_loss": 0.0328180193901062, "train_loss_bc": 0.03281040117144585, "train_loss_llm": 0.000761866569519043, "grad_norm": 0.044697344303131104, "global_step": 14163, "epoch": 393, "lr": 0.001} +{"train_loss": 0.032946713268756866, "train_loss_bc": 0.03294458985328674, "train_loss_llm": 0.00021225214004516602, "grad_norm": 0.04954955726861954, "global_step": 14164, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03629542514681816, "train_loss_bc": 0.03629542514681816, "train_loss_llm": 0.0, "grad_norm": 0.02528034895658493, "global_step": 14165, "epoch": 393, "lr": 0.001} +{"train_loss": 0.032277852296829224, "train_loss_bc": 0.032277852296829224, "train_loss_llm": 0.0, "grad_norm": 0.031037069857120514, "global_step": 14166, "epoch": 393, "lr": 0.001} +{"train_loss": 0.02950732409954071, "train_loss_bc": 0.0295051671564579, "train_loss_llm": 0.00021567940711975098, "grad_norm": 0.04195830225944519, "global_step": 14167, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03515131399035454, "train_loss_bc": 0.03515131399035454, "train_loss_llm": 0.0, "grad_norm": 0.05235103890299797, "global_step": 14168, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03815232589840889, "train_loss_bc": 0.03815232589840889, "train_loss_llm": 0.0, "grad_norm": 0.024293024092912674, "global_step": 14169, "epoch": 393, "lr": 0.001} +{"train_loss": 0.030019335448741913, "train_loss_bc": 0.0300154946744442, "train_loss_llm": 0.0003841519355773926, "grad_norm": 0.029133986681699753, "global_step": 14170, "epoch": 393, "lr": 0.001} +{"train_loss": 0.039467401802539825, "train_loss_bc": 0.039467401802539825, "train_loss_llm": 0.0, "grad_norm": 0.04298831522464752, "global_step": 14171, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03383190184831619, "train_loss_bc": 0.03383190184831619, "train_loss_llm": 0.0, "grad_norm": 0.054719023406505585, "global_step": 14172, "epoch": 393, "lr": 0.001} +{"train_loss": 0.034563157707452774, "train_loss_bc": 0.034561023116111755, "train_loss_llm": 0.00021329522132873535, "grad_norm": 0.02957463450729847, "global_step": 14173, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03520707041025162, "train_loss_bc": 0.03520707041025162, "train_loss_llm": 0.0, "grad_norm": 0.04587429389357567, "global_step": 14174, "epoch": 393, "lr": 0.001} +{"train_loss": 0.0332510843873024, "train_loss_bc": 0.0332510843873024, "train_loss_llm": 0.0, "grad_norm": 0.060464609414339066, "global_step": 14175, "epoch": 393, "lr": 0.001} +{"train_loss": 0.04250200092792511, "train_loss_bc": 0.04244615510106087, "train_loss_llm": 0.0055844783782958984, "grad_norm": 0.08337825536727905, "global_step": 14176, "epoch": 393, "lr": 0.001} +{"train_loss": 0.029905613511800766, "train_loss_bc": 0.029905613511800766, "train_loss_llm": 0.0, "grad_norm": 0.01875363476574421, "global_step": 14177, "epoch": 393, "lr": 0.001} +{"train_loss": 0.039644330739974976, "train_loss_bc": 0.039644330739974976, "train_loss_llm": 0.0, "grad_norm": 0.03346278518438339, "global_step": 14178, "epoch": 393, "lr": 0.001} +{"train_loss": 0.0337119996547699, "train_loss_bc": 0.0337119996547699, "train_loss_llm": 0.0, "grad_norm": 0.049596965312957764, "global_step": 14179, "epoch": 393, "lr": 0.001} +{"train_loss": 0.04216209053993225, "train_loss_bc": 0.04216209053993225, "train_loss_llm": 0.0, "grad_norm": 0.07323858141899109, "global_step": 14180, "epoch": 393, "lr": 0.001} +{"train_loss": 0.03410771116614342, "train_loss_bc": 0.03410771116614342, "train_loss_llm": 0.0, "grad_norm": 0.02517719939351082, "global_step": 14181, "epoch": 393, "lr": 0.001} +{"train_loss": 0.04598962143063545, "train_loss_bc": 0.04598962143063545, "train_loss_llm": 0.0, "grad_norm": 0.04406500607728958, "global_step": 14182, "epoch": 393, "lr": 0.001} +{"train_loss": 0.036070206513007484, "train_loss_bc": 0.034346435219049454, "train_loss_llm": 0.0011414289474487305, "grad_norm": 0.06042955443263054, "global_step": 14183, "epoch": 393, "lr": 0.001, "val_loss": 0.06681002676486969} +{"train_loss": 0.04203763231635094, "train_loss_bc": 0.04203763231635094, "train_loss_llm": 0.0, "grad_norm": 0.07185357064008713, "global_step": 14184, "epoch": 394, "lr": 0.001} +{"train_loss": 0.0322432704269886, "train_loss_bc": 0.0321730375289917, "train_loss_llm": 0.007023215293884277, "grad_norm": 0.018893318250775337, "global_step": 14185, "epoch": 394, "lr": 0.001} +{"train_loss": 0.032964758574962616, "train_loss_bc": 0.03294576331973076, "train_loss_llm": 0.0018996596336364746, "grad_norm": 0.028477083891630173, "global_step": 14186, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03605172038078308, "train_loss_bc": 0.03604935109615326, "train_loss_llm": 0.00023698806762695312, "grad_norm": 0.03698468208312988, "global_step": 14187, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03846683353185654, "train_loss_bc": 0.03845878690481186, "train_loss_llm": 0.0008046925067901611, "grad_norm": 0.04892110079526901, "global_step": 14188, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03578168898820877, "train_loss_bc": 0.03578168898820877, "train_loss_llm": 0.0, "grad_norm": 0.021864736452698708, "global_step": 14189, "epoch": 394, "lr": 0.001} +{"train_loss": 0.031662166118621826, "train_loss_bc": 0.031653884798288345, "train_loss_llm": 0.0008280575275421143, "grad_norm": 0.030659688636660576, "global_step": 14190, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03350387513637543, "train_loss_bc": 0.03350387513637543, "train_loss_llm": 0.0, "grad_norm": 0.03601256385445595, "global_step": 14191, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03628091141581535, "train_loss_bc": 0.03628091141581535, "train_loss_llm": 0.0, "grad_norm": 0.050035011023283005, "global_step": 14192, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03704630956053734, "train_loss_bc": 0.03704630956053734, "train_loss_llm": 0.0, "grad_norm": 0.025104457512497902, "global_step": 14193, "epoch": 394, "lr": 0.001} +{"train_loss": 0.04359564557671547, "train_loss_bc": 0.04359564557671547, "train_loss_llm": 0.0, "grad_norm": 0.03346839174628258, "global_step": 14194, "epoch": 394, "lr": 0.001} +{"train_loss": 0.039063066244125366, "train_loss_bc": 0.03906115144491196, "train_loss_llm": 0.0001913905143737793, "grad_norm": 0.050398677587509155, "global_step": 14195, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03624116629362106, "train_loss_bc": 0.03624116629362106, "train_loss_llm": 0.0, "grad_norm": 0.06237034127116203, "global_step": 14196, "epoch": 394, "lr": 0.001} +{"train_loss": 0.036203816533088684, "train_loss_bc": 0.036203816533088684, "train_loss_llm": 0.0, "grad_norm": 0.025353651493787766, "global_step": 14197, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03310314938426018, "train_loss_bc": 0.03309522196650505, "train_loss_llm": 0.0007927119731903076, "grad_norm": 0.03775705397129059, "global_step": 14198, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03950083628296852, "train_loss_bc": 0.03950083628296852, "train_loss_llm": 0.0, "grad_norm": 0.05138624086976051, "global_step": 14199, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03503379970788956, "train_loss_bc": 0.03502728044986725, "train_loss_llm": 0.0006517469882965088, "grad_norm": 0.06331414729356766, "global_step": 14200, "epoch": 394, "lr": 0.001} +{"train_loss": 0.034977663308382034, "train_loss_bc": 0.034977663308382034, "train_loss_llm": 0.0, "grad_norm": 0.022068362683057785, "global_step": 14201, "epoch": 394, "lr": 0.001} +{"train_loss": 0.036808889359235764, "train_loss_bc": 0.036808889359235764, "train_loss_llm": 0.0, "grad_norm": 0.03126714751124382, "global_step": 14202, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03154789283871651, "train_loss_bc": 0.03153805807232857, "train_loss_llm": 0.0009836256504058838, "grad_norm": 0.04385831952095032, "global_step": 14203, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03730910271406174, "train_loss_bc": 0.03730910271406174, "train_loss_llm": 0.0, "grad_norm": 0.05616261810064316, "global_step": 14204, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03427900746464729, "train_loss_bc": 0.03427900746464729, "train_loss_llm": 0.0, "grad_norm": 0.02763679251074791, "global_step": 14205, "epoch": 394, "lr": 0.001} +{"train_loss": 0.034258533269166946, "train_loss_bc": 0.034258533269166946, "train_loss_llm": 0.0, "grad_norm": 0.03862858563661575, "global_step": 14206, "epoch": 394, "lr": 0.001} +{"train_loss": 0.029117275029420853, "train_loss_bc": 0.029117275029420853, "train_loss_llm": 0.0, "grad_norm": 0.05285833775997162, "global_step": 14207, "epoch": 394, "lr": 0.001} +{"train_loss": 0.038356680423021317, "train_loss_bc": 0.038356680423021317, "train_loss_llm": 0.0, "grad_norm": 0.07150208204984665, "global_step": 14208, "epoch": 394, "lr": 0.001} +{"train_loss": 0.030479591339826584, "train_loss_bc": 0.030479591339826584, "train_loss_llm": 0.0, "grad_norm": 0.024611881002783775, "global_step": 14209, "epoch": 394, "lr": 0.001} +{"train_loss": 0.04142419621348381, "train_loss_bc": 0.04142419621348381, "train_loss_llm": 0.0, "grad_norm": 0.037240222096443176, "global_step": 14210, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03243991732597351, "train_loss_bc": 0.03243991732597351, "train_loss_llm": 0.0, "grad_norm": 0.046182211488485336, "global_step": 14211, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03708212822675705, "train_loss_bc": 0.03708045929670334, "train_loss_llm": 0.00016677379608154297, "grad_norm": 0.061365168541669846, "global_step": 14212, "epoch": 394, "lr": 0.001} +{"train_loss": 0.038601599633693695, "train_loss_bc": 0.038601599633693695, "train_loss_llm": 0.0, "grad_norm": 0.022240733727812767, "global_step": 14213, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03913081809878349, "train_loss_bc": 0.03913081809878349, "train_loss_llm": 0.0, "grad_norm": 0.03984199836850166, "global_step": 14214, "epoch": 394, "lr": 0.001} +{"train_loss": 0.0382886677980423, "train_loss_bc": 0.0382886677980423, "train_loss_llm": 0.0, "grad_norm": 0.05060805007815361, "global_step": 14215, "epoch": 394, "lr": 0.001} +{"train_loss": 0.029547473415732384, "train_loss_bc": 0.02954450249671936, "train_loss_llm": 0.0002970695495605469, "grad_norm": 0.055205218493938446, "global_step": 14216, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03656169772148132, "train_loss_bc": 0.03656169772148132, "train_loss_llm": 0.0, "grad_norm": 0.025533806532621384, "global_step": 14217, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03485145792365074, "train_loss_bc": 0.03485145792365074, "train_loss_llm": 0.0, "grad_norm": 0.042350396513938904, "global_step": 14218, "epoch": 394, "lr": 0.001} +{"train_loss": 0.03599223830840654, "train_loss_bc": 0.04187734052538872, "train_loss_llm": 0.0, "grad_norm": 0.06744997948408127, "global_step": 14219, "epoch": 394, "lr": 0.001, "val_loss": 0.06813626736402512} +{"train_loss": 0.0342240184545517, "train_loss_bc": 0.0342240184545517, "train_loss_llm": 0.0, "grad_norm": 0.08589403331279755, "global_step": 14220, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03548714518547058, "train_loss_bc": 0.03548714518547058, "train_loss_llm": 0.0, "grad_norm": 0.031259894371032715, "global_step": 14221, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03591703251004219, "train_loss_bc": 0.03591703251004219, "train_loss_llm": 0.0, "grad_norm": 0.04834659770131111, "global_step": 14222, "epoch": 395, "lr": 0.001} +{"train_loss": 0.045087479054927826, "train_loss_bc": 0.04507173225283623, "train_loss_llm": 0.0015746057033538818, "grad_norm": 0.07126548886299133, "global_step": 14223, "epoch": 395, "lr": 0.001} +{"train_loss": 0.036298658698797226, "train_loss_bc": 0.036298658698797226, "train_loss_llm": 0.0, "grad_norm": 0.08171902596950531, "global_step": 14224, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03808223828673363, "train_loss_bc": 0.03808223828673363, "train_loss_llm": 0.0, "grad_norm": 0.028938064351677895, "global_step": 14225, "epoch": 395, "lr": 0.001} +{"train_loss": 0.0410810224711895, "train_loss_bc": 0.0410810224711895, "train_loss_llm": 0.0, "grad_norm": 0.05167405679821968, "global_step": 14226, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03455820307135582, "train_loss_bc": 0.03455820307135582, "train_loss_llm": 0.0, "grad_norm": 0.06060653179883957, "global_step": 14227, "epoch": 395, "lr": 0.001} +{"train_loss": 0.0305996835231781, "train_loss_bc": 0.0305996835231781, "train_loss_llm": 0.0, "grad_norm": 0.07803013920783997, "global_step": 14228, "epoch": 395, "lr": 0.001} +{"train_loss": 0.034275591373443604, "train_loss_bc": 0.034275591373443604, "train_loss_llm": 0.0, "grad_norm": 0.027794960886240005, "global_step": 14229, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03852944076061249, "train_loss_bc": 0.03852944076061249, "train_loss_llm": 0.0, "grad_norm": 0.034129489213228226, "global_step": 14230, "epoch": 395, "lr": 0.001} +{"train_loss": 0.04083704203367233, "train_loss_bc": 0.04083704203367233, "train_loss_llm": 0.0, "grad_norm": 0.04568229243159294, "global_step": 14231, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03888482227921486, "train_loss_bc": 0.03888482227921486, "train_loss_llm": 0.0, "grad_norm": 0.05810457095503807, "global_step": 14232, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03247682750225067, "train_loss_bc": 0.03245343267917633, "train_loss_llm": 0.0023395121097564697, "grad_norm": 0.019289012998342514, "global_step": 14233, "epoch": 395, "lr": 0.001} +{"train_loss": 0.036553774029016495, "train_loss_bc": 0.036553774029016495, "train_loss_llm": 0.0, "grad_norm": 0.027757255360484123, "global_step": 14234, "epoch": 395, "lr": 0.001} +{"train_loss": 0.0310046449303627, "train_loss_bc": 0.030983816832304, "train_loss_llm": 0.0020828843116760254, "grad_norm": 0.03820826858282089, "global_step": 14235, "epoch": 395, "lr": 0.001} +{"train_loss": 0.042757708579301834, "train_loss_bc": 0.04274924471974373, "train_loss_llm": 0.0008462667465209961, "grad_norm": 0.04443129524588585, "global_step": 14236, "epoch": 395, "lr": 0.001} +{"train_loss": 0.037648335099220276, "train_loss_bc": 0.037648335099220276, "train_loss_llm": 0.0, "grad_norm": 0.032319311052560806, "global_step": 14237, "epoch": 395, "lr": 0.001} +{"train_loss": 0.036547835916280746, "train_loss_bc": 0.03652206435799599, "train_loss_llm": 0.002577275037765503, "grad_norm": 0.04790271073579788, "global_step": 14238, "epoch": 395, "lr": 0.001} +{"train_loss": 0.034949079155921936, "train_loss_bc": 0.03490886464715004, "train_loss_llm": 0.004021316766738892, "grad_norm": 0.05583275854587555, "global_step": 14239, "epoch": 395, "lr": 0.001} +{"train_loss": 0.0331333763897419, "train_loss_bc": 0.0331333763897419, "train_loss_llm": 0.0, "grad_norm": 0.06750147044658661, "global_step": 14240, "epoch": 395, "lr": 0.001} +{"train_loss": 0.039211440831422806, "train_loss_bc": 0.039211440831422806, "train_loss_llm": 0.0, "grad_norm": 0.02835918217897415, "global_step": 14241, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03453458100557327, "train_loss_bc": 0.03453458100557327, "train_loss_llm": 0.0, "grad_norm": 0.039221685379743576, "global_step": 14242, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03605570271611214, "train_loss_bc": 0.03604676201939583, "train_loss_llm": 0.0008939802646636963, "grad_norm": 0.05305115133523941, "global_step": 14243, "epoch": 395, "lr": 0.001} +{"train_loss": 0.0366373285651207, "train_loss_bc": 0.0366373285651207, "train_loss_llm": 0.0, "grad_norm": 0.0703955590724945, "global_step": 14244, "epoch": 395, "lr": 0.001} +{"train_loss": 0.02292783558368683, "train_loss_bc": 0.02292783558368683, "train_loss_llm": 0.0, "grad_norm": 0.028070682659745216, "global_step": 14245, "epoch": 395, "lr": 0.001} +{"train_loss": 0.037786874920129776, "train_loss_bc": 0.037779800593853, "train_loss_llm": 0.0007074475288391113, "grad_norm": 0.054018694907426834, "global_step": 14246, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03900997340679169, "train_loss_bc": 0.039004288613796234, "train_loss_llm": 0.0005686581134796143, "grad_norm": 0.07981539517641068, "global_step": 14247, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03095727413892746, "train_loss_bc": 0.030947620049118996, "train_loss_llm": 0.000965416431427002, "grad_norm": 0.09899872541427612, "global_step": 14248, "epoch": 395, "lr": 0.001} +{"train_loss": 0.04143286123871803, "train_loss_bc": 0.04143286123871803, "train_loss_llm": 0.0, "grad_norm": 0.028900349512696266, "global_step": 14249, "epoch": 395, "lr": 0.001} +{"train_loss": 0.04178635776042938, "train_loss_bc": 0.04178529977798462, "train_loss_llm": 0.00010567903518676758, "grad_norm": 0.047335632145404816, "global_step": 14250, "epoch": 395, "lr": 0.001} +{"train_loss": 0.028198661282658577, "train_loss_bc": 0.028198661282658577, "train_loss_llm": 0.0, "grad_norm": 0.06208984926342964, "global_step": 14251, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03501037508249283, "train_loss_bc": 0.03501037508249283, "train_loss_llm": 0.0, "grad_norm": 0.08000092953443527, "global_step": 14252, "epoch": 395, "lr": 0.001} +{"train_loss": 0.04153319075703621, "train_loss_bc": 0.04152659699320793, "train_loss_llm": 0.0006591975688934326, "grad_norm": 0.026899810880422592, "global_step": 14253, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03198491781949997, "train_loss_bc": 0.03198491781949997, "train_loss_llm": 0.0, "grad_norm": 0.040927816182374954, "global_step": 14254, "epoch": 395, "lr": 0.001} +{"train_loss": 0.03619526538790928, "train_loss_bc": 0.037023793905973434, "train_loss_llm": 0.0004423856735229492, "grad_norm": 0.05389705300331116, "global_step": 14255, "epoch": 395, "lr": 0.001, "train/cumulative_reward": 24.905963391342727, "train/mean_score": 7.51311052506167, "train/success_rate": 0.2, "test/cumulative_reward": 33.00360493634108, "test/mean_score": 4.220491273793456, "test/success_rate": 0.06, "val_loss": 0.06753794103860855, "train_action_mse_error": 0.03880120441317558} +{"train_loss": 0.04031554237008095, "train_loss_bc": 0.04031554237008095, "train_loss_llm": 0.0, "grad_norm": 0.06105807423591614, "global_step": 14256, "epoch": 396, "lr": 0.001} +{"train_loss": 0.033986322581768036, "train_loss_bc": 0.033986322581768036, "train_loss_llm": 0.0, "grad_norm": 0.023782789707183838, "global_step": 14257, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03662605211138725, "train_loss_bc": 0.03662605211138725, "train_loss_llm": 0.0, "grad_norm": 0.03263644129037857, "global_step": 14258, "epoch": 396, "lr": 0.001} +{"train_loss": 0.035633549094200134, "train_loss_bc": 0.035633549094200134, "train_loss_llm": 0.0, "grad_norm": 0.04432417452335358, "global_step": 14259, "epoch": 396, "lr": 0.001} +{"train_loss": 0.04113466292619705, "train_loss_bc": 0.04112142324447632, "train_loss_llm": 0.001323789358139038, "grad_norm": 0.05143401771783829, "global_step": 14260, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03941802307963371, "train_loss_bc": 0.03941802307963371, "train_loss_llm": 0.0, "grad_norm": 0.028140954673290253, "global_step": 14261, "epoch": 396, "lr": 0.001} +{"train_loss": 0.031999822705984116, "train_loss_bc": 0.03198809176683426, "train_loss_llm": 0.0011731982231140137, "grad_norm": 0.04608139023184776, "global_step": 14262, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03403909504413605, "train_loss_bc": 0.03403773531317711, "train_loss_llm": 0.0001360774040222168, "grad_norm": 0.06222198158502579, "global_step": 14263, "epoch": 396, "lr": 0.001} +{"train_loss": 0.037040870636701584, "train_loss_bc": 0.037040870636701584, "train_loss_llm": 0.0, "grad_norm": 0.07553334534168243, "global_step": 14264, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03929673880338669, "train_loss_bc": 0.03929673880338669, "train_loss_llm": 0.0, "grad_norm": 0.026846542954444885, "global_step": 14265, "epoch": 396, "lr": 0.001} +{"train_loss": 0.028699282556772232, "train_loss_bc": 0.028699282556772232, "train_loss_llm": 0.0, "grad_norm": 0.041351523250341415, "global_step": 14266, "epoch": 396, "lr": 0.001} +{"train_loss": 0.033155664801597595, "train_loss_bc": 0.033155664801597595, "train_loss_llm": 0.0, "grad_norm": 0.058934006839990616, "global_step": 14267, "epoch": 396, "lr": 0.001} +{"train_loss": 0.042193423956632614, "train_loss_bc": 0.042193423956632614, "train_loss_llm": 0.0, "grad_norm": 0.08811096847057343, "global_step": 14268, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03176765516400337, "train_loss_bc": 0.03176765516400337, "train_loss_llm": 0.0, "grad_norm": 0.027172310277819633, "global_step": 14269, "epoch": 396, "lr": 0.001} +{"train_loss": 0.041425544768571854, "train_loss_bc": 0.041425544768571854, "train_loss_llm": 0.0, "grad_norm": 0.04743819311261177, "global_step": 14270, "epoch": 396, "lr": 0.001} +{"train_loss": 0.044637005776166916, "train_loss_bc": 0.044637005776166916, "train_loss_llm": 0.0, "grad_norm": 0.06384695321321487, "global_step": 14271, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03125419095158577, "train_loss_bc": 0.03125419095158577, "train_loss_llm": 0.0, "grad_norm": 0.07839637249708176, "global_step": 14272, "epoch": 396, "lr": 0.001} +{"train_loss": 0.034136779606342316, "train_loss_bc": 0.03411702439188957, "train_loss_llm": 0.001975506544113159, "grad_norm": 0.02761043980717659, "global_step": 14273, "epoch": 396, "lr": 0.001} +{"train_loss": 0.038663409650325775, "train_loss_bc": 0.038663409650325775, "train_loss_llm": 0.0, "grad_norm": 0.041350625455379486, "global_step": 14274, "epoch": 396, "lr": 0.001} +{"train_loss": 0.037012044340372086, "train_loss_bc": 0.037012044340372086, "train_loss_llm": 0.0, "grad_norm": 0.05686633288860321, "global_step": 14275, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03613752871751785, "train_loss_bc": 0.03613752871751785, "train_loss_llm": 0.0, "grad_norm": 0.07571731507778168, "global_step": 14276, "epoch": 396, "lr": 0.001} +{"train_loss": 0.029687900096178055, "train_loss_bc": 0.02968195080757141, "train_loss_llm": 0.0005950033664703369, "grad_norm": 0.02242462709546089, "global_step": 14277, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03246726095676422, "train_loss_bc": 0.03246164321899414, "train_loss_llm": 0.0005617439746856689, "grad_norm": 0.0352323055267334, "global_step": 14278, "epoch": 396, "lr": 0.001} +{"train_loss": 0.0330171138048172, "train_loss_bc": 0.0330171138048172, "train_loss_llm": 0.0, "grad_norm": 0.043857112526893616, "global_step": 14279, "epoch": 396, "lr": 0.001} +{"train_loss": 0.040367256850004196, "train_loss_bc": 0.040367256850004196, "train_loss_llm": 0.0, "grad_norm": 0.055523525923490524, "global_step": 14280, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03937237337231636, "train_loss_bc": 0.03937237337231636, "train_loss_llm": 0.0, "grad_norm": 0.026550795882940292, "global_step": 14281, "epoch": 396, "lr": 0.001} +{"train_loss": 0.04129432141780853, "train_loss_bc": 0.04129432141780853, "train_loss_llm": 0.0, "grad_norm": 0.0356631800532341, "global_step": 14282, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03664768114686012, "train_loss_bc": 0.03664768114686012, "train_loss_llm": 0.0, "grad_norm": 0.04429067671298981, "global_step": 14283, "epoch": 396, "lr": 0.001} +{"train_loss": 0.040640175342559814, "train_loss_bc": 0.040640175342559814, "train_loss_llm": 0.0, "grad_norm": 0.06118463724851608, "global_step": 14284, "epoch": 396, "lr": 0.001} +{"train_loss": 0.036338333040475845, "train_loss_bc": 0.03633647784590721, "train_loss_llm": 0.0001855790615081787, "grad_norm": 0.03344766050577164, "global_step": 14285, "epoch": 396, "lr": 0.001} +{"train_loss": 0.02991269715130329, "train_loss_bc": 0.02991269715130329, "train_loss_llm": 0.0, "grad_norm": 0.04383637756109238, "global_step": 14286, "epoch": 396, "lr": 0.001} +{"train_loss": 0.034055981785058975, "train_loss_bc": 0.034055981785058975, "train_loss_llm": 0.0, "grad_norm": 0.060964833945035934, "global_step": 14287, "epoch": 396, "lr": 0.001} +{"train_loss": 0.031455591320991516, "train_loss_bc": 0.03145388886332512, "train_loss_llm": 0.00017008185386657715, "grad_norm": 0.08279750496149063, "global_step": 14288, "epoch": 396, "lr": 0.001} +{"train_loss": 0.0376642569899559, "train_loss_bc": 0.03764467313885689, "train_loss_llm": 0.001958489418029785, "grad_norm": 0.022300224751234055, "global_step": 14289, "epoch": 396, "lr": 0.001} +{"train_loss": 0.031979914754629135, "train_loss_bc": 0.031976185739040375, "train_loss_llm": 0.00037285685539245605, "grad_norm": 0.03244367986917496, "global_step": 14290, "epoch": 396, "lr": 0.001} +{"train_loss": 0.03619796363636851, "train_loss_bc": 0.0396437868475914, "train_loss_llm": 0.0008838176727294922, "grad_norm": 0.05648583173751831, "global_step": 14291, "epoch": 396, "lr": 0.001, "val_loss": 0.06837177276611328} +{"train_loss": 0.03742546588182449, "train_loss_bc": 0.03742309287190437, "train_loss_llm": 0.00023731589317321777, "grad_norm": 0.08245904743671417, "global_step": 14292, "epoch": 397, "lr": 0.001} +{"train_loss": 0.034049805253744125, "train_loss_bc": 0.034049805253744125, "train_loss_llm": 0.0, "grad_norm": 0.023482931777834892, "global_step": 14293, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03379221260547638, "train_loss_bc": 0.03379221260547638, "train_loss_llm": 0.0, "grad_norm": 0.03305605798959732, "global_step": 14294, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03804757818579674, "train_loss_bc": 0.03804757818579674, "train_loss_llm": 0.0, "grad_norm": 0.046652425080537796, "global_step": 14295, "epoch": 397, "lr": 0.001} +{"train_loss": 0.033221665769815445, "train_loss_bc": 0.033221665769815445, "train_loss_llm": 0.0, "grad_norm": 0.05095670372247696, "global_step": 14296, "epoch": 397, "lr": 0.001} +{"train_loss": 0.037643615156412125, "train_loss_bc": 0.037643615156412125, "train_loss_llm": 0.0, "grad_norm": 0.028383467346429825, "global_step": 14297, "epoch": 397, "lr": 0.001} +{"train_loss": 0.0429813377559185, "train_loss_bc": 0.0429813377559185, "train_loss_llm": 0.0, "grad_norm": 0.04249066486954689, "global_step": 14298, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03225827217102051, "train_loss_bc": 0.03225827217102051, "train_loss_llm": 0.0, "grad_norm": 0.05102855712175369, "global_step": 14299, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03406870365142822, "train_loss_bc": 0.03406870365142822, "train_loss_llm": 0.0, "grad_norm": 0.06918307393789291, "global_step": 14300, "epoch": 397, "lr": 0.001} +{"train_loss": 0.0498184859752655, "train_loss_bc": 0.0498184859752655, "train_loss_llm": 0.0, "grad_norm": 0.03317033126950264, "global_step": 14301, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03420903533697128, "train_loss_bc": 0.03419549763202667, "train_loss_llm": 0.0013537108898162842, "grad_norm": 0.04717352241277695, "global_step": 14302, "epoch": 397, "lr": 0.001} +{"train_loss": 0.033685240894556046, "train_loss_bc": 0.033685240894556046, "train_loss_llm": 0.0, "grad_norm": 0.05583444610238075, "global_step": 14303, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03509821370244026, "train_loss_bc": 0.03509821370244026, "train_loss_llm": 0.0, "grad_norm": 0.06366878002882004, "global_step": 14304, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03718341141939163, "train_loss_bc": 0.03718341141939163, "train_loss_llm": 0.0, "grad_norm": 0.023576796054840088, "global_step": 14305, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03884611278772354, "train_loss_bc": 0.03884611278772354, "train_loss_llm": 0.0, "grad_norm": 0.038530975580215454, "global_step": 14306, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03397513926029205, "train_loss_bc": 0.03397513926029205, "train_loss_llm": 0.0, "grad_norm": 0.04685726761817932, "global_step": 14307, "epoch": 397, "lr": 0.001} +{"train_loss": 0.045610520988702774, "train_loss_bc": 0.04560161754488945, "train_loss_llm": 0.000890195369720459, "grad_norm": 0.05972542241215706, "global_step": 14308, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03968241438269615, "train_loss_bc": 0.03968241438269615, "train_loss_llm": 0.0, "grad_norm": 0.035746388137340546, "global_step": 14309, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03637611120939255, "train_loss_bc": 0.03637611120939255, "train_loss_llm": 0.0, "grad_norm": 0.05202815309166908, "global_step": 14310, "epoch": 397, "lr": 0.001} +{"train_loss": 0.028675366193056107, "train_loss_bc": 0.028675366193056107, "train_loss_llm": 0.0, "grad_norm": 0.06363901495933533, "global_step": 14311, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03888043016195297, "train_loss_bc": 0.03887315094470978, "train_loss_llm": 0.0007281005382537842, "grad_norm": 0.08163883537054062, "global_step": 14312, "epoch": 397, "lr": 0.001} +{"train_loss": 0.02817029319703579, "train_loss_bc": 0.02817029319703579, "train_loss_llm": 0.0, "grad_norm": 0.026468338444828987, "global_step": 14313, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03593580797314644, "train_loss_bc": 0.03593580797314644, "train_loss_llm": 0.0, "grad_norm": 0.047603510320186615, "global_step": 14314, "epoch": 397, "lr": 0.001} +{"train_loss": 0.02998887002468109, "train_loss_bc": 0.029977506026625633, "train_loss_llm": 0.001136481761932373, "grad_norm": 0.06307949870824814, "global_step": 14315, "epoch": 397, "lr": 0.001} +{"train_loss": 0.045118410140275955, "train_loss_bc": 0.045118410140275955, "train_loss_llm": 0.0, "grad_norm": 0.06889773160219193, "global_step": 14316, "epoch": 397, "lr": 0.001} +{"train_loss": 0.034190159291028976, "train_loss_bc": 0.034157294780015945, "train_loss_llm": 0.003286510705947876, "grad_norm": 0.018967457115650177, "global_step": 14317, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03199263662099838, "train_loss_bc": 0.031988851726055145, "train_loss_llm": 0.00037848949432373047, "grad_norm": 0.02700703963637352, "global_step": 14318, "epoch": 397, "lr": 0.001} +{"train_loss": 0.027204202488064766, "train_loss_bc": 0.02719579078257084, "train_loss_llm": 0.0008410811424255371, "grad_norm": 0.041521381586790085, "global_step": 14319, "epoch": 397, "lr": 0.001} +{"train_loss": 0.029008181765675545, "train_loss_bc": 0.029008181765675545, "train_loss_llm": 0.0, "grad_norm": 0.049590177834033966, "global_step": 14320, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03288416564464569, "train_loss_bc": 0.03288416564464569, "train_loss_llm": 0.0, "grad_norm": 0.022102227434515953, "global_step": 14321, "epoch": 397, "lr": 0.001} +{"train_loss": 0.030854620039463043, "train_loss_bc": 0.03084920533001423, "train_loss_llm": 0.0005415380001068115, "grad_norm": 0.0361156091094017, "global_step": 14322, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03903450444340706, "train_loss_bc": 0.03903450444340706, "train_loss_llm": 0.0, "grad_norm": 0.051873743534088135, "global_step": 14323, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03708955645561218, "train_loss_bc": 0.03708955645561218, "train_loss_llm": 0.0, "grad_norm": 0.05613389238715172, "global_step": 14324, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03833715245127678, "train_loss_bc": 0.03833167627453804, "train_loss_llm": 0.0005474984645843506, "grad_norm": 0.023948103189468384, "global_step": 14325, "epoch": 397, "lr": 0.001} +{"train_loss": 0.04113923758268356, "train_loss_bc": 0.041135095059871674, "train_loss_llm": 0.0004143714904785156, "grad_norm": 0.04112657532095909, "global_step": 14326, "epoch": 397, "lr": 0.001} +{"train_loss": 0.03598728118878272, "train_loss_bc": 0.03905407711863518, "train_loss_llm": 0.0011109113693237305, "grad_norm": 0.05337187275290489, "global_step": 14327, "epoch": 397, "lr": 0.001, "val_loss": 0.06755601614713669} +{"train_loss": 0.03212368115782738, "train_loss_bc": 0.03212368115782738, "train_loss_llm": 0.0, "grad_norm": 0.06478134542703629, "global_step": 14328, "epoch": 398, "lr": 0.001} +{"train_loss": 0.030708760023117065, "train_loss_bc": 0.03070140816271305, "train_loss_llm": 0.000735163688659668, "grad_norm": 0.02086782455444336, "global_step": 14329, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03667446970939636, "train_loss_bc": 0.03667446970939636, "train_loss_llm": 0.0, "grad_norm": 0.026618609204888344, "global_step": 14330, "epoch": 398, "lr": 0.001} +{"train_loss": 0.0361863449215889, "train_loss_bc": 0.03617902845144272, "train_loss_llm": 0.0007316768169403076, "grad_norm": 0.044212110340595245, "global_step": 14331, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03959767892956734, "train_loss_bc": 0.039591606706380844, "train_loss_llm": 0.0006073713302612305, "grad_norm": 0.06489431858062744, "global_step": 14332, "epoch": 398, "lr": 0.001} +{"train_loss": 0.02981914021074772, "train_loss_bc": 0.029803704470396042, "train_loss_llm": 0.001543581485748291, "grad_norm": 0.02729782462120056, "global_step": 14333, "epoch": 398, "lr": 0.001} +{"train_loss": 0.041595518589019775, "train_loss_bc": 0.04156336188316345, "train_loss_llm": 0.003215700387954712, "grad_norm": 0.04165930673480034, "global_step": 14334, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03939071670174599, "train_loss_bc": 0.03939071670174599, "train_loss_llm": 0.0, "grad_norm": 0.057792771607637405, "global_step": 14335, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03394297882914543, "train_loss_bc": 0.03394297882914543, "train_loss_llm": 0.0, "grad_norm": 0.07674262672662735, "global_step": 14336, "epoch": 398, "lr": 0.001} +{"train_loss": 0.0354916974902153, "train_loss_bc": 0.0354916974902153, "train_loss_llm": 0.0, "grad_norm": 0.029170040041208267, "global_step": 14337, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03974037244915962, "train_loss_bc": 0.03974037244915962, "train_loss_llm": 0.0, "grad_norm": 0.04547752067446709, "global_step": 14338, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03818564862012863, "train_loss_bc": 0.038139715790748596, "train_loss_llm": 0.004593163728713989, "grad_norm": 0.060083530843257904, "global_step": 14339, "epoch": 398, "lr": 0.001} +{"train_loss": 0.031851544976234436, "train_loss_bc": 0.031851544976234436, "train_loss_llm": 0.0, "grad_norm": 0.08063618838787079, "global_step": 14340, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03955350071191788, "train_loss_bc": 0.03951164335012436, "train_loss_llm": 0.004185795783996582, "grad_norm": 0.022991932928562164, "global_step": 14341, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03607221320271492, "train_loss_bc": 0.03607221320271492, "train_loss_llm": 0.0, "grad_norm": 0.03718823939561844, "global_step": 14342, "epoch": 398, "lr": 0.001} +{"train_loss": 0.026466842740774155, "train_loss_bc": 0.026466842740774155, "train_loss_llm": 0.0, "grad_norm": 0.05326027795672417, "global_step": 14343, "epoch": 398, "lr": 0.001} +{"train_loss": 0.033683810383081436, "train_loss_bc": 0.033683810383081436, "train_loss_llm": 0.0, "grad_norm": 0.06461839377880096, "global_step": 14344, "epoch": 398, "lr": 0.001} +{"train_loss": 0.02942570298910141, "train_loss_bc": 0.029410962015390396, "train_loss_llm": 0.001474142074584961, "grad_norm": 0.017650583758950233, "global_step": 14345, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03317751735448837, "train_loss_bc": 0.033160384744405746, "train_loss_llm": 0.001713186502456665, "grad_norm": 0.031519241631031036, "global_step": 14346, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03574613109230995, "train_loss_bc": 0.03574613109230995, "train_loss_llm": 0.0, "grad_norm": 0.041565071791410446, "global_step": 14347, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03553859144449234, "train_loss_bc": 0.03553859144449234, "train_loss_llm": 0.0, "grad_norm": 0.051238834857940674, "global_step": 14348, "epoch": 398, "lr": 0.001} +{"train_loss": 0.043139323592185974, "train_loss_bc": 0.043139323592185974, "train_loss_llm": 0.0, "grad_norm": 0.025517817586660385, "global_step": 14349, "epoch": 398, "lr": 0.001} +{"train_loss": 0.036271728575229645, "train_loss_bc": 0.036271728575229645, "train_loss_llm": 0.0, "grad_norm": 0.0385318286716938, "global_step": 14350, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03255176171660423, "train_loss_bc": 0.03255176171660423, "train_loss_llm": 0.0, "grad_norm": 0.05226090922951698, "global_step": 14351, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03911007568240166, "train_loss_bc": 0.03911007568240166, "train_loss_llm": 0.0, "grad_norm": 0.06409581750631332, "global_step": 14352, "epoch": 398, "lr": 0.001} +{"train_loss": 0.028332486748695374, "train_loss_bc": 0.028332486748695374, "train_loss_llm": 0.0, "grad_norm": 0.015062359161674976, "global_step": 14353, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03981221094727516, "train_loss_bc": 0.03981221094727516, "train_loss_llm": 0.0, "grad_norm": 0.03676529601216316, "global_step": 14354, "epoch": 398, "lr": 0.001} +{"train_loss": 0.04081900417804718, "train_loss_bc": 0.04081900417804718, "train_loss_llm": 0.0, "grad_norm": 0.05319150909781456, "global_step": 14355, "epoch": 398, "lr": 0.001} +{"train_loss": 0.042422086000442505, "train_loss_bc": 0.0424058698117733, "train_loss_llm": 0.0016217231750488281, "grad_norm": 0.07158064097166061, "global_step": 14356, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03884240984916687, "train_loss_bc": 0.03884240984916687, "train_loss_llm": 0.0, "grad_norm": 0.03697197511792183, "global_step": 14357, "epoch": 398, "lr": 0.001} +{"train_loss": 0.041158322244882584, "train_loss_bc": 0.041158322244882584, "train_loss_llm": 0.0, "grad_norm": 0.04299032688140869, "global_step": 14358, "epoch": 398, "lr": 0.001} +{"train_loss": 0.033843934535980225, "train_loss_bc": 0.033843934535980225, "train_loss_llm": 0.0, "grad_norm": 0.050209011882543564, "global_step": 14359, "epoch": 398, "lr": 0.001} +{"train_loss": 0.034456875175237656, "train_loss_bc": 0.034456875175237656, "train_loss_llm": 0.0, "grad_norm": 0.06771548092365265, "global_step": 14360, "epoch": 398, "lr": 0.001} +{"train_loss": 0.0371476374566555, "train_loss_bc": 0.0371476374566555, "train_loss_llm": 0.0, "grad_norm": 0.020182358101010323, "global_step": 14361, "epoch": 398, "lr": 0.001} +{"train_loss": 0.03709682822227478, "train_loss_bc": 0.03709682822227478, "train_loss_llm": 0.0, "grad_norm": 0.0357583612203598, "global_step": 14362, "epoch": 398, "lr": 0.001} +{"train_loss": 0.036073914573838316, "train_loss_bc": 0.03868337720632553, "train_loss_llm": 0.0, "grad_norm": 0.0513962097465992, "global_step": 14363, "epoch": 398, "lr": 0.001, "val_loss": 0.06853049993515015} +{"train_loss": 0.030724147334694862, "train_loss_bc": 0.030724147334694862, "train_loss_llm": 0.0, "grad_norm": 0.05918756499886513, "global_step": 14364, "epoch": 399, "lr": 0.001} +{"train_loss": 0.034278035163879395, "train_loss_bc": 0.034278035163879395, "train_loss_llm": 0.0, "grad_norm": 0.023517293855547905, "global_step": 14365, "epoch": 399, "lr": 0.001} +{"train_loss": 0.036259230226278305, "train_loss_bc": 0.036259230226278305, "train_loss_llm": 0.0, "grad_norm": 0.03381304815411568, "global_step": 14366, "epoch": 399, "lr": 0.001} +{"train_loss": 0.033772893249988556, "train_loss_bc": 0.033772893249988556, "train_loss_llm": 0.0, "grad_norm": 0.048973605036735535, "global_step": 14367, "epoch": 399, "lr": 0.001} +{"train_loss": 0.04090834781527519, "train_loss_bc": 0.04090834781527519, "train_loss_llm": 0.0, "grad_norm": 0.05998493731021881, "global_step": 14368, "epoch": 399, "lr": 0.001} +{"train_loss": 0.040015287697315216, "train_loss_bc": 0.04001232981681824, "train_loss_llm": 0.0002957284450531006, "grad_norm": 0.03263326361775398, "global_step": 14369, "epoch": 399, "lr": 0.001} +{"train_loss": 0.04145126789808273, "train_loss_bc": 0.04145083948969841, "train_loss_llm": 4.279613494873047e-05, "grad_norm": 0.0472535602748394, "global_step": 14370, "epoch": 399, "lr": 0.001} +{"train_loss": 0.028861278668045998, "train_loss_bc": 0.028861278668045998, "train_loss_llm": 0.0, "grad_norm": 0.054628290235996246, "global_step": 14371, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03977057337760925, "train_loss_bc": 0.03977057337760925, "train_loss_llm": 0.0, "grad_norm": 0.06162967160344124, "global_step": 14372, "epoch": 399, "lr": 0.001} +{"train_loss": 0.026440367102622986, "train_loss_bc": 0.026440367102622986, "train_loss_llm": 0.0, "grad_norm": 0.01912963017821312, "global_step": 14373, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03734784200787544, "train_loss_bc": 0.03734784200787544, "train_loss_llm": 0.0, "grad_norm": 0.03716364502906799, "global_step": 14374, "epoch": 399, "lr": 0.001} +{"train_loss": 0.037350546568632126, "train_loss_bc": 0.037350546568632126, "train_loss_llm": 0.0, "grad_norm": 0.04954482614994049, "global_step": 14375, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03613399341702461, "train_loss_bc": 0.03613399341702461, "train_loss_llm": 0.0, "grad_norm": 0.0518261082470417, "global_step": 14376, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03499720245599747, "train_loss_bc": 0.03499720245599747, "train_loss_llm": 0.0, "grad_norm": 0.01809675805270672, "global_step": 14377, "epoch": 399, "lr": 0.001} +{"train_loss": 0.039054397493600845, "train_loss_bc": 0.039054397493600845, "train_loss_llm": 0.0, "grad_norm": 0.028132135048508644, "global_step": 14378, "epoch": 399, "lr": 0.001} +{"train_loss": 0.029953723773360252, "train_loss_bc": 0.029953723773360252, "train_loss_llm": 0.0, "grad_norm": 0.03248794004321098, "global_step": 14379, "epoch": 399, "lr": 0.001} +{"train_loss": 0.04153769463300705, "train_loss_bc": 0.04153769463300705, "train_loss_llm": 0.0, "grad_norm": 0.04877594858407974, "global_step": 14380, "epoch": 399, "lr": 0.001} +{"train_loss": 0.032024502754211426, "train_loss_bc": 0.03202417492866516, "train_loss_llm": 3.2961368560791016e-05, "grad_norm": 0.02335066720843315, "global_step": 14381, "epoch": 399, "lr": 0.001} +{"train_loss": 0.043400488793849945, "train_loss_bc": 0.043395522981882095, "train_loss_llm": 0.0004965364933013916, "grad_norm": 0.044844865798950195, "global_step": 14382, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03527180850505829, "train_loss_bc": 0.03526860475540161, "train_loss_llm": 0.0003205239772796631, "grad_norm": 0.05235498771071434, "global_step": 14383, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03817513957619667, "train_loss_bc": 0.03817075118422508, "train_loss_llm": 0.0004387497901916504, "grad_norm": 0.0682578831911087, "global_step": 14384, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03370862826704979, "train_loss_bc": 0.0337035208940506, "train_loss_llm": 0.0005109012126922607, "grad_norm": 0.023421257734298706, "global_step": 14385, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03873520717024803, "train_loss_bc": 0.03873340040445328, "train_loss_llm": 0.0001805126667022705, "grad_norm": 0.04145917296409607, "global_step": 14386, "epoch": 399, "lr": 0.001} +{"train_loss": 0.036956772208213806, "train_loss_bc": 0.036956772208213806, "train_loss_llm": 0.0, "grad_norm": 0.046542856842279434, "global_step": 14387, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03051888942718506, "train_loss_bc": 0.03051888942718506, "train_loss_llm": 0.0, "grad_norm": 0.06028897687792778, "global_step": 14388, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03992437943816185, "train_loss_bc": 0.03992437943816185, "train_loss_llm": 0.0, "grad_norm": 0.023392807692289352, "global_step": 14389, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03407251089811325, "train_loss_bc": 0.03407251089811325, "train_loss_llm": 0.0, "grad_norm": 0.04400844871997833, "global_step": 14390, "epoch": 399, "lr": 0.001} +{"train_loss": 0.034148283302783966, "train_loss_bc": 0.034106723964214325, "train_loss_llm": 0.004155784845352173, "grad_norm": 0.058277517557144165, "global_step": 14391, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03585391491651535, "train_loss_bc": 0.03585391491651535, "train_loss_llm": 0.0, "grad_norm": 0.07276058942079544, "global_step": 14392, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03464340418577194, "train_loss_bc": 0.03464340418577194, "train_loss_llm": 0.0, "grad_norm": 0.026899751275777817, "global_step": 14393, "epoch": 399, "lr": 0.001} +{"train_loss": 0.030859295278787613, "train_loss_bc": 0.03085358813405037, "train_loss_llm": 0.0005707442760467529, "grad_norm": 0.03264284133911133, "global_step": 14394, "epoch": 399, "lr": 0.001} +{"train_loss": 0.036415230482816696, "train_loss_bc": 0.03640938922762871, "train_loss_llm": 0.0005840063095092773, "grad_norm": 0.04189372807741165, "global_step": 14395, "epoch": 399, "lr": 0.001} +{"train_loss": 0.037986211478710175, "train_loss_bc": 0.03798206150531769, "train_loss_llm": 0.00041496753692626953, "grad_norm": 0.05317223444581032, "global_step": 14396, "epoch": 399, "lr": 0.001} +{"train_loss": 0.037605758756399155, "train_loss_bc": 0.037605758756399155, "train_loss_llm": 0.0, "grad_norm": 0.02053273282945156, "global_step": 14397, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03509319946169853, "train_loss_bc": 0.03508518636226654, "train_loss_llm": 0.0008012950420379639, "grad_norm": 0.03576745092868805, "global_step": 14398, "epoch": 399, "lr": 0.001} +{"train_loss": 0.03578089607051677, "train_loss_bc": 0.033861804753541946, "train_loss_llm": 0.0, "grad_norm": 0.0500309132039547, "global_step": 14399, "epoch": 399, "lr": 0.001, "val_loss": 0.06737850606441498} +{"train_loss": 0.03375997766852379, "train_loss_bc": 0.03375649452209473, "train_loss_llm": 0.00034815073013305664, "grad_norm": 0.05914991348981857, "global_step": 14400, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03202974051237106, "train_loss_bc": 0.03202974051237106, "train_loss_llm": 0.0, "grad_norm": 0.02323140949010849, "global_step": 14401, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03785935044288635, "train_loss_bc": 0.0378563329577446, "train_loss_llm": 0.00030159950256347656, "grad_norm": 0.0382508859038353, "global_step": 14402, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03600455820560455, "train_loss_bc": 0.03600455820560455, "train_loss_llm": 0.0, "grad_norm": 0.04907892271876335, "global_step": 14403, "epoch": 400, "lr": 0.001} +{"train_loss": 0.036069195717573166, "train_loss_bc": 0.036069195717573166, "train_loss_llm": 0.0, "grad_norm": 0.06406338512897491, "global_step": 14404, "epoch": 400, "lr": 0.001} +{"train_loss": 0.032427091151475906, "train_loss_bc": 0.032427091151475906, "train_loss_llm": 0.0, "grad_norm": 0.024849804118275642, "global_step": 14405, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03599290922284126, "train_loss_bc": 0.035982340574264526, "train_loss_llm": 0.0010568499565124512, "grad_norm": 0.04347722604870796, "global_step": 14406, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03477475419640541, "train_loss_bc": 0.03477475419640541, "train_loss_llm": 0.0, "grad_norm": 0.05644622817635536, "global_step": 14407, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03627518191933632, "train_loss_bc": 0.03627518191933632, "train_loss_llm": 0.0, "grad_norm": 0.07150065898895264, "global_step": 14408, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03598787263035774, "train_loss_bc": 0.03598787263035774, "train_loss_llm": 0.0, "grad_norm": 0.024512289091944695, "global_step": 14409, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03739598020911217, "train_loss_bc": 0.03737788647413254, "train_loss_llm": 0.0018093883991241455, "grad_norm": 0.0396127887070179, "global_step": 14410, "epoch": 400, "lr": 0.001} +{"train_loss": 0.0359463132917881, "train_loss_bc": 0.035931672900915146, "train_loss_llm": 0.0014641284942626953, "grad_norm": 0.054579269140958786, "global_step": 14411, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03496899828314781, "train_loss_bc": 0.03496899828314781, "train_loss_llm": 0.0, "grad_norm": 0.06431598961353302, "global_step": 14412, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03313582390546799, "train_loss_bc": 0.03313582390546799, "train_loss_llm": 0.0, "grad_norm": 0.021540062502026558, "global_step": 14413, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03861519694328308, "train_loss_bc": 0.03861519694328308, "train_loss_llm": 0.0, "grad_norm": 0.032045938074588776, "global_step": 14414, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03538508713245392, "train_loss_bc": 0.03538508713245392, "train_loss_llm": 0.0, "grad_norm": 0.04111788794398308, "global_step": 14415, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03844750300049782, "train_loss_bc": 0.03844750300049782, "train_loss_llm": 0.0, "grad_norm": 0.055432140827178955, "global_step": 14416, "epoch": 400, "lr": 0.001} +{"train_loss": 0.032421886920928955, "train_loss_bc": 0.03241316229104996, "train_loss_llm": 0.0008722841739654541, "grad_norm": 0.018323956057429314, "global_step": 14417, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03990663215517998, "train_loss_bc": 0.03990663215517998, "train_loss_llm": 0.0, "grad_norm": 0.03283548355102539, "global_step": 14418, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03362336382269859, "train_loss_bc": 0.03362336382269859, "train_loss_llm": 0.0, "grad_norm": 0.04148656874895096, "global_step": 14419, "epoch": 400, "lr": 0.001} +{"train_loss": 0.038023896515369415, "train_loss_bc": 0.038023896515369415, "train_loss_llm": 0.0, "grad_norm": 0.054684966802597046, "global_step": 14420, "epoch": 400, "lr": 0.001} +{"train_loss": 0.02728435769677162, "train_loss_bc": 0.02727682888507843, "train_loss_llm": 0.000752866268157959, "grad_norm": 0.023029843345284462, "global_step": 14421, "epoch": 400, "lr": 0.001} +{"train_loss": 0.042207278311252594, "train_loss_bc": 0.042207278311252594, "train_loss_llm": 0.0, "grad_norm": 0.0467182956635952, "global_step": 14422, "epoch": 400, "lr": 0.001} +{"train_loss": 0.034171536564826965, "train_loss_bc": 0.034145016223192215, "train_loss_llm": 0.002652078866958618, "grad_norm": 0.0568789578974247, "global_step": 14423, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03569696098566055, "train_loss_bc": 0.03569696098566055, "train_loss_llm": 0.0, "grad_norm": 0.07532790303230286, "global_step": 14424, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03853149712085724, "train_loss_bc": 0.03852491080760956, "train_loss_llm": 0.0006587803363800049, "grad_norm": 0.026035187765955925, "global_step": 14425, "epoch": 400, "lr": 0.001} +{"train_loss": 0.038543883711099625, "train_loss_bc": 0.038543883711099625, "train_loss_llm": 0.0, "grad_norm": 0.04595881700515747, "global_step": 14426, "epoch": 400, "lr": 0.001} +{"train_loss": 0.036258213222026825, "train_loss_bc": 0.03624793887138367, "train_loss_llm": 0.001027613878250122, "grad_norm": 0.06225832924246788, "global_step": 14427, "epoch": 400, "lr": 0.001} +{"train_loss": 0.040598973631858826, "train_loss_bc": 0.040598973631858826, "train_loss_llm": 0.0, "grad_norm": 0.08365274220705032, "global_step": 14428, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03513050079345703, "train_loss_bc": 0.03513050079345703, "train_loss_llm": 0.0, "grad_norm": 0.021968219429254532, "global_step": 14429, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03719180077314377, "train_loss_bc": 0.03719180077314377, "train_loss_llm": 0.0, "grad_norm": 0.033227019011974335, "global_step": 14430, "epoch": 400, "lr": 0.001} +{"train_loss": 0.036807090044021606, "train_loss_bc": 0.03678128495812416, "train_loss_llm": 0.0025804638862609863, "grad_norm": 0.05000749230384827, "global_step": 14431, "epoch": 400, "lr": 0.001} +{"train_loss": 0.039358410984277725, "train_loss_bc": 0.039358410984277725, "train_loss_llm": 0.0, "grad_norm": 0.06940435618162155, "global_step": 14432, "epoch": 400, "lr": 0.001} +{"train_loss": 0.028905339539051056, "train_loss_bc": 0.028905339539051056, "train_loss_llm": 0.0, "grad_norm": 0.020249273627996445, "global_step": 14433, "epoch": 400, "lr": 0.001} +{"train_loss": 0.0348534993827343, "train_loss_bc": 0.0348534993827343, "train_loss_llm": 0.0, "grad_norm": 0.037020474672317505, "global_step": 14434, "epoch": 400, "lr": 0.001} +{"train_loss": 0.03596439347085026, "train_loss_bc": 0.04012750834226608, "train_loss_llm": 0.0, "grad_norm": 0.05312143638730049, "global_step": 14435, "epoch": 400, "lr": 0.001, "train/cumulative_reward": 70.80543384069779, "train/mean_score": 8.540648387673013, "train/success_rate": 0.2, "test/cumulative_reward": 40.1106659920336, "test/mean_score": 5.23862204187958, "test/success_rate": 0.1, "val_loss": 0.06896382570266724, "train_action_mse_error": 0.038860101252794266} +{"train_loss": 0.03308434411883354, "train_loss_bc": 0.033083777874708176, "train_loss_llm": 5.6803226470947266e-05, "grad_norm": 0.06133541092276573, "global_step": 14436, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03505421057343483, "train_loss_bc": 0.03505218029022217, "train_loss_llm": 0.0002028942108154297, "grad_norm": 0.02066434733569622, "global_step": 14437, "epoch": 401, "lr": 0.001} +{"train_loss": 0.04079636186361313, "train_loss_bc": 0.04079466313123703, "train_loss_llm": 0.00016990303993225098, "grad_norm": 0.034572534263134, "global_step": 14438, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03768264502286911, "train_loss_bc": 0.03768264502286911, "train_loss_llm": 0.0, "grad_norm": 0.046588778495788574, "global_step": 14439, "epoch": 401, "lr": 0.001} +{"train_loss": 0.04016668349504471, "train_loss_bc": 0.04016447439789772, "train_loss_llm": 0.00022101402282714844, "grad_norm": 0.057505905628204346, "global_step": 14440, "epoch": 401, "lr": 0.001} +{"train_loss": 0.032853689044713974, "train_loss_bc": 0.032853689044713974, "train_loss_llm": 0.0, "grad_norm": 0.02647266909480095, "global_step": 14441, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03720412775874138, "train_loss_bc": 0.037198349833488464, "train_loss_llm": 0.0005779266357421875, "grad_norm": 0.03588259592652321, "global_step": 14442, "epoch": 401, "lr": 0.001} +{"train_loss": 0.036302123218774796, "train_loss_bc": 0.036302123218774796, "train_loss_llm": 0.0, "grad_norm": 0.04320443421602249, "global_step": 14443, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03093111701309681, "train_loss_bc": 0.03093111701309681, "train_loss_llm": 0.0, "grad_norm": 0.04998257756233215, "global_step": 14444, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03427509963512421, "train_loss_bc": 0.03427509963512421, "train_loss_llm": 0.0, "grad_norm": 0.025543997064232826, "global_step": 14445, "epoch": 401, "lr": 0.001} +{"train_loss": 0.04445892572402954, "train_loss_bc": 0.04445892572402954, "train_loss_llm": 0.0, "grad_norm": 0.047613535076379776, "global_step": 14446, "epoch": 401, "lr": 0.001} +{"train_loss": 0.04509763792157173, "train_loss_bc": 0.04509763792157173, "train_loss_llm": 0.0, "grad_norm": 0.06470736861228943, "global_step": 14447, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03729407861828804, "train_loss_bc": 0.03729407861828804, "train_loss_llm": 0.0, "grad_norm": 0.08461124449968338, "global_step": 14448, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03161374852061272, "train_loss_bc": 0.03161374852061272, "train_loss_llm": 0.0, "grad_norm": 0.02166736125946045, "global_step": 14449, "epoch": 401, "lr": 0.001} +{"train_loss": 0.030314119532704353, "train_loss_bc": 0.030296260491013527, "train_loss_llm": 0.0017858445644378662, "grad_norm": 0.033867765218019485, "global_step": 14450, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03914007544517517, "train_loss_bc": 0.03914007544517517, "train_loss_llm": 0.0, "grad_norm": 0.04874482750892639, "global_step": 14451, "epoch": 401, "lr": 0.001} +{"train_loss": 0.04170473664999008, "train_loss_bc": 0.04169168695807457, "train_loss_llm": 0.0013049840927124023, "grad_norm": 0.07747602462768555, "global_step": 14452, "epoch": 401, "lr": 0.001} +{"train_loss": 0.04127134010195732, "train_loss_bc": 0.0412636399269104, "train_loss_llm": 0.0007700324058532715, "grad_norm": 0.019541630521416664, "global_step": 14453, "epoch": 401, "lr": 0.001} +{"train_loss": 0.039481841027736664, "train_loss_bc": 0.039481841027736664, "train_loss_llm": 0.0, "grad_norm": 0.03602873906493187, "global_step": 14454, "epoch": 401, "lr": 0.001} +{"train_loss": 0.04005272686481476, "train_loss_bc": 0.04005272686481476, "train_loss_llm": 0.0, "grad_norm": 0.053051941096782684, "global_step": 14455, "epoch": 401, "lr": 0.001} +{"train_loss": 0.027685794979333878, "train_loss_bc": 0.027685794979333878, "train_loss_llm": 0.0, "grad_norm": 0.05942145362496376, "global_step": 14456, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03223244100809097, "train_loss_bc": 0.03223244100809097, "train_loss_llm": 0.0, "grad_norm": 0.022735711187124252, "global_step": 14457, "epoch": 401, "lr": 0.001} +{"train_loss": 0.030391134321689606, "train_loss_bc": 0.030391134321689606, "train_loss_llm": 0.0, "grad_norm": 0.030645037069916725, "global_step": 14458, "epoch": 401, "lr": 0.001} +{"train_loss": 0.031127197667956352, "train_loss_bc": 0.031127197667956352, "train_loss_llm": 0.0, "grad_norm": 0.039739470928907394, "global_step": 14459, "epoch": 401, "lr": 0.001} +{"train_loss": 0.034209683537483215, "train_loss_bc": 0.034209683537483215, "train_loss_llm": 0.0, "grad_norm": 0.04534400254487991, "global_step": 14460, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03440972790122032, "train_loss_bc": 0.03440972790122032, "train_loss_llm": 0.0, "grad_norm": 0.029519185423851013, "global_step": 14461, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03466089814901352, "train_loss_bc": 0.03465826064348221, "train_loss_llm": 0.0002637505531311035, "grad_norm": 0.04444510489702225, "global_step": 14462, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03452540561556816, "train_loss_bc": 0.034516990184783936, "train_loss_llm": 0.0008414983749389648, "grad_norm": 0.06022275239229202, "global_step": 14463, "epoch": 401, "lr": 0.001} +{"train_loss": 0.028447696939110756, "train_loss_bc": 0.028447696939110756, "train_loss_llm": 0.0, "grad_norm": 0.07192502915859222, "global_step": 14464, "epoch": 401, "lr": 0.001} +{"train_loss": 0.0323956198990345, "train_loss_bc": 0.0323956198990345, "train_loss_llm": 0.0, "grad_norm": 0.028029758483171463, "global_step": 14465, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03361282870173454, "train_loss_bc": 0.03360341861844063, "train_loss_llm": 0.0009411275386810303, "grad_norm": 0.04356498643755913, "global_step": 14466, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03696583956480026, "train_loss_bc": 0.03696583956480026, "train_loss_llm": 0.0, "grad_norm": 0.06370595842599869, "global_step": 14467, "epoch": 401, "lr": 0.001} +{"train_loss": 0.044785235077142715, "train_loss_bc": 0.04477733373641968, "train_loss_llm": 0.0007900595664978027, "grad_norm": 0.08073538541793823, "global_step": 14468, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03728916123509407, "train_loss_bc": 0.03728916123509407, "train_loss_llm": 0.0, "grad_norm": 0.026273509487509727, "global_step": 14469, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03454145789146423, "train_loss_bc": 0.03453289717435837, "train_loss_llm": 0.000856250524520874, "grad_norm": 0.04633630812168121, "global_step": 14470, "epoch": 401, "lr": 0.001} +{"train_loss": 0.03595586410827107, "train_loss_bc": 0.038351353257894516, "train_loss_llm": 0.0, "grad_norm": 0.06832807511091232, "global_step": 14471, "epoch": 401, "lr": 0.001, "val_loss": 0.06872855126857758} +{"train_loss": 0.039159275591373444, "train_loss_bc": 0.03915666416287422, "train_loss_llm": 0.00026106834411621094, "grad_norm": 0.08247990906238556, "global_step": 14472, "epoch": 402, "lr": 0.001} +{"train_loss": 0.038969241082668304, "train_loss_bc": 0.038969241082668304, "train_loss_llm": 0.0, "grad_norm": 0.03119942545890808, "global_step": 14473, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03603764623403549, "train_loss_bc": 0.03603764623403549, "train_loss_llm": 0.0, "grad_norm": 0.03909045830368996, "global_step": 14474, "epoch": 402, "lr": 0.001} +{"train_loss": 0.031637903302907944, "train_loss_bc": 0.031637903302907944, "train_loss_llm": 0.0, "grad_norm": 0.05018242076039314, "global_step": 14475, "epoch": 402, "lr": 0.001} +{"train_loss": 0.031693097203969955, "train_loss_bc": 0.031693097203969955, "train_loss_llm": 0.0, "grad_norm": 0.0625353679060936, "global_step": 14476, "epoch": 402, "lr": 0.001} +{"train_loss": 0.04684056341648102, "train_loss_bc": 0.04683474078774452, "train_loss_llm": 0.0005821287631988525, "grad_norm": 0.026932908222079277, "global_step": 14477, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03471241891384125, "train_loss_bc": 0.034691374748945236, "train_loss_llm": 0.002104520797729492, "grad_norm": 0.04607662186026573, "global_step": 14478, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03264554962515831, "train_loss_bc": 0.0326421819627285, "train_loss_llm": 0.00033667683601379395, "grad_norm": 0.05450540781021118, "global_step": 14479, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03595975413918495, "train_loss_bc": 0.03595975413918495, "train_loss_llm": 0.0, "grad_norm": 0.06375432014465332, "global_step": 14480, "epoch": 402, "lr": 0.001} +{"train_loss": 0.030896350741386414, "train_loss_bc": 0.030896350741386414, "train_loss_llm": 0.0, "grad_norm": 0.022931570187211037, "global_step": 14481, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03199879452586174, "train_loss_bc": 0.03199879452586174, "train_loss_llm": 0.0, "grad_norm": 0.031021080911159515, "global_step": 14482, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03933674842119217, "train_loss_bc": 0.03933674842119217, "train_loss_llm": 0.0, "grad_norm": 0.04293382540345192, "global_step": 14483, "epoch": 402, "lr": 0.001} +{"train_loss": 0.0330466628074646, "train_loss_bc": 0.0330466628074646, "train_loss_llm": 0.0, "grad_norm": 0.05581745505332947, "global_step": 14484, "epoch": 402, "lr": 0.001} +{"train_loss": 0.039741773158311844, "train_loss_bc": 0.039741773158311844, "train_loss_llm": 0.0, "grad_norm": 0.02507227472960949, "global_step": 14485, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03817121312022209, "train_loss_bc": 0.03817121312022209, "train_loss_llm": 0.0, "grad_norm": 0.0373275987803936, "global_step": 14486, "epoch": 402, "lr": 0.001} +{"train_loss": 0.037127699702978134, "train_loss_bc": 0.037127699702978134, "train_loss_llm": 0.0, "grad_norm": 0.05040569603443146, "global_step": 14487, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03404176980257034, "train_loss_bc": 0.03404076397418976, "train_loss_llm": 0.00010058283805847168, "grad_norm": 0.06668543070554733, "global_step": 14488, "epoch": 402, "lr": 0.001} +{"train_loss": 0.034770749509334564, "train_loss_bc": 0.034770749509334564, "train_loss_llm": 0.0, "grad_norm": 0.02724200114607811, "global_step": 14489, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03656968101859093, "train_loss_bc": 0.036568913608789444, "train_loss_llm": 7.674098014831543e-05, "grad_norm": 0.048825088888406754, "global_step": 14490, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03199419379234314, "train_loss_bc": 0.03199419379234314, "train_loss_llm": 0.0, "grad_norm": 0.06389375030994415, "global_step": 14491, "epoch": 402, "lr": 0.001} +{"train_loss": 0.035189542919397354, "train_loss_bc": 0.035189542919397354, "train_loss_llm": 0.0, "grad_norm": 0.07720642536878586, "global_step": 14492, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03734414651989937, "train_loss_bc": 0.03734414651989937, "train_loss_llm": 0.0, "grad_norm": 0.026381751522421837, "global_step": 14493, "epoch": 402, "lr": 0.001} +{"train_loss": 0.040158726274967194, "train_loss_bc": 0.040158726274967194, "train_loss_llm": 0.0, "grad_norm": 0.0388956144452095, "global_step": 14494, "epoch": 402, "lr": 0.001} +{"train_loss": 0.029275696724653244, "train_loss_bc": 0.029275696724653244, "train_loss_llm": 0.0, "grad_norm": 0.05345296114683151, "global_step": 14495, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03055514022707939, "train_loss_bc": 0.030554261058568954, "train_loss_llm": 8.794665336608887e-05, "grad_norm": 0.06921732425689697, "global_step": 14496, "epoch": 402, "lr": 0.001} +{"train_loss": 0.04069480299949646, "train_loss_bc": 0.04069480299949646, "train_loss_llm": 0.0, "grad_norm": 0.03182487562298775, "global_step": 14497, "epoch": 402, "lr": 0.001} +{"train_loss": 0.043040599673986435, "train_loss_bc": 0.04302205517888069, "train_loss_llm": 0.0018543899059295654, "grad_norm": 0.04128386825323105, "global_step": 14498, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03797542303800583, "train_loss_bc": 0.03797542303800583, "train_loss_llm": 0.0, "grad_norm": 0.05004892870783806, "global_step": 14499, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03628097474575043, "train_loss_bc": 0.03628097474575043, "train_loss_llm": 0.0, "grad_norm": 0.05904623121023178, "global_step": 14500, "epoch": 402, "lr": 0.001} +{"train_loss": 0.033847205340862274, "train_loss_bc": 0.03383471816778183, "train_loss_llm": 0.0012485384941101074, "grad_norm": 0.0213625505566597, "global_step": 14501, "epoch": 402, "lr": 0.001} +{"train_loss": 0.040717337280511856, "train_loss_bc": 0.0406976081430912, "train_loss_llm": 0.0019730329513549805, "grad_norm": 0.03974602371454239, "global_step": 14502, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03896324709057808, "train_loss_bc": 0.03896324709057808, "train_loss_llm": 0.0, "grad_norm": 0.05308967083692551, "global_step": 14503, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03135793283581734, "train_loss_bc": 0.03135793283581734, "train_loss_llm": 0.0, "grad_norm": 0.0616624541580677, "global_step": 14504, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03130637854337692, "train_loss_bc": 0.03129347413778305, "train_loss_llm": 0.001290440559387207, "grad_norm": 0.026716701686382294, "global_step": 14505, "epoch": 402, "lr": 0.001} +{"train_loss": 0.03198903426527977, "train_loss_bc": 0.03198903426527977, "train_loss_llm": 0.0, "grad_norm": 0.042440015822649, "global_step": 14506, "epoch": 402, "lr": 0.001} +{"train_loss": 0.035912796958453126, "train_loss_bc": 0.038793038576841354, "train_loss_llm": 0.0020376741886138916, "grad_norm": 0.05331631377339363, "global_step": 14507, "epoch": 402, "lr": 0.001, "val_loss": 0.067986398935318} +{"train_loss": 0.04001608118414879, "train_loss_bc": 0.03998463600873947, "train_loss_llm": 0.003144383430480957, "grad_norm": 0.07983947545289993, "global_step": 14508, "epoch": 403, "lr": 0.001} +{"train_loss": 0.04106602072715759, "train_loss_bc": 0.041035138070583344, "train_loss_llm": 0.003088444471359253, "grad_norm": 0.025240683928132057, "global_step": 14509, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03826316446065903, "train_loss_bc": 0.03826316446065903, "train_loss_llm": 0.0, "grad_norm": 0.04142465442419052, "global_step": 14510, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03744003549218178, "train_loss_bc": 0.037413667887449265, "train_loss_llm": 0.0026368796825408936, "grad_norm": 0.05942186713218689, "global_step": 14511, "epoch": 403, "lr": 0.001} +{"train_loss": 0.04249619320034981, "train_loss_bc": 0.04249619320034981, "train_loss_llm": 0.0, "grad_norm": 0.07634641230106354, "global_step": 14512, "epoch": 403, "lr": 0.001} +{"train_loss": 0.034274302423000336, "train_loss_bc": 0.034274302423000336, "train_loss_llm": 0.0, "grad_norm": 0.02504928596317768, "global_step": 14513, "epoch": 403, "lr": 0.001} +{"train_loss": 0.034517642110586166, "train_loss_bc": 0.034517642110586166, "train_loss_llm": 0.0, "grad_norm": 0.03510444611310959, "global_step": 14514, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03444166108965874, "train_loss_bc": 0.03443343564867973, "train_loss_llm": 0.000822603702545166, "grad_norm": 0.043840084224939346, "global_step": 14515, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03758753463625908, "train_loss_bc": 0.03758753463625908, "train_loss_llm": 0.0, "grad_norm": 0.057721663266420364, "global_step": 14516, "epoch": 403, "lr": 0.001} +{"train_loss": 0.0413968563079834, "train_loss_bc": 0.0413968563079834, "train_loss_llm": 0.0, "grad_norm": 0.02928972989320755, "global_step": 14517, "epoch": 403, "lr": 0.001} +{"train_loss": 0.0401601567864418, "train_loss_bc": 0.04015839844942093, "train_loss_llm": 0.00017574429512023926, "grad_norm": 0.04539106413722038, "global_step": 14518, "epoch": 403, "lr": 0.001} +{"train_loss": 0.036024823784828186, "train_loss_bc": 0.036024823784828186, "train_loss_llm": 0.0, "grad_norm": 0.060558419674634933, "global_step": 14519, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03400420770049095, "train_loss_bc": 0.03400420770049095, "train_loss_llm": 0.0, "grad_norm": 0.073720782995224, "global_step": 14520, "epoch": 403, "lr": 0.001} +{"train_loss": 0.035478733479976654, "train_loss_bc": 0.035478733479976654, "train_loss_llm": 0.0, "grad_norm": 0.022649308666586876, "global_step": 14521, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03553459793329239, "train_loss_bc": 0.03553459793329239, "train_loss_llm": 0.0, "grad_norm": 0.0386231429874897, "global_step": 14522, "epoch": 403, "lr": 0.001} +{"train_loss": 0.033894024789333344, "train_loss_bc": 0.033894024789333344, "train_loss_llm": 0.0, "grad_norm": 0.05201467499136925, "global_step": 14523, "epoch": 403, "lr": 0.001} +{"train_loss": 0.030964210629463196, "train_loss_bc": 0.030964210629463196, "train_loss_llm": 0.0, "grad_norm": 0.06884825229644775, "global_step": 14524, "epoch": 403, "lr": 0.001} +{"train_loss": 0.02964300662279129, "train_loss_bc": 0.02964300662279129, "train_loss_llm": 0.0, "grad_norm": 0.0248069167137146, "global_step": 14525, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03496920317411423, "train_loss_bc": 0.03496920317411423, "train_loss_llm": 0.0, "grad_norm": 0.036339301615953445, "global_step": 14526, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03896234929561615, "train_loss_bc": 0.03896234929561615, "train_loss_llm": 0.0, "grad_norm": 0.06007334217429161, "global_step": 14527, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03325163945555687, "train_loss_bc": 0.03325163945555687, "train_loss_llm": 0.0, "grad_norm": 0.0778970867395401, "global_step": 14528, "epoch": 403, "lr": 0.001} +{"train_loss": 0.04214999079704285, "train_loss_bc": 0.04214999079704285, "train_loss_llm": 0.0, "grad_norm": 0.02917936071753502, "global_step": 14529, "epoch": 403, "lr": 0.001} +{"train_loss": 0.033659663051366806, "train_loss_bc": 0.033659663051366806, "train_loss_llm": 0.0, "grad_norm": 0.042306091636419296, "global_step": 14530, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03581719100475311, "train_loss_bc": 0.03581719100475311, "train_loss_llm": 0.0, "grad_norm": 0.05577431619167328, "global_step": 14531, "epoch": 403, "lr": 0.001} +{"train_loss": 0.04476253315806389, "train_loss_bc": 0.04476253315806389, "train_loss_llm": 0.0, "grad_norm": 0.0684216096997261, "global_step": 14532, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03570026904344559, "train_loss_bc": 0.03570026904344559, "train_loss_llm": 0.0, "grad_norm": 0.02605951391160488, "global_step": 14533, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03585483878850937, "train_loss_bc": 0.03585483878850937, "train_loss_llm": 0.0, "grad_norm": 0.04118335247039795, "global_step": 14534, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03715389594435692, "train_loss_bc": 0.03715389594435692, "train_loss_llm": 0.0, "grad_norm": 0.05411383882164955, "global_step": 14535, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03660016879439354, "train_loss_bc": 0.03660016879439354, "train_loss_llm": 0.0, "grad_norm": 0.06550966203212738, "global_step": 14536, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03675279766321182, "train_loss_bc": 0.03675279766321182, "train_loss_llm": 0.0, "grad_norm": 0.023325122892856598, "global_step": 14537, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03177360072731972, "train_loss_bc": 0.03177360072731972, "train_loss_llm": 0.0, "grad_norm": 0.03170619159936905, "global_step": 14538, "epoch": 403, "lr": 0.001} +{"train_loss": 0.031485673040151596, "train_loss_bc": 0.031485673040151596, "train_loss_llm": 0.0, "grad_norm": 0.04140002280473709, "global_step": 14539, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03370840847492218, "train_loss_bc": 0.03370840847492218, "train_loss_llm": 0.0, "grad_norm": 0.049546752125024796, "global_step": 14540, "epoch": 403, "lr": 0.001} +{"train_loss": 0.035065047442913055, "train_loss_bc": 0.035065047442913055, "train_loss_llm": 0.0, "grad_norm": 0.02248489297926426, "global_step": 14541, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03083830326795578, "train_loss_bc": 0.030829912051558495, "train_loss_llm": 0.0008391737937927246, "grad_norm": 0.033472832292318344, "global_step": 14542, "epoch": 403, "lr": 0.001} +{"train_loss": 0.03594844270911482, "train_loss_bc": 0.028435111045837402, "train_loss_llm": 0.0, "grad_norm": 0.054712653160095215, "global_step": 14543, "epoch": 403, "lr": 0.001, "val_loss": 0.06788625568151474} +{"train_loss": 0.03851299732923508, "train_loss_bc": 0.03851299732923508, "train_loss_llm": 0.0, "grad_norm": 0.0611906573176384, "global_step": 14544, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03316209092736244, "train_loss_bc": 0.03316178172826767, "train_loss_llm": 3.090500831604004e-05, "grad_norm": 0.024996263906359673, "global_step": 14545, "epoch": 404, "lr": 0.001} +{"train_loss": 0.035227105021476746, "train_loss_bc": 0.035227105021476746, "train_loss_llm": 0.0, "grad_norm": 0.03486906737089157, "global_step": 14546, "epoch": 404, "lr": 0.001} +{"train_loss": 0.037957221269607544, "train_loss_bc": 0.03794308751821518, "train_loss_llm": 0.0014135241508483887, "grad_norm": 0.042050689458847046, "global_step": 14547, "epoch": 404, "lr": 0.001} +{"train_loss": 0.04205338656902313, "train_loss_bc": 0.04205338656902313, "train_loss_llm": 0.0, "grad_norm": 0.05720805004239082, "global_step": 14548, "epoch": 404, "lr": 0.001} +{"train_loss": 0.035427384078502655, "train_loss_bc": 0.035427384078502655, "train_loss_llm": 0.0, "grad_norm": 0.028151221573352814, "global_step": 14549, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03812526538968086, "train_loss_bc": 0.03812526538968086, "train_loss_llm": 0.0, "grad_norm": 0.040959153324365616, "global_step": 14550, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03357711434364319, "train_loss_bc": 0.03357711434364319, "train_loss_llm": 0.0, "grad_norm": 0.062343377619981766, "global_step": 14551, "epoch": 404, "lr": 0.001} +{"train_loss": 0.036953527480363846, "train_loss_bc": 0.03693409636616707, "train_loss_llm": 0.0019431710243225098, "grad_norm": 0.08352776616811752, "global_step": 14552, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03198198974132538, "train_loss_bc": 0.03198198974132538, "train_loss_llm": 0.0, "grad_norm": 0.018671175464987755, "global_step": 14553, "epoch": 404, "lr": 0.001} +{"train_loss": 0.0409366711974144, "train_loss_bc": 0.04091145098209381, "train_loss_llm": 0.002521991729736328, "grad_norm": 0.0376177541911602, "global_step": 14554, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03673550486564636, "train_loss_bc": 0.03670550510287285, "train_loss_llm": 0.0029999911785125732, "grad_norm": 0.050854288041591644, "global_step": 14555, "epoch": 404, "lr": 0.001} +{"train_loss": 0.039516303688287735, "train_loss_bc": 0.039516303688287735, "train_loss_llm": 0.0, "grad_norm": 0.06537366658449173, "global_step": 14556, "epoch": 404, "lr": 0.001} +{"train_loss": 0.033453378826379776, "train_loss_bc": 0.033453378826379776, "train_loss_llm": 0.0, "grad_norm": 0.020267177373170853, "global_step": 14557, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03886175900697708, "train_loss_bc": 0.03886175900697708, "train_loss_llm": 0.0, "grad_norm": 0.027346866205334663, "global_step": 14558, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03423634544014931, "train_loss_bc": 0.0342247299849987, "train_loss_llm": 0.0011613965034484863, "grad_norm": 0.033386509865522385, "global_step": 14559, "epoch": 404, "lr": 0.001} +{"train_loss": 0.039246391505002975, "train_loss_bc": 0.039246391505002975, "train_loss_llm": 0.0, "grad_norm": 0.047265131026506424, "global_step": 14560, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03209071606397629, "train_loss_bc": 0.032089442014694214, "train_loss_llm": 0.00012728571891784668, "grad_norm": 0.02586785890161991, "global_step": 14561, "epoch": 404, "lr": 0.001} +{"train_loss": 0.0348980613052845, "train_loss_bc": 0.0348980613052845, "train_loss_llm": 0.0, "grad_norm": 0.03953775390982628, "global_step": 14562, "epoch": 404, "lr": 0.001} +{"train_loss": 0.030571894720196724, "train_loss_bc": 0.030571894720196724, "train_loss_llm": 0.0, "grad_norm": 0.055073101073503494, "global_step": 14563, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03151760622859001, "train_loss_bc": 0.03151760622859001, "train_loss_llm": 0.0, "grad_norm": 0.06640110164880753, "global_step": 14564, "epoch": 404, "lr": 0.001} +{"train_loss": 0.035895779728889465, "train_loss_bc": 0.035895779728889465, "train_loss_llm": 0.0, "grad_norm": 0.029558876529335976, "global_step": 14565, "epoch": 404, "lr": 0.001} +{"train_loss": 0.033159587532281876, "train_loss_bc": 0.03315456584095955, "train_loss_llm": 0.0005022883415222168, "grad_norm": 0.04210470989346504, "global_step": 14566, "epoch": 404, "lr": 0.001} +{"train_loss": 0.035735566169023514, "train_loss_bc": 0.035721827298402786, "train_loss_llm": 0.0013737678527832031, "grad_norm": 0.05707988888025284, "global_step": 14567, "epoch": 404, "lr": 0.001} +{"train_loss": 0.035374365746974945, "train_loss_bc": 0.035374365746974945, "train_loss_llm": 0.0, "grad_norm": 0.0658823549747467, "global_step": 14568, "epoch": 404, "lr": 0.001} +{"train_loss": 0.031041845679283142, "train_loss_bc": 0.031041845679283142, "train_loss_llm": 0.0, "grad_norm": 0.02720581740140915, "global_step": 14569, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03735729306936264, "train_loss_bc": 0.03735729306936264, "train_loss_llm": 0.0, "grad_norm": 0.040615200996398926, "global_step": 14570, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03913406655192375, "train_loss_bc": 0.03913406655192375, "train_loss_llm": 0.0, "grad_norm": 0.04949560388922691, "global_step": 14571, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03746958449482918, "train_loss_bc": 0.03746958449482918, "train_loss_llm": 0.0, "grad_norm": 0.052886880934238434, "global_step": 14572, "epoch": 404, "lr": 0.001} +{"train_loss": 0.038252100348472595, "train_loss_bc": 0.038252100348472595, "train_loss_llm": 0.0, "grad_norm": 0.021047933027148247, "global_step": 14573, "epoch": 404, "lr": 0.001} +{"train_loss": 0.035679176449775696, "train_loss_bc": 0.035679176449775696, "train_loss_llm": 0.0, "grad_norm": 0.033926308155059814, "global_step": 14574, "epoch": 404, "lr": 0.001} +{"train_loss": 0.032876353710889816, "train_loss_bc": 0.03287358954548836, "train_loss_llm": 0.00027632713317871094, "grad_norm": 0.04683929309248924, "global_step": 14575, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03573909029364586, "train_loss_bc": 0.03573909029364586, "train_loss_llm": 0.0, "grad_norm": 0.05653835088014603, "global_step": 14576, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03220384940505028, "train_loss_bc": 0.03220384940505028, "train_loss_llm": 0.0, "grad_norm": 0.0211342740803957, "global_step": 14577, "epoch": 404, "lr": 0.001} +{"train_loss": 0.03578152135014534, "train_loss_bc": 0.03573625907301903, "train_loss_llm": 0.004526048898696899, "grad_norm": 0.03533158451318741, "global_step": 14578, "epoch": 404, "lr": 0.001} +{"train_loss": 0.035824082947025694, "train_loss_bc": 0.038924090564250946, "train_loss_llm": 0.0, "grad_norm": 0.0491134412586689, "global_step": 14579, "epoch": 404, "lr": 0.001, "val_loss": 0.06873904913663864} +{"train_loss": 0.04056492820382118, "train_loss_bc": 0.04056492820382118, "train_loss_llm": 0.0, "grad_norm": 0.06330603361129761, "global_step": 14580, "epoch": 405, "lr": 0.001} +{"train_loss": 0.042339447885751724, "train_loss_bc": 0.042339447885751724, "train_loss_llm": 0.0, "grad_norm": 0.037532683461904526, "global_step": 14581, "epoch": 405, "lr": 0.001} +{"train_loss": 0.028493095189332962, "train_loss_bc": 0.028493095189332962, "train_loss_llm": 0.0, "grad_norm": 0.04749318212270737, "global_step": 14582, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03864145651459694, "train_loss_bc": 0.03864145651459694, "train_loss_llm": 0.0, "grad_norm": 0.05372537299990654, "global_step": 14583, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03439318761229515, "train_loss_bc": 0.03439318761229515, "train_loss_llm": 0.0, "grad_norm": 0.07043834030628204, "global_step": 14584, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03316136822104454, "train_loss_bc": 0.03316136822104454, "train_loss_llm": 0.0, "grad_norm": 0.026565028354525566, "global_step": 14585, "epoch": 405, "lr": 0.001} +{"train_loss": 0.032339345663785934, "train_loss_bc": 0.032339345663785934, "train_loss_llm": 0.0, "grad_norm": 0.03660016879439354, "global_step": 14586, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03415362536907196, "train_loss_bc": 0.03415362536907196, "train_loss_llm": 0.0, "grad_norm": 0.04766780510544777, "global_step": 14587, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03181782364845276, "train_loss_bc": 0.03181782364845276, "train_loss_llm": 0.0, "grad_norm": 0.05724289268255234, "global_step": 14588, "epoch": 405, "lr": 0.001} +{"train_loss": 0.04116570204496384, "train_loss_bc": 0.04114159941673279, "train_loss_llm": 0.0024101734161376953, "grad_norm": 0.02828892320394516, "global_step": 14589, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03494742885231972, "train_loss_bc": 0.03494742885231972, "train_loss_llm": 0.0, "grad_norm": 0.0459248386323452, "global_step": 14590, "epoch": 405, "lr": 0.001} +{"train_loss": 0.034350812435150146, "train_loss_bc": 0.034340035170316696, "train_loss_llm": 0.001077890396118164, "grad_norm": 0.04210079461336136, "global_step": 14591, "epoch": 405, "lr": 0.001} +{"train_loss": 0.030313529074192047, "train_loss_bc": 0.030313529074192047, "train_loss_llm": 0.0, "grad_norm": 0.0474059134721756, "global_step": 14592, "epoch": 405, "lr": 0.001} +{"train_loss": 0.04103763401508331, "train_loss_bc": 0.04103763401508331, "train_loss_llm": 0.0, "grad_norm": 0.02540009655058384, "global_step": 14593, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03248272463679314, "train_loss_bc": 0.03247695788741112, "train_loss_llm": 0.000576704740524292, "grad_norm": 0.038563717156648636, "global_step": 14594, "epoch": 405, "lr": 0.001} +{"train_loss": 0.039816588163375854, "train_loss_bc": 0.039816588163375854, "train_loss_llm": 0.0, "grad_norm": 0.0522693395614624, "global_step": 14595, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03339731693267822, "train_loss_bc": 0.033392444252967834, "train_loss_llm": 0.00048729777336120605, "grad_norm": 0.06810643523931503, "global_step": 14596, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03783080726861954, "train_loss_bc": 0.03783080726861954, "train_loss_llm": 0.0, "grad_norm": 0.029275259003043175, "global_step": 14597, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03221413120627403, "train_loss_bc": 0.03221413120627403, "train_loss_llm": 0.0, "grad_norm": 0.043612025678157806, "global_step": 14598, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03313388302922249, "train_loss_bc": 0.03313388302922249, "train_loss_llm": 0.0, "grad_norm": 0.05563308298587799, "global_step": 14599, "epoch": 405, "lr": 0.001} +{"train_loss": 0.040137358009815216, "train_loss_bc": 0.040137358009815216, "train_loss_llm": 0.0, "grad_norm": 0.07054148614406586, "global_step": 14600, "epoch": 405, "lr": 0.001} +{"train_loss": 0.0356808640062809, "train_loss_bc": 0.03567800670862198, "train_loss_llm": 0.00028565526008605957, "grad_norm": 0.028358792886137962, "global_step": 14601, "epoch": 405, "lr": 0.001} +{"train_loss": 0.036963626742362976, "train_loss_bc": 0.036963626742362976, "train_loss_llm": 0.0, "grad_norm": 0.04045536369085312, "global_step": 14602, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03426092863082886, "train_loss_bc": 0.03426092863082886, "train_loss_llm": 0.0, "grad_norm": 0.05372973904013634, "global_step": 14603, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03544032573699951, "train_loss_bc": 0.03544032573699951, "train_loss_llm": 0.0, "grad_norm": 0.06896863132715225, "global_step": 14604, "epoch": 405, "lr": 0.001} +{"train_loss": 0.034285519272089005, "train_loss_bc": 0.034285519272089005, "train_loss_llm": 0.0, "grad_norm": 0.025071149691939354, "global_step": 14605, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03823334723711014, "train_loss_bc": 0.03823334723711014, "train_loss_llm": 0.0, "grad_norm": 0.03738304227590561, "global_step": 14606, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03748207911849022, "train_loss_bc": 0.03748207911849022, "train_loss_llm": 0.0, "grad_norm": 0.044906310737133026, "global_step": 14607, "epoch": 405, "lr": 0.001} +{"train_loss": 0.04384711757302284, "train_loss_bc": 0.043819162994623184, "train_loss_llm": 0.002795279026031494, "grad_norm": 0.05250924453139305, "global_step": 14608, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03782632574439049, "train_loss_bc": 0.03782632574439049, "train_loss_llm": 0.0, "grad_norm": 0.02697259560227394, "global_step": 14609, "epoch": 405, "lr": 0.001} +{"train_loss": 0.02923283353447914, "train_loss_bc": 0.02923283353447914, "train_loss_llm": 0.0, "grad_norm": 0.03481706231832504, "global_step": 14610, "epoch": 405, "lr": 0.001} +{"train_loss": 0.036119621247053146, "train_loss_bc": 0.0361151285469532, "train_loss_llm": 0.00044929981231689453, "grad_norm": 0.0535987913608551, "global_step": 14611, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03693373501300812, "train_loss_bc": 0.03693373501300812, "train_loss_llm": 0.0, "grad_norm": 0.06461825221776962, "global_step": 14612, "epoch": 405, "lr": 0.001} +{"train_loss": 0.04104666784405708, "train_loss_bc": 0.04104666784405708, "train_loss_llm": 0.0, "grad_norm": 0.02615475095808506, "global_step": 14613, "epoch": 405, "lr": 0.001} +{"train_loss": 0.0341818667948246, "train_loss_bc": 0.0341818667948246, "train_loss_llm": 0.0, "grad_norm": 0.041718997061252594, "global_step": 14614, "epoch": 405, "lr": 0.001} +{"train_loss": 0.03578247414487931, "train_loss_bc": 0.0299020167440176, "train_loss_llm": 0.0, "grad_norm": 0.04399386793375015, "global_step": 14615, "epoch": 405, "lr": 0.001, "train/cumulative_reward": 43.20966425855605, "train/mean_score": 8.714131732085168, "train/success_rate": 0.2, "test/cumulative_reward": 41.68934307235594, "test/mean_score": 7.847927684064559, "test/success_rate": 0.16, "val_loss": 0.06874019652605057, "train_action_mse_error": 0.03895267844200134} +{"train_loss": 0.034287068992853165, "train_loss_bc": 0.034287068992853165, "train_loss_llm": 0.0, "grad_norm": 0.06205327436327934, "global_step": 14616, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03323645889759064, "train_loss_bc": 0.03323645889759064, "train_loss_llm": 0.0, "grad_norm": 0.017296627163887024, "global_step": 14617, "epoch": 406, "lr": 0.001} +{"train_loss": 0.0355362743139267, "train_loss_bc": 0.0355362743139267, "train_loss_llm": 0.0, "grad_norm": 0.03451688587665558, "global_step": 14618, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03741626814007759, "train_loss_bc": 0.037415459752082825, "train_loss_llm": 8.091330528259277e-05, "grad_norm": 0.049072835594415665, "global_step": 14619, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03988855704665184, "train_loss_bc": 0.0398729108273983, "train_loss_llm": 0.0015645325183868408, "grad_norm": 0.06620080024003983, "global_step": 14620, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03853534162044525, "train_loss_bc": 0.038535334169864655, "train_loss_llm": 6.556510925292969e-07, "grad_norm": 0.021689679473638535, "global_step": 14621, "epoch": 406, "lr": 0.001} +{"train_loss": 0.037456754595041275, "train_loss_bc": 0.037456754595041275, "train_loss_llm": 0.0, "grad_norm": 0.04611263796687126, "global_step": 14622, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03891497477889061, "train_loss_bc": 0.03891497477889061, "train_loss_llm": 0.0, "grad_norm": 0.055626314133405685, "global_step": 14623, "epoch": 406, "lr": 0.001} +{"train_loss": 0.034404635429382324, "train_loss_bc": 0.034371696412563324, "train_loss_llm": 0.0032939612865448, "grad_norm": 0.05988609418272972, "global_step": 14624, "epoch": 406, "lr": 0.001} +{"train_loss": 0.035368699580430984, "train_loss_bc": 0.035368699580430984, "train_loss_llm": 0.0, "grad_norm": 0.023760072886943817, "global_step": 14625, "epoch": 406, "lr": 0.001} +{"train_loss": 0.028812596574425697, "train_loss_bc": 0.028812596574425697, "train_loss_llm": 0.0, "grad_norm": 0.035641852766275406, "global_step": 14626, "epoch": 406, "lr": 0.001} +{"train_loss": 0.041092805564403534, "train_loss_bc": 0.04109206423163414, "train_loss_llm": 7.426738739013672e-05, "grad_norm": 0.04731274023652077, "global_step": 14627, "epoch": 406, "lr": 0.001} +{"train_loss": 0.04505832493305206, "train_loss_bc": 0.04505832493305206, "train_loss_llm": 0.0, "grad_norm": 0.05680892616510391, "global_step": 14628, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03488871082663536, "train_loss_bc": 0.03488871082663536, "train_loss_llm": 0.0, "grad_norm": 0.027678925544023514, "global_step": 14629, "epoch": 406, "lr": 0.001} +{"train_loss": 0.0339510552585125, "train_loss_bc": 0.0339510552585125, "train_loss_llm": 0.0, "grad_norm": 0.036964256316423416, "global_step": 14630, "epoch": 406, "lr": 0.001} +{"train_loss": 0.034007418900728226, "train_loss_bc": 0.034007418900728226, "train_loss_llm": 0.0, "grad_norm": 0.04662560299038887, "global_step": 14631, "epoch": 406, "lr": 0.001} +{"train_loss": 0.0388101190328598, "train_loss_bc": 0.0388101190328598, "train_loss_llm": 0.0, "grad_norm": 0.061737384647130966, "global_step": 14632, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03407072648406029, "train_loss_bc": 0.03407072648406029, "train_loss_llm": 0.0, "grad_norm": 0.025131015107035637, "global_step": 14633, "epoch": 406, "lr": 0.001} +{"train_loss": 0.034892767667770386, "train_loss_bc": 0.034892767667770386, "train_loss_llm": 0.0, "grad_norm": 0.036873940378427505, "global_step": 14634, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03475246950984001, "train_loss_bc": 0.034720294177532196, "train_loss_llm": 0.0032176971435546875, "grad_norm": 0.04891514778137207, "global_step": 14635, "epoch": 406, "lr": 0.001} +{"train_loss": 0.030048934742808342, "train_loss_bc": 0.030044032260775566, "train_loss_llm": 0.0004902184009552002, "grad_norm": 0.05686160549521446, "global_step": 14636, "epoch": 406, "lr": 0.001} +{"train_loss": 0.037932317703962326, "train_loss_bc": 0.037932317703962326, "train_loss_llm": 0.0, "grad_norm": 0.029825890436768532, "global_step": 14637, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03545865789055824, "train_loss_bc": 0.03545865789055824, "train_loss_llm": 0.0, "grad_norm": 0.04357902705669403, "global_step": 14638, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03299546241760254, "train_loss_bc": 0.03299066051840782, "train_loss_llm": 0.00048035383224487305, "grad_norm": 0.057062771171331406, "global_step": 14639, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03786168992519379, "train_loss_bc": 0.03786168992519379, "train_loss_llm": 0.0, "grad_norm": 0.06972889602184296, "global_step": 14640, "epoch": 406, "lr": 0.001} +{"train_loss": 0.02998625673353672, "train_loss_bc": 0.029979581013321877, "train_loss_llm": 0.0006675422191619873, "grad_norm": 0.01872038096189499, "global_step": 14641, "epoch": 406, "lr": 0.001} +{"train_loss": 0.037701718509197235, "train_loss_bc": 0.037701718509197235, "train_loss_llm": 0.0, "grad_norm": 0.03476779907941818, "global_step": 14642, "epoch": 406, "lr": 0.001} +{"train_loss": 0.0386473722755909, "train_loss_bc": 0.0386473722755909, "train_loss_llm": 0.0, "grad_norm": 0.053697939962148666, "global_step": 14643, "epoch": 406, "lr": 0.001} +{"train_loss": 0.04277554526925087, "train_loss_bc": 0.04277554526925087, "train_loss_llm": 0.0, "grad_norm": 0.05884072929620743, "global_step": 14644, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03313148766756058, "train_loss_bc": 0.033124055713415146, "train_loss_llm": 0.0007430613040924072, "grad_norm": 0.024948468431830406, "global_step": 14645, "epoch": 406, "lr": 0.001} +{"train_loss": 0.036486800760030746, "train_loss_bc": 0.03647187724709511, "train_loss_llm": 0.0014922022819519043, "grad_norm": 0.02947848290205002, "global_step": 14646, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03276893123984337, "train_loss_bc": 0.03276893123984337, "train_loss_llm": 0.0, "grad_norm": 0.03794471174478531, "global_step": 14647, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03551212325692177, "train_loss_bc": 0.03551212325692177, "train_loss_llm": 0.0, "grad_norm": 0.04853159934282303, "global_step": 14648, "epoch": 406, "lr": 0.001} +{"train_loss": 0.027741041034460068, "train_loss_bc": 0.027741041034460068, "train_loss_llm": 0.0, "grad_norm": 0.022150570526719093, "global_step": 14649, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03434043750166893, "train_loss_bc": 0.03434043750166893, "train_loss_llm": 0.0, "grad_norm": 0.04049649089574814, "global_step": 14650, "epoch": 406, "lr": 0.001} +{"train_loss": 0.03578730853688386, "train_loss_bc": 0.039572227746248245, "train_loss_llm": 7.361173629760742e-06, "grad_norm": 0.05449686199426651, "global_step": 14651, "epoch": 406, "lr": 0.001, "val_loss": 0.06798914074897766} +{"train_loss": 0.03553438186645508, "train_loss_bc": 0.03552643209695816, "train_loss_llm": 0.0007948875427246094, "grad_norm": 0.060388922691345215, "global_step": 14652, "epoch": 407, "lr": 0.001} +{"train_loss": 0.033623985946178436, "train_loss_bc": 0.033623743802309036, "train_loss_llm": 2.4080276489257812e-05, "grad_norm": 0.01945645548403263, "global_step": 14653, "epoch": 407, "lr": 0.001} +{"train_loss": 0.032835498452186584, "train_loss_bc": 0.032835498452186584, "train_loss_llm": 0.0, "grad_norm": 0.0302811861038208, "global_step": 14654, "epoch": 407, "lr": 0.001} +{"train_loss": 0.0373004786670208, "train_loss_bc": 0.0373004786670208, "train_loss_llm": 0.0, "grad_norm": 0.03983047977089882, "global_step": 14655, "epoch": 407, "lr": 0.001} +{"train_loss": 0.03699500858783722, "train_loss_bc": 0.03699500858783722, "train_loss_llm": 0.0, "grad_norm": 0.05147639662027359, "global_step": 14656, "epoch": 407, "lr": 0.001} +{"train_loss": 0.030190840363502502, "train_loss_bc": 0.030190840363502502, "train_loss_llm": 0.0, "grad_norm": 0.02419809252023697, "global_step": 14657, "epoch": 407, "lr": 0.001} +{"train_loss": 0.034862320870161057, "train_loss_bc": 0.03485357016324997, "train_loss_llm": 0.0008750259876251221, "grad_norm": 0.04040979966521263, "global_step": 14658, "epoch": 407, "lr": 0.001} +{"train_loss": 0.034082040190696716, "train_loss_bc": 0.03407575190067291, "train_loss_llm": 0.0006289482116699219, "grad_norm": 0.05213043466210365, "global_step": 14659, "epoch": 407, "lr": 0.001} +{"train_loss": 0.036973461508750916, "train_loss_bc": 0.036973461508750916, "train_loss_llm": 0.0, "grad_norm": 0.06225179135799408, "global_step": 14660, "epoch": 407, "lr": 0.001} +{"train_loss": 0.028695523738861084, "train_loss_bc": 0.02869333140552044, "train_loss_llm": 0.0002192854881286621, "grad_norm": 0.026972424238920212, "global_step": 14661, "epoch": 407, "lr": 0.001} +{"train_loss": 0.031214386224746704, "train_loss_bc": 0.03119019605219364, "train_loss_llm": 0.002419024705886841, "grad_norm": 0.041020289063453674, "global_step": 14662, "epoch": 407, "lr": 0.001} +{"train_loss": 0.033661503344774246, "train_loss_bc": 0.033648207783699036, "train_loss_llm": 0.0013294219970703125, "grad_norm": 0.049702707678079605, "global_step": 14663, "epoch": 407, "lr": 0.001} +{"train_loss": 0.0340619757771492, "train_loss_bc": 0.0340619757771492, "train_loss_llm": 0.0, "grad_norm": 0.05891945958137512, "global_step": 14664, "epoch": 407, "lr": 0.001} +{"train_loss": 0.03518978878855705, "train_loss_bc": 0.03518978878855705, "train_loss_llm": 0.0, "grad_norm": 0.0222786832600832, "global_step": 14665, "epoch": 407, "lr": 0.001} +{"train_loss": 0.03545739874243736, "train_loss_bc": 0.03545739874243736, "train_loss_llm": 0.0, "grad_norm": 0.035574715584516525, "global_step": 14666, "epoch": 407, "lr": 0.001} +{"train_loss": 0.039233703166246414, "train_loss_bc": 0.039233703166246414, "train_loss_llm": 0.0, "grad_norm": 0.05187245085835457, "global_step": 14667, "epoch": 407, "lr": 0.001} +{"train_loss": 0.04645344242453575, "train_loss_bc": 0.046426139771938324, "train_loss_llm": 0.0027303099632263184, "grad_norm": 0.07335622608661652, "global_step": 14668, "epoch": 407, "lr": 0.001} +{"train_loss": 0.02878103405237198, "train_loss_bc": 0.028774647042155266, "train_loss_llm": 0.0006386637687683105, "grad_norm": 0.019483473151922226, "global_step": 14669, "epoch": 407, "lr": 0.001} +{"train_loss": 0.02981814183294773, "train_loss_bc": 0.02981814183294773, "train_loss_llm": 0.0, "grad_norm": 0.03048013336956501, "global_step": 14670, "epoch": 407, "lr": 0.001} +{"train_loss": 0.04137887433171272, "train_loss_bc": 0.04137887433171272, "train_loss_llm": 0.0, "grad_norm": 0.04520215094089508, "global_step": 14671, "epoch": 407, "lr": 0.001} +{"train_loss": 0.03547374904155731, "train_loss_bc": 0.03547374904155731, "train_loss_llm": 0.0, "grad_norm": 0.056225523352622986, "global_step": 14672, "epoch": 407, "lr": 0.001} +{"train_loss": 0.04025191813707352, "train_loss_bc": 0.040237005800008774, "train_loss_llm": 0.001491159200668335, "grad_norm": 0.02898094244301319, "global_step": 14673, "epoch": 407, "lr": 0.001} +{"train_loss": 0.030899876728653908, "train_loss_bc": 0.030899876728653908, "train_loss_llm": 0.0, "grad_norm": 0.044260215014219284, "global_step": 14674, "epoch": 407, "lr": 0.001} +{"train_loss": 0.028169024735689163, "train_loss_bc": 0.02816595695912838, "train_loss_llm": 0.00030675530433654785, "grad_norm": 0.06473157554864883, "global_step": 14675, "epoch": 407, "lr": 0.001} +{"train_loss": 0.039011020213365555, "train_loss_bc": 0.039011020213365555, "train_loss_llm": 0.0, "grad_norm": 0.08383329957723618, "global_step": 14676, "epoch": 407, "lr": 0.001} +{"train_loss": 0.038923896849155426, "train_loss_bc": 0.038923896849155426, "train_loss_llm": 0.0, "grad_norm": 0.03409559652209282, "global_step": 14677, "epoch": 407, "lr": 0.001} +{"train_loss": 0.043592412024736404, "train_loss_bc": 0.043573420494794846, "train_loss_llm": 0.0018992722034454346, "grad_norm": 0.04235455393791199, "global_step": 14678, "epoch": 407, "lr": 0.001} +{"train_loss": 0.03909065201878548, "train_loss_bc": 0.03909065201878548, "train_loss_llm": 0.0, "grad_norm": 0.05399026349186897, "global_step": 14679, "epoch": 407, "lr": 0.001} +{"train_loss": 0.03233509510755539, "train_loss_bc": 0.03233509510755539, "train_loss_llm": 0.0, "grad_norm": 0.06488669663667679, "global_step": 14680, "epoch": 407, "lr": 0.001} +{"train_loss": 0.03719697147607803, "train_loss_bc": 0.03719697147607803, "train_loss_llm": 0.0, "grad_norm": 0.022155839949846268, "global_step": 14681, "epoch": 407, "lr": 0.001} +{"train_loss": 0.03304864093661308, "train_loss_bc": 0.03304864093661308, "train_loss_llm": 0.0, "grad_norm": 0.029297886416316032, "global_step": 14682, "epoch": 407, "lr": 0.001} +{"train_loss": 0.044471945613622665, "train_loss_bc": 0.044471945613622665, "train_loss_llm": 0.0, "grad_norm": 0.04213550686836243, "global_step": 14683, "epoch": 407, "lr": 0.001} +{"train_loss": 0.034211404621601105, "train_loss_bc": 0.03418157249689102, "train_loss_llm": 0.0029831230640411377, "grad_norm": 0.04823707044124603, "global_step": 14684, "epoch": 407, "lr": 0.001} +{"train_loss": 0.04089731723070145, "train_loss_bc": 0.04089731723070145, "train_loss_llm": 0.0, "grad_norm": 0.02299969270825386, "global_step": 14685, "epoch": 407, "lr": 0.001} +{"train_loss": 0.03477806597948074, "train_loss_bc": 0.03477806597948074, "train_loss_llm": 0.0, "grad_norm": 0.03401344269514084, "global_step": 14686, "epoch": 407, "lr": 0.001} +{"train_loss": 0.03567745484825638, "train_loss_bc": 0.035692594945430756, "train_loss_llm": 0.0, "grad_norm": 0.055003806948661804, "global_step": 14687, "epoch": 407, "lr": 0.001, "val_loss": 0.06790588796138763} +{"train_loss": 0.03131430596113205, "train_loss_bc": 0.03129765763878822, "train_loss_llm": 0.0016648173332214355, "grad_norm": 0.06556741148233414, "global_step": 14688, "epoch": 408, "lr": 0.001} +{"train_loss": 0.035361405462026596, "train_loss_bc": 0.035361405462026596, "train_loss_llm": 0.0, "grad_norm": 0.02563934028148651, "global_step": 14689, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03555676341056824, "train_loss_bc": 0.03555676341056824, "train_loss_llm": 0.0, "grad_norm": 0.029803184792399406, "global_step": 14690, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03731464594602585, "train_loss_bc": 0.03729876130819321, "train_loss_llm": 0.0015884935855865479, "grad_norm": 0.0428786426782608, "global_step": 14691, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03755737468600273, "train_loss_bc": 0.03755737468600273, "train_loss_llm": 0.0, "grad_norm": 0.05429771915078163, "global_step": 14692, "epoch": 408, "lr": 0.001} +{"train_loss": 0.029935011640191078, "train_loss_bc": 0.0299207903444767, "train_loss_llm": 0.0014221668243408203, "grad_norm": 0.02253713458776474, "global_step": 14693, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03491327539086342, "train_loss_bc": 0.03490869700908661, "train_loss_llm": 0.0004576742649078369, "grad_norm": 0.029111864045262337, "global_step": 14694, "epoch": 408, "lr": 0.001} +{"train_loss": 0.0353228934109211, "train_loss_bc": 0.03531277924776077, "train_loss_llm": 0.0010112524032592773, "grad_norm": 0.04299042001366615, "global_step": 14695, "epoch": 408, "lr": 0.001} +{"train_loss": 0.030718671157956123, "train_loss_bc": 0.030715415254235268, "train_loss_llm": 0.0003255307674407959, "grad_norm": 0.049087271094322205, "global_step": 14696, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03656817600131035, "train_loss_bc": 0.03656817600131035, "train_loss_llm": 0.0, "grad_norm": 0.02586797997355461, "global_step": 14697, "epoch": 408, "lr": 0.001} +{"train_loss": 0.033968228846788406, "train_loss_bc": 0.033968228846788406, "train_loss_llm": 0.0, "grad_norm": 0.03704552352428436, "global_step": 14698, "epoch": 408, "lr": 0.001} +{"train_loss": 0.038770850747823715, "train_loss_bc": 0.038770850747823715, "train_loss_llm": 0.0, "grad_norm": 0.0500171072781086, "global_step": 14699, "epoch": 408, "lr": 0.001} +{"train_loss": 0.034340545535087585, "train_loss_bc": 0.034340545535087585, "train_loss_llm": 0.0, "grad_norm": 0.058803535997867584, "global_step": 14700, "epoch": 408, "lr": 0.001} +{"train_loss": 0.0316154882311821, "train_loss_bc": 0.0316154882311821, "train_loss_llm": 0.0, "grad_norm": 0.02510269358754158, "global_step": 14701, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03974790498614311, "train_loss_bc": 0.03974790498614311, "train_loss_llm": 0.0, "grad_norm": 0.03146505355834961, "global_step": 14702, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03682655468583107, "train_loss_bc": 0.03682655468583107, "train_loss_llm": 0.0, "grad_norm": 0.04335983470082283, "global_step": 14703, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03601931035518646, "train_loss_bc": 0.03601931035518646, "train_loss_llm": 0.0, "grad_norm": 0.05294416472315788, "global_step": 14704, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03274155780673027, "train_loss_bc": 0.03274155780673027, "train_loss_llm": 0.0, "grad_norm": 0.026119880378246307, "global_step": 14705, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03687376528978348, "train_loss_bc": 0.03687376528978348, "train_loss_llm": 0.0, "grad_norm": 0.03273225948214531, "global_step": 14706, "epoch": 408, "lr": 0.001} +{"train_loss": 0.0364893302321434, "train_loss_bc": 0.0364893302321434, "train_loss_llm": 0.0, "grad_norm": 0.048577480018138885, "global_step": 14707, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03730272874236107, "train_loss_bc": 0.03730272874236107, "train_loss_llm": 0.0, "grad_norm": 0.05910811200737953, "global_step": 14708, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03510626405477524, "train_loss_bc": 0.03510626405477524, "train_loss_llm": 0.0, "grad_norm": 0.02347981557250023, "global_step": 14709, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03888949751853943, "train_loss_bc": 0.03888949751853943, "train_loss_llm": 0.0, "grad_norm": 0.03738447651267052, "global_step": 14710, "epoch": 408, "lr": 0.001} +{"train_loss": 0.0341445617377758, "train_loss_bc": 0.0341445617377758, "train_loss_llm": 0.0, "grad_norm": 0.04753740876913071, "global_step": 14711, "epoch": 408, "lr": 0.001} +{"train_loss": 0.038189180195331573, "train_loss_bc": 0.038189180195331573, "train_loss_llm": 0.0, "grad_norm": 0.059375256299972534, "global_step": 14712, "epoch": 408, "lr": 0.001} +{"train_loss": 0.031304098665714264, "train_loss_bc": 0.031304098665714264, "train_loss_llm": 0.0, "grad_norm": 0.01899643987417221, "global_step": 14713, "epoch": 408, "lr": 0.001} +{"train_loss": 0.0307930801063776, "train_loss_bc": 0.030780542641878128, "train_loss_llm": 0.001253753900527954, "grad_norm": 0.03807487711310387, "global_step": 14714, "epoch": 408, "lr": 0.001} +{"train_loss": 0.036123745143413544, "train_loss_bc": 0.036123745143413544, "train_loss_llm": 0.0, "grad_norm": 0.048619404435157776, "global_step": 14715, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03364964947104454, "train_loss_bc": 0.03364964947104454, "train_loss_llm": 0.0, "grad_norm": 0.056727372109889984, "global_step": 14716, "epoch": 408, "lr": 0.001} +{"train_loss": 0.036361124366521835, "train_loss_bc": 0.03635833039879799, "train_loss_llm": 0.0002795755863189697, "grad_norm": 0.027924610301852226, "global_step": 14717, "epoch": 408, "lr": 0.001} +{"train_loss": 0.042954158037900925, "train_loss_bc": 0.0429520346224308, "train_loss_llm": 0.0002123415470123291, "grad_norm": 0.0336967408657074, "global_step": 14718, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03985590115189552, "train_loss_bc": 0.03984571620821953, "train_loss_llm": 0.0010184049606323242, "grad_norm": 0.04694876819849014, "global_step": 14719, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03997208923101425, "train_loss_bc": 0.03997208923101425, "train_loss_llm": 0.0, "grad_norm": 0.058061469346284866, "global_step": 14720, "epoch": 408, "lr": 0.001} +{"train_loss": 0.027822095900774002, "train_loss_bc": 0.027822095900774002, "train_loss_llm": 0.0, "grad_norm": 0.020035913214087486, "global_step": 14721, "epoch": 408, "lr": 0.001} +{"train_loss": 0.03726741671562195, "train_loss_bc": 0.03726741671562195, "train_loss_llm": 0.0, "grad_norm": 0.03303667902946472, "global_step": 14722, "epoch": 408, "lr": 0.001} +{"train_loss": 0.035681219616283975, "train_loss_bc": 0.042832255363464355, "train_loss_llm": 0.0, "grad_norm": 0.0528675802052021, "global_step": 14723, "epoch": 408, "lr": 0.001, "val_loss": 0.06889539957046509} +{"train_loss": 0.03882370516657829, "train_loss_bc": 0.03882370516657829, "train_loss_llm": 0.0, "grad_norm": 0.0593738779425621, "global_step": 14724, "epoch": 409, "lr": 0.001} +{"train_loss": 0.039092615246772766, "train_loss_bc": 0.03908631205558777, "train_loss_llm": 0.0006304085254669189, "grad_norm": 0.01762808859348297, "global_step": 14725, "epoch": 409, "lr": 0.001} +{"train_loss": 0.036728981882333755, "train_loss_bc": 0.036728981882333755, "train_loss_llm": 0.0, "grad_norm": 0.0292012020945549, "global_step": 14726, "epoch": 409, "lr": 0.001} +{"train_loss": 0.030179183930158615, "train_loss_bc": 0.030179183930158615, "train_loss_llm": 0.0, "grad_norm": 0.03909597918391228, "global_step": 14727, "epoch": 409, "lr": 0.001} +{"train_loss": 0.033388640731573105, "train_loss_bc": 0.033388640731573105, "train_loss_llm": 0.0, "grad_norm": 0.04423569142818451, "global_step": 14728, "epoch": 409, "lr": 0.001} +{"train_loss": 0.03130294382572174, "train_loss_bc": 0.03130294382572174, "train_loss_llm": 0.0, "grad_norm": 0.01959323324263096, "global_step": 14729, "epoch": 409, "lr": 0.001} +{"train_loss": 0.032584983855485916, "train_loss_bc": 0.032584983855485916, "train_loss_llm": 0.0, "grad_norm": 0.03336012363433838, "global_step": 14730, "epoch": 409, "lr": 0.001} +{"train_loss": 0.04103514179587364, "train_loss_bc": 0.04103514179587364, "train_loss_llm": 0.0, "grad_norm": 0.04513252526521683, "global_step": 14731, "epoch": 409, "lr": 0.001} +{"train_loss": 0.03465135768055916, "train_loss_bc": 0.03465135768055916, "train_loss_llm": 0.0, "grad_norm": 0.05904397740960121, "global_step": 14732, "epoch": 409, "lr": 0.001} +{"train_loss": 0.038311928510665894, "train_loss_bc": 0.038311928510665894, "train_loss_llm": 0.0, "grad_norm": 0.019920876249670982, "global_step": 14733, "epoch": 409, "lr": 0.001} +{"train_loss": 0.03438983112573624, "train_loss_bc": 0.03438983112573624, "train_loss_llm": 0.0, "grad_norm": 0.03247135505080223, "global_step": 14734, "epoch": 409, "lr": 0.001} +{"train_loss": 0.02651570364832878, "train_loss_bc": 0.02647942490875721, "train_loss_llm": 0.0036278069019317627, "grad_norm": 0.04400773346424103, "global_step": 14735, "epoch": 409, "lr": 0.001} +{"train_loss": 0.038366712629795074, "train_loss_bc": 0.03835253044962883, "train_loss_llm": 0.001418381929397583, "grad_norm": 0.05331828445196152, "global_step": 14736, "epoch": 409, "lr": 0.001} +{"train_loss": 0.04051513969898224, "train_loss_bc": 0.04051513969898224, "train_loss_llm": 0.0, "grad_norm": 0.02242940291762352, "global_step": 14737, "epoch": 409, "lr": 0.001} +{"train_loss": 0.040150463581085205, "train_loss_bc": 0.04012357443571091, "train_loss_llm": 0.002688765525817871, "grad_norm": 0.035071033984422684, "global_step": 14738, "epoch": 409, "lr": 0.001} +{"train_loss": 0.036269087344408035, "train_loss_bc": 0.036268092691898346, "train_loss_llm": 9.945034980773926e-05, "grad_norm": 0.044043030589818954, "global_step": 14739, "epoch": 409, "lr": 0.001} +{"train_loss": 0.03226543590426445, "train_loss_bc": 0.03226543590426445, "train_loss_llm": 0.0, "grad_norm": 0.05745719373226166, "global_step": 14740, "epoch": 409, "lr": 0.001} +{"train_loss": 0.03738122060894966, "train_loss_bc": 0.03738122060894966, "train_loss_llm": 0.0, "grad_norm": 0.020810876041650772, "global_step": 14741, "epoch": 409, "lr": 0.001} +{"train_loss": 0.04090503975749016, "train_loss_bc": 0.04090503975749016, "train_loss_llm": 0.0, "grad_norm": 0.034710854291915894, "global_step": 14742, "epoch": 409, "lr": 0.001} +{"train_loss": 0.03851531445980072, "train_loss_bc": 0.03851531445980072, "train_loss_llm": 0.0, "grad_norm": 0.04551272094249725, "global_step": 14743, "epoch": 409, "lr": 0.001} +{"train_loss": 0.0370655283331871, "train_loss_bc": 0.0370655283331871, "train_loss_llm": 0.0, "grad_norm": 0.05728677660226822, "global_step": 14744, "epoch": 409, "lr": 0.001} +{"train_loss": 0.031175317242741585, "train_loss_bc": 0.031175317242741585, "train_loss_llm": 0.0, "grad_norm": 0.024769004434347153, "global_step": 14745, "epoch": 409, "lr": 0.001} +{"train_loss": 0.036451805382966995, "train_loss_bc": 0.036451805382966995, "train_loss_llm": 0.0, "grad_norm": 0.03819772228598595, "global_step": 14746, "epoch": 409, "lr": 0.001} +{"train_loss": 0.04183180630207062, "train_loss_bc": 0.04183180630207062, "train_loss_llm": 0.0, "grad_norm": 0.06267233192920685, "global_step": 14747, "epoch": 409, "lr": 0.001} +{"train_loss": 0.030354587361216545, "train_loss_bc": 0.030354587361216545, "train_loss_llm": 0.0, "grad_norm": 0.0678284764289856, "global_step": 14748, "epoch": 409, "lr": 0.001} +{"train_loss": 0.033743131905794144, "train_loss_bc": 0.033743131905794144, "train_loss_llm": 0.0, "grad_norm": 0.02237965539097786, "global_step": 14749, "epoch": 409, "lr": 0.001} +{"train_loss": 0.03752438351511955, "train_loss_bc": 0.03752438351511955, "train_loss_llm": 0.0, "grad_norm": 0.03405458480119705, "global_step": 14750, "epoch": 409, "lr": 0.001} +{"train_loss": 0.03850763291120529, "train_loss_bc": 0.038503650575876236, "train_loss_llm": 0.00039833784103393555, "grad_norm": 0.04914143681526184, "global_step": 14751, "epoch": 409, "lr": 0.001} +{"train_loss": 0.032439909875392914, "train_loss_bc": 0.032439909875392914, "train_loss_llm": 0.0, "grad_norm": 0.0545646958053112, "global_step": 14752, "epoch": 409, "lr": 0.001} +{"train_loss": 0.030445318669080734, "train_loss_bc": 0.03043726459145546, "train_loss_llm": 0.0008053481578826904, "grad_norm": 0.02179957740008831, "global_step": 14753, "epoch": 409, "lr": 0.001} +{"train_loss": 0.030912864953279495, "train_loss_bc": 0.030912864953279495, "train_loss_llm": 0.0, "grad_norm": 0.035971976816654205, "global_step": 14754, "epoch": 409, "lr": 0.001} +{"train_loss": 0.0347491055727005, "train_loss_bc": 0.0347491055727005, "train_loss_llm": 0.0, "grad_norm": 0.046871695667505264, "global_step": 14755, "epoch": 409, "lr": 0.001} +{"train_loss": 0.028013672679662704, "train_loss_bc": 0.028013672679662704, "train_loss_llm": 0.0, "grad_norm": 0.05736769363284111, "global_step": 14756, "epoch": 409, "lr": 0.001} +{"train_loss": 0.03886015713214874, "train_loss_bc": 0.03886015713214874, "train_loss_llm": 0.0, "grad_norm": 0.024728888645768166, "global_step": 14757, "epoch": 409, "lr": 0.001} +{"train_loss": 0.040887754410505295, "train_loss_bc": 0.040887754410505295, "train_loss_llm": 0.0, "grad_norm": 0.041505225002765656, "global_step": 14758, "epoch": 409, "lr": 0.001} +{"train_loss": 0.0356456491475304, "train_loss_bc": 0.03890696167945862, "train_loss_llm": 0.0, "grad_norm": 0.06124391779303551, "global_step": 14759, "epoch": 409, "lr": 0.001, "val_loss": 0.06857059895992279} +{"train_loss": 0.034028954803943634, "train_loss_bc": 0.034028954803943634, "train_loss_llm": 0.0, "grad_norm": 0.08178789913654327, "global_step": 14760, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03789761662483215, "train_loss_bc": 0.03789761662483215, "train_loss_llm": 0.0, "grad_norm": 0.028139151632785797, "global_step": 14761, "epoch": 410, "lr": 0.001} +{"train_loss": 0.034261446446180344, "train_loss_bc": 0.034261446446180344, "train_loss_llm": 0.0, "grad_norm": 0.04056954383850098, "global_step": 14762, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03222845867276192, "train_loss_bc": 0.03222520276904106, "train_loss_llm": 0.00032573938369750977, "grad_norm": 0.051305606961250305, "global_step": 14763, "epoch": 410, "lr": 0.001} +{"train_loss": 0.031150080263614655, "train_loss_bc": 0.031150080263614655, "train_loss_llm": 0.0, "grad_norm": 0.055085714906454086, "global_step": 14764, "epoch": 410, "lr": 0.001} +{"train_loss": 0.027213526889681816, "train_loss_bc": 0.027213526889681816, "train_loss_llm": 0.0, "grad_norm": 0.02263016626238823, "global_step": 14765, "epoch": 410, "lr": 0.001} +{"train_loss": 0.037038542330265045, "train_loss_bc": 0.037038542330265045, "train_loss_llm": 0.0, "grad_norm": 0.040050871670246124, "global_step": 14766, "epoch": 410, "lr": 0.001} +{"train_loss": 0.038200899958610535, "train_loss_bc": 0.03818143159151077, "train_loss_llm": 0.0019468963146209717, "grad_norm": 0.05151111260056496, "global_step": 14767, "epoch": 410, "lr": 0.001} +{"train_loss": 0.035385746508836746, "train_loss_bc": 0.035385746508836746, "train_loss_llm": 0.0, "grad_norm": 0.06452056020498276, "global_step": 14768, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03141089901328087, "train_loss_bc": 0.03141089901328087, "train_loss_llm": 0.0, "grad_norm": 0.019345860928297043, "global_step": 14769, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03409959375858307, "train_loss_bc": 0.0340895839035511, "train_loss_llm": 0.0010008513927459717, "grad_norm": 0.02979174256324768, "global_step": 14770, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03451627492904663, "train_loss_bc": 0.03451627492904663, "train_loss_llm": 0.0, "grad_norm": 0.04300334304571152, "global_step": 14771, "epoch": 410, "lr": 0.001} +{"train_loss": 0.033724576234817505, "train_loss_bc": 0.033724576234817505, "train_loss_llm": 0.0, "grad_norm": 0.04946715757250786, "global_step": 14772, "epoch": 410, "lr": 0.001} +{"train_loss": 0.037364374846220016, "train_loss_bc": 0.03735565394163132, "train_loss_llm": 0.0008719861507415771, "grad_norm": 0.027512215077877045, "global_step": 14773, "epoch": 410, "lr": 0.001} +{"train_loss": 0.0328662283718586, "train_loss_bc": 0.03284870460629463, "train_loss_llm": 0.0017524659633636475, "grad_norm": 0.04401979595422745, "global_step": 14774, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03989315405488014, "train_loss_bc": 0.03989315405488014, "train_loss_llm": 0.0, "grad_norm": 0.059391338378190994, "global_step": 14775, "epoch": 410, "lr": 0.001} +{"train_loss": 0.0332314595580101, "train_loss_bc": 0.0332314595580101, "train_loss_llm": 0.0, "grad_norm": 0.07486900687217712, "global_step": 14776, "epoch": 410, "lr": 0.001} +{"train_loss": 0.036293741315603256, "train_loss_bc": 0.036293741315603256, "train_loss_llm": 0.0, "grad_norm": 0.02483987994492054, "global_step": 14777, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03531821072101593, "train_loss_bc": 0.03531821072101593, "train_loss_llm": 0.0, "grad_norm": 0.03665649890899658, "global_step": 14778, "epoch": 410, "lr": 0.001} +{"train_loss": 0.042220909148454666, "train_loss_bc": 0.04221530258655548, "train_loss_llm": 0.0005605518817901611, "grad_norm": 0.0566420741379261, "global_step": 14779, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03556881844997406, "train_loss_bc": 0.03556881844997406, "train_loss_llm": 0.0, "grad_norm": 0.07649090141057968, "global_step": 14780, "epoch": 410, "lr": 0.001} +{"train_loss": 0.035612575709819794, "train_loss_bc": 0.035612575709819794, "train_loss_llm": 0.0, "grad_norm": 0.026004595682024956, "global_step": 14781, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03885247930884361, "train_loss_bc": 0.03885247930884361, "train_loss_llm": 0.0, "grad_norm": 0.04150500148534775, "global_step": 14782, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03917091712355614, "train_loss_bc": 0.03917091712355614, "train_loss_llm": 0.0, "grad_norm": 0.054221395403146744, "global_step": 14783, "epoch": 410, "lr": 0.001} +{"train_loss": 0.034247025847435, "train_loss_bc": 0.034247025847435, "train_loss_llm": 0.0, "grad_norm": 0.06461615860462189, "global_step": 14784, "epoch": 410, "lr": 0.001} +{"train_loss": 0.035194508731365204, "train_loss_bc": 0.035194508731365204, "train_loss_llm": 0.0, "grad_norm": 0.020682241767644882, "global_step": 14785, "epoch": 410, "lr": 0.001} +{"train_loss": 0.031874850392341614, "train_loss_bc": 0.0318746343255043, "train_loss_llm": 2.1725893020629883e-05, "grad_norm": 0.02868085540831089, "global_step": 14786, "epoch": 410, "lr": 0.001} +{"train_loss": 0.04289653152227402, "train_loss_bc": 0.04289653152227402, "train_loss_llm": 0.0, "grad_norm": 0.04605071619153023, "global_step": 14787, "epoch": 410, "lr": 0.001} +{"train_loss": 0.038243819028139114, "train_loss_bc": 0.038243819028139114, "train_loss_llm": 0.0, "grad_norm": 0.05515935644507408, "global_step": 14788, "epoch": 410, "lr": 0.001} +{"train_loss": 0.035181932151317596, "train_loss_bc": 0.0351642444729805, "train_loss_llm": 0.0017687082290649414, "grad_norm": 0.024165164679288864, "global_step": 14789, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03921681270003319, "train_loss_bc": 0.03921681270003319, "train_loss_llm": 0.0, "grad_norm": 0.03841279447078705, "global_step": 14790, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03592304140329361, "train_loss_bc": 0.03592304140329361, "train_loss_llm": 0.0, "grad_norm": 0.05314411595463753, "global_step": 14791, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03444420546293259, "train_loss_bc": 0.03444420546293259, "train_loss_llm": 0.0, "grad_norm": 0.06291354447603226, "global_step": 14792, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03811357170343399, "train_loss_bc": 0.03811357170343399, "train_loss_llm": 0.0, "grad_norm": 0.023047897964715958, "global_step": 14793, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03319894149899483, "train_loss_bc": 0.03319894149899483, "train_loss_llm": 0.0, "grad_norm": 0.040808189660310745, "global_step": 14794, "epoch": 410, "lr": 0.001} +{"train_loss": 0.03564173153912028, "train_loss_bc": 0.03701571002602577, "train_loss_llm": 0.0001901686191558838, "grad_norm": 0.06481794267892838, "global_step": 14795, "epoch": 410, "lr": 0.001, "train/cumulative_reward": 108.993974527299, "train/mean_score": 18.835824475516972, "train/success_rate": 0.5, "test/cumulative_reward": 42.378808310975174, "test/mean_score": 7.4412299304461795, "test/success_rate": 0.16, "val_loss": 0.06852396577596664, "train_action_mse_error": 0.03866693004965782} +{"train_loss": 0.03907347097992897, "train_loss_bc": 0.03907347097992897, "train_loss_llm": 0.0, "grad_norm": 0.07749377191066742, "global_step": 14796, "epoch": 411, "lr": 0.001} +{"train_loss": 0.0314888097345829, "train_loss_bc": 0.0314888097345829, "train_loss_llm": 0.0, "grad_norm": 0.025513026863336563, "global_step": 14797, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03689178079366684, "train_loss_bc": 0.03689178079366684, "train_loss_llm": 0.0, "grad_norm": 0.03861772269010544, "global_step": 14798, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03338796645402908, "train_loss_bc": 0.03338796645402908, "train_loss_llm": 0.0, "grad_norm": 0.04698821157217026, "global_step": 14799, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03850902244448662, "train_loss_bc": 0.03850902244448662, "train_loss_llm": 0.0, "grad_norm": 0.055392876267433167, "global_step": 14800, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03294830396771431, "train_loss_bc": 0.03294273838400841, "train_loss_llm": 0.0005566179752349854, "grad_norm": 0.01923154480755329, "global_step": 14801, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03787510469555855, "train_loss_bc": 0.03787510469555855, "train_loss_llm": 0.0, "grad_norm": 0.02872791886329651, "global_step": 14802, "epoch": 411, "lr": 0.001} +{"train_loss": 0.038413409143686295, "train_loss_bc": 0.038413409143686295, "train_loss_llm": 0.0, "grad_norm": 0.04008615016937256, "global_step": 14803, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03528568148612976, "train_loss_bc": 0.03528568148612976, "train_loss_llm": 0.0, "grad_norm": 0.04938338324427605, "global_step": 14804, "epoch": 411, "lr": 0.001} +{"train_loss": 0.033972758799791336, "train_loss_bc": 0.033972758799791336, "train_loss_llm": 0.0, "grad_norm": 0.021944930776953697, "global_step": 14805, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03974136710166931, "train_loss_bc": 0.03974136710166931, "train_loss_llm": 0.0, "grad_norm": 0.036059580743312836, "global_step": 14806, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03648260980844498, "train_loss_bc": 0.03648260980844498, "train_loss_llm": 0.0, "grad_norm": 0.041968934237957, "global_step": 14807, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03194964677095413, "train_loss_bc": 0.03194964677095413, "train_loss_llm": 0.0, "grad_norm": 0.051874060183763504, "global_step": 14808, "epoch": 411, "lr": 0.001} +{"train_loss": 0.042834293097257614, "train_loss_bc": 0.042831432074308395, "train_loss_llm": 0.0002859830856323242, "grad_norm": 0.023586556315422058, "global_step": 14809, "epoch": 411, "lr": 0.001} +{"train_loss": 0.040945444256067276, "train_loss_bc": 0.040945444256067276, "train_loss_llm": 0.0, "grad_norm": 0.03331393003463745, "global_step": 14810, "epoch": 411, "lr": 0.001} +{"train_loss": 0.031829893589019775, "train_loss_bc": 0.031829893589019775, "train_loss_llm": 0.0, "grad_norm": 0.04200489819049835, "global_step": 14811, "epoch": 411, "lr": 0.001} +{"train_loss": 0.035802800208330154, "train_loss_bc": 0.035802800208330154, "train_loss_llm": 0.0, "grad_norm": 0.053403131663799286, "global_step": 14812, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03139275312423706, "train_loss_bc": 0.03138589859008789, "train_loss_llm": 0.0006854832172393799, "grad_norm": 0.02299749106168747, "global_step": 14813, "epoch": 411, "lr": 0.001} +{"train_loss": 0.040407389402389526, "train_loss_bc": 0.040407389402389526, "train_loss_llm": 0.0, "grad_norm": 0.03950653225183487, "global_step": 14814, "epoch": 411, "lr": 0.001} +{"train_loss": 0.029739825055003166, "train_loss_bc": 0.029739825055003166, "train_loss_llm": 0.0, "grad_norm": 0.049251630902290344, "global_step": 14815, "epoch": 411, "lr": 0.001} +{"train_loss": 0.038413580507040024, "train_loss_bc": 0.038413580507040024, "train_loss_llm": 0.0, "grad_norm": 0.05874794349074364, "global_step": 14816, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03337791934609413, "train_loss_bc": 0.033369630575180054, "train_loss_llm": 0.0008287727832794189, "grad_norm": 0.02438238263130188, "global_step": 14817, "epoch": 411, "lr": 0.001} +{"train_loss": 0.031569287180900574, "train_loss_bc": 0.031554628163576126, "train_loss_llm": 0.0014657676219940186, "grad_norm": 0.03903688117861748, "global_step": 14818, "epoch": 411, "lr": 0.001} +{"train_loss": 0.029385503381490707, "train_loss_bc": 0.029385503381490707, "train_loss_llm": 0.0, "grad_norm": 0.04991770163178444, "global_step": 14819, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03807390108704567, "train_loss_bc": 0.03807390108704567, "train_loss_llm": 0.0, "grad_norm": 0.06078152731060982, "global_step": 14820, "epoch": 411, "lr": 0.001} +{"train_loss": 0.036659374833106995, "train_loss_bc": 0.03665456175804138, "train_loss_llm": 0.0004811286926269531, "grad_norm": 0.02205369435250759, "global_step": 14821, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03272562474012375, "train_loss_bc": 0.03272562474012375, "train_loss_llm": 0.0, "grad_norm": 0.03373958542943001, "global_step": 14822, "epoch": 411, "lr": 0.001} +{"train_loss": 0.04118432104587555, "train_loss_bc": 0.0411786250770092, "train_loss_llm": 0.000569760799407959, "grad_norm": 0.05686425045132637, "global_step": 14823, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03403661027550697, "train_loss_bc": 0.03403661027550697, "train_loss_llm": 0.0, "grad_norm": 0.07262726873159409, "global_step": 14824, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03711409866809845, "train_loss_bc": 0.03711409866809845, "train_loss_llm": 0.0, "grad_norm": 0.019307533279061317, "global_step": 14825, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03414154052734375, "train_loss_bc": 0.03413315489888191, "train_loss_llm": 0.0008387267589569092, "grad_norm": 0.03434554487466812, "global_step": 14826, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03873852267861366, "train_loss_bc": 0.03872454911470413, "train_loss_llm": 0.0013973712921142578, "grad_norm": 0.04687449708580971, "global_step": 14827, "epoch": 411, "lr": 0.001} +{"train_loss": 0.0316469669342041, "train_loss_bc": 0.0316469669342041, "train_loss_llm": 0.0, "grad_norm": 0.057050921022892, "global_step": 14828, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03225604444742203, "train_loss_bc": 0.03225604444742203, "train_loss_llm": 0.0, "grad_norm": 0.02151249535381794, "global_step": 14829, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03223182633519173, "train_loss_bc": 0.03221854567527771, "train_loss_llm": 0.0013280212879180908, "grad_norm": 0.030239881947636604, "global_step": 14830, "epoch": 411, "lr": 0.001} +{"train_loss": 0.03562539303675294, "train_loss_bc": 0.041974037885665894, "train_loss_llm": 0.0012658238410949707, "grad_norm": 0.04985180124640465, "global_step": 14831, "epoch": 411, "lr": 0.001, "val_loss": 0.0683199018239975} +{"train_loss": 0.034841593354940414, "train_loss_bc": 0.034841593354940414, "train_loss_llm": 0.0, "grad_norm": 0.05939396098256111, "global_step": 14832, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03430161252617836, "train_loss_bc": 0.03430161252617836, "train_loss_llm": 0.0, "grad_norm": 0.02401501126587391, "global_step": 14833, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03049721196293831, "train_loss_bc": 0.03049721196293831, "train_loss_llm": 0.0, "grad_norm": 0.035015955567359924, "global_step": 14834, "epoch": 412, "lr": 0.001} +{"train_loss": 0.030283505097031593, "train_loss_bc": 0.030283505097031593, "train_loss_llm": 0.0, "grad_norm": 0.05072341114282608, "global_step": 14835, "epoch": 412, "lr": 0.001} +{"train_loss": 0.04350792244076729, "train_loss_bc": 0.043494705110788345, "train_loss_llm": 0.0013217329978942871, "grad_norm": 0.06947319954633713, "global_step": 14836, "epoch": 412, "lr": 0.001} +{"train_loss": 0.033118631690740585, "train_loss_bc": 0.033117037266492844, "train_loss_llm": 0.00015938282012939453, "grad_norm": 0.02505721151828766, "global_step": 14837, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03668159246444702, "train_loss_bc": 0.036677997559309006, "train_loss_llm": 0.00035953521728515625, "grad_norm": 0.03446607664227486, "global_step": 14838, "epoch": 412, "lr": 0.001} +{"train_loss": 0.037550099194049835, "train_loss_bc": 0.037548214197158813, "train_loss_llm": 0.00018846988677978516, "grad_norm": 0.03932030498981476, "global_step": 14839, "epoch": 412, "lr": 0.001} +{"train_loss": 0.032804783433675766, "train_loss_bc": 0.032804783433675766, "train_loss_llm": 0.0, "grad_norm": 0.05446973070502281, "global_step": 14840, "epoch": 412, "lr": 0.001} +{"train_loss": 0.042609743773937225, "train_loss_bc": 0.042609743773937225, "train_loss_llm": 0.0, "grad_norm": 0.024842141196131706, "global_step": 14841, "epoch": 412, "lr": 0.001} +{"train_loss": 0.039210349321365356, "train_loss_bc": 0.039210349321365356, "train_loss_llm": 0.0, "grad_norm": 0.03759821131825447, "global_step": 14842, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03534570708870888, "train_loss_bc": 0.03534570708870888, "train_loss_llm": 0.0, "grad_norm": 0.049905143678188324, "global_step": 14843, "epoch": 412, "lr": 0.001} +{"train_loss": 0.040022749453783035, "train_loss_bc": 0.040022749453783035, "train_loss_llm": 0.0, "grad_norm": 0.062393467873334885, "global_step": 14844, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03887224942445755, "train_loss_bc": 0.0388689860701561, "train_loss_llm": 0.00032633543014526367, "grad_norm": 0.03109198436141014, "global_step": 14845, "epoch": 412, "lr": 0.001} +{"train_loss": 0.036978527903556824, "train_loss_bc": 0.03696814179420471, "train_loss_llm": 0.0010384619235992432, "grad_norm": 0.043309468775987625, "global_step": 14846, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03650355711579323, "train_loss_bc": 0.03650355711579323, "train_loss_llm": 0.0, "grad_norm": 0.046337660402059555, "global_step": 14847, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03779853135347366, "train_loss_bc": 0.03778154030442238, "train_loss_llm": 0.0016991198062896729, "grad_norm": 0.05508371442556381, "global_step": 14848, "epoch": 412, "lr": 0.001} +{"train_loss": 0.038599807769060135, "train_loss_bc": 0.038586441427469254, "train_loss_llm": 0.0013366639614105225, "grad_norm": 0.02237892523407936, "global_step": 14849, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03396538645029068, "train_loss_bc": 0.03396538645029068, "train_loss_llm": 0.0, "grad_norm": 0.030416205525398254, "global_step": 14850, "epoch": 412, "lr": 0.001} +{"train_loss": 0.040805526077747345, "train_loss_bc": 0.04080445319414139, "train_loss_llm": 0.00010713934898376465, "grad_norm": 0.045067623257637024, "global_step": 14851, "epoch": 412, "lr": 0.001} +{"train_loss": 0.031803786754608154, "train_loss_bc": 0.031803786754608154, "train_loss_llm": 0.0, "grad_norm": 0.052143990993499756, "global_step": 14852, "epoch": 412, "lr": 0.001} +{"train_loss": 0.04004138335585594, "train_loss_bc": 0.04004138335585594, "train_loss_llm": 0.0, "grad_norm": 0.024371322244405746, "global_step": 14853, "epoch": 412, "lr": 0.001} +{"train_loss": 0.02396129071712494, "train_loss_bc": 0.02396129071712494, "train_loss_llm": 0.0, "grad_norm": 0.03912955895066261, "global_step": 14854, "epoch": 412, "lr": 0.001} +{"train_loss": 0.025716496631503105, "train_loss_bc": 0.025711406022310257, "train_loss_llm": 0.0005090236663818359, "grad_norm": 0.04450303688645363, "global_step": 14855, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03551092371344566, "train_loss_bc": 0.03551092371344566, "train_loss_llm": 0.0, "grad_norm": 0.06195225566625595, "global_step": 14856, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03251557797193527, "train_loss_bc": 0.03251557797193527, "train_loss_llm": 0.0, "grad_norm": 0.01854248344898224, "global_step": 14857, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03221740573644638, "train_loss_bc": 0.03220483288168907, "train_loss_llm": 0.0012574195861816406, "grad_norm": 0.02823730930685997, "global_step": 14858, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03312017768621445, "train_loss_bc": 0.03312017768621445, "train_loss_llm": 0.0, "grad_norm": 0.039583828300237656, "global_step": 14859, "epoch": 412, "lr": 0.001} +{"train_loss": 0.040784429758787155, "train_loss_bc": 0.040784429758787155, "train_loss_llm": 0.0, "grad_norm": 0.051870495080947876, "global_step": 14860, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03594016656279564, "train_loss_bc": 0.03594016656279564, "train_loss_llm": 0.0, "grad_norm": 0.029024021700024605, "global_step": 14861, "epoch": 412, "lr": 0.001} +{"train_loss": 0.033808015286922455, "train_loss_bc": 0.033808015286922455, "train_loss_llm": 0.0, "grad_norm": 0.04365755617618561, "global_step": 14862, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03278891742229462, "train_loss_bc": 0.03278891742229462, "train_loss_llm": 0.0, "grad_norm": 0.06103809177875519, "global_step": 14863, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03730995953083038, "train_loss_bc": 0.03730995953083038, "train_loss_llm": 0.0, "grad_norm": 0.06641190499067307, "global_step": 14864, "epoch": 412, "lr": 0.001} +{"train_loss": 0.031551651656627655, "train_loss_bc": 0.031551651656627655, "train_loss_llm": 0.0, "grad_norm": 0.02267022617161274, "global_step": 14865, "epoch": 412, "lr": 0.001} +{"train_loss": 0.04073802009224892, "train_loss_bc": 0.04073802009224892, "train_loss_llm": 0.0, "grad_norm": 0.049764905124902725, "global_step": 14866, "epoch": 412, "lr": 0.001} +{"train_loss": 0.03569025360047817, "train_loss_bc": 0.04271634668111801, "train_loss_llm": 0.0025492310523986816, "grad_norm": 0.0754183679819107, "global_step": 14867, "epoch": 412, "lr": 0.001, "val_loss": 0.06919699162244797} +{"train_loss": 0.034829311072826385, "train_loss_bc": 0.034829311072826385, "train_loss_llm": 0.0, "grad_norm": 0.09075966477394104, "global_step": 14868, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03562371805310249, "train_loss_bc": 0.03561731427907944, "train_loss_llm": 0.00064048171043396, "grad_norm": 0.02519102208316326, "global_step": 14869, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03803715854883194, "train_loss_bc": 0.03803715854883194, "train_loss_llm": 0.0, "grad_norm": 0.04373438283801079, "global_step": 14870, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03522617742419243, "train_loss_bc": 0.03520837053656578, "train_loss_llm": 0.0017807483673095703, "grad_norm": 0.05300610885024071, "global_step": 14871, "epoch": 413, "lr": 0.001} +{"train_loss": 0.036066509783267975, "train_loss_bc": 0.036066509783267975, "train_loss_llm": 0.0, "grad_norm": 0.06396077573299408, "global_step": 14872, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03640243411064148, "train_loss_bc": 0.036393433809280396, "train_loss_llm": 0.000899970531463623, "grad_norm": 0.021394643932580948, "global_step": 14873, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03622707724571228, "train_loss_bc": 0.03622707724571228, "train_loss_llm": 0.0, "grad_norm": 0.03252030164003372, "global_step": 14874, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03479276970028877, "train_loss_bc": 0.03479276970028877, "train_loss_llm": 0.0, "grad_norm": 0.04015108942985535, "global_step": 14875, "epoch": 413, "lr": 0.001} +{"train_loss": 0.0339357852935791, "train_loss_bc": 0.0339357852935791, "train_loss_llm": 0.0, "grad_norm": 0.04534289240837097, "global_step": 14876, "epoch": 413, "lr": 0.001} +{"train_loss": 0.04060596227645874, "train_loss_bc": 0.04060596227645874, "train_loss_llm": 0.0, "grad_norm": 0.031728316098451614, "global_step": 14877, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03858254849910736, "train_loss_bc": 0.03858254849910736, "train_loss_llm": 0.0, "grad_norm": 0.04961370676755905, "global_step": 14878, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03993305191397667, "train_loss_bc": 0.039903976023197174, "train_loss_llm": 0.0029074251651763916, "grad_norm": 0.0683555155992508, "global_step": 14879, "epoch": 413, "lr": 0.001} +{"train_loss": 0.04129607602953911, "train_loss_bc": 0.04129607602953911, "train_loss_llm": 0.0, "grad_norm": 0.09962837398052216, "global_step": 14880, "epoch": 413, "lr": 0.001} +{"train_loss": 0.039030659943819046, "train_loss_bc": 0.03902691230177879, "train_loss_llm": 0.00037485361099243164, "grad_norm": 0.0356040857732296, "global_step": 14881, "epoch": 413, "lr": 0.001} +{"train_loss": 0.0386325977742672, "train_loss_bc": 0.03862490877509117, "train_loss_llm": 0.0007687807083129883, "grad_norm": 0.05516067147254944, "global_step": 14882, "epoch": 413, "lr": 0.001} +{"train_loss": 0.033471062779426575, "train_loss_bc": 0.033471062779426575, "train_loss_llm": 0.0, "grad_norm": 0.07517125457525253, "global_step": 14883, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03183674439787865, "train_loss_bc": 0.03180450201034546, "train_loss_llm": 0.003224194049835205, "grad_norm": 0.10000166296958923, "global_step": 14884, "epoch": 413, "lr": 0.001} +{"train_loss": 0.037274353206157684, "train_loss_bc": 0.037274353206157684, "train_loss_llm": 0.0, "grad_norm": 0.036337096244096756, "global_step": 14885, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03306092321872711, "train_loss_bc": 0.0330524705350399, "train_loss_llm": 0.0008451342582702637, "grad_norm": 0.06301115453243256, "global_step": 14886, "epoch": 413, "lr": 0.001} +{"train_loss": 0.031223472207784653, "train_loss_bc": 0.031223472207784653, "train_loss_llm": 0.0, "grad_norm": 0.08909819275140762, "global_step": 14887, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03873085603117943, "train_loss_bc": 0.03873085603117943, "train_loss_llm": 0.0, "grad_norm": 0.10509148240089417, "global_step": 14888, "epoch": 413, "lr": 0.001} +{"train_loss": 0.036403145641088486, "train_loss_bc": 0.036403145641088486, "train_loss_llm": 0.0, "grad_norm": 0.02493203990161419, "global_step": 14889, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03455054759979248, "train_loss_bc": 0.03455054759979248, "train_loss_llm": 0.0, "grad_norm": 0.041285835206508636, "global_step": 14890, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03409159928560257, "train_loss_bc": 0.03409159928560257, "train_loss_llm": 0.0, "grad_norm": 0.05761827528476715, "global_step": 14891, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03775083273649216, "train_loss_bc": 0.03774498403072357, "train_loss_llm": 0.0005849301815032959, "grad_norm": 0.07329531759023666, "global_step": 14892, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03859908878803253, "train_loss_bc": 0.03859908878803253, "train_loss_llm": 0.0, "grad_norm": 0.023246636614203453, "global_step": 14893, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03427652269601822, "train_loss_bc": 0.03427652269601822, "train_loss_llm": 0.0, "grad_norm": 0.027674544602632523, "global_step": 14894, "epoch": 413, "lr": 0.001} +{"train_loss": 0.029127996414899826, "train_loss_bc": 0.029127996414899826, "train_loss_llm": 0.0, "grad_norm": 0.03637443855404854, "global_step": 14895, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03881247341632843, "train_loss_bc": 0.03881247341632843, "train_loss_llm": 0.0, "grad_norm": 0.04635195806622505, "global_step": 14896, "epoch": 413, "lr": 0.001} +{"train_loss": 0.034771598875522614, "train_loss_bc": 0.034768637269735336, "train_loss_llm": 0.0002962946891784668, "grad_norm": 0.02310999482870102, "global_step": 14897, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03114737756550312, "train_loss_bc": 0.03114737756550312, "train_loss_llm": 0.0, "grad_norm": 0.043695997446775436, "global_step": 14898, "epoch": 413, "lr": 0.001} +{"train_loss": 0.031791411340236664, "train_loss_bc": 0.03178662806749344, "train_loss_llm": 0.00047838687896728516, "grad_norm": 0.05589086934924126, "global_step": 14899, "epoch": 413, "lr": 0.001} +{"train_loss": 0.034667517989873886, "train_loss_bc": 0.034667517989873886, "train_loss_llm": 0.0, "grad_norm": 0.07216966152191162, "global_step": 14900, "epoch": 413, "lr": 0.001} +{"train_loss": 0.04342206194996834, "train_loss_bc": 0.04342206194996834, "train_loss_llm": 0.0, "grad_norm": 0.03466539457440376, "global_step": 14901, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03400041535496712, "train_loss_bc": 0.03400041535496712, "train_loss_llm": 0.0, "grad_norm": 0.06053405627608299, "global_step": 14902, "epoch": 413, "lr": 0.001} +{"train_loss": 0.03600223274487588, "train_loss_bc": 0.03782025724649429, "train_loss_llm": 0.0028284192085266113, "grad_norm": 0.07816728204488754, "global_step": 14903, "epoch": 413, "lr": 0.001, "val_loss": 0.06966271251440048} +{"train_loss": 0.035006601363420486, "train_loss_bc": 0.035006601363420486, "train_loss_llm": 0.0, "grad_norm": 0.09757348150014877, "global_step": 14904, "epoch": 414, "lr": 0.001} +{"train_loss": 0.031177915632724762, "train_loss_bc": 0.031177915632724762, "train_loss_llm": 0.0, "grad_norm": 0.02719728648662567, "global_step": 14905, "epoch": 414, "lr": 0.001} +{"train_loss": 0.041070181876420975, "train_loss_bc": 0.041070181876420975, "train_loss_llm": 0.0, "grad_norm": 0.04885771498084068, "global_step": 14906, "epoch": 414, "lr": 0.001} +{"train_loss": 0.038054920732975006, "train_loss_bc": 0.038044560700654984, "train_loss_llm": 0.001035928726196289, "grad_norm": 0.07104763388633728, "global_step": 14907, "epoch": 414, "lr": 0.001} +{"train_loss": 0.030764035880565643, "train_loss_bc": 0.030764035880565643, "train_loss_llm": 0.0, "grad_norm": 0.09265828132629395, "global_step": 14908, "epoch": 414, "lr": 0.001} +{"train_loss": 0.030196567997336388, "train_loss_bc": 0.03019070439040661, "train_loss_llm": 0.0005863606929779053, "grad_norm": 0.026690904051065445, "global_step": 14909, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03849215805530548, "train_loss_bc": 0.03849215805530548, "train_loss_llm": 0.0, "grad_norm": 0.04641803726553917, "global_step": 14910, "epoch": 414, "lr": 0.001} +{"train_loss": 0.030224263668060303, "train_loss_bc": 0.030224263668060303, "train_loss_llm": 0.0, "grad_norm": 0.06413595378398895, "global_step": 14911, "epoch": 414, "lr": 0.001} +{"train_loss": 0.037309907376766205, "train_loss_bc": 0.037309907376766205, "train_loss_llm": 0.0, "grad_norm": 0.07706272602081299, "global_step": 14912, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03382134065032005, "train_loss_bc": 0.033803123980760574, "train_loss_llm": 0.0018217265605926514, "grad_norm": 0.026733770966529846, "global_step": 14913, "epoch": 414, "lr": 0.001} +{"train_loss": 0.02859445847570896, "train_loss_bc": 0.02859445847570896, "train_loss_llm": 0.0, "grad_norm": 0.035895220935344696, "global_step": 14914, "epoch": 414, "lr": 0.001} +{"train_loss": 0.04033480957150459, "train_loss_bc": 0.04033480957150459, "train_loss_llm": 0.0, "grad_norm": 0.052541255950927734, "global_step": 14915, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03396367281675339, "train_loss_bc": 0.03396367281675339, "train_loss_llm": 0.0, "grad_norm": 0.06420621275901794, "global_step": 14916, "epoch": 414, "lr": 0.001} +{"train_loss": 0.031781043857336044, "train_loss_bc": 0.03177089989185333, "train_loss_llm": 0.0010142624378204346, "grad_norm": 0.020372461527585983, "global_step": 14917, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03984852135181427, "train_loss_bc": 0.03984852135181427, "train_loss_llm": 0.0, "grad_norm": 0.03577035665512085, "global_step": 14918, "epoch": 414, "lr": 0.001} +{"train_loss": 0.042507655918598175, "train_loss_bc": 0.042507655918598175, "train_loss_llm": 0.0, "grad_norm": 0.04488493502140045, "global_step": 14919, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03162079676985741, "train_loss_bc": 0.03161875158548355, "train_loss_llm": 0.00020462274551391602, "grad_norm": 0.046076178550720215, "global_step": 14920, "epoch": 414, "lr": 0.001} +{"train_loss": 0.036491189152002335, "train_loss_bc": 0.036491189152002335, "train_loss_llm": 0.0, "grad_norm": 0.036236055195331573, "global_step": 14921, "epoch": 414, "lr": 0.001} +{"train_loss": 0.04127349704504013, "train_loss_bc": 0.041270751506090164, "train_loss_llm": 0.0002745389938354492, "grad_norm": 0.050922371447086334, "global_step": 14922, "epoch": 414, "lr": 0.001} +{"train_loss": 0.027739400044083595, "train_loss_bc": 0.027739109471440315, "train_loss_llm": 2.899765968322754e-05, "grad_norm": 0.06164199858903885, "global_step": 14923, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03922561928629875, "train_loss_bc": 0.03922557085752487, "train_loss_llm": 4.798173904418945e-06, "grad_norm": 0.07860416173934937, "global_step": 14924, "epoch": 414, "lr": 0.001} +{"train_loss": 0.0375417061150074, "train_loss_bc": 0.03751014545559883, "train_loss_llm": 0.0031559765338897705, "grad_norm": 0.03555884212255478, "global_step": 14925, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03850243613123894, "train_loss_bc": 0.03850243613123894, "train_loss_llm": 0.0, "grad_norm": 0.053267233073711395, "global_step": 14926, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03729568049311638, "train_loss_bc": 0.03728829696774483, "train_loss_llm": 0.0007381737232208252, "grad_norm": 0.07287487387657166, "global_step": 14927, "epoch": 414, "lr": 0.001} +{"train_loss": 0.036986079066991806, "train_loss_bc": 0.036986079066991806, "train_loss_llm": 0.0, "grad_norm": 0.08850449323654175, "global_step": 14928, "epoch": 414, "lr": 0.001} +{"train_loss": 0.044629115611314774, "train_loss_bc": 0.044629115611314774, "train_loss_llm": 0.0, "grad_norm": 0.03774561360478401, "global_step": 14929, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03839677944779396, "train_loss_bc": 0.03838782757520676, "train_loss_llm": 0.0008952915668487549, "grad_norm": 0.05886209383606911, "global_step": 14930, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03580617159605026, "train_loss_bc": 0.03580617159605026, "train_loss_llm": 0.0, "grad_norm": 0.07800310105085373, "global_step": 14931, "epoch": 414, "lr": 0.001} +{"train_loss": 0.033916063606739044, "train_loss_bc": 0.03389483317732811, "train_loss_llm": 0.002123117446899414, "grad_norm": 0.08948645740747452, "global_step": 14932, "epoch": 414, "lr": 0.001} +{"train_loss": 0.035597965121269226, "train_loss_bc": 0.035585809499025345, "train_loss_llm": 0.0012156665325164795, "grad_norm": 0.028734538704156876, "global_step": 14933, "epoch": 414, "lr": 0.001} +{"train_loss": 0.040516577661037445, "train_loss_bc": 0.040512748062610626, "train_loss_llm": 0.00038307905197143555, "grad_norm": 0.049735408276319504, "global_step": 14934, "epoch": 414, "lr": 0.001} +{"train_loss": 0.031017431989312172, "train_loss_bc": 0.031017431989312172, "train_loss_llm": 0.0, "grad_norm": 0.06192491203546524, "global_step": 14935, "epoch": 414, "lr": 0.001} +{"train_loss": 0.035506296902894974, "train_loss_bc": 0.03550105541944504, "train_loss_llm": 0.0005241632461547852, "grad_norm": 0.08006536960601807, "global_step": 14936, "epoch": 414, "lr": 0.001} +{"train_loss": 0.033997464925050735, "train_loss_bc": 0.033997464925050735, "train_loss_llm": 0.0, "grad_norm": 0.02817855402827263, "global_step": 14937, "epoch": 414, "lr": 0.001} +{"train_loss": 0.043729886412620544, "train_loss_bc": 0.043729886412620544, "train_loss_llm": 0.0, "grad_norm": 0.05109087750315666, "global_step": 14938, "epoch": 414, "lr": 0.001} +{"train_loss": 0.03601217445813947, "train_loss_bc": 0.0335000678896904, "train_loss_llm": 0.0, "grad_norm": 0.06550359725952148, "global_step": 14939, "epoch": 414, "lr": 0.001, "val_loss": 0.06953354179859161} +{"train_loss": 0.02969597466289997, "train_loss_bc": 0.029691534116864204, "train_loss_llm": 0.00044414401054382324, "grad_norm": 0.07902488857507706, "global_step": 14940, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03757307305932045, "train_loss_bc": 0.03757139667868614, "train_loss_llm": 0.00016772747039794922, "grad_norm": 0.024923449382185936, "global_step": 14941, "epoch": 415, "lr": 0.001} +{"train_loss": 0.04028908163309097, "train_loss_bc": 0.04028908163309097, "train_loss_llm": 0.0, "grad_norm": 0.03728768974542618, "global_step": 14942, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03268083930015564, "train_loss_bc": 0.03268083930015564, "train_loss_llm": 0.0, "grad_norm": 0.05210596323013306, "global_step": 14943, "epoch": 415, "lr": 0.001} +{"train_loss": 0.027324920520186424, "train_loss_bc": 0.027322957292199135, "train_loss_llm": 0.00019624829292297363, "grad_norm": 0.05317862704396248, "global_step": 14944, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03806213289499283, "train_loss_bc": 0.03806213289499283, "train_loss_llm": 0.0, "grad_norm": 0.029754461720585823, "global_step": 14945, "epoch": 415, "lr": 0.001} +{"train_loss": 0.034593597054481506, "train_loss_bc": 0.034593597054481506, "train_loss_llm": 0.0, "grad_norm": 0.049831219017505646, "global_step": 14946, "epoch": 415, "lr": 0.001} +{"train_loss": 0.033159274607896805, "train_loss_bc": 0.033159274607896805, "train_loss_llm": 0.0, "grad_norm": 0.07204874604940414, "global_step": 14947, "epoch": 415, "lr": 0.001} +{"train_loss": 0.031454168260097504, "train_loss_bc": 0.031454168260097504, "train_loss_llm": 0.0, "grad_norm": 0.09160921722650528, "global_step": 14948, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03653672710061073, "train_loss_bc": 0.03653672710061073, "train_loss_llm": 0.0, "grad_norm": 0.022751105949282646, "global_step": 14949, "epoch": 415, "lr": 0.001} +{"train_loss": 0.04225817695260048, "train_loss_bc": 0.04225817695260048, "train_loss_llm": 0.0, "grad_norm": 0.05034841597080231, "global_step": 14950, "epoch": 415, "lr": 0.001} +{"train_loss": 0.04060019180178642, "train_loss_bc": 0.04058747738599777, "train_loss_llm": 0.0012714564800262451, "grad_norm": 0.06770909577608109, "global_step": 14951, "epoch": 415, "lr": 0.001} +{"train_loss": 0.037913817912340164, "train_loss_bc": 0.037913817912340164, "train_loss_llm": 0.0, "grad_norm": 0.08793666958808899, "global_step": 14952, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03281175345182419, "train_loss_bc": 0.03281175345182419, "train_loss_llm": 0.0, "grad_norm": 0.023531870916485786, "global_step": 14953, "epoch": 415, "lr": 0.001} +{"train_loss": 0.029295941814780235, "train_loss_bc": 0.029295941814780235, "train_loss_llm": 0.0, "grad_norm": 0.03949814662337303, "global_step": 14954, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03612891957163811, "train_loss_bc": 0.03612891957163811, "train_loss_llm": 0.0, "grad_norm": 0.05025976151227951, "global_step": 14955, "epoch": 415, "lr": 0.001} +{"train_loss": 0.038944095373153687, "train_loss_bc": 0.038944095373153687, "train_loss_llm": 0.0, "grad_norm": 0.06981861591339111, "global_step": 14956, "epoch": 415, "lr": 0.001} +{"train_loss": 0.038793712854385376, "train_loss_bc": 0.038793712854385376, "train_loss_llm": 0.0, "grad_norm": 0.025378337129950523, "global_step": 14957, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03949576988816261, "train_loss_bc": 0.03949576988816261, "train_loss_llm": 0.0, "grad_norm": 0.04248151555657387, "global_step": 14958, "epoch": 415, "lr": 0.001} +{"train_loss": 0.044492121785879135, "train_loss_bc": 0.04449198395013809, "train_loss_llm": 1.3649463653564453e-05, "grad_norm": 0.05945207551121712, "global_step": 14959, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03325297310948372, "train_loss_bc": 0.03325297310948372, "train_loss_llm": 0.0, "grad_norm": 0.07552920281887054, "global_step": 14960, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03737223520874977, "train_loss_bc": 0.037350576370954514, "train_loss_llm": 0.0021658241748809814, "grad_norm": 0.026197154074907303, "global_step": 14961, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03398267552256584, "train_loss_bc": 0.03398267552256584, "train_loss_llm": 0.0, "grad_norm": 0.04304085671901703, "global_step": 14962, "epoch": 415, "lr": 0.001} +{"train_loss": 0.0358758270740509, "train_loss_bc": 0.03586716204881668, "train_loss_llm": 0.0008665919303894043, "grad_norm": 0.060937657952308655, "global_step": 14963, "epoch": 415, "lr": 0.001} +{"train_loss": 0.0366240069270134, "train_loss_bc": 0.0366240069270134, "train_loss_llm": 0.0, "grad_norm": 0.08026587218046188, "global_step": 14964, "epoch": 415, "lr": 0.001} +{"train_loss": 0.04076395556330681, "train_loss_bc": 0.0407574325799942, "train_loss_llm": 0.0006523728370666504, "grad_norm": 0.025816315785050392, "global_step": 14965, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03531656041741371, "train_loss_bc": 0.03531656041741371, "train_loss_llm": 0.0, "grad_norm": 0.03619668632745743, "global_step": 14966, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03680090606212616, "train_loss_bc": 0.03679483011364937, "train_loss_llm": 0.0006075203418731689, "grad_norm": 0.04799273982644081, "global_step": 14967, "epoch": 415, "lr": 0.001} +{"train_loss": 0.032119814306497574, "train_loss_bc": 0.03211598098278046, "train_loss_llm": 0.0003832876682281494, "grad_norm": 0.06121918186545372, "global_step": 14968, "epoch": 415, "lr": 0.001} +{"train_loss": 0.029920103028416634, "train_loss_bc": 0.029920103028416634, "train_loss_llm": 0.0, "grad_norm": 0.026643263176083565, "global_step": 14969, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03414568305015564, "train_loss_bc": 0.03414568305015564, "train_loss_llm": 0.0, "grad_norm": 0.03630377724766731, "global_step": 14970, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03721056878566742, "train_loss_bc": 0.03721056878566742, "train_loss_llm": 0.0, "grad_norm": 0.04471064731478691, "global_step": 14971, "epoch": 415, "lr": 0.001} +{"train_loss": 0.0342986099421978, "train_loss_bc": 0.03427856042981148, "train_loss_llm": 0.002004951238632202, "grad_norm": 0.06232709437608719, "global_step": 14972, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03787470981478691, "train_loss_bc": 0.03787470981478691, "train_loss_llm": 0.0, "grad_norm": 0.025858163833618164, "global_step": 14973, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03624966740608215, "train_loss_bc": 0.03624966740608215, "train_loss_llm": 0.0, "grad_norm": 0.040732260793447495, "global_step": 14974, "epoch": 415, "lr": 0.001} +{"train_loss": 0.03580848013775216, "train_loss_bc": 0.03519269824028015, "train_loss_llm": 0.0, "grad_norm": 0.06687121093273163, "global_step": 14975, "epoch": 415, "lr": 0.001, "train/cumulative_reward": 47.9905692721038, "train/mean_score": 10.61298054727418, "train/success_rate": 0.3, "test/cumulative_reward": 22.847199984346513, "test/mean_score": 3.994067532102022, "test/success_rate": 0.08, "val_loss": 0.06940567493438721, "train_action_mse_error": 0.038905732333660126} +{"train_loss": 0.02834038995206356, "train_loss_bc": 0.028332160785794258, "train_loss_llm": 0.0008228719234466553, "grad_norm": 0.08649332821369171, "global_step": 14976, "epoch": 416, "lr": 0.001} +{"train_loss": 0.033424388617277145, "train_loss_bc": 0.033424388617277145, "train_loss_llm": 0.0, "grad_norm": 0.024743901565670967, "global_step": 14977, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03547244146466255, "train_loss_bc": 0.03547244146466255, "train_loss_llm": 0.0, "grad_norm": 0.03924986347556114, "global_step": 14978, "epoch": 416, "lr": 0.001} +{"train_loss": 0.037625476717948914, "train_loss_bc": 0.037625476717948914, "train_loss_llm": 0.0, "grad_norm": 0.060155581682920456, "global_step": 14979, "epoch": 416, "lr": 0.001} +{"train_loss": 0.035655803978443146, "train_loss_bc": 0.035655803978443146, "train_loss_llm": 0.0, "grad_norm": 0.08647412806749344, "global_step": 14980, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03900705277919769, "train_loss_bc": 0.03899388760328293, "train_loss_llm": 0.0013165473937988281, "grad_norm": 0.024903232231736183, "global_step": 14981, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03552872687578201, "train_loss_bc": 0.0355067253112793, "train_loss_llm": 0.0021999776363372803, "grad_norm": 0.04634639993309975, "global_step": 14982, "epoch": 416, "lr": 0.001} +{"train_loss": 0.036472566425800323, "train_loss_bc": 0.036472566425800323, "train_loss_llm": 0.0, "grad_norm": 0.06405037641525269, "global_step": 14983, "epoch": 416, "lr": 0.001} +{"train_loss": 0.037376519292593, "train_loss_bc": 0.037376519292593, "train_loss_llm": 0.0, "grad_norm": 0.08202573657035828, "global_step": 14984, "epoch": 416, "lr": 0.001} +{"train_loss": 0.04086122289299965, "train_loss_bc": 0.04086097329854965, "train_loss_llm": 2.47955322265625e-05, "grad_norm": 0.02647489123046398, "global_step": 14985, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03324689716100693, "train_loss_bc": 0.03323516622185707, "train_loss_llm": 0.0011729896068572998, "grad_norm": 0.0371033176779747, "global_step": 14986, "epoch": 416, "lr": 0.001} +{"train_loss": 0.034022439271211624, "train_loss_bc": 0.034022439271211624, "train_loss_llm": 0.0, "grad_norm": 0.04861341789364815, "global_step": 14987, "epoch": 416, "lr": 0.001} +{"train_loss": 0.038452282547950745, "train_loss_bc": 0.03843272104859352, "train_loss_llm": 0.0019559860229492188, "grad_norm": 0.06163828447461128, "global_step": 14988, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03265688940882683, "train_loss_bc": 0.03265688940882683, "train_loss_llm": 0.0, "grad_norm": 0.018004577606916428, "global_step": 14989, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03289216384291649, "train_loss_bc": 0.03288339078426361, "train_loss_llm": 0.0008774697780609131, "grad_norm": 0.03303001448512077, "global_step": 14990, "epoch": 416, "lr": 0.001} +{"train_loss": 0.0373840406537056, "train_loss_bc": 0.0373840406537056, "train_loss_llm": 0.0, "grad_norm": 0.046999458223581314, "global_step": 14991, "epoch": 416, "lr": 0.001} +{"train_loss": 0.034778889268636703, "train_loss_bc": 0.034778889268636703, "train_loss_llm": 0.0, "grad_norm": 0.05890282243490219, "global_step": 14992, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03668573871254921, "train_loss_bc": 0.03668573871254921, "train_loss_llm": 0.0, "grad_norm": 0.022801371291279793, "global_step": 14993, "epoch": 416, "lr": 0.001} +{"train_loss": 0.033129528164863586, "train_loss_bc": 0.033129528164863586, "train_loss_llm": 0.0, "grad_norm": 0.04062648490071297, "global_step": 14994, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03271283209323883, "train_loss_bc": 0.03271283209323883, "train_loss_llm": 0.0, "grad_norm": 0.05785835161805153, "global_step": 14995, "epoch": 416, "lr": 0.001} +{"train_loss": 0.040093500167131424, "train_loss_bc": 0.040093500167131424, "train_loss_llm": 0.0, "grad_norm": 0.07648751139640808, "global_step": 14996, "epoch": 416, "lr": 0.001} +{"train_loss": 0.04027179256081581, "train_loss_bc": 0.04027179256081581, "train_loss_llm": 0.0, "grad_norm": 0.03477304428815842, "global_step": 14997, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03839309513568878, "train_loss_bc": 0.03838429972529411, "train_loss_llm": 0.0008795559406280518, "grad_norm": 0.050670377910137177, "global_step": 14998, "epoch": 416, "lr": 0.001} +{"train_loss": 0.0424557700753212, "train_loss_bc": 0.0424557700753212, "train_loss_llm": 0.0, "grad_norm": 0.07524038851261139, "global_step": 14999, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03810279816389084, "train_loss_bc": 0.03810279816389084, "train_loss_llm": 0.0, "grad_norm": 0.0961296334862709, "global_step": 15000, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03661948814988136, "train_loss_bc": 0.03661948814988136, "train_loss_llm": 0.0, "grad_norm": 0.029834378510713577, "global_step": 15001, "epoch": 416, "lr": 0.001} +{"train_loss": 0.034068211913108826, "train_loss_bc": 0.0340651273727417, "train_loss_llm": 0.0003082752227783203, "grad_norm": 0.051169876009225845, "global_step": 15002, "epoch": 416, "lr": 0.001} +{"train_loss": 0.0331595316529274, "train_loss_bc": 0.03314859792590141, "train_loss_llm": 0.0010932683944702148, "grad_norm": 0.08080972731113434, "global_step": 15003, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03563084825873375, "train_loss_bc": 0.03563084825873375, "train_loss_llm": 0.0, "grad_norm": 0.09596433490514755, "global_step": 15004, "epoch": 416, "lr": 0.001} +{"train_loss": 0.036312371492385864, "train_loss_bc": 0.036312371492385864, "train_loss_llm": 0.0, "grad_norm": 0.02572309598326683, "global_step": 15005, "epoch": 416, "lr": 0.001} +{"train_loss": 0.032292697578668594, "train_loss_bc": 0.032292697578668594, "train_loss_llm": 0.0, "grad_norm": 0.03997775539755821, "global_step": 15006, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03055032715201378, "train_loss_bc": 0.03054632619023323, "train_loss_llm": 0.00040009617805480957, "grad_norm": 0.06500184535980225, "global_step": 15007, "epoch": 416, "lr": 0.001} +{"train_loss": 0.04022560268640518, "train_loss_bc": 0.04022560268640518, "train_loss_llm": 0.0, "grad_norm": 0.09262983500957489, "global_step": 15008, "epoch": 416, "lr": 0.001} +{"train_loss": 0.037559110671281815, "train_loss_bc": 0.037559110671281815, "train_loss_llm": 0.0, "grad_norm": 0.03015129640698433, "global_step": 15009, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03687150403857231, "train_loss_bc": 0.03686799481511116, "train_loss_llm": 0.0003509521484375, "grad_norm": 0.057385608553886414, "global_step": 15010, "epoch": 416, "lr": 0.001} +{"train_loss": 0.03609683938945333, "train_loss_bc": 0.04114479944109917, "train_loss_llm": 0.0008480250835418701, "grad_norm": 0.07207514345645905, "global_step": 15011, "epoch": 416, "lr": 0.001, "val_loss": 0.06913547217845917} +{"train_loss": 0.041407521814107895, "train_loss_bc": 0.04140675440430641, "train_loss_llm": 7.677078247070312e-05, "grad_norm": 0.09693106263875961, "global_step": 15012, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03659866750240326, "train_loss_bc": 0.03659866750240326, "train_loss_llm": 0.0, "grad_norm": 0.024770153686404228, "global_step": 15013, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03360796347260475, "train_loss_bc": 0.03359931334853172, "train_loss_llm": 0.0008650422096252441, "grad_norm": 0.04562891647219658, "global_step": 15014, "epoch": 417, "lr": 0.001} +{"train_loss": 0.031762655824422836, "train_loss_bc": 0.031762655824422836, "train_loss_llm": 0.0, "grad_norm": 0.05243093520402908, "global_step": 15015, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03504533693194389, "train_loss_bc": 0.03504355996847153, "train_loss_llm": 0.00017768144607543945, "grad_norm": 0.07411234825849533, "global_step": 15016, "epoch": 417, "lr": 0.001} +{"train_loss": 0.035743165761232376, "train_loss_bc": 0.035743165761232376, "train_loss_llm": 0.0, "grad_norm": 0.021391451358795166, "global_step": 15017, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03412453830242157, "train_loss_bc": 0.03412453830242157, "train_loss_llm": 0.0, "grad_norm": 0.030884910374879837, "global_step": 15018, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03532339259982109, "train_loss_bc": 0.03532339259982109, "train_loss_llm": 0.0, "grad_norm": 0.057360921055078506, "global_step": 15019, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03931160271167755, "train_loss_bc": 0.03931160271167755, "train_loss_llm": 0.0, "grad_norm": 0.07703634351491928, "global_step": 15020, "epoch": 417, "lr": 0.001} +{"train_loss": 0.0327359139919281, "train_loss_bc": 0.0327359139919281, "train_loss_llm": 0.0, "grad_norm": 0.025766484439373016, "global_step": 15021, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03207717463374138, "train_loss_bc": 0.03207717463374138, "train_loss_llm": 0.0, "grad_norm": 0.04503069445490837, "global_step": 15022, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03139842301607132, "train_loss_bc": 0.03138958662748337, "train_loss_llm": 0.0008835196495056152, "grad_norm": 0.060651745647192, "global_step": 15023, "epoch": 417, "lr": 0.001} +{"train_loss": 0.025874735787510872, "train_loss_bc": 0.025874735787510872, "train_loss_llm": 0.0, "grad_norm": 0.08155486732721329, "global_step": 15024, "epoch": 417, "lr": 0.001} +{"train_loss": 0.036469172686338425, "train_loss_bc": 0.036469172686338425, "train_loss_llm": 0.0, "grad_norm": 0.03727192059159279, "global_step": 15025, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03771410137414932, "train_loss_bc": 0.03771410137414932, "train_loss_llm": 0.0, "grad_norm": 0.05611900985240936, "global_step": 15026, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03803675249218941, "train_loss_bc": 0.03803675249218941, "train_loss_llm": 0.0, "grad_norm": 0.06550554186105728, "global_step": 15027, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03845445439219475, "train_loss_bc": 0.03845445439219475, "train_loss_llm": 0.0, "grad_norm": 0.0839417353272438, "global_step": 15028, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03799913451075554, "train_loss_bc": 0.03799913451075554, "train_loss_llm": 0.0, "grad_norm": 0.026836557313799858, "global_step": 15029, "epoch": 417, "lr": 0.001} +{"train_loss": 0.037804171442985535, "train_loss_bc": 0.037792425602674484, "train_loss_llm": 0.0011747181415557861, "grad_norm": 0.047913286834955215, "global_step": 15030, "epoch": 417, "lr": 0.001} +{"train_loss": 0.037493523210287094, "train_loss_bc": 0.037493523210287094, "train_loss_llm": 0.0, "grad_norm": 0.05848650634288788, "global_step": 15031, "epoch": 417, "lr": 0.001} +{"train_loss": 0.04003852605819702, "train_loss_bc": 0.040017709136009216, "train_loss_llm": 0.0020817816257476807, "grad_norm": 0.07386060804128647, "global_step": 15032, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03510059788823128, "train_loss_bc": 0.03510059788823128, "train_loss_llm": 0.0, "grad_norm": 0.019999807700514793, "global_step": 15033, "epoch": 417, "lr": 0.001} +{"train_loss": 0.036507654935121536, "train_loss_bc": 0.036507654935121536, "train_loss_llm": 0.0, "grad_norm": 0.029328685253858566, "global_step": 15034, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03712853789329529, "train_loss_bc": 0.03712446987628937, "train_loss_llm": 0.0004069209098815918, "grad_norm": 0.036338113248348236, "global_step": 15035, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03493519127368927, "train_loss_bc": 0.03493048995733261, "train_loss_llm": 0.00047028064727783203, "grad_norm": 0.04445755109190941, "global_step": 15036, "epoch": 417, "lr": 0.001} +{"train_loss": 0.04276062920689583, "train_loss_bc": 0.04276062920689583, "train_loss_llm": 0.0, "grad_norm": 0.022645607590675354, "global_step": 15037, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03820931166410446, "train_loss_bc": 0.038205958902835846, "train_loss_llm": 0.00033530592918395996, "grad_norm": 0.02958703599870205, "global_step": 15038, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03433430939912796, "train_loss_bc": 0.03433430939912796, "train_loss_llm": 0.0, "grad_norm": 0.03874189034104347, "global_step": 15039, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03832679241895676, "train_loss_bc": 0.03832679241895676, "train_loss_llm": 0.0, "grad_norm": 0.05015687271952629, "global_step": 15040, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03379274904727936, "train_loss_bc": 0.03379274904727936, "train_loss_llm": 0.0, "grad_norm": 0.024031618610024452, "global_step": 15041, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03921834006905556, "train_loss_bc": 0.03921139985322952, "train_loss_llm": 0.0006939470767974854, "grad_norm": 0.03676588088274002, "global_step": 15042, "epoch": 417, "lr": 0.001} +{"train_loss": 0.030590249225497246, "train_loss_bc": 0.030590249225497246, "train_loss_llm": 0.0, "grad_norm": 0.04866139218211174, "global_step": 15043, "epoch": 417, "lr": 0.001} +{"train_loss": 0.041421882808208466, "train_loss_bc": 0.041421882808208466, "train_loss_llm": 0.0, "grad_norm": 0.06339670717716217, "global_step": 15044, "epoch": 417, "lr": 0.001} +{"train_loss": 0.021868424490094185, "train_loss_bc": 0.021868424490094185, "train_loss_llm": 0.0, "grad_norm": 0.02284311316907406, "global_step": 15045, "epoch": 417, "lr": 0.001} +{"train_loss": 0.03612660989165306, "train_loss_bc": 0.03612660989165306, "train_loss_llm": 0.0, "grad_norm": 0.04415488615632057, "global_step": 15046, "epoch": 417, "lr": 0.001} +{"train_loss": 0.035802919469359845, "train_loss_bc": 0.0395628921687603, "train_loss_llm": 0.0, "grad_norm": 0.07855509221553802, "global_step": 15047, "epoch": 417, "lr": 0.001, "val_loss": 0.06922517716884613} +{"train_loss": 0.03888145461678505, "train_loss_bc": 0.03888145461678505, "train_loss_llm": 0.0, "grad_norm": 0.09191540628671646, "global_step": 15048, "epoch": 418, "lr": 0.001} +{"train_loss": 0.033042192459106445, "train_loss_bc": 0.033042192459106445, "train_loss_llm": 0.0, "grad_norm": 0.02845296822488308, "global_step": 15049, "epoch": 418, "lr": 0.001} +{"train_loss": 0.040743689984083176, "train_loss_bc": 0.04072137176990509, "train_loss_llm": 0.002231776714324951, "grad_norm": 0.0502258837223053, "global_step": 15050, "epoch": 418, "lr": 0.001} +{"train_loss": 0.04344711825251579, "train_loss_bc": 0.04344711825251579, "train_loss_llm": 0.0, "grad_norm": 0.06795639544725418, "global_step": 15051, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03675108775496483, "train_loss_bc": 0.03673286363482475, "train_loss_llm": 0.0018225908279418945, "grad_norm": 0.08838161081075668, "global_step": 15052, "epoch": 418, "lr": 0.001} +{"train_loss": 0.037245526909828186, "train_loss_bc": 0.037245526909828186, "train_loss_llm": 0.0, "grad_norm": 0.02326485887169838, "global_step": 15053, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03477137163281441, "train_loss_bc": 0.03477137163281441, "train_loss_llm": 0.0, "grad_norm": 0.040831200778484344, "global_step": 15054, "epoch": 418, "lr": 0.001} +{"train_loss": 0.030033092945814133, "train_loss_bc": 0.030033092945814133, "train_loss_llm": 0.0, "grad_norm": 0.044918179512023926, "global_step": 15055, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03553922101855278, "train_loss_bc": 0.03553922101855278, "train_loss_llm": 0.0, "grad_norm": 0.05472160130739212, "global_step": 15056, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03898487240076065, "train_loss_bc": 0.03898487240076065, "train_loss_llm": 0.0, "grad_norm": 0.022067667916417122, "global_step": 15057, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03319887816905975, "train_loss_bc": 0.03319887816905975, "train_loss_llm": 0.0, "grad_norm": 0.04038329795002937, "global_step": 15058, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03711090236902237, "train_loss_bc": 0.03711090236902237, "train_loss_llm": 0.0, "grad_norm": 0.05363183096051216, "global_step": 15059, "epoch": 418, "lr": 0.001} +{"train_loss": 0.04480154439806938, "train_loss_bc": 0.04480154439806938, "train_loss_llm": 0.0, "grad_norm": 0.07283426076173782, "global_step": 15060, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03478708490729332, "train_loss_bc": 0.03478313982486725, "train_loss_llm": 0.00039440393447875977, "grad_norm": 0.025375675410032272, "global_step": 15061, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03238304331898689, "train_loss_bc": 0.03238304331898689, "train_loss_llm": 0.0, "grad_norm": 0.04414881393313408, "global_step": 15062, "epoch": 418, "lr": 0.001} +{"train_loss": 0.0388883538544178, "train_loss_bc": 0.0388883538544178, "train_loss_llm": 0.0, "grad_norm": 0.05948691442608833, "global_step": 15063, "epoch": 418, "lr": 0.001} +{"train_loss": 0.0316060446202755, "train_loss_bc": 0.0316060446202755, "train_loss_llm": 0.0, "grad_norm": 0.07489079236984253, "global_step": 15064, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03362178057432175, "train_loss_bc": 0.03362178057432175, "train_loss_llm": 0.0, "grad_norm": 0.024468397721648216, "global_step": 15065, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03504210337996483, "train_loss_bc": 0.03499181196093559, "train_loss_llm": 0.005029082298278809, "grad_norm": 0.040967315435409546, "global_step": 15066, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03420984372496605, "train_loss_bc": 0.03420669585466385, "train_loss_llm": 0.0003147125244140625, "grad_norm": 0.046855177730321884, "global_step": 15067, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03952788934111595, "train_loss_bc": 0.039510756731033325, "train_loss_llm": 0.001713186502456665, "grad_norm": 0.06035218387842178, "global_step": 15068, "epoch": 418, "lr": 0.001} +{"train_loss": 0.034398920834064484, "train_loss_bc": 0.03439614176750183, "train_loss_llm": 0.0002778470516204834, "grad_norm": 0.03525960445404053, "global_step": 15069, "epoch": 418, "lr": 0.001} +{"train_loss": 0.0377434566617012, "train_loss_bc": 0.037727147340774536, "train_loss_llm": 0.0016308724880218506, "grad_norm": 0.05525623634457588, "global_step": 15070, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03620307147502899, "train_loss_bc": 0.03620307147502899, "train_loss_llm": 0.0, "grad_norm": 0.06992106139659882, "global_step": 15071, "epoch": 418, "lr": 0.001} +{"train_loss": 0.034252434968948364, "train_loss_bc": 0.034252434968948364, "train_loss_llm": 0.0, "grad_norm": 0.09028812497854233, "global_step": 15072, "epoch": 418, "lr": 0.001} +{"train_loss": 0.0333012193441391, "train_loss_bc": 0.0333012193441391, "train_loss_llm": 0.0, "grad_norm": 0.02453724853694439, "global_step": 15073, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03752145171165466, "train_loss_bc": 0.03752145171165466, "train_loss_llm": 0.0, "grad_norm": 0.04075620323419571, "global_step": 15074, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03274604305624962, "train_loss_bc": 0.03273085132241249, "train_loss_llm": 0.0015191435813903809, "grad_norm": 0.06258519738912582, "global_step": 15075, "epoch": 418, "lr": 0.001} +{"train_loss": 0.027631476521492004, "train_loss_bc": 0.027614077553153038, "train_loss_llm": 0.0017399489879608154, "grad_norm": 0.0750681608915329, "global_step": 15076, "epoch": 418, "lr": 0.001} +{"train_loss": 0.029831357300281525, "train_loss_bc": 0.029831357300281525, "train_loss_llm": 0.0, "grad_norm": 0.02137337252497673, "global_step": 15077, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03436998650431633, "train_loss_bc": 0.03436998650431633, "train_loss_llm": 0.0, "grad_norm": 0.03192459046840668, "global_step": 15078, "epoch": 418, "lr": 0.001} +{"train_loss": 0.041185397654771805, "train_loss_bc": 0.041185397654771805, "train_loss_llm": 0.0, "grad_norm": 0.04668738320469856, "global_step": 15079, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03604349493980408, "train_loss_bc": 0.03604349493980408, "train_loss_llm": 0.0, "grad_norm": 0.058458082377910614, "global_step": 15080, "epoch": 418, "lr": 0.001} +{"train_loss": 0.034210383892059326, "train_loss_bc": 0.03419753164052963, "train_loss_llm": 0.001285165548324585, "grad_norm": 0.018998293206095695, "global_step": 15081, "epoch": 418, "lr": 0.001} +{"train_loss": 0.03746232017874718, "train_loss_bc": 0.03745380416512489, "train_loss_llm": 0.0008517801761627197, "grad_norm": 0.030592618510127068, "global_step": 15082, "epoch": 418, "lr": 0.001} +{"train_loss": 0.035621559609555535, "train_loss_bc": 0.030858034268021584, "train_loss_llm": 0.0, "grad_norm": 0.04736977070569992, "global_step": 15083, "epoch": 418, "lr": 0.001, "val_loss": 0.0696275383234024} +{"train_loss": 0.0303660798817873, "train_loss_bc": 0.0303660798817873, "train_loss_llm": 0.0, "grad_norm": 0.0528644397854805, "global_step": 15084, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03720475360751152, "train_loss_bc": 0.03720475360751152, "train_loss_llm": 0.0, "grad_norm": 0.025168495252728462, "global_step": 15085, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03329925611615181, "train_loss_bc": 0.03329925611615181, "train_loss_llm": 0.0, "grad_norm": 0.0371132530272007, "global_step": 15086, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03762176260352135, "train_loss_bc": 0.037604548037052155, "train_loss_llm": 0.0017215609550476074, "grad_norm": 0.043438542634248734, "global_step": 15087, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03699098154902458, "train_loss_bc": 0.03699098154902458, "train_loss_llm": 0.0, "grad_norm": 0.050449859350919724, "global_step": 15088, "epoch": 419, "lr": 0.001} +{"train_loss": 0.024957453832030296, "train_loss_bc": 0.024957453832030296, "train_loss_llm": 0.0, "grad_norm": 0.023060554638504982, "global_step": 15089, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03469730168581009, "train_loss_bc": 0.03469730168581009, "train_loss_llm": 0.0, "grad_norm": 0.04366537928581238, "global_step": 15090, "epoch": 419, "lr": 0.001} +{"train_loss": 0.0433509424328804, "train_loss_bc": 0.043336741626262665, "train_loss_llm": 0.0014200806617736816, "grad_norm": 0.06317611783742905, "global_step": 15091, "epoch": 419, "lr": 0.001} +{"train_loss": 0.036157213151454926, "train_loss_bc": 0.036157213151454926, "train_loss_llm": 0.0, "grad_norm": 0.07742611318826675, "global_step": 15092, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03981165587902069, "train_loss_bc": 0.03979368135333061, "train_loss_llm": 0.0017974674701690674, "grad_norm": 0.02304885722696781, "global_step": 15093, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03626278042793274, "train_loss_bc": 0.03623592108488083, "train_loss_llm": 0.0026858150959014893, "grad_norm": 0.029653210192918777, "global_step": 15094, "epoch": 419, "lr": 0.001} +{"train_loss": 0.041167229413986206, "train_loss_bc": 0.041167229413986206, "train_loss_llm": 0.0, "grad_norm": 0.04555242136120796, "global_step": 15095, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03212937340140343, "train_loss_bc": 0.03212937340140343, "train_loss_llm": 0.0, "grad_norm": 0.05606146901845932, "global_step": 15096, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03364598751068115, "train_loss_bc": 0.03364288806915283, "train_loss_llm": 0.00030997395515441895, "grad_norm": 0.028124427422881126, "global_step": 15097, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03964492678642273, "train_loss_bc": 0.03964492678642273, "train_loss_llm": 0.0, "grad_norm": 0.058583494275808334, "global_step": 15098, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03625587746500969, "train_loss_bc": 0.03625587746500969, "train_loss_llm": 0.0, "grad_norm": 0.07466233521699905, "global_step": 15099, "epoch": 419, "lr": 0.001} +{"train_loss": 0.031599368900060654, "train_loss_bc": 0.031599368900060654, "train_loss_llm": 0.0, "grad_norm": 0.09163662046194077, "global_step": 15100, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03935440629720688, "train_loss_bc": 0.03935440629720688, "train_loss_llm": 0.0, "grad_norm": 0.03614248335361481, "global_step": 15101, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03737969323992729, "train_loss_bc": 0.03733121603727341, "train_loss_llm": 0.00484776496887207, "grad_norm": 0.04770747572183609, "global_step": 15102, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03535570576786995, "train_loss_bc": 0.03535570576786995, "train_loss_llm": 0.0, "grad_norm": 0.05986025184392929, "global_step": 15103, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03389744833111763, "train_loss_bc": 0.03389744833111763, "train_loss_llm": 0.0, "grad_norm": 0.07148982584476471, "global_step": 15104, "epoch": 419, "lr": 0.001} +{"train_loss": 0.029286865144968033, "train_loss_bc": 0.029286865144968033, "train_loss_llm": 0.0, "grad_norm": 0.02332034520804882, "global_step": 15105, "epoch": 419, "lr": 0.001} +{"train_loss": 0.035026516765356064, "train_loss_bc": 0.03500252962112427, "train_loss_llm": 0.002398669719696045, "grad_norm": 0.036663684993982315, "global_step": 15106, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03763287141919136, "train_loss_bc": 0.03763287141919136, "train_loss_llm": 0.0, "grad_norm": 0.04825551062822342, "global_step": 15107, "epoch": 419, "lr": 0.001} +{"train_loss": 0.036425620317459106, "train_loss_bc": 0.036425620317459106, "train_loss_llm": 0.0, "grad_norm": 0.060359664261341095, "global_step": 15108, "epoch": 419, "lr": 0.001} +{"train_loss": 0.035661179572343826, "train_loss_bc": 0.03564423322677612, "train_loss_llm": 0.0016946494579315186, "grad_norm": 0.026153534650802612, "global_step": 15109, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03832157701253891, "train_loss_bc": 0.03832157701253891, "train_loss_llm": 0.0, "grad_norm": 0.04058963805437088, "global_step": 15110, "epoch": 419, "lr": 0.001} +{"train_loss": 0.029664501547813416, "train_loss_bc": 0.029664501547813416, "train_loss_llm": 0.0, "grad_norm": 0.045890867710113525, "global_step": 15111, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03212518244981766, "train_loss_bc": 0.03212518244981766, "train_loss_llm": 0.0, "grad_norm": 0.05653096362948418, "global_step": 15112, "epoch": 419, "lr": 0.001} +{"train_loss": 0.037831857800483704, "train_loss_bc": 0.037831857800483704, "train_loss_llm": 0.0, "grad_norm": 0.023307232186198235, "global_step": 15113, "epoch": 419, "lr": 0.001} +{"train_loss": 0.033744703978300095, "train_loss_bc": 0.0337410643696785, "train_loss_llm": 0.00036406517028808594, "grad_norm": 0.03858669847249985, "global_step": 15114, "epoch": 419, "lr": 0.001} +{"train_loss": 0.033858951181173325, "train_loss_bc": 0.033858951181173325, "train_loss_llm": 0.0, "grad_norm": 0.05389690771698952, "global_step": 15115, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03434883803129196, "train_loss_bc": 0.03434883803129196, "train_loss_llm": 0.0, "grad_norm": 0.0644816979765892, "global_step": 15116, "epoch": 419, "lr": 0.001} +{"train_loss": 0.041743259876966476, "train_loss_bc": 0.041743259876966476, "train_loss_llm": 0.0, "grad_norm": 0.042653366923332214, "global_step": 15117, "epoch": 419, "lr": 0.001} +{"train_loss": 0.038731176406145096, "train_loss_bc": 0.038728129118680954, "train_loss_llm": 0.00030475854873657227, "grad_norm": 0.05955745279788971, "global_step": 15118, "epoch": 419, "lr": 0.001} +{"train_loss": 0.03570594566149844, "train_loss_bc": 0.039854589849710464, "train_loss_llm": 0.0006152689456939697, "grad_norm": 0.08446049690246582, "global_step": 15119, "epoch": 419, "lr": 0.001, "val_loss": 0.06871367990970612} +{"train_loss": 0.0379464291036129, "train_loss_bc": 0.0379464291036129, "train_loss_llm": 0.0, "grad_norm": 0.09671173989772797, "global_step": 15120, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03295456990599632, "train_loss_bc": 0.03295456990599632, "train_loss_llm": 0.0, "grad_norm": 0.027187101542949677, "global_step": 15121, "epoch": 420, "lr": 0.001} +{"train_loss": 0.024901965633034706, "train_loss_bc": 0.024885762482881546, "train_loss_llm": 0.0016203522682189941, "grad_norm": 0.03852282091975212, "global_step": 15122, "epoch": 420, "lr": 0.001} +{"train_loss": 0.028190983459353447, "train_loss_bc": 0.028185728937387466, "train_loss_llm": 0.0005254149436950684, "grad_norm": 0.04828678071498871, "global_step": 15123, "epoch": 420, "lr": 0.001} +{"train_loss": 0.034431032836437225, "train_loss_bc": 0.03441864997148514, "train_loss_llm": 0.0012383460998535156, "grad_norm": 0.058991000056266785, "global_step": 15124, "epoch": 420, "lr": 0.001} +{"train_loss": 0.04092545434832573, "train_loss_bc": 0.04091740399599075, "train_loss_llm": 0.0008048713207244873, "grad_norm": 0.02668674662709236, "global_step": 15125, "epoch": 420, "lr": 0.001} +{"train_loss": 0.038930587470531464, "train_loss_bc": 0.038930587470531464, "train_loss_llm": 0.0, "grad_norm": 0.03951304778456688, "global_step": 15126, "epoch": 420, "lr": 0.001} +{"train_loss": 0.041283514350652695, "train_loss_bc": 0.041283514350652695, "train_loss_llm": 0.0, "grad_norm": 0.05665254220366478, "global_step": 15127, "epoch": 420, "lr": 0.001} +{"train_loss": 0.032740574330091476, "train_loss_bc": 0.032740574330091476, "train_loss_llm": 0.0, "grad_norm": 0.06915754824876785, "global_step": 15128, "epoch": 420, "lr": 0.001} +{"train_loss": 0.02788831852376461, "train_loss_bc": 0.027878591790795326, "train_loss_llm": 0.0009726285934448242, "grad_norm": 0.022182540968060493, "global_step": 15129, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03847046568989754, "train_loss_bc": 0.03847046568989754, "train_loss_llm": 0.0, "grad_norm": 0.03252455219626427, "global_step": 15130, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03879474103450775, "train_loss_bc": 0.038789618760347366, "train_loss_llm": 0.0005120635032653809, "grad_norm": 0.046029508113861084, "global_step": 15131, "epoch": 420, "lr": 0.001} +{"train_loss": 0.039359040558338165, "train_loss_bc": 0.039359040558338165, "train_loss_llm": 0.0, "grad_norm": 0.05516150966286659, "global_step": 15132, "epoch": 420, "lr": 0.001} +{"train_loss": 0.033507440239191055, "train_loss_bc": 0.033507440239191055, "train_loss_llm": 0.0, "grad_norm": 0.022687295451760292, "global_step": 15133, "epoch": 420, "lr": 0.001} +{"train_loss": 0.029874080792069435, "train_loss_bc": 0.029874080792069435, "train_loss_llm": 0.0, "grad_norm": 0.03616360202431679, "global_step": 15134, "epoch": 420, "lr": 0.001} +{"train_loss": 0.032234206795692444, "train_loss_bc": 0.032234206795692444, "train_loss_llm": 0.0, "grad_norm": 0.04370427131652832, "global_step": 15135, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03313567489385605, "train_loss_bc": 0.03313567489385605, "train_loss_llm": 0.0, "grad_norm": 0.05284089967608452, "global_step": 15136, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03867162764072418, "train_loss_bc": 0.03866131603717804, "train_loss_llm": 0.001031041145324707, "grad_norm": 0.027396786957979202, "global_step": 15137, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03871684893965721, "train_loss_bc": 0.03871684893965721, "train_loss_llm": 0.0, "grad_norm": 0.0518195740878582, "global_step": 15138, "epoch": 420, "lr": 0.001} +{"train_loss": 0.036949437111616135, "train_loss_bc": 0.03693680837750435, "train_loss_llm": 0.0012627243995666504, "grad_norm": 0.07024258375167847, "global_step": 15139, "epoch": 420, "lr": 0.001} +{"train_loss": 0.033700790256261826, "train_loss_bc": 0.03369557857513428, "train_loss_llm": 0.0005213022232055664, "grad_norm": 0.08844906836748123, "global_step": 15140, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03541499003767967, "train_loss_bc": 0.03541499003767967, "train_loss_llm": 0.0, "grad_norm": 0.02533753216266632, "global_step": 15141, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03641785308718681, "train_loss_bc": 0.03641785308718681, "train_loss_llm": 0.0, "grad_norm": 0.04191487655043602, "global_step": 15142, "epoch": 420, "lr": 0.001} +{"train_loss": 0.02901388891041279, "train_loss_bc": 0.02900978922843933, "train_loss_llm": 0.000409930944442749, "grad_norm": 0.04963046684861183, "global_step": 15143, "epoch": 420, "lr": 0.001} +{"train_loss": 0.037915609776973724, "train_loss_bc": 0.037915609776973724, "train_loss_llm": 0.0, "grad_norm": 0.07100614905357361, "global_step": 15144, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03750845789909363, "train_loss_bc": 0.03750845789909363, "train_loss_llm": 0.0, "grad_norm": 0.034725867211818695, "global_step": 15145, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03546467050909996, "train_loss_bc": 0.03545476123690605, "train_loss_llm": 0.0009908676147460938, "grad_norm": 0.044126957654953, "global_step": 15146, "epoch": 420, "lr": 0.001} +{"train_loss": 0.04131288826465607, "train_loss_bc": 0.04131288826465607, "train_loss_llm": 0.0, "grad_norm": 0.07128248363733292, "global_step": 15147, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03604861721396446, "train_loss_bc": 0.036029767245054245, "train_loss_llm": 0.001885145902633667, "grad_norm": 0.07482994347810745, "global_step": 15148, "epoch": 420, "lr": 0.001} +{"train_loss": 0.04104055091738701, "train_loss_bc": 0.041035011410713196, "train_loss_llm": 0.0005537867546081543, "grad_norm": 0.03195871412754059, "global_step": 15149, "epoch": 420, "lr": 0.001} +{"train_loss": 0.039095908403396606, "train_loss_bc": 0.039095908403396606, "train_loss_llm": 0.0, "grad_norm": 0.03524018079042435, "global_step": 15150, "epoch": 420, "lr": 0.001} +{"train_loss": 0.039015255868434906, "train_loss_bc": 0.039015255868434906, "train_loss_llm": 0.0, "grad_norm": 0.04907725751399994, "global_step": 15151, "epoch": 420, "lr": 0.001} +{"train_loss": 0.034453846514225006, "train_loss_bc": 0.034453846514225006, "train_loss_llm": 0.0, "grad_norm": 0.05639892444014549, "global_step": 15152, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03712647780776024, "train_loss_bc": 0.03712647780776024, "train_loss_llm": 0.0, "grad_norm": 0.024089548736810684, "global_step": 15153, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03676585853099823, "train_loss_bc": 0.036745503544807434, "train_loss_llm": 0.002035468816757202, "grad_norm": 0.03477952629327774, "global_step": 15154, "epoch": 420, "lr": 0.001} +{"train_loss": 0.03560144573243128, "train_loss_bc": 0.03144938871264458, "train_loss_llm": 0.0, "grad_norm": 0.048087406903505325, "global_step": 15155, "epoch": 420, "lr": 0.001, "train/cumulative_reward": 82.84061233745653, "train/mean_score": 14.736899637685104, "train/success_rate": 0.4, "test/cumulative_reward": 30.704453299674473, "test/mean_score": 7.624157045019584, "test/success_rate": 0.18, "val_loss": 0.07066871970891953, "train_action_mse_error": 0.038786597549915314} +{"train_loss": 0.040787048637866974, "train_loss_bc": 0.040787048637866974, "train_loss_llm": 0.0, "grad_norm": 0.06428945809602737, "global_step": 15156, "epoch": 421, "lr": 0.001} +{"train_loss": 0.030370131134986877, "train_loss_bc": 0.03036760538816452, "train_loss_llm": 0.0002525448799133301, "grad_norm": 0.022505149245262146, "global_step": 15157, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03488849103450775, "train_loss_bc": 0.03488849103450775, "train_loss_llm": 0.0, "grad_norm": 0.03558594360947609, "global_step": 15158, "epoch": 421, "lr": 0.001} +{"train_loss": 0.029720500111579895, "train_loss_bc": 0.029719645157456398, "train_loss_llm": 8.541345596313477e-05, "grad_norm": 0.04836156591773033, "global_step": 15159, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03811532258987427, "train_loss_bc": 0.03810945525765419, "train_loss_llm": 0.0005866289138793945, "grad_norm": 0.06032728776335716, "global_step": 15160, "epoch": 421, "lr": 0.001} +{"train_loss": 0.031694453209638596, "train_loss_bc": 0.031694453209638596, "train_loss_llm": 0.0, "grad_norm": 0.025229763239622116, "global_step": 15161, "epoch": 421, "lr": 0.001} +{"train_loss": 0.031071968376636505, "train_loss_bc": 0.031071968376636505, "train_loss_llm": 0.0, "grad_norm": 0.04346233606338501, "global_step": 15162, "epoch": 421, "lr": 0.001} +{"train_loss": 0.041705172508955, "train_loss_bc": 0.041705172508955, "train_loss_llm": 0.0, "grad_norm": 0.07171224057674408, "global_step": 15163, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03318321704864502, "train_loss_bc": 0.03318321704864502, "train_loss_llm": 0.0, "grad_norm": 0.08945688605308533, "global_step": 15164, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03603705018758774, "train_loss_bc": 0.03603705018758774, "train_loss_llm": 0.0, "grad_norm": 0.018727021291851997, "global_step": 15165, "epoch": 421, "lr": 0.001} +{"train_loss": 0.038452427834272385, "train_loss_bc": 0.038452427834272385, "train_loss_llm": 0.0, "grad_norm": 0.04309021681547165, "global_step": 15166, "epoch": 421, "lr": 0.001} +{"train_loss": 0.032930705696344376, "train_loss_bc": 0.032930705696344376, "train_loss_llm": 0.0, "grad_norm": 0.06012127548456192, "global_step": 15167, "epoch": 421, "lr": 0.001} +{"train_loss": 0.034500602632761, "train_loss_bc": 0.03450030833482742, "train_loss_llm": 2.9265880584716797e-05, "grad_norm": 0.07222861051559448, "global_step": 15168, "epoch": 421, "lr": 0.001} +{"train_loss": 0.02986924722790718, "train_loss_bc": 0.02986924722790718, "train_loss_llm": 0.0, "grad_norm": 0.01580657623708248, "global_step": 15169, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03628097474575043, "train_loss_bc": 0.036248963326215744, "train_loss_llm": 0.0032012462615966797, "grad_norm": 0.038693781942129135, "global_step": 15170, "epoch": 421, "lr": 0.001} +{"train_loss": 0.04458484798669815, "train_loss_bc": 0.044572651386260986, "train_loss_llm": 0.0012197792530059814, "grad_norm": 0.050298746675252914, "global_step": 15171, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03127249330282211, "train_loss_bc": 0.03127249330282211, "train_loss_llm": 0.0, "grad_norm": 0.05956040322780609, "global_step": 15172, "epoch": 421, "lr": 0.001} +{"train_loss": 0.0376487672328949, "train_loss_bc": 0.03764871135354042, "train_loss_llm": 5.424022674560547e-06, "grad_norm": 0.030176471918821335, "global_step": 15173, "epoch": 421, "lr": 0.001} +{"train_loss": 0.032348960638046265, "train_loss_bc": 0.032348960638046265, "train_loss_llm": 0.0, "grad_norm": 0.04853621497750282, "global_step": 15174, "epoch": 421, "lr": 0.001} +{"train_loss": 0.0396156832575798, "train_loss_bc": 0.0396156832575798, "train_loss_llm": 0.0, "grad_norm": 0.06443911790847778, "global_step": 15175, "epoch": 421, "lr": 0.001} +{"train_loss": 0.033531613647937775, "train_loss_bc": 0.033531613647937775, "train_loss_llm": 0.0, "grad_norm": 0.07411990314722061, "global_step": 15176, "epoch": 421, "lr": 0.001} +{"train_loss": 0.02823638916015625, "train_loss_bc": 0.028228064998984337, "train_loss_llm": 0.0008323788642883301, "grad_norm": 0.02126234397292137, "global_step": 15177, "epoch": 421, "lr": 0.001} +{"train_loss": 0.04050154611468315, "train_loss_bc": 0.04050084576010704, "train_loss_llm": 7.000565528869629e-05, "grad_norm": 0.03584985062479973, "global_step": 15178, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03686757758259773, "train_loss_bc": 0.03686757758259773, "train_loss_llm": 0.0, "grad_norm": 0.04374965280294418, "global_step": 15179, "epoch": 421, "lr": 0.001} +{"train_loss": 0.035155221819877625, "train_loss_bc": 0.03514977544546127, "train_loss_llm": 0.0005446672439575195, "grad_norm": 0.054311949759721756, "global_step": 15180, "epoch": 421, "lr": 0.001} +{"train_loss": 0.04180465266108513, "train_loss_bc": 0.04178983345627785, "train_loss_llm": 0.001481860876083374, "grad_norm": 0.02309386432170868, "global_step": 15181, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03697347268462181, "train_loss_bc": 0.03697347268462181, "train_loss_llm": 0.0, "grad_norm": 0.03617551550269127, "global_step": 15182, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03702018782496452, "train_loss_bc": 0.03702018782496452, "train_loss_llm": 0.0, "grad_norm": 0.04662328585982323, "global_step": 15183, "epoch": 421, "lr": 0.001} +{"train_loss": 0.033609624952077866, "train_loss_bc": 0.033609624952077866, "train_loss_llm": 0.0, "grad_norm": 0.06208242475986481, "global_step": 15184, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03513145446777344, "train_loss_bc": 0.03513145446777344, "train_loss_llm": 0.0, "grad_norm": 0.024409664794802666, "global_step": 15185, "epoch": 421, "lr": 0.001} +{"train_loss": 0.04147518053650856, "train_loss_bc": 0.04147518053650856, "train_loss_llm": 0.0, "grad_norm": 0.04078570008277893, "global_step": 15186, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03964391350746155, "train_loss_bc": 0.03964391350746155, "train_loss_llm": 0.0, "grad_norm": 0.0543241910636425, "global_step": 15187, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03336077556014061, "train_loss_bc": 0.03336077556014061, "train_loss_llm": 0.0, "grad_norm": 0.0761336237192154, "global_step": 15188, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03818034380674362, "train_loss_bc": 0.03818034380674362, "train_loss_llm": 0.0, "grad_norm": 0.02559918910264969, "global_step": 15189, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03284564986824989, "train_loss_bc": 0.032845184206962585, "train_loss_llm": 4.649162292480469e-05, "grad_norm": 0.03775894269347191, "global_step": 15190, "epoch": 421, "lr": 0.001} +{"train_loss": 0.03557918965816498, "train_loss_bc": 0.03143515810370445, "train_loss_llm": 0.0, "grad_norm": 0.05392895266413689, "global_step": 15191, "epoch": 421, "lr": 0.001, "val_loss": 0.0688130110502243} +{"train_loss": 0.044071439653635025, "train_loss_bc": 0.044071439653635025, "train_loss_llm": 0.0, "grad_norm": 0.07173392176628113, "global_step": 15192, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03447464480996132, "train_loss_bc": 0.03447464480996132, "train_loss_llm": 0.0, "grad_norm": 0.0223543718457222, "global_step": 15193, "epoch": 422, "lr": 0.001} +{"train_loss": 0.02954983152449131, "train_loss_bc": 0.029534300789237022, "train_loss_llm": 0.0015531182289123535, "grad_norm": 0.037940654903650284, "global_step": 15194, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03522011637687683, "train_loss_bc": 0.03522011637687683, "train_loss_llm": 0.0, "grad_norm": 0.046056412160396576, "global_step": 15195, "epoch": 422, "lr": 0.001} +{"train_loss": 0.0311510618776083, "train_loss_bc": 0.0311510618776083, "train_loss_llm": 0.0, "grad_norm": 0.05205309018492699, "global_step": 15196, "epoch": 422, "lr": 0.001} +{"train_loss": 0.027871763333678246, "train_loss_bc": 0.027840524911880493, "train_loss_llm": 0.0031238198280334473, "grad_norm": 0.022345425561070442, "global_step": 15197, "epoch": 422, "lr": 0.001} +{"train_loss": 0.040062300860881805, "train_loss_bc": 0.040062300860881805, "train_loss_llm": 0.0, "grad_norm": 0.040267378091812134, "global_step": 15198, "epoch": 422, "lr": 0.001} +{"train_loss": 0.0363711416721344, "train_loss_bc": 0.036329612135887146, "train_loss_llm": 0.004152774810791016, "grad_norm": 0.058051422238349915, "global_step": 15199, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03248310089111328, "train_loss_bc": 0.03248310089111328, "train_loss_llm": 0.0, "grad_norm": 0.06621403247117996, "global_step": 15200, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03855927288532257, "train_loss_bc": 0.03855927288532257, "train_loss_llm": 0.0, "grad_norm": 0.028667699545621872, "global_step": 15201, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03673974424600601, "train_loss_bc": 0.0367361418902874, "train_loss_llm": 0.00036031007766723633, "grad_norm": 0.040922343730926514, "global_step": 15202, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03750474750995636, "train_loss_bc": 0.03750474750995636, "train_loss_llm": 0.0, "grad_norm": 0.05459544435143471, "global_step": 15203, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03714409098029137, "train_loss_bc": 0.03714409098029137, "train_loss_llm": 0.0, "grad_norm": 0.06487800180912018, "global_step": 15204, "epoch": 422, "lr": 0.001} +{"train_loss": 0.0328354649245739, "train_loss_bc": 0.0328354649245739, "train_loss_llm": 0.0, "grad_norm": 0.024367565289139748, "global_step": 15205, "epoch": 422, "lr": 0.001} +{"train_loss": 0.0364544577896595, "train_loss_bc": 0.0364544577896595, "train_loss_llm": 0.0, "grad_norm": 0.03859048709273338, "global_step": 15206, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03657319024205208, "train_loss_bc": 0.03657319024205208, "train_loss_llm": 0.0, "grad_norm": 0.05341772362589836, "global_step": 15207, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03256351500749588, "train_loss_bc": 0.03255540877580643, "train_loss_llm": 0.0008105933666229248, "grad_norm": 0.06385087221860886, "global_step": 15208, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03876229003071785, "train_loss_bc": 0.03876229003071785, "train_loss_llm": 0.0, "grad_norm": 0.02876819111406803, "global_step": 15209, "epoch": 422, "lr": 0.001} +{"train_loss": 0.04193553328514099, "train_loss_bc": 0.04193553328514099, "train_loss_llm": 0.0, "grad_norm": 0.041962143033742905, "global_step": 15210, "epoch": 422, "lr": 0.001} +{"train_loss": 0.04199206829071045, "train_loss_bc": 0.04197777435183525, "train_loss_llm": 0.001429438591003418, "grad_norm": 0.05343880131840706, "global_step": 15211, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03542192280292511, "train_loss_bc": 0.03542120009660721, "train_loss_llm": 7.233023643493652e-05, "grad_norm": 0.05906235799193382, "global_step": 15212, "epoch": 422, "lr": 0.001} +{"train_loss": 0.033266909420490265, "train_loss_bc": 0.03325745835900307, "train_loss_llm": 0.0009452700614929199, "grad_norm": 0.029100315645337105, "global_step": 15213, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03570681810379028, "train_loss_bc": 0.035706136375665665, "train_loss_llm": 6.80685043334961e-05, "grad_norm": 0.0376523919403553, "global_step": 15214, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03320986405014992, "train_loss_bc": 0.033188916742801666, "train_loss_llm": 0.002094566822052002, "grad_norm": 0.0450667180120945, "global_step": 15215, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03153405711054802, "train_loss_bc": 0.03153405711054802, "train_loss_llm": 0.0, "grad_norm": 0.050857581198215485, "global_step": 15216, "epoch": 422, "lr": 0.001} +{"train_loss": 0.040477409958839417, "train_loss_bc": 0.040477409958839417, "train_loss_llm": 0.0, "grad_norm": 0.02837827429175377, "global_step": 15217, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03776149824261665, "train_loss_bc": 0.03776149824261665, "train_loss_llm": 0.0, "grad_norm": 0.03994104266166687, "global_step": 15218, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03651537373661995, "train_loss_bc": 0.03651537373661995, "train_loss_llm": 0.0, "grad_norm": 0.05794680491089821, "global_step": 15219, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03539721667766571, "train_loss_bc": 0.03539721667766571, "train_loss_llm": 0.0, "grad_norm": 0.0594540573656559, "global_step": 15220, "epoch": 422, "lr": 0.001} +{"train_loss": 0.04443937912583351, "train_loss_bc": 0.04443937912583351, "train_loss_llm": 0.0, "grad_norm": 0.033698804676532745, "global_step": 15221, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03165590763092041, "train_loss_bc": 0.03165590763092041, "train_loss_llm": 0.0, "grad_norm": 0.046506550163030624, "global_step": 15222, "epoch": 422, "lr": 0.001} +{"train_loss": 0.030625086277723312, "train_loss_bc": 0.030625086277723312, "train_loss_llm": 0.0, "grad_norm": 0.0633183941245079, "global_step": 15223, "epoch": 422, "lr": 0.001} +{"train_loss": 0.029861580580472946, "train_loss_bc": 0.029861580580472946, "train_loss_llm": 0.0, "grad_norm": 0.07689660787582397, "global_step": 15224, "epoch": 422, "lr": 0.001} +{"train_loss": 0.031355030834674835, "train_loss_bc": 0.031355030834674835, "train_loss_llm": 0.0, "grad_norm": 0.023622864857316017, "global_step": 15225, "epoch": 422, "lr": 0.001} +{"train_loss": 0.0333806537091732, "train_loss_bc": 0.0333806537091732, "train_loss_llm": 0.0, "grad_norm": 0.03361840546131134, "global_step": 15226, "epoch": 422, "lr": 0.001} +{"train_loss": 0.03549372171983123, "train_loss_bc": 0.0348454974591732, "train_loss_llm": 0.0, "grad_norm": 0.05450951308012009, "global_step": 15227, "epoch": 422, "lr": 0.001, "val_loss": 0.06935422122478485} +{"train_loss": 0.030512453988194466, "train_loss_bc": 0.030512453988194466, "train_loss_llm": 0.0, "grad_norm": 0.06325828284025192, "global_step": 15228, "epoch": 423, "lr": 0.001} +{"train_loss": 0.030314132571220398, "train_loss_bc": 0.030314132571220398, "train_loss_llm": 0.0, "grad_norm": 0.02142924629151821, "global_step": 15229, "epoch": 423, "lr": 0.001} +{"train_loss": 0.0346975177526474, "train_loss_bc": 0.0346975177526474, "train_loss_llm": 0.0, "grad_norm": 0.03332536295056343, "global_step": 15230, "epoch": 423, "lr": 0.001} +{"train_loss": 0.033459339290857315, "train_loss_bc": 0.033459339290857315, "train_loss_llm": 0.0, "grad_norm": 0.05790761485695839, "global_step": 15231, "epoch": 423, "lr": 0.001} +{"train_loss": 0.04209352657198906, "train_loss_bc": 0.04209352657198906, "train_loss_llm": 0.0, "grad_norm": 0.0667862668633461, "global_step": 15232, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03915194794535637, "train_loss_bc": 0.03915194794535637, "train_loss_llm": 0.0, "grad_norm": 0.024455131962895393, "global_step": 15233, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03462628647685051, "train_loss_bc": 0.03462449833750725, "train_loss_llm": 0.00017893314361572266, "grad_norm": 0.034927334636449814, "global_step": 15234, "epoch": 423, "lr": 0.001} +{"train_loss": 0.02857566811144352, "train_loss_bc": 0.02857566811144352, "train_loss_llm": 0.0, "grad_norm": 0.0513015016913414, "global_step": 15235, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03618422895669937, "train_loss_bc": 0.03618182986974716, "train_loss_llm": 0.00023990869522094727, "grad_norm": 0.06085490435361862, "global_step": 15236, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03182598948478699, "train_loss_bc": 0.031820714473724365, "train_loss_llm": 0.0005274116992950439, "grad_norm": 0.026099462062120438, "global_step": 15237, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03887374699115753, "train_loss_bc": 0.03887374699115753, "train_loss_llm": 0.0, "grad_norm": 0.04031985253095627, "global_step": 15238, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03974065184593201, "train_loss_bc": 0.03974065184593201, "train_loss_llm": 0.0, "grad_norm": 0.05126480758190155, "global_step": 15239, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03697388991713524, "train_loss_bc": 0.03696725517511368, "train_loss_llm": 0.0006634294986724854, "grad_norm": 0.06899937242269516, "global_step": 15240, "epoch": 423, "lr": 0.001} +{"train_loss": 0.02750120498239994, "train_loss_bc": 0.02750120498239994, "train_loss_llm": 0.0, "grad_norm": 0.028160523623228073, "global_step": 15241, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03644011169672012, "train_loss_bc": 0.03644011169672012, "train_loss_llm": 0.0, "grad_norm": 0.04429108276963234, "global_step": 15242, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03425684571266174, "train_loss_bc": 0.03424911946058273, "train_loss_llm": 0.000772625207901001, "grad_norm": 0.06236674636602402, "global_step": 15243, "epoch": 423, "lr": 0.001} +{"train_loss": 0.039197199046611786, "train_loss_bc": 0.0391920730471611, "train_loss_llm": 0.0005125701427459717, "grad_norm": 0.07716874778270721, "global_step": 15244, "epoch": 423, "lr": 0.001} +{"train_loss": 0.02889305166900158, "train_loss_bc": 0.02889305166900158, "train_loss_llm": 0.0, "grad_norm": 0.027875198051333427, "global_step": 15245, "epoch": 423, "lr": 0.001} +{"train_loss": 0.04289625212550163, "train_loss_bc": 0.04289625212550163, "train_loss_llm": 0.0, "grad_norm": 0.04475823789834976, "global_step": 15246, "epoch": 423, "lr": 0.001} +{"train_loss": 0.02888578176498413, "train_loss_bc": 0.0288804043084383, "train_loss_llm": 0.0005378127098083496, "grad_norm": 0.05631113052368164, "global_step": 15247, "epoch": 423, "lr": 0.001} +{"train_loss": 0.035514917224645615, "train_loss_bc": 0.03549374267458916, "train_loss_llm": 0.002117365598678589, "grad_norm": 0.07344117015600204, "global_step": 15248, "epoch": 423, "lr": 0.001} +{"train_loss": 0.04502173885703087, "train_loss_bc": 0.04502173885703087, "train_loss_llm": 0.0, "grad_norm": 0.027247492223978043, "global_step": 15249, "epoch": 423, "lr": 0.001} +{"train_loss": 0.041512150317430496, "train_loss_bc": 0.041512150317430496, "train_loss_llm": 0.0, "grad_norm": 0.03352152556180954, "global_step": 15250, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03696603327989578, "train_loss_bc": 0.03696603327989578, "train_loss_llm": 0.0, "grad_norm": 0.05210035666823387, "global_step": 15251, "epoch": 423, "lr": 0.001} +{"train_loss": 0.037506353110075, "train_loss_bc": 0.037506353110075, "train_loss_llm": 0.0, "grad_norm": 0.0670686662197113, "global_step": 15252, "epoch": 423, "lr": 0.001} +{"train_loss": 0.0395309254527092, "train_loss_bc": 0.0395309254527092, "train_loss_llm": 0.0, "grad_norm": 0.028232458978891373, "global_step": 15253, "epoch": 423, "lr": 0.001} +{"train_loss": 0.036532435566186905, "train_loss_bc": 0.036532435566186905, "train_loss_llm": 0.0, "grad_norm": 0.037455964833498, "global_step": 15254, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03021964244544506, "train_loss_bc": 0.030201027169823647, "train_loss_llm": 0.0018614530563354492, "grad_norm": 0.04434193670749664, "global_step": 15255, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03460044786334038, "train_loss_bc": 0.034598659723997116, "train_loss_llm": 0.00017878413200378418, "grad_norm": 0.0538354329764843, "global_step": 15256, "epoch": 423, "lr": 0.001} +{"train_loss": 0.04223083704710007, "train_loss_bc": 0.042229216545820236, "train_loss_llm": 0.0001620948314666748, "grad_norm": 0.02764342911541462, "global_step": 15257, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03596765547990799, "train_loss_bc": 0.03596765547990799, "train_loss_llm": 0.0, "grad_norm": 0.03730817511677742, "global_step": 15258, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03190040960907936, "train_loss_bc": 0.03189508989453316, "train_loss_llm": 0.0005319714546203613, "grad_norm": 0.04939324036240578, "global_step": 15259, "epoch": 423, "lr": 0.001} +{"train_loss": 0.04115299507975578, "train_loss_bc": 0.04115299507975578, "train_loss_llm": 0.0, "grad_norm": 0.05413196235895157, "global_step": 15260, "epoch": 423, "lr": 0.001} +{"train_loss": 0.030942916870117188, "train_loss_bc": 0.030942916870117188, "train_loss_llm": 0.0, "grad_norm": 0.0202059093862772, "global_step": 15261, "epoch": 423, "lr": 0.001} +{"train_loss": 0.028680894523859024, "train_loss_bc": 0.028680894523859024, "train_loss_llm": 0.0, "grad_norm": 0.034483980387449265, "global_step": 15262, "epoch": 423, "lr": 0.001} +{"train_loss": 0.03547023397146, "train_loss_bc": 0.03354824334383011, "train_loss_llm": 0.0, "grad_norm": 0.049436621367931366, "global_step": 15263, "epoch": 423, "lr": 0.001, "val_loss": 0.06907428056001663} +{"train_loss": 0.03385157138109207, "train_loss_bc": 0.03385157138109207, "train_loss_llm": 0.0, "grad_norm": 0.06311304867267609, "global_step": 15264, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03704875707626343, "train_loss_bc": 0.03703531250357628, "train_loss_llm": 0.0013446211814880371, "grad_norm": 0.022374171763658524, "global_step": 15265, "epoch": 424, "lr": 0.001} +{"train_loss": 0.033992864191532135, "train_loss_bc": 0.033992864191532135, "train_loss_llm": 0.0, "grad_norm": 0.03665575012564659, "global_step": 15266, "epoch": 424, "lr": 0.001} +{"train_loss": 0.037261586636304855, "train_loss_bc": 0.037256937474012375, "train_loss_llm": 0.0004648864269256592, "grad_norm": 0.05989213287830353, "global_step": 15267, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03336462378501892, "train_loss_bc": 0.03335006162524223, "train_loss_llm": 0.0014563202857971191, "grad_norm": 0.07542301714420319, "global_step": 15268, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03394090384244919, "train_loss_bc": 0.03394090384244919, "train_loss_llm": 0.0, "grad_norm": 0.022134888917207718, "global_step": 15269, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03942466154694557, "train_loss_bc": 0.03942466154694557, "train_loss_llm": 0.0, "grad_norm": 0.03810320794582367, "global_step": 15270, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03900973126292229, "train_loss_bc": 0.03900973126292229, "train_loss_llm": 0.0, "grad_norm": 0.048827122896909714, "global_step": 15271, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03626617044210434, "train_loss_bc": 0.0362638421356678, "train_loss_llm": 0.00023278594017028809, "grad_norm": 0.056502941995859146, "global_step": 15272, "epoch": 424, "lr": 0.001} +{"train_loss": 0.04152009263634682, "train_loss_bc": 0.04152009263634682, "train_loss_llm": 0.0, "grad_norm": 0.024273401126265526, "global_step": 15273, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03278940171003342, "train_loss_bc": 0.03278940171003342, "train_loss_llm": 0.0, "grad_norm": 0.03940810263156891, "global_step": 15274, "epoch": 424, "lr": 0.001} +{"train_loss": 0.04452608525753021, "train_loss_bc": 0.04452608525753021, "train_loss_llm": 0.0, "grad_norm": 0.0667496770620346, "global_step": 15275, "epoch": 424, "lr": 0.001} +{"train_loss": 0.02935691736638546, "train_loss_bc": 0.02935691736638546, "train_loss_llm": 0.0, "grad_norm": 0.08015895634889603, "global_step": 15276, "epoch": 424, "lr": 0.001} +{"train_loss": 0.036460958421230316, "train_loss_bc": 0.036460958421230316, "train_loss_llm": 0.0, "grad_norm": 0.024322470650076866, "global_step": 15277, "epoch": 424, "lr": 0.001} +{"train_loss": 0.035078782588243484, "train_loss_bc": 0.035078782588243484, "train_loss_llm": 0.0, "grad_norm": 0.03362332284450531, "global_step": 15278, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03270648419857025, "train_loss_bc": 0.03270648419857025, "train_loss_llm": 0.0, "grad_norm": 0.046857938170433044, "global_step": 15279, "epoch": 424, "lr": 0.001} +{"train_loss": 0.042869698256254196, "train_loss_bc": 0.042869698256254196, "train_loss_llm": 0.0, "grad_norm": 0.06154724955558777, "global_step": 15280, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03185906633734703, "train_loss_bc": 0.03185906633734703, "train_loss_llm": 0.0, "grad_norm": 0.028432896360754967, "global_step": 15281, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03382056951522827, "train_loss_bc": 0.03382056951522827, "train_loss_llm": 0.0, "grad_norm": 0.03807661682367325, "global_step": 15282, "epoch": 424, "lr": 0.001} +{"train_loss": 0.037049103528261185, "train_loss_bc": 0.0370277464389801, "train_loss_llm": 0.002135723829269409, "grad_norm": 0.056311506778001785, "global_step": 15283, "epoch": 424, "lr": 0.001} +{"train_loss": 0.030908675864338875, "train_loss_bc": 0.030904680490493774, "train_loss_llm": 0.00039961934089660645, "grad_norm": 0.072598397731781, "global_step": 15284, "epoch": 424, "lr": 0.001} +{"train_loss": 0.035176727920770645, "train_loss_bc": 0.035176727920770645, "train_loss_llm": 0.0, "grad_norm": 0.027339965105056763, "global_step": 15285, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03786421939730644, "train_loss_bc": 0.03785271197557449, "train_loss_llm": 0.0011506974697113037, "grad_norm": 0.051667094230651855, "global_step": 15286, "epoch": 424, "lr": 0.001} +{"train_loss": 0.04121873527765274, "train_loss_bc": 0.04121873527765274, "train_loss_llm": 0.0, "grad_norm": 0.06299131363630295, "global_step": 15287, "epoch": 424, "lr": 0.001} +{"train_loss": 0.0338892936706543, "train_loss_bc": 0.0338892936706543, "train_loss_llm": 0.0, "grad_norm": 0.09188590198755264, "global_step": 15288, "epoch": 424, "lr": 0.001} +{"train_loss": 0.028629133477807045, "train_loss_bc": 0.028628379106521606, "train_loss_llm": 7.548928260803223e-05, "grad_norm": 0.022645309567451477, "global_step": 15289, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03604528680443764, "train_loss_bc": 0.03604528680443764, "train_loss_llm": 0.0, "grad_norm": 0.0388268381357193, "global_step": 15290, "epoch": 424, "lr": 0.001} +{"train_loss": 0.029717888683080673, "train_loss_bc": 0.029717888683080673, "train_loss_llm": 0.0, "grad_norm": 0.0581204928457737, "global_step": 15291, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03208489343523979, "train_loss_bc": 0.03208489343523979, "train_loss_llm": 0.0, "grad_norm": 0.07809068262577057, "global_step": 15292, "epoch": 424, "lr": 0.001} +{"train_loss": 0.036736875772476196, "train_loss_bc": 0.03672697767615318, "train_loss_llm": 0.000989675521850586, "grad_norm": 0.02510828897356987, "global_step": 15293, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03422167897224426, "train_loss_bc": 0.034195173531770706, "train_loss_llm": 0.0026504993438720703, "grad_norm": 0.03976515680551529, "global_step": 15294, "epoch": 424, "lr": 0.001} +{"train_loss": 0.04213355481624603, "train_loss_bc": 0.04213355481624603, "train_loss_llm": 0.0, "grad_norm": 0.05341615155339241, "global_step": 15295, "epoch": 424, "lr": 0.001} +{"train_loss": 0.034683872014284134, "train_loss_bc": 0.034683872014284134, "train_loss_llm": 0.0, "grad_norm": 0.07044558972120285, "global_step": 15296, "epoch": 424, "lr": 0.001} +{"train_loss": 0.03423977643251419, "train_loss_bc": 0.03423977643251419, "train_loss_llm": 0.0, "grad_norm": 0.028805239126086235, "global_step": 15297, "epoch": 424, "lr": 0.001} +{"train_loss": 0.038342177867889404, "train_loss_bc": 0.038340818136930466, "train_loss_llm": 0.0001360476016998291, "grad_norm": 0.04378847777843475, "global_step": 15298, "epoch": 424, "lr": 0.001} +{"train_loss": 0.035595059757017426, "train_loss_bc": 0.03333133086562157, "train_loss_llm": 0.0, "grad_norm": 0.05385158210992813, "global_step": 15299, "epoch": 424, "lr": 0.001, "val_loss": 0.06960906833410263} +{"train_loss": 0.027232998982071877, "train_loss_bc": 0.027230268344283104, "train_loss_llm": 0.00027304887771606445, "grad_norm": 0.06740617752075195, "global_step": 15300, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03335464000701904, "train_loss_bc": 0.03335464000701904, "train_loss_llm": 0.0, "grad_norm": 0.028816264122724533, "global_step": 15301, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03880087286233902, "train_loss_bc": 0.03880087286233902, "train_loss_llm": 0.0, "grad_norm": 0.04773363471031189, "global_step": 15302, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03797616437077522, "train_loss_bc": 0.03797616437077522, "train_loss_llm": 0.0, "grad_norm": 0.0649200975894928, "global_step": 15303, "epoch": 425, "lr": 0.001} +{"train_loss": 0.036706652492284775, "train_loss_bc": 0.036706652492284775, "train_loss_llm": 0.0, "grad_norm": 0.07312054932117462, "global_step": 15304, "epoch": 425, "lr": 0.001} +{"train_loss": 0.02854289673268795, "train_loss_bc": 0.02854289673268795, "train_loss_llm": 0.0, "grad_norm": 0.02069665677845478, "global_step": 15305, "epoch": 425, "lr": 0.001} +{"train_loss": 0.02908249758183956, "train_loss_bc": 0.02908249758183956, "train_loss_llm": 0.0, "grad_norm": 0.03039877861738205, "global_step": 15306, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03243387117981911, "train_loss_bc": 0.03243088722229004, "train_loss_llm": 0.0002982914447784424, "grad_norm": 0.0445624478161335, "global_step": 15307, "epoch": 425, "lr": 0.001} +{"train_loss": 0.04340332746505737, "train_loss_bc": 0.04340332746505737, "train_loss_llm": 0.0, "grad_norm": 0.06249268725514412, "global_step": 15308, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03385067731142044, "train_loss_bc": 0.033849623054265976, "train_loss_llm": 0.00010529160499572754, "grad_norm": 0.026010911911725998, "global_step": 15309, "epoch": 425, "lr": 0.001} +{"train_loss": 0.037237077951431274, "train_loss_bc": 0.03722483292222023, "train_loss_llm": 0.001224428415298462, "grad_norm": 0.03668811917304993, "global_step": 15310, "epoch": 425, "lr": 0.001} +{"train_loss": 0.037675946950912476, "train_loss_bc": 0.037675946950912476, "train_loss_llm": 0.0, "grad_norm": 0.0500105582177639, "global_step": 15311, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03430328890681267, "train_loss_bc": 0.03426989167928696, "train_loss_llm": 0.003339707851409912, "grad_norm": 0.06004595384001732, "global_step": 15312, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03719295188784599, "train_loss_bc": 0.0371871218085289, "train_loss_llm": 0.0005830526351928711, "grad_norm": 0.020067745819687843, "global_step": 15313, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03396795690059662, "train_loss_bc": 0.03396795690059662, "train_loss_llm": 0.0, "grad_norm": 0.03247927874326706, "global_step": 15314, "epoch": 425, "lr": 0.001} +{"train_loss": 0.027254173532128334, "train_loss_bc": 0.027254173532128334, "train_loss_llm": 0.0, "grad_norm": 0.04730352759361267, "global_step": 15315, "epoch": 425, "lr": 0.001} +{"train_loss": 0.036585476249456406, "train_loss_bc": 0.03657762333750725, "train_loss_llm": 0.0007852911949157715, "grad_norm": 0.05276821553707123, "global_step": 15316, "epoch": 425, "lr": 0.001} +{"train_loss": 0.04323473945260048, "train_loss_bc": 0.04322389140725136, "train_loss_llm": 0.001084834337234497, "grad_norm": 0.035993896424770355, "global_step": 15317, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03863271325826645, "train_loss_bc": 0.03863271325826645, "train_loss_llm": 0.0, "grad_norm": 0.04456420615315437, "global_step": 15318, "epoch": 425, "lr": 0.001} +{"train_loss": 0.033098552376031876, "train_loss_bc": 0.033098552376031876, "train_loss_llm": 0.0, "grad_norm": 0.060635216534137726, "global_step": 15319, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03230934962630272, "train_loss_bc": 0.03230934962630272, "train_loss_llm": 0.0, "grad_norm": 0.07281257212162018, "global_step": 15320, "epoch": 425, "lr": 0.001} +{"train_loss": 0.031092144548892975, "train_loss_bc": 0.03108157031238079, "train_loss_llm": 0.0010574758052825928, "grad_norm": 0.022513587027788162, "global_step": 15321, "epoch": 425, "lr": 0.001} +{"train_loss": 0.027852129191160202, "train_loss_bc": 0.027842212468385696, "train_loss_llm": 0.0009916424751281738, "grad_norm": 0.03804587945342064, "global_step": 15322, "epoch": 425, "lr": 0.001} +{"train_loss": 0.032971955835819244, "train_loss_bc": 0.032957155257463455, "train_loss_llm": 0.001480013132095337, "grad_norm": 0.05163571983575821, "global_step": 15323, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03415398299694061, "train_loss_bc": 0.03415398299694061, "train_loss_llm": 0.0, "grad_norm": 0.05717999115586281, "global_step": 15324, "epoch": 425, "lr": 0.001} +{"train_loss": 0.042529791593551636, "train_loss_bc": 0.042529791593551636, "train_loss_llm": 0.0, "grad_norm": 0.028649533167481422, "global_step": 15325, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03798878937959671, "train_loss_bc": 0.03798878937959671, "train_loss_llm": 0.0, "grad_norm": 0.03898386284708977, "global_step": 15326, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03552890941500664, "train_loss_bc": 0.035516753792762756, "train_loss_llm": 0.0012156665325164795, "grad_norm": 0.05730804428458214, "global_step": 15327, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03867210075259209, "train_loss_bc": 0.03867210075259209, "train_loss_llm": 0.0, "grad_norm": 0.06622175127267838, "global_step": 15328, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03942940756678581, "train_loss_bc": 0.03941816836595535, "train_loss_llm": 0.0011237561702728271, "grad_norm": 0.026865364983677864, "global_step": 15329, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03854100778698921, "train_loss_bc": 0.03854100778698921, "train_loss_llm": 0.0, "grad_norm": 0.04446680471301079, "global_step": 15330, "epoch": 425, "lr": 0.001} +{"train_loss": 0.035823602229356766, "train_loss_bc": 0.035823602229356766, "train_loss_llm": 0.0, "grad_norm": 0.057548608630895615, "global_step": 15331, "epoch": 425, "lr": 0.001} +{"train_loss": 0.041543785482645035, "train_loss_bc": 0.041543785482645035, "train_loss_llm": 0.0, "grad_norm": 0.08301005512475967, "global_step": 15332, "epoch": 425, "lr": 0.001} +{"train_loss": 0.033069510012865067, "train_loss_bc": 0.033069510012865067, "train_loss_llm": 0.0, "grad_norm": 0.018996963277459145, "global_step": 15333, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03892766311764717, "train_loss_bc": 0.03892766311764717, "train_loss_llm": 0.0, "grad_norm": 0.034091901034116745, "global_step": 15334, "epoch": 425, "lr": 0.001} +{"train_loss": 0.03549825358721945, "train_loss_bc": 0.03692888468503952, "train_loss_llm": 0.0006638467311859131, "grad_norm": 0.0457121878862381, "global_step": 15335, "epoch": 425, "lr": 0.001, "train/cumulative_reward": 109.44329243332349, "train/mean_score": 11.818021868075492, "train/success_rate": 0.2, "test/cumulative_reward": 42.76572600420033, "test/mean_score": 6.796200471441167, "test/success_rate": 0.16, "val_loss": 0.07010412961244583, "train_action_mse_error": 0.03835684806108475} +{"train_loss": 0.03692905232310295, "train_loss_bc": 0.03692905232310295, "train_loss_llm": 0.0, "grad_norm": 0.05464617908000946, "global_step": 15336, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03987615928053856, "train_loss_bc": 0.03987615928053856, "train_loss_llm": 0.0, "grad_norm": 0.02370642125606537, "global_step": 15337, "epoch": 426, "lr": 0.001} +{"train_loss": 0.037956491112709045, "train_loss_bc": 0.03791704401373863, "train_loss_llm": 0.0039446353912353516, "grad_norm": 0.04010331630706787, "global_step": 15338, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03263477608561516, "train_loss_bc": 0.03263477608561516, "train_loss_llm": 0.0, "grad_norm": 0.05487284064292908, "global_step": 15339, "epoch": 426, "lr": 0.001} +{"train_loss": 0.035029977560043335, "train_loss_bc": 0.035029977560043335, "train_loss_llm": 0.0, "grad_norm": 0.07018737494945526, "global_step": 15340, "epoch": 426, "lr": 0.001} +{"train_loss": 0.029799923300743103, "train_loss_bc": 0.029799923300743103, "train_loss_llm": 0.0, "grad_norm": 0.02543780207633972, "global_step": 15341, "epoch": 426, "lr": 0.001} +{"train_loss": 0.032760363072156906, "train_loss_bc": 0.032760363072156906, "train_loss_llm": 0.0, "grad_norm": 0.03704769164323807, "global_step": 15342, "epoch": 426, "lr": 0.001} +{"train_loss": 0.0355629064142704, "train_loss_bc": 0.0355629064142704, "train_loss_llm": 0.0, "grad_norm": 0.03875965252518654, "global_step": 15343, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03160473331809044, "train_loss_bc": 0.03158973529934883, "train_loss_llm": 0.0014996826648712158, "grad_norm": 0.05328255891799927, "global_step": 15344, "epoch": 426, "lr": 0.001} +{"train_loss": 0.036798469722270966, "train_loss_bc": 0.036792222410440445, "train_loss_llm": 0.0006248652935028076, "grad_norm": 0.03711216524243355, "global_step": 15345, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03274225816130638, "train_loss_bc": 0.03274225816130638, "train_loss_llm": 0.0, "grad_norm": 0.05508909001946449, "global_step": 15346, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03292020782828331, "train_loss_bc": 0.03292020782828331, "train_loss_llm": 0.0, "grad_norm": 0.06593198329210281, "global_step": 15347, "epoch": 426, "lr": 0.001} +{"train_loss": 0.027953259646892548, "train_loss_bc": 0.02795242704451084, "train_loss_llm": 8.317828178405762e-05, "grad_norm": 0.07966156303882599, "global_step": 15348, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03974871709942818, "train_loss_bc": 0.03974871709942818, "train_loss_llm": 0.0, "grad_norm": 0.028863007202744484, "global_step": 15349, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03390367329120636, "train_loss_bc": 0.03390367329120636, "train_loss_llm": 0.0, "grad_norm": 0.04154083505272865, "global_step": 15350, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03737262636423111, "train_loss_bc": 0.03737262636423111, "train_loss_llm": 0.0, "grad_norm": 0.05556308105587959, "global_step": 15351, "epoch": 426, "lr": 0.001} +{"train_loss": 0.030856233090162277, "train_loss_bc": 0.03084908239543438, "train_loss_llm": 0.0007150471210479736, "grad_norm": 0.06636615097522736, "global_step": 15352, "epoch": 426, "lr": 0.001} +{"train_loss": 0.036485303193330765, "train_loss_bc": 0.036485303193330765, "train_loss_llm": 0.0, "grad_norm": 0.02645385079085827, "global_step": 15353, "epoch": 426, "lr": 0.001} +{"train_loss": 0.037808842957019806, "train_loss_bc": 0.037808842957019806, "train_loss_llm": 0.0, "grad_norm": 0.04041343927383423, "global_step": 15354, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03705336153507233, "train_loss_bc": 0.03705336153507233, "train_loss_llm": 0.0, "grad_norm": 0.04884523153305054, "global_step": 15355, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03518622741103172, "train_loss_bc": 0.03518622741103172, "train_loss_llm": 0.0, "grad_norm": 0.06199740990996361, "global_step": 15356, "epoch": 426, "lr": 0.001} +{"train_loss": 0.040205858647823334, "train_loss_bc": 0.040205858647823334, "train_loss_llm": 0.0, "grad_norm": 0.02713071182370186, "global_step": 15357, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03833348676562309, "train_loss_bc": 0.03833348676562309, "train_loss_llm": 0.0, "grad_norm": 0.04485774785280228, "global_step": 15358, "epoch": 426, "lr": 0.001} +{"train_loss": 0.0325230248272419, "train_loss_bc": 0.0325230248272419, "train_loss_llm": 0.0, "grad_norm": 0.052981093525886536, "global_step": 15359, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03888668492436409, "train_loss_bc": 0.03888668492436409, "train_loss_llm": 0.0, "grad_norm": 0.07103150337934494, "global_step": 15360, "epoch": 426, "lr": 0.001} +{"train_loss": 0.032782603055238724, "train_loss_bc": 0.032782603055238724, "train_loss_llm": 0.0, "grad_norm": 0.02433866821229458, "global_step": 15361, "epoch": 426, "lr": 0.001} +{"train_loss": 0.032341182231903076, "train_loss_bc": 0.032341182231903076, "train_loss_llm": 0.0, "grad_norm": 0.03646445646882057, "global_step": 15362, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03857797011733055, "train_loss_bc": 0.03857797011733055, "train_loss_llm": 0.0, "grad_norm": 0.05170990154147148, "global_step": 15363, "epoch": 426, "lr": 0.001} +{"train_loss": 0.037746626883745193, "train_loss_bc": 0.03774486854672432, "train_loss_llm": 0.00017565488815307617, "grad_norm": 0.06373641639947891, "global_step": 15364, "epoch": 426, "lr": 0.001} +{"train_loss": 0.029937617480754852, "train_loss_bc": 0.029937617480754852, "train_loss_llm": 0.0, "grad_norm": 0.016248585656285286, "global_step": 15365, "epoch": 426, "lr": 0.001} +{"train_loss": 0.04087773337960243, "train_loss_bc": 0.04087177291512489, "train_loss_llm": 0.0005960464477539062, "grad_norm": 0.03514394536614418, "global_step": 15366, "epoch": 426, "lr": 0.001} +{"train_loss": 0.028237713500857353, "train_loss_bc": 0.028220094740390778, "train_loss_llm": 0.0017618834972381592, "grad_norm": 0.04450943320989609, "global_step": 15367, "epoch": 426, "lr": 0.001} +{"train_loss": 0.03932426497340202, "train_loss_bc": 0.03931610286235809, "train_loss_llm": 0.0008162558078765869, "grad_norm": 0.06008673459291458, "global_step": 15368, "epoch": 426, "lr": 0.001} +{"train_loss": 0.04045354947447777, "train_loss_bc": 0.04045354947447777, "train_loss_llm": 0.0, "grad_norm": 0.02495839074254036, "global_step": 15369, "epoch": 426, "lr": 0.001} +{"train_loss": 0.039665475487709045, "train_loss_bc": 0.03965416178107262, "train_loss_llm": 0.001131296157836914, "grad_norm": 0.04720816761255264, "global_step": 15370, "epoch": 426, "lr": 0.001} +{"train_loss": 0.035454695475184254, "train_loss_bc": 0.03549168258905411, "train_loss_llm": 0.0, "grad_norm": 0.05382584035396576, "global_step": 15371, "epoch": 426, "lr": 0.001, "val_loss": 0.0700724869966507} +{"train_loss": 0.035099517554044724, "train_loss_bc": 0.0350920669734478, "train_loss_llm": 0.0007451474666595459, "grad_norm": 0.05624021962285042, "global_step": 15372, "epoch": 427, "lr": 0.001} +{"train_loss": 0.026198089122772217, "train_loss_bc": 0.026198089122772217, "train_loss_llm": 0.0, "grad_norm": 0.021517179906368256, "global_step": 15373, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03253316506743431, "train_loss_bc": 0.03252356871962547, "train_loss_llm": 0.0009596347808837891, "grad_norm": 0.03362544625997543, "global_step": 15374, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03968634828925133, "train_loss_bc": 0.03968634828925133, "train_loss_llm": 0.0, "grad_norm": 0.05234779044985771, "global_step": 15375, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03860514611005783, "train_loss_bc": 0.038594670593738556, "train_loss_llm": 0.0010477006435394287, "grad_norm": 0.06766872107982635, "global_step": 15376, "epoch": 427, "lr": 0.001} +{"train_loss": 0.036720581352710724, "train_loss_bc": 0.036720581352710724, "train_loss_llm": 0.0, "grad_norm": 0.02329406701028347, "global_step": 15377, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03740335628390312, "train_loss_bc": 0.03740335628390312, "train_loss_llm": 0.0, "grad_norm": 0.03072301670908928, "global_step": 15378, "epoch": 427, "lr": 0.001} +{"train_loss": 0.031306829303503036, "train_loss_bc": 0.031306829303503036, "train_loss_llm": 0.0, "grad_norm": 0.03575132042169571, "global_step": 15379, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03462057560682297, "train_loss_bc": 0.03462057560682297, "train_loss_llm": 0.0, "grad_norm": 0.04549464210867882, "global_step": 15380, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03607320040464401, "train_loss_bc": 0.03607320040464401, "train_loss_llm": 0.0, "grad_norm": 0.02124026231467724, "global_step": 15381, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03059728443622589, "train_loss_bc": 0.03059728443622589, "train_loss_llm": 0.0, "grad_norm": 0.033992473036050797, "global_step": 15382, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03305671736598015, "train_loss_bc": 0.03305671736598015, "train_loss_llm": 0.0, "grad_norm": 0.04203315079212189, "global_step": 15383, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03218184411525726, "train_loss_bc": 0.03218184411525726, "train_loss_llm": 0.0, "grad_norm": 0.054818496108055115, "global_step": 15384, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03236553817987442, "train_loss_bc": 0.03236553817987442, "train_loss_llm": 0.0, "grad_norm": 0.024242429062724113, "global_step": 15385, "epoch": 427, "lr": 0.001} +{"train_loss": 0.04032733663916588, "train_loss_bc": 0.04032733663916588, "train_loss_llm": 0.0, "grad_norm": 0.03867504373192787, "global_step": 15386, "epoch": 427, "lr": 0.001} +{"train_loss": 0.032668814063072205, "train_loss_bc": 0.032668814063072205, "train_loss_llm": 0.0, "grad_norm": 0.046135950833559036, "global_step": 15387, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03936275467276573, "train_loss_bc": 0.03936275467276573, "train_loss_llm": 0.0, "grad_norm": 0.06989714503288269, "global_step": 15388, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03779594600200653, "train_loss_bc": 0.03778364509344101, "train_loss_llm": 0.0012300610542297363, "grad_norm": 0.026920277625322342, "global_step": 15389, "epoch": 427, "lr": 0.001} +{"train_loss": 0.02879575826227665, "train_loss_bc": 0.02879575826227665, "train_loss_llm": 0.0, "grad_norm": 0.034069906920194626, "global_step": 15390, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03813256695866585, "train_loss_bc": 0.03811565041542053, "train_loss_llm": 0.0016915202140808105, "grad_norm": 0.04671509563922882, "global_step": 15391, "epoch": 427, "lr": 0.001} +{"train_loss": 0.030145322903990746, "train_loss_bc": 0.03013579174876213, "train_loss_llm": 0.0009530484676361084, "grad_norm": 0.05524882301688194, "global_step": 15392, "epoch": 427, "lr": 0.001} +{"train_loss": 0.04098108410835266, "train_loss_bc": 0.04098108410835266, "train_loss_llm": 0.0, "grad_norm": 0.02425776794552803, "global_step": 15393, "epoch": 427, "lr": 0.001} +{"train_loss": 0.035450778901576996, "train_loss_bc": 0.035450778901576996, "train_loss_llm": 0.0, "grad_norm": 0.04136168956756592, "global_step": 15394, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03062119334936142, "train_loss_bc": 0.03062119334936142, "train_loss_llm": 0.0, "grad_norm": 0.05451379716396332, "global_step": 15395, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03934316337108612, "train_loss_bc": 0.03934316337108612, "train_loss_llm": 0.0, "grad_norm": 0.06819227337837219, "global_step": 15396, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03253978118300438, "train_loss_bc": 0.03251300007104874, "train_loss_llm": 0.0026782453060150146, "grad_norm": 0.02453756518661976, "global_step": 15397, "epoch": 427, "lr": 0.001} +{"train_loss": 0.034650105983018875, "train_loss_bc": 0.03464158624410629, "train_loss_llm": 0.0008520185947418213, "grad_norm": 0.044588517397642136, "global_step": 15398, "epoch": 427, "lr": 0.001} +{"train_loss": 0.04163047671318054, "train_loss_bc": 0.04163047671318054, "train_loss_llm": 0.0, "grad_norm": 0.06207284703850746, "global_step": 15399, "epoch": 427, "lr": 0.001} +{"train_loss": 0.029845979064702988, "train_loss_bc": 0.029845979064702988, "train_loss_llm": 0.0, "grad_norm": 0.06872682273387909, "global_step": 15400, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03394196927547455, "train_loss_bc": 0.03394196927547455, "train_loss_llm": 0.0, "grad_norm": 0.019102411344647408, "global_step": 15401, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03675404191017151, "train_loss_bc": 0.03675404191017151, "train_loss_llm": 0.0, "grad_norm": 0.032989751547575, "global_step": 15402, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03872930258512497, "train_loss_bc": 0.038727056235075, "train_loss_llm": 0.00022462010383605957, "grad_norm": 0.04598473757505417, "global_step": 15403, "epoch": 427, "lr": 0.001} +{"train_loss": 0.04185406491160393, "train_loss_bc": 0.04185406491160393, "train_loss_llm": 0.0, "grad_norm": 0.059832096099853516, "global_step": 15404, "epoch": 427, "lr": 0.001} +{"train_loss": 0.04077785089612007, "train_loss_bc": 0.04077785089612007, "train_loss_llm": 0.0, "grad_norm": 0.025887073948979378, "global_step": 15405, "epoch": 427, "lr": 0.001} +{"train_loss": 0.03393455967307091, "train_loss_bc": 0.03392329439520836, "train_loss_llm": 0.0011264383792877197, "grad_norm": 0.03388194739818573, "global_step": 15406, "epoch": 427, "lr": 0.001} +{"train_loss": 0.0353099981115924, "train_loss_bc": 0.03642888739705086, "train_loss_llm": 0.0, "grad_norm": 0.05094940587878227, "global_step": 15407, "epoch": 427, "lr": 0.001, "val_loss": 0.06972067058086395} +{"train_loss": 0.03573090210556984, "train_loss_bc": 0.035717207938432693, "train_loss_llm": 0.001369386911392212, "grad_norm": 0.06576644629240036, "global_step": 15408, "epoch": 428, "lr": 0.001} +{"train_loss": 0.036942675709724426, "train_loss_bc": 0.036942675709724426, "train_loss_llm": 0.0, "grad_norm": 0.02281852625310421, "global_step": 15409, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03409534692764282, "train_loss_bc": 0.03409534692764282, "train_loss_llm": 0.0, "grad_norm": 0.03258392959833145, "global_step": 15410, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03641234338283539, "train_loss_bc": 0.03641234338283539, "train_loss_llm": 0.0, "grad_norm": 0.042113762348890305, "global_step": 15411, "epoch": 428, "lr": 0.001} +{"train_loss": 0.035791635513305664, "train_loss_bc": 0.03578623756766319, "train_loss_llm": 0.0005398392677307129, "grad_norm": 0.0526118241250515, "global_step": 15412, "epoch": 428, "lr": 0.001} +{"train_loss": 0.02593919448554516, "train_loss_bc": 0.02593059465289116, "train_loss_llm": 0.0008599162101745605, "grad_norm": 0.026051336899399757, "global_step": 15413, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03264831006526947, "train_loss_bc": 0.03264831006526947, "train_loss_llm": 0.0, "grad_norm": 0.043196000158786774, "global_step": 15414, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03937548026442528, "train_loss_bc": 0.039366714656353, "train_loss_llm": 0.0008763968944549561, "grad_norm": 0.05152739956974983, "global_step": 15415, "epoch": 428, "lr": 0.001} +{"train_loss": 0.039084818214178085, "train_loss_bc": 0.039084818214178085, "train_loss_llm": 0.0, "grad_norm": 0.057040099054574966, "global_step": 15416, "epoch": 428, "lr": 0.001} +{"train_loss": 0.039123132824897766, "train_loss_bc": 0.039123132824897766, "train_loss_llm": 0.0, "grad_norm": 0.026078542694449425, "global_step": 15417, "epoch": 428, "lr": 0.001} +{"train_loss": 0.036217495799064636, "train_loss_bc": 0.036217495799064636, "train_loss_llm": 0.0, "grad_norm": 0.040802326053380966, "global_step": 15418, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03565681725740433, "train_loss_bc": 0.03565681725740433, "train_loss_llm": 0.0, "grad_norm": 0.04991250857710838, "global_step": 15419, "epoch": 428, "lr": 0.001} +{"train_loss": 0.029643816873431206, "train_loss_bc": 0.029643816873431206, "train_loss_llm": 0.0, "grad_norm": 0.05882740765810013, "global_step": 15420, "epoch": 428, "lr": 0.001} +{"train_loss": 0.037326034158468246, "train_loss_bc": 0.037326034158468246, "train_loss_llm": 0.0, "grad_norm": 0.02353953756392002, "global_step": 15421, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03829924389719963, "train_loss_bc": 0.03829924389719963, "train_loss_llm": 0.0, "grad_norm": 0.03855309635400772, "global_step": 15422, "epoch": 428, "lr": 0.001} +{"train_loss": 0.035854771733284, "train_loss_bc": 0.035854771733284, "train_loss_llm": 0.0, "grad_norm": 0.04590797424316406, "global_step": 15423, "epoch": 428, "lr": 0.001} +{"train_loss": 0.035074420273303986, "train_loss_bc": 0.035074420273303986, "train_loss_llm": 0.0, "grad_norm": 0.05631807819008827, "global_step": 15424, "epoch": 428, "lr": 0.001} +{"train_loss": 0.026506099849939346, "train_loss_bc": 0.026506099849939346, "train_loss_llm": 0.0, "grad_norm": 0.016623016446828842, "global_step": 15425, "epoch": 428, "lr": 0.001} +{"train_loss": 0.040234897285699844, "train_loss_bc": 0.040230270475149155, "train_loss_llm": 0.0004627704620361328, "grad_norm": 0.03166254609823227, "global_step": 15426, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03800312802195549, "train_loss_bc": 0.03800312802195549, "train_loss_llm": 0.0, "grad_norm": 0.04113024100661278, "global_step": 15427, "epoch": 428, "lr": 0.001} +{"train_loss": 0.033542584627866745, "train_loss_bc": 0.03353061154484749, "train_loss_llm": 0.001197129487991333, "grad_norm": 0.052909739315509796, "global_step": 15428, "epoch": 428, "lr": 0.001} +{"train_loss": 0.04261144623160362, "train_loss_bc": 0.04260780289769173, "train_loss_llm": 0.0003644227981567383, "grad_norm": 0.03040187619626522, "global_step": 15429, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03335488587617874, "train_loss_bc": 0.03335488587617874, "train_loss_llm": 0.0, "grad_norm": 0.036874186247587204, "global_step": 15430, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03825334459543228, "train_loss_bc": 0.03825334459543228, "train_loss_llm": 0.0, "grad_norm": 0.05450047180056572, "global_step": 15431, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03504104167222977, "train_loss_bc": 0.035029202699661255, "train_loss_llm": 0.0011839866638183594, "grad_norm": 0.06802243739366531, "global_step": 15432, "epoch": 428, "lr": 0.001} +{"train_loss": 0.030934525653719902, "train_loss_bc": 0.030934525653719902, "train_loss_llm": 0.0, "grad_norm": 0.01960260421037674, "global_step": 15433, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03555035591125488, "train_loss_bc": 0.03555035591125488, "train_loss_llm": 0.0, "grad_norm": 0.03443099930882454, "global_step": 15434, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03163591027259827, "train_loss_bc": 0.03163591027259827, "train_loss_llm": 0.0, "grad_norm": 0.0461907796561718, "global_step": 15435, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03525126352906227, "train_loss_bc": 0.03523590415716171, "train_loss_llm": 0.0015360116958618164, "grad_norm": 0.06341668218374252, "global_step": 15436, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03472574055194855, "train_loss_bc": 0.034718189388513565, "train_loss_llm": 0.0007549822330474854, "grad_norm": 0.027183817699551582, "global_step": 15437, "epoch": 428, "lr": 0.001} +{"train_loss": 0.033773645758628845, "train_loss_bc": 0.03375348448753357, "train_loss_llm": 0.0020162761211395264, "grad_norm": 0.038202084600925446, "global_step": 15438, "epoch": 428, "lr": 0.001} +{"train_loss": 0.0355028472840786, "train_loss_bc": 0.0355028472840786, "train_loss_llm": 0.0, "grad_norm": 0.04233694076538086, "global_step": 15439, "epoch": 428, "lr": 0.001} +{"train_loss": 0.02726106531918049, "train_loss_bc": 0.02726106531918049, "train_loss_llm": 0.0, "grad_norm": 0.04808437451720238, "global_step": 15440, "epoch": 428, "lr": 0.001} +{"train_loss": 0.041142575442790985, "train_loss_bc": 0.041142575442790985, "train_loss_llm": 0.0, "grad_norm": 0.022839287295937538, "global_step": 15441, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03857967257499695, "train_loss_bc": 0.03857967257499695, "train_loss_llm": 0.0, "grad_norm": 0.030630536377429962, "global_step": 15442, "epoch": 428, "lr": 0.001} +{"train_loss": 0.03521181415352556, "train_loss_bc": 0.03250383958220482, "train_loss_llm": 0.0, "grad_norm": 0.0447176992893219, "global_step": 15443, "epoch": 428, "lr": 0.001, "val_loss": 0.06922556459903717} +{"train_loss": 0.034856416285037994, "train_loss_bc": 0.034840863198041916, "train_loss_llm": 0.0015553534030914307, "grad_norm": 0.054415978491306305, "global_step": 15444, "epoch": 429, "lr": 0.001} +{"train_loss": 0.032945264130830765, "train_loss_bc": 0.032945264130830765, "train_loss_llm": 0.0, "grad_norm": 0.023713132366538048, "global_step": 15445, "epoch": 429, "lr": 0.001} +{"train_loss": 0.032253216952085495, "train_loss_bc": 0.03224846348166466, "train_loss_llm": 0.00047522783279418945, "grad_norm": 0.03168543055653572, "global_step": 15446, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03528888523578644, "train_loss_bc": 0.03526167944073677, "train_loss_llm": 0.002720564603805542, "grad_norm": 0.04195912182331085, "global_step": 15447, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03381212800741196, "train_loss_bc": 0.03381212800741196, "train_loss_llm": 0.0, "grad_norm": 0.05528079718351364, "global_step": 15448, "epoch": 429, "lr": 0.001} +{"train_loss": 0.032644130289554596, "train_loss_bc": 0.032644130289554596, "train_loss_llm": 0.0, "grad_norm": 0.020280666649341583, "global_step": 15449, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03977319598197937, "train_loss_bc": 0.03977319598197937, "train_loss_llm": 0.0, "grad_norm": 0.04001184552907944, "global_step": 15450, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03705377131700516, "train_loss_bc": 0.037025146186351776, "train_loss_llm": 0.0028624534606933594, "grad_norm": 0.04699363559484482, "global_step": 15451, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03285409137606621, "train_loss_bc": 0.03285409137606621, "train_loss_llm": 0.0, "grad_norm": 0.053386326879262924, "global_step": 15452, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03143533319234848, "train_loss_bc": 0.03143533319234848, "train_loss_llm": 0.0, "grad_norm": 0.021511683240532875, "global_step": 15453, "epoch": 429, "lr": 0.001} +{"train_loss": 0.04051332175731659, "train_loss_bc": 0.04051332175731659, "train_loss_llm": 0.0, "grad_norm": 0.033797528594732285, "global_step": 15454, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03345276787877083, "train_loss_bc": 0.03345276787877083, "train_loss_llm": 0.0, "grad_norm": 0.040397997945547104, "global_step": 15455, "epoch": 429, "lr": 0.001} +{"train_loss": 0.034593917429447174, "train_loss_bc": 0.034593917429447174, "train_loss_llm": 0.0, "grad_norm": 0.046227551996707916, "global_step": 15456, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03845283389091492, "train_loss_bc": 0.03844345360994339, "train_loss_llm": 0.0009379684925079346, "grad_norm": 0.017929324880242348, "global_step": 15457, "epoch": 429, "lr": 0.001} +{"train_loss": 0.029790960252285004, "train_loss_bc": 0.029770951718091965, "train_loss_llm": 0.0020008981227874756, "grad_norm": 0.03061920404434204, "global_step": 15458, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03755412995815277, "train_loss_bc": 0.03755412995815277, "train_loss_llm": 0.0, "grad_norm": 0.04909379780292511, "global_step": 15459, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03677689656615257, "train_loss_bc": 0.03677689656615257, "train_loss_llm": 0.0, "grad_norm": 0.06037711352109909, "global_step": 15460, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03942251577973366, "train_loss_bc": 0.03942251577973366, "train_loss_llm": 0.0, "grad_norm": 0.02010323666036129, "global_step": 15461, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03136085718870163, "train_loss_bc": 0.03136085718870163, "train_loss_llm": 0.0, "grad_norm": 0.0283367820084095, "global_step": 15462, "epoch": 429, "lr": 0.001} +{"train_loss": 0.029062122106552124, "train_loss_bc": 0.029062122106552124, "train_loss_llm": 0.0, "grad_norm": 0.0390576496720314, "global_step": 15463, "epoch": 429, "lr": 0.001} +{"train_loss": 0.031181735917925835, "train_loss_bc": 0.031181735917925835, "train_loss_llm": 0.0, "grad_norm": 0.045237865298986435, "global_step": 15464, "epoch": 429, "lr": 0.001} +{"train_loss": 0.037124570459127426, "train_loss_bc": 0.037124570459127426, "train_loss_llm": 0.0, "grad_norm": 0.025157848373055458, "global_step": 15465, "epoch": 429, "lr": 0.001} +{"train_loss": 0.034423187375068665, "train_loss_bc": 0.034423187375068665, "train_loss_llm": 0.0, "grad_norm": 0.03534431383013725, "global_step": 15466, "epoch": 429, "lr": 0.001} +{"train_loss": 0.030979886651039124, "train_loss_bc": 0.03097708709537983, "train_loss_llm": 0.00027993321418762207, "grad_norm": 0.048022035509347916, "global_step": 15467, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03862995281815529, "train_loss_bc": 0.038618072867393494, "train_loss_llm": 0.0011878907680511475, "grad_norm": 0.05701453238725662, "global_step": 15468, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03473706915974617, "train_loss_bc": 0.034731220453977585, "train_loss_llm": 0.0005849003791809082, "grad_norm": 0.020515846088528633, "global_step": 15469, "epoch": 429, "lr": 0.001} +{"train_loss": 0.0290748942643404, "train_loss_bc": 0.029067395254969597, "train_loss_llm": 0.0007498860359191895, "grad_norm": 0.03323546424508095, "global_step": 15470, "epoch": 429, "lr": 0.001} +{"train_loss": 0.039733946323394775, "train_loss_bc": 0.039725739508867264, "train_loss_llm": 0.0008207261562347412, "grad_norm": 0.04817475005984306, "global_step": 15471, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03800319507718086, "train_loss_bc": 0.03800319507718086, "train_loss_llm": 0.0, "grad_norm": 0.0643695592880249, "global_step": 15472, "epoch": 429, "lr": 0.001} +{"train_loss": 0.04045405611395836, "train_loss_bc": 0.04045405611395836, "train_loss_llm": 0.0, "grad_norm": 0.023613285273313522, "global_step": 15473, "epoch": 429, "lr": 0.001} +{"train_loss": 0.029305120930075645, "train_loss_bc": 0.029305120930075645, "train_loss_llm": 0.0, "grad_norm": 0.034236468374729156, "global_step": 15474, "epoch": 429, "lr": 0.001} +{"train_loss": 0.039314065128564835, "train_loss_bc": 0.03931330144405365, "train_loss_llm": 7.63237476348877e-05, "grad_norm": 0.04973117262125015, "global_step": 15475, "epoch": 429, "lr": 0.001} +{"train_loss": 0.0362648107111454, "train_loss_bc": 0.0362648107111454, "train_loss_llm": 0.0, "grad_norm": 0.05584972724318504, "global_step": 15476, "epoch": 429, "lr": 0.001} +{"train_loss": 0.036972057074308395, "train_loss_bc": 0.036972057074308395, "train_loss_llm": 0.0, "grad_norm": 0.030010217800736427, "global_step": 15477, "epoch": 429, "lr": 0.001} +{"train_loss": 0.036566365510225296, "train_loss_bc": 0.036566365510225296, "train_loss_llm": 0.0, "grad_norm": 0.04110021889209747, "global_step": 15478, "epoch": 429, "lr": 0.001} +{"train_loss": 0.03523370111361146, "train_loss_bc": 0.041751570999622345, "train_loss_llm": 0.0, "grad_norm": 0.050356730818748474, "global_step": 15479, "epoch": 429, "lr": 0.001, "val_loss": 0.0706055536866188} +{"train_loss": 0.035993438214063644, "train_loss_bc": 0.035993438214063644, "train_loss_llm": 0.0, "grad_norm": 0.06105746701359749, "global_step": 15480, "epoch": 430, "lr": 0.001} +{"train_loss": 0.036112915724515915, "train_loss_bc": 0.036112915724515915, "train_loss_llm": 0.0, "grad_norm": 0.02473670057952404, "global_step": 15481, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03099309653043747, "train_loss_bc": 0.030992696061730385, "train_loss_llm": 3.999471664428711e-05, "grad_norm": 0.034268394112586975, "global_step": 15482, "epoch": 430, "lr": 0.001} +{"train_loss": 0.02921733632683754, "train_loss_bc": 0.02921733632683754, "train_loss_llm": 0.0, "grad_norm": 0.04664942994713783, "global_step": 15483, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03676334023475647, "train_loss_bc": 0.03676220774650574, "train_loss_llm": 0.00011339783668518066, "grad_norm": 0.05306972190737724, "global_step": 15484, "epoch": 430, "lr": 0.001} +{"train_loss": 0.028501225635409355, "train_loss_bc": 0.028501225635409355, "train_loss_llm": 0.0, "grad_norm": 0.021496722474694252, "global_step": 15485, "epoch": 430, "lr": 0.001} +{"train_loss": 0.036712899804115295, "train_loss_bc": 0.0367087684571743, "train_loss_llm": 0.0004132986068725586, "grad_norm": 0.03470054641366005, "global_step": 15486, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03367013856768608, "train_loss_bc": 0.03367013856768608, "train_loss_llm": 0.0, "grad_norm": 0.03895745053887367, "global_step": 15487, "epoch": 430, "lr": 0.001} +{"train_loss": 0.027895772829651833, "train_loss_bc": 0.027895772829651833, "train_loss_llm": 0.0, "grad_norm": 0.04475728049874306, "global_step": 15488, "epoch": 430, "lr": 0.001} +{"train_loss": 0.036549415439367294, "train_loss_bc": 0.03654608502984047, "train_loss_llm": 0.00033289194107055664, "grad_norm": 0.024568073451519012, "global_step": 15489, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03635305538773537, "train_loss_bc": 0.03634943813085556, "train_loss_llm": 0.0003617703914642334, "grad_norm": 0.03900215029716492, "global_step": 15490, "epoch": 430, "lr": 0.001} +{"train_loss": 0.035112857818603516, "train_loss_bc": 0.035112857818603516, "train_loss_llm": 0.0, "grad_norm": 0.052798353135585785, "global_step": 15491, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03321606665849686, "train_loss_bc": 0.03321606665849686, "train_loss_llm": 0.0, "grad_norm": 0.06482485681772232, "global_step": 15492, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03798540681600571, "train_loss_bc": 0.03798540681600571, "train_loss_llm": 0.0, "grad_norm": 0.025699637830257416, "global_step": 15493, "epoch": 430, "lr": 0.001} +{"train_loss": 0.036007922142744064, "train_loss_bc": 0.036007922142744064, "train_loss_llm": 0.0, "grad_norm": 0.0390993095934391, "global_step": 15494, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03301896154880524, "train_loss_bc": 0.03301896154880524, "train_loss_llm": 0.0, "grad_norm": 0.052462138235569, "global_step": 15495, "epoch": 430, "lr": 0.001} +{"train_loss": 0.032157547771930695, "train_loss_bc": 0.032157547771930695, "train_loss_llm": 0.0, "grad_norm": 0.06281855702400208, "global_step": 15496, "epoch": 430, "lr": 0.001} +{"train_loss": 0.039255447685718536, "train_loss_bc": 0.039255447685718536, "train_loss_llm": 0.0, "grad_norm": 0.02705325558781624, "global_step": 15497, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03633004054427147, "train_loss_bc": 0.03633004054427147, "train_loss_llm": 0.0, "grad_norm": 0.04042930155992508, "global_step": 15498, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03244483470916748, "train_loss_bc": 0.03244483470916748, "train_loss_llm": 0.0, "grad_norm": 0.053667932748794556, "global_step": 15499, "epoch": 430, "lr": 0.001} +{"train_loss": 0.04087122529745102, "train_loss_bc": 0.04087122529745102, "train_loss_llm": 0.0, "grad_norm": 0.06453971564769745, "global_step": 15500, "epoch": 430, "lr": 0.001} +{"train_loss": 0.036562755703926086, "train_loss_bc": 0.036562755703926086, "train_loss_llm": 0.0, "grad_norm": 0.029925629496574402, "global_step": 15501, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03241075575351715, "train_loss_bc": 0.03241075575351715, "train_loss_llm": 0.0, "grad_norm": 0.03736448287963867, "global_step": 15502, "epoch": 430, "lr": 0.001} +{"train_loss": 0.036522313952445984, "train_loss_bc": 0.036517851054668427, "train_loss_llm": 0.0004463493824005127, "grad_norm": 0.048694439232349396, "global_step": 15503, "epoch": 430, "lr": 0.001} +{"train_loss": 0.035624705255031586, "train_loss_bc": 0.035624705255031586, "train_loss_llm": 0.0, "grad_norm": 0.06263397634029388, "global_step": 15504, "epoch": 430, "lr": 0.001} +{"train_loss": 0.04013034328818321, "train_loss_bc": 0.04013034328818321, "train_loss_llm": 0.0, "grad_norm": 0.02431112714111805, "global_step": 15505, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03487710654735565, "train_loss_bc": 0.034872960299253464, "train_loss_llm": 0.0004145205020904541, "grad_norm": 0.03664274513721466, "global_step": 15506, "epoch": 430, "lr": 0.001} +{"train_loss": 0.0340983085334301, "train_loss_bc": 0.0340983085334301, "train_loss_llm": 0.0, "grad_norm": 0.050493162125349045, "global_step": 15507, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03477993980050087, "train_loss_bc": 0.03477993980050087, "train_loss_llm": 0.0, "grad_norm": 0.06160258874297142, "global_step": 15508, "epoch": 430, "lr": 0.001} +{"train_loss": 0.036057669669389725, "train_loss_bc": 0.036057669669389725, "train_loss_llm": 0.0, "grad_norm": 0.02810010313987732, "global_step": 15509, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03827979415655136, "train_loss_bc": 0.038279008120298386, "train_loss_llm": 7.843971252441406e-05, "grad_norm": 0.04824893921613693, "global_step": 15510, "epoch": 430, "lr": 0.001} +{"train_loss": 0.034777916967868805, "train_loss_bc": 0.034777916967868805, "train_loss_llm": 0.0, "grad_norm": 0.06216305494308472, "global_step": 15511, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03972789645195007, "train_loss_bc": 0.03972789645195007, "train_loss_llm": 0.0, "grad_norm": 0.06870977580547333, "global_step": 15512, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03655596822500229, "train_loss_bc": 0.03655596822500229, "train_loss_llm": 0.0, "grad_norm": 0.022971659898757935, "global_step": 15513, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03500565141439438, "train_loss_bc": 0.03500565141439438, "train_loss_llm": 0.0, "grad_norm": 0.031140651553869247, "global_step": 15514, "epoch": 430, "lr": 0.001} +{"train_loss": 0.03525798591888613, "train_loss_bc": 0.038702722638845444, "train_loss_llm": 0.0010700523853302002, "grad_norm": 0.041408609598875046, "global_step": 15515, "epoch": 430, "lr": 0.001, "train/cumulative_reward": 104.6920867393, "train/mean_score": 16.922806654836396, "train/success_rate": 0.4, "test/cumulative_reward": 43.94462250529452, "test/mean_score": 7.054774432931155, "test/success_rate": 0.16, "val_loss": 0.06996146589517593, "train_action_mse_error": 0.03854251280426979} +{"train_loss": 0.032796233892440796, "train_loss_bc": 0.032796233892440796, "train_loss_llm": 0.0, "grad_norm": 0.049630023539066315, "global_step": 15516, "epoch": 431, "lr": 0.001} +{"train_loss": 0.033939559012651443, "train_loss_bc": 0.033939559012651443, "train_loss_llm": 0.0, "grad_norm": 0.0219049584120512, "global_step": 15517, "epoch": 431, "lr": 0.001} +{"train_loss": 0.02970193699002266, "train_loss_bc": 0.02970193699002266, "train_loss_llm": 0.0, "grad_norm": 0.037804849445819855, "global_step": 15518, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03980822116136551, "train_loss_bc": 0.03980822116136551, "train_loss_llm": 0.0, "grad_norm": 0.038010116666555405, "global_step": 15519, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03858794644474983, "train_loss_bc": 0.03858794644474983, "train_loss_llm": 0.0, "grad_norm": 0.04461706057190895, "global_step": 15520, "epoch": 431, "lr": 0.001} +{"train_loss": 0.032107796519994736, "train_loss_bc": 0.032107796519994736, "train_loss_llm": 0.0, "grad_norm": 0.020884031429886818, "global_step": 15521, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03328057378530502, "train_loss_bc": 0.03328057378530502, "train_loss_llm": 0.0, "grad_norm": 0.028194714337587357, "global_step": 15522, "epoch": 431, "lr": 0.001} +{"train_loss": 0.037110015749931335, "train_loss_bc": 0.03708818927407265, "train_loss_llm": 0.0021827518939971924, "grad_norm": 0.04319146275520325, "global_step": 15523, "epoch": 431, "lr": 0.001} +{"train_loss": 0.032865338027477264, "train_loss_bc": 0.03286173939704895, "train_loss_llm": 0.000359952449798584, "grad_norm": 0.046106331050395966, "global_step": 15524, "epoch": 431, "lr": 0.001} +{"train_loss": 0.033443745225667953, "train_loss_bc": 0.033439841121435165, "train_loss_llm": 0.0003902316093444824, "grad_norm": 0.01891956850886345, "global_step": 15525, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03012469969689846, "train_loss_bc": 0.03012469969689846, "train_loss_llm": 0.0, "grad_norm": 0.027679596096277237, "global_step": 15526, "epoch": 431, "lr": 0.001} +{"train_loss": 0.04287390038371086, "train_loss_bc": 0.04284943267703056, "train_loss_llm": 0.002446591854095459, "grad_norm": 0.039955850690603256, "global_step": 15527, "epoch": 431, "lr": 0.001} +{"train_loss": 0.037873364984989166, "train_loss_bc": 0.037873364984989166, "train_loss_llm": 0.0, "grad_norm": 0.051172856241464615, "global_step": 15528, "epoch": 431, "lr": 0.001} +{"train_loss": 0.033786263316869736, "train_loss_bc": 0.03378042206168175, "train_loss_llm": 0.0005843043327331543, "grad_norm": 0.018248237669467926, "global_step": 15529, "epoch": 431, "lr": 0.001} +{"train_loss": 0.032912254333496094, "train_loss_bc": 0.032912254333496094, "train_loss_llm": 0.0, "grad_norm": 0.03066413477063179, "global_step": 15530, "epoch": 431, "lr": 0.001} +{"train_loss": 0.039907023310661316, "train_loss_bc": 0.039907023310661316, "train_loss_llm": 0.0, "grad_norm": 0.047878339886665344, "global_step": 15531, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03864545002579689, "train_loss_bc": 0.03864545002579689, "train_loss_llm": 0.0, "grad_norm": 0.055713482201099396, "global_step": 15532, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03267603740096092, "train_loss_bc": 0.03267603740096092, "train_loss_llm": 0.0, "grad_norm": 0.023863714188337326, "global_step": 15533, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03450793772935867, "train_loss_bc": 0.03450793772935867, "train_loss_llm": 0.0, "grad_norm": 0.03441299498081207, "global_step": 15534, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03453841432929039, "train_loss_bc": 0.03453841432929039, "train_loss_llm": 0.0, "grad_norm": 0.041562628000974655, "global_step": 15535, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03655033931136131, "train_loss_bc": 0.03654219210147858, "train_loss_llm": 0.0008148849010467529, "grad_norm": 0.053457971662282944, "global_step": 15536, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03165633603930473, "train_loss_bc": 0.03165633603930473, "train_loss_llm": 0.0, "grad_norm": 0.02223747782409191, "global_step": 15537, "epoch": 431, "lr": 0.001} +{"train_loss": 0.04090358689427376, "train_loss_bc": 0.04090358689427376, "train_loss_llm": 0.0, "grad_norm": 0.02977898344397545, "global_step": 15538, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03245816007256508, "train_loss_bc": 0.032447293400764465, "train_loss_llm": 0.0010865628719329834, "grad_norm": 0.03967292606830597, "global_step": 15539, "epoch": 431, "lr": 0.001} +{"train_loss": 0.037505168467760086, "train_loss_bc": 0.037488654255867004, "train_loss_llm": 0.0016514360904693604, "grad_norm": 0.04643067345023155, "global_step": 15540, "epoch": 431, "lr": 0.001} +{"train_loss": 0.038914263248443604, "train_loss_bc": 0.03888623043894768, "train_loss_llm": 0.0028031468391418457, "grad_norm": 0.026603111997246742, "global_step": 15541, "epoch": 431, "lr": 0.001} +{"train_loss": 0.035306625068187714, "train_loss_bc": 0.035306625068187714, "train_loss_llm": 0.0, "grad_norm": 0.03140683472156525, "global_step": 15542, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03519679978489876, "train_loss_bc": 0.03519590198993683, "train_loss_llm": 8.982419967651367e-05, "grad_norm": 0.0428379587829113, "global_step": 15543, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03341037780046463, "train_loss_bc": 0.03341037780046463, "train_loss_llm": 0.0, "grad_norm": 0.052348554134368896, "global_step": 15544, "epoch": 431, "lr": 0.001} +{"train_loss": 0.029498321935534477, "train_loss_bc": 0.029498038813471794, "train_loss_llm": 2.8342008590698242e-05, "grad_norm": 0.025893958285450935, "global_step": 15545, "epoch": 431, "lr": 0.001} +{"train_loss": 0.031480006873607635, "train_loss_bc": 0.031480006873607635, "train_loss_llm": 0.0, "grad_norm": 0.036202073097229004, "global_step": 15546, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03131933510303497, "train_loss_bc": 0.03131933510303497, "train_loss_llm": 0.0, "grad_norm": 0.04769737273454666, "global_step": 15547, "epoch": 431, "lr": 0.001} +{"train_loss": 0.030142026022076607, "train_loss_bc": 0.030142026022076607, "train_loss_llm": 0.0, "grad_norm": 0.05109154060482979, "global_step": 15548, "epoch": 431, "lr": 0.001} +{"train_loss": 0.04089254140853882, "train_loss_bc": 0.04089254140853882, "train_loss_llm": 0.0, "grad_norm": 0.02288522757589817, "global_step": 15549, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03696153312921524, "train_loss_bc": 0.03696153312921524, "train_loss_llm": 0.0, "grad_norm": 0.03714735805988312, "global_step": 15550, "epoch": 431, "lr": 0.001} +{"train_loss": 0.03523639925859041, "train_loss_bc": 0.04482823982834816, "train_loss_llm": 0.0, "grad_norm": 0.04502911865711212, "global_step": 15551, "epoch": 431, "lr": 0.001, "val_loss": 0.06954842805862427} +{"train_loss": 0.0362648107111454, "train_loss_bc": 0.0362648107111454, "train_loss_llm": 0.0, "grad_norm": 0.05610031634569168, "global_step": 15552, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03468713164329529, "train_loss_bc": 0.03468373790383339, "train_loss_llm": 0.0003393888473510742, "grad_norm": 0.02609580010175705, "global_step": 15553, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03002139925956726, "train_loss_bc": 0.03002139925956726, "train_loss_llm": 0.0, "grad_norm": 0.03757310286164284, "global_step": 15554, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03503444418311119, "train_loss_bc": 0.03503444418311119, "train_loss_llm": 0.0, "grad_norm": 0.04818810895085335, "global_step": 15555, "epoch": 432, "lr": 0.001} +{"train_loss": 0.04000446945428848, "train_loss_bc": 0.04000446945428848, "train_loss_llm": 0.0, "grad_norm": 0.0522826686501503, "global_step": 15556, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03761771693825722, "train_loss_bc": 0.03761771693825722, "train_loss_llm": 0.0, "grad_norm": 0.028431972488760948, "global_step": 15557, "epoch": 432, "lr": 0.001} +{"train_loss": 0.032037559896707535, "train_loss_bc": 0.032037559896707535, "train_loss_llm": 0.0, "grad_norm": 0.04292786866426468, "global_step": 15558, "epoch": 432, "lr": 0.001} +{"train_loss": 0.024905463680624962, "train_loss_bc": 0.024905463680624962, "train_loss_llm": 0.0, "grad_norm": 0.0542384572327137, "global_step": 15559, "epoch": 432, "lr": 0.001} +{"train_loss": 0.036796994507312775, "train_loss_bc": 0.036796994507312775, "train_loss_llm": 0.0, "grad_norm": 0.06113997474312782, "global_step": 15560, "epoch": 432, "lr": 0.001} +{"train_loss": 0.038432467728853226, "train_loss_bc": 0.038432467728853226, "train_loss_llm": 0.0, "grad_norm": 0.02057819813489914, "global_step": 15561, "epoch": 432, "lr": 0.001} +{"train_loss": 0.031573183834552765, "train_loss_bc": 0.031573183834552765, "train_loss_llm": 0.0, "grad_norm": 0.03019237332046032, "global_step": 15562, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03148315101861954, "train_loss_bc": 0.03148315101861954, "train_loss_llm": 0.0, "grad_norm": 0.04114999622106552, "global_step": 15563, "epoch": 432, "lr": 0.001} +{"train_loss": 0.040833063423633575, "train_loss_bc": 0.040833063423633575, "train_loss_llm": 0.0, "grad_norm": 0.05617998540401459, "global_step": 15564, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03623490780591965, "train_loss_bc": 0.03623490780591965, "train_loss_llm": 0.0, "grad_norm": 0.024214264005422592, "global_step": 15565, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03783976659178734, "train_loss_bc": 0.03783976659178734, "train_loss_llm": 0.0, "grad_norm": 0.040254879742860794, "global_step": 15566, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03283540531992912, "train_loss_bc": 0.03283540531992912, "train_loss_llm": 0.0, "grad_norm": 0.060039542615413666, "global_step": 15567, "epoch": 432, "lr": 0.001} +{"train_loss": 0.040075916796922684, "train_loss_bc": 0.040075916796922684, "train_loss_llm": 0.0, "grad_norm": 0.06767149269580841, "global_step": 15568, "epoch": 432, "lr": 0.001} +{"train_loss": 0.031394124031066895, "train_loss_bc": 0.031394124031066895, "train_loss_llm": 0.0, "grad_norm": 0.021893417462706566, "global_step": 15569, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03870837017893791, "train_loss_bc": 0.03870801255106926, "train_loss_llm": 3.55839729309082e-05, "grad_norm": 0.037472810596227646, "global_step": 15570, "epoch": 432, "lr": 0.001} +{"train_loss": 0.036486390978097916, "train_loss_bc": 0.036486390978097916, "train_loss_llm": 0.0, "grad_norm": 0.04826413840055466, "global_step": 15571, "epoch": 432, "lr": 0.001} +{"train_loss": 0.036625295877456665, "train_loss_bc": 0.036625295877456665, "train_loss_llm": 0.0, "grad_norm": 0.0681777074933052, "global_step": 15572, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03706006333231926, "train_loss_bc": 0.03705889731645584, "train_loss_llm": 0.00011655688285827637, "grad_norm": 0.025568658486008644, "global_step": 15573, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03834269195795059, "train_loss_bc": 0.03834269195795059, "train_loss_llm": 0.0, "grad_norm": 0.04069980978965759, "global_step": 15574, "epoch": 432, "lr": 0.001} +{"train_loss": 0.027956413105130196, "train_loss_bc": 0.027956413105130196, "train_loss_llm": 0.0, "grad_norm": 0.049174755811691284, "global_step": 15575, "epoch": 432, "lr": 0.001} +{"train_loss": 0.038371339440345764, "train_loss_bc": 0.038371339440345764, "train_loss_llm": 0.0, "grad_norm": 0.05850747600197792, "global_step": 15576, "epoch": 432, "lr": 0.001} +{"train_loss": 0.036183543503284454, "train_loss_bc": 0.036183543503284454, "train_loss_llm": 0.0, "grad_norm": 0.023178108036518097, "global_step": 15577, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03659522905945778, "train_loss_bc": 0.036559924483299255, "train_loss_llm": 0.0035306215286254883, "grad_norm": 0.04065771400928497, "global_step": 15578, "epoch": 432, "lr": 0.001} +{"train_loss": 0.0301783699542284, "train_loss_bc": 0.0301783699542284, "train_loss_llm": 0.0, "grad_norm": 0.045919716358184814, "global_step": 15579, "epoch": 432, "lr": 0.001} +{"train_loss": 0.038551874458789825, "train_loss_bc": 0.038551874458789825, "train_loss_llm": 0.0, "grad_norm": 0.054443784058094025, "global_step": 15580, "epoch": 432, "lr": 0.001} +{"train_loss": 0.025675944983959198, "train_loss_bc": 0.025675944983959198, "train_loss_llm": 0.0, "grad_norm": 0.021339068189263344, "global_step": 15581, "epoch": 432, "lr": 0.001} +{"train_loss": 0.04104280471801758, "train_loss_bc": 0.04104234278202057, "train_loss_llm": 4.6372413635253906e-05, "grad_norm": 0.042025912553071976, "global_step": 15582, "epoch": 432, "lr": 0.001} +{"train_loss": 0.027949661016464233, "train_loss_bc": 0.027949661016464233, "train_loss_llm": 0.0, "grad_norm": 0.049873530864715576, "global_step": 15583, "epoch": 432, "lr": 0.001} +{"train_loss": 0.04512704908847809, "train_loss_bc": 0.045124102383852005, "train_loss_llm": 0.0002945065498352051, "grad_norm": 0.060915086418390274, "global_step": 15584, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03601735085248947, "train_loss_bc": 0.03601735085248947, "train_loss_llm": 0.0, "grad_norm": 0.02608058787882328, "global_step": 15585, "epoch": 432, "lr": 0.001} +{"train_loss": 0.037033505737781525, "train_loss_bc": 0.037033505737781525, "train_loss_llm": 0.0, "grad_norm": 0.04177064821124077, "global_step": 15586, "epoch": 432, "lr": 0.001} +{"train_loss": 0.03523244817430774, "train_loss_bc": 0.03239025920629501, "train_loss_llm": 0.0, "grad_norm": 0.06007618457078934, "global_step": 15587, "epoch": 432, "lr": 0.001, "val_loss": 0.07120566070079803} +{"train_loss": 0.04113738611340523, "train_loss_bc": 0.04113263264298439, "train_loss_llm": 0.0004754066467285156, "grad_norm": 0.0740199089050293, "global_step": 15588, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03298303484916687, "train_loss_bc": 0.032969191670417786, "train_loss_llm": 0.0013841986656188965, "grad_norm": 0.02471878193318844, "global_step": 15589, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03706199303269386, "train_loss_bc": 0.03705925494432449, "train_loss_llm": 0.00027373433113098145, "grad_norm": 0.03496314212679863, "global_step": 15590, "epoch": 433, "lr": 0.001} +{"train_loss": 0.031229330226778984, "train_loss_bc": 0.031229330226778984, "train_loss_llm": 0.0, "grad_norm": 0.04496391862630844, "global_step": 15591, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03241332247853279, "train_loss_bc": 0.03241332247853279, "train_loss_llm": 0.0, "grad_norm": 0.05874086543917656, "global_step": 15592, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03739100322127342, "train_loss_bc": 0.03739100322127342, "train_loss_llm": 0.0, "grad_norm": 0.030427707359194756, "global_step": 15593, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03587045520544052, "train_loss_bc": 0.03587045520544052, "train_loss_llm": 0.0, "grad_norm": 0.045351509004831314, "global_step": 15594, "epoch": 433, "lr": 0.001} +{"train_loss": 0.035369716584682465, "train_loss_bc": 0.03534875810146332, "train_loss_llm": 0.002095848321914673, "grad_norm": 0.06299611181020737, "global_step": 15595, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03619302436709404, "train_loss_bc": 0.03617607057094574, "train_loss_llm": 0.0016955137252807617, "grad_norm": 0.0737837702035904, "global_step": 15596, "epoch": 433, "lr": 0.001} +{"train_loss": 0.0353844091296196, "train_loss_bc": 0.0353844091296196, "train_loss_llm": 0.0, "grad_norm": 0.02764957770705223, "global_step": 15597, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03519373759627342, "train_loss_bc": 0.035186734050512314, "train_loss_llm": 0.0007002055644989014, "grad_norm": 0.03370033949613571, "global_step": 15598, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03605511412024498, "train_loss_bc": 0.03605511412024498, "train_loss_llm": 0.0, "grad_norm": 0.0431065671145916, "global_step": 15599, "epoch": 433, "lr": 0.001} +{"train_loss": 0.032706379890441895, "train_loss_bc": 0.032706379890441895, "train_loss_llm": 0.0, "grad_norm": 0.057037901133298874, "global_step": 15600, "epoch": 433, "lr": 0.001} +{"train_loss": 0.02800033800303936, "train_loss_bc": 0.02798985317349434, "train_loss_llm": 0.0010484158992767334, "grad_norm": 0.02544485591351986, "global_step": 15601, "epoch": 433, "lr": 0.001} +{"train_loss": 0.035255175083875656, "train_loss_bc": 0.035255175083875656, "train_loss_llm": 0.0, "grad_norm": 0.036933064460754395, "global_step": 15602, "epoch": 433, "lr": 0.001} +{"train_loss": 0.02875051274895668, "train_loss_bc": 0.02875051274895668, "train_loss_llm": 0.0, "grad_norm": 0.051001716405153275, "global_step": 15603, "epoch": 433, "lr": 0.001} +{"train_loss": 0.033591143786907196, "train_loss_bc": 0.033591143786907196, "train_loss_llm": 0.0, "grad_norm": 0.05914078280329704, "global_step": 15604, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03366585075855255, "train_loss_bc": 0.03366585075855255, "train_loss_llm": 0.0, "grad_norm": 0.019586965441703796, "global_step": 15605, "epoch": 433, "lr": 0.001} +{"train_loss": 0.037451695650815964, "train_loss_bc": 0.037446632981300354, "train_loss_llm": 0.0005063414573669434, "grad_norm": 0.034166134893894196, "global_step": 15606, "epoch": 433, "lr": 0.001} +{"train_loss": 0.030724652111530304, "train_loss_bc": 0.030724652111530304, "train_loss_llm": 0.0, "grad_norm": 0.04230014979839325, "global_step": 15607, "epoch": 433, "lr": 0.001} +{"train_loss": 0.030657531693577766, "train_loss_bc": 0.03064894862473011, "train_loss_llm": 0.000858306884765625, "grad_norm": 0.052080731838941574, "global_step": 15608, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03365533426403999, "train_loss_bc": 0.03365533426403999, "train_loss_llm": 0.0, "grad_norm": 0.029457001015543938, "global_step": 15609, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03976152092218399, "train_loss_bc": 0.03976152092218399, "train_loss_llm": 0.0, "grad_norm": 0.04311332851648331, "global_step": 15610, "epoch": 433, "lr": 0.001} +{"train_loss": 0.031243164092302322, "train_loss_bc": 0.031243164092302322, "train_loss_llm": 0.0, "grad_norm": 0.05847803130745888, "global_step": 15611, "epoch": 433, "lr": 0.001} +{"train_loss": 0.04240851104259491, "train_loss_bc": 0.042407140135765076, "train_loss_llm": 0.00013715028762817383, "grad_norm": 0.07769091427326202, "global_step": 15612, "epoch": 433, "lr": 0.001} +{"train_loss": 0.031687039881944656, "train_loss_bc": 0.031674839556217194, "train_loss_llm": 0.0012201368808746338, "grad_norm": 0.024555565789341927, "global_step": 15613, "epoch": 433, "lr": 0.001} +{"train_loss": 0.038426171988248825, "train_loss_bc": 0.038426171988248825, "train_loss_llm": 0.0, "grad_norm": 0.03689640760421753, "global_step": 15614, "epoch": 433, "lr": 0.001} +{"train_loss": 0.037102654576301575, "train_loss_bc": 0.03708832710981369, "train_loss_llm": 0.0014327764511108398, "grad_norm": 0.05637221783399582, "global_step": 15615, "epoch": 433, "lr": 0.001} +{"train_loss": 0.04536460340023041, "train_loss_bc": 0.04536460340023041, "train_loss_llm": 0.0, "grad_norm": 0.08636649698019028, "global_step": 15616, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03887340426445007, "train_loss_bc": 0.038859281688928604, "train_loss_llm": 0.0014123320579528809, "grad_norm": 0.01965387351810932, "global_step": 15617, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03435841202735901, "train_loss_bc": 0.03435841202735901, "train_loss_llm": 0.0, "grad_norm": 0.035765308886766434, "global_step": 15618, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03943135216832161, "train_loss_bc": 0.03943135216832161, "train_loss_llm": 0.0, "grad_norm": 0.050784457474946976, "global_step": 15619, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03428656607866287, "train_loss_bc": 0.03428656607866287, "train_loss_llm": 0.0, "grad_norm": 0.06537740677595139, "global_step": 15620, "epoch": 433, "lr": 0.001} +{"train_loss": 0.0342753529548645, "train_loss_bc": 0.03426607325673103, "train_loss_llm": 0.000928044319152832, "grad_norm": 0.03312145173549652, "global_step": 15621, "epoch": 433, "lr": 0.001} +{"train_loss": 0.04152149334549904, "train_loss_bc": 0.04152149334549904, "train_loss_llm": 0.0, "grad_norm": 0.052368126809597015, "global_step": 15622, "epoch": 433, "lr": 0.001} +{"train_loss": 0.03543399962493115, "train_loss_bc": 0.03614259883761406, "train_loss_llm": 0.0, "grad_norm": 0.061314698308706284, "global_step": 15623, "epoch": 433, "lr": 0.001, "val_loss": 0.07008372247219086} +{"train_loss": 0.03198681026697159, "train_loss_bc": 0.03198681026697159, "train_loss_llm": 0.0, "grad_norm": 0.0655643418431282, "global_step": 15624, "epoch": 434, "lr": 0.001} +{"train_loss": 0.0337432324886322, "train_loss_bc": 0.0337432324886322, "train_loss_llm": 0.0, "grad_norm": 0.02283143252134323, "global_step": 15625, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03531907498836517, "train_loss_bc": 0.03531907498836517, "train_loss_llm": 0.0, "grad_norm": 0.03644634410738945, "global_step": 15626, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03271026536822319, "train_loss_bc": 0.03271026536822319, "train_loss_llm": 0.0, "grad_norm": 0.05259358137845993, "global_step": 15627, "epoch": 434, "lr": 0.001} +{"train_loss": 0.04055560752749443, "train_loss_bc": 0.04055560752749443, "train_loss_llm": 0.0, "grad_norm": 0.06590649485588074, "global_step": 15628, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03308799862861633, "train_loss_bc": 0.033076923340559006, "train_loss_llm": 0.0011075735092163086, "grad_norm": 0.022393690422177315, "global_step": 15629, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03017950989305973, "train_loss_bc": 0.03017950989305973, "train_loss_llm": 0.0, "grad_norm": 0.033785611391067505, "global_step": 15630, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03388752415776253, "train_loss_bc": 0.03388752415776253, "train_loss_llm": 0.0, "grad_norm": 0.040772221982479095, "global_step": 15631, "epoch": 434, "lr": 0.001} +{"train_loss": 0.02716943435370922, "train_loss_bc": 0.027164718136191368, "train_loss_llm": 0.0004716813564300537, "grad_norm": 0.053477153182029724, "global_step": 15632, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03429928421974182, "train_loss_bc": 0.03429928421974182, "train_loss_llm": 0.0, "grad_norm": 0.024570658802986145, "global_step": 15633, "epoch": 434, "lr": 0.001} +{"train_loss": 0.040364474058151245, "train_loss_bc": 0.040364474058151245, "train_loss_llm": 0.0, "grad_norm": 0.04017679765820503, "global_step": 15634, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03355748578906059, "train_loss_bc": 0.033552322536706924, "train_loss_llm": 0.000516355037689209, "grad_norm": 0.053852543234825134, "global_step": 15635, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03988592326641083, "train_loss_bc": 0.039869628846645355, "train_loss_llm": 0.0016293227672576904, "grad_norm": 0.07054711133241653, "global_step": 15636, "epoch": 434, "lr": 0.001} +{"train_loss": 0.029674682766199112, "train_loss_bc": 0.02966664545238018, "train_loss_llm": 0.0008036494255065918, "grad_norm": 0.025564679875969887, "global_step": 15637, "epoch": 434, "lr": 0.001} +{"train_loss": 0.0371849499642849, "train_loss_bc": 0.0371849499642849, "train_loss_llm": 0.0, "grad_norm": 0.04177140071988106, "global_step": 15638, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03439284488558769, "train_loss_bc": 0.03439284488558769, "train_loss_llm": 0.0, "grad_norm": 0.048126135021448135, "global_step": 15639, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03638598322868347, "train_loss_bc": 0.03638598322868347, "train_loss_llm": 0.0, "grad_norm": 0.05551404878497124, "global_step": 15640, "epoch": 434, "lr": 0.001} +{"train_loss": 0.034231096506118774, "train_loss_bc": 0.03422565013170242, "train_loss_llm": 0.0005446970462799072, "grad_norm": 0.03564316779375076, "global_step": 15641, "epoch": 434, "lr": 0.001} +{"train_loss": 0.034847985953092575, "train_loss_bc": 0.034847985953092575, "train_loss_llm": 0.0, "grad_norm": 0.04491102695465088, "global_step": 15642, "epoch": 434, "lr": 0.001} +{"train_loss": 0.04465194791555405, "train_loss_bc": 0.04465194791555405, "train_loss_llm": 0.0, "grad_norm": 0.05872594565153122, "global_step": 15643, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03338687866926193, "train_loss_bc": 0.03338687866926193, "train_loss_llm": 0.0, "grad_norm": 0.07139148563146591, "global_step": 15644, "epoch": 434, "lr": 0.001} +{"train_loss": 0.02751261182129383, "train_loss_bc": 0.027500668540596962, "train_loss_llm": 0.0011944174766540527, "grad_norm": 0.020450418815016747, "global_step": 15645, "epoch": 434, "lr": 0.001} +{"train_loss": 0.04161550849676132, "train_loss_bc": 0.041605789214372635, "train_loss_llm": 0.0009718239307403564, "grad_norm": 0.0423487089574337, "global_step": 15646, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03919666260480881, "train_loss_bc": 0.03917201608419418, "train_loss_llm": 0.0024648308753967285, "grad_norm": 0.05264357104897499, "global_step": 15647, "epoch": 434, "lr": 0.001} +{"train_loss": 0.04119980335235596, "train_loss_bc": 0.04119980335235596, "train_loss_llm": 0.0, "grad_norm": 0.0690179169178009, "global_step": 15648, "epoch": 434, "lr": 0.001} +{"train_loss": 0.036472730338573456, "train_loss_bc": 0.0364539809525013, "train_loss_llm": 0.001874774694442749, "grad_norm": 0.022341391071677208, "global_step": 15649, "epoch": 434, "lr": 0.001} +{"train_loss": 0.038827236741781235, "train_loss_bc": 0.038827236741781235, "train_loss_llm": 0.0, "grad_norm": 0.03522977605462074, "global_step": 15650, "epoch": 434, "lr": 0.001} +{"train_loss": 0.034280382096767426, "train_loss_bc": 0.03426307439804077, "train_loss_llm": 0.001730591058731079, "grad_norm": 0.04952264949679375, "global_step": 15651, "epoch": 434, "lr": 0.001} +{"train_loss": 0.030825501307845116, "train_loss_bc": 0.030825501307845116, "train_loss_llm": 0.0, "grad_norm": 0.060887012630701065, "global_step": 15652, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03467390686273575, "train_loss_bc": 0.03467390686273575, "train_loss_llm": 0.0, "grad_norm": 0.024875707924365997, "global_step": 15653, "epoch": 434, "lr": 0.001} +{"train_loss": 0.039272408932447433, "train_loss_bc": 0.039272408932447433, "train_loss_llm": 0.0, "grad_norm": 0.04012036323547363, "global_step": 15654, "epoch": 434, "lr": 0.001} +{"train_loss": 0.035736434161663055, "train_loss_bc": 0.035736434161663055, "train_loss_llm": 0.0, "grad_norm": 0.05208039656281471, "global_step": 15655, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03552221134305, "train_loss_bc": 0.03552221134305, "train_loss_llm": 0.0, "grad_norm": 0.06763941794633865, "global_step": 15656, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03515652194619179, "train_loss_bc": 0.03515652194619179, "train_loss_llm": 0.0, "grad_norm": 0.024947011843323708, "global_step": 15657, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03131378814578056, "train_loss_bc": 0.03131378814578056, "train_loss_llm": 0.0, "grad_norm": 0.0402379184961319, "global_step": 15658, "epoch": 434, "lr": 0.001} +{"train_loss": 0.03537214650875992, "train_loss_bc": 0.040288571268320084, "train_loss_llm": 0.0, "grad_norm": 0.05982005223631859, "global_step": 15659, "epoch": 434, "lr": 0.001, "val_loss": 0.07047204673290253} +{"train_loss": 0.029431169852614403, "train_loss_bc": 0.02942790277302265, "train_loss_llm": 0.0003267228603363037, "grad_norm": 0.07216964662075043, "global_step": 15660, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03093057870864868, "train_loss_bc": 0.03093057870864868, "train_loss_llm": 0.0, "grad_norm": 0.03109717182815075, "global_step": 15661, "epoch": 435, "lr": 0.001} +{"train_loss": 0.02559688687324524, "train_loss_bc": 0.02559688687324524, "train_loss_llm": 0.0, "grad_norm": 0.04448416829109192, "global_step": 15662, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03108122944831848, "train_loss_bc": 0.03108122944831848, "train_loss_llm": 0.0, "grad_norm": 0.05159588158130646, "global_step": 15663, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03628992661833763, "train_loss_bc": 0.03628992661833763, "train_loss_llm": 0.0, "grad_norm": 0.06743569672107697, "global_step": 15664, "epoch": 435, "lr": 0.001} +{"train_loss": 0.04429250583052635, "train_loss_bc": 0.04428461566567421, "train_loss_llm": 0.0007890164852142334, "grad_norm": 0.0242955069988966, "global_step": 15665, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03364890441298485, "train_loss_bc": 0.03363044187426567, "train_loss_llm": 0.0018462538719177246, "grad_norm": 0.04116738215088844, "global_step": 15666, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03481409698724747, "train_loss_bc": 0.03481409698724747, "train_loss_llm": 0.0, "grad_norm": 0.0489557608962059, "global_step": 15667, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03272514045238495, "train_loss_bc": 0.032719340175390244, "train_loss_llm": 0.0005801916122436523, "grad_norm": 0.05464169755578041, "global_step": 15668, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03425607830286026, "train_loss_bc": 0.03425607830286026, "train_loss_llm": 0.0, "grad_norm": 0.02321545220911503, "global_step": 15669, "epoch": 435, "lr": 0.001} +{"train_loss": 0.02909759432077408, "train_loss_bc": 0.029093682765960693, "train_loss_llm": 0.00039124488830566406, "grad_norm": 0.03936194255948067, "global_step": 15670, "epoch": 435, "lr": 0.001} +{"train_loss": 0.034529909491539, "train_loss_bc": 0.034529909491539, "train_loss_llm": 0.0, "grad_norm": 0.05100155249238014, "global_step": 15671, "epoch": 435, "lr": 0.001} +{"train_loss": 0.02996901609003544, "train_loss_bc": 0.029943019151687622, "train_loss_llm": 0.002599775791168213, "grad_norm": 0.06257979571819305, "global_step": 15672, "epoch": 435, "lr": 0.001} +{"train_loss": 0.030452817678451538, "train_loss_bc": 0.030452817678451538, "train_loss_llm": 0.0, "grad_norm": 0.02209298312664032, "global_step": 15673, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03785314783453941, "train_loss_bc": 0.03785314783453941, "train_loss_llm": 0.0, "grad_norm": 0.03216037526726723, "global_step": 15674, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03972369432449341, "train_loss_bc": 0.03969518840312958, "train_loss_llm": 0.002850562334060669, "grad_norm": 0.04790632054209709, "global_step": 15675, "epoch": 435, "lr": 0.001} +{"train_loss": 0.044474635273218155, "train_loss_bc": 0.044474635273218155, "train_loss_llm": 0.0, "grad_norm": 0.05477561801671982, "global_step": 15676, "epoch": 435, "lr": 0.001} +{"train_loss": 0.038701094686985016, "train_loss_bc": 0.03869006037712097, "train_loss_llm": 0.0011036098003387451, "grad_norm": 0.02593839168548584, "global_step": 15677, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03240416571497917, "train_loss_bc": 0.03240416571497917, "train_loss_llm": 0.0, "grad_norm": 0.0385749526321888, "global_step": 15678, "epoch": 435, "lr": 0.001} +{"train_loss": 0.04270593822002411, "train_loss_bc": 0.04270095005631447, "train_loss_llm": 0.0004988312721252441, "grad_norm": 0.05779794976115227, "global_step": 15679, "epoch": 435, "lr": 0.001} +{"train_loss": 0.036573391407728195, "train_loss_bc": 0.03657281771302223, "train_loss_llm": 5.7369470596313477e-05, "grad_norm": 0.07055886834859848, "global_step": 15680, "epoch": 435, "lr": 0.001} +{"train_loss": 0.036615755409002304, "train_loss_bc": 0.036615755409002304, "train_loss_llm": 0.0, "grad_norm": 0.02725188061594963, "global_step": 15681, "epoch": 435, "lr": 0.001} +{"train_loss": 0.032657623291015625, "train_loss_bc": 0.03263844549655914, "train_loss_llm": 0.0019176602363586426, "grad_norm": 0.03945653885602951, "global_step": 15682, "epoch": 435, "lr": 0.001} +{"train_loss": 0.037271223962306976, "train_loss_bc": 0.037271223962306976, "train_loss_llm": 0.0, "grad_norm": 0.051547762006521225, "global_step": 15683, "epoch": 435, "lr": 0.001} +{"train_loss": 0.032794367522001266, "train_loss_bc": 0.03278748691082001, "train_loss_llm": 0.0006882250308990479, "grad_norm": 0.06563223153352737, "global_step": 15684, "epoch": 435, "lr": 0.001} +{"train_loss": 0.04258156195282936, "train_loss_bc": 0.04258156195282936, "train_loss_llm": 0.0, "grad_norm": 0.031713906675577164, "global_step": 15685, "epoch": 435, "lr": 0.001} +{"train_loss": 0.038433823734521866, "train_loss_bc": 0.038433823734521866, "train_loss_llm": 0.0, "grad_norm": 0.049302130937576294, "global_step": 15686, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03697402775287628, "train_loss_bc": 0.03697402775287628, "train_loss_llm": 0.0, "grad_norm": 0.06403713673353195, "global_step": 15687, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03482557460665703, "train_loss_bc": 0.03482557460665703, "train_loss_llm": 0.0, "grad_norm": 0.07679508626461029, "global_step": 15688, "epoch": 435, "lr": 0.001} +{"train_loss": 0.04107174649834633, "train_loss_bc": 0.0410563200712204, "train_loss_llm": 0.001542806625366211, "grad_norm": 0.030299659818410873, "global_step": 15689, "epoch": 435, "lr": 0.001} +{"train_loss": 0.032870613038539886, "train_loss_bc": 0.032870613038539886, "train_loss_llm": 0.0, "grad_norm": 0.040822435170412064, "global_step": 15690, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03294406086206436, "train_loss_bc": 0.03294406086206436, "train_loss_llm": 0.0, "grad_norm": 0.05727492272853851, "global_step": 15691, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03865225985646248, "train_loss_bc": 0.03865225985646248, "train_loss_llm": 0.0, "grad_norm": 0.06426681578159332, "global_step": 15692, "epoch": 435, "lr": 0.001} +{"train_loss": 0.026842407882213593, "train_loss_bc": 0.026842407882213593, "train_loss_llm": 0.0, "grad_norm": 0.01936153508722782, "global_step": 15693, "epoch": 435, "lr": 0.001} +{"train_loss": 0.03780851513147354, "train_loss_bc": 0.03780851513147354, "train_loss_llm": 0.0, "grad_norm": 0.03404202312231064, "global_step": 15694, "epoch": 435, "lr": 0.001} +{"train_loss": 0.035298909060657024, "train_loss_bc": 0.03886524215340614, "train_loss_llm": 0.0, "grad_norm": 0.04744618013501167, "global_step": 15695, "epoch": 435, "lr": 0.001, "train/cumulative_reward": 62.02489539791377, "train/mean_score": 8.535159041655724, "train/success_rate": 0.2, "test/cumulative_reward": 33.29488047348364, "test/mean_score": 5.4380388632961045, "test/success_rate": 0.1, "val_loss": 0.07049062848091125, "train_action_mse_error": 0.03835400938987732} +{"train_loss": 0.03424673527479172, "train_loss_bc": 0.03424673527479172, "train_loss_llm": 0.0, "grad_norm": 0.05405667424201965, "global_step": 15696, "epoch": 436, "lr": 0.001} +{"train_loss": 0.035187821835279465, "train_loss_bc": 0.035187821835279465, "train_loss_llm": 0.0, "grad_norm": 0.025435589253902435, "global_step": 15697, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03450547158718109, "train_loss_bc": 0.03450002893805504, "train_loss_llm": 0.0005442202091217041, "grad_norm": 0.04113190993666649, "global_step": 15698, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03855797275900841, "train_loss_bc": 0.03855470195412636, "train_loss_llm": 0.00032714009284973145, "grad_norm": 0.051909010857343674, "global_step": 15699, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03180253505706787, "train_loss_bc": 0.03178833797574043, "train_loss_llm": 0.0014196038246154785, "grad_norm": 0.06864122301340103, "global_step": 15700, "epoch": 436, "lr": 0.001} +{"train_loss": 0.036120954900979996, "train_loss_bc": 0.036120954900979996, "train_loss_llm": 0.0, "grad_norm": 0.025986293330788612, "global_step": 15701, "epoch": 436, "lr": 0.001} +{"train_loss": 0.035236358642578125, "train_loss_bc": 0.035236358642578125, "train_loss_llm": 0.0, "grad_norm": 0.04316309839487076, "global_step": 15702, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03689313679933548, "train_loss_bc": 0.03689313679933548, "train_loss_llm": 0.0, "grad_norm": 0.05884053185582161, "global_step": 15703, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03212524950504303, "train_loss_bc": 0.03211355581879616, "train_loss_llm": 0.0011692941188812256, "grad_norm": 0.07308138161897659, "global_step": 15704, "epoch": 436, "lr": 0.001} +{"train_loss": 0.030286025255918503, "train_loss_bc": 0.030286025255918503, "train_loss_llm": 0.0, "grad_norm": 0.023894978687167168, "global_step": 15705, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03411363065242767, "train_loss_bc": 0.03411363065242767, "train_loss_llm": 0.0, "grad_norm": 0.034674808382987976, "global_step": 15706, "epoch": 436, "lr": 0.001} +{"train_loss": 0.0424501970410347, "train_loss_bc": 0.0424501970410347, "train_loss_llm": 0.0, "grad_norm": 0.0397571325302124, "global_step": 15707, "epoch": 436, "lr": 0.001} +{"train_loss": 0.031271982938051224, "train_loss_bc": 0.031271982938051224, "train_loss_llm": 0.0, "grad_norm": 0.05688716843724251, "global_step": 15708, "epoch": 436, "lr": 0.001} +{"train_loss": 0.027416381984949112, "train_loss_bc": 0.027416381984949112, "train_loss_llm": 0.0, "grad_norm": 0.030801430344581604, "global_step": 15709, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03315455839037895, "train_loss_bc": 0.03315455839037895, "train_loss_llm": 0.0, "grad_norm": 0.04442299157381058, "global_step": 15710, "epoch": 436, "lr": 0.001} +{"train_loss": 0.040744222700595856, "train_loss_bc": 0.040744222700595856, "train_loss_llm": 0.0, "grad_norm": 0.06456688791513443, "global_step": 15711, "epoch": 436, "lr": 0.001} +{"train_loss": 0.04382169246673584, "train_loss_bc": 0.04382169246673584, "train_loss_llm": 0.0, "grad_norm": 0.07998199760913849, "global_step": 15712, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03225778415799141, "train_loss_bc": 0.032245613634586334, "train_loss_llm": 0.001216977834701538, "grad_norm": 0.023184049874544144, "global_step": 15713, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03595346212387085, "train_loss_bc": 0.03595085069537163, "train_loss_llm": 0.00026100873947143555, "grad_norm": 0.04237017035484314, "global_step": 15714, "epoch": 436, "lr": 0.001} +{"train_loss": 0.0382591113448143, "train_loss_bc": 0.0382591113448143, "train_loss_llm": 0.0, "grad_norm": 0.061980511993169785, "global_step": 15715, "epoch": 436, "lr": 0.001} +{"train_loss": 0.02940918318927288, "train_loss_bc": 0.02940918318927288, "train_loss_llm": 0.0, "grad_norm": 0.07140278816223145, "global_step": 15716, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03701218590140343, "train_loss_bc": 0.037009112536907196, "train_loss_llm": 0.0003071725368499756, "grad_norm": 0.027466708794236183, "global_step": 15717, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03426949679851532, "train_loss_bc": 0.03426949679851532, "train_loss_llm": 0.0, "grad_norm": 0.040593720972537994, "global_step": 15718, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03777402266860008, "train_loss_bc": 0.03777402266860008, "train_loss_llm": 0.0, "grad_norm": 0.05715443566441536, "global_step": 15719, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03410192206501961, "train_loss_bc": 0.03409402444958687, "train_loss_llm": 0.0007895827293395996, "grad_norm": 0.06983058154582977, "global_step": 15720, "epoch": 436, "lr": 0.001} +{"train_loss": 0.0364445261657238, "train_loss_bc": 0.036414314061403275, "train_loss_llm": 0.003021240234375, "grad_norm": 0.026712212711572647, "global_step": 15721, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03609737008810043, "train_loss_bc": 0.03609737008810043, "train_loss_llm": 0.0, "grad_norm": 0.042848918586969376, "global_step": 15722, "epoch": 436, "lr": 0.001} +{"train_loss": 0.04045834764838219, "train_loss_bc": 0.04045834764838219, "train_loss_llm": 0.0, "grad_norm": 0.06236124411225319, "global_step": 15723, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03488947078585625, "train_loss_bc": 0.03484248369932175, "train_loss_llm": 0.0046987831592559814, "grad_norm": 0.08876461535692215, "global_step": 15724, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03644389659166336, "train_loss_bc": 0.03643670305609703, "train_loss_llm": 0.0007194578647613525, "grad_norm": 0.031475670635700226, "global_step": 15725, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03344288468360901, "train_loss_bc": 0.03344288468360901, "train_loss_llm": 0.0, "grad_norm": 0.04609842970967293, "global_step": 15726, "epoch": 436, "lr": 0.001} +{"train_loss": 0.0389322005212307, "train_loss_bc": 0.03886648267507553, "train_loss_llm": 0.006571859121322632, "grad_norm": 0.05781242996454239, "global_step": 15727, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03765265271067619, "train_loss_bc": 0.03765265271067619, "train_loss_llm": 0.0, "grad_norm": 0.08205201476812363, "global_step": 15728, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03775951266288757, "train_loss_bc": 0.03775951266288757, "train_loss_llm": 0.0, "grad_norm": 0.02576398104429245, "global_step": 15729, "epoch": 436, "lr": 0.001} +{"train_loss": 0.030440054833889008, "train_loss_bc": 0.030440054833889008, "train_loss_llm": 0.0, "grad_norm": 0.03688811510801315, "global_step": 15730, "epoch": 436, "lr": 0.001} +{"train_loss": 0.03538182579601804, "train_loss_bc": 0.03420110419392586, "train_loss_llm": 0.0011612176895141602, "grad_norm": 0.061945777386426926, "global_step": 15731, "epoch": 436, "lr": 0.001, "val_loss": 0.07005418837070465} +{"train_loss": 0.038302160799503326, "train_loss_bc": 0.038276053965091705, "train_loss_llm": 0.002610713243484497, "grad_norm": 0.08453114330768585, "global_step": 15732, "epoch": 437, "lr": 0.001} +{"train_loss": 0.02909448742866516, "train_loss_bc": 0.02909448742866516, "train_loss_llm": 0.0, "grad_norm": 0.021951640024781227, "global_step": 15733, "epoch": 437, "lr": 0.001} +{"train_loss": 0.033736683428287506, "train_loss_bc": 0.033736683428287506, "train_loss_llm": 0.0, "grad_norm": 0.02921389974653721, "global_step": 15734, "epoch": 437, "lr": 0.001} +{"train_loss": 0.029210345819592476, "train_loss_bc": 0.029210345819592476, "train_loss_llm": 0.0, "grad_norm": 0.04410460218787193, "global_step": 15735, "epoch": 437, "lr": 0.001} +{"train_loss": 0.04351507127285004, "train_loss_bc": 0.04351507127285004, "train_loss_llm": 0.0, "grad_norm": 0.06118471547961235, "global_step": 15736, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03529341146349907, "train_loss_bc": 0.03529341146349907, "train_loss_llm": 0.0, "grad_norm": 0.028428778052330017, "global_step": 15737, "epoch": 437, "lr": 0.001} +{"train_loss": 0.039338789880275726, "train_loss_bc": 0.039338789880275726, "train_loss_llm": 0.0, "grad_norm": 0.05058553069829941, "global_step": 15738, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03797174617648125, "train_loss_bc": 0.03797174617648125, "train_loss_llm": 0.0, "grad_norm": 0.05703239515423775, "global_step": 15739, "epoch": 437, "lr": 0.001} +{"train_loss": 0.031756505370140076, "train_loss_bc": 0.031756505370140076, "train_loss_llm": 0.0, "grad_norm": 0.06834841519594193, "global_step": 15740, "epoch": 437, "lr": 0.001} +{"train_loss": 0.0298239104449749, "train_loss_bc": 0.0298239104449749, "train_loss_llm": 0.0, "grad_norm": 0.024541014805436134, "global_step": 15741, "epoch": 437, "lr": 0.001} +{"train_loss": 0.02379344031214714, "train_loss_bc": 0.02379344031214714, "train_loss_llm": 0.0, "grad_norm": 0.03356533125042915, "global_step": 15742, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03961288556456566, "train_loss_bc": 0.03961080685257912, "train_loss_llm": 0.0002079606056213379, "grad_norm": 0.05840741842985153, "global_step": 15743, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03421195223927498, "train_loss_bc": 0.03421195223927498, "train_loss_llm": 0.0, "grad_norm": 0.07529371231794357, "global_step": 15744, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03749631345272064, "train_loss_bc": 0.03749631345272064, "train_loss_llm": 0.0, "grad_norm": 0.030110575258731842, "global_step": 15745, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03316408023238182, "train_loss_bc": 0.0331357941031456, "train_loss_llm": 0.0028285980224609375, "grad_norm": 0.04257747158408165, "global_step": 15746, "epoch": 437, "lr": 0.001} +{"train_loss": 0.037734489887952805, "train_loss_bc": 0.037726305425167084, "train_loss_llm": 0.0008182823657989502, "grad_norm": 0.057741668075323105, "global_step": 15747, "epoch": 437, "lr": 0.001} +{"train_loss": 0.034148797392845154, "train_loss_bc": 0.034148797392845154, "train_loss_llm": 0.0, "grad_norm": 0.07538621872663498, "global_step": 15748, "epoch": 437, "lr": 0.001} +{"train_loss": 0.040868524461984634, "train_loss_bc": 0.040868524461984634, "train_loss_llm": 0.0, "grad_norm": 0.026635365560650826, "global_step": 15749, "epoch": 437, "lr": 0.001} +{"train_loss": 0.032130226492881775, "train_loss_bc": 0.032130226492881775, "train_loss_llm": 0.0, "grad_norm": 0.04547426849603653, "global_step": 15750, "epoch": 437, "lr": 0.001} +{"train_loss": 0.0359782911837101, "train_loss_bc": 0.0359782911837101, "train_loss_llm": 0.0, "grad_norm": 0.064316526055336, "global_step": 15751, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03309254348278046, "train_loss_bc": 0.03309254348278046, "train_loss_llm": 0.0, "grad_norm": 0.07102024555206299, "global_step": 15752, "epoch": 437, "lr": 0.001} +{"train_loss": 0.038853950798511505, "train_loss_bc": 0.038853950798511505, "train_loss_llm": 0.0, "grad_norm": 0.026607461273670197, "global_step": 15753, "epoch": 437, "lr": 0.001} +{"train_loss": 0.04168204590678215, "train_loss_bc": 0.04168204590678215, "train_loss_llm": 0.0, "grad_norm": 0.041498519480228424, "global_step": 15754, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03579290211200714, "train_loss_bc": 0.03579290211200714, "train_loss_llm": 0.0, "grad_norm": 0.05563819780945778, "global_step": 15755, "epoch": 437, "lr": 0.001} +{"train_loss": 0.034237705171108246, "train_loss_bc": 0.034237705171108246, "train_loss_llm": 0.0, "grad_norm": 0.06892887502908707, "global_step": 15756, "epoch": 437, "lr": 0.001} +{"train_loss": 0.04081617295742035, "train_loss_bc": 0.04081617295742035, "train_loss_llm": 0.0, "grad_norm": 0.025385718792676926, "global_step": 15757, "epoch": 437, "lr": 0.001} +{"train_loss": 0.035701822489500046, "train_loss_bc": 0.035701822489500046, "train_loss_llm": 0.0, "grad_norm": 0.03796078637242317, "global_step": 15758, "epoch": 437, "lr": 0.001} +{"train_loss": 0.030413828790187836, "train_loss_bc": 0.030413828790187836, "train_loss_llm": 0.0, "grad_norm": 0.051081325858831406, "global_step": 15759, "epoch": 437, "lr": 0.001} +{"train_loss": 0.039525121450424194, "train_loss_bc": 0.039525121450424194, "train_loss_llm": 0.0, "grad_norm": 0.06713367253541946, "global_step": 15760, "epoch": 437, "lr": 0.001} +{"train_loss": 0.030167458578944206, "train_loss_bc": 0.030167458578944206, "train_loss_llm": 0.0, "grad_norm": 0.029223546385765076, "global_step": 15761, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03843141719698906, "train_loss_bc": 0.0384044386446476, "train_loss_llm": 0.002697855234146118, "grad_norm": 0.04857949540019035, "global_step": 15762, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03349028155207634, "train_loss_bc": 0.03349011763930321, "train_loss_llm": 1.6510486602783203e-05, "grad_norm": 0.07441448420286179, "global_step": 15763, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03882809728384018, "train_loss_bc": 0.03880702704191208, "train_loss_llm": 0.0021069347858428955, "grad_norm": 0.10235230624675751, "global_step": 15764, "epoch": 437, "lr": 0.001} +{"train_loss": 0.0385943241417408, "train_loss_bc": 0.03859324753284454, "train_loss_llm": 0.00010770559310913086, "grad_norm": 0.030337011441588402, "global_step": 15765, "epoch": 437, "lr": 0.001} +{"train_loss": 0.03135186433792114, "train_loss_bc": 0.03133736923336983, "train_loss_llm": 0.0014493763446807861, "grad_norm": 0.048715464770793915, "global_step": 15766, "epoch": 437, "lr": 0.001} +{"train_loss": 0.035510328908761345, "train_loss_bc": 0.040210191160440445, "train_loss_llm": 0.0, "grad_norm": 0.07455712556838989, "global_step": 15767, "epoch": 437, "lr": 0.001, "val_loss": 0.07030297815799713} +{"train_loss": 0.03394094854593277, "train_loss_bc": 0.03391436114907265, "train_loss_llm": 0.002658843994140625, "grad_norm": 0.09592866897583008, "global_step": 15768, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03109576553106308, "train_loss_bc": 0.03109421581029892, "train_loss_llm": 0.00015500187873840332, "grad_norm": 0.028369860723614693, "global_step": 15769, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03674114868044853, "train_loss_bc": 0.03674114868044853, "train_loss_llm": 0.0, "grad_norm": 0.03981250524520874, "global_step": 15770, "epoch": 438, "lr": 0.001} +{"train_loss": 0.039116282016038895, "train_loss_bc": 0.039116282016038895, "train_loss_llm": 0.0, "grad_norm": 0.058227989822626114, "global_step": 15771, "epoch": 438, "lr": 0.001} +{"train_loss": 0.037772201001644135, "train_loss_bc": 0.0377650111913681, "train_loss_llm": 0.0007190406322479248, "grad_norm": 0.06738237291574478, "global_step": 15772, "epoch": 438, "lr": 0.001} +{"train_loss": 0.034093309193849564, "train_loss_bc": 0.034079134464263916, "train_loss_llm": 0.0014173686504364014, "grad_norm": 0.021296700462698936, "global_step": 15773, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03653664514422417, "train_loss_bc": 0.03653399646282196, "train_loss_llm": 0.00026473402976989746, "grad_norm": 0.03625461831688881, "global_step": 15774, "epoch": 438, "lr": 0.001} +{"train_loss": 0.040443792939186096, "train_loss_bc": 0.040443792939186096, "train_loss_llm": 0.0, "grad_norm": 0.03935905918478966, "global_step": 15775, "epoch": 438, "lr": 0.001} +{"train_loss": 0.028714491054415703, "train_loss_bc": 0.028714491054415703, "train_loss_llm": 0.0, "grad_norm": 0.041659023612737656, "global_step": 15776, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03470555320382118, "train_loss_bc": 0.03470555320382118, "train_loss_llm": 0.0, "grad_norm": 0.028352953493595123, "global_step": 15777, "epoch": 438, "lr": 0.001} +{"train_loss": 0.035193782299757004, "train_loss_bc": 0.035193782299757004, "train_loss_llm": 0.0, "grad_norm": 0.046200044453144073, "global_step": 15778, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03880486264824867, "train_loss_bc": 0.03880486264824867, "train_loss_llm": 0.0, "grad_norm": 0.0647343322634697, "global_step": 15779, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03473944216966629, "train_loss_bc": 0.03473944216966629, "train_loss_llm": 0.0, "grad_norm": 0.08412173390388489, "global_step": 15780, "epoch": 438, "lr": 0.001} +{"train_loss": 0.04111870005726814, "train_loss_bc": 0.04111870005726814, "train_loss_llm": 0.0, "grad_norm": 0.03263852000236511, "global_step": 15781, "epoch": 438, "lr": 0.001} +{"train_loss": 0.036380916833877563, "train_loss_bc": 0.036380916833877563, "train_loss_llm": 0.0, "grad_norm": 0.04373602569103241, "global_step": 15782, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03592698648571968, "train_loss_bc": 0.03592698648571968, "train_loss_llm": 0.0, "grad_norm": 0.062472522258758545, "global_step": 15783, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03361441567540169, "train_loss_bc": 0.03361441567540169, "train_loss_llm": 0.0, "grad_norm": 0.08144378662109375, "global_step": 15784, "epoch": 438, "lr": 0.001} +{"train_loss": 0.032229647040367126, "train_loss_bc": 0.032229647040367126, "train_loss_llm": 0.0, "grad_norm": 0.035496070981025696, "global_step": 15785, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03317848592996597, "train_loss_bc": 0.03317848592996597, "train_loss_llm": 0.0, "grad_norm": 0.060679368674755096, "global_step": 15786, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03852478787302971, "train_loss_bc": 0.03852478787302971, "train_loss_llm": 0.0, "grad_norm": 0.08438091725111008, "global_step": 15787, "epoch": 438, "lr": 0.001} +{"train_loss": 0.039116956293582916, "train_loss_bc": 0.039116956293582916, "train_loss_llm": 0.0, "grad_norm": 0.1010386198759079, "global_step": 15788, "epoch": 438, "lr": 0.001} +{"train_loss": 0.029956305399537086, "train_loss_bc": 0.029956305399537086, "train_loss_llm": 0.0, "grad_norm": 0.022593414410948753, "global_step": 15789, "epoch": 438, "lr": 0.001} +{"train_loss": 0.035323359072208405, "train_loss_bc": 0.035323359072208405, "train_loss_llm": 0.0, "grad_norm": 0.03327632322907448, "global_step": 15790, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03688610717654228, "train_loss_bc": 0.036869946867227554, "train_loss_llm": 0.0016158521175384521, "grad_norm": 0.04799816384911537, "global_step": 15791, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03531358391046524, "train_loss_bc": 0.035301174968481064, "train_loss_llm": 0.001240849494934082, "grad_norm": 0.06493869423866272, "global_step": 15792, "epoch": 438, "lr": 0.001} +{"train_loss": 0.039959076792001724, "train_loss_bc": 0.03995819389820099, "train_loss_llm": 8.812546730041504e-05, "grad_norm": 0.025357922539114952, "global_step": 15793, "epoch": 438, "lr": 0.001} +{"train_loss": 0.026533454656600952, "train_loss_bc": 0.02653193287551403, "train_loss_llm": 0.00015220046043395996, "grad_norm": 0.0360695943236351, "global_step": 15794, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03687756136059761, "train_loss_bc": 0.03687756136059761, "train_loss_llm": 0.0, "grad_norm": 0.04421301186084747, "global_step": 15795, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03607542812824249, "train_loss_bc": 0.03607542812824249, "train_loss_llm": 0.0, "grad_norm": 0.06042000278830528, "global_step": 15796, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03996110334992409, "train_loss_bc": 0.03996110334992409, "train_loss_llm": 0.0, "grad_norm": 0.0367552675306797, "global_step": 15797, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03589849919080734, "train_loss_bc": 0.03589849919080734, "train_loss_llm": 0.0, "grad_norm": 0.05230213701725006, "global_step": 15798, "epoch": 438, "lr": 0.001} +{"train_loss": 0.032846905291080475, "train_loss_bc": 0.032846905291080475, "train_loss_llm": 0.0, "grad_norm": 0.07137466222047806, "global_step": 15799, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03214772418141365, "train_loss_bc": 0.03214772418141365, "train_loss_llm": 0.0, "grad_norm": 0.08418715745210648, "global_step": 15800, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03581337258219719, "train_loss_bc": 0.03579454496502876, "train_loss_llm": 0.0018828511238098145, "grad_norm": 0.034553248435258865, "global_step": 15801, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03357318043708801, "train_loss_bc": 0.03357318043708801, "train_loss_llm": 0.0, "grad_norm": 0.055989205837249756, "global_step": 15802, "epoch": 438, "lr": 0.001} +{"train_loss": 0.03539935344209274, "train_loss_bc": 0.03522194176912308, "train_loss_llm": 0.0, "grad_norm": 0.06401795148849487, "global_step": 15803, "epoch": 438, "lr": 0.001, "val_loss": 0.07020527869462967} +{"train_loss": 0.03596421331167221, "train_loss_bc": 0.035956256091594696, "train_loss_llm": 0.0007956027984619141, "grad_norm": 0.0762854516506195, "global_step": 15804, "epoch": 439, "lr": 0.001} +{"train_loss": 0.034697771072387695, "train_loss_bc": 0.034697771072387695, "train_loss_llm": 0.0, "grad_norm": 0.024851592257618904, "global_step": 15805, "epoch": 439, "lr": 0.001} +{"train_loss": 0.040229640901088715, "train_loss_bc": 0.040229640901088715, "train_loss_llm": 0.0, "grad_norm": 0.053614843636751175, "global_step": 15806, "epoch": 439, "lr": 0.001} +{"train_loss": 0.032024722546339035, "train_loss_bc": 0.032024722546339035, "train_loss_llm": 0.0, "grad_norm": 0.07059682160615921, "global_step": 15807, "epoch": 439, "lr": 0.001} +{"train_loss": 0.030803514644503593, "train_loss_bc": 0.030803514644503593, "train_loss_llm": 0.0, "grad_norm": 0.09178473055362701, "global_step": 15808, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03860683739185333, "train_loss_bc": 0.03860683739185333, "train_loss_llm": 0.0, "grad_norm": 0.03498854115605354, "global_step": 15809, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03768911212682724, "train_loss_bc": 0.0376691073179245, "train_loss_llm": 0.0020003914833068848, "grad_norm": 0.05292971804738045, "global_step": 15810, "epoch": 439, "lr": 0.001} +{"train_loss": 0.040629059076309204, "train_loss_bc": 0.040629059076309204, "train_loss_llm": 0.0, "grad_norm": 0.07218866795301437, "global_step": 15811, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03633585944771767, "train_loss_bc": 0.03633585944771767, "train_loss_llm": 0.0, "grad_norm": 0.08120263367891312, "global_step": 15812, "epoch": 439, "lr": 0.001} +{"train_loss": 0.032281357795000076, "train_loss_bc": 0.032281357795000076, "train_loss_llm": 0.0, "grad_norm": 0.022516289725899696, "global_step": 15813, "epoch": 439, "lr": 0.001} +{"train_loss": 0.032514315098524094, "train_loss_bc": 0.032514315098524094, "train_loss_llm": 0.0, "grad_norm": 0.03921046108007431, "global_step": 15814, "epoch": 439, "lr": 0.001} +{"train_loss": 0.02896064706146717, "train_loss_bc": 0.02896064706146717, "train_loss_llm": 0.0, "grad_norm": 0.05023800581693649, "global_step": 15815, "epoch": 439, "lr": 0.001} +{"train_loss": 0.036833494901657104, "train_loss_bc": 0.036833494901657104, "train_loss_llm": 0.0, "grad_norm": 0.06129692122340202, "global_step": 15816, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03005073592066765, "train_loss_bc": 0.03005073592066765, "train_loss_llm": 0.0, "grad_norm": 0.0258431825786829, "global_step": 15817, "epoch": 439, "lr": 0.001} +{"train_loss": 0.041486237198114395, "train_loss_bc": 0.041486237198114395, "train_loss_llm": 0.0, "grad_norm": 0.04543294012546539, "global_step": 15818, "epoch": 439, "lr": 0.001} +{"train_loss": 0.036730848252773285, "train_loss_bc": 0.03671012818813324, "train_loss_llm": 0.0020719170570373535, "grad_norm": 0.0629449188709259, "global_step": 15819, "epoch": 439, "lr": 0.001} +{"train_loss": 0.035813309252262115, "train_loss_bc": 0.035808056592941284, "train_loss_llm": 0.0005252361297607422, "grad_norm": 0.08443741500377655, "global_step": 15820, "epoch": 439, "lr": 0.001} +{"train_loss": 0.038464829325675964, "train_loss_bc": 0.038464829325675964, "train_loss_llm": 0.0, "grad_norm": 0.02987002395093441, "global_step": 15821, "epoch": 439, "lr": 0.001} +{"train_loss": 0.040416959673166275, "train_loss_bc": 0.04039778932929039, "train_loss_llm": 0.0019170045852661133, "grad_norm": 0.057191506028175354, "global_step": 15822, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03579786792397499, "train_loss_bc": 0.035790327936410904, "train_loss_llm": 0.0007540881633758545, "grad_norm": 0.06146787106990814, "global_step": 15823, "epoch": 439, "lr": 0.001} +{"train_loss": 0.02992052212357521, "train_loss_bc": 0.02992052212357521, "train_loss_llm": 0.0, "grad_norm": 0.08558593690395355, "global_step": 15824, "epoch": 439, "lr": 0.001} +{"train_loss": 0.043032675981521606, "train_loss_bc": 0.043032675981521606, "train_loss_llm": 0.0, "grad_norm": 0.03794126957654953, "global_step": 15825, "epoch": 439, "lr": 0.001} +{"train_loss": 0.0314154252409935, "train_loss_bc": 0.0314154252409935, "train_loss_llm": 0.0, "grad_norm": 0.0590563528239727, "global_step": 15826, "epoch": 439, "lr": 0.001} +{"train_loss": 0.04293133318424225, "train_loss_bc": 0.04293130710721016, "train_loss_llm": 2.592802047729492e-06, "grad_norm": 0.09070619940757751, "global_step": 15827, "epoch": 439, "lr": 0.001} +{"train_loss": 0.02826688438653946, "train_loss_bc": 0.02826688438653946, "train_loss_llm": 0.0, "grad_norm": 0.10233606398105621, "global_step": 15828, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03836576268076897, "train_loss_bc": 0.03836576268076897, "train_loss_llm": 0.0, "grad_norm": 0.029505370184779167, "global_step": 15829, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03213338181376457, "train_loss_bc": 0.03213338181376457, "train_loss_llm": 0.0, "grad_norm": 0.04414012283086777, "global_step": 15830, "epoch": 439, "lr": 0.001} +{"train_loss": 0.033626630902290344, "train_loss_bc": 0.033626630902290344, "train_loss_llm": 0.0, "grad_norm": 0.06234467774629593, "global_step": 15831, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03431692346930504, "train_loss_bc": 0.03431692346930504, "train_loss_llm": 0.0, "grad_norm": 0.08069829642772675, "global_step": 15832, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03380822390317917, "train_loss_bc": 0.033806491643190384, "train_loss_llm": 0.00017306208610534668, "grad_norm": 0.023805413395166397, "global_step": 15833, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03512516990303993, "train_loss_bc": 0.03511202335357666, "train_loss_llm": 0.0013145506381988525, "grad_norm": 0.03676127642393112, "global_step": 15834, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03723713383078575, "train_loss_bc": 0.03723713383078575, "train_loss_llm": 0.0, "grad_norm": 0.05461862310767174, "global_step": 15835, "epoch": 439, "lr": 0.001} +{"train_loss": 0.0298454649746418, "train_loss_bc": 0.0298454649746418, "train_loss_llm": 0.0, "grad_norm": 0.06677225232124329, "global_step": 15836, "epoch": 439, "lr": 0.001} +{"train_loss": 0.041420623660087585, "train_loss_bc": 0.041420623660087585, "train_loss_llm": 0.0, "grad_norm": 0.032285600900650024, "global_step": 15837, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03370051085948944, "train_loss_bc": 0.03370051085948944, "train_loss_llm": 0.0, "grad_norm": 0.040834907442331314, "global_step": 15838, "epoch": 439, "lr": 0.001} +{"train_loss": 0.03548638584713141, "train_loss_bc": 0.03603189066052437, "train_loss_llm": 0.0, "grad_norm": 0.06170228123664856, "global_step": 15839, "epoch": 439, "lr": 0.001, "val_loss": 0.07095244526863098} +{"train_loss": 0.04557817429304123, "train_loss_bc": 0.04556230083107948, "train_loss_llm": 0.0015872716903686523, "grad_norm": 0.08202510327100754, "global_step": 15840, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03785892575979233, "train_loss_bc": 0.03785892575979233, "train_loss_llm": 0.0, "grad_norm": 0.03254418820142746, "global_step": 15841, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03654054179787636, "train_loss_bc": 0.03653779253363609, "train_loss_llm": 0.00027483701705932617, "grad_norm": 0.04854313284158707, "global_step": 15842, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03615137189626694, "train_loss_bc": 0.03615137189626694, "train_loss_llm": 0.0, "grad_norm": 0.07133521884679794, "global_step": 15843, "epoch": 440, "lr": 0.001} +{"train_loss": 0.02963058464229107, "train_loss_bc": 0.029620740562677383, "train_loss_llm": 0.000984489917755127, "grad_norm": 0.08306113630533218, "global_step": 15844, "epoch": 440, "lr": 0.001} +{"train_loss": 0.04054843261837959, "train_loss_bc": 0.04054843261837959, "train_loss_llm": 0.0, "grad_norm": 0.032389361411333084, "global_step": 15845, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03531729429960251, "train_loss_bc": 0.03531729429960251, "train_loss_llm": 0.0, "grad_norm": 0.04968153312802315, "global_step": 15846, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03436078503727913, "train_loss_bc": 0.03436078503727913, "train_loss_llm": 0.0, "grad_norm": 0.06086686626076698, "global_step": 15847, "epoch": 440, "lr": 0.001} +{"train_loss": 0.0314377099275589, "train_loss_bc": 0.0314377099275589, "train_loss_llm": 0.0, "grad_norm": 0.07937554270029068, "global_step": 15848, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03665512055158615, "train_loss_bc": 0.03665512055158615, "train_loss_llm": 0.0, "grad_norm": 0.026086829602718353, "global_step": 15849, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03592545539140701, "train_loss_bc": 0.03592545539140701, "train_loss_llm": 0.0, "grad_norm": 0.038570940494537354, "global_step": 15850, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03436229005455971, "train_loss_bc": 0.03436229005455971, "train_loss_llm": 0.0, "grad_norm": 0.05277364328503609, "global_step": 15851, "epoch": 440, "lr": 0.001} +{"train_loss": 0.036030448973178864, "train_loss_bc": 0.036030448973178864, "train_loss_llm": 0.0, "grad_norm": 0.06679054349660873, "global_step": 15852, "epoch": 440, "lr": 0.001} +{"train_loss": 0.02945532277226448, "train_loss_bc": 0.02945532277226448, "train_loss_llm": 0.0, "grad_norm": 0.02663148194551468, "global_step": 15853, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03086364082992077, "train_loss_bc": 0.03086364082992077, "train_loss_llm": 0.0, "grad_norm": 0.03966651111841202, "global_step": 15854, "epoch": 440, "lr": 0.001} +{"train_loss": 0.031478606164455414, "train_loss_bc": 0.03146403282880783, "train_loss_llm": 0.0014573931694030762, "grad_norm": 0.05034872144460678, "global_step": 15855, "epoch": 440, "lr": 0.001} +{"train_loss": 0.04005107283592224, "train_loss_bc": 0.04005107283592224, "train_loss_llm": 0.0, "grad_norm": 0.05799556151032448, "global_step": 15856, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03560669347643852, "train_loss_bc": 0.03560669347643852, "train_loss_llm": 0.0, "grad_norm": 0.019231021404266357, "global_step": 15857, "epoch": 440, "lr": 0.001} +{"train_loss": 0.027098650112748146, "train_loss_bc": 0.027094675227999687, "train_loss_llm": 0.0003974735736846924, "grad_norm": 0.026953011751174927, "global_step": 15858, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03559344634413719, "train_loss_bc": 0.03559344634413719, "train_loss_llm": 0.0, "grad_norm": 0.0362495519220829, "global_step": 15859, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03617645055055618, "train_loss_bc": 0.03617645055055618, "train_loss_llm": 0.0, "grad_norm": 0.04957208409905434, "global_step": 15860, "epoch": 440, "lr": 0.001} +{"train_loss": 0.04029231145977974, "train_loss_bc": 0.04029231145977974, "train_loss_llm": 0.0, "grad_norm": 0.030427061021327972, "global_step": 15861, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03407638147473335, "train_loss_bc": 0.03407638147473335, "train_loss_llm": 0.0, "grad_norm": 0.04434821382164955, "global_step": 15862, "epoch": 440, "lr": 0.001} +{"train_loss": 0.04076211526989937, "train_loss_bc": 0.04076211526989937, "train_loss_llm": 0.0, "grad_norm": 0.0539252832531929, "global_step": 15863, "epoch": 440, "lr": 0.001} +{"train_loss": 0.034779343754053116, "train_loss_bc": 0.034779343754053116, "train_loss_llm": 0.0, "grad_norm": 0.0664929747581482, "global_step": 15864, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03224717453122139, "train_loss_bc": 0.03224717453122139, "train_loss_llm": 0.0, "grad_norm": 0.02083699032664299, "global_step": 15865, "epoch": 440, "lr": 0.001} +{"train_loss": 0.036640558391809464, "train_loss_bc": 0.03663378208875656, "train_loss_llm": 0.0006777644157409668, "grad_norm": 0.04142516851425171, "global_step": 15866, "epoch": 440, "lr": 0.001} +{"train_loss": 0.033615123480558395, "train_loss_bc": 0.033615123480558395, "train_loss_llm": 0.0, "grad_norm": 0.05913360416889191, "global_step": 15867, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03499167039990425, "train_loss_bc": 0.03499167039990425, "train_loss_llm": 0.0, "grad_norm": 0.07318130135536194, "global_step": 15868, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03701512888073921, "train_loss_bc": 0.03701512888073921, "train_loss_llm": 0.0, "grad_norm": 0.022027598693966866, "global_step": 15869, "epoch": 440, "lr": 0.001} +{"train_loss": 0.03510632738471031, "train_loss_bc": 0.03510632738471031, "train_loss_llm": 0.0, "grad_norm": 0.03275793045759201, "global_step": 15870, "epoch": 440, "lr": 0.001} +{"train_loss": 0.035813264548778534, "train_loss_bc": 0.03580014780163765, "train_loss_llm": 0.0013115406036376953, "grad_norm": 0.04913686588406563, "global_step": 15871, "epoch": 440, "lr": 0.001} +{"train_loss": 0.029022525995969772, "train_loss_bc": 0.02900633215904236, "train_loss_llm": 0.0016193687915802002, "grad_norm": 0.05967063456773758, "global_step": 15872, "epoch": 440, "lr": 0.001} +{"train_loss": 0.034963008016347885, "train_loss_bc": 0.03495102748274803, "train_loss_llm": 0.001197904348373413, "grad_norm": 0.02942614071071148, "global_step": 15873, "epoch": 440, "lr": 0.001} +{"train_loss": 0.04156522452831268, "train_loss_bc": 0.04155820608139038, "train_loss_llm": 0.0007018744945526123, "grad_norm": 0.051038358360528946, "global_step": 15874, "epoch": 440, "lr": 0.001} +{"train_loss": 0.035170188173651695, "train_loss_bc": 0.028499722480773926, "train_loss_llm": 0.001587599515914917, "grad_norm": 0.06219305470585823, "global_step": 15875, "epoch": 440, "lr": 0.001, "train/cumulative_reward": 69.22107190806832, "train/mean_score": 10.809891743319124, "train/success_rate": 0.2, "test/cumulative_reward": 43.09696697628503, "test/mean_score": 6.404758848963208, "test/success_rate": 0.1, "val_loss": 0.07031993567943573, "train_action_mse_error": 0.038038138300180435} +{"train_loss": 0.03892146423459053, "train_loss_bc": 0.03892146423459053, "train_loss_llm": 0.0, "grad_norm": 0.06724061816930771, "global_step": 15876, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03454393893480301, "train_loss_bc": 0.03454034775495529, "train_loss_llm": 0.0003592371940612793, "grad_norm": 0.035413920879364014, "global_step": 15877, "epoch": 441, "lr": 0.001} +{"train_loss": 0.035254329442977905, "train_loss_bc": 0.035254329442977905, "train_loss_llm": 0.0, "grad_norm": 0.05865296721458435, "global_step": 15878, "epoch": 441, "lr": 0.001} +{"train_loss": 0.0419364869594574, "train_loss_bc": 0.0419364869594574, "train_loss_llm": 0.0, "grad_norm": 0.0619497150182724, "global_step": 15879, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03309442102909088, "train_loss_bc": 0.03308809921145439, "train_loss_llm": 0.0006320774555206299, "grad_norm": 0.07405831664800644, "global_step": 15880, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03408720716834068, "train_loss_bc": 0.03408720716834068, "train_loss_llm": 0.0, "grad_norm": 0.023547174409031868, "global_step": 15881, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03559928387403488, "train_loss_bc": 0.03558740019798279, "train_loss_llm": 0.0011883080005645752, "grad_norm": 0.03535152226686478, "global_step": 15882, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03175436705350876, "train_loss_bc": 0.03175436705350876, "train_loss_llm": 0.0, "grad_norm": 0.04160624369978905, "global_step": 15883, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03197527676820755, "train_loss_bc": 0.03197527676820755, "train_loss_llm": 0.0, "grad_norm": 0.052474066615104675, "global_step": 15884, "epoch": 441, "lr": 0.001} +{"train_loss": 0.039946772158145905, "train_loss_bc": 0.039946772158145905, "train_loss_llm": 0.0, "grad_norm": 0.02970113418996334, "global_step": 15885, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03394671529531479, "train_loss_bc": 0.03394145518541336, "train_loss_llm": 0.0005259513854980469, "grad_norm": 0.04676223546266556, "global_step": 15886, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03361023962497711, "train_loss_bc": 0.03361023962497711, "train_loss_llm": 0.0, "grad_norm": 0.057913888245821, "global_step": 15887, "epoch": 441, "lr": 0.001} +{"train_loss": 0.029413042590022087, "train_loss_bc": 0.029413042590022087, "train_loss_llm": 0.0, "grad_norm": 0.06603968143463135, "global_step": 15888, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03837921842932701, "train_loss_bc": 0.03836379572749138, "train_loss_llm": 0.0015422701835632324, "grad_norm": 0.030941801145672798, "global_step": 15889, "epoch": 441, "lr": 0.001} +{"train_loss": 0.04310264065861702, "train_loss_bc": 0.04310264065861702, "train_loss_llm": 0.0, "grad_norm": 0.042278390377759933, "global_step": 15890, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03350766375660896, "train_loss_bc": 0.03347856551408768, "train_loss_llm": 0.002909719944000244, "grad_norm": 0.05608231946825981, "global_step": 15891, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03645601123571396, "train_loss_bc": 0.036448653787374496, "train_loss_llm": 0.0007358789443969727, "grad_norm": 0.07063895463943481, "global_step": 15892, "epoch": 441, "lr": 0.001} +{"train_loss": 0.028886595740914345, "train_loss_bc": 0.028874248266220093, "train_loss_llm": 0.0012347102165222168, "grad_norm": 0.021053068339824677, "global_step": 15893, "epoch": 441, "lr": 0.001} +{"train_loss": 0.04865974187850952, "train_loss_bc": 0.04865974187850952, "train_loss_llm": 0.0, "grad_norm": 0.045724328607320786, "global_step": 15894, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03987662121653557, "train_loss_bc": 0.03987662121653557, "train_loss_llm": 0.0, "grad_norm": 0.06437838077545166, "global_step": 15895, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03741452097892761, "train_loss_bc": 0.03741452097892761, "train_loss_llm": 0.0, "grad_norm": 0.0808621197938919, "global_step": 15896, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03296852111816406, "train_loss_bc": 0.03296852111816406, "train_loss_llm": 0.0, "grad_norm": 0.02211080864071846, "global_step": 15897, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03420352190732956, "train_loss_bc": 0.03420352190732956, "train_loss_llm": 0.0, "grad_norm": 0.032161712646484375, "global_step": 15898, "epoch": 441, "lr": 0.001} +{"train_loss": 0.030135037377476692, "train_loss_bc": 0.030135037377476692, "train_loss_llm": 0.0, "grad_norm": 0.049621377140283585, "global_step": 15899, "epoch": 441, "lr": 0.001} +{"train_loss": 0.039365511387586594, "train_loss_bc": 0.03936208784580231, "train_loss_llm": 0.00034239888191223145, "grad_norm": 0.06555807590484619, "global_step": 15900, "epoch": 441, "lr": 0.001} +{"train_loss": 0.0350683256983757, "train_loss_bc": 0.0350634828209877, "train_loss_llm": 0.00048419833183288574, "grad_norm": 0.025135891512036324, "global_step": 15901, "epoch": 441, "lr": 0.001} +{"train_loss": 0.0372459813952446, "train_loss_bc": 0.03723667562007904, "train_loss_llm": 0.0009307563304901123, "grad_norm": 0.04430294781923294, "global_step": 15902, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03064466454088688, "train_loss_bc": 0.030641017481684685, "train_loss_llm": 0.00036469101905822754, "grad_norm": 0.057152554392814636, "global_step": 15903, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03574365749955177, "train_loss_bc": 0.03574365749955177, "train_loss_llm": 0.0, "grad_norm": 0.06894824653863907, "global_step": 15904, "epoch": 441, "lr": 0.001} +{"train_loss": 0.02763253264129162, "train_loss_bc": 0.02763253264129162, "train_loss_llm": 0.0, "grad_norm": 0.02686147391796112, "global_step": 15905, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03585660830140114, "train_loss_bc": 0.03585660830140114, "train_loss_llm": 0.0, "grad_norm": 0.04355354979634285, "global_step": 15906, "epoch": 441, "lr": 0.001} +{"train_loss": 0.033180344849824905, "train_loss_bc": 0.033180344849824905, "train_loss_llm": 0.0, "grad_norm": 0.05566062033176422, "global_step": 15907, "epoch": 441, "lr": 0.001} +{"train_loss": 0.035802748054265976, "train_loss_bc": 0.035802748054265976, "train_loss_llm": 0.0, "grad_norm": 0.06874574720859528, "global_step": 15908, "epoch": 441, "lr": 0.001} +{"train_loss": 0.030810071155428886, "train_loss_bc": 0.030810071155428886, "train_loss_llm": 0.0, "grad_norm": 0.02416975237429142, "global_step": 15909, "epoch": 441, "lr": 0.001} +{"train_loss": 0.029698116704821587, "train_loss_bc": 0.029698116704821587, "train_loss_llm": 0.0, "grad_norm": 0.03253156319260597, "global_step": 15910, "epoch": 441, "lr": 0.001} +{"train_loss": 0.03515444297550453, "train_loss_bc": 0.03683774545788765, "train_loss_llm": 0.0, "grad_norm": 0.040378812700510025, "global_step": 15911, "epoch": 441, "lr": 0.001, "val_loss": 0.07170011848211288} +{"train_loss": 0.033735353499650955, "train_loss_bc": 0.033735353499650955, "train_loss_llm": 0.0, "grad_norm": 0.04838644713163376, "global_step": 15912, "epoch": 442, "lr": 0.001} +{"train_loss": 0.029180631041526794, "train_loss_bc": 0.029180631041526794, "train_loss_llm": 0.0, "grad_norm": 0.022096751257777214, "global_step": 15913, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03461224585771561, "train_loss_bc": 0.03461224585771561, "train_loss_llm": 0.0, "grad_norm": 0.03544962778687477, "global_step": 15914, "epoch": 442, "lr": 0.001} +{"train_loss": 0.037250250577926636, "train_loss_bc": 0.037250250577926636, "train_loss_llm": 0.0, "grad_norm": 0.0469047836959362, "global_step": 15915, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03581676259636879, "train_loss_bc": 0.03581676259636879, "train_loss_llm": 0.0, "grad_norm": 0.060082338750362396, "global_step": 15916, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03661780431866646, "train_loss_bc": 0.03661780431866646, "train_loss_llm": 0.0, "grad_norm": 0.025023121386766434, "global_step": 15917, "epoch": 442, "lr": 0.001} +{"train_loss": 0.0381169393658638, "train_loss_bc": 0.038105081766843796, "train_loss_llm": 0.001185595989227295, "grad_norm": 0.041479676961898804, "global_step": 15918, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03893575444817543, "train_loss_bc": 0.038933414965867996, "train_loss_llm": 0.00023406744003295898, "grad_norm": 0.05304567515850067, "global_step": 15919, "epoch": 442, "lr": 0.001} +{"train_loss": 0.02972889319062233, "train_loss_bc": 0.02972889319062233, "train_loss_llm": 0.0, "grad_norm": 0.0676233172416687, "global_step": 15920, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03350042924284935, "train_loss_bc": 0.03349555656313896, "train_loss_llm": 0.00048744678497314453, "grad_norm": 0.02411874756217003, "global_step": 15921, "epoch": 442, "lr": 0.001} +{"train_loss": 0.032545436173677444, "train_loss_bc": 0.032545436173677444, "train_loss_llm": 0.0, "grad_norm": 0.03420742601156235, "global_step": 15922, "epoch": 442, "lr": 0.001} +{"train_loss": 0.035711150616407394, "train_loss_bc": 0.035711150616407394, "train_loss_llm": 0.0, "grad_norm": 0.04520811885595322, "global_step": 15923, "epoch": 442, "lr": 0.001} +{"train_loss": 0.04055400937795639, "train_loss_bc": 0.04055400937795639, "train_loss_llm": 0.0, "grad_norm": 0.06622419506311417, "global_step": 15924, "epoch": 442, "lr": 0.001} +{"train_loss": 0.028677647933363914, "train_loss_bc": 0.02867732010781765, "train_loss_llm": 3.275275230407715e-05, "grad_norm": 0.022789837792515755, "global_step": 15925, "epoch": 442, "lr": 0.001} +{"train_loss": 0.04090461507439613, "train_loss_bc": 0.04088855907320976, "train_loss_llm": 0.0016057789325714111, "grad_norm": 0.03452977165579796, "global_step": 15926, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03417011350393295, "train_loss_bc": 0.03417011350393295, "train_loss_llm": 0.0, "grad_norm": 0.046465806663036346, "global_step": 15927, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03189077973365784, "train_loss_bc": 0.03188148885965347, "train_loss_llm": 0.0009292364120483398, "grad_norm": 0.06570267677307129, "global_step": 15928, "epoch": 442, "lr": 0.001} +{"train_loss": 0.031888026744127274, "train_loss_bc": 0.031888026744127274, "train_loss_llm": 0.0, "grad_norm": 0.027644937857985497, "global_step": 15929, "epoch": 442, "lr": 0.001} +{"train_loss": 0.04164735600352287, "train_loss_bc": 0.04164735600352287, "train_loss_llm": 0.0, "grad_norm": 0.05225801840424538, "global_step": 15930, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03524891659617424, "train_loss_bc": 0.035236768424510956, "train_loss_llm": 0.0012148916721343994, "grad_norm": 0.06368815153837204, "global_step": 15931, "epoch": 442, "lr": 0.001} +{"train_loss": 0.02884647250175476, "train_loss_bc": 0.02884647250175476, "train_loss_llm": 0.0, "grad_norm": 0.07397108525037766, "global_step": 15932, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03998999670147896, "train_loss_bc": 0.039973802864551544, "train_loss_llm": 0.0016192793846130371, "grad_norm": 0.025177352130413055, "global_step": 15933, "epoch": 442, "lr": 0.001} +{"train_loss": 0.036474961787462234, "train_loss_bc": 0.036474961787462234, "train_loss_llm": 0.0, "grad_norm": 0.03847377002239227, "global_step": 15934, "epoch": 442, "lr": 0.001} +{"train_loss": 0.028956789523363113, "train_loss_bc": 0.0289445910602808, "train_loss_llm": 0.0012198090553283691, "grad_norm": 0.043659508228302, "global_step": 15935, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03498091921210289, "train_loss_bc": 0.03498091921210289, "train_loss_llm": 0.0, "grad_norm": 0.05576811730861664, "global_step": 15936, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03359372168779373, "train_loss_bc": 0.03359372168779373, "train_loss_llm": 0.0, "grad_norm": 0.023887187242507935, "global_step": 15937, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03857887163758278, "train_loss_bc": 0.03857887163758278, "train_loss_llm": 0.0, "grad_norm": 0.04107574373483658, "global_step": 15938, "epoch": 442, "lr": 0.001} +{"train_loss": 0.0437578447163105, "train_loss_bc": 0.04375508427619934, "train_loss_llm": 0.0002760887145996094, "grad_norm": 0.05320390313863754, "global_step": 15939, "epoch": 442, "lr": 0.001} +{"train_loss": 0.031539831310510635, "train_loss_bc": 0.03153793513774872, "train_loss_llm": 0.0001894533634185791, "grad_norm": 0.06487257033586502, "global_step": 15940, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03788555786013603, "train_loss_bc": 0.03788555786013603, "train_loss_llm": 0.0, "grad_norm": 0.0294895488768816, "global_step": 15941, "epoch": 442, "lr": 0.001} +{"train_loss": 0.035477861762046814, "train_loss_bc": 0.035477861762046814, "train_loss_llm": 0.0, "grad_norm": 0.035828862339258194, "global_step": 15942, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03166235610842705, "train_loss_bc": 0.03166235610842705, "train_loss_llm": 0.0, "grad_norm": 0.04586763679981232, "global_step": 15943, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03727743774652481, "train_loss_bc": 0.03727743774652481, "train_loss_llm": 0.0, "grad_norm": 0.058662667870521545, "global_step": 15944, "epoch": 442, "lr": 0.001} +{"train_loss": 0.038124848157167435, "train_loss_bc": 0.0381225124001503, "train_loss_llm": 0.00023356080055236816, "grad_norm": 0.031522057950496674, "global_step": 15945, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03583377227187157, "train_loss_bc": 0.03583377227187157, "train_loss_llm": 0.0, "grad_norm": 0.04200632870197296, "global_step": 15946, "epoch": 442, "lr": 0.001} +{"train_loss": 0.03498855900640289, "train_loss_bc": 0.025870637968182564, "train_loss_llm": 0.0003127753734588623, "grad_norm": 0.05864142253994942, "global_step": 15947, "epoch": 442, "lr": 0.001, "val_loss": 0.07002531737089157} +{"train_loss": 0.035481713712215424, "train_loss_bc": 0.035481713712215424, "train_loss_llm": 0.0, "grad_norm": 0.08111277967691422, "global_step": 15948, "epoch": 443, "lr": 0.001} +{"train_loss": 0.034037940204143524, "train_loss_bc": 0.034037940204143524, "train_loss_llm": 0.0, "grad_norm": 0.028295550495386124, "global_step": 15949, "epoch": 443, "lr": 0.001} +{"train_loss": 0.029474005103111267, "train_loss_bc": 0.02947189472615719, "train_loss_llm": 0.0002110004425048828, "grad_norm": 0.038132574409246445, "global_step": 15950, "epoch": 443, "lr": 0.001} +{"train_loss": 0.0333770290017128, "train_loss_bc": 0.0333770290017128, "train_loss_llm": 0.0, "grad_norm": 0.04724203795194626, "global_step": 15951, "epoch": 443, "lr": 0.001} +{"train_loss": 0.030197229236364365, "train_loss_bc": 0.030197229236364365, "train_loss_llm": 0.0, "grad_norm": 0.05219632014632225, "global_step": 15952, "epoch": 443, "lr": 0.001} +{"train_loss": 0.036660030484199524, "train_loss_bc": 0.036660030484199524, "train_loss_llm": 0.0, "grad_norm": 0.023257771506905556, "global_step": 15953, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03552166000008583, "train_loss_bc": 0.03552166000008583, "train_loss_llm": 0.0, "grad_norm": 0.03556804358959198, "global_step": 15954, "epoch": 443, "lr": 0.001} +{"train_loss": 0.02459343522787094, "train_loss_bc": 0.024582332000136375, "train_loss_llm": 0.0011102557182312012, "grad_norm": 0.05234174057841301, "global_step": 15955, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03939371928572655, "train_loss_bc": 0.03939371928572655, "train_loss_llm": 0.0, "grad_norm": 0.07559073716402054, "global_step": 15956, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03481830283999443, "train_loss_bc": 0.03481830283999443, "train_loss_llm": 0.0, "grad_norm": 0.024740522727370262, "global_step": 15957, "epoch": 443, "lr": 0.001} +{"train_loss": 0.04183048754930496, "train_loss_bc": 0.04183048754930496, "train_loss_llm": 0.0, "grad_norm": 0.03978752717375755, "global_step": 15958, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03957892954349518, "train_loss_bc": 0.03957892954349518, "train_loss_llm": 0.0, "grad_norm": 0.05041026696562767, "global_step": 15959, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03889850154519081, "train_loss_bc": 0.03888823837041855, "train_loss_llm": 0.0010263025760650635, "grad_norm": 0.05978402867913246, "global_step": 15960, "epoch": 443, "lr": 0.001} +{"train_loss": 0.040036458522081375, "train_loss_bc": 0.040036458522081375, "train_loss_llm": 0.0, "grad_norm": 0.020577644929289818, "global_step": 15961, "epoch": 443, "lr": 0.001} +{"train_loss": 0.029413728043437004, "train_loss_bc": 0.029413728043437004, "train_loss_llm": 0.0, "grad_norm": 0.03385550156235695, "global_step": 15962, "epoch": 443, "lr": 0.001} +{"train_loss": 0.026076288893818855, "train_loss_bc": 0.026076288893818855, "train_loss_llm": 0.0, "grad_norm": 0.04483916237950325, "global_step": 15963, "epoch": 443, "lr": 0.001} +{"train_loss": 0.027653098106384277, "train_loss_bc": 0.027653098106384277, "train_loss_llm": 0.0, "grad_norm": 0.056892018765211105, "global_step": 15964, "epoch": 443, "lr": 0.001} +{"train_loss": 0.04225998744368553, "train_loss_bc": 0.042242471128702164, "train_loss_llm": 0.001751631498336792, "grad_norm": 0.02509705349802971, "global_step": 15965, "epoch": 443, "lr": 0.001} +{"train_loss": 0.026916641741991043, "train_loss_bc": 0.026893863454461098, "train_loss_llm": 0.002277851104736328, "grad_norm": 0.03785223513841629, "global_step": 15966, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03660358488559723, "train_loss_bc": 0.03660358488559723, "train_loss_llm": 0.0, "grad_norm": 0.052476659417152405, "global_step": 15967, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03910104185342789, "train_loss_bc": 0.039100319147109985, "train_loss_llm": 7.224082946777344e-05, "grad_norm": 0.06639354676008224, "global_step": 15968, "epoch": 443, "lr": 0.001} +{"train_loss": 0.029873110353946686, "train_loss_bc": 0.029836369678378105, "train_loss_llm": 0.003674060106277466, "grad_norm": 0.020588045939803123, "global_step": 15969, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03532318398356438, "train_loss_bc": 0.03530428558588028, "train_loss_llm": 0.001889944076538086, "grad_norm": 0.025900257751345634, "global_step": 15970, "epoch": 443, "lr": 0.001} +{"train_loss": 0.02872657962143421, "train_loss_bc": 0.02872657962143421, "train_loss_llm": 0.0, "grad_norm": 0.03315481171011925, "global_step": 15971, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03721599280834198, "train_loss_bc": 0.03721599280834198, "train_loss_llm": 0.0, "grad_norm": 0.05246305093169212, "global_step": 15972, "epoch": 443, "lr": 0.001} +{"train_loss": 0.04016628861427307, "train_loss_bc": 0.04016628861427307, "train_loss_llm": 0.0, "grad_norm": 0.02633846178650856, "global_step": 15973, "epoch": 443, "lr": 0.001} +{"train_loss": 0.04270338639616966, "train_loss_bc": 0.04268590360879898, "train_loss_llm": 0.0017483532428741455, "grad_norm": 0.039928171783685684, "global_step": 15974, "epoch": 443, "lr": 0.001} +{"train_loss": 0.04044358432292938, "train_loss_bc": 0.04044358432292938, "train_loss_llm": 0.0, "grad_norm": 0.05402973294258118, "global_step": 15975, "epoch": 443, "lr": 0.001} +{"train_loss": 0.04038311913609505, "train_loss_bc": 0.04038311913609505, "train_loss_llm": 0.0, "grad_norm": 0.07357587665319443, "global_step": 15976, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03685065358877182, "train_loss_bc": 0.03685065358877182, "train_loss_llm": 0.0, "grad_norm": 0.02929413691163063, "global_step": 15977, "epoch": 443, "lr": 0.001} +{"train_loss": 0.040742337703704834, "train_loss_bc": 0.040742337703704834, "train_loss_llm": 0.0, "grad_norm": 0.03788067400455475, "global_step": 15978, "epoch": 443, "lr": 0.001} +{"train_loss": 0.035230543464422226, "train_loss_bc": 0.035230543464422226, "train_loss_llm": 0.0, "grad_norm": 0.047771986573934555, "global_step": 15979, "epoch": 443, "lr": 0.001} +{"train_loss": 0.032789554446935654, "train_loss_bc": 0.032789554446935654, "train_loss_llm": 0.0, "grad_norm": 0.05558419227600098, "global_step": 15980, "epoch": 443, "lr": 0.001} +{"train_loss": 0.036439161747694016, "train_loss_bc": 0.036439161747694016, "train_loss_llm": 0.0, "grad_norm": 0.03173983842134476, "global_step": 15981, "epoch": 443, "lr": 0.001} +{"train_loss": 0.029066527262330055, "train_loss_bc": 0.029066527262330055, "train_loss_llm": 0.0, "grad_norm": 0.043872375041246414, "global_step": 15982, "epoch": 443, "lr": 0.001} +{"train_loss": 0.03495476901945141, "train_loss_bc": 0.030493848025798798, "train_loss_llm": 0.0, "grad_norm": 0.055903639644384384, "global_step": 15983, "epoch": 443, "lr": 0.001, "val_loss": 0.06986676156520844} +{"train_loss": 0.03338455408811569, "train_loss_bc": 0.03338455408811569, "train_loss_llm": 0.0, "grad_norm": 0.07307087630033493, "global_step": 15984, "epoch": 444, "lr": 0.001} +{"train_loss": 0.035087957978248596, "train_loss_bc": 0.035075198858976364, "train_loss_llm": 0.0012757480144500732, "grad_norm": 0.02733076550066471, "global_step": 15985, "epoch": 444, "lr": 0.001} +{"train_loss": 0.041540514677762985, "train_loss_bc": 0.041540514677762985, "train_loss_llm": 0.0, "grad_norm": 0.041695479303598404, "global_step": 15986, "epoch": 444, "lr": 0.001} +{"train_loss": 0.036105141043663025, "train_loss_bc": 0.036105141043663025, "train_loss_llm": 0.0, "grad_norm": 0.047624047845602036, "global_step": 15987, "epoch": 444, "lr": 0.001} +{"train_loss": 0.032199427485466, "train_loss_bc": 0.032199427485466, "train_loss_llm": 0.0, "grad_norm": 0.060237817466259, "global_step": 15988, "epoch": 444, "lr": 0.001} +{"train_loss": 0.040156036615371704, "train_loss_bc": 0.040156036615371704, "train_loss_llm": 0.0, "grad_norm": 0.030031004920601845, "global_step": 15989, "epoch": 444, "lr": 0.001} +{"train_loss": 0.033687591552734375, "train_loss_bc": 0.033687591552734375, "train_loss_llm": 0.0, "grad_norm": 0.042065300047397614, "global_step": 15990, "epoch": 444, "lr": 0.001} +{"train_loss": 0.02626344934105873, "train_loss_bc": 0.02626344934105873, "train_loss_llm": 0.0, "grad_norm": 0.053600993007421494, "global_step": 15991, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03997818008065224, "train_loss_bc": 0.03997818008065224, "train_loss_llm": 0.0, "grad_norm": 0.07076240330934525, "global_step": 15992, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03348210081458092, "train_loss_bc": 0.03348210081458092, "train_loss_llm": 0.0, "grad_norm": 0.028559530153870583, "global_step": 15993, "epoch": 444, "lr": 0.001} +{"train_loss": 0.04507731273770332, "train_loss_bc": 0.045074909925460815, "train_loss_llm": 0.00024017691612243652, "grad_norm": 0.050932906568050385, "global_step": 15994, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03749954327940941, "train_loss_bc": 0.03749954327940941, "train_loss_llm": 0.0, "grad_norm": 0.07051974534988403, "global_step": 15995, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03427963703870773, "train_loss_bc": 0.03427963703870773, "train_loss_llm": 0.0, "grad_norm": 0.08413753658533096, "global_step": 15996, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03452247381210327, "train_loss_bc": 0.034515198320150375, "train_loss_llm": 0.0007276237010955811, "grad_norm": 0.024635514244437218, "global_step": 15997, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03502707555890083, "train_loss_bc": 0.03502707555890083, "train_loss_llm": 0.0, "grad_norm": 0.03903128206729889, "global_step": 15998, "epoch": 444, "lr": 0.001} +{"train_loss": 0.029719509184360504, "train_loss_bc": 0.029719509184360504, "train_loss_llm": 0.0, "grad_norm": 0.04740826413035393, "global_step": 15999, "epoch": 444, "lr": 0.001} +{"train_loss": 0.029329519718885422, "train_loss_bc": 0.029329519718885422, "train_loss_llm": 0.0, "grad_norm": 0.06130533292889595, "global_step": 16000, "epoch": 444, "lr": 0.001} +{"train_loss": 0.04226727411150932, "train_loss_bc": 0.04226727411150932, "train_loss_llm": 0.0, "grad_norm": 0.03030765801668167, "global_step": 16001, "epoch": 444, "lr": 0.001} +{"train_loss": 0.038397688418626785, "train_loss_bc": 0.038397688418626785, "train_loss_llm": 0.0, "grad_norm": 0.04903086647391319, "global_step": 16002, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03990344703197479, "train_loss_bc": 0.03990344703197479, "train_loss_llm": 0.0, "grad_norm": 0.06242541968822479, "global_step": 16003, "epoch": 444, "lr": 0.001} +{"train_loss": 0.029930101707577705, "train_loss_bc": 0.029930101707577705, "train_loss_llm": 0.0, "grad_norm": 0.06614599376916885, "global_step": 16004, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03363620489835739, "train_loss_bc": 0.03363620489835739, "train_loss_llm": 0.0, "grad_norm": 0.02226804569363594, "global_step": 16005, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03617066517472267, "train_loss_bc": 0.03615310415625572, "train_loss_llm": 0.0017562806606292725, "grad_norm": 0.03711644187569618, "global_step": 16006, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03434840962290764, "train_loss_bc": 0.03434840962290764, "train_loss_llm": 0.0, "grad_norm": 0.052719973027706146, "global_step": 16007, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03807868808507919, "train_loss_bc": 0.03803543746471405, "train_loss_llm": 0.004325240850448608, "grad_norm": 0.06445872783660889, "global_step": 16008, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03371092677116394, "train_loss_bc": 0.03369182348251343, "train_loss_llm": 0.0019104480743408203, "grad_norm": 0.030018409714102745, "global_step": 16009, "epoch": 444, "lr": 0.001} +{"train_loss": 0.036624446511268616, "train_loss_bc": 0.036594074219465256, "train_loss_llm": 0.0030372440814971924, "grad_norm": 0.050541963428258896, "global_step": 16010, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03824559971690178, "train_loss_bc": 0.03824559971690178, "train_loss_llm": 0.0, "grad_norm": 0.0864563062787056, "global_step": 16011, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03358161449432373, "train_loss_bc": 0.03358161449432373, "train_loss_llm": 0.0, "grad_norm": 0.1004604771733284, "global_step": 16012, "epoch": 444, "lr": 0.001} +{"train_loss": 0.029159437865018845, "train_loss_bc": 0.029152918606996536, "train_loss_llm": 0.0006519854068756104, "grad_norm": 0.02283812128007412, "global_step": 16013, "epoch": 444, "lr": 0.001} +{"train_loss": 0.02817097306251526, "train_loss_bc": 0.02817097306251526, "train_loss_llm": 0.0, "grad_norm": 0.028551682829856873, "global_step": 16014, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03375021740794182, "train_loss_bc": 0.03373271971940994, "train_loss_llm": 0.0017496943473815918, "grad_norm": 0.04187054932117462, "global_step": 16015, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03755441680550575, "train_loss_bc": 0.03755441680550575, "train_loss_llm": 0.0, "grad_norm": 0.05719498544931412, "global_step": 16016, "epoch": 444, "lr": 0.001} +{"train_loss": 0.034302968531847, "train_loss_bc": 0.034302968531847, "train_loss_llm": 0.0, "grad_norm": 0.032552070915699005, "global_step": 16017, "epoch": 444, "lr": 0.001} +{"train_loss": 0.037326861172914505, "train_loss_bc": 0.037326861172914505, "train_loss_llm": 0.0, "grad_norm": 0.0391932912170887, "global_step": 16018, "epoch": 444, "lr": 0.001} +{"train_loss": 0.03506447116119994, "train_loss_bc": 0.029820995405316353, "train_loss_llm": 0.0, "grad_norm": 0.05797695368528366, "global_step": 16019, "epoch": 444, "lr": 0.001, "val_loss": 0.07117588818073273} +{"train_loss": 0.03072318434715271, "train_loss_bc": 0.03072318434715271, "train_loss_llm": 0.0, "grad_norm": 0.06779918074607849, "global_step": 16020, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03776255622506142, "train_loss_bc": 0.03776255622506142, "train_loss_llm": 0.0, "grad_norm": 0.0289725624024868, "global_step": 16021, "epoch": 445, "lr": 0.001} +{"train_loss": 0.029958123341202736, "train_loss_bc": 0.029953572899103165, "train_loss_llm": 0.0004550814628601074, "grad_norm": 0.03738648444414139, "global_step": 16022, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03939869999885559, "train_loss_bc": 0.03939869999885559, "train_loss_llm": 0.0, "grad_norm": 0.05287400260567665, "global_step": 16023, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03404141589999199, "train_loss_bc": 0.03404141589999199, "train_loss_llm": 0.0, "grad_norm": 0.062274277210235596, "global_step": 16024, "epoch": 445, "lr": 0.001} +{"train_loss": 0.0328710600733757, "train_loss_bc": 0.032868098467588425, "train_loss_llm": 0.0002962350845336914, "grad_norm": 0.025701157748699188, "global_step": 16025, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03045416809618473, "train_loss_bc": 0.03045416809618473, "train_loss_llm": 0.0, "grad_norm": 0.036912232637405396, "global_step": 16026, "epoch": 445, "lr": 0.001} +{"train_loss": 0.034397877752780914, "train_loss_bc": 0.034397877752780914, "train_loss_llm": 0.0, "grad_norm": 0.04285088926553726, "global_step": 16027, "epoch": 445, "lr": 0.001} +{"train_loss": 0.033169303089380264, "train_loss_bc": 0.033169303089380264, "train_loss_llm": 0.0, "grad_norm": 0.06082438677549362, "global_step": 16028, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03167232871055603, "train_loss_bc": 0.03167232871055603, "train_loss_llm": 0.0, "grad_norm": 0.028109628707170486, "global_step": 16029, "epoch": 445, "lr": 0.001} +{"train_loss": 0.026347927749156952, "train_loss_bc": 0.026347927749156952, "train_loss_llm": 0.0, "grad_norm": 0.04001356288790703, "global_step": 16030, "epoch": 445, "lr": 0.001} +{"train_loss": 0.036848846822977066, "train_loss_bc": 0.036848846822977066, "train_loss_llm": 0.0, "grad_norm": 0.06033986806869507, "global_step": 16031, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03377991542220116, "train_loss_bc": 0.03377991542220116, "train_loss_llm": 0.0, "grad_norm": 0.06821481883525848, "global_step": 16032, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03991616889834404, "train_loss_bc": 0.03991616889834404, "train_loss_llm": 0.0, "grad_norm": 0.03829159587621689, "global_step": 16033, "epoch": 445, "lr": 0.001} +{"train_loss": 0.034215494990348816, "train_loss_bc": 0.034215494990348816, "train_loss_llm": 0.0, "grad_norm": 0.051263246685266495, "global_step": 16034, "epoch": 445, "lr": 0.001} +{"train_loss": 0.04019659385085106, "train_loss_bc": 0.04019659385085106, "train_loss_llm": 0.0, "grad_norm": 0.06740032136440277, "global_step": 16035, "epoch": 445, "lr": 0.001} +{"train_loss": 0.028510915115475655, "train_loss_bc": 0.028510915115475655, "train_loss_llm": 0.0, "grad_norm": 0.0848747044801712, "global_step": 16036, "epoch": 445, "lr": 0.001} +{"train_loss": 0.032199230045080185, "train_loss_bc": 0.032199230045080185, "train_loss_llm": 0.0, "grad_norm": 0.021260935813188553, "global_step": 16037, "epoch": 445, "lr": 0.001} +{"train_loss": 0.036547452211380005, "train_loss_bc": 0.036547452211380005, "train_loss_llm": 0.0, "grad_norm": 0.031189484521746635, "global_step": 16038, "epoch": 445, "lr": 0.001} +{"train_loss": 0.04128243401646614, "train_loss_bc": 0.04128243401646614, "train_loss_llm": 0.0, "grad_norm": 0.04529489204287529, "global_step": 16039, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03829698637127876, "train_loss_bc": 0.03829698637127876, "train_loss_llm": 0.0, "grad_norm": 0.05700254067778587, "global_step": 16040, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03415173664689064, "train_loss_bc": 0.034135788679122925, "train_loss_llm": 0.0015949606895446777, "grad_norm": 0.022434433922171593, "global_step": 16041, "epoch": 445, "lr": 0.001} +{"train_loss": 0.02589631825685501, "train_loss_bc": 0.025890132412314415, "train_loss_llm": 0.0006185472011566162, "grad_norm": 0.030144909396767616, "global_step": 16042, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03712262585759163, "train_loss_bc": 0.03712262585759163, "train_loss_llm": 0.0, "grad_norm": 0.04177433252334595, "global_step": 16043, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03533342480659485, "train_loss_bc": 0.03533342480659485, "train_loss_llm": 0.0, "grad_norm": 0.05110759660601616, "global_step": 16044, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03670605644583702, "train_loss_bc": 0.03670334070920944, "train_loss_llm": 0.000271528959274292, "grad_norm": 0.027274612337350845, "global_step": 16045, "epoch": 445, "lr": 0.001} +{"train_loss": 0.02627432905137539, "train_loss_bc": 0.0262643750756979, "train_loss_llm": 0.0009954869747161865, "grad_norm": 0.03839881345629692, "global_step": 16046, "epoch": 445, "lr": 0.001} +{"train_loss": 0.034124162048101425, "train_loss_bc": 0.034124162048101425, "train_loss_llm": 0.0, "grad_norm": 0.04964152351021767, "global_step": 16047, "epoch": 445, "lr": 0.001} +{"train_loss": 0.0389355830848217, "train_loss_bc": 0.03893547132611275, "train_loss_llm": 1.1056661605834961e-05, "grad_norm": 0.0624067597091198, "global_step": 16048, "epoch": 445, "lr": 0.001} +{"train_loss": 0.043018143624067307, "train_loss_bc": 0.043018143624067307, "train_loss_llm": 0.0, "grad_norm": 0.03211851790547371, "global_step": 16049, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03588300570845604, "train_loss_bc": 0.03586341068148613, "train_loss_llm": 0.0019593238830566406, "grad_norm": 0.04644878953695297, "global_step": 16050, "epoch": 445, "lr": 0.001} +{"train_loss": 0.031771816313266754, "train_loss_bc": 0.031771816313266754, "train_loss_llm": 0.0, "grad_norm": 0.06000794470310211, "global_step": 16051, "epoch": 445, "lr": 0.001} +{"train_loss": 0.039784155786037445, "train_loss_bc": 0.039763644337654114, "train_loss_llm": 0.0020512938499450684, "grad_norm": 0.07458285242319107, "global_step": 16052, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03635698929429054, "train_loss_bc": 0.036351967602968216, "train_loss_llm": 0.0005021393299102783, "grad_norm": 0.025136420503258705, "global_step": 16053, "epoch": 445, "lr": 0.001} +{"train_loss": 0.04549340531229973, "train_loss_bc": 0.04547000676393509, "train_loss_llm": 0.0023398995399475098, "grad_norm": 0.04506024345755577, "global_step": 16054, "epoch": 445, "lr": 0.001} +{"train_loss": 0.03505987084160248, "train_loss_bc": 0.038712915033102036, "train_loss_llm": 0.0, "grad_norm": 0.06753627955913544, "global_step": 16055, "epoch": 445, "lr": 0.001, "train/cumulative_reward": 93.90708080866588, "train/mean_score": 15.832974819776092, "train/success_rate": 0.4, "test/cumulative_reward": 45.48685736274566, "test/mean_score": 6.99189718879992, "test/success_rate": 0.14, "val_loss": 0.07114747166633606, "train_action_mse_error": 0.037921857088804245} +{"train_loss": 0.035445842891931534, "train_loss_bc": 0.035445842891931534, "train_loss_llm": 0.0, "grad_norm": 0.08535850793123245, "global_step": 16056, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03613438457250595, "train_loss_bc": 0.03613297641277313, "train_loss_llm": 0.00014090538024902344, "grad_norm": 0.028928937390446663, "global_step": 16057, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03262654319405556, "train_loss_bc": 0.03262654319405556, "train_loss_llm": 0.0, "grad_norm": 0.045697808265686035, "global_step": 16058, "epoch": 446, "lr": 0.001} +{"train_loss": 0.033930856734514236, "train_loss_bc": 0.03392874822020531, "train_loss_llm": 0.0002110004425048828, "grad_norm": 0.045451804995536804, "global_step": 16059, "epoch": 446, "lr": 0.001} +{"train_loss": 0.037768442183732986, "train_loss_bc": 0.03776305541396141, "train_loss_llm": 0.0005388557910919189, "grad_norm": 0.06469356268644333, "global_step": 16060, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03324934467673302, "train_loss_bc": 0.03324934467673302, "train_loss_llm": 0.0, "grad_norm": 0.02440955862402916, "global_step": 16061, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03542870283126831, "train_loss_bc": 0.03542870283126831, "train_loss_llm": 0.0, "grad_norm": 0.03613337129354477, "global_step": 16062, "epoch": 446, "lr": 0.001} +{"train_loss": 0.04271635785698891, "train_loss_bc": 0.04268992692232132, "train_loss_llm": 0.002643078565597534, "grad_norm": 0.05171693116426468, "global_step": 16063, "epoch": 446, "lr": 0.001} +{"train_loss": 0.040623053908348083, "train_loss_bc": 0.04061431065201759, "train_loss_llm": 0.0008741915225982666, "grad_norm": 0.060543350875377655, "global_step": 16064, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03474186360836029, "train_loss_bc": 0.03474186360836029, "train_loss_llm": 0.0, "grad_norm": 0.023016031831502914, "global_step": 16065, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03318759799003601, "train_loss_bc": 0.03318759799003601, "train_loss_llm": 0.0, "grad_norm": 0.03855510801076889, "global_step": 16066, "epoch": 446, "lr": 0.001} +{"train_loss": 0.031743742525577545, "train_loss_bc": 0.03174268454313278, "train_loss_llm": 0.00010576844215393066, "grad_norm": 0.05081411451101303, "global_step": 16067, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03844565525650978, "train_loss_bc": 0.03844565525650978, "train_loss_llm": 0.0, "grad_norm": 0.06949308514595032, "global_step": 16068, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03601633757352829, "train_loss_bc": 0.03601633757352829, "train_loss_llm": 0.0, "grad_norm": 0.022138191387057304, "global_step": 16069, "epoch": 446, "lr": 0.001} +{"train_loss": 0.033078476786613464, "train_loss_bc": 0.033078476786613464, "train_loss_llm": 0.0, "grad_norm": 0.03257443383336067, "global_step": 16070, "epoch": 446, "lr": 0.001} +{"train_loss": 0.02783544547855854, "train_loss_bc": 0.02783544547855854, "train_loss_llm": 0.0, "grad_norm": 0.04679320752620697, "global_step": 16071, "epoch": 446, "lr": 0.001} +{"train_loss": 0.0323057621717453, "train_loss_bc": 0.0323057621717453, "train_loss_llm": 0.0, "grad_norm": 0.06703610718250275, "global_step": 16072, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03358333930373192, "train_loss_bc": 0.03358333930373192, "train_loss_llm": 0.0, "grad_norm": 0.024148931726813316, "global_step": 16073, "epoch": 446, "lr": 0.001} +{"train_loss": 0.037615157663822174, "train_loss_bc": 0.03759998455643654, "train_loss_llm": 0.00151747465133667, "grad_norm": 0.039553262293338776, "global_step": 16074, "epoch": 446, "lr": 0.001} +{"train_loss": 0.041572947055101395, "train_loss_bc": 0.04156896844506264, "train_loss_llm": 0.00039780139923095703, "grad_norm": 0.058181311935186386, "global_step": 16075, "epoch": 446, "lr": 0.001} +{"train_loss": 0.0400942862033844, "train_loss_bc": 0.04009072482585907, "train_loss_llm": 0.0003561079502105713, "grad_norm": 0.06881112605333328, "global_step": 16076, "epoch": 446, "lr": 0.001} +{"train_loss": 0.040108103305101395, "train_loss_bc": 0.040106695145368576, "train_loss_llm": 0.00014075636863708496, "grad_norm": 0.02810286544263363, "global_step": 16077, "epoch": 446, "lr": 0.001} +{"train_loss": 0.0283163171261549, "train_loss_bc": 0.0283163171261549, "train_loss_llm": 0.0, "grad_norm": 0.045251280069351196, "global_step": 16078, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03848588094115257, "train_loss_bc": 0.03848200663924217, "train_loss_llm": 0.00038745999336242676, "grad_norm": 0.05567030608654022, "global_step": 16079, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03088734857738018, "train_loss_bc": 0.03088734857738018, "train_loss_llm": 0.0, "grad_norm": 0.06707607209682465, "global_step": 16080, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03148679807782173, "train_loss_bc": 0.03148679807782173, "train_loss_llm": 0.0, "grad_norm": 0.02372274175286293, "global_step": 16081, "epoch": 446, "lr": 0.001} +{"train_loss": 0.0284690260887146, "train_loss_bc": 0.0284690260887146, "train_loss_llm": 0.0, "grad_norm": 0.03250473737716675, "global_step": 16082, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03988994285464287, "train_loss_bc": 0.03988994285464287, "train_loss_llm": 0.0, "grad_norm": 0.04337122291326523, "global_step": 16083, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03455089032649994, "train_loss_bc": 0.03455089032649994, "train_loss_llm": 0.0, "grad_norm": 0.053300779312849045, "global_step": 16084, "epoch": 446, "lr": 0.001} +{"train_loss": 0.030191482976078987, "train_loss_bc": 0.030191482976078987, "train_loss_llm": 0.0, "grad_norm": 0.0206514410674572, "global_step": 16085, "epoch": 446, "lr": 0.001} +{"train_loss": 0.029313433915376663, "train_loss_bc": 0.029313433915376663, "train_loss_llm": 0.0, "grad_norm": 0.041444383561611176, "global_step": 16086, "epoch": 446, "lr": 0.001} +{"train_loss": 0.02918391302227974, "train_loss_bc": 0.029173200950026512, "train_loss_llm": 0.0010712742805480957, "grad_norm": 0.05229698494076729, "global_step": 16087, "epoch": 446, "lr": 0.001} +{"train_loss": 0.037397172302007675, "train_loss_bc": 0.037397172302007675, "train_loss_llm": 0.0, "grad_norm": 0.061778221279382706, "global_step": 16088, "epoch": 446, "lr": 0.001} +{"train_loss": 0.03854453191161156, "train_loss_bc": 0.03854453191161156, "train_loss_llm": 0.0, "grad_norm": 0.022706542164087296, "global_step": 16089, "epoch": 446, "lr": 0.001} +{"train_loss": 0.037346549332141876, "train_loss_bc": 0.037346549332141876, "train_loss_llm": 0.0, "grad_norm": 0.03735491633415222, "global_step": 16090, "epoch": 446, "lr": 0.001} +{"train_loss": 0.034954183321032256, "train_loss_bc": 0.03603234887123108, "train_loss_llm": 0.00027182698249816895, "grad_norm": 0.05218229442834854, "global_step": 16091, "epoch": 446, "lr": 0.001, "val_loss": 0.07087808847427368} +{"train_loss": 0.03193119168281555, "train_loss_bc": 0.03193119168281555, "train_loss_llm": 0.0, "grad_norm": 0.06426430493593216, "global_step": 16092, "epoch": 447, "lr": 0.001} +{"train_loss": 0.037130944430828094, "train_loss_bc": 0.037130944430828094, "train_loss_llm": 0.0, "grad_norm": 0.020938610658049583, "global_step": 16093, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03722487762570381, "train_loss_bc": 0.03722487762570381, "train_loss_llm": 0.0, "grad_norm": 0.04237774387001991, "global_step": 16094, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03457419201731682, "train_loss_bc": 0.0345633290708065, "train_loss_llm": 0.0010863542556762695, "grad_norm": 0.05926302447915077, "global_step": 16095, "epoch": 447, "lr": 0.001} +{"train_loss": 0.04120931774377823, "train_loss_bc": 0.04120931774377823, "train_loss_llm": 0.0, "grad_norm": 0.07254098355770111, "global_step": 16096, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03543449565768242, "train_loss_bc": 0.03543449565768242, "train_loss_llm": 0.0, "grad_norm": 0.025645721703767776, "global_step": 16097, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03177367523312569, "train_loss_bc": 0.03177367523312569, "train_loss_llm": 0.0, "grad_norm": 0.035589806735515594, "global_step": 16098, "epoch": 447, "lr": 0.001} +{"train_loss": 0.02955426648259163, "train_loss_bc": 0.02955426648259163, "train_loss_llm": 0.0, "grad_norm": 0.03845388442277908, "global_step": 16099, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03509106859564781, "train_loss_bc": 0.035085733979940414, "train_loss_llm": 0.0005332827568054199, "grad_norm": 0.05572916939854622, "global_step": 16100, "epoch": 447, "lr": 0.001} +{"train_loss": 0.033786218613386154, "train_loss_bc": 0.033786218613386154, "train_loss_llm": 0.0, "grad_norm": 0.025617778301239014, "global_step": 16101, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03958984091877937, "train_loss_bc": 0.03958984091877937, "train_loss_llm": 0.0, "grad_norm": 0.0395657904446125, "global_step": 16102, "epoch": 447, "lr": 0.001} +{"train_loss": 0.04291107878088951, "train_loss_bc": 0.04291107878088951, "train_loss_llm": 0.0, "grad_norm": 0.0504404716193676, "global_step": 16103, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03177790343761444, "train_loss_bc": 0.03177790343761444, "train_loss_llm": 0.0, "grad_norm": 0.06028931960463524, "global_step": 16104, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03841521963477135, "train_loss_bc": 0.03839625045657158, "train_loss_llm": 0.0018970966339111328, "grad_norm": 0.023176783695816994, "global_step": 16105, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03017205186188221, "train_loss_bc": 0.030162731185555458, "train_loss_llm": 0.0009320676326751709, "grad_norm": 0.034535180777311325, "global_step": 16106, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03836800158023834, "train_loss_bc": 0.03836800158023834, "train_loss_llm": 0.0, "grad_norm": 0.046518370509147644, "global_step": 16107, "epoch": 447, "lr": 0.001} +{"train_loss": 0.02928277663886547, "train_loss_bc": 0.02926800400018692, "train_loss_llm": 0.0014772117137908936, "grad_norm": 0.054453570395708084, "global_step": 16108, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03597903251647949, "train_loss_bc": 0.03596385568380356, "train_loss_llm": 0.00151786208152771, "grad_norm": 0.021250762045383453, "global_step": 16109, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03925510495901108, "train_loss_bc": 0.03925510495901108, "train_loss_llm": 0.0, "grad_norm": 0.04067070037126541, "global_step": 16110, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03503478690981865, "train_loss_bc": 0.035013943910598755, "train_loss_llm": 0.0020841658115386963, "grad_norm": 0.05199141427874565, "global_step": 16111, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03364849090576172, "train_loss_bc": 0.03364844620227814, "train_loss_llm": 4.649162292480469e-06, "grad_norm": 0.06271015107631683, "global_step": 16112, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03896886482834816, "train_loss_bc": 0.03896094113588333, "train_loss_llm": 0.0007924437522888184, "grad_norm": 0.025939110666513443, "global_step": 16113, "epoch": 447, "lr": 0.001} +{"train_loss": 0.0359763577580452, "train_loss_bc": 0.0359763577580452, "train_loss_llm": 0.0, "grad_norm": 0.03803481161594391, "global_step": 16114, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03502848744392395, "train_loss_bc": 0.03502848744392395, "train_loss_llm": 0.0, "grad_norm": 0.04952473193407059, "global_step": 16115, "epoch": 447, "lr": 0.001} +{"train_loss": 0.033104412257671356, "train_loss_bc": 0.033104412257671356, "train_loss_llm": 0.0, "grad_norm": 0.05818669870495796, "global_step": 16116, "epoch": 447, "lr": 0.001} +{"train_loss": 0.038814835250377655, "train_loss_bc": 0.038814835250377655, "train_loss_llm": 0.0, "grad_norm": 0.02811066061258316, "global_step": 16117, "epoch": 447, "lr": 0.001} +{"train_loss": 0.04102552309632301, "train_loss_bc": 0.04102552309632301, "train_loss_llm": 0.0, "grad_norm": 0.04774627834558487, "global_step": 16118, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03141549602150917, "train_loss_bc": 0.03140466660261154, "train_loss_llm": 0.001083076000213623, "grad_norm": 0.049800701439380646, "global_step": 16119, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03106066770851612, "train_loss_bc": 0.03106066770851612, "train_loss_llm": 0.0, "grad_norm": 0.06088375672698021, "global_step": 16120, "epoch": 447, "lr": 0.001} +{"train_loss": 0.029411114752292633, "train_loss_bc": 0.029411114752292633, "train_loss_llm": 0.0, "grad_norm": 0.025706898421049118, "global_step": 16121, "epoch": 447, "lr": 0.001} +{"train_loss": 0.041205994784832, "train_loss_bc": 0.041205994784832, "train_loss_llm": 0.0, "grad_norm": 0.045303069055080414, "global_step": 16122, "epoch": 447, "lr": 0.001} +{"train_loss": 0.027035726234316826, "train_loss_bc": 0.027035726234316826, "train_loss_llm": 0.0, "grad_norm": 0.0598745197057724, "global_step": 16123, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03145645558834076, "train_loss_bc": 0.03145645558834076, "train_loss_llm": 0.0, "grad_norm": 0.08000729978084564, "global_step": 16124, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03128131106495857, "train_loss_bc": 0.031273253262043, "train_loss_llm": 0.0008058249950408936, "grad_norm": 0.025684790685772896, "global_step": 16125, "epoch": 447, "lr": 0.001} +{"train_loss": 0.0376691110432148, "train_loss_bc": 0.0376691110432148, "train_loss_llm": 0.0, "grad_norm": 0.04234478250145912, "global_step": 16126, "epoch": 447, "lr": 0.001} +{"train_loss": 0.03515690668589539, "train_loss_bc": 0.04003060609102249, "train_loss_llm": 0.0019140243530273438, "grad_norm": 0.060527391731739044, "global_step": 16127, "epoch": 447, "lr": 0.001, "val_loss": 0.071127749979496} +{"train_loss": 0.03928332403302193, "train_loss_bc": 0.03928149491548538, "train_loss_llm": 0.00018277764320373535, "grad_norm": 0.074272520840168, "global_step": 16128, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03847170248627663, "train_loss_bc": 0.03847170248627663, "train_loss_llm": 0.0, "grad_norm": 0.023885216563940048, "global_step": 16129, "epoch": 448, "lr": 0.001} +{"train_loss": 0.04104676470160484, "train_loss_bc": 0.04104283079504967, "train_loss_llm": 0.0003933906555175781, "grad_norm": 0.03470581769943237, "global_step": 16130, "epoch": 448, "lr": 0.001} +{"train_loss": 0.04145241528749466, "train_loss_bc": 0.04145241528749466, "train_loss_llm": 0.0, "grad_norm": 0.04791855439543724, "global_step": 16131, "epoch": 448, "lr": 0.001} +{"train_loss": 0.037837665528059006, "train_loss_bc": 0.037837665528059006, "train_loss_llm": 0.0, "grad_norm": 0.0641028955578804, "global_step": 16132, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03440077602863312, "train_loss_bc": 0.03440077602863312, "train_loss_llm": 0.0, "grad_norm": 0.027738047763705254, "global_step": 16133, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03947058692574501, "train_loss_bc": 0.03947058692574501, "train_loss_llm": 0.0, "grad_norm": 0.03592041879892349, "global_step": 16134, "epoch": 448, "lr": 0.001} +{"train_loss": 0.039932917803525925, "train_loss_bc": 0.039932917803525925, "train_loss_llm": 0.0, "grad_norm": 0.0463351346552372, "global_step": 16135, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03360427916049957, "train_loss_bc": 0.03360427916049957, "train_loss_llm": 0.0, "grad_norm": 0.05645335465669632, "global_step": 16136, "epoch": 448, "lr": 0.001} +{"train_loss": 0.034259796142578125, "train_loss_bc": 0.03425523638725281, "train_loss_llm": 0.00045609474182128906, "grad_norm": 0.02929120510816574, "global_step": 16137, "epoch": 448, "lr": 0.001} +{"train_loss": 0.039519526064395905, "train_loss_bc": 0.039519526064395905, "train_loss_llm": 0.0, "grad_norm": 0.044473327696323395, "global_step": 16138, "epoch": 448, "lr": 0.001} +{"train_loss": 0.02762727253139019, "train_loss_bc": 0.027617329731583595, "train_loss_llm": 0.0009942948818206787, "grad_norm": 0.05507756769657135, "global_step": 16139, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03180297091603279, "train_loss_bc": 0.03180297091603279, "train_loss_llm": 0.0, "grad_norm": 0.06106722354888916, "global_step": 16140, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03239809349179268, "train_loss_bc": 0.03239809349179268, "train_loss_llm": 0.0, "grad_norm": 0.023970911279320717, "global_step": 16141, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03343605995178223, "train_loss_bc": 0.03343605995178223, "train_loss_llm": 0.0, "grad_norm": 0.0397893562912941, "global_step": 16142, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03936942666769028, "train_loss_bc": 0.039367642253637314, "train_loss_llm": 0.00017827749252319336, "grad_norm": 0.05301105976104736, "global_step": 16143, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03371149301528931, "train_loss_bc": 0.03369256854057312, "train_loss_llm": 0.0018922686576843262, "grad_norm": 0.06672077625989914, "global_step": 16144, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03726356104016304, "train_loss_bc": 0.03725603595376015, "train_loss_llm": 0.0007524490356445312, "grad_norm": 0.026235783472657204, "global_step": 16145, "epoch": 448, "lr": 0.001} +{"train_loss": 0.032605502754449844, "train_loss_bc": 0.032605502754449844, "train_loss_llm": 0.0, "grad_norm": 0.04272177442908287, "global_step": 16146, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03330906108021736, "train_loss_bc": 0.03330906108021736, "train_loss_llm": 0.0, "grad_norm": 0.04779670387506485, "global_step": 16147, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03358040750026703, "train_loss_bc": 0.03358040750026703, "train_loss_llm": 0.0, "grad_norm": 0.0592058002948761, "global_step": 16148, "epoch": 448, "lr": 0.001} +{"train_loss": 0.030894234776496887, "train_loss_bc": 0.030886506661772728, "train_loss_llm": 0.0007728040218353271, "grad_norm": 0.027142349630594254, "global_step": 16149, "epoch": 448, "lr": 0.001} +{"train_loss": 0.030958591029047966, "train_loss_bc": 0.030958591029047966, "train_loss_llm": 0.0, "grad_norm": 0.0407499335706234, "global_step": 16150, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03145294636487961, "train_loss_bc": 0.03145294636487961, "train_loss_llm": 0.0, "grad_norm": 0.0467413030564785, "global_step": 16151, "epoch": 448, "lr": 0.001} +{"train_loss": 0.028979850932955742, "train_loss_bc": 0.028979850932955742, "train_loss_llm": 0.0, "grad_norm": 0.0632072314620018, "global_step": 16152, "epoch": 448, "lr": 0.001} +{"train_loss": 0.032847411930561066, "train_loss_bc": 0.03283805400133133, "train_loss_llm": 0.0009358227252960205, "grad_norm": 0.02743026614189148, "global_step": 16153, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03646059334278107, "train_loss_bc": 0.03646059334278107, "train_loss_llm": 0.0, "grad_norm": 0.04688635841012001, "global_step": 16154, "epoch": 448, "lr": 0.001} +{"train_loss": 0.034921795129776, "train_loss_bc": 0.034914176911115646, "train_loss_llm": 0.0007617473602294922, "grad_norm": 0.05899942293763161, "global_step": 16155, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03844860568642616, "train_loss_bc": 0.038447581231594086, "train_loss_llm": 0.0001023411750793457, "grad_norm": 0.0751894935965538, "global_step": 16156, "epoch": 448, "lr": 0.001} +{"train_loss": 0.02633924037218094, "train_loss_bc": 0.02633924037218094, "train_loss_llm": 0.0, "grad_norm": 0.023208780214190483, "global_step": 16157, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03865857794880867, "train_loss_bc": 0.03865857794880867, "train_loss_llm": 0.0, "grad_norm": 0.03931685537099838, "global_step": 16158, "epoch": 448, "lr": 0.001} +{"train_loss": 0.037320151925086975, "train_loss_bc": 0.037320151925086975, "train_loss_llm": 0.0, "grad_norm": 0.05704895779490471, "global_step": 16159, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03285713493824005, "train_loss_bc": 0.03284095972776413, "train_loss_llm": 0.0016175806522369385, "grad_norm": 0.06866259127855301, "global_step": 16160, "epoch": 448, "lr": 0.001} +{"train_loss": 0.03217798098921776, "train_loss_bc": 0.032174549996852875, "train_loss_llm": 0.00034302473068237305, "grad_norm": 0.028657864779233932, "global_step": 16161, "epoch": 448, "lr": 0.001} +{"train_loss": 0.04148679971694946, "train_loss_bc": 0.04148436710238457, "train_loss_llm": 0.00024318695068359375, "grad_norm": 0.042078468948602676, "global_step": 16162, "epoch": 448, "lr": 0.001} +{"train_loss": 0.034990062574959464, "train_loss_bc": 0.03145473450422287, "train_loss_llm": 0.0, "grad_norm": 0.05823972076177597, "global_step": 16163, "epoch": 448, "lr": 0.001, "val_loss": 0.07073001563549042} +{"train_loss": 0.03690144419670105, "train_loss_bc": 0.036877796053886414, "train_loss_llm": 0.0023647844791412354, "grad_norm": 0.0698898434638977, "global_step": 16164, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03836097940802574, "train_loss_bc": 0.03836097940802574, "train_loss_llm": 0.0, "grad_norm": 0.022180354222655296, "global_step": 16165, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03368772566318512, "train_loss_bc": 0.03368772566318512, "train_loss_llm": 0.0, "grad_norm": 0.038133181631565094, "global_step": 16166, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03198695182800293, "train_loss_bc": 0.03198695182800293, "train_loss_llm": 0.0, "grad_norm": 0.05271708592772484, "global_step": 16167, "epoch": 449, "lr": 0.001} +{"train_loss": 0.035959865897893906, "train_loss_bc": 0.035959865897893906, "train_loss_llm": 0.0, "grad_norm": 0.06524708867073059, "global_step": 16168, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03133564814925194, "train_loss_bc": 0.03133564814925194, "train_loss_llm": 0.0, "grad_norm": 0.028725488111376762, "global_step": 16169, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03688126429915428, "train_loss_bc": 0.03688126429915428, "train_loss_llm": 0.0, "grad_norm": 0.03705188259482384, "global_step": 16170, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03326157480478287, "train_loss_bc": 0.03326157480478287, "train_loss_llm": 0.0, "grad_norm": 0.05016465112566948, "global_step": 16171, "epoch": 449, "lr": 0.001} +{"train_loss": 0.034782521426677704, "train_loss_bc": 0.034782521426677704, "train_loss_llm": 0.0, "grad_norm": 0.06500595062971115, "global_step": 16172, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03967701643705368, "train_loss_bc": 0.03967701643705368, "train_loss_llm": 0.0, "grad_norm": 0.03336720168590546, "global_step": 16173, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03536288067698479, "train_loss_bc": 0.03536288067698479, "train_loss_llm": 0.0, "grad_norm": 0.049396730959415436, "global_step": 16174, "epoch": 449, "lr": 0.001} +{"train_loss": 0.034249916672706604, "train_loss_bc": 0.034249916672706604, "train_loss_llm": 0.0, "grad_norm": 0.06060225889086723, "global_step": 16175, "epoch": 449, "lr": 0.001} +{"train_loss": 0.036955010145902634, "train_loss_bc": 0.036955010145902634, "train_loss_llm": 0.0, "grad_norm": 0.07849345356225967, "global_step": 16176, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03567814081907272, "train_loss_bc": 0.03566784784197807, "train_loss_llm": 0.0010294616222381592, "grad_norm": 0.02986185811460018, "global_step": 16177, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03711126744747162, "train_loss_bc": 0.03711126744747162, "train_loss_llm": 0.0, "grad_norm": 0.04581781476736069, "global_step": 16178, "epoch": 449, "lr": 0.001} +{"train_loss": 0.023919904604554176, "train_loss_bc": 0.023919904604554176, "train_loss_llm": 0.0, "grad_norm": 0.050784237682819366, "global_step": 16179, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03432685136795044, "train_loss_bc": 0.03432685136795044, "train_loss_llm": 0.0, "grad_norm": 0.06550692766904831, "global_step": 16180, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03382869064807892, "train_loss_bc": 0.03382869064807892, "train_loss_llm": 0.0, "grad_norm": 0.021716095507144928, "global_step": 16181, "epoch": 449, "lr": 0.001} +{"train_loss": 0.040562741458415985, "train_loss_bc": 0.040562741458415985, "train_loss_llm": 0.0, "grad_norm": 0.03549748659133911, "global_step": 16182, "epoch": 449, "lr": 0.001} +{"train_loss": 0.02427949570119381, "train_loss_bc": 0.024259250611066818, "train_loss_llm": 0.002024531364440918, "grad_norm": 0.044469308108091354, "global_step": 16183, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03181746229529381, "train_loss_bc": 0.03181746229529381, "train_loss_llm": 0.0, "grad_norm": 0.05443210154771805, "global_step": 16184, "epoch": 449, "lr": 0.001} +{"train_loss": 0.02986457571387291, "train_loss_bc": 0.02986457571387291, "train_loss_llm": 0.0, "grad_norm": 0.02389225736260414, "global_step": 16185, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03539351746439934, "train_loss_bc": 0.03539351746439934, "train_loss_llm": 0.0, "grad_norm": 0.0393909215927124, "global_step": 16186, "epoch": 449, "lr": 0.001} +{"train_loss": 0.032192863523960114, "train_loss_bc": 0.0321783721446991, "train_loss_llm": 0.0014490783214569092, "grad_norm": 0.05310483276844025, "global_step": 16187, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03733779862523079, "train_loss_bc": 0.037336718291044235, "train_loss_llm": 0.00010803341865539551, "grad_norm": 0.06229133531451225, "global_step": 16188, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03970995545387268, "train_loss_bc": 0.03970995545387268, "train_loss_llm": 0.0, "grad_norm": 0.030448513105511665, "global_step": 16189, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03373289853334427, "train_loss_bc": 0.03373289853334427, "train_loss_llm": 0.0, "grad_norm": 0.049155235290527344, "global_step": 16190, "epoch": 449, "lr": 0.001} +{"train_loss": 0.04028953239321709, "train_loss_bc": 0.04028953239321709, "train_loss_llm": 0.0, "grad_norm": 0.07090023159980774, "global_step": 16191, "epoch": 449, "lr": 0.001} +{"train_loss": 0.036542683839797974, "train_loss_bc": 0.036542683839797974, "train_loss_llm": 0.0, "grad_norm": 0.07807579636573792, "global_step": 16192, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03764932602643967, "train_loss_bc": 0.03764932602643967, "train_loss_llm": 0.0, "grad_norm": 0.02956344000995159, "global_step": 16193, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03998618572950363, "train_loss_bc": 0.039973028004169464, "train_loss_llm": 0.0013159215450286865, "grad_norm": 0.04326116293668747, "global_step": 16194, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03454291820526123, "train_loss_bc": 0.03454291820526123, "train_loss_llm": 0.0, "grad_norm": 0.06538104265928268, "global_step": 16195, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03482196107506752, "train_loss_bc": 0.03482196107506752, "train_loss_llm": 0.0, "grad_norm": 0.07585908472537994, "global_step": 16196, "epoch": 449, "lr": 0.001} +{"train_loss": 0.04267893731594086, "train_loss_bc": 0.04267893731594086, "train_loss_llm": 0.0, "grad_norm": 0.028465215116739273, "global_step": 16197, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03142649307847023, "train_loss_bc": 0.0314219705760479, "train_loss_llm": 0.00045233964920043945, "grad_norm": 0.041439443826675415, "global_step": 16198, "epoch": 449, "lr": 0.001} +{"train_loss": 0.03513545770612028, "train_loss_bc": 0.03777947649359703, "train_loss_llm": 0.0, "grad_norm": 0.05577828362584114, "global_step": 16199, "epoch": 449, "lr": 0.001, "val_loss": 0.0721287727355957} +{"train_loss": 0.03818078711628914, "train_loss_bc": 0.03818078711628914, "train_loss_llm": 0.0, "grad_norm": 0.06508077681064606, "global_step": 16200, "epoch": 450, "lr": 0.001} +{"train_loss": 0.035985760390758514, "train_loss_bc": 0.035983048379421234, "train_loss_llm": 0.00027123093605041504, "grad_norm": 0.022917697206139565, "global_step": 16201, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03712776303291321, "train_loss_bc": 0.03712776303291321, "train_loss_llm": 0.0, "grad_norm": 0.04277477040886879, "global_step": 16202, "epoch": 450, "lr": 0.001} +{"train_loss": 0.032433297485113144, "train_loss_bc": 0.03243072330951691, "train_loss_llm": 0.00025731325149536133, "grad_norm": 0.053819440305233, "global_step": 16203, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03681055083870888, "train_loss_bc": 0.03681055083870888, "train_loss_llm": 0.0, "grad_norm": 0.05904601141810417, "global_step": 16204, "epoch": 450, "lr": 0.001} +{"train_loss": 0.02909150719642639, "train_loss_bc": 0.02908935770392418, "train_loss_llm": 0.00021499395370483398, "grad_norm": 0.02888515777885914, "global_step": 16205, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03731181100010872, "train_loss_bc": 0.03731181100010872, "train_loss_llm": 0.0, "grad_norm": 0.04199563339352608, "global_step": 16206, "epoch": 450, "lr": 0.001} +{"train_loss": 0.04255000501871109, "train_loss_bc": 0.04255000501871109, "train_loss_llm": 0.0, "grad_norm": 0.056118059903383255, "global_step": 16207, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03542694076895714, "train_loss_bc": 0.03542694076895714, "train_loss_llm": 0.0, "grad_norm": 0.06481815874576569, "global_step": 16208, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03752019256353378, "train_loss_bc": 0.03751002997159958, "train_loss_llm": 0.0010161995887756348, "grad_norm": 0.03259361535310745, "global_step": 16209, "epoch": 450, "lr": 0.001} +{"train_loss": 0.02812642976641655, "train_loss_bc": 0.02812642976641655, "train_loss_llm": 0.0, "grad_norm": 0.03954770788550377, "global_step": 16210, "epoch": 450, "lr": 0.001} +{"train_loss": 0.02850358933210373, "train_loss_bc": 0.028499318286776543, "train_loss_llm": 0.0004270970821380615, "grad_norm": 0.056027937680482864, "global_step": 16211, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03820241615176201, "train_loss_bc": 0.03820241615176201, "train_loss_llm": 0.0, "grad_norm": 0.0710209384560585, "global_step": 16212, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03515646234154701, "train_loss_bc": 0.03515152260661125, "train_loss_llm": 0.0004940032958984375, "grad_norm": 0.028751375153660774, "global_step": 16213, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03872453048825264, "train_loss_bc": 0.03872453048825264, "train_loss_llm": 0.0, "grad_norm": 0.05681680515408516, "global_step": 16214, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03147498890757561, "train_loss_bc": 0.03147498890757561, "train_loss_llm": 0.0, "grad_norm": 0.06598348915576935, "global_step": 16215, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03231128305196762, "train_loss_bc": 0.03230630233883858, "train_loss_llm": 0.0004981756210327148, "grad_norm": 0.08130111545324326, "global_step": 16216, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03425944969058037, "train_loss_bc": 0.03425944969058037, "train_loss_llm": 0.0, "grad_norm": 0.027651183307170868, "global_step": 16217, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03227296099066734, "train_loss_bc": 0.03227296099066734, "train_loss_llm": 0.0, "grad_norm": 0.0355716273188591, "global_step": 16218, "epoch": 450, "lr": 0.001} +{"train_loss": 0.034548841416835785, "train_loss_bc": 0.034548841416835785, "train_loss_llm": 0.0, "grad_norm": 0.04596704989671707, "global_step": 16219, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03477649390697479, "train_loss_bc": 0.03477649390697479, "train_loss_llm": 0.0, "grad_norm": 0.05948064476251602, "global_step": 16220, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03594211861491203, "train_loss_bc": 0.03593976050615311, "train_loss_llm": 0.00023570656776428223, "grad_norm": 0.021802419796586037, "global_step": 16221, "epoch": 450, "lr": 0.001} +{"train_loss": 0.037029873579740524, "train_loss_bc": 0.03699619323015213, "train_loss_llm": 0.0033680498600006104, "grad_norm": 0.03311288356781006, "global_step": 16222, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03538757190108299, "train_loss_bc": 0.03538757190108299, "train_loss_llm": 0.0, "grad_norm": 0.03939804434776306, "global_step": 16223, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03171064332127571, "train_loss_bc": 0.03171064332127571, "train_loss_llm": 0.0, "grad_norm": 0.048681214451789856, "global_step": 16224, "epoch": 450, "lr": 0.001} +{"train_loss": 0.04072649031877518, "train_loss_bc": 0.04072649031877518, "train_loss_llm": 0.0, "grad_norm": 0.030559536069631577, "global_step": 16225, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03770044073462486, "train_loss_bc": 0.03770044073462486, "train_loss_llm": 0.0, "grad_norm": 0.037790246307849884, "global_step": 16226, "epoch": 450, "lr": 0.001} +{"train_loss": 0.0426882766187191, "train_loss_bc": 0.0426882766187191, "train_loss_llm": 0.0, "grad_norm": 0.0528549887239933, "global_step": 16227, "epoch": 450, "lr": 0.001} +{"train_loss": 0.0319196879863739, "train_loss_bc": 0.03191963583230972, "train_loss_llm": 5.0961971282958984e-06, "grad_norm": 0.06617528200149536, "global_step": 16228, "epoch": 450, "lr": 0.001} +{"train_loss": 0.02742736041545868, "train_loss_bc": 0.027415985241532326, "train_loss_llm": 0.001137465238571167, "grad_norm": 0.019476374611258507, "global_step": 16229, "epoch": 450, "lr": 0.001} +{"train_loss": 0.032282281666994095, "train_loss_bc": 0.032282281666994095, "train_loss_llm": 0.0, "grad_norm": 0.035494763404130936, "global_step": 16230, "epoch": 450, "lr": 0.001} +{"train_loss": 0.034167468547821045, "train_loss_bc": 0.034167468547821045, "train_loss_llm": 0.0, "grad_norm": 0.04754229262471199, "global_step": 16231, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03791913017630577, "train_loss_bc": 0.03791140019893646, "train_loss_llm": 0.0007728636264801025, "grad_norm": 0.0653555691242218, "global_step": 16232, "epoch": 450, "lr": 0.001} +{"train_loss": 0.039122313261032104, "train_loss_bc": 0.03911413624882698, "train_loss_llm": 0.0008175969123840332, "grad_norm": 0.022451067343354225, "global_step": 16233, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03372609615325928, "train_loss_bc": 0.03372609615325928, "train_loss_llm": 0.0, "grad_norm": 0.037761904299259186, "global_step": 16234, "epoch": 450, "lr": 0.001} +{"train_loss": 0.03488566353917122, "train_loss_bc": 0.02731391414999962, "train_loss_llm": 0.002415776252746582, "grad_norm": 0.05030311271548271, "global_step": 16235, "epoch": 450, "lr": 0.001, "train/cumulative_reward": 157.471990322241, "train/mean_score": 18.93654855120486, "train/success_rate": 0.4, "test/cumulative_reward": 47.54347662024142, "test/mean_score": 9.063655486570546, "test/success_rate": 0.22, "val_loss": 0.07121669501066208, "train_action_mse_error": 0.03761240839958191} +{"train_loss": 0.034762635827064514, "train_loss_bc": 0.03475339710712433, "train_loss_llm": 0.0009238123893737793, "grad_norm": 0.04843559488654137, "global_step": 16236, "epoch": 451, "lr": 0.001} +{"train_loss": 0.039222851395606995, "train_loss_bc": 0.03921917453408241, "train_loss_llm": 0.00036767125129699707, "grad_norm": 0.02674342505633831, "global_step": 16237, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03393020108342171, "train_loss_bc": 0.03393020108342171, "train_loss_llm": 0.0, "grad_norm": 0.046812836080789566, "global_step": 16238, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03687426447868347, "train_loss_bc": 0.03687426447868347, "train_loss_llm": 0.0, "grad_norm": 0.04626103863120079, "global_step": 16239, "epoch": 451, "lr": 0.001} +{"train_loss": 0.043319545686244965, "train_loss_bc": 0.043319545686244965, "train_loss_llm": 0.0, "grad_norm": 0.059916868805885315, "global_step": 16240, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03176942095160484, "train_loss_bc": 0.03176942095160484, "train_loss_llm": 0.0, "grad_norm": 0.026716984808444977, "global_step": 16241, "epoch": 451, "lr": 0.001} +{"train_loss": 0.0362565778195858, "train_loss_bc": 0.0362565778195858, "train_loss_llm": 0.0, "grad_norm": 0.04287531599402428, "global_step": 16242, "epoch": 451, "lr": 0.001} +{"train_loss": 0.030798623338341713, "train_loss_bc": 0.030798623338341713, "train_loss_llm": 0.0, "grad_norm": 0.05878667160868645, "global_step": 16243, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03105039708316326, "train_loss_bc": 0.031042521819472313, "train_loss_llm": 0.0007874369621276855, "grad_norm": 0.07343556731939316, "global_step": 16244, "epoch": 451, "lr": 0.001} +{"train_loss": 0.030081747099757195, "train_loss_bc": 0.030081747099757195, "train_loss_llm": 0.0, "grad_norm": 0.020820342004299164, "global_step": 16245, "epoch": 451, "lr": 0.001} +{"train_loss": 0.034096039831638336, "train_loss_bc": 0.03408856317400932, "train_loss_llm": 0.0007478296756744385, "grad_norm": 0.03701955825090408, "global_step": 16246, "epoch": 451, "lr": 0.001} +{"train_loss": 0.04237489402294159, "train_loss_bc": 0.04233958572149277, "train_loss_llm": 0.003530651330947876, "grad_norm": 0.05263315886259079, "global_step": 16247, "epoch": 451, "lr": 0.001} +{"train_loss": 0.029293259605765343, "train_loss_bc": 0.029293259605765343, "train_loss_llm": 0.0, "grad_norm": 0.06615894287824631, "global_step": 16248, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03664841130375862, "train_loss_bc": 0.03664841130375862, "train_loss_llm": 0.0, "grad_norm": 0.024842847138643265, "global_step": 16249, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03241857513785362, "train_loss_bc": 0.03241857513785362, "train_loss_llm": 0.0, "grad_norm": 0.039594318717718124, "global_step": 16250, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03101714514195919, "train_loss_bc": 0.03101457841694355, "train_loss_llm": 0.0002567470073699951, "grad_norm": 0.049492739140987396, "global_step": 16251, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03288638964295387, "train_loss_bc": 0.032879821956157684, "train_loss_llm": 0.0006568431854248047, "grad_norm": 0.06175810471177101, "global_step": 16252, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03543316572904587, "train_loss_bc": 0.03543316572904587, "train_loss_llm": 0.0, "grad_norm": 0.029655052348971367, "global_step": 16253, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03698353469371796, "train_loss_bc": 0.03698353469371796, "train_loss_llm": 0.0, "grad_norm": 0.049107879400253296, "global_step": 16254, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03814256936311722, "train_loss_bc": 0.03814256936311722, "train_loss_llm": 0.0, "grad_norm": 0.06142379343509674, "global_step": 16255, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03853384032845497, "train_loss_bc": 0.03853093832731247, "train_loss_llm": 0.0002900362014770508, "grad_norm": 0.07538548856973648, "global_step": 16256, "epoch": 451, "lr": 0.001} +{"train_loss": 0.04246397688984871, "train_loss_bc": 0.04246397688984871, "train_loss_llm": 0.0, "grad_norm": 0.03023696132004261, "global_step": 16257, "epoch": 451, "lr": 0.001} +{"train_loss": 0.034801971167325974, "train_loss_bc": 0.034801971167325974, "train_loss_llm": 0.0, "grad_norm": 0.0508614182472229, "global_step": 16258, "epoch": 451, "lr": 0.001} +{"train_loss": 0.035627152770757675, "train_loss_bc": 0.03561149910092354, "train_loss_llm": 0.0015652775764465332, "grad_norm": 0.0654030367732048, "global_step": 16259, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03733323514461517, "train_loss_bc": 0.03733323514461517, "train_loss_llm": 0.0, "grad_norm": 0.07427102327346802, "global_step": 16260, "epoch": 451, "lr": 0.001} +{"train_loss": 0.029538514092564583, "train_loss_bc": 0.029538514092564583, "train_loss_llm": 0.0, "grad_norm": 0.023526344448328018, "global_step": 16261, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03486695513129234, "train_loss_bc": 0.03486695513129234, "train_loss_llm": 0.0, "grad_norm": 0.03150685131549835, "global_step": 16262, "epoch": 451, "lr": 0.001} +{"train_loss": 0.027706177905201912, "train_loss_bc": 0.02770376019179821, "train_loss_llm": 0.00024178624153137207, "grad_norm": 0.03607098013162613, "global_step": 16263, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03179550915956497, "train_loss_bc": 0.03177832067012787, "train_loss_llm": 0.001718670129776001, "grad_norm": 0.0492485947906971, "global_step": 16264, "epoch": 451, "lr": 0.001} +{"train_loss": 0.034693360328674316, "train_loss_bc": 0.034693360328674316, "train_loss_llm": 0.0, "grad_norm": 0.027947857975959778, "global_step": 16265, "epoch": 451, "lr": 0.001} +{"train_loss": 0.04011824354529381, "train_loss_bc": 0.04011339694261551, "train_loss_llm": 0.00048482418060302734, "grad_norm": 0.047527167946100235, "global_step": 16266, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03402765467762947, "train_loss_bc": 0.03402765467762947, "train_loss_llm": 0.0, "grad_norm": 0.05099727585911751, "global_step": 16267, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03394729644060135, "train_loss_bc": 0.03394729644060135, "train_loss_llm": 0.0, "grad_norm": 0.06354191899299622, "global_step": 16268, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03527674078941345, "train_loss_bc": 0.03527674078941345, "train_loss_llm": 0.0, "grad_norm": 0.02350306324660778, "global_step": 16269, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03350554779171944, "train_loss_bc": 0.03350554779171944, "train_loss_llm": 0.0, "grad_norm": 0.04424206539988518, "global_step": 16270, "epoch": 451, "lr": 0.001} +{"train_loss": 0.03503337207560738, "train_loss_bc": 0.03960496932268143, "train_loss_llm": 0.0, "grad_norm": 0.06530340015888214, "global_step": 16271, "epoch": 451, "lr": 0.001, "val_loss": 0.07132492959499359} +{"train_loss": 0.037052616477012634, "train_loss_bc": 0.037052616477012634, "train_loss_llm": 0.0, "grad_norm": 0.07494241744279861, "global_step": 16272, "epoch": 452, "lr": 0.001} +{"train_loss": 0.029638709500432014, "train_loss_bc": 0.029638709500432014, "train_loss_llm": 0.0, "grad_norm": 0.025154439732432365, "global_step": 16273, "epoch": 452, "lr": 0.001} +{"train_loss": 0.036408837884664536, "train_loss_bc": 0.036408837884664536, "train_loss_llm": 0.0, "grad_norm": 0.04098615422844887, "global_step": 16274, "epoch": 452, "lr": 0.001} +{"train_loss": 0.0326603539288044, "train_loss_bc": 0.0326603539288044, "train_loss_llm": 0.0, "grad_norm": 0.05140911415219307, "global_step": 16275, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03684358671307564, "train_loss_bc": 0.03684358671307564, "train_loss_llm": 0.0, "grad_norm": 0.06396392732858658, "global_step": 16276, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03839947283267975, "train_loss_bc": 0.03839947283267975, "train_loss_llm": 0.0, "grad_norm": 0.027215363457798958, "global_step": 16277, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03371443971991539, "train_loss_bc": 0.03371443971991539, "train_loss_llm": 0.0, "grad_norm": 0.041419148445129395, "global_step": 16278, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03347620740532875, "train_loss_bc": 0.033472221344709396, "train_loss_llm": 0.0003986358642578125, "grad_norm": 0.05162055790424347, "global_step": 16279, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03793051093816757, "train_loss_bc": 0.03793051093816757, "train_loss_llm": 0.0, "grad_norm": 0.06969985365867615, "global_step": 16280, "epoch": 452, "lr": 0.001} +{"train_loss": 0.0252174511551857, "train_loss_bc": 0.0252174511551857, "train_loss_llm": 0.0, "grad_norm": 0.018432199954986572, "global_step": 16281, "epoch": 452, "lr": 0.001} +{"train_loss": 0.028788236901164055, "train_loss_bc": 0.028788236901164055, "train_loss_llm": 0.0, "grad_norm": 0.031435541808605194, "global_step": 16282, "epoch": 452, "lr": 0.001} +{"train_loss": 0.030078966170549393, "train_loss_bc": 0.030078966170549393, "train_loss_llm": 0.0, "grad_norm": 0.040108099579811096, "global_step": 16283, "epoch": 452, "lr": 0.001} +{"train_loss": 0.030740628018975258, "train_loss_bc": 0.030740628018975258, "train_loss_llm": 0.0, "grad_norm": 0.05020245537161827, "global_step": 16284, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03516019135713577, "train_loss_bc": 0.03516019135713577, "train_loss_llm": 0.0, "grad_norm": 0.02172127179801464, "global_step": 16285, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03633816912770271, "train_loss_bc": 0.03633006662130356, "train_loss_llm": 0.0008101165294647217, "grad_norm": 0.03295215964317322, "global_step": 16286, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03559362515807152, "train_loss_bc": 0.03559362515807152, "train_loss_llm": 0.0, "grad_norm": 0.0518316887319088, "global_step": 16287, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03921861946582794, "train_loss_bc": 0.03921861946582794, "train_loss_llm": 0.0, "grad_norm": 0.07552376389503479, "global_step": 16288, "epoch": 452, "lr": 0.001} +{"train_loss": 0.034626975655555725, "train_loss_bc": 0.034626975655555725, "train_loss_llm": 0.0, "grad_norm": 0.022282233461737633, "global_step": 16289, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03733636811375618, "train_loss_bc": 0.03733636811375618, "train_loss_llm": 0.0, "grad_norm": 0.037750110030174255, "global_step": 16290, "epoch": 452, "lr": 0.001} +{"train_loss": 0.035732582211494446, "train_loss_bc": 0.0357133112847805, "train_loss_llm": 0.0019272267818450928, "grad_norm": 0.05339965596795082, "global_step": 16291, "epoch": 452, "lr": 0.001} +{"train_loss": 0.04051448032259941, "train_loss_bc": 0.04051448032259941, "train_loss_llm": 0.0, "grad_norm": 0.07554080337285995, "global_step": 16292, "epoch": 452, "lr": 0.001} +{"train_loss": 0.036345046013593674, "train_loss_bc": 0.03633926063776016, "train_loss_llm": 0.0005785822868347168, "grad_norm": 0.027783416211605072, "global_step": 16293, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03492989391088486, "train_loss_bc": 0.03492989391088486, "train_loss_llm": 0.0, "grad_norm": 0.039641108363866806, "global_step": 16294, "epoch": 452, "lr": 0.001} +{"train_loss": 0.033902719616889954, "train_loss_bc": 0.033902719616889954, "train_loss_llm": 0.0, "grad_norm": 0.05248226970434189, "global_step": 16295, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03481592610478401, "train_loss_bc": 0.03481592610478401, "train_loss_llm": 0.0, "grad_norm": 0.06993311643600464, "global_step": 16296, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03297087177634239, "train_loss_bc": 0.03296985477209091, "train_loss_llm": 0.0001016855239868164, "grad_norm": 0.023472094908356667, "global_step": 16297, "epoch": 452, "lr": 0.001} +{"train_loss": 0.04150700941681862, "train_loss_bc": 0.04150700941681862, "train_loss_llm": 0.0, "grad_norm": 0.041737888008356094, "global_step": 16298, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03302023559808731, "train_loss_bc": 0.03301605209708214, "train_loss_llm": 0.0004183351993560791, "grad_norm": 0.05887818709015846, "global_step": 16299, "epoch": 452, "lr": 0.001} +{"train_loss": 0.033956099301576614, "train_loss_bc": 0.033956099301576614, "train_loss_llm": 0.0, "grad_norm": 0.0731530413031578, "global_step": 16300, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03272184729576111, "train_loss_bc": 0.03272184729576111, "train_loss_llm": 0.0, "grad_norm": 0.02289740741252899, "global_step": 16301, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03595009073615074, "train_loss_bc": 0.03595009073615074, "train_loss_llm": 0.0, "grad_norm": 0.03432077169418335, "global_step": 16302, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03516062721610069, "train_loss_bc": 0.03516062721610069, "train_loss_llm": 0.0, "grad_norm": 0.04469257593154907, "global_step": 16303, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03712351620197296, "train_loss_bc": 0.03711337596178055, "train_loss_llm": 0.0010140836238861084, "grad_norm": 0.057261258363723755, "global_step": 16304, "epoch": 452, "lr": 0.001} +{"train_loss": 0.031022949144244194, "train_loss_bc": 0.031022708863019943, "train_loss_llm": 2.4020671844482422e-05, "grad_norm": 0.020376713946461678, "global_step": 16305, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03881891444325447, "train_loss_bc": 0.03880498558282852, "train_loss_llm": 0.0013928115367889404, "grad_norm": 0.03130631893873215, "global_step": 16306, "epoch": 452, "lr": 0.001} +{"train_loss": 0.03502586794396242, "train_loss_bc": 0.04321447014808655, "train_loss_llm": 0.0, "grad_norm": 0.045716412365436554, "global_step": 16307, "epoch": 452, "lr": 0.001, "val_loss": 0.07192668318748474} +{"train_loss": 0.03168422356247902, "train_loss_bc": 0.03168422356247902, "train_loss_llm": 0.0, "grad_norm": 0.05731667950749397, "global_step": 16308, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03346889466047287, "train_loss_bc": 0.03346889466047287, "train_loss_llm": 0.0, "grad_norm": 0.024935847148299217, "global_step": 16309, "epoch": 453, "lr": 0.001} +{"train_loss": 0.02788202092051506, "train_loss_bc": 0.027863185852766037, "train_loss_llm": 0.0018835663795471191, "grad_norm": 0.035321492701768875, "global_step": 16310, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03486272320151329, "train_loss_bc": 0.03486272320151329, "train_loss_llm": 0.0, "grad_norm": 0.04940517991781235, "global_step": 16311, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03167887404561043, "train_loss_bc": 0.03165039047598839, "train_loss_llm": 0.002848327159881592, "grad_norm": 0.06081034988164902, "global_step": 16312, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03403108939528465, "train_loss_bc": 0.03403108939528465, "train_loss_llm": 0.0, "grad_norm": 0.02105799876153469, "global_step": 16313, "epoch": 453, "lr": 0.001} +{"train_loss": 0.037140242755413055, "train_loss_bc": 0.037140242755413055, "train_loss_llm": 0.0, "grad_norm": 0.03511561080813408, "global_step": 16314, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03530239313840866, "train_loss_bc": 0.03530239313840866, "train_loss_llm": 0.0, "grad_norm": 0.04750476032495499, "global_step": 16315, "epoch": 453, "lr": 0.001} +{"train_loss": 0.0360870324075222, "train_loss_bc": 0.0360870324075222, "train_loss_llm": 0.0, "grad_norm": 0.05462394654750824, "global_step": 16316, "epoch": 453, "lr": 0.001} +{"train_loss": 0.02681054174900055, "train_loss_bc": 0.02679859846830368, "train_loss_llm": 0.001194387674331665, "grad_norm": 0.022005466744303703, "global_step": 16317, "epoch": 453, "lr": 0.001} +{"train_loss": 0.0414370559155941, "train_loss_bc": 0.0414370559155941, "train_loss_llm": 0.0, "grad_norm": 0.04988699406385422, "global_step": 16318, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03204047679901123, "train_loss_bc": 0.03204047679901123, "train_loss_llm": 0.0, "grad_norm": 0.06376040726900101, "global_step": 16319, "epoch": 453, "lr": 0.001} +{"train_loss": 0.028662651777267456, "train_loss_bc": 0.02865971066057682, "train_loss_llm": 0.00029405951499938965, "grad_norm": 0.07100719958543777, "global_step": 16320, "epoch": 453, "lr": 0.001} +{"train_loss": 0.039307814091444016, "train_loss_bc": 0.039307814091444016, "train_loss_llm": 0.0, "grad_norm": 0.0213757511228323, "global_step": 16321, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03601400926709175, "train_loss_bc": 0.03601400926709175, "train_loss_llm": 0.0, "grad_norm": 0.035641398280858994, "global_step": 16322, "epoch": 453, "lr": 0.001} +{"train_loss": 0.032519277185201645, "train_loss_bc": 0.032519277185201645, "train_loss_llm": 0.0, "grad_norm": 0.046567223966121674, "global_step": 16323, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03669251129031181, "train_loss_bc": 0.03669251129031181, "train_loss_llm": 0.0, "grad_norm": 0.06397959589958191, "global_step": 16324, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03868349641561508, "train_loss_bc": 0.03868349641561508, "train_loss_llm": 0.0, "grad_norm": 0.022250710055232048, "global_step": 16325, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03463082015514374, "train_loss_bc": 0.03463082015514374, "train_loss_llm": 0.0, "grad_norm": 0.0354628786444664, "global_step": 16326, "epoch": 453, "lr": 0.001} +{"train_loss": 0.031324516981840134, "train_loss_bc": 0.03131371736526489, "train_loss_llm": 0.001079857349395752, "grad_norm": 0.043906599283218384, "global_step": 16327, "epoch": 453, "lr": 0.001} +{"train_loss": 0.035195156931877136, "train_loss_bc": 0.035195156931877136, "train_loss_llm": 0.0, "grad_norm": 0.0494900606572628, "global_step": 16328, "epoch": 453, "lr": 0.001} +{"train_loss": 0.038973789662122726, "train_loss_bc": 0.038973789662122726, "train_loss_llm": 0.0, "grad_norm": 0.02759752981364727, "global_step": 16329, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03499363735318184, "train_loss_bc": 0.03499363735318184, "train_loss_llm": 0.0, "grad_norm": 0.03558233752846718, "global_step": 16330, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03525782376527786, "train_loss_bc": 0.03525027632713318, "train_loss_llm": 0.000754624605178833, "grad_norm": 0.044662121683359146, "global_step": 16331, "epoch": 453, "lr": 0.001} +{"train_loss": 0.031033575534820557, "train_loss_bc": 0.031011229380965233, "train_loss_llm": 0.00223463773727417, "grad_norm": 0.05108697712421417, "global_step": 16332, "epoch": 453, "lr": 0.001} +{"train_loss": 0.036227162927389145, "train_loss_bc": 0.036227162927389145, "train_loss_llm": 0.0, "grad_norm": 0.02537742629647255, "global_step": 16333, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03507401421666145, "train_loss_bc": 0.03507401421666145, "train_loss_llm": 0.0, "grad_norm": 0.0376625582575798, "global_step": 16334, "epoch": 453, "lr": 0.001} +{"train_loss": 0.037800077348947525, "train_loss_bc": 0.037800077348947525, "train_loss_llm": 0.0, "grad_norm": 0.04665767773985863, "global_step": 16335, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03577117994427681, "train_loss_bc": 0.03574948385357857, "train_loss_llm": 0.002169668674468994, "grad_norm": 0.06087314337491989, "global_step": 16336, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03743318095803261, "train_loss_bc": 0.03742269426584244, "train_loss_llm": 0.0010485351085662842, "grad_norm": 0.02717587910592556, "global_step": 16337, "epoch": 453, "lr": 0.001} +{"train_loss": 0.034661728888750076, "train_loss_bc": 0.034661728888750076, "train_loss_llm": 0.0, "grad_norm": 0.040963269770145416, "global_step": 16338, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03688137233257294, "train_loss_bc": 0.03688137233257294, "train_loss_llm": 0.0, "grad_norm": 0.06127612665295601, "global_step": 16339, "epoch": 453, "lr": 0.001} +{"train_loss": 0.036224447190761566, "train_loss_bc": 0.03622160106897354, "train_loss_llm": 0.00028446316719055176, "grad_norm": 0.07159500569105148, "global_step": 16340, "epoch": 453, "lr": 0.001} +{"train_loss": 0.043111927807331085, "train_loss_bc": 0.04310043901205063, "train_loss_llm": 0.0011490285396575928, "grad_norm": 0.03152009844779968, "global_step": 16341, "epoch": 453, "lr": 0.001} +{"train_loss": 0.029224086552858353, "train_loss_bc": 0.029224086552858353, "train_loss_llm": 0.0, "grad_norm": 0.04265691712498665, "global_step": 16342, "epoch": 453, "lr": 0.001} +{"train_loss": 0.03474357403400871, "train_loss_bc": 0.032644838094711304, "train_loss_llm": 0.0, "grad_norm": 0.06208585202693939, "global_step": 16343, "epoch": 453, "lr": 0.001, "val_loss": 0.07082201540470123} +{"train_loss": 0.030715441331267357, "train_loss_bc": 0.030715441331267357, "train_loss_llm": 0.0, "grad_norm": 0.073860764503479, "global_step": 16344, "epoch": 454, "lr": 0.001} +{"train_loss": 0.035050418227910995, "train_loss_bc": 0.035050418227910995, "train_loss_llm": 0.0, "grad_norm": 0.021463407203555107, "global_step": 16345, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03145880997180939, "train_loss_bc": 0.03145880997180939, "train_loss_llm": 0.0, "grad_norm": 0.03598623722791672, "global_step": 16346, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03608093410730362, "train_loss_bc": 0.03608093410730362, "train_loss_llm": 0.0, "grad_norm": 0.055575333535671234, "global_step": 16347, "epoch": 454, "lr": 0.001} +{"train_loss": 0.032294027507305145, "train_loss_bc": 0.032294027507305145, "train_loss_llm": 0.0, "grad_norm": 0.06339131295681, "global_step": 16348, "epoch": 454, "lr": 0.001} +{"train_loss": 0.0362342968583107, "train_loss_bc": 0.0362342968583107, "train_loss_llm": 0.0, "grad_norm": 0.024178041145205498, "global_step": 16349, "epoch": 454, "lr": 0.001} +{"train_loss": 0.030769657343626022, "train_loss_bc": 0.03076699748635292, "train_loss_llm": 0.00026598572731018066, "grad_norm": 0.0327911339700222, "global_step": 16350, "epoch": 454, "lr": 0.001} +{"train_loss": 0.030123578384518623, "train_loss_bc": 0.030123578384518623, "train_loss_llm": 0.0, "grad_norm": 0.04276237264275551, "global_step": 16351, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03862916678190231, "train_loss_bc": 0.03862916678190231, "train_loss_llm": 0.0, "grad_norm": 0.04572511091828346, "global_step": 16352, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03517447039484978, "train_loss_bc": 0.03517447039484978, "train_loss_llm": 0.0, "grad_norm": 0.026788106188178062, "global_step": 16353, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03590253368020058, "train_loss_bc": 0.03588822856545448, "train_loss_llm": 0.0014303326606750488, "grad_norm": 0.03864986076951027, "global_step": 16354, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03123364970088005, "train_loss_bc": 0.03123364970088005, "train_loss_llm": 0.0, "grad_norm": 0.04307073354721069, "global_step": 16355, "epoch": 454, "lr": 0.001} +{"train_loss": 0.034198831766843796, "train_loss_bc": 0.034198831766843796, "train_loss_llm": 0.0, "grad_norm": 0.05568201467394829, "global_step": 16356, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03453373908996582, "train_loss_bc": 0.03453373908996582, "train_loss_llm": 0.0, "grad_norm": 0.030949408188462257, "global_step": 16357, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03787383809685707, "train_loss_bc": 0.03787383809685707, "train_loss_llm": 0.0, "grad_norm": 0.048637304455041885, "global_step": 16358, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03195340186357498, "train_loss_bc": 0.031951501965522766, "train_loss_llm": 0.00018984079360961914, "grad_norm": 0.06619858741760254, "global_step": 16359, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03263444826006889, "train_loss_bc": 0.03263444826006889, "train_loss_llm": 0.0, "grad_norm": 0.08251874893903732, "global_step": 16360, "epoch": 454, "lr": 0.001} +{"train_loss": 0.035553544759750366, "train_loss_bc": 0.035553544759750366, "train_loss_llm": 0.0, "grad_norm": 0.023587344214320183, "global_step": 16361, "epoch": 454, "lr": 0.001} +{"train_loss": 0.037829164415597916, "train_loss_bc": 0.037829164415597916, "train_loss_llm": 0.0, "grad_norm": 0.034458648413419724, "global_step": 16362, "epoch": 454, "lr": 0.001} +{"train_loss": 0.04345069080591202, "train_loss_bc": 0.04345069080591202, "train_loss_llm": 0.0, "grad_norm": 0.05768389254808426, "global_step": 16363, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03835141658782959, "train_loss_bc": 0.0383295938372612, "train_loss_llm": 0.002182304859161377, "grad_norm": 0.07578188180923462, "global_step": 16364, "epoch": 454, "lr": 0.001} +{"train_loss": 0.037116050720214844, "train_loss_bc": 0.037112779915332794, "train_loss_llm": 0.00032719969749450684, "grad_norm": 0.030026856809854507, "global_step": 16365, "epoch": 454, "lr": 0.001} +{"train_loss": 0.029089508578181267, "train_loss_bc": 0.029089508578181267, "train_loss_llm": 0.0, "grad_norm": 0.0515897236764431, "global_step": 16366, "epoch": 454, "lr": 0.001} +{"train_loss": 0.04261089488863945, "train_loss_bc": 0.04261089488863945, "train_loss_llm": 0.0, "grad_norm": 0.06864719837903976, "global_step": 16367, "epoch": 454, "lr": 0.001} +{"train_loss": 0.041179824620485306, "train_loss_bc": 0.041179824620485306, "train_loss_llm": 0.0, "grad_norm": 0.07920795679092407, "global_step": 16368, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03601928427815437, "train_loss_bc": 0.03601928427815437, "train_loss_llm": 0.0, "grad_norm": 0.027872227132320404, "global_step": 16369, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03178340196609497, "train_loss_bc": 0.03177530691027641, "train_loss_llm": 0.0008094906806945801, "grad_norm": 0.03802695497870445, "global_step": 16370, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03316493332386017, "train_loss_bc": 0.03316493332386017, "train_loss_llm": 0.0, "grad_norm": 0.052869416773319244, "global_step": 16371, "epoch": 454, "lr": 0.001} +{"train_loss": 0.040791839361190796, "train_loss_bc": 0.040791839361190796, "train_loss_llm": 0.0, "grad_norm": 0.0586492083966732, "global_step": 16372, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03846517205238342, "train_loss_bc": 0.038458723574876785, "train_loss_llm": 0.0006447136402130127, "grad_norm": 0.03606355935335159, "global_step": 16373, "epoch": 454, "lr": 0.001} +{"train_loss": 0.02758920006453991, "train_loss_bc": 0.02758920006453991, "train_loss_llm": 0.0, "grad_norm": 0.049028825014829636, "global_step": 16374, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03411325812339783, "train_loss_bc": 0.03411325812339783, "train_loss_llm": 0.0, "grad_norm": 0.06887784600257874, "global_step": 16375, "epoch": 454, "lr": 0.001} +{"train_loss": 0.02907182276248932, "train_loss_bc": 0.029071403667330742, "train_loss_llm": 4.190206527709961e-05, "grad_norm": 0.09383711963891983, "global_step": 16376, "epoch": 454, "lr": 0.001} +{"train_loss": 0.0374954417347908, "train_loss_bc": 0.0374954417347908, "train_loss_llm": 0.0, "grad_norm": 0.027399517595767975, "global_step": 16377, "epoch": 454, "lr": 0.001} +{"train_loss": 0.033828988671302795, "train_loss_bc": 0.033828988671302795, "train_loss_llm": 0.0, "grad_norm": 0.04487507417798042, "global_step": 16378, "epoch": 454, "lr": 0.001} +{"train_loss": 0.03490927287687858, "train_loss_bc": 0.03436814248561859, "train_loss_llm": 0.0, "grad_norm": 0.06754034012556076, "global_step": 16379, "epoch": 454, "lr": 0.001, "val_loss": 0.07207418233156204} +{"train_loss": 0.037889689207077026, "train_loss_bc": 0.037889689207077026, "train_loss_llm": 0.0, "grad_norm": 0.088472880423069, "global_step": 16380, "epoch": 455, "lr": 0.001} +{"train_loss": 0.031881850212812424, "train_loss_bc": 0.031881850212812424, "train_loss_llm": 0.0, "grad_norm": 0.03046111762523651, "global_step": 16381, "epoch": 455, "lr": 0.001} +{"train_loss": 0.02937684766948223, "train_loss_bc": 0.02937684766948223, "train_loss_llm": 0.0, "grad_norm": 0.05079234391450882, "global_step": 16382, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03778832405805588, "train_loss_bc": 0.03778832405805588, "train_loss_llm": 0.0, "grad_norm": 0.06917667388916016, "global_step": 16383, "epoch": 455, "lr": 0.001} +{"train_loss": 0.038669124245643616, "train_loss_bc": 0.03865846246480942, "train_loss_llm": 0.001066356897354126, "grad_norm": 0.08395310491323471, "global_step": 16384, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03608138859272003, "train_loss_bc": 0.0360800065100193, "train_loss_llm": 0.00013813376426696777, "grad_norm": 0.02553648129105568, "global_step": 16385, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03182683140039444, "train_loss_bc": 0.03182683140039444, "train_loss_llm": 0.0, "grad_norm": 0.038006491959095, "global_step": 16386, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03742395341396332, "train_loss_bc": 0.03742395341396332, "train_loss_llm": 0.0, "grad_norm": 0.04545076936483383, "global_step": 16387, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03723952919244766, "train_loss_bc": 0.03723952919244766, "train_loss_llm": 0.0, "grad_norm": 0.053608451038599014, "global_step": 16388, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03473784402012825, "train_loss_bc": 0.03473784402012825, "train_loss_llm": 0.0, "grad_norm": 0.03317758068442345, "global_step": 16389, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03560422360897064, "train_loss_bc": 0.03559742122888565, "train_loss_llm": 0.0006802380084991455, "grad_norm": 0.04395716264843941, "global_step": 16390, "epoch": 455, "lr": 0.001} +{"train_loss": 0.026088478043675423, "train_loss_bc": 0.026088478043675423, "train_loss_llm": 0.0, "grad_norm": 0.050116222351789474, "global_step": 16391, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03472282737493515, "train_loss_bc": 0.03471604362130165, "train_loss_llm": 0.0006783008575439453, "grad_norm": 0.05536801367998123, "global_step": 16392, "epoch": 455, "lr": 0.001} +{"train_loss": 0.030868306756019592, "train_loss_bc": 0.030865641310811043, "train_loss_llm": 0.00026658177375793457, "grad_norm": 0.02624918520450592, "global_step": 16393, "epoch": 455, "lr": 0.001} +{"train_loss": 0.04071474075317383, "train_loss_bc": 0.04071474075317383, "train_loss_llm": 0.0, "grad_norm": 0.04715773090720177, "global_step": 16394, "epoch": 455, "lr": 0.001} +{"train_loss": 0.036346517503261566, "train_loss_bc": 0.036346517503261566, "train_loss_llm": 0.0, "grad_norm": 0.06413332372903824, "global_step": 16395, "epoch": 455, "lr": 0.001} +{"train_loss": 0.031660087406635284, "train_loss_bc": 0.031660087406635284, "train_loss_llm": 0.0, "grad_norm": 0.07819993048906326, "global_step": 16396, "epoch": 455, "lr": 0.001} +{"train_loss": 0.035511329770088196, "train_loss_bc": 0.035511329770088196, "train_loss_llm": 0.0, "grad_norm": 0.029605738818645477, "global_step": 16397, "epoch": 455, "lr": 0.001} +{"train_loss": 0.02560143731534481, "train_loss_bc": 0.025600839406251907, "train_loss_llm": 5.987286567687988e-05, "grad_norm": 0.0436905100941658, "global_step": 16398, "epoch": 455, "lr": 0.001} +{"train_loss": 0.030781717970967293, "train_loss_bc": 0.030771085992455482, "train_loss_llm": 0.0010632872581481934, "grad_norm": 0.0692746639251709, "global_step": 16399, "epoch": 455, "lr": 0.001} +{"train_loss": 0.04017667472362518, "train_loss_bc": 0.04017667472362518, "train_loss_llm": 0.0, "grad_norm": 0.09208232164382935, "global_step": 16400, "epoch": 455, "lr": 0.001} +{"train_loss": 0.038257796317338943, "train_loss_bc": 0.03821587935090065, "train_loss_llm": 0.004191756248474121, "grad_norm": 0.030010467395186424, "global_step": 16401, "epoch": 455, "lr": 0.001} +{"train_loss": 0.031519632786512375, "train_loss_bc": 0.031519632786512375, "train_loss_llm": 0.0, "grad_norm": 0.05122906714677811, "global_step": 16402, "epoch": 455, "lr": 0.001} +{"train_loss": 0.036403410136699677, "train_loss_bc": 0.03640004247426987, "train_loss_llm": 0.0003367960453033447, "grad_norm": 0.07089077681303024, "global_step": 16403, "epoch": 455, "lr": 0.001} +{"train_loss": 0.034200239926576614, "train_loss_bc": 0.034200239926576614, "train_loss_llm": 0.0, "grad_norm": 0.07753784954547882, "global_step": 16404, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03461828455328941, "train_loss_bc": 0.03461828455328941, "train_loss_llm": 0.0, "grad_norm": 0.028518788516521454, "global_step": 16405, "epoch": 455, "lr": 0.001} +{"train_loss": 0.032451942563056946, "train_loss_bc": 0.032451942563056946, "train_loss_llm": 0.0, "grad_norm": 0.05202052742242813, "global_step": 16406, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03186480328440666, "train_loss_bc": 0.03186480328440666, "train_loss_llm": 0.0, "grad_norm": 0.06644435226917267, "global_step": 16407, "epoch": 455, "lr": 0.001} +{"train_loss": 0.040560513734817505, "train_loss_bc": 0.04055449366569519, "train_loss_llm": 0.0006020069122314453, "grad_norm": 0.07737747579813004, "global_step": 16408, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03163830563426018, "train_loss_bc": 0.03163830563426018, "train_loss_llm": 0.0, "grad_norm": 0.024924200028181076, "global_step": 16409, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03526158630847931, "train_loss_bc": 0.03526158630847931, "train_loss_llm": 0.0, "grad_norm": 0.03671231493353844, "global_step": 16410, "epoch": 455, "lr": 0.001} +{"train_loss": 0.04138495400547981, "train_loss_bc": 0.04138495400547981, "train_loss_llm": 0.0, "grad_norm": 0.055364932864904404, "global_step": 16411, "epoch": 455, "lr": 0.001} +{"train_loss": 0.042556215077638626, "train_loss_bc": 0.04252185672521591, "train_loss_llm": 0.003435671329498291, "grad_norm": 0.0742439329624176, "global_step": 16412, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03609976917505264, "train_loss_bc": 0.036098841577768326, "train_loss_llm": 9.274482727050781e-05, "grad_norm": 0.028047723695635796, "global_step": 16413, "epoch": 455, "lr": 0.001} +{"train_loss": 0.035002242773771286, "train_loss_bc": 0.035002242773771286, "train_loss_llm": 0.0, "grad_norm": 0.03684275597333908, "global_step": 16414, "epoch": 455, "lr": 0.001} +{"train_loss": 0.03494731129871474, "train_loss_bc": 0.03529178723692894, "train_loss_llm": 0.0, "grad_norm": 0.04988045617938042, "global_step": 16415, "epoch": 455, "lr": 0.001, "train/cumulative_reward": 86.21572778639316, "train/mean_score": 22.92106530065309, "train/success_rate": 0.7, "test/cumulative_reward": 62.40647119169851, "test/mean_score": 9.94150527679202, "test/success_rate": 0.24, "val_loss": 0.0729135051369667, "train_action_mse_error": 0.0380462110042572} +{"train_loss": 0.029239222407341003, "train_loss_bc": 0.029239222407341003, "train_loss_llm": 0.0, "grad_norm": 0.06239130347967148, "global_step": 16416, "epoch": 456, "lr": 0.001} +{"train_loss": 0.037205927073955536, "train_loss_bc": 0.037205927073955536, "train_loss_llm": 0.0, "grad_norm": 0.023521974682807922, "global_step": 16417, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03571958839893341, "train_loss_bc": 0.03571958839893341, "train_loss_llm": 0.0, "grad_norm": 0.03037417121231556, "global_step": 16418, "epoch": 456, "lr": 0.001} +{"train_loss": 0.036316730082035065, "train_loss_bc": 0.036316730082035065, "train_loss_llm": 0.0, "grad_norm": 0.0432114452123642, "global_step": 16419, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03716377541422844, "train_loss_bc": 0.03716377541422844, "train_loss_llm": 0.0, "grad_norm": 0.05644836276769638, "global_step": 16420, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03113110549747944, "train_loss_bc": 0.031129643321037292, "train_loss_llm": 0.0001462399959564209, "grad_norm": 0.02796107344329357, "global_step": 16421, "epoch": 456, "lr": 0.001} +{"train_loss": 0.0338953360915184, "train_loss_bc": 0.0338953360915184, "train_loss_llm": 0.0, "grad_norm": 0.046562228351831436, "global_step": 16422, "epoch": 456, "lr": 0.001} +{"train_loss": 0.0355813130736351, "train_loss_bc": 0.0355813130736351, "train_loss_llm": 0.0, "grad_norm": 0.06428549438714981, "global_step": 16423, "epoch": 456, "lr": 0.001} +{"train_loss": 0.02922195754945278, "train_loss_bc": 0.02922195754945278, "train_loss_llm": 0.0, "grad_norm": 0.07881302386522293, "global_step": 16424, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03606927767395973, "train_loss_bc": 0.03606927767395973, "train_loss_llm": 0.0, "grad_norm": 0.0205268245190382, "global_step": 16425, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03531754016876221, "train_loss_bc": 0.03531754016876221, "train_loss_llm": 0.0, "grad_norm": 0.03045986033976078, "global_step": 16426, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03270027041435242, "train_loss_bc": 0.03270027041435242, "train_loss_llm": 0.0, "grad_norm": 0.04766752943396568, "global_step": 16427, "epoch": 456, "lr": 0.001} +{"train_loss": 0.04227351024746895, "train_loss_bc": 0.04224462807178497, "train_loss_llm": 0.002888202667236328, "grad_norm": 0.0714695155620575, "global_step": 16428, "epoch": 456, "lr": 0.001} +{"train_loss": 0.036522358655929565, "train_loss_bc": 0.036514755338430405, "train_loss_llm": 0.0007604062557220459, "grad_norm": 0.025059090927243233, "global_step": 16429, "epoch": 456, "lr": 0.001} +{"train_loss": 0.0347614623606205, "train_loss_bc": 0.03475639224052429, "train_loss_llm": 0.0005069375038146973, "grad_norm": 0.039693642407655716, "global_step": 16430, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03875777870416641, "train_loss_bc": 0.03875777870416641, "train_loss_llm": 0.0, "grad_norm": 0.06424863636493683, "global_step": 16431, "epoch": 456, "lr": 0.001} +{"train_loss": 0.030663257464766502, "train_loss_bc": 0.030663257464766502, "train_loss_llm": 0.0, "grad_norm": 0.07440977543592453, "global_step": 16432, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03533102944493294, "train_loss_bc": 0.03533102944493294, "train_loss_llm": 0.0, "grad_norm": 0.027035102248191833, "global_step": 16433, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03178956359624863, "train_loss_bc": 0.03177860751748085, "train_loss_llm": 0.001095592975616455, "grad_norm": 0.04765469953417778, "global_step": 16434, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03134278953075409, "train_loss_bc": 0.03132777661085129, "train_loss_llm": 0.001501321792602539, "grad_norm": 0.06855448335409164, "global_step": 16435, "epoch": 456, "lr": 0.001} +{"train_loss": 0.04212028533220291, "train_loss_bc": 0.04212028533220291, "train_loss_llm": 0.0, "grad_norm": 0.0836694985628128, "global_step": 16436, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03629295900464058, "train_loss_bc": 0.03627640753984451, "train_loss_llm": 0.0016550123691558838, "grad_norm": 0.03091953508555889, "global_step": 16437, "epoch": 456, "lr": 0.001} +{"train_loss": 0.040134843438863754, "train_loss_bc": 0.040134843438863754, "train_loss_llm": 0.0, "grad_norm": 0.04503714665770531, "global_step": 16438, "epoch": 456, "lr": 0.001} +{"train_loss": 0.0388229601085186, "train_loss_bc": 0.03880877420306206, "train_loss_llm": 0.0014184117317199707, "grad_norm": 0.06724030524492264, "global_step": 16439, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03256353735923767, "train_loss_bc": 0.03256353735923767, "train_loss_llm": 0.0, "grad_norm": 0.08538414537906647, "global_step": 16440, "epoch": 456, "lr": 0.001} +{"train_loss": 0.0315474271774292, "train_loss_bc": 0.0315474271774292, "train_loss_llm": 0.0, "grad_norm": 0.026393769308924675, "global_step": 16441, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03463272377848625, "train_loss_bc": 0.03463272377848625, "train_loss_llm": 0.0, "grad_norm": 0.04503058269619942, "global_step": 16442, "epoch": 456, "lr": 0.001} +{"train_loss": 0.0317574217915535, "train_loss_bc": 0.0317574217915535, "train_loss_llm": 0.0, "grad_norm": 0.06287190318107605, "global_step": 16443, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03438059240579605, "train_loss_bc": 0.03437837213277817, "train_loss_llm": 0.00022202730178833008, "grad_norm": 0.07985062897205353, "global_step": 16444, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03641650080680847, "train_loss_bc": 0.03641650080680847, "train_loss_llm": 0.0, "grad_norm": 0.032784417271614075, "global_step": 16445, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03986124321818352, "train_loss_bc": 0.03986124321818352, "train_loss_llm": 0.0, "grad_norm": 0.04020894318819046, "global_step": 16446, "epoch": 456, "lr": 0.001} +{"train_loss": 0.04110172763466835, "train_loss_bc": 0.04110172763466835, "train_loss_llm": 0.0, "grad_norm": 0.05834732577204704, "global_step": 16447, "epoch": 456, "lr": 0.001} +{"train_loss": 0.030697496607899666, "train_loss_bc": 0.030697496607899666, "train_loss_llm": 0.0, "grad_norm": 0.06146058812737465, "global_step": 16448, "epoch": 456, "lr": 0.001} +{"train_loss": 0.033666711300611496, "train_loss_bc": 0.033666711300611496, "train_loss_llm": 0.0, "grad_norm": 0.02312123216688633, "global_step": 16449, "epoch": 456, "lr": 0.001} +{"train_loss": 0.03304944187402725, "train_loss_bc": 0.033029329031705856, "train_loss_llm": 0.0020112693309783936, "grad_norm": 0.0379410982131958, "global_step": 16450, "epoch": 456, "lr": 0.001} +{"train_loss": 0.034944366508473955, "train_loss_bc": 0.029940834268927574, "train_loss_llm": 0.0006693899631500244, "grad_norm": 0.058079518377780914, "global_step": 16451, "epoch": 456, "lr": 0.001, "val_loss": 0.07061481475830078} +{"train_loss": 0.034130144864320755, "train_loss_bc": 0.034130144864320755, "train_loss_llm": 0.0, "grad_norm": 0.07132779061794281, "global_step": 16452, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03662692382931709, "train_loss_bc": 0.03662655130028725, "train_loss_llm": 3.731250762939453e-05, "grad_norm": 0.0324583575129509, "global_step": 16453, "epoch": 457, "lr": 0.001} +{"train_loss": 0.032008666545152664, "train_loss_bc": 0.032008666545152664, "train_loss_llm": 0.0, "grad_norm": 0.05303601548075676, "global_step": 16454, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03881695494055748, "train_loss_bc": 0.03881695494055748, "train_loss_llm": 0.0, "grad_norm": 0.06630654633045197, "global_step": 16455, "epoch": 457, "lr": 0.001} +{"train_loss": 0.04032288119196892, "train_loss_bc": 0.04032288119196892, "train_loss_llm": 0.0, "grad_norm": 0.09293243288993835, "global_step": 16456, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03384259715676308, "train_loss_bc": 0.03384259715676308, "train_loss_llm": 0.0, "grad_norm": 0.029752612113952637, "global_step": 16457, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03799661621451378, "train_loss_bc": 0.03799661621451378, "train_loss_llm": 0.0, "grad_norm": 0.0546458475291729, "global_step": 16458, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03500613197684288, "train_loss_bc": 0.03500613197684288, "train_loss_llm": 0.0, "grad_norm": 0.0760076642036438, "global_step": 16459, "epoch": 457, "lr": 0.001} +{"train_loss": 0.04118386283516884, "train_loss_bc": 0.04118386283516884, "train_loss_llm": 0.0, "grad_norm": 0.09782816469669342, "global_step": 16460, "epoch": 457, "lr": 0.001} +{"train_loss": 0.026205413043498993, "train_loss_bc": 0.026205413043498993, "train_loss_llm": 0.0, "grad_norm": 0.02263866923749447, "global_step": 16461, "epoch": 457, "lr": 0.001} +{"train_loss": 0.030027739703655243, "train_loss_bc": 0.030027739703655243, "train_loss_llm": 0.0, "grad_norm": 0.04554489254951477, "global_step": 16462, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03953033313155174, "train_loss_bc": 0.03953033313155174, "train_loss_llm": 0.0, "grad_norm": 0.06227932870388031, "global_step": 16463, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03356309235095978, "train_loss_bc": 0.03356309235095978, "train_loss_llm": 0.0, "grad_norm": 0.08130204677581787, "global_step": 16464, "epoch": 457, "lr": 0.001} +{"train_loss": 0.035451192408800125, "train_loss_bc": 0.035451192408800125, "train_loss_llm": 0.0, "grad_norm": 0.022630421444773674, "global_step": 16465, "epoch": 457, "lr": 0.001} +{"train_loss": 0.0324677936732769, "train_loss_bc": 0.0324677936732769, "train_loss_llm": 0.0, "grad_norm": 0.03409869596362114, "global_step": 16466, "epoch": 457, "lr": 0.001} +{"train_loss": 0.031220875680446625, "train_loss_bc": 0.031214699149131775, "train_loss_llm": 0.00061759352684021, "grad_norm": 0.04274412989616394, "global_step": 16467, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03438867628574371, "train_loss_bc": 0.03438867628574371, "train_loss_llm": 0.0, "grad_norm": 0.04862663894891739, "global_step": 16468, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03396441414952278, "train_loss_bc": 0.03396441414952278, "train_loss_llm": 0.0, "grad_norm": 0.025182198733091354, "global_step": 16469, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03072589822113514, "train_loss_bc": 0.03072589822113514, "train_loss_llm": 0.0, "grad_norm": 0.03608709201216698, "global_step": 16470, "epoch": 457, "lr": 0.001} +{"train_loss": 0.039311688393354416, "train_loss_bc": 0.039311688393354416, "train_loss_llm": 0.0, "grad_norm": 0.055445894598960876, "global_step": 16471, "epoch": 457, "lr": 0.001} +{"train_loss": 0.029484082013368607, "train_loss_bc": 0.029480744153261185, "train_loss_llm": 0.0003338456153869629, "grad_norm": 0.06088493764400482, "global_step": 16472, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03336106613278389, "train_loss_bc": 0.03336106613278389, "train_loss_llm": 0.0, "grad_norm": 0.026814093813300133, "global_step": 16473, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03894178941845894, "train_loss_bc": 0.03894178941845894, "train_loss_llm": 0.0, "grad_norm": 0.05235045775771141, "global_step": 16474, "epoch": 457, "lr": 0.001} +{"train_loss": 0.0352986603975296, "train_loss_bc": 0.03529287874698639, "train_loss_llm": 0.0005782246589660645, "grad_norm": 0.06970621645450592, "global_step": 16475, "epoch": 457, "lr": 0.001} +{"train_loss": 0.04167715087532997, "train_loss_bc": 0.041662223637104034, "train_loss_llm": 0.0014927089214324951, "grad_norm": 0.09424302726984024, "global_step": 16476, "epoch": 457, "lr": 0.001} +{"train_loss": 0.0428076833486557, "train_loss_bc": 0.0428076833486557, "train_loss_llm": 0.0, "grad_norm": 0.03271637111902237, "global_step": 16477, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03282269090414047, "train_loss_bc": 0.03281049802899361, "train_loss_llm": 0.0012192726135253906, "grad_norm": 0.05486848205327988, "global_step": 16478, "epoch": 457, "lr": 0.001} +{"train_loss": 0.039588287472724915, "train_loss_bc": 0.03957805037498474, "train_loss_llm": 0.0010235309600830078, "grad_norm": 0.07381601631641388, "global_step": 16479, "epoch": 457, "lr": 0.001} +{"train_loss": 0.030285704880952835, "train_loss_bc": 0.030285704880952835, "train_loss_llm": 0.0, "grad_norm": 0.09772893041372299, "global_step": 16480, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03203963860869408, "train_loss_bc": 0.03203963860869408, "train_loss_llm": 0.0, "grad_norm": 0.020472750067710876, "global_step": 16481, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03302042931318283, "train_loss_bc": 0.03302042931318283, "train_loss_llm": 0.0, "grad_norm": 0.03140256553888321, "global_step": 16482, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03210088238120079, "train_loss_bc": 0.03210088238120079, "train_loss_llm": 0.0, "grad_norm": 0.050807878375053406, "global_step": 16483, "epoch": 457, "lr": 0.001} +{"train_loss": 0.041757870465517044, "train_loss_bc": 0.041757870465517044, "train_loss_llm": 0.0, "grad_norm": 0.07206882536411285, "global_step": 16484, "epoch": 457, "lr": 0.001} +{"train_loss": 0.031776465475559235, "train_loss_bc": 0.031776465475559235, "train_loss_llm": 0.0, "grad_norm": 0.025691621005535126, "global_step": 16485, "epoch": 457, "lr": 0.001} +{"train_loss": 0.038026515394449234, "train_loss_bc": 0.038026515394449234, "train_loss_llm": 0.0, "grad_norm": 0.047021809965372086, "global_step": 16486, "epoch": 457, "lr": 0.001} +{"train_loss": 0.03512879849101106, "train_loss_bc": 0.034855931997299194, "train_loss_llm": 0.0, "grad_norm": 0.06867362558841705, "global_step": 16487, "epoch": 457, "lr": 0.001, "val_loss": 0.07170996069908142} +{"train_loss": 0.03687246888875961, "train_loss_bc": 0.03687246888875961, "train_loss_llm": 0.0, "grad_norm": 0.06883248686790466, "global_step": 16488, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03644802048802376, "train_loss_bc": 0.03644802048802376, "train_loss_llm": 0.0, "grad_norm": 0.03416607901453972, "global_step": 16489, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03015298768877983, "train_loss_bc": 0.03015298768877983, "train_loss_llm": 0.0, "grad_norm": 0.045462094247341156, "global_step": 16490, "epoch": 458, "lr": 0.001} +{"train_loss": 0.030379436910152435, "train_loss_bc": 0.030379436910152435, "train_loss_llm": 0.0, "grad_norm": 0.059780482202768326, "global_step": 16491, "epoch": 458, "lr": 0.001} +{"train_loss": 0.035250455141067505, "train_loss_bc": 0.035250455141067505, "train_loss_llm": 0.0, "grad_norm": 0.07374941557645798, "global_step": 16492, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03115590661764145, "train_loss_bc": 0.031120624393224716, "train_loss_llm": 0.003528207540512085, "grad_norm": 0.03245258331298828, "global_step": 16493, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03619041666388512, "train_loss_bc": 0.03619041666388512, "train_loss_llm": 0.0, "grad_norm": 0.054394934326410294, "global_step": 16494, "epoch": 458, "lr": 0.001} +{"train_loss": 0.034850962460041046, "train_loss_bc": 0.034850962460041046, "train_loss_llm": 0.0, "grad_norm": 0.08251969516277313, "global_step": 16495, "epoch": 458, "lr": 0.001} +{"train_loss": 0.028634047135710716, "train_loss_bc": 0.028634047135710716, "train_loss_llm": 0.0, "grad_norm": 0.09121667593717575, "global_step": 16496, "epoch": 458, "lr": 0.001} +{"train_loss": 0.0406835675239563, "train_loss_bc": 0.0406835675239563, "train_loss_llm": 0.0, "grad_norm": 0.023607278242707253, "global_step": 16497, "epoch": 458, "lr": 0.001} +{"train_loss": 0.036887116730213165, "train_loss_bc": 0.03687944635748863, "train_loss_llm": 0.0007670223712921143, "grad_norm": 0.03219058737158775, "global_step": 16498, "epoch": 458, "lr": 0.001} +{"train_loss": 0.043404433876276016, "train_loss_bc": 0.043404433876276016, "train_loss_llm": 0.0, "grad_norm": 0.05729174613952637, "global_step": 16499, "epoch": 458, "lr": 0.001} +{"train_loss": 0.035691238939762115, "train_loss_bc": 0.035691238939762115, "train_loss_llm": 0.0, "grad_norm": 0.06703899800777435, "global_step": 16500, "epoch": 458, "lr": 0.001} +{"train_loss": 0.036472927778959274, "train_loss_bc": 0.036472927778959274, "train_loss_llm": 0.0, "grad_norm": 0.02497020736336708, "global_step": 16501, "epoch": 458, "lr": 0.001} +{"train_loss": 0.029880089685320854, "train_loss_bc": 0.029861094430088997, "train_loss_llm": 0.0018994808197021484, "grad_norm": 0.04003601521253586, "global_step": 16502, "epoch": 458, "lr": 0.001} +{"train_loss": 0.0295669324696064, "train_loss_bc": 0.029562685638666153, "train_loss_llm": 0.0004246532917022705, "grad_norm": 0.05076484754681587, "global_step": 16503, "epoch": 458, "lr": 0.001} +{"train_loss": 0.035256147384643555, "train_loss_bc": 0.03524656221270561, "train_loss_llm": 0.0009586513042449951, "grad_norm": 0.06693215668201447, "global_step": 16504, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03958423435688019, "train_loss_bc": 0.03958423435688019, "train_loss_llm": 0.0, "grad_norm": 0.026674829423427582, "global_step": 16505, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03461412340402603, "train_loss_bc": 0.03461412340402603, "train_loss_llm": 0.0, "grad_norm": 0.037270158529281616, "global_step": 16506, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03608158230781555, "train_loss_bc": 0.03608158230781555, "train_loss_llm": 0.0, "grad_norm": 0.056629523634910583, "global_step": 16507, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03285461291670799, "train_loss_bc": 0.03285461291670799, "train_loss_llm": 0.0, "grad_norm": 0.06644909083843231, "global_step": 16508, "epoch": 458, "lr": 0.001} +{"train_loss": 0.034285951405763626, "train_loss_bc": 0.034285951405763626, "train_loss_llm": 0.0, "grad_norm": 0.020738234743475914, "global_step": 16509, "epoch": 458, "lr": 0.001} +{"train_loss": 0.036113716661930084, "train_loss_bc": 0.036113716661930084, "train_loss_llm": 0.0, "grad_norm": 0.0392143651843071, "global_step": 16510, "epoch": 458, "lr": 0.001} +{"train_loss": 0.035874415189027786, "train_loss_bc": 0.03585910052061081, "train_loss_llm": 0.0015313923358917236, "grad_norm": 0.05792875960469246, "global_step": 16511, "epoch": 458, "lr": 0.001} +{"train_loss": 0.029996193945407867, "train_loss_bc": 0.029996193945407867, "train_loss_llm": 0.0, "grad_norm": 0.0755254328250885, "global_step": 16512, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03939589112997055, "train_loss_bc": 0.03939157351851463, "train_loss_llm": 0.000431746244430542, "grad_norm": 0.026779616251587868, "global_step": 16513, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03440049663186073, "train_loss_bc": 0.03440049663186073, "train_loss_llm": 0.0, "grad_norm": 0.04274130240082741, "global_step": 16514, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03406186401844025, "train_loss_bc": 0.03406186401844025, "train_loss_llm": 0.0, "grad_norm": 0.07201163470745087, "global_step": 16515, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03617754578590393, "train_loss_bc": 0.03617754578590393, "train_loss_llm": 0.0, "grad_norm": 0.10009068250656128, "global_step": 16516, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03630390018224716, "train_loss_bc": 0.03627970442175865, "train_loss_llm": 0.0024197399616241455, "grad_norm": 0.02780178189277649, "global_step": 16517, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03492819517850876, "train_loss_bc": 0.03492273390293121, "train_loss_llm": 0.0005459487438201904, "grad_norm": 0.048058126121759415, "global_step": 16518, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03880097717046738, "train_loss_bc": 0.03880097717046738, "train_loss_llm": 0.0, "grad_norm": 0.06354445219039917, "global_step": 16519, "epoch": 458, "lr": 0.001} +{"train_loss": 0.04164966568350792, "train_loss_bc": 0.04164966568350792, "train_loss_llm": 0.0, "grad_norm": 0.09084992110729218, "global_step": 16520, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03146723285317421, "train_loss_bc": 0.03146723285317421, "train_loss_llm": 0.0, "grad_norm": 0.0267203189432621, "global_step": 16521, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03320887312293053, "train_loss_bc": 0.03320887312293053, "train_loss_llm": 0.0, "grad_norm": 0.03726363554596901, "global_step": 16522, "epoch": 458, "lr": 0.001} +{"train_loss": 0.03504844361709224, "train_loss_bc": 0.0341673418879509, "train_loss_llm": 0.0, "grad_norm": 0.04726093262434006, "global_step": 16523, "epoch": 458, "lr": 0.001, "val_loss": 0.07240036875009537} +{"train_loss": 0.0351804718375206, "train_loss_bc": 0.0351804718375206, "train_loss_llm": 0.0, "grad_norm": 0.053565628826618195, "global_step": 16524, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03615470230579376, "train_loss_bc": 0.03615470230579376, "train_loss_llm": 0.0, "grad_norm": 0.024220509454607964, "global_step": 16525, "epoch": 459, "lr": 0.001} +{"train_loss": 0.031229926273226738, "train_loss_bc": 0.03122747503221035, "train_loss_llm": 0.00024503469467163086, "grad_norm": 0.03722100332379341, "global_step": 16526, "epoch": 459, "lr": 0.001} +{"train_loss": 0.028469765558838844, "train_loss_bc": 0.02845909260213375, "train_loss_llm": 0.0010672211647033691, "grad_norm": 0.049403101205825806, "global_step": 16527, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03168123960494995, "train_loss_bc": 0.03168123960494995, "train_loss_llm": 0.0, "grad_norm": 0.05979694053530693, "global_step": 16528, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03643307834863663, "train_loss_bc": 0.03643307834863663, "train_loss_llm": 0.0, "grad_norm": 0.027257030829787254, "global_step": 16529, "epoch": 459, "lr": 0.001} +{"train_loss": 0.039355162531137466, "train_loss_bc": 0.039355162531137466, "train_loss_llm": 0.0, "grad_norm": 0.04612826183438301, "global_step": 16530, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03363834321498871, "train_loss_bc": 0.03363834321498871, "train_loss_llm": 0.0, "grad_norm": 0.054868433624506, "global_step": 16531, "epoch": 459, "lr": 0.001} +{"train_loss": 0.04355790093541145, "train_loss_bc": 0.04353794455528259, "train_loss_llm": 0.0019957125186920166, "grad_norm": 0.07004342973232269, "global_step": 16532, "epoch": 459, "lr": 0.001} +{"train_loss": 0.031721312552690506, "train_loss_bc": 0.031721312552690506, "train_loss_llm": 0.0, "grad_norm": 0.029331915080547333, "global_step": 16533, "epoch": 459, "lr": 0.001} +{"train_loss": 0.04006962850689888, "train_loss_bc": 0.04006962850689888, "train_loss_llm": 0.0, "grad_norm": 0.044244568794965744, "global_step": 16534, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03727705031633377, "train_loss_bc": 0.03727705031633377, "train_loss_llm": 0.0, "grad_norm": 0.06511899083852768, "global_step": 16535, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03802819550037384, "train_loss_bc": 0.03801507130265236, "train_loss_llm": 0.0013124644756317139, "grad_norm": 0.08720877021551132, "global_step": 16536, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03271130099892616, "train_loss_bc": 0.03271130099892616, "train_loss_llm": 0.0, "grad_norm": 0.024896468967199326, "global_step": 16537, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03190365061163902, "train_loss_bc": 0.03190365061163902, "train_loss_llm": 0.0, "grad_norm": 0.03925497457385063, "global_step": 16538, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03958137333393097, "train_loss_bc": 0.03955687955021858, "train_loss_llm": 0.0024493038654327393, "grad_norm": 0.05746133625507355, "global_step": 16539, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03736831247806549, "train_loss_bc": 0.03736831247806549, "train_loss_llm": 0.0, "grad_norm": 0.07391931116580963, "global_step": 16540, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03640887141227722, "train_loss_bc": 0.03640887141227722, "train_loss_llm": 0.0, "grad_norm": 0.029084188863635063, "global_step": 16541, "epoch": 459, "lr": 0.001} +{"train_loss": 0.02591974101960659, "train_loss_bc": 0.02591974101960659, "train_loss_llm": 0.0, "grad_norm": 0.037682466208934784, "global_step": 16542, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03319009393453598, "train_loss_bc": 0.03318783640861511, "train_loss_llm": 0.00022590160369873047, "grad_norm": 0.05962390452623367, "global_step": 16543, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03580567240715027, "train_loss_bc": 0.03580567240715027, "train_loss_llm": 0.0, "grad_norm": 0.07156223803758621, "global_step": 16544, "epoch": 459, "lr": 0.001} +{"train_loss": 0.036263901740312576, "train_loss_bc": 0.036263901740312576, "train_loss_llm": 0.0, "grad_norm": 0.02645910158753395, "global_step": 16545, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03731350973248482, "train_loss_bc": 0.03731350973248482, "train_loss_llm": 0.0, "grad_norm": 0.045840293169021606, "global_step": 16546, "epoch": 459, "lr": 0.001} +{"train_loss": 0.030383635312318802, "train_loss_bc": 0.030383635312318802, "train_loss_llm": 0.0, "grad_norm": 0.05588959529995918, "global_step": 16547, "epoch": 459, "lr": 0.001} +{"train_loss": 0.028107719495892525, "train_loss_bc": 0.028107719495892525, "train_loss_llm": 0.0, "grad_norm": 0.06606670469045639, "global_step": 16548, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03504369035363197, "train_loss_bc": 0.03504369035363197, "train_loss_llm": 0.0, "grad_norm": 0.02671905606985092, "global_step": 16549, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03139927238225937, "train_loss_bc": 0.03139927238225937, "train_loss_llm": 0.0, "grad_norm": 0.04142051935195923, "global_step": 16550, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03318610042333603, "train_loss_bc": 0.03318610042333603, "train_loss_llm": 0.0, "grad_norm": 0.05615488439798355, "global_step": 16551, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03638695925474167, "train_loss_bc": 0.03638695925474167, "train_loss_llm": 0.0, "grad_norm": 0.07244295626878738, "global_step": 16552, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03664025291800499, "train_loss_bc": 0.036627497524023056, "train_loss_llm": 0.0012755990028381348, "grad_norm": 0.03466679900884628, "global_step": 16553, "epoch": 459, "lr": 0.001} +{"train_loss": 0.038918301463127136, "train_loss_bc": 0.038918301463127136, "train_loss_llm": 0.0, "grad_norm": 0.04510553181171417, "global_step": 16554, "epoch": 459, "lr": 0.001} +{"train_loss": 0.04445439577102661, "train_loss_bc": 0.04445439577102661, "train_loss_llm": 0.0, "grad_norm": 0.07506134361028671, "global_step": 16555, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03362846374511719, "train_loss_bc": 0.03362846374511719, "train_loss_llm": 0.0, "grad_norm": 0.09241832047700882, "global_step": 16556, "epoch": 459, "lr": 0.001} +{"train_loss": 0.036613233387470245, "train_loss_bc": 0.03660803288221359, "train_loss_llm": 0.0005201399326324463, "grad_norm": 0.029800450429320335, "global_step": 16557, "epoch": 459, "lr": 0.001} +{"train_loss": 0.029215790331363678, "train_loss_bc": 0.029215790331363678, "train_loss_llm": 0.0, "grad_norm": 0.039211615920066833, "global_step": 16558, "epoch": 459, "lr": 0.001} +{"train_loss": 0.03498592310481601, "train_loss_bc": 0.036252211779356, "train_loss_llm": 0.0, "grad_norm": 0.06301851570606232, "global_step": 16559, "epoch": 459, "lr": 0.001, "val_loss": 0.0721442848443985} +{"train_loss": 0.03432561084628105, "train_loss_bc": 0.03432561084628105, "train_loss_llm": 0.0, "grad_norm": 0.08686745166778564, "global_step": 16560, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03962187096476555, "train_loss_bc": 0.03962187096476555, "train_loss_llm": 0.0, "grad_norm": 0.02751285955309868, "global_step": 16561, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03337400034070015, "train_loss_bc": 0.03337400034070015, "train_loss_llm": 0.0, "grad_norm": 0.034137576818466187, "global_step": 16562, "epoch": 460, "lr": 0.001} +{"train_loss": 0.036124296486377716, "train_loss_bc": 0.036124296486377716, "train_loss_llm": 0.0, "grad_norm": 0.05402913689613342, "global_step": 16563, "epoch": 460, "lr": 0.001} +{"train_loss": 0.04274660348892212, "train_loss_bc": 0.04273397475481033, "train_loss_llm": 0.001262962818145752, "grad_norm": 0.05995037034153938, "global_step": 16564, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03255867585539818, "train_loss_bc": 0.03255867585539818, "train_loss_llm": 0.0, "grad_norm": 0.0227363184094429, "global_step": 16565, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03330562636256218, "train_loss_bc": 0.033289484679698944, "train_loss_llm": 0.0016140341758728027, "grad_norm": 0.038323234766721725, "global_step": 16566, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03450200706720352, "train_loss_bc": 0.03447951003909111, "train_loss_llm": 0.0022497475147247314, "grad_norm": 0.047387294471263885, "global_step": 16567, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03459387645125389, "train_loss_bc": 0.03459387645125389, "train_loss_llm": 0.0, "grad_norm": 0.06245366856455803, "global_step": 16568, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03846336528658867, "train_loss_bc": 0.03845694288611412, "train_loss_llm": 0.000642240047454834, "grad_norm": 0.024925597012043, "global_step": 16569, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03832811862230301, "train_loss_bc": 0.03832811862230301, "train_loss_llm": 0.0, "grad_norm": 0.043626897037029266, "global_step": 16570, "epoch": 460, "lr": 0.001} +{"train_loss": 0.0334964282810688, "train_loss_bc": 0.0334964282810688, "train_loss_llm": 0.0, "grad_norm": 0.06695534288883209, "global_step": 16571, "epoch": 460, "lr": 0.001} +{"train_loss": 0.02915642224252224, "train_loss_bc": 0.02915642224252224, "train_loss_llm": 0.0, "grad_norm": 0.08659891039133072, "global_step": 16572, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03447290509939194, "train_loss_bc": 0.03446684405207634, "train_loss_llm": 0.0006062686443328857, "grad_norm": 0.02153244987130165, "global_step": 16573, "epoch": 460, "lr": 0.001} +{"train_loss": 0.030364584177732468, "train_loss_bc": 0.030353764072060585, "train_loss_llm": 0.0010820627212524414, "grad_norm": 0.037788212299346924, "global_step": 16574, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03395579755306244, "train_loss_bc": 0.03395579755306244, "train_loss_llm": 0.0, "grad_norm": 0.05905561149120331, "global_step": 16575, "epoch": 460, "lr": 0.001} +{"train_loss": 0.034042201936244965, "train_loss_bc": 0.034042201936244965, "train_loss_llm": 0.0, "grad_norm": 0.07848930358886719, "global_step": 16576, "epoch": 460, "lr": 0.001} +{"train_loss": 0.040445003658533096, "train_loss_bc": 0.040445003658533096, "train_loss_llm": 0.0, "grad_norm": 0.03458644822239876, "global_step": 16577, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03145106881856918, "train_loss_bc": 0.03145106881856918, "train_loss_llm": 0.0, "grad_norm": 0.04606490582227707, "global_step": 16578, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03205166757106781, "train_loss_bc": 0.03205166757106781, "train_loss_llm": 0.0, "grad_norm": 0.07219020277261734, "global_step": 16579, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03572351858019829, "train_loss_bc": 0.03570224717259407, "train_loss_llm": 0.0021269917488098145, "grad_norm": 0.08594344556331635, "global_step": 16580, "epoch": 460, "lr": 0.001} +{"train_loss": 0.040303878486156464, "train_loss_bc": 0.040303878486156464, "train_loss_llm": 0.0, "grad_norm": 0.026311084628105164, "global_step": 16581, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03935149312019348, "train_loss_bc": 0.03935149312019348, "train_loss_llm": 0.0, "grad_norm": 0.04461505636572838, "global_step": 16582, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03439515456557274, "train_loss_bc": 0.03439515456557274, "train_loss_llm": 0.0, "grad_norm": 0.057881392538547516, "global_step": 16583, "epoch": 460, "lr": 0.001} +{"train_loss": 0.030636316165328026, "train_loss_bc": 0.030636316165328026, "train_loss_llm": 0.0, "grad_norm": 0.06907059252262115, "global_step": 16584, "epoch": 460, "lr": 0.001} +{"train_loss": 0.035256147384643555, "train_loss_bc": 0.035256147384643555, "train_loss_llm": 0.0, "grad_norm": 0.032248977571725845, "global_step": 16585, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03637126460671425, "train_loss_bc": 0.03637126460671425, "train_loss_llm": 0.0, "grad_norm": 0.05056186765432358, "global_step": 16586, "epoch": 460, "lr": 0.001} +{"train_loss": 0.029240205883979797, "train_loss_bc": 0.029201442375779152, "train_loss_llm": 0.0038763582706451416, "grad_norm": 0.056461941450834274, "global_step": 16587, "epoch": 460, "lr": 0.001} +{"train_loss": 0.040169842541217804, "train_loss_bc": 0.040169842541217804, "train_loss_llm": 0.0, "grad_norm": 0.07800929248332977, "global_step": 16588, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03751245513558388, "train_loss_bc": 0.03751245513558388, "train_loss_llm": 0.0, "grad_norm": 0.02786066196858883, "global_step": 16589, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03299441188573837, "train_loss_bc": 0.032991696149110794, "train_loss_llm": 0.00027170777320861816, "grad_norm": 0.04684004932641983, "global_step": 16590, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03332490473985672, "train_loss_bc": 0.03331869840621948, "train_loss_llm": 0.0006206631660461426, "grad_norm": 0.06409875303506851, "global_step": 16591, "epoch": 460, "lr": 0.001} +{"train_loss": 0.029482955113053322, "train_loss_bc": 0.02947167679667473, "train_loss_llm": 0.0011278390884399414, "grad_norm": 0.0780884176492691, "global_step": 16592, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03811813145875931, "train_loss_bc": 0.03811813145875931, "train_loss_llm": 0.0, "grad_norm": 0.03333808854222298, "global_step": 16593, "epoch": 460, "lr": 0.001} +{"train_loss": 0.039469197392463684, "train_loss_bc": 0.039469197392463684, "train_loss_llm": 0.0, "grad_norm": 0.05294843390583992, "global_step": 16594, "epoch": 460, "lr": 0.001} +{"train_loss": 0.03501915926527646, "train_loss_bc": 0.030959725379943848, "train_loss_llm": 0.0, "grad_norm": 0.07918760925531387, "global_step": 16595, "epoch": 460, "lr": 0.001, "train/cumulative_reward": 117.05780159478732, "train/mean_score": 14.746530836278287, "train/success_rate": 0.3, "test/cumulative_reward": 43.63531564636008, "test/mean_score": 8.443613930119675, "test/success_rate": 0.22, "val_loss": 0.07357782125473022, "train_action_mse_error": 0.03776908665895462} +{"train_loss": 0.034714341163635254, "train_loss_bc": 0.034714341163635254, "train_loss_llm": 0.0, "grad_norm": 0.10998793691396713, "global_step": 16596, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03681201487779617, "train_loss_bc": 0.036809127777814865, "train_loss_llm": 0.0002885758876800537, "grad_norm": 0.03130856156349182, "global_step": 16597, "epoch": 461, "lr": 0.001} +{"train_loss": 0.036849938333034515, "train_loss_bc": 0.036849938333034515, "train_loss_llm": 0.0, "grad_norm": 0.0588744655251503, "global_step": 16598, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03885519877076149, "train_loss_bc": 0.03885519877076149, "train_loss_llm": 0.0, "grad_norm": 0.08301272988319397, "global_step": 16599, "epoch": 461, "lr": 0.001} +{"train_loss": 0.031149454414844513, "train_loss_bc": 0.031149454414844513, "train_loss_llm": 0.0, "grad_norm": 0.09960954636335373, "global_step": 16600, "epoch": 461, "lr": 0.001} +{"train_loss": 0.0299881249666214, "train_loss_bc": 0.0299881249666214, "train_loss_llm": 0.0, "grad_norm": 0.02698455937206745, "global_step": 16601, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03763328865170479, "train_loss_bc": 0.03763088583946228, "train_loss_llm": 0.0002403855323791504, "grad_norm": 0.04633602499961853, "global_step": 16602, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03858266770839691, "train_loss_bc": 0.03858266770839691, "train_loss_llm": 0.0, "grad_norm": 0.0652262344956398, "global_step": 16603, "epoch": 461, "lr": 0.001} +{"train_loss": 0.0391288623213768, "train_loss_bc": 0.039113253355026245, "train_loss_llm": 0.0015607178211212158, "grad_norm": 0.07237038016319275, "global_step": 16604, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03814804553985596, "train_loss_bc": 0.03812261298298836, "train_loss_llm": 0.0025430917739868164, "grad_norm": 0.0260581336915493, "global_step": 16605, "epoch": 461, "lr": 0.001} +{"train_loss": 0.034610193222761154, "train_loss_bc": 0.034610193222761154, "train_loss_llm": 0.0, "grad_norm": 0.043950919061899185, "global_step": 16606, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03298791125416756, "train_loss_bc": 0.03298791125416756, "train_loss_llm": 0.0, "grad_norm": 0.05965759605169296, "global_step": 16607, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03267846629023552, "train_loss_bc": 0.03267846629023552, "train_loss_llm": 0.0, "grad_norm": 0.07098914682865143, "global_step": 16608, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03802139312028885, "train_loss_bc": 0.03802139312028885, "train_loss_llm": 0.0, "grad_norm": 0.029162589460611343, "global_step": 16609, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03409770503640175, "train_loss_bc": 0.034083541482686996, "train_loss_llm": 0.0014161765575408936, "grad_norm": 0.05119984596967697, "global_step": 16610, "epoch": 461, "lr": 0.001} +{"train_loss": 0.032973773777484894, "train_loss_bc": 0.032973773777484894, "train_loss_llm": 0.0, "grad_norm": 0.07658211886882782, "global_step": 16611, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03532707691192627, "train_loss_bc": 0.03532707691192627, "train_loss_llm": 0.0, "grad_norm": 0.09197705239057541, "global_step": 16612, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03501329571008682, "train_loss_bc": 0.03500436618924141, "train_loss_llm": 0.0008927881717681885, "grad_norm": 0.02854856662452221, "global_step": 16613, "epoch": 461, "lr": 0.001} +{"train_loss": 0.0351678691804409, "train_loss_bc": 0.0351678691804409, "train_loss_llm": 0.0, "grad_norm": 0.042381010949611664, "global_step": 16614, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03730490431189537, "train_loss_bc": 0.03730490431189537, "train_loss_llm": 0.0, "grad_norm": 0.05534503236413002, "global_step": 16615, "epoch": 461, "lr": 0.001} +{"train_loss": 0.0361701175570488, "train_loss_bc": 0.0361701175570488, "train_loss_llm": 0.0, "grad_norm": 0.07144776731729507, "global_step": 16616, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03485054150223732, "train_loss_bc": 0.03484305739402771, "train_loss_llm": 0.0007483363151550293, "grad_norm": 0.025857599452137947, "global_step": 16617, "epoch": 461, "lr": 0.001} +{"train_loss": 0.035001058131456375, "train_loss_bc": 0.035001058131456375, "train_loss_llm": 0.0, "grad_norm": 0.04917304590344429, "global_step": 16618, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03399938344955444, "train_loss_bc": 0.03399938344955444, "train_loss_llm": 0.0, "grad_norm": 0.06500875949859619, "global_step": 16619, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03918909654021263, "train_loss_bc": 0.03918909654021263, "train_loss_llm": 0.0, "grad_norm": 0.09219740331172943, "global_step": 16620, "epoch": 461, "lr": 0.001} +{"train_loss": 0.033149152994155884, "train_loss_bc": 0.033148184418678284, "train_loss_llm": 9.700655937194824e-05, "grad_norm": 0.026235433295369148, "global_step": 16621, "epoch": 461, "lr": 0.001} +{"train_loss": 0.036372214555740356, "train_loss_bc": 0.036372214555740356, "train_loss_llm": 0.0, "grad_norm": 0.039840586483478546, "global_step": 16622, "epoch": 461, "lr": 0.001} +{"train_loss": 0.032401904463768005, "train_loss_bc": 0.032401904463768005, "train_loss_llm": 0.0, "grad_norm": 0.051248036324977875, "global_step": 16623, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03511229529976845, "train_loss_bc": 0.03511177748441696, "train_loss_llm": 5.179643630981445e-05, "grad_norm": 0.06277735531330109, "global_step": 16624, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03417158126831055, "train_loss_bc": 0.03413891792297363, "train_loss_llm": 0.003266364336013794, "grad_norm": 0.03004629909992218, "global_step": 16625, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03741282597184181, "train_loss_bc": 0.03741282597184181, "train_loss_llm": 0.0, "grad_norm": 0.04281044378876686, "global_step": 16626, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03849048912525177, "train_loss_bc": 0.03849048912525177, "train_loss_llm": 0.0, "grad_norm": 0.06042420491576195, "global_step": 16627, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03284481167793274, "train_loss_bc": 0.03284277394413948, "train_loss_llm": 0.00020393729209899902, "grad_norm": 0.07865852117538452, "global_step": 16628, "epoch": 461, "lr": 0.001} +{"train_loss": 0.027096210047602654, "train_loss_bc": 0.027089722454547882, "train_loss_llm": 0.0006487369537353516, "grad_norm": 0.021985745057463646, "global_step": 16629, "epoch": 461, "lr": 0.001} +{"train_loss": 0.030329860746860504, "train_loss_bc": 0.030329860746860504, "train_loss_llm": 0.0, "grad_norm": 0.037018004804849625, "global_step": 16630, "epoch": 461, "lr": 0.001} +{"train_loss": 0.03520255137441887, "train_loss_bc": 0.04065578058362007, "train_loss_llm": 0.0, "grad_norm": 0.06006650999188423, "global_step": 16631, "epoch": 461, "lr": 0.001, "val_loss": 0.07068240642547607} +{"train_loss": 0.034777335822582245, "train_loss_bc": 0.034777335822582245, "train_loss_llm": 0.0, "grad_norm": 0.0856621265411377, "global_step": 16632, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03608125075697899, "train_loss_bc": 0.036073800176382065, "train_loss_llm": 0.0007452070713043213, "grad_norm": 0.023953385651111603, "global_step": 16633, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03166406229138374, "train_loss_bc": 0.03166406229138374, "train_loss_llm": 0.0, "grad_norm": 0.04283251613378525, "global_step": 16634, "epoch": 462, "lr": 0.001} +{"train_loss": 0.036399781703948975, "train_loss_bc": 0.0363873727619648, "train_loss_llm": 0.0012407898902893066, "grad_norm": 0.05760152265429497, "global_step": 16635, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03079846501350403, "train_loss_bc": 0.03079846501350403, "train_loss_llm": 0.0, "grad_norm": 0.07222450524568558, "global_step": 16636, "epoch": 462, "lr": 0.001} +{"train_loss": 0.038037195801734924, "train_loss_bc": 0.03803078457713127, "train_loss_llm": 0.0006409883499145508, "grad_norm": 0.02553378976881504, "global_step": 16637, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03152325749397278, "train_loss_bc": 0.03152325749397278, "train_loss_llm": 0.0, "grad_norm": 0.03821124881505966, "global_step": 16638, "epoch": 462, "lr": 0.001} +{"train_loss": 0.04050096124410629, "train_loss_bc": 0.04050096124410629, "train_loss_llm": 0.0, "grad_norm": 0.049732938408851624, "global_step": 16639, "epoch": 462, "lr": 0.001} +{"train_loss": 0.025332504883408546, "train_loss_bc": 0.02533084526658058, "train_loss_llm": 0.0001659989356994629, "grad_norm": 0.06120279058814049, "global_step": 16640, "epoch": 462, "lr": 0.001} +{"train_loss": 0.04448361694812775, "train_loss_bc": 0.04448361694812775, "train_loss_llm": 0.0, "grad_norm": 0.04038602486252785, "global_step": 16641, "epoch": 462, "lr": 0.001} +{"train_loss": 0.030584976077079773, "train_loss_bc": 0.030584976077079773, "train_loss_llm": 0.0, "grad_norm": 0.053703851997852325, "global_step": 16642, "epoch": 462, "lr": 0.001} +{"train_loss": 0.036048099398612976, "train_loss_bc": 0.03604079410433769, "train_loss_llm": 0.0007306039333343506, "grad_norm": 0.06697205454111099, "global_step": 16643, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03694857284426689, "train_loss_bc": 0.03694857284426689, "train_loss_llm": 0.0, "grad_norm": 0.0738397017121315, "global_step": 16644, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03618086501955986, "train_loss_bc": 0.03618086501955986, "train_loss_llm": 0.0, "grad_norm": 0.02629210613667965, "global_step": 16645, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03262386843562126, "train_loss_bc": 0.03261247277259827, "train_loss_llm": 0.0011396408081054688, "grad_norm": 0.03944844752550125, "global_step": 16646, "epoch": 462, "lr": 0.001} +{"train_loss": 0.034432049840688705, "train_loss_bc": 0.034432049840688705, "train_loss_llm": 0.0, "grad_norm": 0.047539763152599335, "global_step": 16647, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03266703709959984, "train_loss_bc": 0.03266703709959984, "train_loss_llm": 0.0, "grad_norm": 0.06077498197555542, "global_step": 16648, "epoch": 462, "lr": 0.001} +{"train_loss": 0.029256697744131088, "train_loss_bc": 0.029256697744131088, "train_loss_llm": 0.0, "grad_norm": 0.025985904037952423, "global_step": 16649, "epoch": 462, "lr": 0.001} +{"train_loss": 0.032832372933626175, "train_loss_bc": 0.032832372933626175, "train_loss_llm": 0.0, "grad_norm": 0.03531504422426224, "global_step": 16650, "epoch": 462, "lr": 0.001} +{"train_loss": 0.0265356358140707, "train_loss_bc": 0.0265356358140707, "train_loss_llm": 0.0, "grad_norm": 0.047772228717803955, "global_step": 16651, "epoch": 462, "lr": 0.001} +{"train_loss": 0.034491684287786484, "train_loss_bc": 0.034491684287786484, "train_loss_llm": 0.0, "grad_norm": 0.06676546484231949, "global_step": 16652, "epoch": 462, "lr": 0.001} +{"train_loss": 0.033308256417512894, "train_loss_bc": 0.033308256417512894, "train_loss_llm": 0.0, "grad_norm": 0.026340868324041367, "global_step": 16653, "epoch": 462, "lr": 0.001} +{"train_loss": 0.040414176881313324, "train_loss_bc": 0.040414176881313324, "train_loss_llm": 0.0, "grad_norm": 0.04280303791165352, "global_step": 16654, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03644709289073944, "train_loss_bc": 0.036433324217796326, "train_loss_llm": 0.001376718282699585, "grad_norm": 0.05199596285820007, "global_step": 16655, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03767528012394905, "train_loss_bc": 0.03767528012394905, "train_loss_llm": 0.0, "grad_norm": 0.06715992093086243, "global_step": 16656, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03462442010641098, "train_loss_bc": 0.03462442010641098, "train_loss_llm": 0.0, "grad_norm": 0.03146997094154358, "global_step": 16657, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03448615223169327, "train_loss_bc": 0.03448615223169327, "train_loss_llm": 0.0, "grad_norm": 0.05209483578801155, "global_step": 16658, "epoch": 462, "lr": 0.001} +{"train_loss": 0.036534957587718964, "train_loss_bc": 0.036534957587718964, "train_loss_llm": 0.0, "grad_norm": 0.06919403374195099, "global_step": 16659, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03485061973333359, "train_loss_bc": 0.03485061973333359, "train_loss_llm": 0.0, "grad_norm": 0.09161991626024246, "global_step": 16660, "epoch": 462, "lr": 0.001} +{"train_loss": 0.040271494537591934, "train_loss_bc": 0.040271494537591934, "train_loss_llm": 0.0, "grad_norm": 0.0335535854101181, "global_step": 16661, "epoch": 462, "lr": 0.001} +{"train_loss": 0.036080338060855865, "train_loss_bc": 0.036080338060855865, "train_loss_llm": 0.0, "grad_norm": 0.05318964645266533, "global_step": 16662, "epoch": 462, "lr": 0.001} +{"train_loss": 0.043075453490018845, "train_loss_bc": 0.043075453490018845, "train_loss_llm": 0.0, "grad_norm": 0.07781554013490677, "global_step": 16663, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03799106925725937, "train_loss_bc": 0.03796057775616646, "train_loss_llm": 0.0030489861965179443, "grad_norm": 0.10812235623598099, "global_step": 16664, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03044052980840206, "train_loss_bc": 0.0304377730935812, "train_loss_llm": 0.00027558207511901855, "grad_norm": 0.02552166022360325, "global_step": 16665, "epoch": 462, "lr": 0.001} +{"train_loss": 0.03581826388835907, "train_loss_bc": 0.035814717411994934, "train_loss_llm": 0.00035458803176879883, "grad_norm": 0.049148157238960266, "global_step": 16666, "epoch": 462, "lr": 0.001} +{"train_loss": 0.035110106950418815, "train_loss_bc": 0.039735592901706696, "train_loss_llm": 0.0009861588478088379, "grad_norm": 0.07871095836162567, "global_step": 16667, "epoch": 462, "lr": 0.001, "val_loss": 0.07341309636831284} +{"train_loss": 0.034174755215644836, "train_loss_bc": 0.034174755215644836, "train_loss_llm": 0.0, "grad_norm": 0.10155575722455978, "global_step": 16668, "epoch": 463, "lr": 0.001} +{"train_loss": 0.032794516533613205, "train_loss_bc": 0.03278782591223717, "train_loss_llm": 0.0006690621376037598, "grad_norm": 0.021789003163576126, "global_step": 16669, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03538905084133148, "train_loss_bc": 0.03538905084133148, "train_loss_llm": 0.0, "grad_norm": 0.03970589488744736, "global_step": 16670, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03818145766854286, "train_loss_bc": 0.03818145766854286, "train_loss_llm": 0.0, "grad_norm": 0.0545639842748642, "global_step": 16671, "epoch": 463, "lr": 0.001} +{"train_loss": 0.033267952501773834, "train_loss_bc": 0.03325333446264267, "train_loss_llm": 0.001461714506149292, "grad_norm": 0.06310102343559265, "global_step": 16672, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03505120426416397, "train_loss_bc": 0.03505120426416397, "train_loss_llm": 0.0, "grad_norm": 0.029726436361670494, "global_step": 16673, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03528015315532684, "train_loss_bc": 0.035261120647192, "train_loss_llm": 0.0019034147262573242, "grad_norm": 0.04559871926903725, "global_step": 16674, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03686848282814026, "train_loss_bc": 0.03686848282814026, "train_loss_llm": 0.0, "grad_norm": 0.058327868580818176, "global_step": 16675, "epoch": 463, "lr": 0.001} +{"train_loss": 0.04100585728883743, "train_loss_bc": 0.04100585728883743, "train_loss_llm": 0.0, "grad_norm": 0.07459806650876999, "global_step": 16676, "epoch": 463, "lr": 0.001} +{"train_loss": 0.02967635542154312, "train_loss_bc": 0.02965579926967621, "train_loss_llm": 0.002055644989013672, "grad_norm": 0.030406314879655838, "global_step": 16677, "epoch": 463, "lr": 0.001} +{"train_loss": 0.04253871366381645, "train_loss_bc": 0.04251508787274361, "train_loss_llm": 0.002362459897994995, "grad_norm": 0.054222263395786285, "global_step": 16678, "epoch": 463, "lr": 0.001} +{"train_loss": 0.023803668096661568, "train_loss_bc": 0.023803668096661568, "train_loss_llm": 0.0, "grad_norm": 0.06801243871450424, "global_step": 16679, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03250856697559357, "train_loss_bc": 0.03250376880168915, "train_loss_llm": 0.0004799962043762207, "grad_norm": 0.08194531500339508, "global_step": 16680, "epoch": 463, "lr": 0.001} +{"train_loss": 0.04087826609611511, "train_loss_bc": 0.04085727035999298, "train_loss_llm": 0.002099454402923584, "grad_norm": 0.03367035090923309, "global_step": 16681, "epoch": 463, "lr": 0.001} +{"train_loss": 0.04121370613574982, "train_loss_bc": 0.04120864346623421, "train_loss_llm": 0.0005063712596893311, "grad_norm": 0.0576607808470726, "global_step": 16682, "epoch": 463, "lr": 0.001} +{"train_loss": 0.039815228432416916, "train_loss_bc": 0.039815228432416916, "train_loss_llm": 0.0, "grad_norm": 0.08012954145669937, "global_step": 16683, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03836798295378685, "train_loss_bc": 0.038364604115486145, "train_loss_llm": 0.00033795833587646484, "grad_norm": 0.10482577979564667, "global_step": 16684, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03465092182159424, "train_loss_bc": 0.03465092182159424, "train_loss_llm": 0.0, "grad_norm": 0.03258639946579933, "global_step": 16685, "epoch": 463, "lr": 0.001} +{"train_loss": 0.038260918110609055, "train_loss_bc": 0.03824835643172264, "train_loss_llm": 0.0012561678886413574, "grad_norm": 0.04739543795585632, "global_step": 16686, "epoch": 463, "lr": 0.001} +{"train_loss": 0.029991500079631805, "train_loss_bc": 0.02998918481171131, "train_loss_llm": 0.00023156404495239258, "grad_norm": 0.062429215759038925, "global_step": 16687, "epoch": 463, "lr": 0.001} +{"train_loss": 0.037677425891160965, "train_loss_bc": 0.037677425891160965, "train_loss_llm": 0.0, "grad_norm": 0.08244115114212036, "global_step": 16688, "epoch": 463, "lr": 0.001} +{"train_loss": 0.0332140251994133, "train_loss_bc": 0.033210285007953644, "train_loss_llm": 0.00037410855293273926, "grad_norm": 0.021920066326856613, "global_step": 16689, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03545893728733063, "train_loss_bc": 0.035433579236269, "train_loss_llm": 0.0025356709957122803, "grad_norm": 0.03701500594615936, "global_step": 16690, "epoch": 463, "lr": 0.001} +{"train_loss": 0.028453528881072998, "train_loss_bc": 0.028453528881072998, "train_loss_llm": 0.0, "grad_norm": 0.0426420234143734, "global_step": 16691, "epoch": 463, "lr": 0.001} +{"train_loss": 0.038014039397239685, "train_loss_bc": 0.03799102082848549, "train_loss_llm": 0.002302020788192749, "grad_norm": 0.04620061069726944, "global_step": 16692, "epoch": 463, "lr": 0.001} +{"train_loss": 0.030667997896671295, "train_loss_bc": 0.030667997896671295, "train_loss_llm": 0.0, "grad_norm": 0.020713945850729942, "global_step": 16693, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03430813178420067, "train_loss_bc": 0.03430813178420067, "train_loss_llm": 0.0, "grad_norm": 0.031951211392879486, "global_step": 16694, "epoch": 463, "lr": 0.001} +{"train_loss": 0.04208618029952049, "train_loss_bc": 0.042080819606781006, "train_loss_llm": 0.0005359351634979248, "grad_norm": 0.05149002745747566, "global_step": 16695, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03828028589487076, "train_loss_bc": 0.03828028589487076, "train_loss_llm": 0.0, "grad_norm": 0.060384392738342285, "global_step": 16696, "epoch": 463, "lr": 0.001} +{"train_loss": 0.036985769867897034, "train_loss_bc": 0.036985769867897034, "train_loss_llm": 0.0, "grad_norm": 0.02993454411625862, "global_step": 16697, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03309745714068413, "train_loss_bc": 0.03309745714068413, "train_loss_llm": 0.0, "grad_norm": 0.05538496375083923, "global_step": 16698, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03616189584136009, "train_loss_bc": 0.0361597053706646, "train_loss_llm": 0.00021913647651672363, "grad_norm": 0.06885603815317154, "global_step": 16699, "epoch": 463, "lr": 0.001} +{"train_loss": 0.030258720740675926, "train_loss_bc": 0.030258720740675926, "train_loss_llm": 0.0, "grad_norm": 0.08438907563686371, "global_step": 16700, "epoch": 463, "lr": 0.001} +{"train_loss": 0.028099367395043373, "train_loss_bc": 0.028099367395043373, "train_loss_llm": 0.0, "grad_norm": 0.03294572979211807, "global_step": 16701, "epoch": 463, "lr": 0.001} +{"train_loss": 0.03386741131544113, "train_loss_bc": 0.03386741131544113, "train_loss_llm": 0.0, "grad_norm": 0.04711175337433815, "global_step": 16702, "epoch": 463, "lr": 0.001} +{"train_loss": 0.034956614228172436, "train_loss_bc": 0.02808767929673195, "train_loss_llm": 0.0, "grad_norm": 0.06591133028268814, "global_step": 16703, "epoch": 463, "lr": 0.001, "val_loss": 0.07278473675251007} +{"train_loss": 0.030528130009770393, "train_loss_bc": 0.030528130009770393, "train_loss_llm": 0.0, "grad_norm": 0.08917680382728577, "global_step": 16704, "epoch": 464, "lr": 0.001} +{"train_loss": 0.030094023793935776, "train_loss_bc": 0.030094023793935776, "train_loss_llm": 0.0, "grad_norm": 0.02641129493713379, "global_step": 16705, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03286060690879822, "train_loss_bc": 0.03286060690879822, "train_loss_llm": 0.0, "grad_norm": 0.03807138651609421, "global_step": 16706, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03633889555931091, "train_loss_bc": 0.03633889555931091, "train_loss_llm": 0.0, "grad_norm": 0.05012949928641319, "global_step": 16707, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03404923900961876, "train_loss_bc": 0.03404923900961876, "train_loss_llm": 0.0, "grad_norm": 0.06958205252885818, "global_step": 16708, "epoch": 464, "lr": 0.001} +{"train_loss": 0.0368318147957325, "train_loss_bc": 0.0368318147957325, "train_loss_llm": 0.0, "grad_norm": 0.027248768135905266, "global_step": 16709, "epoch": 464, "lr": 0.001} +{"train_loss": 0.037524159997701645, "train_loss_bc": 0.037524159997701645, "train_loss_llm": 0.0, "grad_norm": 0.03834995999932289, "global_step": 16710, "epoch": 464, "lr": 0.001} +{"train_loss": 0.038404904305934906, "train_loss_bc": 0.038404904305934906, "train_loss_llm": 0.0, "grad_norm": 0.05251115933060646, "global_step": 16711, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03798485919833183, "train_loss_bc": 0.03798485919833183, "train_loss_llm": 0.0, "grad_norm": 0.06854754686355591, "global_step": 16712, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03063330054283142, "train_loss_bc": 0.03063330054283142, "train_loss_llm": 0.0, "grad_norm": 0.022871777415275574, "global_step": 16713, "epoch": 464, "lr": 0.001} +{"train_loss": 0.034570835530757904, "train_loss_bc": 0.03456645831465721, "train_loss_llm": 0.00043761730194091797, "grad_norm": 0.031180385500192642, "global_step": 16714, "epoch": 464, "lr": 0.001} +{"train_loss": 0.039886508136987686, "train_loss_bc": 0.039886508136987686, "train_loss_llm": 0.0, "grad_norm": 0.04958868771791458, "global_step": 16715, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03152593597769737, "train_loss_bc": 0.03152593597769737, "train_loss_llm": 0.0, "grad_norm": 0.059515487402677536, "global_step": 16716, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03516170755028725, "train_loss_bc": 0.03516170755028725, "train_loss_llm": 0.0, "grad_norm": 0.02825002931058407, "global_step": 16717, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03643452376127243, "train_loss_bc": 0.03643452376127243, "train_loss_llm": 0.0, "grad_norm": 0.04507024213671684, "global_step": 16718, "epoch": 464, "lr": 0.001} +{"train_loss": 0.039247311651706696, "train_loss_bc": 0.039247311651706696, "train_loss_llm": 0.0, "grad_norm": 0.06579115241765976, "global_step": 16719, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03621547669172287, "train_loss_bc": 0.03621116280555725, "train_loss_llm": 0.00043135881423950195, "grad_norm": 0.07623910158872604, "global_step": 16720, "epoch": 464, "lr": 0.001} +{"train_loss": 0.037494439631700516, "train_loss_bc": 0.037494439631700516, "train_loss_llm": 0.0, "grad_norm": 0.03817043453454971, "global_step": 16721, "epoch": 464, "lr": 0.001} +{"train_loss": 0.036125387996435165, "train_loss_bc": 0.036121394485235214, "train_loss_llm": 0.0003993213176727295, "grad_norm": 0.04996445029973984, "global_step": 16722, "epoch": 464, "lr": 0.001} +{"train_loss": 0.02913774736225605, "train_loss_bc": 0.02913774736225605, "train_loss_llm": 0.0, "grad_norm": 0.06555458158254623, "global_step": 16723, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03667952120304108, "train_loss_bc": 0.036666389554739, "train_loss_llm": 0.0013131499290466309, "grad_norm": 0.07315012812614441, "global_step": 16724, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03602568432688713, "train_loss_bc": 0.03602568432688713, "train_loss_llm": 0.0, "grad_norm": 0.03092147223651409, "global_step": 16725, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03592626377940178, "train_loss_bc": 0.03592626377940178, "train_loss_llm": 0.0, "grad_norm": 0.03866524621844292, "global_step": 16726, "epoch": 464, "lr": 0.001} +{"train_loss": 0.032015856355428696, "train_loss_bc": 0.032015856355428696, "train_loss_llm": 0.0, "grad_norm": 0.05756679177284241, "global_step": 16727, "epoch": 464, "lr": 0.001} +{"train_loss": 0.030440274626016617, "train_loss_bc": 0.030440274626016617, "train_loss_llm": 0.0, "grad_norm": 0.07539638131856918, "global_step": 16728, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03559325635433197, "train_loss_bc": 0.03558247536420822, "train_loss_llm": 0.001078099012374878, "grad_norm": 0.026144348084926605, "global_step": 16729, "epoch": 464, "lr": 0.001} +{"train_loss": 0.039652660489082336, "train_loss_bc": 0.039652660489082336, "train_loss_llm": 0.0, "grad_norm": 0.04585893079638481, "global_step": 16730, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03958006948232651, "train_loss_bc": 0.03958006948232651, "train_loss_llm": 0.0, "grad_norm": 0.056374672800302505, "global_step": 16731, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03557930141687393, "train_loss_bc": 0.03556642308831215, "train_loss_llm": 0.0012877881526947021, "grad_norm": 0.07573322206735611, "global_step": 16732, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03864165022969246, "train_loss_bc": 0.03863469883799553, "train_loss_llm": 0.0006950795650482178, "grad_norm": 0.027943415567278862, "global_step": 16733, "epoch": 464, "lr": 0.001} +{"train_loss": 0.032716765999794006, "train_loss_bc": 0.032716765999794006, "train_loss_llm": 0.0, "grad_norm": 0.03823905438184738, "global_step": 16734, "epoch": 464, "lr": 0.001} +{"train_loss": 0.028584882616996765, "train_loss_bc": 0.028584882616996765, "train_loss_llm": 0.0, "grad_norm": 0.04642896354198456, "global_step": 16735, "epoch": 464, "lr": 0.001} +{"train_loss": 0.030955670401453972, "train_loss_bc": 0.030955670401453972, "train_loss_llm": 0.0, "grad_norm": 0.0518597848713398, "global_step": 16736, "epoch": 464, "lr": 0.001} +{"train_loss": 0.02806496061384678, "train_loss_bc": 0.02806190401315689, "train_loss_llm": 0.00030562281608581543, "grad_norm": 0.026031071320176125, "global_step": 16737, "epoch": 464, "lr": 0.001} +{"train_loss": 0.037421684712171555, "train_loss_bc": 0.037421684712171555, "train_loss_llm": 0.0, "grad_norm": 0.04211703687906265, "global_step": 16738, "epoch": 464, "lr": 0.001} +{"train_loss": 0.03468122747209337, "train_loss_bc": 0.02959587797522545, "train_loss_llm": 0.0, "grad_norm": 0.05161580443382263, "global_step": 16739, "epoch": 464, "lr": 0.001, "val_loss": 0.07172611355781555} +{"train_loss": 0.03130002319812775, "train_loss_bc": 0.03130002319812775, "train_loss_llm": 0.0, "grad_norm": 0.06049671396613121, "global_step": 16740, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03505953401327133, "train_loss_bc": 0.03505953401327133, "train_loss_llm": 0.0, "grad_norm": 0.0250377357006073, "global_step": 16741, "epoch": 465, "lr": 0.001} +{"train_loss": 0.04260951280593872, "train_loss_bc": 0.04260951280593872, "train_loss_llm": 0.0, "grad_norm": 0.04765242710709572, "global_step": 16742, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03145034983754158, "train_loss_bc": 0.03145034983754158, "train_loss_llm": 0.0, "grad_norm": 0.06414791941642761, "global_step": 16743, "epoch": 465, "lr": 0.001} +{"train_loss": 0.027988828718662262, "train_loss_bc": 0.027984431013464928, "train_loss_llm": 0.00043973326683044434, "grad_norm": 0.07978972792625427, "global_step": 16744, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03131483495235443, "train_loss_bc": 0.03131433203816414, "train_loss_llm": 5.036592483520508e-05, "grad_norm": 0.03558514267206192, "global_step": 16745, "epoch": 465, "lr": 0.001} +{"train_loss": 0.034149572253227234, "train_loss_bc": 0.034149572253227234, "train_loss_llm": 0.0, "grad_norm": 0.05292211100459099, "global_step": 16746, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03973998874425888, "train_loss_bc": 0.03973998874425888, "train_loss_llm": 0.0, "grad_norm": 0.0689908042550087, "global_step": 16747, "epoch": 465, "lr": 0.001} +{"train_loss": 0.0332297645509243, "train_loss_bc": 0.0332297645509243, "train_loss_llm": 0.0, "grad_norm": 0.09947405010461807, "global_step": 16748, "epoch": 465, "lr": 0.001} +{"train_loss": 0.036218829452991486, "train_loss_bc": 0.036218829452991486, "train_loss_llm": 0.0, "grad_norm": 0.035317562520504, "global_step": 16749, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03483272343873978, "train_loss_bc": 0.03483272343873978, "train_loss_llm": 0.0, "grad_norm": 0.049420904368162155, "global_step": 16750, "epoch": 465, "lr": 0.001} +{"train_loss": 0.035069745033979416, "train_loss_bc": 0.035069745033979416, "train_loss_llm": 0.0, "grad_norm": 0.05667275935411453, "global_step": 16751, "epoch": 465, "lr": 0.001} +{"train_loss": 0.037776265293359756, "train_loss_bc": 0.037776265293359756, "train_loss_llm": 0.0, "grad_norm": 0.07159850001335144, "global_step": 16752, "epoch": 465, "lr": 0.001} +{"train_loss": 0.036591093987226486, "train_loss_bc": 0.036591093987226486, "train_loss_llm": 0.0, "grad_norm": 0.023520153015851974, "global_step": 16753, "epoch": 465, "lr": 0.001} +{"train_loss": 0.034238286316394806, "train_loss_bc": 0.034238286316394806, "train_loss_llm": 0.0, "grad_norm": 0.03883369639515877, "global_step": 16754, "epoch": 465, "lr": 0.001} +{"train_loss": 0.037207748740911484, "train_loss_bc": 0.037207748740911484, "train_loss_llm": 0.0, "grad_norm": 0.049805253744125366, "global_step": 16755, "epoch": 465, "lr": 0.001} +{"train_loss": 0.039279162883758545, "train_loss_bc": 0.039278291165828705, "train_loss_llm": 8.729100227355957e-05, "grad_norm": 0.059324536472558975, "global_step": 16756, "epoch": 465, "lr": 0.001} +{"train_loss": 0.0369432158768177, "train_loss_bc": 0.0369432158768177, "train_loss_llm": 0.0, "grad_norm": 0.028520332649350166, "global_step": 16757, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03151492029428482, "train_loss_bc": 0.03151492029428482, "train_loss_llm": 0.0, "grad_norm": 0.04150029644370079, "global_step": 16758, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03808382526040077, "train_loss_bc": 0.03808382526040077, "train_loss_llm": 0.0, "grad_norm": 0.04983526095747948, "global_step": 16759, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03293176740407944, "train_loss_bc": 0.03293176740407944, "train_loss_llm": 0.0, "grad_norm": 0.06217699125409126, "global_step": 16760, "epoch": 465, "lr": 0.001} +{"train_loss": 0.027995452284812927, "train_loss_bc": 0.027995452284812927, "train_loss_llm": 0.0, "grad_norm": 0.024629121646285057, "global_step": 16761, "epoch": 465, "lr": 0.001} +{"train_loss": 0.043236441910266876, "train_loss_bc": 0.043236441910266876, "train_loss_llm": 0.0, "grad_norm": 0.043596651405096054, "global_step": 16762, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03590124100446701, "train_loss_bc": 0.03590124100446701, "train_loss_llm": 0.0, "grad_norm": 0.050545912235975266, "global_step": 16763, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03173360228538513, "train_loss_bc": 0.03173360228538513, "train_loss_llm": 0.0, "grad_norm": 0.06629715859889984, "global_step": 16764, "epoch": 465, "lr": 0.001} +{"train_loss": 0.02964133769273758, "train_loss_bc": 0.029636289924383163, "train_loss_llm": 0.0005047917366027832, "grad_norm": 0.027991795912384987, "global_step": 16765, "epoch": 465, "lr": 0.001} +{"train_loss": 0.028956934809684753, "train_loss_bc": 0.028956934809684753, "train_loss_llm": 0.0, "grad_norm": 0.04574688896536827, "global_step": 16766, "epoch": 465, "lr": 0.001} +{"train_loss": 0.032358959317207336, "train_loss_bc": 0.032358959317207336, "train_loss_llm": 0.0, "grad_norm": 0.06263075023889542, "global_step": 16767, "epoch": 465, "lr": 0.001} +{"train_loss": 0.029759133234620094, "train_loss_bc": 0.02975773625075817, "train_loss_llm": 0.00013968348503112793, "grad_norm": 0.07228145748376846, "global_step": 16768, "epoch": 465, "lr": 0.001} +{"train_loss": 0.034838687628507614, "train_loss_bc": 0.034838687628507614, "train_loss_llm": 0.0, "grad_norm": 0.027423648163676262, "global_step": 16769, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03657417744398117, "train_loss_bc": 0.03655315190553665, "train_loss_llm": 0.002102404832839966, "grad_norm": 0.04024304822087288, "global_step": 16770, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03625442087650299, "train_loss_bc": 0.036238640546798706, "train_loss_llm": 0.0015778839588165283, "grad_norm": 0.058793190866708755, "global_step": 16771, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03729129582643509, "train_loss_bc": 0.03729129582643509, "train_loss_llm": 0.0, "grad_norm": 0.07449910044670105, "global_step": 16772, "epoch": 465, "lr": 0.001} +{"train_loss": 0.038202762603759766, "train_loss_bc": 0.03816922754049301, "train_loss_llm": 0.003353416919708252, "grad_norm": 0.02986079454421997, "global_step": 16773, "epoch": 465, "lr": 0.001} +{"train_loss": 0.03369602560997009, "train_loss_bc": 0.03368828445672989, "train_loss_llm": 0.0007742345333099365, "grad_norm": 0.04086426645517349, "global_step": 16774, "epoch": 465, "lr": 0.001} +{"train_loss": 0.034757019796719156, "train_loss_bc": 0.037282247096300125, "train_loss_llm": 0.0, "grad_norm": 0.05490808188915253, "global_step": 16775, "epoch": 465, "lr": 0.001, "train/cumulative_reward": 88.42561559208, "train/mean_score": 17.836591736271192, "train/success_rate": 0.5, "test/cumulative_reward": 41.93832067827651, "test/mean_score": 9.21251687417378, "test/success_rate": 0.24, "val_loss": 0.07263665646314621, "train_action_mse_error": 0.03720483556389809} +{"train_loss": 0.027806637808680534, "train_loss_bc": 0.027802016586065292, "train_loss_llm": 0.0004621446132659912, "grad_norm": 0.06193862855434418, "global_step": 16776, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03142135590314865, "train_loss_bc": 0.03140755370259285, "train_loss_llm": 0.00138014554977417, "grad_norm": 0.020714566111564636, "global_step": 16777, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03734736517071724, "train_loss_bc": 0.03734736517071724, "train_loss_llm": 0.0, "grad_norm": 0.03182453662157059, "global_step": 16778, "epoch": 466, "lr": 0.001} +{"train_loss": 0.029721051454544067, "train_loss_bc": 0.029721051454544067, "train_loss_llm": 0.0, "grad_norm": 0.03631514310836792, "global_step": 16779, "epoch": 466, "lr": 0.001} +{"train_loss": 0.031010625883936882, "train_loss_bc": 0.031003449112176895, "train_loss_llm": 0.0007176101207733154, "grad_norm": 0.04672766104340553, "global_step": 16780, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03533507511019707, "train_loss_bc": 0.03533507511019707, "train_loss_llm": 0.0, "grad_norm": 0.03338227421045303, "global_step": 16781, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03090803697705269, "train_loss_bc": 0.03090803697705269, "train_loss_llm": 0.0, "grad_norm": 0.04886504262685776, "global_step": 16782, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03551285341382027, "train_loss_bc": 0.03551285341382027, "train_loss_llm": 0.0, "grad_norm": 0.06122004613280296, "global_step": 16783, "epoch": 466, "lr": 0.001} +{"train_loss": 0.033329740166664124, "train_loss_bc": 0.033329740166664124, "train_loss_llm": 0.0, "grad_norm": 0.07397893071174622, "global_step": 16784, "epoch": 466, "lr": 0.001} +{"train_loss": 0.028677083551883698, "train_loss_bc": 0.028677083551883698, "train_loss_llm": 0.0, "grad_norm": 0.022114597260951996, "global_step": 16785, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03189514949917793, "train_loss_bc": 0.03189514949917793, "train_loss_llm": 0.0, "grad_norm": 0.0344691164791584, "global_step": 16786, "epoch": 466, "lr": 0.001} +{"train_loss": 0.02678147330880165, "train_loss_bc": 0.02678147330880165, "train_loss_llm": 0.0, "grad_norm": 0.04766441881656647, "global_step": 16787, "epoch": 466, "lr": 0.001} +{"train_loss": 0.04174795001745224, "train_loss_bc": 0.04174400493502617, "train_loss_llm": 0.000394672155380249, "grad_norm": 0.05651148781180382, "global_step": 16788, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03823032230138779, "train_loss_bc": 0.03823032230138779, "train_loss_llm": 0.0, "grad_norm": 0.03191619738936424, "global_step": 16789, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03513430431485176, "train_loss_bc": 0.03513430431485176, "train_loss_llm": 0.0, "grad_norm": 0.043800510466098785, "global_step": 16790, "epoch": 466, "lr": 0.001} +{"train_loss": 0.04048217460513115, "train_loss_bc": 0.04048217460513115, "train_loss_llm": 0.0, "grad_norm": 0.06348764896392822, "global_step": 16791, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03683630749583244, "train_loss_bc": 0.03683630749583244, "train_loss_llm": 0.0, "grad_norm": 0.07280073314905167, "global_step": 16792, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03623666986823082, "train_loss_bc": 0.03623666986823082, "train_loss_llm": 0.0, "grad_norm": 0.03141021355986595, "global_step": 16793, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03229692578315735, "train_loss_bc": 0.03229692578315735, "train_loss_llm": 0.0, "grad_norm": 0.03597163036465645, "global_step": 16794, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03964973986148834, "train_loss_bc": 0.03964744508266449, "train_loss_llm": 0.00022941827774047852, "grad_norm": 0.05230932682752609, "global_step": 16795, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03209955617785454, "train_loss_bc": 0.03209955617785454, "train_loss_llm": 0.0, "grad_norm": 0.0721585601568222, "global_step": 16796, "epoch": 466, "lr": 0.001} +{"train_loss": 0.04556453600525856, "train_loss_bc": 0.04556453600525856, "train_loss_llm": 0.0, "grad_norm": 0.03449052572250366, "global_step": 16797, "epoch": 466, "lr": 0.001} +{"train_loss": 0.029020803049206734, "train_loss_bc": 0.029020803049206734, "train_loss_llm": 0.0, "grad_norm": 0.04794768989086151, "global_step": 16798, "epoch": 466, "lr": 0.001} +{"train_loss": 0.02916613221168518, "train_loss_bc": 0.02916613221168518, "train_loss_llm": 0.0, "grad_norm": 0.05842268094420433, "global_step": 16799, "epoch": 466, "lr": 0.001} +{"train_loss": 0.031207216903567314, "train_loss_bc": 0.031207216903567314, "train_loss_llm": 0.0, "grad_norm": 0.059673670679330826, "global_step": 16800, "epoch": 466, "lr": 0.001} +{"train_loss": 0.04362354055047035, "train_loss_bc": 0.04362354055047035, "train_loss_llm": 0.0, "grad_norm": 0.030248282477259636, "global_step": 16801, "epoch": 466, "lr": 0.001} +{"train_loss": 0.030867669731378555, "train_loss_bc": 0.030867669731378555, "train_loss_llm": 0.0, "grad_norm": 0.04172886535525322, "global_step": 16802, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03933776542544365, "train_loss_bc": 0.03933776542544365, "train_loss_llm": 0.0, "grad_norm": 0.053449347615242004, "global_step": 16803, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03952435031533241, "train_loss_bc": 0.03952435031533241, "train_loss_llm": 0.0, "grad_norm": 0.07472478598356247, "global_step": 16804, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03523919731378555, "train_loss_bc": 0.03523234650492668, "train_loss_llm": 0.0006852447986602783, "grad_norm": 0.027640903368592262, "global_step": 16805, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03659156709909439, "train_loss_bc": 0.03659156709909439, "train_loss_llm": 0.0, "grad_norm": 0.04488299414515495, "global_step": 16806, "epoch": 466, "lr": 0.001} +{"train_loss": 0.030099378898739815, "train_loss_bc": 0.030099378898739815, "train_loss_llm": 0.0, "grad_norm": 0.06616993248462677, "global_step": 16807, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03742184862494469, "train_loss_bc": 0.03739216551184654, "train_loss_llm": 0.0029682815074920654, "grad_norm": 0.08532437682151794, "global_step": 16808, "epoch": 466, "lr": 0.001} +{"train_loss": 0.039943892508745193, "train_loss_bc": 0.039943892508745193, "train_loss_llm": 0.0, "grad_norm": 0.02699085883796215, "global_step": 16809, "epoch": 466, "lr": 0.001} +{"train_loss": 0.039599910378456116, "train_loss_bc": 0.03957601264119148, "train_loss_llm": 0.0023898184299468994, "grad_norm": 0.03703318163752556, "global_step": 16810, "epoch": 466, "lr": 0.001} +{"train_loss": 0.03452265883485476, "train_loss_bc": 0.023143714293837547, "train_loss_llm": 0.0003795921802520752, "grad_norm": 0.0561930388212204, "global_step": 16811, "epoch": 466, "lr": 0.001, "val_loss": 0.07225322723388672} +{"train_loss": 0.03959428519010544, "train_loss_bc": 0.03959428519010544, "train_loss_llm": 0.0, "grad_norm": 0.0779508426785469, "global_step": 16812, "epoch": 467, "lr": 0.001} +{"train_loss": 0.031581152230501175, "train_loss_bc": 0.031581152230501175, "train_loss_llm": 0.0, "grad_norm": 0.021519053727388382, "global_step": 16813, "epoch": 467, "lr": 0.001} +{"train_loss": 0.035923704504966736, "train_loss_bc": 0.035923704504966736, "train_loss_llm": 0.0, "grad_norm": 0.033891335129737854, "global_step": 16814, "epoch": 467, "lr": 0.001} +{"train_loss": 0.030567413195967674, "train_loss_bc": 0.030551347881555557, "train_loss_llm": 0.0016064941883087158, "grad_norm": 0.04493725672364235, "global_step": 16815, "epoch": 467, "lr": 0.001} +{"train_loss": 0.037138063460588455, "train_loss_bc": 0.03711847960948944, "train_loss_llm": 0.001958221197128296, "grad_norm": 0.05371478945016861, "global_step": 16816, "epoch": 467, "lr": 0.001} +{"train_loss": 0.0333993136882782, "train_loss_bc": 0.0333993136882782, "train_loss_llm": 0.0, "grad_norm": 0.02058645710349083, "global_step": 16817, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03304539620876312, "train_loss_bc": 0.03304539620876312, "train_loss_llm": 0.0, "grad_norm": 0.03317224234342575, "global_step": 16818, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03787537291646004, "train_loss_bc": 0.03787537291646004, "train_loss_llm": 0.0, "grad_norm": 0.0502653568983078, "global_step": 16819, "epoch": 467, "lr": 0.001} +{"train_loss": 0.04331180080771446, "train_loss_bc": 0.04331180080771446, "train_loss_llm": 0.0, "grad_norm": 0.05932362377643585, "global_step": 16820, "epoch": 467, "lr": 0.001} +{"train_loss": 0.0323735810816288, "train_loss_bc": 0.03237348422408104, "train_loss_llm": 9.715557098388672e-06, "grad_norm": 0.024584805592894554, "global_step": 16821, "epoch": 467, "lr": 0.001} +{"train_loss": 0.04322882741689682, "train_loss_bc": 0.04322882741689682, "train_loss_llm": 0.0, "grad_norm": 0.037123534828424454, "global_step": 16822, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03358372300863266, "train_loss_bc": 0.03358372300863266, "train_loss_llm": 0.0, "grad_norm": 0.05573568865656853, "global_step": 16823, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03156985342502594, "train_loss_bc": 0.031564537435770035, "train_loss_llm": 0.0005317032337188721, "grad_norm": 0.0683877021074295, "global_step": 16824, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03306508809328079, "train_loss_bc": 0.0330631323158741, "train_loss_llm": 0.00019571185111999512, "grad_norm": 0.026786677539348602, "global_step": 16825, "epoch": 467, "lr": 0.001} +{"train_loss": 0.040582407265901566, "train_loss_bc": 0.040582407265901566, "train_loss_llm": 0.0, "grad_norm": 0.044041674584150314, "global_step": 16826, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03633129224181175, "train_loss_bc": 0.036328546702861786, "train_loss_llm": 0.00027441978454589844, "grad_norm": 0.0553586371243, "global_step": 16827, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03647914528846741, "train_loss_bc": 0.03647784888744354, "train_loss_llm": 0.0001296699047088623, "grad_norm": 0.0738140344619751, "global_step": 16828, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03161521628499031, "train_loss_bc": 0.03161521628499031, "train_loss_llm": 0.0, "grad_norm": 0.028376305475831032, "global_step": 16829, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03556089475750923, "train_loss_bc": 0.03556089475750923, "train_loss_llm": 0.0, "grad_norm": 0.04417303204536438, "global_step": 16830, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03868440166115761, "train_loss_bc": 0.03868440166115761, "train_loss_llm": 0.0, "grad_norm": 0.04982156306505203, "global_step": 16831, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03359624743461609, "train_loss_bc": 0.03359541296958923, "train_loss_llm": 8.335709571838379e-05, "grad_norm": 0.06523346155881882, "global_step": 16832, "epoch": 467, "lr": 0.001} +{"train_loss": 0.034285105764865875, "train_loss_bc": 0.034285105764865875, "train_loss_llm": 0.0, "grad_norm": 0.02815212309360504, "global_step": 16833, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03928700461983681, "train_loss_bc": 0.03926151618361473, "train_loss_llm": 0.0025488734245300293, "grad_norm": 0.044579099863767624, "global_step": 16834, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03190070390701294, "train_loss_bc": 0.03190070390701294, "train_loss_llm": 0.0, "grad_norm": 0.06261347234249115, "global_step": 16835, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03590500354766846, "train_loss_bc": 0.03590469807386398, "train_loss_llm": 3.0547380447387695e-05, "grad_norm": 0.07297453284263611, "global_step": 16836, "epoch": 467, "lr": 0.001} +{"train_loss": 0.032429978251457214, "train_loss_bc": 0.032422322779893875, "train_loss_llm": 0.000765383243560791, "grad_norm": 0.025082619860768318, "global_step": 16837, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03565865382552147, "train_loss_bc": 0.03565865382552147, "train_loss_llm": 0.0, "grad_norm": 0.037694044411182404, "global_step": 16838, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03177380934357643, "train_loss_bc": 0.03176983818411827, "train_loss_llm": 0.0003972351551055908, "grad_norm": 0.054631732404232025, "global_step": 16839, "epoch": 467, "lr": 0.001} +{"train_loss": 0.029232991859316826, "train_loss_bc": 0.029223959892988205, "train_loss_llm": 0.0009032785892486572, "grad_norm": 0.06370635330677032, "global_step": 16840, "epoch": 467, "lr": 0.001} +{"train_loss": 0.033152397722005844, "train_loss_bc": 0.033152397722005844, "train_loss_llm": 0.0, "grad_norm": 0.02344331331551075, "global_step": 16841, "epoch": 467, "lr": 0.001} +{"train_loss": 0.025969820097088814, "train_loss_bc": 0.025969820097088814, "train_loss_llm": 0.0, "grad_norm": 0.042361561208963394, "global_step": 16842, "epoch": 467, "lr": 0.001} +{"train_loss": 0.036032505333423615, "train_loss_bc": 0.036032505333423615, "train_loss_llm": 0.0, "grad_norm": 0.05183124169707298, "global_step": 16843, "epoch": 467, "lr": 0.001} +{"train_loss": 0.022184332832694054, "train_loss_bc": 0.022184332832694054, "train_loss_llm": 0.0, "grad_norm": 0.06095745041966438, "global_step": 16844, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03252792730927467, "train_loss_bc": 0.03252792730927467, "train_loss_llm": 0.0, "grad_norm": 0.025264276191592216, "global_step": 16845, "epoch": 467, "lr": 0.001} +{"train_loss": 0.04196975380182266, "train_loss_bc": 0.04196931794285774, "train_loss_llm": 4.342198371887207e-05, "grad_norm": 0.044838663190603256, "global_step": 16846, "epoch": 467, "lr": 0.001} +{"train_loss": 0.03472918478979005, "train_loss_bc": 0.038833484053611755, "train_loss_llm": 0.0, "grad_norm": 0.0572676956653595, "global_step": 16847, "epoch": 467, "lr": 0.001, "val_loss": 0.07242253422737122} +{"train_loss": 0.036361489444971085, "train_loss_bc": 0.036361489444971085, "train_loss_llm": 0.0, "grad_norm": 0.07335220277309418, "global_step": 16848, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03845583274960518, "train_loss_bc": 0.038455087691545486, "train_loss_llm": 7.462501525878906e-05, "grad_norm": 0.03131348639726639, "global_step": 16849, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03499780595302582, "train_loss_bc": 0.03499780595302582, "train_loss_llm": 0.0, "grad_norm": 0.04261638969182968, "global_step": 16850, "epoch": 468, "lr": 0.001} +{"train_loss": 0.030988821759819984, "train_loss_bc": 0.030974838882684708, "train_loss_llm": 0.001398324966430664, "grad_norm": 0.05394278094172478, "global_step": 16851, "epoch": 468, "lr": 0.001} +{"train_loss": 0.031096946448087692, "train_loss_bc": 0.0310966819524765, "train_loss_llm": 2.6434659957885742e-05, "grad_norm": 0.07572727650403976, "global_step": 16852, "epoch": 468, "lr": 0.001} +{"train_loss": 0.033628981560468674, "train_loss_bc": 0.033628981560468674, "train_loss_llm": 0.0, "grad_norm": 0.026331502944231033, "global_step": 16853, "epoch": 468, "lr": 0.001} +{"train_loss": 0.033446844667196274, "train_loss_bc": 0.03343433886766434, "train_loss_llm": 0.0012505948543548584, "grad_norm": 0.047944068908691406, "global_step": 16854, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03864450380206108, "train_loss_bc": 0.03864450380206108, "train_loss_llm": 0.0, "grad_norm": 0.062665194272995, "global_step": 16855, "epoch": 468, "lr": 0.001} +{"train_loss": 0.036138784140348434, "train_loss_bc": 0.036138784140348434, "train_loss_llm": 0.0, "grad_norm": 0.07880467176437378, "global_step": 16856, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03435864299535751, "train_loss_bc": 0.03435864299535751, "train_loss_llm": 0.0, "grad_norm": 0.02839099057018757, "global_step": 16857, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03346412256360054, "train_loss_bc": 0.03346412256360054, "train_loss_llm": 0.0, "grad_norm": 0.041636109352111816, "global_step": 16858, "epoch": 468, "lr": 0.001} +{"train_loss": 0.036751870065927505, "train_loss_bc": 0.036751870065927505, "train_loss_llm": 0.0, "grad_norm": 0.060015689581632614, "global_step": 16859, "epoch": 468, "lr": 0.001} +{"train_loss": 0.033272843807935715, "train_loss_bc": 0.0332724004983902, "train_loss_llm": 4.44948673248291e-05, "grad_norm": 0.07522130757570267, "global_step": 16860, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03158405050635338, "train_loss_bc": 0.03158405050635338, "train_loss_llm": 0.0, "grad_norm": 0.028994804248213768, "global_step": 16861, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03642113506793976, "train_loss_bc": 0.03642113506793976, "train_loss_llm": 0.0, "grad_norm": 0.04444064572453499, "global_step": 16862, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03305455297231674, "train_loss_bc": 0.03305455297231674, "train_loss_llm": 0.0, "grad_norm": 0.061954572796821594, "global_step": 16863, "epoch": 468, "lr": 0.001} +{"train_loss": 0.041084617376327515, "train_loss_bc": 0.041084617376327515, "train_loss_llm": 0.0, "grad_norm": 0.07723111659288406, "global_step": 16864, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03639978915452957, "train_loss_bc": 0.03639978915452957, "train_loss_llm": 0.0, "grad_norm": 0.027681348845362663, "global_step": 16865, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03553600236773491, "train_loss_bc": 0.03553600236773491, "train_loss_llm": 0.0, "grad_norm": 0.04710868373513222, "global_step": 16866, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03519798442721367, "train_loss_bc": 0.03519798442721367, "train_loss_llm": 0.0, "grad_norm": 0.06039758771657944, "global_step": 16867, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03809386119246483, "train_loss_bc": 0.03809386119246483, "train_loss_llm": 0.0, "grad_norm": 0.07864076644182205, "global_step": 16868, "epoch": 468, "lr": 0.001} +{"train_loss": 0.02787431888282299, "train_loss_bc": 0.02787431888282299, "train_loss_llm": 0.0, "grad_norm": 0.025188026949763298, "global_step": 16869, "epoch": 468, "lr": 0.001} +{"train_loss": 0.032429635524749756, "train_loss_bc": 0.032422639429569244, "train_loss_llm": 0.0006996393203735352, "grad_norm": 0.05141432210803032, "global_step": 16870, "epoch": 468, "lr": 0.001} +{"train_loss": 0.0324748158454895, "train_loss_bc": 0.0324748158454895, "train_loss_llm": 0.0, "grad_norm": 0.06304080784320831, "global_step": 16871, "epoch": 468, "lr": 0.001} +{"train_loss": 0.039281975477933884, "train_loss_bc": 0.039281975477933884, "train_loss_llm": 0.0, "grad_norm": 0.06981003284454346, "global_step": 16872, "epoch": 468, "lr": 0.001} +{"train_loss": 0.037301305681467056, "train_loss_bc": 0.037301305681467056, "train_loss_llm": 0.0, "grad_norm": 0.03828679770231247, "global_step": 16873, "epoch": 468, "lr": 0.001} +{"train_loss": 0.02859163098037243, "train_loss_bc": 0.02859163098037243, "train_loss_llm": 0.0, "grad_norm": 0.05026566609740257, "global_step": 16874, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03967180848121643, "train_loss_bc": 0.03967180848121643, "train_loss_llm": 0.0, "grad_norm": 0.07040058821439743, "global_step": 16875, "epoch": 468, "lr": 0.001} +{"train_loss": 0.035472188144922256, "train_loss_bc": 0.035472188144922256, "train_loss_llm": 0.0, "grad_norm": 0.07953651249408722, "global_step": 16876, "epoch": 468, "lr": 0.001} +{"train_loss": 0.02959263324737549, "train_loss_bc": 0.029591841623187065, "train_loss_llm": 7.924437522888184e-05, "grad_norm": 0.01835540495812893, "global_step": 16877, "epoch": 468, "lr": 0.001} +{"train_loss": 0.035316623747348785, "train_loss_bc": 0.035316623747348785, "train_loss_llm": 0.0, "grad_norm": 0.03427630290389061, "global_step": 16878, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03572767600417137, "train_loss_bc": 0.035721730440855026, "train_loss_llm": 0.0005946159362792969, "grad_norm": 0.044368401169776917, "global_step": 16879, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03789107874035835, "train_loss_bc": 0.03789107874035835, "train_loss_llm": 0.0, "grad_norm": 0.06116581708192825, "global_step": 16880, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03924369439482689, "train_loss_bc": 0.03923732787370682, "train_loss_llm": 0.0006368160247802734, "grad_norm": 0.02749462239444256, "global_step": 16881, "epoch": 468, "lr": 0.001} +{"train_loss": 0.030768796801567078, "train_loss_bc": 0.030751243233680725, "train_loss_llm": 0.0017553269863128662, "grad_norm": 0.044002097100019455, "global_step": 16882, "epoch": 468, "lr": 0.001} +{"train_loss": 0.03476441460144189, "train_loss_bc": 0.030901260673999786, "train_loss_llm": 0.0, "grad_norm": 0.0564461424946785, "global_step": 16883, "epoch": 468, "lr": 0.001, "val_loss": 0.07273586094379425} +{"train_loss": 0.03310501202940941, "train_loss_bc": 0.03310501202940941, "train_loss_llm": 0.0, "grad_norm": 0.07454775273799896, "global_step": 16884, "epoch": 469, "lr": 0.001} +{"train_loss": 0.04096183925867081, "train_loss_bc": 0.04096183925867081, "train_loss_llm": 0.0, "grad_norm": 0.036348920315504074, "global_step": 16885, "epoch": 469, "lr": 0.001} +{"train_loss": 0.032476555556058884, "train_loss_bc": 0.032476555556058884, "train_loss_llm": 0.0, "grad_norm": 0.06343773007392883, "global_step": 16886, "epoch": 469, "lr": 0.001} +{"train_loss": 0.029640164226293564, "train_loss_bc": 0.029640164226293564, "train_loss_llm": 0.0, "grad_norm": 0.07384238392114639, "global_step": 16887, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03728653863072395, "train_loss_bc": 0.03728653863072395, "train_loss_llm": 0.0, "grad_norm": 0.08649584650993347, "global_step": 16888, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03461478650569916, "train_loss_bc": 0.0346001498401165, "train_loss_llm": 0.0014636814594268799, "grad_norm": 0.026337429881095886, "global_step": 16889, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03474436700344086, "train_loss_bc": 0.034727662801742554, "train_loss_llm": 0.0016703307628631592, "grad_norm": 0.03647642582654953, "global_step": 16890, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03032226301729679, "train_loss_bc": 0.030311070382595062, "train_loss_llm": 0.0011192262172698975, "grad_norm": 0.05137599632143974, "global_step": 16891, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03499447554349899, "train_loss_bc": 0.03499447554349899, "train_loss_llm": 0.0, "grad_norm": 0.05631714314222336, "global_step": 16892, "epoch": 469, "lr": 0.001} +{"train_loss": 0.0343666635453701, "train_loss_bc": 0.0343666635453701, "train_loss_llm": 0.0, "grad_norm": 0.02371375262737274, "global_step": 16893, "epoch": 469, "lr": 0.001} +{"train_loss": 0.04005680978298187, "train_loss_bc": 0.04005680978298187, "train_loss_llm": 0.0, "grad_norm": 0.03959006816148758, "global_step": 16894, "epoch": 469, "lr": 0.001} +{"train_loss": 0.0379321426153183, "train_loss_bc": 0.0379321426153183, "train_loss_llm": 0.0, "grad_norm": 0.0559731051325798, "global_step": 16895, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03834681212902069, "train_loss_bc": 0.03833625093102455, "train_loss_llm": 0.001056075096130371, "grad_norm": 0.0682215616106987, "global_step": 16896, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03631299361586571, "train_loss_bc": 0.03630182519555092, "train_loss_llm": 0.0011166632175445557, "grad_norm": 0.026456983759999275, "global_step": 16897, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03645254671573639, "train_loss_bc": 0.03645254671573639, "train_loss_llm": 0.0, "grad_norm": 0.040881119668483734, "global_step": 16898, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03665788471698761, "train_loss_bc": 0.03665388375520706, "train_loss_llm": 0.0004000663757324219, "grad_norm": 0.05231804400682449, "global_step": 16899, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03418206050992012, "train_loss_bc": 0.03418206050992012, "train_loss_llm": 0.0, "grad_norm": 0.06236362084746361, "global_step": 16900, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03665483370423317, "train_loss_bc": 0.03665483370423317, "train_loss_llm": 0.0, "grad_norm": 0.03284921869635582, "global_step": 16901, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03190663084387779, "train_loss_bc": 0.031893040984869, "train_loss_llm": 0.0013588666915893555, "grad_norm": 0.04846592992544174, "global_step": 16902, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03863803669810295, "train_loss_bc": 0.03863414004445076, "train_loss_llm": 0.0003896951675415039, "grad_norm": 0.06641054153442383, "global_step": 16903, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03704404458403587, "train_loss_bc": 0.0370357520878315, "train_loss_llm": 0.0008292198181152344, "grad_norm": 0.08594048768281937, "global_step": 16904, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03350338712334633, "train_loss_bc": 0.03350338712334633, "train_loss_llm": 0.0, "grad_norm": 0.02420555241405964, "global_step": 16905, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03559669852256775, "train_loss_bc": 0.03559669852256775, "train_loss_llm": 0.0, "grad_norm": 0.044337980449199677, "global_step": 16906, "epoch": 469, "lr": 0.001} +{"train_loss": 0.02957461029291153, "train_loss_bc": 0.02957461029291153, "train_loss_llm": 0.0, "grad_norm": 0.06210862100124359, "global_step": 16907, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03368799015879631, "train_loss_bc": 0.03368799015879631, "train_loss_llm": 0.0, "grad_norm": 0.0746331587433815, "global_step": 16908, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03813239187002182, "train_loss_bc": 0.03813239187002182, "train_loss_llm": 0.0, "grad_norm": 0.03332914784550667, "global_step": 16909, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03307143226265907, "train_loss_bc": 0.03306363523006439, "train_loss_llm": 0.0007796585559844971, "grad_norm": 0.0424666553735733, "global_step": 16910, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03275265917181969, "train_loss_bc": 0.03275265917181969, "train_loss_llm": 0.0, "grad_norm": 0.05944022908806801, "global_step": 16911, "epoch": 469, "lr": 0.001} +{"train_loss": 0.02950657345354557, "train_loss_bc": 0.02950657345354557, "train_loss_llm": 0.0, "grad_norm": 0.06889112293720245, "global_step": 16912, "epoch": 469, "lr": 0.001} +{"train_loss": 0.029868267476558685, "train_loss_bc": 0.029868267476558685, "train_loss_llm": 0.0, "grad_norm": 0.023105856031179428, "global_step": 16913, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03675292059779167, "train_loss_bc": 0.03675292059779167, "train_loss_llm": 0.0, "grad_norm": 0.035574771463871, "global_step": 16914, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03608926013112068, "train_loss_bc": 0.03607107326388359, "train_loss_llm": 0.0018187463283538818, "grad_norm": 0.04891514778137207, "global_step": 16915, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03062622994184494, "train_loss_bc": 0.03062622994184494, "train_loss_llm": 0.0, "grad_norm": 0.05936463922262192, "global_step": 16916, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03378036990761757, "train_loss_bc": 0.03378036990761757, "train_loss_llm": 0.0, "grad_norm": 0.02464909851551056, "global_step": 16917, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03270901367068291, "train_loss_bc": 0.03270901367068291, "train_loss_llm": 0.0, "grad_norm": 0.041639357805252075, "global_step": 16918, "epoch": 469, "lr": 0.001} +{"train_loss": 0.03462200684265958, "train_loss_bc": 0.03404298052191734, "train_loss_llm": 0.0, "grad_norm": 0.05294470489025116, "global_step": 16919, "epoch": 469, "lr": 0.001, "val_loss": 0.07389514148235321} +{"train_loss": 0.03905560448765755, "train_loss_bc": 0.03904886916279793, "train_loss_llm": 0.0006733536720275879, "grad_norm": 0.06616886705160141, "global_step": 16920, "epoch": 470, "lr": 0.001} +{"train_loss": 0.038598451763391495, "train_loss_bc": 0.038598451763391495, "train_loss_llm": 0.0, "grad_norm": 0.02854890376329422, "global_step": 16921, "epoch": 470, "lr": 0.001} +{"train_loss": 0.029139608144760132, "train_loss_bc": 0.029132623225450516, "train_loss_llm": 0.0006985664367675781, "grad_norm": 0.044063519686460495, "global_step": 16922, "epoch": 470, "lr": 0.001} +{"train_loss": 0.04038562998175621, "train_loss_bc": 0.04038562998175621, "train_loss_llm": 0.0, "grad_norm": 0.05659621208906174, "global_step": 16923, "epoch": 470, "lr": 0.001} +{"train_loss": 0.04331826791167259, "train_loss_bc": 0.04330025240778923, "train_loss_llm": 0.0018013715744018555, "grad_norm": 0.08924023807048798, "global_step": 16924, "epoch": 470, "lr": 0.001} +{"train_loss": 0.02964724786579609, "train_loss_bc": 0.02964724786579609, "train_loss_llm": 0.0, "grad_norm": 0.02666947990655899, "global_step": 16925, "epoch": 470, "lr": 0.001} +{"train_loss": 0.035088103264570236, "train_loss_bc": 0.035088103264570236, "train_loss_llm": 0.0, "grad_norm": 0.04206044599413872, "global_step": 16926, "epoch": 470, "lr": 0.001} +{"train_loss": 0.033387910574674606, "train_loss_bc": 0.033387910574674606, "train_loss_llm": 0.0, "grad_norm": 0.06223736330866814, "global_step": 16927, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03361276909708977, "train_loss_bc": 0.03361276909708977, "train_loss_llm": 0.0, "grad_norm": 0.09269863367080688, "global_step": 16928, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03793526813387871, "train_loss_bc": 0.03793526813387871, "train_loss_llm": 0.0, "grad_norm": 0.025518031790852547, "global_step": 16929, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03504632040858269, "train_loss_bc": 0.03504632040858269, "train_loss_llm": 0.0, "grad_norm": 0.04516410827636719, "global_step": 16930, "epoch": 470, "lr": 0.001} +{"train_loss": 0.031430330127477646, "train_loss_bc": 0.031430330127477646, "train_loss_llm": 0.0, "grad_norm": 0.04944148659706116, "global_step": 16931, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03804340958595276, "train_loss_bc": 0.0380387157201767, "train_loss_llm": 0.00046938657760620117, "grad_norm": 0.06264439970254898, "global_step": 16932, "epoch": 470, "lr": 0.001} +{"train_loss": 0.026529209688305855, "train_loss_bc": 0.026516351848840714, "train_loss_llm": 0.001285851001739502, "grad_norm": 0.023119261488318443, "global_step": 16933, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03187001869082451, "train_loss_bc": 0.03187001869082451, "train_loss_llm": 0.0, "grad_norm": 0.03221118450164795, "global_step": 16934, "epoch": 470, "lr": 0.001} +{"train_loss": 0.027326123788952827, "train_loss_bc": 0.027326123788952827, "train_loss_llm": 0.0, "grad_norm": 0.04071890190243721, "global_step": 16935, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03799103572964668, "train_loss_bc": 0.03799103572964668, "train_loss_llm": 0.0, "grad_norm": 0.06462761759757996, "global_step": 16936, "epoch": 470, "lr": 0.001} +{"train_loss": 0.030249323695898056, "train_loss_bc": 0.030242741107940674, "train_loss_llm": 0.0006582140922546387, "grad_norm": 0.02754925936460495, "global_step": 16937, "epoch": 470, "lr": 0.001} +{"train_loss": 0.032390449196100235, "train_loss_bc": 0.032378628849983215, "train_loss_llm": 0.0011821389198303223, "grad_norm": 0.04135045036673546, "global_step": 16938, "epoch": 470, "lr": 0.001} +{"train_loss": 0.034182868897914886, "train_loss_bc": 0.034175943583250046, "train_loss_llm": 0.0006925463676452637, "grad_norm": 0.05607809126377106, "global_step": 16939, "epoch": 470, "lr": 0.001} +{"train_loss": 0.030720094218850136, "train_loss_bc": 0.03071952983736992, "train_loss_llm": 5.650520324707031e-05, "grad_norm": 0.07489736378192902, "global_step": 16940, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03617792949080467, "train_loss_bc": 0.03617792949080467, "train_loss_llm": 0.0, "grad_norm": 0.024177225306630135, "global_step": 16941, "epoch": 470, "lr": 0.001} +{"train_loss": 0.038297321647405624, "train_loss_bc": 0.038297321647405624, "train_loss_llm": 0.0, "grad_norm": 0.03703370690345764, "global_step": 16942, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03753485530614853, "train_loss_bc": 0.03753485530614853, "train_loss_llm": 0.0, "grad_norm": 0.052333392202854156, "global_step": 16943, "epoch": 470, "lr": 0.001} +{"train_loss": 0.039647478610277176, "train_loss_bc": 0.039647478610277176, "train_loss_llm": 0.0, "grad_norm": 0.06478969752788544, "global_step": 16944, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03619745373725891, "train_loss_bc": 0.0361952967941761, "train_loss_llm": 0.00021567940711975098, "grad_norm": 0.026749243959784508, "global_step": 16945, "epoch": 470, "lr": 0.001} +{"train_loss": 0.038395170122385025, "train_loss_bc": 0.03838905692100525, "train_loss_llm": 0.0006112158298492432, "grad_norm": 0.03823385387659073, "global_step": 16946, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03124966472387314, "train_loss_bc": 0.03123895823955536, "train_loss_llm": 0.0010707378387451172, "grad_norm": 0.048039164394140244, "global_step": 16947, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03143936023116112, "train_loss_bc": 0.03142718970775604, "train_loss_llm": 0.001216977834701538, "grad_norm": 0.06206754967570305, "global_step": 16948, "epoch": 470, "lr": 0.001} +{"train_loss": 0.040633074939250946, "train_loss_bc": 0.040633074939250946, "train_loss_llm": 0.0, "grad_norm": 0.034248512238264084, "global_step": 16949, "epoch": 470, "lr": 0.001} +{"train_loss": 0.033995289355516434, "train_loss_bc": 0.033995289355516434, "train_loss_llm": 0.0, "grad_norm": 0.04673635959625244, "global_step": 16950, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03296587988734245, "train_loss_bc": 0.03296587988734245, "train_loss_llm": 0.0, "grad_norm": 0.05852385610342026, "global_step": 16951, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03748296573758125, "train_loss_bc": 0.03748296573758125, "train_loss_llm": 0.0, "grad_norm": 0.07176977396011353, "global_step": 16952, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03334186598658562, "train_loss_bc": 0.03333728387951851, "train_loss_llm": 0.00045818090438842773, "grad_norm": 0.027871081605553627, "global_step": 16953, "epoch": 470, "lr": 0.001} +{"train_loss": 0.031726837158203125, "train_loss_bc": 0.031726837158203125, "train_loss_llm": 0.0, "grad_norm": 0.04697635769844055, "global_step": 16954, "epoch": 470, "lr": 0.001} +{"train_loss": 0.03484095943470796, "train_loss_bc": 0.03924134746193886, "train_loss_llm": 0.0, "grad_norm": 0.0606837272644043, "global_step": 16955, "epoch": 470, "lr": 0.001, "train/cumulative_reward": 119.1080010597863, "train/mean_score": 12.642943611818621, "train/success_rate": 0.3, "test/cumulative_reward": 40.38088309310852, "test/mean_score": 8.447789776293453, "test/success_rate": 0.18, "val_loss": 0.07103695720434189, "train_action_mse_error": 0.03728442266583443} +{"train_loss": 0.02948891930282116, "train_loss_bc": 0.02948891930282116, "train_loss_llm": 0.0, "grad_norm": 0.07859280705451965, "global_step": 16956, "epoch": 471, "lr": 0.001} +{"train_loss": 0.032387860119342804, "train_loss_bc": 0.032387860119342804, "train_loss_llm": 0.0, "grad_norm": 0.026870958507061005, "global_step": 16957, "epoch": 471, "lr": 0.001} +{"train_loss": 0.0413198247551918, "train_loss_bc": 0.0413198247551918, "train_loss_llm": 0.0, "grad_norm": 0.04195637255907059, "global_step": 16958, "epoch": 471, "lr": 0.001} +{"train_loss": 0.026909908279776573, "train_loss_bc": 0.026909908279776573, "train_loss_llm": 0.0, "grad_norm": 0.056382350623607635, "global_step": 16959, "epoch": 471, "lr": 0.001} +{"train_loss": 0.0354161262512207, "train_loss_bc": 0.0354161262512207, "train_loss_llm": 0.0, "grad_norm": 0.07518211752176285, "global_step": 16960, "epoch": 471, "lr": 0.001} +{"train_loss": 0.032387860119342804, "train_loss_bc": 0.032387860119342804, "train_loss_llm": 0.0, "grad_norm": 0.024658340960741043, "global_step": 16961, "epoch": 471, "lr": 0.001} +{"train_loss": 0.030724551528692245, "train_loss_bc": 0.030724551528692245, "train_loss_llm": 0.0, "grad_norm": 0.04056588560342789, "global_step": 16962, "epoch": 471, "lr": 0.001} +{"train_loss": 0.033823903650045395, "train_loss_bc": 0.033823903650045395, "train_loss_llm": 0.0, "grad_norm": 0.05160792917013168, "global_step": 16963, "epoch": 471, "lr": 0.001} +{"train_loss": 0.034217409789562225, "train_loss_bc": 0.03420507535338402, "train_loss_llm": 0.0012335777282714844, "grad_norm": 0.06454936414957047, "global_step": 16964, "epoch": 471, "lr": 0.001} +{"train_loss": 0.028961068019270897, "train_loss_bc": 0.028961068019270897, "train_loss_llm": 0.0, "grad_norm": 0.018075905740261078, "global_step": 16965, "epoch": 471, "lr": 0.001} +{"train_loss": 0.028723839670419693, "train_loss_bc": 0.02871982380747795, "train_loss_llm": 0.0004016757011413574, "grad_norm": 0.03600631281733513, "global_step": 16966, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03825189545750618, "train_loss_bc": 0.03825189545750618, "train_loss_llm": 0.0, "grad_norm": 0.04325612634420395, "global_step": 16967, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03654208779335022, "train_loss_bc": 0.03653819113969803, "train_loss_llm": 0.0003895759582519531, "grad_norm": 0.049420151859521866, "global_step": 16968, "epoch": 471, "lr": 0.001} +{"train_loss": 0.0322660468518734, "train_loss_bc": 0.0322660468518734, "train_loss_llm": 0.0, "grad_norm": 0.025137852877378464, "global_step": 16969, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03266529738903046, "train_loss_bc": 0.03266529738903046, "train_loss_llm": 0.0, "grad_norm": 0.03539254516363144, "global_step": 16970, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03963498771190643, "train_loss_bc": 0.03961389511823654, "train_loss_llm": 0.002109318971633911, "grad_norm": 0.05028810352087021, "global_step": 16971, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03414000943303108, "train_loss_bc": 0.03414000943303108, "train_loss_llm": 0.0, "grad_norm": 0.0624326691031456, "global_step": 16972, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03683118149638176, "train_loss_bc": 0.03683118149638176, "train_loss_llm": 0.0, "grad_norm": 0.028034288436174393, "global_step": 16973, "epoch": 471, "lr": 0.001} +{"train_loss": 0.024380890652537346, "train_loss_bc": 0.024377159774303436, "train_loss_llm": 0.000373154878616333, "grad_norm": 0.03589307516813278, "global_step": 16974, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03862747177481651, "train_loss_bc": 0.03862747177481651, "train_loss_llm": 0.0, "grad_norm": 0.040384262800216675, "global_step": 16975, "epoch": 471, "lr": 0.001} +{"train_loss": 0.036292608827352524, "train_loss_bc": 0.036292608827352524, "train_loss_llm": 0.0, "grad_norm": 0.04907810688018799, "global_step": 16976, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03937505930662155, "train_loss_bc": 0.03937505930662155, "train_loss_llm": 0.0, "grad_norm": 0.023057641461491585, "global_step": 16977, "epoch": 471, "lr": 0.001} +{"train_loss": 0.030333535745739937, "train_loss_bc": 0.030324162915349007, "train_loss_llm": 0.0009372532367706299, "grad_norm": 0.032393865287303925, "global_step": 16978, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03908870369195938, "train_loss_bc": 0.03908870369195938, "train_loss_llm": 0.0, "grad_norm": 0.04148896411061287, "global_step": 16979, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03322150185704231, "train_loss_bc": 0.03322150185704231, "train_loss_llm": 0.0, "grad_norm": 0.05177418142557144, "global_step": 16980, "epoch": 471, "lr": 0.001} +{"train_loss": 0.030977213755249977, "train_loss_bc": 0.030977213755249977, "train_loss_llm": 0.0, "grad_norm": 0.022435517981648445, "global_step": 16981, "epoch": 471, "lr": 0.001} +{"train_loss": 0.037164293229579926, "train_loss_bc": 0.037164293229579926, "train_loss_llm": 0.0, "grad_norm": 0.034046970307826996, "global_step": 16982, "epoch": 471, "lr": 0.001} +{"train_loss": 0.034890640527009964, "train_loss_bc": 0.03488583117723465, "train_loss_llm": 0.00048109889030456543, "grad_norm": 0.04797755554318428, "global_step": 16983, "epoch": 471, "lr": 0.001} +{"train_loss": 0.04434575140476227, "train_loss_bc": 0.04434575140476227, "train_loss_llm": 0.0, "grad_norm": 0.06683630496263504, "global_step": 16984, "epoch": 471, "lr": 0.001} +{"train_loss": 0.0314113050699234, "train_loss_bc": 0.03140651434659958, "train_loss_llm": 0.00047892332077026367, "grad_norm": 0.024788348004221916, "global_step": 16985, "epoch": 471, "lr": 0.001} +{"train_loss": 0.04145912453532219, "train_loss_bc": 0.04145912453532219, "train_loss_llm": 0.0, "grad_norm": 0.046357542276382446, "global_step": 16986, "epoch": 471, "lr": 0.001} +{"train_loss": 0.0379396453499794, "train_loss_bc": 0.0379396453499794, "train_loss_llm": 0.0, "grad_norm": 0.05726519972085953, "global_step": 16987, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03332793712615967, "train_loss_bc": 0.0333215594291687, "train_loss_llm": 0.0006375908851623535, "grad_norm": 0.07350967824459076, "global_step": 16988, "epoch": 471, "lr": 0.001} +{"train_loss": 0.031442441046237946, "train_loss_bc": 0.031419236212968826, "train_loss_llm": 0.002320319414138794, "grad_norm": 0.0221286378800869, "global_step": 16989, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03754296526312828, "train_loss_bc": 0.03754296526312828, "train_loss_llm": 0.0, "grad_norm": 0.041701722890138626, "global_step": 16990, "epoch": 471, "lr": 0.001} +{"train_loss": 0.03458648526834117, "train_loss_bc": 0.03858208656311035, "train_loss_llm": 0.0017514824867248535, "grad_norm": 0.057973042130470276, "global_step": 16991, "epoch": 471, "lr": 0.001, "val_loss": 0.07424945384263992} +{"train_loss": 0.032245513051748276, "train_loss_bc": 0.03224411606788635, "train_loss_llm": 0.00013962388038635254, "grad_norm": 0.06516721099615097, "global_step": 16992, "epoch": 472, "lr": 0.001} +{"train_loss": 0.035195622593164444, "train_loss_bc": 0.03518931567668915, "train_loss_llm": 0.0006305873394012451, "grad_norm": 0.02546553686261177, "global_step": 16993, "epoch": 472, "lr": 0.001} +{"train_loss": 0.030805867165327072, "train_loss_bc": 0.03079015202820301, "train_loss_llm": 0.0015715360641479492, "grad_norm": 0.03794136643409729, "global_step": 16994, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03757460415363312, "train_loss_bc": 0.03757460415363312, "train_loss_llm": 0.0, "grad_norm": 0.046469006687402725, "global_step": 16995, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03514419496059418, "train_loss_bc": 0.03514419496059418, "train_loss_llm": 0.0, "grad_norm": 0.061284393072128296, "global_step": 16996, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03861319273710251, "train_loss_bc": 0.03861319273710251, "train_loss_llm": 0.0, "grad_norm": 0.02413286082446575, "global_step": 16997, "epoch": 472, "lr": 0.001} +{"train_loss": 0.030310355126857758, "train_loss_bc": 0.030310355126857758, "train_loss_llm": 0.0, "grad_norm": 0.03101949952542782, "global_step": 16998, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03457353636622429, "train_loss_bc": 0.03457353636622429, "train_loss_llm": 0.0, "grad_norm": 0.04430403560400009, "global_step": 16999, "epoch": 472, "lr": 0.001} +{"train_loss": 0.028656965121626854, "train_loss_bc": 0.028656965121626854, "train_loss_llm": 0.0, "grad_norm": 0.0502922460436821, "global_step": 17000, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03757248818874359, "train_loss_bc": 0.03757248818874359, "train_loss_llm": 0.0, "grad_norm": 0.032394059002399445, "global_step": 17001, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03141254186630249, "train_loss_bc": 0.03140529990196228, "train_loss_llm": 0.0007240176200866699, "grad_norm": 0.04499068483710289, "global_step": 17002, "epoch": 472, "lr": 0.001} +{"train_loss": 0.029517004266381264, "train_loss_bc": 0.029517004266381264, "train_loss_llm": 0.0, "grad_norm": 0.05745991691946983, "global_step": 17003, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03373775631189346, "train_loss_bc": 0.03373775631189346, "train_loss_llm": 0.0, "grad_norm": 0.06991562992334366, "global_step": 17004, "epoch": 472, "lr": 0.001} +{"train_loss": 0.036759696900844574, "train_loss_bc": 0.036759696900844574, "train_loss_llm": 0.0, "grad_norm": 0.024958252906799316, "global_step": 17005, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03756701201200485, "train_loss_bc": 0.03756701201200485, "train_loss_llm": 0.0, "grad_norm": 0.04239673539996147, "global_step": 17006, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03331375867128372, "train_loss_bc": 0.03331228718161583, "train_loss_llm": 0.00014713406562805176, "grad_norm": 0.06345472484827042, "global_step": 17007, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03873945027589798, "train_loss_bc": 0.03873945027589798, "train_loss_llm": 0.0, "grad_norm": 0.0771726742386818, "global_step": 17008, "epoch": 472, "lr": 0.001} +{"train_loss": 0.033282067626714706, "train_loss_bc": 0.033282067626714706, "train_loss_llm": 0.0, "grad_norm": 0.02164282649755478, "global_step": 17009, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03238845616579056, "train_loss_bc": 0.03238845616579056, "train_loss_llm": 0.0, "grad_norm": 0.03777078166604042, "global_step": 17010, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03286474943161011, "train_loss_bc": 0.03286474943161011, "train_loss_llm": 0.0, "grad_norm": 0.046979766339063644, "global_step": 17011, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03828361630439758, "train_loss_bc": 0.03828361630439758, "train_loss_llm": 0.0, "grad_norm": 0.056496597826480865, "global_step": 17012, "epoch": 472, "lr": 0.001} +{"train_loss": 0.037108320742845535, "train_loss_bc": 0.037108320742845535, "train_loss_llm": 0.0, "grad_norm": 0.029359331354498863, "global_step": 17013, "epoch": 472, "lr": 0.001} +{"train_loss": 0.035944290459156036, "train_loss_bc": 0.035944290459156036, "train_loss_llm": 0.0, "grad_norm": 0.049417901784181595, "global_step": 17014, "epoch": 472, "lr": 0.001} +{"train_loss": 0.039681628346443176, "train_loss_bc": 0.039677709341049194, "train_loss_llm": 0.00039201974868774414, "grad_norm": 0.06047491729259491, "global_step": 17015, "epoch": 472, "lr": 0.001} +{"train_loss": 0.0341600738465786, "train_loss_bc": 0.0341600738465786, "train_loss_llm": 0.0, "grad_norm": 0.07290104031562805, "global_step": 17016, "epoch": 472, "lr": 0.001} +{"train_loss": 0.032838162034749985, "train_loss_bc": 0.032838162034749985, "train_loss_llm": 0.0, "grad_norm": 0.0299974512308836, "global_step": 17017, "epoch": 472, "lr": 0.001} +{"train_loss": 0.0313287079334259, "train_loss_bc": 0.031319357454776764, "train_loss_llm": 0.0009351670742034912, "grad_norm": 0.05526157468557358, "global_step": 17018, "epoch": 472, "lr": 0.001} +{"train_loss": 0.031209660694003105, "train_loss_bc": 0.031209660694003105, "train_loss_llm": 0.0, "grad_norm": 0.07066088914871216, "global_step": 17019, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03772173076868057, "train_loss_bc": 0.03772173076868057, "train_loss_llm": 0.0, "grad_norm": 0.07514024525880814, "global_step": 17020, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03172864764928818, "train_loss_bc": 0.03172864764928818, "train_loss_llm": 0.0, "grad_norm": 0.026150666177272797, "global_step": 17021, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03699963539838791, "train_loss_bc": 0.03699963539838791, "train_loss_llm": 0.0, "grad_norm": 0.04356183484196663, "global_step": 17022, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03671650215983391, "train_loss_bc": 0.03671106696128845, "train_loss_llm": 0.0005436241626739502, "grad_norm": 0.054367806762456894, "global_step": 17023, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03626502677798271, "train_loss_bc": 0.03626502677798271, "train_loss_llm": 0.0, "grad_norm": 0.06887427717447281, "global_step": 17024, "epoch": 472, "lr": 0.001} +{"train_loss": 0.03722900524735451, "train_loss_bc": 0.037213969975709915, "train_loss_llm": 0.00150337815284729, "grad_norm": 0.027716370299458504, "global_step": 17025, "epoch": 472, "lr": 0.001} +{"train_loss": 0.037017934024333954, "train_loss_bc": 0.037017934024333954, "train_loss_llm": 0.0, "grad_norm": 0.03907840698957443, "global_step": 17026, "epoch": 472, "lr": 0.001} +{"train_loss": 0.0344260690940751, "train_loss_bc": 0.024830253794789314, "train_loss_llm": 4.5865774154663086e-05, "grad_norm": 0.049438562244176865, "global_step": 17027, "epoch": 472, "lr": 0.001, "val_loss": 0.0724957063794136} +{"train_loss": 0.03894561156630516, "train_loss_bc": 0.03894561156630516, "train_loss_llm": 0.0, "grad_norm": 0.061018265783786774, "global_step": 17028, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03819754347205162, "train_loss_bc": 0.038184754550457, "train_loss_llm": 0.0012787580490112305, "grad_norm": 0.026561753824353218, "global_step": 17029, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03749842941761017, "train_loss_bc": 0.03749842941761017, "train_loss_llm": 0.0, "grad_norm": 0.04316478222608566, "global_step": 17030, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03178076818585396, "train_loss_bc": 0.03178076818585396, "train_loss_llm": 0.0, "grad_norm": 0.06191074475646019, "global_step": 17031, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03254736587405205, "train_loss_bc": 0.03254736587405205, "train_loss_llm": 0.0, "grad_norm": 0.0742763876914978, "global_step": 17032, "epoch": 473, "lr": 0.001} +{"train_loss": 0.031248874962329865, "train_loss_bc": 0.031248874962329865, "train_loss_llm": 0.0, "grad_norm": 0.02228238433599472, "global_step": 17033, "epoch": 473, "lr": 0.001} +{"train_loss": 0.030944589525461197, "train_loss_bc": 0.030934469774365425, "train_loss_llm": 0.0010120272636413574, "grad_norm": 0.03219950944185257, "global_step": 17034, "epoch": 473, "lr": 0.001} +{"train_loss": 0.0351044237613678, "train_loss_bc": 0.0351044237613678, "train_loss_llm": 0.0, "grad_norm": 0.053334791213274, "global_step": 17035, "epoch": 473, "lr": 0.001} +{"train_loss": 0.036425601691007614, "train_loss_bc": 0.036425601691007614, "train_loss_llm": 0.0, "grad_norm": 0.07107821106910706, "global_step": 17036, "epoch": 473, "lr": 0.001} +{"train_loss": 0.041322991251945496, "train_loss_bc": 0.04132065176963806, "train_loss_llm": 0.00023385882377624512, "grad_norm": 0.024645499885082245, "global_step": 17037, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03379969671368599, "train_loss_bc": 0.03379969671368599, "train_loss_llm": 0.0, "grad_norm": 0.04019579663872719, "global_step": 17038, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03333766758441925, "train_loss_bc": 0.033326830714941025, "train_loss_llm": 0.0010835528373718262, "grad_norm": 0.04926307499408722, "global_step": 17039, "epoch": 473, "lr": 0.001} +{"train_loss": 0.036944255232810974, "train_loss_bc": 0.03692444786429405, "train_loss_llm": 0.0019807517528533936, "grad_norm": 0.06534118205308914, "global_step": 17040, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03378798067569733, "train_loss_bc": 0.03378798067569733, "train_loss_llm": 0.0, "grad_norm": 0.022659987211227417, "global_step": 17041, "epoch": 473, "lr": 0.001} +{"train_loss": 0.030528968200087547, "train_loss_bc": 0.030502062290906906, "train_loss_llm": 0.002690553665161133, "grad_norm": 0.028474194929003716, "global_step": 17042, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03578187897801399, "train_loss_bc": 0.03578187897801399, "train_loss_llm": 0.0, "grad_norm": 0.05477135628461838, "global_step": 17043, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03489586338400841, "train_loss_bc": 0.03489586338400841, "train_loss_llm": 0.0, "grad_norm": 0.06578266620635986, "global_step": 17044, "epoch": 473, "lr": 0.001} +{"train_loss": 0.02672978676855564, "train_loss_bc": 0.026717839762568474, "train_loss_llm": 0.001194685697555542, "grad_norm": 0.022295091301202774, "global_step": 17045, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03533041477203369, "train_loss_bc": 0.03533041477203369, "train_loss_llm": 0.0, "grad_norm": 0.038127850741147995, "global_step": 17046, "epoch": 473, "lr": 0.001} +{"train_loss": 0.045504212379455566, "train_loss_bc": 0.045504212379455566, "train_loss_llm": 0.0, "grad_norm": 0.054418791085481644, "global_step": 17047, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03255242109298706, "train_loss_bc": 0.03255242109298706, "train_loss_llm": 0.0, "grad_norm": 0.06112402677536011, "global_step": 17048, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03493770211935043, "train_loss_bc": 0.03493770211935043, "train_loss_llm": 0.0, "grad_norm": 0.03644874319434166, "global_step": 17049, "epoch": 473, "lr": 0.001} +{"train_loss": 0.04064738750457764, "train_loss_bc": 0.04064738750457764, "train_loss_llm": 0.0, "grad_norm": 0.0514083169400692, "global_step": 17050, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03402033820748329, "train_loss_bc": 0.03399665653705597, "train_loss_llm": 0.002368152141571045, "grad_norm": 0.060374848544597626, "global_step": 17051, "epoch": 473, "lr": 0.001} +{"train_loss": 0.029127134010195732, "train_loss_bc": 0.029127134010195732, "train_loss_llm": 0.0, "grad_norm": 0.06852101534605026, "global_step": 17052, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03455816209316254, "train_loss_bc": 0.03455816209316254, "train_loss_llm": 0.0, "grad_norm": 0.020396865904331207, "global_step": 17053, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03403865545988083, "train_loss_bc": 0.03403865545988083, "train_loss_llm": 0.0, "grad_norm": 0.03417474776506424, "global_step": 17054, "epoch": 473, "lr": 0.001} +{"train_loss": 0.030550271272659302, "train_loss_bc": 0.030550271272659302, "train_loss_llm": 0.0, "grad_norm": 0.0465005598962307, "global_step": 17055, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03250374272465706, "train_loss_bc": 0.03250335156917572, "train_loss_llm": 3.9249658584594727e-05, "grad_norm": 0.06723421066999435, "global_step": 17056, "epoch": 473, "lr": 0.001} +{"train_loss": 0.036251794546842575, "train_loss_bc": 0.036251794546842575, "train_loss_llm": 0.0, "grad_norm": 0.029936637729406357, "global_step": 17057, "epoch": 473, "lr": 0.001} +{"train_loss": 0.031076520681381226, "train_loss_bc": 0.031076520681381226, "train_loss_llm": 0.0, "grad_norm": 0.04479379579424858, "global_step": 17058, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03623376786708832, "train_loss_bc": 0.03623376786708832, "train_loss_llm": 0.0, "grad_norm": 0.05094287544488907, "global_step": 17059, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03788923844695091, "train_loss_bc": 0.03788923844695091, "train_loss_llm": 0.0, "grad_norm": 0.061634525656700134, "global_step": 17060, "epoch": 473, "lr": 0.001} +{"train_loss": 0.032318029552698135, "train_loss_bc": 0.032318029552698135, "train_loss_llm": 0.0, "grad_norm": 0.028553251177072525, "global_step": 17061, "epoch": 473, "lr": 0.001} +{"train_loss": 0.03492332249879837, "train_loss_bc": 0.03489178046584129, "train_loss_llm": 0.003154277801513672, "grad_norm": 0.04172609746456146, "global_step": 17062, "epoch": 473, "lr": 0.001} +{"train_loss": 0.0346046714629564, "train_loss_bc": 0.033482760190963745, "train_loss_llm": 0.0, "grad_norm": 0.06311564892530441, "global_step": 17063, "epoch": 473, "lr": 0.001, "val_loss": 0.07235397398471832} +{"train_loss": 0.033094994723796844, "train_loss_bc": 0.033094994723796844, "train_loss_llm": 0.0, "grad_norm": 0.07230984419584274, "global_step": 17064, "epoch": 474, "lr": 0.001} +{"train_loss": 0.04427175223827362, "train_loss_bc": 0.04427175223827362, "train_loss_llm": 0.0, "grad_norm": 0.0238619614392519, "global_step": 17065, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03063674457371235, "train_loss_bc": 0.03063674457371235, "train_loss_llm": 0.0, "grad_norm": 0.03473196551203728, "global_step": 17066, "epoch": 474, "lr": 0.001} +{"train_loss": 0.0398215614259243, "train_loss_bc": 0.0398215614259243, "train_loss_llm": 0.0, "grad_norm": 0.05318162590265274, "global_step": 17067, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03094613552093506, "train_loss_bc": 0.03094613552093506, "train_loss_llm": 0.0, "grad_norm": 0.05742882192134857, "global_step": 17068, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03501957282423973, "train_loss_bc": 0.03501957282423973, "train_loss_llm": 0.0, "grad_norm": 0.022720254957675934, "global_step": 17069, "epoch": 474, "lr": 0.001} +{"train_loss": 0.034539829939603806, "train_loss_bc": 0.034539829939603806, "train_loss_llm": 0.0, "grad_norm": 0.033885955810546875, "global_step": 17070, "epoch": 474, "lr": 0.001} +{"train_loss": 0.032394297420978546, "train_loss_bc": 0.03238207846879959, "train_loss_llm": 0.0012217462062835693, "grad_norm": 0.04278985783457756, "global_step": 17071, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03327791392803192, "train_loss_bc": 0.03327791392803192, "train_loss_llm": 0.0, "grad_norm": 0.05818420648574829, "global_step": 17072, "epoch": 474, "lr": 0.001} +{"train_loss": 0.02895505167543888, "train_loss_bc": 0.02895505167543888, "train_loss_llm": 0.0, "grad_norm": 0.033597711473703384, "global_step": 17073, "epoch": 474, "lr": 0.001} +{"train_loss": 0.029396841302514076, "train_loss_bc": 0.029377656057476997, "train_loss_llm": 0.0019184350967407227, "grad_norm": 0.04538562148809433, "global_step": 17074, "epoch": 474, "lr": 0.001} +{"train_loss": 0.038083139806985855, "train_loss_bc": 0.0380806103348732, "train_loss_llm": 0.0002530217170715332, "grad_norm": 0.052265532314777374, "global_step": 17075, "epoch": 474, "lr": 0.001} +{"train_loss": 0.04049580171704292, "train_loss_bc": 0.04049580171704292, "train_loss_llm": 0.0, "grad_norm": 0.0677202045917511, "global_step": 17076, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03296291455626488, "train_loss_bc": 0.03296291455626488, "train_loss_llm": 0.0, "grad_norm": 0.037101756781339645, "global_step": 17077, "epoch": 474, "lr": 0.001} +{"train_loss": 0.038316357880830765, "train_loss_bc": 0.03830675780773163, "train_loss_llm": 0.0009601414203643799, "grad_norm": 0.05443550646305084, "global_step": 17078, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03861246630549431, "train_loss_bc": 0.03861246630549431, "train_loss_llm": 0.0, "grad_norm": 0.06954186409711838, "global_step": 17079, "epoch": 474, "lr": 0.001} +{"train_loss": 0.032177891582250595, "train_loss_bc": 0.032177891582250595, "train_loss_llm": 0.0, "grad_norm": 0.08096474409103394, "global_step": 17080, "epoch": 474, "lr": 0.001} +{"train_loss": 0.033207111060619354, "train_loss_bc": 0.033207111060619354, "train_loss_llm": 0.0, "grad_norm": 0.02247893624007702, "global_step": 17081, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03030831180512905, "train_loss_bc": 0.030305715277791023, "train_loss_llm": 0.00025957822799682617, "grad_norm": 0.029085876420140266, "global_step": 17082, "epoch": 474, "lr": 0.001} +{"train_loss": 0.031638436019420624, "train_loss_bc": 0.03163737803697586, "train_loss_llm": 0.00010594725608825684, "grad_norm": 0.03838026523590088, "global_step": 17083, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03568391874432564, "train_loss_bc": 0.03568391874432564, "train_loss_llm": 0.0, "grad_norm": 0.056348931044340134, "global_step": 17084, "epoch": 474, "lr": 0.001} +{"train_loss": 0.032771479338407516, "train_loss_bc": 0.03276851400732994, "train_loss_llm": 0.00029647350311279297, "grad_norm": 0.0239363182336092, "global_step": 17085, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03568531200289726, "train_loss_bc": 0.03568531200289726, "train_loss_llm": 0.0, "grad_norm": 0.04228908568620682, "global_step": 17086, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03745836764574051, "train_loss_bc": 0.03744035214185715, "train_loss_llm": 0.0018014311790466309, "grad_norm": 0.06055912375450134, "global_step": 17087, "epoch": 474, "lr": 0.001} +{"train_loss": 0.031913481652736664, "train_loss_bc": 0.031913481652736664, "train_loss_llm": 0.0, "grad_norm": 0.07754776626825333, "global_step": 17088, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03474534675478935, "train_loss_bc": 0.03474534675478935, "train_loss_llm": 0.0, "grad_norm": 0.02511047013103962, "global_step": 17089, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03631047531962395, "train_loss_bc": 0.03631047531962395, "train_loss_llm": 0.0, "grad_norm": 0.03519565984606743, "global_step": 17090, "epoch": 474, "lr": 0.001} +{"train_loss": 0.0340631864964962, "train_loss_bc": 0.0340631864964962, "train_loss_llm": 0.0, "grad_norm": 0.05043994262814522, "global_step": 17091, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03564038500189781, "train_loss_bc": 0.03564038500189781, "train_loss_llm": 0.0, "grad_norm": 0.06069047749042511, "global_step": 17092, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03923708200454712, "train_loss_bc": 0.039223045110702515, "train_loss_llm": 0.0014038383960723877, "grad_norm": 0.030488234013319016, "global_step": 17093, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03593875840306282, "train_loss_bc": 0.03593875840306282, "train_loss_llm": 0.0, "grad_norm": 0.04906259477138519, "global_step": 17094, "epoch": 474, "lr": 0.001} +{"train_loss": 0.033333394676446915, "train_loss_bc": 0.033333394676446915, "train_loss_llm": 0.0, "grad_norm": 0.06958688050508499, "global_step": 17095, "epoch": 474, "lr": 0.001} +{"train_loss": 0.030337143689393997, "train_loss_bc": 0.030337143689393997, "train_loss_llm": 0.0, "grad_norm": 0.08063328266143799, "global_step": 17096, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03092997893691063, "train_loss_bc": 0.03092997893691063, "train_loss_llm": 0.0, "grad_norm": 0.022208495065569878, "global_step": 17097, "epoch": 474, "lr": 0.001} +{"train_loss": 0.04068242385983467, "train_loss_bc": 0.04066212475299835, "train_loss_llm": 0.0020298361778259277, "grad_norm": 0.039945878088474274, "global_step": 17098, "epoch": 474, "lr": 0.001} +{"train_loss": 0.03457856240371863, "train_loss_bc": 0.031940385699272156, "train_loss_llm": 0.0, "grad_norm": 0.07386666536331177, "global_step": 17099, "epoch": 474, "lr": 0.001, "val_loss": 0.07373829185962677} +{"train_loss": 0.03855007886886597, "train_loss_bc": 0.038512781262397766, "train_loss_llm": 0.003729879856109619, "grad_norm": 0.09369300305843353, "global_step": 17100, "epoch": 475, "lr": 0.001} +{"train_loss": 0.025820240378379822, "train_loss_bc": 0.025820240378379822, "train_loss_llm": 0.0, "grad_norm": 0.023107757791876793, "global_step": 17101, "epoch": 475, "lr": 0.001} +{"train_loss": 0.029661646112799644, "train_loss_bc": 0.029661646112799644, "train_loss_llm": 0.0, "grad_norm": 0.03228442370891571, "global_step": 17102, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03552078455686569, "train_loss_bc": 0.03552078455686569, "train_loss_llm": 0.0, "grad_norm": 0.05518977716565132, "global_step": 17103, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03969161584973335, "train_loss_bc": 0.03969161584973335, "train_loss_llm": 0.0, "grad_norm": 0.07883903384208679, "global_step": 17104, "epoch": 475, "lr": 0.001} +{"train_loss": 0.029920371249318123, "train_loss_bc": 0.02990950271487236, "train_loss_llm": 0.0010868608951568604, "grad_norm": 0.022362645715475082, "global_step": 17105, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03508890047669411, "train_loss_bc": 0.03507024794816971, "train_loss_llm": 0.0018653571605682373, "grad_norm": 0.032608505338430405, "global_step": 17106, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03298236429691315, "train_loss_bc": 0.03298236429691315, "train_loss_llm": 0.0, "grad_norm": 0.04269573092460632, "global_step": 17107, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03632392734289169, "train_loss_bc": 0.036312468349933624, "train_loss_llm": 0.001145780086517334, "grad_norm": 0.05109791085124016, "global_step": 17108, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03784194216132164, "train_loss_bc": 0.03784194216132164, "train_loss_llm": 0.0, "grad_norm": 0.023081816732883453, "global_step": 17109, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03398832678794861, "train_loss_bc": 0.03398832678794861, "train_loss_llm": 0.0, "grad_norm": 0.033030424267053604, "global_step": 17110, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03506385535001755, "train_loss_bc": 0.03506385535001755, "train_loss_llm": 0.0, "grad_norm": 0.04325452446937561, "global_step": 17111, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03572976961731911, "train_loss_bc": 0.03572976961731911, "train_loss_llm": 0.0, "grad_norm": 0.056448329240083694, "global_step": 17112, "epoch": 475, "lr": 0.001} +{"train_loss": 0.0377500057220459, "train_loss_bc": 0.03773162141442299, "train_loss_llm": 0.0018384456634521484, "grad_norm": 0.020841063931584358, "global_step": 17113, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03769198805093765, "train_loss_bc": 0.037663206458091736, "train_loss_llm": 0.002878338098526001, "grad_norm": 0.03564099222421646, "global_step": 17114, "epoch": 475, "lr": 0.001} +{"train_loss": 0.037546779960393906, "train_loss_bc": 0.037546779960393906, "train_loss_llm": 0.0, "grad_norm": 0.048856042325496674, "global_step": 17115, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03155626356601715, "train_loss_bc": 0.031539130955934525, "train_loss_llm": 0.0017132759094238281, "grad_norm": 0.052017249166965485, "global_step": 17116, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03261260688304901, "train_loss_bc": 0.03261260688304901, "train_loss_llm": 0.0, "grad_norm": 0.020069565623998642, "global_step": 17117, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03314835950732231, "train_loss_bc": 0.03314835950732231, "train_loss_llm": 0.0, "grad_norm": 0.03640575706958771, "global_step": 17118, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03520333394408226, "train_loss_bc": 0.03520333394408226, "train_loss_llm": 0.0, "grad_norm": 0.04349980503320694, "global_step": 17119, "epoch": 475, "lr": 0.001} +{"train_loss": 0.04231080785393715, "train_loss_bc": 0.04231080785393715, "train_loss_llm": 0.0, "grad_norm": 0.05560611933469772, "global_step": 17120, "epoch": 475, "lr": 0.001} +{"train_loss": 0.029102176427841187, "train_loss_bc": 0.029102176427841187, "train_loss_llm": 0.0, "grad_norm": 0.02428571507334709, "global_step": 17121, "epoch": 475, "lr": 0.001} +{"train_loss": 0.033975984901189804, "train_loss_bc": 0.03397490456700325, "train_loss_llm": 0.00010785460472106934, "grad_norm": 0.042517829686403275, "global_step": 17122, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03709254413843155, "train_loss_bc": 0.03709254413843155, "train_loss_llm": 0.0, "grad_norm": 0.05775370076298714, "global_step": 17123, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03314533084630966, "train_loss_bc": 0.03314533084630966, "train_loss_llm": 0.0, "grad_norm": 0.07071070373058319, "global_step": 17124, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03313922882080078, "train_loss_bc": 0.03313922882080078, "train_loss_llm": 0.0, "grad_norm": 0.026262249797582626, "global_step": 17125, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03511390462517738, "train_loss_bc": 0.03511390462517738, "train_loss_llm": 0.0, "grad_norm": 0.044511813670396805, "global_step": 17126, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03371299058198929, "train_loss_bc": 0.03371299058198929, "train_loss_llm": 0.0, "grad_norm": 0.060609519481658936, "global_step": 17127, "epoch": 475, "lr": 0.001} +{"train_loss": 0.034579820930957794, "train_loss_bc": 0.034579820930957794, "train_loss_llm": 0.0, "grad_norm": 0.07699952274560928, "global_step": 17128, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03970431163907051, "train_loss_bc": 0.03970431163907051, "train_loss_llm": 0.0, "grad_norm": 0.028429757803678513, "global_step": 17129, "epoch": 475, "lr": 0.001} +{"train_loss": 0.028774822130799294, "train_loss_bc": 0.028774822130799294, "train_loss_llm": 0.0, "grad_norm": 0.04294655844569206, "global_step": 17130, "epoch": 475, "lr": 0.001} +{"train_loss": 0.0347682423889637, "train_loss_bc": 0.0347682423889637, "train_loss_llm": 0.0, "grad_norm": 0.062250085175037384, "global_step": 17131, "epoch": 475, "lr": 0.001} +{"train_loss": 0.031180860474705696, "train_loss_bc": 0.031180860474705696, "train_loss_llm": 0.0, "grad_norm": 0.08094154298305511, "global_step": 17132, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03413998335599899, "train_loss_bc": 0.03413998335599899, "train_loss_llm": 0.0, "grad_norm": 0.028673334047198296, "global_step": 17133, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03209437057375908, "train_loss_bc": 0.032079990953207016, "train_loss_llm": 0.0014380216598510742, "grad_norm": 0.03818449750542641, "global_step": 17134, "epoch": 475, "lr": 0.001} +{"train_loss": 0.03448101753989855, "train_loss_bc": 0.03678809106349945, "train_loss_llm": 0.0, "grad_norm": 0.04405636712908745, "global_step": 17135, "epoch": 475, "lr": 0.001, "train/cumulative_reward": 141.48995738823953, "train/mean_score": 16.841610491645667, "train/success_rate": 0.4, "test/cumulative_reward": 47.256769823163275, "test/mean_score": 9.043791921776796, "test/success_rate": 0.22, "val_loss": 0.07314369082450867, "train_action_mse_error": 0.03709027171134949} +{"train_loss": 0.03832825645804405, "train_loss_bc": 0.03832318261265755, "train_loss_llm": 0.0005072951316833496, "grad_norm": 0.047490473836660385, "global_step": 17136, "epoch": 476, "lr": 0.001} +{"train_loss": 0.031897444278001785, "train_loss_bc": 0.031897444278001785, "train_loss_llm": 0.0, "grad_norm": 0.028492139652371407, "global_step": 17137, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03480173274874687, "train_loss_bc": 0.03480173274874687, "train_loss_llm": 0.0, "grad_norm": 0.03497718274593353, "global_step": 17138, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03573942556977272, "train_loss_bc": 0.03573942556977272, "train_loss_llm": 0.0, "grad_norm": 0.042434968054294586, "global_step": 17139, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03488963842391968, "train_loss_bc": 0.03488963842391968, "train_loss_llm": 0.0, "grad_norm": 0.05461343005299568, "global_step": 17140, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03620503097772598, "train_loss_bc": 0.03620503097772598, "train_loss_llm": 0.0, "grad_norm": 0.03160523995757103, "global_step": 17141, "epoch": 476, "lr": 0.001} +{"train_loss": 0.04113739728927612, "train_loss_bc": 0.04113739728927612, "train_loss_llm": 0.0, "grad_norm": 0.04589433595538139, "global_step": 17142, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03606129810214043, "train_loss_bc": 0.03606129810214043, "train_loss_llm": 0.0, "grad_norm": 0.060784805566072464, "global_step": 17143, "epoch": 476, "lr": 0.001} +{"train_loss": 0.0387415736913681, "train_loss_bc": 0.0387415736913681, "train_loss_llm": 0.0, "grad_norm": 0.0858331099152565, "global_step": 17144, "epoch": 476, "lr": 0.001} +{"train_loss": 0.030800847336649895, "train_loss_bc": 0.030800847336649895, "train_loss_llm": 0.0, "grad_norm": 0.03232545033097267, "global_step": 17145, "epoch": 476, "lr": 0.001} +{"train_loss": 0.034028805792331696, "train_loss_bc": 0.034004196524620056, "train_loss_llm": 0.002461075782775879, "grad_norm": 0.05067393556237221, "global_step": 17146, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03397972136735916, "train_loss_bc": 0.03396992012858391, "train_loss_llm": 0.000979989767074585, "grad_norm": 0.06597165763378143, "global_step": 17147, "epoch": 476, "lr": 0.001} +{"train_loss": 0.033498961478471756, "train_loss_bc": 0.033498961478471756, "train_loss_llm": 0.0, "grad_norm": 0.07455074042081833, "global_step": 17148, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03418746590614319, "train_loss_bc": 0.03418536111712456, "train_loss_llm": 0.0002104043960571289, "grad_norm": 0.023427758365869522, "global_step": 17149, "epoch": 476, "lr": 0.001} +{"train_loss": 0.040157340466976166, "train_loss_bc": 0.040157340466976166, "train_loss_llm": 0.0, "grad_norm": 0.03858036920428276, "global_step": 17150, "epoch": 476, "lr": 0.001} +{"train_loss": 0.039591845124959946, "train_loss_bc": 0.039591845124959946, "train_loss_llm": 0.0, "grad_norm": 0.04839479923248291, "global_step": 17151, "epoch": 476, "lr": 0.001} +{"train_loss": 0.02864435315132141, "train_loss_bc": 0.02864435315132141, "train_loss_llm": 0.0, "grad_norm": 0.05503743514418602, "global_step": 17152, "epoch": 476, "lr": 0.001} +{"train_loss": 0.0334603413939476, "train_loss_bc": 0.0334603413939476, "train_loss_llm": 0.0, "grad_norm": 0.025912966579198837, "global_step": 17153, "epoch": 476, "lr": 0.001} +{"train_loss": 0.0313166081905365, "train_loss_bc": 0.0313166081905365, "train_loss_llm": 0.0, "grad_norm": 0.04983540624380112, "global_step": 17154, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03223124518990517, "train_loss_bc": 0.03223124518990517, "train_loss_llm": 0.0, "grad_norm": 0.06734797358512878, "global_step": 17155, "epoch": 476, "lr": 0.001} +{"train_loss": 0.02558300644159317, "train_loss_bc": 0.025577716529369354, "train_loss_llm": 0.0005290210247039795, "grad_norm": 0.08576475828886032, "global_step": 17156, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03483041748404503, "train_loss_bc": 0.034827638417482376, "train_loss_llm": 0.0002778470516204834, "grad_norm": 0.021302714943885803, "global_step": 17157, "epoch": 476, "lr": 0.001} +{"train_loss": 0.0338413268327713, "train_loss_bc": 0.0338413268327713, "train_loss_llm": 0.0, "grad_norm": 0.03404649347066879, "global_step": 17158, "epoch": 476, "lr": 0.001} +{"train_loss": 0.036189962178468704, "train_loss_bc": 0.036189962178468704, "train_loss_llm": 0.0, "grad_norm": 0.043453626334667206, "global_step": 17159, "epoch": 476, "lr": 0.001} +{"train_loss": 0.031244494020938873, "train_loss_bc": 0.031244494020938873, "train_loss_llm": 0.0, "grad_norm": 0.0552801676094532, "global_step": 17160, "epoch": 476, "lr": 0.001} +{"train_loss": 0.033951252698898315, "train_loss_bc": 0.033951252698898315, "train_loss_llm": 0.0, "grad_norm": 0.03138298541307449, "global_step": 17161, "epoch": 476, "lr": 0.001} +{"train_loss": 0.040430132299661636, "train_loss_bc": 0.040430132299661636, "train_loss_llm": 0.0, "grad_norm": 0.0536385215818882, "global_step": 17162, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03740563616156578, "train_loss_bc": 0.03740563616156578, "train_loss_llm": 0.0, "grad_norm": 0.0634976476430893, "global_step": 17163, "epoch": 476, "lr": 0.001} +{"train_loss": 0.0317867174744606, "train_loss_bc": 0.03175833821296692, "train_loss_llm": 0.0028380155563354492, "grad_norm": 0.07670046389102936, "global_step": 17164, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03228052705526352, "train_loss_bc": 0.032267238944768906, "train_loss_llm": 0.0013288259506225586, "grad_norm": 0.030980272218585014, "global_step": 17165, "epoch": 476, "lr": 0.001} +{"train_loss": 0.035439491271972656, "train_loss_bc": 0.035439491271972656, "train_loss_llm": 0.0, "grad_norm": 0.05141288414597511, "global_step": 17166, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03546302393078804, "train_loss_bc": 0.03546302393078804, "train_loss_llm": 0.0, "grad_norm": 0.06698868423700333, "global_step": 17167, "epoch": 476, "lr": 0.001} +{"train_loss": 0.037673745304346085, "train_loss_bc": 0.03766978532075882, "train_loss_llm": 0.0003961622714996338, "grad_norm": 0.075065478682518, "global_step": 17168, "epoch": 476, "lr": 0.001} +{"train_loss": 0.026949789375066757, "train_loss_bc": 0.026949789375066757, "train_loss_llm": 0.0, "grad_norm": 0.028066108003258705, "global_step": 17169, "epoch": 476, "lr": 0.001} +{"train_loss": 0.030800320208072662, "train_loss_bc": 0.030800320208072662, "train_loss_llm": 0.0, "grad_norm": 0.049746524542570114, "global_step": 17170, "epoch": 476, "lr": 0.001} +{"train_loss": 0.03457633349009686, "train_loss_bc": 0.041178829967975616, "train_loss_llm": 0.0, "grad_norm": 0.06467233598232269, "global_step": 17171, "epoch": 476, "lr": 0.001, "val_loss": 0.07386413216590881} +{"train_loss": 0.03437528759241104, "train_loss_bc": 0.03437528759241104, "train_loss_llm": 0.0, "grad_norm": 0.07746384292840958, "global_step": 17172, "epoch": 477, "lr": 0.001} +{"train_loss": 0.0336637906730175, "train_loss_bc": 0.0336637906730175, "train_loss_llm": 0.0, "grad_norm": 0.02171442285180092, "global_step": 17173, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03392278403043747, "train_loss_bc": 0.03392278403043747, "train_loss_llm": 0.0, "grad_norm": 0.039330076426267624, "global_step": 17174, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03830868750810623, "train_loss_bc": 0.03830868750810623, "train_loss_llm": 0.0, "grad_norm": 0.05560572072863579, "global_step": 17175, "epoch": 477, "lr": 0.001} +{"train_loss": 0.033279143273830414, "train_loss_bc": 0.033279143273830414, "train_loss_llm": 0.0, "grad_norm": 0.07310216128826141, "global_step": 17176, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03442058712244034, "train_loss_bc": 0.03442058712244034, "train_loss_llm": 0.0, "grad_norm": 0.028677020221948624, "global_step": 17177, "epoch": 477, "lr": 0.001} +{"train_loss": 0.030196810141205788, "train_loss_bc": 0.030196810141205788, "train_loss_llm": 0.0, "grad_norm": 0.04284204915165901, "global_step": 17178, "epoch": 477, "lr": 0.001} +{"train_loss": 0.042383261024951935, "train_loss_bc": 0.042383261024951935, "train_loss_llm": 0.0, "grad_norm": 0.06484712660312653, "global_step": 17179, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03689878061413765, "train_loss_bc": 0.03689878061413765, "train_loss_llm": 0.0, "grad_norm": 0.07748255878686905, "global_step": 17180, "epoch": 477, "lr": 0.001} +{"train_loss": 0.033632002770900726, "train_loss_bc": 0.033605124801397324, "train_loss_llm": 0.002687782049179077, "grad_norm": 0.03196531534194946, "global_step": 17181, "epoch": 477, "lr": 0.001} +{"train_loss": 0.04077925905585289, "train_loss_bc": 0.04077925905585289, "train_loss_llm": 0.0, "grad_norm": 0.04156648740172386, "global_step": 17182, "epoch": 477, "lr": 0.001} +{"train_loss": 0.032669153064489365, "train_loss_bc": 0.03264448046684265, "train_loss_llm": 0.0024671852588653564, "grad_norm": 0.05393698811531067, "global_step": 17183, "epoch": 477, "lr": 0.001} +{"train_loss": 0.029225686565041542, "train_loss_bc": 0.029225686565041542, "train_loss_llm": 0.0, "grad_norm": 0.06457305699586868, "global_step": 17184, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03342018276453018, "train_loss_bc": 0.03342018276453018, "train_loss_llm": 0.0, "grad_norm": 0.0365554541349411, "global_step": 17185, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03426150977611542, "train_loss_bc": 0.03426150977611542, "train_loss_llm": 0.0, "grad_norm": 0.053295351564884186, "global_step": 17186, "epoch": 477, "lr": 0.001} +{"train_loss": 0.0370008610188961, "train_loss_bc": 0.03698720410466194, "train_loss_llm": 0.0013656020164489746, "grad_norm": 0.0755324736237526, "global_step": 17187, "epoch": 477, "lr": 0.001} +{"train_loss": 0.026667127385735512, "train_loss_bc": 0.02666671946644783, "train_loss_llm": 4.07099723815918e-05, "grad_norm": 0.08668183535337448, "global_step": 17188, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03650147467851639, "train_loss_bc": 0.03650017827749252, "train_loss_llm": 0.00012949109077453613, "grad_norm": 0.029612598940730095, "global_step": 17189, "epoch": 477, "lr": 0.001} +{"train_loss": 0.030907709151506424, "train_loss_bc": 0.030907709151506424, "train_loss_llm": 0.0, "grad_norm": 0.04317383095622063, "global_step": 17190, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03707444295287132, "train_loss_bc": 0.03707444295287132, "train_loss_llm": 0.0, "grad_norm": 0.06913983821868896, "global_step": 17191, "epoch": 477, "lr": 0.001} +{"train_loss": 0.034211110323667526, "train_loss_bc": 0.03419480472803116, "train_loss_llm": 0.0016304850578308105, "grad_norm": 0.09118581563234329, "global_step": 17192, "epoch": 477, "lr": 0.001} +{"train_loss": 0.038697805255651474, "train_loss_bc": 0.038697805255651474, "train_loss_llm": 0.0, "grad_norm": 0.03331881761550903, "global_step": 17193, "epoch": 477, "lr": 0.001} +{"train_loss": 0.0329475998878479, "train_loss_bc": 0.0329475998878479, "train_loss_llm": 0.0, "grad_norm": 0.049512866884469986, "global_step": 17194, "epoch": 477, "lr": 0.001} +{"train_loss": 0.039087601006031036, "train_loss_bc": 0.039087601006031036, "train_loss_llm": 0.0, "grad_norm": 0.061750151216983795, "global_step": 17195, "epoch": 477, "lr": 0.001} +{"train_loss": 0.035542458295822144, "train_loss_bc": 0.035542458295822144, "train_loss_llm": 0.0, "grad_norm": 0.07886100560426712, "global_step": 17196, "epoch": 477, "lr": 0.001} +{"train_loss": 0.033336132764816284, "train_loss_bc": 0.033336132764816284, "train_loss_llm": 0.0, "grad_norm": 0.02364824339747429, "global_step": 17197, "epoch": 477, "lr": 0.001} +{"train_loss": 0.028997983783483505, "train_loss_bc": 0.028997482731938362, "train_loss_llm": 5.015730857849121e-05, "grad_norm": 0.03663291037082672, "global_step": 17198, "epoch": 477, "lr": 0.001} +{"train_loss": 0.042513057589530945, "train_loss_bc": 0.042505282908678055, "train_loss_llm": 0.0007775425910949707, "grad_norm": 0.05777105689048767, "global_step": 17199, "epoch": 477, "lr": 0.001} +{"train_loss": 0.031937289983034134, "train_loss_bc": 0.031937289983034134, "train_loss_llm": 0.0, "grad_norm": 0.07161809504032135, "global_step": 17200, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03739886358380318, "train_loss_bc": 0.03739886358380318, "train_loss_llm": 0.0, "grad_norm": 0.03338882699608803, "global_step": 17201, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03322119265794754, "train_loss_bc": 0.03322119265794754, "train_loss_llm": 0.0, "grad_norm": 0.051662031561136246, "global_step": 17202, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03312055766582489, "train_loss_bc": 0.03310956060886383, "train_loss_llm": 0.0010995566844940186, "grad_norm": 0.07665248215198517, "global_step": 17203, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03323078900575638, "train_loss_bc": 0.03322981670498848, "train_loss_llm": 9.721517562866211e-05, "grad_norm": 0.09287025779485703, "global_step": 17204, "epoch": 477, "lr": 0.001} +{"train_loss": 0.03916983678936958, "train_loss_bc": 0.03916983678936958, "train_loss_llm": 0.0, "grad_norm": 0.03197433054447174, "global_step": 17205, "epoch": 477, "lr": 0.001} +{"train_loss": 0.0296005979180336, "train_loss_bc": 0.0296005979180336, "train_loss_llm": 0.0, "grad_norm": 0.0506538525223732, "global_step": 17206, "epoch": 477, "lr": 0.001} +{"train_loss": 0.034746812811742224, "train_loss_bc": 0.03828384354710579, "train_loss_llm": 0.0, "grad_norm": 0.06032775714993477, "global_step": 17207, "epoch": 477, "lr": 0.001, "val_loss": 0.07356104254722595} +{"train_loss": 0.03402773663401604, "train_loss_bc": 0.03402773663401604, "train_loss_llm": 0.0, "grad_norm": 0.07306770980358124, "global_step": 17208, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03476172313094139, "train_loss_bc": 0.03476172313094139, "train_loss_llm": 0.0, "grad_norm": 0.02335946448147297, "global_step": 17209, "epoch": 478, "lr": 0.001} +{"train_loss": 0.034075718373060226, "train_loss_bc": 0.034071821719408035, "train_loss_llm": 0.0003898143768310547, "grad_norm": 0.036211349070072174, "global_step": 17210, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03513939306139946, "train_loss_bc": 0.03513769432902336, "train_loss_llm": 0.00016996264457702637, "grad_norm": 0.05414614826440811, "global_step": 17211, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03082672506570816, "train_loss_bc": 0.03082672506570816, "train_loss_llm": 0.0, "grad_norm": 0.06338096410036087, "global_step": 17212, "epoch": 478, "lr": 0.001} +{"train_loss": 0.034554608166217804, "train_loss_bc": 0.03454690799117088, "train_loss_llm": 0.0007699728012084961, "grad_norm": 0.024041293188929558, "global_step": 17213, "epoch": 478, "lr": 0.001} +{"train_loss": 0.035054564476013184, "train_loss_bc": 0.035054564476013184, "train_loss_llm": 0.0, "grad_norm": 0.040464017540216446, "global_step": 17214, "epoch": 478, "lr": 0.001} +{"train_loss": 0.026863612234592438, "train_loss_bc": 0.026863612234592438, "train_loss_llm": 0.0, "grad_norm": 0.05432083457708359, "global_step": 17215, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03837808966636658, "train_loss_bc": 0.03837808966636658, "train_loss_llm": 0.0, "grad_norm": 0.08801007270812988, "global_step": 17216, "epoch": 478, "lr": 0.001} +{"train_loss": 0.02986123412847519, "train_loss_bc": 0.02986123412847519, "train_loss_llm": 0.0, "grad_norm": 0.023741891607642174, "global_step": 17217, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03167056664824486, "train_loss_bc": 0.03167056664824486, "train_loss_llm": 0.0, "grad_norm": 0.05032246932387352, "global_step": 17218, "epoch": 478, "lr": 0.001} +{"train_loss": 0.030670879408717155, "train_loss_bc": 0.0306547824293375, "train_loss_llm": 0.0016096532344818115, "grad_norm": 0.06836660206317902, "global_step": 17219, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03875891864299774, "train_loss_bc": 0.03875715658068657, "train_loss_llm": 0.0001761913299560547, "grad_norm": 0.09461882710456848, "global_step": 17220, "epoch": 478, "lr": 0.001} +{"train_loss": 0.032724443823099136, "train_loss_bc": 0.03271879628300667, "train_loss_llm": 0.0005646347999572754, "grad_norm": 0.029776884242892265, "global_step": 17221, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03537013381719589, "train_loss_bc": 0.03535783290863037, "train_loss_llm": 0.0012301206588745117, "grad_norm": 0.04402877762913704, "global_step": 17222, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03293411061167717, "train_loss_bc": 0.03293411061167717, "train_loss_llm": 0.0, "grad_norm": 0.05839962884783745, "global_step": 17223, "epoch": 478, "lr": 0.001} +{"train_loss": 0.033115193247795105, "train_loss_bc": 0.03309396281838417, "train_loss_llm": 0.0021231770515441895, "grad_norm": 0.07361973077058792, "global_step": 17224, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03300710394978523, "train_loss_bc": 0.03300710394978523, "train_loss_llm": 0.0, "grad_norm": 0.028790654614567757, "global_step": 17225, "epoch": 478, "lr": 0.001} +{"train_loss": 0.0283501036465168, "train_loss_bc": 0.028343016281723976, "train_loss_llm": 0.0007087588310241699, "grad_norm": 0.04502226039767265, "global_step": 17226, "epoch": 478, "lr": 0.001} +{"train_loss": 0.037100858986377716, "train_loss_bc": 0.037100858986377716, "train_loss_llm": 0.0, "grad_norm": 0.060247279703617096, "global_step": 17227, "epoch": 478, "lr": 0.001} +{"train_loss": 0.037227924913167953, "train_loss_bc": 0.037227924913167953, "train_loss_llm": 0.0, "grad_norm": 0.08202103525400162, "global_step": 17228, "epoch": 478, "lr": 0.001} +{"train_loss": 0.035036008805036545, "train_loss_bc": 0.035036008805036545, "train_loss_llm": 0.0, "grad_norm": 0.02776731550693512, "global_step": 17229, "epoch": 478, "lr": 0.001} +{"train_loss": 0.035547465085983276, "train_loss_bc": 0.035547465085983276, "train_loss_llm": 0.0, "grad_norm": 0.05488431826233864, "global_step": 17230, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03771190345287323, "train_loss_bc": 0.03770933300256729, "train_loss_llm": 0.00025704503059387207, "grad_norm": 0.07329579442739487, "global_step": 17231, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03166802227497101, "train_loss_bc": 0.03166802227497101, "train_loss_llm": 0.0, "grad_norm": 0.08871719241142273, "global_step": 17232, "epoch": 478, "lr": 0.001} +{"train_loss": 0.035247351974248886, "train_loss_bc": 0.035245757550001144, "train_loss_llm": 0.00015944242477416992, "grad_norm": 0.021399762481451035, "global_step": 17233, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03409494832158089, "train_loss_bc": 0.03407244756817818, "train_loss_llm": 0.0022500157356262207, "grad_norm": 0.04330723360180855, "global_step": 17234, "epoch": 478, "lr": 0.001} +{"train_loss": 0.04183636233210564, "train_loss_bc": 0.04183636233210564, "train_loss_llm": 0.0, "grad_norm": 0.07214513421058655, "global_step": 17235, "epoch": 478, "lr": 0.001} +{"train_loss": 0.030229900032281876, "train_loss_bc": 0.030229900032281876, "train_loss_llm": 0.0, "grad_norm": 0.10063810646533966, "global_step": 17236, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03761716187000275, "train_loss_bc": 0.03761716187000275, "train_loss_llm": 0.0, "grad_norm": 0.028076494112610817, "global_step": 17237, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03444003686308861, "train_loss_bc": 0.03440986946225166, "train_loss_llm": 0.003016561269760132, "grad_norm": 0.039094071835279465, "global_step": 17238, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03884454816579819, "train_loss_bc": 0.03884454816579819, "train_loss_llm": 0.0, "grad_norm": 0.06338705867528915, "global_step": 17239, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03864521160721779, "train_loss_bc": 0.03864521160721779, "train_loss_llm": 0.0, "grad_norm": 0.08501874655485153, "global_step": 17240, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03534786403179169, "train_loss_bc": 0.03534786403179169, "train_loss_llm": 0.0, "grad_norm": 0.03007168136537075, "global_step": 17241, "epoch": 478, "lr": 0.001} +{"train_loss": 0.039239536970853806, "train_loss_bc": 0.039239536970853806, "train_loss_llm": 0.0, "grad_norm": 0.056386854499578476, "global_step": 17242, "epoch": 478, "lr": 0.001} +{"train_loss": 0.03496814420860675, "train_loss_bc": 0.04887322708964348, "train_loss_llm": 0.0, "grad_norm": 0.07364857941865921, "global_step": 17243, "epoch": 478, "lr": 0.001, "val_loss": 0.07201244682073593} +{"train_loss": 0.03674469515681267, "train_loss_bc": 0.03674469515681267, "train_loss_llm": 0.0, "grad_norm": 0.08909367769956589, "global_step": 17244, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03459147736430168, "train_loss_bc": 0.034571148455142975, "train_loss_llm": 0.002032756805419922, "grad_norm": 0.03145317733287811, "global_step": 17245, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03638105466961861, "train_loss_bc": 0.03638105466961861, "train_loss_llm": 0.0, "grad_norm": 0.05640990287065506, "global_step": 17246, "epoch": 479, "lr": 0.001} +{"train_loss": 0.030790776014328003, "train_loss_bc": 0.030790776014328003, "train_loss_llm": 0.0, "grad_norm": 0.07125100493431091, "global_step": 17247, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03061707131564617, "train_loss_bc": 0.03061707131564617, "train_loss_llm": 0.0, "grad_norm": 0.07982432097196579, "global_step": 17248, "epoch": 479, "lr": 0.001} +{"train_loss": 0.033242687582969666, "train_loss_bc": 0.03320861607789993, "train_loss_llm": 0.003407210111618042, "grad_norm": 0.03281343728303909, "global_step": 17249, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03503287956118584, "train_loss_bc": 0.03503287956118584, "train_loss_llm": 0.0, "grad_norm": 0.05270093306899071, "global_step": 17250, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03505432605743408, "train_loss_bc": 0.03505304455757141, "train_loss_llm": 0.00012809038162231445, "grad_norm": 0.07269526273012161, "global_step": 17251, "epoch": 479, "lr": 0.001} +{"train_loss": 0.029929006472229958, "train_loss_bc": 0.02992851845920086, "train_loss_llm": 4.8786401748657227e-05, "grad_norm": 0.0844898447394371, "global_step": 17252, "epoch": 479, "lr": 0.001} +{"train_loss": 0.027846289798617363, "train_loss_bc": 0.027846289798617363, "train_loss_llm": 0.0, "grad_norm": 0.022444961592555046, "global_step": 17253, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03235064074397087, "train_loss_bc": 0.03235064074397087, "train_loss_llm": 0.0, "grad_norm": 0.04634140059351921, "global_step": 17254, "epoch": 479, "lr": 0.001} +{"train_loss": 0.031927019357681274, "train_loss_bc": 0.031927019357681274, "train_loss_llm": 0.0, "grad_norm": 0.06528980284929276, "global_step": 17255, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03216938301920891, "train_loss_bc": 0.03216938301920891, "train_loss_llm": 0.0, "grad_norm": 0.08324706554412842, "global_step": 17256, "epoch": 479, "lr": 0.001} +{"train_loss": 0.038625966757535934, "train_loss_bc": 0.038625966757535934, "train_loss_llm": 0.0, "grad_norm": 0.032156649976968765, "global_step": 17257, "epoch": 479, "lr": 0.001} +{"train_loss": 0.036764614284038544, "train_loss_bc": 0.036764614284038544, "train_loss_llm": 0.0, "grad_norm": 0.05297001451253891, "global_step": 17258, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03928451985120773, "train_loss_bc": 0.039274487644433975, "train_loss_llm": 0.0010033845901489258, "grad_norm": 0.0741347074508667, "global_step": 17259, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03927048668265343, "train_loss_bc": 0.03927048668265343, "train_loss_llm": 0.0, "grad_norm": 0.0918995589017868, "global_step": 17260, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03660431131720543, "train_loss_bc": 0.03660431131720543, "train_loss_llm": 0.0, "grad_norm": 0.03272799775004387, "global_step": 17261, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03658420220017433, "train_loss_bc": 0.03658420220017433, "train_loss_llm": 0.0, "grad_norm": 0.05213034525513649, "global_step": 17262, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03324836120009422, "train_loss_bc": 0.03324836120009422, "train_loss_llm": 0.0, "grad_norm": 0.06631138175725937, "global_step": 17263, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03708036243915558, "train_loss_bc": 0.03708036243915558, "train_loss_llm": 0.0, "grad_norm": 0.08509968221187592, "global_step": 17264, "epoch": 479, "lr": 0.001} +{"train_loss": 0.040206506848335266, "train_loss_bc": 0.040206506848335266, "train_loss_llm": 0.0, "grad_norm": 0.029369795694947243, "global_step": 17265, "epoch": 479, "lr": 0.001} +{"train_loss": 0.04025528207421303, "train_loss_bc": 0.04025528207421303, "train_loss_llm": 0.0, "grad_norm": 0.045906227082014084, "global_step": 17266, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03578430041670799, "train_loss_bc": 0.03578430041670799, "train_loss_llm": 0.0, "grad_norm": 0.06004355847835541, "global_step": 17267, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03177362680435181, "train_loss_bc": 0.03174610435962677, "train_loss_llm": 0.0027522146701812744, "grad_norm": 0.07626978307962418, "global_step": 17268, "epoch": 479, "lr": 0.001} +{"train_loss": 0.037961557507514954, "train_loss_bc": 0.037961557507514954, "train_loss_llm": 0.0, "grad_norm": 0.030075807124376297, "global_step": 17269, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03145105391740799, "train_loss_bc": 0.031440701335668564, "train_loss_llm": 0.0010353624820709229, "grad_norm": 0.0495864599943161, "global_step": 17270, "epoch": 479, "lr": 0.001} +{"train_loss": 0.033804770559072495, "train_loss_bc": 0.03379638493061066, "train_loss_llm": 0.0008386075496673584, "grad_norm": 0.06211698800325394, "global_step": 17271, "epoch": 479, "lr": 0.001} +{"train_loss": 0.02819085493683815, "train_loss_bc": 0.02819085493683815, "train_loss_llm": 0.0, "grad_norm": 0.07520930469036102, "global_step": 17272, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03895887732505798, "train_loss_bc": 0.03895887732505798, "train_loss_llm": 0.0, "grad_norm": 0.03842075914144516, "global_step": 17273, "epoch": 479, "lr": 0.001} +{"train_loss": 0.036284543573856354, "train_loss_bc": 0.036284543573856354, "train_loss_llm": 0.0, "grad_norm": 0.05927136167883873, "global_step": 17274, "epoch": 479, "lr": 0.001} +{"train_loss": 0.039833422750234604, "train_loss_bc": 0.039833422750234604, "train_loss_llm": 0.0, "grad_norm": 0.08071041107177734, "global_step": 17275, "epoch": 479, "lr": 0.001} +{"train_loss": 0.04228121414780617, "train_loss_bc": 0.04228121414780617, "train_loss_llm": 0.0, "grad_norm": 0.09814807027578354, "global_step": 17276, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03126116469502449, "train_loss_bc": 0.03126116469502449, "train_loss_llm": 0.0, "grad_norm": 0.034224096685647964, "global_step": 17277, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03332185745239258, "train_loss_bc": 0.03332185745239258, "train_loss_llm": 0.0, "grad_norm": 0.04906381666660309, "global_step": 17278, "epoch": 479, "lr": 0.001} +{"train_loss": 0.03491315074885885, "train_loss_bc": 0.031364262104034424, "train_loss_llm": 0.0, "grad_norm": 0.06814530491828918, "global_step": 17279, "epoch": 479, "lr": 0.001, "val_loss": 0.0754176452755928} +{"train_loss": 0.035294532775878906, "train_loss_bc": 0.035294532775878906, "train_loss_llm": 0.0, "grad_norm": 0.08021283149719238, "global_step": 17280, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03771453723311424, "train_loss_bc": 0.03771453723311424, "train_loss_llm": 0.0, "grad_norm": 0.031965143978595734, "global_step": 17281, "epoch": 480, "lr": 0.001} +{"train_loss": 0.029136575758457184, "train_loss_bc": 0.029136575758457184, "train_loss_llm": 0.0, "grad_norm": 0.0395578108727932, "global_step": 17282, "epoch": 480, "lr": 0.001} +{"train_loss": 0.028602391481399536, "train_loss_bc": 0.028602391481399536, "train_loss_llm": 0.0, "grad_norm": 0.05046087130904198, "global_step": 17283, "epoch": 480, "lr": 0.001} +{"train_loss": 0.035709235817193985, "train_loss_bc": 0.035709235817193985, "train_loss_llm": 0.0, "grad_norm": 0.06428007036447525, "global_step": 17284, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03505650907754898, "train_loss_bc": 0.03505650907754898, "train_loss_llm": 0.0, "grad_norm": 0.025447750464081764, "global_step": 17285, "epoch": 480, "lr": 0.001} +{"train_loss": 0.037719421088695526, "train_loss_bc": 0.037719421088695526, "train_loss_llm": 0.0, "grad_norm": 0.0391656719148159, "global_step": 17286, "epoch": 480, "lr": 0.001} +{"train_loss": 0.026490019634366035, "train_loss_bc": 0.026489628478884697, "train_loss_llm": 3.9070844650268555e-05, "grad_norm": 0.04656944423913956, "global_step": 17287, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03600910305976868, "train_loss_bc": 0.03600910305976868, "train_loss_llm": 0.0, "grad_norm": 0.06597797572612762, "global_step": 17288, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03860577195882797, "train_loss_bc": 0.03860577195882797, "train_loss_llm": 0.0, "grad_norm": 0.028549594804644585, "global_step": 17289, "epoch": 480, "lr": 0.001} +{"train_loss": 0.035412441939115524, "train_loss_bc": 0.035412441939115524, "train_loss_llm": 0.0, "grad_norm": 0.052044738084077835, "global_step": 17290, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03417757526040077, "train_loss_bc": 0.03417757526040077, "train_loss_llm": 0.0, "grad_norm": 0.0681748315691948, "global_step": 17291, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03005441278219223, "train_loss_bc": 0.03005441278219223, "train_loss_llm": 0.0, "grad_norm": 0.08081945031881332, "global_step": 17292, "epoch": 480, "lr": 0.001} +{"train_loss": 0.034327756613492966, "train_loss_bc": 0.034323710948228836, "train_loss_llm": 0.0004044771194458008, "grad_norm": 0.03515235707163811, "global_step": 17293, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03332683816552162, "train_loss_bc": 0.03332683816552162, "train_loss_llm": 0.0, "grad_norm": 0.0472969152033329, "global_step": 17294, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03399328142404556, "train_loss_bc": 0.0339658260345459, "train_loss_llm": 0.002745389938354492, "grad_norm": 0.0550231896340847, "global_step": 17295, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03421255573630333, "train_loss_bc": 0.03420353680849075, "train_loss_llm": 0.0009020566940307617, "grad_norm": 0.07095025479793549, "global_step": 17296, "epoch": 480, "lr": 0.001} +{"train_loss": 0.0334034338593483, "train_loss_bc": 0.033388651907444, "train_loss_llm": 0.0014781355857849121, "grad_norm": 0.03019716776907444, "global_step": 17297, "epoch": 480, "lr": 0.001} +{"train_loss": 0.039121996611356735, "train_loss_bc": 0.039120569825172424, "train_loss_llm": 0.00014266371726989746, "grad_norm": 0.050088461488485336, "global_step": 17298, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03699259087443352, "train_loss_bc": 0.03699259087443352, "train_loss_llm": 0.0, "grad_norm": 0.06964311003684998, "global_step": 17299, "epoch": 480, "lr": 0.001} +{"train_loss": 0.0324665866792202, "train_loss_bc": 0.03245850279927254, "train_loss_llm": 0.0008084774017333984, "grad_norm": 0.08550553768873215, "global_step": 17300, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03400392085313797, "train_loss_bc": 0.03400392085313797, "train_loss_llm": 0.0, "grad_norm": 0.029590025544166565, "global_step": 17301, "epoch": 480, "lr": 0.001} +{"train_loss": 0.035248178988695145, "train_loss_bc": 0.035248178988695145, "train_loss_llm": 0.0, "grad_norm": 0.04429188370704651, "global_step": 17302, "epoch": 480, "lr": 0.001} +{"train_loss": 0.0393960177898407, "train_loss_bc": 0.0393960177898407, "train_loss_llm": 0.0, "grad_norm": 0.06707506626844406, "global_step": 17303, "epoch": 480, "lr": 0.001} +{"train_loss": 0.031058909371495247, "train_loss_bc": 0.0310433991253376, "train_loss_llm": 0.0015510916709899902, "grad_norm": 0.07823137938976288, "global_step": 17304, "epoch": 480, "lr": 0.001} +{"train_loss": 0.040109969675540924, "train_loss_bc": 0.040109969675540924, "train_loss_llm": 0.0, "grad_norm": 0.025440918281674385, "global_step": 17305, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03436385840177536, "train_loss_bc": 0.034347087144851685, "train_loss_llm": 0.001677095890045166, "grad_norm": 0.039459969848394394, "global_step": 17306, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03730032965540886, "train_loss_bc": 0.03729573264718056, "train_loss_llm": 0.0004596412181854248, "grad_norm": 0.05181950330734253, "global_step": 17307, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03420105576515198, "train_loss_bc": 0.034200944006443024, "train_loss_llm": 1.1116266250610352e-05, "grad_norm": 0.07327403128147125, "global_step": 17308, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03537466377019882, "train_loss_bc": 0.03535802662372589, "train_loss_llm": 0.0016638338565826416, "grad_norm": 0.02524866722524166, "global_step": 17309, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03815195709466934, "train_loss_bc": 0.03815195709466934, "train_loss_llm": 0.0, "grad_norm": 0.04569930583238602, "global_step": 17310, "epoch": 480, "lr": 0.001} +{"train_loss": 0.039212971925735474, "train_loss_bc": 0.039212971925735474, "train_loss_llm": 0.0, "grad_norm": 0.06712078303098679, "global_step": 17311, "epoch": 480, "lr": 0.001} +{"train_loss": 0.03681414946913719, "train_loss_bc": 0.03681414946913719, "train_loss_llm": 0.0, "grad_norm": 0.07671039551496506, "global_step": 17312, "epoch": 480, "lr": 0.001} +{"train_loss": 0.037673767656087875, "train_loss_bc": 0.037673767656087875, "train_loss_llm": 0.0, "grad_norm": 0.029309524223208427, "global_step": 17313, "epoch": 480, "lr": 0.001} +{"train_loss": 0.02956153266131878, "train_loss_bc": 0.02956153266131878, "train_loss_llm": 0.0, "grad_norm": 0.04262302815914154, "global_step": 17314, "epoch": 480, "lr": 0.001} +{"train_loss": 0.034614679093162216, "train_loss_bc": 0.02582959644496441, "train_loss_llm": 0.0, "grad_norm": 0.05879301577806473, "global_step": 17315, "epoch": 480, "lr": 0.001, "train/cumulative_reward": 71.0162245667535, "train/mean_score": 20.840330096549593, "train/success_rate": 0.6, "test/cumulative_reward": 34.11278616724781, "test/mean_score": 9.396470157637571, "test/success_rate": 0.24, "val_loss": 0.07315823435783386, "train_action_mse_error": 0.038014721125364304} +{"train_loss": 0.0406171977519989, "train_loss_bc": 0.0406171977519989, "train_loss_llm": 0.0, "grad_norm": 0.08013859391212463, "global_step": 17316, "epoch": 481, "lr": 0.001} +{"train_loss": 0.02883126027882099, "train_loss_bc": 0.02883126027882099, "train_loss_llm": 0.0, "grad_norm": 0.02386842854321003, "global_step": 17317, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03733354061841965, "train_loss_bc": 0.03733354061841965, "train_loss_llm": 0.0, "grad_norm": 0.045670390129089355, "global_step": 17318, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03369294852018356, "train_loss_bc": 0.03369294852018356, "train_loss_llm": 0.0, "grad_norm": 0.05195474997162819, "global_step": 17319, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03859862685203552, "train_loss_bc": 0.03859862685203552, "train_loss_llm": 0.0, "grad_norm": 0.06632887572050095, "global_step": 17320, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03466479107737541, "train_loss_bc": 0.03466479107737541, "train_loss_llm": 0.0, "grad_norm": 0.031094172969460487, "global_step": 17321, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03793764486908913, "train_loss_bc": 0.03793764486908913, "train_loss_llm": 0.0, "grad_norm": 0.039949510246515274, "global_step": 17322, "epoch": 481, "lr": 0.001} +{"train_loss": 0.02932034060359001, "train_loss_bc": 0.02932034060359001, "train_loss_llm": 0.0, "grad_norm": 0.05443935841321945, "global_step": 17323, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03351479396224022, "train_loss_bc": 0.033514656126499176, "train_loss_llm": 1.3768672943115234e-05, "grad_norm": 0.06527454406023026, "global_step": 17324, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03221695125102997, "train_loss_bc": 0.03221695125102997, "train_loss_llm": 0.0, "grad_norm": 0.022499248385429382, "global_step": 17325, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03755161538720131, "train_loss_bc": 0.03753480315208435, "train_loss_llm": 0.001681208610534668, "grad_norm": 0.03949544578790665, "global_step": 17326, "epoch": 481, "lr": 0.001} +{"train_loss": 0.028879662975668907, "train_loss_bc": 0.028879662975668907, "train_loss_llm": 0.0, "grad_norm": 0.052736494690179825, "global_step": 17327, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03615190461277962, "train_loss_bc": 0.03615190461277962, "train_loss_llm": 0.0, "grad_norm": 0.05502179265022278, "global_step": 17328, "epoch": 481, "lr": 0.001} +{"train_loss": 0.036418844014406204, "train_loss_bc": 0.036418844014406204, "train_loss_llm": 0.0, "grad_norm": 0.022522062063217163, "global_step": 17329, "epoch": 481, "lr": 0.001} +{"train_loss": 0.037422064691782, "train_loss_bc": 0.037422064691782, "train_loss_llm": 0.0, "grad_norm": 0.03317851200699806, "global_step": 17330, "epoch": 481, "lr": 0.001} +{"train_loss": 0.030399931594729424, "train_loss_bc": 0.030399931594729424, "train_loss_llm": 0.0, "grad_norm": 0.05032384768128395, "global_step": 17331, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03273681551218033, "train_loss_bc": 0.03272605314850807, "train_loss_llm": 0.00107613205909729, "grad_norm": 0.05898546800017357, "global_step": 17332, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03219570219516754, "train_loss_bc": 0.03219570219516754, "train_loss_llm": 0.0, "grad_norm": 0.02475113235414028, "global_step": 17333, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03763125091791153, "train_loss_bc": 0.03763125091791153, "train_loss_llm": 0.0, "grad_norm": 0.03570916876196861, "global_step": 17334, "epoch": 481, "lr": 0.001} +{"train_loss": 0.04218693822622299, "train_loss_bc": 0.04218693822622299, "train_loss_llm": 0.0, "grad_norm": 0.05527326837182045, "global_step": 17335, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03168681263923645, "train_loss_bc": 0.03168681263923645, "train_loss_llm": 0.0, "grad_norm": 0.0677417516708374, "global_step": 17336, "epoch": 481, "lr": 0.001} +{"train_loss": 0.026665810495615005, "train_loss_bc": 0.02662462368607521, "train_loss_llm": 0.0041186511516571045, "grad_norm": 0.02236153744161129, "global_step": 17337, "epoch": 481, "lr": 0.001} +{"train_loss": 0.030526582151651382, "train_loss_bc": 0.030526582151651382, "train_loss_llm": 0.0, "grad_norm": 0.03165711089968681, "global_step": 17338, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03338206931948662, "train_loss_bc": 0.03338206931948662, "train_loss_llm": 0.0, "grad_norm": 0.04599060118198395, "global_step": 17339, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03082878701388836, "train_loss_bc": 0.03082878701388836, "train_loss_llm": 0.0, "grad_norm": 0.05782155692577362, "global_step": 17340, "epoch": 481, "lr": 0.001} +{"train_loss": 0.0363713875412941, "train_loss_bc": 0.0363713875412941, "train_loss_llm": 0.0, "grad_norm": 0.025312410667538643, "global_step": 17341, "epoch": 481, "lr": 0.001} +{"train_loss": 0.032086633145809174, "train_loss_bc": 0.03208252787590027, "train_loss_llm": 0.00041049718856811523, "grad_norm": 0.044109445065259933, "global_step": 17342, "epoch": 481, "lr": 0.001} +{"train_loss": 0.038587622344493866, "train_loss_bc": 0.038580331951379776, "train_loss_llm": 0.0007291734218597412, "grad_norm": 0.05206454545259476, "global_step": 17343, "epoch": 481, "lr": 0.001} +{"train_loss": 0.041554734110832214, "train_loss_bc": 0.041554734110832214, "train_loss_llm": 0.0, "grad_norm": 0.060915734618902206, "global_step": 17344, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03686951473355293, "train_loss_bc": 0.03686951473355293, "train_loss_llm": 0.0, "grad_norm": 0.028825407847762108, "global_step": 17345, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03460923209786415, "train_loss_bc": 0.03460923209786415, "train_loss_llm": 0.0, "grad_norm": 0.04731117561459541, "global_step": 17346, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03311657905578613, "train_loss_bc": 0.03309231251478195, "train_loss_llm": 0.00242653489112854, "grad_norm": 0.06772167235612869, "global_step": 17347, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03304705768823624, "train_loss_bc": 0.03304705768823624, "train_loss_llm": 0.0, "grad_norm": 0.07729798555374146, "global_step": 17348, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03186959773302078, "train_loss_bc": 0.03186959773302078, "train_loss_llm": 0.0, "grad_norm": 0.02914203144609928, "global_step": 17349, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03665405884385109, "train_loss_bc": 0.03665405884385109, "train_loss_llm": 0.0, "grad_norm": 0.04032858833670616, "global_step": 17350, "epoch": 481, "lr": 0.001} +{"train_loss": 0.03444692544225189, "train_loss_bc": 0.03392871841788292, "train_loss_llm": 0.0001291334629058838, "grad_norm": 0.05411205068230629, "global_step": 17351, "epoch": 481, "lr": 0.001, "val_loss": 0.07321840524673462} +{"train_loss": 0.0365779809653759, "train_loss_bc": 0.0365779809653759, "train_loss_llm": 0.0, "grad_norm": 0.06531794369220734, "global_step": 17352, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03475406393408775, "train_loss_bc": 0.03474939987063408, "train_loss_llm": 0.00046622753143310547, "grad_norm": 0.01992480829358101, "global_step": 17353, "epoch": 482, "lr": 0.001} +{"train_loss": 0.031906288117170334, "train_loss_bc": 0.03187672048807144, "train_loss_llm": 0.002956777811050415, "grad_norm": 0.029799936339259148, "global_step": 17354, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03780869022011757, "train_loss_bc": 0.03779672086238861, "train_loss_llm": 0.0011968612670898438, "grad_norm": 0.0366206169128418, "global_step": 17355, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03384851664304733, "train_loss_bc": 0.03384851664304733, "train_loss_llm": 0.0, "grad_norm": 0.046620748937129974, "global_step": 17356, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03254379704594612, "train_loss_bc": 0.03253205120563507, "train_loss_llm": 0.0011744201183319092, "grad_norm": 0.030427122488617897, "global_step": 17357, "epoch": 482, "lr": 0.001} +{"train_loss": 0.040202897042036057, "train_loss_bc": 0.040202897042036057, "train_loss_llm": 0.0, "grad_norm": 0.04112608730792999, "global_step": 17358, "epoch": 482, "lr": 0.001} +{"train_loss": 0.039696138352155685, "train_loss_bc": 0.039688389748334885, "train_loss_llm": 0.0007749795913696289, "grad_norm": 0.05482054501771927, "global_step": 17359, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03024480864405632, "train_loss_bc": 0.03024480864405632, "train_loss_llm": 0.0, "grad_norm": 0.06667092442512512, "global_step": 17360, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03339039534330368, "train_loss_bc": 0.03339039534330368, "train_loss_llm": 0.0, "grad_norm": 0.02807239443063736, "global_step": 17361, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03183157742023468, "train_loss_bc": 0.03183157742023468, "train_loss_llm": 0.0, "grad_norm": 0.05264592543244362, "global_step": 17362, "epoch": 482, "lr": 0.001} +{"train_loss": 0.031604617834091187, "train_loss_bc": 0.031604617834091187, "train_loss_llm": 0.0, "grad_norm": 0.0737801119685173, "global_step": 17363, "epoch": 482, "lr": 0.001} +{"train_loss": 0.0397660918533802, "train_loss_bc": 0.0397660918533802, "train_loss_llm": 0.0, "grad_norm": 0.0913173258304596, "global_step": 17364, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03373568132519722, "train_loss_bc": 0.03373568132519722, "train_loss_llm": 0.0, "grad_norm": 0.03270499408245087, "global_step": 17365, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03603966906666756, "train_loss_bc": 0.03603966906666756, "train_loss_llm": 0.0, "grad_norm": 0.046542394906282425, "global_step": 17366, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03592098876833916, "train_loss_bc": 0.03592098876833916, "train_loss_llm": 0.0, "grad_norm": 0.06445392221212387, "global_step": 17367, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03314092010259628, "train_loss_bc": 0.03314092010259628, "train_loss_llm": 0.0, "grad_norm": 0.07642392814159393, "global_step": 17368, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03379319608211517, "train_loss_bc": 0.03379319608211517, "train_loss_llm": 0.0, "grad_norm": 0.022407708689570427, "global_step": 17369, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03405633196234703, "train_loss_bc": 0.03405633196234703, "train_loss_llm": 0.0, "grad_norm": 0.046013910323381424, "global_step": 17370, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03047185391187668, "train_loss_bc": 0.03047185391187668, "train_loss_llm": 0.0, "grad_norm": 0.06292826682329178, "global_step": 17371, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03733835369348526, "train_loss_bc": 0.03733612224459648, "train_loss_llm": 0.0002231299877166748, "grad_norm": 0.06889658421278, "global_step": 17372, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03676836937665939, "train_loss_bc": 0.03676836937665939, "train_loss_llm": 0.0, "grad_norm": 0.03475832939147949, "global_step": 17373, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03002908267080784, "train_loss_bc": 0.03002908267080784, "train_loss_llm": 0.0, "grad_norm": 0.047895923256874084, "global_step": 17374, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03472190722823143, "train_loss_bc": 0.03472190722823143, "train_loss_llm": 0.0, "grad_norm": 0.05672381445765495, "global_step": 17375, "epoch": 482, "lr": 0.001} +{"train_loss": 0.043401651084423065, "train_loss_bc": 0.0433834008872509, "train_loss_llm": 0.0018251240253448486, "grad_norm": 0.0736711174249649, "global_step": 17376, "epoch": 482, "lr": 0.001} +{"train_loss": 0.0341658815741539, "train_loss_bc": 0.034162357449531555, "train_loss_llm": 0.00035241246223449707, "grad_norm": 0.030811872333288193, "global_step": 17377, "epoch": 482, "lr": 0.001} +{"train_loss": 0.032262176275253296, "train_loss_bc": 0.03226032108068466, "train_loss_llm": 0.00018545985221862793, "grad_norm": 0.04679014906287193, "global_step": 17378, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03747506067156792, "train_loss_bc": 0.037441082298755646, "train_loss_llm": 0.0033979713916778564, "grad_norm": 0.07041706889867783, "global_step": 17379, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03341960534453392, "train_loss_bc": 0.03341960534453392, "train_loss_llm": 0.0, "grad_norm": 0.08048606663942337, "global_step": 17380, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03521118685603142, "train_loss_bc": 0.03521118685603142, "train_loss_llm": 0.0, "grad_norm": 0.02652336098253727, "global_step": 17381, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03453098237514496, "train_loss_bc": 0.03453098237514496, "train_loss_llm": 0.0, "grad_norm": 0.05026029422879219, "global_step": 17382, "epoch": 482, "lr": 0.001} +{"train_loss": 0.030470577999949455, "train_loss_bc": 0.030470577999949455, "train_loss_llm": 0.0, "grad_norm": 0.05989881232380867, "global_step": 17383, "epoch": 482, "lr": 0.001} +{"train_loss": 0.024769164621829987, "train_loss_bc": 0.024769164621829987, "train_loss_llm": 0.0, "grad_norm": 0.07006102800369263, "global_step": 17384, "epoch": 482, "lr": 0.001} +{"train_loss": 0.041608862578868866, "train_loss_bc": 0.041608862578868866, "train_loss_llm": 0.0, "grad_norm": 0.03403420373797417, "global_step": 17385, "epoch": 482, "lr": 0.001} +{"train_loss": 0.03628795966506004, "train_loss_bc": 0.03628795966506004, "train_loss_llm": 0.0, "grad_norm": 0.041424695402383804, "global_step": 17386, "epoch": 482, "lr": 0.001} +{"train_loss": 0.034477395109004445, "train_loss_bc": 0.027386901900172234, "train_loss_llm": 0.00039950013160705566, "grad_norm": 0.05906571075320244, "global_step": 17387, "epoch": 482, "lr": 0.001, "val_loss": 0.07197773456573486} +{"train_loss": 0.031128013506531715, "train_loss_bc": 0.031111102551221848, "train_loss_llm": 0.0016910731792449951, "grad_norm": 0.07383688539266586, "global_step": 17388, "epoch": 483, "lr": 0.001} +{"train_loss": 0.035132139921188354, "train_loss_bc": 0.035132139921188354, "train_loss_llm": 0.0, "grad_norm": 0.0287662073969841, "global_step": 17389, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03754798695445061, "train_loss_bc": 0.037534017115831375, "train_loss_llm": 0.00139695405960083, "grad_norm": 0.04314301535487175, "global_step": 17390, "epoch": 483, "lr": 0.001} +{"train_loss": 0.0347822867333889, "train_loss_bc": 0.03478081524372101, "train_loss_llm": 0.00014704465866088867, "grad_norm": 0.05087362602353096, "global_step": 17391, "epoch": 483, "lr": 0.001} +{"train_loss": 0.033234480768442154, "train_loss_bc": 0.03321672976016998, "train_loss_llm": 0.0017752647399902344, "grad_norm": 0.05651027336716652, "global_step": 17392, "epoch": 483, "lr": 0.001} +{"train_loss": 0.035390809178352356, "train_loss_bc": 0.035390809178352356, "train_loss_llm": 0.0, "grad_norm": 0.028552528470754623, "global_step": 17393, "epoch": 483, "lr": 0.001} +{"train_loss": 0.027886686846613884, "train_loss_bc": 0.027886686846613884, "train_loss_llm": 0.0, "grad_norm": 0.04010561481118202, "global_step": 17394, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03844108060002327, "train_loss_bc": 0.038420408964157104, "train_loss_llm": 0.002067267894744873, "grad_norm": 0.05268694460391998, "global_step": 17395, "epoch": 483, "lr": 0.001} +{"train_loss": 0.02980547957122326, "train_loss_bc": 0.02980547957122326, "train_loss_llm": 0.0, "grad_norm": 0.06757944077253342, "global_step": 17396, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03837328776717186, "train_loss_bc": 0.03837328776717186, "train_loss_llm": 0.0, "grad_norm": 0.03495754301548004, "global_step": 17397, "epoch": 483, "lr": 0.001} +{"train_loss": 0.029346823692321777, "train_loss_bc": 0.0293368399143219, "train_loss_llm": 0.000998377799987793, "grad_norm": 0.055905427783727646, "global_step": 17398, "epoch": 483, "lr": 0.001} +{"train_loss": 0.031991977244615555, "train_loss_bc": 0.03198564052581787, "train_loss_llm": 0.0006336569786071777, "grad_norm": 0.07548099011182785, "global_step": 17399, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03269152715802193, "train_loss_bc": 0.03269152715802193, "train_loss_llm": 0.0, "grad_norm": 0.09568627178668976, "global_step": 17400, "epoch": 483, "lr": 0.001} +{"train_loss": 0.034710921347141266, "train_loss_bc": 0.03469431400299072, "train_loss_llm": 0.001660764217376709, "grad_norm": 0.027083275839686394, "global_step": 17401, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03369263932108879, "train_loss_bc": 0.03369263932108879, "train_loss_llm": 0.0, "grad_norm": 0.04447410628199577, "global_step": 17402, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03262714296579361, "train_loss_bc": 0.03260703757405281, "train_loss_llm": 0.0020104646682739258, "grad_norm": 0.06736045330762863, "global_step": 17403, "epoch": 483, "lr": 0.001} +{"train_loss": 0.035561252385377884, "train_loss_bc": 0.035561252385377884, "train_loss_llm": 0.0, "grad_norm": 0.08207656443119049, "global_step": 17404, "epoch": 483, "lr": 0.001} +{"train_loss": 0.04204143211245537, "train_loss_bc": 0.04204143211245537, "train_loss_llm": 0.0, "grad_norm": 0.029838841408491135, "global_step": 17405, "epoch": 483, "lr": 0.001} +{"train_loss": 0.04082971811294556, "train_loss_bc": 0.04082971811294556, "train_loss_llm": 0.0, "grad_norm": 0.04412570595741272, "global_step": 17406, "epoch": 483, "lr": 0.001} +{"train_loss": 0.030459728091955185, "train_loss_bc": 0.030459728091955185, "train_loss_llm": 0.0, "grad_norm": 0.053715288639068604, "global_step": 17407, "epoch": 483, "lr": 0.001} +{"train_loss": 0.030029520392417908, "train_loss_bc": 0.030029520392417908, "train_loss_llm": 0.0, "grad_norm": 0.07299742847681046, "global_step": 17408, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03497607260942459, "train_loss_bc": 0.03497607260942459, "train_loss_llm": 0.0, "grad_norm": 0.025845296680927277, "global_step": 17409, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03607458248734474, "train_loss_bc": 0.03607458248734474, "train_loss_llm": 0.0, "grad_norm": 0.04362891614437103, "global_step": 17410, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03621505945920944, "train_loss_bc": 0.03621505945920944, "train_loss_llm": 0.0, "grad_norm": 0.062237758189439774, "global_step": 17411, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03586287796497345, "train_loss_bc": 0.03586287796497345, "train_loss_llm": 0.0, "grad_norm": 0.07865633815526962, "global_step": 17412, "epoch": 483, "lr": 0.001} +{"train_loss": 0.027429379522800446, "train_loss_bc": 0.027429379522800446, "train_loss_llm": 0.0, "grad_norm": 0.031739406287670135, "global_step": 17413, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03588567674160004, "train_loss_bc": 0.03588567674160004, "train_loss_llm": 0.0, "grad_norm": 0.05940064415335655, "global_step": 17414, "epoch": 483, "lr": 0.001} +{"train_loss": 0.030027026310563087, "train_loss_bc": 0.030027026310563087, "train_loss_llm": 0.0, "grad_norm": 0.07480088621377945, "global_step": 17415, "epoch": 483, "lr": 0.001} +{"train_loss": 0.033802762627601624, "train_loss_bc": 0.033802762627601624, "train_loss_llm": 0.0, "grad_norm": 0.09049837291240692, "global_step": 17416, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03642318397760391, "train_loss_bc": 0.03642318397760391, "train_loss_llm": 0.0, "grad_norm": 0.027830123901367188, "global_step": 17417, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03767741098999977, "train_loss_bc": 0.03767741098999977, "train_loss_llm": 0.0, "grad_norm": 0.04246428608894348, "global_step": 17418, "epoch": 483, "lr": 0.001} +{"train_loss": 0.03965427726507187, "train_loss_bc": 0.03965136408805847, "train_loss_llm": 0.00029140710830688477, "grad_norm": 0.06270638853311539, "global_step": 17419, "epoch": 483, "lr": 0.001} +{"train_loss": 0.037724655121564865, "train_loss_bc": 0.03771834447979927, "train_loss_llm": 0.0006309449672698975, "grad_norm": 0.0754600465297699, "global_step": 17420, "epoch": 483, "lr": 0.001} +{"train_loss": 0.0359732061624527, "train_loss_bc": 0.035965777933597565, "train_loss_llm": 0.0007429718971252441, "grad_norm": 0.027333220466971397, "global_step": 17421, "epoch": 483, "lr": 0.001} +{"train_loss": 0.032437290996313095, "train_loss_bc": 0.032437290996313095, "train_loss_llm": 0.0, "grad_norm": 0.04806162789463997, "global_step": 17422, "epoch": 483, "lr": 0.001} +{"train_loss": 0.034617968214054905, "train_loss_bc": 0.04037845879793167, "train_loss_llm": 0.0, "grad_norm": 0.06045254319906235, "global_step": 17423, "epoch": 483, "lr": 0.001, "val_loss": 0.07404632866382599} +{"train_loss": 0.02913856692612171, "train_loss_bc": 0.02913856692612171, "train_loss_llm": 0.0, "grad_norm": 0.06989502906799316, "global_step": 17424, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03674079477787018, "train_loss_bc": 0.03674079477787018, "train_loss_llm": 0.0, "grad_norm": 0.02393544651567936, "global_step": 17425, "epoch": 484, "lr": 0.001} +{"train_loss": 0.031923435628414154, "train_loss_bc": 0.031915999948978424, "train_loss_llm": 0.000743478536605835, "grad_norm": 0.035614460706710815, "global_step": 17426, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03165162727236748, "train_loss_bc": 0.03165162727236748, "train_loss_llm": 0.0, "grad_norm": 0.046628110110759735, "global_step": 17427, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03608798235654831, "train_loss_bc": 0.03608798235654831, "train_loss_llm": 0.0, "grad_norm": 0.05629816651344299, "global_step": 17428, "epoch": 484, "lr": 0.001} +{"train_loss": 0.036408547312021255, "train_loss_bc": 0.036408547312021255, "train_loss_llm": 0.0, "grad_norm": 0.025956710800528526, "global_step": 17429, "epoch": 484, "lr": 0.001} +{"train_loss": 0.02958899550139904, "train_loss_bc": 0.02958899550139904, "train_loss_llm": 0.0, "grad_norm": 0.03851970657706261, "global_step": 17430, "epoch": 484, "lr": 0.001} +{"train_loss": 0.023566320538520813, "train_loss_bc": 0.023566320538520813, "train_loss_llm": 0.0, "grad_norm": 0.044266264885663986, "global_step": 17431, "epoch": 484, "lr": 0.001} +{"train_loss": 0.030868016183376312, "train_loss_bc": 0.03085636906325817, "train_loss_llm": 0.0011647939682006836, "grad_norm": 0.05460360273718834, "global_step": 17432, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03476107120513916, "train_loss_bc": 0.03476107120513916, "train_loss_llm": 0.0, "grad_norm": 0.024859534576535225, "global_step": 17433, "epoch": 484, "lr": 0.001} +{"train_loss": 0.02960064820945263, "train_loss_bc": 0.02959059365093708, "train_loss_llm": 0.0010055303573608398, "grad_norm": 0.03525786101818085, "global_step": 17434, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03925817832350731, "train_loss_bc": 0.03925817832350731, "train_loss_llm": 0.0, "grad_norm": 0.05888327211141586, "global_step": 17435, "epoch": 484, "lr": 0.001} +{"train_loss": 0.036161456257104874, "train_loss_bc": 0.036161456257104874, "train_loss_llm": 0.0, "grad_norm": 0.07562660425901413, "global_step": 17436, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03337075561285019, "train_loss_bc": 0.03337075561285019, "train_loss_llm": 0.0, "grad_norm": 0.02417413517832756, "global_step": 17437, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03635353222489357, "train_loss_bc": 0.03635353222489357, "train_loss_llm": 0.0, "grad_norm": 0.040102407336235046, "global_step": 17438, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03967305272817612, "train_loss_bc": 0.03967305272817612, "train_loss_llm": 0.0, "grad_norm": 0.05973565950989723, "global_step": 17439, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03396932780742645, "train_loss_bc": 0.033968012779951096, "train_loss_llm": 0.0001316666603088379, "grad_norm": 0.07033727318048477, "global_step": 17440, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03712617605924606, "train_loss_bc": 0.03712617605924606, "train_loss_llm": 0.0, "grad_norm": 0.028410103172063828, "global_step": 17441, "epoch": 484, "lr": 0.001} +{"train_loss": 0.030399682000279427, "train_loss_bc": 0.030399682000279427, "train_loss_llm": 0.0, "grad_norm": 0.04422035068273544, "global_step": 17442, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03317805379629135, "train_loss_bc": 0.033173568546772, "train_loss_llm": 0.00044861435890197754, "grad_norm": 0.05576581880450249, "global_step": 17443, "epoch": 484, "lr": 0.001} +{"train_loss": 0.0339527502655983, "train_loss_bc": 0.0339527502655983, "train_loss_llm": 0.0, "grad_norm": 0.06794753670692444, "global_step": 17444, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03437509015202522, "train_loss_bc": 0.034371934831142426, "train_loss_llm": 0.00031557679176330566, "grad_norm": 0.031765926629304886, "global_step": 17445, "epoch": 484, "lr": 0.001} +{"train_loss": 0.033907294273376465, "train_loss_bc": 0.03389780968427658, "train_loss_llm": 0.0009486377239227295, "grad_norm": 0.037818800657987595, "global_step": 17446, "epoch": 484, "lr": 0.001} +{"train_loss": 0.035340771079063416, "train_loss_bc": 0.035340771079063416, "train_loss_llm": 0.0, "grad_norm": 0.043453916907310486, "global_step": 17447, "epoch": 484, "lr": 0.001} +{"train_loss": 0.036938365548849106, "train_loss_bc": 0.036938365548849106, "train_loss_llm": 0.0, "grad_norm": 0.050869956612586975, "global_step": 17448, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03811294212937355, "train_loss_bc": 0.03811294212937355, "train_loss_llm": 0.0, "grad_norm": 0.02270541526377201, "global_step": 17449, "epoch": 484, "lr": 0.001} +{"train_loss": 0.0313602052628994, "train_loss_bc": 0.0313602052628994, "train_loss_llm": 0.0, "grad_norm": 0.03501748666167259, "global_step": 17450, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03304734453558922, "train_loss_bc": 0.03304734453558922, "train_loss_llm": 0.0, "grad_norm": 0.03882228583097458, "global_step": 17451, "epoch": 484, "lr": 0.001} +{"train_loss": 0.0381380058825016, "train_loss_bc": 0.0381380058825016, "train_loss_llm": 0.0, "grad_norm": 0.05547044426202774, "global_step": 17452, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03176048398017883, "train_loss_bc": 0.03172370418906212, "train_loss_llm": 0.003677964210510254, "grad_norm": 0.025924528017640114, "global_step": 17453, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03357676789164543, "train_loss_bc": 0.03357253223657608, "train_loss_llm": 0.0004234015941619873, "grad_norm": 0.03785515949130058, "global_step": 17454, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03707320988178253, "train_loss_bc": 0.0370643176138401, "train_loss_llm": 0.0008893907070159912, "grad_norm": 0.04825076833367348, "global_step": 17455, "epoch": 484, "lr": 0.001} +{"train_loss": 0.040702734142541885, "train_loss_bc": 0.040702734142541885, "train_loss_llm": 0.0, "grad_norm": 0.05889817699790001, "global_step": 17456, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03227311000227928, "train_loss_bc": 0.03227311000227928, "train_loss_llm": 0.0, "grad_norm": 0.027691707015037537, "global_step": 17457, "epoch": 484, "lr": 0.001} +{"train_loss": 0.04085212200880051, "train_loss_bc": 0.04084227979183197, "train_loss_llm": 0.0009844005107879639, "grad_norm": 0.0472477450966835, "global_step": 17458, "epoch": 484, "lr": 0.001} +{"train_loss": 0.03423819194237391, "train_loss_bc": 0.031337492167949677, "train_loss_llm": 0.0, "grad_norm": 0.06617943197488785, "global_step": 17459, "epoch": 484, "lr": 0.001, "val_loss": 0.07348404824733734} +{"train_loss": 0.02829079143702984, "train_loss_bc": 0.028284037485718727, "train_loss_llm": 0.0006753504276275635, "grad_norm": 0.07412882894277573, "global_step": 17460, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03495427221059799, "train_loss_bc": 0.03495427221059799, "train_loss_llm": 0.0, "grad_norm": 0.027565306052565575, "global_step": 17461, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03202323243021965, "train_loss_bc": 0.03202323243021965, "train_loss_llm": 0.0, "grad_norm": 0.037216268479824066, "global_step": 17462, "epoch": 485, "lr": 0.001} +{"train_loss": 0.028383810073137283, "train_loss_bc": 0.02837459370493889, "train_loss_llm": 0.0009216666221618652, "grad_norm": 0.05070666968822479, "global_step": 17463, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03253580257296562, "train_loss_bc": 0.03253580257296562, "train_loss_llm": 0.0, "grad_norm": 0.07674326747655869, "global_step": 17464, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03367510810494423, "train_loss_bc": 0.03367510810494423, "train_loss_llm": 0.0, "grad_norm": 0.023793401196599007, "global_step": 17465, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03821973502635956, "train_loss_bc": 0.03821456804871559, "train_loss_llm": 0.0005166232585906982, "grad_norm": 0.03656047582626343, "global_step": 17466, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03509707376360893, "train_loss_bc": 0.03508838638663292, "train_loss_llm": 0.0008686482906341553, "grad_norm": 0.049806807190179825, "global_step": 17467, "epoch": 485, "lr": 0.001} +{"train_loss": 0.029732752591371536, "train_loss_bc": 0.029732752591371536, "train_loss_llm": 0.0, "grad_norm": 0.059402305632829666, "global_step": 17468, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03202827274799347, "train_loss_bc": 0.03202827274799347, "train_loss_llm": 0.0, "grad_norm": 0.024346880614757538, "global_step": 17469, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03238939121365547, "train_loss_bc": 0.03238072618842125, "train_loss_llm": 0.0008666515350341797, "grad_norm": 0.032627008855342865, "global_step": 17470, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03967515006661415, "train_loss_bc": 0.03967515006661415, "train_loss_llm": 0.0, "grad_norm": 0.04550789296627045, "global_step": 17471, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03415815532207489, "train_loss_bc": 0.03415815532207489, "train_loss_llm": 0.0, "grad_norm": 0.055403441190719604, "global_step": 17472, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03417357802391052, "train_loss_bc": 0.03415565565228462, "train_loss_llm": 0.0017921924591064453, "grad_norm": 0.021693997085094452, "global_step": 17473, "epoch": 485, "lr": 0.001} +{"train_loss": 0.0336725153028965, "train_loss_bc": 0.033668145537376404, "train_loss_llm": 0.00043696165084838867, "grad_norm": 0.03440054878592491, "global_step": 17474, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03832784667611122, "train_loss_bc": 0.03832784667611122, "train_loss_llm": 0.0, "grad_norm": 0.05079555884003639, "global_step": 17475, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03083571419119835, "train_loss_bc": 0.030827533453702927, "train_loss_llm": 0.0008181631565093994, "grad_norm": 0.06545041501522064, "global_step": 17476, "epoch": 485, "lr": 0.001} +{"train_loss": 0.032245486974716187, "train_loss_bc": 0.032245486974716187, "train_loss_llm": 0.0, "grad_norm": 0.026495125144720078, "global_step": 17477, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03708206117153168, "train_loss_bc": 0.03708206117153168, "train_loss_llm": 0.0, "grad_norm": 0.04404651001095772, "global_step": 17478, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03178415820002556, "train_loss_bc": 0.03178415820002556, "train_loss_llm": 0.0, "grad_norm": 0.05616046115756035, "global_step": 17479, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03831638768315315, "train_loss_bc": 0.03831638768315315, "train_loss_llm": 0.0, "grad_norm": 0.0662679448723793, "global_step": 17480, "epoch": 485, "lr": 0.001} +{"train_loss": 0.032922014594078064, "train_loss_bc": 0.032922014594078064, "train_loss_llm": 0.0, "grad_norm": 0.02491798996925354, "global_step": 17481, "epoch": 485, "lr": 0.001} +{"train_loss": 0.033208221197128296, "train_loss_bc": 0.033208221197128296, "train_loss_llm": 0.0, "grad_norm": 0.0402134545147419, "global_step": 17482, "epoch": 485, "lr": 0.001} +{"train_loss": 0.034151580184698105, "train_loss_bc": 0.034146178513765335, "train_loss_llm": 0.0005401670932769775, "grad_norm": 0.057351402938365936, "global_step": 17483, "epoch": 485, "lr": 0.001} +{"train_loss": 0.04288297891616821, "train_loss_bc": 0.042879872024059296, "train_loss_llm": 0.00031057000160217285, "grad_norm": 0.06734734773635864, "global_step": 17484, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03890689089894295, "train_loss_bc": 0.03890689089894295, "train_loss_llm": 0.0, "grad_norm": 0.02947826310992241, "global_step": 17485, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03411272540688515, "train_loss_bc": 0.03411272540688515, "train_loss_llm": 0.0, "grad_norm": 0.03772805258631706, "global_step": 17486, "epoch": 485, "lr": 0.001} +{"train_loss": 0.0312960147857666, "train_loss_bc": 0.031291455030441284, "train_loss_llm": 0.000456005334854126, "grad_norm": 0.048603687435388565, "global_step": 17487, "epoch": 485, "lr": 0.001} +{"train_loss": 0.027440326288342476, "train_loss_bc": 0.027440326288342476, "train_loss_llm": 0.0, "grad_norm": 0.053773216903209686, "global_step": 17488, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03114449605345726, "train_loss_bc": 0.03114290162920952, "train_loss_llm": 0.00015944242477416992, "grad_norm": 0.023559367284178734, "global_step": 17489, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03868899494409561, "train_loss_bc": 0.03868744522333145, "train_loss_llm": 0.00015485286712646484, "grad_norm": 0.04107722267508507, "global_step": 17490, "epoch": 485, "lr": 0.001} +{"train_loss": 0.035839181393384933, "train_loss_bc": 0.035826414823532104, "train_loss_llm": 0.0012767910957336426, "grad_norm": 0.05516999214887619, "global_step": 17491, "epoch": 485, "lr": 0.001} +{"train_loss": 0.036939144134521484, "train_loss_bc": 0.03693057596683502, "train_loss_llm": 0.0008569955825805664, "grad_norm": 0.061033736914396286, "global_step": 17492, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03975676745176315, "train_loss_bc": 0.03975676745176315, "train_loss_llm": 0.0, "grad_norm": 0.027003278955817223, "global_step": 17493, "epoch": 485, "lr": 0.001} +{"train_loss": 0.02989804558455944, "train_loss_bc": 0.02989804558455944, "train_loss_llm": 0.0, "grad_norm": 0.03970802575349808, "global_step": 17494, "epoch": 485, "lr": 0.001} +{"train_loss": 0.03432807776456078, "train_loss_bc": 0.0410120002925396, "train_loss_llm": 0.0010122060775756836, "grad_norm": 0.054689712822437286, "global_step": 17495, "epoch": 485, "lr": 0.001, "train/cumulative_reward": 83.31730998318957, "train/mean_score": 10.810951016886818, "train/success_rate": 0.2, "test/cumulative_reward": 39.74938043551988, "test/mean_score": 6.863932545321814, "test/success_rate": 0.14, "val_loss": 0.07397043704986572, "train_action_mse_error": 0.037496116012334824} +{"train_loss": 0.03460868448019028, "train_loss_bc": 0.034598883241415024, "train_loss_llm": 0.0009801685810089111, "grad_norm": 0.06557289510965347, "global_step": 17496, "epoch": 486, "lr": 0.001} +{"train_loss": 0.035566896200180054, "train_loss_bc": 0.035566896200180054, "train_loss_llm": 0.0, "grad_norm": 0.03106950782239437, "global_step": 17497, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03528226912021637, "train_loss_bc": 0.03528226912021637, "train_loss_llm": 0.0, "grad_norm": 0.04933685064315796, "global_step": 17498, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03179149702191353, "train_loss_bc": 0.03179149702191353, "train_loss_llm": 0.0, "grad_norm": 0.06562313437461853, "global_step": 17499, "epoch": 486, "lr": 0.001} +{"train_loss": 0.030876733362674713, "train_loss_bc": 0.030876733362674713, "train_loss_llm": 0.0, "grad_norm": 0.06843824684619904, "global_step": 17500, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03728654980659485, "train_loss_bc": 0.03728654980659485, "train_loss_llm": 0.0, "grad_norm": 0.029108401387929916, "global_step": 17501, "epoch": 486, "lr": 0.001} +{"train_loss": 0.030237320810556412, "train_loss_bc": 0.030237320810556412, "train_loss_llm": 0.0, "grad_norm": 0.04063847288489342, "global_step": 17502, "epoch": 486, "lr": 0.001} +{"train_loss": 0.036373864859342575, "train_loss_bc": 0.036373864859342575, "train_loss_llm": 0.0, "grad_norm": 0.051699474453926086, "global_step": 17503, "epoch": 486, "lr": 0.001} +{"train_loss": 0.039268963038921356, "train_loss_bc": 0.039268963038921356, "train_loss_llm": 0.0, "grad_norm": 0.06369338929653168, "global_step": 17504, "epoch": 486, "lr": 0.001} +{"train_loss": 0.030718881636857986, "train_loss_bc": 0.030718881636857986, "train_loss_llm": 0.0, "grad_norm": 0.022807970643043518, "global_step": 17505, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03253173828125, "train_loss_bc": 0.03253173828125, "train_loss_llm": 0.0, "grad_norm": 0.03767254948616028, "global_step": 17506, "epoch": 486, "lr": 0.001} +{"train_loss": 0.030313963070511818, "train_loss_bc": 0.03028860315680504, "train_loss_llm": 0.0025359392166137695, "grad_norm": 0.046778950840234756, "global_step": 17507, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03383397310972214, "train_loss_bc": 0.03383397310972214, "train_loss_llm": 0.0, "grad_norm": 0.06054788455367088, "global_step": 17508, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03399939090013504, "train_loss_bc": 0.03399939090013504, "train_loss_llm": 0.0, "grad_norm": 0.022067254409193993, "global_step": 17509, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03167922422289848, "train_loss_bc": 0.031661175191402435, "train_loss_llm": 0.0018049776554107666, "grad_norm": 0.028113972395658493, "global_step": 17510, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03923410922288895, "train_loss_bc": 0.03923410922288895, "train_loss_llm": 0.0, "grad_norm": 0.041848067194223404, "global_step": 17511, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03764592111110687, "train_loss_bc": 0.03764592111110687, "train_loss_llm": 0.0, "grad_norm": 0.05773622915148735, "global_step": 17512, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03239351883530617, "train_loss_bc": 0.03239351883530617, "train_loss_llm": 0.0, "grad_norm": 0.02323860488831997, "global_step": 17513, "epoch": 486, "lr": 0.001} +{"train_loss": 0.036125097423791885, "train_loss_bc": 0.036125097423791885, "train_loss_llm": 0.0, "grad_norm": 0.03658728674054146, "global_step": 17514, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03361682593822479, "train_loss_bc": 0.03361682593822479, "train_loss_llm": 0.0, "grad_norm": 0.051753588020801544, "global_step": 17515, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03084302321076393, "train_loss_bc": 0.03084302321076393, "train_loss_llm": 0.0, "grad_norm": 0.053894203156232834, "global_step": 17516, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03439392149448395, "train_loss_bc": 0.03439392149448395, "train_loss_llm": 0.0, "grad_norm": 0.03658994659781456, "global_step": 17517, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03153732791543007, "train_loss_bc": 0.031518422067165375, "train_loss_llm": 0.001890629529953003, "grad_norm": 0.05405198410153389, "global_step": 17518, "epoch": 486, "lr": 0.001} +{"train_loss": 0.02754589170217514, "train_loss_bc": 0.027541108429431915, "train_loss_llm": 0.00047832727432250977, "grad_norm": 0.06133665144443512, "global_step": 17519, "epoch": 486, "lr": 0.001} +{"train_loss": 0.036869969218969345, "train_loss_bc": 0.036858730018138885, "train_loss_llm": 0.0011239945888519287, "grad_norm": 0.06793782114982605, "global_step": 17520, "epoch": 486, "lr": 0.001} +{"train_loss": 0.035464026033878326, "train_loss_bc": 0.03545992821455002, "train_loss_llm": 0.00040984153747558594, "grad_norm": 0.03204916790127754, "global_step": 17521, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03991886228322983, "train_loss_bc": 0.03991886228322983, "train_loss_llm": 0.0, "grad_norm": 0.04759904369711876, "global_step": 17522, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03377358615398407, "train_loss_bc": 0.03376935422420502, "train_loss_llm": 0.0004233121871948242, "grad_norm": 0.05668579787015915, "global_step": 17523, "epoch": 486, "lr": 0.001} +{"train_loss": 0.028481343761086464, "train_loss_bc": 0.028481343761086464, "train_loss_llm": 0.0, "grad_norm": 0.06412499397993088, "global_step": 17524, "epoch": 486, "lr": 0.001} +{"train_loss": 0.038331493735313416, "train_loss_bc": 0.038331493735313416, "train_loss_llm": 0.0, "grad_norm": 0.02380734123289585, "global_step": 17525, "epoch": 486, "lr": 0.001} +{"train_loss": 0.034466881304979324, "train_loss_bc": 0.0344647541642189, "train_loss_llm": 0.00021272897720336914, "grad_norm": 0.03814171999692917, "global_step": 17526, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03950844705104828, "train_loss_bc": 0.03950844705104828, "train_loss_llm": 0.0, "grad_norm": 0.04947113245725632, "global_step": 17527, "epoch": 486, "lr": 0.001} +{"train_loss": 0.032213788479566574, "train_loss_bc": 0.03221233934164047, "train_loss_llm": 0.00014480948448181152, "grad_norm": 0.05694398283958435, "global_step": 17528, "epoch": 486, "lr": 0.001} +{"train_loss": 0.030734656378626823, "train_loss_bc": 0.030734656378626823, "train_loss_llm": 0.0, "grad_norm": 0.022933028638362885, "global_step": 17529, "epoch": 486, "lr": 0.001} +{"train_loss": 0.04089666157960892, "train_loss_bc": 0.04089666157960892, "train_loss_llm": 0.0, "grad_norm": 0.03931356966495514, "global_step": 17530, "epoch": 486, "lr": 0.001} +{"train_loss": 0.03416273551475671, "train_loss_bc": 0.0314931757748127, "train_loss_llm": 0.0, "grad_norm": 0.047500599175691605, "global_step": 17531, "epoch": 486, "lr": 0.001, "val_loss": 0.07387124001979828} +{"train_loss": 0.04303600266575813, "train_loss_bc": 0.04303600266575813, "train_loss_llm": 0.0, "grad_norm": 0.06637626886367798, "global_step": 17532, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03843441978096962, "train_loss_bc": 0.03843441978096962, "train_loss_llm": 0.0, "grad_norm": 0.02697952464222908, "global_step": 17533, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03062431886792183, "train_loss_bc": 0.030608586966991425, "train_loss_llm": 0.001573115587234497, "grad_norm": 0.04323943331837654, "global_step": 17534, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03267231583595276, "train_loss_bc": 0.03267231583595276, "train_loss_llm": 0.0, "grad_norm": 0.05808805301785469, "global_step": 17535, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03559472784399986, "train_loss_bc": 0.03559472784399986, "train_loss_llm": 0.0, "grad_norm": 0.06849982589483261, "global_step": 17536, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03250448778271675, "train_loss_bc": 0.03250448778271675, "train_loss_llm": 0.0, "grad_norm": 0.02847732976078987, "global_step": 17537, "epoch": 487, "lr": 0.001} +{"train_loss": 0.029560301452875137, "train_loss_bc": 0.029554149135947227, "train_loss_llm": 0.0006152689456939697, "grad_norm": 0.0423944815993309, "global_step": 17538, "epoch": 487, "lr": 0.001} +{"train_loss": 0.02774430811405182, "train_loss_bc": 0.02774430811405182, "train_loss_llm": 0.0, "grad_norm": 0.05746643245220184, "global_step": 17539, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03415311127901077, "train_loss_bc": 0.03415311127901077, "train_loss_llm": 0.0, "grad_norm": 0.06676112860441208, "global_step": 17540, "epoch": 487, "lr": 0.001} +{"train_loss": 0.028365490958094597, "train_loss_bc": 0.028365490958094597, "train_loss_llm": 0.0, "grad_norm": 0.0237865149974823, "global_step": 17541, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03505917638540268, "train_loss_bc": 0.03505917638540268, "train_loss_llm": 0.0, "grad_norm": 0.03250105679035187, "global_step": 17542, "epoch": 487, "lr": 0.001} +{"train_loss": 0.033216264098882675, "train_loss_bc": 0.03320534527301788, "train_loss_llm": 0.0010920464992523193, "grad_norm": 0.047830212861299515, "global_step": 17543, "epoch": 487, "lr": 0.001} +{"train_loss": 0.028518659994006157, "train_loss_bc": 0.028512366116046906, "train_loss_llm": 0.0006294548511505127, "grad_norm": 0.05372817814350128, "global_step": 17544, "epoch": 487, "lr": 0.001} +{"train_loss": 0.029368411749601364, "train_loss_bc": 0.029368411749601364, "train_loss_llm": 0.0, "grad_norm": 0.02693275175988674, "global_step": 17545, "epoch": 487, "lr": 0.001} +{"train_loss": 0.034606803208589554, "train_loss_bc": 0.034601014107465744, "train_loss_llm": 0.000578761100769043, "grad_norm": 0.03409943729639053, "global_step": 17546, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03100600093603134, "train_loss_bc": 0.03100600093603134, "train_loss_llm": 0.0, "grad_norm": 0.04183811694383621, "global_step": 17547, "epoch": 487, "lr": 0.001} +{"train_loss": 0.0410684272646904, "train_loss_bc": 0.0410684272646904, "train_loss_llm": 0.0, "grad_norm": 0.054267093539237976, "global_step": 17548, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03971492499113083, "train_loss_bc": 0.03971492499113083, "train_loss_llm": 0.0, "grad_norm": 0.035160038620233536, "global_step": 17549, "epoch": 487, "lr": 0.001} +{"train_loss": 0.04047419875860214, "train_loss_bc": 0.04047419875860214, "train_loss_llm": 0.0, "grad_norm": 0.05401086434721947, "global_step": 17550, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03764942288398743, "train_loss_bc": 0.03764942288398743, "train_loss_llm": 0.0, "grad_norm": 0.05871441960334778, "global_step": 17551, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03862084075808525, "train_loss_bc": 0.03862084075808525, "train_loss_llm": 0.0, "grad_norm": 0.07230351120233536, "global_step": 17552, "epoch": 487, "lr": 0.001} +{"train_loss": 0.035325758159160614, "train_loss_bc": 0.03532210364937782, "train_loss_llm": 0.0003655552864074707, "grad_norm": 0.028369344770908356, "global_step": 17553, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03976474329829216, "train_loss_bc": 0.03976474329829216, "train_loss_llm": 0.0, "grad_norm": 0.041087932884693146, "global_step": 17554, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03741278871893883, "train_loss_bc": 0.03741278871893883, "train_loss_llm": 0.0, "grad_norm": 0.05625045299530029, "global_step": 17555, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03618679195642471, "train_loss_bc": 0.03618679195642471, "train_loss_llm": 0.0, "grad_norm": 0.0682196170091629, "global_step": 17556, "epoch": 487, "lr": 0.001} +{"train_loss": 0.029750827699899673, "train_loss_bc": 0.029750827699899673, "train_loss_llm": 0.0, "grad_norm": 0.02669527567923069, "global_step": 17557, "epoch": 487, "lr": 0.001} +{"train_loss": 0.029819879680871964, "train_loss_bc": 0.029819879680871964, "train_loss_llm": 0.0, "grad_norm": 0.04443483054637909, "global_step": 17558, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03522379696369171, "train_loss_bc": 0.03522379696369171, "train_loss_llm": 0.0, "grad_norm": 0.05454330891370773, "global_step": 17559, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03541591763496399, "train_loss_bc": 0.03541591763496399, "train_loss_llm": 0.0, "grad_norm": 0.07416083663702011, "global_step": 17560, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03245166689157486, "train_loss_bc": 0.03245166689157486, "train_loss_llm": 0.0, "grad_norm": 0.02210734784603119, "global_step": 17561, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03560006618499756, "train_loss_bc": 0.03560006618499756, "train_loss_llm": 0.0, "grad_norm": 0.03965826705098152, "global_step": 17562, "epoch": 487, "lr": 0.001} +{"train_loss": 0.034397803246974945, "train_loss_bc": 0.034362658858299255, "train_loss_llm": 0.0035144686698913574, "grad_norm": 0.04681215435266495, "global_step": 17563, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03352970629930496, "train_loss_bc": 0.03352970629930496, "train_loss_llm": 0.0, "grad_norm": 0.058727942407131195, "global_step": 17564, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03440151736140251, "train_loss_bc": 0.03440151736140251, "train_loss_llm": 0.0, "grad_norm": 0.025624288246035576, "global_step": 17565, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03240969032049179, "train_loss_bc": 0.03240780159831047, "train_loss_llm": 0.0001888871192932129, "grad_norm": 0.041621267795562744, "global_step": 17566, "epoch": 487, "lr": 0.001} +{"train_loss": 0.03418196406629351, "train_loss_bc": 0.026867136359214783, "train_loss_llm": 0.0, "grad_norm": 0.05078296363353729, "global_step": 17567, "epoch": 487, "lr": 0.001, "val_loss": 0.07408826053142548} +{"train_loss": 0.031178995966911316, "train_loss_bc": 0.031174248084425926, "train_loss_llm": 0.000474780797958374, "grad_norm": 0.06621335446834564, "global_step": 17568, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03357093036174774, "train_loss_bc": 0.03357093036174774, "train_loss_llm": 0.0, "grad_norm": 0.025983165949583054, "global_step": 17569, "epoch": 488, "lr": 0.001} +{"train_loss": 0.028895046561956406, "train_loss_bc": 0.028873518109321594, "train_loss_llm": 0.002152860164642334, "grad_norm": 0.04068385437130928, "global_step": 17570, "epoch": 488, "lr": 0.001} +{"train_loss": 0.035926271229982376, "train_loss_bc": 0.035926271229982376, "train_loss_llm": 0.0, "grad_norm": 0.05331263318657875, "global_step": 17571, "epoch": 488, "lr": 0.001} +{"train_loss": 0.033422987908124924, "train_loss_bc": 0.033422987908124924, "train_loss_llm": 0.0, "grad_norm": 0.06730956584215164, "global_step": 17572, "epoch": 488, "lr": 0.001} +{"train_loss": 0.0344688706099987, "train_loss_bc": 0.0344688706099987, "train_loss_llm": 0.0, "grad_norm": 0.026016445830464363, "global_step": 17573, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03108564205467701, "train_loss_bc": 0.031072009354829788, "train_loss_llm": 0.0013632774353027344, "grad_norm": 0.037136685103178024, "global_step": 17574, "epoch": 488, "lr": 0.001} +{"train_loss": 0.0335933193564415, "train_loss_bc": 0.03359034284949303, "train_loss_llm": 0.0002975165843963623, "grad_norm": 0.05149184167385101, "global_step": 17575, "epoch": 488, "lr": 0.001} +{"train_loss": 0.033960700035095215, "train_loss_bc": 0.0339469239115715, "train_loss_llm": 0.0013776123523712158, "grad_norm": 0.06688568741083145, "global_step": 17576, "epoch": 488, "lr": 0.001} +{"train_loss": 0.024110421538352966, "train_loss_bc": 0.024110421538352966, "train_loss_llm": 0.0, "grad_norm": 0.019209619611501694, "global_step": 17577, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03444954752922058, "train_loss_bc": 0.03444954752922058, "train_loss_llm": 0.0, "grad_norm": 0.030183354392647743, "global_step": 17578, "epoch": 488, "lr": 0.001} +{"train_loss": 0.031028825789690018, "train_loss_bc": 0.031028825789690018, "train_loss_llm": 0.0, "grad_norm": 0.037849586457014084, "global_step": 17579, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03715600073337555, "train_loss_bc": 0.03715600073337555, "train_loss_llm": 0.0, "grad_norm": 0.0493604876101017, "global_step": 17580, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03652157261967659, "train_loss_bc": 0.03651786223053932, "train_loss_llm": 0.00037094950675964355, "grad_norm": 0.02841436304152012, "global_step": 17581, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03574744611978531, "train_loss_bc": 0.03574744611978531, "train_loss_llm": 0.0, "grad_norm": 0.03848017007112503, "global_step": 17582, "epoch": 488, "lr": 0.001} +{"train_loss": 0.031273409724235535, "train_loss_bc": 0.031273409724235535, "train_loss_llm": 0.0, "grad_norm": 0.044662296772003174, "global_step": 17583, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03624339401721954, "train_loss_bc": 0.03624339401721954, "train_loss_llm": 0.0, "grad_norm": 0.05643223598599434, "global_step": 17584, "epoch": 488, "lr": 0.001} +{"train_loss": 0.038309771567583084, "train_loss_bc": 0.03827755153179169, "train_loss_llm": 0.0032220184803009033, "grad_norm": 0.02936076931655407, "global_step": 17585, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03122718259692192, "train_loss_bc": 0.03122718259692192, "train_loss_llm": 0.0, "grad_norm": 0.04272813722491264, "global_step": 17586, "epoch": 488, "lr": 0.001} +{"train_loss": 0.026652665808796883, "train_loss_bc": 0.026650525629520416, "train_loss_llm": 0.00021401047706604004, "grad_norm": 0.05266169458627701, "global_step": 17587, "epoch": 488, "lr": 0.001} +{"train_loss": 0.033081989735364914, "train_loss_bc": 0.033081989735364914, "train_loss_llm": 0.0, "grad_norm": 0.06129397451877594, "global_step": 17588, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03458533063530922, "train_loss_bc": 0.03458533063530922, "train_loss_llm": 0.0, "grad_norm": 0.024202724918723106, "global_step": 17589, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03145428001880646, "train_loss_bc": 0.03145049139857292, "train_loss_llm": 0.0003788769245147705, "grad_norm": 0.03862473741173744, "global_step": 17590, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03582077473402023, "train_loss_bc": 0.035808105021715164, "train_loss_llm": 0.0012669563293457031, "grad_norm": 0.05204707384109497, "global_step": 17591, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03494514897465706, "train_loss_bc": 0.03494514897465706, "train_loss_llm": 0.0, "grad_norm": 0.06797821819782257, "global_step": 17592, "epoch": 488, "lr": 0.001} +{"train_loss": 0.039611443877220154, "train_loss_bc": 0.039611443877220154, "train_loss_llm": 0.0, "grad_norm": 0.03583589196205139, "global_step": 17593, "epoch": 488, "lr": 0.001} +{"train_loss": 0.04255722463130951, "train_loss_bc": 0.04255722463130951, "train_loss_llm": 0.0, "grad_norm": 0.06335815787315369, "global_step": 17594, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03313877433538437, "train_loss_bc": 0.03313877433538437, "train_loss_llm": 0.0, "grad_norm": 0.07272795587778091, "global_step": 17595, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03629383444786072, "train_loss_bc": 0.03629383444786072, "train_loss_llm": 0.0, "grad_norm": 0.08829008042812347, "global_step": 17596, "epoch": 488, "lr": 0.001} +{"train_loss": 0.039612941443920135, "train_loss_bc": 0.039612941443920135, "train_loss_llm": 0.0, "grad_norm": 0.027233760803937912, "global_step": 17597, "epoch": 488, "lr": 0.001} +{"train_loss": 0.033752404153347015, "train_loss_bc": 0.033752404153347015, "train_loss_llm": 0.0, "grad_norm": 0.03598728030920029, "global_step": 17598, "epoch": 488, "lr": 0.001} +{"train_loss": 0.036586374044418335, "train_loss_bc": 0.036582041531801224, "train_loss_llm": 0.00043335556983947754, "grad_norm": 0.052034538239240646, "global_step": 17599, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03299855813384056, "train_loss_bc": 0.03299855813384056, "train_loss_llm": 0.0, "grad_norm": 0.06170570105314255, "global_step": 17600, "epoch": 488, "lr": 0.001} +{"train_loss": 0.036351922899484634, "train_loss_bc": 0.036351922899484634, "train_loss_llm": 0.0, "grad_norm": 0.024479636922478676, "global_step": 17601, "epoch": 488, "lr": 0.001} +{"train_loss": 0.03613545373082161, "train_loss_bc": 0.03613545373082161, "train_loss_llm": 0.0, "grad_norm": 0.04127983748912811, "global_step": 17602, "epoch": 488, "lr": 0.001} +{"train_loss": 0.034097047315703496, "train_loss_bc": 0.03174424543976784, "train_loss_llm": 0.0, "grad_norm": 0.052286118268966675, "global_step": 17603, "epoch": 488, "lr": 0.001, "val_loss": 0.0742451399564743} +{"train_loss": 0.03252210468053818, "train_loss_bc": 0.032513730227947235, "train_loss_llm": 0.0008372664451599121, "grad_norm": 0.06430868059396744, "global_step": 17604, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03089248761534691, "train_loss_bc": 0.03089248761534691, "train_loss_llm": 0.0, "grad_norm": 0.0208736602216959, "global_step": 17605, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03512302413582802, "train_loss_bc": 0.03512302413582802, "train_loss_llm": 0.0, "grad_norm": 0.03181273117661476, "global_step": 17606, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03255409747362137, "train_loss_bc": 0.03255409747362137, "train_loss_llm": 0.0, "grad_norm": 0.03659052029252052, "global_step": 17607, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03090662695467472, "train_loss_bc": 0.030898088589310646, "train_loss_llm": 0.0008539259433746338, "grad_norm": 0.04377507418394089, "global_step": 17608, "epoch": 489, "lr": 0.001} +{"train_loss": 0.0411701463162899, "train_loss_bc": 0.0411701463162899, "train_loss_llm": 0.0, "grad_norm": 0.027669083327054977, "global_step": 17609, "epoch": 489, "lr": 0.001} +{"train_loss": 0.02975713275372982, "train_loss_bc": 0.02975713275372982, "train_loss_llm": 0.0, "grad_norm": 0.04055725783109665, "global_step": 17610, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03561608865857124, "train_loss_bc": 0.03561608865857124, "train_loss_llm": 0.0, "grad_norm": 0.0659174695611, "global_step": 17611, "epoch": 489, "lr": 0.001} +{"train_loss": 0.031032662838697433, "train_loss_bc": 0.03102092817425728, "train_loss_llm": 0.0011734962463378906, "grad_norm": 0.0778396725654602, "global_step": 17612, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03468426689505577, "train_loss_bc": 0.03468426689505577, "train_loss_llm": 0.0, "grad_norm": 0.020544229075312614, "global_step": 17613, "epoch": 489, "lr": 0.001} +{"train_loss": 0.0358438715338707, "train_loss_bc": 0.035842739045619965, "train_loss_llm": 0.00011307001113891602, "grad_norm": 0.0393260158598423, "global_step": 17614, "epoch": 489, "lr": 0.001} +{"train_loss": 0.035289376974105835, "train_loss_bc": 0.035289376974105835, "train_loss_llm": 0.0, "grad_norm": 0.05009499192237854, "global_step": 17615, "epoch": 489, "lr": 0.001} +{"train_loss": 0.032870516180992126, "train_loss_bc": 0.032870516180992126, "train_loss_llm": 0.0, "grad_norm": 0.06280437111854553, "global_step": 17616, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03446481004357338, "train_loss_bc": 0.03446481004357338, "train_loss_llm": 0.0, "grad_norm": 0.02671976573765278, "global_step": 17617, "epoch": 489, "lr": 0.001} +{"train_loss": 0.030547086149454117, "train_loss_bc": 0.030521206557750702, "train_loss_llm": 0.0025878846645355225, "grad_norm": 0.03709468990564346, "global_step": 17618, "epoch": 489, "lr": 0.001} +{"train_loss": 0.029309723526239395, "train_loss_bc": 0.029309723526239395, "train_loss_llm": 0.0, "grad_norm": 0.04935945197939873, "global_step": 17619, "epoch": 489, "lr": 0.001} +{"train_loss": 0.034705597907304764, "train_loss_bc": 0.034705597907304764, "train_loss_llm": 0.0, "grad_norm": 0.0490548200905323, "global_step": 17620, "epoch": 489, "lr": 0.001} +{"train_loss": 0.038209009915590286, "train_loss_bc": 0.038209009915590286, "train_loss_llm": 0.0, "grad_norm": 0.028973937034606934, "global_step": 17621, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03379201143980026, "train_loss_bc": 0.03379201143980026, "train_loss_llm": 0.0, "grad_norm": 0.051269158720970154, "global_step": 17622, "epoch": 489, "lr": 0.001} +{"train_loss": 0.0317513644695282, "train_loss_bc": 0.03174467012286186, "train_loss_llm": 0.0006695091724395752, "grad_norm": 0.06778783351182938, "global_step": 17623, "epoch": 489, "lr": 0.001} +{"train_loss": 0.036433491855859756, "train_loss_bc": 0.036433491855859756, "train_loss_llm": 0.0, "grad_norm": 0.08392363041639328, "global_step": 17624, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03471066802740097, "train_loss_bc": 0.034710463136434555, "train_loss_llm": 2.065300941467285e-05, "grad_norm": 0.029904356226325035, "global_step": 17625, "epoch": 489, "lr": 0.001} +{"train_loss": 0.033119797706604004, "train_loss_bc": 0.033119797706604004, "train_loss_llm": 0.0, "grad_norm": 0.04476865753531456, "global_step": 17626, "epoch": 489, "lr": 0.001} +{"train_loss": 0.036107052117586136, "train_loss_bc": 0.036107052117586136, "train_loss_llm": 0.0, "grad_norm": 0.05652556195855141, "global_step": 17627, "epoch": 489, "lr": 0.001} +{"train_loss": 0.030881473794579506, "train_loss_bc": 0.0308698583394289, "train_loss_llm": 0.0011615157127380371, "grad_norm": 0.06840731203556061, "global_step": 17628, "epoch": 489, "lr": 0.001} +{"train_loss": 0.034433357417583466, "train_loss_bc": 0.034433357417583466, "train_loss_llm": 0.0, "grad_norm": 0.02300611138343811, "global_step": 17629, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03409675881266594, "train_loss_bc": 0.03409675881266594, "train_loss_llm": 0.0, "grad_norm": 0.03712945803999901, "global_step": 17630, "epoch": 489, "lr": 0.001} +{"train_loss": 0.033984240144491196, "train_loss_bc": 0.033984240144491196, "train_loss_llm": 0.0, "grad_norm": 0.04608263447880745, "global_step": 17631, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03826049342751503, "train_loss_bc": 0.03826049342751503, "train_loss_llm": 0.0, "grad_norm": 0.05847286060452461, "global_step": 17632, "epoch": 489, "lr": 0.001} +{"train_loss": 0.032574836164712906, "train_loss_bc": 0.03257068991661072, "train_loss_llm": 0.0004144608974456787, "grad_norm": 0.026636330410838127, "global_step": 17633, "epoch": 489, "lr": 0.001} +{"train_loss": 0.038652047514915466, "train_loss_bc": 0.038652047514915466, "train_loss_llm": 0.0, "grad_norm": 0.039397209882736206, "global_step": 17634, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03882370889186859, "train_loss_bc": 0.03882370889186859, "train_loss_llm": 0.0, "grad_norm": 0.049343474209308624, "global_step": 17635, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03520352393388748, "train_loss_bc": 0.03520352393388748, "train_loss_llm": 0.0, "grad_norm": 0.0597788505256176, "global_step": 17636, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03633977100253105, "train_loss_bc": 0.03633977100253105, "train_loss_llm": 0.0, "grad_norm": 0.02474110573530197, "global_step": 17637, "epoch": 489, "lr": 0.001} +{"train_loss": 0.0323561355471611, "train_loss_bc": 0.0323561355471611, "train_loss_llm": 0.0, "grad_norm": 0.0390663743019104, "global_step": 17638, "epoch": 489, "lr": 0.001} +{"train_loss": 0.03412973182275891, "train_loss_bc": 0.03163360804319382, "train_loss_llm": 0.0017375349998474121, "grad_norm": 0.051378365606069565, "global_step": 17639, "epoch": 489, "lr": 0.001, "val_loss": 0.07413333654403687} +{"train_loss": 0.03816644474864006, "train_loss_bc": 0.03816644474864006, "train_loss_llm": 0.0, "grad_norm": 0.07172892242670059, "global_step": 17640, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03350798785686493, "train_loss_bc": 0.03349333629012108, "train_loss_llm": 0.0014652609825134277, "grad_norm": 0.023547951132059097, "global_step": 17641, "epoch": 490, "lr": 0.001} +{"train_loss": 0.030901966616511345, "train_loss_bc": 0.030901966616511345, "train_loss_llm": 0.0, "grad_norm": 0.03480834513902664, "global_step": 17642, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03118017315864563, "train_loss_bc": 0.03118017315864563, "train_loss_llm": 0.0, "grad_norm": 0.040606677532196045, "global_step": 17643, "epoch": 490, "lr": 0.001} +{"train_loss": 0.0369928702712059, "train_loss_bc": 0.0369928702712059, "train_loss_llm": 0.0, "grad_norm": 0.052841830998659134, "global_step": 17644, "epoch": 490, "lr": 0.001} +{"train_loss": 0.036953557282686234, "train_loss_bc": 0.036953557282686234, "train_loss_llm": 0.0, "grad_norm": 0.028654254972934723, "global_step": 17645, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03276587650179863, "train_loss_bc": 0.03276587650179863, "train_loss_llm": 0.0, "grad_norm": 0.048865802586078644, "global_step": 17646, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03552169352769852, "train_loss_bc": 0.03552169352769852, "train_loss_llm": 0.0, "grad_norm": 0.0555681474506855, "global_step": 17647, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03248537331819534, "train_loss_bc": 0.03248537331819534, "train_loss_llm": 0.0, "grad_norm": 0.06796278804540634, "global_step": 17648, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03506099432706833, "train_loss_bc": 0.03506099432706833, "train_loss_llm": 0.0, "grad_norm": 0.02258857898414135, "global_step": 17649, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03282349929213524, "train_loss_bc": 0.032817959785461426, "train_loss_llm": 0.000553816556930542, "grad_norm": 0.03733927011489868, "global_step": 17650, "epoch": 490, "lr": 0.001} +{"train_loss": 0.032276734709739685, "train_loss_bc": 0.032276734709739685, "train_loss_llm": 0.0, "grad_norm": 0.05116399750113487, "global_step": 17651, "epoch": 490, "lr": 0.001} +{"train_loss": 0.037686001509428024, "train_loss_bc": 0.037686001509428024, "train_loss_llm": 0.0, "grad_norm": 0.05897143855690956, "global_step": 17652, "epoch": 490, "lr": 0.001} +{"train_loss": 0.032135769724845886, "train_loss_bc": 0.032135769724845886, "train_loss_llm": 0.0, "grad_norm": 0.02350599318742752, "global_step": 17653, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03717300295829773, "train_loss_bc": 0.03717300295829773, "train_loss_llm": 0.0, "grad_norm": 0.04119010269641876, "global_step": 17654, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03129584714770317, "train_loss_bc": 0.03129584714770317, "train_loss_llm": 0.0, "grad_norm": 0.05495817959308624, "global_step": 17655, "epoch": 490, "lr": 0.001} +{"train_loss": 0.0324740968644619, "train_loss_bc": 0.0324740968644619, "train_loss_llm": 0.0, "grad_norm": 0.06957709789276123, "global_step": 17656, "epoch": 490, "lr": 0.001} +{"train_loss": 0.028143582865595818, "train_loss_bc": 0.02813727967441082, "train_loss_llm": 0.0006302893161773682, "grad_norm": 0.02132909558713436, "global_step": 17657, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03652464598417282, "train_loss_bc": 0.03652464598417282, "train_loss_llm": 0.0, "grad_norm": 0.04569103196263313, "global_step": 17658, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03700295835733414, "train_loss_bc": 0.03700295835733414, "train_loss_llm": 0.0, "grad_norm": 0.058729518204927444, "global_step": 17659, "epoch": 490, "lr": 0.001} +{"train_loss": 0.030373351648449898, "train_loss_bc": 0.030373351648449898, "train_loss_llm": 0.0, "grad_norm": 0.06790206581354141, "global_step": 17660, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03828170895576477, "train_loss_bc": 0.03828170895576477, "train_loss_llm": 0.0, "grad_norm": 0.028432177379727364, "global_step": 17661, "epoch": 490, "lr": 0.001} +{"train_loss": 0.035059213638305664, "train_loss_bc": 0.035059213638305664, "train_loss_llm": 0.0, "grad_norm": 0.0380634181201458, "global_step": 17662, "epoch": 490, "lr": 0.001} +{"train_loss": 0.035582397133111954, "train_loss_bc": 0.035581428557634354, "train_loss_llm": 9.691715240478516e-05, "grad_norm": 0.048393502831459045, "global_step": 17663, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03352154418826103, "train_loss_bc": 0.03352154418826103, "train_loss_llm": 0.0, "grad_norm": 0.05444144830107689, "global_step": 17664, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03248487785458565, "train_loss_bc": 0.03248487785458565, "train_loss_llm": 0.0, "grad_norm": 0.02111297845840454, "global_step": 17665, "epoch": 490, "lr": 0.001} +{"train_loss": 0.033279284834861755, "train_loss_bc": 0.033279284834861755, "train_loss_llm": 0.0, "grad_norm": 0.032511431723833084, "global_step": 17666, "epoch": 490, "lr": 0.001} +{"train_loss": 0.0331522598862648, "train_loss_bc": 0.03312942013144493, "train_loss_llm": 0.0022839903831481934, "grad_norm": 0.04125429689884186, "global_step": 17667, "epoch": 490, "lr": 0.001} +{"train_loss": 0.04051847755908966, "train_loss_bc": 0.04049601033329964, "train_loss_llm": 0.0022468864917755127, "grad_norm": 0.05447673052549362, "global_step": 17668, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03318359702825546, "train_loss_bc": 0.03318359702825546, "train_loss_llm": 0.0, "grad_norm": 0.025659775361418724, "global_step": 17669, "epoch": 490, "lr": 0.001} +{"train_loss": 0.0323391929268837, "train_loss_bc": 0.0323391929268837, "train_loss_llm": 0.0, "grad_norm": 0.03817997872829437, "global_step": 17670, "epoch": 490, "lr": 0.001} +{"train_loss": 0.032859936356544495, "train_loss_bc": 0.032859936356544495, "train_loss_llm": 0.0, "grad_norm": 0.047019895166158676, "global_step": 17671, "epoch": 490, "lr": 0.001} +{"train_loss": 0.032926611602306366, "train_loss_bc": 0.032926611602306366, "train_loss_llm": 0.0, "grad_norm": 0.0630481019616127, "global_step": 17672, "epoch": 490, "lr": 0.001} +{"train_loss": 0.034982454031705856, "train_loss_bc": 0.034982454031705856, "train_loss_llm": 0.0, "grad_norm": 0.024057937785983086, "global_step": 17673, "epoch": 490, "lr": 0.001} +{"train_loss": 0.04144679382443428, "train_loss_bc": 0.041444290429353714, "train_loss_llm": 0.0002505183219909668, "grad_norm": 0.03505699709057808, "global_step": 17674, "epoch": 490, "lr": 0.001} +{"train_loss": 0.03406744822859764, "train_loss_bc": 0.025363357737660408, "train_loss_llm": 0.0, "grad_norm": 0.043992914259433746, "global_step": 17675, "epoch": 490, "lr": 0.001, "train/cumulative_reward": 45.98114463634973, "train/mean_score": 9.611671024054486, "train/success_rate": 0.3, "test/cumulative_reward": 35.80186742428684, "test/mean_score": 9.227864877622475, "test/success_rate": 0.24, "val_loss": 0.07300999760627747, "train_action_mse_error": 0.03761206567287445} +{"train_loss": 0.03616946190595627, "train_loss_bc": 0.03613264858722687, "train_loss_llm": 0.0036814212799072266, "grad_norm": 0.0579865463078022, "global_step": 17676, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03470923379063606, "train_loss_bc": 0.034703440964221954, "train_loss_llm": 0.000579148530960083, "grad_norm": 0.023624103516340256, "global_step": 17677, "epoch": 491, "lr": 0.001} +{"train_loss": 0.02983648143708706, "train_loss_bc": 0.029826046898961067, "train_loss_llm": 0.0010435283184051514, "grad_norm": 0.03807023912668228, "global_step": 17678, "epoch": 491, "lr": 0.001} +{"train_loss": 0.032421696931123734, "train_loss_bc": 0.032421696931123734, "train_loss_llm": 0.0, "grad_norm": 0.06205316632986069, "global_step": 17679, "epoch": 491, "lr": 0.001} +{"train_loss": 0.034218840301036835, "train_loss_bc": 0.034218840301036835, "train_loss_llm": 0.0, "grad_norm": 0.07778855413198471, "global_step": 17680, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03483380749821663, "train_loss_bc": 0.03483380749821663, "train_loss_llm": 0.0, "grad_norm": 0.02260550856590271, "global_step": 17681, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03402324765920639, "train_loss_bc": 0.03402324765920639, "train_loss_llm": 0.0, "grad_norm": 0.04007052257657051, "global_step": 17682, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03592170029878616, "train_loss_bc": 0.03592170029878616, "train_loss_llm": 0.0, "grad_norm": 0.04904293641448021, "global_step": 17683, "epoch": 491, "lr": 0.001} +{"train_loss": 0.0340479277074337, "train_loss_bc": 0.03404713794589043, "train_loss_llm": 7.903575897216797e-05, "grad_norm": 0.05505983158946037, "global_step": 17684, "epoch": 491, "lr": 0.001} +{"train_loss": 0.033415116369724274, "train_loss_bc": 0.033415116369724274, "train_loss_llm": 0.0, "grad_norm": 0.025753486901521683, "global_step": 17685, "epoch": 491, "lr": 0.001} +{"train_loss": 0.026256432756781578, "train_loss_bc": 0.026256432756781578, "train_loss_llm": 0.0, "grad_norm": 0.030549675226211548, "global_step": 17686, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03320007026195526, "train_loss_bc": 0.03320007026195526, "train_loss_llm": 0.0, "grad_norm": 0.03993959724903107, "global_step": 17687, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03455224260687828, "train_loss_bc": 0.03455224260687828, "train_loss_llm": 0.0, "grad_norm": 0.04855357110500336, "global_step": 17688, "epoch": 491, "lr": 0.001} +{"train_loss": 0.033770669251680374, "train_loss_bc": 0.03376461938023567, "train_loss_llm": 0.0006048381328582764, "grad_norm": 0.020466400310397148, "global_step": 17689, "epoch": 491, "lr": 0.001} +{"train_loss": 0.033021602779626846, "train_loss_bc": 0.03302015736699104, "train_loss_llm": 0.00014457106590270996, "grad_norm": 0.030995991080999374, "global_step": 17690, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03505467250943184, "train_loss_bc": 0.03505467250943184, "train_loss_llm": 0.0, "grad_norm": 0.04386984184384346, "global_step": 17691, "epoch": 491, "lr": 0.001} +{"train_loss": 0.038920704275369644, "train_loss_bc": 0.03886415436863899, "train_loss_llm": 0.005655139684677124, "grad_norm": 0.056155044585466385, "global_step": 17692, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03494718670845032, "train_loss_bc": 0.03494718670845032, "train_loss_llm": 0.0, "grad_norm": 0.02419191226363182, "global_step": 17693, "epoch": 491, "lr": 0.001} +{"train_loss": 0.032851289957761765, "train_loss_bc": 0.032851289957761765, "train_loss_llm": 0.0, "grad_norm": 0.033106543123722076, "global_step": 17694, "epoch": 491, "lr": 0.001} +{"train_loss": 0.032395996153354645, "train_loss_bc": 0.032395996153354645, "train_loss_llm": 0.0, "grad_norm": 0.040913958102464676, "global_step": 17695, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03444809839129448, "train_loss_bc": 0.03444809839129448, "train_loss_llm": 0.0, "grad_norm": 0.05282812565565109, "global_step": 17696, "epoch": 491, "lr": 0.001} +{"train_loss": 0.044608570635318756, "train_loss_bc": 0.044608570635318756, "train_loss_llm": 0.0, "grad_norm": 0.03246539458632469, "global_step": 17697, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03313615173101425, "train_loss_bc": 0.03313615173101425, "train_loss_llm": 0.0, "grad_norm": 0.04434796795248985, "global_step": 17698, "epoch": 491, "lr": 0.001} +{"train_loss": 0.032562386244535446, "train_loss_bc": 0.032562386244535446, "train_loss_llm": 0.0, "grad_norm": 0.05075674131512642, "global_step": 17699, "epoch": 491, "lr": 0.001} +{"train_loss": 0.0343286357820034, "train_loss_bc": 0.0343286357820034, "train_loss_llm": 0.0, "grad_norm": 0.05854184180498123, "global_step": 17700, "epoch": 491, "lr": 0.001} +{"train_loss": 0.038664478808641434, "train_loss_bc": 0.038664478808641434, "train_loss_llm": 0.0, "grad_norm": 0.02927899733185768, "global_step": 17701, "epoch": 491, "lr": 0.001} +{"train_loss": 0.033118296414613724, "train_loss_bc": 0.033118296414613724, "train_loss_llm": 0.0, "grad_norm": 0.0395471416413784, "global_step": 17702, "epoch": 491, "lr": 0.001} +{"train_loss": 0.028372090309858322, "train_loss_bc": 0.028372090309858322, "train_loss_llm": 0.0, "grad_norm": 0.05450921878218651, "global_step": 17703, "epoch": 491, "lr": 0.001} +{"train_loss": 0.028673280030488968, "train_loss_bc": 0.028673280030488968, "train_loss_llm": 0.0, "grad_norm": 0.059732574969530106, "global_step": 17704, "epoch": 491, "lr": 0.001} +{"train_loss": 0.038374267518520355, "train_loss_bc": 0.038374267518520355, "train_loss_llm": 0.0, "grad_norm": 0.02528783492743969, "global_step": 17705, "epoch": 491, "lr": 0.001} +{"train_loss": 0.0334322415292263, "train_loss_bc": 0.0334322415292263, "train_loss_llm": 0.0, "grad_norm": 0.03428279608488083, "global_step": 17706, "epoch": 491, "lr": 0.001} +{"train_loss": 0.030525818467140198, "train_loss_bc": 0.030525818467140198, "train_loss_llm": 0.0, "grad_norm": 0.04426321014761925, "global_step": 17707, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03223830461502075, "train_loss_bc": 0.03223830461502075, "train_loss_llm": 0.0, "grad_norm": 0.05086424574255943, "global_step": 17708, "epoch": 491, "lr": 0.001} +{"train_loss": 0.038810163736343384, "train_loss_bc": 0.038810163736343384, "train_loss_llm": 0.0, "grad_norm": 0.02807384356856346, "global_step": 17709, "epoch": 491, "lr": 0.001} +{"train_loss": 0.034742310643196106, "train_loss_bc": 0.034742310643196106, "train_loss_llm": 0.0, "grad_norm": 0.044715337455272675, "global_step": 17710, "epoch": 491, "lr": 0.001} +{"train_loss": 0.03402274764246411, "train_loss_bc": 0.034208789467811584, "train_loss_llm": 0.0006649196147918701, "grad_norm": 0.05800347402691841, "global_step": 17711, "epoch": 491, "lr": 0.001, "val_loss": 0.07496093958616257} +{"train_loss": 0.02942359633743763, "train_loss_bc": 0.02942359633743763, "train_loss_llm": 0.0, "grad_norm": 0.06571297347545624, "global_step": 17712, "epoch": 492, "lr": 0.001} +{"train_loss": 0.02997131645679474, "train_loss_bc": 0.02995382249355316, "train_loss_llm": 0.001749485731124878, "grad_norm": 0.022397330030798912, "global_step": 17713, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03427513316273689, "train_loss_bc": 0.03427513316273689, "train_loss_llm": 0.0, "grad_norm": 0.033544570207595825, "global_step": 17714, "epoch": 492, "lr": 0.001} +{"train_loss": 0.036245930939912796, "train_loss_bc": 0.036245930939912796, "train_loss_llm": 0.0, "grad_norm": 0.04405876249074936, "global_step": 17715, "epoch": 492, "lr": 0.001} +{"train_loss": 0.02974178083240986, "train_loss_bc": 0.02974178083240986, "train_loss_llm": 0.0, "grad_norm": 0.054711803793907166, "global_step": 17716, "epoch": 492, "lr": 0.001} +{"train_loss": 0.033986061811447144, "train_loss_bc": 0.033986061811447144, "train_loss_llm": 0.0, "grad_norm": 0.027928274124860764, "global_step": 17717, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03568997234106064, "train_loss_bc": 0.03568997234106064, "train_loss_llm": 0.0, "grad_norm": 0.03385140374302864, "global_step": 17718, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03310927376151085, "train_loss_bc": 0.033092133700847626, "train_loss_llm": 0.0017139613628387451, "grad_norm": 0.04774325713515282, "global_step": 17719, "epoch": 492, "lr": 0.001} +{"train_loss": 0.036765508353710175, "train_loss_bc": 0.036765508353710175, "train_loss_llm": 0.0, "grad_norm": 0.0653548464179039, "global_step": 17720, "epoch": 492, "lr": 0.001} +{"train_loss": 0.032395437359809875, "train_loss_bc": 0.032395437359809875, "train_loss_llm": 0.0, "grad_norm": 0.02061331272125244, "global_step": 17721, "epoch": 492, "lr": 0.001} +{"train_loss": 0.0397731252014637, "train_loss_bc": 0.0397731252014637, "train_loss_llm": 0.0, "grad_norm": 0.03758002817630768, "global_step": 17722, "epoch": 492, "lr": 0.001} +{"train_loss": 0.04468259960412979, "train_loss_bc": 0.04468259960412979, "train_loss_llm": 0.0, "grad_norm": 0.047714874148368835, "global_step": 17723, "epoch": 492, "lr": 0.001} +{"train_loss": 0.042542099952697754, "train_loss_bc": 0.042542099952697754, "train_loss_llm": 0.0, "grad_norm": 0.062451597303152084, "global_step": 17724, "epoch": 492, "lr": 0.001} +{"train_loss": 0.027503978461027145, "train_loss_bc": 0.027503978461027145, "train_loss_llm": 0.0, "grad_norm": 0.021432863548398018, "global_step": 17725, "epoch": 492, "lr": 0.001} +{"train_loss": 0.034531962126493454, "train_loss_bc": 0.034531962126493454, "train_loss_llm": 0.0, "grad_norm": 0.03148927539587021, "global_step": 17726, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03359701856970787, "train_loss_bc": 0.033571816980838776, "train_loss_llm": 0.002520263195037842, "grad_norm": 0.0460292249917984, "global_step": 17727, "epoch": 492, "lr": 0.001} +{"train_loss": 0.037370070815086365, "train_loss_bc": 0.037370070815086365, "train_loss_llm": 0.0, "grad_norm": 0.05779789760708809, "global_step": 17728, "epoch": 492, "lr": 0.001} +{"train_loss": 0.033849772065877914, "train_loss_bc": 0.03383886069059372, "train_loss_llm": 0.0010912120342254639, "grad_norm": 0.023624584078788757, "global_step": 17729, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03467802330851555, "train_loss_bc": 0.03467802330851555, "train_loss_llm": 0.0, "grad_norm": 0.04014848917722702, "global_step": 17730, "epoch": 492, "lr": 0.001} +{"train_loss": 0.031431734561920166, "train_loss_bc": 0.031431734561920166, "train_loss_llm": 0.0, "grad_norm": 0.04185856133699417, "global_step": 17731, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03762674704194069, "train_loss_bc": 0.037626560777425766, "train_loss_llm": 1.8805265426635742e-05, "grad_norm": 0.052076306194067, "global_step": 17732, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03543948754668236, "train_loss_bc": 0.03543948754668236, "train_loss_llm": 0.0, "grad_norm": 0.030353005975484848, "global_step": 17733, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03559847176074982, "train_loss_bc": 0.03559847176074982, "train_loss_llm": 0.0, "grad_norm": 0.03868157044053078, "global_step": 17734, "epoch": 492, "lr": 0.001} +{"train_loss": 0.031168527901172638, "train_loss_bc": 0.031161407008767128, "train_loss_llm": 0.0007121264934539795, "grad_norm": 0.05510035529732704, "global_step": 17735, "epoch": 492, "lr": 0.001} +{"train_loss": 0.0276547409594059, "train_loss_bc": 0.027639422565698624, "train_loss_llm": 0.0015318095684051514, "grad_norm": 0.06429925560951233, "global_step": 17736, "epoch": 492, "lr": 0.001} +{"train_loss": 0.029382243752479553, "train_loss_bc": 0.029381340369582176, "train_loss_llm": 9.033083915710449e-05, "grad_norm": 0.023647647351026535, "global_step": 17737, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03281472250819206, "train_loss_bc": 0.03281472250819206, "train_loss_llm": 0.0, "grad_norm": 0.03554223105311394, "global_step": 17738, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03498419374227524, "train_loss_bc": 0.03497486934065819, "train_loss_llm": 0.0009323656558990479, "grad_norm": 0.05266626551747322, "global_step": 17739, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03733453527092934, "train_loss_bc": 0.03733453527092934, "train_loss_llm": 0.0, "grad_norm": 0.06947048008441925, "global_step": 17740, "epoch": 492, "lr": 0.001} +{"train_loss": 0.031605593860149384, "train_loss_bc": 0.03159727901220322, "train_loss_llm": 0.0008315742015838623, "grad_norm": 0.024083368480205536, "global_step": 17741, "epoch": 492, "lr": 0.001} +{"train_loss": 0.02963748760521412, "train_loss_bc": 0.02963748760521412, "train_loss_llm": 0.0, "grad_norm": 0.033842772245407104, "global_step": 17742, "epoch": 492, "lr": 0.001} +{"train_loss": 0.033277686685323715, "train_loss_bc": 0.033277686685323715, "train_loss_llm": 0.0, "grad_norm": 0.04933818429708481, "global_step": 17743, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03050508350133896, "train_loss_bc": 0.03050508350133896, "train_loss_llm": 0.0, "grad_norm": 0.06204959750175476, "global_step": 17744, "epoch": 492, "lr": 0.001} +{"train_loss": 0.038723450154066086, "train_loss_bc": 0.038723450154066086, "train_loss_llm": 0.0, "grad_norm": 0.03031228668987751, "global_step": 17745, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03944121301174164, "train_loss_bc": 0.03944121301174164, "train_loss_llm": 0.0, "grad_norm": 0.04677065834403038, "global_step": 17746, "epoch": 492, "lr": 0.001} +{"train_loss": 0.03407776878318853, "train_loss_bc": 0.03004109486937523, "train_loss_llm": 0.0, "grad_norm": 0.05864965543150902, "global_step": 17747, "epoch": 492, "lr": 0.001, "val_loss": 0.07451291382312775} +{"train_loss": 0.039397746324539185, "train_loss_bc": 0.039397746324539185, "train_loss_llm": 0.0, "grad_norm": 0.07219947129487991, "global_step": 17748, "epoch": 493, "lr": 0.001} +{"train_loss": 0.031660329550504684, "train_loss_bc": 0.031660329550504684, "train_loss_llm": 0.0, "grad_norm": 0.025090845301747322, "global_step": 17749, "epoch": 493, "lr": 0.001} +{"train_loss": 0.041595060378313065, "train_loss_bc": 0.04158082604408264, "train_loss_llm": 0.0014232993125915527, "grad_norm": 0.042582303285598755, "global_step": 17750, "epoch": 493, "lr": 0.001} +{"train_loss": 0.036095939576625824, "train_loss_bc": 0.03609386458992958, "train_loss_llm": 0.00020736455917358398, "grad_norm": 0.05314246937632561, "global_step": 17751, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03174935281276703, "train_loss_bc": 0.03174935281276703, "train_loss_llm": 0.0, "grad_norm": 0.06329400092363358, "global_step": 17752, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03843937814235687, "train_loss_bc": 0.03843937814235687, "train_loss_llm": 0.0, "grad_norm": 0.02443028800189495, "global_step": 17753, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03873596712946892, "train_loss_bc": 0.03873596712946892, "train_loss_llm": 0.0, "grad_norm": 0.027658697217702866, "global_step": 17754, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03424397483468056, "train_loss_bc": 0.03424397483468056, "train_loss_llm": 0.0, "grad_norm": 0.03901578113436699, "global_step": 17755, "epoch": 493, "lr": 0.001} +{"train_loss": 0.028260840103030205, "train_loss_bc": 0.028260840103030205, "train_loss_llm": 0.0, "grad_norm": 0.0501205176115036, "global_step": 17756, "epoch": 493, "lr": 0.001} +{"train_loss": 0.038009852170944214, "train_loss_bc": 0.038009852170944214, "train_loss_llm": 0.0, "grad_norm": 0.033392779529094696, "global_step": 17757, "epoch": 493, "lr": 0.001} +{"train_loss": 0.036189816892147064, "train_loss_bc": 0.036189060658216476, "train_loss_llm": 7.56680965423584e-05, "grad_norm": 0.04838161543011665, "global_step": 17758, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03507425636053085, "train_loss_bc": 0.035073354840278625, "train_loss_llm": 9.015202522277832e-05, "grad_norm": 0.060079824179410934, "global_step": 17759, "epoch": 493, "lr": 0.001} +{"train_loss": 0.029139427468180656, "train_loss_bc": 0.029132885858416557, "train_loss_llm": 0.0006541013717651367, "grad_norm": 0.06983157992362976, "global_step": 17760, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03370087966322899, "train_loss_bc": 0.03370087966322899, "train_loss_llm": 0.0, "grad_norm": 0.028796512633562088, "global_step": 17761, "epoch": 493, "lr": 0.001} +{"train_loss": 0.032326050102710724, "train_loss_bc": 0.032326050102710724, "train_loss_llm": 0.0, "grad_norm": 0.04146537929773331, "global_step": 17762, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03283863142132759, "train_loss_bc": 0.03283863142132759, "train_loss_llm": 0.0, "grad_norm": 0.04962586611509323, "global_step": 17763, "epoch": 493, "lr": 0.001} +{"train_loss": 0.034312356263399124, "train_loss_bc": 0.034312356263399124, "train_loss_llm": 0.0, "grad_norm": 0.06436207890510559, "global_step": 17764, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03158968687057495, "train_loss_bc": 0.03158968687057495, "train_loss_llm": 0.0, "grad_norm": 0.023390214890241623, "global_step": 17765, "epoch": 493, "lr": 0.001} +{"train_loss": 0.031086523085832596, "train_loss_bc": 0.031086523085832596, "train_loss_llm": 0.0, "grad_norm": 0.030926132574677467, "global_step": 17766, "epoch": 493, "lr": 0.001} +{"train_loss": 0.039556801319122314, "train_loss_bc": 0.03955230861902237, "train_loss_llm": 0.0004493296146392822, "grad_norm": 0.04519023001194, "global_step": 17767, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03736136481165886, "train_loss_bc": 0.03736136481165886, "train_loss_llm": 0.0, "grad_norm": 0.059624083340168, "global_step": 17768, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03860807046294212, "train_loss_bc": 0.03860727325081825, "train_loss_llm": 7.972121238708496e-05, "grad_norm": 0.027773041278123856, "global_step": 17769, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03415844216942787, "train_loss_bc": 0.03415844216942787, "train_loss_llm": 0.0, "grad_norm": 0.04085661470890045, "global_step": 17770, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03503253683447838, "train_loss_bc": 0.03503253683447838, "train_loss_llm": 0.0, "grad_norm": 0.053222447633743286, "global_step": 17771, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03607560694217682, "train_loss_bc": 0.03607560694217682, "train_loss_llm": 0.0, "grad_norm": 0.07095597684383392, "global_step": 17772, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03598051145672798, "train_loss_bc": 0.03598051145672798, "train_loss_llm": 0.0, "grad_norm": 0.027769695967435837, "global_step": 17773, "epoch": 493, "lr": 0.001} +{"train_loss": 0.036308445036411285, "train_loss_bc": 0.03630302473902702, "train_loss_llm": 0.0005420148372650146, "grad_norm": 0.04100065678358078, "global_step": 17774, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03257890045642853, "train_loss_bc": 0.03257528319954872, "train_loss_llm": 0.0003618896007537842, "grad_norm": 0.05688835680484772, "global_step": 17775, "epoch": 493, "lr": 0.001} +{"train_loss": 0.02843678928911686, "train_loss_bc": 0.02843678928911686, "train_loss_llm": 0.0, "grad_norm": 0.06884563714265823, "global_step": 17776, "epoch": 493, "lr": 0.001} +{"train_loss": 0.028321176767349243, "train_loss_bc": 0.028317933902144432, "train_loss_llm": 0.0003243684768676758, "grad_norm": 0.022124765440821648, "global_step": 17777, "epoch": 493, "lr": 0.001} +{"train_loss": 0.02831067331135273, "train_loss_bc": 0.02829612046480179, "train_loss_llm": 0.0014552474021911621, "grad_norm": 0.031179875135421753, "global_step": 17778, "epoch": 493, "lr": 0.001} +{"train_loss": 0.033060330897569656, "train_loss_bc": 0.03303404524922371, "train_loss_llm": 0.0026285648345947266, "grad_norm": 0.046268995851278305, "global_step": 17779, "epoch": 493, "lr": 0.001} +{"train_loss": 0.027448341250419617, "train_loss_bc": 0.027448341250419617, "train_loss_llm": 0.0, "grad_norm": 0.05278293043375015, "global_step": 17780, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03320557624101639, "train_loss_bc": 0.03320557624101639, "train_loss_llm": 0.0, "grad_norm": 0.024476904422044754, "global_step": 17781, "epoch": 493, "lr": 0.001} +{"train_loss": 0.035256437957286835, "train_loss_bc": 0.03525150939822197, "train_loss_llm": 0.0004927217960357666, "grad_norm": 0.03575722873210907, "global_step": 17782, "epoch": 493, "lr": 0.001} +{"train_loss": 0.03413546323362324, "train_loss_bc": 0.03473060205578804, "train_loss_llm": 0.0, "grad_norm": 0.052349817007780075, "global_step": 17783, "epoch": 493, "lr": 0.001, "val_loss": 0.07375409454107285} +{"train_loss": 0.035429153591394424, "train_loss_bc": 0.035429153591394424, "train_loss_llm": 0.0, "grad_norm": 0.06509070098400116, "global_step": 17784, "epoch": 494, "lr": 0.001} +{"train_loss": 0.041729722172021866, "train_loss_bc": 0.041729722172021866, "train_loss_llm": 0.0, "grad_norm": 0.027236707508563995, "global_step": 17785, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03069743514060974, "train_loss_bc": 0.03069743514060974, "train_loss_llm": 0.0, "grad_norm": 0.035700924694538116, "global_step": 17786, "epoch": 494, "lr": 0.001} +{"train_loss": 0.034557782113552094, "train_loss_bc": 0.034557782113552094, "train_loss_llm": 0.0, "grad_norm": 0.048739004880189896, "global_step": 17787, "epoch": 494, "lr": 0.001} +{"train_loss": 0.040996942669153214, "train_loss_bc": 0.040996942669153214, "train_loss_llm": 0.0, "grad_norm": 0.0646316260099411, "global_step": 17788, "epoch": 494, "lr": 0.001} +{"train_loss": 0.030653290450572968, "train_loss_bc": 0.030653290450572968, "train_loss_llm": 0.0, "grad_norm": 0.024247338995337486, "global_step": 17789, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03646818920969963, "train_loss_bc": 0.03646818920969963, "train_loss_llm": 0.0, "grad_norm": 0.043833762407302856, "global_step": 17790, "epoch": 494, "lr": 0.001} +{"train_loss": 0.041106775403022766, "train_loss_bc": 0.041106775403022766, "train_loss_llm": 0.0, "grad_norm": 0.057811785489320755, "global_step": 17791, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03444244712591171, "train_loss_bc": 0.03444244712591171, "train_loss_llm": 0.0, "grad_norm": 0.07249283045530319, "global_step": 17792, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03201931342482567, "train_loss_bc": 0.03201931342482567, "train_loss_llm": 0.0, "grad_norm": 0.028707271441817284, "global_step": 17793, "epoch": 494, "lr": 0.001} +{"train_loss": 0.028811290860176086, "train_loss_bc": 0.028801998123526573, "train_loss_llm": 0.0009292662143707275, "grad_norm": 0.046104542911052704, "global_step": 17794, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03488917648792267, "train_loss_bc": 0.03488917648792267, "train_loss_llm": 0.0, "grad_norm": 0.052874837070703506, "global_step": 17795, "epoch": 494, "lr": 0.001} +{"train_loss": 0.0352950245141983, "train_loss_bc": 0.035284269601106644, "train_loss_llm": 0.0010753870010375977, "grad_norm": 0.07168538868427277, "global_step": 17796, "epoch": 494, "lr": 0.001} +{"train_loss": 0.030940640717744827, "train_loss_bc": 0.030940640717744827, "train_loss_llm": 0.0, "grad_norm": 0.02534675970673561, "global_step": 17797, "epoch": 494, "lr": 0.001} +{"train_loss": 0.022759458050131798, "train_loss_bc": 0.022759458050131798, "train_loss_llm": 0.0, "grad_norm": 0.0466386117041111, "global_step": 17798, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03710108622908592, "train_loss_bc": 0.03710108622908592, "train_loss_llm": 0.0, "grad_norm": 0.061545368283987045, "global_step": 17799, "epoch": 494, "lr": 0.001} +{"train_loss": 0.027398306876420975, "train_loss_bc": 0.02737918123602867, "train_loss_llm": 0.001912534236907959, "grad_norm": 0.07227355986833572, "global_step": 17800, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03887556493282318, "train_loss_bc": 0.03887556493282318, "train_loss_llm": 0.0, "grad_norm": 0.03190363571047783, "global_step": 17801, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03930264338850975, "train_loss_bc": 0.03929074853658676, "train_loss_llm": 0.0011896193027496338, "grad_norm": 0.043592363595962524, "global_step": 17802, "epoch": 494, "lr": 0.001} +{"train_loss": 0.029582887887954712, "train_loss_bc": 0.029582887887954712, "train_loss_llm": 0.0, "grad_norm": 0.061202727258205414, "global_step": 17803, "epoch": 494, "lr": 0.001} +{"train_loss": 0.029189150780439377, "train_loss_bc": 0.029189150780439377, "train_loss_llm": 0.0, "grad_norm": 0.06868556886911392, "global_step": 17804, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03695131465792656, "train_loss_bc": 0.03695131465792656, "train_loss_llm": 0.0, "grad_norm": 0.026128366589546204, "global_step": 17805, "epoch": 494, "lr": 0.001} +{"train_loss": 0.04348728805780411, "train_loss_bc": 0.04348411783576012, "train_loss_llm": 0.00031694769859313965, "grad_norm": 0.044804662466049194, "global_step": 17806, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03294098377227783, "train_loss_bc": 0.03294098377227783, "train_loss_llm": 0.0, "grad_norm": 0.05769001692533493, "global_step": 17807, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03878023847937584, "train_loss_bc": 0.03878023847937584, "train_loss_llm": 0.0, "grad_norm": 0.0693574771285057, "global_step": 17808, "epoch": 494, "lr": 0.001} +{"train_loss": 0.037253644317388535, "train_loss_bc": 0.037253644317388535, "train_loss_llm": 0.0, "grad_norm": 0.025310015305876732, "global_step": 17809, "epoch": 494, "lr": 0.001} +{"train_loss": 0.025621233507990837, "train_loss_bc": 0.025621233507990837, "train_loss_llm": 0.0, "grad_norm": 0.032429084181785583, "global_step": 17810, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03367908298969269, "train_loss_bc": 0.03367908298969269, "train_loss_llm": 0.0, "grad_norm": 0.04615505039691925, "global_step": 17811, "epoch": 494, "lr": 0.001} +{"train_loss": 0.032456837594509125, "train_loss_bc": 0.032456837594509125, "train_loss_llm": 0.0, "grad_norm": 0.05715553089976311, "global_step": 17812, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03548251464962959, "train_loss_bc": 0.03545787185430527, "train_loss_llm": 0.0024641752243041992, "grad_norm": 0.024816075339913368, "global_step": 17813, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03270206227898598, "train_loss_bc": 0.03269250690937042, "train_loss_llm": 0.000955432653427124, "grad_norm": 0.042340949177742004, "global_step": 17814, "epoch": 494, "lr": 0.001} +{"train_loss": 0.0369020514190197, "train_loss_bc": 0.0369020514190197, "train_loss_llm": 0.0, "grad_norm": 0.05913054198026657, "global_step": 17815, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03246893733739853, "train_loss_bc": 0.03246893733739853, "train_loss_llm": 0.0, "grad_norm": 0.07019267231225967, "global_step": 17816, "epoch": 494, "lr": 0.001} +{"train_loss": 0.034676846116781235, "train_loss_bc": 0.034676846116781235, "train_loss_llm": 0.0, "grad_norm": 0.02781973034143448, "global_step": 17817, "epoch": 494, "lr": 0.001} +{"train_loss": 0.031181517988443375, "train_loss_bc": 0.031150244176387787, "train_loss_llm": 0.0031274259090423584, "grad_norm": 0.04006939008831978, "global_step": 17818, "epoch": 494, "lr": 0.001} +{"train_loss": 0.03428683171255721, "train_loss_bc": 0.03749510645866394, "train_loss_llm": 0.0, "grad_norm": 0.050156403332948685, "global_step": 17819, "epoch": 494, "lr": 0.001, "val_loss": 0.07428982853889465} +{"train_loss": 0.03745926544070244, "train_loss_bc": 0.03745926544070244, "train_loss_llm": 0.0, "grad_norm": 0.06919960677623749, "global_step": 17820, "epoch": 495, "lr": 0.001} +{"train_loss": 0.0329674631357193, "train_loss_bc": 0.0329674631357193, "train_loss_llm": 0.0, "grad_norm": 0.0247124545276165, "global_step": 17821, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03223957493901253, "train_loss_bc": 0.03223029151558876, "train_loss_llm": 0.0009284019470214844, "grad_norm": 0.03937985375523567, "global_step": 17822, "epoch": 495, "lr": 0.001} +{"train_loss": 0.033817119896411896, "train_loss_bc": 0.03380976617336273, "train_loss_llm": 0.0007352828979492188, "grad_norm": 0.05941769853234291, "global_step": 17823, "epoch": 495, "lr": 0.001} +{"train_loss": 0.029590019956231117, "train_loss_bc": 0.029590019956231117, "train_loss_llm": 0.0, "grad_norm": 0.07950849831104279, "global_step": 17824, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03155132755637169, "train_loss_bc": 0.03153308853507042, "train_loss_llm": 0.0018237531185150146, "grad_norm": 0.030735066160559654, "global_step": 17825, "epoch": 495, "lr": 0.001} +{"train_loss": 0.0335034504532814, "train_loss_bc": 0.0335034504532814, "train_loss_llm": 0.0, "grad_norm": 0.03964420408010483, "global_step": 17826, "epoch": 495, "lr": 0.001} +{"train_loss": 0.032492734491825104, "train_loss_bc": 0.032492734491825104, "train_loss_llm": 0.0, "grad_norm": 0.05613235384225845, "global_step": 17827, "epoch": 495, "lr": 0.001} +{"train_loss": 0.036624472588300705, "train_loss_bc": 0.03660503774881363, "train_loss_llm": 0.0019433200359344482, "grad_norm": 0.06578297168016434, "global_step": 17828, "epoch": 495, "lr": 0.001} +{"train_loss": 0.0411628894507885, "train_loss_bc": 0.0411628894507885, "train_loss_llm": 0.0, "grad_norm": 0.030626261606812477, "global_step": 17829, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03619220480322838, "train_loss_bc": 0.03618732467293739, "train_loss_llm": 0.00048801302909851074, "grad_norm": 0.04176870360970497, "global_step": 17830, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03715386614203453, "train_loss_bc": 0.03715386614203453, "train_loss_llm": 0.0, "grad_norm": 0.0526123046875, "global_step": 17831, "epoch": 495, "lr": 0.001} +{"train_loss": 0.036778248846530914, "train_loss_bc": 0.036778248846530914, "train_loss_llm": 0.0, "grad_norm": 0.06036432459950447, "global_step": 17832, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03255484253168106, "train_loss_bc": 0.03255484253168106, "train_loss_llm": 0.0, "grad_norm": 0.02506500482559204, "global_step": 17833, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03707412630319595, "train_loss_bc": 0.03707412630319595, "train_loss_llm": 0.0, "grad_norm": 0.0354899987578392, "global_step": 17834, "epoch": 495, "lr": 0.001} +{"train_loss": 0.029943889006972313, "train_loss_bc": 0.029943889006972313, "train_loss_llm": 0.0, "grad_norm": 0.048347387462854385, "global_step": 17835, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03434956073760986, "train_loss_bc": 0.034347034990787506, "train_loss_llm": 0.0002525448799133301, "grad_norm": 0.0571906678378582, "global_step": 17836, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03148442506790161, "train_loss_bc": 0.03147593140602112, "train_loss_llm": 0.0008494853973388672, "grad_norm": 0.021482408046722412, "global_step": 17837, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03646482899785042, "train_loss_bc": 0.03645400330424309, "train_loss_llm": 0.0010825693607330322, "grad_norm": 0.036495234817266464, "global_step": 17838, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03935755044221878, "train_loss_bc": 0.03935755044221878, "train_loss_llm": 0.0, "grad_norm": 0.04911770671606064, "global_step": 17839, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03420991450548172, "train_loss_bc": 0.03420991450548172, "train_loss_llm": 0.0, "grad_norm": 0.05647001415491104, "global_step": 17840, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03431157395243645, "train_loss_bc": 0.034297823905944824, "train_loss_llm": 0.0013748705387115479, "grad_norm": 0.027393391355872154, "global_step": 17841, "epoch": 495, "lr": 0.001} +{"train_loss": 0.028841745108366013, "train_loss_bc": 0.028841745108366013, "train_loss_llm": 0.0, "grad_norm": 0.04678208380937576, "global_step": 17842, "epoch": 495, "lr": 0.001} +{"train_loss": 0.037492237985134125, "train_loss_bc": 0.037492237985134125, "train_loss_llm": 0.0, "grad_norm": 0.053453583270311356, "global_step": 17843, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03456661105155945, "train_loss_bc": 0.03455336391925812, "train_loss_llm": 0.0013245940208435059, "grad_norm": 0.07395651936531067, "global_step": 17844, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03097020462155342, "train_loss_bc": 0.03097020462155342, "train_loss_llm": 0.0, "grad_norm": 0.02888152003288269, "global_step": 17845, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03416381776332855, "train_loss_bc": 0.03416381776332855, "train_loss_llm": 0.0, "grad_norm": 0.0405702218413353, "global_step": 17846, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03163701295852661, "train_loss_bc": 0.03163701295852661, "train_loss_llm": 0.0, "grad_norm": 0.050184253603219986, "global_step": 17847, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03280993178486824, "train_loss_bc": 0.03280993178486824, "train_loss_llm": 0.0, "grad_norm": 0.06868526339530945, "global_step": 17848, "epoch": 495, "lr": 0.001} +{"train_loss": 0.034260235726833344, "train_loss_bc": 0.034260235726833344, "train_loss_llm": 0.0, "grad_norm": 0.026168648153543472, "global_step": 17849, "epoch": 495, "lr": 0.001} +{"train_loss": 0.026777513325214386, "train_loss_bc": 0.026777513325214386, "train_loss_llm": 0.0, "grad_norm": 0.040137503296136856, "global_step": 17850, "epoch": 495, "lr": 0.001} +{"train_loss": 0.029551535844802856, "train_loss_bc": 0.029551535844802856, "train_loss_llm": 0.0, "grad_norm": 0.05700664222240448, "global_step": 17851, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03604963421821594, "train_loss_bc": 0.03604963421821594, "train_loss_llm": 0.0, "grad_norm": 0.0726599395275116, "global_step": 17852, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03921600803732872, "train_loss_bc": 0.03921600803732872, "train_loss_llm": 0.0, "grad_norm": 0.024636607617139816, "global_step": 17853, "epoch": 495, "lr": 0.001} +{"train_loss": 0.032674603164196014, "train_loss_bc": 0.032674603164196014, "train_loss_llm": 0.0, "grad_norm": 0.032630160450935364, "global_step": 17854, "epoch": 495, "lr": 0.001} +{"train_loss": 0.03421242410937945, "train_loss_bc": 0.04135282710194588, "train_loss_llm": 0.0, "grad_norm": 0.04735277593135834, "global_step": 17855, "epoch": 495, "lr": 0.001, "train/cumulative_reward": 66.0879691945252, "train/mean_score": 11.803433840318343, "train/success_rate": 0.3, "test/cumulative_reward": 37.304819241466, "test/mean_score": 7.7969646074342105, "test/success_rate": 0.2, "val_loss": 0.07394279539585114, "train_action_mse_error": 0.03697832301259041} +{"train_loss": 0.038166169077157974, "train_loss_bc": 0.038166169077157974, "train_loss_llm": 0.0, "grad_norm": 0.06132776290178299, "global_step": 17856, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03148879483342171, "train_loss_bc": 0.03148879483342171, "train_loss_llm": 0.0, "grad_norm": 0.024103408679366112, "global_step": 17857, "epoch": 496, "lr": 0.001} +{"train_loss": 0.035746797919273376, "train_loss_bc": 0.035746797919273376, "train_loss_llm": 0.0, "grad_norm": 0.03372880816459656, "global_step": 17858, "epoch": 496, "lr": 0.001} +{"train_loss": 0.032172586768865585, "train_loss_bc": 0.032172586768865585, "train_loss_llm": 0.0, "grad_norm": 0.03782884031534195, "global_step": 17859, "epoch": 496, "lr": 0.001} +{"train_loss": 0.0314151905477047, "train_loss_bc": 0.0314151905477047, "train_loss_llm": 0.0, "grad_norm": 0.04814765602350235, "global_step": 17860, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03304482623934746, "train_loss_bc": 0.03304482623934746, "train_loss_llm": 0.0, "grad_norm": 0.027947720140218735, "global_step": 17861, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03893309086561203, "train_loss_bc": 0.03893309086561203, "train_loss_llm": 0.0, "grad_norm": 0.03860218822956085, "global_step": 17862, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03624650090932846, "train_loss_bc": 0.03624650090932846, "train_loss_llm": 0.0, "grad_norm": 0.05605106055736542, "global_step": 17863, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03552175685763359, "train_loss_bc": 0.03552175685763359, "train_loss_llm": 0.0, "grad_norm": 0.06697418540716171, "global_step": 17864, "epoch": 496, "lr": 0.001} +{"train_loss": 0.035504184663295746, "train_loss_bc": 0.035504184663295746, "train_loss_llm": 0.0, "grad_norm": 0.024135075509548187, "global_step": 17865, "epoch": 496, "lr": 0.001} +{"train_loss": 0.038254573941230774, "train_loss_bc": 0.03824080899357796, "train_loss_llm": 0.0013764798641204834, "grad_norm": 0.0436103455722332, "global_step": 17866, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03590349480509758, "train_loss_bc": 0.03590349480509758, "train_loss_llm": 0.0, "grad_norm": 0.06432242691516876, "global_step": 17867, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03351278603076935, "train_loss_bc": 0.03351278603076935, "train_loss_llm": 0.0, "grad_norm": 0.08167687058448792, "global_step": 17868, "epoch": 496, "lr": 0.001} +{"train_loss": 0.029506005346775055, "train_loss_bc": 0.029483815655112267, "train_loss_llm": 0.002218902111053467, "grad_norm": 0.02522781491279602, "global_step": 17869, "epoch": 496, "lr": 0.001} +{"train_loss": 0.033436574041843414, "train_loss_bc": 0.033436574041843414, "train_loss_llm": 0.0, "grad_norm": 0.042352620512247086, "global_step": 17870, "epoch": 496, "lr": 0.001} +{"train_loss": 0.031084056943655014, "train_loss_bc": 0.031084056943655014, "train_loss_llm": 0.0, "grad_norm": 0.05516592785716057, "global_step": 17871, "epoch": 496, "lr": 0.001} +{"train_loss": 0.030278215184807777, "train_loss_bc": 0.030274957418441772, "train_loss_llm": 0.00032585859298706055, "grad_norm": 0.070281982421875, "global_step": 17872, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03335075080394745, "train_loss_bc": 0.03333757072687149, "train_loss_llm": 0.0013180673122406006, "grad_norm": 0.02406333014369011, "global_step": 17873, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03479738533496857, "train_loss_bc": 0.03479738533496857, "train_loss_llm": 0.0, "grad_norm": 0.03853025659918785, "global_step": 17874, "epoch": 496, "lr": 0.001} +{"train_loss": 0.038224656134843826, "train_loss_bc": 0.038224656134843826, "train_loss_llm": 0.0, "grad_norm": 0.05513283237814903, "global_step": 17875, "epoch": 496, "lr": 0.001} +{"train_loss": 0.0395108200609684, "train_loss_bc": 0.0395108200609684, "train_loss_llm": 0.0, "grad_norm": 0.0692223459482193, "global_step": 17876, "epoch": 496, "lr": 0.001} +{"train_loss": 0.036808427423238754, "train_loss_bc": 0.036808427423238754, "train_loss_llm": 0.0, "grad_norm": 0.02925998717546463, "global_step": 17877, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03485957533121109, "train_loss_bc": 0.03485957533121109, "train_loss_llm": 0.0, "grad_norm": 0.04970264434814453, "global_step": 17878, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03540902957320213, "train_loss_bc": 0.03540902957320213, "train_loss_llm": 0.0, "grad_norm": 0.07047142088413239, "global_step": 17879, "epoch": 496, "lr": 0.001} +{"train_loss": 0.028532307595014572, "train_loss_bc": 0.028532307595014572, "train_loss_llm": 0.0, "grad_norm": 0.08600097894668579, "global_step": 17880, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03600602224469185, "train_loss_bc": 0.03600602224469185, "train_loss_llm": 0.0, "grad_norm": 0.033809173852205276, "global_step": 17881, "epoch": 496, "lr": 0.001} +{"train_loss": 0.030412258580327034, "train_loss_bc": 0.030387697741389275, "train_loss_llm": 0.0024560093879699707, "grad_norm": 0.05054367706179619, "global_step": 17882, "epoch": 496, "lr": 0.001} +{"train_loss": 0.031201336532831192, "train_loss_bc": 0.031194990500807762, "train_loss_llm": 0.0006345510482788086, "grad_norm": 0.054601311683654785, "global_step": 17883, "epoch": 496, "lr": 0.001} +{"train_loss": 0.035267338156700134, "train_loss_bc": 0.035267338156700134, "train_loss_llm": 0.0, "grad_norm": 0.06365924328565598, "global_step": 17884, "epoch": 496, "lr": 0.001} +{"train_loss": 0.029495874419808388, "train_loss_bc": 0.029495874419808388, "train_loss_llm": 0.0, "grad_norm": 0.027683408930897713, "global_step": 17885, "epoch": 496, "lr": 0.001} +{"train_loss": 0.032042987644672394, "train_loss_bc": 0.032042987644672394, "train_loss_llm": 0.0, "grad_norm": 0.043506741523742676, "global_step": 17886, "epoch": 496, "lr": 0.001} +{"train_loss": 0.03749928995966911, "train_loss_bc": 0.037490103393793106, "train_loss_llm": 0.0009187161922454834, "grad_norm": 0.057752467691898346, "global_step": 17887, "epoch": 496, "lr": 0.001} +{"train_loss": 0.038148198276758194, "train_loss_bc": 0.038148198276758194, "train_loss_llm": 0.0, "grad_norm": 0.07291816174983978, "global_step": 17888, "epoch": 496, "lr": 0.001} +{"train_loss": 0.029829101637005806, "train_loss_bc": 0.029829101637005806, "train_loss_llm": 0.0, "grad_norm": 0.026278581470251083, "global_step": 17889, "epoch": 496, "lr": 0.001} +{"train_loss": 0.033491138368844986, "train_loss_bc": 0.03348353132605553, "train_loss_llm": 0.000760883092880249, "grad_norm": 0.04604075849056244, "global_step": 17890, "epoch": 496, "lr": 0.001} +{"train_loss": 0.034143074726065, "train_loss_bc": 0.03404858708381653, "train_loss_llm": 0.0, "grad_norm": 0.06804519146680832, "global_step": 17891, "epoch": 496, "lr": 0.001, "val_loss": 0.07620339095592499} +{"train_loss": 0.028131840750575066, "train_loss_bc": 0.028131840750575066, "train_loss_llm": 0.0, "grad_norm": 0.0867721438407898, "global_step": 17892, "epoch": 497, "lr": 0.001} +{"train_loss": 0.026892604306340218, "train_loss_bc": 0.026892604306340218, "train_loss_llm": 0.0, "grad_norm": 0.025675833225250244, "global_step": 17893, "epoch": 497, "lr": 0.001} +{"train_loss": 0.034913673996925354, "train_loss_bc": 0.034913673996925354, "train_loss_llm": 0.0, "grad_norm": 0.057954173535108566, "global_step": 17894, "epoch": 497, "lr": 0.001} +{"train_loss": 0.026950320228934288, "train_loss_bc": 0.026950320228934288, "train_loss_llm": 0.0, "grad_norm": 0.06990157812833786, "global_step": 17895, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03707186132669449, "train_loss_bc": 0.03707186132669449, "train_loss_llm": 0.0, "grad_norm": 0.07967878133058548, "global_step": 17896, "epoch": 497, "lr": 0.001} +{"train_loss": 0.033228568732738495, "train_loss_bc": 0.033206623047590256, "train_loss_llm": 0.002194434404373169, "grad_norm": 0.024913819506764412, "global_step": 17897, "epoch": 497, "lr": 0.001} +{"train_loss": 0.0395556204020977, "train_loss_bc": 0.0395556204020977, "train_loss_llm": 0.0, "grad_norm": 0.04255634918808937, "global_step": 17898, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03497292846441269, "train_loss_bc": 0.03497292846441269, "train_loss_llm": 0.0, "grad_norm": 0.05911002680659294, "global_step": 17899, "epoch": 497, "lr": 0.001} +{"train_loss": 0.04281037300825119, "train_loss_bc": 0.04281037300825119, "train_loss_llm": 0.0, "grad_norm": 0.07157152891159058, "global_step": 17900, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03248219192028046, "train_loss_bc": 0.03248219192028046, "train_loss_llm": 0.0, "grad_norm": 0.02340473234653473, "global_step": 17901, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03317805379629135, "train_loss_bc": 0.03317805379629135, "train_loss_llm": 0.0, "grad_norm": 0.03300170972943306, "global_step": 17902, "epoch": 497, "lr": 0.001} +{"train_loss": 0.034958913922309875, "train_loss_bc": 0.034958913922309875, "train_loss_llm": 0.0, "grad_norm": 0.05542459338903427, "global_step": 17903, "epoch": 497, "lr": 0.001} +{"train_loss": 0.0329647958278656, "train_loss_bc": 0.0329647958278656, "train_loss_llm": 0.0, "grad_norm": 0.06799699366092682, "global_step": 17904, "epoch": 497, "lr": 0.001} +{"train_loss": 0.036767445504665375, "train_loss_bc": 0.03675515204668045, "train_loss_llm": 0.001229405403137207, "grad_norm": 0.02700977772474289, "global_step": 17905, "epoch": 497, "lr": 0.001} +{"train_loss": 0.041198380291461945, "train_loss_bc": 0.041198380291461945, "train_loss_llm": 0.0, "grad_norm": 0.04377951845526695, "global_step": 17906, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03170709311962128, "train_loss_bc": 0.03170709311962128, "train_loss_llm": 0.0, "grad_norm": 0.06187734752893448, "global_step": 17907, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03307894617319107, "train_loss_bc": 0.03307894617319107, "train_loss_llm": 0.0, "grad_norm": 0.07873202115297318, "global_step": 17908, "epoch": 497, "lr": 0.001} +{"train_loss": 0.029504209756851196, "train_loss_bc": 0.029504209756851196, "train_loss_llm": 0.0, "grad_norm": 0.027302583679556847, "global_step": 17909, "epoch": 497, "lr": 0.001} +{"train_loss": 0.035063911229372025, "train_loss_bc": 0.035063911229372025, "train_loss_llm": 0.0, "grad_norm": 0.05290127918124199, "global_step": 17910, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03760800138115883, "train_loss_bc": 0.03760800138115883, "train_loss_llm": 0.0, "grad_norm": 0.07359188050031662, "global_step": 17911, "epoch": 497, "lr": 0.001} +{"train_loss": 0.031633611768484116, "train_loss_bc": 0.031633611768484116, "train_loss_llm": 0.0, "grad_norm": 0.0871763601899147, "global_step": 17912, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03794426470994949, "train_loss_bc": 0.03794426470994949, "train_loss_llm": 0.0, "grad_norm": 0.029742322862148285, "global_step": 17913, "epoch": 497, "lr": 0.001} +{"train_loss": 0.04091885685920715, "train_loss_bc": 0.04091885685920715, "train_loss_llm": 0.0, "grad_norm": 0.051805365830659866, "global_step": 17914, "epoch": 497, "lr": 0.001} +{"train_loss": 0.04063957557082176, "train_loss_bc": 0.04061849042773247, "train_loss_llm": 0.002108544111251831, "grad_norm": 0.06433077901601791, "global_step": 17915, "epoch": 497, "lr": 0.001} +{"train_loss": 0.030123677104711533, "train_loss_bc": 0.030123677104711533, "train_loss_llm": 0.0, "grad_norm": 0.08333023637533188, "global_step": 17916, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03294289484620094, "train_loss_bc": 0.03291912004351616, "train_loss_llm": 0.002377539873123169, "grad_norm": 0.030078113079071045, "global_step": 17917, "epoch": 497, "lr": 0.001} +{"train_loss": 0.030088944360613823, "train_loss_bc": 0.03004380315542221, "train_loss_llm": 0.004514187574386597, "grad_norm": 0.037884216755628586, "global_step": 17918, "epoch": 497, "lr": 0.001} +{"train_loss": 0.026827823370695114, "train_loss_bc": 0.026827823370695114, "train_loss_llm": 0.0, "grad_norm": 0.05803484097123146, "global_step": 17919, "epoch": 497, "lr": 0.001} +{"train_loss": 0.036490581929683685, "train_loss_bc": 0.036490581929683685, "train_loss_llm": 0.0, "grad_norm": 0.07755609601736069, "global_step": 17920, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03185058385133743, "train_loss_bc": 0.03183149918913841, "train_loss_llm": 0.0019084811210632324, "grad_norm": 0.026892099529504776, "global_step": 17921, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03338916599750519, "train_loss_bc": 0.03338916599750519, "train_loss_llm": 0.0, "grad_norm": 0.038050577044487, "global_step": 17922, "epoch": 497, "lr": 0.001} +{"train_loss": 0.035701602697372437, "train_loss_bc": 0.035701602697372437, "train_loss_llm": 0.0, "grad_norm": 0.05282944440841675, "global_step": 17923, "epoch": 497, "lr": 0.001} +{"train_loss": 0.032806020230054855, "train_loss_bc": 0.032806020230054855, "train_loss_llm": 0.0, "grad_norm": 0.055280737578868866, "global_step": 17924, "epoch": 497, "lr": 0.001} +{"train_loss": 0.038965433835983276, "train_loss_bc": 0.038965433835983276, "train_loss_llm": 0.0, "grad_norm": 0.03337788209319115, "global_step": 17925, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03754807636141777, "train_loss_bc": 0.03754807636141777, "train_loss_llm": 0.0, "grad_norm": 0.04175259917974472, "global_step": 17926, "epoch": 497, "lr": 0.001} +{"train_loss": 0.03420234197336766, "train_loss_bc": 0.030359605327248573, "train_loss_llm": 0.0013858973979949951, "grad_norm": 0.05339249595999718, "global_step": 17927, "epoch": 497, "lr": 0.001, "val_loss": 0.07450591772794724} +{"train_loss": 0.033155303448438644, "train_loss_bc": 0.033135753124952316, "train_loss_llm": 0.0019549429416656494, "grad_norm": 0.06327206641435623, "global_step": 17928, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03524099290370941, "train_loss_bc": 0.03524099290370941, "train_loss_llm": 0.0, "grad_norm": 0.02858871780335903, "global_step": 17929, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03789297863841057, "train_loss_bc": 0.03789297863841057, "train_loss_llm": 0.0, "grad_norm": 0.05594848841428757, "global_step": 17930, "epoch": 498, "lr": 0.001} +{"train_loss": 0.029808927327394485, "train_loss_bc": 0.029808927327394485, "train_loss_llm": 0.0, "grad_norm": 0.07227206230163574, "global_step": 17931, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03417455404996872, "train_loss_bc": 0.03417455404996872, "train_loss_llm": 0.0, "grad_norm": 0.08404185622930527, "global_step": 17932, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03559417650103569, "train_loss_bc": 0.03559417650103569, "train_loss_llm": 0.0, "grad_norm": 0.02684124931693077, "global_step": 17933, "epoch": 498, "lr": 0.001} +{"train_loss": 0.04040246456861496, "train_loss_bc": 0.04038981348276138, "train_loss_llm": 0.0012651383876800537, "grad_norm": 0.04890637472271919, "global_step": 17934, "epoch": 498, "lr": 0.001} +{"train_loss": 0.032722413539886475, "train_loss_bc": 0.032722413539886475, "train_loss_llm": 0.0, "grad_norm": 0.059995055198669434, "global_step": 17935, "epoch": 498, "lr": 0.001} +{"train_loss": 0.02867002971470356, "train_loss_bc": 0.02867002971470356, "train_loss_llm": 0.0, "grad_norm": 0.07032528519630432, "global_step": 17936, "epoch": 498, "lr": 0.001} +{"train_loss": 0.02845297008752823, "train_loss_bc": 0.02845297008752823, "train_loss_llm": 0.0, "grad_norm": 0.032784152776002884, "global_step": 17937, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03807981312274933, "train_loss_bc": 0.03807981312274933, "train_loss_llm": 0.0, "grad_norm": 0.06507409363985062, "global_step": 17938, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03139696642756462, "train_loss_bc": 0.03139268606901169, "train_loss_llm": 0.00042814016342163086, "grad_norm": 0.08175060153007507, "global_step": 17939, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03176354244351387, "train_loss_bc": 0.03176354244351387, "train_loss_llm": 0.0, "grad_norm": 0.10379860550165176, "global_step": 17940, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03382583335042, "train_loss_bc": 0.033822450786828995, "train_loss_llm": 0.0003381669521331787, "grad_norm": 0.029761886224150658, "global_step": 17941, "epoch": 498, "lr": 0.001} +{"train_loss": 0.036480218172073364, "train_loss_bc": 0.036480218172073364, "train_loss_llm": 0.0, "grad_norm": 0.04351712390780449, "global_step": 17942, "epoch": 498, "lr": 0.001} +{"train_loss": 0.033236946910619736, "train_loss_bc": 0.033236946910619736, "train_loss_llm": 0.0, "grad_norm": 0.05614669993519783, "global_step": 17943, "epoch": 498, "lr": 0.001} +{"train_loss": 0.033839207142591476, "train_loss_bc": 0.033839207142591476, "train_loss_llm": 0.0, "grad_norm": 0.07063787430524826, "global_step": 17944, "epoch": 498, "lr": 0.001} +{"train_loss": 0.04031956568360329, "train_loss_bc": 0.04031465947628021, "train_loss_llm": 0.0004906654357910156, "grad_norm": 0.03474976867437363, "global_step": 17945, "epoch": 498, "lr": 0.001} +{"train_loss": 0.0340019054710865, "train_loss_bc": 0.0340019054710865, "train_loss_llm": 0.0, "grad_norm": 0.05259579047560692, "global_step": 17946, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03815903142094612, "train_loss_bc": 0.03815903142094612, "train_loss_llm": 0.0, "grad_norm": 0.07539594173431396, "global_step": 17947, "epoch": 498, "lr": 0.001} +{"train_loss": 0.031142082065343857, "train_loss_bc": 0.031142082065343857, "train_loss_llm": 0.0, "grad_norm": 0.08399514108896255, "global_step": 17948, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03898027911782265, "train_loss_bc": 0.03898027911782265, "train_loss_llm": 0.0, "grad_norm": 0.03360278531908989, "global_step": 17949, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03374405950307846, "train_loss_bc": 0.033726006746292114, "train_loss_llm": 0.0018053948879241943, "grad_norm": 0.04270055145025253, "global_step": 17950, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03469891846179962, "train_loss_bc": 0.03469891846179962, "train_loss_llm": 0.0, "grad_norm": 0.053502097725868225, "global_step": 17951, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03581508994102478, "train_loss_bc": 0.03581508994102478, "train_loss_llm": 0.0, "grad_norm": 0.0665760263800621, "global_step": 17952, "epoch": 498, "lr": 0.001} +{"train_loss": 0.031252212822437286, "train_loss_bc": 0.031252212822437286, "train_loss_llm": 0.0, "grad_norm": 0.02387009933590889, "global_step": 17953, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03615190088748932, "train_loss_bc": 0.03615190088748932, "train_loss_llm": 0.0, "grad_norm": 0.04270252212882042, "global_step": 17954, "epoch": 498, "lr": 0.001} +{"train_loss": 0.037289850413799286, "train_loss_bc": 0.037289850413799286, "train_loss_llm": 0.0, "grad_norm": 0.06380317360162735, "global_step": 17955, "epoch": 498, "lr": 0.001} +{"train_loss": 0.03651557117700577, "train_loss_bc": 0.03648180514574051, "train_loss_llm": 0.0033764541149139404, "grad_norm": 0.08019214868545532, "global_step": 17956, "epoch": 498, "lr": 0.001} +{"train_loss": 0.0328286774456501, "train_loss_bc": 0.0328286774456501, "train_loss_llm": 0.0, "grad_norm": 0.031612660735845566, "global_step": 17957, "epoch": 498, "lr": 0.001} +{"train_loss": 0.032885923981666565, "train_loss_bc": 0.032885923981666565, "train_loss_llm": 0.0, "grad_norm": 0.05469079688191414, "global_step": 17958, "epoch": 498, "lr": 0.001} +{"train_loss": 0.039011288434267044, "train_loss_bc": 0.039011288434267044, "train_loss_llm": 0.0, "grad_norm": 0.07038183510303497, "global_step": 17959, "epoch": 498, "lr": 0.001} +{"train_loss": 0.031182341277599335, "train_loss_bc": 0.031182341277599335, "train_loss_llm": 0.0, "grad_norm": 0.08813628554344177, "global_step": 17960, "epoch": 498, "lr": 0.001} +{"train_loss": 0.0316811129450798, "train_loss_bc": 0.0316811129450798, "train_loss_llm": 0.0, "grad_norm": 0.0247753094881773, "global_step": 17961, "epoch": 498, "lr": 0.001} +{"train_loss": 0.031680889427661896, "train_loss_bc": 0.031676847487688065, "train_loss_llm": 0.00040435791015625, "grad_norm": 0.03883403167128563, "global_step": 17962, "epoch": 498, "lr": 0.001} +{"train_loss": 0.034320296047048435, "train_loss_bc": 0.03345261886715889, "train_loss_llm": 0.0, "grad_norm": 0.06341533362865448, "global_step": 17963, "epoch": 498, "lr": 0.001, "val_loss": 0.07395961880683899} +{"train_loss": 0.020476052537560463, "train_loss_bc": 0.020476052537560463, "train_loss_llm": 0.0, "grad_norm": 0.08119101077318192, "global_step": 17964, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03939297795295715, "train_loss_bc": 0.03939297795295715, "train_loss_llm": 0.0, "grad_norm": 0.030351243913173676, "global_step": 17965, "epoch": 499, "lr": 0.001} +{"train_loss": 0.0367828905582428, "train_loss_bc": 0.03676662594079971, "train_loss_llm": 0.0016263127326965332, "grad_norm": 0.04871068522334099, "global_step": 17966, "epoch": 499, "lr": 0.001} +{"train_loss": 0.033093612641096115, "train_loss_bc": 0.033093612641096115, "train_loss_llm": 0.0, "grad_norm": 0.0647619441151619, "global_step": 17967, "epoch": 499, "lr": 0.001} +{"train_loss": 0.034878481179475784, "train_loss_bc": 0.034878481179475784, "train_loss_llm": 0.0, "grad_norm": 0.07896532863378525, "global_step": 17968, "epoch": 499, "lr": 0.001} +{"train_loss": 0.033050380647182465, "train_loss_bc": 0.03304874151945114, "train_loss_llm": 0.00016388297080993652, "grad_norm": 0.02724163793027401, "global_step": 17969, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03486492484807968, "train_loss_bc": 0.03486492484807968, "train_loss_llm": 0.0, "grad_norm": 0.04359787330031395, "global_step": 17970, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03243159130215645, "train_loss_bc": 0.03243159130215645, "train_loss_llm": 0.0, "grad_norm": 0.04608975723385811, "global_step": 17971, "epoch": 499, "lr": 0.001} +{"train_loss": 0.040465760976076126, "train_loss_bc": 0.040465760976076126, "train_loss_llm": 0.0, "grad_norm": 0.05359833315014839, "global_step": 17972, "epoch": 499, "lr": 0.001} +{"train_loss": 0.0370238721370697, "train_loss_bc": 0.0370238721370697, "train_loss_llm": 0.0, "grad_norm": 0.028252167627215385, "global_step": 17973, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03721132129430771, "train_loss_bc": 0.03720209747552872, "train_loss_llm": 0.0009222328662872314, "grad_norm": 0.04315691813826561, "global_step": 17974, "epoch": 499, "lr": 0.001} +{"train_loss": 0.0401168130338192, "train_loss_bc": 0.0401168130338192, "train_loss_llm": 0.0, "grad_norm": 0.05874178558588028, "global_step": 17975, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03813907131552696, "train_loss_bc": 0.038137536495923996, "train_loss_llm": 0.00015336275100708008, "grad_norm": 0.07008504867553711, "global_step": 17976, "epoch": 499, "lr": 0.001} +{"train_loss": 0.039154522120952606, "train_loss_bc": 0.039137136191129684, "train_loss_llm": 0.00173872709274292, "grad_norm": 0.029323415830731392, "global_step": 17977, "epoch": 499, "lr": 0.001} +{"train_loss": 0.031247230246663094, "train_loss_bc": 0.03124348446726799, "train_loss_llm": 0.0003746449947357178, "grad_norm": 0.03710125759243965, "global_step": 17978, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03226018324494362, "train_loss_bc": 0.03226018324494362, "train_loss_llm": 0.0, "grad_norm": 0.05095299333333969, "global_step": 17979, "epoch": 499, "lr": 0.001} +{"train_loss": 0.02722945809364319, "train_loss_bc": 0.02722945809364319, "train_loss_llm": 0.0, "grad_norm": 0.058102916926145554, "global_step": 17980, "epoch": 499, "lr": 0.001} +{"train_loss": 0.029009142890572548, "train_loss_bc": 0.029009142890572548, "train_loss_llm": 0.0, "grad_norm": 0.024077320471405983, "global_step": 17981, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03416839987039566, "train_loss_bc": 0.03416839987039566, "train_loss_llm": 0.0, "grad_norm": 0.0334727019071579, "global_step": 17982, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03181823343038559, "train_loss_bc": 0.03181823343038559, "train_loss_llm": 0.0, "grad_norm": 0.05013638734817505, "global_step": 17983, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03194275498390198, "train_loss_bc": 0.031928785145282745, "train_loss_llm": 0.001396864652633667, "grad_norm": 0.06763145327568054, "global_step": 17984, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03373066708445549, "train_loss_bc": 0.03373066708445549, "train_loss_llm": 0.0, "grad_norm": 0.027501866221427917, "global_step": 17985, "epoch": 499, "lr": 0.001} +{"train_loss": 0.029400624334812164, "train_loss_bc": 0.029400624334812164, "train_loss_llm": 0.0, "grad_norm": 0.04278741404414177, "global_step": 17986, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03252776712179184, "train_loss_bc": 0.03252776712179184, "train_loss_llm": 0.0, "grad_norm": 0.055395714938640594, "global_step": 17987, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03440411761403084, "train_loss_bc": 0.03438469022512436, "train_loss_llm": 0.0019426047801971436, "grad_norm": 0.07548316568136215, "global_step": 17988, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03781447187066078, "train_loss_bc": 0.03781447187066078, "train_loss_llm": 0.0, "grad_norm": 0.025199944153428078, "global_step": 17989, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03144275024533272, "train_loss_bc": 0.03144275024533272, "train_loss_llm": 0.0, "grad_norm": 0.03316495195031166, "global_step": 17990, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03613324463367462, "train_loss_bc": 0.03613324463367462, "train_loss_llm": 0.0, "grad_norm": 0.048033472150564194, "global_step": 17991, "epoch": 499, "lr": 0.001} +{"train_loss": 0.033698469400405884, "train_loss_bc": 0.033698469400405884, "train_loss_llm": 0.0, "grad_norm": 0.05914781615138054, "global_step": 17992, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03524814918637276, "train_loss_bc": 0.035197410732507706, "train_loss_llm": 0.005073964595794678, "grad_norm": 0.02785886637866497, "global_step": 17993, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03095274604856968, "train_loss_bc": 0.03095274604856968, "train_loss_llm": 0.0, "grad_norm": 0.046781428158283234, "global_step": 17994, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03587362542748451, "train_loss_bc": 0.03587362542748451, "train_loss_llm": 0.0, "grad_norm": 0.07194705307483673, "global_step": 17995, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03515299782156944, "train_loss_bc": 0.03515299782156944, "train_loss_llm": 0.0, "grad_norm": 0.09068889915943146, "global_step": 17996, "epoch": 499, "lr": 0.001} +{"train_loss": 0.03386503458023071, "train_loss_bc": 0.03386503458023071, "train_loss_llm": 0.0, "grad_norm": 0.026668386533856392, "global_step": 17997, "epoch": 499, "lr": 0.001} +{"train_loss": 0.035860445350408554, "train_loss_bc": 0.035860445350408554, "train_loss_llm": 0.0, "grad_norm": 0.0512753427028656, "global_step": 17998, "epoch": 499, "lr": 0.001} +{"train_loss": 0.034289436828758985, "train_loss_bc": 0.04355693981051445, "train_loss_llm": 0.0, "grad_norm": 0.07202430814504623, "global_step": 17999, "epoch": 499, "lr": 0.001, "val_loss": 0.07503089308738708} +{"train_loss": 0.03393750265240669, "train_loss_bc": 0.03393750265240669, "train_loss_llm": 0.0, "grad_norm": 0.08855319023132324, "global_step": 18000, "epoch": 500, "lr": 0.001} +{"train_loss": 0.029876988381147385, "train_loss_bc": 0.029860172420740128, "train_loss_llm": 0.0016816556453704834, "grad_norm": 0.02309146150946617, "global_step": 18001, "epoch": 500, "lr": 0.001} +{"train_loss": 0.02721318230032921, "train_loss_bc": 0.02721318230032921, "train_loss_llm": 0.0, "grad_norm": 0.03014211915433407, "global_step": 18002, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03846637159585953, "train_loss_bc": 0.03846637159585953, "train_loss_llm": 0.0, "grad_norm": 0.04528224468231201, "global_step": 18003, "epoch": 500, "lr": 0.001} +{"train_loss": 0.032133202999830246, "train_loss_bc": 0.032126300036907196, "train_loss_llm": 0.0006902217864990234, "grad_norm": 0.060891252011060715, "global_step": 18004, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03406485542654991, "train_loss_bc": 0.03406485542654991, "train_loss_llm": 0.0, "grad_norm": 0.02793852426111698, "global_step": 18005, "epoch": 500, "lr": 0.001} +{"train_loss": 0.031274277716875076, "train_loss_bc": 0.031270965933799744, "train_loss_llm": 0.00033104419708251953, "grad_norm": 0.04703140631318092, "global_step": 18006, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03370863199234009, "train_loss_bc": 0.03370863199234009, "train_loss_llm": 0.0, "grad_norm": 0.05203843116760254, "global_step": 18007, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03591158241033554, "train_loss_bc": 0.03591158241033554, "train_loss_llm": 0.0, "grad_norm": 0.0658920556306839, "global_step": 18008, "epoch": 500, "lr": 0.001} +{"train_loss": 0.04005032032728195, "train_loss_bc": 0.04005032032728195, "train_loss_llm": 0.0, "grad_norm": 0.03305082768201828, "global_step": 18009, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03439255803823471, "train_loss_bc": 0.03439255803823471, "train_loss_llm": 0.0, "grad_norm": 0.049619875848293304, "global_step": 18010, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03908770531415939, "train_loss_bc": 0.03908770531415939, "train_loss_llm": 0.0, "grad_norm": 0.063633032143116, "global_step": 18011, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03067093715071678, "train_loss_bc": 0.03067093715071678, "train_loss_llm": 0.0, "grad_norm": 0.06948165595531464, "global_step": 18012, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03683410584926605, "train_loss_bc": 0.0368107408285141, "train_loss_llm": 0.002336353063583374, "grad_norm": 0.02880782075226307, "global_step": 18013, "epoch": 500, "lr": 0.001} +{"train_loss": 0.033083442598581314, "train_loss_bc": 0.03306345269083977, "train_loss_llm": 0.0019988715648651123, "grad_norm": 0.04625962674617767, "global_step": 18014, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03371214494109154, "train_loss_bc": 0.03370808809995651, "train_loss_llm": 0.0004056990146636963, "grad_norm": 0.05987093970179558, "global_step": 18015, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03336789831519127, "train_loss_bc": 0.03336789831519127, "train_loss_llm": 0.0, "grad_norm": 0.08466988056898117, "global_step": 18016, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03609345108270645, "train_loss_bc": 0.03609103709459305, "train_loss_llm": 0.00024139881134033203, "grad_norm": 0.029099762439727783, "global_step": 18017, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03552629426121712, "train_loss_bc": 0.03552629426121712, "train_loss_llm": 0.0, "grad_norm": 0.04187042638659477, "global_step": 18018, "epoch": 500, "lr": 0.001} +{"train_loss": 0.04064200818538666, "train_loss_bc": 0.04064200818538666, "train_loss_llm": 0.0, "grad_norm": 0.056420840322971344, "global_step": 18019, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03740626201033592, "train_loss_bc": 0.037402838468551636, "train_loss_llm": 0.00034233927726745605, "grad_norm": 0.06848778575658798, "global_step": 18020, "epoch": 500, "lr": 0.001} +{"train_loss": 0.030925389379262924, "train_loss_bc": 0.030925389379262924, "train_loss_llm": 0.0, "grad_norm": 0.021656837314367294, "global_step": 18021, "epoch": 500, "lr": 0.001} +{"train_loss": 0.033968787640333176, "train_loss_bc": 0.033968787640333176, "train_loss_llm": 0.0, "grad_norm": 0.03694731742143631, "global_step": 18022, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03411640599370003, "train_loss_bc": 0.03411640599370003, "train_loss_llm": 0.0, "grad_norm": 0.04335630685091019, "global_step": 18023, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03663787990808487, "train_loss_bc": 0.03663787990808487, "train_loss_llm": 0.0, "grad_norm": 0.058146242052316666, "global_step": 18024, "epoch": 500, "lr": 0.001} +{"train_loss": 0.040274299681186676, "train_loss_bc": 0.040257491171360016, "train_loss_llm": 0.0016807615756988525, "grad_norm": 0.03559517115354538, "global_step": 18025, "epoch": 500, "lr": 0.001} +{"train_loss": 0.026184778660535812, "train_loss_bc": 0.026184778660535812, "train_loss_llm": 0.0, "grad_norm": 0.05140381306409836, "global_step": 18026, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03330645710229874, "train_loss_bc": 0.03329691290855408, "train_loss_llm": 0.0009543299674987793, "grad_norm": 0.064747154712677, "global_step": 18027, "epoch": 500, "lr": 0.001} +{"train_loss": 0.034471433609724045, "train_loss_bc": 0.034471433609724045, "train_loss_llm": 0.0, "grad_norm": 0.0817735493183136, "global_step": 18028, "epoch": 500, "lr": 0.001} +{"train_loss": 0.031739141792058945, "train_loss_bc": 0.031739141792058945, "train_loss_llm": 0.0, "grad_norm": 0.022910578176379204, "global_step": 18029, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03217308595776558, "train_loss_bc": 0.03217308595776558, "train_loss_llm": 0.0, "grad_norm": 0.031109625473618507, "global_step": 18030, "epoch": 500, "lr": 0.001} +{"train_loss": 0.030762234702706337, "train_loss_bc": 0.030762234702706337, "train_loss_llm": 0.0, "grad_norm": 0.054365526884794235, "global_step": 18031, "epoch": 500, "lr": 0.001} +{"train_loss": 0.036362726241350174, "train_loss_bc": 0.036362141370773315, "train_loss_llm": 5.835294723510742e-05, "grad_norm": 0.06842798739671707, "global_step": 18032, "epoch": 500, "lr": 0.001} +{"train_loss": 0.02749095857143402, "train_loss_bc": 0.02749095857143402, "train_loss_llm": 0.0, "grad_norm": 0.023342212662100792, "global_step": 18033, "epoch": 500, "lr": 0.001} +{"train_loss": 0.040226828306913376, "train_loss_bc": 0.040226828306913376, "train_loss_llm": 0.0, "grad_norm": 0.04390048235654831, "global_step": 18034, "epoch": 500, "lr": 0.001} +{"train_loss": 0.03420108220436507, "train_loss_bc": 0.035144828259944916, "train_loss_llm": 0.0, "grad_norm": 0.07514958083629608, "global_step": 18035, "epoch": 500, "lr": 0.001, "train/cumulative_reward": 104.2755974518129, "train/mean_score": 15.838549920052879, "train/success_rate": 0.4, "test/cumulative_reward": 39.96468350519717, "test/mean_score": 8.831176706351206, "test/success_rate": 0.2, "val_loss": 0.07498157024383545, "train_action_mse_error": 0.0366705022752285} +{"train_loss": 0.03297394514083862, "train_loss_bc": 0.03297394514083862, "train_loss_llm": 0.0, "grad_norm": 0.08476818352937698, "global_step": 18036, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03954590484499931, "train_loss_bc": 0.03952544555068016, "train_loss_llm": 0.0020457804203033447, "grad_norm": 0.026930397376418114, "global_step": 18037, "epoch": 501, "lr": 0.001} +{"train_loss": 0.0349961593747139, "train_loss_bc": 0.03499443829059601, "train_loss_llm": 0.00017204880714416504, "grad_norm": 0.046152159571647644, "global_step": 18038, "epoch": 501, "lr": 0.001} +{"train_loss": 0.028111770749092102, "train_loss_bc": 0.02810743823647499, "train_loss_llm": 0.00043323636054992676, "grad_norm": 0.06731299310922623, "global_step": 18039, "epoch": 501, "lr": 0.001} +{"train_loss": 0.037188246846199036, "train_loss_bc": 0.037188246846199036, "train_loss_llm": 0.0, "grad_norm": 0.09253878146409988, "global_step": 18040, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03749338537454605, "train_loss_bc": 0.03749338537454605, "train_loss_llm": 0.0, "grad_norm": 0.02890028990805149, "global_step": 18041, "epoch": 501, "lr": 0.001} +{"train_loss": 0.02583223767578602, "train_loss_bc": 0.02582586370408535, "train_loss_llm": 0.000637352466583252, "grad_norm": 0.03786080703139305, "global_step": 18042, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03388827666640282, "train_loss_bc": 0.03388827666640282, "train_loss_llm": 0.0, "grad_norm": 0.0498783178627491, "global_step": 18043, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03899623826146126, "train_loss_bc": 0.03899623826146126, "train_loss_llm": 0.0, "grad_norm": 0.05713051185011864, "global_step": 18044, "epoch": 501, "lr": 0.001} +{"train_loss": 0.033436551690101624, "train_loss_bc": 0.033436551690101624, "train_loss_llm": 0.0, "grad_norm": 0.02347596548497677, "global_step": 18045, "epoch": 501, "lr": 0.001} +{"train_loss": 0.02995971404016018, "train_loss_bc": 0.02995971404016018, "train_loss_llm": 0.0, "grad_norm": 0.03538469597697258, "global_step": 18046, "epoch": 501, "lr": 0.001} +{"train_loss": 0.036409180611371994, "train_loss_bc": 0.036409180611371994, "train_loss_llm": 0.0, "grad_norm": 0.049746714532375336, "global_step": 18047, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03410988673567772, "train_loss_bc": 0.03410988673567772, "train_loss_llm": 0.0, "grad_norm": 0.05797109752893448, "global_step": 18048, "epoch": 501, "lr": 0.001} +{"train_loss": 0.030840450897812843, "train_loss_bc": 0.030840450897812843, "train_loss_llm": 0.0, "grad_norm": 0.029462654143571854, "global_step": 18049, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03131004050374031, "train_loss_bc": 0.03131004050374031, "train_loss_llm": 0.0, "grad_norm": 0.053820475935935974, "global_step": 18050, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03400200232863426, "train_loss_bc": 0.03400200232863426, "train_loss_llm": 0.0, "grad_norm": 0.06961358338594437, "global_step": 18051, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03652353957295418, "train_loss_bc": 0.03651151433587074, "train_loss_llm": 0.0012024641036987305, "grad_norm": 0.08493614941835403, "global_step": 18052, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03246621415019035, "train_loss_bc": 0.032461464405059814, "train_loss_llm": 0.0004750490188598633, "grad_norm": 0.024877222254872322, "global_step": 18053, "epoch": 501, "lr": 0.001} +{"train_loss": 0.037701092660427094, "train_loss_bc": 0.037701092660427094, "train_loss_llm": 0.0, "grad_norm": 0.04261965677142143, "global_step": 18054, "epoch": 501, "lr": 0.001} +{"train_loss": 0.026741065084934235, "train_loss_bc": 0.026718415319919586, "train_loss_llm": 0.0022649765014648438, "grad_norm": 0.052913397550582886, "global_step": 18055, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03673114627599716, "train_loss_bc": 0.03673114627599716, "train_loss_llm": 0.0, "grad_norm": 0.06248871982097626, "global_step": 18056, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03632054850459099, "train_loss_bc": 0.03632054850459099, "train_loss_llm": 0.0, "grad_norm": 0.029708363115787506, "global_step": 18057, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03539323806762695, "train_loss_bc": 0.03539323806762695, "train_loss_llm": 0.0, "grad_norm": 0.05264066532254219, "global_step": 18058, "epoch": 501, "lr": 0.001} +{"train_loss": 0.0271442960947752, "train_loss_bc": 0.0271442960947752, "train_loss_llm": 0.0, "grad_norm": 0.054748546332120895, "global_step": 18059, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03111070580780506, "train_loss_bc": 0.03111070580780506, "train_loss_llm": 0.0, "grad_norm": 0.06156599149107933, "global_step": 18060, "epoch": 501, "lr": 0.001} +{"train_loss": 0.038379356265068054, "train_loss_bc": 0.038379356265068054, "train_loss_llm": 0.0, "grad_norm": 0.026579858735203743, "global_step": 18061, "epoch": 501, "lr": 0.001} +{"train_loss": 0.035857733339071274, "train_loss_bc": 0.035857733339071274, "train_loss_llm": 0.0, "grad_norm": 0.03930852562189102, "global_step": 18062, "epoch": 501, "lr": 0.001} +{"train_loss": 0.037548214197158813, "train_loss_bc": 0.037548214197158813, "train_loss_llm": 0.0, "grad_norm": 0.05374274030327797, "global_step": 18063, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03761029988527298, "train_loss_bc": 0.03761029988527298, "train_loss_llm": 0.0, "grad_norm": 0.073955237865448, "global_step": 18064, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03235973045229912, "train_loss_bc": 0.03235973045229912, "train_loss_llm": 0.0, "grad_norm": 0.025402896106243134, "global_step": 18065, "epoch": 501, "lr": 0.001} +{"train_loss": 0.029335979372262955, "train_loss_bc": 0.029335979372262955, "train_loss_llm": 0.0, "grad_norm": 0.03530211001634598, "global_step": 18066, "epoch": 501, "lr": 0.001} +{"train_loss": 0.0375988632440567, "train_loss_bc": 0.0375988632440567, "train_loss_llm": 0.0, "grad_norm": 0.0481443852186203, "global_step": 18067, "epoch": 501, "lr": 0.001} +{"train_loss": 0.0379822812974453, "train_loss_bc": 0.03794662654399872, "train_loss_llm": 0.0035656094551086426, "grad_norm": 0.060565121471881866, "global_step": 18068, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03457029536366463, "train_loss_bc": 0.034559465944767, "train_loss_llm": 0.0010830461978912354, "grad_norm": 0.031845755875110626, "global_step": 18069, "epoch": 501, "lr": 0.001} +{"train_loss": 0.029789410531520844, "train_loss_bc": 0.02977762557566166, "train_loss_llm": 0.0011784732341766357, "grad_norm": 0.03706759214401245, "global_step": 18070, "epoch": 501, "lr": 0.001} +{"train_loss": 0.03419170958093471, "train_loss_bc": 0.040643543004989624, "train_loss_llm": 0.0, "grad_norm": 0.0554361566901207, "global_step": 18071, "epoch": 501, "lr": 0.001, "val_loss": 0.07405510544776917} +{"train_loss": 0.03790212795138359, "train_loss_bc": 0.03790212795138359, "train_loss_llm": 0.0, "grad_norm": 0.06515451520681381, "global_step": 18072, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03164320066571236, "train_loss_bc": 0.03164320066571236, "train_loss_llm": 0.0, "grad_norm": 0.020630931481719017, "global_step": 18073, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03723832964897156, "train_loss_bc": 0.03720797970890999, "train_loss_llm": 0.003035128116607666, "grad_norm": 0.043069954961538315, "global_step": 18074, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03550826758146286, "train_loss_bc": 0.03550826758146286, "train_loss_llm": 0.0, "grad_norm": 0.05341881886124611, "global_step": 18075, "epoch": 502, "lr": 0.001} +{"train_loss": 0.036071110516786575, "train_loss_bc": 0.036071110516786575, "train_loss_llm": 0.0, "grad_norm": 0.06705757975578308, "global_step": 18076, "epoch": 502, "lr": 0.001} +{"train_loss": 0.035423677414655685, "train_loss_bc": 0.035423677414655685, "train_loss_llm": 0.0, "grad_norm": 0.02682196907699108, "global_step": 18077, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03405826538801193, "train_loss_bc": 0.03405826538801193, "train_loss_llm": 0.0, "grad_norm": 0.046489134430885315, "global_step": 18078, "epoch": 502, "lr": 0.001} +{"train_loss": 0.029269864782691002, "train_loss_bc": 0.029269864782691002, "train_loss_llm": 0.0, "grad_norm": 0.06229046359658241, "global_step": 18079, "epoch": 502, "lr": 0.001} +{"train_loss": 0.033732958137989044, "train_loss_bc": 0.033732958137989044, "train_loss_llm": 0.0, "grad_norm": 0.07531899213790894, "global_step": 18080, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03185660019516945, "train_loss_bc": 0.03185660019516945, "train_loss_llm": 0.0, "grad_norm": 0.02550949901342392, "global_step": 18081, "epoch": 502, "lr": 0.001} +{"train_loss": 0.02939162403345108, "train_loss_bc": 0.02939162403345108, "train_loss_llm": 0.0, "grad_norm": 0.04234139993786812, "global_step": 18082, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03430819511413574, "train_loss_bc": 0.03430819511413574, "train_loss_llm": 0.0, "grad_norm": 0.05976193770766258, "global_step": 18083, "epoch": 502, "lr": 0.001} +{"train_loss": 0.031378407031297684, "train_loss_bc": 0.031354326754808426, "train_loss_llm": 0.0024081170558929443, "grad_norm": 0.06583341211080551, "global_step": 18084, "epoch": 502, "lr": 0.001} +{"train_loss": 0.036205925047397614, "train_loss_bc": 0.036205925047397614, "train_loss_llm": 0.0, "grad_norm": 0.024288712069392204, "global_step": 18085, "epoch": 502, "lr": 0.001} +{"train_loss": 0.023887842893600464, "train_loss_bc": 0.02388550154864788, "train_loss_llm": 0.00023412704467773438, "grad_norm": 0.03149983659386635, "global_step": 18086, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03310345858335495, "train_loss_bc": 0.03310345858335495, "train_loss_llm": 0.0, "grad_norm": 0.045773737132549286, "global_step": 18087, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03539835661649704, "train_loss_bc": 0.0353914350271225, "train_loss_llm": 0.000692218542098999, "grad_norm": 0.059609320014715195, "global_step": 18088, "epoch": 502, "lr": 0.001} +{"train_loss": 0.029976144433021545, "train_loss_bc": 0.029976144433021545, "train_loss_llm": 0.0, "grad_norm": 0.029453525319695473, "global_step": 18089, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03192374110221863, "train_loss_bc": 0.03192374110221863, "train_loss_llm": 0.0, "grad_norm": 0.041640907526016235, "global_step": 18090, "epoch": 502, "lr": 0.001} +{"train_loss": 0.032358378171920776, "train_loss_bc": 0.032358378171920776, "train_loss_llm": 0.0, "grad_norm": 0.05548517778515816, "global_step": 18091, "epoch": 502, "lr": 0.001} +{"train_loss": 0.030459251254796982, "train_loss_bc": 0.030459251254796982, "train_loss_llm": 0.0, "grad_norm": 0.08160386979579926, "global_step": 18092, "epoch": 502, "lr": 0.001} +{"train_loss": 0.033779654651880264, "train_loss_bc": 0.03376961871981621, "train_loss_llm": 0.0010034143924713135, "grad_norm": 0.0335674062371254, "global_step": 18093, "epoch": 502, "lr": 0.001} +{"train_loss": 0.032593097537755966, "train_loss_bc": 0.03258012235164642, "train_loss_llm": 0.001297682523727417, "grad_norm": 0.04525105282664299, "global_step": 18094, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03411103039979935, "train_loss_bc": 0.03411103039979935, "train_loss_llm": 0.0, "grad_norm": 0.05869311839342117, "global_step": 18095, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03134436532855034, "train_loss_bc": 0.03134436532855034, "train_loss_llm": 0.0, "grad_norm": 0.0643961951136589, "global_step": 18096, "epoch": 502, "lr": 0.001} +{"train_loss": 0.04316972941160202, "train_loss_bc": 0.04316972941160202, "train_loss_llm": 0.0, "grad_norm": 0.025410974398255348, "global_step": 18097, "epoch": 502, "lr": 0.001} +{"train_loss": 0.0408439040184021, "train_loss_bc": 0.0408439040184021, "train_loss_llm": 0.0, "grad_norm": 0.04195456579327583, "global_step": 18098, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03674944117665291, "train_loss_bc": 0.036723650991916656, "train_loss_llm": 0.002578914165496826, "grad_norm": 0.050377894192934036, "global_step": 18099, "epoch": 502, "lr": 0.001} +{"train_loss": 0.036739397794008255, "train_loss_bc": 0.036739397794008255, "train_loss_llm": 0.0, "grad_norm": 0.061194054782390594, "global_step": 18100, "epoch": 502, "lr": 0.001} +{"train_loss": 0.034633152186870575, "train_loss_bc": 0.03463118150830269, "train_loss_llm": 0.0001970827579498291, "grad_norm": 0.023474158719182014, "global_step": 18101, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03420570120215416, "train_loss_bc": 0.03420570120215416, "train_loss_llm": 0.0, "grad_norm": 0.03639302775263786, "global_step": 18102, "epoch": 502, "lr": 0.001} +{"train_loss": 0.04196229204535484, "train_loss_bc": 0.04196229204535484, "train_loss_llm": 0.0, "grad_norm": 0.04715075343847275, "global_step": 18103, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03359261155128479, "train_loss_bc": 0.03358849138021469, "train_loss_llm": 0.00041219592094421387, "grad_norm": 0.05072317644953728, "global_step": 18104, "epoch": 502, "lr": 0.001} +{"train_loss": 0.03222733736038208, "train_loss_bc": 0.03222733736038208, "train_loss_llm": 0.0, "grad_norm": 0.025240229442715645, "global_step": 18105, "epoch": 502, "lr": 0.001} +{"train_loss": 0.030882343649864197, "train_loss_bc": 0.030882343649864197, "train_loss_llm": 0.0, "grad_norm": 0.038002464920282364, "global_step": 18106, "epoch": 502, "lr": 0.001} +{"train_loss": 0.034084221710347466, "train_loss_bc": 0.039097435772418976, "train_loss_llm": 0.00047600269317626953, "grad_norm": 0.05697226524353027, "global_step": 18107, "epoch": 502, "lr": 0.001, "val_loss": 0.07479989528656006} +{"train_loss": 0.03471161797642708, "train_loss_bc": 0.03471161797642708, "train_loss_llm": 0.0, "grad_norm": 0.06991378962993622, "global_step": 18108, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03380408510565758, "train_loss_bc": 0.03380408510565758, "train_loss_llm": 0.0, "grad_norm": 0.02658708207309246, "global_step": 18109, "epoch": 503, "lr": 0.001} +{"train_loss": 0.033177699893713, "train_loss_bc": 0.033177699893713, "train_loss_llm": 0.0, "grad_norm": 0.04801413416862488, "global_step": 18110, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03494025394320488, "train_loss_bc": 0.03494025394320488, "train_loss_llm": 0.0, "grad_norm": 0.05338749662041664, "global_step": 18111, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03116666153073311, "train_loss_bc": 0.03116666153073311, "train_loss_llm": 0.0, "grad_norm": 0.06487155705690384, "global_step": 18112, "epoch": 503, "lr": 0.001} +{"train_loss": 0.031140297651290894, "train_loss_bc": 0.031140297651290894, "train_loss_llm": 0.0, "grad_norm": 0.02859177254140377, "global_step": 18113, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03149077296257019, "train_loss_bc": 0.03149077296257019, "train_loss_llm": 0.0, "grad_norm": 0.05064601078629494, "global_step": 18114, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03249537572264671, "train_loss_bc": 0.03249537572264671, "train_loss_llm": 0.0, "grad_norm": 0.057192932814359665, "global_step": 18115, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03799569606781006, "train_loss_bc": 0.03797004371881485, "train_loss_llm": 0.002565354108810425, "grad_norm": 0.07869097590446472, "global_step": 18116, "epoch": 503, "lr": 0.001} +{"train_loss": 0.04461809620261192, "train_loss_bc": 0.044594455510377884, "train_loss_llm": 0.0023640692234039307, "grad_norm": 0.03076891414821148, "global_step": 18117, "epoch": 503, "lr": 0.001} +{"train_loss": 0.033344823867082596, "train_loss_bc": 0.033344823867082596, "train_loss_llm": 0.0, "grad_norm": 0.040298521518707275, "global_step": 18118, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03502281382679939, "train_loss_bc": 0.035020146518945694, "train_loss_llm": 0.0002668797969818115, "grad_norm": 0.053113266825675964, "global_step": 18119, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03438401222229004, "train_loss_bc": 0.03435263782739639, "train_loss_llm": 0.0031375885009765625, "grad_norm": 0.06457194685935974, "global_step": 18120, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03316979855298996, "train_loss_bc": 0.03315658122301102, "train_loss_llm": 0.0013218820095062256, "grad_norm": 0.025974905118346214, "global_step": 18121, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03001626953482628, "train_loss_bc": 0.030011191964149475, "train_loss_llm": 0.0005076825618743896, "grad_norm": 0.04357587546110153, "global_step": 18122, "epoch": 503, "lr": 0.001} +{"train_loss": 0.038774117827415466, "train_loss_bc": 0.038774117827415466, "train_loss_llm": 0.0, "grad_norm": 0.05308777838945389, "global_step": 18123, "epoch": 503, "lr": 0.001} +{"train_loss": 0.032626740634441376, "train_loss_bc": 0.032626740634441376, "train_loss_llm": 0.0, "grad_norm": 0.06015701964497566, "global_step": 18124, "epoch": 503, "lr": 0.001} +{"train_loss": 0.029595771804451942, "train_loss_bc": 0.029595771804451942, "train_loss_llm": 0.0, "grad_norm": 0.03308987617492676, "global_step": 18125, "epoch": 503, "lr": 0.001} +{"train_loss": 0.02780299261212349, "train_loss_bc": 0.02780299261212349, "train_loss_llm": 0.0, "grad_norm": 0.04733218997716904, "global_step": 18126, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03664824739098549, "train_loss_bc": 0.03664824739098549, "train_loss_llm": 0.0, "grad_norm": 0.065673828125, "global_step": 18127, "epoch": 503, "lr": 0.001} +{"train_loss": 0.02991357445716858, "train_loss_bc": 0.02991357445716858, "train_loss_llm": 0.0, "grad_norm": 0.08020996302366257, "global_step": 18128, "epoch": 503, "lr": 0.001} +{"train_loss": 0.034372277557849884, "train_loss_bc": 0.034372277557849884, "train_loss_llm": 0.0, "grad_norm": 0.024197574704885483, "global_step": 18129, "epoch": 503, "lr": 0.001} +{"train_loss": 0.02962072752416134, "train_loss_bc": 0.02962072752416134, "train_loss_llm": 0.0, "grad_norm": 0.04493854194879532, "global_step": 18130, "epoch": 503, "lr": 0.001} +{"train_loss": 0.036187637597322464, "train_loss_bc": 0.036187637597322464, "train_loss_llm": 0.0, "grad_norm": 0.06367471069097519, "global_step": 18131, "epoch": 503, "lr": 0.001} +{"train_loss": 0.04124192148447037, "train_loss_bc": 0.04124192148447037, "train_loss_llm": 0.0, "grad_norm": 0.08334042131900787, "global_step": 18132, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03513733670115471, "train_loss_bc": 0.03513733670115471, "train_loss_llm": 0.0, "grad_norm": 0.026138558983802795, "global_step": 18133, "epoch": 503, "lr": 0.001} +{"train_loss": 0.035996902734041214, "train_loss_bc": 0.035979773849248886, "train_loss_llm": 0.0017129778861999512, "grad_norm": 0.041784971952438354, "global_step": 18134, "epoch": 503, "lr": 0.001} +{"train_loss": 0.037170253694057465, "train_loss_bc": 0.037170253694057465, "train_loss_llm": 0.0, "grad_norm": 0.04639666527509689, "global_step": 18135, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03519608825445175, "train_loss_bc": 0.03519608825445175, "train_loss_llm": 0.0, "grad_norm": 0.06300070881843567, "global_step": 18136, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03776079788804054, "train_loss_bc": 0.03776079788804054, "train_loss_llm": 0.0, "grad_norm": 0.02955903671681881, "global_step": 18137, "epoch": 503, "lr": 0.001} +{"train_loss": 0.033023811876773834, "train_loss_bc": 0.032997626811265945, "train_loss_llm": 0.002618461847305298, "grad_norm": 0.04680228605866432, "global_step": 18138, "epoch": 503, "lr": 0.001} +{"train_loss": 0.034934766590595245, "train_loss_bc": 0.034934766590595245, "train_loss_llm": 0.0, "grad_norm": 0.06520449370145798, "global_step": 18139, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03352801129221916, "train_loss_bc": 0.03352801129221916, "train_loss_llm": 0.0, "grad_norm": 0.07557489722967148, "global_step": 18140, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03003253974020481, "train_loss_bc": 0.03003253974020481, "train_loss_llm": 0.0, "grad_norm": 0.03287842869758606, "global_step": 18141, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03846673294901848, "train_loss_bc": 0.03846673294901848, "train_loss_llm": 0.0, "grad_norm": 0.05025871843099594, "global_step": 18142, "epoch": 503, "lr": 0.001} +{"train_loss": 0.03410540381446481, "train_loss_bc": 0.028283486142754555, "train_loss_llm": 0.000152587890625, "grad_norm": 0.07128334790468216, "global_step": 18143, "epoch": 503, "lr": 0.001, "val_loss": 0.07539483904838562} +{"train_loss": 0.03394549712538719, "train_loss_bc": 0.0339360348880291, "train_loss_llm": 0.0009461045265197754, "grad_norm": 0.09441792964935303, "global_step": 18144, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03226108476519585, "train_loss_bc": 0.03226108476519585, "train_loss_llm": 0.0, "grad_norm": 0.02699250541627407, "global_step": 18145, "epoch": 504, "lr": 0.001} +{"train_loss": 0.030595406889915466, "train_loss_bc": 0.030595406889915466, "train_loss_llm": 0.0, "grad_norm": 0.0430871844291687, "global_step": 18146, "epoch": 504, "lr": 0.001} +{"train_loss": 0.039940446615219116, "train_loss_bc": 0.039940446615219116, "train_loss_llm": 0.0, "grad_norm": 0.04931943863630295, "global_step": 18147, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03665841370820999, "train_loss_bc": 0.03665841370820999, "train_loss_llm": 0.0, "grad_norm": 0.07710354030132294, "global_step": 18148, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03620459511876106, "train_loss_bc": 0.03620459511876106, "train_loss_llm": 0.0, "grad_norm": 0.03801837936043739, "global_step": 18149, "epoch": 504, "lr": 0.001} +{"train_loss": 0.030085213482379913, "train_loss_bc": 0.030085213482379913, "train_loss_llm": 0.0, "grad_norm": 0.044749654829502106, "global_step": 18150, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03270713612437248, "train_loss_bc": 0.03270713612437248, "train_loss_llm": 0.0, "grad_norm": 0.05642073228955269, "global_step": 18151, "epoch": 504, "lr": 0.001} +{"train_loss": 0.027165135368704796, "train_loss_bc": 0.027165135368704796, "train_loss_llm": 0.0, "grad_norm": 0.06748371571302414, "global_step": 18152, "epoch": 504, "lr": 0.001} +{"train_loss": 0.04109535738825798, "train_loss_bc": 0.04109535738825798, "train_loss_llm": 0.0, "grad_norm": 0.028282642364501953, "global_step": 18153, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03244416043162346, "train_loss_bc": 0.03244416043162346, "train_loss_llm": 0.0, "grad_norm": 0.03812054172158241, "global_step": 18154, "epoch": 504, "lr": 0.001} +{"train_loss": 0.028527094051241875, "train_loss_bc": 0.028520889580249786, "train_loss_llm": 0.0006205141544342041, "grad_norm": 0.04544961825013161, "global_step": 18155, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03355452045798302, "train_loss_bc": 0.03355452045798302, "train_loss_llm": 0.0, "grad_norm": 0.05619887262582779, "global_step": 18156, "epoch": 504, "lr": 0.001} +{"train_loss": 0.030052989721298218, "train_loss_bc": 0.030042970553040504, "train_loss_llm": 0.0010018646717071533, "grad_norm": 0.025912903249263763, "global_step": 18157, "epoch": 504, "lr": 0.001} +{"train_loss": 0.0288102924823761, "train_loss_bc": 0.0288102924823761, "train_loss_llm": 0.0, "grad_norm": 0.039791662245988846, "global_step": 18158, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03896954283118248, "train_loss_bc": 0.03896954283118248, "train_loss_llm": 0.0, "grad_norm": 0.05155413597822189, "global_step": 18159, "epoch": 504, "lr": 0.001} +{"train_loss": 0.0368364043533802, "train_loss_bc": 0.0368364043533802, "train_loss_llm": 0.0, "grad_norm": 0.05978025496006012, "global_step": 18160, "epoch": 504, "lr": 0.001} +{"train_loss": 0.029012611135840416, "train_loss_bc": 0.029012611135840416, "train_loss_llm": 0.0, "grad_norm": 0.022174330428242683, "global_step": 18161, "epoch": 504, "lr": 0.001} +{"train_loss": 0.031438667327165604, "train_loss_bc": 0.03143095225095749, "train_loss_llm": 0.0007715821266174316, "grad_norm": 0.03817169740796089, "global_step": 18162, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03686061128973961, "train_loss_bc": 0.03686061128973961, "train_loss_llm": 0.0, "grad_norm": 0.05840439349412918, "global_step": 18163, "epoch": 504, "lr": 0.001} +{"train_loss": 0.030011983588337898, "train_loss_bc": 0.029998959973454475, "train_loss_llm": 0.0013023614883422852, "grad_norm": 0.07044817507266998, "global_step": 18164, "epoch": 504, "lr": 0.001} +{"train_loss": 0.044840648770332336, "train_loss_bc": 0.044840648770332336, "train_loss_llm": 0.0, "grad_norm": 0.033435553312301636, "global_step": 18165, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03637682646512985, "train_loss_bc": 0.03637682646512985, "train_loss_llm": 0.0, "grad_norm": 0.04779348894953728, "global_step": 18166, "epoch": 504, "lr": 0.001} +{"train_loss": 0.0341053307056427, "train_loss_bc": 0.0341053307056427, "train_loss_llm": 0.0, "grad_norm": 0.06078827381134033, "global_step": 18167, "epoch": 504, "lr": 0.001} +{"train_loss": 0.034909121692180634, "train_loss_bc": 0.034909121692180634, "train_loss_llm": 0.0, "grad_norm": 0.07160967588424683, "global_step": 18168, "epoch": 504, "lr": 0.001} +{"train_loss": 0.032145917415618896, "train_loss_bc": 0.032107166945934296, "train_loss_llm": 0.0038749277591705322, "grad_norm": 0.025797229260206223, "global_step": 18169, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03351498767733574, "train_loss_bc": 0.03351498767733574, "train_loss_llm": 0.0, "grad_norm": 0.04417434334754944, "global_step": 18170, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03226565569639206, "train_loss_bc": 0.03226565569639206, "train_loss_llm": 0.0, "grad_norm": 0.049447786062955856, "global_step": 18171, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03280147165060043, "train_loss_bc": 0.03280147165060043, "train_loss_llm": 0.0, "grad_norm": 0.06650852411985397, "global_step": 18172, "epoch": 504, "lr": 0.001} +{"train_loss": 0.0317261666059494, "train_loss_bc": 0.03172292932868004, "train_loss_llm": 0.0003236234188079834, "grad_norm": 0.028361940756440163, "global_step": 18173, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03742736577987671, "train_loss_bc": 0.03742624446749687, "train_loss_llm": 0.00011211633682250977, "grad_norm": 0.043651238083839417, "global_step": 18174, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03295714035630226, "train_loss_bc": 0.03295714035630226, "train_loss_llm": 0.0, "grad_norm": 0.06106178089976311, "global_step": 18175, "epoch": 504, "lr": 0.001} +{"train_loss": 0.034035272896289825, "train_loss_bc": 0.034035272896289825, "train_loss_llm": 0.0, "grad_norm": 0.07094303518533707, "global_step": 18176, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03597221150994301, "train_loss_bc": 0.03597221150994301, "train_loss_llm": 0.0, "grad_norm": 0.03020169958472252, "global_step": 18177, "epoch": 504, "lr": 0.001} +{"train_loss": 0.04128037393093109, "train_loss_bc": 0.04126396402716637, "train_loss_llm": 0.0016408264636993408, "grad_norm": 0.05790657922625542, "global_step": 18178, "epoch": 504, "lr": 0.001} +{"train_loss": 0.03404888603836298, "train_loss_bc": 0.03421933948993683, "train_loss_llm": 0.0004903674125671387, "grad_norm": 0.065254345536232, "global_step": 18179, "epoch": 504, "lr": 0.001, "val_loss": 0.07535669952630997} +{"train_loss": 0.03984639421105385, "train_loss_bc": 0.03984639421105385, "train_loss_llm": 0.0, "grad_norm": 0.08118753880262375, "global_step": 18180, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03829849511384964, "train_loss_bc": 0.03829849511384964, "train_loss_llm": 0.0, "grad_norm": 0.030791079625487328, "global_step": 18181, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03455306217074394, "train_loss_bc": 0.03455306217074394, "train_loss_llm": 0.0, "grad_norm": 0.03998951613903046, "global_step": 18182, "epoch": 505, "lr": 0.001} +{"train_loss": 0.027827341109514236, "train_loss_bc": 0.027827341109514236, "train_loss_llm": 0.0, "grad_norm": 0.0486903190612793, "global_step": 18183, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03123505413532257, "train_loss_bc": 0.03123505413532257, "train_loss_llm": 0.0, "grad_norm": 0.0604294016957283, "global_step": 18184, "epoch": 505, "lr": 0.001} +{"train_loss": 0.02865409106016159, "train_loss_bc": 0.028646260499954224, "train_loss_llm": 0.0007829666137695312, "grad_norm": 0.02525557577610016, "global_step": 18185, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03571133688092232, "train_loss_bc": 0.03571133688092232, "train_loss_llm": 0.0, "grad_norm": 0.05017753317952156, "global_step": 18186, "epoch": 505, "lr": 0.001} +{"train_loss": 0.031053829938173294, "train_loss_bc": 0.031053829938173294, "train_loss_llm": 0.0, "grad_norm": 0.06347605586051941, "global_step": 18187, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03432992100715637, "train_loss_bc": 0.03432992100715637, "train_loss_llm": 0.0, "grad_norm": 0.07689929008483887, "global_step": 18188, "epoch": 505, "lr": 0.001} +{"train_loss": 0.031952112913131714, "train_loss_bc": 0.031952112913131714, "train_loss_llm": 0.0, "grad_norm": 0.034008752554655075, "global_step": 18189, "epoch": 505, "lr": 0.001} +{"train_loss": 0.030073489993810654, "train_loss_bc": 0.030073489993810654, "train_loss_llm": 0.0, "grad_norm": 0.051141805946826935, "global_step": 18190, "epoch": 505, "lr": 0.001} +{"train_loss": 0.036257289350032806, "train_loss_bc": 0.036257289350032806, "train_loss_llm": 0.0, "grad_norm": 0.06802499294281006, "global_step": 18191, "epoch": 505, "lr": 0.001} +{"train_loss": 0.030198348686099052, "train_loss_bc": 0.030198348686099052, "train_loss_llm": 0.0, "grad_norm": 0.08352651447057724, "global_step": 18192, "epoch": 505, "lr": 0.001} +{"train_loss": 0.0352034866809845, "train_loss_bc": 0.0352034866809845, "train_loss_llm": 0.0, "grad_norm": 0.030675506219267845, "global_step": 18193, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03386238217353821, "train_loss_bc": 0.03385048732161522, "train_loss_llm": 0.0011893808841705322, "grad_norm": 0.05398041009902954, "global_step": 18194, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03329404816031456, "train_loss_bc": 0.033285271376371384, "train_loss_llm": 0.0008775591850280762, "grad_norm": 0.06933699548244476, "global_step": 18195, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03467576205730438, "train_loss_bc": 0.03467576205730438, "train_loss_llm": 0.0, "grad_norm": 0.08925820142030716, "global_step": 18196, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03440166264772415, "train_loss_bc": 0.03440166264772415, "train_loss_llm": 0.0, "grad_norm": 0.02701888233423233, "global_step": 18197, "epoch": 505, "lr": 0.001} +{"train_loss": 0.04057936742901802, "train_loss_bc": 0.04057936742901802, "train_loss_llm": 0.0, "grad_norm": 0.04542124643921852, "global_step": 18198, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03784754127264023, "train_loss_bc": 0.03782299533486366, "train_loss_llm": 0.002454429864883423, "grad_norm": 0.0586581714451313, "global_step": 18199, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03559952601790428, "train_loss_bc": 0.03559952601790428, "train_loss_llm": 0.0, "grad_norm": 0.07202904671430588, "global_step": 18200, "epoch": 505, "lr": 0.001} +{"train_loss": 0.0343475304543972, "train_loss_bc": 0.034341465681791306, "train_loss_llm": 0.0006065666675567627, "grad_norm": 0.0246815737336874, "global_step": 18201, "epoch": 505, "lr": 0.001} +{"train_loss": 0.04042831063270569, "train_loss_bc": 0.04040102660655975, "train_loss_llm": 0.002728372812271118, "grad_norm": 0.043394774198532104, "global_step": 18202, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03331238031387329, "train_loss_bc": 0.03331238031387329, "train_loss_llm": 0.0, "grad_norm": 0.05326690152287483, "global_step": 18203, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03578827157616615, "train_loss_bc": 0.03577699512243271, "train_loss_llm": 0.0011276602745056152, "grad_norm": 0.07229471206665039, "global_step": 18204, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03431424871087074, "train_loss_bc": 0.034312982112169266, "train_loss_llm": 0.00012668967247009277, "grad_norm": 0.022370433434844017, "global_step": 18205, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03435821458697319, "train_loss_bc": 0.03435821458697319, "train_loss_llm": 0.0, "grad_norm": 0.04553333669900894, "global_step": 18206, "epoch": 505, "lr": 0.001} +{"train_loss": 0.035002775490283966, "train_loss_bc": 0.035002775490283966, "train_loss_llm": 0.0, "grad_norm": 0.05965753644704819, "global_step": 18207, "epoch": 505, "lr": 0.001} +{"train_loss": 0.032433342188596725, "train_loss_bc": 0.032433342188596725, "train_loss_llm": 0.0, "grad_norm": 0.07394308596849442, "global_step": 18208, "epoch": 505, "lr": 0.001} +{"train_loss": 0.030767900869250298, "train_loss_bc": 0.030767900869250298, "train_loss_llm": 0.0, "grad_norm": 0.03275799751281738, "global_step": 18209, "epoch": 505, "lr": 0.001} +{"train_loss": 0.034499477595090866, "train_loss_bc": 0.03447803482413292, "train_loss_llm": 0.0021441280841827393, "grad_norm": 0.05705321952700615, "global_step": 18210, "epoch": 505, "lr": 0.001} +{"train_loss": 0.02839057520031929, "train_loss_bc": 0.02839057520031929, "train_loss_llm": 0.0, "grad_norm": 0.07256338745355606, "global_step": 18211, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03498983383178711, "train_loss_bc": 0.03498983383178711, "train_loss_llm": 0.0, "grad_norm": 0.0941312238574028, "global_step": 18212, "epoch": 505, "lr": 0.001} +{"train_loss": 0.03686787188053131, "train_loss_bc": 0.03686030954122543, "train_loss_llm": 0.0007562637329101562, "grad_norm": 0.03532365337014198, "global_step": 18213, "epoch": 505, "lr": 0.001} +{"train_loss": 0.036125775426626205, "train_loss_bc": 0.03612019494175911, "train_loss_llm": 0.0005580484867095947, "grad_norm": 0.057184454053640366, "global_step": 18214, "epoch": 505, "lr": 0.001} +{"train_loss": 0.034216748654014535, "train_loss_bc": 0.03472189977765083, "train_loss_llm": 0.0, "grad_norm": 0.0740484669804573, "global_step": 18215, "epoch": 505, "lr": 0.001, "train/cumulative_reward": 61.698936643830145, "train/mean_score": 14.729064788643083, "train/success_rate": 0.4, "test/cumulative_reward": 41.58624176049203, "test/mean_score": 5.643039045505532, "test/success_rate": 0.1, "val_loss": 0.07626928389072418, "train_action_mse_error": 0.037928465753793716} +{"train_loss": 0.0332561731338501, "train_loss_bc": 0.0332561731338501, "train_loss_llm": 0.0, "grad_norm": 0.09693876653909683, "global_step": 18216, "epoch": 506, "lr": 0.001} +{"train_loss": 0.036651741713285446, "train_loss_bc": 0.03664397820830345, "train_loss_llm": 0.0007762610912322998, "grad_norm": 0.03647051379084587, "global_step": 18217, "epoch": 506, "lr": 0.001} +{"train_loss": 0.030716799199581146, "train_loss_bc": 0.030716799199581146, "train_loss_llm": 0.0, "grad_norm": 0.04545845463871956, "global_step": 18218, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03631817549467087, "train_loss_bc": 0.03629983216524124, "train_loss_llm": 0.0018343925476074219, "grad_norm": 0.061379313468933105, "global_step": 18219, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03900062292814255, "train_loss_bc": 0.03900062292814255, "train_loss_llm": 0.0, "grad_norm": 0.07214250415563583, "global_step": 18220, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03149962052702904, "train_loss_bc": 0.03149962052702904, "train_loss_llm": 0.0, "grad_norm": 0.025154229253530502, "global_step": 18221, "epoch": 506, "lr": 0.001} +{"train_loss": 0.032680485397577286, "train_loss_bc": 0.032678768038749695, "train_loss_llm": 0.0001716315746307373, "grad_norm": 0.042775657027959824, "global_step": 18222, "epoch": 506, "lr": 0.001} +{"train_loss": 0.036300964653491974, "train_loss_bc": 0.036300964653491974, "train_loss_llm": 0.0, "grad_norm": 0.0588284507393837, "global_step": 18223, "epoch": 506, "lr": 0.001} +{"train_loss": 0.035149239003658295, "train_loss_bc": 0.035149239003658295, "train_loss_llm": 0.0, "grad_norm": 0.06936910003423691, "global_step": 18224, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03376234695315361, "train_loss_bc": 0.03376234695315361, "train_loss_llm": 0.0, "grad_norm": 0.028043054044246674, "global_step": 18225, "epoch": 506, "lr": 0.001} +{"train_loss": 0.0318257138133049, "train_loss_bc": 0.0318257138133049, "train_loss_llm": 0.0, "grad_norm": 0.039088375866413116, "global_step": 18226, "epoch": 506, "lr": 0.001} +{"train_loss": 0.02973167970776558, "train_loss_bc": 0.02972298301756382, "train_loss_llm": 0.0008697211742401123, "grad_norm": 0.05432866886258125, "global_step": 18227, "epoch": 506, "lr": 0.001} +{"train_loss": 0.02700924687087536, "train_loss_bc": 0.02700924687087536, "train_loss_llm": 0.0, "grad_norm": 0.06950891017913818, "global_step": 18228, "epoch": 506, "lr": 0.001} +{"train_loss": 0.02911739982664585, "train_loss_bc": 0.02911739982664585, "train_loss_llm": 0.0, "grad_norm": 0.024463241919875145, "global_step": 18229, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03413652628660202, "train_loss_bc": 0.03411306068301201, "train_loss_llm": 0.0023465752601623535, "grad_norm": 0.03789360448718071, "global_step": 18230, "epoch": 506, "lr": 0.001} +{"train_loss": 0.037260502576828, "train_loss_bc": 0.037260502576828, "train_loss_llm": 0.0, "grad_norm": 0.05290308967232704, "global_step": 18231, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03535076230764389, "train_loss_bc": 0.03535076230764389, "train_loss_llm": 0.0, "grad_norm": 0.06848397105932236, "global_step": 18232, "epoch": 506, "lr": 0.001} +{"train_loss": 0.028216620907187462, "train_loss_bc": 0.028216620907187462, "train_loss_llm": 0.0, "grad_norm": 0.025277215987443924, "global_step": 18233, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03175972402095795, "train_loss_bc": 0.03175972402095795, "train_loss_llm": 0.0, "grad_norm": 0.034129541367292404, "global_step": 18234, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03161131963133812, "train_loss_bc": 0.03161131963133812, "train_loss_llm": 0.0, "grad_norm": 0.05246273800730705, "global_step": 18235, "epoch": 506, "lr": 0.001} +{"train_loss": 0.04237345978617668, "train_loss_bc": 0.04237345978617668, "train_loss_llm": 0.0, "grad_norm": 0.07510075718164444, "global_step": 18236, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03096582181751728, "train_loss_bc": 0.03096582181751728, "train_loss_llm": 0.0, "grad_norm": 0.03252837806940079, "global_step": 18237, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03332708775997162, "train_loss_bc": 0.03332708775997162, "train_loss_llm": 0.0, "grad_norm": 0.04831109568476677, "global_step": 18238, "epoch": 506, "lr": 0.001} +{"train_loss": 0.031469039618968964, "train_loss_bc": 0.031469039618968964, "train_loss_llm": 0.0, "grad_norm": 0.05427950248122215, "global_step": 18239, "epoch": 506, "lr": 0.001} +{"train_loss": 0.0345035083591938, "train_loss_bc": 0.0345035083591938, "train_loss_llm": 0.0, "grad_norm": 0.06863515824079514, "global_step": 18240, "epoch": 506, "lr": 0.001} +{"train_loss": 0.033450622111558914, "train_loss_bc": 0.033450622111558914, "train_loss_llm": 0.0, "grad_norm": 0.03160686790943146, "global_step": 18241, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03584326431155205, "train_loss_bc": 0.03584326431155205, "train_loss_llm": 0.0, "grad_norm": 0.0436864010989666, "global_step": 18242, "epoch": 506, "lr": 0.001} +{"train_loss": 0.02959255501627922, "train_loss_bc": 0.02959255501627922, "train_loss_llm": 0.0, "grad_norm": 0.05602572485804558, "global_step": 18243, "epoch": 506, "lr": 0.001} +{"train_loss": 0.028483683243393898, "train_loss_bc": 0.028483683243393898, "train_loss_llm": 0.0, "grad_norm": 0.06858690083026886, "global_step": 18244, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03427970036864281, "train_loss_bc": 0.03427970036864281, "train_loss_llm": 0.0, "grad_norm": 0.023090612143278122, "global_step": 18245, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03786157816648483, "train_loss_bc": 0.03786157816648483, "train_loss_llm": 0.0, "grad_norm": 0.040775179862976074, "global_step": 18246, "epoch": 506, "lr": 0.001} +{"train_loss": 0.039105914533138275, "train_loss_bc": 0.039105914533138275, "train_loss_llm": 0.0, "grad_norm": 0.05724853277206421, "global_step": 18247, "epoch": 506, "lr": 0.001} +{"train_loss": 0.0384659469127655, "train_loss_bc": 0.0384649783372879, "train_loss_llm": 9.694695472717285e-05, "grad_norm": 0.09174001961946487, "global_step": 18248, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03599225729703903, "train_loss_bc": 0.03599225729703903, "train_loss_llm": 0.0, "grad_norm": 0.025312988087534904, "global_step": 18249, "epoch": 506, "lr": 0.001} +{"train_loss": 0.0427108071744442, "train_loss_bc": 0.0427108071744442, "train_loss_llm": 0.0, "grad_norm": 0.03824957460165024, "global_step": 18250, "epoch": 506, "lr": 0.001} +{"train_loss": 0.03400965485100945, "train_loss_bc": 0.03456666320562363, "train_loss_llm": 0.0, "grad_norm": 0.05053910240530968, "global_step": 18251, "epoch": 506, "lr": 0.001, "val_loss": 0.07546526938676834} +{"train_loss": 0.03467094898223877, "train_loss_bc": 0.0346655398607254, "train_loss_llm": 0.0005408823490142822, "grad_norm": 0.058207664638757706, "global_step": 18252, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03634963184595108, "train_loss_bc": 0.03634963184595108, "train_loss_llm": 0.0, "grad_norm": 0.028706712648272514, "global_step": 18253, "epoch": 507, "lr": 0.001} +{"train_loss": 0.038723867386579514, "train_loss_bc": 0.038723867386579514, "train_loss_llm": 0.0, "grad_norm": 0.048043277114629745, "global_step": 18254, "epoch": 507, "lr": 0.001} +{"train_loss": 0.0407736599445343, "train_loss_bc": 0.040771715342998505, "train_loss_llm": 0.0001945197582244873, "grad_norm": 0.06338825821876526, "global_step": 18255, "epoch": 507, "lr": 0.001} +{"train_loss": 0.02871762588620186, "train_loss_bc": 0.02871762588620186, "train_loss_llm": 0.0, "grad_norm": 0.07570235431194305, "global_step": 18256, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03942582756280899, "train_loss_bc": 0.03942582756280899, "train_loss_llm": 0.0, "grad_norm": 0.027132505550980568, "global_step": 18257, "epoch": 507, "lr": 0.001} +{"train_loss": 0.033495329320430756, "train_loss_bc": 0.033495329320430756, "train_loss_llm": 0.0, "grad_norm": 0.04613430052995682, "global_step": 18258, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03223578259348869, "train_loss_bc": 0.03223578259348869, "train_loss_llm": 0.0, "grad_norm": 0.06247635558247566, "global_step": 18259, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03359995037317276, "train_loss_bc": 0.03359995037317276, "train_loss_llm": 0.0, "grad_norm": 0.07085900008678436, "global_step": 18260, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03613389655947685, "train_loss_bc": 0.03613389655947685, "train_loss_llm": 0.0, "grad_norm": 0.02474263124167919, "global_step": 18261, "epoch": 507, "lr": 0.001} +{"train_loss": 0.031223095953464508, "train_loss_bc": 0.031223095953464508, "train_loss_llm": 0.0, "grad_norm": 0.03490104153752327, "global_step": 18262, "epoch": 507, "lr": 0.001} +{"train_loss": 0.030492540448904037, "train_loss_bc": 0.030474171042442322, "train_loss_llm": 0.0018369555473327637, "grad_norm": 0.044391240924596786, "global_step": 18263, "epoch": 507, "lr": 0.001} +{"train_loss": 0.038635898381471634, "train_loss_bc": 0.038635898381471634, "train_loss_llm": 0.0, "grad_norm": 0.0662064179778099, "global_step": 18264, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03450760990381241, "train_loss_bc": 0.03450760990381241, "train_loss_llm": 0.0, "grad_norm": 0.02987714298069477, "global_step": 18265, "epoch": 507, "lr": 0.001} +{"train_loss": 0.02925056964159012, "train_loss_bc": 0.029250377789139748, "train_loss_llm": 1.9252300262451172e-05, "grad_norm": 0.04514945298433304, "global_step": 18266, "epoch": 507, "lr": 0.001} +{"train_loss": 0.030367335304617882, "train_loss_bc": 0.030349500477313995, "train_loss_llm": 0.0017834603786468506, "grad_norm": 0.060245055705308914, "global_step": 18267, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03607025742530823, "train_loss_bc": 0.03607025742530823, "train_loss_llm": 0.0, "grad_norm": 0.08427654951810837, "global_step": 18268, "epoch": 507, "lr": 0.001} +{"train_loss": 0.02809915691614151, "train_loss_bc": 0.02809915691614151, "train_loss_llm": 0.0, "grad_norm": 0.02677949145436287, "global_step": 18269, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03445515036582947, "train_loss_bc": 0.034453488886356354, "train_loss_llm": 0.00016617774963378906, "grad_norm": 0.04000767320394516, "global_step": 18270, "epoch": 507, "lr": 0.001} +{"train_loss": 0.0344122089445591, "train_loss_bc": 0.03440765291452408, "train_loss_llm": 0.00045558810234069824, "grad_norm": 0.055619016289711, "global_step": 18271, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03361741080880165, "train_loss_bc": 0.03361741080880165, "train_loss_llm": 0.0, "grad_norm": 0.06624539941549301, "global_step": 18272, "epoch": 507, "lr": 0.001} +{"train_loss": 0.031012319028377533, "train_loss_bc": 0.031012319028377533, "train_loss_llm": 0.0, "grad_norm": 0.029141735285520554, "global_step": 18273, "epoch": 507, "lr": 0.001} +{"train_loss": 0.030142419040203094, "train_loss_bc": 0.030142419040203094, "train_loss_llm": 0.0, "grad_norm": 0.04526229947805405, "global_step": 18274, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03172750025987625, "train_loss_bc": 0.03172750025987625, "train_loss_llm": 0.0, "grad_norm": 0.05719604715704918, "global_step": 18275, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03684266656637192, "train_loss_bc": 0.03684266656637192, "train_loss_llm": 0.0, "grad_norm": 0.06945318728685379, "global_step": 18276, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03295395150780678, "train_loss_bc": 0.03293062001466751, "train_loss_llm": 0.002333134412765503, "grad_norm": 0.026973484084010124, "global_step": 18277, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03588929399847984, "train_loss_bc": 0.03588893637061119, "train_loss_llm": 3.5643577575683594e-05, "grad_norm": 0.0531550869345665, "global_step": 18278, "epoch": 507, "lr": 0.001} +{"train_loss": 0.032979853451251984, "train_loss_bc": 0.032979853451251984, "train_loss_llm": 0.0, "grad_norm": 0.061174292117357254, "global_step": 18279, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03008611686527729, "train_loss_bc": 0.03008611686527729, "train_loss_llm": 0.0, "grad_norm": 0.0765383318066597, "global_step": 18280, "epoch": 507, "lr": 0.001} +{"train_loss": 0.04470345377922058, "train_loss_bc": 0.04470345377922058, "train_loss_llm": 0.0, "grad_norm": 0.036499541252851486, "global_step": 18281, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03173665329813957, "train_loss_bc": 0.03173353150486946, "train_loss_llm": 0.0003122389316558838, "grad_norm": 0.0569586381316185, "global_step": 18282, "epoch": 507, "lr": 0.001} +{"train_loss": 0.03074382245540619, "train_loss_bc": 0.03074382245540619, "train_loss_llm": 0.0, "grad_norm": 0.06857718527317047, "global_step": 18283, "epoch": 507, "lr": 0.001} +{"train_loss": 0.036208607256412506, "train_loss_bc": 0.036208607256412506, "train_loss_llm": 0.0, "grad_norm": 0.0913560688495636, "global_step": 18284, "epoch": 507, "lr": 0.001} +{"train_loss": 0.038926299661397934, "train_loss_bc": 0.038926299661397934, "train_loss_llm": 0.0, "grad_norm": 0.026322320103645325, "global_step": 18285, "epoch": 507, "lr": 0.001} +{"train_loss": 0.033168453723192215, "train_loss_bc": 0.033168453723192215, "train_loss_llm": 0.0, "grad_norm": 0.04121362045407295, "global_step": 18286, "epoch": 507, "lr": 0.001} +{"train_loss": 0.034070837414926954, "train_loss_bc": 0.03416964039206505, "train_loss_llm": 0.000134199857711792, "grad_norm": 0.054865047335624695, "global_step": 18287, "epoch": 507, "lr": 0.001, "val_loss": 0.07566823065280914} +{"train_loss": 0.03366877883672714, "train_loss_bc": 0.03366877883672714, "train_loss_llm": 0.0, "grad_norm": 0.07472355663776398, "global_step": 18288, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03606708347797394, "train_loss_bc": 0.03606708347797394, "train_loss_llm": 0.0, "grad_norm": 0.036822449415922165, "global_step": 18289, "epoch": 508, "lr": 0.001} +{"train_loss": 0.0414801761507988, "train_loss_bc": 0.0414801761507988, "train_loss_llm": 0.0, "grad_norm": 0.053079310804605484, "global_step": 18290, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03561481088399887, "train_loss_bc": 0.03561363369226456, "train_loss_llm": 0.00011777877807617188, "grad_norm": 0.06469079107046127, "global_step": 18291, "epoch": 508, "lr": 0.001} +{"train_loss": 0.030683236196637154, "train_loss_bc": 0.030683236196637154, "train_loss_llm": 0.0, "grad_norm": 0.0763474777340889, "global_step": 18292, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03723285719752312, "train_loss_bc": 0.03723285719752312, "train_loss_llm": 0.0, "grad_norm": 0.039835888892412186, "global_step": 18293, "epoch": 508, "lr": 0.001} +{"train_loss": 0.033323608338832855, "train_loss_bc": 0.033323608338832855, "train_loss_llm": 0.0, "grad_norm": 0.06035524234175682, "global_step": 18294, "epoch": 508, "lr": 0.001} +{"train_loss": 0.026497822254896164, "train_loss_bc": 0.026497822254896164, "train_loss_llm": 0.0, "grad_norm": 0.07531221210956573, "global_step": 18295, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03359699249267578, "train_loss_bc": 0.03359699249267578, "train_loss_llm": 0.0, "grad_norm": 0.09303538501262665, "global_step": 18296, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03272693604230881, "train_loss_bc": 0.03272693604230881, "train_loss_llm": 0.0, "grad_norm": 0.02919081598520279, "global_step": 18297, "epoch": 508, "lr": 0.001} +{"train_loss": 0.027133572846651077, "train_loss_bc": 0.027133572846651077, "train_loss_llm": 0.0, "grad_norm": 0.04062824696302414, "global_step": 18298, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03450809419155121, "train_loss_bc": 0.03450809419155121, "train_loss_llm": 0.0, "grad_norm": 0.06750844419002533, "global_step": 18299, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03544227406382561, "train_loss_bc": 0.03544227406382561, "train_loss_llm": 0.0, "grad_norm": 0.09283405542373657, "global_step": 18300, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03444267064332962, "train_loss_bc": 0.034439343959093094, "train_loss_llm": 0.00033277273178100586, "grad_norm": 0.028198793530464172, "global_step": 18301, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03158938139677048, "train_loss_bc": 0.03158938139677048, "train_loss_llm": 0.0, "grad_norm": 0.03812389820814133, "global_step": 18302, "epoch": 508, "lr": 0.001} +{"train_loss": 0.034839849919080734, "train_loss_bc": 0.03483216091990471, "train_loss_llm": 0.0007689893245697021, "grad_norm": 0.05193459615111351, "global_step": 18303, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03796827048063278, "train_loss_bc": 0.03796827048063278, "train_loss_llm": 0.0, "grad_norm": 0.05854516103863716, "global_step": 18304, "epoch": 508, "lr": 0.001} +{"train_loss": 0.02932037226855755, "train_loss_bc": 0.02932037226855755, "train_loss_llm": 0.0, "grad_norm": 0.025954220443964005, "global_step": 18305, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03816727176308632, "train_loss_bc": 0.03816727176308632, "train_loss_llm": 0.0, "grad_norm": 0.03738538548350334, "global_step": 18306, "epoch": 508, "lr": 0.001} +{"train_loss": 0.04043687507510185, "train_loss_bc": 0.04043687507510185, "train_loss_llm": 0.0, "grad_norm": 0.06381989270448685, "global_step": 18307, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03620021417737007, "train_loss_bc": 0.03620021417737007, "train_loss_llm": 0.0, "grad_norm": 0.08051279932260513, "global_step": 18308, "epoch": 508, "lr": 0.001} +{"train_loss": 0.0364995114505291, "train_loss_bc": 0.0364995114505291, "train_loss_llm": 0.0, "grad_norm": 0.025533830747008324, "global_step": 18309, "epoch": 508, "lr": 0.001} +{"train_loss": 0.037694212049245834, "train_loss_bc": 0.03767244145274162, "train_loss_llm": 0.0021771490573883057, "grad_norm": 0.04917902871966362, "global_step": 18310, "epoch": 508, "lr": 0.001} +{"train_loss": 0.02811402641236782, "train_loss_bc": 0.02811402641236782, "train_loss_llm": 0.0, "grad_norm": 0.05412852764129639, "global_step": 18311, "epoch": 508, "lr": 0.001} +{"train_loss": 0.032948754727840424, "train_loss_bc": 0.032948754727840424, "train_loss_llm": 0.0, "grad_norm": 0.07220358401536942, "global_step": 18312, "epoch": 508, "lr": 0.001} +{"train_loss": 0.026439756155014038, "train_loss_bc": 0.026439756155014038, "train_loss_llm": 0.0, "grad_norm": 0.023518703877925873, "global_step": 18313, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03439340740442276, "train_loss_bc": 0.03439261391758919, "train_loss_llm": 7.930397987365723e-05, "grad_norm": 0.03634285554289818, "global_step": 18314, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03009284846484661, "train_loss_bc": 0.03009284846484661, "train_loss_llm": 0.0, "grad_norm": 0.052363358438014984, "global_step": 18315, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03459014743566513, "train_loss_bc": 0.03459014743566513, "train_loss_llm": 0.0, "grad_norm": 0.06541454046964645, "global_step": 18316, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03468957543373108, "train_loss_bc": 0.03468957543373108, "train_loss_llm": 0.0, "grad_norm": 0.03293851390480995, "global_step": 18317, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03457639738917351, "train_loss_bc": 0.03457639738917351, "train_loss_llm": 0.0, "grad_norm": 0.05222500115633011, "global_step": 18318, "epoch": 508, "lr": 0.001} +{"train_loss": 0.035566192120313644, "train_loss_bc": 0.035566192120313644, "train_loss_llm": 0.0, "grad_norm": 0.07276353240013123, "global_step": 18319, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03656145930290222, "train_loss_bc": 0.03656145930290222, "train_loss_llm": 0.0, "grad_norm": 0.0802757516503334, "global_step": 18320, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03718153387308121, "train_loss_bc": 0.03715500980615616, "train_loss_llm": 0.0026523470878601074, "grad_norm": 0.026825224980711937, "global_step": 18321, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03174164891242981, "train_loss_bc": 0.031727664172649384, "train_loss_llm": 0.0013985931873321533, "grad_norm": 0.035200681537389755, "global_step": 18322, "epoch": 508, "lr": 0.001} +{"train_loss": 0.03422095347195864, "train_loss_bc": 0.03990118205547333, "train_loss_llm": 0.0022524595260620117, "grad_norm": 0.05489826574921608, "global_step": 18323, "epoch": 508, "lr": 0.001, "val_loss": 0.07470294833183289} +{"train_loss": 0.03982746973633766, "train_loss_bc": 0.039822064340114594, "train_loss_llm": 0.0005404949188232422, "grad_norm": 0.07705339044332504, "global_step": 18324, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03752259537577629, "train_loss_bc": 0.03752259537577629, "train_loss_llm": 0.0, "grad_norm": 0.031910475343465805, "global_step": 18325, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03267696872353554, "train_loss_bc": 0.03266280144453049, "train_loss_llm": 0.0014166831970214844, "grad_norm": 0.046717554330825806, "global_step": 18326, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03291483595967293, "train_loss_bc": 0.03291483595967293, "train_loss_llm": 0.0, "grad_norm": 0.061030369251966476, "global_step": 18327, "epoch": 509, "lr": 0.001} +{"train_loss": 0.02721055969595909, "train_loss_bc": 0.02721055969595909, "train_loss_llm": 0.0, "grad_norm": 0.06799586862325668, "global_step": 18328, "epoch": 509, "lr": 0.001} +{"train_loss": 0.024278409779071808, "train_loss_bc": 0.024278409779071808, "train_loss_llm": 0.0, "grad_norm": 0.01883193850517273, "global_step": 18329, "epoch": 509, "lr": 0.001} +{"train_loss": 0.035921331495046616, "train_loss_bc": 0.03591718152165413, "train_loss_llm": 0.00041493773460388184, "grad_norm": 0.03834809735417366, "global_step": 18330, "epoch": 509, "lr": 0.001} +{"train_loss": 0.030206602066755295, "train_loss_bc": 0.030197646468877792, "train_loss_llm": 0.0008956491947174072, "grad_norm": 0.04744631424546242, "global_step": 18331, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03918104246258736, "train_loss_bc": 0.03918104246258736, "train_loss_llm": 0.0, "grad_norm": 0.06622710824012756, "global_step": 18332, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03641252592206001, "train_loss_bc": 0.03641252592206001, "train_loss_llm": 0.0, "grad_norm": 0.03202902153134346, "global_step": 18333, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03499401733279228, "train_loss_bc": 0.03499401733279228, "train_loss_llm": 0.0, "grad_norm": 0.042030371725559235, "global_step": 18334, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03498821333050728, "train_loss_bc": 0.03498821333050728, "train_loss_llm": 0.0, "grad_norm": 0.05486157536506653, "global_step": 18335, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03799418359994888, "train_loss_bc": 0.03799418359994888, "train_loss_llm": 0.0, "grad_norm": 0.07753928005695343, "global_step": 18336, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03656397759914398, "train_loss_bc": 0.03656397759914398, "train_loss_llm": 0.0, "grad_norm": 0.025541966781020164, "global_step": 18337, "epoch": 509, "lr": 0.001} +{"train_loss": 0.031149813905358315, "train_loss_bc": 0.031149813905358315, "train_loss_llm": 0.0, "grad_norm": 0.04615673050284386, "global_step": 18338, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03707530349493027, "train_loss_bc": 0.037061192095279694, "train_loss_llm": 0.0014111995697021484, "grad_norm": 0.06253182142972946, "global_step": 18339, "epoch": 509, "lr": 0.001} +{"train_loss": 0.035808540880680084, "train_loss_bc": 0.035808540880680084, "train_loss_llm": 0.0, "grad_norm": 0.07659447193145752, "global_step": 18340, "epoch": 509, "lr": 0.001} +{"train_loss": 0.031015096232295036, "train_loss_bc": 0.031015096232295036, "train_loss_llm": 0.0, "grad_norm": 0.027265215292572975, "global_step": 18341, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03539358824491501, "train_loss_bc": 0.03539172559976578, "train_loss_llm": 0.0001862645149230957, "grad_norm": 0.040006302297115326, "global_step": 18342, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03998604416847229, "train_loss_bc": 0.03998604416847229, "train_loss_llm": 0.0, "grad_norm": 0.062284018844366074, "global_step": 18343, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03028511442244053, "train_loss_bc": 0.03028511442244053, "train_loss_llm": 0.0, "grad_norm": 0.08054137974977493, "global_step": 18344, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03528464958071709, "train_loss_bc": 0.03528464958071709, "train_loss_llm": 0.0, "grad_norm": 0.02635609544813633, "global_step": 18345, "epoch": 509, "lr": 0.001} +{"train_loss": 0.033364418894052505, "train_loss_bc": 0.03334594890475273, "train_loss_llm": 0.0018471181392669678, "grad_norm": 0.0430595800280571, "global_step": 18346, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03192285820841789, "train_loss_bc": 0.03189771622419357, "train_loss_llm": 0.0025143325328826904, "grad_norm": 0.060152191668748856, "global_step": 18347, "epoch": 509, "lr": 0.001} +{"train_loss": 0.030623827129602432, "train_loss_bc": 0.030622845515608788, "train_loss_llm": 9.819865226745605e-05, "grad_norm": 0.07158593088388443, "global_step": 18348, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03048647753894329, "train_loss_bc": 0.03048647753894329, "train_loss_llm": 0.0, "grad_norm": 0.029116211459040642, "global_step": 18349, "epoch": 509, "lr": 0.001} +{"train_loss": 0.023510245606303215, "train_loss_bc": 0.023510245606303215, "train_loss_llm": 0.0, "grad_norm": 0.032602664083242416, "global_step": 18350, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03403859585523605, "train_loss_bc": 0.03401778265833855, "train_loss_llm": 0.0020813345909118652, "grad_norm": 0.04808706417679787, "global_step": 18351, "epoch": 509, "lr": 0.001} +{"train_loss": 0.030099516734480858, "train_loss_bc": 0.030099516734480858, "train_loss_llm": 0.0, "grad_norm": 0.050533294677734375, "global_step": 18352, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03236881271004677, "train_loss_bc": 0.03236881271004677, "train_loss_llm": 0.0, "grad_norm": 0.023317331448197365, "global_step": 18353, "epoch": 509, "lr": 0.001} +{"train_loss": 0.0375681035220623, "train_loss_bc": 0.0375681035220623, "train_loss_llm": 0.0, "grad_norm": 0.0355951227247715, "global_step": 18354, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03423250839114189, "train_loss_bc": 0.03423250839114189, "train_loss_llm": 0.0, "grad_norm": 0.04187091439962387, "global_step": 18355, "epoch": 509, "lr": 0.001} +{"train_loss": 0.04130950942635536, "train_loss_bc": 0.04130950942635536, "train_loss_llm": 0.0, "grad_norm": 0.05683270841836929, "global_step": 18356, "epoch": 509, "lr": 0.001} +{"train_loss": 0.039393890649080276, "train_loss_bc": 0.039393890649080276, "train_loss_llm": 0.0, "grad_norm": 0.031717363744974136, "global_step": 18357, "epoch": 509, "lr": 0.001} +{"train_loss": 0.03188765048980713, "train_loss_bc": 0.03188765048980713, "train_loss_llm": 0.0, "grad_norm": 0.0486920140683651, "global_step": 18358, "epoch": 509, "lr": 0.001} +{"train_loss": 0.034079268471234374, "train_loss_bc": 0.03935636579990387, "train_loss_llm": 0.0, "grad_norm": 0.06358107924461365, "global_step": 18359, "epoch": 509, "lr": 0.001, "val_loss": 0.07555588334798813} +{"train_loss": 0.03326166793704033, "train_loss_bc": 0.03326166793704033, "train_loss_llm": 0.0, "grad_norm": 0.0802793875336647, "global_step": 18360, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03226996213197708, "train_loss_bc": 0.03226996213197708, "train_loss_llm": 0.0, "grad_norm": 0.029132135212421417, "global_step": 18361, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03974192962050438, "train_loss_bc": 0.03974192962050438, "train_loss_llm": 0.0, "grad_norm": 0.04711737483739853, "global_step": 18362, "epoch": 510, "lr": 0.001} +{"train_loss": 0.029213672503829002, "train_loss_bc": 0.02921096608042717, "train_loss_llm": 0.00027060508728027344, "grad_norm": 0.0611078217625618, "global_step": 18363, "epoch": 510, "lr": 0.001} +{"train_loss": 0.039431795477867126, "train_loss_bc": 0.039431795477867126, "train_loss_llm": 0.0, "grad_norm": 0.0742429867386818, "global_step": 18364, "epoch": 510, "lr": 0.001} +{"train_loss": 0.02840263582766056, "train_loss_bc": 0.02840263582766056, "train_loss_llm": 0.0, "grad_norm": 0.028631815686821938, "global_step": 18365, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03703469783067703, "train_loss_bc": 0.03703469783067703, "train_loss_llm": 0.0, "grad_norm": 0.040180038660764694, "global_step": 18366, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03872614726424217, "train_loss_bc": 0.03872614726424217, "train_loss_llm": 0.0, "grad_norm": 0.052821241319179535, "global_step": 18367, "epoch": 510, "lr": 0.001} +{"train_loss": 0.030225230380892754, "train_loss_bc": 0.030225230380892754, "train_loss_llm": 0.0, "grad_norm": 0.0640064999461174, "global_step": 18368, "epoch": 510, "lr": 0.001} +{"train_loss": 0.031016496941447258, "train_loss_bc": 0.031016496941447258, "train_loss_llm": 0.0, "grad_norm": 0.02498823031783104, "global_step": 18369, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03377384692430496, "train_loss_bc": 0.033767737448215485, "train_loss_llm": 0.0006110668182373047, "grad_norm": 0.04551655054092407, "global_step": 18370, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03526033088564873, "train_loss_bc": 0.03524758294224739, "train_loss_llm": 0.0012746155261993408, "grad_norm": 0.06824208796024323, "global_step": 18371, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03751697018742561, "train_loss_bc": 0.03751697018742561, "train_loss_llm": 0.0, "grad_norm": 0.07966689020395279, "global_step": 18372, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03323609381914139, "train_loss_bc": 0.03322465345263481, "train_loss_llm": 0.0011440813541412354, "grad_norm": 0.02388189174234867, "global_step": 18373, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03868011385202408, "train_loss_bc": 0.03866194188594818, "train_loss_llm": 0.0018170475959777832, "grad_norm": 0.040660083293914795, "global_step": 18374, "epoch": 510, "lr": 0.001} +{"train_loss": 0.044553887099027634, "train_loss_bc": 0.044553887099027634, "train_loss_llm": 0.0, "grad_norm": 0.06341834366321564, "global_step": 18375, "epoch": 510, "lr": 0.001} +{"train_loss": 0.029197363182902336, "train_loss_bc": 0.029197363182902336, "train_loss_llm": 0.0, "grad_norm": 0.07431330531835556, "global_step": 18376, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03412829712033272, "train_loss_bc": 0.03412829712033272, "train_loss_llm": 0.0, "grad_norm": 0.026761015877127647, "global_step": 18377, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03371449187397957, "train_loss_bc": 0.03369392082095146, "train_loss_llm": 0.0020570456981658936, "grad_norm": 0.035106100142002106, "global_step": 18378, "epoch": 510, "lr": 0.001} +{"train_loss": 0.032964736223220825, "train_loss_bc": 0.032964736223220825, "train_loss_llm": 0.0, "grad_norm": 0.04898617044091225, "global_step": 18379, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03146203234791756, "train_loss_bc": 0.03146203234791756, "train_loss_llm": 0.0, "grad_norm": 0.06707489490509033, "global_step": 18380, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03472067043185234, "train_loss_bc": 0.03469258174300194, "train_loss_llm": 0.00280877947807312, "grad_norm": 0.025577187538146973, "global_step": 18381, "epoch": 510, "lr": 0.001} +{"train_loss": 0.030361920595169067, "train_loss_bc": 0.030361920595169067, "train_loss_llm": 0.0, "grad_norm": 0.04079379513859749, "global_step": 18382, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03145391494035721, "train_loss_bc": 0.03145391494035721, "train_loss_llm": 0.0, "grad_norm": 0.05112576484680176, "global_step": 18383, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03406086936593056, "train_loss_bc": 0.034059636294841766, "train_loss_llm": 0.0001233220100402832, "grad_norm": 0.05841507017612457, "global_step": 18384, "epoch": 510, "lr": 0.001} +{"train_loss": 0.030391596257686615, "train_loss_bc": 0.030391596257686615, "train_loss_llm": 0.0, "grad_norm": 0.02943316660821438, "global_step": 18385, "epoch": 510, "lr": 0.001} +{"train_loss": 0.0368085652589798, "train_loss_bc": 0.03679927438497543, "train_loss_llm": 0.0009290874004364014, "grad_norm": 0.03777308017015457, "global_step": 18386, "epoch": 510, "lr": 0.001} +{"train_loss": 0.044316716492176056, "train_loss_bc": 0.044316716492176056, "train_loss_llm": 0.0, "grad_norm": 0.057550083845853806, "global_step": 18387, "epoch": 510, "lr": 0.001} +{"train_loss": 0.02854018844664097, "train_loss_bc": 0.02854018844664097, "train_loss_llm": 0.0, "grad_norm": 0.07220485061407089, "global_step": 18388, "epoch": 510, "lr": 0.001} +{"train_loss": 0.029830202460289, "train_loss_bc": 0.029830202460289, "train_loss_llm": 0.0, "grad_norm": 0.023460038006305695, "global_step": 18389, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03758849576115608, "train_loss_bc": 0.03758849576115608, "train_loss_llm": 0.0, "grad_norm": 0.045334719121456146, "global_step": 18390, "epoch": 510, "lr": 0.001} +{"train_loss": 0.031530801206827164, "train_loss_bc": 0.031507086008787155, "train_loss_llm": 0.002371490001678467, "grad_norm": 0.0588684156537056, "global_step": 18391, "epoch": 510, "lr": 0.001} +{"train_loss": 0.04006184637546539, "train_loss_bc": 0.04006184637546539, "train_loss_llm": 0.0, "grad_norm": 0.07322590053081512, "global_step": 18392, "epoch": 510, "lr": 0.001} +{"train_loss": 0.03150373324751854, "train_loss_bc": 0.031491659581661224, "train_loss_llm": 0.0012074410915374756, "grad_norm": 0.03321301192045212, "global_step": 18393, "epoch": 510, "lr": 0.001} +{"train_loss": 0.026635251939296722, "train_loss_bc": 0.026635251939296722, "train_loss_llm": 0.0, "grad_norm": 0.05152211710810661, "global_step": 18394, "epoch": 510, "lr": 0.001} +{"train_loss": 0.0339059302997258, "train_loss_bc": 0.028996318578720093, "train_loss_llm": 0.0, "grad_norm": 0.06627020984888077, "global_step": 18395, "epoch": 510, "lr": 0.001, "train/cumulative_reward": 61.33183409000814, "train/mean_score": 19.834803628176875, "train/success_rate": 0.6, "test/cumulative_reward": 30.70367971250508, "test/mean_score": 6.535758742426799, "test/success_rate": 0.16, "val_loss": 0.07540136575698853, "train_action_mse_error": 0.036936502903699875} +{"train_loss": 0.035009536892175674, "train_loss_bc": 0.035009536892175674, "train_loss_llm": 0.0, "grad_norm": 0.07993244379758835, "global_step": 18396, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03566910699009895, "train_loss_bc": 0.035668548196554184, "train_loss_llm": 5.5849552154541016e-05, "grad_norm": 0.03211110830307007, "global_step": 18397, "epoch": 511, "lr": 0.001} +{"train_loss": 0.0321856252849102, "train_loss_bc": 0.0321856252849102, "train_loss_llm": 0.0, "grad_norm": 0.04339595139026642, "global_step": 18398, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03251277655363083, "train_loss_bc": 0.03251277655363083, "train_loss_llm": 0.0, "grad_norm": 0.05183576047420502, "global_step": 18399, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03951592743396759, "train_loss_bc": 0.03951592743396759, "train_loss_llm": 0.0, "grad_norm": 0.0681244358420372, "global_step": 18400, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03139181435108185, "train_loss_bc": 0.03139181435108185, "train_loss_llm": 0.0, "grad_norm": 0.029696330428123474, "global_step": 18401, "epoch": 511, "lr": 0.001} +{"train_loss": 0.0364559143781662, "train_loss_bc": 0.0364559143781662, "train_loss_llm": 0.0, "grad_norm": 0.04532170668244362, "global_step": 18402, "epoch": 511, "lr": 0.001} +{"train_loss": 0.02967653051018715, "train_loss_bc": 0.029632478952407837, "train_loss_llm": 0.004405081272125244, "grad_norm": 0.06386236846446991, "global_step": 18403, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03352470323443413, "train_loss_bc": 0.03352470323443413, "train_loss_llm": 0.0, "grad_norm": 0.07353992015123367, "global_step": 18404, "epoch": 511, "lr": 0.001} +{"train_loss": 0.02655266970396042, "train_loss_bc": 0.02655266970396042, "train_loss_llm": 0.0, "grad_norm": 0.02651715651154518, "global_step": 18405, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03002026304602623, "train_loss_bc": 0.03002026304602623, "train_loss_llm": 0.0, "grad_norm": 0.04107152298092842, "global_step": 18406, "epoch": 511, "lr": 0.001} +{"train_loss": 0.0355590358376503, "train_loss_bc": 0.0355590358376503, "train_loss_llm": 0.0, "grad_norm": 0.05374770238995552, "global_step": 18407, "epoch": 511, "lr": 0.001} +{"train_loss": 0.031386420130729675, "train_loss_bc": 0.031386420130729675, "train_loss_llm": 0.0, "grad_norm": 0.0672992691397667, "global_step": 18408, "epoch": 511, "lr": 0.001} +{"train_loss": 0.04004592448472977, "train_loss_bc": 0.04004592448472977, "train_loss_llm": 0.0, "grad_norm": 0.035563286393880844, "global_step": 18409, "epoch": 511, "lr": 0.001} +{"train_loss": 0.040395740419626236, "train_loss_bc": 0.040395740419626236, "train_loss_llm": 0.0, "grad_norm": 0.060004860162734985, "global_step": 18410, "epoch": 511, "lr": 0.001} +{"train_loss": 0.029287969693541527, "train_loss_bc": 0.029287969693541527, "train_loss_llm": 0.0, "grad_norm": 0.06875503808259964, "global_step": 18411, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03492295742034912, "train_loss_bc": 0.03492295742034912, "train_loss_llm": 0.0, "grad_norm": 0.08600273728370667, "global_step": 18412, "epoch": 511, "lr": 0.001} +{"train_loss": 0.029091423377394676, "train_loss_bc": 0.029091423377394676, "train_loss_llm": 0.0, "grad_norm": 0.026527434587478638, "global_step": 18413, "epoch": 511, "lr": 0.001} +{"train_loss": 0.04522966593503952, "train_loss_bc": 0.04521482065320015, "train_loss_llm": 0.001484513282775879, "grad_norm": 0.05447878688573837, "global_step": 18414, "epoch": 511, "lr": 0.001} +{"train_loss": 0.033139292150735855, "train_loss_bc": 0.033139292150735855, "train_loss_llm": 0.0, "grad_norm": 0.05634000524878502, "global_step": 18415, "epoch": 511, "lr": 0.001} +{"train_loss": 0.030886705964803696, "train_loss_bc": 0.030886705964803696, "train_loss_llm": 0.0, "grad_norm": 0.06894625723361969, "global_step": 18416, "epoch": 511, "lr": 0.001} +{"train_loss": 0.033566784113645554, "train_loss_bc": 0.033550333231687546, "train_loss_llm": 0.0016449391841888428, "grad_norm": 0.03402339667081833, "global_step": 18417, "epoch": 511, "lr": 0.001} +{"train_loss": 0.0347994789481163, "train_loss_bc": 0.0347994789481163, "train_loss_llm": 0.0, "grad_norm": 0.04886207729578018, "global_step": 18418, "epoch": 511, "lr": 0.001} +{"train_loss": 0.02811906300485134, "train_loss_bc": 0.02811906300485134, "train_loss_llm": 0.0, "grad_norm": 0.06248876079916954, "global_step": 18419, "epoch": 511, "lr": 0.001} +{"train_loss": 0.030331742018461227, "train_loss_bc": 0.030331742018461227, "train_loss_llm": 0.0, "grad_norm": 0.0751950815320015, "global_step": 18420, "epoch": 511, "lr": 0.001} +{"train_loss": 0.037398841232061386, "train_loss_bc": 0.037398841232061386, "train_loss_llm": 0.0, "grad_norm": 0.03026185743510723, "global_step": 18421, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03348890691995621, "train_loss_bc": 0.03348890691995621, "train_loss_llm": 0.0, "grad_norm": 0.04381652921438217, "global_step": 18422, "epoch": 511, "lr": 0.001} +{"train_loss": 0.036253560334444046, "train_loss_bc": 0.036253560334444046, "train_loss_llm": 0.0, "grad_norm": 0.04909219965338707, "global_step": 18423, "epoch": 511, "lr": 0.001} +{"train_loss": 0.032228726893663406, "train_loss_bc": 0.032228726893663406, "train_loss_llm": 0.0, "grad_norm": 0.06402285397052765, "global_step": 18424, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03325337544083595, "train_loss_bc": 0.03325337544083595, "train_loss_llm": 0.0, "grad_norm": 0.024885663762688637, "global_step": 18425, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03587911278009415, "train_loss_bc": 0.03587911278009415, "train_loss_llm": 0.0, "grad_norm": 0.04493223875761032, "global_step": 18426, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03486635163426399, "train_loss_bc": 0.03486635163426399, "train_loss_llm": 0.0, "grad_norm": 0.06386356800794601, "global_step": 18427, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03403718024492264, "train_loss_bc": 0.034032758325338364, "train_loss_llm": 0.00044226646423339844, "grad_norm": 0.07486144453287125, "global_step": 18428, "epoch": 511, "lr": 0.001} +{"train_loss": 0.038163576275110245, "train_loss_bc": 0.038163576275110245, "train_loss_llm": 0.0, "grad_norm": 0.029454750940203667, "global_step": 18429, "epoch": 511, "lr": 0.001} +{"train_loss": 0.0299014113843441, "train_loss_bc": 0.029890157282352448, "train_loss_llm": 0.001125425100326538, "grad_norm": 0.05254366993904114, "global_step": 18430, "epoch": 511, "lr": 0.001} +{"train_loss": 0.03404259314346644, "train_loss_bc": 0.04078523814678192, "train_loss_llm": 0.0, "grad_norm": 0.07629379630088806, "global_step": 18431, "epoch": 511, "lr": 0.001, "val_loss": 0.076311394572258} +{"train_loss": 0.03351351618766785, "train_loss_bc": 0.03351351618766785, "train_loss_llm": 0.0, "grad_norm": 0.09771741926670074, "global_step": 18432, "epoch": 512, "lr": 0.001} +{"train_loss": 0.036904655396938324, "train_loss_bc": 0.036904655396938324, "train_loss_llm": 0.0, "grad_norm": 0.03549641743302345, "global_step": 18433, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03186911344528198, "train_loss_bc": 0.03185494989156723, "train_loss_llm": 0.0014164149761199951, "grad_norm": 0.05577344819903374, "global_step": 18434, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03391021862626076, "train_loss_bc": 0.03390797972679138, "train_loss_llm": 0.00022399425506591797, "grad_norm": 0.07167451083660126, "global_step": 18435, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03585820645093918, "train_loss_bc": 0.03585820645093918, "train_loss_llm": 0.0, "grad_norm": 0.09125520288944244, "global_step": 18436, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03763087838888168, "train_loss_bc": 0.03763087838888168, "train_loss_llm": 0.0, "grad_norm": 0.022140473127365112, "global_step": 18437, "epoch": 512, "lr": 0.001} +{"train_loss": 0.036401402205228806, "train_loss_bc": 0.036401402205228806, "train_loss_llm": 0.0, "grad_norm": 0.04025665670633316, "global_step": 18438, "epoch": 512, "lr": 0.001} +{"train_loss": 0.036168839782476425, "train_loss_bc": 0.036168839782476425, "train_loss_llm": 0.0, "grad_norm": 0.058066047728061676, "global_step": 18439, "epoch": 512, "lr": 0.001} +{"train_loss": 0.036748744547367096, "train_loss_bc": 0.036748744547367096, "train_loss_llm": 0.0, "grad_norm": 0.07601374387741089, "global_step": 18440, "epoch": 512, "lr": 0.001} +{"train_loss": 0.034049127250909805, "train_loss_bc": 0.034049127250909805, "train_loss_llm": 0.0, "grad_norm": 0.026711277663707733, "global_step": 18441, "epoch": 512, "lr": 0.001} +{"train_loss": 0.026933319866657257, "train_loss_bc": 0.026933319866657257, "train_loss_llm": 0.0, "grad_norm": 0.04153043404221535, "global_step": 18442, "epoch": 512, "lr": 0.001} +{"train_loss": 0.04212436452507973, "train_loss_bc": 0.04212436452507973, "train_loss_llm": 0.0, "grad_norm": 0.05582195147871971, "global_step": 18443, "epoch": 512, "lr": 0.001} +{"train_loss": 0.026139959692955017, "train_loss_bc": 0.026139959692955017, "train_loss_llm": 0.0, "grad_norm": 0.06969975680112839, "global_step": 18444, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03278134763240814, "train_loss_bc": 0.03278134763240814, "train_loss_llm": 0.0, "grad_norm": 0.024168381467461586, "global_step": 18445, "epoch": 512, "lr": 0.001} +{"train_loss": 0.028406430035829544, "train_loss_bc": 0.028406430035829544, "train_loss_llm": 0.0, "grad_norm": 0.04105686396360397, "global_step": 18446, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03430454432964325, "train_loss_bc": 0.03430454432964325, "train_loss_llm": 0.0, "grad_norm": 0.05880363658070564, "global_step": 18447, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03164668008685112, "train_loss_bc": 0.03164668008685112, "train_loss_llm": 0.0, "grad_norm": 0.08051898330450058, "global_step": 18448, "epoch": 512, "lr": 0.001} +{"train_loss": 0.0359354242682457, "train_loss_bc": 0.0359354242682457, "train_loss_llm": 0.0, "grad_norm": 0.03708893433213234, "global_step": 18449, "epoch": 512, "lr": 0.001} +{"train_loss": 0.037681516259908676, "train_loss_bc": 0.037681516259908676, "train_loss_llm": 0.0, "grad_norm": 0.05152903497219086, "global_step": 18450, "epoch": 512, "lr": 0.001} +{"train_loss": 0.031084390357136726, "train_loss_bc": 0.03108365833759308, "train_loss_llm": 7.31348991394043e-05, "grad_norm": 0.06468499451875687, "global_step": 18451, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03669131174683571, "train_loss_bc": 0.03668396174907684, "train_loss_llm": 0.0007349848747253418, "grad_norm": 0.0851338803768158, "global_step": 18452, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03592558205127716, "train_loss_bc": 0.03592558205127716, "train_loss_llm": 0.0, "grad_norm": 0.031029492616653442, "global_step": 18453, "epoch": 512, "lr": 0.001} +{"train_loss": 0.036100443452596664, "train_loss_bc": 0.036100443452596664, "train_loss_llm": 0.0, "grad_norm": 0.05713126063346863, "global_step": 18454, "epoch": 512, "lr": 0.001} +{"train_loss": 0.032244227826595306, "train_loss_bc": 0.032244227826595306, "train_loss_llm": 0.0, "grad_norm": 0.07992620766162872, "global_step": 18455, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03251899033784866, "train_loss_bc": 0.032513078302145004, "train_loss_llm": 0.0005910694599151611, "grad_norm": 0.10565824061632156, "global_step": 18456, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03991920128464699, "train_loss_bc": 0.03991654887795448, "train_loss_llm": 0.000265270471572876, "grad_norm": 0.03337639570236206, "global_step": 18457, "epoch": 512, "lr": 0.001} +{"train_loss": 0.033678509294986725, "train_loss_bc": 0.033678509294986725, "train_loss_llm": 0.0, "grad_norm": 0.054629161953926086, "global_step": 18458, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03950326517224312, "train_loss_bc": 0.03950326517224312, "train_loss_llm": 0.0, "grad_norm": 0.07962878048419952, "global_step": 18459, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03743252903223038, "train_loss_bc": 0.03742561489343643, "train_loss_llm": 0.0006913244724273682, "grad_norm": 0.10145904868841171, "global_step": 18460, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03443131595849991, "train_loss_bc": 0.03443131595849991, "train_loss_llm": 0.0, "grad_norm": 0.023601308465003967, "global_step": 18461, "epoch": 512, "lr": 0.001} +{"train_loss": 0.026495007798075676, "train_loss_bc": 0.02649468369781971, "train_loss_llm": 3.2454729080200195e-05, "grad_norm": 0.030346062034368515, "global_step": 18462, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03317086026072502, "train_loss_bc": 0.03317086026072502, "train_loss_llm": 0.0, "grad_norm": 0.04001585766673088, "global_step": 18463, "epoch": 512, "lr": 0.001} +{"train_loss": 0.033393871039152145, "train_loss_bc": 0.033390264958143234, "train_loss_llm": 0.0003605186939239502, "grad_norm": 0.05851496383547783, "global_step": 18464, "epoch": 512, "lr": 0.001} +{"train_loss": 0.03709683567285538, "train_loss_bc": 0.03709683567285538, "train_loss_llm": 0.0, "grad_norm": 0.030102701857686043, "global_step": 18465, "epoch": 512, "lr": 0.001} +{"train_loss": 0.024973146617412567, "train_loss_bc": 0.024973146617412567, "train_loss_llm": 0.0, "grad_norm": 0.04097012057900429, "global_step": 18466, "epoch": 512, "lr": 0.001} +{"train_loss": 0.034139970731404096, "train_loss_bc": 0.035371169447898865, "train_loss_llm": 0.0, "grad_norm": 0.06218786910176277, "global_step": 18467, "epoch": 512, "lr": 0.001, "val_loss": 0.07534857839345932} +{"train_loss": 0.0298027154058218, "train_loss_bc": 0.0298027154058218, "train_loss_llm": 0.0, "grad_norm": 0.07853223383426666, "global_step": 18468, "epoch": 513, "lr": 0.001} +{"train_loss": 0.031798236072063446, "train_loss_bc": 0.031798236072063446, "train_loss_llm": 0.0, "grad_norm": 0.0321904718875885, "global_step": 18469, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03306915611028671, "train_loss_bc": 0.03306915611028671, "train_loss_llm": 0.0, "grad_norm": 0.04076097533106804, "global_step": 18470, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03416036069393158, "train_loss_bc": 0.03415331244468689, "train_loss_llm": 0.0007047653198242188, "grad_norm": 0.05824366956949234, "global_step": 18471, "epoch": 513, "lr": 0.001} +{"train_loss": 0.029917072504758835, "train_loss_bc": 0.02991531603038311, "train_loss_llm": 0.00017559528350830078, "grad_norm": 0.0739978775382042, "global_step": 18472, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03980617970228195, "train_loss_bc": 0.03980522230267525, "train_loss_llm": 9.572505950927734e-05, "grad_norm": 0.037485286593437195, "global_step": 18473, "epoch": 513, "lr": 0.001} +{"train_loss": 0.04089394956827164, "train_loss_bc": 0.04087967425584793, "train_loss_llm": 0.0014273524284362793, "grad_norm": 0.061900656670331955, "global_step": 18474, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03278113529086113, "train_loss_bc": 0.03278113529086113, "train_loss_llm": 0.0, "grad_norm": 0.08702223002910614, "global_step": 18475, "epoch": 513, "lr": 0.001} +{"train_loss": 0.04218047857284546, "train_loss_bc": 0.04218047857284546, "train_loss_llm": 0.0, "grad_norm": 0.10853256285190582, "global_step": 18476, "epoch": 513, "lr": 0.001} +{"train_loss": 0.039195939898490906, "train_loss_bc": 0.039192814379930496, "train_loss_llm": 0.00031241774559020996, "grad_norm": 0.04297392815351486, "global_step": 18477, "epoch": 513, "lr": 0.001} +{"train_loss": 0.036123622208833694, "train_loss_bc": 0.036123622208833694, "train_loss_llm": 0.0, "grad_norm": 0.07262171059846878, "global_step": 18478, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03933008015155792, "train_loss_bc": 0.03933008015155792, "train_loss_llm": 0.0, "grad_norm": 0.08493102341890335, "global_step": 18479, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03990718349814415, "train_loss_bc": 0.03990718349814415, "train_loss_llm": 0.0, "grad_norm": 0.11230973899364471, "global_step": 18480, "epoch": 513, "lr": 0.001} +{"train_loss": 0.037217818200588226, "train_loss_bc": 0.037217818200588226, "train_loss_llm": 0.0, "grad_norm": 0.03149048611521721, "global_step": 18481, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03715593367815018, "train_loss_bc": 0.03715593367815018, "train_loss_llm": 0.0, "grad_norm": 0.06880143284797668, "global_step": 18482, "epoch": 513, "lr": 0.001} +{"train_loss": 0.028613552451133728, "train_loss_bc": 0.028613552451133728, "train_loss_llm": 0.0, "grad_norm": 0.09091611206531525, "global_step": 18483, "epoch": 513, "lr": 0.001} +{"train_loss": 0.028626739978790283, "train_loss_bc": 0.028617896139621735, "train_loss_llm": 0.0008844435214996338, "grad_norm": 0.11304176598787308, "global_step": 18484, "epoch": 513, "lr": 0.001} +{"train_loss": 0.031764961779117584, "train_loss_bc": 0.031759582459926605, "train_loss_llm": 0.000537872314453125, "grad_norm": 0.027185456827282906, "global_step": 18485, "epoch": 513, "lr": 0.001} +{"train_loss": 0.0282317902892828, "train_loss_bc": 0.0282317902892828, "train_loss_llm": 0.0, "grad_norm": 0.042063891887664795, "global_step": 18486, "epoch": 513, "lr": 0.001} +{"train_loss": 0.035101283341646194, "train_loss_bc": 0.035101283341646194, "train_loss_llm": 0.0, "grad_norm": 0.053220346570014954, "global_step": 18487, "epoch": 513, "lr": 0.001} +{"train_loss": 0.0340418666601181, "train_loss_bc": 0.0340418666601181, "train_loss_llm": 0.0, "grad_norm": 0.06773484498262405, "global_step": 18488, "epoch": 513, "lr": 0.001} +{"train_loss": 0.039226774126291275, "train_loss_bc": 0.039226774126291275, "train_loss_llm": 0.0, "grad_norm": 0.03836934641003609, "global_step": 18489, "epoch": 513, "lr": 0.001} +{"train_loss": 0.0268721804022789, "train_loss_bc": 0.0268721804022789, "train_loss_llm": 0.0, "grad_norm": 0.04993151128292084, "global_step": 18490, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03533879294991493, "train_loss_bc": 0.03533879294991493, "train_loss_llm": 0.0, "grad_norm": 0.07062686234712601, "global_step": 18491, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03673822060227394, "train_loss_bc": 0.03673822060227394, "train_loss_llm": 0.0, "grad_norm": 0.09718748182058334, "global_step": 18492, "epoch": 513, "lr": 0.001} +{"train_loss": 0.0331135094165802, "train_loss_bc": 0.0331135094165802, "train_loss_llm": 0.0, "grad_norm": 0.037975139915943146, "global_step": 18493, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03245479241013527, "train_loss_bc": 0.03245479241013527, "train_loss_llm": 0.0, "grad_norm": 0.063873790204525, "global_step": 18494, "epoch": 513, "lr": 0.001} +{"train_loss": 0.030924007296562195, "train_loss_bc": 0.030924007296562195, "train_loss_llm": 0.0, "grad_norm": 0.08731182664632797, "global_step": 18495, "epoch": 513, "lr": 0.001} +{"train_loss": 0.034103889018297195, "train_loss_bc": 0.03410330042243004, "train_loss_llm": 5.8770179748535156e-05, "grad_norm": 0.09904663264751434, "global_step": 18496, "epoch": 513, "lr": 0.001} +{"train_loss": 0.02957249991595745, "train_loss_bc": 0.02957249991595745, "train_loss_llm": 0.0, "grad_norm": 0.03558594733476639, "global_step": 18497, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03711671382188797, "train_loss_bc": 0.03711671382188797, "train_loss_llm": 0.0, "grad_norm": 0.06695813685655594, "global_step": 18498, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03518560528755188, "train_loss_bc": 0.03518560528755188, "train_loss_llm": 0.0, "grad_norm": 0.08507563173770905, "global_step": 18499, "epoch": 513, "lr": 0.001} +{"train_loss": 0.037673093378543854, "train_loss_bc": 0.03766290843486786, "train_loss_llm": 0.001018613576889038, "grad_norm": 0.10610566288232803, "global_step": 18500, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03788020461797714, "train_loss_bc": 0.03787367790937424, "train_loss_llm": 0.0006528496742248535, "grad_norm": 0.042547788470983505, "global_step": 18501, "epoch": 513, "lr": 0.001} +{"train_loss": 0.0349581204354763, "train_loss_bc": 0.0349581204354763, "train_loss_llm": 0.0, "grad_norm": 0.06219109147787094, "global_step": 18502, "epoch": 513, "lr": 0.001} +{"train_loss": 0.03427386371832755, "train_loss_bc": 0.02298063412308693, "train_loss_llm": 0.0, "grad_norm": 0.091615229845047, "global_step": 18503, "epoch": 513, "lr": 0.001, "val_loss": 0.07605836540460587} +{"train_loss": 0.039668601006269455, "train_loss_bc": 0.039668601006269455, "train_loss_llm": 0.0, "grad_norm": 0.12733174860477448, "global_step": 18504, "epoch": 514, "lr": 0.001} +{"train_loss": 0.04048033058643341, "train_loss_bc": 0.04048033058643341, "train_loss_llm": 0.0, "grad_norm": 0.03403298184275627, "global_step": 18505, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03422629088163376, "train_loss_bc": 0.03422629088163376, "train_loss_llm": 0.0, "grad_norm": 0.05318620800971985, "global_step": 18506, "epoch": 514, "lr": 0.001} +{"train_loss": 0.030327359214425087, "train_loss_bc": 0.03032459132373333, "train_loss_llm": 0.00027683377265930176, "grad_norm": 0.06879116594791412, "global_step": 18507, "epoch": 514, "lr": 0.001} +{"train_loss": 0.02893681265413761, "train_loss_bc": 0.02893681265413761, "train_loss_llm": 0.0, "grad_norm": 0.08187811821699142, "global_step": 18508, "epoch": 514, "lr": 0.001} +{"train_loss": 0.033505864441394806, "train_loss_bc": 0.033505864441394806, "train_loss_llm": 0.0, "grad_norm": 0.02830328792333603, "global_step": 18509, "epoch": 514, "lr": 0.001} +{"train_loss": 0.030987609177827835, "train_loss_bc": 0.03096259944140911, "train_loss_llm": 0.0025010406970977783, "grad_norm": 0.040768638253211975, "global_step": 18510, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03186483308672905, "train_loss_bc": 0.03186483308672905, "train_loss_llm": 0.0, "grad_norm": 0.050687287002801895, "global_step": 18511, "epoch": 514, "lr": 0.001} +{"train_loss": 0.034532006829977036, "train_loss_bc": 0.034532006829977036, "train_loss_llm": 0.0, "grad_norm": 0.06380844116210938, "global_step": 18512, "epoch": 514, "lr": 0.001} +{"train_loss": 0.036745842546224594, "train_loss_bc": 0.036745842546224594, "train_loss_llm": 0.0, "grad_norm": 0.034715067595243454, "global_step": 18513, "epoch": 514, "lr": 0.001} +{"train_loss": 0.031550001353025436, "train_loss_bc": 0.03151657432317734, "train_loss_llm": 0.003342658281326294, "grad_norm": 0.04973084479570389, "global_step": 18514, "epoch": 514, "lr": 0.001} +{"train_loss": 0.036280084401369095, "train_loss_bc": 0.036275871098041534, "train_loss_llm": 0.00042119622230529785, "grad_norm": 0.07158264517784119, "global_step": 18515, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03333118185400963, "train_loss_bc": 0.0333198644220829, "train_loss_llm": 0.0011317431926727295, "grad_norm": 0.0949016660451889, "global_step": 18516, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03388090431690216, "train_loss_bc": 0.03386998921632767, "train_loss_llm": 0.0010916590690612793, "grad_norm": 0.036238934844732285, "global_step": 18517, "epoch": 514, "lr": 0.001} +{"train_loss": 0.036437466740608215, "train_loss_bc": 0.036437466740608215, "train_loss_llm": 0.0, "grad_norm": 0.05858822539448738, "global_step": 18518, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03764265403151512, "train_loss_bc": 0.03764265403151512, "train_loss_llm": 0.0, "grad_norm": 0.07757445424795151, "global_step": 18519, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03417511284351349, "train_loss_bc": 0.03417511284351349, "train_loss_llm": 0.0, "grad_norm": 0.10037684440612793, "global_step": 18520, "epoch": 514, "lr": 0.001} +{"train_loss": 0.032923389226198196, "train_loss_bc": 0.032923389226198196, "train_loss_llm": 0.0, "grad_norm": 0.03514905646443367, "global_step": 18521, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03299974650144577, "train_loss_bc": 0.03299974650144577, "train_loss_llm": 0.0, "grad_norm": 0.06496590375900269, "global_step": 18522, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03548572212457657, "train_loss_bc": 0.0354752317070961, "train_loss_llm": 0.001049041748046875, "grad_norm": 0.09490112960338593, "global_step": 18523, "epoch": 514, "lr": 0.001} +{"train_loss": 0.026820484548807144, "train_loss_bc": 0.026820484548807144, "train_loss_llm": 0.0, "grad_norm": 0.10400390625, "global_step": 18524, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03461586683988571, "train_loss_bc": 0.03461586683988571, "train_loss_llm": 0.0, "grad_norm": 0.027158230543136597, "global_step": 18525, "epoch": 514, "lr": 0.001} +{"train_loss": 0.031015753746032715, "train_loss_bc": 0.031006719917058945, "train_loss_llm": 0.0009033083915710449, "grad_norm": 0.04343392327427864, "global_step": 18526, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03617973253130913, "train_loss_bc": 0.036179251968860626, "train_loss_llm": 4.813075065612793e-05, "grad_norm": 0.061852406710386276, "global_step": 18527, "epoch": 514, "lr": 0.001} +{"train_loss": 0.042319782078266144, "train_loss_bc": 0.04231584072113037, "train_loss_llm": 0.00039395689964294434, "grad_norm": 0.08237214386463165, "global_step": 18528, "epoch": 514, "lr": 0.001} +{"train_loss": 0.036467164754867554, "train_loss_bc": 0.036467164754867554, "train_loss_llm": 0.0, "grad_norm": 0.026231808587908745, "global_step": 18529, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03553173691034317, "train_loss_bc": 0.03553173691034317, "train_loss_llm": 0.0, "grad_norm": 0.04211635887622833, "global_step": 18530, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03579714894294739, "train_loss_bc": 0.03579084202647209, "train_loss_llm": 0.0006306469440460205, "grad_norm": 0.059480708092451096, "global_step": 18531, "epoch": 514, "lr": 0.001} +{"train_loss": 0.033953677862882614, "train_loss_bc": 0.03393528610467911, "train_loss_llm": 0.0018392205238342285, "grad_norm": 0.06910291314125061, "global_step": 18532, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03415084630250931, "train_loss_bc": 0.03415084630250931, "train_loss_llm": 0.0, "grad_norm": 0.028163887560367584, "global_step": 18533, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03330143541097641, "train_loss_bc": 0.03330143541097641, "train_loss_llm": 0.0, "grad_norm": 0.04096430167555809, "global_step": 18534, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03213139995932579, "train_loss_bc": 0.03213139995932579, "train_loss_llm": 0.0, "grad_norm": 0.06637243181467056, "global_step": 18535, "epoch": 514, "lr": 0.001} +{"train_loss": 0.029259782284498215, "train_loss_bc": 0.02925977110862732, "train_loss_llm": 1.1026859283447266e-06, "grad_norm": 0.08955002576112747, "global_step": 18536, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03911258280277252, "train_loss_bc": 0.039102546870708466, "train_loss_llm": 0.0010036826133728027, "grad_norm": 0.0441746711730957, "global_step": 18537, "epoch": 514, "lr": 0.001} +{"train_loss": 0.029521020129323006, "train_loss_bc": 0.029521020129323006, "train_loss_llm": 0.0, "grad_norm": 0.06894545257091522, "global_step": 18538, "epoch": 514, "lr": 0.001} +{"train_loss": 0.03408871472088827, "train_loss_bc": 0.031024863943457603, "train_loss_llm": 0.0008637905120849609, "grad_norm": 0.07273118942975998, "global_step": 18539, "epoch": 514, "lr": 0.001, "val_loss": 0.0754300057888031} +{"train_loss": 0.036460306495428085, "train_loss_bc": 0.036460306495428085, "train_loss_llm": 0.0, "grad_norm": 0.08205319195985794, "global_step": 18540, "epoch": 515, "lr": 0.001} +{"train_loss": 0.04043753445148468, "train_loss_bc": 0.04043753445148468, "train_loss_llm": 0.0, "grad_norm": 0.027641676366329193, "global_step": 18541, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03103015199303627, "train_loss_bc": 0.03103015199303627, "train_loss_llm": 0.0, "grad_norm": 0.061260756105184555, "global_step": 18542, "epoch": 515, "lr": 0.001} +{"train_loss": 0.035866979509592056, "train_loss_bc": 0.035866979509592056, "train_loss_llm": 0.0, "grad_norm": 0.089827761054039, "global_step": 18543, "epoch": 515, "lr": 0.001} +{"train_loss": 0.036924418061971664, "train_loss_bc": 0.03690551593899727, "train_loss_llm": 0.001890331506729126, "grad_norm": 0.12259899824857712, "global_step": 18544, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03368218243122101, "train_loss_bc": 0.03368218243122101, "train_loss_llm": 0.0, "grad_norm": 0.030402418226003647, "global_step": 18545, "epoch": 515, "lr": 0.001} +{"train_loss": 0.036197029054164886, "train_loss_bc": 0.036197029054164886, "train_loss_llm": 0.0, "grad_norm": 0.05989484116435051, "global_step": 18546, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03889834135770798, "train_loss_bc": 0.03889834135770798, "train_loss_llm": 0.0, "grad_norm": 0.08731856942176819, "global_step": 18547, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03742354363203049, "train_loss_bc": 0.03742354363203049, "train_loss_llm": 0.0, "grad_norm": 0.10893862694501877, "global_step": 18548, "epoch": 515, "lr": 0.001} +{"train_loss": 0.030783405527472496, "train_loss_bc": 0.030782906338572502, "train_loss_llm": 4.991888999938965e-05, "grad_norm": 0.02318195253610611, "global_step": 18549, "epoch": 515, "lr": 0.001} +{"train_loss": 0.034463830292224884, "train_loss_bc": 0.034463830292224884, "train_loss_llm": 0.0, "grad_norm": 0.03239439055323601, "global_step": 18550, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03307410702109337, "train_loss_bc": 0.03307410702109337, "train_loss_llm": 0.0, "grad_norm": 0.04435652494430542, "global_step": 18551, "epoch": 515, "lr": 0.001} +{"train_loss": 0.033776070922613144, "train_loss_bc": 0.03376564383506775, "train_loss_llm": 0.0010428130626678467, "grad_norm": 0.05868978798389435, "global_step": 18552, "epoch": 515, "lr": 0.001} +{"train_loss": 0.036099471151828766, "train_loss_bc": 0.036099471151828766, "train_loss_llm": 0.0, "grad_norm": 0.029442528262734413, "global_step": 18553, "epoch": 515, "lr": 0.001} +{"train_loss": 0.037948332726955414, "train_loss_bc": 0.037948332726955414, "train_loss_llm": 0.0, "grad_norm": 0.03978167474269867, "global_step": 18554, "epoch": 515, "lr": 0.001} +{"train_loss": 0.023273784667253494, "train_loss_bc": 0.023270051926374435, "train_loss_llm": 0.0003733634948730469, "grad_norm": 0.04990445822477341, "global_step": 18555, "epoch": 515, "lr": 0.001} +{"train_loss": 0.026113657280802727, "train_loss_bc": 0.026113657280802727, "train_loss_llm": 0.0, "grad_norm": 0.06327319890260696, "global_step": 18556, "epoch": 515, "lr": 0.001} +{"train_loss": 0.029355289414525032, "train_loss_bc": 0.029355289414525032, "train_loss_llm": 0.0, "grad_norm": 0.03282454237341881, "global_step": 18557, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03128395229578018, "train_loss_bc": 0.031249528750777245, "train_loss_llm": 0.0034421682357788086, "grad_norm": 0.05579639971256256, "global_step": 18558, "epoch": 515, "lr": 0.001} +{"train_loss": 0.037207119166851044, "train_loss_bc": 0.037207119166851044, "train_loss_llm": 0.0, "grad_norm": 0.079667828977108, "global_step": 18559, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03225291892886162, "train_loss_bc": 0.03224242478609085, "train_loss_llm": 0.0010494589805603027, "grad_norm": 0.1060730591416359, "global_step": 18560, "epoch": 515, "lr": 0.001} +{"train_loss": 0.039383046329021454, "train_loss_bc": 0.03935264050960541, "train_loss_llm": 0.0030404627323150635, "grad_norm": 0.03890093415975571, "global_step": 18561, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03362247347831726, "train_loss_bc": 0.03362247347831726, "train_loss_llm": 0.0, "grad_norm": 0.07002615183591843, "global_step": 18562, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03289283066987991, "train_loss_bc": 0.03289283066987991, "train_loss_llm": 0.0, "grad_norm": 0.1027621403336525, "global_step": 18563, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03360636904835701, "train_loss_bc": 0.03360636904835701, "train_loss_llm": 0.0, "grad_norm": 0.1342480629682541, "global_step": 18564, "epoch": 515, "lr": 0.001} +{"train_loss": 0.035275936126708984, "train_loss_bc": 0.035275936126708984, "train_loss_llm": 0.0, "grad_norm": 0.02572079934179783, "global_step": 18565, "epoch": 515, "lr": 0.001} +{"train_loss": 0.031185349449515343, "train_loss_bc": 0.031175920739769936, "train_loss_llm": 0.0009428560733795166, "grad_norm": 0.03839879110455513, "global_step": 18566, "epoch": 515, "lr": 0.001} +{"train_loss": 0.036675289273262024, "train_loss_bc": 0.036675289273262024, "train_loss_llm": 0.0, "grad_norm": 0.052571795880794525, "global_step": 18567, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03338046371936798, "train_loss_bc": 0.03338046371936798, "train_loss_llm": 0.0, "grad_norm": 0.06859192997217178, "global_step": 18568, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03717418760061264, "train_loss_bc": 0.03717418760061264, "train_loss_llm": 0.0, "grad_norm": 0.02736716903746128, "global_step": 18569, "epoch": 515, "lr": 0.001} +{"train_loss": 0.02964073419570923, "train_loss_bc": 0.0296342633664608, "train_loss_llm": 0.0006470978260040283, "grad_norm": 0.04180888459086418, "global_step": 18570, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03396247699856758, "train_loss_bc": 0.03396247699856758, "train_loss_llm": 0.0, "grad_norm": 0.055369406938552856, "global_step": 18571, "epoch": 515, "lr": 0.001} +{"train_loss": 0.0386974923312664, "train_loss_bc": 0.0386974923312664, "train_loss_llm": 0.0, "grad_norm": 0.06984034925699234, "global_step": 18572, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03667261078953743, "train_loss_bc": 0.03667261078953743, "train_loss_llm": 0.0, "grad_norm": 0.02769426815211773, "global_step": 18573, "epoch": 515, "lr": 0.001} +{"train_loss": 0.03524722903966904, "train_loss_bc": 0.03524722903966904, "train_loss_llm": 0.0, "grad_norm": 0.04337616637349129, "global_step": 18574, "epoch": 515, "lr": 0.001} +{"train_loss": 0.034225359145137996, "train_loss_bc": 0.03214401379227638, "train_loss_llm": 0.0, "grad_norm": 0.05776868388056755, "global_step": 18575, "epoch": 515, "lr": 0.001, "train/cumulative_reward": 68.52628368911272, "train/mean_score": 9.895831627643853, "train/success_rate": 0.2, "test/cumulative_reward": 32.24151666928411, "test/mean_score": 8.781778531695805, "test/success_rate": 0.22, "val_loss": 0.07611565291881561, "train_action_mse_error": 0.03674401715397835} +{"train_loss": 0.03487803414463997, "train_loss_bc": 0.03487101569771767, "train_loss_llm": 0.0007018744945526123, "grad_norm": 0.07127945870161057, "global_step": 18576, "epoch": 516, "lr": 0.001} +{"train_loss": 0.026052555069327354, "train_loss_bc": 0.026052555069327354, "train_loss_llm": 0.0, "grad_norm": 0.025036441162228584, "global_step": 18577, "epoch": 516, "lr": 0.001} +{"train_loss": 0.02907002530992031, "train_loss_bc": 0.02907002530992031, "train_loss_llm": 0.0, "grad_norm": 0.04093910753726959, "global_step": 18578, "epoch": 516, "lr": 0.001} +{"train_loss": 0.030748916789889336, "train_loss_bc": 0.030748916789889336, "train_loss_llm": 0.0, "grad_norm": 0.05704447999596596, "global_step": 18579, "epoch": 516, "lr": 0.001} +{"train_loss": 0.037520881742239, "train_loss_bc": 0.037520881742239, "train_loss_llm": 0.0, "grad_norm": 0.07717571407556534, "global_step": 18580, "epoch": 516, "lr": 0.001} +{"train_loss": 0.037008002400398254, "train_loss_bc": 0.037008002400398254, "train_loss_llm": 0.0, "grad_norm": 0.026901736855506897, "global_step": 18581, "epoch": 516, "lr": 0.001} +{"train_loss": 0.035558052361011505, "train_loss_bc": 0.035558052361011505, "train_loss_llm": 0.0, "grad_norm": 0.0518437996506691, "global_step": 18582, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03056327998638153, "train_loss_bc": 0.03056327998638153, "train_loss_llm": 0.0, "grad_norm": 0.06894435733556747, "global_step": 18583, "epoch": 516, "lr": 0.001} +{"train_loss": 0.039717987179756165, "train_loss_bc": 0.039717987179756165, "train_loss_llm": 0.0, "grad_norm": 0.10197283327579498, "global_step": 18584, "epoch": 516, "lr": 0.001} +{"train_loss": 0.039021603763103485, "train_loss_bc": 0.039021603763103485, "train_loss_llm": 0.0, "grad_norm": 0.036125872284173965, "global_step": 18585, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03447100892663002, "train_loss_bc": 0.03447100892663002, "train_loss_llm": 0.0, "grad_norm": 0.05767584592103958, "global_step": 18586, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03506290912628174, "train_loss_bc": 0.03506290912628174, "train_loss_llm": 0.0, "grad_norm": 0.08066220581531525, "global_step": 18587, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03376469016075134, "train_loss_bc": 0.03376469016075134, "train_loss_llm": 0.0, "grad_norm": 0.0985887199640274, "global_step": 18588, "epoch": 516, "lr": 0.001} +{"train_loss": 0.030533790588378906, "train_loss_bc": 0.030533790588378906, "train_loss_llm": 0.0, "grad_norm": 0.030370812863111496, "global_step": 18589, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03488365560770035, "train_loss_bc": 0.03488365560770035, "train_loss_llm": 0.0, "grad_norm": 0.04597724974155426, "global_step": 18590, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03239666298031807, "train_loss_bc": 0.03239666298031807, "train_loss_llm": 0.0, "grad_norm": 0.06333662569522858, "global_step": 18591, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03391910716891289, "train_loss_bc": 0.033900924026966095, "train_loss_llm": 0.0018183887004852295, "grad_norm": 0.08467941731214523, "global_step": 18592, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03134628385305405, "train_loss_bc": 0.03131874278187752, "train_loss_llm": 0.0027539730072021484, "grad_norm": 0.02911393903195858, "global_step": 18593, "epoch": 516, "lr": 0.001} +{"train_loss": 0.0341971181333065, "train_loss_bc": 0.0341971181333065, "train_loss_llm": 0.0, "grad_norm": 0.039289139211177826, "global_step": 18594, "epoch": 516, "lr": 0.001} +{"train_loss": 0.0345759280025959, "train_loss_bc": 0.0345759280025959, "train_loss_llm": 0.0, "grad_norm": 0.06355557590723038, "global_step": 18595, "epoch": 516, "lr": 0.001} +{"train_loss": 0.024013880640268326, "train_loss_bc": 0.023994149640202522, "train_loss_llm": 0.001973181962966919, "grad_norm": 0.076667420566082, "global_step": 18596, "epoch": 516, "lr": 0.001} +{"train_loss": 0.043095193803310394, "train_loss_bc": 0.043095193803310394, "train_loss_llm": 0.0, "grad_norm": 0.03477388247847557, "global_step": 18597, "epoch": 516, "lr": 0.001} +{"train_loss": 0.038017209619283676, "train_loss_bc": 0.038017209619283676, "train_loss_llm": 0.0, "grad_norm": 0.06750660389661789, "global_step": 18598, "epoch": 516, "lr": 0.001} +{"train_loss": 0.02855510078370571, "train_loss_bc": 0.02855510078370571, "train_loss_llm": 0.0, "grad_norm": 0.07964496314525604, "global_step": 18599, "epoch": 516, "lr": 0.001} +{"train_loss": 0.04151913523674011, "train_loss_bc": 0.04151913523674011, "train_loss_llm": 0.0, "grad_norm": 0.09864253550767899, "global_step": 18600, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03517140448093414, "train_loss_bc": 0.03517140448093414, "train_loss_llm": 0.0, "grad_norm": 0.02628907933831215, "global_step": 18601, "epoch": 516, "lr": 0.001} +{"train_loss": 0.030735809355974197, "train_loss_bc": 0.03072217106819153, "train_loss_llm": 0.0013637542724609375, "grad_norm": 0.04016212746500969, "global_step": 18602, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03790806233882904, "train_loss_bc": 0.03790806233882904, "train_loss_llm": 0.0, "grad_norm": 0.05577549710869789, "global_step": 18603, "epoch": 516, "lr": 0.001} +{"train_loss": 0.038554441183805466, "train_loss_bc": 0.038554441183805466, "train_loss_llm": 0.0, "grad_norm": 0.067679263651371, "global_step": 18604, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03167632222175598, "train_loss_bc": 0.03167632222175598, "train_loss_llm": 0.0, "grad_norm": 0.026300406083464622, "global_step": 18605, "epoch": 516, "lr": 0.001} +{"train_loss": 0.030695006251335144, "train_loss_bc": 0.03069317899644375, "train_loss_llm": 0.00018271803855895996, "grad_norm": 0.04092865437269211, "global_step": 18606, "epoch": 516, "lr": 0.001} +{"train_loss": 0.027366630733013153, "train_loss_bc": 0.027366630733013153, "train_loss_llm": 0.0, "grad_norm": 0.05990997701883316, "global_step": 18607, "epoch": 516, "lr": 0.001} +{"train_loss": 0.0345284566283226, "train_loss_bc": 0.0345284566283226, "train_loss_llm": 0.0, "grad_norm": 0.07437711954116821, "global_step": 18608, "epoch": 516, "lr": 0.001} +{"train_loss": 0.03816363215446472, "train_loss_bc": 0.03816363215446472, "train_loss_llm": 0.0, "grad_norm": 0.03669113293290138, "global_step": 18609, "epoch": 516, "lr": 0.001} +{"train_loss": 0.036113087087869644, "train_loss_bc": 0.036113087087869644, "train_loss_llm": 0.0, "grad_norm": 0.0570807121694088, "global_step": 18610, "epoch": 516, "lr": 0.001} +{"train_loss": 0.0340524377922217, "train_loss_bc": 0.0344838947057724, "train_loss_llm": 0.0, "grad_norm": 0.08417855948209763, "global_step": 18611, "epoch": 516, "lr": 0.001, "val_loss": 0.07699733972549438} +{"train_loss": 0.03504784032702446, "train_loss_bc": 0.03504784032702446, "train_loss_llm": 0.0, "grad_norm": 0.09852174669504166, "global_step": 18612, "epoch": 517, "lr": 0.001} +{"train_loss": 0.04112842306494713, "train_loss_bc": 0.04112411290407181, "train_loss_llm": 0.0004309117794036865, "grad_norm": 0.03252078965306282, "global_step": 18613, "epoch": 517, "lr": 0.001} +{"train_loss": 0.032182611525058746, "train_loss_bc": 0.032182611525058746, "train_loss_llm": 0.0, "grad_norm": 0.05023537948727608, "global_step": 18614, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03598552569746971, "train_loss_bc": 0.035976506769657135, "train_loss_llm": 0.0009019374847412109, "grad_norm": 0.06763795018196106, "global_step": 18615, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03152962028980255, "train_loss_bc": 0.03152962028980255, "train_loss_llm": 0.0, "grad_norm": 0.08517186343669891, "global_step": 18616, "epoch": 517, "lr": 0.001} +{"train_loss": 0.028704024851322174, "train_loss_bc": 0.028703469783067703, "train_loss_llm": 5.555152893066406e-05, "grad_norm": 0.026569515466690063, "global_step": 18617, "epoch": 517, "lr": 0.001} +{"train_loss": 0.030394531786441803, "train_loss_bc": 0.030394531786441803, "train_loss_llm": 0.0, "grad_norm": 0.041297122836112976, "global_step": 18618, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03629454970359802, "train_loss_bc": 0.03629454970359802, "train_loss_llm": 0.0, "grad_norm": 0.059823449701070786, "global_step": 18619, "epoch": 517, "lr": 0.001} +{"train_loss": 0.02724560722708702, "train_loss_bc": 0.027240026742219925, "train_loss_llm": 0.000558018684387207, "grad_norm": 0.06977006047964096, "global_step": 18620, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03714824467897415, "train_loss_bc": 0.037143073976039886, "train_loss_llm": 0.0005171298980712891, "grad_norm": 0.02751321531832218, "global_step": 18621, "epoch": 517, "lr": 0.001} +{"train_loss": 0.029298512265086174, "train_loss_bc": 0.029298512265086174, "train_loss_llm": 0.0, "grad_norm": 0.04384453967213631, "global_step": 18622, "epoch": 517, "lr": 0.001} +{"train_loss": 0.037130001932382584, "train_loss_bc": 0.037124235183000565, "train_loss_llm": 0.0005765855312347412, "grad_norm": 0.0569177009165287, "global_step": 18623, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03865249454975128, "train_loss_bc": 0.03865249454975128, "train_loss_llm": 0.0, "grad_norm": 0.0681072324514389, "global_step": 18624, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03515711799263954, "train_loss_bc": 0.03515711799263954, "train_loss_llm": 0.0, "grad_norm": 0.030341600999236107, "global_step": 18625, "epoch": 517, "lr": 0.001} +{"train_loss": 0.033621810376644135, "train_loss_bc": 0.033621810376644135, "train_loss_llm": 0.0, "grad_norm": 0.046506550163030624, "global_step": 18626, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03181305155158043, "train_loss_bc": 0.03181224688887596, "train_loss_llm": 8.052587509155273e-05, "grad_norm": 0.061912406235933304, "global_step": 18627, "epoch": 517, "lr": 0.001} +{"train_loss": 0.034547772258520126, "train_loss_bc": 0.034547772258520126, "train_loss_llm": 0.0, "grad_norm": 0.0784604474902153, "global_step": 18628, "epoch": 517, "lr": 0.001} +{"train_loss": 0.029410667717456818, "train_loss_bc": 0.029410667717456818, "train_loss_llm": 0.0, "grad_norm": 0.03493405506014824, "global_step": 18629, "epoch": 517, "lr": 0.001} +{"train_loss": 0.0287530105561018, "train_loss_bc": 0.0287511944770813, "train_loss_llm": 0.00018167495727539062, "grad_norm": 0.058664336800575256, "global_step": 18630, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03396458551287651, "train_loss_bc": 0.03396458551287651, "train_loss_llm": 0.0, "grad_norm": 0.08204984664916992, "global_step": 18631, "epoch": 517, "lr": 0.001} +{"train_loss": 0.027839917689561844, "train_loss_bc": 0.027839917689561844, "train_loss_llm": 0.0, "grad_norm": 0.10128478705883026, "global_step": 18632, "epoch": 517, "lr": 0.001} +{"train_loss": 0.04170664772391319, "train_loss_bc": 0.04170664772391319, "train_loss_llm": 0.0, "grad_norm": 0.031560786068439484, "global_step": 18633, "epoch": 517, "lr": 0.001} +{"train_loss": 0.033737894147634506, "train_loss_bc": 0.0337349958717823, "train_loss_llm": 0.0002898275852203369, "grad_norm": 0.057401105761528015, "global_step": 18634, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03784535825252533, "train_loss_bc": 0.037839002907276154, "train_loss_llm": 0.0006354451179504395, "grad_norm": 0.07771032303571701, "global_step": 18635, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03503043204545975, "train_loss_bc": 0.03503043204545975, "train_loss_llm": 0.0, "grad_norm": 0.0959559753537178, "global_step": 18636, "epoch": 517, "lr": 0.001} +{"train_loss": 0.04178421571850777, "train_loss_bc": 0.04178421571850777, "train_loss_llm": 0.0, "grad_norm": 0.02475898154079914, "global_step": 18637, "epoch": 517, "lr": 0.001} +{"train_loss": 0.035757504403591156, "train_loss_bc": 0.035757504403591156, "train_loss_llm": 0.0, "grad_norm": 0.04230010136961937, "global_step": 18638, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03574589639902115, "train_loss_bc": 0.03574589639902115, "train_loss_llm": 0.0, "grad_norm": 0.05723490193486214, "global_step": 18639, "epoch": 517, "lr": 0.001} +{"train_loss": 0.038044195622205734, "train_loss_bc": 0.0380207784473896, "train_loss_llm": 0.0023418962955474854, "grad_norm": 0.07074324786663055, "global_step": 18640, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03585987910628319, "train_loss_bc": 0.03585987910628319, "train_loss_llm": 0.0, "grad_norm": 0.030528170987963676, "global_step": 18641, "epoch": 517, "lr": 0.001} +{"train_loss": 0.02991335093975067, "train_loss_bc": 0.02991335093975067, "train_loss_llm": 0.0, "grad_norm": 0.045724499970674515, "global_step": 18642, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03657589852809906, "train_loss_bc": 0.03655777871608734, "train_loss_llm": 0.0018120408058166504, "grad_norm": 0.064543217420578, "global_step": 18643, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03385847061872482, "train_loss_bc": 0.033854711800813675, "train_loss_llm": 0.00037598609924316406, "grad_norm": 0.07356493175029755, "global_step": 18644, "epoch": 517, "lr": 0.001} +{"train_loss": 0.03185756877064705, "train_loss_bc": 0.03185756877064705, "train_loss_llm": 0.0, "grad_norm": 0.02445121295750141, "global_step": 18645, "epoch": 517, "lr": 0.001} +{"train_loss": 0.028718210756778717, "train_loss_bc": 0.028718210756778717, "train_loss_llm": 0.0, "grad_norm": 0.03508748114109039, "global_step": 18646, "epoch": 517, "lr": 0.001} +{"train_loss": 0.034037359783218965, "train_loss_bc": 0.0330377072095871, "train_loss_llm": 0.002180129289627075, "grad_norm": 0.05440746620297432, "global_step": 18647, "epoch": 517, "lr": 0.001, "val_loss": 0.07754303514957428} +{"train_loss": 0.037122029811143875, "train_loss_bc": 0.037122029811143875, "train_loss_llm": 0.0, "grad_norm": 0.06683926284313202, "global_step": 18648, "epoch": 518, "lr": 0.001} +{"train_loss": 0.034502122551202774, "train_loss_bc": 0.03446339815855026, "train_loss_llm": 0.0038725435733795166, "grad_norm": 0.02755168080329895, "global_step": 18649, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03526102751493454, "train_loss_bc": 0.03526102751493454, "train_loss_llm": 0.0, "grad_norm": 0.05325683206319809, "global_step": 18650, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03604556992650032, "train_loss_bc": 0.03604556992650032, "train_loss_llm": 0.0, "grad_norm": 0.07199638336896896, "global_step": 18651, "epoch": 518, "lr": 0.001} +{"train_loss": 0.033825572580099106, "train_loss_bc": 0.03382284194231033, "train_loss_llm": 0.00027292966842651367, "grad_norm": 0.09346222132444382, "global_step": 18652, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03395521640777588, "train_loss_bc": 0.03395521640777588, "train_loss_llm": 0.0, "grad_norm": 0.04241052642464638, "global_step": 18653, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03451187536120415, "train_loss_bc": 0.03451187536120415, "train_loss_llm": 0.0, "grad_norm": 0.06424514204263687, "global_step": 18654, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03693849593400955, "train_loss_bc": 0.03693849593400955, "train_loss_llm": 0.0, "grad_norm": 0.09185447543859482, "global_step": 18655, "epoch": 518, "lr": 0.001} +{"train_loss": 0.0322142168879509, "train_loss_bc": 0.0322142168879509, "train_loss_llm": 0.0, "grad_norm": 0.11282849311828613, "global_step": 18656, "epoch": 518, "lr": 0.001} +{"train_loss": 0.029324257746338844, "train_loss_bc": 0.029317641630768776, "train_loss_llm": 0.000661700963973999, "grad_norm": 0.025477182120084763, "global_step": 18657, "epoch": 518, "lr": 0.001} +{"train_loss": 0.035230088979005814, "train_loss_bc": 0.035230088979005814, "train_loss_llm": 0.0, "grad_norm": 0.05331505462527275, "global_step": 18658, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03411057963967323, "train_loss_bc": 0.03411057963967323, "train_loss_llm": 0.0, "grad_norm": 0.0667884424328804, "global_step": 18659, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03110266476869583, "train_loss_bc": 0.03110266476869583, "train_loss_llm": 0.0, "grad_norm": 0.08248294144868851, "global_step": 18660, "epoch": 518, "lr": 0.001} +{"train_loss": 0.035569097846746445, "train_loss_bc": 0.03555591404438019, "train_loss_llm": 0.0013182759284973145, "grad_norm": 0.025200776755809784, "global_step": 18661, "epoch": 518, "lr": 0.001} +{"train_loss": 0.034726209938526154, "train_loss_bc": 0.034726209938526154, "train_loss_llm": 0.0, "grad_norm": 0.04404020309448242, "global_step": 18662, "epoch": 518, "lr": 0.001} +{"train_loss": 0.04130259528756142, "train_loss_bc": 0.04129760339856148, "train_loss_llm": 0.0004992783069610596, "grad_norm": 0.054716795682907104, "global_step": 18663, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03136139363050461, "train_loss_bc": 0.03132707625627518, "train_loss_llm": 0.0034318864345550537, "grad_norm": 0.07307854294776917, "global_step": 18664, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03851448372006416, "train_loss_bc": 0.03851448372006416, "train_loss_llm": 0.0, "grad_norm": 0.028533935546875, "global_step": 18665, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03231703117489815, "train_loss_bc": 0.032308559864759445, "train_loss_llm": 0.0008469521999359131, "grad_norm": 0.0375346802175045, "global_step": 18666, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03712129965424538, "train_loss_bc": 0.03712129965424538, "train_loss_llm": 0.0, "grad_norm": 0.04759349301457405, "global_step": 18667, "epoch": 518, "lr": 0.001} +{"train_loss": 0.02923162840306759, "train_loss_bc": 0.02923162840306759, "train_loss_llm": 0.0, "grad_norm": 0.05529271811246872, "global_step": 18668, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03156612068414688, "train_loss_bc": 0.03156612068414688, "train_loss_llm": 0.0, "grad_norm": 0.030173903331160545, "global_step": 18669, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03257092833518982, "train_loss_bc": 0.03254695609211922, "train_loss_llm": 0.0023973584175109863, "grad_norm": 0.03936579078435898, "global_step": 18670, "epoch": 518, "lr": 0.001} +{"train_loss": 0.029663652181625366, "train_loss_bc": 0.029663652181625366, "train_loss_llm": 0.0, "grad_norm": 0.06263343244791031, "global_step": 18671, "epoch": 518, "lr": 0.001} +{"train_loss": 0.036929626017808914, "train_loss_bc": 0.036929626017808914, "train_loss_llm": 0.0, "grad_norm": 0.08404611051082611, "global_step": 18672, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03269527480006218, "train_loss_bc": 0.03269527480006218, "train_loss_llm": 0.0, "grad_norm": 0.030350813642144203, "global_step": 18673, "epoch": 518, "lr": 0.001} +{"train_loss": 0.0350969135761261, "train_loss_bc": 0.03508434444665909, "train_loss_llm": 0.0012568235397338867, "grad_norm": 0.056380320340394974, "global_step": 18674, "epoch": 518, "lr": 0.001} +{"train_loss": 0.02786119095981121, "train_loss_bc": 0.02786119095981121, "train_loss_llm": 0.0, "grad_norm": 0.0737665519118309, "global_step": 18675, "epoch": 518, "lr": 0.001} +{"train_loss": 0.038774482905864716, "train_loss_bc": 0.038774482905864716, "train_loss_llm": 0.0, "grad_norm": 0.11146704107522964, "global_step": 18676, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03385036438703537, "train_loss_bc": 0.03385036438703537, "train_loss_llm": 0.0, "grad_norm": 0.026029124855995178, "global_step": 18677, "epoch": 518, "lr": 0.001} +{"train_loss": 0.02914941869676113, "train_loss_bc": 0.02914941869676113, "train_loss_llm": 0.0, "grad_norm": 0.055327776819467545, "global_step": 18678, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03743882477283478, "train_loss_bc": 0.03743882477283478, "train_loss_llm": 0.0, "grad_norm": 0.07775159925222397, "global_step": 18679, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03377046436071396, "train_loss_bc": 0.03377046436071396, "train_loss_llm": 0.0, "grad_norm": 0.09690635651350021, "global_step": 18680, "epoch": 518, "lr": 0.001} +{"train_loss": 0.03826300427317619, "train_loss_bc": 0.03826300427317619, "train_loss_llm": 0.0, "grad_norm": 0.032420143485069275, "global_step": 18681, "epoch": 518, "lr": 0.001} +{"train_loss": 0.031915225088596344, "train_loss_bc": 0.0319119356572628, "train_loss_llm": 0.00032892823219299316, "grad_norm": 0.051905788481235504, "global_step": 18682, "epoch": 518, "lr": 0.001} +{"train_loss": 0.034015372292035155, "train_loss_bc": 0.030720453709363937, "train_loss_llm": 0.0, "grad_norm": 0.0847831517457962, "global_step": 18683, "epoch": 518, "lr": 0.001, "val_loss": 0.07503242045640945} +{"train_loss": 0.033919382840394974, "train_loss_bc": 0.033919382840394974, "train_loss_llm": 0.0, "grad_norm": 0.09495984762907028, "global_step": 18684, "epoch": 519, "lr": 0.001} +{"train_loss": 0.030888432636857033, "train_loss_bc": 0.030888432636857033, "train_loss_llm": 0.0, "grad_norm": 0.021648695692420006, "global_step": 18685, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03343018889427185, "train_loss_bc": 0.03341484069824219, "train_loss_llm": 0.001534789800643921, "grad_norm": 0.03382818028330803, "global_step": 18686, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03606010600924492, "train_loss_bc": 0.03606010600924492, "train_loss_llm": 0.0, "grad_norm": 0.042274344712495804, "global_step": 18687, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03449752926826477, "train_loss_bc": 0.03449752926826477, "train_loss_llm": 0.0, "grad_norm": 0.053170330822467804, "global_step": 18688, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03763819485902786, "train_loss_bc": 0.03763819485902786, "train_loss_llm": 0.0, "grad_norm": 0.035737793892621994, "global_step": 18689, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03611258417367935, "train_loss_bc": 0.03611258417367935, "train_loss_llm": 0.0, "grad_norm": 0.0493243932723999, "global_step": 18690, "epoch": 519, "lr": 0.001} +{"train_loss": 0.035905029624700546, "train_loss_bc": 0.035905029624700546, "train_loss_llm": 0.0, "grad_norm": 0.07018690556287766, "global_step": 18691, "epoch": 519, "lr": 0.001} +{"train_loss": 0.0365208238363266, "train_loss_bc": 0.0365208238363266, "train_loss_llm": 0.0, "grad_norm": 0.09368627518415451, "global_step": 18692, "epoch": 519, "lr": 0.001} +{"train_loss": 0.04122036695480347, "train_loss_bc": 0.04122036695480347, "train_loss_llm": 0.0, "grad_norm": 0.03631245717406273, "global_step": 18693, "epoch": 519, "lr": 0.001} +{"train_loss": 0.028750495985150337, "train_loss_bc": 0.028750495985150337, "train_loss_llm": 0.0, "grad_norm": 0.05630787834525108, "global_step": 18694, "epoch": 519, "lr": 0.001} +{"train_loss": 0.02950059063732624, "train_loss_bc": 0.02950059063732624, "train_loss_llm": 0.0, "grad_norm": 0.07840125262737274, "global_step": 18695, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03267862647771835, "train_loss_bc": 0.03267862647771835, "train_loss_llm": 0.0, "grad_norm": 0.10134384781122208, "global_step": 18696, "epoch": 519, "lr": 0.001} +{"train_loss": 0.035425346344709396, "train_loss_bc": 0.035425346344709396, "train_loss_llm": 0.0, "grad_norm": 0.0444863922894001, "global_step": 18697, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03568440303206444, "train_loss_bc": 0.03567776829004288, "train_loss_llm": 0.0006635487079620361, "grad_norm": 0.0671691820025444, "global_step": 18698, "epoch": 519, "lr": 0.001} +{"train_loss": 0.024787750095129013, "train_loss_bc": 0.02478465437889099, "train_loss_llm": 0.0003095269203186035, "grad_norm": 0.10038553178310394, "global_step": 18699, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03237389028072357, "train_loss_bc": 0.03237144276499748, "train_loss_llm": 0.000244826078414917, "grad_norm": 0.12501764297485352, "global_step": 18700, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03701155632734299, "train_loss_bc": 0.03697969391942024, "train_loss_llm": 0.003186255693435669, "grad_norm": 0.02918008342385292, "global_step": 18701, "epoch": 519, "lr": 0.001} +{"train_loss": 0.033796511590480804, "train_loss_bc": 0.033796511590480804, "train_loss_llm": 0.0, "grad_norm": 0.057369425892829895, "global_step": 18702, "epoch": 519, "lr": 0.001} +{"train_loss": 0.040188949555158615, "train_loss_bc": 0.04018387198448181, "train_loss_llm": 0.0005077123641967773, "grad_norm": 0.07607502490282059, "global_step": 18703, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03421146422624588, "train_loss_bc": 0.03421146422624588, "train_loss_llm": 0.0, "grad_norm": 0.0977158322930336, "global_step": 18704, "epoch": 519, "lr": 0.001} +{"train_loss": 0.034121014177799225, "train_loss_bc": 0.034121014177799225, "train_loss_llm": 0.0, "grad_norm": 0.027485316619277, "global_step": 18705, "epoch": 519, "lr": 0.001} +{"train_loss": 0.031102774664759636, "train_loss_bc": 0.031102774664759636, "train_loss_llm": 0.0, "grad_norm": 0.037717513740062714, "global_step": 18706, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03644436225295067, "train_loss_bc": 0.03644436225295067, "train_loss_llm": 0.0, "grad_norm": 0.045065537095069885, "global_step": 18707, "epoch": 519, "lr": 0.001} +{"train_loss": 0.034893520176410675, "train_loss_bc": 0.034893520176410675, "train_loss_llm": 0.0, "grad_norm": 0.050190117210149765, "global_step": 18708, "epoch": 519, "lr": 0.001} +{"train_loss": 0.026916489005088806, "train_loss_bc": 0.026916489005088806, "train_loss_llm": 0.0, "grad_norm": 0.02383185178041458, "global_step": 18709, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03473205119371414, "train_loss_bc": 0.03473205119371414, "train_loss_llm": 0.0, "grad_norm": 0.03807878866791725, "global_step": 18710, "epoch": 519, "lr": 0.001} +{"train_loss": 0.0348779521882534, "train_loss_bc": 0.0348779521882534, "train_loss_llm": 0.0, "grad_norm": 0.06222737580537796, "global_step": 18711, "epoch": 519, "lr": 0.001} +{"train_loss": 0.033892352133989334, "train_loss_bc": 0.033892352133989334, "train_loss_llm": 0.0, "grad_norm": 0.08418727666139603, "global_step": 18712, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03422459214925766, "train_loss_bc": 0.03421623632311821, "train_loss_llm": 0.0008355081081390381, "grad_norm": 0.03394201770424843, "global_step": 18713, "epoch": 519, "lr": 0.001} +{"train_loss": 0.031191743910312653, "train_loss_bc": 0.03118470497429371, "train_loss_llm": 0.0007038414478302002, "grad_norm": 0.0572892427444458, "global_step": 18714, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03952552005648613, "train_loss_bc": 0.03952552005648613, "train_loss_llm": 0.0, "grad_norm": 0.08948275446891785, "global_step": 18715, "epoch": 519, "lr": 0.001} +{"train_loss": 0.036312002688646317, "train_loss_bc": 0.036312002688646317, "train_loss_llm": 0.0, "grad_norm": 0.10853094607591629, "global_step": 18716, "epoch": 519, "lr": 0.001} +{"train_loss": 0.02812410145998001, "train_loss_bc": 0.02812410145998001, "train_loss_llm": 0.0, "grad_norm": 0.027219373732805252, "global_step": 18717, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03703043609857559, "train_loss_bc": 0.03703043609857559, "train_loss_llm": 0.0, "grad_norm": 0.06409052014350891, "global_step": 18718, "epoch": 519, "lr": 0.001} +{"train_loss": 0.03410755480743117, "train_loss_bc": 0.03388083726167679, "train_loss_llm": 0.0, "grad_norm": 0.0901050716638565, "global_step": 18719, "epoch": 519, "lr": 0.001, "val_loss": 0.07576867938041687} +{"train_loss": 0.036588381975889206, "train_loss_bc": 0.03658698499202728, "train_loss_llm": 0.00013956427574157715, "grad_norm": 0.11851365864276886, "global_step": 18720, "epoch": 520, "lr": 0.001} +{"train_loss": 0.04055706411600113, "train_loss_bc": 0.04055706411600113, "train_loss_llm": 0.0, "grad_norm": 0.03314461186528206, "global_step": 18721, "epoch": 520, "lr": 0.001} +{"train_loss": 0.036069002002477646, "train_loss_bc": 0.036069002002477646, "train_loss_llm": 0.0, "grad_norm": 0.04402243718504906, "global_step": 18722, "epoch": 520, "lr": 0.001} +{"train_loss": 0.031699106097221375, "train_loss_bc": 0.031699106097221375, "train_loss_llm": 0.0, "grad_norm": 0.05943192541599274, "global_step": 18723, "epoch": 520, "lr": 0.001} +{"train_loss": 0.028794946148991585, "train_loss_bc": 0.028794946148991585, "train_loss_llm": 0.0, "grad_norm": 0.07939848303794861, "global_step": 18724, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03413436561822891, "train_loss_bc": 0.03412109613418579, "train_loss_llm": 0.0013269484043121338, "grad_norm": 0.026888469234108925, "global_step": 18725, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03230739384889603, "train_loss_bc": 0.03230680525302887, "train_loss_llm": 5.8710575103759766e-05, "grad_norm": 0.03979774937033653, "global_step": 18726, "epoch": 520, "lr": 0.001} +{"train_loss": 0.029712701216340065, "train_loss_bc": 0.029712701216340065, "train_loss_llm": 0.0, "grad_norm": 0.05250580236315727, "global_step": 18727, "epoch": 520, "lr": 0.001} +{"train_loss": 0.02575281634926796, "train_loss_bc": 0.02575281634926796, "train_loss_llm": 0.0, "grad_norm": 0.0628352165222168, "global_step": 18728, "epoch": 520, "lr": 0.001} +{"train_loss": 0.031531739979982376, "train_loss_bc": 0.03151966631412506, "train_loss_llm": 0.0012074708938598633, "grad_norm": 0.025321491062641144, "global_step": 18729, "epoch": 520, "lr": 0.001} +{"train_loss": 0.034000463783741, "train_loss_bc": 0.034000463783741, "train_loss_llm": 0.0, "grad_norm": 0.03552595525979996, "global_step": 18730, "epoch": 520, "lr": 0.001} +{"train_loss": 0.033444780856370926, "train_loss_bc": 0.03342938423156738, "train_loss_llm": 0.0015395283699035645, "grad_norm": 0.049592405557632446, "global_step": 18731, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03365787863731384, "train_loss_bc": 0.03365129977464676, "train_loss_llm": 0.0006580650806427002, "grad_norm": 0.06263698637485504, "global_step": 18732, "epoch": 520, "lr": 0.001} +{"train_loss": 0.032507799565792084, "train_loss_bc": 0.032507799565792084, "train_loss_llm": 0.0, "grad_norm": 0.028687968850135803, "global_step": 18733, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03143496811389923, "train_loss_bc": 0.03143496811389923, "train_loss_llm": 0.0, "grad_norm": 0.04451310634613037, "global_step": 18734, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03348885476589203, "train_loss_bc": 0.03348885476589203, "train_loss_llm": 0.0, "grad_norm": 0.05970295891165733, "global_step": 18735, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03315359354019165, "train_loss_bc": 0.03315350040793419, "train_loss_llm": 9.238719940185547e-06, "grad_norm": 0.07525096833705902, "global_step": 18736, "epoch": 520, "lr": 0.001} +{"train_loss": 0.030075816437602043, "train_loss_bc": 0.030075816437602043, "train_loss_llm": 0.0, "grad_norm": 0.03106815181672573, "global_step": 18737, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03925565257668495, "train_loss_bc": 0.03925565257668495, "train_loss_llm": 0.0, "grad_norm": 0.05764217674732208, "global_step": 18738, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03597661852836609, "train_loss_bc": 0.03597661852836609, "train_loss_llm": 0.0, "grad_norm": 0.08205690234899521, "global_step": 18739, "epoch": 520, "lr": 0.001} +{"train_loss": 0.041472744196653366, "train_loss_bc": 0.041472047567367554, "train_loss_llm": 6.970763206481934e-05, "grad_norm": 0.10696892440319061, "global_step": 18740, "epoch": 520, "lr": 0.001} +{"train_loss": 0.04235360026359558, "train_loss_bc": 0.04235360026359558, "train_loss_llm": 0.0, "grad_norm": 0.033910367637872696, "global_step": 18741, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03586767613887787, "train_loss_bc": 0.03586767613887787, "train_loss_llm": 0.0, "grad_norm": 0.051029328256845474, "global_step": 18742, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03546203672885895, "train_loss_bc": 0.0354568287730217, "train_loss_llm": 0.0005208849906921387, "grad_norm": 0.06952346861362457, "global_step": 18743, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03186040744185448, "train_loss_bc": 0.0318511426448822, "train_loss_llm": 0.0009264349937438965, "grad_norm": 0.08617702126502991, "global_step": 18744, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03417406603693962, "train_loss_bc": 0.03417300432920456, "train_loss_llm": 0.00010603666305541992, "grad_norm": 0.025380518287420273, "global_step": 18745, "epoch": 520, "lr": 0.001} +{"train_loss": 0.034199588000774384, "train_loss_bc": 0.034199588000774384, "train_loss_llm": 0.0, "grad_norm": 0.044854965060949326, "global_step": 18746, "epoch": 520, "lr": 0.001} +{"train_loss": 0.037515368312597275, "train_loss_bc": 0.037515368312597275, "train_loss_llm": 0.0, "grad_norm": 0.06127142533659935, "global_step": 18747, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03680875524878502, "train_loss_bc": 0.03680875524878502, "train_loss_llm": 0.0, "grad_norm": 0.07393155246973038, "global_step": 18748, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03312422335147858, "train_loss_bc": 0.03312422335147858, "train_loss_llm": 0.0, "grad_norm": 0.025965947657823563, "global_step": 18749, "epoch": 520, "lr": 0.001} +{"train_loss": 0.028685979545116425, "train_loss_bc": 0.028676092624664307, "train_loss_llm": 0.000988781452178955, "grad_norm": 0.0386735200881958, "global_step": 18750, "epoch": 520, "lr": 0.001} +{"train_loss": 0.034154877066612244, "train_loss_bc": 0.034154877066612244, "train_loss_llm": 0.0, "grad_norm": 0.053401801735162735, "global_step": 18751, "epoch": 520, "lr": 0.001} +{"train_loss": 0.026838505640625954, "train_loss_bc": 0.02683534286916256, "train_loss_llm": 0.00031626224517822266, "grad_norm": 0.06263677775859833, "global_step": 18752, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03495631366968155, "train_loss_bc": 0.03495631366968155, "train_loss_llm": 0.0, "grad_norm": 0.034216053783893585, "global_step": 18753, "epoch": 520, "lr": 0.001} +{"train_loss": 0.03608868643641472, "train_loss_bc": 0.03608868643641472, "train_loss_llm": 0.0, "grad_norm": 0.04834095388650894, "global_step": 18754, "epoch": 520, "lr": 0.001} +{"train_loss": 0.034045926502181426, "train_loss_bc": 0.03794658184051514, "train_loss_llm": 0.0, "grad_norm": 0.07788382470607758, "global_step": 18755, "epoch": 520, "lr": 0.001, "train/cumulative_reward": 64.79243533373908, "train/mean_score": 13.733220433777294, "train/success_rate": 0.4, "test/cumulative_reward": 47.23849190307126, "test/mean_score": 9.036982242078654, "test/success_rate": 0.22, "val_loss": 0.07554280012845993, "train_action_mse_error": 0.03690320998430252} +{"train_loss": 0.03212039917707443, "train_loss_bc": 0.03212039917707443, "train_loss_llm": 0.0, "grad_norm": 0.10321712493896484, "global_step": 18756, "epoch": 521, "lr": 0.001} +{"train_loss": 0.02823786437511444, "train_loss_bc": 0.02823428064584732, "train_loss_llm": 0.00035831332206726074, "grad_norm": 0.027537332847714424, "global_step": 18757, "epoch": 521, "lr": 0.001} +{"train_loss": 0.035595569759607315, "train_loss_bc": 0.03557470813393593, "train_loss_llm": 0.002086132764816284, "grad_norm": 0.04500061273574829, "global_step": 18758, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03688889369368553, "train_loss_bc": 0.03688889369368553, "train_loss_llm": 0.0, "grad_norm": 0.0591457337141037, "global_step": 18759, "epoch": 521, "lr": 0.001} +{"train_loss": 0.023219279944896698, "train_loss_bc": 0.023219279944896698, "train_loss_llm": 0.0, "grad_norm": 0.08540275692939758, "global_step": 18760, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03401142731308937, "train_loss_bc": 0.03400660678744316, "train_loss_llm": 0.0004820525646209717, "grad_norm": 0.03912336379289627, "global_step": 18761, "epoch": 521, "lr": 0.001} +{"train_loss": 0.031663648784160614, "train_loss_bc": 0.031663648784160614, "train_loss_llm": 0.0, "grad_norm": 0.05870002880692482, "global_step": 18762, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03308665379881859, "train_loss_bc": 0.03308665379881859, "train_loss_llm": 0.0, "grad_norm": 0.07992337644100189, "global_step": 18763, "epoch": 521, "lr": 0.001} +{"train_loss": 0.039607271552085876, "train_loss_bc": 0.039607271552085876, "train_loss_llm": 0.0, "grad_norm": 0.09764517098665237, "global_step": 18764, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03654276952147484, "train_loss_bc": 0.03654276952147484, "train_loss_llm": 0.0, "grad_norm": 0.03407948091626167, "global_step": 18765, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03135623037815094, "train_loss_bc": 0.03134911507368088, "train_loss_llm": 0.0007115602493286133, "grad_norm": 0.04842628538608551, "global_step": 18766, "epoch": 521, "lr": 0.001} +{"train_loss": 0.031417716294527054, "train_loss_bc": 0.031417716294527054, "train_loss_llm": 0.0, "grad_norm": 0.05756442993879318, "global_step": 18767, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03570785000920296, "train_loss_bc": 0.03570785000920296, "train_loss_llm": 0.0, "grad_norm": 0.07662536948919296, "global_step": 18768, "epoch": 521, "lr": 0.001} +{"train_loss": 0.034469615668058395, "train_loss_bc": 0.034458667039871216, "train_loss_llm": 0.0010949373245239258, "grad_norm": 0.027152083814144135, "global_step": 18769, "epoch": 521, "lr": 0.001} +{"train_loss": 0.033415332436561584, "train_loss_bc": 0.033407773822546005, "train_loss_llm": 0.0007558465003967285, "grad_norm": 0.0396784283220768, "global_step": 18770, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03657306358218193, "train_loss_bc": 0.0365411601960659, "train_loss_llm": 0.0031902194023132324, "grad_norm": 0.05273255705833435, "global_step": 18771, "epoch": 521, "lr": 0.001} +{"train_loss": 0.023625334724783897, "train_loss_bc": 0.023625334724783897, "train_loss_llm": 0.0, "grad_norm": 0.06466683745384216, "global_step": 18772, "epoch": 521, "lr": 0.001} +{"train_loss": 0.029242970049381256, "train_loss_bc": 0.029242970049381256, "train_loss_llm": 0.0, "grad_norm": 0.02592085488140583, "global_step": 18773, "epoch": 521, "lr": 0.001} +{"train_loss": 0.04100142791867256, "train_loss_bc": 0.04100142791867256, "train_loss_llm": 0.0, "grad_norm": 0.043559420853853226, "global_step": 18774, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03775708004832268, "train_loss_bc": 0.03775708004832268, "train_loss_llm": 0.0, "grad_norm": 0.05822264030575752, "global_step": 18775, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03438748046755791, "train_loss_bc": 0.03438748046755791, "train_loss_llm": 0.0, "grad_norm": 0.07411455363035202, "global_step": 18776, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03154012933373451, "train_loss_bc": 0.03154012933373451, "train_loss_llm": 0.0, "grad_norm": 0.03463074564933777, "global_step": 18777, "epoch": 521, "lr": 0.001} +{"train_loss": 0.032117150723934174, "train_loss_bc": 0.032098427414894104, "train_loss_llm": 0.001872420310974121, "grad_norm": 0.049121443182229996, "global_step": 18778, "epoch": 521, "lr": 0.001} +{"train_loss": 0.035799529403448105, "train_loss_bc": 0.035799529403448105, "train_loss_llm": 0.0, "grad_norm": 0.06919559091329575, "global_step": 18779, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03751964494585991, "train_loss_bc": 0.03751964494585991, "train_loss_llm": 0.0, "grad_norm": 0.09163004159927368, "global_step": 18780, "epoch": 521, "lr": 0.001} +{"train_loss": 0.034903962165117264, "train_loss_bc": 0.034887589514255524, "train_loss_llm": 0.001637279987335205, "grad_norm": 0.026155468076467514, "global_step": 18781, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03645835816860199, "train_loss_bc": 0.03644392639398575, "train_loss_llm": 0.0014431178569793701, "grad_norm": 0.04498569294810295, "global_step": 18782, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03289388120174408, "train_loss_bc": 0.03289388120174408, "train_loss_llm": 0.0, "grad_norm": 0.06467577815055847, "global_step": 18783, "epoch": 521, "lr": 0.001} +{"train_loss": 0.031784091144800186, "train_loss_bc": 0.03177700191736221, "train_loss_llm": 0.0007089674472808838, "grad_norm": 0.0913422480225563, "global_step": 18784, "epoch": 521, "lr": 0.001} +{"train_loss": 0.030485494062304497, "train_loss_bc": 0.030482767149806023, "train_loss_llm": 0.0002726912498474121, "grad_norm": 0.023879604414105415, "global_step": 18785, "epoch": 521, "lr": 0.001} +{"train_loss": 0.037448637187480927, "train_loss_bc": 0.037448637187480927, "train_loss_llm": 0.0, "grad_norm": 0.045942530035972595, "global_step": 18786, "epoch": 521, "lr": 0.001} +{"train_loss": 0.038645777851343155, "train_loss_bc": 0.038645777851343155, "train_loss_llm": 0.0, "grad_norm": 0.0654081180691719, "global_step": 18787, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03790728747844696, "train_loss_bc": 0.037890929728746414, "train_loss_llm": 0.0016358494758605957, "grad_norm": 0.08123911172151566, "global_step": 18788, "epoch": 521, "lr": 0.001} +{"train_loss": 0.03746210038661957, "train_loss_bc": 0.03746210038661957, "train_loss_llm": 0.0, "grad_norm": 0.029451623558998108, "global_step": 18789, "epoch": 521, "lr": 0.001} +{"train_loss": 0.034749116748571396, "train_loss_bc": 0.034747444093227386, "train_loss_llm": 0.00016710162162780762, "grad_norm": 0.041673362255096436, "global_step": 18790, "epoch": 521, "lr": 0.001} +{"train_loss": 0.034108693401018776, "train_loss_bc": 0.03826992213726044, "train_loss_llm": 0.0, "grad_norm": 0.08107081800699234, "global_step": 18791, "epoch": 521, "lr": 0.001, "val_loss": 0.07719557732343674} +{"train_loss": 0.027406133711338043, "train_loss_bc": 0.027406133711338043, "train_loss_llm": 0.0, "grad_norm": 0.08454283326864243, "global_step": 18792, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03855975717306137, "train_loss_bc": 0.03855946287512779, "train_loss_llm": 2.956390380859375e-05, "grad_norm": 0.03259885683655739, "global_step": 18793, "epoch": 522, "lr": 0.001} +{"train_loss": 0.0294338446110487, "train_loss_bc": 0.0294338446110487, "train_loss_llm": 0.0, "grad_norm": 0.03737088292837143, "global_step": 18794, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03157883509993553, "train_loss_bc": 0.03157883509993553, "train_loss_llm": 0.0, "grad_norm": 0.04441624507308006, "global_step": 18795, "epoch": 522, "lr": 0.001} +{"train_loss": 0.038602013140916824, "train_loss_bc": 0.038602013140916824, "train_loss_llm": 0.0, "grad_norm": 0.06200841814279556, "global_step": 18796, "epoch": 522, "lr": 0.001} +{"train_loss": 0.02678610198199749, "train_loss_bc": 0.02678610198199749, "train_loss_llm": 0.0, "grad_norm": 0.0256772693246603, "global_step": 18797, "epoch": 522, "lr": 0.001} +{"train_loss": 0.032844290137290955, "train_loss_bc": 0.032844290137290955, "train_loss_llm": 0.0, "grad_norm": 0.04062913358211517, "global_step": 18798, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03262793645262718, "train_loss_bc": 0.03262793645262718, "train_loss_llm": 0.0, "grad_norm": 0.06309563666582108, "global_step": 18799, "epoch": 522, "lr": 0.001} +{"train_loss": 0.04414476081728935, "train_loss_bc": 0.04414476081728935, "train_loss_llm": 0.0, "grad_norm": 0.10810085386037827, "global_step": 18800, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03582385182380676, "train_loss_bc": 0.03582385182380676, "train_loss_llm": 0.0, "grad_norm": 0.028769711032509804, "global_step": 18801, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03552914783358574, "train_loss_bc": 0.03552914783358574, "train_loss_llm": 0.0, "grad_norm": 0.06438906490802765, "global_step": 18802, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03426326811313629, "train_loss_bc": 0.03426326811313629, "train_loss_llm": 0.0, "grad_norm": 0.07312125712633133, "global_step": 18803, "epoch": 522, "lr": 0.001} +{"train_loss": 0.02829066850244999, "train_loss_bc": 0.02829066850244999, "train_loss_llm": 0.0, "grad_norm": 0.09566256403923035, "global_step": 18804, "epoch": 522, "lr": 0.001} +{"train_loss": 0.037047918885946274, "train_loss_bc": 0.037047918885946274, "train_loss_llm": 0.0, "grad_norm": 0.02745206654071808, "global_step": 18805, "epoch": 522, "lr": 0.001} +{"train_loss": 0.028303762897849083, "train_loss_bc": 0.028303762897849083, "train_loss_llm": 0.0, "grad_norm": 0.04191982001066208, "global_step": 18806, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03264027088880539, "train_loss_bc": 0.03264027088880539, "train_loss_llm": 0.0, "grad_norm": 0.05511106550693512, "global_step": 18807, "epoch": 522, "lr": 0.001} +{"train_loss": 0.030453339219093323, "train_loss_bc": 0.030453339219093323, "train_loss_llm": 0.0, "grad_norm": 0.06251686066389084, "global_step": 18808, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03328632935881615, "train_loss_bc": 0.03328632935881615, "train_loss_llm": 0.0, "grad_norm": 0.029690470546483994, "global_step": 18809, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03143337741494179, "train_loss_bc": 0.03143337741494179, "train_loss_llm": 0.0, "grad_norm": 0.047686006873846054, "global_step": 18810, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03496532514691353, "train_loss_bc": 0.03496532514691353, "train_loss_llm": 0.0, "grad_norm": 0.06089827045798302, "global_step": 18811, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03172868490219116, "train_loss_bc": 0.031705595552921295, "train_loss_llm": 0.0023089051246643066, "grad_norm": 0.07427962869405746, "global_step": 18812, "epoch": 522, "lr": 0.001} +{"train_loss": 0.034980304539203644, "train_loss_bc": 0.034980304539203644, "train_loss_llm": 0.0, "grad_norm": 0.027626803144812584, "global_step": 18813, "epoch": 522, "lr": 0.001} +{"train_loss": 0.04079819470643997, "train_loss_bc": 0.04077458754181862, "train_loss_llm": 0.0023608505725860596, "grad_norm": 0.043572671711444855, "global_step": 18814, "epoch": 522, "lr": 0.001} +{"train_loss": 0.037735894322395325, "train_loss_bc": 0.037735894322395325, "train_loss_llm": 0.0, "grad_norm": 0.052303120493888855, "global_step": 18815, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03438004478812218, "train_loss_bc": 0.034367404878139496, "train_loss_llm": 0.001264035701751709, "grad_norm": 0.0707792341709137, "global_step": 18816, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03182799369096756, "train_loss_bc": 0.0318170040845871, "train_loss_llm": 0.0010989010334014893, "grad_norm": 0.03099638968706131, "global_step": 18817, "epoch": 522, "lr": 0.001} +{"train_loss": 0.036667488515377045, "train_loss_bc": 0.036650437861680984, "train_loss_llm": 0.001705169677734375, "grad_norm": 0.055506449192762375, "global_step": 18818, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03268013894557953, "train_loss_bc": 0.03268013894557953, "train_loss_llm": 0.0, "grad_norm": 0.08385975658893585, "global_step": 18819, "epoch": 522, "lr": 0.001} +{"train_loss": 0.034114278852939606, "train_loss_bc": 0.034114278852939606, "train_loss_llm": 0.0, "grad_norm": 0.08794083446264267, "global_step": 18820, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03224269673228264, "train_loss_bc": 0.03220450505614281, "train_loss_llm": 0.0038193166255950928, "grad_norm": 0.030199063941836357, "global_step": 18821, "epoch": 522, "lr": 0.001} +{"train_loss": 0.031451769173145294, "train_loss_bc": 0.031451769173145294, "train_loss_llm": 0.0, "grad_norm": 0.049006253480911255, "global_step": 18822, "epoch": 522, "lr": 0.001} +{"train_loss": 0.034699153155088425, "train_loss_bc": 0.034699153155088425, "train_loss_llm": 0.0, "grad_norm": 0.07758410274982452, "global_step": 18823, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03213643282651901, "train_loss_bc": 0.03212779760360718, "train_loss_llm": 0.0008634328842163086, "grad_norm": 0.10286244750022888, "global_step": 18824, "epoch": 522, "lr": 0.001} +{"train_loss": 0.039151694625616074, "train_loss_bc": 0.039151694625616074, "train_loss_llm": 0.0, "grad_norm": 0.029641158878803253, "global_step": 18825, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03615650534629822, "train_loss_bc": 0.0361279733479023, "train_loss_llm": 0.0028530657291412354, "grad_norm": 0.03996722400188446, "global_step": 18826, "epoch": 522, "lr": 0.001} +{"train_loss": 0.03400482806480593, "train_loss_bc": 0.03940160199999809, "train_loss_llm": 0.0, "grad_norm": 0.06226518750190735, "global_step": 18827, "epoch": 522, "lr": 0.001, "val_loss": 0.07621912658214569} +{"train_loss": 0.03213634341955185, "train_loss_bc": 0.03213634341955185, "train_loss_llm": 0.0, "grad_norm": 0.0775841623544693, "global_step": 18828, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03069290891289711, "train_loss_bc": 0.03069290891289711, "train_loss_llm": 0.0, "grad_norm": 0.026249047368764877, "global_step": 18829, "epoch": 523, "lr": 0.001} +{"train_loss": 0.026907123625278473, "train_loss_bc": 0.026907123625278473, "train_loss_llm": 0.0, "grad_norm": 0.0417093001306057, "global_step": 18830, "epoch": 523, "lr": 0.001} +{"train_loss": 0.035119153559207916, "train_loss_bc": 0.035119153559207916, "train_loss_llm": 0.0, "grad_norm": 0.041489049792289734, "global_step": 18831, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03360685706138611, "train_loss_bc": 0.03360685706138611, "train_loss_llm": 0.0, "grad_norm": 0.04830430448055267, "global_step": 18832, "epoch": 523, "lr": 0.001} +{"train_loss": 0.038530703634023666, "train_loss_bc": 0.038530703634023666, "train_loss_llm": 0.0, "grad_norm": 0.027385054156184196, "global_step": 18833, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03696431592106819, "train_loss_bc": 0.036946993321180344, "train_loss_llm": 0.0017320811748504639, "grad_norm": 0.037182603031396866, "global_step": 18834, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03603702783584595, "train_loss_bc": 0.03603227064013481, "train_loss_llm": 0.00047582387924194336, "grad_norm": 0.04593871533870697, "global_step": 18835, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03649153187870979, "train_loss_bc": 0.03649153187870979, "train_loss_llm": 0.0, "grad_norm": 0.05856627598404884, "global_step": 18836, "epoch": 523, "lr": 0.001} +{"train_loss": 0.0281742624938488, "train_loss_bc": 0.0281742624938488, "train_loss_llm": 0.0, "grad_norm": 0.023746857419610023, "global_step": 18837, "epoch": 523, "lr": 0.001} +{"train_loss": 0.034002844244241714, "train_loss_bc": 0.03398933261632919, "train_loss_llm": 0.0013510584831237793, "grad_norm": 0.03931145742535591, "global_step": 18838, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03440450504422188, "train_loss_bc": 0.03440450504422188, "train_loss_llm": 0.0, "grad_norm": 0.05568821728229523, "global_step": 18839, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03241611644625664, "train_loss_bc": 0.032408129423856735, "train_loss_llm": 0.0007985830307006836, "grad_norm": 0.06990963220596313, "global_step": 18840, "epoch": 523, "lr": 0.001} +{"train_loss": 0.032836321741342545, "train_loss_bc": 0.032836321741342545, "train_loss_llm": 0.0, "grad_norm": 0.025852395221590996, "global_step": 18841, "epoch": 523, "lr": 0.001} +{"train_loss": 0.028887758031487465, "train_loss_bc": 0.02886742167174816, "train_loss_llm": 0.0020335912704467773, "grad_norm": 0.05013812705874443, "global_step": 18842, "epoch": 523, "lr": 0.001} +{"train_loss": 0.033631499856710434, "train_loss_bc": 0.033631499856710434, "train_loss_llm": 0.0, "grad_norm": 0.07413654029369354, "global_step": 18843, "epoch": 523, "lr": 0.001} +{"train_loss": 0.032229091972112656, "train_loss_bc": 0.032229091972112656, "train_loss_llm": 0.0, "grad_norm": 0.09571797400712967, "global_step": 18844, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03404850512742996, "train_loss_bc": 0.03404850512742996, "train_loss_llm": 0.0, "grad_norm": 0.02404164895415306, "global_step": 18845, "epoch": 523, "lr": 0.001} +{"train_loss": 0.0359128899872303, "train_loss_bc": 0.0359128899872303, "train_loss_llm": 0.0, "grad_norm": 0.04619358479976654, "global_step": 18846, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03289469704031944, "train_loss_bc": 0.03289469704031944, "train_loss_llm": 0.0, "grad_norm": 0.06546193361282349, "global_step": 18847, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03794414550065994, "train_loss_bc": 0.03794414550065994, "train_loss_llm": 0.0, "grad_norm": 0.08544816076755524, "global_step": 18848, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03920997306704521, "train_loss_bc": 0.03920997306704521, "train_loss_llm": 0.0, "grad_norm": 0.029074812307953835, "global_step": 18849, "epoch": 523, "lr": 0.001} +{"train_loss": 0.0322115495800972, "train_loss_bc": 0.032206516712903976, "train_loss_llm": 0.0005031824111938477, "grad_norm": 0.042825352400541306, "global_step": 18850, "epoch": 523, "lr": 0.001} +{"train_loss": 0.032704830169677734, "train_loss_bc": 0.032704830169677734, "train_loss_llm": 0.0, "grad_norm": 0.06140802055597305, "global_step": 18851, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03831811994314194, "train_loss_bc": 0.03831811994314194, "train_loss_llm": 0.0, "grad_norm": 0.07871563732624054, "global_step": 18852, "epoch": 523, "lr": 0.001} +{"train_loss": 0.02893252484500408, "train_loss_bc": 0.02893252484500408, "train_loss_llm": 0.0, "grad_norm": 0.027420973405241966, "global_step": 18853, "epoch": 523, "lr": 0.001} +{"train_loss": 0.02966918610036373, "train_loss_bc": 0.02966918610036373, "train_loss_llm": 0.0, "grad_norm": 0.044693224132061005, "global_step": 18854, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03431205078959465, "train_loss_bc": 0.03431205078959465, "train_loss_llm": 0.0, "grad_norm": 0.05053002014756203, "global_step": 18855, "epoch": 523, "lr": 0.001} +{"train_loss": 0.038409605622291565, "train_loss_bc": 0.03840210288763046, "train_loss_llm": 0.0007503926753997803, "grad_norm": 0.06144407391548157, "global_step": 18856, "epoch": 523, "lr": 0.001} +{"train_loss": 0.038431279361248016, "train_loss_bc": 0.038431279361248016, "train_loss_llm": 0.0, "grad_norm": 0.03033161163330078, "global_step": 18857, "epoch": 523, "lr": 0.001} +{"train_loss": 0.039344582706689835, "train_loss_bc": 0.039344582706689835, "train_loss_llm": 0.0, "grad_norm": 0.05772882699966431, "global_step": 18858, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03173491358757019, "train_loss_bc": 0.03173491358757019, "train_loss_llm": 0.0, "grad_norm": 0.08137625455856323, "global_step": 18859, "epoch": 523, "lr": 0.001} +{"train_loss": 0.03424866124987602, "train_loss_bc": 0.03424866124987602, "train_loss_llm": 0.0, "grad_norm": 0.08938977122306824, "global_step": 18860, "epoch": 523, "lr": 0.001} +{"train_loss": 0.032926060259342194, "train_loss_bc": 0.03291377052664757, "train_loss_llm": 0.0012288689613342285, "grad_norm": 0.02966618724167347, "global_step": 18861, "epoch": 523, "lr": 0.001} +{"train_loss": 0.034878309816122055, "train_loss_bc": 0.034878309816122055, "train_loss_llm": 0.0, "grad_norm": 0.06048068404197693, "global_step": 18862, "epoch": 523, "lr": 0.001} +{"train_loss": 0.033866249283568725, "train_loss_bc": 0.029984723776578903, "train_loss_llm": 0.0, "grad_norm": 0.07256543636322021, "global_step": 18863, "epoch": 523, "lr": 0.001, "val_loss": 0.0780428946018219} +{"train_loss": 0.027169089764356613, "train_loss_bc": 0.027163347229361534, "train_loss_llm": 0.0005742311477661133, "grad_norm": 0.09657645225524902, "global_step": 18864, "epoch": 524, "lr": 0.001} +{"train_loss": 0.0343126617372036, "train_loss_bc": 0.034290604293346405, "train_loss_llm": 0.0022059082984924316, "grad_norm": 0.027806542813777924, "global_step": 18865, "epoch": 524, "lr": 0.001} +{"train_loss": 0.034265048801898956, "train_loss_bc": 0.03424466773867607, "train_loss_llm": 0.002037942409515381, "grad_norm": 0.05151178315281868, "global_step": 18866, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03689121827483177, "train_loss_bc": 0.03689121827483177, "train_loss_llm": 0.0, "grad_norm": 0.0750405490398407, "global_step": 18867, "epoch": 524, "lr": 0.001} +{"train_loss": 0.029974307864904404, "train_loss_bc": 0.029974307864904404, "train_loss_llm": 0.0, "grad_norm": 0.08985331654548645, "global_step": 18868, "epoch": 524, "lr": 0.001} +{"train_loss": 0.035305045545101166, "train_loss_bc": 0.03530352935194969, "train_loss_llm": 0.00015145540237426758, "grad_norm": 0.026890818029642105, "global_step": 18869, "epoch": 524, "lr": 0.001} +{"train_loss": 0.032473910599946976, "train_loss_bc": 0.032473910599946976, "train_loss_llm": 0.0, "grad_norm": 0.03989550843834877, "global_step": 18870, "epoch": 524, "lr": 0.001} +{"train_loss": 0.0368243083357811, "train_loss_bc": 0.0368243083357811, "train_loss_llm": 0.0, "grad_norm": 0.04969343543052673, "global_step": 18871, "epoch": 524, "lr": 0.001} +{"train_loss": 0.02647250145673752, "train_loss_bc": 0.02647250145673752, "train_loss_llm": 0.0, "grad_norm": 0.058365751057863235, "global_step": 18872, "epoch": 524, "lr": 0.001} +{"train_loss": 0.033266086131334305, "train_loss_bc": 0.033262692391872406, "train_loss_llm": 0.0003393590450286865, "grad_norm": 0.03787953779101372, "global_step": 18873, "epoch": 524, "lr": 0.001} +{"train_loss": 0.0347503125667572, "train_loss_bc": 0.0347503125667572, "train_loss_llm": 0.0, "grad_norm": 0.0586123913526535, "global_step": 18874, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03392989933490753, "train_loss_bc": 0.03392989933490753, "train_loss_llm": 0.0, "grad_norm": 0.082133948802948, "global_step": 18875, "epoch": 524, "lr": 0.001} +{"train_loss": 0.034633245319128036, "train_loss_bc": 0.034633245319128036, "train_loss_llm": 0.0, "grad_norm": 0.09876164048910141, "global_step": 18876, "epoch": 524, "lr": 0.001} +{"train_loss": 0.027151137590408325, "train_loss_bc": 0.027151137590408325, "train_loss_llm": 0.0, "grad_norm": 0.0279579758644104, "global_step": 18877, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03295351192355156, "train_loss_bc": 0.03295351192355156, "train_loss_llm": 0.0, "grad_norm": 0.04212804138660431, "global_step": 18878, "epoch": 524, "lr": 0.001} +{"train_loss": 0.032839711755514145, "train_loss_bc": 0.032839711755514145, "train_loss_llm": 0.0, "grad_norm": 0.0635422095656395, "global_step": 18879, "epoch": 524, "lr": 0.001} +{"train_loss": 0.029790282249450684, "train_loss_bc": 0.029790282249450684, "train_loss_llm": 0.0, "grad_norm": 0.08306101709604263, "global_step": 18880, "epoch": 524, "lr": 0.001} +{"train_loss": 0.040899887681007385, "train_loss_bc": 0.040899887681007385, "train_loss_llm": 0.0, "grad_norm": 0.045418884605169296, "global_step": 18881, "epoch": 524, "lr": 0.001} +{"train_loss": 0.0386245995759964, "train_loss_bc": 0.0386245995759964, "train_loss_llm": 0.0, "grad_norm": 0.07033799588680267, "global_step": 18882, "epoch": 524, "lr": 0.001} +{"train_loss": 0.031305402517318726, "train_loss_bc": 0.03129425272345543, "train_loss_llm": 0.001114964485168457, "grad_norm": 0.0815817266702652, "global_step": 18883, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03651939332485199, "train_loss_bc": 0.03651939332485199, "train_loss_llm": 0.0, "grad_norm": 0.09894168376922607, "global_step": 18884, "epoch": 524, "lr": 0.001} +{"train_loss": 0.02847774140536785, "train_loss_bc": 0.02847774140536785, "train_loss_llm": 0.0, "grad_norm": 0.021888604387640953, "global_step": 18885, "epoch": 524, "lr": 0.001} +{"train_loss": 0.036792438477277756, "train_loss_bc": 0.036792438477277756, "train_loss_llm": 0.0, "grad_norm": 0.04322011396288872, "global_step": 18886, "epoch": 524, "lr": 0.001} +{"train_loss": 0.04240727424621582, "train_loss_bc": 0.04240727424621582, "train_loss_llm": 0.0, "grad_norm": 0.06304509937763214, "global_step": 18887, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03823573887348175, "train_loss_bc": 0.03823573887348175, "train_loss_llm": 0.0, "grad_norm": 0.08421993255615234, "global_step": 18888, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03757490590214729, "train_loss_bc": 0.03757490590214729, "train_loss_llm": 0.0, "grad_norm": 0.027731090784072876, "global_step": 18889, "epoch": 524, "lr": 0.001} +{"train_loss": 0.029738685116171837, "train_loss_bc": 0.02973177097737789, "train_loss_llm": 0.0006913542747497559, "grad_norm": 0.04109228402376175, "global_step": 18890, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03214307874441147, "train_loss_bc": 0.03214307874441147, "train_loss_llm": 0.0, "grad_norm": 0.05606864392757416, "global_step": 18891, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03868046775460243, "train_loss_bc": 0.03868046775460243, "train_loss_llm": 0.0, "grad_norm": 0.0692705363035202, "global_step": 18892, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03429068997502327, "train_loss_bc": 0.03429068997502327, "train_loss_llm": 0.0, "grad_norm": 0.030974000692367554, "global_step": 18893, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03732271119952202, "train_loss_bc": 0.0373208224773407, "train_loss_llm": 0.0001887679100036621, "grad_norm": 0.05607229471206665, "global_step": 18894, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03409629687666893, "train_loss_bc": 0.03409629687666893, "train_loss_llm": 0.0, "grad_norm": 0.08490074425935745, "global_step": 18895, "epoch": 524, "lr": 0.001} +{"train_loss": 0.031172938644886017, "train_loss_bc": 0.031172938644886017, "train_loss_llm": 0.0, "grad_norm": 0.11025150865316391, "global_step": 18896, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03896845504641533, "train_loss_bc": 0.03896327689290047, "train_loss_llm": 0.0005176663398742676, "grad_norm": 0.03282172977924347, "global_step": 18897, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03055099956691265, "train_loss_bc": 0.03055099956691265, "train_loss_llm": 0.0, "grad_norm": 0.05160515755414963, "global_step": 18898, "epoch": 524, "lr": 0.001} +{"train_loss": 0.03417948690346546, "train_loss_bc": 0.03965754434466362, "train_loss_llm": 0.0, "grad_norm": 0.070126973092556, "global_step": 18899, "epoch": 524, "lr": 0.001, "val_loss": 0.07590951025485992} +{"train_loss": 0.03153245896100998, "train_loss_bc": 0.03152124583721161, "train_loss_llm": 0.0011214017868041992, "grad_norm": 0.09141882508993149, "global_step": 18900, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03469958156347275, "train_loss_bc": 0.03469958156347275, "train_loss_llm": 0.0, "grad_norm": 0.03685944154858589, "global_step": 18901, "epoch": 525, "lr": 0.001} +{"train_loss": 0.032040320336818695, "train_loss_bc": 0.032040320336818695, "train_loss_llm": 0.0, "grad_norm": 0.05143381282687187, "global_step": 18902, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03527426719665527, "train_loss_bc": 0.03527426719665527, "train_loss_llm": 0.0, "grad_norm": 0.07562573999166489, "global_step": 18903, "epoch": 525, "lr": 0.001} +{"train_loss": 0.034952402114868164, "train_loss_bc": 0.034935079514980316, "train_loss_llm": 0.0017322003841400146, "grad_norm": 0.08269014209508896, "global_step": 18904, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03387626260519028, "train_loss_bc": 0.03387626260519028, "train_loss_llm": 0.0, "grad_norm": 0.03151315450668335, "global_step": 18905, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03682061284780502, "train_loss_bc": 0.03682061284780502, "train_loss_llm": 0.0, "grad_norm": 0.05939441919326782, "global_step": 18906, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03621754050254822, "train_loss_bc": 0.03621754050254822, "train_loss_llm": 0.0, "grad_norm": 0.085484579205513, "global_step": 18907, "epoch": 525, "lr": 0.001} +{"train_loss": 0.031940121203660965, "train_loss_bc": 0.031940121203660965, "train_loss_llm": 0.0, "grad_norm": 0.10672123730182648, "global_step": 18908, "epoch": 525, "lr": 0.001} +{"train_loss": 0.030297361314296722, "train_loss_bc": 0.030284054577350616, "train_loss_llm": 0.001330733299255371, "grad_norm": 0.026268942281603813, "global_step": 18909, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03522559255361557, "train_loss_bc": 0.03519126772880554, "train_loss_llm": 0.0034325122833251953, "grad_norm": 0.04713796451687813, "global_step": 18910, "epoch": 525, "lr": 0.001} +{"train_loss": 0.02909702993929386, "train_loss_bc": 0.02909702993929386, "train_loss_llm": 0.0, "grad_norm": 0.06248500198125839, "global_step": 18911, "epoch": 525, "lr": 0.001} +{"train_loss": 0.039013151079416275, "train_loss_bc": 0.039013151079416275, "train_loss_llm": 0.0, "grad_norm": 0.08357801288366318, "global_step": 18912, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03227565437555313, "train_loss_bc": 0.03227565437555313, "train_loss_llm": 0.0, "grad_norm": 0.03130690008401871, "global_step": 18913, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03482426330447197, "train_loss_bc": 0.03482426330447197, "train_loss_llm": 0.0, "grad_norm": 0.05675896629691124, "global_step": 18914, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03290580213069916, "train_loss_bc": 0.03290580213069916, "train_loss_llm": 0.0, "grad_norm": 0.07690750062465668, "global_step": 18915, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03379904478788376, "train_loss_bc": 0.03379904478788376, "train_loss_llm": 0.0, "grad_norm": 0.10166040062904358, "global_step": 18916, "epoch": 525, "lr": 0.001} +{"train_loss": 0.029477950185537338, "train_loss_bc": 0.029463518410921097, "train_loss_llm": 0.0014430880546569824, "grad_norm": 0.028138292953372, "global_step": 18917, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03875161334872246, "train_loss_bc": 0.03875161334872246, "train_loss_llm": 0.0, "grad_norm": 0.05511678382754326, "global_step": 18918, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03490043431520462, "train_loss_bc": 0.03490043431520462, "train_loss_llm": 0.0, "grad_norm": 0.08794807642698288, "global_step": 18919, "epoch": 525, "lr": 0.001} +{"train_loss": 0.035742565989494324, "train_loss_bc": 0.035742565989494324, "train_loss_llm": 0.0, "grad_norm": 0.11199583858251572, "global_step": 18920, "epoch": 525, "lr": 0.001} +{"train_loss": 0.034798331558704376, "train_loss_bc": 0.034798331558704376, "train_loss_llm": 0.0, "grad_norm": 0.031274326145648956, "global_step": 18921, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03671820089221001, "train_loss_bc": 0.03671686723828316, "train_loss_llm": 0.00013327598571777344, "grad_norm": 0.0603969469666481, "global_step": 18922, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03439156711101532, "train_loss_bc": 0.03439156711101532, "train_loss_llm": 0.0, "grad_norm": 0.08101711422204971, "global_step": 18923, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03164675086736679, "train_loss_bc": 0.03164675086736679, "train_loss_llm": 0.0, "grad_norm": 0.09772350639104843, "global_step": 18924, "epoch": 525, "lr": 0.001} +{"train_loss": 0.035317786037921906, "train_loss_bc": 0.035317786037921906, "train_loss_llm": 0.0, "grad_norm": 0.031936243176460266, "global_step": 18925, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03560788929462433, "train_loss_bc": 0.03560788929462433, "train_loss_llm": 0.0, "grad_norm": 0.047315292060375214, "global_step": 18926, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03441711142659187, "train_loss_bc": 0.03441622480750084, "train_loss_llm": 8.881092071533203e-05, "grad_norm": 0.07142029702663422, "global_step": 18927, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03275922313332558, "train_loss_bc": 0.03275046870112419, "train_loss_llm": 0.0008752942085266113, "grad_norm": 0.08951913565397263, "global_step": 18928, "epoch": 525, "lr": 0.001} +{"train_loss": 0.035983260720968246, "train_loss_bc": 0.035983260720968246, "train_loss_llm": 0.0, "grad_norm": 0.02759542688727379, "global_step": 18929, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03566400334239006, "train_loss_bc": 0.03566400334239006, "train_loss_llm": 0.0, "grad_norm": 0.04255719855427742, "global_step": 18930, "epoch": 525, "lr": 0.001} +{"train_loss": 0.027965327724814415, "train_loss_bc": 0.027962064370512962, "train_loss_llm": 0.00032639503479003906, "grad_norm": 0.056974250823259354, "global_step": 18931, "epoch": 525, "lr": 0.001} +{"train_loss": 0.031171215698122978, "train_loss_bc": 0.031171215698122978, "train_loss_llm": 0.0, "grad_norm": 0.07744643837213516, "global_step": 18932, "epoch": 525, "lr": 0.001} +{"train_loss": 0.03638378158211708, "train_loss_bc": 0.03638378158211708, "train_loss_llm": 0.0, "grad_norm": 0.035777509212493896, "global_step": 18933, "epoch": 525, "lr": 0.001} +{"train_loss": 0.037384163588285446, "train_loss_bc": 0.037384163588285446, "train_loss_llm": 0.0, "grad_norm": 0.05265409126877785, "global_step": 18934, "epoch": 525, "lr": 0.001} +{"train_loss": 0.034270521408567824, "train_loss_bc": 0.03985114023089409, "train_loss_llm": 0.0014988183975219727, "grad_norm": 0.07289617508649826, "global_step": 18935, "epoch": 525, "lr": 0.001, "train/cumulative_reward": 59.66350305275044, "train/mean_score": 13.722813291986066, "train/success_rate": 0.4, "test/cumulative_reward": 29.26380341482306, "test/mean_score": 7.4364554820696585, "test/success_rate": 0.2, "val_loss": 0.07763787358999252, "train_action_mse_error": 0.03760708123445511} +{"train_loss": 0.035574980080127716, "train_loss_bc": 0.035574980080127716, "train_loss_llm": 0.0, "grad_norm": 0.08307530730962753, "global_step": 18936, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03574454039335251, "train_loss_bc": 0.03574454039335251, "train_loss_llm": 0.0, "grad_norm": 0.03144717961549759, "global_step": 18937, "epoch": 526, "lr": 0.001} +{"train_loss": 0.032046135514974594, "train_loss_bc": 0.03204233944416046, "train_loss_llm": 0.0003794431686401367, "grad_norm": 0.05461277440190315, "global_step": 18938, "epoch": 526, "lr": 0.001} +{"train_loss": 0.029809098690748215, "train_loss_bc": 0.029809098690748215, "train_loss_llm": 0.0, "grad_norm": 0.07138533145189285, "global_step": 18939, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03613635152578354, "train_loss_bc": 0.03613635152578354, "train_loss_llm": 0.0, "grad_norm": 0.10459811985492706, "global_step": 18940, "epoch": 526, "lr": 0.001} +{"train_loss": 0.038161300122737885, "train_loss_bc": 0.038142986595630646, "train_loss_llm": 0.0018315017223358154, "grad_norm": 0.03591056168079376, "global_step": 18941, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03730181232094765, "train_loss_bc": 0.03723880648612976, "train_loss_llm": 0.006300419569015503, "grad_norm": 0.06799359619617462, "global_step": 18942, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03257094696164131, "train_loss_bc": 0.03257094696164131, "train_loss_llm": 0.0, "grad_norm": 0.08540354669094086, "global_step": 18943, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03441485017538071, "train_loss_bc": 0.034398674964904785, "train_loss_llm": 0.0016174614429473877, "grad_norm": 0.1096758022904396, "global_step": 18944, "epoch": 526, "lr": 0.001} +{"train_loss": 0.042825985699892044, "train_loss_bc": 0.042825985699892044, "train_loss_llm": 0.0, "grad_norm": 0.03581700101494789, "global_step": 18945, "epoch": 526, "lr": 0.001} +{"train_loss": 0.029716528952121735, "train_loss_bc": 0.029716528952121735, "train_loss_llm": 0.0, "grad_norm": 0.049078285694122314, "global_step": 18946, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03669688105583191, "train_loss_bc": 0.03669688105583191, "train_loss_llm": 0.0, "grad_norm": 0.06643397361040115, "global_step": 18947, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03330615162849426, "train_loss_bc": 0.03330615162849426, "train_loss_llm": 0.0, "grad_norm": 0.08680643141269684, "global_step": 18948, "epoch": 526, "lr": 0.001} +{"train_loss": 0.04251149296760559, "train_loss_bc": 0.04249643534421921, "train_loss_llm": 0.0015055835247039795, "grad_norm": 0.028596913442015648, "global_step": 18949, "epoch": 526, "lr": 0.001} +{"train_loss": 0.027797887101769447, "train_loss_bc": 0.027797887101769447, "train_loss_llm": 0.0, "grad_norm": 0.04127572104334831, "global_step": 18950, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03204658627510071, "train_loss_bc": 0.03204658627510071, "train_loss_llm": 0.0, "grad_norm": 0.0575002059340477, "global_step": 18951, "epoch": 526, "lr": 0.001} +{"train_loss": 0.028162484988570213, "train_loss_bc": 0.028162484988570213, "train_loss_llm": 0.0, "grad_norm": 0.07018634676933289, "global_step": 18952, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03592268005013466, "train_loss_bc": 0.03591874614357948, "train_loss_llm": 0.00039333105087280273, "grad_norm": 0.025767231360077858, "global_step": 18953, "epoch": 526, "lr": 0.001} +{"train_loss": 0.033714912831783295, "train_loss_bc": 0.033714912831783295, "train_loss_llm": 0.0, "grad_norm": 0.03791258484125137, "global_step": 18954, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03009599633514881, "train_loss_bc": 0.03009599633514881, "train_loss_llm": 0.0, "grad_norm": 0.052177391946315765, "global_step": 18955, "epoch": 526, "lr": 0.001} +{"train_loss": 0.033229853957891464, "train_loss_bc": 0.033229853957891464, "train_loss_llm": 0.0, "grad_norm": 0.0733552798628807, "global_step": 18956, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03838503733277321, "train_loss_bc": 0.03838503733277321, "train_loss_llm": 0.0, "grad_norm": 0.030313100665807724, "global_step": 18957, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03742557391524315, "train_loss_bc": 0.03742557391524315, "train_loss_llm": 0.0, "grad_norm": 0.05627942085266113, "global_step": 18958, "epoch": 526, "lr": 0.001} +{"train_loss": 0.032420773059129715, "train_loss_bc": 0.032420773059129715, "train_loss_llm": 0.0, "grad_norm": 0.0685463398694992, "global_step": 18959, "epoch": 526, "lr": 0.001} +{"train_loss": 0.030636943876743317, "train_loss_bc": 0.030634181573987007, "train_loss_llm": 0.00027626752853393555, "grad_norm": 0.0833209678530693, "global_step": 18960, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03691282868385315, "train_loss_bc": 0.03691282868385315, "train_loss_llm": 0.0, "grad_norm": 0.03030521422624588, "global_step": 18961, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03494682535529137, "train_loss_bc": 0.03494682535529137, "train_loss_llm": 0.0, "grad_norm": 0.049375712871551514, "global_step": 18962, "epoch": 526, "lr": 0.001} +{"train_loss": 0.032966695725917816, "train_loss_bc": 0.032966695725917816, "train_loss_llm": 0.0, "grad_norm": 0.07467859983444214, "global_step": 18963, "epoch": 526, "lr": 0.001} +{"train_loss": 0.032118044793605804, "train_loss_bc": 0.03209436684846878, "train_loss_llm": 0.0023679137229919434, "grad_norm": 0.08884922415018082, "global_step": 18964, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03235543891787529, "train_loss_bc": 0.03235543891787529, "train_loss_llm": 0.0, "grad_norm": 0.03474002704024315, "global_step": 18965, "epoch": 526, "lr": 0.001} +{"train_loss": 0.038288962095975876, "train_loss_bc": 0.038288962095975876, "train_loss_llm": 0.0, "grad_norm": 0.06176352500915527, "global_step": 18966, "epoch": 526, "lr": 0.001} +{"train_loss": 0.029040107503533363, "train_loss_bc": 0.029040107503533363, "train_loss_llm": 0.0, "grad_norm": 0.06364850699901581, "global_step": 18967, "epoch": 526, "lr": 0.001} +{"train_loss": 0.03523670509457588, "train_loss_bc": 0.03523670509457588, "train_loss_llm": 0.0, "grad_norm": 0.07608489692211151, "global_step": 18968, "epoch": 526, "lr": 0.001} +{"train_loss": 0.029211563989520073, "train_loss_bc": 0.029211563989520073, "train_loss_llm": 0.0, "grad_norm": 0.02395748719573021, "global_step": 18969, "epoch": 526, "lr": 0.001} +{"train_loss": 0.035703349858522415, "train_loss_bc": 0.035703349858522415, "train_loss_llm": 0.0, "grad_norm": 0.04544010013341904, "global_step": 18970, "epoch": 526, "lr": 0.001} +{"train_loss": 0.034082378829932876, "train_loss_bc": 0.03352658823132515, "train_loss_llm": 0.00027433037757873535, "grad_norm": 0.052080411463975906, "global_step": 18971, "epoch": 526, "lr": 0.001, "val_loss": 0.07646685838699341} +{"train_loss": 0.03160437196493149, "train_loss_bc": 0.03160437196493149, "train_loss_llm": 0.0, "grad_norm": 0.06372488290071487, "global_step": 18972, "epoch": 527, "lr": 0.001} +{"train_loss": 0.037036214023828506, "train_loss_bc": 0.037036214023828506, "train_loss_llm": 0.0, "grad_norm": 0.030592046678066254, "global_step": 18973, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03124091401696205, "train_loss_bc": 0.03124091401696205, "train_loss_llm": 0.0, "grad_norm": 0.046395692974328995, "global_step": 18974, "epoch": 527, "lr": 0.001} +{"train_loss": 0.027664240449666977, "train_loss_bc": 0.027664240449666977, "train_loss_llm": 0.0, "grad_norm": 0.0545215904712677, "global_step": 18975, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03345569968223572, "train_loss_bc": 0.03343958780169487, "train_loss_llm": 0.0016112923622131348, "grad_norm": 0.06419640779495239, "global_step": 18976, "epoch": 527, "lr": 0.001} +{"train_loss": 0.034207917749881744, "train_loss_bc": 0.034207917749881744, "train_loss_llm": 0.0, "grad_norm": 0.03358776494860649, "global_step": 18977, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03271449729800224, "train_loss_bc": 0.03269707038998604, "train_loss_llm": 0.0017427802085876465, "grad_norm": 0.05471620336174965, "global_step": 18978, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03537243604660034, "train_loss_bc": 0.03537243604660034, "train_loss_llm": 0.0, "grad_norm": 0.07045932859182358, "global_step": 18979, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03669841215014458, "train_loss_bc": 0.03669841215014458, "train_loss_llm": 0.0, "grad_norm": 0.08756767213344574, "global_step": 18980, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03196079283952713, "train_loss_bc": 0.03196079283952713, "train_loss_llm": 0.0, "grad_norm": 0.02502158284187317, "global_step": 18981, "epoch": 527, "lr": 0.001} +{"train_loss": 0.034232866019010544, "train_loss_bc": 0.034232866019010544, "train_loss_llm": 0.0, "grad_norm": 0.0438423715531826, "global_step": 18982, "epoch": 527, "lr": 0.001} +{"train_loss": 0.031601350754499435, "train_loss_bc": 0.031601350754499435, "train_loss_llm": 0.0, "grad_norm": 0.06495147198438644, "global_step": 18983, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03592268005013466, "train_loss_bc": 0.03592268005013466, "train_loss_llm": 0.0, "grad_norm": 0.07870453596115112, "global_step": 18984, "epoch": 527, "lr": 0.001} +{"train_loss": 0.033761169761419296, "train_loss_bc": 0.033761169761419296, "train_loss_llm": 0.0, "grad_norm": 0.03591950610280037, "global_step": 18985, "epoch": 527, "lr": 0.001} +{"train_loss": 0.029674917459487915, "train_loss_bc": 0.029665380716323853, "train_loss_llm": 0.0009536147117614746, "grad_norm": 0.05516698211431503, "global_step": 18986, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03359932452440262, "train_loss_bc": 0.033589452505111694, "train_loss_llm": 0.0009873807430267334, "grad_norm": 0.06951101124286652, "global_step": 18987, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03583088144659996, "train_loss_bc": 0.03583088144659996, "train_loss_llm": 0.0, "grad_norm": 0.09157443046569824, "global_step": 18988, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03612862527370453, "train_loss_bc": 0.03612862527370453, "train_loss_llm": 0.0, "grad_norm": 0.02856059931218624, "global_step": 18989, "epoch": 527, "lr": 0.001} +{"train_loss": 0.041416820138692856, "train_loss_bc": 0.041408274322748184, "train_loss_llm": 0.0008546113967895508, "grad_norm": 0.049906667321920395, "global_step": 18990, "epoch": 527, "lr": 0.001} +{"train_loss": 0.037546318024396896, "train_loss_bc": 0.037546318024396896, "train_loss_llm": 0.0, "grad_norm": 0.07124315947294235, "global_step": 18991, "epoch": 527, "lr": 0.001} +{"train_loss": 0.028652263805270195, "train_loss_bc": 0.028652263805270195, "train_loss_llm": 0.0, "grad_norm": 0.08658653497695923, "global_step": 18992, "epoch": 527, "lr": 0.001} +{"train_loss": 0.030544089153409004, "train_loss_bc": 0.030544089153409004, "train_loss_llm": 0.0, "grad_norm": 0.029895378276705742, "global_step": 18993, "epoch": 527, "lr": 0.001} +{"train_loss": 0.029396850615739822, "train_loss_bc": 0.029396850615739822, "train_loss_llm": 0.0, "grad_norm": 0.05081866309046745, "global_step": 18994, "epoch": 527, "lr": 0.001} +{"train_loss": 0.033824969083070755, "train_loss_bc": 0.033824969083070755, "train_loss_llm": 0.0, "grad_norm": 0.06901560723781586, "global_step": 18995, "epoch": 527, "lr": 0.001} +{"train_loss": 0.032913271337747574, "train_loss_bc": 0.032911669462919235, "train_loss_llm": 0.00016003847122192383, "grad_norm": 0.08608154952526093, "global_step": 18996, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03277282789349556, "train_loss_bc": 0.032770391553640366, "train_loss_llm": 0.00024366378784179688, "grad_norm": 0.0214979387819767, "global_step": 18997, "epoch": 527, "lr": 0.001} +{"train_loss": 0.034378159791231155, "train_loss_bc": 0.034366339445114136, "train_loss_llm": 0.0011820793151855469, "grad_norm": 0.041642073541879654, "global_step": 18998, "epoch": 527, "lr": 0.001} +{"train_loss": 0.039204906672239304, "train_loss_bc": 0.03918231278657913, "train_loss_llm": 0.0022593438625335693, "grad_norm": 0.06210210174322128, "global_step": 18999, "epoch": 527, "lr": 0.001} +{"train_loss": 0.035005874931812286, "train_loss_bc": 0.03500451147556305, "train_loss_llm": 0.00013631582260131836, "grad_norm": 0.08081075549125671, "global_step": 19000, "epoch": 527, "lr": 0.001} +{"train_loss": 0.030251240357756615, "train_loss_bc": 0.030245155096054077, "train_loss_llm": 0.0006085336208343506, "grad_norm": 0.03473919630050659, "global_step": 19001, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03814694285392761, "train_loss_bc": 0.03814694285392761, "train_loss_llm": 0.0, "grad_norm": 0.052283015102148056, "global_step": 19002, "epoch": 527, "lr": 0.001} +{"train_loss": 0.0319003090262413, "train_loss_bc": 0.0319003090262413, "train_loss_llm": 0.0, "grad_norm": 0.05984969064593315, "global_step": 19003, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03742131590843201, "train_loss_bc": 0.037415649741888046, "train_loss_llm": 0.0005666613578796387, "grad_norm": 0.07635387033224106, "global_step": 19004, "epoch": 527, "lr": 0.001} +{"train_loss": 0.04222964122891426, "train_loss_bc": 0.04222964122891426, "train_loss_llm": 0.0, "grad_norm": 0.03286956995725632, "global_step": 19005, "epoch": 527, "lr": 0.001} +{"train_loss": 0.03009798936545849, "train_loss_bc": 0.03009798936545849, "train_loss_llm": 0.0, "grad_norm": 0.04924721643328667, "global_step": 19006, "epoch": 527, "lr": 0.001} +{"train_loss": 0.033878065542214446, "train_loss_bc": 0.031187497079372406, "train_loss_llm": 0.0011760890483856201, "grad_norm": 0.06542488932609558, "global_step": 19007, "epoch": 527, "lr": 0.001, "val_loss": 0.0765305906534195} +{"train_loss": 0.031005872413516045, "train_loss_bc": 0.031005872413516045, "train_loss_llm": 0.0, "grad_norm": 0.08413231372833252, "global_step": 19008, "epoch": 528, "lr": 0.001} +{"train_loss": 0.030802864581346512, "train_loss_bc": 0.03078615851700306, "train_loss_llm": 0.0016705989837646484, "grad_norm": 0.021281596273183823, "global_step": 19009, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03662116080522537, "train_loss_bc": 0.03662116080522537, "train_loss_llm": 0.0, "grad_norm": 0.04062775894999504, "global_step": 19010, "epoch": 528, "lr": 0.001} +{"train_loss": 0.028780151158571243, "train_loss_bc": 0.028780151158571243, "train_loss_llm": 0.0, "grad_norm": 0.05198180675506592, "global_step": 19011, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03957153484225273, "train_loss_bc": 0.03957153484225273, "train_loss_llm": 0.0, "grad_norm": 0.075459785759449, "global_step": 19012, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03270815312862396, "train_loss_bc": 0.03270815312862396, "train_loss_llm": 0.0, "grad_norm": 0.02733045257627964, "global_step": 19013, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03515682741999626, "train_loss_bc": 0.03515682741999626, "train_loss_llm": 0.0, "grad_norm": 0.04109809175133705, "global_step": 19014, "epoch": 528, "lr": 0.001} +{"train_loss": 0.030764581635594368, "train_loss_bc": 0.030764581635594368, "train_loss_llm": 0.0, "grad_norm": 0.0458100289106369, "global_step": 19015, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03352954611182213, "train_loss_bc": 0.033526502549648285, "train_loss_llm": 0.00030422210693359375, "grad_norm": 0.055303409695625305, "global_step": 19016, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03595622256398201, "train_loss_bc": 0.03595352545380592, "train_loss_llm": 0.00026986002922058105, "grad_norm": 0.02473043091595173, "global_step": 19017, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03422198444604874, "train_loss_bc": 0.03422198444604874, "train_loss_llm": 0.0, "grad_norm": 0.033921051770448685, "global_step": 19018, "epoch": 528, "lr": 0.001} +{"train_loss": 0.036682624369859695, "train_loss_bc": 0.036682624369859695, "train_loss_llm": 0.0, "grad_norm": 0.05817602947354317, "global_step": 19019, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03374475613236427, "train_loss_bc": 0.03371456637978554, "train_loss_llm": 0.003018796443939209, "grad_norm": 0.07460050284862518, "global_step": 19020, "epoch": 528, "lr": 0.001} +{"train_loss": 0.04040970280766487, "train_loss_bc": 0.040391772985458374, "train_loss_llm": 0.001792818307876587, "grad_norm": 0.03386431559920311, "global_step": 19021, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03906258940696716, "train_loss_bc": 0.03906258940696716, "train_loss_llm": 0.0, "grad_norm": 0.044078513979911804, "global_step": 19022, "epoch": 528, "lr": 0.001} +{"train_loss": 0.035342443734407425, "train_loss_bc": 0.035342443734407425, "train_loss_llm": 0.0, "grad_norm": 0.06247151270508766, "global_step": 19023, "epoch": 528, "lr": 0.001} +{"train_loss": 0.030021164566278458, "train_loss_bc": 0.030020778998732567, "train_loss_llm": 3.853440284729004e-05, "grad_norm": 0.06481805443763733, "global_step": 19024, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03309499844908714, "train_loss_bc": 0.03309499844908714, "train_loss_llm": 0.0, "grad_norm": 0.03331184387207031, "global_step": 19025, "epoch": 528, "lr": 0.001} +{"train_loss": 0.034464966505765915, "train_loss_bc": 0.034464966505765915, "train_loss_llm": 0.0, "grad_norm": 0.05486070737242699, "global_step": 19026, "epoch": 528, "lr": 0.001} +{"train_loss": 0.031673744320869446, "train_loss_bc": 0.031673744320869446, "train_loss_llm": 0.0, "grad_norm": 0.07386037707328796, "global_step": 19027, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03513167425990105, "train_loss_bc": 0.03513167425990105, "train_loss_llm": 0.0, "grad_norm": 0.08928332477807999, "global_step": 19028, "epoch": 528, "lr": 0.001} +{"train_loss": 0.034257858991622925, "train_loss_bc": 0.034257858991622925, "train_loss_llm": 0.0, "grad_norm": 0.027679534628987312, "global_step": 19029, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03400159999728203, "train_loss_bc": 0.03400159999728203, "train_loss_llm": 0.0, "grad_norm": 0.04237682744860649, "global_step": 19030, "epoch": 528, "lr": 0.001} +{"train_loss": 0.028126928955316544, "train_loss_bc": 0.028126928955316544, "train_loss_llm": 0.0, "grad_norm": 0.05336017534136772, "global_step": 19031, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03594468533992767, "train_loss_bc": 0.03594468533992767, "train_loss_llm": 0.0, "grad_norm": 0.06942009180784225, "global_step": 19032, "epoch": 528, "lr": 0.001} +{"train_loss": 0.026844125241041183, "train_loss_bc": 0.026844125241041183, "train_loss_llm": 0.0, "grad_norm": 0.027891656383872032, "global_step": 19033, "epoch": 528, "lr": 0.001} +{"train_loss": 0.033290814608335495, "train_loss_bc": 0.03327777609229088, "train_loss_llm": 0.0013038218021392822, "grad_norm": 0.038258228451013565, "global_step": 19034, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03288080543279648, "train_loss_bc": 0.03288080543279648, "train_loss_llm": 0.0, "grad_norm": 0.06193454563617706, "global_step": 19035, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03456655889749527, "train_loss_bc": 0.03456655889749527, "train_loss_llm": 0.0, "grad_norm": 0.07603131234645844, "global_step": 19036, "epoch": 528, "lr": 0.001} +{"train_loss": 0.027912793681025505, "train_loss_bc": 0.027902251109480858, "train_loss_llm": 0.0010542571544647217, "grad_norm": 0.01960715278983116, "global_step": 19037, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03032342903316021, "train_loss_bc": 0.03032342903316021, "train_loss_llm": 0.0, "grad_norm": 0.044747963547706604, "global_step": 19038, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03670218959450722, "train_loss_bc": 0.03669394925236702, "train_loss_llm": 0.0008240342140197754, "grad_norm": 0.05670468881726265, "global_step": 19039, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03527814894914627, "train_loss_bc": 0.03527814894914627, "train_loss_llm": 0.0, "grad_norm": 0.07339829206466675, "global_step": 19040, "epoch": 528, "lr": 0.001} +{"train_loss": 0.035015448927879333, "train_loss_bc": 0.035015448927879333, "train_loss_llm": 0.0, "grad_norm": 0.02452515810728073, "global_step": 19041, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03442025184631348, "train_loss_bc": 0.03442025184631348, "train_loss_llm": 0.0, "grad_norm": 0.039108145982027054, "global_step": 19042, "epoch": 528, "lr": 0.001} +{"train_loss": 0.03371071939667066, "train_loss_bc": 0.03527269512414932, "train_loss_llm": 0.0, "grad_norm": 0.053115665912628174, "global_step": 19043, "epoch": 528, "lr": 0.001, "val_loss": 0.0769772008061409} +{"train_loss": 0.03038502112030983, "train_loss_bc": 0.03038502112030983, "train_loss_llm": 0.0, "grad_norm": 0.0690634697675705, "global_step": 19044, "epoch": 529, "lr": 0.001} +{"train_loss": 0.028781535103917122, "train_loss_bc": 0.028781535103917122, "train_loss_llm": 0.0, "grad_norm": 0.025059005245566368, "global_step": 19045, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03282563388347626, "train_loss_bc": 0.03280284255743027, "train_loss_llm": 0.0022791028022766113, "grad_norm": 0.04277089983224869, "global_step": 19046, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03393496200442314, "train_loss_bc": 0.03393496200442314, "train_loss_llm": 0.0, "grad_norm": 0.054338760673999786, "global_step": 19047, "epoch": 529, "lr": 0.001} +{"train_loss": 0.029851971194148064, "train_loss_bc": 0.029851971194148064, "train_loss_llm": 0.0, "grad_norm": 0.06305091083049774, "global_step": 19048, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03695366904139519, "train_loss_bc": 0.03695366904139519, "train_loss_llm": 0.0, "grad_norm": 0.026936445385217667, "global_step": 19049, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03606618568301201, "train_loss_bc": 0.03606618568301201, "train_loss_llm": 0.0, "grad_norm": 0.04820236563682556, "global_step": 19050, "epoch": 529, "lr": 0.001} +{"train_loss": 0.04040667414665222, "train_loss_bc": 0.04040667414665222, "train_loss_llm": 0.0, "grad_norm": 0.061046525835990906, "global_step": 19051, "epoch": 529, "lr": 0.001} +{"train_loss": 0.02629346027970314, "train_loss_bc": 0.02629346027970314, "train_loss_llm": 0.0, "grad_norm": 0.06999381631612778, "global_step": 19052, "epoch": 529, "lr": 0.001} +{"train_loss": 0.037422165274620056, "train_loss_bc": 0.0374196358025074, "train_loss_llm": 0.0002530515193939209, "grad_norm": 0.03058202750980854, "global_step": 19053, "epoch": 529, "lr": 0.001} +{"train_loss": 0.037297338247299194, "train_loss_bc": 0.03729502111673355, "train_loss_llm": 0.00023174285888671875, "grad_norm": 0.05405566841363907, "global_step": 19054, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03616248071193695, "train_loss_bc": 0.03616233542561531, "train_loss_llm": 1.436471939086914e-05, "grad_norm": 0.07028934359550476, "global_step": 19055, "epoch": 529, "lr": 0.001} +{"train_loss": 0.036351222544908524, "train_loss_bc": 0.036351222544908524, "train_loss_llm": 0.0, "grad_norm": 0.08748073130846024, "global_step": 19056, "epoch": 529, "lr": 0.001} +{"train_loss": 0.030725128948688507, "train_loss_bc": 0.030725128948688507, "train_loss_llm": 0.0, "grad_norm": 0.027226688340306282, "global_step": 19057, "epoch": 529, "lr": 0.001} +{"train_loss": 0.027925362810492516, "train_loss_bc": 0.027917681261897087, "train_loss_llm": 0.0007681548595428467, "grad_norm": 0.04304075613617897, "global_step": 19058, "epoch": 529, "lr": 0.001} +{"train_loss": 0.028754979372024536, "train_loss_bc": 0.028754979372024536, "train_loss_llm": 0.0, "grad_norm": 0.05829428881406784, "global_step": 19059, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03298661485314369, "train_loss_bc": 0.03298661485314369, "train_loss_llm": 0.0, "grad_norm": 0.07781863957643509, "global_step": 19060, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03143531084060669, "train_loss_bc": 0.03143531084060669, "train_loss_llm": 0.0, "grad_norm": 0.028802374377846718, "global_step": 19061, "epoch": 529, "lr": 0.001} +{"train_loss": 0.031034233048558235, "train_loss_bc": 0.031034233048558235, "train_loss_llm": 0.0, "grad_norm": 0.0428541898727417, "global_step": 19062, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03442232683300972, "train_loss_bc": 0.03441300615668297, "train_loss_llm": 0.0009321272373199463, "grad_norm": 0.05847093462944031, "global_step": 19063, "epoch": 529, "lr": 0.001} +{"train_loss": 0.035207077860832214, "train_loss_bc": 0.035207077860832214, "train_loss_llm": 0.0, "grad_norm": 0.06868641823530197, "global_step": 19064, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03463538736104965, "train_loss_bc": 0.03463505581021309, "train_loss_llm": 3.314018249511719e-05, "grad_norm": 0.026034049689769745, "global_step": 19065, "epoch": 529, "lr": 0.001} +{"train_loss": 0.029419470578432083, "train_loss_bc": 0.029419470578432083, "train_loss_llm": 0.0, "grad_norm": 0.044073812663555145, "global_step": 19066, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03796844556927681, "train_loss_bc": 0.03794877603650093, "train_loss_llm": 0.001967012882232666, "grad_norm": 0.05671735852956772, "global_step": 19067, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03426886722445488, "train_loss_bc": 0.03426768258213997, "train_loss_llm": 0.00011837482452392578, "grad_norm": 0.06944389641284943, "global_step": 19068, "epoch": 529, "lr": 0.001} +{"train_loss": 0.034664373844861984, "train_loss_bc": 0.034635353833436966, "train_loss_llm": 0.0029021501541137695, "grad_norm": 0.031785063445568085, "global_step": 19069, "epoch": 529, "lr": 0.001} +{"train_loss": 0.029519787058234215, "train_loss_bc": 0.029519787058234215, "train_loss_llm": 0.0, "grad_norm": 0.044420745223760605, "global_step": 19070, "epoch": 529, "lr": 0.001} +{"train_loss": 0.039495352655649185, "train_loss_bc": 0.03949468955397606, "train_loss_llm": 6.619095802307129e-05, "grad_norm": 0.05559490993618965, "global_step": 19071, "epoch": 529, "lr": 0.001} +{"train_loss": 0.04047052934765816, "train_loss_bc": 0.04047052934765816, "train_loss_llm": 0.0, "grad_norm": 0.06529192626476288, "global_step": 19072, "epoch": 529, "lr": 0.001} +{"train_loss": 0.030690135434269905, "train_loss_bc": 0.030690135434269905, "train_loss_llm": 0.0, "grad_norm": 0.0248691588640213, "global_step": 19073, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03365525230765343, "train_loss_bc": 0.03365525230765343, "train_loss_llm": 0.0, "grad_norm": 0.04220563918352127, "global_step": 19074, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03675241768360138, "train_loss_bc": 0.03675241768360138, "train_loss_llm": 0.0, "grad_norm": 0.06183118000626564, "global_step": 19075, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03189992532134056, "train_loss_bc": 0.03189992532134056, "train_loss_llm": 0.0, "grad_norm": 0.07222993671894073, "global_step": 19076, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03601068630814552, "train_loss_bc": 0.03601068630814552, "train_loss_llm": 0.0, "grad_norm": 0.02849939651787281, "global_step": 19077, "epoch": 529, "lr": 0.001} +{"train_loss": 0.03624358028173447, "train_loss_bc": 0.03624358028173447, "train_loss_llm": 0.0, "grad_norm": 0.0460423119366169, "global_step": 19078, "epoch": 529, "lr": 0.001} +{"train_loss": 0.0337638236168358, "train_loss_bc": 0.034580085426568985, "train_loss_llm": 0.0, "grad_norm": 0.05839899927377701, "global_step": 19079, "epoch": 529, "lr": 0.001, "val_loss": 0.07695205509662628} +{"train_loss": 0.03064146637916565, "train_loss_bc": 0.030641306191682816, "train_loss_llm": 1.6003847122192383e-05, "grad_norm": 0.0739586353302002, "global_step": 19080, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03448594734072685, "train_loss_bc": 0.034473031759262085, "train_loss_llm": 0.001291424036026001, "grad_norm": 0.034547485411167145, "global_step": 19081, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03231905400753021, "train_loss_bc": 0.03231905400753021, "train_loss_llm": 0.0, "grad_norm": 0.056117184460163116, "global_step": 19082, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03771604970097542, "train_loss_bc": 0.03771604970097542, "train_loss_llm": 0.0, "grad_norm": 0.08775819838047028, "global_step": 19083, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03307252749800682, "train_loss_bc": 0.0330657921731472, "train_loss_llm": 0.0006734728813171387, "grad_norm": 0.10476072877645493, "global_step": 19084, "epoch": 530, "lr": 0.001} +{"train_loss": 0.034727297723293304, "train_loss_bc": 0.034727297723293304, "train_loss_llm": 0.0, "grad_norm": 0.03158174827694893, "global_step": 19085, "epoch": 530, "lr": 0.001} +{"train_loss": 0.038371652364730835, "train_loss_bc": 0.038371652364730835, "train_loss_llm": 0.0, "grad_norm": 0.05198756232857704, "global_step": 19086, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03394779562950134, "train_loss_bc": 0.03394779562950134, "train_loss_llm": 0.0, "grad_norm": 0.06185189262032509, "global_step": 19087, "epoch": 530, "lr": 0.001} +{"train_loss": 0.032168738543987274, "train_loss_bc": 0.032168738543987274, "train_loss_llm": 0.0, "grad_norm": 0.07770177721977234, "global_step": 19088, "epoch": 530, "lr": 0.001} +{"train_loss": 0.030373694375157356, "train_loss_bc": 0.030373694375157356, "train_loss_llm": 0.0, "grad_norm": 0.02498452551662922, "global_step": 19089, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03261258080601692, "train_loss_bc": 0.03261258080601692, "train_loss_llm": 0.0, "grad_norm": 0.04377123713493347, "global_step": 19090, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03753003478050232, "train_loss_bc": 0.03753003478050232, "train_loss_llm": 0.0, "grad_norm": 0.06651636213064194, "global_step": 19091, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03177312761545181, "train_loss_bc": 0.03177312761545181, "train_loss_llm": 0.0, "grad_norm": 0.08562836796045303, "global_step": 19092, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03134267404675484, "train_loss_bc": 0.03134267404675484, "train_loss_llm": 0.0, "grad_norm": 0.024081716313958168, "global_step": 19093, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03525695204734802, "train_loss_bc": 0.035237330943346024, "train_loss_llm": 0.001962006092071533, "grad_norm": 0.0352962501347065, "global_step": 19094, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03189386799931526, "train_loss_bc": 0.03189386799931526, "train_loss_llm": 0.0, "grad_norm": 0.05450514703989029, "global_step": 19095, "epoch": 530, "lr": 0.001} +{"train_loss": 0.04117809981107712, "train_loss_bc": 0.04117809981107712, "train_loss_llm": 0.0, "grad_norm": 0.06814489513635635, "global_step": 19096, "epoch": 530, "lr": 0.001} +{"train_loss": 0.033329516649246216, "train_loss_bc": 0.033320244401693344, "train_loss_llm": 0.0009271204471588135, "grad_norm": 0.027714373543858528, "global_step": 19097, "epoch": 530, "lr": 0.001} +{"train_loss": 0.033630289137363434, "train_loss_bc": 0.033630289137363434, "train_loss_llm": 0.0, "grad_norm": 0.04729624465107918, "global_step": 19098, "epoch": 530, "lr": 0.001} +{"train_loss": 0.032159510999917984, "train_loss_bc": 0.03214726597070694, "train_loss_llm": 0.001224428415298462, "grad_norm": 0.06385787576436996, "global_step": 19099, "epoch": 530, "lr": 0.001} +{"train_loss": 0.02962609939277172, "train_loss_bc": 0.02962609939277172, "train_loss_llm": 0.0, "grad_norm": 0.07861162722110748, "global_step": 19100, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03428562358021736, "train_loss_bc": 0.03428562358021736, "train_loss_llm": 0.0, "grad_norm": 0.034990038722753525, "global_step": 19101, "epoch": 530, "lr": 0.001} +{"train_loss": 0.034645359963178635, "train_loss_bc": 0.03463812172412872, "train_loss_llm": 0.0007237493991851807, "grad_norm": 0.05183497071266174, "global_step": 19102, "epoch": 530, "lr": 0.001} +{"train_loss": 0.035957466810941696, "train_loss_bc": 0.035957466810941696, "train_loss_llm": 0.0, "grad_norm": 0.06856382638216019, "global_step": 19103, "epoch": 530, "lr": 0.001} +{"train_loss": 0.035951048135757446, "train_loss_bc": 0.03593240678310394, "train_loss_llm": 0.0018639862537384033, "grad_norm": 0.08629734069108963, "global_step": 19104, "epoch": 530, "lr": 0.001} +{"train_loss": 0.0342220664024353, "train_loss_bc": 0.0342220664024353, "train_loss_llm": 0.0, "grad_norm": 0.0332508459687233, "global_step": 19105, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03575967624783516, "train_loss_bc": 0.03575967624783516, "train_loss_llm": 0.0, "grad_norm": 0.05521828681230545, "global_step": 19106, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03631051629781723, "train_loss_bc": 0.03631051629781723, "train_loss_llm": 0.0, "grad_norm": 0.06901876628398895, "global_step": 19107, "epoch": 530, "lr": 0.001} +{"train_loss": 0.038335222750902176, "train_loss_bc": 0.038335222750902176, "train_loss_llm": 0.0, "grad_norm": 0.0942142903804779, "global_step": 19108, "epoch": 530, "lr": 0.001} +{"train_loss": 0.033133234828710556, "train_loss_bc": 0.033133234828710556, "train_loss_llm": 0.0, "grad_norm": 0.0253733079880476, "global_step": 19109, "epoch": 530, "lr": 0.001} +{"train_loss": 0.02764221467077732, "train_loss_bc": 0.02763206884264946, "train_loss_llm": 0.001014620065689087, "grad_norm": 0.03620556369423866, "global_step": 19110, "epoch": 530, "lr": 0.001} +{"train_loss": 0.032963842153549194, "train_loss_bc": 0.032963842153549194, "train_loss_llm": 0.0, "grad_norm": 0.04728449136018753, "global_step": 19111, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03271203115582466, "train_loss_bc": 0.032709769904613495, "train_loss_llm": 0.00022614002227783203, "grad_norm": 0.06156691163778305, "global_step": 19112, "epoch": 530, "lr": 0.001} +{"train_loss": 0.032545801252126694, "train_loss_bc": 0.03254188224673271, "train_loss_llm": 0.00039196014404296875, "grad_norm": 0.025633612647652626, "global_step": 19113, "epoch": 530, "lr": 0.001} +{"train_loss": 0.032395508140325546, "train_loss_bc": 0.032395508140325546, "train_loss_llm": 0.0, "grad_norm": 0.03940943628549576, "global_step": 19114, "epoch": 530, "lr": 0.001} +{"train_loss": 0.03393805011486014, "train_loss_bc": 0.03675321489572525, "train_loss_llm": 0.0, "grad_norm": 0.05230114981532097, "global_step": 19115, "epoch": 530, "lr": 0.001, "train/cumulative_reward": 26.688689872209164, "train/mean_score": 7.607813939477703, "train/success_rate": 0.2, "test/cumulative_reward": 32.51144377619617, "test/mean_score": 5.504347167164559, "test/success_rate": 0.1, "val_loss": 0.07732018083333969, "train_action_mse_error": 0.03654084354639053} +{"train_loss": 0.040181998163461685, "train_loss_bc": 0.040181998163461685, "train_loss_llm": 0.0, "grad_norm": 0.05844055116176605, "global_step": 19116, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03325382620096207, "train_loss_bc": 0.033240363001823425, "train_loss_llm": 0.0013463795185089111, "grad_norm": 0.030696846544742584, "global_step": 19117, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03306414186954498, "train_loss_bc": 0.03304307162761688, "train_loss_llm": 0.0021068453788757324, "grad_norm": 0.047518957406282425, "global_step": 19118, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03387383744120598, "train_loss_bc": 0.03387383744120598, "train_loss_llm": 0.0, "grad_norm": 0.06599452346563339, "global_step": 19119, "epoch": 531, "lr": 0.001} +{"train_loss": 0.0333135612308979, "train_loss_bc": 0.0333135612308979, "train_loss_llm": 0.0, "grad_norm": 0.0788191556930542, "global_step": 19120, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03501896560192108, "train_loss_bc": 0.03501896560192108, "train_loss_llm": 0.0, "grad_norm": 0.03277812898159027, "global_step": 19121, "epoch": 531, "lr": 0.001} +{"train_loss": 0.032202042639255524, "train_loss_bc": 0.032202042639255524, "train_loss_llm": 0.0, "grad_norm": 0.05203525349497795, "global_step": 19122, "epoch": 531, "lr": 0.001} +{"train_loss": 0.02939768135547638, "train_loss_bc": 0.02939768135547638, "train_loss_llm": 0.0, "grad_norm": 0.067278653383255, "global_step": 19123, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03672364354133606, "train_loss_bc": 0.03671105578541756, "train_loss_llm": 0.0012589097023010254, "grad_norm": 0.0830686092376709, "global_step": 19124, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03289148956537247, "train_loss_bc": 0.03289148956537247, "train_loss_llm": 0.0, "grad_norm": 0.040033046156167984, "global_step": 19125, "epoch": 531, "lr": 0.001} +{"train_loss": 0.030895058065652847, "train_loss_bc": 0.030884545296430588, "train_loss_llm": 0.0010512173175811768, "grad_norm": 0.06043250858783722, "global_step": 19126, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03493017330765724, "train_loss_bc": 0.034913770854473114, "train_loss_llm": 0.0016403496265411377, "grad_norm": 0.08229635655879974, "global_step": 19127, "epoch": 531, "lr": 0.001} +{"train_loss": 0.031246302649378777, "train_loss_bc": 0.031246302649378777, "train_loss_llm": 0.0, "grad_norm": 0.10656780004501343, "global_step": 19128, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03462210297584534, "train_loss_bc": 0.03462210297584534, "train_loss_llm": 0.0, "grad_norm": 0.0274106003344059, "global_step": 19129, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03738686442375183, "train_loss_bc": 0.03738686442375183, "train_loss_llm": 0.0, "grad_norm": 0.03495148569345474, "global_step": 19130, "epoch": 531, "lr": 0.001} +{"train_loss": 0.041788436472415924, "train_loss_bc": 0.041788436472415924, "train_loss_llm": 0.0, "grad_norm": 0.06094643101096153, "global_step": 19131, "epoch": 531, "lr": 0.001} +{"train_loss": 0.034108709543943405, "train_loss_bc": 0.034098587930202484, "train_loss_llm": 0.0010122954845428467, "grad_norm": 0.07773374766111374, "global_step": 19132, "epoch": 531, "lr": 0.001} +{"train_loss": 0.02920696511864662, "train_loss_bc": 0.02920696511864662, "train_loss_llm": 0.0, "grad_norm": 0.022133545950055122, "global_step": 19133, "epoch": 531, "lr": 0.001} +{"train_loss": 0.031960565596818924, "train_loss_bc": 0.031960565596818924, "train_loss_llm": 0.0, "grad_norm": 0.03670945763587952, "global_step": 19134, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03271833807229996, "train_loss_bc": 0.03271833807229996, "train_loss_llm": 0.0, "grad_norm": 0.0494181290268898, "global_step": 19135, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03524451330304146, "train_loss_bc": 0.03524451330304146, "train_loss_llm": 0.0, "grad_norm": 0.06605106592178345, "global_step": 19136, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03819561377167702, "train_loss_bc": 0.03819561377167702, "train_loss_llm": 0.0, "grad_norm": 0.03281863033771515, "global_step": 19137, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03283515200018883, "train_loss_bc": 0.03283515200018883, "train_loss_llm": 0.0, "grad_norm": 0.046979282051324844, "global_step": 19138, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03348998352885246, "train_loss_bc": 0.03348998352885246, "train_loss_llm": 0.0, "grad_norm": 0.060912489891052246, "global_step": 19139, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03274235501885414, "train_loss_bc": 0.03274235501885414, "train_loss_llm": 0.0, "grad_norm": 0.06781569868326187, "global_step": 19140, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03249727189540863, "train_loss_bc": 0.03249727189540863, "train_loss_llm": 0.0, "grad_norm": 0.024014117196202278, "global_step": 19141, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03366314619779587, "train_loss_bc": 0.0336572490632534, "train_loss_llm": 0.0005897879600524902, "grad_norm": 0.034001514315605164, "global_step": 19142, "epoch": 531, "lr": 0.001} +{"train_loss": 0.029399342834949493, "train_loss_bc": 0.029399342834949493, "train_loss_llm": 0.0, "grad_norm": 0.03800510615110397, "global_step": 19143, "epoch": 531, "lr": 0.001} +{"train_loss": 0.031022267416119576, "train_loss_bc": 0.031022267416119576, "train_loss_llm": 0.0, "grad_norm": 0.05239808186888695, "global_step": 19144, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03492132946848869, "train_loss_bc": 0.03492132946848869, "train_loss_llm": 0.0, "grad_norm": 0.029765095561742783, "global_step": 19145, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03155829757452011, "train_loss_bc": 0.03155829757452011, "train_loss_llm": 0.0, "grad_norm": 0.0470380075275898, "global_step": 19146, "epoch": 531, "lr": 0.001} +{"train_loss": 0.0380648709833622, "train_loss_bc": 0.0380648709833622, "train_loss_llm": 0.0, "grad_norm": 0.06049551069736481, "global_step": 19147, "epoch": 531, "lr": 0.001} +{"train_loss": 0.030907589942216873, "train_loss_bc": 0.030907589942216873, "train_loss_llm": 0.0, "grad_norm": 0.08279966562986374, "global_step": 19148, "epoch": 531, "lr": 0.001} +{"train_loss": 0.032398007810115814, "train_loss_bc": 0.032398007810115814, "train_loss_llm": 0.0, "grad_norm": 0.03831927478313446, "global_step": 19149, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03799177333712578, "train_loss_bc": 0.03799067810177803, "train_loss_llm": 0.00010955333709716797, "grad_norm": 0.07229820638895035, "global_step": 19150, "epoch": 531, "lr": 0.001} +{"train_loss": 0.03380823269900349, "train_loss_bc": 0.03338015824556351, "train_loss_llm": 0.0, "grad_norm": 0.09935179352760315, "global_step": 19151, "epoch": 531, "lr": 0.001, "val_loss": 0.0775613933801651} +{"train_loss": 0.03636769950389862, "train_loss_bc": 0.03636769950389862, "train_loss_llm": 0.0, "grad_norm": 0.12215545028448105, "global_step": 19152, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03277513012290001, "train_loss_bc": 0.03277513012290001, "train_loss_llm": 0.0, "grad_norm": 0.026584897190332413, "global_step": 19153, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03100566379725933, "train_loss_bc": 0.03100566379725933, "train_loss_llm": 0.0, "grad_norm": 0.04143231734633446, "global_step": 19154, "epoch": 532, "lr": 0.001} +{"train_loss": 0.028099054470658302, "train_loss_bc": 0.028099054470658302, "train_loss_llm": 0.0, "grad_norm": 0.053578585386276245, "global_step": 19155, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03266477957367897, "train_loss_bc": 0.03266477957367897, "train_loss_llm": 0.0, "grad_norm": 0.07027649134397507, "global_step": 19156, "epoch": 532, "lr": 0.001} +{"train_loss": 0.04023697227239609, "train_loss_bc": 0.04023697227239609, "train_loss_llm": 0.0, "grad_norm": 0.029411617666482925, "global_step": 19157, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03399326279759407, "train_loss_bc": 0.03399326279759407, "train_loss_llm": 0.0, "grad_norm": 0.03600481152534485, "global_step": 19158, "epoch": 532, "lr": 0.001} +{"train_loss": 0.02816878631711006, "train_loss_bc": 0.02816878631711006, "train_loss_llm": 0.0, "grad_norm": 0.04710850492119789, "global_step": 19159, "epoch": 532, "lr": 0.001} +{"train_loss": 0.030360279604792595, "train_loss_bc": 0.030360279604792595, "train_loss_llm": 0.0, "grad_norm": 0.0519375279545784, "global_step": 19160, "epoch": 532, "lr": 0.001} +{"train_loss": 0.030773192644119263, "train_loss_bc": 0.030773192644119263, "train_loss_llm": 0.0, "grad_norm": 0.019938278943300247, "global_step": 19161, "epoch": 532, "lr": 0.001} +{"train_loss": 0.029961122199892998, "train_loss_bc": 0.02995418757200241, "train_loss_llm": 0.0006935000419616699, "grad_norm": 0.032665833830833435, "global_step": 19162, "epoch": 532, "lr": 0.001} +{"train_loss": 0.0387398861348629, "train_loss_bc": 0.038711097091436386, "train_loss_llm": 0.002878725528717041, "grad_norm": 0.05655808374285698, "global_step": 19163, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03469518944621086, "train_loss_bc": 0.03469518944621086, "train_loss_llm": 0.0, "grad_norm": 0.076045460999012, "global_step": 19164, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03176514804363251, "train_loss_bc": 0.03176514804363251, "train_loss_llm": 0.0, "grad_norm": 0.027462324127554893, "global_step": 19165, "epoch": 532, "lr": 0.001} +{"train_loss": 0.034741517156362534, "train_loss_bc": 0.034740742295980453, "train_loss_llm": 7.739663124084473e-05, "grad_norm": 0.04734135791659355, "global_step": 19166, "epoch": 532, "lr": 0.001} +{"train_loss": 0.04154496267437935, "train_loss_bc": 0.04154496267437935, "train_loss_llm": 0.0, "grad_norm": 0.06380896270275116, "global_step": 19167, "epoch": 532, "lr": 0.001} +{"train_loss": 0.030071565881371498, "train_loss_bc": 0.03006852976977825, "train_loss_llm": 0.00030365586280822754, "grad_norm": 0.07657387852668762, "global_step": 19168, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03313790261745453, "train_loss_bc": 0.03313539922237396, "train_loss_llm": 0.0002503395080566406, "grad_norm": 0.03607304394245148, "global_step": 19169, "epoch": 532, "lr": 0.001} +{"train_loss": 0.0353793241083622, "train_loss_bc": 0.03537152707576752, "train_loss_llm": 0.0007798373699188232, "grad_norm": 0.055981531739234924, "global_step": 19170, "epoch": 532, "lr": 0.001} +{"train_loss": 0.033719174563884735, "train_loss_bc": 0.033719174563884735, "train_loss_llm": 0.0, "grad_norm": 0.06613963097333908, "global_step": 19171, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03160763159394264, "train_loss_bc": 0.03160763159394264, "train_loss_llm": 0.0, "grad_norm": 0.08512869477272034, "global_step": 19172, "epoch": 532, "lr": 0.001} +{"train_loss": 0.034477848559617996, "train_loss_bc": 0.034477848559617996, "train_loss_llm": 0.0, "grad_norm": 0.025414705276489258, "global_step": 19173, "epoch": 532, "lr": 0.001} +{"train_loss": 0.032872043550014496, "train_loss_bc": 0.032872043550014496, "train_loss_llm": 0.0, "grad_norm": 0.04810555651783943, "global_step": 19174, "epoch": 532, "lr": 0.001} +{"train_loss": 0.030499164015054703, "train_loss_bc": 0.030499164015054703, "train_loss_llm": 0.0, "grad_norm": 0.056423891335725784, "global_step": 19175, "epoch": 532, "lr": 0.001} +{"train_loss": 0.040162645280361176, "train_loss_bc": 0.040158696472644806, "train_loss_llm": 0.00039499998092651367, "grad_norm": 0.06762383133172989, "global_step": 19176, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03546895459294319, "train_loss_bc": 0.03546895459294319, "train_loss_llm": 0.0, "grad_norm": 0.029345879331231117, "global_step": 19177, "epoch": 532, "lr": 0.001} +{"train_loss": 0.035909321159124374, "train_loss_bc": 0.035909321159124374, "train_loss_llm": 0.0, "grad_norm": 0.0483478344976902, "global_step": 19178, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03154542297124863, "train_loss_bc": 0.03152793273329735, "train_loss_llm": 0.001748889684677124, "grad_norm": 0.061291929334402084, "global_step": 19179, "epoch": 532, "lr": 0.001} +{"train_loss": 0.028047097846865654, "train_loss_bc": 0.028047097846865654, "train_loss_llm": 0.0, "grad_norm": 0.07105524092912674, "global_step": 19180, "epoch": 532, "lr": 0.001} +{"train_loss": 0.027968307957053185, "train_loss_bc": 0.027963930740952492, "train_loss_llm": 0.00043779611587524414, "grad_norm": 0.022363023832440376, "global_step": 19181, "epoch": 532, "lr": 0.001} +{"train_loss": 0.029574941843748093, "train_loss_bc": 0.029574941843748093, "train_loss_llm": 0.0, "grad_norm": 0.03380163013935089, "global_step": 19182, "epoch": 532, "lr": 0.001} +{"train_loss": 0.036216992884874344, "train_loss_bc": 0.03620195761322975, "train_loss_llm": 0.0015034973621368408, "grad_norm": 0.04348073899745941, "global_step": 19183, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03716828301548958, "train_loss_bc": 0.03716828301548958, "train_loss_llm": 0.0, "grad_norm": 0.053875524550676346, "global_step": 19184, "epoch": 532, "lr": 0.001} +{"train_loss": 0.04140383005142212, "train_loss_bc": 0.04140383005142212, "train_loss_llm": 0.0, "grad_norm": 0.03906554728746414, "global_step": 19185, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03307937830686569, "train_loss_bc": 0.03306092694401741, "train_loss_llm": 0.0018450617790222168, "grad_norm": 0.04727218672633171, "global_step": 19186, "epoch": 532, "lr": 0.001} +{"train_loss": 0.03375564293108053, "train_loss_bc": 0.0409875251352787, "train_loss_llm": 0.0013141632080078125, "grad_norm": 0.061161261051893234, "global_step": 19187, "epoch": 532, "lr": 0.001, "val_loss": 0.07633661478757858} +{"train_loss": 0.04094787687063217, "train_loss_bc": 0.04094787687063217, "train_loss_llm": 0.0, "grad_norm": 0.07326439768075943, "global_step": 19188, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03400026261806488, "train_loss_bc": 0.03400026261806488, "train_loss_llm": 0.0, "grad_norm": 0.027647508308291435, "global_step": 19189, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03703103959560394, "train_loss_bc": 0.03702019900083542, "train_loss_llm": 0.0010841190814971924, "grad_norm": 0.04124998673796654, "global_step": 19190, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03542068228125572, "train_loss_bc": 0.03542068228125572, "train_loss_llm": 0.0, "grad_norm": 0.05765147879719734, "global_step": 19191, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03251716494560242, "train_loss_bc": 0.03251716494560242, "train_loss_llm": 0.0, "grad_norm": 0.07298052310943604, "global_step": 19192, "epoch": 533, "lr": 0.001} +{"train_loss": 0.034474678337574005, "train_loss_bc": 0.034474678337574005, "train_loss_llm": 0.0, "grad_norm": 0.030423762276768684, "global_step": 19193, "epoch": 533, "lr": 0.001} +{"train_loss": 0.033937226980924606, "train_loss_bc": 0.033937226980924606, "train_loss_llm": 0.0, "grad_norm": 0.04374818131327629, "global_step": 19194, "epoch": 533, "lr": 0.001} +{"train_loss": 0.027676831930875778, "train_loss_bc": 0.027676831930875778, "train_loss_llm": 0.0, "grad_norm": 0.059937190264463425, "global_step": 19195, "epoch": 533, "lr": 0.001} +{"train_loss": 0.023332208395004272, "train_loss_bc": 0.023331671953201294, "train_loss_llm": 5.367398262023926e-05, "grad_norm": 0.07863257825374603, "global_step": 19196, "epoch": 533, "lr": 0.001} +{"train_loss": 0.038340840488672256, "train_loss_bc": 0.038311801850795746, "train_loss_llm": 0.0029036998748779297, "grad_norm": 0.02509990707039833, "global_step": 19197, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03203997015953064, "train_loss_bc": 0.032036200165748596, "train_loss_llm": 0.0003768503665924072, "grad_norm": 0.033900752663612366, "global_step": 19198, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03644375130534172, "train_loss_bc": 0.036425381898880005, "train_loss_llm": 0.0018367767333984375, "grad_norm": 0.049272552132606506, "global_step": 19199, "epoch": 533, "lr": 0.001} +{"train_loss": 0.037293680012226105, "train_loss_bc": 0.037293680012226105, "train_loss_llm": 0.0, "grad_norm": 0.06727355718612671, "global_step": 19200, "epoch": 533, "lr": 0.001} +{"train_loss": 0.0272742360830307, "train_loss_bc": 0.0272742360830307, "train_loss_llm": 0.0, "grad_norm": 0.022820686921477318, "global_step": 19201, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03566254675388336, "train_loss_bc": 0.03566254675388336, "train_loss_llm": 0.0, "grad_norm": 0.034432701766490936, "global_step": 19202, "epoch": 533, "lr": 0.001} +{"train_loss": 0.030705496668815613, "train_loss_bc": 0.030705496668815613, "train_loss_llm": 0.0, "grad_norm": 0.049929194152355194, "global_step": 19203, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03521936014294624, "train_loss_bc": 0.03521936014294624, "train_loss_llm": 0.0, "grad_norm": 0.05715996399521828, "global_step": 19204, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03626479208469391, "train_loss_bc": 0.03626479208469391, "train_loss_llm": 0.0, "grad_norm": 0.03288598358631134, "global_step": 19205, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03267153352499008, "train_loss_bc": 0.03267153352499008, "train_loss_llm": 0.0, "grad_norm": 0.0489116795361042, "global_step": 19206, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03268193081021309, "train_loss_bc": 0.03268193081021309, "train_loss_llm": 0.0, "grad_norm": 0.06165130063891411, "global_step": 19207, "epoch": 533, "lr": 0.001} +{"train_loss": 0.030469326302409172, "train_loss_bc": 0.030463790521025658, "train_loss_llm": 0.0005535483360290527, "grad_norm": 0.08065223693847656, "global_step": 19208, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03281555697321892, "train_loss_bc": 0.03281010314822197, "train_loss_llm": 0.000545501708984375, "grad_norm": 0.026409810408949852, "global_step": 19209, "epoch": 533, "lr": 0.001} +{"train_loss": 0.039481014013290405, "train_loss_bc": 0.039481014013290405, "train_loss_llm": 0.0, "grad_norm": 0.040009498596191406, "global_step": 19210, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03054071217775345, "train_loss_bc": 0.030536534264683723, "train_loss_llm": 0.0004177987575531006, "grad_norm": 0.05265335366129875, "global_step": 19211, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03330343961715698, "train_loss_bc": 0.03330343961715698, "train_loss_llm": 0.0, "grad_norm": 0.07243721932172775, "global_step": 19212, "epoch": 533, "lr": 0.001} +{"train_loss": 0.031705789268016815, "train_loss_bc": 0.031705789268016815, "train_loss_llm": 0.0, "grad_norm": 0.02934115193784237, "global_step": 19213, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03787173703312874, "train_loss_bc": 0.03787173703312874, "train_loss_llm": 0.0, "grad_norm": 0.05186570808291435, "global_step": 19214, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03726942837238312, "train_loss_bc": 0.03726942837238312, "train_loss_llm": 0.0, "grad_norm": 0.08739671111106873, "global_step": 19215, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03443208709359169, "train_loss_bc": 0.03443208709359169, "train_loss_llm": 0.0, "grad_norm": 0.11230301111936569, "global_step": 19216, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03515166789293289, "train_loss_bc": 0.03512578830122948, "train_loss_llm": 0.0025880932807922363, "grad_norm": 0.027362501248717308, "global_step": 19217, "epoch": 533, "lr": 0.001} +{"train_loss": 0.032495759427547455, "train_loss_bc": 0.032484572380781174, "train_loss_llm": 0.0011188387870788574, "grad_norm": 0.05050908029079437, "global_step": 19218, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03647429868578911, "train_loss_bc": 0.03647429868578911, "train_loss_llm": 0.0, "grad_norm": 0.0760992169380188, "global_step": 19219, "epoch": 533, "lr": 0.001} +{"train_loss": 0.031026197597384453, "train_loss_bc": 0.031025972217321396, "train_loss_llm": 2.2470951080322266e-05, "grad_norm": 0.08215238898992538, "global_step": 19220, "epoch": 533, "lr": 0.001} +{"train_loss": 0.02913215570151806, "train_loss_bc": 0.029131243005394936, "train_loss_llm": 9.128451347351074e-05, "grad_norm": 0.02741750329732895, "global_step": 19221, "epoch": 533, "lr": 0.001} +{"train_loss": 0.03575786575675011, "train_loss_bc": 0.03575786575675011, "train_loss_llm": 0.0, "grad_norm": 0.04565049707889557, "global_step": 19222, "epoch": 533, "lr": 0.001} +{"train_loss": 0.033620100313176714, "train_loss_bc": 0.028460465371608734, "train_loss_llm": 0.0, "grad_norm": 0.06502583622932434, "global_step": 19223, "epoch": 533, "lr": 0.001, "val_loss": 0.07766563445329666} +{"train_loss": 0.03262300789356232, "train_loss_bc": 0.03261321783065796, "train_loss_llm": 0.0009788274765014648, "grad_norm": 0.07965634763240814, "global_step": 19224, "epoch": 534, "lr": 0.001} +{"train_loss": 0.033326536417007446, "train_loss_bc": 0.033326536417007446, "train_loss_llm": 0.0, "grad_norm": 0.024739759042859077, "global_step": 19225, "epoch": 534, "lr": 0.001} +{"train_loss": 0.0352834090590477, "train_loss_bc": 0.0352834090590477, "train_loss_llm": 0.0, "grad_norm": 0.04925544187426567, "global_step": 19226, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03167884796857834, "train_loss_bc": 0.03167884796857834, "train_loss_llm": 0.0, "grad_norm": 0.06877035647630692, "global_step": 19227, "epoch": 534, "lr": 0.001} +{"train_loss": 0.031364765018224716, "train_loss_bc": 0.031364765018224716, "train_loss_llm": 0.0, "grad_norm": 0.08463985472917557, "global_step": 19228, "epoch": 534, "lr": 0.001} +{"train_loss": 0.032708559185266495, "train_loss_bc": 0.032708559185266495, "train_loss_llm": 0.0, "grad_norm": 0.02872556447982788, "global_step": 19229, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03430880978703499, "train_loss_bc": 0.03430880978703499, "train_loss_llm": 0.0, "grad_norm": 0.039880603551864624, "global_step": 19230, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03623836115002632, "train_loss_bc": 0.03623836115002632, "train_loss_llm": 0.0, "grad_norm": 0.05361071601510048, "global_step": 19231, "epoch": 534, "lr": 0.001} +{"train_loss": 0.02821424789726734, "train_loss_bc": 0.02821424789726734, "train_loss_llm": 0.0, "grad_norm": 0.06648526340723038, "global_step": 19232, "epoch": 534, "lr": 0.001} +{"train_loss": 0.030962079763412476, "train_loss_bc": 0.030955303460359573, "train_loss_llm": 0.0006777048110961914, "grad_norm": 0.031993888318538666, "global_step": 19233, "epoch": 534, "lr": 0.001} +{"train_loss": 0.029583130031824112, "train_loss_bc": 0.029583130031824112, "train_loss_llm": 0.0, "grad_norm": 0.047853924334049225, "global_step": 19234, "epoch": 534, "lr": 0.001} +{"train_loss": 0.0370364785194397, "train_loss_bc": 0.0370311513543129, "train_loss_llm": 0.0005328655242919922, "grad_norm": 0.06793699413537979, "global_step": 19235, "epoch": 534, "lr": 0.001} +{"train_loss": 0.034081656485795975, "train_loss_bc": 0.034081656485795975, "train_loss_llm": 0.0, "grad_norm": 0.08367703855037689, "global_step": 19236, "epoch": 534, "lr": 0.001} +{"train_loss": 0.032842423766851425, "train_loss_bc": 0.032826077193021774, "train_loss_llm": 0.001634836196899414, "grad_norm": 0.029927369207143784, "global_step": 19237, "epoch": 534, "lr": 0.001} +{"train_loss": 0.029955459758639336, "train_loss_bc": 0.029955459758639336, "train_loss_llm": 0.0, "grad_norm": 0.05236838385462761, "global_step": 19238, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03672054782509804, "train_loss_bc": 0.03672054782509804, "train_loss_llm": 0.0, "grad_norm": 0.06887121498584747, "global_step": 19239, "epoch": 534, "lr": 0.001} +{"train_loss": 0.0392642579972744, "train_loss_bc": 0.039251383394002914, "train_loss_llm": 0.001287609338760376, "grad_norm": 0.08473018556833267, "global_step": 19240, "epoch": 534, "lr": 0.001} +{"train_loss": 0.029016541317105293, "train_loss_bc": 0.029016416519880295, "train_loss_llm": 1.2487173080444336e-05, "grad_norm": 0.026084035634994507, "global_step": 19241, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03612422198057175, "train_loss_bc": 0.03612193465232849, "train_loss_llm": 0.00022867321968078613, "grad_norm": 0.04812919721007347, "global_step": 19242, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03154458478093147, "train_loss_bc": 0.03154347091913223, "train_loss_llm": 0.0001112520694732666, "grad_norm": 0.06101055443286896, "global_step": 19243, "epoch": 534, "lr": 0.001} +{"train_loss": 0.032477863132953644, "train_loss_bc": 0.032477863132953644, "train_loss_llm": 0.0, "grad_norm": 0.07448846101760864, "global_step": 19244, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03393957018852234, "train_loss_bc": 0.03393957018852234, "train_loss_llm": 0.0, "grad_norm": 0.02415839396417141, "global_step": 19245, "epoch": 534, "lr": 0.001} +{"train_loss": 0.04090753570199013, "train_loss_bc": 0.04090753570199013, "train_loss_llm": 0.0, "grad_norm": 0.03650125861167908, "global_step": 19246, "epoch": 534, "lr": 0.001} +{"train_loss": 0.0425337553024292, "train_loss_bc": 0.0425337553024292, "train_loss_llm": 0.0, "grad_norm": 0.05062270909547806, "global_step": 19247, "epoch": 534, "lr": 0.001} +{"train_loss": 0.033603888005018234, "train_loss_bc": 0.03359919786453247, "train_loss_llm": 0.0004691183567047119, "grad_norm": 0.06688375771045685, "global_step": 19248, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03860205039381981, "train_loss_bc": 0.03860205039381981, "train_loss_llm": 0.0, "grad_norm": 0.03100564330816269, "global_step": 19249, "epoch": 534, "lr": 0.001} +{"train_loss": 0.027290932834148407, "train_loss_bc": 0.027290932834148407, "train_loss_llm": 0.0, "grad_norm": 0.03781527653336525, "global_step": 19250, "epoch": 534, "lr": 0.001} +{"train_loss": 0.0323750376701355, "train_loss_bc": 0.0323750376701355, "train_loss_llm": 0.0, "grad_norm": 0.0454065203666687, "global_step": 19251, "epoch": 534, "lr": 0.001} +{"train_loss": 0.032941412180662155, "train_loss_bc": 0.032938096672296524, "train_loss_llm": 0.00033152103424072266, "grad_norm": 0.053785622119903564, "global_step": 19252, "epoch": 534, "lr": 0.001} +{"train_loss": 0.031583912670612335, "train_loss_bc": 0.031583912670612335, "train_loss_llm": 0.0, "grad_norm": 0.03424955531954765, "global_step": 19253, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03740791976451874, "train_loss_bc": 0.03740791976451874, "train_loss_llm": 0.0, "grad_norm": 0.04762282967567444, "global_step": 19254, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03597161918878555, "train_loss_bc": 0.03597161918878555, "train_loss_llm": 0.0, "grad_norm": 0.05172644928097725, "global_step": 19255, "epoch": 534, "lr": 0.001} +{"train_loss": 0.031484536826610565, "train_loss_bc": 0.031484536826610565, "train_loss_llm": 0.0, "grad_norm": 0.06429706513881683, "global_step": 19256, "epoch": 534, "lr": 0.001} +{"train_loss": 0.02630038559436798, "train_loss_bc": 0.02630038559436798, "train_loss_llm": 0.0, "grad_norm": 0.023306675255298615, "global_step": 19257, "epoch": 534, "lr": 0.001} +{"train_loss": 0.03364313766360283, "train_loss_bc": 0.03364313766360283, "train_loss_llm": 0.0, "grad_norm": 0.040658533573150635, "global_step": 19258, "epoch": 534, "lr": 0.001} +{"train_loss": 0.033558942636268005, "train_loss_bc": 0.03416711837053299, "train_loss_llm": 0.0013331770896911621, "grad_norm": 0.05330304056406021, "global_step": 19259, "epoch": 534, "lr": 0.001, "val_loss": 0.0766187310218811} +{"train_loss": 0.033792704343795776, "train_loss_bc": 0.033792704343795776, "train_loss_llm": 0.0, "grad_norm": 0.06726895272731781, "global_step": 19260, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03127622604370117, "train_loss_bc": 0.03127622604370117, "train_loss_llm": 0.0, "grad_norm": 0.02469899132847786, "global_step": 19261, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03032633662223816, "train_loss_bc": 0.030322160571813583, "train_loss_llm": 0.0004176795482635498, "grad_norm": 0.034536153078079224, "global_step": 19262, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03663712367415428, "train_loss_bc": 0.03663712367415428, "train_loss_llm": 0.0, "grad_norm": 0.05279494449496269, "global_step": 19263, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03890835866332054, "train_loss_bc": 0.03890835866332054, "train_loss_llm": 0.0, "grad_norm": 0.06363552063703537, "global_step": 19264, "epoch": 535, "lr": 0.001} +{"train_loss": 0.036186572164297104, "train_loss_bc": 0.036186572164297104, "train_loss_llm": 0.0, "grad_norm": 0.029654407873749733, "global_step": 19265, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03744398057460785, "train_loss_bc": 0.03744398057460785, "train_loss_llm": 0.0, "grad_norm": 0.03880512714385986, "global_step": 19266, "epoch": 535, "lr": 0.001} +{"train_loss": 0.025004694238305092, "train_loss_bc": 0.025004694238305092, "train_loss_llm": 0.0, "grad_norm": 0.04452791437506676, "global_step": 19267, "epoch": 535, "lr": 0.001} +{"train_loss": 0.038500793278217316, "train_loss_bc": 0.038500793278217316, "train_loss_llm": 0.0, "grad_norm": 0.0506775863468647, "global_step": 19268, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03115672431886196, "train_loss_bc": 0.03115672431886196, "train_loss_llm": 0.0, "grad_norm": 0.02773222140967846, "global_step": 19269, "epoch": 535, "lr": 0.001} +{"train_loss": 0.026803620159626007, "train_loss_bc": 0.026803620159626007, "train_loss_llm": 0.0, "grad_norm": 0.03978889063000679, "global_step": 19270, "epoch": 535, "lr": 0.001} +{"train_loss": 0.040933676064014435, "train_loss_bc": 0.04091477766633034, "train_loss_llm": 0.0018900036811828613, "grad_norm": 0.047661956399679184, "global_step": 19271, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03292608633637428, "train_loss_bc": 0.032917942851781845, "train_loss_llm": 0.0008143484592437744, "grad_norm": 0.06603124737739563, "global_step": 19272, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03508466109633446, "train_loss_bc": 0.03508466109633446, "train_loss_llm": 0.0, "grad_norm": 0.026239529252052307, "global_step": 19273, "epoch": 535, "lr": 0.001} +{"train_loss": 0.030735939741134644, "train_loss_bc": 0.030735939741134644, "train_loss_llm": 0.0, "grad_norm": 0.043523091822862625, "global_step": 19274, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03653726726770401, "train_loss_bc": 0.03653726726770401, "train_loss_llm": 0.0, "grad_norm": 0.052236977964639664, "global_step": 19275, "epoch": 535, "lr": 0.001} +{"train_loss": 0.038766179233789444, "train_loss_bc": 0.038766179233789444, "train_loss_llm": 0.0, "grad_norm": 0.06607681512832642, "global_step": 19276, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03653180971741676, "train_loss_bc": 0.03653180971741676, "train_loss_llm": 0.0, "grad_norm": 0.03006688691675663, "global_step": 19277, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03882816806435585, "train_loss_bc": 0.03882816806435585, "train_loss_llm": 0.0, "grad_norm": 0.04684208333492279, "global_step": 19278, "epoch": 535, "lr": 0.001} +{"train_loss": 0.02731468342244625, "train_loss_bc": 0.02731468342244625, "train_loss_llm": 0.0, "grad_norm": 0.05695652961730957, "global_step": 19279, "epoch": 535, "lr": 0.001} +{"train_loss": 0.027192512527108192, "train_loss_bc": 0.027192512527108192, "train_loss_llm": 0.0, "grad_norm": 0.0649474635720253, "global_step": 19280, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03513909503817558, "train_loss_bc": 0.03513909503817558, "train_loss_llm": 0.0, "grad_norm": 0.028496403247117996, "global_step": 19281, "epoch": 535, "lr": 0.001} +{"train_loss": 0.04380522668361664, "train_loss_bc": 0.04380522668361664, "train_loss_llm": 0.0, "grad_norm": 0.04238640516996384, "global_step": 19282, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03328186646103859, "train_loss_bc": 0.03327937051653862, "train_loss_llm": 0.00024962425231933594, "grad_norm": 0.05658143386244774, "global_step": 19283, "epoch": 535, "lr": 0.001} +{"train_loss": 0.031642455607652664, "train_loss_bc": 0.031642455607652664, "train_loss_llm": 0.0, "grad_norm": 0.07557043433189392, "global_step": 19284, "epoch": 535, "lr": 0.001} +{"train_loss": 0.035218045115470886, "train_loss_bc": 0.035218045115470886, "train_loss_llm": 0.0, "grad_norm": 0.025412390008568764, "global_step": 19285, "epoch": 535, "lr": 0.001} +{"train_loss": 0.02924143709242344, "train_loss_bc": 0.029232000932097435, "train_loss_llm": 0.0009436309337615967, "grad_norm": 0.043898895382881165, "global_step": 19286, "epoch": 535, "lr": 0.001} +{"train_loss": 0.02987513691186905, "train_loss_bc": 0.029857710003852844, "train_loss_llm": 0.001742631196975708, "grad_norm": 0.06547080725431442, "global_step": 19287, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03024771809577942, "train_loss_bc": 0.03024771809577942, "train_loss_llm": 0.0, "grad_norm": 0.07862221449613571, "global_step": 19288, "epoch": 535, "lr": 0.001} +{"train_loss": 0.028670068830251694, "train_loss_bc": 0.028670068830251694, "train_loss_llm": 0.0, "grad_norm": 0.023379217833280563, "global_step": 19289, "epoch": 535, "lr": 0.001} +{"train_loss": 0.02972184121608734, "train_loss_bc": 0.029716962948441505, "train_loss_llm": 0.00048789381980895996, "grad_norm": 0.04498002678155899, "global_step": 19290, "epoch": 535, "lr": 0.001} +{"train_loss": 0.0388018861413002, "train_loss_bc": 0.0388018861413002, "train_loss_llm": 0.0, "grad_norm": 0.05807669833302498, "global_step": 19291, "epoch": 535, "lr": 0.001} +{"train_loss": 0.03024255856871605, "train_loss_bc": 0.030237160623073578, "train_loss_llm": 0.0005398392677307129, "grad_norm": 0.07117585092782974, "global_step": 19292, "epoch": 535, "lr": 0.001} +{"train_loss": 0.028789645060896873, "train_loss_bc": 0.028789645060896873, "train_loss_llm": 0.0, "grad_norm": 0.028916260227560997, "global_step": 19293, "epoch": 535, "lr": 0.001} +{"train_loss": 0.036546606570482254, "train_loss_bc": 0.036546606570482254, "train_loss_llm": 0.0, "grad_norm": 0.04580572247505188, "global_step": 19294, "epoch": 535, "lr": 0.001} +{"train_loss": 0.033447711211111814, "train_loss_bc": 0.03199752792716026, "train_loss_llm": 0.0008369088172912598, "grad_norm": 0.06627613306045532, "global_step": 19295, "epoch": 535, "lr": 0.001, "train/cumulative_reward": 85.0694195272022, "train/mean_score": 15.8185424536884, "train/success_rate": 0.4, "test/cumulative_reward": 39.4690134071177, "test/mean_score": 8.379585969725712, "test/success_rate": 0.2, "val_loss": 0.07678446173667908, "train_action_mse_error": 0.0364421084523201} +{"train_loss": 0.027283422648906708, "train_loss_bc": 0.027283422648906708, "train_loss_llm": 0.0, "grad_norm": 0.07791152596473694, "global_step": 19296, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03193410113453865, "train_loss_bc": 0.03193410113453865, "train_loss_llm": 0.0, "grad_norm": 0.028313087299466133, "global_step": 19297, "epoch": 536, "lr": 0.001} +{"train_loss": 0.028904441744089127, "train_loss_bc": 0.028904441744089127, "train_loss_llm": 0.0, "grad_norm": 0.04733804240822792, "global_step": 19298, "epoch": 536, "lr": 0.001} +{"train_loss": 0.031239550560712814, "train_loss_bc": 0.031239550560712814, "train_loss_llm": 0.0, "grad_norm": 0.06155761703848839, "global_step": 19299, "epoch": 536, "lr": 0.001} +{"train_loss": 0.035535767674446106, "train_loss_bc": 0.035535767674446106, "train_loss_llm": 0.0, "grad_norm": 0.08005103468894958, "global_step": 19300, "epoch": 536, "lr": 0.001} +{"train_loss": 0.04359470680356026, "train_loss_bc": 0.04359470680356026, "train_loss_llm": 0.0, "grad_norm": 0.02729114331305027, "global_step": 19301, "epoch": 536, "lr": 0.001} +{"train_loss": 0.02885020710527897, "train_loss_bc": 0.028846189379692078, "train_loss_llm": 0.0004017651081085205, "grad_norm": 0.04077095165848732, "global_step": 19302, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03444759175181389, "train_loss_bc": 0.03444759175181389, "train_loss_llm": 0.0, "grad_norm": 0.044268131256103516, "global_step": 19303, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03478403016924858, "train_loss_bc": 0.03478403016924858, "train_loss_llm": 0.0, "grad_norm": 0.05320388078689575, "global_step": 19304, "epoch": 536, "lr": 0.001} +{"train_loss": 0.0343637652695179, "train_loss_bc": 0.0343637652695179, "train_loss_llm": 0.0, "grad_norm": 0.018840020522475243, "global_step": 19305, "epoch": 536, "lr": 0.001} +{"train_loss": 0.029060203582048416, "train_loss_bc": 0.029059862717986107, "train_loss_llm": 3.403425216674805e-05, "grad_norm": 0.03330431878566742, "global_step": 19306, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03420299291610718, "train_loss_bc": 0.034189146012067795, "train_loss_llm": 0.0013846158981323242, "grad_norm": 0.041729722172021866, "global_step": 19307, "epoch": 536, "lr": 0.001} +{"train_loss": 0.0373978465795517, "train_loss_bc": 0.0373978465795517, "train_loss_llm": 0.0, "grad_norm": 0.0607026182115078, "global_step": 19308, "epoch": 536, "lr": 0.001} +{"train_loss": 0.02995407022535801, "train_loss_bc": 0.02994448132812977, "train_loss_llm": 0.0009589791297912598, "grad_norm": 0.024401972070336342, "global_step": 19309, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03379882127046585, "train_loss_bc": 0.03379882127046585, "train_loss_llm": 0.0, "grad_norm": 0.04298238083720207, "global_step": 19310, "epoch": 536, "lr": 0.001} +{"train_loss": 0.031184742227196693, "train_loss_bc": 0.031184742227196693, "train_loss_llm": 0.0, "grad_norm": 0.05231722816824913, "global_step": 19311, "epoch": 536, "lr": 0.001} +{"train_loss": 0.038718514144420624, "train_loss_bc": 0.038718514144420624, "train_loss_llm": 0.0, "grad_norm": 0.06615292280912399, "global_step": 19312, "epoch": 536, "lr": 0.001} +{"train_loss": 0.038270726799964905, "train_loss_bc": 0.038270726799964905, "train_loss_llm": 0.0, "grad_norm": 0.030590247362852097, "global_step": 19313, "epoch": 536, "lr": 0.001} +{"train_loss": 0.031664811074733734, "train_loss_bc": 0.031664811074733734, "train_loss_llm": 0.0, "grad_norm": 0.040051572024822235, "global_step": 19314, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03205014392733574, "train_loss_bc": 0.03205014392733574, "train_loss_llm": 0.0, "grad_norm": 0.056151535362005234, "global_step": 19315, "epoch": 536, "lr": 0.001} +{"train_loss": 0.033979203552007675, "train_loss_bc": 0.033979203552007675, "train_loss_llm": 0.0, "grad_norm": 0.061023470014333725, "global_step": 19316, "epoch": 536, "lr": 0.001} +{"train_loss": 0.034839559346437454, "train_loss_bc": 0.034839559346437454, "train_loss_llm": 0.0, "grad_norm": 0.02808511257171631, "global_step": 19317, "epoch": 536, "lr": 0.001} +{"train_loss": 0.0440986342728138, "train_loss_bc": 0.0440986342728138, "train_loss_llm": 0.0, "grad_norm": 0.05090424418449402, "global_step": 19318, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03599905967712402, "train_loss_bc": 0.03597584739327431, "train_loss_llm": 0.0023210644721984863, "grad_norm": 0.055001936852931976, "global_step": 19319, "epoch": 536, "lr": 0.001} +{"train_loss": 0.023207014426589012, "train_loss_bc": 0.023207014426589012, "train_loss_llm": 0.0, "grad_norm": 0.06425640732049942, "global_step": 19320, "epoch": 536, "lr": 0.001} +{"train_loss": 0.037736695259809494, "train_loss_bc": 0.037736695259809494, "train_loss_llm": 0.0, "grad_norm": 0.028324425220489502, "global_step": 19321, "epoch": 536, "lr": 0.001} +{"train_loss": 0.037128787487745285, "train_loss_bc": 0.037128787487745285, "train_loss_llm": 0.0, "grad_norm": 0.04307884722948074, "global_step": 19322, "epoch": 536, "lr": 0.001} +{"train_loss": 0.02710496261715889, "train_loss_bc": 0.02710496261715889, "train_loss_llm": 0.0, "grad_norm": 0.04874959960579872, "global_step": 19323, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03575678914785385, "train_loss_bc": 0.03575678914785385, "train_loss_llm": 0.0, "grad_norm": 0.0580914169549942, "global_step": 19324, "epoch": 536, "lr": 0.001} +{"train_loss": 0.031859394162893295, "train_loss_bc": 0.03185581788420677, "train_loss_llm": 0.00035762786865234375, "grad_norm": 0.03471532464027405, "global_step": 19325, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03448576480150223, "train_loss_bc": 0.034463606774806976, "train_loss_llm": 0.0022158920764923096, "grad_norm": 0.05153703689575195, "global_step": 19326, "epoch": 536, "lr": 0.001} +{"train_loss": 0.028044160455465317, "train_loss_bc": 0.028043026104569435, "train_loss_llm": 0.00011348724365234375, "grad_norm": 0.06310512125492096, "global_step": 19327, "epoch": 536, "lr": 0.001} +{"train_loss": 0.030249979346990585, "train_loss_bc": 0.030249979346990585, "train_loss_llm": 0.0, "grad_norm": 0.07831677794456482, "global_step": 19328, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03445233404636383, "train_loss_bc": 0.03445233404636383, "train_loss_llm": 0.0, "grad_norm": 0.03552142530679703, "global_step": 19329, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03309532254934311, "train_loss_bc": 0.03308014199137688, "train_loss_llm": 0.0015180110931396484, "grad_norm": 0.047631628811359406, "global_step": 19330, "epoch": 536, "lr": 0.001} +{"train_loss": 0.03339991780618826, "train_loss_bc": 0.033118922263383865, "train_loss_llm": 0.0, "grad_norm": 0.06246216595172882, "global_step": 19331, "epoch": 536, "lr": 0.001, "val_loss": 0.07675367593765259} +{"train_loss": 0.03432818502187729, "train_loss_bc": 0.03432818502187729, "train_loss_llm": 0.0, "grad_norm": 0.07254762947559357, "global_step": 19332, "epoch": 537, "lr": 0.001} +{"train_loss": 0.02865259349346161, "train_loss_bc": 0.028647610917687416, "train_loss_llm": 0.0004982054233551025, "grad_norm": 0.023434564471244812, "global_step": 19333, "epoch": 537, "lr": 0.001} +{"train_loss": 0.031775835901498795, "train_loss_bc": 0.03177354857325554, "train_loss_llm": 0.00022876262664794922, "grad_norm": 0.03408848121762276, "global_step": 19334, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03452572599053383, "train_loss_bc": 0.03452485427260399, "train_loss_llm": 8.702278137207031e-05, "grad_norm": 0.0576915368437767, "global_step": 19335, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03379245847463608, "train_loss_bc": 0.03379245847463608, "train_loss_llm": 0.0, "grad_norm": 0.06738230586051941, "global_step": 19336, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03411143273115158, "train_loss_bc": 0.03411143273115158, "train_loss_llm": 0.0, "grad_norm": 0.028580553829669952, "global_step": 19337, "epoch": 537, "lr": 0.001} +{"train_loss": 0.029707932844758034, "train_loss_bc": 0.029707932844758034, "train_loss_llm": 0.0, "grad_norm": 0.04322313144803047, "global_step": 19338, "epoch": 537, "lr": 0.001} +{"train_loss": 0.031143691390752792, "train_loss_bc": 0.031143691390752792, "train_loss_llm": 0.0, "grad_norm": 0.06093846634030342, "global_step": 19339, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03255002945661545, "train_loss_bc": 0.032546039670705795, "train_loss_llm": 0.00039905309677124023, "grad_norm": 0.07046815752983093, "global_step": 19340, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03599422797560692, "train_loss_bc": 0.0359804704785347, "train_loss_llm": 0.0013755857944488525, "grad_norm": 0.03088352270424366, "global_step": 19341, "epoch": 537, "lr": 0.001} +{"train_loss": 0.038755372166633606, "train_loss_bc": 0.038755372166633606, "train_loss_llm": 0.0, "grad_norm": 0.05117261782288551, "global_step": 19342, "epoch": 537, "lr": 0.001} +{"train_loss": 0.029850179329514503, "train_loss_bc": 0.029850179329514503, "train_loss_llm": 0.0, "grad_norm": 0.07532760500907898, "global_step": 19343, "epoch": 537, "lr": 0.001} +{"train_loss": 0.029414936900138855, "train_loss_bc": 0.029408007860183716, "train_loss_llm": 0.0006928741931915283, "grad_norm": 0.08856464177370071, "global_step": 19344, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03227945417165756, "train_loss_bc": 0.03227945417165756, "train_loss_llm": 0.0, "grad_norm": 0.025579100474715233, "global_step": 19345, "epoch": 537, "lr": 0.001} +{"train_loss": 0.035537239164114, "train_loss_bc": 0.035537239164114, "train_loss_llm": 0.0, "grad_norm": 0.04035671055316925, "global_step": 19346, "epoch": 537, "lr": 0.001} +{"train_loss": 0.037309326231479645, "train_loss_bc": 0.037309326231479645, "train_loss_llm": 0.0, "grad_norm": 0.0572628378868103, "global_step": 19347, "epoch": 537, "lr": 0.001} +{"train_loss": 0.033143848180770874, "train_loss_bc": 0.033143848180770874, "train_loss_llm": 0.0, "grad_norm": 0.0730532854795456, "global_step": 19348, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03150467202067375, "train_loss_bc": 0.03150467202067375, "train_loss_llm": 0.0, "grad_norm": 0.025643659755587578, "global_step": 19349, "epoch": 537, "lr": 0.001} +{"train_loss": 0.033175207674503326, "train_loss_bc": 0.033175207674503326, "train_loss_llm": 0.0, "grad_norm": 0.03406727686524391, "global_step": 19350, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03886118903756142, "train_loss_bc": 0.03886118903756142, "train_loss_llm": 0.0, "grad_norm": 0.0513196624815464, "global_step": 19351, "epoch": 537, "lr": 0.001} +{"train_loss": 0.02981262281537056, "train_loss_bc": 0.02980729192495346, "train_loss_llm": 0.0005330443382263184, "grad_norm": 0.058348581194877625, "global_step": 19352, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03614441677927971, "train_loss_bc": 0.03614441677927971, "train_loss_llm": 0.0, "grad_norm": 0.03292740508913994, "global_step": 19353, "epoch": 537, "lr": 0.001} +{"train_loss": 0.034375350922346115, "train_loss_bc": 0.034375350922346115, "train_loss_llm": 0.0, "grad_norm": 0.05666821449995041, "global_step": 19354, "epoch": 537, "lr": 0.001} +{"train_loss": 0.033818524330854416, "train_loss_bc": 0.03381355106830597, "train_loss_llm": 0.0004971623420715332, "grad_norm": 0.07098247855901718, "global_step": 19355, "epoch": 537, "lr": 0.001} +{"train_loss": 0.033544957637786865, "train_loss_bc": 0.033544957637786865, "train_loss_llm": 0.0, "grad_norm": 0.07969390600919724, "global_step": 19356, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03673592209815979, "train_loss_bc": 0.03671182319521904, "train_loss_llm": 0.0024098455905914307, "grad_norm": 0.02740832045674324, "global_step": 19357, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03932203724980354, "train_loss_bc": 0.03932203724980354, "train_loss_llm": 0.0, "grad_norm": 0.0399087630212307, "global_step": 19358, "epoch": 537, "lr": 0.001} +{"train_loss": 0.025442304089665413, "train_loss_bc": 0.025442304089665413, "train_loss_llm": 0.0, "grad_norm": 0.047293078154325485, "global_step": 19359, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03269614651799202, "train_loss_bc": 0.03269614651799202, "train_loss_llm": 0.0, "grad_norm": 0.06415373086929321, "global_step": 19360, "epoch": 537, "lr": 0.001} +{"train_loss": 0.02941916510462761, "train_loss_bc": 0.029403001070022583, "train_loss_llm": 0.0016163885593414307, "grad_norm": 0.02545296400785446, "global_step": 19361, "epoch": 537, "lr": 0.001} +{"train_loss": 0.035760052502155304, "train_loss_bc": 0.035750437527894974, "train_loss_llm": 0.0009615421295166016, "grad_norm": 0.04047190025448799, "global_step": 19362, "epoch": 537, "lr": 0.001} +{"train_loss": 0.027955394238233566, "train_loss_bc": 0.027955394238233566, "train_loss_llm": 0.0, "grad_norm": 0.055586110800504684, "global_step": 19363, "epoch": 537, "lr": 0.001} +{"train_loss": 0.04300522059202194, "train_loss_bc": 0.04300522059202194, "train_loss_llm": 0.0, "grad_norm": 0.08621589839458466, "global_step": 19364, "epoch": 537, "lr": 0.001} +{"train_loss": 0.0380336232483387, "train_loss_bc": 0.0380336232483387, "train_loss_llm": 0.0, "grad_norm": 0.033443909138441086, "global_step": 19365, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03308553993701935, "train_loss_bc": 0.03308553993701935, "train_loss_llm": 0.0, "grad_norm": 0.0512973889708519, "global_step": 19366, "epoch": 537, "lr": 0.001} +{"train_loss": 0.03351812628615233, "train_loss_bc": 0.031087730079889297, "train_loss_llm": 0.0, "grad_norm": 0.0716390535235405, "global_step": 19367, "epoch": 537, "lr": 0.001, "val_loss": 0.07723861187696457} +{"train_loss": 0.03085414133965969, "train_loss_bc": 0.03083854354918003, "train_loss_llm": 0.0015597939491271973, "grad_norm": 0.0880349725484848, "global_step": 19368, "epoch": 538, "lr": 0.001} +{"train_loss": 0.0418880321085453, "train_loss_bc": 0.04186030849814415, "train_loss_llm": 0.0027725398540496826, "grad_norm": 0.028485232964158058, "global_step": 19369, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03131178021430969, "train_loss_bc": 0.03131178021430969, "train_loss_llm": 0.0, "grad_norm": 0.04074518382549286, "global_step": 19370, "epoch": 538, "lr": 0.001} +{"train_loss": 0.028234044089913368, "train_loss_bc": 0.028221268206834793, "train_loss_llm": 0.0012776553630828857, "grad_norm": 0.06088931858539581, "global_step": 19371, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03465167060494423, "train_loss_bc": 0.03463008999824524, "train_loss_llm": 0.0021581053733825684, "grad_norm": 0.07714926451444626, "global_step": 19372, "epoch": 538, "lr": 0.001} +{"train_loss": 0.02897518500685692, "train_loss_bc": 0.02897518500685692, "train_loss_llm": 0.0, "grad_norm": 0.02909969538450241, "global_step": 19373, "epoch": 538, "lr": 0.001} +{"train_loss": 0.027729840949177742, "train_loss_bc": 0.027729840949177742, "train_loss_llm": 0.0, "grad_norm": 0.04001612588763237, "global_step": 19374, "epoch": 538, "lr": 0.001} +{"train_loss": 0.029270684346556664, "train_loss_bc": 0.029270684346556664, "train_loss_llm": 0.0, "grad_norm": 0.056873902678489685, "global_step": 19375, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03567809984087944, "train_loss_bc": 0.035665467381477356, "train_loss_llm": 0.001263350248336792, "grad_norm": 0.08460730314254761, "global_step": 19376, "epoch": 538, "lr": 0.001} +{"train_loss": 0.032437071204185486, "train_loss_bc": 0.03243584930896759, "train_loss_llm": 0.00012230873107910156, "grad_norm": 0.027841243892908096, "global_step": 19377, "epoch": 538, "lr": 0.001} +{"train_loss": 0.032600320875644684, "train_loss_bc": 0.032600320875644684, "train_loss_llm": 0.0, "grad_norm": 0.03979964554309845, "global_step": 19378, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03189637511968613, "train_loss_bc": 0.03189637511968613, "train_loss_llm": 0.0, "grad_norm": 0.053183041512966156, "global_step": 19379, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03566018491983414, "train_loss_bc": 0.03565434738993645, "train_loss_llm": 0.0005837380886077881, "grad_norm": 0.06704701483249664, "global_step": 19380, "epoch": 538, "lr": 0.001} +{"train_loss": 0.031205372884869576, "train_loss_bc": 0.031205372884869576, "train_loss_llm": 0.0, "grad_norm": 0.022458985447883606, "global_step": 19381, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03310267999768257, "train_loss_bc": 0.03308229148387909, "train_loss_llm": 0.0020388364791870117, "grad_norm": 0.0318593792617321, "global_step": 19382, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03446606546640396, "train_loss_bc": 0.03446606546640396, "train_loss_llm": 0.0, "grad_norm": 0.0389469750225544, "global_step": 19383, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03501688688993454, "train_loss_bc": 0.03501688688993454, "train_loss_llm": 0.0, "grad_norm": 0.049028314650058746, "global_step": 19384, "epoch": 538, "lr": 0.001} +{"train_loss": 0.035968512296676636, "train_loss_bc": 0.035968512296676636, "train_loss_llm": 0.0, "grad_norm": 0.03350117430090904, "global_step": 19385, "epoch": 538, "lr": 0.001} +{"train_loss": 0.039656367152929306, "train_loss_bc": 0.03964655101299286, "train_loss_llm": 0.0009815394878387451, "grad_norm": 0.05755074694752693, "global_step": 19386, "epoch": 538, "lr": 0.001} +{"train_loss": 0.032990530133247375, "train_loss_bc": 0.032990530133247375, "train_loss_llm": 0.0, "grad_norm": 0.08379708975553513, "global_step": 19387, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03687692806124687, "train_loss_bc": 0.03687692806124687, "train_loss_llm": 0.0, "grad_norm": 0.09900903701782227, "global_step": 19388, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03074861690402031, "train_loss_bc": 0.03074861690402031, "train_loss_llm": 0.0, "grad_norm": 0.02796095982193947, "global_step": 19389, "epoch": 538, "lr": 0.001} +{"train_loss": 0.036081600934267044, "train_loss_bc": 0.036081600934267044, "train_loss_llm": 0.0, "grad_norm": 0.04217499494552612, "global_step": 19390, "epoch": 538, "lr": 0.001} +{"train_loss": 0.033132560551166534, "train_loss_bc": 0.033132560551166534, "train_loss_llm": 0.0, "grad_norm": 0.06053576618432999, "global_step": 19391, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03553468734025955, "train_loss_bc": 0.03553468734025955, "train_loss_llm": 0.0, "grad_norm": 0.07948459684848785, "global_step": 19392, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03291028365492821, "train_loss_bc": 0.03291028365492821, "train_loss_llm": 0.0, "grad_norm": 0.025250140577554703, "global_step": 19393, "epoch": 538, "lr": 0.001} +{"train_loss": 0.02806704305112362, "train_loss_bc": 0.02806704305112362, "train_loss_llm": 0.0, "grad_norm": 0.03628097474575043, "global_step": 19394, "epoch": 538, "lr": 0.001} +{"train_loss": 0.036568135023117065, "train_loss_bc": 0.036541860550642014, "train_loss_llm": 0.0026273727416992188, "grad_norm": 0.058742765337228775, "global_step": 19395, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03417374938726425, "train_loss_bc": 0.034165527671575546, "train_loss_llm": 0.0008221268653869629, "grad_norm": 0.07810945063829422, "global_step": 19396, "epoch": 538, "lr": 0.001} +{"train_loss": 0.030466124415397644, "train_loss_bc": 0.030466124415397644, "train_loss_llm": 0.0, "grad_norm": 0.027650581672787666, "global_step": 19397, "epoch": 538, "lr": 0.001} +{"train_loss": 0.037994664162397385, "train_loss_bc": 0.037994664162397385, "train_loss_llm": 0.0, "grad_norm": 0.046554211527109146, "global_step": 19398, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03648966923356056, "train_loss_bc": 0.03648966923356056, "train_loss_llm": 0.0, "grad_norm": 0.059947822242975235, "global_step": 19399, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03639157861471176, "train_loss_bc": 0.03639157861471176, "train_loss_llm": 0.0, "grad_norm": 0.07043622434139252, "global_step": 19400, "epoch": 538, "lr": 0.001} +{"train_loss": 0.03201092407107353, "train_loss_bc": 0.03201092407107353, "train_loss_llm": 0.0, "grad_norm": 0.02787550538778305, "global_step": 19401, "epoch": 538, "lr": 0.001} +{"train_loss": 0.035004496574401855, "train_loss_bc": 0.035004496574401855, "train_loss_llm": 0.0, "grad_norm": 0.049813348799943924, "global_step": 19402, "epoch": 538, "lr": 0.001} +{"train_loss": 0.033651970223420195, "train_loss_bc": 0.03542602062225342, "train_loss_llm": 0.0, "grad_norm": 0.06824280321598053, "global_step": 19403, "epoch": 538, "lr": 0.001, "val_loss": 0.07711388170719147} +{"train_loss": 0.035467710345983505, "train_loss_bc": 0.035467710345983505, "train_loss_llm": 0.0, "grad_norm": 0.07652336359024048, "global_step": 19404, "epoch": 539, "lr": 0.001} +{"train_loss": 0.029036983847618103, "train_loss_bc": 0.02903367020189762, "train_loss_llm": 0.0003312826156616211, "grad_norm": 0.03031899780035019, "global_step": 19405, "epoch": 539, "lr": 0.001} +{"train_loss": 0.028787963092327118, "train_loss_bc": 0.028765037655830383, "train_loss_llm": 0.0022925138473510742, "grad_norm": 0.038908492773771286, "global_step": 19406, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03694239631295204, "train_loss_bc": 0.03694239631295204, "train_loss_llm": 0.0, "grad_norm": 0.06095949932932854, "global_step": 19407, "epoch": 539, "lr": 0.001} +{"train_loss": 0.039333928376436234, "train_loss_bc": 0.039333928376436234, "train_loss_llm": 0.0, "grad_norm": 0.07786644995212555, "global_step": 19408, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03007245622575283, "train_loss_bc": 0.03007245622575283, "train_loss_llm": 0.0, "grad_norm": 0.027373407036066055, "global_step": 19409, "epoch": 539, "lr": 0.001} +{"train_loss": 0.036976952105760574, "train_loss_bc": 0.036976952105760574, "train_loss_llm": 0.0, "grad_norm": 0.04542917013168335, "global_step": 19410, "epoch": 539, "lr": 0.001} +{"train_loss": 0.035107746720314026, "train_loss_bc": 0.035096462815999985, "train_loss_llm": 0.0011284351348876953, "grad_norm": 0.058616816997528076, "global_step": 19411, "epoch": 539, "lr": 0.001} +{"train_loss": 0.027331802994012833, "train_loss_bc": 0.027331802994012833, "train_loss_llm": 0.0, "grad_norm": 0.08189722895622253, "global_step": 19412, "epoch": 539, "lr": 0.001} +{"train_loss": 0.030332256108522415, "train_loss_bc": 0.030332256108522415, "train_loss_llm": 0.0, "grad_norm": 0.024118473753333092, "global_step": 19413, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03507627919316292, "train_loss_bc": 0.03507627919316292, "train_loss_llm": 0.0, "grad_norm": 0.04762432724237442, "global_step": 19414, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03581452742218971, "train_loss_bc": 0.03581452742218971, "train_loss_llm": 0.0, "grad_norm": 0.06526510417461395, "global_step": 19415, "epoch": 539, "lr": 0.001} +{"train_loss": 0.033028438687324524, "train_loss_bc": 0.033028438687324524, "train_loss_llm": 0.0, "grad_norm": 0.08621533960103989, "global_step": 19416, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03570683300495148, "train_loss_bc": 0.035690873861312866, "train_loss_llm": 0.0015957951545715332, "grad_norm": 0.029529033228754997, "global_step": 19417, "epoch": 539, "lr": 0.001} +{"train_loss": 0.034421321004629135, "train_loss_bc": 0.034421321004629135, "train_loss_llm": 0.0, "grad_norm": 0.045501504093408585, "global_step": 19418, "epoch": 539, "lr": 0.001} +{"train_loss": 0.037021197378635406, "train_loss_bc": 0.03700552135705948, "train_loss_llm": 0.001567453145980835, "grad_norm": 0.05475730821490288, "global_step": 19419, "epoch": 539, "lr": 0.001} +{"train_loss": 0.029590943828225136, "train_loss_bc": 0.029574094340205193, "train_loss_llm": 0.001685023307800293, "grad_norm": 0.06266278773546219, "global_step": 19420, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03343678265810013, "train_loss_bc": 0.03343557193875313, "train_loss_llm": 0.00012090802192687988, "grad_norm": 0.02657087706029415, "global_step": 19421, "epoch": 539, "lr": 0.001} +{"train_loss": 0.028869403526186943, "train_loss_bc": 0.028869403526186943, "train_loss_llm": 0.0, "grad_norm": 0.04462827369570732, "global_step": 19422, "epoch": 539, "lr": 0.001} +{"train_loss": 0.033195700496435165, "train_loss_bc": 0.033195700496435165, "train_loss_llm": 0.0, "grad_norm": 0.049511782824993134, "global_step": 19423, "epoch": 539, "lr": 0.001} +{"train_loss": 0.032112665474414825, "train_loss_bc": 0.032112665474414825, "train_loss_llm": 0.0, "grad_norm": 0.05845483019948006, "global_step": 19424, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03354061767458916, "train_loss_bc": 0.03354061767458916, "train_loss_llm": 0.0, "grad_norm": 0.027497800067067146, "global_step": 19425, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03481777757406235, "train_loss_bc": 0.03479873389005661, "train_loss_llm": 0.0019044578075408936, "grad_norm": 0.04641779884696007, "global_step": 19426, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03187967464327812, "train_loss_bc": 0.03187967464327812, "train_loss_llm": 0.0, "grad_norm": 0.05855625495314598, "global_step": 19427, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03726721554994583, "train_loss_bc": 0.03726242110133171, "train_loss_llm": 0.0004793405532836914, "grad_norm": 0.0736248642206192, "global_step": 19428, "epoch": 539, "lr": 0.001} +{"train_loss": 0.034812089055776596, "train_loss_bc": 0.034812089055776596, "train_loss_llm": 0.0, "grad_norm": 0.026265183463692665, "global_step": 19429, "epoch": 539, "lr": 0.001} +{"train_loss": 0.031823690980672836, "train_loss_bc": 0.03181846812367439, "train_loss_llm": 0.0005221366882324219, "grad_norm": 0.03811352699995041, "global_step": 19430, "epoch": 539, "lr": 0.001} +{"train_loss": 0.0328078530728817, "train_loss_bc": 0.0328078530728817, "train_loss_llm": 0.0, "grad_norm": 0.0485166572034359, "global_step": 19431, "epoch": 539, "lr": 0.001} +{"train_loss": 0.033386118710041046, "train_loss_bc": 0.03336489200592041, "train_loss_llm": 0.002122849225997925, "grad_norm": 0.06047205626964569, "global_step": 19432, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03371385484933853, "train_loss_bc": 0.03371385484933853, "train_loss_llm": 0.0, "grad_norm": 0.02371279150247574, "global_step": 19433, "epoch": 539, "lr": 0.001} +{"train_loss": 0.0372421033680439, "train_loss_bc": 0.0372421033680439, "train_loss_llm": 0.0, "grad_norm": 0.03783417493104935, "global_step": 19434, "epoch": 539, "lr": 0.001} +{"train_loss": 0.04008420556783676, "train_loss_bc": 0.04008420556783676, "train_loss_llm": 0.0, "grad_norm": 0.051150690764188766, "global_step": 19435, "epoch": 539, "lr": 0.001} +{"train_loss": 0.03364705294370651, "train_loss_bc": 0.03364315629005432, "train_loss_llm": 0.000389784574508667, "grad_norm": 0.06999720633029938, "global_step": 19436, "epoch": 539, "lr": 0.001} +{"train_loss": 0.028911055997014046, "train_loss_bc": 0.028911055997014046, "train_loss_llm": 0.0, "grad_norm": 0.028689784929156303, "global_step": 19437, "epoch": 539, "lr": 0.001} +{"train_loss": 0.030571941286325455, "train_loss_bc": 0.030571941286325455, "train_loss_llm": 0.0, "grad_norm": 0.04586734622716904, "global_step": 19438, "epoch": 539, "lr": 0.001} +{"train_loss": 0.0334781791186995, "train_loss_bc": 0.033044908195734024, "train_loss_llm": 0.0, "grad_norm": 0.051059573888778687, "global_step": 19439, "epoch": 539, "lr": 0.001, "val_loss": 0.07669055461883545} +{"train_loss": 0.033513929694890976, "train_loss_bc": 0.033513929694890976, "train_loss_llm": 0.0, "grad_norm": 0.06744703650474548, "global_step": 19440, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03319254145026207, "train_loss_bc": 0.033191900700330734, "train_loss_llm": 6.395578384399414e-05, "grad_norm": 0.027238115668296814, "global_step": 19441, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03637785464525223, "train_loss_bc": 0.03637785464525223, "train_loss_llm": 0.0, "grad_norm": 0.03863850608468056, "global_step": 19442, "epoch": 540, "lr": 0.001} +{"train_loss": 0.039324138313531876, "train_loss_bc": 0.039324138313531876, "train_loss_llm": 0.0, "grad_norm": 0.054479777812957764, "global_step": 19443, "epoch": 540, "lr": 0.001} +{"train_loss": 0.04007972404360771, "train_loss_bc": 0.04007972404360771, "train_loss_llm": 0.0, "grad_norm": 0.06960470229387283, "global_step": 19444, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03008866496384144, "train_loss_bc": 0.03008866496384144, "train_loss_llm": 0.0, "grad_norm": 0.02499181404709816, "global_step": 19445, "epoch": 540, "lr": 0.001} +{"train_loss": 0.027378173545002937, "train_loss_bc": 0.027365252375602722, "train_loss_llm": 0.0012921392917633057, "grad_norm": 0.03830462694168091, "global_step": 19446, "epoch": 540, "lr": 0.001} +{"train_loss": 0.036231424659490585, "train_loss_bc": 0.036229152232408524, "train_loss_llm": 0.00022721290588378906, "grad_norm": 0.04613320529460907, "global_step": 19447, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03474999964237213, "train_loss_bc": 0.03474999964237213, "train_loss_llm": 0.0, "grad_norm": 0.05335952341556549, "global_step": 19448, "epoch": 540, "lr": 0.001} +{"train_loss": 0.024803241714835167, "train_loss_bc": 0.024794569239020348, "train_loss_llm": 0.0008672177791595459, "grad_norm": 0.022793954238295555, "global_step": 19449, "epoch": 540, "lr": 0.001} +{"train_loss": 0.025429757311940193, "train_loss_bc": 0.025429757311940193, "train_loss_llm": 0.0, "grad_norm": 0.03615722805261612, "global_step": 19450, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03336039558053017, "train_loss_bc": 0.03336039558053017, "train_loss_llm": 0.0, "grad_norm": 0.05337942764163017, "global_step": 19451, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03008388914167881, "train_loss_bc": 0.030066417530179024, "train_loss_llm": 0.0017470717430114746, "grad_norm": 0.0689520537853241, "global_step": 19452, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03708692267537117, "train_loss_bc": 0.03708692267537117, "train_loss_llm": 0.0, "grad_norm": 0.0329708606004715, "global_step": 19453, "epoch": 540, "lr": 0.001} +{"train_loss": 0.030804766342043877, "train_loss_bc": 0.030804766342043877, "train_loss_llm": 0.0, "grad_norm": 0.06277145445346832, "global_step": 19454, "epoch": 540, "lr": 0.001} +{"train_loss": 0.032632503658533096, "train_loss_bc": 0.032632503658533096, "train_loss_llm": 0.0, "grad_norm": 0.07766877114772797, "global_step": 19455, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03331894427537918, "train_loss_bc": 0.03331894427537918, "train_loss_llm": 0.0, "grad_norm": 0.09136494249105453, "global_step": 19456, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03412291407585144, "train_loss_bc": 0.03412291407585144, "train_loss_llm": 0.0, "grad_norm": 0.024844251573085785, "global_step": 19457, "epoch": 540, "lr": 0.001} +{"train_loss": 0.02824786864221096, "train_loss_bc": 0.02824786864221096, "train_loss_llm": 0.0, "grad_norm": 0.03588880971074104, "global_step": 19458, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03297841176390648, "train_loss_bc": 0.032970476895570755, "train_loss_llm": 0.00079345703125, "grad_norm": 0.05062670633196831, "global_step": 19459, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03714141622185707, "train_loss_bc": 0.03714141622185707, "train_loss_llm": 0.0, "grad_norm": 0.0682145431637764, "global_step": 19460, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03363072872161865, "train_loss_bc": 0.03363072872161865, "train_loss_llm": 0.0, "grad_norm": 0.027239063754677773, "global_step": 19461, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03226417675614357, "train_loss_bc": 0.03224935755133629, "train_loss_llm": 0.0014820992946624756, "grad_norm": 0.03446962311863899, "global_step": 19462, "epoch": 540, "lr": 0.001} +{"train_loss": 0.034402526915073395, "train_loss_bc": 0.034402526915073395, "train_loss_llm": 0.0, "grad_norm": 0.05397395044565201, "global_step": 19463, "epoch": 540, "lr": 0.001} +{"train_loss": 0.025366872549057007, "train_loss_bc": 0.025366872549057007, "train_loss_llm": 0.0, "grad_norm": 0.06642858684062958, "global_step": 19464, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03245272859930992, "train_loss_bc": 0.032446783035993576, "train_loss_llm": 0.00059470534324646, "grad_norm": 0.030247393995523453, "global_step": 19465, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03418901190161705, "train_loss_bc": 0.034181151539087296, "train_loss_llm": 0.0007860660552978516, "grad_norm": 0.04529577121138573, "global_step": 19466, "epoch": 540, "lr": 0.001} +{"train_loss": 0.037297967821359634, "train_loss_bc": 0.037297967821359634, "train_loss_llm": 0.0, "grad_norm": 0.058583054691553116, "global_step": 19467, "epoch": 540, "lr": 0.001} +{"train_loss": 0.04009609296917915, "train_loss_bc": 0.04009438678622246, "train_loss_llm": 0.00017073750495910645, "grad_norm": 0.07441842555999756, "global_step": 19468, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03474296256899834, "train_loss_bc": 0.03474296256899834, "train_loss_llm": 0.0, "grad_norm": 0.02922755479812622, "global_step": 19469, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03472841903567314, "train_loss_bc": 0.03472841903567314, "train_loss_llm": 0.0, "grad_norm": 0.04675228148698807, "global_step": 19470, "epoch": 540, "lr": 0.001} +{"train_loss": 0.03333425149321556, "train_loss_bc": 0.03333425149321556, "train_loss_llm": 0.0, "grad_norm": 0.0610460489988327, "global_step": 19471, "epoch": 540, "lr": 0.001} +{"train_loss": 0.02666877768933773, "train_loss_bc": 0.02665761485695839, "train_loss_llm": 0.001116245985031128, "grad_norm": 0.07001154869794846, "global_step": 19472, "epoch": 540, "lr": 0.001} +{"train_loss": 0.033770784735679626, "train_loss_bc": 0.033770784735679626, "train_loss_llm": 0.0, "grad_norm": 0.029032927006483078, "global_step": 19473, "epoch": 540, "lr": 0.001} +{"train_loss": 0.04429912567138672, "train_loss_bc": 0.04429912567138672, "train_loss_llm": 0.0, "grad_norm": 0.049606263637542725, "global_step": 19474, "epoch": 540, "lr": 0.001} +{"train_loss": 0.033435097999042936, "train_loss_bc": 0.03547161817550659, "train_loss_llm": 0.0, "grad_norm": 0.07726003974676132, "global_step": 19475, "epoch": 540, "lr": 0.001, "train/cumulative_reward": 116.15675639470828, "train/mean_score": 11.826965661229657, "train/success_rate": 0.2, "test/cumulative_reward": 31.36577406985699, "test/mean_score": 8.28000649451267, "test/success_rate": 0.22, "val_loss": 0.07873259484767914, "train_action_mse_error": 0.03648906201124191} +{"train_loss": 0.03567101061344147, "train_loss_bc": 0.03567101061344147, "train_loss_llm": 0.0, "grad_norm": 0.09241660684347153, "global_step": 19476, "epoch": 541, "lr": 0.001} +{"train_loss": 0.036971502006053925, "train_loss_bc": 0.036971502006053925, "train_loss_llm": 0.0, "grad_norm": 0.03282255306839943, "global_step": 19477, "epoch": 541, "lr": 0.001} +{"train_loss": 0.030515583232045174, "train_loss_bc": 0.030515583232045174, "train_loss_llm": 0.0, "grad_norm": 0.05303313210606575, "global_step": 19478, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03561340272426605, "train_loss_bc": 0.03561340272426605, "train_loss_llm": 0.0, "grad_norm": 0.06583224982023239, "global_step": 19479, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03706586733460426, "train_loss_bc": 0.03706586733460426, "train_loss_llm": 0.0, "grad_norm": 0.06517952680587769, "global_step": 19480, "epoch": 541, "lr": 0.001} +{"train_loss": 0.031168470159173012, "train_loss_bc": 0.031168470159173012, "train_loss_llm": 0.0, "grad_norm": 0.02012668550014496, "global_step": 19481, "epoch": 541, "lr": 0.001} +{"train_loss": 0.027155641466379166, "train_loss_bc": 0.027155641466379166, "train_loss_llm": 0.0, "grad_norm": 0.03119608201086521, "global_step": 19482, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03508823364973068, "train_loss_bc": 0.03508823364973068, "train_loss_llm": 0.0, "grad_norm": 0.0454648993909359, "global_step": 19483, "epoch": 541, "lr": 0.001} +{"train_loss": 0.032527849078178406, "train_loss_bc": 0.032527849078178406, "train_loss_llm": 0.0, "grad_norm": 0.06060187891125679, "global_step": 19484, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03333347290754318, "train_loss_bc": 0.03333347290754318, "train_loss_llm": 0.0, "grad_norm": 0.024292409420013428, "global_step": 19485, "epoch": 541, "lr": 0.001} +{"train_loss": 0.035040583461523056, "train_loss_bc": 0.035040583461523056, "train_loss_llm": 0.0, "grad_norm": 0.03436172753572464, "global_step": 19486, "epoch": 541, "lr": 0.001} +{"train_loss": 0.035339683294296265, "train_loss_bc": 0.035339683294296265, "train_loss_llm": 0.0, "grad_norm": 0.0617404542863369, "global_step": 19487, "epoch": 541, "lr": 0.001} +{"train_loss": 0.039520759135484695, "train_loss_bc": 0.039520759135484695, "train_loss_llm": 0.0, "grad_norm": 0.0826645940542221, "global_step": 19488, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03492928296327591, "train_loss_bc": 0.03492928296327591, "train_loss_llm": 0.0, "grad_norm": 0.03451383486390114, "global_step": 19489, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03182116523385048, "train_loss_bc": 0.031810153275728226, "train_loss_llm": 0.0011011064052581787, "grad_norm": 0.05389372259378433, "global_step": 19490, "epoch": 541, "lr": 0.001} +{"train_loss": 0.029674094170331955, "train_loss_bc": 0.029674094170331955, "train_loss_llm": 0.0, "grad_norm": 0.08031228184700012, "global_step": 19491, "epoch": 541, "lr": 0.001} +{"train_loss": 0.02811364084482193, "train_loss_bc": 0.02811364084482193, "train_loss_llm": 0.0, "grad_norm": 0.09721999615430832, "global_step": 19492, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03335786983370781, "train_loss_bc": 0.03335786983370781, "train_loss_llm": 0.0, "grad_norm": 0.029099974781274796, "global_step": 19493, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03276718780398369, "train_loss_bc": 0.03275653347373009, "train_loss_llm": 0.001065373420715332, "grad_norm": 0.04485034942626953, "global_step": 19494, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03121398575603962, "train_loss_bc": 0.03121398575603962, "train_loss_llm": 0.0, "grad_norm": 0.06281062960624695, "global_step": 19495, "epoch": 541, "lr": 0.001} +{"train_loss": 0.04207639396190643, "train_loss_bc": 0.042053721845149994, "train_loss_llm": 0.0022670626640319824, "grad_norm": 0.08009660243988037, "global_step": 19496, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03660355508327484, "train_loss_bc": 0.03660355508327484, "train_loss_llm": 0.0, "grad_norm": 0.027963882312178612, "global_step": 19497, "epoch": 541, "lr": 0.001} +{"train_loss": 0.032731421291828156, "train_loss_bc": 0.03272537514567375, "train_loss_llm": 0.0006046593189239502, "grad_norm": 0.04356846958398819, "global_step": 19498, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03032071888446808, "train_loss_bc": 0.03032071888446808, "train_loss_llm": 0.0, "grad_norm": 0.06045565754175186, "global_step": 19499, "epoch": 541, "lr": 0.001} +{"train_loss": 0.033665239810943604, "train_loss_bc": 0.03365873545408249, "train_loss_llm": 0.0006506145000457764, "grad_norm": 0.07924877852201462, "global_step": 19500, "epoch": 541, "lr": 0.001} +{"train_loss": 0.030589628964662552, "train_loss_bc": 0.030589628964662552, "train_loss_llm": 0.0, "grad_norm": 0.026924286037683487, "global_step": 19501, "epoch": 541, "lr": 0.001} +{"train_loss": 0.034190308302640915, "train_loss_bc": 0.03418707102537155, "train_loss_llm": 0.00032386183738708496, "grad_norm": 0.037804778665304184, "global_step": 19502, "epoch": 541, "lr": 0.001} +{"train_loss": 0.0340331606566906, "train_loss_bc": 0.0340331606566906, "train_loss_llm": 0.0, "grad_norm": 0.05226609855890274, "global_step": 19503, "epoch": 541, "lr": 0.001} +{"train_loss": 0.031226539984345436, "train_loss_bc": 0.03122568689286709, "train_loss_llm": 8.538365364074707e-05, "grad_norm": 0.061220332980155945, "global_step": 19504, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03447064384818077, "train_loss_bc": 0.03447064384818077, "train_loss_llm": 0.0, "grad_norm": 0.02642141655087471, "global_step": 19505, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03235389292240143, "train_loss_bc": 0.03235389292240143, "train_loss_llm": 0.0, "grad_norm": 0.04959932342171669, "global_step": 19506, "epoch": 541, "lr": 0.001} +{"train_loss": 0.036219313740730286, "train_loss_bc": 0.036219313740730286, "train_loss_llm": 0.0, "grad_norm": 0.05934949219226837, "global_step": 19507, "epoch": 541, "lr": 0.001} +{"train_loss": 0.037108469754457474, "train_loss_bc": 0.03708358108997345, "train_loss_llm": 0.0024889111518859863, "grad_norm": 0.06891565769910812, "global_step": 19508, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03724290803074837, "train_loss_bc": 0.03724290803074837, "train_loss_llm": 0.0, "grad_norm": 0.029041694477200508, "global_step": 19509, "epoch": 541, "lr": 0.001} +{"train_loss": 0.02864280343055725, "train_loss_bc": 0.02864280343055725, "train_loss_llm": 0.0, "grad_norm": 0.049218155443668365, "global_step": 19510, "epoch": 541, "lr": 0.001} +{"train_loss": 0.03347414550888869, "train_loss_bc": 0.026704953983426094, "train_loss_llm": 0.0, "grad_norm": 0.07157326489686966, "global_step": 19511, "epoch": 541, "lr": 0.001, "val_loss": 0.07777895033359528} +{"train_loss": 0.031293440610170364, "train_loss_bc": 0.03128642588853836, "train_loss_llm": 0.0007013380527496338, "grad_norm": 0.08065734058618546, "global_step": 19512, "epoch": 542, "lr": 0.001} +{"train_loss": 0.034699052572250366, "train_loss_bc": 0.034699052572250366, "train_loss_llm": 0.0, "grad_norm": 0.02368634007871151, "global_step": 19513, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03421410918235779, "train_loss_bc": 0.03421410918235779, "train_loss_llm": 0.0, "grad_norm": 0.040972739458084106, "global_step": 19514, "epoch": 542, "lr": 0.001} +{"train_loss": 0.032938286662101746, "train_loss_bc": 0.032938286662101746, "train_loss_llm": 0.0, "grad_norm": 0.04462345317006111, "global_step": 19515, "epoch": 542, "lr": 0.001} +{"train_loss": 0.037579283118247986, "train_loss_bc": 0.037579283118247986, "train_loss_llm": 0.0, "grad_norm": 0.06832463294267654, "global_step": 19516, "epoch": 542, "lr": 0.001} +{"train_loss": 0.031096437945961952, "train_loss_bc": 0.031096437945961952, "train_loss_llm": 0.0, "grad_norm": 0.025229405611753464, "global_step": 19517, "epoch": 542, "lr": 0.001} +{"train_loss": 0.035796426236629486, "train_loss_bc": 0.035796426236629486, "train_loss_llm": 0.0, "grad_norm": 0.04245028272271156, "global_step": 19518, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03510654345154762, "train_loss_bc": 0.03506820648908615, "train_loss_llm": 0.003833770751953125, "grad_norm": 0.05594934523105621, "global_step": 19519, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03759988769888878, "train_loss_bc": 0.03759988769888878, "train_loss_llm": 0.0, "grad_norm": 0.06676188856363297, "global_step": 19520, "epoch": 542, "lr": 0.001} +{"train_loss": 0.034087445586919785, "train_loss_bc": 0.03407694771885872, "train_loss_llm": 0.0010498762130737305, "grad_norm": 0.02479388378560543, "global_step": 19521, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03387131914496422, "train_loss_bc": 0.03386222571134567, "train_loss_llm": 0.0009093582630157471, "grad_norm": 0.04248207062482834, "global_step": 19522, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03117329627275467, "train_loss_bc": 0.03117329627275467, "train_loss_llm": 0.0, "grad_norm": 0.058556269854307175, "global_step": 19523, "epoch": 542, "lr": 0.001} +{"train_loss": 0.037559930235147476, "train_loss_bc": 0.03753937408328056, "train_loss_llm": 0.0020557641983032227, "grad_norm": 0.07509179413318634, "global_step": 19524, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03084954246878624, "train_loss_bc": 0.03084954246878624, "train_loss_llm": 0.0, "grad_norm": 0.03241267427802086, "global_step": 19525, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03190908953547478, "train_loss_bc": 0.03190908953547478, "train_loss_llm": 0.0, "grad_norm": 0.04783826693892479, "global_step": 19526, "epoch": 542, "lr": 0.001} +{"train_loss": 0.026287034153938293, "train_loss_bc": 0.026284009218215942, "train_loss_llm": 0.0003025829792022705, "grad_norm": 0.05966774374246597, "global_step": 19527, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03342896327376366, "train_loss_bc": 0.03342896327376366, "train_loss_llm": 0.0, "grad_norm": 0.07019241899251938, "global_step": 19528, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03421056270599365, "train_loss_bc": 0.03421056270599365, "train_loss_llm": 0.0, "grad_norm": 0.02918589487671852, "global_step": 19529, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03070206381380558, "train_loss_bc": 0.03070063330233097, "train_loss_llm": 0.00014314055442810059, "grad_norm": 0.03431783989071846, "global_step": 19530, "epoch": 542, "lr": 0.001} +{"train_loss": 0.035194676369428635, "train_loss_bc": 0.035194605588912964, "train_loss_llm": 7.092952728271484e-06, "grad_norm": 0.05180303752422333, "global_step": 19531, "epoch": 542, "lr": 0.001} +{"train_loss": 0.030984705314040184, "train_loss_bc": 0.030984705314040184, "train_loss_llm": 0.0, "grad_norm": 0.061836060136556625, "global_step": 19532, "epoch": 542, "lr": 0.001} +{"train_loss": 0.030103113502264023, "train_loss_bc": 0.030081026256084442, "train_loss_llm": 0.002208709716796875, "grad_norm": 0.022663164883852005, "global_step": 19533, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03333625569939613, "train_loss_bc": 0.03333625569939613, "train_loss_llm": 0.0, "grad_norm": 0.03636394068598747, "global_step": 19534, "epoch": 542, "lr": 0.001} +{"train_loss": 0.033766113221645355, "train_loss_bc": 0.033766113221645355, "train_loss_llm": 0.0, "grad_norm": 0.04450435936450958, "global_step": 19535, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03552965447306633, "train_loss_bc": 0.03552965447306633, "train_loss_llm": 0.0, "grad_norm": 0.060611896216869354, "global_step": 19536, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03769734874367714, "train_loss_bc": 0.03769734874367714, "train_loss_llm": 0.0, "grad_norm": 0.02777303196489811, "global_step": 19537, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03733097389340401, "train_loss_bc": 0.03733097389340401, "train_loss_llm": 0.0, "grad_norm": 0.04441770911216736, "global_step": 19538, "epoch": 542, "lr": 0.001} +{"train_loss": 0.031064460054039955, "train_loss_bc": 0.031064460054039955, "train_loss_llm": 0.0, "grad_norm": 0.058563895523548126, "global_step": 19539, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03647578880190849, "train_loss_bc": 0.03647578880190849, "train_loss_llm": 0.0, "grad_norm": 0.06867878884077072, "global_step": 19540, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03783075138926506, "train_loss_bc": 0.03783075138926506, "train_loss_llm": 0.0, "grad_norm": 0.023416463285684586, "global_step": 19541, "epoch": 542, "lr": 0.001} +{"train_loss": 0.02956925891339779, "train_loss_bc": 0.02956925891339779, "train_loss_llm": 0.0, "grad_norm": 0.03998512774705887, "global_step": 19542, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03383602574467659, "train_loss_bc": 0.033820535987615585, "train_loss_llm": 0.0015491247177124023, "grad_norm": 0.048343777656555176, "global_step": 19543, "epoch": 542, "lr": 0.001} +{"train_loss": 0.030749881640076637, "train_loss_bc": 0.030749881640076637, "train_loss_llm": 0.0, "grad_norm": 0.062485698610544205, "global_step": 19544, "epoch": 542, "lr": 0.001} +{"train_loss": 0.029311634600162506, "train_loss_bc": 0.029311634600162506, "train_loss_llm": 0.0, "grad_norm": 0.021623551845550537, "global_step": 19545, "epoch": 542, "lr": 0.001} +{"train_loss": 0.02974659763276577, "train_loss_bc": 0.029734235256910324, "train_loss_llm": 0.0012362003326416016, "grad_norm": 0.036690328270196915, "global_step": 19546, "epoch": 542, "lr": 0.001} +{"train_loss": 0.03346401437496146, "train_loss_bc": 0.03777165338397026, "train_loss_llm": 0.0002911090850830078, "grad_norm": 0.05437163636088371, "global_step": 19547, "epoch": 542, "lr": 0.001, "val_loss": 0.07716270536184311} +{"train_loss": 0.025074971839785576, "train_loss_bc": 0.025074971839785576, "train_loss_llm": 0.0, "grad_norm": 0.06210964545607567, "global_step": 19548, "epoch": 543, "lr": 0.001} +{"train_loss": 0.033456962555646896, "train_loss_bc": 0.033456962555646896, "train_loss_llm": 0.0, "grad_norm": 0.022226544097065926, "global_step": 19549, "epoch": 543, "lr": 0.001} +{"train_loss": 0.0376870296895504, "train_loss_bc": 0.03768071532249451, "train_loss_llm": 0.000631481409072876, "grad_norm": 0.04108366370201111, "global_step": 19550, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03503987193107605, "train_loss_bc": 0.03503987193107605, "train_loss_llm": 0.0, "grad_norm": 0.053686272352933884, "global_step": 19551, "epoch": 543, "lr": 0.001} +{"train_loss": 0.029647067189216614, "train_loss_bc": 0.029647067189216614, "train_loss_llm": 0.0, "grad_norm": 0.06035503372550011, "global_step": 19552, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03118942119181156, "train_loss_bc": 0.03118942119181156, "train_loss_llm": 0.0, "grad_norm": 0.0220609363168478, "global_step": 19553, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03345715254545212, "train_loss_bc": 0.03345715254545212, "train_loss_llm": 0.0, "grad_norm": 0.036581266671419144, "global_step": 19554, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03573745861649513, "train_loss_bc": 0.03573745861649513, "train_loss_llm": 0.0, "grad_norm": 0.055832017213106155, "global_step": 19555, "epoch": 543, "lr": 0.001} +{"train_loss": 0.028824731707572937, "train_loss_bc": 0.028824731707572937, "train_loss_llm": 0.0, "grad_norm": 0.06159142404794693, "global_step": 19556, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03461020067334175, "train_loss_bc": 0.03461020067334175, "train_loss_llm": 0.0, "grad_norm": 0.02848726697266102, "global_step": 19557, "epoch": 543, "lr": 0.001} +{"train_loss": 0.035563696175813675, "train_loss_bc": 0.035563696175813675, "train_loss_llm": 0.0, "grad_norm": 0.046227529644966125, "global_step": 19558, "epoch": 543, "lr": 0.001} +{"train_loss": 0.034168798476457596, "train_loss_bc": 0.034168798476457596, "train_loss_llm": 0.0, "grad_norm": 0.05879822000861168, "global_step": 19559, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03226245939731598, "train_loss_bc": 0.03226245939731598, "train_loss_llm": 0.0, "grad_norm": 0.07480970025062561, "global_step": 19560, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03482633829116821, "train_loss_bc": 0.034821439534425735, "train_loss_llm": 0.0004897117614746094, "grad_norm": 0.031023748219013214, "global_step": 19561, "epoch": 543, "lr": 0.001} +{"train_loss": 0.033467214554548264, "train_loss_bc": 0.033467214554548264, "train_loss_llm": 0.0, "grad_norm": 0.0485185906291008, "global_step": 19562, "epoch": 543, "lr": 0.001} +{"train_loss": 0.02995152398943901, "train_loss_bc": 0.02993379347026348, "train_loss_llm": 0.0017730295658111572, "grad_norm": 0.05732860416173935, "global_step": 19563, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03222065418958664, "train_loss_bc": 0.03222065418958664, "train_loss_llm": 0.0, "grad_norm": 0.06375729292631149, "global_step": 19564, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03190639615058899, "train_loss_bc": 0.03190639615058899, "train_loss_llm": 0.0, "grad_norm": 0.02398771047592163, "global_step": 19565, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03211056813597679, "train_loss_bc": 0.03211056813597679, "train_loss_llm": 0.0, "grad_norm": 0.04072076454758644, "global_step": 19566, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03994826599955559, "train_loss_bc": 0.03994826599955559, "train_loss_llm": 0.0, "grad_norm": 0.052996132522821426, "global_step": 19567, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03366726636886597, "train_loss_bc": 0.03366117551922798, "train_loss_llm": 0.0006091296672821045, "grad_norm": 0.06168549135327339, "global_step": 19568, "epoch": 543, "lr": 0.001} +{"train_loss": 0.030660489574074745, "train_loss_bc": 0.030660489574074745, "train_loss_llm": 0.0, "grad_norm": 0.025789285078644753, "global_step": 19569, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03289713338017464, "train_loss_bc": 0.03289713338017464, "train_loss_llm": 0.0, "grad_norm": 0.03645925968885422, "global_step": 19570, "epoch": 543, "lr": 0.001} +{"train_loss": 0.0380859375, "train_loss_bc": 0.03807339072227478, "train_loss_llm": 0.0012546777725219727, "grad_norm": 0.053561434149742126, "global_step": 19571, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03549329563975334, "train_loss_bc": 0.03549095243215561, "train_loss_llm": 0.00023436546325683594, "grad_norm": 0.0634596198797226, "global_step": 19572, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03382193669676781, "train_loss_bc": 0.03381834551692009, "train_loss_llm": 0.0003592371940612793, "grad_norm": 0.029674958437681198, "global_step": 19573, "epoch": 543, "lr": 0.001} +{"train_loss": 0.029080940410494804, "train_loss_bc": 0.029080940410494804, "train_loss_llm": 0.0, "grad_norm": 0.0444830097258091, "global_step": 19574, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03686034306883812, "train_loss_bc": 0.03683174401521683, "train_loss_llm": 0.0028600096702575684, "grad_norm": 0.05936822667717934, "global_step": 19575, "epoch": 543, "lr": 0.001} +{"train_loss": 0.028357788920402527, "train_loss_bc": 0.028357788920402527, "train_loss_llm": 0.0, "grad_norm": 0.06767689436674118, "global_step": 19576, "epoch": 543, "lr": 0.001} +{"train_loss": 0.028566576540470123, "train_loss_bc": 0.028566576540470123, "train_loss_llm": 0.0, "grad_norm": 0.024197956547141075, "global_step": 19577, "epoch": 543, "lr": 0.001} +{"train_loss": 0.037355005741119385, "train_loss_bc": 0.037344224750995636, "train_loss_llm": 0.001078188419342041, "grad_norm": 0.048703111708164215, "global_step": 19578, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03644232451915741, "train_loss_bc": 0.03644232451915741, "train_loss_llm": 0.0, "grad_norm": 0.05743430554866791, "global_step": 19579, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03390410169959068, "train_loss_bc": 0.03390410169959068, "train_loss_llm": 0.0, "grad_norm": 0.07661737501621246, "global_step": 19580, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03737403079867363, "train_loss_bc": 0.03736359253525734, "train_loss_llm": 0.0010440051555633545, "grad_norm": 0.040949929505586624, "global_step": 19581, "epoch": 543, "lr": 0.001} +{"train_loss": 0.034218449145555496, "train_loss_bc": 0.034218449145555496, "train_loss_llm": 0.0, "grad_norm": 0.06596174836158752, "global_step": 19582, "epoch": 543, "lr": 0.001} +{"train_loss": 0.03329395652852125, "train_loss_bc": 0.030644448474049568, "train_loss_llm": 0.0001583099365234375, "grad_norm": 0.08030808717012405, "global_step": 19583, "epoch": 543, "lr": 0.001, "val_loss": 0.07717296481132507} +{"train_loss": 0.03293031081557274, "train_loss_bc": 0.03293031081557274, "train_loss_llm": 0.0, "grad_norm": 0.08617035299539566, "global_step": 19584, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03023909404873848, "train_loss_bc": 0.03023909404873848, "train_loss_llm": 0.0, "grad_norm": 0.02105625532567501, "global_step": 19585, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03476816043257713, "train_loss_bc": 0.03476816043257713, "train_loss_llm": 0.0, "grad_norm": 0.03096991591155529, "global_step": 19586, "epoch": 544, "lr": 0.001} +{"train_loss": 0.037902239710092545, "train_loss_bc": 0.03788664564490318, "train_loss_llm": 0.001559227705001831, "grad_norm": 0.044814012944698334, "global_step": 19587, "epoch": 544, "lr": 0.001} +{"train_loss": 0.0368085578083992, "train_loss_bc": 0.0368085578083992, "train_loss_llm": 0.0, "grad_norm": 0.05232784152030945, "global_step": 19588, "epoch": 544, "lr": 0.001} +{"train_loss": 0.030476512387394905, "train_loss_bc": 0.030470333993434906, "train_loss_llm": 0.0006178915500640869, "grad_norm": 0.02226504497230053, "global_step": 19589, "epoch": 544, "lr": 0.001} +{"train_loss": 0.030554212629795074, "train_loss_bc": 0.030542530119419098, "train_loss_llm": 0.001168280839920044, "grad_norm": 0.03980526328086853, "global_step": 19590, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03585871681571007, "train_loss_bc": 0.03585871681571007, "train_loss_llm": 0.0, "grad_norm": 0.06114824488759041, "global_step": 19591, "epoch": 544, "lr": 0.001} +{"train_loss": 0.035873737186193466, "train_loss_bc": 0.035873737186193466, "train_loss_llm": 0.0, "grad_norm": 0.07835689187049866, "global_step": 19592, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03148571401834488, "train_loss_bc": 0.03148571401834488, "train_loss_llm": 0.0, "grad_norm": 0.030559487640857697, "global_step": 19593, "epoch": 544, "lr": 0.001} +{"train_loss": 0.04006842151284218, "train_loss_bc": 0.040062807500362396, "train_loss_llm": 0.000561445951461792, "grad_norm": 0.04516113921999931, "global_step": 19594, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03970751166343689, "train_loss_bc": 0.03970751166343689, "train_loss_llm": 0.0, "grad_norm": 0.07075010240077972, "global_step": 19595, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03958366811275482, "train_loss_bc": 0.03958366811275482, "train_loss_llm": 0.0, "grad_norm": 0.09809213131666183, "global_step": 19596, "epoch": 544, "lr": 0.001} +{"train_loss": 0.04013986140489578, "train_loss_bc": 0.04013754799962044, "train_loss_llm": 0.0002314746379852295, "grad_norm": 0.036302536725997925, "global_step": 19597, "epoch": 544, "lr": 0.001} +{"train_loss": 0.036356858909130096, "train_loss_bc": 0.036356858909130096, "train_loss_llm": 0.0, "grad_norm": 0.05330556631088257, "global_step": 19598, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03306003659963608, "train_loss_bc": 0.03305060416460037, "train_loss_llm": 0.0009432733058929443, "grad_norm": 0.06408628821372986, "global_step": 19599, "epoch": 544, "lr": 0.001} +{"train_loss": 0.031915877014398575, "train_loss_bc": 0.03191071003675461, "train_loss_llm": 0.0005166232585906982, "grad_norm": 0.07498792558908463, "global_step": 19600, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03652838245034218, "train_loss_bc": 0.03652838245034218, "train_loss_llm": 0.0, "grad_norm": 0.026520518586039543, "global_step": 19601, "epoch": 544, "lr": 0.001} +{"train_loss": 0.025256790220737457, "train_loss_bc": 0.025256790220737457, "train_loss_llm": 0.0, "grad_norm": 0.04158426821231842, "global_step": 19602, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03359401598572731, "train_loss_bc": 0.033580224961042404, "train_loss_llm": 0.0013791024684906006, "grad_norm": 0.05195721238851547, "global_step": 19603, "epoch": 544, "lr": 0.001} +{"train_loss": 0.02858855575323105, "train_loss_bc": 0.028557846322655678, "train_loss_llm": 0.003070920705795288, "grad_norm": 0.06625811010599136, "global_step": 19604, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03222015127539635, "train_loss_bc": 0.03222015127539635, "train_loss_llm": 0.0, "grad_norm": 0.03130875900387764, "global_step": 19605, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03183065354824066, "train_loss_bc": 0.031813107430934906, "train_loss_llm": 0.0017544329166412354, "grad_norm": 0.04121848940849304, "global_step": 19606, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03412796929478645, "train_loss_bc": 0.0341123528778553, "train_loss_llm": 0.001561582088470459, "grad_norm": 0.06334401667118073, "global_step": 19607, "epoch": 544, "lr": 0.001} +{"train_loss": 0.028092771768569946, "train_loss_bc": 0.028092771768569946, "train_loss_llm": 0.0, "grad_norm": 0.08044235408306122, "global_step": 19608, "epoch": 544, "lr": 0.001} +{"train_loss": 0.038068853318691254, "train_loss_bc": 0.038060497492551804, "train_loss_llm": 0.0008354783058166504, "grad_norm": 0.02840474061667919, "global_step": 19609, "epoch": 544, "lr": 0.001} +{"train_loss": 0.02855415642261505, "train_loss_bc": 0.028548559173941612, "train_loss_llm": 0.0005597770214080811, "grad_norm": 0.038233399391174316, "global_step": 19610, "epoch": 544, "lr": 0.001} +{"train_loss": 0.0301485788077116, "train_loss_bc": 0.0301485788077116, "train_loss_llm": 0.0, "grad_norm": 0.05145338177680969, "global_step": 19611, "epoch": 544, "lr": 0.001} +{"train_loss": 0.035553932189941406, "train_loss_bc": 0.03555034473538399, "train_loss_llm": 0.00035881996154785156, "grad_norm": 0.07234654575586319, "global_step": 19612, "epoch": 544, "lr": 0.001} +{"train_loss": 0.027401944622397423, "train_loss_bc": 0.027401944622397423, "train_loss_llm": 0.0, "grad_norm": 0.01840389147400856, "global_step": 19613, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03560285642743111, "train_loss_bc": 0.03560285642743111, "train_loss_llm": 0.0, "grad_norm": 0.03846902772784233, "global_step": 19614, "epoch": 544, "lr": 0.001} +{"train_loss": 0.030476192012429237, "train_loss_bc": 0.030476192012429237, "train_loss_llm": 0.0, "grad_norm": 0.05475505441427231, "global_step": 19615, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03609594702720642, "train_loss_bc": 0.03609594702720642, "train_loss_llm": 0.0, "grad_norm": 0.07235579192638397, "global_step": 19616, "epoch": 544, "lr": 0.001} +{"train_loss": 0.029627980664372444, "train_loss_bc": 0.029627980664372444, "train_loss_llm": 0.0, "grad_norm": 0.02747371233999729, "global_step": 19617, "epoch": 544, "lr": 0.001} +{"train_loss": 0.03114449791610241, "train_loss_bc": 0.03114449791610241, "train_loss_llm": 0.0, "grad_norm": 0.04493056237697601, "global_step": 19618, "epoch": 544, "lr": 0.001} +{"train_loss": 0.033534572873678475, "train_loss_bc": 0.036600902676582336, "train_loss_llm": 0.0, "grad_norm": 0.07200734317302704, "global_step": 19619, "epoch": 544, "lr": 0.001, "val_loss": 0.07937836647033691} +{"train_loss": 0.03353625535964966, "train_loss_bc": 0.03353625535964966, "train_loss_llm": 0.0, "grad_norm": 0.08827629685401917, "global_step": 19620, "epoch": 545, "lr": 0.001} +{"train_loss": 0.02803759276866913, "train_loss_bc": 0.02803759276866913, "train_loss_llm": 0.0, "grad_norm": 0.023870408535003662, "global_step": 19621, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03230202570557594, "train_loss_bc": 0.03230202570557594, "train_loss_llm": 0.0, "grad_norm": 0.03967592865228653, "global_step": 19622, "epoch": 545, "lr": 0.001} +{"train_loss": 0.032758720219135284, "train_loss_bc": 0.032758720219135284, "train_loss_llm": 0.0, "grad_norm": 0.05608489364385605, "global_step": 19623, "epoch": 545, "lr": 0.001} +{"train_loss": 0.029733432456851006, "train_loss_bc": 0.029733432456851006, "train_loss_llm": 0.0, "grad_norm": 0.06793799996376038, "global_step": 19624, "epoch": 545, "lr": 0.001} +{"train_loss": 0.033394306898117065, "train_loss_bc": 0.033394306898117065, "train_loss_llm": 0.0, "grad_norm": 0.026219205930829048, "global_step": 19625, "epoch": 545, "lr": 0.001} +{"train_loss": 0.036390528082847595, "train_loss_bc": 0.036390528082847595, "train_loss_llm": 0.0, "grad_norm": 0.04389689117670059, "global_step": 19626, "epoch": 545, "lr": 0.001} +{"train_loss": 0.029316965490579605, "train_loss_bc": 0.029316965490579605, "train_loss_llm": 0.0, "grad_norm": 0.05975041538476944, "global_step": 19627, "epoch": 545, "lr": 0.001} +{"train_loss": 0.032507676631212234, "train_loss_bc": 0.032507676631212234, "train_loss_llm": 0.0, "grad_norm": 0.07436282932758331, "global_step": 19628, "epoch": 545, "lr": 0.001} +{"train_loss": 0.036109957844018936, "train_loss_bc": 0.036109957844018936, "train_loss_llm": 0.0, "grad_norm": 0.035336896777153015, "global_step": 19629, "epoch": 545, "lr": 0.001} +{"train_loss": 0.033496733754873276, "train_loss_bc": 0.033496733754873276, "train_loss_llm": 0.0, "grad_norm": 0.058774933218955994, "global_step": 19630, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03660038486123085, "train_loss_bc": 0.036597926169633865, "train_loss_llm": 0.0002459883689880371, "grad_norm": 0.08124420791864395, "global_step": 19631, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03718515485525131, "train_loss_bc": 0.03718515485525131, "train_loss_llm": 0.0, "grad_norm": 0.10056767612695694, "global_step": 19632, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03471823036670685, "train_loss_bc": 0.03471823036670685, "train_loss_llm": 0.0, "grad_norm": 0.033025652170181274, "global_step": 19633, "epoch": 545, "lr": 0.001} +{"train_loss": 0.0370686799287796, "train_loss_bc": 0.0370686799287796, "train_loss_llm": 0.0, "grad_norm": 0.06377848982810974, "global_step": 19634, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03146744146943092, "train_loss_bc": 0.031453877687454224, "train_loss_llm": 0.0013562440872192383, "grad_norm": 0.08303189277648926, "global_step": 19635, "epoch": 545, "lr": 0.001} +{"train_loss": 0.028289534151554108, "train_loss_bc": 0.028289534151554108, "train_loss_llm": 0.0, "grad_norm": 0.10377786308526993, "global_step": 19636, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03274351358413696, "train_loss_bc": 0.03274351358413696, "train_loss_llm": 0.0, "grad_norm": 0.02560603991150856, "global_step": 19637, "epoch": 545, "lr": 0.001} +{"train_loss": 0.032178934663534164, "train_loss_bc": 0.032178934663534164, "train_loss_llm": 0.0, "grad_norm": 0.04060116782784462, "global_step": 19638, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03596896305680275, "train_loss_bc": 0.03596896305680275, "train_loss_llm": 0.0, "grad_norm": 0.05248898267745972, "global_step": 19639, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03485014662146568, "train_loss_bc": 0.03485014662146568, "train_loss_llm": 0.0, "grad_norm": 0.06956180930137634, "global_step": 19640, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03596906736493111, "train_loss_bc": 0.03596906736493111, "train_loss_llm": 0.0, "grad_norm": 0.026989592239260674, "global_step": 19641, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03280312940478325, "train_loss_bc": 0.03279935568571091, "train_loss_llm": 0.00037747621536254883, "grad_norm": 0.03874225914478302, "global_step": 19642, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03851528838276863, "train_loss_bc": 0.03851393982768059, "train_loss_llm": 0.00013503432273864746, "grad_norm": 0.0632760226726532, "global_step": 19643, "epoch": 545, "lr": 0.001} +{"train_loss": 0.039050038903951645, "train_loss_bc": 0.039050038903951645, "train_loss_llm": 0.0, "grad_norm": 0.07593756914138794, "global_step": 19644, "epoch": 545, "lr": 0.001} +{"train_loss": 0.02671549655497074, "train_loss_bc": 0.02671549655497074, "train_loss_llm": 0.0, "grad_norm": 0.0298552718013525, "global_step": 19645, "epoch": 545, "lr": 0.001} +{"train_loss": 0.027870340272784233, "train_loss_bc": 0.027870340272784233, "train_loss_llm": 0.0, "grad_norm": 0.04570724070072174, "global_step": 19646, "epoch": 545, "lr": 0.001} +{"train_loss": 0.028339896351099014, "train_loss_bc": 0.028339896351099014, "train_loss_llm": 0.0, "grad_norm": 0.06333357095718384, "global_step": 19647, "epoch": 545, "lr": 0.001} +{"train_loss": 0.035973820835351944, "train_loss_bc": 0.035973820835351944, "train_loss_llm": 0.0, "grad_norm": 0.08205922693014145, "global_step": 19648, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03537654131650925, "train_loss_bc": 0.03536926954984665, "train_loss_llm": 0.0007272958755493164, "grad_norm": 0.03003484569489956, "global_step": 19649, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03676446154713631, "train_loss_bc": 0.036757420748472214, "train_loss_llm": 0.0007039904594421387, "grad_norm": 0.053125448524951935, "global_step": 19650, "epoch": 545, "lr": 0.001} +{"train_loss": 0.032791618257761, "train_loss_bc": 0.032791618257761, "train_loss_llm": 0.0, "grad_norm": 0.07279975712299347, "global_step": 19651, "epoch": 545, "lr": 0.001} +{"train_loss": 0.033031731843948364, "train_loss_bc": 0.033031731843948364, "train_loss_llm": 0.0, "grad_norm": 0.09676654636859894, "global_step": 19652, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03995626047253609, "train_loss_bc": 0.03995626047253609, "train_loss_llm": 0.0, "grad_norm": 0.03864274173974991, "global_step": 19653, "epoch": 545, "lr": 0.001} +{"train_loss": 0.034266307950019836, "train_loss_bc": 0.034266307950019836, "train_loss_llm": 0.0, "grad_norm": 0.047241680324077606, "global_step": 19654, "epoch": 545, "lr": 0.001} +{"train_loss": 0.03359779130874409, "train_loss_bc": 0.033420853316783905, "train_loss_llm": 0.0020456016063690186, "grad_norm": 0.05704501271247864, "global_step": 19655, "epoch": 545, "lr": 0.001, "train/cumulative_reward": 48.14810453904853, "train/mean_score": 19.744212123706355, "train/success_rate": 0.6, "test/cumulative_reward": 39.12950598580573, "test/mean_score": 8.227854393096374, "test/success_rate": 0.22, "val_loss": 0.07661016285419464, "train_action_mse_error": 0.036274369806051254} +{"train_loss": 0.03370572254061699, "train_loss_bc": 0.03370572254061699, "train_loss_llm": 0.0, "grad_norm": 0.0717955082654953, "global_step": 19656, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03375282138586044, "train_loss_bc": 0.03373991698026657, "train_loss_llm": 0.0012905001640319824, "grad_norm": 0.030219390988349915, "global_step": 19657, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03474771976470947, "train_loss_bc": 0.03474771976470947, "train_loss_llm": 0.0, "grad_norm": 0.04886347055435181, "global_step": 19658, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03249945864081383, "train_loss_bc": 0.03249945864081383, "train_loss_llm": 0.0, "grad_norm": 0.06194470822811127, "global_step": 19659, "epoch": 546, "lr": 0.001} +{"train_loss": 0.0334310457110405, "train_loss_bc": 0.0334310457110405, "train_loss_llm": 0.0, "grad_norm": 0.08029662817716599, "global_step": 19660, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03403259441256523, "train_loss_bc": 0.03403259441256523, "train_loss_llm": 0.0, "grad_norm": 0.030229244381189346, "global_step": 19661, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03409496694803238, "train_loss_bc": 0.03406941890716553, "train_loss_llm": 0.0025546252727508545, "grad_norm": 0.04262230917811394, "global_step": 19662, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03297749161720276, "train_loss_bc": 0.032941725105047226, "train_loss_llm": 0.0035767853260040283, "grad_norm": 0.057769253849983215, "global_step": 19663, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03443697467446327, "train_loss_bc": 0.03443697467446327, "train_loss_llm": 0.0, "grad_norm": 0.08106133341789246, "global_step": 19664, "epoch": 546, "lr": 0.001} +{"train_loss": 0.030852671712636948, "train_loss_bc": 0.030841108411550522, "train_loss_llm": 0.0011563897132873535, "grad_norm": 0.02753865160048008, "global_step": 19665, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03106684423983097, "train_loss_bc": 0.03106684423983097, "train_loss_llm": 0.0, "grad_norm": 0.04677283391356468, "global_step": 19666, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03471216559410095, "train_loss_bc": 0.03471216559410095, "train_loss_llm": 0.0, "grad_norm": 0.06352338194847107, "global_step": 19667, "epoch": 546, "lr": 0.001} +{"train_loss": 0.0336744599044323, "train_loss_bc": 0.033674370497465134, "train_loss_llm": 8.761882781982422e-06, "grad_norm": 0.07835444062948227, "global_step": 19668, "epoch": 546, "lr": 0.001} +{"train_loss": 0.035803742706775665, "train_loss_bc": 0.035803742706775665, "train_loss_llm": 0.0, "grad_norm": 0.029498446732759476, "global_step": 19669, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03699956089258194, "train_loss_bc": 0.03699956089258194, "train_loss_llm": 0.0, "grad_norm": 0.05290819704532623, "global_step": 19670, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03476449102163315, "train_loss_bc": 0.03476449102163315, "train_loss_llm": 0.0, "grad_norm": 0.07484421879053116, "global_step": 19671, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03479330986738205, "train_loss_bc": 0.03479330986738205, "train_loss_llm": 0.0, "grad_norm": 0.0918143019080162, "global_step": 19672, "epoch": 546, "lr": 0.001} +{"train_loss": 0.026705985888838768, "train_loss_bc": 0.026705985888838768, "train_loss_llm": 0.0, "grad_norm": 0.02439499832689762, "global_step": 19673, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03680495545268059, "train_loss_bc": 0.03680495545268059, "train_loss_llm": 0.0, "grad_norm": 0.04499544948339462, "global_step": 19674, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03859971836209297, "train_loss_bc": 0.03859971836209297, "train_loss_llm": 0.0, "grad_norm": 0.06512802094221115, "global_step": 19675, "epoch": 546, "lr": 0.001} +{"train_loss": 0.029950667172670364, "train_loss_bc": 0.029950667172670364, "train_loss_llm": 0.0, "grad_norm": 0.08104889839887619, "global_step": 19676, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03305497393012047, "train_loss_bc": 0.03305497393012047, "train_loss_llm": 0.0, "grad_norm": 0.029133982956409454, "global_step": 19677, "epoch": 546, "lr": 0.001} +{"train_loss": 0.036578819155693054, "train_loss_bc": 0.036578819155693054, "train_loss_llm": 0.0, "grad_norm": 0.045906681567430496, "global_step": 19678, "epoch": 546, "lr": 0.001} +{"train_loss": 0.030161529779434204, "train_loss_bc": 0.030155329033732414, "train_loss_llm": 0.0006200969219207764, "grad_norm": 0.0674484372138977, "global_step": 19679, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03566882386803627, "train_loss_bc": 0.03565409407019615, "train_loss_llm": 0.0014728307723999023, "grad_norm": 0.08529380708932877, "global_step": 19680, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03424086794257164, "train_loss_bc": 0.03424086794257164, "train_loss_llm": 0.0, "grad_norm": 0.03394566848874092, "global_step": 19681, "epoch": 546, "lr": 0.001} +{"train_loss": 0.030582930892705917, "train_loss_bc": 0.030582930892705917, "train_loss_llm": 0.0, "grad_norm": 0.04549644514918327, "global_step": 19682, "epoch": 546, "lr": 0.001} +{"train_loss": 0.034122347831726074, "train_loss_bc": 0.034122347831726074, "train_loss_llm": 0.0, "grad_norm": 0.05872667208313942, "global_step": 19683, "epoch": 546, "lr": 0.001} +{"train_loss": 0.034688811749219894, "train_loss_bc": 0.034688811749219894, "train_loss_llm": 0.0, "grad_norm": 0.07008742541074753, "global_step": 19684, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03170426934957504, "train_loss_bc": 0.03170426934957504, "train_loss_llm": 0.0, "grad_norm": 0.02143871784210205, "global_step": 19685, "epoch": 546, "lr": 0.001} +{"train_loss": 0.02948846109211445, "train_loss_bc": 0.02948846109211445, "train_loss_llm": 0.0, "grad_norm": 0.038021329790353775, "global_step": 19686, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03380609303712845, "train_loss_bc": 0.03380609303712845, "train_loss_llm": 0.0, "grad_norm": 0.04817570745944977, "global_step": 19687, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03759064897894859, "train_loss_bc": 0.03759064897894859, "train_loss_llm": 0.0, "grad_norm": 0.06618949770927429, "global_step": 19688, "epoch": 546, "lr": 0.001} +{"train_loss": 0.023705149069428444, "train_loss_bc": 0.023705149069428444, "train_loss_llm": 0.0, "grad_norm": 0.027487797662615776, "global_step": 19689, "epoch": 546, "lr": 0.001} +{"train_loss": 0.044159699231386185, "train_loss_bc": 0.044159699231386185, "train_loss_llm": 0.0, "grad_norm": 0.0555923767387867, "global_step": 19690, "epoch": 546, "lr": 0.001} +{"train_loss": 0.03365238745593362, "train_loss_bc": 0.03352515399456024, "train_loss_llm": 0.0, "grad_norm": 0.08948365598917007, "global_step": 19691, "epoch": 546, "lr": 0.001, "val_loss": 0.07915704697370529} +{"train_loss": 0.03752635419368744, "train_loss_bc": 0.03752635419368744, "train_loss_llm": 0.0, "grad_norm": 0.1105978861451149, "global_step": 19692, "epoch": 547, "lr": 0.001} +{"train_loss": 0.0316171757876873, "train_loss_bc": 0.03161509707570076, "train_loss_llm": 0.00020802021026611328, "grad_norm": 0.02987244538962841, "global_step": 19693, "epoch": 547, "lr": 0.001} +{"train_loss": 0.040706224739551544, "train_loss_bc": 0.040706224739551544, "train_loss_llm": 0.0, "grad_norm": 0.04827772453427315, "global_step": 19694, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03042943775653839, "train_loss_bc": 0.03042943775653839, "train_loss_llm": 0.0, "grad_norm": 0.06069853901863098, "global_step": 19695, "epoch": 547, "lr": 0.001} +{"train_loss": 0.026742223650217056, "train_loss_bc": 0.026742223650217056, "train_loss_llm": 0.0, "grad_norm": 0.07133294641971588, "global_step": 19696, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03438063710927963, "train_loss_bc": 0.03438063710927963, "train_loss_llm": 0.0, "grad_norm": 0.03033752553164959, "global_step": 19697, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03475590795278549, "train_loss_bc": 0.03475590795278549, "train_loss_llm": 0.0, "grad_norm": 0.041373878717422485, "global_step": 19698, "epoch": 547, "lr": 0.001} +{"train_loss": 0.027761735022068024, "train_loss_bc": 0.027761735022068024, "train_loss_llm": 0.0, "grad_norm": 0.05004272982478142, "global_step": 19699, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03262760490179062, "train_loss_bc": 0.03261394798755646, "train_loss_llm": 0.0013655126094818115, "grad_norm": 0.061579007655382156, "global_step": 19700, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03348005935549736, "train_loss_bc": 0.03348005935549736, "train_loss_llm": 0.0, "grad_norm": 0.03245645388960838, "global_step": 19701, "epoch": 547, "lr": 0.001} +{"train_loss": 0.032169777899980545, "train_loss_bc": 0.032169777899980545, "train_loss_llm": 0.0, "grad_norm": 0.04285969212651253, "global_step": 19702, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03584465757012367, "train_loss_bc": 0.03582988679409027, "train_loss_llm": 0.0014772415161132812, "grad_norm": 0.058624379336833954, "global_step": 19703, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03374755382537842, "train_loss_bc": 0.033746324479579926, "train_loss_llm": 0.00012299418449401855, "grad_norm": 0.07305009663105011, "global_step": 19704, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03625290468335152, "train_loss_bc": 0.036232490092515945, "train_loss_llm": 0.002041369676589966, "grad_norm": 0.02766706608235836, "global_step": 19705, "epoch": 547, "lr": 0.001} +{"train_loss": 0.031254373490810394, "train_loss_bc": 0.031254373490810394, "train_loss_llm": 0.0, "grad_norm": 0.04023066908121109, "global_step": 19706, "epoch": 547, "lr": 0.001} +{"train_loss": 0.030915791168808937, "train_loss_bc": 0.030915791168808937, "train_loss_llm": 0.0, "grad_norm": 0.04998397454619408, "global_step": 19707, "epoch": 547, "lr": 0.001} +{"train_loss": 0.033724427223205566, "train_loss_bc": 0.033724427223205566, "train_loss_llm": 0.0, "grad_norm": 0.060763098299503326, "global_step": 19708, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03780612349510193, "train_loss_bc": 0.03780612349510193, "train_loss_llm": 0.0, "grad_norm": 0.03278493881225586, "global_step": 19709, "epoch": 547, "lr": 0.001} +{"train_loss": 0.031009318307042122, "train_loss_bc": 0.03099389187991619, "train_loss_llm": 0.0015426874160766602, "grad_norm": 0.051450133323669434, "global_step": 19710, "epoch": 547, "lr": 0.001} +{"train_loss": 0.031041720882058144, "train_loss_bc": 0.031041720882058144, "train_loss_llm": 0.0, "grad_norm": 0.07425787299871445, "global_step": 19711, "epoch": 547, "lr": 0.001} +{"train_loss": 0.02853865921497345, "train_loss_bc": 0.02853865921497345, "train_loss_llm": 0.0, "grad_norm": 0.09203086793422699, "global_step": 19712, "epoch": 547, "lr": 0.001} +{"train_loss": 0.035478927195072174, "train_loss_bc": 0.035478927195072174, "train_loss_llm": 0.0, "grad_norm": 0.02546861208975315, "global_step": 19713, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03277265280485153, "train_loss_bc": 0.03277265280485153, "train_loss_llm": 0.0, "grad_norm": 0.046840403228998184, "global_step": 19714, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03335655480623245, "train_loss_bc": 0.03335655480623245, "train_loss_llm": 0.0, "grad_norm": 0.05930546298623085, "global_step": 19715, "epoch": 547, "lr": 0.001} +{"train_loss": 0.0323471799492836, "train_loss_bc": 0.0323471799492836, "train_loss_llm": 0.0, "grad_norm": 0.06499405950307846, "global_step": 19716, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03241131827235222, "train_loss_bc": 0.03241034597158432, "train_loss_llm": 9.718537330627441e-05, "grad_norm": 0.026065589860081673, "global_step": 19717, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03196881711483002, "train_loss_bc": 0.031947918236255646, "train_loss_llm": 0.002089768648147583, "grad_norm": 0.04448205232620239, "global_step": 19718, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03741484507918358, "train_loss_bc": 0.03741392120718956, "train_loss_llm": 9.223818778991699e-05, "grad_norm": 0.05942634865641594, "global_step": 19719, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03689466416835785, "train_loss_bc": 0.03689466416835785, "train_loss_llm": 0.0, "grad_norm": 0.07031102478504181, "global_step": 19720, "epoch": 547, "lr": 0.001} +{"train_loss": 0.035616859793663025, "train_loss_bc": 0.035616859793663025, "train_loss_llm": 0.0, "grad_norm": 0.028205331414937973, "global_step": 19721, "epoch": 547, "lr": 0.001} +{"train_loss": 0.02744857594370842, "train_loss_bc": 0.02744857594370842, "train_loss_llm": 0.0, "grad_norm": 0.04237500950694084, "global_step": 19722, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03327441215515137, "train_loss_bc": 0.03325803205370903, "train_loss_llm": 0.001638174057006836, "grad_norm": 0.05770178139209747, "global_step": 19723, "epoch": 547, "lr": 0.001} +{"train_loss": 0.029451614245772362, "train_loss_bc": 0.029451614245772362, "train_loss_llm": 0.0, "grad_norm": 0.07984421402215958, "global_step": 19724, "epoch": 547, "lr": 0.001} +{"train_loss": 0.039326008409261703, "train_loss_bc": 0.039326008409261703, "train_loss_llm": 0.0, "grad_norm": 0.0415324829518795, "global_step": 19725, "epoch": 547, "lr": 0.001} +{"train_loss": 0.03782467916607857, "train_loss_bc": 0.03782467916607857, "train_loss_llm": 0.0, "grad_norm": 0.062286101281642914, "global_step": 19726, "epoch": 547, "lr": 0.001} +{"train_loss": 0.033550304257207446, "train_loss_bc": 0.03988366574048996, "train_loss_llm": 0.0007511675357818604, "grad_norm": 0.0872935950756073, "global_step": 19727, "epoch": 547, "lr": 0.001, "val_loss": 0.07895476371049881} +{"train_loss": 0.038334451615810394, "train_loss_bc": 0.038334451615810394, "train_loss_llm": 0.0, "grad_norm": 0.1076752096414566, "global_step": 19728, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03461837023496628, "train_loss_bc": 0.03461837023496628, "train_loss_llm": 0.0, "grad_norm": 0.030735721811652184, "global_step": 19729, "epoch": 548, "lr": 0.001} +{"train_loss": 0.028183015063405037, "train_loss_bc": 0.028183015063405037, "train_loss_llm": 0.0, "grad_norm": 0.03731045871973038, "global_step": 19730, "epoch": 548, "lr": 0.001} +{"train_loss": 0.02827930636703968, "train_loss_bc": 0.028274180367588997, "train_loss_llm": 0.000512540340423584, "grad_norm": 0.05065222829580307, "global_step": 19731, "epoch": 548, "lr": 0.001} +{"train_loss": 0.033632438629865646, "train_loss_bc": 0.033632438629865646, "train_loss_llm": 0.0, "grad_norm": 0.05923857167363167, "global_step": 19732, "epoch": 548, "lr": 0.001} +{"train_loss": 0.030468441545963287, "train_loss_bc": 0.030457912012934685, "train_loss_llm": 0.0010530352592468262, "grad_norm": 0.023055454716086388, "global_step": 19733, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03873622789978981, "train_loss_bc": 0.03873622789978981, "train_loss_llm": 0.0, "grad_norm": 0.032678455114364624, "global_step": 19734, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03863746300339699, "train_loss_bc": 0.03863746300339699, "train_loss_llm": 0.0, "grad_norm": 0.045445214956998825, "global_step": 19735, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03437507152557373, "train_loss_bc": 0.03437507152557373, "train_loss_llm": 0.0, "grad_norm": 0.056155599653720856, "global_step": 19736, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03004305064678192, "train_loss_bc": 0.030024591833353043, "train_loss_llm": 0.0018458962440490723, "grad_norm": 0.028019843623042107, "global_step": 19737, "epoch": 548, "lr": 0.001} +{"train_loss": 0.034303028136491776, "train_loss_bc": 0.034303028136491776, "train_loss_llm": 0.0, "grad_norm": 0.0454653725028038, "global_step": 19738, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03327758237719536, "train_loss_bc": 0.03327758237719536, "train_loss_llm": 0.0, "grad_norm": 0.0665152370929718, "global_step": 19739, "epoch": 548, "lr": 0.001} +{"train_loss": 0.02949737384915352, "train_loss_bc": 0.02949737384915352, "train_loss_llm": 0.0, "grad_norm": 0.07734990864992142, "global_step": 19740, "epoch": 548, "lr": 0.001} +{"train_loss": 0.033038340508937836, "train_loss_bc": 0.033038340508937836, "train_loss_llm": 0.0, "grad_norm": 0.029414799064397812, "global_step": 19741, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03683863580226898, "train_loss_bc": 0.03683863580226898, "train_loss_llm": 0.0, "grad_norm": 0.047003891319036484, "global_step": 19742, "epoch": 548, "lr": 0.001} +{"train_loss": 0.029955148696899414, "train_loss_bc": 0.029955148696899414, "train_loss_llm": 0.0, "grad_norm": 0.0700107142329216, "global_step": 19743, "epoch": 548, "lr": 0.001} +{"train_loss": 0.0330054797232151, "train_loss_bc": 0.0330054797232151, "train_loss_llm": 0.0, "grad_norm": 0.10153621435165405, "global_step": 19744, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03152981400489807, "train_loss_bc": 0.03152981400489807, "train_loss_llm": 0.0, "grad_norm": 0.026586607098579407, "global_step": 19745, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03672443702816963, "train_loss_bc": 0.03672443702816963, "train_loss_llm": 0.0, "grad_norm": 0.04580172896385193, "global_step": 19746, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03532341122627258, "train_loss_bc": 0.03530590608716011, "train_loss_llm": 0.0017505288124084473, "grad_norm": 0.0607534758746624, "global_step": 19747, "epoch": 548, "lr": 0.001} +{"train_loss": 0.035699669271707535, "train_loss_bc": 0.03568246215581894, "train_loss_llm": 0.0017205774784088135, "grad_norm": 0.07730147242546082, "global_step": 19748, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03607418015599251, "train_loss_bc": 0.03607418015599251, "train_loss_llm": 0.0, "grad_norm": 0.030527304857969284, "global_step": 19749, "epoch": 548, "lr": 0.001} +{"train_loss": 0.038437407463788986, "train_loss_bc": 0.03841415420174599, "train_loss_llm": 0.002325505018234253, "grad_norm": 0.053541939705610275, "global_step": 19750, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03576819598674774, "train_loss_bc": 0.03575141727924347, "train_loss_llm": 0.0016780495643615723, "grad_norm": 0.07239922881126404, "global_step": 19751, "epoch": 548, "lr": 0.001} +{"train_loss": 0.02991546504199505, "train_loss_bc": 0.02991546504199505, "train_loss_llm": 0.0, "grad_norm": 0.08355079591274261, "global_step": 19752, "epoch": 548, "lr": 0.001} +{"train_loss": 0.030618954449892044, "train_loss_bc": 0.030612196773290634, "train_loss_llm": 0.0006757974624633789, "grad_norm": 0.030792800709605217, "global_step": 19753, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03428085893392563, "train_loss_bc": 0.03427625820040703, "train_loss_llm": 0.00045990943908691406, "grad_norm": 0.04592055827379227, "global_step": 19754, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03496139496564865, "train_loss_bc": 0.03496139496564865, "train_loss_llm": 0.0, "grad_norm": 0.0640704557299614, "global_step": 19755, "epoch": 548, "lr": 0.001} +{"train_loss": 0.02840140275657177, "train_loss_bc": 0.02839629538357258, "train_loss_llm": 0.00051078200340271, "grad_norm": 0.07745697349309921, "global_step": 19756, "epoch": 548, "lr": 0.001} +{"train_loss": 0.027734527364373207, "train_loss_bc": 0.027734527364373207, "train_loss_llm": 0.0, "grad_norm": 0.02909233421087265, "global_step": 19757, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03256465494632721, "train_loss_bc": 0.03256465494632721, "train_loss_llm": 0.0, "grad_norm": 0.04703158140182495, "global_step": 19758, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03314482420682907, "train_loss_bc": 0.03313692659139633, "train_loss_llm": 0.0007897019386291504, "grad_norm": 0.05880783870816231, "global_step": 19759, "epoch": 548, "lr": 0.001} +{"train_loss": 0.036653511226177216, "train_loss_bc": 0.036653511226177216, "train_loss_llm": 0.0, "grad_norm": 0.07412619143724442, "global_step": 19760, "epoch": 548, "lr": 0.001} +{"train_loss": 0.027093984186649323, "train_loss_bc": 0.027072254568338394, "train_loss_llm": 0.0021728873252868652, "grad_norm": 0.027385495603084564, "global_step": 19761, "epoch": 548, "lr": 0.001} +{"train_loss": 0.03785661980509758, "train_loss_bc": 0.03784474730491638, "train_loss_llm": 0.0011873841285705566, "grad_norm": 0.04328348487615585, "global_step": 19762, "epoch": 548, "lr": 0.001} +{"train_loss": 0.033514831887765065, "train_loss_bc": 0.03852720931172371, "train_loss_llm": 0.0, "grad_norm": 0.07121064513921738, "global_step": 19763, "epoch": 548, "lr": 0.001, "val_loss": 0.07928022742271423} +{"train_loss": 0.02536863274872303, "train_loss_bc": 0.02536863274872303, "train_loss_llm": 0.0, "grad_norm": 0.0830470621585846, "global_step": 19764, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03105813078582287, "train_loss_bc": 0.03105813078582287, "train_loss_llm": 0.0, "grad_norm": 0.031982019543647766, "global_step": 19765, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03747108206152916, "train_loss_bc": 0.03747108206152916, "train_loss_llm": 0.0, "grad_norm": 0.056301623582839966, "global_step": 19766, "epoch": 549, "lr": 0.001} +{"train_loss": 0.032241545617580414, "train_loss_bc": 0.03221748396754265, "train_loss_llm": 0.0024061501026153564, "grad_norm": 0.06946583092212677, "global_step": 19767, "epoch": 549, "lr": 0.001} +{"train_loss": 0.029012538492679596, "train_loss_bc": 0.029012538492679596, "train_loss_llm": 0.0, "grad_norm": 0.08235632628202438, "global_step": 19768, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03293464705348015, "train_loss_bc": 0.03293464705348015, "train_loss_llm": 0.0, "grad_norm": 0.0269653107970953, "global_step": 19769, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03458801284432411, "train_loss_bc": 0.03458801284432411, "train_loss_llm": 0.0, "grad_norm": 0.0443791039288044, "global_step": 19770, "epoch": 549, "lr": 0.001} +{"train_loss": 0.027409737929701805, "train_loss_bc": 0.027393687516450882, "train_loss_llm": 0.0016049742698669434, "grad_norm": 0.051740612834692, "global_step": 19771, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03852904215455055, "train_loss_bc": 0.03852904215455055, "train_loss_llm": 0.0, "grad_norm": 0.06688926368951797, "global_step": 19772, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03843732923269272, "train_loss_bc": 0.03843732923269272, "train_loss_llm": 0.0, "grad_norm": 0.02798600308597088, "global_step": 19773, "epoch": 549, "lr": 0.001} +{"train_loss": 0.033625684678554535, "train_loss_bc": 0.033606432378292084, "train_loss_llm": 0.001925140619277954, "grad_norm": 0.04619119316339493, "global_step": 19774, "epoch": 549, "lr": 0.001} +{"train_loss": 0.04118501394987106, "train_loss_bc": 0.0411737821996212, "train_loss_llm": 0.0011231005191802979, "grad_norm": 0.061563994735479355, "global_step": 19775, "epoch": 549, "lr": 0.001} +{"train_loss": 0.027967169880867004, "train_loss_bc": 0.027967169880867004, "train_loss_llm": 0.0, "grad_norm": 0.07684896886348724, "global_step": 19776, "epoch": 549, "lr": 0.001} +{"train_loss": 0.033653080463409424, "train_loss_bc": 0.033653080463409424, "train_loss_llm": 0.0, "grad_norm": 0.03271970897912979, "global_step": 19777, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03236997872591019, "train_loss_bc": 0.03236997872591019, "train_loss_llm": 0.0, "grad_norm": 0.04835159704089165, "global_step": 19778, "epoch": 549, "lr": 0.001} +{"train_loss": 0.034978922456502914, "train_loss_bc": 0.034978076815605164, "train_loss_llm": 8.472800254821777e-05, "grad_norm": 0.06280739605426788, "global_step": 19779, "epoch": 549, "lr": 0.001} +{"train_loss": 0.032381571829319, "train_loss_bc": 0.03236880898475647, "train_loss_llm": 0.0012762248516082764, "grad_norm": 0.07350777834653854, "global_step": 19780, "epoch": 549, "lr": 0.001} +{"train_loss": 0.031837183982133865, "train_loss_bc": 0.031837183982133865, "train_loss_llm": 0.0, "grad_norm": 0.02806830033659935, "global_step": 19781, "epoch": 549, "lr": 0.001} +{"train_loss": 0.037042852491140366, "train_loss_bc": 0.037042852491140366, "train_loss_llm": 0.0, "grad_norm": 0.04943346977233887, "global_step": 19782, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03148575872182846, "train_loss_bc": 0.03148575872182846, "train_loss_llm": 0.0, "grad_norm": 0.06537949293851852, "global_step": 19783, "epoch": 549, "lr": 0.001} +{"train_loss": 0.0312916561961174, "train_loss_bc": 0.0312916561961174, "train_loss_llm": 0.0, "grad_norm": 0.08474703878164291, "global_step": 19784, "epoch": 549, "lr": 0.001} +{"train_loss": 0.04120144620537758, "train_loss_bc": 0.04120144620537758, "train_loss_llm": 0.0, "grad_norm": 0.03506218641996384, "global_step": 19785, "epoch": 549, "lr": 0.001} +{"train_loss": 0.029306799173355103, "train_loss_bc": 0.029306799173355103, "train_loss_llm": 0.0, "grad_norm": 0.045920416712760925, "global_step": 19786, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03045673668384552, "train_loss_bc": 0.03045673668384552, "train_loss_llm": 0.0, "grad_norm": 0.06533048301935196, "global_step": 19787, "epoch": 549, "lr": 0.001} +{"train_loss": 0.035519979894161224, "train_loss_bc": 0.035519979894161224, "train_loss_llm": 0.0, "grad_norm": 0.08397877216339111, "global_step": 19788, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03150084242224693, "train_loss_bc": 0.03150084242224693, "train_loss_llm": 0.0, "grad_norm": 0.02844785898923874, "global_step": 19789, "epoch": 549, "lr": 0.001} +{"train_loss": 0.036914534866809845, "train_loss_bc": 0.036914534866809845, "train_loss_llm": 0.0, "grad_norm": 0.040158383548259735, "global_step": 19790, "epoch": 549, "lr": 0.001} +{"train_loss": 0.02892177924513817, "train_loss_bc": 0.02892177924513817, "train_loss_llm": 0.0, "grad_norm": 0.05861301347613335, "global_step": 19791, "epoch": 549, "lr": 0.001} +{"train_loss": 0.034630730748176575, "train_loss_bc": 0.034630730748176575, "train_loss_llm": 0.0, "grad_norm": 0.07229892909526825, "global_step": 19792, "epoch": 549, "lr": 0.001} +{"train_loss": 0.033603470772504807, "train_loss_bc": 0.03360239043831825, "train_loss_llm": 0.00010794401168823242, "grad_norm": 0.02709823101758957, "global_step": 19793, "epoch": 549, "lr": 0.001} +{"train_loss": 0.036151785403490067, "train_loss_bc": 0.03615100309252739, "train_loss_llm": 7.829070091247559e-05, "grad_norm": 0.03999171033501625, "global_step": 19794, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03436298295855522, "train_loss_bc": 0.03434951603412628, "train_loss_llm": 0.0013467967510223389, "grad_norm": 0.05133336782455444, "global_step": 19795, "epoch": 549, "lr": 0.001} +{"train_loss": 0.02956010214984417, "train_loss_bc": 0.02956010214984417, "train_loss_llm": 0.0, "grad_norm": 0.06319282203912735, "global_step": 19796, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03942793980240822, "train_loss_bc": 0.03942793980240822, "train_loss_llm": 0.0, "grad_norm": 0.028068041428923607, "global_step": 19797, "epoch": 549, "lr": 0.001} +{"train_loss": 0.031056726351380348, "train_loss_bc": 0.031056726351380348, "train_loss_llm": 0.0, "grad_norm": 0.0372149758040905, "global_step": 19798, "epoch": 549, "lr": 0.001} +{"train_loss": 0.03341141027501888, "train_loss_bc": 0.03530935198068619, "train_loss_llm": 0.001599043607711792, "grad_norm": 0.05312461778521538, "global_step": 19799, "epoch": 549, "lr": 0.001, "val_loss": 0.07866300642490387} +{"train_loss": 0.038825903087854385, "train_loss_bc": 0.03882462531328201, "train_loss_llm": 0.0001277923583984375, "grad_norm": 0.07365360110998154, "global_step": 19800, "epoch": 550, "lr": 0.001} +{"train_loss": 0.034252386540174484, "train_loss_bc": 0.03423970565199852, "train_loss_llm": 0.001268237829208374, "grad_norm": 0.02272135205566883, "global_step": 19801, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03338675573468208, "train_loss_bc": 0.03338675573468208, "train_loss_llm": 0.0, "grad_norm": 0.04351012781262398, "global_step": 19802, "epoch": 550, "lr": 0.001} +{"train_loss": 0.031083442270755768, "train_loss_bc": 0.031083442270755768, "train_loss_llm": 0.0, "grad_norm": 0.05724114552140236, "global_step": 19803, "epoch": 550, "lr": 0.001} +{"train_loss": 0.033918384462594986, "train_loss_bc": 0.03391185775399208, "train_loss_llm": 0.0006526410579681396, "grad_norm": 0.07653164863586426, "global_step": 19804, "epoch": 550, "lr": 0.001} +{"train_loss": 0.029388997703790665, "train_loss_bc": 0.029388997703790665, "train_loss_llm": 0.0, "grad_norm": 0.026829343289136887, "global_step": 19805, "epoch": 550, "lr": 0.001} +{"train_loss": 0.032964736223220825, "train_loss_bc": 0.032964736223220825, "train_loss_llm": 0.0, "grad_norm": 0.042172398418188095, "global_step": 19806, "epoch": 550, "lr": 0.001} +{"train_loss": 0.030349459499120712, "train_loss_bc": 0.030349459499120712, "train_loss_llm": 0.0, "grad_norm": 0.06600606441497803, "global_step": 19807, "epoch": 550, "lr": 0.001} +{"train_loss": 0.034927498549222946, "train_loss_bc": 0.034927498549222946, "train_loss_llm": 0.0, "grad_norm": 0.07010062038898468, "global_step": 19808, "epoch": 550, "lr": 0.001} +{"train_loss": 0.039141591638326645, "train_loss_bc": 0.039141591638326645, "train_loss_llm": 0.0, "grad_norm": 0.0372462160885334, "global_step": 19809, "epoch": 550, "lr": 0.001} +{"train_loss": 0.039885539561510086, "train_loss_bc": 0.039885539561510086, "train_loss_llm": 0.0, "grad_norm": 0.0557282380759716, "global_step": 19810, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03229103237390518, "train_loss_bc": 0.03229103237390518, "train_loss_llm": 0.0, "grad_norm": 0.07320285588502884, "global_step": 19811, "epoch": 550, "lr": 0.001} +{"train_loss": 0.038078710436820984, "train_loss_bc": 0.038078710436820984, "train_loss_llm": 0.0, "grad_norm": 0.0811033621430397, "global_step": 19812, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03889467939734459, "train_loss_bc": 0.03886649012565613, "train_loss_llm": 0.0028189420700073242, "grad_norm": 0.029744701460003853, "global_step": 19813, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03363557904958725, "train_loss_bc": 0.03363557904958725, "train_loss_llm": 0.0, "grad_norm": 0.03848978504538536, "global_step": 19814, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03150027617812157, "train_loss_bc": 0.031497787684202194, "train_loss_llm": 0.00024896860122680664, "grad_norm": 0.04269479587674141, "global_step": 19815, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03460134193301201, "train_loss_bc": 0.03460134193301201, "train_loss_llm": 0.0, "grad_norm": 0.04955105483531952, "global_step": 19816, "epoch": 550, "lr": 0.001} +{"train_loss": 0.02440257929265499, "train_loss_bc": 0.02440257929265499, "train_loss_llm": 0.0, "grad_norm": 0.023501751944422722, "global_step": 19817, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03325415775179863, "train_loss_bc": 0.03324401006102562, "train_loss_llm": 0.0010147392749786377, "grad_norm": 0.03830547630786896, "global_step": 19818, "epoch": 550, "lr": 0.001} +{"train_loss": 0.030315838754177094, "train_loss_bc": 0.030315838754177094, "train_loss_llm": 0.0, "grad_norm": 0.053199075162410736, "global_step": 19819, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03578560799360275, "train_loss_bc": 0.035780973732471466, "train_loss_llm": 0.0004633963108062744, "grad_norm": 0.05959668010473251, "global_step": 19820, "epoch": 550, "lr": 0.001} +{"train_loss": 0.035249073058366776, "train_loss_bc": 0.035249073058366776, "train_loss_llm": 0.0, "grad_norm": 0.028349250555038452, "global_step": 19821, "epoch": 550, "lr": 0.001} +{"train_loss": 0.0254458449780941, "train_loss_bc": 0.025439923629164696, "train_loss_llm": 0.0005921721458435059, "grad_norm": 0.042350154370069504, "global_step": 19822, "epoch": 550, "lr": 0.001} +{"train_loss": 0.030604081228375435, "train_loss_bc": 0.03059159405529499, "train_loss_llm": 0.0012486577033996582, "grad_norm": 0.05552574247121811, "global_step": 19823, "epoch": 550, "lr": 0.001} +{"train_loss": 0.037500932812690735, "train_loss_bc": 0.037500932812690735, "train_loss_llm": 0.0, "grad_norm": 0.07737795263528824, "global_step": 19824, "epoch": 550, "lr": 0.001} +{"train_loss": 0.02802922949194908, "train_loss_bc": 0.02802922949194908, "train_loss_llm": 0.0, "grad_norm": 0.03419909253716469, "global_step": 19825, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03978947177529335, "train_loss_bc": 0.039789386093616486, "train_loss_llm": 8.434057235717773e-06, "grad_norm": 0.04737749695777893, "global_step": 19826, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03235494717955589, "train_loss_bc": 0.03235494717955589, "train_loss_llm": 0.0, "grad_norm": 0.06638623028993607, "global_step": 19827, "epoch": 550, "lr": 0.001} +{"train_loss": 0.0346856489777565, "train_loss_bc": 0.0346856489777565, "train_loss_llm": 0.0, "grad_norm": 0.08399716019630432, "global_step": 19828, "epoch": 550, "lr": 0.001} +{"train_loss": 0.029935985803604126, "train_loss_bc": 0.029935985803604126, "train_loss_llm": 0.0, "grad_norm": 0.029644792899489403, "global_step": 19829, "epoch": 550, "lr": 0.001} +{"train_loss": 0.02587100863456726, "train_loss_bc": 0.02587100863456726, "train_loss_llm": 0.0, "grad_norm": 0.045833658427000046, "global_step": 19830, "epoch": 550, "lr": 0.001} +{"train_loss": 0.0333426296710968, "train_loss_bc": 0.03333583474159241, "train_loss_llm": 0.0006796419620513916, "grad_norm": 0.04844841733574867, "global_step": 19831, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03742089495062828, "train_loss_bc": 0.03742089495062828, "train_loss_llm": 0.0, "grad_norm": 0.0648888647556305, "global_step": 19832, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03169681876897812, "train_loss_bc": 0.03169681876897812, "train_loss_llm": 0.0, "grad_norm": 0.028894640505313873, "global_step": 19833, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03254988044500351, "train_loss_bc": 0.03254988044500351, "train_loss_llm": 0.0, "grad_norm": 0.043971553444862366, "global_step": 19834, "epoch": 550, "lr": 0.001} +{"train_loss": 0.03321767417300078, "train_loss_bc": 0.030466144904494286, "train_loss_llm": 0.0009178519248962402, "grad_norm": 0.057527121156454086, "global_step": 19835, "epoch": 550, "lr": 0.001, "train/cumulative_reward": 85.43471527380666, "train/mean_score": 16.918684319025864, "train/success_rate": 0.4, "test/cumulative_reward": 32.07624539546128, "test/mean_score": 8.413957296271318, "test/success_rate": 0.2, "val_loss": 0.07784674316644669, "train_action_mse_error": 0.03590290993452072} +{"train_loss": 0.02965979278087616, "train_loss_bc": 0.02965979278087616, "train_loss_llm": 0.0, "grad_norm": 0.06810874491930008, "global_step": 19836, "epoch": 551, "lr": 0.001} +{"train_loss": 0.02890312671661377, "train_loss_bc": 0.0289025716483593, "train_loss_llm": 5.552172660827637e-05, "grad_norm": 0.02435600385069847, "global_step": 19837, "epoch": 551, "lr": 0.001} +{"train_loss": 0.032280951738357544, "train_loss_bc": 0.032280951738357544, "train_loss_llm": 0.0, "grad_norm": 0.036390334367752075, "global_step": 19838, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03377877175807953, "train_loss_bc": 0.03377877175807953, "train_loss_llm": 0.0, "grad_norm": 0.052064262330532074, "global_step": 19839, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03363737091422081, "train_loss_bc": 0.03363737091422081, "train_loss_llm": 0.0, "grad_norm": 0.06570398807525635, "global_step": 19840, "epoch": 551, "lr": 0.001} +{"train_loss": 0.02688218466937542, "train_loss_bc": 0.02688218466937542, "train_loss_llm": 0.0, "grad_norm": 0.025145500898361206, "global_step": 19841, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03019983321428299, "train_loss_bc": 0.03019983321428299, "train_loss_llm": 0.0, "grad_norm": 0.041577331721782684, "global_step": 19842, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03190923482179642, "train_loss_bc": 0.03190923482179642, "train_loss_llm": 0.0, "grad_norm": 0.06194455921649933, "global_step": 19843, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03615264967083931, "train_loss_bc": 0.03615264967083931, "train_loss_llm": 0.0, "grad_norm": 0.07229335606098175, "global_step": 19844, "epoch": 551, "lr": 0.001} +{"train_loss": 0.030527569353580475, "train_loss_bc": 0.030514732003211975, "train_loss_llm": 0.0012837350368499756, "grad_norm": 0.03684661537408829, "global_step": 19845, "epoch": 551, "lr": 0.001} +{"train_loss": 0.029866190627217293, "train_loss_bc": 0.029866190627217293, "train_loss_llm": 0.0, "grad_norm": 0.06176162511110306, "global_step": 19846, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03429557755589485, "train_loss_bc": 0.03429557755589485, "train_loss_llm": 0.0, "grad_norm": 0.08053163439035416, "global_step": 19847, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03261272609233856, "train_loss_bc": 0.03261272609233856, "train_loss_llm": 0.0, "grad_norm": 0.1043761745095253, "global_step": 19848, "epoch": 551, "lr": 0.001} +{"train_loss": 0.02991260401904583, "train_loss_bc": 0.029887383803725243, "train_loss_llm": 0.0025219619274139404, "grad_norm": 0.024589357897639275, "global_step": 19849, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03374180197715759, "train_loss_bc": 0.03374180197715759, "train_loss_llm": 0.0, "grad_norm": 0.05050995573401451, "global_step": 19850, "epoch": 551, "lr": 0.001} +{"train_loss": 0.029194580391049385, "train_loss_bc": 0.029189802706241608, "train_loss_llm": 0.00047773122787475586, "grad_norm": 0.06790915131568909, "global_step": 19851, "epoch": 551, "lr": 0.001} +{"train_loss": 0.038486238569021225, "train_loss_bc": 0.038462165743112564, "train_loss_llm": 0.0024074018001556396, "grad_norm": 0.07416892051696777, "global_step": 19852, "epoch": 551, "lr": 0.001} +{"train_loss": 0.04259556904435158, "train_loss_bc": 0.042586226016283035, "train_loss_llm": 0.0009341537952423096, "grad_norm": 0.03086024336516857, "global_step": 19853, "epoch": 551, "lr": 0.001} +{"train_loss": 0.026360411196947098, "train_loss_bc": 0.0263526514172554, "train_loss_llm": 0.0007759630680084229, "grad_norm": 0.049137432128190994, "global_step": 19854, "epoch": 551, "lr": 0.001} +{"train_loss": 0.036505699157714844, "train_loss_bc": 0.03650491312146187, "train_loss_llm": 7.846951484680176e-05, "grad_norm": 0.07453929632902145, "global_step": 19855, "epoch": 551, "lr": 0.001} +{"train_loss": 0.034091003239154816, "train_loss_bc": 0.034091003239154816, "train_loss_llm": 0.0, "grad_norm": 0.08728396892547607, "global_step": 19856, "epoch": 551, "lr": 0.001} +{"train_loss": 0.036971818655729294, "train_loss_bc": 0.036971818655729294, "train_loss_llm": 0.0, "grad_norm": 0.03050108253955841, "global_step": 19857, "epoch": 551, "lr": 0.001} +{"train_loss": 0.036259669810533524, "train_loss_bc": 0.036259669810533524, "train_loss_llm": 0.0, "grad_norm": 0.04740424081683159, "global_step": 19858, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03874421492218971, "train_loss_bc": 0.03874421492218971, "train_loss_llm": 0.0, "grad_norm": 0.07095565646886826, "global_step": 19859, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03119525872170925, "train_loss_bc": 0.03119525872170925, "train_loss_llm": 0.0, "grad_norm": 0.08885642141103745, "global_step": 19860, "epoch": 551, "lr": 0.001} +{"train_loss": 0.029460148885846138, "train_loss_bc": 0.02943606488406658, "train_loss_llm": 0.0024084746837615967, "grad_norm": 0.029584215953946114, "global_step": 19861, "epoch": 551, "lr": 0.001} +{"train_loss": 0.033620044589042664, "train_loss_bc": 0.033620044589042664, "train_loss_llm": 0.0, "grad_norm": 0.04817962646484375, "global_step": 19862, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03989516943693161, "train_loss_bc": 0.03988691419363022, "train_loss_llm": 0.0008256733417510986, "grad_norm": 0.06942674517631531, "global_step": 19863, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03544607385993004, "train_loss_bc": 0.03544607385993004, "train_loss_llm": 0.0, "grad_norm": 0.08464417606592178, "global_step": 19864, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03883744031190872, "train_loss_bc": 0.03883744031190872, "train_loss_llm": 0.0, "grad_norm": 0.03356386348605156, "global_step": 19865, "epoch": 551, "lr": 0.001} +{"train_loss": 0.02462073042988777, "train_loss_bc": 0.02461681142449379, "train_loss_llm": 0.0003918111324310303, "grad_norm": 0.046667806804180145, "global_step": 19866, "epoch": 551, "lr": 0.001} +{"train_loss": 0.04199022799730301, "train_loss_bc": 0.04199022799730301, "train_loss_llm": 0.0, "grad_norm": 0.06679531931877136, "global_step": 19867, "epoch": 551, "lr": 0.001} +{"train_loss": 0.033760614693164825, "train_loss_bc": 0.033760614693164825, "train_loss_llm": 0.0, "grad_norm": 0.08488857001066208, "global_step": 19868, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03319837898015976, "train_loss_bc": 0.03319837898015976, "train_loss_llm": 0.0, "grad_norm": 0.027448222041130066, "global_step": 19869, "epoch": 551, "lr": 0.001} +{"train_loss": 0.038544051349163055, "train_loss_bc": 0.038542550057172775, "train_loss_llm": 0.00015026330947875977, "grad_norm": 0.0379023477435112, "global_step": 19870, "epoch": 551, "lr": 0.001} +{"train_loss": 0.03327756323334244, "train_loss_bc": 0.02385454624891281, "train_loss_llm": 0.0, "grad_norm": 0.052444931119680405, "global_step": 19871, "epoch": 551, "lr": 0.001, "val_loss": 0.07861769199371338} +{"train_loss": 0.030539866536855698, "train_loss_bc": 0.030539866536855698, "train_loss_llm": 0.0, "grad_norm": 0.06647785753011703, "global_step": 19872, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03698711842298508, "train_loss_bc": 0.03698711842298508, "train_loss_llm": 0.0, "grad_norm": 0.03465424105525017, "global_step": 19873, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03345567733049393, "train_loss_bc": 0.03345567733049393, "train_loss_llm": 0.0, "grad_norm": 0.05250298231840134, "global_step": 19874, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03851712867617607, "train_loss_bc": 0.03851712867617607, "train_loss_llm": 0.0, "grad_norm": 0.07677292078733444, "global_step": 19875, "epoch": 552, "lr": 0.001} +{"train_loss": 0.0325995571911335, "train_loss_bc": 0.0325995571911335, "train_loss_llm": 0.0, "grad_norm": 0.09388884156942368, "global_step": 19876, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03580362722277641, "train_loss_bc": 0.03578187897801399, "train_loss_llm": 0.0021747946739196777, "grad_norm": 0.030139796435832977, "global_step": 19877, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03449377417564392, "train_loss_bc": 0.03449377417564392, "train_loss_llm": 0.0, "grad_norm": 0.0486304797232151, "global_step": 19878, "epoch": 552, "lr": 0.001} +{"train_loss": 0.04001511633396149, "train_loss_bc": 0.04001511633396149, "train_loss_llm": 0.0, "grad_norm": 0.06622474640607834, "global_step": 19879, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03124266304075718, "train_loss_bc": 0.03124266304075718, "train_loss_llm": 0.0, "grad_norm": 0.08073023706674576, "global_step": 19880, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03157756105065346, "train_loss_bc": 0.03157429024577141, "train_loss_llm": 0.00032719969749450684, "grad_norm": 0.027083929628133774, "global_step": 19881, "epoch": 552, "lr": 0.001} +{"train_loss": 0.02567501924932003, "train_loss_bc": 0.02567501924932003, "train_loss_llm": 0.0, "grad_norm": 0.046406980603933334, "global_step": 19882, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03259989246726036, "train_loss_bc": 0.03259989246726036, "train_loss_llm": 0.0, "grad_norm": 0.05865424498915672, "global_step": 19883, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03976453095674515, "train_loss_bc": 0.03976453095674515, "train_loss_llm": 0.0, "grad_norm": 0.07518725097179413, "global_step": 19884, "epoch": 552, "lr": 0.001} +{"train_loss": 0.04026716575026512, "train_loss_bc": 0.04026716575026512, "train_loss_llm": 0.0, "grad_norm": 0.03750372678041458, "global_step": 19885, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03663298115134239, "train_loss_bc": 0.03663298115134239, "train_loss_llm": 0.0, "grad_norm": 0.06101518124341965, "global_step": 19886, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03490939736366272, "train_loss_bc": 0.03490939736366272, "train_loss_llm": 0.0, "grad_norm": 0.08527674525976181, "global_step": 19887, "epoch": 552, "lr": 0.001} +{"train_loss": 0.032537732273340225, "train_loss_bc": 0.032531388103961945, "train_loss_llm": 0.0006342828273773193, "grad_norm": 0.10304983705282211, "global_step": 19888, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03486499935388565, "train_loss_bc": 0.03486499935388565, "train_loss_llm": 0.0, "grad_norm": 0.024951612576842308, "global_step": 19889, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03591835871338844, "train_loss_bc": 0.03591835871338844, "train_loss_llm": 0.0, "grad_norm": 0.04844512790441513, "global_step": 19890, "epoch": 552, "lr": 0.001} +{"train_loss": 0.02674288861453533, "train_loss_bc": 0.02674288861453533, "train_loss_llm": 0.0, "grad_norm": 0.056687694042921066, "global_step": 19891, "epoch": 552, "lr": 0.001} +{"train_loss": 0.0248337984085083, "train_loss_bc": 0.0248337984085083, "train_loss_llm": 0.0, "grad_norm": 0.06571374833583832, "global_step": 19892, "epoch": 552, "lr": 0.001} +{"train_loss": 0.030829912051558495, "train_loss_bc": 0.03081262856721878, "train_loss_llm": 0.0017283856868743896, "grad_norm": 0.03409711271524429, "global_step": 19893, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03258296847343445, "train_loss_bc": 0.03258296847343445, "train_loss_llm": 0.0, "grad_norm": 0.050314657390117645, "global_step": 19894, "epoch": 552, "lr": 0.001} +{"train_loss": 0.028746170923113823, "train_loss_bc": 0.028746170923113823, "train_loss_llm": 0.0, "grad_norm": 0.0640188455581665, "global_step": 19895, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03115168586373329, "train_loss_bc": 0.031150076538324356, "train_loss_llm": 0.00016096234321594238, "grad_norm": 0.07558534294366837, "global_step": 19896, "epoch": 552, "lr": 0.001} +{"train_loss": 0.036903731524944305, "train_loss_bc": 0.036903731524944305, "train_loss_llm": 0.0, "grad_norm": 0.035653602331876755, "global_step": 19897, "epoch": 552, "lr": 0.001} +{"train_loss": 0.0335703045129776, "train_loss_bc": 0.03355403244495392, "train_loss_llm": 0.0016273856163024902, "grad_norm": 0.05172712355852127, "global_step": 19898, "epoch": 552, "lr": 0.001} +{"train_loss": 0.030925307422876358, "train_loss_bc": 0.030925307422876358, "train_loss_llm": 0.0, "grad_norm": 0.06545484811067581, "global_step": 19899, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03431267663836479, "train_loss_bc": 0.03431267663836479, "train_loss_llm": 0.0, "grad_norm": 0.08247502148151398, "global_step": 19900, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03230554610490799, "train_loss_bc": 0.03230554610490799, "train_loss_llm": 0.0, "grad_norm": 0.032339394092559814, "global_step": 19901, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03226137533783913, "train_loss_bc": 0.03226137533783913, "train_loss_llm": 0.0, "grad_norm": 0.05308949574828148, "global_step": 19902, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03307782858610153, "train_loss_bc": 0.033064424991607666, "train_loss_llm": 0.0013405084609985352, "grad_norm": 0.07085469365119934, "global_step": 19903, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03530983626842499, "train_loss_bc": 0.03530983626842499, "train_loss_llm": 0.0, "grad_norm": 0.08521870523691177, "global_step": 19904, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03153010085225105, "train_loss_bc": 0.03153010085225105, "train_loss_llm": 0.0, "grad_norm": 0.035168908536434174, "global_step": 19905, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03421701863408089, "train_loss_bc": 0.03421701863408089, "train_loss_llm": 0.0, "grad_norm": 0.05495622009038925, "global_step": 19906, "epoch": 552, "lr": 0.001} +{"train_loss": 0.03338750353496936, "train_loss_bc": 0.03424680978059769, "train_loss_llm": 0.0, "grad_norm": 0.06629014760255814, "global_step": 19907, "epoch": 552, "lr": 0.001, "val_loss": 0.07773347944021225} +{"train_loss": 0.039151888340711594, "train_loss_bc": 0.039151888340711594, "train_loss_llm": 0.0, "grad_norm": 0.09239187836647034, "global_step": 19908, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03518816456198692, "train_loss_bc": 0.035184990614652634, "train_loss_llm": 0.000317305326461792, "grad_norm": 0.038238074630498886, "global_step": 19909, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03265121951699257, "train_loss_bc": 0.03265121951699257, "train_loss_llm": 0.0, "grad_norm": 0.05450421944260597, "global_step": 19910, "epoch": 553, "lr": 0.001} +{"train_loss": 0.036372411996126175, "train_loss_bc": 0.036372411996126175, "train_loss_llm": 0.0, "grad_norm": 0.08016584068536758, "global_step": 19911, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03333611041307449, "train_loss_bc": 0.03333611041307449, "train_loss_llm": 0.0, "grad_norm": 0.0973280519247055, "global_step": 19912, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03769902139902115, "train_loss_bc": 0.03769902139902115, "train_loss_llm": 0.0, "grad_norm": 0.033095914870500565, "global_step": 19913, "epoch": 553, "lr": 0.001} +{"train_loss": 0.039431098848581314, "train_loss_bc": 0.039431098848581314, "train_loss_llm": 0.0, "grad_norm": 0.05564974620938301, "global_step": 19914, "epoch": 553, "lr": 0.001} +{"train_loss": 0.02815129980444908, "train_loss_bc": 0.02815129980444908, "train_loss_llm": 0.0, "grad_norm": 0.06993826478719711, "global_step": 19915, "epoch": 553, "lr": 0.001} +{"train_loss": 0.039391182363033295, "train_loss_bc": 0.039389677345752716, "train_loss_llm": 0.00015050172805786133, "grad_norm": 0.09176265448331833, "global_step": 19916, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03576589375734329, "train_loss_bc": 0.03576589375734329, "train_loss_llm": 0.0, "grad_norm": 0.02387118712067604, "global_step": 19917, "epoch": 553, "lr": 0.001} +{"train_loss": 0.02583504095673561, "train_loss_bc": 0.025822270661592484, "train_loss_llm": 0.0012770295143127441, "grad_norm": 0.032036714255809784, "global_step": 19918, "epoch": 553, "lr": 0.001} +{"train_loss": 0.0314662866294384, "train_loss_bc": 0.03146051988005638, "train_loss_llm": 0.0005768537521362305, "grad_norm": 0.045806292444467545, "global_step": 19919, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03533802554011345, "train_loss_bc": 0.03533802554011345, "train_loss_llm": 0.0, "grad_norm": 0.06306710094213486, "global_step": 19920, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03061697632074356, "train_loss_bc": 0.03061697632074356, "train_loss_llm": 0.0, "grad_norm": 0.029549479484558105, "global_step": 19921, "epoch": 553, "lr": 0.001} +{"train_loss": 0.023644080385565758, "train_loss_bc": 0.023644080385565758, "train_loss_llm": 0.0, "grad_norm": 0.05110251158475876, "global_step": 19922, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03490259125828743, "train_loss_bc": 0.03490259125828743, "train_loss_llm": 0.0, "grad_norm": 0.0764213353395462, "global_step": 19923, "epoch": 553, "lr": 0.001} +{"train_loss": 0.0320541113615036, "train_loss_bc": 0.032041650265455246, "train_loss_llm": 0.001245945692062378, "grad_norm": 0.08816369622945786, "global_step": 19924, "epoch": 553, "lr": 0.001} +{"train_loss": 0.034461598843336105, "train_loss_bc": 0.034461598843336105, "train_loss_llm": 0.0, "grad_norm": 0.033377453684806824, "global_step": 19925, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03447024151682854, "train_loss_bc": 0.03447024151682854, "train_loss_llm": 0.0, "grad_norm": 0.04991709440946579, "global_step": 19926, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03051275759935379, "train_loss_bc": 0.03051275759935379, "train_loss_llm": 0.0, "grad_norm": 0.059954509139060974, "global_step": 19927, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03339560702443123, "train_loss_bc": 0.033380430191755295, "train_loss_llm": 0.0015177726745605469, "grad_norm": 0.06992024183273315, "global_step": 19928, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03609360009431839, "train_loss_bc": 0.03608682379126549, "train_loss_llm": 0.0006776750087738037, "grad_norm": 0.0299000795930624, "global_step": 19929, "epoch": 553, "lr": 0.001} +{"train_loss": 0.02920582704246044, "train_loss_bc": 0.02920582704246044, "train_loss_llm": 0.0, "grad_norm": 0.0473577119410038, "global_step": 19930, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03604261577129364, "train_loss_bc": 0.036037035286426544, "train_loss_llm": 0.0005579292774200439, "grad_norm": 0.07104744762182236, "global_step": 19931, "epoch": 553, "lr": 0.001} +{"train_loss": 0.033611927181482315, "train_loss_bc": 0.033611927181482315, "train_loss_llm": 0.0, "grad_norm": 0.08735153824090958, "global_step": 19932, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03654031455516815, "train_loss_bc": 0.03653618320822716, "train_loss_llm": 0.0004131793975830078, "grad_norm": 0.022896939888596535, "global_step": 19933, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03205104544758797, "train_loss_bc": 0.03205104544758797, "train_loss_llm": 0.0, "grad_norm": 0.0414789654314518, "global_step": 19934, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03217011317610741, "train_loss_bc": 0.03217011317610741, "train_loss_llm": 0.0, "grad_norm": 0.06592831760644913, "global_step": 19935, "epoch": 553, "lr": 0.001} +{"train_loss": 0.028588278219103813, "train_loss_bc": 0.028588278219103813, "train_loss_llm": 0.0, "grad_norm": 0.07735943049192429, "global_step": 19936, "epoch": 553, "lr": 0.001} +{"train_loss": 0.0345952995121479, "train_loss_bc": 0.03456830978393555, "train_loss_llm": 0.002698957920074463, "grad_norm": 0.03672850504517555, "global_step": 19937, "epoch": 553, "lr": 0.001} +{"train_loss": 0.02891451120376587, "train_loss_bc": 0.02891451120376587, "train_loss_llm": 0.0, "grad_norm": 0.052692711353302, "global_step": 19938, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03469344228506088, "train_loss_bc": 0.03469344228506088, "train_loss_llm": 0.0, "grad_norm": 0.0667005106806755, "global_step": 19939, "epoch": 553, "lr": 0.001} +{"train_loss": 0.0349264070391655, "train_loss_bc": 0.0349264070391655, "train_loss_llm": 0.0, "grad_norm": 0.08755916357040405, "global_step": 19940, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03623185306787491, "train_loss_bc": 0.03623185306787491, "train_loss_llm": 0.0, "grad_norm": 0.03078276664018631, "global_step": 19941, "epoch": 553, "lr": 0.001} +{"train_loss": 0.02814299426972866, "train_loss_bc": 0.02814299426972866, "train_loss_llm": 0.0, "grad_norm": 0.04427523910999298, "global_step": 19942, "epoch": 553, "lr": 0.001} +{"train_loss": 0.03347535969482528, "train_loss_bc": 0.03946782276034355, "train_loss_llm": 0.00012892484664916992, "grad_norm": 0.07340432703495026, "global_step": 19943, "epoch": 553, "lr": 0.001, "val_loss": 0.0781329870223999} +{"train_loss": 0.035342585295438766, "train_loss_bc": 0.035342585295438766, "train_loss_llm": 0.0, "grad_norm": 0.09177964180707932, "global_step": 19944, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03659401834011078, "train_loss_bc": 0.03659401834011078, "train_loss_llm": 0.0, "grad_norm": 0.027879875153303146, "global_step": 19945, "epoch": 554, "lr": 0.001} +{"train_loss": 0.031049935147166252, "train_loss_bc": 0.031049935147166252, "train_loss_llm": 0.0, "grad_norm": 0.04378100857138634, "global_step": 19946, "epoch": 554, "lr": 0.001} +{"train_loss": 0.0339677594602108, "train_loss_bc": 0.03396008163690567, "train_loss_llm": 0.0007678866386413574, "grad_norm": 0.04979526996612549, "global_step": 19947, "epoch": 554, "lr": 0.001} +{"train_loss": 0.026677437126636505, "train_loss_bc": 0.02666620910167694, "train_loss_llm": 0.0011227726936340332, "grad_norm": 0.057974923402071, "global_step": 19948, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03398064523935318, "train_loss_bc": 0.03398064523935318, "train_loss_llm": 0.0, "grad_norm": 0.03380716219544411, "global_step": 19949, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03482387214899063, "train_loss_bc": 0.034802813082933426, "train_loss_llm": 0.002106010913848877, "grad_norm": 0.044533152133226395, "global_step": 19950, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03272493556141853, "train_loss_bc": 0.03272493556141853, "train_loss_llm": 0.0, "grad_norm": 0.05795678868889809, "global_step": 19951, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03420475125312805, "train_loss_bc": 0.03418734297156334, "train_loss_llm": 0.0017407536506652832, "grad_norm": 0.06929910182952881, "global_step": 19952, "epoch": 554, "lr": 0.001} +{"train_loss": 0.030888061970472336, "train_loss_bc": 0.030888061970472336, "train_loss_llm": 0.0, "grad_norm": 0.02585355006158352, "global_step": 19953, "epoch": 554, "lr": 0.001} +{"train_loss": 0.029355021193623543, "train_loss_bc": 0.029355021193623543, "train_loss_llm": 0.0, "grad_norm": 0.03680714964866638, "global_step": 19954, "epoch": 554, "lr": 0.001} +{"train_loss": 0.036757998168468475, "train_loss_bc": 0.03675152733922005, "train_loss_llm": 0.0006472170352935791, "grad_norm": 0.04913729429244995, "global_step": 19955, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03358827531337738, "train_loss_bc": 0.03358827531337738, "train_loss_llm": 0.0, "grad_norm": 0.06854242831468582, "global_step": 19956, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03532266616821289, "train_loss_bc": 0.03532266616821289, "train_loss_llm": 0.0, "grad_norm": 0.03469553962349892, "global_step": 19957, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03959338739514351, "train_loss_bc": 0.03959338739514351, "train_loss_llm": 0.0, "grad_norm": 0.05810287222266197, "global_step": 19958, "epoch": 554, "lr": 0.001} +{"train_loss": 0.029446855187416077, "train_loss_bc": 0.029444372281432152, "train_loss_llm": 0.00024831295013427734, "grad_norm": 0.0789186879992485, "global_step": 19959, "epoch": 554, "lr": 0.001} +{"train_loss": 0.04275735840201378, "train_loss_bc": 0.04275735840201378, "train_loss_llm": 0.0, "grad_norm": 0.09331781417131424, "global_step": 19960, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03002816252410412, "train_loss_bc": 0.03002816252410412, "train_loss_llm": 0.0, "grad_norm": 0.03067714162170887, "global_step": 19961, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03051803819835186, "train_loss_bc": 0.03051803819835186, "train_loss_llm": 0.0, "grad_norm": 0.05261491984128952, "global_step": 19962, "epoch": 554, "lr": 0.001} +{"train_loss": 0.0324275903403759, "train_loss_bc": 0.0324275903403759, "train_loss_llm": 0.0, "grad_norm": 0.0685771107673645, "global_step": 19963, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03179222345352173, "train_loss_bc": 0.03179222345352173, "train_loss_llm": 0.0, "grad_norm": 0.08832445740699768, "global_step": 19964, "epoch": 554, "lr": 0.001} +{"train_loss": 0.028940048068761826, "train_loss_bc": 0.028940048068761826, "train_loss_llm": 0.0, "grad_norm": 0.02596038207411766, "global_step": 19965, "epoch": 554, "lr": 0.001} +{"train_loss": 0.031693052500486374, "train_loss_bc": 0.031693052500486374, "train_loss_llm": 0.0, "grad_norm": 0.040538396686315536, "global_step": 19966, "epoch": 554, "lr": 0.001} +{"train_loss": 0.032211896032094955, "train_loss_bc": 0.032211896032094955, "train_loss_llm": 0.0, "grad_norm": 0.05336834862828255, "global_step": 19967, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03224680572748184, "train_loss_bc": 0.03224680572748184, "train_loss_llm": 0.0, "grad_norm": 0.06297022104263306, "global_step": 19968, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03644895926117897, "train_loss_bc": 0.03644895926117897, "train_loss_llm": 0.0, "grad_norm": 0.033615339547395706, "global_step": 19969, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03408610075712204, "train_loss_bc": 0.03408610075712204, "train_loss_llm": 0.0, "grad_norm": 0.04556138813495636, "global_step": 19970, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03248540312051773, "train_loss_bc": 0.03248540312051773, "train_loss_llm": 0.0, "grad_norm": 0.06570475548505783, "global_step": 19971, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03643261268734932, "train_loss_bc": 0.03643261268734932, "train_loss_llm": 0.0, "grad_norm": 0.08657928556203842, "global_step": 19972, "epoch": 554, "lr": 0.001} +{"train_loss": 0.028430325910449028, "train_loss_bc": 0.028430325910449028, "train_loss_llm": 0.0, "grad_norm": 0.028164315968751907, "global_step": 19973, "epoch": 554, "lr": 0.001} +{"train_loss": 0.025960151106119156, "train_loss_bc": 0.02595198154449463, "train_loss_llm": 0.0008169710636138916, "grad_norm": 0.03924664482474327, "global_step": 19974, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03635553643107414, "train_loss_bc": 0.03635553643107414, "train_loss_llm": 0.0, "grad_norm": 0.06003303453326225, "global_step": 19975, "epoch": 554, "lr": 0.001} +{"train_loss": 0.031156348064541817, "train_loss_bc": 0.031156348064541817, "train_loss_llm": 0.0, "grad_norm": 0.0726129338145256, "global_step": 19976, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03599361702799797, "train_loss_bc": 0.035993076860904694, "train_loss_llm": 5.3942203521728516e-05, "grad_norm": 0.0350346639752388, "global_step": 19977, "epoch": 554, "lr": 0.001} +{"train_loss": 0.03827088698744774, "train_loss_bc": 0.03827088698744774, "train_loss_llm": 0.0, "grad_norm": 0.04833786562085152, "global_step": 19978, "epoch": 554, "lr": 0.001} +{"train_loss": 0.033288222530649766, "train_loss_bc": 0.0358111709356308, "train_loss_llm": 0.00115203857421875, "grad_norm": 0.07170961797237396, "global_step": 19979, "epoch": 554, "lr": 0.001, "val_loss": 0.08046306669712067} +{"train_loss": 0.04087597504258156, "train_loss_bc": 0.04087597504258156, "train_loss_llm": 0.0, "grad_norm": 0.0897158682346344, "global_step": 19980, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03654602915048599, "train_loss_bc": 0.036544300615787506, "train_loss_llm": 0.00017279386520385742, "grad_norm": 0.027449866756796837, "global_step": 19981, "epoch": 555, "lr": 0.001} +{"train_loss": 0.034928396344184875, "train_loss_bc": 0.034921519458293915, "train_loss_llm": 0.0006875395774841309, "grad_norm": 0.05006726086139679, "global_step": 19982, "epoch": 555, "lr": 0.001} +{"train_loss": 0.028087574988603592, "train_loss_bc": 0.028087574988603592, "train_loss_llm": 0.0, "grad_norm": 0.06722670048475266, "global_step": 19983, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03138118237257004, "train_loss_bc": 0.03138118237257004, "train_loss_llm": 0.0, "grad_norm": 0.08156099170446396, "global_step": 19984, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03089788928627968, "train_loss_bc": 0.03088955767452717, "train_loss_llm": 0.0008330941200256348, "grad_norm": 0.02638152800500393, "global_step": 19985, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03686998412013054, "train_loss_bc": 0.03685728460550308, "train_loss_llm": 0.0012698769569396973, "grad_norm": 0.03912109136581421, "global_step": 19986, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03443939611315727, "train_loss_bc": 0.03443939611315727, "train_loss_llm": 0.0, "grad_norm": 0.05129483342170715, "global_step": 19987, "epoch": 555, "lr": 0.001} +{"train_loss": 0.0346614234149456, "train_loss_bc": 0.0346614234149456, "train_loss_llm": 0.0, "grad_norm": 0.0647878646850586, "global_step": 19988, "epoch": 555, "lr": 0.001} +{"train_loss": 0.0329989418387413, "train_loss_bc": 0.0329989418387413, "train_loss_llm": 0.0, "grad_norm": 0.025655928999185562, "global_step": 19989, "epoch": 555, "lr": 0.001} +{"train_loss": 0.02942502312362194, "train_loss_bc": 0.02942502312362194, "train_loss_llm": 0.0, "grad_norm": 0.04608700051903725, "global_step": 19990, "epoch": 555, "lr": 0.001} +{"train_loss": 0.034486740827560425, "train_loss_bc": 0.034486740827560425, "train_loss_llm": 0.0, "grad_norm": 0.0639396384358406, "global_step": 19991, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03161861374974251, "train_loss_bc": 0.031603433191776276, "train_loss_llm": 0.0015180408954620361, "grad_norm": 0.0761638805270195, "global_step": 19992, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03176026791334152, "train_loss_bc": 0.03176026791334152, "train_loss_llm": 0.0, "grad_norm": 0.0357857346534729, "global_step": 19993, "epoch": 555, "lr": 0.001} +{"train_loss": 0.031035510823130608, "train_loss_bc": 0.031013326719403267, "train_loss_llm": 0.002218395471572876, "grad_norm": 0.051481693983078, "global_step": 19994, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03794298321008682, "train_loss_bc": 0.03794298321008682, "train_loss_llm": 0.0, "grad_norm": 0.05882330611348152, "global_step": 19995, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03197174146771431, "train_loss_bc": 0.03197174146771431, "train_loss_llm": 0.0, "grad_norm": 0.06726378202438354, "global_step": 19996, "epoch": 555, "lr": 0.001} +{"train_loss": 0.032163698226213455, "train_loss_bc": 0.032163698226213455, "train_loss_llm": 0.0, "grad_norm": 0.03171864151954651, "global_step": 19997, "epoch": 555, "lr": 0.001} +{"train_loss": 0.02906760945916176, "train_loss_bc": 0.02906760945916176, "train_loss_llm": 0.0, "grad_norm": 0.03805294632911682, "global_step": 19998, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03198447823524475, "train_loss_bc": 0.03197664022445679, "train_loss_llm": 0.0007837414741516113, "grad_norm": 0.049773458391427994, "global_step": 19999, "epoch": 555, "lr": 0.001} +{"train_loss": 0.04393569007515907, "train_loss_bc": 0.04393569007515907, "train_loss_llm": 0.0, "grad_norm": 0.05926036462187767, "global_step": 20000, "epoch": 555, "lr": 0.001} +{"train_loss": 0.035753536969423294, "train_loss_bc": 0.035753536969423294, "train_loss_llm": 0.0, "grad_norm": 0.028763478621840477, "global_step": 20001, "epoch": 555, "lr": 0.001} +{"train_loss": 0.029818926006555557, "train_loss_bc": 0.029818926006555557, "train_loss_llm": 0.0, "grad_norm": 0.04245571792125702, "global_step": 20002, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03177259489893913, "train_loss_bc": 0.03177259489893913, "train_loss_llm": 0.0, "grad_norm": 0.05405576154589653, "global_step": 20003, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03342089056968689, "train_loss_bc": 0.03342089056968689, "train_loss_llm": 0.0, "grad_norm": 0.06577934324741364, "global_step": 20004, "epoch": 555, "lr": 0.001} +{"train_loss": 0.033349037170410156, "train_loss_bc": 0.033349037170410156, "train_loss_llm": 0.0, "grad_norm": 0.02793055772781372, "global_step": 20005, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03706185892224312, "train_loss_bc": 0.03706185892224312, "train_loss_llm": 0.0, "grad_norm": 0.042311280965805054, "global_step": 20006, "epoch": 555, "lr": 0.001} +{"train_loss": 0.030810730531811714, "train_loss_bc": 0.030810730531811714, "train_loss_llm": 0.0, "grad_norm": 0.056696560233831406, "global_step": 20007, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03542398661375046, "train_loss_bc": 0.03540681675076485, "train_loss_llm": 0.0017168521881103516, "grad_norm": 0.06897247582674026, "global_step": 20008, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03321986272931099, "train_loss_bc": 0.03321986272931099, "train_loss_llm": 0.0, "grad_norm": 0.02471734583377838, "global_step": 20009, "epoch": 555, "lr": 0.001} +{"train_loss": 0.029582997784018517, "train_loss_bc": 0.029572611674666405, "train_loss_llm": 0.001038581132888794, "grad_norm": 0.046482112258672714, "global_step": 20010, "epoch": 555, "lr": 0.001} +{"train_loss": 0.02909495308995247, "train_loss_bc": 0.029080118983983994, "train_loss_llm": 0.0014834105968475342, "grad_norm": 0.06166375055909157, "global_step": 20011, "epoch": 555, "lr": 0.001} +{"train_loss": 0.030139490962028503, "train_loss_bc": 0.030139490962028503, "train_loss_llm": 0.0, "grad_norm": 0.07507229596376419, "global_step": 20012, "epoch": 555, "lr": 0.001} +{"train_loss": 0.028421804308891296, "train_loss_bc": 0.02842169627547264, "train_loss_llm": 1.0758638381958008e-05, "grad_norm": 0.027592333033680916, "global_step": 20013, "epoch": 555, "lr": 0.001} +{"train_loss": 0.03632538393139839, "train_loss_bc": 0.03632538393139839, "train_loss_llm": 0.0, "grad_norm": 0.0406804084777832, "global_step": 20014, "epoch": 555, "lr": 0.001} +{"train_loss": 0.033140304084453315, "train_loss_bc": 0.03177574276924133, "train_loss_llm": 0.0, "grad_norm": 0.04733559116721153, "global_step": 20015, "epoch": 555, "lr": 0.001, "train/cumulative_reward": 104.67979369128486, "train/mean_score": 18.917091527654016, "train/success_rate": 0.5, "test/cumulative_reward": 32.522278231644684, "test/mean_score": 9.46971192120326, "test/success_rate": 0.26, "val_loss": 0.07811348140239716, "train_action_mse_error": 0.0358930379152298} +{"train_loss": 0.034118153154850006, "train_loss_bc": 0.034118153154850006, "train_loss_llm": 0.0, "grad_norm": 0.05969872698187828, "global_step": 20016, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03539664298295975, "train_loss_bc": 0.03539664298295975, "train_loss_llm": 0.0, "grad_norm": 0.0262927133589983, "global_step": 20017, "epoch": 556, "lr": 0.001} +{"train_loss": 0.02561456523835659, "train_loss_bc": 0.02561456523835659, "train_loss_llm": 0.0, "grad_norm": 0.03421289101243019, "global_step": 20018, "epoch": 556, "lr": 0.001} +{"train_loss": 0.028813017532229424, "train_loss_bc": 0.028805194422602654, "train_loss_llm": 0.0007822513580322266, "grad_norm": 0.044312722980976105, "global_step": 20019, "epoch": 556, "lr": 0.001} +{"train_loss": 0.029122620820999146, "train_loss_bc": 0.029122620820999146, "train_loss_llm": 0.0, "grad_norm": 0.05356607958674431, "global_step": 20020, "epoch": 556, "lr": 0.001} +{"train_loss": 0.036503866314888, "train_loss_bc": 0.036503866314888, "train_loss_llm": 0.0, "grad_norm": 0.03417658433318138, "global_step": 20021, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03778444975614548, "train_loss_bc": 0.03778444975614548, "train_loss_llm": 0.0, "grad_norm": 0.056932516396045685, "global_step": 20022, "epoch": 556, "lr": 0.001} +{"train_loss": 0.026599591597914696, "train_loss_bc": 0.026599591597914696, "train_loss_llm": 0.0, "grad_norm": 0.06352058053016663, "global_step": 20023, "epoch": 556, "lr": 0.001} +{"train_loss": 0.033670224249362946, "train_loss_bc": 0.03366212919354439, "train_loss_llm": 0.0008094608783721924, "grad_norm": 0.07841699570417404, "global_step": 20024, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03879895061254501, "train_loss_bc": 0.038793694227933884, "train_loss_llm": 0.0005257725715637207, "grad_norm": 0.030592912808060646, "global_step": 20025, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03092363476753235, "train_loss_bc": 0.03092363476753235, "train_loss_llm": 0.0, "grad_norm": 0.039467643946409225, "global_step": 20026, "epoch": 556, "lr": 0.001} +{"train_loss": 0.034112896770238876, "train_loss_bc": 0.034112896770238876, "train_loss_llm": 0.0, "grad_norm": 0.05375494062900543, "global_step": 20027, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03101707063615322, "train_loss_bc": 0.03101707063615322, "train_loss_llm": 0.0, "grad_norm": 0.055605851113796234, "global_step": 20028, "epoch": 556, "lr": 0.001} +{"train_loss": 0.037218812853097916, "train_loss_bc": 0.037218812853097916, "train_loss_llm": 0.0, "grad_norm": 0.027618922293186188, "global_step": 20029, "epoch": 556, "lr": 0.001} +{"train_loss": 0.030728131532669067, "train_loss_bc": 0.030728131532669067, "train_loss_llm": 0.0, "grad_norm": 0.041363779455423355, "global_step": 20030, "epoch": 556, "lr": 0.001} +{"train_loss": 0.02936636283993721, "train_loss_bc": 0.02936636283993721, "train_loss_llm": 0.0, "grad_norm": 0.0560213178396225, "global_step": 20031, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03464781865477562, "train_loss_bc": 0.03464781865477562, "train_loss_llm": 0.0, "grad_norm": 0.06512576341629028, "global_step": 20032, "epoch": 556, "lr": 0.001} +{"train_loss": 0.037456754595041275, "train_loss_bc": 0.037456754595041275, "train_loss_llm": 0.0, "grad_norm": 0.032578691840171814, "global_step": 20033, "epoch": 556, "lr": 0.001} +{"train_loss": 0.026528313755989075, "train_loss_bc": 0.026528313755989075, "train_loss_llm": 0.0, "grad_norm": 0.042258404195308685, "global_step": 20034, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03263521566987038, "train_loss_bc": 0.03263521566987038, "train_loss_llm": 0.0, "grad_norm": 0.050053782761096954, "global_step": 20035, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03156549111008644, "train_loss_bc": 0.0315614715218544, "train_loss_llm": 0.00040209293365478516, "grad_norm": 0.061723265796899796, "global_step": 20036, "epoch": 556, "lr": 0.001} +{"train_loss": 0.028776591643691063, "train_loss_bc": 0.0287765059620142, "train_loss_llm": 8.493661880493164e-06, "grad_norm": 0.028018197044730186, "global_step": 20037, "epoch": 556, "lr": 0.001} +{"train_loss": 0.0376395545899868, "train_loss_bc": 0.0376395545899868, "train_loss_llm": 0.0, "grad_norm": 0.0409754142165184, "global_step": 20038, "epoch": 556, "lr": 0.001} +{"train_loss": 0.031845495104789734, "train_loss_bc": 0.03183183819055557, "train_loss_llm": 0.0013658702373504639, "grad_norm": 0.05156084522604942, "global_step": 20039, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03316862881183624, "train_loss_bc": 0.03316862881183624, "train_loss_llm": 0.0, "grad_norm": 0.07454248517751694, "global_step": 20040, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03507109731435776, "train_loss_bc": 0.03507109731435776, "train_loss_llm": 0.0, "grad_norm": 0.03318843990564346, "global_step": 20041, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03223040699958801, "train_loss_bc": 0.03223040699958801, "train_loss_llm": 0.0, "grad_norm": 0.04968570917844772, "global_step": 20042, "epoch": 556, "lr": 0.001} +{"train_loss": 0.04116744175553322, "train_loss_bc": 0.04116744175553322, "train_loss_llm": 0.0, "grad_norm": 0.06245875731110573, "global_step": 20043, "epoch": 556, "lr": 0.001} +{"train_loss": 0.036563750356435776, "train_loss_bc": 0.036527570337057114, "train_loss_llm": 0.0036181211471557617, "grad_norm": 0.07421235740184784, "global_step": 20044, "epoch": 556, "lr": 0.001} +{"train_loss": 0.04008306935429573, "train_loss_bc": 0.04008306935429573, "train_loss_llm": 0.0, "grad_norm": 0.027326950803399086, "global_step": 20045, "epoch": 556, "lr": 0.001} +{"train_loss": 0.035808153450489044, "train_loss_bc": 0.03578341379761696, "train_loss_llm": 0.0024740993976593018, "grad_norm": 0.04511088877916336, "global_step": 20046, "epoch": 556, "lr": 0.001} +{"train_loss": 0.036022309213876724, "train_loss_bc": 0.036022309213876724, "train_loss_llm": 0.0, "grad_norm": 0.0594315268099308, "global_step": 20047, "epoch": 556, "lr": 0.001} +{"train_loss": 0.027467628940939903, "train_loss_bc": 0.027467628940939903, "train_loss_llm": 0.0, "grad_norm": 0.06506513059139252, "global_step": 20048, "epoch": 556, "lr": 0.001} +{"train_loss": 0.030318541452288628, "train_loss_bc": 0.030307568609714508, "train_loss_llm": 0.0010972917079925537, "grad_norm": 0.022559234872460365, "global_step": 20049, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03129260987043381, "train_loss_bc": 0.031289976090192795, "train_loss_llm": 0.0002633333206176758, "grad_norm": 0.03870532289147377, "global_step": 20050, "epoch": 556, "lr": 0.001} +{"train_loss": 0.03300175986563166, "train_loss_bc": 0.02798549085855484, "train_loss_llm": 0.0, "grad_norm": 0.04966197535395622, "global_step": 20051, "epoch": 556, "lr": 0.001, "val_loss": 0.0787692740559578} +{"train_loss": 0.02822982333600521, "train_loss_bc": 0.02822982333600521, "train_loss_llm": 0.0, "grad_norm": 0.058616749942302704, "global_step": 20052, "epoch": 557, "lr": 0.001} +{"train_loss": 0.030569955706596375, "train_loss_bc": 0.030569955706596375, "train_loss_llm": 0.0, "grad_norm": 0.024675648659467697, "global_step": 20053, "epoch": 557, "lr": 0.001} +{"train_loss": 0.04177679494023323, "train_loss_bc": 0.04177679494023323, "train_loss_llm": 0.0, "grad_norm": 0.04680422693490982, "global_step": 20054, "epoch": 557, "lr": 0.001} +{"train_loss": 0.028370322659611702, "train_loss_bc": 0.02836522087454796, "train_loss_llm": 0.0005101561546325684, "grad_norm": 0.058167796581983566, "global_step": 20055, "epoch": 557, "lr": 0.001} +{"train_loss": 0.031469572335481644, "train_loss_bc": 0.031469572335481644, "train_loss_llm": 0.0, "grad_norm": 0.07448624074459076, "global_step": 20056, "epoch": 557, "lr": 0.001} +{"train_loss": 0.034182701259851456, "train_loss_bc": 0.034182701259851456, "train_loss_llm": 0.0, "grad_norm": 0.023698847740888596, "global_step": 20057, "epoch": 557, "lr": 0.001} +{"train_loss": 0.0312671884894371, "train_loss_bc": 0.03125552088022232, "train_loss_llm": 0.0011668205261230469, "grad_norm": 0.03688699007034302, "global_step": 20058, "epoch": 557, "lr": 0.001} +{"train_loss": 0.033338576555252075, "train_loss_bc": 0.033338576555252075, "train_loss_llm": 0.0, "grad_norm": 0.048575811088085175, "global_step": 20059, "epoch": 557, "lr": 0.001} +{"train_loss": 0.034816402941942215, "train_loss_bc": 0.034816402941942215, "train_loss_llm": 0.0, "grad_norm": 0.05457953363656998, "global_step": 20060, "epoch": 557, "lr": 0.001} +{"train_loss": 0.033431436866521835, "train_loss_bc": 0.03341533988714218, "train_loss_llm": 0.0016097426414489746, "grad_norm": 0.03074645809829235, "global_step": 20061, "epoch": 557, "lr": 0.001} +{"train_loss": 0.030355531722307205, "train_loss_bc": 0.03033255785703659, "train_loss_llm": 0.002297312021255493, "grad_norm": 0.041548170149326324, "global_step": 20062, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03376693278551102, "train_loss_bc": 0.03376420959830284, "train_loss_llm": 0.00027248263359069824, "grad_norm": 0.05315899848937988, "global_step": 20063, "epoch": 557, "lr": 0.001} +{"train_loss": 0.037165865302085876, "train_loss_bc": 0.037165865302085876, "train_loss_llm": 0.0, "grad_norm": 0.06732288002967834, "global_step": 20064, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03562480956315994, "train_loss_bc": 0.03562480956315994, "train_loss_llm": 0.0, "grad_norm": 0.02812737226486206, "global_step": 20065, "epoch": 557, "lr": 0.001} +{"train_loss": 0.030930425971746445, "train_loss_bc": 0.030930425971746445, "train_loss_llm": 0.0, "grad_norm": 0.04665736109018326, "global_step": 20066, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03418496996164322, "train_loss_bc": 0.034181468188762665, "train_loss_llm": 0.0003503561019897461, "grad_norm": 0.05552643910050392, "global_step": 20067, "epoch": 557, "lr": 0.001} +{"train_loss": 0.032050929963588715, "train_loss_bc": 0.032050929963588715, "train_loss_llm": 0.0, "grad_norm": 0.06833842396736145, "global_step": 20068, "epoch": 557, "lr": 0.001} +{"train_loss": 0.033969901502132416, "train_loss_bc": 0.033969901502132416, "train_loss_llm": 0.0, "grad_norm": 0.027317099273204803, "global_step": 20069, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03182424232363701, "train_loss_bc": 0.03182424232363701, "train_loss_llm": 0.0, "grad_norm": 0.034694768488407135, "global_step": 20070, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03239817917346954, "train_loss_bc": 0.032395850867033005, "train_loss_llm": 0.0002326667308807373, "grad_norm": 0.04634467884898186, "global_step": 20071, "epoch": 557, "lr": 0.001} +{"train_loss": 0.035477474331855774, "train_loss_bc": 0.035477474331855774, "train_loss_llm": 0.0, "grad_norm": 0.06073390692472458, "global_step": 20072, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03290568292140961, "train_loss_bc": 0.03289477154612541, "train_loss_llm": 0.00109100341796875, "grad_norm": 0.02911550924181938, "global_step": 20073, "epoch": 557, "lr": 0.001} +{"train_loss": 0.035813044756650925, "train_loss_bc": 0.03580222651362419, "train_loss_llm": 0.0010817348957061768, "grad_norm": 0.03700549900531769, "global_step": 20074, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03158453106880188, "train_loss_bc": 0.03158453106880188, "train_loss_llm": 0.0, "grad_norm": 0.04563591629266739, "global_step": 20075, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03629693016409874, "train_loss_bc": 0.03629693016409874, "train_loss_llm": 0.0, "grad_norm": 0.055478814989328384, "global_step": 20076, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03130539879202843, "train_loss_bc": 0.03130539879202843, "train_loss_llm": 0.0, "grad_norm": 0.027723640203475952, "global_step": 20077, "epoch": 557, "lr": 0.001} +{"train_loss": 0.04044601321220398, "train_loss_bc": 0.04044454172253609, "train_loss_llm": 0.00014725327491760254, "grad_norm": 0.051842305809259415, "global_step": 20078, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03599514812231064, "train_loss_bc": 0.03599514812231064, "train_loss_llm": 0.0, "grad_norm": 0.07255270332098007, "global_step": 20079, "epoch": 557, "lr": 0.001} +{"train_loss": 0.029021279886364937, "train_loss_bc": 0.029021279886364937, "train_loss_llm": 0.0, "grad_norm": 0.07923384010791779, "global_step": 20080, "epoch": 557, "lr": 0.001} +{"train_loss": 0.028955478221178055, "train_loss_bc": 0.02894551493227482, "train_loss_llm": 0.0009962916374206543, "grad_norm": 0.026529472321271896, "global_step": 20081, "epoch": 557, "lr": 0.001} +{"train_loss": 0.031221378594636917, "train_loss_bc": 0.0312206968665123, "train_loss_llm": 6.824731826782227e-05, "grad_norm": 0.04585107043385506, "global_step": 20082, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03321577608585358, "train_loss_bc": 0.03321438655257225, "train_loss_llm": 0.00013905763626098633, "grad_norm": 0.060025859624147415, "global_step": 20083, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03294029086828232, "train_loss_bc": 0.032932717353105545, "train_loss_llm": 0.0007572174072265625, "grad_norm": 0.06648138910531998, "global_step": 20084, "epoch": 557, "lr": 0.001} +{"train_loss": 0.026416370645165443, "train_loss_bc": 0.026409586891531944, "train_loss_llm": 0.000678330659866333, "grad_norm": 0.02354428358376026, "global_step": 20085, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03368911147117615, "train_loss_bc": 0.03368911147117615, "train_loss_llm": 0.0, "grad_norm": 0.03722508251667023, "global_step": 20086, "epoch": 557, "lr": 0.001} +{"train_loss": 0.03303550587346157, "train_loss_bc": 0.03426973894238472, "train_loss_llm": 0.0, "grad_norm": 0.058658625930547714, "global_step": 20087, "epoch": 557, "lr": 0.001, "val_loss": 0.07850909233093262} +{"train_loss": 0.036372266709804535, "train_loss_bc": 0.036372266709804535, "train_loss_llm": 0.0, "grad_norm": 0.0700400173664093, "global_step": 20088, "epoch": 558, "lr": 0.001} +{"train_loss": 0.031395286321640015, "train_loss_bc": 0.031395286321640015, "train_loss_llm": 0.0, "grad_norm": 0.022912422195076942, "global_step": 20089, "epoch": 558, "lr": 0.001} +{"train_loss": 0.032696012407541275, "train_loss_bc": 0.032696012407541275, "train_loss_llm": 0.0, "grad_norm": 0.03386415168642998, "global_step": 20090, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03153294697403908, "train_loss_bc": 0.0315268374979496, "train_loss_llm": 0.0006109476089477539, "grad_norm": 0.037002455443143845, "global_step": 20091, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03512733429670334, "train_loss_bc": 0.03512733429670334, "train_loss_llm": 0.0, "grad_norm": 0.05873145908117294, "global_step": 20092, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03177975118160248, "train_loss_bc": 0.03177975118160248, "train_loss_llm": 0.0, "grad_norm": 0.026707591488957405, "global_step": 20093, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03426305204629898, "train_loss_bc": 0.03426305204629898, "train_loss_llm": 0.0, "grad_norm": 0.04132860153913498, "global_step": 20094, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03270304948091507, "train_loss_bc": 0.03270304948091507, "train_loss_llm": 0.0, "grad_norm": 0.05939464643597603, "global_step": 20095, "epoch": 558, "lr": 0.001} +{"train_loss": 0.039669398218393326, "train_loss_bc": 0.03965963050723076, "train_loss_llm": 0.0009767413139343262, "grad_norm": 0.07086234539747238, "global_step": 20096, "epoch": 558, "lr": 0.001} +{"train_loss": 0.037861667573451996, "train_loss_bc": 0.037854768335819244, "train_loss_llm": 0.0006899833679199219, "grad_norm": 0.03661017492413521, "global_step": 20097, "epoch": 558, "lr": 0.001} +{"train_loss": 0.02941730245947838, "train_loss_bc": 0.02941730245947838, "train_loss_llm": 0.0, "grad_norm": 0.037770334631204605, "global_step": 20098, "epoch": 558, "lr": 0.001} +{"train_loss": 0.028718505054712296, "train_loss_bc": 0.028718505054712296, "train_loss_llm": 0.0, "grad_norm": 0.04848750680685043, "global_step": 20099, "epoch": 558, "lr": 0.001} +{"train_loss": 0.0313677154481411, "train_loss_bc": 0.0313677154481411, "train_loss_llm": 0.0, "grad_norm": 0.06099071353673935, "global_step": 20100, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03405090421438217, "train_loss_bc": 0.03404759243130684, "train_loss_llm": 0.0003310739994049072, "grad_norm": 0.02576945722103119, "global_step": 20101, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03123548813164234, "train_loss_bc": 0.03123548813164234, "train_loss_llm": 0.0, "grad_norm": 0.045335784554481506, "global_step": 20102, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03502235561609268, "train_loss_bc": 0.03502235561609268, "train_loss_llm": 0.0, "grad_norm": 0.04805648326873779, "global_step": 20103, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03029976412653923, "train_loss_bc": 0.03029976412653923, "train_loss_llm": 0.0, "grad_norm": 0.05580184981226921, "global_step": 20104, "epoch": 558, "lr": 0.001} +{"train_loss": 0.032652124762535095, "train_loss_bc": 0.03264157846570015, "train_loss_llm": 0.0010546743869781494, "grad_norm": 0.03074244223535061, "global_step": 20105, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03068666160106659, "train_loss_bc": 0.03068666160106659, "train_loss_llm": 0.0, "grad_norm": 0.04701140895485878, "global_step": 20106, "epoch": 558, "lr": 0.001} +{"train_loss": 0.02904263138771057, "train_loss_bc": 0.02904263138771057, "train_loss_llm": 0.0, "grad_norm": 0.052779246121644974, "global_step": 20107, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03382055088877678, "train_loss_bc": 0.03382055088877678, "train_loss_llm": 0.0, "grad_norm": 0.0666949450969696, "global_step": 20108, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03199564665555954, "train_loss_bc": 0.03199564665555954, "train_loss_llm": 0.0, "grad_norm": 0.028919367119669914, "global_step": 20109, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03514615818858147, "train_loss_bc": 0.03514615818858147, "train_loss_llm": 0.0, "grad_norm": 0.04806468263268471, "global_step": 20110, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03211408108472824, "train_loss_bc": 0.0320916473865509, "train_loss_llm": 0.002243250608444214, "grad_norm": 0.05555340275168419, "global_step": 20111, "epoch": 558, "lr": 0.001} +{"train_loss": 0.032086338847875595, "train_loss_bc": 0.032086338847875595, "train_loss_llm": 0.0, "grad_norm": 0.06971535831689835, "global_step": 20112, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03479122370481491, "train_loss_bc": 0.03477604687213898, "train_loss_llm": 0.0015177726745605469, "grad_norm": 0.028013013303279877, "global_step": 20113, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03334566205739975, "train_loss_bc": 0.03334348648786545, "train_loss_llm": 0.00021761655807495117, "grad_norm": 0.04428242892026901, "global_step": 20114, "epoch": 558, "lr": 0.001} +{"train_loss": 0.030768493190407753, "train_loss_bc": 0.030768493190407753, "train_loss_llm": 0.0, "grad_norm": 0.06182769313454628, "global_step": 20115, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03828834742307663, "train_loss_bc": 0.03828834742307663, "train_loss_llm": 0.0, "grad_norm": 0.07124623656272888, "global_step": 20116, "epoch": 558, "lr": 0.001} +{"train_loss": 0.02870018221437931, "train_loss_bc": 0.02870018221437931, "train_loss_llm": 0.0, "grad_norm": 0.030925637111067772, "global_step": 20117, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03525908291339874, "train_loss_bc": 0.03525908291339874, "train_loss_llm": 0.0, "grad_norm": 0.04356147348880768, "global_step": 20118, "epoch": 558, "lr": 0.001} +{"train_loss": 0.037986163049936295, "train_loss_bc": 0.03796859830617905, "train_loss_llm": 0.0017563402652740479, "grad_norm": 0.06727667152881622, "global_step": 20119, "epoch": 558, "lr": 0.001} +{"train_loss": 0.034686971455812454, "train_loss_bc": 0.034686971455812454, "train_loss_llm": 0.0, "grad_norm": 0.08101947605609894, "global_step": 20120, "epoch": 558, "lr": 0.001} +{"train_loss": 0.02822064980864525, "train_loss_bc": 0.028217878192663193, "train_loss_llm": 0.0002772510051727295, "grad_norm": 0.02902192436158657, "global_step": 20121, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03201659768819809, "train_loss_bc": 0.032015781849622726, "train_loss_llm": 8.168816566467285e-05, "grad_norm": 0.04270013049244881, "global_step": 20122, "epoch": 558, "lr": 0.001} +{"train_loss": 0.03308254682148496, "train_loss_bc": 0.03585131838917732, "train_loss_llm": 0.0, "grad_norm": 0.06423458456993103, "global_step": 20123, "epoch": 558, "lr": 0.001, "val_loss": 0.07846462726593018} +{"train_loss": 0.03014502301812172, "train_loss_bc": 0.03014502301812172, "train_loss_llm": 0.0, "grad_norm": 0.07520877569913864, "global_step": 20124, "epoch": 559, "lr": 0.001} +{"train_loss": 0.0373084656894207, "train_loss_bc": 0.03730126470327377, "train_loss_llm": 0.0007199347019195557, "grad_norm": 0.02970612235367298, "global_step": 20125, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03125507012009621, "train_loss_bc": 0.03125384822487831, "train_loss_llm": 0.00012215971946716309, "grad_norm": 0.04249417036771774, "global_step": 20126, "epoch": 559, "lr": 0.001} +{"train_loss": 0.033138830214738846, "train_loss_bc": 0.033138830214738846, "train_loss_llm": 0.0, "grad_norm": 0.05591624975204468, "global_step": 20127, "epoch": 559, "lr": 0.001} +{"train_loss": 0.033064451068639755, "train_loss_bc": 0.033064451068639755, "train_loss_llm": 0.0, "grad_norm": 0.06779898703098297, "global_step": 20128, "epoch": 559, "lr": 0.001} +{"train_loss": 0.030990440398454666, "train_loss_bc": 0.030990440398454666, "train_loss_llm": 0.0, "grad_norm": 0.024291031062602997, "global_step": 20129, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03874791041016579, "train_loss_bc": 0.03874791041016579, "train_loss_llm": 0.0, "grad_norm": 0.04780126363039017, "global_step": 20130, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03434675559401512, "train_loss_bc": 0.03434675559401512, "train_loss_llm": 0.0, "grad_norm": 0.05684434249997139, "global_step": 20131, "epoch": 559, "lr": 0.001} +{"train_loss": 0.024802109226584435, "train_loss_bc": 0.024802109226584435, "train_loss_llm": 0.0, "grad_norm": 0.07360067963600159, "global_step": 20132, "epoch": 559, "lr": 0.001} +{"train_loss": 0.031023062765598297, "train_loss_bc": 0.031023062765598297, "train_loss_llm": 0.0, "grad_norm": 0.028547294437885284, "global_step": 20133, "epoch": 559, "lr": 0.001} +{"train_loss": 0.031843509525060654, "train_loss_bc": 0.03182562440633774, "train_loss_llm": 0.0017885565757751465, "grad_norm": 0.03850004822015762, "global_step": 20134, "epoch": 559, "lr": 0.001} +{"train_loss": 0.030865835025906563, "train_loss_bc": 0.030864687636494637, "train_loss_llm": 0.00011476874351501465, "grad_norm": 0.05989883095026016, "global_step": 20135, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03534414991736412, "train_loss_bc": 0.03534414991736412, "train_loss_llm": 0.0, "grad_norm": 0.07485903054475784, "global_step": 20136, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03849506005644798, "train_loss_bc": 0.03849506005644798, "train_loss_llm": 0.0, "grad_norm": 0.02874579280614853, "global_step": 20137, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03946619853377342, "train_loss_bc": 0.03945808485150337, "train_loss_llm": 0.0008114278316497803, "grad_norm": 0.057932041585445404, "global_step": 20138, "epoch": 559, "lr": 0.001} +{"train_loss": 0.0321236215531826, "train_loss_bc": 0.03211529180407524, "train_loss_llm": 0.0008330643177032471, "grad_norm": 0.07522106170654297, "global_step": 20139, "epoch": 559, "lr": 0.001} +{"train_loss": 0.030330751091241837, "train_loss_bc": 0.030328940600156784, "train_loss_llm": 0.00018110871315002441, "grad_norm": 0.0908554196357727, "global_step": 20140, "epoch": 559, "lr": 0.001} +{"train_loss": 0.032654426991939545, "train_loss_bc": 0.032643985003232956, "train_loss_llm": 0.001044243574142456, "grad_norm": 0.027180414646863937, "global_step": 20141, "epoch": 559, "lr": 0.001} +{"train_loss": 0.0314340740442276, "train_loss_bc": 0.031419407576322556, "train_loss_llm": 0.0014665722846984863, "grad_norm": 0.03530530259013176, "global_step": 20142, "epoch": 559, "lr": 0.001} +{"train_loss": 0.0405985526740551, "train_loss_bc": 0.040579717606306076, "train_loss_llm": 0.0018835663795471191, "grad_norm": 0.05861425772309303, "global_step": 20143, "epoch": 559, "lr": 0.001} +{"train_loss": 0.030148549005389214, "train_loss_bc": 0.030148549005389214, "train_loss_llm": 0.0, "grad_norm": 0.07382441312074661, "global_step": 20144, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03680445998907089, "train_loss_bc": 0.03680445998907089, "train_loss_llm": 0.0, "grad_norm": 0.028110502287745476, "global_step": 20145, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03271622955799103, "train_loss_bc": 0.03270949795842171, "train_loss_llm": 0.0006732940673828125, "grad_norm": 0.04355926066637039, "global_step": 20146, "epoch": 559, "lr": 0.001} +{"train_loss": 0.026962583884596825, "train_loss_bc": 0.026962583884596825, "train_loss_llm": 0.0, "grad_norm": 0.05466832220554352, "global_step": 20147, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03576069325208664, "train_loss_bc": 0.03576069325208664, "train_loss_llm": 0.0, "grad_norm": 0.07350042462348938, "global_step": 20148, "epoch": 559, "lr": 0.001} +{"train_loss": 0.02767827920615673, "train_loss_bc": 0.02767827920615673, "train_loss_llm": 0.0, "grad_norm": 0.02541721798479557, "global_step": 20149, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03191869705915451, "train_loss_bc": 0.03191869705915451, "train_loss_llm": 0.0, "grad_norm": 0.04784412309527397, "global_step": 20150, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03582735359668732, "train_loss_bc": 0.03582735359668732, "train_loss_llm": 0.0, "grad_norm": 0.06356089562177658, "global_step": 20151, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03189116343855858, "train_loss_bc": 0.03188996762037277, "train_loss_llm": 0.0001195371150970459, "grad_norm": 0.0769241452217102, "global_step": 20152, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03245515748858452, "train_loss_bc": 0.03245515748858452, "train_loss_llm": 0.0, "grad_norm": 0.03124576434493065, "global_step": 20153, "epoch": 559, "lr": 0.001} +{"train_loss": 0.027812756597995758, "train_loss_bc": 0.027812756597995758, "train_loss_llm": 0.0, "grad_norm": 0.048992812633514404, "global_step": 20154, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03144054114818573, "train_loss_bc": 0.03144054114818573, "train_loss_llm": 0.0, "grad_norm": 0.07509037852287292, "global_step": 20155, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03521084040403366, "train_loss_bc": 0.0351966992020607, "train_loss_llm": 0.001413971185684204, "grad_norm": 0.09289680421352386, "global_step": 20156, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03979996219277382, "train_loss_bc": 0.03979996219277382, "train_loss_llm": 0.0, "grad_norm": 0.0298061054199934, "global_step": 20157, "epoch": 559, "lr": 0.001} +{"train_loss": 0.04062943905591965, "train_loss_bc": 0.04062943905591965, "train_loss_llm": 0.0, "grad_norm": 0.06325682252645493, "global_step": 20158, "epoch": 559, "lr": 0.001} +{"train_loss": 0.03324922582962447, "train_loss_bc": 0.031937126070261, "train_loss_llm": 0.0, "grad_norm": 0.081039659678936, "global_step": 20159, "epoch": 559, "lr": 0.001, "val_loss": 0.08196721971035004} +{"train_loss": 0.02589152380824089, "train_loss_bc": 0.02589152380824089, "train_loss_llm": 0.0, "grad_norm": 0.09587231278419495, "global_step": 20160, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03268219158053398, "train_loss_bc": 0.03268219158053398, "train_loss_llm": 0.0, "grad_norm": 0.02774696797132492, "global_step": 20161, "epoch": 560, "lr": 0.001} +{"train_loss": 0.024053804576396942, "train_loss_bc": 0.024053804576396942, "train_loss_llm": 0.0, "grad_norm": 0.03870639204978943, "global_step": 20162, "epoch": 560, "lr": 0.001} +{"train_loss": 0.033388786017894745, "train_loss_bc": 0.0333879180252552, "train_loss_llm": 8.672475814819336e-05, "grad_norm": 0.058583371341228485, "global_step": 20163, "epoch": 560, "lr": 0.001} +{"train_loss": 0.030918627977371216, "train_loss_bc": 0.030918627977371216, "train_loss_llm": 0.0, "grad_norm": 0.07084643840789795, "global_step": 20164, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03574143350124359, "train_loss_bc": 0.03574143350124359, "train_loss_llm": 0.0, "grad_norm": 0.03105541132390499, "global_step": 20165, "epoch": 560, "lr": 0.001} +{"train_loss": 0.0313505157828331, "train_loss_bc": 0.0313505157828331, "train_loss_llm": 0.0, "grad_norm": 0.039760712534189224, "global_step": 20166, "epoch": 560, "lr": 0.001} +{"train_loss": 0.037464458495378494, "train_loss_bc": 0.037447910755872726, "train_loss_llm": 0.001654893159866333, "grad_norm": 0.058513324707746506, "global_step": 20167, "epoch": 560, "lr": 0.001} +{"train_loss": 0.036521293222904205, "train_loss_bc": 0.036521293222904205, "train_loss_llm": 0.0, "grad_norm": 0.07126953452825546, "global_step": 20168, "epoch": 560, "lr": 0.001} +{"train_loss": 0.030381249263882637, "train_loss_bc": 0.030381249263882637, "train_loss_llm": 0.0, "grad_norm": 0.029811840504407883, "global_step": 20169, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03138737753033638, "train_loss_bc": 0.03138737753033638, "train_loss_llm": 0.0, "grad_norm": 0.04372406378388405, "global_step": 20170, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03166702389717102, "train_loss_bc": 0.03166702389717102, "train_loss_llm": 0.0, "grad_norm": 0.059126053005456924, "global_step": 20171, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03623794764280319, "train_loss_bc": 0.03623794764280319, "train_loss_llm": 0.0, "grad_norm": 0.06928893178701401, "global_step": 20172, "epoch": 560, "lr": 0.001} +{"train_loss": 0.037242524325847626, "train_loss_bc": 0.03723497688770294, "train_loss_llm": 0.0007548332214355469, "grad_norm": 0.03523174300789833, "global_step": 20173, "epoch": 560, "lr": 0.001} +{"train_loss": 0.029869377613067627, "train_loss_bc": 0.029869377613067627, "train_loss_llm": 0.0, "grad_norm": 0.053779952228069305, "global_step": 20174, "epoch": 560, "lr": 0.001} +{"train_loss": 0.035855043679475784, "train_loss_bc": 0.03585333377122879, "train_loss_llm": 0.00017082691192626953, "grad_norm": 0.07576601952314377, "global_step": 20175, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03829951584339142, "train_loss_bc": 0.03829951584339142, "train_loss_llm": 0.0, "grad_norm": 0.09622060507535934, "global_step": 20176, "epoch": 560, "lr": 0.001} +{"train_loss": 0.038659416139125824, "train_loss_bc": 0.038659416139125824, "train_loss_llm": 0.0, "grad_norm": 0.03704289719462395, "global_step": 20177, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03848559781908989, "train_loss_bc": 0.03848559781908989, "train_loss_llm": 0.0, "grad_norm": 0.06598236411809921, "global_step": 20178, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03208765760064125, "train_loss_bc": 0.03208765760064125, "train_loss_llm": 0.0, "grad_norm": 0.08112037926912308, "global_step": 20179, "epoch": 560, "lr": 0.001} +{"train_loss": 0.038691747933626175, "train_loss_bc": 0.038691747933626175, "train_loss_llm": 0.0, "grad_norm": 0.0914074257016182, "global_step": 20180, "epoch": 560, "lr": 0.001} +{"train_loss": 0.0347047820687294, "train_loss_bc": 0.0347047820687294, "train_loss_llm": 0.0, "grad_norm": 0.026756582781672478, "global_step": 20181, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03182690218091011, "train_loss_bc": 0.03180764243006706, "train_loss_llm": 0.001926124095916748, "grad_norm": 0.04730278626084328, "global_step": 20182, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03728442266583443, "train_loss_bc": 0.03728442266583443, "train_loss_llm": 0.0, "grad_norm": 0.0631696805357933, "global_step": 20183, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03405585512518883, "train_loss_bc": 0.03404328227043152, "train_loss_llm": 0.0012573003768920898, "grad_norm": 0.08772221952676773, "global_step": 20184, "epoch": 560, "lr": 0.001} +{"train_loss": 0.030875707045197487, "train_loss_bc": 0.030875707045197487, "train_loss_llm": 0.0, "grad_norm": 0.02020443044602871, "global_step": 20185, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03241089731454849, "train_loss_bc": 0.03241089731454849, "train_loss_llm": 0.0, "grad_norm": 0.028756607323884964, "global_step": 20186, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03711758926510811, "train_loss_bc": 0.03711758926510811, "train_loss_llm": 0.0, "grad_norm": 0.040075868368148804, "global_step": 20187, "epoch": 560, "lr": 0.001} +{"train_loss": 0.029191549867391586, "train_loss_bc": 0.02917974628508091, "train_loss_llm": 0.0011803805828094482, "grad_norm": 0.05106872692704201, "global_step": 20188, "epoch": 560, "lr": 0.001} +{"train_loss": 0.02810039184987545, "train_loss_bc": 0.02810039184987545, "train_loss_llm": 0.0, "grad_norm": 0.024837175384163857, "global_step": 20189, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03657033294439316, "train_loss_bc": 0.03657033294439316, "train_loss_llm": 0.0, "grad_norm": 0.0449177511036396, "global_step": 20190, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03344538435339928, "train_loss_bc": 0.03344538435339928, "train_loss_llm": 0.0, "grad_norm": 0.06055624410510063, "global_step": 20191, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03426958620548248, "train_loss_bc": 0.03426958620548248, "train_loss_llm": 0.0, "grad_norm": 0.0761369839310646, "global_step": 20192, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03309140354394913, "train_loss_bc": 0.03309140354394913, "train_loss_llm": 0.0, "grad_norm": 0.03201024606823921, "global_step": 20193, "epoch": 560, "lr": 0.001} +{"train_loss": 0.024571595713496208, "train_loss_bc": 0.024571595713496208, "train_loss_llm": 0.0, "grad_norm": 0.051060959696769714, "global_step": 20194, "epoch": 560, "lr": 0.001} +{"train_loss": 0.03316987119615078, "train_loss_bc": 0.029721848666667938, "train_loss_llm": 0.0, "grad_norm": 0.06576798111200333, "global_step": 20195, "epoch": 560, "lr": 0.001, "train/cumulative_reward": 46.213044162895116, "train/mean_score": 19.739691273381467, "train/success_rate": 0.6, "test/cumulative_reward": 35.147264946365254, "test/mean_score": 5.756854052673386, "test/success_rate": 0.14, "val_loss": 0.0786566510796547, "train_action_mse_error": 0.03629150614142418} +{"train_loss": 0.029152823612093925, "train_loss_bc": 0.029152823612093925, "train_loss_llm": 0.0, "grad_norm": 0.08280202746391296, "global_step": 20196, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03712170198559761, "train_loss_bc": 0.03712170198559761, "train_loss_llm": 0.0, "grad_norm": 0.03279801830649376, "global_step": 20197, "epoch": 561, "lr": 0.001} +{"train_loss": 0.029932217672467232, "train_loss_bc": 0.029932217672467232, "train_loss_llm": 0.0, "grad_norm": 0.04409192129969597, "global_step": 20198, "epoch": 561, "lr": 0.001} +{"train_loss": 0.031267162412405014, "train_loss_bc": 0.031267162412405014, "train_loss_llm": 0.0, "grad_norm": 0.060028936713933945, "global_step": 20199, "epoch": 561, "lr": 0.001} +{"train_loss": 0.02998228184878826, "train_loss_bc": 0.02998228184878826, "train_loss_llm": 0.0, "grad_norm": 0.06925737857818604, "global_step": 20200, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03550034388899803, "train_loss_bc": 0.03550034388899803, "train_loss_llm": 0.0, "grad_norm": 0.02890416607260704, "global_step": 20201, "epoch": 561, "lr": 0.001} +{"train_loss": 0.0353737398982048, "train_loss_bc": 0.0353737398982048, "train_loss_llm": 0.0, "grad_norm": 0.043491777032613754, "global_step": 20202, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03427768871188164, "train_loss_bc": 0.034264951944351196, "train_loss_llm": 0.0012737810611724854, "grad_norm": 0.057099636644124985, "global_step": 20203, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03263307735323906, "train_loss_bc": 0.03263307735323906, "train_loss_llm": 0.0, "grad_norm": 0.06861160695552826, "global_step": 20204, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03192067891359329, "train_loss_bc": 0.03192067891359329, "train_loss_llm": 0.0, "grad_norm": 0.027883421629667282, "global_step": 20205, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03281945362687111, "train_loss_bc": 0.03280750662088394, "train_loss_llm": 0.0011948645114898682, "grad_norm": 0.035350214689970016, "global_step": 20206, "epoch": 561, "lr": 0.001} +{"train_loss": 0.029341230168938637, "train_loss_bc": 0.029341230168938637, "train_loss_llm": 0.0, "grad_norm": 0.043548375368118286, "global_step": 20207, "epoch": 561, "lr": 0.001} +{"train_loss": 0.02784026227891445, "train_loss_bc": 0.02784026227891445, "train_loss_llm": 0.0, "grad_norm": 0.05066375806927681, "global_step": 20208, "epoch": 561, "lr": 0.001} +{"train_loss": 0.035189270973205566, "train_loss_bc": 0.035189270973205566, "train_loss_llm": 0.0, "grad_norm": 0.02942272648215294, "global_step": 20209, "epoch": 561, "lr": 0.001} +{"train_loss": 0.037642333656549454, "train_loss_bc": 0.0376402921974659, "train_loss_llm": 0.0002041459083557129, "grad_norm": 0.043525390326976776, "global_step": 20210, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03363471105694771, "train_loss_bc": 0.03363471105694771, "train_loss_llm": 0.0, "grad_norm": 0.04516809806227684, "global_step": 20211, "epoch": 561, "lr": 0.001} +{"train_loss": 0.035649530589580536, "train_loss_bc": 0.035649530589580536, "train_loss_llm": 0.0, "grad_norm": 0.05796051025390625, "global_step": 20212, "epoch": 561, "lr": 0.001} +{"train_loss": 0.0337129607796669, "train_loss_bc": 0.0337129607796669, "train_loss_llm": 0.0, "grad_norm": 0.02594088390469551, "global_step": 20213, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03434319794178009, "train_loss_bc": 0.03434319794178009, "train_loss_llm": 0.0, "grad_norm": 0.04180503636598587, "global_step": 20214, "epoch": 561, "lr": 0.001} +{"train_loss": 0.036109644919633865, "train_loss_bc": 0.036109644919633865, "train_loss_llm": 0.0, "grad_norm": 0.053311076015233994, "global_step": 20215, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03340035304427147, "train_loss_bc": 0.03340035304427147, "train_loss_llm": 0.0, "grad_norm": 0.06893882900476456, "global_step": 20216, "epoch": 561, "lr": 0.001} +{"train_loss": 0.029400603845715523, "train_loss_bc": 0.029400603845715523, "train_loss_llm": 0.0, "grad_norm": 0.025268342345952988, "global_step": 20217, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03613895922899246, "train_loss_bc": 0.03613895922899246, "train_loss_llm": 0.0, "grad_norm": 0.03806032985448837, "global_step": 20218, "epoch": 561, "lr": 0.001} +{"train_loss": 0.028257157653570175, "train_loss_bc": 0.02824786677956581, "train_loss_llm": 0.0009291172027587891, "grad_norm": 0.05283872410655022, "global_step": 20219, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03102181665599346, "train_loss_bc": 0.03102181665599346, "train_loss_llm": 0.0, "grad_norm": 0.06297943741083145, "global_step": 20220, "epoch": 561, "lr": 0.001} +{"train_loss": 0.034180380403995514, "train_loss_bc": 0.03417694568634033, "train_loss_llm": 0.0003434717655181885, "grad_norm": 0.03234769403934479, "global_step": 20221, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03512852266430855, "train_loss_bc": 0.035122644156217575, "train_loss_llm": 0.0005879700183868408, "grad_norm": 0.05937262251973152, "global_step": 20222, "epoch": 561, "lr": 0.001} +{"train_loss": 0.029096439480781555, "train_loss_bc": 0.029096439480781555, "train_loss_llm": 0.0, "grad_norm": 0.0697692334651947, "global_step": 20223, "epoch": 561, "lr": 0.001} +{"train_loss": 0.032961975783109665, "train_loss_bc": 0.032961975783109665, "train_loss_llm": 0.0, "grad_norm": 0.07897495478391647, "global_step": 20224, "epoch": 561, "lr": 0.001} +{"train_loss": 0.034947775304317474, "train_loss_bc": 0.034947775304317474, "train_loss_llm": 0.0, "grad_norm": 0.03043457865715027, "global_step": 20225, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03797135129570961, "train_loss_bc": 0.03797135129570961, "train_loss_llm": 0.0, "grad_norm": 0.052135150879621506, "global_step": 20226, "epoch": 561, "lr": 0.001} +{"train_loss": 0.040634721517562866, "train_loss_bc": 0.040634721517562866, "train_loss_llm": 0.0, "grad_norm": 0.061547547578811646, "global_step": 20227, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03027142398059368, "train_loss_bc": 0.03024403192102909, "train_loss_llm": 0.0027392804622650146, "grad_norm": 0.06048572435975075, "global_step": 20228, "epoch": 561, "lr": 0.001} +{"train_loss": 0.031807780265808105, "train_loss_bc": 0.0317741297185421, "train_loss_llm": 0.0033652186393737793, "grad_norm": 0.02972867153584957, "global_step": 20229, "epoch": 561, "lr": 0.001} +{"train_loss": 0.026244690641760826, "train_loss_bc": 0.026223069056868553, "train_loss_llm": 0.0021620988845825195, "grad_norm": 0.03709590062499046, "global_step": 20230, "epoch": 561, "lr": 0.001} +{"train_loss": 0.03297567031242781, "train_loss_bc": 0.032215867191553116, "train_loss_llm": 0.0, "grad_norm": 0.053450632840394974, "global_step": 20231, "epoch": 561, "lr": 0.001, "val_loss": 0.07907037436962128} +{"train_loss": 0.0230886097997427, "train_loss_bc": 0.023086359724402428, "train_loss_llm": 0.0002250075340270996, "grad_norm": 0.06010173633694649, "global_step": 20232, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03624033182859421, "train_loss_bc": 0.03623875975608826, "train_loss_llm": 0.0001571178436279297, "grad_norm": 0.029597334563732147, "global_step": 20233, "epoch": 562, "lr": 0.001} +{"train_loss": 0.032363567501306534, "train_loss_bc": 0.032363567501306534, "train_loss_llm": 0.0, "grad_norm": 0.041506510227918625, "global_step": 20234, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03626984730362892, "train_loss_bc": 0.03625606745481491, "train_loss_llm": 0.0013780593872070312, "grad_norm": 0.05094997584819794, "global_step": 20235, "epoch": 562, "lr": 0.001} +{"train_loss": 0.027864472940564156, "train_loss_bc": 0.027864472940564156, "train_loss_llm": 0.0, "grad_norm": 0.0606229342520237, "global_step": 20236, "epoch": 562, "lr": 0.001} +{"train_loss": 0.031249618157744408, "train_loss_bc": 0.031249618157744408, "train_loss_llm": 0.0, "grad_norm": 0.02637339010834694, "global_step": 20237, "epoch": 562, "lr": 0.001} +{"train_loss": 0.027688849717378616, "train_loss_bc": 0.027677301317453384, "train_loss_llm": 0.0011548399925231934, "grad_norm": 0.03821806237101555, "global_step": 20238, "epoch": 562, "lr": 0.001} +{"train_loss": 0.0333322137594223, "train_loss_bc": 0.0333322137594223, "train_loss_llm": 0.0, "grad_norm": 0.04811631515622139, "global_step": 20239, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03425908461213112, "train_loss_bc": 0.03423202037811279, "train_loss_llm": 0.0027062594890594482, "grad_norm": 0.05944642052054405, "global_step": 20240, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03715827688574791, "train_loss_bc": 0.03715827688574791, "train_loss_llm": 0.0, "grad_norm": 0.03726062923669815, "global_step": 20241, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03600175678730011, "train_loss_bc": 0.03600175678730011, "train_loss_llm": 0.0, "grad_norm": 0.06379196792840958, "global_step": 20242, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03351402282714844, "train_loss_bc": 0.03351402282714844, "train_loss_llm": 0.0, "grad_norm": 0.08373977988958359, "global_step": 20243, "epoch": 562, "lr": 0.001} +{"train_loss": 0.025962496176362038, "train_loss_bc": 0.025946615263819695, "train_loss_llm": 0.0015881359577178955, "grad_norm": 0.09228038042783737, "global_step": 20244, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03569503501057625, "train_loss_bc": 0.03569391369819641, "train_loss_llm": 0.00011208653450012207, "grad_norm": 0.02509026974439621, "global_step": 20245, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03298872709274292, "train_loss_bc": 0.03298872709274292, "train_loss_llm": 0.0, "grad_norm": 0.04225130006670952, "global_step": 20246, "epoch": 562, "lr": 0.001} +{"train_loss": 0.034053437411785126, "train_loss_bc": 0.034053437411785126, "train_loss_llm": 0.0, "grad_norm": 0.06085122749209404, "global_step": 20247, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03376510739326477, "train_loss_bc": 0.03375290334224701, "train_loss_llm": 0.0012202858924865723, "grad_norm": 0.08205688744783401, "global_step": 20248, "epoch": 562, "lr": 0.001} +{"train_loss": 0.026682719588279724, "train_loss_bc": 0.026643753051757812, "train_loss_llm": 0.0038965940475463867, "grad_norm": 0.022118864580988884, "global_step": 20249, "epoch": 562, "lr": 0.001} +{"train_loss": 0.031154990196228027, "train_loss_bc": 0.031154990196228027, "train_loss_llm": 0.0, "grad_norm": 0.04091385751962662, "global_step": 20250, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03787663206458092, "train_loss_bc": 0.03787663206458092, "train_loss_llm": 0.0, "grad_norm": 0.05715407803654671, "global_step": 20251, "epoch": 562, "lr": 0.001} +{"train_loss": 0.033440206199884415, "train_loss_bc": 0.03343740478157997, "train_loss_llm": 0.00028011202812194824, "grad_norm": 0.0675615519285202, "global_step": 20252, "epoch": 562, "lr": 0.001} +{"train_loss": 0.033877287060022354, "train_loss_bc": 0.033877287060022354, "train_loss_llm": 0.0, "grad_norm": 0.025005940347909927, "global_step": 20253, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03155026212334633, "train_loss_bc": 0.03155026212334633, "train_loss_llm": 0.0, "grad_norm": 0.0403982475399971, "global_step": 20254, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03634778410196304, "train_loss_bc": 0.03634778410196304, "train_loss_llm": 0.0, "grad_norm": 0.06698554754257202, "global_step": 20255, "epoch": 562, "lr": 0.001} +{"train_loss": 0.035649802535772324, "train_loss_bc": 0.035649802535772324, "train_loss_llm": 0.0, "grad_norm": 0.07827498018741608, "global_step": 20256, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03691734001040459, "train_loss_bc": 0.03691734001040459, "train_loss_llm": 0.0, "grad_norm": 0.03079833649098873, "global_step": 20257, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03462787717580795, "train_loss_bc": 0.03462787717580795, "train_loss_llm": 0.0, "grad_norm": 0.049949370324611664, "global_step": 20258, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03521343693137169, "train_loss_bc": 0.03521343693137169, "train_loss_llm": 0.0, "grad_norm": 0.07263045012950897, "global_step": 20259, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03467395901679993, "train_loss_bc": 0.03467395901679993, "train_loss_llm": 0.0, "grad_norm": 0.09294743835926056, "global_step": 20260, "epoch": 562, "lr": 0.001} +{"train_loss": 0.030737992376089096, "train_loss_bc": 0.030737992376089096, "train_loss_llm": 0.0, "grad_norm": 0.024060769006609917, "global_step": 20261, "epoch": 562, "lr": 0.001} +{"train_loss": 0.032416753470897675, "train_loss_bc": 0.032416753470897675, "train_loss_llm": 0.0, "grad_norm": 0.03990127146244049, "global_step": 20262, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03023073822259903, "train_loss_bc": 0.03023073822259903, "train_loss_llm": 0.0, "grad_norm": 0.05210205912590027, "global_step": 20263, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03890625387430191, "train_loss_bc": 0.03890625387430191, "train_loss_llm": 0.0, "grad_norm": 0.0705055519938469, "global_step": 20264, "epoch": 562, "lr": 0.001} +{"train_loss": 0.031446050852537155, "train_loss_bc": 0.031446050852537155, "train_loss_llm": 0.0, "grad_norm": 0.02795722335577011, "global_step": 20265, "epoch": 562, "lr": 0.001} +{"train_loss": 0.03676453232765198, "train_loss_bc": 0.03676453232765198, "train_loss_llm": 0.0, "grad_norm": 0.0474419891834259, "global_step": 20266, "epoch": 562, "lr": 0.001} +{"train_loss": 0.033145670468608536, "train_loss_bc": 0.033234063535928726, "train_loss_llm": 0.0, "grad_norm": 0.0599031038582325, "global_step": 20267, "epoch": 562, "lr": 0.001, "val_loss": 0.0800834372639656} +{"train_loss": 0.03768960013985634, "train_loss_bc": 0.03768773749470711, "train_loss_llm": 0.00018611550331115723, "grad_norm": 0.06974197179079056, "global_step": 20268, "epoch": 563, "lr": 0.001} +{"train_loss": 0.032626986503601074, "train_loss_bc": 0.032626986503601074, "train_loss_llm": 0.0, "grad_norm": 0.029362661764025688, "global_step": 20269, "epoch": 563, "lr": 0.001} +{"train_loss": 0.031358905136585236, "train_loss_bc": 0.031358905136585236, "train_loss_llm": 0.0, "grad_norm": 0.048559945076704025, "global_step": 20270, "epoch": 563, "lr": 0.001} +{"train_loss": 0.030162464827299118, "train_loss_bc": 0.03016124665737152, "train_loss_llm": 0.00012183189392089844, "grad_norm": 0.07057678699493408, "global_step": 20271, "epoch": 563, "lr": 0.001} +{"train_loss": 0.029947729781270027, "train_loss_bc": 0.029947729781270027, "train_loss_llm": 0.0, "grad_norm": 0.08706238120794296, "global_step": 20272, "epoch": 563, "lr": 0.001} +{"train_loss": 0.034237515181303024, "train_loss_bc": 0.034237515181303024, "train_loss_llm": 0.0, "grad_norm": 0.029432309791445732, "global_step": 20273, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03403649479150772, "train_loss_bc": 0.03402860090136528, "train_loss_llm": 0.0007895231246948242, "grad_norm": 0.04538809508085251, "global_step": 20274, "epoch": 563, "lr": 0.001} +{"train_loss": 0.030976979061961174, "train_loss_bc": 0.030950376763939857, "train_loss_llm": 0.002660304307937622, "grad_norm": 0.059302110224962234, "global_step": 20275, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03348153457045555, "train_loss_bc": 0.03346695378422737, "train_loss_llm": 0.0014579296112060547, "grad_norm": 0.06661217659711838, "global_step": 20276, "epoch": 563, "lr": 0.001} +{"train_loss": 0.027432631701231003, "train_loss_bc": 0.027432631701231003, "train_loss_llm": 0.0, "grad_norm": 0.03205975517630577, "global_step": 20277, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03384430706501007, "train_loss_bc": 0.03384147584438324, "train_loss_llm": 0.0002830624580383301, "grad_norm": 0.05316096544265747, "global_step": 20278, "epoch": 563, "lr": 0.001} +{"train_loss": 0.0333135724067688, "train_loss_bc": 0.0333135724067688, "train_loss_llm": 0.0, "grad_norm": 0.06828024983406067, "global_step": 20279, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03854638710618019, "train_loss_bc": 0.03854638710618019, "train_loss_llm": 0.0, "grad_norm": 0.09563682228326797, "global_step": 20280, "epoch": 563, "lr": 0.001} +{"train_loss": 0.02471938729286194, "train_loss_bc": 0.02471938729286194, "train_loss_llm": 0.0, "grad_norm": 0.02416795864701271, "global_step": 20281, "epoch": 563, "lr": 0.001} +{"train_loss": 0.029719198122620583, "train_loss_bc": 0.02970735915005207, "train_loss_llm": 0.0011838674545288086, "grad_norm": 0.03725658729672432, "global_step": 20282, "epoch": 563, "lr": 0.001} +{"train_loss": 0.02997461147606373, "train_loss_bc": 0.02997461147606373, "train_loss_llm": 0.0, "grad_norm": 0.052775051444768906, "global_step": 20283, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03660235181450844, "train_loss_bc": 0.03660235181450844, "train_loss_llm": 0.0, "grad_norm": 0.06950901448726654, "global_step": 20284, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03400014340877533, "train_loss_bc": 0.03400014340877533, "train_loss_llm": 0.0, "grad_norm": 0.031419191509485245, "global_step": 20285, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03229789808392525, "train_loss_bc": 0.03229789808392525, "train_loss_llm": 0.0, "grad_norm": 0.04560881480574608, "global_step": 20286, "epoch": 563, "lr": 0.001} +{"train_loss": 0.028984583914279938, "train_loss_bc": 0.028984583914279938, "train_loss_llm": 0.0, "grad_norm": 0.0668393075466156, "global_step": 20287, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03364584967494011, "train_loss_bc": 0.03364584967494011, "train_loss_llm": 0.0, "grad_norm": 0.07740610092878342, "global_step": 20288, "epoch": 563, "lr": 0.001} +{"train_loss": 0.032895393669605255, "train_loss_bc": 0.03287655860185623, "train_loss_llm": 0.0018836557865142822, "grad_norm": 0.030242495238780975, "global_step": 20289, "epoch": 563, "lr": 0.001} +{"train_loss": 0.035648126155138016, "train_loss_bc": 0.035648126155138016, "train_loss_llm": 0.0, "grad_norm": 0.04893626272678375, "global_step": 20290, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03551281616091728, "train_loss_bc": 0.03551281616091728, "train_loss_llm": 0.0, "grad_norm": 0.06320053339004517, "global_step": 20291, "epoch": 563, "lr": 0.001} +{"train_loss": 0.030377063900232315, "train_loss_bc": 0.030377063900232315, "train_loss_llm": 0.0, "grad_norm": 0.07928527891635895, "global_step": 20292, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03816809877753258, "train_loss_bc": 0.03816809877753258, "train_loss_llm": 0.0, "grad_norm": 0.026065845042467117, "global_step": 20293, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03268968313932419, "train_loss_bc": 0.03268968313932419, "train_loss_llm": 0.0, "grad_norm": 0.04801501706242561, "global_step": 20294, "epoch": 563, "lr": 0.001} +{"train_loss": 0.032044313848018646, "train_loss_bc": 0.032044313848018646, "train_loss_llm": 0.0, "grad_norm": 0.06790676712989807, "global_step": 20295, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03803664818406105, "train_loss_bc": 0.03803664818406105, "train_loss_llm": 0.0, "grad_norm": 0.089685820043087, "global_step": 20296, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03485429286956787, "train_loss_bc": 0.03485429286956787, "train_loss_llm": 0.0, "grad_norm": 0.03556741774082184, "global_step": 20297, "epoch": 563, "lr": 0.001} +{"train_loss": 0.0380820594727993, "train_loss_bc": 0.0380820594727993, "train_loss_llm": 0.0, "grad_norm": 0.058935198932886124, "global_step": 20298, "epoch": 563, "lr": 0.001} +{"train_loss": 0.030696386471390724, "train_loss_bc": 0.030696386471390724, "train_loss_llm": 0.0, "grad_norm": 0.06317361444234848, "global_step": 20299, "epoch": 563, "lr": 0.001} +{"train_loss": 0.02919916994869709, "train_loss_bc": 0.02919916994869709, "train_loss_llm": 0.0, "grad_norm": 0.07798779010772705, "global_step": 20300, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03687646612524986, "train_loss_bc": 0.03687646612524986, "train_loss_llm": 0.0, "grad_norm": 0.036377910524606705, "global_step": 20301, "epoch": 563, "lr": 0.001} +{"train_loss": 0.038656625896692276, "train_loss_bc": 0.038656625896692276, "train_loss_llm": 0.0, "grad_norm": 0.06329672783613205, "global_step": 20302, "epoch": 563, "lr": 0.001} +{"train_loss": 0.03332289235873355, "train_loss_bc": 0.038287848234176636, "train_loss_llm": 0.0, "grad_norm": 0.08048661053180695, "global_step": 20303, "epoch": 563, "lr": 0.001, "val_loss": 0.07978160679340363} +{"train_loss": 0.03236844390630722, "train_loss_bc": 0.03235561400651932, "train_loss_llm": 0.0012829303741455078, "grad_norm": 0.093626469373703, "global_step": 20304, "epoch": 564, "lr": 0.001} +{"train_loss": 0.0382043682038784, "train_loss_bc": 0.03820367529988289, "train_loss_llm": 6.940960884094238e-05, "grad_norm": 0.031268369406461716, "global_step": 20305, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03229039907455444, "train_loss_bc": 0.03229039907455444, "train_loss_llm": 0.0, "grad_norm": 0.03637026250362396, "global_step": 20306, "epoch": 564, "lr": 0.001} +{"train_loss": 0.028216911479830742, "train_loss_bc": 0.028202908113598824, "train_loss_llm": 0.0014002621173858643, "grad_norm": 0.052034892141819, "global_step": 20307, "epoch": 564, "lr": 0.001} +{"train_loss": 0.029298890382051468, "train_loss_bc": 0.029298890382051468, "train_loss_llm": 0.0, "grad_norm": 0.06154311075806618, "global_step": 20308, "epoch": 564, "lr": 0.001} +{"train_loss": 0.0331934317946434, "train_loss_bc": 0.03316827118396759, "train_loss_llm": 0.002516031265258789, "grad_norm": 0.029809478670358658, "global_step": 20309, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03358197212219238, "train_loss_bc": 0.03358197212219238, "train_loss_llm": 0.0, "grad_norm": 0.03618936613202095, "global_step": 20310, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03251957520842552, "train_loss_bc": 0.03251957520842552, "train_loss_llm": 0.0, "grad_norm": 0.04681205004453659, "global_step": 20311, "epoch": 564, "lr": 0.001} +{"train_loss": 0.030740711838006973, "train_loss_bc": 0.030740711838006973, "train_loss_llm": 0.0, "grad_norm": 0.05056195333600044, "global_step": 20312, "epoch": 564, "lr": 0.001} +{"train_loss": 0.02534276619553566, "train_loss_bc": 0.025338873267173767, "train_loss_llm": 0.00038930773735046387, "grad_norm": 0.026062043383717537, "global_step": 20313, "epoch": 564, "lr": 0.001} +{"train_loss": 0.039801448583602905, "train_loss_bc": 0.039801448583602905, "train_loss_llm": 0.0, "grad_norm": 0.04831905663013458, "global_step": 20314, "epoch": 564, "lr": 0.001} +{"train_loss": 0.0276329405605793, "train_loss_bc": 0.02762036956846714, "train_loss_llm": 0.0012570321559906006, "grad_norm": 0.06291422992944717, "global_step": 20315, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03071877546608448, "train_loss_bc": 0.030704282224178314, "train_loss_llm": 0.0014493763446807861, "grad_norm": 0.08226212859153748, "global_step": 20316, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03039643168449402, "train_loss_bc": 0.030392711982131004, "train_loss_llm": 0.0003719627857208252, "grad_norm": 0.032246481627225876, "global_step": 20317, "epoch": 564, "lr": 0.001} +{"train_loss": 0.02988303452730179, "train_loss_bc": 0.02988303452730179, "train_loss_llm": 0.0, "grad_norm": 0.06141791492700577, "global_step": 20318, "epoch": 564, "lr": 0.001} +{"train_loss": 0.031025391072034836, "train_loss_bc": 0.031025391072034836, "train_loss_llm": 0.0, "grad_norm": 0.0806642696261406, "global_step": 20319, "epoch": 564, "lr": 0.001} +{"train_loss": 0.034688800573349, "train_loss_bc": 0.034688800573349, "train_loss_llm": 0.0, "grad_norm": 0.10623785108327866, "global_step": 20320, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03187936544418335, "train_loss_bc": 0.03187936544418335, "train_loss_llm": 0.0, "grad_norm": 0.02629811316728592, "global_step": 20321, "epoch": 564, "lr": 0.001} +{"train_loss": 0.032912079244852066, "train_loss_bc": 0.032912079244852066, "train_loss_llm": 0.0, "grad_norm": 0.04438726231455803, "global_step": 20322, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03461150825023651, "train_loss_bc": 0.03461150825023651, "train_loss_llm": 0.0, "grad_norm": 0.06512444466352463, "global_step": 20323, "epoch": 564, "lr": 0.001} +{"train_loss": 0.037658534944057465, "train_loss_bc": 0.03763266280293465, "train_loss_llm": 0.0025873184204101562, "grad_norm": 0.09374329447746277, "global_step": 20324, "epoch": 564, "lr": 0.001} +{"train_loss": 0.033855415880680084, "train_loss_bc": 0.033855415880680084, "train_loss_llm": 0.0, "grad_norm": 0.03087678737938404, "global_step": 20325, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03523623198270798, "train_loss_bc": 0.03523623198270798, "train_loss_llm": 0.0, "grad_norm": 0.04276862367987633, "global_step": 20326, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03286086767911911, "train_loss_bc": 0.03286086767911911, "train_loss_llm": 0.0, "grad_norm": 0.04645306617021561, "global_step": 20327, "epoch": 564, "lr": 0.001} +{"train_loss": 0.035374630242586136, "train_loss_bc": 0.035374630242586136, "train_loss_llm": 0.0, "grad_norm": 0.05620180815458298, "global_step": 20328, "epoch": 564, "lr": 0.001} +{"train_loss": 0.029732078313827515, "train_loss_bc": 0.029732078313827515, "train_loss_llm": 0.0, "grad_norm": 0.023986918851733208, "global_step": 20329, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03564285486936569, "train_loss_bc": 0.03563067317008972, "train_loss_llm": 0.0012180805206298828, "grad_norm": 0.0382160060107708, "global_step": 20330, "epoch": 564, "lr": 0.001} +{"train_loss": 0.031146030873060226, "train_loss_bc": 0.03113693557679653, "train_loss_llm": 0.0009095072746276855, "grad_norm": 0.05081145465373993, "global_step": 20331, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03592558205127716, "train_loss_bc": 0.03592558205127716, "train_loss_llm": 0.0, "grad_norm": 0.0640076994895935, "global_step": 20332, "epoch": 564, "lr": 0.001} +{"train_loss": 0.031207604333758354, "train_loss_bc": 0.031207604333758354, "train_loss_llm": 0.0, "grad_norm": 0.02897052653133869, "global_step": 20333, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03993593901395798, "train_loss_bc": 0.039926718920469284, "train_loss_llm": 0.0009221732616424561, "grad_norm": 0.05129845440387726, "global_step": 20334, "epoch": 564, "lr": 0.001} +{"train_loss": 0.0386541374027729, "train_loss_bc": 0.0386541374027729, "train_loss_llm": 0.0, "grad_norm": 0.06509724259376526, "global_step": 20335, "epoch": 564, "lr": 0.001} +{"train_loss": 0.029650339856743813, "train_loss_bc": 0.029650339856743813, "train_loss_llm": 0.0, "grad_norm": 0.075168177485466, "global_step": 20336, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03479825705289841, "train_loss_bc": 0.03479825705289841, "train_loss_llm": 0.0, "grad_norm": 0.036121588200330734, "global_step": 20337, "epoch": 564, "lr": 0.001} +{"train_loss": 0.03667249158024788, "train_loss_bc": 0.03667249158024788, "train_loss_llm": 0.0, "grad_norm": 0.05760112404823303, "global_step": 20338, "epoch": 564, "lr": 0.001} +{"train_loss": 0.033141139377322465, "train_loss_bc": 0.03542277589440346, "train_loss_llm": 0.0, "grad_norm": 0.08650464564561844, "global_step": 20339, "epoch": 564, "lr": 0.001, "val_loss": 0.0793205201625824} +{"train_loss": 0.028110351413488388, "train_loss_bc": 0.028106102719902992, "train_loss_llm": 0.00042489171028137207, "grad_norm": 0.09472990781068802, "global_step": 20340, "epoch": 565, "lr": 0.001} +{"train_loss": 0.0244439747184515, "train_loss_bc": 0.02444382756948471, "train_loss_llm": 1.4662742614746094e-05, "grad_norm": 0.029703926295042038, "global_step": 20341, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03675028309226036, "train_loss_bc": 0.03674225136637688, "train_loss_llm": 0.0008030235767364502, "grad_norm": 0.047552138566970825, "global_step": 20342, "epoch": 565, "lr": 0.001} +{"train_loss": 0.029282746836543083, "train_loss_bc": 0.029282746836543083, "train_loss_llm": 0.0, "grad_norm": 0.06952761113643646, "global_step": 20343, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03465491533279419, "train_loss_bc": 0.03465491533279419, "train_loss_llm": 0.0, "grad_norm": 0.08250121772289276, "global_step": 20344, "epoch": 565, "lr": 0.001} +{"train_loss": 0.037197258323431015, "train_loss_bc": 0.037197258323431015, "train_loss_llm": 0.0, "grad_norm": 0.03045041672885418, "global_step": 20345, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03253614157438278, "train_loss_bc": 0.03253614157438278, "train_loss_llm": 0.0, "grad_norm": 0.048535387963056564, "global_step": 20346, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03569365665316582, "train_loss_bc": 0.03569365665316582, "train_loss_llm": 0.0, "grad_norm": 0.06879159063100815, "global_step": 20347, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03208301216363907, "train_loss_bc": 0.03208301216363907, "train_loss_llm": 0.0, "grad_norm": 0.08830702304840088, "global_step": 20348, "epoch": 565, "lr": 0.001} +{"train_loss": 0.02522822469472885, "train_loss_bc": 0.02522822469472885, "train_loss_llm": 0.0, "grad_norm": 0.025399809703230858, "global_step": 20349, "epoch": 565, "lr": 0.001} +{"train_loss": 0.0268594678491354, "train_loss_bc": 0.02685399167239666, "train_loss_llm": 0.0005476772785186768, "grad_norm": 0.04371827840805054, "global_step": 20350, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03086032159626484, "train_loss_bc": 0.03086032159626484, "train_loss_llm": 0.0, "grad_norm": 0.05380077660083771, "global_step": 20351, "epoch": 565, "lr": 0.001} +{"train_loss": 0.031659506261348724, "train_loss_bc": 0.031659506261348724, "train_loss_llm": 0.0, "grad_norm": 0.06111616641283035, "global_step": 20352, "epoch": 565, "lr": 0.001} +{"train_loss": 0.034008659422397614, "train_loss_bc": 0.03400658071041107, "train_loss_llm": 0.00020769238471984863, "grad_norm": 0.0277172289788723, "global_step": 20353, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03035525418817997, "train_loss_bc": 0.03035525418817997, "train_loss_llm": 0.0, "grad_norm": 0.040070608258247375, "global_step": 20354, "epoch": 565, "lr": 0.001} +{"train_loss": 0.034887805581092834, "train_loss_bc": 0.03487785905599594, "train_loss_llm": 0.0009945034980773926, "grad_norm": 0.0505082868039608, "global_step": 20355, "epoch": 565, "lr": 0.001} +{"train_loss": 0.0337994247674942, "train_loss_bc": 0.0337994247674942, "train_loss_llm": 0.0, "grad_norm": 0.061909738928079605, "global_step": 20356, "epoch": 565, "lr": 0.001} +{"train_loss": 0.039606012403964996, "train_loss_bc": 0.039606012403964996, "train_loss_llm": 0.0, "grad_norm": 0.030247658491134644, "global_step": 20357, "epoch": 565, "lr": 0.001} +{"train_loss": 0.0304915439337492, "train_loss_bc": 0.0304915439337492, "train_loss_llm": 0.0, "grad_norm": 0.05339943990111351, "global_step": 20358, "epoch": 565, "lr": 0.001} +{"train_loss": 0.029292525723576546, "train_loss_bc": 0.02929086424410343, "train_loss_llm": 0.00016614794731140137, "grad_norm": 0.07739435881376266, "global_step": 20359, "epoch": 565, "lr": 0.001} +{"train_loss": 0.028384847566485405, "train_loss_bc": 0.028384847566485405, "train_loss_llm": 0.0, "grad_norm": 0.08380775898694992, "global_step": 20360, "epoch": 565, "lr": 0.001} +{"train_loss": 0.033443715423345566, "train_loss_bc": 0.033443715423345566, "train_loss_llm": 0.0, "grad_norm": 0.026865631341934204, "global_step": 20361, "epoch": 565, "lr": 0.001} +{"train_loss": 0.035930849611759186, "train_loss_bc": 0.035930849611759186, "train_loss_llm": 0.0, "grad_norm": 0.038526102900505066, "global_step": 20362, "epoch": 565, "lr": 0.001} +{"train_loss": 0.030797459185123444, "train_loss_bc": 0.030786361545324326, "train_loss_llm": 0.0011097192764282227, "grad_norm": 0.05776914209127426, "global_step": 20363, "epoch": 565, "lr": 0.001} +{"train_loss": 0.0381774939596653, "train_loss_bc": 0.0381774939596653, "train_loss_llm": 0.0, "grad_norm": 0.07180288434028625, "global_step": 20364, "epoch": 565, "lr": 0.001} +{"train_loss": 0.033042557537555695, "train_loss_bc": 0.033042557537555695, "train_loss_llm": 0.0, "grad_norm": 0.023948485031723976, "global_step": 20365, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03671012073755264, "train_loss_bc": 0.03671012073755264, "train_loss_llm": 0.0, "grad_norm": 0.03951331228017807, "global_step": 20366, "epoch": 565, "lr": 0.001} +{"train_loss": 0.0370238795876503, "train_loss_bc": 0.0370238795876503, "train_loss_llm": 0.0, "grad_norm": 0.055462684482336044, "global_step": 20367, "epoch": 565, "lr": 0.001} +{"train_loss": 0.0347965806722641, "train_loss_bc": 0.0347965806722641, "train_loss_llm": 0.0, "grad_norm": 0.07638723403215408, "global_step": 20368, "epoch": 565, "lr": 0.001} +{"train_loss": 0.029120204970240593, "train_loss_bc": 0.029120204970240593, "train_loss_llm": 0.0, "grad_norm": 0.024372298270463943, "global_step": 20369, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03637878596782684, "train_loss_bc": 0.03637878596782684, "train_loss_llm": 0.0, "grad_norm": 0.03977147489786148, "global_step": 20370, "epoch": 565, "lr": 0.001} +{"train_loss": 0.036983199417591095, "train_loss_bc": 0.036983199417591095, "train_loss_llm": 0.0, "grad_norm": 0.05640514940023422, "global_step": 20371, "epoch": 565, "lr": 0.001} +{"train_loss": 0.037156011909246445, "train_loss_bc": 0.037156011909246445, "train_loss_llm": 0.0, "grad_norm": 0.07192166149616241, "global_step": 20372, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03511920943856239, "train_loss_bc": 0.03511920943856239, "train_loss_llm": 0.0, "grad_norm": 0.03756795823574066, "global_step": 20373, "epoch": 565, "lr": 0.001} +{"train_loss": 0.03625091537833214, "train_loss_bc": 0.03623626381158829, "train_loss_llm": 0.0014649927616119385, "grad_norm": 0.05891403183341026, "global_step": 20374, "epoch": 565, "lr": 0.001} +{"train_loss": 0.033164395898994475, "train_loss_bc": 0.03680133447051048, "train_loss_llm": 0.0, "grad_norm": 0.07468847930431366, "global_step": 20375, "epoch": 565, "lr": 0.001, "train/cumulative_reward": 84.99644217970491, "train/mean_score": 14.812147025421723, "train/success_rate": 0.4, "test/cumulative_reward": 39.657814967820165, "test/mean_score": 7.794962930135192, "test/success_rate": 0.18, "val_loss": 0.08009812235832214, "train_action_mse_error": 0.036302220076322556} +{"train_loss": 0.03174743801355362, "train_loss_bc": 0.03174743801355362, "train_loss_llm": 0.0, "grad_norm": 0.08760744333267212, "global_step": 20376, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03624198958277702, "train_loss_bc": 0.0362330786883831, "train_loss_llm": 0.0008911490440368652, "grad_norm": 0.03320581093430519, "global_step": 20377, "epoch": 566, "lr": 0.001} +{"train_loss": 0.033304594457149506, "train_loss_bc": 0.033304594457149506, "train_loss_llm": 0.0, "grad_norm": 0.047483738511800766, "global_step": 20378, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03139781579375267, "train_loss_bc": 0.03139781579375267, "train_loss_llm": 0.0, "grad_norm": 0.06854100525379181, "global_step": 20379, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03402292728424072, "train_loss_bc": 0.03402292728424072, "train_loss_llm": 0.0, "grad_norm": 0.08979245275259018, "global_step": 20380, "epoch": 566, "lr": 0.001} +{"train_loss": 0.035982850939035416, "train_loss_bc": 0.035982850939035416, "train_loss_llm": 0.0, "grad_norm": 0.03461664915084839, "global_step": 20381, "epoch": 566, "lr": 0.001} +{"train_loss": 0.030072137713432312, "train_loss_bc": 0.030072137713432312, "train_loss_llm": 0.0, "grad_norm": 0.05323699116706848, "global_step": 20382, "epoch": 566, "lr": 0.001} +{"train_loss": 0.031841978430747986, "train_loss_bc": 0.031841978430747986, "train_loss_llm": 0.0, "grad_norm": 0.06815074384212494, "global_step": 20383, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03133562207221985, "train_loss_bc": 0.03133562207221985, "train_loss_llm": 0.0, "grad_norm": 0.08399265259504318, "global_step": 20384, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03204188123345375, "train_loss_bc": 0.03204188123345375, "train_loss_llm": 0.0, "grad_norm": 0.03103294037282467, "global_step": 20385, "epoch": 566, "lr": 0.001} +{"train_loss": 0.037389907985925674, "train_loss_bc": 0.037389907985925674, "train_loss_llm": 0.0, "grad_norm": 0.04695862904191017, "global_step": 20386, "epoch": 566, "lr": 0.001} +{"train_loss": 0.025542860850691795, "train_loss_bc": 0.025542860850691795, "train_loss_llm": 0.0, "grad_norm": 0.06080435961484909, "global_step": 20387, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03211412578821182, "train_loss_bc": 0.03208327293395996, "train_loss_llm": 0.0030852556228637695, "grad_norm": 0.07507451623678207, "global_step": 20388, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03243060037493706, "train_loss_bc": 0.03240956738591194, "train_loss_llm": 0.002103179693222046, "grad_norm": 0.02613910101354122, "global_step": 20389, "epoch": 566, "lr": 0.001} +{"train_loss": 0.026744982227683067, "train_loss_bc": 0.02674172818660736, "train_loss_llm": 0.0003253817558288574, "grad_norm": 0.04352080076932907, "global_step": 20390, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03236039727926254, "train_loss_bc": 0.03236039727926254, "train_loss_llm": 0.0, "grad_norm": 0.06311450153589249, "global_step": 20391, "epoch": 566, "lr": 0.001} +{"train_loss": 0.034127071499824524, "train_loss_bc": 0.034127071499824524, "train_loss_llm": 0.0, "grad_norm": 0.0695032849907875, "global_step": 20392, "epoch": 566, "lr": 0.001} +{"train_loss": 0.030259229242801666, "train_loss_bc": 0.030259229242801666, "train_loss_llm": 0.0, "grad_norm": 0.030135657638311386, "global_step": 20393, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03737529367208481, "train_loss_bc": 0.03737529367208481, "train_loss_llm": 0.0, "grad_norm": 0.04242240637540817, "global_step": 20394, "epoch": 566, "lr": 0.001} +{"train_loss": 0.035978589206933975, "train_loss_bc": 0.035978589206933975, "train_loss_llm": 0.0, "grad_norm": 0.049413811415433884, "global_step": 20395, "epoch": 566, "lr": 0.001} +{"train_loss": 0.031322773545980453, "train_loss_bc": 0.031322773545980453, "train_loss_llm": 0.0, "grad_norm": 0.06306058168411255, "global_step": 20396, "epoch": 566, "lr": 0.001} +{"train_loss": 0.030994784086942673, "train_loss_bc": 0.030994784086942673, "train_loss_llm": 0.0, "grad_norm": 0.03003203123807907, "global_step": 20397, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03186862915754318, "train_loss_bc": 0.03185351565480232, "train_loss_llm": 0.0015114843845367432, "grad_norm": 0.05182552710175514, "global_step": 20398, "epoch": 566, "lr": 0.001} +{"train_loss": 0.033426493406295776, "train_loss_bc": 0.033426493406295776, "train_loss_llm": 0.0, "grad_norm": 0.06524408608675003, "global_step": 20399, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03711431846022606, "train_loss_bc": 0.03711431846022606, "train_loss_llm": 0.0, "grad_norm": 0.0699920579791069, "global_step": 20400, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03186419606208801, "train_loss_bc": 0.031857024878263474, "train_loss_llm": 0.0007172226905822754, "grad_norm": 0.03120107203722, "global_step": 20401, "epoch": 566, "lr": 0.001} +{"train_loss": 0.030229220166802406, "train_loss_bc": 0.030229220166802406, "train_loss_llm": 0.0, "grad_norm": 0.04453837126493454, "global_step": 20402, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03649688884615898, "train_loss_bc": 0.03649688884615898, "train_loss_llm": 0.0, "grad_norm": 0.06634128838777542, "global_step": 20403, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03534575179219246, "train_loss_bc": 0.035327088087797165, "train_loss_llm": 0.0018662214279174805, "grad_norm": 0.08161426335573196, "global_step": 20404, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03434763848781586, "train_loss_bc": 0.03434763848781586, "train_loss_llm": 0.0, "grad_norm": 0.028612155467271805, "global_step": 20405, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03618635609745979, "train_loss_bc": 0.03618635609745979, "train_loss_llm": 0.0, "grad_norm": 0.050876397639513016, "global_step": 20406, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03002639301121235, "train_loss_bc": 0.030025916174054146, "train_loss_llm": 4.7713518142700195e-05, "grad_norm": 0.07183607667684555, "global_step": 20407, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03855711966753006, "train_loss_bc": 0.03855711966753006, "train_loss_llm": 0.0, "grad_norm": 0.09426713734865189, "global_step": 20408, "epoch": 566, "lr": 0.001} +{"train_loss": 0.03387187421321869, "train_loss_bc": 0.03387187421321869, "train_loss_llm": 0.0, "grad_norm": 0.029847797006368637, "global_step": 20409, "epoch": 566, "lr": 0.001} +{"train_loss": 0.033803489059209824, "train_loss_bc": 0.033803489059209824, "train_loss_llm": 0.0, "grad_norm": 0.04067030921578407, "global_step": 20410, "epoch": 566, "lr": 0.001} +{"train_loss": 0.033193856581217714, "train_loss_bc": 0.0372106172144413, "train_loss_llm": 0.0, "grad_norm": 0.05839678272604942, "global_step": 20411, "epoch": 566, "lr": 0.001, "val_loss": 0.08016248047351837} +{"train_loss": 0.02955213189125061, "train_loss_bc": 0.02955213189125061, "train_loss_llm": 0.0, "grad_norm": 0.06861318647861481, "global_step": 20412, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03368232026696205, "train_loss_bc": 0.033682022243738174, "train_loss_llm": 2.9653310775756836e-05, "grad_norm": 0.029543709009885788, "global_step": 20413, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03610522672533989, "train_loss_bc": 0.03610522672533989, "train_loss_llm": 0.0, "grad_norm": 0.041636619716882706, "global_step": 20414, "epoch": 567, "lr": 0.001} +{"train_loss": 0.030423520132899284, "train_loss_bc": 0.03040580078959465, "train_loss_llm": 0.0017720162868499756, "grad_norm": 0.05619330331683159, "global_step": 20415, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03798853978514671, "train_loss_bc": 0.03798523172736168, "train_loss_llm": 0.00033092498779296875, "grad_norm": 0.06651253253221512, "global_step": 20416, "epoch": 567, "lr": 0.001} +{"train_loss": 0.027844449505209923, "train_loss_bc": 0.027844449505209923, "train_loss_llm": 0.0, "grad_norm": 0.021729251369833946, "global_step": 20417, "epoch": 567, "lr": 0.001} +{"train_loss": 0.029834726825356483, "train_loss_bc": 0.029832376167178154, "train_loss_llm": 0.00023514032363891602, "grad_norm": 0.03575928881764412, "global_step": 20418, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03057853691279888, "train_loss_bc": 0.03057853691279888, "train_loss_llm": 0.0, "grad_norm": 0.057697318494319916, "global_step": 20419, "epoch": 567, "lr": 0.001} +{"train_loss": 0.032797057181596756, "train_loss_bc": 0.032797057181596756, "train_loss_llm": 0.0, "grad_norm": 0.07699093222618103, "global_step": 20420, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03820344805717468, "train_loss_bc": 0.03820344805717468, "train_loss_llm": 0.0, "grad_norm": 0.038731664419174194, "global_step": 20421, "epoch": 567, "lr": 0.001} +{"train_loss": 0.024324338883161545, "train_loss_bc": 0.024317245930433273, "train_loss_llm": 0.0007092058658599854, "grad_norm": 0.05559727922081947, "global_step": 20422, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03289422020316124, "train_loss_bc": 0.03289422020316124, "train_loss_llm": 0.0, "grad_norm": 0.07515040785074234, "global_step": 20423, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03330078348517418, "train_loss_bc": 0.0332932285964489, "train_loss_llm": 0.0007555186748504639, "grad_norm": 0.09496249258518219, "global_step": 20424, "epoch": 567, "lr": 0.001} +{"train_loss": 0.037009138613939285, "train_loss_bc": 0.037009138613939285, "train_loss_llm": 0.0, "grad_norm": 0.02972772717475891, "global_step": 20425, "epoch": 567, "lr": 0.001} +{"train_loss": 0.032669782638549805, "train_loss_bc": 0.03265991434454918, "train_loss_llm": 0.0009866654872894287, "grad_norm": 0.045389339327812195, "global_step": 20426, "epoch": 567, "lr": 0.001} +{"train_loss": 0.036554496735334396, "train_loss_bc": 0.036554496735334396, "train_loss_llm": 0.0, "grad_norm": 0.0640595480799675, "global_step": 20427, "epoch": 567, "lr": 0.001} +{"train_loss": 0.029736042022705078, "train_loss_bc": 0.029733551666140556, "train_loss_llm": 0.0002490580081939697, "grad_norm": 0.08533670008182526, "global_step": 20428, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03036169707775116, "train_loss_bc": 0.03036169707775116, "train_loss_llm": 0.0, "grad_norm": 0.019335756078362465, "global_step": 20429, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03222809359431267, "train_loss_bc": 0.03222809359431267, "train_loss_llm": 0.0, "grad_norm": 0.03741997480392456, "global_step": 20430, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03314157575368881, "train_loss_bc": 0.03314157575368881, "train_loss_llm": 0.0, "grad_norm": 0.056000255048274994, "global_step": 20431, "epoch": 567, "lr": 0.001} +{"train_loss": 0.033968206495046616, "train_loss_bc": 0.033961862325668335, "train_loss_llm": 0.0006344616413116455, "grad_norm": 0.06880184262990952, "global_step": 20432, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03093131072819233, "train_loss_bc": 0.03093131072819233, "train_loss_llm": 0.0, "grad_norm": 0.026735899969935417, "global_step": 20433, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03563595935702324, "train_loss_bc": 0.03563595935702324, "train_loss_llm": 0.0, "grad_norm": 0.042337510734796524, "global_step": 20434, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03280893340706825, "train_loss_bc": 0.03280893340706825, "train_loss_llm": 0.0, "grad_norm": 0.050452396273612976, "global_step": 20435, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03171492740511894, "train_loss_bc": 0.03171492740511894, "train_loss_llm": 0.0, "grad_norm": 0.06445134431123734, "global_step": 20436, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03100365400314331, "train_loss_bc": 0.030985556542873383, "train_loss_llm": 0.0018096566200256348, "grad_norm": 0.02330329269170761, "global_step": 20437, "epoch": 567, "lr": 0.001} +{"train_loss": 0.02451569400727749, "train_loss_bc": 0.02451569400727749, "train_loss_llm": 0.0, "grad_norm": 0.0374668650329113, "global_step": 20438, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03600851446390152, "train_loss_bc": 0.03599319979548454, "train_loss_llm": 0.0015316009521484375, "grad_norm": 0.05012059584259987, "global_step": 20439, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03511432930827141, "train_loss_bc": 0.03511432930827141, "train_loss_llm": 0.0, "grad_norm": 0.06645895540714264, "global_step": 20440, "epoch": 567, "lr": 0.001} +{"train_loss": 0.0312582328915596, "train_loss_bc": 0.0312582328915596, "train_loss_llm": 0.0, "grad_norm": 0.03557142987847328, "global_step": 20441, "epoch": 567, "lr": 0.001} +{"train_loss": 0.037690576165914536, "train_loss_bc": 0.037690576165914536, "train_loss_llm": 0.0, "grad_norm": 0.045782420784235, "global_step": 20442, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03424924612045288, "train_loss_bc": 0.03424924612045288, "train_loss_llm": 0.0, "grad_norm": 0.07253440469503403, "global_step": 20443, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03721454739570618, "train_loss_bc": 0.037203021347522736, "train_loss_llm": 0.0011525750160217285, "grad_norm": 0.091609887778759, "global_step": 20444, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03600293770432472, "train_loss_bc": 0.03599997237324715, "train_loss_llm": 0.00029659271240234375, "grad_norm": 0.028460631147027016, "global_step": 20445, "epoch": 567, "lr": 0.001} +{"train_loss": 0.039949171245098114, "train_loss_bc": 0.039949171245098114, "train_loss_llm": 0.0, "grad_norm": 0.04696628451347351, "global_step": 20446, "epoch": 567, "lr": 0.001} +{"train_loss": 0.03301746894915899, "train_loss_bc": 0.03133251518011093, "train_loss_llm": 0.0, "grad_norm": 0.058406684547662735, "global_step": 20447, "epoch": 567, "lr": 0.001, "val_loss": 0.07944721728563309} +{"train_loss": 0.034402038902044296, "train_loss_bc": 0.034402038902044296, "train_loss_llm": 0.0, "grad_norm": 0.07101187109947205, "global_step": 20448, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03490981459617615, "train_loss_bc": 0.03490981459617615, "train_loss_llm": 0.0, "grad_norm": 0.0303702000528574, "global_step": 20449, "epoch": 568, "lr": 0.001} +{"train_loss": 0.02621820569038391, "train_loss_bc": 0.026197124272584915, "train_loss_llm": 0.002108067274093628, "grad_norm": 0.037613674998283386, "global_step": 20450, "epoch": 568, "lr": 0.001} +{"train_loss": 0.028105121105909348, "train_loss_bc": 0.028100913390517235, "train_loss_llm": 0.0004208087921142578, "grad_norm": 0.05008366331458092, "global_step": 20451, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03088456392288208, "train_loss_bc": 0.030844062566757202, "train_loss_llm": 0.004050225019454956, "grad_norm": 0.06306909024715424, "global_step": 20452, "epoch": 568, "lr": 0.001} +{"train_loss": 0.038653306663036346, "train_loss_bc": 0.038653306663036346, "train_loss_llm": 0.0, "grad_norm": 0.028513053432106972, "global_step": 20453, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03319938853383064, "train_loss_bc": 0.03319938853383064, "train_loss_llm": 0.0, "grad_norm": 0.040450260043144226, "global_step": 20454, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03293471038341522, "train_loss_bc": 0.03293471038341522, "train_loss_llm": 0.0, "grad_norm": 0.052285127341747284, "global_step": 20455, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03443332761526108, "train_loss_bc": 0.03443332761526108, "train_loss_llm": 0.0, "grad_norm": 0.05991946533322334, "global_step": 20456, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03512075915932655, "train_loss_bc": 0.03512075915932655, "train_loss_llm": 0.0, "grad_norm": 0.0280783548951149, "global_step": 20457, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03235545754432678, "train_loss_bc": 0.03235545754432678, "train_loss_llm": 0.0, "grad_norm": 0.044937364757061005, "global_step": 20458, "epoch": 568, "lr": 0.001} +{"train_loss": 0.02686373144388199, "train_loss_bc": 0.02686373144388199, "train_loss_llm": 0.0, "grad_norm": 0.047080788761377335, "global_step": 20459, "epoch": 568, "lr": 0.001} +{"train_loss": 0.0315030999481678, "train_loss_bc": 0.03150266408920288, "train_loss_llm": 4.374980926513672e-05, "grad_norm": 0.0715734139084816, "global_step": 20460, "epoch": 568, "lr": 0.001} +{"train_loss": 0.033623769879341125, "train_loss_bc": 0.033623769879341125, "train_loss_llm": 0.0, "grad_norm": 0.027809901162981987, "global_step": 20461, "epoch": 568, "lr": 0.001} +{"train_loss": 0.035140447318553925, "train_loss_bc": 0.035139620304107666, "train_loss_llm": 8.279085159301758e-05, "grad_norm": 0.043483905494213104, "global_step": 20462, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03344319388270378, "train_loss_bc": 0.0334119014441967, "train_loss_llm": 0.0031292736530303955, "grad_norm": 0.05358674377202988, "global_step": 20463, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03914451226592064, "train_loss_bc": 0.03914451226592064, "train_loss_llm": 0.0, "grad_norm": 0.0662466436624527, "global_step": 20464, "epoch": 568, "lr": 0.001} +{"train_loss": 0.032030701637268066, "train_loss_bc": 0.032030701637268066, "train_loss_llm": 0.0, "grad_norm": 0.036233752965927124, "global_step": 20465, "epoch": 568, "lr": 0.001} +{"train_loss": 0.02953423000872135, "train_loss_bc": 0.02953423000872135, "train_loss_llm": 0.0, "grad_norm": 0.05186526104807854, "global_step": 20466, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03335420787334442, "train_loss_bc": 0.03335420787334442, "train_loss_llm": 0.0, "grad_norm": 0.07188591361045837, "global_step": 20467, "epoch": 568, "lr": 0.001} +{"train_loss": 0.0317106693983078, "train_loss_bc": 0.0317106693983078, "train_loss_llm": 0.0, "grad_norm": 0.09735335409641266, "global_step": 20468, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03629574179649353, "train_loss_bc": 0.03629574179649353, "train_loss_llm": 0.0, "grad_norm": 0.03118692897260189, "global_step": 20469, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03644755855202675, "train_loss_bc": 0.03644755855202675, "train_loss_llm": 0.0, "grad_norm": 0.051596034318208694, "global_step": 20470, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03289226070046425, "train_loss_bc": 0.03289226070046425, "train_loss_llm": 0.0, "grad_norm": 0.0624757781624794, "global_step": 20471, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03061944805085659, "train_loss_bc": 0.03061944805085659, "train_loss_llm": 0.0, "grad_norm": 0.07762395590543747, "global_step": 20472, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03275294229388237, "train_loss_bc": 0.03275294229388237, "train_loss_llm": 0.0, "grad_norm": 0.03444498032331467, "global_step": 20473, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03598809242248535, "train_loss_bc": 0.035978928208351135, "train_loss_llm": 0.0009163916110992432, "grad_norm": 0.04303676262497902, "global_step": 20474, "epoch": 568, "lr": 0.001} +{"train_loss": 0.030043458566069603, "train_loss_bc": 0.030043458566069603, "train_loss_llm": 0.0, "grad_norm": 0.0612310916185379, "global_step": 20475, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03532642871141434, "train_loss_bc": 0.0353071466088295, "train_loss_llm": 0.0019282698631286621, "grad_norm": 0.0704832524061203, "global_step": 20476, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03536365181207657, "train_loss_bc": 0.03536365181207657, "train_loss_llm": 0.0, "grad_norm": 0.03300721198320389, "global_step": 20477, "epoch": 568, "lr": 0.001} +{"train_loss": 0.033176083117723465, "train_loss_bc": 0.033176083117723465, "train_loss_llm": 0.0, "grad_norm": 0.03831245377659798, "global_step": 20478, "epoch": 568, "lr": 0.001} +{"train_loss": 0.036619823426008224, "train_loss_bc": 0.036619823426008224, "train_loss_llm": 0.0, "grad_norm": 0.05108853429555893, "global_step": 20479, "epoch": 568, "lr": 0.001} +{"train_loss": 0.033658213913440704, "train_loss_bc": 0.033658213913440704, "train_loss_llm": 0.0, "grad_norm": 0.06886708736419678, "global_step": 20480, "epoch": 568, "lr": 0.001} +{"train_loss": 0.030828362330794334, "train_loss_bc": 0.03082801029086113, "train_loss_llm": 3.528594970703125e-05, "grad_norm": 0.02779717929661274, "global_step": 20481, "epoch": 568, "lr": 0.001} +{"train_loss": 0.026836801320314407, "train_loss_bc": 0.026836801320314407, "train_loss_llm": 0.0, "grad_norm": 0.03768350929021835, "global_step": 20482, "epoch": 568, "lr": 0.001} +{"train_loss": 0.03302688368906578, "train_loss_bc": 0.03452867642045021, "train_loss_llm": 0.002501159906387329, "grad_norm": 0.05541471764445305, "global_step": 20483, "epoch": 568, "lr": 0.001, "val_loss": 0.07982286810874939} +{"train_loss": 0.03429511561989784, "train_loss_bc": 0.034281324595212936, "train_loss_llm": 0.0013790130615234375, "grad_norm": 0.06874608248472214, "global_step": 20484, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03896329551935196, "train_loss_bc": 0.03896329551935196, "train_loss_llm": 0.0, "grad_norm": 0.0309918150305748, "global_step": 20485, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03202882036566734, "train_loss_bc": 0.03200520947575569, "train_loss_llm": 0.0023610591888427734, "grad_norm": 0.046656303107738495, "global_step": 20486, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03255906328558922, "train_loss_bc": 0.03255906328558922, "train_loss_llm": 0.0, "grad_norm": 0.05413149297237396, "global_step": 20487, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03183095529675484, "train_loss_bc": 0.031829096376895905, "train_loss_llm": 0.00018593668937683105, "grad_norm": 0.06880175322294235, "global_step": 20488, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03022252395749092, "train_loss_bc": 0.03020748682320118, "train_loss_llm": 0.001503676176071167, "grad_norm": 0.0280950665473938, "global_step": 20489, "epoch": 569, "lr": 0.001} +{"train_loss": 0.030352728441357613, "train_loss_bc": 0.030352728441357613, "train_loss_llm": 0.0, "grad_norm": 0.044960808008909225, "global_step": 20490, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03395235538482666, "train_loss_bc": 0.03395235538482666, "train_loss_llm": 0.0, "grad_norm": 0.06319843977689743, "global_step": 20491, "epoch": 569, "lr": 0.001} +{"train_loss": 0.02757607400417328, "train_loss_bc": 0.02757607400417328, "train_loss_llm": 0.0, "grad_norm": 0.07741709798574448, "global_step": 20492, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03573431447148323, "train_loss_bc": 0.03573431447148323, "train_loss_llm": 0.0, "grad_norm": 0.03237329423427582, "global_step": 20493, "epoch": 569, "lr": 0.001} +{"train_loss": 0.0362260639667511, "train_loss_bc": 0.0362260639667511, "train_loss_llm": 0.0, "grad_norm": 0.04857124388217926, "global_step": 20494, "epoch": 569, "lr": 0.001} +{"train_loss": 0.028613034635782242, "train_loss_bc": 0.028613034635782242, "train_loss_llm": 0.0, "grad_norm": 0.056493278592824936, "global_step": 20495, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03641185536980629, "train_loss_bc": 0.03641185536980629, "train_loss_llm": 0.0, "grad_norm": 0.06454101949930191, "global_step": 20496, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03103557787835598, "train_loss_bc": 0.03103557787835598, "train_loss_llm": 0.0, "grad_norm": 0.02712184377014637, "global_step": 20497, "epoch": 569, "lr": 0.001} +{"train_loss": 0.029003430157899857, "train_loss_bc": 0.029003430157899857, "train_loss_llm": 0.0, "grad_norm": 0.04751015082001686, "global_step": 20498, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03217160701751709, "train_loss_bc": 0.03217160701751709, "train_loss_llm": 0.0, "grad_norm": 0.06057271733880043, "global_step": 20499, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03476683050394058, "train_loss_bc": 0.03476683050394058, "train_loss_llm": 0.0, "grad_norm": 0.07410846650600433, "global_step": 20500, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03059735894203186, "train_loss_bc": 0.03059735894203186, "train_loss_llm": 0.0, "grad_norm": 0.027980484068393707, "global_step": 20501, "epoch": 569, "lr": 0.001} +{"train_loss": 0.036474041640758514, "train_loss_bc": 0.036474041640758514, "train_loss_llm": 0.0, "grad_norm": 0.045491673052310944, "global_step": 20502, "epoch": 569, "lr": 0.001} +{"train_loss": 0.035221390426158905, "train_loss_bc": 0.035220734775066376, "train_loss_llm": 6.562471389770508e-05, "grad_norm": 0.063479945063591, "global_step": 20503, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03257361054420471, "train_loss_bc": 0.03257361054420471, "train_loss_llm": 0.0, "grad_norm": 0.06921670585870743, "global_step": 20504, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03481846675276756, "train_loss_bc": 0.03481846675276756, "train_loss_llm": 0.0, "grad_norm": 0.032789260149002075, "global_step": 20505, "epoch": 569, "lr": 0.001} +{"train_loss": 0.029802978038787842, "train_loss_bc": 0.029802978038787842, "train_loss_llm": 0.0, "grad_norm": 0.05090655758976936, "global_step": 20506, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03382653743028641, "train_loss_bc": 0.03382653743028641, "train_loss_llm": 0.0, "grad_norm": 0.06318862736225128, "global_step": 20507, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03328167647123337, "train_loss_bc": 0.03328167647123337, "train_loss_llm": 0.0, "grad_norm": 0.06659533828496933, "global_step": 20508, "epoch": 569, "lr": 0.001} +{"train_loss": 0.025517508387565613, "train_loss_bc": 0.025517508387565613, "train_loss_llm": 0.0, "grad_norm": 0.020471222698688507, "global_step": 20509, "epoch": 569, "lr": 0.001} +{"train_loss": 0.031217049807310104, "train_loss_bc": 0.031217049807310104, "train_loss_llm": 0.0, "grad_norm": 0.03386823087930679, "global_step": 20510, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03469165414571762, "train_loss_bc": 0.03469165414571762, "train_loss_llm": 0.0, "grad_norm": 0.047492098063230515, "global_step": 20511, "epoch": 569, "lr": 0.001} +{"train_loss": 0.024459021165966988, "train_loss_bc": 0.024459021165966988, "train_loss_llm": 0.0, "grad_norm": 0.05409013107419014, "global_step": 20512, "epoch": 569, "lr": 0.001} +{"train_loss": 0.034914858639240265, "train_loss_bc": 0.034914858639240265, "train_loss_llm": 0.0, "grad_norm": 0.024392293766140938, "global_step": 20513, "epoch": 569, "lr": 0.001} +{"train_loss": 0.033575087785720825, "train_loss_bc": 0.03356616944074631, "train_loss_llm": 0.0008918046951293945, "grad_norm": 0.038221001625061035, "global_step": 20514, "epoch": 569, "lr": 0.001} +{"train_loss": 0.038866784423589706, "train_loss_bc": 0.038866784423589706, "train_loss_llm": 0.0, "grad_norm": 0.06128263473510742, "global_step": 20515, "epoch": 569, "lr": 0.001} +{"train_loss": 0.03209691122174263, "train_loss_bc": 0.03209691122174263, "train_loss_llm": 0.0, "grad_norm": 0.07656178623437881, "global_step": 20516, "epoch": 569, "lr": 0.001} +{"train_loss": 0.04278162121772766, "train_loss_bc": 0.04278162121772766, "train_loss_llm": 0.0, "grad_norm": 0.03299534320831299, "global_step": 20517, "epoch": 569, "lr": 0.001} +{"train_loss": 0.02929438091814518, "train_loss_bc": 0.02926916629076004, "train_loss_llm": 0.002521425485610962, "grad_norm": 0.05135282874107361, "global_step": 20518, "epoch": 569, "lr": 0.001} +{"train_loss": 0.032948789911137685, "train_loss_bc": 0.03640185296535492, "train_loss_llm": 0.0, "grad_norm": 0.0758247971534729, "global_step": 20519, "epoch": 569, "lr": 0.001, "val_loss": 0.07993052899837494} +{"train_loss": 0.03400087356567383, "train_loss_bc": 0.03400087356567383, "train_loss_llm": 0.0, "grad_norm": 0.09580447524785995, "global_step": 20520, "epoch": 570, "lr": 0.001} +{"train_loss": 0.0323522686958313, "train_loss_bc": 0.0323522686958313, "train_loss_llm": 0.0, "grad_norm": 0.025679128244519234, "global_step": 20521, "epoch": 570, "lr": 0.001} +{"train_loss": 0.033423181623220444, "train_loss_bc": 0.033399634063243866, "train_loss_llm": 0.0023548901081085205, "grad_norm": 0.03927258402109146, "global_step": 20522, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03125422075390816, "train_loss_bc": 0.03125422075390816, "train_loss_llm": 0.0, "grad_norm": 0.05475971847772598, "global_step": 20523, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03172188624739647, "train_loss_bc": 0.03172007575631142, "train_loss_llm": 0.0001811981201171875, "grad_norm": 0.06987664103507996, "global_step": 20524, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03393834829330444, "train_loss_bc": 0.03393834829330444, "train_loss_llm": 0.0, "grad_norm": 0.02662632428109646, "global_step": 20525, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03558480367064476, "train_loss_bc": 0.03558480367064476, "train_loss_llm": 0.0, "grad_norm": 0.04571393132209778, "global_step": 20526, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03509976342320442, "train_loss_bc": 0.03509976342320442, "train_loss_llm": 0.0, "grad_norm": 0.05289899930357933, "global_step": 20527, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03242671117186546, "train_loss_bc": 0.03242671117186546, "train_loss_llm": 0.0, "grad_norm": 0.06863948702812195, "global_step": 20528, "epoch": 570, "lr": 0.001} +{"train_loss": 0.029472747817635536, "train_loss_bc": 0.029472747817635536, "train_loss_llm": 0.0, "grad_norm": 0.027313021942973137, "global_step": 20529, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03404390811920166, "train_loss_bc": 0.03404390811920166, "train_loss_llm": 0.0, "grad_norm": 0.041628770530223846, "global_step": 20530, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03138958662748337, "train_loss_bc": 0.03138958662748337, "train_loss_llm": 0.0, "grad_norm": 0.04910947382450104, "global_step": 20531, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03383515775203705, "train_loss_bc": 0.03383515775203705, "train_loss_llm": 0.0, "grad_norm": 0.056537363678216934, "global_step": 20532, "epoch": 570, "lr": 0.001} +{"train_loss": 0.026689395308494568, "train_loss_bc": 0.026689395308494568, "train_loss_llm": 0.0, "grad_norm": 0.026196299120783806, "global_step": 20533, "epoch": 570, "lr": 0.001} +{"train_loss": 0.02744279056787491, "train_loss_bc": 0.02744279056787491, "train_loss_llm": 0.0, "grad_norm": 0.03757791593670845, "global_step": 20534, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03474507853388786, "train_loss_bc": 0.03474507853388786, "train_loss_llm": 0.0, "grad_norm": 0.05503446236252785, "global_step": 20535, "epoch": 570, "lr": 0.001} +{"train_loss": 0.029306570068001747, "train_loss_bc": 0.029286785051226616, "train_loss_llm": 0.0019784271717071533, "grad_norm": 0.07201114296913147, "global_step": 20536, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03258829191327095, "train_loss_bc": 0.03258829191327095, "train_loss_llm": 0.0, "grad_norm": 0.026076046749949455, "global_step": 20537, "epoch": 570, "lr": 0.001} +{"train_loss": 0.028511518612504005, "train_loss_bc": 0.0284848902374506, "train_loss_llm": 0.0026629269123077393, "grad_norm": 0.04227425158023834, "global_step": 20538, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03278796374797821, "train_loss_bc": 0.03278796374797821, "train_loss_llm": 0.0, "grad_norm": 0.05860264226794243, "global_step": 20539, "epoch": 570, "lr": 0.001} +{"train_loss": 0.035989683121442795, "train_loss_bc": 0.035989683121442795, "train_loss_llm": 0.0, "grad_norm": 0.06341142952442169, "global_step": 20540, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03222121298313141, "train_loss_bc": 0.03222121298313141, "train_loss_llm": 0.0, "grad_norm": 0.026274345815181732, "global_step": 20541, "epoch": 570, "lr": 0.001} +{"train_loss": 0.034627702087163925, "train_loss_bc": 0.034627702087163925, "train_loss_llm": 0.0, "grad_norm": 0.03762900456786156, "global_step": 20542, "epoch": 570, "lr": 0.001} +{"train_loss": 0.0358365960419178, "train_loss_bc": 0.0358365960419178, "train_loss_llm": 0.0, "grad_norm": 0.052476901561021805, "global_step": 20543, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03485717996954918, "train_loss_bc": 0.03485717996954918, "train_loss_llm": 0.0, "grad_norm": 0.06339552253484726, "global_step": 20544, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03622552007436752, "train_loss_bc": 0.03622552007436752, "train_loss_llm": 0.0, "grad_norm": 0.02660500630736351, "global_step": 20545, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03159795701503754, "train_loss_bc": 0.03159795701503754, "train_loss_llm": 0.0, "grad_norm": 0.04007111117243767, "global_step": 20546, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03390883281826973, "train_loss_bc": 0.03390883281826973, "train_loss_llm": 0.0, "grad_norm": 0.05825299769639969, "global_step": 20547, "epoch": 570, "lr": 0.001} +{"train_loss": 0.028263019397854805, "train_loss_bc": 0.028263019397854805, "train_loss_llm": 0.0, "grad_norm": 0.06848978251218796, "global_step": 20548, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03530489653348923, "train_loss_bc": 0.03528347983956337, "train_loss_llm": 0.0021416842937469482, "grad_norm": 0.025425247848033905, "global_step": 20549, "epoch": 570, "lr": 0.001} +{"train_loss": 0.032263945788145065, "train_loss_bc": 0.032263945788145065, "train_loss_llm": 0.0, "grad_norm": 0.03816413879394531, "global_step": 20550, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03779544308781624, "train_loss_bc": 0.03779544308781624, "train_loss_llm": 0.0, "grad_norm": 0.05632966756820679, "global_step": 20551, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03317093104124069, "train_loss_bc": 0.03315138444304466, "train_loss_llm": 0.001954793930053711, "grad_norm": 0.07769909501075745, "global_step": 20552, "epoch": 570, "lr": 0.001} +{"train_loss": 0.03406985104084015, "train_loss_bc": 0.03406836837530136, "train_loss_llm": 0.00014832615852355957, "grad_norm": 0.02616424486041069, "global_step": 20553, "epoch": 570, "lr": 0.001} +{"train_loss": 0.030210550874471664, "train_loss_bc": 0.030205899849534035, "train_loss_llm": 0.00046518445014953613, "grad_norm": 0.0382811464369297, "global_step": 20554, "epoch": 570, "lr": 0.001} +{"train_loss": 0.032922140426105924, "train_loss_bc": 0.03823866695165634, "train_loss_llm": 0.0, "grad_norm": 0.05560522526502609, "global_step": 20555, "epoch": 570, "lr": 0.001, "train/cumulative_reward": 81.17624262825147, "train/mean_score": 21.92454801510317, "train/success_rate": 0.6, "test/cumulative_reward": 40.52117107417273, "test/mean_score": 9.687207978960394, "test/success_rate": 0.26, "val_loss": 0.07924192398786545, "train_action_mse_error": 0.03513290733098984} +{"train_loss": 0.02817595563828945, "train_loss_bc": 0.02817595563828945, "train_loss_llm": 0.0, "grad_norm": 0.06161406636238098, "global_step": 20556, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03939404711127281, "train_loss_bc": 0.0393824428319931, "train_loss_llm": 0.0011605024337768555, "grad_norm": 0.03813885152339935, "global_step": 20557, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03177395835518837, "train_loss_bc": 0.03177395835518837, "train_loss_llm": 0.0, "grad_norm": 0.05698856711387634, "global_step": 20558, "epoch": 571, "lr": 0.001} +{"train_loss": 0.027994932606816292, "train_loss_bc": 0.027994932606816292, "train_loss_llm": 0.0, "grad_norm": 0.07938025146722794, "global_step": 20559, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03672081604599953, "train_loss_bc": 0.03672081604599953, "train_loss_llm": 0.0, "grad_norm": 0.09715849161148071, "global_step": 20560, "epoch": 571, "lr": 0.001} +{"train_loss": 0.029469812288880348, "train_loss_bc": 0.02946922369301319, "train_loss_llm": 5.882978439331055e-05, "grad_norm": 0.025667089968919754, "global_step": 20561, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03828645497560501, "train_loss_bc": 0.03828645497560501, "train_loss_llm": 0.0, "grad_norm": 0.04193374514579773, "global_step": 20562, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03149309754371643, "train_loss_bc": 0.03149309754371643, "train_loss_llm": 0.0, "grad_norm": 0.05556778982281685, "global_step": 20563, "epoch": 571, "lr": 0.001} +{"train_loss": 0.0380757711827755, "train_loss_bc": 0.0380757711827755, "train_loss_llm": 0.0, "grad_norm": 0.07131736725568771, "global_step": 20564, "epoch": 571, "lr": 0.001} +{"train_loss": 0.033163439482450485, "train_loss_bc": 0.033163439482450485, "train_loss_llm": 0.0, "grad_norm": 0.028332913294434547, "global_step": 20565, "epoch": 571, "lr": 0.001} +{"train_loss": 0.0334196500480175, "train_loss_bc": 0.0334196500480175, "train_loss_llm": 0.0, "grad_norm": 0.043648477643728256, "global_step": 20566, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03469434380531311, "train_loss_bc": 0.03469434380531311, "train_loss_llm": 0.0, "grad_norm": 0.06081978976726532, "global_step": 20567, "epoch": 571, "lr": 0.001} +{"train_loss": 0.030340353026986122, "train_loss_bc": 0.030332770198583603, "train_loss_llm": 0.0007582902908325195, "grad_norm": 0.07135696709156036, "global_step": 20568, "epoch": 571, "lr": 0.001} +{"train_loss": 0.036406200379133224, "train_loss_bc": 0.03638935461640358, "train_loss_llm": 0.0016845464706420898, "grad_norm": 0.03038046695291996, "global_step": 20569, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03562987968325615, "train_loss_bc": 0.03562987968325615, "train_loss_llm": 0.0, "grad_norm": 0.04357446730136871, "global_step": 20570, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03311542421579361, "train_loss_bc": 0.03311542421579361, "train_loss_llm": 0.0, "grad_norm": 0.06994457542896271, "global_step": 20571, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03285392373800278, "train_loss_bc": 0.03285392373800278, "train_loss_llm": 0.0, "grad_norm": 0.10030391812324524, "global_step": 20572, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03165845572948456, "train_loss_bc": 0.031648240983486176, "train_loss_llm": 0.0010213255882263184, "grad_norm": 0.0246463343501091, "global_step": 20573, "epoch": 571, "lr": 0.001} +{"train_loss": 0.028575299307703972, "train_loss_bc": 0.028575299307703972, "train_loss_llm": 0.0, "grad_norm": 0.04304211959242821, "global_step": 20574, "epoch": 571, "lr": 0.001} +{"train_loss": 0.029807470738887787, "train_loss_bc": 0.02980026789009571, "train_loss_llm": 0.0007202029228210449, "grad_norm": 0.05481995642185211, "global_step": 20575, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03242715448141098, "train_loss_bc": 0.03242660313844681, "train_loss_llm": 5.501508712768555e-05, "grad_norm": 0.06995254755020142, "global_step": 20576, "epoch": 571, "lr": 0.001} +{"train_loss": 0.027156921103596687, "train_loss_bc": 0.027147579938173294, "train_loss_llm": 0.0009340941905975342, "grad_norm": 0.026584863662719727, "global_step": 20577, "epoch": 571, "lr": 0.001} +{"train_loss": 0.028635447844862938, "train_loss_bc": 0.028635447844862938, "train_loss_llm": 0.0, "grad_norm": 0.03563311696052551, "global_step": 20578, "epoch": 571, "lr": 0.001} +{"train_loss": 0.0398714542388916, "train_loss_bc": 0.03986411914229393, "train_loss_llm": 0.000733494758605957, "grad_norm": 0.057358890771865845, "global_step": 20579, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03967425227165222, "train_loss_bc": 0.03967425227165222, "train_loss_llm": 0.0, "grad_norm": 0.06588788330554962, "global_step": 20580, "epoch": 571, "lr": 0.001} +{"train_loss": 0.030947605147957802, "train_loss_bc": 0.030947605147957802, "train_loss_llm": 0.0, "grad_norm": 0.02756376378238201, "global_step": 20581, "epoch": 571, "lr": 0.001} +{"train_loss": 0.030088843777775764, "train_loss_bc": 0.030088843777775764, "train_loss_llm": 0.0, "grad_norm": 0.04649317264556885, "global_step": 20582, "epoch": 571, "lr": 0.001} +{"train_loss": 0.031156891956925392, "train_loss_bc": 0.031156891956925392, "train_loss_llm": 0.0, "grad_norm": 0.06032193824648857, "global_step": 20583, "epoch": 571, "lr": 0.001} +{"train_loss": 0.029911678284406662, "train_loss_bc": 0.029911678284406662, "train_loss_llm": 0.0, "grad_norm": 0.07542618364095688, "global_step": 20584, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03531666100025177, "train_loss_bc": 0.03531666100025177, "train_loss_llm": 0.0, "grad_norm": 0.03969673067331314, "global_step": 20585, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03258943185210228, "train_loss_bc": 0.03258943185210228, "train_loss_llm": 0.0, "grad_norm": 0.056123264133930206, "global_step": 20586, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03277164697647095, "train_loss_bc": 0.03277164697647095, "train_loss_llm": 0.0, "grad_norm": 0.07903024554252625, "global_step": 20587, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03749122843146324, "train_loss_bc": 0.03749122843146324, "train_loss_llm": 0.0, "grad_norm": 0.09603803604841232, "global_step": 20588, "epoch": 571, "lr": 0.001} +{"train_loss": 0.035103604197502136, "train_loss_bc": 0.035103604197502136, "train_loss_llm": 0.0, "grad_norm": 0.02875102311372757, "global_step": 20589, "epoch": 571, "lr": 0.001} +{"train_loss": 0.031481850892305374, "train_loss_bc": 0.031481850892305374, "train_loss_llm": 0.0, "grad_norm": 0.04774977266788483, "global_step": 20590, "epoch": 571, "lr": 0.001} +{"train_loss": 0.03291791925827662, "train_loss_bc": 0.02937113493680954, "train_loss_llm": 0.0, "grad_norm": 0.07393437623977661, "global_step": 20591, "epoch": 571, "lr": 0.001, "val_loss": 0.08029814064502716} +{"train_loss": 0.03368762508034706, "train_loss_bc": 0.03368762508034706, "train_loss_llm": 0.0, "grad_norm": 0.09781429171562195, "global_step": 20592, "epoch": 572, "lr": 0.001} +{"train_loss": 0.029976937919855118, "train_loss_bc": 0.029976937919855118, "train_loss_llm": 0.0, "grad_norm": 0.029497040435671806, "global_step": 20593, "epoch": 572, "lr": 0.001} +{"train_loss": 0.026596106588840485, "train_loss_bc": 0.026594366878271103, "train_loss_llm": 0.00017395615577697754, "grad_norm": 0.04503704607486725, "global_step": 20594, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03252263367176056, "train_loss_bc": 0.03252263367176056, "train_loss_llm": 0.0, "grad_norm": 0.05579061806201935, "global_step": 20595, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03329868242144585, "train_loss_bc": 0.033291030675172806, "train_loss_llm": 0.0007653534412384033, "grad_norm": 0.0762404352426529, "global_step": 20596, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03689856827259064, "train_loss_bc": 0.03689856827259064, "train_loss_llm": 0.0, "grad_norm": 0.025546828284859657, "global_step": 20597, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03137042373418808, "train_loss_bc": 0.03137042373418808, "train_loss_llm": 0.0, "grad_norm": 0.04084793105721474, "global_step": 20598, "epoch": 572, "lr": 0.001} +{"train_loss": 0.02788585238158703, "train_loss_bc": 0.02787274494767189, "train_loss_llm": 0.0013107061386108398, "grad_norm": 0.049457427114248276, "global_step": 20599, "epoch": 572, "lr": 0.001} +{"train_loss": 0.036238089203834534, "train_loss_bc": 0.036233969032764435, "train_loss_llm": 0.0004120767116546631, "grad_norm": 0.061538487672805786, "global_step": 20600, "epoch": 572, "lr": 0.001} +{"train_loss": 0.02830694615840912, "train_loss_bc": 0.02830694615840912, "train_loss_llm": 0.0, "grad_norm": 0.02582547999918461, "global_step": 20601, "epoch": 572, "lr": 0.001} +{"train_loss": 0.030537046492099762, "train_loss_bc": 0.030537046492099762, "train_loss_llm": 0.0, "grad_norm": 0.040523380041122437, "global_step": 20602, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03688536584377289, "train_loss_bc": 0.03688536584377289, "train_loss_llm": 0.0, "grad_norm": 0.06309258937835693, "global_step": 20603, "epoch": 572, "lr": 0.001} +{"train_loss": 0.029441095888614655, "train_loss_bc": 0.029441095888614655, "train_loss_llm": 0.0, "grad_norm": 0.06398770958185196, "global_step": 20604, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03756420314311981, "train_loss_bc": 0.03756420314311981, "train_loss_llm": 0.0, "grad_norm": 0.030508989468216896, "global_step": 20605, "epoch": 572, "lr": 0.001} +{"train_loss": 0.032250069081783295, "train_loss_bc": 0.032245852053165436, "train_loss_llm": 0.00042188167572021484, "grad_norm": 0.05928850173950195, "global_step": 20606, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03712773695588112, "train_loss_bc": 0.03712773695588112, "train_loss_llm": 0.0, "grad_norm": 0.07837962359189987, "global_step": 20607, "epoch": 572, "lr": 0.001} +{"train_loss": 0.030080348253250122, "train_loss_bc": 0.030080348253250122, "train_loss_llm": 0.0, "grad_norm": 0.11179688572883606, "global_step": 20608, "epoch": 572, "lr": 0.001} +{"train_loss": 0.037068214267492294, "train_loss_bc": 0.037068214267492294, "train_loss_llm": 0.0, "grad_norm": 0.039949528872966766, "global_step": 20609, "epoch": 572, "lr": 0.001} +{"train_loss": 0.0276761706918478, "train_loss_bc": 0.0276761706918478, "train_loss_llm": 0.0, "grad_norm": 0.05460299924015999, "global_step": 20610, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03797541931271553, "train_loss_bc": 0.03797541931271553, "train_loss_llm": 0.0, "grad_norm": 0.09055279940366745, "global_step": 20611, "epoch": 572, "lr": 0.001} +{"train_loss": 0.039615973830223083, "train_loss_bc": 0.039615973830223083, "train_loss_llm": 0.0, "grad_norm": 0.12232357263565063, "global_step": 20612, "epoch": 572, "lr": 0.001} +{"train_loss": 0.031648874282836914, "train_loss_bc": 0.031648874282836914, "train_loss_llm": 0.0, "grad_norm": 0.027738085016608238, "global_step": 20613, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03416295722126961, "train_loss_bc": 0.03416223078966141, "train_loss_llm": 7.2479248046875e-05, "grad_norm": 0.04969656094908714, "global_step": 20614, "epoch": 572, "lr": 0.001} +{"train_loss": 0.035167548805475235, "train_loss_bc": 0.035167548805475235, "train_loss_llm": 0.0, "grad_norm": 0.07234589010477066, "global_step": 20615, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03493209183216095, "train_loss_bc": 0.03493209183216095, "train_loss_llm": 0.0, "grad_norm": 0.09651210904121399, "global_step": 20616, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03306123986840248, "train_loss_bc": 0.03305879980325699, "train_loss_llm": 0.0002441108226776123, "grad_norm": 0.027326181530952454, "global_step": 20617, "epoch": 572, "lr": 0.001} +{"train_loss": 0.029857657849788666, "train_loss_bc": 0.0298355370759964, "train_loss_llm": 0.0022121071815490723, "grad_norm": 0.04005034640431404, "global_step": 20618, "epoch": 572, "lr": 0.001} +{"train_loss": 0.033100683242082596, "train_loss_bc": 0.033100683242082596, "train_loss_llm": 0.0, "grad_norm": 0.06159653142094612, "global_step": 20619, "epoch": 572, "lr": 0.001} +{"train_loss": 0.033844947814941406, "train_loss_bc": 0.033844947814941406, "train_loss_llm": 0.0, "grad_norm": 0.07031792402267456, "global_step": 20620, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03228205442428589, "train_loss_bc": 0.032266855239868164, "train_loss_llm": 0.0015200972557067871, "grad_norm": 0.03347137197852135, "global_step": 20621, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03502684459090233, "train_loss_bc": 0.03502684459090233, "train_loss_llm": 0.0, "grad_norm": 0.06236915662884712, "global_step": 20622, "epoch": 572, "lr": 0.001} +{"train_loss": 0.02619406022131443, "train_loss_bc": 0.026180116459727287, "train_loss_llm": 0.0013943016529083252, "grad_norm": 0.08531522005796432, "global_step": 20623, "epoch": 572, "lr": 0.001} +{"train_loss": 0.037530891597270966, "train_loss_bc": 0.037530891597270966, "train_loss_llm": 0.0, "grad_norm": 0.11393925547599792, "global_step": 20624, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03685452416539192, "train_loss_bc": 0.03685452416539192, "train_loss_llm": 0.0, "grad_norm": 0.029863663017749786, "global_step": 20625, "epoch": 572, "lr": 0.001} +{"train_loss": 0.03587368503212929, "train_loss_bc": 0.03587368503212929, "train_loss_llm": 0.0, "grad_norm": 0.05400718376040459, "global_step": 20626, "epoch": 572, "lr": 0.001} +{"train_loss": 0.033044662016133465, "train_loss_bc": 0.027066262438893318, "train_loss_llm": 0.0, "grad_norm": 0.07468829303979874, "global_step": 20627, "epoch": 572, "lr": 0.001, "val_loss": 0.07867804914712906} +{"train_loss": 0.03573151305317879, "train_loss_bc": 0.03573151305317879, "train_loss_llm": 0.0, "grad_norm": 0.09128564596176147, "global_step": 20628, "epoch": 573, "lr": 0.001} +{"train_loss": 0.0333150178194046, "train_loss_bc": 0.0333150178194046, "train_loss_llm": 0.0, "grad_norm": 0.03095347061753273, "global_step": 20629, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03538450598716736, "train_loss_bc": 0.03538450598716736, "train_loss_llm": 0.0, "grad_norm": 0.03961091488599777, "global_step": 20630, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03043838031589985, "train_loss_bc": 0.03043838031589985, "train_loss_llm": 0.0, "grad_norm": 0.06287732720375061, "global_step": 20631, "epoch": 573, "lr": 0.001} +{"train_loss": 0.036858607083559036, "train_loss_bc": 0.036858607083559036, "train_loss_llm": 0.0, "grad_norm": 0.07676329463720322, "global_step": 20632, "epoch": 573, "lr": 0.001} +{"train_loss": 0.029978010803461075, "train_loss_bc": 0.029978010803461075, "train_loss_llm": 0.0, "grad_norm": 0.02744702808558941, "global_step": 20633, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03744499385356903, "train_loss_bc": 0.03744499385356903, "train_loss_llm": 0.0, "grad_norm": 0.04567930847406387, "global_step": 20634, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03434440866112709, "train_loss_bc": 0.03434440866112709, "train_loss_llm": 0.0, "grad_norm": 0.057801272720098495, "global_step": 20635, "epoch": 573, "lr": 0.001} +{"train_loss": 0.028814364224672318, "train_loss_bc": 0.028814364224672318, "train_loss_llm": 0.0, "grad_norm": 0.06673681735992432, "global_step": 20636, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03849196806550026, "train_loss_bc": 0.03847906365990639, "train_loss_llm": 0.0012904703617095947, "grad_norm": 0.031000304967164993, "global_step": 20637, "epoch": 573, "lr": 0.001} +{"train_loss": 0.028938665986061096, "train_loss_bc": 0.028938665986061096, "train_loss_llm": 0.0, "grad_norm": 0.04596078395843506, "global_step": 20638, "epoch": 573, "lr": 0.001} +{"train_loss": 0.030546529218554497, "train_loss_bc": 0.03053407184779644, "train_loss_llm": 0.0012456774711608887, "grad_norm": 0.05598612502217293, "global_step": 20639, "epoch": 573, "lr": 0.001} +{"train_loss": 0.037558309733867645, "train_loss_bc": 0.037558309733867645, "train_loss_llm": 0.0, "grad_norm": 0.07041679322719574, "global_step": 20640, "epoch": 573, "lr": 0.001} +{"train_loss": 0.031077614054083824, "train_loss_bc": 0.031075621023774147, "train_loss_llm": 0.00019922852516174316, "grad_norm": 0.028631776571273804, "global_step": 20641, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03209192305803299, "train_loss_bc": 0.03209192305803299, "train_loss_llm": 0.0, "grad_norm": 0.045927874743938446, "global_step": 20642, "epoch": 573, "lr": 0.001} +{"train_loss": 0.029705597087740898, "train_loss_bc": 0.029697071760892868, "train_loss_llm": 0.0008526146411895752, "grad_norm": 0.06067361682653427, "global_step": 20643, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03210880979895592, "train_loss_bc": 0.03210880979895592, "train_loss_llm": 0.0, "grad_norm": 0.08158392459154129, "global_step": 20644, "epoch": 573, "lr": 0.001} +{"train_loss": 0.029875366017222404, "train_loss_bc": 0.029856711626052856, "train_loss_llm": 0.0018654465675354004, "grad_norm": 0.03183747082948685, "global_step": 20645, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03166848421096802, "train_loss_bc": 0.03166848421096802, "train_loss_llm": 0.0, "grad_norm": 0.05577366054058075, "global_step": 20646, "epoch": 573, "lr": 0.001} +{"train_loss": 0.027080362662672997, "train_loss_bc": 0.027080362662672997, "train_loss_llm": 0.0, "grad_norm": 0.06872835010290146, "global_step": 20647, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03547985106706619, "train_loss_bc": 0.03547985106706619, "train_loss_llm": 0.0, "grad_norm": 0.08688937872648239, "global_step": 20648, "epoch": 573, "lr": 0.001} +{"train_loss": 0.02950993925333023, "train_loss_bc": 0.02950993925333023, "train_loss_llm": 0.0, "grad_norm": 0.025438111275434494, "global_step": 20649, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03925586864352226, "train_loss_bc": 0.03925586864352226, "train_loss_llm": 0.0, "grad_norm": 0.04686402529478073, "global_step": 20650, "epoch": 573, "lr": 0.001} +{"train_loss": 0.033510155975818634, "train_loss_bc": 0.033492282032966614, "train_loss_llm": 0.0017872750759124756, "grad_norm": 0.061105310916900635, "global_step": 20651, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03711037337779999, "train_loss_bc": 0.03710580989718437, "train_loss_llm": 0.0004564821720123291, "grad_norm": 0.08881430327892303, "global_step": 20652, "epoch": 573, "lr": 0.001} +{"train_loss": 0.028608011081814766, "train_loss_bc": 0.028608011081814766, "train_loss_llm": 0.0, "grad_norm": 0.025490324944257736, "global_step": 20653, "epoch": 573, "lr": 0.001} +{"train_loss": 0.02924957312643528, "train_loss_bc": 0.029221726581454277, "train_loss_llm": 0.0027846097946166992, "grad_norm": 0.04246263951063156, "global_step": 20654, "epoch": 573, "lr": 0.001} +{"train_loss": 0.033740848302841187, "train_loss_bc": 0.0337301567196846, "train_loss_llm": 0.001069188117980957, "grad_norm": 0.0665404200553894, "global_step": 20655, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03024076111614704, "train_loss_bc": 0.030228249728679657, "train_loss_llm": 0.0012511909008026123, "grad_norm": 0.0851089358329773, "global_step": 20656, "epoch": 573, "lr": 0.001} +{"train_loss": 0.034041810780763626, "train_loss_bc": 0.034041810780763626, "train_loss_llm": 0.0, "grad_norm": 0.02403857745230198, "global_step": 20657, "epoch": 573, "lr": 0.001} +{"train_loss": 0.028874913230538368, "train_loss_bc": 0.028867082670331, "train_loss_llm": 0.0007831454277038574, "grad_norm": 0.0386478528380394, "global_step": 20658, "epoch": 573, "lr": 0.001} +{"train_loss": 0.037072714418172836, "train_loss_bc": 0.03706168383359909, "train_loss_llm": 0.0011028945446014404, "grad_norm": 0.051237788051366806, "global_step": 20659, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03434623405337334, "train_loss_bc": 0.03434623405337334, "train_loss_llm": 0.0, "grad_norm": 0.06354266405105591, "global_step": 20660, "epoch": 573, "lr": 0.001} +{"train_loss": 0.03681756556034088, "train_loss_bc": 0.03681756556034088, "train_loss_llm": 0.0, "grad_norm": 0.027690205723047256, "global_step": 20661, "epoch": 573, "lr": 0.001} +{"train_loss": 0.030958889052271843, "train_loss_bc": 0.030958889052271843, "train_loss_llm": 0.0, "grad_norm": 0.04140044376254082, "global_step": 20662, "epoch": 573, "lr": 0.001} +{"train_loss": 0.0330358595173392, "train_loss_bc": 0.039020001888275146, "train_loss_llm": 0.0, "grad_norm": 0.07647489756345749, "global_step": 20663, "epoch": 573, "lr": 0.001, "val_loss": 0.08070388436317444} +{"train_loss": 0.031193191185593605, "train_loss_bc": 0.031193191185593605, "train_loss_llm": 0.0, "grad_norm": 0.09689629822969437, "global_step": 20664, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03626777604222298, "train_loss_bc": 0.03626777604222298, "train_loss_llm": 0.0, "grad_norm": 0.04037163779139519, "global_step": 20665, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03369126096367836, "train_loss_bc": 0.03369126096367836, "train_loss_llm": 0.0, "grad_norm": 0.07063017785549164, "global_step": 20666, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03283441439270973, "train_loss_bc": 0.032828766852617264, "train_loss_llm": 0.0005646646022796631, "grad_norm": 0.09044261276721954, "global_step": 20667, "epoch": 574, "lr": 0.001} +{"train_loss": 0.036536067724227905, "train_loss_bc": 0.036536067724227905, "train_loss_llm": 0.0, "grad_norm": 0.11120841652154922, "global_step": 20668, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03446441516280174, "train_loss_bc": 0.03446441516280174, "train_loss_llm": 0.0, "grad_norm": 0.03497280925512314, "global_step": 20669, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03591226041316986, "train_loss_bc": 0.03591226041316986, "train_loss_llm": 0.0, "grad_norm": 0.049509067088365555, "global_step": 20670, "epoch": 574, "lr": 0.001} +{"train_loss": 0.031675174832344055, "train_loss_bc": 0.031675174832344055, "train_loss_llm": 0.0, "grad_norm": 0.06557408720254898, "global_step": 20671, "epoch": 574, "lr": 0.001} +{"train_loss": 0.028706209734082222, "train_loss_bc": 0.028706209734082222, "train_loss_llm": 0.0, "grad_norm": 0.0889555960893631, "global_step": 20672, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03274475410580635, "train_loss_bc": 0.03273536637425423, "train_loss_llm": 0.0009386837482452393, "grad_norm": 0.027293993160128593, "global_step": 20673, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03820253163576126, "train_loss_bc": 0.03820253163576126, "train_loss_llm": 0.0, "grad_norm": 0.04166227579116821, "global_step": 20674, "epoch": 574, "lr": 0.001} +{"train_loss": 0.031114395707845688, "train_loss_bc": 0.031114395707845688, "train_loss_llm": 0.0, "grad_norm": 0.05131400004029274, "global_step": 20675, "epoch": 574, "lr": 0.001} +{"train_loss": 0.028668466955423355, "train_loss_bc": 0.028668466955423355, "train_loss_llm": 0.0, "grad_norm": 0.061695314943790436, "global_step": 20676, "epoch": 574, "lr": 0.001} +{"train_loss": 0.034058067947626114, "train_loss_bc": 0.034058067947626114, "train_loss_llm": 0.0, "grad_norm": 0.026625914499163628, "global_step": 20677, "epoch": 574, "lr": 0.001} +{"train_loss": 0.027295976877212524, "train_loss_bc": 0.027295976877212524, "train_loss_llm": 0.0, "grad_norm": 0.0398174524307251, "global_step": 20678, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03505704924464226, "train_loss_bc": 0.03502924367785454, "train_loss_llm": 0.002780526876449585, "grad_norm": 0.05040972679853439, "global_step": 20679, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03695165365934372, "train_loss_bc": 0.03695165365934372, "train_loss_llm": 0.0, "grad_norm": 0.06678157299757004, "global_step": 20680, "epoch": 574, "lr": 0.001} +{"train_loss": 0.029557745903730392, "train_loss_bc": 0.029557745903730392, "train_loss_llm": 0.0, "grad_norm": 0.037272825837135315, "global_step": 20681, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03480512648820877, "train_loss_bc": 0.034801680594682693, "train_loss_llm": 0.0003446638584136963, "grad_norm": 0.06303826719522476, "global_step": 20682, "epoch": 574, "lr": 0.001} +{"train_loss": 0.032158832997083664, "train_loss_bc": 0.032158832997083664, "train_loss_llm": 0.0, "grad_norm": 0.07411415874958038, "global_step": 20683, "epoch": 574, "lr": 0.001} +{"train_loss": 0.033341821283102036, "train_loss_bc": 0.033341821283102036, "train_loss_llm": 0.0, "grad_norm": 0.09180570393800735, "global_step": 20684, "epoch": 574, "lr": 0.001} +{"train_loss": 0.04131123423576355, "train_loss_bc": 0.04131123423576355, "train_loss_llm": 0.0, "grad_norm": 0.03943166136741638, "global_step": 20685, "epoch": 574, "lr": 0.001} +{"train_loss": 0.030056554824113846, "train_loss_bc": 0.030046019703149796, "train_loss_llm": 0.001053452491760254, "grad_norm": 0.06897483021020889, "global_step": 20686, "epoch": 574, "lr": 0.001} +{"train_loss": 0.031089402735233307, "train_loss_bc": 0.031075991690158844, "train_loss_llm": 0.0013410747051239014, "grad_norm": 0.0975712239742279, "global_step": 20687, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03526360169053078, "train_loss_bc": 0.035253047943115234, "train_loss_llm": 0.0010554194450378418, "grad_norm": 0.1190679594874382, "global_step": 20688, "epoch": 574, "lr": 0.001} +{"train_loss": 0.036582719534635544, "train_loss_bc": 0.03654033690690994, "train_loss_llm": 0.00423818826675415, "grad_norm": 0.03326289355754852, "global_step": 20689, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03690454363822937, "train_loss_bc": 0.03690454363822937, "train_loss_llm": 0.0, "grad_norm": 0.05951355770230293, "global_step": 20690, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03211401030421257, "train_loss_bc": 0.03209095820784569, "train_loss_llm": 0.002305328845977783, "grad_norm": 0.08046039938926697, "global_step": 20691, "epoch": 574, "lr": 0.001} +{"train_loss": 0.029789691790938377, "train_loss_bc": 0.029789691790938377, "train_loss_llm": 0.0, "grad_norm": 0.09035571664571762, "global_step": 20692, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03308909013867378, "train_loss_bc": 0.03308909013867378, "train_loss_llm": 0.0, "grad_norm": 0.028409617021679878, "global_step": 20693, "epoch": 574, "lr": 0.001} +{"train_loss": 0.02955436520278454, "train_loss_bc": 0.02955436520278454, "train_loss_llm": 0.0, "grad_norm": 0.05104559287428856, "global_step": 20694, "epoch": 574, "lr": 0.001} +{"train_loss": 0.032637160271406174, "train_loss_bc": 0.032637160271406174, "train_loss_llm": 0.0, "grad_norm": 0.05778571963310242, "global_step": 20695, "epoch": 574, "lr": 0.001} +{"train_loss": 0.027316579595208168, "train_loss_bc": 0.027316579595208168, "train_loss_llm": 0.0, "grad_norm": 0.06877786666154861, "global_step": 20696, "epoch": 574, "lr": 0.001} +{"train_loss": 0.031668514013290405, "train_loss_bc": 0.031668514013290405, "train_loss_llm": 0.0, "grad_norm": 0.0324205681681633, "global_step": 20697, "epoch": 574, "lr": 0.001} +{"train_loss": 0.03341376408934593, "train_loss_bc": 0.03341376408934593, "train_loss_llm": 0.0, "grad_norm": 0.04163791984319687, "global_step": 20698, "epoch": 574, "lr": 0.001} +{"train_loss": 0.033132384065538645, "train_loss_bc": 0.03672604635357857, "train_loss_llm": 0.001135319471359253, "grad_norm": 0.06160898879170418, "global_step": 20699, "epoch": 574, "lr": 0.001, "val_loss": 0.07946601510047913} +{"train_loss": 0.02827460691332817, "train_loss_bc": 0.02826501987874508, "train_loss_llm": 0.0009587705135345459, "grad_norm": 0.0698404535651207, "global_step": 20700, "epoch": 575, "lr": 0.001} +{"train_loss": 0.029335767030715942, "train_loss_bc": 0.029335767030715942, "train_loss_llm": 0.0, "grad_norm": 0.029949188232421875, "global_step": 20701, "epoch": 575, "lr": 0.001} +{"train_loss": 0.030900971964001656, "train_loss_bc": 0.030900971964001656, "train_loss_llm": 0.0, "grad_norm": 0.04720840975642204, "global_step": 20702, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03502678498625755, "train_loss_bc": 0.03502678498625755, "train_loss_llm": 0.0, "grad_norm": 0.06954977661371231, "global_step": 20703, "epoch": 575, "lr": 0.001} +{"train_loss": 0.029908781871199608, "train_loss_bc": 0.029908781871199608, "train_loss_llm": 0.0, "grad_norm": 0.08525259792804718, "global_step": 20704, "epoch": 575, "lr": 0.001} +{"train_loss": 0.027283988893032074, "train_loss_bc": 0.027275696396827698, "train_loss_llm": 0.000829160213470459, "grad_norm": 0.037413205951452255, "global_step": 20705, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03137081861495972, "train_loss_bc": 0.03137081861495972, "train_loss_llm": 0.0, "grad_norm": 0.06501442939043045, "global_step": 20706, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03353746980428696, "train_loss_bc": 0.03353746980428696, "train_loss_llm": 0.0, "grad_norm": 0.10015121847391129, "global_step": 20707, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03630419075489044, "train_loss_bc": 0.03630419075489044, "train_loss_llm": 0.0, "grad_norm": 0.1387018859386444, "global_step": 20708, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03853100910782814, "train_loss_bc": 0.03853100910782814, "train_loss_llm": 0.0, "grad_norm": 0.037913478910923004, "global_step": 20709, "epoch": 575, "lr": 0.001} +{"train_loss": 0.028725091367959976, "train_loss_bc": 0.028725091367959976, "train_loss_llm": 0.0, "grad_norm": 0.06476461887359619, "global_step": 20710, "epoch": 575, "lr": 0.001} +{"train_loss": 0.028628230094909668, "train_loss_bc": 0.028628230094909668, "train_loss_llm": 0.0, "grad_norm": 0.08603077381849289, "global_step": 20711, "epoch": 575, "lr": 0.001} +{"train_loss": 0.035385340452194214, "train_loss_bc": 0.035385340452194214, "train_loss_llm": 0.0, "grad_norm": 0.12228026986122131, "global_step": 20712, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03804628551006317, "train_loss_bc": 0.03804303705692291, "train_loss_llm": 0.0003248751163482666, "grad_norm": 0.02885616384446621, "global_step": 20713, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03996896371245384, "train_loss_bc": 0.03996386006474495, "train_loss_llm": 0.0005102157592773438, "grad_norm": 0.04800419136881828, "global_step": 20714, "epoch": 575, "lr": 0.001} +{"train_loss": 0.033556096255779266, "train_loss_bc": 0.033556096255779266, "train_loss_llm": 0.0, "grad_norm": 0.06441701948642731, "global_step": 20715, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03778096288442612, "train_loss_bc": 0.03778096288442612, "train_loss_llm": 0.0, "grad_norm": 0.08098363876342773, "global_step": 20716, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03398909047245979, "train_loss_bc": 0.03398909047245979, "train_loss_llm": 0.0, "grad_norm": 0.030293142423033714, "global_step": 20717, "epoch": 575, "lr": 0.001} +{"train_loss": 0.033393122255802155, "train_loss_bc": 0.033393122255802155, "train_loss_llm": 0.0, "grad_norm": 0.04079025611281395, "global_step": 20718, "epoch": 575, "lr": 0.001} +{"train_loss": 0.030840681865811348, "train_loss_bc": 0.030840681865811348, "train_loss_llm": 0.0, "grad_norm": 0.05000022426247597, "global_step": 20719, "epoch": 575, "lr": 0.001} +{"train_loss": 0.02770211361348629, "train_loss_bc": 0.02770211361348629, "train_loss_llm": 0.0, "grad_norm": 0.05758003517985344, "global_step": 20720, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03528475761413574, "train_loss_bc": 0.03528475761413574, "train_loss_llm": 0.0, "grad_norm": 0.03451177850365639, "global_step": 20721, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03185689076781273, "train_loss_bc": 0.03184531256556511, "train_loss_llm": 0.0011578202247619629, "grad_norm": 0.06219610571861267, "global_step": 20722, "epoch": 575, "lr": 0.001} +{"train_loss": 0.033489689230918884, "train_loss_bc": 0.033489689230918884, "train_loss_llm": 0.0, "grad_norm": 0.08006688952445984, "global_step": 20723, "epoch": 575, "lr": 0.001} +{"train_loss": 0.04211067408323288, "train_loss_bc": 0.04211067408323288, "train_loss_llm": 0.0, "grad_norm": 0.10216225683689117, "global_step": 20724, "epoch": 575, "lr": 0.001} +{"train_loss": 0.025581829249858856, "train_loss_bc": 0.02557430975139141, "train_loss_llm": 0.0007520318031311035, "grad_norm": 0.03536365181207657, "global_step": 20725, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03770162910223007, "train_loss_bc": 0.037691544741392136, "train_loss_llm": 0.0010084807872772217, "grad_norm": 0.0651533380150795, "global_step": 20726, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03759627044200897, "train_loss_bc": 0.03759627044200897, "train_loss_llm": 0.0, "grad_norm": 0.10260415822267532, "global_step": 20727, "epoch": 575, "lr": 0.001} +{"train_loss": 0.039780355989933014, "train_loss_bc": 0.039780355989933014, "train_loss_llm": 0.0, "grad_norm": 0.13405339419841766, "global_step": 20728, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03830746188759804, "train_loss_bc": 0.03830746188759804, "train_loss_llm": 0.0, "grad_norm": 0.028025178238749504, "global_step": 20729, "epoch": 575, "lr": 0.001} +{"train_loss": 0.030303651466965675, "train_loss_bc": 0.030303651466965675, "train_loss_llm": 0.0, "grad_norm": 0.04506118595600128, "global_step": 20730, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03421717509627342, "train_loss_bc": 0.03421292454004288, "train_loss_llm": 0.00042515993118286133, "grad_norm": 0.05957983806729317, "global_step": 20731, "epoch": 575, "lr": 0.001} +{"train_loss": 0.032217059284448624, "train_loss_bc": 0.03221191465854645, "train_loss_llm": 0.0005143880844116211, "grad_norm": 0.07676184177398682, "global_step": 20732, "epoch": 575, "lr": 0.001} +{"train_loss": 0.030343487858772278, "train_loss_bc": 0.030324123799800873, "train_loss_llm": 0.0019363164901733398, "grad_norm": 0.03551189973950386, "global_step": 20733, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03097500279545784, "train_loss_bc": 0.03097500279545784, "train_loss_llm": 0.0, "grad_norm": 0.06037425622344017, "global_step": 20734, "epoch": 575, "lr": 0.001} +{"train_loss": 0.03335347238721119, "train_loss_bc": 0.03244614973664284, "train_loss_llm": 0.002255260944366455, "grad_norm": 0.08088699728250504, "global_step": 20735, "epoch": 575, "lr": 0.001, "train/cumulative_reward": 57.544569219735614, "train/mean_score": 17.817131416364628, "train/success_rate": 0.5, "test/cumulative_reward": 44.57994919599255, "test/mean_score": 7.827387817592109, "test/success_rate": 0.18, "val_loss": 0.08117058873176575, "train_action_mse_error": 0.036171846091747284} +{"train_loss": 0.030739350244402885, "train_loss_bc": 0.030728314071893692, "train_loss_llm": 0.0011036992073059082, "grad_norm": 0.10842946916818619, "global_step": 20736, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03636033087968826, "train_loss_bc": 0.03636033087968826, "train_loss_llm": 0.0, "grad_norm": 0.028150619938969612, "global_step": 20737, "epoch": 576, "lr": 0.001} +{"train_loss": 0.029544267803430557, "train_loss_bc": 0.02952336147427559, "train_loss_llm": 0.0020906031131744385, "grad_norm": 0.04109526425600052, "global_step": 20738, "epoch": 576, "lr": 0.001} +{"train_loss": 0.027756480500102043, "train_loss_bc": 0.027726227417588234, "train_loss_llm": 0.003025233745574951, "grad_norm": 0.05284019559621811, "global_step": 20739, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03174055740237236, "train_loss_bc": 0.03174055740237236, "train_loss_llm": 0.0, "grad_norm": 0.0649736300110817, "global_step": 20740, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03186710551381111, "train_loss_bc": 0.03186710551381111, "train_loss_llm": 0.0, "grad_norm": 0.03113834746181965, "global_step": 20741, "epoch": 576, "lr": 0.001} +{"train_loss": 0.026783665642142296, "train_loss_bc": 0.026778623461723328, "train_loss_llm": 0.0005041360855102539, "grad_norm": 0.05418414995074272, "global_step": 20742, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03990887850522995, "train_loss_bc": 0.03990887850522995, "train_loss_llm": 0.0, "grad_norm": 0.07900028675794601, "global_step": 20743, "epoch": 576, "lr": 0.001} +{"train_loss": 0.035915572196245193, "train_loss_bc": 0.035915572196245193, "train_loss_llm": 0.0, "grad_norm": 0.11093905568122864, "global_step": 20744, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03298667073249817, "train_loss_bc": 0.032961759716272354, "train_loss_llm": 0.0024910271167755127, "grad_norm": 0.028794609010219574, "global_step": 20745, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03879685327410698, "train_loss_bc": 0.03879685327410698, "train_loss_llm": 0.0, "grad_norm": 0.05761455371975899, "global_step": 20746, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03276104852557182, "train_loss_bc": 0.03276104852557182, "train_loss_llm": 0.0, "grad_norm": 0.07620551437139511, "global_step": 20747, "epoch": 576, "lr": 0.001} +{"train_loss": 0.02554582804441452, "train_loss_bc": 0.02554582804441452, "train_loss_llm": 0.0, "grad_norm": 0.08876660466194153, "global_step": 20748, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03333436697721481, "train_loss_bc": 0.0333099439740181, "train_loss_llm": 0.002442389726638794, "grad_norm": 0.02822890318930149, "global_step": 20749, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03144346550107002, "train_loss_bc": 0.03144346550107002, "train_loss_llm": 0.0, "grad_norm": 0.05351259931921959, "global_step": 20750, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03445596247911453, "train_loss_bc": 0.03445596247911453, "train_loss_llm": 0.0, "grad_norm": 0.08431296795606613, "global_step": 20751, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03253398463129997, "train_loss_bc": 0.03253398463129997, "train_loss_llm": 0.0, "grad_norm": 0.10503558814525604, "global_step": 20752, "epoch": 576, "lr": 0.001} +{"train_loss": 0.02947843261063099, "train_loss_bc": 0.02947843261063099, "train_loss_llm": 0.0, "grad_norm": 0.03742930665612221, "global_step": 20753, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03096965327858925, "train_loss_bc": 0.03096965327858925, "train_loss_llm": 0.0, "grad_norm": 0.06511819362640381, "global_step": 20754, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03447966277599335, "train_loss_bc": 0.03447839617729187, "train_loss_llm": 0.00012665987014770508, "grad_norm": 0.07310834527015686, "global_step": 20755, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03383858501911163, "train_loss_bc": 0.03383858501911163, "train_loss_llm": 0.0, "grad_norm": 0.08391959220170975, "global_step": 20756, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03299092501401901, "train_loss_bc": 0.03299092501401901, "train_loss_llm": 0.0, "grad_norm": 0.03151548653841019, "global_step": 20757, "epoch": 576, "lr": 0.001} +{"train_loss": 0.039441682398319244, "train_loss_bc": 0.039441682398319244, "train_loss_llm": 0.0, "grad_norm": 0.05438496917486191, "global_step": 20758, "epoch": 576, "lr": 0.001} +{"train_loss": 0.030743146315217018, "train_loss_bc": 0.030743146315217018, "train_loss_llm": 0.0, "grad_norm": 0.07171422988176346, "global_step": 20759, "epoch": 576, "lr": 0.001} +{"train_loss": 0.030405372381210327, "train_loss_bc": 0.030405372381210327, "train_loss_llm": 0.0, "grad_norm": 0.08088236302137375, "global_step": 20760, "epoch": 576, "lr": 0.001} +{"train_loss": 0.0319259837269783, "train_loss_bc": 0.0319259837269783, "train_loss_llm": 0.0, "grad_norm": 0.03351116180419922, "global_step": 20761, "epoch": 576, "lr": 0.001} +{"train_loss": 0.04299946129322052, "train_loss_bc": 0.04299946129322052, "train_loss_llm": 0.0, "grad_norm": 0.056507889181375504, "global_step": 20762, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03302827477455139, "train_loss_bc": 0.03302827477455139, "train_loss_llm": 0.0, "grad_norm": 0.08302989602088928, "global_step": 20763, "epoch": 576, "lr": 0.001} +{"train_loss": 0.031171578913927078, "train_loss_bc": 0.031171578913927078, "train_loss_llm": 0.0, "grad_norm": 0.08722740411758423, "global_step": 20764, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03705976903438568, "train_loss_bc": 0.03705976903438568, "train_loss_llm": 0.0, "grad_norm": 0.0398406945168972, "global_step": 20765, "epoch": 576, "lr": 0.001} +{"train_loss": 0.035031758248806, "train_loss_bc": 0.0350193977355957, "train_loss_llm": 0.0012360811233520508, "grad_norm": 0.06452947109937668, "global_step": 20766, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03857531026005745, "train_loss_bc": 0.03857531026005745, "train_loss_llm": 0.0, "grad_norm": 0.10306358337402344, "global_step": 20767, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03000129759311676, "train_loss_bc": 0.03000129759311676, "train_loss_llm": 0.0, "grad_norm": 0.13440124690532684, "global_step": 20768, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03713800385594368, "train_loss_bc": 0.03712129220366478, "train_loss_llm": 0.0016713142395019531, "grad_norm": 0.04675726220011711, "global_step": 20769, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03413759171962738, "train_loss_bc": 0.03413759171962738, "train_loss_llm": 0.0, "grad_norm": 0.07231883704662323, "global_step": 20770, "epoch": 576, "lr": 0.001} +{"train_loss": 0.03335243644606736, "train_loss_bc": 0.034776993095874786, "train_loss_llm": 0.001984119415283203, "grad_norm": 0.09784577786922455, "global_step": 20771, "epoch": 576, "lr": 0.001, "val_loss": 0.07999081164598465} +{"train_loss": 0.03034665435552597, "train_loss_bc": 0.03034665435552597, "train_loss_llm": 0.0, "grad_norm": 0.12084373086690903, "global_step": 20772, "epoch": 577, "lr": 0.001} +{"train_loss": 0.038542333990335464, "train_loss_bc": 0.038517311215400696, "train_loss_llm": 0.0025023818016052246, "grad_norm": 0.04690657928586006, "global_step": 20773, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03957284986972809, "train_loss_bc": 0.03957284986972809, "train_loss_llm": 0.0, "grad_norm": 0.08364854753017426, "global_step": 20774, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03279808163642883, "train_loss_bc": 0.03279808163642883, "train_loss_llm": 0.0, "grad_norm": 0.10342513024806976, "global_step": 20775, "epoch": 577, "lr": 0.001} +{"train_loss": 0.036643270403146744, "train_loss_bc": 0.036643270403146744, "train_loss_llm": 0.0, "grad_norm": 0.1178852841258049, "global_step": 20776, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03318411111831665, "train_loss_bc": 0.03318411111831665, "train_loss_llm": 0.0, "grad_norm": 0.03170635551214218, "global_step": 20777, "epoch": 577, "lr": 0.001} +{"train_loss": 0.034125108271837234, "train_loss_bc": 0.034125108271837234, "train_loss_llm": 0.0, "grad_norm": 0.05717584863305092, "global_step": 20778, "epoch": 577, "lr": 0.001} +{"train_loss": 0.029739653691649437, "train_loss_bc": 0.029728878289461136, "train_loss_llm": 0.0010774731636047363, "grad_norm": 0.08124382793903351, "global_step": 20779, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03655000403523445, "train_loss_bc": 0.03653542324900627, "train_loss_llm": 0.0014580786228179932, "grad_norm": 0.09259242564439774, "global_step": 20780, "epoch": 577, "lr": 0.001} +{"train_loss": 0.0342080295085907, "train_loss_bc": 0.0342080295085907, "train_loss_llm": 0.0, "grad_norm": 0.03376839682459831, "global_step": 20781, "epoch": 577, "lr": 0.001} +{"train_loss": 0.033277880400419235, "train_loss_bc": 0.033277880400419235, "train_loss_llm": 0.0, "grad_norm": 0.04644598439335823, "global_step": 20782, "epoch": 577, "lr": 0.001} +{"train_loss": 0.030027668923139572, "train_loss_bc": 0.030020201578736305, "train_loss_llm": 0.0007467865943908691, "grad_norm": 0.06338085979223251, "global_step": 20783, "epoch": 577, "lr": 0.001} +{"train_loss": 0.033193912357091904, "train_loss_bc": 0.033193912357091904, "train_loss_llm": 0.0, "grad_norm": 0.08820270746946335, "global_step": 20784, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03593261167407036, "train_loss_bc": 0.03593261167407036, "train_loss_llm": 0.0, "grad_norm": 0.045513398945331573, "global_step": 20785, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03752615302801132, "train_loss_bc": 0.03752615302801132, "train_loss_llm": 0.0, "grad_norm": 0.08296573162078857, "global_step": 20786, "epoch": 577, "lr": 0.001} +{"train_loss": 0.029372574761509895, "train_loss_bc": 0.029372574761509895, "train_loss_llm": 0.0, "grad_norm": 0.10810953378677368, "global_step": 20787, "epoch": 577, "lr": 0.001} +{"train_loss": 0.028260311111807823, "train_loss_bc": 0.028260311111807823, "train_loss_llm": 0.0, "grad_norm": 0.1437089592218399, "global_step": 20788, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03121647983789444, "train_loss_bc": 0.03121647983789444, "train_loss_llm": 0.0, "grad_norm": 0.03304188698530197, "global_step": 20789, "epoch": 577, "lr": 0.001} +{"train_loss": 0.037439316511154175, "train_loss_bc": 0.037439316511154175, "train_loss_llm": 0.0, "grad_norm": 0.06744384765625, "global_step": 20790, "epoch": 577, "lr": 0.001} +{"train_loss": 0.032689835876226425, "train_loss_bc": 0.032689835876226425, "train_loss_llm": 0.0, "grad_norm": 0.08572764694690704, "global_step": 20791, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03290934860706329, "train_loss_bc": 0.03290934860706329, "train_loss_llm": 0.0, "grad_norm": 0.09996776282787323, "global_step": 20792, "epoch": 577, "lr": 0.001} +{"train_loss": 0.030150825157761574, "train_loss_bc": 0.030150825157761574, "train_loss_llm": 0.0, "grad_norm": 0.030064312741160393, "global_step": 20793, "epoch": 577, "lr": 0.001} +{"train_loss": 0.033119503408670425, "train_loss_bc": 0.033119503408670425, "train_loss_llm": 0.0, "grad_norm": 0.046384699642658234, "global_step": 20794, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03272707387804985, "train_loss_bc": 0.03271947801113129, "train_loss_llm": 0.0007596313953399658, "grad_norm": 0.0698293149471283, "global_step": 20795, "epoch": 577, "lr": 0.001} +{"train_loss": 0.037279196083545685, "train_loss_bc": 0.037279196083545685, "train_loss_llm": 0.0, "grad_norm": 0.0958520844578743, "global_step": 20796, "epoch": 577, "lr": 0.001} +{"train_loss": 0.030440151691436768, "train_loss_bc": 0.030440151691436768, "train_loss_llm": 0.0, "grad_norm": 0.02416403591632843, "global_step": 20797, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03427991643548012, "train_loss_bc": 0.03427991643548012, "train_loss_llm": 0.0, "grad_norm": 0.041395403444767, "global_step": 20798, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03374966233968735, "train_loss_bc": 0.03374966233968735, "train_loss_llm": 0.0, "grad_norm": 0.05473407357931137, "global_step": 20799, "epoch": 577, "lr": 0.001} +{"train_loss": 0.028523942455649376, "train_loss_bc": 0.028523942455649376, "train_loss_llm": 0.0, "grad_norm": 0.06703580170869827, "global_step": 20800, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03004581667482853, "train_loss_bc": 0.03004581667482853, "train_loss_llm": 0.0, "grad_norm": 0.03607267886400223, "global_step": 20801, "epoch": 577, "lr": 0.001} +{"train_loss": 0.033038780093193054, "train_loss_bc": 0.033038780093193054, "train_loss_llm": 0.0, "grad_norm": 0.0457027442753315, "global_step": 20802, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03017028607428074, "train_loss_bc": 0.03017028607428074, "train_loss_llm": 0.0, "grad_norm": 0.05946677550673485, "global_step": 20803, "epoch": 577, "lr": 0.001} +{"train_loss": 0.035838596522808075, "train_loss_bc": 0.035838596522808075, "train_loss_llm": 0.0, "grad_norm": 0.07425843179225922, "global_step": 20804, "epoch": 577, "lr": 0.001} +{"train_loss": 0.03797470033168793, "train_loss_bc": 0.03797470033168793, "train_loss_llm": 0.0, "grad_norm": 0.037094760686159134, "global_step": 20805, "epoch": 577, "lr": 0.001} +{"train_loss": 0.029167726635932922, "train_loss_bc": 0.029157785698771477, "train_loss_llm": 0.0009940862655639648, "grad_norm": 0.0508037768304348, "global_step": 20806, "epoch": 577, "lr": 0.001} +{"train_loss": 0.033317659929808646, "train_loss_bc": 0.03537338972091675, "train_loss_llm": 0.0, "grad_norm": 0.07079203426837921, "global_step": 20807, "epoch": 577, "lr": 0.001, "val_loss": 0.08245860785245895} +{"train_loss": 0.04029717668890953, "train_loss_bc": 0.04027441516518593, "train_loss_llm": 0.0022760629653930664, "grad_norm": 0.09229861199855804, "global_step": 20808, "epoch": 578, "lr": 0.001} +{"train_loss": 0.028112679719924927, "train_loss_bc": 0.028112679719924927, "train_loss_llm": 0.0, "grad_norm": 0.025924760848283768, "global_step": 20809, "epoch": 578, "lr": 0.001} +{"train_loss": 0.032402653247117996, "train_loss_bc": 0.032402653247117996, "train_loss_llm": 0.0, "grad_norm": 0.06003818288445473, "global_step": 20810, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03389876335859299, "train_loss_bc": 0.03389876335859299, "train_loss_llm": 0.0, "grad_norm": 0.08179077506065369, "global_step": 20811, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03153222054243088, "train_loss_bc": 0.03153222054243088, "train_loss_llm": 0.0, "grad_norm": 0.11322803050279617, "global_step": 20812, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03623978793621063, "train_loss_bc": 0.03621954470872879, "train_loss_llm": 0.0020244717597961426, "grad_norm": 0.029867960140109062, "global_step": 20813, "epoch": 578, "lr": 0.001} +{"train_loss": 0.036495864391326904, "train_loss_bc": 0.03649098426103592, "train_loss_llm": 0.00048786401748657227, "grad_norm": 0.04912492632865906, "global_step": 20814, "epoch": 578, "lr": 0.001} +{"train_loss": 0.027127182111144066, "train_loss_bc": 0.027119698002934456, "train_loss_llm": 0.000748366117477417, "grad_norm": 0.06089353933930397, "global_step": 20815, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03805062174797058, "train_loss_bc": 0.03805062174797058, "train_loss_llm": 0.0, "grad_norm": 0.08597443997859955, "global_step": 20816, "epoch": 578, "lr": 0.001} +{"train_loss": 0.027751076966524124, "train_loss_bc": 0.027751076966524124, "train_loss_llm": 0.0, "grad_norm": 0.024785909801721573, "global_step": 20817, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03209499269723892, "train_loss_bc": 0.03209499269723892, "train_loss_llm": 0.0, "grad_norm": 0.039374206215143204, "global_step": 20818, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03057674691081047, "train_loss_bc": 0.030559102073311806, "train_loss_llm": 0.001764446496963501, "grad_norm": 0.07523132115602493, "global_step": 20819, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03598371148109436, "train_loss_bc": 0.03598371148109436, "train_loss_llm": 0.0, "grad_norm": 0.09855878353118896, "global_step": 20820, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03306432440876961, "train_loss_bc": 0.03306432440876961, "train_loss_llm": 0.0, "grad_norm": 0.02732744812965393, "global_step": 20821, "epoch": 578, "lr": 0.001} +{"train_loss": 0.029743632301688194, "train_loss_bc": 0.029743632301688194, "train_loss_llm": 0.0, "grad_norm": 0.03535442799329758, "global_step": 20822, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03353775292634964, "train_loss_bc": 0.03353775292634964, "train_loss_llm": 0.0, "grad_norm": 0.053664714097976685, "global_step": 20823, "epoch": 578, "lr": 0.001} +{"train_loss": 0.034667063504457474, "train_loss_bc": 0.03466647118330002, "train_loss_llm": 5.906820297241211e-05, "grad_norm": 0.06265533715486526, "global_step": 20824, "epoch": 578, "lr": 0.001} +{"train_loss": 0.0374181866645813, "train_loss_bc": 0.0373992845416069, "train_loss_llm": 0.0018901228904724121, "grad_norm": 0.03319753333926201, "global_step": 20825, "epoch": 578, "lr": 0.001} +{"train_loss": 0.033358581364154816, "train_loss_bc": 0.033356353640556335, "train_loss_llm": 0.00022280216217041016, "grad_norm": 0.0575011782348156, "global_step": 20826, "epoch": 578, "lr": 0.001} +{"train_loss": 0.028329772874712944, "train_loss_bc": 0.028329772874712944, "train_loss_llm": 0.0, "grad_norm": 0.07350879162549973, "global_step": 20827, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03322721645236015, "train_loss_bc": 0.03322721645236015, "train_loss_llm": 0.0, "grad_norm": 0.09861424565315247, "global_step": 20828, "epoch": 578, "lr": 0.001} +{"train_loss": 0.02716420590877533, "train_loss_bc": 0.02716420590877533, "train_loss_llm": 0.0, "grad_norm": 0.03741990402340889, "global_step": 20829, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03159888833761215, "train_loss_bc": 0.03159888833761215, "train_loss_llm": 0.0, "grad_norm": 0.06250950694084167, "global_step": 20830, "epoch": 578, "lr": 0.001} +{"train_loss": 0.030885810032486916, "train_loss_bc": 0.030885810032486916, "train_loss_llm": 0.0, "grad_norm": 0.0885867029428482, "global_step": 20831, "epoch": 578, "lr": 0.001} +{"train_loss": 0.040931057184934616, "train_loss_bc": 0.04092720150947571, "train_loss_llm": 0.00038552284240722656, "grad_norm": 0.11806142330169678, "global_step": 20832, "epoch": 578, "lr": 0.001} +{"train_loss": 0.0272473506629467, "train_loss_bc": 0.0272473506629467, "train_loss_llm": 0.0, "grad_norm": 0.024298368021845818, "global_step": 20833, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03868646174669266, "train_loss_bc": 0.03868646174669266, "train_loss_llm": 0.0, "grad_norm": 0.04643535986542702, "global_step": 20834, "epoch": 578, "lr": 0.001} +{"train_loss": 0.0319882370531559, "train_loss_bc": 0.0319882370531559, "train_loss_llm": 0.0, "grad_norm": 0.06412918120622635, "global_step": 20835, "epoch": 578, "lr": 0.001} +{"train_loss": 0.030690006911754608, "train_loss_bc": 0.030684206634759903, "train_loss_llm": 0.0005801022052764893, "grad_norm": 0.0794801190495491, "global_step": 20836, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03249680623412132, "train_loss_bc": 0.03249680623412132, "train_loss_llm": 0.0, "grad_norm": 0.028629634529352188, "global_step": 20837, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03403514623641968, "train_loss_bc": 0.03403514623641968, "train_loss_llm": 0.0, "grad_norm": 0.04876391589641571, "global_step": 20838, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03277280554175377, "train_loss_bc": 0.03277280554175377, "train_loss_llm": 0.0, "grad_norm": 0.05744121968746185, "global_step": 20839, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03430226072669029, "train_loss_bc": 0.03430226072669029, "train_loss_llm": 0.0, "grad_norm": 0.07192854583263397, "global_step": 20840, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03533072769641876, "train_loss_bc": 0.03533072769641876, "train_loss_llm": 0.0, "grad_norm": 0.03096618689596653, "global_step": 20841, "epoch": 578, "lr": 0.001} +{"train_loss": 0.03752448037266731, "train_loss_bc": 0.03750038519501686, "train_loss_llm": 0.0024096667766571045, "grad_norm": 0.049988459795713425, "global_step": 20842, "epoch": 578, "lr": 0.001} +{"train_loss": 0.033239762402243085, "train_loss_bc": 0.037067193537950516, "train_loss_llm": 0.0, "grad_norm": 0.06822056323289871, "global_step": 20843, "epoch": 578, "lr": 0.001, "val_loss": 0.08093810081481934} +{"train_loss": 0.03520023822784424, "train_loss_bc": 0.03520023822784424, "train_loss_llm": 0.0, "grad_norm": 0.08876948058605194, "global_step": 20844, "epoch": 579, "lr": 0.001} +{"train_loss": 0.031875915825366974, "train_loss_bc": 0.031871769577264786, "train_loss_llm": 0.0004144906997680664, "grad_norm": 0.03086210787296295, "global_step": 20845, "epoch": 579, "lr": 0.001} +{"train_loss": 0.035945601761341095, "train_loss_bc": 0.035945601761341095, "train_loss_llm": 0.0, "grad_norm": 0.05511902645230293, "global_step": 20846, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03678346425294876, "train_loss_bc": 0.03677884116768837, "train_loss_llm": 0.0004622042179107666, "grad_norm": 0.06654593348503113, "global_step": 20847, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03202633559703827, "train_loss_bc": 0.03202611953020096, "train_loss_llm": 2.1725893020629883e-05, "grad_norm": 0.08075898140668869, "global_step": 20848, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03389010950922966, "train_loss_bc": 0.03388090059161186, "train_loss_llm": 0.0009210705757141113, "grad_norm": 0.03594178706407547, "global_step": 20849, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03162837401032448, "train_loss_bc": 0.03162674233317375, "train_loss_llm": 0.00016301870346069336, "grad_norm": 0.06023475155234337, "global_step": 20850, "epoch": 579, "lr": 0.001} +{"train_loss": 0.030242186039686203, "train_loss_bc": 0.030242186039686203, "train_loss_llm": 0.0, "grad_norm": 0.08479276299476624, "global_step": 20851, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03456072136759758, "train_loss_bc": 0.03456072136759758, "train_loss_llm": 0.0, "grad_norm": 0.11571892350912094, "global_step": 20852, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03413369879126549, "train_loss_bc": 0.03413369879126549, "train_loss_llm": 0.0, "grad_norm": 0.02921607904136181, "global_step": 20853, "epoch": 579, "lr": 0.001} +{"train_loss": 0.02862301841378212, "train_loss_bc": 0.02862301841378212, "train_loss_llm": 0.0, "grad_norm": 0.04645043984055519, "global_step": 20854, "epoch": 579, "lr": 0.001} +{"train_loss": 0.031438954174518585, "train_loss_bc": 0.03139574080705643, "train_loss_llm": 0.004321157932281494, "grad_norm": 0.05721366032958031, "global_step": 20855, "epoch": 579, "lr": 0.001} +{"train_loss": 0.030069950968027115, "train_loss_bc": 0.030069950968027115, "train_loss_llm": 0.0, "grad_norm": 0.06884542852640152, "global_step": 20856, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03379683196544647, "train_loss_bc": 0.03378797695040703, "train_loss_llm": 0.0008855760097503662, "grad_norm": 0.027345845475792885, "global_step": 20857, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03241574391722679, "train_loss_bc": 0.032414983958005905, "train_loss_llm": 7.590651512145996e-05, "grad_norm": 0.04651448503136635, "global_step": 20858, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03334622457623482, "train_loss_bc": 0.03334622457623482, "train_loss_llm": 0.0, "grad_norm": 0.06106055527925491, "global_step": 20859, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03260774165391922, "train_loss_bc": 0.03260774165391922, "train_loss_llm": 0.0, "grad_norm": 0.0753980353474617, "global_step": 20860, "epoch": 579, "lr": 0.001} +{"train_loss": 0.033857736736536026, "train_loss_bc": 0.033853091299533844, "train_loss_llm": 0.00046452879905700684, "grad_norm": 0.029047247022390366, "global_step": 20861, "epoch": 579, "lr": 0.001} +{"train_loss": 0.0378461554646492, "train_loss_bc": 0.0378461554646492, "train_loss_llm": 0.0, "grad_norm": 0.04452645033597946, "global_step": 20862, "epoch": 579, "lr": 0.001} +{"train_loss": 0.029659397900104523, "train_loss_bc": 0.029636455699801445, "train_loss_llm": 0.002294182777404785, "grad_norm": 0.056378740817308426, "global_step": 20863, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03317912295460701, "train_loss_bc": 0.03317912295460701, "train_loss_llm": 0.0, "grad_norm": 0.06937568634748459, "global_step": 20864, "epoch": 579, "lr": 0.001} +{"train_loss": 0.033401794731616974, "train_loss_bc": 0.033401794731616974, "train_loss_llm": 0.0, "grad_norm": 0.031380049884319305, "global_step": 20865, "epoch": 579, "lr": 0.001} +{"train_loss": 0.02891657128930092, "train_loss_bc": 0.02891657128930092, "train_loss_llm": 0.0, "grad_norm": 0.0526709109544754, "global_step": 20866, "epoch": 579, "lr": 0.001} +{"train_loss": 0.030005941167473793, "train_loss_bc": 0.030005941167473793, "train_loss_llm": 0.0, "grad_norm": 0.0719529315829277, "global_step": 20867, "epoch": 579, "lr": 0.001} +{"train_loss": 0.031371407210826874, "train_loss_bc": 0.031371407210826874, "train_loss_llm": 0.0, "grad_norm": 0.08200521767139435, "global_step": 20868, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03617677837610245, "train_loss_bc": 0.03617677837610245, "train_loss_llm": 0.0, "grad_norm": 0.03248301148414612, "global_step": 20869, "epoch": 579, "lr": 0.001} +{"train_loss": 0.039280202239751816, "train_loss_bc": 0.039280202239751816, "train_loss_llm": 0.0, "grad_norm": 0.04633179306983948, "global_step": 20870, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03392855077981949, "train_loss_bc": 0.03392855077981949, "train_loss_llm": 0.0, "grad_norm": 0.05464982986450195, "global_step": 20871, "epoch": 579, "lr": 0.001} +{"train_loss": 0.032703813165426254, "train_loss_bc": 0.032703813165426254, "train_loss_llm": 0.0, "grad_norm": 0.06913096457719803, "global_step": 20872, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03557002544403076, "train_loss_bc": 0.03557002544403076, "train_loss_llm": 0.0, "grad_norm": 0.03076779842376709, "global_step": 20873, "epoch": 579, "lr": 0.001} +{"train_loss": 0.028712371364235878, "train_loss_bc": 0.028712371364235878, "train_loss_llm": 0.0, "grad_norm": 0.04605253040790558, "global_step": 20874, "epoch": 579, "lr": 0.001} +{"train_loss": 0.030321015045046806, "train_loss_bc": 0.030321015045046806, "train_loss_llm": 0.0, "grad_norm": 0.06313082575798035, "global_step": 20875, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03155326098203659, "train_loss_bc": 0.031548742204904556, "train_loss_llm": 0.0004520118236541748, "grad_norm": 0.08459189534187317, "global_step": 20876, "epoch": 579, "lr": 0.001} +{"train_loss": 0.035271454602479935, "train_loss_bc": 0.03524918481707573, "train_loss_llm": 0.0022271275520324707, "grad_norm": 0.030708247795701027, "global_step": 20877, "epoch": 579, "lr": 0.001} +{"train_loss": 0.033590152859687805, "train_loss_bc": 0.033590152859687805, "train_loss_llm": 0.0, "grad_norm": 0.04794874042272568, "global_step": 20878, "epoch": 579, "lr": 0.001} +{"train_loss": 0.03308597098415097, "train_loss_bc": 0.03714122995734215, "train_loss_llm": 0.0022863149642944336, "grad_norm": 0.06363734602928162, "global_step": 20879, "epoch": 579, "lr": 0.001, "val_loss": 0.08078303188085556} +{"train_loss": 0.03174909949302673, "train_loss_bc": 0.03174909949302673, "train_loss_llm": 0.0, "grad_norm": 0.07983306795358658, "global_step": 20880, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03502178192138672, "train_loss_bc": 0.03502178192138672, "train_loss_llm": 0.0, "grad_norm": 0.02564259059727192, "global_step": 20881, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03099019266664982, "train_loss_bc": 0.03099019266664982, "train_loss_llm": 0.0, "grad_norm": 0.042810626327991486, "global_step": 20882, "epoch": 580, "lr": 0.001} +{"train_loss": 0.027444999665021896, "train_loss_bc": 0.027444999665021896, "train_loss_llm": 0.0, "grad_norm": 0.057454392313957214, "global_step": 20883, "epoch": 580, "lr": 0.001} +{"train_loss": 0.030834652483463287, "train_loss_bc": 0.030834652483463287, "train_loss_llm": 0.0, "grad_norm": 0.06776801496744156, "global_step": 20884, "epoch": 580, "lr": 0.001} +{"train_loss": 0.0312386155128479, "train_loss_bc": 0.031231889501214027, "train_loss_llm": 0.0006725192070007324, "grad_norm": 0.02759481593966484, "global_step": 20885, "epoch": 580, "lr": 0.001} +{"train_loss": 0.04153985157608986, "train_loss_bc": 0.04153985157608986, "train_loss_llm": 0.0, "grad_norm": 0.053037699311971664, "global_step": 20886, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03719193488359451, "train_loss_bc": 0.03719193488359451, "train_loss_llm": 0.0, "grad_norm": 0.07886739820241928, "global_step": 20887, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03476220369338989, "train_loss_bc": 0.03476220369338989, "train_loss_llm": 0.0, "grad_norm": 0.09754890948534012, "global_step": 20888, "epoch": 580, "lr": 0.001} +{"train_loss": 0.033671651035547256, "train_loss_bc": 0.033671651035547256, "train_loss_llm": 0.0, "grad_norm": 0.024747610092163086, "global_step": 20889, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03535563871264458, "train_loss_bc": 0.03535563871264458, "train_loss_llm": 0.0, "grad_norm": 0.03892946243286133, "global_step": 20890, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03276322782039642, "train_loss_bc": 0.03276322782039642, "train_loss_llm": 0.0, "grad_norm": 0.057503506541252136, "global_step": 20891, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03201861307024956, "train_loss_bc": 0.03201861307024956, "train_loss_llm": 0.0, "grad_norm": 0.07918893545866013, "global_step": 20892, "epoch": 580, "lr": 0.001} +{"train_loss": 0.02766195498406887, "train_loss_bc": 0.027659695595502853, "train_loss_llm": 0.00022596120834350586, "grad_norm": 0.031150931492447853, "global_step": 20893, "epoch": 580, "lr": 0.001} +{"train_loss": 0.037008244544267654, "train_loss_bc": 0.0370064303278923, "train_loss_llm": 0.00018140673637390137, "grad_norm": 0.06139521300792694, "global_step": 20894, "epoch": 580, "lr": 0.001} +{"train_loss": 0.036145295947790146, "train_loss_bc": 0.03614308312535286, "train_loss_llm": 0.00022143125534057617, "grad_norm": 0.08149497956037521, "global_step": 20895, "epoch": 580, "lr": 0.001} +{"train_loss": 0.032989028841257095, "train_loss_bc": 0.032989028841257095, "train_loss_llm": 0.0, "grad_norm": 0.09885116666555405, "global_step": 20896, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03127684444189072, "train_loss_bc": 0.03127684444189072, "train_loss_llm": 0.0, "grad_norm": 0.03190583735704422, "global_step": 20897, "epoch": 580, "lr": 0.001} +{"train_loss": 0.030076388269662857, "train_loss_bc": 0.030063459649682045, "train_loss_llm": 0.0012928545475006104, "grad_norm": 0.047697514295578, "global_step": 20898, "epoch": 580, "lr": 0.001} +{"train_loss": 0.032200463116168976, "train_loss_bc": 0.03219177946448326, "train_loss_llm": 0.0008683204650878906, "grad_norm": 0.06605309993028641, "global_step": 20899, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03469807654619217, "train_loss_bc": 0.03469807654619217, "train_loss_llm": 0.0, "grad_norm": 0.08828499168157578, "global_step": 20900, "epoch": 580, "lr": 0.001} +{"train_loss": 0.028222376480698586, "train_loss_bc": 0.028212806209921837, "train_loss_llm": 0.0009570717811584473, "grad_norm": 0.027965117245912552, "global_step": 20901, "epoch": 580, "lr": 0.001} +{"train_loss": 0.032796554267406464, "train_loss_bc": 0.032788146287202835, "train_loss_llm": 0.0008406341075897217, "grad_norm": 0.04580676928162575, "global_step": 20902, "epoch": 580, "lr": 0.001} +{"train_loss": 0.031840745359659195, "train_loss_bc": 0.031840745359659195, "train_loss_llm": 0.0, "grad_norm": 0.059536002576351166, "global_step": 20903, "epoch": 580, "lr": 0.001} +{"train_loss": 0.0340300127863884, "train_loss_bc": 0.0340300127863884, "train_loss_llm": 0.0, "grad_norm": 0.07095082104206085, "global_step": 20904, "epoch": 580, "lr": 0.001} +{"train_loss": 0.0323549322783947, "train_loss_bc": 0.0323549322783947, "train_loss_llm": 0.0, "grad_norm": 0.025245044380426407, "global_step": 20905, "epoch": 580, "lr": 0.001} +{"train_loss": 0.040573589503765106, "train_loss_bc": 0.040573589503765106, "train_loss_llm": 0.0, "grad_norm": 0.04743442311882973, "global_step": 20906, "epoch": 580, "lr": 0.001} +{"train_loss": 0.0317271426320076, "train_loss_bc": 0.03171806037425995, "train_loss_llm": 0.0009084045886993408, "grad_norm": 0.05985734611749649, "global_step": 20907, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03385155275464058, "train_loss_bc": 0.03385155275464058, "train_loss_llm": 0.0, "grad_norm": 0.06177512928843498, "global_step": 20908, "epoch": 580, "lr": 0.001} +{"train_loss": 0.02844863012433052, "train_loss_bc": 0.02844863012433052, "train_loss_llm": 0.0, "grad_norm": 0.032239459455013275, "global_step": 20909, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03372485935688019, "train_loss_bc": 0.03372485935688019, "train_loss_llm": 0.0, "grad_norm": 0.04497040808200836, "global_step": 20910, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03434650972485542, "train_loss_bc": 0.03434650972485542, "train_loss_llm": 0.0, "grad_norm": 0.06402093917131424, "global_step": 20911, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03089490532875061, "train_loss_bc": 0.030882520601153374, "train_loss_llm": 0.001238405704498291, "grad_norm": 0.07535470277070999, "global_step": 20912, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03370703384280205, "train_loss_bc": 0.03370703384280205, "train_loss_llm": 0.0, "grad_norm": 0.0318387895822525, "global_step": 20913, "epoch": 580, "lr": 0.001} +{"train_loss": 0.036959532648324966, "train_loss_bc": 0.036959532648324966, "train_loss_llm": 0.0, "grad_norm": 0.048402246087789536, "global_step": 20914, "epoch": 580, "lr": 0.001} +{"train_loss": 0.03290882887732652, "train_loss_bc": 0.02460070326924324, "train_loss_llm": 0.0, "grad_norm": 0.06635530292987823, "global_step": 20915, "epoch": 580, "lr": 0.001, "train/cumulative_reward": 94.72169109370313, "train/mean_score": 16.83755034401946, "train/success_rate": 0.4, "test/cumulative_reward": 31.830845074546218, "test/mean_score": 8.598954522453399, "test/success_rate": 0.24, "val_loss": 0.08029237389564514, "train_action_mse_error": 0.03569714352488518} +{"train_loss": 0.02801007404923439, "train_loss_bc": 0.028009701520204544, "train_loss_llm": 3.731250762939453e-05, "grad_norm": 0.07886994630098343, "global_step": 20916, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03552854433655739, "train_loss_bc": 0.03552188351750374, "train_loss_llm": 0.0006659030914306641, "grad_norm": 0.032228484749794006, "global_step": 20917, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03186969459056854, "train_loss_bc": 0.03186969459056854, "train_loss_llm": 0.0, "grad_norm": 0.04922409728169441, "global_step": 20918, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03231639415025711, "train_loss_bc": 0.03231639415025711, "train_loss_llm": 0.0, "grad_norm": 0.06307370215654373, "global_step": 20919, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03621814399957657, "train_loss_bc": 0.03621814399957657, "train_loss_llm": 0.0, "grad_norm": 0.07998084276914597, "global_step": 20920, "epoch": 581, "lr": 0.001} +{"train_loss": 0.043410733342170715, "train_loss_bc": 0.043410733342170715, "train_loss_llm": 0.0, "grad_norm": 0.03759271651506424, "global_step": 20921, "epoch": 581, "lr": 0.001} +{"train_loss": 0.028790419921278954, "train_loss_bc": 0.028790419921278954, "train_loss_llm": 0.0, "grad_norm": 0.042190950363874435, "global_step": 20922, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03614730015397072, "train_loss_bc": 0.03614155575633049, "train_loss_llm": 0.0005744397640228271, "grad_norm": 0.06327203661203384, "global_step": 20923, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03343478962779045, "train_loss_bc": 0.03343478962779045, "train_loss_llm": 0.0, "grad_norm": 0.08102067559957504, "global_step": 20924, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03101908601820469, "train_loss_bc": 0.03101908601820469, "train_loss_llm": 0.0, "grad_norm": 0.023594669997692108, "global_step": 20925, "epoch": 581, "lr": 0.001} +{"train_loss": 0.04052557796239853, "train_loss_bc": 0.04052557796239853, "train_loss_llm": 0.0, "grad_norm": 0.044610992074012756, "global_step": 20926, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03262219578027725, "train_loss_bc": 0.03262219578027725, "train_loss_llm": 0.0, "grad_norm": 0.05785654857754707, "global_step": 20927, "epoch": 581, "lr": 0.001} +{"train_loss": 0.026416916400194168, "train_loss_bc": 0.026406630873680115, "train_loss_llm": 0.0010285675525665283, "grad_norm": 0.07586509734392166, "global_step": 20928, "epoch": 581, "lr": 0.001} +{"train_loss": 0.029086381196975708, "train_loss_bc": 0.029086381196975708, "train_loss_llm": 0.0, "grad_norm": 0.0250189658254385, "global_step": 20929, "epoch": 581, "lr": 0.001} +{"train_loss": 0.031474530696868896, "train_loss_bc": 0.031474530696868896, "train_loss_llm": 0.0, "grad_norm": 0.03477218374609947, "global_step": 20930, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03395156189799309, "train_loss_bc": 0.03395156189799309, "train_loss_llm": 0.0, "grad_norm": 0.050689127296209335, "global_step": 20931, "epoch": 581, "lr": 0.001} +{"train_loss": 0.028346657752990723, "train_loss_bc": 0.028346657752990723, "train_loss_llm": 0.0, "grad_norm": 0.06373120844364166, "global_step": 20932, "epoch": 581, "lr": 0.001} +{"train_loss": 0.034297794103622437, "train_loss_bc": 0.034297794103622437, "train_loss_llm": 0.0, "grad_norm": 0.035003721714019775, "global_step": 20933, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03504561632871628, "train_loss_bc": 0.03504561632871628, "train_loss_llm": 0.0, "grad_norm": 0.057341501116752625, "global_step": 20934, "epoch": 581, "lr": 0.001} +{"train_loss": 0.029803140088915825, "train_loss_bc": 0.02980036847293377, "train_loss_llm": 0.0002771317958831787, "grad_norm": 0.0724339634180069, "global_step": 20935, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03190910443663597, "train_loss_bc": 0.03190910443663597, "train_loss_llm": 0.0, "grad_norm": 0.08324968814849854, "global_step": 20936, "epoch": 581, "lr": 0.001} +{"train_loss": 0.029904969036579132, "train_loss_bc": 0.029894251376390457, "train_loss_llm": 0.001071840524673462, "grad_norm": 0.028710445389151573, "global_step": 20937, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03235575929284096, "train_loss_bc": 0.0323377326130867, "train_loss_llm": 0.0018026232719421387, "grad_norm": 0.05025016516447067, "global_step": 20938, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03143329545855522, "train_loss_bc": 0.03143329545855522, "train_loss_llm": 0.0, "grad_norm": 0.0770312249660492, "global_step": 20939, "epoch": 581, "lr": 0.001} +{"train_loss": 0.030517814680933952, "train_loss_bc": 0.030517814680933952, "train_loss_llm": 0.0, "grad_norm": 0.09489230811595917, "global_step": 20940, "epoch": 581, "lr": 0.001} +{"train_loss": 0.033511847257614136, "train_loss_bc": 0.033511847257614136, "train_loss_llm": 0.0, "grad_norm": 0.03051520138978958, "global_step": 20941, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03613300248980522, "train_loss_bc": 0.036128561943769455, "train_loss_llm": 0.00044408440589904785, "grad_norm": 0.0433829091489315, "global_step": 20942, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03176175057888031, "train_loss_bc": 0.03176175057888031, "train_loss_llm": 0.0, "grad_norm": 0.05558709055185318, "global_step": 20943, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03313247486948967, "train_loss_bc": 0.03313247486948967, "train_loss_llm": 0.0, "grad_norm": 0.07158540189266205, "global_step": 20944, "epoch": 581, "lr": 0.001} +{"train_loss": 0.032591257244348526, "train_loss_bc": 0.032591257244348526, "train_loss_llm": 0.0, "grad_norm": 0.03481290489435196, "global_step": 20945, "epoch": 581, "lr": 0.001} +{"train_loss": 0.0331437848508358, "train_loss_bc": 0.033138617873191833, "train_loss_llm": 0.0005166828632354736, "grad_norm": 0.046520061790943146, "global_step": 20946, "epoch": 581, "lr": 0.001} +{"train_loss": 0.02592676505446434, "train_loss_bc": 0.02592313475906849, "train_loss_llm": 0.0003630518913269043, "grad_norm": 0.06309109181165695, "global_step": 20947, "epoch": 581, "lr": 0.001} +{"train_loss": 0.034491702914237976, "train_loss_bc": 0.034491702914237976, "train_loss_llm": 0.0, "grad_norm": 0.06818234175443649, "global_step": 20948, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03273205831646919, "train_loss_bc": 0.03273205831646919, "train_loss_llm": 0.0, "grad_norm": 0.0293403510004282, "global_step": 20949, "epoch": 581, "lr": 0.001} +{"train_loss": 0.035168759524822235, "train_loss_bc": 0.035168759524822235, "train_loss_llm": 0.0, "grad_norm": 0.05144760385155678, "global_step": 20950, "epoch": 581, "lr": 0.001} +{"train_loss": 0.03298535642938481, "train_loss_bc": 0.044442933052778244, "train_loss_llm": 0.0, "grad_norm": 0.06338001042604446, "global_step": 20951, "epoch": 581, "lr": 0.001, "val_loss": 0.08032466471195221} +{"train_loss": 0.027366798371076584, "train_loss_bc": 0.027366798371076584, "train_loss_llm": 0.0, "grad_norm": 0.08264503628015518, "global_step": 20952, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03449336439371109, "train_loss_bc": 0.03449336439371109, "train_loss_llm": 0.0, "grad_norm": 0.03256405144929886, "global_step": 20953, "epoch": 582, "lr": 0.001} +{"train_loss": 0.029989134520292282, "train_loss_bc": 0.029989134520292282, "train_loss_llm": 0.0, "grad_norm": 0.041011616587638855, "global_step": 20954, "epoch": 582, "lr": 0.001} +{"train_loss": 0.034804247319698334, "train_loss_bc": 0.034804247319698334, "train_loss_llm": 0.0, "grad_norm": 0.06317231804132462, "global_step": 20955, "epoch": 582, "lr": 0.001} +{"train_loss": 0.036340612918138504, "train_loss_bc": 0.036340612918138504, "train_loss_llm": 0.0, "grad_norm": 0.07845784723758698, "global_step": 20956, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03000481426715851, "train_loss_bc": 0.02997700683772564, "train_loss_llm": 0.0027807950973510742, "grad_norm": 0.034055452793836594, "global_step": 20957, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03355351462960243, "train_loss_bc": 0.03355351462960243, "train_loss_llm": 0.0, "grad_norm": 0.05036790668964386, "global_step": 20958, "epoch": 582, "lr": 0.001} +{"train_loss": 0.028319843113422394, "train_loss_bc": 0.028309348970651627, "train_loss_llm": 0.0010494589805603027, "grad_norm": 0.06482575088739395, "global_step": 20959, "epoch": 582, "lr": 0.001} +{"train_loss": 0.035458073019981384, "train_loss_bc": 0.035458073019981384, "train_loss_llm": 0.0, "grad_norm": 0.083103246986866, "global_step": 20960, "epoch": 582, "lr": 0.001} +{"train_loss": 0.04248465970158577, "train_loss_bc": 0.04248465970158577, "train_loss_llm": 0.0, "grad_norm": 0.03478582575917244, "global_step": 20961, "epoch": 582, "lr": 0.001} +{"train_loss": 0.030937256291508675, "train_loss_bc": 0.0309237502515316, "train_loss_llm": 0.0013506710529327393, "grad_norm": 0.04916035011410713, "global_step": 20962, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03694057837128639, "train_loss_bc": 0.03694057837128639, "train_loss_llm": 0.0, "grad_norm": 0.06259707361459732, "global_step": 20963, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03237403929233551, "train_loss_bc": 0.03236459940671921, "train_loss_llm": 0.0009441375732421875, "grad_norm": 0.0832231268286705, "global_step": 20964, "epoch": 582, "lr": 0.001} +{"train_loss": 0.02446766197681427, "train_loss_bc": 0.024442588910460472, "train_loss_llm": 0.0025072991847991943, "grad_norm": 0.02337724156677723, "global_step": 20965, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03050297684967518, "train_loss_bc": 0.03050297684967518, "train_loss_llm": 0.0, "grad_norm": 0.035730283707380295, "global_step": 20966, "epoch": 582, "lr": 0.001} +{"train_loss": 0.0335388109087944, "train_loss_bc": 0.0335388109087944, "train_loss_llm": 0.0, "grad_norm": 0.05019142106175423, "global_step": 20967, "epoch": 582, "lr": 0.001} +{"train_loss": 0.0347803458571434, "train_loss_bc": 0.034763529896736145, "train_loss_llm": 0.0016816556453704834, "grad_norm": 0.06211322173476219, "global_step": 20968, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03425348922610283, "train_loss_bc": 0.03425348922610283, "train_loss_llm": 0.0, "grad_norm": 0.0253597442060709, "global_step": 20969, "epoch": 582, "lr": 0.001} +{"train_loss": 0.030536429956555367, "train_loss_bc": 0.030536429956555367, "train_loss_llm": 0.0, "grad_norm": 0.03397008404135704, "global_step": 20970, "epoch": 582, "lr": 0.001} +{"train_loss": 0.031693343073129654, "train_loss_bc": 0.031693343073129654, "train_loss_llm": 0.0, "grad_norm": 0.04455619305372238, "global_step": 20971, "epoch": 582, "lr": 0.001} +{"train_loss": 0.0359971709549427, "train_loss_bc": 0.0359971709549427, "train_loss_llm": 0.0, "grad_norm": 0.05885670706629753, "global_step": 20972, "epoch": 582, "lr": 0.001} +{"train_loss": 0.034702278673648834, "train_loss_bc": 0.03469822183251381, "train_loss_llm": 0.0004056692123413086, "grad_norm": 0.03049723617732525, "global_step": 20973, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03142888471484184, "train_loss_bc": 0.031424380838871, "train_loss_llm": 0.00045049190521240234, "grad_norm": 0.04686667397618294, "global_step": 20974, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03166580945253372, "train_loss_bc": 0.03166580945253372, "train_loss_llm": 0.0, "grad_norm": 0.05782892927527428, "global_step": 20975, "epoch": 582, "lr": 0.001} +{"train_loss": 0.033966001123189926, "train_loss_bc": 0.033966001123189926, "train_loss_llm": 0.0, "grad_norm": 0.07104987651109695, "global_step": 20976, "epoch": 582, "lr": 0.001} +{"train_loss": 0.028583871200680733, "train_loss_bc": 0.028576280921697617, "train_loss_llm": 0.0007590055465698242, "grad_norm": 0.026186008006334305, "global_step": 20977, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03915242850780487, "train_loss_bc": 0.03915242850780487, "train_loss_llm": 0.0, "grad_norm": 0.04361647740006447, "global_step": 20978, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03050280176103115, "train_loss_bc": 0.03050280176103115, "train_loss_llm": 0.0, "grad_norm": 0.05410515144467354, "global_step": 20979, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03145338222384453, "train_loss_bc": 0.03145338222384453, "train_loss_llm": 0.0, "grad_norm": 0.06289161741733551, "global_step": 20980, "epoch": 582, "lr": 0.001} +{"train_loss": 0.0276893712580204, "train_loss_bc": 0.0276893712580204, "train_loss_llm": 0.0, "grad_norm": 0.026991816237568855, "global_step": 20981, "epoch": 582, "lr": 0.001} +{"train_loss": 0.029969705268740654, "train_loss_bc": 0.029969705268740654, "train_loss_llm": 0.0, "grad_norm": 0.04031170532107353, "global_step": 20982, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03352374583482742, "train_loss_bc": 0.03352363780140877, "train_loss_llm": 1.0907649993896484e-05, "grad_norm": 0.05264914780855179, "global_step": 20983, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03341890871524811, "train_loss_bc": 0.033399175852537155, "train_loss_llm": 0.0019731223583221436, "grad_norm": 0.06363889575004578, "global_step": 20984, "epoch": 582, "lr": 0.001} +{"train_loss": 0.038954511284828186, "train_loss_bc": 0.038951653987169266, "train_loss_llm": 0.00028568506240844727, "grad_norm": 0.03087654337286949, "global_step": 20985, "epoch": 582, "lr": 0.001} +{"train_loss": 0.03763381764292717, "train_loss_bc": 0.03763381764292717, "train_loss_llm": 0.0, "grad_norm": 0.04519958421587944, "global_step": 20986, "epoch": 582, "lr": 0.001} +{"train_loss": 0.0327464547008276, "train_loss_bc": 0.027389686554670334, "train_loss_llm": 0.0, "grad_norm": 0.06126983091235161, "global_step": 20987, "epoch": 582, "lr": 0.001, "val_loss": 0.08169584721326828} +{"train_loss": 0.02746790461242199, "train_loss_bc": 0.02746790461242199, "train_loss_llm": 0.0, "grad_norm": 0.06585133820772171, "global_step": 20988, "epoch": 583, "lr": 0.001} +{"train_loss": 0.030206965282559395, "train_loss_bc": 0.030206412076950073, "train_loss_llm": 5.53131103515625e-05, "grad_norm": 0.02399856224656105, "global_step": 20989, "epoch": 583, "lr": 0.001} +{"train_loss": 0.026142075657844543, "train_loss_bc": 0.026142075657844543, "train_loss_llm": 0.0, "grad_norm": 0.03403506428003311, "global_step": 20990, "epoch": 583, "lr": 0.001} +{"train_loss": 0.036800116300582886, "train_loss_bc": 0.036800116300582886, "train_loss_llm": 0.0, "grad_norm": 0.04205924645066261, "global_step": 20991, "epoch": 583, "lr": 0.001} +{"train_loss": 0.032131921499967575, "train_loss_bc": 0.032131921499967575, "train_loss_llm": 0.0, "grad_norm": 0.055228035897016525, "global_step": 20992, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03480914607644081, "train_loss_bc": 0.03480914607644081, "train_loss_llm": 0.0, "grad_norm": 0.03022242709994316, "global_step": 20993, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03178112953901291, "train_loss_bc": 0.031777314841747284, "train_loss_llm": 0.0003813207149505615, "grad_norm": 0.05102212354540825, "global_step": 20994, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03413937985897064, "train_loss_bc": 0.03413937985897064, "train_loss_llm": 0.0, "grad_norm": 0.06933356076478958, "global_step": 20995, "epoch": 583, "lr": 0.001} +{"train_loss": 0.030115216970443726, "train_loss_bc": 0.030115216970443726, "train_loss_llm": 0.0, "grad_norm": 0.08106859773397446, "global_step": 20996, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03214486688375473, "train_loss_bc": 0.032141249626874924, "train_loss_llm": 0.0003618597984313965, "grad_norm": 0.026543796062469482, "global_step": 20997, "epoch": 583, "lr": 0.001} +{"train_loss": 0.035977691411972046, "train_loss_bc": 0.035975147038698196, "train_loss_llm": 0.0002543330192565918, "grad_norm": 0.04569639638066292, "global_step": 20998, "epoch": 583, "lr": 0.001} +{"train_loss": 0.030426405370235443, "train_loss_bc": 0.030426405370235443, "train_loss_llm": 0.0, "grad_norm": 0.06430863589048386, "global_step": 20999, "epoch": 583, "lr": 0.001} +{"train_loss": 0.02918209508061409, "train_loss_bc": 0.029179314151406288, "train_loss_llm": 0.00027814507484436035, "grad_norm": 0.08001089096069336, "global_step": 21000, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03550029546022415, "train_loss_bc": 0.035499412566423416, "train_loss_llm": 8.842349052429199e-05, "grad_norm": 0.030654698610305786, "global_step": 21001, "epoch": 583, "lr": 0.001} +{"train_loss": 0.0366552509367466, "train_loss_bc": 0.036653511226177216, "train_loss_llm": 0.00017407536506652832, "grad_norm": 0.05104301497340202, "global_step": 21002, "epoch": 583, "lr": 0.001} +{"train_loss": 0.030856436118483543, "train_loss_bc": 0.030856436118483543, "train_loss_llm": 0.0, "grad_norm": 0.059416770935058594, "global_step": 21003, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03440834954380989, "train_loss_bc": 0.03440834954380989, "train_loss_llm": 0.0, "grad_norm": 0.07789524644613266, "global_step": 21004, "epoch": 583, "lr": 0.001} +{"train_loss": 0.031004661694169044, "train_loss_bc": 0.031004661694169044, "train_loss_llm": 0.0, "grad_norm": 0.03132341429591179, "global_step": 21005, "epoch": 583, "lr": 0.001} +{"train_loss": 0.029928546398878098, "train_loss_bc": 0.029928546398878098, "train_loss_llm": 0.0, "grad_norm": 0.05608561262488365, "global_step": 21006, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03849558159708977, "train_loss_bc": 0.03848733752965927, "train_loss_llm": 0.0008243322372436523, "grad_norm": 0.07435537874698639, "global_step": 21007, "epoch": 583, "lr": 0.001} +{"train_loss": 0.030659832060337067, "train_loss_bc": 0.030659832060337067, "train_loss_llm": 0.0, "grad_norm": 0.0970921739935875, "global_step": 21008, "epoch": 583, "lr": 0.001} +{"train_loss": 0.0384950265288353, "train_loss_bc": 0.038464318960905075, "train_loss_llm": 0.0030707120895385742, "grad_norm": 0.02867945097386837, "global_step": 21009, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03269048407673836, "train_loss_bc": 0.03269048407673836, "train_loss_llm": 0.0, "grad_norm": 0.048673614859580994, "global_step": 21010, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03955816477537155, "train_loss_bc": 0.039548393338918686, "train_loss_llm": 0.0009772181510925293, "grad_norm": 0.07191459834575653, "global_step": 21011, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03451720252633095, "train_loss_bc": 0.03451720252633095, "train_loss_llm": 0.0, "grad_norm": 0.08652043342590332, "global_step": 21012, "epoch": 583, "lr": 0.001} +{"train_loss": 0.031170900911092758, "train_loss_bc": 0.031170900911092758, "train_loss_llm": 0.0, "grad_norm": 0.025291675701737404, "global_step": 21013, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03306131809949875, "train_loss_bc": 0.03306131809949875, "train_loss_llm": 0.0, "grad_norm": 0.03719368204474449, "global_step": 21014, "epoch": 583, "lr": 0.001} +{"train_loss": 0.030989551916718483, "train_loss_bc": 0.030989551916718483, "train_loss_llm": 0.0, "grad_norm": 0.045926764607429504, "global_step": 21015, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03320995718240738, "train_loss_bc": 0.03320995718240738, "train_loss_llm": 0.0, "grad_norm": 0.05926930159330368, "global_step": 21016, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03701344504952431, "train_loss_bc": 0.03701161593198776, "train_loss_llm": 0.00018292665481567383, "grad_norm": 0.02841363102197647, "global_step": 21017, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03476003557443619, "train_loss_bc": 0.03476003557443619, "train_loss_llm": 0.0, "grad_norm": 0.053339000791311264, "global_step": 21018, "epoch": 583, "lr": 0.001} +{"train_loss": 0.02684410661458969, "train_loss_bc": 0.02684016339480877, "train_loss_llm": 0.00039437413215637207, "grad_norm": 0.068557508289814, "global_step": 21019, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03230581805109978, "train_loss_bc": 0.03230581805109978, "train_loss_llm": 0.0, "grad_norm": 0.0979626327753067, "global_step": 21020, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03647109121084213, "train_loss_bc": 0.03647109121084213, "train_loss_llm": 0.0, "grad_norm": 0.030826430767774582, "global_step": 21021, "epoch": 583, "lr": 0.001} +{"train_loss": 0.030744293704628944, "train_loss_bc": 0.03073505312204361, "train_loss_llm": 0.0009240806102752686, "grad_norm": 0.04790119454264641, "global_step": 21022, "epoch": 583, "lr": 0.001} +{"train_loss": 0.03285937890824345, "train_loss_bc": 0.032272376120090485, "train_loss_llm": 0.0, "grad_norm": 0.06535834819078445, "global_step": 21023, "epoch": 583, "lr": 0.001, "val_loss": 0.08064398169517517} +{"train_loss": 0.03533507511019707, "train_loss_bc": 0.03532636910676956, "train_loss_llm": 0.0008704662322998047, "grad_norm": 0.08391246199607849, "global_step": 21024, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03144945204257965, "train_loss_bc": 0.03143348917365074, "train_loss_llm": 0.0015961825847625732, "grad_norm": 0.02606479823589325, "global_step": 21025, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03721783310174942, "train_loss_bc": 0.03721369802951813, "train_loss_llm": 0.00041344761848449707, "grad_norm": 0.049001339823007584, "global_step": 21026, "epoch": 584, "lr": 0.001} +{"train_loss": 0.036544885486364365, "train_loss_bc": 0.036544885486364365, "train_loss_llm": 0.0, "grad_norm": 0.07540017366409302, "global_step": 21027, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03166475519537926, "train_loss_bc": 0.03166475519537926, "train_loss_llm": 0.0, "grad_norm": 0.09809352457523346, "global_step": 21028, "epoch": 584, "lr": 0.001} +{"train_loss": 0.026425549760460854, "train_loss_bc": 0.026425549760460854, "train_loss_llm": 0.0, "grad_norm": 0.030374085530638695, "global_step": 21029, "epoch": 584, "lr": 0.001} +{"train_loss": 0.030745459720492363, "train_loss_bc": 0.030745459720492363, "train_loss_llm": 0.0, "grad_norm": 0.04326696693897247, "global_step": 21030, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03367491066455841, "train_loss_bc": 0.03367491066455841, "train_loss_llm": 0.0, "grad_norm": 0.048626694828271866, "global_step": 21031, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03776158019900322, "train_loss_bc": 0.03776158019900322, "train_loss_llm": 0.0, "grad_norm": 0.06486185640096664, "global_step": 21032, "epoch": 584, "lr": 0.001} +{"train_loss": 0.022984813898801804, "train_loss_bc": 0.022984813898801804, "train_loss_llm": 0.0, "grad_norm": 0.023483887314796448, "global_step": 21033, "epoch": 584, "lr": 0.001} +{"train_loss": 0.028442105278372765, "train_loss_bc": 0.028442105278372765, "train_loss_llm": 0.0, "grad_norm": 0.03637078031897545, "global_step": 21034, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03589995205402374, "train_loss_bc": 0.03587542846798897, "train_loss_llm": 0.002452254295349121, "grad_norm": 0.04650066792964935, "global_step": 21035, "epoch": 584, "lr": 0.001} +{"train_loss": 0.032191433012485504, "train_loss_bc": 0.032191433012485504, "train_loss_llm": 0.0, "grad_norm": 0.05871051922440529, "global_step": 21036, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03119828552007675, "train_loss_bc": 0.03119097650051117, "train_loss_llm": 0.0007309317588806152, "grad_norm": 0.027454614639282227, "global_step": 21037, "epoch": 584, "lr": 0.001} +{"train_loss": 0.033891163766384125, "train_loss_bc": 0.033891163766384125, "train_loss_llm": 0.0, "grad_norm": 0.04680197313427925, "global_step": 21038, "epoch": 584, "lr": 0.001} +{"train_loss": 0.029152914881706238, "train_loss_bc": 0.029152914881706238, "train_loss_llm": 0.0, "grad_norm": 0.05784974247217178, "global_step": 21039, "epoch": 584, "lr": 0.001} +{"train_loss": 0.028969189152121544, "train_loss_bc": 0.028969189152121544, "train_loss_llm": 0.0, "grad_norm": 0.0710458755493164, "global_step": 21040, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03692823275923729, "train_loss_bc": 0.03692823275923729, "train_loss_llm": 0.0, "grad_norm": 0.02690211310982704, "global_step": 21041, "epoch": 584, "lr": 0.001} +{"train_loss": 0.027945350855588913, "train_loss_bc": 0.027945350855588913, "train_loss_llm": 0.0, "grad_norm": 0.039074450731277466, "global_step": 21042, "epoch": 584, "lr": 0.001} +{"train_loss": 0.032596439123153687, "train_loss_bc": 0.032596439123153687, "train_loss_llm": 0.0, "grad_norm": 0.05473364144563675, "global_step": 21043, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03852496296167374, "train_loss_bc": 0.03852496296167374, "train_loss_llm": 0.0, "grad_norm": 0.07265771180391312, "global_step": 21044, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03714033216238022, "train_loss_bc": 0.037139978259801865, "train_loss_llm": 3.5315752029418945e-05, "grad_norm": 0.029695112258195877, "global_step": 21045, "epoch": 584, "lr": 0.001} +{"train_loss": 0.038906026631593704, "train_loss_bc": 0.038906026631593704, "train_loss_llm": 0.0, "grad_norm": 0.06080983579158783, "global_step": 21046, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03405025228857994, "train_loss_bc": 0.03405025228857994, "train_loss_llm": 0.0, "grad_norm": 0.08601807802915573, "global_step": 21047, "epoch": 584, "lr": 0.001} +{"train_loss": 0.028929423540830612, "train_loss_bc": 0.028929423540830612, "train_loss_llm": 0.0, "grad_norm": 0.10380919277667999, "global_step": 21048, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03275730460882187, "train_loss_bc": 0.032750871032476425, "train_loss_llm": 0.0006431937217712402, "grad_norm": 0.028893565759062767, "global_step": 21049, "epoch": 584, "lr": 0.001} +{"train_loss": 0.035860687494277954, "train_loss_bc": 0.035860687494277954, "train_loss_llm": 0.0, "grad_norm": 0.05074068531394005, "global_step": 21050, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03194470331072807, "train_loss_bc": 0.03194470331072807, "train_loss_llm": 0.0, "grad_norm": 0.06487181782722473, "global_step": 21051, "epoch": 584, "lr": 0.001} +{"train_loss": 0.032159075140953064, "train_loss_bc": 0.03215783089399338, "train_loss_llm": 0.00012430548667907715, "grad_norm": 0.08955688029527664, "global_step": 21052, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03405655920505524, "train_loss_bc": 0.03405655920505524, "train_loss_llm": 0.0, "grad_norm": 0.031835637986660004, "global_step": 21053, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03150465339422226, "train_loss_bc": 0.03150465339422226, "train_loss_llm": 0.0, "grad_norm": 0.0419740155339241, "global_step": 21054, "epoch": 584, "lr": 0.001} +{"train_loss": 0.02830781415104866, "train_loss_bc": 0.028301117941737175, "train_loss_llm": 0.0006696581840515137, "grad_norm": 0.0547853522002697, "global_step": 21055, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03399351239204407, "train_loss_bc": 0.03399351239204407, "train_loss_llm": 0.0, "grad_norm": 0.06004782393574715, "global_step": 21056, "epoch": 584, "lr": 0.001} +{"train_loss": 0.039080660790205, "train_loss_bc": 0.039073992520570755, "train_loss_llm": 0.0006667673587799072, "grad_norm": 0.03466358780860901, "global_step": 21057, "epoch": 584, "lr": 0.001} +{"train_loss": 0.035997577011585236, "train_loss_bc": 0.03597664088010788, "train_loss_llm": 0.0020935237407684326, "grad_norm": 0.04867083206772804, "global_step": 21058, "epoch": 584, "lr": 0.001} +{"train_loss": 0.03280394757166505, "train_loss_bc": 0.026709185913205147, "train_loss_llm": 0.0, "grad_norm": 0.06266012787818909, "global_step": 21059, "epoch": 584, "lr": 0.001, "val_loss": 0.0802910178899765} +{"train_loss": 0.03709682077169418, "train_loss_bc": 0.03709682077169418, "train_loss_llm": 0.0, "grad_norm": 0.07646995782852173, "global_step": 21060, "epoch": 585, "lr": 0.001} +{"train_loss": 0.030614027753472328, "train_loss_bc": 0.030614027753472328, "train_loss_llm": 0.0, "grad_norm": 0.03034193441271782, "global_step": 21061, "epoch": 585, "lr": 0.001} +{"train_loss": 0.032586719840765, "train_loss_bc": 0.032584529370069504, "train_loss_llm": 0.00021907687187194824, "grad_norm": 0.049411576241254807, "global_step": 21062, "epoch": 585, "lr": 0.001} +{"train_loss": 0.034269705414772034, "train_loss_bc": 0.034259893000125885, "train_loss_llm": 0.0009810924530029297, "grad_norm": 0.06803051382303238, "global_step": 21063, "epoch": 585, "lr": 0.001} +{"train_loss": 0.0327925868332386, "train_loss_bc": 0.0327925868332386, "train_loss_llm": 0.0, "grad_norm": 0.07863157987594604, "global_step": 21064, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03391189128160477, "train_loss_bc": 0.03391189128160477, "train_loss_llm": 0.0, "grad_norm": 0.02960287034511566, "global_step": 21065, "epoch": 585, "lr": 0.001} +{"train_loss": 0.032441396266222, "train_loss_bc": 0.03243328258395195, "train_loss_llm": 0.0008114278316497803, "grad_norm": 0.050073932856321335, "global_step": 21066, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03234333544969559, "train_loss_bc": 0.03234333544969559, "train_loss_llm": 0.0, "grad_norm": 0.06451358646154404, "global_step": 21067, "epoch": 585, "lr": 0.001} +{"train_loss": 0.027562178671360016, "train_loss_bc": 0.027549924328923225, "train_loss_llm": 0.0012254416942596436, "grad_norm": 0.08083714544773102, "global_step": 21068, "epoch": 585, "lr": 0.001} +{"train_loss": 0.028744246810674667, "train_loss_bc": 0.028744246810674667, "train_loss_llm": 0.0, "grad_norm": 0.02560974471271038, "global_step": 21069, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03813864290714264, "train_loss_bc": 0.03813864290714264, "train_loss_llm": 0.0, "grad_norm": 0.0530872605741024, "global_step": 21070, "epoch": 585, "lr": 0.001} +{"train_loss": 0.032206594944000244, "train_loss_bc": 0.032206594944000244, "train_loss_llm": 0.0, "grad_norm": 0.07385413348674774, "global_step": 21071, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03262271359562874, "train_loss_bc": 0.03262271359562874, "train_loss_llm": 0.0, "grad_norm": 0.08527074754238129, "global_step": 21072, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03519773110747337, "train_loss_bc": 0.03519773110747337, "train_loss_llm": 0.0, "grad_norm": 0.02951723337173462, "global_step": 21073, "epoch": 585, "lr": 0.001} +{"train_loss": 0.032479360699653625, "train_loss_bc": 0.032479360699653625, "train_loss_llm": 0.0, "grad_norm": 0.04977975785732269, "global_step": 21074, "epoch": 585, "lr": 0.001} +{"train_loss": 0.028543150052428246, "train_loss_bc": 0.02853899635374546, "train_loss_llm": 0.0004152953624725342, "grad_norm": 0.05553237348794937, "global_step": 21075, "epoch": 585, "lr": 0.001} +{"train_loss": 0.031257979571819305, "train_loss_bc": 0.031257979571819305, "train_loss_llm": 0.0, "grad_norm": 0.06488274782896042, "global_step": 21076, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03527276962995529, "train_loss_bc": 0.0352616012096405, "train_loss_llm": 0.0011169910430908203, "grad_norm": 0.02631426230072975, "global_step": 21077, "epoch": 585, "lr": 0.001} +{"train_loss": 0.034272450953722, "train_loss_bc": 0.034272450953722, "train_loss_llm": 0.0, "grad_norm": 0.03616529703140259, "global_step": 21078, "epoch": 585, "lr": 0.001} +{"train_loss": 0.029060443863272667, "train_loss_bc": 0.029060443863272667, "train_loss_llm": 0.0, "grad_norm": 0.047023072838783264, "global_step": 21079, "epoch": 585, "lr": 0.001} +{"train_loss": 0.0324615053832531, "train_loss_bc": 0.032459359616041183, "train_loss_llm": 0.00021454691886901855, "grad_norm": 0.055235955864191055, "global_step": 21080, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03666968271136284, "train_loss_bc": 0.03666968271136284, "train_loss_llm": 0.0, "grad_norm": 0.024743562564253807, "global_step": 21081, "epoch": 585, "lr": 0.001} +{"train_loss": 0.02412683703005314, "train_loss_bc": 0.02412683703005314, "train_loss_llm": 0.0, "grad_norm": 0.038565948605537415, "global_step": 21082, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03521043062210083, "train_loss_bc": 0.03521043062210083, "train_loss_llm": 0.0, "grad_norm": 0.04735949635505676, "global_step": 21083, "epoch": 585, "lr": 0.001} +{"train_loss": 0.035920944064855576, "train_loss_bc": 0.03590485826134682, "train_loss_llm": 0.0016086995601654053, "grad_norm": 0.058999352157115936, "global_step": 21084, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03180401399731636, "train_loss_bc": 0.03180401399731636, "train_loss_llm": 0.0, "grad_norm": 0.033485520631074905, "global_step": 21085, "epoch": 585, "lr": 0.001} +{"train_loss": 0.031134024262428284, "train_loss_bc": 0.031134024262428284, "train_loss_llm": 0.0, "grad_norm": 0.052396394312381744, "global_step": 21086, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03345005214214325, "train_loss_bc": 0.03345005214214325, "train_loss_llm": 0.0, "grad_norm": 0.06720058619976044, "global_step": 21087, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03758348152041435, "train_loss_bc": 0.03758348152041435, "train_loss_llm": 0.0, "grad_norm": 0.0840587392449379, "global_step": 21088, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03407804295420647, "train_loss_bc": 0.03407804295420647, "train_loss_llm": 0.0, "grad_norm": 0.03134516626596451, "global_step": 21089, "epoch": 585, "lr": 0.001} +{"train_loss": 0.034810733050107956, "train_loss_bc": 0.034810733050107956, "train_loss_llm": 0.0, "grad_norm": 0.05351465940475464, "global_step": 21090, "epoch": 585, "lr": 0.001} +{"train_loss": 0.028865549713373184, "train_loss_bc": 0.028865549713373184, "train_loss_llm": 0.0, "grad_norm": 0.07383152842521667, "global_step": 21091, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03364557772874832, "train_loss_bc": 0.03363043814897537, "train_loss_llm": 0.0015140175819396973, "grad_norm": 0.10025567561388016, "global_step": 21092, "epoch": 585, "lr": 0.001} +{"train_loss": 0.028171826153993607, "train_loss_bc": 0.028171826153993607, "train_loss_llm": 0.0, "grad_norm": 0.027477199211716652, "global_step": 21093, "epoch": 585, "lr": 0.001} +{"train_loss": 0.036832232028245926, "train_loss_bc": 0.03682733699679375, "train_loss_llm": 0.0004894137382507324, "grad_norm": 0.05564509332180023, "global_step": 21094, "epoch": 585, "lr": 0.001} +{"train_loss": 0.03283687587827444, "train_loss_bc": 0.03593596816062927, "train_loss_llm": 0.0011883378028869629, "grad_norm": 0.08801009505987167, "global_step": 21095, "epoch": 585, "lr": 0.001, "train/cumulative_reward": 93.99324395633289, "train/mean_score": 9.623043518809519, "train/success_rate": 0.2, "test/cumulative_reward": 46.808943169439196, "test/mean_score": 7.446606015550349, "test/success_rate": 0.16, "val_loss": 0.08276726305484772, "train_action_mse_error": 0.03563021495938301} +{"train_loss": 0.03377129137516022, "train_loss_bc": 0.03377129137516022, "train_loss_llm": 0.0, "grad_norm": 0.10801680386066437, "global_step": 21096, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03998717665672302, "train_loss_bc": 0.03998717665672302, "train_loss_llm": 0.0, "grad_norm": 0.037358518689870834, "global_step": 21097, "epoch": 586, "lr": 0.001} +{"train_loss": 0.031024683266878128, "train_loss_bc": 0.031024683266878128, "train_loss_llm": 0.0, "grad_norm": 0.046651508659124374, "global_step": 21098, "epoch": 586, "lr": 0.001} +{"train_loss": 0.032750070095062256, "train_loss_bc": 0.032750070095062256, "train_loss_llm": 0.0, "grad_norm": 0.05937466397881508, "global_step": 21099, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03152771294116974, "train_loss_bc": 0.03152771294116974, "train_loss_llm": 0.0, "grad_norm": 0.071633480489254, "global_step": 21100, "epoch": 586, "lr": 0.001} +{"train_loss": 0.030624331906437874, "train_loss_bc": 0.030620109289884567, "train_loss_llm": 0.0004222393035888672, "grad_norm": 0.029723219573497772, "global_step": 21101, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03392835706472397, "train_loss_bc": 0.03392835706472397, "train_loss_llm": 0.0, "grad_norm": 0.04875922575592995, "global_step": 21102, "epoch": 586, "lr": 0.001} +{"train_loss": 0.02605314552783966, "train_loss_bc": 0.026043463498353958, "train_loss_llm": 0.000968247652053833, "grad_norm": 0.06017115339636803, "global_step": 21103, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03455377742648125, "train_loss_bc": 0.03455377742648125, "train_loss_llm": 0.0, "grad_norm": 0.08160977065563202, "global_step": 21104, "epoch": 586, "lr": 0.001} +{"train_loss": 0.036426842212677, "train_loss_bc": 0.036411352455616, "train_loss_llm": 0.0015487968921661377, "grad_norm": 0.023700939491391182, "global_step": 21105, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03456956893205643, "train_loss_bc": 0.03456956893205643, "train_loss_llm": 0.0, "grad_norm": 0.03904372453689575, "global_step": 21106, "epoch": 586, "lr": 0.001} +{"train_loss": 0.029905308037996292, "train_loss_bc": 0.029905308037996292, "train_loss_llm": 0.0, "grad_norm": 0.05005548521876335, "global_step": 21107, "epoch": 586, "lr": 0.001} +{"train_loss": 0.0331389494240284, "train_loss_bc": 0.0331389494240284, "train_loss_llm": 0.0, "grad_norm": 0.06364452093839645, "global_step": 21108, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03151771053671837, "train_loss_bc": 0.03151771053671837, "train_loss_llm": 0.0, "grad_norm": 0.02921006642282009, "global_step": 21109, "epoch": 586, "lr": 0.001} +{"train_loss": 0.032150235027074814, "train_loss_bc": 0.032150235027074814, "train_loss_llm": 0.0, "grad_norm": 0.048551108688116074, "global_step": 21110, "epoch": 586, "lr": 0.001} +{"train_loss": 0.0317278727889061, "train_loss_bc": 0.0317278727889061, "train_loss_llm": 0.0, "grad_norm": 0.06937001645565033, "global_step": 21111, "epoch": 586, "lr": 0.001} +{"train_loss": 0.036889370530843735, "train_loss_bc": 0.036889370530843735, "train_loss_llm": 0.0, "grad_norm": 0.09011150151491165, "global_step": 21112, "epoch": 586, "lr": 0.001} +{"train_loss": 0.033129025250673294, "train_loss_bc": 0.033129025250673294, "train_loss_llm": 0.0, "grad_norm": 0.025906391441822052, "global_step": 21113, "epoch": 586, "lr": 0.001} +{"train_loss": 0.036191653460264206, "train_loss_bc": 0.036191653460264206, "train_loss_llm": 0.0, "grad_norm": 0.042976584285497665, "global_step": 21114, "epoch": 586, "lr": 0.001} +{"train_loss": 0.02898266911506653, "train_loss_bc": 0.028982574120163918, "train_loss_llm": 9.47713851928711e-06, "grad_norm": 0.061262067407369614, "global_step": 21115, "epoch": 586, "lr": 0.001} +{"train_loss": 0.035068579018116, "train_loss_bc": 0.035068579018116, "train_loss_llm": 0.0, "grad_norm": 0.08546249568462372, "global_step": 21116, "epoch": 586, "lr": 0.001} +{"train_loss": 0.030375594273209572, "train_loss_bc": 0.030375594273209572, "train_loss_llm": 0.0, "grad_norm": 0.025359489023685455, "global_step": 21117, "epoch": 586, "lr": 0.001} +{"train_loss": 0.031107453629374504, "train_loss_bc": 0.031095482409000397, "train_loss_llm": 0.00119704008102417, "grad_norm": 0.04135233536362648, "global_step": 21118, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03215877711772919, "train_loss_bc": 0.03215877711772919, "train_loss_llm": 0.0, "grad_norm": 0.05574818700551987, "global_step": 21119, "epoch": 586, "lr": 0.001} +{"train_loss": 0.029955832287669182, "train_loss_bc": 0.029955832287669182, "train_loss_llm": 0.0, "grad_norm": 0.07517629116773605, "global_step": 21120, "epoch": 586, "lr": 0.001} +{"train_loss": 0.028638143092393875, "train_loss_bc": 0.02862132526934147, "train_loss_llm": 0.0016817748546600342, "grad_norm": 0.02751130424439907, "global_step": 21121, "epoch": 586, "lr": 0.001} +{"train_loss": 0.034026578068733215, "train_loss_bc": 0.034026578068733215, "train_loss_llm": 0.0, "grad_norm": 0.04839450493454933, "global_step": 21122, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03512830287218094, "train_loss_bc": 0.035127487033605576, "train_loss_llm": 8.162856101989746e-05, "grad_norm": 0.06748031079769135, "global_step": 21123, "epoch": 586, "lr": 0.001} +{"train_loss": 0.028595510870218277, "train_loss_bc": 0.028595510870218277, "train_loss_llm": 0.0, "grad_norm": 0.09138202667236328, "global_step": 21124, "epoch": 586, "lr": 0.001} +{"train_loss": 0.035694386810064316, "train_loss_bc": 0.035694386810064316, "train_loss_llm": 0.0, "grad_norm": 0.03063732199370861, "global_step": 21125, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03398837149143219, "train_loss_bc": 0.033983319997787476, "train_loss_llm": 0.0005049705505371094, "grad_norm": 0.05603747442364693, "global_step": 21126, "epoch": 586, "lr": 0.001} +{"train_loss": 0.035864997655153275, "train_loss_bc": 0.0358472615480423, "train_loss_llm": 0.001773744821548462, "grad_norm": 0.07488708198070526, "global_step": 21127, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03273759037256241, "train_loss_bc": 0.03273759037256241, "train_loss_llm": 0.0, "grad_norm": 0.08693192899227142, "global_step": 21128, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03538672998547554, "train_loss_bc": 0.03538672998547554, "train_loss_llm": 0.0, "grad_norm": 0.025508636608719826, "global_step": 21129, "epoch": 586, "lr": 0.001} +{"train_loss": 0.03209472447633743, "train_loss_bc": 0.03207894414663315, "train_loss_llm": 0.0015780925750732422, "grad_norm": 0.03804149106144905, "global_step": 21130, "epoch": 586, "lr": 0.001} +{"train_loss": 0.032853949090672865, "train_loss_bc": 0.033070843666791916, "train_loss_llm": 0.0, "grad_norm": 0.05513722077012062, "global_step": 21131, "epoch": 586, "lr": 0.001, "val_loss": 0.08034655451774597} +{"train_loss": 0.03413459286093712, "train_loss_bc": 0.03413459286093712, "train_loss_llm": 0.0, "grad_norm": 0.06967783719301224, "global_step": 21132, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03680901974439621, "train_loss_bc": 0.03680901974439621, "train_loss_llm": 0.0, "grad_norm": 0.03756135329604149, "global_step": 21133, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03225007653236389, "train_loss_bc": 0.03225007653236389, "train_loss_llm": 0.0, "grad_norm": 0.050380077213048935, "global_step": 21134, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03370044007897377, "train_loss_bc": 0.03370044007897377, "train_loss_llm": 0.0, "grad_norm": 0.0678211897611618, "global_step": 21135, "epoch": 587, "lr": 0.001} +{"train_loss": 0.04126956686377525, "train_loss_bc": 0.04126956686377525, "train_loss_llm": 0.0, "grad_norm": 0.07633579522371292, "global_step": 21136, "epoch": 587, "lr": 0.001} +{"train_loss": 0.030804449692368507, "train_loss_bc": 0.030804449692368507, "train_loss_llm": 0.0, "grad_norm": 0.02984491176903248, "global_step": 21137, "epoch": 587, "lr": 0.001} +{"train_loss": 0.0410919114947319, "train_loss_bc": 0.04108484834432602, "train_loss_llm": 0.0007061362266540527, "grad_norm": 0.05687562748789787, "global_step": 21138, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03030080907046795, "train_loss_bc": 0.03030080907046795, "train_loss_llm": 0.0, "grad_norm": 0.06717512011528015, "global_step": 21139, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03326891362667084, "train_loss_bc": 0.03326891362667084, "train_loss_llm": 0.0, "grad_norm": 0.0759422779083252, "global_step": 21140, "epoch": 587, "lr": 0.001} +{"train_loss": 0.033404380083084106, "train_loss_bc": 0.033404380083084106, "train_loss_llm": 0.0, "grad_norm": 0.030336707830429077, "global_step": 21141, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03410610556602478, "train_loss_bc": 0.03410610556602478, "train_loss_llm": 0.0, "grad_norm": 0.060140591114759445, "global_step": 21142, "epoch": 587, "lr": 0.001} +{"train_loss": 0.0312185175716877, "train_loss_bc": 0.031216328963637352, "train_loss_llm": 0.0002187788486480713, "grad_norm": 0.07802842557430267, "global_step": 21143, "epoch": 587, "lr": 0.001} +{"train_loss": 0.0343158096075058, "train_loss_bc": 0.0343158096075058, "train_loss_llm": 0.0, "grad_norm": 0.09693238884210587, "global_step": 21144, "epoch": 587, "lr": 0.001} +{"train_loss": 0.027707692235708237, "train_loss_bc": 0.027707692235708237, "train_loss_llm": 0.0, "grad_norm": 0.024799970909953117, "global_step": 21145, "epoch": 587, "lr": 0.001} +{"train_loss": 0.034591902047395706, "train_loss_bc": 0.034591902047395706, "train_loss_llm": 0.0, "grad_norm": 0.050285521894693375, "global_step": 21146, "epoch": 587, "lr": 0.001} +{"train_loss": 0.029618332162499428, "train_loss_bc": 0.029589688405394554, "train_loss_llm": 0.002864331007003784, "grad_norm": 0.06452714651823044, "global_step": 21147, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03288082405924797, "train_loss_bc": 0.03288082405924797, "train_loss_llm": 0.0, "grad_norm": 0.07343118637800217, "global_step": 21148, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03500118479132652, "train_loss_bc": 0.03500118479132652, "train_loss_llm": 0.0, "grad_norm": 0.031322743743658066, "global_step": 21149, "epoch": 587, "lr": 0.001} +{"train_loss": 0.036553945392370224, "train_loss_bc": 0.036553945392370224, "train_loss_llm": 0.0, "grad_norm": 0.0462145172059536, "global_step": 21150, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03319603204727173, "train_loss_bc": 0.03319169580936432, "train_loss_llm": 0.0004335939884185791, "grad_norm": 0.06721264868974686, "global_step": 21151, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03239923343062401, "train_loss_bc": 0.03235514834523201, "train_loss_llm": 0.004408329725265503, "grad_norm": 0.08227693289518356, "global_step": 21152, "epoch": 587, "lr": 0.001} +{"train_loss": 0.037054941058158875, "train_loss_bc": 0.037054941058158875, "train_loss_llm": 0.0, "grad_norm": 0.033784009516239166, "global_step": 21153, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03631759062409401, "train_loss_bc": 0.03630039840936661, "train_loss_llm": 0.0017192065715789795, "grad_norm": 0.05142471194267273, "global_step": 21154, "epoch": 587, "lr": 0.001} +{"train_loss": 0.033280566334724426, "train_loss_bc": 0.033280566334724426, "train_loss_llm": 0.0, "grad_norm": 0.07334724068641663, "global_step": 21155, "epoch": 587, "lr": 0.001} +{"train_loss": 0.02520640753209591, "train_loss_bc": 0.02520640753209591, "train_loss_llm": 0.0, "grad_norm": 0.08318323642015457, "global_step": 21156, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03206612542271614, "train_loss_bc": 0.032051652669906616, "train_loss_llm": 0.0014472603797912598, "grad_norm": 0.025219110772013664, "global_step": 21157, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03127208724617958, "train_loss_bc": 0.03127208724617958, "train_loss_llm": 0.0, "grad_norm": 0.04511471837759018, "global_step": 21158, "epoch": 587, "lr": 0.001} +{"train_loss": 0.030525309965014458, "train_loss_bc": 0.030525309965014458, "train_loss_llm": 0.0, "grad_norm": 0.05590279400348663, "global_step": 21159, "epoch": 587, "lr": 0.001} +{"train_loss": 0.024257486686110497, "train_loss_bc": 0.024257486686110497, "train_loss_llm": 0.0, "grad_norm": 0.06453199684619904, "global_step": 21160, "epoch": 587, "lr": 0.001} +{"train_loss": 0.031088368967175484, "train_loss_bc": 0.031088368967175484, "train_loss_llm": 0.0, "grad_norm": 0.02722794935107231, "global_step": 21161, "epoch": 587, "lr": 0.001} +{"train_loss": 0.027563560754060745, "train_loss_bc": 0.027563560754060745, "train_loss_llm": 0.0, "grad_norm": 0.044964749366045, "global_step": 21162, "epoch": 587, "lr": 0.001} +{"train_loss": 0.031032243743538857, "train_loss_bc": 0.03103134036064148, "train_loss_llm": 9.036064147949219e-05, "grad_norm": 0.05446091666817665, "global_step": 21163, "epoch": 587, "lr": 0.001} +{"train_loss": 0.027492905035614967, "train_loss_bc": 0.02748981863260269, "train_loss_llm": 0.00030869245529174805, "grad_norm": 0.0700409859418869, "global_step": 21164, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03863464295864105, "train_loss_bc": 0.03863464295864105, "train_loss_llm": 0.0, "grad_norm": 0.03805431351065636, "global_step": 21165, "epoch": 587, "lr": 0.001} +{"train_loss": 0.035397984087467194, "train_loss_bc": 0.035397984087467194, "train_loss_llm": 0.0, "grad_norm": 0.06466402858495712, "global_step": 21166, "epoch": 587, "lr": 0.001} +{"train_loss": 0.03286370691946811, "train_loss_bc": 0.03327948972582817, "train_loss_llm": 0.0, "grad_norm": 0.08349590748548508, "global_step": 21167, "epoch": 587, "lr": 0.001, "val_loss": 0.08316807448863983} +{"train_loss": 0.034228499978780746, "train_loss_bc": 0.034228499978780746, "train_loss_llm": 0.0, "grad_norm": 0.09942736476659775, "global_step": 21168, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03689814731478691, "train_loss_bc": 0.03689267113804817, "train_loss_llm": 0.0005474686622619629, "grad_norm": 0.03156457841396332, "global_step": 21169, "epoch": 588, "lr": 0.001} +{"train_loss": 0.031246758997440338, "train_loss_bc": 0.031246758997440338, "train_loss_llm": 0.0, "grad_norm": 0.045198604464530945, "global_step": 21170, "epoch": 588, "lr": 0.001} +{"train_loss": 0.031508974730968475, "train_loss_bc": 0.031508974730968475, "train_loss_llm": 0.0, "grad_norm": 0.06132609024643898, "global_step": 21171, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03179825097322464, "train_loss_bc": 0.03179825097322464, "train_loss_llm": 0.0, "grad_norm": 0.07068083435297012, "global_step": 21172, "epoch": 588, "lr": 0.001} +{"train_loss": 0.029148738831281662, "train_loss_bc": 0.029148738831281662, "train_loss_llm": 0.0, "grad_norm": 0.0325557142496109, "global_step": 21173, "epoch": 588, "lr": 0.001} +{"train_loss": 0.035366665571928024, "train_loss_bc": 0.035366665571928024, "train_loss_llm": 0.0, "grad_norm": 0.048556145280599594, "global_step": 21174, "epoch": 588, "lr": 0.001} +{"train_loss": 0.037234317511320114, "train_loss_bc": 0.037234317511320114, "train_loss_llm": 0.0, "grad_norm": 0.06306523829698563, "global_step": 21175, "epoch": 588, "lr": 0.001} +{"train_loss": 0.029502758756279945, "train_loss_bc": 0.029471414163708687, "train_loss_llm": 0.00313451886177063, "grad_norm": 0.08025353401899338, "global_step": 21176, "epoch": 588, "lr": 0.001} +{"train_loss": 0.035806380212306976, "train_loss_bc": 0.035806380212306976, "train_loss_llm": 0.0, "grad_norm": 0.03137883171439171, "global_step": 21177, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03772396594285965, "train_loss_bc": 0.03772396594285965, "train_loss_llm": 0.0, "grad_norm": 0.051639098674058914, "global_step": 21178, "epoch": 588, "lr": 0.001} +{"train_loss": 0.029242131859064102, "train_loss_bc": 0.029238909482955933, "train_loss_llm": 0.0003222227096557617, "grad_norm": 0.06393127143383026, "global_step": 21179, "epoch": 588, "lr": 0.001} +{"train_loss": 0.027765726670622826, "train_loss_bc": 0.027765726670622826, "train_loss_llm": 0.0, "grad_norm": 0.08054052293300629, "global_step": 21180, "epoch": 588, "lr": 0.001} +{"train_loss": 0.0372699573636055, "train_loss_bc": 0.0372699573636055, "train_loss_llm": 0.0, "grad_norm": 0.039184167981147766, "global_step": 21181, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03373673930764198, "train_loss_bc": 0.03373673930764198, "train_loss_llm": 0.0, "grad_norm": 0.059477537870407104, "global_step": 21182, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03726630285382271, "train_loss_bc": 0.03726630285382271, "train_loss_llm": 0.0, "grad_norm": 0.07695749402046204, "global_step": 21183, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03255545720458031, "train_loss_bc": 0.03255545720458031, "train_loss_llm": 0.0, "grad_norm": 0.09618877619504929, "global_step": 21184, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03262787684798241, "train_loss_bc": 0.03262787684798241, "train_loss_llm": 0.0, "grad_norm": 0.03065524436533451, "global_step": 21185, "epoch": 588, "lr": 0.001} +{"train_loss": 0.030652852728962898, "train_loss_bc": 0.030652852728962898, "train_loss_llm": 0.0, "grad_norm": 0.045939020812511444, "global_step": 21186, "epoch": 588, "lr": 0.001} +{"train_loss": 0.029224257916212082, "train_loss_bc": 0.029196063056588173, "train_loss_llm": 0.0028194785118103027, "grad_norm": 0.06046903878450394, "global_step": 21187, "epoch": 588, "lr": 0.001} +{"train_loss": 0.034686069935560226, "train_loss_bc": 0.03467436134815216, "train_loss_llm": 0.0011706948280334473, "grad_norm": 0.07875698059797287, "global_step": 21188, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03909005969762802, "train_loss_bc": 0.03909005969762802, "train_loss_llm": 0.0, "grad_norm": 0.03680751100182533, "global_step": 21189, "epoch": 588, "lr": 0.001} +{"train_loss": 0.028207233175635338, "train_loss_bc": 0.028207233175635338, "train_loss_llm": 0.0, "grad_norm": 0.056289415806531906, "global_step": 21190, "epoch": 588, "lr": 0.001} +{"train_loss": 0.031366899609565735, "train_loss_bc": 0.031366899609565735, "train_loss_llm": 0.0, "grad_norm": 0.07576417177915573, "global_step": 21191, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03462003916501999, "train_loss_bc": 0.03462003916501999, "train_loss_llm": 0.0, "grad_norm": 0.08370755612850189, "global_step": 21192, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03407348319888115, "train_loss_bc": 0.03405658155679703, "train_loss_llm": 0.001690000295639038, "grad_norm": 0.03325475752353668, "global_step": 21193, "epoch": 588, "lr": 0.001} +{"train_loss": 0.030112385749816895, "train_loss_bc": 0.030112385749816895, "train_loss_llm": 0.0, "grad_norm": 0.05943619832396507, "global_step": 21194, "epoch": 588, "lr": 0.001} +{"train_loss": 0.026255980134010315, "train_loss_bc": 0.026255980134010315, "train_loss_llm": 0.0, "grad_norm": 0.07989084720611572, "global_step": 21195, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03445718437433243, "train_loss_bc": 0.03444906324148178, "train_loss_llm": 0.0008122622966766357, "grad_norm": 0.09638611972332001, "global_step": 21196, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03463900461792946, "train_loss_bc": 0.03463603928685188, "train_loss_llm": 0.00029659271240234375, "grad_norm": 0.03065667115151882, "global_step": 21197, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03557584807276726, "train_loss_bc": 0.03557584807276726, "train_loss_llm": 0.0, "grad_norm": 0.06046173349022865, "global_step": 21198, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03342943266034126, "train_loss_bc": 0.03342943266034126, "train_loss_llm": 0.0, "grad_norm": 0.0873168334364891, "global_step": 21199, "epoch": 588, "lr": 0.001} +{"train_loss": 0.028294207528233528, "train_loss_bc": 0.028294207528233528, "train_loss_llm": 0.0, "grad_norm": 0.11050364375114441, "global_step": 21200, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03420853614807129, "train_loss_bc": 0.03420853614807129, "train_loss_llm": 0.0, "grad_norm": 0.03797987475991249, "global_step": 21201, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03473633527755737, "train_loss_bc": 0.03473633527755737, "train_loss_llm": 0.0, "grad_norm": 0.04608628898859024, "global_step": 21202, "epoch": 588, "lr": 0.001} +{"train_loss": 0.03297106999283036, "train_loss_bc": 0.032401058822870255, "train_loss_llm": 0.0, "grad_norm": 0.05865052714943886, "global_step": 21203, "epoch": 588, "lr": 0.001, "val_loss": 0.0802413821220398} +{"train_loss": 0.03354139253497124, "train_loss_bc": 0.03354139253497124, "train_loss_llm": 0.0, "grad_norm": 0.0665372833609581, "global_step": 21204, "epoch": 589, "lr": 0.001} +{"train_loss": 0.034612759947776794, "train_loss_bc": 0.034612759947776794, "train_loss_llm": 0.0, "grad_norm": 0.028903035447001457, "global_step": 21205, "epoch": 589, "lr": 0.001} +{"train_loss": 0.032755427062511444, "train_loss_bc": 0.032755427062511444, "train_loss_llm": 0.0, "grad_norm": 0.04876158758997917, "global_step": 21206, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03415951505303383, "train_loss_bc": 0.03415951505303383, "train_loss_llm": 0.0, "grad_norm": 0.05539647862315178, "global_step": 21207, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03370640054345131, "train_loss_bc": 0.03370640054345131, "train_loss_llm": 0.0, "grad_norm": 0.06542101502418518, "global_step": 21208, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03333150967955589, "train_loss_bc": 0.033309295773506165, "train_loss_llm": 0.0022214651107788086, "grad_norm": 0.031647324562072754, "global_step": 21209, "epoch": 589, "lr": 0.001} +{"train_loss": 0.028134338557720184, "train_loss_bc": 0.028131410479545593, "train_loss_llm": 0.00029274821281433105, "grad_norm": 0.042213015258312225, "global_step": 21210, "epoch": 589, "lr": 0.001} +{"train_loss": 0.028426919132471085, "train_loss_bc": 0.028426919132471085, "train_loss_llm": 0.0, "grad_norm": 0.057000838220119476, "global_step": 21211, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03944776952266693, "train_loss_bc": 0.03944776952266693, "train_loss_llm": 0.0, "grad_norm": 0.06953418254852295, "global_step": 21212, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03213953971862793, "train_loss_bc": 0.03213685378432274, "train_loss_llm": 0.00026866793632507324, "grad_norm": 0.03261323273181915, "global_step": 21213, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03747137635946274, "train_loss_bc": 0.03744248300790787, "train_loss_llm": 0.002889305353164673, "grad_norm": 0.0473364032804966, "global_step": 21214, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03143788129091263, "train_loss_bc": 0.031426843255758286, "train_loss_llm": 0.0011037588119506836, "grad_norm": 0.07066649198532104, "global_step": 21215, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03742083162069321, "train_loss_bc": 0.03742083162069321, "train_loss_llm": 0.0, "grad_norm": 0.07926943153142929, "global_step": 21216, "epoch": 589, "lr": 0.001} +{"train_loss": 0.027739090844988823, "train_loss_bc": 0.027739090844988823, "train_loss_llm": 0.0, "grad_norm": 0.028347210958600044, "global_step": 21217, "epoch": 589, "lr": 0.001} +{"train_loss": 0.034483298659324646, "train_loss_bc": 0.03447221592068672, "train_loss_llm": 0.0011082291603088379, "grad_norm": 0.044451531022787094, "global_step": 21218, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03593406453728676, "train_loss_bc": 0.03593406453728676, "train_loss_llm": 0.0, "grad_norm": 0.06118744984269142, "global_step": 21219, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03207222744822502, "train_loss_bc": 0.03207222744822502, "train_loss_llm": 0.0, "grad_norm": 0.07921268790960312, "global_step": 21220, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03121710754930973, "train_loss_bc": 0.03121710754930973, "train_loss_llm": 0.0, "grad_norm": 0.03353901952505112, "global_step": 21221, "epoch": 589, "lr": 0.001} +{"train_loss": 0.029490895569324493, "train_loss_bc": 0.029490895569324493, "train_loss_llm": 0.0, "grad_norm": 0.050229430198669434, "global_step": 21222, "epoch": 589, "lr": 0.001} +{"train_loss": 0.038100920617580414, "train_loss_bc": 0.038100920617580414, "train_loss_llm": 0.0, "grad_norm": 0.08248574286699295, "global_step": 21223, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03022724948823452, "train_loss_bc": 0.03022724948823452, "train_loss_llm": 0.0, "grad_norm": 0.09808363020420074, "global_step": 21224, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03346771374344826, "train_loss_bc": 0.03345979377627373, "train_loss_llm": 0.0007919371128082275, "grad_norm": 0.030421782284975052, "global_step": 21225, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03614124655723572, "train_loss_bc": 0.03613383322954178, "train_loss_llm": 0.0007412731647491455, "grad_norm": 0.046018343418836594, "global_step": 21226, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03060518205165863, "train_loss_bc": 0.03060518205165863, "train_loss_llm": 0.0, "grad_norm": 0.05539625510573387, "global_step": 21227, "epoch": 589, "lr": 0.001} +{"train_loss": 0.031071145087480545, "train_loss_bc": 0.03106755018234253, "train_loss_llm": 0.00035950541496276855, "grad_norm": 0.061211567372083664, "global_step": 21228, "epoch": 589, "lr": 0.001} +{"train_loss": 0.036709561944007874, "train_loss_bc": 0.03670766204595566, "train_loss_llm": 0.0001901686191558838, "grad_norm": 0.03434574976563454, "global_step": 21229, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03359489142894745, "train_loss_bc": 0.03359489142894745, "train_loss_llm": 0.0, "grad_norm": 0.047774720937013626, "global_step": 21230, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03320174291729927, "train_loss_bc": 0.033199284225702286, "train_loss_llm": 0.0002460479736328125, "grad_norm": 0.06657792627811432, "global_step": 21231, "epoch": 589, "lr": 0.001} +{"train_loss": 0.02372831478714943, "train_loss_bc": 0.023720385506749153, "train_loss_llm": 0.0007929801940917969, "grad_norm": 0.07149876654148102, "global_step": 21232, "epoch": 589, "lr": 0.001} +{"train_loss": 0.034513141959905624, "train_loss_bc": 0.034513141959905624, "train_loss_llm": 0.0, "grad_norm": 0.025244098156690598, "global_step": 21233, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03649686276912689, "train_loss_bc": 0.03649686276912689, "train_loss_llm": 0.0, "grad_norm": 0.04841964319348335, "global_step": 21234, "epoch": 589, "lr": 0.001} +{"train_loss": 0.032333046197891235, "train_loss_bc": 0.0323183499276638, "train_loss_llm": 0.0014696717262268066, "grad_norm": 0.06032561510801315, "global_step": 21235, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03495294973254204, "train_loss_bc": 0.03495294973254204, "train_loss_llm": 0.0, "grad_norm": 0.07955079525709152, "global_step": 21236, "epoch": 589, "lr": 0.001} +{"train_loss": 0.029046593233942986, "train_loss_bc": 0.029046593233942986, "train_loss_llm": 0.0, "grad_norm": 0.03299419581890106, "global_step": 21237, "epoch": 589, "lr": 0.001} +{"train_loss": 0.024421272799372673, "train_loss_bc": 0.024416988715529442, "train_loss_llm": 0.0004284381866455078, "grad_norm": 0.053800325840711594, "global_step": 21238, "epoch": 589, "lr": 0.001} +{"train_loss": 0.03284236659399337, "train_loss_bc": 0.03619101643562317, "train_loss_llm": 0.0, "grad_norm": 0.0856480821967125, "global_step": 21239, "epoch": 589, "lr": 0.001, "val_loss": 0.08239313215017319} +{"train_loss": 0.033490102738142014, "train_loss_bc": 0.033490102738142014, "train_loss_llm": 0.0, "grad_norm": 0.11062609404325485, "global_step": 21240, "epoch": 590, "lr": 0.001} +{"train_loss": 0.036167845129966736, "train_loss_bc": 0.036167845129966736, "train_loss_llm": 0.0, "grad_norm": 0.031327035278081894, "global_step": 21241, "epoch": 590, "lr": 0.001} +{"train_loss": 0.025781363248825073, "train_loss_bc": 0.025781363248825073, "train_loss_llm": 0.0, "grad_norm": 0.047832656651735306, "global_step": 21242, "epoch": 590, "lr": 0.001} +{"train_loss": 0.02980947680771351, "train_loss_bc": 0.02980947680771351, "train_loss_llm": 0.0, "grad_norm": 0.06199925020337105, "global_step": 21243, "epoch": 590, "lr": 0.001} +{"train_loss": 0.031010441482067108, "train_loss_bc": 0.031004205346107483, "train_loss_llm": 0.0006236135959625244, "grad_norm": 0.07834926247596741, "global_step": 21244, "epoch": 590, "lr": 0.001} +{"train_loss": 0.04239414632320404, "train_loss_bc": 0.04239414632320404, "train_loss_llm": 0.0, "grad_norm": 0.03389429673552513, "global_step": 21245, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03666699305176735, "train_loss_bc": 0.03666321188211441, "train_loss_llm": 0.00037807226181030273, "grad_norm": 0.05392354354262352, "global_step": 21246, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03242579475045204, "train_loss_bc": 0.03242579475045204, "train_loss_llm": 0.0, "grad_norm": 0.06338498741388321, "global_step": 21247, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03466983884572983, "train_loss_bc": 0.03465135022997856, "train_loss_llm": 0.0018488764762878418, "grad_norm": 0.07663843780755997, "global_step": 21248, "epoch": 590, "lr": 0.001} +{"train_loss": 0.032152943313121796, "train_loss_bc": 0.032152943313121796, "train_loss_llm": 0.0, "grad_norm": 0.032423730939626694, "global_step": 21249, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03209039196372032, "train_loss_bc": 0.03209039196372032, "train_loss_llm": 0.0, "grad_norm": 0.050215598195791245, "global_step": 21250, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03145921230316162, "train_loss_bc": 0.03145921230316162, "train_loss_llm": 0.0, "grad_norm": 0.058279622346162796, "global_step": 21251, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03528454527258873, "train_loss_bc": 0.03527330607175827, "train_loss_llm": 0.001123875379562378, "grad_norm": 0.07793383300304413, "global_step": 21252, "epoch": 590, "lr": 0.001} +{"train_loss": 0.033643290400505066, "train_loss_bc": 0.033643290400505066, "train_loss_llm": 0.0, "grad_norm": 0.03167369216680527, "global_step": 21253, "epoch": 590, "lr": 0.001} +{"train_loss": 0.028147699311375618, "train_loss_bc": 0.028147699311375618, "train_loss_llm": 0.0, "grad_norm": 0.05185304582118988, "global_step": 21254, "epoch": 590, "lr": 0.001} +{"train_loss": 0.039029382169246674, "train_loss_bc": 0.03901614248752594, "train_loss_llm": 0.001324087381362915, "grad_norm": 0.06611545383930206, "global_step": 21255, "epoch": 590, "lr": 0.001} +{"train_loss": 0.02732945792376995, "train_loss_bc": 0.02732945792376995, "train_loss_llm": 0.0, "grad_norm": 0.07572000473737717, "global_step": 21256, "epoch": 590, "lr": 0.001} +{"train_loss": 0.026887381449341774, "train_loss_bc": 0.02688298001885414, "train_loss_llm": 0.000440061092376709, "grad_norm": 0.027828287333250046, "global_step": 21257, "epoch": 590, "lr": 0.001} +{"train_loss": 0.02939721755683422, "train_loss_bc": 0.02939721755683422, "train_loss_llm": 0.0, "grad_norm": 0.0393889918923378, "global_step": 21258, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03646466135978699, "train_loss_bc": 0.03646466135978699, "train_loss_llm": 0.0, "grad_norm": 0.06252697855234146, "global_step": 21259, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03578058257699013, "train_loss_bc": 0.03578058257699013, "train_loss_llm": 0.0, "grad_norm": 0.08246763795614243, "global_step": 21260, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03445720300078392, "train_loss_bc": 0.03445720300078392, "train_loss_llm": 0.0, "grad_norm": 0.030205760151147842, "global_step": 21261, "epoch": 590, "lr": 0.001} +{"train_loss": 0.04357326030731201, "train_loss_bc": 0.04356527701020241, "train_loss_llm": 0.0007984340190887451, "grad_norm": 0.051470570266246796, "global_step": 21262, "epoch": 590, "lr": 0.001} +{"train_loss": 0.033366329967975616, "train_loss_bc": 0.033366329967975616, "train_loss_llm": 0.0, "grad_norm": 0.06868307292461395, "global_step": 21263, "epoch": 590, "lr": 0.001} +{"train_loss": 0.030960150063037872, "train_loss_bc": 0.030959371477365494, "train_loss_llm": 7.784366607666016e-05, "grad_norm": 0.07430258393287659, "global_step": 21264, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03206109255552292, "train_loss_bc": 0.03206109255552292, "train_loss_llm": 0.0, "grad_norm": 0.025265762582421303, "global_step": 21265, "epoch": 590, "lr": 0.001} +{"train_loss": 0.0285941231995821, "train_loss_bc": 0.028586799278855324, "train_loss_llm": 0.0007324516773223877, "grad_norm": 0.044996753334999084, "global_step": 21266, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03159383311867714, "train_loss_bc": 0.03159383311867714, "train_loss_llm": 0.0, "grad_norm": 0.056507308036088943, "global_step": 21267, "epoch": 590, "lr": 0.001} +{"train_loss": 0.033322036266326904, "train_loss_bc": 0.033322036266326904, "train_loss_llm": 0.0, "grad_norm": 0.07195547223091125, "global_step": 21268, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03097144141793251, "train_loss_bc": 0.03097144141793251, "train_loss_llm": 0.0, "grad_norm": 0.02151568979024887, "global_step": 21269, "epoch": 590, "lr": 0.001} +{"train_loss": 0.02971760369837284, "train_loss_bc": 0.029696281999349594, "train_loss_llm": 0.0021320879459381104, "grad_norm": 0.03866465017199516, "global_step": 21270, "epoch": 590, "lr": 0.001} +{"train_loss": 0.03158948943018913, "train_loss_bc": 0.031568434089422226, "train_loss_llm": 0.002105712890625, "grad_norm": 0.04741974547505379, "global_step": 21271, "epoch": 590, "lr": 0.001} +{"train_loss": 0.034239862114191055, "train_loss_bc": 0.03422645106911659, "train_loss_llm": 0.0013411641120910645, "grad_norm": 0.06290558725595474, "global_step": 21272, "epoch": 590, "lr": 0.001} +{"train_loss": 0.027979431673884392, "train_loss_bc": 0.027979431673884392, "train_loss_llm": 0.0, "grad_norm": 0.023049371317029, "global_step": 21273, "epoch": 590, "lr": 0.001} +{"train_loss": 0.028849726542830467, "train_loss_bc": 0.028849726542830467, "train_loss_llm": 0.0, "grad_norm": 0.03818309307098389, "global_step": 21274, "epoch": 590, "lr": 0.001} +{"train_loss": 0.032756732414580055, "train_loss_bc": 0.03788401558995247, "train_loss_llm": 0.0, "grad_norm": 0.05789405107498169, "global_step": 21275, "epoch": 590, "lr": 0.001, "train/cumulative_reward": 71.84904946783941, "train/mean_score": 6.514883002881199, "train/success_rate": 0.1, "test/cumulative_reward": 33.87331907723882, "test/mean_score": 9.412169536083386, "test/success_rate": 0.26, "val_loss": 0.08122594654560089, "train_action_mse_error": 0.03526175022125244} +{"train_loss": 0.038556475192308426, "train_loss_bc": 0.038556475192308426, "train_loss_llm": 0.0, "grad_norm": 0.0705530121922493, "global_step": 21276, "epoch": 591, "lr": 0.001} +{"train_loss": 0.034220002591609955, "train_loss_bc": 0.03421410918235779, "train_loss_llm": 0.0005893111228942871, "grad_norm": 0.027590878307819366, "global_step": 21277, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03617528825998306, "train_loss_bc": 0.03617528825998306, "train_loss_llm": 0.0, "grad_norm": 0.04133884236216545, "global_step": 21278, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03517986088991165, "train_loss_bc": 0.03517986088991165, "train_loss_llm": 0.0, "grad_norm": 0.053395502269268036, "global_step": 21279, "epoch": 591, "lr": 0.001} +{"train_loss": 0.032273780554533005, "train_loss_bc": 0.03224436566233635, "train_loss_llm": 0.002941310405731201, "grad_norm": 0.06267189234495163, "global_step": 21280, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03508908674120903, "train_loss_bc": 0.03508908674120903, "train_loss_llm": 0.0, "grad_norm": 0.02999875321984291, "global_step": 21281, "epoch": 591, "lr": 0.001} +{"train_loss": 0.031326644122600555, "train_loss_bc": 0.031326644122600555, "train_loss_llm": 0.0, "grad_norm": 0.046649765223264694, "global_step": 21282, "epoch": 591, "lr": 0.001} +{"train_loss": 0.031713731586933136, "train_loss_bc": 0.031713731586933136, "train_loss_llm": 0.0, "grad_norm": 0.06562357395887375, "global_step": 21283, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03439636528491974, "train_loss_bc": 0.03439636528491974, "train_loss_llm": 0.0, "grad_norm": 0.07332224398851395, "global_step": 21284, "epoch": 591, "lr": 0.001} +{"train_loss": 0.036368973553180695, "train_loss_bc": 0.036368973553180695, "train_loss_llm": 0.0, "grad_norm": 0.0259376373142004, "global_step": 21285, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03378012776374817, "train_loss_bc": 0.03378012776374817, "train_loss_llm": 0.0, "grad_norm": 0.042527176439762115, "global_step": 21286, "epoch": 591, "lr": 0.001} +{"train_loss": 0.029827222228050232, "train_loss_bc": 0.029827222228050232, "train_loss_llm": 0.0, "grad_norm": 0.055428482592105865, "global_step": 21287, "epoch": 591, "lr": 0.001} +{"train_loss": 0.027761291712522507, "train_loss_bc": 0.027761291712522507, "train_loss_llm": 0.0, "grad_norm": 0.06801111996173859, "global_step": 21288, "epoch": 591, "lr": 0.001} +{"train_loss": 0.030556607991456985, "train_loss_bc": 0.030556607991456985, "train_loss_llm": 0.0, "grad_norm": 0.024852078408002853, "global_step": 21289, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03210184723138809, "train_loss_bc": 0.03210184723138809, "train_loss_llm": 0.0, "grad_norm": 0.039950065314769745, "global_step": 21290, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03551393374800682, "train_loss_bc": 0.03551393374800682, "train_loss_llm": 0.0, "grad_norm": 0.047608185559511185, "global_step": 21291, "epoch": 591, "lr": 0.001} +{"train_loss": 0.0379776656627655, "train_loss_bc": 0.0379776656627655, "train_loss_llm": 0.0, "grad_norm": 0.06263288110494614, "global_step": 21292, "epoch": 591, "lr": 0.001} +{"train_loss": 0.027793530374765396, "train_loss_bc": 0.027793530374765396, "train_loss_llm": 0.0, "grad_norm": 0.03498318791389465, "global_step": 21293, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03607913479208946, "train_loss_bc": 0.03607913479208946, "train_loss_llm": 0.0, "grad_norm": 0.05526650696992874, "global_step": 21294, "epoch": 591, "lr": 0.001} +{"train_loss": 0.025812817737460136, "train_loss_bc": 0.025812817737460136, "train_loss_llm": 0.0, "grad_norm": 0.07566581666469574, "global_step": 21295, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03560878708958626, "train_loss_bc": 0.035590846091508865, "train_loss_llm": 0.0017942190170288086, "grad_norm": 0.08971436321735382, "global_step": 21296, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03633476048707962, "train_loss_bc": 0.03633476048707962, "train_loss_llm": 0.0, "grad_norm": 0.035959772765636444, "global_step": 21297, "epoch": 591, "lr": 0.001} +{"train_loss": 0.031345434486866, "train_loss_bc": 0.031345434486866, "train_loss_llm": 0.0, "grad_norm": 0.0496470108628273, "global_step": 21298, "epoch": 591, "lr": 0.001} +{"train_loss": 0.0342305451631546, "train_loss_bc": 0.0342305451631546, "train_loss_llm": 0.0, "grad_norm": 0.0637291744351387, "global_step": 21299, "epoch": 591, "lr": 0.001} +{"train_loss": 0.02988251857459545, "train_loss_bc": 0.029878996312618256, "train_loss_llm": 0.0003522932529449463, "grad_norm": 0.0770864263176918, "global_step": 21300, "epoch": 591, "lr": 0.001} +{"train_loss": 0.026551438495516777, "train_loss_bc": 0.026551438495516777, "train_loss_llm": 0.0, "grad_norm": 0.023616524413228035, "global_step": 21301, "epoch": 591, "lr": 0.001} +{"train_loss": 0.02762780711054802, "train_loss_bc": 0.02762780711054802, "train_loss_llm": 0.0, "grad_norm": 0.037153031677007675, "global_step": 21302, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03126765042543411, "train_loss_bc": 0.03125875070691109, "train_loss_llm": 0.0008900761604309082, "grad_norm": 0.05114202946424484, "global_step": 21303, "epoch": 591, "lr": 0.001} +{"train_loss": 0.026822317391633987, "train_loss_bc": 0.026822317391633987, "train_loss_llm": 0.0, "grad_norm": 0.06182631105184555, "global_step": 21304, "epoch": 591, "lr": 0.001} +{"train_loss": 0.029855070635676384, "train_loss_bc": 0.029855070635676384, "train_loss_llm": 0.0, "grad_norm": 0.025326473638415337, "global_step": 21305, "epoch": 591, "lr": 0.001} +{"train_loss": 0.033084720373153687, "train_loss_bc": 0.033084720373153687, "train_loss_llm": 0.0, "grad_norm": 0.03704393282532692, "global_step": 21306, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03339247405529022, "train_loss_bc": 0.03337249532341957, "train_loss_llm": 0.0019978582859039307, "grad_norm": 0.05290502682328224, "global_step": 21307, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03355387598276138, "train_loss_bc": 0.03355387598276138, "train_loss_llm": 0.0, "grad_norm": 0.06788696348667145, "global_step": 21308, "epoch": 591, "lr": 0.001} +{"train_loss": 0.031708378344774246, "train_loss_bc": 0.031708378344774246, "train_loss_llm": 0.0, "grad_norm": 0.02916882187128067, "global_step": 21309, "epoch": 591, "lr": 0.001} +{"train_loss": 0.038069210946559906, "train_loss_bc": 0.038069210946559906, "train_loss_llm": 0.0, "grad_norm": 0.05721093341708183, "global_step": 21310, "epoch": 591, "lr": 0.001} +{"train_loss": 0.03254831970358888, "train_loss_bc": 0.02987435646355152, "train_loss_llm": 0.002577483654022217, "grad_norm": 0.0573330782353878, "global_step": 21311, "epoch": 591, "lr": 0.001, "val_loss": 0.08046958595514297} +{"train_loss": 0.027128908783197403, "train_loss_bc": 0.027128908783197403, "train_loss_llm": 0.0, "grad_norm": 0.0717928484082222, "global_step": 21312, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03326357156038284, "train_loss_bc": 0.03324972838163376, "train_loss_llm": 0.0013841688632965088, "grad_norm": 0.026050027459859848, "global_step": 21313, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03477191925048828, "train_loss_bc": 0.03477191925048828, "train_loss_llm": 0.0, "grad_norm": 0.04499751329421997, "global_step": 21314, "epoch": 592, "lr": 0.001} +{"train_loss": 0.034392815083265305, "train_loss_bc": 0.034392815083265305, "train_loss_llm": 0.0, "grad_norm": 0.06649433076381683, "global_step": 21315, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03917805850505829, "train_loss_bc": 0.03917805850505829, "train_loss_llm": 0.0, "grad_norm": 0.07049474120140076, "global_step": 21316, "epoch": 592, "lr": 0.001} +{"train_loss": 0.02944132871925831, "train_loss_bc": 0.029439067468047142, "train_loss_llm": 0.00022619962692260742, "grad_norm": 0.022492006421089172, "global_step": 21317, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03150641545653343, "train_loss_bc": 0.03150641545653343, "train_loss_llm": 0.0, "grad_norm": 0.038100942969322205, "global_step": 21318, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03451228141784668, "train_loss_bc": 0.03451228141784668, "train_loss_llm": 0.0, "grad_norm": 0.05753423646092415, "global_step": 21319, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03174611181020737, "train_loss_bc": 0.03174611181020737, "train_loss_llm": 0.0, "grad_norm": 0.0858728364109993, "global_step": 21320, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03201962262392044, "train_loss_bc": 0.03201690688729286, "train_loss_llm": 0.0002715587615966797, "grad_norm": 0.03119843266904354, "global_step": 21321, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03335298225283623, "train_loss_bc": 0.03335298225283623, "train_loss_llm": 0.0, "grad_norm": 0.045716602355241776, "global_step": 21322, "epoch": 592, "lr": 0.001} +{"train_loss": 0.027810974046587944, "train_loss_bc": 0.027807915583252907, "train_loss_llm": 0.0003058910369873047, "grad_norm": 0.06200765073299408, "global_step": 21323, "epoch": 592, "lr": 0.001} +{"train_loss": 0.033677320927381516, "train_loss_bc": 0.03367158770561218, "train_loss_llm": 0.0005732476711273193, "grad_norm": 0.07608728855848312, "global_step": 21324, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03521345555782318, "train_loss_bc": 0.03521345555782318, "train_loss_llm": 0.0, "grad_norm": 0.0311002004891634, "global_step": 21325, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03402382507920265, "train_loss_bc": 0.034012727439403534, "train_loss_llm": 0.0011098682880401611, "grad_norm": 0.03871196135878563, "global_step": 21326, "epoch": 592, "lr": 0.001} +{"train_loss": 0.031163077801465988, "train_loss_bc": 0.031160108745098114, "train_loss_llm": 0.0002969205379486084, "grad_norm": 0.06084637716412544, "global_step": 21327, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03655527904629707, "train_loss_bc": 0.03655527904629707, "train_loss_llm": 0.0, "grad_norm": 0.07356040179729462, "global_step": 21328, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03142143413424492, "train_loss_bc": 0.03142143413424492, "train_loss_llm": 0.0, "grad_norm": 0.026483841240406036, "global_step": 21329, "epoch": 592, "lr": 0.001} +{"train_loss": 0.027771741151809692, "train_loss_bc": 0.027771741151809692, "train_loss_llm": 0.0, "grad_norm": 0.04300433769822121, "global_step": 21330, "epoch": 592, "lr": 0.001} +{"train_loss": 0.030389711260795593, "train_loss_bc": 0.030362948775291443, "train_loss_llm": 0.0026762783527374268, "grad_norm": 0.05137378349900246, "global_step": 21331, "epoch": 592, "lr": 0.001} +{"train_loss": 0.030489902943372726, "train_loss_bc": 0.030477892607450485, "train_loss_llm": 0.0012009739875793457, "grad_norm": 0.06795423477888107, "global_step": 21332, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03143598139286041, "train_loss_bc": 0.03143598139286041, "train_loss_llm": 0.0, "grad_norm": 0.028798962011933327, "global_step": 21333, "epoch": 592, "lr": 0.001} +{"train_loss": 0.0268010925501585, "train_loss_bc": 0.0268010925501585, "train_loss_llm": 0.0, "grad_norm": 0.04930495098233223, "global_step": 21334, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03329060226678848, "train_loss_bc": 0.03328590467572212, "train_loss_llm": 0.00046959519386291504, "grad_norm": 0.06980142742395401, "global_step": 21335, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03338242694735527, "train_loss_bc": 0.03338242694735527, "train_loss_llm": 0.0, "grad_norm": 0.09012177586555481, "global_step": 21336, "epoch": 592, "lr": 0.001} +{"train_loss": 0.0353812649846077, "train_loss_bc": 0.0353812649846077, "train_loss_llm": 0.0, "grad_norm": 0.0325007401406765, "global_step": 21337, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03322100639343262, "train_loss_bc": 0.033213309943675995, "train_loss_llm": 0.0007696449756622314, "grad_norm": 0.05118532106280327, "global_step": 21338, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03623565658926964, "train_loss_bc": 0.03623565658926964, "train_loss_llm": 0.0, "grad_norm": 0.07036492973566055, "global_step": 21339, "epoch": 592, "lr": 0.001} +{"train_loss": 0.034406278282403946, "train_loss_bc": 0.034406278282403946, "train_loss_llm": 0.0, "grad_norm": 0.08388222754001617, "global_step": 21340, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03866991773247719, "train_loss_bc": 0.03866991773247719, "train_loss_llm": 0.0, "grad_norm": 0.03207247331738472, "global_step": 21341, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03247346729040146, "train_loss_bc": 0.03247346729040146, "train_loss_llm": 0.0, "grad_norm": 0.05201908200979233, "global_step": 21342, "epoch": 592, "lr": 0.001} +{"train_loss": 0.029605446383357048, "train_loss_bc": 0.029605446383357048, "train_loss_llm": 0.0, "grad_norm": 0.06033261492848396, "global_step": 21343, "epoch": 592, "lr": 0.001} +{"train_loss": 0.029086852446198463, "train_loss_bc": 0.029086142778396606, "train_loss_llm": 7.104873657226562e-05, "grad_norm": 0.07751413434743881, "global_step": 21344, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03248260170221329, "train_loss_bc": 0.032472968101501465, "train_loss_llm": 0.0009635090827941895, "grad_norm": 0.031456660479307175, "global_step": 21345, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03300848230719566, "train_loss_bc": 0.033004701137542725, "train_loss_llm": 0.00037795305252075195, "grad_norm": 0.048271987587213516, "global_step": 21346, "epoch": 592, "lr": 0.001} +{"train_loss": 0.03279049751452274, "train_loss_bc": 0.04114609584212303, "train_loss_llm": 0.0, "grad_norm": 0.07326534390449524, "global_step": 21347, "epoch": 592, "lr": 0.001, "val_loss": 0.08047233521938324} +{"train_loss": 0.03355251997709274, "train_loss_bc": 0.03355251997709274, "train_loss_llm": 0.0, "grad_norm": 0.09133902937173843, "global_step": 21348, "epoch": 593, "lr": 0.001} +{"train_loss": 0.035600584000349045, "train_loss_bc": 0.035600584000349045, "train_loss_llm": 0.0, "grad_norm": 0.030832916498184204, "global_step": 21349, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03620777279138565, "train_loss_bc": 0.03620777279138565, "train_loss_llm": 0.0, "grad_norm": 0.052909303456544876, "global_step": 21350, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03140847384929657, "train_loss_bc": 0.03140078857541084, "train_loss_llm": 0.0007686614990234375, "grad_norm": 0.07169084995985031, "global_step": 21351, "epoch": 593, "lr": 0.001} +{"train_loss": 0.031326666474342346, "train_loss_bc": 0.031326666474342346, "train_loss_llm": 0.0, "grad_norm": 0.08546864986419678, "global_step": 21352, "epoch": 593, "lr": 0.001} +{"train_loss": 0.029939334839582443, "train_loss_bc": 0.029939334839582443, "train_loss_llm": 0.0, "grad_norm": 0.02463238313794136, "global_step": 21353, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03824440762400627, "train_loss_bc": 0.03824440762400627, "train_loss_llm": 0.0, "grad_norm": 0.051024340093135834, "global_step": 21354, "epoch": 593, "lr": 0.001} +{"train_loss": 0.032414622604846954, "train_loss_bc": 0.03240644186735153, "train_loss_llm": 0.0008178949356079102, "grad_norm": 0.06198930740356445, "global_step": 21355, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03296256810426712, "train_loss_bc": 0.032931700348854065, "train_loss_llm": 0.003086954355239868, "grad_norm": 0.07911620289087296, "global_step": 21356, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03095855936408043, "train_loss_bc": 0.03095855936408043, "train_loss_llm": 0.0, "grad_norm": 0.02835794724524021, "global_step": 21357, "epoch": 593, "lr": 0.001} +{"train_loss": 0.028964316472411156, "train_loss_bc": 0.02895028330385685, "train_loss_llm": 0.0014033317565917969, "grad_norm": 0.040827829390764236, "global_step": 21358, "epoch": 593, "lr": 0.001} +{"train_loss": 0.029689405113458633, "train_loss_bc": 0.029689405113458633, "train_loss_llm": 0.0, "grad_norm": 0.05570929870009422, "global_step": 21359, "epoch": 593, "lr": 0.001} +{"train_loss": 0.02911096252501011, "train_loss_bc": 0.02911096252501011, "train_loss_llm": 0.0, "grad_norm": 0.06706439703702927, "global_step": 21360, "epoch": 593, "lr": 0.001} +{"train_loss": 0.021638572216033936, "train_loss_bc": 0.021638572216033936, "train_loss_llm": 0.0, "grad_norm": 0.028701072558760643, "global_step": 21361, "epoch": 593, "lr": 0.001} +{"train_loss": 0.030757594853639603, "train_loss_bc": 0.030757594853639603, "train_loss_llm": 0.0, "grad_norm": 0.03732157126069069, "global_step": 21362, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03290216624736786, "train_loss_bc": 0.03290216624736786, "train_loss_llm": 0.0, "grad_norm": 0.05028960853815079, "global_step": 21363, "epoch": 593, "lr": 0.001} +{"train_loss": 0.037865035235881805, "train_loss_bc": 0.037865035235881805, "train_loss_llm": 0.0, "grad_norm": 0.06586168706417084, "global_step": 21364, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03223112225532532, "train_loss_bc": 0.03223112225532532, "train_loss_llm": 0.0, "grad_norm": 0.030469294637441635, "global_step": 21365, "epoch": 593, "lr": 0.001} +{"train_loss": 0.0305922981351614, "train_loss_bc": 0.0305922981351614, "train_loss_llm": 0.0, "grad_norm": 0.050826545804739, "global_step": 21366, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03753722086548805, "train_loss_bc": 0.03751802816987038, "train_loss_llm": 0.0019192993640899658, "grad_norm": 0.06703652441501617, "global_step": 21367, "epoch": 593, "lr": 0.001} +{"train_loss": 0.028331687673926353, "train_loss_bc": 0.028321746736764908, "train_loss_llm": 0.0009940862655639648, "grad_norm": 0.07523317635059357, "global_step": 21368, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03154991194605827, "train_loss_bc": 0.03154991194605827, "train_loss_llm": 0.0, "grad_norm": 0.025882620364427567, "global_step": 21369, "epoch": 593, "lr": 0.001} +{"train_loss": 0.02997724339365959, "train_loss_bc": 0.02997724339365959, "train_loss_llm": 0.0, "grad_norm": 0.036994047462940216, "global_step": 21370, "epoch": 593, "lr": 0.001} +{"train_loss": 0.02445286698639393, "train_loss_bc": 0.02445286698639393, "train_loss_llm": 0.0, "grad_norm": 0.045680221170186996, "global_step": 21371, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03874495252966881, "train_loss_bc": 0.03871724754571915, "train_loss_llm": 0.002770543098449707, "grad_norm": 0.06913234293460846, "global_step": 21372, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03588700667023659, "train_loss_bc": 0.03588700667023659, "train_loss_llm": 0.0, "grad_norm": 0.035219766199588776, "global_step": 21373, "epoch": 593, "lr": 0.001} +{"train_loss": 0.030933748930692673, "train_loss_bc": 0.030933748930692673, "train_loss_llm": 0.0, "grad_norm": 0.05299578607082367, "global_step": 21374, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03299253061413765, "train_loss_bc": 0.03299253061413765, "train_loss_llm": 0.0, "grad_norm": 0.06411866098642349, "global_step": 21375, "epoch": 593, "lr": 0.001} +{"train_loss": 0.033806461840867996, "train_loss_bc": 0.033806461840867996, "train_loss_llm": 0.0, "grad_norm": 0.07757864147424698, "global_step": 21376, "epoch": 593, "lr": 0.001} +{"train_loss": 0.034801531583070755, "train_loss_bc": 0.034801531583070755, "train_loss_llm": 0.0, "grad_norm": 0.025712644681334496, "global_step": 21377, "epoch": 593, "lr": 0.001} +{"train_loss": 0.0345340371131897, "train_loss_bc": 0.0345340371131897, "train_loss_llm": 0.0, "grad_norm": 0.040150538086891174, "global_step": 21378, "epoch": 593, "lr": 0.001} +{"train_loss": 0.04284350201487541, "train_loss_bc": 0.04284350201487541, "train_loss_llm": 0.0, "grad_norm": 0.05468090996146202, "global_step": 21379, "epoch": 593, "lr": 0.001} +{"train_loss": 0.029896773397922516, "train_loss_bc": 0.029896773397922516, "train_loss_llm": 0.0, "grad_norm": 0.06751012802124023, "global_step": 21380, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03930061683058739, "train_loss_bc": 0.03930061683058739, "train_loss_llm": 0.0, "grad_norm": 0.030394356697797775, "global_step": 21381, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03345238044857979, "train_loss_bc": 0.03345238044857979, "train_loss_llm": 0.0, "grad_norm": 0.04652855545282364, "global_step": 21382, "epoch": 593, "lr": 0.001} +{"train_loss": 0.03274248504183359, "train_loss_bc": 0.03332000598311424, "train_loss_llm": 0.0, "grad_norm": 0.05543438717722893, "global_step": 21383, "epoch": 593, "lr": 0.001, "val_loss": 0.08179639279842377} +{"train_loss": 0.03200583532452583, "train_loss_bc": 0.03200583532452583, "train_loss_llm": 0.0, "grad_norm": 0.06970228999853134, "global_step": 21384, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03724905475974083, "train_loss_bc": 0.03724905475974083, "train_loss_llm": 0.0, "grad_norm": 0.03444255515933037, "global_step": 21385, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03710634633898735, "train_loss_bc": 0.03710634633898735, "train_loss_llm": 0.0, "grad_norm": 0.05283007025718689, "global_step": 21386, "epoch": 594, "lr": 0.001} +{"train_loss": 0.029681604355573654, "train_loss_bc": 0.029681604355573654, "train_loss_llm": 0.0, "grad_norm": 0.06447667628526688, "global_step": 21387, "epoch": 594, "lr": 0.001} +{"train_loss": 0.028240522369742393, "train_loss_bc": 0.028240522369742393, "train_loss_llm": 0.0, "grad_norm": 0.0820140391588211, "global_step": 21388, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03844089061021805, "train_loss_bc": 0.03844089061021805, "train_loss_llm": 0.0, "grad_norm": 0.02919880859553814, "global_step": 21389, "epoch": 594, "lr": 0.001} +{"train_loss": 0.031392235308885574, "train_loss_bc": 0.031392235308885574, "train_loss_llm": 0.0, "grad_norm": 0.04428788647055626, "global_step": 21390, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03357597440481186, "train_loss_bc": 0.03357597440481186, "train_loss_llm": 0.0, "grad_norm": 0.06246021389961243, "global_step": 21391, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03081059269607067, "train_loss_bc": 0.03081059269607067, "train_loss_llm": 0.0, "grad_norm": 0.07653935998678207, "global_step": 21392, "epoch": 594, "lr": 0.001} +{"train_loss": 0.033229049295186996, "train_loss_bc": 0.033229049295186996, "train_loss_llm": 0.0, "grad_norm": 0.0277253407984972, "global_step": 21393, "epoch": 594, "lr": 0.001} +{"train_loss": 0.02953873947262764, "train_loss_bc": 0.02953873947262764, "train_loss_llm": 0.0, "grad_norm": 0.04382885619997978, "global_step": 21394, "epoch": 594, "lr": 0.001} +{"train_loss": 0.034122928977012634, "train_loss_bc": 0.034122928977012634, "train_loss_llm": 0.0, "grad_norm": 0.06349536031484604, "global_step": 21395, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03062228113412857, "train_loss_bc": 0.03062228113412857, "train_loss_llm": 0.0, "grad_norm": 0.07422637939453125, "global_step": 21396, "epoch": 594, "lr": 0.001} +{"train_loss": 0.030734756961464882, "train_loss_bc": 0.03073323518037796, "train_loss_llm": 0.00015223026275634766, "grad_norm": 0.028069835156202316, "global_step": 21397, "epoch": 594, "lr": 0.001} +{"train_loss": 0.030336588621139526, "train_loss_bc": 0.030336588621139526, "train_loss_llm": 0.0, "grad_norm": 0.04233196750283241, "global_step": 21398, "epoch": 594, "lr": 0.001} +{"train_loss": 0.029479676857590675, "train_loss_bc": 0.029479218646883965, "train_loss_llm": 4.583597183227539e-05, "grad_norm": 0.05538090318441391, "global_step": 21399, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03524799272418022, "train_loss_bc": 0.03524772450327873, "train_loss_llm": 2.6702880859375e-05, "grad_norm": 0.06894469261169434, "global_step": 21400, "epoch": 594, "lr": 0.001} +{"train_loss": 0.032613739371299744, "train_loss_bc": 0.03259052336215973, "train_loss_llm": 0.0023217201232910156, "grad_norm": 0.028503786772489548, "global_step": 21401, "epoch": 594, "lr": 0.001} +{"train_loss": 0.029352683573961258, "train_loss_bc": 0.029352683573961258, "train_loss_llm": 0.0, "grad_norm": 0.040888577699661255, "global_step": 21402, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03206539526581764, "train_loss_bc": 0.032043591141700745, "train_loss_llm": 0.0021802783012390137, "grad_norm": 0.056753192096948624, "global_step": 21403, "epoch": 594, "lr": 0.001} +{"train_loss": 0.0366244800388813, "train_loss_bc": 0.03662101551890373, "train_loss_llm": 0.0003463327884674072, "grad_norm": 0.07891539484262466, "global_step": 21404, "epoch": 594, "lr": 0.001} +{"train_loss": 0.0344245620071888, "train_loss_bc": 0.0344245620071888, "train_loss_llm": 0.0, "grad_norm": 0.03149547800421715, "global_step": 21405, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03429468721151352, "train_loss_bc": 0.03429100662469864, "train_loss_llm": 0.00036790966987609863, "grad_norm": 0.04678122699260712, "global_step": 21406, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03395973891019821, "train_loss_bc": 0.03394608944654465, "train_loss_llm": 0.0013647675514221191, "grad_norm": 0.0636032447218895, "global_step": 21407, "epoch": 594, "lr": 0.001} +{"train_loss": 0.04103035479784012, "train_loss_bc": 0.04103035479784012, "train_loss_llm": 0.0, "grad_norm": 0.07421378046274185, "global_step": 21408, "epoch": 594, "lr": 0.001} +{"train_loss": 0.0335443839430809, "train_loss_bc": 0.0335443839430809, "train_loss_llm": 0.0, "grad_norm": 0.036114342510700226, "global_step": 21409, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03954607993364334, "train_loss_bc": 0.03954607993364334, "train_loss_llm": 0.0, "grad_norm": 0.062037739902734756, "global_step": 21410, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03384871035814285, "train_loss_bc": 0.03384366258978844, "train_loss_llm": 0.0005048513412475586, "grad_norm": 0.0695132464170456, "global_step": 21411, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03539653122425079, "train_loss_bc": 0.035368289798498154, "train_loss_llm": 0.0028243064880371094, "grad_norm": 0.08253886550664902, "global_step": 21412, "epoch": 594, "lr": 0.001} +{"train_loss": 0.03261547535657883, "train_loss_bc": 0.03261547535657883, "train_loss_llm": 0.0, "grad_norm": 0.02714371308684349, "global_step": 21413, "epoch": 594, "lr": 0.001} +{"train_loss": 0.02518511936068535, "train_loss_bc": 0.02516109123826027, "train_loss_llm": 0.0024028122425079346, "grad_norm": 0.03743590787053108, "global_step": 21414, "epoch": 594, "lr": 0.001} +{"train_loss": 0.02534976601600647, "train_loss_bc": 0.02534976601600647, "train_loss_llm": 0.0, "grad_norm": 0.045621465891599655, "global_step": 21415, "epoch": 594, "lr": 0.001} +{"train_loss": 0.028655242174863815, "train_loss_bc": 0.028655242174863815, "train_loss_llm": 0.0, "grad_norm": 0.05479268357157707, "global_step": 21416, "epoch": 594, "lr": 0.001} +{"train_loss": 0.030593205243349075, "train_loss_bc": 0.03057342767715454, "train_loss_llm": 0.001977771520614624, "grad_norm": 0.024518026039004326, "global_step": 21417, "epoch": 594, "lr": 0.001} +{"train_loss": 0.030208734795451164, "train_loss_bc": 0.030208734795451164, "train_loss_llm": 0.0, "grad_norm": 0.03491515293717384, "global_step": 21418, "epoch": 594, "lr": 0.001} +{"train_loss": 0.032645225059241056, "train_loss_bc": 0.03410414233803749, "train_loss_llm": 0.0, "grad_norm": 0.05246676877140999, "global_step": 21419, "epoch": 594, "lr": 0.001, "val_loss": 0.08181615173816681} +{"train_loss": 0.032317932695150375, "train_loss_bc": 0.032317932695150375, "train_loss_llm": 0.0, "grad_norm": 0.05682796984910965, "global_step": 21420, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03101356141269207, "train_loss_bc": 0.03101356141269207, "train_loss_llm": 0.0, "grad_norm": 0.03160737827420235, "global_step": 21421, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03163468465209007, "train_loss_bc": 0.03161716088652611, "train_loss_llm": 0.0017524957656860352, "grad_norm": 0.040316659957170486, "global_step": 21422, "epoch": 595, "lr": 0.001} +{"train_loss": 0.04039672017097473, "train_loss_bc": 0.04039672017097473, "train_loss_llm": 0.0, "grad_norm": 0.04953662306070328, "global_step": 21423, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03189731016755104, "train_loss_bc": 0.03189731016755104, "train_loss_llm": 0.0, "grad_norm": 0.059881359338760376, "global_step": 21424, "epoch": 595, "lr": 0.001} +{"train_loss": 0.024602480232715607, "train_loss_bc": 0.024602480232715607, "train_loss_llm": 0.0, "grad_norm": 0.02487283945083618, "global_step": 21425, "epoch": 595, "lr": 0.001} +{"train_loss": 0.030795078724622726, "train_loss_bc": 0.030795078724622726, "train_loss_llm": 0.0, "grad_norm": 0.04595310240983963, "global_step": 21426, "epoch": 595, "lr": 0.001} +{"train_loss": 0.032478224486112595, "train_loss_bc": 0.032478224486112595, "train_loss_llm": 0.0, "grad_norm": 0.05486498773097992, "global_step": 21427, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03209349885582924, "train_loss_bc": 0.03209349885582924, "train_loss_llm": 0.0, "grad_norm": 0.0634603276848793, "global_step": 21428, "epoch": 595, "lr": 0.001} +{"train_loss": 0.035287197679281235, "train_loss_bc": 0.03528480604290962, "train_loss_llm": 0.0002391040325164795, "grad_norm": 0.03444358706474304, "global_step": 21429, "epoch": 595, "lr": 0.001} +{"train_loss": 0.026833761483430862, "train_loss_bc": 0.026833761483430862, "train_loss_llm": 0.0, "grad_norm": 0.05115794762969017, "global_step": 21430, "epoch": 595, "lr": 0.001} +{"train_loss": 0.02879815176129341, "train_loss_bc": 0.028786765411496162, "train_loss_llm": 0.0011387169361114502, "grad_norm": 0.06870380789041519, "global_step": 21431, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03580978512763977, "train_loss_bc": 0.03579786419868469, "train_loss_llm": 0.0011919140815734863, "grad_norm": 0.07674463093280792, "global_step": 21432, "epoch": 595, "lr": 0.001} +{"train_loss": 0.035123176872730255, "train_loss_bc": 0.035123176872730255, "train_loss_llm": 0.0, "grad_norm": 0.03134068101644516, "global_step": 21433, "epoch": 595, "lr": 0.001} +{"train_loss": 0.0333503820002079, "train_loss_bc": 0.0333503820002079, "train_loss_llm": 0.0, "grad_norm": 0.04339461401104927, "global_step": 21434, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03073645755648613, "train_loss_bc": 0.030725974589586258, "train_loss_llm": 0.0010483264923095703, "grad_norm": 0.054455701261758804, "global_step": 21435, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03178210183978081, "train_loss_bc": 0.03178210183978081, "train_loss_llm": 0.0, "grad_norm": 0.055335287004709244, "global_step": 21436, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03357042372226715, "train_loss_bc": 0.03357042372226715, "train_loss_llm": 0.0, "grad_norm": 0.02912815846502781, "global_step": 21437, "epoch": 595, "lr": 0.001} +{"train_loss": 0.030639272183179855, "train_loss_bc": 0.030639272183179855, "train_loss_llm": 0.0, "grad_norm": 0.03849167004227638, "global_step": 21438, "epoch": 595, "lr": 0.001} +{"train_loss": 0.033063821494579315, "train_loss_bc": 0.033060118556022644, "train_loss_llm": 0.0003701150417327881, "grad_norm": 0.05012325569987297, "global_step": 21439, "epoch": 595, "lr": 0.001} +{"train_loss": 0.031089814379811287, "train_loss_bc": 0.031063653528690338, "train_loss_llm": 0.0026161670684814453, "grad_norm": 0.05929330736398697, "global_step": 21440, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03422660380601883, "train_loss_bc": 0.03422660380601883, "train_loss_llm": 0.0, "grad_norm": 0.031193861737847328, "global_step": 21441, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03172621130943298, "train_loss_bc": 0.03172621130943298, "train_loss_llm": 0.0, "grad_norm": 0.040488652884960175, "global_step": 21442, "epoch": 595, "lr": 0.001} +{"train_loss": 0.025096358731389046, "train_loss_bc": 0.025096358731389046, "train_loss_llm": 0.0, "grad_norm": 0.05512204393744469, "global_step": 21443, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03118104673922062, "train_loss_bc": 0.03118104673922062, "train_loss_llm": 0.0, "grad_norm": 0.0696253553032875, "global_step": 21444, "epoch": 595, "lr": 0.001} +{"train_loss": 0.034435536712408066, "train_loss_bc": 0.034435536712408066, "train_loss_llm": 0.0, "grad_norm": 0.04076117277145386, "global_step": 21445, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03282398357987404, "train_loss_bc": 0.03282398357987404, "train_loss_llm": 0.0, "grad_norm": 0.04732869565486908, "global_step": 21446, "epoch": 595, "lr": 0.001} +{"train_loss": 0.031871501356363297, "train_loss_bc": 0.03185664862394333, "train_loss_llm": 0.0014853477478027344, "grad_norm": 0.06914183497428894, "global_step": 21447, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03868862986564636, "train_loss_bc": 0.03868402913212776, "train_loss_llm": 0.0004601478576660156, "grad_norm": 0.08761166036128998, "global_step": 21448, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03322632238268852, "train_loss_bc": 0.03322632238268852, "train_loss_llm": 0.0, "grad_norm": 0.028857894241809845, "global_step": 21449, "epoch": 595, "lr": 0.001} +{"train_loss": 0.03946224972605705, "train_loss_bc": 0.03946224972605705, "train_loss_llm": 0.0, "grad_norm": 0.04786837846040726, "global_step": 21450, "epoch": 595, "lr": 0.001} +{"train_loss": 0.027132710441946983, "train_loss_bc": 0.027132710441946983, "train_loss_llm": 0.0, "grad_norm": 0.05137500539422035, "global_step": 21451, "epoch": 595, "lr": 0.001} +{"train_loss": 0.034191545099020004, "train_loss_bc": 0.034191545099020004, "train_loss_llm": 0.0, "grad_norm": 0.062195610255002975, "global_step": 21452, "epoch": 595, "lr": 0.001} +{"train_loss": 0.033486537635326385, "train_loss_bc": 0.033486537635326385, "train_loss_llm": 0.0, "grad_norm": 0.026604151353240013, "global_step": 21453, "epoch": 595, "lr": 0.001} +{"train_loss": 0.037787050008773804, "train_loss_bc": 0.037775881588459015, "train_loss_llm": 0.0011167526245117188, "grad_norm": 0.050413962453603745, "global_step": 21454, "epoch": 595, "lr": 0.001} +{"train_loss": 0.0325957003256513, "train_loss_bc": 0.034795086830854416, "train_loss_llm": 0.0, "grad_norm": 0.07375927269458771, "global_step": 21455, "epoch": 595, "lr": 0.001, "train/cumulative_reward": 93.58822301924441, "train/mean_score": 20.846547954198865, "train/success_rate": 0.5, "test/cumulative_reward": 29.04939112178887, "test/mean_score": 9.380511918934488, "test/success_rate": 0.28, "val_loss": 0.08117783814668655, "train_action_mse_error": 0.03525335341691971} +{"train_loss": 0.02501668594777584, "train_loss_bc": 0.02501348964869976, "train_loss_llm": 0.0003196299076080322, "grad_norm": 0.0932384729385376, "global_step": 21456, "epoch": 596, "lr": 0.001} +{"train_loss": 0.027193795889616013, "train_loss_bc": 0.0271883737295866, "train_loss_llm": 0.0005421936511993408, "grad_norm": 0.025105677545070648, "global_step": 21457, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03250886872410774, "train_loss_bc": 0.03250886872410774, "train_loss_llm": 0.0, "grad_norm": 0.03804890438914299, "global_step": 21458, "epoch": 596, "lr": 0.001} +{"train_loss": 0.030740171670913696, "train_loss_bc": 0.030740171670913696, "train_loss_llm": 0.0, "grad_norm": 0.04517779499292374, "global_step": 21459, "epoch": 596, "lr": 0.001} +{"train_loss": 0.0341603085398674, "train_loss_bc": 0.0341603085398674, "train_loss_llm": 0.0, "grad_norm": 0.054310575127601624, "global_step": 21460, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03931780159473419, "train_loss_bc": 0.03931780159473419, "train_loss_llm": 0.0, "grad_norm": 0.02829771675169468, "global_step": 21461, "epoch": 596, "lr": 0.001} +{"train_loss": 0.031927093863487244, "train_loss_bc": 0.03190368413925171, "train_loss_llm": 0.002340853214263916, "grad_norm": 0.038709163665771484, "global_step": 21462, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03510763868689537, "train_loss_bc": 0.03510763868689537, "train_loss_llm": 0.0, "grad_norm": 0.0604836568236351, "global_step": 21463, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03409957140684128, "train_loss_bc": 0.03409957140684128, "train_loss_llm": 0.0, "grad_norm": 0.0796624943614006, "global_step": 21464, "epoch": 596, "lr": 0.001} +{"train_loss": 0.035354726016521454, "train_loss_bc": 0.035354726016521454, "train_loss_llm": 0.0, "grad_norm": 0.0338941253721714, "global_step": 21465, "epoch": 596, "lr": 0.001} +{"train_loss": 0.030138816684484482, "train_loss_bc": 0.030138816684484482, "train_loss_llm": 0.0, "grad_norm": 0.04816856235265732, "global_step": 21466, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03514779359102249, "train_loss_bc": 0.03514779359102249, "train_loss_llm": 0.0, "grad_norm": 0.0616205632686615, "global_step": 21467, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03177763521671295, "train_loss_bc": 0.03176611289381981, "train_loss_llm": 0.0011521577835083008, "grad_norm": 0.08116265386343002, "global_step": 21468, "epoch": 596, "lr": 0.001} +{"train_loss": 0.035036757588386536, "train_loss_bc": 0.035036757588386536, "train_loss_llm": 0.0, "grad_norm": 0.02924017608165741, "global_step": 21469, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03365395963191986, "train_loss_bc": 0.03365395963191986, "train_loss_llm": 0.0, "grad_norm": 0.0549023412168026, "global_step": 21470, "epoch": 596, "lr": 0.001} +{"train_loss": 0.034342873841524124, "train_loss_bc": 0.034336067736148834, "train_loss_llm": 0.0006804466247558594, "grad_norm": 0.07230391353368759, "global_step": 21471, "epoch": 596, "lr": 0.001} +{"train_loss": 0.031282663345336914, "train_loss_bc": 0.031282663345336914, "train_loss_llm": 0.0, "grad_norm": 0.08336415141820908, "global_step": 21472, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03009040094912052, "train_loss_bc": 0.03009040094912052, "train_loss_llm": 0.0, "grad_norm": 0.030788181349635124, "global_step": 21473, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03237059712409973, "train_loss_bc": 0.03237041458487511, "train_loss_llm": 1.8358230590820312e-05, "grad_norm": 0.05033496022224426, "global_step": 21474, "epoch": 596, "lr": 0.001} +{"train_loss": 0.033872585743665695, "train_loss_bc": 0.033872585743665695, "train_loss_llm": 0.0, "grad_norm": 0.06945917755365372, "global_step": 21475, "epoch": 596, "lr": 0.001} +{"train_loss": 0.036248400807380676, "train_loss_bc": 0.036248400807380676, "train_loss_llm": 0.0, "grad_norm": 0.07667279988527298, "global_step": 21476, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03150122985243797, "train_loss_bc": 0.03150122985243797, "train_loss_llm": 0.0, "grad_norm": 0.031592290848493576, "global_step": 21477, "epoch": 596, "lr": 0.001} +{"train_loss": 0.035942479968070984, "train_loss_bc": 0.035942479968070984, "train_loss_llm": 0.0, "grad_norm": 0.04686542600393295, "global_step": 21478, "epoch": 596, "lr": 0.001} +{"train_loss": 0.025858189910650253, "train_loss_bc": 0.025855522602796555, "train_loss_llm": 0.00026670098304748535, "grad_norm": 0.06261416524648666, "global_step": 21479, "epoch": 596, "lr": 0.001} +{"train_loss": 0.028434190899133682, "train_loss_bc": 0.028434190899133682, "train_loss_llm": 0.0, "grad_norm": 0.07191735506057739, "global_step": 21480, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03371541202068329, "train_loss_bc": 0.03371541202068329, "train_loss_llm": 0.0, "grad_norm": 0.030270645394921303, "global_step": 21481, "epoch": 596, "lr": 0.001} +{"train_loss": 0.0258079394698143, "train_loss_bc": 0.0258079394698143, "train_loss_llm": 0.0, "grad_norm": 0.041527122259140015, "global_step": 21482, "epoch": 596, "lr": 0.001} +{"train_loss": 0.030582280829548836, "train_loss_bc": 0.030580667778849602, "train_loss_llm": 0.00016126036643981934, "grad_norm": 0.05026344954967499, "global_step": 21483, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03725337237119675, "train_loss_bc": 0.03725337237119675, "train_loss_llm": 0.0, "grad_norm": 0.0613723061978817, "global_step": 21484, "epoch": 596, "lr": 0.001} +{"train_loss": 0.036267947405576706, "train_loss_bc": 0.036267947405576706, "train_loss_llm": 0.0, "grad_norm": 0.03546419367194176, "global_step": 21485, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03527006134390831, "train_loss_bc": 0.03527006134390831, "train_loss_llm": 0.0, "grad_norm": 0.05701858177781105, "global_step": 21486, "epoch": 596, "lr": 0.001} +{"train_loss": 0.026329822838306427, "train_loss_bc": 0.026329822838306427, "train_loss_llm": 0.0, "grad_norm": 0.06801760196685791, "global_step": 21487, "epoch": 596, "lr": 0.001} +{"train_loss": 0.031619179993867874, "train_loss_bc": 0.031619179993867874, "train_loss_llm": 0.0, "grad_norm": 0.0784786120057106, "global_step": 21488, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03424821048974991, "train_loss_bc": 0.03424821048974991, "train_loss_llm": 0.0, "grad_norm": 0.03904964029788971, "global_step": 21489, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03740910068154335, "train_loss_bc": 0.03740910068154335, "train_loss_llm": 0.0, "grad_norm": 0.061834435909986496, "global_step": 21490, "epoch": 596, "lr": 0.001} +{"train_loss": 0.03268069944654902, "train_loss_bc": 0.0368766151368618, "train_loss_llm": 0.0, "grad_norm": 0.09228896349668503, "global_step": 21491, "epoch": 596, "lr": 0.001, "val_loss": 0.08245407044887543} +{"train_loss": 0.036306485533714294, "train_loss_bc": 0.036306485533714294, "train_loss_llm": 0.0, "grad_norm": 0.11303053796291351, "global_step": 21492, "epoch": 597, "lr": 0.001} +{"train_loss": 0.030078008770942688, "train_loss_bc": 0.03006565384566784, "train_loss_llm": 0.0012355446815490723, "grad_norm": 0.03625849261879921, "global_step": 21493, "epoch": 597, "lr": 0.001} +{"train_loss": 0.038220904767513275, "train_loss_bc": 0.038220904767513275, "train_loss_llm": 0.0, "grad_norm": 0.06279419362545013, "global_step": 21494, "epoch": 597, "lr": 0.001} +{"train_loss": 0.030606742948293686, "train_loss_bc": 0.030600693076848984, "train_loss_llm": 0.0006050169467926025, "grad_norm": 0.07493456453084946, "global_step": 21495, "epoch": 597, "lr": 0.001} +{"train_loss": 0.032989658415317535, "train_loss_bc": 0.032989658415317535, "train_loss_llm": 0.0, "grad_norm": 0.08716464787721634, "global_step": 21496, "epoch": 597, "lr": 0.001} +{"train_loss": 0.031598255038261414, "train_loss_bc": 0.03157629817724228, "train_loss_llm": 0.0021957457065582275, "grad_norm": 0.02937163971364498, "global_step": 21497, "epoch": 597, "lr": 0.001} +{"train_loss": 0.034904997795820236, "train_loss_bc": 0.034904997795820236, "train_loss_llm": 0.0, "grad_norm": 0.046325813978910446, "global_step": 21498, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03410620242357254, "train_loss_bc": 0.03410620242357254, "train_loss_llm": 0.0, "grad_norm": 0.0709274411201477, "global_step": 21499, "epoch": 597, "lr": 0.001} +{"train_loss": 0.033962078392505646, "train_loss_bc": 0.033962078392505646, "train_loss_llm": 0.0, "grad_norm": 0.08404171466827393, "global_step": 21500, "epoch": 597, "lr": 0.001} +{"train_loss": 0.029781797900795937, "train_loss_bc": 0.02976841852068901, "train_loss_llm": 0.001337975263595581, "grad_norm": 0.024419395253062248, "global_step": 21501, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03382957726716995, "train_loss_bc": 0.03382957726716995, "train_loss_llm": 0.0, "grad_norm": 0.047658320516347885, "global_step": 21502, "epoch": 597, "lr": 0.001} +{"train_loss": 0.038807518780231476, "train_loss_bc": 0.038807518780231476, "train_loss_llm": 0.0, "grad_norm": 0.07274764031171799, "global_step": 21503, "epoch": 597, "lr": 0.001} +{"train_loss": 0.0310249961912632, "train_loss_bc": 0.0310249961912632, "train_loss_llm": 0.0, "grad_norm": 0.08643937855958939, "global_step": 21504, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03483175113797188, "train_loss_bc": 0.03483175113797188, "train_loss_llm": 0.0, "grad_norm": 0.03487182781100273, "global_step": 21505, "epoch": 597, "lr": 0.001} +{"train_loss": 0.028360186144709587, "train_loss_bc": 0.028360186144709587, "train_loss_llm": 0.0, "grad_norm": 0.054486680775880814, "global_step": 21506, "epoch": 597, "lr": 0.001} +{"train_loss": 0.029677286744117737, "train_loss_bc": 0.029677286744117737, "train_loss_llm": 0.0, "grad_norm": 0.06953095644712448, "global_step": 21507, "epoch": 597, "lr": 0.001} +{"train_loss": 0.030967891216278076, "train_loss_bc": 0.03096311166882515, "train_loss_llm": 0.0004779994487762451, "grad_norm": 0.08487503975629807, "global_step": 21508, "epoch": 597, "lr": 0.001} +{"train_loss": 0.02787862718105316, "train_loss_bc": 0.02787862718105316, "train_loss_llm": 0.0, "grad_norm": 0.031598050147295, "global_step": 21509, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03703051432967186, "train_loss_bc": 0.03703051432967186, "train_loss_llm": 0.0, "grad_norm": 0.049469199031591415, "global_step": 21510, "epoch": 597, "lr": 0.001} +{"train_loss": 0.02864931896328926, "train_loss_bc": 0.02864931896328926, "train_loss_llm": 0.0, "grad_norm": 0.0645337924361229, "global_step": 21511, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03941332548856735, "train_loss_bc": 0.03941332548856735, "train_loss_llm": 0.0, "grad_norm": 0.08519376814365387, "global_step": 21512, "epoch": 597, "lr": 0.001} +{"train_loss": 0.036305177956819534, "train_loss_bc": 0.036305177956819534, "train_loss_llm": 0.0, "grad_norm": 0.03441811352968216, "global_step": 21513, "epoch": 597, "lr": 0.001} +{"train_loss": 0.028917517513036728, "train_loss_bc": 0.028917517513036728, "train_loss_llm": 0.0, "grad_norm": 0.05154294893145561, "global_step": 21514, "epoch": 597, "lr": 0.001} +{"train_loss": 0.034345775842666626, "train_loss_bc": 0.034345775842666626, "train_loss_llm": 0.0, "grad_norm": 0.07044245302677155, "global_step": 21515, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03576485440135002, "train_loss_bc": 0.03576485440135002, "train_loss_llm": 0.0, "grad_norm": 0.09270699322223663, "global_step": 21516, "epoch": 597, "lr": 0.001} +{"train_loss": 0.029879817739129066, "train_loss_bc": 0.02986265905201435, "train_loss_llm": 0.00171583890914917, "grad_norm": 0.029449980705976486, "global_step": 21517, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03525562584400177, "train_loss_bc": 0.03525562584400177, "train_loss_llm": 0.0, "grad_norm": 0.05015046149492264, "global_step": 21518, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03175114840269089, "train_loss_bc": 0.03175114840269089, "train_loss_llm": 0.0, "grad_norm": 0.06490634381771088, "global_step": 21519, "epoch": 597, "lr": 0.001} +{"train_loss": 0.033604271709918976, "train_loss_bc": 0.033604271709918976, "train_loss_llm": 0.0, "grad_norm": 0.08944752812385559, "global_step": 21520, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03196985647082329, "train_loss_bc": 0.031962212175130844, "train_loss_llm": 0.0007645785808563232, "grad_norm": 0.02858765423297882, "global_step": 21521, "epoch": 597, "lr": 0.001} +{"train_loss": 0.0267754215747118, "train_loss_bc": 0.0267754215747118, "train_loss_llm": 0.0, "grad_norm": 0.04004374518990517, "global_step": 21522, "epoch": 597, "lr": 0.001} +{"train_loss": 0.029325351119041443, "train_loss_bc": 0.02932184748351574, "train_loss_llm": 0.0003503561019897461, "grad_norm": 0.05019212141633034, "global_step": 21523, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03437890112400055, "train_loss_bc": 0.03437890112400055, "train_loss_llm": 0.0, "grad_norm": 0.06165755167603493, "global_step": 21524, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03517281264066696, "train_loss_bc": 0.03517281264066696, "train_loss_llm": 0.0, "grad_norm": 0.035727377980947495, "global_step": 21525, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03331995755434036, "train_loss_bc": 0.03330688923597336, "train_loss_llm": 0.0013067424297332764, "grad_norm": 0.04243231564760208, "global_step": 21526, "epoch": 597, "lr": 0.001} +{"train_loss": 0.03266838379204273, "train_loss_bc": 0.02626788429915905, "train_loss_llm": 0.00013175606727600098, "grad_norm": 0.057932429015636444, "global_step": 21527, "epoch": 597, "lr": 0.001, "val_loss": 0.08290562033653259} +{"train_loss": 0.0321849100291729, "train_loss_bc": 0.0321849100291729, "train_loss_llm": 0.0, "grad_norm": 0.07386963069438934, "global_step": 21528, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03202347829937935, "train_loss_bc": 0.03200386464595795, "train_loss_llm": 0.001961320638656616, "grad_norm": 0.03347386047244072, "global_step": 21529, "epoch": 598, "lr": 0.001} +{"train_loss": 0.030105944722890854, "train_loss_bc": 0.030105944722890854, "train_loss_llm": 0.0, "grad_norm": 0.0515955425798893, "global_step": 21530, "epoch": 598, "lr": 0.001} +{"train_loss": 0.02887308970093727, "train_loss_bc": 0.02887308970093727, "train_loss_llm": 0.0, "grad_norm": 0.06992534548044205, "global_step": 21531, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03482756018638611, "train_loss_bc": 0.03482756018638611, "train_loss_llm": 0.0, "grad_norm": 0.09465758502483368, "global_step": 21532, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03161634877324104, "train_loss_bc": 0.03160592541098595, "train_loss_llm": 0.0010423362255096436, "grad_norm": 0.040853437036275864, "global_step": 21533, "epoch": 598, "lr": 0.001} +{"train_loss": 0.037755466997623444, "train_loss_bc": 0.037747327238321304, "train_loss_llm": 0.0008138418197631836, "grad_norm": 0.07271885126829147, "global_step": 21534, "epoch": 598, "lr": 0.001} +{"train_loss": 0.02852352149784565, "train_loss_bc": 0.02851230278611183, "train_loss_llm": 0.00112190842628479, "grad_norm": 0.08664926141500473, "global_step": 21535, "epoch": 598, "lr": 0.001} +{"train_loss": 0.029474174603819847, "train_loss_bc": 0.029465924948453903, "train_loss_llm": 0.0008249878883361816, "grad_norm": 0.0955418199300766, "global_step": 21536, "epoch": 598, "lr": 0.001} +{"train_loss": 0.034997738897800446, "train_loss_bc": 0.034997738897800446, "train_loss_llm": 0.0, "grad_norm": 0.023335909470915794, "global_step": 21537, "epoch": 598, "lr": 0.001} +{"train_loss": 0.0394420400261879, "train_loss_bc": 0.0394420400261879, "train_loss_llm": 0.0, "grad_norm": 0.041071414947509766, "global_step": 21538, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03419923037290573, "train_loss_bc": 0.03419923037290573, "train_loss_llm": 0.0, "grad_norm": 0.05314048007130623, "global_step": 21539, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03200212121009827, "train_loss_bc": 0.03200212121009827, "train_loss_llm": 0.0, "grad_norm": 0.07308081537485123, "global_step": 21540, "epoch": 598, "lr": 0.001} +{"train_loss": 0.0335073247551918, "train_loss_bc": 0.0335073247551918, "train_loss_llm": 0.0, "grad_norm": 0.03370774909853935, "global_step": 21541, "epoch": 598, "lr": 0.001} +{"train_loss": 0.035252489149570465, "train_loss_bc": 0.035252489149570465, "train_loss_llm": 0.0, "grad_norm": 0.05890725925564766, "global_step": 21542, "epoch": 598, "lr": 0.001} +{"train_loss": 0.034368302673101425, "train_loss_bc": 0.034368302673101425, "train_loss_llm": 0.0, "grad_norm": 0.07092662900686264, "global_step": 21543, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03379199653863907, "train_loss_bc": 0.03379199653863907, "train_loss_llm": 0.0, "grad_norm": 0.07962267845869064, "global_step": 21544, "epoch": 598, "lr": 0.001} +{"train_loss": 0.027328984811902046, "train_loss_bc": 0.027328984811902046, "train_loss_llm": 0.0, "grad_norm": 0.026871101930737495, "global_step": 21545, "epoch": 598, "lr": 0.001} +{"train_loss": 0.023489227518439293, "train_loss_bc": 0.023489227518439293, "train_loss_llm": 0.0, "grad_norm": 0.03539503365755081, "global_step": 21546, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03303717076778412, "train_loss_bc": 0.03303717076778412, "train_loss_llm": 0.0, "grad_norm": 0.05064201354980469, "global_step": 21547, "epoch": 598, "lr": 0.001} +{"train_loss": 0.030079152435064316, "train_loss_bc": 0.030079152435064316, "train_loss_llm": 0.0, "grad_norm": 0.05993545427918434, "global_step": 21548, "epoch": 598, "lr": 0.001} +{"train_loss": 0.0311944168061018, "train_loss_bc": 0.0311944168061018, "train_loss_llm": 0.0, "grad_norm": 0.035373441874980927, "global_step": 21549, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03437914699316025, "train_loss_bc": 0.034364305436611176, "train_loss_llm": 0.0014840960502624512, "grad_norm": 0.048507437109947205, "global_step": 21550, "epoch": 598, "lr": 0.001} +{"train_loss": 0.034449510276317596, "train_loss_bc": 0.03443361446261406, "train_loss_llm": 0.001589447259902954, "grad_norm": 0.05983247235417366, "global_step": 21551, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03172769770026207, "train_loss_bc": 0.03172769770026207, "train_loss_llm": 0.0, "grad_norm": 0.0719694271683693, "global_step": 21552, "epoch": 598, "lr": 0.001} +{"train_loss": 0.0425095334649086, "train_loss_bc": 0.04249722138047218, "train_loss_llm": 0.0012312829494476318, "grad_norm": 0.04000895097851753, "global_step": 21553, "epoch": 598, "lr": 0.001} +{"train_loss": 0.028335964307188988, "train_loss_bc": 0.028335964307188988, "train_loss_llm": 0.0, "grad_norm": 0.05604934319853783, "global_step": 21554, "epoch": 598, "lr": 0.001} +{"train_loss": 0.0371854268014431, "train_loss_bc": 0.0371854268014431, "train_loss_llm": 0.0, "grad_norm": 0.0816071629524231, "global_step": 21555, "epoch": 598, "lr": 0.001} +{"train_loss": 0.028965892270207405, "train_loss_bc": 0.028965892270207405, "train_loss_llm": 0.0, "grad_norm": 0.09783639013767242, "global_step": 21556, "epoch": 598, "lr": 0.001} +{"train_loss": 0.028851693496108055, "train_loss_bc": 0.028851693496108055, "train_loss_llm": 0.0, "grad_norm": 0.02740999311208725, "global_step": 21557, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03174784034490585, "train_loss_bc": 0.03174766153097153, "train_loss_llm": 1.7702579498291016e-05, "grad_norm": 0.04321230202913284, "global_step": 21558, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03665245696902275, "train_loss_bc": 0.03665245696902275, "train_loss_llm": 0.0, "grad_norm": 0.06315486133098602, "global_step": 21559, "epoch": 598, "lr": 0.001} +{"train_loss": 0.03447258472442627, "train_loss_bc": 0.03447258472442627, "train_loss_llm": 0.0, "grad_norm": 0.08916432410478592, "global_step": 21560, "epoch": 598, "lr": 0.001} +{"train_loss": 0.035655777901411057, "train_loss_bc": 0.035655777901411057, "train_loss_llm": 0.0, "grad_norm": 0.03040125034749508, "global_step": 21561, "epoch": 598, "lr": 0.001} +{"train_loss": 0.033793747425079346, "train_loss_bc": 0.033793747425079346, "train_loss_llm": 0.0, "grad_norm": 0.05656247213482857, "global_step": 21562, "epoch": 598, "lr": 0.001} +{"train_loss": 0.032752865097588964, "train_loss_bc": 0.03230118006467819, "train_loss_llm": 0.0, "grad_norm": 0.08721870929002762, "global_step": 21563, "epoch": 598, "lr": 0.001, "val_loss": 0.08174650371074677} +{"train_loss": 0.027772512286901474, "train_loss_bc": 0.027772512286901474, "train_loss_llm": 0.0, "grad_norm": 0.10060080885887146, "global_step": 21564, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03305008262395859, "train_loss_bc": 0.03305008262395859, "train_loss_llm": 0.0, "grad_norm": 0.034957241266965866, "global_step": 21565, "epoch": 599, "lr": 0.001} +{"train_loss": 0.02761022374033928, "train_loss_bc": 0.02761022374033928, "train_loss_llm": 0.0, "grad_norm": 0.03918442502617836, "global_step": 21566, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03136991709470749, "train_loss_bc": 0.03136991709470749, "train_loss_llm": 0.0, "grad_norm": 0.048941727727651596, "global_step": 21567, "epoch": 599, "lr": 0.001} +{"train_loss": 0.030741287395358086, "train_loss_bc": 0.03072209842503071, "train_loss_llm": 0.0019189715385437012, "grad_norm": 0.05827491730451584, "global_step": 21568, "epoch": 599, "lr": 0.001} +{"train_loss": 0.02990897186100483, "train_loss_bc": 0.02990897186100483, "train_loss_llm": 0.0, "grad_norm": 0.03429586440324783, "global_step": 21569, "epoch": 599, "lr": 0.001} +{"train_loss": 0.030810102820396423, "train_loss_bc": 0.030810102820396423, "train_loss_llm": 0.0, "grad_norm": 0.05323394015431404, "global_step": 21570, "epoch": 599, "lr": 0.001} +{"train_loss": 0.02716902084648609, "train_loss_bc": 0.02715720422565937, "train_loss_llm": 0.0011816024780273438, "grad_norm": 0.06536085903644562, "global_step": 21571, "epoch": 599, "lr": 0.001} +{"train_loss": 0.0313413143157959, "train_loss_bc": 0.0313413143157959, "train_loss_llm": 0.0, "grad_norm": 0.0842951312661171, "global_step": 21572, "epoch": 599, "lr": 0.001} +{"train_loss": 0.034663714468479156, "train_loss_bc": 0.034663714468479156, "train_loss_llm": 0.0, "grad_norm": 0.033837731927633286, "global_step": 21573, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03653968498110771, "train_loss_bc": 0.03653968498110771, "train_loss_llm": 0.0, "grad_norm": 0.05757445842027664, "global_step": 21574, "epoch": 599, "lr": 0.001} +{"train_loss": 0.031108899042010307, "train_loss_bc": 0.031108899042010307, "train_loss_llm": 0.0, "grad_norm": 0.07761716842651367, "global_step": 21575, "epoch": 599, "lr": 0.001} +{"train_loss": 0.034542154520750046, "train_loss_bc": 0.034542154520750046, "train_loss_llm": 0.0, "grad_norm": 0.09370429813861847, "global_step": 21576, "epoch": 599, "lr": 0.001} +{"train_loss": 0.0330699123442173, "train_loss_bc": 0.0330699123442173, "train_loss_llm": 0.0, "grad_norm": 0.03783205896615982, "global_step": 21577, "epoch": 599, "lr": 0.001} +{"train_loss": 0.036031294614076614, "train_loss_bc": 0.036031294614076614, "train_loss_llm": 0.0, "grad_norm": 0.06071098893880844, "global_step": 21578, "epoch": 599, "lr": 0.001} +{"train_loss": 0.02655402198433876, "train_loss_bc": 0.02655402198433876, "train_loss_llm": 0.0, "grad_norm": 0.0894094854593277, "global_step": 21579, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03032880835235119, "train_loss_bc": 0.030319634824991226, "train_loss_llm": 0.0009173154830932617, "grad_norm": 0.10415655374526978, "global_step": 21580, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03258131444454193, "train_loss_bc": 0.03258131444454193, "train_loss_llm": 0.0, "grad_norm": 0.027698827907443047, "global_step": 21581, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03135326877236366, "train_loss_bc": 0.03134346008300781, "train_loss_llm": 0.0009808242321014404, "grad_norm": 0.04044583812355995, "global_step": 21582, "epoch": 599, "lr": 0.001} +{"train_loss": 0.030665121972560883, "train_loss_bc": 0.030665121972560883, "train_loss_llm": 0.0, "grad_norm": 0.0587264746427536, "global_step": 21583, "epoch": 599, "lr": 0.001} +{"train_loss": 0.034381162375211716, "train_loss_bc": 0.034381162375211716, "train_loss_llm": 0.0, "grad_norm": 0.07583355903625488, "global_step": 21584, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03277484327554703, "train_loss_bc": 0.03277484327554703, "train_loss_llm": 0.0, "grad_norm": 0.02646489441394806, "global_step": 21585, "epoch": 599, "lr": 0.001} +{"train_loss": 0.02713613584637642, "train_loss_bc": 0.02713613584637642, "train_loss_llm": 0.0, "grad_norm": 0.03842820227146149, "global_step": 21586, "epoch": 599, "lr": 0.001} +{"train_loss": 0.04403886944055557, "train_loss_bc": 0.04403886944055557, "train_loss_llm": 0.0, "grad_norm": 0.05516975373029709, "global_step": 21587, "epoch": 599, "lr": 0.001} +{"train_loss": 0.025276482105255127, "train_loss_bc": 0.025276482105255127, "train_loss_llm": 0.0, "grad_norm": 0.06772512197494507, "global_step": 21588, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03503038361668587, "train_loss_bc": 0.03503038361668587, "train_loss_llm": 0.0, "grad_norm": 0.027325481176376343, "global_step": 21589, "epoch": 599, "lr": 0.001} +{"train_loss": 0.04026014357805252, "train_loss_bc": 0.0402577742934227, "train_loss_llm": 0.00023689866065979004, "grad_norm": 0.05276843532919884, "global_step": 21590, "epoch": 599, "lr": 0.001} +{"train_loss": 0.030454397201538086, "train_loss_bc": 0.030453424900770187, "train_loss_llm": 9.718537330627441e-05, "grad_norm": 0.060971785336732864, "global_step": 21591, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03135579079389572, "train_loss_bc": 0.03135579079389572, "train_loss_llm": 0.0, "grad_norm": 0.06980614364147186, "global_step": 21592, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03257997706532478, "train_loss_bc": 0.03257997706532478, "train_loss_llm": 0.0, "grad_norm": 0.03435609117150307, "global_step": 21593, "epoch": 599, "lr": 0.001} +{"train_loss": 0.034951210021972656, "train_loss_bc": 0.03494676202535629, "train_loss_llm": 0.0004449784755706787, "grad_norm": 0.05853515490889549, "global_step": 21594, "epoch": 599, "lr": 0.001} +{"train_loss": 0.041999250650405884, "train_loss_bc": 0.041999250650405884, "train_loss_llm": 0.0, "grad_norm": 0.07941673696041107, "global_step": 21595, "epoch": 599, "lr": 0.001} +{"train_loss": 0.040611643344163895, "train_loss_bc": 0.040611643344163895, "train_loss_llm": 0.0, "grad_norm": 0.10502005368471146, "global_step": 21596, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03328140452504158, "train_loss_bc": 0.03328140452504158, "train_loss_llm": 0.0, "grad_norm": 0.033973269164562225, "global_step": 21597, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03489362448453903, "train_loss_bc": 0.03486336022615433, "train_loss_llm": 0.003026425838470459, "grad_norm": 0.06217200309038162, "global_step": 21598, "epoch": 599, "lr": 0.001} +{"train_loss": 0.03269687822709481, "train_loss_bc": 0.030850669369101524, "train_loss_llm": 0.0, "grad_norm": 0.0848701000213623, "global_step": 21599, "epoch": 599, "lr": 0.001, "val_loss": 0.08208715915679932} +{"train_loss": 0.03929537162184715, "train_loss_bc": 0.03929537162184715, "train_loss_llm": 0.0, "grad_norm": 0.1017720177769661, "global_step": 21600, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03397919982671738, "train_loss_bc": 0.03395683690905571, "train_loss_llm": 0.0022362470626831055, "grad_norm": 0.03670363128185272, "global_step": 21601, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03413143754005432, "train_loss_bc": 0.034130048006772995, "train_loss_llm": 0.00013899803161621094, "grad_norm": 0.06514200568199158, "global_step": 21602, "epoch": 600, "lr": 0.001} +{"train_loss": 0.033112891018390656, "train_loss_bc": 0.03310638666152954, "train_loss_llm": 0.0006504654884338379, "grad_norm": 0.08179651945829391, "global_step": 21603, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03094233013689518, "train_loss_bc": 0.03094233013689518, "train_loss_llm": 0.0, "grad_norm": 0.09686525166034698, "global_step": 21604, "epoch": 600, "lr": 0.001} +{"train_loss": 0.0369766503572464, "train_loss_bc": 0.0369766503572464, "train_loss_llm": 0.0, "grad_norm": 0.031388428062200546, "global_step": 21605, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03356495499610901, "train_loss_bc": 0.03355861455202103, "train_loss_llm": 0.0006341934204101562, "grad_norm": 0.03744347020983696, "global_step": 21606, "epoch": 600, "lr": 0.001} +{"train_loss": 0.030833320692181587, "train_loss_bc": 0.030833320692181587, "train_loss_llm": 0.0, "grad_norm": 0.048889704048633575, "global_step": 21607, "epoch": 600, "lr": 0.001} +{"train_loss": 0.028566503897309303, "train_loss_bc": 0.028547341004014015, "train_loss_llm": 0.001916348934173584, "grad_norm": 0.057738933712244034, "global_step": 21608, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03457992523908615, "train_loss_bc": 0.03455361723899841, "train_loss_llm": 0.002630949020385742, "grad_norm": 0.03375992178916931, "global_step": 21609, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03238881006836891, "train_loss_bc": 0.03238881006836891, "train_loss_llm": 0.0, "grad_norm": 0.05585077777504921, "global_step": 21610, "epoch": 600, "lr": 0.001} +{"train_loss": 0.030789345502853394, "train_loss_bc": 0.030786190181970596, "train_loss_llm": 0.0003155171871185303, "grad_norm": 0.06617995351552963, "global_step": 21611, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03463700786232948, "train_loss_bc": 0.03463700786232948, "train_loss_llm": 0.0, "grad_norm": 0.07854195684194565, "global_step": 21612, "epoch": 600, "lr": 0.001} +{"train_loss": 0.028145622462034225, "train_loss_bc": 0.028142672032117844, "train_loss_llm": 0.0002950727939605713, "grad_norm": 0.025419006124138832, "global_step": 21613, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03286027908325195, "train_loss_bc": 0.03286027908325195, "train_loss_llm": 0.0, "grad_norm": 0.053761743009090424, "global_step": 21614, "epoch": 600, "lr": 0.001} +{"train_loss": 0.02857288159430027, "train_loss_bc": 0.02857288159430027, "train_loss_llm": 0.0, "grad_norm": 0.07469689846038818, "global_step": 21615, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03404537960886955, "train_loss_bc": 0.034018196165561676, "train_loss_llm": 0.0027181804180145264, "grad_norm": 0.1051085814833641, "global_step": 21616, "epoch": 600, "lr": 0.001} +{"train_loss": 0.030117154121398926, "train_loss_bc": 0.030117154121398926, "train_loss_llm": 0.0, "grad_norm": 0.03398147597908974, "global_step": 21617, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03023291938006878, "train_loss_bc": 0.03023291938006878, "train_loss_llm": 0.0, "grad_norm": 0.05455645173788071, "global_step": 21618, "epoch": 600, "lr": 0.001} +{"train_loss": 0.031987227499485016, "train_loss_bc": 0.031987227499485016, "train_loss_llm": 0.0, "grad_norm": 0.08366365730762482, "global_step": 21619, "epoch": 600, "lr": 0.001} +{"train_loss": 0.0358256958425045, "train_loss_bc": 0.0358256958425045, "train_loss_llm": 0.0, "grad_norm": 0.10209298878908157, "global_step": 21620, "epoch": 600, "lr": 0.001} +{"train_loss": 0.035012438893318176, "train_loss_bc": 0.035012438893318176, "train_loss_llm": 0.0, "grad_norm": 0.03738611936569214, "global_step": 21621, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03192894905805588, "train_loss_bc": 0.03192894905805588, "train_loss_llm": 0.0, "grad_norm": 0.04614603519439697, "global_step": 21622, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03024953603744507, "train_loss_bc": 0.03024953603744507, "train_loss_llm": 0.0, "grad_norm": 0.0701102614402771, "global_step": 21623, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03394926339387894, "train_loss_bc": 0.03393565118312836, "train_loss_llm": 0.0013610422611236572, "grad_norm": 0.08990394324064255, "global_step": 21624, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03618357703089714, "train_loss_bc": 0.03618357703089714, "train_loss_llm": 0.0, "grad_norm": 0.03169569745659828, "global_step": 21625, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03690801188349724, "train_loss_bc": 0.03690801188349724, "train_loss_llm": 0.0, "grad_norm": 0.0547255240380764, "global_step": 21626, "epoch": 600, "lr": 0.001} +{"train_loss": 0.035604383796453476, "train_loss_bc": 0.035604383796453476, "train_loss_llm": 0.0, "grad_norm": 0.06687598675489426, "global_step": 21627, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03328019380569458, "train_loss_bc": 0.03328019380569458, "train_loss_llm": 0.0, "grad_norm": 0.07800491154193878, "global_step": 21628, "epoch": 600, "lr": 0.001} +{"train_loss": 0.02907792665064335, "train_loss_bc": 0.02907792665064335, "train_loss_llm": 0.0, "grad_norm": 0.029932759702205658, "global_step": 21629, "epoch": 600, "lr": 0.001} +{"train_loss": 0.0333758108317852, "train_loss_bc": 0.0333758108317852, "train_loss_llm": 0.0, "grad_norm": 0.05272921547293663, "global_step": 21630, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03559405356645584, "train_loss_bc": 0.03559405356645584, "train_loss_llm": 0.0, "grad_norm": 0.060857608914375305, "global_step": 21631, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03219782933592796, "train_loss_bc": 0.03219782933592796, "train_loss_llm": 0.0, "grad_norm": 0.07557152956724167, "global_step": 21632, "epoch": 600, "lr": 0.001} +{"train_loss": 0.03115583024919033, "train_loss_bc": 0.03115583024919033, "train_loss_llm": 0.0, "grad_norm": 0.046596135944128036, "global_step": 21633, "epoch": 600, "lr": 0.001} +{"train_loss": 0.029765736311674118, "train_loss_bc": 0.029765736311674118, "train_loss_llm": 0.0, "grad_norm": 0.06728345900774002, "global_step": 21634, "epoch": 600, "lr": 0.001} +{"train_loss": 0.0326949831408759, "train_loss_bc": 0.027150943875312805, "train_loss_llm": 0.0, "grad_norm": 0.08937343209981918, "global_step": 21635, "epoch": 600, "lr": 0.001, "train/cumulative_reward": 89.6904914146699, "train/mean_score": 22.934628649347008, "train/success_rate": 0.6, "test/cumulative_reward": 28.32772767794576, "test/mean_score": 7.842188165153578, "test/success_rate": 0.22, "val_loss": 0.08368716388940811, "train_action_mse_error": 0.036202769726514816} +{"train_loss": 0.033253271132707596, "train_loss_bc": 0.033249519765377045, "train_loss_llm": 0.0003751516342163086, "grad_norm": 0.1070769727230072, "global_step": 21636, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03317449614405632, "train_loss_bc": 0.03317449614405632, "train_loss_llm": 0.0, "grad_norm": 0.03763333335518837, "global_step": 21637, "epoch": 601, "lr": 0.001} +{"train_loss": 0.034550126641988754, "train_loss_bc": 0.034550126641988754, "train_loss_llm": 0.0, "grad_norm": 0.06846597790718079, "global_step": 21638, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03546404093503952, "train_loss_bc": 0.03546404093503952, "train_loss_llm": 0.0, "grad_norm": 0.08248955011367798, "global_step": 21639, "epoch": 601, "lr": 0.001} +{"train_loss": 0.033794429153203964, "train_loss_bc": 0.03378886729478836, "train_loss_llm": 0.0005561709403991699, "grad_norm": 0.09411268681287766, "global_step": 21640, "epoch": 601, "lr": 0.001} +{"train_loss": 0.029550891369581223, "train_loss_bc": 0.029550891369581223, "train_loss_llm": 0.0, "grad_norm": 0.030010677874088287, "global_step": 21641, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03136347606778145, "train_loss_bc": 0.03134673088788986, "train_loss_llm": 0.001674562692642212, "grad_norm": 0.03884660452604294, "global_step": 21642, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03306181728839874, "train_loss_bc": 0.03305470943450928, "train_loss_llm": 0.0007109642028808594, "grad_norm": 0.06281069666147232, "global_step": 21643, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03446802869439125, "train_loss_bc": 0.03446599096059799, "train_loss_llm": 0.00020384788513183594, "grad_norm": 0.07801687717437744, "global_step": 21644, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03381258249282837, "train_loss_bc": 0.03381258249282837, "train_loss_llm": 0.0, "grad_norm": 0.02761409431695938, "global_step": 21645, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03281499817967415, "train_loss_bc": 0.03281499817967415, "train_loss_llm": 0.0, "grad_norm": 0.040333058685064316, "global_step": 21646, "epoch": 601, "lr": 0.001} +{"train_loss": 0.035500772297382355, "train_loss_bc": 0.03549608588218689, "train_loss_llm": 0.0004685819149017334, "grad_norm": 0.05139879882335663, "global_step": 21647, "epoch": 601, "lr": 0.001} +{"train_loss": 0.02664400264620781, "train_loss_bc": 0.02664400264620781, "train_loss_llm": 0.0, "grad_norm": 0.0662795677781105, "global_step": 21648, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03669776767492294, "train_loss_bc": 0.03668979927897453, "train_loss_llm": 0.0007966756820678711, "grad_norm": 0.036087315529584885, "global_step": 21649, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03153638914227486, "train_loss_bc": 0.03153638914227486, "train_loss_llm": 0.0, "grad_norm": 0.04174249619245529, "global_step": 21650, "epoch": 601, "lr": 0.001} +{"train_loss": 0.029175616800785065, "train_loss_bc": 0.029175616800785065, "train_loss_llm": 0.0, "grad_norm": 0.05012477934360504, "global_step": 21651, "epoch": 601, "lr": 0.001} +{"train_loss": 0.028097305446863174, "train_loss_bc": 0.028097305446863174, "train_loss_llm": 0.0, "grad_norm": 0.06371303647756577, "global_step": 21652, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03204634040594101, "train_loss_bc": 0.03204634040594101, "train_loss_llm": 0.0, "grad_norm": 0.028693947941064835, "global_step": 21653, "epoch": 601, "lr": 0.001} +{"train_loss": 0.02685818448662758, "train_loss_bc": 0.026844680309295654, "train_loss_llm": 0.001350492238998413, "grad_norm": 0.04217953234910965, "global_step": 21654, "epoch": 601, "lr": 0.001} +{"train_loss": 0.0299836378544569, "train_loss_bc": 0.0299836378544569, "train_loss_llm": 0.0, "grad_norm": 0.060762662440538406, "global_step": 21655, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03629628196358681, "train_loss_bc": 0.03629628196358681, "train_loss_llm": 0.0, "grad_norm": 0.07912857085466385, "global_step": 21656, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03309063985943794, "train_loss_bc": 0.03309063985943794, "train_loss_llm": 0.0, "grad_norm": 0.028824390843510628, "global_step": 21657, "epoch": 601, "lr": 0.001} +{"train_loss": 0.026320334523916245, "train_loss_bc": 0.026309218257665634, "train_loss_llm": 0.0011115670204162598, "grad_norm": 0.04498526453971863, "global_step": 21658, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03210870549082756, "train_loss_bc": 0.03210870549082756, "train_loss_llm": 0.0, "grad_norm": 0.06104021519422531, "global_step": 21659, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03251313790678978, "train_loss_bc": 0.03251313790678978, "train_loss_llm": 0.0, "grad_norm": 0.08067528158426285, "global_step": 21660, "epoch": 601, "lr": 0.001} +{"train_loss": 0.033995140343904495, "train_loss_bc": 0.033995140343904495, "train_loss_llm": 0.0, "grad_norm": 0.0267783310264349, "global_step": 21661, "epoch": 601, "lr": 0.001} +{"train_loss": 0.035735122859478, "train_loss_bc": 0.035735122859478, "train_loss_llm": 0.0, "grad_norm": 0.046388424932956696, "global_step": 21662, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03622184693813324, "train_loss_bc": 0.03622184693813324, "train_loss_llm": 0.0, "grad_norm": 0.06499843299388885, "global_step": 21663, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03337300196290016, "train_loss_bc": 0.033365435898303986, "train_loss_llm": 0.0007567107677459717, "grad_norm": 0.09062167257070541, "global_step": 21664, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03412451595067978, "train_loss_bc": 0.03412451595067978, "train_loss_llm": 0.0, "grad_norm": 0.037128448486328125, "global_step": 21665, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03526471555233002, "train_loss_bc": 0.03526471555233002, "train_loss_llm": 0.0, "grad_norm": 0.05045183002948761, "global_step": 21666, "epoch": 601, "lr": 0.001} +{"train_loss": 0.033549364656209946, "train_loss_bc": 0.033549364656209946, "train_loss_llm": 0.0, "grad_norm": 0.06788438558578491, "global_step": 21667, "epoch": 601, "lr": 0.001} +{"train_loss": 0.02742372825741768, "train_loss_bc": 0.027422770857810974, "train_loss_llm": 9.569525718688965e-05, "grad_norm": 0.08132654428482056, "global_step": 21668, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03692895174026489, "train_loss_bc": 0.036916010081768036, "train_loss_llm": 0.0012940764427185059, "grad_norm": 0.03522127866744995, "global_step": 21669, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03273054584860802, "train_loss_bc": 0.03273054584860802, "train_loss_llm": 0.0, "grad_norm": 0.05329761281609535, "global_step": 21670, "epoch": 601, "lr": 0.001} +{"train_loss": 0.03259193064231011, "train_loss_bc": 0.03178263455629349, "train_loss_llm": 0.0002659261226654053, "grad_norm": 0.06728111952543259, "global_step": 21671, "epoch": 601, "lr": 0.001, "val_loss": 0.08217642456293106} +{"train_loss": 0.03474223241209984, "train_loss_bc": 0.03474223241209984, "train_loss_llm": 0.0, "grad_norm": 0.08410225063562393, "global_step": 21672, "epoch": 602, "lr": 0.001} +{"train_loss": 0.029517242684960365, "train_loss_bc": 0.029517242684960365, "train_loss_llm": 0.0, "grad_norm": 0.035071298480033875, "global_step": 21673, "epoch": 602, "lr": 0.001} +{"train_loss": 0.030923301354050636, "train_loss_bc": 0.030923301354050636, "train_loss_llm": 0.0, "grad_norm": 0.05567095801234245, "global_step": 21674, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03412376344203949, "train_loss_bc": 0.034116484224796295, "train_loss_llm": 0.0007277727127075195, "grad_norm": 0.0794897973537445, "global_step": 21675, "epoch": 602, "lr": 0.001} +{"train_loss": 0.02788032591342926, "train_loss_bc": 0.02788032591342926, "train_loss_llm": 0.0, "grad_norm": 0.09160684049129486, "global_step": 21676, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03057677671313286, "train_loss_bc": 0.03057153895497322, "train_loss_llm": 0.0005237758159637451, "grad_norm": 0.026478085666894913, "global_step": 21677, "epoch": 602, "lr": 0.001} +{"train_loss": 0.02525288425385952, "train_loss_bc": 0.02525288425385952, "train_loss_llm": 0.0, "grad_norm": 0.03913446143269539, "global_step": 21678, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03339657932519913, "train_loss_bc": 0.03339657932519913, "train_loss_llm": 0.0, "grad_norm": 0.04798052832484245, "global_step": 21679, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03265687823295593, "train_loss_bc": 0.03265687823295593, "train_loss_llm": 0.0, "grad_norm": 0.06267408281564713, "global_step": 21680, "epoch": 602, "lr": 0.001} +{"train_loss": 0.02679903246462345, "train_loss_bc": 0.02679903246462345, "train_loss_llm": 0.0, "grad_norm": 0.02758735977113247, "global_step": 21681, "epoch": 602, "lr": 0.001} +{"train_loss": 0.02936282753944397, "train_loss_bc": 0.02936282753944397, "train_loss_llm": 0.0, "grad_norm": 0.04881851747632027, "global_step": 21682, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03314075618982315, "train_loss_bc": 0.03314075618982315, "train_loss_llm": 0.0, "grad_norm": 0.07360657304525375, "global_step": 21683, "epoch": 602, "lr": 0.001} +{"train_loss": 0.033703964203596115, "train_loss_bc": 0.033703964203596115, "train_loss_llm": 0.0, "grad_norm": 0.08578961342573166, "global_step": 21684, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03497385233640671, "train_loss_bc": 0.03497190773487091, "train_loss_llm": 0.00019437074661254883, "grad_norm": 0.031387653201818466, "global_step": 21685, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03890661522746086, "train_loss_bc": 0.03890661522746086, "train_loss_llm": 0.0, "grad_norm": 0.06613367795944214, "global_step": 21686, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03068256564438343, "train_loss_bc": 0.03068256564438343, "train_loss_llm": 0.0, "grad_norm": 0.07776468247175217, "global_step": 21687, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03152845427393913, "train_loss_bc": 0.03152845427393913, "train_loss_llm": 0.0, "grad_norm": 0.10128427296876907, "global_step": 21688, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03532644733786583, "train_loss_bc": 0.035324692726135254, "train_loss_llm": 0.00017547607421875, "grad_norm": 0.04300309345126152, "global_step": 21689, "epoch": 602, "lr": 0.001} +{"train_loss": 0.045847196131944656, "train_loss_bc": 0.045847196131944656, "train_loss_llm": 0.0, "grad_norm": 0.05714322626590729, "global_step": 21690, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03160158917307854, "train_loss_bc": 0.03160158917307854, "train_loss_llm": 0.0, "grad_norm": 0.06565474718809128, "global_step": 21691, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03351572901010513, "train_loss_bc": 0.03351572901010513, "train_loss_llm": 0.0, "grad_norm": 0.07233566045761108, "global_step": 21692, "epoch": 602, "lr": 0.001} +{"train_loss": 0.028165126219391823, "train_loss_bc": 0.028165126219391823, "train_loss_llm": 0.0, "grad_norm": 0.029000744223594666, "global_step": 21693, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03381508216261864, "train_loss_bc": 0.03381508216261864, "train_loss_llm": 0.0, "grad_norm": 0.03954535350203514, "global_step": 21694, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03248259425163269, "train_loss_bc": 0.0324784480035305, "train_loss_llm": 0.0004145205020904541, "grad_norm": 0.048254042863845825, "global_step": 21695, "epoch": 602, "lr": 0.001} +{"train_loss": 0.04035324975848198, "train_loss_bc": 0.04035324975848198, "train_loss_llm": 0.0, "grad_norm": 0.05749544873833656, "global_step": 21696, "epoch": 602, "lr": 0.001} +{"train_loss": 0.027852369472384453, "train_loss_bc": 0.027852369472384453, "train_loss_llm": 0.0, "grad_norm": 0.025891181081533432, "global_step": 21697, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03415951132774353, "train_loss_bc": 0.03415951132774353, "train_loss_llm": 0.0, "grad_norm": 0.0370854027569294, "global_step": 21698, "epoch": 602, "lr": 0.001} +{"train_loss": 0.030676433816552162, "train_loss_bc": 0.030676433816552162, "train_loss_llm": 0.0, "grad_norm": 0.04949972778558731, "global_step": 21699, "epoch": 602, "lr": 0.001} +{"train_loss": 0.028045527637004852, "train_loss_bc": 0.02803811803460121, "train_loss_llm": 0.0007410049438476562, "grad_norm": 0.061535853892564774, "global_step": 21700, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03028031811118126, "train_loss_bc": 0.03027850016951561, "train_loss_llm": 0.0001818239688873291, "grad_norm": 0.02857905812561512, "global_step": 21701, "epoch": 602, "lr": 0.001} +{"train_loss": 0.029452959075570107, "train_loss_bc": 0.029440781101584435, "train_loss_llm": 0.0012177228927612305, "grad_norm": 0.048611003905534744, "global_step": 21702, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03494250774383545, "train_loss_bc": 0.03494250774383545, "train_loss_llm": 0.0, "grad_norm": 0.06737709790468216, "global_step": 21703, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03226407617330551, "train_loss_bc": 0.03226407617330551, "train_loss_llm": 0.0, "grad_norm": 0.08202303946018219, "global_step": 21704, "epoch": 602, "lr": 0.001} +{"train_loss": 0.04146653413772583, "train_loss_bc": 0.041458260267972946, "train_loss_llm": 0.0008272230625152588, "grad_norm": 0.03717624023556709, "global_step": 21705, "epoch": 602, "lr": 0.001} +{"train_loss": 0.039681095629930496, "train_loss_bc": 0.03965630754828453, "train_loss_llm": 0.00247877836227417, "grad_norm": 0.06338756531476974, "global_step": 21706, "epoch": 602, "lr": 0.001} +{"train_loss": 0.03249087960769733, "train_loss_bc": 0.021575266495347023, "train_loss_llm": 0.0, "grad_norm": 0.08181103318929672, "global_step": 21707, "epoch": 602, "lr": 0.001, "val_loss": 0.08327433466911316} +{"train_loss": 0.03236134350299835, "train_loss_bc": 0.03236134350299835, "train_loss_llm": 0.0, "grad_norm": 0.10131852328777313, "global_step": 21708, "epoch": 603, "lr": 0.001} +{"train_loss": 0.033303793519735336, "train_loss_bc": 0.033303793519735336, "train_loss_llm": 0.0, "grad_norm": 0.04412267729640007, "global_step": 21709, "epoch": 603, "lr": 0.001} +{"train_loss": 0.030288195237517357, "train_loss_bc": 0.030288195237517357, "train_loss_llm": 0.0, "grad_norm": 0.05821238458156586, "global_step": 21710, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03145190700888634, "train_loss_bc": 0.03145190700888634, "train_loss_llm": 0.0, "grad_norm": 0.0844452902674675, "global_step": 21711, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03177740052342415, "train_loss_bc": 0.03177740052342415, "train_loss_llm": 0.0, "grad_norm": 0.09768155962228775, "global_step": 21712, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03313281014561653, "train_loss_bc": 0.03313281014561653, "train_loss_llm": 0.0, "grad_norm": 0.03491000831127167, "global_step": 21713, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03832433000206947, "train_loss_bc": 0.03830356150865555, "train_loss_llm": 0.002076864242553711, "grad_norm": 0.048984095454216, "global_step": 21714, "epoch": 603, "lr": 0.001} +{"train_loss": 0.035460807383060455, "train_loss_bc": 0.035460807383060455, "train_loss_llm": 0.0, "grad_norm": 0.06050533056259155, "global_step": 21715, "epoch": 603, "lr": 0.001} +{"train_loss": 0.031013796105980873, "train_loss_bc": 0.031013796105980873, "train_loss_llm": 0.0, "grad_norm": 0.08447293192148209, "global_step": 21716, "epoch": 603, "lr": 0.001} +{"train_loss": 0.030027659609913826, "train_loss_bc": 0.030027659609913826, "train_loss_llm": 0.0, "grad_norm": 0.03181576728820801, "global_step": 21717, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03654865175485611, "train_loss_bc": 0.03654865175485611, "train_loss_llm": 0.0, "grad_norm": 0.05416331812739372, "global_step": 21718, "epoch": 603, "lr": 0.001} +{"train_loss": 0.029185930266976357, "train_loss_bc": 0.029185930266976357, "train_loss_llm": 0.0, "grad_norm": 0.0652998834848404, "global_step": 21719, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03321128338575363, "train_loss_bc": 0.033200349658727646, "train_loss_llm": 0.001093447208404541, "grad_norm": 0.07631688565015793, "global_step": 21720, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03380442410707474, "train_loss_bc": 0.03379657864570618, "train_loss_llm": 0.000784456729888916, "grad_norm": 0.025980953127145767, "global_step": 21721, "epoch": 603, "lr": 0.001} +{"train_loss": 0.032465871423482895, "train_loss_bc": 0.032465871423482895, "train_loss_llm": 0.0, "grad_norm": 0.046080153435468674, "global_step": 21722, "epoch": 603, "lr": 0.001} +{"train_loss": 0.028140883892774582, "train_loss_bc": 0.028139585629105568, "train_loss_llm": 0.00012990832328796387, "grad_norm": 0.06339086592197418, "global_step": 21723, "epoch": 603, "lr": 0.001} +{"train_loss": 0.02426554448902607, "train_loss_bc": 0.02425656095147133, "train_loss_llm": 0.0008982717990875244, "grad_norm": 0.07986260950565338, "global_step": 21724, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03431406989693642, "train_loss_bc": 0.034311190247535706, "train_loss_llm": 0.0002880990505218506, "grad_norm": 0.037177860736846924, "global_step": 21725, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03218637779355049, "train_loss_bc": 0.03218637779355049, "train_loss_llm": 0.0, "grad_norm": 0.06055488437414169, "global_step": 21726, "epoch": 603, "lr": 0.001} +{"train_loss": 0.04186040908098221, "train_loss_bc": 0.04186040908098221, "train_loss_llm": 0.0, "grad_norm": 0.08402884006500244, "global_step": 21727, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03590387850999832, "train_loss_bc": 0.03590387850999832, "train_loss_llm": 0.0, "grad_norm": 0.10836627334356308, "global_step": 21728, "epoch": 603, "lr": 0.001} +{"train_loss": 0.02678445540368557, "train_loss_bc": 0.026780206710100174, "train_loss_llm": 0.00042492151260375977, "grad_norm": 0.04212859272956848, "global_step": 21729, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03373777121305466, "train_loss_bc": 0.03372826799750328, "train_loss_llm": 0.0009503066539764404, "grad_norm": 0.08023077994585037, "global_step": 21730, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03810569643974304, "train_loss_bc": 0.03810569643974304, "train_loss_llm": 0.0, "grad_norm": 0.10427023470401764, "global_step": 21731, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03761836886405945, "train_loss_bc": 0.03761836886405945, "train_loss_llm": 0.0, "grad_norm": 0.14759740233421326, "global_step": 21732, "epoch": 603, "lr": 0.001} +{"train_loss": 0.034020766615867615, "train_loss_bc": 0.034020766615867615, "train_loss_llm": 0.0, "grad_norm": 0.03831423446536064, "global_step": 21733, "epoch": 603, "lr": 0.001} +{"train_loss": 0.033177439123392105, "train_loss_bc": 0.03315630182623863, "train_loss_llm": 0.0021137595176696777, "grad_norm": 0.06864666938781738, "global_step": 21734, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03700738027691841, "train_loss_bc": 0.03700738027691841, "train_loss_llm": 0.0, "grad_norm": 0.09271271526813507, "global_step": 21735, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03719059005379677, "train_loss_bc": 0.03719059005379677, "train_loss_llm": 0.0, "grad_norm": 0.12260230630636215, "global_step": 21736, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03285257890820503, "train_loss_bc": 0.03285257890820503, "train_loss_llm": 0.0, "grad_norm": 0.027472862973809242, "global_step": 21737, "epoch": 603, "lr": 0.001} +{"train_loss": 0.028814947232604027, "train_loss_bc": 0.028803464025259018, "train_loss_llm": 0.0011483728885650635, "grad_norm": 0.04027529060840607, "global_step": 21738, "epoch": 603, "lr": 0.001} +{"train_loss": 0.030844012275338173, "train_loss_bc": 0.0308295376598835, "train_loss_llm": 0.0014474987983703613, "grad_norm": 0.05999428778886795, "global_step": 21739, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03517986088991165, "train_loss_bc": 0.03517325222492218, "train_loss_llm": 0.0006608963012695312, "grad_norm": 0.06407172232866287, "global_step": 21740, "epoch": 603, "lr": 0.001} +{"train_loss": 0.029049014672636986, "train_loss_bc": 0.029049014672636986, "train_loss_llm": 0.0, "grad_norm": 0.038088686764240265, "global_step": 21741, "epoch": 603, "lr": 0.001} +{"train_loss": 0.02682831510901451, "train_loss_bc": 0.02682831510901451, "train_loss_llm": 0.0, "grad_norm": 0.05998603254556656, "global_step": 21742, "epoch": 603, "lr": 0.001} +{"train_loss": 0.03294311510398984, "train_loss_bc": 0.03571157902479172, "train_loss_llm": 0.0, "grad_norm": 0.08629988878965378, "global_step": 21743, "epoch": 603, "lr": 0.001, "val_loss": 0.08316604793071747} +{"train_loss": 0.040210407227277756, "train_loss_bc": 0.04018911346793175, "train_loss_llm": 0.002129495143890381, "grad_norm": 0.11108899116516113, "global_step": 21744, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03334828466176987, "train_loss_bc": 0.03334828466176987, "train_loss_llm": 0.0, "grad_norm": 0.0358319915831089, "global_step": 21745, "epoch": 604, "lr": 0.001} +{"train_loss": 0.02778274193406105, "train_loss_bc": 0.02777932956814766, "train_loss_llm": 0.0003412961959838867, "grad_norm": 0.06172576919198036, "global_step": 21746, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03162282705307007, "train_loss_bc": 0.03162282705307007, "train_loss_llm": 0.0, "grad_norm": 0.09351038932800293, "global_step": 21747, "epoch": 604, "lr": 0.001} +{"train_loss": 0.029974715784192085, "train_loss_bc": 0.029974715784192085, "train_loss_llm": 0.0, "grad_norm": 0.10350668430328369, "global_step": 21748, "epoch": 604, "lr": 0.001} +{"train_loss": 0.035419363528490067, "train_loss_bc": 0.035417716950178146, "train_loss_llm": 0.00016483664512634277, "grad_norm": 0.037218526005744934, "global_step": 21749, "epoch": 604, "lr": 0.001} +{"train_loss": 0.031089317053556442, "train_loss_bc": 0.03107317164540291, "train_loss_llm": 0.001614600419998169, "grad_norm": 0.06588561832904816, "global_step": 21750, "epoch": 604, "lr": 0.001} +{"train_loss": 0.038440726697444916, "train_loss_bc": 0.038440726697444916, "train_loss_llm": 0.0, "grad_norm": 0.10973107814788818, "global_step": 21751, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03328865021467209, "train_loss_bc": 0.03328865021467209, "train_loss_llm": 0.0, "grad_norm": 0.13953284919261932, "global_step": 21752, "epoch": 604, "lr": 0.001} +{"train_loss": 0.030654745176434517, "train_loss_bc": 0.030654745176434517, "train_loss_llm": 0.0, "grad_norm": 0.03027361072599888, "global_step": 21753, "epoch": 604, "lr": 0.001} +{"train_loss": 0.029491212218999863, "train_loss_bc": 0.02947303280234337, "train_loss_llm": 0.001817852258682251, "grad_norm": 0.04648423567414284, "global_step": 21754, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03584261238574982, "train_loss_bc": 0.03584261238574982, "train_loss_llm": 0.0, "grad_norm": 0.07695039361715317, "global_step": 21755, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03433666005730629, "train_loss_bc": 0.03433666005730629, "train_loss_llm": 0.0, "grad_norm": 0.09716257452964783, "global_step": 21756, "epoch": 604, "lr": 0.001} +{"train_loss": 0.02741597779095173, "train_loss_bc": 0.02741597779095173, "train_loss_llm": 0.0, "grad_norm": 0.032406967133283615, "global_step": 21757, "epoch": 604, "lr": 0.001} +{"train_loss": 0.0326564684510231, "train_loss_bc": 0.0326564684510231, "train_loss_llm": 0.0, "grad_norm": 0.04996352642774582, "global_step": 21758, "epoch": 604, "lr": 0.001} +{"train_loss": 0.0335107259452343, "train_loss_bc": 0.0335107259452343, "train_loss_llm": 0.0, "grad_norm": 0.0655059888958931, "global_step": 21759, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03597058728337288, "train_loss_bc": 0.03597058728337288, "train_loss_llm": 0.0, "grad_norm": 0.08261386305093765, "global_step": 21760, "epoch": 604, "lr": 0.001} +{"train_loss": 0.030230730772018433, "train_loss_bc": 0.030230730772018433, "train_loss_llm": 0.0, "grad_norm": 0.028179796412587166, "global_step": 21761, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03045475296676159, "train_loss_bc": 0.03044595941901207, "train_loss_llm": 0.0008793771266937256, "grad_norm": 0.044275444000959396, "global_step": 21762, "epoch": 604, "lr": 0.001} +{"train_loss": 0.04154649004340172, "train_loss_bc": 0.04151427745819092, "train_loss_llm": 0.0032214224338531494, "grad_norm": 0.06652067601680756, "global_step": 21763, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03420276939868927, "train_loss_bc": 0.03420276939868927, "train_loss_llm": 0.0, "grad_norm": 0.07621178776025772, "global_step": 21764, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03268880397081375, "train_loss_bc": 0.03268880397081375, "train_loss_llm": 0.0, "grad_norm": 0.03622543811798096, "global_step": 21765, "epoch": 604, "lr": 0.001} +{"train_loss": 0.028887061402201653, "train_loss_bc": 0.028887061402201653, "train_loss_llm": 0.0, "grad_norm": 0.05835805460810661, "global_step": 21766, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03205142170190811, "train_loss_bc": 0.032038364559412, "train_loss_llm": 0.0013056397438049316, "grad_norm": 0.07765959203243256, "global_step": 21767, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03377693146467209, "train_loss_bc": 0.03377693146467209, "train_loss_llm": 0.0, "grad_norm": 0.09597916156053543, "global_step": 21768, "epoch": 604, "lr": 0.001} +{"train_loss": 0.027575399726629257, "train_loss_bc": 0.027565976604819298, "train_loss_llm": 0.0009422898292541504, "grad_norm": 0.026764092966914177, "global_step": 21769, "epoch": 604, "lr": 0.001} +{"train_loss": 0.035124462097883224, "train_loss_bc": 0.035124462097883224, "train_loss_llm": 0.0, "grad_norm": 0.061836253851652145, "global_step": 21770, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03803727403283119, "train_loss_bc": 0.03803645074367523, "train_loss_llm": 8.225440979003906e-05, "grad_norm": 0.09117349982261658, "global_step": 21771, "epoch": 604, "lr": 0.001} +{"train_loss": 0.030731091275811195, "train_loss_bc": 0.030731091275811195, "train_loss_llm": 0.0, "grad_norm": 0.11427470296621323, "global_step": 21772, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03954543173313141, "train_loss_bc": 0.03954543173313141, "train_loss_llm": 0.0, "grad_norm": 0.04195495322346687, "global_step": 21773, "epoch": 604, "lr": 0.001} +{"train_loss": 0.034151673316955566, "train_loss_bc": 0.03414715826511383, "train_loss_llm": 0.0004514753818511963, "grad_norm": 0.06562869995832443, "global_step": 21774, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03334439918398857, "train_loss_bc": 0.03334439918398857, "train_loss_llm": 0.0, "grad_norm": 0.09255219995975494, "global_step": 21775, "epoch": 604, "lr": 0.001} +{"train_loss": 0.031125564128160477, "train_loss_bc": 0.031123671680688858, "train_loss_llm": 0.00018933415412902832, "grad_norm": 0.11929113417863846, "global_step": 21776, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03239484503865242, "train_loss_bc": 0.032389990985393524, "train_loss_llm": 0.00048542022705078125, "grad_norm": 0.03455489128828049, "global_step": 21777, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03107813559472561, "train_loss_bc": 0.03107813559472561, "train_loss_llm": 0.0, "grad_norm": 0.06342105567455292, "global_step": 21778, "epoch": 604, "lr": 0.001} +{"train_loss": 0.03302085704894529, "train_loss_bc": 0.03074580617249012, "train_loss_llm": 0.00017854571342468262, "grad_norm": 0.09337462484836578, "global_step": 21779, "epoch": 604, "lr": 0.001, "val_loss": 0.08271367102861404} +{"train_loss": 0.033496588468551636, "train_loss_bc": 0.033496588468551636, "train_loss_llm": 0.0, "grad_norm": 0.12055902183055878, "global_step": 21780, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03565962240099907, "train_loss_bc": 0.03565523773431778, "train_loss_llm": 0.0004386007785797119, "grad_norm": 0.03374755010008812, "global_step": 21781, "epoch": 605, "lr": 0.001} +{"train_loss": 0.029110928997397423, "train_loss_bc": 0.029107369482517242, "train_loss_llm": 0.0003558695316314697, "grad_norm": 0.05846293643116951, "global_step": 21782, "epoch": 605, "lr": 0.001} +{"train_loss": 0.036641765385866165, "train_loss_bc": 0.036641765385866165, "train_loss_llm": 0.0, "grad_norm": 0.0660100057721138, "global_step": 21783, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03143949434161186, "train_loss_bc": 0.03143949434161186, "train_loss_llm": 0.0, "grad_norm": 0.08869282901287079, "global_step": 21784, "epoch": 605, "lr": 0.001} +{"train_loss": 0.032731883227825165, "train_loss_bc": 0.032731883227825165, "train_loss_llm": 0.0, "grad_norm": 0.028965238481760025, "global_step": 21785, "epoch": 605, "lr": 0.001} +{"train_loss": 0.038827262818813324, "train_loss_bc": 0.038827262818813324, "train_loss_llm": 0.0, "grad_norm": 0.06704287976026535, "global_step": 21786, "epoch": 605, "lr": 0.001} +{"train_loss": 0.035839833319187164, "train_loss_bc": 0.035839833319187164, "train_loss_llm": 0.0, "grad_norm": 0.09753133356571198, "global_step": 21787, "epoch": 605, "lr": 0.001} +{"train_loss": 0.026676781475543976, "train_loss_bc": 0.026676781475543976, "train_loss_llm": 0.0, "grad_norm": 0.11566457152366638, "global_step": 21788, "epoch": 605, "lr": 0.001} +{"train_loss": 0.033414777368307114, "train_loss_bc": 0.033414777368307114, "train_loss_llm": 0.0, "grad_norm": 0.03991404548287392, "global_step": 21789, "epoch": 605, "lr": 0.001} +{"train_loss": 0.0329422727227211, "train_loss_bc": 0.0329422727227211, "train_loss_llm": 0.0, "grad_norm": 0.07644623517990112, "global_step": 21790, "epoch": 605, "lr": 0.001} +{"train_loss": 0.030619345605373383, "train_loss_bc": 0.030608469620347023, "train_loss_llm": 0.0010876059532165527, "grad_norm": 0.11166351288557053, "global_step": 21791, "epoch": 605, "lr": 0.001} +{"train_loss": 0.035053953528404236, "train_loss_bc": 0.035045817494392395, "train_loss_llm": 0.0008135139942169189, "grad_norm": 0.14451994001865387, "global_step": 21792, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03702045977115631, "train_loss_bc": 0.03702045977115631, "train_loss_llm": 0.0, "grad_norm": 0.04626985266804695, "global_step": 21793, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03455350548028946, "train_loss_bc": 0.03455350548028946, "train_loss_llm": 0.0, "grad_norm": 0.0772021934390068, "global_step": 21794, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03541083261370659, "train_loss_bc": 0.03539803996682167, "train_loss_llm": 0.0012792348861694336, "grad_norm": 0.08942107856273651, "global_step": 21795, "epoch": 605, "lr": 0.001} +{"train_loss": 0.029455572366714478, "train_loss_bc": 0.029455572366714478, "train_loss_llm": 0.0, "grad_norm": 0.11399637907743454, "global_step": 21796, "epoch": 605, "lr": 0.001} +{"train_loss": 0.027101635932922363, "train_loss_bc": 0.027101635932922363, "train_loss_llm": 0.0, "grad_norm": 0.03035043179988861, "global_step": 21797, "epoch": 605, "lr": 0.001} +{"train_loss": 0.030651308596134186, "train_loss_bc": 0.030651308596134186, "train_loss_llm": 0.0, "grad_norm": 0.04301765188574791, "global_step": 21798, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03140396624803543, "train_loss_bc": 0.03140396624803543, "train_loss_llm": 0.0, "grad_norm": 0.058626968413591385, "global_step": 21799, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03593309596180916, "train_loss_bc": 0.03593309596180916, "train_loss_llm": 0.0, "grad_norm": 0.07740938663482666, "global_step": 21800, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03365321084856987, "train_loss_bc": 0.03365321084856987, "train_loss_llm": 0.0, "grad_norm": 0.03143986687064171, "global_step": 21801, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03442240506410599, "train_loss_bc": 0.03442240506410599, "train_loss_llm": 0.0, "grad_norm": 0.05763203278183937, "global_step": 21802, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03281367942690849, "train_loss_bc": 0.03281367942690849, "train_loss_llm": 0.0, "grad_norm": 0.08484060317277908, "global_step": 21803, "epoch": 605, "lr": 0.001} +{"train_loss": 0.030358845368027687, "train_loss_bc": 0.030358845368027687, "train_loss_llm": 0.0, "grad_norm": 0.10249588638544083, "global_step": 21804, "epoch": 605, "lr": 0.001} +{"train_loss": 0.029725918546319008, "train_loss_bc": 0.029725080356001854, "train_loss_llm": 8.374452590942383e-05, "grad_norm": 0.04027530550956726, "global_step": 21805, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03890875354409218, "train_loss_bc": 0.03890875354409218, "train_loss_llm": 0.0, "grad_norm": 0.08236552774906158, "global_step": 21806, "epoch": 605, "lr": 0.001} +{"train_loss": 0.0326215997338295, "train_loss_bc": 0.0326215997338295, "train_loss_llm": 0.0, "grad_norm": 0.10641257464885712, "global_step": 21807, "epoch": 605, "lr": 0.001} +{"train_loss": 0.031078044325113297, "train_loss_bc": 0.031078044325113297, "train_loss_llm": 0.0, "grad_norm": 0.12926645576953888, "global_step": 21808, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03254680335521698, "train_loss_bc": 0.03254286199808121, "train_loss_llm": 0.0003942549228668213, "grad_norm": 0.033284299075603485, "global_step": 21809, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03482820466160774, "train_loss_bc": 0.03482820466160774, "train_loss_llm": 0.0, "grad_norm": 0.050278276205062866, "global_step": 21810, "epoch": 605, "lr": 0.001} +{"train_loss": 0.02957608737051487, "train_loss_bc": 0.029569821432232857, "train_loss_llm": 0.0006265044212341309, "grad_norm": 0.06899582594633102, "global_step": 21811, "epoch": 605, "lr": 0.001} +{"train_loss": 0.040278807282447815, "train_loss_bc": 0.040278807282447815, "train_loss_llm": 0.0, "grad_norm": 0.09513765573501587, "global_step": 21812, "epoch": 605, "lr": 0.001} +{"train_loss": 0.02887038327753544, "train_loss_bc": 0.0288661140948534, "train_loss_llm": 0.00042682886123657227, "grad_norm": 0.03773891553282738, "global_step": 21813, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03509794920682907, "train_loss_bc": 0.03508315235376358, "train_loss_llm": 0.0014796257019042969, "grad_norm": 0.06544558703899384, "global_step": 21814, "epoch": 605, "lr": 0.001} +{"train_loss": 0.03323381099229058, "train_loss_bc": 0.03765161707997322, "train_loss_llm": 0.0, "grad_norm": 0.09251338988542557, "global_step": 21815, "epoch": 605, "lr": 0.001, "train/cumulative_reward": 104.31508101009258, "train/mean_score": 6.816282578860883, "train/success_rate": 0.0, "test/cumulative_reward": 43.66583921559094, "test/mean_score": 8.652646965348907, "test/success_rate": 0.18, "val_loss": 0.08226144313812256, "train_action_mse_error": 0.0357334166765213} +{"train_loss": 0.027623683214187622, "train_loss_bc": 0.027623683214187622, "train_loss_llm": 0.0, "grad_norm": 0.11697117239236832, "global_step": 21816, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03809728845953941, "train_loss_bc": 0.03809728845953941, "train_loss_llm": 0.0, "grad_norm": 0.029224378988146782, "global_step": 21817, "epoch": 606, "lr": 0.001} +{"train_loss": 0.037315141409635544, "train_loss_bc": 0.03730551525950432, "train_loss_llm": 0.0009625256061553955, "grad_norm": 0.05723658949136734, "global_step": 21818, "epoch": 606, "lr": 0.001} +{"train_loss": 0.035911910235881805, "train_loss_bc": 0.035911910235881805, "train_loss_llm": 0.0, "grad_norm": 0.07348188757896423, "global_step": 21819, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03197595849633217, "train_loss_bc": 0.03196754679083824, "train_loss_llm": 0.0008410215377807617, "grad_norm": 0.08384889364242554, "global_step": 21820, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03541747108101845, "train_loss_bc": 0.03541747108101845, "train_loss_llm": 0.0, "grad_norm": 0.02929433435201645, "global_step": 21821, "epoch": 606, "lr": 0.001} +{"train_loss": 0.0334332175552845, "train_loss_bc": 0.0334332175552845, "train_loss_llm": 0.0, "grad_norm": 0.054079070687294006, "global_step": 21822, "epoch": 606, "lr": 0.001} +{"train_loss": 0.029886914417147636, "train_loss_bc": 0.029886914417147636, "train_loss_llm": 0.0, "grad_norm": 0.07106319814920425, "global_step": 21823, "epoch": 606, "lr": 0.001} +{"train_loss": 0.032120805233716965, "train_loss_bc": 0.032120805233716965, "train_loss_llm": 0.0, "grad_norm": 0.0839165672659874, "global_step": 21824, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03083568997681141, "train_loss_bc": 0.03083568997681141, "train_loss_llm": 0.0, "grad_norm": 0.030806679278612137, "global_step": 21825, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03734470158815384, "train_loss_bc": 0.037343498319387436, "train_loss_llm": 0.00012028217315673828, "grad_norm": 0.043983977288007736, "global_step": 21826, "epoch": 606, "lr": 0.001} +{"train_loss": 0.028065746650099754, "train_loss_bc": 0.028065746650099754, "train_loss_llm": 0.0, "grad_norm": 0.06102961674332619, "global_step": 21827, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03115878999233246, "train_loss_bc": 0.03115878999233246, "train_loss_llm": 0.0, "grad_norm": 0.07808151841163635, "global_step": 21828, "epoch": 606, "lr": 0.001} +{"train_loss": 0.031989503651857376, "train_loss_bc": 0.031989503651857376, "train_loss_llm": 0.0, "grad_norm": 0.04782147333025932, "global_step": 21829, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03375041484832764, "train_loss_bc": 0.03375041484832764, "train_loss_llm": 0.0, "grad_norm": 0.08374010026454926, "global_step": 21830, "epoch": 606, "lr": 0.001} +{"train_loss": 0.035810135304927826, "train_loss_bc": 0.035810135304927826, "train_loss_llm": 0.0, "grad_norm": 0.11270496994256973, "global_step": 21831, "epoch": 606, "lr": 0.001} +{"train_loss": 0.028600389137864113, "train_loss_bc": 0.028585294261574745, "train_loss_llm": 0.001509547233581543, "grad_norm": 0.1277398020029068, "global_step": 21832, "epoch": 606, "lr": 0.001} +{"train_loss": 0.030675994232296944, "train_loss_bc": 0.030675994232296944, "train_loss_llm": 0.0, "grad_norm": 0.02893074043095112, "global_step": 21833, "epoch": 606, "lr": 0.001} +{"train_loss": 0.037255655974149704, "train_loss_bc": 0.037255655974149704, "train_loss_llm": 0.0, "grad_norm": 0.05078674852848053, "global_step": 21834, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03777705878019333, "train_loss_bc": 0.03777705878019333, "train_loss_llm": 0.0, "grad_norm": 0.06601467728614807, "global_step": 21835, "epoch": 606, "lr": 0.001} +{"train_loss": 0.02366124838590622, "train_loss_bc": 0.02365867793560028, "train_loss_llm": 0.0002569854259490967, "grad_norm": 0.07769297063350677, "global_step": 21836, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03543788194656372, "train_loss_bc": 0.03541271388530731, "train_loss_llm": 0.0025169849395751953, "grad_norm": 0.02865849994122982, "global_step": 21837, "epoch": 606, "lr": 0.001} +{"train_loss": 0.033368319272994995, "train_loss_bc": 0.033368319272994995, "train_loss_llm": 0.0, "grad_norm": 0.046647921204566956, "global_step": 21838, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03489267826080322, "train_loss_bc": 0.03489267826080322, "train_loss_llm": 0.0, "grad_norm": 0.06503982096910477, "global_step": 21839, "epoch": 606, "lr": 0.001} +{"train_loss": 0.032254502177238464, "train_loss_bc": 0.032254502177238464, "train_loss_llm": 0.0, "grad_norm": 0.07202441245317459, "global_step": 21840, "epoch": 606, "lr": 0.001} +{"train_loss": 0.028777627274394035, "train_loss_bc": 0.02877500094473362, "train_loss_llm": 0.0002627074718475342, "grad_norm": 0.028418397530913353, "global_step": 21841, "epoch": 606, "lr": 0.001} +{"train_loss": 0.026259586215019226, "train_loss_bc": 0.026259586215019226, "train_loss_llm": 0.0, "grad_norm": 0.033891599625349045, "global_step": 21842, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03644002228975296, "train_loss_bc": 0.03644002228975296, "train_loss_llm": 0.0, "grad_norm": 0.05590221658349037, "global_step": 21843, "epoch": 606, "lr": 0.001} +{"train_loss": 0.034196943044662476, "train_loss_bc": 0.034196943044662476, "train_loss_llm": 0.0, "grad_norm": 0.07331054657697678, "global_step": 21844, "epoch": 606, "lr": 0.001} +{"train_loss": 0.032430000603199005, "train_loss_bc": 0.03241083770990372, "train_loss_llm": 0.0019162893295288086, "grad_norm": 0.0292847640812397, "global_step": 21845, "epoch": 606, "lr": 0.001} +{"train_loss": 0.026200244203209877, "train_loss_bc": 0.02619338221848011, "train_loss_llm": 0.0006862282752990723, "grad_norm": 0.04654529318213463, "global_step": 21846, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03195810317993164, "train_loss_bc": 0.03191151097416878, "train_loss_llm": 0.004659324884414673, "grad_norm": 0.06877178698778152, "global_step": 21847, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03859610855579376, "train_loss_bc": 0.03859610855579376, "train_loss_llm": 0.0, "grad_norm": 0.09385393559932709, "global_step": 21848, "epoch": 606, "lr": 0.001} +{"train_loss": 0.03226889297366142, "train_loss_bc": 0.03226889297366142, "train_loss_llm": 0.0, "grad_norm": 0.029512805864214897, "global_step": 21849, "epoch": 606, "lr": 0.001} +{"train_loss": 0.035914890468120575, "train_loss_bc": 0.03590403497219086, "train_loss_llm": 0.001085430383682251, "grad_norm": 0.04768016189336777, "global_step": 21850, "epoch": 606, "lr": 0.001} +{"train_loss": 0.032669668603274554, "train_loss_bc": 0.028378095477819443, "train_loss_llm": 0.002645432949066162, "grad_norm": 0.06779276579618454, "global_step": 21851, "epoch": 606, "lr": 0.001, "val_loss": 0.08287438750267029} +{"train_loss": 0.03268951177597046, "train_loss_bc": 0.03268749266862869, "train_loss_llm": 0.00020173192024230957, "grad_norm": 0.09203372895717621, "global_step": 21852, "epoch": 607, "lr": 0.001} +{"train_loss": 0.029145270586013794, "train_loss_bc": 0.029145270586013794, "train_loss_llm": 0.0, "grad_norm": 0.042859289795160294, "global_step": 21853, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03522821143269539, "train_loss_bc": 0.03522821143269539, "train_loss_llm": 0.0, "grad_norm": 0.0822598785161972, "global_step": 21854, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03491148352622986, "train_loss_bc": 0.03491148352622986, "train_loss_llm": 0.0, "grad_norm": 0.09902073442935944, "global_step": 21855, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03187268599867821, "train_loss_bc": 0.03187268599867821, "train_loss_llm": 0.0, "grad_norm": 0.11876672506332397, "global_step": 21856, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03271789476275444, "train_loss_bc": 0.032710157334804535, "train_loss_llm": 0.0007737874984741211, "grad_norm": 0.030979041010141373, "global_step": 21857, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03567756339907646, "train_loss_bc": 0.03567756339907646, "train_loss_llm": 0.0, "grad_norm": 0.04965684190392494, "global_step": 21858, "epoch": 607, "lr": 0.001} +{"train_loss": 0.030689844861626625, "train_loss_bc": 0.030689844861626625, "train_loss_llm": 0.0, "grad_norm": 0.0647585541009903, "global_step": 21859, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03428218141198158, "train_loss_bc": 0.03428218141198158, "train_loss_llm": 0.0, "grad_norm": 0.0815039798617363, "global_step": 21860, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03225623071193695, "train_loss_bc": 0.03225623071193695, "train_loss_llm": 0.0, "grad_norm": 0.0335119366645813, "global_step": 21861, "epoch": 607, "lr": 0.001} +{"train_loss": 0.02969515137374401, "train_loss_bc": 0.02969515137374401, "train_loss_llm": 0.0, "grad_norm": 0.04112931340932846, "global_step": 21862, "epoch": 607, "lr": 0.001} +{"train_loss": 0.02759489417076111, "train_loss_bc": 0.02759489417076111, "train_loss_llm": 0.0, "grad_norm": 0.05087117478251457, "global_step": 21863, "epoch": 607, "lr": 0.001} +{"train_loss": 0.02955797128379345, "train_loss_bc": 0.02955797128379345, "train_loss_llm": 0.0, "grad_norm": 0.058562275022268295, "global_step": 21864, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03213276341557503, "train_loss_bc": 0.03213276341557503, "train_loss_llm": 0.0, "grad_norm": 0.03578786924481392, "global_step": 21865, "epoch": 607, "lr": 0.001} +{"train_loss": 0.02882777340710163, "train_loss_bc": 0.028817998245358467, "train_loss_llm": 0.0009774267673492432, "grad_norm": 0.06677217781543732, "global_step": 21866, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03829607367515564, "train_loss_bc": 0.03827012702822685, "train_loss_llm": 0.0025947093963623047, "grad_norm": 0.09893608838319778, "global_step": 21867, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03155377507209778, "train_loss_bc": 0.031547170132398605, "train_loss_llm": 0.0006605088710784912, "grad_norm": 0.11998534947633743, "global_step": 21868, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03696690872311592, "train_loss_bc": 0.03696690872311592, "train_loss_llm": 0.0, "grad_norm": 0.028668014332652092, "global_step": 21869, "epoch": 607, "lr": 0.001} +{"train_loss": 0.029620962217450142, "train_loss_bc": 0.029620962217450142, "train_loss_llm": 0.0, "grad_norm": 0.03855714574456215, "global_step": 21870, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03105570189654827, "train_loss_bc": 0.03105570189654827, "train_loss_llm": 0.0, "grad_norm": 0.05941079929471016, "global_step": 21871, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03216847777366638, "train_loss_bc": 0.03216847777366638, "train_loss_llm": 0.0, "grad_norm": 0.07556556910276413, "global_step": 21872, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03917239233851433, "train_loss_bc": 0.03917239233851433, "train_loss_llm": 0.0, "grad_norm": 0.03979918360710144, "global_step": 21873, "epoch": 607, "lr": 0.001} +{"train_loss": 0.02798212505877018, "train_loss_bc": 0.02796199917793274, "train_loss_llm": 0.0020126402378082275, "grad_norm": 0.050221480429172516, "global_step": 21874, "epoch": 607, "lr": 0.001} +{"train_loss": 0.028768640011548996, "train_loss_bc": 0.028768640011548996, "train_loss_llm": 0.0, "grad_norm": 0.067502461373806, "global_step": 21875, "epoch": 607, "lr": 0.001} +{"train_loss": 0.0383513942360878, "train_loss_bc": 0.0383513942360878, "train_loss_llm": 0.0, "grad_norm": 0.09773438423871994, "global_step": 21876, "epoch": 607, "lr": 0.001} +{"train_loss": 0.031030453741550446, "train_loss_bc": 0.031030453741550446, "train_loss_llm": 0.0, "grad_norm": 0.027230296283960342, "global_step": 21877, "epoch": 607, "lr": 0.001} +{"train_loss": 0.027279527857899666, "train_loss_bc": 0.027270948514342308, "train_loss_llm": 0.000857919454574585, "grad_norm": 0.04453495144844055, "global_step": 21878, "epoch": 607, "lr": 0.001} +{"train_loss": 0.033652450889348984, "train_loss_bc": 0.033652450889348984, "train_loss_llm": 0.0, "grad_norm": 0.061935536563396454, "global_step": 21879, "epoch": 607, "lr": 0.001} +{"train_loss": 0.028534356504678726, "train_loss_bc": 0.028531687334179878, "train_loss_llm": 0.0002669095993041992, "grad_norm": 0.07248334586620331, "global_step": 21880, "epoch": 607, "lr": 0.001} +{"train_loss": 0.033964160829782486, "train_loss_bc": 0.03394947201013565, "train_loss_llm": 0.001468747854232788, "grad_norm": 0.029716117307543755, "global_step": 21881, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03942029923200607, "train_loss_bc": 0.03942029923200607, "train_loss_llm": 0.0, "grad_norm": 0.06089259311556816, "global_step": 21882, "epoch": 607, "lr": 0.001} +{"train_loss": 0.04015678912401199, "train_loss_bc": 0.04015678912401199, "train_loss_llm": 0.0, "grad_norm": 0.08390171080827713, "global_step": 21883, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03913763538002968, "train_loss_bc": 0.03913763538002968, "train_loss_llm": 0.0, "grad_norm": 0.09728100895881653, "global_step": 21884, "epoch": 607, "lr": 0.001} +{"train_loss": 0.030935967341065407, "train_loss_bc": 0.030935967341065407, "train_loss_llm": 0.0, "grad_norm": 0.02817811444401741, "global_step": 21885, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03195538371801376, "train_loss_bc": 0.03195538371801376, "train_loss_llm": 0.0, "grad_norm": 0.05269235372543335, "global_step": 21886, "epoch": 607, "lr": 0.001} +{"train_loss": 0.03266039925316969, "train_loss_bc": 0.028491465374827385, "train_loss_llm": 0.0, "grad_norm": 0.07757027447223663, "global_step": 21887, "epoch": 607, "lr": 0.001, "val_loss": 0.08440953493118286} +{"train_loss": 0.038511015474796295, "train_loss_bc": 0.038511015474796295, "train_loss_llm": 0.0, "grad_norm": 0.09377903491258621, "global_step": 21888, "epoch": 608, "lr": 0.001} +{"train_loss": 0.034921132028102875, "train_loss_bc": 0.034921132028102875, "train_loss_llm": 0.0, "grad_norm": 0.03139825165271759, "global_step": 21889, "epoch": 608, "lr": 0.001} +{"train_loss": 0.031087907031178474, "train_loss_bc": 0.031087907031178474, "train_loss_llm": 0.0, "grad_norm": 0.05047078803181648, "global_step": 21890, "epoch": 608, "lr": 0.001} +{"train_loss": 0.034920837730169296, "train_loss_bc": 0.034920837730169296, "train_loss_llm": 0.0, "grad_norm": 0.07495994120836258, "global_step": 21891, "epoch": 608, "lr": 0.001} +{"train_loss": 0.027458610013127327, "train_loss_bc": 0.027453819289803505, "train_loss_llm": 0.00047907233238220215, "grad_norm": 0.0864064022898674, "global_step": 21892, "epoch": 608, "lr": 0.001} +{"train_loss": 0.033746976405382156, "train_loss_bc": 0.033746976405382156, "train_loss_llm": 0.0, "grad_norm": 0.03428454324603081, "global_step": 21893, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03130977973341942, "train_loss_bc": 0.03130902722477913, "train_loss_llm": 7.528066635131836e-05, "grad_norm": 0.057395171374082565, "global_step": 21894, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03568831458687782, "train_loss_bc": 0.035683177411556244, "train_loss_llm": 0.0005136728286743164, "grad_norm": 0.08897437155246735, "global_step": 21895, "epoch": 608, "lr": 0.001} +{"train_loss": 0.035670749843120575, "train_loss_bc": 0.035670749843120575, "train_loss_llm": 0.0, "grad_norm": 0.10109329968690872, "global_step": 21896, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03690053150057793, "train_loss_bc": 0.03689364716410637, "train_loss_llm": 0.0006882548332214355, "grad_norm": 0.03155058994889259, "global_step": 21897, "epoch": 608, "lr": 0.001} +{"train_loss": 0.028295759111642838, "train_loss_bc": 0.028295759111642838, "train_loss_llm": 0.0, "grad_norm": 0.046934932470321655, "global_step": 21898, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03527147322893143, "train_loss_bc": 0.03527147322893143, "train_loss_llm": 0.0, "grad_norm": 0.0629243329167366, "global_step": 21899, "epoch": 608, "lr": 0.001} +{"train_loss": 0.028362859040498734, "train_loss_bc": 0.028357578441500664, "train_loss_llm": 0.0005280375480651855, "grad_norm": 0.07843699306249619, "global_step": 21900, "epoch": 608, "lr": 0.001} +{"train_loss": 0.032544322311878204, "train_loss_bc": 0.032544322311878204, "train_loss_llm": 0.0, "grad_norm": 0.029640818014740944, "global_step": 21901, "epoch": 608, "lr": 0.001} +{"train_loss": 0.029814092442393303, "train_loss_bc": 0.029814092442393303, "train_loss_llm": 0.0, "grad_norm": 0.052248988300561905, "global_step": 21902, "epoch": 608, "lr": 0.001} +{"train_loss": 0.028194716200232506, "train_loss_bc": 0.028194716200232506, "train_loss_llm": 0.0, "grad_norm": 0.07298292964696884, "global_step": 21903, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03302909806370735, "train_loss_bc": 0.03302909806370735, "train_loss_llm": 0.0, "grad_norm": 0.0989164412021637, "global_step": 21904, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03220946341753006, "train_loss_bc": 0.03220946341753006, "train_loss_llm": 0.0, "grad_norm": 0.03817909583449364, "global_step": 21905, "epoch": 608, "lr": 0.001} +{"train_loss": 0.02781941369175911, "train_loss_bc": 0.027809591963887215, "train_loss_llm": 0.0009821951389312744, "grad_norm": 0.06836989521980286, "global_step": 21906, "epoch": 608, "lr": 0.001} +{"train_loss": 0.031262196600437164, "train_loss_bc": 0.031262196600437164, "train_loss_llm": 0.0, "grad_norm": 0.08713344484567642, "global_step": 21907, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03957000747323036, "train_loss_bc": 0.03957000747323036, "train_loss_llm": 0.0, "grad_norm": 0.10652274638414383, "global_step": 21908, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03767932951450348, "train_loss_bc": 0.03767932951450348, "train_loss_llm": 0.0, "grad_norm": 0.05031372606754303, "global_step": 21909, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03447464480996132, "train_loss_bc": 0.03447464480996132, "train_loss_llm": 0.0, "grad_norm": 0.08959755301475525, "global_step": 21910, "epoch": 608, "lr": 0.001} +{"train_loss": 0.028374848887324333, "train_loss_bc": 0.028374848887324333, "train_loss_llm": 0.0, "grad_norm": 0.10575909167528152, "global_step": 21911, "epoch": 608, "lr": 0.001} +{"train_loss": 0.026539459824562073, "train_loss_bc": 0.02651936374604702, "train_loss_llm": 0.0020096898078918457, "grad_norm": 0.12343238294124603, "global_step": 21912, "epoch": 608, "lr": 0.001} +{"train_loss": 0.030144276097416878, "train_loss_bc": 0.030144276097416878, "train_loss_llm": 0.0, "grad_norm": 0.03327970579266548, "global_step": 21913, "epoch": 608, "lr": 0.001} +{"train_loss": 0.029526900500059128, "train_loss_bc": 0.029526900500059128, "train_loss_llm": 0.0, "grad_norm": 0.05356863886117935, "global_step": 21914, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03875282034277916, "train_loss_bc": 0.03875282034277916, "train_loss_llm": 0.0, "grad_norm": 0.06661391258239746, "global_step": 21915, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03560304641723633, "train_loss_bc": 0.03560304641723633, "train_loss_llm": 0.0, "grad_norm": 0.09779155999422073, "global_step": 21916, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03173273056745529, "train_loss_bc": 0.03171852231025696, "train_loss_llm": 0.0014207065105438232, "grad_norm": 0.026638302952051163, "global_step": 21917, "epoch": 608, "lr": 0.001} +{"train_loss": 0.02916492335498333, "train_loss_bc": 0.02916492335498333, "train_loss_llm": 0.0, "grad_norm": 0.04221880063414574, "global_step": 21918, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03363331779837608, "train_loss_bc": 0.03363331779837608, "train_loss_llm": 0.0, "grad_norm": 0.05453009530901909, "global_step": 21919, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03326081857085228, "train_loss_bc": 0.03326081857085228, "train_loss_llm": 0.0, "grad_norm": 0.07005015760660172, "global_step": 21920, "epoch": 608, "lr": 0.001} +{"train_loss": 0.036871179938316345, "train_loss_bc": 0.036871179938316345, "train_loss_llm": 0.0, "grad_norm": 0.03367201238870621, "global_step": 21921, "epoch": 608, "lr": 0.001} +{"train_loss": 0.035929787904024124, "train_loss_bc": 0.035924263298511505, "train_loss_llm": 0.0005523562431335449, "grad_norm": 0.058174487203359604, "global_step": 21922, "epoch": 608, "lr": 0.001} +{"train_loss": 0.03266177191916439, "train_loss_bc": 0.02754644863307476, "train_loss_llm": 0.0, "grad_norm": 0.0744062140583992, "global_step": 21923, "epoch": 608, "lr": 0.001, "val_loss": 0.08350981771945953} +{"train_loss": 0.035422347486019135, "train_loss_bc": 0.035393476486206055, "train_loss_llm": 0.002886950969696045, "grad_norm": 0.09230571240186691, "global_step": 21924, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03645305708050728, "train_loss_bc": 0.03645305708050728, "train_loss_llm": 0.0, "grad_norm": 0.039051465690135956, "global_step": 21925, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03804628551006317, "train_loss_bc": 0.03804628551006317, "train_loss_llm": 0.0, "grad_norm": 0.08171077072620392, "global_step": 21926, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03720161318778992, "train_loss_bc": 0.03720161318778992, "train_loss_llm": 0.0, "grad_norm": 0.11157043278217316, "global_step": 21927, "epoch": 609, "lr": 0.001} +{"train_loss": 0.026508722454309464, "train_loss_bc": 0.02650299482047558, "train_loss_llm": 0.0005728006362915039, "grad_norm": 0.13389797508716583, "global_step": 21928, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03730163350701332, "train_loss_bc": 0.03730163350701332, "train_loss_llm": 0.0, "grad_norm": 0.04304748401045799, "global_step": 21929, "epoch": 609, "lr": 0.001} +{"train_loss": 0.029059264808893204, "train_loss_bc": 0.029059264808893204, "train_loss_llm": 0.0, "grad_norm": 0.06843696534633636, "global_step": 21930, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03301187604665756, "train_loss_bc": 0.03301187604665756, "train_loss_llm": 0.0, "grad_norm": 0.09285546094179153, "global_step": 21931, "epoch": 609, "lr": 0.001} +{"train_loss": 0.02995881997048855, "train_loss_bc": 0.02995881997048855, "train_loss_llm": 0.0, "grad_norm": 0.1282622516155243, "global_step": 21932, "epoch": 609, "lr": 0.001} +{"train_loss": 0.030019376426935196, "train_loss_bc": 0.030019376426935196, "train_loss_llm": 0.0, "grad_norm": 0.034476783126592636, "global_step": 21933, "epoch": 609, "lr": 0.001} +{"train_loss": 0.027470098808407784, "train_loss_bc": 0.027470098808407784, "train_loss_llm": 0.0, "grad_norm": 0.05571632832288742, "global_step": 21934, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03454342111945152, "train_loss_bc": 0.03454342111945152, "train_loss_llm": 0.0, "grad_norm": 0.0856136828660965, "global_step": 21935, "epoch": 609, "lr": 0.001} +{"train_loss": 0.028307510539889336, "train_loss_bc": 0.028307510539889336, "train_loss_llm": 0.0, "grad_norm": 0.10954256355762482, "global_step": 21936, "epoch": 609, "lr": 0.001} +{"train_loss": 0.032753244042396545, "train_loss_bc": 0.032753244042396545, "train_loss_llm": 0.0, "grad_norm": 0.032982148230075836, "global_step": 21937, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03207485377788544, "train_loss_bc": 0.03207485377788544, "train_loss_llm": 0.0, "grad_norm": 0.053270142525434494, "global_step": 21938, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03407899662852287, "train_loss_bc": 0.03407899662852287, "train_loss_llm": 0.0, "grad_norm": 0.05449400842189789, "global_step": 21939, "epoch": 609, "lr": 0.001} +{"train_loss": 0.027287807315587997, "train_loss_bc": 0.027287807315587997, "train_loss_llm": 0.0, "grad_norm": 0.06238114461302757, "global_step": 21940, "epoch": 609, "lr": 0.001} +{"train_loss": 0.036318305879831314, "train_loss_bc": 0.036318305879831314, "train_loss_llm": 0.0, "grad_norm": 0.034676533192396164, "global_step": 21941, "epoch": 609, "lr": 0.001} +{"train_loss": 0.036105927079916, "train_loss_bc": 0.036105927079916, "train_loss_llm": 0.0, "grad_norm": 0.06233791634440422, "global_step": 21942, "epoch": 609, "lr": 0.001} +{"train_loss": 0.02623145654797554, "train_loss_bc": 0.02623145654797554, "train_loss_llm": 0.0, "grad_norm": 0.0903831347823143, "global_step": 21943, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03483901917934418, "train_loss_bc": 0.03483901917934418, "train_loss_llm": 0.0, "grad_norm": 0.10749034583568573, "global_step": 21944, "epoch": 609, "lr": 0.001} +{"train_loss": 0.036581944674253464, "train_loss_bc": 0.036581944674253464, "train_loss_llm": 0.0, "grad_norm": 0.0391993410885334, "global_step": 21945, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03054746240377426, "train_loss_bc": 0.03054746240377426, "train_loss_llm": 0.0, "grad_norm": 0.06691230833530426, "global_step": 21946, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03351788967847824, "train_loss_bc": 0.03351788967847824, "train_loss_llm": 0.0, "grad_norm": 0.10098350793123245, "global_step": 21947, "epoch": 609, "lr": 0.001} +{"train_loss": 0.030256183817982674, "train_loss_bc": 0.03024372085928917, "train_loss_llm": 0.0012463629245758057, "grad_norm": 0.13219349086284637, "global_step": 21948, "epoch": 609, "lr": 0.001} +{"train_loss": 0.036389898508787155, "train_loss_bc": 0.036389898508787155, "train_loss_llm": 0.0, "grad_norm": 0.031905245035886765, "global_step": 21949, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03758065029978752, "train_loss_bc": 0.03756486251950264, "train_loss_llm": 0.001578688621520996, "grad_norm": 0.0658801794052124, "global_step": 21950, "epoch": 609, "lr": 0.001} +{"train_loss": 0.031585246324539185, "train_loss_bc": 0.03156711533665657, "train_loss_llm": 0.0018131136894226074, "grad_norm": 0.08869797736406326, "global_step": 21951, "epoch": 609, "lr": 0.001} +{"train_loss": 0.0325799360871315, "train_loss_bc": 0.0325799360871315, "train_loss_llm": 0.0, "grad_norm": 0.10391658544540405, "global_step": 21952, "epoch": 609, "lr": 0.001} +{"train_loss": 0.034275345504283905, "train_loss_bc": 0.03424393758177757, "train_loss_llm": 0.0031409263610839844, "grad_norm": 0.0366014800965786, "global_step": 21953, "epoch": 609, "lr": 0.001} +{"train_loss": 0.029223576188087463, "train_loss_bc": 0.029223576188087463, "train_loss_llm": 0.0, "grad_norm": 0.05001272261142731, "global_step": 21954, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03533061593770981, "train_loss_bc": 0.03533061593770981, "train_loss_llm": 0.0, "grad_norm": 0.06737615913152695, "global_step": 21955, "epoch": 609, "lr": 0.001} +{"train_loss": 0.040067318826913834, "train_loss_bc": 0.040067318826913834, "train_loss_llm": 0.0, "grad_norm": 0.08703381568193436, "global_step": 21956, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03208865225315094, "train_loss_bc": 0.03208865225315094, "train_loss_llm": 0.0, "grad_norm": 0.02934243530035019, "global_step": 21957, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03446299955248833, "train_loss_bc": 0.03445848077535629, "train_loss_llm": 0.0004519820213317871, "grad_norm": 0.04677262157201767, "global_step": 21958, "epoch": 609, "lr": 0.001} +{"train_loss": 0.03295022611402803, "train_loss_bc": 0.028726782649755478, "train_loss_llm": 0.0, "grad_norm": 0.06450992077589035, "global_step": 21959, "epoch": 609, "lr": 0.001, "val_loss": 0.08310368657112122} +{"train_loss": 0.03668144717812538, "train_loss_bc": 0.03668144717812538, "train_loss_llm": 0.0, "grad_norm": 0.08468494564294815, "global_step": 21960, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03571125864982605, "train_loss_bc": 0.035689521580934525, "train_loss_llm": 0.002173870801925659, "grad_norm": 0.029598990455269814, "global_step": 21961, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03170253708958626, "train_loss_bc": 0.03169791027903557, "train_loss_llm": 0.00046265125274658203, "grad_norm": 0.03969471529126167, "global_step": 21962, "epoch": 610, "lr": 0.001} +{"train_loss": 0.032189056277275085, "train_loss_bc": 0.032189056277275085, "train_loss_llm": 0.0, "grad_norm": 0.05720558762550354, "global_step": 21963, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03348923847079277, "train_loss_bc": 0.03348923847079277, "train_loss_llm": 0.0, "grad_norm": 0.07210306823253632, "global_step": 21964, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03471023589372635, "train_loss_bc": 0.03471023589372635, "train_loss_llm": 0.0, "grad_norm": 0.0244976207613945, "global_step": 21965, "epoch": 610, "lr": 0.001} +{"train_loss": 0.02484070509672165, "train_loss_bc": 0.024821683764457703, "train_loss_llm": 0.001902073621749878, "grad_norm": 0.03953752666711807, "global_step": 21966, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03442324697971344, "train_loss_bc": 0.03442225605249405, "train_loss_llm": 9.909272193908691e-05, "grad_norm": 0.06005365028977394, "global_step": 21967, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03569726273417473, "train_loss_bc": 0.03569726273417473, "train_loss_llm": 0.0, "grad_norm": 0.07459088414907455, "global_step": 21968, "epoch": 610, "lr": 0.001} +{"train_loss": 0.031220361590385437, "train_loss_bc": 0.031220361590385437, "train_loss_llm": 0.0, "grad_norm": 0.03740039840340614, "global_step": 21969, "epoch": 610, "lr": 0.001} +{"train_loss": 0.034441929310560226, "train_loss_bc": 0.03442549705505371, "train_loss_llm": 0.0016431808471679688, "grad_norm": 0.059534139931201935, "global_step": 21970, "epoch": 610, "lr": 0.001} +{"train_loss": 0.0321650356054306, "train_loss_bc": 0.0321650356054306, "train_loss_llm": 0.0, "grad_norm": 0.08281561732292175, "global_step": 21971, "epoch": 610, "lr": 0.001} +{"train_loss": 0.0332600362598896, "train_loss_bc": 0.03325832635164261, "train_loss_llm": 0.000170975923538208, "grad_norm": 0.10499302297830582, "global_step": 21972, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03368585929274559, "train_loss_bc": 0.03368585929274559, "train_loss_llm": 0.0, "grad_norm": 0.030913623049855232, "global_step": 21973, "epoch": 610, "lr": 0.001} +{"train_loss": 0.029684409499168396, "train_loss_bc": 0.029684409499168396, "train_loss_llm": 0.0, "grad_norm": 0.05243192985653877, "global_step": 21974, "epoch": 610, "lr": 0.001} +{"train_loss": 0.027804404497146606, "train_loss_bc": 0.027804404497146606, "train_loss_llm": 0.0, "grad_norm": 0.07170969247817993, "global_step": 21975, "epoch": 610, "lr": 0.001} +{"train_loss": 0.0316535122692585, "train_loss_bc": 0.0316535122692585, "train_loss_llm": 0.0, "grad_norm": 0.09442318230867386, "global_step": 21976, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03622199222445488, "train_loss_bc": 0.03622199222445488, "train_loss_llm": 0.0, "grad_norm": 0.03208385035395622, "global_step": 21977, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03568003699183464, "train_loss_bc": 0.03562676161527634, "train_loss_llm": 0.00532764196395874, "grad_norm": 0.06047842279076576, "global_step": 21978, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03814845159649849, "train_loss_bc": 0.038138248026371, "train_loss_llm": 0.0010203719139099121, "grad_norm": 0.08065232634544373, "global_step": 21979, "epoch": 610, "lr": 0.001} +{"train_loss": 0.037912286818027496, "train_loss_bc": 0.037912286818027496, "train_loss_llm": 0.0, "grad_norm": 0.09758686274290085, "global_step": 21980, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03297818452119827, "train_loss_bc": 0.03297818452119827, "train_loss_llm": 0.0, "grad_norm": 0.02799152582883835, "global_step": 21981, "epoch": 610, "lr": 0.001} +{"train_loss": 0.029182590544223785, "train_loss_bc": 0.029182590544223785, "train_loss_llm": 0.0, "grad_norm": 0.044917069375514984, "global_step": 21982, "epoch": 610, "lr": 0.001} +{"train_loss": 0.031043952330946922, "train_loss_bc": 0.03104039654135704, "train_loss_llm": 0.0003555715084075928, "grad_norm": 0.05785176903009415, "global_step": 21983, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03204412758350372, "train_loss_bc": 0.03204412758350372, "train_loss_llm": 0.0, "grad_norm": 0.07739603519439697, "global_step": 21984, "epoch": 610, "lr": 0.001} +{"train_loss": 0.038541581481695175, "train_loss_bc": 0.038541581481695175, "train_loss_llm": 0.0, "grad_norm": 0.040800921618938446, "global_step": 21985, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03260308504104614, "train_loss_bc": 0.03260093182325363, "train_loss_llm": 0.00021523237228393555, "grad_norm": 0.06036345660686493, "global_step": 21986, "epoch": 610, "lr": 0.001} +{"train_loss": 0.02665190026164055, "train_loss_bc": 0.02665190026164055, "train_loss_llm": 0.0, "grad_norm": 0.08461527526378632, "global_step": 21987, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03258603438735008, "train_loss_bc": 0.03258603438735008, "train_loss_llm": 0.0, "grad_norm": 0.12046091258525848, "global_step": 21988, "epoch": 610, "lr": 0.001} +{"train_loss": 0.031378064304590225, "train_loss_bc": 0.031378064304590225, "train_loss_llm": 0.0, "grad_norm": 0.03155510127544403, "global_step": 21989, "epoch": 610, "lr": 0.001} +{"train_loss": 0.02907249517738819, "train_loss_bc": 0.02907249517738819, "train_loss_llm": 0.0, "grad_norm": 0.04801906645298004, "global_step": 21990, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03058314509689808, "train_loss_bc": 0.03058314509689808, "train_loss_llm": 0.0, "grad_norm": 0.06728847324848175, "global_step": 21991, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03562910482287407, "train_loss_bc": 0.03562910482287407, "train_loss_llm": 0.0, "grad_norm": 0.08567682653665543, "global_step": 21992, "epoch": 610, "lr": 0.001} +{"train_loss": 0.028027748689055443, "train_loss_bc": 0.028027748689055443, "train_loss_llm": 0.0, "grad_norm": 0.02561045065522194, "global_step": 21993, "epoch": 610, "lr": 0.001} +{"train_loss": 0.030877452343702316, "train_loss_bc": 0.030877452343702316, "train_loss_llm": 0.0, "grad_norm": 0.05155950039625168, "global_step": 21994, "epoch": 610, "lr": 0.001} +{"train_loss": 0.03274290284348859, "train_loss_bc": 0.03621755540370941, "train_loss_llm": 0.0004177689552307129, "grad_norm": 0.06722468882799149, "global_step": 21995, "epoch": 610, "lr": 0.001, "train/cumulative_reward": 76.61055980377395, "train/mean_score": 23.844133055235982, "train/success_rate": 0.7, "test/cumulative_reward": 50.682020962029554, "test/mean_score": 7.869509585238584, "test/success_rate": 0.18, "val_loss": 0.08174562454223633, "train_action_mse_error": 0.034995947033166885} +{"train_loss": 0.02938813716173172, "train_loss_bc": 0.02938813716173172, "train_loss_llm": 0.0, "grad_norm": 0.09091275930404663, "global_step": 21996, "epoch": 611, "lr": 0.001} +{"train_loss": 0.030116209760308266, "train_loss_bc": 0.030116209760308266, "train_loss_llm": 0.0, "grad_norm": 0.02711268700659275, "global_step": 21997, "epoch": 611, "lr": 0.001} +{"train_loss": 0.031942255795001984, "train_loss_bc": 0.031936515122652054, "train_loss_llm": 0.0005739927291870117, "grad_norm": 0.04229367896914482, "global_step": 21998, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03364604711532593, "train_loss_bc": 0.033643219619989395, "train_loss_llm": 0.00028261542320251465, "grad_norm": 0.05799275264143944, "global_step": 21999, "epoch": 611, "lr": 0.001} +{"train_loss": 0.030595041811466217, "train_loss_bc": 0.030595041811466217, "train_loss_llm": 0.0, "grad_norm": 0.07136622816324234, "global_step": 22000, "epoch": 611, "lr": 0.001} +{"train_loss": 0.026702942326664925, "train_loss_bc": 0.026702942326664925, "train_loss_llm": 0.0, "grad_norm": 0.027854209765791893, "global_step": 22001, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03295201435685158, "train_loss_bc": 0.032951436936855316, "train_loss_llm": 5.790591239929199e-05, "grad_norm": 0.047675423324108124, "global_step": 22002, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03215840458869934, "train_loss_bc": 0.03215840458869934, "train_loss_llm": 0.0, "grad_norm": 0.06172254681587219, "global_step": 22003, "epoch": 611, "lr": 0.001} +{"train_loss": 0.030983079224824905, "train_loss_bc": 0.030983079224824905, "train_loss_llm": 0.0, "grad_norm": 0.07404343783855438, "global_step": 22004, "epoch": 611, "lr": 0.001} +{"train_loss": 0.02900213561952114, "train_loss_bc": 0.02900213561952114, "train_loss_llm": 0.0, "grad_norm": 0.031474728137254715, "global_step": 22005, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03398139029741287, "train_loss_bc": 0.03398139029741287, "train_loss_llm": 0.0, "grad_norm": 0.05463593080639839, "global_step": 22006, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03407658264040947, "train_loss_bc": 0.03406636416912079, "train_loss_llm": 0.001021742820739746, "grad_norm": 0.07407069951295853, "global_step": 22007, "epoch": 611, "lr": 0.001} +{"train_loss": 0.030488885939121246, "train_loss_bc": 0.030488885939121246, "train_loss_llm": 0.0, "grad_norm": 0.09101183712482452, "global_step": 22008, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03416484221816063, "train_loss_bc": 0.03416484221816063, "train_loss_llm": 0.0, "grad_norm": 0.03788319602608681, "global_step": 22009, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03465234860777855, "train_loss_bc": 0.0346510224044323, "train_loss_llm": 0.00013250112533569336, "grad_norm": 0.05989793315529823, "global_step": 22010, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03733200207352638, "train_loss_bc": 0.03733200207352638, "train_loss_llm": 0.0, "grad_norm": 0.07662274688482285, "global_step": 22011, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03177463635802269, "train_loss_bc": 0.03177463635802269, "train_loss_llm": 0.0, "grad_norm": 0.10215432196855545, "global_step": 22012, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03709438443183899, "train_loss_bc": 0.03708912059664726, "train_loss_llm": 0.0005263984203338623, "grad_norm": 0.029027285054326057, "global_step": 22013, "epoch": 611, "lr": 0.001} +{"train_loss": 0.029881563037633896, "train_loss_bc": 0.02987806312739849, "train_loss_llm": 0.00034993886947631836, "grad_norm": 0.044332001358270645, "global_step": 22014, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03128771483898163, "train_loss_bc": 0.031281907111406326, "train_loss_llm": 0.0005808174610137939, "grad_norm": 0.054084062576293945, "global_step": 22015, "epoch": 611, "lr": 0.001} +{"train_loss": 0.034737132489681244, "train_loss_bc": 0.034737132489681244, "train_loss_llm": 0.0, "grad_norm": 0.0611189529299736, "global_step": 22016, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03964439779520035, "train_loss_bc": 0.03964439779520035, "train_loss_llm": 0.0, "grad_norm": 0.03169969096779823, "global_step": 22017, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03603397682309151, "train_loss_bc": 0.03603397682309151, "train_loss_llm": 0.0, "grad_norm": 0.04907750338315964, "global_step": 22018, "epoch": 611, "lr": 0.001} +{"train_loss": 0.035045325756073, "train_loss_bc": 0.035037439316511154, "train_loss_llm": 0.0007886290550231934, "grad_norm": 0.05402769148349762, "global_step": 22019, "epoch": 611, "lr": 0.001} +{"train_loss": 0.030349498614668846, "train_loss_bc": 0.030334502458572388, "train_loss_llm": 0.0014996230602264404, "grad_norm": 0.06611205637454987, "global_step": 22020, "epoch": 611, "lr": 0.001} +{"train_loss": 0.032999251037836075, "train_loss_bc": 0.032999251037836075, "train_loss_llm": 0.0, "grad_norm": 0.03499717637896538, "global_step": 22021, "epoch": 611, "lr": 0.001} +{"train_loss": 0.02770501933991909, "train_loss_bc": 0.02770501933991909, "train_loss_llm": 0.0, "grad_norm": 0.04854699596762657, "global_step": 22022, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03694895654916763, "train_loss_bc": 0.03694895654916763, "train_loss_llm": 0.0, "grad_norm": 0.06686555594205856, "global_step": 22023, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03333471715450287, "train_loss_bc": 0.03333471715450287, "train_loss_llm": 0.0, "grad_norm": 0.08358598500490189, "global_step": 22024, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03537285327911377, "train_loss_bc": 0.03537285327911377, "train_loss_llm": 0.0, "grad_norm": 0.03367595374584198, "global_step": 22025, "epoch": 611, "lr": 0.001} +{"train_loss": 0.028956327587366104, "train_loss_bc": 0.028956327587366104, "train_loss_llm": 0.0, "grad_norm": 0.04568830132484436, "global_step": 22026, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03456730395555496, "train_loss_bc": 0.03456730395555496, "train_loss_llm": 0.0, "grad_norm": 0.06746966391801834, "global_step": 22027, "epoch": 611, "lr": 0.001} +{"train_loss": 0.029313569888472557, "train_loss_bc": 0.029308104887604713, "train_loss_llm": 0.000546574592590332, "grad_norm": 0.08016519248485565, "global_step": 22028, "epoch": 611, "lr": 0.001} +{"train_loss": 0.0286172516644001, "train_loss_bc": 0.0286172516644001, "train_loss_llm": 0.0, "grad_norm": 0.028767064213752747, "global_step": 22029, "epoch": 611, "lr": 0.001} +{"train_loss": 0.03339546546339989, "train_loss_bc": 0.03339546546339989, "train_loss_llm": 0.0, "grad_norm": 0.04341355338692665, "global_step": 22030, "epoch": 611, "lr": 0.001} +{"train_loss": 0.032451373938885, "train_loss_bc": 0.028991635888814926, "train_loss_llm": 0.0016160905361175537, "grad_norm": 0.06742087751626968, "global_step": 22031, "epoch": 611, "lr": 0.001, "val_loss": 0.08306236565113068} +{"train_loss": 0.03397423028945923, "train_loss_bc": 0.03397423028945923, "train_loss_llm": 0.0, "grad_norm": 0.09042477607727051, "global_step": 22032, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03503260761499405, "train_loss_bc": 0.03502601012587547, "train_loss_llm": 0.0006597936153411865, "grad_norm": 0.02986440435051918, "global_step": 22033, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03503584489226341, "train_loss_bc": 0.03503584489226341, "train_loss_llm": 0.0, "grad_norm": 0.048313602805137634, "global_step": 22034, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03370755910873413, "train_loss_bc": 0.03370755910873413, "train_loss_llm": 0.0, "grad_norm": 0.06398402899503708, "global_step": 22035, "epoch": 612, "lr": 0.001} +{"train_loss": 0.035249557346105576, "train_loss_bc": 0.035248029977083206, "train_loss_llm": 0.00015291571617126465, "grad_norm": 0.07768146693706512, "global_step": 22036, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03215039148926735, "train_loss_bc": 0.03215039148926735, "train_loss_llm": 0.0, "grad_norm": 0.02567024528980255, "global_step": 22037, "epoch": 612, "lr": 0.001} +{"train_loss": 0.030965032055974007, "train_loss_bc": 0.030965032055974007, "train_loss_llm": 0.0, "grad_norm": 0.03817988187074661, "global_step": 22038, "epoch": 612, "lr": 0.001} +{"train_loss": 0.033961258828639984, "train_loss_bc": 0.033961258828639984, "train_loss_llm": 0.0, "grad_norm": 0.05092921480536461, "global_step": 22039, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03164910525083542, "train_loss_bc": 0.03164910525083542, "train_loss_llm": 0.0, "grad_norm": 0.061059728264808655, "global_step": 22040, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03307424113154411, "train_loss_bc": 0.03307424113154411, "train_loss_llm": 0.0, "grad_norm": 0.035572599619627, "global_step": 22041, "epoch": 612, "lr": 0.001} +{"train_loss": 0.029565881937742233, "train_loss_bc": 0.029565881937742233, "train_loss_llm": 0.0, "grad_norm": 0.05361796170473099, "global_step": 22042, "epoch": 612, "lr": 0.001} +{"train_loss": 0.026439715176820755, "train_loss_bc": 0.02643563039600849, "train_loss_llm": 0.00040844082832336426, "grad_norm": 0.06548258662223816, "global_step": 22043, "epoch": 612, "lr": 0.001} +{"train_loss": 0.035725463181734085, "train_loss_bc": 0.035725463181734085, "train_loss_llm": 0.0, "grad_norm": 0.08507945388555527, "global_step": 22044, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03332749754190445, "train_loss_bc": 0.03332749754190445, "train_loss_llm": 0.0, "grad_norm": 0.032267022877931595, "global_step": 22045, "epoch": 612, "lr": 0.001} +{"train_loss": 0.026728443801403046, "train_loss_bc": 0.02672692947089672, "train_loss_llm": 0.00015145540237426758, "grad_norm": 0.046864114701747894, "global_step": 22046, "epoch": 612, "lr": 0.001} +{"train_loss": 0.02575063705444336, "train_loss_bc": 0.025747928768396378, "train_loss_llm": 0.0002708137035369873, "grad_norm": 0.054048798978328705, "global_step": 22047, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03429977223277092, "train_loss_bc": 0.03429977223277092, "train_loss_llm": 0.0, "grad_norm": 0.07291815429925919, "global_step": 22048, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03480246290564537, "train_loss_bc": 0.03480246290564537, "train_loss_llm": 0.0, "grad_norm": 0.03373867645859718, "global_step": 22049, "epoch": 612, "lr": 0.001} +{"train_loss": 0.028207501396536827, "train_loss_bc": 0.028196455910801888, "train_loss_llm": 0.0011046230792999268, "grad_norm": 0.0527971051633358, "global_step": 22050, "epoch": 612, "lr": 0.001} +{"train_loss": 0.030344801023602486, "train_loss_bc": 0.030335569754242897, "train_loss_llm": 0.0009232163429260254, "grad_norm": 0.07408670336008072, "global_step": 22051, "epoch": 612, "lr": 0.001} +{"train_loss": 0.035919420421123505, "train_loss_bc": 0.035915110260248184, "train_loss_llm": 0.0004311203956604004, "grad_norm": 0.08916943520307541, "global_step": 22052, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03739801421761513, "train_loss_bc": 0.03739801421761513, "train_loss_llm": 0.0, "grad_norm": 0.031565096229314804, "global_step": 22053, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03198070451617241, "train_loss_bc": 0.03198070451617241, "train_loss_llm": 0.0, "grad_norm": 0.04654688015580177, "global_step": 22054, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03235558792948723, "train_loss_bc": 0.03235558792948723, "train_loss_llm": 0.0, "grad_norm": 0.06873400509357452, "global_step": 22055, "epoch": 612, "lr": 0.001} +{"train_loss": 0.033407121896743774, "train_loss_bc": 0.03339480236172676, "train_loss_llm": 0.0012320876121520996, "grad_norm": 0.08854883164167404, "global_step": 22056, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03327170014381409, "train_loss_bc": 0.03327170014381409, "train_loss_llm": 0.0, "grad_norm": 0.031618632376194, "global_step": 22057, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03165813907980919, "train_loss_bc": 0.031634993851184845, "train_loss_llm": 0.0023144185543060303, "grad_norm": 0.04075929895043373, "global_step": 22058, "epoch": 612, "lr": 0.001} +{"train_loss": 0.02997545897960663, "train_loss_bc": 0.02997545897960663, "train_loss_llm": 0.0, "grad_norm": 0.054907556623220444, "global_step": 22059, "epoch": 612, "lr": 0.001} +{"train_loss": 0.033336859196424484, "train_loss_bc": 0.033321674913167953, "train_loss_llm": 0.0015183985233306885, "grad_norm": 0.07557574659585953, "global_step": 22060, "epoch": 612, "lr": 0.001} +{"train_loss": 0.029856961220502853, "train_loss_bc": 0.029856961220502853, "train_loss_llm": 0.0, "grad_norm": 0.030708037316799164, "global_step": 22061, "epoch": 612, "lr": 0.001} +{"train_loss": 0.02785983495414257, "train_loss_bc": 0.02785983495414257, "train_loss_llm": 0.0, "grad_norm": 0.05079947039484978, "global_step": 22062, "epoch": 612, "lr": 0.001} +{"train_loss": 0.031201239675283432, "train_loss_bc": 0.031201239675283432, "train_loss_llm": 0.0, "grad_norm": 0.06919045001268387, "global_step": 22063, "epoch": 612, "lr": 0.001} +{"train_loss": 0.033983014523983, "train_loss_bc": 0.033983014523983, "train_loss_llm": 0.0, "grad_norm": 0.0828649178147316, "global_step": 22064, "epoch": 612, "lr": 0.001} +{"train_loss": 0.038513701409101486, "train_loss_bc": 0.038513701409101486, "train_loss_llm": 0.0, "grad_norm": 0.03431934863328934, "global_step": 22065, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03618660196661949, "train_loss_bc": 0.03618660196661949, "train_loss_llm": 0.0, "grad_norm": 0.05835513770580292, "global_step": 22066, "epoch": 612, "lr": 0.001} +{"train_loss": 0.03249142298267947, "train_loss_bc": 0.03279486298561096, "train_loss_llm": 0.0, "grad_norm": 0.08259508013725281, "global_step": 22067, "epoch": 612, "lr": 0.001, "val_loss": 0.08214224874973297} +{"train_loss": 0.03174874186515808, "train_loss_bc": 0.03174874186515808, "train_loss_llm": 0.0, "grad_norm": 0.10060045123100281, "global_step": 22068, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03425215557217598, "train_loss_bc": 0.03425215557217598, "train_loss_llm": 0.0, "grad_norm": 0.02728065848350525, "global_step": 22069, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03446482494473457, "train_loss_bc": 0.03445452079176903, "train_loss_llm": 0.0010304152965545654, "grad_norm": 0.03601812571287155, "global_step": 22070, "epoch": 613, "lr": 0.001} +{"train_loss": 0.027153829112648964, "train_loss_bc": 0.02714499831199646, "train_loss_llm": 0.0008831620216369629, "grad_norm": 0.05204305797815323, "global_step": 22071, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03327272832393646, "train_loss_bc": 0.03327241539955139, "train_loss_llm": 3.123283386230469e-05, "grad_norm": 0.06403885036706924, "global_step": 22072, "epoch": 613, "lr": 0.001} +{"train_loss": 0.030410002917051315, "train_loss_bc": 0.030407700687646866, "train_loss_llm": 0.0002301633358001709, "grad_norm": 0.028638068586587906, "global_step": 22073, "epoch": 613, "lr": 0.001} +{"train_loss": 0.029514627531170845, "train_loss_bc": 0.029514627531170845, "train_loss_llm": 0.0, "grad_norm": 0.03962545096874237, "global_step": 22074, "epoch": 613, "lr": 0.001} +{"train_loss": 0.029825450852513313, "train_loss_bc": 0.029825450852513313, "train_loss_llm": 0.0, "grad_norm": 0.050943899899721146, "global_step": 22075, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03399315103888512, "train_loss_bc": 0.03399315103888512, "train_loss_llm": 0.0, "grad_norm": 0.06280648708343506, "global_step": 22076, "epoch": 613, "lr": 0.001} +{"train_loss": 0.031017985194921494, "train_loss_bc": 0.031017985194921494, "train_loss_llm": 0.0, "grad_norm": 0.028571242466568947, "global_step": 22077, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03009832836687565, "train_loss_bc": 0.030079839751124382, "train_loss_llm": 0.0018489360809326172, "grad_norm": 0.04526825249195099, "global_step": 22078, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03294890746474266, "train_loss_bc": 0.03294890746474266, "train_loss_llm": 0.0, "grad_norm": 0.06807850301265717, "global_step": 22079, "epoch": 613, "lr": 0.001} +{"train_loss": 0.035326018929481506, "train_loss_bc": 0.035326018929481506, "train_loss_llm": 0.0, "grad_norm": 0.08693332970142365, "global_step": 22080, "epoch": 613, "lr": 0.001} +{"train_loss": 0.034904252737760544, "train_loss_bc": 0.03488701581954956, "train_loss_llm": 0.0017237663269042969, "grad_norm": 0.031058909371495247, "global_step": 22081, "epoch": 613, "lr": 0.001} +{"train_loss": 0.030297208577394485, "train_loss_bc": 0.030297208577394485, "train_loss_llm": 0.0, "grad_norm": 0.046948596835136414, "global_step": 22082, "epoch": 613, "lr": 0.001} +{"train_loss": 0.026644570752978325, "train_loss_bc": 0.02663704752922058, "train_loss_llm": 0.0007523894309997559, "grad_norm": 0.058102332055568695, "global_step": 22083, "epoch": 613, "lr": 0.001} +{"train_loss": 0.029203126206994057, "train_loss_bc": 0.029203126206994057, "train_loss_llm": 0.0, "grad_norm": 0.07470369338989258, "global_step": 22084, "epoch": 613, "lr": 0.001} +{"train_loss": 0.025095121935009956, "train_loss_bc": 0.025076644495129585, "train_loss_llm": 0.001847773790359497, "grad_norm": 0.022113725543022156, "global_step": 22085, "epoch": 613, "lr": 0.001} +{"train_loss": 0.0308661088347435, "train_loss_bc": 0.0308661088347435, "train_loss_llm": 0.0, "grad_norm": 0.036618418991565704, "global_step": 22086, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03388093039393425, "train_loss_bc": 0.03387180343270302, "train_loss_llm": 0.000912785530090332, "grad_norm": 0.0554695688188076, "global_step": 22087, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03447529673576355, "train_loss_bc": 0.03447529673576355, "train_loss_llm": 0.0, "grad_norm": 0.06659633666276932, "global_step": 22088, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03173438459634781, "train_loss_bc": 0.03173438459634781, "train_loss_llm": 0.0, "grad_norm": 0.02571277692914009, "global_step": 22089, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03398096561431885, "train_loss_bc": 0.03398096561431885, "train_loss_llm": 0.0, "grad_norm": 0.040660154074430466, "global_step": 22090, "epoch": 613, "lr": 0.001} +{"train_loss": 0.039808839559555054, "train_loss_bc": 0.039808839559555054, "train_loss_llm": 0.0, "grad_norm": 0.054618772119283676, "global_step": 22091, "epoch": 613, "lr": 0.001} +{"train_loss": 0.031943291425704956, "train_loss_bc": 0.031943291425704956, "train_loss_llm": 0.0, "grad_norm": 0.06709630787372589, "global_step": 22092, "epoch": 613, "lr": 0.001} +{"train_loss": 0.036303501576185226, "train_loss_bc": 0.036303501576185226, "train_loss_llm": 0.0, "grad_norm": 0.03683541342616081, "global_step": 22093, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03582855314016342, "train_loss_bc": 0.03580935299396515, "train_loss_llm": 0.0019200146198272705, "grad_norm": 0.05877090245485306, "global_step": 22094, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03775092959403992, "train_loss_bc": 0.03775092959403992, "train_loss_llm": 0.0, "grad_norm": 0.08180014044046402, "global_step": 22095, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03345927968621254, "train_loss_bc": 0.03345927968621254, "train_loss_llm": 0.0, "grad_norm": 0.09732157737016678, "global_step": 22096, "epoch": 613, "lr": 0.001} +{"train_loss": 0.02881513722240925, "train_loss_bc": 0.02879934571683407, "train_loss_llm": 0.0015791356563568115, "grad_norm": 0.03056839108467102, "global_step": 22097, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03269617632031441, "train_loss_bc": 0.03269617632031441, "train_loss_llm": 0.0, "grad_norm": 0.038659073412418365, "global_step": 22098, "epoch": 613, "lr": 0.001} +{"train_loss": 0.04003199189901352, "train_loss_bc": 0.04003199189901352, "train_loss_llm": 0.0, "grad_norm": 0.06728605926036835, "global_step": 22099, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03334081172943115, "train_loss_bc": 0.03334081172943115, "train_loss_llm": 0.0, "grad_norm": 0.09415356814861298, "global_step": 22100, "epoch": 613, "lr": 0.001} +{"train_loss": 0.02786610648036003, "train_loss_bc": 0.02786610648036003, "train_loss_llm": 0.0, "grad_norm": 0.025964757427573204, "global_step": 22101, "epoch": 613, "lr": 0.001} +{"train_loss": 0.03250640258193016, "train_loss_bc": 0.03250640258193016, "train_loss_llm": 0.0, "grad_norm": 0.04408583417534828, "global_step": 22102, "epoch": 613, "lr": 0.001} +{"train_loss": 0.032342914698852435, "train_loss_bc": 0.028885139152407646, "train_loss_llm": 5.036592483520508e-06, "grad_norm": 0.06502926349639893, "global_step": 22103, "epoch": 613, "lr": 0.001, "val_loss": 0.08266925066709518} +{"train_loss": 0.03819349780678749, "train_loss_bc": 0.03819349780678749, "train_loss_llm": 0.0, "grad_norm": 0.08700412511825562, "global_step": 22104, "epoch": 614, "lr": 0.001} +{"train_loss": 0.0319073349237442, "train_loss_bc": 0.031899526715278625, "train_loss_llm": 0.000780642032623291, "grad_norm": 0.02574140951037407, "global_step": 22105, "epoch": 614, "lr": 0.001} +{"train_loss": 0.02542215771973133, "train_loss_bc": 0.025417903438210487, "train_loss_llm": 0.0004253685474395752, "grad_norm": 0.039640799164772034, "global_step": 22106, "epoch": 614, "lr": 0.001} +{"train_loss": 0.030424684286117554, "train_loss_bc": 0.030424684286117554, "train_loss_llm": 0.0, "grad_norm": 0.052610933780670166, "global_step": 22107, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03168801963329315, "train_loss_bc": 0.03168801963329315, "train_loss_llm": 0.0, "grad_norm": 0.06759101897478104, "global_step": 22108, "epoch": 614, "lr": 0.001} +{"train_loss": 0.033685971051454544, "train_loss_bc": 0.033685971051454544, "train_loss_llm": 0.0, "grad_norm": 0.02730651944875717, "global_step": 22109, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03396983817219734, "train_loss_bc": 0.033966224640607834, "train_loss_llm": 0.00036150217056274414, "grad_norm": 0.04401082545518875, "global_step": 22110, "epoch": 614, "lr": 0.001} +{"train_loss": 0.02999234013259411, "train_loss_bc": 0.029978495091199875, "train_loss_llm": 0.0013845860958099365, "grad_norm": 0.06158392131328583, "global_step": 22111, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03707686439156532, "train_loss_bc": 0.03707686439156532, "train_loss_llm": 0.0, "grad_norm": 0.07412475347518921, "global_step": 22112, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03571990504860878, "train_loss_bc": 0.035711683332920074, "train_loss_llm": 0.0008220076560974121, "grad_norm": 0.03831661865115166, "global_step": 22113, "epoch": 614, "lr": 0.001} +{"train_loss": 0.032953813672065735, "train_loss_bc": 0.032951630651950836, "train_loss_llm": 0.00021845102310180664, "grad_norm": 0.05707288160920143, "global_step": 22114, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03305383026599884, "train_loss_bc": 0.03305383026599884, "train_loss_llm": 0.0, "grad_norm": 0.07561157643795013, "global_step": 22115, "epoch": 614, "lr": 0.001} +{"train_loss": 0.029717573896050453, "train_loss_bc": 0.029717573896050453, "train_loss_llm": 0.0, "grad_norm": 0.0957740917801857, "global_step": 22116, "epoch": 614, "lr": 0.001} +{"train_loss": 0.032440733164548874, "train_loss_bc": 0.032440733164548874, "train_loss_llm": 0.0, "grad_norm": 0.03620685636997223, "global_step": 22117, "epoch": 614, "lr": 0.001} +{"train_loss": 0.029765013605356216, "train_loss_bc": 0.029765013605356216, "train_loss_llm": 0.0, "grad_norm": 0.052382323890924454, "global_step": 22118, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03481864929199219, "train_loss_bc": 0.03481864929199219, "train_loss_llm": 0.0, "grad_norm": 0.07666503638029099, "global_step": 22119, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03681770712137222, "train_loss_bc": 0.03680340200662613, "train_loss_llm": 0.001430422067642212, "grad_norm": 0.08886303752660751, "global_step": 22120, "epoch": 614, "lr": 0.001} +{"train_loss": 0.031506698578596115, "train_loss_bc": 0.031506698578596115, "train_loss_llm": 0.0, "grad_norm": 0.022385748103260994, "global_step": 22121, "epoch": 614, "lr": 0.001} +{"train_loss": 0.033116959035396576, "train_loss_bc": 0.033116020262241364, "train_loss_llm": 9.372830390930176e-05, "grad_norm": 0.03273259848356247, "global_step": 22122, "epoch": 614, "lr": 0.001} +{"train_loss": 0.023487379774451256, "train_loss_bc": 0.023487379774451256, "train_loss_llm": 0.0, "grad_norm": 0.041153497993946075, "global_step": 22123, "epoch": 614, "lr": 0.001} +{"train_loss": 0.030345570296049118, "train_loss_bc": 0.030341167002916336, "train_loss_llm": 0.00044023990631103516, "grad_norm": 0.052984628826379776, "global_step": 22124, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03355790674686432, "train_loss_bc": 0.03355790674686432, "train_loss_llm": 0.0, "grad_norm": 0.033306512981653214, "global_step": 22125, "epoch": 614, "lr": 0.001} +{"train_loss": 0.030417896807193756, "train_loss_bc": 0.030417896807193756, "train_loss_llm": 0.0, "grad_norm": 0.050869591534137726, "global_step": 22126, "epoch": 614, "lr": 0.001} +{"train_loss": 0.028967250138521194, "train_loss_bc": 0.028960902243852615, "train_loss_llm": 0.0006348788738250732, "grad_norm": 0.062215693295001984, "global_step": 22127, "epoch": 614, "lr": 0.001} +{"train_loss": 0.029349178075790405, "train_loss_bc": 0.029349178075790405, "train_loss_llm": 0.0, "grad_norm": 0.06753720343112946, "global_step": 22128, "epoch": 614, "lr": 0.001} +{"train_loss": 0.031635478138923645, "train_loss_bc": 0.031635478138923645, "train_loss_llm": 0.0, "grad_norm": 0.029513895511627197, "global_step": 22129, "epoch": 614, "lr": 0.001} +{"train_loss": 0.035715725272893906, "train_loss_bc": 0.035715725272893906, "train_loss_llm": 0.0, "grad_norm": 0.04985897243022919, "global_step": 22130, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03622772917151451, "train_loss_bc": 0.03622772917151451, "train_loss_llm": 0.0, "grad_norm": 0.06815043836832047, "global_step": 22131, "epoch": 614, "lr": 0.001} +{"train_loss": 0.033890485763549805, "train_loss_bc": 0.033890485763549805, "train_loss_llm": 0.0, "grad_norm": 0.07657261937856674, "global_step": 22132, "epoch": 614, "lr": 0.001} +{"train_loss": 0.029177188873291016, "train_loss_bc": 0.029177188873291016, "train_loss_llm": 0.0, "grad_norm": 0.02869740128517151, "global_step": 22133, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03603605180978775, "train_loss_bc": 0.03603605180978775, "train_loss_llm": 0.0, "grad_norm": 0.04512948542833328, "global_step": 22134, "epoch": 614, "lr": 0.001} +{"train_loss": 0.0355222150683403, "train_loss_bc": 0.0355222150683403, "train_loss_llm": 0.0, "grad_norm": 0.06451927125453949, "global_step": 22135, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03172909840941429, "train_loss_bc": 0.03172909840941429, "train_loss_llm": 0.0, "grad_norm": 0.07791346311569214, "global_step": 22136, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03594231978058815, "train_loss_bc": 0.03594231978058815, "train_loss_llm": 0.0, "grad_norm": 0.03256456181406975, "global_step": 22137, "epoch": 614, "lr": 0.001} +{"train_loss": 0.028634052723646164, "train_loss_bc": 0.028634052723646164, "train_loss_llm": 0.0, "grad_norm": 0.04498744755983353, "global_step": 22138, "epoch": 614, "lr": 0.001} +{"train_loss": 0.03248540901889404, "train_loss_bc": 0.036567606031894684, "train_loss_llm": 0.0, "grad_norm": 0.06488745659589767, "global_step": 22139, "epoch": 614, "lr": 0.001, "val_loss": 0.08497726917266846} +{"train_loss": 0.031560588628053665, "train_loss_bc": 0.031560588628053665, "train_loss_llm": 0.0, "grad_norm": 0.08802305161952972, "global_step": 22140, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03492092713713646, "train_loss_bc": 0.03492092713713646, "train_loss_llm": 0.0, "grad_norm": 0.03633967787027359, "global_step": 22141, "epoch": 615, "lr": 0.001} +{"train_loss": 0.032587286084890366, "train_loss_bc": 0.03258691728115082, "train_loss_llm": 3.6776065826416016e-05, "grad_norm": 0.052579235285520554, "global_step": 22142, "epoch": 615, "lr": 0.001} +{"train_loss": 0.02734716422855854, "train_loss_bc": 0.02734716422855854, "train_loss_llm": 0.0, "grad_norm": 0.07729340344667435, "global_step": 22143, "epoch": 615, "lr": 0.001} +{"train_loss": 0.02979438193142414, "train_loss_bc": 0.02979438193142414, "train_loss_llm": 0.0, "grad_norm": 0.08979339152574539, "global_step": 22144, "epoch": 615, "lr": 0.001} +{"train_loss": 0.026437358930706978, "train_loss_bc": 0.026437358930706978, "train_loss_llm": 0.0, "grad_norm": 0.025603430345654488, "global_step": 22145, "epoch": 615, "lr": 0.001} +{"train_loss": 0.030057238414883614, "train_loss_bc": 0.03004320338368416, "train_loss_llm": 0.0014034509658813477, "grad_norm": 0.03894810751080513, "global_step": 22146, "epoch": 615, "lr": 0.001} +{"train_loss": 0.029297517612576485, "train_loss_bc": 0.029297517612576485, "train_loss_llm": 0.0, "grad_norm": 0.057064272463321686, "global_step": 22147, "epoch": 615, "lr": 0.001} +{"train_loss": 0.031087379902601242, "train_loss_bc": 0.03108328953385353, "train_loss_llm": 0.00040906667709350586, "grad_norm": 0.0656026154756546, "global_step": 22148, "epoch": 615, "lr": 0.001} +{"train_loss": 0.037434112280607224, "train_loss_bc": 0.037434112280607224, "train_loss_llm": 0.0, "grad_norm": 0.040955834090709686, "global_step": 22149, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03518322482705116, "train_loss_bc": 0.03518322482705116, "train_loss_llm": 0.0, "grad_norm": 0.05330808460712433, "global_step": 22150, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03470242768526077, "train_loss_bc": 0.03470242768526077, "train_loss_llm": 0.0, "grad_norm": 0.06766732037067413, "global_step": 22151, "epoch": 615, "lr": 0.001} +{"train_loss": 0.02853361889719963, "train_loss_bc": 0.02852180041372776, "train_loss_llm": 0.0011818110942840576, "grad_norm": 0.080134317278862, "global_step": 22152, "epoch": 615, "lr": 0.001} +{"train_loss": 0.02921397238969803, "train_loss_bc": 0.029208900406956673, "train_loss_llm": 0.0005071163177490234, "grad_norm": 0.0338563397526741, "global_step": 22153, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03335534408688545, "train_loss_bc": 0.03334995359182358, "train_loss_llm": 0.0005390942096710205, "grad_norm": 0.05421968176960945, "global_step": 22154, "epoch": 615, "lr": 0.001} +{"train_loss": 0.028066566213965416, "train_loss_bc": 0.028066566213965416, "train_loss_llm": 0.0, "grad_norm": 0.07132723182439804, "global_step": 22155, "epoch": 615, "lr": 0.001} +{"train_loss": 0.031102579087018967, "train_loss_bc": 0.031102579087018967, "train_loss_llm": 0.0, "grad_norm": 0.08209710568189621, "global_step": 22156, "epoch": 615, "lr": 0.001} +{"train_loss": 0.029336778447031975, "train_loss_bc": 0.029336778447031975, "train_loss_llm": 0.0, "grad_norm": 0.03330123797059059, "global_step": 22157, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03730132803320885, "train_loss_bc": 0.03730132803320885, "train_loss_llm": 0.0, "grad_norm": 0.04946954548358917, "global_step": 22158, "epoch": 615, "lr": 0.001} +{"train_loss": 0.028994586318731308, "train_loss_bc": 0.028994586318731308, "train_loss_llm": 0.0, "grad_norm": 0.07316520810127258, "global_step": 22159, "epoch": 615, "lr": 0.001} +{"train_loss": 0.032823819667100906, "train_loss_bc": 0.032823819667100906, "train_loss_llm": 0.0, "grad_norm": 0.08572757244110107, "global_step": 22160, "epoch": 615, "lr": 0.001} +{"train_loss": 0.02878168225288391, "train_loss_bc": 0.02877626195549965, "train_loss_llm": 0.0005420446395874023, "grad_norm": 0.028888020664453506, "global_step": 22161, "epoch": 615, "lr": 0.001} +{"train_loss": 0.0344783179461956, "train_loss_bc": 0.0344783179461956, "train_loss_llm": 0.0, "grad_norm": 0.0498429574072361, "global_step": 22162, "epoch": 615, "lr": 0.001} +{"train_loss": 0.038054004311561584, "train_loss_bc": 0.03804314509034157, "train_loss_llm": 0.0010858774185180664, "grad_norm": 0.07728815078735352, "global_step": 22163, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03886152058839798, "train_loss_bc": 0.03884950280189514, "train_loss_llm": 0.0012015998363494873, "grad_norm": 0.09014735370874405, "global_step": 22164, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03285643830895424, "train_loss_bc": 0.03285643830895424, "train_loss_llm": 0.0, "grad_norm": 0.03417867422103882, "global_step": 22165, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03174732252955437, "train_loss_bc": 0.03174732252955437, "train_loss_llm": 0.0, "grad_norm": 0.05713025480508804, "global_step": 22166, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03228259086608887, "train_loss_bc": 0.03228259086608887, "train_loss_llm": 0.0, "grad_norm": 0.07281219214200974, "global_step": 22167, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03728101775050163, "train_loss_bc": 0.03728101775050163, "train_loss_llm": 0.0, "grad_norm": 0.09209809452295303, "global_step": 22168, "epoch": 615, "lr": 0.001} +{"train_loss": 0.0354691818356514, "train_loss_bc": 0.0354691818356514, "train_loss_llm": 0.0, "grad_norm": 0.03268387168645859, "global_step": 22169, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03093709982931614, "train_loss_bc": 0.03093709982931614, "train_loss_llm": 0.0, "grad_norm": 0.06003749743103981, "global_step": 22170, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03159837797284126, "train_loss_bc": 0.03159291297197342, "train_loss_llm": 0.000546574592590332, "grad_norm": 0.07170850783586502, "global_step": 22171, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03668268769979477, "train_loss_bc": 0.03668268769979477, "train_loss_llm": 0.0, "grad_norm": 0.08595030754804611, "global_step": 22172, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03324422612786293, "train_loss_bc": 0.03324422612786293, "train_loss_llm": 0.0, "grad_norm": 0.034970421344041824, "global_step": 22173, "epoch": 615, "lr": 0.001} +{"train_loss": 0.03841627761721611, "train_loss_bc": 0.03841627761721611, "train_loss_llm": 0.0, "grad_norm": 0.04933825507760048, "global_step": 22174, "epoch": 615, "lr": 0.001} +{"train_loss": 0.032543620819018945, "train_loss_bc": 0.03173450380563736, "train_loss_llm": 0.0006898045539855957, "grad_norm": 0.061967141926288605, "global_step": 22175, "epoch": 615, "lr": 0.001, "train/cumulative_reward": 88.55316455317976, "train/mean_score": 17.834296491067608, "train/success_rate": 0.5, "test/cumulative_reward": 35.47868415159221, "test/mean_score": 9.848034910598878, "test/success_rate": 0.28, "val_loss": 0.08446468412876129, "train_action_mse_error": 0.0358240082859993} +{"train_loss": 0.037817392498254776, "train_loss_bc": 0.03779216855764389, "train_loss_llm": 0.0025224387645721436, "grad_norm": 0.07539104670286179, "global_step": 22176, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03140539675951004, "train_loss_bc": 0.03140539675951004, "train_loss_llm": 0.0, "grad_norm": 0.02925333008170128, "global_step": 22177, "epoch": 616, "lr": 0.001} +{"train_loss": 0.028988197445869446, "train_loss_bc": 0.028988197445869446, "train_loss_llm": 0.0, "grad_norm": 0.049601223319768906, "global_step": 22178, "epoch": 616, "lr": 0.001} +{"train_loss": 0.02937389723956585, "train_loss_bc": 0.02937389723956585, "train_loss_llm": 0.0, "grad_norm": 0.0606486052274704, "global_step": 22179, "epoch": 616, "lr": 0.001} +{"train_loss": 0.029371200129389763, "train_loss_bc": 0.029358070343732834, "train_loss_llm": 0.001312941312789917, "grad_norm": 0.08700447529554367, "global_step": 22180, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03260653465986252, "train_loss_bc": 0.03259681910276413, "train_loss_llm": 0.0009716451168060303, "grad_norm": 0.03480460122227669, "global_step": 22181, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03788994997739792, "train_loss_bc": 0.03788994997739792, "train_loss_llm": 0.0, "grad_norm": 0.06692468374967575, "global_step": 22182, "epoch": 616, "lr": 0.001} +{"train_loss": 0.02914527989923954, "train_loss_bc": 0.02914527989923954, "train_loss_llm": 0.0, "grad_norm": 0.08594338595867157, "global_step": 22183, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03452082350850105, "train_loss_bc": 0.03452082350850105, "train_loss_llm": 0.0, "grad_norm": 0.11092443019151688, "global_step": 22184, "epoch": 616, "lr": 0.001} +{"train_loss": 0.029948558658361435, "train_loss_bc": 0.029948558658361435, "train_loss_llm": 0.0, "grad_norm": 0.028288915753364563, "global_step": 22185, "epoch": 616, "lr": 0.001} +{"train_loss": 0.038806430995464325, "train_loss_bc": 0.038806430995464325, "train_loss_llm": 0.0, "grad_norm": 0.05258606746792793, "global_step": 22186, "epoch": 616, "lr": 0.001} +{"train_loss": 0.02851332165300846, "train_loss_bc": 0.028513045981526375, "train_loss_llm": 2.765655517578125e-05, "grad_norm": 0.06623506546020508, "global_step": 22187, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03361808881163597, "train_loss_bc": 0.03361808881163597, "train_loss_llm": 0.0, "grad_norm": 0.0813707560300827, "global_step": 22188, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03450819104909897, "train_loss_bc": 0.0344889834523201, "train_loss_llm": 0.0019207000732421875, "grad_norm": 0.030465660616755486, "global_step": 22189, "epoch": 616, "lr": 0.001} +{"train_loss": 0.033724334090948105, "train_loss_bc": 0.03371027112007141, "train_loss_llm": 0.0014061331748962402, "grad_norm": 0.04468298703432083, "global_step": 22190, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03185021132230759, "train_loss_bc": 0.03185021132230759, "train_loss_llm": 0.0, "grad_norm": 0.06219090521335602, "global_step": 22191, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03308359161019325, "train_loss_bc": 0.03307308256626129, "train_loss_llm": 0.0010509192943572998, "grad_norm": 0.07627829164266586, "global_step": 22192, "epoch": 616, "lr": 0.001} +{"train_loss": 0.032402366399765015, "train_loss_bc": 0.032388053834438324, "train_loss_llm": 0.0014314353466033936, "grad_norm": 0.026750100776553154, "global_step": 22193, "epoch": 616, "lr": 0.001} +{"train_loss": 0.0340425968170166, "train_loss_bc": 0.034037038683891296, "train_loss_llm": 0.0005557835102081299, "grad_norm": 0.05117511749267578, "global_step": 22194, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03650733828544617, "train_loss_bc": 0.03650733828544617, "train_loss_llm": 0.0, "grad_norm": 0.0714297890663147, "global_step": 22195, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03425799682736397, "train_loss_bc": 0.03425799682736397, "train_loss_llm": 0.0, "grad_norm": 0.09248736500740051, "global_step": 22196, "epoch": 616, "lr": 0.001} +{"train_loss": 0.036367446184158325, "train_loss_bc": 0.036367446184158325, "train_loss_llm": 0.0, "grad_norm": 0.03581764176487923, "global_step": 22197, "epoch": 616, "lr": 0.001} +{"train_loss": 0.034297049045562744, "train_loss_bc": 0.034280385822057724, "train_loss_llm": 0.001666337251663208, "grad_norm": 0.05611313134431839, "global_step": 22198, "epoch": 616, "lr": 0.001} +{"train_loss": 0.031748440116643906, "train_loss_bc": 0.031748440116643906, "train_loss_llm": 0.0, "grad_norm": 0.07601986080408096, "global_step": 22199, "epoch": 616, "lr": 0.001} +{"train_loss": 0.036722294986248016, "train_loss_bc": 0.036722294986248016, "train_loss_llm": 0.0, "grad_norm": 0.09787408262491226, "global_step": 22200, "epoch": 616, "lr": 0.001} +{"train_loss": 0.028967872262001038, "train_loss_bc": 0.028967872262001038, "train_loss_llm": 0.0, "grad_norm": 0.029347695410251617, "global_step": 22201, "epoch": 616, "lr": 0.001} +{"train_loss": 0.031155461445450783, "train_loss_bc": 0.03115250915288925, "train_loss_llm": 0.00029531121253967285, "grad_norm": 0.04562871530652046, "global_step": 22202, "epoch": 616, "lr": 0.001} +{"train_loss": 0.031199604272842407, "train_loss_bc": 0.031199604272842407, "train_loss_llm": 0.0, "grad_norm": 0.06987720727920532, "global_step": 22203, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03324374929070473, "train_loss_bc": 0.03324374929070473, "train_loss_llm": 0.0, "grad_norm": 0.09311044961214066, "global_step": 22204, "epoch": 616, "lr": 0.001} +{"train_loss": 0.030575884506106377, "train_loss_bc": 0.030575884506106377, "train_loss_llm": 0.0, "grad_norm": 0.0284360833466053, "global_step": 22205, "epoch": 616, "lr": 0.001} +{"train_loss": 0.033717505633831024, "train_loss_bc": 0.033713214099407196, "train_loss_llm": 0.0004292130470275879, "grad_norm": 0.045455001294612885, "global_step": 22206, "epoch": 616, "lr": 0.001} +{"train_loss": 0.032658353447914124, "train_loss_bc": 0.03264665603637695, "train_loss_llm": 0.0011698603630065918, "grad_norm": 0.05774720385670662, "global_step": 22207, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03457801043987274, "train_loss_bc": 0.03457801043987274, "train_loss_llm": 0.0, "grad_norm": 0.07550301402807236, "global_step": 22208, "epoch": 616, "lr": 0.001} +{"train_loss": 0.030858539044857025, "train_loss_bc": 0.030858539044857025, "train_loss_llm": 0.0, "grad_norm": 0.03394103795289993, "global_step": 22209, "epoch": 616, "lr": 0.001} +{"train_loss": 0.026732148602604866, "train_loss_bc": 0.026718098670244217, "train_loss_llm": 0.0014049410820007324, "grad_norm": 0.055670976638793945, "global_step": 22210, "epoch": 616, "lr": 0.001} +{"train_loss": 0.03271027028353678, "train_loss_bc": 0.032365672290325165, "train_loss_llm": 0.0, "grad_norm": 0.09303779155015945, "global_step": 22211, "epoch": 616, "lr": 0.001, "val_loss": 0.0835242047905922} +{"train_loss": 0.03100672923028469, "train_loss_bc": 0.03100672923028469, "train_loss_llm": 0.0, "grad_norm": 0.1057564914226532, "global_step": 22212, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03329167515039444, "train_loss_bc": 0.03329167515039444, "train_loss_llm": 0.0, "grad_norm": 0.03150190785527229, "global_step": 22213, "epoch": 617, "lr": 0.001} +{"train_loss": 0.02689671330153942, "train_loss_bc": 0.02689671330153942, "train_loss_llm": 0.0, "grad_norm": 0.038146138191223145, "global_step": 22214, "epoch": 617, "lr": 0.001} +{"train_loss": 0.02730688638985157, "train_loss_bc": 0.02730688638985157, "train_loss_llm": 0.0, "grad_norm": 0.05434482917189598, "global_step": 22215, "epoch": 617, "lr": 0.001} +{"train_loss": 0.033015020191669464, "train_loss_bc": 0.033015020191669464, "train_loss_llm": 0.0, "grad_norm": 0.0771268904209137, "global_step": 22216, "epoch": 617, "lr": 0.001} +{"train_loss": 0.0394706204533577, "train_loss_bc": 0.0394706204533577, "train_loss_llm": 0.0, "grad_norm": 0.0388597697019577, "global_step": 22217, "epoch": 617, "lr": 0.001} +{"train_loss": 0.025901105254888535, "train_loss_bc": 0.025901105254888535, "train_loss_llm": 0.0, "grad_norm": 0.05123072862625122, "global_step": 22218, "epoch": 617, "lr": 0.001} +{"train_loss": 0.031221244484186172, "train_loss_bc": 0.031214559450745583, "train_loss_llm": 0.0006684362888336182, "grad_norm": 0.07937546819448471, "global_step": 22219, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03454558923840523, "train_loss_bc": 0.03454558923840523, "train_loss_llm": 0.0, "grad_norm": 0.10505259782075882, "global_step": 22220, "epoch": 617, "lr": 0.001} +{"train_loss": 0.031089795753359795, "train_loss_bc": 0.031089795753359795, "train_loss_llm": 0.0, "grad_norm": 0.02761632576584816, "global_step": 22221, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03427528217434883, "train_loss_bc": 0.03427528217434883, "train_loss_llm": 0.0, "grad_norm": 0.050076086074113846, "global_step": 22222, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03707882761955261, "train_loss_bc": 0.03707882761955261, "train_loss_llm": 0.0, "grad_norm": 0.06318967789411545, "global_step": 22223, "epoch": 617, "lr": 0.001} +{"train_loss": 0.029366053640842438, "train_loss_bc": 0.029366053640842438, "train_loss_llm": 0.0, "grad_norm": 0.07356175035238266, "global_step": 22224, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03327105566859245, "train_loss_bc": 0.03327105566859245, "train_loss_llm": 0.0, "grad_norm": 0.031080452725291252, "global_step": 22225, "epoch": 617, "lr": 0.001} +{"train_loss": 0.036002691835165024, "train_loss_bc": 0.036002691835165024, "train_loss_llm": 0.0, "grad_norm": 0.06073129549622536, "global_step": 22226, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03492536023259163, "train_loss_bc": 0.03492536023259163, "train_loss_llm": 0.0, "grad_norm": 0.081161268055439, "global_step": 22227, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03738649934530258, "train_loss_bc": 0.03738649934530258, "train_loss_llm": 0.0, "grad_norm": 0.0999542623758316, "global_step": 22228, "epoch": 617, "lr": 0.001} +{"train_loss": 0.030827969312667847, "train_loss_bc": 0.030827969312667847, "train_loss_llm": 0.0, "grad_norm": 0.034706342965364456, "global_step": 22229, "epoch": 617, "lr": 0.001} +{"train_loss": 0.02804727293550968, "train_loss_bc": 0.02804727293550968, "train_loss_llm": 0.0, "grad_norm": 0.050183799117803574, "global_step": 22230, "epoch": 617, "lr": 0.001} +{"train_loss": 0.030284198001027107, "train_loss_bc": 0.030283614993095398, "train_loss_llm": 5.829334259033203e-05, "grad_norm": 0.07230363041162491, "global_step": 22231, "epoch": 617, "lr": 0.001} +{"train_loss": 0.028738167136907578, "train_loss_bc": 0.028738167136907578, "train_loss_llm": 0.0, "grad_norm": 0.08391252905130386, "global_step": 22232, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03797076642513275, "train_loss_bc": 0.037963684648275375, "train_loss_llm": 0.000708162784576416, "grad_norm": 0.046549875289201736, "global_step": 22233, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03234501928091049, "train_loss_bc": 0.03234501928091049, "train_loss_llm": 0.0, "grad_norm": 0.06056655943393707, "global_step": 22234, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03627030551433563, "train_loss_bc": 0.03627030551433563, "train_loss_llm": 0.0, "grad_norm": 0.08779029548168182, "global_step": 22235, "epoch": 617, "lr": 0.001} +{"train_loss": 0.02658689208328724, "train_loss_bc": 0.02658689208328724, "train_loss_llm": 0.0, "grad_norm": 0.10629210621118546, "global_step": 22236, "epoch": 617, "lr": 0.001} +{"train_loss": 0.0367569625377655, "train_loss_bc": 0.0367569625377655, "train_loss_llm": 0.0, "grad_norm": 0.03439728170633316, "global_step": 22237, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03627897426486015, "train_loss_bc": 0.03627897426486015, "train_loss_llm": 0.0, "grad_norm": 0.05309770628809929, "global_step": 22238, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03015485778450966, "train_loss_bc": 0.03015485778450966, "train_loss_llm": 0.0, "grad_norm": 0.06707175821065903, "global_step": 22239, "epoch": 617, "lr": 0.001} +{"train_loss": 0.035119690001010895, "train_loss_bc": 0.035119690001010895, "train_loss_llm": 0.0, "grad_norm": 0.08461786061525345, "global_step": 22240, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03237612545490265, "train_loss_bc": 0.03237207233905792, "train_loss_llm": 0.00040543079376220703, "grad_norm": 0.03373552858829498, "global_step": 22241, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03517918661236763, "train_loss_bc": 0.0351705439388752, "train_loss_llm": 0.0008640885353088379, "grad_norm": 0.05289444327354431, "global_step": 22242, "epoch": 617, "lr": 0.001} +{"train_loss": 0.029431264847517014, "train_loss_bc": 0.029429880902171135, "train_loss_llm": 0.00013843178749084473, "grad_norm": 0.062217675149440765, "global_step": 22243, "epoch": 617, "lr": 0.001} +{"train_loss": 0.030735591426491737, "train_loss_bc": 0.030729182064533234, "train_loss_llm": 0.0006410181522369385, "grad_norm": 0.06770654767751694, "global_step": 22244, "epoch": 617, "lr": 0.001} +{"train_loss": 0.03694099932909012, "train_loss_bc": 0.03691408410668373, "train_loss_llm": 0.0026915371417999268, "grad_norm": 0.03844233602285385, "global_step": 22245, "epoch": 617, "lr": 0.001} +{"train_loss": 0.035144973546266556, "train_loss_bc": 0.03514232859015465, "train_loss_llm": 0.0002645552158355713, "grad_norm": 0.058424100279808044, "global_step": 22246, "epoch": 617, "lr": 0.001} +{"train_loss": 0.032563754357397556, "train_loss_bc": 0.027054790407419205, "train_loss_llm": 0.0, "grad_norm": 0.07154840230941772, "global_step": 22247, "epoch": 617, "lr": 0.001, "val_loss": 0.08308937400579453} +{"train_loss": 0.03075530380010605, "train_loss_bc": 0.03075530380010605, "train_loss_llm": 0.0, "grad_norm": 0.08629705756902695, "global_step": 22248, "epoch": 618, "lr": 0.001} +{"train_loss": 0.031050819903612137, "train_loss_bc": 0.03103308193385601, "train_loss_llm": 0.001773744821548462, "grad_norm": 0.041600301861763, "global_step": 22249, "epoch": 618, "lr": 0.001} +{"train_loss": 0.035683490335941315, "train_loss_bc": 0.035656657069921494, "train_loss_llm": 0.0026833415031433105, "grad_norm": 0.06093291565775871, "global_step": 22250, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03348951414227486, "train_loss_bc": 0.03348951414227486, "train_loss_llm": 0.0, "grad_norm": 0.08891481906175613, "global_step": 22251, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03202740475535393, "train_loss_bc": 0.03201206400990486, "train_loss_llm": 0.001534193754196167, "grad_norm": 0.10312814265489578, "global_step": 22252, "epoch": 618, "lr": 0.001} +{"train_loss": 0.033009301871061325, "train_loss_bc": 0.033009301871061325, "train_loss_llm": 0.0, "grad_norm": 0.03237198293209076, "global_step": 22253, "epoch": 618, "lr": 0.001} +{"train_loss": 0.030126815661787987, "train_loss_bc": 0.030126815661787987, "train_loss_llm": 0.0, "grad_norm": 0.04253816232085228, "global_step": 22254, "epoch": 618, "lr": 0.001} +{"train_loss": 0.02818671241402626, "train_loss_bc": 0.02818671241402626, "train_loss_llm": 0.0, "grad_norm": 0.05360756441950798, "global_step": 22255, "epoch": 618, "lr": 0.001} +{"train_loss": 0.028605101630091667, "train_loss_bc": 0.028605101630091667, "train_loss_llm": 0.0, "grad_norm": 0.0703926756978035, "global_step": 22256, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03532697632908821, "train_loss_bc": 0.03532697632908821, "train_loss_llm": 0.0, "grad_norm": 0.02872306853532791, "global_step": 22257, "epoch": 618, "lr": 0.001} +{"train_loss": 0.02474323660135269, "train_loss_bc": 0.02474323660135269, "train_loss_llm": 0.0, "grad_norm": 0.04227457940578461, "global_step": 22258, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03329063951969147, "train_loss_bc": 0.033279817551374435, "train_loss_llm": 0.0010822117328643799, "grad_norm": 0.06066729873418808, "global_step": 22259, "epoch": 618, "lr": 0.001} +{"train_loss": 0.02636684849858284, "train_loss_bc": 0.02636684849858284, "train_loss_llm": 0.0, "grad_norm": 0.08320816606283188, "global_step": 22260, "epoch": 618, "lr": 0.001} +{"train_loss": 0.024315444752573967, "train_loss_bc": 0.024315444752573967, "train_loss_llm": 0.0, "grad_norm": 0.02773956209421158, "global_step": 22261, "epoch": 618, "lr": 0.001} +{"train_loss": 0.030494647100567818, "train_loss_bc": 0.030494647100567818, "train_loss_llm": 0.0, "grad_norm": 0.05107003077864647, "global_step": 22262, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03143565356731415, "train_loss_bc": 0.03143565356731415, "train_loss_llm": 0.0, "grad_norm": 0.07257471978664398, "global_step": 22263, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03106771968305111, "train_loss_bc": 0.031061965972185135, "train_loss_llm": 0.000575333833694458, "grad_norm": 0.08637072890996933, "global_step": 22264, "epoch": 618, "lr": 0.001} +{"train_loss": 0.032507505267858505, "train_loss_bc": 0.032486725598573685, "train_loss_llm": 0.0020779073238372803, "grad_norm": 0.02824411727488041, "global_step": 22265, "epoch": 618, "lr": 0.001} +{"train_loss": 0.032948799431324005, "train_loss_bc": 0.0329369492828846, "train_loss_llm": 0.001184999942779541, "grad_norm": 0.051087163388729095, "global_step": 22266, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03478071466088295, "train_loss_bc": 0.03478071466088295, "train_loss_llm": 0.0, "grad_norm": 0.05924307182431221, "global_step": 22267, "epoch": 618, "lr": 0.001} +{"train_loss": 0.0389280803501606, "train_loss_bc": 0.0389280803501606, "train_loss_llm": 0.0, "grad_norm": 0.07804340869188309, "global_step": 22268, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03642284497618675, "train_loss_bc": 0.036422230303287506, "train_loss_llm": 6.133317947387695e-05, "grad_norm": 0.037025172263383865, "global_step": 22269, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03554213047027588, "train_loss_bc": 0.03554213047027588, "train_loss_llm": 0.0, "grad_norm": 0.052835024893283844, "global_step": 22270, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03959815576672554, "train_loss_bc": 0.03959815576672554, "train_loss_llm": 0.0, "grad_norm": 0.059204719960689545, "global_step": 22271, "epoch": 618, "lr": 0.001} +{"train_loss": 0.027346093207597733, "train_loss_bc": 0.027346093207597733, "train_loss_llm": 0.0, "grad_norm": 0.07510709762573242, "global_step": 22272, "epoch": 618, "lr": 0.001} +{"train_loss": 0.030854010954499245, "train_loss_bc": 0.030854010954499245, "train_loss_llm": 0.0, "grad_norm": 0.028724070638418198, "global_step": 22273, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03698165714740753, "train_loss_bc": 0.03698165714740753, "train_loss_llm": 0.0, "grad_norm": 0.04977771267294884, "global_step": 22274, "epoch": 618, "lr": 0.001} +{"train_loss": 0.036017704755067825, "train_loss_bc": 0.036017704755067825, "train_loss_llm": 0.0, "grad_norm": 0.05105294659733772, "global_step": 22275, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03442808613181114, "train_loss_bc": 0.03442808613181114, "train_loss_llm": 0.0, "grad_norm": 0.06516891717910767, "global_step": 22276, "epoch": 618, "lr": 0.001} +{"train_loss": 0.034665241837501526, "train_loss_bc": 0.034665241837501526, "train_loss_llm": 0.0, "grad_norm": 0.03162054717540741, "global_step": 22277, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03304385766386986, "train_loss_bc": 0.03302692249417305, "train_loss_llm": 0.001693427562713623, "grad_norm": 0.04826262220740318, "global_step": 22278, "epoch": 618, "lr": 0.001} +{"train_loss": 0.031132826581597328, "train_loss_bc": 0.031132826581597328, "train_loss_llm": 0.0, "grad_norm": 0.06519773602485657, "global_step": 22279, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03196215257048607, "train_loss_bc": 0.03195291385054588, "train_loss_llm": 0.000923842191696167, "grad_norm": 0.0794716477394104, "global_step": 22280, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03480737283825874, "train_loss_bc": 0.03480737283825874, "train_loss_llm": 0.0, "grad_norm": 0.03496963903307915, "global_step": 22281, "epoch": 618, "lr": 0.001} +{"train_loss": 0.033147748559713364, "train_loss_bc": 0.033139459788799286, "train_loss_llm": 0.0008287727832794189, "grad_norm": 0.055247094482183456, "global_step": 22282, "epoch": 618, "lr": 0.001} +{"train_loss": 0.03257803924174772, "train_loss_bc": 0.03771879896521568, "train_loss_llm": 0.0, "grad_norm": 0.08322756737470627, "global_step": 22283, "epoch": 618, "lr": 0.001, "val_loss": 0.08331488072872162} +{"train_loss": 0.034840457141399384, "train_loss_bc": 0.034840457141399384, "train_loss_llm": 0.0, "grad_norm": 0.10261807590723038, "global_step": 22284, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03221970796585083, "train_loss_bc": 0.03221970796585083, "train_loss_llm": 0.0, "grad_norm": 0.027492234483361244, "global_step": 22285, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03775692358613014, "train_loss_bc": 0.03775692358613014, "train_loss_llm": 0.0, "grad_norm": 0.04941311851143837, "global_step": 22286, "epoch": 619, "lr": 0.001} +{"train_loss": 0.028709469363093376, "train_loss_bc": 0.028709469363093376, "train_loss_llm": 0.0, "grad_norm": 0.06209074705839157, "global_step": 22287, "epoch": 619, "lr": 0.001} +{"train_loss": 0.02726350910961628, "train_loss_bc": 0.02726350910961628, "train_loss_llm": 0.0, "grad_norm": 0.07706480473279953, "global_step": 22288, "epoch": 619, "lr": 0.001} +{"train_loss": 0.030400792136788368, "train_loss_bc": 0.030400792136788368, "train_loss_llm": 0.0, "grad_norm": 0.026372939348220825, "global_step": 22289, "epoch": 619, "lr": 0.001} +{"train_loss": 0.02858424186706543, "train_loss_bc": 0.02858424186706543, "train_loss_llm": 0.0, "grad_norm": 0.03688628971576691, "global_step": 22290, "epoch": 619, "lr": 0.001} +{"train_loss": 0.035865191370248795, "train_loss_bc": 0.035865191370248795, "train_loss_llm": 0.0, "grad_norm": 0.055650047957897186, "global_step": 22291, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03434263542294502, "train_loss_bc": 0.03434263542294502, "train_loss_llm": 0.0, "grad_norm": 0.06499435007572174, "global_step": 22292, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03267578408122063, "train_loss_bc": 0.03267578408122063, "train_loss_llm": 0.0, "grad_norm": 0.0304495207965374, "global_step": 22293, "epoch": 619, "lr": 0.001} +{"train_loss": 0.028106989338994026, "train_loss_bc": 0.028106989338994026, "train_loss_llm": 0.0, "grad_norm": 0.04556316137313843, "global_step": 22294, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03233368322253227, "train_loss_bc": 0.03233368322253227, "train_loss_llm": 0.0, "grad_norm": 0.06398944556713104, "global_step": 22295, "epoch": 619, "lr": 0.001} +{"train_loss": 0.035797279328107834, "train_loss_bc": 0.035797279328107834, "train_loss_llm": 0.0, "grad_norm": 0.07835497707128525, "global_step": 22296, "epoch": 619, "lr": 0.001} +{"train_loss": 0.027953483164310455, "train_loss_bc": 0.02794976346194744, "train_loss_llm": 0.0003719031810760498, "grad_norm": 0.027848485857248306, "global_step": 22297, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03201523423194885, "train_loss_bc": 0.03201523423194885, "train_loss_llm": 0.0, "grad_norm": 0.042714230716228485, "global_step": 22298, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03517792746424675, "train_loss_bc": 0.03517792746424675, "train_loss_llm": 0.0, "grad_norm": 0.06160059943795204, "global_step": 22299, "epoch": 619, "lr": 0.001} +{"train_loss": 0.033448848873376846, "train_loss_bc": 0.03344037011265755, "train_loss_llm": 0.0008480250835418701, "grad_norm": 0.07973127067089081, "global_step": 22300, "epoch": 619, "lr": 0.001} +{"train_loss": 0.033950239419937134, "train_loss_bc": 0.033950239419937134, "train_loss_llm": 0.0, "grad_norm": 0.03811553120613098, "global_step": 22301, "epoch": 619, "lr": 0.001} +{"train_loss": 0.034489408135414124, "train_loss_bc": 0.034489408135414124, "train_loss_llm": 0.0, "grad_norm": 0.046917613595724106, "global_step": 22302, "epoch": 619, "lr": 0.001} +{"train_loss": 0.028381705284118652, "train_loss_bc": 0.028381705284118652, "train_loss_llm": 0.0, "grad_norm": 0.06921228766441345, "global_step": 22303, "epoch": 619, "lr": 0.001} +{"train_loss": 0.032097648829221725, "train_loss_bc": 0.032097648829221725, "train_loss_llm": 0.0, "grad_norm": 0.08312032371759415, "global_step": 22304, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03762452304363251, "train_loss_bc": 0.03760829195380211, "train_loss_llm": 0.0016230642795562744, "grad_norm": 0.03873932361602783, "global_step": 22305, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03074369579553604, "train_loss_bc": 0.03074369579553604, "train_loss_llm": 0.0, "grad_norm": 0.05696948245167732, "global_step": 22306, "epoch": 619, "lr": 0.001} +{"train_loss": 0.027794573456048965, "train_loss_bc": 0.027790576219558716, "train_loss_llm": 0.00039967894554138184, "grad_norm": 0.08336944878101349, "global_step": 22307, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03057762421667576, "train_loss_bc": 0.03057762421667576, "train_loss_llm": 0.0, "grad_norm": 0.10316412895917892, "global_step": 22308, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03416235372424126, "train_loss_bc": 0.03416235372424126, "train_loss_llm": 0.0, "grad_norm": 0.03677196800708771, "global_step": 22309, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03641914948821068, "train_loss_bc": 0.03641914948821068, "train_loss_llm": 0.0, "grad_norm": 0.05981244146823883, "global_step": 22310, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03281501308083534, "train_loss_bc": 0.03281501308083534, "train_loss_llm": 0.0, "grad_norm": 0.07678157091140747, "global_step": 22311, "epoch": 619, "lr": 0.001} +{"train_loss": 0.030347269028425217, "train_loss_bc": 0.030344031751155853, "train_loss_llm": 0.0003237128257751465, "grad_norm": 0.0873725488781929, "global_step": 22312, "epoch": 619, "lr": 0.001} +{"train_loss": 0.0348912738263607, "train_loss_bc": 0.0348912738263607, "train_loss_llm": 0.0, "grad_norm": 0.030504528433084488, "global_step": 22313, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03372054174542427, "train_loss_bc": 0.03372054174542427, "train_loss_llm": 0.0, "grad_norm": 0.04528683051466942, "global_step": 22314, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03026123158633709, "train_loss_bc": 0.030248433351516724, "train_loss_llm": 0.0012798905372619629, "grad_norm": 0.05990255996584892, "global_step": 22315, "epoch": 619, "lr": 0.001} +{"train_loss": 0.03541019558906555, "train_loss_bc": 0.03540242835879326, "train_loss_llm": 0.0007767081260681152, "grad_norm": 0.07148437201976776, "global_step": 22316, "epoch": 619, "lr": 0.001} +{"train_loss": 0.035689130425453186, "train_loss_bc": 0.035689130425453186, "train_loss_llm": 0.0, "grad_norm": 0.039673611521720886, "global_step": 22317, "epoch": 619, "lr": 0.001} +{"train_loss": 0.032102975994348526, "train_loss_bc": 0.032071150839328766, "train_loss_llm": 0.003182649612426758, "grad_norm": 0.05749969929456711, "global_step": 22318, "epoch": 619, "lr": 0.001} +{"train_loss": 0.032596764568653375, "train_loss_bc": 0.034475233405828476, "train_loss_llm": 0.003758072853088379, "grad_norm": 0.07239121943712234, "global_step": 22319, "epoch": 619, "lr": 0.001, "val_loss": 0.0836082473397255} +{"train_loss": 0.034128718078136444, "train_loss_bc": 0.034128718078136444, "train_loss_llm": 0.0, "grad_norm": 0.08533651381731033, "global_step": 22320, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03338916227221489, "train_loss_bc": 0.03338916227221489, "train_loss_llm": 0.0, "grad_norm": 0.02562114968895912, "global_step": 22321, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03529289737343788, "train_loss_bc": 0.03528275713324547, "train_loss_llm": 0.0010139942169189453, "grad_norm": 0.046629901975393295, "global_step": 22322, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03183232247829437, "train_loss_bc": 0.03182454779744148, "train_loss_llm": 0.0007774233818054199, "grad_norm": 0.05381752923130989, "global_step": 22323, "epoch": 620, "lr": 0.001} +{"train_loss": 0.027896776795387268, "train_loss_bc": 0.027896776795387268, "train_loss_llm": 0.0, "grad_norm": 0.07038529962301254, "global_step": 22324, "epoch": 620, "lr": 0.001} +{"train_loss": 0.030445637181401253, "train_loss_bc": 0.030445637181401253, "train_loss_llm": 0.0, "grad_norm": 0.029594723135232925, "global_step": 22325, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03541705012321472, "train_loss_bc": 0.03541705012321472, "train_loss_llm": 0.0, "grad_norm": 0.0502709299325943, "global_step": 22326, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03706377372145653, "train_loss_bc": 0.03706377372145653, "train_loss_llm": 0.0, "grad_norm": 0.06842191517353058, "global_step": 22327, "epoch": 620, "lr": 0.001} +{"train_loss": 0.041389450430870056, "train_loss_bc": 0.041389450430870056, "train_loss_llm": 0.0, "grad_norm": 0.0854642242193222, "global_step": 22328, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03879518061876297, "train_loss_bc": 0.038765911012887955, "train_loss_llm": 0.002927124500274658, "grad_norm": 0.03749047592282295, "global_step": 22329, "epoch": 620, "lr": 0.001} +{"train_loss": 0.02906382642686367, "train_loss_bc": 0.02906382642686367, "train_loss_llm": 0.0, "grad_norm": 0.05380018427968025, "global_step": 22330, "epoch": 620, "lr": 0.001} +{"train_loss": 0.030884288251399994, "train_loss_bc": 0.030884288251399994, "train_loss_llm": 0.0, "grad_norm": 0.07145887613296509, "global_step": 22331, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03235643729567528, "train_loss_bc": 0.03235643729567528, "train_loss_llm": 0.0, "grad_norm": 0.09791739284992218, "global_step": 22332, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03342604264616966, "train_loss_bc": 0.03342604264616966, "train_loss_llm": 0.0, "grad_norm": 0.04295196384191513, "global_step": 22333, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03806828707456589, "train_loss_bc": 0.03806828707456589, "train_loss_llm": 0.0, "grad_norm": 0.06737116724252701, "global_step": 22334, "epoch": 620, "lr": 0.001} +{"train_loss": 0.02749377302825451, "train_loss_bc": 0.02749377302825451, "train_loss_llm": 0.0, "grad_norm": 0.08596478402614594, "global_step": 22335, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03167068958282471, "train_loss_bc": 0.03166091442108154, "train_loss_llm": 0.0009776949882507324, "grad_norm": 0.10133405774831772, "global_step": 22336, "epoch": 620, "lr": 0.001} +{"train_loss": 0.030125293880701065, "train_loss_bc": 0.030125293880701065, "train_loss_llm": 0.0, "grad_norm": 0.028103871271014214, "global_step": 22337, "epoch": 620, "lr": 0.001} +{"train_loss": 0.02834061160683632, "train_loss_bc": 0.02834061160683632, "train_loss_llm": 0.0, "grad_norm": 0.045965585857629776, "global_step": 22338, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03671552613377571, "train_loss_bc": 0.03671552613377571, "train_loss_llm": 0.0, "grad_norm": 0.07058080285787582, "global_step": 22339, "epoch": 620, "lr": 0.001} +{"train_loss": 0.0333368293941021, "train_loss_bc": 0.0333368293941021, "train_loss_llm": 0.0, "grad_norm": 0.07694797217845917, "global_step": 22340, "epoch": 620, "lr": 0.001} +{"train_loss": 0.032534413039684296, "train_loss_bc": 0.032534413039684296, "train_loss_llm": 0.0, "grad_norm": 0.02600398287177086, "global_step": 22341, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03126714378595352, "train_loss_bc": 0.03126714378595352, "train_loss_llm": 0.0, "grad_norm": 0.04021758586168289, "global_step": 22342, "epoch": 620, "lr": 0.001} +{"train_loss": 0.028669510036706924, "train_loss_bc": 0.028663156554102898, "train_loss_llm": 0.0006353557109832764, "grad_norm": 0.05123916640877724, "global_step": 22343, "epoch": 620, "lr": 0.001} +{"train_loss": 0.02777121216058731, "train_loss_bc": 0.02777121216058731, "train_loss_llm": 0.0, "grad_norm": 0.06198381632566452, "global_step": 22344, "epoch": 620, "lr": 0.001} +{"train_loss": 0.034148190170526505, "train_loss_bc": 0.034148190170526505, "train_loss_llm": 0.0, "grad_norm": 0.033253587782382965, "global_step": 22345, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03356117382645607, "train_loss_bc": 0.03355719521641731, "train_loss_llm": 0.00039768218994140625, "grad_norm": 0.056785136461257935, "global_step": 22346, "epoch": 620, "lr": 0.001} +{"train_loss": 0.032636143267154694, "train_loss_bc": 0.032636143267154694, "train_loss_llm": 0.0, "grad_norm": 0.06552980095148087, "global_step": 22347, "epoch": 620, "lr": 0.001} +{"train_loss": 0.02745288982987404, "train_loss_bc": 0.02745288982987404, "train_loss_llm": 0.0, "grad_norm": 0.08113439381122589, "global_step": 22348, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03355547785758972, "train_loss_bc": 0.03355547785758972, "train_loss_llm": 0.0, "grad_norm": 0.03788195177912712, "global_step": 22349, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03516911342740059, "train_loss_bc": 0.03516911342740059, "train_loss_llm": 0.0, "grad_norm": 0.056507911533117294, "global_step": 22350, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03231169655919075, "train_loss_bc": 0.03231169655919075, "train_loss_llm": 0.0, "grad_norm": 0.07718997448682785, "global_step": 22351, "epoch": 620, "lr": 0.001} +{"train_loss": 0.033728327602148056, "train_loss_bc": 0.033728327602148056, "train_loss_llm": 0.0, "grad_norm": 0.09027977287769318, "global_step": 22352, "epoch": 620, "lr": 0.001} +{"train_loss": 0.031108297407627106, "train_loss_bc": 0.031108297407627106, "train_loss_llm": 0.0, "grad_norm": 0.02793600969016552, "global_step": 22353, "epoch": 620, "lr": 0.001} +{"train_loss": 0.02688712440431118, "train_loss_bc": 0.02688712440431118, "train_loss_llm": 0.0, "grad_norm": 0.04952166974544525, "global_step": 22354, "epoch": 620, "lr": 0.001} +{"train_loss": 0.03253154642879963, "train_loss_bc": 0.033202383667230606, "train_loss_llm": 0.0, "grad_norm": 0.07279644906520844, "global_step": 22355, "epoch": 620, "lr": 0.001, "train/cumulative_reward": 35.989870351301015, "train/mean_score": 21.74416226024477, "train/success_rate": 0.7, "test/cumulative_reward": 32.77364969304686, "test/mean_score": 7.792009232467604, "test/success_rate": 0.18, "val_loss": 0.0838853120803833, "train_action_mse_error": 0.036159809678792953} +{"train_loss": 0.0342676155269146, "train_loss_bc": 0.0342676155269146, "train_loss_llm": 0.0, "grad_norm": 0.08617746829986572, "global_step": 22356, "epoch": 621, "lr": 0.001} +{"train_loss": 0.037356507033109665, "train_loss_bc": 0.037356507033109665, "train_loss_llm": 0.0, "grad_norm": 0.038130711764097214, "global_step": 22357, "epoch": 621, "lr": 0.001} +{"train_loss": 0.02409394644200802, "train_loss_bc": 0.02409394644200802, "train_loss_llm": 0.0, "grad_norm": 0.052981141954660416, "global_step": 22358, "epoch": 621, "lr": 0.001} +{"train_loss": 0.03697682172060013, "train_loss_bc": 0.036953944712877274, "train_loss_llm": 0.00228765606880188, "grad_norm": 0.07973912358283997, "global_step": 22359, "epoch": 621, "lr": 0.001} +{"train_loss": 0.031233040615916252, "train_loss_bc": 0.031223716214299202, "train_loss_llm": 0.0009325146675109863, "grad_norm": 0.09680407494306564, "global_step": 22360, "epoch": 621, "lr": 0.001} +{"train_loss": 0.027051331475377083, "train_loss_bc": 0.02704768255352974, "train_loss_llm": 0.0003648996353149414, "grad_norm": 0.030875109136104584, "global_step": 22361, "epoch": 621, "lr": 0.001} +{"train_loss": 0.033675599843263626, "train_loss_bc": 0.03366704657673836, "train_loss_llm": 0.0008554756641387939, "grad_norm": 0.0480528399348259, "global_step": 22362, "epoch": 621, "lr": 0.001} +{"train_loss": 0.029185375198721886, "train_loss_bc": 0.029185375198721886, "train_loss_llm": 0.0, "grad_norm": 0.07055819034576416, "global_step": 22363, "epoch": 621, "lr": 0.001} +{"train_loss": 0.039663080126047134, "train_loss_bc": 0.039663080126047134, "train_loss_llm": 0.0, "grad_norm": 0.09121657907962799, "global_step": 22364, "epoch": 621, "lr": 0.001} +{"train_loss": 0.038106150925159454, "train_loss_bc": 0.038106150925159454, "train_loss_llm": 0.0, "grad_norm": 0.027657147496938705, "global_step": 22365, "epoch": 621, "lr": 0.001} +{"train_loss": 0.036500222980976105, "train_loss_bc": 0.036480460315942764, "train_loss_llm": 0.001976102590560913, "grad_norm": 0.05794328823685646, "global_step": 22366, "epoch": 621, "lr": 0.001} +{"train_loss": 0.036708638072013855, "train_loss_bc": 0.036708638072013855, "train_loss_llm": 0.0, "grad_norm": 0.062495991587638855, "global_step": 22367, "epoch": 621, "lr": 0.001} +{"train_loss": 0.030208975076675415, "train_loss_bc": 0.030208975076675415, "train_loss_llm": 0.0, "grad_norm": 0.07233477383852005, "global_step": 22368, "epoch": 621, "lr": 0.001} +{"train_loss": 0.029990628361701965, "train_loss_bc": 0.029990628361701965, "train_loss_llm": 0.0, "grad_norm": 0.03011898882687092, "global_step": 22369, "epoch": 621, "lr": 0.001} +{"train_loss": 0.03786030784249306, "train_loss_bc": 0.03786030784249306, "train_loss_llm": 0.0, "grad_norm": 0.04774272441864014, "global_step": 22370, "epoch": 621, "lr": 0.001} +{"train_loss": 0.028147030621767044, "train_loss_bc": 0.028147030621767044, "train_loss_llm": 0.0, "grad_norm": 0.06138353794813156, "global_step": 22371, "epoch": 621, "lr": 0.001} +{"train_loss": 0.033503517508506775, "train_loss_bc": 0.033503517508506775, "train_loss_llm": 0.0, "grad_norm": 0.07185641676187515, "global_step": 22372, "epoch": 621, "lr": 0.001} +{"train_loss": 0.033818673342466354, "train_loss_bc": 0.03381473571062088, "train_loss_llm": 0.0003936886787414551, "grad_norm": 0.02727319486439228, "global_step": 22373, "epoch": 621, "lr": 0.001} +{"train_loss": 0.036458902060985565, "train_loss_bc": 0.036458902060985565, "train_loss_llm": 0.0, "grad_norm": 0.056271743029356, "global_step": 22374, "epoch": 621, "lr": 0.001} +{"train_loss": 0.02928946167230606, "train_loss_bc": 0.02928946167230606, "train_loss_llm": 0.0, "grad_norm": 0.08133797347545624, "global_step": 22375, "epoch": 621, "lr": 0.001} +{"train_loss": 0.034026362001895905, "train_loss_bc": 0.034026362001895905, "train_loss_llm": 0.0, "grad_norm": 0.09261280298233032, "global_step": 22376, "epoch": 621, "lr": 0.001} +{"train_loss": 0.03216111660003662, "train_loss_bc": 0.03216111660003662, "train_loss_llm": 0.0, "grad_norm": 0.028648439794778824, "global_step": 22377, "epoch": 621, "lr": 0.001} +{"train_loss": 0.029303349554538727, "train_loss_bc": 0.029303349554538727, "train_loss_llm": 0.0, "grad_norm": 0.04668472707271576, "global_step": 22378, "epoch": 621, "lr": 0.001} +{"train_loss": 0.03157360851764679, "train_loss_bc": 0.03155731037259102, "train_loss_llm": 0.0016298890113830566, "grad_norm": 0.06199636310338974, "global_step": 22379, "epoch": 621, "lr": 0.001} +{"train_loss": 0.034415680915117264, "train_loss_bc": 0.034415680915117264, "train_loss_llm": 0.0, "grad_norm": 0.08048293739557266, "global_step": 22380, "epoch": 621, "lr": 0.001} +{"train_loss": 0.032234400510787964, "train_loss_bc": 0.032234400510787964, "train_loss_llm": 0.0, "grad_norm": 0.03078155405819416, "global_step": 22381, "epoch": 621, "lr": 0.001} +{"train_loss": 0.030956145375967026, "train_loss_bc": 0.030935952439904213, "train_loss_llm": 0.002019345760345459, "grad_norm": 0.04475073516368866, "global_step": 22382, "epoch": 621, "lr": 0.001} +{"train_loss": 0.031215745955705643, "train_loss_bc": 0.031215745955705643, "train_loss_llm": 0.0, "grad_norm": 0.0669708102941513, "global_step": 22383, "epoch": 621, "lr": 0.001} +{"train_loss": 0.03273642808198929, "train_loss_bc": 0.03273642808198929, "train_loss_llm": 0.0, "grad_norm": 0.08376804739236832, "global_step": 22384, "epoch": 621, "lr": 0.001} +{"train_loss": 0.02356998436152935, "train_loss_bc": 0.02356998436152935, "train_loss_llm": 0.0, "grad_norm": 0.03159305825829506, "global_step": 22385, "epoch": 621, "lr": 0.001} +{"train_loss": 0.03201720491051674, "train_loss_bc": 0.032016903162002563, "train_loss_llm": 3.0249357223510742e-05, "grad_norm": 0.05209815502166748, "global_step": 22386, "epoch": 621, "lr": 0.001} +{"train_loss": 0.031138507649302483, "train_loss_bc": 0.031138507649302483, "train_loss_llm": 0.0, "grad_norm": 0.06629542261362076, "global_step": 22387, "epoch": 621, "lr": 0.001} +{"train_loss": 0.033895548433065414, "train_loss_bc": 0.03385753557085991, "train_loss_llm": 0.0038012266159057617, "grad_norm": 0.07700511068105698, "global_step": 22388, "epoch": 621, "lr": 0.001} +{"train_loss": 0.03327741101384163, "train_loss_bc": 0.03327741101384163, "train_loss_llm": 0.0, "grad_norm": 0.034840065985918045, "global_step": 22389, "epoch": 621, "lr": 0.001} +{"train_loss": 0.030488567426800728, "train_loss_bc": 0.030488567426800728, "train_loss_llm": 0.0, "grad_norm": 0.05193376913666725, "global_step": 22390, "epoch": 621, "lr": 0.001} +{"train_loss": 0.032510897092935115, "train_loss_bc": 0.033286407589912415, "train_loss_llm": 0.0, "grad_norm": 0.0787343978881836, "global_step": 22391, "epoch": 621, "lr": 0.001, "val_loss": 0.0830475389957428} +{"train_loss": 0.03565556928515434, "train_loss_bc": 0.03565556928515434, "train_loss_llm": 0.0, "grad_norm": 0.08683926612138748, "global_step": 22392, "epoch": 622, "lr": 0.001} +{"train_loss": 0.035132165998220444, "train_loss_bc": 0.035132165998220444, "train_loss_llm": 0.0, "grad_norm": 0.03133506327867508, "global_step": 22393, "epoch": 622, "lr": 0.001} +{"train_loss": 0.033112600445747375, "train_loss_bc": 0.033112600445747375, "train_loss_llm": 0.0, "grad_norm": 0.06365008652210236, "global_step": 22394, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03271471709012985, "train_loss_bc": 0.03271471709012985, "train_loss_llm": 0.0, "grad_norm": 0.06802926957607269, "global_step": 22395, "epoch": 622, "lr": 0.001} +{"train_loss": 0.033879734575748444, "train_loss_bc": 0.033879734575748444, "train_loss_llm": 0.0, "grad_norm": 0.08396077156066895, "global_step": 22396, "epoch": 622, "lr": 0.001} +{"train_loss": 0.024121921509504318, "train_loss_bc": 0.024119438603520393, "train_loss_llm": 0.00024831295013427734, "grad_norm": 0.026265159249305725, "global_step": 22397, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03404688835144043, "train_loss_bc": 0.03401288390159607, "train_loss_llm": 0.0034006237983703613, "grad_norm": 0.03902752697467804, "global_step": 22398, "epoch": 622, "lr": 0.001} +{"train_loss": 0.030990619212388992, "train_loss_bc": 0.030980035662651062, "train_loss_llm": 0.0010583996772766113, "grad_norm": 0.06498883664608002, "global_step": 22399, "epoch": 622, "lr": 0.001} +{"train_loss": 0.034450992941856384, "train_loss_bc": 0.034450992941856384, "train_loss_llm": 0.0, "grad_norm": 0.07451319694519043, "global_step": 22400, "epoch": 622, "lr": 0.001} +{"train_loss": 0.02865133062005043, "train_loss_bc": 0.02865133062005043, "train_loss_llm": 0.0, "grad_norm": 0.027102090418338776, "global_step": 22401, "epoch": 622, "lr": 0.001} +{"train_loss": 0.027463799342513084, "train_loss_bc": 0.027463799342513084, "train_loss_llm": 0.0, "grad_norm": 0.04068372771143913, "global_step": 22402, "epoch": 622, "lr": 0.001} +{"train_loss": 0.035901330411434174, "train_loss_bc": 0.035889290273189545, "train_loss_llm": 0.0012040436267852783, "grad_norm": 0.06165892630815506, "global_step": 22403, "epoch": 622, "lr": 0.001} +{"train_loss": 0.031019339337944984, "train_loss_bc": 0.031008385121822357, "train_loss_llm": 0.0010953843593597412, "grad_norm": 0.08105259388685226, "global_step": 22404, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03615408390760422, "train_loss_bc": 0.03615408390760422, "train_loss_llm": 0.0, "grad_norm": 0.03662394359707832, "global_step": 22405, "epoch": 622, "lr": 0.001} +{"train_loss": 0.031385257840156555, "train_loss_bc": 0.03137975558638573, "train_loss_llm": 0.0005501508712768555, "grad_norm": 0.060318369418382645, "global_step": 22406, "epoch": 622, "lr": 0.001} +{"train_loss": 0.030752910301089287, "train_loss_bc": 0.030752910301089287, "train_loss_llm": 0.0, "grad_norm": 0.07216785103082657, "global_step": 22407, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03317291662096977, "train_loss_bc": 0.03317201882600784, "train_loss_llm": 8.979439735412598e-05, "grad_norm": 0.08375760912895203, "global_step": 22408, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03786958381533623, "train_loss_bc": 0.03786958381533623, "train_loss_llm": 0.0, "grad_norm": 0.032549161463975906, "global_step": 22409, "epoch": 622, "lr": 0.001} +{"train_loss": 0.028983114287257195, "train_loss_bc": 0.028976788744330406, "train_loss_llm": 0.0006324946880340576, "grad_norm": 0.045069243758916855, "global_step": 22410, "epoch": 622, "lr": 0.001} +{"train_loss": 0.036223579198122025, "train_loss_bc": 0.036223579198122025, "train_loss_llm": 0.0, "grad_norm": 0.0647398829460144, "global_step": 22411, "epoch": 622, "lr": 0.001} +{"train_loss": 0.026775436475872993, "train_loss_bc": 0.026775436475872993, "train_loss_llm": 0.0, "grad_norm": 0.07367609441280365, "global_step": 22412, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03534852713346481, "train_loss_bc": 0.03534852713346481, "train_loss_llm": 0.0, "grad_norm": 0.034060247242450714, "global_step": 22413, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03010222315788269, "train_loss_bc": 0.03010222315788269, "train_loss_llm": 0.0, "grad_norm": 0.05642346665263176, "global_step": 22414, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03355475515127182, "train_loss_bc": 0.03355475515127182, "train_loss_llm": 0.0, "grad_norm": 0.07700944691896439, "global_step": 22415, "epoch": 622, "lr": 0.001} +{"train_loss": 0.033431537449359894, "train_loss_bc": 0.033431537449359894, "train_loss_llm": 0.0, "grad_norm": 0.09938862919807434, "global_step": 22416, "epoch": 622, "lr": 0.001} +{"train_loss": 0.032752878963947296, "train_loss_bc": 0.032749827951192856, "train_loss_llm": 0.00030496716499328613, "grad_norm": 0.04085579141974449, "global_step": 22417, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03015669621527195, "train_loss_bc": 0.03015669621527195, "train_loss_llm": 0.0, "grad_norm": 0.06296118348836899, "global_step": 22418, "epoch": 622, "lr": 0.001} +{"train_loss": 0.0334191732108593, "train_loss_bc": 0.0334191732108593, "train_loss_llm": 0.0, "grad_norm": 0.08206978440284729, "global_step": 22419, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03118700347840786, "train_loss_bc": 0.03118700347840786, "train_loss_llm": 0.0, "grad_norm": 0.10069488734006882, "global_step": 22420, "epoch": 622, "lr": 0.001} +{"train_loss": 0.0348505824804306, "train_loss_bc": 0.0348505824804306, "train_loss_llm": 0.0, "grad_norm": 0.03466881439089775, "global_step": 22421, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03511461615562439, "train_loss_bc": 0.03509886562824249, "train_loss_llm": 0.0015749931335449219, "grad_norm": 0.06467931717634201, "global_step": 22422, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03409957140684128, "train_loss_bc": 0.03409957140684128, "train_loss_llm": 0.0, "grad_norm": 0.08191552758216858, "global_step": 22423, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03281435742974281, "train_loss_bc": 0.03280013054609299, "train_loss_llm": 0.0014226734638214111, "grad_norm": 0.09101012349128723, "global_step": 22424, "epoch": 622, "lr": 0.001} +{"train_loss": 0.03420986980199814, "train_loss_bc": 0.03420986980199814, "train_loss_llm": 0.0, "grad_norm": 0.032871995121240616, "global_step": 22425, "epoch": 622, "lr": 0.001} +{"train_loss": 0.031701039522886276, "train_loss_bc": 0.031701039522886276, "train_loss_llm": 0.0, "grad_norm": 0.050049424171447754, "global_step": 22426, "epoch": 622, "lr": 0.001} +{"train_loss": 0.032421034346852035, "train_loss_bc": 0.025956513360142708, "train_loss_llm": 0.0, "grad_norm": 0.0633433386683464, "global_step": 22427, "epoch": 622, "lr": 0.001, "val_loss": 0.08233897387981415} +{"train_loss": 0.03297758474946022, "train_loss_bc": 0.03297758474946022, "train_loss_llm": 0.0, "grad_norm": 0.07731079310178757, "global_step": 22428, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03391506150364876, "train_loss_bc": 0.03391506150364876, "train_loss_llm": 0.0, "grad_norm": 0.029789291322231293, "global_step": 22429, "epoch": 623, "lr": 0.001} +{"train_loss": 0.025930417701601982, "train_loss_bc": 0.025930417701601982, "train_loss_llm": 0.0, "grad_norm": 0.04686156287789345, "global_step": 22430, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03470511734485626, "train_loss_bc": 0.03470511734485626, "train_loss_llm": 0.0, "grad_norm": 0.059297531843185425, "global_step": 22431, "epoch": 623, "lr": 0.001} +{"train_loss": 0.038563717156648636, "train_loss_bc": 0.038563717156648636, "train_loss_llm": 0.0, "grad_norm": 0.07300891727209091, "global_step": 22432, "epoch": 623, "lr": 0.001} +{"train_loss": 0.022754566743969917, "train_loss_bc": 0.022754566743969917, "train_loss_llm": 0.0, "grad_norm": 0.021989794448018074, "global_step": 22433, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03227602690458298, "train_loss_bc": 0.03227602690458298, "train_loss_llm": 0.0, "grad_norm": 0.040062662214040756, "global_step": 22434, "epoch": 623, "lr": 0.001} +{"train_loss": 0.026129908859729767, "train_loss_bc": 0.02612607553601265, "train_loss_llm": 0.0003833174705505371, "grad_norm": 0.049371782690286636, "global_step": 22435, "epoch": 623, "lr": 0.001} +{"train_loss": 0.027311421930789948, "train_loss_bc": 0.027311421930789948, "train_loss_llm": 0.0, "grad_norm": 0.06676162779331207, "global_step": 22436, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03475392982363701, "train_loss_bc": 0.03475392982363701, "train_loss_llm": 0.0, "grad_norm": 0.033220067620277405, "global_step": 22437, "epoch": 623, "lr": 0.001} +{"train_loss": 0.032294727861881256, "train_loss_bc": 0.03227415680885315, "train_loss_llm": 0.0020571649074554443, "grad_norm": 0.054871831089258194, "global_step": 22438, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03467357158660889, "train_loss_bc": 0.03467357158660889, "train_loss_llm": 0.0, "grad_norm": 0.07637090981006622, "global_step": 22439, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03274504467844963, "train_loss_bc": 0.03274504467844963, "train_loss_llm": 0.0, "grad_norm": 0.09951886534690857, "global_step": 22440, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03697595372796059, "train_loss_bc": 0.03697595372796059, "train_loss_llm": 0.0, "grad_norm": 0.028519080951809883, "global_step": 22441, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03033248707652092, "train_loss_bc": 0.03033248707652092, "train_loss_llm": 0.0, "grad_norm": 0.05053827166557312, "global_step": 22442, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03385600447654724, "train_loss_bc": 0.033829957246780396, "train_loss_llm": 0.0026046037673950195, "grad_norm": 0.07524748891592026, "global_step": 22443, "epoch": 623, "lr": 0.001} +{"train_loss": 0.026897480711340904, "train_loss_bc": 0.026897480711340904, "train_loss_llm": 0.0, "grad_norm": 0.10312121361494064, "global_step": 22444, "epoch": 623, "lr": 0.001} +{"train_loss": 0.036349616944789886, "train_loss_bc": 0.036349616944789886, "train_loss_llm": 0.0, "grad_norm": 0.03764161840081215, "global_step": 22445, "epoch": 623, "lr": 0.001} +{"train_loss": 0.02908865362405777, "train_loss_bc": 0.02908865362405777, "train_loss_llm": 0.0, "grad_norm": 0.05227908492088318, "global_step": 22446, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03023657388985157, "train_loss_bc": 0.03022669069468975, "train_loss_llm": 0.0009882748126983643, "grad_norm": 0.06586132198572159, "global_step": 22447, "epoch": 623, "lr": 0.001} +{"train_loss": 0.02467694692313671, "train_loss_bc": 0.02467694692313671, "train_loss_llm": 0.0, "grad_norm": 0.07321527600288391, "global_step": 22448, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03021024912595749, "train_loss_bc": 0.030206099152565002, "train_loss_llm": 0.0004150569438934326, "grad_norm": 0.029541978612542152, "global_step": 22449, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03393183648586273, "train_loss_bc": 0.03393183648586273, "train_loss_llm": 0.0, "grad_norm": 0.04473533481359482, "global_step": 22450, "epoch": 623, "lr": 0.001} +{"train_loss": 0.037498392164707184, "train_loss_bc": 0.03748944029211998, "train_loss_llm": 0.000895082950592041, "grad_norm": 0.06093555688858032, "global_step": 22451, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03544272854924202, "train_loss_bc": 0.03544272854924202, "train_loss_llm": 0.0, "grad_norm": 0.07099022716283798, "global_step": 22452, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03551771491765976, "train_loss_bc": 0.03551448509097099, "train_loss_llm": 0.000322878360748291, "grad_norm": 0.027965115383267403, "global_step": 22453, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03639741986989975, "train_loss_bc": 0.03639741986989975, "train_loss_llm": 0.0, "grad_norm": 0.051419004797935486, "global_step": 22454, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03358055278658867, "train_loss_bc": 0.03358055278658867, "train_loss_llm": 0.0, "grad_norm": 0.07158607989549637, "global_step": 22455, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03435872867703438, "train_loss_bc": 0.03434513509273529, "train_loss_llm": 0.0013592839241027832, "grad_norm": 0.08907651156187057, "global_step": 22456, "epoch": 623, "lr": 0.001} +{"train_loss": 0.0336504802107811, "train_loss_bc": 0.033649101853370667, "train_loss_llm": 0.00013765692710876465, "grad_norm": 0.026782754808664322, "global_step": 22457, "epoch": 623, "lr": 0.001} +{"train_loss": 0.029849538579583168, "train_loss_bc": 0.02984509989619255, "train_loss_llm": 0.000443875789642334, "grad_norm": 0.05096236988902092, "global_step": 22458, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03240625560283661, "train_loss_bc": 0.03240625560283661, "train_loss_llm": 0.0, "grad_norm": 0.06371335685253143, "global_step": 22459, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03253619000315666, "train_loss_bc": 0.03253619000315666, "train_loss_llm": 0.0, "grad_norm": 0.07403786480426788, "global_step": 22460, "epoch": 623, "lr": 0.001} +{"train_loss": 0.037006884813308716, "train_loss_bc": 0.037006884813308716, "train_loss_llm": 0.0, "grad_norm": 0.038998883217573166, "global_step": 22461, "epoch": 623, "lr": 0.001} +{"train_loss": 0.032213740050792694, "train_loss_bc": 0.032213740050792694, "train_loss_llm": 0.0, "grad_norm": 0.05616751313209534, "global_step": 22462, "epoch": 623, "lr": 0.001} +{"train_loss": 0.03229549309859673, "train_loss_bc": 0.03059222549200058, "train_loss_llm": 0.0, "grad_norm": 0.063750721514225, "global_step": 22463, "epoch": 623, "lr": 0.001, "val_loss": 0.08388777822256088} +{"train_loss": 0.032421328127384186, "train_loss_bc": 0.032421328127384186, "train_loss_llm": 0.0, "grad_norm": 0.07708077877759933, "global_step": 22464, "epoch": 624, "lr": 0.001} +{"train_loss": 0.0351550430059433, "train_loss_bc": 0.0351550430059433, "train_loss_llm": 0.0, "grad_norm": 0.03588750213384628, "global_step": 22465, "epoch": 624, "lr": 0.001} +{"train_loss": 0.031420379877090454, "train_loss_bc": 0.031420379877090454, "train_loss_llm": 0.0, "grad_norm": 0.05918924883008003, "global_step": 22466, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03381714969873428, "train_loss_bc": 0.03381714969873428, "train_loss_llm": 0.0, "grad_norm": 0.0792880430817604, "global_step": 22467, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03370814397931099, "train_loss_bc": 0.03370814397931099, "train_loss_llm": 0.0, "grad_norm": 0.09586180001497269, "global_step": 22468, "epoch": 624, "lr": 0.001} +{"train_loss": 0.026584140956401825, "train_loss_bc": 0.026575889438390732, "train_loss_llm": 0.0008251368999481201, "grad_norm": 0.02571013756096363, "global_step": 22469, "epoch": 624, "lr": 0.001} +{"train_loss": 0.02768503874540329, "train_loss_bc": 0.027684418484568596, "train_loss_llm": 6.207823753356934e-05, "grad_norm": 0.039985448122024536, "global_step": 22470, "epoch": 624, "lr": 0.001} +{"train_loss": 0.029737522825598717, "train_loss_bc": 0.029737522825598717, "train_loss_llm": 0.0, "grad_norm": 0.04313274100422859, "global_step": 22471, "epoch": 624, "lr": 0.001} +{"train_loss": 0.029441028833389282, "train_loss_bc": 0.029441028833389282, "train_loss_llm": 0.0, "grad_norm": 0.057826969772577286, "global_step": 22472, "epoch": 624, "lr": 0.001} +{"train_loss": 0.0331093966960907, "train_loss_bc": 0.03310498967766762, "train_loss_llm": 0.00044077634811401367, "grad_norm": 0.033414777368307114, "global_step": 22473, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03574265167117119, "train_loss_bc": 0.03573901206254959, "train_loss_llm": 0.0003637969493865967, "grad_norm": 0.0549546480178833, "global_step": 22474, "epoch": 624, "lr": 0.001} +{"train_loss": 0.029857682064175606, "train_loss_bc": 0.0298474058508873, "train_loss_llm": 0.001027613878250122, "grad_norm": 0.06237218901515007, "global_step": 22475, "epoch": 624, "lr": 0.001} +{"train_loss": 0.035463739186525345, "train_loss_bc": 0.035463739186525345, "train_loss_llm": 0.0, "grad_norm": 0.0748869776725769, "global_step": 22476, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03109588846564293, "train_loss_bc": 0.03109588846564293, "train_loss_llm": 0.0, "grad_norm": 0.036305587738752365, "global_step": 22477, "epoch": 624, "lr": 0.001} +{"train_loss": 0.0352356843650341, "train_loss_bc": 0.03523459658026695, "train_loss_llm": 0.00010883808135986328, "grad_norm": 0.05186736583709717, "global_step": 22478, "epoch": 624, "lr": 0.001} +{"train_loss": 0.033952370285987854, "train_loss_bc": 0.03395089879631996, "train_loss_llm": 0.00014719367027282715, "grad_norm": 0.06478963792324066, "global_step": 22479, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03167347609996796, "train_loss_bc": 0.03167347609996796, "train_loss_llm": 0.0, "grad_norm": 0.09085498005151749, "global_step": 22480, "epoch": 624, "lr": 0.001} +{"train_loss": 0.028617270290851593, "train_loss_bc": 0.028617270290851593, "train_loss_llm": 0.0, "grad_norm": 0.026321763172745705, "global_step": 22481, "epoch": 624, "lr": 0.001} +{"train_loss": 0.029670529067516327, "train_loss_bc": 0.029666166752576828, "train_loss_llm": 0.0004362761974334717, "grad_norm": 0.045577287673950195, "global_step": 22482, "epoch": 624, "lr": 0.001} +{"train_loss": 0.026733405888080597, "train_loss_bc": 0.026726024225354195, "train_loss_llm": 0.0007382333278656006, "grad_norm": 0.057538580149412155, "global_step": 22483, "epoch": 624, "lr": 0.001} +{"train_loss": 0.039233338087797165, "train_loss_bc": 0.039233338087797165, "train_loss_llm": 0.0, "grad_norm": 0.08042217046022415, "global_step": 22484, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03173525258898735, "train_loss_bc": 0.031734783202409744, "train_loss_llm": 4.696846008300781e-05, "grad_norm": 0.026558352634310722, "global_step": 22485, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03350536897778511, "train_loss_bc": 0.03350536897778511, "train_loss_llm": 0.0, "grad_norm": 0.04228508472442627, "global_step": 22486, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03455951809883118, "train_loss_bc": 0.0345575213432312, "train_loss_llm": 0.00019952654838562012, "grad_norm": 0.051651615649461746, "global_step": 22487, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03758608177304268, "train_loss_bc": 0.03758608177304268, "train_loss_llm": 0.0, "grad_norm": 0.061837200075387955, "global_step": 22488, "epoch": 624, "lr": 0.001} +{"train_loss": 0.030903583392500877, "train_loss_bc": 0.030903583392500877, "train_loss_llm": 0.0, "grad_norm": 0.03643312677741051, "global_step": 22489, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03434496745467186, "train_loss_bc": 0.03434496745467186, "train_loss_llm": 0.0, "grad_norm": 0.05342236906290054, "global_step": 22490, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03714510053396225, "train_loss_bc": 0.03714510053396225, "train_loss_llm": 0.0, "grad_norm": 0.07320832461118698, "global_step": 22491, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03322689235210419, "train_loss_bc": 0.03322689235210419, "train_loss_llm": 0.0, "grad_norm": 0.0918891504406929, "global_step": 22492, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03255713731050491, "train_loss_bc": 0.03255713731050491, "train_loss_llm": 0.0, "grad_norm": 0.031317442655563354, "global_step": 22493, "epoch": 624, "lr": 0.001} +{"train_loss": 0.03022455982863903, "train_loss_bc": 0.03022455982863903, "train_loss_llm": 0.0, "grad_norm": 0.04747166857123375, "global_step": 22494, "epoch": 624, "lr": 0.001} +{"train_loss": 0.030507978051900864, "train_loss_bc": 0.030507978051900864, "train_loss_llm": 0.0, "grad_norm": 0.06111954525113106, "global_step": 22495, "epoch": 624, "lr": 0.001} +{"train_loss": 0.031077934429049492, "train_loss_bc": 0.031077934429049492, "train_loss_llm": 0.0, "grad_norm": 0.07485237717628479, "global_step": 22496, "epoch": 624, "lr": 0.001} +{"train_loss": 0.027170220389962196, "train_loss_bc": 0.027170220389962196, "train_loss_llm": 0.0, "grad_norm": 0.026048919185996056, "global_step": 22497, "epoch": 624, "lr": 0.001} +{"train_loss": 0.033541712909936905, "train_loss_bc": 0.033541712909936905, "train_loss_llm": 0.0, "grad_norm": 0.04608331248164177, "global_step": 22498, "epoch": 624, "lr": 0.001} +{"train_loss": 0.032324643081261054, "train_loss_bc": 0.0352449044585228, "train_loss_llm": 7.30752944946289e-05, "grad_norm": 0.06269857287406921, "global_step": 22499, "epoch": 624, "lr": 0.001, "val_loss": 0.08386673778295517} +{"train_loss": 0.029504545032978058, "train_loss_bc": 0.029498854652047157, "train_loss_llm": 0.0005689859390258789, "grad_norm": 0.07665906846523285, "global_step": 22500, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03126811236143112, "train_loss_bc": 0.03126811236143112, "train_loss_llm": 0.0, "grad_norm": 0.028882896527647972, "global_step": 22501, "epoch": 625, "lr": 0.001} +{"train_loss": 0.030177416279911995, "train_loss_bc": 0.030177416279911995, "train_loss_llm": 0.0, "grad_norm": 0.05683865770697594, "global_step": 22502, "epoch": 625, "lr": 0.001} +{"train_loss": 0.025164328515529633, "train_loss_bc": 0.025164328515529633, "train_loss_llm": 0.0, "grad_norm": 0.07215377688407898, "global_step": 22503, "epoch": 625, "lr": 0.001} +{"train_loss": 0.02884877473115921, "train_loss_bc": 0.02884877473115921, "train_loss_llm": 0.0, "grad_norm": 0.08882886171340942, "global_step": 22504, "epoch": 625, "lr": 0.001} +{"train_loss": 0.0284484401345253, "train_loss_bc": 0.0284484401345253, "train_loss_llm": 0.0, "grad_norm": 0.027725255116820335, "global_step": 22505, "epoch": 625, "lr": 0.001} +{"train_loss": 0.04098912328481674, "train_loss_bc": 0.04098912328481674, "train_loss_llm": 0.0, "grad_norm": 0.05398916080594063, "global_step": 22506, "epoch": 625, "lr": 0.001} +{"train_loss": 0.02990121580660343, "train_loss_bc": 0.02990121580660343, "train_loss_llm": 0.0, "grad_norm": 0.06761854887008667, "global_step": 22507, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03109133429825306, "train_loss_bc": 0.031070634722709656, "train_loss_llm": 0.0020698904991149902, "grad_norm": 0.08764881640672684, "global_step": 22508, "epoch": 625, "lr": 0.001} +{"train_loss": 0.04423858970403671, "train_loss_bc": 0.04423581808805466, "train_loss_llm": 0.0002771615982055664, "grad_norm": 0.05425908789038658, "global_step": 22509, "epoch": 625, "lr": 0.001} +{"train_loss": 0.034989096224308014, "train_loss_bc": 0.034989096224308014, "train_loss_llm": 0.0, "grad_norm": 0.07827859371900558, "global_step": 22510, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03205667808651924, "train_loss_bc": 0.03204558789730072, "train_loss_llm": 0.0011090338230133057, "grad_norm": 0.0913105234503746, "global_step": 22511, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03211382031440735, "train_loss_bc": 0.03210742026567459, "train_loss_llm": 0.0006398856639862061, "grad_norm": 0.09948799014091492, "global_step": 22512, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03781807795166969, "train_loss_bc": 0.03781807795166969, "train_loss_llm": 0.0, "grad_norm": 0.031998857855796814, "global_step": 22513, "epoch": 625, "lr": 0.001} +{"train_loss": 0.033626705408096313, "train_loss_bc": 0.03362107276916504, "train_loss_llm": 0.0005632936954498291, "grad_norm": 0.04680582135915756, "global_step": 22514, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03004138171672821, "train_loss_bc": 0.030037226155400276, "train_loss_llm": 0.00041550397872924805, "grad_norm": 0.06383264064788818, "global_step": 22515, "epoch": 625, "lr": 0.001} +{"train_loss": 0.029410865157842636, "train_loss_bc": 0.029403982684016228, "train_loss_llm": 0.0006883144378662109, "grad_norm": 0.08000721782445908, "global_step": 22516, "epoch": 625, "lr": 0.001} +{"train_loss": 0.028669258579611778, "train_loss_bc": 0.028669258579611778, "train_loss_llm": 0.0, "grad_norm": 0.02747529186308384, "global_step": 22517, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03237985447049141, "train_loss_bc": 0.03237985447049141, "train_loss_llm": 0.0, "grad_norm": 0.048447076231241226, "global_step": 22518, "epoch": 625, "lr": 0.001} +{"train_loss": 0.039424970746040344, "train_loss_bc": 0.039424970746040344, "train_loss_llm": 0.0, "grad_norm": 0.06460240483283997, "global_step": 22519, "epoch": 625, "lr": 0.001} +{"train_loss": 0.0326697938144207, "train_loss_bc": 0.0326697938144207, "train_loss_llm": 0.0, "grad_norm": 0.08397690951824188, "global_step": 22520, "epoch": 625, "lr": 0.001} +{"train_loss": 0.0341169610619545, "train_loss_bc": 0.0341169610619545, "train_loss_llm": 0.0, "grad_norm": 0.0354049876332283, "global_step": 22521, "epoch": 625, "lr": 0.001} +{"train_loss": 0.02978363260626793, "train_loss_bc": 0.02978363260626793, "train_loss_llm": 0.0, "grad_norm": 0.054246410727500916, "global_step": 22522, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03168593347072601, "train_loss_bc": 0.03168593347072601, "train_loss_llm": 0.0, "grad_norm": 0.08495352417230606, "global_step": 22523, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03403901681303978, "train_loss_bc": 0.03403901681303978, "train_loss_llm": 0.0, "grad_norm": 0.10671251267194748, "global_step": 22524, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03706821799278259, "train_loss_bc": 0.037061937153339386, "train_loss_llm": 0.000628054141998291, "grad_norm": 0.03933068364858627, "global_step": 22525, "epoch": 625, "lr": 0.001} +{"train_loss": 0.033912915736436844, "train_loss_bc": 0.033912915736436844, "train_loss_llm": 0.0, "grad_norm": 0.07060351967811584, "global_step": 22526, "epoch": 625, "lr": 0.001} +{"train_loss": 0.02815544418990612, "train_loss_bc": 0.02815162017941475, "train_loss_llm": 0.00038239359855651855, "grad_norm": 0.0886220633983612, "global_step": 22527, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03779938444495201, "train_loss_bc": 0.03779938444495201, "train_loss_llm": 0.0, "grad_norm": 0.10380742698907852, "global_step": 22528, "epoch": 625, "lr": 0.001} +{"train_loss": 0.030569756403565407, "train_loss_bc": 0.030569756403565407, "train_loss_llm": 0.0, "grad_norm": 0.028513571247458458, "global_step": 22529, "epoch": 625, "lr": 0.001} +{"train_loss": 0.029760265722870827, "train_loss_bc": 0.029759984463453293, "train_loss_llm": 2.816319465637207e-05, "grad_norm": 0.047952957451343536, "global_step": 22530, "epoch": 625, "lr": 0.001} +{"train_loss": 0.026028957217931747, "train_loss_bc": 0.026028957217931747, "train_loss_llm": 0.0, "grad_norm": 0.06295747309923172, "global_step": 22531, "epoch": 625, "lr": 0.001} +{"train_loss": 0.032168302685022354, "train_loss_bc": 0.032168302685022354, "train_loss_llm": 0.0, "grad_norm": 0.08186056464910507, "global_step": 22532, "epoch": 625, "lr": 0.001} +{"train_loss": 0.031767621636390686, "train_loss_bc": 0.03175046667456627, "train_loss_llm": 0.001715630292892456, "grad_norm": 0.03482445701956749, "global_step": 22533, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03176051005721092, "train_loss_bc": 0.03176051005721092, "train_loss_llm": 0.0, "grad_norm": 0.04632314667105675, "global_step": 22534, "epoch": 625, "lr": 0.001} +{"train_loss": 0.03250852356561356, "train_loss_bc": 0.03885754570364952, "train_loss_llm": 0.0, "grad_norm": 0.06198390945792198, "global_step": 22535, "epoch": 625, "lr": 0.001, "train/cumulative_reward": 65.53781040733436, "train/mean_score": 7.705696782033074, "train/success_rate": 0.1, "test/cumulative_reward": 49.17322092733558, "test/mean_score": 8.631625109948745, "test/success_rate": 0.22, "val_loss": 0.08290616422891617, "train_action_mse_error": 0.03541547805070877} +{"train_loss": 0.028725681826472282, "train_loss_bc": 0.028725681826472282, "train_loss_llm": 0.0, "grad_norm": 0.07183313369750977, "global_step": 22536, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03339056298136711, "train_loss_bc": 0.03339056298136711, "train_loss_llm": 0.0, "grad_norm": 0.028228310868144035, "global_step": 22537, "epoch": 626, "lr": 0.001} +{"train_loss": 0.035693343728780746, "train_loss_bc": 0.035693343728780746, "train_loss_llm": 0.0, "grad_norm": 0.050584375858306885, "global_step": 22538, "epoch": 626, "lr": 0.001} +{"train_loss": 0.035491641610860825, "train_loss_bc": 0.035491641610860825, "train_loss_llm": 0.0, "grad_norm": 0.058747440576553345, "global_step": 22539, "epoch": 626, "lr": 0.001} +{"train_loss": 0.026965931057929993, "train_loss_bc": 0.026965931057929993, "train_loss_llm": 0.0, "grad_norm": 0.07507170736789703, "global_step": 22540, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03082985244691372, "train_loss_bc": 0.03082985244691372, "train_loss_llm": 0.0, "grad_norm": 0.0294465571641922, "global_step": 22541, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03079576976597309, "train_loss_bc": 0.03079358860850334, "train_loss_llm": 0.00021818280220031738, "grad_norm": 0.04735128581523895, "global_step": 22542, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03260711580514908, "train_loss_bc": 0.03260711580514908, "train_loss_llm": 0.0, "grad_norm": 0.06741771847009659, "global_step": 22543, "epoch": 626, "lr": 0.001} +{"train_loss": 0.030564073473215103, "train_loss_bc": 0.030564073473215103, "train_loss_llm": 0.0, "grad_norm": 0.08356305211782455, "global_step": 22544, "epoch": 626, "lr": 0.001} +{"train_loss": 0.030993755906820297, "train_loss_bc": 0.030993755906820297, "train_loss_llm": 0.0, "grad_norm": 0.024009743705391884, "global_step": 22545, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03012014739215374, "train_loss_bc": 0.03012014739215374, "train_loss_llm": 0.0, "grad_norm": 0.04254233092069626, "global_step": 22546, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03295547515153885, "train_loss_bc": 0.03295547515153885, "train_loss_llm": 0.0, "grad_norm": 0.06161482632160187, "global_step": 22547, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03427109867334366, "train_loss_bc": 0.03426224738359451, "train_loss_llm": 0.0008850991725921631, "grad_norm": 0.08357217907905579, "global_step": 22548, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03047170490026474, "train_loss_bc": 0.03047170490026474, "train_loss_llm": 0.0, "grad_norm": 0.026201298460364342, "global_step": 22549, "epoch": 626, "lr": 0.001} +{"train_loss": 0.033000435680150986, "train_loss_bc": 0.033000435680150986, "train_loss_llm": 0.0, "grad_norm": 0.04384255409240723, "global_step": 22550, "epoch": 626, "lr": 0.001} +{"train_loss": 0.02949019894003868, "train_loss_bc": 0.02949019894003868, "train_loss_llm": 0.0, "grad_norm": 0.058511607348918915, "global_step": 22551, "epoch": 626, "lr": 0.001} +{"train_loss": 0.036191731691360474, "train_loss_bc": 0.03618813306093216, "train_loss_llm": 0.0003599226474761963, "grad_norm": 0.07432632148265839, "global_step": 22552, "epoch": 626, "lr": 0.001} +{"train_loss": 0.032959043979644775, "train_loss_bc": 0.03295678272843361, "train_loss_llm": 0.00022596120834350586, "grad_norm": 0.027655037119984627, "global_step": 22553, "epoch": 626, "lr": 0.001} +{"train_loss": 0.030120551586151123, "train_loss_bc": 0.030110390856862068, "train_loss_llm": 0.0010161399841308594, "grad_norm": 0.040929362177848816, "global_step": 22554, "epoch": 626, "lr": 0.001} +{"train_loss": 0.035357970744371414, "train_loss_bc": 0.03534742072224617, "train_loss_llm": 0.0010550916194915771, "grad_norm": 0.053345974534749985, "global_step": 22555, "epoch": 626, "lr": 0.001} +{"train_loss": 0.035000383853912354, "train_loss_bc": 0.03499205410480499, "train_loss_llm": 0.0008330643177032471, "grad_norm": 0.06500175595283508, "global_step": 22556, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03291020542383194, "train_loss_bc": 0.03291020542383194, "train_loss_llm": 0.0, "grad_norm": 0.034145794808864594, "global_step": 22557, "epoch": 626, "lr": 0.001} +{"train_loss": 0.031248871237039566, "train_loss_bc": 0.031244002282619476, "train_loss_llm": 0.000486910343170166, "grad_norm": 0.05425259470939636, "global_step": 22558, "epoch": 626, "lr": 0.001} +{"train_loss": 0.033830199390649796, "train_loss_bc": 0.033830199390649796, "train_loss_llm": 0.0, "grad_norm": 0.08365407586097717, "global_step": 22559, "epoch": 626, "lr": 0.001} +{"train_loss": 0.034584857523441315, "train_loss_bc": 0.034584857523441315, "train_loss_llm": 0.0, "grad_norm": 0.10000302642583847, "global_step": 22560, "epoch": 626, "lr": 0.001} +{"train_loss": 0.030824976041913033, "train_loss_bc": 0.030824976041913033, "train_loss_llm": 0.0, "grad_norm": 0.025675207376480103, "global_step": 22561, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03457396849989891, "train_loss_bc": 0.03457396849989891, "train_loss_llm": 0.0, "grad_norm": 0.04177245870232582, "global_step": 22562, "epoch": 626, "lr": 0.001} +{"train_loss": 0.032544318586587906, "train_loss_bc": 0.032544318586587906, "train_loss_llm": 0.0, "grad_norm": 0.0691877007484436, "global_step": 22563, "epoch": 626, "lr": 0.001} +{"train_loss": 0.02968381717801094, "train_loss_bc": 0.029660390689969063, "train_loss_llm": 0.002342700958251953, "grad_norm": 0.08507107943296432, "global_step": 22564, "epoch": 626, "lr": 0.001} +{"train_loss": 0.036829426884651184, "train_loss_bc": 0.036829426884651184, "train_loss_llm": 0.0, "grad_norm": 0.029569199308753014, "global_step": 22565, "epoch": 626, "lr": 0.001} +{"train_loss": 0.034160953015089035, "train_loss_bc": 0.034127093851566315, "train_loss_llm": 0.0033858120441436768, "grad_norm": 0.048300229012966156, "global_step": 22566, "epoch": 626, "lr": 0.001} +{"train_loss": 0.031820595264434814, "train_loss_bc": 0.03179382160305977, "train_loss_llm": 0.002677232027053833, "grad_norm": 0.06546550989151001, "global_step": 22567, "epoch": 626, "lr": 0.001} +{"train_loss": 0.027074819430708885, "train_loss_bc": 0.02706088311970234, "train_loss_llm": 0.0013936758041381836, "grad_norm": 0.0829181894659996, "global_step": 22568, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03441356495022774, "train_loss_bc": 0.03441356495022774, "train_loss_llm": 0.0, "grad_norm": 0.032662730664014816, "global_step": 22569, "epoch": 626, "lr": 0.001} +{"train_loss": 0.033861152827739716, "train_loss_bc": 0.033861152827739716, "train_loss_llm": 0.0, "grad_norm": 0.0638803020119667, "global_step": 22570, "epoch": 626, "lr": 0.001} +{"train_loss": 0.03218655795272854, "train_loss_bc": 0.024357888847589493, "train_loss_llm": 0.0, "grad_norm": 0.0803077444434166, "global_step": 22571, "epoch": 626, "lr": 0.001, "val_loss": 0.085110142827034} +{"train_loss": 0.035157203674316406, "train_loss_bc": 0.035157203674316406, "train_loss_llm": 0.0, "grad_norm": 0.10621796548366547, "global_step": 22572, "epoch": 627, "lr": 0.001} +{"train_loss": 0.031795039772987366, "train_loss_bc": 0.031795039772987366, "train_loss_llm": 0.0, "grad_norm": 0.03507547825574875, "global_step": 22573, "epoch": 627, "lr": 0.001} +{"train_loss": 0.030323946848511696, "train_loss_bc": 0.030308468267321587, "train_loss_llm": 0.0015478730201721191, "grad_norm": 0.06027505174279213, "global_step": 22574, "epoch": 627, "lr": 0.001} +{"train_loss": 0.028292188420891762, "train_loss_bc": 0.028291067108511925, "train_loss_llm": 0.00011205673217773438, "grad_norm": 0.08521903306245804, "global_step": 22575, "epoch": 627, "lr": 0.001} +{"train_loss": 0.032895468175411224, "train_loss_bc": 0.032889124006032944, "train_loss_llm": 0.0006345212459564209, "grad_norm": 0.10859427601099014, "global_step": 22576, "epoch": 627, "lr": 0.001} +{"train_loss": 0.028762206435203552, "train_loss_bc": 0.028749270364642143, "train_loss_llm": 0.001293569803237915, "grad_norm": 0.03408553823828697, "global_step": 22577, "epoch": 627, "lr": 0.001} +{"train_loss": 0.030896848067641258, "train_loss_bc": 0.030896848067641258, "train_loss_llm": 0.0, "grad_norm": 0.061294540762901306, "global_step": 22578, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03209344670176506, "train_loss_bc": 0.03209344670176506, "train_loss_llm": 0.0, "grad_norm": 0.0852404534816742, "global_step": 22579, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03037407249212265, "train_loss_bc": 0.03037407249212265, "train_loss_llm": 0.0, "grad_norm": 0.09778283536434174, "global_step": 22580, "epoch": 627, "lr": 0.001} +{"train_loss": 0.031701408326625824, "train_loss_bc": 0.031701408326625824, "train_loss_llm": 0.0, "grad_norm": 0.034469567239284515, "global_step": 22581, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03489776700735092, "train_loss_bc": 0.03486939147114754, "train_loss_llm": 0.002837449312210083, "grad_norm": 0.0630272626876831, "global_step": 22582, "epoch": 627, "lr": 0.001} +{"train_loss": 0.02735173888504505, "train_loss_bc": 0.02735173888504505, "train_loss_llm": 0.0, "grad_norm": 0.07637663185596466, "global_step": 22583, "epoch": 627, "lr": 0.001} +{"train_loss": 0.028412315994501114, "train_loss_bc": 0.028409255668520927, "train_loss_llm": 0.00030609965324401855, "grad_norm": 0.09641596674919128, "global_step": 22584, "epoch": 627, "lr": 0.001} +{"train_loss": 0.036466084420681, "train_loss_bc": 0.03646388277411461, "train_loss_llm": 0.00022009015083312988, "grad_norm": 0.036028940230607986, "global_step": 22585, "epoch": 627, "lr": 0.001} +{"train_loss": 0.029526039958000183, "train_loss_bc": 0.029526039958000183, "train_loss_llm": 0.0, "grad_norm": 0.05075448006391525, "global_step": 22586, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03608545660972595, "train_loss_bc": 0.03608545660972595, "train_loss_llm": 0.0, "grad_norm": 0.07718023657798767, "global_step": 22587, "epoch": 627, "lr": 0.001} +{"train_loss": 0.033301886171102524, "train_loss_bc": 0.033301886171102524, "train_loss_llm": 0.0, "grad_norm": 0.09661180526018143, "global_step": 22588, "epoch": 627, "lr": 0.001} +{"train_loss": 0.04299946129322052, "train_loss_bc": 0.04299946129322052, "train_loss_llm": 0.0, "grad_norm": 0.04536718875169754, "global_step": 22589, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03258577734231949, "train_loss_bc": 0.03258577734231949, "train_loss_llm": 0.0, "grad_norm": 0.07867993414402008, "global_step": 22590, "epoch": 627, "lr": 0.001} +{"train_loss": 0.030662760138511658, "train_loss_bc": 0.030662760138511658, "train_loss_llm": 0.0, "grad_norm": 0.08935398608446121, "global_step": 22591, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03973163291811943, "train_loss_bc": 0.03972754627466202, "train_loss_llm": 0.00040858983993530273, "grad_norm": 0.10910691320896149, "global_step": 22592, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03586475923657417, "train_loss_bc": 0.03586475923657417, "train_loss_llm": 0.0, "grad_norm": 0.04071574658155441, "global_step": 22593, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03506965935230255, "train_loss_bc": 0.03503483161330223, "train_loss_llm": 0.003482609987258911, "grad_norm": 0.06205621734261513, "global_step": 22594, "epoch": 627, "lr": 0.001} +{"train_loss": 0.031133856624364853, "train_loss_bc": 0.031133856624364853, "train_loss_llm": 0.0, "grad_norm": 0.08709876239299774, "global_step": 22595, "epoch": 627, "lr": 0.001} +{"train_loss": 0.02943587861955166, "train_loss_bc": 0.02943587861955166, "train_loss_llm": 0.0, "grad_norm": 0.10982083529233932, "global_step": 22596, "epoch": 627, "lr": 0.001} +{"train_loss": 0.037785567343235016, "train_loss_bc": 0.037785567343235016, "train_loss_llm": 0.0, "grad_norm": 0.028747981414198875, "global_step": 22597, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03538062050938606, "train_loss_bc": 0.035375356674194336, "train_loss_llm": 0.00052642822265625, "grad_norm": 0.051498208194971085, "global_step": 22598, "epoch": 627, "lr": 0.001} +{"train_loss": 0.02998618222773075, "train_loss_bc": 0.02998531423509121, "train_loss_llm": 8.684396743774414e-05, "grad_norm": 0.06959789991378784, "global_step": 22599, "epoch": 627, "lr": 0.001} +{"train_loss": 0.030282896012067795, "train_loss_bc": 0.030282896012067795, "train_loss_llm": 0.0, "grad_norm": 0.07740658521652222, "global_step": 22600, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03301924467086792, "train_loss_bc": 0.03299723565578461, "train_loss_llm": 0.002200901508331299, "grad_norm": 0.03459753841161728, "global_step": 22601, "epoch": 627, "lr": 0.001} +{"train_loss": 0.035893458873033524, "train_loss_bc": 0.035893458873033524, "train_loss_llm": 0.0, "grad_norm": 0.05818052589893341, "global_step": 22602, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03055432438850403, "train_loss_bc": 0.03055432438850403, "train_loss_llm": 0.0, "grad_norm": 0.07642737030982971, "global_step": 22603, "epoch": 627, "lr": 0.001} +{"train_loss": 0.029652919620275497, "train_loss_bc": 0.029652919620275497, "train_loss_llm": 0.0, "grad_norm": 0.09255808591842651, "global_step": 22604, "epoch": 627, "lr": 0.001} +{"train_loss": 0.02896270342171192, "train_loss_bc": 0.02893095090985298, "train_loss_llm": 0.0031752586364746094, "grad_norm": 0.033922869712114334, "global_step": 22605, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03425102308392525, "train_loss_bc": 0.03425102308392525, "train_loss_llm": 0.0, "grad_norm": 0.05644458532333374, "global_step": 22606, "epoch": 627, "lr": 0.001} +{"train_loss": 0.03259781836014655, "train_loss_bc": 0.031935617327690125, "train_loss_llm": 0.0, "grad_norm": 0.0920417383313179, "global_step": 22607, "epoch": 627, "lr": 0.001, "val_loss": 0.08488468825817108} +{"train_loss": 0.02959347888827324, "train_loss_bc": 0.029567930847406387, "train_loss_llm": 0.002554774284362793, "grad_norm": 0.11328330636024475, "global_step": 22608, "epoch": 628, "lr": 0.001} +{"train_loss": 0.037496935576200485, "train_loss_bc": 0.037496935576200485, "train_loss_llm": 0.0, "grad_norm": 0.04363565146923065, "global_step": 22609, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03755994886159897, "train_loss_bc": 0.03755994886159897, "train_loss_llm": 0.0, "grad_norm": 0.060553207993507385, "global_step": 22610, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03175514563918114, "train_loss_bc": 0.03175514563918114, "train_loss_llm": 0.0, "grad_norm": 0.08989735692739487, "global_step": 22611, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03204526752233505, "train_loss_bc": 0.03204526752233505, "train_loss_llm": 0.0, "grad_norm": 0.1160973459482193, "global_step": 22612, "epoch": 628, "lr": 0.001} +{"train_loss": 0.031804271042346954, "train_loss_bc": 0.03179207444190979, "train_loss_llm": 0.0012195110321044922, "grad_norm": 0.031190356239676476, "global_step": 22613, "epoch": 628, "lr": 0.001} +{"train_loss": 0.034663379192352295, "train_loss_bc": 0.03463990241289139, "train_loss_llm": 0.0023478269577026367, "grad_norm": 0.046637386083602905, "global_step": 22614, "epoch": 628, "lr": 0.001} +{"train_loss": 0.031736068427562714, "train_loss_bc": 0.031736068427562714, "train_loss_llm": 0.0, "grad_norm": 0.0602484866976738, "global_step": 22615, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03621574118733406, "train_loss_bc": 0.03621574118733406, "train_loss_llm": 0.0, "grad_norm": 0.07124695926904678, "global_step": 22616, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03343048691749573, "train_loss_bc": 0.03342508524656296, "train_loss_llm": 0.0005401372909545898, "grad_norm": 0.03133498877286911, "global_step": 22617, "epoch": 628, "lr": 0.001} +{"train_loss": 0.032680027186870575, "train_loss_bc": 0.032680027186870575, "train_loss_llm": 0.0, "grad_norm": 0.04769568890333176, "global_step": 22618, "epoch": 628, "lr": 0.001} +{"train_loss": 0.031536590307950974, "train_loss_bc": 0.031536590307950974, "train_loss_llm": 0.0, "grad_norm": 0.055947091430425644, "global_step": 22619, "epoch": 628, "lr": 0.001} +{"train_loss": 0.02883894182741642, "train_loss_bc": 0.02883894182741642, "train_loss_llm": 0.0, "grad_norm": 0.06693559139966965, "global_step": 22620, "epoch": 628, "lr": 0.001} +{"train_loss": 0.030194826424121857, "train_loss_bc": 0.030194826424121857, "train_loss_llm": 0.0, "grad_norm": 0.024937273934483528, "global_step": 22621, "epoch": 628, "lr": 0.001} +{"train_loss": 0.029362784698605537, "train_loss_bc": 0.029362784698605537, "train_loss_llm": 0.0, "grad_norm": 0.045452918857336044, "global_step": 22622, "epoch": 628, "lr": 0.001} +{"train_loss": 0.029437631368637085, "train_loss_bc": 0.029437631368637085, "train_loss_llm": 0.0, "grad_norm": 0.056286830455064774, "global_step": 22623, "epoch": 628, "lr": 0.001} +{"train_loss": 0.034045856446027756, "train_loss_bc": 0.034045856446027756, "train_loss_llm": 0.0, "grad_norm": 0.0730295330286026, "global_step": 22624, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03631693869829178, "train_loss_bc": 0.03631693869829178, "train_loss_llm": 0.0, "grad_norm": 0.05000051483511925, "global_step": 22625, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03078712709248066, "train_loss_bc": 0.03078712709248066, "train_loss_llm": 0.0, "grad_norm": 0.07645733654499054, "global_step": 22626, "epoch": 628, "lr": 0.001} +{"train_loss": 0.02722654864192009, "train_loss_bc": 0.02722654864192009, "train_loss_llm": 0.0, "grad_norm": 0.10987354069948196, "global_step": 22627, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03689103201031685, "train_loss_bc": 0.03688749670982361, "train_loss_llm": 0.0003534257411956787, "grad_norm": 0.14858366549015045, "global_step": 22628, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03821196034550667, "train_loss_bc": 0.03821196034550667, "train_loss_llm": 0.0, "grad_norm": 0.04176585376262665, "global_step": 22629, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03245571255683899, "train_loss_bc": 0.03245571255683899, "train_loss_llm": 0.0, "grad_norm": 0.06392741948366165, "global_step": 22630, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03704279288649559, "train_loss_bc": 0.03704279288649559, "train_loss_llm": 0.0, "grad_norm": 0.09573496878147125, "global_step": 22631, "epoch": 628, "lr": 0.001} +{"train_loss": 0.032117247581481934, "train_loss_bc": 0.032117247581481934, "train_loss_llm": 0.0, "grad_norm": 0.11662932485342026, "global_step": 22632, "epoch": 628, "lr": 0.001} +{"train_loss": 0.029482994228601456, "train_loss_bc": 0.029482994228601456, "train_loss_llm": 0.0, "grad_norm": 0.03038855269551277, "global_step": 22633, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03251930698752403, "train_loss_bc": 0.03251317888498306, "train_loss_llm": 0.0006129443645477295, "grad_norm": 0.04690020531415939, "global_step": 22634, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03032868728041649, "train_loss_bc": 0.03032868728041649, "train_loss_llm": 0.0, "grad_norm": 0.0565732978284359, "global_step": 22635, "epoch": 628, "lr": 0.001} +{"train_loss": 0.02961997501552105, "train_loss_bc": 0.029618363827466965, "train_loss_llm": 0.00016120076179504395, "grad_norm": 0.07460135966539383, "global_step": 22636, "epoch": 628, "lr": 0.001} +{"train_loss": 0.030890746042132378, "train_loss_bc": 0.030890746042132378, "train_loss_llm": 0.0, "grad_norm": 0.0284404493868351, "global_step": 22637, "epoch": 628, "lr": 0.001} +{"train_loss": 0.023605450987815857, "train_loss_bc": 0.023605450987815857, "train_loss_llm": 0.0, "grad_norm": 0.04773672670125961, "global_step": 22638, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03209633380174637, "train_loss_bc": 0.03209633380174637, "train_loss_llm": 0.0, "grad_norm": 0.06069079786539078, "global_step": 22639, "epoch": 628, "lr": 0.001} +{"train_loss": 0.033616214990615845, "train_loss_bc": 0.033604878932237625, "train_loss_llm": 0.0011335909366607666, "grad_norm": 0.0861373096704483, "global_step": 22640, "epoch": 628, "lr": 0.001} +{"train_loss": 0.038712311536073685, "train_loss_bc": 0.038712311536073685, "train_loss_llm": 0.0, "grad_norm": 0.042177800089120865, "global_step": 22641, "epoch": 628, "lr": 0.001} +{"train_loss": 0.037727680057287216, "train_loss_bc": 0.03772575408220291, "train_loss_llm": 0.00019273161888122559, "grad_norm": 0.08673303574323654, "global_step": 22642, "epoch": 628, "lr": 0.001} +{"train_loss": 0.03258132805219955, "train_loss_bc": 0.03086220473051071, "train_loss_llm": 0.0019163787364959717, "grad_norm": 0.11548461019992828, "global_step": 22643, "epoch": 628, "lr": 0.001, "val_loss": 0.08587167412042618} +{"train_loss": 0.04145360365509987, "train_loss_bc": 0.04145360365509987, "train_loss_llm": 0.0, "grad_norm": 0.15124739706516266, "global_step": 22644, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03447575867176056, "train_loss_bc": 0.03447575867176056, "train_loss_llm": 0.0, "grad_norm": 0.029848091304302216, "global_step": 22645, "epoch": 629, "lr": 0.001} +{"train_loss": 0.035500239580869675, "train_loss_bc": 0.035486429929733276, "train_loss_llm": 0.001381009817123413, "grad_norm": 0.06295593082904816, "global_step": 22646, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03545399382710457, "train_loss_bc": 0.03545399382710457, "train_loss_llm": 0.0, "grad_norm": 0.09477778524160385, "global_step": 22647, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03639887273311615, "train_loss_bc": 0.036396704614162445, "train_loss_llm": 0.0002167224884033203, "grad_norm": 0.11864995211362839, "global_step": 22648, "epoch": 629, "lr": 0.001} +{"train_loss": 0.02728455513715744, "train_loss_bc": 0.02728455513715744, "train_loss_llm": 0.0, "grad_norm": 0.03204527869820595, "global_step": 22649, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03542937710881233, "train_loss_bc": 0.035427022725343704, "train_loss_llm": 0.00023546814918518066, "grad_norm": 0.05680926889181137, "global_step": 22650, "epoch": 629, "lr": 0.001} +{"train_loss": 0.0308707095682621, "train_loss_bc": 0.0308707095682621, "train_loss_llm": 0.0, "grad_norm": 0.06689316034317017, "global_step": 22651, "epoch": 629, "lr": 0.001} +{"train_loss": 0.030003810301423073, "train_loss_bc": 0.03000284545123577, "train_loss_llm": 9.655952453613281e-05, "grad_norm": 0.08199457079172134, "global_step": 22652, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03269669786095619, "train_loss_bc": 0.03269638866186142, "train_loss_llm": 3.081560134887695e-05, "grad_norm": 0.030500179156661034, "global_step": 22653, "epoch": 629, "lr": 0.001} +{"train_loss": 0.031870000064373016, "train_loss_bc": 0.03186280280351639, "train_loss_llm": 0.0007198750972747803, "grad_norm": 0.05118599161505699, "global_step": 22654, "epoch": 629, "lr": 0.001} +{"train_loss": 0.028147678822278976, "train_loss_bc": 0.02814173884689808, "train_loss_llm": 0.0005940496921539307, "grad_norm": 0.07041038572788239, "global_step": 22655, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03236657381057739, "train_loss_bc": 0.03236657381057739, "train_loss_llm": 0.0, "grad_norm": 0.08086360991001129, "global_step": 22656, "epoch": 629, "lr": 0.001} +{"train_loss": 0.026371212676167488, "train_loss_bc": 0.026363639160990715, "train_loss_llm": 0.000757366418838501, "grad_norm": 0.027613040059804916, "global_step": 22657, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03211018070578575, "train_loss_bc": 0.03211018070578575, "train_loss_llm": 0.0, "grad_norm": 0.04855431616306305, "global_step": 22658, "epoch": 629, "lr": 0.001} +{"train_loss": 0.0350925475358963, "train_loss_bc": 0.0350925475358963, "train_loss_llm": 0.0, "grad_norm": 0.06527473777532578, "global_step": 22659, "epoch": 629, "lr": 0.001} +{"train_loss": 0.027539042755961418, "train_loss_bc": 0.02753637731075287, "train_loss_llm": 0.00026658177375793457, "grad_norm": 0.07170751690864563, "global_step": 22660, "epoch": 629, "lr": 0.001} +{"train_loss": 0.0338003970682621, "train_loss_bc": 0.0338003970682621, "train_loss_llm": 0.0, "grad_norm": 0.03736342862248421, "global_step": 22661, "epoch": 629, "lr": 0.001} +{"train_loss": 0.038026269525289536, "train_loss_bc": 0.038026269525289536, "train_loss_llm": 0.0, "grad_norm": 0.06576657295227051, "global_step": 22662, "epoch": 629, "lr": 0.001} +{"train_loss": 0.029622424393892288, "train_loss_bc": 0.029622424393892288, "train_loss_llm": 0.0, "grad_norm": 0.0908212959766388, "global_step": 22663, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03046433813869953, "train_loss_bc": 0.03046092391014099, "train_loss_llm": 0.0003415048122406006, "grad_norm": 0.1141834631562233, "global_step": 22664, "epoch": 629, "lr": 0.001} +{"train_loss": 0.030661555007100105, "train_loss_bc": 0.030661555007100105, "train_loss_llm": 0.0, "grad_norm": 0.032943084836006165, "global_step": 22665, "epoch": 629, "lr": 0.001} +{"train_loss": 0.0371960774064064, "train_loss_bc": 0.0371687188744545, "train_loss_llm": 0.002735823392868042, "grad_norm": 0.047294262796640396, "global_step": 22666, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03251989558339119, "train_loss_bc": 0.03251989558339119, "train_loss_llm": 0.0, "grad_norm": 0.07519696652889252, "global_step": 22667, "epoch": 629, "lr": 0.001} +{"train_loss": 0.04054867476224899, "train_loss_bc": 0.04054867476224899, "train_loss_llm": 0.0, "grad_norm": 0.11095724999904633, "global_step": 22668, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03455168008804321, "train_loss_bc": 0.03455168008804321, "train_loss_llm": 0.0, "grad_norm": 0.03136969357728958, "global_step": 22669, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03370232135057449, "train_loss_bc": 0.03370232135057449, "train_loss_llm": 0.0, "grad_norm": 0.05302590876817703, "global_step": 22670, "epoch": 629, "lr": 0.001} +{"train_loss": 0.031070539727807045, "train_loss_bc": 0.031070539727807045, "train_loss_llm": 0.0, "grad_norm": 0.08177545666694641, "global_step": 22671, "epoch": 629, "lr": 0.001} +{"train_loss": 0.030746417120099068, "train_loss_bc": 0.030746417120099068, "train_loss_llm": 0.0, "grad_norm": 0.09351398795843124, "global_step": 22672, "epoch": 629, "lr": 0.001} +{"train_loss": 0.017513738945126534, "train_loss_bc": 0.017510542646050453, "train_loss_llm": 0.0003196299076080322, "grad_norm": 0.02391396276652813, "global_step": 22673, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03223870322108269, "train_loss_bc": 0.03223870322108269, "train_loss_llm": 0.0, "grad_norm": 0.04369301348924637, "global_step": 22674, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03548227250576019, "train_loss_bc": 0.03548227250576019, "train_loss_llm": 0.0, "grad_norm": 0.06499553471803665, "global_step": 22675, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03195463493466377, "train_loss_bc": 0.03194376453757286, "train_loss_llm": 0.0010869204998016357, "grad_norm": 0.07947957515716553, "global_step": 22676, "epoch": 629, "lr": 0.001} +{"train_loss": 0.030773011967539787, "train_loss_bc": 0.030773011967539787, "train_loss_llm": 0.0, "grad_norm": 0.03063865751028061, "global_step": 22677, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03425827622413635, "train_loss_bc": 0.03425827622413635, "train_loss_llm": 0.0, "grad_norm": 0.04401030018925667, "global_step": 22678, "epoch": 629, "lr": 0.001} +{"train_loss": 0.03260564954123563, "train_loss_bc": 0.03560730069875717, "train_loss_llm": 0.0, "grad_norm": 0.06734922528266907, "global_step": 22679, "epoch": 629, "lr": 0.001, "val_loss": 0.08454027771949768} +{"train_loss": 0.02836911752820015, "train_loss_bc": 0.02836911752820015, "train_loss_llm": 0.0, "grad_norm": 0.08191703259944916, "global_step": 22680, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03143410012125969, "train_loss_bc": 0.03143410012125969, "train_loss_llm": 0.0, "grad_norm": 0.03594806790351868, "global_step": 22681, "epoch": 630, "lr": 0.001} +{"train_loss": 0.031989935785532, "train_loss_bc": 0.03197421133518219, "train_loss_llm": 0.0015723705291748047, "grad_norm": 0.05713571235537529, "global_step": 22682, "epoch": 630, "lr": 0.001} +{"train_loss": 0.02974874898791313, "train_loss_bc": 0.02974874898791313, "train_loss_llm": 0.0, "grad_norm": 0.08552508056163788, "global_step": 22683, "epoch": 630, "lr": 0.001} +{"train_loss": 0.026615096256136894, "train_loss_bc": 0.026615096256136894, "train_loss_llm": 0.0, "grad_norm": 0.09250510483980179, "global_step": 22684, "epoch": 630, "lr": 0.001} +{"train_loss": 0.02995876409113407, "train_loss_bc": 0.02995876409113407, "train_loss_llm": 0.0, "grad_norm": 0.03507189080119133, "global_step": 22685, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03305025026202202, "train_loss_bc": 0.03303148224949837, "train_loss_llm": 0.0018767118453979492, "grad_norm": 0.0637042298913002, "global_step": 22686, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03819165751338005, "train_loss_bc": 0.03818044811487198, "train_loss_llm": 0.0011208653450012207, "grad_norm": 0.07761717587709427, "global_step": 22687, "epoch": 630, "lr": 0.001} +{"train_loss": 0.028620680794119835, "train_loss_bc": 0.028619883581995964, "train_loss_llm": 7.981061935424805e-05, "grad_norm": 0.1119619607925415, "global_step": 22688, "epoch": 630, "lr": 0.001} +{"train_loss": 0.0308055616915226, "train_loss_bc": 0.030804447829723358, "train_loss_llm": 0.00011140108108520508, "grad_norm": 0.033506572246551514, "global_step": 22689, "epoch": 630, "lr": 0.001} +{"train_loss": 0.034399278461933136, "train_loss_bc": 0.034399278461933136, "train_loss_llm": 0.0, "grad_norm": 0.051950905472040176, "global_step": 22690, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03047007881104946, "train_loss_bc": 0.03047007881104946, "train_loss_llm": 0.0, "grad_norm": 0.06982164829969406, "global_step": 22691, "epoch": 630, "lr": 0.001} +{"train_loss": 0.034795500338077545, "train_loss_bc": 0.034795500338077545, "train_loss_llm": 0.0, "grad_norm": 0.09898699074983597, "global_step": 22692, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03335069864988327, "train_loss_bc": 0.03335069864988327, "train_loss_llm": 0.0, "grad_norm": 0.03517535701394081, "global_step": 22693, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03774428740143776, "train_loss_bc": 0.03774428740143776, "train_loss_llm": 0.0, "grad_norm": 0.06358129531145096, "global_step": 22694, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03299062326550484, "train_loss_bc": 0.03299062326550484, "train_loss_llm": 0.0, "grad_norm": 0.07850132882595062, "global_step": 22695, "epoch": 630, "lr": 0.001} +{"train_loss": 0.031124860048294067, "train_loss_bc": 0.031124860048294067, "train_loss_llm": 0.0, "grad_norm": 0.09281913191080093, "global_step": 22696, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03414614871144295, "train_loss_bc": 0.03414614871144295, "train_loss_llm": 0.0, "grad_norm": 0.034732431173324585, "global_step": 22697, "epoch": 630, "lr": 0.001} +{"train_loss": 0.02708258293569088, "train_loss_bc": 0.02708258293569088, "train_loss_llm": 0.0, "grad_norm": 0.056186895817518234, "global_step": 22698, "epoch": 630, "lr": 0.001} +{"train_loss": 0.029594052582979202, "train_loss_bc": 0.029592499136924744, "train_loss_llm": 0.00015538930892944336, "grad_norm": 0.07568421214818954, "global_step": 22699, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03267045319080353, "train_loss_bc": 0.03267045319080353, "train_loss_llm": 0.0, "grad_norm": 0.09998082369565964, "global_step": 22700, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03263098746538162, "train_loss_bc": 0.03263098746538162, "train_loss_llm": 0.0, "grad_norm": 0.027576856315135956, "global_step": 22701, "epoch": 630, "lr": 0.001} +{"train_loss": 0.036416150629520416, "train_loss_bc": 0.036416150629520416, "train_loss_llm": 0.0, "grad_norm": 0.04624749347567558, "global_step": 22702, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03454899415373802, "train_loss_bc": 0.0345458909869194, "train_loss_llm": 0.0003103911876678467, "grad_norm": 0.06279369443655014, "global_step": 22703, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03796989470720291, "train_loss_bc": 0.03796989470720291, "train_loss_llm": 0.0, "grad_norm": 0.07245563715696335, "global_step": 22704, "epoch": 630, "lr": 0.001} +{"train_loss": 0.030564110726118088, "train_loss_bc": 0.030546795576810837, "train_loss_llm": 0.00173148512840271, "grad_norm": 0.031373873353004456, "global_step": 22705, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03163124993443489, "train_loss_bc": 0.03163124993443489, "train_loss_llm": 0.0, "grad_norm": 0.05490244925022125, "global_step": 22706, "epoch": 630, "lr": 0.001} +{"train_loss": 0.02937011420726776, "train_loss_bc": 0.02936709299683571, "train_loss_llm": 0.000302046537399292, "grad_norm": 0.07563945651054382, "global_step": 22707, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03290339186787605, "train_loss_bc": 0.03290339186787605, "train_loss_llm": 0.0, "grad_norm": 0.09493546932935715, "global_step": 22708, "epoch": 630, "lr": 0.001} +{"train_loss": 0.032595470547676086, "train_loss_bc": 0.03258853405714035, "train_loss_llm": 0.0006938278675079346, "grad_norm": 0.03178771212697029, "global_step": 22709, "epoch": 630, "lr": 0.001} +{"train_loss": 0.036109670996665955, "train_loss_bc": 0.036109670996665955, "train_loss_llm": 0.0, "grad_norm": 0.06251420825719833, "global_step": 22710, "epoch": 630, "lr": 0.001} +{"train_loss": 0.0344211682677269, "train_loss_bc": 0.0344211682677269, "train_loss_llm": 0.0, "grad_norm": 0.07827767729759216, "global_step": 22711, "epoch": 630, "lr": 0.001} +{"train_loss": 0.03676281124353409, "train_loss_bc": 0.03675025328993797, "train_loss_llm": 0.0012557506561279297, "grad_norm": 0.10334412008523941, "global_step": 22712, "epoch": 630, "lr": 0.001} +{"train_loss": 0.033074479550123215, "train_loss_bc": 0.033046502619981766, "train_loss_llm": 0.0027978718280792236, "grad_norm": 0.03875851258635521, "global_step": 22713, "epoch": 630, "lr": 0.001} +{"train_loss": 0.033931344747543335, "train_loss_bc": 0.033931344747543335, "train_loss_llm": 0.0, "grad_norm": 0.06450244784355164, "global_step": 22714, "epoch": 630, "lr": 0.001} +{"train_loss": 0.032508861325267285, "train_loss_bc": 0.030228864401578903, "train_loss_llm": 0.000782698392868042, "grad_norm": 0.09111401438713074, "global_step": 22715, "epoch": 630, "lr": 0.001, "train/cumulative_reward": 114.75923789128599, "train/mean_score": 14.745612713195694, "train/success_rate": 0.3, "test/cumulative_reward": 49.85429197778992, "test/mean_score": 7.410258582651404, "test/success_rate": 0.18, "val_loss": 0.08427022397518158, "train_action_mse_error": 0.03559250384569168} +{"train_loss": 0.033784233033657074, "train_loss_bc": 0.033784233033657074, "train_loss_llm": 0.0, "grad_norm": 0.11819552630186081, "global_step": 22716, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03407231345772743, "train_loss_bc": 0.03407147154211998, "train_loss_llm": 8.401274681091309e-05, "grad_norm": 0.034278567880392075, "global_step": 22717, "epoch": 631, "lr": 0.001} +{"train_loss": 0.024789106100797653, "train_loss_bc": 0.024789106100797653, "train_loss_llm": 0.0, "grad_norm": 0.04825234040617943, "global_step": 22718, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03371361643075943, "train_loss_bc": 0.03371361643075943, "train_loss_llm": 0.0, "grad_norm": 0.0777471512556076, "global_step": 22719, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03132137283682823, "train_loss_bc": 0.03132137283682823, "train_loss_llm": 0.0, "grad_norm": 0.09833309799432755, "global_step": 22720, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03616030141711235, "train_loss_bc": 0.03615657985210419, "train_loss_llm": 0.000372081995010376, "grad_norm": 0.029729072004556656, "global_step": 22721, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03097555600106716, "train_loss_bc": 0.03097555600106716, "train_loss_llm": 0.0, "grad_norm": 0.04239383712410927, "global_step": 22722, "epoch": 631, "lr": 0.001} +{"train_loss": 0.025468364357948303, "train_loss_bc": 0.025468364357948303, "train_loss_llm": 0.0, "grad_norm": 0.04595692828297615, "global_step": 22723, "epoch": 631, "lr": 0.001} +{"train_loss": 0.030796784907579422, "train_loss_bc": 0.030796784907579422, "train_loss_llm": 0.0, "grad_norm": 0.060688816010951996, "global_step": 22724, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03664345294237137, "train_loss_bc": 0.03664345294237137, "train_loss_llm": 0.0, "grad_norm": 0.029344767332077026, "global_step": 22725, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03847447410225868, "train_loss_bc": 0.03847447410225868, "train_loss_llm": 0.0, "grad_norm": 0.04298631474375725, "global_step": 22726, "epoch": 631, "lr": 0.001} +{"train_loss": 0.035143107175827026, "train_loss_bc": 0.035143107175827026, "train_loss_llm": 0.0, "grad_norm": 0.057567574083805084, "global_step": 22727, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03545178845524788, "train_loss_bc": 0.035433605313301086, "train_loss_llm": 0.0018182992935180664, "grad_norm": 0.06840731203556061, "global_step": 22728, "epoch": 631, "lr": 0.001} +{"train_loss": 0.030833769589662552, "train_loss_bc": 0.030822589993476868, "train_loss_llm": 0.0011178851127624512, "grad_norm": 0.03364961966872215, "global_step": 22729, "epoch": 631, "lr": 0.001} +{"train_loss": 0.033698584884405136, "train_loss_bc": 0.033698584884405136, "train_loss_llm": 0.0, "grad_norm": 0.05567470192909241, "global_step": 22730, "epoch": 631, "lr": 0.001} +{"train_loss": 0.030275195837020874, "train_loss_bc": 0.030252009630203247, "train_loss_llm": 0.0023185312747955322, "grad_norm": 0.07573549449443817, "global_step": 22731, "epoch": 631, "lr": 0.001} +{"train_loss": 0.031430717557668686, "train_loss_bc": 0.031430717557668686, "train_loss_llm": 0.0, "grad_norm": 0.09736249595880508, "global_step": 22732, "epoch": 631, "lr": 0.001} +{"train_loss": 0.029834309592843056, "train_loss_bc": 0.029834309592843056, "train_loss_llm": 0.0, "grad_norm": 0.03816185146570206, "global_step": 22733, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03702850267291069, "train_loss_bc": 0.03702850267291069, "train_loss_llm": 0.0, "grad_norm": 0.06703483313322067, "global_step": 22734, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03508631885051727, "train_loss_bc": 0.03508631885051727, "train_loss_llm": 0.0, "grad_norm": 0.08061148226261139, "global_step": 22735, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03258141502737999, "train_loss_bc": 0.03258141502737999, "train_loss_llm": 0.0, "grad_norm": 0.09712934494018555, "global_step": 22736, "epoch": 631, "lr": 0.001} +{"train_loss": 0.0317496620118618, "train_loss_bc": 0.0317496620118618, "train_loss_llm": 0.0, "grad_norm": 0.02927982248365879, "global_step": 22737, "epoch": 631, "lr": 0.001} +{"train_loss": 0.030963005498051643, "train_loss_bc": 0.030963005498051643, "train_loss_llm": 0.0, "grad_norm": 0.05003149062395096, "global_step": 22738, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03145791217684746, "train_loss_bc": 0.031452372670173645, "train_loss_llm": 0.0005539953708648682, "grad_norm": 0.07301207631826401, "global_step": 22739, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03372165560722351, "train_loss_bc": 0.03372165560722351, "train_loss_llm": 0.0, "grad_norm": 0.10094007849693298, "global_step": 22740, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03188939020037651, "train_loss_bc": 0.03188939020037651, "train_loss_llm": 0.0, "grad_norm": 0.028998281806707382, "global_step": 22741, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03341328725218773, "train_loss_bc": 0.03341328725218773, "train_loss_llm": 0.0, "grad_norm": 0.05550633370876312, "global_step": 22742, "epoch": 631, "lr": 0.001} +{"train_loss": 0.029068542644381523, "train_loss_bc": 0.029068542644381523, "train_loss_llm": 0.0, "grad_norm": 0.0725683942437172, "global_step": 22743, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03429821506142616, "train_loss_bc": 0.03429821506142616, "train_loss_llm": 0.0, "grad_norm": 0.09445172548294067, "global_step": 22744, "epoch": 631, "lr": 0.001} +{"train_loss": 0.02715594694018364, "train_loss_bc": 0.02714596316218376, "train_loss_llm": 0.0009984076023101807, "grad_norm": 0.031659793108701706, "global_step": 22745, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03600621595978737, "train_loss_bc": 0.03600621595978737, "train_loss_llm": 0.0, "grad_norm": 0.051342789083719254, "global_step": 22746, "epoch": 631, "lr": 0.001} +{"train_loss": 0.030385881662368774, "train_loss_bc": 0.03037297911942005, "train_loss_llm": 0.0012902021408081055, "grad_norm": 0.07529302686452866, "global_step": 22747, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03178465738892555, "train_loss_bc": 0.03178135305643082, "train_loss_llm": 0.000330507755279541, "grad_norm": 0.0921916663646698, "global_step": 22748, "epoch": 631, "lr": 0.001} +{"train_loss": 0.031587325036525726, "train_loss_bc": 0.031587325036525726, "train_loss_llm": 0.0, "grad_norm": 0.033157411962747574, "global_step": 22749, "epoch": 631, "lr": 0.001} +{"train_loss": 0.03198549523949623, "train_loss_bc": 0.03197872266173363, "train_loss_llm": 0.0006771087646484375, "grad_norm": 0.04870786517858505, "global_step": 22750, "epoch": 631, "lr": 0.001} +{"train_loss": 0.032401897427108556, "train_loss_bc": 0.033437829464673996, "train_loss_llm": 0.0, "grad_norm": 0.06527706980705261, "global_step": 22751, "epoch": 631, "lr": 0.001, "val_loss": 0.0840061604976654} +{"train_loss": 0.034365054219961166, "train_loss_bc": 0.034356530755758286, "train_loss_llm": 0.0008522868156433105, "grad_norm": 0.07471582293510437, "global_step": 22752, "epoch": 632, "lr": 0.001} +{"train_loss": 0.029566867277026176, "train_loss_bc": 0.029566867277026176, "train_loss_llm": 0.0, "grad_norm": 0.031081773340702057, "global_step": 22753, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03819628432393074, "train_loss_bc": 0.03819628432393074, "train_loss_llm": 0.0, "grad_norm": 0.05583915114402771, "global_step": 22754, "epoch": 632, "lr": 0.001} +{"train_loss": 0.024875618517398834, "train_loss_bc": 0.024875618517398834, "train_loss_llm": 0.0, "grad_norm": 0.06659527122974396, "global_step": 22755, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03117898479104042, "train_loss_bc": 0.03117898479104042, "train_loss_llm": 0.0, "grad_norm": 0.08275064080953598, "global_step": 22756, "epoch": 632, "lr": 0.001} +{"train_loss": 0.02956305630505085, "train_loss_bc": 0.02956305630505085, "train_loss_llm": 0.0, "grad_norm": 0.032180361449718475, "global_step": 22757, "epoch": 632, "lr": 0.001} +{"train_loss": 0.033964093774557114, "train_loss_bc": 0.033964093774557114, "train_loss_llm": 0.0, "grad_norm": 0.06285407394170761, "global_step": 22758, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03510873019695282, "train_loss_bc": 0.03510873019695282, "train_loss_llm": 0.0, "grad_norm": 0.1022147461771965, "global_step": 22759, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03215378522872925, "train_loss_bc": 0.032147858291864395, "train_loss_llm": 0.000592648983001709, "grad_norm": 0.12956641614437103, "global_step": 22760, "epoch": 632, "lr": 0.001} +{"train_loss": 0.035027116537094116, "train_loss_bc": 0.035027116537094116, "train_loss_llm": 0.0, "grad_norm": 0.03055846504867077, "global_step": 22761, "epoch": 632, "lr": 0.001} +{"train_loss": 0.02726152166724205, "train_loss_bc": 0.027259420603513718, "train_loss_llm": 0.00021010637283325195, "grad_norm": 0.05663789063692093, "global_step": 22762, "epoch": 632, "lr": 0.001} +{"train_loss": 0.028486456722021103, "train_loss_bc": 0.028484975919127464, "train_loss_llm": 0.0001481473445892334, "grad_norm": 0.08493597060441971, "global_step": 22763, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03301864117383957, "train_loss_bc": 0.03300430625677109, "train_loss_llm": 0.0014336109161376953, "grad_norm": 0.10768062621355057, "global_step": 22764, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03192688897252083, "train_loss_bc": 0.03192688897252083, "train_loss_llm": 0.0, "grad_norm": 0.034334175288677216, "global_step": 22765, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03338639438152313, "train_loss_bc": 0.03338639438152313, "train_loss_llm": 0.0, "grad_norm": 0.058639656752347946, "global_step": 22766, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03269913047552109, "train_loss_bc": 0.03269913047552109, "train_loss_llm": 0.0, "grad_norm": 0.08286129683256149, "global_step": 22767, "epoch": 632, "lr": 0.001} +{"train_loss": 0.032979339361190796, "train_loss_bc": 0.032979339361190796, "train_loss_llm": 0.0, "grad_norm": 0.09543190896511078, "global_step": 22768, "epoch": 632, "lr": 0.001} +{"train_loss": 0.035102736204862595, "train_loss_bc": 0.035102736204862595, "train_loss_llm": 0.0, "grad_norm": 0.026317866519093513, "global_step": 22769, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03496008738875389, "train_loss_bc": 0.03496008738875389, "train_loss_llm": 0.0, "grad_norm": 0.04162054881453514, "global_step": 22770, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03297985717654228, "train_loss_bc": 0.03297985717654228, "train_loss_llm": 0.0, "grad_norm": 0.048766158521175385, "global_step": 22771, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03840979188680649, "train_loss_bc": 0.03840979188680649, "train_loss_llm": 0.0, "grad_norm": 0.07681215554475784, "global_step": 22772, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03792145848274231, "train_loss_bc": 0.03792145848274231, "train_loss_llm": 0.0, "grad_norm": 0.04974031075835228, "global_step": 22773, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03235166147351265, "train_loss_bc": 0.03235166147351265, "train_loss_llm": 0.0, "grad_norm": 0.06162329390645027, "global_step": 22774, "epoch": 632, "lr": 0.001} +{"train_loss": 0.034923311322927475, "train_loss_bc": 0.034923311322927475, "train_loss_llm": 0.0, "grad_norm": 0.07686569541692734, "global_step": 22775, "epoch": 632, "lr": 0.001} +{"train_loss": 0.031760893762111664, "train_loss_bc": 0.031760893762111664, "train_loss_llm": 0.0, "grad_norm": 0.10270552337169647, "global_step": 22776, "epoch": 632, "lr": 0.001} +{"train_loss": 0.02787429466843605, "train_loss_bc": 0.02787429466843605, "train_loss_llm": 0.0, "grad_norm": 0.033395666629076004, "global_step": 22777, "epoch": 632, "lr": 0.001} +{"train_loss": 0.036080580204725266, "train_loss_bc": 0.036065325140953064, "train_loss_llm": 0.0015254020690917969, "grad_norm": 0.056234680116176605, "global_step": 22778, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03443089500069618, "train_loss_bc": 0.03442702442407608, "train_loss_llm": 0.0003871321678161621, "grad_norm": 0.07650772482156754, "global_step": 22779, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03378668799996376, "train_loss_bc": 0.03378668799996376, "train_loss_llm": 0.0, "grad_norm": 0.09121914207935333, "global_step": 22780, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03182602673768997, "train_loss_bc": 0.03182602673768997, "train_loss_llm": 0.0, "grad_norm": 0.03585192188620567, "global_step": 22781, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03352027013897896, "train_loss_bc": 0.03352027013897896, "train_loss_llm": 0.0, "grad_norm": 0.05934527516365051, "global_step": 22782, "epoch": 632, "lr": 0.001} +{"train_loss": 0.028537040576338768, "train_loss_bc": 0.028537040576338768, "train_loss_llm": 0.0, "grad_norm": 0.08341016620397568, "global_step": 22783, "epoch": 632, "lr": 0.001} +{"train_loss": 0.027344077825546265, "train_loss_bc": 0.027344077825546265, "train_loss_llm": 0.0, "grad_norm": 0.1044909730553627, "global_step": 22784, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03215892240405083, "train_loss_bc": 0.03215461224317551, "train_loss_llm": 0.0004310011863708496, "grad_norm": 0.031008172780275345, "global_step": 22785, "epoch": 632, "lr": 0.001} +{"train_loss": 0.030719606205821037, "train_loss_bc": 0.030689099803566933, "train_loss_llm": 0.0030506253242492676, "grad_norm": 0.05983907729387283, "global_step": 22786, "epoch": 632, "lr": 0.001} +{"train_loss": 0.03249230473819706, "train_loss_bc": 0.03127680718898773, "train_loss_llm": 0.0, "grad_norm": 0.08519164472818375, "global_step": 22787, "epoch": 632, "lr": 0.001, "val_loss": 0.08440186083316803} +{"train_loss": 0.03118562139570713, "train_loss_bc": 0.031181881204247475, "train_loss_llm": 0.00037401914596557617, "grad_norm": 0.09927385300397873, "global_step": 22788, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03293989226222038, "train_loss_bc": 0.03293989226222038, "train_loss_llm": 0.0, "grad_norm": 0.029941605404019356, "global_step": 22789, "epoch": 633, "lr": 0.001} +{"train_loss": 0.02931828424334526, "train_loss_bc": 0.029293088242411613, "train_loss_llm": 0.0025196373462677, "grad_norm": 0.03361565247178078, "global_step": 22790, "epoch": 633, "lr": 0.001} +{"train_loss": 0.035552773624658585, "train_loss_bc": 0.03555114567279816, "train_loss_llm": 0.00016266107559204102, "grad_norm": 0.05194459483027458, "global_step": 22791, "epoch": 633, "lr": 0.001} +{"train_loss": 0.032671596854925156, "train_loss_bc": 0.032671596854925156, "train_loss_llm": 0.0, "grad_norm": 0.07384718209505081, "global_step": 22792, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03243761137127876, "train_loss_bc": 0.03242691606283188, "train_loss_llm": 0.001069486141204834, "grad_norm": 0.03162573650479317, "global_step": 22793, "epoch": 633, "lr": 0.001} +{"train_loss": 0.033886417746543884, "train_loss_bc": 0.033878669142723083, "train_loss_llm": 0.0007749497890472412, "grad_norm": 0.05068708211183548, "global_step": 22794, "epoch": 633, "lr": 0.001} +{"train_loss": 0.031687162816524506, "train_loss_bc": 0.031687162816524506, "train_loss_llm": 0.0, "grad_norm": 0.0710582286119461, "global_step": 22795, "epoch": 633, "lr": 0.001} +{"train_loss": 0.031807344406843185, "train_loss_bc": 0.031807344406843185, "train_loss_llm": 0.0, "grad_norm": 0.09740079194307327, "global_step": 22796, "epoch": 633, "lr": 0.001} +{"train_loss": 0.02729550376534462, "train_loss_bc": 0.027275828644633293, "train_loss_llm": 0.0019674599170684814, "grad_norm": 0.03133115917444229, "global_step": 22797, "epoch": 633, "lr": 0.001} +{"train_loss": 0.034176845103502274, "train_loss_bc": 0.034176845103502274, "train_loss_llm": 0.0, "grad_norm": 0.044821809977293015, "global_step": 22798, "epoch": 633, "lr": 0.001} +{"train_loss": 0.027626266703009605, "train_loss_bc": 0.027626266703009605, "train_loss_llm": 0.0, "grad_norm": 0.06236429885029793, "global_step": 22799, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03260510787367821, "train_loss_bc": 0.03260510787367821, "train_loss_llm": 0.0, "grad_norm": 0.08176687359809875, "global_step": 22800, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03342871740460396, "train_loss_bc": 0.03342871740460396, "train_loss_llm": 0.0, "grad_norm": 0.03202451765537262, "global_step": 22801, "epoch": 633, "lr": 0.001} +{"train_loss": 0.030080046504735947, "train_loss_bc": 0.030080046504735947, "train_loss_llm": 0.0, "grad_norm": 0.04654400050640106, "global_step": 22802, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03593695908784866, "train_loss_bc": 0.035932693630456924, "train_loss_llm": 0.0004266798496246338, "grad_norm": 0.06938443332910538, "global_step": 22803, "epoch": 633, "lr": 0.001} +{"train_loss": 0.02744985930621624, "train_loss_bc": 0.02744985930621624, "train_loss_llm": 0.0, "grad_norm": 0.08266831189393997, "global_step": 22804, "epoch": 633, "lr": 0.001} +{"train_loss": 0.035526759922504425, "train_loss_bc": 0.035526759922504425, "train_loss_llm": 0.0, "grad_norm": 0.02757364884018898, "global_step": 22805, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03461233526468277, "train_loss_bc": 0.03461233526468277, "train_loss_llm": 0.0, "grad_norm": 0.03600117936730385, "global_step": 22806, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03579438105225563, "train_loss_bc": 0.03578244522213936, "train_loss_llm": 0.0011935830116271973, "grad_norm": 0.05648990720510483, "global_step": 22807, "epoch": 633, "lr": 0.001} +{"train_loss": 0.02614836022257805, "train_loss_bc": 0.02614836022257805, "train_loss_llm": 0.0, "grad_norm": 0.06415822356939316, "global_step": 22808, "epoch": 633, "lr": 0.001} +{"train_loss": 0.030986079946160316, "train_loss_bc": 0.030986079946160316, "train_loss_llm": 0.0, "grad_norm": 0.03084029257297516, "global_step": 22809, "epoch": 633, "lr": 0.001} +{"train_loss": 0.031012142077088356, "train_loss_bc": 0.031012142077088356, "train_loss_llm": 0.0, "grad_norm": 0.04945465177297592, "global_step": 22810, "epoch": 633, "lr": 0.001} +{"train_loss": 0.029953282326459885, "train_loss_bc": 0.029953282326459885, "train_loss_llm": 0.0, "grad_norm": 0.06576000899076462, "global_step": 22811, "epoch": 633, "lr": 0.001} +{"train_loss": 0.030305514112114906, "train_loss_bc": 0.030305514112114906, "train_loss_llm": 0.0, "grad_norm": 0.09359432756900787, "global_step": 22812, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03764767199754715, "train_loss_bc": 0.03764767199754715, "train_loss_llm": 0.0, "grad_norm": 0.032975904643535614, "global_step": 22813, "epoch": 633, "lr": 0.001} +{"train_loss": 0.028212031349539757, "train_loss_bc": 0.028212031349539757, "train_loss_llm": 0.0, "grad_norm": 0.04693208262324333, "global_step": 22814, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03187299892306328, "train_loss_bc": 0.03186861053109169, "train_loss_llm": 0.00043901801109313965, "grad_norm": 0.0624535009264946, "global_step": 22815, "epoch": 633, "lr": 0.001} +{"train_loss": 0.030236070975661278, "train_loss_bc": 0.030236070975661278, "train_loss_llm": 0.0, "grad_norm": 0.07613786309957504, "global_step": 22816, "epoch": 633, "lr": 0.001} +{"train_loss": 0.032660167664289474, "train_loss_bc": 0.032660167664289474, "train_loss_llm": 0.0, "grad_norm": 0.034120239317417145, "global_step": 22817, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03700733929872513, "train_loss_bc": 0.03700733929872513, "train_loss_llm": 0.0, "grad_norm": 0.054055046290159225, "global_step": 22818, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03359808772802353, "train_loss_bc": 0.033585868775844574, "train_loss_llm": 0.0012217462062835693, "grad_norm": 0.06596261262893677, "global_step": 22819, "epoch": 633, "lr": 0.001} +{"train_loss": 0.03114232048392296, "train_loss_bc": 0.03114232048392296, "train_loss_llm": 0.0, "grad_norm": 0.07211491465568542, "global_step": 22820, "epoch": 633, "lr": 0.001} +{"train_loss": 0.033976759761571884, "train_loss_bc": 0.033976759761571884, "train_loss_llm": 0.0, "grad_norm": 0.03106124885380268, "global_step": 22821, "epoch": 633, "lr": 0.001} +{"train_loss": 0.032423216849565506, "train_loss_bc": 0.032423216849565506, "train_loss_llm": 0.0, "grad_norm": 0.03944740444421768, "global_step": 22822, "epoch": 633, "lr": 0.001} +{"train_loss": 0.032106497738924294, "train_loss_bc": 0.032632388174533844, "train_loss_llm": 0.0, "grad_norm": 0.04796072468161583, "global_step": 22823, "epoch": 633, "lr": 0.001, "val_loss": 0.08313116431236267} +{"train_loss": 0.025897588580846786, "train_loss_bc": 0.025897588580846786, "train_loss_llm": 0.0, "grad_norm": 0.052358683198690414, "global_step": 22824, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03256304934620857, "train_loss_bc": 0.03256304934620857, "train_loss_llm": 0.0, "grad_norm": 0.030032334849238396, "global_step": 22825, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03627365082502365, "train_loss_bc": 0.03627365082502365, "train_loss_llm": 0.0, "grad_norm": 0.04604632779955864, "global_step": 22826, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03159371390938759, "train_loss_bc": 0.03159371390938759, "train_loss_llm": 0.0, "grad_norm": 0.056535642594099045, "global_step": 22827, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03067510947585106, "train_loss_bc": 0.03067510947585106, "train_loss_llm": 0.0, "grad_norm": 0.06418241560459137, "global_step": 22828, "epoch": 634, "lr": 0.001} +{"train_loss": 0.02998625673353672, "train_loss_bc": 0.02998625673353672, "train_loss_llm": 0.0, "grad_norm": 0.028510423377156258, "global_step": 22829, "epoch": 634, "lr": 0.001} +{"train_loss": 0.02872452698647976, "train_loss_bc": 0.02872452698647976, "train_loss_llm": 0.0, "grad_norm": 0.04509981349110603, "global_step": 22830, "epoch": 634, "lr": 0.001} +{"train_loss": 0.028648508712649345, "train_loss_bc": 0.028648508712649345, "train_loss_llm": 0.0, "grad_norm": 0.06778981536626816, "global_step": 22831, "epoch": 634, "lr": 0.001} +{"train_loss": 0.027770724147558212, "train_loss_bc": 0.02776581235229969, "train_loss_llm": 0.0004912614822387695, "grad_norm": 0.07696811854839325, "global_step": 22832, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03765111789107323, "train_loss_bc": 0.03765111789107323, "train_loss_llm": 0.0, "grad_norm": 0.03696862980723381, "global_step": 22833, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03412375599145889, "train_loss_bc": 0.03412375599145889, "train_loss_llm": 0.0, "grad_norm": 0.05362909659743309, "global_step": 22834, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03306598588824272, "train_loss_bc": 0.03306598588824272, "train_loss_llm": 0.0, "grad_norm": 0.06864876300096512, "global_step": 22835, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03633023425936699, "train_loss_bc": 0.03633023425936699, "train_loss_llm": 0.0, "grad_norm": 0.08224959671497345, "global_step": 22836, "epoch": 634, "lr": 0.001} +{"train_loss": 0.029297174885869026, "train_loss_bc": 0.029297174885869026, "train_loss_llm": 0.0, "grad_norm": 0.030211051926016808, "global_step": 22837, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03316522017121315, "train_loss_bc": 0.03315852954983711, "train_loss_llm": 0.0006691813468933105, "grad_norm": 0.04421742260456085, "global_step": 22838, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03224002942442894, "train_loss_bc": 0.03224002942442894, "train_loss_llm": 0.0, "grad_norm": 0.04942404106259346, "global_step": 22839, "epoch": 634, "lr": 0.001} +{"train_loss": 0.030890000984072685, "train_loss_bc": 0.030890000984072685, "train_loss_llm": 0.0, "grad_norm": 0.0642126277089119, "global_step": 22840, "epoch": 634, "lr": 0.001} +{"train_loss": 0.029984181746840477, "train_loss_bc": 0.029984181746840477, "train_loss_llm": 0.0, "grad_norm": 0.02589784376323223, "global_step": 22841, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03149939700961113, "train_loss_bc": 0.03149939700961113, "train_loss_llm": 0.0, "grad_norm": 0.03567374125123024, "global_step": 22842, "epoch": 634, "lr": 0.001} +{"train_loss": 0.026278460398316383, "train_loss_bc": 0.02625994384288788, "train_loss_llm": 0.0018517374992370605, "grad_norm": 0.046103861182928085, "global_step": 22843, "epoch": 634, "lr": 0.001} +{"train_loss": 0.029477031901478767, "train_loss_bc": 0.029477031901478767, "train_loss_llm": 0.0, "grad_norm": 0.06483481079339981, "global_step": 22844, "epoch": 634, "lr": 0.001} +{"train_loss": 0.034116800874471664, "train_loss_bc": 0.034116800874471664, "train_loss_llm": 0.0, "grad_norm": 0.034527990967035294, "global_step": 22845, "epoch": 634, "lr": 0.001} +{"train_loss": 0.02806929498910904, "train_loss_bc": 0.02806929498910904, "train_loss_llm": 0.0, "grad_norm": 0.047099947929382324, "global_step": 22846, "epoch": 634, "lr": 0.001} +{"train_loss": 0.033626358956098557, "train_loss_bc": 0.03361806273460388, "train_loss_llm": 0.0008296966552734375, "grad_norm": 0.0639088973402977, "global_step": 22847, "epoch": 634, "lr": 0.001} +{"train_loss": 0.034479059278964996, "train_loss_bc": 0.034479059278964996, "train_loss_llm": 0.0, "grad_norm": 0.07963719964027405, "global_step": 22848, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03821179270744324, "train_loss_bc": 0.03821179270744324, "train_loss_llm": 0.0, "grad_norm": 0.03689274564385414, "global_step": 22849, "epoch": 634, "lr": 0.001} +{"train_loss": 0.030888957902789116, "train_loss_bc": 0.030888957902789116, "train_loss_llm": 0.0, "grad_norm": 0.05116930976510048, "global_step": 22850, "epoch": 634, "lr": 0.001} +{"train_loss": 0.025715269148349762, "train_loss_bc": 0.025715269148349762, "train_loss_llm": 0.0, "grad_norm": 0.0653042271733284, "global_step": 22851, "epoch": 634, "lr": 0.001} +{"train_loss": 0.035749588161706924, "train_loss_bc": 0.035749588161706924, "train_loss_llm": 0.0, "grad_norm": 0.08734799176454544, "global_step": 22852, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03313329815864563, "train_loss_bc": 0.03312728926539421, "train_loss_llm": 0.0006010234355926514, "grad_norm": 0.02811049111187458, "global_step": 22853, "epoch": 634, "lr": 0.001} +{"train_loss": 0.04016720503568649, "train_loss_bc": 0.04016720503568649, "train_loss_llm": 0.0, "grad_norm": 0.04478877782821655, "global_step": 22854, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03303397446870804, "train_loss_bc": 0.03303199261426926, "train_loss_llm": 0.00019824504852294922, "grad_norm": 0.05821916088461876, "global_step": 22855, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03212624415755272, "train_loss_bc": 0.03212624415755272, "train_loss_llm": 0.0, "grad_norm": 0.07673719525337219, "global_step": 22856, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03237825632095337, "train_loss_bc": 0.03237825632095337, "train_loss_llm": 0.0, "grad_norm": 0.02701423689723015, "global_step": 22857, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03149772435426712, "train_loss_bc": 0.03149602934718132, "train_loss_llm": 0.00016963481903076172, "grad_norm": 0.04048750922083855, "global_step": 22858, "epoch": 634, "lr": 0.001} +{"train_loss": 0.03211872998832001, "train_loss_bc": 0.03692004829645157, "train_loss_llm": 0.002468675374984741, "grad_norm": 0.052607327699661255, "global_step": 22859, "epoch": 634, "lr": 0.001, "val_loss": 0.08512076735496521} +{"train_loss": 0.033602479845285416, "train_loss_bc": 0.033587079495191574, "train_loss_llm": 0.001539856195449829, "grad_norm": 0.07037746161222458, "global_step": 22860, "epoch": 635, "lr": 0.001} +{"train_loss": 0.0322575680911541, "train_loss_bc": 0.0322575680911541, "train_loss_llm": 0.0, "grad_norm": 0.0234421007335186, "global_step": 22861, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03287278488278389, "train_loss_bc": 0.03287278488278389, "train_loss_llm": 0.0, "grad_norm": 0.03997541218996048, "global_step": 22862, "epoch": 635, "lr": 0.001} +{"train_loss": 0.02787630446255207, "train_loss_bc": 0.02787630446255207, "train_loss_llm": 0.0, "grad_norm": 0.047835275530815125, "global_step": 22863, "epoch": 635, "lr": 0.001} +{"train_loss": 0.0349346324801445, "train_loss_bc": 0.0349346324801445, "train_loss_llm": 0.0, "grad_norm": 0.05677562206983566, "global_step": 22864, "epoch": 635, "lr": 0.001} +{"train_loss": 0.034271832555532455, "train_loss_bc": 0.034271832555532455, "train_loss_llm": 0.0, "grad_norm": 0.03260978311300278, "global_step": 22865, "epoch": 635, "lr": 0.001} +{"train_loss": 0.030419403687119484, "train_loss_bc": 0.030419403687119484, "train_loss_llm": 0.0, "grad_norm": 0.045220039784908295, "global_step": 22866, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03562578558921814, "train_loss_bc": 0.03562578558921814, "train_loss_llm": 0.0, "grad_norm": 0.05770249292254448, "global_step": 22867, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03248574212193489, "train_loss_bc": 0.03248574212193489, "train_loss_llm": 0.0, "grad_norm": 0.06468383222818375, "global_step": 22868, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03381776064634323, "train_loss_bc": 0.03381776064634323, "train_loss_llm": 0.0, "grad_norm": 0.0364416278898716, "global_step": 22869, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03609957918524742, "train_loss_bc": 0.03609957918524742, "train_loss_llm": 0.0, "grad_norm": 0.05596986785531044, "global_step": 22870, "epoch": 635, "lr": 0.001} +{"train_loss": 0.030485980212688446, "train_loss_bc": 0.030485980212688446, "train_loss_llm": 0.0, "grad_norm": 0.06311143189668655, "global_step": 22871, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03261609748005867, "train_loss_bc": 0.03261609748005867, "train_loss_llm": 0.0, "grad_norm": 0.0795949324965477, "global_step": 22872, "epoch": 635, "lr": 0.001} +{"train_loss": 0.028040209785103798, "train_loss_bc": 0.028040209785103798, "train_loss_llm": 0.0, "grad_norm": 0.02801814302802086, "global_step": 22873, "epoch": 635, "lr": 0.001} +{"train_loss": 0.036235783249139786, "train_loss_bc": 0.036235783249139786, "train_loss_llm": 0.0, "grad_norm": 0.0393749475479126, "global_step": 22874, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03187716752290726, "train_loss_bc": 0.0318574421107769, "train_loss_llm": 0.0019724667072296143, "grad_norm": 0.05002981051802635, "global_step": 22875, "epoch": 635, "lr": 0.001} +{"train_loss": 0.025132520124316216, "train_loss_bc": 0.025132520124316216, "train_loss_llm": 0.0, "grad_norm": 0.0653953105211258, "global_step": 22876, "epoch": 635, "lr": 0.001} +{"train_loss": 0.036111000925302505, "train_loss_bc": 0.036111000925302505, "train_loss_llm": 0.0, "grad_norm": 0.028491495177149773, "global_step": 22877, "epoch": 635, "lr": 0.001} +{"train_loss": 0.0290656927973032, "train_loss_bc": 0.029039181768894196, "train_loss_llm": 0.002651035785675049, "grad_norm": 0.044840432703495026, "global_step": 22878, "epoch": 635, "lr": 0.001} +{"train_loss": 0.028880935162305832, "train_loss_bc": 0.028880935162305832, "train_loss_llm": 0.0, "grad_norm": 0.06718889623880386, "global_step": 22879, "epoch": 635, "lr": 0.001} +{"train_loss": 0.02895374409854412, "train_loss_bc": 0.02895374409854412, "train_loss_llm": 0.0, "grad_norm": 0.07916122674942017, "global_step": 22880, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03161714971065521, "train_loss_bc": 0.03161714971065521, "train_loss_llm": 0.0, "grad_norm": 0.029649144038558006, "global_step": 22881, "epoch": 635, "lr": 0.001} +{"train_loss": 0.032234977930784225, "train_loss_bc": 0.032216399908065796, "train_loss_llm": 0.0018579363822937012, "grad_norm": 0.046203430742025375, "global_step": 22882, "epoch": 635, "lr": 0.001} +{"train_loss": 0.0326794795691967, "train_loss_bc": 0.0326794795691967, "train_loss_llm": 0.0, "grad_norm": 0.06314191967248917, "global_step": 22883, "epoch": 635, "lr": 0.001} +{"train_loss": 0.030149023979902267, "train_loss_bc": 0.030149023979902267, "train_loss_llm": 0.0, "grad_norm": 0.07481490075588226, "global_step": 22884, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03598804399371147, "train_loss_bc": 0.03598645702004433, "train_loss_llm": 0.00015881657600402832, "grad_norm": 0.03659367188811302, "global_step": 22885, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03223232552409172, "train_loss_bc": 0.03223232552409172, "train_loss_llm": 0.0, "grad_norm": 0.05431697890162468, "global_step": 22886, "epoch": 635, "lr": 0.001} +{"train_loss": 0.0308469757437706, "train_loss_bc": 0.0308469757437706, "train_loss_llm": 0.0, "grad_norm": 0.06370946764945984, "global_step": 22887, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03203887492418289, "train_loss_bc": 0.032019831240177155, "train_loss_llm": 0.001904517412185669, "grad_norm": 0.07798656076192856, "global_step": 22888, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03680039569735527, "train_loss_bc": 0.03680039569735527, "train_loss_llm": 0.0, "grad_norm": 0.03386690095067024, "global_step": 22889, "epoch": 635, "lr": 0.001} +{"train_loss": 0.028440650552511215, "train_loss_bc": 0.02843642048537731, "train_loss_llm": 0.0004229247570037842, "grad_norm": 0.055808618664741516, "global_step": 22890, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03461034223437309, "train_loss_bc": 0.034590944647789, "train_loss_llm": 0.001939624547958374, "grad_norm": 0.07707768678665161, "global_step": 22891, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03315963223576546, "train_loss_bc": 0.03315963223576546, "train_loss_llm": 0.0, "grad_norm": 0.09217237681150436, "global_step": 22892, "epoch": 635, "lr": 0.001} +{"train_loss": 0.02949756756424904, "train_loss_bc": 0.02949756756424904, "train_loss_llm": 0.0, "grad_norm": 0.03220997005701065, "global_step": 22893, "epoch": 635, "lr": 0.001} +{"train_loss": 0.02914915606379509, "train_loss_bc": 0.02914915606379509, "train_loss_llm": 0.0, "grad_norm": 0.041707489639520645, "global_step": 22894, "epoch": 635, "lr": 0.001} +{"train_loss": 0.03207083625925912, "train_loss_bc": 0.033442504703998566, "train_loss_llm": 0.0, "grad_norm": 0.06276658922433853, "global_step": 22895, "epoch": 635, "lr": 0.001, "train/cumulative_reward": 119.65055317242381, "train/mean_score": 16.752792755715557, "train/success_rate": 0.4, "test/cumulative_reward": 34.5934748481094, "test/mean_score": 7.410438477416472, "test/success_rate": 0.18, "val_loss": 0.08411070704460144, "train_action_mse_error": 0.03523363545536995} +{"train_loss": 0.029201749712228775, "train_loss_bc": 0.029201749712228775, "train_loss_llm": 0.0, "grad_norm": 0.07381903380155563, "global_step": 22896, "epoch": 636, "lr": 0.001} +{"train_loss": 0.030407201498746872, "train_loss_bc": 0.030406074598431587, "train_loss_llm": 0.00011274218559265137, "grad_norm": 0.03301522135734558, "global_step": 22897, "epoch": 636, "lr": 0.001} +{"train_loss": 0.028333066031336784, "train_loss_bc": 0.028333066031336784, "train_loss_llm": 0.0, "grad_norm": 0.044972971081733704, "global_step": 22898, "epoch": 636, "lr": 0.001} +{"train_loss": 0.030930863693356514, "train_loss_bc": 0.030930863693356514, "train_loss_llm": 0.0, "grad_norm": 0.05728398263454437, "global_step": 22899, "epoch": 636, "lr": 0.001} +{"train_loss": 0.030608870089054108, "train_loss_bc": 0.030608870089054108, "train_loss_llm": 0.0, "grad_norm": 0.0660533681511879, "global_step": 22900, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03021545521914959, "train_loss_bc": 0.03021545521914959, "train_loss_llm": 0.0, "grad_norm": 0.026989901438355446, "global_step": 22901, "epoch": 636, "lr": 0.001} +{"train_loss": 0.02785516530275345, "train_loss_bc": 0.027841797098517418, "train_loss_llm": 0.0013368427753448486, "grad_norm": 0.048851240426301956, "global_step": 22902, "epoch": 636, "lr": 0.001} +{"train_loss": 0.0374433808028698, "train_loss_bc": 0.03743325173854828, "train_loss_llm": 0.0010128915309906006, "grad_norm": 0.05726587399840355, "global_step": 22903, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03678230568766594, "train_loss_bc": 0.036767520010471344, "train_loss_llm": 0.0014785826206207275, "grad_norm": 0.07436513900756836, "global_step": 22904, "epoch": 636, "lr": 0.001} +{"train_loss": 0.02953866682946682, "train_loss_bc": 0.029503140598535538, "train_loss_llm": 0.003552645444869995, "grad_norm": 0.033707454800605774, "global_step": 22905, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03048994392156601, "train_loss_bc": 0.03048994392156601, "train_loss_llm": 0.0, "grad_norm": 0.044642120599746704, "global_step": 22906, "epoch": 636, "lr": 0.001} +{"train_loss": 0.037502557039260864, "train_loss_bc": 0.0374874584376812, "train_loss_llm": 0.0015097260475158691, "grad_norm": 0.06036832556128502, "global_step": 22907, "epoch": 636, "lr": 0.001} +{"train_loss": 0.023551413789391518, "train_loss_bc": 0.023551413789391518, "train_loss_llm": 0.0, "grad_norm": 0.07594019919633865, "global_step": 22908, "epoch": 636, "lr": 0.001} +{"train_loss": 0.02519257925450802, "train_loss_bc": 0.02519257925450802, "train_loss_llm": 0.0, "grad_norm": 0.038174353539943695, "global_step": 22909, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03712156042456627, "train_loss_bc": 0.03712156042456627, "train_loss_llm": 0.0, "grad_norm": 0.06173170357942581, "global_step": 22910, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03264607489109039, "train_loss_bc": 0.032639678567647934, "train_loss_llm": 0.0006395578384399414, "grad_norm": 0.08375341445207596, "global_step": 22911, "epoch": 636, "lr": 0.001} +{"train_loss": 0.029812095686793327, "train_loss_bc": 0.029812095686793327, "train_loss_llm": 0.0, "grad_norm": 0.10720805078744888, "global_step": 22912, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03694275766611099, "train_loss_bc": 0.03694275766611099, "train_loss_llm": 0.0, "grad_norm": 0.03176935762166977, "global_step": 22913, "epoch": 636, "lr": 0.001} +{"train_loss": 0.02048700861632824, "train_loss_bc": 0.020483877509832382, "train_loss_llm": 0.00031304359436035156, "grad_norm": 0.04967351257801056, "global_step": 22914, "epoch": 636, "lr": 0.001} +{"train_loss": 0.036153994500637054, "train_loss_bc": 0.03615378588438034, "train_loss_llm": 2.0772218704223633e-05, "grad_norm": 0.06528933346271515, "global_step": 22915, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03663577139377594, "train_loss_bc": 0.03663577139377594, "train_loss_llm": 0.0, "grad_norm": 0.08269081264734268, "global_step": 22916, "epoch": 636, "lr": 0.001} +{"train_loss": 0.0336761474609375, "train_loss_bc": 0.03366943821310997, "train_loss_llm": 0.0006708800792694092, "grad_norm": 0.02808350883424282, "global_step": 22917, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03239097073674202, "train_loss_bc": 0.03239097073674202, "train_loss_llm": 0.0, "grad_norm": 0.04455331340432167, "global_step": 22918, "epoch": 636, "lr": 0.001} +{"train_loss": 0.032998912036418915, "train_loss_bc": 0.032998912036418915, "train_loss_llm": 0.0, "grad_norm": 0.06188932806253433, "global_step": 22919, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03138997033238411, "train_loss_bc": 0.03138997033238411, "train_loss_llm": 0.0, "grad_norm": 0.0701596662402153, "global_step": 22920, "epoch": 636, "lr": 0.001} +{"train_loss": 0.04517146199941635, "train_loss_bc": 0.04517146199941635, "train_loss_llm": 0.0, "grad_norm": 0.042218927294015884, "global_step": 22921, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03876528516411781, "train_loss_bc": 0.03876528516411781, "train_loss_llm": 0.0, "grad_norm": 0.06653475016355515, "global_step": 22922, "epoch": 636, "lr": 0.001} +{"train_loss": 0.04049587622284889, "train_loss_bc": 0.04049587622284889, "train_loss_llm": 0.0, "grad_norm": 0.09505472332239151, "global_step": 22923, "epoch": 636, "lr": 0.001} +{"train_loss": 0.032134801149368286, "train_loss_bc": 0.032134801149368286, "train_loss_llm": 0.0, "grad_norm": 0.11527170240879059, "global_step": 22924, "epoch": 636, "lr": 0.001} +{"train_loss": 0.035403743386268616, "train_loss_bc": 0.03540117293596268, "train_loss_llm": 0.00025710463523864746, "grad_norm": 0.03152085840702057, "global_step": 22925, "epoch": 636, "lr": 0.001} +{"train_loss": 0.032324567437171936, "train_loss_bc": 0.032324567437171936, "train_loss_llm": 0.0, "grad_norm": 0.04838138818740845, "global_step": 22926, "epoch": 636, "lr": 0.001} +{"train_loss": 0.02817455679178238, "train_loss_bc": 0.02817455679178238, "train_loss_llm": 0.0, "grad_norm": 0.06772881746292114, "global_step": 22927, "epoch": 636, "lr": 0.001} +{"train_loss": 0.033843476325273514, "train_loss_bc": 0.033843476325273514, "train_loss_llm": 0.0, "grad_norm": 0.09244397282600403, "global_step": 22928, "epoch": 636, "lr": 0.001} +{"train_loss": 0.029238233342766762, "train_loss_bc": 0.029238233342766762, "train_loss_llm": 0.0, "grad_norm": 0.02990104630589485, "global_step": 22929, "epoch": 636, "lr": 0.001} +{"train_loss": 0.029465999454259872, "train_loss_bc": 0.029465999454259872, "train_loss_llm": 0.0, "grad_norm": 0.057694729417562485, "global_step": 22930, "epoch": 636, "lr": 0.001} +{"train_loss": 0.03219249927335315, "train_loss_bc": 0.02558835968375206, "train_loss_llm": 0.001512974500656128, "grad_norm": 0.06878143548965454, "global_step": 22931, "epoch": 636, "lr": 0.001, "val_loss": 0.0852740928530693} +{"train_loss": 0.0314803346991539, "train_loss_bc": 0.0314803346991539, "train_loss_llm": 0.0, "grad_norm": 0.08360965549945831, "global_step": 22932, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03542640060186386, "train_loss_bc": 0.03542640060186386, "train_loss_llm": 0.0, "grad_norm": 0.03280731663107872, "global_step": 22933, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03790052607655525, "train_loss_bc": 0.03790052607655525, "train_loss_llm": 0.0, "grad_norm": 0.05329931154847145, "global_step": 22934, "epoch": 637, "lr": 0.001} +{"train_loss": 0.029835687950253487, "train_loss_bc": 0.029835687950253487, "train_loss_llm": 0.0, "grad_norm": 0.06593797355890274, "global_step": 22935, "epoch": 637, "lr": 0.001} +{"train_loss": 0.031513605266809464, "train_loss_bc": 0.031513605266809464, "train_loss_llm": 0.0, "grad_norm": 0.09129761159420013, "global_step": 22936, "epoch": 637, "lr": 0.001} +{"train_loss": 0.02731599658727646, "train_loss_bc": 0.02731599658727646, "train_loss_llm": 0.0, "grad_norm": 0.02485452964901924, "global_step": 22937, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03468972444534302, "train_loss_bc": 0.03468972444534302, "train_loss_llm": 0.0, "grad_norm": 0.05205710232257843, "global_step": 22938, "epoch": 637, "lr": 0.001} +{"train_loss": 0.029543880373239517, "train_loss_bc": 0.029543880373239517, "train_loss_llm": 0.0, "grad_norm": 0.06765116751194, "global_step": 22939, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03372922167181969, "train_loss_bc": 0.03372922167181969, "train_loss_llm": 0.0, "grad_norm": 0.08376099914312363, "global_step": 22940, "epoch": 637, "lr": 0.001} +{"train_loss": 0.035076793283224106, "train_loss_bc": 0.03507190942764282, "train_loss_llm": 0.0004883706569671631, "grad_norm": 0.029876885935664177, "global_step": 22941, "epoch": 637, "lr": 0.001} +{"train_loss": 0.031921520829200745, "train_loss_bc": 0.031921520829200745, "train_loss_llm": 0.0, "grad_norm": 0.049079082906246185, "global_step": 22942, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03375966101884842, "train_loss_bc": 0.03375966101884842, "train_loss_llm": 0.0, "grad_norm": 0.06519664824008942, "global_step": 22943, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03173981234431267, "train_loss_bc": 0.03173981234431267, "train_loss_llm": 0.0, "grad_norm": 0.08160365372896194, "global_step": 22944, "epoch": 637, "lr": 0.001} +{"train_loss": 0.031086498871445656, "train_loss_bc": 0.031086498871445656, "train_loss_llm": 0.0, "grad_norm": 0.032946743071079254, "global_step": 22945, "epoch": 637, "lr": 0.001} +{"train_loss": 0.035522762686014175, "train_loss_bc": 0.035516392439603806, "train_loss_llm": 0.0006370246410369873, "grad_norm": 0.05160708352923393, "global_step": 22946, "epoch": 637, "lr": 0.001} +{"train_loss": 0.030636638402938843, "train_loss_bc": 0.030636638402938843, "train_loss_llm": 0.0, "grad_norm": 0.06491701304912567, "global_step": 22947, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03086269646883011, "train_loss_bc": 0.03086269646883011, "train_loss_llm": 0.0, "grad_norm": 0.0788378119468689, "global_step": 22948, "epoch": 637, "lr": 0.001} +{"train_loss": 0.030929256230592728, "train_loss_bc": 0.030929256230592728, "train_loss_llm": 0.0, "grad_norm": 0.024498876184225082, "global_step": 22949, "epoch": 637, "lr": 0.001} +{"train_loss": 0.034923166036605835, "train_loss_bc": 0.034908559173345566, "train_loss_llm": 0.001460641622543335, "grad_norm": 0.0471121221780777, "global_step": 22950, "epoch": 637, "lr": 0.001} +{"train_loss": 0.030259890481829643, "train_loss_bc": 0.030259890481829643, "train_loss_llm": 0.0, "grad_norm": 0.06488404422998428, "global_step": 22951, "epoch": 637, "lr": 0.001} +{"train_loss": 0.029623324051499367, "train_loss_bc": 0.029623324051499367, "train_loss_llm": 0.0, "grad_norm": 0.07996053248643875, "global_step": 22952, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03364480659365654, "train_loss_bc": 0.03364480659365654, "train_loss_llm": 0.0, "grad_norm": 0.030001472681760788, "global_step": 22953, "epoch": 637, "lr": 0.001} +{"train_loss": 0.029791340231895447, "train_loss_bc": 0.029784660786390305, "train_loss_llm": 0.0006678998470306396, "grad_norm": 0.04621540755033493, "global_step": 22954, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03281344100832939, "train_loss_bc": 0.03281344100832939, "train_loss_llm": 0.0, "grad_norm": 0.05831895023584366, "global_step": 22955, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03155132755637169, "train_loss_bc": 0.03155132755637169, "train_loss_llm": 0.0, "grad_norm": 0.0781237781047821, "global_step": 22956, "epoch": 637, "lr": 0.001} +{"train_loss": 0.029579706490039825, "train_loss_bc": 0.029574928805232048, "train_loss_llm": 0.00047782063484191895, "grad_norm": 0.026956714689731598, "global_step": 22957, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03270259499549866, "train_loss_bc": 0.03269384056329727, "train_loss_llm": 0.0008753538131713867, "grad_norm": 0.050008390098810196, "global_step": 22958, "epoch": 637, "lr": 0.001} +{"train_loss": 0.033268071711063385, "train_loss_bc": 0.03325939550995827, "train_loss_llm": 0.0008676648139953613, "grad_norm": 0.06193839758634567, "global_step": 22959, "epoch": 637, "lr": 0.001} +{"train_loss": 0.02739548496901989, "train_loss_bc": 0.02739548496901989, "train_loss_llm": 0.0, "grad_norm": 0.07463425397872925, "global_step": 22960, "epoch": 637, "lr": 0.001} +{"train_loss": 0.030690856277942657, "train_loss_bc": 0.030690856277942657, "train_loss_llm": 0.0, "grad_norm": 0.03461063280701637, "global_step": 22961, "epoch": 637, "lr": 0.001} +{"train_loss": 0.029793517664074898, "train_loss_bc": 0.029788140207529068, "train_loss_llm": 0.0005377829074859619, "grad_norm": 0.06141308322548866, "global_step": 22962, "epoch": 637, "lr": 0.001} +{"train_loss": 0.0372571162879467, "train_loss_bc": 0.0372433103621006, "train_loss_llm": 0.0013805627822875977, "grad_norm": 0.0704236626625061, "global_step": 22963, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03577514737844467, "train_loss_bc": 0.035770315676927567, "train_loss_llm": 0.00048300623893737793, "grad_norm": 0.08579502254724503, "global_step": 22964, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03463747352361679, "train_loss_bc": 0.03463747352361679, "train_loss_llm": 0.0, "grad_norm": 0.037005335092544556, "global_step": 22965, "epoch": 637, "lr": 0.001} +{"train_loss": 0.032549478113651276, "train_loss_bc": 0.032549478113651276, "train_loss_llm": 0.0, "grad_norm": 0.05145774036645889, "global_step": 22966, "epoch": 637, "lr": 0.001} +{"train_loss": 0.03225456550717354, "train_loss_bc": 0.03190923109650612, "train_loss_llm": 0.002736717462539673, "grad_norm": 0.07450401782989502, "global_step": 22967, "epoch": 637, "lr": 0.001, "val_loss": 0.08419529348611832} +{"train_loss": 0.03353618085384369, "train_loss_bc": 0.03353618085384369, "train_loss_llm": 0.0, "grad_norm": 0.09152622520923615, "global_step": 22968, "epoch": 638, "lr": 0.001} +{"train_loss": 0.029945597052574158, "train_loss_bc": 0.029911689460277557, "train_loss_llm": 0.003390699625015259, "grad_norm": 0.029907114803791046, "global_step": 22969, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03470967337489128, "train_loss_bc": 0.03470967337489128, "train_loss_llm": 0.0, "grad_norm": 0.04611963406205177, "global_step": 22970, "epoch": 638, "lr": 0.001} +{"train_loss": 0.027813734486699104, "train_loss_bc": 0.027813734486699104, "train_loss_llm": 0.0, "grad_norm": 0.06030019745230675, "global_step": 22971, "epoch": 638, "lr": 0.001} +{"train_loss": 0.02787623554468155, "train_loss_bc": 0.02787623554468155, "train_loss_llm": 0.0, "grad_norm": 0.07635955512523651, "global_step": 22972, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03579425811767578, "train_loss_bc": 0.03579336777329445, "train_loss_llm": 8.90493392944336e-05, "grad_norm": 0.04303920269012451, "global_step": 22973, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03311356157064438, "train_loss_bc": 0.03311356157064438, "train_loss_llm": 0.0, "grad_norm": 0.05726932734251022, "global_step": 22974, "epoch": 638, "lr": 0.001} +{"train_loss": 0.025775523856282234, "train_loss_bc": 0.02577214688062668, "train_loss_llm": 0.0003376901149749756, "grad_norm": 0.06612905114889145, "global_step": 22975, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03055255115032196, "train_loss_bc": 0.03055255115032196, "train_loss_llm": 0.0, "grad_norm": 0.07996517419815063, "global_step": 22976, "epoch": 638, "lr": 0.001} +{"train_loss": 0.032184235751628876, "train_loss_bc": 0.03217417001724243, "train_loss_llm": 0.0010066628456115723, "grad_norm": 0.024473484605550766, "global_step": 22977, "epoch": 638, "lr": 0.001} +{"train_loss": 0.031749941408634186, "train_loss_bc": 0.031749941408634186, "train_loss_llm": 0.0, "grad_norm": 0.04142173007130623, "global_step": 22978, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03256332501769066, "train_loss_bc": 0.03256332501769066, "train_loss_llm": 0.0, "grad_norm": 0.055437441915273666, "global_step": 22979, "epoch": 638, "lr": 0.001} +{"train_loss": 0.030451755970716476, "train_loss_bc": 0.030444664880633354, "train_loss_llm": 0.0007090270519256592, "grad_norm": 0.07502435892820358, "global_step": 22980, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03447583317756653, "train_loss_bc": 0.03447583317756653, "train_loss_llm": 0.0, "grad_norm": 0.03510046377778053, "global_step": 22981, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03348392993211746, "train_loss_bc": 0.03348392993211746, "train_loss_llm": 0.0, "grad_norm": 0.050763096660375595, "global_step": 22982, "epoch": 638, "lr": 0.001} +{"train_loss": 0.04217936843633652, "train_loss_bc": 0.04217936843633652, "train_loss_llm": 0.0, "grad_norm": 0.06165274232625961, "global_step": 22983, "epoch": 638, "lr": 0.001} +{"train_loss": 0.02794833295047283, "train_loss_bc": 0.02794833295047283, "train_loss_llm": 0.0, "grad_norm": 0.0784517303109169, "global_step": 22984, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03703271225094795, "train_loss_bc": 0.03703271225094795, "train_loss_llm": 0.0, "grad_norm": 0.03693162277340889, "global_step": 22985, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03593001514673233, "train_loss_bc": 0.03593001514673233, "train_loss_llm": 0.0, "grad_norm": 0.057678770273923874, "global_step": 22986, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03714173659682274, "train_loss_bc": 0.03714173659682274, "train_loss_llm": 0.0, "grad_norm": 0.07723429054021835, "global_step": 22987, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03464309498667717, "train_loss_bc": 0.03464309498667717, "train_loss_llm": 0.0, "grad_norm": 0.10771653801202774, "global_step": 22988, "epoch": 638, "lr": 0.001} +{"train_loss": 0.0310710147023201, "train_loss_bc": 0.031060706824064255, "train_loss_llm": 0.0010307133197784424, "grad_norm": 0.033060360699892044, "global_step": 22989, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03672860562801361, "train_loss_bc": 0.03672860562801361, "train_loss_llm": 0.0, "grad_norm": 0.053786590695381165, "global_step": 22990, "epoch": 638, "lr": 0.001} +{"train_loss": 0.026475021615624428, "train_loss_bc": 0.026475021615624428, "train_loss_llm": 0.0, "grad_norm": 0.0741826742887497, "global_step": 22991, "epoch": 638, "lr": 0.001} +{"train_loss": 0.0343487448990345, "train_loss_bc": 0.03433142974972725, "train_loss_llm": 0.0017316639423370361, "grad_norm": 0.09972234815359116, "global_step": 22992, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03218010440468788, "train_loss_bc": 0.0321674607694149, "train_loss_llm": 0.0012642145156860352, "grad_norm": 0.033582571893930435, "global_step": 22993, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03257375955581665, "train_loss_bc": 0.03257375955581665, "train_loss_llm": 0.0, "grad_norm": 0.05542929098010063, "global_step": 22994, "epoch": 638, "lr": 0.001} +{"train_loss": 0.030292753130197525, "train_loss_bc": 0.030292753130197525, "train_loss_llm": 0.0, "grad_norm": 0.0737030878663063, "global_step": 22995, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03528059646487236, "train_loss_bc": 0.03528059646487236, "train_loss_llm": 0.0, "grad_norm": 0.0917491540312767, "global_step": 22996, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03739732876420021, "train_loss_bc": 0.03739732876420021, "train_loss_llm": 0.0, "grad_norm": 0.03571383282542229, "global_step": 22997, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03387239947915077, "train_loss_bc": 0.03387239947915077, "train_loss_llm": 0.0, "grad_norm": 0.05257035419344902, "global_step": 22998, "epoch": 638, "lr": 0.001} +{"train_loss": 0.026703255251049995, "train_loss_bc": 0.026703255251049995, "train_loss_llm": 0.0, "grad_norm": 0.06961718201637268, "global_step": 22999, "epoch": 638, "lr": 0.001} +{"train_loss": 0.02539791725575924, "train_loss_bc": 0.02539791725575924, "train_loss_llm": 0.0, "grad_norm": 0.0844646692276001, "global_step": 23000, "epoch": 638, "lr": 0.001} +{"train_loss": 0.029696306213736534, "train_loss_bc": 0.029696306213736534, "train_loss_llm": 0.0, "grad_norm": 0.029675813391804695, "global_step": 23001, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03459372743964195, "train_loss_bc": 0.03459372743964195, "train_loss_llm": 0.0, "grad_norm": 0.05098552629351616, "global_step": 23002, "epoch": 638, "lr": 0.001} +{"train_loss": 0.03225516476151016, "train_loss_bc": 0.025672798976302147, "train_loss_llm": 0.0, "grad_norm": 0.0795893445611, "global_step": 23003, "epoch": 638, "lr": 0.001, "val_loss": 0.0849752426147461} +{"train_loss": 0.03422420099377632, "train_loss_bc": 0.03422420099377632, "train_loss_llm": 0.0, "grad_norm": 0.10066615045070648, "global_step": 23004, "epoch": 639, "lr": 0.001} +{"train_loss": 0.027453098446130753, "train_loss_bc": 0.027439486235380173, "train_loss_llm": 0.0013611912727355957, "grad_norm": 0.028375443071126938, "global_step": 23005, "epoch": 639, "lr": 0.001} +{"train_loss": 0.027728551998734474, "train_loss_bc": 0.027728551998734474, "train_loss_llm": 0.0, "grad_norm": 0.05142102763056755, "global_step": 23006, "epoch": 639, "lr": 0.001} +{"train_loss": 0.031175242736935616, "train_loss_bc": 0.031147489324212074, "train_loss_llm": 0.002775341272354126, "grad_norm": 0.0640435516834259, "global_step": 23007, "epoch": 639, "lr": 0.001} +{"train_loss": 0.030527984723448753, "train_loss_bc": 0.030527984723448753, "train_loss_llm": 0.0, "grad_norm": 0.08553077280521393, "global_step": 23008, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03482270985841751, "train_loss_bc": 0.03482270985841751, "train_loss_llm": 0.0, "grad_norm": 0.03132764995098114, "global_step": 23009, "epoch": 639, "lr": 0.001} +{"train_loss": 0.04581483453512192, "train_loss_bc": 0.04581483453512192, "train_loss_llm": 0.0, "grad_norm": 0.05216478183865547, "global_step": 23010, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03411940112709999, "train_loss_bc": 0.03411940112709999, "train_loss_llm": 0.0, "grad_norm": 0.06566085666418076, "global_step": 23011, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03290103003382683, "train_loss_bc": 0.03290103003382683, "train_loss_llm": 0.0, "grad_norm": 0.08283405750989914, "global_step": 23012, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03660782426595688, "train_loss_bc": 0.03659434616565704, "train_loss_llm": 0.0013478994369506836, "grad_norm": 0.03588399663567543, "global_step": 23013, "epoch": 639, "lr": 0.001} +{"train_loss": 0.029661376029253006, "train_loss_bc": 0.029661376029253006, "train_loss_llm": 0.0, "grad_norm": 0.05390075221657753, "global_step": 23014, "epoch": 639, "lr": 0.001} +{"train_loss": 0.035264045000076294, "train_loss_bc": 0.03524990379810333, "train_loss_llm": 0.0014140009880065918, "grad_norm": 0.07186148315668106, "global_step": 23015, "epoch": 639, "lr": 0.001} +{"train_loss": 0.02962217479944229, "train_loss_bc": 0.02962217479944229, "train_loss_llm": 0.0, "grad_norm": 0.07961328327655792, "global_step": 23016, "epoch": 639, "lr": 0.001} +{"train_loss": 0.029101748019456863, "train_loss_bc": 0.029101748019456863, "train_loss_llm": 0.0, "grad_norm": 0.03254641592502594, "global_step": 23017, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03073287568986416, "train_loss_bc": 0.03073287568986416, "train_loss_llm": 0.0, "grad_norm": 0.05152386426925659, "global_step": 23018, "epoch": 639, "lr": 0.001} +{"train_loss": 0.028483988717198372, "train_loss_bc": 0.028483988717198372, "train_loss_llm": 0.0, "grad_norm": 0.061770930886268616, "global_step": 23019, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03385814651846886, "train_loss_bc": 0.03385233134031296, "train_loss_llm": 0.0005816221237182617, "grad_norm": 0.07713346183300018, "global_step": 23020, "epoch": 639, "lr": 0.001} +{"train_loss": 0.024710102006793022, "train_loss_bc": 0.024710102006793022, "train_loss_llm": 0.0, "grad_norm": 0.026333222165703773, "global_step": 23021, "epoch": 639, "lr": 0.001} +{"train_loss": 0.02992410585284233, "train_loss_bc": 0.02992410585284233, "train_loss_llm": 0.0, "grad_norm": 0.04737885296344757, "global_step": 23022, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03482251241803169, "train_loss_bc": 0.03482251241803169, "train_loss_llm": 0.0, "grad_norm": 0.07047636806964874, "global_step": 23023, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03234345465898514, "train_loss_bc": 0.03234345465898514, "train_loss_llm": 0.0, "grad_norm": 0.09656299650669098, "global_step": 23024, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03516363352537155, "train_loss_bc": 0.03516363352537155, "train_loss_llm": 0.0, "grad_norm": 0.034398019313812256, "global_step": 23025, "epoch": 639, "lr": 0.001} +{"train_loss": 0.0302191860973835, "train_loss_bc": 0.0302191860973835, "train_loss_llm": 0.0, "grad_norm": 0.055613502860069275, "global_step": 23026, "epoch": 639, "lr": 0.001} +{"train_loss": 0.025801243260502815, "train_loss_bc": 0.025801243260502815, "train_loss_llm": 0.0, "grad_norm": 0.06527302414178848, "global_step": 23027, "epoch": 639, "lr": 0.001} +{"train_loss": 0.030258044600486755, "train_loss_bc": 0.030258044600486755, "train_loss_llm": 0.0, "grad_norm": 0.09088514745235443, "global_step": 23028, "epoch": 639, "lr": 0.001} +{"train_loss": 0.036413345485925674, "train_loss_bc": 0.036413345485925674, "train_loss_llm": 0.0, "grad_norm": 0.03648366034030914, "global_step": 23029, "epoch": 639, "lr": 0.001} +{"train_loss": 0.036565281450748444, "train_loss_bc": 0.036565281450748444, "train_loss_llm": 0.0, "grad_norm": 0.046087950468063354, "global_step": 23030, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03260290250182152, "train_loss_bc": 0.03258777782320976, "train_loss_llm": 0.0015124380588531494, "grad_norm": 0.07099203020334244, "global_step": 23031, "epoch": 639, "lr": 0.001} +{"train_loss": 0.035779573023319244, "train_loss_bc": 0.035779573023319244, "train_loss_llm": 0.0, "grad_norm": 0.08307640254497528, "global_step": 23032, "epoch": 639, "lr": 0.001} +{"train_loss": 0.031182313337922096, "train_loss_bc": 0.031179675832390785, "train_loss_llm": 0.0002638399600982666, "grad_norm": 0.03219980373978615, "global_step": 23033, "epoch": 639, "lr": 0.001} +{"train_loss": 0.030647363513708115, "train_loss_bc": 0.030647363513708115, "train_loss_llm": 0.0, "grad_norm": 0.05349672958254814, "global_step": 23034, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03579668328166008, "train_loss_bc": 0.03576432541012764, "train_loss_llm": 0.0032359659671783447, "grad_norm": 0.07088030129671097, "global_step": 23035, "epoch": 639, "lr": 0.001} +{"train_loss": 0.03204597905278206, "train_loss_bc": 0.03203543648123741, "train_loss_llm": 0.0010543763637542725, "grad_norm": 0.08321601897478104, "global_step": 23036, "epoch": 639, "lr": 0.001} +{"train_loss": 0.032287321984767914, "train_loss_bc": 0.032287321984767914, "train_loss_llm": 0.0, "grad_norm": 0.024944089353084564, "global_step": 23037, "epoch": 639, "lr": 0.001} +{"train_loss": 0.02621600590646267, "train_loss_bc": 0.026213714852929115, "train_loss_llm": 0.00022906064987182617, "grad_norm": 0.03687220439314842, "global_step": 23038, "epoch": 639, "lr": 0.001} +{"train_loss": 0.032338957871413894, "train_loss_bc": 0.03932419791817665, "train_loss_llm": 0.0, "grad_norm": 0.060794614255428314, "global_step": 23039, "epoch": 639, "lr": 0.001, "val_loss": 0.08407796174287796} +{"train_loss": 0.033200714737176895, "train_loss_bc": 0.033200714737176895, "train_loss_llm": 0.0, "grad_norm": 0.07986543327569962, "global_step": 23040, "epoch": 640, "lr": 0.001} +{"train_loss": 0.029890337958931923, "train_loss_bc": 0.029890337958931923, "train_loss_llm": 0.0, "grad_norm": 0.02701808698475361, "global_step": 23041, "epoch": 640, "lr": 0.001} +{"train_loss": 0.03154300898313522, "train_loss_bc": 0.031524039804935455, "train_loss_llm": 0.0018970072269439697, "grad_norm": 0.04812255874276161, "global_step": 23042, "epoch": 640, "lr": 0.001} +{"train_loss": 0.026803556829690933, "train_loss_bc": 0.026802729815244675, "train_loss_llm": 8.26716423034668e-05, "grad_norm": 0.06188300997018814, "global_step": 23043, "epoch": 640, "lr": 0.001} +{"train_loss": 0.036685001105070114, "train_loss_bc": 0.036685001105070114, "train_loss_llm": 0.0, "grad_norm": 0.08674633502960205, "global_step": 23044, "epoch": 640, "lr": 0.001} +{"train_loss": 0.03814051300287247, "train_loss_bc": 0.038137663155794144, "train_loss_llm": 0.0002849996089935303, "grad_norm": 0.047643642872571945, "global_step": 23045, "epoch": 640, "lr": 0.001} +{"train_loss": 0.03184007853269577, "train_loss_bc": 0.03184007853269577, "train_loss_llm": 0.0, "grad_norm": 0.0715940073132515, "global_step": 23046, "epoch": 640, "lr": 0.001} +{"train_loss": 0.030381642282009125, "train_loss_bc": 0.030381642282009125, "train_loss_llm": 0.0, "grad_norm": 0.09591987729072571, "global_step": 23047, "epoch": 640, "lr": 0.001} +{"train_loss": 0.031907934695482254, "train_loss_bc": 0.031907934695482254, "train_loss_llm": 0.0, "grad_norm": 0.1188230887055397, "global_step": 23048, "epoch": 640, "lr": 0.001} +{"train_loss": 0.022174827754497528, "train_loss_bc": 0.022164886817336082, "train_loss_llm": 0.000994175672531128, "grad_norm": 0.029479311779141426, "global_step": 23049, "epoch": 640, "lr": 0.001} +{"train_loss": 0.02961784228682518, "train_loss_bc": 0.02959497831761837, "train_loss_llm": 0.0022864341735839844, "grad_norm": 0.043393123894929886, "global_step": 23050, "epoch": 640, "lr": 0.001} +{"train_loss": 0.04123447462916374, "train_loss_bc": 0.04123447462916374, "train_loss_llm": 0.0, "grad_norm": 0.06520817428827286, "global_step": 23051, "epoch": 640, "lr": 0.001} +{"train_loss": 0.02953268401324749, "train_loss_bc": 0.02953268401324749, "train_loss_llm": 0.0, "grad_norm": 0.091453917324543, "global_step": 23052, "epoch": 640, "lr": 0.001} +{"train_loss": 0.03330540657043457, "train_loss_bc": 0.03330540657043457, "train_loss_llm": 0.0, "grad_norm": 0.03718290105462074, "global_step": 23053, "epoch": 640, "lr": 0.001} +{"train_loss": 0.030201224610209465, "train_loss_bc": 0.030201224610209465, "train_loss_llm": 0.0, "grad_norm": 0.057410601526498795, "global_step": 23054, "epoch": 640, "lr": 0.001} +{"train_loss": 0.02821946330368519, "train_loss_bc": 0.02821946330368519, "train_loss_llm": 0.0, "grad_norm": 0.08369684219360352, "global_step": 23055, "epoch": 640, "lr": 0.001} +{"train_loss": 0.026897341012954712, "train_loss_bc": 0.026897341012954712, "train_loss_llm": 0.0, "grad_norm": 0.10068516433238983, "global_step": 23056, "epoch": 640, "lr": 0.001} +{"train_loss": 0.027732526883482933, "train_loss_bc": 0.027726495638489723, "train_loss_llm": 0.0006030797958374023, "grad_norm": 0.02949623018503189, "global_step": 23057, "epoch": 640, "lr": 0.001} +{"train_loss": 0.0364946685731411, "train_loss_bc": 0.0364946685731411, "train_loss_llm": 0.0, "grad_norm": 0.055817969143390656, "global_step": 23058, "epoch": 640, "lr": 0.001} +{"train_loss": 0.03581256791949272, "train_loss_bc": 0.03581256791949272, "train_loss_llm": 0.0, "grad_norm": 0.07111398875713348, "global_step": 23059, "epoch": 640, "lr": 0.001} +{"train_loss": 0.03022577613592148, "train_loss_bc": 0.03022577613592148, "train_loss_llm": 0.0, "grad_norm": 0.08939259499311447, "global_step": 23060, "epoch": 640, "lr": 0.001} +{"train_loss": 0.034949012100696564, "train_loss_bc": 0.034949012100696564, "train_loss_llm": 0.0, "grad_norm": 0.03242860734462738, "global_step": 23061, "epoch": 640, "lr": 0.001} +{"train_loss": 0.029535485431551933, "train_loss_bc": 0.029535485431551933, "train_loss_llm": 0.0, "grad_norm": 0.0455678328871727, "global_step": 23062, "epoch": 640, "lr": 0.001} +{"train_loss": 0.036787718534469604, "train_loss_bc": 0.036787718534469604, "train_loss_llm": 0.0, "grad_norm": 0.06848414987325668, "global_step": 23063, "epoch": 640, "lr": 0.001} +{"train_loss": 0.032435595989227295, "train_loss_bc": 0.032435595989227295, "train_loss_llm": 0.0, "grad_norm": 0.08843842148780823, "global_step": 23064, "epoch": 640, "lr": 0.001} +{"train_loss": 0.027981270104646683, "train_loss_bc": 0.02797756716609001, "train_loss_llm": 0.00037020444869995117, "grad_norm": 0.030445711687207222, "global_step": 23065, "epoch": 640, "lr": 0.001} +{"train_loss": 0.035708677023649216, "train_loss_bc": 0.035708677023649216, "train_loss_llm": 0.0, "grad_norm": 0.050081685185432434, "global_step": 23066, "epoch": 640, "lr": 0.001} +{"train_loss": 0.040250781923532486, "train_loss_bc": 0.040250781923532486, "train_loss_llm": 0.0, "grad_norm": 0.0627031922340393, "global_step": 23067, "epoch": 640, "lr": 0.001} +{"train_loss": 0.029557647183537483, "train_loss_bc": 0.029557647183537483, "train_loss_llm": 0.0, "grad_norm": 0.06710086017847061, "global_step": 23068, "epoch": 640, "lr": 0.001} +{"train_loss": 0.02910587191581726, "train_loss_bc": 0.029103580862283707, "train_loss_llm": 0.00022909045219421387, "grad_norm": 0.03723606467247009, "global_step": 23069, "epoch": 640, "lr": 0.001} +{"train_loss": 0.034106362611055374, "train_loss_bc": 0.034106362611055374, "train_loss_llm": 0.0, "grad_norm": 0.054261885583400726, "global_step": 23070, "epoch": 640, "lr": 0.001} +{"train_loss": 0.03492777422070503, "train_loss_bc": 0.03492777422070503, "train_loss_llm": 0.0, "grad_norm": 0.08241203427314758, "global_step": 23071, "epoch": 640, "lr": 0.001} +{"train_loss": 0.030297711491584778, "train_loss_bc": 0.030297711491584778, "train_loss_llm": 0.0, "grad_norm": 0.10563549399375916, "global_step": 23072, "epoch": 640, "lr": 0.001} +{"train_loss": 0.03288135305047035, "train_loss_bc": 0.03288135305047035, "train_loss_llm": 0.0, "grad_norm": 0.03195646032691002, "global_step": 23073, "epoch": 640, "lr": 0.001} +{"train_loss": 0.03797433525323868, "train_loss_bc": 0.03797433525323868, "train_loss_llm": 0.0, "grad_norm": 0.05695980042219162, "global_step": 23074, "epoch": 640, "lr": 0.001} +{"train_loss": 0.03252320338247551, "train_loss_bc": 0.04249413311481476, "train_loss_llm": 0.0, "grad_norm": 0.0957711860537529, "global_step": 23075, "epoch": 640, "lr": 0.001, "train/cumulative_reward": 55.34782794418618, "train/mean_score": 13.557223290336514, "train/success_rate": 0.4, "test/cumulative_reward": 47.94202193621998, "test/mean_score": 9.449287742754375, "test/success_rate": 0.2, "val_loss": 0.08563597500324249, "train_action_mse_error": 0.036960992962121964} +{"train_loss": 0.03139641135931015, "train_loss_bc": 0.03139641135931015, "train_loss_llm": 0.0, "grad_norm": 0.12424896657466888, "global_step": 23076, "epoch": 641, "lr": 0.001} +{"train_loss": 0.027959562838077545, "train_loss_bc": 0.027959562838077545, "train_loss_llm": 0.0, "grad_norm": 0.0315777026116848, "global_step": 23077, "epoch": 641, "lr": 0.001} +{"train_loss": 0.028053410351276398, "train_loss_bc": 0.028053410351276398, "train_loss_llm": 0.0, "grad_norm": 0.04777849465608597, "global_step": 23078, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03324302285909653, "train_loss_bc": 0.03324302285909653, "train_loss_llm": 0.0, "grad_norm": 0.06454736739397049, "global_step": 23079, "epoch": 641, "lr": 0.001} +{"train_loss": 0.032187122851610184, "train_loss_bc": 0.032159049063920975, "train_loss_llm": 0.002807408571243286, "grad_norm": 0.08438495546579361, "global_step": 23080, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03146253153681755, "train_loss_bc": 0.03146253153681755, "train_loss_llm": 0.0, "grad_norm": 0.0315413773059845, "global_step": 23081, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03868329897522926, "train_loss_bc": 0.03868329897522926, "train_loss_llm": 0.0, "grad_norm": 0.0511389784514904, "global_step": 23082, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03317989036440849, "train_loss_bc": 0.033179737627506256, "train_loss_llm": 1.5139579772949219e-05, "grad_norm": 0.062030620872974396, "global_step": 23083, "epoch": 641, "lr": 0.001} +{"train_loss": 0.038879603147506714, "train_loss_bc": 0.038879603147506714, "train_loss_llm": 0.0, "grad_norm": 0.08453427255153656, "global_step": 23084, "epoch": 641, "lr": 0.001} +{"train_loss": 0.027000579982995987, "train_loss_bc": 0.027000579982995987, "train_loss_llm": 0.0, "grad_norm": 0.03539082407951355, "global_step": 23085, "epoch": 641, "lr": 0.001} +{"train_loss": 0.031853120774030685, "train_loss_bc": 0.031853120774030685, "train_loss_llm": 0.0, "grad_norm": 0.05534270778298378, "global_step": 23086, "epoch": 641, "lr": 0.001} +{"train_loss": 0.026992227882146835, "train_loss_bc": 0.026992227882146835, "train_loss_llm": 0.0, "grad_norm": 0.06765425205230713, "global_step": 23087, "epoch": 641, "lr": 0.001} +{"train_loss": 0.034884754568338394, "train_loss_bc": 0.034874726086854935, "train_loss_llm": 0.0010026991367340088, "grad_norm": 0.09994203597307205, "global_step": 23088, "epoch": 641, "lr": 0.001} +{"train_loss": 0.035168036818504333, "train_loss_bc": 0.035168036818504333, "train_loss_llm": 0.0, "grad_norm": 0.030660931020975113, "global_step": 23089, "epoch": 641, "lr": 0.001} +{"train_loss": 0.029352247714996338, "train_loss_bc": 0.029352247714996338, "train_loss_llm": 0.0, "grad_norm": 0.04636559635400772, "global_step": 23090, "epoch": 641, "lr": 0.001} +{"train_loss": 0.0345430001616478, "train_loss_bc": 0.0345430001616478, "train_loss_llm": 0.0, "grad_norm": 0.06485272198915482, "global_step": 23091, "epoch": 641, "lr": 0.001} +{"train_loss": 0.0285577904433012, "train_loss_bc": 0.028536679223179817, "train_loss_llm": 0.0021111369132995605, "grad_norm": 0.07966786623001099, "global_step": 23092, "epoch": 641, "lr": 0.001} +{"train_loss": 0.027763016521930695, "train_loss_bc": 0.027763016521930695, "train_loss_llm": 0.0, "grad_norm": 0.03424978628754616, "global_step": 23093, "epoch": 641, "lr": 0.001} +{"train_loss": 0.0347275510430336, "train_loss_bc": 0.0347275510430336, "train_loss_llm": 0.0, "grad_norm": 0.05657690390944481, "global_step": 23094, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03762363642454147, "train_loss_bc": 0.03762363642454147, "train_loss_llm": 0.0, "grad_norm": 0.08066175878047943, "global_step": 23095, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03191077709197998, "train_loss_bc": 0.031899407505989075, "train_loss_llm": 0.0011368989944458008, "grad_norm": 0.09042626619338989, "global_step": 23096, "epoch": 641, "lr": 0.001} +{"train_loss": 0.02925211377441883, "train_loss_bc": 0.02925211377441883, "train_loss_llm": 0.0, "grad_norm": 0.02832547016441822, "global_step": 23097, "epoch": 641, "lr": 0.001} +{"train_loss": 0.030579596757888794, "train_loss_bc": 0.030579596757888794, "train_loss_llm": 0.0, "grad_norm": 0.04304085671901703, "global_step": 23098, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03246695175766945, "train_loss_bc": 0.03246695175766945, "train_loss_llm": 0.0, "grad_norm": 0.05788418650627136, "global_step": 23099, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03442183509469032, "train_loss_bc": 0.03442183509469032, "train_loss_llm": 0.0, "grad_norm": 0.07179215550422668, "global_step": 23100, "epoch": 641, "lr": 0.001} +{"train_loss": 0.0289569403976202, "train_loss_bc": 0.0289569403976202, "train_loss_llm": 0.0, "grad_norm": 0.03381224349141121, "global_step": 23101, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03602197393774986, "train_loss_bc": 0.036012016236782074, "train_loss_llm": 0.0009958744049072266, "grad_norm": 0.05575660988688469, "global_step": 23102, "epoch": 641, "lr": 0.001} +{"train_loss": 0.032965607941150665, "train_loss_bc": 0.032965607941150665, "train_loss_llm": 0.0, "grad_norm": 0.068633072078228, "global_step": 23103, "epoch": 641, "lr": 0.001} +{"train_loss": 0.028696861118078232, "train_loss_bc": 0.028696861118078232, "train_loss_llm": 0.0, "grad_norm": 0.08557892590761185, "global_step": 23104, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03473448008298874, "train_loss_bc": 0.03473448008298874, "train_loss_llm": 0.0, "grad_norm": 0.03230865299701691, "global_step": 23105, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03599689155817032, "train_loss_bc": 0.03599689155817032, "train_loss_llm": 0.0, "grad_norm": 0.0514407604932785, "global_step": 23106, "epoch": 641, "lr": 0.001} +{"train_loss": 0.030439266934990883, "train_loss_bc": 0.030439266934990883, "train_loss_llm": 0.0, "grad_norm": 0.06793249398469925, "global_step": 23107, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03210920840501785, "train_loss_bc": 0.03210316598415375, "train_loss_llm": 0.0006040632724761963, "grad_norm": 0.08394508063793182, "global_step": 23108, "epoch": 641, "lr": 0.001} +{"train_loss": 0.032422251999378204, "train_loss_bc": 0.032422251999378204, "train_loss_llm": 0.0, "grad_norm": 0.029567088931798935, "global_step": 23109, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03459669277071953, "train_loss_bc": 0.03459669277071953, "train_loss_llm": 0.0, "grad_norm": 0.04495830461382866, "global_step": 23110, "epoch": 641, "lr": 0.001} +{"train_loss": 0.03223837767210272, "train_loss_bc": 0.03149932995438576, "train_loss_llm": 0.0, "grad_norm": 0.06332679092884064, "global_step": 23111, "epoch": 641, "lr": 0.001, "val_loss": 0.0852336436510086} +{"train_loss": 0.03144142031669617, "train_loss_bc": 0.03144142031669617, "train_loss_llm": 0.0, "grad_norm": 0.07892455160617828, "global_step": 23112, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03239907696843147, "train_loss_bc": 0.03239907696843147, "train_loss_llm": 0.0, "grad_norm": 0.03511270508170128, "global_step": 23113, "epoch": 642, "lr": 0.001} +{"train_loss": 0.032704487442970276, "train_loss_bc": 0.032704487442970276, "train_loss_llm": 0.0, "grad_norm": 0.05561862885951996, "global_step": 23114, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03303966298699379, "train_loss_bc": 0.03303966298699379, "train_loss_llm": 0.0, "grad_norm": 0.06284662336111069, "global_step": 23115, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03099805861711502, "train_loss_bc": 0.03099805861711502, "train_loss_llm": 0.0, "grad_norm": 0.06725280731916428, "global_step": 23116, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03964761272072792, "train_loss_bc": 0.03964761272072792, "train_loss_llm": 0.0, "grad_norm": 0.03582257404923439, "global_step": 23117, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03245534002780914, "train_loss_bc": 0.03245534002780914, "train_loss_llm": 0.0, "grad_norm": 0.057399146258831024, "global_step": 23118, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03472289815545082, "train_loss_bc": 0.03471171855926514, "train_loss_llm": 0.0011178851127624512, "grad_norm": 0.07896002382040024, "global_step": 23119, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03116110898554325, "train_loss_bc": 0.03116110898554325, "train_loss_llm": 0.0, "grad_norm": 0.08789440244436264, "global_step": 23120, "epoch": 642, "lr": 0.001} +{"train_loss": 0.028701037168502808, "train_loss_bc": 0.028701037168502808, "train_loss_llm": 0.0, "grad_norm": 0.028260407969355583, "global_step": 23121, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03477199003100395, "train_loss_bc": 0.03477199003100395, "train_loss_llm": 0.0, "grad_norm": 0.058170244097709656, "global_step": 23122, "epoch": 642, "lr": 0.001} +{"train_loss": 0.032215990126132965, "train_loss_bc": 0.03221292421221733, "train_loss_llm": 0.000306546688079834, "grad_norm": 0.07848263531923294, "global_step": 23123, "epoch": 642, "lr": 0.001} +{"train_loss": 0.026168642565608025, "train_loss_bc": 0.026168642565608025, "train_loss_llm": 0.0, "grad_norm": 0.0889447033405304, "global_step": 23124, "epoch": 642, "lr": 0.001} +{"train_loss": 0.02979324385523796, "train_loss_bc": 0.02979324385523796, "train_loss_llm": 0.0, "grad_norm": 0.028623076155781746, "global_step": 23125, "epoch": 642, "lr": 0.001} +{"train_loss": 0.02616969682276249, "train_loss_bc": 0.02616969682276249, "train_loss_llm": 0.0, "grad_norm": 0.04235946387052536, "global_step": 23126, "epoch": 642, "lr": 0.001} +{"train_loss": 0.030385633930563927, "train_loss_bc": 0.030374668538570404, "train_loss_llm": 0.0010965466499328613, "grad_norm": 0.0598282590508461, "global_step": 23127, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03526467829942703, "train_loss_bc": 0.03526467829942703, "train_loss_llm": 0.0, "grad_norm": 0.08158107101917267, "global_step": 23128, "epoch": 642, "lr": 0.001} +{"train_loss": 0.031015153974294662, "train_loss_bc": 0.031015153974294662, "train_loss_llm": 0.0, "grad_norm": 0.03565814346075058, "global_step": 23129, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03466605395078659, "train_loss_bc": 0.03466605395078659, "train_loss_llm": 0.0, "grad_norm": 0.04912788048386574, "global_step": 23130, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03553301841020584, "train_loss_bc": 0.035519521683454514, "train_loss_llm": 0.0013497769832611084, "grad_norm": 0.07291106134653091, "global_step": 23131, "epoch": 642, "lr": 0.001} +{"train_loss": 0.026896649971604347, "train_loss_bc": 0.026887377724051476, "train_loss_llm": 0.0009272396564483643, "grad_norm": 0.09118951857089996, "global_step": 23132, "epoch": 642, "lr": 0.001} +{"train_loss": 0.031133387237787247, "train_loss_bc": 0.031120670959353447, "train_loss_llm": 0.0012716352939605713, "grad_norm": 0.02832181006669998, "global_step": 23133, "epoch": 642, "lr": 0.001} +{"train_loss": 0.030637158080935478, "train_loss_bc": 0.030637158080935478, "train_loss_llm": 0.0, "grad_norm": 0.04519061744213104, "global_step": 23134, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03283727914094925, "train_loss_bc": 0.03283727914094925, "train_loss_llm": 0.0, "grad_norm": 0.06239672005176544, "global_step": 23135, "epoch": 642, "lr": 0.001} +{"train_loss": 0.029284900054335594, "train_loss_bc": 0.029284900054335594, "train_loss_llm": 0.0, "grad_norm": 0.07971415668725967, "global_step": 23136, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03441435098648071, "train_loss_bc": 0.03438311442732811, "train_loss_llm": 0.003123641014099121, "grad_norm": 0.0314667634665966, "global_step": 23137, "epoch": 642, "lr": 0.001} +{"train_loss": 0.02936752326786518, "train_loss_bc": 0.02936752326786518, "train_loss_llm": 0.0, "grad_norm": 0.0484851598739624, "global_step": 23138, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03781170770525932, "train_loss_bc": 0.03780992701649666, "train_loss_llm": 0.0001780092716217041, "grad_norm": 0.06925107538700104, "global_step": 23139, "epoch": 642, "lr": 0.001} +{"train_loss": 0.031130239367485046, "train_loss_bc": 0.031130239367485046, "train_loss_llm": 0.0, "grad_norm": 0.08522819727659225, "global_step": 23140, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03275888413190842, "train_loss_bc": 0.03275499492883682, "train_loss_llm": 0.0003890097141265869, "grad_norm": 0.025782929733395576, "global_step": 23141, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03675379976630211, "train_loss_bc": 0.03675379976630211, "train_loss_llm": 0.0, "grad_norm": 0.04498350992798805, "global_step": 23142, "epoch": 642, "lr": 0.001} +{"train_loss": 0.028588788583874702, "train_loss_bc": 0.028564030304551125, "train_loss_llm": 0.002475738525390625, "grad_norm": 0.060921575874090195, "global_step": 23143, "epoch": 642, "lr": 0.001} +{"train_loss": 0.029711531475186348, "train_loss_bc": 0.029711531475186348, "train_loss_llm": 0.0, "grad_norm": 0.0694713443517685, "global_step": 23144, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03199723735451698, "train_loss_bc": 0.03199723735451698, "train_loss_llm": 0.0, "grad_norm": 0.03803597763180733, "global_step": 23145, "epoch": 642, "lr": 0.001} +{"train_loss": 0.04137428477406502, "train_loss_bc": 0.04137428477406502, "train_loss_llm": 0.0, "grad_norm": 0.06344912946224213, "global_step": 23146, "epoch": 642, "lr": 0.001} +{"train_loss": 0.03220031058622731, "train_loss_bc": 0.03125865384936333, "train_loss_llm": 0.0, "grad_norm": 0.09454020857810974, "global_step": 23147, "epoch": 642, "lr": 0.001, "val_loss": 0.08597314357757568} +{"train_loss": 0.028406906872987747, "train_loss_bc": 0.028406906872987747, "train_loss_llm": 0.0, "grad_norm": 0.11210896074771881, "global_step": 23148, "epoch": 643, "lr": 0.001} +{"train_loss": 0.0361933559179306, "train_loss_bc": 0.03615301474928856, "train_loss_llm": 0.004034161567687988, "grad_norm": 0.0385453961789608, "global_step": 23149, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03298119828104973, "train_loss_bc": 0.03298119828104973, "train_loss_llm": 0.0, "grad_norm": 0.05381527915596962, "global_step": 23150, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03170708939433098, "train_loss_bc": 0.03170708939433098, "train_loss_llm": 0.0, "grad_norm": 0.07361616939306259, "global_step": 23151, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03245539218187332, "train_loss_bc": 0.03245539218187332, "train_loss_llm": 0.0, "grad_norm": 0.0937853753566742, "global_step": 23152, "epoch": 643, "lr": 0.001} +{"train_loss": 0.029638834297657013, "train_loss_bc": 0.029638834297657013, "train_loss_llm": 0.0, "grad_norm": 0.02488267607986927, "global_step": 23153, "epoch": 643, "lr": 0.001} +{"train_loss": 0.034974683076143265, "train_loss_bc": 0.034974683076143265, "train_loss_llm": 0.0, "grad_norm": 0.04271922633051872, "global_step": 23154, "epoch": 643, "lr": 0.001} +{"train_loss": 0.028952764347195625, "train_loss_bc": 0.028952764347195625, "train_loss_llm": 0.0, "grad_norm": 0.05401110276579857, "global_step": 23155, "epoch": 643, "lr": 0.001} +{"train_loss": 0.029357347637414932, "train_loss_bc": 0.029357347637414932, "train_loss_llm": 0.0, "grad_norm": 0.0689735934138298, "global_step": 23156, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03253777697682381, "train_loss_bc": 0.03253777697682381, "train_loss_llm": 0.0, "grad_norm": 0.02691132389008999, "global_step": 23157, "epoch": 643, "lr": 0.001} +{"train_loss": 0.030237041413784027, "train_loss_bc": 0.03021496720612049, "train_loss_llm": 0.0022073686122894287, "grad_norm": 0.04125013202428818, "global_step": 23158, "epoch": 643, "lr": 0.001} +{"train_loss": 0.030446186661720276, "train_loss_bc": 0.030432909727096558, "train_loss_llm": 0.0013277530670166016, "grad_norm": 0.04930489882826805, "global_step": 23159, "epoch": 643, "lr": 0.001} +{"train_loss": 0.028615590184926987, "train_loss_bc": 0.028615590184926987, "train_loss_llm": 0.0, "grad_norm": 0.05980938300490379, "global_step": 23160, "epoch": 643, "lr": 0.001} +{"train_loss": 0.028289319947361946, "train_loss_bc": 0.028289319947361946, "train_loss_llm": 0.0, "grad_norm": 0.029540902003645897, "global_step": 23161, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03289681673049927, "train_loss_bc": 0.03289681673049927, "train_loss_llm": 0.0, "grad_norm": 0.04666246846318245, "global_step": 23162, "epoch": 643, "lr": 0.001} +{"train_loss": 0.033969055861234665, "train_loss_bc": 0.033969055861234665, "train_loss_llm": 0.0, "grad_norm": 0.06154385581612587, "global_step": 23163, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03374752402305603, "train_loss_bc": 0.03374752402305603, "train_loss_llm": 0.0, "grad_norm": 0.0787472054362297, "global_step": 23164, "epoch": 643, "lr": 0.001} +{"train_loss": 0.028763268142938614, "train_loss_bc": 0.028763268142938614, "train_loss_llm": 0.0, "grad_norm": 0.03107677586376667, "global_step": 23165, "epoch": 643, "lr": 0.001} +{"train_loss": 0.025347361341118813, "train_loss_bc": 0.025335730984807014, "train_loss_llm": 0.0011631250381469727, "grad_norm": 0.048793476074934006, "global_step": 23166, "epoch": 643, "lr": 0.001} +{"train_loss": 0.031870342791080475, "train_loss_bc": 0.031870342791080475, "train_loss_llm": 0.0, "grad_norm": 0.060722991824150085, "global_step": 23167, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03627926483750343, "train_loss_bc": 0.03624464198946953, "train_loss_llm": 0.0034624338150024414, "grad_norm": 0.08275964111089706, "global_step": 23168, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03464218229055405, "train_loss_bc": 0.03464218229055405, "train_loss_llm": 0.0, "grad_norm": 0.034441988915205, "global_step": 23169, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03740563616156578, "train_loss_bc": 0.03740563616156578, "train_loss_llm": 0.0, "grad_norm": 0.06166263297200203, "global_step": 23170, "epoch": 643, "lr": 0.001} +{"train_loss": 0.026546921581029892, "train_loss_bc": 0.026527801528573036, "train_loss_llm": 0.0019119977951049805, "grad_norm": 0.07471157610416412, "global_step": 23171, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03211900591850281, "train_loss_bc": 0.03211900591850281, "train_loss_llm": 0.0, "grad_norm": 0.08465605229139328, "global_step": 23172, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03140159323811531, "train_loss_bc": 0.03140159323811531, "train_loss_llm": 0.0, "grad_norm": 0.02981904335319996, "global_step": 23173, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03108019195497036, "train_loss_bc": 0.03108019195497036, "train_loss_llm": 0.0, "grad_norm": 0.04143938794732094, "global_step": 23174, "epoch": 643, "lr": 0.001} +{"train_loss": 0.02632910944521427, "train_loss_bc": 0.02632910944521427, "train_loss_llm": 0.0, "grad_norm": 0.05291257053613663, "global_step": 23175, "epoch": 643, "lr": 0.001} +{"train_loss": 0.04374304413795471, "train_loss_bc": 0.04374304413795471, "train_loss_llm": 0.0, "grad_norm": 0.08115247637033463, "global_step": 23176, "epoch": 643, "lr": 0.001} +{"train_loss": 0.02682698518037796, "train_loss_bc": 0.026819583028554916, "train_loss_llm": 0.0007401704788208008, "grad_norm": 0.024646997451782227, "global_step": 23177, "epoch": 643, "lr": 0.001} +{"train_loss": 0.031153570860624313, "train_loss_bc": 0.031153570860624313, "train_loss_llm": 0.0, "grad_norm": 0.04772738367319107, "global_step": 23178, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03148188814520836, "train_loss_bc": 0.03148188814520836, "train_loss_llm": 0.0, "grad_norm": 0.07432222366333008, "global_step": 23179, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03581177443265915, "train_loss_bc": 0.03581177443265915, "train_loss_llm": 0.0, "grad_norm": 0.09283328056335449, "global_step": 23180, "epoch": 643, "lr": 0.001} +{"train_loss": 0.0341297872364521, "train_loss_bc": 0.0341297872364521, "train_loss_llm": 0.0, "grad_norm": 0.034452732652425766, "global_step": 23181, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03904394805431366, "train_loss_bc": 0.03904394805431366, "train_loss_llm": 0.0, "grad_norm": 0.05985612794756889, "global_step": 23182, "epoch": 643, "lr": 0.001} +{"train_loss": 0.03215170460235742, "train_loss_bc": 0.03807860612869263, "train_loss_llm": 0.0, "grad_norm": 0.0871410220861435, "global_step": 23183, "epoch": 643, "lr": 0.001, "val_loss": 0.08494344353675842} +{"train_loss": 0.03212412819266319, "train_loss_bc": 0.03210408613085747, "train_loss_llm": 0.002004176378250122, "grad_norm": 0.09953297674655914, "global_step": 23184, "epoch": 644, "lr": 0.001} +{"train_loss": 0.026877643540501595, "train_loss_bc": 0.026877643540501595, "train_loss_llm": 0.0, "grad_norm": 0.03209814056754112, "global_step": 23185, "epoch": 644, "lr": 0.001} +{"train_loss": 0.029061542823910713, "train_loss_bc": 0.029061542823910713, "train_loss_llm": 0.0, "grad_norm": 0.04943566024303436, "global_step": 23186, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03126097470521927, "train_loss_bc": 0.03126097470521927, "train_loss_llm": 0.0, "grad_norm": 0.06658756732940674, "global_step": 23187, "epoch": 644, "lr": 0.001} +{"train_loss": 0.02984306402504444, "train_loss_bc": 0.02984306402504444, "train_loss_llm": 0.0, "grad_norm": 0.07967453449964523, "global_step": 23188, "epoch": 644, "lr": 0.001} +{"train_loss": 0.032776378095149994, "train_loss_bc": 0.032776378095149994, "train_loss_llm": 0.0, "grad_norm": 0.02787957340478897, "global_step": 23189, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03094692900776863, "train_loss_bc": 0.030942656099796295, "train_loss_llm": 0.0004273653030395508, "grad_norm": 0.03984976187348366, "global_step": 23190, "epoch": 644, "lr": 0.001} +{"train_loss": 0.030040789395570755, "train_loss_bc": 0.030040789395570755, "train_loss_llm": 0.0, "grad_norm": 0.05537237226963043, "global_step": 23191, "epoch": 644, "lr": 0.001} +{"train_loss": 0.029661275446414948, "train_loss_bc": 0.029661275446414948, "train_loss_llm": 0.0, "grad_norm": 0.05788023769855499, "global_step": 23192, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03204088658094406, "train_loss_bc": 0.03204088658094406, "train_loss_llm": 0.0, "grad_norm": 0.035709187388420105, "global_step": 23193, "epoch": 644, "lr": 0.001} +{"train_loss": 0.028228847309947014, "train_loss_bc": 0.028227748349308968, "train_loss_llm": 0.00010988116264343262, "grad_norm": 0.05661870539188385, "global_step": 23194, "epoch": 644, "lr": 0.001} +{"train_loss": 0.030850065872073174, "train_loss_bc": 0.03084484115242958, "train_loss_llm": 0.0005225539207458496, "grad_norm": 0.06615055352449417, "global_step": 23195, "epoch": 644, "lr": 0.001} +{"train_loss": 0.032864250242710114, "train_loss_bc": 0.032864250242710114, "train_loss_llm": 0.0, "grad_norm": 0.07931206375360489, "global_step": 23196, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03267163783311844, "train_loss_bc": 0.03267163783311844, "train_loss_llm": 0.0, "grad_norm": 0.04093611612915993, "global_step": 23197, "epoch": 644, "lr": 0.001} +{"train_loss": 0.0365806519985199, "train_loss_bc": 0.0365806519985199, "train_loss_llm": 0.0, "grad_norm": 0.07684766501188278, "global_step": 23198, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03369302302598953, "train_loss_bc": 0.03369302302598953, "train_loss_llm": 0.0, "grad_norm": 0.0803651437163353, "global_step": 23199, "epoch": 644, "lr": 0.001} +{"train_loss": 0.0356568768620491, "train_loss_bc": 0.0356568768620491, "train_loss_llm": 0.0, "grad_norm": 0.11277693510055542, "global_step": 23200, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03408903628587723, "train_loss_bc": 0.03408903628587723, "train_loss_llm": 0.0, "grad_norm": 0.032498233020305634, "global_step": 23201, "epoch": 644, "lr": 0.001} +{"train_loss": 0.037054844200611115, "train_loss_bc": 0.037054844200611115, "train_loss_llm": 0.0, "grad_norm": 0.050783250480890274, "global_step": 23202, "epoch": 644, "lr": 0.001} +{"train_loss": 0.031612854450941086, "train_loss_bc": 0.031612854450941086, "train_loss_llm": 0.0, "grad_norm": 0.0743623748421669, "global_step": 23203, "epoch": 644, "lr": 0.001} +{"train_loss": 0.031142186373472214, "train_loss_bc": 0.031142186373472214, "train_loss_llm": 0.0, "grad_norm": 0.09326517581939697, "global_step": 23204, "epoch": 644, "lr": 0.001} +{"train_loss": 0.028980860486626625, "train_loss_bc": 0.028980860486626625, "train_loss_llm": 0.0, "grad_norm": 0.0288739912211895, "global_step": 23205, "epoch": 644, "lr": 0.001} +{"train_loss": 0.038289979100227356, "train_loss_bc": 0.038289979100227356, "train_loss_llm": 0.0, "grad_norm": 0.053006671369075775, "global_step": 23206, "epoch": 644, "lr": 0.001} +{"train_loss": 0.028820646926760674, "train_loss_bc": 0.028820646926760674, "train_loss_llm": 0.0, "grad_norm": 0.07395625859498978, "global_step": 23207, "epoch": 644, "lr": 0.001} +{"train_loss": 0.0305558443069458, "train_loss_bc": 0.0305558443069458, "train_loss_llm": 0.0, "grad_norm": 0.0811176523566246, "global_step": 23208, "epoch": 644, "lr": 0.001} +{"train_loss": 0.0308225080370903, "train_loss_bc": 0.0308225080370903, "train_loss_llm": 0.0, "grad_norm": 0.03748827055096626, "global_step": 23209, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03412298113107681, "train_loss_bc": 0.03412298113107681, "train_loss_llm": 0.0, "grad_norm": 0.04690501466393471, "global_step": 23210, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03274591639637947, "train_loss_bc": 0.03274591639637947, "train_loss_llm": 0.0, "grad_norm": 0.06057966127991676, "global_step": 23211, "epoch": 644, "lr": 0.001} +{"train_loss": 0.031727347522974014, "train_loss_bc": 0.031727347522974014, "train_loss_llm": 0.0, "grad_norm": 0.08085068315267563, "global_step": 23212, "epoch": 644, "lr": 0.001} +{"train_loss": 0.034510012716054916, "train_loss_bc": 0.034510012716054916, "train_loss_llm": 0.0, "grad_norm": 0.030591551214456558, "global_step": 23213, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03398607298731804, "train_loss_bc": 0.03397895023226738, "train_loss_llm": 0.0007121562957763672, "grad_norm": 0.04429199919104576, "global_step": 23214, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03806610032916069, "train_loss_bc": 0.038057275116443634, "train_loss_llm": 0.0008826255798339844, "grad_norm": 0.058063920587301254, "global_step": 23215, "epoch": 644, "lr": 0.001} +{"train_loss": 0.033770833164453506, "train_loss_bc": 0.033770833164453506, "train_loss_llm": 0.0, "grad_norm": 0.06220678985118866, "global_step": 23216, "epoch": 644, "lr": 0.001} +{"train_loss": 0.026526229456067085, "train_loss_bc": 0.026526229456067085, "train_loss_llm": 0.0, "grad_norm": 0.025992820039391518, "global_step": 23217, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03136196732521057, "train_loss_bc": 0.03136016055941582, "train_loss_llm": 0.00018078088760375977, "grad_norm": 0.04345312714576721, "global_step": 23218, "epoch": 644, "lr": 0.001} +{"train_loss": 0.03205899444098274, "train_loss_bc": 0.030758613720536232, "train_loss_llm": 0.0, "grad_norm": 0.06941661238670349, "global_step": 23219, "epoch": 644, "lr": 0.001, "val_loss": 0.0854264348745346} +{"train_loss": 0.03267133980989456, "train_loss_bc": 0.03267133980989456, "train_loss_llm": 0.0, "grad_norm": 0.08107315003871918, "global_step": 23220, "epoch": 645, "lr": 0.001} +{"train_loss": 0.030764825642108917, "train_loss_bc": 0.03075951710343361, "train_loss_llm": 0.0005307793617248535, "grad_norm": 0.033796876668930054, "global_step": 23221, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03173329681158066, "train_loss_bc": 0.03172345831990242, "train_loss_llm": 0.0009839534759521484, "grad_norm": 0.05497639998793602, "global_step": 23222, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03483401983976364, "train_loss_bc": 0.03483401983976364, "train_loss_llm": 0.0, "grad_norm": 0.07330670952796936, "global_step": 23223, "epoch": 645, "lr": 0.001} +{"train_loss": 0.029546324163675308, "train_loss_bc": 0.029546324163675308, "train_loss_llm": 0.0, "grad_norm": 0.08345034718513489, "global_step": 23224, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03284968435764313, "train_loss_bc": 0.03284968435764313, "train_loss_llm": 0.0, "grad_norm": 0.03362007066607475, "global_step": 23225, "epoch": 645, "lr": 0.001} +{"train_loss": 0.028408825397491455, "train_loss_bc": 0.02840539440512657, "train_loss_llm": 0.00034302473068237305, "grad_norm": 0.04754794016480446, "global_step": 23226, "epoch": 645, "lr": 0.001} +{"train_loss": 0.02520846575498581, "train_loss_bc": 0.02520846575498581, "train_loss_llm": 0.0, "grad_norm": 0.0639021173119545, "global_step": 23227, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03580179810523987, "train_loss_bc": 0.03579168766736984, "train_loss_llm": 0.0010111331939697266, "grad_norm": 0.09439900517463684, "global_step": 23228, "epoch": 645, "lr": 0.001} +{"train_loss": 0.032152559608221054, "train_loss_bc": 0.0321514829993248, "train_loss_llm": 0.00010752677917480469, "grad_norm": 0.028057808056473732, "global_step": 23229, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03464987501502037, "train_loss_bc": 0.034630436450242996, "train_loss_llm": 0.0019439160823822021, "grad_norm": 0.040102988481521606, "global_step": 23230, "epoch": 645, "lr": 0.001} +{"train_loss": 0.032860372215509415, "train_loss_bc": 0.032860372215509415, "train_loss_llm": 0.0, "grad_norm": 0.04881583899259567, "global_step": 23231, "epoch": 645, "lr": 0.001} +{"train_loss": 0.030664268881082535, "train_loss_bc": 0.030658092349767685, "train_loss_llm": 0.000617682933807373, "grad_norm": 0.0579400472342968, "global_step": 23232, "epoch": 645, "lr": 0.001} +{"train_loss": 0.0332651324570179, "train_loss_bc": 0.0332651324570179, "train_loss_llm": 0.0, "grad_norm": 0.025770219042897224, "global_step": 23233, "epoch": 645, "lr": 0.001} +{"train_loss": 0.029112467542290688, "train_loss_bc": 0.029112467542290688, "train_loss_llm": 0.0, "grad_norm": 0.03733595833182335, "global_step": 23234, "epoch": 645, "lr": 0.001} +{"train_loss": 0.030201120302081108, "train_loss_bc": 0.030201120302081108, "train_loss_llm": 0.0, "grad_norm": 0.05297474190592766, "global_step": 23235, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03224830701947212, "train_loss_bc": 0.03224830701947212, "train_loss_llm": 0.0, "grad_norm": 0.07174273580312729, "global_step": 23236, "epoch": 645, "lr": 0.001} +{"train_loss": 0.034636240452528, "train_loss_bc": 0.034636240452528, "train_loss_llm": 0.0, "grad_norm": 0.0251909252256155, "global_step": 23237, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03422676771879196, "train_loss_bc": 0.03422676771879196, "train_loss_llm": 0.0, "grad_norm": 0.0491766519844532, "global_step": 23238, "epoch": 645, "lr": 0.001} +{"train_loss": 0.029318805783987045, "train_loss_bc": 0.029318805783987045, "train_loss_llm": 0.0, "grad_norm": 0.06634920835494995, "global_step": 23239, "epoch": 645, "lr": 0.001} +{"train_loss": 0.0349106602370739, "train_loss_bc": 0.0349106602370739, "train_loss_llm": 0.0, "grad_norm": 0.07307597994804382, "global_step": 23240, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03052632510662079, "train_loss_bc": 0.03052632510662079, "train_loss_llm": 0.0, "grad_norm": 0.03157445788383484, "global_step": 23241, "epoch": 645, "lr": 0.001} +{"train_loss": 0.027970893308520317, "train_loss_bc": 0.027970893308520317, "train_loss_llm": 0.0, "grad_norm": 0.05418282002210617, "global_step": 23242, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03908262029290199, "train_loss_bc": 0.03908172622323036, "train_loss_llm": 8.928775787353516e-05, "grad_norm": 0.08086865395307541, "global_step": 23243, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03025033324956894, "train_loss_bc": 0.03025033324956894, "train_loss_llm": 0.0, "grad_norm": 0.08761224150657654, "global_step": 23244, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03573251888155937, "train_loss_bc": 0.03573251888155937, "train_loss_llm": 0.0, "grad_norm": 0.03384001553058624, "global_step": 23245, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03380906209349632, "train_loss_bc": 0.03380906209349632, "train_loss_llm": 0.0, "grad_norm": 0.060858868062496185, "global_step": 23246, "epoch": 645, "lr": 0.001} +{"train_loss": 0.02961141988635063, "train_loss_bc": 0.029593564569950104, "train_loss_llm": 0.0017854571342468262, "grad_norm": 0.07914438098669052, "global_step": 23247, "epoch": 645, "lr": 0.001} +{"train_loss": 0.034378260374069214, "train_loss_bc": 0.034378260374069214, "train_loss_llm": 0.0, "grad_norm": 0.09618709236383438, "global_step": 23248, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03564535826444626, "train_loss_bc": 0.035634592175483704, "train_loss_llm": 0.001076728105545044, "grad_norm": 0.027450673282146454, "global_step": 23249, "epoch": 645, "lr": 0.001} +{"train_loss": 0.029558023437857628, "train_loss_bc": 0.02955775149166584, "train_loss_llm": 2.7120113372802734e-05, "grad_norm": 0.04288958013057709, "global_step": 23250, "epoch": 645, "lr": 0.001} +{"train_loss": 0.028058620169758797, "train_loss_bc": 0.028050461784005165, "train_loss_llm": 0.0008159279823303223, "grad_norm": 0.05621477961540222, "global_step": 23251, "epoch": 645, "lr": 0.001} +{"train_loss": 0.033173080533742905, "train_loss_bc": 0.033173080533742905, "train_loss_llm": 0.0, "grad_norm": 0.06900741159915924, "global_step": 23252, "epoch": 645, "lr": 0.001} +{"train_loss": 0.029861565679311752, "train_loss_bc": 0.029861565679311752, "train_loss_llm": 0.0, "grad_norm": 0.026552964001893997, "global_step": 23253, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03140652924776077, "train_loss_bc": 0.03140652924776077, "train_loss_llm": 0.0, "grad_norm": 0.04055231437087059, "global_step": 23254, "epoch": 645, "lr": 0.001} +{"train_loss": 0.03199809960400065, "train_loss_bc": 0.03280181810259819, "train_loss_llm": 0.0, "grad_norm": 0.06797546148300171, "global_step": 23255, "epoch": 645, "lr": 0.001, "train/cumulative_reward": 86.61450129612332, "train/mean_score": 12.824613167967419, "train/success_rate": 0.3, "test/cumulative_reward": 33.46807565709573, "test/mean_score": 7.848375842610328, "test/success_rate": 0.2, "val_loss": 0.08465412259101868, "train_action_mse_error": 0.03449777141213417} +{"train_loss": 0.025774940848350525, "train_loss_bc": 0.025774940848350525, "train_loss_llm": 0.0, "grad_norm": 0.08455607295036316, "global_step": 23256, "epoch": 646, "lr": 0.001} +{"train_loss": 0.031610749661922455, "train_loss_bc": 0.031610749661922455, "train_loss_llm": 0.0, "grad_norm": 0.02814585715532303, "global_step": 23257, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03416541591286659, "train_loss_bc": 0.03416541591286659, "train_loss_llm": 0.0, "grad_norm": 0.044263359159231186, "global_step": 23258, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03398127108812332, "train_loss_bc": 0.03397496044635773, "train_loss_llm": 0.0006311237812042236, "grad_norm": 0.05907366797327995, "global_step": 23259, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03189114108681679, "train_loss_bc": 0.031886305660009384, "train_loss_llm": 0.0004836916923522949, "grad_norm": 0.07476143538951874, "global_step": 23260, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03553029149770737, "train_loss_bc": 0.03553029149770737, "train_loss_llm": 0.0, "grad_norm": 0.0369243398308754, "global_step": 23261, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03158501908183098, "train_loss_bc": 0.03158501908183098, "train_loss_llm": 0.0, "grad_norm": 0.04800468310713768, "global_step": 23262, "epoch": 646, "lr": 0.001} +{"train_loss": 0.0310993455350399, "train_loss_bc": 0.0310993455350399, "train_loss_llm": 0.0, "grad_norm": 0.06884276121854782, "global_step": 23263, "epoch": 646, "lr": 0.001} +{"train_loss": 0.028992580249905586, "train_loss_bc": 0.02899092622101307, "train_loss_llm": 0.00016540288925170898, "grad_norm": 0.07781904935836792, "global_step": 23264, "epoch": 646, "lr": 0.001} +{"train_loss": 0.0327320471405983, "train_loss_bc": 0.03272752836346626, "train_loss_llm": 0.00045183300971984863, "grad_norm": 0.03735532611608505, "global_step": 23265, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03150084242224693, "train_loss_bc": 0.03148579224944115, "train_loss_llm": 0.0015051960945129395, "grad_norm": 0.05845925956964493, "global_step": 23266, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03185934200882912, "train_loss_bc": 0.03185878321528435, "train_loss_llm": 5.602836608886719e-05, "grad_norm": 0.0668097659945488, "global_step": 23267, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03814142569899559, "train_loss_bc": 0.03814142569899559, "train_loss_llm": 0.0, "grad_norm": 0.08339149504899979, "global_step": 23268, "epoch": 646, "lr": 0.001} +{"train_loss": 0.02772560343146324, "train_loss_bc": 0.02772560343146324, "train_loss_llm": 0.0, "grad_norm": 0.0238577239215374, "global_step": 23269, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03162641450762749, "train_loss_bc": 0.03162641450762749, "train_loss_llm": 0.0, "grad_norm": 0.03542916104197502, "global_step": 23270, "epoch": 646, "lr": 0.001} +{"train_loss": 0.02847549319267273, "train_loss_bc": 0.02847549319267273, "train_loss_llm": 0.0, "grad_norm": 0.053035587072372437, "global_step": 23271, "epoch": 646, "lr": 0.001} +{"train_loss": 0.026733310893177986, "train_loss_bc": 0.02673054300248623, "train_loss_llm": 0.00027686357498168945, "grad_norm": 0.0693734735250473, "global_step": 23272, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03519903123378754, "train_loss_bc": 0.03519903123378754, "train_loss_llm": 0.0, "grad_norm": 0.028157388791441917, "global_step": 23273, "epoch": 646, "lr": 0.001} +{"train_loss": 0.026971500366926193, "train_loss_bc": 0.026971500366926193, "train_loss_llm": 0.0, "grad_norm": 0.044638097286224365, "global_step": 23274, "epoch": 646, "lr": 0.001} +{"train_loss": 0.0335901714861393, "train_loss_bc": 0.0335901714861393, "train_loss_llm": 0.0, "grad_norm": 0.06533988565206528, "global_step": 23275, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03421640396118164, "train_loss_bc": 0.03421640396118164, "train_loss_llm": 0.0, "grad_norm": 0.06922367215156555, "global_step": 23276, "epoch": 646, "lr": 0.001} +{"train_loss": 0.026231294497847557, "train_loss_bc": 0.026231294497847557, "train_loss_llm": 0.0, "grad_norm": 0.02750953659415245, "global_step": 23277, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03428121283650398, "train_loss_bc": 0.03428121283650398, "train_loss_llm": 0.0, "grad_norm": 0.0452740378677845, "global_step": 23278, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03497699275612831, "train_loss_bc": 0.03497699275612831, "train_loss_llm": 0.0, "grad_norm": 0.0656488686800003, "global_step": 23279, "epoch": 646, "lr": 0.001} +{"train_loss": 0.0338250957429409, "train_loss_bc": 0.03381707891821861, "train_loss_llm": 0.0008015930652618408, "grad_norm": 0.09027640521526337, "global_step": 23280, "epoch": 646, "lr": 0.001} +{"train_loss": 0.030276786535978317, "train_loss_bc": 0.030276786535978317, "train_loss_llm": 0.0, "grad_norm": 0.029152341187000275, "global_step": 23281, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03155025839805603, "train_loss_bc": 0.03154682740569115, "train_loss_llm": 0.0003431737422943115, "grad_norm": 0.044424161314964294, "global_step": 23282, "epoch": 646, "lr": 0.001} +{"train_loss": 0.02755458652973175, "train_loss_bc": 0.02755458652973175, "train_loss_llm": 0.0, "grad_norm": 0.0526726096868515, "global_step": 23283, "epoch": 646, "lr": 0.001} +{"train_loss": 0.030573777854442596, "train_loss_bc": 0.030573777854442596, "train_loss_llm": 0.0, "grad_norm": 0.07120891660451889, "global_step": 23284, "epoch": 646, "lr": 0.001} +{"train_loss": 0.027280477806925774, "train_loss_bc": 0.02726648934185505, "train_loss_llm": 0.0013988912105560303, "grad_norm": 0.03020019456744194, "global_step": 23285, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03261509910225868, "train_loss_bc": 0.03261509910225868, "train_loss_llm": 0.0, "grad_norm": 0.05432935804128647, "global_step": 23286, "epoch": 646, "lr": 0.001} +{"train_loss": 0.038295987993478775, "train_loss_bc": 0.038295987993478775, "train_loss_llm": 0.0, "grad_norm": 0.07570430636405945, "global_step": 23287, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03565414622426033, "train_loss_bc": 0.03565414622426033, "train_loss_llm": 0.0, "grad_norm": 0.08589350432157516, "global_step": 23288, "epoch": 646, "lr": 0.001} +{"train_loss": 0.03730519488453865, "train_loss_bc": 0.03730519488453865, "train_loss_llm": 0.0, "grad_norm": 0.032121289521455765, "global_step": 23289, "epoch": 646, "lr": 0.001} +{"train_loss": 0.028186922892928123, "train_loss_bc": 0.028186922892928123, "train_loss_llm": 0.0, "grad_norm": 0.0481940396130085, "global_step": 23290, "epoch": 646, "lr": 0.001} +{"train_loss": 0.032066353441526495, "train_loss_bc": 0.04237398877739906, "train_loss_llm": 0.0004558563232421875, "grad_norm": 0.06743326038122177, "global_step": 23291, "epoch": 646, "lr": 0.001, "val_loss": 0.08522684872150421} +{"train_loss": 0.03173333406448364, "train_loss_bc": 0.03173322230577469, "train_loss_llm": 1.1056661605834961e-05, "grad_norm": 0.08020181208848953, "global_step": 23292, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03421536087989807, "train_loss_bc": 0.034209586679935455, "train_loss_llm": 0.0005773007869720459, "grad_norm": 0.032428160309791565, "global_step": 23293, "epoch": 647, "lr": 0.001} +{"train_loss": 0.028028802946209908, "train_loss_bc": 0.028028802946209908, "train_loss_llm": 0.0, "grad_norm": 0.04898681864142418, "global_step": 23294, "epoch": 647, "lr": 0.001} +{"train_loss": 0.030272191390395164, "train_loss_bc": 0.030272191390395164, "train_loss_llm": 0.0, "grad_norm": 0.05745982006192207, "global_step": 23295, "epoch": 647, "lr": 0.001} +{"train_loss": 0.0318138524889946, "train_loss_bc": 0.0318138524889946, "train_loss_llm": 0.0, "grad_norm": 0.07755585014820099, "global_step": 23296, "epoch": 647, "lr": 0.001} +{"train_loss": 0.028962574899196625, "train_loss_bc": 0.028957247734069824, "train_loss_llm": 0.0005326569080352783, "grad_norm": 0.026815244928002357, "global_step": 23297, "epoch": 647, "lr": 0.001} +{"train_loss": 0.0384581983089447, "train_loss_bc": 0.0384581983089447, "train_loss_llm": 0.0, "grad_norm": 0.03920796141028404, "global_step": 23298, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03167682886123657, "train_loss_bc": 0.03167682886123657, "train_loss_llm": 0.0, "grad_norm": 0.055766794830560684, "global_step": 23299, "epoch": 647, "lr": 0.001} +{"train_loss": 0.030617456883192062, "train_loss_bc": 0.03060821071267128, "train_loss_llm": 0.0009246766567230225, "grad_norm": 0.06827520579099655, "global_step": 23300, "epoch": 647, "lr": 0.001} +{"train_loss": 0.029111821204423904, "train_loss_bc": 0.029111821204423904, "train_loss_llm": 0.0, "grad_norm": 0.03640192002058029, "global_step": 23301, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03199410066008568, "train_loss_bc": 0.03199410066008568, "train_loss_llm": 0.0, "grad_norm": 0.049813415855169296, "global_step": 23302, "epoch": 647, "lr": 0.001} +{"train_loss": 0.025631621479988098, "train_loss_bc": 0.025631621479988098, "train_loss_llm": 0.0, "grad_norm": 0.05623599886894226, "global_step": 23303, "epoch": 647, "lr": 0.001} +{"train_loss": 0.02831391990184784, "train_loss_bc": 0.02831391990184784, "train_loss_llm": 0.0, "grad_norm": 0.07406623661518097, "global_step": 23304, "epoch": 647, "lr": 0.001} +{"train_loss": 0.029565779492259026, "train_loss_bc": 0.029560722410678864, "train_loss_llm": 0.0005056262016296387, "grad_norm": 0.025219373404979706, "global_step": 23305, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03349560126662254, "train_loss_bc": 0.03349560126662254, "train_loss_llm": 0.0, "grad_norm": 0.03466440737247467, "global_step": 23306, "epoch": 647, "lr": 0.001} +{"train_loss": 0.027282780036330223, "train_loss_bc": 0.027282780036330223, "train_loss_llm": 0.0, "grad_norm": 0.045188985764980316, "global_step": 23307, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03257841616868973, "train_loss_bc": 0.03257841616868973, "train_loss_llm": 0.0, "grad_norm": 0.062125883996486664, "global_step": 23308, "epoch": 647, "lr": 0.001} +{"train_loss": 0.037531428039073944, "train_loss_bc": 0.037531428039073944, "train_loss_llm": 0.0, "grad_norm": 0.026658831164240837, "global_step": 23309, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03972669318318367, "train_loss_bc": 0.03972669318318367, "train_loss_llm": 0.0, "grad_norm": 0.04545529559254646, "global_step": 23310, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03428415581583977, "train_loss_bc": 0.03428415581583977, "train_loss_llm": 0.0, "grad_norm": 0.05446292832493782, "global_step": 23311, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03593902289867401, "train_loss_bc": 0.0359298512339592, "train_loss_llm": 0.0009171068668365479, "grad_norm": 0.07249090820550919, "global_step": 23312, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03743519261479378, "train_loss_bc": 0.03743519261479378, "train_loss_llm": 0.0, "grad_norm": 0.03527044877409935, "global_step": 23313, "epoch": 647, "lr": 0.001} +{"train_loss": 0.02725297212600708, "train_loss_bc": 0.027233075350522995, "train_loss_llm": 0.001989692449569702, "grad_norm": 0.05751391127705574, "global_step": 23314, "epoch": 647, "lr": 0.001} +{"train_loss": 0.031700581312179565, "train_loss_bc": 0.031700581312179565, "train_loss_llm": 0.0, "grad_norm": 0.06130426749587059, "global_step": 23315, "epoch": 647, "lr": 0.001} +{"train_loss": 0.028253892436623573, "train_loss_bc": 0.028253892436623573, "train_loss_llm": 0.0, "grad_norm": 0.0731947124004364, "global_step": 23316, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03322776034474373, "train_loss_bc": 0.033211760222911835, "train_loss_llm": 0.0015999078750610352, "grad_norm": 0.029869114980101585, "global_step": 23317, "epoch": 647, "lr": 0.001} +{"train_loss": 0.036513082683086395, "train_loss_bc": 0.036513082683086395, "train_loss_llm": 0.0, "grad_norm": 0.046339891850948334, "global_step": 23318, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03570745885372162, "train_loss_bc": 0.03570745885372162, "train_loss_llm": 0.0, "grad_norm": 0.0621831938624382, "global_step": 23319, "epoch": 647, "lr": 0.001} +{"train_loss": 0.02579953894019127, "train_loss_bc": 0.02579953894019127, "train_loss_llm": 0.0, "grad_norm": 0.08285313844680786, "global_step": 23320, "epoch": 647, "lr": 0.001} +{"train_loss": 0.02619076892733574, "train_loss_bc": 0.02617730014026165, "train_loss_llm": 0.0013468563556671143, "grad_norm": 0.02581850253045559, "global_step": 23321, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03063581883907318, "train_loss_bc": 0.03063581883907318, "train_loss_llm": 0.0, "grad_norm": 0.039625853300094604, "global_step": 23322, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03168675675988197, "train_loss_bc": 0.03168675675988197, "train_loss_llm": 0.0, "grad_norm": 0.06186148524284363, "global_step": 23323, "epoch": 647, "lr": 0.001} +{"train_loss": 0.037197839468717575, "train_loss_bc": 0.037197839468717575, "train_loss_llm": 0.0, "grad_norm": 0.08467775583267212, "global_step": 23324, "epoch": 647, "lr": 0.001} +{"train_loss": 0.03191910311579704, "train_loss_bc": 0.03191910311579704, "train_loss_llm": 0.0, "grad_norm": 0.03441014513373375, "global_step": 23325, "epoch": 647, "lr": 0.001} +{"train_loss": 0.028870444744825363, "train_loss_bc": 0.028870444744825363, "train_loss_llm": 0.0, "grad_norm": 0.03843889385461807, "global_step": 23326, "epoch": 647, "lr": 0.001} +{"train_loss": 0.031826340748618044, "train_loss_bc": 0.03212311491370201, "train_loss_llm": 0.0, "grad_norm": 0.04864763841032982, "global_step": 23327, "epoch": 647, "lr": 0.001, "val_loss": 0.0850447416305542} +{"train_loss": 0.03223690018057823, "train_loss_bc": 0.03223690018057823, "train_loss_llm": 0.0, "grad_norm": 0.05674801394343376, "global_step": 23328, "epoch": 648, "lr": 0.001} +{"train_loss": 0.036940839141607285, "train_loss_bc": 0.036926623433828354, "train_loss_llm": 0.0014214813709259033, "grad_norm": 0.035273514688014984, "global_step": 23329, "epoch": 648, "lr": 0.001} +{"train_loss": 0.033895015716552734, "train_loss_bc": 0.033895015716552734, "train_loss_llm": 0.0, "grad_norm": 0.05185773968696594, "global_step": 23330, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03273196145892143, "train_loss_bc": 0.032731495797634125, "train_loss_llm": 4.646182060241699e-05, "grad_norm": 0.06851169466972351, "global_step": 23331, "epoch": 648, "lr": 0.001} +{"train_loss": 0.029641613364219666, "train_loss_bc": 0.029641613364219666, "train_loss_llm": 0.0, "grad_norm": 0.08085864782333374, "global_step": 23332, "epoch": 648, "lr": 0.001} +{"train_loss": 0.02937721274793148, "train_loss_bc": 0.02937721274793148, "train_loss_llm": 0.0, "grad_norm": 0.02767268195748329, "global_step": 23333, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03273078799247742, "train_loss_bc": 0.0327211394906044, "train_loss_llm": 0.0009647905826568604, "grad_norm": 0.04923612251877785, "global_step": 23334, "epoch": 648, "lr": 0.001} +{"train_loss": 0.029478374868631363, "train_loss_bc": 0.029478374868631363, "train_loss_llm": 0.0, "grad_norm": 0.06570952385663986, "global_step": 23335, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03099462017416954, "train_loss_bc": 0.03099462017416954, "train_loss_llm": 0.0, "grad_norm": 0.08146682381629944, "global_step": 23336, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03312874957919121, "train_loss_bc": 0.03311418741941452, "train_loss_llm": 0.0014561116695404053, "grad_norm": 0.026574067771434784, "global_step": 23337, "epoch": 648, "lr": 0.001} +{"train_loss": 0.02656792476773262, "train_loss_bc": 0.02656601183116436, "train_loss_llm": 0.00019127130508422852, "grad_norm": 0.04177698865532875, "global_step": 23338, "epoch": 648, "lr": 0.001} +{"train_loss": 0.034682054072618484, "train_loss_bc": 0.034682054072618484, "train_loss_llm": 0.0, "grad_norm": 0.04993262141942978, "global_step": 23339, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03297773003578186, "train_loss_bc": 0.03297773003578186, "train_loss_llm": 0.0, "grad_norm": 0.06619106233119965, "global_step": 23340, "epoch": 648, "lr": 0.001} +{"train_loss": 0.04041634500026703, "train_loss_bc": 0.04041634500026703, "train_loss_llm": 0.0, "grad_norm": 0.0400443896651268, "global_step": 23341, "epoch": 648, "lr": 0.001} +{"train_loss": 0.028642669320106506, "train_loss_bc": 0.028630496934056282, "train_loss_llm": 0.001217186450958252, "grad_norm": 0.05210326611995697, "global_step": 23342, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03003661520779133, "train_loss_bc": 0.030030466616153717, "train_loss_llm": 0.0006147921085357666, "grad_norm": 0.06470668315887451, "global_step": 23343, "epoch": 648, "lr": 0.001} +{"train_loss": 0.031071113422513008, "train_loss_bc": 0.031071113422513008, "train_loss_llm": 0.0, "grad_norm": 0.08581427484750748, "global_step": 23344, "epoch": 648, "lr": 0.001} +{"train_loss": 0.029904259368777275, "train_loss_bc": 0.029904259368777275, "train_loss_llm": 0.0, "grad_norm": 0.027075083926320076, "global_step": 23345, "epoch": 648, "lr": 0.001} +{"train_loss": 0.02420128509402275, "train_loss_bc": 0.02420128509402275, "train_loss_llm": 0.0, "grad_norm": 0.04291309043765068, "global_step": 23346, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03495151922106743, "train_loss_bc": 0.03494681790471077, "train_loss_llm": 0.0004703104496002197, "grad_norm": 0.06423961371183395, "global_step": 23347, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03265988826751709, "train_loss_bc": 0.03265988826751709, "train_loss_llm": 0.0, "grad_norm": 0.08368553966283798, "global_step": 23348, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03021589107811451, "train_loss_bc": 0.030212443321943283, "train_loss_llm": 0.0003447234630584717, "grad_norm": 0.02679113857448101, "global_step": 23349, "epoch": 648, "lr": 0.001} +{"train_loss": 0.030671538785099983, "train_loss_bc": 0.030671538785099983, "train_loss_llm": 0.0, "grad_norm": 0.04046845808625221, "global_step": 23350, "epoch": 648, "lr": 0.001} +{"train_loss": 0.031546127051115036, "train_loss_bc": 0.03154611587524414, "train_loss_llm": 1.0132789611816406e-06, "grad_norm": 0.054643843322992325, "global_step": 23351, "epoch": 648, "lr": 0.001} +{"train_loss": 0.029295826330780983, "train_loss_bc": 0.029295826330780983, "train_loss_llm": 0.0, "grad_norm": 0.06846750527620316, "global_step": 23352, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03517128899693489, "train_loss_bc": 0.03517128899693489, "train_loss_llm": 0.0, "grad_norm": 0.03722156211733818, "global_step": 23353, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03053143061697483, "train_loss_bc": 0.03053143061697483, "train_loss_llm": 0.0, "grad_norm": 0.05355559289455414, "global_step": 23354, "epoch": 648, "lr": 0.001} +{"train_loss": 0.033396631479263306, "train_loss_bc": 0.033396631479263306, "train_loss_llm": 0.0, "grad_norm": 0.07774832099676132, "global_step": 23355, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03365698829293251, "train_loss_bc": 0.03365698829293251, "train_loss_llm": 0.0, "grad_norm": 0.09967884421348572, "global_step": 23356, "epoch": 648, "lr": 0.001} +{"train_loss": 0.03637300431728363, "train_loss_bc": 0.03637300431728363, "train_loss_llm": 0.0, "grad_norm": 0.03830191120505333, "global_step": 23357, "epoch": 648, "lr": 0.001} +{"train_loss": 0.027429288253188133, "train_loss_bc": 0.027429288253188133, "train_loss_llm": 0.0, "grad_norm": 0.050009116530418396, "global_step": 23358, "epoch": 648, "lr": 0.001} +{"train_loss": 0.030717162415385246, "train_loss_bc": 0.030717162415385246, "train_loss_llm": 0.0, "grad_norm": 0.069096140563488, "global_step": 23359, "epoch": 648, "lr": 0.001} +{"train_loss": 0.031241802498698235, "train_loss_bc": 0.03122408501803875, "train_loss_llm": 0.001771777868270874, "grad_norm": 0.09112010151147842, "global_step": 23360, "epoch": 648, "lr": 0.001} +{"train_loss": 0.037767443805933, "train_loss_bc": 0.037767443805933, "train_loss_llm": 0.0, "grad_norm": 0.03127047047019005, "global_step": 23361, "epoch": 648, "lr": 0.001} +{"train_loss": 0.029290419071912766, "train_loss_bc": 0.029290419071912766, "train_loss_llm": 0.0, "grad_norm": 0.042406048625707626, "global_step": 23362, "epoch": 648, "lr": 0.001} +{"train_loss": 0.031922991646246776, "train_loss_bc": 0.0346553772687912, "train_loss_llm": 0.0, "grad_norm": 0.05673637241125107, "global_step": 23363, "epoch": 648, "lr": 0.001, "val_loss": 0.08558763563632965} +{"train_loss": 0.034159913659095764, "train_loss_bc": 0.034159913659095764, "train_loss_llm": 0.0, "grad_norm": 0.08154967427253723, "global_step": 23364, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03314078226685524, "train_loss_bc": 0.03310475870966911, "train_loss_llm": 0.003602534532546997, "grad_norm": 0.03939233720302582, "global_step": 23365, "epoch": 649, "lr": 0.001} +{"train_loss": 0.027933260425925255, "train_loss_bc": 0.027920851483941078, "train_loss_llm": 0.0012409687042236328, "grad_norm": 0.05491112917661667, "global_step": 23366, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03240349888801575, "train_loss_bc": 0.03240349888801575, "train_loss_llm": 0.0, "grad_norm": 0.07878269255161285, "global_step": 23367, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03074464201927185, "train_loss_bc": 0.03074464201927185, "train_loss_llm": 0.0, "grad_norm": 0.08831547945737839, "global_step": 23368, "epoch": 649, "lr": 0.001} +{"train_loss": 0.028536207973957062, "train_loss_bc": 0.028536207973957062, "train_loss_llm": 0.0, "grad_norm": 0.026299219578504562, "global_step": 23369, "epoch": 649, "lr": 0.001} +{"train_loss": 0.02708575502038002, "train_loss_bc": 0.02708575502038002, "train_loss_llm": 0.0, "grad_norm": 0.042552556842565536, "global_step": 23370, "epoch": 649, "lr": 0.001} +{"train_loss": 0.030692698433995247, "train_loss_bc": 0.030692698433995247, "train_loss_llm": 0.0, "grad_norm": 0.06046338006854057, "global_step": 23371, "epoch": 649, "lr": 0.001} +{"train_loss": 0.032541945576667786, "train_loss_bc": 0.032541945576667786, "train_loss_llm": 0.0, "grad_norm": 0.0682140588760376, "global_step": 23372, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03401020169258118, "train_loss_bc": 0.03398987650871277, "train_loss_llm": 0.002032458782196045, "grad_norm": 0.041591912508010864, "global_step": 23373, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03175994008779526, "train_loss_bc": 0.03175994008779526, "train_loss_llm": 0.0, "grad_norm": 0.05184578523039818, "global_step": 23374, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03994011506438255, "train_loss_bc": 0.03994011506438255, "train_loss_llm": 0.0, "grad_norm": 0.06847986578941345, "global_step": 23375, "epoch": 649, "lr": 0.001} +{"train_loss": 0.027924207970499992, "train_loss_bc": 0.027916651219129562, "train_loss_llm": 0.00075569748878479, "grad_norm": 0.08150655776262283, "global_step": 23376, "epoch": 649, "lr": 0.001} +{"train_loss": 0.029726671054959297, "train_loss_bc": 0.02971923165023327, "train_loss_llm": 0.0007438957691192627, "grad_norm": 0.029001224786043167, "global_step": 23377, "epoch": 649, "lr": 0.001} +{"train_loss": 0.030330615118145943, "train_loss_bc": 0.030330615118145943, "train_loss_llm": 0.0, "grad_norm": 0.049592770636081696, "global_step": 23378, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03468073904514313, "train_loss_bc": 0.03468073904514313, "train_loss_llm": 0.0, "grad_norm": 0.06738285720348358, "global_step": 23379, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03431888297200203, "train_loss_bc": 0.03431888297200203, "train_loss_llm": 0.0, "grad_norm": 0.0847228467464447, "global_step": 23380, "epoch": 649, "lr": 0.001} +{"train_loss": 0.032606903463602066, "train_loss_bc": 0.032606903463602066, "train_loss_llm": 0.0, "grad_norm": 0.03156847879290581, "global_step": 23381, "epoch": 649, "lr": 0.001} +{"train_loss": 0.034226298332214355, "train_loss_bc": 0.03422244265675545, "train_loss_llm": 0.00038561224937438965, "grad_norm": 0.045412007719278336, "global_step": 23382, "epoch": 649, "lr": 0.001} +{"train_loss": 0.031843509525060654, "train_loss_bc": 0.031843509525060654, "train_loss_llm": 0.0, "grad_norm": 0.06097107008099556, "global_step": 23383, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03517618402838707, "train_loss_bc": 0.03517618402838707, "train_loss_llm": 0.0, "grad_norm": 0.07629571855068207, "global_step": 23384, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03420415148139, "train_loss_bc": 0.03420415148139, "train_loss_llm": 0.0, "grad_norm": 0.02840569242835045, "global_step": 23385, "epoch": 649, "lr": 0.001} +{"train_loss": 0.02696296200156212, "train_loss_bc": 0.02696296200156212, "train_loss_llm": 0.0, "grad_norm": 0.03589114174246788, "global_step": 23386, "epoch": 649, "lr": 0.001} +{"train_loss": 0.026529859751462936, "train_loss_bc": 0.026529695838689804, "train_loss_llm": 1.633167266845703e-05, "grad_norm": 0.0499454140663147, "global_step": 23387, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03541508689522743, "train_loss_bc": 0.03541508689522743, "train_loss_llm": 0.0, "grad_norm": 0.07010658085346222, "global_step": 23388, "epoch": 649, "lr": 0.001} +{"train_loss": 0.02606135606765747, "train_loss_bc": 0.02606135606765747, "train_loss_llm": 0.0, "grad_norm": 0.031099669635295868, "global_step": 23389, "epoch": 649, "lr": 0.001} +{"train_loss": 0.035393040627241135, "train_loss_bc": 0.035393040627241135, "train_loss_llm": 0.0, "grad_norm": 0.04630471393465996, "global_step": 23390, "epoch": 649, "lr": 0.001} +{"train_loss": 0.02700856328010559, "train_loss_bc": 0.02700856328010559, "train_loss_llm": 0.0, "grad_norm": 0.06869689375162125, "global_step": 23391, "epoch": 649, "lr": 0.001} +{"train_loss": 0.028657475486397743, "train_loss_bc": 0.028657475486397743, "train_loss_llm": 0.0, "grad_norm": 0.08797908574342728, "global_step": 23392, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03380744531750679, "train_loss_bc": 0.03380744531750679, "train_loss_llm": 0.0, "grad_norm": 0.03600502759218216, "global_step": 23393, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03417385742068291, "train_loss_bc": 0.03417385742068291, "train_loss_llm": 0.0, "grad_norm": 0.04714403301477432, "global_step": 23394, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03835673630237579, "train_loss_bc": 0.03835673630237579, "train_loss_llm": 0.0, "grad_norm": 0.07319286465644836, "global_step": 23395, "epoch": 649, "lr": 0.001} +{"train_loss": 0.02779233269393444, "train_loss_bc": 0.02779233269393444, "train_loss_llm": 0.0, "grad_norm": 0.08624964207410812, "global_step": 23396, "epoch": 649, "lr": 0.001} +{"train_loss": 0.036581024527549744, "train_loss_bc": 0.036581024527549744, "train_loss_llm": 0.0, "grad_norm": 0.03991640731692314, "global_step": 23397, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03216201439499855, "train_loss_bc": 0.032146528363227844, "train_loss_llm": 0.0015485286712646484, "grad_norm": 0.0627451092004776, "global_step": 23398, "epoch": 649, "lr": 0.001} +{"train_loss": 0.03201762126344773, "train_loss_bc": 0.03574548661708832, "train_loss_llm": 0.0, "grad_norm": 0.08736936748027802, "global_step": 23399, "epoch": 649, "lr": 0.001, "val_loss": 0.08653247356414795} +{"train_loss": 0.030480246990919113, "train_loss_bc": 0.030480246990919113, "train_loss_llm": 0.0, "grad_norm": 0.09997354447841644, "global_step": 23400, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03469904512166977, "train_loss_bc": 0.03469904512166977, "train_loss_llm": 0.0, "grad_norm": 0.0324220135807991, "global_step": 23401, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03416874632239342, "train_loss_bc": 0.03416874632239342, "train_loss_llm": 0.0, "grad_norm": 0.045600928366184235, "global_step": 23402, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03215079382061958, "train_loss_bc": 0.03215079382061958, "train_loss_llm": 0.0, "grad_norm": 0.06209857761859894, "global_step": 23403, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03284536302089691, "train_loss_bc": 0.03284536302089691, "train_loss_llm": 0.0, "grad_norm": 0.0806891918182373, "global_step": 23404, "epoch": 650, "lr": 0.001} +{"train_loss": 0.027934452518820763, "train_loss_bc": 0.027934452518820763, "train_loss_llm": 0.0, "grad_norm": 0.022454719990491867, "global_step": 23405, "epoch": 650, "lr": 0.001} +{"train_loss": 0.029776956886053085, "train_loss_bc": 0.029776956886053085, "train_loss_llm": 0.0, "grad_norm": 0.0392887219786644, "global_step": 23406, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03648217022418976, "train_loss_bc": 0.03648217022418976, "train_loss_llm": 0.0, "grad_norm": 0.055468831211328506, "global_step": 23407, "epoch": 650, "lr": 0.001} +{"train_loss": 0.028402578085660934, "train_loss_bc": 0.028399255126714706, "train_loss_llm": 0.00033226609230041504, "grad_norm": 0.07032836973667145, "global_step": 23408, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03339935839176178, "train_loss_bc": 0.03339935839176178, "train_loss_llm": 0.0, "grad_norm": 0.031008915975689888, "global_step": 23409, "epoch": 650, "lr": 0.001} +{"train_loss": 0.0322294719517231, "train_loss_bc": 0.0322294719517231, "train_loss_llm": 0.0, "grad_norm": 0.05248691514134407, "global_step": 23410, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03348340466618538, "train_loss_bc": 0.033479221165180206, "train_loss_llm": 0.0004184544086456299, "grad_norm": 0.06692738831043243, "global_step": 23411, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03222127631306648, "train_loss_bc": 0.03222127631306648, "train_loss_llm": 0.0, "grad_norm": 0.08277982473373413, "global_step": 23412, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03525484725832939, "train_loss_bc": 0.03525484725832939, "train_loss_llm": 0.0, "grad_norm": 0.03892112150788307, "global_step": 23413, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03302166238427162, "train_loss_bc": 0.03299788758158684, "train_loss_llm": 0.002377539873123169, "grad_norm": 0.047120921313762665, "global_step": 23414, "epoch": 650, "lr": 0.001} +{"train_loss": 0.033770907670259476, "train_loss_bc": 0.033770907670259476, "train_loss_llm": 0.0, "grad_norm": 0.07540906220674515, "global_step": 23415, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03377648442983627, "train_loss_bc": 0.03377648442983627, "train_loss_llm": 0.0, "grad_norm": 0.08589192479848862, "global_step": 23416, "epoch": 650, "lr": 0.001} +{"train_loss": 0.026224475353956223, "train_loss_bc": 0.026224475353956223, "train_loss_llm": 0.0, "grad_norm": 0.032151009887456894, "global_step": 23417, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03595687076449394, "train_loss_bc": 0.03595687076449394, "train_loss_llm": 0.0, "grad_norm": 0.051113311201334, "global_step": 23418, "epoch": 650, "lr": 0.001} +{"train_loss": 0.030039308592677116, "train_loss_bc": 0.030039308592677116, "train_loss_llm": 0.0, "grad_norm": 0.07528088986873627, "global_step": 23419, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03221282362937927, "train_loss_bc": 0.03221282362937927, "train_loss_llm": 0.0, "grad_norm": 0.09561237692832947, "global_step": 23420, "epoch": 650, "lr": 0.001} +{"train_loss": 0.024711159989237785, "train_loss_bc": 0.02470739185810089, "train_loss_llm": 0.0003768801689147949, "grad_norm": 0.025362646207213402, "global_step": 23421, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03176785260438919, "train_loss_bc": 0.03176551312208176, "train_loss_llm": 0.0002340376377105713, "grad_norm": 0.0411897674202919, "global_step": 23422, "epoch": 650, "lr": 0.001} +{"train_loss": 0.026061279699206352, "train_loss_bc": 0.026053238660097122, "train_loss_llm": 0.0008040964603424072, "grad_norm": 0.05583825334906578, "global_step": 23423, "epoch": 650, "lr": 0.001} +{"train_loss": 0.04066259413957596, "train_loss_bc": 0.04066259413957596, "train_loss_llm": 0.0, "grad_norm": 0.07463479787111282, "global_step": 23424, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03153682500123978, "train_loss_bc": 0.03153606131672859, "train_loss_llm": 7.620453834533691e-05, "grad_norm": 0.030251450836658478, "global_step": 23425, "epoch": 650, "lr": 0.001} +{"train_loss": 0.030524101108312607, "train_loss_bc": 0.030524101108312607, "train_loss_llm": 0.0, "grad_norm": 0.04393058270215988, "global_step": 23426, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03484027832746506, "train_loss_bc": 0.03484027832746506, "train_loss_llm": 0.0, "grad_norm": 0.0687747374176979, "global_step": 23427, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03069341368973255, "train_loss_bc": 0.030655674636363983, "train_loss_llm": 0.003773987293243408, "grad_norm": 0.08205045759677887, "global_step": 23428, "epoch": 650, "lr": 0.001} +{"train_loss": 0.024262119084596634, "train_loss_bc": 0.024262119084596634, "train_loss_llm": 0.0, "grad_norm": 0.027221661061048508, "global_step": 23429, "epoch": 650, "lr": 0.001} +{"train_loss": 0.030413895845413208, "train_loss_bc": 0.030413895845413208, "train_loss_llm": 0.0, "grad_norm": 0.03981100767850876, "global_step": 23430, "epoch": 650, "lr": 0.001} +{"train_loss": 0.02737516723573208, "train_loss_bc": 0.02737058885395527, "train_loss_llm": 0.0004577934741973877, "grad_norm": 0.04795708507299423, "global_step": 23431, "epoch": 650, "lr": 0.001} +{"train_loss": 0.02923939749598503, "train_loss_bc": 0.02923797443509102, "train_loss_llm": 0.0001423656940460205, "grad_norm": 0.05818692594766617, "global_step": 23432, "epoch": 650, "lr": 0.001} +{"train_loss": 0.034567344933748245, "train_loss_bc": 0.034564513713121414, "train_loss_llm": 0.000282973051071167, "grad_norm": 0.03352908790111542, "global_step": 23433, "epoch": 650, "lr": 0.001} +{"train_loss": 0.03769926726818085, "train_loss_bc": 0.037687912583351135, "train_loss_llm": 0.0011353492736816406, "grad_norm": 0.05690008029341698, "global_step": 23434, "epoch": 650, "lr": 0.001} +{"train_loss": 0.032046407668126956, "train_loss_bc": 0.04078473523259163, "train_loss_llm": 0.0, "grad_norm": 0.08584656566381454, "global_step": 23435, "epoch": 650, "lr": 0.001, "train/cumulative_reward": 67.14490436591237, "train/mean_score": 11.813636253562596, "train/success_rate": 0.3, "test/cumulative_reward": 31.263044837973467, "test/mean_score": 8.595566488988794, "test/success_rate": 0.26, "val_loss": 0.08502309024333954, "train_action_mse_error": 0.03469938784837723} +{"train_loss": 0.027987176552414894, "train_loss_bc": 0.027983371168375015, "train_loss_llm": 0.00038051605224609375, "grad_norm": 0.09911206364631653, "global_step": 23436, "epoch": 651, "lr": 0.001} +{"train_loss": 0.03154613822698593, "train_loss_bc": 0.03154613822698593, "train_loss_llm": 0.0, "grad_norm": 0.027935486286878586, "global_step": 23437, "epoch": 651, "lr": 0.001} +{"train_loss": 0.030474497005343437, "train_loss_bc": 0.030474497005343437, "train_loss_llm": 0.0, "grad_norm": 0.04274629056453705, "global_step": 23438, "epoch": 651, "lr": 0.001} +{"train_loss": 0.03257179260253906, "train_loss_bc": 0.03256195783615112, "train_loss_llm": 0.0009836554527282715, "grad_norm": 0.054631538689136505, "global_step": 23439, "epoch": 651, "lr": 0.001} +{"train_loss": 0.036463089287281036, "train_loss_bc": 0.036463089287281036, "train_loss_llm": 0.0, "grad_norm": 0.0724697858095169, "global_step": 23440, "epoch": 651, "lr": 0.001} +{"train_loss": 0.02783406339585781, "train_loss_bc": 0.02783406339585781, "train_loss_llm": 0.0, "grad_norm": 0.03322187811136246, "global_step": 23441, "epoch": 651, "lr": 0.001} +{"train_loss": 0.02923549897968769, "train_loss_bc": 0.02923549897968769, "train_loss_llm": 0.0, "grad_norm": 0.05265128239989281, "global_step": 23442, "epoch": 651, "lr": 0.001} +{"train_loss": 0.031086541712284088, "train_loss_bc": 0.031086541712284088, "train_loss_llm": 0.0, "grad_norm": 0.07367239892482758, "global_step": 23443, "epoch": 651, "lr": 0.001} +{"train_loss": 0.034531161189079285, "train_loss_bc": 0.034525156021118164, "train_loss_llm": 0.0006005465984344482, "grad_norm": 0.09773585200309753, "global_step": 23444, "epoch": 651, "lr": 0.001} +{"train_loss": 0.035605207085609436, "train_loss_bc": 0.035605207085609436, "train_loss_llm": 0.0, "grad_norm": 0.02968216873705387, "global_step": 23445, "epoch": 651, "lr": 0.001} +{"train_loss": 0.025795290246605873, "train_loss_bc": 0.025795290246605873, "train_loss_llm": 0.0, "grad_norm": 0.04469536989927292, "global_step": 23446, "epoch": 651, "lr": 0.001} +{"train_loss": 0.028256867080926895, "train_loss_bc": 0.028256867080926895, "train_loss_llm": 0.0, "grad_norm": 0.06259874999523163, "global_step": 23447, "epoch": 651, "lr": 0.001} +{"train_loss": 0.030226992443203926, "train_loss_bc": 0.030226992443203926, "train_loss_llm": 0.0, "grad_norm": 0.07602453976869583, "global_step": 23448, "epoch": 651, "lr": 0.001} +{"train_loss": 0.03428083658218384, "train_loss_bc": 0.03427231311798096, "train_loss_llm": 0.0008521974086761475, "grad_norm": 0.0452861487865448, "global_step": 23449, "epoch": 651, "lr": 0.001} +{"train_loss": 0.0334845706820488, "train_loss_bc": 0.0334845706820488, "train_loss_llm": 0.0, "grad_norm": 0.07176018506288528, "global_step": 23450, "epoch": 651, "lr": 0.001} +{"train_loss": 0.03495318442583084, "train_loss_bc": 0.03495318442583084, "train_loss_llm": 0.0, "grad_norm": 0.10380542278289795, "global_step": 23451, "epoch": 651, "lr": 0.001} +{"train_loss": 0.03971369192004204, "train_loss_bc": 0.0396660752594471, "train_loss_llm": 0.004761487245559692, "grad_norm": 0.13907788693904877, "global_step": 23452, "epoch": 651, "lr": 0.001} +{"train_loss": 0.038298919796943665, "train_loss_bc": 0.038298919796943665, "train_loss_llm": 0.0, "grad_norm": 0.027568073943257332, "global_step": 23453, "epoch": 651, "lr": 0.001} +{"train_loss": 0.03706948086619377, "train_loss_bc": 0.03706948086619377, "train_loss_llm": 0.0, "grad_norm": 0.051887452602386475, "global_step": 23454, "epoch": 651, "lr": 0.001} +{"train_loss": 0.0281326025724411, "train_loss_bc": 0.0281326025724411, "train_loss_llm": 0.0, "grad_norm": 0.06469248235225677, "global_step": 23455, "epoch": 651, "lr": 0.001} +{"train_loss": 0.036394305527210236, "train_loss_bc": 0.036394305527210236, "train_loss_llm": 0.0, "grad_norm": 0.0880221426486969, "global_step": 23456, "epoch": 651, "lr": 0.001} +{"train_loss": 0.030792487785220146, "train_loss_bc": 0.030792487785220146, "train_loss_llm": 0.0, "grad_norm": 0.03875086084008217, "global_step": 23457, "epoch": 651, "lr": 0.001} +{"train_loss": 0.026518434286117554, "train_loss_bc": 0.026518434286117554, "train_loss_llm": 0.0, "grad_norm": 0.050565727055072784, "global_step": 23458, "epoch": 651, "lr": 0.001} +{"train_loss": 0.03377237543463707, "train_loss_bc": 0.03377237543463707, "train_loss_llm": 0.0, "grad_norm": 0.06563378870487213, "global_step": 23459, "epoch": 651, "lr": 0.001} +{"train_loss": 0.028919650241732597, "train_loss_bc": 0.02891233190894127, "train_loss_llm": 0.0007317662239074707, "grad_norm": 0.07394734025001526, "global_step": 23460, "epoch": 651, "lr": 0.001} +{"train_loss": 0.03330519050359726, "train_loss_bc": 0.03330519050359726, "train_loss_llm": 0.0, "grad_norm": 0.03312985971570015, "global_step": 23461, "epoch": 651, "lr": 0.001} +{"train_loss": 0.029447240754961967, "train_loss_bc": 0.029435353353619576, "train_loss_llm": 0.0011887550354003906, "grad_norm": 0.0439458005130291, "global_step": 23462, "epoch": 651, "lr": 0.001} +{"train_loss": 0.02954990230500698, "train_loss_bc": 0.02954990230500698, "train_loss_llm": 0.0, "grad_norm": 0.06125175207853317, "global_step": 23463, "epoch": 651, "lr": 0.001} +{"train_loss": 0.030603516846895218, "train_loss_bc": 0.030603516846895218, "train_loss_llm": 0.0, "grad_norm": 0.08374176919460297, "global_step": 23464, "epoch": 651, "lr": 0.001} +{"train_loss": 0.03744989261031151, "train_loss_bc": 0.037417229264974594, "train_loss_llm": 0.0032663345336914062, "grad_norm": 0.039275579154491425, "global_step": 23465, "epoch": 651, "lr": 0.001} +{"train_loss": 0.0384024903178215, "train_loss_bc": 0.0384024903178215, "train_loss_llm": 0.0, "grad_norm": 0.0737854540348053, "global_step": 23466, "epoch": 651, "lr": 0.001} +{"train_loss": 0.033798765391111374, "train_loss_bc": 0.033798765391111374, "train_loss_llm": 0.0, "grad_norm": 0.10170914977788925, "global_step": 23467, "epoch": 651, "lr": 0.001} +{"train_loss": 0.031185537576675415, "train_loss_bc": 0.031182993203401566, "train_loss_llm": 0.0002543926239013672, "grad_norm": 0.1245640218257904, "global_step": 23468, "epoch": 651, "lr": 0.001} +{"train_loss": 0.034563127905130386, "train_loss_bc": 0.03456192836165428, "train_loss_llm": 0.00011992454528808594, "grad_norm": 0.03540169075131416, "global_step": 23469, "epoch": 651, "lr": 0.001} +{"train_loss": 0.025677556172013283, "train_loss_bc": 0.02566804178059101, "train_loss_llm": 0.0009514689445495605, "grad_norm": 0.06902264803647995, "global_step": 23470, "epoch": 651, "lr": 0.001} +{"train_loss": 0.032147177339841924, "train_loss_bc": 0.02937030792236328, "train_loss_llm": 0.0, "grad_norm": 0.09712410718202591, "global_step": 23471, "epoch": 651, "lr": 0.001, "val_loss": 0.0855490118265152} +{"train_loss": 0.032716818153858185, "train_loss_bc": 0.03269956633448601, "train_loss_llm": 0.0017251372337341309, "grad_norm": 0.13180485367774963, "global_step": 23472, "epoch": 652, "lr": 0.001} +{"train_loss": 0.031319282948970795, "train_loss_bc": 0.031319282948970795, "train_loss_llm": 0.0, "grad_norm": 0.028313517570495605, "global_step": 23473, "epoch": 652, "lr": 0.001} +{"train_loss": 0.035782117396593094, "train_loss_bc": 0.035782117396593094, "train_loss_llm": 0.0, "grad_norm": 0.053961873054504395, "global_step": 23474, "epoch": 652, "lr": 0.001} +{"train_loss": 0.02857286110520363, "train_loss_bc": 0.028557296842336655, "train_loss_llm": 0.0015564560890197754, "grad_norm": 0.06587137281894684, "global_step": 23475, "epoch": 652, "lr": 0.001} +{"train_loss": 0.028552023693919182, "train_loss_bc": 0.028552023693919182, "train_loss_llm": 0.0, "grad_norm": 0.07876866310834885, "global_step": 23476, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03111952356994152, "train_loss_bc": 0.03111952356994152, "train_loss_llm": 0.0, "grad_norm": 0.02643759548664093, "global_step": 23477, "epoch": 652, "lr": 0.001} +{"train_loss": 0.036292385309934616, "train_loss_bc": 0.036292385309934616, "train_loss_llm": 0.0, "grad_norm": 0.043083254247903824, "global_step": 23478, "epoch": 652, "lr": 0.001} +{"train_loss": 0.029360605403780937, "train_loss_bc": 0.029350128024816513, "train_loss_llm": 0.0010478198528289795, "grad_norm": 0.05831994116306305, "global_step": 23479, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03100631572306156, "train_loss_bc": 0.03100631572306156, "train_loss_llm": 0.0, "grad_norm": 0.0784372016787529, "global_step": 23480, "epoch": 652, "lr": 0.001} +{"train_loss": 0.0368737168610096, "train_loss_bc": 0.0368737168610096, "train_loss_llm": 0.0, "grad_norm": 0.03963615000247955, "global_step": 23481, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03294998034834862, "train_loss_bc": 0.03294998034834862, "train_loss_llm": 0.0, "grad_norm": 0.058615099638700485, "global_step": 23482, "epoch": 652, "lr": 0.001} +{"train_loss": 0.030742090195417404, "train_loss_bc": 0.030742090195417404, "train_loss_llm": 0.0, "grad_norm": 0.08969230204820633, "global_step": 23483, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03804965317249298, "train_loss_bc": 0.03804965317249298, "train_loss_llm": 0.0, "grad_norm": 0.11628881841897964, "global_step": 23484, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03193319961428642, "train_loss_bc": 0.03193319961428642, "train_loss_llm": 0.0, "grad_norm": 0.043369464576244354, "global_step": 23485, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03229733556509018, "train_loss_bc": 0.03229733556509018, "train_loss_llm": 0.0, "grad_norm": 0.06913456320762634, "global_step": 23486, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03649727627635002, "train_loss_bc": 0.03649727627635002, "train_loss_llm": 0.0, "grad_norm": 0.10608029365539551, "global_step": 23487, "epoch": 652, "lr": 0.001} +{"train_loss": 0.0382741317152977, "train_loss_bc": 0.03826165571808815, "train_loss_llm": 0.0012475550174713135, "grad_norm": 0.13254332542419434, "global_step": 23488, "epoch": 652, "lr": 0.001} +{"train_loss": 0.02456270530819893, "train_loss_bc": 0.02456270530819893, "train_loss_llm": 0.0, "grad_norm": 0.03172194957733154, "global_step": 23489, "epoch": 652, "lr": 0.001} +{"train_loss": 0.02909254841506481, "train_loss_bc": 0.02909254841506481, "train_loss_llm": 0.0, "grad_norm": 0.051242146641016006, "global_step": 23490, "epoch": 652, "lr": 0.001} +{"train_loss": 0.02598660998046398, "train_loss_bc": 0.02598660998046398, "train_loss_llm": 0.0, "grad_norm": 0.06913787871599197, "global_step": 23491, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03429553285241127, "train_loss_bc": 0.03429553285241127, "train_loss_llm": 0.0, "grad_norm": 0.08645080029964447, "global_step": 23492, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03324824199080467, "train_loss_bc": 0.03324824199080467, "train_loss_llm": 0.0, "grad_norm": 0.026881394907832146, "global_step": 23493, "epoch": 652, "lr": 0.001} +{"train_loss": 0.02892116829752922, "train_loss_bc": 0.02892116829752922, "train_loss_llm": 0.0, "grad_norm": 0.04111802577972412, "global_step": 23494, "epoch": 652, "lr": 0.001} +{"train_loss": 0.029994195327162743, "train_loss_bc": 0.029994195327162743, "train_loss_llm": 0.0, "grad_norm": 0.06214381381869316, "global_step": 23495, "epoch": 652, "lr": 0.001} +{"train_loss": 0.033875685185194016, "train_loss_bc": 0.033866025507450104, "train_loss_llm": 0.0009661018848419189, "grad_norm": 0.07902055233716965, "global_step": 23496, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03224385529756546, "train_loss_bc": 0.03224208578467369, "train_loss_llm": 0.00017711520195007324, "grad_norm": 0.03358045965433121, "global_step": 23497, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03253341466188431, "train_loss_bc": 0.03252941370010376, "train_loss_llm": 0.0004000663757324219, "grad_norm": 0.04797767102718353, "global_step": 23498, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03749643266201019, "train_loss_bc": 0.03749643266201019, "train_loss_llm": 0.0, "grad_norm": 0.07713888585567474, "global_step": 23499, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03284386172890663, "train_loss_bc": 0.03284386172890663, "train_loss_llm": 0.0, "grad_norm": 0.10111664235591888, "global_step": 23500, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03308512270450592, "train_loss_bc": 0.03308512270450592, "train_loss_llm": 0.0, "grad_norm": 0.031186364591121674, "global_step": 23501, "epoch": 652, "lr": 0.001} +{"train_loss": 0.032404836267232895, "train_loss_bc": 0.032404836267232895, "train_loss_llm": 0.0, "grad_norm": 0.05339399352669716, "global_step": 23502, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03634043037891388, "train_loss_bc": 0.0363326258957386, "train_loss_llm": 0.0007803738117218018, "grad_norm": 0.08363655209541321, "global_step": 23503, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03235567733645439, "train_loss_bc": 0.03235567733645439, "train_loss_llm": 0.0, "grad_norm": 0.11018111556768417, "global_step": 23504, "epoch": 652, "lr": 0.001} +{"train_loss": 0.031337056308984756, "train_loss_bc": 0.03133590891957283, "train_loss_llm": 0.00011482834815979004, "grad_norm": 0.03837893530726433, "global_step": 23505, "epoch": 652, "lr": 0.001} +{"train_loss": 0.023857897147536278, "train_loss_bc": 0.023841703310608864, "train_loss_llm": 0.0016193985939025879, "grad_norm": 0.06298099458217621, "global_step": 23506, "epoch": 652, "lr": 0.001} +{"train_loss": 0.03232362483524614, "train_loss_bc": 0.03682474046945572, "train_loss_llm": 0.0011165142059326172, "grad_norm": 0.08499689400196075, "global_step": 23507, "epoch": 652, "lr": 0.001, "val_loss": 0.08446693420410156} +{"train_loss": 0.02867899276316166, "train_loss_bc": 0.028660474345088005, "train_loss_llm": 0.0018518269062042236, "grad_norm": 0.10641340166330338, "global_step": 23508, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03344271332025528, "train_loss_bc": 0.03344271332025528, "train_loss_llm": 0.0, "grad_norm": 0.03253283351659775, "global_step": 23509, "epoch": 653, "lr": 0.001} +{"train_loss": 0.0355125330388546, "train_loss_bc": 0.035507313907146454, "train_loss_llm": 0.0005218684673309326, "grad_norm": 0.04982079938054085, "global_step": 23510, "epoch": 653, "lr": 0.001} +{"train_loss": 0.028276974335312843, "train_loss_bc": 0.028276974335312843, "train_loss_llm": 0.0, "grad_norm": 0.06328998506069183, "global_step": 23511, "epoch": 653, "lr": 0.001} +{"train_loss": 0.034115687012672424, "train_loss_bc": 0.0341109037399292, "train_loss_llm": 0.00047835707664489746, "grad_norm": 0.09823548793792725, "global_step": 23512, "epoch": 653, "lr": 0.001} +{"train_loss": 0.033927157521247864, "train_loss_bc": 0.03390505537390709, "train_loss_llm": 0.0022100508213043213, "grad_norm": 0.03266220539808273, "global_step": 23513, "epoch": 653, "lr": 0.001} +{"train_loss": 0.032144706696271896, "train_loss_bc": 0.032144706696271896, "train_loss_llm": 0.0, "grad_norm": 0.0537353940308094, "global_step": 23514, "epoch": 653, "lr": 0.001} +{"train_loss": 0.035144489258527756, "train_loss_bc": 0.035144489258527756, "train_loss_llm": 0.0, "grad_norm": 0.0798841193318367, "global_step": 23515, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03007793240249157, "train_loss_bc": 0.03007793240249157, "train_loss_llm": 0.0, "grad_norm": 0.1020963042974472, "global_step": 23516, "epoch": 653, "lr": 0.001} +{"train_loss": 0.028446577489376068, "train_loss_bc": 0.028445400297641754, "train_loss_llm": 0.00011774897575378418, "grad_norm": 0.02664433978497982, "global_step": 23517, "epoch": 653, "lr": 0.001} +{"train_loss": 0.02893526665866375, "train_loss_bc": 0.02893526665866375, "train_loss_llm": 0.0, "grad_norm": 0.05162734165787697, "global_step": 23518, "epoch": 653, "lr": 0.001} +{"train_loss": 0.030691152438521385, "train_loss_bc": 0.030691152438521385, "train_loss_llm": 0.0, "grad_norm": 0.06958996504545212, "global_step": 23519, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03251131623983383, "train_loss_bc": 0.03251131623983383, "train_loss_llm": 0.0, "grad_norm": 0.08889679610729218, "global_step": 23520, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03051777370274067, "train_loss_bc": 0.03051777370274067, "train_loss_llm": 0.0, "grad_norm": 0.04163477569818497, "global_step": 23521, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03136902302503586, "train_loss_bc": 0.03136902302503586, "train_loss_llm": 0.0, "grad_norm": 0.06153995543718338, "global_step": 23522, "epoch": 653, "lr": 0.001} +{"train_loss": 0.028726736083626747, "train_loss_bc": 0.028726736083626747, "train_loss_llm": 0.0, "grad_norm": 0.0704067125916481, "global_step": 23523, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03254528343677521, "train_loss_bc": 0.03254528343677521, "train_loss_llm": 0.0, "grad_norm": 0.087193064391613, "global_step": 23524, "epoch": 653, "lr": 0.001} +{"train_loss": 0.032845690846443176, "train_loss_bc": 0.032845690846443176, "train_loss_llm": 0.0, "grad_norm": 0.029594993218779564, "global_step": 23525, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03190992400050163, "train_loss_bc": 0.03190992400050163, "train_loss_llm": 0.0, "grad_norm": 0.055585213005542755, "global_step": 23526, "epoch": 653, "lr": 0.001} +{"train_loss": 0.023494431748986244, "train_loss_bc": 0.023494431748986244, "train_loss_llm": 0.0, "grad_norm": 0.07099496573209763, "global_step": 23527, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03324317932128906, "train_loss_bc": 0.03324317932128906, "train_loss_llm": 0.0, "grad_norm": 0.09635014832019806, "global_step": 23528, "epoch": 653, "lr": 0.001} +{"train_loss": 0.033965833485126495, "train_loss_bc": 0.033965833485126495, "train_loss_llm": 0.0, "grad_norm": 0.039008259773254395, "global_step": 23529, "epoch": 653, "lr": 0.001} +{"train_loss": 0.032400406897068024, "train_loss_bc": 0.032400406897068024, "train_loss_llm": 0.0, "grad_norm": 0.06850193440914154, "global_step": 23530, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03865205496549606, "train_loss_bc": 0.03865205496549606, "train_loss_llm": 0.0, "grad_norm": 0.1033790335059166, "global_step": 23531, "epoch": 653, "lr": 0.001} +{"train_loss": 0.034422796219587326, "train_loss_bc": 0.034422796219587326, "train_loss_llm": 0.0, "grad_norm": 0.14308775961399078, "global_step": 23532, "epoch": 653, "lr": 0.001} +{"train_loss": 0.034820303320884705, "train_loss_bc": 0.034820303320884705, "train_loss_llm": 0.0, "grad_norm": 0.04164908826351166, "global_step": 23533, "epoch": 653, "lr": 0.001} +{"train_loss": 0.033882468938827515, "train_loss_bc": 0.03387441113591194, "train_loss_llm": 0.0008058249950408936, "grad_norm": 0.06236984208226204, "global_step": 23534, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03600199520587921, "train_loss_bc": 0.035981982946395874, "train_loss_llm": 0.0020012259483337402, "grad_norm": 0.07914283871650696, "global_step": 23535, "epoch": 653, "lr": 0.001} +{"train_loss": 0.030911562964320183, "train_loss_bc": 0.030911562964320183, "train_loss_llm": 0.0, "grad_norm": 0.10137932747602463, "global_step": 23536, "epoch": 653, "lr": 0.001} +{"train_loss": 0.030521254986524582, "train_loss_bc": 0.030506378039717674, "train_loss_llm": 0.0014877617359161377, "grad_norm": 0.041046392172575, "global_step": 23537, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03252607211470604, "train_loss_bc": 0.03252607211470604, "train_loss_llm": 0.0, "grad_norm": 0.06308130919933319, "global_step": 23538, "epoch": 653, "lr": 0.001} +{"train_loss": 0.02850821614265442, "train_loss_bc": 0.02850821614265442, "train_loss_llm": 0.0, "grad_norm": 0.07027847319841385, "global_step": 23539, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03081779181957245, "train_loss_bc": 0.03081779181957245, "train_loss_llm": 0.0, "grad_norm": 0.0850563794374466, "global_step": 23540, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03570788353681564, "train_loss_bc": 0.03570788353681564, "train_loss_llm": 0.0, "grad_norm": 0.03126624599099159, "global_step": 23541, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03575034439563751, "train_loss_bc": 0.03575034439563751, "train_loss_llm": 0.0, "grad_norm": 0.040247783064842224, "global_step": 23542, "epoch": 653, "lr": 0.001} +{"train_loss": 0.03229003384088477, "train_loss_bc": 0.0369911715388298, "train_loss_llm": 0.0003520846366882324, "grad_norm": 0.07076502591371536, "global_step": 23543, "epoch": 653, "lr": 0.001, "val_loss": 0.0846792683005333} +{"train_loss": 0.031809158623218536, "train_loss_bc": 0.03180734068155289, "train_loss_llm": 0.0001817643642425537, "grad_norm": 0.09089446812868118, "global_step": 23544, "epoch": 654, "lr": 0.001} +{"train_loss": 0.028560083359479904, "train_loss_bc": 0.028560083359479904, "train_loss_llm": 0.0, "grad_norm": 0.04046686738729477, "global_step": 23545, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03266887366771698, "train_loss_bc": 0.03266887366771698, "train_loss_llm": 0.0, "grad_norm": 0.06013215705752373, "global_step": 23546, "epoch": 654, "lr": 0.001} +{"train_loss": 0.029455827549099922, "train_loss_bc": 0.029455827549099922, "train_loss_llm": 0.0, "grad_norm": 0.08480033278465271, "global_step": 23547, "epoch": 654, "lr": 0.001} +{"train_loss": 0.030238769948482513, "train_loss_bc": 0.030238769948482513, "train_loss_llm": 0.0, "grad_norm": 0.11174990236759186, "global_step": 23548, "epoch": 654, "lr": 0.001} +{"train_loss": 0.02999960072338581, "train_loss_bc": 0.02999960072338581, "train_loss_llm": 0.0, "grad_norm": 0.03221839293837547, "global_step": 23549, "epoch": 654, "lr": 0.001} +{"train_loss": 0.029491638764739037, "train_loss_bc": 0.029464343562722206, "train_loss_llm": 0.002729475498199463, "grad_norm": 0.053417179733514786, "global_step": 23550, "epoch": 654, "lr": 0.001} +{"train_loss": 0.028402792289853096, "train_loss_bc": 0.02838253788650036, "train_loss_llm": 0.0020254552364349365, "grad_norm": 0.08310329914093018, "global_step": 23551, "epoch": 654, "lr": 0.001} +{"train_loss": 0.0320284329354763, "train_loss_bc": 0.0320284329354763, "train_loss_llm": 0.0, "grad_norm": 0.11400651931762695, "global_step": 23552, "epoch": 654, "lr": 0.001} +{"train_loss": 0.02987072244286537, "train_loss_bc": 0.02987072244286537, "train_loss_llm": 0.0, "grad_norm": 0.04037351533770561, "global_step": 23553, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03298959881067276, "train_loss_bc": 0.03298959881067276, "train_loss_llm": 0.0, "grad_norm": 0.06508620083332062, "global_step": 23554, "epoch": 654, "lr": 0.001} +{"train_loss": 0.035357698798179626, "train_loss_bc": 0.03534325957298279, "train_loss_llm": 0.001444011926651001, "grad_norm": 0.09959253668785095, "global_step": 23555, "epoch": 654, "lr": 0.001} +{"train_loss": 0.04093153029680252, "train_loss_bc": 0.04093153029680252, "train_loss_llm": 0.0, "grad_norm": 0.13936825096607208, "global_step": 23556, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03596847876906395, "train_loss_bc": 0.03596847876906395, "train_loss_llm": 0.0, "grad_norm": 0.04341001436114311, "global_step": 23557, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03250940516591072, "train_loss_bc": 0.03250940516591072, "train_loss_llm": 0.0, "grad_norm": 0.06662388890981674, "global_step": 23558, "epoch": 654, "lr": 0.001} +{"train_loss": 0.032570980489254, "train_loss_bc": 0.032570980489254, "train_loss_llm": 0.0, "grad_norm": 0.07300467789173126, "global_step": 23559, "epoch": 654, "lr": 0.001} +{"train_loss": 0.02786310203373432, "train_loss_bc": 0.02786310203373432, "train_loss_llm": 0.0, "grad_norm": 0.08885190635919571, "global_step": 23560, "epoch": 654, "lr": 0.001} +{"train_loss": 0.036001525819301605, "train_loss_bc": 0.036001525819301605, "train_loss_llm": 0.0, "grad_norm": 0.03684324398636818, "global_step": 23561, "epoch": 654, "lr": 0.001} +{"train_loss": 0.031863465905189514, "train_loss_bc": 0.031863465905189514, "train_loss_llm": 0.0, "grad_norm": 0.05216611549258232, "global_step": 23562, "epoch": 654, "lr": 0.001} +{"train_loss": 0.032727256417274475, "train_loss_bc": 0.03270940110087395, "train_loss_llm": 0.0017855167388916016, "grad_norm": 0.07701566070318222, "global_step": 23563, "epoch": 654, "lr": 0.001} +{"train_loss": 0.033757440745830536, "train_loss_bc": 0.03375343233346939, "train_loss_llm": 0.00040087103843688965, "grad_norm": 0.09631240367889404, "global_step": 23564, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03350691869854927, "train_loss_bc": 0.033500541001558304, "train_loss_llm": 0.0006378591060638428, "grad_norm": 0.037128206342458725, "global_step": 23565, "epoch": 654, "lr": 0.001} +{"train_loss": 0.028634056448936462, "train_loss_bc": 0.028634056448936462, "train_loss_llm": 0.0, "grad_norm": 0.06471352279186249, "global_step": 23566, "epoch": 654, "lr": 0.001} +{"train_loss": 0.037613511085510254, "train_loss_bc": 0.037609733641147614, "train_loss_llm": 0.0003775954246520996, "grad_norm": 0.09306184947490692, "global_step": 23567, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03644483909010887, "train_loss_bc": 0.03644483909010887, "train_loss_llm": 0.0, "grad_norm": 0.12425881624221802, "global_step": 23568, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03184602037072182, "train_loss_bc": 0.03184602037072182, "train_loss_llm": 0.0, "grad_norm": 0.029027454555034637, "global_step": 23569, "epoch": 654, "lr": 0.001} +{"train_loss": 0.036450669169425964, "train_loss_bc": 0.0364399291574955, "train_loss_llm": 0.001073986291885376, "grad_norm": 0.0735463872551918, "global_step": 23570, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03731102496385574, "train_loss_bc": 0.037298962473869324, "train_loss_llm": 0.0012063086032867432, "grad_norm": 0.09842628985643387, "global_step": 23571, "epoch": 654, "lr": 0.001} +{"train_loss": 0.029330089688301086, "train_loss_bc": 0.029330089688301086, "train_loss_llm": 0.0, "grad_norm": 0.12111889570951462, "global_step": 23572, "epoch": 654, "lr": 0.001} +{"train_loss": 0.02824478968977928, "train_loss_bc": 0.02824478968977928, "train_loss_llm": 0.0, "grad_norm": 0.025922372937202454, "global_step": 23573, "epoch": 654, "lr": 0.001} +{"train_loss": 0.028063304722309113, "train_loss_bc": 0.028063304722309113, "train_loss_llm": 0.0, "grad_norm": 0.043757952749729156, "global_step": 23574, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03226228430867195, "train_loss_bc": 0.03226228430867195, "train_loss_llm": 0.0, "grad_norm": 0.0744088888168335, "global_step": 23575, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03421301394701004, "train_loss_bc": 0.03421301394701004, "train_loss_llm": 0.0, "grad_norm": 0.09540882706642151, "global_step": 23576, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03273160755634308, "train_loss_bc": 0.03273160755634308, "train_loss_llm": 0.0, "grad_norm": 0.03734714165329933, "global_step": 23577, "epoch": 654, "lr": 0.001} +{"train_loss": 0.029505014419555664, "train_loss_bc": 0.029505014419555664, "train_loss_llm": 0.0, "grad_norm": 0.05986648052930832, "global_step": 23578, "epoch": 654, "lr": 0.001} +{"train_loss": 0.03246009831006328, "train_loss_bc": 0.037316642701625824, "train_loss_llm": 0.0023369193077087402, "grad_norm": 0.09985591471195221, "global_step": 23579, "epoch": 654, "lr": 0.001, "val_loss": 0.08563277870416641} +{"train_loss": 0.03052815981209278, "train_loss_bc": 0.03052815981209278, "train_loss_llm": 0.0, "grad_norm": 0.12753824889659882, "global_step": 23580, "epoch": 655, "lr": 0.001} +{"train_loss": 0.027229875326156616, "train_loss_bc": 0.027229875326156616, "train_loss_llm": 0.0, "grad_norm": 0.02474878542125225, "global_step": 23581, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03250863030552864, "train_loss_bc": 0.03248210623860359, "train_loss_llm": 0.0026525557041168213, "grad_norm": 0.04233720898628235, "global_step": 23582, "epoch": 655, "lr": 0.001} +{"train_loss": 0.031362198293209076, "train_loss_bc": 0.03135228529572487, "train_loss_llm": 0.0009913146495819092, "grad_norm": 0.05983736738562584, "global_step": 23583, "epoch": 655, "lr": 0.001} +{"train_loss": 0.033120296895504, "train_loss_bc": 0.033120296895504, "train_loss_llm": 0.0, "grad_norm": 0.08069147914648056, "global_step": 23584, "epoch": 655, "lr": 0.001} +{"train_loss": 0.0335589274764061, "train_loss_bc": 0.0335589274764061, "train_loss_llm": 0.0, "grad_norm": 0.04534098133444786, "global_step": 23585, "epoch": 655, "lr": 0.001} +{"train_loss": 0.033118389546871185, "train_loss_bc": 0.033118389546871185, "train_loss_llm": 0.0, "grad_norm": 0.06952967494726181, "global_step": 23586, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03246298432350159, "train_loss_bc": 0.032453037798404694, "train_loss_llm": 0.0009945333003997803, "grad_norm": 0.09966710954904556, "global_step": 23587, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03146088495850563, "train_loss_bc": 0.03146088495850563, "train_loss_llm": 0.0, "grad_norm": 0.11264932155609131, "global_step": 23588, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03310869634151459, "train_loss_bc": 0.03310869634151459, "train_loss_llm": 0.0, "grad_norm": 0.03941638767719269, "global_step": 23589, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03242453560233116, "train_loss_bc": 0.03242453560233116, "train_loss_llm": 0.0, "grad_norm": 0.07240588217973709, "global_step": 23590, "epoch": 655, "lr": 0.001} +{"train_loss": 0.031260717660188675, "train_loss_bc": 0.031260717660188675, "train_loss_llm": 0.0, "grad_norm": 0.08896350860595703, "global_step": 23591, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03214448690414429, "train_loss_bc": 0.032134443521499634, "train_loss_llm": 0.0010044574737548828, "grad_norm": 0.12584936618804932, "global_step": 23592, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03338589891791344, "train_loss_bc": 0.033382583409547806, "train_loss_llm": 0.00033161044120788574, "grad_norm": 0.0379759706556797, "global_step": 23593, "epoch": 655, "lr": 0.001} +{"train_loss": 0.029097866266965866, "train_loss_bc": 0.029097866266965866, "train_loss_llm": 0.0, "grad_norm": 0.06072014197707176, "global_step": 23594, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03771587461233139, "train_loss_bc": 0.03771587461233139, "train_loss_llm": 0.0, "grad_norm": 0.0967942625284195, "global_step": 23595, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03386765718460083, "train_loss_bc": 0.03386765718460083, "train_loss_llm": 0.0, "grad_norm": 0.12038197368383408, "global_step": 23596, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03255101665854454, "train_loss_bc": 0.032547760754823685, "train_loss_llm": 0.000325620174407959, "grad_norm": 0.03118436224758625, "global_step": 23597, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03346656262874603, "train_loss_bc": 0.033455539494752884, "train_loss_llm": 0.0011022984981536865, "grad_norm": 0.04541155695915222, "global_step": 23598, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03203391283750534, "train_loss_bc": 0.03203033655881882, "train_loss_llm": 0.00035765767097473145, "grad_norm": 0.06464333087205887, "global_step": 23599, "epoch": 655, "lr": 0.001} +{"train_loss": 0.02908547781407833, "train_loss_bc": 0.02908547781407833, "train_loss_llm": 0.0, "grad_norm": 0.08458470553159714, "global_step": 23600, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03201689571142197, "train_loss_bc": 0.031995825469493866, "train_loss_llm": 0.0021069347858428955, "grad_norm": 0.02796911634504795, "global_step": 23601, "epoch": 655, "lr": 0.001} +{"train_loss": 0.028156118467450142, "train_loss_bc": 0.028156118467450142, "train_loss_llm": 0.0, "grad_norm": 0.044538579881191254, "global_step": 23602, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03296046704053879, "train_loss_bc": 0.03296046704053879, "train_loss_llm": 0.0, "grad_norm": 0.05881952866911888, "global_step": 23603, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03222054988145828, "train_loss_bc": 0.03222054988145828, "train_loss_llm": 0.0, "grad_norm": 0.07173240929841995, "global_step": 23604, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03292388468980789, "train_loss_bc": 0.032896410673856735, "train_loss_llm": 0.0027475357055664062, "grad_norm": 0.03697340190410614, "global_step": 23605, "epoch": 655, "lr": 0.001} +{"train_loss": 0.04088224843144417, "train_loss_bc": 0.04088224843144417, "train_loss_llm": 0.0, "grad_norm": 0.06110209971666336, "global_step": 23606, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03220167011022568, "train_loss_bc": 0.03220167011022568, "train_loss_llm": 0.0, "grad_norm": 0.09052486717700958, "global_step": 23607, "epoch": 655, "lr": 0.001} +{"train_loss": 0.032274726778268814, "train_loss_bc": 0.032274726778268814, "train_loss_llm": 0.0, "grad_norm": 0.11408819258213043, "global_step": 23608, "epoch": 655, "lr": 0.001} +{"train_loss": 0.031294405460357666, "train_loss_bc": 0.031294405460357666, "train_loss_llm": 0.0, "grad_norm": 0.03555756062269211, "global_step": 23609, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03586112707853317, "train_loss_bc": 0.035844381898641586, "train_loss_llm": 0.0016745328903198242, "grad_norm": 0.0655762255191803, "global_step": 23610, "epoch": 655, "lr": 0.001} +{"train_loss": 0.03168575465679169, "train_loss_bc": 0.03168575465679169, "train_loss_llm": 0.0, "grad_norm": 0.10061155259609222, "global_step": 23611, "epoch": 655, "lr": 0.001} +{"train_loss": 0.029485275968909264, "train_loss_bc": 0.029485275968909264, "train_loss_llm": 0.0, "grad_norm": 0.13315363228321075, "global_step": 23612, "epoch": 655, "lr": 0.001} +{"train_loss": 0.029835479333996773, "train_loss_bc": 0.029835479333996773, "train_loss_llm": 0.0, "grad_norm": 0.037696074694395065, "global_step": 23613, "epoch": 655, "lr": 0.001} +{"train_loss": 0.034837376326322556, "train_loss_bc": 0.034834302961826324, "train_loss_llm": 0.00030750036239624023, "grad_norm": 0.07060859352350235, "global_step": 23614, "epoch": 655, "lr": 0.001} +{"train_loss": 0.032182484658228025, "train_loss_bc": 0.026440417394042015, "train_loss_llm": 0.0, "grad_norm": 0.09131012856960297, "global_step": 23615, "epoch": 655, "lr": 0.001, "train/cumulative_reward": 91.00195505006074, "train/mean_score": 12.882833836036388, "train/success_rate": 0.3, "test/cumulative_reward": 33.65354274768483, "test/mean_score": 8.227185522092482, "test/success_rate": 0.22, "val_loss": 0.08666443824768066, "train_action_mse_error": 0.03608991578221321} +{"train_loss": 0.02821039967238903, "train_loss_bc": 0.02821039967238903, "train_loss_llm": 0.0, "grad_norm": 0.11060269922018051, "global_step": 23616, "epoch": 656, "lr": 0.001} +{"train_loss": 0.033356960862874985, "train_loss_bc": 0.03333587199449539, "train_loss_llm": 0.002108752727508545, "grad_norm": 0.030305936932563782, "global_step": 23617, "epoch": 656, "lr": 0.001} +{"train_loss": 0.031031345948576927, "train_loss_bc": 0.031031345948576927, "train_loss_llm": 0.0, "grad_norm": 0.05170988664031029, "global_step": 23618, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03173729032278061, "train_loss_bc": 0.03173729032278061, "train_loss_llm": 0.0, "grad_norm": 0.06413400173187256, "global_step": 23619, "epoch": 656, "lr": 0.001} +{"train_loss": 0.028057480230927467, "train_loss_bc": 0.028048621490597725, "train_loss_llm": 0.0008858144283294678, "grad_norm": 0.07573701441287994, "global_step": 23620, "epoch": 656, "lr": 0.001} +{"train_loss": 0.028897935524582863, "train_loss_bc": 0.028897935524582863, "train_loss_llm": 0.0, "grad_norm": 0.02741820551455021, "global_step": 23621, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03264138847589493, "train_loss_bc": 0.03264138847589493, "train_loss_llm": 0.0, "grad_norm": 0.041530296206474304, "global_step": 23622, "epoch": 656, "lr": 0.001} +{"train_loss": 0.02969585917890072, "train_loss_bc": 0.02969585917890072, "train_loss_llm": 0.0, "grad_norm": 0.05320402607321739, "global_step": 23623, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03517969697713852, "train_loss_bc": 0.03517969697713852, "train_loss_llm": 0.0, "grad_norm": 0.06984790414571762, "global_step": 23624, "epoch": 656, "lr": 0.001} +{"train_loss": 0.029177922755479813, "train_loss_bc": 0.02917705848813057, "train_loss_llm": 8.64863395690918e-05, "grad_norm": 0.03811319172382355, "global_step": 23625, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03402954339981079, "train_loss_bc": 0.03402795270085335, "train_loss_llm": 0.00015920400619506836, "grad_norm": 0.05884663015604019, "global_step": 23626, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03230070322751999, "train_loss_bc": 0.03230070322751999, "train_loss_llm": 0.0, "grad_norm": 0.0776662826538086, "global_step": 23627, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03010023757815361, "train_loss_bc": 0.03010023757815361, "train_loss_llm": 0.0, "grad_norm": 0.10636802017688751, "global_step": 23628, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03611934557557106, "train_loss_bc": 0.03611934557557106, "train_loss_llm": 0.0, "grad_norm": 0.04360366240143776, "global_step": 23629, "epoch": 656, "lr": 0.001} +{"train_loss": 0.02979259192943573, "train_loss_bc": 0.02979259192943573, "train_loss_llm": 0.0, "grad_norm": 0.07596316933631897, "global_step": 23630, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03308870270848274, "train_loss_bc": 0.033084988594055176, "train_loss_llm": 0.00037154555320739746, "grad_norm": 0.10763877630233765, "global_step": 23631, "epoch": 656, "lr": 0.001} +{"train_loss": 0.029145898297429085, "train_loss_bc": 0.029145898297429085, "train_loss_llm": 0.0, "grad_norm": 0.13221518695354462, "global_step": 23632, "epoch": 656, "lr": 0.001} +{"train_loss": 0.035198528319597244, "train_loss_bc": 0.035197094082832336, "train_loss_llm": 0.00014355778694152832, "grad_norm": 0.029349008575081825, "global_step": 23633, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03377487137913704, "train_loss_bc": 0.03377487137913704, "train_loss_llm": 0.0, "grad_norm": 0.05229560285806656, "global_step": 23634, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03273763135075569, "train_loss_bc": 0.03273763135075569, "train_loss_llm": 0.0, "grad_norm": 0.0757468119263649, "global_step": 23635, "epoch": 656, "lr": 0.001} +{"train_loss": 0.0326661579310894, "train_loss_bc": 0.0326661579310894, "train_loss_llm": 0.0, "grad_norm": 0.0929829403758049, "global_step": 23636, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03276767581701279, "train_loss_bc": 0.03276767581701279, "train_loss_llm": 0.0, "grad_norm": 0.0315861739218235, "global_step": 23637, "epoch": 656, "lr": 0.001} +{"train_loss": 0.02856089174747467, "train_loss_bc": 0.02856089174747467, "train_loss_llm": 0.0, "grad_norm": 0.041421908885240555, "global_step": 23638, "epoch": 656, "lr": 0.001} +{"train_loss": 0.030946141108870506, "train_loss_bc": 0.030946141108870506, "train_loss_llm": 0.0, "grad_norm": 0.05192241817712784, "global_step": 23639, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03056425414979458, "train_loss_bc": 0.030559802427887917, "train_loss_llm": 0.0004451572895050049, "grad_norm": 0.06348951905965805, "global_step": 23640, "epoch": 656, "lr": 0.001} +{"train_loss": 0.0330343134701252, "train_loss_bc": 0.0330343134701252, "train_loss_llm": 0.0, "grad_norm": 0.035089168697595596, "global_step": 23641, "epoch": 656, "lr": 0.001} +{"train_loss": 0.031649596989154816, "train_loss_bc": 0.031624361872673035, "train_loss_llm": 0.002523481845855713, "grad_norm": 0.05170702189207077, "global_step": 23642, "epoch": 656, "lr": 0.001} +{"train_loss": 0.036620575934648514, "train_loss_bc": 0.036620575934648514, "train_loss_llm": 0.0, "grad_norm": 0.06490758061408997, "global_step": 23643, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03544589877128601, "train_loss_bc": 0.03544589877128601, "train_loss_llm": 0.0, "grad_norm": 0.07886119186878204, "global_step": 23644, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03309891000390053, "train_loss_bc": 0.033079613000154495, "train_loss_llm": 0.0019297897815704346, "grad_norm": 0.04290859028697014, "global_step": 23645, "epoch": 656, "lr": 0.001} +{"train_loss": 0.0373874269425869, "train_loss_bc": 0.037374719977378845, "train_loss_llm": 0.001270592212677002, "grad_norm": 0.07785109430551529, "global_step": 23646, "epoch": 656, "lr": 0.001} +{"train_loss": 0.034237246960401535, "train_loss_bc": 0.034237246960401535, "train_loss_llm": 0.0, "grad_norm": 0.1002642884850502, "global_step": 23647, "epoch": 656, "lr": 0.001} +{"train_loss": 0.02984747290611267, "train_loss_bc": 0.029837943613529205, "train_loss_llm": 0.0009528994560241699, "grad_norm": 0.1236891895532608, "global_step": 23648, "epoch": 656, "lr": 0.001} +{"train_loss": 0.03337472304701805, "train_loss_bc": 0.03337472304701805, "train_loss_llm": 0.0, "grad_norm": 0.034864772111177444, "global_step": 23649, "epoch": 656, "lr": 0.001} +{"train_loss": 0.02761995606124401, "train_loss_bc": 0.02761995606124401, "train_loss_llm": 0.0, "grad_norm": 0.05032668635249138, "global_step": 23650, "epoch": 656, "lr": 0.001} +{"train_loss": 0.031949597928259105, "train_loss_bc": 0.02808104269206524, "train_loss_llm": 0.0008907318115234375, "grad_norm": 0.07724887132644653, "global_step": 23651, "epoch": 656, "lr": 0.001, "val_loss": 0.08748909831047058} +{"train_loss": 0.0291601549834013, "train_loss_bc": 0.0291601549834013, "train_loss_llm": 0.0, "grad_norm": 0.10282441228628159, "global_step": 23652, "epoch": 657, "lr": 0.001} +{"train_loss": 0.033824313431978226, "train_loss_bc": 0.033824313431978226, "train_loss_llm": 0.0, "grad_norm": 0.03827688470482826, "global_step": 23653, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03471563011407852, "train_loss_bc": 0.03471563011407852, "train_loss_llm": 0.0, "grad_norm": 0.06168896704912186, "global_step": 23654, "epoch": 657, "lr": 0.001} +{"train_loss": 0.027996907010674477, "train_loss_bc": 0.027976447716355324, "train_loss_llm": 0.0020460188388824463, "grad_norm": 0.10283517092466354, "global_step": 23655, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03090125508606434, "train_loss_bc": 0.03090125508606434, "train_loss_llm": 0.0, "grad_norm": 0.12737198173999786, "global_step": 23656, "epoch": 657, "lr": 0.001} +{"train_loss": 0.032400354743003845, "train_loss_bc": 0.032400354743003845, "train_loss_llm": 0.0, "grad_norm": 0.035736918449401855, "global_step": 23657, "epoch": 657, "lr": 0.001} +{"train_loss": 0.02971310168504715, "train_loss_bc": 0.02971310168504715, "train_loss_llm": 0.0, "grad_norm": 0.052931249141693115, "global_step": 23658, "epoch": 657, "lr": 0.001} +{"train_loss": 0.02301780879497528, "train_loss_bc": 0.02301780879497528, "train_loss_llm": 0.0, "grad_norm": 0.060413915663957596, "global_step": 23659, "epoch": 657, "lr": 0.001} +{"train_loss": 0.02908107452094555, "train_loss_bc": 0.02907586097717285, "train_loss_llm": 0.0005213022232055664, "grad_norm": 0.06750565767288208, "global_step": 23660, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03548319637775421, "train_loss_bc": 0.03548319637775421, "train_loss_llm": 0.0, "grad_norm": 0.03326207771897316, "global_step": 23661, "epoch": 657, "lr": 0.001} +{"train_loss": 0.028178073465824127, "train_loss_bc": 0.028178073465824127, "train_loss_llm": 0.0, "grad_norm": 0.03740473836660385, "global_step": 23662, "epoch": 657, "lr": 0.001} +{"train_loss": 0.028036922216415405, "train_loss_bc": 0.028036922216415405, "train_loss_llm": 0.0, "grad_norm": 0.049620386213064194, "global_step": 23663, "epoch": 657, "lr": 0.001} +{"train_loss": 0.029319724068045616, "train_loss_bc": 0.029319724068045616, "train_loss_llm": 0.0, "grad_norm": 0.05896686017513275, "global_step": 23664, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03506554663181305, "train_loss_bc": 0.03506217151880264, "train_loss_llm": 0.0003376305103302002, "grad_norm": 0.03689806908369064, "global_step": 23665, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03621198609471321, "train_loss_bc": 0.03621198609471321, "train_loss_llm": 0.0, "grad_norm": 0.0615423321723938, "global_step": 23666, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03302354738116264, "train_loss_bc": 0.03302354738116264, "train_loss_llm": 0.0, "grad_norm": 0.08211251348257065, "global_step": 23667, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03318795561790466, "train_loss_bc": 0.0331820547580719, "train_loss_llm": 0.0005901157855987549, "grad_norm": 0.11668156832456589, "global_step": 23668, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03371461480855942, "train_loss_bc": 0.03371461480855942, "train_loss_llm": 0.0, "grad_norm": 0.03599056228995323, "global_step": 23669, "epoch": 657, "lr": 0.001} +{"train_loss": 0.030843673273921013, "train_loss_bc": 0.030843673273921013, "train_loss_llm": 0.0, "grad_norm": 0.05600041523575783, "global_step": 23670, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03244409337639809, "train_loss_bc": 0.032435014843940735, "train_loss_llm": 0.0009077787399291992, "grad_norm": 0.08531273156404495, "global_step": 23671, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03372954577207565, "train_loss_bc": 0.03371838107705116, "train_loss_llm": 0.0011163949966430664, "grad_norm": 0.12166944891214371, "global_step": 23672, "epoch": 657, "lr": 0.001} +{"train_loss": 0.036119136959314346, "train_loss_bc": 0.036119136959314346, "train_loss_llm": 0.0, "grad_norm": 0.039263226091861725, "global_step": 23673, "epoch": 657, "lr": 0.001} +{"train_loss": 0.02892274595797062, "train_loss_bc": 0.02892274595797062, "train_loss_llm": 0.0, "grad_norm": 0.057191021740436554, "global_step": 23674, "epoch": 657, "lr": 0.001} +{"train_loss": 0.0256380345672369, "train_loss_bc": 0.0256380345672369, "train_loss_llm": 0.0, "grad_norm": 0.07646588236093521, "global_step": 23675, "epoch": 657, "lr": 0.001} +{"train_loss": 0.036387618631124496, "train_loss_bc": 0.036387618631124496, "train_loss_llm": 0.0, "grad_norm": 0.1076681911945343, "global_step": 23676, "epoch": 657, "lr": 0.001} +{"train_loss": 0.032862503081560135, "train_loss_bc": 0.032862503081560135, "train_loss_llm": 0.0, "grad_norm": 0.0318460687994957, "global_step": 23677, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03477051481604576, "train_loss_bc": 0.03477051481604576, "train_loss_llm": 0.0, "grad_norm": 0.054737918078899384, "global_step": 23678, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03811876103281975, "train_loss_bc": 0.038113731890916824, "train_loss_llm": 0.0005029439926147461, "grad_norm": 0.08286167681217194, "global_step": 23679, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03534076362848282, "train_loss_bc": 0.03534076362848282, "train_loss_llm": 0.0, "grad_norm": 0.0994635745882988, "global_step": 23680, "epoch": 657, "lr": 0.001} +{"train_loss": 0.027133753523230553, "train_loss_bc": 0.02712327428162098, "train_loss_llm": 0.0010478496551513672, "grad_norm": 0.025129521265625954, "global_step": 23681, "epoch": 657, "lr": 0.001} +{"train_loss": 0.030920052900910378, "train_loss_bc": 0.030920052900910378, "train_loss_llm": 0.0, "grad_norm": 0.03974462300539017, "global_step": 23682, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03175878897309303, "train_loss_bc": 0.03175878897309303, "train_loss_llm": 0.0, "grad_norm": 0.04655923321843147, "global_step": 23683, "epoch": 657, "lr": 0.001} +{"train_loss": 0.034945376217365265, "train_loss_bc": 0.03492876514792442, "train_loss_llm": 0.0016611814498901367, "grad_norm": 0.06514792889356613, "global_step": 23684, "epoch": 657, "lr": 0.001} +{"train_loss": 0.033013682812452316, "train_loss_bc": 0.033013682812452316, "train_loss_llm": 0.0, "grad_norm": 0.034464478492736816, "global_step": 23685, "epoch": 657, "lr": 0.001} +{"train_loss": 0.03250202536582947, "train_loss_bc": 0.03250202536582947, "train_loss_llm": 0.0, "grad_norm": 0.05165800452232361, "global_step": 23686, "epoch": 657, "lr": 0.001} +{"train_loss": 0.032085205842223435, "train_loss_bc": 0.03655364364385605, "train_loss_llm": 0.0030528008937835693, "grad_norm": 0.08940301835536957, "global_step": 23687, "epoch": 657, "lr": 0.001, "val_loss": 0.08734767138957977} +{"train_loss": 0.03613533079624176, "train_loss_bc": 0.03613533079624176, "train_loss_llm": 0.0, "grad_norm": 0.1214471310377121, "global_step": 23688, "epoch": 658, "lr": 0.001} +{"train_loss": 0.026702985167503357, "train_loss_bc": 0.026702985167503357, "train_loss_llm": 0.0, "grad_norm": 0.031150424852967262, "global_step": 23689, "epoch": 658, "lr": 0.001} +{"train_loss": 0.024017248302698135, "train_loss_bc": 0.024017248302698135, "train_loss_llm": 0.0, "grad_norm": 0.05344421789050102, "global_step": 23690, "epoch": 658, "lr": 0.001} +{"train_loss": 0.030385049059987068, "train_loss_bc": 0.03037060983479023, "train_loss_llm": 0.0014439523220062256, "grad_norm": 0.07423003017902374, "global_step": 23691, "epoch": 658, "lr": 0.001} +{"train_loss": 0.030935876071453094, "train_loss_bc": 0.030935876071453094, "train_loss_llm": 0.0, "grad_norm": 0.08226849883794785, "global_step": 23692, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03134768456220627, "train_loss_bc": 0.03134768456220627, "train_loss_llm": 0.0, "grad_norm": 0.0373896062374115, "global_step": 23693, "epoch": 658, "lr": 0.001} +{"train_loss": 0.033712975680828094, "train_loss_bc": 0.03370879590511322, "train_loss_llm": 0.00041791796684265137, "grad_norm": 0.05526811257004738, "global_step": 23694, "epoch": 658, "lr": 0.001} +{"train_loss": 0.026610152795910835, "train_loss_bc": 0.026610152795910835, "train_loss_llm": 0.0, "grad_norm": 0.07141892611980438, "global_step": 23695, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03617092967033386, "train_loss_bc": 0.03617092967033386, "train_loss_llm": 0.0, "grad_norm": 0.09642568975687027, "global_step": 23696, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03226703032851219, "train_loss_bc": 0.03226703032851219, "train_loss_llm": 0.0, "grad_norm": 0.04286026209592819, "global_step": 23697, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03754368796944618, "train_loss_bc": 0.03754368796944618, "train_loss_llm": 0.0, "grad_norm": 0.06129879876971245, "global_step": 23698, "epoch": 658, "lr": 0.001} +{"train_loss": 0.0295629370957613, "train_loss_bc": 0.0295629370957613, "train_loss_llm": 0.0, "grad_norm": 0.08264529705047607, "global_step": 23699, "epoch": 658, "lr": 0.001} +{"train_loss": 0.02811896800994873, "train_loss_bc": 0.028108233585953712, "train_loss_llm": 0.0010734796524047852, "grad_norm": 0.09289263188838959, "global_step": 23700, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03234955668449402, "train_loss_bc": 0.03234955668449402, "train_loss_llm": 0.0, "grad_norm": 0.02756139263510704, "global_step": 23701, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03817109018564224, "train_loss_bc": 0.03817109018564224, "train_loss_llm": 0.0, "grad_norm": 0.056714631617069244, "global_step": 23702, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03362279012799263, "train_loss_bc": 0.03362279012799263, "train_loss_llm": 0.0, "grad_norm": 0.07936681061983109, "global_step": 23703, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03345368057489395, "train_loss_bc": 0.03345368057489395, "train_loss_llm": 0.0, "grad_norm": 0.09759628027677536, "global_step": 23704, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03514742851257324, "train_loss_bc": 0.03514742851257324, "train_loss_llm": 0.0, "grad_norm": 0.03154705464839935, "global_step": 23705, "epoch": 658, "lr": 0.001} +{"train_loss": 0.035658471286296844, "train_loss_bc": 0.035658471286296844, "train_loss_llm": 0.0, "grad_norm": 0.051977671682834625, "global_step": 23706, "epoch": 658, "lr": 0.001} +{"train_loss": 0.026173004880547523, "train_loss_bc": 0.026173004880547523, "train_loss_llm": 0.0, "grad_norm": 0.06998372077941895, "global_step": 23707, "epoch": 658, "lr": 0.001} +{"train_loss": 0.025822779163718224, "train_loss_bc": 0.025822779163718224, "train_loss_llm": 0.0, "grad_norm": 0.09062701463699341, "global_step": 23708, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03561103716492653, "train_loss_bc": 0.03560733050107956, "train_loss_llm": 0.0003705024719238281, "grad_norm": 0.032450173050165176, "global_step": 23709, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03152424842119217, "train_loss_bc": 0.03152424842119217, "train_loss_llm": 0.0, "grad_norm": 0.05945485830307007, "global_step": 23710, "epoch": 658, "lr": 0.001} +{"train_loss": 0.029137039557099342, "train_loss_bc": 0.029137039557099342, "train_loss_llm": 0.0, "grad_norm": 0.07473190128803253, "global_step": 23711, "epoch": 658, "lr": 0.001} +{"train_loss": 0.029159456491470337, "train_loss_bc": 0.0291549451649189, "train_loss_llm": 0.00045108795166015625, "grad_norm": 0.09361695498228073, "global_step": 23712, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03390676900744438, "train_loss_bc": 0.03390676900744438, "train_loss_llm": 0.0, "grad_norm": 0.042720645666122437, "global_step": 23713, "epoch": 658, "lr": 0.001} +{"train_loss": 0.0262327641248703, "train_loss_bc": 0.0262327641248703, "train_loss_llm": 0.0, "grad_norm": 0.06318585574626923, "global_step": 23714, "epoch": 658, "lr": 0.001} +{"train_loss": 0.037228360772132874, "train_loss_bc": 0.037228360772132874, "train_loss_llm": 0.0, "grad_norm": 0.08256521821022034, "global_step": 23715, "epoch": 658, "lr": 0.001} +{"train_loss": 0.030368926003575325, "train_loss_bc": 0.030368926003575325, "train_loss_llm": 0.0, "grad_norm": 0.09301815181970596, "global_step": 23716, "epoch": 658, "lr": 0.001} +{"train_loss": 0.037440113723278046, "train_loss_bc": 0.03742636367678642, "train_loss_llm": 0.0013748407363891602, "grad_norm": 0.04172267019748688, "global_step": 23717, "epoch": 658, "lr": 0.001} +{"train_loss": 0.035076502710580826, "train_loss_bc": 0.03507203608751297, "train_loss_llm": 0.00044661760330200195, "grad_norm": 0.06906761974096298, "global_step": 23718, "epoch": 658, "lr": 0.001} +{"train_loss": 0.02849002182483673, "train_loss_bc": 0.028483694419264793, "train_loss_llm": 0.0006328225135803223, "grad_norm": 0.10237312316894531, "global_step": 23719, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03248325735330582, "train_loss_bc": 0.03246958926320076, "train_loss_llm": 0.001366943120956421, "grad_norm": 0.1341029405593872, "global_step": 23720, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03609741851687431, "train_loss_bc": 0.03609594330191612, "train_loss_llm": 0.00014737248420715332, "grad_norm": 0.03352704644203186, "global_step": 23721, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03118620067834854, "train_loss_bc": 0.031185414642095566, "train_loss_llm": 7.855892181396484e-05, "grad_norm": 0.05502510070800781, "global_step": 23722, "epoch": 658, "lr": 0.001} +{"train_loss": 0.03211994940001103, "train_loss_bc": 0.038466405123472214, "train_loss_llm": 0.0, "grad_norm": 0.08847982436418533, "global_step": 23723, "epoch": 658, "lr": 0.001, "val_loss": 0.08689364045858383} +{"train_loss": 0.030976327136158943, "train_loss_bc": 0.030963126569986343, "train_loss_llm": 0.0013200938701629639, "grad_norm": 0.10891085118055344, "global_step": 23724, "epoch": 659, "lr": 0.001} +{"train_loss": 0.027264779433608055, "train_loss_bc": 0.027258511632680893, "train_loss_llm": 0.0006267428398132324, "grad_norm": 0.02859729342162609, "global_step": 23725, "epoch": 659, "lr": 0.001} +{"train_loss": 0.0310214851051569, "train_loss_bc": 0.031018303707242012, "train_loss_llm": 0.00031822919845581055, "grad_norm": 0.044747184962034225, "global_step": 23726, "epoch": 659, "lr": 0.001} +{"train_loss": 0.031031325459480286, "train_loss_bc": 0.031031325459480286, "train_loss_llm": 0.0, "grad_norm": 0.06630066782236099, "global_step": 23727, "epoch": 659, "lr": 0.001} +{"train_loss": 0.030972642824053764, "train_loss_bc": 0.030972642824053764, "train_loss_llm": 0.0, "grad_norm": 0.08690965920686722, "global_step": 23728, "epoch": 659, "lr": 0.001} +{"train_loss": 0.032618265599012375, "train_loss_bc": 0.032618265599012375, "train_loss_llm": 0.0, "grad_norm": 0.04498017206788063, "global_step": 23729, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03507711738348007, "train_loss_bc": 0.03507711738348007, "train_loss_llm": 0.0, "grad_norm": 0.07346230745315552, "global_step": 23730, "epoch": 659, "lr": 0.001} +{"train_loss": 0.02891436591744423, "train_loss_bc": 0.02891436591744423, "train_loss_llm": 0.0, "grad_norm": 0.08083546161651611, "global_step": 23731, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03307603299617767, "train_loss_bc": 0.03307603299617767, "train_loss_llm": 0.0, "grad_norm": 0.10814238339662552, "global_step": 23732, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03368457406759262, "train_loss_bc": 0.03368457406759262, "train_loss_llm": 0.0, "grad_norm": 0.03454430773854256, "global_step": 23733, "epoch": 659, "lr": 0.001} +{"train_loss": 0.026504207402467728, "train_loss_bc": 0.026504207402467728, "train_loss_llm": 0.0, "grad_norm": 0.0541524700820446, "global_step": 23734, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03301505371928215, "train_loss_bc": 0.03301505371928215, "train_loss_llm": 0.0, "grad_norm": 0.08023330569267273, "global_step": 23735, "epoch": 659, "lr": 0.001} +{"train_loss": 0.02966264635324478, "train_loss_bc": 0.02966170199215412, "train_loss_llm": 9.441375732421875e-05, "grad_norm": 0.10529990494251251, "global_step": 23736, "epoch": 659, "lr": 0.001} +{"train_loss": 0.032439038157463074, "train_loss_bc": 0.032439038157463074, "train_loss_llm": 0.0, "grad_norm": 0.03549504652619362, "global_step": 23737, "epoch": 659, "lr": 0.001} +{"train_loss": 0.031547896564006805, "train_loss_bc": 0.031539395451545715, "train_loss_llm": 0.0008501112461090088, "grad_norm": 0.04591120406985283, "global_step": 23738, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03314739093184471, "train_loss_bc": 0.0331471748650074, "train_loss_llm": 2.1755695343017578e-05, "grad_norm": 0.06114203855395317, "global_step": 23739, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03850468993186951, "train_loss_bc": 0.03849221020936966, "train_loss_llm": 0.0012480318546295166, "grad_norm": 0.09003926813602448, "global_step": 23740, "epoch": 659, "lr": 0.001} +{"train_loss": 0.028644178062677383, "train_loss_bc": 0.028644178062677383, "train_loss_llm": 0.0, "grad_norm": 0.03486783802509308, "global_step": 23741, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03320641443133354, "train_loss_bc": 0.03320641443133354, "train_loss_llm": 0.0, "grad_norm": 0.054848793894052505, "global_step": 23742, "epoch": 659, "lr": 0.001} +{"train_loss": 0.030036818236112595, "train_loss_bc": 0.030031288042664528, "train_loss_llm": 0.0005530714988708496, "grad_norm": 0.07334630191326141, "global_step": 23743, "epoch": 659, "lr": 0.001} +{"train_loss": 0.027846230193972588, "train_loss_bc": 0.027846230193972588, "train_loss_llm": 0.0, "grad_norm": 0.09308057278394699, "global_step": 23744, "epoch": 659, "lr": 0.001} +{"train_loss": 0.029348449781537056, "train_loss_bc": 0.029348449781537056, "train_loss_llm": 0.0, "grad_norm": 0.03257725015282631, "global_step": 23745, "epoch": 659, "lr": 0.001} +{"train_loss": 0.0311060082167387, "train_loss_bc": 0.0311060082167387, "train_loss_llm": 0.0, "grad_norm": 0.05150990188121796, "global_step": 23746, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03346941992640495, "train_loss_bc": 0.03345761075615883, "train_loss_llm": 0.0011808276176452637, "grad_norm": 0.07942907512187958, "global_step": 23747, "epoch": 659, "lr": 0.001} +{"train_loss": 0.031109189614653587, "train_loss_bc": 0.031102042645215988, "train_loss_llm": 0.0007147490978240967, "grad_norm": 0.1011556014418602, "global_step": 23748, "epoch": 659, "lr": 0.001} +{"train_loss": 0.036656349897384644, "train_loss_bc": 0.036656349897384644, "train_loss_llm": 0.0, "grad_norm": 0.03652413561940193, "global_step": 23749, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03268667310476303, "train_loss_bc": 0.032678041607141495, "train_loss_llm": 0.000863194465637207, "grad_norm": 0.0581669919192791, "global_step": 23750, "epoch": 659, "lr": 0.001} +{"train_loss": 0.026850996538996696, "train_loss_bc": 0.026850996538996696, "train_loss_llm": 0.0, "grad_norm": 0.07911998778581619, "global_step": 23751, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03836045414209366, "train_loss_bc": 0.03835654631257057, "train_loss_llm": 0.000390857458114624, "grad_norm": 0.09781720489263535, "global_step": 23752, "epoch": 659, "lr": 0.001} +{"train_loss": 0.028208257630467415, "train_loss_bc": 0.028208257630467415, "train_loss_llm": 0.0, "grad_norm": 0.030582966282963753, "global_step": 23753, "epoch": 659, "lr": 0.001} +{"train_loss": 0.041101742535829544, "train_loss_bc": 0.041101742535829544, "train_loss_llm": 0.0, "grad_norm": 0.05782699212431908, "global_step": 23754, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03524194285273552, "train_loss_bc": 0.03524194285273552, "train_loss_llm": 0.0, "grad_norm": 0.0730922520160675, "global_step": 23755, "epoch": 659, "lr": 0.001} +{"train_loss": 0.029808318242430687, "train_loss_bc": 0.029777400195598602, "train_loss_llm": 0.003091782331466675, "grad_norm": 0.08532131463289261, "global_step": 23756, "epoch": 659, "lr": 0.001} +{"train_loss": 0.03788908198475838, "train_loss_bc": 0.03788908198475838, "train_loss_llm": 0.0, "grad_norm": 0.03855651989579201, "global_step": 23757, "epoch": 659, "lr": 0.001} +{"train_loss": 0.029800720512866974, "train_loss_bc": 0.029800720512866974, "train_loss_llm": 0.0, "grad_norm": 0.05285707488656044, "global_step": 23758, "epoch": 659, "lr": 0.001} +{"train_loss": 0.032102011816783085, "train_loss_bc": 0.03483709320425987, "train_loss_llm": 0.0006247758865356445, "grad_norm": 0.07612477242946625, "global_step": 23759, "epoch": 659, "lr": 0.001, "val_loss": 0.08516772091388702} +{"train_loss": 0.036307550966739655, "train_loss_bc": 0.036307550966739655, "train_loss_llm": 0.0, "grad_norm": 0.10090998560190201, "global_step": 23760, "epoch": 660, "lr": 0.001} +{"train_loss": 0.027458971366286278, "train_loss_bc": 0.027458971366286278, "train_loss_llm": 0.0, "grad_norm": 0.035224542021751404, "global_step": 23761, "epoch": 660, "lr": 0.001} +{"train_loss": 0.032114237546920776, "train_loss_bc": 0.032103169709444046, "train_loss_llm": 0.0011069178581237793, "grad_norm": 0.053365010768175125, "global_step": 23762, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03270452097058296, "train_loss_bc": 0.03270452097058296, "train_loss_llm": 0.0, "grad_norm": 0.07027583569288254, "global_step": 23763, "epoch": 660, "lr": 0.001} +{"train_loss": 0.030562782660126686, "train_loss_bc": 0.030562782660126686, "train_loss_llm": 0.0, "grad_norm": 0.09525209665298462, "global_step": 23764, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03360519930720329, "train_loss_bc": 0.03360519930720329, "train_loss_llm": 0.0, "grad_norm": 0.03774727135896683, "global_step": 23765, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03221147507429123, "train_loss_bc": 0.032200682908296585, "train_loss_llm": 0.0010790526866912842, "grad_norm": 0.051932454109191895, "global_step": 23766, "epoch": 660, "lr": 0.001} +{"train_loss": 0.030176984146237373, "train_loss_bc": 0.030176984146237373, "train_loss_llm": 0.0, "grad_norm": 0.06413668394088745, "global_step": 23767, "epoch": 660, "lr": 0.001} +{"train_loss": 0.031239228323101997, "train_loss_bc": 0.031239228323101997, "train_loss_llm": 0.0, "grad_norm": 0.07512983679771423, "global_step": 23768, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03215177729725838, "train_loss_bc": 0.03212838992476463, "train_loss_llm": 0.0023388564586639404, "grad_norm": 0.033093519508838654, "global_step": 23769, "epoch": 660, "lr": 0.001} +{"train_loss": 0.028003666549921036, "train_loss_bc": 0.028003666549921036, "train_loss_llm": 0.0, "grad_norm": 0.0564267672598362, "global_step": 23770, "epoch": 660, "lr": 0.001} +{"train_loss": 0.0318184569478035, "train_loss_bc": 0.0318184569478035, "train_loss_llm": 0.0, "grad_norm": 0.06710667908191681, "global_step": 23771, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03410026431083679, "train_loss_bc": 0.03410026431083679, "train_loss_llm": 0.0, "grad_norm": 0.09595605731010437, "global_step": 23772, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03127781301736832, "train_loss_bc": 0.03127781301736832, "train_loss_llm": 0.0, "grad_norm": 0.02498384565114975, "global_step": 23773, "epoch": 660, "lr": 0.001} +{"train_loss": 0.031099990010261536, "train_loss_bc": 0.031099990010261536, "train_loss_llm": 0.0, "grad_norm": 0.04695170745253563, "global_step": 23774, "epoch": 660, "lr": 0.001} +{"train_loss": 0.036126576364040375, "train_loss_bc": 0.036126576364040375, "train_loss_llm": 0.0, "grad_norm": 0.07153996080160141, "global_step": 23775, "epoch": 660, "lr": 0.001} +{"train_loss": 0.02954411692917347, "train_loss_bc": 0.02954411692917347, "train_loss_llm": 0.0, "grad_norm": 0.0912899523973465, "global_step": 23776, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03419920802116394, "train_loss_bc": 0.03419920802116394, "train_loss_llm": 0.0, "grad_norm": 0.03337779641151428, "global_step": 23777, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03375208005309105, "train_loss_bc": 0.03375208005309105, "train_loss_llm": 0.0, "grad_norm": 0.05190345272421837, "global_step": 23778, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03830081596970558, "train_loss_bc": 0.03828030452132225, "train_loss_llm": 0.002051055431365967, "grad_norm": 0.06791303306818008, "global_step": 23779, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03395269811153412, "train_loss_bc": 0.03395269811153412, "train_loss_llm": 0.0, "grad_norm": 0.08260700851678848, "global_step": 23780, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03355202451348305, "train_loss_bc": 0.03355202451348305, "train_loss_llm": 0.0, "grad_norm": 0.030761752277612686, "global_step": 23781, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03549402579665184, "train_loss_bc": 0.03549402579665184, "train_loss_llm": 0.0, "grad_norm": 0.0516657717525959, "global_step": 23782, "epoch": 660, "lr": 0.001} +{"train_loss": 0.035916782915592194, "train_loss_bc": 0.035916782915592194, "train_loss_llm": 0.0, "grad_norm": 0.06977897137403488, "global_step": 23783, "epoch": 660, "lr": 0.001} +{"train_loss": 0.030089110136032104, "train_loss_bc": 0.030089110136032104, "train_loss_llm": 0.0, "grad_norm": 0.08554822951555252, "global_step": 23784, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03620064631104469, "train_loss_bc": 0.03620064631104469, "train_loss_llm": 0.0, "grad_norm": 0.036688800901174545, "global_step": 23785, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03058566339313984, "train_loss_bc": 0.03058566339313984, "train_loss_llm": 0.0, "grad_norm": 0.054493580013513565, "global_step": 23786, "epoch": 660, "lr": 0.001} +{"train_loss": 0.030283216387033463, "train_loss_bc": 0.030283216387033463, "train_loss_llm": 0.0, "grad_norm": 0.07217986136674881, "global_step": 23787, "epoch": 660, "lr": 0.001} +{"train_loss": 0.029387401416897774, "train_loss_bc": 0.029387401416897774, "train_loss_llm": 0.0, "grad_norm": 0.08618898689746857, "global_step": 23788, "epoch": 660, "lr": 0.001} +{"train_loss": 0.031578559428453445, "train_loss_bc": 0.031578559428453445, "train_loss_llm": 0.0, "grad_norm": 0.035301852971315384, "global_step": 23789, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03385774418711662, "train_loss_bc": 0.03385774418711662, "train_loss_llm": 0.0, "grad_norm": 0.05990726500749588, "global_step": 23790, "epoch": 660, "lr": 0.001} +{"train_loss": 0.029464298859238625, "train_loss_bc": 0.029464298859238625, "train_loss_llm": 0.0, "grad_norm": 0.07970090210437775, "global_step": 23791, "epoch": 660, "lr": 0.001} +{"train_loss": 0.022325247526168823, "train_loss_bc": 0.022325247526168823, "train_loss_llm": 0.0, "grad_norm": 0.10450741648674011, "global_step": 23792, "epoch": 660, "lr": 0.001} +{"train_loss": 0.02804088033735752, "train_loss_bc": 0.028031744062900543, "train_loss_llm": 0.0009136199951171875, "grad_norm": 0.023123228922486305, "global_step": 23793, "epoch": 660, "lr": 0.001} +{"train_loss": 0.029953664168715477, "train_loss_bc": 0.029953664168715477, "train_loss_llm": 0.0, "grad_norm": 0.0378909669816494, "global_step": 23794, "epoch": 660, "lr": 0.001} +{"train_loss": 0.03200497581726975, "train_loss_bc": 0.03472452238202095, "train_loss_llm": 0.0016926229000091553, "grad_norm": 0.05749029666185379, "global_step": 23795, "epoch": 660, "lr": 0.001, "train/cumulative_reward": 68.45366901486712, "train/mean_score": 16.92096681471929, "train/success_rate": 0.4, "test/cumulative_reward": 37.99658113251919, "test/mean_score": 8.646034746213806, "test/success_rate": 0.22, "val_loss": 0.08524809032678604, "train_action_mse_error": 0.03491697087883949} +{"train_loss": 0.030726075172424316, "train_loss_bc": 0.030726075172424316, "train_loss_llm": 0.0, "grad_norm": 0.06767607480287552, "global_step": 23796, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03198517858982086, "train_loss_bc": 0.03198517858982086, "train_loss_llm": 0.0, "grad_norm": 0.0401243157684803, "global_step": 23797, "epoch": 661, "lr": 0.001} +{"train_loss": 0.033634692430496216, "train_loss_bc": 0.033622369170188904, "train_loss_llm": 0.0012323558330535889, "grad_norm": 0.05591851845383644, "global_step": 23798, "epoch": 661, "lr": 0.001} +{"train_loss": 0.033413179218769073, "train_loss_bc": 0.03339661657810211, "train_loss_llm": 0.0016562342643737793, "grad_norm": 0.07911314070224762, "global_step": 23799, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03239509463310242, "train_loss_bc": 0.03239509463310242, "train_loss_llm": 0.0, "grad_norm": 0.09799961745738983, "global_step": 23800, "epoch": 661, "lr": 0.001} +{"train_loss": 0.0266833808273077, "train_loss_bc": 0.026676032692193985, "train_loss_llm": 0.0007348358631134033, "grad_norm": 0.029582485556602478, "global_step": 23801, "epoch": 661, "lr": 0.001} +{"train_loss": 0.030781395733356476, "train_loss_bc": 0.030781395733356476, "train_loss_llm": 0.0, "grad_norm": 0.048971544951200485, "global_step": 23802, "epoch": 661, "lr": 0.001} +{"train_loss": 0.033020488917827606, "train_loss_bc": 0.03301233798265457, "train_loss_llm": 0.0008152425289154053, "grad_norm": 0.054990001022815704, "global_step": 23803, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03438775613903999, "train_loss_bc": 0.03438775613903999, "train_loss_llm": 0.0, "grad_norm": 0.06614068895578384, "global_step": 23804, "epoch": 661, "lr": 0.001} +{"train_loss": 0.02927802875638008, "train_loss_bc": 0.029262911528348923, "train_loss_llm": 0.0015117526054382324, "grad_norm": 0.028588207438588142, "global_step": 23805, "epoch": 661, "lr": 0.001} +{"train_loss": 0.035210803151130676, "train_loss_bc": 0.035210803151130676, "train_loss_llm": 0.0, "grad_norm": 0.05197017267346382, "global_step": 23806, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03474990278482437, "train_loss_bc": 0.034741662442684174, "train_loss_llm": 0.0008238554000854492, "grad_norm": 0.07371189445257187, "global_step": 23807, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03002338297665119, "train_loss_bc": 0.030021902173757553, "train_loss_llm": 0.0001480579376220703, "grad_norm": 0.08573473989963531, "global_step": 23808, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03129120543599129, "train_loss_bc": 0.03125942870974541, "train_loss_llm": 0.003177732229232788, "grad_norm": 0.03300367668271065, "global_step": 23809, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03337162733078003, "train_loss_bc": 0.03337162733078003, "train_loss_llm": 0.0, "grad_norm": 0.049877818673849106, "global_step": 23810, "epoch": 661, "lr": 0.001} +{"train_loss": 0.029003843665122986, "train_loss_bc": 0.029003843665122986, "train_loss_llm": 0.0, "grad_norm": 0.06095414236187935, "global_step": 23811, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03434595465660095, "train_loss_bc": 0.03434595465660095, "train_loss_llm": 0.0, "grad_norm": 0.07876467704772949, "global_step": 23812, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03549870476126671, "train_loss_bc": 0.03549870476126671, "train_loss_llm": 0.0, "grad_norm": 0.03126465901732445, "global_step": 23813, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03000396490097046, "train_loss_bc": 0.03000396490097046, "train_loss_llm": 0.0, "grad_norm": 0.04695197194814682, "global_step": 23814, "epoch": 661, "lr": 0.001} +{"train_loss": 0.034251127392053604, "train_loss_bc": 0.034251127392053604, "train_loss_llm": 0.0, "grad_norm": 0.06455212831497192, "global_step": 23815, "epoch": 661, "lr": 0.001} +{"train_loss": 0.026356594637036324, "train_loss_bc": 0.026356594637036324, "train_loss_llm": 0.0, "grad_norm": 0.08034097403287888, "global_step": 23816, "epoch": 661, "lr": 0.001} +{"train_loss": 0.032713744789361954, "train_loss_bc": 0.032713744789361954, "train_loss_llm": 0.0, "grad_norm": 0.03488248959183693, "global_step": 23817, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03749749809503555, "train_loss_bc": 0.03747980669140816, "train_loss_llm": 0.001769036054611206, "grad_norm": 0.05018903687596321, "global_step": 23818, "epoch": 661, "lr": 0.001} +{"train_loss": 0.02298603020608425, "train_loss_bc": 0.02298603020608425, "train_loss_llm": 0.0, "grad_norm": 0.06848343461751938, "global_step": 23819, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03242399916052818, "train_loss_bc": 0.03242399916052818, "train_loss_llm": 0.0, "grad_norm": 0.09021046757698059, "global_step": 23820, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03557944297790527, "train_loss_bc": 0.03557944297790527, "train_loss_llm": 0.0, "grad_norm": 0.026488007977604866, "global_step": 23821, "epoch": 661, "lr": 0.001} +{"train_loss": 0.028735384345054626, "train_loss_bc": 0.028735384345054626, "train_loss_llm": 0.0, "grad_norm": 0.040275052189826965, "global_step": 23822, "epoch": 661, "lr": 0.001} +{"train_loss": 0.02636886201798916, "train_loss_bc": 0.02636284939944744, "train_loss_llm": 0.0006012022495269775, "grad_norm": 0.057986047118902206, "global_step": 23823, "epoch": 661, "lr": 0.001} +{"train_loss": 0.030391747131943703, "train_loss_bc": 0.030391747131943703, "train_loss_llm": 0.0, "grad_norm": 0.07402448356151581, "global_step": 23824, "epoch": 661, "lr": 0.001} +{"train_loss": 0.035291314125061035, "train_loss_bc": 0.035291314125061035, "train_loss_llm": 0.0, "grad_norm": 0.034434232860803604, "global_step": 23825, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03016408532857895, "train_loss_bc": 0.03016408532857895, "train_loss_llm": 0.0, "grad_norm": 0.05177268385887146, "global_step": 23826, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03597265109419823, "train_loss_bc": 0.03597265109419823, "train_loss_llm": 0.0, "grad_norm": 0.06458865106105804, "global_step": 23827, "epoch": 661, "lr": 0.001} +{"train_loss": 0.03258242458105087, "train_loss_bc": 0.03258242458105087, "train_loss_llm": 0.0, "grad_norm": 0.07481507956981659, "global_step": 23828, "epoch": 661, "lr": 0.001} +{"train_loss": 0.031964316964149475, "train_loss_bc": 0.03195470571517944, "train_loss_llm": 0.000961005687713623, "grad_norm": 0.02978392504155636, "global_step": 23829, "epoch": 661, "lr": 0.001} +{"train_loss": 0.02821827121078968, "train_loss_bc": 0.02821827121078968, "train_loss_llm": 0.0, "grad_norm": 0.04753050208091736, "global_step": 23830, "epoch": 661, "lr": 0.001} +{"train_loss": 0.031918841465893716, "train_loss_bc": 0.03777613863348961, "train_loss_llm": 0.0, "grad_norm": 0.07554391771554947, "global_step": 23831, "epoch": 661, "lr": 0.001, "val_loss": 0.08584681153297424} +{"train_loss": 0.026386573910713196, "train_loss_bc": 0.026386573910713196, "train_loss_llm": 0.0, "grad_norm": 0.0938500463962555, "global_step": 23832, "epoch": 662, "lr": 0.001} +{"train_loss": 0.030821382999420166, "train_loss_bc": 0.030821382999420166, "train_loss_llm": 0.0, "grad_norm": 0.03428127244114876, "global_step": 23833, "epoch": 662, "lr": 0.001} +{"train_loss": 0.04059569537639618, "train_loss_bc": 0.04059569537639618, "train_loss_llm": 0.0, "grad_norm": 0.049105916172266006, "global_step": 23834, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03211138769984245, "train_loss_bc": 0.03211138769984245, "train_loss_llm": 0.0, "grad_norm": 0.07443375885486603, "global_step": 23835, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03454658389091492, "train_loss_bc": 0.03454658389091492, "train_loss_llm": 0.0, "grad_norm": 0.09645325690507889, "global_step": 23836, "epoch": 662, "lr": 0.001} +{"train_loss": 0.033622488379478455, "train_loss_bc": 0.033622488379478455, "train_loss_llm": 0.0, "grad_norm": 0.039855681359767914, "global_step": 23837, "epoch": 662, "lr": 0.001} +{"train_loss": 0.02679150179028511, "train_loss_bc": 0.02679150179028511, "train_loss_llm": 0.0, "grad_norm": 0.05443985387682915, "global_step": 23838, "epoch": 662, "lr": 0.001} +{"train_loss": 0.02521703578531742, "train_loss_bc": 0.02521703578531742, "train_loss_llm": 0.0, "grad_norm": 0.07163245230913162, "global_step": 23839, "epoch": 662, "lr": 0.001} +{"train_loss": 0.030071523040533066, "train_loss_bc": 0.030071523040533066, "train_loss_llm": 0.0, "grad_norm": 0.08233704417943954, "global_step": 23840, "epoch": 662, "lr": 0.001} +{"train_loss": 0.027004359290003777, "train_loss_bc": 0.027004359290003777, "train_loss_llm": 0.0, "grad_norm": 0.02621806040406227, "global_step": 23841, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03583516180515289, "train_loss_bc": 0.03583516180515289, "train_loss_llm": 0.0, "grad_norm": 0.05501055717468262, "global_step": 23842, "epoch": 662, "lr": 0.001} +{"train_loss": 0.02817394584417343, "train_loss_bc": 0.028173178434371948, "train_loss_llm": 7.665157318115234e-05, "grad_norm": 0.07234895974397659, "global_step": 23843, "epoch": 662, "lr": 0.001} +{"train_loss": 0.034951284527778625, "train_loss_bc": 0.03494981676340103, "train_loss_llm": 0.0001468062400817871, "grad_norm": 0.09247838705778122, "global_step": 23844, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03359382227063179, "train_loss_bc": 0.03359382227063179, "train_loss_llm": 0.0, "grad_norm": 0.03154664486646652, "global_step": 23845, "epoch": 662, "lr": 0.001} +{"train_loss": 0.031666893512010574, "train_loss_bc": 0.03166269138455391, "train_loss_llm": 0.0004201233386993408, "grad_norm": 0.04576433077454567, "global_step": 23846, "epoch": 662, "lr": 0.001} +{"train_loss": 0.025726504623889923, "train_loss_bc": 0.025726504623889923, "train_loss_llm": 0.0, "grad_norm": 0.05682935565710068, "global_step": 23847, "epoch": 662, "lr": 0.001} +{"train_loss": 0.024869950488209724, "train_loss_bc": 0.02486356534063816, "train_loss_llm": 0.0006384551525115967, "grad_norm": 0.06343864649534225, "global_step": 23848, "epoch": 662, "lr": 0.001} +{"train_loss": 0.04084732383489609, "train_loss_bc": 0.04084732383489609, "train_loss_llm": 0.0, "grad_norm": 0.033901870250701904, "global_step": 23849, "epoch": 662, "lr": 0.001} +{"train_loss": 0.034921497106552124, "train_loss_bc": 0.034921497106552124, "train_loss_llm": 0.0, "grad_norm": 0.04970868304371834, "global_step": 23850, "epoch": 662, "lr": 0.001} +{"train_loss": 0.031176820397377014, "train_loss_bc": 0.031176811084151268, "train_loss_llm": 9.834766387939453e-07, "grad_norm": 0.08026149868965149, "global_step": 23851, "epoch": 662, "lr": 0.001} +{"train_loss": 0.029290176928043365, "train_loss_bc": 0.029260730370879173, "train_loss_llm": 0.002944648265838623, "grad_norm": 0.09526382386684418, "global_step": 23852, "epoch": 662, "lr": 0.001} +{"train_loss": 0.030384469777345657, "train_loss_bc": 0.030362842604517937, "train_loss_llm": 0.0021626651287078857, "grad_norm": 0.026132915169000626, "global_step": 23853, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03125733882188797, "train_loss_bc": 0.03125733882188797, "train_loss_llm": 0.0, "grad_norm": 0.04966915771365166, "global_step": 23854, "epoch": 662, "lr": 0.001} +{"train_loss": 0.02610950544476509, "train_loss_bc": 0.02610950544476509, "train_loss_llm": 0.0, "grad_norm": 0.06398790329694748, "global_step": 23855, "epoch": 662, "lr": 0.001} +{"train_loss": 0.04090036824345589, "train_loss_bc": 0.040887586772441864, "train_loss_llm": 0.0012780725955963135, "grad_norm": 0.07868500053882599, "global_step": 23856, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03245095908641815, "train_loss_bc": 0.03245095908641815, "train_loss_llm": 0.0, "grad_norm": 0.03151745721697807, "global_step": 23857, "epoch": 662, "lr": 0.001} +{"train_loss": 0.02982291579246521, "train_loss_bc": 0.029804034158587456, "train_loss_llm": 0.001888185739517212, "grad_norm": 0.045733965933322906, "global_step": 23858, "epoch": 662, "lr": 0.001} +{"train_loss": 0.028495054692029953, "train_loss_bc": 0.028495054692029953, "train_loss_llm": 0.0, "grad_norm": 0.06449252367019653, "global_step": 23859, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03738848492503166, "train_loss_bc": 0.03738848492503166, "train_loss_llm": 0.0, "grad_norm": 0.08258523792028427, "global_step": 23860, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03407951816916466, "train_loss_bc": 0.034071143716573715, "train_loss_llm": 0.0008375644683837891, "grad_norm": 0.02813471294939518, "global_step": 23861, "epoch": 662, "lr": 0.001} +{"train_loss": 0.030025362968444824, "train_loss_bc": 0.03001144900918007, "train_loss_llm": 0.0013914704322814941, "grad_norm": 0.054174087941646576, "global_step": 23862, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03769988194108009, "train_loss_bc": 0.037696078419685364, "train_loss_llm": 0.0003802478313446045, "grad_norm": 0.07908666133880615, "global_step": 23863, "epoch": 662, "lr": 0.001} +{"train_loss": 0.032186899334192276, "train_loss_bc": 0.032186899334192276, "train_loss_llm": 0.0, "grad_norm": 0.10467517375946045, "global_step": 23864, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03090451844036579, "train_loss_bc": 0.03090451844036579, "train_loss_llm": 0.0, "grad_norm": 0.03029242530465126, "global_step": 23865, "epoch": 662, "lr": 0.001} +{"train_loss": 0.032099224627017975, "train_loss_bc": 0.032099224627017975, "train_loss_llm": 0.0, "grad_norm": 0.04551353305578232, "global_step": 23866, "epoch": 662, "lr": 0.001} +{"train_loss": 0.03193807426012225, "train_loss_bc": 0.03774425759911537, "train_loss_llm": 0.0, "grad_norm": 0.07598968595266342, "global_step": 23867, "epoch": 662, "lr": 0.001, "val_loss": 0.08610443770885468} +{"train_loss": 0.03266922011971474, "train_loss_bc": 0.03266922011971474, "train_loss_llm": 0.0, "grad_norm": 0.08927012234926224, "global_step": 23868, "epoch": 663, "lr": 0.001} +{"train_loss": 0.036083802580833435, "train_loss_bc": 0.036083802580833435, "train_loss_llm": 0.0, "grad_norm": 0.04109971970319748, "global_step": 23869, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03244961053133011, "train_loss_bc": 0.03244961053133011, "train_loss_llm": 0.0, "grad_norm": 0.055257897824048996, "global_step": 23870, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03615840524435043, "train_loss_bc": 0.03615840524435043, "train_loss_llm": 0.0, "grad_norm": 0.0696929320693016, "global_step": 23871, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03657856583595276, "train_loss_bc": 0.03657856583595276, "train_loss_llm": 0.0, "grad_norm": 0.0952831581234932, "global_step": 23872, "epoch": 663, "lr": 0.001} +{"train_loss": 0.031599219888448715, "train_loss_bc": 0.03159169852733612, "train_loss_llm": 0.0007520318031311035, "grad_norm": 0.037956975400447845, "global_step": 23873, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03004896081984043, "train_loss_bc": 0.03004896081984043, "train_loss_llm": 0.0, "grad_norm": 0.059346023947000504, "global_step": 23874, "epoch": 663, "lr": 0.001} +{"train_loss": 0.029782511293888092, "train_loss_bc": 0.029782511293888092, "train_loss_llm": 0.0, "grad_norm": 0.0753503367304802, "global_step": 23875, "epoch": 663, "lr": 0.001} +{"train_loss": 0.034883540123701096, "train_loss_bc": 0.034883540123701096, "train_loss_llm": 0.0, "grad_norm": 0.0913119688630104, "global_step": 23876, "epoch": 663, "lr": 0.001} +{"train_loss": 0.031244367361068726, "train_loss_bc": 0.031244367361068726, "train_loss_llm": 0.0, "grad_norm": 0.03513260930776596, "global_step": 23877, "epoch": 663, "lr": 0.001} +{"train_loss": 0.02660486102104187, "train_loss_bc": 0.02660486102104187, "train_loss_llm": 0.0, "grad_norm": 0.042426399886608124, "global_step": 23878, "epoch": 663, "lr": 0.001} +{"train_loss": 0.028482820838689804, "train_loss_bc": 0.028482820838689804, "train_loss_llm": 0.0, "grad_norm": 0.05633305385708809, "global_step": 23879, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03190205246210098, "train_loss_bc": 0.03190205246210098, "train_loss_llm": 0.0, "grad_norm": 0.06484387814998627, "global_step": 23880, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03476297855377197, "train_loss_bc": 0.03476297855377197, "train_loss_llm": 0.0, "grad_norm": 0.036788854748010635, "global_step": 23881, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03678331524133682, "train_loss_bc": 0.03678331524133682, "train_loss_llm": 0.0, "grad_norm": 0.04286833480000496, "global_step": 23882, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03179822489619255, "train_loss_bc": 0.03179822489619255, "train_loss_llm": 0.0, "grad_norm": 0.05674542486667633, "global_step": 23883, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03276662155985832, "train_loss_bc": 0.03276662155985832, "train_loss_llm": 0.0, "grad_norm": 0.07274561375379562, "global_step": 23884, "epoch": 663, "lr": 0.001} +{"train_loss": 0.028394892811775208, "train_loss_bc": 0.028394892811775208, "train_loss_llm": 0.0, "grad_norm": 0.034538205713033676, "global_step": 23885, "epoch": 663, "lr": 0.001} +{"train_loss": 0.028402598574757576, "train_loss_bc": 0.02840234898030758, "train_loss_llm": 2.4944543838500977e-05, "grad_norm": 0.04167386144399643, "global_step": 23886, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03916521370410919, "train_loss_bc": 0.03916521370410919, "train_loss_llm": 0.0, "grad_norm": 0.07498732954263687, "global_step": 23887, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03400391712784767, "train_loss_bc": 0.03399671986699104, "train_loss_llm": 0.000719904899597168, "grad_norm": 0.09026465564966202, "global_step": 23888, "epoch": 663, "lr": 0.001} +{"train_loss": 0.027088342234492302, "train_loss_bc": 0.027088342234492302, "train_loss_llm": 0.0, "grad_norm": 0.028665969148278236, "global_step": 23889, "epoch": 663, "lr": 0.001} +{"train_loss": 0.0331365242600441, "train_loss_bc": 0.033134691417217255, "train_loss_llm": 0.0001831650733947754, "grad_norm": 0.050660815089941025, "global_step": 23890, "epoch": 663, "lr": 0.001} +{"train_loss": 0.02854071743786335, "train_loss_bc": 0.02852976880967617, "train_loss_llm": 0.0010949373245239258, "grad_norm": 0.056086134165525436, "global_step": 23891, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03433934599161148, "train_loss_bc": 0.03433404862880707, "train_loss_llm": 0.0005295872688293457, "grad_norm": 0.0709761306643486, "global_step": 23892, "epoch": 663, "lr": 0.001} +{"train_loss": 0.029381180182099342, "train_loss_bc": 0.029381180182099342, "train_loss_llm": 0.0, "grad_norm": 0.03035258874297142, "global_step": 23893, "epoch": 663, "lr": 0.001} +{"train_loss": 0.0277722030878067, "train_loss_bc": 0.027772145345807076, "train_loss_llm": 5.7220458984375e-06, "grad_norm": 0.044602494686841965, "global_step": 23894, "epoch": 663, "lr": 0.001} +{"train_loss": 0.029877156019210815, "train_loss_bc": 0.029877156019210815, "train_loss_llm": 0.0, "grad_norm": 0.053772445768117905, "global_step": 23895, "epoch": 663, "lr": 0.001} +{"train_loss": 0.034462492913007736, "train_loss_bc": 0.034462492913007736, "train_loss_llm": 0.0, "grad_norm": 0.07070287317037582, "global_step": 23896, "epoch": 663, "lr": 0.001} +{"train_loss": 0.02646271325647831, "train_loss_bc": 0.02646271325647831, "train_loss_llm": 0.0, "grad_norm": 0.027778929099440575, "global_step": 23897, "epoch": 663, "lr": 0.001} +{"train_loss": 0.030564039945602417, "train_loss_bc": 0.03055723011493683, "train_loss_llm": 0.0006809532642364502, "grad_norm": 0.04630419611930847, "global_step": 23898, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03071039356291294, "train_loss_bc": 0.030700812116265297, "train_loss_llm": 0.000958174467086792, "grad_norm": 0.055504102259874344, "global_step": 23899, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03284763544797897, "train_loss_bc": 0.03284763544797897, "train_loss_llm": 0.0, "grad_norm": 0.0788908377289772, "global_step": 23900, "epoch": 663, "lr": 0.001} +{"train_loss": 0.025479603558778763, "train_loss_bc": 0.025479603558778763, "train_loss_llm": 0.0, "grad_norm": 0.0234474316239357, "global_step": 23901, "epoch": 663, "lr": 0.001} +{"train_loss": 0.03437545895576477, "train_loss_bc": 0.034337129443883896, "train_loss_llm": 0.0038330256938934326, "grad_norm": 0.045808788388967514, "global_step": 23902, "epoch": 663, "lr": 0.001} +{"train_loss": 0.031684619271092944, "train_loss_bc": 0.031042786315083504, "train_loss_llm": 0.0, "grad_norm": 0.05673719570040703, "global_step": 23903, "epoch": 663, "lr": 0.001, "val_loss": 0.08767230063676834} +{"train_loss": 0.024012725800275803, "train_loss_bc": 0.024007417261600494, "train_loss_llm": 0.0005308091640472412, "grad_norm": 0.0675802081823349, "global_step": 23904, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03081805258989334, "train_loss_bc": 0.03081805258989334, "train_loss_llm": 0.0, "grad_norm": 0.03578665107488632, "global_step": 23905, "epoch": 664, "lr": 0.001} +{"train_loss": 0.033936355262994766, "train_loss_bc": 0.03391578420996666, "train_loss_llm": 0.0020572245121002197, "grad_norm": 0.05986678600311279, "global_step": 23906, "epoch": 664, "lr": 0.001} +{"train_loss": 0.030376356095075607, "train_loss_bc": 0.030376356095075607, "train_loss_llm": 0.0, "grad_norm": 0.07088080793619156, "global_step": 23907, "epoch": 664, "lr": 0.001} +{"train_loss": 0.036041244864463806, "train_loss_bc": 0.036041244864463806, "train_loss_llm": 0.0, "grad_norm": 0.07934185862541199, "global_step": 23908, "epoch": 664, "lr": 0.001} +{"train_loss": 0.021318664774298668, "train_loss_bc": 0.021313147619366646, "train_loss_llm": 0.0005516409873962402, "grad_norm": 0.03632176294922829, "global_step": 23909, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03160218149423599, "train_loss_bc": 0.03160218149423599, "train_loss_llm": 0.0, "grad_norm": 0.05864609032869339, "global_step": 23910, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03676604852080345, "train_loss_bc": 0.036751288920640945, "train_loss_llm": 0.0014758110046386719, "grad_norm": 0.07469306141138077, "global_step": 23911, "epoch": 664, "lr": 0.001} +{"train_loss": 0.039090126752853394, "train_loss_bc": 0.039090126752853394, "train_loss_llm": 0.0, "grad_norm": 0.08734913170337677, "global_step": 23912, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03061429038643837, "train_loss_bc": 0.03061429038643837, "train_loss_llm": 0.0, "grad_norm": 0.029486259445548058, "global_step": 23913, "epoch": 664, "lr": 0.001} +{"train_loss": 0.031186722218990326, "train_loss_bc": 0.031186722218990326, "train_loss_llm": 0.0, "grad_norm": 0.05445915833115578, "global_step": 23914, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03053121455013752, "train_loss_bc": 0.030526626855134964, "train_loss_llm": 0.00045883655548095703, "grad_norm": 0.07346052676439285, "global_step": 23915, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03290070220828056, "train_loss_bc": 0.03289796784520149, "train_loss_llm": 0.000273287296295166, "grad_norm": 0.09013356268405914, "global_step": 23916, "epoch": 664, "lr": 0.001} +{"train_loss": 0.024539345875382423, "train_loss_bc": 0.024532871320843697, "train_loss_llm": 0.0006475448608398438, "grad_norm": 0.030473224818706512, "global_step": 23917, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03814388066530228, "train_loss_bc": 0.03814388066530228, "train_loss_llm": 0.0, "grad_norm": 0.04837091639637947, "global_step": 23918, "epoch": 664, "lr": 0.001} +{"train_loss": 0.027824468910694122, "train_loss_bc": 0.027824468910694122, "train_loss_llm": 0.0, "grad_norm": 0.05553021281957626, "global_step": 23919, "epoch": 664, "lr": 0.001} +{"train_loss": 0.029994402080774307, "train_loss_bc": 0.029994402080774307, "train_loss_llm": 0.0, "grad_norm": 0.06599260121583939, "global_step": 23920, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03687882795929909, "train_loss_bc": 0.03687882795929909, "train_loss_llm": 0.0, "grad_norm": 0.032500457018613815, "global_step": 23921, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03559489920735359, "train_loss_bc": 0.03559489920735359, "train_loss_llm": 0.0, "grad_norm": 0.046118758618831635, "global_step": 23922, "epoch": 664, "lr": 0.001} +{"train_loss": 0.030622392892837524, "train_loss_bc": 0.030622392892837524, "train_loss_llm": 0.0, "grad_norm": 0.06014981493353844, "global_step": 23923, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03435724973678589, "train_loss_bc": 0.03435724973678589, "train_loss_llm": 0.0, "grad_norm": 0.07357356697320938, "global_step": 23924, "epoch": 664, "lr": 0.001} +{"train_loss": 0.029828375205397606, "train_loss_bc": 0.029828375205397606, "train_loss_llm": 0.0, "grad_norm": 0.0390167310833931, "global_step": 23925, "epoch": 664, "lr": 0.001} +{"train_loss": 0.030749112367630005, "train_loss_bc": 0.030749112367630005, "train_loss_llm": 0.0, "grad_norm": 0.06286812573671341, "global_step": 23926, "epoch": 664, "lr": 0.001} +{"train_loss": 0.028805941343307495, "train_loss_bc": 0.028792500495910645, "train_loss_llm": 0.0013440251350402832, "grad_norm": 0.08839026093482971, "global_step": 23927, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03240881860256195, "train_loss_bc": 0.03240881860256195, "train_loss_llm": 0.0, "grad_norm": 0.10958459228277206, "global_step": 23928, "epoch": 664, "lr": 0.001} +{"train_loss": 0.034849248826503754, "train_loss_bc": 0.034849248826503754, "train_loss_llm": 0.0, "grad_norm": 0.03721730411052704, "global_step": 23929, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03785791993141174, "train_loss_bc": 0.03785791993141174, "train_loss_llm": 0.0, "grad_norm": 0.06324893981218338, "global_step": 23930, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03452068194746971, "train_loss_bc": 0.03452068194746971, "train_loss_llm": 0.0, "grad_norm": 0.08010585606098175, "global_step": 23931, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03416435420513153, "train_loss_bc": 0.03416435420513153, "train_loss_llm": 0.0, "grad_norm": 0.09360483288764954, "global_step": 23932, "epoch": 664, "lr": 0.001} +{"train_loss": 0.026373473927378654, "train_loss_bc": 0.026368599385023117, "train_loss_llm": 0.0004874765872955322, "grad_norm": 0.028587447479367256, "global_step": 23933, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03390875831246376, "train_loss_bc": 0.03390875831246376, "train_loss_llm": 0.0, "grad_norm": 0.04660748690366745, "global_step": 23934, "epoch": 664, "lr": 0.001} +{"train_loss": 0.029352674260735512, "train_loss_bc": 0.029352674260735512, "train_loss_llm": 0.0, "grad_norm": 0.06402062624692917, "global_step": 23935, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03194381669163704, "train_loss_bc": 0.03194381669163704, "train_loss_llm": 0.0, "grad_norm": 0.0822005420923233, "global_step": 23936, "epoch": 664, "lr": 0.001} +{"train_loss": 0.028874972835183144, "train_loss_bc": 0.028874972835183144, "train_loss_llm": 0.0, "grad_norm": 0.02690235897898674, "global_step": 23937, "epoch": 664, "lr": 0.001} +{"train_loss": 0.032204896211624146, "train_loss_bc": 0.03220449760556221, "train_loss_llm": 3.975629806518555e-05, "grad_norm": 0.04181944578886032, "global_step": 23938, "epoch": 664, "lr": 0.001} +{"train_loss": 0.03180039099728068, "train_loss_bc": 0.03181227669119835, "train_loss_llm": 0.0008601546287536621, "grad_norm": 0.06430528312921524, "global_step": 23939, "epoch": 664, "lr": 0.001, "val_loss": 0.08563326299190521} +{"train_loss": 0.03120185062289238, "train_loss_bc": 0.03120185062289238, "train_loss_llm": 0.0, "grad_norm": 0.07985717803239822, "global_step": 23940, "epoch": 665, "lr": 0.001} +{"train_loss": 0.027762332931160927, "train_loss_bc": 0.027762332931160927, "train_loss_llm": 0.0, "grad_norm": 0.0367480143904686, "global_step": 23941, "epoch": 665, "lr": 0.001} +{"train_loss": 0.031365904957056046, "train_loss_bc": 0.03136114776134491, "train_loss_llm": 0.0004755854606628418, "grad_norm": 0.05152905359864235, "global_step": 23942, "epoch": 665, "lr": 0.001} +{"train_loss": 0.040899552404880524, "train_loss_bc": 0.040888771414756775, "train_loss_llm": 0.0010781288146972656, "grad_norm": 0.07450516521930695, "global_step": 23943, "epoch": 665, "lr": 0.001} +{"train_loss": 0.02680697664618492, "train_loss_bc": 0.02680697664618492, "train_loss_llm": 0.0, "grad_norm": 0.08550559729337692, "global_step": 23944, "epoch": 665, "lr": 0.001} +{"train_loss": 0.02602703869342804, "train_loss_bc": 0.02602703869342804, "train_loss_llm": 0.0, "grad_norm": 0.03469287231564522, "global_step": 23945, "epoch": 665, "lr": 0.001} +{"train_loss": 0.030731866136193275, "train_loss_bc": 0.030731866136193275, "train_loss_llm": 0.0, "grad_norm": 0.04312688484787941, "global_step": 23946, "epoch": 665, "lr": 0.001} +{"train_loss": 0.032361045479774475, "train_loss_bc": 0.032361045479774475, "train_loss_llm": 0.0, "grad_norm": 0.05531235411763191, "global_step": 23947, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03633839637041092, "train_loss_bc": 0.03633839637041092, "train_loss_llm": 0.0, "grad_norm": 0.07061300426721573, "global_step": 23948, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03262835368514061, "train_loss_bc": 0.03262048587203026, "train_loss_llm": 0.0007866024971008301, "grad_norm": 0.036389537155628204, "global_step": 23949, "epoch": 665, "lr": 0.001} +{"train_loss": 0.028957653790712357, "train_loss_bc": 0.028957653790712357, "train_loss_llm": 0.0, "grad_norm": 0.06044219434261322, "global_step": 23950, "epoch": 665, "lr": 0.001} +{"train_loss": 0.032394956797361374, "train_loss_bc": 0.03238561004400253, "train_loss_llm": 0.0009348392486572266, "grad_norm": 0.08107947558164597, "global_step": 23951, "epoch": 665, "lr": 0.001} +{"train_loss": 0.030546536669135094, "train_loss_bc": 0.030541926622390747, "train_loss_llm": 0.0004610717296600342, "grad_norm": 0.10162800550460815, "global_step": 23952, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03803713619709015, "train_loss_bc": 0.03803713619709015, "train_loss_llm": 0.0, "grad_norm": 0.034585412591695786, "global_step": 23953, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03083146922290325, "train_loss_bc": 0.03083146922290325, "train_loss_llm": 0.0, "grad_norm": 0.04398535192012787, "global_step": 23954, "epoch": 665, "lr": 0.001} +{"train_loss": 0.02632443979382515, "train_loss_bc": 0.02630911022424698, "train_loss_llm": 0.0015329420566558838, "grad_norm": 0.05813616141676903, "global_step": 23955, "epoch": 665, "lr": 0.001} +{"train_loss": 0.032095879316329956, "train_loss_bc": 0.032095879316329956, "train_loss_llm": 0.0, "grad_norm": 0.07748395949602127, "global_step": 23956, "epoch": 665, "lr": 0.001} +{"train_loss": 0.034624695777893066, "train_loss_bc": 0.034624695777893066, "train_loss_llm": 0.0, "grad_norm": 0.03604105859994888, "global_step": 23957, "epoch": 665, "lr": 0.001} +{"train_loss": 0.02560557797551155, "train_loss_bc": 0.02560557797551155, "train_loss_llm": 0.0, "grad_norm": 0.05673140287399292, "global_step": 23958, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03476221114397049, "train_loss_bc": 0.03476221114397049, "train_loss_llm": 0.0, "grad_norm": 0.07754403352737427, "global_step": 23959, "epoch": 665, "lr": 0.001} +{"train_loss": 0.02761039510369301, "train_loss_bc": 0.0276088397949934, "train_loss_llm": 0.00015550851821899414, "grad_norm": 0.0895562618970871, "global_step": 23960, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03278859332203865, "train_loss_bc": 0.03278859332203865, "train_loss_llm": 0.0, "grad_norm": 0.030562903732061386, "global_step": 23961, "epoch": 665, "lr": 0.001} +{"train_loss": 0.0367867536842823, "train_loss_bc": 0.0367867536842823, "train_loss_llm": 0.0, "grad_norm": 0.042371347546577454, "global_step": 23962, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03717043995857239, "train_loss_bc": 0.03717043995857239, "train_loss_llm": 0.0, "grad_norm": 0.06560266762971878, "global_step": 23963, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03576088324189186, "train_loss_bc": 0.03576088324189186, "train_loss_llm": 0.0, "grad_norm": 0.08668769150972366, "global_step": 23964, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03180838003754616, "train_loss_bc": 0.03180838003754616, "train_loss_llm": 0.0, "grad_norm": 0.02863306738436222, "global_step": 23965, "epoch": 665, "lr": 0.001} +{"train_loss": 0.026198003441095352, "train_loss_bc": 0.026198003441095352, "train_loss_llm": 0.0, "grad_norm": 0.04062474146485329, "global_step": 23966, "epoch": 665, "lr": 0.001} +{"train_loss": 0.026547618210315704, "train_loss_bc": 0.026537559926509857, "train_loss_llm": 0.0010057389736175537, "grad_norm": 0.05362094193696976, "global_step": 23967, "epoch": 665, "lr": 0.001} +{"train_loss": 0.030762001872062683, "train_loss_bc": 0.030762001872062683, "train_loss_llm": 0.0, "grad_norm": 0.07271384447813034, "global_step": 23968, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03914869576692581, "train_loss_bc": 0.03914869576692581, "train_loss_llm": 0.0, "grad_norm": 0.046083901077508926, "global_step": 23969, "epoch": 665, "lr": 0.001} +{"train_loss": 0.037707116454839706, "train_loss_bc": 0.037707116454839706, "train_loss_llm": 0.0, "grad_norm": 0.06771456450223923, "global_step": 23970, "epoch": 665, "lr": 0.001} +{"train_loss": 0.026907041668891907, "train_loss_bc": 0.026906240731477737, "train_loss_llm": 8.01384449005127e-05, "grad_norm": 0.08315823972225189, "global_step": 23971, "epoch": 665, "lr": 0.001} +{"train_loss": 0.032495755702257156, "train_loss_bc": 0.032495755702257156, "train_loss_llm": 0.0, "grad_norm": 0.09763083606958389, "global_step": 23972, "epoch": 665, "lr": 0.001} +{"train_loss": 0.033745840191841125, "train_loss_bc": 0.033745840191841125, "train_loss_llm": 0.0, "grad_norm": 0.0378330834209919, "global_step": 23973, "epoch": 665, "lr": 0.001} +{"train_loss": 0.02851617895066738, "train_loss_bc": 0.02851617895066738, "train_loss_llm": 0.0, "grad_norm": 0.051634207367897034, "global_step": 23974, "epoch": 665, "lr": 0.001} +{"train_loss": 0.03193123095358411, "train_loss_bc": 0.03526674211025238, "train_loss_llm": 0.0, "grad_norm": 0.07656458020210266, "global_step": 23975, "epoch": 665, "lr": 0.001, "train/cumulative_reward": 67.97433315401626, "train/mean_score": 22.840117736701508, "train/success_rate": 0.7, "test/cumulative_reward": 30.565752352890865, "test/mean_score": 8.605404806860376, "test/success_rate": 0.24, "val_loss": 0.087407186627388, "train_action_mse_error": 0.03523017093539238} +{"train_loss": 0.0307538490742445, "train_loss_bc": 0.0307538490742445, "train_loss_llm": 0.0, "grad_norm": 0.09299336373806, "global_step": 23976, "epoch": 666, "lr": 0.001} +{"train_loss": 0.029146907851099968, "train_loss_bc": 0.029146907851099968, "train_loss_llm": 0.0, "grad_norm": 0.03291869908571243, "global_step": 23977, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03581193462014198, "train_loss_bc": 0.03581193462014198, "train_loss_llm": 0.0, "grad_norm": 0.05115688219666481, "global_step": 23978, "epoch": 666, "lr": 0.001} +{"train_loss": 0.026766203343868256, "train_loss_bc": 0.026765048503875732, "train_loss_llm": 0.00011551380157470703, "grad_norm": 0.06424718350172043, "global_step": 23979, "epoch": 666, "lr": 0.001} +{"train_loss": 0.030098671093583107, "train_loss_bc": 0.030098671093583107, "train_loss_llm": 0.0, "grad_norm": 0.07961125671863556, "global_step": 23980, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03781159967184067, "train_loss_bc": 0.03781159967184067, "train_loss_llm": 0.0, "grad_norm": 0.035528723150491714, "global_step": 23981, "epoch": 666, "lr": 0.001} +{"train_loss": 0.032841987907886505, "train_loss_bc": 0.032832954078912735, "train_loss_llm": 0.0009034872055053711, "grad_norm": 0.053996771574020386, "global_step": 23982, "epoch": 666, "lr": 0.001} +{"train_loss": 0.02850436046719551, "train_loss_bc": 0.02850436046719551, "train_loss_llm": 0.0, "grad_norm": 0.06841632723808289, "global_step": 23983, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03094593808054924, "train_loss_bc": 0.030924925580620766, "train_loss_llm": 0.0021012723445892334, "grad_norm": 0.07706696540117264, "global_step": 23984, "epoch": 666, "lr": 0.001} +{"train_loss": 0.033363305032253265, "train_loss_bc": 0.03336232528090477, "train_loss_llm": 9.813904762268066e-05, "grad_norm": 0.040180452167987823, "global_step": 23985, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03338806703686714, "train_loss_bc": 0.03338806703686714, "train_loss_llm": 0.0, "grad_norm": 0.06051640585064888, "global_step": 23986, "epoch": 666, "lr": 0.001} +{"train_loss": 0.028406400233507156, "train_loss_bc": 0.02839541621506214, "train_loss_llm": 0.0010983645915985107, "grad_norm": 0.07617005705833435, "global_step": 23987, "epoch": 666, "lr": 0.001} +{"train_loss": 0.030183900147676468, "train_loss_bc": 0.030183900147676468, "train_loss_llm": 0.0, "grad_norm": 0.09379708766937256, "global_step": 23988, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03254067152738571, "train_loss_bc": 0.03254067152738571, "train_loss_llm": 0.0, "grad_norm": 0.033642612397670746, "global_step": 23989, "epoch": 666, "lr": 0.001} +{"train_loss": 0.0359836146235466, "train_loss_bc": 0.035980887711048126, "train_loss_llm": 0.0002726912498474121, "grad_norm": 0.06854531168937683, "global_step": 23990, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03719218447804451, "train_loss_bc": 0.03718840330839157, "train_loss_llm": 0.0003782510757446289, "grad_norm": 0.08236218243837357, "global_step": 23991, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03106873854994774, "train_loss_bc": 0.03106873854994774, "train_loss_llm": 0.0, "grad_norm": 0.09930254518985748, "global_step": 23992, "epoch": 666, "lr": 0.001} +{"train_loss": 0.032288916409015656, "train_loss_bc": 0.032284606248140335, "train_loss_llm": 0.0004310309886932373, "grad_norm": 0.024540789425373077, "global_step": 23993, "epoch": 666, "lr": 0.001} +{"train_loss": 0.031176690012216568, "train_loss_bc": 0.03115626610815525, "train_loss_llm": 0.0020423829555511475, "grad_norm": 0.04743422940373421, "global_step": 23994, "epoch": 666, "lr": 0.001} +{"train_loss": 0.022298261523246765, "train_loss_bc": 0.022292448207736015, "train_loss_llm": 0.0005813837051391602, "grad_norm": 0.056858986616134644, "global_step": 23995, "epoch": 666, "lr": 0.001} +{"train_loss": 0.027847226709127426, "train_loss_bc": 0.027847226709127426, "train_loss_llm": 0.0, "grad_norm": 0.07599256932735443, "global_step": 23996, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03378220275044441, "train_loss_bc": 0.03376767784357071, "train_loss_llm": 0.0014525651931762695, "grad_norm": 0.03184399753808975, "global_step": 23997, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03812004625797272, "train_loss_bc": 0.038116101175546646, "train_loss_llm": 0.00039464235305786133, "grad_norm": 0.047418855130672455, "global_step": 23998, "epoch": 666, "lr": 0.001} +{"train_loss": 0.033373646438121796, "train_loss_bc": 0.03334083780646324, "train_loss_llm": 0.0032809972763061523, "grad_norm": 0.06462498754262924, "global_step": 23999, "epoch": 666, "lr": 0.001} +{"train_loss": 0.028919043019413948, "train_loss_bc": 0.028919043019413948, "train_loss_llm": 0.0, "grad_norm": 0.08677414804697037, "global_step": 24000, "epoch": 666, "lr": 0.001} +{"train_loss": 0.028337446972727776, "train_loss_bc": 0.028337446972727776, "train_loss_llm": 0.0, "grad_norm": 0.029739802703261375, "global_step": 24001, "epoch": 666, "lr": 0.001} +{"train_loss": 0.032419413328170776, "train_loss_bc": 0.032419413328170776, "train_loss_llm": 0.0, "grad_norm": 0.06130300834774971, "global_step": 24002, "epoch": 666, "lr": 0.001} +{"train_loss": 0.0349886529147625, "train_loss_bc": 0.0349886529147625, "train_loss_llm": 0.0, "grad_norm": 0.08897752314805984, "global_step": 24003, "epoch": 666, "lr": 0.001} +{"train_loss": 0.033271100372076035, "train_loss_bc": 0.033271100372076035, "train_loss_llm": 0.0, "grad_norm": 0.09225235879421234, "global_step": 24004, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03659917786717415, "train_loss_bc": 0.03659917786717415, "train_loss_llm": 0.0, "grad_norm": 0.03492216020822525, "global_step": 24005, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03405454382300377, "train_loss_bc": 0.03405454382300377, "train_loss_llm": 0.0, "grad_norm": 0.0622926726937294, "global_step": 24006, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03234162554144859, "train_loss_bc": 0.032331835478544235, "train_loss_llm": 0.0009791254997253418, "grad_norm": 0.08294755220413208, "global_step": 24007, "epoch": 666, "lr": 0.001} +{"train_loss": 0.032077938318252563, "train_loss_bc": 0.032077938318252563, "train_loss_llm": 0.0, "grad_norm": 0.09462809562683105, "global_step": 24008, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03324305638670921, "train_loss_bc": 0.03324305638670921, "train_loss_llm": 0.0, "grad_norm": 0.030013008043169975, "global_step": 24009, "epoch": 666, "lr": 0.001} +{"train_loss": 0.02927437610924244, "train_loss_bc": 0.02927437610924244, "train_loss_llm": 0.0, "grad_norm": 0.04955719783902168, "global_step": 24010, "epoch": 666, "lr": 0.001} +{"train_loss": 0.03190304778723253, "train_loss_bc": 0.029288021847605705, "train_loss_llm": 0.0, "grad_norm": 0.07585636526346207, "global_step": 24011, "epoch": 666, "lr": 0.001, "val_loss": 0.08837579190731049} +{"train_loss": 0.03188314288854599, "train_loss_bc": 0.03188314288854599, "train_loss_llm": 0.0, "grad_norm": 0.09253142774105072, "global_step": 24012, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03417224436998367, "train_loss_bc": 0.03417224436998367, "train_loss_llm": 0.0, "grad_norm": 0.03536175563931465, "global_step": 24013, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03256090357899666, "train_loss_bc": 0.03256090357899666, "train_loss_llm": 0.0, "grad_norm": 0.056634534150362015, "global_step": 24014, "epoch": 667, "lr": 0.001} +{"train_loss": 0.027387237176299095, "train_loss_bc": 0.027370763942599297, "train_loss_llm": 0.0016473233699798584, "grad_norm": 0.07543299347162247, "global_step": 24015, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03435579314827919, "train_loss_bc": 0.03435579314827919, "train_loss_llm": 0.0, "grad_norm": 0.09904024004936218, "global_step": 24016, "epoch": 667, "lr": 0.001} +{"train_loss": 0.027873296290636063, "train_loss_bc": 0.027873296290636063, "train_loss_llm": 0.0, "grad_norm": 0.027498818933963776, "global_step": 24017, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03025701269507408, "train_loss_bc": 0.03025701269507408, "train_loss_llm": 0.0, "grad_norm": 0.05427826941013336, "global_step": 24018, "epoch": 667, "lr": 0.001} +{"train_loss": 0.030761050060391426, "train_loss_bc": 0.030761050060391426, "train_loss_llm": 0.0, "grad_norm": 0.07100501656532288, "global_step": 24019, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03273289278149605, "train_loss_bc": 0.03273289278149605, "train_loss_llm": 0.0, "grad_norm": 0.08873704075813293, "global_step": 24020, "epoch": 667, "lr": 0.001} +{"train_loss": 0.025622492656111717, "train_loss_bc": 0.025622492656111717, "train_loss_llm": 0.0, "grad_norm": 0.025103261694312096, "global_step": 24021, "epoch": 667, "lr": 0.001} +{"train_loss": 0.02889859490096569, "train_loss_bc": 0.02889859490096569, "train_loss_llm": 0.0, "grad_norm": 0.043036505579948425, "global_step": 24022, "epoch": 667, "lr": 0.001} +{"train_loss": 0.028851192444562912, "train_loss_bc": 0.028851192444562912, "train_loss_llm": 0.0, "grad_norm": 0.062394704669713974, "global_step": 24023, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03171199560165405, "train_loss_bc": 0.03171199560165405, "train_loss_llm": 0.0, "grad_norm": 0.07772228121757507, "global_step": 24024, "epoch": 667, "lr": 0.001} +{"train_loss": 0.0336158461868763, "train_loss_bc": 0.033599209040403366, "train_loss_llm": 0.001663893461227417, "grad_norm": 0.030345303937792778, "global_step": 24025, "epoch": 667, "lr": 0.001} +{"train_loss": 0.030833270400762558, "train_loss_bc": 0.030833270400762558, "train_loss_llm": 0.0, "grad_norm": 0.05336307734251022, "global_step": 24026, "epoch": 667, "lr": 0.001} +{"train_loss": 0.028359772637486458, "train_loss_bc": 0.028359772637486458, "train_loss_llm": 0.0, "grad_norm": 0.0773693397641182, "global_step": 24027, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03466086462140083, "train_loss_bc": 0.03466086462140083, "train_loss_llm": 0.0, "grad_norm": 0.100675068795681, "global_step": 24028, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03318020701408386, "train_loss_bc": 0.03318020701408386, "train_loss_llm": 0.0, "grad_norm": 0.03457937017083168, "global_step": 24029, "epoch": 667, "lr": 0.001} +{"train_loss": 0.031104907393455505, "train_loss_bc": 0.031104907393455505, "train_loss_llm": 0.0, "grad_norm": 0.04831352084875107, "global_step": 24030, "epoch": 667, "lr": 0.001} +{"train_loss": 0.028752829879522324, "train_loss_bc": 0.028752829879522324, "train_loss_llm": 0.0, "grad_norm": 0.06260643154382706, "global_step": 24031, "epoch": 667, "lr": 0.001} +{"train_loss": 0.043444693088531494, "train_loss_bc": 0.043444693088531494, "train_loss_llm": 0.0, "grad_norm": 0.0742870345711708, "global_step": 24032, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03494807705283165, "train_loss_bc": 0.0349261537194252, "train_loss_llm": 0.002192169427871704, "grad_norm": 0.027906756848096848, "global_step": 24033, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03147226199507713, "train_loss_bc": 0.03147226199507713, "train_loss_llm": 0.0, "grad_norm": 0.04092065244913101, "global_step": 24034, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03326971083879471, "train_loss_bc": 0.03326971083879471, "train_loss_llm": 0.0, "grad_norm": 0.05979832261800766, "global_step": 24035, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03257100284099579, "train_loss_bc": 0.03257100284099579, "train_loss_llm": 0.0, "grad_norm": 0.07188515365123749, "global_step": 24036, "epoch": 667, "lr": 0.001} +{"train_loss": 0.0310289915651083, "train_loss_bc": 0.0310289915651083, "train_loss_llm": 0.0, "grad_norm": 0.03155801445245743, "global_step": 24037, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03287778049707413, "train_loss_bc": 0.03287778049707413, "train_loss_llm": 0.0, "grad_norm": 0.05674351379275322, "global_step": 24038, "epoch": 667, "lr": 0.001} +{"train_loss": 0.030178755521774292, "train_loss_bc": 0.030150752514600754, "train_loss_llm": 0.0028002560138702393, "grad_norm": 0.07914476841688156, "global_step": 24039, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03362297639250755, "train_loss_bc": 0.03362297639250755, "train_loss_llm": 0.0, "grad_norm": 0.09605327248573303, "global_step": 24040, "epoch": 667, "lr": 0.001} +{"train_loss": 0.02900458127260208, "train_loss_bc": 0.02900458127260208, "train_loss_llm": 0.0, "grad_norm": 0.02850024588406086, "global_step": 24041, "epoch": 667, "lr": 0.001} +{"train_loss": 0.030044225975871086, "train_loss_bc": 0.030044225975871086, "train_loss_llm": 0.0, "grad_norm": 0.03669777885079384, "global_step": 24042, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03534955531358719, "train_loss_bc": 0.03534955531358719, "train_loss_llm": 0.0, "grad_norm": 0.05475911498069763, "global_step": 24043, "epoch": 667, "lr": 0.001} +{"train_loss": 0.03462052345275879, "train_loss_bc": 0.03462052345275879, "train_loss_llm": 0.0, "grad_norm": 0.06529282033443451, "global_step": 24044, "epoch": 667, "lr": 0.001} +{"train_loss": 0.029047364369034767, "train_loss_bc": 0.029043033719062805, "train_loss_llm": 0.0004330277442932129, "grad_norm": 0.031189249828457832, "global_step": 24045, "epoch": 667, "lr": 0.001} +{"train_loss": 0.038416482508182526, "train_loss_bc": 0.03839895501732826, "train_loss_llm": 0.0017528235912322998, "grad_norm": 0.05207331106066704, "global_step": 24046, "epoch": 667, "lr": 0.001} +{"train_loss": 0.031835687864157886, "train_loss_bc": 0.02859020046889782, "train_loss_llm": 0.00220334529876709, "grad_norm": 0.06786786019802094, "global_step": 24047, "epoch": 667, "lr": 0.001, "val_loss": 0.0867237076163292} +{"train_loss": 0.038318730890750885, "train_loss_bc": 0.038318730890750885, "train_loss_llm": 0.0, "grad_norm": 0.10036160796880722, "global_step": 24048, "epoch": 668, "lr": 0.001} +{"train_loss": 0.02962150052189827, "train_loss_bc": 0.02962150052189827, "train_loss_llm": 0.0, "grad_norm": 0.03793518245220184, "global_step": 24049, "epoch": 668, "lr": 0.001} +{"train_loss": 0.030910421162843704, "train_loss_bc": 0.030910421162843704, "train_loss_llm": 0.0, "grad_norm": 0.05839625000953674, "global_step": 24050, "epoch": 668, "lr": 0.001} +{"train_loss": 0.025682669132947922, "train_loss_bc": 0.025682669132947922, "train_loss_llm": 0.0, "grad_norm": 0.06271247565746307, "global_step": 24051, "epoch": 668, "lr": 0.001} +{"train_loss": 0.030153900384902954, "train_loss_bc": 0.030153611674904823, "train_loss_llm": 2.8789043426513672e-05, "grad_norm": 0.08450180292129517, "global_step": 24052, "epoch": 668, "lr": 0.001} +{"train_loss": 0.036344822496175766, "train_loss_bc": 0.03634383901953697, "train_loss_llm": 9.825825691223145e-05, "grad_norm": 0.026233136653900146, "global_step": 24053, "epoch": 668, "lr": 0.001} +{"train_loss": 0.02479444444179535, "train_loss_bc": 0.02479444444179535, "train_loss_llm": 0.0, "grad_norm": 0.041491296142339706, "global_step": 24054, "epoch": 668, "lr": 0.001} +{"train_loss": 0.02472950704395771, "train_loss_bc": 0.02472844161093235, "train_loss_llm": 0.00010663270950317383, "grad_norm": 0.06182121858000755, "global_step": 24055, "epoch": 668, "lr": 0.001} +{"train_loss": 0.028191085904836655, "train_loss_bc": 0.028187626972794533, "train_loss_llm": 0.0003459751605987549, "grad_norm": 0.07174918800592422, "global_step": 24056, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03464461490511894, "train_loss_bc": 0.03464461490511894, "train_loss_llm": 0.0, "grad_norm": 0.028900589793920517, "global_step": 24057, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03171268478035927, "train_loss_bc": 0.031701743602752686, "train_loss_llm": 0.0010940134525299072, "grad_norm": 0.05066462606191635, "global_step": 24058, "epoch": 668, "lr": 0.001} +{"train_loss": 0.028288282454013824, "train_loss_bc": 0.028288282454013824, "train_loss_llm": 0.0, "grad_norm": 0.06574644148349762, "global_step": 24059, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03406330570578575, "train_loss_bc": 0.034035760909318924, "train_loss_llm": 0.0027544498443603516, "grad_norm": 0.07917206734418869, "global_step": 24060, "epoch": 668, "lr": 0.001} +{"train_loss": 0.026355169713497162, "train_loss_bc": 0.026350652799010277, "train_loss_llm": 0.00045174360275268555, "grad_norm": 0.034539226442575455, "global_step": 24061, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03987501189112663, "train_loss_bc": 0.03987501189112663, "train_loss_llm": 0.0, "grad_norm": 0.060379065573215485, "global_step": 24062, "epoch": 668, "lr": 0.001} +{"train_loss": 0.032378651201725006, "train_loss_bc": 0.032378651201725006, "train_loss_llm": 0.0, "grad_norm": 0.08377454429864883, "global_step": 24063, "epoch": 668, "lr": 0.001} +{"train_loss": 0.035390421748161316, "train_loss_bc": 0.03538975864648819, "train_loss_llm": 6.642937660217285e-05, "grad_norm": 0.09727176278829575, "global_step": 24064, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03249456733465195, "train_loss_bc": 0.03249456733465195, "train_loss_llm": 0.0, "grad_norm": 0.03156178444623947, "global_step": 24065, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03739536926150322, "train_loss_bc": 0.03739536926150322, "train_loss_llm": 0.0, "grad_norm": 0.04763608053326607, "global_step": 24066, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03283924609422684, "train_loss_bc": 0.03283924609422684, "train_loss_llm": 0.0, "grad_norm": 0.06230941042304039, "global_step": 24067, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03384044021368027, "train_loss_bc": 0.03384044021368027, "train_loss_llm": 0.0, "grad_norm": 0.07113790512084961, "global_step": 24068, "epoch": 668, "lr": 0.001} +{"train_loss": 0.037278641015291214, "train_loss_bc": 0.037278641015291214, "train_loss_llm": 0.0, "grad_norm": 0.030710389837622643, "global_step": 24069, "epoch": 668, "lr": 0.001} +{"train_loss": 0.02420402318239212, "train_loss_bc": 0.024191539734601974, "train_loss_llm": 0.0012483298778533936, "grad_norm": 0.04052707925438881, "global_step": 24070, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03258673846721649, "train_loss_bc": 0.03258673846721649, "train_loss_llm": 0.0, "grad_norm": 0.05143474414944649, "global_step": 24071, "epoch": 668, "lr": 0.001} +{"train_loss": 0.034168682992458344, "train_loss_bc": 0.034168682992458344, "train_loss_llm": 0.0, "grad_norm": 0.0698365867137909, "global_step": 24072, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03314993530511856, "train_loss_bc": 0.03314993530511856, "train_loss_llm": 0.0, "grad_norm": 0.0338076576590538, "global_step": 24073, "epoch": 668, "lr": 0.001} +{"train_loss": 0.030830146744847298, "train_loss_bc": 0.030830146744847298, "train_loss_llm": 0.0, "grad_norm": 0.04612454026937485, "global_step": 24074, "epoch": 668, "lr": 0.001} +{"train_loss": 0.026407627388834953, "train_loss_bc": 0.026407627388834953, "train_loss_llm": 0.0, "grad_norm": 0.06152394413948059, "global_step": 24075, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03389476239681244, "train_loss_bc": 0.03389476239681244, "train_loss_llm": 0.0, "grad_norm": 0.07321526855230331, "global_step": 24076, "epoch": 668, "lr": 0.001} +{"train_loss": 0.026788590475916862, "train_loss_bc": 0.026788590475916862, "train_loss_llm": 0.0, "grad_norm": 0.024061055853962898, "global_step": 24077, "epoch": 668, "lr": 0.001} +{"train_loss": 0.032076481729745865, "train_loss_bc": 0.03205621987581253, "train_loss_llm": 0.0020261406898498535, "grad_norm": 0.038781292736530304, "global_step": 24078, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03588918596506119, "train_loss_bc": 0.03588918596506119, "train_loss_llm": 0.0, "grad_norm": 0.056497588753700256, "global_step": 24079, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03860849887132645, "train_loss_bc": 0.03860849887132645, "train_loss_llm": 0.0, "grad_norm": 0.07658378779888153, "global_step": 24080, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03159327432513237, "train_loss_bc": 0.03159327432513237, "train_loss_llm": 0.0, "grad_norm": 0.02670336328446865, "global_step": 24081, "epoch": 668, "lr": 0.001} +{"train_loss": 0.02715732343494892, "train_loss_bc": 0.02714492939412594, "train_loss_llm": 0.0012393295764923096, "grad_norm": 0.04246722534298897, "global_step": 24082, "epoch": 668, "lr": 0.001} +{"train_loss": 0.03176816568399469, "train_loss_bc": 0.0309952050447464, "train_loss_llm": 0.0, "grad_norm": 0.05307459458708763, "global_step": 24083, "epoch": 668, "lr": 0.001, "val_loss": 0.08770252019166946} +{"train_loss": 0.034950919449329376, "train_loss_bc": 0.034950919449329376, "train_loss_llm": 0.0, "grad_norm": 0.06732524931430817, "global_step": 24084, "epoch": 669, "lr": 0.001} +{"train_loss": 0.031735945492982864, "train_loss_bc": 0.031735051423311234, "train_loss_llm": 8.943676948547363e-05, "grad_norm": 0.032708168029785156, "global_step": 24085, "epoch": 669, "lr": 0.001} +{"train_loss": 0.031869035214185715, "train_loss_bc": 0.031869035214185715, "train_loss_llm": 0.0, "grad_norm": 0.05062641203403473, "global_step": 24086, "epoch": 669, "lr": 0.001} +{"train_loss": 0.024032188579440117, "train_loss_bc": 0.024032188579440117, "train_loss_llm": 0.0, "grad_norm": 0.06842774897813797, "global_step": 24087, "epoch": 669, "lr": 0.001} +{"train_loss": 0.029294664040207863, "train_loss_bc": 0.029294664040207863, "train_loss_llm": 0.0, "grad_norm": 0.08274510502815247, "global_step": 24088, "epoch": 669, "lr": 0.001} +{"train_loss": 0.029730556532740593, "train_loss_bc": 0.029730556532740593, "train_loss_llm": 0.0, "grad_norm": 0.026856284588575363, "global_step": 24089, "epoch": 669, "lr": 0.001} +{"train_loss": 0.032074831426143646, "train_loss_bc": 0.03206557780504227, "train_loss_llm": 0.000925213098526001, "grad_norm": 0.04615425318479538, "global_step": 24090, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03372030705213547, "train_loss_bc": 0.03372030705213547, "train_loss_llm": 0.0, "grad_norm": 0.07748939841985703, "global_step": 24091, "epoch": 669, "lr": 0.001} +{"train_loss": 0.027249297127127647, "train_loss_bc": 0.027249297127127647, "train_loss_llm": 0.0, "grad_norm": 0.09068966656923294, "global_step": 24092, "epoch": 669, "lr": 0.001} +{"train_loss": 0.0340772420167923, "train_loss_bc": 0.0340772420167923, "train_loss_llm": 0.0, "grad_norm": 0.041024092584848404, "global_step": 24093, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03849022462964058, "train_loss_bc": 0.03849022462964058, "train_loss_llm": 0.0, "grad_norm": 0.05140670761466026, "global_step": 24094, "epoch": 669, "lr": 0.001} +{"train_loss": 0.028869753703475, "train_loss_bc": 0.028869753703475, "train_loss_llm": 0.0, "grad_norm": 0.0537884421646595, "global_step": 24095, "epoch": 669, "lr": 0.001} +{"train_loss": 0.030990956351161003, "train_loss_bc": 0.03098813071846962, "train_loss_llm": 0.00028258562088012695, "grad_norm": 0.06491908431053162, "global_step": 24096, "epoch": 669, "lr": 0.001} +{"train_loss": 0.030932359397411346, "train_loss_bc": 0.030923573300242424, "train_loss_llm": 0.0008786618709564209, "grad_norm": 0.028909876942634583, "global_step": 24097, "epoch": 669, "lr": 0.001} +{"train_loss": 0.024878207594156265, "train_loss_bc": 0.024878207594156265, "train_loss_llm": 0.0, "grad_norm": 0.04335888475179672, "global_step": 24098, "epoch": 669, "lr": 0.001} +{"train_loss": 0.02717064879834652, "train_loss_bc": 0.02715359814465046, "train_loss_llm": 0.0017051100730895996, "grad_norm": 0.0545538067817688, "global_step": 24099, "epoch": 669, "lr": 0.001} +{"train_loss": 0.026992950588464737, "train_loss_bc": 0.026992950588464737, "train_loss_llm": 0.0, "grad_norm": 0.06435976922512054, "global_step": 24100, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03736771270632744, "train_loss_bc": 0.03736771270632744, "train_loss_llm": 0.0, "grad_norm": 0.04170013591647148, "global_step": 24101, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03471202403306961, "train_loss_bc": 0.03471202403306961, "train_loss_llm": 0.0, "grad_norm": 0.06683129072189331, "global_step": 24102, "epoch": 669, "lr": 0.001} +{"train_loss": 0.036440979689359665, "train_loss_bc": 0.03643980622291565, "train_loss_llm": 0.00011730194091796875, "grad_norm": 0.08754970878362656, "global_step": 24103, "epoch": 669, "lr": 0.001} +{"train_loss": 0.028161006048321724, "train_loss_bc": 0.028161006048321724, "train_loss_llm": 0.0, "grad_norm": 0.10581187903881073, "global_step": 24104, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03674323484301567, "train_loss_bc": 0.03674323484301567, "train_loss_llm": 0.0, "grad_norm": 0.04482867941260338, "global_step": 24105, "epoch": 669, "lr": 0.001} +{"train_loss": 0.033534735441207886, "train_loss_bc": 0.033531781286001205, "train_loss_llm": 0.0002955794334411621, "grad_norm": 0.06639663875102997, "global_step": 24106, "epoch": 669, "lr": 0.001} +{"train_loss": 0.031165724620223045, "train_loss_bc": 0.031165724620223045, "train_loss_llm": 0.0, "grad_norm": 0.07906562834978104, "global_step": 24107, "epoch": 669, "lr": 0.001} +{"train_loss": 0.0309832114726305, "train_loss_bc": 0.030962690711021423, "train_loss_llm": 0.002052009105682373, "grad_norm": 0.09807837009429932, "global_step": 24108, "epoch": 669, "lr": 0.001} +{"train_loss": 0.0341423936188221, "train_loss_bc": 0.0341423936188221, "train_loss_llm": 0.0, "grad_norm": 0.04106433317065239, "global_step": 24109, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03075043484568596, "train_loss_bc": 0.030750323086977005, "train_loss_llm": 1.1175870895385742e-05, "grad_norm": 0.06415310502052307, "global_step": 24110, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03476296737790108, "train_loss_bc": 0.03476296737790108, "train_loss_llm": 0.0, "grad_norm": 0.08367133885622025, "global_step": 24111, "epoch": 669, "lr": 0.001} +{"train_loss": 0.02740844339132309, "train_loss_bc": 0.02740451693534851, "train_loss_llm": 0.00039267539978027344, "grad_norm": 0.0971716046333313, "global_step": 24112, "epoch": 669, "lr": 0.001} +{"train_loss": 0.033827219158411026, "train_loss_bc": 0.033827219158411026, "train_loss_llm": 0.0, "grad_norm": 0.03047426976263523, "global_step": 24113, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03351108357310295, "train_loss_bc": 0.03351108357310295, "train_loss_llm": 0.0, "grad_norm": 0.045859452337026596, "global_step": 24114, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03480253368616104, "train_loss_bc": 0.03480253368616104, "train_loss_llm": 0.0, "grad_norm": 0.05975675210356712, "global_step": 24115, "epoch": 669, "lr": 0.001} +{"train_loss": 0.02925604209303856, "train_loss_bc": 0.02925604209303856, "train_loss_llm": 0.0, "grad_norm": 0.059668973088264465, "global_step": 24116, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03277044743299484, "train_loss_bc": 0.03277044743299484, "train_loss_llm": 0.0, "grad_norm": 0.029668834060430527, "global_step": 24117, "epoch": 669, "lr": 0.001} +{"train_loss": 0.03449021279811859, "train_loss_bc": 0.03449021279811859, "train_loss_llm": 0.0, "grad_norm": 0.051258303225040436, "global_step": 24118, "epoch": 669, "lr": 0.001} +{"train_loss": 0.031825784800781146, "train_loss_bc": 0.033837758004665375, "train_loss_llm": 0.0, "grad_norm": 0.06510740518569946, "global_step": 24119, "epoch": 669, "lr": 0.001, "val_loss": 0.08641892671585083} +{"train_loss": 0.02748391032218933, "train_loss_bc": 0.02748391032218933, "train_loss_llm": 0.0, "grad_norm": 0.07794935256242752, "global_step": 24120, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03360382094979286, "train_loss_bc": 0.03360382094979286, "train_loss_llm": 0.0, "grad_norm": 0.03172020614147186, "global_step": 24121, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03165779262781143, "train_loss_bc": 0.03164852038025856, "train_loss_llm": 0.0009272992610931396, "grad_norm": 0.03944754600524902, "global_step": 24122, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03136109560728073, "train_loss_bc": 0.03134915605187416, "train_loss_llm": 0.001194089651107788, "grad_norm": 0.04726856201887131, "global_step": 24123, "epoch": 670, "lr": 0.001} +{"train_loss": 0.026656584814190865, "train_loss_bc": 0.026656584814190865, "train_loss_llm": 0.0, "grad_norm": 0.05880703032016754, "global_step": 24124, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03156541287899017, "train_loss_bc": 0.03156541287899017, "train_loss_llm": 0.0, "grad_norm": 0.03586803376674652, "global_step": 24125, "epoch": 670, "lr": 0.001} +{"train_loss": 0.033842362463474274, "train_loss_bc": 0.033842362463474274, "train_loss_llm": 0.0, "grad_norm": 0.0641072541475296, "global_step": 24126, "epoch": 670, "lr": 0.001} +{"train_loss": 0.035037510097026825, "train_loss_bc": 0.035037510097026825, "train_loss_llm": 0.0, "grad_norm": 0.0869661346077919, "global_step": 24127, "epoch": 670, "lr": 0.001} +{"train_loss": 0.029330771416425705, "train_loss_bc": 0.029330771416425705, "train_loss_llm": 0.0, "grad_norm": 0.10592115670442581, "global_step": 24128, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03453439101576805, "train_loss_bc": 0.03453439101576805, "train_loss_llm": 0.0, "grad_norm": 0.039021655917167664, "global_step": 24129, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03487586975097656, "train_loss_bc": 0.03487586975097656, "train_loss_llm": 0.0, "grad_norm": 0.06865976750850677, "global_step": 24130, "epoch": 670, "lr": 0.001} +{"train_loss": 0.02651403471827507, "train_loss_bc": 0.02651403471827507, "train_loss_llm": 0.0, "grad_norm": 0.08050136268138885, "global_step": 24131, "epoch": 670, "lr": 0.001} +{"train_loss": 0.033855337649583817, "train_loss_bc": 0.033855337649583817, "train_loss_llm": 0.0, "grad_norm": 0.09619586169719696, "global_step": 24132, "epoch": 670, "lr": 0.001} +{"train_loss": 0.0329277478158474, "train_loss_bc": 0.0329277478158474, "train_loss_llm": 0.0, "grad_norm": 0.03802790120244026, "global_step": 24133, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03146568685770035, "train_loss_bc": 0.03146568685770035, "train_loss_llm": 0.0, "grad_norm": 0.056146834045648575, "global_step": 24134, "epoch": 670, "lr": 0.001} +{"train_loss": 0.031621720641851425, "train_loss_bc": 0.03159770742058754, "train_loss_llm": 0.002401202917098999, "grad_norm": 0.0673680528998375, "global_step": 24135, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03147029131650925, "train_loss_bc": 0.03147029131650925, "train_loss_llm": 0.0, "grad_norm": 0.09211576730012894, "global_step": 24136, "epoch": 670, "lr": 0.001} +{"train_loss": 0.02863754890859127, "train_loss_bc": 0.02863616682589054, "train_loss_llm": 0.00013822317123413086, "grad_norm": 0.030385032296180725, "global_step": 24137, "epoch": 670, "lr": 0.001} +{"train_loss": 0.039122048765420914, "train_loss_bc": 0.039122048765420914, "train_loss_llm": 0.0, "grad_norm": 0.04567022621631622, "global_step": 24138, "epoch": 670, "lr": 0.001} +{"train_loss": 0.0313839390873909, "train_loss_bc": 0.0313839390873909, "train_loss_llm": 0.0, "grad_norm": 0.04761650413274765, "global_step": 24139, "epoch": 670, "lr": 0.001} +{"train_loss": 0.02641109749674797, "train_loss_bc": 0.02641109749674797, "train_loss_llm": 0.0, "grad_norm": 0.05626592040061951, "global_step": 24140, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03817884251475334, "train_loss_bc": 0.03817884251475334, "train_loss_llm": 0.0, "grad_norm": 0.03958863765001297, "global_step": 24141, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03105185367166996, "train_loss_bc": 0.03105185367166996, "train_loss_llm": 0.0, "grad_norm": 0.06487862020730972, "global_step": 24142, "epoch": 670, "lr": 0.001} +{"train_loss": 0.038775485008955, "train_loss_bc": 0.038775485008955, "train_loss_llm": 0.0, "grad_norm": 0.08451051265001297, "global_step": 24143, "epoch": 670, "lr": 0.001} +{"train_loss": 0.026110060513019562, "train_loss_bc": 0.026110060513019562, "train_loss_llm": 0.0, "grad_norm": 0.09840397536754608, "global_step": 24144, "epoch": 670, "lr": 0.001} +{"train_loss": 0.026629479601979256, "train_loss_bc": 0.026629479601979256, "train_loss_llm": 0.0, "grad_norm": 0.02569635771214962, "global_step": 24145, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03413365036249161, "train_loss_bc": 0.03413365036249161, "train_loss_llm": 0.0, "grad_norm": 0.0448545441031456, "global_step": 24146, "epoch": 670, "lr": 0.001} +{"train_loss": 0.034348972141742706, "train_loss_bc": 0.034348972141742706, "train_loss_llm": 0.0, "grad_norm": 0.06435108184814453, "global_step": 24147, "epoch": 670, "lr": 0.001} +{"train_loss": 0.028715813532471657, "train_loss_bc": 0.028709659352898598, "train_loss_llm": 0.0006154477596282959, "grad_norm": 0.0803721696138382, "global_step": 24148, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03182077780365944, "train_loss_bc": 0.031819455325603485, "train_loss_llm": 0.0001322329044342041, "grad_norm": 0.03260834142565727, "global_step": 24149, "epoch": 670, "lr": 0.001} +{"train_loss": 0.034491825848817825, "train_loss_bc": 0.034479156136512756, "train_loss_llm": 0.0012669563293457031, "grad_norm": 0.06198592483997345, "global_step": 24150, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03253506124019623, "train_loss_bc": 0.03253506124019623, "train_loss_llm": 0.0, "grad_norm": 0.09237875044345856, "global_step": 24151, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03140374273061752, "train_loss_bc": 0.03140374273061752, "train_loss_llm": 0.0, "grad_norm": 0.1157386526465416, "global_step": 24152, "epoch": 670, "lr": 0.001} +{"train_loss": 0.029313720762729645, "train_loss_bc": 0.029313720762729645, "train_loss_llm": 0.0, "grad_norm": 0.03165433555841446, "global_step": 24153, "epoch": 670, "lr": 0.001} +{"train_loss": 0.032541826367378235, "train_loss_bc": 0.03253629803657532, "train_loss_llm": 0.0005526542663574219, "grad_norm": 0.060889530926942825, "global_step": 24154, "epoch": 670, "lr": 0.001} +{"train_loss": 0.03177104579905669, "train_loss_bc": 0.030336864292621613, "train_loss_llm": 0.0010696351528167725, "grad_norm": 0.08198685199022293, "global_step": 24155, "epoch": 670, "lr": 0.001, "train/cumulative_reward": 28.528886798990435, "train/mean_score": 8.527413046185657, "train/success_rate": 0.2, "test/cumulative_reward": 29.015095952898395, "test/mean_score": 9.670366829506921, "test/success_rate": 0.26, "val_loss": 0.087018221616745, "train_action_mse_error": 0.03483780473470688} +{"train_loss": 0.02732544019818306, "train_loss_bc": 0.02732544019818306, "train_loss_llm": 0.0, "grad_norm": 0.0935346931219101, "global_step": 24156, "epoch": 671, "lr": 0.001} +{"train_loss": 0.030210034921765327, "train_loss_bc": 0.030210034921765327, "train_loss_llm": 0.0, "grad_norm": 0.03923080489039421, "global_step": 24157, "epoch": 671, "lr": 0.001} +{"train_loss": 0.0308894831687212, "train_loss_bc": 0.0308894831687212, "train_loss_llm": 0.0, "grad_norm": 0.059334978461265564, "global_step": 24158, "epoch": 671, "lr": 0.001} +{"train_loss": 0.02907741069793701, "train_loss_bc": 0.02907741069793701, "train_loss_llm": 0.0, "grad_norm": 0.07210616767406464, "global_step": 24159, "epoch": 671, "lr": 0.001} +{"train_loss": 0.029238982126116753, "train_loss_bc": 0.029237201437354088, "train_loss_llm": 0.00017815828323364258, "grad_norm": 0.08000541478395462, "global_step": 24160, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03447125852108002, "train_loss_bc": 0.03447125852108002, "train_loss_llm": 0.0, "grad_norm": 0.03257155045866966, "global_step": 24161, "epoch": 671, "lr": 0.001} +{"train_loss": 0.030060356482863426, "train_loss_bc": 0.030060356482863426, "train_loss_llm": 0.0, "grad_norm": 0.05284950137138367, "global_step": 24162, "epoch": 671, "lr": 0.001} +{"train_loss": 0.023076122626662254, "train_loss_bc": 0.023076122626662254, "train_loss_llm": 0.0, "grad_norm": 0.06491647660732269, "global_step": 24163, "epoch": 671, "lr": 0.001} +{"train_loss": 0.02985437400639057, "train_loss_bc": 0.02984839305281639, "train_loss_llm": 0.0005980432033538818, "grad_norm": 0.08858650177717209, "global_step": 24164, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03230776637792587, "train_loss_bc": 0.03230598196387291, "train_loss_llm": 0.000178605318069458, "grad_norm": 0.02895008958876133, "global_step": 24165, "epoch": 671, "lr": 0.001} +{"train_loss": 0.035005033016204834, "train_loss_bc": 0.035005033016204834, "train_loss_llm": 0.0, "grad_norm": 0.04287128895521164, "global_step": 24166, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03076951391994953, "train_loss_bc": 0.03076951391994953, "train_loss_llm": 0.0, "grad_norm": 0.05581262335181236, "global_step": 24167, "epoch": 671, "lr": 0.001} +{"train_loss": 0.030278827995061874, "train_loss_bc": 0.030278827995061874, "train_loss_llm": 0.0, "grad_norm": 0.0687607005238533, "global_step": 24168, "epoch": 671, "lr": 0.001} +{"train_loss": 0.028694478794932365, "train_loss_bc": 0.028694478794932365, "train_loss_llm": 0.0, "grad_norm": 0.02956997975707054, "global_step": 24169, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03172534704208374, "train_loss_bc": 0.0317206047475338, "train_loss_llm": 0.0004743337631225586, "grad_norm": 0.03726811334490776, "global_step": 24170, "epoch": 671, "lr": 0.001} +{"train_loss": 0.029073821380734444, "train_loss_bc": 0.029073821380734444, "train_loss_llm": 0.0, "grad_norm": 0.05403425544500351, "global_step": 24171, "epoch": 671, "lr": 0.001} +{"train_loss": 0.034532126039266586, "train_loss_bc": 0.034532126039266586, "train_loss_llm": 0.0, "grad_norm": 0.080045685172081, "global_step": 24172, "epoch": 671, "lr": 0.001} +{"train_loss": 0.030174052342772484, "train_loss_bc": 0.030159544199705124, "train_loss_llm": 0.0014508068561553955, "grad_norm": 0.031743232160806656, "global_step": 24173, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03237645700573921, "train_loss_bc": 0.03237645700573921, "train_loss_llm": 0.0, "grad_norm": 0.06365422904491425, "global_step": 24174, "epoch": 671, "lr": 0.001} +{"train_loss": 0.02506837248802185, "train_loss_bc": 0.02506837248802185, "train_loss_llm": 0.0, "grad_norm": 0.07546515762805939, "global_step": 24175, "epoch": 671, "lr": 0.001} +{"train_loss": 0.035137273371219635, "train_loss_bc": 0.035137273371219635, "train_loss_llm": 0.0, "grad_norm": 0.09621544182300568, "global_step": 24176, "epoch": 671, "lr": 0.001} +{"train_loss": 0.026171773672103882, "train_loss_bc": 0.026171773672103882, "train_loss_llm": 0.0, "grad_norm": 0.028615422546863556, "global_step": 24177, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03445085883140564, "train_loss_bc": 0.03445085883140564, "train_loss_llm": 0.0, "grad_norm": 0.04631514847278595, "global_step": 24178, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03816569969058037, "train_loss_bc": 0.03816569969058037, "train_loss_llm": 0.0, "grad_norm": 0.0612635463476181, "global_step": 24179, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03524835780262947, "train_loss_bc": 0.03524835780262947, "train_loss_llm": 0.0, "grad_norm": 0.07619491219520569, "global_step": 24180, "epoch": 671, "lr": 0.001} +{"train_loss": 0.030424153432250023, "train_loss_bc": 0.030418092384934425, "train_loss_llm": 0.0006060898303985596, "grad_norm": 0.031909700483083725, "global_step": 24181, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03546861186623573, "train_loss_bc": 0.035462941974401474, "train_loss_llm": 0.0005668103694915771, "grad_norm": 0.05166575685143471, "global_step": 24182, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03021318092942238, "train_loss_bc": 0.03021318092942238, "train_loss_llm": 0.0, "grad_norm": 0.07432537525892258, "global_step": 24183, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03384564444422722, "train_loss_bc": 0.03384564444422722, "train_loss_llm": 0.0, "grad_norm": 0.09493385255336761, "global_step": 24184, "epoch": 671, "lr": 0.001} +{"train_loss": 0.028236402198672295, "train_loss_bc": 0.028232790529727936, "train_loss_llm": 0.0003611147403717041, "grad_norm": 0.03371533378958702, "global_step": 24185, "epoch": 671, "lr": 0.001} +{"train_loss": 0.028924712911248207, "train_loss_bc": 0.028924712911248207, "train_loss_llm": 0.0, "grad_norm": 0.06075473502278328, "global_step": 24186, "epoch": 671, "lr": 0.001} +{"train_loss": 0.038668230175971985, "train_loss_bc": 0.038668230175971985, "train_loss_llm": 0.0, "grad_norm": 0.07499872893095016, "global_step": 24187, "epoch": 671, "lr": 0.001} +{"train_loss": 0.0354483388364315, "train_loss_bc": 0.0354483388364315, "train_loss_llm": 0.0, "grad_norm": 0.10300370305776596, "global_step": 24188, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03731641173362732, "train_loss_bc": 0.03731641173362732, "train_loss_llm": 0.0, "grad_norm": 0.04618748649954796, "global_step": 24189, "epoch": 671, "lr": 0.001} +{"train_loss": 0.034517802298069, "train_loss_bc": 0.034517802298069, "train_loss_llm": 0.0, "grad_norm": 0.06481434404850006, "global_step": 24190, "epoch": 671, "lr": 0.001} +{"train_loss": 0.03169538116910391, "train_loss_bc": 0.034587010741233826, "train_loss_llm": 0.0, "grad_norm": 0.09197995811700821, "global_step": 24191, "epoch": 671, "lr": 0.001, "val_loss": 0.08709210902452469} +{"train_loss": 0.026733044534921646, "train_loss_bc": 0.026715349406003952, "train_loss_llm": 0.0017695724964141846, "grad_norm": 0.1120343804359436, "global_step": 24192, "epoch": 672, "lr": 0.001} +{"train_loss": 0.035815075039863586, "train_loss_bc": 0.035815075039863586, "train_loss_llm": 0.0, "grad_norm": 0.02975097857415676, "global_step": 24193, "epoch": 672, "lr": 0.001} +{"train_loss": 0.026658130809664726, "train_loss_bc": 0.026658130809664726, "train_loss_llm": 0.0, "grad_norm": 0.05427875742316246, "global_step": 24194, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03461083024740219, "train_loss_bc": 0.03461083024740219, "train_loss_llm": 0.0, "grad_norm": 0.07423289865255356, "global_step": 24195, "epoch": 672, "lr": 0.001} +{"train_loss": 0.030767405405640602, "train_loss_bc": 0.030767405405640602, "train_loss_llm": 0.0, "grad_norm": 0.09929194301366806, "global_step": 24196, "epoch": 672, "lr": 0.001} +{"train_loss": 0.028144309297204018, "train_loss_bc": 0.028144309297204018, "train_loss_llm": 0.0, "grad_norm": 0.02501547522842884, "global_step": 24197, "epoch": 672, "lr": 0.001} +{"train_loss": 0.025587286800146103, "train_loss_bc": 0.025587286800146103, "train_loss_llm": 0.0, "grad_norm": 0.039628803730010986, "global_step": 24198, "epoch": 672, "lr": 0.001} +{"train_loss": 0.030313648283481598, "train_loss_bc": 0.03029319830238819, "train_loss_llm": 0.00204506516456604, "grad_norm": 0.06065196543931961, "global_step": 24199, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03601456806063652, "train_loss_bc": 0.03601456806063652, "train_loss_llm": 0.0, "grad_norm": 0.07856869697570801, "global_step": 24200, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03659496456384659, "train_loss_bc": 0.03659496456384659, "train_loss_llm": 0.0, "grad_norm": 0.03189793601632118, "global_step": 24201, "epoch": 672, "lr": 0.001} +{"train_loss": 0.030870504677295685, "train_loss_bc": 0.030870504677295685, "train_loss_llm": 0.0, "grad_norm": 0.05791022628545761, "global_step": 24202, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03932509943842888, "train_loss_bc": 0.03932509943842888, "train_loss_llm": 0.0, "grad_norm": 0.07948078960180283, "global_step": 24203, "epoch": 672, "lr": 0.001} +{"train_loss": 0.02672196738421917, "train_loss_bc": 0.02672196738421917, "train_loss_llm": 0.0, "grad_norm": 0.09443536400794983, "global_step": 24204, "epoch": 672, "lr": 0.001} +{"train_loss": 0.026141852140426636, "train_loss_bc": 0.026141852140426636, "train_loss_llm": 0.0, "grad_norm": 0.029532283544540405, "global_step": 24205, "epoch": 672, "lr": 0.001} +{"train_loss": 0.032166220247745514, "train_loss_bc": 0.032166220247745514, "train_loss_llm": 0.0, "grad_norm": 0.05196535214781761, "global_step": 24206, "epoch": 672, "lr": 0.001} +{"train_loss": 0.029608994722366333, "train_loss_bc": 0.029608994722366333, "train_loss_llm": 0.0, "grad_norm": 0.0668170377612114, "global_step": 24207, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03291551396250725, "train_loss_bc": 0.03291551396250725, "train_loss_llm": 0.0, "grad_norm": 0.08445889502763748, "global_step": 24208, "epoch": 672, "lr": 0.001} +{"train_loss": 0.031720664352178574, "train_loss_bc": 0.031720664352178574, "train_loss_llm": 0.0, "grad_norm": 0.028241997584700584, "global_step": 24209, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03626615181565285, "train_loss_bc": 0.03626615181565285, "train_loss_llm": 0.0, "grad_norm": 0.04849829897284508, "global_step": 24210, "epoch": 672, "lr": 0.001} +{"train_loss": 0.029891975224018097, "train_loss_bc": 0.029891975224018097, "train_loss_llm": 0.0, "grad_norm": 0.05819569155573845, "global_step": 24211, "epoch": 672, "lr": 0.001} +{"train_loss": 0.028752470389008522, "train_loss_bc": 0.028742283582687378, "train_loss_llm": 0.0010187029838562012, "grad_norm": 0.06552629172801971, "global_step": 24212, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03629176691174507, "train_loss_bc": 0.03629176691174507, "train_loss_llm": 0.0, "grad_norm": 0.04257906973361969, "global_step": 24213, "epoch": 672, "lr": 0.001} +{"train_loss": 0.032635290175676346, "train_loss_bc": 0.032635290175676346, "train_loss_llm": 0.0, "grad_norm": 0.048579029738903046, "global_step": 24214, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03483321890234947, "train_loss_bc": 0.03483321890234947, "train_loss_llm": 0.0, "grad_norm": 0.0630357638001442, "global_step": 24215, "epoch": 672, "lr": 0.001} +{"train_loss": 0.032346419990062714, "train_loss_bc": 0.032346419990062714, "train_loss_llm": 0.0, "grad_norm": 0.08533631265163422, "global_step": 24216, "epoch": 672, "lr": 0.001} +{"train_loss": 0.027918515726923943, "train_loss_bc": 0.027918515726923943, "train_loss_llm": 0.0, "grad_norm": 0.03112609125673771, "global_step": 24217, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03007112443447113, "train_loss_bc": 0.03007112443447113, "train_loss_llm": 0.0, "grad_norm": 0.0500355027616024, "global_step": 24218, "epoch": 672, "lr": 0.001} +{"train_loss": 0.030737845227122307, "train_loss_bc": 0.03072531335055828, "train_loss_llm": 0.0012532174587249756, "grad_norm": 0.07191696017980576, "global_step": 24219, "epoch": 672, "lr": 0.001} +{"train_loss": 0.034421760588884354, "train_loss_bc": 0.034421760588884354, "train_loss_llm": 0.0, "grad_norm": 0.09581007808446884, "global_step": 24220, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03052881546318531, "train_loss_bc": 0.03052881546318531, "train_loss_llm": 0.0, "grad_norm": 0.029491985216736794, "global_step": 24221, "epoch": 672, "lr": 0.001} +{"train_loss": 0.036375563591718674, "train_loss_bc": 0.03637220337986946, "train_loss_llm": 0.00033602118492126465, "grad_norm": 0.04448925331234932, "global_step": 24222, "epoch": 672, "lr": 0.001} +{"train_loss": 0.030659066513180733, "train_loss_bc": 0.030659066513180733, "train_loss_llm": 0.0, "grad_norm": 0.055998098105192184, "global_step": 24223, "epoch": 672, "lr": 0.001} +{"train_loss": 0.036374956369400024, "train_loss_bc": 0.036374956369400024, "train_loss_llm": 0.0, "grad_norm": 0.07337654381990433, "global_step": 24224, "epoch": 672, "lr": 0.001} +{"train_loss": 0.02581658959388733, "train_loss_bc": 0.02581658959388733, "train_loss_llm": 0.0, "grad_norm": 0.039291396737098694, "global_step": 24225, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03390353545546532, "train_loss_bc": 0.03389054164290428, "train_loss_llm": 0.0012993812561035156, "grad_norm": 0.06488444656133652, "global_step": 24226, "epoch": 672, "lr": 0.001} +{"train_loss": 0.03160343427831928, "train_loss_bc": 0.02918848767876625, "train_loss_llm": 0.0, "grad_norm": 0.09046770632266998, "global_step": 24227, "epoch": 672, "lr": 0.001, "val_loss": 0.08709630370140076} +{"train_loss": 0.032209329307079315, "train_loss_bc": 0.032209329307079315, "train_loss_llm": 0.0, "grad_norm": 0.11985749006271362, "global_step": 24228, "epoch": 673, "lr": 0.001} +{"train_loss": 0.033495452255010605, "train_loss_bc": 0.033495452255010605, "train_loss_llm": 0.0, "grad_norm": 0.03342006355524063, "global_step": 24229, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03540439158678055, "train_loss_bc": 0.03540439158678055, "train_loss_llm": 0.0, "grad_norm": 0.06171518936753273, "global_step": 24230, "epoch": 673, "lr": 0.001} +{"train_loss": 0.032483670860528946, "train_loss_bc": 0.03247954323887825, "train_loss_llm": 0.0004127919673919678, "grad_norm": 0.07878458499908447, "global_step": 24231, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03679681941866875, "train_loss_bc": 0.03679681941866875, "train_loss_llm": 0.0, "grad_norm": 0.08181991428136826, "global_step": 24232, "epoch": 673, "lr": 0.001} +{"train_loss": 0.030880015343427658, "train_loss_bc": 0.030880015343427658, "train_loss_llm": 0.0, "grad_norm": 0.02991987019777298, "global_step": 24233, "epoch": 673, "lr": 0.001} +{"train_loss": 0.02953614480793476, "train_loss_bc": 0.02953614480793476, "train_loss_llm": 0.0, "grad_norm": 0.056218989193439484, "global_step": 24234, "epoch": 673, "lr": 0.001} +{"train_loss": 0.02963963709771633, "train_loss_bc": 0.029638953506946564, "train_loss_llm": 6.839632987976074e-05, "grad_norm": 0.07786986231803894, "global_step": 24235, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03094385378062725, "train_loss_bc": 0.030918246135115623, "train_loss_llm": 0.002560734748840332, "grad_norm": 0.09472200274467468, "global_step": 24236, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03285324573516846, "train_loss_bc": 0.03285324573516846, "train_loss_llm": 0.0, "grad_norm": 0.02634698897600174, "global_step": 24237, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03123163990676403, "train_loss_bc": 0.03121793456375599, "train_loss_llm": 0.001370459794998169, "grad_norm": 0.04384227469563484, "global_step": 24238, "epoch": 673, "lr": 0.001} +{"train_loss": 0.027839774265885353, "train_loss_bc": 0.027839774265885353, "train_loss_llm": 0.0, "grad_norm": 0.06028589978814125, "global_step": 24239, "epoch": 673, "lr": 0.001} +{"train_loss": 0.028268052265048027, "train_loss_bc": 0.028268052265048027, "train_loss_llm": 0.0, "grad_norm": 0.07349544018507004, "global_step": 24240, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03704185038805008, "train_loss_bc": 0.03704185038805008, "train_loss_llm": 0.0, "grad_norm": 0.027671704068779945, "global_step": 24241, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03383564203977585, "train_loss_bc": 0.03383564203977585, "train_loss_llm": 0.0, "grad_norm": 0.04099077358841896, "global_step": 24242, "epoch": 673, "lr": 0.001} +{"train_loss": 0.027438975870609283, "train_loss_bc": 0.027423720806837082, "train_loss_llm": 0.001525580883026123, "grad_norm": 0.05689115822315216, "global_step": 24243, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03120299056172371, "train_loss_bc": 0.03120299056172371, "train_loss_llm": 0.0, "grad_norm": 0.07184752076864243, "global_step": 24244, "epoch": 673, "lr": 0.001} +{"train_loss": 0.02758335880935192, "train_loss_bc": 0.02758335880935192, "train_loss_llm": 0.0, "grad_norm": 0.033783964812755585, "global_step": 24245, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03527291864156723, "train_loss_bc": 0.03527291864156723, "train_loss_llm": 0.0, "grad_norm": 0.0681694969534874, "global_step": 24246, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03209614381194115, "train_loss_bc": 0.03209614381194115, "train_loss_llm": 0.0, "grad_norm": 0.09498903900384903, "global_step": 24247, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03340476006269455, "train_loss_bc": 0.03340476006269455, "train_loss_llm": 0.0, "grad_norm": 0.11991077661514282, "global_step": 24248, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03258313611149788, "train_loss_bc": 0.03258313611149788, "train_loss_llm": 0.0, "grad_norm": 0.03310195356607437, "global_step": 24249, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03598196059465408, "train_loss_bc": 0.03598196059465408, "train_loss_llm": 0.0, "grad_norm": 0.060444172471761703, "global_step": 24250, "epoch": 673, "lr": 0.001} +{"train_loss": 0.027050921693444252, "train_loss_bc": 0.027050921693444252, "train_loss_llm": 0.0, "grad_norm": 0.07404211908578873, "global_step": 24251, "epoch": 673, "lr": 0.001} +{"train_loss": 0.025737127289175987, "train_loss_bc": 0.025737127289175987, "train_loss_llm": 0.0, "grad_norm": 0.094930499792099, "global_step": 24252, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03369157388806343, "train_loss_bc": 0.03367630019783974, "train_loss_llm": 0.0015274882316589355, "grad_norm": 0.03298821672797203, "global_step": 24253, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03386529162526131, "train_loss_bc": 0.03384055569767952, "train_loss_llm": 0.0024735331535339355, "grad_norm": 0.049995556473731995, "global_step": 24254, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03347693756222725, "train_loss_bc": 0.03347693756222725, "train_loss_llm": 0.0, "grad_norm": 0.06625150889158249, "global_step": 24255, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03145294263958931, "train_loss_bc": 0.03145294263958931, "train_loss_llm": 0.0, "grad_norm": 0.08091951161623001, "global_step": 24256, "epoch": 673, "lr": 0.001} +{"train_loss": 0.032847579568624496, "train_loss_bc": 0.032847579568624496, "train_loss_llm": 0.0, "grad_norm": 0.026502525433897972, "global_step": 24257, "epoch": 673, "lr": 0.001} +{"train_loss": 0.025870071724057198, "train_loss_bc": 0.025870071724057198, "train_loss_llm": 0.0, "grad_norm": 0.03363645076751709, "global_step": 24258, "epoch": 673, "lr": 0.001} +{"train_loss": 0.032789330929517746, "train_loss_bc": 0.032789330929517746, "train_loss_llm": 0.0, "grad_norm": 0.0500478632748127, "global_step": 24259, "epoch": 673, "lr": 0.001} +{"train_loss": 0.03007333353161812, "train_loss_bc": 0.03006647899746895, "train_loss_llm": 0.0006855130195617676, "grad_norm": 0.056835949420928955, "global_step": 24260, "epoch": 673, "lr": 0.001} +{"train_loss": 0.029370000585913658, "train_loss_bc": 0.029370000585913658, "train_loss_llm": 0.0, "grad_norm": 0.04165712743997574, "global_step": 24261, "epoch": 673, "lr": 0.001} +{"train_loss": 0.036723919212818146, "train_loss_bc": 0.036723919212818146, "train_loss_llm": 0.0, "grad_norm": 0.0654960349202156, "global_step": 24262, "epoch": 673, "lr": 0.001} +{"train_loss": 0.031728621003114514, "train_loss_bc": 0.03125756233930588, "train_loss_llm": 0.0, "grad_norm": 0.07799746841192245, "global_step": 24263, "epoch": 673, "lr": 0.001, "val_loss": 0.08820739388465881} +{"train_loss": 0.029337260872125626, "train_loss_bc": 0.029330410063266754, "train_loss_llm": 0.0006851553916931152, "grad_norm": 0.10335992276668549, "global_step": 24264, "epoch": 674, "lr": 0.001} +{"train_loss": 0.033569734543561935, "train_loss_bc": 0.033569734543561935, "train_loss_llm": 0.0, "grad_norm": 0.03346527740359306, "global_step": 24265, "epoch": 674, "lr": 0.001} +{"train_loss": 0.037075258791446686, "train_loss_bc": 0.03707282617688179, "train_loss_llm": 0.00024318695068359375, "grad_norm": 0.06078936904668808, "global_step": 24266, "epoch": 674, "lr": 0.001} +{"train_loss": 0.024786485359072685, "train_loss_bc": 0.024786485359072685, "train_loss_llm": 0.0, "grad_norm": 0.08184865117073059, "global_step": 24267, "epoch": 674, "lr": 0.001} +{"train_loss": 0.03047565370798111, "train_loss_bc": 0.030441945418715477, "train_loss_llm": 0.0033707618713378906, "grad_norm": 0.09952060878276825, "global_step": 24268, "epoch": 674, "lr": 0.001} +{"train_loss": 0.02986709401011467, "train_loss_bc": 0.02986709401011467, "train_loss_llm": 0.0, "grad_norm": 0.03366634249687195, "global_step": 24269, "epoch": 674, "lr": 0.001} +{"train_loss": 0.03598377853631973, "train_loss_bc": 0.03597906604409218, "train_loss_llm": 0.0004711151123046875, "grad_norm": 0.06306101381778717, "global_step": 24270, "epoch": 674, "lr": 0.001} +{"train_loss": 0.030510976910591125, "train_loss_bc": 0.030510976910591125, "train_loss_llm": 0.0, "grad_norm": 0.0800337865948677, "global_step": 24271, "epoch": 674, "lr": 0.001} +{"train_loss": 0.038146719336509705, "train_loss_bc": 0.038146719336509705, "train_loss_llm": 0.0, "grad_norm": 0.10692890733480453, "global_step": 24272, "epoch": 674, "lr": 0.001} +{"train_loss": 0.027301738038659096, "train_loss_bc": 0.027301738038659096, "train_loss_llm": 0.0, "grad_norm": 0.033089764416217804, "global_step": 24273, "epoch": 674, "lr": 0.001} +{"train_loss": 0.030952181667089462, "train_loss_bc": 0.030952181667089462, "train_loss_llm": 0.0, "grad_norm": 0.04794710874557495, "global_step": 24274, "epoch": 674, "lr": 0.001} +{"train_loss": 0.02744562365114689, "train_loss_bc": 0.02744562365114689, "train_loss_llm": 0.0, "grad_norm": 0.06880152970552444, "global_step": 24275, "epoch": 674, "lr": 0.001} +{"train_loss": 0.031106416136026382, "train_loss_bc": 0.031106416136026382, "train_loss_llm": 0.0, "grad_norm": 0.0869869589805603, "global_step": 24276, "epoch": 674, "lr": 0.001} +{"train_loss": 0.02943500317633152, "train_loss_bc": 0.02943500317633152, "train_loss_llm": 0.0, "grad_norm": 0.03514266014099121, "global_step": 24277, "epoch": 674, "lr": 0.001} +{"train_loss": 0.035459719598293304, "train_loss_bc": 0.035459719598293304, "train_loss_llm": 0.0, "grad_norm": 0.06526254862546921, "global_step": 24278, "epoch": 674, "lr": 0.001} +{"train_loss": 0.028529318049550056, "train_loss_bc": 0.02852555736899376, "train_loss_llm": 0.00037607550621032715, "grad_norm": 0.08207671344280243, "global_step": 24279, "epoch": 674, "lr": 0.001} +{"train_loss": 0.03307772055268288, "train_loss_bc": 0.0330745205283165, "train_loss_llm": 0.00032007694244384766, "grad_norm": 0.09957882761955261, "global_step": 24280, "epoch": 674, "lr": 0.001} +{"train_loss": 0.03181946277618408, "train_loss_bc": 0.03180444985628128, "train_loss_llm": 0.0015013813972473145, "grad_norm": 0.03505139425396919, "global_step": 24281, "epoch": 674, "lr": 0.001} +{"train_loss": 0.02960258536040783, "train_loss_bc": 0.02960258536040783, "train_loss_llm": 0.0, "grad_norm": 0.0470598042011261, "global_step": 24282, "epoch": 674, "lr": 0.001} +{"train_loss": 0.028629329055547714, "train_loss_bc": 0.028629329055547714, "train_loss_llm": 0.0, "grad_norm": 0.05932916700839996, "global_step": 24283, "epoch": 674, "lr": 0.001} +{"train_loss": 0.03827942535281181, "train_loss_bc": 0.03827942535281181, "train_loss_llm": 0.0, "grad_norm": 0.07864312082529068, "global_step": 24284, "epoch": 674, "lr": 0.001} +{"train_loss": 0.03441523388028145, "train_loss_bc": 0.03440149873495102, "train_loss_llm": 0.001373589038848877, "grad_norm": 0.03631836920976639, "global_step": 24285, "epoch": 674, "lr": 0.001} +{"train_loss": 0.038880810141563416, "train_loss_bc": 0.038880810141563416, "train_loss_llm": 0.0, "grad_norm": 0.05483540892601013, "global_step": 24286, "epoch": 674, "lr": 0.001} +{"train_loss": 0.03473184257745743, "train_loss_bc": 0.03473184257745743, "train_loss_llm": 0.0, "grad_norm": 0.07008878886699677, "global_step": 24287, "epoch": 674, "lr": 0.001} +{"train_loss": 0.030797703191637993, "train_loss_bc": 0.03078746236860752, "train_loss_llm": 0.0010241568088531494, "grad_norm": 0.09087111055850983, "global_step": 24288, "epoch": 674, "lr": 0.001} +{"train_loss": 0.030712736770510674, "train_loss_bc": 0.030712736770510674, "train_loss_llm": 0.0, "grad_norm": 0.04296359047293663, "global_step": 24289, "epoch": 674, "lr": 0.001} +{"train_loss": 0.03578832373023033, "train_loss_bc": 0.03578832373023033, "train_loss_llm": 0.0, "grad_norm": 0.06981115788221359, "global_step": 24290, "epoch": 674, "lr": 0.001} +{"train_loss": 0.026366638019680977, "train_loss_bc": 0.026366638019680977, "train_loss_llm": 0.0, "grad_norm": 0.08661661297082901, "global_step": 24291, "epoch": 674, "lr": 0.001} +{"train_loss": 0.029914490878582, "train_loss_bc": 0.029914490878582, "train_loss_llm": 0.0, "grad_norm": 0.10894624143838882, "global_step": 24292, "epoch": 674, "lr": 0.001} +{"train_loss": 0.031093571335077286, "train_loss_bc": 0.031093571335077286, "train_loss_llm": 0.0, "grad_norm": 0.03159897029399872, "global_step": 24293, "epoch": 674, "lr": 0.001} +{"train_loss": 0.03435890004038811, "train_loss_bc": 0.03435890004038811, "train_loss_llm": 0.0, "grad_norm": 0.05196820944547653, "global_step": 24294, "epoch": 674, "lr": 0.001} +{"train_loss": 0.030950602144002914, "train_loss_bc": 0.030950602144002914, "train_loss_llm": 0.0, "grad_norm": 0.06085466220974922, "global_step": 24295, "epoch": 674, "lr": 0.001} +{"train_loss": 0.02944757603108883, "train_loss_bc": 0.02944757603108883, "train_loss_llm": 0.0, "grad_norm": 0.07905230671167374, "global_step": 24296, "epoch": 674, "lr": 0.001} +{"train_loss": 0.027927562594413757, "train_loss_bc": 0.027927562594413757, "train_loss_llm": 0.0, "grad_norm": 0.02946379967033863, "global_step": 24297, "epoch": 674, "lr": 0.001} +{"train_loss": 0.0338623933494091, "train_loss_bc": 0.0338623933494091, "train_loss_llm": 0.0, "grad_norm": 0.0444512665271759, "global_step": 24298, "epoch": 674, "lr": 0.001} +{"train_loss": 0.03178583230409357, "train_loss_bc": 0.03365009278059006, "train_loss_llm": 0.0, "grad_norm": 0.061924487352371216, "global_step": 24299, "epoch": 674, "lr": 0.001, "val_loss": 0.08746527135372162} +{"train_loss": 0.032539643347263336, "train_loss_bc": 0.032539643347263336, "train_loss_llm": 0.0, "grad_norm": 0.07692889124155045, "global_step": 24300, "epoch": 675, "lr": 0.001} +{"train_loss": 0.032961633056402206, "train_loss_bc": 0.032951802015304565, "train_loss_llm": 0.0009829699993133545, "grad_norm": 0.03779628872871399, "global_step": 24301, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03333210200071335, "train_loss_bc": 0.03333210200071335, "train_loss_llm": 0.0, "grad_norm": 0.05411585420370102, "global_step": 24302, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03700239211320877, "train_loss_bc": 0.03700239211320877, "train_loss_llm": 0.0, "grad_norm": 0.07246806472539902, "global_step": 24303, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03269766643643379, "train_loss_bc": 0.03269766643643379, "train_loss_llm": 0.0, "grad_norm": 0.09242118149995804, "global_step": 24304, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03436638414859772, "train_loss_bc": 0.03436095267534256, "train_loss_llm": 0.0005429983139038086, "grad_norm": 0.03174630552530289, "global_step": 24305, "epoch": 675, "lr": 0.001} +{"train_loss": 0.033888883888721466, "train_loss_bc": 0.033888883888721466, "train_loss_llm": 0.0, "grad_norm": 0.05165403336286545, "global_step": 24306, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03181243687868118, "train_loss_bc": 0.03181243687868118, "train_loss_llm": 0.0, "grad_norm": 0.06849689781665802, "global_step": 24307, "epoch": 675, "lr": 0.001} +{"train_loss": 0.029947061091661453, "train_loss_bc": 0.029932189732789993, "train_loss_llm": 0.001487135887145996, "grad_norm": 0.07753810286521912, "global_step": 24308, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03154774382710457, "train_loss_bc": 0.03154774382710457, "train_loss_llm": 0.0, "grad_norm": 0.03135250508785248, "global_step": 24309, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03285779058933258, "train_loss_bc": 0.03284045681357384, "train_loss_llm": 0.0017334818840026855, "grad_norm": 0.047520190477371216, "global_step": 24310, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03387519717216492, "train_loss_bc": 0.03387519717216492, "train_loss_llm": 0.0, "grad_norm": 0.06776444613933563, "global_step": 24311, "epoch": 675, "lr": 0.001} +{"train_loss": 0.028729185461997986, "train_loss_bc": 0.028729185461997986, "train_loss_llm": 0.0, "grad_norm": 0.08971423655748367, "global_step": 24312, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03198988363146782, "train_loss_bc": 0.03198988363146782, "train_loss_llm": 0.0, "grad_norm": 0.03240916505455971, "global_step": 24313, "epoch": 675, "lr": 0.001} +{"train_loss": 0.030817711725831032, "train_loss_bc": 0.030803345143795013, "train_loss_llm": 0.0014366209506988525, "grad_norm": 0.05308561772108078, "global_step": 24314, "epoch": 675, "lr": 0.001} +{"train_loss": 0.0306710172444582, "train_loss_bc": 0.0306710172444582, "train_loss_llm": 0.0, "grad_norm": 0.07067033648490906, "global_step": 24315, "epoch": 675, "lr": 0.001} +{"train_loss": 0.026469381526112556, "train_loss_bc": 0.026469381526112556, "train_loss_llm": 0.0, "grad_norm": 0.09510715305805206, "global_step": 24316, "epoch": 675, "lr": 0.001} +{"train_loss": 0.027389364317059517, "train_loss_bc": 0.027381183579564095, "train_loss_llm": 0.000818103551864624, "grad_norm": 0.029304448515176773, "global_step": 24317, "epoch": 675, "lr": 0.001} +{"train_loss": 0.030608775094151497, "train_loss_bc": 0.030583903193473816, "train_loss_llm": 0.0024872124195098877, "grad_norm": 0.038805942982435226, "global_step": 24318, "epoch": 675, "lr": 0.001} +{"train_loss": 0.02620730921626091, "train_loss_bc": 0.02620730921626091, "train_loss_llm": 0.0, "grad_norm": 0.05520293489098549, "global_step": 24319, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03070649318397045, "train_loss_bc": 0.030698265880346298, "train_loss_llm": 0.0008227229118347168, "grad_norm": 0.0624314621090889, "global_step": 24320, "epoch": 675, "lr": 0.001} +{"train_loss": 0.021020682528614998, "train_loss_bc": 0.021020682528614998, "train_loss_llm": 0.0, "grad_norm": 0.022750306874513626, "global_step": 24321, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03006727434694767, "train_loss_bc": 0.030055297538638115, "train_loss_llm": 0.0011976659297943115, "grad_norm": 0.043440647423267365, "global_step": 24322, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03445645421743393, "train_loss_bc": 0.03445645421743393, "train_loss_llm": 0.0, "grad_norm": 0.06586425006389618, "global_step": 24323, "epoch": 675, "lr": 0.001} +{"train_loss": 0.032634783536195755, "train_loss_bc": 0.032634783536195755, "train_loss_llm": 0.0, "grad_norm": 0.08226682990789413, "global_step": 24324, "epoch": 675, "lr": 0.001} +{"train_loss": 0.037978265434503555, "train_loss_bc": 0.037978265434503555, "train_loss_llm": 0.0, "grad_norm": 0.04276148974895477, "global_step": 24325, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03313148766756058, "train_loss_bc": 0.03313148766756058, "train_loss_llm": 0.0, "grad_norm": 0.06093209981918335, "global_step": 24326, "epoch": 675, "lr": 0.001} +{"train_loss": 0.034336019307374954, "train_loss_bc": 0.034336019307374954, "train_loss_llm": 0.0, "grad_norm": 0.07963048666715622, "global_step": 24327, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03147032856941223, "train_loss_bc": 0.03147032856941223, "train_loss_llm": 0.0, "grad_norm": 0.10419502854347229, "global_step": 24328, "epoch": 675, "lr": 0.001} +{"train_loss": 0.032421600073575974, "train_loss_bc": 0.032421600073575974, "train_loss_llm": 0.0, "grad_norm": 0.03307724744081497, "global_step": 24329, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03693282976746559, "train_loss_bc": 0.03693282976746559, "train_loss_llm": 0.0, "grad_norm": 0.06358475983142853, "global_step": 24330, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03178562596440315, "train_loss_bc": 0.03178562596440315, "train_loss_llm": 0.0, "grad_norm": 0.09208144247531891, "global_step": 24331, "epoch": 675, "lr": 0.001} +{"train_loss": 0.031206566840410233, "train_loss_bc": 0.031206566840410233, "train_loss_llm": 0.0, "grad_norm": 0.11505777388811111, "global_step": 24332, "epoch": 675, "lr": 0.001} +{"train_loss": 0.031133607029914856, "train_loss_bc": 0.031133607029914856, "train_loss_llm": 0.0, "grad_norm": 0.0325935035943985, "global_step": 24333, "epoch": 675, "lr": 0.001} +{"train_loss": 0.027046605944633484, "train_loss_bc": 0.027046605944633484, "train_loss_llm": 0.0, "grad_norm": 0.04699556902050972, "global_step": 24334, "epoch": 675, "lr": 0.001} +{"train_loss": 0.03189402839375867, "train_loss_bc": 0.042144834995269775, "train_loss_llm": 0.0, "grad_norm": 0.08434022963047028, "global_step": 24335, "epoch": 675, "lr": 0.001, "train/cumulative_reward": 45.844374412422006, "train/mean_score": 10.712090945717378, "train/success_rate": 0.3, "test/cumulative_reward": 48.98574062943023, "test/mean_score": 10.470643754741754, "test/success_rate": 0.28, "val_loss": 0.0866207554936409, "train_action_mse_error": 0.03515377268195152} +{"train_loss": 0.024596629664301872, "train_loss_bc": 0.024596629664301872, "train_loss_llm": 0.0, "grad_norm": 0.09781577438116074, "global_step": 24336, "epoch": 676, "lr": 0.001} +{"train_loss": 0.031377825886011124, "train_loss_bc": 0.031377825886011124, "train_loss_llm": 0.0, "grad_norm": 0.03046673722565174, "global_step": 24337, "epoch": 676, "lr": 0.001} +{"train_loss": 0.0333174467086792, "train_loss_bc": 0.0333174467086792, "train_loss_llm": 0.0, "grad_norm": 0.043159425258636475, "global_step": 24338, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03236069902777672, "train_loss_bc": 0.03235220909118652, "train_loss_llm": 0.0008489489555358887, "grad_norm": 0.057176291942596436, "global_step": 24339, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03175326809287071, "train_loss_bc": 0.03175126388669014, "train_loss_llm": 0.00020059943199157715, "grad_norm": 0.08140254765748978, "global_step": 24340, "epoch": 676, "lr": 0.001} +{"train_loss": 0.028229031711816788, "train_loss_bc": 0.028229031711816788, "train_loss_llm": 0.0, "grad_norm": 0.03311380743980408, "global_step": 24341, "epoch": 676, "lr": 0.001} +{"train_loss": 0.030753040686249733, "train_loss_bc": 0.030753040686249733, "train_loss_llm": 0.0, "grad_norm": 0.0604393444955349, "global_step": 24342, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03336931765079498, "train_loss_bc": 0.033342085778713226, "train_loss_llm": 0.0027230679988861084, "grad_norm": 0.08230758458375931, "global_step": 24343, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03125442937016487, "train_loss_bc": 0.03125442937016487, "train_loss_llm": 0.0, "grad_norm": 0.10841972380876541, "global_step": 24344, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03487888723611832, "train_loss_bc": 0.03487888723611832, "train_loss_llm": 0.0, "grad_norm": 0.037605348974466324, "global_step": 24345, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03516950458288193, "train_loss_bc": 0.03516950458288193, "train_loss_llm": 0.0, "grad_norm": 0.06856221705675125, "global_step": 24346, "epoch": 676, "lr": 0.001} +{"train_loss": 0.030499082058668137, "train_loss_bc": 0.030499082058668137, "train_loss_llm": 0.0, "grad_norm": 0.09109333902597427, "global_step": 24347, "epoch": 676, "lr": 0.001} +{"train_loss": 0.033837106078863144, "train_loss_bc": 0.033837106078863144, "train_loss_llm": 0.0, "grad_norm": 0.10826434940099716, "global_step": 24348, "epoch": 676, "lr": 0.001} +{"train_loss": 0.0336809977889061, "train_loss_bc": 0.0336809977889061, "train_loss_llm": 0.0, "grad_norm": 0.04388546198606491, "global_step": 24349, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03012070246040821, "train_loss_bc": 0.03012070246040821, "train_loss_llm": 0.0, "grad_norm": 0.06932485848665237, "global_step": 24350, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03249248117208481, "train_loss_bc": 0.03249248117208481, "train_loss_llm": 0.0, "grad_norm": 0.09007097780704498, "global_step": 24351, "epoch": 676, "lr": 0.001} +{"train_loss": 0.030177006497979164, "train_loss_bc": 0.030177006497979164, "train_loss_llm": 0.0, "grad_norm": 0.10954516381025314, "global_step": 24352, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03437073528766632, "train_loss_bc": 0.03437073528766632, "train_loss_llm": 0.0, "grad_norm": 0.03596531227231026, "global_step": 24353, "epoch": 676, "lr": 0.001} +{"train_loss": 0.028295272961258888, "train_loss_bc": 0.02829163148999214, "train_loss_llm": 0.0003641843795776367, "grad_norm": 0.04975374788045883, "global_step": 24354, "epoch": 676, "lr": 0.001} +{"train_loss": 0.031129024922847748, "train_loss_bc": 0.031129024922847748, "train_loss_llm": 0.0, "grad_norm": 0.06856698542833328, "global_step": 24355, "epoch": 676, "lr": 0.001} +{"train_loss": 0.029183203354477882, "train_loss_bc": 0.029181726276874542, "train_loss_llm": 0.00014778971672058105, "grad_norm": 0.0807836726307869, "global_step": 24356, "epoch": 676, "lr": 0.001} +{"train_loss": 0.036590807139873505, "train_loss_bc": 0.036590807139873505, "train_loss_llm": 0.0, "grad_norm": 0.02761935628950596, "global_step": 24357, "epoch": 676, "lr": 0.001} +{"train_loss": 0.02546870708465576, "train_loss_bc": 0.025436509400606155, "train_loss_llm": 0.003219813108444214, "grad_norm": 0.04275789484381676, "global_step": 24358, "epoch": 676, "lr": 0.001} +{"train_loss": 0.029671156778931618, "train_loss_bc": 0.029668157920241356, "train_loss_llm": 0.00029984116554260254, "grad_norm": 0.056217629462480545, "global_step": 24359, "epoch": 676, "lr": 0.001} +{"train_loss": 0.035088684409856796, "train_loss_bc": 0.035088684409856796, "train_loss_llm": 0.0, "grad_norm": 0.0664730817079544, "global_step": 24360, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03341686353087425, "train_loss_bc": 0.03341686353087425, "train_loss_llm": 0.0, "grad_norm": 0.031183550134301186, "global_step": 24361, "epoch": 676, "lr": 0.001} +{"train_loss": 0.026855502277612686, "train_loss_bc": 0.026855502277612686, "train_loss_llm": 0.0, "grad_norm": 0.04613681882619858, "global_step": 24362, "epoch": 676, "lr": 0.001} +{"train_loss": 0.031192319467663765, "train_loss_bc": 0.03119022585451603, "train_loss_llm": 0.00020939111709594727, "grad_norm": 0.07402614504098892, "global_step": 24363, "epoch": 676, "lr": 0.001} +{"train_loss": 0.04047401249408722, "train_loss_bc": 0.04047401249408722, "train_loss_llm": 0.0, "grad_norm": 0.09380374103784561, "global_step": 24364, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03045746125280857, "train_loss_bc": 0.03045746125280857, "train_loss_llm": 0.0, "grad_norm": 0.04334031790494919, "global_step": 24365, "epoch": 676, "lr": 0.001} +{"train_loss": 0.038334883749485016, "train_loss_bc": 0.038334883749485016, "train_loss_llm": 0.0, "grad_norm": 0.06604806333780289, "global_step": 24366, "epoch": 676, "lr": 0.001} +{"train_loss": 0.029272576794028282, "train_loss_bc": 0.029272576794028282, "train_loss_llm": 0.0, "grad_norm": 0.08048481494188309, "global_step": 24367, "epoch": 676, "lr": 0.001} +{"train_loss": 0.026494722813367844, "train_loss_bc": 0.026494722813367844, "train_loss_llm": 0.0, "grad_norm": 0.0863947942852974, "global_step": 24368, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03596469759941101, "train_loss_bc": 0.03596469759941101, "train_loss_llm": 0.0, "grad_norm": 0.05074489489197731, "global_step": 24369, "epoch": 676, "lr": 0.001} +{"train_loss": 0.029526857659220695, "train_loss_bc": 0.029517896473407745, "train_loss_llm": 0.000896066427230835, "grad_norm": 0.059917718172073364, "global_step": 24370, "epoch": 676, "lr": 0.001} +{"train_loss": 0.03171491048609217, "train_loss_bc": 0.032081831246614456, "train_loss_llm": 0.0, "grad_norm": 0.08710882067680359, "global_step": 24371, "epoch": 676, "lr": 0.001, "val_loss": 0.0872492641210556} +{"train_loss": 0.029208308085799217, "train_loss_bc": 0.029208308085799217, "train_loss_llm": 0.0, "grad_norm": 0.10286863148212433, "global_step": 24372, "epoch": 677, "lr": 0.001} +{"train_loss": 0.027173491194844246, "train_loss_bc": 0.027173491194844246, "train_loss_llm": 0.0, "grad_norm": 0.02457752265036106, "global_step": 24373, "epoch": 677, "lr": 0.001} +{"train_loss": 0.02718476578593254, "train_loss_bc": 0.02718476578593254, "train_loss_llm": 0.0, "grad_norm": 0.04199756309390068, "global_step": 24374, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03300408646464348, "train_loss_bc": 0.0329991839826107, "train_loss_llm": 0.000490337610244751, "grad_norm": 0.06495025753974915, "global_step": 24375, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03471015393733978, "train_loss_bc": 0.03471015393733978, "train_loss_llm": 0.0, "grad_norm": 0.08204133808612823, "global_step": 24376, "epoch": 677, "lr": 0.001} +{"train_loss": 0.029469527304172516, "train_loss_bc": 0.029436612501740456, "train_loss_llm": 0.0032914578914642334, "grad_norm": 0.036638762801885605, "global_step": 24377, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03196287900209427, "train_loss_bc": 0.03195897117257118, "train_loss_llm": 0.00039067864418029785, "grad_norm": 0.0601942203938961, "global_step": 24378, "epoch": 677, "lr": 0.001} +{"train_loss": 0.030404135584831238, "train_loss_bc": 0.030404135584831238, "train_loss_llm": 0.0, "grad_norm": 0.07838816940784454, "global_step": 24379, "epoch": 677, "lr": 0.001} +{"train_loss": 0.038447055965662, "train_loss_bc": 0.038447055965662, "train_loss_llm": 0.0, "grad_norm": 0.11061923950910568, "global_step": 24380, "epoch": 677, "lr": 0.001} +{"train_loss": 0.031501226127147675, "train_loss_bc": 0.031501226127147675, "train_loss_llm": 0.0, "grad_norm": 0.033255189657211304, "global_step": 24381, "epoch": 677, "lr": 0.001} +{"train_loss": 0.027393469586968422, "train_loss_bc": 0.027393469586968422, "train_loss_llm": 0.0, "grad_norm": 0.052942294627428055, "global_step": 24382, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03281614929437637, "train_loss_bc": 0.03281614929437637, "train_loss_llm": 0.0, "grad_norm": 0.07450483739376068, "global_step": 24383, "epoch": 677, "lr": 0.001} +{"train_loss": 0.025889812037348747, "train_loss_bc": 0.025889812037348747, "train_loss_llm": 0.0, "grad_norm": 0.0865025445818901, "global_step": 24384, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03456966578960419, "train_loss_bc": 0.03456966578960419, "train_loss_llm": 0.0, "grad_norm": 0.03929564356803894, "global_step": 24385, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03538699448108673, "train_loss_bc": 0.03538699448108673, "train_loss_llm": 0.0, "grad_norm": 0.05559413135051727, "global_step": 24386, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03265903890132904, "train_loss_bc": 0.03265903890132904, "train_loss_llm": 0.0, "grad_norm": 0.08147193491458893, "global_step": 24387, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03400726616382599, "train_loss_bc": 0.03400726616382599, "train_loss_llm": 0.0, "grad_norm": 0.09398305416107178, "global_step": 24388, "epoch": 677, "lr": 0.001} +{"train_loss": 0.029211197048425674, "train_loss_bc": 0.029206139966845512, "train_loss_llm": 0.0005057454109191895, "grad_norm": 0.03344980627298355, "global_step": 24389, "epoch": 677, "lr": 0.001} +{"train_loss": 0.029525846242904663, "train_loss_bc": 0.02952400967478752, "train_loss_llm": 0.00018361210823059082, "grad_norm": 0.04965226352214813, "global_step": 24390, "epoch": 677, "lr": 0.001} +{"train_loss": 0.030212383717298508, "train_loss_bc": 0.030212383717298508, "train_loss_llm": 0.0, "grad_norm": 0.07335042208433151, "global_step": 24391, "epoch": 677, "lr": 0.001} +{"train_loss": 0.036850880831480026, "train_loss_bc": 0.03683376684784889, "train_loss_llm": 0.0017113983631134033, "grad_norm": 0.09825548529624939, "global_step": 24392, "epoch": 677, "lr": 0.001} +{"train_loss": 0.027768658474087715, "train_loss_bc": 0.027764083817601204, "train_loss_llm": 0.00045755505561828613, "grad_norm": 0.03752993047237396, "global_step": 24393, "epoch": 677, "lr": 0.001} +{"train_loss": 0.027493709698319435, "train_loss_bc": 0.027478963136672974, "train_loss_llm": 0.0014746785163879395, "grad_norm": 0.06609632074832916, "global_step": 24394, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03497713804244995, "train_loss_bc": 0.03497713804244995, "train_loss_llm": 0.0, "grad_norm": 0.09839405119419098, "global_step": 24395, "epoch": 677, "lr": 0.001} +{"train_loss": 0.029604192823171616, "train_loss_bc": 0.029598549008369446, "train_loss_llm": 0.0005644261837005615, "grad_norm": 0.11890404671430588, "global_step": 24396, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03563987463712692, "train_loss_bc": 0.03563987463712692, "train_loss_llm": 0.0, "grad_norm": 0.03150354325771332, "global_step": 24397, "epoch": 677, "lr": 0.001} +{"train_loss": 0.030539467930793762, "train_loss_bc": 0.03052794747054577, "train_loss_llm": 0.00115203857421875, "grad_norm": 0.0613616481423378, "global_step": 24398, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03419668972492218, "train_loss_bc": 0.03419668972492218, "train_loss_llm": 0.0, "grad_norm": 0.07322248816490173, "global_step": 24399, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03528657555580139, "train_loss_bc": 0.03528370335698128, "train_loss_llm": 0.00028714537620544434, "grad_norm": 0.09737078845500946, "global_step": 24400, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03227400779724121, "train_loss_bc": 0.03227400779724121, "train_loss_llm": 0.0, "grad_norm": 0.027667462825775146, "global_step": 24401, "epoch": 677, "lr": 0.001} +{"train_loss": 0.035883828997612, "train_loss_bc": 0.035883828997612, "train_loss_llm": 0.0, "grad_norm": 0.043038275092840195, "global_step": 24402, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03137968108057976, "train_loss_bc": 0.03137968108057976, "train_loss_llm": 0.0, "grad_norm": 0.06392723321914673, "global_step": 24403, "epoch": 677, "lr": 0.001} +{"train_loss": 0.030429894104599953, "train_loss_bc": 0.030429894104599953, "train_loss_llm": 0.0, "grad_norm": 0.08366760611534119, "global_step": 24404, "epoch": 677, "lr": 0.001} +{"train_loss": 0.027804991230368614, "train_loss_bc": 0.027804991230368614, "train_loss_llm": 0.0, "grad_norm": 0.03064141422510147, "global_step": 24405, "epoch": 677, "lr": 0.001} +{"train_loss": 0.03379274159669876, "train_loss_bc": 0.03379274159669876, "train_loss_llm": 0.0, "grad_norm": 0.04613015055656433, "global_step": 24406, "epoch": 677, "lr": 0.001} +{"train_loss": 0.031750179102851286, "train_loss_bc": 0.03433331847190857, "train_loss_llm": 0.0013343393802642822, "grad_norm": 0.0652083158493042, "global_step": 24407, "epoch": 677, "lr": 0.001, "val_loss": 0.08699090033769608} +{"train_loss": 0.030683154240250587, "train_loss_bc": 0.030683154240250587, "train_loss_llm": 0.0, "grad_norm": 0.07862492650747299, "global_step": 24408, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03413509950041771, "train_loss_bc": 0.03413509950041771, "train_loss_llm": 0.0, "grad_norm": 0.02974223718047142, "global_step": 24409, "epoch": 678, "lr": 0.001} +{"train_loss": 0.029273690655827522, "train_loss_bc": 0.029268953949213028, "train_loss_llm": 0.0004737377166748047, "grad_norm": 0.03921202942728996, "global_step": 24410, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03603024408221245, "train_loss_bc": 0.03602832183241844, "train_loss_llm": 0.00019225478172302246, "grad_norm": 0.05297144129872322, "global_step": 24411, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03944332152605057, "train_loss_bc": 0.03944332152605057, "train_loss_llm": 0.0, "grad_norm": 0.07656851410865784, "global_step": 24412, "epoch": 678, "lr": 0.001} +{"train_loss": 0.026517212390899658, "train_loss_bc": 0.026517212390899658, "train_loss_llm": 0.0, "grad_norm": 0.024857735261321068, "global_step": 24413, "epoch": 678, "lr": 0.001} +{"train_loss": 0.030393749475479126, "train_loss_bc": 0.030393749475479126, "train_loss_llm": 0.0, "grad_norm": 0.05418921262025833, "global_step": 24414, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03241705149412155, "train_loss_bc": 0.03241705149412155, "train_loss_llm": 0.0, "grad_norm": 0.06501892954111099, "global_step": 24415, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03190546855330467, "train_loss_bc": 0.03190546855330467, "train_loss_llm": 0.0, "grad_norm": 0.08827938884496689, "global_step": 24416, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03568428382277489, "train_loss_bc": 0.03567785024642944, "train_loss_llm": 0.0006432533264160156, "grad_norm": 0.029508018866181374, "global_step": 24417, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03137427195906639, "train_loss_bc": 0.03137427195906639, "train_loss_llm": 0.0, "grad_norm": 0.04507071152329445, "global_step": 24418, "epoch": 678, "lr": 0.001} +{"train_loss": 0.033345066010951996, "train_loss_bc": 0.033345066010951996, "train_loss_llm": 0.0, "grad_norm": 0.05760518088936806, "global_step": 24419, "epoch": 678, "lr": 0.001} +{"train_loss": 0.02863074280321598, "train_loss_bc": 0.02863074280321598, "train_loss_llm": 0.0, "grad_norm": 0.07911131531000137, "global_step": 24420, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03322112187743187, "train_loss_bc": 0.03322112187743187, "train_loss_llm": 0.0, "grad_norm": 0.044328972697257996, "global_step": 24421, "epoch": 678, "lr": 0.001} +{"train_loss": 0.026133589446544647, "train_loss_bc": 0.026133589446544647, "train_loss_llm": 0.0, "grad_norm": 0.05913415551185608, "global_step": 24422, "epoch": 678, "lr": 0.001} +{"train_loss": 0.028477756306529045, "train_loss_bc": 0.02846274897456169, "train_loss_llm": 0.0015008151531219482, "grad_norm": 0.06716766208410263, "global_step": 24423, "epoch": 678, "lr": 0.001} +{"train_loss": 0.029816463589668274, "train_loss_bc": 0.029816463589668274, "train_loss_llm": 0.0, "grad_norm": 0.08210963755846024, "global_step": 24424, "epoch": 678, "lr": 0.001} +{"train_loss": 0.032099299132823944, "train_loss_bc": 0.032099299132823944, "train_loss_llm": 0.0, "grad_norm": 0.028367498889565468, "global_step": 24425, "epoch": 678, "lr": 0.001} +{"train_loss": 0.024673601612448692, "train_loss_bc": 0.024673601612448692, "train_loss_llm": 0.0, "grad_norm": 0.040683563798666, "global_step": 24426, "epoch": 678, "lr": 0.001} +{"train_loss": 0.029678113758563995, "train_loss_bc": 0.029678113758563995, "train_loss_llm": 0.0, "grad_norm": 0.05466838926076889, "global_step": 24427, "epoch": 678, "lr": 0.001} +{"train_loss": 0.029985325410962105, "train_loss_bc": 0.029985325410962105, "train_loss_llm": 0.0, "grad_norm": 0.06393878906965256, "global_step": 24428, "epoch": 678, "lr": 0.001} +{"train_loss": 0.033800262957811356, "train_loss_bc": 0.033800262957811356, "train_loss_llm": 0.0, "grad_norm": 0.029591776430606842, "global_step": 24429, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03567345067858696, "train_loss_bc": 0.03567345067858696, "train_loss_llm": 0.0, "grad_norm": 0.04203283041715622, "global_step": 24430, "epoch": 678, "lr": 0.001} +{"train_loss": 0.0287698395550251, "train_loss_bc": 0.0287698395550251, "train_loss_llm": 0.0, "grad_norm": 0.054115764796733856, "global_step": 24431, "epoch": 678, "lr": 0.001} +{"train_loss": 0.027625232934951782, "train_loss_bc": 0.027625232934951782, "train_loss_llm": 0.0, "grad_norm": 0.06734619289636612, "global_step": 24432, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03267786279320717, "train_loss_bc": 0.03267041966319084, "train_loss_llm": 0.0007444322109222412, "grad_norm": 0.03845876455307007, "global_step": 24433, "epoch": 678, "lr": 0.001} +{"train_loss": 0.0373711958527565, "train_loss_bc": 0.0373711958527565, "train_loss_llm": 0.0, "grad_norm": 0.06118415296077728, "global_step": 24434, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03672418370842934, "train_loss_bc": 0.03672418370842934, "train_loss_llm": 0.0, "grad_norm": 0.07814732193946838, "global_step": 24435, "epoch": 678, "lr": 0.001} +{"train_loss": 0.023250751197338104, "train_loss_bc": 0.023250751197338104, "train_loss_llm": 0.0, "grad_norm": 0.08967979997396469, "global_step": 24436, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03733735904097557, "train_loss_bc": 0.03733391314744949, "train_loss_llm": 0.0003446042537689209, "grad_norm": 0.039112187922000885, "global_step": 24437, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03674710541963577, "train_loss_bc": 0.03671782836318016, "train_loss_llm": 0.0029277503490448, "grad_norm": 0.05147649347782135, "global_step": 24438, "epoch": 678, "lr": 0.001} +{"train_loss": 0.0296489167958498, "train_loss_bc": 0.0296489167958498, "train_loss_llm": 0.0, "grad_norm": 0.062071871012449265, "global_step": 24439, "epoch": 678, "lr": 0.001} +{"train_loss": 0.034345563501119614, "train_loss_bc": 0.034345563501119614, "train_loss_llm": 0.0, "grad_norm": 0.08046115189790726, "global_step": 24440, "epoch": 678, "lr": 0.001} +{"train_loss": 0.02947358787059784, "train_loss_bc": 0.02947358787059784, "train_loss_llm": 0.0, "grad_norm": 0.027761185541749, "global_step": 24441, "epoch": 678, "lr": 0.001} +{"train_loss": 0.03039555251598358, "train_loss_bc": 0.03039555251598358, "train_loss_llm": 0.0, "grad_norm": 0.04605782404541969, "global_step": 24442, "epoch": 678, "lr": 0.001} +{"train_loss": 0.031424868696679674, "train_loss_bc": 0.023535780608654022, "train_loss_llm": 0.0, "grad_norm": 0.0672709196805954, "global_step": 24443, "epoch": 678, "lr": 0.001, "val_loss": 0.08753011375665665} +{"train_loss": 0.03359493240714073, "train_loss_bc": 0.03358917310833931, "train_loss_llm": 0.0005759596824645996, "grad_norm": 0.08079113066196442, "global_step": 24444, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03211395442485809, "train_loss_bc": 0.03211395442485809, "train_loss_llm": 0.0, "grad_norm": 0.03410130739212036, "global_step": 24445, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03211560100317001, "train_loss_bc": 0.032109055668115616, "train_loss_llm": 0.0006545186042785645, "grad_norm": 0.050169263035058975, "global_step": 24446, "epoch": 679, "lr": 0.001} +{"train_loss": 0.02903902344405651, "train_loss_bc": 0.02903902344405651, "train_loss_llm": 0.0, "grad_norm": 0.06740231812000275, "global_step": 24447, "epoch": 679, "lr": 0.001} +{"train_loss": 0.02622899040579796, "train_loss_bc": 0.02622899040579796, "train_loss_llm": 0.0, "grad_norm": 0.07871399819850922, "global_step": 24448, "epoch": 679, "lr": 0.001} +{"train_loss": 0.031970515847206116, "train_loss_bc": 0.031970515847206116, "train_loss_llm": 0.0, "grad_norm": 0.024951426312327385, "global_step": 24449, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03504817560315132, "train_loss_bc": 0.03503122180700302, "train_loss_llm": 0.0016953349113464355, "grad_norm": 0.043486762791872025, "global_step": 24450, "epoch": 679, "lr": 0.001} +{"train_loss": 0.023527266457676888, "train_loss_bc": 0.023527266457676888, "train_loss_llm": 0.0, "grad_norm": 0.0517861433327198, "global_step": 24451, "epoch": 679, "lr": 0.001} +{"train_loss": 0.033723991364240646, "train_loss_bc": 0.033723991364240646, "train_loss_llm": 0.0, "grad_norm": 0.07164362818002701, "global_step": 24452, "epoch": 679, "lr": 0.001} +{"train_loss": 0.030461177229881287, "train_loss_bc": 0.030461177229881287, "train_loss_llm": 0.0, "grad_norm": 0.03149512782692909, "global_step": 24453, "epoch": 679, "lr": 0.001} +{"train_loss": 0.02625609189271927, "train_loss_bc": 0.02625609189271927, "train_loss_llm": 0.0, "grad_norm": 0.04437757283449173, "global_step": 24454, "epoch": 679, "lr": 0.001} +{"train_loss": 0.026097409427165985, "train_loss_bc": 0.026097409427165985, "train_loss_llm": 0.0, "grad_norm": 0.06214478239417076, "global_step": 24455, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03494865819811821, "train_loss_bc": 0.034944891929626465, "train_loss_llm": 0.00037676095962524414, "grad_norm": 0.07218442857265472, "global_step": 24456, "epoch": 679, "lr": 0.001} +{"train_loss": 0.032075244933366776, "train_loss_bc": 0.032075244933366776, "train_loss_llm": 0.0, "grad_norm": 0.0304724108427763, "global_step": 24457, "epoch": 679, "lr": 0.001} +{"train_loss": 0.02960696630179882, "train_loss_bc": 0.029590262100100517, "train_loss_llm": 0.00167044997215271, "grad_norm": 0.04388781264424324, "global_step": 24458, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03466976806521416, "train_loss_bc": 0.034666698426008224, "train_loss_llm": 0.00030681490898132324, "grad_norm": 0.052710674703121185, "global_step": 24459, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03194112330675125, "train_loss_bc": 0.03194112330675125, "train_loss_llm": 0.0, "grad_norm": 0.05988354608416557, "global_step": 24460, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03130302578210831, "train_loss_bc": 0.03130302578210831, "train_loss_llm": 0.0, "grad_norm": 0.03242220729589462, "global_step": 24461, "epoch": 679, "lr": 0.001} +{"train_loss": 0.02885465696454048, "train_loss_bc": 0.02885465696454048, "train_loss_llm": 0.0, "grad_norm": 0.051419466733932495, "global_step": 24462, "epoch": 679, "lr": 0.001} +{"train_loss": 0.030960168689489365, "train_loss_bc": 0.030960168689489365, "train_loss_llm": 0.0, "grad_norm": 0.06315509974956512, "global_step": 24463, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03433700278401375, "train_loss_bc": 0.034334320574998856, "train_loss_llm": 0.0002682507038116455, "grad_norm": 0.07737784832715988, "global_step": 24464, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03334604948759079, "train_loss_bc": 0.03334604948759079, "train_loss_llm": 0.0, "grad_norm": 0.034240178763866425, "global_step": 24465, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03369482234120369, "train_loss_bc": 0.03369482234120369, "train_loss_llm": 0.0, "grad_norm": 0.05392758175730705, "global_step": 24466, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03343413397669792, "train_loss_bc": 0.03343413397669792, "train_loss_llm": 0.0, "grad_norm": 0.07774271070957184, "global_step": 24467, "epoch": 679, "lr": 0.001} +{"train_loss": 0.030932145193219185, "train_loss_bc": 0.030932145193219185, "train_loss_llm": 0.0, "grad_norm": 0.09743335843086243, "global_step": 24468, "epoch": 679, "lr": 0.001} +{"train_loss": 0.029799191281199455, "train_loss_bc": 0.029799191281199455, "train_loss_llm": 0.0, "grad_norm": 0.027963273227214813, "global_step": 24469, "epoch": 679, "lr": 0.001} +{"train_loss": 0.027901126071810722, "train_loss_bc": 0.027901126071810722, "train_loss_llm": 0.0, "grad_norm": 0.04273110628128052, "global_step": 24470, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03138263151049614, "train_loss_bc": 0.03138263151049614, "train_loss_llm": 0.0, "grad_norm": 0.06300576031208038, "global_step": 24471, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03491462394595146, "train_loss_bc": 0.03491462394595146, "train_loss_llm": 0.0, "grad_norm": 0.08756718784570694, "global_step": 24472, "epoch": 679, "lr": 0.001} +{"train_loss": 0.026813553646206856, "train_loss_bc": 0.026813553646206856, "train_loss_llm": 0.0, "grad_norm": 0.03048773482441902, "global_step": 24473, "epoch": 679, "lr": 0.001} +{"train_loss": 0.040259938687086105, "train_loss_bc": 0.040259938687086105, "train_loss_llm": 0.0, "grad_norm": 0.05692040175199509, "global_step": 24474, "epoch": 679, "lr": 0.001} +{"train_loss": 0.036993954330682755, "train_loss_bc": 0.036993954330682755, "train_loss_llm": 0.0, "grad_norm": 0.07773784548044205, "global_step": 24475, "epoch": 679, "lr": 0.001} +{"train_loss": 0.029913602396845818, "train_loss_bc": 0.029913602396845818, "train_loss_llm": 0.0, "grad_norm": 0.09126856923103333, "global_step": 24476, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03615601733326912, "train_loss_bc": 0.03615601733326912, "train_loss_llm": 0.0, "grad_norm": 0.03568549454212189, "global_step": 24477, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03475179523229599, "train_loss_bc": 0.03475051000714302, "train_loss_llm": 0.0001283586025238037, "grad_norm": 0.06055815890431404, "global_step": 24478, "epoch": 679, "lr": 0.001} +{"train_loss": 0.03154937415901157, "train_loss_bc": 0.02661013975739479, "train_loss_llm": 0.0, "grad_norm": 0.07072145491838455, "global_step": 24479, "epoch": 679, "lr": 0.001, "val_loss": 0.08632639050483704} +{"train_loss": 0.0226135291159153, "train_loss_bc": 0.0226135291159153, "train_loss_llm": 0.0, "grad_norm": 0.07692208141088486, "global_step": 24480, "epoch": 680, "lr": 0.001} +{"train_loss": 0.033159151673316956, "train_loss_bc": 0.033159151673316956, "train_loss_llm": 0.0, "grad_norm": 0.036224037408828735, "global_step": 24481, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03343086689710617, "train_loss_bc": 0.03343086689710617, "train_loss_llm": 0.0, "grad_norm": 0.051517803221940994, "global_step": 24482, "epoch": 680, "lr": 0.001} +{"train_loss": 0.024656614288687706, "train_loss_bc": 0.024656614288687706, "train_loss_llm": 0.0, "grad_norm": 0.06265749037265778, "global_step": 24483, "epoch": 680, "lr": 0.001} +{"train_loss": 0.029907358810305595, "train_loss_bc": 0.029907358810305595, "train_loss_llm": 0.0, "grad_norm": 0.07716796547174454, "global_step": 24484, "epoch": 680, "lr": 0.001} +{"train_loss": 0.027690082788467407, "train_loss_bc": 0.027690082788467407, "train_loss_llm": 0.0, "grad_norm": 0.026384513825178146, "global_step": 24485, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03172491118311882, "train_loss_bc": 0.03172491118311882, "train_loss_llm": 0.0, "grad_norm": 0.04325525090098381, "global_step": 24486, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03250570222735405, "train_loss_bc": 0.03250570222735405, "train_loss_llm": 0.0, "grad_norm": 0.06546939164400101, "global_step": 24487, "epoch": 680, "lr": 0.001} +{"train_loss": 0.030704742297530174, "train_loss_bc": 0.030675603076815605, "train_loss_llm": 0.002913922071456909, "grad_norm": 0.08137935400009155, "global_step": 24488, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03072730265557766, "train_loss_bc": 0.030713265761733055, "train_loss_llm": 0.0014036893844604492, "grad_norm": 0.03340423107147217, "global_step": 24489, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03599894419312477, "train_loss_bc": 0.03599879518151283, "train_loss_llm": 1.5050172805786133e-05, "grad_norm": 0.052985843271017075, "global_step": 24490, "epoch": 680, "lr": 0.001} +{"train_loss": 0.02861127257347107, "train_loss_bc": 0.02861127257347107, "train_loss_llm": 0.0, "grad_norm": 0.06178051978349686, "global_step": 24491, "epoch": 680, "lr": 0.001} +{"train_loss": 0.033002160489559174, "train_loss_bc": 0.033002160489559174, "train_loss_llm": 0.0, "grad_norm": 0.08861920982599258, "global_step": 24492, "epoch": 680, "lr": 0.001} +{"train_loss": 0.032708846032619476, "train_loss_bc": 0.03268548473715782, "train_loss_llm": 0.002336174249649048, "grad_norm": 0.029444904997944832, "global_step": 24493, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03408528491854668, "train_loss_bc": 0.03408528491854668, "train_loss_llm": 0.0, "grad_norm": 0.041889239102602005, "global_step": 24494, "epoch": 680, "lr": 0.001} +{"train_loss": 0.026753025129437447, "train_loss_bc": 0.026753025129437447, "train_loss_llm": 0.0, "grad_norm": 0.05626341700553894, "global_step": 24495, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03581008315086365, "train_loss_bc": 0.03581008315086365, "train_loss_llm": 0.0, "grad_norm": 0.0763130933046341, "global_step": 24496, "epoch": 680, "lr": 0.001} +{"train_loss": 0.032551705837249756, "train_loss_bc": 0.032551705837249756, "train_loss_llm": 0.0, "grad_norm": 0.030444933101534843, "global_step": 24497, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03315101936459541, "train_loss_bc": 0.03313448652625084, "train_loss_llm": 0.0016532540321350098, "grad_norm": 0.04191018268465996, "global_step": 24498, "epoch": 680, "lr": 0.001} +{"train_loss": 0.04142799600958824, "train_loss_bc": 0.04142799600958824, "train_loss_llm": 0.0, "grad_norm": 0.05949897691607475, "global_step": 24499, "epoch": 680, "lr": 0.001} +{"train_loss": 0.025260677561163902, "train_loss_bc": 0.025260677561163902, "train_loss_llm": 0.0, "grad_norm": 0.07237672805786133, "global_step": 24500, "epoch": 680, "lr": 0.001} +{"train_loss": 0.035271987318992615, "train_loss_bc": 0.035271987318992615, "train_loss_llm": 0.0, "grad_norm": 0.029955130070447922, "global_step": 24501, "epoch": 680, "lr": 0.001} +{"train_loss": 0.034215379506349564, "train_loss_bc": 0.03421412408351898, "train_loss_llm": 0.00012546777725219727, "grad_norm": 0.0482916533946991, "global_step": 24502, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03170058876276016, "train_loss_bc": 0.03170058876276016, "train_loss_llm": 0.0, "grad_norm": 0.06178349256515503, "global_step": 24503, "epoch": 680, "lr": 0.001} +{"train_loss": 0.027496369555592537, "train_loss_bc": 0.027496369555592537, "train_loss_llm": 0.0, "grad_norm": 0.08002301305532455, "global_step": 24504, "epoch": 680, "lr": 0.001} +{"train_loss": 0.031348444521427155, "train_loss_bc": 0.031348444521427155, "train_loss_llm": 0.0, "grad_norm": 0.029465630650520325, "global_step": 24505, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03353080898523331, "train_loss_bc": 0.03353080898523331, "train_loss_llm": 0.0, "grad_norm": 0.044556692242622375, "global_step": 24506, "epoch": 680, "lr": 0.001} +{"train_loss": 0.040274478495121, "train_loss_bc": 0.040274478495121, "train_loss_llm": 0.0, "grad_norm": 0.07102487236261368, "global_step": 24507, "epoch": 680, "lr": 0.001} +{"train_loss": 0.031265441328287125, "train_loss_bc": 0.031244631856679916, "train_loss_llm": 0.002081066370010376, "grad_norm": 0.08477019518613815, "global_step": 24508, "epoch": 680, "lr": 0.001} +{"train_loss": 0.031742535531520844, "train_loss_bc": 0.031741250306367874, "train_loss_llm": 0.0001284480094909668, "grad_norm": 0.034736327826976776, "global_step": 24509, "epoch": 680, "lr": 0.001} +{"train_loss": 0.02825143001973629, "train_loss_bc": 0.02825143001973629, "train_loss_llm": 0.0, "grad_norm": 0.054937127977609634, "global_step": 24510, "epoch": 680, "lr": 0.001} +{"train_loss": 0.024551400914788246, "train_loss_bc": 0.024551400914788246, "train_loss_llm": 0.0, "grad_norm": 0.06169520318508148, "global_step": 24511, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03419999033212662, "train_loss_bc": 0.03419999033212662, "train_loss_llm": 0.0, "grad_norm": 0.08797693252563477, "global_step": 24512, "epoch": 680, "lr": 0.001} +{"train_loss": 0.028836330398917198, "train_loss_bc": 0.028836330398917198, "train_loss_llm": 0.0, "grad_norm": 0.029698876664042473, "global_step": 24513, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03423868864774704, "train_loss_bc": 0.03423868864774704, "train_loss_llm": 0.0, "grad_norm": 0.048771876841783524, "global_step": 24514, "epoch": 680, "lr": 0.001} +{"train_loss": 0.03155128026588096, "train_loss_bc": 0.03243599832057953, "train_loss_llm": 0.0004938840866088867, "grad_norm": 0.06393341720104218, "global_step": 24515, "epoch": 680, "lr": 0.001, "train/cumulative_reward": 25.083013595920193, "train/mean_score": 4.497063758766768, "train/success_rate": 0.1, "test/cumulative_reward": 40.321998718508354, "test/mean_score": 9.855071037792468, "test/success_rate": 0.26, "val_loss": 0.0887848436832428, "train_action_mse_error": 0.03456512466073036} +{"train_loss": 0.03153480589389801, "train_loss_bc": 0.03153480589389801, "train_loss_llm": 0.0, "grad_norm": 0.07130316644906998, "global_step": 24516, "epoch": 681, "lr": 0.001} +{"train_loss": 0.026172874495387077, "train_loss_bc": 0.026172874495387077, "train_loss_llm": 0.0, "grad_norm": 0.027085652574896812, "global_step": 24517, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03203095495700836, "train_loss_bc": 0.032017529010772705, "train_loss_llm": 0.0013424456119537354, "grad_norm": 0.049105312675237656, "global_step": 24518, "epoch": 681, "lr": 0.001} +{"train_loss": 0.027455920353531837, "train_loss_bc": 0.027452923357486725, "train_loss_llm": 0.00029969215393066406, "grad_norm": 0.06342744082212448, "global_step": 24519, "epoch": 681, "lr": 0.001} +{"train_loss": 0.031117718666791916, "train_loss_bc": 0.031117718666791916, "train_loss_llm": 0.0, "grad_norm": 0.06957287341356277, "global_step": 24520, "epoch": 681, "lr": 0.001} +{"train_loss": 0.028750009834766388, "train_loss_bc": 0.028750009834766388, "train_loss_llm": 0.0, "grad_norm": 0.028688890859484673, "global_step": 24521, "epoch": 681, "lr": 0.001} +{"train_loss": 0.030062565580010414, "train_loss_bc": 0.030062565580010414, "train_loss_llm": 0.0, "grad_norm": 0.04956984147429466, "global_step": 24522, "epoch": 681, "lr": 0.001} +{"train_loss": 0.032903097569942474, "train_loss_bc": 0.032899532467126846, "train_loss_llm": 0.0003566741943359375, "grad_norm": 0.06997073441743851, "global_step": 24523, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03148706257343292, "train_loss_bc": 0.03146694228053093, "train_loss_llm": 0.0020121335983276367, "grad_norm": 0.0884314626455307, "global_step": 24524, "epoch": 681, "lr": 0.001} +{"train_loss": 0.028237445279955864, "train_loss_bc": 0.028237445279955864, "train_loss_llm": 0.0, "grad_norm": 0.027577439323067665, "global_step": 24525, "epoch": 681, "lr": 0.001} +{"train_loss": 0.037045616656541824, "train_loss_bc": 0.037045616656541824, "train_loss_llm": 0.0, "grad_norm": 0.0522458516061306, "global_step": 24526, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03219791129231453, "train_loss_bc": 0.0321771577000618, "train_loss_llm": 0.002075403928756714, "grad_norm": 0.06825440376996994, "global_step": 24527, "epoch": 681, "lr": 0.001} +{"train_loss": 0.030123364180326462, "train_loss_bc": 0.03012307919561863, "train_loss_llm": 2.8520822525024414e-05, "grad_norm": 0.08461117744445801, "global_step": 24528, "epoch": 681, "lr": 0.001} +{"train_loss": 0.029604479670524597, "train_loss_bc": 0.029604479670524597, "train_loss_llm": 0.0, "grad_norm": 0.030907373875379562, "global_step": 24529, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03173667937517166, "train_loss_bc": 0.03173667937517166, "train_loss_llm": 0.0, "grad_norm": 0.04432544857263565, "global_step": 24530, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03574066236615181, "train_loss_bc": 0.03574066236615181, "train_loss_llm": 0.0, "grad_norm": 0.07243615388870239, "global_step": 24531, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03145400807261467, "train_loss_bc": 0.03144337981939316, "train_loss_llm": 0.0010628700256347656, "grad_norm": 0.09291823953390121, "global_step": 24532, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03542838245630264, "train_loss_bc": 0.03542838245630264, "train_loss_llm": 0.0, "grad_norm": 0.037145745009183884, "global_step": 24533, "epoch": 681, "lr": 0.001} +{"train_loss": 0.031729210168123245, "train_loss_bc": 0.031729210168123245, "train_loss_llm": 0.0, "grad_norm": 0.05124744027853012, "global_step": 24534, "epoch": 681, "lr": 0.001} +{"train_loss": 0.029095128178596497, "train_loss_bc": 0.029095128178596497, "train_loss_llm": 0.0, "grad_norm": 0.06586721539497375, "global_step": 24535, "epoch": 681, "lr": 0.001} +{"train_loss": 0.030341248959302902, "train_loss_bc": 0.030341248959302902, "train_loss_llm": 0.0, "grad_norm": 0.08877776563167572, "global_step": 24536, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03591211885213852, "train_loss_bc": 0.03591033071279526, "train_loss_llm": 0.00017893314361572266, "grad_norm": 0.03253631293773651, "global_step": 24537, "epoch": 681, "lr": 0.001} +{"train_loss": 0.02883446402847767, "train_loss_bc": 0.02883446402847767, "train_loss_llm": 0.0, "grad_norm": 0.046588603407144547, "global_step": 24538, "epoch": 681, "lr": 0.001} +{"train_loss": 0.04159482568502426, "train_loss_bc": 0.04159482568502426, "train_loss_llm": 0.0, "grad_norm": 0.07610922306776047, "global_step": 24539, "epoch": 681, "lr": 0.001} +{"train_loss": 0.027573836967349052, "train_loss_bc": 0.027573836967349052, "train_loss_llm": 0.0, "grad_norm": 0.08559152483940125, "global_step": 24540, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03555035591125488, "train_loss_bc": 0.03555035591125488, "train_loss_llm": 0.0, "grad_norm": 0.03796425089240074, "global_step": 24541, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03364719823002815, "train_loss_bc": 0.03364719823002815, "train_loss_llm": 0.0, "grad_norm": 0.054534584283828735, "global_step": 24542, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03233421966433525, "train_loss_bc": 0.03233421966433525, "train_loss_llm": 0.0, "grad_norm": 0.07559584826231003, "global_step": 24543, "epoch": 681, "lr": 0.001} +{"train_loss": 0.031466685235500336, "train_loss_bc": 0.031466685235500336, "train_loss_llm": 0.0, "grad_norm": 0.09875413775444031, "global_step": 24544, "epoch": 681, "lr": 0.001} +{"train_loss": 0.029061313718557358, "train_loss_bc": 0.029058970510959625, "train_loss_llm": 0.00023424625396728516, "grad_norm": 0.03408747538924217, "global_step": 24545, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03006136044859886, "train_loss_bc": 0.03005794808268547, "train_loss_llm": 0.00034114718437194824, "grad_norm": 0.055188070982694626, "global_step": 24546, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03238421678543091, "train_loss_bc": 0.03238421678543091, "train_loss_llm": 0.0, "grad_norm": 0.07586638629436493, "global_step": 24547, "epoch": 681, "lr": 0.001} +{"train_loss": 0.028647398576140404, "train_loss_bc": 0.028647398576140404, "train_loss_llm": 0.0, "grad_norm": 0.0888984352350235, "global_step": 24548, "epoch": 681, "lr": 0.001} +{"train_loss": 0.03659186139702797, "train_loss_bc": 0.03657858818769455, "train_loss_llm": 0.0013274550437927246, "grad_norm": 0.04640301689505577, "global_step": 24549, "epoch": 681, "lr": 0.001} +{"train_loss": 0.030613355338573456, "train_loss_bc": 0.030613355338573456, "train_loss_llm": 0.0, "grad_norm": 0.0718802884221077, "global_step": 24550, "epoch": 681, "lr": 0.001} +{"train_loss": 0.031699929748558335, "train_loss_bc": 0.032675113528966904, "train_loss_llm": 0.0, "grad_norm": 0.09494639188051224, "global_step": 24551, "epoch": 681, "lr": 0.001, "val_loss": 0.08794175088405609} +{"train_loss": 0.0319414921104908, "train_loss_bc": 0.03194147348403931, "train_loss_llm": 1.6987323760986328e-06, "grad_norm": 0.11475814878940582, "global_step": 24552, "epoch": 682, "lr": 0.001} +{"train_loss": 0.0325041264295578, "train_loss_bc": 0.0325041264295578, "train_loss_llm": 0.0, "grad_norm": 0.028507430106401443, "global_step": 24553, "epoch": 682, "lr": 0.001} +{"train_loss": 0.029097488150000572, "train_loss_bc": 0.029097488150000572, "train_loss_llm": 0.0, "grad_norm": 0.045066867023706436, "global_step": 24554, "epoch": 682, "lr": 0.001} +{"train_loss": 0.034428589046001434, "train_loss_bc": 0.0343945249915123, "train_loss_llm": 0.003406226634979248, "grad_norm": 0.0649527907371521, "global_step": 24555, "epoch": 682, "lr": 0.001} +{"train_loss": 0.021262213587760925, "train_loss_bc": 0.021262213587760925, "train_loss_llm": 0.0, "grad_norm": 0.07645716518163681, "global_step": 24556, "epoch": 682, "lr": 0.001} +{"train_loss": 0.034056372940540314, "train_loss_bc": 0.034056372940540314, "train_loss_llm": 0.0, "grad_norm": 0.0367039255797863, "global_step": 24557, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03286292403936386, "train_loss_bc": 0.03286292403936386, "train_loss_llm": 0.0, "grad_norm": 0.05688800662755966, "global_step": 24558, "epoch": 682, "lr": 0.001} +{"train_loss": 0.034914590418338776, "train_loss_bc": 0.034914590418338776, "train_loss_llm": 0.0, "grad_norm": 0.07240287959575653, "global_step": 24559, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03681763634085655, "train_loss_bc": 0.03681763634085655, "train_loss_llm": 0.0, "grad_norm": 0.10610797256231308, "global_step": 24560, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03220849856734276, "train_loss_bc": 0.03220849856734276, "train_loss_llm": 0.0, "grad_norm": 0.028440747410058975, "global_step": 24561, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03455038368701935, "train_loss_bc": 0.03455038368701935, "train_loss_llm": 0.0, "grad_norm": 0.03430793806910515, "global_step": 24562, "epoch": 682, "lr": 0.001} +{"train_loss": 0.030300969257950783, "train_loss_bc": 0.030300969257950783, "train_loss_llm": 0.0, "grad_norm": 0.05046474561095238, "global_step": 24563, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03513573110103607, "train_loss_bc": 0.03513573110103607, "train_loss_llm": 0.0, "grad_norm": 0.0736505538225174, "global_step": 24564, "epoch": 682, "lr": 0.001} +{"train_loss": 0.032606955617666245, "train_loss_bc": 0.032606955617666245, "train_loss_llm": 0.0, "grad_norm": 0.030705688521265984, "global_step": 24565, "epoch": 682, "lr": 0.001} +{"train_loss": 0.030912458896636963, "train_loss_bc": 0.030901610851287842, "train_loss_llm": 0.001084834337234497, "grad_norm": 0.05400872603058815, "global_step": 24566, "epoch": 682, "lr": 0.001} +{"train_loss": 0.030339963734149933, "train_loss_bc": 0.030324703082442284, "train_loss_llm": 0.0015259981155395508, "grad_norm": 0.06995895504951477, "global_step": 24567, "epoch": 682, "lr": 0.001} +{"train_loss": 0.030952217057347298, "train_loss_bc": 0.030952217057347298, "train_loss_llm": 0.0, "grad_norm": 0.0725889727473259, "global_step": 24568, "epoch": 682, "lr": 0.001} +{"train_loss": 0.030650708824396133, "train_loss_bc": 0.030650708824396133, "train_loss_llm": 0.0, "grad_norm": 0.030114151537418365, "global_step": 24569, "epoch": 682, "lr": 0.001} +{"train_loss": 0.0300757996737957, "train_loss_bc": 0.0300757996737957, "train_loss_llm": 0.0, "grad_norm": 0.05202257260680199, "global_step": 24570, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03128383308649063, "train_loss_bc": 0.03128383308649063, "train_loss_llm": 0.0, "grad_norm": 0.07572989910840988, "global_step": 24571, "epoch": 682, "lr": 0.001} +{"train_loss": 0.0280628465116024, "train_loss_bc": 0.0280628465116024, "train_loss_llm": 0.0, "grad_norm": 0.08689473569393158, "global_step": 24572, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03249479830265045, "train_loss_bc": 0.03249479830265045, "train_loss_llm": 0.0, "grad_norm": 0.038224585354328156, "global_step": 24573, "epoch": 682, "lr": 0.001} +{"train_loss": 0.030235473066568375, "train_loss_bc": 0.030235473066568375, "train_loss_llm": 0.0, "grad_norm": 0.06394695490598679, "global_step": 24574, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03055022656917572, "train_loss_bc": 0.030548226088285446, "train_loss_llm": 0.00020003318786621094, "grad_norm": 0.08724575489759445, "global_step": 24575, "epoch": 682, "lr": 0.001} +{"train_loss": 0.024733969941735268, "train_loss_bc": 0.02471587434411049, "train_loss_llm": 0.0018095076084136963, "grad_norm": 0.10247524827718735, "global_step": 24576, "epoch": 682, "lr": 0.001} +{"train_loss": 0.028722723945975304, "train_loss_bc": 0.028722723945975304, "train_loss_llm": 0.0, "grad_norm": 0.02602097950875759, "global_step": 24577, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03728942945599556, "train_loss_bc": 0.03728942945599556, "train_loss_llm": 0.0, "grad_norm": 0.04991047456860542, "global_step": 24578, "epoch": 682, "lr": 0.001} +{"train_loss": 0.034925613552331924, "train_loss_bc": 0.034925613552331924, "train_loss_llm": 0.0, "grad_norm": 0.06247221305966377, "global_step": 24579, "epoch": 682, "lr": 0.001} +{"train_loss": 0.02748883329331875, "train_loss_bc": 0.02748883329331875, "train_loss_llm": 0.0, "grad_norm": 0.08287426084280014, "global_step": 24580, "epoch": 682, "lr": 0.001} +{"train_loss": 0.02896416187286377, "train_loss_bc": 0.028962450101971626, "train_loss_llm": 0.00017115473747253418, "grad_norm": 0.028114372864365578, "global_step": 24581, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03003925271332264, "train_loss_bc": 0.03003925271332264, "train_loss_llm": 0.0, "grad_norm": 0.04473348334431648, "global_step": 24582, "epoch": 682, "lr": 0.001} +{"train_loss": 0.027036620303988457, "train_loss_bc": 0.027036620303988457, "train_loss_llm": 0.0, "grad_norm": 0.05590198189020157, "global_step": 24583, "epoch": 682, "lr": 0.001} +{"train_loss": 0.031161362305283546, "train_loss_bc": 0.031161362305283546, "train_loss_llm": 0.0, "grad_norm": 0.06608922779560089, "global_step": 24584, "epoch": 682, "lr": 0.001} +{"train_loss": 0.042816706001758575, "train_loss_bc": 0.042816706001758575, "train_loss_llm": 0.0, "grad_norm": 0.04547649994492531, "global_step": 24585, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03346018120646477, "train_loss_bc": 0.03346018120646477, "train_loss_llm": 0.0, "grad_norm": 0.06295283138751984, "global_step": 24586, "epoch": 682, "lr": 0.001} +{"train_loss": 0.03172255426438318, "train_loss_bc": 0.037126801908016205, "train_loss_llm": 0.0, "grad_norm": 0.07478701323270798, "global_step": 24587, "epoch": 682, "lr": 0.001, "val_loss": 0.08816242963075638} +{"train_loss": 0.027166087180376053, "train_loss_bc": 0.027166087180376053, "train_loss_llm": 0.0, "grad_norm": 0.09335238486528397, "global_step": 24588, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03137139603495598, "train_loss_bc": 0.03136724233627319, "train_loss_llm": 0.0004152655601501465, "grad_norm": 0.032497748732566833, "global_step": 24589, "epoch": 683, "lr": 0.001} +{"train_loss": 0.0317581407725811, "train_loss_bc": 0.0317581407725811, "train_loss_llm": 0.0, "grad_norm": 0.041610755026340485, "global_step": 24590, "epoch": 683, "lr": 0.001} +{"train_loss": 0.031356241554021835, "train_loss_bc": 0.031356241554021835, "train_loss_llm": 0.0, "grad_norm": 0.04825585335493088, "global_step": 24591, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03444574028253555, "train_loss_bc": 0.034442830830812454, "train_loss_llm": 0.00029096007347106934, "grad_norm": 0.061475951224565506, "global_step": 24592, "epoch": 683, "lr": 0.001} +{"train_loss": 0.02854711003601551, "train_loss_bc": 0.02854711003601551, "train_loss_llm": 0.0, "grad_norm": 0.024915020912885666, "global_step": 24593, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03841697424650192, "train_loss_bc": 0.03840763121843338, "train_loss_llm": 0.0009342730045318604, "grad_norm": 0.047390490770339966, "global_step": 24594, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03448648750782013, "train_loss_bc": 0.03448648750782013, "train_loss_llm": 0.0, "grad_norm": 0.06503631174564362, "global_step": 24595, "epoch": 683, "lr": 0.001} +{"train_loss": 0.029376758262515068, "train_loss_bc": 0.029376758262515068, "train_loss_llm": 0.0, "grad_norm": 0.0767202079296112, "global_step": 24596, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03460273891687393, "train_loss_bc": 0.03460273891687393, "train_loss_llm": 0.0, "grad_norm": 0.02832808718085289, "global_step": 24597, "epoch": 683, "lr": 0.001} +{"train_loss": 0.022214442491531372, "train_loss_bc": 0.02220858447253704, "train_loss_llm": 0.0005858838558197021, "grad_norm": 0.047980889678001404, "global_step": 24598, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03640284761786461, "train_loss_bc": 0.03640284761786461, "train_loss_llm": 0.0, "grad_norm": 0.07882428914308548, "global_step": 24599, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03258175775408745, "train_loss_bc": 0.03258175775408745, "train_loss_llm": 0.0, "grad_norm": 0.09589478373527527, "global_step": 24600, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03148432821035385, "train_loss_bc": 0.03148432821035385, "train_loss_llm": 0.0, "grad_norm": 0.03211941197514534, "global_step": 24601, "epoch": 683, "lr": 0.001} +{"train_loss": 0.02735736593604088, "train_loss_bc": 0.027348218485713005, "train_loss_llm": 0.0009146928787231445, "grad_norm": 0.05284586176276207, "global_step": 24602, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03323155641555786, "train_loss_bc": 0.03323155641555786, "train_loss_llm": 0.0, "grad_norm": 0.07813363522291183, "global_step": 24603, "epoch": 683, "lr": 0.001} +{"train_loss": 0.032535817474126816, "train_loss_bc": 0.03253554925322533, "train_loss_llm": 2.6911497116088867e-05, "grad_norm": 0.1086844652891159, "global_step": 24604, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03420557454228401, "train_loss_bc": 0.034175947308540344, "train_loss_llm": 0.002962768077850342, "grad_norm": 0.038622692227363586, "global_step": 24605, "epoch": 683, "lr": 0.001} +{"train_loss": 0.02870587259531021, "train_loss_bc": 0.02870587259531021, "train_loss_llm": 0.0, "grad_norm": 0.054158881306648254, "global_step": 24606, "epoch": 683, "lr": 0.001} +{"train_loss": 0.029041379690170288, "train_loss_bc": 0.029041379690170288, "train_loss_llm": 0.0, "grad_norm": 0.06798837333917618, "global_step": 24607, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03101547807455063, "train_loss_bc": 0.03101547807455063, "train_loss_llm": 0.0, "grad_norm": 0.08466067165136337, "global_step": 24608, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03048587031662464, "train_loss_bc": 0.03048587031662464, "train_loss_llm": 0.0, "grad_norm": 0.033383529633283615, "global_step": 24609, "epoch": 683, "lr": 0.001} +{"train_loss": 0.029549282044172287, "train_loss_bc": 0.02954365499317646, "train_loss_llm": 0.0005626678466796875, "grad_norm": 0.05355140194296837, "global_step": 24610, "epoch": 683, "lr": 0.001} +{"train_loss": 0.027337484061717987, "train_loss_bc": 0.02732723578810692, "train_loss_llm": 0.0010249018669128418, "grad_norm": 0.05879714712500572, "global_step": 24611, "epoch": 683, "lr": 0.001} +{"train_loss": 0.030677417293190956, "train_loss_bc": 0.030667362734675407, "train_loss_llm": 0.0010054707527160645, "grad_norm": 0.07748913764953613, "global_step": 24612, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03350768983364105, "train_loss_bc": 0.03350768983364105, "train_loss_llm": 0.0, "grad_norm": 0.04056249558925629, "global_step": 24613, "epoch": 683, "lr": 0.001} +{"train_loss": 0.034821320325136185, "train_loss_bc": 0.034821320325136185, "train_loss_llm": 0.0, "grad_norm": 0.07733093202114105, "global_step": 24614, "epoch": 683, "lr": 0.001} +{"train_loss": 0.031043801456689835, "train_loss_bc": 0.031043801456689835, "train_loss_llm": 0.0, "grad_norm": 0.09787219762802124, "global_step": 24615, "epoch": 683, "lr": 0.001} +{"train_loss": 0.031946245580911636, "train_loss_bc": 0.031946245580911636, "train_loss_llm": 0.0, "grad_norm": 0.10981329530477524, "global_step": 24616, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03629966825246811, "train_loss_bc": 0.03629966825246811, "train_loss_llm": 0.0, "grad_norm": 0.039179787039756775, "global_step": 24617, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03037761151790619, "train_loss_bc": 0.03037164732813835, "train_loss_llm": 0.0005963444709777832, "grad_norm": 0.06824985146522522, "global_step": 24618, "epoch": 683, "lr": 0.001} +{"train_loss": 0.031945060938596725, "train_loss_bc": 0.031945060938596725, "train_loss_llm": 0.0, "grad_norm": 0.08387143164873123, "global_step": 24619, "epoch": 683, "lr": 0.001} +{"train_loss": 0.02812248095870018, "train_loss_bc": 0.02812248095870018, "train_loss_llm": 0.0, "grad_norm": 0.10556115955114365, "global_step": 24620, "epoch": 683, "lr": 0.001} +{"train_loss": 0.03838859498500824, "train_loss_bc": 0.03838738426566124, "train_loss_llm": 0.00012090802192687988, "grad_norm": 0.03900029510259628, "global_step": 24621, "epoch": 683, "lr": 0.001} +{"train_loss": 0.031602732837200165, "train_loss_bc": 0.031602732837200165, "train_loss_llm": 0.0, "grad_norm": 0.055145472288131714, "global_step": 24622, "epoch": 683, "lr": 0.001} +{"train_loss": 0.0316406452200479, "train_loss_bc": 0.03264867514371872, "train_loss_llm": 0.0005195438861846924, "grad_norm": 0.07910791039466858, "global_step": 24623, "epoch": 683, "lr": 0.001, "val_loss": 0.08763514459133148} +{"train_loss": 0.033941760659217834, "train_loss_bc": 0.033941760659217834, "train_loss_llm": 0.0, "grad_norm": 0.10594962537288666, "global_step": 24624, "epoch": 684, "lr": 0.001} +{"train_loss": 0.025791753083467484, "train_loss_bc": 0.025782335549592972, "train_loss_llm": 0.000941842794418335, "grad_norm": 0.027935488149523735, "global_step": 24625, "epoch": 684, "lr": 0.001} +{"train_loss": 0.0330905020236969, "train_loss_bc": 0.03308100625872612, "train_loss_llm": 0.0009497404098510742, "grad_norm": 0.04842740669846535, "global_step": 24626, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03142092749476433, "train_loss_bc": 0.031413666903972626, "train_loss_llm": 0.0007262229919433594, "grad_norm": 0.06848803162574768, "global_step": 24627, "epoch": 684, "lr": 0.001} +{"train_loss": 0.02744143456220627, "train_loss_bc": 0.027434449642896652, "train_loss_llm": 0.000698477029800415, "grad_norm": 0.08671414852142334, "global_step": 24628, "epoch": 684, "lr": 0.001} +{"train_loss": 0.033415865153074265, "train_loss_bc": 0.033415865153074265, "train_loss_llm": 0.0, "grad_norm": 0.03663210570812225, "global_step": 24629, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03121054358780384, "train_loss_bc": 0.03121054358780384, "train_loss_llm": 0.0, "grad_norm": 0.0509665384888649, "global_step": 24630, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03501841798424721, "train_loss_bc": 0.03501841798424721, "train_loss_llm": 0.0, "grad_norm": 0.0773058831691742, "global_step": 24631, "epoch": 684, "lr": 0.001} +{"train_loss": 0.04069802165031433, "train_loss_bc": 0.04069802165031433, "train_loss_llm": 0.0, "grad_norm": 0.1022363007068634, "global_step": 24632, "epoch": 684, "lr": 0.001} +{"train_loss": 0.032058943063020706, "train_loss_bc": 0.032058943063020706, "train_loss_llm": 0.0, "grad_norm": 0.030225137248635292, "global_step": 24633, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03332604095339775, "train_loss_bc": 0.033317845314741135, "train_loss_llm": 0.000819474458694458, "grad_norm": 0.05150142312049866, "global_step": 24634, "epoch": 684, "lr": 0.001} +{"train_loss": 0.029482152312994003, "train_loss_bc": 0.029482152312994003, "train_loss_llm": 0.0, "grad_norm": 0.08373427391052246, "global_step": 24635, "epoch": 684, "lr": 0.001} +{"train_loss": 0.029356282204389572, "train_loss_bc": 0.029356282204389572, "train_loss_llm": 0.0, "grad_norm": 0.09656193107366562, "global_step": 24636, "epoch": 684, "lr": 0.001} +{"train_loss": 0.02751891501247883, "train_loss_bc": 0.02747018076479435, "train_loss_llm": 0.004873424768447876, "grad_norm": 0.03479544073343277, "global_step": 24637, "epoch": 684, "lr": 0.001} +{"train_loss": 0.0292514655739069, "train_loss_bc": 0.029233098030090332, "train_loss_llm": 0.0018366873264312744, "grad_norm": 0.05749940872192383, "global_step": 24638, "epoch": 684, "lr": 0.001} +{"train_loss": 0.032774344086647034, "train_loss_bc": 0.032774344086647034, "train_loss_llm": 0.0, "grad_norm": 0.06811297684907913, "global_step": 24639, "epoch": 684, "lr": 0.001} +{"train_loss": 0.0340074859559536, "train_loss_bc": 0.0340074859559536, "train_loss_llm": 0.0, "grad_norm": 0.0888008251786232, "global_step": 24640, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03584353253245354, "train_loss_bc": 0.03584353253245354, "train_loss_llm": 0.0, "grad_norm": 0.031716808676719666, "global_step": 24641, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03345610946416855, "train_loss_bc": 0.03345610946416855, "train_loss_llm": 0.0, "grad_norm": 0.046541858464479446, "global_step": 24642, "epoch": 684, "lr": 0.001} +{"train_loss": 0.025105992332100868, "train_loss_bc": 0.025105992332100868, "train_loss_llm": 0.0, "grad_norm": 0.05730926990509033, "global_step": 24643, "epoch": 684, "lr": 0.001} +{"train_loss": 0.022525912150740623, "train_loss_bc": 0.022525912150740623, "train_loss_llm": 0.0, "grad_norm": 0.07327255606651306, "global_step": 24644, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03687736764550209, "train_loss_bc": 0.03687736764550209, "train_loss_llm": 0.0, "grad_norm": 0.03232358768582344, "global_step": 24645, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03013433702290058, "train_loss_bc": 0.03013433702290058, "train_loss_llm": 0.0, "grad_norm": 0.04397929832339287, "global_step": 24646, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03175976499915123, "train_loss_bc": 0.03175976499915123, "train_loss_llm": 0.0, "grad_norm": 0.06050011143088341, "global_step": 24647, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03027428686618805, "train_loss_bc": 0.03027428686618805, "train_loss_llm": 0.0, "grad_norm": 0.07837599515914917, "global_step": 24648, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03395240381360054, "train_loss_bc": 0.03394424915313721, "train_loss_llm": 0.0008156299591064453, "grad_norm": 0.044895850121974945, "global_step": 24649, "epoch": 684, "lr": 0.001} +{"train_loss": 0.026184631511569023, "train_loss_bc": 0.026184631511569023, "train_loss_llm": 0.0, "grad_norm": 0.06440360099077225, "global_step": 24650, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03444283455610275, "train_loss_bc": 0.03443829342722893, "train_loss_llm": 0.00045424699783325195, "grad_norm": 0.07904770970344543, "global_step": 24651, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03203978016972542, "train_loss_bc": 0.03203978016972542, "train_loss_llm": 0.0, "grad_norm": 0.09947515279054642, "global_step": 24652, "epoch": 684, "lr": 0.001} +{"train_loss": 0.030364736914634705, "train_loss_bc": 0.030364736914634705, "train_loss_llm": 0.0, "grad_norm": 0.032372210174798965, "global_step": 24653, "epoch": 684, "lr": 0.001} +{"train_loss": 0.037931714206933975, "train_loss_bc": 0.037931714206933975, "train_loss_llm": 0.0, "grad_norm": 0.06415259093046188, "global_step": 24654, "epoch": 684, "lr": 0.001} +{"train_loss": 0.035333890467882156, "train_loss_bc": 0.035333890467882156, "train_loss_llm": 0.0, "grad_norm": 0.0904129147529602, "global_step": 24655, "epoch": 684, "lr": 0.001} +{"train_loss": 0.026918498799204826, "train_loss_bc": 0.026918498799204826, "train_loss_llm": 0.0, "grad_norm": 0.10669386386871338, "global_step": 24656, "epoch": 684, "lr": 0.001} +{"train_loss": 0.034863244742155075, "train_loss_bc": 0.034863244742155075, "train_loss_llm": 0.0, "grad_norm": 0.04020976647734642, "global_step": 24657, "epoch": 684, "lr": 0.001} +{"train_loss": 0.03099861554801464, "train_loss_bc": 0.03099861554801464, "train_loss_llm": 0.0, "grad_norm": 0.05350268632173538, "global_step": 24658, "epoch": 684, "lr": 0.001} +{"train_loss": 0.031717212612016335, "train_loss_bc": 0.03300129622220993, "train_loss_llm": 0.0005849003791809082, "grad_norm": 0.07687725871801376, "global_step": 24659, "epoch": 684, "lr": 0.001, "val_loss": 0.08688735216856003} +{"train_loss": 0.032020047307014465, "train_loss_bc": 0.032020047307014465, "train_loss_llm": 0.0, "grad_norm": 0.08735162019729614, "global_step": 24660, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03311844542622566, "train_loss_bc": 0.03311844542622566, "train_loss_llm": 0.0, "grad_norm": 0.02819748967885971, "global_step": 24661, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03748834878206253, "train_loss_bc": 0.03748834878206253, "train_loss_llm": 0.0, "grad_norm": 0.045606862753629684, "global_step": 24662, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03497971594333649, "train_loss_bc": 0.03495445102453232, "train_loss_llm": 0.0025266706943511963, "grad_norm": 0.06289048492908478, "global_step": 24663, "epoch": 685, "lr": 0.001} +{"train_loss": 0.02936691790819168, "train_loss_bc": 0.02936691790819168, "train_loss_llm": 0.0, "grad_norm": 0.07674401253461838, "global_step": 24664, "epoch": 685, "lr": 0.001} +{"train_loss": 0.036559704691171646, "train_loss_bc": 0.036559704691171646, "train_loss_llm": 0.0, "grad_norm": 0.036860376596450806, "global_step": 24665, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03028896078467369, "train_loss_bc": 0.03028896078467369, "train_loss_llm": 0.0, "grad_norm": 0.05161780118942261, "global_step": 24666, "epoch": 685, "lr": 0.001} +{"train_loss": 0.025230295956134796, "train_loss_bc": 0.025230295956134796, "train_loss_llm": 0.0, "grad_norm": 0.060609810054302216, "global_step": 24667, "epoch": 685, "lr": 0.001} +{"train_loss": 0.02982945367693901, "train_loss_bc": 0.02982945367693901, "train_loss_llm": 0.0, "grad_norm": 0.07481829822063446, "global_step": 24668, "epoch": 685, "lr": 0.001} +{"train_loss": 0.024210156872868538, "train_loss_bc": 0.024210156872868538, "train_loss_llm": 0.0, "grad_norm": 0.031555116176605225, "global_step": 24669, "epoch": 685, "lr": 0.001} +{"train_loss": 0.028905656188726425, "train_loss_bc": 0.028905656188726425, "train_loss_llm": 0.0, "grad_norm": 0.04018377512693405, "global_step": 24670, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03536899387836456, "train_loss_bc": 0.03536899387836456, "train_loss_llm": 0.0, "grad_norm": 0.055483318865299225, "global_step": 24671, "epoch": 685, "lr": 0.001} +{"train_loss": 0.02824372984468937, "train_loss_bc": 0.02824372984468937, "train_loss_llm": 0.0, "grad_norm": 0.07091338187456131, "global_step": 24672, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03803962841629982, "train_loss_bc": 0.03803630173206329, "train_loss_llm": 0.00033274292945861816, "grad_norm": 0.03701283410191536, "global_step": 24673, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03264891356229782, "train_loss_bc": 0.03264891356229782, "train_loss_llm": 0.0, "grad_norm": 0.05148159712553024, "global_step": 24674, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03240472450852394, "train_loss_bc": 0.03240472450852394, "train_loss_llm": 0.0, "grad_norm": 0.06534804403781891, "global_step": 24675, "epoch": 685, "lr": 0.001} +{"train_loss": 0.031198665499687195, "train_loss_bc": 0.031198665499687195, "train_loss_llm": 0.0, "grad_norm": 0.08507242053747177, "global_step": 24676, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03367069736123085, "train_loss_bc": 0.03367069736123085, "train_loss_llm": 0.0, "grad_norm": 0.03797660022974014, "global_step": 24677, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03974561765789986, "train_loss_bc": 0.03973780944943428, "train_loss_llm": 0.0007809996604919434, "grad_norm": 0.053112827241420746, "global_step": 24678, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03274101763963699, "train_loss_bc": 0.03274101763963699, "train_loss_llm": 0.0, "grad_norm": 0.0666334480047226, "global_step": 24679, "epoch": 685, "lr": 0.001} +{"train_loss": 0.031250737607479095, "train_loss_bc": 0.031250737607479095, "train_loss_llm": 0.0, "grad_norm": 0.087016262114048, "global_step": 24680, "epoch": 685, "lr": 0.001} +{"train_loss": 0.025326460599899292, "train_loss_bc": 0.02530067041516304, "train_loss_llm": 0.0025789737701416016, "grad_norm": 0.02782733179628849, "global_step": 24681, "epoch": 685, "lr": 0.001} +{"train_loss": 0.030605385079979897, "train_loss_bc": 0.030591275542974472, "train_loss_llm": 0.0014109015464782715, "grad_norm": 0.046436384320259094, "global_step": 24682, "epoch": 685, "lr": 0.001} +{"train_loss": 0.027748407796025276, "train_loss_bc": 0.027748407796025276, "train_loss_llm": 0.0, "grad_norm": 0.06009218841791153, "global_step": 24683, "epoch": 685, "lr": 0.001} +{"train_loss": 0.031249139457941055, "train_loss_bc": 0.031249139457941055, "train_loss_llm": 0.0, "grad_norm": 0.08864092081785202, "global_step": 24684, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03689322620630264, "train_loss_bc": 0.03689322620630264, "train_loss_llm": 0.0, "grad_norm": 0.03868703171610832, "global_step": 24685, "epoch": 685, "lr": 0.001} +{"train_loss": 0.030356619507074356, "train_loss_bc": 0.030356619507074356, "train_loss_llm": 0.0, "grad_norm": 0.05591634660959244, "global_step": 24686, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03639950230717659, "train_loss_bc": 0.0363847054541111, "train_loss_llm": 0.0014798343181610107, "grad_norm": 0.07594044506549835, "global_step": 24687, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03111596591770649, "train_loss_bc": 0.031112510710954666, "train_loss_llm": 0.00034555792808532715, "grad_norm": 0.0900537371635437, "global_step": 24688, "epoch": 685, "lr": 0.001} +{"train_loss": 0.024647140875458717, "train_loss_bc": 0.024647140875458717, "train_loss_llm": 0.0, "grad_norm": 0.02560078352689743, "global_step": 24689, "epoch": 685, "lr": 0.001} +{"train_loss": 0.029542837291955948, "train_loss_bc": 0.029542837291955948, "train_loss_llm": 0.0, "grad_norm": 0.04011824354529381, "global_step": 24690, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03369865566492081, "train_loss_bc": 0.03369865566492081, "train_loss_llm": 0.0, "grad_norm": 0.05990726873278618, "global_step": 24691, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03251255676150322, "train_loss_bc": 0.03251255676150322, "train_loss_llm": 0.0, "grad_norm": 0.07998139411211014, "global_step": 24692, "epoch": 685, "lr": 0.001} +{"train_loss": 0.02596283331513405, "train_loss_bc": 0.02595679461956024, "train_loss_llm": 0.0006039142608642578, "grad_norm": 0.03274287283420563, "global_step": 24693, "epoch": 685, "lr": 0.001} +{"train_loss": 0.030016852542757988, "train_loss_bc": 0.030011970549821854, "train_loss_llm": 0.0004881918430328369, "grad_norm": 0.053736984729766846, "global_step": 24694, "epoch": 685, "lr": 0.001} +{"train_loss": 0.03159022843465209, "train_loss_bc": 0.03386221081018448, "train_loss_llm": 0.0, "grad_norm": 0.07535284757614136, "global_step": 24695, "epoch": 685, "lr": 0.001, "train/cumulative_reward": 86.07230604980322, "train/mean_score": 16.82112885602096, "train/success_rate": 0.4, "test/cumulative_reward": 40.87115237700561, "test/mean_score": 8.635135662283771, "test/success_rate": 0.22, "val_loss": 0.08810455352067947, "train_action_mse_error": 0.03382081910967827} +{"train_loss": 0.03253655880689621, "train_loss_bc": 0.03253655880689621, "train_loss_llm": 0.0, "grad_norm": 0.09034259617328644, "global_step": 24696, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03232554346323013, "train_loss_bc": 0.032308973371982574, "train_loss_llm": 0.0016571283340454102, "grad_norm": 0.03668986260890961, "global_step": 24697, "epoch": 686, "lr": 0.001} +{"train_loss": 0.028788067400455475, "train_loss_bc": 0.028788067400455475, "train_loss_llm": 0.0, "grad_norm": 0.05475127696990967, "global_step": 24698, "epoch": 686, "lr": 0.001} +{"train_loss": 0.030522562563419342, "train_loss_bc": 0.030522562563419342, "train_loss_llm": 0.0, "grad_norm": 0.07213303446769714, "global_step": 24699, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03539638593792915, "train_loss_bc": 0.03539638593792915, "train_loss_llm": 0.0, "grad_norm": 0.09736715257167816, "global_step": 24700, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03002321533858776, "train_loss_bc": 0.03002321533858776, "train_loss_llm": 0.0, "grad_norm": 0.03364463523030281, "global_step": 24701, "epoch": 686, "lr": 0.001} +{"train_loss": 0.029475774616003036, "train_loss_bc": 0.029467852786183357, "train_loss_llm": 0.0007921159267425537, "grad_norm": 0.044939424842596054, "global_step": 24702, "epoch": 686, "lr": 0.001} +{"train_loss": 0.034542109817266464, "train_loss_bc": 0.034542109817266464, "train_loss_llm": 0.0, "grad_norm": 0.06435892730951309, "global_step": 24703, "epoch": 686, "lr": 0.001} +{"train_loss": 0.026924489066004753, "train_loss_bc": 0.026924489066004753, "train_loss_llm": 0.0, "grad_norm": 0.07649663090705872, "global_step": 24704, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03065897524356842, "train_loss_bc": 0.03065897524356842, "train_loss_llm": 0.0, "grad_norm": 0.03171665221452713, "global_step": 24705, "epoch": 686, "lr": 0.001} +{"train_loss": 0.031181085854768753, "train_loss_bc": 0.031176665797829628, "train_loss_llm": 0.0004419386386871338, "grad_norm": 0.0508137121796608, "global_step": 24706, "epoch": 686, "lr": 0.001} +{"train_loss": 0.031193602830171585, "train_loss_bc": 0.031193602830171585, "train_loss_llm": 0.0, "grad_norm": 0.06181516870856285, "global_step": 24707, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03306214511394501, "train_loss_bc": 0.03306214511394501, "train_loss_llm": 0.0, "grad_norm": 0.08312124758958817, "global_step": 24708, "epoch": 686, "lr": 0.001} +{"train_loss": 0.031786657869815826, "train_loss_bc": 0.031786657869815826, "train_loss_llm": 0.0, "grad_norm": 0.030878763645887375, "global_step": 24709, "epoch": 686, "lr": 0.001} +{"train_loss": 0.030131349340081215, "train_loss_bc": 0.030131349340081215, "train_loss_llm": 0.0, "grad_norm": 0.0495445542037487, "global_step": 24710, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03316900134086609, "train_loss_bc": 0.03316900134086609, "train_loss_llm": 0.0, "grad_norm": 0.06880000233650208, "global_step": 24711, "epoch": 686, "lr": 0.001} +{"train_loss": 0.030643906444311142, "train_loss_bc": 0.030643906444311142, "train_loss_llm": 0.0, "grad_norm": 0.0757066160440445, "global_step": 24712, "epoch": 686, "lr": 0.001} +{"train_loss": 0.031393345445394516, "train_loss_bc": 0.031393345445394516, "train_loss_llm": 0.0, "grad_norm": 0.03501654788851738, "global_step": 24713, "epoch": 686, "lr": 0.001} +{"train_loss": 0.036388639360666275, "train_loss_bc": 0.03638524189591408, "train_loss_llm": 0.0003395676612854004, "grad_norm": 0.0518917441368103, "global_step": 24714, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03507551923394203, "train_loss_bc": 0.03507551923394203, "train_loss_llm": 0.0, "grad_norm": 0.07125689089298248, "global_step": 24715, "epoch": 686, "lr": 0.001} +{"train_loss": 0.025097686797380447, "train_loss_bc": 0.025094609707593918, "train_loss_llm": 0.0003076493740081787, "grad_norm": 0.09536806493997574, "global_step": 24716, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03458918258547783, "train_loss_bc": 0.03458918258547783, "train_loss_llm": 0.0, "grad_norm": 0.03566918522119522, "global_step": 24717, "epoch": 686, "lr": 0.001} +{"train_loss": 0.031084828078746796, "train_loss_bc": 0.031084828078746796, "train_loss_llm": 0.0, "grad_norm": 0.05085268244147301, "global_step": 24718, "epoch": 686, "lr": 0.001} +{"train_loss": 0.032671187072992325, "train_loss_bc": 0.032644834369421005, "train_loss_llm": 0.0026351511478424072, "grad_norm": 0.06082334741950035, "global_step": 24719, "epoch": 686, "lr": 0.001} +{"train_loss": 0.031808432191610336, "train_loss_bc": 0.031808432191610336, "train_loss_llm": 0.0, "grad_norm": 0.09039396047592163, "global_step": 24720, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03241356462240219, "train_loss_bc": 0.03241182863712311, "train_loss_llm": 0.00017374753952026367, "grad_norm": 0.02759464643895626, "global_step": 24721, "epoch": 686, "lr": 0.001} +{"train_loss": 0.02852451242506504, "train_loss_bc": 0.02852451242506504, "train_loss_llm": 0.0, "grad_norm": 0.04751456156373024, "global_step": 24722, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03060588613152504, "train_loss_bc": 0.03060588613152504, "train_loss_llm": 0.0, "grad_norm": 0.0659525915980339, "global_step": 24723, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03473420813679695, "train_loss_bc": 0.03470396250486374, "train_loss_llm": 0.003024578094482422, "grad_norm": 0.09190365672111511, "global_step": 24724, "epoch": 686, "lr": 0.001} +{"train_loss": 0.029995454475283623, "train_loss_bc": 0.029995454475283623, "train_loss_llm": 0.0, "grad_norm": 0.028918547555804253, "global_step": 24725, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03548499196767807, "train_loss_bc": 0.03548499196767807, "train_loss_llm": 0.0, "grad_norm": 0.04770958796143532, "global_step": 24726, "epoch": 686, "lr": 0.001} +{"train_loss": 0.026487603783607483, "train_loss_bc": 0.026487603783607483, "train_loss_llm": 0.0, "grad_norm": 0.058793384581804276, "global_step": 24727, "epoch": 686, "lr": 0.001} +{"train_loss": 0.0315806046128273, "train_loss_bc": 0.0315806046128273, "train_loss_llm": 0.0, "grad_norm": 0.07106798887252808, "global_step": 24728, "epoch": 686, "lr": 0.001} +{"train_loss": 0.032260093837976456, "train_loss_bc": 0.032235287129879, "train_loss_llm": 0.0024805665016174316, "grad_norm": 0.03106808103621006, "global_step": 24729, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03129302337765694, "train_loss_bc": 0.03129302337765694, "train_loss_llm": 0.0, "grad_norm": 0.04805025830864906, "global_step": 24730, "epoch": 686, "lr": 0.001} +{"train_loss": 0.03160683101870947, "train_loss_bc": 0.033995721489191055, "train_loss_llm": 0.0, "grad_norm": 0.06341009587049484, "global_step": 24731, "epoch": 686, "lr": 0.001, "val_loss": 0.08814803510904312} +{"train_loss": 0.032018229365348816, "train_loss_bc": 0.032018229365348816, "train_loss_llm": 0.0, "grad_norm": 0.08728701621294022, "global_step": 24732, "epoch": 687, "lr": 0.001} +{"train_loss": 0.030463112518191338, "train_loss_bc": 0.030463112518191338, "train_loss_llm": 0.0, "grad_norm": 0.026217201724648476, "global_step": 24733, "epoch": 687, "lr": 0.001} +{"train_loss": 0.0331084281206131, "train_loss_bc": 0.0331084281206131, "train_loss_llm": 0.0, "grad_norm": 0.05603475123643875, "global_step": 24734, "epoch": 687, "lr": 0.001} +{"train_loss": 0.040802352130413055, "train_loss_bc": 0.040802352130413055, "train_loss_llm": 0.0, "grad_norm": 0.08570123463869095, "global_step": 24735, "epoch": 687, "lr": 0.001} +{"train_loss": 0.02778695710003376, "train_loss_bc": 0.02778695710003376, "train_loss_llm": 0.0, "grad_norm": 0.10240020602941513, "global_step": 24736, "epoch": 687, "lr": 0.001} +{"train_loss": 0.031407780945301056, "train_loss_bc": 0.03139067068696022, "train_loss_llm": 0.0017111599445343018, "grad_norm": 0.03237001970410347, "global_step": 24737, "epoch": 687, "lr": 0.001} +{"train_loss": 0.03153399005532265, "train_loss_bc": 0.03153399005532265, "train_loss_llm": 0.0, "grad_norm": 0.050034403800964355, "global_step": 24738, "epoch": 687, "lr": 0.001} +{"train_loss": 0.0365026630461216, "train_loss_bc": 0.0365026630461216, "train_loss_llm": 0.0, "grad_norm": 0.0908840149641037, "global_step": 24739, "epoch": 687, "lr": 0.001} +{"train_loss": 0.036252621561288834, "train_loss_bc": 0.036252621561288834, "train_loss_llm": 0.0, "grad_norm": 0.1146029606461525, "global_step": 24740, "epoch": 687, "lr": 0.001} +{"train_loss": 0.03212691843509674, "train_loss_bc": 0.03212691843509674, "train_loss_llm": 0.0, "grad_norm": 0.04619977995753288, "global_step": 24741, "epoch": 687, "lr": 0.001} +{"train_loss": 0.031528498977422714, "train_loss_bc": 0.031528498977422714, "train_loss_llm": 0.0, "grad_norm": 0.06604444980621338, "global_step": 24742, "epoch": 687, "lr": 0.001} +{"train_loss": 0.031833335757255554, "train_loss_bc": 0.031833335757255554, "train_loss_llm": 0.0, "grad_norm": 0.09217511862516403, "global_step": 24743, "epoch": 687, "lr": 0.001} +{"train_loss": 0.025366060435771942, "train_loss_bc": 0.025366060435771942, "train_loss_llm": 0.0, "grad_norm": 0.11214646697044373, "global_step": 24744, "epoch": 687, "lr": 0.001} +{"train_loss": 0.03199619799852371, "train_loss_bc": 0.03199619799852371, "train_loss_llm": 0.0, "grad_norm": 0.030870236456394196, "global_step": 24745, "epoch": 687, "lr": 0.001} +{"train_loss": 0.030474692583084106, "train_loss_bc": 0.030474692583084106, "train_loss_llm": 0.0, "grad_norm": 0.05872243270277977, "global_step": 24746, "epoch": 687, "lr": 0.001} +{"train_loss": 0.030648643150925636, "train_loss_bc": 0.03063122369349003, "train_loss_llm": 0.0017420053482055664, "grad_norm": 0.0774010643362999, "global_step": 24747, "epoch": 687, "lr": 0.001} +{"train_loss": 0.03260427340865135, "train_loss_bc": 0.03260427340865135, "train_loss_llm": 0.0, "grad_norm": 0.09094448387622833, "global_step": 24748, "epoch": 687, "lr": 0.001} +{"train_loss": 0.029321886599063873, "train_loss_bc": 0.029321160167455673, "train_loss_llm": 7.256865501403809e-05, "grad_norm": 0.033484794199466705, "global_step": 24749, "epoch": 687, "lr": 0.001} +{"train_loss": 0.0309344120323658, "train_loss_bc": 0.0309344120323658, "train_loss_llm": 0.0, "grad_norm": 0.051880136132240295, "global_step": 24750, "epoch": 687, "lr": 0.001} +{"train_loss": 0.03739898279309273, "train_loss_bc": 0.03739898279309273, "train_loss_llm": 0.0, "grad_norm": 0.07514332979917526, "global_step": 24751, "epoch": 687, "lr": 0.001} +{"train_loss": 0.03570636734366417, "train_loss_bc": 0.03570636734366417, "train_loss_llm": 0.0, "grad_norm": 0.08911679685115814, "global_step": 24752, "epoch": 687, "lr": 0.001} +{"train_loss": 0.030774351209402084, "train_loss_bc": 0.030771220102906227, "train_loss_llm": 0.00031304359436035156, "grad_norm": 0.032894961535930634, "global_step": 24753, "epoch": 687, "lr": 0.001} +{"train_loss": 0.031597476452589035, "train_loss_bc": 0.0315873809158802, "train_loss_llm": 0.0010097324848175049, "grad_norm": 0.0573260523378849, "global_step": 24754, "epoch": 687, "lr": 0.001} +{"train_loss": 0.0335635282099247, "train_loss_bc": 0.0335635282099247, "train_loss_llm": 0.0, "grad_norm": 0.0652075707912445, "global_step": 24755, "epoch": 687, "lr": 0.001} +{"train_loss": 0.02890698052942753, "train_loss_bc": 0.02890698052942753, "train_loss_llm": 0.0, "grad_norm": 0.08216531574726105, "global_step": 24756, "epoch": 687, "lr": 0.001} +{"train_loss": 0.02408597245812416, "train_loss_bc": 0.02408597245812416, "train_loss_llm": 0.0, "grad_norm": 0.023741448298096657, "global_step": 24757, "epoch": 687, "lr": 0.001} +{"train_loss": 0.03480216860771179, "train_loss_bc": 0.03480216860771179, "train_loss_llm": 0.0, "grad_norm": 0.044857196509838104, "global_step": 24758, "epoch": 687, "lr": 0.001} +{"train_loss": 0.028155837208032608, "train_loss_bc": 0.028155837208032608, "train_loss_llm": 0.0, "grad_norm": 0.06070088967680931, "global_step": 24759, "epoch": 687, "lr": 0.001} +{"train_loss": 0.03338552638888359, "train_loss_bc": 0.03338181972503662, "train_loss_llm": 0.0003705024719238281, "grad_norm": 0.0775526762008667, "global_step": 24760, "epoch": 687, "lr": 0.001} +{"train_loss": 0.0304790660738945, "train_loss_bc": 0.030460864305496216, "train_loss_llm": 0.0018201172351837158, "grad_norm": 0.030881810933351517, "global_step": 24761, "epoch": 687, "lr": 0.001} +{"train_loss": 0.0264964010566473, "train_loss_bc": 0.02648531086742878, "train_loss_llm": 0.0011089444160461426, "grad_norm": 0.04583042487502098, "global_step": 24762, "epoch": 687, "lr": 0.001} +{"train_loss": 0.037423644214868546, "train_loss_bc": 0.03741365298628807, "train_loss_llm": 0.0009992122650146484, "grad_norm": 0.08536779880523682, "global_step": 24763, "epoch": 687, "lr": 0.001} +{"train_loss": 0.03074583038687706, "train_loss_bc": 0.03074583038687706, "train_loss_llm": 0.0, "grad_norm": 0.09737727791070938, "global_step": 24764, "epoch": 687, "lr": 0.001} +{"train_loss": 0.029546106234192848, "train_loss_bc": 0.029545804485678673, "train_loss_llm": 3.0249357223510742e-05, "grad_norm": 0.04061001166701317, "global_step": 24765, "epoch": 687, "lr": 0.001} +{"train_loss": 0.031158417463302612, "train_loss_bc": 0.03115447610616684, "train_loss_llm": 0.0003942251205444336, "grad_norm": 0.07126282155513763, "global_step": 24766, "epoch": 687, "lr": 0.001} +{"train_loss": 0.0318931360832519, "train_loss_bc": 0.03721518814563751, "train_loss_llm": 0.0, "grad_norm": 0.10099700093269348, "global_step": 24767, "epoch": 687, "lr": 0.001, "val_loss": 0.08808544278144836} +{"train_loss": 0.031736891716718674, "train_loss_bc": 0.03173677623271942, "train_loss_llm": 1.1533498764038086e-05, "grad_norm": 0.1251203566789627, "global_step": 24768, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03243669867515564, "train_loss_bc": 0.03243669867515564, "train_loss_llm": 0.0, "grad_norm": 0.037066228687763214, "global_step": 24769, "epoch": 688, "lr": 0.001} +{"train_loss": 0.032175272703170776, "train_loss_bc": 0.032175272703170776, "train_loss_llm": 0.0, "grad_norm": 0.05670595169067383, "global_step": 24770, "epoch": 688, "lr": 0.001} +{"train_loss": 0.034175995737314224, "train_loss_bc": 0.03416016697883606, "train_loss_llm": 0.0015829205513000488, "grad_norm": 0.07018408179283142, "global_step": 24771, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03591917082667351, "train_loss_bc": 0.03591039404273033, "train_loss_llm": 0.0008775591850280762, "grad_norm": 0.0908137708902359, "global_step": 24772, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03083556890487671, "train_loss_bc": 0.03083556890487671, "train_loss_llm": 0.0, "grad_norm": 0.035536009818315506, "global_step": 24773, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03336597606539726, "train_loss_bc": 0.03336597606539726, "train_loss_llm": 0.0, "grad_norm": 0.05917420610785484, "global_step": 24774, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03377222269773483, "train_loss_bc": 0.03377222269773483, "train_loss_llm": 0.0, "grad_norm": 0.07080845534801483, "global_step": 24775, "epoch": 688, "lr": 0.001} +{"train_loss": 0.028640279546380043, "train_loss_bc": 0.028640279546380043, "train_loss_llm": 0.0, "grad_norm": 0.07983977347612381, "global_step": 24776, "epoch": 688, "lr": 0.001} +{"train_loss": 0.030491823330521584, "train_loss_bc": 0.030491823330521584, "train_loss_llm": 0.0, "grad_norm": 0.04130541533231735, "global_step": 24777, "epoch": 688, "lr": 0.001} +{"train_loss": 0.030541984364390373, "train_loss_bc": 0.030537467449903488, "train_loss_llm": 0.00045171380043029785, "grad_norm": 0.05971982330083847, "global_step": 24778, "epoch": 688, "lr": 0.001} +{"train_loss": 0.028586706146597862, "train_loss_bc": 0.028586706146597862, "train_loss_llm": 0.0, "grad_norm": 0.08516020327806473, "global_step": 24779, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03489203751087189, "train_loss_bc": 0.03489203751087189, "train_loss_llm": 0.0, "grad_norm": 0.10234712809324265, "global_step": 24780, "epoch": 688, "lr": 0.001} +{"train_loss": 0.030812153592705727, "train_loss_bc": 0.030812153592705727, "train_loss_llm": 0.0, "grad_norm": 0.03975910693407059, "global_step": 24781, "epoch": 688, "lr": 0.001} +{"train_loss": 0.030278950929641724, "train_loss_bc": 0.030278950929641724, "train_loss_llm": 0.0, "grad_norm": 0.06773403286933899, "global_step": 24782, "epoch": 688, "lr": 0.001} +{"train_loss": 0.034981075674295425, "train_loss_bc": 0.034976206719875336, "train_loss_llm": 0.0004869997501373291, "grad_norm": 0.08407758176326752, "global_step": 24783, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03282870724797249, "train_loss_bc": 0.03282870724797249, "train_loss_llm": 0.0, "grad_norm": 0.11790014803409576, "global_step": 24784, "epoch": 688, "lr": 0.001} +{"train_loss": 0.02815353311598301, "train_loss_bc": 0.028141314163804054, "train_loss_llm": 0.001221984624862671, "grad_norm": 0.03497505187988281, "global_step": 24785, "epoch": 688, "lr": 0.001} +{"train_loss": 0.022795788943767548, "train_loss_bc": 0.022793404757976532, "train_loss_llm": 0.0002384483814239502, "grad_norm": 0.05044136196374893, "global_step": 24786, "epoch": 688, "lr": 0.001} +{"train_loss": 0.027806008234620094, "train_loss_bc": 0.027806008234620094, "train_loss_llm": 0.0, "grad_norm": 0.06555184721946716, "global_step": 24787, "epoch": 688, "lr": 0.001} +{"train_loss": 0.02730521745979786, "train_loss_bc": 0.02730521745979786, "train_loss_llm": 0.0, "grad_norm": 0.08253191411495209, "global_step": 24788, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03287432715296745, "train_loss_bc": 0.03287432715296745, "train_loss_llm": 0.0, "grad_norm": 0.03977304324507713, "global_step": 24789, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03269895538687706, "train_loss_bc": 0.03269895538687706, "train_loss_llm": 0.0, "grad_norm": 0.055031225085258484, "global_step": 24790, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03244080767035484, "train_loss_bc": 0.03244080767035484, "train_loss_llm": 0.0, "grad_norm": 0.07327151298522949, "global_step": 24791, "epoch": 688, "lr": 0.001} +{"train_loss": 0.033360205590724945, "train_loss_bc": 0.03335871174931526, "train_loss_llm": 0.00014930963516235352, "grad_norm": 0.08712118119001389, "global_step": 24792, "epoch": 688, "lr": 0.001} +{"train_loss": 0.026888152584433556, "train_loss_bc": 0.026888152584433556, "train_loss_llm": 0.0, "grad_norm": 0.027484886348247528, "global_step": 24793, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03088836371898651, "train_loss_bc": 0.03088836371898651, "train_loss_llm": 0.0, "grad_norm": 0.0441298745572567, "global_step": 24794, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03534761443734169, "train_loss_bc": 0.03534761443734169, "train_loss_llm": 0.0, "grad_norm": 0.05814559757709503, "global_step": 24795, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03732937201857567, "train_loss_bc": 0.03732937201857567, "train_loss_llm": 0.0, "grad_norm": 0.07315675169229507, "global_step": 24796, "epoch": 688, "lr": 0.001} +{"train_loss": 0.029848871752619743, "train_loss_bc": 0.029848871752619743, "train_loss_llm": 0.0, "grad_norm": 0.029085703194141388, "global_step": 24797, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03279750794172287, "train_loss_bc": 0.032791830599308014, "train_loss_llm": 0.0005675852298736572, "grad_norm": 0.050312917679548264, "global_step": 24798, "epoch": 688, "lr": 0.001} +{"train_loss": 0.0285349041223526, "train_loss_bc": 0.0285349041223526, "train_loss_llm": 0.0, "grad_norm": 0.0696222186088562, "global_step": 24799, "epoch": 688, "lr": 0.001} +{"train_loss": 0.031254880130290985, "train_loss_bc": 0.031247932463884354, "train_loss_llm": 0.0006946921348571777, "grad_norm": 0.09122225642204285, "global_step": 24800, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03313908353447914, "train_loss_bc": 0.03313908353447914, "train_loss_llm": 0.0, "grad_norm": 0.0472278967499733, "global_step": 24801, "epoch": 688, "lr": 0.001} +{"train_loss": 0.033381856977939606, "train_loss_bc": 0.033381856977939606, "train_loss_llm": 0.0, "grad_norm": 0.07449131458997726, "global_step": 24802, "epoch": 688, "lr": 0.001} +{"train_loss": 0.03181442949506971, "train_loss_bc": 0.042002495378255844, "train_loss_llm": 0.0, "grad_norm": 0.08937633037567139, "global_step": 24803, "epoch": 688, "lr": 0.001, "val_loss": 0.08927616477012634} +{"train_loss": 0.027662279084324837, "train_loss_bc": 0.027662279084324837, "train_loss_llm": 0.0, "grad_norm": 0.10748820006847382, "global_step": 24804, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03176756575703621, "train_loss_bc": 0.03175976872444153, "train_loss_llm": 0.0007796287536621094, "grad_norm": 0.03418155014514923, "global_step": 24805, "epoch": 689, "lr": 0.001} +{"train_loss": 0.032610103487968445, "train_loss_bc": 0.032610103487968445, "train_loss_llm": 0.0, "grad_norm": 0.05709502100944519, "global_step": 24806, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03535570576786995, "train_loss_bc": 0.03535030782222748, "train_loss_llm": 0.0005398988723754883, "grad_norm": 0.07209325581789017, "global_step": 24807, "epoch": 689, "lr": 0.001} +{"train_loss": 0.030197713524103165, "train_loss_bc": 0.030191391706466675, "train_loss_llm": 0.0006322562694549561, "grad_norm": 0.10437420755624771, "global_step": 24808, "epoch": 689, "lr": 0.001} +{"train_loss": 0.030479947105050087, "train_loss_bc": 0.030479947105050087, "train_loss_llm": 0.0, "grad_norm": 0.03249146044254303, "global_step": 24809, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03195858746767044, "train_loss_bc": 0.03195858746767044, "train_loss_llm": 0.0, "grad_norm": 0.04975070059299469, "global_step": 24810, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03572624549269676, "train_loss_bc": 0.035719309002161026, "train_loss_llm": 0.0006937384605407715, "grad_norm": 0.05993643030524254, "global_step": 24811, "epoch": 689, "lr": 0.001} +{"train_loss": 0.030090104788541794, "train_loss_bc": 0.030090104788541794, "train_loss_llm": 0.0, "grad_norm": 0.06789898872375488, "global_step": 24812, "epoch": 689, "lr": 0.001} +{"train_loss": 0.02775469794869423, "train_loss_bc": 0.02775469794869423, "train_loss_llm": 0.0, "grad_norm": 0.030018243938684464, "global_step": 24813, "epoch": 689, "lr": 0.001} +{"train_loss": 0.037021707743406296, "train_loss_bc": 0.037021707743406296, "train_loss_llm": 0.0, "grad_norm": 0.060958050191402435, "global_step": 24814, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03503715246915817, "train_loss_bc": 0.03503715246915817, "train_loss_llm": 0.0, "grad_norm": 0.07453597337007523, "global_step": 24815, "epoch": 689, "lr": 0.001} +{"train_loss": 0.033800557255744934, "train_loss_bc": 0.033800557255744934, "train_loss_llm": 0.0, "grad_norm": 0.09830982983112335, "global_step": 24816, "epoch": 689, "lr": 0.001} +{"train_loss": 0.031386617571115494, "train_loss_bc": 0.031386617571115494, "train_loss_llm": 0.0, "grad_norm": 0.031745269894599915, "global_step": 24817, "epoch": 689, "lr": 0.001} +{"train_loss": 0.0369112603366375, "train_loss_bc": 0.03690759837627411, "train_loss_llm": 0.0003662109375, "grad_norm": 0.06033315509557724, "global_step": 24818, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03216969221830368, "train_loss_bc": 0.03216969221830368, "train_loss_llm": 0.0, "grad_norm": 0.08471833914518356, "global_step": 24819, "epoch": 689, "lr": 0.001} +{"train_loss": 0.0297438595443964, "train_loss_bc": 0.0297438595443964, "train_loss_llm": 0.0, "grad_norm": 0.09871161729097366, "global_step": 24820, "epoch": 689, "lr": 0.001} +{"train_loss": 0.029702916741371155, "train_loss_bc": 0.029702916741371155, "train_loss_llm": 0.0, "grad_norm": 0.036928433924913406, "global_step": 24821, "epoch": 689, "lr": 0.001} +{"train_loss": 0.0350700318813324, "train_loss_bc": 0.0350700318813324, "train_loss_llm": 0.0, "grad_norm": 0.061177946627140045, "global_step": 24822, "epoch": 689, "lr": 0.001} +{"train_loss": 0.026951495558023453, "train_loss_bc": 0.026938792318105698, "train_loss_llm": 0.0012703537940979004, "grad_norm": 0.08174742013216019, "global_step": 24823, "epoch": 689, "lr": 0.001} +{"train_loss": 0.028241349384188652, "train_loss_bc": 0.028241349384188652, "train_loss_llm": 0.0, "grad_norm": 0.1126689538359642, "global_step": 24824, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03714551776647568, "train_loss_bc": 0.03714551776647568, "train_loss_llm": 0.0, "grad_norm": 0.035910554230213165, "global_step": 24825, "epoch": 689, "lr": 0.001} +{"train_loss": 0.029191428795456886, "train_loss_bc": 0.029191428795456886, "train_loss_llm": 0.0, "grad_norm": 0.050822533667087555, "global_step": 24826, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03636244311928749, "train_loss_bc": 0.03636244311928749, "train_loss_llm": 0.0, "grad_norm": 0.07840150594711304, "global_step": 24827, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03523240610957146, "train_loss_bc": 0.03523240610957146, "train_loss_llm": 0.0, "grad_norm": 0.08553317189216614, "global_step": 24828, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03229324519634247, "train_loss_bc": 0.03229053318500519, "train_loss_llm": 0.00027111172676086426, "grad_norm": 0.026999471709132195, "global_step": 24829, "epoch": 689, "lr": 0.001} +{"train_loss": 0.02808079868555069, "train_loss_bc": 0.02808079868555069, "train_loss_llm": 0.0, "grad_norm": 0.045406755059957504, "global_step": 24830, "epoch": 689, "lr": 0.001} +{"train_loss": 0.027127377688884735, "train_loss_bc": 0.027116093784570694, "train_loss_llm": 0.001128464937210083, "grad_norm": 0.06522127985954285, "global_step": 24831, "epoch": 689, "lr": 0.001} +{"train_loss": 0.036442797631025314, "train_loss_bc": 0.036442797631025314, "train_loss_llm": 0.0, "grad_norm": 0.09313060343265533, "global_step": 24832, "epoch": 689, "lr": 0.001} +{"train_loss": 0.02426134981215, "train_loss_bc": 0.024255873635411263, "train_loss_llm": 0.0005475282669067383, "grad_norm": 0.03023880533874035, "global_step": 24833, "epoch": 689, "lr": 0.001} +{"train_loss": 0.026880977675318718, "train_loss_bc": 0.026880977675318718, "train_loss_llm": 0.0, "grad_norm": 0.04430954158306122, "global_step": 24834, "epoch": 689, "lr": 0.001} +{"train_loss": 0.030470702797174454, "train_loss_bc": 0.030470702797174454, "train_loss_llm": 0.0, "grad_norm": 0.06453566253185272, "global_step": 24835, "epoch": 689, "lr": 0.001} +{"train_loss": 0.03323003649711609, "train_loss_bc": 0.03322211652994156, "train_loss_llm": 0.0007921755313873291, "grad_norm": 0.09342886507511139, "global_step": 24836, "epoch": 689, "lr": 0.001} +{"train_loss": 0.028989385813474655, "train_loss_bc": 0.028989385813474655, "train_loss_llm": 0.0, "grad_norm": 0.0268530435860157, "global_step": 24837, "epoch": 689, "lr": 0.001} +{"train_loss": 0.030412251129746437, "train_loss_bc": 0.030412251129746437, "train_loss_llm": 0.0, "grad_norm": 0.04328206926584244, "global_step": 24838, "epoch": 689, "lr": 0.001} +{"train_loss": 0.0315404094548689, "train_loss_bc": 0.02969443053007126, "train_loss_llm": 0.0, "grad_norm": 0.07805778831243515, "global_step": 24839, "epoch": 689, "lr": 0.001, "val_loss": 0.0898638591170311} +{"train_loss": 0.02974719926714897, "train_loss_bc": 0.029745664447546005, "train_loss_llm": 0.00015342235565185547, "grad_norm": 0.10076597332954407, "global_step": 24840, "epoch": 690, "lr": 0.001} +{"train_loss": 0.028859302401542664, "train_loss_bc": 0.028859302401542664, "train_loss_llm": 0.0, "grad_norm": 0.02824360877275467, "global_step": 24841, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03182930499315262, "train_loss_bc": 0.03182930499315262, "train_loss_llm": 0.0, "grad_norm": 0.041954148560762405, "global_step": 24842, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03511080890893936, "train_loss_bc": 0.03511080890893936, "train_loss_llm": 0.0, "grad_norm": 0.06529471278190613, "global_step": 24843, "epoch": 690, "lr": 0.001} +{"train_loss": 0.030561992898583412, "train_loss_bc": 0.030556509271264076, "train_loss_llm": 0.0005483925342559814, "grad_norm": 0.08341457694768906, "global_step": 24844, "epoch": 690, "lr": 0.001} +{"train_loss": 0.026089338585734367, "train_loss_bc": 0.026089338585734367, "train_loss_llm": 0.0, "grad_norm": 0.03196375072002411, "global_step": 24845, "epoch": 690, "lr": 0.001} +{"train_loss": 0.029890503734350204, "train_loss_bc": 0.029890503734350204, "train_loss_llm": 0.0, "grad_norm": 0.0518387034535408, "global_step": 24846, "epoch": 690, "lr": 0.001} +{"train_loss": 0.026345353573560715, "train_loss_bc": 0.026345353573560715, "train_loss_llm": 0.0, "grad_norm": 0.08230031281709671, "global_step": 24847, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03640342876315117, "train_loss_bc": 0.03640342876315117, "train_loss_llm": 0.0, "grad_norm": 0.11739031225442886, "global_step": 24848, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03324497863650322, "train_loss_bc": 0.03324497863650322, "train_loss_llm": 0.0, "grad_norm": 0.03353887051343918, "global_step": 24849, "epoch": 690, "lr": 0.001} +{"train_loss": 0.025788376107811928, "train_loss_bc": 0.025788376107811928, "train_loss_llm": 0.0, "grad_norm": 0.05655152350664139, "global_step": 24850, "epoch": 690, "lr": 0.001} +{"train_loss": 0.033461425453424454, "train_loss_bc": 0.03345351293683052, "train_loss_llm": 0.0007913708686828613, "grad_norm": 0.06564222276210785, "global_step": 24851, "epoch": 690, "lr": 0.001} +{"train_loss": 0.031825751066207886, "train_loss_bc": 0.031825751066207886, "train_loss_llm": 0.0, "grad_norm": 0.07703866064548492, "global_step": 24852, "epoch": 690, "lr": 0.001} +{"train_loss": 0.02532229945063591, "train_loss_bc": 0.02531309239566326, "train_loss_llm": 0.0009207427501678467, "grad_norm": 0.0290005411952734, "global_step": 24853, "epoch": 690, "lr": 0.001} +{"train_loss": 0.035981856286525726, "train_loss_bc": 0.035981856286525726, "train_loss_llm": 0.0, "grad_norm": 0.0476502887904644, "global_step": 24854, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03545692190527916, "train_loss_bc": 0.03545692190527916, "train_loss_llm": 0.0, "grad_norm": 0.061092671006917953, "global_step": 24855, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03229482099413872, "train_loss_bc": 0.03229482099413872, "train_loss_llm": 0.0, "grad_norm": 0.07578030228614807, "global_step": 24856, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03440461307764053, "train_loss_bc": 0.03440461307764053, "train_loss_llm": 0.0, "grad_norm": 0.029941868036985397, "global_step": 24857, "epoch": 690, "lr": 0.001} +{"train_loss": 0.0337548702955246, "train_loss_bc": 0.0337548702955246, "train_loss_llm": 0.0, "grad_norm": 0.04909774288535118, "global_step": 24858, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03306734561920166, "train_loss_bc": 0.03306734561920166, "train_loss_llm": 0.0, "grad_norm": 0.07021328061819077, "global_step": 24859, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03135199099779129, "train_loss_bc": 0.03135199099779129, "train_loss_llm": 0.0, "grad_norm": 0.08683669567108154, "global_step": 24860, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03430716693401337, "train_loss_bc": 0.03430716693401337, "train_loss_llm": 0.0, "grad_norm": 0.03474871441721916, "global_step": 24861, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03207842633128166, "train_loss_bc": 0.03207842633128166, "train_loss_llm": 0.0, "grad_norm": 0.05817320942878723, "global_step": 24862, "epoch": 690, "lr": 0.001} +{"train_loss": 0.02894781343638897, "train_loss_bc": 0.02894781343638897, "train_loss_llm": 0.0, "grad_norm": 0.07927821576595306, "global_step": 24863, "epoch": 690, "lr": 0.001} +{"train_loss": 0.031521450728178024, "train_loss_bc": 0.03150526061654091, "train_loss_llm": 0.0016188621520996094, "grad_norm": 0.11562492698431015, "global_step": 24864, "epoch": 690, "lr": 0.001} +{"train_loss": 0.035869013518095016, "train_loss_bc": 0.035869013518095016, "train_loss_llm": 0.0, "grad_norm": 0.043653372675180435, "global_step": 24865, "epoch": 690, "lr": 0.001} +{"train_loss": 0.024591093882918358, "train_loss_bc": 0.024591093882918358, "train_loss_llm": 0.0, "grad_norm": 0.05732375383377075, "global_step": 24866, "epoch": 690, "lr": 0.001} +{"train_loss": 0.026705626398324966, "train_loss_bc": 0.026705626398324966, "train_loss_llm": 0.0, "grad_norm": 0.07720116525888443, "global_step": 24867, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03302881121635437, "train_loss_bc": 0.03299752622842789, "train_loss_llm": 0.003128528594970703, "grad_norm": 0.0867496207356453, "global_step": 24868, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03346069157123566, "train_loss_bc": 0.03346069157123566, "train_loss_llm": 0.0, "grad_norm": 0.030874839052557945, "global_step": 24869, "epoch": 690, "lr": 0.001} +{"train_loss": 0.032795339822769165, "train_loss_bc": 0.03279382362961769, "train_loss_llm": 0.00015154480934143066, "grad_norm": 0.04535643011331558, "global_step": 24870, "epoch": 690, "lr": 0.001} +{"train_loss": 0.030681857839226723, "train_loss_bc": 0.030681857839226723, "train_loss_llm": 0.0, "grad_norm": 0.0629318356513977, "global_step": 24871, "epoch": 690, "lr": 0.001} +{"train_loss": 0.032425008714199066, "train_loss_bc": 0.032425008714199066, "train_loss_llm": 0.0, "grad_norm": 0.08651971817016602, "global_step": 24872, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03505025804042816, "train_loss_bc": 0.0350382998585701, "train_loss_llm": 0.0011958479881286621, "grad_norm": 0.03428448736667633, "global_step": 24873, "epoch": 690, "lr": 0.001} +{"train_loss": 0.03242815285921097, "train_loss_bc": 0.032411109656095505, "train_loss_llm": 0.0017043650150299072, "grad_norm": 0.05477684363722801, "global_step": 24874, "epoch": 690, "lr": 0.001} +{"train_loss": 0.0316485275203983, "train_loss_bc": 0.03466379642486572, "train_loss_llm": 0.0, "grad_norm": 0.08535110950469971, "global_step": 24875, "epoch": 690, "lr": 0.001, "train/cumulative_reward": 113.37090863580201, "train/mean_score": 15.824119123492949, "train/success_rate": 0.3, "test/cumulative_reward": 34.832731089430496, "test/mean_score": 7.444156166153731, "test/success_rate": 0.2, "val_loss": 0.08778470754623413, "train_action_mse_error": 0.03467576578259468} +{"train_loss": 0.03279329091310501, "train_loss_bc": 0.03277589753270149, "train_loss_llm": 0.0017392635345458984, "grad_norm": 0.10106867551803589, "global_step": 24876, "epoch": 691, "lr": 0.001} +{"train_loss": 0.02145913429558277, "train_loss_bc": 0.02145913429558277, "train_loss_llm": 0.0, "grad_norm": 0.026134172454476357, "global_step": 24877, "epoch": 691, "lr": 0.001} +{"train_loss": 0.0286638792604208, "train_loss_bc": 0.0286638792604208, "train_loss_llm": 0.0, "grad_norm": 0.048202112317085266, "global_step": 24878, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03292752802371979, "train_loss_bc": 0.03292752802371979, "train_loss_llm": 0.0, "grad_norm": 0.05986684933304787, "global_step": 24879, "epoch": 691, "lr": 0.001} +{"train_loss": 0.032609257847070694, "train_loss_bc": 0.0325908288359642, "train_loss_llm": 0.0018429458141326904, "grad_norm": 0.08259772509336472, "global_step": 24880, "epoch": 691, "lr": 0.001} +{"train_loss": 0.0367136113345623, "train_loss_bc": 0.0367136113345623, "train_loss_llm": 0.0, "grad_norm": 0.040362607687711716, "global_step": 24881, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03611993044614792, "train_loss_bc": 0.036113958805799484, "train_loss_llm": 0.0005971789360046387, "grad_norm": 0.06463708728551865, "global_step": 24882, "epoch": 691, "lr": 0.001} +{"train_loss": 0.028238581493496895, "train_loss_bc": 0.028222301974892616, "train_loss_llm": 0.0016279816627502441, "grad_norm": 0.09067711979150772, "global_step": 24883, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03055817447602749, "train_loss_bc": 0.030552681535482407, "train_loss_llm": 0.0005492866039276123, "grad_norm": 0.10748021304607391, "global_step": 24884, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03610773757100105, "train_loss_bc": 0.03610773757100105, "train_loss_llm": 0.0, "grad_norm": 0.058396030217409134, "global_step": 24885, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03149854391813278, "train_loss_bc": 0.03148691728711128, "train_loss_llm": 0.0011625587940216064, "grad_norm": 0.08978656679391861, "global_step": 24886, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03065062128007412, "train_loss_bc": 0.03065062128007412, "train_loss_llm": 0.0, "grad_norm": 0.11517146229743958, "global_step": 24887, "epoch": 691, "lr": 0.001} +{"train_loss": 0.026362277567386627, "train_loss_bc": 0.026362277567386627, "train_loss_llm": 0.0, "grad_norm": 0.1433744728565216, "global_step": 24888, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03940306603908539, "train_loss_bc": 0.03940306603908539, "train_loss_llm": 0.0, "grad_norm": 0.03432705998420715, "global_step": 24889, "epoch": 691, "lr": 0.001} +{"train_loss": 0.030572786927223206, "train_loss_bc": 0.030563170090317726, "train_loss_llm": 0.00096169114112854, "grad_norm": 0.05236503481864929, "global_step": 24890, "epoch": 691, "lr": 0.001} +{"train_loss": 0.031050678342580795, "train_loss_bc": 0.031050678342580795, "train_loss_llm": 0.0, "grad_norm": 0.07470899075269699, "global_step": 24891, "epoch": 691, "lr": 0.001} +{"train_loss": 0.031103871762752533, "train_loss_bc": 0.031103871762752533, "train_loss_llm": 0.0, "grad_norm": 0.09058007597923279, "global_step": 24892, "epoch": 691, "lr": 0.001} +{"train_loss": 0.025134362280368805, "train_loss_bc": 0.025134362280368805, "train_loss_llm": 0.0, "grad_norm": 0.027662700042128563, "global_step": 24893, "epoch": 691, "lr": 0.001} +{"train_loss": 0.026304759085178375, "train_loss_bc": 0.026304759085178375, "train_loss_llm": 0.0, "grad_norm": 0.043263159692287445, "global_step": 24894, "epoch": 691, "lr": 0.001} +{"train_loss": 0.031875044107437134, "train_loss_bc": 0.031875044107437134, "train_loss_llm": 0.0, "grad_norm": 0.060246553272008896, "global_step": 24895, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03635846823453903, "train_loss_bc": 0.03635846823453903, "train_loss_llm": 0.0, "grad_norm": 0.08217962831258774, "global_step": 24896, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03302702307701111, "train_loss_bc": 0.03302702307701111, "train_loss_llm": 0.0, "grad_norm": 0.03732902184128761, "global_step": 24897, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03268415853381157, "train_loss_bc": 0.03267728164792061, "train_loss_llm": 0.0006875693798065186, "grad_norm": 0.07046693563461304, "global_step": 24898, "epoch": 691, "lr": 0.001} +{"train_loss": 0.0312640555202961, "train_loss_bc": 0.0312640555202961, "train_loss_llm": 0.0, "grad_norm": 0.10638833791017532, "global_step": 24899, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03409373015165329, "train_loss_bc": 0.03409373015165329, "train_loss_llm": 0.0, "grad_norm": 0.12497984617948532, "global_step": 24900, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03143671527504921, "train_loss_bc": 0.031432587653398514, "train_loss_llm": 0.0004127323627471924, "grad_norm": 0.03935150429606438, "global_step": 24901, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03369150683283806, "train_loss_bc": 0.03369150683283806, "train_loss_llm": 0.0, "grad_norm": 0.058007534593343735, "global_step": 24902, "epoch": 691, "lr": 0.001} +{"train_loss": 0.029119238257408142, "train_loss_bc": 0.029119238257408142, "train_loss_llm": 0.0, "grad_norm": 0.08748841285705566, "global_step": 24903, "epoch": 691, "lr": 0.001} +{"train_loss": 0.032716214656829834, "train_loss_bc": 0.032716214656829834, "train_loss_llm": 0.0, "grad_norm": 0.10308592021465302, "global_step": 24904, "epoch": 691, "lr": 0.001} +{"train_loss": 0.032118577510118484, "train_loss_bc": 0.032118577510118484, "train_loss_llm": 0.0, "grad_norm": 0.041705675423145294, "global_step": 24905, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03550365939736366, "train_loss_bc": 0.03550365939736366, "train_loss_llm": 0.0, "grad_norm": 0.0657874122262001, "global_step": 24906, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03074732981622219, "train_loss_bc": 0.030743485316634178, "train_loss_llm": 0.0003845393657684326, "grad_norm": 0.0883345678448677, "global_step": 24907, "epoch": 691, "lr": 0.001} +{"train_loss": 0.02960224635899067, "train_loss_bc": 0.02960224635899067, "train_loss_llm": 0.0, "grad_norm": 0.10688374191522598, "global_step": 24908, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03383537381887436, "train_loss_bc": 0.03383537381887436, "train_loss_llm": 0.0, "grad_norm": 0.03703468665480614, "global_step": 24909, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03278731927275658, "train_loss_bc": 0.03277921676635742, "train_loss_llm": 0.0008102059364318848, "grad_norm": 0.055102623999118805, "global_step": 24910, "epoch": 691, "lr": 0.001} +{"train_loss": 0.03170026016111175, "train_loss_bc": 0.03207661211490631, "train_loss_llm": 0.0, "grad_norm": 0.07184705883264542, "global_step": 24911, "epoch": 691, "lr": 0.001, "val_loss": 0.08877807855606079} +{"train_loss": 0.028591899201273918, "train_loss_bc": 0.028591899201273918, "train_loss_llm": 0.0, "grad_norm": 0.08117852360010147, "global_step": 24912, "epoch": 692, "lr": 0.001} +{"train_loss": 0.031443558633327484, "train_loss_bc": 0.031443558633327484, "train_loss_llm": 0.0, "grad_norm": 0.036904510110616684, "global_step": 24913, "epoch": 692, "lr": 0.001} +{"train_loss": 0.02898288145661354, "train_loss_bc": 0.028977014124393463, "train_loss_llm": 0.0005868077278137207, "grad_norm": 0.0528961680829525, "global_step": 24914, "epoch": 692, "lr": 0.001} +{"train_loss": 0.033052220940589905, "train_loss_bc": 0.033048197627067566, "train_loss_llm": 0.0004023611545562744, "grad_norm": 0.06516170501708984, "global_step": 24915, "epoch": 692, "lr": 0.001} +{"train_loss": 0.027593161910772324, "train_loss_bc": 0.02756141498684883, "train_loss_llm": 0.0031747817993164062, "grad_norm": 0.07855217158794403, "global_step": 24916, "epoch": 692, "lr": 0.001} +{"train_loss": 0.029978200793266296, "train_loss_bc": 0.029978200793266296, "train_loss_llm": 0.0, "grad_norm": 0.027420081198215485, "global_step": 24917, "epoch": 692, "lr": 0.001} +{"train_loss": 0.02761325053870678, "train_loss_bc": 0.027605587616562843, "train_loss_llm": 0.0007663369178771973, "grad_norm": 0.04870907962322235, "global_step": 24918, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03209949657320976, "train_loss_bc": 0.03209949657320976, "train_loss_llm": 0.0, "grad_norm": 0.0651235580444336, "global_step": 24919, "epoch": 692, "lr": 0.001} +{"train_loss": 0.026372095569968224, "train_loss_bc": 0.026372095569968224, "train_loss_llm": 0.0, "grad_norm": 0.08611228317022324, "global_step": 24920, "epoch": 692, "lr": 0.001} +{"train_loss": 0.032060280442237854, "train_loss_bc": 0.032060280442237854, "train_loss_llm": 0.0, "grad_norm": 0.04391532763838768, "global_step": 24921, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03488323092460632, "train_loss_bc": 0.03488323092460632, "train_loss_llm": 0.0, "grad_norm": 0.06426363438367844, "global_step": 24922, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03401317074894905, "train_loss_bc": 0.03401317074894905, "train_loss_llm": 0.0, "grad_norm": 0.09279723465442657, "global_step": 24923, "epoch": 692, "lr": 0.001} +{"train_loss": 0.02672533504664898, "train_loss_bc": 0.02672533504664898, "train_loss_llm": 0.0, "grad_norm": 0.12357472628355026, "global_step": 24924, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03760966286063194, "train_loss_bc": 0.03759665787220001, "train_loss_llm": 0.0013006329536437988, "grad_norm": 0.04378761351108551, "global_step": 24925, "epoch": 692, "lr": 0.001} +{"train_loss": 0.02890348620712757, "train_loss_bc": 0.02890348620712757, "train_loss_llm": 0.0, "grad_norm": 0.06477434188127518, "global_step": 24926, "epoch": 692, "lr": 0.001} +{"train_loss": 0.030389370396733284, "train_loss_bc": 0.03037085570394993, "train_loss_llm": 0.0018513798713684082, "grad_norm": 0.07864946126937866, "global_step": 24927, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03149275854229927, "train_loss_bc": 0.03149275854229927, "train_loss_llm": 0.0, "grad_norm": 0.10522983223199844, "global_step": 24928, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03260935842990875, "train_loss_bc": 0.03260935842990875, "train_loss_llm": 0.0, "grad_norm": 0.029609395191073418, "global_step": 24929, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03182307258248329, "train_loss_bc": 0.03182307258248329, "train_loss_llm": 0.0, "grad_norm": 0.04856925830245018, "global_step": 24930, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03667277842760086, "train_loss_bc": 0.03667277842760086, "train_loss_llm": 0.0, "grad_norm": 0.07495100796222687, "global_step": 24931, "epoch": 692, "lr": 0.001} +{"train_loss": 0.028845971450209618, "train_loss_bc": 0.028845971450209618, "train_loss_llm": 0.0, "grad_norm": 0.09762008488178253, "global_step": 24932, "epoch": 692, "lr": 0.001} +{"train_loss": 0.032782670110464096, "train_loss_bc": 0.03277663141489029, "train_loss_llm": 0.0006038844585418701, "grad_norm": 0.03740844875574112, "global_step": 24933, "epoch": 692, "lr": 0.001} +{"train_loss": 0.027587220072746277, "train_loss_bc": 0.027587220072746277, "train_loss_llm": 0.0, "grad_norm": 0.05730945244431496, "global_step": 24934, "epoch": 692, "lr": 0.001} +{"train_loss": 0.038434948772192, "train_loss_bc": 0.038434948772192, "train_loss_llm": 0.0, "grad_norm": 0.08064909279346466, "global_step": 24935, "epoch": 692, "lr": 0.001} +{"train_loss": 0.036245305091142654, "train_loss_bc": 0.03624487295746803, "train_loss_llm": 4.330277442932129e-05, "grad_norm": 0.1034199669957161, "global_step": 24936, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03383568301796913, "train_loss_bc": 0.03383568301796913, "train_loss_llm": 0.0, "grad_norm": 0.030987365171313286, "global_step": 24937, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03330536559224129, "train_loss_bc": 0.03330536559224129, "train_loss_llm": 0.0, "grad_norm": 0.04818650335073471, "global_step": 24938, "epoch": 692, "lr": 0.001} +{"train_loss": 0.02836780995130539, "train_loss_bc": 0.02836102806031704, "train_loss_llm": 0.0006782114505767822, "grad_norm": 0.06421326100826263, "global_step": 24939, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03775437921285629, "train_loss_bc": 0.03775437921285629, "train_loss_llm": 0.0, "grad_norm": 0.0880894735455513, "global_step": 24940, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03159666061401367, "train_loss_bc": 0.031580567359924316, "train_loss_llm": 0.0016093850135803223, "grad_norm": 0.03288527578115463, "global_step": 24941, "epoch": 692, "lr": 0.001} +{"train_loss": 0.0357171855866909, "train_loss_bc": 0.035709843039512634, "train_loss_llm": 0.0007344186305999756, "grad_norm": 0.06114714965224266, "global_step": 24942, "epoch": 692, "lr": 0.001} +{"train_loss": 0.03155564144253731, "train_loss_bc": 0.03155564144253731, "train_loss_llm": 0.0, "grad_norm": 0.08458492159843445, "global_step": 24943, "epoch": 692, "lr": 0.001} +{"train_loss": 0.029890678822994232, "train_loss_bc": 0.029890678822994232, "train_loss_llm": 0.0, "grad_norm": 0.0997038409113884, "global_step": 24944, "epoch": 692, "lr": 0.001} +{"train_loss": 0.028399700298905373, "train_loss_bc": 0.028399700298905373, "train_loss_llm": 0.0, "grad_norm": 0.02859249897301197, "global_step": 24945, "epoch": 692, "lr": 0.001} +{"train_loss": 0.029886197298765182, "train_loss_bc": 0.029881387948989868, "train_loss_llm": 0.00048086047172546387, "grad_norm": 0.05570835992693901, "global_step": 24946, "epoch": 692, "lr": 0.001} +{"train_loss": 0.031612855278783374, "train_loss_bc": 0.030948102474212646, "train_loss_llm": 0.0, "grad_norm": 0.08458174765110016, "global_step": 24947, "epoch": 692, "lr": 0.001, "val_loss": 0.08860616385936737} +{"train_loss": 0.03266752138733864, "train_loss_bc": 0.032641734927892685, "train_loss_llm": 0.002578824758529663, "grad_norm": 0.10477486252784729, "global_step": 24948, "epoch": 693, "lr": 0.001} +{"train_loss": 0.034270018339157104, "train_loss_bc": 0.034270018339157104, "train_loss_llm": 0.0, "grad_norm": 0.03551693260669708, "global_step": 24949, "epoch": 693, "lr": 0.001} +{"train_loss": 0.028114810585975647, "train_loss_bc": 0.028114810585975647, "train_loss_llm": 0.0, "grad_norm": 0.05956375598907471, "global_step": 24950, "epoch": 693, "lr": 0.001} +{"train_loss": 0.034000128507614136, "train_loss_bc": 0.034000128507614136, "train_loss_llm": 0.0, "grad_norm": 0.09080910682678223, "global_step": 24951, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03378906100988388, "train_loss_bc": 0.03378906100988388, "train_loss_llm": 0.0, "grad_norm": 0.11188429594039917, "global_step": 24952, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03458908945322037, "train_loss_bc": 0.03458098694682121, "train_loss_llm": 0.0008103549480438232, "grad_norm": 0.038199856877326965, "global_step": 24953, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03106401115655899, "train_loss_bc": 0.03106401115655899, "train_loss_llm": 0.0, "grad_norm": 0.0687013491988182, "global_step": 24954, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03571430221199989, "train_loss_bc": 0.03571430221199989, "train_loss_llm": 0.0, "grad_norm": 0.10281793773174286, "global_step": 24955, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03412569686770439, "train_loss_bc": 0.03412569686770439, "train_loss_llm": 0.0, "grad_norm": 0.1346290558576584, "global_step": 24956, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03132687509059906, "train_loss_bc": 0.03132687509059906, "train_loss_llm": 0.0, "grad_norm": 0.04391801357269287, "global_step": 24957, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03341967612504959, "train_loss_bc": 0.03341967612504959, "train_loss_llm": 0.0, "grad_norm": 0.08206551522016525, "global_step": 24958, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03176990896463394, "train_loss_bc": 0.03176990896463394, "train_loss_llm": 0.0, "grad_norm": 0.11717464029788971, "global_step": 24959, "epoch": 693, "lr": 0.001} +{"train_loss": 0.02674979157745838, "train_loss_bc": 0.02674979157745838, "train_loss_llm": 0.0, "grad_norm": 0.14542636275291443, "global_step": 24960, "epoch": 693, "lr": 0.001} +{"train_loss": 0.026464812457561493, "train_loss_bc": 0.026464812457561493, "train_loss_llm": 0.0, "grad_norm": 0.027989333495497704, "global_step": 24961, "epoch": 693, "lr": 0.001} +{"train_loss": 0.027567191049456596, "train_loss_bc": 0.02755529247224331, "train_loss_llm": 0.0011899471282958984, "grad_norm": 0.06039825454354286, "global_step": 24962, "epoch": 693, "lr": 0.001} +{"train_loss": 0.026904890313744545, "train_loss_bc": 0.026885030791163445, "train_loss_llm": 0.001985877752304077, "grad_norm": 0.07368310540914536, "global_step": 24963, "epoch": 693, "lr": 0.001} +{"train_loss": 0.033484701067209244, "train_loss_bc": 0.033465396612882614, "train_loss_llm": 0.00193062424659729, "grad_norm": 0.10425693541765213, "global_step": 24964, "epoch": 693, "lr": 0.001} +{"train_loss": 0.030906908214092255, "train_loss_bc": 0.030906908214092255, "train_loss_llm": 0.0, "grad_norm": 0.037513647228479385, "global_step": 24965, "epoch": 693, "lr": 0.001} +{"train_loss": 0.0314132384955883, "train_loss_bc": 0.03140757605433464, "train_loss_llm": 0.0005661547183990479, "grad_norm": 0.07264655083417892, "global_step": 24966, "epoch": 693, "lr": 0.001} +{"train_loss": 0.028897790238261223, "train_loss_bc": 0.028897790238261223, "train_loss_llm": 0.0, "grad_norm": 0.09678826481103897, "global_step": 24967, "epoch": 693, "lr": 0.001} +{"train_loss": 0.033245474100112915, "train_loss_bc": 0.033245474100112915, "train_loss_llm": 0.0, "grad_norm": 0.12543661892414093, "global_step": 24968, "epoch": 693, "lr": 0.001} +{"train_loss": 0.02970125339925289, "train_loss_bc": 0.02970125339925289, "train_loss_llm": 0.0, "grad_norm": 0.04266786202788353, "global_step": 24969, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03312430530786514, "train_loss_bc": 0.03312430530786514, "train_loss_llm": 0.0, "grad_norm": 0.07369764149188995, "global_step": 24970, "epoch": 693, "lr": 0.001} +{"train_loss": 0.037881169468164444, "train_loss_bc": 0.037862248718738556, "train_loss_llm": 0.00189208984375, "grad_norm": 0.1163639947772026, "global_step": 24971, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03208760917186737, "train_loss_bc": 0.03208760917186737, "train_loss_llm": 0.0, "grad_norm": 0.13861727714538574, "global_step": 24972, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03321506083011627, "train_loss_bc": 0.03321506083011627, "train_loss_llm": 0.0, "grad_norm": 0.05281310901045799, "global_step": 24973, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03144938126206398, "train_loss_bc": 0.03144938126206398, "train_loss_llm": 0.0, "grad_norm": 0.08817490935325623, "global_step": 24974, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03332316130399704, "train_loss_bc": 0.03332316130399704, "train_loss_llm": 0.0, "grad_norm": 0.11905968189239502, "global_step": 24975, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03508950024843216, "train_loss_bc": 0.03508950024843216, "train_loss_llm": 0.0, "grad_norm": 0.15185265243053436, "global_step": 24976, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03278675675392151, "train_loss_bc": 0.03277093917131424, "train_loss_llm": 0.0015816688537597656, "grad_norm": 0.03307240083813667, "global_step": 24977, "epoch": 693, "lr": 0.001} +{"train_loss": 0.039351098239421844, "train_loss_bc": 0.039351098239421844, "train_loss_llm": 0.0, "grad_norm": 0.07844512164592743, "global_step": 24978, "epoch": 693, "lr": 0.001} +{"train_loss": 0.02720271795988083, "train_loss_bc": 0.02720271795988083, "train_loss_llm": 0.0, "grad_norm": 0.10863395780324936, "global_step": 24979, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03479556366801262, "train_loss_bc": 0.03479556366801262, "train_loss_llm": 0.0, "grad_norm": 0.1533578336238861, "global_step": 24980, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03678297623991966, "train_loss_bc": 0.03678297623991966, "train_loss_llm": 0.0, "grad_norm": 0.03184136748313904, "global_step": 24981, "epoch": 693, "lr": 0.001} +{"train_loss": 0.03284839168190956, "train_loss_bc": 0.03284839168190956, "train_loss_llm": 0.0, "grad_norm": 0.06263445317745209, "global_step": 24982, "epoch": 693, "lr": 0.001} +{"train_loss": 0.032327884549482, "train_loss_bc": 0.03366519510746002, "train_loss_llm": 0.0013804733753204346, "grad_norm": 0.09519711136817932, "global_step": 24983, "epoch": 693, "lr": 0.001, "val_loss": 0.0883660763502121} +{"train_loss": 0.03198264539241791, "train_loss_bc": 0.03198264539241791, "train_loss_llm": 0.0, "grad_norm": 0.123566634953022, "global_step": 24984, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03233005478978157, "train_loss_bc": 0.03233005478978157, "train_loss_llm": 0.0, "grad_norm": 0.03871386498212814, "global_step": 24985, "epoch": 694, "lr": 0.001} +{"train_loss": 0.032518789172172546, "train_loss_bc": 0.032518789172172546, "train_loss_llm": 0.0, "grad_norm": 0.06659526377916336, "global_step": 24986, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03191068768501282, "train_loss_bc": 0.03191068768501282, "train_loss_llm": 0.0, "grad_norm": 0.07289624214172363, "global_step": 24987, "epoch": 694, "lr": 0.001} +{"train_loss": 0.027261804789304733, "train_loss_bc": 0.027261804789304733, "train_loss_llm": 0.0, "grad_norm": 0.09251870959997177, "global_step": 24988, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03397263213992119, "train_loss_bc": 0.03397263213992119, "train_loss_llm": 0.0, "grad_norm": 0.029658488929271698, "global_step": 24989, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03559359535574913, "train_loss_bc": 0.03559359535574913, "train_loss_llm": 0.0, "grad_norm": 0.06351146101951599, "global_step": 24990, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03542691096663475, "train_loss_bc": 0.03542691096663475, "train_loss_llm": 0.0, "grad_norm": 0.08463142812252045, "global_step": 24991, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03131349757313728, "train_loss_bc": 0.03131349757313728, "train_loss_llm": 0.0, "grad_norm": 0.11973472684621811, "global_step": 24992, "epoch": 694, "lr": 0.001} +{"train_loss": 0.027794258669018745, "train_loss_bc": 0.027794258669018745, "train_loss_llm": 0.0, "grad_norm": 0.03894414007663727, "global_step": 24993, "epoch": 694, "lr": 0.001} +{"train_loss": 0.036567311733961105, "train_loss_bc": 0.036567311733961105, "train_loss_llm": 0.0, "grad_norm": 0.07131989300251007, "global_step": 24994, "epoch": 694, "lr": 0.001} +{"train_loss": 0.032966095954179764, "train_loss_bc": 0.032966095954179764, "train_loss_llm": 0.0, "grad_norm": 0.10207783430814743, "global_step": 24995, "epoch": 694, "lr": 0.001} +{"train_loss": 0.037441760301589966, "train_loss_bc": 0.037441760301589966, "train_loss_llm": 0.0, "grad_norm": 0.14576363563537598, "global_step": 24996, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03585363179445267, "train_loss_bc": 0.03585363179445267, "train_loss_llm": 0.0, "grad_norm": 0.03995155170559883, "global_step": 24997, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03276357427239418, "train_loss_bc": 0.03276357427239418, "train_loss_llm": 0.0, "grad_norm": 0.07695712894201279, "global_step": 24998, "epoch": 694, "lr": 0.001} +{"train_loss": 0.031429048627614975, "train_loss_bc": 0.031429048627614975, "train_loss_llm": 0.0, "grad_norm": 0.10596676915884018, "global_step": 24999, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03285090997815132, "train_loss_bc": 0.0328388474881649, "train_loss_llm": 0.0012063980102539062, "grad_norm": 0.13977892696857452, "global_step": 25000, "epoch": 694, "lr": 0.001} +{"train_loss": 0.028371546417474747, "train_loss_bc": 0.028371036052703857, "train_loss_llm": 5.1021575927734375e-05, "grad_norm": 0.03448588028550148, "global_step": 25001, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03559049591422081, "train_loss_bc": 0.03559049591422081, "train_loss_llm": 0.0, "grad_norm": 0.05400073900818825, "global_step": 25002, "epoch": 694, "lr": 0.001} +{"train_loss": 0.031175781041383743, "train_loss_bc": 0.031175781041383743, "train_loss_llm": 0.0, "grad_norm": 0.07428549975156784, "global_step": 25003, "epoch": 694, "lr": 0.001} +{"train_loss": 0.026885390281677246, "train_loss_bc": 0.026869036257267, "train_loss_llm": 0.0016354918479919434, "grad_norm": 0.09579812735319138, "global_step": 25004, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03122781775891781, "train_loss_bc": 0.031204789876937866, "train_loss_llm": 0.002302795648574829, "grad_norm": 0.03398597612977028, "global_step": 25005, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03627534210681915, "train_loss_bc": 0.03627346083521843, "train_loss_llm": 0.0001881420612335205, "grad_norm": 0.05367688462138176, "global_step": 25006, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03240673989057541, "train_loss_bc": 0.03240673989057541, "train_loss_llm": 0.0, "grad_norm": 0.0669502541422844, "global_step": 25007, "epoch": 694, "lr": 0.001} +{"train_loss": 0.02884078398346901, "train_loss_bc": 0.02884078398346901, "train_loss_llm": 0.0, "grad_norm": 0.08508264273405075, "global_step": 25008, "epoch": 694, "lr": 0.001} +{"train_loss": 0.029942546039819717, "train_loss_bc": 0.029942546039819717, "train_loss_llm": 0.0, "grad_norm": 0.02915376052260399, "global_step": 25009, "epoch": 694, "lr": 0.001} +{"train_loss": 0.029107198119163513, "train_loss_bc": 0.029107198119163513, "train_loss_llm": 0.0, "grad_norm": 0.05611952766776085, "global_step": 25010, "epoch": 694, "lr": 0.001} +{"train_loss": 0.036027904599905014, "train_loss_bc": 0.036027904599905014, "train_loss_llm": 0.0, "grad_norm": 0.08054452389478683, "global_step": 25011, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03289263695478439, "train_loss_bc": 0.032891467213630676, "train_loss_llm": 0.00011709332466125488, "grad_norm": 0.10772515088319778, "global_step": 25012, "epoch": 694, "lr": 0.001} +{"train_loss": 0.029705669730901718, "train_loss_bc": 0.029705669730901718, "train_loss_llm": 0.0, "grad_norm": 0.04464617744088173, "global_step": 25013, "epoch": 694, "lr": 0.001} +{"train_loss": 0.029952602460980415, "train_loss_bc": 0.029952602460980415, "train_loss_llm": 0.0, "grad_norm": 0.06784117966890335, "global_step": 25014, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03183046728372574, "train_loss_bc": 0.03183046728372574, "train_loss_llm": 0.0, "grad_norm": 0.1021617203950882, "global_step": 25015, "epoch": 694, "lr": 0.001} +{"train_loss": 0.031130127608776093, "train_loss_bc": 0.031108170747756958, "train_loss_llm": 0.0021957755088806152, "grad_norm": 0.1291494369506836, "global_step": 25016, "epoch": 694, "lr": 0.001} +{"train_loss": 0.028695380315184593, "train_loss_bc": 0.02868749387562275, "train_loss_llm": 0.0007886886596679688, "grad_norm": 0.03361323103308678, "global_step": 25017, "epoch": 694, "lr": 0.001} +{"train_loss": 0.032759957015514374, "train_loss_bc": 0.03272782638669014, "train_loss_llm": 0.003213047981262207, "grad_norm": 0.0671267881989479, "global_step": 25018, "epoch": 694, "lr": 0.001} +{"train_loss": 0.03197123162034485, "train_loss_bc": 0.028155595064163208, "train_loss_llm": 0.0013146400451660156, "grad_norm": 0.08851784467697144, "global_step": 25019, "epoch": 694, "lr": 0.001, "val_loss": 0.08934345841407776} +{"train_loss": 0.02873258851468563, "train_loss_bc": 0.02873258851468563, "train_loss_llm": 0.0, "grad_norm": 0.11221746355295181, "global_step": 25020, "epoch": 695, "lr": 0.001} +{"train_loss": 0.0261518731713295, "train_loss_bc": 0.026147829368710518, "train_loss_llm": 0.0004043877124786377, "grad_norm": 0.03394351899623871, "global_step": 25021, "epoch": 695, "lr": 0.001} +{"train_loss": 0.028054162859916687, "train_loss_bc": 0.028054162859916687, "train_loss_llm": 0.0, "grad_norm": 0.05772247910499573, "global_step": 25022, "epoch": 695, "lr": 0.001} +{"train_loss": 0.033967453986406326, "train_loss_bc": 0.033967453986406326, "train_loss_llm": 0.0, "grad_norm": 0.07738574594259262, "global_step": 25023, "epoch": 695, "lr": 0.001} +{"train_loss": 0.030958887189626694, "train_loss_bc": 0.030958887189626694, "train_loss_llm": 0.0, "grad_norm": 0.10583113133907318, "global_step": 25024, "epoch": 695, "lr": 0.001} +{"train_loss": 0.036660585552453995, "train_loss_bc": 0.036660585552453995, "train_loss_llm": 0.0, "grad_norm": 0.03465653955936432, "global_step": 25025, "epoch": 695, "lr": 0.001} +{"train_loss": 0.035591378808021545, "train_loss_bc": 0.035591378808021545, "train_loss_llm": 0.0, "grad_norm": 0.06113477423787117, "global_step": 25026, "epoch": 695, "lr": 0.001} +{"train_loss": 0.030223287642002106, "train_loss_bc": 0.030223287642002106, "train_loss_llm": 0.0, "grad_norm": 0.09104487299919128, "global_step": 25027, "epoch": 695, "lr": 0.001} +{"train_loss": 0.029009070247411728, "train_loss_bc": 0.029009070247411728, "train_loss_llm": 0.0, "grad_norm": 0.11050543189048767, "global_step": 25028, "epoch": 695, "lr": 0.001} +{"train_loss": 0.027643317356705666, "train_loss_bc": 0.027643317356705666, "train_loss_llm": 0.0, "grad_norm": 0.02773359790444374, "global_step": 25029, "epoch": 695, "lr": 0.001} +{"train_loss": 0.030746186152100563, "train_loss_bc": 0.030746186152100563, "train_loss_llm": 0.0, "grad_norm": 0.04738641530275345, "global_step": 25030, "epoch": 695, "lr": 0.001} +{"train_loss": 0.0300055593252182, "train_loss_bc": 0.03000463731586933, "train_loss_llm": 9.223818778991699e-05, "grad_norm": 0.06414443254470825, "global_step": 25031, "epoch": 695, "lr": 0.001} +{"train_loss": 0.033803053200244904, "train_loss_bc": 0.03379274532198906, "train_loss_llm": 0.001030951738357544, "grad_norm": 0.0948764979839325, "global_step": 25032, "epoch": 695, "lr": 0.001} +{"train_loss": 0.034980203956365585, "train_loss_bc": 0.034980203956365585, "train_loss_llm": 0.0, "grad_norm": 0.03502720594406128, "global_step": 25033, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03207856044173241, "train_loss_bc": 0.03207856044173241, "train_loss_llm": 0.0, "grad_norm": 0.0647067278623581, "global_step": 25034, "epoch": 695, "lr": 0.001} +{"train_loss": 0.036888934671878815, "train_loss_bc": 0.036888934671878815, "train_loss_llm": 0.0, "grad_norm": 0.09157738089561462, "global_step": 25035, "epoch": 695, "lr": 0.001} +{"train_loss": 0.033089529722929, "train_loss_bc": 0.033082034438848495, "train_loss_llm": 0.0007496178150177002, "grad_norm": 0.11322696506977081, "global_step": 25036, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03398527204990387, "train_loss_bc": 0.03398527204990387, "train_loss_llm": 0.0, "grad_norm": 0.029369181022047997, "global_step": 25037, "epoch": 695, "lr": 0.001} +{"train_loss": 0.030688468366861343, "train_loss_bc": 0.030688468366861343, "train_loss_llm": 0.0, "grad_norm": 0.053790099918842316, "global_step": 25038, "epoch": 695, "lr": 0.001} +{"train_loss": 0.035464636981487274, "train_loss_bc": 0.035464636981487274, "train_loss_llm": 0.0, "grad_norm": 0.07498064637184143, "global_step": 25039, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03334705904126167, "train_loss_bc": 0.03332515060901642, "train_loss_llm": 0.002190709114074707, "grad_norm": 0.10658722370862961, "global_step": 25040, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03000068850815296, "train_loss_bc": 0.03000068850815296, "train_loss_llm": 0.0, "grad_norm": 0.03207916393876076, "global_step": 25041, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03167575225234032, "train_loss_bc": 0.03167575225234032, "train_loss_llm": 0.0, "grad_norm": 0.05223040655255318, "global_step": 25042, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03194599598646164, "train_loss_bc": 0.03194599598646164, "train_loss_llm": 0.0, "grad_norm": 0.0724525973200798, "global_step": 25043, "epoch": 695, "lr": 0.001} +{"train_loss": 0.027776213362812996, "train_loss_bc": 0.027776142582297325, "train_loss_llm": 7.152557373046875e-06, "grad_norm": 0.09276486188173294, "global_step": 25044, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03067951090633869, "train_loss_bc": 0.03067951090633869, "train_loss_llm": 0.0, "grad_norm": 0.033720873296260834, "global_step": 25045, "epoch": 695, "lr": 0.001} +{"train_loss": 0.031323425471782684, "train_loss_bc": 0.031323425471782684, "train_loss_llm": 0.0, "grad_norm": 0.057184264063835144, "global_step": 25046, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03309909626841545, "train_loss_bc": 0.03309909626841545, "train_loss_llm": 0.0, "grad_norm": 0.0802055075764656, "global_step": 25047, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03522403538227081, "train_loss_bc": 0.03521673008799553, "train_loss_llm": 0.0007304847240447998, "grad_norm": 0.10071031749248505, "global_step": 25048, "epoch": 695, "lr": 0.001} +{"train_loss": 0.034628402441740036, "train_loss_bc": 0.03461805358529091, "train_loss_llm": 0.0010348856449127197, "grad_norm": 0.035785019397735596, "global_step": 25049, "epoch": 695, "lr": 0.001} +{"train_loss": 0.029611075296998024, "train_loss_bc": 0.029602207243442535, "train_loss_llm": 0.0008868873119354248, "grad_norm": 0.06694356352090836, "global_step": 25050, "epoch": 695, "lr": 0.001} +{"train_loss": 0.030504126101732254, "train_loss_bc": 0.030504126101732254, "train_loss_llm": 0.0, "grad_norm": 0.08313333243131638, "global_step": 25051, "epoch": 695, "lr": 0.001} +{"train_loss": 0.026588771492242813, "train_loss_bc": 0.026588771492242813, "train_loss_llm": 0.0, "grad_norm": 0.09778676927089691, "global_step": 25052, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03284497931599617, "train_loss_bc": 0.03284497931599617, "train_loss_llm": 0.0, "grad_norm": 0.03813461586833, "global_step": 25053, "epoch": 695, "lr": 0.001} +{"train_loss": 0.028723454102873802, "train_loss_bc": 0.028723454102873802, "train_loss_llm": 0.0, "grad_norm": 0.06210334971547127, "global_step": 25054, "epoch": 695, "lr": 0.001} +{"train_loss": 0.03189348915798797, "train_loss_bc": 0.041470013558864594, "train_loss_llm": 0.0, "grad_norm": 0.08922597765922546, "global_step": 25055, "epoch": 695, "lr": 0.001, "train/cumulative_reward": 52.26305757252068, "train/mean_score": 15.815002965480204, "train/success_rate": 0.4, "test/cumulative_reward": 42.44584916784766, "test/mean_score": 6.792371362773988, "test/success_rate": 0.14, "val_loss": 0.08947501331567764, "train_action_mse_error": 0.034388456493616104} +{"train_loss": 0.02534564584493637, "train_loss_bc": 0.02534564584493637, "train_loss_llm": 0.0, "grad_norm": 0.10874105244874954, "global_step": 25056, "epoch": 696, "lr": 0.001} +{"train_loss": 0.02778754197061062, "train_loss_bc": 0.02778754197061062, "train_loss_llm": 0.0, "grad_norm": 0.025823958218097687, "global_step": 25057, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03384914621710777, "train_loss_bc": 0.03384914621710777, "train_loss_llm": 0.0, "grad_norm": 0.044834550470113754, "global_step": 25058, "epoch": 696, "lr": 0.001} +{"train_loss": 0.02994687855243683, "train_loss_bc": 0.029914889484643936, "train_loss_llm": 0.0031988322734832764, "grad_norm": 0.06605101376771927, "global_step": 25059, "epoch": 696, "lr": 0.001} +{"train_loss": 0.02983577363193035, "train_loss_bc": 0.02983577363193035, "train_loss_llm": 0.0, "grad_norm": 0.07676585018634796, "global_step": 25060, "epoch": 696, "lr": 0.001} +{"train_loss": 0.026711050420999527, "train_loss_bc": 0.026711050420999527, "train_loss_llm": 0.0, "grad_norm": 0.030024468898773193, "global_step": 25061, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03404064476490021, "train_loss_bc": 0.03403977304697037, "train_loss_llm": 8.70823860168457e-05, "grad_norm": 0.04884745553135872, "global_step": 25062, "epoch": 696, "lr": 0.001} +{"train_loss": 0.028324691578745842, "train_loss_bc": 0.028324691578745842, "train_loss_llm": 0.0, "grad_norm": 0.06796741485595703, "global_step": 25063, "epoch": 696, "lr": 0.001} +{"train_loss": 0.0380241796374321, "train_loss_bc": 0.03800730034708977, "train_loss_llm": 0.001687765121459961, "grad_norm": 0.09343820810317993, "global_step": 25064, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03697403147816658, "train_loss_bc": 0.03697403147816658, "train_loss_llm": 0.0, "grad_norm": 0.03706448897719383, "global_step": 25065, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03121446818113327, "train_loss_bc": 0.03121446818113327, "train_loss_llm": 0.0, "grad_norm": 0.05131020396947861, "global_step": 25066, "epoch": 696, "lr": 0.001} +{"train_loss": 0.02814934216439724, "train_loss_bc": 0.02814362198114395, "train_loss_llm": 0.0005719363689422607, "grad_norm": 0.068679578602314, "global_step": 25067, "epoch": 696, "lr": 0.001} +{"train_loss": 0.02558930404484272, "train_loss_bc": 0.02558930404484272, "train_loss_llm": 0.0, "grad_norm": 0.08794679492712021, "global_step": 25068, "epoch": 696, "lr": 0.001} +{"train_loss": 0.035688720643520355, "train_loss_bc": 0.035688720643520355, "train_loss_llm": 0.0, "grad_norm": 0.027511732652783394, "global_step": 25069, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03213788941502571, "train_loss_bc": 0.03213788941502571, "train_loss_llm": 0.0, "grad_norm": 0.0499090738594532, "global_step": 25070, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03429558128118515, "train_loss_bc": 0.03429558128118515, "train_loss_llm": 0.0, "grad_norm": 0.0681978240609169, "global_step": 25071, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03396141529083252, "train_loss_bc": 0.03396141529083252, "train_loss_llm": 0.0, "grad_norm": 0.10081061720848083, "global_step": 25072, "epoch": 696, "lr": 0.001} +{"train_loss": 0.033248044550418854, "train_loss_bc": 0.033248044550418854, "train_loss_llm": 0.0, "grad_norm": 0.027890954166650772, "global_step": 25073, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03198027238249779, "train_loss_bc": 0.03198027238249779, "train_loss_llm": 0.0, "grad_norm": 0.054450396448373795, "global_step": 25074, "epoch": 696, "lr": 0.001} +{"train_loss": 0.021895265206694603, "train_loss_bc": 0.021895265206694603, "train_loss_llm": 0.0, "grad_norm": 0.07651769369840622, "global_step": 25075, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03458067774772644, "train_loss_bc": 0.03458067774772644, "train_loss_llm": 0.0, "grad_norm": 0.1024317517876625, "global_step": 25076, "epoch": 696, "lr": 0.001} +{"train_loss": 0.032152608036994934, "train_loss_bc": 0.032152608036994934, "train_loss_llm": 0.0, "grad_norm": 0.03270659223198891, "global_step": 25077, "epoch": 696, "lr": 0.001} +{"train_loss": 0.035823315382003784, "train_loss_bc": 0.035823315382003784, "train_loss_llm": 0.0, "grad_norm": 0.04832674190402031, "global_step": 25078, "epoch": 696, "lr": 0.001} +{"train_loss": 0.029406970366835594, "train_loss_bc": 0.02940329536795616, "train_loss_llm": 0.0003674328327178955, "grad_norm": 0.06738923490047455, "global_step": 25079, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03848268836736679, "train_loss_bc": 0.03848268836736679, "train_loss_llm": 0.0, "grad_norm": 0.08932828158140182, "global_step": 25080, "epoch": 696, "lr": 0.001} +{"train_loss": 0.028457392007112503, "train_loss_bc": 0.028456123545765877, "train_loss_llm": 0.00012692809104919434, "grad_norm": 0.027375707402825356, "global_step": 25081, "epoch": 696, "lr": 0.001} +{"train_loss": 0.024965649470686913, "train_loss_bc": 0.024955658242106438, "train_loss_llm": 0.000999152660369873, "grad_norm": 0.04380599781870842, "global_step": 25082, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03269165754318237, "train_loss_bc": 0.03269165754318237, "train_loss_llm": 0.0, "grad_norm": 0.06020677089691162, "global_step": 25083, "epoch": 696, "lr": 0.001} +{"train_loss": 0.036146290600299835, "train_loss_bc": 0.03613946959376335, "train_loss_llm": 0.0006820559501647949, "grad_norm": 0.0853784903883934, "global_step": 25084, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03417181596159935, "train_loss_bc": 0.03417181596159935, "train_loss_llm": 0.0, "grad_norm": 0.03294392675161362, "global_step": 25085, "epoch": 696, "lr": 0.001} +{"train_loss": 0.028184624388813972, "train_loss_bc": 0.028184624388813972, "train_loss_llm": 0.0, "grad_norm": 0.03971100226044655, "global_step": 25086, "epoch": 696, "lr": 0.001} +{"train_loss": 0.028410792350769043, "train_loss_bc": 0.028410792350769043, "train_loss_llm": 0.0, "grad_norm": 0.058756500482559204, "global_step": 25087, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03535642474889755, "train_loss_bc": 0.03535548225045204, "train_loss_llm": 9.414553642272949e-05, "grad_norm": 0.07949848473072052, "global_step": 25088, "epoch": 696, "lr": 0.001} +{"train_loss": 0.03198704868555069, "train_loss_bc": 0.03198704868555069, "train_loss_llm": 0.0, "grad_norm": 0.03941240534186363, "global_step": 25089, "epoch": 696, "lr": 0.001} +{"train_loss": 0.031977005302906036, "train_loss_bc": 0.03196307271718979, "train_loss_llm": 0.0013934075832366943, "grad_norm": 0.06018005311489105, "global_step": 25090, "epoch": 696, "lr": 0.001} +{"train_loss": 0.031538599325964846, "train_loss_bc": 0.03356602042913437, "train_loss_llm": 0.0028708279132843018, "grad_norm": 0.08260176330804825, "global_step": 25091, "epoch": 696, "lr": 0.001, "val_loss": 0.08839982748031616} +{"train_loss": 0.02850855328142643, "train_loss_bc": 0.028493938967585564, "train_loss_llm": 0.0014613568782806396, "grad_norm": 0.09915895015001297, "global_step": 25092, "epoch": 697, "lr": 0.001} +{"train_loss": 0.029677940532565117, "train_loss_bc": 0.029677940532565117, "train_loss_llm": 0.0, "grad_norm": 0.03126468509435654, "global_step": 25093, "epoch": 697, "lr": 0.001} +{"train_loss": 0.031119097024202347, "train_loss_bc": 0.031119097024202347, "train_loss_llm": 0.0, "grad_norm": 0.04992899298667908, "global_step": 25094, "epoch": 697, "lr": 0.001} +{"train_loss": 0.02860189601778984, "train_loss_bc": 0.02860189601778984, "train_loss_llm": 0.0, "grad_norm": 0.06968522071838379, "global_step": 25095, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03048316016793251, "train_loss_bc": 0.03048316016793251, "train_loss_llm": 0.0, "grad_norm": 0.09452582150697708, "global_step": 25096, "epoch": 697, "lr": 0.001} +{"train_loss": 0.033351995050907135, "train_loss_bc": 0.033351995050907135, "train_loss_llm": 0.0, "grad_norm": 0.03553440421819687, "global_step": 25097, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03651963919401169, "train_loss_bc": 0.036517348140478134, "train_loss_llm": 0.00022915005683898926, "grad_norm": 0.05390232428908348, "global_step": 25098, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03036108985543251, "train_loss_bc": 0.030339006334543228, "train_loss_llm": 0.0022084414958953857, "grad_norm": 0.07492290437221527, "global_step": 25099, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03908567503094673, "train_loss_bc": 0.03908567503094673, "train_loss_llm": 0.0, "grad_norm": 0.09499727189540863, "global_step": 25100, "epoch": 697, "lr": 0.001} +{"train_loss": 0.030609017238020897, "train_loss_bc": 0.030609017238020897, "train_loss_llm": 0.0, "grad_norm": 0.037716254591941833, "global_step": 25101, "epoch": 697, "lr": 0.001} +{"train_loss": 0.028151297941803932, "train_loss_bc": 0.028151297941803932, "train_loss_llm": 0.0, "grad_norm": 0.059777289628982544, "global_step": 25102, "epoch": 697, "lr": 0.001} +{"train_loss": 0.030052898451685905, "train_loss_bc": 0.030052898451685905, "train_loss_llm": 0.0, "grad_norm": 0.07804954051971436, "global_step": 25103, "epoch": 697, "lr": 0.001} +{"train_loss": 0.027173645794391632, "train_loss_bc": 0.027154002338647842, "train_loss_llm": 0.0019644200801849365, "grad_norm": 0.09085237979888916, "global_step": 25104, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03192388638854027, "train_loss_bc": 0.03192267194390297, "train_loss_llm": 0.00012126564979553223, "grad_norm": 0.032827962189912796, "global_step": 25105, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03282167762517929, "train_loss_bc": 0.03281562030315399, "train_loss_llm": 0.0006055831909179688, "grad_norm": 0.052012231200933456, "global_step": 25106, "epoch": 697, "lr": 0.001} +{"train_loss": 0.02516847290098667, "train_loss_bc": 0.025153042748570442, "train_loss_llm": 0.0015430152416229248, "grad_norm": 0.05790676921606064, "global_step": 25107, "epoch": 697, "lr": 0.001} +{"train_loss": 0.028611542657017708, "train_loss_bc": 0.028611542657017708, "train_loss_llm": 0.0, "grad_norm": 0.071259506046772, "global_step": 25108, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03220568224787712, "train_loss_bc": 0.03220568224787712, "train_loss_llm": 0.0, "grad_norm": 0.03298371657729149, "global_step": 25109, "epoch": 697, "lr": 0.001} +{"train_loss": 0.031240463256835938, "train_loss_bc": 0.031240463256835938, "train_loss_llm": 0.0, "grad_norm": 0.04721938818693161, "global_step": 25110, "epoch": 697, "lr": 0.001} +{"train_loss": 0.041274864226579666, "train_loss_bc": 0.04125530272722244, "train_loss_llm": 0.001956254243850708, "grad_norm": 0.0625777617096901, "global_step": 25111, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03034074790775776, "train_loss_bc": 0.03034074790775776, "train_loss_llm": 0.0, "grad_norm": 0.07363449037075043, "global_step": 25112, "epoch": 697, "lr": 0.001} +{"train_loss": 0.032751698046922684, "train_loss_bc": 0.032751698046922684, "train_loss_llm": 0.0, "grad_norm": 0.03050987794995308, "global_step": 25113, "epoch": 697, "lr": 0.001} +{"train_loss": 0.034307464957237244, "train_loss_bc": 0.034307464957237244, "train_loss_llm": 0.0, "grad_norm": 0.04768195003271103, "global_step": 25114, "epoch": 697, "lr": 0.001} +{"train_loss": 0.02856653556227684, "train_loss_bc": 0.02856653556227684, "train_loss_llm": 0.0, "grad_norm": 0.06059201434254646, "global_step": 25115, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03214560076594353, "train_loss_bc": 0.03213298320770264, "train_loss_llm": 0.001261889934539795, "grad_norm": 0.07390569895505905, "global_step": 25116, "epoch": 697, "lr": 0.001} +{"train_loss": 0.039665646851062775, "train_loss_bc": 0.039665646851062775, "train_loss_llm": 0.0, "grad_norm": 0.044289033859968185, "global_step": 25117, "epoch": 697, "lr": 0.001} +{"train_loss": 0.02474874071776867, "train_loss_bc": 0.02474874071776867, "train_loss_llm": 0.0, "grad_norm": 0.06317037343978882, "global_step": 25118, "epoch": 697, "lr": 0.001} +{"train_loss": 0.030068833380937576, "train_loss_bc": 0.030068732798099518, "train_loss_llm": 1.0013580322265625e-05, "grad_norm": 0.08248774707317352, "global_step": 25119, "epoch": 697, "lr": 0.001} +{"train_loss": 0.033511172980070114, "train_loss_bc": 0.033511172980070114, "train_loss_llm": 0.0, "grad_norm": 0.09931450337171555, "global_step": 25120, "epoch": 697, "lr": 0.001} +{"train_loss": 0.028578121215105057, "train_loss_bc": 0.028578121215105057, "train_loss_llm": 0.0, "grad_norm": 0.02998555824160576, "global_step": 25121, "epoch": 697, "lr": 0.001} +{"train_loss": 0.028298014774918556, "train_loss_bc": 0.028298014774918556, "train_loss_llm": 0.0, "grad_norm": 0.04707653447985649, "global_step": 25122, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03465777263045311, "train_loss_bc": 0.03465777263045311, "train_loss_llm": 0.0, "grad_norm": 0.06625146418809891, "global_step": 25123, "epoch": 697, "lr": 0.001} +{"train_loss": 0.0339997299015522, "train_loss_bc": 0.0339997299015522, "train_loss_llm": 0.0, "grad_norm": 0.0947890505194664, "global_step": 25124, "epoch": 697, "lr": 0.001} +{"train_loss": 0.03136270120739937, "train_loss_bc": 0.03136270120739937, "train_loss_llm": 0.0, "grad_norm": 0.02946886233985424, "global_step": 25125, "epoch": 697, "lr": 0.001} +{"train_loss": 0.032017000019550323, "train_loss_bc": 0.032017000019550323, "train_loss_llm": 0.0, "grad_norm": 0.04572361335158348, "global_step": 25126, "epoch": 697, "lr": 0.001} +{"train_loss": 0.031431669265859656, "train_loss_bc": 0.029577817767858505, "train_loss_llm": 0.0, "grad_norm": 0.07478093355894089, "global_step": 25127, "epoch": 697, "lr": 0.001, "val_loss": 0.08799146115779877} +{"train_loss": 0.03343929350376129, "train_loss_bc": 0.03343929350376129, "train_loss_llm": 0.0, "grad_norm": 0.09172310680150986, "global_step": 25128, "epoch": 698, "lr": 0.001} +{"train_loss": 0.03696005046367645, "train_loss_bc": 0.03696005046367645, "train_loss_llm": 0.0, "grad_norm": 0.038150280714035034, "global_step": 25129, "epoch": 698, "lr": 0.001} +{"train_loss": 0.03390836715698242, "train_loss_bc": 0.03390836715698242, "train_loss_llm": 0.0, "grad_norm": 0.06893341988325119, "global_step": 25130, "epoch": 698, "lr": 0.001} +{"train_loss": 0.03476893901824951, "train_loss_bc": 0.03476893901824951, "train_loss_llm": 0.0, "grad_norm": 0.09522232413291931, "global_step": 25131, "epoch": 698, "lr": 0.001} +{"train_loss": 0.0321878083050251, "train_loss_bc": 0.0321878083050251, "train_loss_llm": 0.0, "grad_norm": 0.11529205739498138, "global_step": 25132, "epoch": 698, "lr": 0.001} +{"train_loss": 0.031240489333868027, "train_loss_bc": 0.031240489333868027, "train_loss_llm": 0.0, "grad_norm": 0.027060633525252342, "global_step": 25133, "epoch": 698, "lr": 0.001} +{"train_loss": 0.03630194813013077, "train_loss_bc": 0.03630194813013077, "train_loss_llm": 0.0, "grad_norm": 0.04131532832980156, "global_step": 25134, "epoch": 698, "lr": 0.001} +{"train_loss": 0.02820647321641445, "train_loss_bc": 0.02820647321641445, "train_loss_llm": 0.0, "grad_norm": 0.05113594979047775, "global_step": 25135, "epoch": 698, "lr": 0.001} +{"train_loss": 0.03863024339079857, "train_loss_bc": 0.03863024339079857, "train_loss_llm": 0.0, "grad_norm": 0.0638888031244278, "global_step": 25136, "epoch": 698, "lr": 0.001} +{"train_loss": 0.02987655997276306, "train_loss_bc": 0.02987655997276306, "train_loss_llm": 0.0, "grad_norm": 0.025013025850057602, "global_step": 25137, "epoch": 698, "lr": 0.001} +{"train_loss": 0.03569769859313965, "train_loss_bc": 0.03569769859313965, "train_loss_llm": 0.0, "grad_norm": 0.04208023101091385, "global_step": 25138, "epoch": 698, "lr": 0.001} +{"train_loss": 0.029013261198997498, "train_loss_bc": 0.029013261198997498, "train_loss_llm": 0.0, "grad_norm": 0.04932169243693352, "global_step": 25139, "epoch": 698, "lr": 0.001} +{"train_loss": 0.02472870610654354, "train_loss_bc": 0.02472870610654354, "train_loss_llm": 0.0, "grad_norm": 0.06092873588204384, "global_step": 25140, "epoch": 698, "lr": 0.001} +{"train_loss": 0.0312824472784996, "train_loss_bc": 0.031277846544981, "train_loss_llm": 0.0004602372646331787, "grad_norm": 0.032763928174972534, "global_step": 25141, "epoch": 698, "lr": 0.001} +{"train_loss": 0.03187711536884308, "train_loss_bc": 0.03187711536884308, "train_loss_llm": 0.0, "grad_norm": 0.054226212203502655, "global_step": 25142, "epoch": 698, "lr": 0.001} +{"train_loss": 0.02995014749467373, "train_loss_bc": 0.02992243506014347, "train_loss_llm": 0.002771228551864624, "grad_norm": 0.06461751461029053, "global_step": 25143, "epoch": 698, "lr": 0.001} +{"train_loss": 0.038051147013902664, "train_loss_bc": 0.03804665431380272, "train_loss_llm": 0.00044909119606018066, "grad_norm": 0.08324681967496872, "global_step": 25144, "epoch": 698, "lr": 0.001} +{"train_loss": 0.03292666748166084, "train_loss_bc": 0.03292666748166084, "train_loss_llm": 0.0, "grad_norm": 0.03330226615071297, "global_step": 25145, "epoch": 698, "lr": 0.001} +{"train_loss": 0.033586565405130386, "train_loss_bc": 0.03357699513435364, "train_loss_llm": 0.0009569227695465088, "grad_norm": 0.05509575083851814, "global_step": 25146, "epoch": 698, "lr": 0.001} +{"train_loss": 0.03253798931837082, "train_loss_bc": 0.03252295404672623, "train_loss_llm": 0.0015037059783935547, "grad_norm": 0.0750461220741272, "global_step": 25147, "epoch": 698, "lr": 0.001} +{"train_loss": 0.025588037446141243, "train_loss_bc": 0.025588037446141243, "train_loss_llm": 0.0, "grad_norm": 0.08775986731052399, "global_step": 25148, "epoch": 698, "lr": 0.001} +{"train_loss": 0.02591126039624214, "train_loss_bc": 0.02591126039624214, "train_loss_llm": 0.0, "grad_norm": 0.0298523660749197, "global_step": 25149, "epoch": 698, "lr": 0.001} +{"train_loss": 0.033486541360616684, "train_loss_bc": 0.033474575728178024, "train_loss_llm": 0.0011964738368988037, "grad_norm": 0.047909028828144073, "global_step": 25150, "epoch": 698, "lr": 0.001} +{"train_loss": 0.026589004322886467, "train_loss_bc": 0.026589004322886467, "train_loss_llm": 0.0, "grad_norm": 0.060597099363803864, "global_step": 25151, "epoch": 698, "lr": 0.001} +{"train_loss": 0.025980183854699135, "train_loss_bc": 0.025976575911045074, "train_loss_llm": 0.00036081671714782715, "grad_norm": 0.06467955559492111, "global_step": 25152, "epoch": 698, "lr": 0.001} +{"train_loss": 0.029097843915224075, "train_loss_bc": 0.029097843915224075, "train_loss_llm": 0.0, "grad_norm": 0.024268602952361107, "global_step": 25153, "epoch": 698, "lr": 0.001} +{"train_loss": 0.026327377185225487, "train_loss_bc": 0.026327377185225487, "train_loss_llm": 0.0, "grad_norm": 0.03772301599383354, "global_step": 25154, "epoch": 698, "lr": 0.001} +{"train_loss": 0.027813555672764778, "train_loss_bc": 0.027813555672764778, "train_loss_llm": 0.0, "grad_norm": 0.04967350885272026, "global_step": 25155, "epoch": 698, "lr": 0.001} +{"train_loss": 0.02914458140730858, "train_loss_bc": 0.02914458140730858, "train_loss_llm": 0.0, "grad_norm": 0.05578167364001274, "global_step": 25156, "epoch": 698, "lr": 0.001} +{"train_loss": 0.026533734053373337, "train_loss_bc": 0.026533734053373337, "train_loss_llm": 0.0, "grad_norm": 0.030241213738918304, "global_step": 25157, "epoch": 698, "lr": 0.001} +{"train_loss": 0.03321876749396324, "train_loss_bc": 0.03321876749396324, "train_loss_llm": 0.0, "grad_norm": 0.049324244260787964, "global_step": 25158, "epoch": 698, "lr": 0.001} +{"train_loss": 0.0281574297696352, "train_loss_bc": 0.0281574297696352, "train_loss_llm": 0.0, "grad_norm": 0.060318879783153534, "global_step": 25159, "epoch": 698, "lr": 0.001} +{"train_loss": 0.030937086790800095, "train_loss_bc": 0.030937086790800095, "train_loss_llm": 0.0, "grad_norm": 0.07684154063463211, "global_step": 25160, "epoch": 698, "lr": 0.001} +{"train_loss": 0.0370858833193779, "train_loss_bc": 0.0370858833193779, "train_loss_llm": 0.0, "grad_norm": 0.03702688589692116, "global_step": 25161, "epoch": 698, "lr": 0.001} +{"train_loss": 0.02813809923827648, "train_loss_bc": 0.02813809923827648, "train_loss_llm": 0.0, "grad_norm": 0.042674191296100616, "global_step": 25162, "epoch": 698, "lr": 0.001} +{"train_loss": 0.031209740994705096, "train_loss_bc": 0.034369371831417084, "train_loss_llm": 0.0, "grad_norm": 0.05853062495589256, "global_step": 25163, "epoch": 698, "lr": 0.001, "val_loss": 0.0891677513718605} +{"train_loss": 0.03072551265358925, "train_loss_bc": 0.03072551265358925, "train_loss_llm": 0.0, "grad_norm": 0.06687743961811066, "global_step": 25164, "epoch": 699, "lr": 0.001} +{"train_loss": 0.030694536864757538, "train_loss_bc": 0.030691877007484436, "train_loss_llm": 0.00026595592498779297, "grad_norm": 0.02608446031808853, "global_step": 25165, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03433558717370033, "train_loss_bc": 0.034333135932683945, "train_loss_llm": 0.00024500489234924316, "grad_norm": 0.042778536677360535, "global_step": 25166, "epoch": 699, "lr": 0.001} +{"train_loss": 0.024125197902321815, "train_loss_bc": 0.024125197902321815, "train_loss_llm": 0.0, "grad_norm": 0.0581500343978405, "global_step": 25167, "epoch": 699, "lr": 0.001} +{"train_loss": 0.029983246698975563, "train_loss_bc": 0.029983246698975563, "train_loss_llm": 0.0, "grad_norm": 0.08821437507867813, "global_step": 25168, "epoch": 699, "lr": 0.001} +{"train_loss": 0.031720783561468124, "train_loss_bc": 0.031710412353277206, "train_loss_llm": 0.0010369420051574707, "grad_norm": 0.030631838366389275, "global_step": 25169, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03093365952372551, "train_loss_bc": 0.03093365952372551, "train_loss_llm": 0.0, "grad_norm": 0.044857338070869446, "global_step": 25170, "epoch": 699, "lr": 0.001} +{"train_loss": 0.030026929453015327, "train_loss_bc": 0.030026929453015327, "train_loss_llm": 0.0, "grad_norm": 0.05969233438372612, "global_step": 25171, "epoch": 699, "lr": 0.001} +{"train_loss": 0.037922121584415436, "train_loss_bc": 0.03791360929608345, "train_loss_llm": 0.0008512735366821289, "grad_norm": 0.06522581726312637, "global_step": 25172, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03312153369188309, "train_loss_bc": 0.03312153369188309, "train_loss_llm": 0.0, "grad_norm": 0.028311124071478844, "global_step": 25173, "epoch": 699, "lr": 0.001} +{"train_loss": 0.034684788435697556, "train_loss_bc": 0.034684788435697556, "train_loss_llm": 0.0, "grad_norm": 0.04930436983704567, "global_step": 25174, "epoch": 699, "lr": 0.001} +{"train_loss": 0.027652284130454063, "train_loss_bc": 0.027644723653793335, "train_loss_llm": 0.0007560849189758301, "grad_norm": 0.063032366335392, "global_step": 25175, "epoch": 699, "lr": 0.001} +{"train_loss": 0.029101690277457237, "train_loss_bc": 0.029101690277457237, "train_loss_llm": 0.0, "grad_norm": 0.07778724282979965, "global_step": 25176, "epoch": 699, "lr": 0.001} +{"train_loss": 0.029677700251340866, "train_loss_bc": 0.029677700251340866, "train_loss_llm": 0.0, "grad_norm": 0.033004555851221085, "global_step": 25177, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03223114088177681, "train_loss_bc": 0.032227251678705215, "train_loss_llm": 0.00038880109786987305, "grad_norm": 0.05012192949652672, "global_step": 25178, "epoch": 699, "lr": 0.001} +{"train_loss": 0.0332825593650341, "train_loss_bc": 0.0332825593650341, "train_loss_llm": 0.0, "grad_norm": 0.06476712226867676, "global_step": 25179, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03175221011042595, "train_loss_bc": 0.03175221011042595, "train_loss_llm": 0.0, "grad_norm": 0.08045423030853271, "global_step": 25180, "epoch": 699, "lr": 0.001} +{"train_loss": 0.025873737409710884, "train_loss_bc": 0.025873737409710884, "train_loss_llm": 0.0, "grad_norm": 0.025099867954850197, "global_step": 25181, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03230277821421623, "train_loss_bc": 0.03230277821421623, "train_loss_llm": 0.0, "grad_norm": 0.04121559113264084, "global_step": 25182, "epoch": 699, "lr": 0.001} +{"train_loss": 0.025220979005098343, "train_loss_bc": 0.025220979005098343, "train_loss_llm": 0.0, "grad_norm": 0.05246799439191818, "global_step": 25183, "epoch": 699, "lr": 0.001} +{"train_loss": 0.032070763409137726, "train_loss_bc": 0.032070763409137726, "train_loss_llm": 0.0, "grad_norm": 0.061483707278966904, "global_step": 25184, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03176214173436165, "train_loss_bc": 0.03176214173436165, "train_loss_llm": 0.0, "grad_norm": 0.033041082322597504, "global_step": 25185, "epoch": 699, "lr": 0.001} +{"train_loss": 0.025289461016654968, "train_loss_bc": 0.025289461016654968, "train_loss_llm": 0.0, "grad_norm": 0.044950757175683975, "global_step": 25186, "epoch": 699, "lr": 0.001} +{"train_loss": 0.031838949769735336, "train_loss_bc": 0.031838949769735336, "train_loss_llm": 0.0, "grad_norm": 0.0649113655090332, "global_step": 25187, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03429850563406944, "train_loss_bc": 0.03429850563406944, "train_loss_llm": 0.0, "grad_norm": 0.09013240039348602, "global_step": 25188, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03842969611287117, "train_loss_bc": 0.03842969611287117, "train_loss_llm": 0.0, "grad_norm": 0.02731650322675705, "global_step": 25189, "epoch": 699, "lr": 0.001} +{"train_loss": 0.030566928908228874, "train_loss_bc": 0.03055364266037941, "train_loss_llm": 0.001328587532043457, "grad_norm": 0.03767542541027069, "global_step": 25190, "epoch": 699, "lr": 0.001} +{"train_loss": 0.02774682082235813, "train_loss_bc": 0.02774682082235813, "train_loss_llm": 0.0, "grad_norm": 0.04495500773191452, "global_step": 25191, "epoch": 699, "lr": 0.001} +{"train_loss": 0.032425958663225174, "train_loss_bc": 0.03241473436355591, "train_loss_llm": 0.0011223554611206055, "grad_norm": 0.057483553886413574, "global_step": 25192, "epoch": 699, "lr": 0.001} +{"train_loss": 0.035244133323431015, "train_loss_bc": 0.035244133323431015, "train_loss_llm": 0.0, "grad_norm": 0.029767584055662155, "global_step": 25193, "epoch": 699, "lr": 0.001} +{"train_loss": 0.029257820919156075, "train_loss_bc": 0.029250947758555412, "train_loss_llm": 0.0006873905658721924, "grad_norm": 0.03948523476719856, "global_step": 25194, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03896396979689598, "train_loss_bc": 0.03896396979689598, "train_loss_llm": 0.0, "grad_norm": 0.05811465159058571, "global_step": 25195, "epoch": 699, "lr": 0.001} +{"train_loss": 0.02750195376574993, "train_loss_bc": 0.027488049119710922, "train_loss_llm": 0.0013904273509979248, "grad_norm": 0.0768953263759613, "global_step": 25196, "epoch": 699, "lr": 0.001} +{"train_loss": 0.02924615703523159, "train_loss_bc": 0.02924615703523159, "train_loss_llm": 0.0, "grad_norm": 0.03026236779987812, "global_step": 25197, "epoch": 699, "lr": 0.001} +{"train_loss": 0.033616259694099426, "train_loss_bc": 0.033578500151634216, "train_loss_llm": 0.003775954246520996, "grad_norm": 0.056088924407958984, "global_step": 25198, "epoch": 699, "lr": 0.001} +{"train_loss": 0.03122315328154299, "train_loss_bc": 0.030402276664972305, "train_loss_llm": 0.00027441978454589844, "grad_norm": 0.06857717037200928, "global_step": 25199, "epoch": 699, "lr": 0.001, "val_loss": 0.08964608609676361} +{"train_loss": 0.03016987442970276, "train_loss_bc": 0.03016987442970276, "train_loss_llm": 0.0, "grad_norm": 0.07455428689718246, "global_step": 25200, "epoch": 700, "lr": 0.001} +{"train_loss": 0.025912977755069733, "train_loss_bc": 0.0259097907692194, "train_loss_llm": 0.0003186464309692383, "grad_norm": 0.02654803730547428, "global_step": 25201, "epoch": 700, "lr": 0.001} +{"train_loss": 0.0279262512922287, "train_loss_bc": 0.0279262512922287, "train_loss_llm": 0.0, "grad_norm": 0.03939875587821007, "global_step": 25202, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03209258243441582, "train_loss_bc": 0.03208402916789055, "train_loss_llm": 0.0008553862571716309, "grad_norm": 0.05200732499361038, "global_step": 25203, "epoch": 700, "lr": 0.001} +{"train_loss": 0.02687283419072628, "train_loss_bc": 0.02687283419072628, "train_loss_llm": 0.0, "grad_norm": 0.06401478499174118, "global_step": 25204, "epoch": 700, "lr": 0.001} +{"train_loss": 0.030354194343090057, "train_loss_bc": 0.03035195916891098, "train_loss_llm": 0.00022351741790771484, "grad_norm": 0.03891601786017418, "global_step": 25205, "epoch": 700, "lr": 0.001} +{"train_loss": 0.034922607243061066, "train_loss_bc": 0.034922607243061066, "train_loss_llm": 0.0, "grad_norm": 0.06118901073932648, "global_step": 25206, "epoch": 700, "lr": 0.001} +{"train_loss": 0.030192352831363678, "train_loss_bc": 0.03019116446375847, "train_loss_llm": 0.00011876225471496582, "grad_norm": 0.0849113017320633, "global_step": 25207, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03361250460147858, "train_loss_bc": 0.03361250460147858, "train_loss_llm": 0.0, "grad_norm": 0.09576544165611267, "global_step": 25208, "epoch": 700, "lr": 0.001} +{"train_loss": 0.030785709619522095, "train_loss_bc": 0.030785709619522095, "train_loss_llm": 0.0, "grad_norm": 0.03498351201415062, "global_step": 25209, "epoch": 700, "lr": 0.001} +{"train_loss": 0.027832627296447754, "train_loss_bc": 0.027831781655550003, "train_loss_llm": 8.45789909362793e-05, "grad_norm": 0.049595851451158524, "global_step": 25210, "epoch": 700, "lr": 0.001} +{"train_loss": 0.030621614307165146, "train_loss_bc": 0.030621614307165146, "train_loss_llm": 0.0, "grad_norm": 0.06578930467367172, "global_step": 25211, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03817898407578468, "train_loss_bc": 0.03817898407578468, "train_loss_llm": 0.0, "grad_norm": 0.08701677620410919, "global_step": 25212, "epoch": 700, "lr": 0.001} +{"train_loss": 0.027853084728121758, "train_loss_bc": 0.027853084728121758, "train_loss_llm": 0.0, "grad_norm": 0.027099039405584335, "global_step": 25213, "epoch": 700, "lr": 0.001} +{"train_loss": 0.030336814001202583, "train_loss_bc": 0.030336814001202583, "train_loss_llm": 0.0, "grad_norm": 0.04083221033215523, "global_step": 25214, "epoch": 700, "lr": 0.001} +{"train_loss": 0.030593786388635635, "train_loss_bc": 0.030592558905482292, "train_loss_llm": 0.0001227259635925293, "grad_norm": 0.05482042208313942, "global_step": 25215, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03552325442433357, "train_loss_bc": 0.03551557660102844, "train_loss_llm": 0.0007679164409637451, "grad_norm": 0.06884967535734177, "global_step": 25216, "epoch": 700, "lr": 0.001} +{"train_loss": 0.031509436666965485, "train_loss_bc": 0.031509436666965485, "train_loss_llm": 0.0, "grad_norm": 0.027706176042556763, "global_step": 25217, "epoch": 700, "lr": 0.001} +{"train_loss": 0.028144394978880882, "train_loss_bc": 0.028144394978880882, "train_loss_llm": 0.0, "grad_norm": 0.04431935027241707, "global_step": 25218, "epoch": 700, "lr": 0.001} +{"train_loss": 0.030143773183226585, "train_loss_bc": 0.030143773183226585, "train_loss_llm": 0.0, "grad_norm": 0.0561053566634655, "global_step": 25219, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03131435811519623, "train_loss_bc": 0.03131435811519623, "train_loss_llm": 0.0, "grad_norm": 0.0810994803905487, "global_step": 25220, "epoch": 700, "lr": 0.001} +{"train_loss": 0.037132032215595245, "train_loss_bc": 0.037132032215595245, "train_loss_llm": 0.0, "grad_norm": 0.040979351848363876, "global_step": 25221, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03456146642565727, "train_loss_bc": 0.03456146642565727, "train_loss_llm": 0.0, "grad_norm": 0.05319087207317352, "global_step": 25222, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03897413983941078, "train_loss_bc": 0.03897413983941078, "train_loss_llm": 0.0, "grad_norm": 0.07286443561315536, "global_step": 25223, "epoch": 700, "lr": 0.001} +{"train_loss": 0.02896764688193798, "train_loss_bc": 0.028942661359906197, "train_loss_llm": 0.0024985969066619873, "grad_norm": 0.0812133327126503, "global_step": 25224, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03220127150416374, "train_loss_bc": 0.03220127150416374, "train_loss_llm": 0.0, "grad_norm": 0.031221849843859673, "global_step": 25225, "epoch": 700, "lr": 0.001} +{"train_loss": 0.027943333610892296, "train_loss_bc": 0.027943333610892296, "train_loss_llm": 0.0, "grad_norm": 0.04395002871751785, "global_step": 25226, "epoch": 700, "lr": 0.001} +{"train_loss": 0.02969716303050518, "train_loss_bc": 0.029692301526665688, "train_loss_llm": 0.000486224889755249, "grad_norm": 0.07494929432868958, "global_step": 25227, "epoch": 700, "lr": 0.001} +{"train_loss": 0.023902958258986473, "train_loss_bc": 0.023902958258986473, "train_loss_llm": 0.0, "grad_norm": 0.09021472185850143, "global_step": 25228, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03676478937268257, "train_loss_bc": 0.03676478937268257, "train_loss_llm": 0.0, "grad_norm": 0.02856322191655636, "global_step": 25229, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03507951274514198, "train_loss_bc": 0.03507951274514198, "train_loss_llm": 0.0, "grad_norm": 0.052417803555727005, "global_step": 25230, "epoch": 700, "lr": 0.001} +{"train_loss": 0.028831522911787033, "train_loss_bc": 0.028815722092986107, "train_loss_llm": 0.0015801191329956055, "grad_norm": 0.06303740292787552, "global_step": 25231, "epoch": 700, "lr": 0.001} +{"train_loss": 0.03265836089849472, "train_loss_bc": 0.032650694251060486, "train_loss_llm": 0.000766754150390625, "grad_norm": 0.0788569226861, "global_step": 25232, "epoch": 700, "lr": 0.001} +{"train_loss": 0.031520962715148926, "train_loss_bc": 0.031520962715148926, "train_loss_llm": 0.0, "grad_norm": 0.034669406712055206, "global_step": 25233, "epoch": 700, "lr": 0.001} +{"train_loss": 0.027978166937828064, "train_loss_bc": 0.02796417661011219, "train_loss_llm": 0.0013990998268127441, "grad_norm": 0.043998368084430695, "global_step": 25234, "epoch": 700, "lr": 0.001} +{"train_loss": 0.0312658726858596, "train_loss_bc": 0.034464072436094284, "train_loss_llm": 0.0, "grad_norm": 0.05895906314253807, "global_step": 25235, "epoch": 700, "lr": 0.001, "train/cumulative_reward": 83.32392312513109, "train/mean_score": 7.630046243612218, "train/success_rate": 0.1, "test/cumulative_reward": 37.13062826919946, "test/mean_score": 10.235628908821779, "test/success_rate": 0.28, "val_loss": 0.08927315473556519, "train_action_mse_error": 0.033796798437833786} +{"train_loss": 0.03066241554915905, "train_loss_bc": 0.030625801533460617, "train_loss_llm": 0.0036614537239074707, "grad_norm": 0.06509391218423843, "global_step": 25236, "epoch": 701, "lr": 0.001} +{"train_loss": 0.030740533024072647, "train_loss_bc": 0.030740533024072647, "train_loss_llm": 0.0, "grad_norm": 0.03608983755111694, "global_step": 25237, "epoch": 701, "lr": 0.001} +{"train_loss": 0.024212080985307693, "train_loss_bc": 0.024212080985307693, "train_loss_llm": 0.0, "grad_norm": 0.04480348899960518, "global_step": 25238, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03161012753844261, "train_loss_bc": 0.03161012753844261, "train_loss_llm": 0.0, "grad_norm": 0.04207296296954155, "global_step": 25239, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03189653158187866, "train_loss_bc": 0.03188949450850487, "train_loss_llm": 0.000703662633895874, "grad_norm": 0.05585302785038948, "global_step": 25240, "epoch": 701, "lr": 0.001} +{"train_loss": 0.030639396980404854, "train_loss_bc": 0.030639396980404854, "train_loss_llm": 0.0, "grad_norm": 0.027888936921954155, "global_step": 25241, "epoch": 701, "lr": 0.001} +{"train_loss": 0.02768629789352417, "train_loss_bc": 0.02768534980714321, "train_loss_llm": 9.47415828704834e-05, "grad_norm": 0.04076438397169113, "global_step": 25242, "epoch": 701, "lr": 0.001} +{"train_loss": 0.028328174725174904, "train_loss_bc": 0.028325380757451057, "train_loss_llm": 0.00027939677238464355, "grad_norm": 0.05556650832295418, "global_step": 25243, "epoch": 701, "lr": 0.001} +{"train_loss": 0.02074190229177475, "train_loss_bc": 0.02073616161942482, "train_loss_llm": 0.0005740821361541748, "grad_norm": 0.0673268511891365, "global_step": 25244, "epoch": 701, "lr": 0.001} +{"train_loss": 0.02798992209136486, "train_loss_bc": 0.02798992209136486, "train_loss_llm": 0.0, "grad_norm": 0.020752370357513428, "global_step": 25245, "epoch": 701, "lr": 0.001} +{"train_loss": 0.02943652682006359, "train_loss_bc": 0.029427239671349525, "train_loss_llm": 0.0009286403656005859, "grad_norm": 0.03393867239356041, "global_step": 25246, "epoch": 701, "lr": 0.001} +{"train_loss": 0.031088965013623238, "train_loss_bc": 0.031088965013623238, "train_loss_llm": 0.0, "grad_norm": 0.04885297268629074, "global_step": 25247, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03475465625524521, "train_loss_bc": 0.03475465625524521, "train_loss_llm": 0.0, "grad_norm": 0.055992692708969116, "global_step": 25248, "epoch": 701, "lr": 0.001} +{"train_loss": 0.031658757477998734, "train_loss_bc": 0.031658757477998734, "train_loss_llm": 0.0, "grad_norm": 0.026705538854002953, "global_step": 25249, "epoch": 701, "lr": 0.001} +{"train_loss": 0.028650833293795586, "train_loss_bc": 0.028650833293795586, "train_loss_llm": 0.0, "grad_norm": 0.03868700563907623, "global_step": 25250, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03663056716322899, "train_loss_bc": 0.03663056716322899, "train_loss_llm": 0.0, "grad_norm": 0.06014169380068779, "global_step": 25251, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03342478722333908, "train_loss_bc": 0.033422138541936874, "train_loss_llm": 0.00026479363441467285, "grad_norm": 0.0788343995809555, "global_step": 25252, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03404288366436958, "train_loss_bc": 0.03404288366436958, "train_loss_llm": 0.0, "grad_norm": 0.034813087433576584, "global_step": 25253, "epoch": 701, "lr": 0.001} +{"train_loss": 0.030445797368884087, "train_loss_bc": 0.030440760776400566, "train_loss_llm": 0.0005037486553192139, "grad_norm": 0.05959039181470871, "global_step": 25254, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03747798874974251, "train_loss_bc": 0.03747798874974251, "train_loss_llm": 0.0, "grad_norm": 0.08068173378705978, "global_step": 25255, "epoch": 701, "lr": 0.001} +{"train_loss": 0.030511828139424324, "train_loss_bc": 0.030511828139424324, "train_loss_llm": 0.0, "grad_norm": 0.09032474458217621, "global_step": 25256, "epoch": 701, "lr": 0.001} +{"train_loss": 0.0316704586148262, "train_loss_bc": 0.0316704586148262, "train_loss_llm": 0.0, "grad_norm": 0.03911645710468292, "global_step": 25257, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03489888831973076, "train_loss_bc": 0.03489888831973076, "train_loss_llm": 0.0, "grad_norm": 0.05678797513246536, "global_step": 25258, "epoch": 701, "lr": 0.001} +{"train_loss": 0.0301517304033041, "train_loss_bc": 0.0301517304033041, "train_loss_llm": 0.0, "grad_norm": 0.07805437594652176, "global_step": 25259, "epoch": 701, "lr": 0.001} +{"train_loss": 0.032374244183301926, "train_loss_bc": 0.032374244183301926, "train_loss_llm": 0.0, "grad_norm": 0.0946175754070282, "global_step": 25260, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03452687710523605, "train_loss_bc": 0.03452687710523605, "train_loss_llm": 0.0, "grad_norm": 0.032684117555618286, "global_step": 25261, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03237134590744972, "train_loss_bc": 0.032364070415496826, "train_loss_llm": 0.0007277131080627441, "grad_norm": 0.04238856956362724, "global_step": 25262, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03212222829461098, "train_loss_bc": 0.03212222829461098, "train_loss_llm": 0.0, "grad_norm": 0.056741274893283844, "global_step": 25263, "epoch": 701, "lr": 0.001} +{"train_loss": 0.029379786923527718, "train_loss_bc": 0.029379786923527718, "train_loss_llm": 0.0, "grad_norm": 0.06244892254471779, "global_step": 25264, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03161423280835152, "train_loss_bc": 0.03161423280835152, "train_loss_llm": 0.0, "grad_norm": 0.029091184958815575, "global_step": 25265, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03380625322461128, "train_loss_bc": 0.03380625322461128, "train_loss_llm": 0.0, "grad_norm": 0.055610235780477524, "global_step": 25266, "epoch": 701, "lr": 0.001} +{"train_loss": 0.03275809809565544, "train_loss_bc": 0.03274281695485115, "train_loss_llm": 0.0015282928943634033, "grad_norm": 0.07312140613794327, "global_step": 25267, "epoch": 701, "lr": 0.001} +{"train_loss": 0.02996472641825676, "train_loss_bc": 0.02996472641825676, "train_loss_llm": 0.0, "grad_norm": 0.09523407369852066, "global_step": 25268, "epoch": 701, "lr": 0.001} +{"train_loss": 0.0332644060254097, "train_loss_bc": 0.033264197409152985, "train_loss_llm": 2.0772218704223633e-05, "grad_norm": 0.035939786583185196, "global_step": 25269, "epoch": 701, "lr": 0.001} +{"train_loss": 0.0260569266974926, "train_loss_bc": 0.026054145768284798, "train_loss_llm": 0.00027802586555480957, "grad_norm": 0.053344979882240295, "global_step": 25270, "epoch": 701, "lr": 0.001} +{"train_loss": 0.031157904770225286, "train_loss_bc": 0.03408849984407425, "train_loss_llm": 0.0004896819591522217, "grad_norm": 0.08138024061918259, "global_step": 25271, "epoch": 701, "lr": 0.001, "val_loss": 0.08847944438457489} +{"train_loss": 0.03668878972530365, "train_loss_bc": 0.03668878972530365, "train_loss_llm": 0.0, "grad_norm": 0.10159045457839966, "global_step": 25272, "epoch": 702, "lr": 0.001} +{"train_loss": 0.029158342629671097, "train_loss_bc": 0.029157729819417, "train_loss_llm": 6.130337715148926e-05, "grad_norm": 0.026666652411222458, "global_step": 25273, "epoch": 702, "lr": 0.001} +{"train_loss": 0.029070870950818062, "train_loss_bc": 0.029067493975162506, "train_loss_llm": 0.0003377199172973633, "grad_norm": 0.05568281561136246, "global_step": 25274, "epoch": 702, "lr": 0.001} +{"train_loss": 0.03643330931663513, "train_loss_bc": 0.03643330931663513, "train_loss_llm": 0.0, "grad_norm": 0.06954235583543777, "global_step": 25275, "epoch": 702, "lr": 0.001} +{"train_loss": 0.0259648896753788, "train_loss_bc": 0.0259648896753788, "train_loss_llm": 0.0, "grad_norm": 0.08572285622358322, "global_step": 25276, "epoch": 702, "lr": 0.001} +{"train_loss": 0.0349276103079319, "train_loss_bc": 0.034917619079351425, "train_loss_llm": 0.0009991824626922607, "grad_norm": 0.0346241258084774, "global_step": 25277, "epoch": 702, "lr": 0.001} +{"train_loss": 0.035329632461071014, "train_loss_bc": 0.03531757369637489, "train_loss_llm": 0.0012059211730957031, "grad_norm": 0.06080102175474167, "global_step": 25278, "epoch": 702, "lr": 0.001} +{"train_loss": 0.036407940089702606, "train_loss_bc": 0.036407940089702606, "train_loss_llm": 0.0, "grad_norm": 0.07494030147790909, "global_step": 25279, "epoch": 702, "lr": 0.001} +{"train_loss": 0.03925992548465729, "train_loss_bc": 0.03923344984650612, "train_loss_llm": 0.002647697925567627, "grad_norm": 0.09317133575677872, "global_step": 25280, "epoch": 702, "lr": 0.001} +{"train_loss": 0.03170683979988098, "train_loss_bc": 0.03170683979988098, "train_loss_llm": 0.0, "grad_norm": 0.027382027357816696, "global_step": 25281, "epoch": 702, "lr": 0.001} +{"train_loss": 0.033715106546878815, "train_loss_bc": 0.03369719535112381, "train_loss_llm": 0.001791149377822876, "grad_norm": 0.046391721814870834, "global_step": 25282, "epoch": 702, "lr": 0.001} +{"train_loss": 0.030865313485264778, "train_loss_bc": 0.030865313485264778, "train_loss_llm": 0.0, "grad_norm": 0.06170837953686714, "global_step": 25283, "epoch": 702, "lr": 0.001} +{"train_loss": 0.031939972192049026, "train_loss_bc": 0.031939972192049026, "train_loss_llm": 0.0, "grad_norm": 0.07474730908870697, "global_step": 25284, "epoch": 702, "lr": 0.001} +{"train_loss": 0.03096715919673443, "train_loss_bc": 0.03096715919673443, "train_loss_llm": 0.0, "grad_norm": 0.03341366723179817, "global_step": 25285, "epoch": 702, "lr": 0.001} +{"train_loss": 0.029647812247276306, "train_loss_bc": 0.029643723741173744, "train_loss_llm": 0.0004088282585144043, "grad_norm": 0.05025402829051018, "global_step": 25286, "epoch": 702, "lr": 0.001} +{"train_loss": 0.026280734688043594, "train_loss_bc": 0.02627645432949066, "train_loss_llm": 0.0004279613494873047, "grad_norm": 0.06558231264352798, "global_step": 25287, "epoch": 702, "lr": 0.001} +{"train_loss": 0.025467095896601677, "train_loss_bc": 0.025467095896601677, "train_loss_llm": 0.0, "grad_norm": 0.08012406527996063, "global_step": 25288, "epoch": 702, "lr": 0.001} +{"train_loss": 0.028369946405291557, "train_loss_bc": 0.028369946405291557, "train_loss_llm": 0.0, "grad_norm": 0.03651238977909088, "global_step": 25289, "epoch": 702, "lr": 0.001} +{"train_loss": 0.0314149484038353, "train_loss_bc": 0.03139529004693031, "train_loss_llm": 0.001965939998626709, "grad_norm": 0.05126471444964409, "global_step": 25290, "epoch": 702, "lr": 0.001} +{"train_loss": 0.030554458498954773, "train_loss_bc": 0.030554458498954773, "train_loss_llm": 0.0, "grad_norm": 0.06773701310157776, "global_step": 25291, "epoch": 702, "lr": 0.001} +{"train_loss": 0.02556576021015644, "train_loss_bc": 0.02556576021015644, "train_loss_llm": 0.0, "grad_norm": 0.07587583363056183, "global_step": 25292, "epoch": 702, "lr": 0.001} +{"train_loss": 0.0307427030056715, "train_loss_bc": 0.0307427030056715, "train_loss_llm": 0.0, "grad_norm": 0.02550278976559639, "global_step": 25293, "epoch": 702, "lr": 0.001} +{"train_loss": 0.032423235476017, "train_loss_bc": 0.032423235476017, "train_loss_llm": 0.0, "grad_norm": 0.04348332807421684, "global_step": 25294, "epoch": 702, "lr": 0.001} +{"train_loss": 0.03452501818537712, "train_loss_bc": 0.03449656441807747, "train_loss_llm": 0.0028453469276428223, "grad_norm": 0.05658009648323059, "global_step": 25295, "epoch": 702, "lr": 0.001} +{"train_loss": 0.029173867776989937, "train_loss_bc": 0.029173867776989937, "train_loss_llm": 0.0, "grad_norm": 0.07584360241889954, "global_step": 25296, "epoch": 702, "lr": 0.001} +{"train_loss": 0.031134996563196182, "train_loss_bc": 0.031134996563196182, "train_loss_llm": 0.0, "grad_norm": 0.038209639489650726, "global_step": 25297, "epoch": 702, "lr": 0.001} +{"train_loss": 0.030016474425792694, "train_loss_bc": 0.030016474425792694, "train_loss_llm": 0.0, "grad_norm": 0.06126460060477257, "global_step": 25298, "epoch": 702, "lr": 0.001} +{"train_loss": 0.032690975815057755, "train_loss_bc": 0.032690975815057755, "train_loss_llm": 0.0, "grad_norm": 0.08757851272821426, "global_step": 25299, "epoch": 702, "lr": 0.001} +{"train_loss": 0.029950037598609924, "train_loss_bc": 0.029950037598609924, "train_loss_llm": 0.0, "grad_norm": 0.09968676418066025, "global_step": 25300, "epoch": 702, "lr": 0.001} +{"train_loss": 0.03345082700252533, "train_loss_bc": 0.03345082700252533, "train_loss_llm": 0.0, "grad_norm": 0.04147820919752121, "global_step": 25301, "epoch": 702, "lr": 0.001} +{"train_loss": 0.03391615301370621, "train_loss_bc": 0.03391615301370621, "train_loss_llm": 0.0, "grad_norm": 0.056612804532051086, "global_step": 25302, "epoch": 702, "lr": 0.001} +{"train_loss": 0.029242807999253273, "train_loss_bc": 0.029217908158898354, "train_loss_llm": 0.002489924430847168, "grad_norm": 0.0752621442079544, "global_step": 25303, "epoch": 702, "lr": 0.001} +{"train_loss": 0.03085934929549694, "train_loss_bc": 0.03085934929549694, "train_loss_llm": 0.0, "grad_norm": 0.08985545486211777, "global_step": 25304, "epoch": 702, "lr": 0.001} +{"train_loss": 0.030375033617019653, "train_loss_bc": 0.030375033617019653, "train_loss_llm": 0.0, "grad_norm": 0.035867881029844284, "global_step": 25305, "epoch": 702, "lr": 0.001} +{"train_loss": 0.028798235580325127, "train_loss_bc": 0.028798235580325127, "train_loss_llm": 0.0, "grad_norm": 0.05680301412940025, "global_step": 25306, "epoch": 702, "lr": 0.001} +{"train_loss": 0.031262159502754606, "train_loss_bc": 0.02839931845664978, "train_loss_llm": 0.00022491812705993652, "grad_norm": 0.07895068824291229, "global_step": 25307, "epoch": 702, "lr": 0.001, "val_loss": 0.0890716165304184} +{"train_loss": 0.035272784531116486, "train_loss_bc": 0.035272784531116486, "train_loss_llm": 0.0, "grad_norm": 0.09709854423999786, "global_step": 25308, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03352397307753563, "train_loss_bc": 0.03352074697613716, "train_loss_llm": 0.00032255053520202637, "grad_norm": 0.031380411237478256, "global_step": 25309, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03505471348762512, "train_loss_bc": 0.03503436967730522, "train_loss_llm": 0.002034395933151245, "grad_norm": 0.05556980147957802, "global_step": 25310, "epoch": 703, "lr": 0.001} +{"train_loss": 0.02951900288462639, "train_loss_bc": 0.02951900288462639, "train_loss_llm": 0.0, "grad_norm": 0.07760578393936157, "global_step": 25311, "epoch": 703, "lr": 0.001} +{"train_loss": 0.026332108303904533, "train_loss_bc": 0.026332108303904533, "train_loss_llm": 0.0, "grad_norm": 0.09994515031576157, "global_step": 25312, "epoch": 703, "lr": 0.001} +{"train_loss": 0.0342668741941452, "train_loss_bc": 0.0342668741941452, "train_loss_llm": 0.0, "grad_norm": 0.03440454974770546, "global_step": 25313, "epoch": 703, "lr": 0.001} +{"train_loss": 0.030076108872890472, "train_loss_bc": 0.030076108872890472, "train_loss_llm": 0.0, "grad_norm": 0.05562448129057884, "global_step": 25314, "epoch": 703, "lr": 0.001} +{"train_loss": 0.037041641771793365, "train_loss_bc": 0.037041641771793365, "train_loss_llm": 0.0, "grad_norm": 0.07098656892776489, "global_step": 25315, "epoch": 703, "lr": 0.001} +{"train_loss": 0.026545684784650803, "train_loss_bc": 0.026545684784650803, "train_loss_llm": 0.0, "grad_norm": 0.08917738497257233, "global_step": 25316, "epoch": 703, "lr": 0.001} +{"train_loss": 0.031185921281576157, "train_loss_bc": 0.031185921281576157, "train_loss_llm": 0.0, "grad_norm": 0.031441062688827515, "global_step": 25317, "epoch": 703, "lr": 0.001} +{"train_loss": 0.029253806918859482, "train_loss_bc": 0.029253806918859482, "train_loss_llm": 0.0, "grad_norm": 0.042856551706790924, "global_step": 25318, "epoch": 703, "lr": 0.001} +{"train_loss": 0.0272152591496706, "train_loss_bc": 0.0272152591496706, "train_loss_llm": 0.0, "grad_norm": 0.06539309024810791, "global_step": 25319, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03152577206492424, "train_loss_bc": 0.031523071229457855, "train_loss_llm": 0.0002700984477996826, "grad_norm": 0.08026447892189026, "global_step": 25320, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03213007003068924, "train_loss_bc": 0.03213007003068924, "train_loss_llm": 0.0, "grad_norm": 0.03226989507675171, "global_step": 25321, "epoch": 703, "lr": 0.001} +{"train_loss": 0.028752991929650307, "train_loss_bc": 0.028747469186782837, "train_loss_llm": 0.0005522072315216064, "grad_norm": 0.04369213059544563, "global_step": 25322, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03368759900331497, "train_loss_bc": 0.03368524834513664, "train_loss_llm": 0.0002351999282836914, "grad_norm": 0.054539795964956284, "global_step": 25323, "epoch": 703, "lr": 0.001} +{"train_loss": 0.029740549623966217, "train_loss_bc": 0.029735995456576347, "train_loss_llm": 0.0004553794860839844, "grad_norm": 0.07225043326616287, "global_step": 25324, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03419370576739311, "train_loss_bc": 0.0341823436319828, "train_loss_llm": 0.00113639235496521, "grad_norm": 0.040843356400728226, "global_step": 25325, "epoch": 703, "lr": 0.001} +{"train_loss": 0.0347614549100399, "train_loss_bc": 0.0347614549100399, "train_loss_llm": 0.0, "grad_norm": 0.07481788098812103, "global_step": 25326, "epoch": 703, "lr": 0.001} +{"train_loss": 0.030773688107728958, "train_loss_bc": 0.030773688107728958, "train_loss_llm": 0.0, "grad_norm": 0.1011725440621376, "global_step": 25327, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03287539631128311, "train_loss_bc": 0.03287539631128311, "train_loss_llm": 0.0, "grad_norm": 0.1090056449174881, "global_step": 25328, "epoch": 703, "lr": 0.001} +{"train_loss": 0.0349569171667099, "train_loss_bc": 0.0349569171667099, "train_loss_llm": 0.0, "grad_norm": 0.051536064594984055, "global_step": 25329, "epoch": 703, "lr": 0.001} +{"train_loss": 0.027774164453148842, "train_loss_bc": 0.027774164453148842, "train_loss_llm": 0.0, "grad_norm": 0.07391298562288284, "global_step": 25330, "epoch": 703, "lr": 0.001} +{"train_loss": 0.027024222537875175, "train_loss_bc": 0.027009956538677216, "train_loss_llm": 0.0014265477657318115, "grad_norm": 0.09482266008853912, "global_step": 25331, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03335052356123924, "train_loss_bc": 0.03335052356123924, "train_loss_llm": 0.0, "grad_norm": 0.10480967164039612, "global_step": 25332, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03144998848438263, "train_loss_bc": 0.03144998848438263, "train_loss_llm": 0.0, "grad_norm": 0.0412772111594677, "global_step": 25333, "epoch": 703, "lr": 0.001} +{"train_loss": 0.033265165984630585, "train_loss_bc": 0.033265165984630585, "train_loss_llm": 0.0, "grad_norm": 0.06455360352993011, "global_step": 25334, "epoch": 703, "lr": 0.001} +{"train_loss": 0.0309891477227211, "train_loss_bc": 0.0309891477227211, "train_loss_llm": 0.0, "grad_norm": 0.07532605528831482, "global_step": 25335, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03299375995993614, "train_loss_bc": 0.03299375995993614, "train_loss_llm": 0.0, "grad_norm": 0.09970871359109879, "global_step": 25336, "epoch": 703, "lr": 0.001} +{"train_loss": 0.026214005425572395, "train_loss_bc": 0.026214005425572395, "train_loss_llm": 0.0, "grad_norm": 0.03276527673006058, "global_step": 25337, "epoch": 703, "lr": 0.001} +{"train_loss": 0.03134891018271446, "train_loss_bc": 0.03134891018271446, "train_loss_llm": 0.0, "grad_norm": 0.04759377986192703, "global_step": 25338, "epoch": 703, "lr": 0.001} +{"train_loss": 0.028923949226737022, "train_loss_bc": 0.028923949226737022, "train_loss_llm": 0.0, "grad_norm": 0.05957445502281189, "global_step": 25339, "epoch": 703, "lr": 0.001} +{"train_loss": 0.031568918377161026, "train_loss_bc": 0.03156713768839836, "train_loss_llm": 0.00017815828323364258, "grad_norm": 0.06428031623363495, "global_step": 25340, "epoch": 703, "lr": 0.001} +{"train_loss": 0.032328322529792786, "train_loss_bc": 0.032328322529792786, "train_loss_llm": 0.0, "grad_norm": 0.0321008525788784, "global_step": 25341, "epoch": 703, "lr": 0.001} +{"train_loss": 0.032728053629398346, "train_loss_bc": 0.032728053629398346, "train_loss_llm": 0.0, "grad_norm": 0.058514565229415894, "global_step": 25342, "epoch": 703, "lr": 0.001} +{"train_loss": 0.031324227133558855, "train_loss_bc": 0.0290000569075346, "train_loss_llm": 0.0026962757110595703, "grad_norm": 0.06858261674642563, "global_step": 25343, "epoch": 703, "lr": 0.001, "val_loss": 0.08829464018344879} +{"train_loss": 0.03553543984889984, "train_loss_bc": 0.03553543984889984, "train_loss_llm": 0.0, "grad_norm": 0.0975634753704071, "global_step": 25344, "epoch": 704, "lr": 0.001} +{"train_loss": 0.034538231790065765, "train_loss_bc": 0.034538231790065765, "train_loss_llm": 0.0, "grad_norm": 0.039489731192588806, "global_step": 25345, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03179237246513367, "train_loss_bc": 0.03179237246513367, "train_loss_llm": 0.0, "grad_norm": 0.06561828404664993, "global_step": 25346, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03666345402598381, "train_loss_bc": 0.03666345402598381, "train_loss_llm": 0.0, "grad_norm": 0.08022323995828629, "global_step": 25347, "epoch": 704, "lr": 0.001} +{"train_loss": 0.02991025149822235, "train_loss_bc": 0.02991025149822235, "train_loss_llm": 0.0, "grad_norm": 0.10668598860502243, "global_step": 25348, "epoch": 704, "lr": 0.001} +{"train_loss": 0.024297652766108513, "train_loss_bc": 0.024297652766108513, "train_loss_llm": 0.0, "grad_norm": 0.03075335919857025, "global_step": 25349, "epoch": 704, "lr": 0.001} +{"train_loss": 0.030466943979263306, "train_loss_bc": 0.030466943979263306, "train_loss_llm": 0.0, "grad_norm": 0.05767100676894188, "global_step": 25350, "epoch": 704, "lr": 0.001} +{"train_loss": 0.031176162883639336, "train_loss_bc": 0.031176162883639336, "train_loss_llm": 0.0, "grad_norm": 0.08431799709796906, "global_step": 25351, "epoch": 704, "lr": 0.001} +{"train_loss": 0.02985971048474312, "train_loss_bc": 0.02985971048474312, "train_loss_llm": 0.0, "grad_norm": 0.11750299483537674, "global_step": 25352, "epoch": 704, "lr": 0.001} +{"train_loss": 0.02439332939684391, "train_loss_bc": 0.024388549849390984, "train_loss_llm": 0.0004779696464538574, "grad_norm": 0.030745316296815872, "global_step": 25353, "epoch": 704, "lr": 0.001} +{"train_loss": 0.026770835742354393, "train_loss_bc": 0.026770835742354393, "train_loss_llm": 0.0, "grad_norm": 0.052337899804115295, "global_step": 25354, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03373512998223305, "train_loss_bc": 0.0337304025888443, "train_loss_llm": 0.0004728734493255615, "grad_norm": 0.06576865166425705, "global_step": 25355, "epoch": 704, "lr": 0.001} +{"train_loss": 0.033834539353847504, "train_loss_bc": 0.033834539353847504, "train_loss_llm": 0.0, "grad_norm": 0.07794010639190674, "global_step": 25356, "epoch": 704, "lr": 0.001} +{"train_loss": 0.029733261093497276, "train_loss_bc": 0.029733261093497276, "train_loss_llm": 0.0, "grad_norm": 0.03082149475812912, "global_step": 25357, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03104040026664734, "train_loss_bc": 0.03104040026664734, "train_loss_llm": 0.0, "grad_norm": 0.05969837307929993, "global_step": 25358, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03498335927724838, "train_loss_bc": 0.03498335927724838, "train_loss_llm": 0.0, "grad_norm": 0.07856635749340057, "global_step": 25359, "epoch": 704, "lr": 0.001} +{"train_loss": 0.025799252092838287, "train_loss_bc": 0.025799252092838287, "train_loss_llm": 0.0, "grad_norm": 0.08481424301862717, "global_step": 25360, "epoch": 704, "lr": 0.001} +{"train_loss": 0.033852867782115936, "train_loss_bc": 0.033852867782115936, "train_loss_llm": 0.0, "grad_norm": 0.02689790353178978, "global_step": 25361, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03520131856203079, "train_loss_bc": 0.03520131856203079, "train_loss_llm": 0.0, "grad_norm": 0.0530259795486927, "global_step": 25362, "epoch": 704, "lr": 0.001} +{"train_loss": 0.024390671402215958, "train_loss_bc": 0.024390671402215958, "train_loss_llm": 0.0, "grad_norm": 0.0668756365776062, "global_step": 25363, "epoch": 704, "lr": 0.001} +{"train_loss": 0.033950041979551315, "train_loss_bc": 0.03394535556435585, "train_loss_llm": 0.00046876072883605957, "grad_norm": 0.08783848583698273, "global_step": 25364, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03591642528772354, "train_loss_bc": 0.03591642528772354, "train_loss_llm": 0.0, "grad_norm": 0.058139584958553314, "global_step": 25365, "epoch": 704, "lr": 0.001} +{"train_loss": 0.026203341782093048, "train_loss_bc": 0.026201842352747917, "train_loss_llm": 0.0001499950885772705, "grad_norm": 0.06144417077302933, "global_step": 25366, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03152475506067276, "train_loss_bc": 0.03149258717894554, "train_loss_llm": 0.0032168924808502197, "grad_norm": 0.08512204140424728, "global_step": 25367, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03463064506649971, "train_loss_bc": 0.03463064506649971, "train_loss_llm": 0.0, "grad_norm": 0.10071501135826111, "global_step": 25368, "epoch": 704, "lr": 0.001} +{"train_loss": 0.033044278621673584, "train_loss_bc": 0.03303762897849083, "train_loss_llm": 0.0006650090217590332, "grad_norm": 0.04144054651260376, "global_step": 25369, "epoch": 704, "lr": 0.001} +{"train_loss": 0.038518767803907394, "train_loss_bc": 0.038508091121912, "train_loss_llm": 0.00106772780418396, "grad_norm": 0.0807456225156784, "global_step": 25370, "epoch": 704, "lr": 0.001} +{"train_loss": 0.031413957476615906, "train_loss_bc": 0.031413957476615906, "train_loss_llm": 0.0, "grad_norm": 0.09461924433708191, "global_step": 25371, "epoch": 704, "lr": 0.001} +{"train_loss": 0.028650138527154922, "train_loss_bc": 0.028650138527154922, "train_loss_llm": 0.0, "grad_norm": 0.11212374269962311, "global_step": 25372, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03463037684559822, "train_loss_bc": 0.034601084887981415, "train_loss_llm": 0.0029290318489074707, "grad_norm": 0.03673003241419792, "global_step": 25373, "epoch": 704, "lr": 0.001} +{"train_loss": 0.031134240329265594, "train_loss_bc": 0.031134240329265594, "train_loss_llm": 0.0, "grad_norm": 0.06141579896211624, "global_step": 25374, "epoch": 704, "lr": 0.001} +{"train_loss": 0.033614255487918854, "train_loss_bc": 0.033614255487918854, "train_loss_llm": 0.0, "grad_norm": 0.06647589802742004, "global_step": 25375, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03037092089653015, "train_loss_bc": 0.03037092089653015, "train_loss_llm": 0.0, "grad_norm": 0.08687083423137665, "global_step": 25376, "epoch": 704, "lr": 0.001} +{"train_loss": 0.03235737979412079, "train_loss_bc": 0.032339908182621, "train_loss_llm": 0.0017473101615905762, "grad_norm": 0.04473699629306793, "global_step": 25377, "epoch": 704, "lr": 0.001} +{"train_loss": 0.02949511632323265, "train_loss_bc": 0.029471850022673607, "train_loss_llm": 0.0023265480995178223, "grad_norm": 0.05072803422808647, "global_step": 25378, "epoch": 704, "lr": 0.001} +{"train_loss": 0.031385111053370766, "train_loss_bc": 0.026434171944856644, "train_loss_llm": 0.0, "grad_norm": 0.06264737993478775, "global_step": 25379, "epoch": 704, "lr": 0.001, "val_loss": 0.08999604731798172} +{"train_loss": 0.030689559876918793, "train_loss_bc": 0.030689559876918793, "train_loss_llm": 0.0, "grad_norm": 0.07650440186262131, "global_step": 25380, "epoch": 705, "lr": 0.001} +{"train_loss": 0.034006476402282715, "train_loss_bc": 0.03399483114480972, "train_loss_llm": 0.0011644959449768066, "grad_norm": 0.03816523030400276, "global_step": 25381, "epoch": 705, "lr": 0.001} +{"train_loss": 0.02832537516951561, "train_loss_bc": 0.02832537516951561, "train_loss_llm": 0.0, "grad_norm": 0.051015812903642654, "global_step": 25382, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03365117311477661, "train_loss_bc": 0.03365117311477661, "train_loss_llm": 0.0, "grad_norm": 0.06079555302858353, "global_step": 25383, "epoch": 705, "lr": 0.001} +{"train_loss": 0.030659908428788185, "train_loss_bc": 0.030659908428788185, "train_loss_llm": 0.0, "grad_norm": 0.07787160575389862, "global_step": 25384, "epoch": 705, "lr": 0.001} +{"train_loss": 0.028827333822846413, "train_loss_bc": 0.028827333822846413, "train_loss_llm": 0.0, "grad_norm": 0.029935043305158615, "global_step": 25385, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03357962518930435, "train_loss_bc": 0.03357962518930435, "train_loss_llm": 0.0, "grad_norm": 0.05123551934957504, "global_step": 25386, "epoch": 705, "lr": 0.001} +{"train_loss": 0.030233334749937057, "train_loss_bc": 0.030233334749937057, "train_loss_llm": 0.0, "grad_norm": 0.08001979440450668, "global_step": 25387, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03577880188822746, "train_loss_bc": 0.03577880188822746, "train_loss_llm": 0.0, "grad_norm": 0.09636099636554718, "global_step": 25388, "epoch": 705, "lr": 0.001} +{"train_loss": 0.027621524408459663, "train_loss_bc": 0.027621524408459663, "train_loss_llm": 0.0, "grad_norm": 0.03076770342886448, "global_step": 25389, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03329891711473465, "train_loss_bc": 0.03329891711473465, "train_loss_llm": 0.0, "grad_norm": 0.04996064677834511, "global_step": 25390, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03514750301837921, "train_loss_bc": 0.03514750301837921, "train_loss_llm": 0.0, "grad_norm": 0.07384808361530304, "global_step": 25391, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03012729622423649, "train_loss_bc": 0.03012729622423649, "train_loss_llm": 0.0, "grad_norm": 0.09477278590202332, "global_step": 25392, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03168322518467903, "train_loss_bc": 0.03168148547410965, "train_loss_llm": 0.00017401576042175293, "grad_norm": 0.03176717460155487, "global_step": 25393, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03184901177883148, "train_loss_bc": 0.03184901177883148, "train_loss_llm": 0.0, "grad_norm": 0.05421419069170952, "global_step": 25394, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03288250043988228, "train_loss_bc": 0.03288250043988228, "train_loss_llm": 0.0, "grad_norm": 0.08103369921445847, "global_step": 25395, "epoch": 705, "lr": 0.001} +{"train_loss": 0.02743789181113243, "train_loss_bc": 0.02743789181113243, "train_loss_llm": 0.0, "grad_norm": 0.10174631327390671, "global_step": 25396, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03302444517612457, "train_loss_bc": 0.03302444517612457, "train_loss_llm": 0.0, "grad_norm": 0.03487870842218399, "global_step": 25397, "epoch": 705, "lr": 0.001} +{"train_loss": 0.0304153710603714, "train_loss_bc": 0.0304153710603714, "train_loss_llm": 0.0, "grad_norm": 0.04951619729399681, "global_step": 25398, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03498632833361626, "train_loss_bc": 0.03498632833361626, "train_loss_llm": 0.0, "grad_norm": 0.06171635910868645, "global_step": 25399, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03426089510321617, "train_loss_bc": 0.03426089510321617, "train_loss_llm": 0.0, "grad_norm": 0.07267522066831589, "global_step": 25400, "epoch": 705, "lr": 0.001} +{"train_loss": 0.02915176749229431, "train_loss_bc": 0.02915176749229431, "train_loss_llm": 0.0, "grad_norm": 0.03042016178369522, "global_step": 25401, "epoch": 705, "lr": 0.001} +{"train_loss": 0.02856115624308586, "train_loss_bc": 0.02854219451546669, "train_loss_llm": 0.0018960833549499512, "grad_norm": 0.04485302418470383, "global_step": 25402, "epoch": 705, "lr": 0.001} +{"train_loss": 0.032882947474718094, "train_loss_bc": 0.032882947474718094, "train_loss_llm": 0.0, "grad_norm": 0.06517724692821503, "global_step": 25403, "epoch": 705, "lr": 0.001} +{"train_loss": 0.0313325934112072, "train_loss_bc": 0.0313209667801857, "train_loss_llm": 0.001162707805633545, "grad_norm": 0.08010157942771912, "global_step": 25404, "epoch": 705, "lr": 0.001} +{"train_loss": 0.027542414143681526, "train_loss_bc": 0.027542414143681526, "train_loss_llm": 0.0, "grad_norm": 0.032025519758462906, "global_step": 25405, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03044172003865242, "train_loss_bc": 0.03044050559401512, "train_loss_llm": 0.0001214444637298584, "grad_norm": 0.047534577548503876, "global_step": 25406, "epoch": 705, "lr": 0.001} +{"train_loss": 0.033826734870672226, "train_loss_bc": 0.03381135314702988, "train_loss_llm": 0.001538097858428955, "grad_norm": 0.0632045567035675, "global_step": 25407, "epoch": 705, "lr": 0.001} +{"train_loss": 0.02843294106423855, "train_loss_bc": 0.028414355590939522, "train_loss_llm": 0.0018585622310638428, "grad_norm": 0.08622129261493683, "global_step": 25408, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03739991411566734, "train_loss_bc": 0.03739991411566734, "train_loss_llm": 0.0, "grad_norm": 0.0424196794629097, "global_step": 25409, "epoch": 705, "lr": 0.001} +{"train_loss": 0.033222127705812454, "train_loss_bc": 0.03322206810116768, "train_loss_llm": 6.139278411865234e-06, "grad_norm": 0.06571385264396667, "global_step": 25410, "epoch": 705, "lr": 0.001} +{"train_loss": 0.029169712215662003, "train_loss_bc": 0.029169712215662003, "train_loss_llm": 0.0, "grad_norm": 0.08310534805059433, "global_step": 25411, "epoch": 705, "lr": 0.001} +{"train_loss": 0.029704762622714043, "train_loss_bc": 0.029704762622714043, "train_loss_llm": 0.0, "grad_norm": 0.10670183598995209, "global_step": 25412, "epoch": 705, "lr": 0.001} +{"train_loss": 0.03393582999706268, "train_loss_bc": 0.03393237292766571, "train_loss_llm": 0.00034552812576293945, "grad_norm": 0.04163634404540062, "global_step": 25413, "epoch": 705, "lr": 0.001} +{"train_loss": 0.026122255250811577, "train_loss_bc": 0.026105985045433044, "train_loss_llm": 0.0016270279884338379, "grad_norm": 0.06508871912956238, "global_step": 25414, "epoch": 705, "lr": 0.001} +{"train_loss": 0.031426791929536395, "train_loss_bc": 0.03112969733774662, "train_loss_llm": 0.0021437406539916992, "grad_norm": 0.08593886345624924, "global_step": 25415, "epoch": 705, "lr": 0.001, "train/cumulative_reward": 74.57457627525346, "train/mean_score": 15.823631116168135, "train/success_rate": 0.4, "test/cumulative_reward": 32.79179908576139, "test/mean_score": 8.241733723235688, "test/success_rate": 0.18, "val_loss": 0.0897119790315628, "train_action_mse_error": 0.034457605332136154} +{"train_loss": 0.035096220672130585, "train_loss_bc": 0.03508323058485985, "train_loss_llm": 0.0012988746166229248, "grad_norm": 0.11180777847766876, "global_step": 25416, "epoch": 706, "lr": 0.001} +{"train_loss": 0.02728382498025894, "train_loss_bc": 0.02728382498025894, "train_loss_llm": 0.0, "grad_norm": 0.02919887565076351, "global_step": 25417, "epoch": 706, "lr": 0.001} +{"train_loss": 0.02422792837023735, "train_loss_bc": 0.02422792837023735, "train_loss_llm": 0.0, "grad_norm": 0.03798552230000496, "global_step": 25418, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03038725256919861, "train_loss_bc": 0.03038725256919861, "train_loss_llm": 0.0, "grad_norm": 0.05257450416684151, "global_step": 25419, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03389716148376465, "train_loss_bc": 0.03389716148376465, "train_loss_llm": 0.0, "grad_norm": 0.06632868945598602, "global_step": 25420, "epoch": 706, "lr": 0.001} +{"train_loss": 0.031075570732355118, "train_loss_bc": 0.031072299927473068, "train_loss_llm": 0.00032708048820495605, "grad_norm": 0.027029043063521385, "global_step": 25421, "epoch": 706, "lr": 0.001} +{"train_loss": 0.02971472777426243, "train_loss_bc": 0.02971472777426243, "train_loss_llm": 0.0, "grad_norm": 0.04149889200925827, "global_step": 25422, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03011728636920452, "train_loss_bc": 0.03011009842157364, "train_loss_llm": 0.0007187724113464355, "grad_norm": 0.05516324192285538, "global_step": 25423, "epoch": 706, "lr": 0.001} +{"train_loss": 0.031608693301677704, "train_loss_bc": 0.031608693301677704, "train_loss_llm": 0.0, "grad_norm": 0.0667831301689148, "global_step": 25424, "epoch": 706, "lr": 0.001} +{"train_loss": 0.030702315270900726, "train_loss_bc": 0.030702315270900726, "train_loss_llm": 0.0, "grad_norm": 0.03385395556688309, "global_step": 25425, "epoch": 706, "lr": 0.001} +{"train_loss": 0.02856854349374771, "train_loss_bc": 0.02856854349374771, "train_loss_llm": 0.0, "grad_norm": 0.06297201663255692, "global_step": 25426, "epoch": 706, "lr": 0.001} +{"train_loss": 0.030697394162416458, "train_loss_bc": 0.030697394162416458, "train_loss_llm": 0.0, "grad_norm": 0.08608588576316833, "global_step": 25427, "epoch": 706, "lr": 0.001} +{"train_loss": 0.034945297986269, "train_loss_bc": 0.034945297986269, "train_loss_llm": 0.0, "grad_norm": 0.11190619319677353, "global_step": 25428, "epoch": 706, "lr": 0.001} +{"train_loss": 0.0312263872474432, "train_loss_bc": 0.031217124313116074, "train_loss_llm": 0.0009263157844543457, "grad_norm": 0.04237143695354462, "global_step": 25429, "epoch": 706, "lr": 0.001} +{"train_loss": 0.029327183961868286, "train_loss_bc": 0.029310163110494614, "train_loss_llm": 0.0017020106315612793, "grad_norm": 0.06316027790307999, "global_step": 25430, "epoch": 706, "lr": 0.001} +{"train_loss": 0.02945828065276146, "train_loss_bc": 0.02945828065276146, "train_loss_llm": 0.0, "grad_norm": 0.08813841640949249, "global_step": 25431, "epoch": 706, "lr": 0.001} +{"train_loss": 0.02972688525915146, "train_loss_bc": 0.029722899198532104, "train_loss_llm": 0.0003986954689025879, "grad_norm": 0.11110478639602661, "global_step": 25432, "epoch": 706, "lr": 0.001} +{"train_loss": 0.0373646505177021, "train_loss_bc": 0.0373646505177021, "train_loss_llm": 0.0, "grad_norm": 0.03440731391310692, "global_step": 25433, "epoch": 706, "lr": 0.001} +{"train_loss": 0.0346694178879261, "train_loss_bc": 0.0346694178879261, "train_loss_llm": 0.0, "grad_norm": 0.05312090367078781, "global_step": 25434, "epoch": 706, "lr": 0.001} +{"train_loss": 0.02926419861614704, "train_loss_bc": 0.02926419861614704, "train_loss_llm": 0.0, "grad_norm": 0.06571044027805328, "global_step": 25435, "epoch": 706, "lr": 0.001} +{"train_loss": 0.034987181425094604, "train_loss_bc": 0.034987181425094604, "train_loss_llm": 0.0, "grad_norm": 0.07982815057039261, "global_step": 25436, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03110424056649208, "train_loss_bc": 0.03110424056649208, "train_loss_llm": 0.0, "grad_norm": 0.03684081882238388, "global_step": 25437, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03092067316174507, "train_loss_bc": 0.03092067316174507, "train_loss_llm": 0.0, "grad_norm": 0.058194246143102646, "global_step": 25438, "epoch": 706, "lr": 0.001} +{"train_loss": 0.02988121658563614, "train_loss_bc": 0.029860137030482292, "train_loss_llm": 0.002107977867126465, "grad_norm": 0.0720338299870491, "global_step": 25439, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03134242445230484, "train_loss_bc": 0.03134242445230484, "train_loss_llm": 0.0, "grad_norm": 0.08093826472759247, "global_step": 25440, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03608056157827377, "train_loss_bc": 0.03608056157827377, "train_loss_llm": 0.0, "grad_norm": 0.03261011838912964, "global_step": 25441, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03204470872879028, "train_loss_bc": 0.03204470872879028, "train_loss_llm": 0.0, "grad_norm": 0.053156644105911255, "global_step": 25442, "epoch": 706, "lr": 0.001} +{"train_loss": 0.028733639046549797, "train_loss_bc": 0.028733639046549797, "train_loss_llm": 0.0, "grad_norm": 0.07289570569992065, "global_step": 25443, "epoch": 706, "lr": 0.001} +{"train_loss": 0.0333859845995903, "train_loss_bc": 0.0333859845995903, "train_loss_llm": 0.0, "grad_norm": 0.09041251987218857, "global_step": 25444, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03436015173792839, "train_loss_bc": 0.034354016184806824, "train_loss_llm": 0.0006134212017059326, "grad_norm": 0.033193331211805344, "global_step": 25445, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03146759793162346, "train_loss_bc": 0.03146759793162346, "train_loss_llm": 0.0, "grad_norm": 0.04962100833654404, "global_step": 25446, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03264864534139633, "train_loss_bc": 0.03263816609978676, "train_loss_llm": 0.001047968864440918, "grad_norm": 0.07296539098024368, "global_step": 25447, "epoch": 706, "lr": 0.001} +{"train_loss": 0.0295998714864254, "train_loss_bc": 0.0295998714864254, "train_loss_llm": 0.0, "grad_norm": 0.08411339670419693, "global_step": 25448, "epoch": 706, "lr": 0.001} +{"train_loss": 0.03383929654955864, "train_loss_bc": 0.03381384536623955, "train_loss_llm": 0.002545088529586792, "grad_norm": 0.038812942802906036, "global_step": 25449, "epoch": 706, "lr": 0.001} +{"train_loss": 0.02702905796468258, "train_loss_bc": 0.02702905796468258, "train_loss_llm": 0.0, "grad_norm": 0.05362313985824585, "global_step": 25450, "epoch": 706, "lr": 0.001} +{"train_loss": 0.031298088271998696, "train_loss_bc": 0.02994001843035221, "train_loss_llm": 0.0006686151027679443, "grad_norm": 0.06714864075183868, "global_step": 25451, "epoch": 706, "lr": 0.001, "val_loss": 0.09156569838523865} +{"train_loss": 0.031222566962242126, "train_loss_bc": 0.03119470179080963, "train_loss_llm": 0.002786487340927124, "grad_norm": 0.08642186224460602, "global_step": 25452, "epoch": 707, "lr": 0.001} +{"train_loss": 0.027857761830091476, "train_loss_bc": 0.027857761830091476, "train_loss_llm": 0.0, "grad_norm": 0.029217040166258812, "global_step": 25453, "epoch": 707, "lr": 0.001} +{"train_loss": 0.027479542419314384, "train_loss_bc": 0.027479542419314384, "train_loss_llm": 0.0, "grad_norm": 0.04470197483897209, "global_step": 25454, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03167266771197319, "train_loss_bc": 0.03167266771197319, "train_loss_llm": 0.0, "grad_norm": 0.06713971495628357, "global_step": 25455, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03001919761300087, "train_loss_bc": 0.030017320066690445, "train_loss_llm": 0.00018772482872009277, "grad_norm": 0.08229565620422363, "global_step": 25456, "epoch": 707, "lr": 0.001} +{"train_loss": 0.036722395569086075, "train_loss_bc": 0.036718934774398804, "train_loss_llm": 0.00034621357917785645, "grad_norm": 0.03975534066557884, "global_step": 25457, "epoch": 707, "lr": 0.001} +{"train_loss": 0.026043636724352837, "train_loss_bc": 0.026043636724352837, "train_loss_llm": 0.0, "grad_norm": 0.050263911485672, "global_step": 25458, "epoch": 707, "lr": 0.001} +{"train_loss": 0.028669938445091248, "train_loss_bc": 0.028669938445091248, "train_loss_llm": 0.0, "grad_norm": 0.06496760994195938, "global_step": 25459, "epoch": 707, "lr": 0.001} +{"train_loss": 0.029275067150592804, "train_loss_bc": 0.029275067150592804, "train_loss_llm": 0.0, "grad_norm": 0.0846751257777214, "global_step": 25460, "epoch": 707, "lr": 0.001} +{"train_loss": 0.0333397313952446, "train_loss_bc": 0.0333397313952446, "train_loss_llm": 0.0, "grad_norm": 0.03759181499481201, "global_step": 25461, "epoch": 707, "lr": 0.001} +{"train_loss": 0.02903340756893158, "train_loss_bc": 0.02903340756893158, "train_loss_llm": 0.0, "grad_norm": 0.05780500918626785, "global_step": 25462, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03131140395998955, "train_loss_bc": 0.03131140395998955, "train_loss_llm": 0.0, "grad_norm": 0.08566183596849442, "global_step": 25463, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03072727657854557, "train_loss_bc": 0.030707262456417084, "train_loss_llm": 0.002001464366912842, "grad_norm": 0.11576137691736221, "global_step": 25464, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03352263569831848, "train_loss_bc": 0.03352263569831848, "train_loss_llm": 0.0, "grad_norm": 0.03018238954246044, "global_step": 25465, "epoch": 707, "lr": 0.001} +{"train_loss": 0.02674783207476139, "train_loss_bc": 0.02674783207476139, "train_loss_llm": 0.0, "grad_norm": 0.054661162197589874, "global_step": 25466, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03080182522535324, "train_loss_bc": 0.03080085664987564, "train_loss_llm": 9.691715240478516e-05, "grad_norm": 0.06863849610090256, "global_step": 25467, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03542301803827286, "train_loss_bc": 0.03542301803827286, "train_loss_llm": 0.0, "grad_norm": 0.0805533304810524, "global_step": 25468, "epoch": 707, "lr": 0.001} +{"train_loss": 0.027830958366394043, "train_loss_bc": 0.027822988107800484, "train_loss_llm": 0.0007970631122589111, "grad_norm": 0.03087100200355053, "global_step": 25469, "epoch": 707, "lr": 0.001} +{"train_loss": 0.029816443100571632, "train_loss_bc": 0.029803162440657616, "train_loss_llm": 0.001328110694885254, "grad_norm": 0.05206039547920227, "global_step": 25470, "epoch": 707, "lr": 0.001} +{"train_loss": 0.0323839895427227, "train_loss_bc": 0.0323839895427227, "train_loss_llm": 0.0, "grad_norm": 0.067723847925663, "global_step": 25471, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03477318957448006, "train_loss_bc": 0.03477318957448006, "train_loss_llm": 0.0, "grad_norm": 0.07941730320453644, "global_step": 25472, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03388969972729683, "train_loss_bc": 0.03388969972729683, "train_loss_llm": 0.0, "grad_norm": 0.034163009375333786, "global_step": 25473, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03276215121150017, "train_loss_bc": 0.03276215121150017, "train_loss_llm": 0.0, "grad_norm": 0.05508521944284439, "global_step": 25474, "epoch": 707, "lr": 0.001} +{"train_loss": 0.037462662905454636, "train_loss_bc": 0.037462662905454636, "train_loss_llm": 0.0, "grad_norm": 0.07788848131895065, "global_step": 25475, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03406066447496414, "train_loss_bc": 0.03404688835144043, "train_loss_llm": 0.001377791166305542, "grad_norm": 0.10252294689416885, "global_step": 25476, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03470732644200325, "train_loss_bc": 0.03470732644200325, "train_loss_llm": 0.0, "grad_norm": 0.030554622411727905, "global_step": 25477, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03537367656826973, "train_loss_bc": 0.03537367656826973, "train_loss_llm": 0.0, "grad_norm": 0.05007752403616905, "global_step": 25478, "epoch": 707, "lr": 0.001} +{"train_loss": 0.028268931433558464, "train_loss_bc": 0.028268931433558464, "train_loss_llm": 0.0, "grad_norm": 0.05570191517472267, "global_step": 25479, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03223447874188423, "train_loss_bc": 0.03223447874188423, "train_loss_llm": 0.0, "grad_norm": 0.06589414924383163, "global_step": 25480, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03037680685520172, "train_loss_bc": 0.03037680685520172, "train_loss_llm": 0.0, "grad_norm": 0.037195902317762375, "global_step": 25481, "epoch": 707, "lr": 0.001} +{"train_loss": 0.029455944895744324, "train_loss_bc": 0.029455944895744324, "train_loss_llm": 0.0, "grad_norm": 0.04506112262606621, "global_step": 25482, "epoch": 707, "lr": 0.001} +{"train_loss": 0.02921075001358986, "train_loss_bc": 0.02921075001358986, "train_loss_llm": 0.0, "grad_norm": 0.05777047947049141, "global_step": 25483, "epoch": 707, "lr": 0.001} +{"train_loss": 0.03322652727365494, "train_loss_bc": 0.03322652727365494, "train_loss_llm": 0.0, "grad_norm": 0.07787302136421204, "global_step": 25484, "epoch": 707, "lr": 0.001} +{"train_loss": 0.028369441628456116, "train_loss_bc": 0.028369441628456116, "train_loss_llm": 0.0, "grad_norm": 0.030962489545345306, "global_step": 25485, "epoch": 707, "lr": 0.001} +{"train_loss": 0.026552755385637283, "train_loss_bc": 0.026552755385637283, "train_loss_llm": 0.0, "grad_norm": 0.04633674770593643, "global_step": 25486, "epoch": 707, "lr": 0.001} +{"train_loss": 0.031339209940698415, "train_loss_bc": 0.03758525475859642, "train_loss_llm": 0.0, "grad_norm": 0.07625273615121841, "global_step": 25487, "epoch": 707, "lr": 0.001, "val_loss": 0.08923112601041794} +{"train_loss": 0.033497072756290436, "train_loss_bc": 0.033497072756290436, "train_loss_llm": 0.0, "grad_norm": 0.09156699478626251, "global_step": 25488, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03315877541899681, "train_loss_bc": 0.03315877541899681, "train_loss_llm": 0.0, "grad_norm": 0.033501505851745605, "global_step": 25489, "epoch": 708, "lr": 0.001} +{"train_loss": 0.031575899571180344, "train_loss_bc": 0.031575899571180344, "train_loss_llm": 0.0, "grad_norm": 0.045819781720638275, "global_step": 25490, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03360666707158089, "train_loss_bc": 0.03360385820269585, "train_loss_llm": 0.0002810359001159668, "grad_norm": 0.06470469385385513, "global_step": 25491, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03468586131930351, "train_loss_bc": 0.0346793495118618, "train_loss_llm": 0.0006512105464935303, "grad_norm": 0.07636237889528275, "global_step": 25492, "epoch": 708, "lr": 0.001} +{"train_loss": 0.034349896013736725, "train_loss_bc": 0.034349896013736725, "train_loss_llm": 0.0, "grad_norm": 0.03327232971787453, "global_step": 25493, "epoch": 708, "lr": 0.001} +{"train_loss": 0.02905982919037342, "train_loss_bc": 0.02905982919037342, "train_loss_llm": 0.0, "grad_norm": 0.047500357031822205, "global_step": 25494, "epoch": 708, "lr": 0.001} +{"train_loss": 0.031136296689510345, "train_loss_bc": 0.031136296689510345, "train_loss_llm": 0.0, "grad_norm": 0.07720005512237549, "global_step": 25495, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03196890652179718, "train_loss_bc": 0.03196890652179718, "train_loss_llm": 0.0, "grad_norm": 0.08663053810596466, "global_step": 25496, "epoch": 708, "lr": 0.001} +{"train_loss": 0.030785784125328064, "train_loss_bc": 0.030785784125328064, "train_loss_llm": 0.0, "grad_norm": 0.0260790828615427, "global_step": 25497, "epoch": 708, "lr": 0.001} +{"train_loss": 0.02864946983754635, "train_loss_bc": 0.02864946983754635, "train_loss_llm": 0.0, "grad_norm": 0.042727384716272354, "global_step": 25498, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03247921168804169, "train_loss_bc": 0.03247921168804169, "train_loss_llm": 0.0, "grad_norm": 0.05717131122946739, "global_step": 25499, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03048258274793625, "train_loss_bc": 0.03048258274793625, "train_loss_llm": 0.0, "grad_norm": 0.06305526196956635, "global_step": 25500, "epoch": 708, "lr": 0.001} +{"train_loss": 0.032114967703819275, "train_loss_bc": 0.032114967703819275, "train_loss_llm": 0.0, "grad_norm": 0.03183882310986519, "global_step": 25501, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03332771360874176, "train_loss_bc": 0.03332771360874176, "train_loss_llm": 0.0, "grad_norm": 0.049751535058021545, "global_step": 25502, "epoch": 708, "lr": 0.001} +{"train_loss": 0.032630328088998795, "train_loss_bc": 0.03260071203112602, "train_loss_llm": 0.002961456775665283, "grad_norm": 0.07140886038541794, "global_step": 25503, "epoch": 708, "lr": 0.001} +{"train_loss": 0.028813805431127548, "train_loss_bc": 0.028813805431127548, "train_loss_llm": 0.0, "grad_norm": 0.08676133304834366, "global_step": 25504, "epoch": 708, "lr": 0.001} +{"train_loss": 0.026203501969575882, "train_loss_bc": 0.026203501969575882, "train_loss_llm": 0.0, "grad_norm": 0.032608762383461, "global_step": 25505, "epoch": 708, "lr": 0.001} +{"train_loss": 0.02703225240111351, "train_loss_bc": 0.027027349919080734, "train_loss_llm": 0.0004903078079223633, "grad_norm": 0.042260732501745224, "global_step": 25506, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03414847329258919, "train_loss_bc": 0.03414847329258919, "train_loss_llm": 0.0, "grad_norm": 0.051674067974090576, "global_step": 25507, "epoch": 708, "lr": 0.001} +{"train_loss": 0.032540179789066315, "train_loss_bc": 0.032540179789066315, "train_loss_llm": 0.0, "grad_norm": 0.06131384149193764, "global_step": 25508, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03633023053407669, "train_loss_bc": 0.036304108798503876, "train_loss_llm": 0.0026121437549591064, "grad_norm": 0.031123381108045578, "global_step": 25509, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03128569945693016, "train_loss_bc": 0.03128569945693016, "train_loss_llm": 0.0, "grad_norm": 0.04870050027966499, "global_step": 25510, "epoch": 708, "lr": 0.001} +{"train_loss": 0.031590577214956284, "train_loss_bc": 0.031590577214956284, "train_loss_llm": 0.0, "grad_norm": 0.06111106276512146, "global_step": 25511, "epoch": 708, "lr": 0.001} +{"train_loss": 0.028117815032601357, "train_loss_bc": 0.028110576793551445, "train_loss_llm": 0.0007237493991851807, "grad_norm": 0.07542630285024643, "global_step": 25512, "epoch": 708, "lr": 0.001} +{"train_loss": 0.032836247235536575, "train_loss_bc": 0.032836247235536575, "train_loss_llm": 0.0, "grad_norm": 0.030001206323504448, "global_step": 25513, "epoch": 708, "lr": 0.001} +{"train_loss": 0.0366516038775444, "train_loss_bc": 0.0366516038775444, "train_loss_llm": 0.0, "grad_norm": 0.042952314019203186, "global_step": 25514, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03065161406993866, "train_loss_bc": 0.03065161406993866, "train_loss_llm": 0.0, "grad_norm": 0.05896467715501785, "global_step": 25515, "epoch": 708, "lr": 0.001} +{"train_loss": 0.025731904432177544, "train_loss_bc": 0.025731904432177544, "train_loss_llm": 0.0, "grad_norm": 0.07653472572565079, "global_step": 25516, "epoch": 708, "lr": 0.001} +{"train_loss": 0.0255690049380064, "train_loss_bc": 0.0255690049380064, "train_loss_llm": 0.0, "grad_norm": 0.032769203186035156, "global_step": 25517, "epoch": 708, "lr": 0.001} +{"train_loss": 0.02783825807273388, "train_loss_bc": 0.02783825807273388, "train_loss_llm": 0.0, "grad_norm": 0.04578772932291031, "global_step": 25518, "epoch": 708, "lr": 0.001} +{"train_loss": 0.02885008230805397, "train_loss_bc": 0.02885008230805397, "train_loss_llm": 0.0, "grad_norm": 0.05544479936361313, "global_step": 25519, "epoch": 708, "lr": 0.001} +{"train_loss": 0.03135480359196663, "train_loss_bc": 0.031350743025541306, "train_loss_llm": 0.00040593743324279785, "grad_norm": 0.06081673875451088, "global_step": 25520, "epoch": 708, "lr": 0.001} +{"train_loss": 0.028300702571868896, "train_loss_bc": 0.028295893222093582, "train_loss_llm": 0.00048092007637023926, "grad_norm": 0.03300841897726059, "global_step": 25521, "epoch": 708, "lr": 0.001} +{"train_loss": 0.029084550216794014, "train_loss_bc": 0.02908237837255001, "train_loss_llm": 0.00021716952323913574, "grad_norm": 0.04924110695719719, "global_step": 25522, "epoch": 708, "lr": 0.001} +{"train_loss": 0.031225003043396607, "train_loss_bc": 0.03765954077243805, "train_loss_llm": 0.0, "grad_norm": 0.08759837597608566, "global_step": 25523, "epoch": 708, "lr": 0.001, "val_loss": 0.08929914236068726} +{"train_loss": 0.02760634757578373, "train_loss_bc": 0.02760634757578373, "train_loss_llm": 0.0, "grad_norm": 0.09622449427843094, "global_step": 25524, "epoch": 709, "lr": 0.001} +{"train_loss": 0.030996331945061684, "train_loss_bc": 0.030996331945061684, "train_loss_llm": 0.0, "grad_norm": 0.03080935776233673, "global_step": 25525, "epoch": 709, "lr": 0.001} +{"train_loss": 0.032088398933410645, "train_loss_bc": 0.03208611533045769, "train_loss_llm": 0.0002282857894897461, "grad_norm": 0.039802923798561096, "global_step": 25526, "epoch": 709, "lr": 0.001} +{"train_loss": 0.03245969116687775, "train_loss_bc": 0.03245969116687775, "train_loss_llm": 0.0, "grad_norm": 0.05183779075741768, "global_step": 25527, "epoch": 709, "lr": 0.001} +{"train_loss": 0.036476705223321915, "train_loss_bc": 0.036476705223321915, "train_loss_llm": 0.0, "grad_norm": 0.06565945595502853, "global_step": 25528, "epoch": 709, "lr": 0.001} +{"train_loss": 0.03566589206457138, "train_loss_bc": 0.03566589206457138, "train_loss_llm": 0.0, "grad_norm": 0.03152526542544365, "global_step": 25529, "epoch": 709, "lr": 0.001} +{"train_loss": 0.03112293966114521, "train_loss_bc": 0.03112293966114521, "train_loss_llm": 0.0, "grad_norm": 0.05265706032514572, "global_step": 25530, "epoch": 709, "lr": 0.001} +{"train_loss": 0.03238222002983093, "train_loss_bc": 0.03238222002983093, "train_loss_llm": 0.0, "grad_norm": 0.06612537056207657, "global_step": 25531, "epoch": 709, "lr": 0.001} +{"train_loss": 0.030860770493745804, "train_loss_bc": 0.030860770493745804, "train_loss_llm": 0.0, "grad_norm": 0.07624433189630508, "global_step": 25532, "epoch": 709, "lr": 0.001} +{"train_loss": 0.02876953035593033, "train_loss_bc": 0.02876953035593033, "train_loss_llm": 0.0, "grad_norm": 0.02988487109541893, "global_step": 25533, "epoch": 709, "lr": 0.001} +{"train_loss": 0.035351864993572235, "train_loss_bc": 0.035351864993572235, "train_loss_llm": 0.0, "grad_norm": 0.04593076556921005, "global_step": 25534, "epoch": 709, "lr": 0.001} +{"train_loss": 0.0293306652456522, "train_loss_bc": 0.0293306652456522, "train_loss_llm": 0.0, "grad_norm": 0.07075490802526474, "global_step": 25535, "epoch": 709, "lr": 0.001} +{"train_loss": 0.026904599741101265, "train_loss_bc": 0.026904599741101265, "train_loss_llm": 0.0, "grad_norm": 0.08153070509433746, "global_step": 25536, "epoch": 709, "lr": 0.001} +{"train_loss": 0.02734960801899433, "train_loss_bc": 0.027341440320014954, "train_loss_llm": 0.0008167624473571777, "grad_norm": 0.029604854062199593, "global_step": 25537, "epoch": 709, "lr": 0.001} +{"train_loss": 0.03278147429227829, "train_loss_bc": 0.03278147429227829, "train_loss_llm": 0.0, "grad_norm": 0.05724874511361122, "global_step": 25538, "epoch": 709, "lr": 0.001} +{"train_loss": 0.033079199492931366, "train_loss_bc": 0.033079199492931366, "train_loss_llm": 0.0, "grad_norm": 0.08265001326799393, "global_step": 25539, "epoch": 709, "lr": 0.001} +{"train_loss": 0.03374525159597397, "train_loss_bc": 0.03373291343450546, "train_loss_llm": 0.0012337863445281982, "grad_norm": 0.09375970810651779, "global_step": 25540, "epoch": 709, "lr": 0.001} +{"train_loss": 0.030954638496041298, "train_loss_bc": 0.030954638496041298, "train_loss_llm": 0.0, "grad_norm": 0.03429611027240753, "global_step": 25541, "epoch": 709, "lr": 0.001} +{"train_loss": 0.03792206570506096, "train_loss_bc": 0.03792206570506096, "train_loss_llm": 0.0, "grad_norm": 0.05681004002690315, "global_step": 25542, "epoch": 709, "lr": 0.001} +{"train_loss": 0.028077181428670883, "train_loss_bc": 0.028065714985132217, "train_loss_llm": 0.0011466741561889648, "grad_norm": 0.06645332276821136, "global_step": 25543, "epoch": 709, "lr": 0.001} +{"train_loss": 0.02968628890812397, "train_loss_bc": 0.029669126495718956, "train_loss_llm": 0.0017161667346954346, "grad_norm": 0.0734158456325531, "global_step": 25544, "epoch": 709, "lr": 0.001} +{"train_loss": 0.030073702335357666, "train_loss_bc": 0.030073702335357666, "train_loss_llm": 0.0, "grad_norm": 0.02844761498272419, "global_step": 25545, "epoch": 709, "lr": 0.001} +{"train_loss": 0.026199020445346832, "train_loss_bc": 0.026187485083937645, "train_loss_llm": 0.0011535882949829102, "grad_norm": 0.03999185189604759, "global_step": 25546, "epoch": 709, "lr": 0.001} +{"train_loss": 0.032784443348646164, "train_loss_bc": 0.032784443348646164, "train_loss_llm": 0.0, "grad_norm": 0.05223919451236725, "global_step": 25547, "epoch": 709, "lr": 0.001} +{"train_loss": 0.032352618873119354, "train_loss_bc": 0.032352618873119354, "train_loss_llm": 0.0, "grad_norm": 0.0643400326371193, "global_step": 25548, "epoch": 709, "lr": 0.001} +{"train_loss": 0.02882581576704979, "train_loss_bc": 0.028815878555178642, "train_loss_llm": 0.0009937286376953125, "grad_norm": 0.03047834150493145, "global_step": 25549, "epoch": 709, "lr": 0.001} +{"train_loss": 0.02880989946424961, "train_loss_bc": 0.02880989946424961, "train_loss_llm": 0.0, "grad_norm": 0.04095015302300453, "global_step": 25550, "epoch": 709, "lr": 0.001} +{"train_loss": 0.02917782962322235, "train_loss_bc": 0.02917782962322235, "train_loss_llm": 0.0, "grad_norm": 0.049433618783950806, "global_step": 25551, "epoch": 709, "lr": 0.001} +{"train_loss": 0.034942954778671265, "train_loss_bc": 0.034942954778671265, "train_loss_llm": 0.0, "grad_norm": 0.06957253813743591, "global_step": 25552, "epoch": 709, "lr": 0.001} +{"train_loss": 0.029160872101783752, "train_loss_bc": 0.029160872101783752, "train_loss_llm": 0.0, "grad_norm": 0.024402838200330734, "global_step": 25553, "epoch": 709, "lr": 0.001} +{"train_loss": 0.03159020468592644, "train_loss_bc": 0.03157437592744827, "train_loss_llm": 0.0015829503536224365, "grad_norm": 0.0400223508477211, "global_step": 25554, "epoch": 709, "lr": 0.001} +{"train_loss": 0.02751944586634636, "train_loss_bc": 0.02751944586634636, "train_loss_llm": 0.0, "grad_norm": 0.05321183055639267, "global_step": 25555, "epoch": 709, "lr": 0.001} +{"train_loss": 0.029678136110305786, "train_loss_bc": 0.029666529968380928, "train_loss_llm": 0.0011605620384216309, "grad_norm": 0.0647333562374115, "global_step": 25556, "epoch": 709, "lr": 0.001} +{"train_loss": 0.02856219746172428, "train_loss_bc": 0.02856219746172428, "train_loss_llm": 0.0, "grad_norm": 0.033961884677410126, "global_step": 25557, "epoch": 709, "lr": 0.001} +{"train_loss": 0.030986571684479713, "train_loss_bc": 0.030986571684479713, "train_loss_llm": 0.0, "grad_norm": 0.05519833043217659, "global_step": 25558, "epoch": 709, "lr": 0.001} +{"train_loss": 0.03115575196635392, "train_loss_bc": 0.03732515126466751, "train_loss_llm": 0.0006542801856994629, "grad_norm": 0.08834870159626007, "global_step": 25559, "epoch": 709, "lr": 0.001, "val_loss": 0.09041306376457214} +{"train_loss": 0.028339682146906853, "train_loss_bc": 0.028339682146906853, "train_loss_llm": 0.0, "grad_norm": 0.0994889885187149, "global_step": 25560, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03818192705512047, "train_loss_bc": 0.03818192705512047, "train_loss_llm": 0.0, "grad_norm": 0.033481232821941376, "global_step": 25561, "epoch": 710, "lr": 0.001} +{"train_loss": 0.02566000632941723, "train_loss_bc": 0.02566000632941723, "train_loss_llm": 0.0, "grad_norm": 0.04256817325949669, "global_step": 25562, "epoch": 710, "lr": 0.001} +{"train_loss": 0.0332045778632164, "train_loss_bc": 0.033199336379766464, "train_loss_llm": 0.0005242228507995605, "grad_norm": 0.05553276091814041, "global_step": 25563, "epoch": 710, "lr": 0.001} +{"train_loss": 0.028488943353295326, "train_loss_bc": 0.028488943353295326, "train_loss_llm": 0.0, "grad_norm": 0.07040515542030334, "global_step": 25564, "epoch": 710, "lr": 0.001} +{"train_loss": 0.028027508407831192, "train_loss_bc": 0.028027508407831192, "train_loss_llm": 0.0, "grad_norm": 0.027899840846657753, "global_step": 25565, "epoch": 710, "lr": 0.001} +{"train_loss": 0.035042695701122284, "train_loss_bc": 0.035042695701122284, "train_loss_llm": 0.0, "grad_norm": 0.054354071617126465, "global_step": 25566, "epoch": 710, "lr": 0.001} +{"train_loss": 0.032498352229595184, "train_loss_bc": 0.03248231112957001, "train_loss_llm": 0.0016039609909057617, "grad_norm": 0.0690225213766098, "global_step": 25567, "epoch": 710, "lr": 0.001} +{"train_loss": 0.032804060727357864, "train_loss_bc": 0.032804060727357864, "train_loss_llm": 0.0, "grad_norm": 0.07900463789701462, "global_step": 25568, "epoch": 710, "lr": 0.001} +{"train_loss": 0.033085912466049194, "train_loss_bc": 0.033085912466049194, "train_loss_llm": 0.0, "grad_norm": 0.03405594080686569, "global_step": 25569, "epoch": 710, "lr": 0.001} +{"train_loss": 0.02921757474541664, "train_loss_bc": 0.02921757474541664, "train_loss_llm": 0.0, "grad_norm": 0.054935526102781296, "global_step": 25570, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03216426819562912, "train_loss_bc": 0.03216426819562912, "train_loss_llm": 0.0, "grad_norm": 0.06616917252540588, "global_step": 25571, "epoch": 710, "lr": 0.001} +{"train_loss": 0.027587980031967163, "train_loss_bc": 0.027587980031967163, "train_loss_llm": 0.0, "grad_norm": 0.07719304412603378, "global_step": 25572, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03137307986617088, "train_loss_bc": 0.03137307986617088, "train_loss_llm": 0.0, "grad_norm": 0.03844147175550461, "global_step": 25573, "epoch": 710, "lr": 0.001} +{"train_loss": 0.033366575837135315, "train_loss_bc": 0.0333622582256794, "train_loss_llm": 0.0004316568374633789, "grad_norm": 0.053514789789915085, "global_step": 25574, "epoch": 710, "lr": 0.001} +{"train_loss": 0.02803957834839821, "train_loss_bc": 0.028031548485159874, "train_loss_llm": 0.0008029937744140625, "grad_norm": 0.07235245406627655, "global_step": 25575, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03329908847808838, "train_loss_bc": 0.03329908847808838, "train_loss_llm": 0.0, "grad_norm": 0.09199723601341248, "global_step": 25576, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03199387714266777, "train_loss_bc": 0.03199387714266777, "train_loss_llm": 0.0, "grad_norm": 0.032798558473587036, "global_step": 25577, "epoch": 710, "lr": 0.001} +{"train_loss": 0.031214920803904533, "train_loss_bc": 0.03120707906782627, "train_loss_llm": 0.0007842481136322021, "grad_norm": 0.0499398373067379, "global_step": 25578, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03297053277492523, "train_loss_bc": 0.03297053277492523, "train_loss_llm": 0.0, "grad_norm": 0.0775889903306961, "global_step": 25579, "epoch": 710, "lr": 0.001} +{"train_loss": 0.028112251311540604, "train_loss_bc": 0.028104394674301147, "train_loss_llm": 0.0007856488227844238, "grad_norm": 0.08894188702106476, "global_step": 25580, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03391564264893532, "train_loss_bc": 0.03391256555914879, "train_loss_llm": 0.0003076493740081787, "grad_norm": 0.03299233689904213, "global_step": 25581, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03076772205531597, "train_loss_bc": 0.030766956508159637, "train_loss_llm": 7.650256156921387e-05, "grad_norm": 0.04690919443964958, "global_step": 25582, "epoch": 710, "lr": 0.001} +{"train_loss": 0.026501448825001717, "train_loss_bc": 0.026494812220335007, "train_loss_llm": 0.0006636381149291992, "grad_norm": 0.06411481648683548, "global_step": 25583, "epoch": 710, "lr": 0.001} +{"train_loss": 0.030832070857286453, "train_loss_bc": 0.030832070857286453, "train_loss_llm": 0.0, "grad_norm": 0.07876229286193848, "global_step": 25584, "epoch": 710, "lr": 0.001} +{"train_loss": 0.02926485240459442, "train_loss_bc": 0.029263818636536598, "train_loss_llm": 0.00010332465171813965, "grad_norm": 0.03115600347518921, "global_step": 25585, "epoch": 710, "lr": 0.001} +{"train_loss": 0.029508667066693306, "train_loss_bc": 0.02949136681854725, "train_loss_llm": 0.0017300248146057129, "grad_norm": 0.04513397067785263, "global_step": 25586, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03426168113946915, "train_loss_bc": 0.03426168113946915, "train_loss_llm": 0.0, "grad_norm": 0.062159955501556396, "global_step": 25587, "epoch": 710, "lr": 0.001} +{"train_loss": 0.02898099087178707, "train_loss_bc": 0.02898099087178707, "train_loss_llm": 0.0, "grad_norm": 0.0761646181344986, "global_step": 25588, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03276804834604263, "train_loss_bc": 0.03276804834604263, "train_loss_llm": 0.0, "grad_norm": 0.03179977089166641, "global_step": 25589, "epoch": 710, "lr": 0.001} +{"train_loss": 0.02693289704620838, "train_loss_bc": 0.02693289704620838, "train_loss_llm": 0.0, "grad_norm": 0.0533677339553833, "global_step": 25590, "epoch": 710, "lr": 0.001} +{"train_loss": 0.029899122193455696, "train_loss_bc": 0.029899122193455696, "train_loss_llm": 0.0, "grad_norm": 0.07336989790201187, "global_step": 25591, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03851401060819626, "train_loss_bc": 0.038504041731357574, "train_loss_llm": 0.0009970664978027344, "grad_norm": 0.08845751732587814, "global_step": 25592, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03227980434894562, "train_loss_bc": 0.03227980434894562, "train_loss_llm": 0.0, "grad_norm": 0.03168230503797531, "global_step": 25593, "epoch": 710, "lr": 0.001} +{"train_loss": 0.03421485051512718, "train_loss_bc": 0.03421485051512718, "train_loss_llm": 0.0, "grad_norm": 0.04265475273132324, "global_step": 25594, "epoch": 710, "lr": 0.001} +{"train_loss": 0.031069435955335695, "train_loss_bc": 0.025175368413329124, "train_loss_llm": 0.0009123682975769043, "grad_norm": 0.056714605540037155, "global_step": 25595, "epoch": 710, "lr": 0.001, "train/cumulative_reward": 47.73153567151397, "train/mean_score": 7.615427681244005, "train/success_rate": 0.2, "test/cumulative_reward": 49.11247264583383, "test/mean_score": 11.12543520685282, "test/success_rate": 0.26, "val_loss": 0.09077457338571548, "train_action_mse_error": 0.03458525985479355} +{"train_loss": 0.026643767952919006, "train_loss_bc": 0.026643767952919006, "train_loss_llm": 0.0, "grad_norm": 0.07332036644220352, "global_step": 25596, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03375701233744621, "train_loss_bc": 0.03375596925616264, "train_loss_llm": 0.00010445713996887207, "grad_norm": 0.0371607206761837, "global_step": 25597, "epoch": 711, "lr": 0.001} +{"train_loss": 0.028721274808049202, "train_loss_bc": 0.028721274808049202, "train_loss_llm": 0.0, "grad_norm": 0.06023843213915825, "global_step": 25598, "epoch": 711, "lr": 0.001} +{"train_loss": 0.02857324108481407, "train_loss_bc": 0.02857324108481407, "train_loss_llm": 0.0, "grad_norm": 0.0790998563170433, "global_step": 25599, "epoch": 711, "lr": 0.001} +{"train_loss": 0.030266478657722473, "train_loss_bc": 0.030253348872065544, "train_loss_llm": 0.0013130605220794678, "grad_norm": 0.09777816385030746, "global_step": 25600, "epoch": 711, "lr": 0.001} +{"train_loss": 0.035166118294000626, "train_loss_bc": 0.03513967990875244, "train_loss_llm": 0.0026438534259796143, "grad_norm": 0.03189493715763092, "global_step": 25601, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03553245589137077, "train_loss_bc": 0.03553245589137077, "train_loss_llm": 0.0, "grad_norm": 0.040918175131082535, "global_step": 25602, "epoch": 711, "lr": 0.001} +{"train_loss": 0.02879994921386242, "train_loss_bc": 0.02879994921386242, "train_loss_llm": 0.0, "grad_norm": 0.05303071066737175, "global_step": 25603, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03089849278330803, "train_loss_bc": 0.03089849278330803, "train_loss_llm": 0.0, "grad_norm": 0.06944072246551514, "global_step": 25604, "epoch": 711, "lr": 0.001} +{"train_loss": 0.032724108546972275, "train_loss_bc": 0.032714106142520905, "train_loss_llm": 0.0010001063346862793, "grad_norm": 0.036468639969825745, "global_step": 25605, "epoch": 711, "lr": 0.001} +{"train_loss": 0.02632122114300728, "train_loss_bc": 0.02632122114300728, "train_loss_llm": 0.0, "grad_norm": 0.056392569094896317, "global_step": 25606, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03081131912767887, "train_loss_bc": 0.03081131912767887, "train_loss_llm": 0.0, "grad_norm": 0.07553686946630478, "global_step": 25607, "epoch": 711, "lr": 0.001} +{"train_loss": 0.029136206954717636, "train_loss_bc": 0.029136206954717636, "train_loss_llm": 0.0, "grad_norm": 0.09028127789497375, "global_step": 25608, "epoch": 711, "lr": 0.001} +{"train_loss": 0.02855970524251461, "train_loss_bc": 0.02855970524251461, "train_loss_llm": 0.0, "grad_norm": 0.032186124473810196, "global_step": 25609, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03386810049414635, "train_loss_bc": 0.03386810049414635, "train_loss_llm": 0.0, "grad_norm": 0.05373165383934975, "global_step": 25610, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03631165251135826, "train_loss_bc": 0.03631165251135826, "train_loss_llm": 0.0, "grad_norm": 0.072734534740448, "global_step": 25611, "epoch": 711, "lr": 0.001} +{"train_loss": 0.029841599985957146, "train_loss_bc": 0.029841599985957146, "train_loss_llm": 0.0, "grad_norm": 0.0956488698720932, "global_step": 25612, "epoch": 711, "lr": 0.001} +{"train_loss": 0.035394344478845596, "train_loss_bc": 0.03536094352602959, "train_loss_llm": 0.003340214490890503, "grad_norm": 0.042838405817747116, "global_step": 25613, "epoch": 711, "lr": 0.001} +{"train_loss": 0.032101891934871674, "train_loss_bc": 0.03209627419710159, "train_loss_llm": 0.0005618631839752197, "grad_norm": 0.057135265320539474, "global_step": 25614, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03360218554735184, "train_loss_bc": 0.03360218554735184, "train_loss_llm": 0.0, "grad_norm": 0.07666955888271332, "global_step": 25615, "epoch": 711, "lr": 0.001} +{"train_loss": 0.032575320452451706, "train_loss_bc": 0.032575320452451706, "train_loss_llm": 0.0, "grad_norm": 0.09849459677934647, "global_step": 25616, "epoch": 711, "lr": 0.001} +{"train_loss": 0.031813185662031174, "train_loss_bc": 0.031813185662031174, "train_loss_llm": 0.0, "grad_norm": 0.03328821063041687, "global_step": 25617, "epoch": 711, "lr": 0.001} +{"train_loss": 0.028495285660028458, "train_loss_bc": 0.028495285660028458, "train_loss_llm": 0.0, "grad_norm": 0.05214507877826691, "global_step": 25618, "epoch": 711, "lr": 0.001} +{"train_loss": 0.025996893644332886, "train_loss_bc": 0.02598053775727749, "train_loss_llm": 0.0016356110572814941, "grad_norm": 0.06933098286390305, "global_step": 25619, "epoch": 711, "lr": 0.001} +{"train_loss": 0.02729090116918087, "train_loss_bc": 0.02729090116918087, "train_loss_llm": 0.0, "grad_norm": 0.08147953450679779, "global_step": 25620, "epoch": 711, "lr": 0.001} +{"train_loss": 0.026751693338155746, "train_loss_bc": 0.026751190423965454, "train_loss_llm": 5.02467155456543e-05, "grad_norm": 0.031982988119125366, "global_step": 25621, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03234581649303436, "train_loss_bc": 0.03234581649303436, "train_loss_llm": 0.0, "grad_norm": 0.042456988245248795, "global_step": 25622, "epoch": 711, "lr": 0.001} +{"train_loss": 0.038026176393032074, "train_loss_bc": 0.038005903363227844, "train_loss_llm": 0.0020272433757781982, "grad_norm": 0.06374692916870117, "global_step": 25623, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03342059627175331, "train_loss_bc": 0.03342059627175331, "train_loss_llm": 0.0, "grad_norm": 0.07719036936759949, "global_step": 25624, "epoch": 711, "lr": 0.001} +{"train_loss": 0.0314948633313179, "train_loss_bc": 0.0314948633313179, "train_loss_llm": 0.0, "grad_norm": 0.039832521229982376, "global_step": 25625, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03478283807635307, "train_loss_bc": 0.03478283807635307, "train_loss_llm": 0.0, "grad_norm": 0.05749401077628136, "global_step": 25626, "epoch": 711, "lr": 0.001} +{"train_loss": 0.029280709102749825, "train_loss_bc": 0.0292768906801939, "train_loss_llm": 0.00038191676139831543, "grad_norm": 0.08311337232589722, "global_step": 25627, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03928573802113533, "train_loss_bc": 0.03928573802113533, "train_loss_llm": 0.0, "grad_norm": 0.1061684712767601, "global_step": 25628, "epoch": 711, "lr": 0.001} +{"train_loss": 0.03151164576411247, "train_loss_bc": 0.03151055797934532, "train_loss_llm": 0.00010886788368225098, "grad_norm": 0.025334838777780533, "global_step": 25629, "epoch": 711, "lr": 0.001} +{"train_loss": 0.029795927926898003, "train_loss_bc": 0.029795927926898003, "train_loss_llm": 0.0, "grad_norm": 0.04529527947306633, "global_step": 25630, "epoch": 711, "lr": 0.001} +{"train_loss": 0.031148666360725958, "train_loss_bc": 0.021453270688652992, "train_loss_llm": 0.0, "grad_norm": 0.05492235720157623, "global_step": 25631, "epoch": 711, "lr": 0.001, "val_loss": 0.08917497843503952} +{"train_loss": 0.03198571875691414, "train_loss_bc": 0.03198571875691414, "train_loss_llm": 0.0, "grad_norm": 0.06797121465206146, "global_step": 25632, "epoch": 712, "lr": 0.001} +{"train_loss": 0.028857549652457237, "train_loss_bc": 0.028857549652457237, "train_loss_llm": 0.0, "grad_norm": 0.02879173681139946, "global_step": 25633, "epoch": 712, "lr": 0.001} +{"train_loss": 0.028193561360239983, "train_loss_bc": 0.028193561360239983, "train_loss_llm": 0.0, "grad_norm": 0.04065887629985809, "global_step": 25634, "epoch": 712, "lr": 0.001} +{"train_loss": 0.027187012135982513, "train_loss_bc": 0.027187012135982513, "train_loss_llm": 0.0, "grad_norm": 0.06132541969418526, "global_step": 25635, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03387250378727913, "train_loss_bc": 0.03387250378727913, "train_loss_llm": 0.0, "grad_norm": 0.0640987828373909, "global_step": 25636, "epoch": 712, "lr": 0.001} +{"train_loss": 0.031478241086006165, "train_loss_bc": 0.031478241086006165, "train_loss_llm": 0.0, "grad_norm": 0.028784634545445442, "global_step": 25637, "epoch": 712, "lr": 0.001} +{"train_loss": 0.028577269986271858, "train_loss_bc": 0.028577269986271858, "train_loss_llm": 0.0, "grad_norm": 0.049975331872701645, "global_step": 25638, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03093850426375866, "train_loss_bc": 0.030936937779188156, "train_loss_llm": 0.00015658140182495117, "grad_norm": 0.06136900186538696, "global_step": 25639, "epoch": 712, "lr": 0.001} +{"train_loss": 0.02915246970951557, "train_loss_bc": 0.02915246970951557, "train_loss_llm": 0.0, "grad_norm": 0.06801588833332062, "global_step": 25640, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03814629092812538, "train_loss_bc": 0.03814629092812538, "train_loss_llm": 0.0, "grad_norm": 0.04126875847578049, "global_step": 25641, "epoch": 712, "lr": 0.001} +{"train_loss": 0.04032520577311516, "train_loss_bc": 0.04032520577311516, "train_loss_llm": 0.0, "grad_norm": 0.05773257836699486, "global_step": 25642, "epoch": 712, "lr": 0.001} +{"train_loss": 0.029232896864414215, "train_loss_bc": 0.02922065183520317, "train_loss_llm": 0.001224517822265625, "grad_norm": 0.07192052155733109, "global_step": 25643, "epoch": 712, "lr": 0.001} +{"train_loss": 0.029870932921767235, "train_loss_bc": 0.02986515872180462, "train_loss_llm": 0.0005774796009063721, "grad_norm": 0.08990512788295746, "global_step": 25644, "epoch": 712, "lr": 0.001} +{"train_loss": 0.027423981577157974, "train_loss_bc": 0.027421502396464348, "train_loss_llm": 0.000247955322265625, "grad_norm": 0.02665822021663189, "global_step": 25645, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03529621660709381, "train_loss_bc": 0.03529621660709381, "train_loss_llm": 0.0, "grad_norm": 0.0481872595846653, "global_step": 25646, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03057153709232807, "train_loss_bc": 0.03057153709232807, "train_loss_llm": 0.0, "grad_norm": 0.060786452144384384, "global_step": 25647, "epoch": 712, "lr": 0.001} +{"train_loss": 0.031666528433561325, "train_loss_bc": 0.031666528433561325, "train_loss_llm": 0.0, "grad_norm": 0.07199636101722717, "global_step": 25648, "epoch": 712, "lr": 0.001} +{"train_loss": 0.033595409244298935, "train_loss_bc": 0.033595409244298935, "train_loss_llm": 0.0, "grad_norm": 0.03372780606150627, "global_step": 25649, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03289439529180527, "train_loss_bc": 0.03289439529180527, "train_loss_llm": 0.0, "grad_norm": 0.04814499244093895, "global_step": 25650, "epoch": 712, "lr": 0.001} +{"train_loss": 0.032122474163770676, "train_loss_bc": 0.032112423330545425, "train_loss_llm": 0.0010051131248474121, "grad_norm": 0.06234009563922882, "global_step": 25651, "epoch": 712, "lr": 0.001} +{"train_loss": 0.027980607002973557, "train_loss_bc": 0.027980607002973557, "train_loss_llm": 0.0, "grad_norm": 0.08175468444824219, "global_step": 25652, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03140392526984215, "train_loss_bc": 0.03140392526984215, "train_loss_llm": 0.0, "grad_norm": 0.02989344485104084, "global_step": 25653, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03531277924776077, "train_loss_bc": 0.03531216084957123, "train_loss_llm": 6.189942359924316e-05, "grad_norm": 0.06375070661306381, "global_step": 25654, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03143696114420891, "train_loss_bc": 0.03143696114420891, "train_loss_llm": 0.0, "grad_norm": 0.09318108856678009, "global_step": 25655, "epoch": 712, "lr": 0.001} +{"train_loss": 0.025927914306521416, "train_loss_bc": 0.025927914306521416, "train_loss_llm": 0.0, "grad_norm": 0.11691485345363617, "global_step": 25656, "epoch": 712, "lr": 0.001} +{"train_loss": 0.026348253712058067, "train_loss_bc": 0.026336507871747017, "train_loss_llm": 0.0011745095252990723, "grad_norm": 0.031171096488833427, "global_step": 25657, "epoch": 712, "lr": 0.001} +{"train_loss": 0.0272564385086298, "train_loss_bc": 0.0272564385086298, "train_loss_llm": 0.0, "grad_norm": 0.045802488923072815, "global_step": 25658, "epoch": 712, "lr": 0.001} +{"train_loss": 0.035991471260786057, "train_loss_bc": 0.035991471260786057, "train_loss_llm": 0.0, "grad_norm": 0.0667482241988182, "global_step": 25659, "epoch": 712, "lr": 0.001} +{"train_loss": 0.031360823661088943, "train_loss_bc": 0.031360823661088943, "train_loss_llm": 0.0, "grad_norm": 0.09152832627296448, "global_step": 25660, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03066924400627613, "train_loss_bc": 0.03066924400627613, "train_loss_llm": 0.0, "grad_norm": 0.03761691600084305, "global_step": 25661, "epoch": 712, "lr": 0.001} +{"train_loss": 0.030140675604343414, "train_loss_bc": 0.030140675604343414, "train_loss_llm": 0.0, "grad_norm": 0.05705224350094795, "global_step": 25662, "epoch": 712, "lr": 0.001} +{"train_loss": 0.03490539267659187, "train_loss_bc": 0.03490539267659187, "train_loss_llm": 0.0, "grad_norm": 0.06506185233592987, "global_step": 25663, "epoch": 712, "lr": 0.001} +{"train_loss": 0.034198954701423645, "train_loss_bc": 0.034198954701423645, "train_loss_llm": 0.0, "grad_norm": 0.08416903018951416, "global_step": 25664, "epoch": 712, "lr": 0.001} +{"train_loss": 0.02608686313033104, "train_loss_bc": 0.026086358353495598, "train_loss_llm": 5.0455331802368164e-05, "grad_norm": 0.029634464532136917, "global_step": 25665, "epoch": 712, "lr": 0.001} +{"train_loss": 0.033081330358982086, "train_loss_bc": 0.033081330358982086, "train_loss_llm": 0.0, "grad_norm": 0.04018944874405861, "global_step": 25666, "epoch": 712, "lr": 0.001} +{"train_loss": 0.031151750331951514, "train_loss_bc": 0.02796621434390545, "train_loss_llm": 0.0006462633609771729, "grad_norm": 0.05848720669746399, "global_step": 25667, "epoch": 712, "lr": 0.001, "val_loss": 0.09036752581596375} +{"train_loss": 0.028050070628523827, "train_loss_bc": 0.028050070628523827, "train_loss_llm": 0.0, "grad_norm": 0.07485867291688919, "global_step": 25668, "epoch": 713, "lr": 0.001} +{"train_loss": 0.029105374589562416, "train_loss_bc": 0.029105374589562416, "train_loss_llm": 0.0, "grad_norm": 0.03313164412975311, "global_step": 25669, "epoch": 713, "lr": 0.001} +{"train_loss": 0.02920437976717949, "train_loss_bc": 0.02920437976717949, "train_loss_llm": 0.0, "grad_norm": 0.04854931682348251, "global_step": 25670, "epoch": 713, "lr": 0.001} +{"train_loss": 0.031484588980674744, "train_loss_bc": 0.031478144228458405, "train_loss_llm": 0.0006445646286010742, "grad_norm": 0.07180683314800262, "global_step": 25671, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03584437817335129, "train_loss_bc": 0.03584437817335129, "train_loss_llm": 0.0, "grad_norm": 0.0946815237402916, "global_step": 25672, "epoch": 713, "lr": 0.001} +{"train_loss": 0.036245714873075485, "train_loss_bc": 0.036245714873075485, "train_loss_llm": 0.0, "grad_norm": 0.03821146860718727, "global_step": 25673, "epoch": 713, "lr": 0.001} +{"train_loss": 0.030761267989873886, "train_loss_bc": 0.030761267989873886, "train_loss_llm": 0.0, "grad_norm": 0.06675251573324203, "global_step": 25674, "epoch": 713, "lr": 0.001} +{"train_loss": 0.02884713187813759, "train_loss_bc": 0.02884713187813759, "train_loss_llm": 0.0, "grad_norm": 0.0864546149969101, "global_step": 25675, "epoch": 713, "lr": 0.001} +{"train_loss": 0.031497810035943985, "train_loss_bc": 0.031497810035943985, "train_loss_llm": 0.0, "grad_norm": 0.11432898789644241, "global_step": 25676, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03128979727625847, "train_loss_bc": 0.03128979727625847, "train_loss_llm": 0.0, "grad_norm": 0.03716116398572922, "global_step": 25677, "epoch": 713, "lr": 0.001} +{"train_loss": 0.031389329582452774, "train_loss_bc": 0.031389329582452774, "train_loss_llm": 0.0, "grad_norm": 0.059532102197408676, "global_step": 25678, "epoch": 713, "lr": 0.001} +{"train_loss": 0.0316401943564415, "train_loss_bc": 0.0316401943564415, "train_loss_llm": 0.0, "grad_norm": 0.07315099984407425, "global_step": 25679, "epoch": 713, "lr": 0.001} +{"train_loss": 0.02856585942208767, "train_loss_bc": 0.02856585942208767, "train_loss_llm": 0.0, "grad_norm": 0.09295859932899475, "global_step": 25680, "epoch": 713, "lr": 0.001} +{"train_loss": 0.029293615370988846, "train_loss_bc": 0.029290352016687393, "train_loss_llm": 0.00032636523246765137, "grad_norm": 0.029570305719971657, "global_step": 25681, "epoch": 713, "lr": 0.001} +{"train_loss": 0.034993018954992294, "train_loss_bc": 0.034989964216947556, "train_loss_llm": 0.00030547380447387695, "grad_norm": 0.06044211983680725, "global_step": 25682, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03522935509681702, "train_loss_bc": 0.0352267287671566, "train_loss_llm": 0.0002626478672027588, "grad_norm": 0.08309512585401535, "global_step": 25683, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03195083513855934, "train_loss_bc": 0.03194432705640793, "train_loss_llm": 0.0006509721279144287, "grad_norm": 0.1046360433101654, "global_step": 25684, "epoch": 713, "lr": 0.001} +{"train_loss": 0.033032551407814026, "train_loss_bc": 0.03303075581789017, "train_loss_llm": 0.00017973780632019043, "grad_norm": 0.04584595188498497, "global_step": 25685, "epoch": 713, "lr": 0.001} +{"train_loss": 0.030102306976914406, "train_loss_bc": 0.030102306976914406, "train_loss_llm": 0.0, "grad_norm": 0.06182321906089783, "global_step": 25686, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03167552500963211, "train_loss_bc": 0.03167552500963211, "train_loss_llm": 0.0, "grad_norm": 0.08503615856170654, "global_step": 25687, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03160901740193367, "train_loss_bc": 0.03160901740193367, "train_loss_llm": 0.0, "grad_norm": 0.10546069592237473, "global_step": 25688, "epoch": 713, "lr": 0.001} +{"train_loss": 0.030529510229825974, "train_loss_bc": 0.030529510229825974, "train_loss_llm": 0.0, "grad_norm": 0.02717990055680275, "global_step": 25689, "epoch": 713, "lr": 0.001} +{"train_loss": 0.02641889452934265, "train_loss_bc": 0.02641889452934265, "train_loss_llm": 0.0, "grad_norm": 0.04665270075201988, "global_step": 25690, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03326234966516495, "train_loss_bc": 0.03326234966516495, "train_loss_llm": 0.0, "grad_norm": 0.06463127583265305, "global_step": 25691, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03525814414024353, "train_loss_bc": 0.035249728709459305, "train_loss_llm": 0.0008414685726165771, "grad_norm": 0.09125486761331558, "global_step": 25692, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03237641602754593, "train_loss_bc": 0.03237641602754593, "train_loss_llm": 0.0, "grad_norm": 0.03530024737119675, "global_step": 25693, "epoch": 713, "lr": 0.001} +{"train_loss": 0.031019816175103188, "train_loss_bc": 0.031019816175103188, "train_loss_llm": 0.0, "grad_norm": 0.060167647898197174, "global_step": 25694, "epoch": 713, "lr": 0.001} +{"train_loss": 0.030437909066677094, "train_loss_bc": 0.03043481521308422, "train_loss_llm": 0.00030934810638427734, "grad_norm": 0.08183862268924713, "global_step": 25695, "epoch": 713, "lr": 0.001} +{"train_loss": 0.025185471400618553, "train_loss_bc": 0.025185471400618553, "train_loss_llm": 0.0, "grad_norm": 0.10868081450462341, "global_step": 25696, "epoch": 713, "lr": 0.001} +{"train_loss": 0.0329766683280468, "train_loss_bc": 0.0329766683280468, "train_loss_llm": 0.0, "grad_norm": 0.03196931630373001, "global_step": 25697, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03264614939689636, "train_loss_bc": 0.03264614939689636, "train_loss_llm": 0.0, "grad_norm": 0.048179786652326584, "global_step": 25698, "epoch": 713, "lr": 0.001} +{"train_loss": 0.02585570514202118, "train_loss_bc": 0.02585570514202118, "train_loss_llm": 0.0, "grad_norm": 0.06212720274925232, "global_step": 25699, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03059322200715542, "train_loss_bc": 0.0305740125477314, "train_loss_llm": 0.0019208788871765137, "grad_norm": 0.06952322274446487, "global_step": 25700, "epoch": 713, "lr": 0.001} +{"train_loss": 0.029704738408327103, "train_loss_bc": 0.029687410220503807, "train_loss_llm": 0.0017327368259429932, "grad_norm": 0.03049320913851261, "global_step": 25701, "epoch": 713, "lr": 0.001} +{"train_loss": 0.03995968773961067, "train_loss_bc": 0.03995968773961067, "train_loss_llm": 0.0, "grad_norm": 0.056920312345027924, "global_step": 25702, "epoch": 713, "lr": 0.001} +{"train_loss": 0.031380798615929156, "train_loss_bc": 0.03166475519537926, "train_loss_llm": 0.0007191598415374756, "grad_norm": 0.09384983777999878, "global_step": 25703, "epoch": 713, "lr": 0.001, "val_loss": 0.09050776064395905} +{"train_loss": 0.024144621565937996, "train_loss_bc": 0.024126490578055382, "train_loss_llm": 0.0018131732940673828, "grad_norm": 0.10969245433807373, "global_step": 25704, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03275897353887558, "train_loss_bc": 0.03275011479854584, "train_loss_llm": 0.0008859336376190186, "grad_norm": 0.04703928530216217, "global_step": 25705, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03331174701452255, "train_loss_bc": 0.03329402580857277, "train_loss_llm": 0.0017721056938171387, "grad_norm": 0.06795509904623032, "global_step": 25706, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03398682177066803, "train_loss_bc": 0.03398682177066803, "train_loss_llm": 0.0, "grad_norm": 0.0952395498752594, "global_step": 25707, "epoch": 714, "lr": 0.001} +{"train_loss": 0.030527912080287933, "train_loss_bc": 0.030527912080287933, "train_loss_llm": 0.0, "grad_norm": 0.11706355214118958, "global_step": 25708, "epoch": 714, "lr": 0.001} +{"train_loss": 0.028395235538482666, "train_loss_bc": 0.028395235538482666, "train_loss_llm": 0.0, "grad_norm": 0.03614247217774391, "global_step": 25709, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03542314097285271, "train_loss_bc": 0.03542314097285271, "train_loss_llm": 0.0, "grad_norm": 0.06527479737997055, "global_step": 25710, "epoch": 714, "lr": 0.001} +{"train_loss": 0.0343632735311985, "train_loss_bc": 0.034351419657468796, "train_loss_llm": 0.0011853575706481934, "grad_norm": 0.09187179803848267, "global_step": 25711, "epoch": 714, "lr": 0.001} +{"train_loss": 0.0330481082201004, "train_loss_bc": 0.0330481082201004, "train_loss_llm": 0.0, "grad_norm": 0.1063922867178917, "global_step": 25712, "epoch": 714, "lr": 0.001} +{"train_loss": 0.02925094962120056, "train_loss_bc": 0.02925094962120056, "train_loss_llm": 0.0, "grad_norm": 0.031374115496873856, "global_step": 25713, "epoch": 714, "lr": 0.001} +{"train_loss": 0.038617897778749466, "train_loss_bc": 0.038617897778749466, "train_loss_llm": 0.0, "grad_norm": 0.06949545443058014, "global_step": 25714, "epoch": 714, "lr": 0.001} +{"train_loss": 0.031183850020170212, "train_loss_bc": 0.031165171414613724, "train_loss_llm": 0.0018678903579711914, "grad_norm": 0.08590128272771835, "global_step": 25715, "epoch": 714, "lr": 0.001} +{"train_loss": 0.0323605053126812, "train_loss_bc": 0.0323605053126812, "train_loss_llm": 0.0, "grad_norm": 0.10150346159934998, "global_step": 25716, "epoch": 714, "lr": 0.001} +{"train_loss": 0.026282764971256256, "train_loss_bc": 0.026282764971256256, "train_loss_llm": 0.0, "grad_norm": 0.02912251278758049, "global_step": 25717, "epoch": 714, "lr": 0.001} +{"train_loss": 0.02528957463800907, "train_loss_bc": 0.02528913877904415, "train_loss_llm": 4.363059997558594e-05, "grad_norm": 0.0422615185379982, "global_step": 25718, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03222689777612686, "train_loss_bc": 0.03222689777612686, "train_loss_llm": 0.0, "grad_norm": 0.059854261577129364, "global_step": 25719, "epoch": 714, "lr": 0.001} +{"train_loss": 0.031741026788949966, "train_loss_bc": 0.031741026788949966, "train_loss_llm": 0.0, "grad_norm": 0.07949323207139969, "global_step": 25720, "epoch": 714, "lr": 0.001} +{"train_loss": 0.027987075969576836, "train_loss_bc": 0.027987075969576836, "train_loss_llm": 0.0, "grad_norm": 0.03383903205394745, "global_step": 25721, "epoch": 714, "lr": 0.001} +{"train_loss": 0.032873425632715225, "train_loss_bc": 0.032873425632715225, "train_loss_llm": 0.0, "grad_norm": 0.04964228346943855, "global_step": 25722, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03272746875882149, "train_loss_bc": 0.03272746875882149, "train_loss_llm": 0.0, "grad_norm": 0.06945134699344635, "global_step": 25723, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03159036487340927, "train_loss_bc": 0.031579792499542236, "train_loss_llm": 0.0010574162006378174, "grad_norm": 0.10122301429510117, "global_step": 25724, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03450344502925873, "train_loss_bc": 0.03450344502925873, "train_loss_llm": 0.0, "grad_norm": 0.04029235988855362, "global_step": 25725, "epoch": 714, "lr": 0.001} +{"train_loss": 0.02844899334013462, "train_loss_bc": 0.02844899334013462, "train_loss_llm": 0.0, "grad_norm": 0.05856465920805931, "global_step": 25726, "epoch": 714, "lr": 0.001} +{"train_loss": 0.027059119194746017, "train_loss_bc": 0.027059119194746017, "train_loss_llm": 0.0, "grad_norm": 0.08358773589134216, "global_step": 25727, "epoch": 714, "lr": 0.001} +{"train_loss": 0.034575849771499634, "train_loss_bc": 0.034575849771499634, "train_loss_llm": 0.0, "grad_norm": 0.1059441864490509, "global_step": 25728, "epoch": 714, "lr": 0.001} +{"train_loss": 0.028588080778717995, "train_loss_bc": 0.028566874563694, "train_loss_llm": 0.002120673656463623, "grad_norm": 0.0347854346036911, "global_step": 25729, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03150897100567818, "train_loss_bc": 0.03148704767227173, "train_loss_llm": 0.0021923184394836426, "grad_norm": 0.05524137243628502, "global_step": 25730, "epoch": 714, "lr": 0.001} +{"train_loss": 0.02824511006474495, "train_loss_bc": 0.02824511006474495, "train_loss_llm": 0.0, "grad_norm": 0.07106456160545349, "global_step": 25731, "epoch": 714, "lr": 0.001} +{"train_loss": 0.030739882960915565, "train_loss_bc": 0.030739882960915565, "train_loss_llm": 0.0, "grad_norm": 0.08962516486644745, "global_step": 25732, "epoch": 714, "lr": 0.001} +{"train_loss": 0.02819799818098545, "train_loss_bc": 0.028195807710289955, "train_loss_llm": 0.00021904706954956055, "grad_norm": 0.039325159043073654, "global_step": 25733, "epoch": 714, "lr": 0.001} +{"train_loss": 0.038062404841184616, "train_loss_bc": 0.03804558143019676, "train_loss_llm": 0.0016822218894958496, "grad_norm": 0.054372381418943405, "global_step": 25734, "epoch": 714, "lr": 0.001} +{"train_loss": 0.0317954421043396, "train_loss_bc": 0.0317954421043396, "train_loss_llm": 0.0, "grad_norm": 0.07820617407560349, "global_step": 25735, "epoch": 714, "lr": 0.001} +{"train_loss": 0.028305603191256523, "train_loss_bc": 0.028299039229750633, "train_loss_llm": 0.000656425952911377, "grad_norm": 0.11138483136892319, "global_step": 25736, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03926612064242363, "train_loss_bc": 0.03926612064242363, "train_loss_llm": 0.0, "grad_norm": 0.041480861604213715, "global_step": 25737, "epoch": 714, "lr": 0.001} +{"train_loss": 0.03782576695084572, "train_loss_bc": 0.03782576695084572, "train_loss_llm": 0.0, "grad_norm": 0.07215063273906708, "global_step": 25738, "epoch": 714, "lr": 0.001} +{"train_loss": 0.031356000600175724, "train_loss_bc": 0.023600397631525993, "train_loss_llm": 0.00012001395225524902, "grad_norm": 0.09082462638616562, "global_step": 25739, "epoch": 714, "lr": 0.001, "val_loss": 0.09234841912984848} +{"train_loss": 0.025079505518078804, "train_loss_bc": 0.025079505518078804, "train_loss_llm": 0.0, "grad_norm": 0.11409652978181839, "global_step": 25740, "epoch": 715, "lr": 0.001} +{"train_loss": 0.032296888530254364, "train_loss_bc": 0.032296888530254364, "train_loss_llm": 0.0, "grad_norm": 0.035884592682123184, "global_step": 25741, "epoch": 715, "lr": 0.001} +{"train_loss": 0.027592282742261887, "train_loss_bc": 0.027592282742261887, "train_loss_llm": 0.0, "grad_norm": 0.06724312901496887, "global_step": 25742, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03240747004747391, "train_loss_bc": 0.032404180616140366, "train_loss_llm": 0.00032895803451538086, "grad_norm": 0.09178575128316879, "global_step": 25743, "epoch": 715, "lr": 0.001} +{"train_loss": 0.026622526347637177, "train_loss_bc": 0.02661691978573799, "train_loss_llm": 0.0005606114864349365, "grad_norm": 0.11332488805055618, "global_step": 25744, "epoch": 715, "lr": 0.001} +{"train_loss": 0.030151108279824257, "train_loss_bc": 0.030150890350341797, "train_loss_llm": 2.1845102310180664e-05, "grad_norm": 0.028518011793494225, "global_step": 25745, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03326673433184624, "train_loss_bc": 0.03326673433184624, "train_loss_llm": 0.0, "grad_norm": 0.041100334376096725, "global_step": 25746, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03448408469557762, "train_loss_bc": 0.03448408469557762, "train_loss_llm": 0.0, "grad_norm": 0.06477010250091553, "global_step": 25747, "epoch": 715, "lr": 0.001} +{"train_loss": 0.028589880093932152, "train_loss_bc": 0.028589880093932152, "train_loss_llm": 0.0, "grad_norm": 0.0731205865740776, "global_step": 25748, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03405485674738884, "train_loss_bc": 0.03405485674738884, "train_loss_llm": 0.0, "grad_norm": 0.032945819199085236, "global_step": 25749, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03419644385576248, "train_loss_bc": 0.03419644385576248, "train_loss_llm": 0.0, "grad_norm": 0.05262000486254692, "global_step": 25750, "epoch": 715, "lr": 0.001} +{"train_loss": 0.033414535224437714, "train_loss_bc": 0.03340836986899376, "train_loss_llm": 0.0006163716316223145, "grad_norm": 0.059712860733270645, "global_step": 25751, "epoch": 715, "lr": 0.001} +{"train_loss": 0.025554701685905457, "train_loss_bc": 0.025553885847330093, "train_loss_llm": 8.156895637512207e-05, "grad_norm": 0.07070302218198776, "global_step": 25752, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03265365585684776, "train_loss_bc": 0.03265365585684776, "train_loss_llm": 0.0, "grad_norm": 0.03368700295686722, "global_step": 25753, "epoch": 715, "lr": 0.001} +{"train_loss": 0.031175939366221428, "train_loss_bc": 0.031175939366221428, "train_loss_llm": 0.0, "grad_norm": 0.07839346677064896, "global_step": 25754, "epoch": 715, "lr": 0.001} +{"train_loss": 0.026079382747411728, "train_loss_bc": 0.026079382747411728, "train_loss_llm": 0.0, "grad_norm": 0.09704519808292389, "global_step": 25755, "epoch": 715, "lr": 0.001} +{"train_loss": 0.030071919783949852, "train_loss_bc": 0.030071919783949852, "train_loss_llm": 0.0, "grad_norm": 0.14067159593105316, "global_step": 25756, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03240344300866127, "train_loss_bc": 0.03240344300866127, "train_loss_llm": 0.0, "grad_norm": 0.03912069648504257, "global_step": 25757, "epoch": 715, "lr": 0.001} +{"train_loss": 0.029780976474285126, "train_loss_bc": 0.029773514717817307, "train_loss_llm": 0.0007461309432983398, "grad_norm": 0.056497108191251755, "global_step": 25758, "epoch": 715, "lr": 0.001} +{"train_loss": 0.029314689338207245, "train_loss_bc": 0.029314106330275536, "train_loss_llm": 5.8323144912719727e-05, "grad_norm": 0.07695706933736801, "global_step": 25759, "epoch": 715, "lr": 0.001} +{"train_loss": 0.032243214547634125, "train_loss_bc": 0.032243214547634125, "train_loss_llm": 0.0, "grad_norm": 0.09470389038324356, "global_step": 25760, "epoch": 715, "lr": 0.001} +{"train_loss": 0.025339780375361443, "train_loss_bc": 0.025338266044855118, "train_loss_llm": 0.00015145540237426758, "grad_norm": 0.028986506164073944, "global_step": 25761, "epoch": 715, "lr": 0.001} +{"train_loss": 0.02443239465355873, "train_loss_bc": 0.02443239465355873, "train_loss_llm": 0.0, "grad_norm": 0.05026628077030182, "global_step": 25762, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03630738705396652, "train_loss_bc": 0.03630738705396652, "train_loss_llm": 0.0, "grad_norm": 0.07089048624038696, "global_step": 25763, "epoch": 715, "lr": 0.001} +{"train_loss": 0.02459236979484558, "train_loss_bc": 0.02459236979484558, "train_loss_llm": 0.0, "grad_norm": 0.08957362920045853, "global_step": 25764, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03331596031785011, "train_loss_bc": 0.033306919038295746, "train_loss_llm": 0.0009041726589202881, "grad_norm": 0.03789530694484711, "global_step": 25765, "epoch": 715, "lr": 0.001} +{"train_loss": 0.034819748252630234, "train_loss_bc": 0.034819748252630234, "train_loss_llm": 0.0, "grad_norm": 0.05589183047413826, "global_step": 25766, "epoch": 715, "lr": 0.001} +{"train_loss": 0.033191241323947906, "train_loss_bc": 0.033191241323947906, "train_loss_llm": 0.0, "grad_norm": 0.0815519317984581, "global_step": 25767, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03464962914586067, "train_loss_bc": 0.03464962914586067, "train_loss_llm": 0.0, "grad_norm": 0.10131452232599258, "global_step": 25768, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03497948870062828, "train_loss_bc": 0.03497948870062828, "train_loss_llm": 0.0, "grad_norm": 0.030577341094613075, "global_step": 25769, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03171467408537865, "train_loss_bc": 0.03171467408537865, "train_loss_llm": 0.0, "grad_norm": 0.0536075197160244, "global_step": 25770, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03784901648759842, "train_loss_bc": 0.03784901648759842, "train_loss_llm": 0.0, "grad_norm": 0.07581575214862823, "global_step": 25771, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03675714135169983, "train_loss_bc": 0.03675714135169983, "train_loss_llm": 0.0, "grad_norm": 0.0973043218255043, "global_step": 25772, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03480500727891922, "train_loss_bc": 0.03480500727891922, "train_loss_llm": 0.0, "grad_norm": 0.025774035602808, "global_step": 25773, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03416988626122475, "train_loss_bc": 0.03416988626122475, "train_loss_llm": 0.0, "grad_norm": 0.047551319003105164, "global_step": 25774, "epoch": 715, "lr": 0.001} +{"train_loss": 0.03124406902740399, "train_loss_bc": 0.026428520679473877, "train_loss_llm": 0.0, "grad_norm": 0.06267935037612915, "global_step": 25775, "epoch": 715, "lr": 0.001, "train/cumulative_reward": 60.093963375069336, "train/mean_score": 6.717314674261999, "train/success_rate": 0.1, "test/cumulative_reward": 40.62886801691518, "test/mean_score": 9.635862243864437, "test/success_rate": 0.26, "val_loss": 0.08973377197980881, "train_action_mse_error": 0.03379182144999504} +{"train_loss": 0.03096126951277256, "train_loss_bc": 0.030953051522374153, "train_loss_llm": 0.0008218288421630859, "grad_norm": 0.07286807149648666, "global_step": 25776, "epoch": 716, "lr": 0.001} +{"train_loss": 0.028348887339234352, "train_loss_bc": 0.028330054134130478, "train_loss_llm": 0.0018832385540008545, "grad_norm": 0.027783837169408798, "global_step": 25777, "epoch": 716, "lr": 0.001} +{"train_loss": 0.02926909364759922, "train_loss_bc": 0.02926909364759922, "train_loss_llm": 0.0, "grad_norm": 0.0474911667406559, "global_step": 25778, "epoch": 716, "lr": 0.001} +{"train_loss": 0.03493588790297508, "train_loss_bc": 0.03493588790297508, "train_loss_llm": 0.0, "grad_norm": 0.07603752613067627, "global_step": 25779, "epoch": 716, "lr": 0.001} +{"train_loss": 0.026994286105036736, "train_loss_bc": 0.026994286105036736, "train_loss_llm": 0.0, "grad_norm": 0.09668058156967163, "global_step": 25780, "epoch": 716, "lr": 0.001} +{"train_loss": 0.028208233416080475, "train_loss_bc": 0.028208233416080475, "train_loss_llm": 0.0, "grad_norm": 0.03913034871220589, "global_step": 25781, "epoch": 716, "lr": 0.001} +{"train_loss": 0.030877765268087387, "train_loss_bc": 0.030877765268087387, "train_loss_llm": 0.0, "grad_norm": 0.0485382154583931, "global_step": 25782, "epoch": 716, "lr": 0.001} +{"train_loss": 0.03334777429699898, "train_loss_bc": 0.03334777429699898, "train_loss_llm": 0.0, "grad_norm": 0.07851708680391312, "global_step": 25783, "epoch": 716, "lr": 0.001} +{"train_loss": 0.024811865761876106, "train_loss_bc": 0.02479664608836174, "train_loss_llm": 0.0015219151973724365, "grad_norm": 0.09787091612815857, "global_step": 25784, "epoch": 716, "lr": 0.001} +{"train_loss": 0.02907934971153736, "train_loss_bc": 0.02907934971153736, "train_loss_llm": 0.0, "grad_norm": 0.03148588910698891, "global_step": 25785, "epoch": 716, "lr": 0.001} +{"train_loss": 0.03275078162550926, "train_loss_bc": 0.03275078162550926, "train_loss_llm": 0.0, "grad_norm": 0.048467252403497696, "global_step": 25786, "epoch": 716, "lr": 0.001} +{"train_loss": 0.02869061380624771, "train_loss_bc": 0.02869061380624771, "train_loss_llm": 0.0, "grad_norm": 0.06343448162078857, "global_step": 25787, "epoch": 716, "lr": 0.001} +{"train_loss": 0.02673722431063652, "train_loss_bc": 0.02673722431063652, "train_loss_llm": 0.0, "grad_norm": 0.08480513840913773, "global_step": 25788, "epoch": 716, "lr": 0.001} +{"train_loss": 0.026043258607387543, "train_loss_bc": 0.02603522315621376, "train_loss_llm": 0.0008035004138946533, "grad_norm": 0.03245564550161362, "global_step": 25789, "epoch": 716, "lr": 0.001} +{"train_loss": 0.030711177736520767, "train_loss_bc": 0.03069460578262806, "train_loss_llm": 0.0016572773456573486, "grad_norm": 0.051606435328722, "global_step": 25790, "epoch": 716, "lr": 0.001} +{"train_loss": 0.037165626883506775, "train_loss_bc": 0.037165626883506775, "train_loss_llm": 0.0, "grad_norm": 0.056667253375053406, "global_step": 25791, "epoch": 716, "lr": 0.001} +{"train_loss": 0.03452570363879204, "train_loss_bc": 0.03452570363879204, "train_loss_llm": 0.0, "grad_norm": 0.07414846122264862, "global_step": 25792, "epoch": 716, "lr": 0.001} +{"train_loss": 0.037492021918296814, "train_loss_bc": 0.037492021918296814, "train_loss_llm": 0.0, "grad_norm": 0.03622902184724808, "global_step": 25793, "epoch": 716, "lr": 0.001} +{"train_loss": 0.038157783448696136, "train_loss_bc": 0.038157783448696136, "train_loss_llm": 0.0, "grad_norm": 0.07290180027484894, "global_step": 25794, "epoch": 716, "lr": 0.001} +{"train_loss": 0.02818201668560505, "train_loss_bc": 0.028152380138635635, "train_loss_llm": 0.0029636919498443604, "grad_norm": 0.08669481426477432, "global_step": 25795, "epoch": 716, "lr": 0.001} +{"train_loss": 0.02972337231040001, "train_loss_bc": 0.02972337231040001, "train_loss_llm": 0.0, "grad_norm": 0.10559514164924622, "global_step": 25796, "epoch": 716, "lr": 0.001} +{"train_loss": 0.03222270682454109, "train_loss_bc": 0.032211918383836746, "train_loss_llm": 0.0010790228843688965, "grad_norm": 0.0398075096309185, "global_step": 25797, "epoch": 716, "lr": 0.001} +{"train_loss": 0.030295444652438164, "train_loss_bc": 0.030292753130197525, "train_loss_llm": 0.00026923418045043945, "grad_norm": 0.05954025685787201, "global_step": 25798, "epoch": 716, "lr": 0.001} +{"train_loss": 0.03236079961061478, "train_loss_bc": 0.03236079961061478, "train_loss_llm": 0.0, "grad_norm": 0.09025610983371735, "global_step": 25799, "epoch": 716, "lr": 0.001} +{"train_loss": 0.034013886004686356, "train_loss_bc": 0.034013886004686356, "train_loss_llm": 0.0, "grad_norm": 0.10750001668930054, "global_step": 25800, "epoch": 716, "lr": 0.001} +{"train_loss": 0.040173761546611786, "train_loss_bc": 0.040173761546611786, "train_loss_llm": 0.0, "grad_norm": 0.04182140529155731, "global_step": 25801, "epoch": 716, "lr": 0.001} +{"train_loss": 0.03931437432765961, "train_loss_bc": 0.03931437432765961, "train_loss_llm": 0.0, "grad_norm": 0.06917892396450043, "global_step": 25802, "epoch": 716, "lr": 0.001} +{"train_loss": 0.03221709653735161, "train_loss_bc": 0.03221709653735161, "train_loss_llm": 0.0, "grad_norm": 0.08716262876987457, "global_step": 25803, "epoch": 716, "lr": 0.001} +{"train_loss": 0.036134470254182816, "train_loss_bc": 0.03613370284438133, "train_loss_llm": 7.659196853637695e-05, "grad_norm": 0.116205133497715, "global_step": 25804, "epoch": 716, "lr": 0.001} +{"train_loss": 0.030994053930044174, "train_loss_bc": 0.030994053930044174, "train_loss_llm": 0.0, "grad_norm": 0.03578678146004677, "global_step": 25805, "epoch": 716, "lr": 0.001} +{"train_loss": 0.026520105078816414, "train_loss_bc": 0.026520105078816414, "train_loss_llm": 0.0, "grad_norm": 0.059591103345155716, "global_step": 25806, "epoch": 716, "lr": 0.001} +{"train_loss": 0.02825125679373741, "train_loss_bc": 0.02825125679373741, "train_loss_llm": 0.0, "grad_norm": 0.09365370124578476, "global_step": 25807, "epoch": 716, "lr": 0.001} +{"train_loss": 0.029870856553316116, "train_loss_bc": 0.029870856553316116, "train_loss_llm": 0.0, "grad_norm": 0.12116920948028564, "global_step": 25808, "epoch": 716, "lr": 0.001} +{"train_loss": 0.03188496083021164, "train_loss_bc": 0.03188496083021164, "train_loss_llm": 0.0, "grad_norm": 0.0383107028901577, "global_step": 25809, "epoch": 716, "lr": 0.001} +{"train_loss": 0.02658894658088684, "train_loss_bc": 0.02658894658088684, "train_loss_llm": 0.0, "grad_norm": 0.057539474219083786, "global_step": 25810, "epoch": 716, "lr": 0.001} +{"train_loss": 0.03131549645008312, "train_loss_bc": 0.029501158744096756, "train_loss_llm": 0.0, "grad_norm": 0.0700085461139679, "global_step": 25811, "epoch": 716, "lr": 0.001, "val_loss": 0.09066610038280487} +{"train_loss": 0.029771197587251663, "train_loss_bc": 0.029771197587251663, "train_loss_llm": 0.0, "grad_norm": 0.08738288283348083, "global_step": 25812, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03117421269416809, "train_loss_bc": 0.031172459945082664, "train_loss_llm": 0.00017529726028442383, "grad_norm": 0.03626929968595505, "global_step": 25813, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03084925189614296, "train_loss_bc": 0.03084925189614296, "train_loss_llm": 0.0, "grad_norm": 0.06469645351171494, "global_step": 25814, "epoch": 717, "lr": 0.001} +{"train_loss": 0.028806185349822044, "train_loss_bc": 0.028806185349822044, "train_loss_llm": 0.0, "grad_norm": 0.08625538647174835, "global_step": 25815, "epoch": 717, "lr": 0.001} +{"train_loss": 0.031795695424079895, "train_loss_bc": 0.031795695424079895, "train_loss_llm": 0.0, "grad_norm": 0.10368110984563828, "global_step": 25816, "epoch": 717, "lr": 0.001} +{"train_loss": 0.033249616622924805, "train_loss_bc": 0.033249616622924805, "train_loss_llm": 0.0, "grad_norm": 0.031919214874506, "global_step": 25817, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03451753407716751, "train_loss_bc": 0.03448346629738808, "train_loss_llm": 0.0034066736698150635, "grad_norm": 0.06029292196035385, "global_step": 25818, "epoch": 717, "lr": 0.001} +{"train_loss": 0.02617533877491951, "train_loss_bc": 0.02617533877491951, "train_loss_llm": 0.0, "grad_norm": 0.07091259211301804, "global_step": 25819, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03638843074440956, "train_loss_bc": 0.03638843074440956, "train_loss_llm": 0.0, "grad_norm": 0.09963927417993546, "global_step": 25820, "epoch": 717, "lr": 0.001} +{"train_loss": 0.031445492058992386, "train_loss_bc": 0.031445492058992386, "train_loss_llm": 0.0, "grad_norm": 0.03462729975581169, "global_step": 25821, "epoch": 717, "lr": 0.001} +{"train_loss": 0.0338691882789135, "train_loss_bc": 0.0338691882789135, "train_loss_llm": 0.0, "grad_norm": 0.05771549046039581, "global_step": 25822, "epoch": 717, "lr": 0.001} +{"train_loss": 0.027617668733000755, "train_loss_bc": 0.027617668733000755, "train_loss_llm": 0.0, "grad_norm": 0.07586433738470078, "global_step": 25823, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03773162141442299, "train_loss_bc": 0.03772404417395592, "train_loss_llm": 0.0007575452327728271, "grad_norm": 0.1027836948633194, "global_step": 25824, "epoch": 717, "lr": 0.001} +{"train_loss": 0.02993864379823208, "train_loss_bc": 0.02993864379823208, "train_loss_llm": 0.0, "grad_norm": 0.0347704142332077, "global_step": 25825, "epoch": 717, "lr": 0.001} +{"train_loss": 0.027943449094891548, "train_loss_bc": 0.027943449094891548, "train_loss_llm": 0.0, "grad_norm": 0.05615193769335747, "global_step": 25826, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03708512708544731, "train_loss_bc": 0.037063729017972946, "train_loss_llm": 0.002139955759048462, "grad_norm": 0.08040830492973328, "global_step": 25827, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03220824897289276, "train_loss_bc": 0.03220824897289276, "train_loss_llm": 0.0, "grad_norm": 0.09836256504058838, "global_step": 25828, "epoch": 717, "lr": 0.001} +{"train_loss": 0.031579095870256424, "train_loss_bc": 0.031579095870256424, "train_loss_llm": 0.0, "grad_norm": 0.040757983922958374, "global_step": 25829, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03207220137119293, "train_loss_bc": 0.03207220137119293, "train_loss_llm": 0.0, "grad_norm": 0.06025142967700958, "global_step": 25830, "epoch": 717, "lr": 0.001} +{"train_loss": 0.034843530505895615, "train_loss_bc": 0.034843530505895615, "train_loss_llm": 0.0, "grad_norm": 0.09099157899618149, "global_step": 25831, "epoch": 717, "lr": 0.001} +{"train_loss": 0.024963201954960823, "train_loss_bc": 0.024963201954960823, "train_loss_llm": 0.0, "grad_norm": 0.09949545562267303, "global_step": 25832, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03647879138588905, "train_loss_bc": 0.03647879138588905, "train_loss_llm": 0.0, "grad_norm": 0.03712589293718338, "global_step": 25833, "epoch": 717, "lr": 0.001} +{"train_loss": 0.0338263064622879, "train_loss_bc": 0.033821672201156616, "train_loss_llm": 0.0004635453224182129, "grad_norm": 0.05506075918674469, "global_step": 25834, "epoch": 717, "lr": 0.001} +{"train_loss": 0.029816502705216408, "train_loss_bc": 0.029816502705216408, "train_loss_llm": 0.0, "grad_norm": 0.06498399376869202, "global_step": 25835, "epoch": 717, "lr": 0.001} +{"train_loss": 0.027325965464115143, "train_loss_bc": 0.027325965464115143, "train_loss_llm": 0.0, "grad_norm": 0.08037487417459488, "global_step": 25836, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03327813372015953, "train_loss_bc": 0.03327813372015953, "train_loss_llm": 0.0, "grad_norm": 0.043770529329776764, "global_step": 25837, "epoch": 717, "lr": 0.001} +{"train_loss": 0.033887237310409546, "train_loss_bc": 0.033887237310409546, "train_loss_llm": 0.0, "grad_norm": 0.06345584988594055, "global_step": 25838, "epoch": 717, "lr": 0.001} +{"train_loss": 0.021063316613435745, "train_loss_bc": 0.021055186167359352, "train_loss_llm": 0.0008130669593811035, "grad_norm": 0.09452569484710693, "global_step": 25839, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03617452457547188, "train_loss_bc": 0.03617452457547188, "train_loss_llm": 0.0, "grad_norm": 0.12716074287891388, "global_step": 25840, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03392650559544563, "train_loss_bc": 0.03392650559544563, "train_loss_llm": 0.0, "grad_norm": 0.03699425235390663, "global_step": 25841, "epoch": 717, "lr": 0.001} +{"train_loss": 0.02702668122947216, "train_loss_bc": 0.02700095623731613, "train_loss_llm": 0.0025724470615386963, "grad_norm": 0.056299470365047455, "global_step": 25842, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03500041365623474, "train_loss_bc": 0.03500041365623474, "train_loss_llm": 0.0, "grad_norm": 0.08229227364063263, "global_step": 25843, "epoch": 717, "lr": 0.001} +{"train_loss": 0.025723574683070183, "train_loss_bc": 0.025723574683070183, "train_loss_llm": 0.0, "grad_norm": 0.10223332792520523, "global_step": 25844, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03041691705584526, "train_loss_bc": 0.030394354835152626, "train_loss_llm": 0.002256244421005249, "grad_norm": 0.032968923449516296, "global_step": 25845, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03361957520246506, "train_loss_bc": 0.0336097851395607, "train_loss_llm": 0.0009788870811462402, "grad_norm": 0.05137175694108009, "global_step": 25846, "epoch": 717, "lr": 0.001} +{"train_loss": 0.03137333308243089, "train_loss_bc": 0.027850612998008728, "train_loss_llm": 0.0, "grad_norm": 0.0716622844338417, "global_step": 25847, "epoch": 717, "lr": 0.001, "val_loss": 0.09058763831853867} +{"train_loss": 0.03424496948719025, "train_loss_bc": 0.03424496948719025, "train_loss_llm": 0.0, "grad_norm": 0.09752029180526733, "global_step": 25848, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03268377482891083, "train_loss_bc": 0.03268377482891083, "train_loss_llm": 0.0, "grad_norm": 0.02893219329416752, "global_step": 25849, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03177451714873314, "train_loss_bc": 0.03177451714873314, "train_loss_llm": 0.0, "grad_norm": 0.05222456157207489, "global_step": 25850, "epoch": 718, "lr": 0.001} +{"train_loss": 0.030721215531229973, "train_loss_bc": 0.030721215531229973, "train_loss_llm": 0.0, "grad_norm": 0.05908215045928955, "global_step": 25851, "epoch": 718, "lr": 0.001} +{"train_loss": 0.029875950887799263, "train_loss_bc": 0.029875950887799263, "train_loss_llm": 0.0, "grad_norm": 0.07781131565570831, "global_step": 25852, "epoch": 718, "lr": 0.001} +{"train_loss": 0.02361108362674713, "train_loss_bc": 0.02360406517982483, "train_loss_llm": 0.0007017552852630615, "grad_norm": 0.024853110313415527, "global_step": 25853, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03329842910170555, "train_loss_bc": 0.03329842910170555, "train_loss_llm": 0.0, "grad_norm": 0.04220148175954819, "global_step": 25854, "epoch": 718, "lr": 0.001} +{"train_loss": 0.02843465469777584, "train_loss_bc": 0.02843465469777584, "train_loss_llm": 0.0, "grad_norm": 0.05852461978793144, "global_step": 25855, "epoch": 718, "lr": 0.001} +{"train_loss": 0.025333397090435028, "train_loss_bc": 0.025333397090435028, "train_loss_llm": 0.0, "grad_norm": 0.07508039474487305, "global_step": 25856, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03393133357167244, "train_loss_bc": 0.03391797095537186, "train_loss_llm": 0.0013360977172851562, "grad_norm": 0.03809530660510063, "global_step": 25857, "epoch": 718, "lr": 0.001} +{"train_loss": 0.02966071479022503, "train_loss_bc": 0.029654785990715027, "train_loss_llm": 0.0005928575992584229, "grad_norm": 0.06425340473651886, "global_step": 25858, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03163435682654381, "train_loss_bc": 0.03163435682654381, "train_loss_llm": 0.0, "grad_norm": 0.08433876931667328, "global_step": 25859, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03199905529618263, "train_loss_bc": 0.03195963427424431, "train_loss_llm": 0.0039421916007995605, "grad_norm": 0.0969281941652298, "global_step": 25860, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03513233736157417, "train_loss_bc": 0.03513233736157417, "train_loss_llm": 0.0, "grad_norm": 0.032484084367752075, "global_step": 25861, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03263629227876663, "train_loss_bc": 0.03263629227876663, "train_loss_llm": 0.0, "grad_norm": 0.06478539854288101, "global_step": 25862, "epoch": 718, "lr": 0.001} +{"train_loss": 0.02876347117125988, "train_loss_bc": 0.02876347117125988, "train_loss_llm": 0.0, "grad_norm": 0.08385280519723892, "global_step": 25863, "epoch": 718, "lr": 0.001} +{"train_loss": 0.02906726486980915, "train_loss_bc": 0.02906726486980915, "train_loss_llm": 0.0, "grad_norm": 0.10155162215232849, "global_step": 25864, "epoch": 718, "lr": 0.001} +{"train_loss": 0.029029633849859238, "train_loss_bc": 0.029029633849859238, "train_loss_llm": 0.0, "grad_norm": 0.03619125112891197, "global_step": 25865, "epoch": 718, "lr": 0.001} +{"train_loss": 0.029937420040369034, "train_loss_bc": 0.029937420040369034, "train_loss_llm": 0.0, "grad_norm": 0.06088106706738472, "global_step": 25866, "epoch": 718, "lr": 0.001} +{"train_loss": 0.023810304701328278, "train_loss_bc": 0.023810304701328278, "train_loss_llm": 0.0, "grad_norm": 0.08398605138063431, "global_step": 25867, "epoch": 718, "lr": 0.001} +{"train_loss": 0.028730913996696472, "train_loss_bc": 0.028730913996696472, "train_loss_llm": 0.0, "grad_norm": 0.09918073564767838, "global_step": 25868, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03101276606321335, "train_loss_bc": 0.03101276606321335, "train_loss_llm": 0.0, "grad_norm": 0.03639742732048035, "global_step": 25869, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03755437210202217, "train_loss_bc": 0.03755300119519234, "train_loss_llm": 0.00013720989227294922, "grad_norm": 0.057165924459695816, "global_step": 25870, "epoch": 718, "lr": 0.001} +{"train_loss": 0.033440444618463516, "train_loss_bc": 0.033440444618463516, "train_loss_llm": 0.0, "grad_norm": 0.07722901552915573, "global_step": 25871, "epoch": 718, "lr": 0.001} +{"train_loss": 0.025413161143660545, "train_loss_bc": 0.025413161143660545, "train_loss_llm": 0.0, "grad_norm": 0.0964614748954773, "global_step": 25872, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03057238459587097, "train_loss_bc": 0.0305705014616251, "train_loss_llm": 0.00018838047981262207, "grad_norm": 0.03287775442004204, "global_step": 25873, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03152346983551979, "train_loss_bc": 0.03151325508952141, "train_loss_llm": 0.0010212957859039307, "grad_norm": 0.046563662588596344, "global_step": 25874, "epoch": 718, "lr": 0.001} +{"train_loss": 0.030338941141963005, "train_loss_bc": 0.030338941141963005, "train_loss_llm": 0.0, "grad_norm": 0.06581837683916092, "global_step": 25875, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03364038094878197, "train_loss_bc": 0.03364007547497749, "train_loss_llm": 3.072619438171387e-05, "grad_norm": 0.08387201279401779, "global_step": 25876, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03770683333277702, "train_loss_bc": 0.03770683333277702, "train_loss_llm": 0.0, "grad_norm": 0.0446741059422493, "global_step": 25877, "epoch": 718, "lr": 0.001} +{"train_loss": 0.0320073664188385, "train_loss_bc": 0.0320073664188385, "train_loss_llm": 0.0, "grad_norm": 0.06411013752222061, "global_step": 25878, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03189846873283386, "train_loss_bc": 0.0318906307220459, "train_loss_llm": 0.0007836818695068359, "grad_norm": 0.074675552546978, "global_step": 25879, "epoch": 718, "lr": 0.001} +{"train_loss": 0.030491136014461517, "train_loss_bc": 0.030491136014461517, "train_loss_llm": 0.0, "grad_norm": 0.09035282582044601, "global_step": 25880, "epoch": 718, "lr": 0.001} +{"train_loss": 0.0404234416782856, "train_loss_bc": 0.0404234416782856, "train_loss_llm": 0.0, "grad_norm": 0.04684212803840637, "global_step": 25881, "epoch": 718, "lr": 0.001} +{"train_loss": 0.03397952765226364, "train_loss_bc": 0.03397952765226364, "train_loss_llm": 0.0, "grad_norm": 0.06402692943811417, "global_step": 25882, "epoch": 718, "lr": 0.001} +{"train_loss": 0.031164857196725078, "train_loss_bc": 0.02762107364833355, "train_loss_llm": 0.0, "grad_norm": 0.08653492480516434, "global_step": 25883, "epoch": 718, "lr": 0.001, "val_loss": 0.09192252159118652} +{"train_loss": 0.033012792468070984, "train_loss_bc": 0.033012792468070984, "train_loss_llm": 0.0, "grad_norm": 0.10456489771604538, "global_step": 25884, "epoch": 719, "lr": 0.001} +{"train_loss": 0.030668560415506363, "train_loss_bc": 0.030665015801787376, "train_loss_llm": 0.00035452842712402344, "grad_norm": 0.04344598948955536, "global_step": 25885, "epoch": 719, "lr": 0.001} +{"train_loss": 0.029149390757083893, "train_loss_bc": 0.029149390757083893, "train_loss_llm": 0.0, "grad_norm": 0.06478986144065857, "global_step": 25886, "epoch": 719, "lr": 0.001} +{"train_loss": 0.028566893190145493, "train_loss_bc": 0.02855224907398224, "train_loss_llm": 0.0014643669128417969, "grad_norm": 0.08140610158443451, "global_step": 25887, "epoch": 719, "lr": 0.001} +{"train_loss": 0.026101214811205864, "train_loss_bc": 0.026101214811205864, "train_loss_llm": 0.0, "grad_norm": 0.10081633180379868, "global_step": 25888, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03930984064936638, "train_loss_bc": 0.03930984064936638, "train_loss_llm": 0.0, "grad_norm": 0.04290064051747322, "global_step": 25889, "epoch": 719, "lr": 0.001} +{"train_loss": 0.021953746676445007, "train_loss_bc": 0.021953746676445007, "train_loss_llm": 0.0, "grad_norm": 0.056099291890859604, "global_step": 25890, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03470982238650322, "train_loss_bc": 0.03470982238650322, "train_loss_llm": 0.0, "grad_norm": 0.06850052624940872, "global_step": 25891, "epoch": 719, "lr": 0.001} +{"train_loss": 0.02605774998664856, "train_loss_bc": 0.02605774998664856, "train_loss_llm": 0.0, "grad_norm": 0.08807002753019333, "global_step": 25892, "epoch": 719, "lr": 0.001} +{"train_loss": 0.031601183116436005, "train_loss_bc": 0.031601183116436005, "train_loss_llm": 0.0, "grad_norm": 0.03105432167649269, "global_step": 25893, "epoch": 719, "lr": 0.001} +{"train_loss": 0.033148229122161865, "train_loss_bc": 0.033148229122161865, "train_loss_llm": 0.0, "grad_norm": 0.048415809869766235, "global_step": 25894, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03512436896562576, "train_loss_bc": 0.0351235568523407, "train_loss_llm": 8.13901424407959e-05, "grad_norm": 0.06414708495140076, "global_step": 25895, "epoch": 719, "lr": 0.001} +{"train_loss": 0.034196171909570694, "train_loss_bc": 0.034196171909570694, "train_loss_llm": 0.0, "grad_norm": 0.08376975357532501, "global_step": 25896, "epoch": 719, "lr": 0.001} +{"train_loss": 0.033430315554142, "train_loss_bc": 0.033415429294109344, "train_loss_llm": 0.0014885962009429932, "grad_norm": 0.03188963234424591, "global_step": 25897, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03464869037270546, "train_loss_bc": 0.03464869037270546, "train_loss_llm": 0.0, "grad_norm": 0.04865264892578125, "global_step": 25898, "epoch": 719, "lr": 0.001} +{"train_loss": 0.031109729781746864, "train_loss_bc": 0.031104667112231255, "train_loss_llm": 0.0005063116550445557, "grad_norm": 0.06188289448618889, "global_step": 25899, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03275889903306961, "train_loss_bc": 0.03275889903306961, "train_loss_llm": 0.0, "grad_norm": 0.07628859579563141, "global_step": 25900, "epoch": 719, "lr": 0.001} +{"train_loss": 0.031057491898536682, "train_loss_bc": 0.031057491898536682, "train_loss_llm": 0.0, "grad_norm": 0.03173626586794853, "global_step": 25901, "epoch": 719, "lr": 0.001} +{"train_loss": 0.028254268690943718, "train_loss_bc": 0.02825249917805195, "train_loss_llm": 0.00017702579498291016, "grad_norm": 0.055614374577999115, "global_step": 25902, "epoch": 719, "lr": 0.001} +{"train_loss": 0.02535686455667019, "train_loss_bc": 0.025354037061333656, "train_loss_llm": 0.0002827942371368408, "grad_norm": 0.07045487314462662, "global_step": 25903, "epoch": 719, "lr": 0.001} +{"train_loss": 0.027266718447208405, "train_loss_bc": 0.027266718447208405, "train_loss_llm": 0.0, "grad_norm": 0.08373238146305084, "global_step": 25904, "epoch": 719, "lr": 0.001} +{"train_loss": 0.029438409954309464, "train_loss_bc": 0.029438409954309464, "train_loss_llm": 0.0, "grad_norm": 0.026775667443871498, "global_step": 25905, "epoch": 719, "lr": 0.001} +{"train_loss": 0.031136278063058853, "train_loss_bc": 0.031131980940699577, "train_loss_llm": 0.0004297196865081787, "grad_norm": 0.05123473331332207, "global_step": 25906, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03323781117796898, "train_loss_bc": 0.03323781117796898, "train_loss_llm": 0.0, "grad_norm": 0.07940825819969177, "global_step": 25907, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03150408715009689, "train_loss_bc": 0.03149691969156265, "train_loss_llm": 0.0007165670394897461, "grad_norm": 0.11099987477064133, "global_step": 25908, "epoch": 719, "lr": 0.001} +{"train_loss": 0.030646605417132378, "train_loss_bc": 0.030646605417132378, "train_loss_llm": 0.0, "grad_norm": 0.02844422124326229, "global_step": 25909, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03667769208550453, "train_loss_bc": 0.0366683192551136, "train_loss_llm": 0.0009372830390930176, "grad_norm": 0.05975028872489929, "global_step": 25910, "epoch": 719, "lr": 0.001} +{"train_loss": 0.033467087894678116, "train_loss_bc": 0.033431731164455414, "train_loss_llm": 0.0035355091094970703, "grad_norm": 0.08451449126005173, "global_step": 25911, "epoch": 719, "lr": 0.001} +{"train_loss": 0.027280574664473534, "train_loss_bc": 0.0272680576890707, "train_loss_llm": 0.0012517273426055908, "grad_norm": 0.10748188942670822, "global_step": 25912, "epoch": 719, "lr": 0.001} +{"train_loss": 0.030446378514170647, "train_loss_bc": 0.030446378514170647, "train_loss_llm": 0.0, "grad_norm": 0.0316363088786602, "global_step": 25913, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03668823465704918, "train_loss_bc": 0.03668823465704918, "train_loss_llm": 0.0, "grad_norm": 0.05167276784777641, "global_step": 25914, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03023671545088291, "train_loss_bc": 0.030230775475502014, "train_loss_llm": 0.0005940794944763184, "grad_norm": 0.06022046506404877, "global_step": 25915, "epoch": 719, "lr": 0.001} +{"train_loss": 0.028856122866272926, "train_loss_bc": 0.02884858287870884, "train_loss_llm": 0.0007539689540863037, "grad_norm": 0.06949039548635483, "global_step": 25916, "epoch": 719, "lr": 0.001} +{"train_loss": 0.034685492515563965, "train_loss_bc": 0.034685492515563965, "train_loss_llm": 0.0, "grad_norm": 0.03649648278951645, "global_step": 25917, "epoch": 719, "lr": 0.001} +{"train_loss": 0.03156851977109909, "train_loss_bc": 0.03156851977109909, "train_loss_llm": 0.0, "grad_norm": 0.05647001042962074, "global_step": 25918, "epoch": 719, "lr": 0.001} +{"train_loss": 0.031104086587826412, "train_loss_bc": 0.026392275467514992, "train_loss_llm": 0.0001887977123260498, "grad_norm": 0.061982959508895874, "global_step": 25919, "epoch": 719, "lr": 0.001, "val_loss": 0.091349296271801} +{"train_loss": 0.03394710272550583, "train_loss_bc": 0.03394710272550583, "train_loss_llm": 0.0, "grad_norm": 0.07732238620519638, "global_step": 25920, "epoch": 720, "lr": 0.001} +{"train_loss": 0.0322074368596077, "train_loss_bc": 0.0322074368596077, "train_loss_llm": 0.0, "grad_norm": 0.03770845755934715, "global_step": 25921, "epoch": 720, "lr": 0.001} +{"train_loss": 0.034400902688503265, "train_loss_bc": 0.034400902688503265, "train_loss_llm": 0.0, "grad_norm": 0.0604378767311573, "global_step": 25922, "epoch": 720, "lr": 0.001} +{"train_loss": 0.02971077337861061, "train_loss_bc": 0.02970794588327408, "train_loss_llm": 0.0002828240394592285, "grad_norm": 0.0862591341137886, "global_step": 25923, "epoch": 720, "lr": 0.001} +{"train_loss": 0.02795158140361309, "train_loss_bc": 0.02794906683266163, "train_loss_llm": 0.00025144219398498535, "grad_norm": 0.10315897315740585, "global_step": 25924, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03472847491502762, "train_loss_bc": 0.03472847491502762, "train_loss_llm": 0.0, "grad_norm": 0.04184655100107193, "global_step": 25925, "epoch": 720, "lr": 0.001} +{"train_loss": 0.029286950826644897, "train_loss_bc": 0.0292868260294199, "train_loss_llm": 1.2487173080444336e-05, "grad_norm": 0.055414650589227676, "global_step": 25926, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03209027275443077, "train_loss_bc": 0.03207971528172493, "train_loss_llm": 0.0010557174682617188, "grad_norm": 0.06378670036792755, "global_step": 25927, "epoch": 720, "lr": 0.001} +{"train_loss": 0.029835626482963562, "train_loss_bc": 0.029835626482963562, "train_loss_llm": 0.0, "grad_norm": 0.0814780667424202, "global_step": 25928, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03167928010225296, "train_loss_bc": 0.03167928010225296, "train_loss_llm": 0.0, "grad_norm": 0.03714349493384361, "global_step": 25929, "epoch": 720, "lr": 0.001} +{"train_loss": 0.033085860311985016, "train_loss_bc": 0.033085860311985016, "train_loss_llm": 0.0, "grad_norm": 0.05889709293842316, "global_step": 25930, "epoch": 720, "lr": 0.001} +{"train_loss": 0.030639557167887688, "train_loss_bc": 0.03062635101377964, "train_loss_llm": 0.001320570707321167, "grad_norm": 0.08068584650754929, "global_step": 25931, "epoch": 720, "lr": 0.001} +{"train_loss": 0.041680414229631424, "train_loss_bc": 0.041680414229631424, "train_loss_llm": 0.0, "grad_norm": 0.1058024913072586, "global_step": 25932, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03160467371344566, "train_loss_bc": 0.03160467371344566, "train_loss_llm": 0.0, "grad_norm": 0.027868248522281647, "global_step": 25933, "epoch": 720, "lr": 0.001} +{"train_loss": 0.030078083276748657, "train_loss_bc": 0.030078083276748657, "train_loss_llm": 0.0, "grad_norm": 0.04293476417660713, "global_step": 25934, "epoch": 720, "lr": 0.001} +{"train_loss": 0.027588047087192535, "train_loss_bc": 0.027588047087192535, "train_loss_llm": 0.0, "grad_norm": 0.052524734288454056, "global_step": 25935, "epoch": 720, "lr": 0.001} +{"train_loss": 0.027326218783855438, "train_loss_bc": 0.027320029214024544, "train_loss_llm": 0.0006189942359924316, "grad_norm": 0.06681410223245621, "global_step": 25936, "epoch": 720, "lr": 0.001} +{"train_loss": 0.021552035585045815, "train_loss_bc": 0.021552035585045815, "train_loss_llm": 0.0, "grad_norm": 0.023712987080216408, "global_step": 25937, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03130513057112694, "train_loss_bc": 0.03130513057112694, "train_loss_llm": 0.0, "grad_norm": 0.04632938653230667, "global_step": 25938, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03094281628727913, "train_loss_bc": 0.03094281628727913, "train_loss_llm": 0.0, "grad_norm": 0.05860898643732071, "global_step": 25939, "epoch": 720, "lr": 0.001} +{"train_loss": 0.02633105032145977, "train_loss_bc": 0.02633105032145977, "train_loss_llm": 0.0, "grad_norm": 0.07429532706737518, "global_step": 25940, "epoch": 720, "lr": 0.001} +{"train_loss": 0.032138753682374954, "train_loss_bc": 0.032138753682374954, "train_loss_llm": 0.0, "grad_norm": 0.031988125294446945, "global_step": 25941, "epoch": 720, "lr": 0.001} +{"train_loss": 0.026128234341740608, "train_loss_bc": 0.02612782083451748, "train_loss_llm": 4.139542579650879e-05, "grad_norm": 0.04900328069925308, "global_step": 25942, "epoch": 720, "lr": 0.001} +{"train_loss": 0.0235611442476511, "train_loss_bc": 0.0235611442476511, "train_loss_llm": 0.0, "grad_norm": 0.07416039705276489, "global_step": 25943, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03543970361351967, "train_loss_bc": 0.03543970361351967, "train_loss_llm": 0.0, "grad_norm": 0.08774451166391373, "global_step": 25944, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03422674164175987, "train_loss_bc": 0.03421479091048241, "train_loss_llm": 0.001194983720779419, "grad_norm": 0.03963163122534752, "global_step": 25945, "epoch": 720, "lr": 0.001} +{"train_loss": 0.028089964762330055, "train_loss_bc": 0.028089964762330055, "train_loss_llm": 0.0, "grad_norm": 0.06564770638942719, "global_step": 25946, "epoch": 720, "lr": 0.001} +{"train_loss": 0.030902767553925514, "train_loss_bc": 0.030902767553925514, "train_loss_llm": 0.0, "grad_norm": 0.08936882764101028, "global_step": 25947, "epoch": 720, "lr": 0.001} +{"train_loss": 0.034563906490802765, "train_loss_bc": 0.034560058265924454, "train_loss_llm": 0.0003847777843475342, "grad_norm": 0.11139196902513504, "global_step": 25948, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03495415672659874, "train_loss_bc": 0.03495415672659874, "train_loss_llm": 0.0, "grad_norm": 0.036469172686338425, "global_step": 25949, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03336811065673828, "train_loss_bc": 0.03336139768362045, "train_loss_llm": 0.00067138671875, "grad_norm": 0.0631883516907692, "global_step": 25950, "epoch": 720, "lr": 0.001} +{"train_loss": 0.031081680208444595, "train_loss_bc": 0.031081680208444595, "train_loss_llm": 0.0, "grad_norm": 0.08299602568149567, "global_step": 25951, "epoch": 720, "lr": 0.001} +{"train_loss": 0.03401722013950348, "train_loss_bc": 0.03401722013950348, "train_loss_llm": 0.0, "grad_norm": 0.10922703146934509, "global_step": 25952, "epoch": 720, "lr": 0.001} +{"train_loss": 0.035033803433179855, "train_loss_bc": 0.035033803433179855, "train_loss_llm": 0.0, "grad_norm": 0.038910869508981705, "global_step": 25953, "epoch": 720, "lr": 0.001} +{"train_loss": 0.028432175517082214, "train_loss_bc": 0.028432175517082214, "train_loss_llm": 0.0, "grad_norm": 0.04961599409580231, "global_step": 25954, "epoch": 720, "lr": 0.001} +{"train_loss": 0.031230647427340347, "train_loss_bc": 0.03439265489578247, "train_loss_llm": 0.0, "grad_norm": 0.06389342993497849, "global_step": 25955, "epoch": 720, "lr": 0.001, "train/cumulative_reward": 70.64524130893155, "train/mean_score": 15.81854661623828, "train/success_rate": 0.3, "test/cumulative_reward": 35.73942399878453, "test/mean_score": 7.208331049401055, "test/success_rate": 0.18, "val_loss": 0.09146172553300858, "train_action_mse_error": 0.03390716016292572} +{"train_loss": 0.02818009816110134, "train_loss_bc": 0.02818009816110134, "train_loss_llm": 0.0, "grad_norm": 0.08259813487529755, "global_step": 25956, "epoch": 721, "lr": 0.001} +{"train_loss": 0.028227785602211952, "train_loss_bc": 0.028226856142282486, "train_loss_llm": 9.298324584960938e-05, "grad_norm": 0.027721812948584557, "global_step": 25957, "epoch": 721, "lr": 0.001} +{"train_loss": 0.02445758320391178, "train_loss_bc": 0.02445758320391178, "train_loss_llm": 0.0, "grad_norm": 0.047794319689273834, "global_step": 25958, "epoch": 721, "lr": 0.001} +{"train_loss": 0.028565527871251106, "train_loss_bc": 0.028565527871251106, "train_loss_llm": 0.0, "grad_norm": 0.05999518930912018, "global_step": 25959, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03263683617115021, "train_loss_bc": 0.03263683617115021, "train_loss_llm": 0.0, "grad_norm": 0.07730626314878464, "global_step": 25960, "epoch": 721, "lr": 0.001} +{"train_loss": 0.037125032395124435, "train_loss_bc": 0.037125032395124435, "train_loss_llm": 0.0, "grad_norm": 0.03532424196600914, "global_step": 25961, "epoch": 721, "lr": 0.001} +{"train_loss": 0.02996409498155117, "train_loss_bc": 0.02996409498155117, "train_loss_llm": 0.0, "grad_norm": 0.07998790591955185, "global_step": 25962, "epoch": 721, "lr": 0.001} +{"train_loss": 0.02754244953393936, "train_loss_bc": 0.02754244953393936, "train_loss_llm": 0.0, "grad_norm": 0.1144665777683258, "global_step": 25963, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03446507453918457, "train_loss_bc": 0.03446507453918457, "train_loss_llm": 0.0, "grad_norm": 0.13973747193813324, "global_step": 25964, "epoch": 721, "lr": 0.001} +{"train_loss": 0.025659877806901932, "train_loss_bc": 0.025654245167970657, "train_loss_llm": 0.0005632340908050537, "grad_norm": 0.03599059581756592, "global_step": 25965, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03184078261256218, "train_loss_bc": 0.03184078261256218, "train_loss_llm": 0.0, "grad_norm": 0.059604380279779434, "global_step": 25966, "epoch": 721, "lr": 0.001} +{"train_loss": 0.037059977650642395, "train_loss_bc": 0.037059977650642395, "train_loss_llm": 0.0, "grad_norm": 0.09258037060499191, "global_step": 25967, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03275250270962715, "train_loss_bc": 0.03275250270962715, "train_loss_llm": 0.0, "grad_norm": 0.12872326374053955, "global_step": 25968, "epoch": 721, "lr": 0.001} +{"train_loss": 0.034053049981594086, "train_loss_bc": 0.034053049981594086, "train_loss_llm": 0.0, "grad_norm": 0.03587063401937485, "global_step": 25969, "epoch": 721, "lr": 0.001} +{"train_loss": 0.032209813594818115, "train_loss_bc": 0.032209813594818115, "train_loss_llm": 0.0, "grad_norm": 0.06501348316669464, "global_step": 25970, "epoch": 721, "lr": 0.001} +{"train_loss": 0.028623491525650024, "train_loss_bc": 0.028622370213270187, "train_loss_llm": 0.00011211633682250977, "grad_norm": 0.08747650682926178, "global_step": 25971, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03454745188355446, "train_loss_bc": 0.03453335911035538, "train_loss_llm": 0.0014093220233917236, "grad_norm": 0.10331536829471588, "global_step": 25972, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03338482603430748, "train_loss_bc": 0.03338482603430748, "train_loss_llm": 0.0, "grad_norm": 0.030324790626764297, "global_step": 25973, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03483930602669716, "train_loss_bc": 0.034817617386579514, "train_loss_llm": 0.002168893814086914, "grad_norm": 0.04046071693301201, "global_step": 25974, "epoch": 721, "lr": 0.001} +{"train_loss": 0.02598397620022297, "train_loss_bc": 0.02594846300780773, "train_loss_llm": 0.003551274538040161, "grad_norm": 0.05477512627840042, "global_step": 25975, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03392593562602997, "train_loss_bc": 0.033903393894433975, "train_loss_llm": 0.002254277467727661, "grad_norm": 0.0766448974609375, "global_step": 25976, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03683558106422424, "train_loss_bc": 0.03683558106422424, "train_loss_llm": 0.0, "grad_norm": 0.04981590062379837, "global_step": 25977, "epoch": 721, "lr": 0.001} +{"train_loss": 0.02792542614042759, "train_loss_bc": 0.02792542614042759, "train_loss_llm": 0.0, "grad_norm": 0.06832771748304367, "global_step": 25978, "epoch": 721, "lr": 0.001} +{"train_loss": 0.02979763224720955, "train_loss_bc": 0.029792843386530876, "train_loss_llm": 0.00047895312309265137, "grad_norm": 0.09690418839454651, "global_step": 25979, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03153596818447113, "train_loss_bc": 0.03153596818447113, "train_loss_llm": 0.0, "grad_norm": 0.12306573987007141, "global_step": 25980, "epoch": 721, "lr": 0.001} +{"train_loss": 0.0363650843501091, "train_loss_bc": 0.0363650843501091, "train_loss_llm": 0.0, "grad_norm": 0.04359032213687897, "global_step": 25981, "epoch": 721, "lr": 0.001} +{"train_loss": 0.037912413477897644, "train_loss_bc": 0.037912413477897644, "train_loss_llm": 0.0, "grad_norm": 0.07820137590169907, "global_step": 25982, "epoch": 721, "lr": 0.001} +{"train_loss": 0.02965608425438404, "train_loss_bc": 0.02965160645544529, "train_loss_llm": 0.00044777989387512207, "grad_norm": 0.11042439192533493, "global_step": 25983, "epoch": 721, "lr": 0.001} +{"train_loss": 0.02536953054368496, "train_loss_bc": 0.02536953054368496, "train_loss_llm": 0.0, "grad_norm": 0.1403055042028427, "global_step": 25984, "epoch": 721, "lr": 0.001} +{"train_loss": 0.040021732449531555, "train_loss_bc": 0.040021732449531555, "train_loss_llm": 0.0, "grad_norm": 0.05965489149093628, "global_step": 25985, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03553764894604683, "train_loss_bc": 0.03553764894604683, "train_loss_llm": 0.0, "grad_norm": 0.09549028426408768, "global_step": 25986, "epoch": 721, "lr": 0.001} +{"train_loss": 0.0294971726834774, "train_loss_bc": 0.0294971726834774, "train_loss_llm": 0.0, "grad_norm": 0.11830998212099075, "global_step": 25987, "epoch": 721, "lr": 0.001} +{"train_loss": 0.02838628925383091, "train_loss_bc": 0.02838628925383091, "train_loss_llm": 0.0, "grad_norm": 0.13133074343204498, "global_step": 25988, "epoch": 721, "lr": 0.001} +{"train_loss": 0.0361282154917717, "train_loss_bc": 0.0361282154917717, "train_loss_llm": 0.0, "grad_norm": 0.039282314479351044, "global_step": 25989, "epoch": 721, "lr": 0.001} +{"train_loss": 0.0308538768440485, "train_loss_bc": 0.0308538768440485, "train_loss_llm": 0.0, "grad_norm": 0.06461977958679199, "global_step": 25990, "epoch": 721, "lr": 0.001} +{"train_loss": 0.03167350543662906, "train_loss_bc": 0.02836550399661064, "train_loss_llm": 0.0012562274932861328, "grad_norm": 0.0867338553071022, "global_step": 25991, "epoch": 721, "lr": 0.001, "val_loss": 0.088996022939682} +{"train_loss": 0.03155416250228882, "train_loss_bc": 0.03155241906642914, "train_loss_llm": 0.0001742243766784668, "grad_norm": 0.11613886803388596, "global_step": 25992, "epoch": 722, "lr": 0.001} +{"train_loss": 0.028052374720573425, "train_loss_bc": 0.028052374720573425, "train_loss_llm": 0.0, "grad_norm": 0.03504214808344841, "global_step": 25993, "epoch": 722, "lr": 0.001} +{"train_loss": 0.027485663071274757, "train_loss_bc": 0.027482762932777405, "train_loss_llm": 0.0002900660037994385, "grad_norm": 0.04003685712814331, "global_step": 25994, "epoch": 722, "lr": 0.001} +{"train_loss": 0.032704804092645645, "train_loss_bc": 0.032704804092645645, "train_loss_llm": 0.0, "grad_norm": 0.051478900015354156, "global_step": 25995, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03345251455903053, "train_loss_bc": 0.03342966362833977, "train_loss_llm": 0.0022851526737213135, "grad_norm": 0.05922364071011543, "global_step": 25996, "epoch": 722, "lr": 0.001} +{"train_loss": 0.027885351330041885, "train_loss_bc": 0.027885351330041885, "train_loss_llm": 0.0, "grad_norm": 0.03220110014081001, "global_step": 25997, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03014824353158474, "train_loss_bc": 0.03014824353158474, "train_loss_llm": 0.0, "grad_norm": 0.05617799609899521, "global_step": 25998, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03276454284787178, "train_loss_bc": 0.03276454284787178, "train_loss_llm": 0.0, "grad_norm": 0.085111603140831, "global_step": 25999, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03221042454242706, "train_loss_bc": 0.032198064029216766, "train_loss_llm": 0.0012360811233520508, "grad_norm": 0.09956823289394379, "global_step": 26000, "epoch": 722, "lr": 0.001} +{"train_loss": 0.029978187754750252, "train_loss_bc": 0.029978187754750252, "train_loss_llm": 0.0, "grad_norm": 0.04118824750185013, "global_step": 26001, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03233553469181061, "train_loss_bc": 0.03233553469181061, "train_loss_llm": 0.0, "grad_norm": 0.06213577091693878, "global_step": 26002, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03562881425023079, "train_loss_bc": 0.03562881425023079, "train_loss_llm": 0.0, "grad_norm": 0.0934983417391777, "global_step": 26003, "epoch": 722, "lr": 0.001} +{"train_loss": 0.033745087683200836, "train_loss_bc": 0.03373188525438309, "train_loss_llm": 0.0013203322887420654, "grad_norm": 0.12726807594299316, "global_step": 26004, "epoch": 722, "lr": 0.001} +{"train_loss": 0.026659151539206505, "train_loss_bc": 0.026659151539206505, "train_loss_llm": 0.0, "grad_norm": 0.04132772609591484, "global_step": 26005, "epoch": 722, "lr": 0.001} +{"train_loss": 0.02818235754966736, "train_loss_bc": 0.028178835287690163, "train_loss_llm": 0.0003522932529449463, "grad_norm": 0.06840091198682785, "global_step": 26006, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03236041218042374, "train_loss_bc": 0.03236041218042374, "train_loss_llm": 0.0, "grad_norm": 0.09426512569189072, "global_step": 26007, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03646604344248772, "train_loss_bc": 0.03646604344248772, "train_loss_llm": 0.0, "grad_norm": 0.12560856342315674, "global_step": 26008, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03492414951324463, "train_loss_bc": 0.03492414951324463, "train_loss_llm": 0.0, "grad_norm": 0.03603288158774376, "global_step": 26009, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03130045533180237, "train_loss_bc": 0.03128882497549057, "train_loss_llm": 0.0011630654335021973, "grad_norm": 0.05817260965704918, "global_step": 26010, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03168987110257149, "train_loss_bc": 0.03168987110257149, "train_loss_llm": 0.0, "grad_norm": 0.08618953078985214, "global_step": 26011, "epoch": 722, "lr": 0.001} +{"train_loss": 0.028964228928089142, "train_loss_bc": 0.028951315209269524, "train_loss_llm": 0.0012913644313812256, "grad_norm": 0.10510082542896271, "global_step": 26012, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03398707136511803, "train_loss_bc": 0.03398707136511803, "train_loss_llm": 0.0, "grad_norm": 0.03754079341888428, "global_step": 26013, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03224816545844078, "train_loss_bc": 0.03224816545844078, "train_loss_llm": 0.0, "grad_norm": 0.06815364211797714, "global_step": 26014, "epoch": 722, "lr": 0.001} +{"train_loss": 0.027928687632083893, "train_loss_bc": 0.027928687632083893, "train_loss_llm": 0.0, "grad_norm": 0.08699639141559601, "global_step": 26015, "epoch": 722, "lr": 0.001} +{"train_loss": 0.036077938973903656, "train_loss_bc": 0.036071471869945526, "train_loss_llm": 0.0006468296051025391, "grad_norm": 0.11059743165969849, "global_step": 26016, "epoch": 722, "lr": 0.001} +{"train_loss": 0.029260212555527687, "train_loss_bc": 0.029260212555527687, "train_loss_llm": 0.0, "grad_norm": 0.028380535542964935, "global_step": 26017, "epoch": 722, "lr": 0.001} +{"train_loss": 0.02823794074356556, "train_loss_bc": 0.028237076476216316, "train_loss_llm": 8.651614189147949e-05, "grad_norm": 0.05434392765164375, "global_step": 26018, "epoch": 722, "lr": 0.001} +{"train_loss": 0.033770833164453506, "train_loss_bc": 0.033761799335479736, "train_loss_llm": 0.0009034276008605957, "grad_norm": 0.07066171616315842, "global_step": 26019, "epoch": 722, "lr": 0.001} +{"train_loss": 0.023606782779097557, "train_loss_bc": 0.023606782779097557, "train_loss_llm": 0.0, "grad_norm": 0.08376051485538483, "global_step": 26020, "epoch": 722, "lr": 0.001} +{"train_loss": 0.02653723955154419, "train_loss_bc": 0.026536259800195694, "train_loss_llm": 9.79304313659668e-05, "grad_norm": 0.04109399393200874, "global_step": 26021, "epoch": 722, "lr": 0.001} +{"train_loss": 0.030738530680537224, "train_loss_bc": 0.030738530680537224, "train_loss_llm": 0.0, "grad_norm": 0.06436113268136978, "global_step": 26022, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03736596554517746, "train_loss_bc": 0.037357836961746216, "train_loss_llm": 0.0008126795291900635, "grad_norm": 0.09471515566110611, "global_step": 26023, "epoch": 722, "lr": 0.001} +{"train_loss": 0.0365896001458168, "train_loss_bc": 0.0365896001458168, "train_loss_llm": 0.0, "grad_norm": 0.12330269068479538, "global_step": 26024, "epoch": 722, "lr": 0.001} +{"train_loss": 0.031928591430187225, "train_loss_bc": 0.031928591430187225, "train_loss_llm": 0.0, "grad_norm": 0.044474903494119644, "global_step": 26025, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03728146851062775, "train_loss_bc": 0.037273697555065155, "train_loss_llm": 0.0007771551609039307, "grad_norm": 0.08944766968488693, "global_step": 26026, "epoch": 722, "lr": 0.001} +{"train_loss": 0.03153219022270706, "train_loss_bc": 0.031104501336812973, "train_loss_llm": 0.0002943277359008789, "grad_norm": 0.13893257081508636, "global_step": 26027, "epoch": 722, "lr": 0.001, "val_loss": 0.09006290137767792} +{"train_loss": 0.029676983132958412, "train_loss_bc": 0.029676983132958412, "train_loss_llm": 0.0, "grad_norm": 0.152326300740242, "global_step": 26028, "epoch": 723, "lr": 0.001} +{"train_loss": 0.02842462994158268, "train_loss_bc": 0.02842462994158268, "train_loss_llm": 0.0, "grad_norm": 0.03717983141541481, "global_step": 26029, "epoch": 723, "lr": 0.001} +{"train_loss": 0.031938012689352036, "train_loss_bc": 0.031938012689352036, "train_loss_llm": 0.0, "grad_norm": 0.06247551739215851, "global_step": 26030, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03326570615172386, "train_loss_bc": 0.033261023461818695, "train_loss_llm": 0.00046816468238830566, "grad_norm": 0.09346122294664383, "global_step": 26031, "epoch": 723, "lr": 0.001} +{"train_loss": 0.030955394729971886, "train_loss_bc": 0.03094950132071972, "train_loss_llm": 0.0005893111228942871, "grad_norm": 0.11615365743637085, "global_step": 26032, "epoch": 723, "lr": 0.001} +{"train_loss": 0.0314839743077755, "train_loss_bc": 0.0314839743077755, "train_loss_llm": 0.0, "grad_norm": 0.036538757383823395, "global_step": 26033, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03290499374270439, "train_loss_bc": 0.03290499374270439, "train_loss_llm": 0.0, "grad_norm": 0.05214247107505798, "global_step": 26034, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03362452983856201, "train_loss_bc": 0.03362452983856201, "train_loss_llm": 0.0, "grad_norm": 0.07201994210481644, "global_step": 26035, "epoch": 723, "lr": 0.001} +{"train_loss": 0.031074954196810722, "train_loss_bc": 0.031074954196810722, "train_loss_llm": 0.0, "grad_norm": 0.09346376359462738, "global_step": 26036, "epoch": 723, "lr": 0.001} +{"train_loss": 0.033823300153017044, "train_loss_bc": 0.033823300153017044, "train_loss_llm": 0.0, "grad_norm": 0.03877618908882141, "global_step": 26037, "epoch": 723, "lr": 0.001} +{"train_loss": 0.033811621367931366, "train_loss_bc": 0.033811621367931366, "train_loss_llm": 0.0, "grad_norm": 0.06524768471717834, "global_step": 26038, "epoch": 723, "lr": 0.001} +{"train_loss": 0.029907308518886566, "train_loss_bc": 0.029907308518886566, "train_loss_llm": 0.0, "grad_norm": 0.09329821169376373, "global_step": 26039, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03409428894519806, "train_loss_bc": 0.03409428894519806, "train_loss_llm": 0.0, "grad_norm": 0.11344143748283386, "global_step": 26040, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03282106667757034, "train_loss_bc": 0.03282106667757034, "train_loss_llm": 0.0, "grad_norm": 0.03510832041501999, "global_step": 26041, "epoch": 723, "lr": 0.001} +{"train_loss": 0.034454427659511566, "train_loss_bc": 0.034454427659511566, "train_loss_llm": 0.0, "grad_norm": 0.062408242374658585, "global_step": 26042, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03042188100516796, "train_loss_bc": 0.03042188100516796, "train_loss_llm": 0.0, "grad_norm": 0.0909322202205658, "global_step": 26043, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03514206409454346, "train_loss_bc": 0.03514206409454346, "train_loss_llm": 0.0, "grad_norm": 0.10235140472650528, "global_step": 26044, "epoch": 723, "lr": 0.001} +{"train_loss": 0.030066734179854393, "train_loss_bc": 0.030066734179854393, "train_loss_llm": 0.0, "grad_norm": 0.05318117141723633, "global_step": 26045, "epoch": 723, "lr": 0.001} +{"train_loss": 0.025005504488945007, "train_loss_bc": 0.025005504488945007, "train_loss_llm": 0.0, "grad_norm": 0.07916150987148285, "global_step": 26046, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03188891336321831, "train_loss_bc": 0.03188891336321831, "train_loss_llm": 0.0, "grad_norm": 0.11458507180213928, "global_step": 26047, "epoch": 723, "lr": 0.001} +{"train_loss": 0.02925785258412361, "train_loss_bc": 0.029256191104650497, "train_loss_llm": 0.00016620755195617676, "grad_norm": 0.14312832057476044, "global_step": 26048, "epoch": 723, "lr": 0.001} +{"train_loss": 0.034195780754089355, "train_loss_bc": 0.03418930619955063, "train_loss_llm": 0.0006476044654846191, "grad_norm": 0.041674837470054626, "global_step": 26049, "epoch": 723, "lr": 0.001} +{"train_loss": 0.031304217875003815, "train_loss_bc": 0.031304217875003815, "train_loss_llm": 0.0, "grad_norm": 0.051489777863025665, "global_step": 26050, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03456990420818329, "train_loss_bc": 0.034553490579128265, "train_loss_llm": 0.0016414821147918701, "grad_norm": 0.0726323053240776, "global_step": 26051, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03309081494808197, "train_loss_bc": 0.03308015689253807, "train_loss_llm": 0.0010659396648406982, "grad_norm": 0.10152511298656464, "global_step": 26052, "epoch": 723, "lr": 0.001} +{"train_loss": 0.02944614179432392, "train_loss_bc": 0.02944614179432392, "train_loss_llm": 0.0, "grad_norm": 0.02908390201628208, "global_step": 26053, "epoch": 723, "lr": 0.001} +{"train_loss": 0.02848493866622448, "train_loss_bc": 0.02848493866622448, "train_loss_llm": 0.0, "grad_norm": 0.04669405519962311, "global_step": 26054, "epoch": 723, "lr": 0.001} +{"train_loss": 0.030416375026106834, "train_loss_bc": 0.03041067346930504, "train_loss_llm": 0.0005702078342437744, "grad_norm": 0.06779739260673523, "global_step": 26055, "epoch": 723, "lr": 0.001} +{"train_loss": 0.028168676421046257, "train_loss_bc": 0.028168676421046257, "train_loss_llm": 0.0, "grad_norm": 0.08392792195081711, "global_step": 26056, "epoch": 723, "lr": 0.001} +{"train_loss": 0.03493674844503403, "train_loss_bc": 0.034918587654829025, "train_loss_llm": 0.0018161535263061523, "grad_norm": 0.04699724540114403, "global_step": 26057, "epoch": 723, "lr": 0.001} +{"train_loss": 0.02665783278644085, "train_loss_bc": 0.02665783278644085, "train_loss_llm": 0.0, "grad_norm": 0.05673344060778618, "global_step": 26058, "epoch": 723, "lr": 0.001} +{"train_loss": 0.031169533729553223, "train_loss_bc": 0.031169533729553223, "train_loss_llm": 0.0, "grad_norm": 0.07948029786348343, "global_step": 26059, "epoch": 723, "lr": 0.001} +{"train_loss": 0.02934923954308033, "train_loss_bc": 0.029346423223614693, "train_loss_llm": 0.0002816617488861084, "grad_norm": 0.10712942481040955, "global_step": 26060, "epoch": 723, "lr": 0.001} +{"train_loss": 0.038783613592386246, "train_loss_bc": 0.038783613592386246, "train_loss_llm": 0.0, "grad_norm": 0.03706645965576172, "global_step": 26061, "epoch": 723, "lr": 0.001} +{"train_loss": 0.029085546731948853, "train_loss_bc": 0.029085546731948853, "train_loss_llm": 0.0, "grad_norm": 0.05829690024256706, "global_step": 26062, "epoch": 723, "lr": 0.001} +{"train_loss": 0.031552542828851275, "train_loss_bc": 0.03218803554773331, "train_loss_llm": 0.0, "grad_norm": 0.08634328842163086, "global_step": 26063, "epoch": 723, "lr": 0.001, "val_loss": 0.09024426341056824} +{"train_loss": 0.028951086103916168, "train_loss_bc": 0.028945256024599075, "train_loss_llm": 0.0005830824375152588, "grad_norm": 0.10353459417819977, "global_step": 26064, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03390178084373474, "train_loss_bc": 0.03390178084373474, "train_loss_llm": 0.0, "grad_norm": 0.047490086406469345, "global_step": 26065, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03226632997393608, "train_loss_bc": 0.03226632997393608, "train_loss_llm": 0.0, "grad_norm": 0.07703503221273422, "global_step": 26066, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03258984163403511, "train_loss_bc": 0.03258984163403511, "train_loss_llm": 0.0, "grad_norm": 0.10438954085111618, "global_step": 26067, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03169294074177742, "train_loss_bc": 0.03169294074177742, "train_loss_llm": 0.0, "grad_norm": 0.11258631199598312, "global_step": 26068, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03274714946746826, "train_loss_bc": 0.03274714946746826, "train_loss_llm": 0.0, "grad_norm": 0.03716515004634857, "global_step": 26069, "epoch": 724, "lr": 0.001} +{"train_loss": 0.0336892232298851, "train_loss_bc": 0.0336892232298851, "train_loss_llm": 0.0, "grad_norm": 0.04570659622550011, "global_step": 26070, "epoch": 724, "lr": 0.001} +{"train_loss": 0.023469790816307068, "train_loss_bc": 0.023466099053621292, "train_loss_llm": 0.00036919116973876953, "grad_norm": 0.06026393920183182, "global_step": 26071, "epoch": 724, "lr": 0.001} +{"train_loss": 0.02484080009162426, "train_loss_bc": 0.02483745850622654, "train_loss_llm": 0.00033411383628845215, "grad_norm": 0.0667426586151123, "global_step": 26072, "epoch": 724, "lr": 0.001} +{"train_loss": 0.0374821238219738, "train_loss_bc": 0.0374821200966835, "train_loss_llm": 3.2782554626464844e-07, "grad_norm": 0.036900363862514496, "global_step": 26073, "epoch": 724, "lr": 0.001} +{"train_loss": 0.030025547370314598, "train_loss_bc": 0.030025547370314598, "train_loss_llm": 0.0, "grad_norm": 0.05048166960477829, "global_step": 26074, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03279808536171913, "train_loss_bc": 0.032788317650556564, "train_loss_llm": 0.0009769201278686523, "grad_norm": 0.0648803785443306, "global_step": 26075, "epoch": 724, "lr": 0.001} +{"train_loss": 0.031492844223976135, "train_loss_bc": 0.031492844223976135, "train_loss_llm": 0.0, "grad_norm": 0.0787004828453064, "global_step": 26076, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03078494593501091, "train_loss_bc": 0.03078494593501091, "train_loss_llm": 0.0, "grad_norm": 0.05342298746109009, "global_step": 26077, "epoch": 724, "lr": 0.001} +{"train_loss": 0.030377093702554703, "train_loss_bc": 0.030369669198989868, "train_loss_llm": 0.000742495059967041, "grad_norm": 0.08108007907867432, "global_step": 26078, "epoch": 724, "lr": 0.001} +{"train_loss": 0.027926404029130936, "train_loss_bc": 0.02792275883257389, "train_loss_llm": 0.000364452600479126, "grad_norm": 0.09575521945953369, "global_step": 26079, "epoch": 724, "lr": 0.001} +{"train_loss": 0.034719597548246384, "train_loss_bc": 0.034719597548246384, "train_loss_llm": 0.0, "grad_norm": 0.11889036744832993, "global_step": 26080, "epoch": 724, "lr": 0.001} +{"train_loss": 0.0259380042552948, "train_loss_bc": 0.025929855182766914, "train_loss_llm": 0.0008148252964019775, "grad_norm": 0.03069745935499668, "global_step": 26081, "epoch": 724, "lr": 0.001} +{"train_loss": 0.033831752836704254, "train_loss_bc": 0.033824581652879715, "train_loss_llm": 0.0007170438766479492, "grad_norm": 0.06309300661087036, "global_step": 26082, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03100620210170746, "train_loss_bc": 0.03100620210170746, "train_loss_llm": 0.0, "grad_norm": 0.08006865531206131, "global_step": 26083, "epoch": 724, "lr": 0.001} +{"train_loss": 0.02995261549949646, "train_loss_bc": 0.02995261549949646, "train_loss_llm": 0.0, "grad_norm": 0.09800443053245544, "global_step": 26084, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03347564861178398, "train_loss_bc": 0.03347564861178398, "train_loss_llm": 0.0, "grad_norm": 0.03999011591076851, "global_step": 26085, "epoch": 724, "lr": 0.001} +{"train_loss": 0.02931460551917553, "train_loss_bc": 0.029313569888472557, "train_loss_llm": 0.0001036524772644043, "grad_norm": 0.060385480523109436, "global_step": 26086, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03211337327957153, "train_loss_bc": 0.03211337327957153, "train_loss_llm": 0.0, "grad_norm": 0.07361277937889099, "global_step": 26087, "epoch": 724, "lr": 0.001} +{"train_loss": 0.026010004803538322, "train_loss_bc": 0.026010004803538322, "train_loss_llm": 0.0, "grad_norm": 0.09145115315914154, "global_step": 26088, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03262704610824585, "train_loss_bc": 0.03262704610824585, "train_loss_llm": 0.0, "grad_norm": 0.032532528042793274, "global_step": 26089, "epoch": 724, "lr": 0.001} +{"train_loss": 0.02907230332493782, "train_loss_bc": 0.02907230332493782, "train_loss_llm": 0.0, "grad_norm": 0.05837729200720787, "global_step": 26090, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03515899181365967, "train_loss_bc": 0.03515899181365967, "train_loss_llm": 0.0, "grad_norm": 0.0813029482960701, "global_step": 26091, "epoch": 724, "lr": 0.001} +{"train_loss": 0.032240815460681915, "train_loss_bc": 0.032232608646154404, "train_loss_llm": 0.0008206367492675781, "grad_norm": 0.0964595153927803, "global_step": 26092, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03329112008213997, "train_loss_bc": 0.03329112008213997, "train_loss_llm": 0.0, "grad_norm": 0.04022025689482689, "global_step": 26093, "epoch": 724, "lr": 0.001} +{"train_loss": 0.03053385391831398, "train_loss_bc": 0.03053385391831398, "train_loss_llm": 0.0, "grad_norm": 0.062469884753227234, "global_step": 26094, "epoch": 724, "lr": 0.001} +{"train_loss": 0.034131795167922974, "train_loss_bc": 0.03412768617272377, "train_loss_llm": 0.00041097402572631836, "grad_norm": 0.08389777690172195, "global_step": 26095, "epoch": 724, "lr": 0.001} +{"train_loss": 0.030891792848706245, "train_loss_bc": 0.030886901542544365, "train_loss_llm": 0.0004890859127044678, "grad_norm": 0.10858622938394547, "global_step": 26096, "epoch": 724, "lr": 0.001} +{"train_loss": 0.0318622887134552, "train_loss_bc": 0.0318622887134552, "train_loss_llm": 0.0, "grad_norm": 0.03708285838365555, "global_step": 26097, "epoch": 724, "lr": 0.001} +{"train_loss": 0.027128268033266068, "train_loss_bc": 0.027123747393488884, "train_loss_llm": 0.0004520118236541748, "grad_norm": 0.0489746518433094, "global_step": 26098, "epoch": 724, "lr": 0.001} +{"train_loss": 0.031341215750823416, "train_loss_bc": 0.03994769975543022, "train_loss_llm": 0.0, "grad_norm": 0.06853222101926804, "global_step": 26099, "epoch": 724, "lr": 0.001, "val_loss": 0.09189744293689728} +{"train_loss": 0.028611106798052788, "train_loss_bc": 0.028608376160264015, "train_loss_llm": 0.00027313828468322754, "grad_norm": 0.08259677141904831, "global_step": 26100, "epoch": 725, "lr": 0.001} +{"train_loss": 0.025455566123127937, "train_loss_bc": 0.025455566123127937, "train_loss_llm": 0.0, "grad_norm": 0.02890249527990818, "global_step": 26101, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03375007584691048, "train_loss_bc": 0.03375007584691048, "train_loss_llm": 0.0, "grad_norm": 0.051426276564598083, "global_step": 26102, "epoch": 725, "lr": 0.001} +{"train_loss": 0.031172752380371094, "train_loss_bc": 0.031172752380371094, "train_loss_llm": 0.0, "grad_norm": 0.06999565660953522, "global_step": 26103, "epoch": 725, "lr": 0.001} +{"train_loss": 0.028961142525076866, "train_loss_bc": 0.028952162712812424, "train_loss_llm": 0.0008979439735412598, "grad_norm": 0.08545177429914474, "global_step": 26104, "epoch": 725, "lr": 0.001} +{"train_loss": 0.030204832553863525, "train_loss_bc": 0.030197495594620705, "train_loss_llm": 0.0007336437702178955, "grad_norm": 0.02936185523867607, "global_step": 26105, "epoch": 725, "lr": 0.001} +{"train_loss": 0.027816560119390488, "train_loss_bc": 0.027816560119390488, "train_loss_llm": 0.0, "grad_norm": 0.051540058106184006, "global_step": 26106, "epoch": 725, "lr": 0.001} +{"train_loss": 0.027228308841586113, "train_loss_bc": 0.027228308841586113, "train_loss_llm": 0.0, "grad_norm": 0.06783454120159149, "global_step": 26107, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03180769458413124, "train_loss_bc": 0.03180769458413124, "train_loss_llm": 0.0, "grad_norm": 0.08714257925748825, "global_step": 26108, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03154897689819336, "train_loss_bc": 0.03153739869594574, "train_loss_llm": 0.0011576712131500244, "grad_norm": 0.031910717487335205, "global_step": 26109, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03520461916923523, "train_loss_bc": 0.03520461916923523, "train_loss_llm": 0.0, "grad_norm": 0.05566456541419029, "global_step": 26110, "epoch": 725, "lr": 0.001} +{"train_loss": 0.02848178707063198, "train_loss_bc": 0.02848178707063198, "train_loss_llm": 0.0, "grad_norm": 0.08086545765399933, "global_step": 26111, "epoch": 725, "lr": 0.001} +{"train_loss": 0.028529299423098564, "train_loss_bc": 0.028529299423098564, "train_loss_llm": 0.0, "grad_norm": 0.09314000606536865, "global_step": 26112, "epoch": 725, "lr": 0.001} +{"train_loss": 0.031198719516396523, "train_loss_bc": 0.031198719516396523, "train_loss_llm": 0.0, "grad_norm": 0.023953299969434738, "global_step": 26113, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03261379525065422, "train_loss_bc": 0.03261379525065422, "train_loss_llm": 0.0, "grad_norm": 0.04984525963664055, "global_step": 26114, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03342868387699127, "train_loss_bc": 0.03342868387699127, "train_loss_llm": 0.0, "grad_norm": 0.07254184037446976, "global_step": 26115, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03604651987552643, "train_loss_bc": 0.03604651987552643, "train_loss_llm": 0.0, "grad_norm": 0.08960786461830139, "global_step": 26116, "epoch": 725, "lr": 0.001} +{"train_loss": 0.024454424157738686, "train_loss_bc": 0.024454424157738686, "train_loss_llm": 0.0, "grad_norm": 0.029844269156455994, "global_step": 26117, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03490762040019035, "train_loss_bc": 0.03490762040019035, "train_loss_llm": 0.0, "grad_norm": 0.046625103801488876, "global_step": 26118, "epoch": 725, "lr": 0.001} +{"train_loss": 0.030802566558122635, "train_loss_bc": 0.030802566558122635, "train_loss_llm": 0.0, "grad_norm": 0.058209896087646484, "global_step": 26119, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03752497583627701, "train_loss_bc": 0.03752497583627701, "train_loss_llm": 0.0, "grad_norm": 0.08335305005311966, "global_step": 26120, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03125565126538277, "train_loss_bc": 0.03125565126538277, "train_loss_llm": 0.0, "grad_norm": 0.02803410217165947, "global_step": 26121, "epoch": 725, "lr": 0.001} +{"train_loss": 0.02993476204574108, "train_loss_bc": 0.02993476204574108, "train_loss_llm": 0.0, "grad_norm": 0.04724213108420372, "global_step": 26122, "epoch": 725, "lr": 0.001} +{"train_loss": 0.035506438463926315, "train_loss_bc": 0.035506438463926315, "train_loss_llm": 0.0, "grad_norm": 0.06354406476020813, "global_step": 26123, "epoch": 725, "lr": 0.001} +{"train_loss": 0.031745463609695435, "train_loss_bc": 0.031723540276288986, "train_loss_llm": 0.002192378044128418, "grad_norm": 0.07731623202562332, "global_step": 26124, "epoch": 725, "lr": 0.001} +{"train_loss": 0.035778991878032684, "train_loss_bc": 0.035778991878032684, "train_loss_llm": 0.0, "grad_norm": 0.04048752784729004, "global_step": 26125, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03134659677743912, "train_loss_bc": 0.03134659677743912, "train_loss_llm": 0.0, "grad_norm": 0.057615768164396286, "global_step": 26126, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03413405641913414, "train_loss_bc": 0.034123510122299194, "train_loss_llm": 0.0010547935962677002, "grad_norm": 0.07159478217363358, "global_step": 26127, "epoch": 725, "lr": 0.001} +{"train_loss": 0.02904720976948738, "train_loss_bc": 0.029035909101366997, "train_loss_llm": 0.0011300742626190186, "grad_norm": 0.08000074326992035, "global_step": 26128, "epoch": 725, "lr": 0.001} +{"train_loss": 0.029568975791335106, "train_loss_bc": 0.029568975791335106, "train_loss_llm": 0.0, "grad_norm": 0.03502729907631874, "global_step": 26129, "epoch": 725, "lr": 0.001} +{"train_loss": 0.02867124415934086, "train_loss_bc": 0.02867124415934086, "train_loss_llm": 0.0, "grad_norm": 0.0557468943297863, "global_step": 26130, "epoch": 725, "lr": 0.001} +{"train_loss": 0.034694574773311615, "train_loss_bc": 0.034690070897340775, "train_loss_llm": 0.00045040249824523926, "grad_norm": 0.08358561992645264, "global_step": 26131, "epoch": 725, "lr": 0.001} +{"train_loss": 0.030541453510522842, "train_loss_bc": 0.030535373836755753, "train_loss_llm": 0.000607907772064209, "grad_norm": 0.10962029546499252, "global_step": 26132, "epoch": 725, "lr": 0.001} +{"train_loss": 0.025493966415524483, "train_loss_bc": 0.025488603860139847, "train_loss_llm": 0.0005362629890441895, "grad_norm": 0.023097515106201172, "global_step": 26133, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03458578139543533, "train_loss_bc": 0.03458578139543533, "train_loss_llm": 0.0, "grad_norm": 0.041895151138305664, "global_step": 26134, "epoch": 725, "lr": 0.001} +{"train_loss": 0.03115908397982518, "train_loss_bc": 0.02967182919383049, "train_loss_llm": 0.0, "grad_norm": 0.07617518305778503, "global_step": 26135, "epoch": 725, "lr": 0.001, "train/cumulative_reward": 55.3789821225286, "train/mean_score": 11.79058166565503, "train/success_rate": 0.3, "test/cumulative_reward": 35.429141767836484, "test/mean_score": 7.996761102710193, "test/success_rate": 0.2, "val_loss": 0.09051387757062912, "train_action_mse_error": 0.03498665615916252} +{"train_loss": 0.027527857571840286, "train_loss_bc": 0.027527857571840286, "train_loss_llm": 0.0, "grad_norm": 0.09750878810882568, "global_step": 26136, "epoch": 726, "lr": 0.001} +{"train_loss": 0.032006874680519104, "train_loss_bc": 0.032006874680519104, "train_loss_llm": 0.0, "grad_norm": 0.037628840655088425, "global_step": 26137, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03510842099785805, "train_loss_bc": 0.03510842099785805, "train_loss_llm": 0.0, "grad_norm": 0.05738203600049019, "global_step": 26138, "epoch": 726, "lr": 0.001} +{"train_loss": 0.02805081196129322, "train_loss_bc": 0.02804511971771717, "train_loss_llm": 0.0005693137645721436, "grad_norm": 0.07438540458679199, "global_step": 26139, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03311030939221382, "train_loss_bc": 0.03309791907668114, "train_loss_llm": 0.0012389719486236572, "grad_norm": 0.09050141274929047, "global_step": 26140, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03406786173582077, "train_loss_bc": 0.03404686599969864, "train_loss_llm": 0.0020994842052459717, "grad_norm": 0.03622344508767128, "global_step": 26141, "epoch": 726, "lr": 0.001} +{"train_loss": 0.025503188371658325, "train_loss_bc": 0.025503188371658325, "train_loss_llm": 0.0, "grad_norm": 0.048197437077760696, "global_step": 26142, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03286783769726753, "train_loss_bc": 0.03286783769726753, "train_loss_llm": 0.0, "grad_norm": 0.08415808528661728, "global_step": 26143, "epoch": 726, "lr": 0.001} +{"train_loss": 0.032914839684963226, "train_loss_bc": 0.032912902534008026, "train_loss_llm": 0.00019362568855285645, "grad_norm": 0.10123658180236816, "global_step": 26144, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03051201067864895, "train_loss_bc": 0.03051201067864895, "train_loss_llm": 0.0, "grad_norm": 0.036535296589136124, "global_step": 26145, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03119947761297226, "train_loss_bc": 0.031193500384688377, "train_loss_llm": 0.0005978047847747803, "grad_norm": 0.06108398735523224, "global_step": 26146, "epoch": 726, "lr": 0.001} +{"train_loss": 0.027434581890702248, "train_loss_bc": 0.027434581890702248, "train_loss_llm": 0.0, "grad_norm": 0.08364012092351913, "global_step": 26147, "epoch": 726, "lr": 0.001} +{"train_loss": 0.02664838917553425, "train_loss_bc": 0.026644466444849968, "train_loss_llm": 0.000392228364944458, "grad_norm": 0.11193158477544785, "global_step": 26148, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03290455415844917, "train_loss_bc": 0.03290455415844917, "train_loss_llm": 0.0, "grad_norm": 0.039973657578229904, "global_step": 26149, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03757772594690323, "train_loss_bc": 0.03757772594690323, "train_loss_llm": 0.0, "grad_norm": 0.0679549127817154, "global_step": 26150, "epoch": 726, "lr": 0.001} +{"train_loss": 0.02615220658481121, "train_loss_bc": 0.02615220658481121, "train_loss_llm": 0.0, "grad_norm": 0.09206908196210861, "global_step": 26151, "epoch": 726, "lr": 0.001} +{"train_loss": 0.029188772663474083, "train_loss_bc": 0.029188772663474083, "train_loss_llm": 0.0, "grad_norm": 0.10545320063829422, "global_step": 26152, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03357822820544243, "train_loss_bc": 0.03357822820544243, "train_loss_llm": 0.0, "grad_norm": 0.042793337255716324, "global_step": 26153, "epoch": 726, "lr": 0.001} +{"train_loss": 0.028695495799183846, "train_loss_bc": 0.028695495799183846, "train_loss_llm": 0.0, "grad_norm": 0.07204117625951767, "global_step": 26154, "epoch": 726, "lr": 0.001} +{"train_loss": 0.032254595309495926, "train_loss_bc": 0.032254595309495926, "train_loss_llm": 0.0, "grad_norm": 0.09302922338247299, "global_step": 26155, "epoch": 726, "lr": 0.001} +{"train_loss": 0.02778424322605133, "train_loss_bc": 0.027779169380664825, "train_loss_llm": 0.0005074143409729004, "grad_norm": 0.10866992920637131, "global_step": 26156, "epoch": 726, "lr": 0.001} +{"train_loss": 0.0370263047516346, "train_loss_bc": 0.036997754126787186, "train_loss_llm": 0.0028549134731292725, "grad_norm": 0.03888973966240883, "global_step": 26157, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03618168085813522, "train_loss_bc": 0.036170318722724915, "train_loss_llm": 0.0011362135410308838, "grad_norm": 0.06255653500556946, "global_step": 26158, "epoch": 726, "lr": 0.001} +{"train_loss": 0.029235726222395897, "train_loss_bc": 0.029235726222395897, "train_loss_llm": 0.0, "grad_norm": 0.09051129966974258, "global_step": 26159, "epoch": 726, "lr": 0.001} +{"train_loss": 0.037059634923934937, "train_loss_bc": 0.037059634923934937, "train_loss_llm": 0.0, "grad_norm": 0.12543654441833496, "global_step": 26160, "epoch": 726, "lr": 0.001} +{"train_loss": 0.0374099500477314, "train_loss_bc": 0.0374099500477314, "train_loss_llm": 0.0, "grad_norm": 0.037694089114665985, "global_step": 26161, "epoch": 726, "lr": 0.001} +{"train_loss": 0.028866447508335114, "train_loss_bc": 0.028866447508335114, "train_loss_llm": 0.0, "grad_norm": 0.05758967623114586, "global_step": 26162, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03135530278086662, "train_loss_bc": 0.03135530278086662, "train_loss_llm": 0.0, "grad_norm": 0.07522865384817123, "global_step": 26163, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03371588885784149, "train_loss_bc": 0.03371588885784149, "train_loss_llm": 0.0, "grad_norm": 0.10268167406320572, "global_step": 26164, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03010631538927555, "train_loss_bc": 0.03010631538927555, "train_loss_llm": 0.0, "grad_norm": 0.032041385769844055, "global_step": 26165, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03848284110426903, "train_loss_bc": 0.03848284110426903, "train_loss_llm": 0.0, "grad_norm": 0.05505704879760742, "global_step": 26166, "epoch": 726, "lr": 0.001} +{"train_loss": 0.02704784646630287, "train_loss_bc": 0.02704784646630287, "train_loss_llm": 0.0, "grad_norm": 0.08015239983797073, "global_step": 26167, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03440125286579132, "train_loss_bc": 0.03440125286579132, "train_loss_llm": 0.0, "grad_norm": 0.09065653383731842, "global_step": 26168, "epoch": 726, "lr": 0.001} +{"train_loss": 0.026044918224215508, "train_loss_bc": 0.026037197560071945, "train_loss_llm": 0.0007720589637756348, "grad_norm": 0.03685775399208069, "global_step": 26169, "epoch": 726, "lr": 0.001} +{"train_loss": 0.028072848916053772, "train_loss_bc": 0.028072848916053772, "train_loss_llm": 0.0, "grad_norm": 0.058748017996549606, "global_step": 26170, "epoch": 726, "lr": 0.001} +{"train_loss": 0.03132015492560135, "train_loss_bc": 0.027430335059762, "train_loss_llm": 0.0, "grad_norm": 0.07539709657430649, "global_step": 26171, "epoch": 726, "lr": 0.001, "val_loss": 0.09088286012411118} +{"train_loss": 0.037454672157764435, "train_loss_bc": 0.037454672157764435, "train_loss_llm": 0.0, "grad_norm": 0.1012127548456192, "global_step": 26172, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03242020681500435, "train_loss_bc": 0.03242020681500435, "train_loss_llm": 0.0, "grad_norm": 0.034387342631816864, "global_step": 26173, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03591713681817055, "train_loss_bc": 0.03591713681817055, "train_loss_llm": 0.0, "grad_norm": 0.06572716683149338, "global_step": 26174, "epoch": 727, "lr": 0.001} +{"train_loss": 0.030299901962280273, "train_loss_bc": 0.030299901962280273, "train_loss_llm": 0.0, "grad_norm": 0.10397989302873611, "global_step": 26175, "epoch": 727, "lr": 0.001} +{"train_loss": 0.026586180552840233, "train_loss_bc": 0.026586180552840233, "train_loss_llm": 0.0, "grad_norm": 0.11677782237529755, "global_step": 26176, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03403586894273758, "train_loss_bc": 0.03403586894273758, "train_loss_llm": 0.0, "grad_norm": 0.040474534034729004, "global_step": 26177, "epoch": 727, "lr": 0.001} +{"train_loss": 0.027850661426782608, "train_loss_bc": 0.02781742438673973, "train_loss_llm": 0.003323674201965332, "grad_norm": 0.059136763215065, "global_step": 26178, "epoch": 727, "lr": 0.001} +{"train_loss": 0.0334140881896019, "train_loss_bc": 0.0334140881896019, "train_loss_llm": 0.0, "grad_norm": 0.08079155534505844, "global_step": 26179, "epoch": 727, "lr": 0.001} +{"train_loss": 0.032595355063676834, "train_loss_bc": 0.032590653747320175, "train_loss_llm": 0.00047025084495544434, "grad_norm": 0.10125627368688583, "global_step": 26180, "epoch": 727, "lr": 0.001} +{"train_loss": 0.030591046437621117, "train_loss_bc": 0.030583973973989487, "train_loss_llm": 0.0007072091102600098, "grad_norm": 0.04057775065302849, "global_step": 26181, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03139640390872955, "train_loss_bc": 0.03139640390872955, "train_loss_llm": 0.0, "grad_norm": 0.06118541955947876, "global_step": 26182, "epoch": 727, "lr": 0.001} +{"train_loss": 0.030400337651371956, "train_loss_bc": 0.030400337651371956, "train_loss_llm": 0.0, "grad_norm": 0.08929034322500229, "global_step": 26183, "epoch": 727, "lr": 0.001} +{"train_loss": 0.029046587646007538, "train_loss_bc": 0.029024960473179817, "train_loss_llm": 0.002162635326385498, "grad_norm": 0.1115584671497345, "global_step": 26184, "epoch": 727, "lr": 0.001} +{"train_loss": 0.029991652816534042, "train_loss_bc": 0.029991652816534042, "train_loss_llm": 0.0, "grad_norm": 0.026017623022198677, "global_step": 26185, "epoch": 727, "lr": 0.001} +{"train_loss": 0.028645390644669533, "train_loss_bc": 0.028645390644669533, "train_loss_llm": 0.0, "grad_norm": 0.040209054946899414, "global_step": 26186, "epoch": 727, "lr": 0.001} +{"train_loss": 0.02936480939388275, "train_loss_bc": 0.02936480939388275, "train_loss_llm": 0.0, "grad_norm": 0.05689910426735878, "global_step": 26187, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03328700736165047, "train_loss_bc": 0.03326792269945145, "train_loss_llm": 0.0019083917140960693, "grad_norm": 0.07760054618120193, "global_step": 26188, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03197905793786049, "train_loss_bc": 0.03197905793786049, "train_loss_llm": 0.0, "grad_norm": 0.034044407308101654, "global_step": 26189, "epoch": 727, "lr": 0.001} +{"train_loss": 0.04082290083169937, "train_loss_bc": 0.0407988615334034, "train_loss_llm": 0.002403944730758667, "grad_norm": 0.055384717881679535, "global_step": 26190, "epoch": 727, "lr": 0.001} +{"train_loss": 0.030242199078202248, "train_loss_bc": 0.030242199078202248, "train_loss_llm": 0.0, "grad_norm": 0.07044059783220291, "global_step": 26191, "epoch": 727, "lr": 0.001} +{"train_loss": 0.028145259246230125, "train_loss_bc": 0.028138624504208565, "train_loss_llm": 0.0006634891033172607, "grad_norm": 0.08985168486833572, "global_step": 26192, "epoch": 727, "lr": 0.001} +{"train_loss": 0.030727392062544823, "train_loss_bc": 0.030727392062544823, "train_loss_llm": 0.0, "grad_norm": 0.037215106189250946, "global_step": 26193, "epoch": 727, "lr": 0.001} +{"train_loss": 0.025078868493437767, "train_loss_bc": 0.025078868493437767, "train_loss_llm": 0.0, "grad_norm": 0.05422096699476242, "global_step": 26194, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03603747859597206, "train_loss_bc": 0.03603747859597206, "train_loss_llm": 0.0, "grad_norm": 0.09322218596935272, "global_step": 26195, "epoch": 727, "lr": 0.001} +{"train_loss": 0.02974999137222767, "train_loss_bc": 0.029743855819106102, "train_loss_llm": 0.0006136000156402588, "grad_norm": 0.12390967458486557, "global_step": 26196, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03612476959824562, "train_loss_bc": 0.03612476959824562, "train_loss_llm": 0.0, "grad_norm": 0.038816120475530624, "global_step": 26197, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03005995601415634, "train_loss_bc": 0.03005995601415634, "train_loss_llm": 0.0, "grad_norm": 0.06196393072605133, "global_step": 26198, "epoch": 727, "lr": 0.001} +{"train_loss": 0.0314984992146492, "train_loss_bc": 0.0314984992146492, "train_loss_llm": 0.0, "grad_norm": 0.08539609611034393, "global_step": 26199, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03586530312895775, "train_loss_bc": 0.03586402162909508, "train_loss_llm": 0.00012823939323425293, "grad_norm": 0.11205099523067474, "global_step": 26200, "epoch": 727, "lr": 0.001} +{"train_loss": 0.033368710428476334, "train_loss_bc": 0.03336765617132187, "train_loss_llm": 0.00010541081428527832, "grad_norm": 0.03394021838903427, "global_step": 26201, "epoch": 727, "lr": 0.001} +{"train_loss": 0.024195970967411995, "train_loss_bc": 0.024195970967411995, "train_loss_llm": 0.0, "grad_norm": 0.060608524829149246, "global_step": 26202, "epoch": 727, "lr": 0.001} +{"train_loss": 0.029226910322904587, "train_loss_bc": 0.029226910322904587, "train_loss_llm": 0.0, "grad_norm": 0.07746535539627075, "global_step": 26203, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03466169163584709, "train_loss_bc": 0.03466169163584709, "train_loss_llm": 0.0, "grad_norm": 0.11461623013019562, "global_step": 26204, "epoch": 727, "lr": 0.001} +{"train_loss": 0.027384402230381966, "train_loss_bc": 0.027384402230381966, "train_loss_llm": 0.0, "grad_norm": 0.030757838860154152, "global_step": 26205, "epoch": 727, "lr": 0.001} +{"train_loss": 0.034565653651952744, "train_loss_bc": 0.034565653651952744, "train_loss_llm": 0.0, "grad_norm": 0.04770871624350548, "global_step": 26206, "epoch": 727, "lr": 0.001} +{"train_loss": 0.03127756921781434, "train_loss_bc": 0.022951705381274223, "train_loss_llm": 0.0008462965488433838, "grad_norm": 0.07331110537052155, "global_step": 26207, "epoch": 727, "lr": 0.001, "val_loss": 0.09315335750579834} +{"train_loss": 0.030697446316480637, "train_loss_bc": 0.030697446316480637, "train_loss_llm": 0.0, "grad_norm": 0.10165246576070786, "global_step": 26208, "epoch": 728, "lr": 0.001} +{"train_loss": 0.034801874309778214, "train_loss_bc": 0.034801874309778214, "train_loss_llm": 0.0, "grad_norm": 0.04551621526479721, "global_step": 26209, "epoch": 728, "lr": 0.001} +{"train_loss": 0.030224908143281937, "train_loss_bc": 0.030224908143281937, "train_loss_llm": 0.0, "grad_norm": 0.06794838607311249, "global_step": 26210, "epoch": 728, "lr": 0.001} +{"train_loss": 0.030693987384438515, "train_loss_bc": 0.030693987384438515, "train_loss_llm": 0.0, "grad_norm": 0.09470783919095993, "global_step": 26211, "epoch": 728, "lr": 0.001} +{"train_loss": 0.024516884237527847, "train_loss_bc": 0.024516884237527847, "train_loss_llm": 0.0, "grad_norm": 0.11864890158176422, "global_step": 26212, "epoch": 728, "lr": 0.001} +{"train_loss": 0.026386121287941933, "train_loss_bc": 0.026386121287941933, "train_loss_llm": 0.0, "grad_norm": 0.02787845954298973, "global_step": 26213, "epoch": 728, "lr": 0.001} +{"train_loss": 0.02627604454755783, "train_loss_bc": 0.02627604454755783, "train_loss_llm": 0.0, "grad_norm": 0.05363813415169716, "global_step": 26214, "epoch": 728, "lr": 0.001} +{"train_loss": 0.02977599948644638, "train_loss_bc": 0.02977599948644638, "train_loss_llm": 0.0, "grad_norm": 0.07881622016429901, "global_step": 26215, "epoch": 728, "lr": 0.001} +{"train_loss": 0.030183939263224602, "train_loss_bc": 0.030183939263224602, "train_loss_llm": 0.0, "grad_norm": 0.10062280297279358, "global_step": 26216, "epoch": 728, "lr": 0.001} +{"train_loss": 0.03002138063311577, "train_loss_bc": 0.030014541000127792, "train_loss_llm": 0.0006839334964752197, "grad_norm": 0.028791578486561775, "global_step": 26217, "epoch": 728, "lr": 0.001} +{"train_loss": 0.030272461473941803, "train_loss_bc": 0.030272461473941803, "train_loss_llm": 0.0, "grad_norm": 0.05649907886981964, "global_step": 26218, "epoch": 728, "lr": 0.001} +{"train_loss": 0.032545991241931915, "train_loss_bc": 0.032545991241931915, "train_loss_llm": 0.0, "grad_norm": 0.09301085770130157, "global_step": 26219, "epoch": 728, "lr": 0.001} +{"train_loss": 0.028726419433951378, "train_loss_bc": 0.028726419433951378, "train_loss_llm": 0.0, "grad_norm": 0.10624944418668747, "global_step": 26220, "epoch": 728, "lr": 0.001} +{"train_loss": 0.029455171898007393, "train_loss_bc": 0.029455171898007393, "train_loss_llm": 0.0, "grad_norm": 0.03567542880773544, "global_step": 26221, "epoch": 728, "lr": 0.001} +{"train_loss": 0.03227446600794792, "train_loss_bc": 0.03226654231548309, "train_loss_llm": 0.0007925033569335938, "grad_norm": 0.06528221070766449, "global_step": 26222, "epoch": 728, "lr": 0.001} +{"train_loss": 0.03165128827095032, "train_loss_bc": 0.03164137899875641, "train_loss_llm": 0.0009909570217132568, "grad_norm": 0.09683360159397125, "global_step": 26223, "epoch": 728, "lr": 0.001} +{"train_loss": 0.02846946008503437, "train_loss_bc": 0.02846946008503437, "train_loss_llm": 0.0, "grad_norm": 0.11845258623361588, "global_step": 26224, "epoch": 728, "lr": 0.001} +{"train_loss": 0.030419526621699333, "train_loss_bc": 0.030390817672014236, "train_loss_llm": 0.002870917320251465, "grad_norm": 0.029969027265906334, "global_step": 26225, "epoch": 728, "lr": 0.001} +{"train_loss": 0.0373944528400898, "train_loss_bc": 0.0373944528400898, "train_loss_llm": 0.0, "grad_norm": 0.05697980523109436, "global_step": 26226, "epoch": 728, "lr": 0.001} +{"train_loss": 0.0336008258163929, "train_loss_bc": 0.0335959754884243, "train_loss_llm": 0.0004849433898925781, "grad_norm": 0.08034174889326096, "global_step": 26227, "epoch": 728, "lr": 0.001} +{"train_loss": 0.03122241236269474, "train_loss_bc": 0.03122241236269474, "train_loss_llm": 0.0, "grad_norm": 0.10739614069461823, "global_step": 26228, "epoch": 728, "lr": 0.001} +{"train_loss": 0.035982050001621246, "train_loss_bc": 0.03596067801117897, "train_loss_llm": 0.0021370649337768555, "grad_norm": 0.035485751926898956, "global_step": 26229, "epoch": 728, "lr": 0.001} +{"train_loss": 0.02655761130154133, "train_loss_bc": 0.02655761130154133, "train_loss_llm": 0.0, "grad_norm": 0.05231206864118576, "global_step": 26230, "epoch": 728, "lr": 0.001} +{"train_loss": 0.033563464879989624, "train_loss_bc": 0.03355564549565315, "train_loss_llm": 0.0007819533348083496, "grad_norm": 0.07470739632844925, "global_step": 26231, "epoch": 728, "lr": 0.001} +{"train_loss": 0.03723321855068207, "train_loss_bc": 0.03723321855068207, "train_loss_llm": 0.0, "grad_norm": 0.08374514430761337, "global_step": 26232, "epoch": 728, "lr": 0.001} +{"train_loss": 0.030720915645360947, "train_loss_bc": 0.030720915645360947, "train_loss_llm": 0.0, "grad_norm": 0.034044358879327774, "global_step": 26233, "epoch": 728, "lr": 0.001} +{"train_loss": 0.035336632281541824, "train_loss_bc": 0.035336632281541824, "train_loss_llm": 0.0, "grad_norm": 0.058007773011922836, "global_step": 26234, "epoch": 728, "lr": 0.001} +{"train_loss": 0.03081393614411354, "train_loss_bc": 0.03078276477754116, "train_loss_llm": 0.003117084503173828, "grad_norm": 0.08056892454624176, "global_step": 26235, "epoch": 728, "lr": 0.001} +{"train_loss": 0.03289191424846649, "train_loss_bc": 0.03289191424846649, "train_loss_llm": 0.0, "grad_norm": 0.10576178133487701, "global_step": 26236, "epoch": 728, "lr": 0.001} +{"train_loss": 0.03466587886214256, "train_loss_bc": 0.03466587886214256, "train_loss_llm": 0.0, "grad_norm": 0.04807319864630699, "global_step": 26237, "epoch": 728, "lr": 0.001} +{"train_loss": 0.031332507729530334, "train_loss_bc": 0.03132243454456329, "train_loss_llm": 0.0010072886943817139, "grad_norm": 0.07488428801298141, "global_step": 26238, "epoch": 728, "lr": 0.001} +{"train_loss": 0.03713428974151611, "train_loss_bc": 0.03710184991359711, "train_loss_llm": 0.0032441318035125732, "grad_norm": 0.12095290422439575, "global_step": 26239, "epoch": 728, "lr": 0.001} +{"train_loss": 0.030150724574923515, "train_loss_bc": 0.030150724574923515, "train_loss_llm": 0.0, "grad_norm": 0.15204881131649017, "global_step": 26240, "epoch": 728, "lr": 0.001} +{"train_loss": 0.029272260144352913, "train_loss_bc": 0.029254473745822906, "train_loss_llm": 0.001778632402420044, "grad_norm": 0.03139490634202957, "global_step": 26241, "epoch": 728, "lr": 0.001} +{"train_loss": 0.03449534997344017, "train_loss_bc": 0.0344928577542305, "train_loss_llm": 0.000249326229095459, "grad_norm": 0.06267134100198746, "global_step": 26242, "epoch": 728, "lr": 0.001} +{"train_loss": 0.031303764010469116, "train_loss_bc": 0.02717009000480175, "train_loss_llm": 0.0003598034381866455, "grad_norm": 0.0850369855761528, "global_step": 26243, "epoch": 728, "lr": 0.001, "val_loss": 0.09149828553199768} +{"train_loss": 0.025689449161291122, "train_loss_bc": 0.025689449161291122, "train_loss_llm": 0.0, "grad_norm": 0.10336976498365402, "global_step": 26244, "epoch": 729, "lr": 0.001} +{"train_loss": 0.02880651317536831, "train_loss_bc": 0.028776202350854874, "train_loss_llm": 0.003031134605407715, "grad_norm": 0.029881348833441734, "global_step": 26245, "epoch": 729, "lr": 0.001} +{"train_loss": 0.02636490762233734, "train_loss_bc": 0.02636490762233734, "train_loss_llm": 0.0, "grad_norm": 0.042115550488233566, "global_step": 26246, "epoch": 729, "lr": 0.001} +{"train_loss": 0.031445663422346115, "train_loss_bc": 0.03144212067127228, "train_loss_llm": 0.0003541707992553711, "grad_norm": 0.0589459165930748, "global_step": 26247, "epoch": 729, "lr": 0.001} +{"train_loss": 0.028442833572626114, "train_loss_bc": 0.028442833572626114, "train_loss_llm": 0.0, "grad_norm": 0.0762397050857544, "global_step": 26248, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03396957740187645, "train_loss_bc": 0.03396931663155556, "train_loss_llm": 2.6106834411621094e-05, "grad_norm": 0.03321400657296181, "global_step": 26249, "epoch": 729, "lr": 0.001} +{"train_loss": 0.022358588874340057, "train_loss_bc": 0.022358588874340057, "train_loss_llm": 0.0, "grad_norm": 0.04296288266777992, "global_step": 26250, "epoch": 729, "lr": 0.001} +{"train_loss": 0.030342737212777138, "train_loss_bc": 0.030342737212777138, "train_loss_llm": 0.0, "grad_norm": 0.06585659831762314, "global_step": 26251, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03150011599063873, "train_loss_bc": 0.03150011599063873, "train_loss_llm": 0.0, "grad_norm": 0.0789685994386673, "global_step": 26252, "epoch": 729, "lr": 0.001} +{"train_loss": 0.0364004485309124, "train_loss_bc": 0.0364004485309124, "train_loss_llm": 0.0, "grad_norm": 0.033480603247880936, "global_step": 26253, "epoch": 729, "lr": 0.001} +{"train_loss": 0.02437354438006878, "train_loss_bc": 0.02437354438006878, "train_loss_llm": 0.0, "grad_norm": 0.060596153140068054, "global_step": 26254, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03361744433641434, "train_loss_bc": 0.03361744433641434, "train_loss_llm": 0.0, "grad_norm": 0.08742249011993408, "global_step": 26255, "epoch": 729, "lr": 0.001} +{"train_loss": 0.037447646260261536, "train_loss_bc": 0.037447646260261536, "train_loss_llm": 0.0, "grad_norm": 0.13450032472610474, "global_step": 26256, "epoch": 729, "lr": 0.001} +{"train_loss": 0.033213794231414795, "train_loss_bc": 0.033213794231414795, "train_loss_llm": 0.0, "grad_norm": 0.0399908684194088, "global_step": 26257, "epoch": 729, "lr": 0.001} +{"train_loss": 0.02846652828156948, "train_loss_bc": 0.02846652828156948, "train_loss_llm": 0.0, "grad_norm": 0.06503435224294662, "global_step": 26258, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03197814151644707, "train_loss_bc": 0.03197814151644707, "train_loss_llm": 0.0, "grad_norm": 0.10715622454881668, "global_step": 26259, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03353477269411087, "train_loss_bc": 0.03353477269411087, "train_loss_llm": 0.0, "grad_norm": 0.13670402765274048, "global_step": 26260, "epoch": 729, "lr": 0.001} +{"train_loss": 0.0394851379096508, "train_loss_bc": 0.0394851379096508, "train_loss_llm": 0.0, "grad_norm": 0.05948103964328766, "global_step": 26261, "epoch": 729, "lr": 0.001} +{"train_loss": 0.026882454752922058, "train_loss_bc": 0.026882454752922058, "train_loss_llm": 0.0, "grad_norm": 0.08838249742984772, "global_step": 26262, "epoch": 729, "lr": 0.001} +{"train_loss": 0.025480076670646667, "train_loss_bc": 0.025480076670646667, "train_loss_llm": 0.0, "grad_norm": 0.10537493228912354, "global_step": 26263, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03428393602371216, "train_loss_bc": 0.03428393602371216, "train_loss_llm": 0.0, "grad_norm": 0.13695232570171356, "global_step": 26264, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03352005034685135, "train_loss_bc": 0.03352005034685135, "train_loss_llm": 0.0, "grad_norm": 0.031379688531160355, "global_step": 26265, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03164456784725189, "train_loss_bc": 0.03164456784725189, "train_loss_llm": 0.0, "grad_norm": 0.04812382906675339, "global_step": 26266, "epoch": 729, "lr": 0.001} +{"train_loss": 0.034642092883586884, "train_loss_bc": 0.034642092883586884, "train_loss_llm": 0.0, "grad_norm": 0.06890376657247543, "global_step": 26267, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03263714909553528, "train_loss_bc": 0.03263714909553528, "train_loss_llm": 0.0, "grad_norm": 0.08504755795001984, "global_step": 26268, "epoch": 729, "lr": 0.001} +{"train_loss": 0.027398964390158653, "train_loss_bc": 0.027398964390158653, "train_loss_llm": 0.0, "grad_norm": 0.034437745809555054, "global_step": 26269, "epoch": 729, "lr": 0.001} +{"train_loss": 0.027959425002336502, "train_loss_bc": 0.027936026453971863, "train_loss_llm": 0.002339780330657959, "grad_norm": 0.05778875574469566, "global_step": 26270, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03325773775577545, "train_loss_bc": 0.03324884921312332, "train_loss_llm": 0.000888824462890625, "grad_norm": 0.0845128744840622, "global_step": 26271, "epoch": 729, "lr": 0.001} +{"train_loss": 0.02640264667570591, "train_loss_bc": 0.026387928053736687, "train_loss_llm": 0.0014719367027282715, "grad_norm": 0.11222914606332779, "global_step": 26272, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03552810475230217, "train_loss_bc": 0.03552810475230217, "train_loss_llm": 0.0, "grad_norm": 0.036189012229442596, "global_step": 26273, "epoch": 729, "lr": 0.001} +{"train_loss": 0.029066869989037514, "train_loss_bc": 0.029066869989037514, "train_loss_llm": 0.0, "grad_norm": 0.07050928473472595, "global_step": 26274, "epoch": 729, "lr": 0.001} +{"train_loss": 0.037865933030843735, "train_loss_bc": 0.037865933030843735, "train_loss_llm": 0.0, "grad_norm": 0.10155723243951797, "global_step": 26275, "epoch": 729, "lr": 0.001} +{"train_loss": 0.032837383449077606, "train_loss_bc": 0.032837383449077606, "train_loss_llm": 0.0, "grad_norm": 0.12167390435934067, "global_step": 26276, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03060329519212246, "train_loss_bc": 0.03060329519212246, "train_loss_llm": 0.0, "grad_norm": 0.04610220342874527, "global_step": 26277, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03964127600193024, "train_loss_bc": 0.03964127600193024, "train_loss_llm": 0.0, "grad_norm": 0.08640351891517639, "global_step": 26278, "epoch": 729, "lr": 0.001} +{"train_loss": 0.03133510715431637, "train_loss_bc": 0.030964909121394157, "train_loss_llm": 0.0009140074253082275, "grad_norm": 0.13242334127426147, "global_step": 26279, "epoch": 729, "lr": 0.001, "val_loss": 0.09150437265634537} +{"train_loss": 0.030644668266177177, "train_loss_bc": 0.030644668266177177, "train_loss_llm": 0.0, "grad_norm": 0.16552501916885376, "global_step": 26280, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03271759673953056, "train_loss_bc": 0.03271759673953056, "train_loss_llm": 0.0, "grad_norm": 0.038158271461725235, "global_step": 26281, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03269205987453461, "train_loss_bc": 0.032673560082912445, "train_loss_llm": 0.001849830150604248, "grad_norm": 0.07537970691919327, "global_step": 26282, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03614293038845062, "train_loss_bc": 0.03614293038845062, "train_loss_llm": 0.0, "grad_norm": 0.0969323068857193, "global_step": 26283, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03091122955083847, "train_loss_bc": 0.030887989327311516, "train_loss_llm": 0.0023239850997924805, "grad_norm": 0.1211560070514679, "global_step": 26284, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03435787186026573, "train_loss_bc": 0.034338317811489105, "train_loss_llm": 0.001955270767211914, "grad_norm": 0.03530547767877579, "global_step": 26285, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03124251402914524, "train_loss_bc": 0.03124251402914524, "train_loss_llm": 0.0, "grad_norm": 0.05315239727497101, "global_step": 26286, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03469680994749069, "train_loss_bc": 0.03469680994749069, "train_loss_llm": 0.0, "grad_norm": 0.07518339157104492, "global_step": 26287, "epoch": 730, "lr": 0.001} +{"train_loss": 0.028394710272550583, "train_loss_bc": 0.028394710272550583, "train_loss_llm": 0.0, "grad_norm": 0.0832161009311676, "global_step": 26288, "epoch": 730, "lr": 0.001} +{"train_loss": 0.02838674746453762, "train_loss_bc": 0.02838674746453762, "train_loss_llm": 0.0, "grad_norm": 0.028171921148896217, "global_step": 26289, "epoch": 730, "lr": 0.001} +{"train_loss": 0.030135517939925194, "train_loss_bc": 0.030135517939925194, "train_loss_llm": 0.0, "grad_norm": 0.05972667410969734, "global_step": 26290, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03572604060173035, "train_loss_bc": 0.03572604060173035, "train_loss_llm": 0.0, "grad_norm": 0.0931389108300209, "global_step": 26291, "epoch": 730, "lr": 0.001} +{"train_loss": 0.033665768802165985, "train_loss_bc": 0.033665768802165985, "train_loss_llm": 0.0, "grad_norm": 0.11451531946659088, "global_step": 26292, "epoch": 730, "lr": 0.001} +{"train_loss": 0.030778849497437477, "train_loss_bc": 0.030778849497437477, "train_loss_llm": 0.0, "grad_norm": 0.04589129611849785, "global_step": 26293, "epoch": 730, "lr": 0.001} +{"train_loss": 0.04066195711493492, "train_loss_bc": 0.04066195711493492, "train_loss_llm": 0.0, "grad_norm": 0.07815736532211304, "global_step": 26294, "epoch": 730, "lr": 0.001} +{"train_loss": 0.031104233115911484, "train_loss_bc": 0.031104233115911484, "train_loss_llm": 0.0, "grad_norm": 0.10999501496553421, "global_step": 26295, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03619914874434471, "train_loss_bc": 0.03619914874434471, "train_loss_llm": 0.0, "grad_norm": 0.15287983417510986, "global_step": 26296, "epoch": 730, "lr": 0.001} +{"train_loss": 0.027494432404637337, "train_loss_bc": 0.027494432404637337, "train_loss_llm": 0.0, "grad_norm": 0.04493092745542526, "global_step": 26297, "epoch": 730, "lr": 0.001} +{"train_loss": 0.025025567039847374, "train_loss_bc": 0.025025567039847374, "train_loss_llm": 0.0, "grad_norm": 0.05791385844349861, "global_step": 26298, "epoch": 730, "lr": 0.001} +{"train_loss": 0.024029748514294624, "train_loss_bc": 0.024029748514294624, "train_loss_llm": 0.0, "grad_norm": 0.08405088633298874, "global_step": 26299, "epoch": 730, "lr": 0.001} +{"train_loss": 0.029119258746504784, "train_loss_bc": 0.029119258746504784, "train_loss_llm": 0.0, "grad_norm": 0.1100311428308487, "global_step": 26300, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03302370011806488, "train_loss_bc": 0.03302370011806488, "train_loss_llm": 0.0, "grad_norm": 0.037895843386650085, "global_step": 26301, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03461509197950363, "train_loss_bc": 0.03461509197950363, "train_loss_llm": 0.0, "grad_norm": 0.08325116336345673, "global_step": 26302, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03635787591338158, "train_loss_bc": 0.03635787591338158, "train_loss_llm": 0.0, "grad_norm": 0.1241549551486969, "global_step": 26303, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03737252578139305, "train_loss_bc": 0.03737252578139305, "train_loss_llm": 0.0, "grad_norm": 0.15123671293258667, "global_step": 26304, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03292214125394821, "train_loss_bc": 0.03292214125394821, "train_loss_llm": 0.0, "grad_norm": 0.028517961502075195, "global_step": 26305, "epoch": 730, "lr": 0.001} +{"train_loss": 0.022348511964082718, "train_loss_bc": 0.022348511964082718, "train_loss_llm": 0.0, "grad_norm": 0.0391341969370842, "global_step": 26306, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03268499672412872, "train_loss_bc": 0.03268499672412872, "train_loss_llm": 0.0, "grad_norm": 0.05939493328332901, "global_step": 26307, "epoch": 730, "lr": 0.001} +{"train_loss": 0.030566509813070297, "train_loss_bc": 0.030566509813070297, "train_loss_llm": 0.0, "grad_norm": 0.0736946165561676, "global_step": 26308, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03053886629641056, "train_loss_bc": 0.03050130419433117, "train_loss_llm": 0.003756195306777954, "grad_norm": 0.047320831567049026, "global_step": 26309, "epoch": 730, "lr": 0.001} +{"train_loss": 0.026212261989712715, "train_loss_bc": 0.026212261989712715, "train_loss_llm": 0.0, "grad_norm": 0.07751026749610901, "global_step": 26310, "epoch": 730, "lr": 0.001} +{"train_loss": 0.028296053409576416, "train_loss_bc": 0.028296053409576416, "train_loss_llm": 0.0, "grad_norm": 0.10062087327241898, "global_step": 26311, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03128072991967201, "train_loss_bc": 0.031273066997528076, "train_loss_llm": 0.0007663965225219727, "grad_norm": 0.13771818578243256, "global_step": 26312, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03138351812958717, "train_loss_bc": 0.03136520832777023, "train_loss_llm": 0.0018308758735656738, "grad_norm": 0.05320826172828674, "global_step": 26313, "epoch": 730, "lr": 0.001} +{"train_loss": 0.03228100389242172, "train_loss_bc": 0.03228100389242172, "train_loss_llm": 0.0, "grad_norm": 0.09797237813472748, "global_step": 26314, "epoch": 730, "lr": 0.001} +{"train_loss": 0.031732472435881696, "train_loss_bc": 0.03834479674696922, "train_loss_llm": 0.001276165246963501, "grad_norm": 0.12145987153053284, "global_step": 26315, "epoch": 730, "lr": 0.001, "train/cumulative_reward": 93.29600657886154, "train/mean_score": 13.894248408791544, "train/success_rate": 0.3, "test/cumulative_reward": 41.13043660036487, "test/mean_score": 6.391325202772715, "test/success_rate": 0.14, "val_loss": 0.09385774284601212, "train_action_mse_error": 0.03598422557115555} +{"train_loss": 0.0319335050880909, "train_loss_bc": 0.03192755952477455, "train_loss_llm": 0.0005946755409240723, "grad_norm": 0.1543082594871521, "global_step": 26316, "epoch": 731, "lr": 0.001} +{"train_loss": 0.03256401792168617, "train_loss_bc": 0.03256401792168617, "train_loss_llm": 0.0, "grad_norm": 0.03298544883728027, "global_step": 26317, "epoch": 731, "lr": 0.001} +{"train_loss": 0.03037930838763714, "train_loss_bc": 0.030375773087143898, "train_loss_llm": 0.0003534853458404541, "grad_norm": 0.05606430023908615, "global_step": 26318, "epoch": 731, "lr": 0.001} +{"train_loss": 0.0345049612224102, "train_loss_bc": 0.0345035083591938, "train_loss_llm": 0.00014525651931762695, "grad_norm": 0.10416378825902939, "global_step": 26319, "epoch": 731, "lr": 0.001} +{"train_loss": 0.032178305089473724, "train_loss_bc": 0.032178305089473724, "train_loss_llm": 0.0, "grad_norm": 0.1367575079202652, "global_step": 26320, "epoch": 731, "lr": 0.001} +{"train_loss": 0.03414298593997955, "train_loss_bc": 0.034118253737688065, "train_loss_llm": 0.0024733245372772217, "grad_norm": 0.03792978450655937, "global_step": 26321, "epoch": 731, "lr": 0.001} +{"train_loss": 0.031791336834430695, "train_loss_bc": 0.031791336834430695, "train_loss_llm": 0.0, "grad_norm": 0.06514185667037964, "global_step": 26322, "epoch": 731, "lr": 0.001} +{"train_loss": 0.033799365162849426, "train_loss_bc": 0.03378356993198395, "train_loss_llm": 0.0015796124935150146, "grad_norm": 0.08176791667938232, "global_step": 26323, "epoch": 731, "lr": 0.001} +{"train_loss": 0.026398740708827972, "train_loss_bc": 0.026397790759801865, "train_loss_llm": 9.506940841674805e-05, "grad_norm": 0.09838621318340302, "global_step": 26324, "epoch": 731, "lr": 0.001} +{"train_loss": 0.0322725884616375, "train_loss_bc": 0.0322725884616375, "train_loss_llm": 0.0, "grad_norm": 0.03455181419849396, "global_step": 26325, "epoch": 731, "lr": 0.001} +{"train_loss": 0.02918533980846405, "train_loss_bc": 0.029173709452152252, "train_loss_llm": 0.0011630356311798096, "grad_norm": 0.05031896010041237, "global_step": 26326, "epoch": 731, "lr": 0.001} +{"train_loss": 0.03336907550692558, "train_loss_bc": 0.033325761556625366, "train_loss_llm": 0.004331499338150024, "grad_norm": 0.06938629597425461, "global_step": 26327, "epoch": 731, "lr": 0.001} +{"train_loss": 0.033618614077568054, "train_loss_bc": 0.0336170457303524, "train_loss_llm": 0.0001569688320159912, "grad_norm": 0.09345179796218872, "global_step": 26328, "epoch": 731, "lr": 0.001} +{"train_loss": 0.037675850093364716, "train_loss_bc": 0.037675850093364716, "train_loss_llm": 0.0, "grad_norm": 0.04968959838151932, "global_step": 26329, "epoch": 731, "lr": 0.001} +{"train_loss": 0.033316612243652344, "train_loss_bc": 0.033316612243652344, "train_loss_llm": 0.0, "grad_norm": 0.07418287545442581, "global_step": 26330, "epoch": 731, "lr": 0.001} +{"train_loss": 0.03091469779610634, "train_loss_bc": 0.03091469779610634, "train_loss_llm": 0.0, "grad_norm": 0.104298435151577, "global_step": 26331, "epoch": 731, "lr": 0.001} +{"train_loss": 0.032242655754089355, "train_loss_bc": 0.032242655754089355, "train_loss_llm": 0.0, "grad_norm": 0.12949992716312408, "global_step": 26332, "epoch": 731, "lr": 0.001} +{"train_loss": 0.030450720340013504, "train_loss_bc": 0.030450720340013504, "train_loss_llm": 0.0, "grad_norm": 0.037619758397340775, "global_step": 26333, "epoch": 731, "lr": 0.001} +{"train_loss": 0.032089412212371826, "train_loss_bc": 0.032089412212371826, "train_loss_llm": 0.0, "grad_norm": 0.07617189735174179, "global_step": 26334, "epoch": 731, "lr": 0.001} +{"train_loss": 0.02474263682961464, "train_loss_bc": 0.02474263682961464, "train_loss_llm": 0.0, "grad_norm": 0.10314471274614334, "global_step": 26335, "epoch": 731, "lr": 0.001} +{"train_loss": 0.030228879302740097, "train_loss_bc": 0.030228879302740097, "train_loss_llm": 0.0, "grad_norm": 0.132248193025589, "global_step": 26336, "epoch": 731, "lr": 0.001} +{"train_loss": 0.031048091128468513, "train_loss_bc": 0.031048091128468513, "train_loss_llm": 0.0, "grad_norm": 0.04614308103919029, "global_step": 26337, "epoch": 731, "lr": 0.001} +{"train_loss": 0.03715265169739723, "train_loss_bc": 0.03715265169739723, "train_loss_llm": 0.0, "grad_norm": 0.07973352074623108, "global_step": 26338, "epoch": 731, "lr": 0.001} +{"train_loss": 0.029516061767935753, "train_loss_bc": 0.029516061767935753, "train_loss_llm": 0.0, "grad_norm": 0.1253872960805893, "global_step": 26339, "epoch": 731, "lr": 0.001} +{"train_loss": 0.033392008394002914, "train_loss_bc": 0.03337009996175766, "train_loss_llm": 0.0021909475326538086, "grad_norm": 0.16674980521202087, "global_step": 26340, "epoch": 731, "lr": 0.001} +{"train_loss": 0.030669186264276505, "train_loss_bc": 0.030669186264276505, "train_loss_llm": 0.0, "grad_norm": 0.03960466384887695, "global_step": 26341, "epoch": 731, "lr": 0.001} +{"train_loss": 0.032921236008405685, "train_loss_bc": 0.032921236008405685, "train_loss_llm": 0.0, "grad_norm": 0.06354622542858124, "global_step": 26342, "epoch": 731, "lr": 0.001} +{"train_loss": 0.032118059694767, "train_loss_bc": 0.032118059694767, "train_loss_llm": 0.0, "grad_norm": 0.09222082048654556, "global_step": 26343, "epoch": 731, "lr": 0.001} +{"train_loss": 0.025876380503177643, "train_loss_bc": 0.025876380503177643, "train_loss_llm": 0.0, "grad_norm": 0.11230268329381943, "global_step": 26344, "epoch": 731, "lr": 0.001} +{"train_loss": 0.02660706266760826, "train_loss_bc": 0.02660706266760826, "train_loss_llm": 0.0, "grad_norm": 0.0294137354940176, "global_step": 26345, "epoch": 731, "lr": 0.001} +{"train_loss": 0.02917347475886345, "train_loss_bc": 0.02917347475886345, "train_loss_llm": 0.0, "grad_norm": 0.055250126868486404, "global_step": 26346, "epoch": 731, "lr": 0.001} +{"train_loss": 0.03193662315607071, "train_loss_bc": 0.03193662315607071, "train_loss_llm": 0.0, "grad_norm": 0.06685417890548706, "global_step": 26347, "epoch": 731, "lr": 0.001} +{"train_loss": 0.029946014285087585, "train_loss_bc": 0.02993505448102951, "train_loss_llm": 0.0010959208011627197, "grad_norm": 0.08205993473529816, "global_step": 26348, "epoch": 731, "lr": 0.001} +{"train_loss": 0.03146352246403694, "train_loss_bc": 0.03146352246403694, "train_loss_llm": 0.0, "grad_norm": 0.0392073355615139, "global_step": 26349, "epoch": 731, "lr": 0.001} +{"train_loss": 0.03322288393974304, "train_loss_bc": 0.03322288393974304, "train_loss_llm": 0.0, "grad_norm": 0.07522232830524445, "global_step": 26350, "epoch": 731, "lr": 0.001} +{"train_loss": 0.03158922384803494, "train_loss_bc": 0.03436318784952164, "train_loss_llm": 0.0006004571914672852, "grad_norm": 0.10850328952074051, "global_step": 26351, "epoch": 731, "lr": 0.001, "val_loss": 0.09350384026765823} +{"train_loss": 0.033150773495435715, "train_loss_bc": 0.033150773495435715, "train_loss_llm": 0.0, "grad_norm": 0.1465311497449875, "global_step": 26352, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03170245885848999, "train_loss_bc": 0.03168994188308716, "train_loss_llm": 0.001251816749572754, "grad_norm": 0.05971216782927513, "global_step": 26353, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03459666669368744, "train_loss_bc": 0.03459249809384346, "train_loss_llm": 0.0004170238971710205, "grad_norm": 0.0961000993847847, "global_step": 26354, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03387425094842911, "train_loss_bc": 0.03387425094842911, "train_loss_llm": 0.0, "grad_norm": 0.1324545443058014, "global_step": 26355, "epoch": 732, "lr": 0.001} +{"train_loss": 0.02939859963953495, "train_loss_bc": 0.02939458005130291, "train_loss_llm": 0.00040200352668762207, "grad_norm": 0.16730989515781403, "global_step": 26356, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03343496471643448, "train_loss_bc": 0.033400848507881165, "train_loss_llm": 0.0034114718437194824, "grad_norm": 0.042403239756822586, "global_step": 26357, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03484685719013214, "train_loss_bc": 0.03484251722693443, "train_loss_llm": 0.00043395161628723145, "grad_norm": 0.07437709718942642, "global_step": 26358, "epoch": 732, "lr": 0.001} +{"train_loss": 0.0317719429731369, "train_loss_bc": 0.0317719429731369, "train_loss_llm": 0.0, "grad_norm": 0.09943272173404694, "global_step": 26359, "epoch": 732, "lr": 0.001} +{"train_loss": 0.031231679022312164, "train_loss_bc": 0.031231679022312164, "train_loss_llm": 0.0, "grad_norm": 0.1262994110584259, "global_step": 26360, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03438865393400192, "train_loss_bc": 0.03438865393400192, "train_loss_llm": 0.0, "grad_norm": 0.04469429329037666, "global_step": 26361, "epoch": 732, "lr": 0.001} +{"train_loss": 0.02568790502846241, "train_loss_bc": 0.02568790502846241, "train_loss_llm": 0.0, "grad_norm": 0.0687568336725235, "global_step": 26362, "epoch": 732, "lr": 0.001} +{"train_loss": 0.032913800328969955, "train_loss_bc": 0.032913800328969955, "train_loss_llm": 0.0, "grad_norm": 0.096431203186512, "global_step": 26363, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03123946487903595, "train_loss_bc": 0.03123946487903595, "train_loss_llm": 0.0, "grad_norm": 0.12286151945590973, "global_step": 26364, "epoch": 732, "lr": 0.001} +{"train_loss": 0.02114110067486763, "train_loss_bc": 0.02114110067486763, "train_loss_llm": 0.0, "grad_norm": 0.023468362167477608, "global_step": 26365, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03141423314809799, "train_loss_bc": 0.03141423314809799, "train_loss_llm": 0.0, "grad_norm": 0.05463612079620361, "global_step": 26366, "epoch": 732, "lr": 0.001} +{"train_loss": 0.02552676759660244, "train_loss_bc": 0.02552676759660244, "train_loss_llm": 0.0, "grad_norm": 0.07694844901561737, "global_step": 26367, "epoch": 732, "lr": 0.001} +{"train_loss": 0.033122334629297256, "train_loss_bc": 0.033122334629297256, "train_loss_llm": 0.0, "grad_norm": 0.0908408984541893, "global_step": 26368, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03013642132282257, "train_loss_bc": 0.030126113444566727, "train_loss_llm": 0.0010308325290679932, "grad_norm": 0.04750286415219307, "global_step": 26369, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03148816153407097, "train_loss_bc": 0.03148816153407097, "train_loss_llm": 0.0, "grad_norm": 0.07099762558937073, "global_step": 26370, "epoch": 732, "lr": 0.001} +{"train_loss": 0.030481960624456406, "train_loss_bc": 0.030481960624456406, "train_loss_llm": 0.0, "grad_norm": 0.09283806383609772, "global_step": 26371, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03682868555188179, "train_loss_bc": 0.03682868555188179, "train_loss_llm": 0.0, "grad_norm": 0.10213388502597809, "global_step": 26372, "epoch": 732, "lr": 0.001} +{"train_loss": 0.029973339289426804, "train_loss_bc": 0.029960546642541885, "train_loss_llm": 0.0012791752815246582, "grad_norm": 0.04509874805808067, "global_step": 26373, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03470342233777046, "train_loss_bc": 0.03470342233777046, "train_loss_llm": 0.0, "grad_norm": 0.07543926686048508, "global_step": 26374, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03399591147899628, "train_loss_bc": 0.03399591147899628, "train_loss_llm": 0.0, "grad_norm": 0.11413808166980743, "global_step": 26375, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03748316317796707, "train_loss_bc": 0.03748316317796707, "train_loss_llm": 0.0, "grad_norm": 0.1594792604446411, "global_step": 26376, "epoch": 732, "lr": 0.001} +{"train_loss": 0.0346447117626667, "train_loss_bc": 0.0346447117626667, "train_loss_llm": 0.0, "grad_norm": 0.04739365726709366, "global_step": 26377, "epoch": 732, "lr": 0.001} +{"train_loss": 0.02915196120738983, "train_loss_bc": 0.02915196120738983, "train_loss_llm": 0.0, "grad_norm": 0.0756182074546814, "global_step": 26378, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03441403806209564, "train_loss_bc": 0.03441403806209564, "train_loss_llm": 0.0, "grad_norm": 0.11115805059671402, "global_step": 26379, "epoch": 732, "lr": 0.001} +{"train_loss": 0.02604881301522255, "train_loss_bc": 0.02604881301522255, "train_loss_llm": 0.0, "grad_norm": 0.13684998452663422, "global_step": 26380, "epoch": 732, "lr": 0.001} +{"train_loss": 0.033787474036216736, "train_loss_bc": 0.03377974033355713, "train_loss_llm": 0.0007734000682830811, "grad_norm": 0.032156724482774734, "global_step": 26381, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03428829461336136, "train_loss_bc": 0.03428829461336136, "train_loss_llm": 0.0, "grad_norm": 0.059093572199344635, "global_step": 26382, "epoch": 732, "lr": 0.001} +{"train_loss": 0.029835285618901253, "train_loss_bc": 0.029835285618901253, "train_loss_llm": 0.0, "grad_norm": 0.08485615998506546, "global_step": 26383, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03255538269877434, "train_loss_bc": 0.03254613280296326, "train_loss_llm": 0.000924915075302124, "grad_norm": 0.1120956763625145, "global_step": 26384, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03413834422826767, "train_loss_bc": 0.03412405028939247, "train_loss_llm": 0.001429438591003418, "grad_norm": 0.03806636482477188, "global_step": 26385, "epoch": 732, "lr": 0.001} +{"train_loss": 0.027721593156456947, "train_loss_bc": 0.027721593156456947, "train_loss_llm": 0.0, "grad_norm": 0.0489484965801239, "global_step": 26386, "epoch": 732, "lr": 0.001} +{"train_loss": 0.03168806656160288, "train_loss_bc": 0.02965097874403, "train_loss_llm": 0.0, "grad_norm": 0.07513566315174103, "global_step": 26387, "epoch": 732, "lr": 0.001, "val_loss": 0.09054966270923615} +{"train_loss": 0.0251504834741354, "train_loss_bc": 0.0251504834741354, "train_loss_llm": 0.0, "grad_norm": 0.09315536916255951, "global_step": 26388, "epoch": 733, "lr": 0.001} +{"train_loss": 0.02766682393848896, "train_loss_bc": 0.02766682393848896, "train_loss_llm": 0.0, "grad_norm": 0.032846082001924515, "global_step": 26389, "epoch": 733, "lr": 0.001} +{"train_loss": 0.027162425220012665, "train_loss_bc": 0.027162425220012665, "train_loss_llm": 0.0, "grad_norm": 0.053761839866638184, "global_step": 26390, "epoch": 733, "lr": 0.001} +{"train_loss": 0.033677682280540466, "train_loss_bc": 0.033677682280540466, "train_loss_llm": 0.0, "grad_norm": 0.08879946917295456, "global_step": 26391, "epoch": 733, "lr": 0.001} +{"train_loss": 0.0344984196126461, "train_loss_bc": 0.03448963165283203, "train_loss_llm": 0.0008789002895355225, "grad_norm": 0.12278908491134644, "global_step": 26392, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03179743140935898, "train_loss_bc": 0.03179516643285751, "train_loss_llm": 0.00022667646408081055, "grad_norm": 0.04049285873770714, "global_step": 26393, "epoch": 733, "lr": 0.001} +{"train_loss": 0.034823089838027954, "train_loss_bc": 0.034823089838027954, "train_loss_llm": 0.0, "grad_norm": 0.06822968274354935, "global_step": 26394, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03175116702914238, "train_loss_bc": 0.03174767643213272, "train_loss_llm": 0.0003490447998046875, "grad_norm": 0.10322831571102142, "global_step": 26395, "epoch": 733, "lr": 0.001} +{"train_loss": 0.030816389247775078, "train_loss_bc": 0.030816389247775078, "train_loss_llm": 0.0, "grad_norm": 0.12771552801132202, "global_step": 26396, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03469693660736084, "train_loss_bc": 0.03469693660736084, "train_loss_llm": 0.0, "grad_norm": 0.04694385081529617, "global_step": 26397, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03344191983342171, "train_loss_bc": 0.03344191983342171, "train_loss_llm": 0.0, "grad_norm": 0.07248027622699738, "global_step": 26398, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03166865184903145, "train_loss_bc": 0.03166865184903145, "train_loss_llm": 0.0, "grad_norm": 0.09188660979270935, "global_step": 26399, "epoch": 733, "lr": 0.001} +{"train_loss": 0.028006816282868385, "train_loss_bc": 0.028006816282868385, "train_loss_llm": 0.0, "grad_norm": 0.13262583315372467, "global_step": 26400, "epoch": 733, "lr": 0.001} +{"train_loss": 0.035261370241642, "train_loss_bc": 0.035261370241642, "train_loss_llm": 0.0, "grad_norm": 0.03056507743895054, "global_step": 26401, "epoch": 733, "lr": 0.001} +{"train_loss": 0.027201127260923386, "train_loss_bc": 0.027201127260923386, "train_loss_llm": 0.0, "grad_norm": 0.05386632680892944, "global_step": 26402, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03135767579078674, "train_loss_bc": 0.03135767579078674, "train_loss_llm": 0.0, "grad_norm": 0.06584364175796509, "global_step": 26403, "epoch": 733, "lr": 0.001} +{"train_loss": 0.02725987881422043, "train_loss_bc": 0.02725987881422043, "train_loss_llm": 0.0, "grad_norm": 0.08638882637023926, "global_step": 26404, "epoch": 733, "lr": 0.001} +{"train_loss": 0.033196836709976196, "train_loss_bc": 0.03318929299712181, "train_loss_llm": 0.0007542669773101807, "grad_norm": 0.03797352686524391, "global_step": 26405, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03605102375149727, "train_loss_bc": 0.03605102375149727, "train_loss_llm": 0.0, "grad_norm": 0.060398366302251816, "global_step": 26406, "epoch": 733, "lr": 0.001} +{"train_loss": 0.029144976288080215, "train_loss_bc": 0.029144976288080215, "train_loss_llm": 0.0, "grad_norm": 0.0732768103480339, "global_step": 26407, "epoch": 733, "lr": 0.001} +{"train_loss": 0.029409002512693405, "train_loss_bc": 0.029409002512693405, "train_loss_llm": 0.0, "grad_norm": 0.09856173396110535, "global_step": 26408, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03212780877947807, "train_loss_bc": 0.032116036862134933, "train_loss_llm": 0.0011772513389587402, "grad_norm": 0.04672427102923393, "global_step": 26409, "epoch": 733, "lr": 0.001} +{"train_loss": 0.031416457146406174, "train_loss_bc": 0.031416457146406174, "train_loss_llm": 0.0, "grad_norm": 0.06509576737880707, "global_step": 26410, "epoch": 733, "lr": 0.001} +{"train_loss": 0.034084025770425797, "train_loss_bc": 0.03406490385532379, "train_loss_llm": 0.0019123256206512451, "grad_norm": 0.09276331961154938, "global_step": 26411, "epoch": 733, "lr": 0.001} +{"train_loss": 0.028490973636507988, "train_loss_bc": 0.028490973636507988, "train_loss_llm": 0.0, "grad_norm": 0.12424392998218536, "global_step": 26412, "epoch": 733, "lr": 0.001} +{"train_loss": 0.024135688319802284, "train_loss_bc": 0.024119121953845024, "train_loss_llm": 0.0016566216945648193, "grad_norm": 0.034325242042541504, "global_step": 26413, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03048856370151043, "train_loss_bc": 0.030454004183411598, "train_loss_llm": 0.003456026315689087, "grad_norm": 0.06792601197957993, "global_step": 26414, "epoch": 733, "lr": 0.001} +{"train_loss": 0.02951030619442463, "train_loss_bc": 0.02951030619442463, "train_loss_llm": 0.0, "grad_norm": 0.09177044034004211, "global_step": 26415, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03731244429945946, "train_loss_bc": 0.03731244429945946, "train_loss_llm": 0.0, "grad_norm": 0.118568554520607, "global_step": 26416, "epoch": 733, "lr": 0.001} +{"train_loss": 0.035736069083213806, "train_loss_bc": 0.035736069083213806, "train_loss_llm": 0.0, "grad_norm": 0.04308304563164711, "global_step": 26417, "epoch": 733, "lr": 0.001} +{"train_loss": 0.031920719891786575, "train_loss_bc": 0.03191116079688072, "train_loss_llm": 0.0009560286998748779, "grad_norm": 0.07670409977436066, "global_step": 26418, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03394626826047897, "train_loss_bc": 0.03394626826047897, "train_loss_llm": 0.0, "grad_norm": 0.10532090812921524, "global_step": 26419, "epoch": 733, "lr": 0.001} +{"train_loss": 0.035477567464113235, "train_loss_bc": 0.03546721488237381, "train_loss_llm": 0.0010353326797485352, "grad_norm": 0.14297069609165192, "global_step": 26420, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03488645702600479, "train_loss_bc": 0.03488645702600479, "train_loss_llm": 0.0, "grad_norm": 0.04269908368587494, "global_step": 26421, "epoch": 733, "lr": 0.001} +{"train_loss": 0.03075949102640152, "train_loss_bc": 0.03073612041771412, "train_loss_llm": 0.0023369789123535156, "grad_norm": 0.061539940536022186, "global_step": 26422, "epoch": 733, "lr": 0.001} +{"train_loss": 0.031608962557382055, "train_loss_bc": 0.03357812762260437, "train_loss_llm": 0.001155465841293335, "grad_norm": 0.0924341231584549, "global_step": 26423, "epoch": 733, "lr": 0.001, "val_loss": 0.0908244401216507} +{"train_loss": 0.025978650897741318, "train_loss_bc": 0.02597207762300968, "train_loss_llm": 0.0006573200225830078, "grad_norm": 0.10946276038885117, "global_step": 26424, "epoch": 734, "lr": 0.001} +{"train_loss": 0.032683782279491425, "train_loss_bc": 0.032683782279491425, "train_loss_llm": 0.0, "grad_norm": 0.03329550847411156, "global_step": 26425, "epoch": 734, "lr": 0.001} +{"train_loss": 0.025184396654367447, "train_loss_bc": 0.025184396654367447, "train_loss_llm": 0.0, "grad_norm": 0.04980038106441498, "global_step": 26426, "epoch": 734, "lr": 0.001} +{"train_loss": 0.02999655343592167, "train_loss_bc": 0.02999655343592167, "train_loss_llm": 0.0, "grad_norm": 0.05712810158729553, "global_step": 26427, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03156667202711105, "train_loss_bc": 0.031547561287879944, "train_loss_llm": 0.001911252737045288, "grad_norm": 0.06354779750108719, "global_step": 26428, "epoch": 734, "lr": 0.001} +{"train_loss": 0.035506948828697205, "train_loss_bc": 0.035506948828697205, "train_loss_llm": 0.0, "grad_norm": 0.03590143471956253, "global_step": 26429, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03330151364207268, "train_loss_bc": 0.03330151364207268, "train_loss_llm": 0.0, "grad_norm": 0.05794215202331543, "global_step": 26430, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03574449196457863, "train_loss_bc": 0.035740725696086884, "train_loss_llm": 0.0003765225410461426, "grad_norm": 0.08941931277513504, "global_step": 26431, "epoch": 734, "lr": 0.001} +{"train_loss": 0.027331247925758362, "train_loss_bc": 0.027331247925758362, "train_loss_llm": 0.0, "grad_norm": 0.11702066659927368, "global_step": 26432, "epoch": 734, "lr": 0.001} +{"train_loss": 0.033952414989471436, "train_loss_bc": 0.033943068236112595, "train_loss_llm": 0.0009346306324005127, "grad_norm": 0.037263739854097366, "global_step": 26433, "epoch": 734, "lr": 0.001} +{"train_loss": 0.02305232360959053, "train_loss_bc": 0.02305232360959053, "train_loss_llm": 0.0, "grad_norm": 0.0640394389629364, "global_step": 26434, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03415786474943161, "train_loss_bc": 0.03415786474943161, "train_loss_llm": 0.0, "grad_norm": 0.0874977633357048, "global_step": 26435, "epoch": 734, "lr": 0.001} +{"train_loss": 0.029622824862599373, "train_loss_bc": 0.029622824862599373, "train_loss_llm": 0.0, "grad_norm": 0.10966784507036209, "global_step": 26436, "epoch": 734, "lr": 0.001} +{"train_loss": 0.034955769777297974, "train_loss_bc": 0.034955769777297974, "train_loss_llm": 0.0, "grad_norm": 0.0411582812666893, "global_step": 26437, "epoch": 734, "lr": 0.001} +{"train_loss": 0.02847525291144848, "train_loss_bc": 0.02847525291144848, "train_loss_llm": 0.0, "grad_norm": 0.07003958523273468, "global_step": 26438, "epoch": 734, "lr": 0.001} +{"train_loss": 0.035729870200157166, "train_loss_bc": 0.035729870200157166, "train_loss_llm": 0.0, "grad_norm": 0.10713528841733932, "global_step": 26439, "epoch": 734, "lr": 0.001} +{"train_loss": 0.02911880798637867, "train_loss_bc": 0.029117411002516747, "train_loss_llm": 0.00013974308967590332, "grad_norm": 0.12823767960071564, "global_step": 26440, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03482216224074364, "train_loss_bc": 0.03482216224074364, "train_loss_llm": 0.0, "grad_norm": 0.03624339401721954, "global_step": 26441, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03211195394396782, "train_loss_bc": 0.03211195394396782, "train_loss_llm": 0.0, "grad_norm": 0.0627584382891655, "global_step": 26442, "epoch": 734, "lr": 0.001} +{"train_loss": 0.026914892718195915, "train_loss_bc": 0.026914892718195915, "train_loss_llm": 0.0, "grad_norm": 0.07820649445056915, "global_step": 26443, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03735443577170372, "train_loss_bc": 0.037352047860622406, "train_loss_llm": 0.00023877620697021484, "grad_norm": 0.09608446061611176, "global_step": 26444, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03188458830118179, "train_loss_bc": 0.03188458830118179, "train_loss_llm": 0.0, "grad_norm": 0.028568092733621597, "global_step": 26445, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03130004554986954, "train_loss_bc": 0.0312681682407856, "train_loss_llm": 0.003187835216522217, "grad_norm": 0.04444955289363861, "global_step": 26446, "epoch": 734, "lr": 0.001} +{"train_loss": 0.028778547421097755, "train_loss_bc": 0.028778547421097755, "train_loss_llm": 0.0, "grad_norm": 0.058656416833400726, "global_step": 26447, "epoch": 734, "lr": 0.001} +{"train_loss": 0.02780509740114212, "train_loss_bc": 0.02780509740114212, "train_loss_llm": 0.0, "grad_norm": 0.06625797599554062, "global_step": 26448, "epoch": 734, "lr": 0.001} +{"train_loss": 0.034628141671419144, "train_loss_bc": 0.034623559564352036, "train_loss_llm": 0.0004583299160003662, "grad_norm": 0.029459446668624878, "global_step": 26449, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03078390844166279, "train_loss_bc": 0.03078390844166279, "train_loss_llm": 0.0, "grad_norm": 0.042756374925374985, "global_step": 26450, "epoch": 734, "lr": 0.001} +{"train_loss": 0.034448057413101196, "train_loss_bc": 0.034448057413101196, "train_loss_llm": 0.0, "grad_norm": 0.056564681231975555, "global_step": 26451, "epoch": 734, "lr": 0.001} +{"train_loss": 0.029412467032670975, "train_loss_bc": 0.029412467032670975, "train_loss_llm": 0.0, "grad_norm": 0.07353915274143219, "global_step": 26452, "epoch": 734, "lr": 0.001} +{"train_loss": 0.031668998301029205, "train_loss_bc": 0.03166528791189194, "train_loss_llm": 0.00037109851837158203, "grad_norm": 0.03475054353475571, "global_step": 26453, "epoch": 734, "lr": 0.001} +{"train_loss": 0.02857966721057892, "train_loss_bc": 0.02857966721057892, "train_loss_llm": 0.0, "grad_norm": 0.05290617793798447, "global_step": 26454, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03355386108160019, "train_loss_bc": 0.03355386108160019, "train_loss_llm": 0.0, "grad_norm": 0.07951118052005768, "global_step": 26455, "epoch": 734, "lr": 0.001} +{"train_loss": 0.027855737134814262, "train_loss_bc": 0.02785404585301876, "train_loss_llm": 0.0001691579818725586, "grad_norm": 0.09963070601224899, "global_step": 26456, "epoch": 734, "lr": 0.001} +{"train_loss": 0.036098331212997437, "train_loss_bc": 0.036098331212997437, "train_loss_llm": 0.0, "grad_norm": 0.04368533566594124, "global_step": 26457, "epoch": 734, "lr": 0.001} +{"train_loss": 0.02846326120197773, "train_loss_bc": 0.028452731668949127, "train_loss_llm": 0.0010529756546020508, "grad_norm": 0.0698588415980339, "global_step": 26458, "epoch": 734, "lr": 0.001} +{"train_loss": 0.03114664347635375, "train_loss_bc": 0.028455624356865883, "train_loss_llm": 0.0, "grad_norm": 0.08672896027565002, "global_step": 26459, "epoch": 734, "lr": 0.001, "val_loss": 0.0914110317826271} +{"train_loss": 0.029973993077874184, "train_loss_bc": 0.029973993077874184, "train_loss_llm": 0.0, "grad_norm": 0.1107550784945488, "global_step": 26460, "epoch": 735, "lr": 0.001} +{"train_loss": 0.026177383959293365, "train_loss_bc": 0.026177383959293365, "train_loss_llm": 0.0, "grad_norm": 0.037797778844833374, "global_step": 26461, "epoch": 735, "lr": 0.001} +{"train_loss": 0.038542669266462326, "train_loss_bc": 0.038542669266462326, "train_loss_llm": 0.0, "grad_norm": 0.06252092123031616, "global_step": 26462, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03079092875123024, "train_loss_bc": 0.030785243958234787, "train_loss_llm": 0.0005684792995452881, "grad_norm": 0.0825977548956871, "global_step": 26463, "epoch": 735, "lr": 0.001} +{"train_loss": 0.029389163479208946, "train_loss_bc": 0.029389163479208946, "train_loss_llm": 0.0, "grad_norm": 0.09393996000289917, "global_step": 26464, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03341563045978546, "train_loss_bc": 0.03341563045978546, "train_loss_llm": 0.0, "grad_norm": 0.03227446228265762, "global_step": 26465, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03392349183559418, "train_loss_bc": 0.03391770273447037, "train_loss_llm": 0.0005787312984466553, "grad_norm": 0.04588199779391289, "global_step": 26466, "epoch": 735, "lr": 0.001} +{"train_loss": 0.030390150845050812, "train_loss_bc": 0.030388180166482925, "train_loss_llm": 0.0001970529556274414, "grad_norm": 0.052599649876356125, "global_step": 26467, "epoch": 735, "lr": 0.001} +{"train_loss": 0.028798861429095268, "train_loss_bc": 0.028796305879950523, "train_loss_llm": 0.000255584716796875, "grad_norm": 0.06462706625461578, "global_step": 26468, "epoch": 735, "lr": 0.001} +{"train_loss": 0.030248604714870453, "train_loss_bc": 0.030248604714870453, "train_loss_llm": 0.0, "grad_norm": 0.0280731450766325, "global_step": 26469, "epoch": 735, "lr": 0.001} +{"train_loss": 0.030896490439772606, "train_loss_bc": 0.030896490439772606, "train_loss_llm": 0.0, "grad_norm": 0.039158206433057785, "global_step": 26470, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03129240870475769, "train_loss_bc": 0.03129107132554054, "train_loss_llm": 0.00013375282287597656, "grad_norm": 0.055752482265233994, "global_step": 26471, "epoch": 735, "lr": 0.001} +{"train_loss": 0.030793504789471626, "train_loss_bc": 0.030793504789471626, "train_loss_llm": 0.0, "grad_norm": 0.06334928423166275, "global_step": 26472, "epoch": 735, "lr": 0.001} +{"train_loss": 0.034363582730293274, "train_loss_bc": 0.034363582730293274, "train_loss_llm": 0.0, "grad_norm": 0.03302708640694618, "global_step": 26473, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03035755641758442, "train_loss_bc": 0.03034398891031742, "train_loss_llm": 0.0013568401336669922, "grad_norm": 0.054596997797489166, "global_step": 26474, "epoch": 735, "lr": 0.001} +{"train_loss": 0.031381554901599884, "train_loss_bc": 0.03137468919157982, "train_loss_llm": 0.0006866455078125, "grad_norm": 0.07461366057395935, "global_step": 26475, "epoch": 735, "lr": 0.001} +{"train_loss": 0.027572473511099815, "train_loss_bc": 0.027572473511099815, "train_loss_llm": 0.0, "grad_norm": 0.09206221252679825, "global_step": 26476, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03427467495203018, "train_loss_bc": 0.03427467495203018, "train_loss_llm": 0.0, "grad_norm": 0.037060029804706573, "global_step": 26477, "epoch": 735, "lr": 0.001} +{"train_loss": 0.030904008075594902, "train_loss_bc": 0.030904008075594902, "train_loss_llm": 0.0, "grad_norm": 0.06809498369693756, "global_step": 26478, "epoch": 735, "lr": 0.001} +{"train_loss": 0.02930869720876217, "train_loss_bc": 0.02930869720876217, "train_loss_llm": 0.0, "grad_norm": 0.0905875563621521, "global_step": 26479, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03325284644961357, "train_loss_bc": 0.03325284644961357, "train_loss_llm": 0.0, "grad_norm": 0.10729038715362549, "global_step": 26480, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03231348842382431, "train_loss_bc": 0.03231348842382431, "train_loss_llm": 0.0, "grad_norm": 0.04438970983028412, "global_step": 26481, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03125755116343498, "train_loss_bc": 0.03125755116343498, "train_loss_llm": 0.0, "grad_norm": 0.07170499861240387, "global_step": 26482, "epoch": 735, "lr": 0.001} +{"train_loss": 0.029224228113889694, "train_loss_bc": 0.029224054887890816, "train_loss_llm": 1.7315149307250977e-05, "grad_norm": 0.09224151074886322, "global_step": 26483, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03405836224555969, "train_loss_bc": 0.03405836224555969, "train_loss_llm": 0.0, "grad_norm": 0.110262930393219, "global_step": 26484, "epoch": 735, "lr": 0.001} +{"train_loss": 0.02914782427251339, "train_loss_bc": 0.02914782427251339, "train_loss_llm": 0.0, "grad_norm": 0.032829463481903076, "global_step": 26485, "epoch": 735, "lr": 0.001} +{"train_loss": 0.033136144280433655, "train_loss_bc": 0.033136144280433655, "train_loss_llm": 0.0, "grad_norm": 0.05733151733875275, "global_step": 26486, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03156331181526184, "train_loss_bc": 0.03153681755065918, "train_loss_llm": 0.0026495158672332764, "grad_norm": 0.07980196923017502, "global_step": 26487, "epoch": 735, "lr": 0.001} +{"train_loss": 0.026686888188123703, "train_loss_bc": 0.026686888188123703, "train_loss_llm": 0.0, "grad_norm": 0.09127090126276016, "global_step": 26488, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03442993387579918, "train_loss_bc": 0.03442993387579918, "train_loss_llm": 0.0, "grad_norm": 0.03687019646167755, "global_step": 26489, "epoch": 735, "lr": 0.001} +{"train_loss": 0.030947599560022354, "train_loss_bc": 0.030947599560022354, "train_loss_llm": 0.0, "grad_norm": 0.05628121271729469, "global_step": 26490, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03003781847655773, "train_loss_bc": 0.03003781847655773, "train_loss_llm": 0.0, "grad_norm": 0.07610534876585007, "global_step": 26491, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03249563276767731, "train_loss_bc": 0.03249563276767731, "train_loss_llm": 0.0, "grad_norm": 0.09342952072620392, "global_step": 26492, "epoch": 735, "lr": 0.001} +{"train_loss": 0.024235935881733894, "train_loss_bc": 0.02422390878200531, "train_loss_llm": 0.0012027919292449951, "grad_norm": 0.036290813237428665, "global_step": 26493, "epoch": 735, "lr": 0.001} +{"train_loss": 0.03166281804442406, "train_loss_bc": 0.03163396567106247, "train_loss_llm": 0.0028851330280303955, "grad_norm": 0.05058532580733299, "global_step": 26494, "epoch": 735, "lr": 0.001} +{"train_loss": 0.031051431368622515, "train_loss_bc": 0.03059377707540989, "train_loss_llm": 0.0011538565158843994, "grad_norm": 0.07122811675071716, "global_step": 26495, "epoch": 735, "lr": 0.001, "train/cumulative_reward": 72.47694453844976, "train/mean_score": 21.8378550982006, "train/success_rate": 0.6, "test/cumulative_reward": 37.53747133639673, "test/mean_score": 6.925945750876764, "test/success_rate": 0.14, "val_loss": 0.09383602440357208, "train_action_mse_error": 0.03448418900370598} +{"train_loss": 0.02567100338637829, "train_loss_bc": 0.02567100338637829, "train_loss_llm": 0.0, "grad_norm": 0.09459858387708664, "global_step": 26496, "epoch": 736, "lr": 0.001} +{"train_loss": 0.031175268813967705, "train_loss_bc": 0.031175268813967705, "train_loss_llm": 0.0, "grad_norm": 0.04482085257768631, "global_step": 26497, "epoch": 736, "lr": 0.001} +{"train_loss": 0.037379685789346695, "train_loss_bc": 0.037366461008787155, "train_loss_llm": 0.001322329044342041, "grad_norm": 0.07309078425168991, "global_step": 26498, "epoch": 736, "lr": 0.001} +{"train_loss": 0.031129246577620506, "train_loss_bc": 0.031129246577620506, "train_loss_llm": 0.0, "grad_norm": 0.0928678810596466, "global_step": 26499, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03232383728027344, "train_loss_bc": 0.03231344372034073, "train_loss_llm": 0.0010394454002380371, "grad_norm": 0.11009139567613602, "global_step": 26500, "epoch": 736, "lr": 0.001} +{"train_loss": 0.038297139108181, "train_loss_bc": 0.038297139108181, "train_loss_llm": 0.0, "grad_norm": 0.03487897664308548, "global_step": 26501, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03191518783569336, "train_loss_bc": 0.03191518783569336, "train_loss_llm": 0.0, "grad_norm": 0.05382699519395828, "global_step": 26502, "epoch": 736, "lr": 0.001} +{"train_loss": 0.034562837332487106, "train_loss_bc": 0.034562207758426666, "train_loss_llm": 6.303191184997559e-05, "grad_norm": 0.09443778544664383, "global_step": 26503, "epoch": 736, "lr": 0.001} +{"train_loss": 0.030938133597373962, "train_loss_bc": 0.030936401337385178, "train_loss_llm": 0.00017321109771728516, "grad_norm": 0.11595001071691513, "global_step": 26504, "epoch": 736, "lr": 0.001} +{"train_loss": 0.028819600120186806, "train_loss_bc": 0.028806278482079506, "train_loss_llm": 0.0013321340084075928, "grad_norm": 0.025804033502936363, "global_step": 26505, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03248472511768341, "train_loss_bc": 0.03248472511768341, "train_loss_llm": 0.0, "grad_norm": 0.050483450293540955, "global_step": 26506, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03199514001607895, "train_loss_bc": 0.0319807305932045, "train_loss_llm": 0.0014408230781555176, "grad_norm": 0.0701037347316742, "global_step": 26507, "epoch": 736, "lr": 0.001} +{"train_loss": 0.02928096242249012, "train_loss_bc": 0.02928096242249012, "train_loss_llm": 0.0, "grad_norm": 0.09371377527713776, "global_step": 26508, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03027396649122238, "train_loss_bc": 0.03027396649122238, "train_loss_llm": 0.0, "grad_norm": 0.03582340106368065, "global_step": 26509, "epoch": 736, "lr": 0.001} +{"train_loss": 0.035588588565588, "train_loss_bc": 0.03557794913649559, "train_loss_llm": 0.0010638535022735596, "grad_norm": 0.06133069097995758, "global_step": 26510, "epoch": 736, "lr": 0.001} +{"train_loss": 0.026720480993390083, "train_loss_bc": 0.026720480993390083, "train_loss_llm": 0.0, "grad_norm": 0.078239805996418, "global_step": 26511, "epoch": 736, "lr": 0.001} +{"train_loss": 0.033574990928173065, "train_loss_bc": 0.033574990928173065, "train_loss_llm": 0.0, "grad_norm": 0.09267327189445496, "global_step": 26512, "epoch": 736, "lr": 0.001} +{"train_loss": 0.031277112662792206, "train_loss_bc": 0.031277112662792206, "train_loss_llm": 0.0, "grad_norm": 0.034421537071466446, "global_step": 26513, "epoch": 736, "lr": 0.001} +{"train_loss": 0.027222013100981712, "train_loss_bc": 0.027222013100981712, "train_loss_llm": 0.0, "grad_norm": 0.046158041805028915, "global_step": 26514, "epoch": 736, "lr": 0.001} +{"train_loss": 0.029549505561590195, "train_loss_bc": 0.029549505561590195, "train_loss_llm": 0.0, "grad_norm": 0.05122236907482147, "global_step": 26515, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03001495823264122, "train_loss_bc": 0.03001495823264122, "train_loss_llm": 0.0, "grad_norm": 0.0605509988963604, "global_step": 26516, "epoch": 736, "lr": 0.001} +{"train_loss": 0.0282004252076149, "train_loss_bc": 0.0282004252076149, "train_loss_llm": 0.0, "grad_norm": 0.03701676428318024, "global_step": 26517, "epoch": 736, "lr": 0.001} +{"train_loss": 0.031072651967406273, "train_loss_bc": 0.03105979971587658, "train_loss_llm": 0.001285254955291748, "grad_norm": 0.061448123306035995, "global_step": 26518, "epoch": 736, "lr": 0.001} +{"train_loss": 0.0233688373118639, "train_loss_bc": 0.023351352661848068, "train_loss_llm": 0.0017485320568084717, "grad_norm": 0.07452591508626938, "global_step": 26519, "epoch": 736, "lr": 0.001} +{"train_loss": 0.030878810212016106, "train_loss_bc": 0.030878810212016106, "train_loss_llm": 0.0, "grad_norm": 0.08491212874650955, "global_step": 26520, "epoch": 736, "lr": 0.001} +{"train_loss": 0.026950068771839142, "train_loss_bc": 0.026931971311569214, "train_loss_llm": 0.0018097460269927979, "grad_norm": 0.03045671060681343, "global_step": 26521, "epoch": 736, "lr": 0.001} +{"train_loss": 0.030783390626311302, "train_loss_bc": 0.030783390626311302, "train_loss_llm": 0.0, "grad_norm": 0.05838838964700699, "global_step": 26522, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03176591172814369, "train_loss_bc": 0.031756650656461716, "train_loss_llm": 0.000926285982131958, "grad_norm": 0.07084188610315323, "global_step": 26523, "epoch": 736, "lr": 0.001} +{"train_loss": 0.0350252166390419, "train_loss_bc": 0.0350252166390419, "train_loss_llm": 0.0, "grad_norm": 0.08780224621295929, "global_step": 26524, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03608547896146774, "train_loss_bc": 0.03608547896146774, "train_loss_llm": 0.0, "grad_norm": 0.036241017282009125, "global_step": 26525, "epoch": 736, "lr": 0.001} +{"train_loss": 0.028871815651655197, "train_loss_bc": 0.028871815651655197, "train_loss_llm": 0.0, "grad_norm": 0.05693080648779869, "global_step": 26526, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03071569837629795, "train_loss_bc": 0.03071569837629795, "train_loss_llm": 0.0, "grad_norm": 0.07246267050504684, "global_step": 26527, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03351765125989914, "train_loss_bc": 0.03351765125989914, "train_loss_llm": 0.0, "grad_norm": 0.08711793273687363, "global_step": 26528, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03009221889078617, "train_loss_bc": 0.03009221889078617, "train_loss_llm": 0.0, "grad_norm": 0.035650476813316345, "global_step": 26529, "epoch": 736, "lr": 0.001} +{"train_loss": 0.032065216451883316, "train_loss_bc": 0.03206144645810127, "train_loss_llm": 0.0003769099712371826, "grad_norm": 0.050788842141628265, "global_step": 26530, "epoch": 736, "lr": 0.001} +{"train_loss": 0.03112655847022931, "train_loss_bc": 0.030968328937888145, "train_loss_llm": 0.0, "grad_norm": 0.07260710000991821, "global_step": 26531, "epoch": 736, "lr": 0.001, "val_loss": 0.09175266325473785} +{"train_loss": 0.03012932650744915, "train_loss_bc": 0.03012932650744915, "train_loss_llm": 0.0, "grad_norm": 0.08606596291065216, "global_step": 26532, "epoch": 737, "lr": 0.001} +{"train_loss": 0.032906532287597656, "train_loss_bc": 0.032906532287597656, "train_loss_llm": 0.0, "grad_norm": 0.03306934982538223, "global_step": 26533, "epoch": 737, "lr": 0.001} +{"train_loss": 0.033094193786382675, "train_loss_bc": 0.033071670681238174, "train_loss_llm": 0.002252250909805298, "grad_norm": 0.04668246954679489, "global_step": 26534, "epoch": 737, "lr": 0.001} +{"train_loss": 0.026507830247282982, "train_loss_bc": 0.026507830247282982, "train_loss_llm": 0.0, "grad_norm": 0.05116735398769379, "global_step": 26535, "epoch": 737, "lr": 0.001} +{"train_loss": 0.0324290469288826, "train_loss_bc": 0.0324290469288826, "train_loss_llm": 0.0, "grad_norm": 0.07544989883899689, "global_step": 26536, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03618769720196724, "train_loss_bc": 0.03618769720196724, "train_loss_llm": 0.0, "grad_norm": 0.03623485937714577, "global_step": 26537, "epoch": 737, "lr": 0.001} +{"train_loss": 0.02923322655260563, "train_loss_bc": 0.02923322655260563, "train_loss_llm": 0.0, "grad_norm": 0.0536087304353714, "global_step": 26538, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03281141072511673, "train_loss_bc": 0.03277623653411865, "train_loss_llm": 0.003517240285873413, "grad_norm": 0.06898299604654312, "global_step": 26539, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03337840735912323, "train_loss_bc": 0.03337840735912323, "train_loss_llm": 0.0, "grad_norm": 0.07857711613178253, "global_step": 26540, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03229064494371414, "train_loss_bc": 0.03227458894252777, "train_loss_llm": 0.001605689525604248, "grad_norm": 0.03183586150407791, "global_step": 26541, "epoch": 737, "lr": 0.001} +{"train_loss": 0.02911144681274891, "train_loss_bc": 0.02911144681274891, "train_loss_llm": 0.0, "grad_norm": 0.05159098654985428, "global_step": 26542, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03179004415869713, "train_loss_bc": 0.03179004415869713, "train_loss_llm": 0.0, "grad_norm": 0.07512716203927994, "global_step": 26543, "epoch": 737, "lr": 0.001} +{"train_loss": 0.026077983900904655, "train_loss_bc": 0.026077983900904655, "train_loss_llm": 0.0, "grad_norm": 0.09039545059204102, "global_step": 26544, "epoch": 737, "lr": 0.001} +{"train_loss": 0.027982931584119797, "train_loss_bc": 0.027982931584119797, "train_loss_llm": 0.0, "grad_norm": 0.03639758750796318, "global_step": 26545, "epoch": 737, "lr": 0.001} +{"train_loss": 0.028400328010320663, "train_loss_bc": 0.028400328010320663, "train_loss_llm": 0.0, "grad_norm": 0.061646293848752975, "global_step": 26546, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03069823421537876, "train_loss_bc": 0.03069823421537876, "train_loss_llm": 0.0, "grad_norm": 0.07880671322345734, "global_step": 26547, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03151378780603409, "train_loss_bc": 0.031497396528720856, "train_loss_llm": 0.0016390681266784668, "grad_norm": 0.09435367584228516, "global_step": 26548, "epoch": 737, "lr": 0.001} +{"train_loss": 0.031358592212200165, "train_loss_bc": 0.031358592212200165, "train_loss_llm": 0.0, "grad_norm": 0.03204956650733948, "global_step": 26549, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03685959428548813, "train_loss_bc": 0.03685374930500984, "train_loss_llm": 0.0005844533443450928, "grad_norm": 0.05472062528133392, "global_step": 26550, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03117753565311432, "train_loss_bc": 0.031158804893493652, "train_loss_llm": 0.0018731653690338135, "grad_norm": 0.0653589516878128, "global_step": 26551, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03387711942195892, "train_loss_bc": 0.03387711942195892, "train_loss_llm": 0.0, "grad_norm": 0.08718432486057281, "global_step": 26552, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03530272841453552, "train_loss_bc": 0.03530272841453552, "train_loss_llm": 0.0, "grad_norm": 0.03110629692673683, "global_step": 26553, "epoch": 737, "lr": 0.001} +{"train_loss": 0.02913893572986126, "train_loss_bc": 0.02912646345794201, "train_loss_llm": 0.0012471377849578857, "grad_norm": 0.0389985665678978, "global_step": 26554, "epoch": 737, "lr": 0.001} +{"train_loss": 0.028388092294335365, "train_loss_bc": 0.02837500348687172, "train_loss_llm": 0.0013089478015899658, "grad_norm": 0.054791975766420364, "global_step": 26555, "epoch": 737, "lr": 0.001} +{"train_loss": 0.02234284207224846, "train_loss_bc": 0.02234284207224846, "train_loss_llm": 0.0, "grad_norm": 0.06955075263977051, "global_step": 26556, "epoch": 737, "lr": 0.001} +{"train_loss": 0.030046798288822174, "train_loss_bc": 0.030046798288822174, "train_loss_llm": 0.0, "grad_norm": 0.030562175437808037, "global_step": 26557, "epoch": 737, "lr": 0.001} +{"train_loss": 0.031176256015896797, "train_loss_bc": 0.031176256015896797, "train_loss_llm": 0.0, "grad_norm": 0.04647737368941307, "global_step": 26558, "epoch": 737, "lr": 0.001} +{"train_loss": 0.02800864912569523, "train_loss_bc": 0.02800864912569523, "train_loss_llm": 0.0, "grad_norm": 0.06271931529045105, "global_step": 26559, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03066754899919033, "train_loss_bc": 0.03066754899919033, "train_loss_llm": 0.0, "grad_norm": 0.08243066072463989, "global_step": 26560, "epoch": 737, "lr": 0.001} +{"train_loss": 0.027605343610048294, "train_loss_bc": 0.027605343610048294, "train_loss_llm": 0.0, "grad_norm": 0.032506950199604034, "global_step": 26561, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03252275288105011, "train_loss_bc": 0.03251883387565613, "train_loss_llm": 0.00039207935333251953, "grad_norm": 0.059945184737443924, "global_step": 26562, "epoch": 737, "lr": 0.001} +{"train_loss": 0.036931104958057404, "train_loss_bc": 0.036931104958057404, "train_loss_llm": 0.0, "grad_norm": 0.0798073559999466, "global_step": 26563, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03286480903625488, "train_loss_bc": 0.03286480903625488, "train_loss_llm": 0.0, "grad_norm": 0.09798174351453781, "global_step": 26564, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03202714025974274, "train_loss_bc": 0.03202714025974274, "train_loss_llm": 0.0, "grad_norm": 0.03041881136596203, "global_step": 26565, "epoch": 737, "lr": 0.001} +{"train_loss": 0.032180823385715485, "train_loss_bc": 0.032180823385715485, "train_loss_llm": 0.0, "grad_norm": 0.047980234026908875, "global_step": 26566, "epoch": 737, "lr": 0.001} +{"train_loss": 0.03089361534350448, "train_loss_bc": 0.025150412693619728, "train_loss_llm": 0.0, "grad_norm": 0.07618006318807602, "global_step": 26567, "epoch": 737, "lr": 0.001, "val_loss": 0.09118807315826416} +{"train_loss": 0.028156280517578125, "train_loss_bc": 0.028156280517578125, "train_loss_llm": 0.0, "grad_norm": 0.09140490740537643, "global_step": 26568, "epoch": 738, "lr": 0.001} +{"train_loss": 0.033239226788282394, "train_loss_bc": 0.033239226788282394, "train_loss_llm": 0.0, "grad_norm": 0.03137945756316185, "global_step": 26569, "epoch": 738, "lr": 0.001} +{"train_loss": 0.03485756739974022, "train_loss_bc": 0.03485756739974022, "train_loss_llm": 0.0, "grad_norm": 0.047117751091718674, "global_step": 26570, "epoch": 738, "lr": 0.001} +{"train_loss": 0.03728834167122841, "train_loss_bc": 0.03728834167122841, "train_loss_llm": 0.0, "grad_norm": 0.06510796397924423, "global_step": 26571, "epoch": 738, "lr": 0.001} +{"train_loss": 0.027538113296031952, "train_loss_bc": 0.027530044317245483, "train_loss_llm": 0.000806957483291626, "grad_norm": 0.07915522903203964, "global_step": 26572, "epoch": 738, "lr": 0.001} +{"train_loss": 0.034832097589969635, "train_loss_bc": 0.034832097589969635, "train_loss_llm": 0.0, "grad_norm": 0.03656558692455292, "global_step": 26573, "epoch": 738, "lr": 0.001} +{"train_loss": 0.029749467968940735, "train_loss_bc": 0.029749467968940735, "train_loss_llm": 0.0, "grad_norm": 0.04621552675962448, "global_step": 26574, "epoch": 738, "lr": 0.001} +{"train_loss": 0.023641513660550117, "train_loss_bc": 0.023640846833586693, "train_loss_llm": 6.660819053649902e-05, "grad_norm": 0.06013811007142067, "global_step": 26575, "epoch": 738, "lr": 0.001} +{"train_loss": 0.026871107518672943, "train_loss_bc": 0.026871107518672943, "train_loss_llm": 0.0, "grad_norm": 0.07689189910888672, "global_step": 26576, "epoch": 738, "lr": 0.001} +{"train_loss": 0.028704319149255753, "train_loss_bc": 0.028704319149255753, "train_loss_llm": 0.0, "grad_norm": 0.03220772743225098, "global_step": 26577, "epoch": 738, "lr": 0.001} +{"train_loss": 0.034768808633089066, "train_loss_bc": 0.034768808633089066, "train_loss_llm": 0.0, "grad_norm": 0.06403359770774841, "global_step": 26578, "epoch": 738, "lr": 0.001} +{"train_loss": 0.03202901780605316, "train_loss_bc": 0.03202901780605316, "train_loss_llm": 0.0, "grad_norm": 0.0713980421423912, "global_step": 26579, "epoch": 738, "lr": 0.001} +{"train_loss": 0.029654737561941147, "train_loss_bc": 0.029649948701262474, "train_loss_llm": 0.000478893518447876, "grad_norm": 0.07333643734455109, "global_step": 26580, "epoch": 738, "lr": 0.001} +{"train_loss": 0.03175227716565132, "train_loss_bc": 0.03175227716565132, "train_loss_llm": 0.0, "grad_norm": 0.03620561957359314, "global_step": 26581, "epoch": 738, "lr": 0.001} +{"train_loss": 0.032128773629665375, "train_loss_bc": 0.032128773629665375, "train_loss_llm": 0.0, "grad_norm": 0.056171715259552, "global_step": 26582, "epoch": 738, "lr": 0.001} +{"train_loss": 0.03196420520544052, "train_loss_bc": 0.03196420520544052, "train_loss_llm": 0.0, "grad_norm": 0.07680772989988327, "global_step": 26583, "epoch": 738, "lr": 0.001} +{"train_loss": 0.019230833277106285, "train_loss_bc": 0.019229033961892128, "train_loss_llm": 0.0001798570156097412, "grad_norm": 0.09944218397140503, "global_step": 26584, "epoch": 738, "lr": 0.001} +{"train_loss": 0.030438929796218872, "train_loss_bc": 0.030438929796218872, "train_loss_llm": 0.0, "grad_norm": 0.03494488075375557, "global_step": 26585, "epoch": 738, "lr": 0.001} +{"train_loss": 0.0346103236079216, "train_loss_bc": 0.0346103236079216, "train_loss_llm": 0.0, "grad_norm": 0.046686507761478424, "global_step": 26586, "epoch": 738, "lr": 0.001} +{"train_loss": 0.03985258936882019, "train_loss_bc": 0.03985258936882019, "train_loss_llm": 0.0, "grad_norm": 0.06820368766784668, "global_step": 26587, "epoch": 738, "lr": 0.001} +{"train_loss": 0.025364089757204056, "train_loss_bc": 0.025364089757204056, "train_loss_llm": 0.0, "grad_norm": 0.0849660187959671, "global_step": 26588, "epoch": 738, "lr": 0.001} +{"train_loss": 0.024472398683428764, "train_loss_bc": 0.024472398683428764, "train_loss_llm": 0.0, "grad_norm": 0.03201918303966522, "global_step": 26589, "epoch": 738, "lr": 0.001} +{"train_loss": 0.028126269578933716, "train_loss_bc": 0.02811894193291664, "train_loss_llm": 0.0007327497005462646, "grad_norm": 0.04879433661699295, "global_step": 26590, "epoch": 738, "lr": 0.001} +{"train_loss": 0.030502229928970337, "train_loss_bc": 0.030502229928970337, "train_loss_llm": 0.0, "grad_norm": 0.05998311564326286, "global_step": 26591, "epoch": 738, "lr": 0.001} +{"train_loss": 0.03173433989286423, "train_loss_bc": 0.03173433989286423, "train_loss_llm": 0.0, "grad_norm": 0.07378628104925156, "global_step": 26592, "epoch": 738, "lr": 0.001} +{"train_loss": 0.03385554999113083, "train_loss_bc": 0.03385305404663086, "train_loss_llm": 0.00024953484535217285, "grad_norm": 0.03718441724777222, "global_step": 26593, "epoch": 738, "lr": 0.001} +{"train_loss": 0.035176001489162445, "train_loss_bc": 0.035168010741472244, "train_loss_llm": 0.0007989704608917236, "grad_norm": 0.056127697229385376, "global_step": 26594, "epoch": 738, "lr": 0.001} +{"train_loss": 0.024922246113419533, "train_loss_bc": 0.024922246113419533, "train_loss_llm": 0.0, "grad_norm": 0.06613440811634064, "global_step": 26595, "epoch": 738, "lr": 0.001} +{"train_loss": 0.02583431825041771, "train_loss_bc": 0.025822483003139496, "train_loss_llm": 0.0011835694313049316, "grad_norm": 0.0803612619638443, "global_step": 26596, "epoch": 738, "lr": 0.001} +{"train_loss": 0.041263189166784286, "train_loss_bc": 0.041263189166784286, "train_loss_llm": 0.0, "grad_norm": 0.03873550519347191, "global_step": 26597, "epoch": 738, "lr": 0.001} +{"train_loss": 0.03075166791677475, "train_loss_bc": 0.03075166791677475, "train_loss_llm": 0.0, "grad_norm": 0.06201409921050072, "global_step": 26598, "epoch": 738, "lr": 0.001} +{"train_loss": 0.031692102551460266, "train_loss_bc": 0.031678177416324615, "train_loss_llm": 0.0013923943042755127, "grad_norm": 0.08097716420888901, "global_step": 26599, "epoch": 738, "lr": 0.001} +{"train_loss": 0.0280108954757452, "train_loss_bc": 0.0280108954757452, "train_loss_llm": 0.0, "grad_norm": 0.09691139310598373, "global_step": 26600, "epoch": 738, "lr": 0.001} +{"train_loss": 0.03621174395084381, "train_loss_bc": 0.03621174395084381, "train_loss_llm": 0.0, "grad_norm": 0.032457415014505386, "global_step": 26601, "epoch": 738, "lr": 0.001} +{"train_loss": 0.028273209929466248, "train_loss_bc": 0.028265362605452538, "train_loss_llm": 0.0007847249507904053, "grad_norm": 0.0479435995221138, "global_step": 26602, "epoch": 738, "lr": 0.001} +{"train_loss": 0.030965175945311785, "train_loss_bc": 0.03724754974246025, "train_loss_llm": 0.0, "grad_norm": 0.0693812295794487, "global_step": 26603, "epoch": 738, "lr": 0.001, "val_loss": 0.0920819565653801} +{"train_loss": 0.026556653901934624, "train_loss_bc": 0.026556214317679405, "train_loss_llm": 4.3958425521850586e-05, "grad_norm": 0.08475641906261444, "global_step": 26604, "epoch": 739, "lr": 0.001} +{"train_loss": 0.030814973637461662, "train_loss_bc": 0.030814973637461662, "train_loss_llm": 0.0, "grad_norm": 0.038556743413209915, "global_step": 26605, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03491625562310219, "train_loss_bc": 0.034907061606645584, "train_loss_llm": 0.0009195804595947266, "grad_norm": 0.0679023265838623, "global_step": 26606, "epoch": 739, "lr": 0.001} +{"train_loss": 0.026005961000919342, "train_loss_bc": 0.026005961000919342, "train_loss_llm": 0.0, "grad_norm": 0.07432959973812103, "global_step": 26607, "epoch": 739, "lr": 0.001} +{"train_loss": 0.027117881923913956, "train_loss_bc": 0.027112843468785286, "train_loss_llm": 0.000503838062286377, "grad_norm": 0.08936106413602829, "global_step": 26608, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03127562999725342, "train_loss_bc": 0.03125687688589096, "train_loss_llm": 0.0018754303455352783, "grad_norm": 0.04130249097943306, "global_step": 26609, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03593282774090767, "train_loss_bc": 0.03593282774090767, "train_loss_llm": 0.0, "grad_norm": 0.07522594928741455, "global_step": 26610, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03274056315422058, "train_loss_bc": 0.03274056315422058, "train_loss_llm": 0.0, "grad_norm": 0.09781581163406372, "global_step": 26611, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03346821293234825, "train_loss_bc": 0.03346821293234825, "train_loss_llm": 0.0, "grad_norm": 0.12654145061969757, "global_step": 26612, "epoch": 739, "lr": 0.001} +{"train_loss": 0.032167959958314896, "train_loss_bc": 0.032167959958314896, "train_loss_llm": 0.0, "grad_norm": 0.04461252689361572, "global_step": 26613, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03084130771458149, "train_loss_bc": 0.03084130771458149, "train_loss_llm": 0.0, "grad_norm": 0.05553577095270157, "global_step": 26614, "epoch": 739, "lr": 0.001} +{"train_loss": 0.029329897835850716, "train_loss_bc": 0.029314326122403145, "train_loss_llm": 0.0015571415424346924, "grad_norm": 0.060354601591825485, "global_step": 26615, "epoch": 739, "lr": 0.001} +{"train_loss": 0.033014871180057526, "train_loss_bc": 0.033014871180057526, "train_loss_llm": 0.0, "grad_norm": 0.07640552520751953, "global_step": 26616, "epoch": 739, "lr": 0.001} +{"train_loss": 0.0333896167576313, "train_loss_bc": 0.0333896167576313, "train_loss_llm": 0.0, "grad_norm": 0.035113897174596786, "global_step": 26617, "epoch": 739, "lr": 0.001} +{"train_loss": 0.030832339078187943, "train_loss_bc": 0.030832339078187943, "train_loss_llm": 0.0, "grad_norm": 0.05834909528493881, "global_step": 26618, "epoch": 739, "lr": 0.001} +{"train_loss": 0.031000377610325813, "train_loss_bc": 0.031000377610325813, "train_loss_llm": 0.0, "grad_norm": 0.07188215851783752, "global_step": 26619, "epoch": 739, "lr": 0.001} +{"train_loss": 0.034978099167346954, "train_loss_bc": 0.034978099167346954, "train_loss_llm": 0.0, "grad_norm": 0.08539997041225433, "global_step": 26620, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03183393552899361, "train_loss_bc": 0.03182903677225113, "train_loss_llm": 0.0004897117614746094, "grad_norm": 0.035135917365550995, "global_step": 26621, "epoch": 739, "lr": 0.001} +{"train_loss": 0.027120130136609077, "train_loss_bc": 0.027120130136609077, "train_loss_llm": 0.0, "grad_norm": 0.055601511150598526, "global_step": 26622, "epoch": 739, "lr": 0.001} +{"train_loss": 0.026743099093437195, "train_loss_bc": 0.026743099093437195, "train_loss_llm": 0.0, "grad_norm": 0.0735090896487236, "global_step": 26623, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03196363523602486, "train_loss_bc": 0.031955476850271225, "train_loss_llm": 0.0008158385753631592, "grad_norm": 0.09908340126276016, "global_step": 26624, "epoch": 739, "lr": 0.001} +{"train_loss": 0.029897533357143402, "train_loss_bc": 0.029895871877670288, "train_loss_llm": 0.00016611814498901367, "grad_norm": 0.03896337375044823, "global_step": 26625, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03288893774151802, "train_loss_bc": 0.03288893774151802, "train_loss_llm": 0.0, "grad_norm": 0.058632224798202515, "global_step": 26626, "epoch": 739, "lr": 0.001} +{"train_loss": 0.030977612361311913, "train_loss_bc": 0.030977612361311913, "train_loss_llm": 0.0, "grad_norm": 0.08186336606740952, "global_step": 26627, "epoch": 739, "lr": 0.001} +{"train_loss": 0.032145362347364426, "train_loss_bc": 0.03214501217007637, "train_loss_llm": 3.5196542739868164e-05, "grad_norm": 0.1092984527349472, "global_step": 26628, "epoch": 739, "lr": 0.001} +{"train_loss": 0.02968864142894745, "train_loss_bc": 0.02968864142894745, "train_loss_llm": 0.0, "grad_norm": 0.03173905983567238, "global_step": 26629, "epoch": 739, "lr": 0.001} +{"train_loss": 0.033906325697898865, "train_loss_bc": 0.033906325697898865, "train_loss_llm": 0.0, "grad_norm": 0.060663577169179916, "global_step": 26630, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03408881649374962, "train_loss_bc": 0.03408881649374962, "train_loss_llm": 0.0, "grad_norm": 0.07631947100162506, "global_step": 26631, "epoch": 739, "lr": 0.001} +{"train_loss": 0.024630462750792503, "train_loss_bc": 0.024630462750792503, "train_loss_llm": 0.0, "grad_norm": 0.09245790541172028, "global_step": 26632, "epoch": 739, "lr": 0.001} +{"train_loss": 0.02795034646987915, "train_loss_bc": 0.027946515008807182, "train_loss_llm": 0.000383228063583374, "grad_norm": 0.039830077439546585, "global_step": 26633, "epoch": 739, "lr": 0.001} +{"train_loss": 0.026316283270716667, "train_loss_bc": 0.02630283497273922, "train_loss_llm": 0.0013447403907775879, "grad_norm": 0.0555109865963459, "global_step": 26634, "epoch": 739, "lr": 0.001} +{"train_loss": 0.028351057320833206, "train_loss_bc": 0.028351057320833206, "train_loss_llm": 0.0, "grad_norm": 0.074038065969944, "global_step": 26635, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03351669758558273, "train_loss_bc": 0.03351669758558273, "train_loss_llm": 0.0, "grad_norm": 0.09068908542394638, "global_step": 26636, "epoch": 739, "lr": 0.001} +{"train_loss": 0.03926726430654526, "train_loss_bc": 0.03926726430654526, "train_loss_llm": 0.0, "grad_norm": 0.030122796073555946, "global_step": 26637, "epoch": 739, "lr": 0.001} +{"train_loss": 0.028799105435609818, "train_loss_bc": 0.028788205236196518, "train_loss_llm": 0.0010900795459747314, "grad_norm": 0.04529786482453346, "global_step": 26638, "epoch": 739, "lr": 0.001} +{"train_loss": 0.030986140119946666, "train_loss_bc": 0.031032368540763855, "train_loss_llm": 0.0, "grad_norm": 0.06645233929157257, "global_step": 26639, "epoch": 739, "lr": 0.001, "val_loss": 0.09273060411214828} +{"train_loss": 0.02646806836128235, "train_loss_bc": 0.02646806836128235, "train_loss_llm": 0.0, "grad_norm": 0.07852985709905624, "global_step": 26640, "epoch": 740, "lr": 0.001} +{"train_loss": 0.022833645343780518, "train_loss_bc": 0.02283029817044735, "train_loss_llm": 0.00033479928970336914, "grad_norm": 0.028638286516070366, "global_step": 26641, "epoch": 740, "lr": 0.001} +{"train_loss": 0.031418755650520325, "train_loss_bc": 0.031418755650520325, "train_loss_llm": 0.0, "grad_norm": 0.05255155637860298, "global_step": 26642, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03362766280770302, "train_loss_bc": 0.03362766280770302, "train_loss_llm": 0.0, "grad_norm": 0.07198397070169449, "global_step": 26643, "epoch": 740, "lr": 0.001} +{"train_loss": 0.036080799996852875, "train_loss_bc": 0.036080799996852875, "train_loss_llm": 0.0, "grad_norm": 0.0906418114900589, "global_step": 26644, "epoch": 740, "lr": 0.001} +{"train_loss": 0.027374062687158585, "train_loss_bc": 0.02735207974910736, "train_loss_llm": 0.0021983683109283447, "grad_norm": 0.029419301077723503, "global_step": 26645, "epoch": 740, "lr": 0.001} +{"train_loss": 0.026160279288887978, "train_loss_bc": 0.026158835738897324, "train_loss_llm": 0.0001443326473236084, "grad_norm": 0.04775413125753403, "global_step": 26646, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03698960319161415, "train_loss_bc": 0.03698960319161415, "train_loss_llm": 0.0, "grad_norm": 0.07908005267381668, "global_step": 26647, "epoch": 740, "lr": 0.001} +{"train_loss": 0.024980368092656136, "train_loss_bc": 0.024976644665002823, "train_loss_llm": 0.00037229061126708984, "grad_norm": 0.10646037757396698, "global_step": 26648, "epoch": 740, "lr": 0.001} +{"train_loss": 0.030979735776782036, "train_loss_bc": 0.030979735776782036, "train_loss_llm": 0.0, "grad_norm": 0.02704283595085144, "global_step": 26649, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03205518424510956, "train_loss_bc": 0.03203558176755905, "train_loss_llm": 0.00196036696434021, "grad_norm": 0.04148244485259056, "global_step": 26650, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03490089625120163, "train_loss_bc": 0.03490089625120163, "train_loss_llm": 0.0, "grad_norm": 0.06984370946884155, "global_step": 26651, "epoch": 740, "lr": 0.001} +{"train_loss": 0.032326970249414444, "train_loss_bc": 0.032326970249414444, "train_loss_llm": 0.0, "grad_norm": 0.0918310210108757, "global_step": 26652, "epoch": 740, "lr": 0.001} +{"train_loss": 0.025776531547307968, "train_loss_bc": 0.025776531547307968, "train_loss_llm": 0.0, "grad_norm": 0.0343555249273777, "global_step": 26653, "epoch": 740, "lr": 0.001} +{"train_loss": 0.02766190841794014, "train_loss_bc": 0.02766190841794014, "train_loss_llm": 0.0, "grad_norm": 0.06351009011268616, "global_step": 26654, "epoch": 740, "lr": 0.001} +{"train_loss": 0.037694528698921204, "train_loss_bc": 0.037694528698921204, "train_loss_llm": 0.0, "grad_norm": 0.09777675569057465, "global_step": 26655, "epoch": 740, "lr": 0.001} +{"train_loss": 0.031113801524043083, "train_loss_bc": 0.031113801524043083, "train_loss_llm": 0.0, "grad_norm": 0.11425568163394928, "global_step": 26656, "epoch": 740, "lr": 0.001} +{"train_loss": 0.02928692288696766, "train_loss_bc": 0.02928692288696766, "train_loss_llm": 0.0, "grad_norm": 0.03972557187080383, "global_step": 26657, "epoch": 740, "lr": 0.001} +{"train_loss": 0.030549127608537674, "train_loss_bc": 0.030549127608537674, "train_loss_llm": 0.0, "grad_norm": 0.05405652895569801, "global_step": 26658, "epoch": 740, "lr": 0.001} +{"train_loss": 0.029172208160161972, "train_loss_bc": 0.029172208160161972, "train_loss_llm": 0.0, "grad_norm": 0.06867478042840958, "global_step": 26659, "epoch": 740, "lr": 0.001} +{"train_loss": 0.0268084853887558, "train_loss_bc": 0.026805048808455467, "train_loss_llm": 0.00034368038177490234, "grad_norm": 0.07877928763628006, "global_step": 26660, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03223247453570366, "train_loss_bc": 0.03223247453570366, "train_loss_llm": 0.0, "grad_norm": 0.04070586338639259, "global_step": 26661, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03792605549097061, "train_loss_bc": 0.03792605549097061, "train_loss_llm": 0.0, "grad_norm": 0.06723730266094208, "global_step": 26662, "epoch": 740, "lr": 0.001} +{"train_loss": 0.032245077192783356, "train_loss_bc": 0.032245077192783356, "train_loss_llm": 0.0, "grad_norm": 0.09646132588386536, "global_step": 26663, "epoch": 740, "lr": 0.001} +{"train_loss": 0.024652725085616112, "train_loss_bc": 0.024636147543787956, "train_loss_llm": 0.0016576945781707764, "grad_norm": 0.1092139482498169, "global_step": 26664, "epoch": 740, "lr": 0.001} +{"train_loss": 0.032411105930805206, "train_loss_bc": 0.03239404410123825, "train_loss_llm": 0.0017060041427612305, "grad_norm": 0.03830959275364876, "global_step": 26665, "epoch": 740, "lr": 0.001} +{"train_loss": 0.030629035085439682, "train_loss_bc": 0.030629035085439682, "train_loss_llm": 0.0, "grad_norm": 0.05505962669849396, "global_step": 26666, "epoch": 740, "lr": 0.001} +{"train_loss": 0.035744789987802505, "train_loss_bc": 0.035744789987802505, "train_loss_llm": 0.0, "grad_norm": 0.07999545335769653, "global_step": 26667, "epoch": 740, "lr": 0.001} +{"train_loss": 0.037119556218385696, "train_loss_bc": 0.03711128607392311, "train_loss_llm": 0.000827103853225708, "grad_norm": 0.10959437489509583, "global_step": 26668, "epoch": 740, "lr": 0.001} +{"train_loss": 0.028566686436533928, "train_loss_bc": 0.028566686436533928, "train_loss_llm": 0.0, "grad_norm": 0.032108109444379807, "global_step": 26669, "epoch": 740, "lr": 0.001} +{"train_loss": 0.032203178852796555, "train_loss_bc": 0.032203178852796555, "train_loss_llm": 0.0, "grad_norm": 0.05690371245145798, "global_step": 26670, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03091580979526043, "train_loss_bc": 0.03091580979526043, "train_loss_llm": 0.0, "grad_norm": 0.0735412985086441, "global_step": 26671, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03379397466778755, "train_loss_bc": 0.03379397466778755, "train_loss_llm": 0.0, "grad_norm": 0.10415177047252655, "global_step": 26672, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03310049697756767, "train_loss_bc": 0.03310049697756767, "train_loss_llm": 0.0, "grad_norm": 0.03769103065133095, "global_step": 26673, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03140128776431084, "train_loss_bc": 0.03140128776431084, "train_loss_llm": 0.0, "grad_norm": 0.054777536541223526, "global_step": 26674, "epoch": 740, "lr": 0.001} +{"train_loss": 0.03101580274394817, "train_loss_bc": 0.029367098584771156, "train_loss_llm": 0.0, "grad_norm": 0.06958582997322083, "global_step": 26675, "epoch": 740, "lr": 0.001, "train/cumulative_reward": 44.87150676202208, "train/mean_score": 13.784906795015454, "train/success_rate": 0.4, "test/cumulative_reward": 41.2956037623521, "test/mean_score": 8.011471639663432, "test/success_rate": 0.2, "val_loss": 0.09046322107315063, "train_action_mse_error": 0.033513374626636505} +{"train_loss": 0.025959108024835587, "train_loss_bc": 0.025959108024835587, "train_loss_llm": 0.0, "grad_norm": 0.0811699852347374, "global_step": 26676, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03149920701980591, "train_loss_bc": 0.03149920701980591, "train_loss_llm": 0.0, "grad_norm": 0.03850304335355759, "global_step": 26677, "epoch": 741, "lr": 0.001} +{"train_loss": 0.02860289439558983, "train_loss_bc": 0.02860289439558983, "train_loss_llm": 0.0, "grad_norm": 0.05021965131163597, "global_step": 26678, "epoch": 741, "lr": 0.001} +{"train_loss": 0.032185714691877365, "train_loss_bc": 0.03218002989888191, "train_loss_llm": 0.0005685091018676758, "grad_norm": 0.07816888391971588, "global_step": 26679, "epoch": 741, "lr": 0.001} +{"train_loss": 0.02658589370548725, "train_loss_bc": 0.02658589370548725, "train_loss_llm": 0.0, "grad_norm": 0.09718134999275208, "global_step": 26680, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03707952797412872, "train_loss_bc": 0.03707952797412872, "train_loss_llm": 0.0, "grad_norm": 0.0362553633749485, "global_step": 26681, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03200080618262291, "train_loss_bc": 0.03199697658419609, "train_loss_llm": 0.0003829002380371094, "grad_norm": 0.057702165096998215, "global_step": 26682, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03249821811914444, "train_loss_bc": 0.03249821811914444, "train_loss_llm": 0.0, "grad_norm": 0.07214212417602539, "global_step": 26683, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03703797236084938, "train_loss_bc": 0.03703797236084938, "train_loss_llm": 0.0, "grad_norm": 0.09912281483411789, "global_step": 26684, "epoch": 741, "lr": 0.001} +{"train_loss": 0.027661046013236046, "train_loss_bc": 0.027636194601655006, "train_loss_llm": 0.002485126256942749, "grad_norm": 0.02538236230611801, "global_step": 26685, "epoch": 741, "lr": 0.001} +{"train_loss": 0.034943435341119766, "train_loss_bc": 0.034939393401145935, "train_loss_llm": 0.0004042983055114746, "grad_norm": 0.047229964286088943, "global_step": 26686, "epoch": 741, "lr": 0.001} +{"train_loss": 0.029094766825437546, "train_loss_bc": 0.029094766825437546, "train_loss_llm": 0.0, "grad_norm": 0.06566350162029266, "global_step": 26687, "epoch": 741, "lr": 0.001} +{"train_loss": 0.031545430421829224, "train_loss_bc": 0.03153814375400543, "train_loss_llm": 0.0007284879684448242, "grad_norm": 0.09069240838289261, "global_step": 26688, "epoch": 741, "lr": 0.001} +{"train_loss": 0.02975619025528431, "train_loss_bc": 0.02975619025528431, "train_loss_llm": 0.0, "grad_norm": 0.030979283154010773, "global_step": 26689, "epoch": 741, "lr": 0.001} +{"train_loss": 0.0304284505546093, "train_loss_bc": 0.0304284505546093, "train_loss_llm": 0.0, "grad_norm": 0.05012348294258118, "global_step": 26690, "epoch": 741, "lr": 0.001} +{"train_loss": 0.029993440955877304, "train_loss_bc": 0.029993440955877304, "train_loss_llm": 0.0, "grad_norm": 0.07392711937427521, "global_step": 26691, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03249815106391907, "train_loss_bc": 0.03249815106391907, "train_loss_llm": 0.0, "grad_norm": 0.10665131360292435, "global_step": 26692, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03161189705133438, "train_loss_bc": 0.03161189705133438, "train_loss_llm": 0.0, "grad_norm": 0.033656515181064606, "global_step": 26693, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03194619342684746, "train_loss_bc": 0.03194619342684746, "train_loss_llm": 0.0, "grad_norm": 0.056492093950510025, "global_step": 26694, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03265327587723732, "train_loss_bc": 0.032636143267154694, "train_loss_llm": 0.0017133355140686035, "grad_norm": 0.09271931648254395, "global_step": 26695, "epoch": 741, "lr": 0.001} +{"train_loss": 0.028410468250513077, "train_loss_bc": 0.028410468250513077, "train_loss_llm": 0.0, "grad_norm": 0.1208149790763855, "global_step": 26696, "epoch": 741, "lr": 0.001} +{"train_loss": 0.02569938637316227, "train_loss_bc": 0.02569938637316227, "train_loss_llm": 0.0, "grad_norm": 0.04160568490624428, "global_step": 26697, "epoch": 741, "lr": 0.001} +{"train_loss": 0.02575179748237133, "train_loss_bc": 0.02575179748237133, "train_loss_llm": 0.0, "grad_norm": 0.0586739256978035, "global_step": 26698, "epoch": 741, "lr": 0.001} +{"train_loss": 0.031349971890449524, "train_loss_bc": 0.031349971890449524, "train_loss_llm": 0.0, "grad_norm": 0.07676150649785995, "global_step": 26699, "epoch": 741, "lr": 0.001} +{"train_loss": 0.031164417043328285, "train_loss_bc": 0.03114611655473709, "train_loss_llm": 0.0018300414085388184, "grad_norm": 0.08559463918209076, "global_step": 26700, "epoch": 741, "lr": 0.001} +{"train_loss": 0.032878439873456955, "train_loss_bc": 0.032878439873456955, "train_loss_llm": 0.0, "grad_norm": 0.0355268232524395, "global_step": 26701, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03834712132811546, "train_loss_bc": 0.03833978250622749, "train_loss_llm": 0.000733792781829834, "grad_norm": 0.05444271117448807, "global_step": 26702, "epoch": 741, "lr": 0.001} +{"train_loss": 0.027798384428024292, "train_loss_bc": 0.027798384428024292, "train_loss_llm": 0.0, "grad_norm": 0.06454229354858398, "global_step": 26703, "epoch": 741, "lr": 0.001} +{"train_loss": 0.031897954642772675, "train_loss_bc": 0.031897954642772675, "train_loss_llm": 0.0, "grad_norm": 0.0872240662574768, "global_step": 26704, "epoch": 741, "lr": 0.001} +{"train_loss": 0.031665414571762085, "train_loss_bc": 0.031665414571762085, "train_loss_llm": 0.0, "grad_norm": 0.03340497240424156, "global_step": 26705, "epoch": 741, "lr": 0.001} +{"train_loss": 0.025914261117577553, "train_loss_bc": 0.02589533105492592, "train_loss_llm": 0.0018929243087768555, "grad_norm": 0.06261415034532547, "global_step": 26706, "epoch": 741, "lr": 0.001} +{"train_loss": 0.036947112530469894, "train_loss_bc": 0.036947112530469894, "train_loss_llm": 0.0, "grad_norm": 0.09496632218360901, "global_step": 26707, "epoch": 741, "lr": 0.001} +{"train_loss": 0.028306150808930397, "train_loss_bc": 0.02829800359904766, "train_loss_llm": 0.0008147656917572021, "grad_norm": 0.12299512326717377, "global_step": 26708, "epoch": 741, "lr": 0.001} +{"train_loss": 0.031556639820337296, "train_loss_bc": 0.031556639820337296, "train_loss_llm": 0.0, "grad_norm": 0.04603222757577896, "global_step": 26709, "epoch": 741, "lr": 0.001} +{"train_loss": 0.03011760488152504, "train_loss_bc": 0.03011036105453968, "train_loss_llm": 0.00072440505027771, "grad_norm": 0.07014968991279602, "global_step": 26710, "epoch": 741, "lr": 0.001} +{"train_loss": 0.031216120761301782, "train_loss_bc": 0.04038568586111069, "train_loss_llm": 0.0007915198802947998, "grad_norm": 0.10199524462223053, "global_step": 26711, "epoch": 741, "lr": 0.001, "val_loss": 0.09237580001354218} +{"train_loss": 0.03028133511543274, "train_loss_bc": 0.03027748502790928, "train_loss_llm": 0.00038507580757141113, "grad_norm": 0.11968252062797546, "global_step": 26712, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03188103064894676, "train_loss_bc": 0.03188103064894676, "train_loss_llm": 0.0, "grad_norm": 0.04385678097605705, "global_step": 26713, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03788262978196144, "train_loss_bc": 0.03787412494421005, "train_loss_llm": 0.0008505880832672119, "grad_norm": 0.07511269301176071, "global_step": 26714, "epoch": 742, "lr": 0.001} +{"train_loss": 0.027710335329174995, "train_loss_bc": 0.027710335329174995, "train_loss_llm": 0.0, "grad_norm": 0.10369540005922318, "global_step": 26715, "epoch": 742, "lr": 0.001} +{"train_loss": 0.02815377712249756, "train_loss_bc": 0.028148874640464783, "train_loss_llm": 0.0004903078079223633, "grad_norm": 0.12538090348243713, "global_step": 26716, "epoch": 742, "lr": 0.001} +{"train_loss": 0.031897835433483124, "train_loss_bc": 0.031897835433483124, "train_loss_llm": 0.0, "grad_norm": 0.03884948790073395, "global_step": 26717, "epoch": 742, "lr": 0.001} +{"train_loss": 0.02493811771273613, "train_loss_bc": 0.02493811771273613, "train_loss_llm": 0.0, "grad_norm": 0.06178456172347069, "global_step": 26718, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03365372121334076, "train_loss_bc": 0.03365372121334076, "train_loss_llm": 0.0, "grad_norm": 0.08353369683027267, "global_step": 26719, "epoch": 742, "lr": 0.001} +{"train_loss": 0.029011305421590805, "train_loss_bc": 0.029007909819483757, "train_loss_llm": 0.000339508056640625, "grad_norm": 0.1079263910651207, "global_step": 26720, "epoch": 742, "lr": 0.001} +{"train_loss": 0.032264310866594315, "train_loss_bc": 0.032264310866594315, "train_loss_llm": 0.0, "grad_norm": 0.04144042357802391, "global_step": 26721, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03424958884716034, "train_loss_bc": 0.03423891216516495, "train_loss_llm": 0.0010678470134735107, "grad_norm": 0.07352378219366074, "global_step": 26722, "epoch": 742, "lr": 0.001} +{"train_loss": 0.029143258929252625, "train_loss_bc": 0.029143258929252625, "train_loss_llm": 0.0, "grad_norm": 0.09037435054779053, "global_step": 26723, "epoch": 742, "lr": 0.001} +{"train_loss": 0.030487939715385437, "train_loss_bc": 0.030482217669487, "train_loss_llm": 0.0005722343921661377, "grad_norm": 0.11228574067354202, "global_step": 26724, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03532472625374794, "train_loss_bc": 0.03532472625374794, "train_loss_llm": 0.0, "grad_norm": 0.03889907896518707, "global_step": 26725, "epoch": 742, "lr": 0.001} +{"train_loss": 0.033237919211387634, "train_loss_bc": 0.033237919211387634, "train_loss_llm": 0.0, "grad_norm": 0.059565164148807526, "global_step": 26726, "epoch": 742, "lr": 0.001} +{"train_loss": 0.026005901396274567, "train_loss_bc": 0.02599305659532547, "train_loss_llm": 0.0012845396995544434, "grad_norm": 0.08110161870718002, "global_step": 26727, "epoch": 742, "lr": 0.001} +{"train_loss": 0.02674887515604496, "train_loss_bc": 0.02674887515604496, "train_loss_llm": 0.0, "grad_norm": 0.11012192815542221, "global_step": 26728, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03689544275403023, "train_loss_bc": 0.03689544275403023, "train_loss_llm": 0.0, "grad_norm": 0.044593147933483124, "global_step": 26729, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03119892254471779, "train_loss_bc": 0.03119892254471779, "train_loss_llm": 0.0, "grad_norm": 0.054294854402542114, "global_step": 26730, "epoch": 742, "lr": 0.001} +{"train_loss": 0.0296975988894701, "train_loss_bc": 0.0296975988894701, "train_loss_llm": 0.0, "grad_norm": 0.07494651526212692, "global_step": 26731, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03683175891637802, "train_loss_bc": 0.03683175891637802, "train_loss_llm": 0.0, "grad_norm": 0.09631617367267609, "global_step": 26732, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03419310599565506, "train_loss_bc": 0.03419310599565506, "train_loss_llm": 0.0, "grad_norm": 0.038838762789964676, "global_step": 26733, "epoch": 742, "lr": 0.001} +{"train_loss": 0.031141402199864388, "train_loss_bc": 0.031141402199864388, "train_loss_llm": 0.0, "grad_norm": 0.06265344470739365, "global_step": 26734, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03397950530052185, "train_loss_bc": 0.03397950530052185, "train_loss_llm": 0.0, "grad_norm": 0.08225604146718979, "global_step": 26735, "epoch": 742, "lr": 0.001} +{"train_loss": 0.029409607872366905, "train_loss_bc": 0.029409607872366905, "train_loss_llm": 0.0, "grad_norm": 0.10333151370286942, "global_step": 26736, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03411117568612099, "train_loss_bc": 0.03411117568612099, "train_loss_llm": 0.0, "grad_norm": 0.033400021493434906, "global_step": 26737, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03473793715238571, "train_loss_bc": 0.0347338542342186, "train_loss_llm": 0.0004082322120666504, "grad_norm": 0.05406036972999573, "global_step": 26738, "epoch": 742, "lr": 0.001} +{"train_loss": 0.024052301421761513, "train_loss_bc": 0.02404840476810932, "train_loss_llm": 0.0003896057605743408, "grad_norm": 0.06858694553375244, "global_step": 26739, "epoch": 742, "lr": 0.001} +{"train_loss": 0.03085888922214508, "train_loss_bc": 0.03085888922214508, "train_loss_llm": 0.0, "grad_norm": 0.09087049961090088, "global_step": 26740, "epoch": 742, "lr": 0.001} +{"train_loss": 0.030162466689944267, "train_loss_bc": 0.030162466689944267, "train_loss_llm": 0.0, "grad_norm": 0.02714761719107628, "global_step": 26741, "epoch": 742, "lr": 0.001} +{"train_loss": 0.034626953303813934, "train_loss_bc": 0.034626953303813934, "train_loss_llm": 0.0, "grad_norm": 0.04300348460674286, "global_step": 26742, "epoch": 742, "lr": 0.001} +{"train_loss": 0.026361476629972458, "train_loss_bc": 0.02635994926095009, "train_loss_llm": 0.0001526474952697754, "grad_norm": 0.053316231817007065, "global_step": 26743, "epoch": 742, "lr": 0.001} +{"train_loss": 0.030046561732888222, "train_loss_bc": 0.030046561732888222, "train_loss_llm": 0.0, "grad_norm": 0.0676453486084938, "global_step": 26744, "epoch": 742, "lr": 0.001} +{"train_loss": 0.029190951958298683, "train_loss_bc": 0.029190951958298683, "train_loss_llm": 0.0, "grad_norm": 0.02787916734814644, "global_step": 26745, "epoch": 742, "lr": 0.001} +{"train_loss": 0.030311429873108864, "train_loss_bc": 0.030311429873108864, "train_loss_llm": 0.0, "grad_norm": 0.04399878531694412, "global_step": 26746, "epoch": 742, "lr": 0.001} +{"train_loss": 0.031134392755726974, "train_loss_bc": 0.030154680833220482, "train_loss_llm": 0.00033614039421081543, "grad_norm": 0.06370387971401215, "global_step": 26747, "epoch": 742, "lr": 0.001, "val_loss": 0.09062933176755905} +{"train_loss": 0.026903530582785606, "train_loss_bc": 0.026903530582785606, "train_loss_llm": 0.0, "grad_norm": 0.07361207157373428, "global_step": 26748, "epoch": 743, "lr": 0.001} +{"train_loss": 0.03036312758922577, "train_loss_bc": 0.030360711738467216, "train_loss_llm": 0.0002415776252746582, "grad_norm": 0.031751688569784164, "global_step": 26749, "epoch": 743, "lr": 0.001} +{"train_loss": 0.028615839779376984, "train_loss_bc": 0.0286149512976408, "train_loss_llm": 8.890032768249512e-05, "grad_norm": 0.057056643068790436, "global_step": 26750, "epoch": 743, "lr": 0.001} +{"train_loss": 0.03115348145365715, "train_loss_bc": 0.031146235764026642, "train_loss_llm": 0.0007245242595672607, "grad_norm": 0.07449313998222351, "global_step": 26751, "epoch": 743, "lr": 0.001} +{"train_loss": 0.026933874934911728, "train_loss_bc": 0.026915838941931725, "train_loss_llm": 0.0018035173416137695, "grad_norm": 0.08548305183649063, "global_step": 26752, "epoch": 743, "lr": 0.001} +{"train_loss": 0.029588567093014717, "train_loss_bc": 0.02958761155605316, "train_loss_llm": 9.551644325256348e-05, "grad_norm": 0.03694751858711243, "global_step": 26753, "epoch": 743, "lr": 0.001} +{"train_loss": 0.03132596239447594, "train_loss_bc": 0.03132596239447594, "train_loss_llm": 0.0, "grad_norm": 0.05967850983142853, "global_step": 26754, "epoch": 743, "lr": 0.001} +{"train_loss": 0.02906094677746296, "train_loss_bc": 0.02906094677746296, "train_loss_llm": 0.0, "grad_norm": 0.08043110370635986, "global_step": 26755, "epoch": 743, "lr": 0.001} +{"train_loss": 0.03332488611340523, "train_loss_bc": 0.03332488611340523, "train_loss_llm": 0.0, "grad_norm": 0.10215875506401062, "global_step": 26756, "epoch": 743, "lr": 0.001} +{"train_loss": 0.028642840683460236, "train_loss_bc": 0.028642840683460236, "train_loss_llm": 0.0, "grad_norm": 0.02636188454926014, "global_step": 26757, "epoch": 743, "lr": 0.001} +{"train_loss": 0.03794924542307854, "train_loss_bc": 0.03794924542307854, "train_loss_llm": 0.0, "grad_norm": 0.05138067901134491, "global_step": 26758, "epoch": 743, "lr": 0.001} +{"train_loss": 0.0261062178760767, "train_loss_bc": 0.0261062178760767, "train_loss_llm": 0.0, "grad_norm": 0.07209858298301697, "global_step": 26759, "epoch": 743, "lr": 0.001} +{"train_loss": 0.026931695640087128, "train_loss_bc": 0.026931695640087128, "train_loss_llm": 0.0, "grad_norm": 0.09041038155555725, "global_step": 26760, "epoch": 743, "lr": 0.001} +{"train_loss": 0.03562743961811066, "train_loss_bc": 0.03562743961811066, "train_loss_llm": 0.0, "grad_norm": 0.03045487590134144, "global_step": 26761, "epoch": 743, "lr": 0.001} +{"train_loss": 0.028070125728845596, "train_loss_bc": 0.028070125728845596, "train_loss_llm": 0.0, "grad_norm": 0.050516337156295776, "global_step": 26762, "epoch": 743, "lr": 0.001} +{"train_loss": 0.026707220822572708, "train_loss_bc": 0.026707220822572708, "train_loss_llm": 0.0, "grad_norm": 0.07230553030967712, "global_step": 26763, "epoch": 743, "lr": 0.001} +{"train_loss": 0.030062472447752953, "train_loss_bc": 0.030062472447752953, "train_loss_llm": 0.0, "grad_norm": 0.09630639851093292, "global_step": 26764, "epoch": 743, "lr": 0.001} +{"train_loss": 0.03181244432926178, "train_loss_bc": 0.03181244432926178, "train_loss_llm": 0.0, "grad_norm": 0.03548668324947357, "global_step": 26765, "epoch": 743, "lr": 0.001} +{"train_loss": 0.032518744468688965, "train_loss_bc": 0.032518744468688965, "train_loss_llm": 0.0, "grad_norm": 0.04801427945494652, "global_step": 26766, "epoch": 743, "lr": 0.001} +{"train_loss": 0.030492179095745087, "train_loss_bc": 0.030492179095745087, "train_loss_llm": 0.0, "grad_norm": 0.06180458888411522, "global_step": 26767, "epoch": 743, "lr": 0.001} +{"train_loss": 0.0275832898914814, "train_loss_bc": 0.0275832898914814, "train_loss_llm": 0.0, "grad_norm": 0.08081848919391632, "global_step": 26768, "epoch": 743, "lr": 0.001} +{"train_loss": 0.037639688700437546, "train_loss_bc": 0.037639688700437546, "train_loss_llm": 0.0, "grad_norm": 0.044265441596508026, "global_step": 26769, "epoch": 743, "lr": 0.001} +{"train_loss": 0.030802695080637932, "train_loss_bc": 0.030802695080637932, "train_loss_llm": 0.0, "grad_norm": 0.05717219412326813, "global_step": 26770, "epoch": 743, "lr": 0.001} +{"train_loss": 0.03250226750969887, "train_loss_bc": 0.03250226750969887, "train_loss_llm": 0.0, "grad_norm": 0.08440093696117401, "global_step": 26771, "epoch": 743, "lr": 0.001} +{"train_loss": 0.032337725162506104, "train_loss_bc": 0.032337725162506104, "train_loss_llm": 0.0, "grad_norm": 0.10177033394575119, "global_step": 26772, "epoch": 743, "lr": 0.001} +{"train_loss": 0.027748610824346542, "train_loss_bc": 0.027748610824346542, "train_loss_llm": 0.0, "grad_norm": 0.035606641322374344, "global_step": 26773, "epoch": 743, "lr": 0.001} +{"train_loss": 0.03366735205054283, "train_loss_bc": 0.03366735205054283, "train_loss_llm": 0.0, "grad_norm": 0.05548481270670891, "global_step": 26774, "epoch": 743, "lr": 0.001} +{"train_loss": 0.039043817669153214, "train_loss_bc": 0.039043817669153214, "train_loss_llm": 0.0, "grad_norm": 0.07854949682950974, "global_step": 26775, "epoch": 743, "lr": 0.001} +{"train_loss": 0.031204355880618095, "train_loss_bc": 0.031204355880618095, "train_loss_llm": 0.0, "grad_norm": 0.09243267774581909, "global_step": 26776, "epoch": 743, "lr": 0.001} +{"train_loss": 0.032537396997213364, "train_loss_bc": 0.032537396997213364, "train_loss_llm": 0.0, "grad_norm": 0.044158775359392166, "global_step": 26777, "epoch": 743, "lr": 0.001} +{"train_loss": 0.027463437989354134, "train_loss_bc": 0.027463437989354134, "train_loss_llm": 0.0, "grad_norm": 0.06623607128858566, "global_step": 26778, "epoch": 743, "lr": 0.001} +{"train_loss": 0.031301215291023254, "train_loss_bc": 0.031301215291023254, "train_loss_llm": 0.0, "grad_norm": 0.08426206558942795, "global_step": 26779, "epoch": 743, "lr": 0.001} +{"train_loss": 0.030070001259446144, "train_loss_bc": 0.030070001259446144, "train_loss_llm": 0.0, "grad_norm": 0.10371068120002747, "global_step": 26780, "epoch": 743, "lr": 0.001} +{"train_loss": 0.031163115054368973, "train_loss_bc": 0.031146742403507233, "train_loss_llm": 0.0016372203826904297, "grad_norm": 0.032160475850105286, "global_step": 26781, "epoch": 743, "lr": 0.001} +{"train_loss": 0.034127723425626755, "train_loss_bc": 0.034127723425626755, "train_loss_llm": 0.0, "grad_norm": 0.055516425520181656, "global_step": 26782, "epoch": 743, "lr": 0.001} +{"train_loss": 0.031056237562249105, "train_loss_bc": 0.03663194924592972, "train_loss_llm": 0.0009067654609680176, "grad_norm": 0.0820876732468605, "global_step": 26783, "epoch": 743, "lr": 0.001, "val_loss": 0.09227463603019714} +{"train_loss": 0.03229164704680443, "train_loss_bc": 0.03227456659078598, "train_loss_llm": 0.001707911491394043, "grad_norm": 0.1055421382188797, "global_step": 26784, "epoch": 744, "lr": 0.001} +{"train_loss": 0.028824687004089355, "train_loss_bc": 0.028824687004089355, "train_loss_llm": 0.0, "grad_norm": 0.02812127023935318, "global_step": 26785, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03258615359663963, "train_loss_bc": 0.03256189450621605, "train_loss_llm": 0.0024257302284240723, "grad_norm": 0.047141022980213165, "global_step": 26786, "epoch": 744, "lr": 0.001} +{"train_loss": 0.032304711639881134, "train_loss_bc": 0.032304711639881134, "train_loss_llm": 0.0, "grad_norm": 0.06014583632349968, "global_step": 26787, "epoch": 744, "lr": 0.001} +{"train_loss": 0.02990019880235195, "train_loss_bc": 0.02990019880235195, "train_loss_llm": 0.0, "grad_norm": 0.07724635303020477, "global_step": 26788, "epoch": 744, "lr": 0.001} +{"train_loss": 0.035854872316122055, "train_loss_bc": 0.035854872316122055, "train_loss_llm": 0.0, "grad_norm": 0.03133732080459595, "global_step": 26789, "epoch": 744, "lr": 0.001} +{"train_loss": 0.033119916915893555, "train_loss_bc": 0.033119916915893555, "train_loss_llm": 0.0, "grad_norm": 0.04507909342646599, "global_step": 26790, "epoch": 744, "lr": 0.001} +{"train_loss": 0.026910755783319473, "train_loss_bc": 0.026910755783319473, "train_loss_llm": 0.0, "grad_norm": 0.07155798375606537, "global_step": 26791, "epoch": 744, "lr": 0.001} +{"train_loss": 0.030668729916214943, "train_loss_bc": 0.030668729916214943, "train_loss_llm": 0.0, "grad_norm": 0.08595772832632065, "global_step": 26792, "epoch": 744, "lr": 0.001} +{"train_loss": 0.02971707470715046, "train_loss_bc": 0.029703916981816292, "train_loss_llm": 0.0013157427310943604, "grad_norm": 0.03696326166391373, "global_step": 26793, "epoch": 744, "lr": 0.001} +{"train_loss": 0.034399401396512985, "train_loss_bc": 0.034399401396512985, "train_loss_llm": 0.0, "grad_norm": 0.06912310421466827, "global_step": 26794, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03256820887327194, "train_loss_bc": 0.03256820887327194, "train_loss_llm": 0.0, "grad_norm": 0.0838077962398529, "global_step": 26795, "epoch": 744, "lr": 0.001} +{"train_loss": 0.029880067333579063, "train_loss_bc": 0.02984822355210781, "train_loss_llm": 0.003184288740158081, "grad_norm": 0.08837294578552246, "global_step": 26796, "epoch": 744, "lr": 0.001} +{"train_loss": 0.030920766294002533, "train_loss_bc": 0.030920766294002533, "train_loss_llm": 0.0, "grad_norm": 0.03543861582875252, "global_step": 26797, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03452920541167259, "train_loss_bc": 0.03452332317829132, "train_loss_llm": 0.0005880892276763916, "grad_norm": 0.05376154184341431, "global_step": 26798, "epoch": 744, "lr": 0.001} +{"train_loss": 0.028370456770062447, "train_loss_bc": 0.028370456770062447, "train_loss_llm": 0.0, "grad_norm": 0.083279088139534, "global_step": 26799, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03080608695745468, "train_loss_bc": 0.03080608695745468, "train_loss_llm": 0.0, "grad_norm": 0.11045882105827332, "global_step": 26800, "epoch": 744, "lr": 0.001} +{"train_loss": 0.028854670003056526, "train_loss_bc": 0.028854670003056526, "train_loss_llm": 0.0, "grad_norm": 0.031377218663692474, "global_step": 26801, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03732900321483612, "train_loss_bc": 0.03732183948159218, "train_loss_llm": 0.0007164180278778076, "grad_norm": 0.05907774716615677, "global_step": 26802, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03179744631052017, "train_loss_bc": 0.03179744631052017, "train_loss_llm": 0.0, "grad_norm": 0.07782929390668869, "global_step": 26803, "epoch": 744, "lr": 0.001} +{"train_loss": 0.025687959045171738, "train_loss_bc": 0.025687959045171738, "train_loss_llm": 0.0, "grad_norm": 0.10427960753440857, "global_step": 26804, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03514157235622406, "train_loss_bc": 0.03514157235622406, "train_loss_llm": 0.0, "grad_norm": 0.03621981665492058, "global_step": 26805, "epoch": 744, "lr": 0.001} +{"train_loss": 0.026126263663172722, "train_loss_bc": 0.026126263663172722, "train_loss_llm": 0.0, "grad_norm": 0.041621264070272446, "global_step": 26806, "epoch": 744, "lr": 0.001} +{"train_loss": 0.030239032581448555, "train_loss_bc": 0.030239032581448555, "train_loss_llm": 0.0, "grad_norm": 0.06605735421180725, "global_step": 26807, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03220297023653984, "train_loss_bc": 0.032202936708927155, "train_loss_llm": 3.4570693969726562e-06, "grad_norm": 0.08258110284805298, "global_step": 26808, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03458587825298309, "train_loss_bc": 0.03458587825298309, "train_loss_llm": 0.0, "grad_norm": 0.033299870789051056, "global_step": 26809, "epoch": 744, "lr": 0.001} +{"train_loss": 0.034318797290325165, "train_loss_bc": 0.034318797290325165, "train_loss_llm": 0.0, "grad_norm": 0.05196120962500572, "global_step": 26810, "epoch": 744, "lr": 0.001} +{"train_loss": 0.02662409469485283, "train_loss_bc": 0.02662409469485283, "train_loss_llm": 0.0, "grad_norm": 0.06758037209510803, "global_step": 26811, "epoch": 744, "lr": 0.001} +{"train_loss": 0.033905278891325, "train_loss_bc": 0.033905278891325, "train_loss_llm": 0.0, "grad_norm": 0.07866540551185608, "global_step": 26812, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03142442926764488, "train_loss_bc": 0.03142442926764488, "train_loss_llm": 0.0, "grad_norm": 0.032996732741594315, "global_step": 26813, "epoch": 744, "lr": 0.001} +{"train_loss": 0.02185199409723282, "train_loss_bc": 0.02185199409723282, "train_loss_llm": 0.0, "grad_norm": 0.05317556858062744, "global_step": 26814, "epoch": 744, "lr": 0.001} +{"train_loss": 0.030061133205890656, "train_loss_bc": 0.030049718916416168, "train_loss_llm": 0.0011414289474487305, "grad_norm": 0.06910517066717148, "global_step": 26815, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03306279331445694, "train_loss_bc": 0.03306279331445694, "train_loss_llm": 0.0, "grad_norm": 0.09045517444610596, "global_step": 26816, "epoch": 744, "lr": 0.001} +{"train_loss": 0.028449878096580505, "train_loss_bc": 0.028449878096580505, "train_loss_llm": 0.0, "grad_norm": 0.038250554352998734, "global_step": 26817, "epoch": 744, "lr": 0.001} +{"train_loss": 0.03218609467148781, "train_loss_bc": 0.03218609467148781, "train_loss_llm": 0.0, "grad_norm": 0.07187522947788239, "global_step": 26818, "epoch": 744, "lr": 0.001} +{"train_loss": 0.030992068712496094, "train_loss_bc": 0.02821224369108677, "train_loss_llm": 0.0, "grad_norm": 0.10497970879077911, "global_step": 26819, "epoch": 744, "lr": 0.001, "val_loss": 0.09304492175579071} +{"train_loss": 0.030084675177931786, "train_loss_bc": 0.030084675177931786, "train_loss_llm": 0.0, "grad_norm": 0.13826224207878113, "global_step": 26820, "epoch": 745, "lr": 0.001} +{"train_loss": 0.029847729951143265, "train_loss_bc": 0.029847729951143265, "train_loss_llm": 0.0, "grad_norm": 0.03610438108444214, "global_step": 26821, "epoch": 745, "lr": 0.001} +{"train_loss": 0.02984347566962242, "train_loss_bc": 0.02984347566962242, "train_loss_llm": 0.0, "grad_norm": 0.06359146535396576, "global_step": 26822, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03075028955936432, "train_loss_bc": 0.03075028955936432, "train_loss_llm": 0.0, "grad_norm": 0.10499655455350876, "global_step": 26823, "epoch": 745, "lr": 0.001} +{"train_loss": 0.028731027618050575, "train_loss_bc": 0.028731027618050575, "train_loss_llm": 0.0, "grad_norm": 0.12960302829742432, "global_step": 26824, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03741590306162834, "train_loss_bc": 0.03740760684013367, "train_loss_llm": 0.0008294880390167236, "grad_norm": 0.03980430215597153, "global_step": 26825, "epoch": 745, "lr": 0.001} +{"train_loss": 0.028984861448407173, "train_loss_bc": 0.028984861448407173, "train_loss_llm": 0.0, "grad_norm": 0.05498014762997627, "global_step": 26826, "epoch": 745, "lr": 0.001} +{"train_loss": 0.029954081401228905, "train_loss_bc": 0.029954081401228905, "train_loss_llm": 0.0, "grad_norm": 0.07469771802425385, "global_step": 26827, "epoch": 745, "lr": 0.001} +{"train_loss": 0.02850787155330181, "train_loss_bc": 0.02850787155330181, "train_loss_llm": 0.0, "grad_norm": 0.08881392329931259, "global_step": 26828, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03147696703672409, "train_loss_bc": 0.03147696703672409, "train_loss_llm": 0.0, "grad_norm": 0.029724223539233208, "global_step": 26829, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03614857792854309, "train_loss_bc": 0.036146070808172226, "train_loss_llm": 0.0002505779266357422, "grad_norm": 0.043030839413404465, "global_step": 26830, "epoch": 745, "lr": 0.001} +{"train_loss": 0.033242613077163696, "train_loss_bc": 0.033242613077163696, "train_loss_llm": 0.0, "grad_norm": 0.054544996470212936, "global_step": 26831, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03294362872838974, "train_loss_bc": 0.03294362872838974, "train_loss_llm": 0.0, "grad_norm": 0.07458539307117462, "global_step": 26832, "epoch": 745, "lr": 0.001} +{"train_loss": 0.029042378067970276, "train_loss_bc": 0.02902997098863125, "train_loss_llm": 0.0012407004833221436, "grad_norm": 0.03852711617946625, "global_step": 26833, "epoch": 745, "lr": 0.001} +{"train_loss": 0.027286427095532417, "train_loss_bc": 0.027286427095532417, "train_loss_llm": 0.0, "grad_norm": 0.05203397944569588, "global_step": 26834, "epoch": 745, "lr": 0.001} +{"train_loss": 0.02933192066848278, "train_loss_bc": 0.02933192066848278, "train_loss_llm": 0.0, "grad_norm": 0.07274070382118225, "global_step": 26835, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03181968629360199, "train_loss_bc": 0.03181968629360199, "train_loss_llm": 0.0, "grad_norm": 0.09497439116239548, "global_step": 26836, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03261503949761391, "train_loss_bc": 0.032613497227430344, "train_loss_llm": 0.00015440583229064941, "grad_norm": 0.04324990138411522, "global_step": 26837, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03088918887078762, "train_loss_bc": 0.03087611310184002, "train_loss_llm": 0.0013076066970825195, "grad_norm": 0.06519007682800293, "global_step": 26838, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03263260796666145, "train_loss_bc": 0.03263260796666145, "train_loss_llm": 0.0, "grad_norm": 0.0991009771823883, "global_step": 26839, "epoch": 745, "lr": 0.001} +{"train_loss": 0.02594885416328907, "train_loss_bc": 0.02594885416328907, "train_loss_llm": 0.0, "grad_norm": 0.11923081427812576, "global_step": 26840, "epoch": 745, "lr": 0.001} +{"train_loss": 0.029041480273008347, "train_loss_bc": 0.02900182083249092, "train_loss_llm": 0.003965914249420166, "grad_norm": 0.042801663279533386, "global_step": 26841, "epoch": 745, "lr": 0.001} +{"train_loss": 0.036538101732730865, "train_loss_bc": 0.036538101732730865, "train_loss_llm": 0.0, "grad_norm": 0.07951907068490982, "global_step": 26842, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03013521246612072, "train_loss_bc": 0.03013521246612072, "train_loss_llm": 0.0, "grad_norm": 0.11658868938684464, "global_step": 26843, "epoch": 745, "lr": 0.001} +{"train_loss": 0.035902589559555054, "train_loss_bc": 0.035902589559555054, "train_loss_llm": 0.0, "grad_norm": 0.1555713564157486, "global_step": 26844, "epoch": 745, "lr": 0.001} +{"train_loss": 0.033517178148031235, "train_loss_bc": 0.033511705696582794, "train_loss_llm": 0.0005473792552947998, "grad_norm": 0.03320760279893875, "global_step": 26845, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03372644633054733, "train_loss_bc": 0.03372644633054733, "train_loss_llm": 0.0, "grad_norm": 0.05924924090504646, "global_step": 26846, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03793849050998688, "train_loss_bc": 0.03793557733297348, "train_loss_llm": 0.0002912282943725586, "grad_norm": 0.07952243089675903, "global_step": 26847, "epoch": 745, "lr": 0.001} +{"train_loss": 0.023216629400849342, "train_loss_bc": 0.023199373856186867, "train_loss_llm": 0.0017254948616027832, "grad_norm": 0.09730371087789536, "global_step": 26848, "epoch": 745, "lr": 0.001} +{"train_loss": 0.029963476583361626, "train_loss_bc": 0.029963476583361626, "train_loss_llm": 0.0, "grad_norm": 0.036486975848674774, "global_step": 26849, "epoch": 745, "lr": 0.001} +{"train_loss": 0.024093493819236755, "train_loss_bc": 0.024081913754343987, "train_loss_llm": 0.0011579692363739014, "grad_norm": 0.04454950615763664, "global_step": 26850, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03524697944521904, "train_loss_bc": 0.03523179516196251, "train_loss_llm": 0.001518547534942627, "grad_norm": 0.058145295828580856, "global_step": 26851, "epoch": 745, "lr": 0.001} +{"train_loss": 0.027948008850216866, "train_loss_bc": 0.027928540483117104, "train_loss_llm": 0.001946777105331421, "grad_norm": 0.0685615986585617, "global_step": 26852, "epoch": 745, "lr": 0.001} +{"train_loss": 0.029295392334461212, "train_loss_bc": 0.029295392334461212, "train_loss_llm": 0.0, "grad_norm": 0.03931117057800293, "global_step": 26853, "epoch": 745, "lr": 0.001} +{"train_loss": 0.04098357632756233, "train_loss_bc": 0.04098357632756233, "train_loss_llm": 0.0, "grad_norm": 0.07496377825737, "global_step": 26854, "epoch": 745, "lr": 0.001} +{"train_loss": 0.03120478206417627, "train_loss_bc": 0.028327293694019318, "train_loss_llm": 0.0, "grad_norm": 0.10026664286851883, "global_step": 26855, "epoch": 745, "lr": 0.001, "train/cumulative_reward": 40.15808936049843, "train/mean_score": 7.533360128724868, "train/success_rate": 0.2, "test/cumulative_reward": 57.12083639188436, "test/mean_score": 8.89926628526802, "test/success_rate": 0.22, "val_loss": 0.09360498189926147, "train_action_mse_error": 0.034677453339099884} +{"train_loss": 0.028975773602724075, "train_loss_bc": 0.028975773602724075, "train_loss_llm": 0.0, "grad_norm": 0.12275799363851547, "global_step": 26856, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03385506942868233, "train_loss_bc": 0.03385506942868233, "train_loss_llm": 0.0, "grad_norm": 0.036177296191453934, "global_step": 26857, "epoch": 746, "lr": 0.001} +{"train_loss": 0.026793166995048523, "train_loss_bc": 0.026793166995048523, "train_loss_llm": 0.0, "grad_norm": 0.06913875043392181, "global_step": 26858, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03450258821249008, "train_loss_bc": 0.03450226038694382, "train_loss_llm": 3.293156623840332e-05, "grad_norm": 0.09971770644187927, "global_step": 26859, "epoch": 746, "lr": 0.001} +{"train_loss": 0.029544075950980186, "train_loss_bc": 0.029544075950980186, "train_loss_llm": 0.0, "grad_norm": 0.13179413974285126, "global_step": 26860, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03136146813631058, "train_loss_bc": 0.03135344013571739, "train_loss_llm": 0.0008027851581573486, "grad_norm": 0.037745166569948196, "global_step": 26861, "epoch": 746, "lr": 0.001} +{"train_loss": 0.028309470042586327, "train_loss_bc": 0.028309470042586327, "train_loss_llm": 0.0, "grad_norm": 0.06344593316316605, "global_step": 26862, "epoch": 746, "lr": 0.001} +{"train_loss": 0.02499326504766941, "train_loss_bc": 0.02499326504766941, "train_loss_llm": 0.0, "grad_norm": 0.07821159064769745, "global_step": 26863, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03551879897713661, "train_loss_bc": 0.03551879897713661, "train_loss_llm": 0.0, "grad_norm": 0.11199183762073517, "global_step": 26864, "epoch": 746, "lr": 0.001} +{"train_loss": 0.028767379000782967, "train_loss_bc": 0.028761515393853188, "train_loss_llm": 0.0005864202976226807, "grad_norm": 0.030624907463788986, "global_step": 26865, "epoch": 746, "lr": 0.001} +{"train_loss": 0.02687816508114338, "train_loss_bc": 0.02687816508114338, "train_loss_llm": 0.0, "grad_norm": 0.042827434837818146, "global_step": 26866, "epoch": 746, "lr": 0.001} +{"train_loss": 0.040690500289201736, "train_loss_bc": 0.040690500289201736, "train_loss_llm": 0.0, "grad_norm": 0.06282716989517212, "global_step": 26867, "epoch": 746, "lr": 0.001} +{"train_loss": 0.034774065017700195, "train_loss_bc": 0.034774065017700195, "train_loss_llm": 0.0, "grad_norm": 0.07897862046957016, "global_step": 26868, "epoch": 746, "lr": 0.001} +{"train_loss": 0.032657086849212646, "train_loss_bc": 0.032657086849212646, "train_loss_llm": 0.0, "grad_norm": 0.03429397940635681, "global_step": 26869, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03196064755320549, "train_loss_bc": 0.03196064755320549, "train_loss_llm": 0.0, "grad_norm": 0.05568120256066322, "global_step": 26870, "epoch": 746, "lr": 0.001} +{"train_loss": 0.02865549921989441, "train_loss_bc": 0.02865549921989441, "train_loss_llm": 0.0, "grad_norm": 0.07174720615148544, "global_step": 26871, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03361567109823227, "train_loss_bc": 0.03361567109823227, "train_loss_llm": 0.0, "grad_norm": 0.09384110569953918, "global_step": 26872, "epoch": 746, "lr": 0.001} +{"train_loss": 0.029850538820028305, "train_loss_bc": 0.029850538820028305, "train_loss_llm": 0.0, "grad_norm": 0.040375545620918274, "global_step": 26873, "epoch": 746, "lr": 0.001} +{"train_loss": 0.031041014939546585, "train_loss_bc": 0.03103497624397278, "train_loss_llm": 0.000603795051574707, "grad_norm": 0.05567875877022743, "global_step": 26874, "epoch": 746, "lr": 0.001} +{"train_loss": 0.027867795899510384, "train_loss_bc": 0.027867795899510384, "train_loss_llm": 0.0, "grad_norm": 0.07569628208875656, "global_step": 26875, "epoch": 746, "lr": 0.001} +{"train_loss": 0.031981006264686584, "train_loss_bc": 0.031981006264686584, "train_loss_llm": 0.0, "grad_norm": 0.1057853102684021, "global_step": 26876, "epoch": 746, "lr": 0.001} +{"train_loss": 0.032729096710681915, "train_loss_bc": 0.03272135928273201, "train_loss_llm": 0.0007738173007965088, "grad_norm": 0.03543353080749512, "global_step": 26877, "epoch": 746, "lr": 0.001} +{"train_loss": 0.033136460930109024, "train_loss_bc": 0.0331108458340168, "train_loss_llm": 0.0025613605976104736, "grad_norm": 0.06030585616827011, "global_step": 26878, "epoch": 746, "lr": 0.001} +{"train_loss": 0.025111595168709755, "train_loss_bc": 0.025111595168709755, "train_loss_llm": 0.0, "grad_norm": 0.07127954065799713, "global_step": 26879, "epoch": 746, "lr": 0.001} +{"train_loss": 0.033532142639160156, "train_loss_bc": 0.033532142639160156, "train_loss_llm": 0.0, "grad_norm": 0.09181054681539536, "global_step": 26880, "epoch": 746, "lr": 0.001} +{"train_loss": 0.028096208348870277, "train_loss_bc": 0.028096208348870277, "train_loss_llm": 0.0, "grad_norm": 0.032857757061719894, "global_step": 26881, "epoch": 746, "lr": 0.001} +{"train_loss": 0.029819367453455925, "train_loss_bc": 0.029819367453455925, "train_loss_llm": 0.0, "grad_norm": 0.04965609312057495, "global_step": 26882, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03348815441131592, "train_loss_bc": 0.03347991406917572, "train_loss_llm": 0.0008239150047302246, "grad_norm": 0.07139527052640915, "global_step": 26883, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03156881406903267, "train_loss_bc": 0.03156881406903267, "train_loss_llm": 0.0, "grad_norm": 0.09457647800445557, "global_step": 26884, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03147190809249878, "train_loss_bc": 0.03147190809249878, "train_loss_llm": 0.0, "grad_norm": 0.035342030227184296, "global_step": 26885, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03503561019897461, "train_loss_bc": 0.03503561019897461, "train_loss_llm": 0.0, "grad_norm": 0.06171658635139465, "global_step": 26886, "epoch": 746, "lr": 0.001} +{"train_loss": 0.03360113501548767, "train_loss_bc": 0.03360113501548767, "train_loss_llm": 0.0, "grad_norm": 0.08394508063793182, "global_step": 26887, "epoch": 746, "lr": 0.001} +{"train_loss": 0.030062278732657433, "train_loss_bc": 0.03005860187113285, "train_loss_llm": 0.0003676414489746094, "grad_norm": 0.11462078988552094, "global_step": 26888, "epoch": 746, "lr": 0.001} +{"train_loss": 0.02832251414656639, "train_loss_bc": 0.02831912972033024, "train_loss_llm": 0.00033843517303466797, "grad_norm": 0.02976306341588497, "global_step": 26889, "epoch": 746, "lr": 0.001} +{"train_loss": 0.032159361988306046, "train_loss_bc": 0.0321408174932003, "train_loss_llm": 0.0018542706966400146, "grad_norm": 0.04295121878385544, "global_step": 26890, "epoch": 746, "lr": 0.001} +{"train_loss": 0.031110375912653074, "train_loss_bc": 0.029245838522911072, "train_loss_llm": 0.0, "grad_norm": 0.07013437151908875, "global_step": 26891, "epoch": 746, "lr": 0.001, "val_loss": 0.08980882912874222} +{"train_loss": 0.029752323403954506, "train_loss_bc": 0.029752323403954506, "train_loss_llm": 0.0, "grad_norm": 0.08661525696516037, "global_step": 26892, "epoch": 747, "lr": 0.001} +{"train_loss": 0.03073885850608349, "train_loss_bc": 0.03073885850608349, "train_loss_llm": 0.0, "grad_norm": 0.03800386190414429, "global_step": 26893, "epoch": 747, "lr": 0.001} +{"train_loss": 0.03195738047361374, "train_loss_bc": 0.03195738047361374, "train_loss_llm": 0.0, "grad_norm": 0.06623473018407822, "global_step": 26894, "epoch": 747, "lr": 0.001} +{"train_loss": 0.031409502029418945, "train_loss_bc": 0.031409502029418945, "train_loss_llm": 0.0, "grad_norm": 0.08717522770166397, "global_step": 26895, "epoch": 747, "lr": 0.001} +{"train_loss": 0.024507125839591026, "train_loss_bc": 0.024507125839591026, "train_loss_llm": 0.0, "grad_norm": 0.10676795989274979, "global_step": 26896, "epoch": 747, "lr": 0.001} +{"train_loss": 0.02547559328377247, "train_loss_bc": 0.02547559328377247, "train_loss_llm": 0.0, "grad_norm": 0.02964148111641407, "global_step": 26897, "epoch": 747, "lr": 0.001} +{"train_loss": 0.032654523849487305, "train_loss_bc": 0.03265080600976944, "train_loss_llm": 0.00037169456481933594, "grad_norm": 0.05975053086876869, "global_step": 26898, "epoch": 747, "lr": 0.001} +{"train_loss": 0.02466646395623684, "train_loss_bc": 0.024663731455802917, "train_loss_llm": 0.00027319788932800293, "grad_norm": 0.07877171784639359, "global_step": 26899, "epoch": 747, "lr": 0.001} +{"train_loss": 0.028022175654768944, "train_loss_bc": 0.028022175654768944, "train_loss_llm": 0.0, "grad_norm": 0.0913546234369278, "global_step": 26900, "epoch": 747, "lr": 0.001} +{"train_loss": 0.030191943049430847, "train_loss_bc": 0.030191943049430847, "train_loss_llm": 0.0, "grad_norm": 0.03344003111124039, "global_step": 26901, "epoch": 747, "lr": 0.001} +{"train_loss": 0.03601263090968132, "train_loss_bc": 0.03600698336958885, "train_loss_llm": 0.0005648434162139893, "grad_norm": 0.05685587227344513, "global_step": 26902, "epoch": 747, "lr": 0.001} +{"train_loss": 0.037215933203697205, "train_loss_bc": 0.03720524162054062, "train_loss_llm": 0.0010692775249481201, "grad_norm": 0.07272996008396149, "global_step": 26903, "epoch": 747, "lr": 0.001} +{"train_loss": 0.0276870708912611, "train_loss_bc": 0.0276870708912611, "train_loss_llm": 0.0, "grad_norm": 0.09076446294784546, "global_step": 26904, "epoch": 747, "lr": 0.001} +{"train_loss": 0.038050953298807144, "train_loss_bc": 0.038050953298807144, "train_loss_llm": 0.0, "grad_norm": 0.04606939107179642, "global_step": 26905, "epoch": 747, "lr": 0.001} +{"train_loss": 0.032399941235780716, "train_loss_bc": 0.032399941235780716, "train_loss_llm": 0.0, "grad_norm": 0.06888298690319061, "global_step": 26906, "epoch": 747, "lr": 0.001} +{"train_loss": 0.030207065865397453, "train_loss_bc": 0.030203357338905334, "train_loss_llm": 0.00037088990211486816, "grad_norm": 0.10145454108715057, "global_step": 26907, "epoch": 747, "lr": 0.001} +{"train_loss": 0.03524642437696457, "train_loss_bc": 0.03524642437696457, "train_loss_llm": 0.0, "grad_norm": 0.13592594861984253, "global_step": 26908, "epoch": 747, "lr": 0.001} +{"train_loss": 0.029626639559864998, "train_loss_bc": 0.029626639559864998, "train_loss_llm": 0.0, "grad_norm": 0.028654959052801132, "global_step": 26909, "epoch": 747, "lr": 0.001} +{"train_loss": 0.030227508395910263, "train_loss_bc": 0.030227508395910263, "train_loss_llm": 0.0, "grad_norm": 0.04120238870382309, "global_step": 26910, "epoch": 747, "lr": 0.001} +{"train_loss": 0.030919481068849564, "train_loss_bc": 0.03090706840157509, "train_loss_llm": 0.0012412667274475098, "grad_norm": 0.07027260959148407, "global_step": 26911, "epoch": 747, "lr": 0.001} +{"train_loss": 0.03676459565758705, "train_loss_bc": 0.03676459565758705, "train_loss_llm": 0.0, "grad_norm": 0.08582807332277298, "global_step": 26912, "epoch": 747, "lr": 0.001} +{"train_loss": 0.02904699556529522, "train_loss_bc": 0.02904699556529522, "train_loss_llm": 0.0, "grad_norm": 0.036503829061985016, "global_step": 26913, "epoch": 747, "lr": 0.001} +{"train_loss": 0.030694499611854553, "train_loss_bc": 0.030694499611854553, "train_loss_llm": 0.0, "grad_norm": 0.0641108974814415, "global_step": 26914, "epoch": 747, "lr": 0.001} +{"train_loss": 0.028512855991721153, "train_loss_bc": 0.028512855991721153, "train_loss_llm": 0.0, "grad_norm": 0.07124347239732742, "global_step": 26915, "epoch": 747, "lr": 0.001} +{"train_loss": 0.034457165747880936, "train_loss_bc": 0.03445332869887352, "train_loss_llm": 0.0003838837146759033, "grad_norm": 0.08709974586963654, "global_step": 26916, "epoch": 747, "lr": 0.001} +{"train_loss": 0.0252063125371933, "train_loss_bc": 0.0252063125371933, "train_loss_llm": 0.0, "grad_norm": 0.029882969334721565, "global_step": 26917, "epoch": 747, "lr": 0.001} +{"train_loss": 0.03273049741983414, "train_loss_bc": 0.03272883594036102, "train_loss_llm": 0.00016629695892333984, "grad_norm": 0.04020417109131813, "global_step": 26918, "epoch": 747, "lr": 0.001} +{"train_loss": 0.035700999200344086, "train_loss_bc": 0.035700999200344086, "train_loss_llm": 0.0, "grad_norm": 0.04880781099200249, "global_step": 26919, "epoch": 747, "lr": 0.001} +{"train_loss": 0.029058346524834633, "train_loss_bc": 0.029058346524834633, "train_loss_llm": 0.0, "grad_norm": 0.06461021304130554, "global_step": 26920, "epoch": 747, "lr": 0.001} +{"train_loss": 0.03665095195174217, "train_loss_bc": 0.03665095195174217, "train_loss_llm": 0.0, "grad_norm": 0.04306657984852791, "global_step": 26921, "epoch": 747, "lr": 0.001} +{"train_loss": 0.030851570889353752, "train_loss_bc": 0.030851570889353752, "train_loss_llm": 0.0, "grad_norm": 0.06250841915607452, "global_step": 26922, "epoch": 747, "lr": 0.001} +{"train_loss": 0.026737330481410027, "train_loss_bc": 0.026737330481410027, "train_loss_llm": 0.0, "grad_norm": 0.07666613906621933, "global_step": 26923, "epoch": 747, "lr": 0.001} +{"train_loss": 0.03126813471317291, "train_loss_bc": 0.03126813471317291, "train_loss_llm": 0.0, "grad_norm": 0.09481775015592575, "global_step": 26924, "epoch": 747, "lr": 0.001} +{"train_loss": 0.026346607133746147, "train_loss_bc": 0.026346607133746147, "train_loss_llm": 0.0, "grad_norm": 0.03161275386810303, "global_step": 26925, "epoch": 747, "lr": 0.001} +{"train_loss": 0.03231048583984375, "train_loss_bc": 0.03231048583984375, "train_loss_llm": 0.0, "grad_norm": 0.04949239268898964, "global_step": 26926, "epoch": 747, "lr": 0.001} +{"train_loss": 0.031084067343423765, "train_loss_bc": 0.03571953624486923, "train_loss_llm": 0.0, "grad_norm": 0.07325153052806854, "global_step": 26927, "epoch": 747, "lr": 0.001, "val_loss": 0.09287456423044205} +{"train_loss": 0.030467435717582703, "train_loss_bc": 0.030467435717582703, "train_loss_llm": 0.0, "grad_norm": 0.08063969016075134, "global_step": 26928, "epoch": 748, "lr": 0.001} +{"train_loss": 0.03314513340592384, "train_loss_bc": 0.03314513340592384, "train_loss_llm": 0.0, "grad_norm": 0.03358989208936691, "global_step": 26929, "epoch": 748, "lr": 0.001} +{"train_loss": 0.03289831802248955, "train_loss_bc": 0.03289418667554855, "train_loss_llm": 0.00041306018829345703, "grad_norm": 0.05350828543305397, "global_step": 26930, "epoch": 748, "lr": 0.001} +{"train_loss": 0.025208845734596252, "train_loss_bc": 0.025208845734596252, "train_loss_llm": 0.0, "grad_norm": 0.07114856690168381, "global_step": 26931, "epoch": 748, "lr": 0.001} +{"train_loss": 0.02862093225121498, "train_loss_bc": 0.02862093225121498, "train_loss_llm": 0.0, "grad_norm": 0.08513588458299637, "global_step": 26932, "epoch": 748, "lr": 0.001} +{"train_loss": 0.031640857458114624, "train_loss_bc": 0.031640857458114624, "train_loss_llm": 0.0, "grad_norm": 0.036375656723976135, "global_step": 26933, "epoch": 748, "lr": 0.001} +{"train_loss": 0.02677309140563011, "train_loss_bc": 0.02677309140563011, "train_loss_llm": 0.0, "grad_norm": 0.05936616659164429, "global_step": 26934, "epoch": 748, "lr": 0.001} +{"train_loss": 0.030015621334314346, "train_loss_bc": 0.030015621334314346, "train_loss_llm": 0.0, "grad_norm": 0.06966292858123779, "global_step": 26935, "epoch": 748, "lr": 0.001} +{"train_loss": 0.031517963856458664, "train_loss_bc": 0.031517963856458664, "train_loss_llm": 0.0, "grad_norm": 0.08718807250261307, "global_step": 26936, "epoch": 748, "lr": 0.001} +{"train_loss": 0.029181286692619324, "train_loss_bc": 0.029181286692619324, "train_loss_llm": 0.0, "grad_norm": 0.029565101489424706, "global_step": 26937, "epoch": 748, "lr": 0.001} +{"train_loss": 0.035094618797302246, "train_loss_bc": 0.03508917614817619, "train_loss_llm": 0.0005443096160888672, "grad_norm": 0.051072217524051666, "global_step": 26938, "epoch": 748, "lr": 0.001} +{"train_loss": 0.031502604484558105, "train_loss_bc": 0.031502604484558105, "train_loss_llm": 0.0, "grad_norm": 0.06948273628950119, "global_step": 26939, "epoch": 748, "lr": 0.001} +{"train_loss": 0.02891268953680992, "train_loss_bc": 0.02891268953680992, "train_loss_llm": 0.0, "grad_norm": 0.07692892849445343, "global_step": 26940, "epoch": 748, "lr": 0.001} +{"train_loss": 0.031375691294670105, "train_loss_bc": 0.031375691294670105, "train_loss_llm": 0.0, "grad_norm": 0.04022533819079399, "global_step": 26941, "epoch": 748, "lr": 0.001} +{"train_loss": 0.02799285762012005, "train_loss_bc": 0.02799285762012005, "train_loss_llm": 0.0, "grad_norm": 0.06493466347455978, "global_step": 26942, "epoch": 748, "lr": 0.001} +{"train_loss": 0.034332651644945145, "train_loss_bc": 0.034332651644945145, "train_loss_llm": 0.0, "grad_norm": 0.08286058902740479, "global_step": 26943, "epoch": 748, "lr": 0.001} +{"train_loss": 0.0330997034907341, "train_loss_bc": 0.0330997034907341, "train_loss_llm": 0.0, "grad_norm": 0.10144032537937164, "global_step": 26944, "epoch": 748, "lr": 0.001} +{"train_loss": 0.03543347865343094, "train_loss_bc": 0.03543347865343094, "train_loss_llm": 0.0, "grad_norm": 0.04087413474917412, "global_step": 26945, "epoch": 748, "lr": 0.001} +{"train_loss": 0.033034831285476685, "train_loss_bc": 0.033034831285476685, "train_loss_llm": 0.0, "grad_norm": 0.06453064829111099, "global_step": 26946, "epoch": 748, "lr": 0.001} +{"train_loss": 0.027330618351697922, "train_loss_bc": 0.027330618351697922, "train_loss_llm": 0.0, "grad_norm": 0.07717013359069824, "global_step": 26947, "epoch": 748, "lr": 0.001} +{"train_loss": 0.029316486790776253, "train_loss_bc": 0.029316486790776253, "train_loss_llm": 0.0, "grad_norm": 0.09815529733896255, "global_step": 26948, "epoch": 748, "lr": 0.001} +{"train_loss": 0.029655136168003082, "train_loss_bc": 0.029655136168003082, "train_loss_llm": 0.0, "grad_norm": 0.03168242424726486, "global_step": 26949, "epoch": 748, "lr": 0.001} +{"train_loss": 0.02947995811700821, "train_loss_bc": 0.02947995811700821, "train_loss_llm": 0.0, "grad_norm": 0.06674826145172119, "global_step": 26950, "epoch": 748, "lr": 0.001} +{"train_loss": 0.03417186439037323, "train_loss_bc": 0.03417186439037323, "train_loss_llm": 0.0, "grad_norm": 0.08680255711078644, "global_step": 26951, "epoch": 748, "lr": 0.001} +{"train_loss": 0.031045088544487953, "train_loss_bc": 0.0310443677008152, "train_loss_llm": 7.200241088867188e-05, "grad_norm": 0.11964300274848938, "global_step": 26952, "epoch": 748, "lr": 0.001} +{"train_loss": 0.026741493493318558, "train_loss_bc": 0.026741493493318558, "train_loss_llm": 0.0, "grad_norm": 0.03814174234867096, "global_step": 26953, "epoch": 748, "lr": 0.001} +{"train_loss": 0.029729431495070457, "train_loss_bc": 0.029729431495070457, "train_loss_llm": 0.0, "grad_norm": 0.062353868037462234, "global_step": 26954, "epoch": 748, "lr": 0.001} +{"train_loss": 0.03352605551481247, "train_loss_bc": 0.03352605551481247, "train_loss_llm": 0.0, "grad_norm": 0.07821296900510788, "global_step": 26955, "epoch": 748, "lr": 0.001} +{"train_loss": 0.029930012300610542, "train_loss_bc": 0.02992205135524273, "train_loss_llm": 0.0007961392402648926, "grad_norm": 0.0967506393790245, "global_step": 26956, "epoch": 748, "lr": 0.001} +{"train_loss": 0.029264669865369797, "train_loss_bc": 0.029246162623167038, "train_loss_llm": 0.0018506944179534912, "grad_norm": 0.03723887354135513, "global_step": 26957, "epoch": 748, "lr": 0.001} +{"train_loss": 0.029621699824929237, "train_loss_bc": 0.029604513198137283, "train_loss_llm": 0.001718670129776001, "grad_norm": 0.059899430721998215, "global_step": 26958, "epoch": 748, "lr": 0.001} +{"train_loss": 0.033230721950531006, "train_loss_bc": 0.033230721950531006, "train_loss_llm": 0.0, "grad_norm": 0.08010156452655792, "global_step": 26959, "epoch": 748, "lr": 0.001} +{"train_loss": 0.029961345717310905, "train_loss_bc": 0.029961345717310905, "train_loss_llm": 0.0, "grad_norm": 0.10959804058074951, "global_step": 26960, "epoch": 748, "lr": 0.001} +{"train_loss": 0.036888912320137024, "train_loss_bc": 0.036888912320137024, "train_loss_llm": 0.0, "grad_norm": 0.03075672686100006, "global_step": 26961, "epoch": 748, "lr": 0.001} +{"train_loss": 0.03257278352975845, "train_loss_bc": 0.03257278352975845, "train_loss_llm": 0.0, "grad_norm": 0.056530922651290894, "global_step": 26962, "epoch": 748, "lr": 0.001} +{"train_loss": 0.031138870916846726, "train_loss_bc": 0.038235507905483246, "train_loss_llm": 0.004895240068435669, "grad_norm": 0.08475562930107117, "global_step": 26963, "epoch": 748, "lr": 0.001, "val_loss": 0.09271036088466644} +{"train_loss": 0.03452935069799423, "train_loss_bc": 0.03452383726835251, "train_loss_llm": 0.0005513131618499756, "grad_norm": 0.11049079895019531, "global_step": 26964, "epoch": 749, "lr": 0.001} +{"train_loss": 0.030301343649625778, "train_loss_bc": 0.030301343649625778, "train_loss_llm": 0.0, "grad_norm": 0.03292127326130867, "global_step": 26965, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03272928297519684, "train_loss_bc": 0.03272344544529915, "train_loss_llm": 0.0005835890769958496, "grad_norm": 0.04769573733210564, "global_step": 26966, "epoch": 749, "lr": 0.001} +{"train_loss": 0.02305416204035282, "train_loss_bc": 0.023041149601340294, "train_loss_llm": 0.0013012886047363281, "grad_norm": 0.06313417106866837, "global_step": 26967, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03051300160586834, "train_loss_bc": 0.03051300160586834, "train_loss_llm": 0.0, "grad_norm": 0.07956047356128693, "global_step": 26968, "epoch": 749, "lr": 0.001} +{"train_loss": 0.025479739531874657, "train_loss_bc": 0.02546732872724533, "train_loss_llm": 0.001241147518157959, "grad_norm": 0.03223621845245361, "global_step": 26969, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03248731791973114, "train_loss_bc": 0.03248731791973114, "train_loss_llm": 0.0, "grad_norm": 0.05158447101712227, "global_step": 26970, "epoch": 749, "lr": 0.001} +{"train_loss": 0.02986126020550728, "train_loss_bc": 0.029853885993361473, "train_loss_llm": 0.0007374584674835205, "grad_norm": 0.06895852088928223, "global_step": 26971, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03208167105913162, "train_loss_bc": 0.03208167105913162, "train_loss_llm": 0.0, "grad_norm": 0.09413929283618927, "global_step": 26972, "epoch": 749, "lr": 0.001} +{"train_loss": 0.0277091097086668, "train_loss_bc": 0.027694521471858025, "train_loss_llm": 0.0014587938785552979, "grad_norm": 0.031872622668743134, "global_step": 26973, "epoch": 749, "lr": 0.001} +{"train_loss": 0.035168662667274475, "train_loss_bc": 0.035168662667274475, "train_loss_llm": 0.0, "grad_norm": 0.0565405935049057, "global_step": 26974, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03509090468287468, "train_loss_bc": 0.03509005904197693, "train_loss_llm": 8.472800254821777e-05, "grad_norm": 0.07180161774158478, "global_step": 26975, "epoch": 749, "lr": 0.001} +{"train_loss": 0.0322851836681366, "train_loss_bc": 0.0322851836681366, "train_loss_llm": 0.0, "grad_norm": 0.09091579169034958, "global_step": 26976, "epoch": 749, "lr": 0.001} +{"train_loss": 0.028750862926244736, "train_loss_bc": 0.028750862926244736, "train_loss_llm": 0.0, "grad_norm": 0.0422821044921875, "global_step": 26977, "epoch": 749, "lr": 0.001} +{"train_loss": 0.0347944013774395, "train_loss_bc": 0.03478993475437164, "train_loss_llm": 0.00044652819633483887, "grad_norm": 0.059841372072696686, "global_step": 26978, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03252214938402176, "train_loss_bc": 0.03250643610954285, "train_loss_llm": 0.0015712082386016846, "grad_norm": 0.068572036921978, "global_step": 26979, "epoch": 749, "lr": 0.001} +{"train_loss": 0.033357374370098114, "train_loss_bc": 0.033357374370098114, "train_loss_llm": 0.0, "grad_norm": 0.0927400141954422, "global_step": 26980, "epoch": 749, "lr": 0.001} +{"train_loss": 0.029983360320329666, "train_loss_bc": 0.029983360320329666, "train_loss_llm": 0.0, "grad_norm": 0.03216452896595001, "global_step": 26981, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03692994266748428, "train_loss_bc": 0.03692994266748428, "train_loss_llm": 0.0, "grad_norm": 0.05604083091020584, "global_step": 26982, "epoch": 749, "lr": 0.001} +{"train_loss": 0.030286237597465515, "train_loss_bc": 0.030273817479610443, "train_loss_llm": 0.0012420713901519775, "grad_norm": 0.07128095626831055, "global_step": 26983, "epoch": 749, "lr": 0.001} +{"train_loss": 0.030569327995181084, "train_loss_bc": 0.030569327995181084, "train_loss_llm": 0.0, "grad_norm": 0.0942501574754715, "global_step": 26984, "epoch": 749, "lr": 0.001} +{"train_loss": 0.02865869738161564, "train_loss_bc": 0.02865869738161564, "train_loss_llm": 0.0, "grad_norm": 0.037314943969249725, "global_step": 26985, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03281569108366966, "train_loss_bc": 0.03281569108366966, "train_loss_llm": 0.0, "grad_norm": 0.0574336014688015, "global_step": 26986, "epoch": 749, "lr": 0.001} +{"train_loss": 0.026321373879909515, "train_loss_bc": 0.02630992792546749, "train_loss_llm": 0.0011446475982666016, "grad_norm": 0.07101915031671524, "global_step": 26987, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03262731432914734, "train_loss_bc": 0.03262731432914734, "train_loss_llm": 0.0, "grad_norm": 0.09058400243520737, "global_step": 26988, "epoch": 749, "lr": 0.001} +{"train_loss": 0.031534045934677124, "train_loss_bc": 0.031534045934677124, "train_loss_llm": 0.0, "grad_norm": 0.03292585909366608, "global_step": 26989, "epoch": 749, "lr": 0.001} +{"train_loss": 0.031874414533376694, "train_loss_bc": 0.031874414533376694, "train_loss_llm": 0.0, "grad_norm": 0.04941559582948685, "global_step": 26990, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03372669219970703, "train_loss_bc": 0.03372669219970703, "train_loss_llm": 0.0, "grad_norm": 0.06577055901288986, "global_step": 26991, "epoch": 749, "lr": 0.001} +{"train_loss": 0.03178359568119049, "train_loss_bc": 0.03178359568119049, "train_loss_llm": 0.0, "grad_norm": 0.08143404871225357, "global_step": 26992, "epoch": 749, "lr": 0.001} +{"train_loss": 0.02549484744668007, "train_loss_bc": 0.02549484744668007, "train_loss_llm": 0.0, "grad_norm": 0.027374625205993652, "global_step": 26993, "epoch": 749, "lr": 0.001} +{"train_loss": 0.027780188247561455, "train_loss_bc": 0.027780188247561455, "train_loss_llm": 0.0, "grad_norm": 0.04481334984302521, "global_step": 26994, "epoch": 749, "lr": 0.001} +{"train_loss": 0.02802131697535515, "train_loss_bc": 0.02802131697535515, "train_loss_llm": 0.0, "grad_norm": 0.05739959329366684, "global_step": 26995, "epoch": 749, "lr": 0.001} +{"train_loss": 0.027202486991882324, "train_loss_bc": 0.027202486991882324, "train_loss_llm": 0.0, "grad_norm": 0.06322792917490005, "global_step": 26996, "epoch": 749, "lr": 0.001} +{"train_loss": 0.033844590187072754, "train_loss_bc": 0.033844590187072754, "train_loss_llm": 0.0, "grad_norm": 0.03263850137591362, "global_step": 26997, "epoch": 749, "lr": 0.001} +{"train_loss": 0.0351664237678051, "train_loss_bc": 0.0351664237678051, "train_loss_llm": 0.0, "grad_norm": 0.04400372505187988, "global_step": 26998, "epoch": 749, "lr": 0.001} +{"train_loss": 0.030965601456248097, "train_loss_bc": 0.029425326734781265, "train_loss_llm": 0.0, "grad_norm": 0.06606088578701019, "global_step": 26999, "epoch": 749, "lr": 0.001, "val_loss": 0.09322044253349304} +{"train_loss": 0.03326054662466049, "train_loss_bc": 0.03326054662466049, "train_loss_llm": 0.0, "grad_norm": 0.07894131541252136, "global_step": 27000, "epoch": 750, "lr": 0.001} +{"train_loss": 0.032887496054172516, "train_loss_bc": 0.032887496054172516, "train_loss_llm": 0.0, "grad_norm": 0.03150193393230438, "global_step": 27001, "epoch": 750, "lr": 0.001} +{"train_loss": 0.023847343400120735, "train_loss_bc": 0.023847343400120735, "train_loss_llm": 0.0, "grad_norm": 0.04958978667855263, "global_step": 27002, "epoch": 750, "lr": 0.001} +{"train_loss": 0.03441404178738594, "train_loss_bc": 0.03441404178738594, "train_loss_llm": 0.0, "grad_norm": 0.0702192634344101, "global_step": 27003, "epoch": 750, "lr": 0.001} +{"train_loss": 0.030483199283480644, "train_loss_bc": 0.030483199283480644, "train_loss_llm": 0.0, "grad_norm": 0.091159388422966, "global_step": 27004, "epoch": 750, "lr": 0.001} +{"train_loss": 0.027367958799004555, "train_loss_bc": 0.027354778721928596, "train_loss_llm": 0.0013179779052734375, "grad_norm": 0.027771661058068275, "global_step": 27005, "epoch": 750, "lr": 0.001} +{"train_loss": 0.03548168018460274, "train_loss_bc": 0.03548168018460274, "train_loss_llm": 0.0, "grad_norm": 0.052268531173467636, "global_step": 27006, "epoch": 750, "lr": 0.001} +{"train_loss": 0.031038936227560043, "train_loss_bc": 0.031038936227560043, "train_loss_llm": 0.0, "grad_norm": 0.07558353990316391, "global_step": 27007, "epoch": 750, "lr": 0.001} +{"train_loss": 0.029420647770166397, "train_loss_bc": 0.029420647770166397, "train_loss_llm": 0.0, "grad_norm": 0.09356635063886642, "global_step": 27008, "epoch": 750, "lr": 0.001} +{"train_loss": 0.031080136075615883, "train_loss_bc": 0.031080136075615883, "train_loss_llm": 0.0, "grad_norm": 0.037611305713653564, "global_step": 27009, "epoch": 750, "lr": 0.001} +{"train_loss": 0.030882814899086952, "train_loss_bc": 0.030882814899086952, "train_loss_llm": 0.0, "grad_norm": 0.060249213129282, "global_step": 27010, "epoch": 750, "lr": 0.001} +{"train_loss": 0.03128749132156372, "train_loss_bc": 0.03128749132156372, "train_loss_llm": 0.0, "grad_norm": 0.06963442265987396, "global_step": 27011, "epoch": 750, "lr": 0.001} +{"train_loss": 0.023389389738440514, "train_loss_bc": 0.023389389738440514, "train_loss_llm": 0.0, "grad_norm": 0.08717738837003708, "global_step": 27012, "epoch": 750, "lr": 0.001} +{"train_loss": 0.02565688081085682, "train_loss_bc": 0.02565469592809677, "train_loss_llm": 0.00021857023239135742, "grad_norm": 0.02715761959552765, "global_step": 27013, "epoch": 750, "lr": 0.001} +{"train_loss": 0.0397336520254612, "train_loss_bc": 0.0397336520254612, "train_loss_llm": 0.0, "grad_norm": 0.04934978857636452, "global_step": 27014, "epoch": 750, "lr": 0.001} +{"train_loss": 0.035367291420698166, "train_loss_bc": 0.035367291420698166, "train_loss_llm": 0.0, "grad_norm": 0.061840739101171494, "global_step": 27015, "epoch": 750, "lr": 0.001} +{"train_loss": 0.029361242428421974, "train_loss_bc": 0.029345529153943062, "train_loss_llm": 0.001571357250213623, "grad_norm": 0.07441070675849915, "global_step": 27016, "epoch": 750, "lr": 0.001} +{"train_loss": 0.025848649442195892, "train_loss_bc": 0.025847841054201126, "train_loss_llm": 8.085370063781738e-05, "grad_norm": 0.03041812591254711, "global_step": 27017, "epoch": 750, "lr": 0.001} +{"train_loss": 0.029270634055137634, "train_loss_bc": 0.029270634055137634, "train_loss_llm": 0.0, "grad_norm": 0.04900149628520012, "global_step": 27018, "epoch": 750, "lr": 0.001} +{"train_loss": 0.026359498500823975, "train_loss_bc": 0.026359498500823975, "train_loss_llm": 0.0, "grad_norm": 0.06000622361898422, "global_step": 27019, "epoch": 750, "lr": 0.001} +{"train_loss": 0.030801383778452873, "train_loss_bc": 0.03079042211174965, "train_loss_llm": 0.0010961592197418213, "grad_norm": 0.06977251172065735, "global_step": 27020, "epoch": 750, "lr": 0.001} +{"train_loss": 0.03022916428744793, "train_loss_bc": 0.03022916428744793, "train_loss_llm": 0.0, "grad_norm": 0.03275780379772186, "global_step": 27021, "epoch": 750, "lr": 0.001} +{"train_loss": 0.029429083690047264, "train_loss_bc": 0.029418211430311203, "train_loss_llm": 0.001087188720703125, "grad_norm": 0.04722900316119194, "global_step": 27022, "epoch": 750, "lr": 0.001} +{"train_loss": 0.03675328567624092, "train_loss_bc": 0.036749426275491714, "train_loss_llm": 0.0003857612609863281, "grad_norm": 0.06441837549209595, "global_step": 27023, "epoch": 750, "lr": 0.001} +{"train_loss": 0.027607198804616928, "train_loss_bc": 0.02760346792638302, "train_loss_llm": 0.0003730356693267822, "grad_norm": 0.08344312012195587, "global_step": 27024, "epoch": 750, "lr": 0.001} +{"train_loss": 0.030893966555595398, "train_loss_bc": 0.030893966555595398, "train_loss_llm": 0.0, "grad_norm": 0.029479531571269035, "global_step": 27025, "epoch": 750, "lr": 0.001} +{"train_loss": 0.025884469971060753, "train_loss_bc": 0.025884469971060753, "train_loss_llm": 0.0, "grad_norm": 0.049269866198301315, "global_step": 27026, "epoch": 750, "lr": 0.001} +{"train_loss": 0.035592690110206604, "train_loss_bc": 0.035592690110206604, "train_loss_llm": 0.0, "grad_norm": 0.06858424842357635, "global_step": 27027, "epoch": 750, "lr": 0.001} +{"train_loss": 0.03467598557472229, "train_loss_bc": 0.03467598557472229, "train_loss_llm": 0.0, "grad_norm": 0.08475055545568466, "global_step": 27028, "epoch": 750, "lr": 0.001} +{"train_loss": 0.032508864998817444, "train_loss_bc": 0.032508864998817444, "train_loss_llm": 0.0, "grad_norm": 0.049213796854019165, "global_step": 27029, "epoch": 750, "lr": 0.001} +{"train_loss": 0.030551061034202576, "train_loss_bc": 0.030551061034202576, "train_loss_llm": 0.0, "grad_norm": 0.07222874462604523, "global_step": 27030, "epoch": 750, "lr": 0.001} +{"train_loss": 0.034948308020830154, "train_loss_bc": 0.034948308020830154, "train_loss_llm": 0.0, "grad_norm": 0.09592083096504211, "global_step": 27031, "epoch": 750, "lr": 0.001} +{"train_loss": 0.029118821024894714, "train_loss_bc": 0.02910757251083851, "train_loss_llm": 0.0011247992515563965, "grad_norm": 0.11318980157375336, "global_step": 27032, "epoch": 750, "lr": 0.001} +{"train_loss": 0.030802486464381218, "train_loss_bc": 0.030802486464381218, "train_loss_llm": 0.0, "grad_norm": 0.02696896344423294, "global_step": 27033, "epoch": 750, "lr": 0.001} +{"train_loss": 0.03300700709223747, "train_loss_bc": 0.03300700709223747, "train_loss_llm": 0.0, "grad_norm": 0.04601641371846199, "global_step": 27034, "epoch": 750, "lr": 0.001} +{"train_loss": 0.03090269723907113, "train_loss_bc": 0.03325379267334938, "train_loss_llm": 0.0, "grad_norm": 0.06365016102790833, "global_step": 27035, "epoch": 750, "lr": 0.001, "train/cumulative_reward": 86.9855784512214, "train/mean_score": 14.732799373832785, "train/success_rate": 0.3, "test/cumulative_reward": 36.40981499821848, "test/mean_score": 7.649084747903492, "test/success_rate": 0.18, "val_loss": 0.09290903061628342, "train_action_mse_error": 0.0342271588742733} +{"train_loss": 0.03486275672912598, "train_loss_bc": 0.03486275672912598, "train_loss_llm": 0.0, "grad_norm": 0.08886966854333878, "global_step": 27036, "epoch": 751, "lr": 0.001} +{"train_loss": 0.02698025479912758, "train_loss_bc": 0.02698025479912758, "train_loss_llm": 0.0, "grad_norm": 0.029615500941872597, "global_step": 27037, "epoch": 751, "lr": 0.001} +{"train_loss": 0.025522662326693535, "train_loss_bc": 0.025522662326693535, "train_loss_llm": 0.0, "grad_norm": 0.03647766634821892, "global_step": 27038, "epoch": 751, "lr": 0.001} +{"train_loss": 0.036430343985557556, "train_loss_bc": 0.036430343985557556, "train_loss_llm": 0.0, "grad_norm": 0.059473369270563126, "global_step": 27039, "epoch": 751, "lr": 0.001} +{"train_loss": 0.026494411751627922, "train_loss_bc": 0.026479484513401985, "train_loss_llm": 0.0014926493167877197, "grad_norm": 0.07214102149009705, "global_step": 27040, "epoch": 751, "lr": 0.001} +{"train_loss": 0.026897702366113663, "train_loss_bc": 0.026897702366113663, "train_loss_llm": 0.0, "grad_norm": 0.02646966651082039, "global_step": 27041, "epoch": 751, "lr": 0.001} +{"train_loss": 0.031517378985881805, "train_loss_bc": 0.031517378985881805, "train_loss_llm": 0.0, "grad_norm": 0.04577397182583809, "global_step": 27042, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03042544052004814, "train_loss_bc": 0.03042544052004814, "train_loss_llm": 0.0, "grad_norm": 0.062092218548059464, "global_step": 27043, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03180912882089615, "train_loss_bc": 0.03180912882089615, "train_loss_llm": 0.0, "grad_norm": 0.08009839057922363, "global_step": 27044, "epoch": 751, "lr": 0.001} +{"train_loss": 0.028957046568393707, "train_loss_bc": 0.028953533619642258, "train_loss_llm": 0.00035125017166137695, "grad_norm": 0.03124801069498062, "global_step": 27045, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03247631713747978, "train_loss_bc": 0.03247631713747978, "train_loss_llm": 0.0, "grad_norm": 0.04633105546236038, "global_step": 27046, "epoch": 751, "lr": 0.001} +{"train_loss": 0.0324874222278595, "train_loss_bc": 0.0324874222278595, "train_loss_llm": 0.0, "grad_norm": 0.06868784874677658, "global_step": 27047, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03359929844737053, "train_loss_bc": 0.03358147665858269, "train_loss_llm": 0.0017823576927185059, "grad_norm": 0.08004719018936157, "global_step": 27048, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03159891441464424, "train_loss_bc": 0.03159891441464424, "train_loss_llm": 0.0, "grad_norm": 0.03511470556259155, "global_step": 27049, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03384978324174881, "train_loss_bc": 0.03384978324174881, "train_loss_llm": 0.0, "grad_norm": 0.058036044239997864, "global_step": 27050, "epoch": 751, "lr": 0.001} +{"train_loss": 0.032941896468400955, "train_loss_bc": 0.032934173941612244, "train_loss_llm": 0.0007721781730651855, "grad_norm": 0.08083539456129074, "global_step": 27051, "epoch": 751, "lr": 0.001} +{"train_loss": 0.027711158618330956, "train_loss_bc": 0.027711158618330956, "train_loss_llm": 0.0, "grad_norm": 0.10140783339738846, "global_step": 27052, "epoch": 751, "lr": 0.001} +{"train_loss": 0.025892682373523712, "train_loss_bc": 0.025877323001623154, "train_loss_llm": 0.001535862684249878, "grad_norm": 0.030450468882918358, "global_step": 27053, "epoch": 751, "lr": 0.001} +{"train_loss": 0.029849683865904808, "train_loss_bc": 0.029849683865904808, "train_loss_llm": 0.0, "grad_norm": 0.05192854627966881, "global_step": 27054, "epoch": 751, "lr": 0.001} +{"train_loss": 0.032378457486629486, "train_loss_bc": 0.032378457486629486, "train_loss_llm": 0.0, "grad_norm": 0.06507139652967453, "global_step": 27055, "epoch": 751, "lr": 0.001} +{"train_loss": 0.030660545453429222, "train_loss_bc": 0.030628666281700134, "train_loss_llm": 0.0031879842281341553, "grad_norm": 0.08020893484354019, "global_step": 27056, "epoch": 751, "lr": 0.001} +{"train_loss": 0.02652391791343689, "train_loss_bc": 0.02652275748550892, "train_loss_llm": 0.00011596083641052246, "grad_norm": 0.029028458520770073, "global_step": 27057, "epoch": 751, "lr": 0.001} +{"train_loss": 0.031011255457997322, "train_loss_bc": 0.031011255457997322, "train_loss_llm": 0.0, "grad_norm": 0.05143633857369423, "global_step": 27058, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03265390545129776, "train_loss_bc": 0.03265390545129776, "train_loss_llm": 0.0, "grad_norm": 0.05930371582508087, "global_step": 27059, "epoch": 751, "lr": 0.001} +{"train_loss": 0.02926144190132618, "train_loss_bc": 0.02926144190132618, "train_loss_llm": 0.0, "grad_norm": 0.0717364251613617, "global_step": 27060, "epoch": 751, "lr": 0.001} +{"train_loss": 0.027686892077326775, "train_loss_bc": 0.027686892077326775, "train_loss_llm": 0.0, "grad_norm": 0.029259968549013138, "global_step": 27061, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03230375796556473, "train_loss_bc": 0.032287318259477615, "train_loss_llm": 0.0016438663005828857, "grad_norm": 0.04597089812159538, "global_step": 27062, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03008456900715828, "train_loss_bc": 0.030072994530200958, "train_loss_llm": 0.0011574029922485352, "grad_norm": 0.06665900349617004, "global_step": 27063, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03279957175254822, "train_loss_bc": 0.03279957175254822, "train_loss_llm": 0.0, "grad_norm": 0.08995427936315536, "global_step": 27064, "epoch": 751, "lr": 0.001} +{"train_loss": 0.031896986067295074, "train_loss_bc": 0.031896986067295074, "train_loss_llm": 0.0, "grad_norm": 0.039303265511989594, "global_step": 27065, "epoch": 751, "lr": 0.001} +{"train_loss": 0.032811325043439865, "train_loss_bc": 0.032811325043439865, "train_loss_llm": 0.0, "grad_norm": 0.0730910450220108, "global_step": 27066, "epoch": 751, "lr": 0.001} +{"train_loss": 0.028951674699783325, "train_loss_bc": 0.028951674699783325, "train_loss_llm": 0.0, "grad_norm": 0.08989124745130539, "global_step": 27067, "epoch": 751, "lr": 0.001} +{"train_loss": 0.029540298506617546, "train_loss_bc": 0.029540298506617546, "train_loss_llm": 0.0, "grad_norm": 0.11127004772424698, "global_step": 27068, "epoch": 751, "lr": 0.001} +{"train_loss": 0.03330502659082413, "train_loss_bc": 0.03330502659082413, "train_loss_llm": 0.0, "grad_norm": 0.03932638466358185, "global_step": 27069, "epoch": 751, "lr": 0.001} +{"train_loss": 0.033764228224754333, "train_loss_bc": 0.033764228224754333, "train_loss_llm": 0.0, "grad_norm": 0.052052970975637436, "global_step": 27070, "epoch": 751, "lr": 0.001} +{"train_loss": 0.030920483223679993, "train_loss_bc": 0.03719925880432129, "train_loss_llm": 0.0, "grad_norm": 0.09217706322669983, "global_step": 27071, "epoch": 751, "lr": 0.001, "val_loss": 0.09217258542776108} +{"train_loss": 0.03185570240020752, "train_loss_bc": 0.03185570240020752, "train_loss_llm": 0.0, "grad_norm": 0.12638241052627563, "global_step": 27072, "epoch": 752, "lr": 0.001} +{"train_loss": 0.02962016873061657, "train_loss_bc": 0.02962016873061657, "train_loss_llm": 0.0, "grad_norm": 0.031191300600767136, "global_step": 27073, "epoch": 752, "lr": 0.001} +{"train_loss": 0.03190267086029053, "train_loss_bc": 0.03190267086029053, "train_loss_llm": 0.0, "grad_norm": 0.04958476126194, "global_step": 27074, "epoch": 752, "lr": 0.001} +{"train_loss": 0.03572060167789459, "train_loss_bc": 0.03570114076137543, "train_loss_llm": 0.0019461512565612793, "grad_norm": 0.05281340330839157, "global_step": 27075, "epoch": 752, "lr": 0.001} +{"train_loss": 0.029105164110660553, "train_loss_bc": 0.029105164110660553, "train_loss_llm": 0.0, "grad_norm": 0.061285555362701416, "global_step": 27076, "epoch": 752, "lr": 0.001} +{"train_loss": 0.028082193806767464, "train_loss_bc": 0.028080271556973457, "train_loss_llm": 0.00019213557243347168, "grad_norm": 0.027669541537761688, "global_step": 27077, "epoch": 752, "lr": 0.001} +{"train_loss": 0.026046471670269966, "train_loss_bc": 0.026046471670269966, "train_loss_llm": 0.0, "grad_norm": 0.041536398231983185, "global_step": 27078, "epoch": 752, "lr": 0.001} +{"train_loss": 0.032191645354032516, "train_loss_bc": 0.032191645354032516, "train_loss_llm": 0.0, "grad_norm": 0.049434542655944824, "global_step": 27079, "epoch": 752, "lr": 0.001} +{"train_loss": 0.027698321267962456, "train_loss_bc": 0.02768293023109436, "train_loss_llm": 0.0015390515327453613, "grad_norm": 0.05444733053445816, "global_step": 27080, "epoch": 752, "lr": 0.001} +{"train_loss": 0.030499063432216644, "train_loss_bc": 0.030499063432216644, "train_loss_llm": 0.0, "grad_norm": 0.03267475217580795, "global_step": 27081, "epoch": 752, "lr": 0.001} +{"train_loss": 0.026241546496748924, "train_loss_bc": 0.026241546496748924, "train_loss_llm": 0.0, "grad_norm": 0.05143839120864868, "global_step": 27082, "epoch": 752, "lr": 0.001} +{"train_loss": 0.029837792739272118, "train_loss_bc": 0.029832623898983, "train_loss_llm": 0.0005168318748474121, "grad_norm": 0.06149088218808174, "global_step": 27083, "epoch": 752, "lr": 0.001} +{"train_loss": 0.03581719100475311, "train_loss_bc": 0.03581719100475311, "train_loss_llm": 0.0, "grad_norm": 0.07618537545204163, "global_step": 27084, "epoch": 752, "lr": 0.001} +{"train_loss": 0.03085792064666748, "train_loss_bc": 0.0308536309748888, "train_loss_llm": 0.00042897462844848633, "grad_norm": 0.032009828835725784, "global_step": 27085, "epoch": 752, "lr": 0.001} +{"train_loss": 0.035599276423454285, "train_loss_bc": 0.035599276423454285, "train_loss_llm": 0.0, "grad_norm": 0.05787636712193489, "global_step": 27086, "epoch": 752, "lr": 0.001} +{"train_loss": 0.02212035097181797, "train_loss_bc": 0.02212035097181797, "train_loss_llm": 0.0, "grad_norm": 0.07175175100564957, "global_step": 27087, "epoch": 752, "lr": 0.001} +{"train_loss": 0.029427694156765938, "train_loss_bc": 0.029427694156765938, "train_loss_llm": 0.0, "grad_norm": 0.09034106880426407, "global_step": 27088, "epoch": 752, "lr": 0.001} +{"train_loss": 0.03307278826832771, "train_loss_bc": 0.0330621600151062, "train_loss_llm": 0.0010627806186676025, "grad_norm": 0.0382881835103035, "global_step": 27089, "epoch": 752, "lr": 0.001} +{"train_loss": 0.035726819187402725, "train_loss_bc": 0.035726819187402725, "train_loss_llm": 0.0, "grad_norm": 0.0666080117225647, "global_step": 27090, "epoch": 752, "lr": 0.001} +{"train_loss": 0.035228531807661057, "train_loss_bc": 0.035228531807661057, "train_loss_llm": 0.0, "grad_norm": 0.07406733930110931, "global_step": 27091, "epoch": 752, "lr": 0.001} +{"train_loss": 0.030388817191123962, "train_loss_bc": 0.030388817191123962, "train_loss_llm": 0.0, "grad_norm": 0.09453792124986649, "global_step": 27092, "epoch": 752, "lr": 0.001} +{"train_loss": 0.027840524911880493, "train_loss_bc": 0.027840524911880493, "train_loss_llm": 0.0, "grad_norm": 0.036350324749946594, "global_step": 27093, "epoch": 752, "lr": 0.001} +{"train_loss": 0.029940558597445488, "train_loss_bc": 0.029940558597445488, "train_loss_llm": 0.0, "grad_norm": 0.05738198384642601, "global_step": 27094, "epoch": 752, "lr": 0.001} +{"train_loss": 0.03429289907217026, "train_loss_bc": 0.03429289907217026, "train_loss_llm": 0.0, "grad_norm": 0.07249881327152252, "global_step": 27095, "epoch": 752, "lr": 0.001} +{"train_loss": 0.03590911999344826, "train_loss_bc": 0.03590911999344826, "train_loss_llm": 0.0, "grad_norm": 0.10588132590055466, "global_step": 27096, "epoch": 752, "lr": 0.001} +{"train_loss": 0.02886851318180561, "train_loss_bc": 0.02886851318180561, "train_loss_llm": 0.0, "grad_norm": 0.02869885042309761, "global_step": 27097, "epoch": 752, "lr": 0.001} +{"train_loss": 0.026942458003759384, "train_loss_bc": 0.026942458003759384, "train_loss_llm": 0.0, "grad_norm": 0.04450023174285889, "global_step": 27098, "epoch": 752, "lr": 0.001} +{"train_loss": 0.034833360463380814, "train_loss_bc": 0.034833360463380814, "train_loss_llm": 0.0, "grad_norm": 0.05941736698150635, "global_step": 27099, "epoch": 752, "lr": 0.001} +{"train_loss": 0.03138258308172226, "train_loss_bc": 0.03137388452887535, "train_loss_llm": 0.0008698403835296631, "grad_norm": 0.07143068313598633, "global_step": 27100, "epoch": 752, "lr": 0.001} +{"train_loss": 0.032042354345321655, "train_loss_bc": 0.032042354345321655, "train_loss_llm": 0.0, "grad_norm": 0.03196899965405464, "global_step": 27101, "epoch": 752, "lr": 0.001} +{"train_loss": 0.03163669630885124, "train_loss_bc": 0.03163669630885124, "train_loss_llm": 0.0, "grad_norm": 0.05329175665974617, "global_step": 27102, "epoch": 752, "lr": 0.001} +{"train_loss": 0.0333922877907753, "train_loss_bc": 0.03337162733078003, "train_loss_llm": 0.0020662248134613037, "grad_norm": 0.07576065510511398, "global_step": 27103, "epoch": 752, "lr": 0.001} +{"train_loss": 0.028354037553071976, "train_loss_bc": 0.02834988199174404, "train_loss_llm": 0.00041559338569641113, "grad_norm": 0.09032117575407028, "global_step": 27104, "epoch": 752, "lr": 0.001} +{"train_loss": 0.02495690993964672, "train_loss_bc": 0.02495110221207142, "train_loss_llm": 0.0005808472633361816, "grad_norm": 0.03330354392528534, "global_step": 27105, "epoch": 752, "lr": 0.001} +{"train_loss": 0.027491608634591103, "train_loss_bc": 0.027491608634591103, "train_loss_llm": 0.0, "grad_norm": 0.04517361894249916, "global_step": 27106, "epoch": 752, "lr": 0.001} +{"train_loss": 0.03052608435973525, "train_loss_bc": 0.024007832631468773, "train_loss_llm": 0.0006559491157531738, "grad_norm": 0.06675231456756592, "global_step": 27107, "epoch": 752, "lr": 0.001, "val_loss": 0.09521780163049698} +{"train_loss": 0.029691051691770554, "train_loss_bc": 0.029691051691770554, "train_loss_llm": 0.0, "grad_norm": 0.09122804552316666, "global_step": 27108, "epoch": 753, "lr": 0.001} +{"train_loss": 0.0270917396992445, "train_loss_bc": 0.0270917396992445, "train_loss_llm": 0.0, "grad_norm": 0.03770502656698227, "global_step": 27109, "epoch": 753, "lr": 0.001} +{"train_loss": 0.030503693968057632, "train_loss_bc": 0.030503693968057632, "train_loss_llm": 0.0, "grad_norm": 0.05822324380278587, "global_step": 27110, "epoch": 753, "lr": 0.001} +{"train_loss": 0.024184804409742355, "train_loss_bc": 0.024184804409742355, "train_loss_llm": 0.0, "grad_norm": 0.08302836120128632, "global_step": 27111, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03211882710456848, "train_loss_bc": 0.03211198374629021, "train_loss_llm": 0.0006844997406005859, "grad_norm": 0.11076639592647552, "global_step": 27112, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03496859222650528, "train_loss_bc": 0.03496859222650528, "train_loss_llm": 0.0, "grad_norm": 0.0376221239566803, "global_step": 27113, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03275167569518089, "train_loss_bc": 0.03275167569518089, "train_loss_llm": 0.0, "grad_norm": 0.042075902223587036, "global_step": 27114, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03000481240451336, "train_loss_bc": 0.03000481240451336, "train_loss_llm": 0.0, "grad_norm": 0.05375256389379501, "global_step": 27115, "epoch": 753, "lr": 0.001} +{"train_loss": 0.030480412766337395, "train_loss_bc": 0.030437864363193512, "train_loss_llm": 0.004254847764968872, "grad_norm": 0.07073906809091568, "global_step": 27116, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03164464980363846, "train_loss_bc": 0.03164464980363846, "train_loss_llm": 0.0, "grad_norm": 0.049069419503211975, "global_step": 27117, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03212970495223999, "train_loss_bc": 0.03212970495223999, "train_loss_llm": 0.0, "grad_norm": 0.062086351215839386, "global_step": 27118, "epoch": 753, "lr": 0.001} +{"train_loss": 0.02775266021490097, "train_loss_bc": 0.027750583365559578, "train_loss_llm": 0.00020775198936462402, "grad_norm": 0.07615336030721664, "global_step": 27119, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03277859091758728, "train_loss_bc": 0.03277859091758728, "train_loss_llm": 0.0, "grad_norm": 0.0842093750834465, "global_step": 27120, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03179996833205223, "train_loss_bc": 0.03179533779621124, "train_loss_llm": 0.0004629194736480713, "grad_norm": 0.03580315038561821, "global_step": 27121, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03159819170832634, "train_loss_bc": 0.03159819170832634, "train_loss_llm": 0.0, "grad_norm": 0.05148980766534805, "global_step": 27122, "epoch": 753, "lr": 0.001} +{"train_loss": 0.025757543742656708, "train_loss_bc": 0.025748956948518753, "train_loss_llm": 0.0008586347103118896, "grad_norm": 0.069393590092659, "global_step": 27123, "epoch": 753, "lr": 0.001} +{"train_loss": 0.028679467737674713, "train_loss_bc": 0.02867523767054081, "train_loss_llm": 0.00042307376861572266, "grad_norm": 0.07665953040122986, "global_step": 27124, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03143489360809326, "train_loss_bc": 0.03143489360809326, "train_loss_llm": 0.0, "grad_norm": 0.03243798390030861, "global_step": 27125, "epoch": 753, "lr": 0.001} +{"train_loss": 0.028575846925377846, "train_loss_bc": 0.028573935851454735, "train_loss_llm": 0.00019109249114990234, "grad_norm": 0.05152237415313721, "global_step": 27126, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03446969762444496, "train_loss_bc": 0.03446969762444496, "train_loss_llm": 0.0, "grad_norm": 0.0749431625008583, "global_step": 27127, "epoch": 753, "lr": 0.001} +{"train_loss": 0.032944515347480774, "train_loss_bc": 0.032944515347480774, "train_loss_llm": 0.0, "grad_norm": 0.10324851423501968, "global_step": 27128, "epoch": 753, "lr": 0.001} +{"train_loss": 0.030772928148508072, "train_loss_bc": 0.030771467834711075, "train_loss_llm": 0.00014594197273254395, "grad_norm": 0.03424665704369545, "global_step": 27129, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03102222830057144, "train_loss_bc": 0.03102222830057144, "train_loss_llm": 0.0, "grad_norm": 0.0532381571829319, "global_step": 27130, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03144560009241104, "train_loss_bc": 0.03144412115216255, "train_loss_llm": 0.00014793872833251953, "grad_norm": 0.06636471301317215, "global_step": 27131, "epoch": 753, "lr": 0.001} +{"train_loss": 0.025246117264032364, "train_loss_bc": 0.025246117264032364, "train_loss_llm": 0.0, "grad_norm": 0.08126495033502579, "global_step": 27132, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03172338753938675, "train_loss_bc": 0.03172338753938675, "train_loss_llm": 0.0, "grad_norm": 0.04823040962219238, "global_step": 27133, "epoch": 753, "lr": 0.001} +{"train_loss": 0.02974216267466545, "train_loss_bc": 0.02974216267466545, "train_loss_llm": 0.0, "grad_norm": 0.06804141402244568, "global_step": 27134, "epoch": 753, "lr": 0.001} +{"train_loss": 0.033986471593379974, "train_loss_bc": 0.033986471593379974, "train_loss_llm": 0.0, "grad_norm": 0.09572741389274597, "global_step": 27135, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03363717719912529, "train_loss_bc": 0.03363717719912529, "train_loss_llm": 0.0, "grad_norm": 0.11577504128217697, "global_step": 27136, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03556401655077934, "train_loss_bc": 0.035560354590415955, "train_loss_llm": 0.0003661811351776123, "grad_norm": 0.04123373329639435, "global_step": 27137, "epoch": 753, "lr": 0.001} +{"train_loss": 0.02988453023135662, "train_loss_bc": 0.02988453023135662, "train_loss_llm": 0.0, "grad_norm": 0.06244499608874321, "global_step": 27138, "epoch": 753, "lr": 0.001} +{"train_loss": 0.027117766439914703, "train_loss_bc": 0.027117766439914703, "train_loss_llm": 0.0, "grad_norm": 0.08217636495828629, "global_step": 27139, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03394013270735741, "train_loss_bc": 0.03394013270735741, "train_loss_llm": 0.0, "grad_norm": 0.09697875380516052, "global_step": 27140, "epoch": 753, "lr": 0.001} +{"train_loss": 0.027385348454117775, "train_loss_bc": 0.027385348454117775, "train_loss_llm": 0.0, "grad_norm": 0.030710114166140556, "global_step": 27141, "epoch": 753, "lr": 0.001} +{"train_loss": 0.03548740968108177, "train_loss_bc": 0.03548740968108177, "train_loss_llm": 0.0, "grad_norm": 0.05225192755460739, "global_step": 27142, "epoch": 753, "lr": 0.001} +{"train_loss": 0.030876811386810407, "train_loss_bc": 0.03324859216809273, "train_loss_llm": 0.0, "grad_norm": 0.0790504515171051, "global_step": 27143, "epoch": 753, "lr": 0.001, "val_loss": 0.09217497706413269} +{"train_loss": 0.029966499656438828, "train_loss_bc": 0.029966499656438828, "train_loss_llm": 0.0, "grad_norm": 0.09249550849199295, "global_step": 27144, "epoch": 754, "lr": 0.001} +{"train_loss": 0.02971702627837658, "train_loss_bc": 0.02971702627837658, "train_loss_llm": 0.0, "grad_norm": 0.03153383731842041, "global_step": 27145, "epoch": 754, "lr": 0.001} +{"train_loss": 0.0294053852558136, "train_loss_bc": 0.0294053852558136, "train_loss_llm": 0.0, "grad_norm": 0.0445609949529171, "global_step": 27146, "epoch": 754, "lr": 0.001} +{"train_loss": 0.029536614194512367, "train_loss_bc": 0.029536614194512367, "train_loss_llm": 0.0, "grad_norm": 0.05669627711176872, "global_step": 27147, "epoch": 754, "lr": 0.001} +{"train_loss": 0.033488526940345764, "train_loss_bc": 0.033488526940345764, "train_loss_llm": 0.0, "grad_norm": 0.07741439342498779, "global_step": 27148, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03309789299964905, "train_loss_bc": 0.03309789299964905, "train_loss_llm": 0.0, "grad_norm": 0.045270051807165146, "global_step": 27149, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03164781257510185, "train_loss_bc": 0.03164781257510185, "train_loss_llm": 0.0, "grad_norm": 0.0677984356880188, "global_step": 27150, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03430483490228653, "train_loss_bc": 0.03430483490228653, "train_loss_llm": 0.0, "grad_norm": 0.09578083455562592, "global_step": 27151, "epoch": 754, "lr": 0.001} +{"train_loss": 0.02941114827990532, "train_loss_bc": 0.02941114827990532, "train_loss_llm": 0.0, "grad_norm": 0.12355269491672516, "global_step": 27152, "epoch": 754, "lr": 0.001} +{"train_loss": 0.02932922914624214, "train_loss_bc": 0.02932922914624214, "train_loss_llm": 0.0, "grad_norm": 0.03834649175405502, "global_step": 27153, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03631768375635147, "train_loss_bc": 0.03631768375635147, "train_loss_llm": 0.0, "grad_norm": 0.07546314597129822, "global_step": 27154, "epoch": 754, "lr": 0.001} +{"train_loss": 0.027992932125926018, "train_loss_bc": 0.027989905327558517, "train_loss_llm": 0.000302731990814209, "grad_norm": 0.08878292888402939, "global_step": 27155, "epoch": 754, "lr": 0.001} +{"train_loss": 0.0278809554874897, "train_loss_bc": 0.0278809554874897, "train_loss_llm": 0.0, "grad_norm": 0.1076936274766922, "global_step": 27156, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03284363076090813, "train_loss_bc": 0.03284363076090813, "train_loss_llm": 0.0, "grad_norm": 0.033013034611940384, "global_step": 27157, "epoch": 754, "lr": 0.001} +{"train_loss": 0.030739758163690567, "train_loss_bc": 0.030735492706298828, "train_loss_llm": 0.0004265904426574707, "grad_norm": 0.05387618765234947, "global_step": 27158, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03572012111544609, "train_loss_bc": 0.03572012111544609, "train_loss_llm": 0.0, "grad_norm": 0.07602342218160629, "global_step": 27159, "epoch": 754, "lr": 0.001} +{"train_loss": 0.027059808373451233, "train_loss_bc": 0.027059808373451233, "train_loss_llm": 0.0, "grad_norm": 0.09295672178268433, "global_step": 27160, "epoch": 754, "lr": 0.001} +{"train_loss": 0.02951912209391594, "train_loss_bc": 0.02951912209391594, "train_loss_llm": 0.0, "grad_norm": 0.02604716829955578, "global_step": 27161, "epoch": 754, "lr": 0.001} +{"train_loss": 0.0254156906157732, "train_loss_bc": 0.0254156906157732, "train_loss_llm": 0.0, "grad_norm": 0.03870688006281853, "global_step": 27162, "epoch": 754, "lr": 0.001} +{"train_loss": 0.032163191586732864, "train_loss_bc": 0.03215441852807999, "train_loss_llm": 0.0008771419525146484, "grad_norm": 0.04771878942847252, "global_step": 27163, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03419838100671768, "train_loss_bc": 0.03419838100671768, "train_loss_llm": 0.0, "grad_norm": 0.0624387152493, "global_step": 27164, "epoch": 754, "lr": 0.001} +{"train_loss": 0.02904924564063549, "train_loss_bc": 0.02904924564063549, "train_loss_llm": 0.0, "grad_norm": 0.031539931893348694, "global_step": 27165, "epoch": 754, "lr": 0.001} +{"train_loss": 0.030222442001104355, "train_loss_bc": 0.030222442001104355, "train_loss_llm": 0.0, "grad_norm": 0.050117362290620804, "global_step": 27166, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03049371764063835, "train_loss_bc": 0.03049371764063835, "train_loss_llm": 0.0, "grad_norm": 0.08134375512599945, "global_step": 27167, "epoch": 754, "lr": 0.001} +{"train_loss": 0.030605480074882507, "train_loss_bc": 0.030605480074882507, "train_loss_llm": 0.0, "grad_norm": 0.09056181460618973, "global_step": 27168, "epoch": 754, "lr": 0.001} +{"train_loss": 0.02744016982614994, "train_loss_bc": 0.027426233515143394, "train_loss_llm": 0.0013937056064605713, "grad_norm": 0.033987391740083694, "global_step": 27169, "epoch": 754, "lr": 0.001} +{"train_loss": 0.02897639013826847, "train_loss_bc": 0.028972718864679337, "train_loss_llm": 0.00036710500717163086, "grad_norm": 0.0547390915453434, "global_step": 27170, "epoch": 754, "lr": 0.001} +{"train_loss": 0.031327877193689346, "train_loss_bc": 0.031327877193689346, "train_loss_llm": 0.0, "grad_norm": 0.07987228035926819, "global_step": 27171, "epoch": 754, "lr": 0.001} +{"train_loss": 0.028822125867009163, "train_loss_bc": 0.028822125867009163, "train_loss_llm": 0.0, "grad_norm": 0.10303717106580734, "global_step": 27172, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03317587450146675, "train_loss_bc": 0.03317023441195488, "train_loss_llm": 0.0005639791488647461, "grad_norm": 0.03747307136654854, "global_step": 27173, "epoch": 754, "lr": 0.001} +{"train_loss": 0.02704356238245964, "train_loss_bc": 0.027033481746912003, "train_loss_llm": 0.0010079741477966309, "grad_norm": 0.05606754124164581, "global_step": 27174, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03365296870470047, "train_loss_bc": 0.03362509608268738, "train_loss_llm": 0.002787083387374878, "grad_norm": 0.07118454575538635, "global_step": 27175, "epoch": 754, "lr": 0.001} +{"train_loss": 0.0347728431224823, "train_loss_bc": 0.03475700318813324, "train_loss_llm": 0.001583874225616455, "grad_norm": 0.08923329412937164, "global_step": 27176, "epoch": 754, "lr": 0.001} +{"train_loss": 0.028940781950950623, "train_loss_bc": 0.028934726491570473, "train_loss_llm": 0.0006056129932403564, "grad_norm": 0.029086971655488014, "global_step": 27177, "epoch": 754, "lr": 0.001} +{"train_loss": 0.03140321746468544, "train_loss_bc": 0.031390782445669174, "train_loss_llm": 0.001243680715560913, "grad_norm": 0.05296243727207184, "global_step": 27178, "epoch": 754, "lr": 0.001} +{"train_loss": 0.030807342276804976, "train_loss_bc": 0.0333813801407814, "train_loss_llm": 0.0, "grad_norm": 0.060008455067873, "global_step": 27179, "epoch": 754, "lr": 0.001, "val_loss": 0.09174215793609619} +{"train_loss": 0.029273992404341698, "train_loss_bc": 0.029273992404341698, "train_loss_llm": 0.0, "grad_norm": 0.07560022175312042, "global_step": 27180, "epoch": 755, "lr": 0.001} +{"train_loss": 0.02821473777294159, "train_loss_bc": 0.02821473777294159, "train_loss_llm": 0.0, "grad_norm": 0.03864862024784088, "global_step": 27181, "epoch": 755, "lr": 0.001} +{"train_loss": 0.034228578209877014, "train_loss_bc": 0.03422544151544571, "train_loss_llm": 0.00031381845474243164, "grad_norm": 0.06376859545707703, "global_step": 27182, "epoch": 755, "lr": 0.001} +{"train_loss": 0.023846523836255074, "train_loss_bc": 0.023841099813580513, "train_loss_llm": 0.0005424916744232178, "grad_norm": 0.07499762624502182, "global_step": 27183, "epoch": 755, "lr": 0.001} +{"train_loss": 0.02918918989598751, "train_loss_bc": 0.02918918989598751, "train_loss_llm": 0.0, "grad_norm": 0.08686905354261398, "global_step": 27184, "epoch": 755, "lr": 0.001} +{"train_loss": 0.035646941512823105, "train_loss_bc": 0.035646941512823105, "train_loss_llm": 0.0, "grad_norm": 0.02987555041909218, "global_step": 27185, "epoch": 755, "lr": 0.001} +{"train_loss": 0.02781994827091694, "train_loss_bc": 0.02781994827091694, "train_loss_llm": 0.0, "grad_norm": 0.04406196251511574, "global_step": 27186, "epoch": 755, "lr": 0.001} +{"train_loss": 0.030120141804218292, "train_loss_bc": 0.030120141804218292, "train_loss_llm": 0.0, "grad_norm": 0.067702516913414, "global_step": 27187, "epoch": 755, "lr": 0.001} +{"train_loss": 0.032471831887960434, "train_loss_bc": 0.032471831887960434, "train_loss_llm": 0.0, "grad_norm": 0.08039291948080063, "global_step": 27188, "epoch": 755, "lr": 0.001} +{"train_loss": 0.029147082939743996, "train_loss_bc": 0.029147082939743996, "train_loss_llm": 0.0, "grad_norm": 0.027176955714821815, "global_step": 27189, "epoch": 755, "lr": 0.001} +{"train_loss": 0.02895457111299038, "train_loss_bc": 0.02895457111299038, "train_loss_llm": 0.0, "grad_norm": 0.03880331292748451, "global_step": 27190, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03490042686462402, "train_loss_bc": 0.03490042686462402, "train_loss_llm": 0.0, "grad_norm": 0.05089208111166954, "global_step": 27191, "epoch": 755, "lr": 0.001} +{"train_loss": 0.02872704155743122, "train_loss_bc": 0.02872704155743122, "train_loss_llm": 0.0, "grad_norm": 0.05967526137828827, "global_step": 27192, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03303658962249756, "train_loss_bc": 0.03303658962249756, "train_loss_llm": 0.0, "grad_norm": 0.0324988029897213, "global_step": 27193, "epoch": 755, "lr": 0.001} +{"train_loss": 0.033167630434036255, "train_loss_bc": 0.033167630434036255, "train_loss_llm": 0.0, "grad_norm": 0.05243724212050438, "global_step": 27194, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03388892859220505, "train_loss_bc": 0.03388751670718193, "train_loss_llm": 0.0001411139965057373, "grad_norm": 0.07067487388849258, "global_step": 27195, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03102613054215908, "train_loss_bc": 0.03102613054215908, "train_loss_llm": 0.0, "grad_norm": 0.08635250478982925, "global_step": 27196, "epoch": 755, "lr": 0.001} +{"train_loss": 0.032272495329380035, "train_loss_bc": 0.032272495329380035, "train_loss_llm": 0.0, "grad_norm": 0.0414414256811142, "global_step": 27197, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03106384351849556, "train_loss_bc": 0.03106384351849556, "train_loss_llm": 0.0, "grad_norm": 0.0618368461728096, "global_step": 27198, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03275410085916519, "train_loss_bc": 0.03275410085916519, "train_loss_llm": 0.0, "grad_norm": 0.07995657622814178, "global_step": 27199, "epoch": 755, "lr": 0.001} +{"train_loss": 0.029376888647675514, "train_loss_bc": 0.029376696795225143, "train_loss_llm": 1.9103288650512695e-05, "grad_norm": 0.09539885818958282, "global_step": 27200, "epoch": 755, "lr": 0.001} +{"train_loss": 0.030701352283358574, "train_loss_bc": 0.030701352283358574, "train_loss_llm": 0.0, "grad_norm": 0.030768416821956635, "global_step": 27201, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03302731737494469, "train_loss_bc": 0.03302731737494469, "train_loss_llm": 0.0, "grad_norm": 0.05015384405851364, "global_step": 27202, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03011161834001541, "train_loss_bc": 0.03011161834001541, "train_loss_llm": 0.0, "grad_norm": 0.06841028481721878, "global_step": 27203, "epoch": 755, "lr": 0.001} +{"train_loss": 0.025609958916902542, "train_loss_bc": 0.025609958916902542, "train_loss_llm": 0.0, "grad_norm": 0.08485674113035202, "global_step": 27204, "epoch": 755, "lr": 0.001} +{"train_loss": 0.030620152130723, "train_loss_bc": 0.030620152130723, "train_loss_llm": 0.0, "grad_norm": 0.03843645751476288, "global_step": 27205, "epoch": 755, "lr": 0.001} +{"train_loss": 0.025594618171453476, "train_loss_bc": 0.025594618171453476, "train_loss_llm": 0.0, "grad_norm": 0.05896484851837158, "global_step": 27206, "epoch": 755, "lr": 0.001} +{"train_loss": 0.02806505374610424, "train_loss_bc": 0.02806505374610424, "train_loss_llm": 0.0, "grad_norm": 0.08155789226293564, "global_step": 27207, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03335057944059372, "train_loss_bc": 0.03335057944059372, "train_loss_llm": 0.0, "grad_norm": 0.09890925884246826, "global_step": 27208, "epoch": 755, "lr": 0.001} +{"train_loss": 0.02500796876847744, "train_loss_bc": 0.02499864622950554, "train_loss_llm": 0.0009323358535766602, "grad_norm": 0.02799031510949135, "global_step": 27209, "epoch": 755, "lr": 0.001} +{"train_loss": 0.030407920479774475, "train_loss_bc": 0.030407920479774475, "train_loss_llm": 0.0, "grad_norm": 0.04088391363620758, "global_step": 27210, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03260338306427002, "train_loss_bc": 0.03260268643498421, "train_loss_llm": 6.949901580810547e-05, "grad_norm": 0.059962887316942215, "global_step": 27211, "epoch": 755, "lr": 0.001} +{"train_loss": 0.037073325365781784, "train_loss_bc": 0.03707244619727135, "train_loss_llm": 8.782744407653809e-05, "grad_norm": 0.07413171231746674, "global_step": 27212, "epoch": 755, "lr": 0.001} +{"train_loss": 0.030235935002565384, "train_loss_bc": 0.030235935002565384, "train_loss_llm": 0.0, "grad_norm": 0.032853394746780396, "global_step": 27213, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03111456148326397, "train_loss_bc": 0.031112462282180786, "train_loss_llm": 0.0002098679542541504, "grad_norm": 0.05373794957995415, "global_step": 27214, "epoch": 755, "lr": 0.001} +{"train_loss": 0.03059636087467273, "train_loss_bc": 0.02881763130426407, "train_loss_llm": 0.0, "grad_norm": 0.06408418715000153, "global_step": 27215, "epoch": 755, "lr": 0.001, "train/cumulative_reward": 113.67821212637666, "train/mean_score": 13.733469624271706, "train/success_rate": 0.2, "test/cumulative_reward": 41.37267201855782, "test/mean_score": 11.086246294823786, "test/success_rate": 0.3, "val_loss": 0.09325502067804337, "train_action_mse_error": 0.03354542702436447} +{"train_loss": 0.025045841932296753, "train_loss_bc": 0.025045841932296753, "train_loss_llm": 0.0, "grad_norm": 0.07895708829164505, "global_step": 27216, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03148488327860832, "train_loss_bc": 0.03148488327860832, "train_loss_llm": 0.0, "grad_norm": 0.03544051572680473, "global_step": 27217, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03529169037938118, "train_loss_bc": 0.03529169037938118, "train_loss_llm": 0.0, "grad_norm": 0.05827828124165535, "global_step": 27218, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03241041302680969, "train_loss_bc": 0.03241041302680969, "train_loss_llm": 0.0, "grad_norm": 0.07710050791501999, "global_step": 27219, "epoch": 756, "lr": 0.001} +{"train_loss": 0.031071720644831657, "train_loss_bc": 0.031071720644831657, "train_loss_llm": 0.0, "grad_norm": 0.08704601973295212, "global_step": 27220, "epoch": 756, "lr": 0.001} +{"train_loss": 0.02747228369116783, "train_loss_bc": 0.02747228369116783, "train_loss_llm": 0.0, "grad_norm": 0.04580928012728691, "global_step": 27221, "epoch": 756, "lr": 0.001} +{"train_loss": 0.028024036437273026, "train_loss_bc": 0.028024036437273026, "train_loss_llm": 0.0, "grad_norm": 0.06622222810983658, "global_step": 27222, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03417351841926575, "train_loss_bc": 0.03417351841926575, "train_loss_llm": 0.0, "grad_norm": 0.08626007288694382, "global_step": 27223, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03373172506690025, "train_loss_bc": 0.03373172506690025, "train_loss_llm": 0.0, "grad_norm": 0.11036847531795502, "global_step": 27224, "epoch": 756, "lr": 0.001} +{"train_loss": 0.028911028057336807, "train_loss_bc": 0.028911028057336807, "train_loss_llm": 0.0, "grad_norm": 0.03282785788178444, "global_step": 27225, "epoch": 756, "lr": 0.001} +{"train_loss": 0.029934804886579514, "train_loss_bc": 0.029934804886579514, "train_loss_llm": 0.0, "grad_norm": 0.05345960706472397, "global_step": 27226, "epoch": 756, "lr": 0.001} +{"train_loss": 0.034902140498161316, "train_loss_bc": 0.034902140498161316, "train_loss_llm": 0.0, "grad_norm": 0.0692133754491806, "global_step": 27227, "epoch": 756, "lr": 0.001} +{"train_loss": 0.029647793620824814, "train_loss_bc": 0.029647793620824814, "train_loss_llm": 0.0, "grad_norm": 0.0866800993680954, "global_step": 27228, "epoch": 756, "lr": 0.001} +{"train_loss": 0.027875732630491257, "train_loss_bc": 0.027875732630491257, "train_loss_llm": 0.0, "grad_norm": 0.03153347969055176, "global_step": 27229, "epoch": 756, "lr": 0.001} +{"train_loss": 0.031107617542147636, "train_loss_bc": 0.031107617542147636, "train_loss_llm": 0.0, "grad_norm": 0.0428180955350399, "global_step": 27230, "epoch": 756, "lr": 0.001} +{"train_loss": 0.030923303216695786, "train_loss_bc": 0.030923303216695786, "train_loss_llm": 0.0, "grad_norm": 0.05984567850828171, "global_step": 27231, "epoch": 756, "lr": 0.001} +{"train_loss": 0.033897243440151215, "train_loss_bc": 0.03388813138008118, "train_loss_llm": 0.0009111166000366211, "grad_norm": 0.07508357614278793, "global_step": 27232, "epoch": 756, "lr": 0.001} +{"train_loss": 0.026369914412498474, "train_loss_bc": 0.026356562972068787, "train_loss_llm": 0.0013350844383239746, "grad_norm": 0.03240644931793213, "global_step": 27233, "epoch": 756, "lr": 0.001} +{"train_loss": 0.028546763584017754, "train_loss_bc": 0.028546763584017754, "train_loss_llm": 0.0, "grad_norm": 0.05059092119336128, "global_step": 27234, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03097875416278839, "train_loss_bc": 0.030976595357060432, "train_loss_llm": 0.00021579861640930176, "grad_norm": 0.06582242250442505, "global_step": 27235, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03353423997759819, "train_loss_bc": 0.03353423997759819, "train_loss_llm": 0.0, "grad_norm": 0.09519431740045547, "global_step": 27236, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03198825567960739, "train_loss_bc": 0.03198825567960739, "train_loss_llm": 0.0, "grad_norm": 0.03643415868282318, "global_step": 27237, "epoch": 756, "lr": 0.001} +{"train_loss": 0.030412787571549416, "train_loss_bc": 0.030412787571549416, "train_loss_llm": 0.0, "grad_norm": 0.04679074138402939, "global_step": 27238, "epoch": 756, "lr": 0.001} +{"train_loss": 0.030857769772410393, "train_loss_bc": 0.030857769772410393, "train_loss_llm": 0.0, "grad_norm": 0.06320903450250626, "global_step": 27239, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03595248982310295, "train_loss_bc": 0.03593764826655388, "train_loss_llm": 0.001484215259552002, "grad_norm": 0.08477386832237244, "global_step": 27240, "epoch": 756, "lr": 0.001} +{"train_loss": 0.02680359221994877, "train_loss_bc": 0.02678559347987175, "train_loss_llm": 0.0017998218536376953, "grad_norm": 0.033522941172122955, "global_step": 27241, "epoch": 756, "lr": 0.001} +{"train_loss": 0.029884031042456627, "train_loss_bc": 0.029880410060286522, "train_loss_llm": 0.00036209821701049805, "grad_norm": 0.043941058218479156, "global_step": 27242, "epoch": 756, "lr": 0.001} +{"train_loss": 0.029705185443162918, "train_loss_bc": 0.029681723564863205, "train_loss_llm": 0.0023460984230041504, "grad_norm": 0.05874599888920784, "global_step": 27243, "epoch": 756, "lr": 0.001} +{"train_loss": 0.030422674492001534, "train_loss_bc": 0.030422674492001534, "train_loss_llm": 0.0, "grad_norm": 0.07056435942649841, "global_step": 27244, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03045816905796528, "train_loss_bc": 0.03045816905796528, "train_loss_llm": 0.0, "grad_norm": 0.034083858132362366, "global_step": 27245, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03638368472456932, "train_loss_bc": 0.03638368472456932, "train_loss_llm": 0.0, "grad_norm": 0.06273715943098068, "global_step": 27246, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03058476746082306, "train_loss_bc": 0.030562691390514374, "train_loss_llm": 0.002207517623901367, "grad_norm": 0.07976052910089493, "global_step": 27247, "epoch": 756, "lr": 0.001} +{"train_loss": 0.03294100984930992, "train_loss_bc": 0.03294100984930992, "train_loss_llm": 0.0, "grad_norm": 0.09384563565254211, "global_step": 27248, "epoch": 756, "lr": 0.001} +{"train_loss": 0.02919759973883629, "train_loss_bc": 0.02919759973883629, "train_loss_llm": 0.0, "grad_norm": 0.02721763402223587, "global_step": 27249, "epoch": 756, "lr": 0.001} +{"train_loss": 0.031722377985715866, "train_loss_bc": 0.031717438250780106, "train_loss_llm": 0.0004937946796417236, "grad_norm": 0.044982828199863434, "global_step": 27250, "epoch": 756, "lr": 0.001} +{"train_loss": 0.030656161821550794, "train_loss_bc": 0.02094617858529091, "train_loss_llm": 0.00258028507232666, "grad_norm": 0.07697407156229019, "global_step": 27251, "epoch": 756, "lr": 0.001, "val_loss": 0.09378643333911896} +{"train_loss": 0.03599543869495392, "train_loss_bc": 0.03599543869495392, "train_loss_llm": 0.0, "grad_norm": 0.09216942638158798, "global_step": 27252, "epoch": 757, "lr": 0.001} +{"train_loss": 0.029249340295791626, "train_loss_bc": 0.02923942543566227, "train_loss_llm": 0.0009914934635162354, "grad_norm": 0.03265923634171486, "global_step": 27253, "epoch": 757, "lr": 0.001} +{"train_loss": 0.025898287072777748, "train_loss_bc": 0.025898287072777748, "train_loss_llm": 0.0, "grad_norm": 0.045915909111499786, "global_step": 27254, "epoch": 757, "lr": 0.001} +{"train_loss": 0.026008542627096176, "train_loss_bc": 0.026008542627096176, "train_loss_llm": 0.0, "grad_norm": 0.07394370436668396, "global_step": 27255, "epoch": 757, "lr": 0.001} +{"train_loss": 0.02632366493344307, "train_loss_bc": 0.02632366493344307, "train_loss_llm": 0.0, "grad_norm": 0.08891666680574417, "global_step": 27256, "epoch": 757, "lr": 0.001} +{"train_loss": 0.031233344227075577, "train_loss_bc": 0.031224805861711502, "train_loss_llm": 0.0008537769317626953, "grad_norm": 0.034966085106134415, "global_step": 27257, "epoch": 757, "lr": 0.001} +{"train_loss": 0.030677057802677155, "train_loss_bc": 0.030677057802677155, "train_loss_llm": 0.0, "grad_norm": 0.05340581387281418, "global_step": 27258, "epoch": 757, "lr": 0.001} +{"train_loss": 0.03126094490289688, "train_loss_bc": 0.03126094490289688, "train_loss_llm": 0.0, "grad_norm": 0.07004354894161224, "global_step": 27259, "epoch": 757, "lr": 0.001} +{"train_loss": 0.03174271434545517, "train_loss_bc": 0.03174271434545517, "train_loss_llm": 0.0, "grad_norm": 0.08895144611597061, "global_step": 27260, "epoch": 757, "lr": 0.001} +{"train_loss": 0.027460362762212753, "train_loss_bc": 0.027460362762212753, "train_loss_llm": 0.0, "grad_norm": 0.03108394518494606, "global_step": 27261, "epoch": 757, "lr": 0.001} +{"train_loss": 0.03360215947031975, "train_loss_bc": 0.03358711674809456, "train_loss_llm": 0.001504272222518921, "grad_norm": 0.05734844133257866, "global_step": 27262, "epoch": 757, "lr": 0.001} +{"train_loss": 0.030838312581181526, "train_loss_bc": 0.030838312581181526, "train_loss_llm": 0.0, "grad_norm": 0.07120789587497711, "global_step": 27263, "epoch": 757, "lr": 0.001} +{"train_loss": 0.032323382794857025, "train_loss_bc": 0.032323382794857025, "train_loss_llm": 0.0, "grad_norm": 0.08550480008125305, "global_step": 27264, "epoch": 757, "lr": 0.001} +{"train_loss": 0.03363518789410591, "train_loss_bc": 0.03363518789410591, "train_loss_llm": 0.0, "grad_norm": 0.04420209303498268, "global_step": 27265, "epoch": 757, "lr": 0.001} +{"train_loss": 0.03141488879919052, "train_loss_bc": 0.03141488879919052, "train_loss_llm": 0.0, "grad_norm": 0.06049986183643341, "global_step": 27266, "epoch": 757, "lr": 0.001} +{"train_loss": 0.031019197776913643, "train_loss_bc": 0.031019197776913643, "train_loss_llm": 0.0, "grad_norm": 0.07899411022663116, "global_step": 27267, "epoch": 757, "lr": 0.001} +{"train_loss": 0.032146062701940536, "train_loss_bc": 0.032144661992788315, "train_loss_llm": 0.00014013051986694336, "grad_norm": 0.09252055734395981, "global_step": 27268, "epoch": 757, "lr": 0.001} +{"train_loss": 0.02574080601334572, "train_loss_bc": 0.02574080601334572, "train_loss_llm": 0.0, "grad_norm": 0.028132809326052666, "global_step": 27269, "epoch": 757, "lr": 0.001} +{"train_loss": 0.03342060372233391, "train_loss_bc": 0.03342060372233391, "train_loss_llm": 0.0, "grad_norm": 0.05241252854466438, "global_step": 27270, "epoch": 757, "lr": 0.001} +{"train_loss": 0.032857462763786316, "train_loss_bc": 0.032857462763786316, "train_loss_llm": 0.0, "grad_norm": 0.07101809233427048, "global_step": 27271, "epoch": 757, "lr": 0.001} +{"train_loss": 0.03369027003645897, "train_loss_bc": 0.03368020057678223, "train_loss_llm": 0.0010069310665130615, "grad_norm": 0.08601784706115723, "global_step": 27272, "epoch": 757, "lr": 0.001} +{"train_loss": 0.029615866020321846, "train_loss_bc": 0.029615866020321846, "train_loss_llm": 0.0, "grad_norm": 0.03417927026748657, "global_step": 27273, "epoch": 757, "lr": 0.001} +{"train_loss": 0.031800080090761185, "train_loss_bc": 0.031800080090761185, "train_loss_llm": 0.0, "grad_norm": 0.0600697323679924, "global_step": 27274, "epoch": 757, "lr": 0.001} +{"train_loss": 0.026335714384913445, "train_loss_bc": 0.026335714384913445, "train_loss_llm": 0.0, "grad_norm": 0.08527675271034241, "global_step": 27275, "epoch": 757, "lr": 0.001} +{"train_loss": 0.033137425780296326, "train_loss_bc": 0.03313330560922623, "train_loss_llm": 0.0004121065139770508, "grad_norm": 0.10241327434778214, "global_step": 27276, "epoch": 757, "lr": 0.001} +{"train_loss": 0.037545207887887955, "train_loss_bc": 0.037545207887887955, "train_loss_llm": 0.0, "grad_norm": 0.03324521705508232, "global_step": 27277, "epoch": 757, "lr": 0.001} +{"train_loss": 0.028770389035344124, "train_loss_bc": 0.028770389035344124, "train_loss_llm": 0.0, "grad_norm": 0.042762406170368195, "global_step": 27278, "epoch": 757, "lr": 0.001} +{"train_loss": 0.029346495866775513, "train_loss_bc": 0.029346495866775513, "train_loss_llm": 0.0, "grad_norm": 0.056148894131183624, "global_step": 27279, "epoch": 757, "lr": 0.001} +{"train_loss": 0.028159048408269882, "train_loss_bc": 0.028159048408269882, "train_loss_llm": 0.0, "grad_norm": 0.06306323409080505, "global_step": 27280, "epoch": 757, "lr": 0.001} +{"train_loss": 0.03134733811020851, "train_loss_bc": 0.03134733811020851, "train_loss_llm": 0.0, "grad_norm": 0.02700289525091648, "global_step": 27281, "epoch": 757, "lr": 0.001} +{"train_loss": 0.030682003125548363, "train_loss_bc": 0.030682003125548363, "train_loss_llm": 0.0, "grad_norm": 0.04070952907204628, "global_step": 27282, "epoch": 757, "lr": 0.001} +{"train_loss": 0.028350505977869034, "train_loss_bc": 0.028348177671432495, "train_loss_llm": 0.00023281574249267578, "grad_norm": 0.06931915134191513, "global_step": 27283, "epoch": 757, "lr": 0.001} +{"train_loss": 0.03530513122677803, "train_loss_bc": 0.03530513122677803, "train_loss_llm": 0.0, "grad_norm": 0.09164486825466156, "global_step": 27284, "epoch": 757, "lr": 0.001} +{"train_loss": 0.031019065529108047, "train_loss_bc": 0.031019065529108047, "train_loss_llm": 0.0, "grad_norm": 0.03874371945858002, "global_step": 27285, "epoch": 757, "lr": 0.001} +{"train_loss": 0.026962583884596825, "train_loss_bc": 0.026962583884596825, "train_loss_llm": 0.0, "grad_norm": 0.05033892020583153, "global_step": 27286, "epoch": 757, "lr": 0.001} +{"train_loss": 0.030777318713565666, "train_loss_bc": 0.03306858614087105, "train_loss_llm": 0.0, "grad_norm": 0.07804600149393082, "global_step": 27287, "epoch": 757, "lr": 0.001, "val_loss": 0.0932898074388504} +{"train_loss": 0.024456985294818878, "train_loss_bc": 0.024456985294818878, "train_loss_llm": 0.0, "grad_norm": 0.0933908149600029, "global_step": 27288, "epoch": 758, "lr": 0.001} +{"train_loss": 0.024025723338127136, "train_loss_bc": 0.024025723338127136, "train_loss_llm": 0.0, "grad_norm": 0.03125164657831192, "global_step": 27289, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03279836103320122, "train_loss_bc": 0.03277897089719772, "train_loss_llm": 0.0019389688968658447, "grad_norm": 0.05246058851480484, "global_step": 27290, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03435918316245079, "train_loss_bc": 0.03435918316245079, "train_loss_llm": 0.0, "grad_norm": 0.06678636372089386, "global_step": 27291, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03383404389023781, "train_loss_bc": 0.033815138041973114, "train_loss_llm": 0.0018907487392425537, "grad_norm": 0.08602772653102875, "global_step": 27292, "epoch": 758, "lr": 0.001} +{"train_loss": 0.02923785150051117, "train_loss_bc": 0.02921169064939022, "train_loss_llm": 0.00261610746383667, "grad_norm": 0.033108871430158615, "global_step": 27293, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03597068041563034, "train_loss_bc": 0.03597068041563034, "train_loss_llm": 0.0, "grad_norm": 0.057269856333732605, "global_step": 27294, "epoch": 758, "lr": 0.001} +{"train_loss": 0.0340389609336853, "train_loss_bc": 0.0340389609336853, "train_loss_llm": 0.0, "grad_norm": 0.0788843035697937, "global_step": 27295, "epoch": 758, "lr": 0.001} +{"train_loss": 0.024022618308663368, "train_loss_bc": 0.024022618308663368, "train_loss_llm": 0.0, "grad_norm": 0.09987249970436096, "global_step": 27296, "epoch": 758, "lr": 0.001} +{"train_loss": 0.029179053381085396, "train_loss_bc": 0.029179053381085396, "train_loss_llm": 0.0, "grad_norm": 0.033788952976465225, "global_step": 27297, "epoch": 758, "lr": 0.001} +{"train_loss": 0.022410979494452477, "train_loss_bc": 0.022410979494452477, "train_loss_llm": 0.0, "grad_norm": 0.052526168525218964, "global_step": 27298, "epoch": 758, "lr": 0.001} +{"train_loss": 0.031076587736606598, "train_loss_bc": 0.031076587736606598, "train_loss_llm": 0.0, "grad_norm": 0.08000815659761429, "global_step": 27299, "epoch": 758, "lr": 0.001} +{"train_loss": 0.034744951874017715, "train_loss_bc": 0.03474065661430359, "train_loss_llm": 0.00042936205863952637, "grad_norm": 0.08969803899526596, "global_step": 27300, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03435877338051796, "train_loss_bc": 0.03435877338051796, "train_loss_llm": 0.0, "grad_norm": 0.03173474594950676, "global_step": 27301, "epoch": 758, "lr": 0.001} +{"train_loss": 0.02637399174273014, "train_loss_bc": 0.02637399174273014, "train_loss_llm": 0.0, "grad_norm": 0.04885469004511833, "global_step": 27302, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03529553860425949, "train_loss_bc": 0.03529553860425949, "train_loss_llm": 0.0, "grad_norm": 0.07842794060707092, "global_step": 27303, "epoch": 758, "lr": 0.001} +{"train_loss": 0.025380540639162064, "train_loss_bc": 0.025380540639162064, "train_loss_llm": 0.0, "grad_norm": 0.08892757445573807, "global_step": 27304, "epoch": 758, "lr": 0.001} +{"train_loss": 0.030693091452121735, "train_loss_bc": 0.030693091452121735, "train_loss_llm": 0.0, "grad_norm": 0.032585300505161285, "global_step": 27305, "epoch": 758, "lr": 0.001} +{"train_loss": 0.034206949174404144, "train_loss_bc": 0.03419662266969681, "train_loss_llm": 0.001032501459121704, "grad_norm": 0.05114220827817917, "global_step": 27306, "epoch": 758, "lr": 0.001} +{"train_loss": 0.028971610590815544, "train_loss_bc": 0.028971610590815544, "train_loss_llm": 0.0, "grad_norm": 0.06454242020845413, "global_step": 27307, "epoch": 758, "lr": 0.001} +{"train_loss": 0.02628607489168644, "train_loss_bc": 0.02628607489168644, "train_loss_llm": 0.0, "grad_norm": 0.07923686504364014, "global_step": 27308, "epoch": 758, "lr": 0.001} +{"train_loss": 0.028417246416211128, "train_loss_bc": 0.028417246416211128, "train_loss_llm": 0.0, "grad_norm": 0.04052145779132843, "global_step": 27309, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03235578164458275, "train_loss_bc": 0.03235578164458275, "train_loss_llm": 0.0, "grad_norm": 0.06510742753744125, "global_step": 27310, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03329696133732796, "train_loss_bc": 0.033281199634075165, "train_loss_llm": 0.0015761852264404297, "grad_norm": 0.08401239663362503, "global_step": 27311, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03244704753160477, "train_loss_bc": 0.03242868557572365, "train_loss_llm": 0.001836329698562622, "grad_norm": 0.10408876091241837, "global_step": 27312, "epoch": 758, "lr": 0.001} +{"train_loss": 0.028075292706489563, "train_loss_bc": 0.028075292706489563, "train_loss_llm": 0.0, "grad_norm": 0.03681638464331627, "global_step": 27313, "epoch": 758, "lr": 0.001} +{"train_loss": 0.026721563190221786, "train_loss_bc": 0.026721563190221786, "train_loss_llm": 0.0, "grad_norm": 0.06319865584373474, "global_step": 27314, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03558659180998802, "train_loss_bc": 0.0355558879673481, "train_loss_llm": 0.003070354461669922, "grad_norm": 0.09383574873209, "global_step": 27315, "epoch": 758, "lr": 0.001} +{"train_loss": 0.02857600525021553, "train_loss_bc": 0.02857600525021553, "train_loss_llm": 0.0, "grad_norm": 0.12383897602558136, "global_step": 27316, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03816766291856766, "train_loss_bc": 0.03816766291856766, "train_loss_llm": 0.0, "grad_norm": 0.034131646156311035, "global_step": 27317, "epoch": 758, "lr": 0.001} +{"train_loss": 0.031037230044603348, "train_loss_bc": 0.031037230044603348, "train_loss_llm": 0.0, "grad_norm": 0.048602279275655746, "global_step": 27318, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03656907007098198, "train_loss_bc": 0.03656907007098198, "train_loss_llm": 0.0, "grad_norm": 0.07740273326635361, "global_step": 27319, "epoch": 758, "lr": 0.001} +{"train_loss": 0.028793781995773315, "train_loss_bc": 0.028793781995773315, "train_loss_llm": 0.0, "grad_norm": 0.09287716448307037, "global_step": 27320, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03415249288082123, "train_loss_bc": 0.03414276987314224, "train_loss_llm": 0.0009722411632537842, "grad_norm": 0.03807631880044937, "global_step": 27321, "epoch": 758, "lr": 0.001} +{"train_loss": 0.0300426222383976, "train_loss_bc": 0.0300426222383976, "train_loss_llm": 0.0, "grad_norm": 0.05131139978766441, "global_step": 27322, "epoch": 758, "lr": 0.001} +{"train_loss": 0.03083308594715264, "train_loss_bc": 0.034029193222522736, "train_loss_llm": 0.0, "grad_norm": 0.07457994669675827, "global_step": 27323, "epoch": 758, "lr": 0.001, "val_loss": 0.09280754625797272} +{"train_loss": 0.02994959056377411, "train_loss_bc": 0.02994959056377411, "train_loss_llm": 0.0, "grad_norm": 0.09138357639312744, "global_step": 27324, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03484431654214859, "train_loss_bc": 0.03484431654214859, "train_loss_llm": 0.0, "grad_norm": 0.04327549412846565, "global_step": 27325, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03077573888003826, "train_loss_bc": 0.030761385336518288, "train_loss_llm": 0.001435399055480957, "grad_norm": 0.06572513282299042, "global_step": 27326, "epoch": 759, "lr": 0.001} +{"train_loss": 0.030898744240403175, "train_loss_bc": 0.030898744240403175, "train_loss_llm": 0.0, "grad_norm": 0.07190023362636566, "global_step": 27327, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03135009482502937, "train_loss_bc": 0.03135009482502937, "train_loss_llm": 0.0, "grad_norm": 0.07900100201368332, "global_step": 27328, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03417668864130974, "train_loss_bc": 0.03417668864130974, "train_loss_llm": 0.0, "grad_norm": 0.04463138431310654, "global_step": 27329, "epoch": 759, "lr": 0.001} +{"train_loss": 0.02771056443452835, "train_loss_bc": 0.027704993262887, "train_loss_llm": 0.0005570650100708008, "grad_norm": 0.0703573077917099, "global_step": 27330, "epoch": 759, "lr": 0.001} +{"train_loss": 0.028039153665304184, "train_loss_bc": 0.028039153665304184, "train_loss_llm": 0.0, "grad_norm": 0.09363927692174911, "global_step": 27331, "epoch": 759, "lr": 0.001} +{"train_loss": 0.032161734998226166, "train_loss_bc": 0.032161734998226166, "train_loss_llm": 0.0, "grad_norm": 0.11529140919446945, "global_step": 27332, "epoch": 759, "lr": 0.001} +{"train_loss": 0.030062345787882805, "train_loss_bc": 0.03005553036928177, "train_loss_llm": 0.0006815791130065918, "grad_norm": 0.03772524371743202, "global_step": 27333, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03654082864522934, "train_loss_bc": 0.03654082864522934, "train_loss_llm": 0.0, "grad_norm": 0.07349567115306854, "global_step": 27334, "epoch": 759, "lr": 0.001} +{"train_loss": 0.024120952934026718, "train_loss_bc": 0.024107083678245544, "train_loss_llm": 0.001386880874633789, "grad_norm": 0.09346345067024231, "global_step": 27335, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03164060041308403, "train_loss_bc": 0.03164060041308403, "train_loss_llm": 0.0, "grad_norm": 0.12494359165430069, "global_step": 27336, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03157372772693634, "train_loss_bc": 0.03157372772693634, "train_loss_llm": 0.0, "grad_norm": 0.040833793580532074, "global_step": 27337, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03512885421514511, "train_loss_bc": 0.0351286344230175, "train_loss_llm": 2.1845102310180664e-05, "grad_norm": 0.06347841769456863, "global_step": 27338, "epoch": 759, "lr": 0.001} +{"train_loss": 0.033625103533267975, "train_loss_bc": 0.033625103533267975, "train_loss_llm": 0.0, "grad_norm": 0.08714014291763306, "global_step": 27339, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03001079335808754, "train_loss_bc": 0.03001079335808754, "train_loss_llm": 0.0, "grad_norm": 0.110218346118927, "global_step": 27340, "epoch": 759, "lr": 0.001} +{"train_loss": 0.026889128610491753, "train_loss_bc": 0.026863083243370056, "train_loss_llm": 0.0026046037673950195, "grad_norm": 0.0345342680811882, "global_step": 27341, "epoch": 759, "lr": 0.001} +{"train_loss": 0.029540421441197395, "train_loss_bc": 0.029540421441197395, "train_loss_llm": 0.0, "grad_norm": 0.05422479286789894, "global_step": 27342, "epoch": 759, "lr": 0.001} +{"train_loss": 0.034077875316143036, "train_loss_bc": 0.0340706929564476, "train_loss_llm": 0.0007183253765106201, "grad_norm": 0.06578753888607025, "global_step": 27343, "epoch": 759, "lr": 0.001} +{"train_loss": 0.033681947737932205, "train_loss_bc": 0.033681947737932205, "train_loss_llm": 0.0, "grad_norm": 0.0949309915304184, "global_step": 27344, "epoch": 759, "lr": 0.001} +{"train_loss": 0.030236193910241127, "train_loss_bc": 0.030236193910241127, "train_loss_llm": 0.0, "grad_norm": 0.03225426748394966, "global_step": 27345, "epoch": 759, "lr": 0.001} +{"train_loss": 0.028476644307374954, "train_loss_bc": 0.028476644307374954, "train_loss_llm": 0.0, "grad_norm": 0.04538823291659355, "global_step": 27346, "epoch": 759, "lr": 0.001} +{"train_loss": 0.02654336951673031, "train_loss_bc": 0.02654336951673031, "train_loss_llm": 0.0, "grad_norm": 0.059970639646053314, "global_step": 27347, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03227273002266884, "train_loss_bc": 0.03227273002266884, "train_loss_llm": 0.0, "grad_norm": 0.07583051174879074, "global_step": 27348, "epoch": 759, "lr": 0.001} +{"train_loss": 0.02473151683807373, "train_loss_bc": 0.02473151683807373, "train_loss_llm": 0.0, "grad_norm": 0.029096558690071106, "global_step": 27349, "epoch": 759, "lr": 0.001} +{"train_loss": 0.027247795835137367, "train_loss_bc": 0.027247490361332893, "train_loss_llm": 3.045797348022461e-05, "grad_norm": 0.044590312987565994, "global_step": 27350, "epoch": 759, "lr": 0.001} +{"train_loss": 0.026309553533792496, "train_loss_bc": 0.026309553533792496, "train_loss_llm": 0.0, "grad_norm": 0.05925602838397026, "global_step": 27351, "epoch": 759, "lr": 0.001} +{"train_loss": 0.028510548174381256, "train_loss_bc": 0.028493419289588928, "train_loss_llm": 0.0017129182815551758, "grad_norm": 0.08594892919063568, "global_step": 27352, "epoch": 759, "lr": 0.001} +{"train_loss": 0.028501786291599274, "train_loss_bc": 0.028501786291599274, "train_loss_llm": 0.0, "grad_norm": 0.03676624968647957, "global_step": 27353, "epoch": 759, "lr": 0.001} +{"train_loss": 0.030088895931839943, "train_loss_bc": 0.030088895931839943, "train_loss_llm": 0.0, "grad_norm": 0.06947511434555054, "global_step": 27354, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03663221001625061, "train_loss_bc": 0.03661871328949928, "train_loss_llm": 0.0013496577739715576, "grad_norm": 0.10950405150651932, "global_step": 27355, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03450923040509224, "train_loss_bc": 0.03450923040509224, "train_loss_llm": 0.0, "grad_norm": 0.14507217705249786, "global_step": 27356, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03130432218313217, "train_loss_bc": 0.03130432218313217, "train_loss_llm": 0.0, "grad_norm": 0.04186626151204109, "global_step": 27357, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03465377539396286, "train_loss_bc": 0.03465377539396286, "train_loss_llm": 0.0, "grad_norm": 0.06944011896848679, "global_step": 27358, "epoch": 759, "lr": 0.001} +{"train_loss": 0.03093534413104256, "train_loss_bc": 0.03685460984706879, "train_loss_llm": 0.0, "grad_norm": 0.0982070192694664, "global_step": 27359, "epoch": 759, "lr": 0.001, "val_loss": 0.09319566190242767} +{"train_loss": 0.03213447704911232, "train_loss_bc": 0.03212370723485947, "train_loss_llm": 0.0010771453380584717, "grad_norm": 0.12088459730148315, "global_step": 27360, "epoch": 760, "lr": 0.001} +{"train_loss": 0.033603668212890625, "train_loss_bc": 0.033603668212890625, "train_loss_llm": 0.0, "grad_norm": 0.04073958843946457, "global_step": 27361, "epoch": 760, "lr": 0.001} +{"train_loss": 0.030571481212973595, "train_loss_bc": 0.03056085668504238, "train_loss_llm": 0.0010623931884765625, "grad_norm": 0.06097816303372383, "global_step": 27362, "epoch": 760, "lr": 0.001} +{"train_loss": 0.030663058161735535, "train_loss_bc": 0.030663058161735535, "train_loss_llm": 0.0, "grad_norm": 0.07452071458101273, "global_step": 27363, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03223699331283569, "train_loss_bc": 0.03223699331283569, "train_loss_llm": 0.0, "grad_norm": 0.08635979890823364, "global_step": 27364, "epoch": 760, "lr": 0.001} +{"train_loss": 0.026362275704741478, "train_loss_bc": 0.026362275704741478, "train_loss_llm": 0.0, "grad_norm": 0.03397701680660248, "global_step": 27365, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03222454711794853, "train_loss_bc": 0.03222454711794853, "train_loss_llm": 0.0, "grad_norm": 0.061932213604450226, "global_step": 27366, "epoch": 760, "lr": 0.001} +{"train_loss": 0.02940778061747551, "train_loss_bc": 0.02940778061747551, "train_loss_llm": 0.0, "grad_norm": 0.09593795239925385, "global_step": 27367, "epoch": 760, "lr": 0.001} +{"train_loss": 0.02766214683651924, "train_loss_bc": 0.02766214683651924, "train_loss_llm": 0.0, "grad_norm": 0.11503476649522781, "global_step": 27368, "epoch": 760, "lr": 0.001} +{"train_loss": 0.023388205096125603, "train_loss_bc": 0.023388205096125603, "train_loss_llm": 0.0, "grad_norm": 0.030847040936350822, "global_step": 27369, "epoch": 760, "lr": 0.001} +{"train_loss": 0.0264761783182621, "train_loss_bc": 0.026472924277186394, "train_loss_llm": 0.0003253519535064697, "grad_norm": 0.06758272647857666, "global_step": 27370, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03204920142889023, "train_loss_bc": 0.03204920142889023, "train_loss_llm": 0.0, "grad_norm": 0.1018073558807373, "global_step": 27371, "epoch": 760, "lr": 0.001} +{"train_loss": 0.034144483506679535, "train_loss_bc": 0.034144483506679535, "train_loss_llm": 0.0, "grad_norm": 0.11169011890888214, "global_step": 27372, "epoch": 760, "lr": 0.001} +{"train_loss": 0.026490654796361923, "train_loss_bc": 0.02647969499230385, "train_loss_llm": 0.0010959804058074951, "grad_norm": 0.02558796852827072, "global_step": 27373, "epoch": 760, "lr": 0.001} +{"train_loss": 0.030572835355997086, "train_loss_bc": 0.030572835355997086, "train_loss_llm": 0.0, "grad_norm": 0.048463620245456696, "global_step": 27374, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03120173141360283, "train_loss_bc": 0.03120173141360283, "train_loss_llm": 0.0, "grad_norm": 0.0643451139330864, "global_step": 27375, "epoch": 760, "lr": 0.001} +{"train_loss": 0.0360228568315506, "train_loss_bc": 0.0360228568315506, "train_loss_llm": 0.0, "grad_norm": 0.08983064442873001, "global_step": 27376, "epoch": 760, "lr": 0.001} +{"train_loss": 0.029240207746624947, "train_loss_bc": 0.02923532761633396, "train_loss_llm": 0.00048795342445373535, "grad_norm": 0.037176117300987244, "global_step": 27377, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03320992365479469, "train_loss_bc": 0.03320992365479469, "train_loss_llm": 0.0, "grad_norm": 0.058925457298755646, "global_step": 27378, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03598029538989067, "train_loss_bc": 0.03598029538989067, "train_loss_llm": 0.0, "grad_norm": 0.07445404678583145, "global_step": 27379, "epoch": 760, "lr": 0.001} +{"train_loss": 0.0323319137096405, "train_loss_bc": 0.0323319137096405, "train_loss_llm": 0.0, "grad_norm": 0.09693606942892075, "global_step": 27380, "epoch": 760, "lr": 0.001} +{"train_loss": 0.0302373506128788, "train_loss_bc": 0.0302373506128788, "train_loss_llm": 0.0, "grad_norm": 0.03276335448026657, "global_step": 27381, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03223889321088791, "train_loss_bc": 0.03223889321088791, "train_loss_llm": 0.0, "grad_norm": 0.05822914093732834, "global_step": 27382, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03535142168402672, "train_loss_bc": 0.03535142168402672, "train_loss_llm": 0.0, "grad_norm": 0.08379247039556503, "global_step": 27383, "epoch": 760, "lr": 0.001} +{"train_loss": 0.024693768471479416, "train_loss_bc": 0.02468736842274666, "train_loss_llm": 0.0006399750709533691, "grad_norm": 0.11160304397344589, "global_step": 27384, "epoch": 760, "lr": 0.001} +{"train_loss": 0.027568131685256958, "train_loss_bc": 0.02755255065858364, "train_loss_llm": 0.0015580952167510986, "grad_norm": 0.03144033998250961, "global_step": 27385, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03822117671370506, "train_loss_bc": 0.03822117671370506, "train_loss_llm": 0.0, "grad_norm": 0.05354240909218788, "global_step": 27386, "epoch": 760, "lr": 0.001} +{"train_loss": 0.026356300339102745, "train_loss_bc": 0.026356300339102745, "train_loss_llm": 0.0, "grad_norm": 0.08079204708337784, "global_step": 27387, "epoch": 760, "lr": 0.001} +{"train_loss": 0.026163166388869286, "train_loss_bc": 0.026163166388869286, "train_loss_llm": 0.0, "grad_norm": 0.10876983404159546, "global_step": 27388, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03169253095984459, "train_loss_bc": 0.03167892247438431, "train_loss_llm": 0.0013608336448669434, "grad_norm": 0.03018317185342312, "global_step": 27389, "epoch": 760, "lr": 0.001} +{"train_loss": 0.030923936516046524, "train_loss_bc": 0.030923936516046524, "train_loss_llm": 0.0, "grad_norm": 0.06159859150648117, "global_step": 27390, "epoch": 760, "lr": 0.001} +{"train_loss": 0.033017102628946304, "train_loss_bc": 0.03301291912794113, "train_loss_llm": 0.0004183948040008545, "grad_norm": 0.08291669934988022, "global_step": 27391, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03159281238913536, "train_loss_bc": 0.03159281238913536, "train_loss_llm": 0.0, "grad_norm": 0.1049891859292984, "global_step": 27392, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03446513041853905, "train_loss_bc": 0.03446513041853905, "train_loss_llm": 0.0, "grad_norm": 0.03506961837410927, "global_step": 27393, "epoch": 760, "lr": 0.001} +{"train_loss": 0.03482187166810036, "train_loss_bc": 0.03480770066380501, "train_loss_llm": 0.0014169514179229736, "grad_norm": 0.04336641728878021, "global_step": 27394, "epoch": 760, "lr": 0.001} +{"train_loss": 0.030770464127676353, "train_loss_bc": 0.024418199434876442, "train_loss_llm": 0.0, "grad_norm": 0.05797777324914932, "global_step": 27395, "epoch": 760, "lr": 0.001, "train/cumulative_reward": 56.01342690685864, "train/mean_score": 6.448055663505462, "train/success_rate": 0.1, "test/cumulative_reward": 23.3497426208411, "test/mean_score": 6.53848123110537, "test/success_rate": 0.16, "val_loss": 0.09311772137880325, "train_action_mse_error": 0.03387174755334854} +{"train_loss": 0.031071139499545097, "train_loss_bc": 0.031071139499545097, "train_loss_llm": 0.0, "grad_norm": 0.07785285264253616, "global_step": 27396, "epoch": 761, "lr": 0.001} +{"train_loss": 0.026585768908262253, "train_loss_bc": 0.026569396257400513, "train_loss_llm": 0.0016373395919799805, "grad_norm": 0.037102144211530685, "global_step": 27397, "epoch": 761, "lr": 0.001} +{"train_loss": 0.030167637392878532, "train_loss_bc": 0.030167637392878532, "train_loss_llm": 0.0, "grad_norm": 0.045256681740283966, "global_step": 27398, "epoch": 761, "lr": 0.001} +{"train_loss": 0.030818907544016838, "train_loss_bc": 0.03079920820891857, "train_loss_llm": 0.0019698739051818848, "grad_norm": 0.05345143750309944, "global_step": 27399, "epoch": 761, "lr": 0.001} +{"train_loss": 0.029736170545220375, "train_loss_bc": 0.029736170545220375, "train_loss_llm": 0.0, "grad_norm": 0.0680883452296257, "global_step": 27400, "epoch": 761, "lr": 0.001} +{"train_loss": 0.030887220054864883, "train_loss_bc": 0.030882710590958595, "train_loss_llm": 0.00045096874237060547, "grad_norm": 0.0371289998292923, "global_step": 27401, "epoch": 761, "lr": 0.001} +{"train_loss": 0.029470544308423996, "train_loss_bc": 0.029458362609148026, "train_loss_llm": 0.0012182295322418213, "grad_norm": 0.054282404482364655, "global_step": 27402, "epoch": 761, "lr": 0.001} +{"train_loss": 0.02984648197889328, "train_loss_bc": 0.02984648197889328, "train_loss_llm": 0.0, "grad_norm": 0.07362066954374313, "global_step": 27403, "epoch": 761, "lr": 0.001} +{"train_loss": 0.028176860883831978, "train_loss_bc": 0.028176860883831978, "train_loss_llm": 0.0, "grad_norm": 0.08485332131385803, "global_step": 27404, "epoch": 761, "lr": 0.001} +{"train_loss": 0.02930973470211029, "train_loss_bc": 0.02930973470211029, "train_loss_llm": 0.0, "grad_norm": 0.03691345453262329, "global_step": 27405, "epoch": 761, "lr": 0.001} +{"train_loss": 0.027036579325795174, "train_loss_bc": 0.027033613994717598, "train_loss_llm": 0.00029647350311279297, "grad_norm": 0.0625881627202034, "global_step": 27406, "epoch": 761, "lr": 0.001} +{"train_loss": 0.03442041948437691, "train_loss_bc": 0.03442041948437691, "train_loss_llm": 0.0, "grad_norm": 0.08157715946435928, "global_step": 27407, "epoch": 761, "lr": 0.001} +{"train_loss": 0.0298618171364069, "train_loss_bc": 0.02985444851219654, "train_loss_llm": 0.0007368624210357666, "grad_norm": 0.11034414172172546, "global_step": 27408, "epoch": 761, "lr": 0.001} +{"train_loss": 0.03624127805233002, "train_loss_bc": 0.03624127805233002, "train_loss_llm": 0.0, "grad_norm": 0.04185562953352928, "global_step": 27409, "epoch": 761, "lr": 0.001} +{"train_loss": 0.039236146956682205, "train_loss_bc": 0.039236146956682205, "train_loss_llm": 0.0, "grad_norm": 0.07994286715984344, "global_step": 27410, "epoch": 761, "lr": 0.001} +{"train_loss": 0.027271557599306107, "train_loss_bc": 0.027271557599306107, "train_loss_llm": 0.0, "grad_norm": 0.10361215472221375, "global_step": 27411, "epoch": 761, "lr": 0.001} +{"train_loss": 0.026313286274671555, "train_loss_bc": 0.026313286274671555, "train_loss_llm": 0.0, "grad_norm": 0.12052319943904877, "global_step": 27412, "epoch": 761, "lr": 0.001} +{"train_loss": 0.02737041562795639, "train_loss_bc": 0.027366336435079575, "train_loss_llm": 0.00040790438652038574, "grad_norm": 0.036773160099983215, "global_step": 27413, "epoch": 761, "lr": 0.001} +{"train_loss": 0.030939802527427673, "train_loss_bc": 0.030939802527427673, "train_loss_llm": 0.0, "grad_norm": 0.052763693034648895, "global_step": 27414, "epoch": 761, "lr": 0.001} +{"train_loss": 0.028458978980779648, "train_loss_bc": 0.028458978980779648, "train_loss_llm": 0.0, "grad_norm": 0.06606501340866089, "global_step": 27415, "epoch": 761, "lr": 0.001} +{"train_loss": 0.031662505120038986, "train_loss_bc": 0.03166133537888527, "train_loss_llm": 0.0001170039176940918, "grad_norm": 0.07466661930084229, "global_step": 27416, "epoch": 761, "lr": 0.001} +{"train_loss": 0.03451136499643326, "train_loss_bc": 0.03451136499643326, "train_loss_llm": 0.0, "grad_norm": 0.03286770358681679, "global_step": 27417, "epoch": 761, "lr": 0.001} +{"train_loss": 0.04032004624605179, "train_loss_bc": 0.04032004624605179, "train_loss_llm": 0.0, "grad_norm": 0.047026995569467545, "global_step": 27418, "epoch": 761, "lr": 0.001} +{"train_loss": 0.03283030539751053, "train_loss_bc": 0.03283030539751053, "train_loss_llm": 0.0, "grad_norm": 0.06187978759407997, "global_step": 27419, "epoch": 761, "lr": 0.001} +{"train_loss": 0.026379169896245003, "train_loss_bc": 0.026379169896245003, "train_loss_llm": 0.0, "grad_norm": 0.0729154571890831, "global_step": 27420, "epoch": 761, "lr": 0.001} +{"train_loss": 0.03265051171183586, "train_loss_bc": 0.03264768794178963, "train_loss_llm": 0.0002824366092681885, "grad_norm": 0.0360325425863266, "global_step": 27421, "epoch": 761, "lr": 0.001} +{"train_loss": 0.028946181759238243, "train_loss_bc": 0.028946181759238243, "train_loss_llm": 0.0, "grad_norm": 0.047837745398283005, "global_step": 27422, "epoch": 761, "lr": 0.001} +{"train_loss": 0.031050389632582664, "train_loss_bc": 0.031050389632582664, "train_loss_llm": 0.0, "grad_norm": 0.07666045427322388, "global_step": 27423, "epoch": 761, "lr": 0.001} +{"train_loss": 0.03741532936692238, "train_loss_bc": 0.03741272911429405, "train_loss_llm": 0.0002600550651550293, "grad_norm": 0.0931268259882927, "global_step": 27424, "epoch": 761, "lr": 0.001} +{"train_loss": 0.024860017001628876, "train_loss_bc": 0.024860017001628876, "train_loss_llm": 0.0, "grad_norm": 0.03127729147672653, "global_step": 27425, "epoch": 761, "lr": 0.001} +{"train_loss": 0.028655875474214554, "train_loss_bc": 0.028655875474214554, "train_loss_llm": 0.0, "grad_norm": 0.05646712705492973, "global_step": 27426, "epoch": 761, "lr": 0.001} +{"train_loss": 0.02968905121088028, "train_loss_bc": 0.02968905121088028, "train_loss_llm": 0.0, "grad_norm": 0.08914157748222351, "global_step": 27427, "epoch": 761, "lr": 0.001} +{"train_loss": 0.028928883373737335, "train_loss_bc": 0.028928883373737335, "train_loss_llm": 0.0, "grad_norm": 0.11592642962932587, "global_step": 27428, "epoch": 761, "lr": 0.001} +{"train_loss": 0.03569291532039642, "train_loss_bc": 0.03569291532039642, "train_loss_llm": 0.0, "grad_norm": 0.04301095008850098, "global_step": 27429, "epoch": 761, "lr": 0.001} +{"train_loss": 0.03141632676124573, "train_loss_bc": 0.03141632676124573, "train_loss_llm": 0.0, "grad_norm": 0.06369093805551529, "global_step": 27430, "epoch": 761, "lr": 0.001} +{"train_loss": 0.030800972133874893, "train_loss_bc": 0.03059542365372181, "train_loss_llm": 0.0009951889514923096, "grad_norm": 0.09459413588047028, "global_step": 27431, "epoch": 761, "lr": 0.001, "val_loss": 0.09569618850946426} +{"train_loss": 0.028454571962356567, "train_loss_bc": 0.028454571962356567, "train_loss_llm": 0.0, "grad_norm": 0.11008501052856445, "global_step": 27432, "epoch": 762, "lr": 0.001} +{"train_loss": 0.028211193159222603, "train_loss_bc": 0.028211193159222603, "train_loss_llm": 0.0, "grad_norm": 0.03418067842721939, "global_step": 27433, "epoch": 762, "lr": 0.001} +{"train_loss": 0.028889086097478867, "train_loss_bc": 0.028889086097478867, "train_loss_llm": 0.0, "grad_norm": 0.044929616153240204, "global_step": 27434, "epoch": 762, "lr": 0.001} +{"train_loss": 0.0300204586237669, "train_loss_bc": 0.0300116129219532, "train_loss_llm": 0.0008846521377563477, "grad_norm": 0.06514621526002884, "global_step": 27435, "epoch": 762, "lr": 0.001} +{"train_loss": 0.030973222106695175, "train_loss_bc": 0.030973222106695175, "train_loss_llm": 0.0, "grad_norm": 0.07317359000444412, "global_step": 27436, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03487451747059822, "train_loss_bc": 0.03486825153231621, "train_loss_llm": 0.0006266236305236816, "grad_norm": 0.028917018324136734, "global_step": 27437, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03907250985503197, "train_loss_bc": 0.03907250985503197, "train_loss_llm": 0.0, "grad_norm": 0.04045246168971062, "global_step": 27438, "epoch": 762, "lr": 0.001} +{"train_loss": 0.02350355125963688, "train_loss_bc": 0.02350355125963688, "train_loss_llm": 0.0, "grad_norm": 0.05098818987607956, "global_step": 27439, "epoch": 762, "lr": 0.001} +{"train_loss": 0.028416892513632774, "train_loss_bc": 0.028416892513632774, "train_loss_llm": 0.0, "grad_norm": 0.06908253580331802, "global_step": 27440, "epoch": 762, "lr": 0.001} +{"train_loss": 0.024169988930225372, "train_loss_bc": 0.024156764149665833, "train_loss_llm": 0.0013224482536315918, "grad_norm": 0.026075756177306175, "global_step": 27441, "epoch": 762, "lr": 0.001} +{"train_loss": 0.032180532813072205, "train_loss_bc": 0.03217681124806404, "train_loss_llm": 0.00037217140197753906, "grad_norm": 0.03824843838810921, "global_step": 27442, "epoch": 762, "lr": 0.001} +{"train_loss": 0.0345289371907711, "train_loss_bc": 0.034528203308582306, "train_loss_llm": 7.355213165283203e-05, "grad_norm": 0.06022365391254425, "global_step": 27443, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03306088596582413, "train_loss_bc": 0.03306088596582413, "train_loss_llm": 0.0, "grad_norm": 0.0749121829867363, "global_step": 27444, "epoch": 762, "lr": 0.001} +{"train_loss": 0.027352945879101753, "train_loss_bc": 0.027352945879101753, "train_loss_llm": 0.0, "grad_norm": 0.02851404808461666, "global_step": 27445, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03414565324783325, "train_loss_bc": 0.03414565324783325, "train_loss_llm": 0.0, "grad_norm": 0.04611529782414436, "global_step": 27446, "epoch": 762, "lr": 0.001} +{"train_loss": 0.0297402273863554, "train_loss_bc": 0.0297402273863554, "train_loss_llm": 0.0, "grad_norm": 0.06368652731180191, "global_step": 27447, "epoch": 762, "lr": 0.001} +{"train_loss": 0.026895634829998016, "train_loss_bc": 0.026895634829998016, "train_loss_llm": 0.0, "grad_norm": 0.08870794624090195, "global_step": 27448, "epoch": 762, "lr": 0.001} +{"train_loss": 0.025206230580806732, "train_loss_bc": 0.025206230580806732, "train_loss_llm": 0.0, "grad_norm": 0.037833236157894135, "global_step": 27449, "epoch": 762, "lr": 0.001} +{"train_loss": 0.030731908977031708, "train_loss_bc": 0.030712340027093887, "train_loss_llm": 0.0019568800926208496, "grad_norm": 0.05761298909783363, "global_step": 27450, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03356664627790451, "train_loss_bc": 0.03356664627790451, "train_loss_llm": 0.0, "grad_norm": 0.08006349951028824, "global_step": 27451, "epoch": 762, "lr": 0.001} +{"train_loss": 0.031836267560720444, "train_loss_bc": 0.031836267560720444, "train_loss_llm": 0.0, "grad_norm": 0.11488796770572662, "global_step": 27452, "epoch": 762, "lr": 0.001} +{"train_loss": 0.029547283425927162, "train_loss_bc": 0.029547283425927162, "train_loss_llm": 0.0, "grad_norm": 0.02553672343492508, "global_step": 27453, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03235974907875061, "train_loss_bc": 0.032348476350307465, "train_loss_llm": 0.0011274516582489014, "grad_norm": 0.05362687632441521, "global_step": 27454, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03478477895259857, "train_loss_bc": 0.03478477895259857, "train_loss_llm": 0.0, "grad_norm": 0.07700259983539581, "global_step": 27455, "epoch": 762, "lr": 0.001} +{"train_loss": 0.029291916638612747, "train_loss_bc": 0.029291916638612747, "train_loss_llm": 0.0, "grad_norm": 0.0937562957406044, "global_step": 27456, "epoch": 762, "lr": 0.001} +{"train_loss": 0.032924823462963104, "train_loss_bc": 0.032924823462963104, "train_loss_llm": 0.0, "grad_norm": 0.03732012212276459, "global_step": 27457, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03086675889790058, "train_loss_bc": 0.03084806352853775, "train_loss_llm": 0.0018696188926696777, "grad_norm": 0.04850190877914429, "global_step": 27458, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03454194217920303, "train_loss_bc": 0.034524619579315186, "train_loss_llm": 0.0017322897911071777, "grad_norm": 0.06951314955949783, "global_step": 27459, "epoch": 762, "lr": 0.001} +{"train_loss": 0.026463164016604424, "train_loss_bc": 0.026457084342837334, "train_loss_llm": 0.0006078779697418213, "grad_norm": 0.08658262342214584, "global_step": 27460, "epoch": 762, "lr": 0.001} +{"train_loss": 0.028107155114412308, "train_loss_bc": 0.028107155114412308, "train_loss_llm": 0.0, "grad_norm": 0.03892222046852112, "global_step": 27461, "epoch": 762, "lr": 0.001} +{"train_loss": 0.0331963449716568, "train_loss_bc": 0.03319492191076279, "train_loss_llm": 0.00014230608940124512, "grad_norm": 0.06511251628398895, "global_step": 27462, "epoch": 762, "lr": 0.001} +{"train_loss": 0.029999908059835434, "train_loss_bc": 0.029999908059835434, "train_loss_llm": 0.0, "grad_norm": 0.07349458336830139, "global_step": 27463, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03557974100112915, "train_loss_bc": 0.03557974100112915, "train_loss_llm": 0.0, "grad_norm": 0.0930674597620964, "global_step": 27464, "epoch": 762, "lr": 0.001} +{"train_loss": 0.03344568982720375, "train_loss_bc": 0.03343789279460907, "train_loss_llm": 0.0007798075675964355, "grad_norm": 0.03748875856399536, "global_step": 27465, "epoch": 762, "lr": 0.001} +{"train_loss": 0.02454228326678276, "train_loss_bc": 0.02454228326678276, "train_loss_llm": 0.0, "grad_norm": 0.05172593519091606, "global_step": 27466, "epoch": 762, "lr": 0.001} +{"train_loss": 0.030666493862453435, "train_loss_bc": 0.03250885009765625, "train_loss_llm": 0.00015291571617126465, "grad_norm": 0.0745176374912262, "global_step": 27467, "epoch": 762, "lr": 0.001, "val_loss": 0.09340134263038635} +{"train_loss": 0.02821035124361515, "train_loss_bc": 0.02821035124361515, "train_loss_llm": 0.0, "grad_norm": 0.09262912720441818, "global_step": 27468, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03235996514558792, "train_loss_bc": 0.032334957271814346, "train_loss_llm": 0.0025006234645843506, "grad_norm": 0.032387182116508484, "global_step": 27469, "epoch": 763, "lr": 0.001} +{"train_loss": 0.02461227960884571, "train_loss_bc": 0.02460511401295662, "train_loss_llm": 0.000716477632522583, "grad_norm": 0.04856644198298454, "global_step": 27470, "epoch": 763, "lr": 0.001} +{"train_loss": 0.033037614077329636, "train_loss_bc": 0.03303103148937225, "train_loss_llm": 0.0006582438945770264, "grad_norm": 0.06947997212409973, "global_step": 27471, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03079436719417572, "train_loss_bc": 0.03079436719417572, "train_loss_llm": 0.0, "grad_norm": 0.0913042277097702, "global_step": 27472, "epoch": 763, "lr": 0.001} +{"train_loss": 0.033851634711027145, "train_loss_bc": 0.033851634711027145, "train_loss_llm": 0.0, "grad_norm": 0.03481798246502876, "global_step": 27473, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03375392034649849, "train_loss_bc": 0.03374607488512993, "train_loss_llm": 0.0007843673229217529, "grad_norm": 0.05868827551603317, "global_step": 27474, "epoch": 763, "lr": 0.001} +{"train_loss": 0.02656332217156887, "train_loss_bc": 0.02656332217156887, "train_loss_llm": 0.0, "grad_norm": 0.07708145678043365, "global_step": 27475, "epoch": 763, "lr": 0.001} +{"train_loss": 0.026589449495077133, "train_loss_bc": 0.026589449495077133, "train_loss_llm": 0.0, "grad_norm": 0.09615425765514374, "global_step": 27476, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03908729925751686, "train_loss_bc": 0.03908729925751686, "train_loss_llm": 0.0, "grad_norm": 0.037320513278245926, "global_step": 27477, "epoch": 763, "lr": 0.001} +{"train_loss": 0.02765684202313423, "train_loss_bc": 0.02765684202313423, "train_loss_llm": 0.0, "grad_norm": 0.05513709783554077, "global_step": 27478, "epoch": 763, "lr": 0.001} +{"train_loss": 0.0281804408878088, "train_loss_bc": 0.0281804408878088, "train_loss_llm": 0.0, "grad_norm": 0.06715676933526993, "global_step": 27479, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03067106008529663, "train_loss_bc": 0.03067106008529663, "train_loss_llm": 0.0, "grad_norm": 0.08969676494598389, "global_step": 27480, "epoch": 763, "lr": 0.001} +{"train_loss": 0.031177692115306854, "train_loss_bc": 0.031172625720500946, "train_loss_llm": 0.0005066990852355957, "grad_norm": 0.04991476982831955, "global_step": 27481, "epoch": 763, "lr": 0.001} +{"train_loss": 0.028560951352119446, "train_loss_bc": 0.028554778546094894, "train_loss_llm": 0.0006173253059387207, "grad_norm": 0.06330060958862305, "global_step": 27482, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03510252758860588, "train_loss_bc": 0.03510252758860588, "train_loss_llm": 0.0, "grad_norm": 0.08178708702325821, "global_step": 27483, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03042801469564438, "train_loss_bc": 0.030407847836613655, "train_loss_llm": 0.002016723155975342, "grad_norm": 0.10037911683320999, "global_step": 27484, "epoch": 763, "lr": 0.001} +{"train_loss": 0.036967504769563675, "train_loss_bc": 0.036967504769563675, "train_loss_llm": 0.0, "grad_norm": 0.05305149778723717, "global_step": 27485, "epoch": 763, "lr": 0.001} +{"train_loss": 0.026367459446191788, "train_loss_bc": 0.026367459446191788, "train_loss_llm": 0.0, "grad_norm": 0.07676397264003754, "global_step": 27486, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03730151429772377, "train_loss_bc": 0.03730151429772377, "train_loss_llm": 0.0, "grad_norm": 0.10935865342617035, "global_step": 27487, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03166627138853073, "train_loss_bc": 0.03165893256664276, "train_loss_llm": 0.0007338225841522217, "grad_norm": 0.1254427582025528, "global_step": 27488, "epoch": 763, "lr": 0.001} +{"train_loss": 0.02717849239706993, "train_loss_bc": 0.02717849239706993, "train_loss_llm": 0.0, "grad_norm": 0.03647850081324577, "global_step": 27489, "epoch": 763, "lr": 0.001} +{"train_loss": 0.031179366633296013, "train_loss_bc": 0.031179366633296013, "train_loss_llm": 0.0, "grad_norm": 0.048528365790843964, "global_step": 27490, "epoch": 763, "lr": 0.001} +{"train_loss": 0.033658575266599655, "train_loss_bc": 0.033658575266599655, "train_loss_llm": 0.0, "grad_norm": 0.07345616072416306, "global_step": 27491, "epoch": 763, "lr": 0.001} +{"train_loss": 0.02533499151468277, "train_loss_bc": 0.02533499151468277, "train_loss_llm": 0.0, "grad_norm": 0.0910106971859932, "global_step": 27492, "epoch": 763, "lr": 0.001} +{"train_loss": 0.02588985301554203, "train_loss_bc": 0.02588985301554203, "train_loss_llm": 0.0, "grad_norm": 0.033818844705820084, "global_step": 27493, "epoch": 763, "lr": 0.001} +{"train_loss": 0.025855030864477158, "train_loss_bc": 0.025846373289823532, "train_loss_llm": 0.0008657574653625488, "grad_norm": 0.04045401141047478, "global_step": 27494, "epoch": 763, "lr": 0.001} +{"train_loss": 0.037019651383161545, "train_loss_bc": 0.037019651383161545, "train_loss_llm": 0.0, "grad_norm": 0.058158740401268005, "global_step": 27495, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03694489225745201, "train_loss_bc": 0.03694489225745201, "train_loss_llm": 0.0, "grad_norm": 0.06199011951684952, "global_step": 27496, "epoch": 763, "lr": 0.001} +{"train_loss": 0.029462609440088272, "train_loss_bc": 0.029462609440088272, "train_loss_llm": 0.0, "grad_norm": 0.028704630210995674, "global_step": 27497, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03412507101893425, "train_loss_bc": 0.03412507101893425, "train_loss_llm": 0.0, "grad_norm": 0.050953932106494904, "global_step": 27498, "epoch": 763, "lr": 0.001} +{"train_loss": 0.029579196125268936, "train_loss_bc": 0.029579196125268936, "train_loss_llm": 0.0, "grad_norm": 0.06516389548778534, "global_step": 27499, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03020092099905014, "train_loss_bc": 0.03020092099905014, "train_loss_llm": 0.0, "grad_norm": 0.08926866203546524, "global_step": 27500, "epoch": 763, "lr": 0.001} +{"train_loss": 0.029662620276212692, "train_loss_bc": 0.029662620276212692, "train_loss_llm": 0.0, "grad_norm": 0.03274478390812874, "global_step": 27501, "epoch": 763, "lr": 0.001} +{"train_loss": 0.02981148473918438, "train_loss_bc": 0.02981148473918438, "train_loss_llm": 0.0, "grad_norm": 0.05043525993824005, "global_step": 27502, "epoch": 763, "lr": 0.001} +{"train_loss": 0.03069328475329611, "train_loss_bc": 0.026111753657460213, "train_loss_llm": 0.0003260374069213867, "grad_norm": 0.07971710711717606, "global_step": 27503, "epoch": 763, "lr": 0.001, "val_loss": 0.09491419792175293} +{"train_loss": 0.039253395050764084, "train_loss_bc": 0.039253395050764084, "train_loss_llm": 0.0, "grad_norm": 0.10183253884315491, "global_step": 27504, "epoch": 764, "lr": 0.001} +{"train_loss": 0.030932782217860222, "train_loss_bc": 0.030932782217860222, "train_loss_llm": 0.0, "grad_norm": 0.04618845134973526, "global_step": 27505, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03299214318394661, "train_loss_bc": 0.03299214318394661, "train_loss_llm": 0.0, "grad_norm": 0.07177117466926575, "global_step": 27506, "epoch": 764, "lr": 0.001} +{"train_loss": 0.02601771429181099, "train_loss_bc": 0.02601771429181099, "train_loss_llm": 0.0, "grad_norm": 0.09235956519842148, "global_step": 27507, "epoch": 764, "lr": 0.001} +{"train_loss": 0.0286102294921875, "train_loss_bc": 0.0286102294921875, "train_loss_llm": 0.0, "grad_norm": 0.11923940479755402, "global_step": 27508, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03061632812023163, "train_loss_bc": 0.03061632812023163, "train_loss_llm": 0.0, "grad_norm": 0.03363019973039627, "global_step": 27509, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03271195665001869, "train_loss_bc": 0.03271195665001869, "train_loss_llm": 0.0, "grad_norm": 0.060846783220767975, "global_step": 27510, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03193741664290428, "train_loss_bc": 0.03192129731178284, "train_loss_llm": 0.0016120672225952148, "grad_norm": 0.08460023999214172, "global_step": 27511, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03266795724630356, "train_loss_bc": 0.03266795724630356, "train_loss_llm": 0.0, "grad_norm": 0.1014273390173912, "global_step": 27512, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03097837045788765, "train_loss_bc": 0.03096240945160389, "train_loss_llm": 0.0015960335731506348, "grad_norm": 0.03305435925722122, "global_step": 27513, "epoch": 764, "lr": 0.001} +{"train_loss": 0.02839871682226658, "train_loss_bc": 0.028397927060723305, "train_loss_llm": 7.900595664978027e-05, "grad_norm": 0.05002293363213539, "global_step": 27514, "epoch": 764, "lr": 0.001} +{"train_loss": 0.030491137877106667, "train_loss_bc": 0.030482904985547066, "train_loss_llm": 0.0008232593536376953, "grad_norm": 0.06520530581474304, "global_step": 27515, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03870897740125656, "train_loss_bc": 0.03870897740125656, "train_loss_llm": 0.0, "grad_norm": 0.08786641061306, "global_step": 27516, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03575693443417549, "train_loss_bc": 0.03575022146105766, "train_loss_llm": 0.0006712377071380615, "grad_norm": 0.024872902780771255, "global_step": 27517, "epoch": 764, "lr": 0.001} +{"train_loss": 0.026162318885326385, "train_loss_bc": 0.026162318885326385, "train_loss_llm": 0.0, "grad_norm": 0.042211584746837616, "global_step": 27518, "epoch": 764, "lr": 0.001} +{"train_loss": 0.02534482628107071, "train_loss_bc": 0.02534482628107071, "train_loss_llm": 0.0, "grad_norm": 0.06125302240252495, "global_step": 27519, "epoch": 764, "lr": 0.001} +{"train_loss": 0.023182731121778488, "train_loss_bc": 0.023182731121778488, "train_loss_llm": 0.0, "grad_norm": 0.08392476290464401, "global_step": 27520, "epoch": 764, "lr": 0.001} +{"train_loss": 0.02843642048537731, "train_loss_bc": 0.02843642048537731, "train_loss_llm": 0.0, "grad_norm": 0.03583308681845665, "global_step": 27521, "epoch": 764, "lr": 0.001} +{"train_loss": 0.027398796752095222, "train_loss_bc": 0.027398796752095222, "train_loss_llm": 0.0, "grad_norm": 0.05769909918308258, "global_step": 27522, "epoch": 764, "lr": 0.001} +{"train_loss": 0.027967795729637146, "train_loss_bc": 0.027967795729637146, "train_loss_llm": 0.0, "grad_norm": 0.08063587546348572, "global_step": 27523, "epoch": 764, "lr": 0.001} +{"train_loss": 0.026786334812641144, "train_loss_bc": 0.026786334812641144, "train_loss_llm": 0.0, "grad_norm": 0.09004046022891998, "global_step": 27524, "epoch": 764, "lr": 0.001} +{"train_loss": 0.0290431696921587, "train_loss_bc": 0.0290431696921587, "train_loss_llm": 0.0, "grad_norm": 0.03267800807952881, "global_step": 27525, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03278770297765732, "train_loss_bc": 0.0327732153236866, "train_loss_llm": 0.0014488399028778076, "grad_norm": 0.06634286046028137, "global_step": 27526, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03484740108251572, "train_loss_bc": 0.03484271466732025, "train_loss_llm": 0.0004684627056121826, "grad_norm": 0.08851724117994308, "global_step": 27527, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03631097823381424, "train_loss_bc": 0.03631097823381424, "train_loss_llm": 0.0, "grad_norm": 0.10630770027637482, "global_step": 27528, "epoch": 764, "lr": 0.001} +{"train_loss": 0.025605347007513046, "train_loss_bc": 0.025605347007513046, "train_loss_llm": 0.0, "grad_norm": 0.032353490591049194, "global_step": 27529, "epoch": 764, "lr": 0.001} +{"train_loss": 0.02810974419116974, "train_loss_bc": 0.028107473626732826, "train_loss_llm": 0.0002270042896270752, "grad_norm": 0.056706443428993225, "global_step": 27530, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03253313899040222, "train_loss_bc": 0.03253313899040222, "train_loss_llm": 0.0, "grad_norm": 0.07290087640285492, "global_step": 27531, "epoch": 764, "lr": 0.001} +{"train_loss": 0.028692608699202538, "train_loss_bc": 0.028692608699202538, "train_loss_llm": 0.0, "grad_norm": 0.09442104399204254, "global_step": 27532, "epoch": 764, "lr": 0.001} +{"train_loss": 0.035189926624298096, "train_loss_bc": 0.03518534451723099, "train_loss_llm": 0.0004582405090332031, "grad_norm": 0.043115463107824326, "global_step": 27533, "epoch": 764, "lr": 0.001} +{"train_loss": 0.031853850930929184, "train_loss_bc": 0.03184264898300171, "train_loss_llm": 0.001120060682296753, "grad_norm": 0.06829003989696503, "global_step": 27534, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03208151087164879, "train_loss_bc": 0.03208151087164879, "train_loss_llm": 0.0, "grad_norm": 0.08759916573762894, "global_step": 27535, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03150268644094467, "train_loss_bc": 0.031495559960603714, "train_loss_llm": 0.0007127821445465088, "grad_norm": 0.10890786349773407, "global_step": 27536, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03598583862185478, "train_loss_bc": 0.03598583862185478, "train_loss_llm": 0.0, "grad_norm": 0.03828219324350357, "global_step": 27537, "epoch": 764, "lr": 0.001} +{"train_loss": 0.03416256234049797, "train_loss_bc": 0.03416256234049797, "train_loss_llm": 0.0, "grad_norm": 0.07484741508960724, "global_step": 27538, "epoch": 764, "lr": 0.001} +{"train_loss": 0.030896447961115174, "train_loss_bc": 0.028212375938892365, "train_loss_llm": 0.0, "grad_norm": 0.0941900908946991, "global_step": 27539, "epoch": 764, "lr": 0.001, "val_loss": 0.09290873259305954} +{"train_loss": 0.03346949815750122, "train_loss_bc": 0.03346799686551094, "train_loss_llm": 0.0001499652862548828, "grad_norm": 0.12252050638198853, "global_step": 27540, "epoch": 765, "lr": 0.001} +{"train_loss": 0.028330601751804352, "train_loss_bc": 0.028330601751804352, "train_loss_llm": 0.0, "grad_norm": 0.032787494361400604, "global_step": 27541, "epoch": 765, "lr": 0.001} +{"train_loss": 0.029058687388896942, "train_loss_bc": 0.029046496376395226, "train_loss_llm": 0.0012190639972686768, "grad_norm": 0.057319995015859604, "global_step": 27542, "epoch": 765, "lr": 0.001} +{"train_loss": 0.027750862762331963, "train_loss_bc": 0.027749396860599518, "train_loss_llm": 0.00014659762382507324, "grad_norm": 0.06311823427677155, "global_step": 27543, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03577360510826111, "train_loss_bc": 0.03577360510826111, "train_loss_llm": 0.0, "grad_norm": 0.07711344212293625, "global_step": 27544, "epoch": 765, "lr": 0.001} +{"train_loss": 0.02854233980178833, "train_loss_bc": 0.02854233980178833, "train_loss_llm": 0.0, "grad_norm": 0.03382314741611481, "global_step": 27545, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03011048212647438, "train_loss_bc": 0.030102189630270004, "train_loss_llm": 0.0008293390274047852, "grad_norm": 0.04855028912425041, "global_step": 27546, "epoch": 765, "lr": 0.001} +{"train_loss": 0.021213281899690628, "train_loss_bc": 0.021213281899690628, "train_loss_llm": 0.0, "grad_norm": 0.06221257522702217, "global_step": 27547, "epoch": 765, "lr": 0.001} +{"train_loss": 0.027028094977140427, "train_loss_bc": 0.027028094977140427, "train_loss_llm": 0.0, "grad_norm": 0.07479091733694077, "global_step": 27548, "epoch": 765, "lr": 0.001} +{"train_loss": 0.02535782754421234, "train_loss_bc": 0.02535782754421234, "train_loss_llm": 0.0, "grad_norm": 0.03378935530781746, "global_step": 27549, "epoch": 765, "lr": 0.001} +{"train_loss": 0.028434747830033302, "train_loss_bc": 0.028434747830033302, "train_loss_llm": 0.0, "grad_norm": 0.053176965564489365, "global_step": 27550, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03279958665370941, "train_loss_bc": 0.03279958665370941, "train_loss_llm": 0.0, "grad_norm": 0.07002433389425278, "global_step": 27551, "epoch": 765, "lr": 0.001} +{"train_loss": 0.030186444520950317, "train_loss_bc": 0.030186444520950317, "train_loss_llm": 0.0, "grad_norm": 0.08762701600790024, "global_step": 27552, "epoch": 765, "lr": 0.001} +{"train_loss": 0.0337475948035717, "train_loss_bc": 0.0337475948035717, "train_loss_llm": 0.0, "grad_norm": 0.04533076286315918, "global_step": 27553, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03171319514513016, "train_loss_bc": 0.03171319514513016, "train_loss_llm": 0.0, "grad_norm": 0.08178219199180603, "global_step": 27554, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03476386517286301, "train_loss_bc": 0.03476386517286301, "train_loss_llm": 0.0, "grad_norm": 0.11903493851423264, "global_step": 27555, "epoch": 765, "lr": 0.001} +{"train_loss": 0.027834482491016388, "train_loss_bc": 0.027834482491016388, "train_loss_llm": 0.0, "grad_norm": 0.14191269874572754, "global_step": 27556, "epoch": 765, "lr": 0.001} +{"train_loss": 0.02944403700530529, "train_loss_bc": 0.02944403700530529, "train_loss_llm": 0.0, "grad_norm": 0.04154927283525467, "global_step": 27557, "epoch": 765, "lr": 0.001} +{"train_loss": 0.026026546955108643, "train_loss_bc": 0.026023171842098236, "train_loss_llm": 0.0003374814987182617, "grad_norm": 0.05914347618818283, "global_step": 27558, "epoch": 765, "lr": 0.001} +{"train_loss": 0.034387536346912384, "train_loss_bc": 0.034387536346912384, "train_loss_llm": 0.0, "grad_norm": 0.08389028906822205, "global_step": 27559, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03479933738708496, "train_loss_bc": 0.03479933738708496, "train_loss_llm": 0.0, "grad_norm": 0.10996580868959427, "global_step": 27560, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03352348506450653, "train_loss_bc": 0.03352348506450653, "train_loss_llm": 0.0, "grad_norm": 0.035578686743974686, "global_step": 27561, "epoch": 765, "lr": 0.001} +{"train_loss": 0.027302047237753868, "train_loss_bc": 0.027302047237753868, "train_loss_llm": 0.0, "grad_norm": 0.06247568503022194, "global_step": 27562, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03589612990617752, "train_loss_bc": 0.03589612990617752, "train_loss_llm": 0.0, "grad_norm": 0.08883249014616013, "global_step": 27563, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03134358674287796, "train_loss_bc": 0.031339164823293686, "train_loss_llm": 0.00044226646423339844, "grad_norm": 0.1085527315735817, "global_step": 27564, "epoch": 765, "lr": 0.001} +{"train_loss": 0.030223393812775612, "train_loss_bc": 0.030223393812775612, "train_loss_llm": 0.0, "grad_norm": 0.029219098389148712, "global_step": 27565, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03063865192234516, "train_loss_bc": 0.03063865192234516, "train_loss_llm": 0.0, "grad_norm": 0.040755100548267365, "global_step": 27566, "epoch": 765, "lr": 0.001} +{"train_loss": 0.029733849689364433, "train_loss_bc": 0.029733849689364433, "train_loss_llm": 0.0, "grad_norm": 0.05090278014540672, "global_step": 27567, "epoch": 765, "lr": 0.001} +{"train_loss": 0.034804146736860275, "train_loss_bc": 0.03479503095149994, "train_loss_llm": 0.0009116828441619873, "grad_norm": 0.07052067667245865, "global_step": 27568, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03606725111603737, "train_loss_bc": 0.03606725111603737, "train_loss_llm": 0.0, "grad_norm": 0.033280499279499054, "global_step": 27569, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03417070209980011, "train_loss_bc": 0.03417070209980011, "train_loss_llm": 0.0, "grad_norm": 0.05061781778931618, "global_step": 27570, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03322688862681389, "train_loss_bc": 0.03322688862681389, "train_loss_llm": 0.0, "grad_norm": 0.06185660511255264, "global_step": 27571, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03167247027158737, "train_loss_bc": 0.03165404498577118, "train_loss_llm": 0.001842409372329712, "grad_norm": 0.07781930267810822, "global_step": 27572, "epoch": 765, "lr": 0.001} +{"train_loss": 0.034888796508312225, "train_loss_bc": 0.034888796508312225, "train_loss_llm": 0.0, "grad_norm": 0.04603564366698265, "global_step": 27573, "epoch": 765, "lr": 0.001} +{"train_loss": 0.026300320401787758, "train_loss_bc": 0.026300320401787758, "train_loss_llm": 0.0, "grad_norm": 0.07450518012046814, "global_step": 27574, "epoch": 765, "lr": 0.001} +{"train_loss": 0.03080290136858821, "train_loss_bc": 0.028334220871329308, "train_loss_llm": 0.0005852282047271729, "grad_norm": 0.0964030846953392, "global_step": 27575, "epoch": 765, "lr": 0.001, "train/cumulative_reward": 95.32704582324047, "train/mean_score": 11.793702079314327, "train/success_rate": 0.2, "test/cumulative_reward": 35.87210821450576, "test/mean_score": 6.757646805798192, "test/success_rate": 0.16, "val_loss": 0.09432664513587952, "train_action_mse_error": 0.03343851491808891} +{"train_loss": 0.03171868994832039, "train_loss_bc": 0.03171868994832039, "train_loss_llm": 0.0, "grad_norm": 0.11548160016536713, "global_step": 27576, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03280553221702576, "train_loss_bc": 0.03277546167373657, "train_loss_llm": 0.003007054328918457, "grad_norm": 0.032272838056087494, "global_step": 27577, "epoch": 766, "lr": 0.001} +{"train_loss": 0.028314068913459778, "train_loss_bc": 0.028314068913459778, "train_loss_llm": 0.0, "grad_norm": 0.05717146769165993, "global_step": 27578, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03586822375655174, "train_loss_bc": 0.03586822375655174, "train_loss_llm": 0.0, "grad_norm": 0.07835667580366135, "global_step": 27579, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03481534123420715, "train_loss_bc": 0.03480818495154381, "train_loss_llm": 0.0007154941558837891, "grad_norm": 0.10842426866292953, "global_step": 27580, "epoch": 766, "lr": 0.001} +{"train_loss": 0.023543424904346466, "train_loss_bc": 0.023543424904346466, "train_loss_llm": 0.0, "grad_norm": 0.035617463290691376, "global_step": 27581, "epoch": 766, "lr": 0.001} +{"train_loss": 0.030066603794693947, "train_loss_bc": 0.030066603794693947, "train_loss_llm": 0.0, "grad_norm": 0.06283171474933624, "global_step": 27582, "epoch": 766, "lr": 0.001} +{"train_loss": 0.02964371256530285, "train_loss_bc": 0.02964371256530285, "train_loss_llm": 0.0, "grad_norm": 0.08767254650592804, "global_step": 27583, "epoch": 766, "lr": 0.001} +{"train_loss": 0.028474021703004837, "train_loss_bc": 0.028460294008255005, "train_loss_llm": 0.0013727545738220215, "grad_norm": 0.1143605187535286, "global_step": 27584, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03262896090745926, "train_loss_bc": 0.03260627016425133, "train_loss_llm": 0.0022692084312438965, "grad_norm": 0.038268569856882095, "global_step": 27585, "epoch": 766, "lr": 0.001} +{"train_loss": 0.030542224645614624, "train_loss_bc": 0.030542224645614624, "train_loss_llm": 0.0, "grad_norm": 0.06321120262145996, "global_step": 27586, "epoch": 766, "lr": 0.001} +{"train_loss": 0.0317532978951931, "train_loss_bc": 0.031726136803627014, "train_loss_llm": 0.0027162134647369385, "grad_norm": 0.07992582768201828, "global_step": 27587, "epoch": 766, "lr": 0.001} +{"train_loss": 0.02990121953189373, "train_loss_bc": 0.02990121953189373, "train_loss_llm": 0.0, "grad_norm": 0.08635476976633072, "global_step": 27588, "epoch": 766, "lr": 0.001} +{"train_loss": 0.027557887136936188, "train_loss_bc": 0.027555324137210846, "train_loss_llm": 0.0002562999725341797, "grad_norm": 0.03315869718790054, "global_step": 27589, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03982307389378548, "train_loss_bc": 0.03982307389378548, "train_loss_llm": 0.0, "grad_norm": 0.06858741492033005, "global_step": 27590, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03091956116259098, "train_loss_bc": 0.03091956116259098, "train_loss_llm": 0.0, "grad_norm": 0.08237562328577042, "global_step": 27591, "epoch": 766, "lr": 0.001} +{"train_loss": 0.030218934640288353, "train_loss_bc": 0.030218934640288353, "train_loss_llm": 0.0, "grad_norm": 0.09150873869657516, "global_step": 27592, "epoch": 766, "lr": 0.001} +{"train_loss": 0.032101087272167206, "train_loss_bc": 0.032101087272167206, "train_loss_llm": 0.0, "grad_norm": 0.03231317549943924, "global_step": 27593, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03147495537996292, "train_loss_bc": 0.03147495537996292, "train_loss_llm": 0.0, "grad_norm": 0.05124868452548981, "global_step": 27594, "epoch": 766, "lr": 0.001} +{"train_loss": 0.028988756239414215, "train_loss_bc": 0.028988756239414215, "train_loss_llm": 0.0, "grad_norm": 0.06729941815137863, "global_step": 27595, "epoch": 766, "lr": 0.001} +{"train_loss": 0.02867352031171322, "train_loss_bc": 0.028666554018855095, "train_loss_llm": 0.0006966590881347656, "grad_norm": 0.09275785833597183, "global_step": 27596, "epoch": 766, "lr": 0.001} +{"train_loss": 0.029652869328856468, "train_loss_bc": 0.029652869328856468, "train_loss_llm": 0.0, "grad_norm": 0.0379793718457222, "global_step": 27597, "epoch": 766, "lr": 0.001} +{"train_loss": 0.026421692222356796, "train_loss_bc": 0.026421692222356796, "train_loss_llm": 0.0, "grad_norm": 0.04475873336195946, "global_step": 27598, "epoch": 766, "lr": 0.001} +{"train_loss": 0.029184944927692413, "train_loss_bc": 0.029184944927692413, "train_loss_llm": 0.0, "grad_norm": 0.06250322610139847, "global_step": 27599, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03261741250753403, "train_loss_bc": 0.03261741250753403, "train_loss_llm": 0.0, "grad_norm": 0.08621599525213242, "global_step": 27600, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03614874184131622, "train_loss_bc": 0.0361311137676239, "train_loss_llm": 0.001762986183166504, "grad_norm": 0.0452333502471447, "global_step": 27601, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03024919517338276, "train_loss_bc": 0.03024919517338276, "train_loss_llm": 0.0, "grad_norm": 0.05687739700078964, "global_step": 27602, "epoch": 766, "lr": 0.001} +{"train_loss": 0.028311146423220634, "train_loss_bc": 0.028311146423220634, "train_loss_llm": 0.0, "grad_norm": 0.06317992508411407, "global_step": 27603, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03192237764596939, "train_loss_bc": 0.03192237764596939, "train_loss_llm": 0.0, "grad_norm": 0.07746264338493347, "global_step": 27604, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03293328359723091, "train_loss_bc": 0.03293328359723091, "train_loss_llm": 0.0, "grad_norm": 0.03722697123885155, "global_step": 27605, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03226078301668167, "train_loss_bc": 0.032250452786684036, "train_loss_llm": 0.0010331571102142334, "grad_norm": 0.05848940834403038, "global_step": 27606, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03357582166790962, "train_loss_bc": 0.03357582166790962, "train_loss_llm": 0.0, "grad_norm": 0.07305672019720078, "global_step": 27607, "epoch": 766, "lr": 0.001} +{"train_loss": 0.025433724746108055, "train_loss_bc": 0.02542681246995926, "train_loss_llm": 0.0006912648677825928, "grad_norm": 0.08215036988258362, "global_step": 27608, "epoch": 766, "lr": 0.001} +{"train_loss": 0.030058836564421654, "train_loss_bc": 0.030058836564421654, "train_loss_llm": 0.0, "grad_norm": 0.031147252768278122, "global_step": 27609, "epoch": 766, "lr": 0.001} +{"train_loss": 0.027156274765729904, "train_loss_bc": 0.027156274765729904, "train_loss_llm": 0.0, "grad_norm": 0.04521968960762024, "global_step": 27610, "epoch": 766, "lr": 0.001} +{"train_loss": 0.03055260045867827, "train_loss_bc": 0.024133414030075073, "train_loss_llm": 0.0, "grad_norm": 0.05482609197497368, "global_step": 27611, "epoch": 766, "lr": 0.001, "val_loss": 0.09474605321884155} +{"train_loss": 0.022409621626138687, "train_loss_bc": 0.022409621626138687, "train_loss_llm": 0.0, "grad_norm": 0.06882532685995102, "global_step": 27612, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03395788371562958, "train_loss_bc": 0.03395788371562958, "train_loss_llm": 0.0, "grad_norm": 0.03766671195626259, "global_step": 27613, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03342237323522568, "train_loss_bc": 0.03342237323522568, "train_loss_llm": 0.0, "grad_norm": 0.0578082837164402, "global_step": 27614, "epoch": 767, "lr": 0.001} +{"train_loss": 0.02816125564277172, "train_loss_bc": 0.028154876083135605, "train_loss_llm": 0.0006379783153533936, "grad_norm": 0.07156975567340851, "global_step": 27615, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03173574432730675, "train_loss_bc": 0.03173574432730675, "train_loss_llm": 0.0, "grad_norm": 0.0824243351817131, "global_step": 27616, "epoch": 767, "lr": 0.001} +{"train_loss": 0.025433635339140892, "train_loss_bc": 0.025433635339140892, "train_loss_llm": 0.0, "grad_norm": 0.03463032841682434, "global_step": 27617, "epoch": 767, "lr": 0.001} +{"train_loss": 0.035377588123083115, "train_loss_bc": 0.035377588123083115, "train_loss_llm": 0.0, "grad_norm": 0.05874129757285118, "global_step": 27618, "epoch": 767, "lr": 0.001} +{"train_loss": 0.02757992409169674, "train_loss_bc": 0.027574926614761353, "train_loss_llm": 0.0004997849464416504, "grad_norm": 0.08411141484975815, "global_step": 27619, "epoch": 767, "lr": 0.001} +{"train_loss": 0.038423605263233185, "train_loss_bc": 0.038421425968408585, "train_loss_llm": 0.0002180635929107666, "grad_norm": 0.10294186323881149, "global_step": 27620, "epoch": 767, "lr": 0.001} +{"train_loss": 0.029284285381436348, "train_loss_bc": 0.029281431809067726, "train_loss_llm": 0.0002853572368621826, "grad_norm": 0.031066929921507835, "global_step": 27621, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03287845477461815, "train_loss_bc": 0.03287845477461815, "train_loss_llm": 0.0, "grad_norm": 0.0614045187830925, "global_step": 27622, "epoch": 767, "lr": 0.001} +{"train_loss": 0.025699470192193985, "train_loss_bc": 0.02569681592285633, "train_loss_llm": 0.00026550889015197754, "grad_norm": 0.07279618084430695, "global_step": 27623, "epoch": 767, "lr": 0.001} +{"train_loss": 0.032105546444654465, "train_loss_bc": 0.032105546444654465, "train_loss_llm": 0.0, "grad_norm": 0.09072453528642654, "global_step": 27624, "epoch": 767, "lr": 0.001} +{"train_loss": 0.027540262788534164, "train_loss_bc": 0.027540262788534164, "train_loss_llm": 0.0, "grad_norm": 0.03673398494720459, "global_step": 27625, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03261175751686096, "train_loss_bc": 0.03261175751686096, "train_loss_llm": 0.0, "grad_norm": 0.06041855365037918, "global_step": 27626, "epoch": 767, "lr": 0.001} +{"train_loss": 0.02870224416255951, "train_loss_bc": 0.02870224416255951, "train_loss_llm": 0.0, "grad_norm": 0.07423588633537292, "global_step": 27627, "epoch": 767, "lr": 0.001} +{"train_loss": 0.025385119020938873, "train_loss_bc": 0.025385119020938873, "train_loss_llm": 0.0, "grad_norm": 0.09039860218763351, "global_step": 27628, "epoch": 767, "lr": 0.001} +{"train_loss": 0.029376832768321037, "train_loss_bc": 0.029376832768321037, "train_loss_llm": 0.0, "grad_norm": 0.03844386711716652, "global_step": 27629, "epoch": 767, "lr": 0.001} +{"train_loss": 0.031234856694936752, "train_loss_bc": 0.031234856694936752, "train_loss_llm": 0.0, "grad_norm": 0.05728766694664955, "global_step": 27630, "epoch": 767, "lr": 0.001} +{"train_loss": 0.025896118953824043, "train_loss_bc": 0.025896118953824043, "train_loss_llm": 0.0, "grad_norm": 0.06485584378242493, "global_step": 27631, "epoch": 767, "lr": 0.001} +{"train_loss": 0.02930448018014431, "train_loss_bc": 0.02930448018014431, "train_loss_llm": 0.0, "grad_norm": 0.08383333683013916, "global_step": 27632, "epoch": 767, "lr": 0.001} +{"train_loss": 0.036490101367235184, "train_loss_bc": 0.036490101367235184, "train_loss_llm": 0.0, "grad_norm": 0.050367988646030426, "global_step": 27633, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03431712090969086, "train_loss_bc": 0.03431712090969086, "train_loss_llm": 0.0, "grad_norm": 0.08517075330018997, "global_step": 27634, "epoch": 767, "lr": 0.001} +{"train_loss": 0.037245605140924454, "train_loss_bc": 0.037245605140924454, "train_loss_llm": 0.0, "grad_norm": 0.1022382527589798, "global_step": 27635, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03370348736643791, "train_loss_bc": 0.033667098730802536, "train_loss_llm": 0.0036386847496032715, "grad_norm": 0.13495337963104248, "global_step": 27636, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03247637301683426, "train_loss_bc": 0.03247637301683426, "train_loss_llm": 0.0, "grad_norm": 0.04389754310250282, "global_step": 27637, "epoch": 767, "lr": 0.001} +{"train_loss": 0.030854027718305588, "train_loss_bc": 0.030854027718305588, "train_loss_llm": 0.0, "grad_norm": 0.04913181811571121, "global_step": 27638, "epoch": 767, "lr": 0.001} +{"train_loss": 0.02329946495592594, "train_loss_bc": 0.02329946495592594, "train_loss_llm": 0.0, "grad_norm": 0.05639887973666191, "global_step": 27639, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03193400055170059, "train_loss_bc": 0.031919676810503006, "train_loss_llm": 0.0014322996139526367, "grad_norm": 0.08727128803730011, "global_step": 27640, "epoch": 767, "lr": 0.001} +{"train_loss": 0.030761243775486946, "train_loss_bc": 0.030761243775486946, "train_loss_llm": 0.0, "grad_norm": 0.03490602597594261, "global_step": 27641, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03285660222172737, "train_loss_bc": 0.03285660222172737, "train_loss_llm": 0.0, "grad_norm": 0.06563416123390198, "global_step": 27642, "epoch": 767, "lr": 0.001} +{"train_loss": 0.030506087467074394, "train_loss_bc": 0.030506087467074394, "train_loss_llm": 0.0, "grad_norm": 0.08854807168245316, "global_step": 27643, "epoch": 767, "lr": 0.001} +{"train_loss": 0.02903922274708748, "train_loss_bc": 0.02903258241713047, "train_loss_llm": 0.0006640851497650146, "grad_norm": 0.11089237779378891, "global_step": 27644, "epoch": 767, "lr": 0.001} +{"train_loss": 0.027838561683893204, "train_loss_bc": 0.027838561683893204, "train_loss_llm": 0.0, "grad_norm": 0.03550352528691292, "global_step": 27645, "epoch": 767, "lr": 0.001} +{"train_loss": 0.03166131302714348, "train_loss_bc": 0.03165096417069435, "train_loss_llm": 0.0010347068309783936, "grad_norm": 0.05323474481701851, "global_step": 27646, "epoch": 767, "lr": 0.001} +{"train_loss": 0.030696340629623994, "train_loss_bc": 0.03556399047374725, "train_loss_llm": 0.0, "grad_norm": 0.08523469418287277, "global_step": 27647, "epoch": 767, "lr": 0.001, "val_loss": 0.09541476517915726} +{"train_loss": 0.027503060176968575, "train_loss_bc": 0.027503060176968575, "train_loss_llm": 0.0, "grad_norm": 0.10996352136135101, "global_step": 27648, "epoch": 768, "lr": 0.001} +{"train_loss": 0.031736940145492554, "train_loss_bc": 0.03172073885798454, "train_loss_llm": 0.0016200840473175049, "grad_norm": 0.05079777538776398, "global_step": 27649, "epoch": 768, "lr": 0.001} +{"train_loss": 0.028200775384902954, "train_loss_bc": 0.02819618210196495, "train_loss_llm": 0.00045934319496154785, "grad_norm": 0.0675114169716835, "global_step": 27650, "epoch": 768, "lr": 0.001} +{"train_loss": 0.03041810914874077, "train_loss_bc": 0.03041308932006359, "train_loss_llm": 0.0005020201206207275, "grad_norm": 0.08409620076417923, "global_step": 27651, "epoch": 768, "lr": 0.001} +{"train_loss": 0.030629819259047508, "train_loss_bc": 0.030629819259047508, "train_loss_llm": 0.0, "grad_norm": 0.11242004483938217, "global_step": 27652, "epoch": 768, "lr": 0.001} +{"train_loss": 0.03080415166914463, "train_loss_bc": 0.03080415166914463, "train_loss_llm": 0.0, "grad_norm": 0.028915178030729294, "global_step": 27653, "epoch": 768, "lr": 0.001} +{"train_loss": 0.02672574296593666, "train_loss_bc": 0.02672574296593666, "train_loss_llm": 0.0, "grad_norm": 0.05028205364942551, "global_step": 27654, "epoch": 768, "lr": 0.001} +{"train_loss": 0.031980257481336594, "train_loss_bc": 0.031980257481336594, "train_loss_llm": 0.0, "grad_norm": 0.06497593969106674, "global_step": 27655, "epoch": 768, "lr": 0.001} +{"train_loss": 0.027990302070975304, "train_loss_bc": 0.027977067977190018, "train_loss_llm": 0.0013233423233032227, "grad_norm": 0.09484488517045975, "global_step": 27656, "epoch": 768, "lr": 0.001} +{"train_loss": 0.027662761509418488, "train_loss_bc": 0.02765706181526184, "train_loss_llm": 0.0005699396133422852, "grad_norm": 0.0526106134057045, "global_step": 27657, "epoch": 768, "lr": 0.001} +{"train_loss": 0.03206050768494606, "train_loss_bc": 0.03206050768494606, "train_loss_llm": 0.0, "grad_norm": 0.07400871813297272, "global_step": 27658, "epoch": 768, "lr": 0.001} +{"train_loss": 0.032570980489254, "train_loss_bc": 0.032570980489254, "train_loss_llm": 0.0, "grad_norm": 0.09111405164003372, "global_step": 27659, "epoch": 768, "lr": 0.001} +{"train_loss": 0.03466834872961044, "train_loss_bc": 0.03466834872961044, "train_loss_llm": 0.0, "grad_norm": 0.12441005557775497, "global_step": 27660, "epoch": 768, "lr": 0.001} +{"train_loss": 0.031944554299116135, "train_loss_bc": 0.031944554299116135, "train_loss_llm": 0.0, "grad_norm": 0.04040653258562088, "global_step": 27661, "epoch": 768, "lr": 0.001} +{"train_loss": 0.03339727595448494, "train_loss_bc": 0.03339727595448494, "train_loss_llm": 0.0, "grad_norm": 0.06967387348413467, "global_step": 27662, "epoch": 768, "lr": 0.001} +{"train_loss": 0.02948506362736225, "train_loss_bc": 0.02948506362736225, "train_loss_llm": 0.0, "grad_norm": 0.09278438240289688, "global_step": 27663, "epoch": 768, "lr": 0.001} +{"train_loss": 0.030995532870292664, "train_loss_bc": 0.030995532870292664, "train_loss_llm": 0.0, "grad_norm": 0.09962767362594604, "global_step": 27664, "epoch": 768, "lr": 0.001} +{"train_loss": 0.027516627684235573, "train_loss_bc": 0.027516627684235573, "train_loss_llm": 0.0, "grad_norm": 0.030327340587973595, "global_step": 27665, "epoch": 768, "lr": 0.001} +{"train_loss": 0.030372852459549904, "train_loss_bc": 0.03037065453827381, "train_loss_llm": 0.00021982192993164062, "grad_norm": 0.045222897082567215, "global_step": 27666, "epoch": 768, "lr": 0.001} +{"train_loss": 0.022752664983272552, "train_loss_bc": 0.022752664983272552, "train_loss_llm": 0.0, "grad_norm": 0.06305325776338577, "global_step": 27667, "epoch": 768, "lr": 0.001} +{"train_loss": 0.033454105257987976, "train_loss_bc": 0.033454105257987976, "train_loss_llm": 0.0, "grad_norm": 0.07615628093481064, "global_step": 27668, "epoch": 768, "lr": 0.001} +{"train_loss": 0.035131774842739105, "train_loss_bc": 0.035131774842739105, "train_loss_llm": 0.0, "grad_norm": 0.02980201132595539, "global_step": 27669, "epoch": 768, "lr": 0.001} +{"train_loss": 0.027571124956011772, "train_loss_bc": 0.027555890381336212, "train_loss_llm": 0.001523524522781372, "grad_norm": 0.0472799688577652, "global_step": 27670, "epoch": 768, "lr": 0.001} +{"train_loss": 0.033253263682127, "train_loss_bc": 0.033253263682127, "train_loss_llm": 0.0, "grad_norm": 0.06957920640707016, "global_step": 27671, "epoch": 768, "lr": 0.001} +{"train_loss": 0.03408805653452873, "train_loss_bc": 0.03408805653452873, "train_loss_llm": 0.0, "grad_norm": 0.09908034652471542, "global_step": 27672, "epoch": 768, "lr": 0.001} +{"train_loss": 0.03579961881041527, "train_loss_bc": 0.03579659387469292, "train_loss_llm": 0.00030243396759033203, "grad_norm": 0.037602704018354416, "global_step": 27673, "epoch": 768, "lr": 0.001} +{"train_loss": 0.030088286846876144, "train_loss_bc": 0.030064700171351433, "train_loss_llm": 0.0023587048053741455, "grad_norm": 0.07742766290903091, "global_step": 27674, "epoch": 768, "lr": 0.001} +{"train_loss": 0.03397146984934807, "train_loss_bc": 0.03397146984934807, "train_loss_llm": 0.0, "grad_norm": 0.1085735633969307, "global_step": 27675, "epoch": 768, "lr": 0.001} +{"train_loss": 0.02747614122927189, "train_loss_bc": 0.02747614122927189, "train_loss_llm": 0.0, "grad_norm": 0.14352849125862122, "global_step": 27676, "epoch": 768, "lr": 0.001} +{"train_loss": 0.030191808938980103, "train_loss_bc": 0.030191808938980103, "train_loss_llm": 0.0, "grad_norm": 0.041601259261369705, "global_step": 27677, "epoch": 768, "lr": 0.001} +{"train_loss": 0.026764268055558205, "train_loss_bc": 0.026756005361676216, "train_loss_llm": 0.0008262693881988525, "grad_norm": 0.060111187398433685, "global_step": 27678, "epoch": 768, "lr": 0.001} +{"train_loss": 0.035798054188489914, "train_loss_bc": 0.03579479083418846, "train_loss_llm": 0.0003262162208557129, "grad_norm": 0.09069997072219849, "global_step": 27679, "epoch": 768, "lr": 0.001} +{"train_loss": 0.03626277670264244, "train_loss_bc": 0.036260150372982025, "train_loss_llm": 0.000262528657913208, "grad_norm": 0.13365373015403748, "global_step": 27680, "epoch": 768, "lr": 0.001} +{"train_loss": 0.028007302433252335, "train_loss_bc": 0.02799789234995842, "train_loss_llm": 0.0009410679340362549, "grad_norm": 0.03373553603887558, "global_step": 27681, "epoch": 768, "lr": 0.001} +{"train_loss": 0.030698714777827263, "train_loss_bc": 0.030698714777827263, "train_loss_llm": 0.0, "grad_norm": 0.04991956800222397, "global_step": 27682, "epoch": 768, "lr": 0.001} +{"train_loss": 0.030875411660720904, "train_loss_bc": 0.0335417240858078, "train_loss_llm": 0.0, "grad_norm": 0.07072465866804123, "global_step": 27683, "epoch": 768, "lr": 0.001, "val_loss": 0.09409111738204956} +{"train_loss": 0.02908479794859886, "train_loss_bc": 0.02908479794859886, "train_loss_llm": 0.0, "grad_norm": 0.08522872626781464, "global_step": 27684, "epoch": 769, "lr": 0.001} +{"train_loss": 0.03408382833003998, "train_loss_bc": 0.03408382833003998, "train_loss_llm": 0.0, "grad_norm": 0.028819626197218895, "global_step": 27685, "epoch": 769, "lr": 0.001} +{"train_loss": 0.03059329278767109, "train_loss_bc": 0.03059329278767109, "train_loss_llm": 0.0, "grad_norm": 0.03881587088108063, "global_step": 27686, "epoch": 769, "lr": 0.001} +{"train_loss": 0.027823016047477722, "train_loss_bc": 0.027823016047477722, "train_loss_llm": 0.0, "grad_norm": 0.047091394662857056, "global_step": 27687, "epoch": 769, "lr": 0.001} +{"train_loss": 0.0276393610984087, "train_loss_bc": 0.0276393610984087, "train_loss_llm": 0.0, "grad_norm": 0.06301292777061462, "global_step": 27688, "epoch": 769, "lr": 0.001} +{"train_loss": 0.026785260066390038, "train_loss_bc": 0.026785260066390038, "train_loss_llm": 0.0, "grad_norm": 0.03095148876309395, "global_step": 27689, "epoch": 769, "lr": 0.001} +{"train_loss": 0.033758968114852905, "train_loss_bc": 0.033758968114852905, "train_loss_llm": 0.0, "grad_norm": 0.06186496466398239, "global_step": 27690, "epoch": 769, "lr": 0.001} +{"train_loss": 0.03094727173447609, "train_loss_bc": 0.03094727173447609, "train_loss_llm": 0.0, "grad_norm": 0.075057752430439, "global_step": 27691, "epoch": 769, "lr": 0.001} +{"train_loss": 0.0275362990796566, "train_loss_bc": 0.0275362990796566, "train_loss_llm": 0.0, "grad_norm": 0.07917748391628265, "global_step": 27692, "epoch": 769, "lr": 0.001} +{"train_loss": 0.029864709824323654, "train_loss_bc": 0.029864709824323654, "train_loss_llm": 0.0, "grad_norm": 0.03966208174824715, "global_step": 27693, "epoch": 769, "lr": 0.001} +{"train_loss": 0.03321095556020737, "train_loss_bc": 0.03321095556020737, "train_loss_llm": 0.0, "grad_norm": 0.07105284184217453, "global_step": 27694, "epoch": 769, "lr": 0.001} +{"train_loss": 0.02601940929889679, "train_loss_bc": 0.02601940929889679, "train_loss_llm": 0.0, "grad_norm": 0.09746196866035461, "global_step": 27695, "epoch": 769, "lr": 0.001} +{"train_loss": 0.027171675115823746, "train_loss_bc": 0.027171675115823746, "train_loss_llm": 0.0, "grad_norm": 0.11774320900440216, "global_step": 27696, "epoch": 769, "lr": 0.001} +{"train_loss": 0.03702617436647415, "train_loss_bc": 0.03702549263834953, "train_loss_llm": 6.80387020111084e-05, "grad_norm": 0.043139584362506866, "global_step": 27697, "epoch": 769, "lr": 0.001} +{"train_loss": 0.032789196819067, "train_loss_bc": 0.032789196819067, "train_loss_llm": 0.0, "grad_norm": 0.07712452113628387, "global_step": 27698, "epoch": 769, "lr": 0.001} +{"train_loss": 0.03069363534450531, "train_loss_bc": 0.03069363534450531, "train_loss_llm": 0.0, "grad_norm": 0.10793235898017883, "global_step": 27699, "epoch": 769, "lr": 0.001} +{"train_loss": 0.03403487801551819, "train_loss_bc": 0.03400852158665657, "train_loss_llm": 0.0026355981826782227, "grad_norm": 0.13716581463813782, "global_step": 27700, "epoch": 769, "lr": 0.001} +{"train_loss": 0.030925240367650986, "train_loss_bc": 0.030925240367650986, "train_loss_llm": 0.0, "grad_norm": 0.036139361560344696, "global_step": 27701, "epoch": 769, "lr": 0.001} +{"train_loss": 0.02870824560523033, "train_loss_bc": 0.02870824560523033, "train_loss_llm": 0.0, "grad_norm": 0.05659119412302971, "global_step": 27702, "epoch": 769, "lr": 0.001} +{"train_loss": 0.02950437180697918, "train_loss_bc": 0.02950437180697918, "train_loss_llm": 0.0, "grad_norm": 0.07205097377300262, "global_step": 27703, "epoch": 769, "lr": 0.001} +{"train_loss": 0.03568043187260628, "train_loss_bc": 0.03566896170377731, "train_loss_llm": 0.0011468827724456787, "grad_norm": 0.09701412171125412, "global_step": 27704, "epoch": 769, "lr": 0.001} +{"train_loss": 0.04163380712270737, "train_loss_bc": 0.04163380712270737, "train_loss_llm": 0.0, "grad_norm": 0.04209008440375328, "global_step": 27705, "epoch": 769, "lr": 0.001} +{"train_loss": 0.03374837338924408, "train_loss_bc": 0.03374837338924408, "train_loss_llm": 0.0, "grad_norm": 0.05489729344844818, "global_step": 27706, "epoch": 769, "lr": 0.001} +{"train_loss": 0.028340114280581474, "train_loss_bc": 0.028340114280581474, "train_loss_llm": 0.0, "grad_norm": 0.06937985122203827, "global_step": 27707, "epoch": 769, "lr": 0.001} +{"train_loss": 0.0292810071259737, "train_loss_bc": 0.02927359938621521, "train_loss_llm": 0.0007407069206237793, "grad_norm": 0.07729000598192215, "global_step": 27708, "epoch": 769, "lr": 0.001} +{"train_loss": 0.02416052483022213, "train_loss_bc": 0.024159792810678482, "train_loss_llm": 7.316470146179199e-05, "grad_norm": 0.03113601543009281, "global_step": 27709, "epoch": 769, "lr": 0.001} +{"train_loss": 0.02823537401854992, "train_loss_bc": 0.02823537401854992, "train_loss_llm": 0.0, "grad_norm": 0.057286642491817474, "global_step": 27710, "epoch": 769, "lr": 0.001} +{"train_loss": 0.029492203146219254, "train_loss_bc": 0.029486872255802155, "train_loss_llm": 0.0005330741405487061, "grad_norm": 0.06480687856674194, "global_step": 27711, "epoch": 769, "lr": 0.001} +{"train_loss": 0.02462175488471985, "train_loss_bc": 0.02462175488471985, "train_loss_llm": 0.0, "grad_norm": 0.08138123899698257, "global_step": 27712, "epoch": 769, "lr": 0.001} +{"train_loss": 0.031380582600831985, "train_loss_bc": 0.03134125471115112, "train_loss_llm": 0.003932952880859375, "grad_norm": 0.051090825349092484, "global_step": 27713, "epoch": 769, "lr": 0.001} +{"train_loss": 0.038112085312604904, "train_loss_bc": 0.038100991398096085, "train_loss_llm": 0.0011094212532043457, "grad_norm": 0.07417894899845123, "global_step": 27714, "epoch": 769, "lr": 0.001} +{"train_loss": 0.030690139159560204, "train_loss_bc": 0.03068382851779461, "train_loss_llm": 0.0006309747695922852, "grad_norm": 0.09981033205986023, "global_step": 27715, "epoch": 769, "lr": 0.001} +{"train_loss": 0.02962910570204258, "train_loss_bc": 0.02962910570204258, "train_loss_llm": 0.0, "grad_norm": 0.12240120768547058, "global_step": 27716, "epoch": 769, "lr": 0.001} +{"train_loss": 0.032233864068984985, "train_loss_bc": 0.032233864068984985, "train_loss_llm": 0.0, "grad_norm": 0.039570536464452744, "global_step": 27717, "epoch": 769, "lr": 0.001} +{"train_loss": 0.02793731354176998, "train_loss_bc": 0.02793731354176998, "train_loss_llm": 0.0, "grad_norm": 0.05677410587668419, "global_step": 27718, "epoch": 769, "lr": 0.001} +{"train_loss": 0.03073273418057296, "train_loss_bc": 0.03300096094608307, "train_loss_llm": 2.0176172256469727e-05, "grad_norm": 0.09753556549549103, "global_step": 27719, "epoch": 769, "lr": 0.001, "val_loss": 0.09467872232198715} +{"train_loss": 0.03174309432506561, "train_loss_bc": 0.0317317433655262, "train_loss_llm": 0.0011352300643920898, "grad_norm": 0.10968971252441406, "global_step": 27720, "epoch": 770, "lr": 0.001} +{"train_loss": 0.030837828293442726, "train_loss_bc": 0.030837828293442726, "train_loss_llm": 0.0, "grad_norm": 0.03952256217598915, "global_step": 27721, "epoch": 770, "lr": 0.001} +{"train_loss": 0.028967643156647682, "train_loss_bc": 0.028967643156647682, "train_loss_llm": 0.0, "grad_norm": 0.05293550342321396, "global_step": 27722, "epoch": 770, "lr": 0.001} +{"train_loss": 0.03553711622953415, "train_loss_bc": 0.03553711622953415, "train_loss_llm": 0.0, "grad_norm": 0.07569419592618942, "global_step": 27723, "epoch": 770, "lr": 0.001} +{"train_loss": 0.035827115178108215, "train_loss_bc": 0.035827115178108215, "train_loss_llm": 0.0, "grad_norm": 0.09974852204322815, "global_step": 27724, "epoch": 770, "lr": 0.001} +{"train_loss": 0.03471151366829872, "train_loss_bc": 0.03471151366829872, "train_loss_llm": 0.0, "grad_norm": 0.041690174490213394, "global_step": 27725, "epoch": 770, "lr": 0.001} +{"train_loss": 0.028666676953434944, "train_loss_bc": 0.02866341732442379, "train_loss_llm": 0.00032597780227661133, "grad_norm": 0.059101272374391556, "global_step": 27726, "epoch": 770, "lr": 0.001} +{"train_loss": 0.029566312208771706, "train_loss_bc": 0.029566312208771706, "train_loss_llm": 0.0, "grad_norm": 0.08426586538553238, "global_step": 27727, "epoch": 770, "lr": 0.001} +{"train_loss": 0.03495872765779495, "train_loss_bc": 0.03495872765779495, "train_loss_llm": 0.0, "grad_norm": 0.10970956832170486, "global_step": 27728, "epoch": 770, "lr": 0.001} +{"train_loss": 0.027048178017139435, "train_loss_bc": 0.027036573737859726, "train_loss_llm": 0.0011604726314544678, "grad_norm": 0.03741013631224632, "global_step": 27729, "epoch": 770, "lr": 0.001} +{"train_loss": 0.029724011197686195, "train_loss_bc": 0.029724011197686195, "train_loss_llm": 0.0, "grad_norm": 0.04875025898218155, "global_step": 27730, "epoch": 770, "lr": 0.001} +{"train_loss": 0.034774646162986755, "train_loss_bc": 0.034774646162986755, "train_loss_llm": 0.0, "grad_norm": 0.06060495227575302, "global_step": 27731, "epoch": 770, "lr": 0.001} +{"train_loss": 0.031184542924165726, "train_loss_bc": 0.031184542924165726, "train_loss_llm": 0.0, "grad_norm": 0.08046306669712067, "global_step": 27732, "epoch": 770, "lr": 0.001} +{"train_loss": 0.03255125507712364, "train_loss_bc": 0.03255125507712364, "train_loss_llm": 0.0, "grad_norm": 0.02788342721760273, "global_step": 27733, "epoch": 770, "lr": 0.001} +{"train_loss": 0.031946130096912384, "train_loss_bc": 0.031946130096912384, "train_loss_llm": 0.0, "grad_norm": 0.04872647672891617, "global_step": 27734, "epoch": 770, "lr": 0.001} +{"train_loss": 0.027186185121536255, "train_loss_bc": 0.027186185121536255, "train_loss_llm": 0.0, "grad_norm": 0.0648462101817131, "global_step": 27735, "epoch": 770, "lr": 0.001} +{"train_loss": 0.028850749135017395, "train_loss_bc": 0.028850749135017395, "train_loss_llm": 0.0, "grad_norm": 0.08292566239833832, "global_step": 27736, "epoch": 770, "lr": 0.001} +{"train_loss": 0.029813598841428757, "train_loss_bc": 0.029813598841428757, "train_loss_llm": 0.0, "grad_norm": 0.03980232775211334, "global_step": 27737, "epoch": 770, "lr": 0.001} +{"train_loss": 0.029657866805791855, "train_loss_bc": 0.029657866805791855, "train_loss_llm": 0.0, "grad_norm": 0.06093503534793854, "global_step": 27738, "epoch": 770, "lr": 0.001} +{"train_loss": 0.03088228404521942, "train_loss_bc": 0.03088228404521942, "train_loss_llm": 0.0, "grad_norm": 0.08039963990449905, "global_step": 27739, "epoch": 770, "lr": 0.001} +{"train_loss": 0.024480383843183517, "train_loss_bc": 0.024480383843183517, "train_loss_llm": 0.0, "grad_norm": 0.10529382526874542, "global_step": 27740, "epoch": 770, "lr": 0.001} +{"train_loss": 0.030893975868821144, "train_loss_bc": 0.030893975868821144, "train_loss_llm": 0.0, "grad_norm": 0.03714773431420326, "global_step": 27741, "epoch": 770, "lr": 0.001} +{"train_loss": 0.026929941028356552, "train_loss_bc": 0.026929941028356552, "train_loss_llm": 0.0, "grad_norm": 0.05515854060649872, "global_step": 27742, "epoch": 770, "lr": 0.001} +{"train_loss": 0.025970133021473885, "train_loss_bc": 0.025970133021473885, "train_loss_llm": 0.0, "grad_norm": 0.07558408379554749, "global_step": 27743, "epoch": 770, "lr": 0.001} +{"train_loss": 0.02736637182533741, "train_loss_bc": 0.02736637182533741, "train_loss_llm": 0.0, "grad_norm": 0.0851612463593483, "global_step": 27744, "epoch": 770, "lr": 0.001} +{"train_loss": 0.028079625219106674, "train_loss_bc": 0.02807893417775631, "train_loss_llm": 6.914138793945312e-05, "grad_norm": 0.04299066215753555, "global_step": 27745, "epoch": 770, "lr": 0.001} +{"train_loss": 0.02898983843624592, "train_loss_bc": 0.02898983843624592, "train_loss_llm": 0.0, "grad_norm": 0.0725645050406456, "global_step": 27746, "epoch": 770, "lr": 0.001} +{"train_loss": 0.03449335694313049, "train_loss_bc": 0.03449335694313049, "train_loss_llm": 0.0, "grad_norm": 0.0941772386431694, "global_step": 27747, "epoch": 770, "lr": 0.001} +{"train_loss": 0.038099776953458786, "train_loss_bc": 0.038099776953458786, "train_loss_llm": 0.0, "grad_norm": 0.11609594523906708, "global_step": 27748, "epoch": 770, "lr": 0.001} +{"train_loss": 0.030782010406255722, "train_loss_bc": 0.030782010406255722, "train_loss_llm": 0.0, "grad_norm": 0.03555372357368469, "global_step": 27749, "epoch": 770, "lr": 0.001} +{"train_loss": 0.03166762366890907, "train_loss_bc": 0.03166762366890907, "train_loss_llm": 0.0, "grad_norm": 0.049151450395584106, "global_step": 27750, "epoch": 770, "lr": 0.001} +{"train_loss": 0.02965502440929413, "train_loss_bc": 0.029642948880791664, "train_loss_llm": 0.001207590103149414, "grad_norm": 0.05998365208506584, "global_step": 27751, "epoch": 770, "lr": 0.001} +{"train_loss": 0.02804761566221714, "train_loss_bc": 0.02804761566221714, "train_loss_llm": 0.0, "grad_norm": 0.0815761536359787, "global_step": 27752, "epoch": 770, "lr": 0.001} +{"train_loss": 0.03455159068107605, "train_loss_bc": 0.03455159068107605, "train_loss_llm": 0.0, "grad_norm": 0.03661451116204262, "global_step": 27753, "epoch": 770, "lr": 0.001} +{"train_loss": 0.02934134565293789, "train_loss_bc": 0.02934134565293789, "train_loss_llm": 0.0, "grad_norm": 0.0523914098739624, "global_step": 27754, "epoch": 770, "lr": 0.001} +{"train_loss": 0.030567155374834936, "train_loss_bc": 0.02662729099392891, "train_loss_llm": 0.0006214380264282227, "grad_norm": 0.061837635934352875, "global_step": 27755, "epoch": 770, "lr": 0.001, "train/cumulative_reward": 65.15851576747148, "train/mean_score": 8.545785923075245, "train/success_rate": 0.2, "test/cumulative_reward": 43.356584560290884, "test/mean_score": 8.664189520562504, "test/success_rate": 0.24, "val_loss": 0.0942593663930893, "train_action_mse_error": 0.03341526538133621} +{"train_loss": 0.02669050171971321, "train_loss_bc": 0.026686077937483788, "train_loss_llm": 0.0004424452781677246, "grad_norm": 0.08089146763086319, "global_step": 27756, "epoch": 771, "lr": 0.001} +{"train_loss": 0.028272617608308792, "train_loss_bc": 0.028272617608308792, "train_loss_llm": 0.0, "grad_norm": 0.02622300386428833, "global_step": 27757, "epoch": 771, "lr": 0.001} +{"train_loss": 0.029442908242344856, "train_loss_bc": 0.029442908242344856, "train_loss_llm": 0.0, "grad_norm": 0.037109069526195526, "global_step": 27758, "epoch": 771, "lr": 0.001} +{"train_loss": 0.03270791843533516, "train_loss_bc": 0.03270791843533516, "train_loss_llm": 0.0, "grad_norm": 0.050488825887441635, "global_step": 27759, "epoch": 771, "lr": 0.001} +{"train_loss": 0.029908714815974236, "train_loss_bc": 0.029908714815974236, "train_loss_llm": 0.0, "grad_norm": 0.06952287256717682, "global_step": 27760, "epoch": 771, "lr": 0.001} +{"train_loss": 0.027567202225327492, "train_loss_bc": 0.027567202225327492, "train_loss_llm": 0.0, "grad_norm": 0.025044968351721764, "global_step": 27761, "epoch": 771, "lr": 0.001} +{"train_loss": 0.037423089146614075, "train_loss_bc": 0.037423089146614075, "train_loss_llm": 0.0, "grad_norm": 0.04142554849386215, "global_step": 27762, "epoch": 771, "lr": 0.001} +{"train_loss": 0.030661704018712044, "train_loss_bc": 0.030649341642856598, "train_loss_llm": 0.0012362897396087646, "grad_norm": 0.053553640842437744, "global_step": 27763, "epoch": 771, "lr": 0.001} +{"train_loss": 0.03157078102231026, "train_loss_bc": 0.03157078102231026, "train_loss_llm": 0.0, "grad_norm": 0.06822890043258667, "global_step": 27764, "epoch": 771, "lr": 0.001} +{"train_loss": 0.03011043556034565, "train_loss_bc": 0.03011043556034565, "train_loss_llm": 0.0, "grad_norm": 0.03198573365807533, "global_step": 27765, "epoch": 771, "lr": 0.001} +{"train_loss": 0.03216128051280975, "train_loss_bc": 0.03216128051280975, "train_loss_llm": 0.0, "grad_norm": 0.04966814070940018, "global_step": 27766, "epoch": 771, "lr": 0.001} +{"train_loss": 0.03460627794265747, "train_loss_bc": 0.03460569307208061, "train_loss_llm": 5.835294723510742e-05, "grad_norm": 0.07365020364522934, "global_step": 27767, "epoch": 771, "lr": 0.001} +{"train_loss": 0.033700622618198395, "train_loss_bc": 0.033700622618198395, "train_loss_llm": 0.0, "grad_norm": 0.09770894795656204, "global_step": 27768, "epoch": 771, "lr": 0.001} +{"train_loss": 0.031326550990343094, "train_loss_bc": 0.031326550990343094, "train_loss_llm": 0.0, "grad_norm": 0.0513179637491703, "global_step": 27769, "epoch": 771, "lr": 0.001} +{"train_loss": 0.03799992799758911, "train_loss_bc": 0.03798432648181915, "train_loss_llm": 0.0015602409839630127, "grad_norm": 0.08251899480819702, "global_step": 27770, "epoch": 771, "lr": 0.001} +{"train_loss": 0.029849065467715263, "train_loss_bc": 0.029849065467715263, "train_loss_llm": 0.0, "grad_norm": 0.10474086552858353, "global_step": 27771, "epoch": 771, "lr": 0.001} +{"train_loss": 0.026104016229510307, "train_loss_bc": 0.026104016229510307, "train_loss_llm": 0.0, "grad_norm": 0.1173301488161087, "global_step": 27772, "epoch": 771, "lr": 0.001} +{"train_loss": 0.028624113649129868, "train_loss_bc": 0.028616780415177345, "train_loss_llm": 0.0007333159446716309, "grad_norm": 0.03242763876914978, "global_step": 27773, "epoch": 771, "lr": 0.001} +{"train_loss": 0.029714569449424744, "train_loss_bc": 0.029714569449424744, "train_loss_llm": 0.0, "grad_norm": 0.044755637645721436, "global_step": 27774, "epoch": 771, "lr": 0.001} +{"train_loss": 0.03150598332285881, "train_loss_bc": 0.03150515630841255, "train_loss_llm": 8.288025856018066e-05, "grad_norm": 0.06901969015598297, "global_step": 27775, "epoch": 771, "lr": 0.001} +{"train_loss": 0.027454763650894165, "train_loss_bc": 0.027454763650894165, "train_loss_llm": 0.0, "grad_norm": 0.08294237405061722, "global_step": 27776, "epoch": 771, "lr": 0.001} +{"train_loss": 0.03086732141673565, "train_loss_bc": 0.03086732141673565, "train_loss_llm": 0.0, "grad_norm": 0.03345152363181114, "global_step": 27777, "epoch": 771, "lr": 0.001} +{"train_loss": 0.026703357696533203, "train_loss_bc": 0.026703357696533203, "train_loss_llm": 0.0, "grad_norm": 0.04796845093369484, "global_step": 27778, "epoch": 771, "lr": 0.001} +{"train_loss": 0.03424282744526863, "train_loss_bc": 0.03424282744526863, "train_loss_llm": 0.0, "grad_norm": 0.06955073028802872, "global_step": 27779, "epoch": 771, "lr": 0.001} +{"train_loss": 0.029651351273059845, "train_loss_bc": 0.029651351273059845, "train_loss_llm": 0.0, "grad_norm": 0.09238523989915848, "global_step": 27780, "epoch": 771, "lr": 0.001} +{"train_loss": 0.03238969296216965, "train_loss_bc": 0.03238969296216965, "train_loss_llm": 0.0, "grad_norm": 0.03566490858793259, "global_step": 27781, "epoch": 771, "lr": 0.001} +{"train_loss": 0.030507391318678856, "train_loss_bc": 0.030507391318678856, "train_loss_llm": 0.0, "grad_norm": 0.04814194515347481, "global_step": 27782, "epoch": 771, "lr": 0.001} +{"train_loss": 0.026106636971235275, "train_loss_bc": 0.026101944968104362, "train_loss_llm": 0.0004691779613494873, "grad_norm": 0.06268732994794846, "global_step": 27783, "epoch": 771, "lr": 0.001} +{"train_loss": 0.02917461469769478, "train_loss_bc": 0.02917461469769478, "train_loss_llm": 0.0, "grad_norm": 0.07211385667324066, "global_step": 27784, "epoch": 771, "lr": 0.001} +{"train_loss": 0.02534162998199463, "train_loss_bc": 0.02534162998199463, "train_loss_llm": 0.0, "grad_norm": 0.031318992376327515, "global_step": 27785, "epoch": 771, "lr": 0.001} +{"train_loss": 0.031792741268873215, "train_loss_bc": 0.031792741268873215, "train_loss_llm": 0.0, "grad_norm": 0.0497465617954731, "global_step": 27786, "epoch": 771, "lr": 0.001} +{"train_loss": 0.029550649225711823, "train_loss_bc": 0.029550649225711823, "train_loss_llm": 0.0, "grad_norm": 0.0643877387046814, "global_step": 27787, "epoch": 771, "lr": 0.001} +{"train_loss": 0.029501069337129593, "train_loss_bc": 0.029501069337129593, "train_loss_llm": 0.0, "grad_norm": 0.07604236155748367, "global_step": 27788, "epoch": 771, "lr": 0.001} +{"train_loss": 0.02921995520591736, "train_loss_bc": 0.029201563447713852, "train_loss_llm": 0.0018391013145446777, "grad_norm": 0.027772465720772743, "global_step": 27789, "epoch": 771, "lr": 0.001} +{"train_loss": 0.0339454747736454, "train_loss_bc": 0.0339343324303627, "train_loss_llm": 0.0011140704154968262, "grad_norm": 0.0551769845187664, "global_step": 27790, "epoch": 771, "lr": 0.001} +{"train_loss": 0.030544295855280425, "train_loss_bc": 0.03319689258933067, "train_loss_llm": 0.0, "grad_norm": 0.08276092261075974, "global_step": 27791, "epoch": 771, "lr": 0.001, "val_loss": 0.0946120098233223} +{"train_loss": 0.03143379092216492, "train_loss_bc": 0.03143040090799332, "train_loss_llm": 0.0003388822078704834, "grad_norm": 0.11449350416660309, "global_step": 27792, "epoch": 772, "lr": 0.001} +{"train_loss": 0.029162302613258362, "train_loss_bc": 0.02915286086499691, "train_loss_llm": 0.0009441673755645752, "grad_norm": 0.026581401005387306, "global_step": 27793, "epoch": 772, "lr": 0.001} +{"train_loss": 0.026812706142663956, "train_loss_bc": 0.026812706142663956, "train_loss_llm": 0.0, "grad_norm": 0.03877854719758034, "global_step": 27794, "epoch": 772, "lr": 0.001} +{"train_loss": 0.02669372782111168, "train_loss_bc": 0.02669372782111168, "train_loss_llm": 0.0, "grad_norm": 0.0511951670050621, "global_step": 27795, "epoch": 772, "lr": 0.001} +{"train_loss": 0.03193075582385063, "train_loss_bc": 0.03193075582385063, "train_loss_llm": 0.0, "grad_norm": 0.07173419743776321, "global_step": 27796, "epoch": 772, "lr": 0.001} +{"train_loss": 0.029666094109416008, "train_loss_bc": 0.029662497341632843, "train_loss_llm": 0.00035959482192993164, "grad_norm": 0.029632404446601868, "global_step": 27797, "epoch": 772, "lr": 0.001} +{"train_loss": 0.026563134044408798, "train_loss_bc": 0.026563134044408798, "train_loss_llm": 0.0, "grad_norm": 0.044963154941797256, "global_step": 27798, "epoch": 772, "lr": 0.001} +{"train_loss": 0.031002063304185867, "train_loss_bc": 0.03099711425602436, "train_loss_llm": 0.0004949569702148438, "grad_norm": 0.06405039876699448, "global_step": 27799, "epoch": 772, "lr": 0.001} +{"train_loss": 0.02572435513138771, "train_loss_bc": 0.02572435513138771, "train_loss_llm": 0.0, "grad_norm": 0.07008028775453568, "global_step": 27800, "epoch": 772, "lr": 0.001} +{"train_loss": 0.029056856408715248, "train_loss_bc": 0.029042374342679977, "train_loss_llm": 0.001448124647140503, "grad_norm": 0.036369286477565765, "global_step": 27801, "epoch": 772, "lr": 0.001} +{"train_loss": 0.03349744528532028, "train_loss_bc": 0.03349744528532028, "train_loss_llm": 0.0, "grad_norm": 0.054882343858480453, "global_step": 27802, "epoch": 772, "lr": 0.001} +{"train_loss": 0.03589949011802673, "train_loss_bc": 0.03589949011802673, "train_loss_llm": 0.0, "grad_norm": 0.08124949783086777, "global_step": 27803, "epoch": 772, "lr": 0.001} +{"train_loss": 0.027459323406219482, "train_loss_bc": 0.027451135218143463, "train_loss_llm": 0.0008188188076019287, "grad_norm": 0.08825403451919556, "global_step": 27804, "epoch": 772, "lr": 0.001} +{"train_loss": 0.03213014453649521, "train_loss_bc": 0.03211430832743645, "train_loss_llm": 0.0015837550163269043, "grad_norm": 0.04064369574189186, "global_step": 27805, "epoch": 772, "lr": 0.001} +{"train_loss": 0.02962784469127655, "train_loss_bc": 0.02962784469127655, "train_loss_llm": 0.0, "grad_norm": 0.07509535551071167, "global_step": 27806, "epoch": 772, "lr": 0.001} +{"train_loss": 0.0322200208902359, "train_loss_bc": 0.0322200208902359, "train_loss_llm": 0.0, "grad_norm": 0.08656172454357147, "global_step": 27807, "epoch": 772, "lr": 0.001} +{"train_loss": 0.028182098641991615, "train_loss_bc": 0.028182098641991615, "train_loss_llm": 0.0, "grad_norm": 0.09494156390428543, "global_step": 27808, "epoch": 772, "lr": 0.001} +{"train_loss": 0.029663890600204468, "train_loss_bc": 0.029663890600204468, "train_loss_llm": 0.0, "grad_norm": 0.03749255836009979, "global_step": 27809, "epoch": 772, "lr": 0.001} +{"train_loss": 0.034594014286994934, "train_loss_bc": 0.034594014286994934, "train_loss_llm": 0.0, "grad_norm": 0.05934879556298256, "global_step": 27810, "epoch": 772, "lr": 0.001} +{"train_loss": 0.0333210751414299, "train_loss_bc": 0.0333210751414299, "train_loss_llm": 0.0, "grad_norm": 0.08077068626880646, "global_step": 27811, "epoch": 772, "lr": 0.001} +{"train_loss": 0.028869036585092545, "train_loss_bc": 0.028865572065114975, "train_loss_llm": 0.000346451997756958, "grad_norm": 0.0991072878241539, "global_step": 27812, "epoch": 772, "lr": 0.001} +{"train_loss": 0.029989823698997498, "train_loss_bc": 0.029989823698997498, "train_loss_llm": 0.0, "grad_norm": 0.04059799388051033, "global_step": 27813, "epoch": 772, "lr": 0.001} +{"train_loss": 0.03804543986916542, "train_loss_bc": 0.03803990036249161, "train_loss_llm": 0.0005538463592529297, "grad_norm": 0.06426529586315155, "global_step": 27814, "epoch": 772, "lr": 0.001} +{"train_loss": 0.025786032900214195, "train_loss_bc": 0.025763344019651413, "train_loss_llm": 0.002268850803375244, "grad_norm": 0.07297530770301819, "global_step": 27815, "epoch": 772, "lr": 0.001} +{"train_loss": 0.029225733131170273, "train_loss_bc": 0.02922171726822853, "train_loss_llm": 0.0004016757011413574, "grad_norm": 0.08510784059762955, "global_step": 27816, "epoch": 772, "lr": 0.001} +{"train_loss": 0.03167147561907768, "train_loss_bc": 0.03167147561907768, "train_loss_llm": 0.0, "grad_norm": 0.044755078852176666, "global_step": 27817, "epoch": 772, "lr": 0.001} +{"train_loss": 0.036678116768598557, "train_loss_bc": 0.036678116768598557, "train_loss_llm": 0.0, "grad_norm": 0.06466521322727203, "global_step": 27818, "epoch": 772, "lr": 0.001} +{"train_loss": 0.02211562544107437, "train_loss_bc": 0.02211562544107437, "train_loss_llm": 0.0, "grad_norm": 0.07617425918579102, "global_step": 27819, "epoch": 772, "lr": 0.001} +{"train_loss": 0.03497089073061943, "train_loss_bc": 0.03497089073061943, "train_loss_llm": 0.0, "grad_norm": 0.09843911230564117, "global_step": 27820, "epoch": 772, "lr": 0.001} +{"train_loss": 0.028843076899647713, "train_loss_bc": 0.028843076899647713, "train_loss_llm": 0.0, "grad_norm": 0.03603305295109749, "global_step": 27821, "epoch": 772, "lr": 0.001} +{"train_loss": 0.03418666869401932, "train_loss_bc": 0.03418666869401932, "train_loss_llm": 0.0, "grad_norm": 0.059482648968696594, "global_step": 27822, "epoch": 772, "lr": 0.001} +{"train_loss": 0.028220992535352707, "train_loss_bc": 0.028220992535352707, "train_loss_llm": 0.0, "grad_norm": 0.08238881826400757, "global_step": 27823, "epoch": 772, "lr": 0.001} +{"train_loss": 0.032356198877096176, "train_loss_bc": 0.032356198877096176, "train_loss_llm": 0.0, "grad_norm": 0.11288569122552872, "global_step": 27824, "epoch": 772, "lr": 0.001} +{"train_loss": 0.02896382100880146, "train_loss_bc": 0.02896382100880146, "train_loss_llm": 0.0, "grad_norm": 0.028215765953063965, "global_step": 27825, "epoch": 772, "lr": 0.001} +{"train_loss": 0.030899014323949814, "train_loss_bc": 0.030899014323949814, "train_loss_llm": 0.0, "grad_norm": 0.05526380613446236, "global_step": 27826, "epoch": 772, "lr": 0.001} +{"train_loss": 0.030589490611520078, "train_loss_bc": 0.03575858101248741, "train_loss_llm": 0.0, "grad_norm": 0.07784683257341385, "global_step": 27827, "epoch": 772, "lr": 0.001, "val_loss": 0.09701882302761078} +{"train_loss": 0.032452940940856934, "train_loss_bc": 0.032452940940856934, "train_loss_llm": 0.0, "grad_norm": 0.10110319405794144, "global_step": 27828, "epoch": 773, "lr": 0.001} +{"train_loss": 0.034616418182849884, "train_loss_bc": 0.03457754850387573, "train_loss_llm": 0.0038871169090270996, "grad_norm": 0.050396811217069626, "global_step": 27829, "epoch": 773, "lr": 0.001} +{"train_loss": 0.03076554648578167, "train_loss_bc": 0.03075840324163437, "train_loss_llm": 0.0007143318653106689, "grad_norm": 0.07879605144262314, "global_step": 27830, "epoch": 773, "lr": 0.001} +{"train_loss": 0.031589575111866, "train_loss_bc": 0.031589575111866, "train_loss_llm": 0.0, "grad_norm": 0.1087174043059349, "global_step": 27831, "epoch": 773, "lr": 0.001} +{"train_loss": 0.030409831553697586, "train_loss_bc": 0.030409831553697586, "train_loss_llm": 0.0, "grad_norm": 0.13617634773254395, "global_step": 27832, "epoch": 773, "lr": 0.001} +{"train_loss": 0.034190207719802856, "train_loss_bc": 0.034190207719802856, "train_loss_llm": 0.0, "grad_norm": 0.04096192121505737, "global_step": 27833, "epoch": 773, "lr": 0.001} +{"train_loss": 0.0245802104473114, "train_loss_bc": 0.0245802104473114, "train_loss_llm": 0.0, "grad_norm": 0.05193905159831047, "global_step": 27834, "epoch": 773, "lr": 0.001} +{"train_loss": 0.03494556248188019, "train_loss_bc": 0.03494556248188019, "train_loss_llm": 0.0, "grad_norm": 0.07563499361276627, "global_step": 27835, "epoch": 773, "lr": 0.001} +{"train_loss": 0.029010774567723274, "train_loss_bc": 0.029010774567723274, "train_loss_llm": 0.0, "grad_norm": 0.09192319959402084, "global_step": 27836, "epoch": 773, "lr": 0.001} +{"train_loss": 0.030261728912591934, "train_loss_bc": 0.030257757753133774, "train_loss_llm": 0.00039714574813842773, "grad_norm": 0.03211770951747894, "global_step": 27837, "epoch": 773, "lr": 0.001} +{"train_loss": 0.03607459366321564, "train_loss_bc": 0.03607459366321564, "train_loss_llm": 0.0, "grad_norm": 0.06014573946595192, "global_step": 27838, "epoch": 773, "lr": 0.001} +{"train_loss": 0.032427601516246796, "train_loss_bc": 0.032427601516246796, "train_loss_llm": 0.0, "grad_norm": 0.08231411874294281, "global_step": 27839, "epoch": 773, "lr": 0.001} +{"train_loss": 0.02805262617766857, "train_loss_bc": 0.02805262617766857, "train_loss_llm": 0.0, "grad_norm": 0.1035960465669632, "global_step": 27840, "epoch": 773, "lr": 0.001} +{"train_loss": 0.027457060292363167, "train_loss_bc": 0.027457060292363167, "train_loss_llm": 0.0, "grad_norm": 0.03637213632464409, "global_step": 27841, "epoch": 773, "lr": 0.001} +{"train_loss": 0.03621522709727287, "train_loss_bc": 0.03620683401823044, "train_loss_llm": 0.0008394420146942139, "grad_norm": 0.07203114032745361, "global_step": 27842, "epoch": 773, "lr": 0.001} +{"train_loss": 0.029513437300920486, "train_loss_bc": 0.02951250597834587, "train_loss_llm": 9.316205978393555e-05, "grad_norm": 0.08660165220499039, "global_step": 27843, "epoch": 773, "lr": 0.001} +{"train_loss": 0.03554002195596695, "train_loss_bc": 0.03554002195596695, "train_loss_llm": 0.0, "grad_norm": 0.11518003046512604, "global_step": 27844, "epoch": 773, "lr": 0.001} +{"train_loss": 0.028315791860222816, "train_loss_bc": 0.028315791860222816, "train_loss_llm": 0.0, "grad_norm": 0.03748461231589317, "global_step": 27845, "epoch": 773, "lr": 0.001} +{"train_loss": 0.03564591705799103, "train_loss_bc": 0.03564591705799103, "train_loss_llm": 0.0, "grad_norm": 0.08583562076091766, "global_step": 27846, "epoch": 773, "lr": 0.001} +{"train_loss": 0.035862989723682404, "train_loss_bc": 0.035862989723682404, "train_loss_llm": 0.0, "grad_norm": 0.13031189143657684, "global_step": 27847, "epoch": 773, "lr": 0.001} +{"train_loss": 0.026550281792879105, "train_loss_bc": 0.026550281792879105, "train_loss_llm": 0.0, "grad_norm": 0.15979616343975067, "global_step": 27848, "epoch": 773, "lr": 0.001} +{"train_loss": 0.03162440285086632, "train_loss_bc": 0.03162440285086632, "train_loss_llm": 0.0, "grad_norm": 0.031699057668447495, "global_step": 27849, "epoch": 773, "lr": 0.001} +{"train_loss": 0.02869831584393978, "train_loss_bc": 0.02868502214550972, "train_loss_llm": 0.0013293921947479248, "grad_norm": 0.05875971540808678, "global_step": 27850, "epoch": 773, "lr": 0.001} +{"train_loss": 0.02544957958161831, "train_loss_bc": 0.02544957958161831, "train_loss_llm": 0.0, "grad_norm": 0.08163083344697952, "global_step": 27851, "epoch": 773, "lr": 0.001} +{"train_loss": 0.035370174795389175, "train_loss_bc": 0.03535740450024605, "train_loss_llm": 0.0012771785259246826, "grad_norm": 0.1036219596862793, "global_step": 27852, "epoch": 773, "lr": 0.001} +{"train_loss": 0.03348424285650253, "train_loss_bc": 0.033474355936050415, "train_loss_llm": 0.000988602638244629, "grad_norm": 0.03453746438026428, "global_step": 27853, "epoch": 773, "lr": 0.001} +{"train_loss": 0.030587583780288696, "train_loss_bc": 0.030581848695874214, "train_loss_llm": 0.0005734562873840332, "grad_norm": 0.04451626166701317, "global_step": 27854, "epoch": 773, "lr": 0.001} +{"train_loss": 0.030956821516156197, "train_loss_bc": 0.030956821516156197, "train_loss_llm": 0.0, "grad_norm": 0.06118138134479523, "global_step": 27855, "epoch": 773, "lr": 0.001} +{"train_loss": 0.028229428455233574, "train_loss_bc": 0.028229428455233574, "train_loss_llm": 0.0, "grad_norm": 0.07381550967693329, "global_step": 27856, "epoch": 773, "lr": 0.001} +{"train_loss": 0.020536258816719055, "train_loss_bc": 0.020535727962851524, "train_loss_llm": 5.307793617248535e-05, "grad_norm": 0.02717544510960579, "global_step": 27857, "epoch": 773, "lr": 0.001} +{"train_loss": 0.033480312675237656, "train_loss_bc": 0.033480312675237656, "train_loss_llm": 0.0, "grad_norm": 0.04340318217873573, "global_step": 27858, "epoch": 773, "lr": 0.001} +{"train_loss": 0.03242366015911102, "train_loss_bc": 0.03242366015911102, "train_loss_llm": 0.0, "grad_norm": 0.05857575684785843, "global_step": 27859, "epoch": 773, "lr": 0.001} +{"train_loss": 0.030247174203395844, "train_loss_bc": 0.030247174203395844, "train_loss_llm": 0.0, "grad_norm": 0.0732874944806099, "global_step": 27860, "epoch": 773, "lr": 0.001} +{"train_loss": 0.028215592727065086, "train_loss_bc": 0.02821105532348156, "train_loss_llm": 0.0004538297653198242, "grad_norm": 0.03276799991726875, "global_step": 27861, "epoch": 773, "lr": 0.001} +{"train_loss": 0.027320537716150284, "train_loss_bc": 0.027320537716150284, "train_loss_llm": 0.0, "grad_norm": 0.0524463914334774, "global_step": 27862, "epoch": 773, "lr": 0.001} +{"train_loss": 0.03070838511404064, "train_loss_bc": 0.02439943328499794, "train_loss_llm": 0.0, "grad_norm": 0.0624878965318203, "global_step": 27863, "epoch": 773, "lr": 0.001, "val_loss": 0.0939728245139122} +{"train_loss": 0.03096833825111389, "train_loss_bc": 0.03096833825111389, "train_loss_llm": 0.0, "grad_norm": 0.08058292418718338, "global_step": 27864, "epoch": 774, "lr": 0.001} +{"train_loss": 0.033099569380283356, "train_loss_bc": 0.033099569380283356, "train_loss_llm": 0.0, "grad_norm": 0.053557414561510086, "global_step": 27865, "epoch": 774, "lr": 0.001} +{"train_loss": 0.02971261367201805, "train_loss_bc": 0.02971261367201805, "train_loss_llm": 0.0, "grad_norm": 0.07967700064182281, "global_step": 27866, "epoch": 774, "lr": 0.001} +{"train_loss": 0.024591568857431412, "train_loss_bc": 0.024570433422923088, "train_loss_llm": 0.0021134912967681885, "grad_norm": 0.09191898256540298, "global_step": 27867, "epoch": 774, "lr": 0.001} +{"train_loss": 0.02850167080760002, "train_loss_bc": 0.02850167080760002, "train_loss_llm": 0.0, "grad_norm": 0.1099388599395752, "global_step": 27868, "epoch": 774, "lr": 0.001} +{"train_loss": 0.029435250908136368, "train_loss_bc": 0.02942800708115101, "train_loss_llm": 0.00072440505027771, "grad_norm": 0.03586718812584877, "global_step": 27869, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03135713189840317, "train_loss_bc": 0.03132260590791702, "train_loss_llm": 0.003452509641647339, "grad_norm": 0.06140589714050293, "global_step": 27870, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03251432999968529, "train_loss_bc": 0.03251432999968529, "train_loss_llm": 0.0, "grad_norm": 0.07667598128318787, "global_step": 27871, "epoch": 774, "lr": 0.001} +{"train_loss": 0.028814647346735, "train_loss_bc": 0.028814647346735, "train_loss_llm": 0.0, "grad_norm": 0.09782962501049042, "global_step": 27872, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03270871564745903, "train_loss_bc": 0.03270871564745903, "train_loss_llm": 0.0, "grad_norm": 0.038522690534591675, "global_step": 27873, "epoch": 774, "lr": 0.001} +{"train_loss": 0.029640398919582367, "train_loss_bc": 0.02963772416114807, "train_loss_llm": 0.0002675652503967285, "grad_norm": 0.061756640672683716, "global_step": 27874, "epoch": 774, "lr": 0.001} +{"train_loss": 0.0351959727704525, "train_loss_bc": 0.0351959727704525, "train_loss_llm": 0.0, "grad_norm": 0.07555107027292252, "global_step": 27875, "epoch": 774, "lr": 0.001} +{"train_loss": 0.026547690853476524, "train_loss_bc": 0.026547690853476524, "train_loss_llm": 0.0, "grad_norm": 0.09873385727405548, "global_step": 27876, "epoch": 774, "lr": 0.001} +{"train_loss": 0.0321330651640892, "train_loss_bc": 0.0321330651640892, "train_loss_llm": 0.0, "grad_norm": 0.027142800390720367, "global_step": 27877, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03034938871860504, "train_loss_bc": 0.030343394726514816, "train_loss_llm": 0.0005993247032165527, "grad_norm": 0.04288623854517937, "global_step": 27878, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03328617662191391, "train_loss_bc": 0.03328617662191391, "train_loss_llm": 0.0, "grad_norm": 0.06700438261032104, "global_step": 27879, "epoch": 774, "lr": 0.001} +{"train_loss": 0.029111934825778008, "train_loss_bc": 0.029111934825778008, "train_loss_llm": 0.0, "grad_norm": 0.07175296545028687, "global_step": 27880, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03461229056119919, "train_loss_bc": 0.03461229056119919, "train_loss_llm": 0.0, "grad_norm": 0.04898503050208092, "global_step": 27881, "epoch": 774, "lr": 0.001} +{"train_loss": 0.02451472543179989, "train_loss_bc": 0.02451472543179989, "train_loss_llm": 0.0, "grad_norm": 0.06365534663200378, "global_step": 27882, "epoch": 774, "lr": 0.001} +{"train_loss": 0.035488780587911606, "train_loss_bc": 0.035488780587911606, "train_loss_llm": 0.0, "grad_norm": 0.08444476872682571, "global_step": 27883, "epoch": 774, "lr": 0.001} +{"train_loss": 0.026474127545952797, "train_loss_bc": 0.026474127545952797, "train_loss_llm": 0.0, "grad_norm": 0.10504326224327087, "global_step": 27884, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03928854316473007, "train_loss_bc": 0.03928835690021515, "train_loss_llm": 1.856684684753418e-05, "grad_norm": 0.03956087306141853, "global_step": 27885, "epoch": 774, "lr": 0.001} +{"train_loss": 0.032751940190792084, "train_loss_bc": 0.032744038850069046, "train_loss_llm": 0.0007902383804321289, "grad_norm": 0.06631015241146088, "global_step": 27886, "epoch": 774, "lr": 0.001} +{"train_loss": 0.02308758720755577, "train_loss_bc": 0.023082612082362175, "train_loss_llm": 0.0004974305629730225, "grad_norm": 0.08004459738731384, "global_step": 27887, "epoch": 774, "lr": 0.001} +{"train_loss": 0.030056625604629517, "train_loss_bc": 0.030056625604629517, "train_loss_llm": 0.0, "grad_norm": 0.110390804708004, "global_step": 27888, "epoch": 774, "lr": 0.001} +{"train_loss": 0.028695860877633095, "train_loss_bc": 0.028695860877633095, "train_loss_llm": 0.0, "grad_norm": 0.032780032604932785, "global_step": 27889, "epoch": 774, "lr": 0.001} +{"train_loss": 0.029127663001418114, "train_loss_bc": 0.029127663001418114, "train_loss_llm": 0.0, "grad_norm": 0.04509437084197998, "global_step": 27890, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03344536945223808, "train_loss_bc": 0.03344536945223808, "train_loss_llm": 0.0, "grad_norm": 0.0590168796479702, "global_step": 27891, "epoch": 774, "lr": 0.001} +{"train_loss": 0.026649655774235725, "train_loss_bc": 0.026649655774235725, "train_loss_llm": 0.0, "grad_norm": 0.07958182692527771, "global_step": 27892, "epoch": 774, "lr": 0.001} +{"train_loss": 0.033383507281541824, "train_loss_bc": 0.033383507281541824, "train_loss_llm": 0.0, "grad_norm": 0.038127440959215164, "global_step": 27893, "epoch": 774, "lr": 0.001} +{"train_loss": 0.02754014916718006, "train_loss_bc": 0.02754014916718006, "train_loss_llm": 0.0, "grad_norm": 0.050897594541311264, "global_step": 27894, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03185714781284332, "train_loss_bc": 0.03185714781284332, "train_loss_llm": 0.0, "grad_norm": 0.06784103810787201, "global_step": 27895, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03738509863615036, "train_loss_bc": 0.03736712783575058, "train_loss_llm": 0.0017970502376556396, "grad_norm": 0.08113353699445724, "global_step": 27896, "epoch": 774, "lr": 0.001} +{"train_loss": 0.026546098291873932, "train_loss_bc": 0.026546098291873932, "train_loss_llm": 0.0, "grad_norm": 0.029815519228577614, "global_step": 27897, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03298766911029816, "train_loss_bc": 0.03298228606581688, "train_loss_llm": 0.0005384385585784912, "grad_norm": 0.04784495383501053, "global_step": 27898, "epoch": 774, "lr": 0.001} +{"train_loss": 0.03047249274742272, "train_loss_bc": 0.025148434564471245, "train_loss_llm": 0.0, "grad_norm": 0.06857674568891525, "global_step": 27899, "epoch": 774, "lr": 0.001, "val_loss": 0.09471873939037323} +{"train_loss": 0.02488652989268303, "train_loss_bc": 0.02488652989268303, "train_loss_llm": 0.0, "grad_norm": 0.09274160116910934, "global_step": 27900, "epoch": 775, "lr": 0.001} +{"train_loss": 0.03039470687508583, "train_loss_bc": 0.03039470687508583, "train_loss_llm": 0.0, "grad_norm": 0.027750281617045403, "global_step": 27901, "epoch": 775, "lr": 0.001} +{"train_loss": 0.03159519284963608, "train_loss_bc": 0.03159519284963608, "train_loss_llm": 0.0, "grad_norm": 0.03786462917923927, "global_step": 27902, "epoch": 775, "lr": 0.001} +{"train_loss": 0.025007149204611778, "train_loss_bc": 0.025007149204611778, "train_loss_llm": 0.0, "grad_norm": 0.05211612582206726, "global_step": 27903, "epoch": 775, "lr": 0.001} +{"train_loss": 0.028036504983901978, "train_loss_bc": 0.028036504983901978, "train_loss_llm": 0.0, "grad_norm": 0.06784852594137192, "global_step": 27904, "epoch": 775, "lr": 0.001} +{"train_loss": 0.025964098051190376, "train_loss_bc": 0.025958266109228134, "train_loss_llm": 0.0005832016468048096, "grad_norm": 0.039099059998989105, "global_step": 27905, "epoch": 775, "lr": 0.001} +{"train_loss": 0.029144084081053734, "train_loss_bc": 0.029144084081053734, "train_loss_llm": 0.0, "grad_norm": 0.054879993200302124, "global_step": 27906, "epoch": 775, "lr": 0.001} +{"train_loss": 0.029745731502771378, "train_loss_bc": 0.029745731502771378, "train_loss_llm": 0.0, "grad_norm": 0.0638071596622467, "global_step": 27907, "epoch": 775, "lr": 0.001} +{"train_loss": 0.033025939017534256, "train_loss_bc": 0.033025939017534256, "train_loss_llm": 0.0, "grad_norm": 0.07565806061029434, "global_step": 27908, "epoch": 775, "lr": 0.001} +{"train_loss": 0.040832389146089554, "train_loss_bc": 0.040832389146089554, "train_loss_llm": 0.0, "grad_norm": 0.03693242371082306, "global_step": 27909, "epoch": 775, "lr": 0.001} +{"train_loss": 0.02754545770585537, "train_loss_bc": 0.02754545770585537, "train_loss_llm": 0.0, "grad_norm": 0.04798835515975952, "global_step": 27910, "epoch": 775, "lr": 0.001} +{"train_loss": 0.029966209083795547, "train_loss_bc": 0.029946547001600266, "train_loss_llm": 0.001966148614883423, "grad_norm": 0.06489625573158264, "global_step": 27911, "epoch": 775, "lr": 0.001} +{"train_loss": 0.03247600421309471, "train_loss_bc": 0.032472122460603714, "train_loss_llm": 0.00038802623748779297, "grad_norm": 0.07211855798959732, "global_step": 27912, "epoch": 775, "lr": 0.001} +{"train_loss": 0.03092162497341633, "train_loss_bc": 0.030911002308130264, "train_loss_llm": 0.0010622143745422363, "grad_norm": 0.035763975232839584, "global_step": 27913, "epoch": 775, "lr": 0.001} +{"train_loss": 0.036088064312934875, "train_loss_bc": 0.036088064312934875, "train_loss_llm": 0.0, "grad_norm": 0.06032443791627884, "global_step": 27914, "epoch": 775, "lr": 0.001} +{"train_loss": 0.029362553730607033, "train_loss_bc": 0.029356518760323524, "train_loss_llm": 0.0006035268306732178, "grad_norm": 0.06962880492210388, "global_step": 27915, "epoch": 775, "lr": 0.001} +{"train_loss": 0.03428851440548897, "train_loss_bc": 0.03428284078836441, "train_loss_llm": 0.0005672276020050049, "grad_norm": 0.08291351795196533, "global_step": 27916, "epoch": 775, "lr": 0.001} +{"train_loss": 0.030652130022644997, "train_loss_bc": 0.030652130022644997, "train_loss_llm": 0.0, "grad_norm": 0.031241105869412422, "global_step": 27917, "epoch": 775, "lr": 0.001} +{"train_loss": 0.032403092831373215, "train_loss_bc": 0.032403092831373215, "train_loss_llm": 0.0, "grad_norm": 0.042929865419864655, "global_step": 27918, "epoch": 775, "lr": 0.001} +{"train_loss": 0.031266920268535614, "train_loss_bc": 0.031266920268535614, "train_loss_llm": 0.0, "grad_norm": 0.05945443734526634, "global_step": 27919, "epoch": 775, "lr": 0.001} +{"train_loss": 0.027991170063614845, "train_loss_bc": 0.027991170063614845, "train_loss_llm": 0.0, "grad_norm": 0.07132594287395477, "global_step": 27920, "epoch": 775, "lr": 0.001} +{"train_loss": 0.027666930109262466, "train_loss_bc": 0.027666930109262466, "train_loss_llm": 0.0, "grad_norm": 0.04007677733898163, "global_step": 27921, "epoch": 775, "lr": 0.001} +{"train_loss": 0.03257784992456436, "train_loss_bc": 0.03257184475660324, "train_loss_llm": 0.0006006062030792236, "grad_norm": 0.05740325152873993, "global_step": 27922, "epoch": 775, "lr": 0.001} +{"train_loss": 0.03138373792171478, "train_loss_bc": 0.031382255256175995, "train_loss_llm": 0.00014823675155639648, "grad_norm": 0.09344475716352463, "global_step": 27923, "epoch": 775, "lr": 0.001} +{"train_loss": 0.03138047456741333, "train_loss_bc": 0.03138047456741333, "train_loss_llm": 0.0, "grad_norm": 0.10152952373027802, "global_step": 27924, "epoch": 775, "lr": 0.001} +{"train_loss": 0.037230897694826126, "train_loss_bc": 0.037229109555482864, "train_loss_llm": 0.0001786351203918457, "grad_norm": 0.036473654210567474, "global_step": 27925, "epoch": 775, "lr": 0.001} +{"train_loss": 0.0282068382948637, "train_loss_bc": 0.0282068382948637, "train_loss_llm": 0.0, "grad_norm": 0.05244584381580353, "global_step": 27926, "epoch": 775, "lr": 0.001} +{"train_loss": 0.025137724354863167, "train_loss_bc": 0.025118626654148102, "train_loss_llm": 0.0019097328186035156, "grad_norm": 0.058553826063871384, "global_step": 27927, "epoch": 775, "lr": 0.001} +{"train_loss": 0.032533109188079834, "train_loss_bc": 0.032533109188079834, "train_loss_llm": 0.0, "grad_norm": 0.06932946294546127, "global_step": 27928, "epoch": 775, "lr": 0.001} +{"train_loss": 0.023911109194159508, "train_loss_bc": 0.023911109194159508, "train_loss_llm": 0.0, "grad_norm": 0.025647595524787903, "global_step": 27929, "epoch": 775, "lr": 0.001} +{"train_loss": 0.03513763099908829, "train_loss_bc": 0.03513763099908829, "train_loss_llm": 0.0, "grad_norm": 0.05448317155241966, "global_step": 27930, "epoch": 775, "lr": 0.001} +{"train_loss": 0.025651780888438225, "train_loss_bc": 0.025651780888438225, "train_loss_llm": 0.0, "grad_norm": 0.061342690140008926, "global_step": 27931, "epoch": 775, "lr": 0.001} +{"train_loss": 0.02859673649072647, "train_loss_bc": 0.02859322726726532, "train_loss_llm": 0.0003509819507598877, "grad_norm": 0.074180968105793, "global_step": 27932, "epoch": 775, "lr": 0.001} +{"train_loss": 0.027432644739747047, "train_loss_bc": 0.027432644739747047, "train_loss_llm": 0.0, "grad_norm": 0.030085092410445213, "global_step": 27933, "epoch": 775, "lr": 0.001} +{"train_loss": 0.029785148799419403, "train_loss_bc": 0.029768921434879303, "train_loss_llm": 0.0016226470470428467, "grad_norm": 0.04914020001888275, "global_step": 27934, "epoch": 775, "lr": 0.001} +{"train_loss": 0.030266707090453968, "train_loss_bc": 0.03137477487325668, "train_loss_llm": 0.0, "grad_norm": 0.05954194813966751, "global_step": 27935, "epoch": 775, "lr": 0.001, "train/cumulative_reward": 51.68498450150522, "train/mean_score": 20.82740798144411, "train/success_rate": 0.6, "test/cumulative_reward": 49.34837396457659, "test/mean_score": 7.844828269564804, "test/success_rate": 0.16, "val_loss": 0.09433288127183914, "train_action_mse_error": 0.033604368567466736} +{"train_loss": 0.028882216662168503, "train_loss_bc": 0.028882216662168503, "train_loss_llm": 0.0, "grad_norm": 0.07142399251461029, "global_step": 27936, "epoch": 776, "lr": 0.001} +{"train_loss": 0.026267394423484802, "train_loss_bc": 0.02625637874007225, "train_loss_llm": 0.0011015236377716064, "grad_norm": 0.035316869616508484, "global_step": 27937, "epoch": 776, "lr": 0.001} +{"train_loss": 0.026167433708906174, "train_loss_bc": 0.026164554059505463, "train_loss_llm": 0.0002880394458770752, "grad_norm": 0.058474570512771606, "global_step": 27938, "epoch": 776, "lr": 0.001} +{"train_loss": 0.037528492510318756, "train_loss_bc": 0.0375279001891613, "train_loss_llm": 5.9157609939575195e-05, "grad_norm": 0.08787840604782104, "global_step": 27939, "epoch": 776, "lr": 0.001} +{"train_loss": 0.02708149142563343, "train_loss_bc": 0.02708149142563343, "train_loss_llm": 0.0, "grad_norm": 0.09255407750606537, "global_step": 27940, "epoch": 776, "lr": 0.001} +{"train_loss": 0.02555559016764164, "train_loss_bc": 0.02555559016764164, "train_loss_llm": 0.0, "grad_norm": 0.025987306609749794, "global_step": 27941, "epoch": 776, "lr": 0.001} +{"train_loss": 0.03400786966085434, "train_loss_bc": 0.03400786966085434, "train_loss_llm": 0.0, "grad_norm": 0.04253591224551201, "global_step": 27942, "epoch": 776, "lr": 0.001} +{"train_loss": 0.03274845704436302, "train_loss_bc": 0.03274845704436302, "train_loss_llm": 0.0, "grad_norm": 0.05821749195456505, "global_step": 27943, "epoch": 776, "lr": 0.001} +{"train_loss": 0.03128330036997795, "train_loss_bc": 0.03128330036997795, "train_loss_llm": 0.0, "grad_norm": 0.0722988024353981, "global_step": 27944, "epoch": 776, "lr": 0.001} +{"train_loss": 0.029916048049926758, "train_loss_bc": 0.029916048049926758, "train_loss_llm": 0.0, "grad_norm": 0.041526198387145996, "global_step": 27945, "epoch": 776, "lr": 0.001} +{"train_loss": 0.03123290278017521, "train_loss_bc": 0.03123290278017521, "train_loss_llm": 0.0, "grad_norm": 0.07036568224430084, "global_step": 27946, "epoch": 776, "lr": 0.001} +{"train_loss": 0.02554752677679062, "train_loss_bc": 0.02554752677679062, "train_loss_llm": 0.0, "grad_norm": 0.07988999038934708, "global_step": 27947, "epoch": 776, "lr": 0.001} +{"train_loss": 0.027859754860401154, "train_loss_bc": 0.027859754860401154, "train_loss_llm": 0.0, "grad_norm": 0.09449579566717148, "global_step": 27948, "epoch": 776, "lr": 0.001} +{"train_loss": 0.024942276999354362, "train_loss_bc": 0.024942276999354362, "train_loss_llm": 0.0, "grad_norm": 0.02663009613752365, "global_step": 27949, "epoch": 776, "lr": 0.001} +{"train_loss": 0.029648633673787117, "train_loss_bc": 0.029648633673787117, "train_loss_llm": 0.0, "grad_norm": 0.043905749917030334, "global_step": 27950, "epoch": 776, "lr": 0.001} +{"train_loss": 0.02857995592057705, "train_loss_bc": 0.02857321873307228, "train_loss_llm": 0.0006738007068634033, "grad_norm": 0.06013726443052292, "global_step": 27951, "epoch": 776, "lr": 0.001} +{"train_loss": 0.03108326345682144, "train_loss_bc": 0.03108326345682144, "train_loss_llm": 0.0, "grad_norm": 0.06392250955104828, "global_step": 27952, "epoch": 776, "lr": 0.001} +{"train_loss": 0.03336566314101219, "train_loss_bc": 0.03336566314101219, "train_loss_llm": 0.0, "grad_norm": 0.02920253947377205, "global_step": 27953, "epoch": 776, "lr": 0.001} +{"train_loss": 0.031089777126908302, "train_loss_bc": 0.031071728095412254, "train_loss_llm": 0.0018049180507659912, "grad_norm": 0.037943143397569656, "global_step": 27954, "epoch": 776, "lr": 0.001} +{"train_loss": 0.030373092740774155, "train_loss_bc": 0.030373092740774155, "train_loss_llm": 0.0, "grad_norm": 0.05093783512711525, "global_step": 27955, "epoch": 776, "lr": 0.001} +{"train_loss": 0.033983826637268066, "train_loss_bc": 0.033983826637268066, "train_loss_llm": 0.0, "grad_norm": 0.06040116026997566, "global_step": 27956, "epoch": 776, "lr": 0.001} +{"train_loss": 0.034580737352371216, "train_loss_bc": 0.034580737352371216, "train_loss_llm": 0.0, "grad_norm": 0.0335613451898098, "global_step": 27957, "epoch": 776, "lr": 0.001} +{"train_loss": 0.029079576954245567, "train_loss_bc": 0.029079576954245567, "train_loss_llm": 0.0, "grad_norm": 0.04074263945221901, "global_step": 27958, "epoch": 776, "lr": 0.001} +{"train_loss": 0.03526797145605087, "train_loss_bc": 0.03525146096944809, "train_loss_llm": 0.0016510486602783203, "grad_norm": 0.06340622156858444, "global_step": 27959, "epoch": 776, "lr": 0.001} +{"train_loss": 0.029879320412874222, "train_loss_bc": 0.02986486442387104, "train_loss_llm": 0.0014455318450927734, "grad_norm": 0.07584436237812042, "global_step": 27960, "epoch": 776, "lr": 0.001} +{"train_loss": 0.024113932624459267, "train_loss_bc": 0.024110088124871254, "train_loss_llm": 0.00038439035415649414, "grad_norm": 0.03051898628473282, "global_step": 27961, "epoch": 776, "lr": 0.001} +{"train_loss": 0.032979510724544525, "train_loss_bc": 0.032979510724544525, "train_loss_llm": 0.0, "grad_norm": 0.0486905463039875, "global_step": 27962, "epoch": 776, "lr": 0.001} +{"train_loss": 0.029560096561908722, "train_loss_bc": 0.029560096561908722, "train_loss_llm": 0.0, "grad_norm": 0.06958553940057755, "global_step": 27963, "epoch": 776, "lr": 0.001} +{"train_loss": 0.030626913532614708, "train_loss_bc": 0.030617378652095795, "train_loss_llm": 0.0009534657001495361, "grad_norm": 0.08832378685474396, "global_step": 27964, "epoch": 776, "lr": 0.001} +{"train_loss": 0.03759945556521416, "train_loss_bc": 0.03759945556521416, "train_loss_llm": 0.0, "grad_norm": 0.04642328992486, "global_step": 27965, "epoch": 776, "lr": 0.001} +{"train_loss": 0.031221864745020866, "train_loss_bc": 0.031221864745020866, "train_loss_llm": 0.0, "grad_norm": 0.05783231183886528, "global_step": 27966, "epoch": 776, "lr": 0.001} +{"train_loss": 0.03202241659164429, "train_loss_bc": 0.03202241659164429, "train_loss_llm": 0.0, "grad_norm": 0.07701601088047028, "global_step": 27967, "epoch": 776, "lr": 0.001} +{"train_loss": 0.029392223805189133, "train_loss_bc": 0.029392223805189133, "train_loss_llm": 0.0, "grad_norm": 0.09368278831243515, "global_step": 27968, "epoch": 776, "lr": 0.001} +{"train_loss": 0.026773016899824142, "train_loss_bc": 0.02677241712808609, "train_loss_llm": 5.9992074966430664e-05, "grad_norm": 0.03252902999520302, "global_step": 27969, "epoch": 776, "lr": 0.001} +{"train_loss": 0.032778121531009674, "train_loss_bc": 0.03274114429950714, "train_loss_llm": 0.003697693347930908, "grad_norm": 0.05057229474186897, "global_step": 27970, "epoch": 776, "lr": 0.001} +{"train_loss": 0.03036049406768547, "train_loss_bc": 0.029956307262182236, "train_loss_llm": 0.00013521313667297363, "grad_norm": 0.09164148569107056, "global_step": 27971, "epoch": 776, "lr": 0.001, "val_loss": 0.09456318616867065} +{"train_loss": 0.027080688625574112, "train_loss_bc": 0.02707449160516262, "train_loss_llm": 0.0006196498870849609, "grad_norm": 0.10174534469842911, "global_step": 27972, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03078463301062584, "train_loss_bc": 0.03077627904713154, "train_loss_llm": 0.0008353590965270996, "grad_norm": 0.033673565834760666, "global_step": 27973, "epoch": 777, "lr": 0.001} +{"train_loss": 0.0297213327139616, "train_loss_bc": 0.0297213327139616, "train_loss_llm": 0.0, "grad_norm": 0.04944883659482002, "global_step": 27974, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03407030552625656, "train_loss_bc": 0.03407030552625656, "train_loss_llm": 0.0, "grad_norm": 0.07264447212219238, "global_step": 27975, "epoch": 777, "lr": 0.001} +{"train_loss": 0.036786966025829315, "train_loss_bc": 0.036786966025829315, "train_loss_llm": 0.0, "grad_norm": 0.09124690294265747, "global_step": 27976, "epoch": 777, "lr": 0.001} +{"train_loss": 0.024853602051734924, "train_loss_bc": 0.024845317006111145, "train_loss_llm": 0.0008285045623779297, "grad_norm": 0.03286245837807655, "global_step": 27977, "epoch": 777, "lr": 0.001} +{"train_loss": 0.032082539051771164, "train_loss_bc": 0.032082539051771164, "train_loss_llm": 0.0, "grad_norm": 0.04700562730431557, "global_step": 27978, "epoch": 777, "lr": 0.001} +{"train_loss": 0.032652612775564194, "train_loss_bc": 0.032652612775564194, "train_loss_llm": 0.0, "grad_norm": 0.0534803532063961, "global_step": 27979, "epoch": 777, "lr": 0.001} +{"train_loss": 0.0332469567656517, "train_loss_bc": 0.03324167802929878, "train_loss_llm": 0.000527799129486084, "grad_norm": 0.0693572610616684, "global_step": 27980, "epoch": 777, "lr": 0.001} +{"train_loss": 0.038080018013715744, "train_loss_bc": 0.038080018013715744, "train_loss_llm": 0.0, "grad_norm": 0.037479061633348465, "global_step": 27981, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03152584657073021, "train_loss_bc": 0.03152584657073021, "train_loss_llm": 0.0, "grad_norm": 0.07259858399629593, "global_step": 27982, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03543074429035187, "train_loss_bc": 0.035423994064331055, "train_loss_llm": 0.0006749629974365234, "grad_norm": 0.09659886360168457, "global_step": 27983, "epoch": 777, "lr": 0.001} +{"train_loss": 0.028231626376509666, "train_loss_bc": 0.028231626376509666, "train_loss_llm": 0.0, "grad_norm": 0.11646319925785065, "global_step": 27984, "epoch": 777, "lr": 0.001} +{"train_loss": 0.0260810274630785, "train_loss_bc": 0.0260810274630785, "train_loss_llm": 0.0, "grad_norm": 0.029995523393154144, "global_step": 27985, "epoch": 777, "lr": 0.001} +{"train_loss": 0.02971271611750126, "train_loss_bc": 0.02971271611750126, "train_loss_llm": 0.0, "grad_norm": 0.06100962311029434, "global_step": 27986, "epoch": 777, "lr": 0.001} +{"train_loss": 0.026209883391857147, "train_loss_bc": 0.02620316483080387, "train_loss_llm": 0.0006718635559082031, "grad_norm": 0.08365954458713531, "global_step": 27987, "epoch": 777, "lr": 0.001} +{"train_loss": 0.031908392906188965, "train_loss_bc": 0.031908392906188965, "train_loss_llm": 0.0, "grad_norm": 0.11157304793596268, "global_step": 27988, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03345070034265518, "train_loss_bc": 0.033446021378040314, "train_loss_llm": 0.0004678964614868164, "grad_norm": 0.04382488131523132, "global_step": 27989, "epoch": 777, "lr": 0.001} +{"train_loss": 0.02783583663403988, "train_loss_bc": 0.02783583663403988, "train_loss_llm": 0.0, "grad_norm": 0.05592285096645355, "global_step": 27990, "epoch": 777, "lr": 0.001} +{"train_loss": 0.030134456232190132, "train_loss_bc": 0.030134456232190132, "train_loss_llm": 0.0, "grad_norm": 0.07965072989463806, "global_step": 27991, "epoch": 777, "lr": 0.001} +{"train_loss": 0.028903251513838768, "train_loss_bc": 0.028900068253278732, "train_loss_llm": 0.0003184080123901367, "grad_norm": 0.11034335196018219, "global_step": 27992, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03146689757704735, "train_loss_bc": 0.03146221861243248, "train_loss_llm": 0.0004678666591644287, "grad_norm": 0.03444670885801315, "global_step": 27993, "epoch": 777, "lr": 0.001} +{"train_loss": 0.030290670692920685, "train_loss_bc": 0.030289195477962494, "train_loss_llm": 0.0001475512981414795, "grad_norm": 0.04672440141439438, "global_step": 27994, "epoch": 777, "lr": 0.001} +{"train_loss": 0.02503514476120472, "train_loss_bc": 0.02503514476120472, "train_loss_llm": 0.0, "grad_norm": 0.055073369294404984, "global_step": 27995, "epoch": 777, "lr": 0.001} +{"train_loss": 0.030965175479650497, "train_loss_bc": 0.030961211770772934, "train_loss_llm": 0.00039634108543395996, "grad_norm": 0.0764736756682396, "global_step": 27996, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03231535479426384, "train_loss_bc": 0.03231535479426384, "train_loss_llm": 0.0, "grad_norm": 0.03999589383602142, "global_step": 27997, "epoch": 777, "lr": 0.001} +{"train_loss": 0.02366134710609913, "train_loss_bc": 0.02366134710609913, "train_loss_llm": 0.0, "grad_norm": 0.054237741976976395, "global_step": 27998, "epoch": 777, "lr": 0.001} +{"train_loss": 0.0273343026638031, "train_loss_bc": 0.0273343026638031, "train_loss_llm": 0.0, "grad_norm": 0.06567096710205078, "global_step": 27999, "epoch": 777, "lr": 0.001} +{"train_loss": 0.02803548239171505, "train_loss_bc": 0.02803548239171505, "train_loss_llm": 0.0, "grad_norm": 0.07813816517591476, "global_step": 28000, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03215426206588745, "train_loss_bc": 0.03215426206588745, "train_loss_llm": 0.0, "grad_norm": 0.029308026656508446, "global_step": 28001, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03080417588353157, "train_loss_bc": 0.03080417588353157, "train_loss_llm": 0.0, "grad_norm": 0.05248104780912399, "global_step": 28002, "epoch": 777, "lr": 0.001} +{"train_loss": 0.02987656369805336, "train_loss_bc": 0.02987656369805336, "train_loss_llm": 0.0, "grad_norm": 0.07341737300157547, "global_step": 28003, "epoch": 777, "lr": 0.001} +{"train_loss": 0.030936073511838913, "train_loss_bc": 0.030936073511838913, "train_loss_llm": 0.0, "grad_norm": 0.09949839860200882, "global_step": 28004, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03172962740063667, "train_loss_bc": 0.03172962740063667, "train_loss_llm": 0.0, "grad_norm": 0.04047507047653198, "global_step": 28005, "epoch": 777, "lr": 0.001} +{"train_loss": 0.031056927517056465, "train_loss_bc": 0.031038524582982063, "train_loss_llm": 0.0018402934074401855, "grad_norm": 0.05520186945796013, "global_step": 28006, "epoch": 777, "lr": 0.001} +{"train_loss": 0.03040264831441972, "train_loss_bc": 0.030017007142305374, "train_loss_llm": 0.003219008445739746, "grad_norm": 0.06899571418762207, "global_step": 28007, "epoch": 777, "lr": 0.001, "val_loss": 0.09505181014537811} +{"train_loss": 0.029254188761115074, "train_loss_bc": 0.029250063002109528, "train_loss_llm": 0.0004126429557800293, "grad_norm": 0.08551843464374542, "global_step": 28008, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03399232402443886, "train_loss_bc": 0.03399232402443886, "train_loss_llm": 0.0, "grad_norm": 0.030370566993951797, "global_step": 28009, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03144124895334244, "train_loss_bc": 0.03144124895334244, "train_loss_llm": 0.0, "grad_norm": 0.04342624172568321, "global_step": 28010, "epoch": 778, "lr": 0.001} +{"train_loss": 0.030994804576039314, "train_loss_bc": 0.030994804576039314, "train_loss_llm": 0.0, "grad_norm": 0.06908595561981201, "global_step": 28011, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03334599360823631, "train_loss_bc": 0.03334599360823631, "train_loss_llm": 0.0, "grad_norm": 0.08455023914575577, "global_step": 28012, "epoch": 778, "lr": 0.001} +{"train_loss": 0.02469019778072834, "train_loss_bc": 0.02469019778072834, "train_loss_llm": 0.0, "grad_norm": 0.024636724963784218, "global_step": 28013, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03277020528912544, "train_loss_bc": 0.03277020528912544, "train_loss_llm": 0.0, "grad_norm": 0.04317750036716461, "global_step": 28014, "epoch": 778, "lr": 0.001} +{"train_loss": 0.02752697467803955, "train_loss_bc": 0.02752697467803955, "train_loss_llm": 0.0, "grad_norm": 0.05553028732538223, "global_step": 28015, "epoch": 778, "lr": 0.001} +{"train_loss": 0.034634266048669815, "train_loss_bc": 0.034634266048669815, "train_loss_llm": 0.0, "grad_norm": 0.06991511583328247, "global_step": 28016, "epoch": 778, "lr": 0.001} +{"train_loss": 0.02975725568830967, "train_loss_bc": 0.02975725568830967, "train_loss_llm": 0.0, "grad_norm": 0.0300399512052536, "global_step": 28017, "epoch": 778, "lr": 0.001} +{"train_loss": 0.02969088964164257, "train_loss_bc": 0.02969088964164257, "train_loss_llm": 0.0, "grad_norm": 0.0506969690322876, "global_step": 28018, "epoch": 778, "lr": 0.001} +{"train_loss": 0.029573578387498856, "train_loss_bc": 0.029562577605247498, "train_loss_llm": 0.0011001527309417725, "grad_norm": 0.06526554375886917, "global_step": 28019, "epoch": 778, "lr": 0.001} +{"train_loss": 0.024540385231375694, "train_loss_bc": 0.024540385231375694, "train_loss_llm": 0.0, "grad_norm": 0.08172839134931564, "global_step": 28020, "epoch": 778, "lr": 0.001} +{"train_loss": 0.026065828278660774, "train_loss_bc": 0.026060130447149277, "train_loss_llm": 0.0005697906017303467, "grad_norm": 0.029636366292834282, "global_step": 28021, "epoch": 778, "lr": 0.001} +{"train_loss": 0.02587740123271942, "train_loss_bc": 0.02587740123271942, "train_loss_llm": 0.0, "grad_norm": 0.037621017545461655, "global_step": 28022, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03473009169101715, "train_loss_bc": 0.03473009169101715, "train_loss_llm": 0.0, "grad_norm": 0.054844994097948074, "global_step": 28023, "epoch": 778, "lr": 0.001} +{"train_loss": 0.025726288557052612, "train_loss_bc": 0.025726288557052612, "train_loss_llm": 0.0, "grad_norm": 0.05955427139997482, "global_step": 28024, "epoch": 778, "lr": 0.001} +{"train_loss": 0.029597558081150055, "train_loss_bc": 0.029597558081150055, "train_loss_llm": 0.0, "grad_norm": 0.04470660537481308, "global_step": 28025, "epoch": 778, "lr": 0.001} +{"train_loss": 0.035955242812633514, "train_loss_bc": 0.035955242812633514, "train_loss_llm": 0.0, "grad_norm": 0.07009190320968628, "global_step": 28026, "epoch": 778, "lr": 0.001} +{"train_loss": 0.0374305322766304, "train_loss_bc": 0.0374305322766304, "train_loss_llm": 0.0, "grad_norm": 0.09723597764968872, "global_step": 28027, "epoch": 778, "lr": 0.001} +{"train_loss": 0.028262628242373466, "train_loss_bc": 0.028262628242373466, "train_loss_llm": 0.0, "grad_norm": 0.11877517402172089, "global_step": 28028, "epoch": 778, "lr": 0.001} +{"train_loss": 0.02720041573047638, "train_loss_bc": 0.02720041573047638, "train_loss_llm": 0.0, "grad_norm": 0.036441054195165634, "global_step": 28029, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03059130720794201, "train_loss_bc": 0.03059130720794201, "train_loss_llm": 0.0, "grad_norm": 0.05383015051484108, "global_step": 28030, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03289174661040306, "train_loss_bc": 0.03289174661040306, "train_loss_llm": 0.0, "grad_norm": 0.0776878371834755, "global_step": 28031, "epoch": 778, "lr": 0.001} +{"train_loss": 0.030742548406124115, "train_loss_bc": 0.030742548406124115, "train_loss_llm": 0.0, "grad_norm": 0.09896349161863327, "global_step": 28032, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03296434506773949, "train_loss_bc": 0.03296434506773949, "train_loss_llm": 0.0, "grad_norm": 0.03211916610598564, "global_step": 28033, "epoch": 778, "lr": 0.001} +{"train_loss": 0.02669139765202999, "train_loss_bc": 0.02669139765202999, "train_loss_llm": 0.0, "grad_norm": 0.04450060427188873, "global_step": 28034, "epoch": 778, "lr": 0.001} +{"train_loss": 0.037131018936634064, "train_loss_bc": 0.03712517395615578, "train_loss_llm": 0.0005846619606018066, "grad_norm": 0.059657420963048935, "global_step": 28035, "epoch": 778, "lr": 0.001} +{"train_loss": 0.028550153598189354, "train_loss_bc": 0.028550153598189354, "train_loss_llm": 0.0, "grad_norm": 0.06831233203411102, "global_step": 28036, "epoch": 778, "lr": 0.001} +{"train_loss": 0.031747203320264816, "train_loss_bc": 0.031747203320264816, "train_loss_llm": 0.0, "grad_norm": 0.034098897129297256, "global_step": 28037, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03178144246339798, "train_loss_bc": 0.03178144246339798, "train_loss_llm": 0.0, "grad_norm": 0.051588937640190125, "global_step": 28038, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03081105463206768, "train_loss_bc": 0.030810289084911346, "train_loss_llm": 7.653236389160156e-05, "grad_norm": 0.06900708377361298, "global_step": 28039, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03130126744508743, "train_loss_bc": 0.03130126744508743, "train_loss_llm": 0.0, "grad_norm": 0.08671469241380692, "global_step": 28040, "epoch": 778, "lr": 0.001} +{"train_loss": 0.028178010135889053, "train_loss_bc": 0.028170404955744743, "train_loss_llm": 0.0007605552673339844, "grad_norm": 0.03206798806786537, "global_step": 28041, "epoch": 778, "lr": 0.001} +{"train_loss": 0.03049153834581375, "train_loss_bc": 0.03046775609254837, "train_loss_llm": 0.0023781955242156982, "grad_norm": 0.06400550901889801, "global_step": 28042, "epoch": 778, "lr": 0.001} +{"train_loss": 0.030305681698438194, "train_loss_bc": 0.024045037105679512, "train_loss_llm": 0.0028169751167297363, "grad_norm": 0.09528020769357681, "global_step": 28043, "epoch": 778, "lr": 0.001, "val_loss": 0.0938255786895752} +{"train_loss": 0.025934403762221336, "train_loss_bc": 0.025932379066944122, "train_loss_llm": 0.00020247697830200195, "grad_norm": 0.10262402147054672, "global_step": 28044, "epoch": 779, "lr": 0.001} +{"train_loss": 0.029780808836221695, "train_loss_bc": 0.029771123081445694, "train_loss_llm": 0.0009686648845672607, "grad_norm": 0.03503704071044922, "global_step": 28045, "epoch": 779, "lr": 0.001} +{"train_loss": 0.031128212809562683, "train_loss_bc": 0.031128212809562683, "train_loss_llm": 0.0, "grad_norm": 0.046152953058481216, "global_step": 28046, "epoch": 779, "lr": 0.001} +{"train_loss": 0.026103168725967407, "train_loss_bc": 0.026103168725967407, "train_loss_llm": 0.0, "grad_norm": 0.05681537464261055, "global_step": 28047, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03449102118611336, "train_loss_bc": 0.034489329904317856, "train_loss_llm": 0.00016927719116210938, "grad_norm": 0.08340880274772644, "global_step": 28048, "epoch": 779, "lr": 0.001} +{"train_loss": 0.028800249099731445, "train_loss_bc": 0.028800249099731445, "train_loss_llm": 0.0, "grad_norm": 0.029619751498103142, "global_step": 28049, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03201163560152054, "train_loss_bc": 0.03201163560152054, "train_loss_llm": 0.0, "grad_norm": 0.049325134605169296, "global_step": 28050, "epoch": 779, "lr": 0.001} +{"train_loss": 0.023522287607192993, "train_loss_bc": 0.023522287607192993, "train_loss_llm": 0.0, "grad_norm": 0.06823774427175522, "global_step": 28051, "epoch": 779, "lr": 0.001} +{"train_loss": 0.026524724438786507, "train_loss_bc": 0.026524724438786507, "train_loss_llm": 0.0, "grad_norm": 0.07764680683612823, "global_step": 28052, "epoch": 779, "lr": 0.001} +{"train_loss": 0.030975915491580963, "train_loss_bc": 0.030975915491580963, "train_loss_llm": 0.0, "grad_norm": 0.03432154655456543, "global_step": 28053, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03366727754473686, "train_loss_bc": 0.03366727754473686, "train_loss_llm": 0.0, "grad_norm": 0.04859631508588791, "global_step": 28054, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03210818022489548, "train_loss_bc": 0.03210818022489548, "train_loss_llm": 0.0, "grad_norm": 0.06474456936120987, "global_step": 28055, "epoch": 779, "lr": 0.001} +{"train_loss": 0.033736202865839005, "train_loss_bc": 0.033732663840055466, "train_loss_llm": 0.00035372376441955566, "grad_norm": 0.08530145138502121, "global_step": 28056, "epoch": 779, "lr": 0.001} +{"train_loss": 0.030684255063533783, "train_loss_bc": 0.030684255063533783, "train_loss_llm": 0.0, "grad_norm": 0.026154978200793266, "global_step": 28057, "epoch": 779, "lr": 0.001} +{"train_loss": 0.02782604470849037, "train_loss_bc": 0.02782604470849037, "train_loss_llm": 0.0, "grad_norm": 0.039800141006708145, "global_step": 28058, "epoch": 779, "lr": 0.001} +{"train_loss": 0.033531978726387024, "train_loss_bc": 0.03349609300494194, "train_loss_llm": 0.0035884976387023926, "grad_norm": 0.06452673673629761, "global_step": 28059, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03002999536693096, "train_loss_bc": 0.030028725042939186, "train_loss_llm": 0.00012701749801635742, "grad_norm": 0.0831521674990654, "global_step": 28060, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03147554025053978, "train_loss_bc": 0.03146868944168091, "train_loss_llm": 0.0006849467754364014, "grad_norm": 0.03528508171439171, "global_step": 28061, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03293043375015259, "train_loss_bc": 0.03293043375015259, "train_loss_llm": 0.0, "grad_norm": 0.05279872193932533, "global_step": 28062, "epoch": 779, "lr": 0.001} +{"train_loss": 0.026044826954603195, "train_loss_bc": 0.026044826954603195, "train_loss_llm": 0.0, "grad_norm": 0.06852647662162781, "global_step": 28063, "epoch": 779, "lr": 0.001} +{"train_loss": 0.031470246613025665, "train_loss_bc": 0.031470246613025665, "train_loss_llm": 0.0, "grad_norm": 0.08311450481414795, "global_step": 28064, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03174356743693352, "train_loss_bc": 0.03173593804240227, "train_loss_llm": 0.0007629096508026123, "grad_norm": 0.039133697748184204, "global_step": 28065, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03432009741663933, "train_loss_bc": 0.034315746277570724, "train_loss_llm": 0.00043529272079467773, "grad_norm": 0.05531655624508858, "global_step": 28066, "epoch": 779, "lr": 0.001} +{"train_loss": 0.028941264376044273, "train_loss_bc": 0.02893337793648243, "train_loss_llm": 0.0007887184619903564, "grad_norm": 0.07284563779830933, "global_step": 28067, "epoch": 779, "lr": 0.001} +{"train_loss": 0.026092205196619034, "train_loss_bc": 0.02608920820057392, "train_loss_llm": 0.00029969215393066406, "grad_norm": 0.08972485363483429, "global_step": 28068, "epoch": 779, "lr": 0.001} +{"train_loss": 0.0333041176199913, "train_loss_bc": 0.0333041176199913, "train_loss_llm": 0.0, "grad_norm": 0.042673107236623764, "global_step": 28069, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03901902213692665, "train_loss_bc": 0.03901902213692665, "train_loss_llm": 0.0, "grad_norm": 0.06890945136547089, "global_step": 28070, "epoch": 779, "lr": 0.001} +{"train_loss": 0.025686349719762802, "train_loss_bc": 0.025686349719762802, "train_loss_llm": 0.0, "grad_norm": 0.08713173121213913, "global_step": 28071, "epoch": 779, "lr": 0.001} +{"train_loss": 0.030604448169469833, "train_loss_bc": 0.030604448169469833, "train_loss_llm": 0.0, "grad_norm": 0.10707870870828629, "global_step": 28072, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03543763607740402, "train_loss_bc": 0.03543763607740402, "train_loss_llm": 0.0, "grad_norm": 0.05165063217282295, "global_step": 28073, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03680935874581337, "train_loss_bc": 0.03680935874581337, "train_loss_llm": 0.0, "grad_norm": 0.07888086885213852, "global_step": 28074, "epoch": 779, "lr": 0.001} +{"train_loss": 0.028345772996544838, "train_loss_bc": 0.028345772996544838, "train_loss_llm": 0.0, "grad_norm": 0.10514269769191742, "global_step": 28075, "epoch": 779, "lr": 0.001} +{"train_loss": 0.02795412205159664, "train_loss_bc": 0.02795412205159664, "train_loss_llm": 0.0, "grad_norm": 0.1481182873249054, "global_step": 28076, "epoch": 779, "lr": 0.001} +{"train_loss": 0.022796019911766052, "train_loss_bc": 0.022796019911766052, "train_loss_llm": 0.0, "grad_norm": 0.023884892463684082, "global_step": 28077, "epoch": 779, "lr": 0.001} +{"train_loss": 0.031622324138879776, "train_loss_bc": 0.031622324138879776, "train_loss_llm": 0.0, "grad_norm": 0.043185118585824966, "global_step": 28078, "epoch": 779, "lr": 0.001} +{"train_loss": 0.03046057823424538, "train_loss_bc": 0.031122710555791855, "train_loss_llm": 0.00043904781341552734, "grad_norm": 0.08392511308193207, "global_step": 28079, "epoch": 779, "lr": 0.001, "val_loss": 0.09526025503873825} +{"train_loss": 0.02595485933125019, "train_loss_bc": 0.02595485933125019, "train_loss_llm": 0.0, "grad_norm": 0.10318725556135178, "global_step": 28080, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03491661697626114, "train_loss_bc": 0.03491661697626114, "train_loss_llm": 0.0, "grad_norm": 0.035296715795993805, "global_step": 28081, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03476592153310776, "train_loss_bc": 0.03476592153310776, "train_loss_llm": 0.0, "grad_norm": 0.04345645755529404, "global_step": 28082, "epoch": 780, "lr": 0.001} +{"train_loss": 0.02396428771317005, "train_loss_bc": 0.02395651489496231, "train_loss_llm": 0.0007772147655487061, "grad_norm": 0.06498590111732483, "global_step": 28083, "epoch": 780, "lr": 0.001} +{"train_loss": 0.025571443140506744, "train_loss_bc": 0.025571443140506744, "train_loss_llm": 0.0, "grad_norm": 0.07504902780056, "global_step": 28084, "epoch": 780, "lr": 0.001} +{"train_loss": 0.031942807137966156, "train_loss_bc": 0.031942807137966156, "train_loss_llm": 0.0, "grad_norm": 0.026226093992590904, "global_step": 28085, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03319352865219116, "train_loss_bc": 0.03319352865219116, "train_loss_llm": 0.0, "grad_norm": 0.06536393612623215, "global_step": 28086, "epoch": 780, "lr": 0.001} +{"train_loss": 0.026097605004906654, "train_loss_bc": 0.026094811037182808, "train_loss_llm": 0.00027936697006225586, "grad_norm": 0.08380047231912613, "global_step": 28087, "epoch": 780, "lr": 0.001} +{"train_loss": 0.026463449001312256, "train_loss_bc": 0.026462649926543236, "train_loss_llm": 7.990002632141113e-05, "grad_norm": 0.11736372113227844, "global_step": 28088, "epoch": 780, "lr": 0.001} +{"train_loss": 0.029704120010137558, "train_loss_bc": 0.02969861589372158, "train_loss_llm": 0.0005504488945007324, "grad_norm": 0.041486918926239014, "global_step": 28089, "epoch": 780, "lr": 0.001} +{"train_loss": 0.029986295849084854, "train_loss_bc": 0.0299833957105875, "train_loss_llm": 0.0002899467945098877, "grad_norm": 0.07571502774953842, "global_step": 28090, "epoch": 780, "lr": 0.001} +{"train_loss": 0.031567469239234924, "train_loss_bc": 0.031560640782117844, "train_loss_llm": 0.0006829202175140381, "grad_norm": 0.10988525301218033, "global_step": 28091, "epoch": 780, "lr": 0.001} +{"train_loss": 0.02610434778034687, "train_loss_bc": 0.02610434778034687, "train_loss_llm": 0.0, "grad_norm": 0.1361960768699646, "global_step": 28092, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03050556406378746, "train_loss_bc": 0.03050556406378746, "train_loss_llm": 0.0, "grad_norm": 0.04912753403186798, "global_step": 28093, "epoch": 780, "lr": 0.001} +{"train_loss": 0.028970468789339066, "train_loss_bc": 0.028970468789339066, "train_loss_llm": 0.0, "grad_norm": 0.06712990254163742, "global_step": 28094, "epoch": 780, "lr": 0.001} +{"train_loss": 0.029949339106678963, "train_loss_bc": 0.029942374676465988, "train_loss_llm": 0.0006964802742004395, "grad_norm": 0.07633037120103836, "global_step": 28095, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03360095992684364, "train_loss_bc": 0.03360095992684364, "train_loss_llm": 0.0, "grad_norm": 0.10881571471691132, "global_step": 28096, "epoch": 780, "lr": 0.001} +{"train_loss": 0.0327732115983963, "train_loss_bc": 0.0327732115983963, "train_loss_llm": 0.0, "grad_norm": 0.0407630018889904, "global_step": 28097, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03049185499548912, "train_loss_bc": 0.03049185499548912, "train_loss_llm": 0.0, "grad_norm": 0.0570220910012722, "global_step": 28098, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03415884077548981, "train_loss_bc": 0.03415270894765854, "train_loss_llm": 0.0006130039691925049, "grad_norm": 0.07487019151449203, "global_step": 28099, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03337998688220978, "train_loss_bc": 0.03336795046925545, "train_loss_llm": 0.0012037456035614014, "grad_norm": 0.09230819344520569, "global_step": 28100, "epoch": 780, "lr": 0.001} +{"train_loss": 0.028518831357359886, "train_loss_bc": 0.028518831357359886, "train_loss_llm": 0.0, "grad_norm": 0.027033086866140366, "global_step": 28101, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03371260315179825, "train_loss_bc": 0.03371260315179825, "train_loss_llm": 0.0, "grad_norm": 0.05608600378036499, "global_step": 28102, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03251894935965538, "train_loss_bc": 0.03251894935965538, "train_loss_llm": 0.0, "grad_norm": 0.07765037566423416, "global_step": 28103, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03088599257171154, "train_loss_bc": 0.030884120613336563, "train_loss_llm": 0.00018727779388427734, "grad_norm": 0.08626049011945724, "global_step": 28104, "epoch": 780, "lr": 0.001} +{"train_loss": 0.030482430011034012, "train_loss_bc": 0.03047129325568676, "train_loss_llm": 0.0011137425899505615, "grad_norm": 0.03484718129038811, "global_step": 28105, "epoch": 780, "lr": 0.001} +{"train_loss": 0.028294015675783157, "train_loss_bc": 0.028277847915887833, "train_loss_llm": 0.0016167759895324707, "grad_norm": 0.05642762780189514, "global_step": 28106, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03270651400089264, "train_loss_bc": 0.032702889293432236, "train_loss_llm": 0.0003625154495239258, "grad_norm": 0.08166326582431793, "global_step": 28107, "epoch": 780, "lr": 0.001} +{"train_loss": 0.029831741005182266, "train_loss_bc": 0.029831741005182266, "train_loss_llm": 0.0, "grad_norm": 0.10496719926595688, "global_step": 28108, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03159315139055252, "train_loss_bc": 0.03159315139055252, "train_loss_llm": 0.0, "grad_norm": 0.03760354965925217, "global_step": 28109, "epoch": 780, "lr": 0.001} +{"train_loss": 0.031001105904579163, "train_loss_bc": 0.030980581417679787, "train_loss_llm": 0.002052396535873413, "grad_norm": 0.06794673204421997, "global_step": 28110, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03688407316803932, "train_loss_bc": 0.03688407316803932, "train_loss_llm": 0.0, "grad_norm": 0.10333534330129623, "global_step": 28111, "epoch": 780, "lr": 0.001} +{"train_loss": 0.033890098333358765, "train_loss_bc": 0.033890098333358765, "train_loss_llm": 0.0, "grad_norm": 0.13860094547271729, "global_step": 28112, "epoch": 780, "lr": 0.001} +{"train_loss": 0.027512220665812492, "train_loss_bc": 0.027512220665812492, "train_loss_llm": 0.0, "grad_norm": 0.03691878914833069, "global_step": 28113, "epoch": 780, "lr": 0.001} +{"train_loss": 0.02429899387061596, "train_loss_bc": 0.024285882711410522, "train_loss_llm": 0.0013110637664794922, "grad_norm": 0.05120834335684776, "global_step": 28114, "epoch": 780, "lr": 0.001} +{"train_loss": 0.03055729204788804, "train_loss_bc": 0.03386881574988365, "train_loss_llm": 0.0, "grad_norm": 0.09091595560312271, "global_step": 28115, "epoch": 780, "lr": 0.001, "train/cumulative_reward": 98.26304087178994, "train/mean_score": 17.827892145820037, "train/success_rate": 0.4, "test/cumulative_reward": 43.26313060894456, "test/mean_score": 7.429191303442523, "test/success_rate": 0.16, "val_loss": 0.09487836807966232, "train_action_mse_error": 0.03392994403839111} +{"train_loss": 0.03139542043209076, "train_loss_bc": 0.03139542043209076, "train_loss_llm": 0.0, "grad_norm": 0.11308638751506805, "global_step": 28116, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03446848690509796, "train_loss_bc": 0.03446848690509796, "train_loss_llm": 0.0, "grad_norm": 0.0389312282204628, "global_step": 28117, "epoch": 781, "lr": 0.001} +{"train_loss": 0.030538691207766533, "train_loss_bc": 0.030538691207766533, "train_loss_llm": 0.0, "grad_norm": 0.0544545017182827, "global_step": 28118, "epoch": 781, "lr": 0.001} +{"train_loss": 0.028814619407057762, "train_loss_bc": 0.028814619407057762, "train_loss_llm": 0.0, "grad_norm": 0.08543706685304642, "global_step": 28119, "epoch": 781, "lr": 0.001} +{"train_loss": 0.030335666611790657, "train_loss_bc": 0.030333397909998894, "train_loss_llm": 0.0002269446849822998, "grad_norm": 0.09558039903640747, "global_step": 28120, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03679927438497543, "train_loss_bc": 0.03679927438497543, "train_loss_llm": 0.0, "grad_norm": 0.038272958248853683, "global_step": 28121, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03026832826435566, "train_loss_bc": 0.03026832826435566, "train_loss_llm": 0.0, "grad_norm": 0.05909178406000137, "global_step": 28122, "epoch": 781, "lr": 0.001} +{"train_loss": 0.031330883502960205, "train_loss_bc": 0.031330883502960205, "train_loss_llm": 0.0, "grad_norm": 0.08070175349712372, "global_step": 28123, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03076675534248352, "train_loss_bc": 0.03076675534248352, "train_loss_llm": 0.0, "grad_norm": 0.09252786636352539, "global_step": 28124, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03214770555496216, "train_loss_bc": 0.03214770555496216, "train_loss_llm": 0.0, "grad_norm": 0.04393378645181656, "global_step": 28125, "epoch": 781, "lr": 0.001} +{"train_loss": 0.02898545376956463, "train_loss_bc": 0.02898545376956463, "train_loss_llm": 0.0, "grad_norm": 0.07103583216667175, "global_step": 28126, "epoch": 781, "lr": 0.001} +{"train_loss": 0.026659907773137093, "train_loss_bc": 0.026659907773137093, "train_loss_llm": 0.0, "grad_norm": 0.09839960187673569, "global_step": 28127, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03045414388179779, "train_loss_bc": 0.03045414388179779, "train_loss_llm": 0.0, "grad_norm": 0.12198708206415176, "global_step": 28128, "epoch": 781, "lr": 0.001} +{"train_loss": 0.029927261173725128, "train_loss_bc": 0.029927261173725128, "train_loss_llm": 0.0, "grad_norm": 0.03667966648936272, "global_step": 28129, "epoch": 781, "lr": 0.001} +{"train_loss": 0.025031793862581253, "train_loss_bc": 0.025031793862581253, "train_loss_llm": 0.0, "grad_norm": 0.0442696213722229, "global_step": 28130, "epoch": 781, "lr": 0.001} +{"train_loss": 0.026057245209813118, "train_loss_bc": 0.026057245209813118, "train_loss_llm": 0.0, "grad_norm": 0.06544467061758041, "global_step": 28131, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03184761106967926, "train_loss_bc": 0.03184761106967926, "train_loss_llm": 0.0, "grad_norm": 0.09014491736888885, "global_step": 28132, "epoch": 781, "lr": 0.001} +{"train_loss": 0.026665138080716133, "train_loss_bc": 0.026665138080716133, "train_loss_llm": 0.0, "grad_norm": 0.03136346861720085, "global_step": 28133, "epoch": 781, "lr": 0.001} +{"train_loss": 0.031665973365306854, "train_loss_bc": 0.03165257349610329, "train_loss_llm": 0.0013400614261627197, "grad_norm": 0.051693808287382126, "global_step": 28134, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03672011196613312, "train_loss_bc": 0.03672011196613312, "train_loss_llm": 0.0, "grad_norm": 0.08601228892803192, "global_step": 28135, "epoch": 781, "lr": 0.001} +{"train_loss": 0.028985140845179558, "train_loss_bc": 0.028985140845179558, "train_loss_llm": 0.0, "grad_norm": 0.10705018043518066, "global_step": 28136, "epoch": 781, "lr": 0.001} +{"train_loss": 0.029546305537223816, "train_loss_bc": 0.029546305537223816, "train_loss_llm": 0.0, "grad_norm": 0.035357605665922165, "global_step": 28137, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03375449404120445, "train_loss_bc": 0.03375449404120445, "train_loss_llm": 0.0, "grad_norm": 0.059619802981615067, "global_step": 28138, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03432397544384003, "train_loss_bc": 0.03432397544384003, "train_loss_llm": 0.0, "grad_norm": 0.07370918244123459, "global_step": 28139, "epoch": 781, "lr": 0.001} +{"train_loss": 0.030475573614239693, "train_loss_bc": 0.030465757474303246, "train_loss_llm": 0.0009815394878387451, "grad_norm": 0.08208611607551575, "global_step": 28140, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03112003766000271, "train_loss_bc": 0.031077027320861816, "train_loss_llm": 0.004301100969314575, "grad_norm": 0.03601346164941788, "global_step": 28141, "epoch": 781, "lr": 0.001} +{"train_loss": 0.029403842985630035, "train_loss_bc": 0.029403842985630035, "train_loss_llm": 0.0, "grad_norm": 0.059222228825092316, "global_step": 28142, "epoch": 781, "lr": 0.001} +{"train_loss": 0.027713842689990997, "train_loss_bc": 0.02770945616066456, "train_loss_llm": 0.0004385709762573242, "grad_norm": 0.07327177375555038, "global_step": 28143, "epoch": 781, "lr": 0.001} +{"train_loss": 0.02716321498155594, "train_loss_bc": 0.027162611484527588, "train_loss_llm": 6.0439109802246094e-05, "grad_norm": 0.08782392740249634, "global_step": 28144, "epoch": 781, "lr": 0.001} +{"train_loss": 0.026610594242811203, "train_loss_bc": 0.026587773114442825, "train_loss_llm": 0.0022822022438049316, "grad_norm": 0.047484345734119415, "global_step": 28145, "epoch": 781, "lr": 0.001} +{"train_loss": 0.030372144654393196, "train_loss_bc": 0.030372144654393196, "train_loss_llm": 0.0, "grad_norm": 0.08067955821752548, "global_step": 28146, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03424949198961258, "train_loss_bc": 0.03424949198961258, "train_loss_llm": 0.0, "grad_norm": 0.08993667364120483, "global_step": 28147, "epoch": 781, "lr": 0.001} +{"train_loss": 0.027073225006461143, "train_loss_bc": 0.027073225006461143, "train_loss_llm": 0.0, "grad_norm": 0.10568737983703613, "global_step": 28148, "epoch": 781, "lr": 0.001} +{"train_loss": 0.02906320057809353, "train_loss_bc": 0.02906227670609951, "train_loss_llm": 9.238719940185547e-05, "grad_norm": 0.034903936088085175, "global_step": 28149, "epoch": 781, "lr": 0.001} +{"train_loss": 0.03252015635371208, "train_loss_bc": 0.032512322068214417, "train_loss_llm": 0.0007834136486053467, "grad_norm": 0.050504572689533234, "global_step": 28150, "epoch": 781, "lr": 0.001} +{"train_loss": 0.030418940405878756, "train_loss_bc": 0.031826142221689224, "train_loss_llm": 0.0, "grad_norm": 0.06427069008350372, "global_step": 28151, "epoch": 781, "lr": 0.001, "val_loss": 0.09453164786100388} +{"train_loss": 0.0317041352391243, "train_loss_bc": 0.031688936054706573, "train_loss_llm": 0.0015200972557067871, "grad_norm": 0.06983022391796112, "global_step": 28152, "epoch": 782, "lr": 0.001} +{"train_loss": 0.029165957123041153, "train_loss_bc": 0.029165957123041153, "train_loss_llm": 0.0, "grad_norm": 0.03477708622813225, "global_step": 28153, "epoch": 782, "lr": 0.001} +{"train_loss": 0.02540234848856926, "train_loss_bc": 0.02540234848856926, "train_loss_llm": 0.0, "grad_norm": 0.05044377222657204, "global_step": 28154, "epoch": 782, "lr": 0.001} +{"train_loss": 0.026000121608376503, "train_loss_bc": 0.026000121608376503, "train_loss_llm": 0.0, "grad_norm": 0.06266595423221588, "global_step": 28155, "epoch": 782, "lr": 0.001} +{"train_loss": 0.03026125766336918, "train_loss_bc": 0.030251070857048035, "train_loss_llm": 0.0010187327861785889, "grad_norm": 0.07748492062091827, "global_step": 28156, "epoch": 782, "lr": 0.001} +{"train_loss": 0.03215836361050606, "train_loss_bc": 0.03215836361050606, "train_loss_llm": 0.0, "grad_norm": 0.03381616622209549, "global_step": 28157, "epoch": 782, "lr": 0.001} +{"train_loss": 0.026011578738689423, "train_loss_bc": 0.026007546111941338, "train_loss_llm": 0.0004032254219055176, "grad_norm": 0.05140763893723488, "global_step": 28158, "epoch": 782, "lr": 0.001} +{"train_loss": 0.035815123468637466, "train_loss_bc": 0.035815123468637466, "train_loss_llm": 0.0, "grad_norm": 0.06194809079170227, "global_step": 28159, "epoch": 782, "lr": 0.001} +{"train_loss": 0.030265692621469498, "train_loss_bc": 0.030265692621469498, "train_loss_llm": 0.0, "grad_norm": 0.07419829070568085, "global_step": 28160, "epoch": 782, "lr": 0.001} +{"train_loss": 0.036564283072948456, "train_loss_bc": 0.03655439242720604, "train_loss_llm": 0.0009892284870147705, "grad_norm": 0.034117598086595535, "global_step": 28161, "epoch": 782, "lr": 0.001} +{"train_loss": 0.029874028638005257, "train_loss_bc": 0.029874028638005257, "train_loss_llm": 0.0, "grad_norm": 0.04806291311979294, "global_step": 28162, "epoch": 782, "lr": 0.001} +{"train_loss": 0.02544650249183178, "train_loss_bc": 0.025430068373680115, "train_loss_llm": 0.001643359661102295, "grad_norm": 0.058247681707143784, "global_step": 28163, "epoch": 782, "lr": 0.001} +{"train_loss": 0.029766377061605453, "train_loss_bc": 0.029754456132650375, "train_loss_llm": 0.0011921525001525879, "grad_norm": 0.0746968686580658, "global_step": 28164, "epoch": 782, "lr": 0.001} +{"train_loss": 0.028157902881503105, "train_loss_bc": 0.028151648119091988, "train_loss_llm": 0.0006254017353057861, "grad_norm": 0.03323255106806755, "global_step": 28165, "epoch": 782, "lr": 0.001} +{"train_loss": 0.02403280884027481, "train_loss_bc": 0.02403280884027481, "train_loss_llm": 0.0, "grad_norm": 0.04446686431765556, "global_step": 28166, "epoch": 782, "lr": 0.001} +{"train_loss": 0.02936832420527935, "train_loss_bc": 0.02936832420527935, "train_loss_llm": 0.0, "grad_norm": 0.05955148860812187, "global_step": 28167, "epoch": 782, "lr": 0.001} +{"train_loss": 0.03053411841392517, "train_loss_bc": 0.03053411841392517, "train_loss_llm": 0.0, "grad_norm": 0.08031157404184341, "global_step": 28168, "epoch": 782, "lr": 0.001} +{"train_loss": 0.03690119460225105, "train_loss_bc": 0.03690119460225105, "train_loss_llm": 0.0, "grad_norm": 0.04677081108093262, "global_step": 28169, "epoch": 782, "lr": 0.001} +{"train_loss": 0.03321484848856926, "train_loss_bc": 0.03320600464940071, "train_loss_llm": 0.0008843541145324707, "grad_norm": 0.06738711893558502, "global_step": 28170, "epoch": 782, "lr": 0.001} +{"train_loss": 0.028900576755404472, "train_loss_bc": 0.0288994163274765, "train_loss_llm": 0.00011605024337768555, "grad_norm": 0.08929675817489624, "global_step": 28171, "epoch": 782, "lr": 0.001} +{"train_loss": 0.02836921438574791, "train_loss_bc": 0.02836921438574791, "train_loss_llm": 0.0, "grad_norm": 0.11680372059345245, "global_step": 28172, "epoch": 782, "lr": 0.001} +{"train_loss": 0.02851104363799095, "train_loss_bc": 0.02851104363799095, "train_loss_llm": 0.0, "grad_norm": 0.028488360345363617, "global_step": 28173, "epoch": 782, "lr": 0.001} +{"train_loss": 0.03217863664031029, "train_loss_bc": 0.03217863664031029, "train_loss_llm": 0.0, "grad_norm": 0.03741126507520676, "global_step": 28174, "epoch": 782, "lr": 0.001} +{"train_loss": 0.025976214557886124, "train_loss_bc": 0.025976214557886124, "train_loss_llm": 0.0, "grad_norm": 0.051652103662490845, "global_step": 28175, "epoch": 782, "lr": 0.001} +{"train_loss": 0.030990319326519966, "train_loss_bc": 0.030990319326519966, "train_loss_llm": 0.0, "grad_norm": 0.07164809852838516, "global_step": 28176, "epoch": 782, "lr": 0.001} +{"train_loss": 0.030490979552268982, "train_loss_bc": 0.03047787956893444, "train_loss_llm": 0.0013099312782287598, "grad_norm": 0.032210931181907654, "global_step": 28177, "epoch": 782, "lr": 0.001} +{"train_loss": 0.030279522761702538, "train_loss_bc": 0.030279522761702538, "train_loss_llm": 0.0, "grad_norm": 0.05067598819732666, "global_step": 28178, "epoch": 782, "lr": 0.001} +{"train_loss": 0.03452203422784805, "train_loss_bc": 0.03452203422784805, "train_loss_llm": 0.0, "grad_norm": 0.06636647135019302, "global_step": 28179, "epoch": 782, "lr": 0.001} +{"train_loss": 0.03260963782668114, "train_loss_bc": 0.03260963782668114, "train_loss_llm": 0.0, "grad_norm": 0.09199082106351852, "global_step": 28180, "epoch": 782, "lr": 0.001} +{"train_loss": 0.02910042367875576, "train_loss_bc": 0.02910042367875576, "train_loss_llm": 0.0, "grad_norm": 0.03350856527686119, "global_step": 28181, "epoch": 782, "lr": 0.001} +{"train_loss": 0.028760937973856926, "train_loss_bc": 0.028760937973856926, "train_loss_llm": 0.0, "grad_norm": 0.05450917407870293, "global_step": 28182, "epoch": 782, "lr": 0.001} +{"train_loss": 0.035084325820207596, "train_loss_bc": 0.035084325820207596, "train_loss_llm": 0.0, "grad_norm": 0.057663559913635254, "global_step": 28183, "epoch": 782, "lr": 0.001} +{"train_loss": 0.030364669859409332, "train_loss_bc": 0.03034742921590805, "train_loss_llm": 0.0017239749431610107, "grad_norm": 0.07634413987398148, "global_step": 28184, "epoch": 782, "lr": 0.001} +{"train_loss": 0.03648092970252037, "train_loss_bc": 0.03648092970252037, "train_loss_llm": 0.0, "grad_norm": 0.03597303852438927, "global_step": 28185, "epoch": 782, "lr": 0.001} +{"train_loss": 0.02541406825184822, "train_loss_bc": 0.02541406825184822, "train_loss_llm": 0.0, "grad_norm": 0.04593238979578018, "global_step": 28186, "epoch": 782, "lr": 0.001} +{"train_loss": 0.030090863454259105, "train_loss_bc": 0.024592582136392593, "train_loss_llm": 0.0, "grad_norm": 0.060013748705387115, "global_step": 28187, "epoch": 782, "lr": 0.001, "val_loss": 0.09558980166912079} +{"train_loss": 0.027878297492861748, "train_loss_bc": 0.02787373587489128, "train_loss_llm": 0.00045621395111083984, "grad_norm": 0.07710424810647964, "global_step": 28188, "epoch": 783, "lr": 0.001} +{"train_loss": 0.026251981034874916, "train_loss_bc": 0.026251981034874916, "train_loss_llm": 0.0, "grad_norm": 0.026988552883267403, "global_step": 28189, "epoch": 783, "lr": 0.001} +{"train_loss": 0.03352949768304825, "train_loss_bc": 0.03352949768304825, "train_loss_llm": 0.0, "grad_norm": 0.05123553425073624, "global_step": 28190, "epoch": 783, "lr": 0.001} +{"train_loss": 0.028868844732642174, "train_loss_bc": 0.028859766200184822, "train_loss_llm": 0.00090789794921875, "grad_norm": 0.06306131184101105, "global_step": 28191, "epoch": 783, "lr": 0.001} +{"train_loss": 0.030824089422822, "train_loss_bc": 0.030824089422822, "train_loss_llm": 0.0, "grad_norm": 0.06622060388326645, "global_step": 28192, "epoch": 783, "lr": 0.001} +{"train_loss": 0.028339743614196777, "train_loss_bc": 0.02829691395163536, "train_loss_llm": 0.004282921552658081, "grad_norm": 0.026728402823209763, "global_step": 28193, "epoch": 783, "lr": 0.001} +{"train_loss": 0.03197537735104561, "train_loss_bc": 0.03197537735104561, "train_loss_llm": 0.0, "grad_norm": 0.03791502118110657, "global_step": 28194, "epoch": 783, "lr": 0.001} +{"train_loss": 0.028206396847963333, "train_loss_bc": 0.02819867990911007, "train_loss_llm": 0.0007716715335845947, "grad_norm": 0.047374334186315536, "global_step": 28195, "epoch": 783, "lr": 0.001} +{"train_loss": 0.026903320103883743, "train_loss_bc": 0.026903320103883743, "train_loss_llm": 0.0, "grad_norm": 0.060257457196712494, "global_step": 28196, "epoch": 783, "lr": 0.001} +{"train_loss": 0.03302731737494469, "train_loss_bc": 0.03302731737494469, "train_loss_llm": 0.0, "grad_norm": 0.046190861612558365, "global_step": 28197, "epoch": 783, "lr": 0.001} +{"train_loss": 0.03633270040154457, "train_loss_bc": 0.03632604703307152, "train_loss_llm": 0.0006651580333709717, "grad_norm": 0.07386341691017151, "global_step": 28198, "epoch": 783, "lr": 0.001} +{"train_loss": 0.03229110687971115, "train_loss_bc": 0.03227644041180611, "train_loss_llm": 0.0014667809009552002, "grad_norm": 0.09849414229393005, "global_step": 28199, "epoch": 783, "lr": 0.001} +{"train_loss": 0.027691926807165146, "train_loss_bc": 0.027691926807165146, "train_loss_llm": 0.0, "grad_norm": 0.10768193006515503, "global_step": 28200, "epoch": 783, "lr": 0.001} +{"train_loss": 0.030616529285907745, "train_loss_bc": 0.030616529285907745, "train_loss_llm": 0.0, "grad_norm": 0.03723134845495224, "global_step": 28201, "epoch": 783, "lr": 0.001} +{"train_loss": 0.038219962269067764, "train_loss_bc": 0.038219962269067764, "train_loss_llm": 0.0, "grad_norm": 0.0742594301700592, "global_step": 28202, "epoch": 783, "lr": 0.001} +{"train_loss": 0.031025396659970284, "train_loss_bc": 0.031025396659970284, "train_loss_llm": 0.0, "grad_norm": 0.0976349338889122, "global_step": 28203, "epoch": 783, "lr": 0.001} +{"train_loss": 0.03461511805653572, "train_loss_bc": 0.03461511805653572, "train_loss_llm": 0.0, "grad_norm": 0.11837687343358994, "global_step": 28204, "epoch": 783, "lr": 0.001} +{"train_loss": 0.029481887817382812, "train_loss_bc": 0.029481887817382812, "train_loss_llm": 0.0, "grad_norm": 0.032423537224531174, "global_step": 28205, "epoch": 783, "lr": 0.001} +{"train_loss": 0.02695460617542267, "train_loss_bc": 0.026951294392347336, "train_loss_llm": 0.0003311038017272949, "grad_norm": 0.04697840288281441, "global_step": 28206, "epoch": 783, "lr": 0.001} +{"train_loss": 0.028283780440688133, "train_loss_bc": 0.028283780440688133, "train_loss_llm": 0.0, "grad_norm": 0.06383325159549713, "global_step": 28207, "epoch": 783, "lr": 0.001} +{"train_loss": 0.03460746258497238, "train_loss_bc": 0.03460746258497238, "train_loss_llm": 0.0, "grad_norm": 0.08597607165575027, "global_step": 28208, "epoch": 783, "lr": 0.001} +{"train_loss": 0.028103971853852272, "train_loss_bc": 0.028103971853852272, "train_loss_llm": 0.0, "grad_norm": 0.030552364885807037, "global_step": 28209, "epoch": 783, "lr": 0.001} +{"train_loss": 0.027313925325870514, "train_loss_bc": 0.027313925325870514, "train_loss_llm": 0.0, "grad_norm": 0.04857749864459038, "global_step": 28210, "epoch": 783, "lr": 0.001} +{"train_loss": 0.029841110110282898, "train_loss_bc": 0.029841110110282898, "train_loss_llm": 0.0, "grad_norm": 0.06072471663355827, "global_step": 28211, "epoch": 783, "lr": 0.001} +{"train_loss": 0.02792435698211193, "train_loss_bc": 0.02792435698211193, "train_loss_llm": 0.0, "grad_norm": 0.0749252513051033, "global_step": 28212, "epoch": 783, "lr": 0.001} +{"train_loss": 0.026542287319898605, "train_loss_bc": 0.026542287319898605, "train_loss_llm": 0.0, "grad_norm": 0.031747352331876755, "global_step": 28213, "epoch": 783, "lr": 0.001} +{"train_loss": 0.02989005669951439, "train_loss_bc": 0.02989005669951439, "train_loss_llm": 0.0, "grad_norm": 0.05866729095578194, "global_step": 28214, "epoch": 783, "lr": 0.001} +{"train_loss": 0.03270965814590454, "train_loss_bc": 0.03270965814590454, "train_loss_llm": 0.0, "grad_norm": 0.07915875315666199, "global_step": 28215, "epoch": 783, "lr": 0.001} +{"train_loss": 0.035652805119752884, "train_loss_bc": 0.035652805119752884, "train_loss_llm": 0.0, "grad_norm": 0.0960220918059349, "global_step": 28216, "epoch": 783, "lr": 0.001} +{"train_loss": 0.029497992247343063, "train_loss_bc": 0.029497992247343063, "train_loss_llm": 0.0, "grad_norm": 0.029309602454304695, "global_step": 28217, "epoch": 783, "lr": 0.001} +{"train_loss": 0.029828110709786415, "train_loss_bc": 0.029821516945958138, "train_loss_llm": 0.0006594061851501465, "grad_norm": 0.040228310972452164, "global_step": 28218, "epoch": 783, "lr": 0.001} +{"train_loss": 0.029367150738835335, "train_loss_bc": 0.029367150738835335, "train_loss_llm": 0.0, "grad_norm": 0.060492947697639465, "global_step": 28219, "epoch": 783, "lr": 0.001} +{"train_loss": 0.02578580752015114, "train_loss_bc": 0.02578321471810341, "train_loss_llm": 0.0002592504024505615, "grad_norm": 0.07390046119689941, "global_step": 28220, "epoch": 783, "lr": 0.001} +{"train_loss": 0.030842499807476997, "train_loss_bc": 0.030842499807476997, "train_loss_llm": 0.0, "grad_norm": 0.03399594873189926, "global_step": 28221, "epoch": 783, "lr": 0.001} +{"train_loss": 0.031516265124082565, "train_loss_bc": 0.031516265124082565, "train_loss_llm": 0.0, "grad_norm": 0.06648769974708557, "global_step": 28222, "epoch": 783, "lr": 0.001} +{"train_loss": 0.030348332029663853, "train_loss_bc": 0.03179861232638359, "train_loss_llm": 0.0, "grad_norm": 0.08119552582502365, "global_step": 28223, "epoch": 783, "lr": 0.001, "val_loss": 0.09633032232522964} +{"train_loss": 0.024474531412124634, "train_loss_bc": 0.024474531412124634, "train_loss_llm": 0.0, "grad_norm": 0.09685053676366806, "global_step": 28224, "epoch": 784, "lr": 0.001} +{"train_loss": 0.033916447311639786, "train_loss_bc": 0.033916447311639786, "train_loss_llm": 0.0, "grad_norm": 0.03629723936319351, "global_step": 28225, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03224841132760048, "train_loss_bc": 0.03224522992968559, "train_loss_llm": 0.00031808018684387207, "grad_norm": 0.06201634556055069, "global_step": 28226, "epoch": 784, "lr": 0.001} +{"train_loss": 0.02857849933207035, "train_loss_bc": 0.02857849933207035, "train_loss_llm": 0.0, "grad_norm": 0.0743207037448883, "global_step": 28227, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03073672577738762, "train_loss_bc": 0.03073004260659218, "train_loss_llm": 0.0006682872772216797, "grad_norm": 0.087348073720932, "global_step": 28228, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03331676870584488, "train_loss_bc": 0.03331676870584488, "train_loss_llm": 0.0, "grad_norm": 0.03887199983000755, "global_step": 28229, "epoch": 784, "lr": 0.001} +{"train_loss": 0.028708264231681824, "train_loss_bc": 0.028708264231681824, "train_loss_llm": 0.0, "grad_norm": 0.0663759857416153, "global_step": 28230, "epoch": 784, "lr": 0.001} +{"train_loss": 0.032718725502491, "train_loss_bc": 0.032718725502491, "train_loss_llm": 0.0, "grad_norm": 0.08092576265335083, "global_step": 28231, "epoch": 784, "lr": 0.001} +{"train_loss": 0.030095016583800316, "train_loss_bc": 0.030095016583800316, "train_loss_llm": 0.0, "grad_norm": 0.10097484290599823, "global_step": 28232, "epoch": 784, "lr": 0.001} +{"train_loss": 0.02335485816001892, "train_loss_bc": 0.023344596847891808, "train_loss_llm": 0.0010260939598083496, "grad_norm": 0.031969018280506134, "global_step": 28233, "epoch": 784, "lr": 0.001} +{"train_loss": 0.028400491923093796, "train_loss_bc": 0.028400491923093796, "train_loss_llm": 0.0, "grad_norm": 0.0527062863111496, "global_step": 28234, "epoch": 784, "lr": 0.001} +{"train_loss": 0.028727296739816666, "train_loss_bc": 0.028727296739816666, "train_loss_llm": 0.0, "grad_norm": 0.0712052434682846, "global_step": 28235, "epoch": 784, "lr": 0.001} +{"train_loss": 0.032232243567705154, "train_loss_bc": 0.032232243567705154, "train_loss_llm": 0.0, "grad_norm": 0.09356816858053207, "global_step": 28236, "epoch": 784, "lr": 0.001} +{"train_loss": 0.028541678562760353, "train_loss_bc": 0.028541678562760353, "train_loss_llm": 0.0, "grad_norm": 0.027937598526477814, "global_step": 28237, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03114788793027401, "train_loss_bc": 0.031136874109506607, "train_loss_llm": 0.0011013150215148926, "grad_norm": 0.04723266139626503, "global_step": 28238, "epoch": 784, "lr": 0.001} +{"train_loss": 0.030772479251027107, "train_loss_bc": 0.030772479251027107, "train_loss_llm": 0.0, "grad_norm": 0.06607041507959366, "global_step": 28239, "epoch": 784, "lr": 0.001} +{"train_loss": 0.030544839799404144, "train_loss_bc": 0.030544839799404144, "train_loss_llm": 0.0, "grad_norm": 0.08471894264221191, "global_step": 28240, "epoch": 784, "lr": 0.001} +{"train_loss": 0.031169334426522255, "train_loss_bc": 0.031151214614510536, "train_loss_llm": 0.001811891794204712, "grad_norm": 0.028851818293333054, "global_step": 28241, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03348683565855026, "train_loss_bc": 0.03348683565855026, "train_loss_llm": 0.0, "grad_norm": 0.042848315089941025, "global_step": 28242, "epoch": 784, "lr": 0.001} +{"train_loss": 0.028262902051210403, "train_loss_bc": 0.028262902051210403, "train_loss_llm": 0.0, "grad_norm": 0.06612075120210648, "global_step": 28243, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03190883621573448, "train_loss_bc": 0.03190883621573448, "train_loss_llm": 0.0, "grad_norm": 0.0798996165394783, "global_step": 28244, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03075174242258072, "train_loss_bc": 0.03075174242258072, "train_loss_llm": 0.0, "grad_norm": 0.03910934552550316, "global_step": 28245, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03462347388267517, "train_loss_bc": 0.03462347388267517, "train_loss_llm": 0.0, "grad_norm": 0.06144971027970314, "global_step": 28246, "epoch": 784, "lr": 0.001} +{"train_loss": 0.028339780867099762, "train_loss_bc": 0.028339780867099762, "train_loss_llm": 0.0, "grad_norm": 0.07003402709960938, "global_step": 28247, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03214128687977791, "train_loss_bc": 0.03214128687977791, "train_loss_llm": 0.0, "grad_norm": 0.07951020449399948, "global_step": 28248, "epoch": 784, "lr": 0.001} +{"train_loss": 0.026013582944869995, "train_loss_bc": 0.026013582944869995, "train_loss_llm": 0.0, "grad_norm": 0.029938243329524994, "global_step": 28249, "epoch": 784, "lr": 0.001} +{"train_loss": 0.034521955996751785, "train_loss_bc": 0.03451460227370262, "train_loss_llm": 0.0007354021072387695, "grad_norm": 0.053787171840667725, "global_step": 28250, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03355669975280762, "train_loss_bc": 0.03355669975280762, "train_loss_llm": 0.0, "grad_norm": 0.07817909121513367, "global_step": 28251, "epoch": 784, "lr": 0.001} +{"train_loss": 0.0318976566195488, "train_loss_bc": 0.0318976566195488, "train_loss_llm": 0.0, "grad_norm": 0.10240441560745239, "global_step": 28252, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03664614260196686, "train_loss_bc": 0.03664614260196686, "train_loss_llm": 0.0, "grad_norm": 0.03782319277524948, "global_step": 28253, "epoch": 784, "lr": 0.001} +{"train_loss": 0.025753382593393326, "train_loss_bc": 0.025742873549461365, "train_loss_llm": 0.0010509192943572998, "grad_norm": 0.04959212616086006, "global_step": 28254, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03017589822411537, "train_loss_bc": 0.03017589822411537, "train_loss_llm": 0.0, "grad_norm": 0.06968573480844498, "global_step": 28255, "epoch": 784, "lr": 0.001} +{"train_loss": 0.025108834728598595, "train_loss_bc": 0.025108834728598595, "train_loss_llm": 0.0, "grad_norm": 0.08673804253339767, "global_step": 28256, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03225616365671158, "train_loss_bc": 0.03225616365671158, "train_loss_llm": 0.0, "grad_norm": 0.03513045608997345, "global_step": 28257, "epoch": 784, "lr": 0.001} +{"train_loss": 0.03160581737756729, "train_loss_bc": 0.03160581737756729, "train_loss_llm": 0.0, "grad_norm": 0.06135258451104164, "global_step": 28258, "epoch": 784, "lr": 0.001} +{"train_loss": 0.0303822025242779, "train_loss_bc": 0.026990849524736404, "train_loss_llm": 0.0032947659492492676, "grad_norm": 0.08229821175336838, "global_step": 28259, "epoch": 784, "lr": 0.001, "val_loss": 0.09497088193893433} +{"train_loss": 0.02758021280169487, "train_loss_bc": 0.02758021280169487, "train_loss_llm": 0.0, "grad_norm": 0.09843742102384567, "global_step": 28260, "epoch": 785, "lr": 0.001} +{"train_loss": 0.031631287187337875, "train_loss_bc": 0.031631287187337875, "train_loss_llm": 0.0, "grad_norm": 0.03148658946156502, "global_step": 28261, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03224596008658409, "train_loss_bc": 0.03224029019474983, "train_loss_llm": 0.0005670785903930664, "grad_norm": 0.04477576166391373, "global_step": 28262, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03185327723622322, "train_loss_bc": 0.03185327723622322, "train_loss_llm": 0.0, "grad_norm": 0.056923337280750275, "global_step": 28263, "epoch": 785, "lr": 0.001} +{"train_loss": 0.027517275884747505, "train_loss_bc": 0.027517275884747505, "train_loss_llm": 0.0, "grad_norm": 0.06586573272943497, "global_step": 28264, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03534341976046562, "train_loss_bc": 0.03534341976046562, "train_loss_llm": 0.0, "grad_norm": 0.03354305773973465, "global_step": 28265, "epoch": 785, "lr": 0.001} +{"train_loss": 0.029226111248135567, "train_loss_bc": 0.029226111248135567, "train_loss_llm": 0.0, "grad_norm": 0.046570129692554474, "global_step": 28266, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03593705594539642, "train_loss_bc": 0.03593705594539642, "train_loss_llm": 0.0, "grad_norm": 0.06912389397621155, "global_step": 28267, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03044041246175766, "train_loss_bc": 0.030431879684329033, "train_loss_llm": 0.0008532404899597168, "grad_norm": 0.08875420689582825, "global_step": 28268, "epoch": 785, "lr": 0.001} +{"train_loss": 0.030954206362366676, "train_loss_bc": 0.030947750434279442, "train_loss_llm": 0.0006455183029174805, "grad_norm": 0.035758256912231445, "global_step": 28269, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03450030833482742, "train_loss_bc": 0.03450030833482742, "train_loss_llm": 0.0, "grad_norm": 0.062092605978250504, "global_step": 28270, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03421182557940483, "train_loss_bc": 0.03421182557940483, "train_loss_llm": 0.0, "grad_norm": 0.08377116173505783, "global_step": 28271, "epoch": 785, "lr": 0.001} +{"train_loss": 0.028004029765725136, "train_loss_bc": 0.028004029765725136, "train_loss_llm": 0.0, "grad_norm": 0.09304837882518768, "global_step": 28272, "epoch": 785, "lr": 0.001} +{"train_loss": 0.030487049371004105, "train_loss_bc": 0.030487049371004105, "train_loss_llm": 0.0, "grad_norm": 0.042610593140125275, "global_step": 28273, "epoch": 785, "lr": 0.001} +{"train_loss": 0.02840523049235344, "train_loss_bc": 0.02840523049235344, "train_loss_llm": 0.0, "grad_norm": 0.0660938024520874, "global_step": 28274, "epoch": 785, "lr": 0.001} +{"train_loss": 0.028206992894411087, "train_loss_bc": 0.028206992894411087, "train_loss_llm": 0.0, "grad_norm": 0.07868434488773346, "global_step": 28275, "epoch": 785, "lr": 0.001} +{"train_loss": 0.031824927777051926, "train_loss_bc": 0.031824927777051926, "train_loss_llm": 0.0, "grad_norm": 0.09957533329725266, "global_step": 28276, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03459416329860687, "train_loss_bc": 0.03459416329860687, "train_loss_llm": 0.0, "grad_norm": 0.03902606666088104, "global_step": 28277, "epoch": 785, "lr": 0.001} +{"train_loss": 0.028468796983361244, "train_loss_bc": 0.028468796983361244, "train_loss_llm": 0.0, "grad_norm": 0.06357727944850922, "global_step": 28278, "epoch": 785, "lr": 0.001} +{"train_loss": 0.026066308841109276, "train_loss_bc": 0.026058759540319443, "train_loss_llm": 0.0007548928260803223, "grad_norm": 0.0705694630742073, "global_step": 28279, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03037392534315586, "train_loss_bc": 0.03037392534315586, "train_loss_llm": 0.0, "grad_norm": 0.08465515077114105, "global_step": 28280, "epoch": 785, "lr": 0.001} +{"train_loss": 0.034640323370695114, "train_loss_bc": 0.034640323370695114, "train_loss_llm": 0.0, "grad_norm": 0.03330017998814583, "global_step": 28281, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03161725401878357, "train_loss_bc": 0.03161725401878357, "train_loss_llm": 0.0, "grad_norm": 0.046223726123571396, "global_step": 28282, "epoch": 785, "lr": 0.001} +{"train_loss": 0.02526715211570263, "train_loss_bc": 0.02526715211570263, "train_loss_llm": 0.0, "grad_norm": 0.06120894476771355, "global_step": 28283, "epoch": 785, "lr": 0.001} +{"train_loss": 0.029804952442646027, "train_loss_bc": 0.029804952442646027, "train_loss_llm": 0.0, "grad_norm": 0.08905494958162308, "global_step": 28284, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03062460944056511, "train_loss_bc": 0.030609216541051865, "train_loss_llm": 0.0015392899513244629, "grad_norm": 0.035502344369888306, "global_step": 28285, "epoch": 785, "lr": 0.001} +{"train_loss": 0.030780093744397163, "train_loss_bc": 0.03077108971774578, "train_loss_llm": 0.0009003877639770508, "grad_norm": 0.05911487713456154, "global_step": 28286, "epoch": 785, "lr": 0.001} +{"train_loss": 0.02940281108021736, "train_loss_bc": 0.029401259496808052, "train_loss_llm": 0.0001550912857055664, "grad_norm": 0.07534248381853104, "global_step": 28287, "epoch": 785, "lr": 0.001} +{"train_loss": 0.03443583473563194, "train_loss_bc": 0.03443583473563194, "train_loss_llm": 0.0, "grad_norm": 0.1042524203658104, "global_step": 28288, "epoch": 785, "lr": 0.001} +{"train_loss": 0.01995672658085823, "train_loss_bc": 0.019949769601225853, "train_loss_llm": 0.0006956756114959717, "grad_norm": 0.032050054520368576, "global_step": 28289, "epoch": 785, "lr": 0.001} +{"train_loss": 0.02713361755013466, "train_loss_bc": 0.027113012969493866, "train_loss_llm": 0.002060502767562866, "grad_norm": 0.05075260251760483, "global_step": 28290, "epoch": 785, "lr": 0.001} +{"train_loss": 0.028335750102996826, "train_loss_bc": 0.028335750102996826, "train_loss_llm": 0.0, "grad_norm": 0.06481096893548965, "global_step": 28291, "epoch": 785, "lr": 0.001} +{"train_loss": 0.028921162709593773, "train_loss_bc": 0.028921162709593773, "train_loss_llm": 0.0, "grad_norm": 0.07884137332439423, "global_step": 28292, "epoch": 785, "lr": 0.001} +{"train_loss": 0.02788889966905117, "train_loss_bc": 0.02785194292664528, "train_loss_llm": 0.003695666790008545, "grad_norm": 0.029714757576584816, "global_step": 28293, "epoch": 785, "lr": 0.001} +{"train_loss": 0.030529197305440903, "train_loss_bc": 0.030518364161252975, "train_loss_llm": 0.0010834038257598877, "grad_norm": 0.0474737249314785, "global_step": 28294, "epoch": 785, "lr": 0.001} +{"train_loss": 0.030334851963238582, "train_loss_bc": 0.03480930253863335, "train_loss_llm": 0.0034204721450805664, "grad_norm": 0.07297290116548538, "global_step": 28295, "epoch": 785, "lr": 0.001, "train/cumulative_reward": 41.5342577859066, "train/mean_score": 7.532773083944377, "train/success_rate": 0.2, "test/cumulative_reward": 33.48930306345915, "test/mean_score": 8.429933106031285, "test/success_rate": 0.24, "val_loss": 0.0946185290813446, "train_action_mse_error": 0.033693864941596985} +{"train_loss": 0.026798641309142113, "train_loss_bc": 0.0267893485724926, "train_loss_llm": 0.0009292960166931152, "grad_norm": 0.09292344003915787, "global_step": 28296, "epoch": 786, "lr": 0.001} +{"train_loss": 0.028490619733929634, "train_loss_bc": 0.028490619733929634, "train_loss_llm": 0.0, "grad_norm": 0.03387506678700447, "global_step": 28297, "epoch": 786, "lr": 0.001} +{"train_loss": 0.027228444814682007, "train_loss_bc": 0.027228444814682007, "train_loss_llm": 0.0, "grad_norm": 0.04881146922707558, "global_step": 28298, "epoch": 786, "lr": 0.001} +{"train_loss": 0.026821400970220566, "train_loss_bc": 0.02681327611207962, "train_loss_llm": 0.0008125007152557373, "grad_norm": 0.062289174646139145, "global_step": 28299, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03080064244568348, "train_loss_bc": 0.03080064244568348, "train_loss_llm": 0.0, "grad_norm": 0.07412134110927582, "global_step": 28300, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03005244955420494, "train_loss_bc": 0.03005244955420494, "train_loss_llm": 0.0, "grad_norm": 0.039043575525283813, "global_step": 28301, "epoch": 786, "lr": 0.001} +{"train_loss": 0.02964695170521736, "train_loss_bc": 0.02963251806795597, "train_loss_llm": 0.0014434456825256348, "grad_norm": 0.0502835251390934, "global_step": 28302, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03286179155111313, "train_loss_bc": 0.03286179155111313, "train_loss_llm": 0.0, "grad_norm": 0.07410719245672226, "global_step": 28303, "epoch": 786, "lr": 0.001} +{"train_loss": 0.02703150548040867, "train_loss_bc": 0.027013834565877914, "train_loss_llm": 0.0017670989036560059, "grad_norm": 0.09060759842395782, "global_step": 28304, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03035731241106987, "train_loss_bc": 0.03035731241106987, "train_loss_llm": 0.0, "grad_norm": 0.03378493711352348, "global_step": 28305, "epoch": 786, "lr": 0.001} +{"train_loss": 0.028311055153608322, "train_loss_bc": 0.028311055153608322, "train_loss_llm": 0.0, "grad_norm": 0.052126601338386536, "global_step": 28306, "epoch": 786, "lr": 0.001} +{"train_loss": 0.026430431753396988, "train_loss_bc": 0.026430431753396988, "train_loss_llm": 0.0, "grad_norm": 0.07903090119361877, "global_step": 28307, "epoch": 786, "lr": 0.001} +{"train_loss": 0.029719533398747444, "train_loss_bc": 0.029719533398747444, "train_loss_llm": 0.0, "grad_norm": 0.09565018117427826, "global_step": 28308, "epoch": 786, "lr": 0.001} +{"train_loss": 0.026011692360043526, "train_loss_bc": 0.026010029017925262, "train_loss_llm": 0.00016635656356811523, "grad_norm": 0.02704152837395668, "global_step": 28309, "epoch": 786, "lr": 0.001} +{"train_loss": 0.028733136132359505, "train_loss_bc": 0.028733136132359505, "train_loss_llm": 0.0, "grad_norm": 0.042490843683481216, "global_step": 28310, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03455095738172531, "train_loss_bc": 0.03455095738172531, "train_loss_llm": 0.0, "grad_norm": 0.0550072081387043, "global_step": 28311, "epoch": 786, "lr": 0.001} +{"train_loss": 0.034031618386507034, "train_loss_bc": 0.034031618386507034, "train_loss_llm": 0.0, "grad_norm": 0.06926736980676651, "global_step": 28312, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03371373564004898, "train_loss_bc": 0.03371373564004898, "train_loss_llm": 0.0, "grad_norm": 0.042575571686029434, "global_step": 28313, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03697901591658592, "train_loss_bc": 0.03697901591658592, "train_loss_llm": 0.0, "grad_norm": 0.06284291297197342, "global_step": 28314, "epoch": 786, "lr": 0.001} +{"train_loss": 0.031345807015895844, "train_loss_bc": 0.031345807015895844, "train_loss_llm": 0.0, "grad_norm": 0.07994233816862106, "global_step": 28315, "epoch": 786, "lr": 0.001} +{"train_loss": 0.034266214817762375, "train_loss_bc": 0.034266214817762375, "train_loss_llm": 0.0, "grad_norm": 0.08824627846479416, "global_step": 28316, "epoch": 786, "lr": 0.001} +{"train_loss": 0.029210181906819344, "train_loss_bc": 0.029207123443484306, "train_loss_llm": 0.0003059208393096924, "grad_norm": 0.030371394008398056, "global_step": 28317, "epoch": 786, "lr": 0.001} +{"train_loss": 0.030402017757296562, "train_loss_bc": 0.030402017757296562, "train_loss_llm": 0.0, "grad_norm": 0.04716818034648895, "global_step": 28318, "epoch": 786, "lr": 0.001} +{"train_loss": 0.028058089315891266, "train_loss_bc": 0.028058089315891266, "train_loss_llm": 0.0, "grad_norm": 0.05917815491557121, "global_step": 28319, "epoch": 786, "lr": 0.001} +{"train_loss": 0.02366592176258564, "train_loss_bc": 0.02366592176258564, "train_loss_llm": 0.0, "grad_norm": 0.0691831111907959, "global_step": 28320, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03469199687242508, "train_loss_bc": 0.03469199687242508, "train_loss_llm": 0.0, "grad_norm": 0.036153681576251984, "global_step": 28321, "epoch": 786, "lr": 0.001} +{"train_loss": 0.023410705849528313, "train_loss_bc": 0.023410705849528313, "train_loss_llm": 0.0, "grad_norm": 0.05032045394182205, "global_step": 28322, "epoch": 786, "lr": 0.001} +{"train_loss": 0.02662297897040844, "train_loss_bc": 0.02662297897040844, "train_loss_llm": 0.0, "grad_norm": 0.07020693272352219, "global_step": 28323, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03354596719145775, "train_loss_bc": 0.03354596719145775, "train_loss_llm": 0.0, "grad_norm": 0.10306508839130402, "global_step": 28324, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03124283067882061, "train_loss_bc": 0.03124283067882061, "train_loss_llm": 0.0, "grad_norm": 0.037281785160303116, "global_step": 28325, "epoch": 786, "lr": 0.001} +{"train_loss": 0.030159929767251015, "train_loss_bc": 0.03014930710196495, "train_loss_llm": 0.0010623037815093994, "grad_norm": 0.057180386036634445, "global_step": 28326, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03217293694615364, "train_loss_bc": 0.03217293694615364, "train_loss_llm": 0.0, "grad_norm": 0.07608699053525925, "global_step": 28327, "epoch": 786, "lr": 0.001} +{"train_loss": 0.02851654402911663, "train_loss_bc": 0.02851654402911663, "train_loss_llm": 0.0, "grad_norm": 0.08225341886281967, "global_step": 28328, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03470238670706749, "train_loss_bc": 0.03469834849238396, "train_loss_llm": 0.0004038512706756592, "grad_norm": 0.029057925567030907, "global_step": 28329, "epoch": 786, "lr": 0.001} +{"train_loss": 0.03152262791991234, "train_loss_bc": 0.03150270879268646, "train_loss_llm": 0.001991957426071167, "grad_norm": 0.04335152730345726, "global_step": 28330, "epoch": 786, "lr": 0.001} +{"train_loss": 0.030320986018826563, "train_loss_bc": 0.039447423070669174, "train_loss_llm": 0.0, "grad_norm": 0.06899791955947876, "global_step": 28331, "epoch": 786, "lr": 0.001, "val_loss": 0.0961008369922638} +{"train_loss": 0.03029303066432476, "train_loss_bc": 0.03029303066432476, "train_loss_llm": 0.0, "grad_norm": 0.0860561653971672, "global_step": 28332, "epoch": 787, "lr": 0.001} +{"train_loss": 0.02709079720079899, "train_loss_bc": 0.02709079720079899, "train_loss_llm": 0.0, "grad_norm": 0.028874365612864494, "global_step": 28333, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03593830391764641, "train_loss_bc": 0.03593830391764641, "train_loss_llm": 0.0, "grad_norm": 0.04397512972354889, "global_step": 28334, "epoch": 787, "lr": 0.001} +{"train_loss": 0.027210427448153496, "train_loss_bc": 0.027210427448153496, "train_loss_llm": 0.0, "grad_norm": 0.049517035484313965, "global_step": 28335, "epoch": 787, "lr": 0.001} +{"train_loss": 0.024082275107502937, "train_loss_bc": 0.02405693009495735, "train_loss_llm": 0.002534419298171997, "grad_norm": 0.05903828889131546, "global_step": 28336, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03086075559258461, "train_loss_bc": 0.03086075559258461, "train_loss_llm": 0.0, "grad_norm": 0.027551859617233276, "global_step": 28337, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03054174780845642, "train_loss_bc": 0.03054174780845642, "train_loss_llm": 0.0, "grad_norm": 0.043456338346004486, "global_step": 28338, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03157171234488487, "train_loss_bc": 0.03157081827521324, "train_loss_llm": 8.925795555114746e-05, "grad_norm": 0.053205084055662155, "global_step": 28339, "epoch": 787, "lr": 0.001} +{"train_loss": 0.023267339915037155, "train_loss_bc": 0.023267339915037155, "train_loss_llm": 0.0, "grad_norm": 0.06746283918619156, "global_step": 28340, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03694925829768181, "train_loss_bc": 0.03694925829768181, "train_loss_llm": 0.0, "grad_norm": 0.04763905704021454, "global_step": 28341, "epoch": 787, "lr": 0.001} +{"train_loss": 0.027629796415567398, "train_loss_bc": 0.027600441128015518, "train_loss_llm": 0.0029356181621551514, "grad_norm": 0.0632990300655365, "global_step": 28342, "epoch": 787, "lr": 0.001} +{"train_loss": 0.028916701674461365, "train_loss_bc": 0.028904367238283157, "train_loss_llm": 0.0012333989143371582, "grad_norm": 0.08338655531406403, "global_step": 28343, "epoch": 787, "lr": 0.001} +{"train_loss": 0.0252061877399683, "train_loss_bc": 0.0252061877399683, "train_loss_llm": 0.0, "grad_norm": 0.09363160282373428, "global_step": 28344, "epoch": 787, "lr": 0.001} +{"train_loss": 0.031098676845431328, "train_loss_bc": 0.031098676845431328, "train_loss_llm": 0.0, "grad_norm": 0.03038608469069004, "global_step": 28345, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03277398273348808, "train_loss_bc": 0.03277398273348808, "train_loss_llm": 0.0, "grad_norm": 0.04590768739581108, "global_step": 28346, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03208811581134796, "train_loss_bc": 0.03208811581134796, "train_loss_llm": 0.0, "grad_norm": 0.06435103714466095, "global_step": 28347, "epoch": 787, "lr": 0.001} +{"train_loss": 0.030322259292006493, "train_loss_bc": 0.030322259292006493, "train_loss_llm": 0.0, "grad_norm": 0.07842937856912613, "global_step": 28348, "epoch": 787, "lr": 0.001} +{"train_loss": 0.031991418451070786, "train_loss_bc": 0.031991418451070786, "train_loss_llm": 0.0, "grad_norm": 0.037740226835012436, "global_step": 28349, "epoch": 787, "lr": 0.001} +{"train_loss": 0.028536666184663773, "train_loss_bc": 0.028536666184663773, "train_loss_llm": 0.0, "grad_norm": 0.06231674924492836, "global_step": 28350, "epoch": 787, "lr": 0.001} +{"train_loss": 0.027910472825169563, "train_loss_bc": 0.027910472825169563, "train_loss_llm": 0.0, "grad_norm": 0.07203678786754608, "global_step": 28351, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03460345417261124, "train_loss_bc": 0.03459231182932854, "train_loss_llm": 0.0011140704154968262, "grad_norm": 0.080149345099926, "global_step": 28352, "epoch": 787, "lr": 0.001} +{"train_loss": 0.035024359822273254, "train_loss_bc": 0.035024359822273254, "train_loss_llm": 0.0, "grad_norm": 0.03809911757707596, "global_step": 28353, "epoch": 787, "lr": 0.001} +{"train_loss": 0.030216315761208534, "train_loss_bc": 0.030216315761208534, "train_loss_llm": 0.0, "grad_norm": 0.053509995341300964, "global_step": 28354, "epoch": 787, "lr": 0.001} +{"train_loss": 0.02912602201104164, "train_loss_bc": 0.02912602201104164, "train_loss_llm": 0.0, "grad_norm": 0.06413614749908447, "global_step": 28355, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03420158103108406, "train_loss_bc": 0.03420158103108406, "train_loss_llm": 0.0, "grad_norm": 0.09465856850147247, "global_step": 28356, "epoch": 787, "lr": 0.001} +{"train_loss": 0.02788924239575863, "train_loss_bc": 0.02788924239575863, "train_loss_llm": 0.0, "grad_norm": 0.034098416566848755, "global_step": 28357, "epoch": 787, "lr": 0.001} +{"train_loss": 0.030745409429073334, "train_loss_bc": 0.030745409429073334, "train_loss_llm": 0.0, "grad_norm": 0.0499664805829525, "global_step": 28358, "epoch": 787, "lr": 0.001} +{"train_loss": 0.037512876093387604, "train_loss_bc": 0.037508703768253326, "train_loss_llm": 0.0004171133041381836, "grad_norm": 0.07074379920959473, "global_step": 28359, "epoch": 787, "lr": 0.001} +{"train_loss": 0.028773445636034012, "train_loss_bc": 0.028773445636034012, "train_loss_llm": 0.0, "grad_norm": 0.08256620913743973, "global_step": 28360, "epoch": 787, "lr": 0.001} +{"train_loss": 0.024220725521445274, "train_loss_bc": 0.024220725521445274, "train_loss_llm": 0.0, "grad_norm": 0.03597720339894295, "global_step": 28361, "epoch": 787, "lr": 0.001} +{"train_loss": 0.026938121765851974, "train_loss_bc": 0.026938121765851974, "train_loss_llm": 0.0, "grad_norm": 0.06073862314224243, "global_step": 28362, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03286845609545708, "train_loss_bc": 0.032839883118867874, "train_loss_llm": 0.0028571486473083496, "grad_norm": 0.07972975075244904, "global_step": 28363, "epoch": 787, "lr": 0.001} +{"train_loss": 0.030433373525738716, "train_loss_bc": 0.030433373525738716, "train_loss_llm": 0.0, "grad_norm": 0.0920325368642807, "global_step": 28364, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03019579127430916, "train_loss_bc": 0.03019579127430916, "train_loss_llm": 0.0, "grad_norm": 0.042939309030771255, "global_step": 28365, "epoch": 787, "lr": 0.001} +{"train_loss": 0.027757836505770683, "train_loss_bc": 0.027757836505770683, "train_loss_llm": 0.0, "grad_norm": 0.05005229264497757, "global_step": 28366, "epoch": 787, "lr": 0.001} +{"train_loss": 0.03024636979939209, "train_loss_bc": 0.03405945003032684, "train_loss_llm": 0.0022923946380615234, "grad_norm": 0.06953790783882141, "global_step": 28367, "epoch": 787, "lr": 0.001, "val_loss": 0.09667462855577469} +{"train_loss": 0.033093325793743134, "train_loss_bc": 0.033093325793743134, "train_loss_llm": 0.0, "grad_norm": 0.08646023273468018, "global_step": 28368, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03229958191514015, "train_loss_bc": 0.03229958191514015, "train_loss_llm": 0.0, "grad_norm": 0.03676418587565422, "global_step": 28369, "epoch": 788, "lr": 0.001} +{"train_loss": 0.0319165363907814, "train_loss_bc": 0.03191062808036804, "train_loss_llm": 0.0005909204483032227, "grad_norm": 0.06112419068813324, "global_step": 28370, "epoch": 788, "lr": 0.001} +{"train_loss": 0.028054354712367058, "train_loss_bc": 0.02805088274180889, "train_loss_llm": 0.0003472268581390381, "grad_norm": 0.08064498752355576, "global_step": 28371, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03896709531545639, "train_loss_bc": 0.03896709531545639, "train_loss_llm": 0.0, "grad_norm": 0.10816685855388641, "global_step": 28372, "epoch": 788, "lr": 0.001} +{"train_loss": 0.0266150813549757, "train_loss_bc": 0.0266150813549757, "train_loss_llm": 0.0, "grad_norm": 0.0288540106266737, "global_step": 28373, "epoch": 788, "lr": 0.001} +{"train_loss": 0.027229130268096924, "train_loss_bc": 0.027229130268096924, "train_loss_llm": 0.0, "grad_norm": 0.04354357346892357, "global_step": 28374, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03693576902151108, "train_loss_bc": 0.03693576902151108, "train_loss_llm": 0.0, "grad_norm": 0.07694847136735916, "global_step": 28375, "epoch": 788, "lr": 0.001} +{"train_loss": 0.027834942564368248, "train_loss_bc": 0.02782457508146763, "train_loss_llm": 0.0010367929935455322, "grad_norm": 0.09236356616020203, "global_step": 28376, "epoch": 788, "lr": 0.001} +{"train_loss": 0.031187068670988083, "train_loss_bc": 0.031187068670988083, "train_loss_llm": 0.0, "grad_norm": 0.03225480020046234, "global_step": 28377, "epoch": 788, "lr": 0.001} +{"train_loss": 0.02699900232255459, "train_loss_bc": 0.02699900232255459, "train_loss_llm": 0.0, "grad_norm": 0.053148459643125534, "global_step": 28378, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03150574862957001, "train_loss_bc": 0.03150574862957001, "train_loss_llm": 0.0, "grad_norm": 0.07024654000997543, "global_step": 28379, "epoch": 788, "lr": 0.001} +{"train_loss": 0.02789030782878399, "train_loss_bc": 0.02789030782878399, "train_loss_llm": 0.0, "grad_norm": 0.07183679193258286, "global_step": 28380, "epoch": 788, "lr": 0.001} +{"train_loss": 0.02953789010643959, "train_loss_bc": 0.029526567086577415, "train_loss_llm": 0.0011323392391204834, "grad_norm": 0.03303025662899017, "global_step": 28381, "epoch": 788, "lr": 0.001} +{"train_loss": 0.028837203979492188, "train_loss_bc": 0.028837203979492188, "train_loss_llm": 0.0, "grad_norm": 0.0545954704284668, "global_step": 28382, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03428076207637787, "train_loss_bc": 0.03428076207637787, "train_loss_llm": 0.0, "grad_norm": 0.06969279050827026, "global_step": 28383, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03464384377002716, "train_loss_bc": 0.03464384377002716, "train_loss_llm": 0.0, "grad_norm": 0.07258020341396332, "global_step": 28384, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03069869801402092, "train_loss_bc": 0.03069869801402092, "train_loss_llm": 0.0, "grad_norm": 0.03245971351861954, "global_step": 28385, "epoch": 788, "lr": 0.001} +{"train_loss": 0.02940240316092968, "train_loss_bc": 0.02940240316092968, "train_loss_llm": 0.0, "grad_norm": 0.05007147416472435, "global_step": 28386, "epoch": 788, "lr": 0.001} +{"train_loss": 0.024481238797307014, "train_loss_bc": 0.024481238797307014, "train_loss_llm": 0.0, "grad_norm": 0.061506547033786774, "global_step": 28387, "epoch": 788, "lr": 0.001} +{"train_loss": 0.032180123031139374, "train_loss_bc": 0.03217356279492378, "train_loss_llm": 0.0006561577320098877, "grad_norm": 0.08674081414937973, "global_step": 28388, "epoch": 788, "lr": 0.001} +{"train_loss": 0.02805134654045105, "train_loss_bc": 0.02805134654045105, "train_loss_llm": 0.0, "grad_norm": 0.02861339971423149, "global_step": 28389, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03275837004184723, "train_loss_bc": 0.03275837004184723, "train_loss_llm": 0.0, "grad_norm": 0.059336621314287186, "global_step": 28390, "epoch": 788, "lr": 0.001} +{"train_loss": 0.02850518561899662, "train_loss_bc": 0.02850518561899662, "train_loss_llm": 0.0, "grad_norm": 0.06715971231460571, "global_step": 28391, "epoch": 788, "lr": 0.001} +{"train_loss": 0.031200258061289787, "train_loss_bc": 0.031200258061289787, "train_loss_llm": 0.0, "grad_norm": 0.08049095422029495, "global_step": 28392, "epoch": 788, "lr": 0.001} +{"train_loss": 0.025687137618660927, "train_loss_bc": 0.025687137618660927, "train_loss_llm": 0.0, "grad_norm": 0.03737279400229454, "global_step": 28393, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03232154622673988, "train_loss_bc": 0.03232154622673988, "train_loss_llm": 0.0, "grad_norm": 0.05766484513878822, "global_step": 28394, "epoch": 788, "lr": 0.001} +{"train_loss": 0.030238235369324684, "train_loss_bc": 0.03023359179496765, "train_loss_llm": 0.00046432018280029297, "grad_norm": 0.07767325639724731, "global_step": 28395, "epoch": 788, "lr": 0.001} +{"train_loss": 0.028882145881652832, "train_loss_bc": 0.028873974457383156, "train_loss_llm": 0.0008170604705810547, "grad_norm": 0.09645416587591171, "global_step": 28396, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03277529403567314, "train_loss_bc": 0.03277529403567314, "train_loss_llm": 0.0, "grad_norm": 0.03646489605307579, "global_step": 28397, "epoch": 788, "lr": 0.001} +{"train_loss": 0.02749708667397499, "train_loss_bc": 0.02749708667397499, "train_loss_llm": 0.0, "grad_norm": 0.054812509566545486, "global_step": 28398, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03161574900150299, "train_loss_bc": 0.03161574900150299, "train_loss_llm": 0.0, "grad_norm": 0.0758204311132431, "global_step": 28399, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03372133895754814, "train_loss_bc": 0.03372133895754814, "train_loss_llm": 0.0, "grad_norm": 0.09347814321517944, "global_step": 28400, "epoch": 788, "lr": 0.001} +{"train_loss": 0.02680007740855217, "train_loss_bc": 0.02680007740855217, "train_loss_llm": 0.0, "grad_norm": 0.030162349343299866, "global_step": 28401, "epoch": 788, "lr": 0.001} +{"train_loss": 0.02643476612865925, "train_loss_bc": 0.02643047831952572, "train_loss_llm": 0.00042882561683654785, "grad_norm": 0.047440677881240845, "global_step": 28402, "epoch": 788, "lr": 0.001} +{"train_loss": 0.03021866036579013, "train_loss_bc": 0.02679312787950039, "train_loss_llm": 0.0, "grad_norm": 0.06266868114471436, "global_step": 28403, "epoch": 788, "lr": 0.001, "val_loss": 0.09620711207389832} +{"train_loss": 0.025633182376623154, "train_loss_bc": 0.025633182376623154, "train_loss_llm": 0.0, "grad_norm": 0.07992121577262878, "global_step": 28404, "epoch": 789, "lr": 0.001} +{"train_loss": 0.03737625479698181, "train_loss_bc": 0.03737127035856247, "train_loss_llm": 0.0004985332489013672, "grad_norm": 0.03847961500287056, "global_step": 28405, "epoch": 789, "lr": 0.001} +{"train_loss": 0.03001009300351143, "train_loss_bc": 0.03001009300351143, "train_loss_llm": 0.0, "grad_norm": 0.05520249158143997, "global_step": 28406, "epoch": 789, "lr": 0.001} +{"train_loss": 0.03509034588932991, "train_loss_bc": 0.0350714772939682, "train_loss_llm": 0.0018868446350097656, "grad_norm": 0.07169099152088165, "global_step": 28407, "epoch": 789, "lr": 0.001} +{"train_loss": 0.030093597248196602, "train_loss_bc": 0.030087396502494812, "train_loss_llm": 0.0006200075149536133, "grad_norm": 0.09115158021450043, "global_step": 28408, "epoch": 789, "lr": 0.001} +{"train_loss": 0.02957838401198387, "train_loss_bc": 0.029575815424323082, "train_loss_llm": 0.0002568960189819336, "grad_norm": 0.03772248700261116, "global_step": 28409, "epoch": 789, "lr": 0.001} +{"train_loss": 0.02887766994535923, "train_loss_bc": 0.02887766994535923, "train_loss_llm": 0.0, "grad_norm": 0.06391177326440811, "global_step": 28410, "epoch": 789, "lr": 0.001} +{"train_loss": 0.030269060283899307, "train_loss_bc": 0.030269060283899307, "train_loss_llm": 0.0, "grad_norm": 0.09250719845294952, "global_step": 28411, "epoch": 789, "lr": 0.001} +{"train_loss": 0.02945951744914055, "train_loss_bc": 0.02945588156580925, "train_loss_llm": 0.0003634989261627197, "grad_norm": 0.10499689728021622, "global_step": 28412, "epoch": 789, "lr": 0.001} +{"train_loss": 0.02964705415070057, "train_loss_bc": 0.029645798727869987, "train_loss_llm": 0.00012549757957458496, "grad_norm": 0.03237377107143402, "global_step": 28413, "epoch": 789, "lr": 0.001} +{"train_loss": 0.024981312453746796, "train_loss_bc": 0.024968422949314117, "train_loss_llm": 0.0012888610363006592, "grad_norm": 0.05139481648802757, "global_step": 28414, "epoch": 789, "lr": 0.001} +{"train_loss": 0.03278252109885216, "train_loss_bc": 0.03277713060379028, "train_loss_llm": 0.0005391240119934082, "grad_norm": 0.06621941179037094, "global_step": 28415, "epoch": 789, "lr": 0.001} +{"train_loss": 0.029169149696826935, "train_loss_bc": 0.029145844280719757, "train_loss_llm": 0.0023305416107177734, "grad_norm": 0.07774829119443893, "global_step": 28416, "epoch": 789, "lr": 0.001} +{"train_loss": 0.031672120094299316, "train_loss_bc": 0.031667560338974, "train_loss_llm": 0.0004559457302093506, "grad_norm": 0.03367859870195389, "global_step": 28417, "epoch": 789, "lr": 0.001} +{"train_loss": 0.03101676143705845, "train_loss_bc": 0.03101676143705845, "train_loss_llm": 0.0, "grad_norm": 0.05023153871297836, "global_step": 28418, "epoch": 789, "lr": 0.001} +{"train_loss": 0.030051255598664284, "train_loss_bc": 0.030026651918888092, "train_loss_llm": 0.0024604499340057373, "grad_norm": 0.06651605665683746, "global_step": 28419, "epoch": 789, "lr": 0.001} +{"train_loss": 0.033951565623283386, "train_loss_bc": 0.033951565623283386, "train_loss_llm": 0.0, "grad_norm": 0.08310233056545258, "global_step": 28420, "epoch": 789, "lr": 0.001} +{"train_loss": 0.028685448691248894, "train_loss_bc": 0.028685448691248894, "train_loss_llm": 0.0, "grad_norm": 0.02738732285797596, "global_step": 28421, "epoch": 789, "lr": 0.001} +{"train_loss": 0.025726983323693275, "train_loss_bc": 0.025726983323693275, "train_loss_llm": 0.0, "grad_norm": 0.05116987228393555, "global_step": 28422, "epoch": 789, "lr": 0.001} +{"train_loss": 0.03250871226191521, "train_loss_bc": 0.03250871226191521, "train_loss_llm": 0.0, "grad_norm": 0.06321713328361511, "global_step": 28423, "epoch": 789, "lr": 0.001} +{"train_loss": 0.029716771095991135, "train_loss_bc": 0.029705453664064407, "train_loss_llm": 0.0011317431926727295, "grad_norm": 0.07642979174852371, "global_step": 28424, "epoch": 789, "lr": 0.001} +{"train_loss": 0.03101472556591034, "train_loss_bc": 0.03101472556591034, "train_loss_llm": 0.0, "grad_norm": 0.0387662909924984, "global_step": 28425, "epoch": 789, "lr": 0.001} +{"train_loss": 0.031624358147382736, "train_loss_bc": 0.031624358147382736, "train_loss_llm": 0.0, "grad_norm": 0.05798163264989853, "global_step": 28426, "epoch": 789, "lr": 0.001} +{"train_loss": 0.032997976988554, "train_loss_bc": 0.032997976988554, "train_loss_llm": 0.0, "grad_norm": 0.08063215762376785, "global_step": 28427, "epoch": 789, "lr": 0.001} +{"train_loss": 0.028003813698887825, "train_loss_bc": 0.028003813698887825, "train_loss_llm": 0.0, "grad_norm": 0.09695515036582947, "global_step": 28428, "epoch": 789, "lr": 0.001} +{"train_loss": 0.029073631390929222, "train_loss_bc": 0.029073631390929222, "train_loss_llm": 0.0, "grad_norm": 0.03205209597945213, "global_step": 28429, "epoch": 789, "lr": 0.001} +{"train_loss": 0.026298431679606438, "train_loss_bc": 0.026298431679606438, "train_loss_llm": 0.0, "grad_norm": 0.04361133277416229, "global_step": 28430, "epoch": 789, "lr": 0.001} +{"train_loss": 0.02816350944340229, "train_loss_bc": 0.02816350944340229, "train_loss_llm": 0.0, "grad_norm": 0.06267762184143066, "global_step": 28431, "epoch": 789, "lr": 0.001} +{"train_loss": 0.032396070659160614, "train_loss_bc": 0.032396070659160614, "train_loss_llm": 0.0, "grad_norm": 0.0781821608543396, "global_step": 28432, "epoch": 789, "lr": 0.001} +{"train_loss": 0.034137990325689316, "train_loss_bc": 0.034137990325689316, "train_loss_llm": 0.0, "grad_norm": 0.04300512373447418, "global_step": 28433, "epoch": 789, "lr": 0.001} +{"train_loss": 0.026503274217247963, "train_loss_bc": 0.026503274217247963, "train_loss_llm": 0.0, "grad_norm": 0.05237658694386482, "global_step": 28434, "epoch": 789, "lr": 0.001} +{"train_loss": 0.0294544268399477, "train_loss_bc": 0.0294544268399477, "train_loss_llm": 0.0, "grad_norm": 0.07358667254447937, "global_step": 28435, "epoch": 789, "lr": 0.001} +{"train_loss": 0.02804552949965, "train_loss_bc": 0.02804552949965, "train_loss_llm": 0.0, "grad_norm": 0.08403979986906052, "global_step": 28436, "epoch": 789, "lr": 0.001} +{"train_loss": 0.03459208086133003, "train_loss_bc": 0.03459208086133003, "train_loss_llm": 0.0, "grad_norm": 0.031342871487140656, "global_step": 28437, "epoch": 789, "lr": 0.001} +{"train_loss": 0.02794002927839756, "train_loss_bc": 0.02794002927839756, "train_loss_llm": 0.0, "grad_norm": 0.04963866248726845, "global_step": 28438, "epoch": 789, "lr": 0.001} +{"train_loss": 0.030332915847086243, "train_loss_bc": 0.035441361367702484, "train_loss_llm": 0.0, "grad_norm": 0.0749376192688942, "global_step": 28439, "epoch": 789, "lr": 0.001, "val_loss": 0.09441196918487549} +{"train_loss": 0.030063066631555557, "train_loss_bc": 0.030043641105294228, "train_loss_llm": 0.0019425451755523682, "grad_norm": 0.0873228907585144, "global_step": 28440, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03177256882190704, "train_loss_bc": 0.03177256882190704, "train_loss_llm": 0.0, "grad_norm": 0.036714106798172, "global_step": 28441, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03203200548887253, "train_loss_bc": 0.03203200548887253, "train_loss_llm": 0.0, "grad_norm": 0.0607411190867424, "global_step": 28442, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03410189598798752, "train_loss_bc": 0.03409932181239128, "train_loss_llm": 0.0002575516700744629, "grad_norm": 0.08632098138332367, "global_step": 28443, "epoch": 790, "lr": 0.001} +{"train_loss": 0.031054005026817322, "train_loss_bc": 0.03103911317884922, "train_loss_llm": 0.0014891624450683594, "grad_norm": 0.09545211493968964, "global_step": 28444, "epoch": 790, "lr": 0.001} +{"train_loss": 0.028108755126595497, "train_loss_bc": 0.028083238750696182, "train_loss_llm": 0.002551645040512085, "grad_norm": 0.030536197125911713, "global_step": 28445, "epoch": 790, "lr": 0.001} +{"train_loss": 0.026394473388791084, "train_loss_bc": 0.026394473388791084, "train_loss_llm": 0.0, "grad_norm": 0.04640361666679382, "global_step": 28446, "epoch": 790, "lr": 0.001} +{"train_loss": 0.02588401921093464, "train_loss_bc": 0.02588401921093464, "train_loss_llm": 0.0, "grad_norm": 0.0595499686896801, "global_step": 28447, "epoch": 790, "lr": 0.001} +{"train_loss": 0.030529920011758804, "train_loss_bc": 0.030529920011758804, "train_loss_llm": 0.0, "grad_norm": 0.07248520106077194, "global_step": 28448, "epoch": 790, "lr": 0.001} +{"train_loss": 0.028330903500318527, "train_loss_bc": 0.028330903500318527, "train_loss_llm": 0.0, "grad_norm": 0.03244638815522194, "global_step": 28449, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03053325042128563, "train_loss_bc": 0.030521903187036514, "train_loss_llm": 0.001134723424911499, "grad_norm": 0.05261518433690071, "global_step": 28450, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03469468653202057, "train_loss_bc": 0.03469468653202057, "train_loss_llm": 0.0, "grad_norm": 0.0735059380531311, "global_step": 28451, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03295231983065605, "train_loss_bc": 0.03295231983065605, "train_loss_llm": 0.0, "grad_norm": 0.09651314467191696, "global_step": 28452, "epoch": 790, "lr": 0.001} +{"train_loss": 0.0317910760641098, "train_loss_bc": 0.03177009895443916, "train_loss_llm": 0.0020976662635803223, "grad_norm": 0.048205435276031494, "global_step": 28453, "epoch": 790, "lr": 0.001} +{"train_loss": 0.030147578567266464, "train_loss_bc": 0.030147578567266464, "train_loss_llm": 0.0, "grad_norm": 0.07892943173646927, "global_step": 28454, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03176487982273102, "train_loss_bc": 0.03175736218690872, "train_loss_llm": 0.0007516443729400635, "grad_norm": 0.09803630411624908, "global_step": 28455, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03188767656683922, "train_loss_bc": 0.03188767656683922, "train_loss_llm": 0.0, "grad_norm": 0.12157711386680603, "global_step": 28456, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03088286519050598, "train_loss_bc": 0.03088286519050598, "train_loss_llm": 0.0, "grad_norm": 0.035390544682741165, "global_step": 28457, "epoch": 790, "lr": 0.001} +{"train_loss": 0.030696602538228035, "train_loss_bc": 0.030696602538228035, "train_loss_llm": 0.0, "grad_norm": 0.07347773760557175, "global_step": 28458, "epoch": 790, "lr": 0.001} +{"train_loss": 0.032262641936540604, "train_loss_bc": 0.032262641936540604, "train_loss_llm": 0.0, "grad_norm": 0.08704913407564163, "global_step": 28459, "epoch": 790, "lr": 0.001} +{"train_loss": 0.02751707099378109, "train_loss_bc": 0.027515973895788193, "train_loss_llm": 0.00010964274406433105, "grad_norm": 0.10773469507694244, "global_step": 28460, "epoch": 790, "lr": 0.001} +{"train_loss": 0.031863849610090256, "train_loss_bc": 0.03185715898871422, "train_loss_llm": 0.0006691217422485352, "grad_norm": 0.03696136549115181, "global_step": 28461, "epoch": 790, "lr": 0.001} +{"train_loss": 0.029530446976423264, "train_loss_bc": 0.029530446976423264, "train_loss_llm": 0.0, "grad_norm": 0.04755576327443123, "global_step": 28462, "epoch": 790, "lr": 0.001} +{"train_loss": 0.029405442997813225, "train_loss_bc": 0.029405442997813225, "train_loss_llm": 0.0, "grad_norm": 0.06081995740532875, "global_step": 28463, "epoch": 790, "lr": 0.001} +{"train_loss": 0.025295399129390717, "train_loss_bc": 0.025295399129390717, "train_loss_llm": 0.0, "grad_norm": 0.0740295872092247, "global_step": 28464, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03012726828455925, "train_loss_bc": 0.03012726828455925, "train_loss_llm": 0.0, "grad_norm": 0.02352687530219555, "global_step": 28465, "epoch": 790, "lr": 0.001} +{"train_loss": 0.032036054879426956, "train_loss_bc": 0.032036054879426956, "train_loss_llm": 0.0, "grad_norm": 0.05135142803192139, "global_step": 28466, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03218291327357292, "train_loss_bc": 0.03218291327357292, "train_loss_llm": 0.0, "grad_norm": 0.07861826568841934, "global_step": 28467, "epoch": 790, "lr": 0.001} +{"train_loss": 0.02725742757320404, "train_loss_bc": 0.02725742757320404, "train_loss_llm": 0.0, "grad_norm": 0.10165528208017349, "global_step": 28468, "epoch": 790, "lr": 0.001} +{"train_loss": 0.02628241665661335, "train_loss_bc": 0.02628241665661335, "train_loss_llm": 0.0, "grad_norm": 0.03299064189195633, "global_step": 28469, "epoch": 790, "lr": 0.001} +{"train_loss": 0.036334794014692307, "train_loss_bc": 0.036334794014692307, "train_loss_llm": 0.0, "grad_norm": 0.05638345703482628, "global_step": 28470, "epoch": 790, "lr": 0.001} +{"train_loss": 0.029934478923678398, "train_loss_bc": 0.029934478923678398, "train_loss_llm": 0.0, "grad_norm": 0.08520931750535965, "global_step": 28471, "epoch": 790, "lr": 0.001} +{"train_loss": 0.029346946626901627, "train_loss_bc": 0.029346946626901627, "train_loss_llm": 0.0, "grad_norm": 0.10236524790525436, "global_step": 28472, "epoch": 790, "lr": 0.001} +{"train_loss": 0.029704928398132324, "train_loss_bc": 0.029704928398132324, "train_loss_llm": 0.0, "grad_norm": 0.04566178470849991, "global_step": 28473, "epoch": 790, "lr": 0.001} +{"train_loss": 0.03245360776782036, "train_loss_bc": 0.03245360776782036, "train_loss_llm": 0.0, "grad_norm": 0.07335799187421799, "global_step": 28474, "epoch": 790, "lr": 0.001} +{"train_loss": 0.030321590912838776, "train_loss_bc": 0.026419036090373993, "train_loss_llm": 0.0, "grad_norm": 0.09553845226764679, "global_step": 28475, "epoch": 790, "lr": 0.001, "train/cumulative_reward": 43.55608623989982, "train/mean_score": 10.698703750937174, "train/success_rate": 0.3, "test/cumulative_reward": 41.657273246405076, "test/mean_score": 7.8421501476038085, "test/success_rate": 0.2, "val_loss": 0.09535334259271622, "train_action_mse_error": 0.03316648304462433} +{"train_loss": 0.03648430481553078, "train_loss_bc": 0.03648430481553078, "train_loss_llm": 0.0, "grad_norm": 0.12776324152946472, "global_step": 28476, "epoch": 791, "lr": 0.001} +{"train_loss": 0.032396093010902405, "train_loss_bc": 0.032396093010902405, "train_loss_llm": 0.0, "grad_norm": 0.03089188225567341, "global_step": 28477, "epoch": 791, "lr": 0.001} +{"train_loss": 0.02988182008266449, "train_loss_bc": 0.02988182008266449, "train_loss_llm": 0.0, "grad_norm": 0.05222563445568085, "global_step": 28478, "epoch": 791, "lr": 0.001} +{"train_loss": 0.02958100661635399, "train_loss_bc": 0.029576733708381653, "train_loss_llm": 0.0004272758960723877, "grad_norm": 0.0656181275844574, "global_step": 28479, "epoch": 791, "lr": 0.001} +{"train_loss": 0.030096063390374184, "train_loss_bc": 0.030096063390374184, "train_loss_llm": 0.0, "grad_norm": 0.0712442472577095, "global_step": 28480, "epoch": 791, "lr": 0.001} +{"train_loss": 0.032820411026477814, "train_loss_bc": 0.032820411026477814, "train_loss_llm": 0.0, "grad_norm": 0.036369260400533676, "global_step": 28481, "epoch": 791, "lr": 0.001} +{"train_loss": 0.02869960479438305, "train_loss_bc": 0.02869960479438305, "train_loss_llm": 0.0, "grad_norm": 0.06528943032026291, "global_step": 28482, "epoch": 791, "lr": 0.001} +{"train_loss": 0.02698947861790657, "train_loss_bc": 0.02698947861790657, "train_loss_llm": 0.0, "grad_norm": 0.086154043674469, "global_step": 28483, "epoch": 791, "lr": 0.001} +{"train_loss": 0.03246772661805153, "train_loss_bc": 0.03246772661805153, "train_loss_llm": 0.0, "grad_norm": 0.12367747724056244, "global_step": 28484, "epoch": 791, "lr": 0.001} +{"train_loss": 0.02688528411090374, "train_loss_bc": 0.02688528411090374, "train_loss_llm": 0.0, "grad_norm": 0.025215651839971542, "global_step": 28485, "epoch": 791, "lr": 0.001} +{"train_loss": 0.0325385145843029, "train_loss_bc": 0.0325385145843029, "train_loss_llm": 0.0, "grad_norm": 0.05229802429676056, "global_step": 28486, "epoch": 791, "lr": 0.001} +{"train_loss": 0.03034374490380287, "train_loss_bc": 0.03034374490380287, "train_loss_llm": 0.0, "grad_norm": 0.08106040209531784, "global_step": 28487, "epoch": 791, "lr": 0.001} +{"train_loss": 0.029907049611210823, "train_loss_bc": 0.029907049611210823, "train_loss_llm": 0.0, "grad_norm": 0.10540268570184708, "global_step": 28488, "epoch": 791, "lr": 0.001} +{"train_loss": 0.028508733958005905, "train_loss_bc": 0.028499681502580643, "train_loss_llm": 0.0009052455425262451, "grad_norm": 0.03236560523509979, "global_step": 28489, "epoch": 791, "lr": 0.001} +{"train_loss": 0.03445960208773613, "train_loss_bc": 0.03444972634315491, "train_loss_llm": 0.0009875893592834473, "grad_norm": 0.051509156823158264, "global_step": 28490, "epoch": 791, "lr": 0.001} +{"train_loss": 0.03345848247408867, "train_loss_bc": 0.03345481678843498, "train_loss_llm": 0.00036641955375671387, "grad_norm": 0.07638204842805862, "global_step": 28491, "epoch": 791, "lr": 0.001} +{"train_loss": 0.025378793478012085, "train_loss_bc": 0.025378793478012085, "train_loss_llm": 0.0, "grad_norm": 0.09231404215097427, "global_step": 28492, "epoch": 791, "lr": 0.001} +{"train_loss": 0.028615353628993034, "train_loss_bc": 0.028615353628993034, "train_loss_llm": 0.0, "grad_norm": 0.026147983968257904, "global_step": 28493, "epoch": 791, "lr": 0.001} +{"train_loss": 0.02900264970958233, "train_loss_bc": 0.02900264970958233, "train_loss_llm": 0.0, "grad_norm": 0.050210196524858475, "global_step": 28494, "epoch": 791, "lr": 0.001} +{"train_loss": 0.030282272025942802, "train_loss_bc": 0.030276577919721603, "train_loss_llm": 0.0005693435668945312, "grad_norm": 0.06858762353658676, "global_step": 28495, "epoch": 791, "lr": 0.001} +{"train_loss": 0.024034176021814346, "train_loss_bc": 0.024032628163695335, "train_loss_llm": 0.00015482306480407715, "grad_norm": 0.07594048231840134, "global_step": 28496, "epoch": 791, "lr": 0.001} +{"train_loss": 0.031991053372621536, "train_loss_bc": 0.031991053372621536, "train_loss_llm": 0.0, "grad_norm": 0.0463627465069294, "global_step": 28497, "epoch": 791, "lr": 0.001} +{"train_loss": 0.03143979609012604, "train_loss_bc": 0.03143979609012604, "train_loss_llm": 0.0, "grad_norm": 0.05581936985254288, "global_step": 28498, "epoch": 791, "lr": 0.001} +{"train_loss": 0.029603250324726105, "train_loss_bc": 0.02960321493446827, "train_loss_llm": 3.546476364135742e-06, "grad_norm": 0.07334063947200775, "global_step": 28499, "epoch": 791, "lr": 0.001} +{"train_loss": 0.03387397527694702, "train_loss_bc": 0.03387397527694702, "train_loss_llm": 0.0, "grad_norm": 0.09010296314954758, "global_step": 28500, "epoch": 791, "lr": 0.001} +{"train_loss": 0.030817342922091484, "train_loss_bc": 0.030817342922091484, "train_loss_llm": 0.0, "grad_norm": 0.03194320946931839, "global_step": 28501, "epoch": 791, "lr": 0.001} +{"train_loss": 0.028159648180007935, "train_loss_bc": 0.028159648180007935, "train_loss_llm": 0.0, "grad_norm": 0.04619229957461357, "global_step": 28502, "epoch": 791, "lr": 0.001} +{"train_loss": 0.027690615504980087, "train_loss_bc": 0.0276752058416605, "train_loss_llm": 0.0015409290790557861, "grad_norm": 0.06373773515224457, "global_step": 28503, "epoch": 791, "lr": 0.001} +{"train_loss": 0.02818654477596283, "train_loss_bc": 0.02818654477596283, "train_loss_llm": 0.0, "grad_norm": 0.0830269381403923, "global_step": 28504, "epoch": 791, "lr": 0.001} +{"train_loss": 0.03500126302242279, "train_loss_bc": 0.03500126302242279, "train_loss_llm": 0.0, "grad_norm": 0.0388125516474247, "global_step": 28505, "epoch": 791, "lr": 0.001} +{"train_loss": 0.029849518090486526, "train_loss_bc": 0.029849518090486526, "train_loss_llm": 0.0, "grad_norm": 0.045318201184272766, "global_step": 28506, "epoch": 791, "lr": 0.001} +{"train_loss": 0.03112812153995037, "train_loss_bc": 0.031120486557483673, "train_loss_llm": 0.0007635056972503662, "grad_norm": 0.055123716592788696, "global_step": 28507, "epoch": 791, "lr": 0.001} +{"train_loss": 0.024879086762666702, "train_loss_bc": 0.024879086762666702, "train_loss_llm": 0.0, "grad_norm": 0.0708954930305481, "global_step": 28508, "epoch": 791, "lr": 0.001} +{"train_loss": 0.034797582775354385, "train_loss_bc": 0.034797582775354385, "train_loss_llm": 0.0, "grad_norm": 0.036945417523384094, "global_step": 28509, "epoch": 791, "lr": 0.001} +{"train_loss": 0.03213386610150337, "train_loss_bc": 0.03213386610150337, "train_loss_llm": 0.0, "grad_norm": 0.061903197318315506, "global_step": 28510, "epoch": 791, "lr": 0.001} +{"train_loss": 0.030320063336855836, "train_loss_bc": 0.029139449819922447, "train_loss_llm": 0.0, "grad_norm": 0.07974401861429214, "global_step": 28511, "epoch": 791, "lr": 0.001, "val_loss": 0.09538234770298004} +{"train_loss": 0.034469421952962875, "train_loss_bc": 0.034467071294784546, "train_loss_llm": 0.0002352297306060791, "grad_norm": 0.09972627460956573, "global_step": 28512, "epoch": 792, "lr": 0.001} +{"train_loss": 0.026414521038532257, "train_loss_bc": 0.026414521038532257, "train_loss_llm": 0.0, "grad_norm": 0.035585254430770874, "global_step": 28513, "epoch": 792, "lr": 0.001} +{"train_loss": 0.028008487075567245, "train_loss_bc": 0.028002513572573662, "train_loss_llm": 0.0005973577499389648, "grad_norm": 0.06291743367910385, "global_step": 28514, "epoch": 792, "lr": 0.001} +{"train_loss": 0.026105796918272972, "train_loss_bc": 0.026105796918272972, "train_loss_llm": 0.0, "grad_norm": 0.08181264996528625, "global_step": 28515, "epoch": 792, "lr": 0.001} +{"train_loss": 0.028238743543624878, "train_loss_bc": 0.028238743543624878, "train_loss_llm": 0.0, "grad_norm": 0.10450589656829834, "global_step": 28516, "epoch": 792, "lr": 0.001} +{"train_loss": 0.025441093370318413, "train_loss_bc": 0.02541411481797695, "train_loss_llm": 0.0026977956295013428, "grad_norm": 0.031005995348095894, "global_step": 28517, "epoch": 792, "lr": 0.001} +{"train_loss": 0.02976844646036625, "train_loss_bc": 0.02976844646036625, "train_loss_llm": 0.0, "grad_norm": 0.04794522002339363, "global_step": 28518, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03308352828025818, "train_loss_bc": 0.03307594358921051, "train_loss_llm": 0.0007584989070892334, "grad_norm": 0.0644868016242981, "global_step": 28519, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03271970897912979, "train_loss_bc": 0.03271970897912979, "train_loss_llm": 0.0, "grad_norm": 0.07646481692790985, "global_step": 28520, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03356826305389404, "train_loss_bc": 0.03356826305389404, "train_loss_llm": 0.0, "grad_norm": 0.035365279763936996, "global_step": 28521, "epoch": 792, "lr": 0.001} +{"train_loss": 0.027211884036660194, "train_loss_bc": 0.02720119245350361, "train_loss_llm": 0.0010692477226257324, "grad_norm": 0.05482431873679161, "global_step": 28522, "epoch": 792, "lr": 0.001} +{"train_loss": 0.033433765172958374, "train_loss_bc": 0.033433765172958374, "train_loss_llm": 0.0, "grad_norm": 0.0662459209561348, "global_step": 28523, "epoch": 792, "lr": 0.001} +{"train_loss": 0.029278073459863663, "train_loss_bc": 0.029278073459863663, "train_loss_llm": 0.0, "grad_norm": 0.08083037286996841, "global_step": 28524, "epoch": 792, "lr": 0.001} +{"train_loss": 0.023079968988895416, "train_loss_bc": 0.023079968988895416, "train_loss_llm": 0.0, "grad_norm": 0.028575554490089417, "global_step": 28525, "epoch": 792, "lr": 0.001} +{"train_loss": 0.032547201961278915, "train_loss_bc": 0.032547201961278915, "train_loss_llm": 0.0, "grad_norm": 0.05274272337555885, "global_step": 28526, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03483991324901581, "train_loss_bc": 0.03483991324901581, "train_loss_llm": 0.0, "grad_norm": 0.07120128720998764, "global_step": 28527, "epoch": 792, "lr": 0.001} +{"train_loss": 0.026468902826309204, "train_loss_bc": 0.026468902826309204, "train_loss_llm": 0.0, "grad_norm": 0.08695812523365021, "global_step": 28528, "epoch": 792, "lr": 0.001} +{"train_loss": 0.030577490106225014, "train_loss_bc": 0.030577490106225014, "train_loss_llm": 0.0, "grad_norm": 0.029082918539643288, "global_step": 28529, "epoch": 792, "lr": 0.001} +{"train_loss": 0.02730417810380459, "train_loss_bc": 0.02728128246963024, "train_loss_llm": 0.0022895634174346924, "grad_norm": 0.04438120499253273, "global_step": 28530, "epoch": 792, "lr": 0.001} +{"train_loss": 0.031084824353456497, "train_loss_bc": 0.031084824353456497, "train_loss_llm": 0.0, "grad_norm": 0.05993121862411499, "global_step": 28531, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03074154444038868, "train_loss_bc": 0.03074154444038868, "train_loss_llm": 0.0, "grad_norm": 0.08146559447050095, "global_step": 28532, "epoch": 792, "lr": 0.001} +{"train_loss": 0.026962552219629288, "train_loss_bc": 0.026962552219629288, "train_loss_llm": 0.0, "grad_norm": 0.03219659626483917, "global_step": 28533, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03577936440706253, "train_loss_bc": 0.03577936440706253, "train_loss_llm": 0.0, "grad_norm": 0.04678456857800484, "global_step": 28534, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03159799054265022, "train_loss_bc": 0.03159799054265022, "train_loss_llm": 0.0, "grad_norm": 0.0660049319267273, "global_step": 28535, "epoch": 792, "lr": 0.001} +{"train_loss": 0.032073941081762314, "train_loss_bc": 0.03207392245531082, "train_loss_llm": 1.9073486328125e-06, "grad_norm": 0.0799470767378807, "global_step": 28536, "epoch": 792, "lr": 0.001} +{"train_loss": 0.027516337111592293, "train_loss_bc": 0.027516337111592293, "train_loss_llm": 0.0, "grad_norm": 0.028318820521235466, "global_step": 28537, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03196961432695389, "train_loss_bc": 0.03196742758154869, "train_loss_llm": 0.00021883845329284668, "grad_norm": 0.04479282349348068, "global_step": 28538, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03123701922595501, "train_loss_bc": 0.03123701922595501, "train_loss_llm": 0.0, "grad_norm": 0.057948678731918335, "global_step": 28539, "epoch": 792, "lr": 0.001} +{"train_loss": 0.022447261959314346, "train_loss_bc": 0.022447261959314346, "train_loss_llm": 0.0, "grad_norm": 0.07439642399549484, "global_step": 28540, "epoch": 792, "lr": 0.001} +{"train_loss": 0.038946326822042465, "train_loss_bc": 0.038946326822042465, "train_loss_llm": 0.0, "grad_norm": 0.03666471317410469, "global_step": 28541, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03402351588010788, "train_loss_bc": 0.03402351588010788, "train_loss_llm": 0.0, "grad_norm": 0.05633898451924324, "global_step": 28542, "epoch": 792, "lr": 0.001} +{"train_loss": 0.03267394378781319, "train_loss_bc": 0.03267394378781319, "train_loss_llm": 0.0, "grad_norm": 0.07647997885942459, "global_step": 28543, "epoch": 792, "lr": 0.001} +{"train_loss": 0.030738888308405876, "train_loss_bc": 0.030738888308405876, "train_loss_llm": 0.0, "grad_norm": 0.09620697051286697, "global_step": 28544, "epoch": 792, "lr": 0.001} +{"train_loss": 0.029884420335292816, "train_loss_bc": 0.029884420335292816, "train_loss_llm": 0.0, "grad_norm": 0.03582645207643509, "global_step": 28545, "epoch": 792, "lr": 0.001} +{"train_loss": 0.028610974550247192, "train_loss_bc": 0.028610974550247192, "train_loss_llm": 0.0, "grad_norm": 0.053634416311979294, "global_step": 28546, "epoch": 792, "lr": 0.001} +{"train_loss": 0.030185027047991753, "train_loss_bc": 0.02781306579709053, "train_loss_llm": 0.0, "grad_norm": 0.08485978841781616, "global_step": 28547, "epoch": 792, "lr": 0.001, "val_loss": 0.09527587890625} +{"train_loss": 0.030204834416508675, "train_loss_bc": 0.030191997066140175, "train_loss_llm": 0.0012836456298828125, "grad_norm": 0.10167381167411804, "global_step": 28548, "epoch": 793, "lr": 0.001} +{"train_loss": 0.030302001163363457, "train_loss_bc": 0.030302001163363457, "train_loss_llm": 0.0, "grad_norm": 0.034222301095724106, "global_step": 28549, "epoch": 793, "lr": 0.001} +{"train_loss": 0.03128528222441673, "train_loss_bc": 0.03128528222441673, "train_loss_llm": 0.0, "grad_norm": 0.05259367823600769, "global_step": 28550, "epoch": 793, "lr": 0.001} +{"train_loss": 0.031169908121228218, "train_loss_bc": 0.031153684481978416, "train_loss_llm": 0.0016223788261413574, "grad_norm": 0.07416684180498123, "global_step": 28551, "epoch": 793, "lr": 0.001} +{"train_loss": 0.021232761442661285, "train_loss_bc": 0.021232761442661285, "train_loss_llm": 0.0, "grad_norm": 0.08574660122394562, "global_step": 28552, "epoch": 793, "lr": 0.001} +{"train_loss": 0.027764683589339256, "train_loss_bc": 0.027764683589339256, "train_loss_llm": 0.0, "grad_norm": 0.03431989625096321, "global_step": 28553, "epoch": 793, "lr": 0.001} +{"train_loss": 0.029225673526525497, "train_loss_bc": 0.029224541038274765, "train_loss_llm": 0.00011333823204040527, "grad_norm": 0.05245677009224892, "global_step": 28554, "epoch": 793, "lr": 0.001} +{"train_loss": 0.029218420386314392, "train_loss_bc": 0.029218420386314392, "train_loss_llm": 0.0, "grad_norm": 0.0666528269648552, "global_step": 28555, "epoch": 793, "lr": 0.001} +{"train_loss": 0.0346585214138031, "train_loss_bc": 0.0346585214138031, "train_loss_llm": 0.0, "grad_norm": 0.0763094425201416, "global_step": 28556, "epoch": 793, "lr": 0.001} +{"train_loss": 0.025166397914290428, "train_loss_bc": 0.025166397914290428, "train_loss_llm": 0.0, "grad_norm": 0.027985895052552223, "global_step": 28557, "epoch": 793, "lr": 0.001} +{"train_loss": 0.02935904823243618, "train_loss_bc": 0.02935904823243618, "train_loss_llm": 0.0, "grad_norm": 0.04696538299322128, "global_step": 28558, "epoch": 793, "lr": 0.001} +{"train_loss": 0.031209390610456467, "train_loss_bc": 0.031209390610456467, "train_loss_llm": 0.0, "grad_norm": 0.06719768047332764, "global_step": 28559, "epoch": 793, "lr": 0.001} +{"train_loss": 0.034000322222709656, "train_loss_bc": 0.03399036079645157, "train_loss_llm": 0.0009959638118743896, "grad_norm": 0.08522661030292511, "global_step": 28560, "epoch": 793, "lr": 0.001} +{"train_loss": 0.029610812664031982, "train_loss_bc": 0.029610812664031982, "train_loss_llm": 0.0, "grad_norm": 0.02849642001092434, "global_step": 28561, "epoch": 793, "lr": 0.001} +{"train_loss": 0.033492229878902435, "train_loss_bc": 0.033479686826467514, "train_loss_llm": 0.0012544691562652588, "grad_norm": 0.05082013085484505, "global_step": 28562, "epoch": 793, "lr": 0.001} +{"train_loss": 0.03337176889181137, "train_loss_bc": 0.03337176889181137, "train_loss_llm": 0.0, "grad_norm": 0.06362123042345047, "global_step": 28563, "epoch": 793, "lr": 0.001} +{"train_loss": 0.026933111250400543, "train_loss_bc": 0.026933111250400543, "train_loss_llm": 0.0, "grad_norm": 0.08256860077381134, "global_step": 28564, "epoch": 793, "lr": 0.001} +{"train_loss": 0.028252344578504562, "train_loss_bc": 0.028252344578504562, "train_loss_llm": 0.0, "grad_norm": 0.04135502129793167, "global_step": 28565, "epoch": 793, "lr": 0.001} +{"train_loss": 0.028763355687260628, "train_loss_bc": 0.028761256486177444, "train_loss_llm": 0.00020992755889892578, "grad_norm": 0.07414846867322922, "global_step": 28566, "epoch": 793, "lr": 0.001} +{"train_loss": 0.025857100263237953, "train_loss_bc": 0.02584272436797619, "train_loss_llm": 0.0014375746250152588, "grad_norm": 0.09258030354976654, "global_step": 28567, "epoch": 793, "lr": 0.001} +{"train_loss": 0.03280625492334366, "train_loss_bc": 0.03280625492334366, "train_loss_llm": 0.0, "grad_norm": 0.1307525485754013, "global_step": 28568, "epoch": 793, "lr": 0.001} +{"train_loss": 0.030464019626379013, "train_loss_bc": 0.030464019626379013, "train_loss_llm": 0.0, "grad_norm": 0.03075014054775238, "global_step": 28569, "epoch": 793, "lr": 0.001} +{"train_loss": 0.03118680790066719, "train_loss_bc": 0.03118680790066719, "train_loss_llm": 0.0, "grad_norm": 0.060995377600193024, "global_step": 28570, "epoch": 793, "lr": 0.001} +{"train_loss": 0.03873779624700546, "train_loss_bc": 0.03869315981864929, "train_loss_llm": 0.004463493824005127, "grad_norm": 0.0800626277923584, "global_step": 28571, "epoch": 793, "lr": 0.001} +{"train_loss": 0.03319622948765755, "train_loss_bc": 0.03319622948765755, "train_loss_llm": 0.0, "grad_norm": 0.10518491268157959, "global_step": 28572, "epoch": 793, "lr": 0.001} +{"train_loss": 0.029109491035342216, "train_loss_bc": 0.029100436717271805, "train_loss_llm": 0.0009055137634277344, "grad_norm": 0.035829830914735794, "global_step": 28573, "epoch": 793, "lr": 0.001} +{"train_loss": 0.02864084020256996, "train_loss_bc": 0.02864084020256996, "train_loss_llm": 0.0, "grad_norm": 0.052077483385801315, "global_step": 28574, "epoch": 793, "lr": 0.001} +{"train_loss": 0.02697913348674774, "train_loss_bc": 0.02697913348674774, "train_loss_llm": 0.0, "grad_norm": 0.07259947806596756, "global_step": 28575, "epoch": 793, "lr": 0.001} +{"train_loss": 0.024473898112773895, "train_loss_bc": 0.024473898112773895, "train_loss_llm": 0.0, "grad_norm": 0.08427383005619049, "global_step": 28576, "epoch": 793, "lr": 0.001} +{"train_loss": 0.03595783933997154, "train_loss_bc": 0.03595783933997154, "train_loss_llm": 0.0, "grad_norm": 0.032853852957487106, "global_step": 28577, "epoch": 793, "lr": 0.001} +{"train_loss": 0.03099190443754196, "train_loss_bc": 0.03099190443754196, "train_loss_llm": 0.0, "grad_norm": 0.05460511893033981, "global_step": 28578, "epoch": 793, "lr": 0.001} +{"train_loss": 0.0345640555024147, "train_loss_bc": 0.0345640555024147, "train_loss_llm": 0.0, "grad_norm": 0.07780241966247559, "global_step": 28579, "epoch": 793, "lr": 0.001} +{"train_loss": 0.023449039086699486, "train_loss_bc": 0.023449039086699486, "train_loss_llm": 0.0, "grad_norm": 0.08731019496917725, "global_step": 28580, "epoch": 793, "lr": 0.001} +{"train_loss": 0.031246082857251167, "train_loss_bc": 0.03124326840043068, "train_loss_llm": 0.00028136372566223145, "grad_norm": 0.039093293249607086, "global_step": 28581, "epoch": 793, "lr": 0.001} +{"train_loss": 0.03359554708003998, "train_loss_bc": 0.03359249606728554, "train_loss_llm": 0.0003052651882171631, "grad_norm": 0.0620599202811718, "global_step": 28582, "epoch": 793, "lr": 0.001} +{"train_loss": 0.030228726307137146, "train_loss_bc": 0.03175723925232887, "train_loss_llm": 0.0, "grad_norm": 0.0757918730378151, "global_step": 28583, "epoch": 793, "lr": 0.001, "val_loss": 0.09524521231651306} +{"train_loss": 0.030249670147895813, "train_loss_bc": 0.030238473787903786, "train_loss_llm": 0.0011195540428161621, "grad_norm": 0.10574201494455338, "global_step": 28584, "epoch": 794, "lr": 0.001} +{"train_loss": 0.024057619273662567, "train_loss_bc": 0.024057619273662567, "train_loss_llm": 0.0, "grad_norm": 0.033578671514987946, "global_step": 28585, "epoch": 794, "lr": 0.001} +{"train_loss": 0.030156323686242104, "train_loss_bc": 0.030156323686242104, "train_loss_llm": 0.0, "grad_norm": 0.04628454148769379, "global_step": 28586, "epoch": 794, "lr": 0.001} +{"train_loss": 0.0345291942358017, "train_loss_bc": 0.0345291942358017, "train_loss_llm": 0.0, "grad_norm": 0.07213221490383148, "global_step": 28587, "epoch": 794, "lr": 0.001} +{"train_loss": 0.02983664907515049, "train_loss_bc": 0.029824916273355484, "train_loss_llm": 0.0011733472347259521, "grad_norm": 0.09929867833852768, "global_step": 28588, "epoch": 794, "lr": 0.001} +{"train_loss": 0.028840845450758934, "train_loss_bc": 0.028840845450758934, "train_loss_llm": 0.0, "grad_norm": 0.04124906286597252, "global_step": 28589, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03316006809473038, "train_loss_bc": 0.03314122185111046, "train_loss_llm": 0.0018845200538635254, "grad_norm": 0.07671920955181122, "global_step": 28590, "epoch": 794, "lr": 0.001} +{"train_loss": 0.030543560162186623, "train_loss_bc": 0.030543560162186623, "train_loss_llm": 0.0, "grad_norm": 0.08754400908946991, "global_step": 28591, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03191043436527252, "train_loss_bc": 0.03189389780163765, "train_loss_llm": 0.0016535818576812744, "grad_norm": 0.09730684757232666, "global_step": 28592, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03122442215681076, "train_loss_bc": 0.031210457906126976, "train_loss_llm": 0.0013964176177978516, "grad_norm": 0.033850520849227905, "global_step": 28593, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03243054822087288, "train_loss_bc": 0.03243054822087288, "train_loss_llm": 0.0, "grad_norm": 0.052795592695474625, "global_step": 28594, "epoch": 794, "lr": 0.001} +{"train_loss": 0.025902343913912773, "train_loss_bc": 0.025902343913912773, "train_loss_llm": 0.0, "grad_norm": 0.06467882543802261, "global_step": 28595, "epoch": 794, "lr": 0.001} +{"train_loss": 0.035782478749752045, "train_loss_bc": 0.03576323390007019, "train_loss_llm": 0.001924663782119751, "grad_norm": 0.08113574981689453, "global_step": 28596, "epoch": 794, "lr": 0.001} +{"train_loss": 0.02915147878229618, "train_loss_bc": 0.02915147878229618, "train_loss_llm": 0.0, "grad_norm": 0.029545435681939125, "global_step": 28597, "epoch": 794, "lr": 0.001} +{"train_loss": 0.028212588280439377, "train_loss_bc": 0.028212588280439377, "train_loss_llm": 0.0, "grad_norm": 0.046964194625616074, "global_step": 28598, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03352494165301323, "train_loss_bc": 0.03352494165301323, "train_loss_llm": 0.0, "grad_norm": 0.06975704431533813, "global_step": 28599, "epoch": 794, "lr": 0.001} +{"train_loss": 0.037899263203144073, "train_loss_bc": 0.037899263203144073, "train_loss_llm": 0.0, "grad_norm": 0.0754198208451271, "global_step": 28600, "epoch": 794, "lr": 0.001} +{"train_loss": 0.02764408104121685, "train_loss_bc": 0.02764408104121685, "train_loss_llm": 0.0, "grad_norm": 0.028879469260573387, "global_step": 28601, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03257898986339569, "train_loss_bc": 0.03257523477077484, "train_loss_llm": 0.0003756582736968994, "grad_norm": 0.048836786299943924, "global_step": 28602, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03203393891453743, "train_loss_bc": 0.03203393891453743, "train_loss_llm": 0.0, "grad_norm": 0.06633445620536804, "global_step": 28603, "epoch": 794, "lr": 0.001} +{"train_loss": 0.032244209200143814, "train_loss_bc": 0.032234109938144684, "train_loss_llm": 0.0010100305080413818, "grad_norm": 0.07705357670783997, "global_step": 28604, "epoch": 794, "lr": 0.001} +{"train_loss": 0.02806163765490055, "train_loss_bc": 0.02806163765490055, "train_loss_llm": 0.0, "grad_norm": 0.02446492575109005, "global_step": 28605, "epoch": 794, "lr": 0.001} +{"train_loss": 0.027974680066108704, "train_loss_bc": 0.027974680066108704, "train_loss_llm": 0.0, "grad_norm": 0.03957454487681389, "global_step": 28606, "epoch": 794, "lr": 0.001} +{"train_loss": 0.026313306763768196, "train_loss_bc": 0.026313306763768196, "train_loss_llm": 0.0, "grad_norm": 0.05779136344790459, "global_step": 28607, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03199257329106331, "train_loss_bc": 0.03199257329106331, "train_loss_llm": 0.0, "grad_norm": 0.0734630674123764, "global_step": 28608, "epoch": 794, "lr": 0.001} +{"train_loss": 0.0254841111600399, "train_loss_bc": 0.0254841111600399, "train_loss_llm": 0.0, "grad_norm": 0.027552908286452293, "global_step": 28609, "epoch": 794, "lr": 0.001} +{"train_loss": 0.028274109587073326, "train_loss_bc": 0.02826978825032711, "train_loss_llm": 0.00043210387229919434, "grad_norm": 0.05048860237002373, "global_step": 28610, "epoch": 794, "lr": 0.001} +{"train_loss": 0.031794678419828415, "train_loss_bc": 0.031794678419828415, "train_loss_llm": 0.0, "grad_norm": 0.06711127609014511, "global_step": 28611, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03290800750255585, "train_loss_bc": 0.03290800750255585, "train_loss_llm": 0.0, "grad_norm": 0.0846063569188118, "global_step": 28612, "epoch": 794, "lr": 0.001} +{"train_loss": 0.030967561528086662, "train_loss_bc": 0.030959054827690125, "train_loss_llm": 0.000850677490234375, "grad_norm": 0.033138420432806015, "global_step": 28613, "epoch": 794, "lr": 0.001} +{"train_loss": 0.029092812910676003, "train_loss_bc": 0.029092812910676003, "train_loss_llm": 0.0, "grad_norm": 0.056460559368133545, "global_step": 28614, "epoch": 794, "lr": 0.001} +{"train_loss": 0.02824244275689125, "train_loss_bc": 0.02824244275689125, "train_loss_llm": 0.0, "grad_norm": 0.07440640777349472, "global_step": 28615, "epoch": 794, "lr": 0.001} +{"train_loss": 0.030044078826904297, "train_loss_bc": 0.030039874836802483, "train_loss_llm": 0.00042045116424560547, "grad_norm": 0.0900978073477745, "global_step": 28616, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03227096050977707, "train_loss_bc": 0.03225833922624588, "train_loss_llm": 0.0012621283531188965, "grad_norm": 0.03207769989967346, "global_step": 28617, "epoch": 794, "lr": 0.001} +{"train_loss": 0.02136506885290146, "train_loss_bc": 0.02136506885290146, "train_loss_llm": 0.0, "grad_norm": 0.04378904402256012, "global_step": 28618, "epoch": 794, "lr": 0.001} +{"train_loss": 0.03007427468481991, "train_loss_bc": 0.02597421035170555, "train_loss_llm": 0.0, "grad_norm": 0.0648442879319191, "global_step": 28619, "epoch": 794, "lr": 0.001, "val_loss": 0.09599953144788742} +{"train_loss": 0.03040960058569908, "train_loss_bc": 0.030396025627851486, "train_loss_llm": 0.0013574063777923584, "grad_norm": 0.08900099992752075, "global_step": 28620, "epoch": 795, "lr": 0.001} +{"train_loss": 0.030192118138074875, "train_loss_bc": 0.030192118138074875, "train_loss_llm": 0.0, "grad_norm": 0.03337550535798073, "global_step": 28621, "epoch": 795, "lr": 0.001} +{"train_loss": 0.030824098736047745, "train_loss_bc": 0.030824098736047745, "train_loss_llm": 0.0, "grad_norm": 0.040994174778461456, "global_step": 28622, "epoch": 795, "lr": 0.001} +{"train_loss": 0.029825445264577866, "train_loss_bc": 0.029825445264577866, "train_loss_llm": 0.0, "grad_norm": 0.05878862366080284, "global_step": 28623, "epoch": 795, "lr": 0.001} +{"train_loss": 0.025318121537566185, "train_loss_bc": 0.025318121537566185, "train_loss_llm": 0.0, "grad_norm": 0.06886833161115646, "global_step": 28624, "epoch": 795, "lr": 0.001} +{"train_loss": 0.0259805116802454, "train_loss_bc": 0.02597719617187977, "train_loss_llm": 0.00033149123191833496, "grad_norm": 0.030247200280427933, "global_step": 28625, "epoch": 795, "lr": 0.001} +{"train_loss": 0.035815224051475525, "train_loss_bc": 0.035815224051475525, "train_loss_llm": 0.0, "grad_norm": 0.05282074585556984, "global_step": 28626, "epoch": 795, "lr": 0.001} +{"train_loss": 0.03270982950925827, "train_loss_bc": 0.03270982950925827, "train_loss_llm": 0.0, "grad_norm": 0.07252202183008194, "global_step": 28627, "epoch": 795, "lr": 0.001} +{"train_loss": 0.02709689736366272, "train_loss_bc": 0.027079133316874504, "train_loss_llm": 0.0017763972282409668, "grad_norm": 0.082664854824543, "global_step": 28628, "epoch": 795, "lr": 0.001} +{"train_loss": 0.03506248816847801, "train_loss_bc": 0.03506248816847801, "train_loss_llm": 0.0, "grad_norm": 0.03861450031399727, "global_step": 28629, "epoch": 795, "lr": 0.001} +{"train_loss": 0.03170229122042656, "train_loss_bc": 0.03170229122042656, "train_loss_llm": 0.0, "grad_norm": 0.05765498802065849, "global_step": 28630, "epoch": 795, "lr": 0.001} +{"train_loss": 0.029622819274663925, "train_loss_bc": 0.02962072379887104, "train_loss_llm": 0.00020956993103027344, "grad_norm": 0.07489844411611557, "global_step": 28631, "epoch": 795, "lr": 0.001} +{"train_loss": 0.030369197949767113, "train_loss_bc": 0.030347799882292747, "train_loss_llm": 0.0021397769451141357, "grad_norm": 0.09834855049848557, "global_step": 28632, "epoch": 795, "lr": 0.001} +{"train_loss": 0.0322955921292305, "train_loss_bc": 0.0322955921292305, "train_loss_llm": 0.0, "grad_norm": 0.03384539112448692, "global_step": 28633, "epoch": 795, "lr": 0.001} +{"train_loss": 0.029290085658431053, "train_loss_bc": 0.029287999495863914, "train_loss_llm": 0.00020864605903625488, "grad_norm": 0.055498845875263214, "global_step": 28634, "epoch": 795, "lr": 0.001} +{"train_loss": 0.030607236549258232, "train_loss_bc": 0.030607236549258232, "train_loss_llm": 0.0, "grad_norm": 0.07048764079809189, "global_step": 28635, "epoch": 795, "lr": 0.001} +{"train_loss": 0.026025105267763138, "train_loss_bc": 0.026025105267763138, "train_loss_llm": 0.0, "grad_norm": 0.08171282708644867, "global_step": 28636, "epoch": 795, "lr": 0.001} +{"train_loss": 0.03497955948114395, "train_loss_bc": 0.03497955948114395, "train_loss_llm": 0.0, "grad_norm": 0.0333484411239624, "global_step": 28637, "epoch": 795, "lr": 0.001} +{"train_loss": 0.030051110312342644, "train_loss_bc": 0.030051110312342644, "train_loss_llm": 0.0, "grad_norm": 0.04441996291279793, "global_step": 28638, "epoch": 795, "lr": 0.001} +{"train_loss": 0.03444603085517883, "train_loss_bc": 0.03444603085517883, "train_loss_llm": 0.0, "grad_norm": 0.06698638200759888, "global_step": 28639, "epoch": 795, "lr": 0.001} +{"train_loss": 0.02853182703256607, "train_loss_bc": 0.028517499566078186, "train_loss_llm": 0.0014326870441436768, "grad_norm": 0.07981053739786148, "global_step": 28640, "epoch": 795, "lr": 0.001} +{"train_loss": 0.03850898891687393, "train_loss_bc": 0.03850898891687393, "train_loss_llm": 0.0, "grad_norm": 0.03400035202503204, "global_step": 28641, "epoch": 795, "lr": 0.001} +{"train_loss": 0.028241436928510666, "train_loss_bc": 0.028241436928510666, "train_loss_llm": 0.0, "grad_norm": 0.05810768902301788, "global_step": 28642, "epoch": 795, "lr": 0.001} +{"train_loss": 0.02339421771466732, "train_loss_bc": 0.023384585976600647, "train_loss_llm": 0.0009632408618927002, "grad_norm": 0.06923802942037582, "global_step": 28643, "epoch": 795, "lr": 0.001} +{"train_loss": 0.03120392933487892, "train_loss_bc": 0.03120392933487892, "train_loss_llm": 0.0, "grad_norm": 0.08578341454267502, "global_step": 28644, "epoch": 795, "lr": 0.001} +{"train_loss": 0.027167484164237976, "train_loss_bc": 0.027167484164237976, "train_loss_llm": 0.0, "grad_norm": 0.03141271322965622, "global_step": 28645, "epoch": 795, "lr": 0.001} +{"train_loss": 0.030514754354953766, "train_loss_bc": 0.030514754354953766, "train_loss_llm": 0.0, "grad_norm": 0.05144565552473068, "global_step": 28646, "epoch": 795, "lr": 0.001} +{"train_loss": 0.03088601678609848, "train_loss_bc": 0.03088601678609848, "train_loss_llm": 0.0, "grad_norm": 0.08216255158185959, "global_step": 28647, "epoch": 795, "lr": 0.001} +{"train_loss": 0.03380388766527176, "train_loss_bc": 0.03380388766527176, "train_loss_llm": 0.0, "grad_norm": 0.10668971389532089, "global_step": 28648, "epoch": 795, "lr": 0.001} +{"train_loss": 0.025995729491114616, "train_loss_bc": 0.025995729491114616, "train_loss_llm": 0.0, "grad_norm": 0.0325118787586689, "global_step": 28649, "epoch": 795, "lr": 0.001} +{"train_loss": 0.027973858639597893, "train_loss_bc": 0.02795606292784214, "train_loss_llm": 0.0017795264720916748, "grad_norm": 0.05897248908877373, "global_step": 28650, "epoch": 795, "lr": 0.001} +{"train_loss": 0.03445925936102867, "train_loss_bc": 0.03445925936102867, "train_loss_llm": 0.0, "grad_norm": 0.07574436813592911, "global_step": 28651, "epoch": 795, "lr": 0.001} +{"train_loss": 0.02704787068068981, "train_loss_bc": 0.02704787068068981, "train_loss_llm": 0.0, "grad_norm": 0.09537072479724884, "global_step": 28652, "epoch": 795, "lr": 0.001} +{"train_loss": 0.028894132003188133, "train_loss_bc": 0.02888769656419754, "train_loss_llm": 0.0006435811519622803, "grad_norm": 0.03189472854137421, "global_step": 28653, "epoch": 795, "lr": 0.001} +{"train_loss": 0.02788923867046833, "train_loss_bc": 0.02788923867046833, "train_loss_llm": 0.0, "grad_norm": 0.051774609833955765, "global_step": 28654, "epoch": 795, "lr": 0.001} +{"train_loss": 0.030253841645187803, "train_loss_bc": 0.03200219199061394, "train_loss_llm": 1.1146068572998047e-05, "grad_norm": 0.0798415020108223, "global_step": 28655, "epoch": 795, "lr": 0.001, "train/cumulative_reward": 84.0881664925796, "train/mean_score": 18.839775886686077, "train/success_rate": 0.5, "test/cumulative_reward": 35.093442424233196, "test/mean_score": 7.814641262991453, "test/success_rate": 0.18, "val_loss": 0.09514353424310684, "train_action_mse_error": 0.03304779529571533} +{"train_loss": 0.030303772538900375, "train_loss_bc": 0.030285917222499847, "train_loss_llm": 0.0017855465412139893, "grad_norm": 0.10112901777029037, "global_step": 28656, "epoch": 796, "lr": 0.001} +{"train_loss": 0.03114664927124977, "train_loss_bc": 0.03114664927124977, "train_loss_llm": 0.0, "grad_norm": 0.032478176057338715, "global_step": 28657, "epoch": 796, "lr": 0.001} +{"train_loss": 0.029955996200442314, "train_loss_bc": 0.029955996200442314, "train_loss_llm": 0.0, "grad_norm": 0.04113028198480606, "global_step": 28658, "epoch": 796, "lr": 0.001} +{"train_loss": 0.02794867940247059, "train_loss_bc": 0.02794867940247059, "train_loss_llm": 0.0, "grad_norm": 0.05038853734731674, "global_step": 28659, "epoch": 796, "lr": 0.001} +{"train_loss": 0.03634270653128624, "train_loss_bc": 0.03634270653128624, "train_loss_llm": 0.0, "grad_norm": 0.06828927248716354, "global_step": 28660, "epoch": 796, "lr": 0.001} +{"train_loss": 0.026687832549214363, "train_loss_bc": 0.026687832549214363, "train_loss_llm": 0.0, "grad_norm": 0.03199085593223572, "global_step": 28661, "epoch": 796, "lr": 0.001} +{"train_loss": 0.02941405214369297, "train_loss_bc": 0.02941405214369297, "train_loss_llm": 0.0, "grad_norm": 0.051757123321294785, "global_step": 28662, "epoch": 796, "lr": 0.001} +{"train_loss": 0.02953450381755829, "train_loss_bc": 0.02953450381755829, "train_loss_llm": 0.0, "grad_norm": 0.06069698557257652, "global_step": 28663, "epoch": 796, "lr": 0.001} +{"train_loss": 0.026500865817070007, "train_loss_bc": 0.026500865817070007, "train_loss_llm": 0.0, "grad_norm": 0.07775150239467621, "global_step": 28664, "epoch": 796, "lr": 0.001} +{"train_loss": 0.02763865888118744, "train_loss_bc": 0.02763865888118744, "train_loss_llm": 0.0, "grad_norm": 0.036173056811094284, "global_step": 28665, "epoch": 796, "lr": 0.001} +{"train_loss": 0.030274292454123497, "train_loss_bc": 0.030263185501098633, "train_loss_llm": 0.001110672950744629, "grad_norm": 0.05577210336923599, "global_step": 28666, "epoch": 796, "lr": 0.001} +{"train_loss": 0.024815134704113007, "train_loss_bc": 0.024803511798381805, "train_loss_llm": 0.0011623203754425049, "grad_norm": 0.07254146039485931, "global_step": 28667, "epoch": 796, "lr": 0.001} +{"train_loss": 0.031138934195041656, "train_loss_bc": 0.031138934195041656, "train_loss_llm": 0.0, "grad_norm": 0.08571454137563705, "global_step": 28668, "epoch": 796, "lr": 0.001} +{"train_loss": 0.02959449402987957, "train_loss_bc": 0.02959449402987957, "train_loss_llm": 0.0, "grad_norm": 0.03536532446742058, "global_step": 28669, "epoch": 796, "lr": 0.001} +{"train_loss": 0.029625538736581802, "train_loss_bc": 0.029623473063111305, "train_loss_llm": 0.0002065598964691162, "grad_norm": 0.06047090142965317, "global_step": 28670, "epoch": 796, "lr": 0.001} +{"train_loss": 0.033943645656108856, "train_loss_bc": 0.033943645656108856, "train_loss_llm": 0.0, "grad_norm": 0.09695836901664734, "global_step": 28671, "epoch": 796, "lr": 0.001} +{"train_loss": 0.030150054022669792, "train_loss_bc": 0.03014342486858368, "train_loss_llm": 0.0006629228591918945, "grad_norm": 0.1238127201795578, "global_step": 28672, "epoch": 796, "lr": 0.001} +{"train_loss": 0.030449654906988144, "train_loss_bc": 0.03044651262462139, "train_loss_llm": 0.0003142356872558594, "grad_norm": 0.039079781621694565, "global_step": 28673, "epoch": 796, "lr": 0.001} +{"train_loss": 0.033184148371219635, "train_loss_bc": 0.033184148371219635, "train_loss_llm": 0.0, "grad_norm": 0.06806688755750656, "global_step": 28674, "epoch": 796, "lr": 0.001} +{"train_loss": 0.029737379401922226, "train_loss_bc": 0.029737379401922226, "train_loss_llm": 0.0, "grad_norm": 0.09787940979003906, "global_step": 28675, "epoch": 796, "lr": 0.001} +{"train_loss": 0.03546978160738945, "train_loss_bc": 0.03546978160738945, "train_loss_llm": 0.0, "grad_norm": 0.11844322830438614, "global_step": 28676, "epoch": 796, "lr": 0.001} +{"train_loss": 0.02835182473063469, "train_loss_bc": 0.028332432731986046, "train_loss_llm": 0.001939147710800171, "grad_norm": 0.028763923794031143, "global_step": 28677, "epoch": 796, "lr": 0.001} +{"train_loss": 0.03043077513575554, "train_loss_bc": 0.03043077513575554, "train_loss_llm": 0.0, "grad_norm": 0.04937367886304855, "global_step": 28678, "epoch": 796, "lr": 0.001} +{"train_loss": 0.029504982754588127, "train_loss_bc": 0.029504982754588127, "train_loss_llm": 0.0, "grad_norm": 0.06896160542964935, "global_step": 28679, "epoch": 796, "lr": 0.001} +{"train_loss": 0.025403207167983055, "train_loss_bc": 0.025399329140782356, "train_loss_llm": 0.000387728214263916, "grad_norm": 0.07675383985042572, "global_step": 28680, "epoch": 796, "lr": 0.001} +{"train_loss": 0.03378013148903847, "train_loss_bc": 0.03378013148903847, "train_loss_llm": 0.0, "grad_norm": 0.03274080902338028, "global_step": 28681, "epoch": 796, "lr": 0.001} +{"train_loss": 0.02789856307208538, "train_loss_bc": 0.02789856307208538, "train_loss_llm": 0.0, "grad_norm": 0.048041827976703644, "global_step": 28682, "epoch": 796, "lr": 0.001} +{"train_loss": 0.03235340490937233, "train_loss_bc": 0.03233303874731064, "train_loss_llm": 0.0020365118980407715, "grad_norm": 0.06092435494065285, "global_step": 28683, "epoch": 796, "lr": 0.001} +{"train_loss": 0.03327315300703049, "train_loss_bc": 0.03327315300703049, "train_loss_llm": 0.0, "grad_norm": 0.08013713359832764, "global_step": 28684, "epoch": 796, "lr": 0.001} +{"train_loss": 0.02356879971921444, "train_loss_bc": 0.02355128899216652, "train_loss_llm": 0.0017511248588562012, "grad_norm": 0.02854408696293831, "global_step": 28685, "epoch": 796, "lr": 0.001} +{"train_loss": 0.03427128121256828, "train_loss_bc": 0.03427128121256828, "train_loss_llm": 0.0, "grad_norm": 0.06820723414421082, "global_step": 28686, "epoch": 796, "lr": 0.001} +{"train_loss": 0.027517439797520638, "train_loss_bc": 0.027517439797520638, "train_loss_llm": 0.0, "grad_norm": 0.11346578598022461, "global_step": 28687, "epoch": 796, "lr": 0.001} +{"train_loss": 0.030205056071281433, "train_loss_bc": 0.030205056071281433, "train_loss_llm": 0.0, "grad_norm": 0.14836527407169342, "global_step": 28688, "epoch": 796, "lr": 0.001} +{"train_loss": 0.03628033772110939, "train_loss_bc": 0.03627795726060867, "train_loss_llm": 0.00023788213729858398, "grad_norm": 0.039906859397888184, "global_step": 28689, "epoch": 796, "lr": 0.001} +{"train_loss": 0.03430422395467758, "train_loss_bc": 0.03430422395467758, "train_loss_llm": 0.0, "grad_norm": 0.06348360329866409, "global_step": 28690, "epoch": 796, "lr": 0.001} +{"train_loss": 0.030389596708118916, "train_loss_bc": 0.03302552551031113, "train_loss_llm": 0.0, "grad_norm": 0.09689673036336899, "global_step": 28691, "epoch": 796, "lr": 0.001, "val_loss": 0.09571618586778641} +{"train_loss": 0.03316029533743858, "train_loss_bc": 0.03316029533743858, "train_loss_llm": 0.0, "grad_norm": 0.12434417754411697, "global_step": 28692, "epoch": 797, "lr": 0.001} +{"train_loss": 0.02864600531756878, "train_loss_bc": 0.02864600531756878, "train_loss_llm": 0.0, "grad_norm": 0.03587201610207558, "global_step": 28693, "epoch": 797, "lr": 0.001} +{"train_loss": 0.0312647745013237, "train_loss_bc": 0.03126351535320282, "train_loss_llm": 0.00012603402137756348, "grad_norm": 0.05949319526553154, "global_step": 28694, "epoch": 797, "lr": 0.001} +{"train_loss": 0.029159491881728172, "train_loss_bc": 0.029159491881728172, "train_loss_llm": 0.0, "grad_norm": 0.08509168028831482, "global_step": 28695, "epoch": 797, "lr": 0.001} +{"train_loss": 0.02932192198932171, "train_loss_bc": 0.02932192198932171, "train_loss_llm": 0.0, "grad_norm": 0.10205187648534775, "global_step": 28696, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03249939903616905, "train_loss_bc": 0.03249286860227585, "train_loss_llm": 0.0006531476974487305, "grad_norm": 0.027187645435333252, "global_step": 28697, "epoch": 797, "lr": 0.001} +{"train_loss": 0.032999955117702484, "train_loss_bc": 0.032995808869600296, "train_loss_llm": 0.0004146695137023926, "grad_norm": 0.056615766137838364, "global_step": 28698, "epoch": 797, "lr": 0.001} +{"train_loss": 0.027903035283088684, "train_loss_bc": 0.027903035283088684, "train_loss_llm": 0.0, "grad_norm": 0.06080654636025429, "global_step": 28699, "epoch": 797, "lr": 0.001} +{"train_loss": 0.02584615722298622, "train_loss_bc": 0.025844279676675797, "train_loss_llm": 0.00018784403800964355, "grad_norm": 0.07838250696659088, "global_step": 28700, "epoch": 797, "lr": 0.001} +{"train_loss": 0.02770579420030117, "train_loss_bc": 0.027697285637259483, "train_loss_llm": 0.0008509159088134766, "grad_norm": 0.035744134336709976, "global_step": 28701, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03340444713830948, "train_loss_bc": 0.03339913487434387, "train_loss_llm": 0.0005310773849487305, "grad_norm": 0.06769981980323792, "global_step": 28702, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03876926004886627, "train_loss_bc": 0.03876926004886627, "train_loss_llm": 0.0, "grad_norm": 0.09196478873491287, "global_step": 28703, "epoch": 797, "lr": 0.001} +{"train_loss": 0.02495403401553631, "train_loss_bc": 0.02495403401553631, "train_loss_llm": 0.0, "grad_norm": 0.10669088363647461, "global_step": 28704, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03352252393960953, "train_loss_bc": 0.03352252393960953, "train_loss_llm": 0.0, "grad_norm": 0.03802118077874184, "global_step": 28705, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03275574743747711, "train_loss_bc": 0.032743070274591446, "train_loss_llm": 0.0012678802013397217, "grad_norm": 0.05319778621196747, "global_step": 28706, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03129959851503372, "train_loss_bc": 0.03129959851503372, "train_loss_llm": 0.0, "grad_norm": 0.07515694946050644, "global_step": 28707, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03437988832592964, "train_loss_bc": 0.03437988832592964, "train_loss_llm": 0.0, "grad_norm": 0.10082376003265381, "global_step": 28708, "epoch": 797, "lr": 0.001} +{"train_loss": 0.028072107583284378, "train_loss_bc": 0.028042349964380264, "train_loss_llm": 0.002975761890411377, "grad_norm": 0.037211984395980835, "global_step": 28709, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03856567665934563, "train_loss_bc": 0.038558412343263626, "train_loss_llm": 0.0007262825965881348, "grad_norm": 0.07499422878026962, "global_step": 28710, "epoch": 797, "lr": 0.001} +{"train_loss": 0.028568897396326065, "train_loss_bc": 0.028568897396326065, "train_loss_llm": 0.0, "grad_norm": 0.10252497345209122, "global_step": 28711, "epoch": 797, "lr": 0.001} +{"train_loss": 0.034617599099874496, "train_loss_bc": 0.034617599099874496, "train_loss_llm": 0.0, "grad_norm": 0.12710051238536835, "global_step": 28712, "epoch": 797, "lr": 0.001} +{"train_loss": 0.028793416917324066, "train_loss_bc": 0.028773833066225052, "train_loss_llm": 0.0019584596157073975, "grad_norm": 0.03538701683282852, "global_step": 28713, "epoch": 797, "lr": 0.001} +{"train_loss": 0.02277030050754547, "train_loss_bc": 0.0227659922093153, "train_loss_llm": 0.00043082237243652344, "grad_norm": 0.05669911578297615, "global_step": 28714, "epoch": 797, "lr": 0.001} +{"train_loss": 0.030793139711022377, "train_loss_bc": 0.030793139711022377, "train_loss_llm": 0.0, "grad_norm": 0.07964371889829636, "global_step": 28715, "epoch": 797, "lr": 0.001} +{"train_loss": 0.02923944965004921, "train_loss_bc": 0.029228007420897484, "train_loss_llm": 0.0011442601680755615, "grad_norm": 0.10269172489643097, "global_step": 28716, "epoch": 797, "lr": 0.001} +{"train_loss": 0.027313191443681717, "train_loss_bc": 0.027310650795698166, "train_loss_llm": 0.00025406479835510254, "grad_norm": 0.03233722224831581, "global_step": 28717, "epoch": 797, "lr": 0.001} +{"train_loss": 0.02922951988875866, "train_loss_bc": 0.02922951988875866, "train_loss_llm": 0.0, "grad_norm": 0.05460144579410553, "global_step": 28718, "epoch": 797, "lr": 0.001} +{"train_loss": 0.032316904515028, "train_loss_bc": 0.032316904515028, "train_loss_llm": 0.0, "grad_norm": 0.08465095609426498, "global_step": 28719, "epoch": 797, "lr": 0.001} +{"train_loss": 0.026928912848234177, "train_loss_bc": 0.0269243735820055, "train_loss_llm": 0.0004538893699645996, "grad_norm": 0.0912463366985321, "global_step": 28720, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03108995407819748, "train_loss_bc": 0.03108995407819748, "train_loss_llm": 0.0, "grad_norm": 0.032585784792900085, "global_step": 28721, "epoch": 797, "lr": 0.001} +{"train_loss": 0.02865702100098133, "train_loss_bc": 0.028650769963860512, "train_loss_llm": 0.0006250441074371338, "grad_norm": 0.048388414084911346, "global_step": 28722, "epoch": 797, "lr": 0.001} +{"train_loss": 0.028393354266881943, "train_loss_bc": 0.028393354266881943, "train_loss_llm": 0.0, "grad_norm": 0.06625329703092575, "global_step": 28723, "epoch": 797, "lr": 0.001} +{"train_loss": 0.027365880087018013, "train_loss_bc": 0.027365880087018013, "train_loss_llm": 0.0, "grad_norm": 0.08300310373306274, "global_step": 28724, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03412869945168495, "train_loss_bc": 0.03412869945168495, "train_loss_llm": 0.0, "grad_norm": 0.043947719037532806, "global_step": 28725, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03336533159017563, "train_loss_bc": 0.03336533159017563, "train_loss_llm": 0.0, "grad_norm": 0.07668871432542801, "global_step": 28726, "epoch": 797, "lr": 0.001} +{"train_loss": 0.03041359342427717, "train_loss_bc": 0.027087677270174026, "train_loss_llm": 0.0, "grad_norm": 0.09300907701253891, "global_step": 28727, "epoch": 797, "lr": 0.001, "val_loss": 0.09723880141973495} +{"train_loss": 0.031305018812417984, "train_loss_bc": 0.031295470893383026, "train_loss_llm": 0.0009548664093017578, "grad_norm": 0.13331621885299683, "global_step": 28728, "epoch": 798, "lr": 0.001} +{"train_loss": 0.030567573383450508, "train_loss_bc": 0.030567573383450508, "train_loss_llm": 0.0, "grad_norm": 0.04422064870595932, "global_step": 28729, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03086063824594021, "train_loss_bc": 0.03086063824594021, "train_loss_llm": 0.0, "grad_norm": 0.06703367084264755, "global_step": 28730, "epoch": 798, "lr": 0.001} +{"train_loss": 0.030964845791459084, "train_loss_bc": 0.030964845791459084, "train_loss_llm": 0.0, "grad_norm": 0.08860857784748077, "global_step": 28731, "epoch": 798, "lr": 0.001} +{"train_loss": 0.02711348421871662, "train_loss_bc": 0.027105432003736496, "train_loss_llm": 0.0008052289485931396, "grad_norm": 0.10649213939905167, "global_step": 28732, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03040909767150879, "train_loss_bc": 0.03040909767150879, "train_loss_llm": 0.0, "grad_norm": 0.04226858913898468, "global_step": 28733, "epoch": 798, "lr": 0.001} +{"train_loss": 0.02515716850757599, "train_loss_bc": 0.02515716850757599, "train_loss_llm": 0.0, "grad_norm": 0.07328962534666061, "global_step": 28734, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03152753412723541, "train_loss_bc": 0.03152753412723541, "train_loss_llm": 0.0, "grad_norm": 0.10874233394861221, "global_step": 28735, "epoch": 798, "lr": 0.001} +{"train_loss": 0.0348052941262722, "train_loss_bc": 0.034803833812475204, "train_loss_llm": 0.00014588236808776855, "grad_norm": 0.14866530895233154, "global_step": 28736, "epoch": 798, "lr": 0.001} +{"train_loss": 0.030044803395867348, "train_loss_bc": 0.030044803395867348, "train_loss_llm": 0.0, "grad_norm": 0.04946305230259895, "global_step": 28737, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03314579650759697, "train_loss_bc": 0.03314579650759697, "train_loss_llm": 0.0, "grad_norm": 0.07616311311721802, "global_step": 28738, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03147659823298454, "train_loss_bc": 0.03147659823298454, "train_loss_llm": 0.0, "grad_norm": 0.0807948112487793, "global_step": 28739, "epoch": 798, "lr": 0.001} +{"train_loss": 0.027068747207522392, "train_loss_bc": 0.027068747207522392, "train_loss_llm": 0.0, "grad_norm": 0.08492832630872726, "global_step": 28740, "epoch": 798, "lr": 0.001} +{"train_loss": 0.028862999752163887, "train_loss_bc": 0.028857210651040077, "train_loss_llm": 0.0005789995193481445, "grad_norm": 0.033365439623594284, "global_step": 28741, "epoch": 798, "lr": 0.001} +{"train_loss": 0.030703045427799225, "train_loss_bc": 0.030703045427799225, "train_loss_llm": 0.0, "grad_norm": 0.053274646401405334, "global_step": 28742, "epoch": 798, "lr": 0.001} +{"train_loss": 0.034066468477249146, "train_loss_bc": 0.034066468477249146, "train_loss_llm": 0.0, "grad_norm": 0.0689760148525238, "global_step": 28743, "epoch": 798, "lr": 0.001} +{"train_loss": 0.030780363827943802, "train_loss_bc": 0.030780363827943802, "train_loss_llm": 0.0, "grad_norm": 0.08201459050178528, "global_step": 28744, "epoch": 798, "lr": 0.001} +{"train_loss": 0.031652823090553284, "train_loss_bc": 0.031652823090553284, "train_loss_llm": 0.0, "grad_norm": 0.0333428792655468, "global_step": 28745, "epoch": 798, "lr": 0.001} +{"train_loss": 0.031666237860918045, "train_loss_bc": 0.031666237860918045, "train_loss_llm": 0.0, "grad_norm": 0.060710467398166656, "global_step": 28746, "epoch": 798, "lr": 0.001} +{"train_loss": 0.030311888083815575, "train_loss_bc": 0.030310310423374176, "train_loss_llm": 0.00015780329704284668, "grad_norm": 0.07408303767442703, "global_step": 28747, "epoch": 798, "lr": 0.001} +{"train_loss": 0.02557995729148388, "train_loss_bc": 0.025575993582606316, "train_loss_llm": 0.00039634108543395996, "grad_norm": 0.08655591309070587, "global_step": 28748, "epoch": 798, "lr": 0.001} +{"train_loss": 0.028379516676068306, "train_loss_bc": 0.028379516676068306, "train_loss_llm": 0.0, "grad_norm": 0.03353829309344292, "global_step": 28749, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03017280623316765, "train_loss_bc": 0.03017280623316765, "train_loss_llm": 0.0, "grad_norm": 0.05739487707614899, "global_step": 28750, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03652657940983772, "train_loss_bc": 0.03652657940983772, "train_loss_llm": 0.0, "grad_norm": 0.08839002996683121, "global_step": 28751, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03453753888607025, "train_loss_bc": 0.03453753888607025, "train_loss_llm": 0.0, "grad_norm": 0.10078456997871399, "global_step": 28752, "epoch": 798, "lr": 0.001} +{"train_loss": 0.025884196162223816, "train_loss_bc": 0.025884196162223816, "train_loss_llm": 0.0, "grad_norm": 0.036137185990810394, "global_step": 28753, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03091801516711712, "train_loss_bc": 0.030880311504006386, "train_loss_llm": 0.0037704408168792725, "grad_norm": 0.05976669862866402, "global_step": 28754, "epoch": 798, "lr": 0.001} +{"train_loss": 0.030965667217969894, "train_loss_bc": 0.030965667217969894, "train_loss_llm": 0.0, "grad_norm": 0.08222701400518417, "global_step": 28755, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03430155664682388, "train_loss_bc": 0.03429514914751053, "train_loss_llm": 0.000640571117401123, "grad_norm": 0.10745545476675034, "global_step": 28756, "epoch": 798, "lr": 0.001} +{"train_loss": 0.02999074198305607, "train_loss_bc": 0.02997269481420517, "train_loss_llm": 0.0018047094345092773, "grad_norm": 0.0288421381264925, "global_step": 28757, "epoch": 798, "lr": 0.001} +{"train_loss": 0.030000539496541023, "train_loss_bc": 0.030000539496541023, "train_loss_llm": 0.0, "grad_norm": 0.05216585099697113, "global_step": 28758, "epoch": 798, "lr": 0.001} +{"train_loss": 0.02851426787674427, "train_loss_bc": 0.02851426787674427, "train_loss_llm": 0.0, "grad_norm": 0.06599926948547363, "global_step": 28759, "epoch": 798, "lr": 0.001} +{"train_loss": 0.030472751706838608, "train_loss_bc": 0.030472751706838608, "train_loss_llm": 0.0, "grad_norm": 0.0805833712220192, "global_step": 28760, "epoch": 798, "lr": 0.001} +{"train_loss": 0.027322299778461456, "train_loss_bc": 0.027322299778461456, "train_loss_llm": 0.0, "grad_norm": 0.03253685683012009, "global_step": 28761, "epoch": 798, "lr": 0.001} +{"train_loss": 0.03142885863780975, "train_loss_bc": 0.03142017871141434, "train_loss_llm": 0.0008679330348968506, "grad_norm": 0.056533459573984146, "global_step": 28762, "epoch": 798, "lr": 0.001} +{"train_loss": 0.030376704616679087, "train_loss_bc": 0.02607564628124237, "train_loss_llm": 0.0, "grad_norm": 0.07101687043905258, "global_step": 28763, "epoch": 798, "lr": 0.001, "val_loss": 0.09613369405269623} +{"train_loss": 0.031077498570084572, "train_loss_bc": 0.031077498570084572, "train_loss_llm": 0.0, "grad_norm": 0.08411174267530441, "global_step": 28764, "epoch": 799, "lr": 0.001} +{"train_loss": 0.02649807371199131, "train_loss_bc": 0.02647349238395691, "train_loss_llm": 0.0024580955505371094, "grad_norm": 0.03156628832221031, "global_step": 28765, "epoch": 799, "lr": 0.001} +{"train_loss": 0.029056180268526077, "train_loss_bc": 0.029056180268526077, "train_loss_llm": 0.0, "grad_norm": 0.04596972465515137, "global_step": 28766, "epoch": 799, "lr": 0.001} +{"train_loss": 0.030959704890847206, "train_loss_bc": 0.030959704890847206, "train_loss_llm": 0.0, "grad_norm": 0.0626024454832077, "global_step": 28767, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03768845647573471, "train_loss_bc": 0.03768845647573471, "train_loss_llm": 0.0, "grad_norm": 0.07974737137556076, "global_step": 28768, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03218589723110199, "train_loss_bc": 0.032185155898332596, "train_loss_llm": 7.408857345581055e-05, "grad_norm": 0.03302065283060074, "global_step": 28769, "epoch": 799, "lr": 0.001} +{"train_loss": 0.031168341636657715, "train_loss_bc": 0.031168341636657715, "train_loss_llm": 0.0, "grad_norm": 0.05601406842470169, "global_step": 28770, "epoch": 799, "lr": 0.001} +{"train_loss": 0.028167972341179848, "train_loss_bc": 0.028167972341179848, "train_loss_llm": 0.0, "grad_norm": 0.08038534224033356, "global_step": 28771, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03312583267688751, "train_loss_bc": 0.03312583267688751, "train_loss_llm": 0.0, "grad_norm": 0.097419872879982, "global_step": 28772, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03386693820357323, "train_loss_bc": 0.03386693820357323, "train_loss_llm": 0.0, "grad_norm": 0.04450157657265663, "global_step": 28773, "epoch": 799, "lr": 0.001} +{"train_loss": 0.02958790771663189, "train_loss_bc": 0.02958790771663189, "train_loss_llm": 0.0, "grad_norm": 0.06628328561782837, "global_step": 28774, "epoch": 799, "lr": 0.001} +{"train_loss": 0.02569815143942833, "train_loss_bc": 0.02569815143942833, "train_loss_llm": 0.0, "grad_norm": 0.08519555628299713, "global_step": 28775, "epoch": 799, "lr": 0.001} +{"train_loss": 0.029351403936743736, "train_loss_bc": 0.029351403936743736, "train_loss_llm": 0.0, "grad_norm": 0.11402910202741623, "global_step": 28776, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03045588731765747, "train_loss_bc": 0.03045588731765747, "train_loss_llm": 0.0, "grad_norm": 0.039861392229795456, "global_step": 28777, "epoch": 799, "lr": 0.001} +{"train_loss": 0.034868426620960236, "train_loss_bc": 0.034868426620960236, "train_loss_llm": 0.0, "grad_norm": 0.06411547213792801, "global_step": 28778, "epoch": 799, "lr": 0.001} +{"train_loss": 0.0366322360932827, "train_loss_bc": 0.0366322360932827, "train_loss_llm": 0.0, "grad_norm": 0.08731506019830704, "global_step": 28779, "epoch": 799, "lr": 0.001} +{"train_loss": 0.028190894052386284, "train_loss_bc": 0.028190894052386284, "train_loss_llm": 0.0, "grad_norm": 0.11003243178129196, "global_step": 28780, "epoch": 799, "lr": 0.001} +{"train_loss": 0.028142403811216354, "train_loss_bc": 0.028142403811216354, "train_loss_llm": 0.0, "grad_norm": 0.034144170582294464, "global_step": 28781, "epoch": 799, "lr": 0.001} +{"train_loss": 0.024736210703849792, "train_loss_bc": 0.02473418042063713, "train_loss_llm": 0.00020298361778259277, "grad_norm": 0.048222847282886505, "global_step": 28782, "epoch": 799, "lr": 0.001} +{"train_loss": 0.031785063445568085, "train_loss_bc": 0.031785063445568085, "train_loss_llm": 0.0, "grad_norm": 0.06463640928268433, "global_step": 28783, "epoch": 799, "lr": 0.001} +{"train_loss": 0.030998798087239265, "train_loss_bc": 0.030998798087239265, "train_loss_llm": 0.0, "grad_norm": 0.08175968378782272, "global_step": 28784, "epoch": 799, "lr": 0.001} +{"train_loss": 0.02357162907719612, "train_loss_bc": 0.023568706586956978, "train_loss_llm": 0.0002923309803009033, "grad_norm": 0.028421679511666298, "global_step": 28785, "epoch": 799, "lr": 0.001} +{"train_loss": 0.028839046135544777, "train_loss_bc": 0.028839046135544777, "train_loss_llm": 0.0, "grad_norm": 0.04473893716931343, "global_step": 28786, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03576621413230896, "train_loss_bc": 0.03576621413230896, "train_loss_llm": 0.0, "grad_norm": 0.055983226746320724, "global_step": 28787, "epoch": 799, "lr": 0.001} +{"train_loss": 0.027348054572939873, "train_loss_bc": 0.027348054572939873, "train_loss_llm": 0.0, "grad_norm": 0.07511795312166214, "global_step": 28788, "epoch": 799, "lr": 0.001} +{"train_loss": 0.025623025372624397, "train_loss_bc": 0.025622861459851265, "train_loss_llm": 1.633167266845703e-05, "grad_norm": 0.02831888757646084, "global_step": 28789, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03061356395483017, "train_loss_bc": 0.03061356395483017, "train_loss_llm": 0.0, "grad_norm": 0.04294859245419502, "global_step": 28790, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03163741156458855, "train_loss_bc": 0.03163741156458855, "train_loss_llm": 0.0, "grad_norm": 0.05759917572140694, "global_step": 28791, "epoch": 799, "lr": 0.001} +{"train_loss": 0.024026066064834595, "train_loss_bc": 0.02400960959494114, "train_loss_llm": 0.0016456246376037598, "grad_norm": 0.06904356926679611, "global_step": 28792, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03848464414477348, "train_loss_bc": 0.03848464414477348, "train_loss_llm": 0.0, "grad_norm": 0.04263380914926529, "global_step": 28793, "epoch": 799, "lr": 0.001} +{"train_loss": 0.0292271189391613, "train_loss_bc": 0.0292271189391613, "train_loss_llm": 0.0, "grad_norm": 0.0644470825791359, "global_step": 28794, "epoch": 799, "lr": 0.001} +{"train_loss": 0.029697826132178307, "train_loss_bc": 0.02967674843966961, "train_loss_llm": 0.0021078288555145264, "grad_norm": 0.08016590774059296, "global_step": 28795, "epoch": 799, "lr": 0.001} +{"train_loss": 0.025837279856204987, "train_loss_bc": 0.025837279856204987, "train_loss_llm": 0.0, "grad_norm": 0.11010144650936127, "global_step": 28796, "epoch": 799, "lr": 0.001} +{"train_loss": 0.030076580122113228, "train_loss_bc": 0.030056899413466454, "train_loss_llm": 0.0019681453704833984, "grad_norm": 0.037804003804922104, "global_step": 28797, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03106102906167507, "train_loss_bc": 0.03105454333126545, "train_loss_llm": 0.0006486475467681885, "grad_norm": 0.05973401665687561, "global_step": 28798, "epoch": 799, "lr": 0.001} +{"train_loss": 0.03038214985281229, "train_loss_bc": 0.03768528252840042, "train_loss_llm": 0.0020344555377960205, "grad_norm": 0.09466750174760818, "global_step": 28799, "epoch": 799, "lr": 0.001, "val_loss": 0.09550921618938446} +{"train_loss": 0.030795414000749588, "train_loss_bc": 0.030786560848355293, "train_loss_llm": 0.0008853673934936523, "grad_norm": 0.11768737435340881, "global_step": 28800, "epoch": 800, "lr": 0.001} +{"train_loss": 0.036996714770793915, "train_loss_bc": 0.036996714770793915, "train_loss_llm": 0.0, "grad_norm": 0.03535865992307663, "global_step": 28801, "epoch": 800, "lr": 0.001} +{"train_loss": 0.0324927493929863, "train_loss_bc": 0.0324927493929863, "train_loss_llm": 0.0, "grad_norm": 0.052907396107912064, "global_step": 28802, "epoch": 800, "lr": 0.001} +{"train_loss": 0.02154751494526863, "train_loss_bc": 0.02154751494526863, "train_loss_llm": 0.0, "grad_norm": 0.055047258734703064, "global_step": 28803, "epoch": 800, "lr": 0.001} +{"train_loss": 0.031344491988420486, "train_loss_bc": 0.03133789077401161, "train_loss_llm": 0.0006602108478546143, "grad_norm": 0.06612595915794373, "global_step": 28804, "epoch": 800, "lr": 0.001} +{"train_loss": 0.025272462517023087, "train_loss_bc": 0.025259174406528473, "train_loss_llm": 0.0013288557529449463, "grad_norm": 0.03009728528559208, "global_step": 28805, "epoch": 800, "lr": 0.001} +{"train_loss": 0.028289005160331726, "train_loss_bc": 0.028289005160331726, "train_loss_llm": 0.0, "grad_norm": 0.04300611838698387, "global_step": 28806, "epoch": 800, "lr": 0.001} +{"train_loss": 0.02580600045621395, "train_loss_bc": 0.025796517729759216, "train_loss_llm": 0.0009483098983764648, "grad_norm": 0.06467831879854202, "global_step": 28807, "epoch": 800, "lr": 0.001} +{"train_loss": 0.026853486895561218, "train_loss_bc": 0.026853486895561218, "train_loss_llm": 0.0, "grad_norm": 0.07945649325847626, "global_step": 28808, "epoch": 800, "lr": 0.001} +{"train_loss": 0.031188055872917175, "train_loss_bc": 0.031185485422611237, "train_loss_llm": 0.00025707483291625977, "grad_norm": 0.037126101553440094, "global_step": 28809, "epoch": 800, "lr": 0.001} +{"train_loss": 0.031824544072151184, "train_loss_bc": 0.031824544072151184, "train_loss_llm": 0.0, "grad_norm": 0.05272799730300903, "global_step": 28810, "epoch": 800, "lr": 0.001} +{"train_loss": 0.03380162641406059, "train_loss_bc": 0.03380162641406059, "train_loss_llm": 0.0, "grad_norm": 0.07180267572402954, "global_step": 28811, "epoch": 800, "lr": 0.001} +{"train_loss": 0.027728689834475517, "train_loss_bc": 0.027710510417819023, "train_loss_llm": 0.0018179118633270264, "grad_norm": 0.08889301866292953, "global_step": 28812, "epoch": 800, "lr": 0.001} +{"train_loss": 0.029280364513397217, "train_loss_bc": 0.029280364513397217, "train_loss_llm": 0.0, "grad_norm": 0.0362614169716835, "global_step": 28813, "epoch": 800, "lr": 0.001} +{"train_loss": 0.027137862518429756, "train_loss_bc": 0.027137862518429756, "train_loss_llm": 0.0, "grad_norm": 0.05484491214156151, "global_step": 28814, "epoch": 800, "lr": 0.001} +{"train_loss": 0.025377541780471802, "train_loss_bc": 0.025377541780471802, "train_loss_llm": 0.0, "grad_norm": 0.07341615855693817, "global_step": 28815, "epoch": 800, "lr": 0.001} +{"train_loss": 0.028179537504911423, "train_loss_bc": 0.028179537504911423, "train_loss_llm": 0.0, "grad_norm": 0.08777287602424622, "global_step": 28816, "epoch": 800, "lr": 0.001} +{"train_loss": 0.030117148533463478, "train_loss_bc": 0.030117148533463478, "train_loss_llm": 0.0, "grad_norm": 0.03635294362902641, "global_step": 28817, "epoch": 800, "lr": 0.001} +{"train_loss": 0.029902784153819084, "train_loss_bc": 0.029902784153819084, "train_loss_llm": 0.0, "grad_norm": 0.06092865765094757, "global_step": 28818, "epoch": 800, "lr": 0.001} +{"train_loss": 0.027457421645522118, "train_loss_bc": 0.027457421645522118, "train_loss_llm": 0.0, "grad_norm": 0.07231573760509491, "global_step": 28819, "epoch": 800, "lr": 0.001} +{"train_loss": 0.03250424563884735, "train_loss_bc": 0.03250424563884735, "train_loss_llm": 0.0, "grad_norm": 0.09219532459974289, "global_step": 28820, "epoch": 800, "lr": 0.001} +{"train_loss": 0.03017466701567173, "train_loss_bc": 0.03017466701567173, "train_loss_llm": 0.0, "grad_norm": 0.029673870652914047, "global_step": 28821, "epoch": 800, "lr": 0.001} +{"train_loss": 0.033512670546770096, "train_loss_bc": 0.033512670546770096, "train_loss_llm": 0.0, "grad_norm": 0.05135979503393173, "global_step": 28822, "epoch": 800, "lr": 0.001} +{"train_loss": 0.03232935070991516, "train_loss_bc": 0.03232935070991516, "train_loss_llm": 0.0, "grad_norm": 0.05983437970280647, "global_step": 28823, "epoch": 800, "lr": 0.001} +{"train_loss": 0.027141181752085686, "train_loss_bc": 0.027141181752085686, "train_loss_llm": 0.0, "grad_norm": 0.07617450505495071, "global_step": 28824, "epoch": 800, "lr": 0.001} +{"train_loss": 0.024925146251916885, "train_loss_bc": 0.02492327243089676, "train_loss_llm": 0.00018736720085144043, "grad_norm": 0.025872480124235153, "global_step": 28825, "epoch": 800, "lr": 0.001} +{"train_loss": 0.03015214577317238, "train_loss_bc": 0.03015214577317238, "train_loss_llm": 0.0, "grad_norm": 0.050863441079854965, "global_step": 28826, "epoch": 800, "lr": 0.001} +{"train_loss": 0.030127165839076042, "train_loss_bc": 0.030127165839076042, "train_loss_llm": 0.0, "grad_norm": 0.06342432647943497, "global_step": 28827, "epoch": 800, "lr": 0.001} +{"train_loss": 0.031039569526910782, "train_loss_bc": 0.031039569526910782, "train_loss_llm": 0.0, "grad_norm": 0.08151597529649734, "global_step": 28828, "epoch": 800, "lr": 0.001} +{"train_loss": 0.03681304305791855, "train_loss_bc": 0.03680809959769249, "train_loss_llm": 0.0004943311214447021, "grad_norm": 0.05035369098186493, "global_step": 28829, "epoch": 800, "lr": 0.001} +{"train_loss": 0.030794601887464523, "train_loss_bc": 0.030794601887464523, "train_loss_llm": 0.0, "grad_norm": 0.06327138096094131, "global_step": 28830, "epoch": 800, "lr": 0.001} +{"train_loss": 0.030399587005376816, "train_loss_bc": 0.030399587005376816, "train_loss_llm": 0.0, "grad_norm": 0.07796363532543182, "global_step": 28831, "epoch": 800, "lr": 0.001} +{"train_loss": 0.031015191227197647, "train_loss_bc": 0.031015191227197647, "train_loss_llm": 0.0, "grad_norm": 0.09586609154939651, "global_step": 28832, "epoch": 800, "lr": 0.001} +{"train_loss": 0.03778858855366707, "train_loss_bc": 0.03778858855366707, "train_loss_llm": 0.0, "grad_norm": 0.0344441719353199, "global_step": 28833, "epoch": 800, "lr": 0.001} +{"train_loss": 0.03138577938079834, "train_loss_bc": 0.03138577938079834, "train_loss_llm": 0.0, "grad_norm": 0.05019373446702957, "global_step": 28834, "epoch": 800, "lr": 0.001} +{"train_loss": 0.030080649329142436, "train_loss_bc": 0.03140702471137047, "train_loss_llm": 0.0, "grad_norm": 0.0645688995718956, "global_step": 28835, "epoch": 800, "lr": 0.001, "train/cumulative_reward": 82.36894612928015, "train/mean_score": 13.82536197947571, "train/success_rate": 0.3, "test/cumulative_reward": 40.41183342957514, "test/mean_score": 8.410309968587834, "test/success_rate": 0.22, "val_loss": 0.09608162939548492, "train_action_mse_error": 0.032475490123033524} +{"train_loss": 0.02785746566951275, "train_loss_bc": 0.02785746566951275, "train_loss_llm": 0.0, "grad_norm": 0.0796770229935646, "global_step": 28836, "epoch": 801, "lr": 0.001} +{"train_loss": 0.032316844910383224, "train_loss_bc": 0.032316844910383224, "train_loss_llm": 0.0, "grad_norm": 0.03677720949053764, "global_step": 28837, "epoch": 801, "lr": 0.001} +{"train_loss": 0.030401287600398064, "train_loss_bc": 0.030401287600398064, "train_loss_llm": 0.0, "grad_norm": 0.055253732949495316, "global_step": 28838, "epoch": 801, "lr": 0.001} +{"train_loss": 0.028966054320335388, "train_loss_bc": 0.028966054320335388, "train_loss_llm": 0.0, "grad_norm": 0.07160748541355133, "global_step": 28839, "epoch": 801, "lr": 0.001} +{"train_loss": 0.02171418070793152, "train_loss_bc": 0.02171418070793152, "train_loss_llm": 0.0, "grad_norm": 0.08049260079860687, "global_step": 28840, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03302443400025368, "train_loss_bc": 0.03302443400025368, "train_loss_llm": 0.0, "grad_norm": 0.037778932601213455, "global_step": 28841, "epoch": 801, "lr": 0.001} +{"train_loss": 0.033157240599393845, "train_loss_bc": 0.03314344212412834, "train_loss_llm": 0.0013796985149383545, "grad_norm": 0.05691688507795334, "global_step": 28842, "epoch": 801, "lr": 0.001} +{"train_loss": 0.029949627816677094, "train_loss_bc": 0.029949627816677094, "train_loss_llm": 0.0, "grad_norm": 0.0799766257405281, "global_step": 28843, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03351825103163719, "train_loss_bc": 0.03351825103163719, "train_loss_llm": 0.0, "grad_norm": 0.10632066428661346, "global_step": 28844, "epoch": 801, "lr": 0.001} +{"train_loss": 0.026732850819826126, "train_loss_bc": 0.026732850819826126, "train_loss_llm": 0.0, "grad_norm": 0.027384454384446144, "global_step": 28845, "epoch": 801, "lr": 0.001} +{"train_loss": 0.027536798268556595, "train_loss_bc": 0.02753455564379692, "train_loss_llm": 0.0002243518829345703, "grad_norm": 0.04424603655934334, "global_step": 28846, "epoch": 801, "lr": 0.001} +{"train_loss": 0.024920519441366196, "train_loss_bc": 0.024911820888519287, "train_loss_llm": 0.0008697807788848877, "grad_norm": 0.050769489258527756, "global_step": 28847, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03437057137489319, "train_loss_bc": 0.03437057137489319, "train_loss_llm": 0.0, "grad_norm": 0.07025487720966339, "global_step": 28848, "epoch": 801, "lr": 0.001} +{"train_loss": 0.027176687493920326, "train_loss_bc": 0.027176687493920326, "train_loss_llm": 0.0, "grad_norm": 0.037547457963228226, "global_step": 28849, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03395289182662964, "train_loss_bc": 0.03395289182662964, "train_loss_llm": 0.0, "grad_norm": 0.060493580996990204, "global_step": 28850, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03181992471218109, "train_loss_bc": 0.03181115910410881, "train_loss_llm": 0.0008766651153564453, "grad_norm": 0.0980614572763443, "global_step": 28851, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03295712545514107, "train_loss_bc": 0.032953932881355286, "train_loss_llm": 0.0003191828727722168, "grad_norm": 0.11193732917308807, "global_step": 28852, "epoch": 801, "lr": 0.001} +{"train_loss": 0.032445039600133896, "train_loss_bc": 0.032445039600133896, "train_loss_llm": 0.0, "grad_norm": 0.04283501207828522, "global_step": 28853, "epoch": 801, "lr": 0.001} +{"train_loss": 0.02784481830894947, "train_loss_bc": 0.02784481830894947, "train_loss_llm": 0.0, "grad_norm": 0.06481727957725525, "global_step": 28854, "epoch": 801, "lr": 0.001} +{"train_loss": 0.032739654183387756, "train_loss_bc": 0.032739654183387756, "train_loss_llm": 0.0, "grad_norm": 0.06954751163721085, "global_step": 28855, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03149711340665817, "train_loss_bc": 0.03149711340665817, "train_loss_llm": 0.0, "grad_norm": 0.08678402006626129, "global_step": 28856, "epoch": 801, "lr": 0.001} +{"train_loss": 0.029047565534710884, "train_loss_bc": 0.029047565534710884, "train_loss_llm": 0.0, "grad_norm": 0.03140659257769585, "global_step": 28857, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03098784014582634, "train_loss_bc": 0.03098784014582634, "train_loss_llm": 0.0, "grad_norm": 0.05058877170085907, "global_step": 28858, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03156678006052971, "train_loss_bc": 0.03156678006052971, "train_loss_llm": 0.0, "grad_norm": 0.07502791285514832, "global_step": 28859, "epoch": 801, "lr": 0.001} +{"train_loss": 0.031558867543935776, "train_loss_bc": 0.031558867543935776, "train_loss_llm": 0.0, "grad_norm": 0.09727449715137482, "global_step": 28860, "epoch": 801, "lr": 0.001} +{"train_loss": 0.027395127341151237, "train_loss_bc": 0.027395127341151237, "train_loss_llm": 0.0, "grad_norm": 0.03608788922429085, "global_step": 28861, "epoch": 801, "lr": 0.001} +{"train_loss": 0.02952100895345211, "train_loss_bc": 0.02952100895345211, "train_loss_llm": 0.0, "grad_norm": 0.0493042916059494, "global_step": 28862, "epoch": 801, "lr": 0.001} +{"train_loss": 0.032263901084661484, "train_loss_bc": 0.032262690365314484, "train_loss_llm": 0.00012123584747314453, "grad_norm": 0.06409160792827606, "global_step": 28863, "epoch": 801, "lr": 0.001} +{"train_loss": 0.029798315837979317, "train_loss_bc": 0.02979416400194168, "train_loss_llm": 0.0004151761531829834, "grad_norm": 0.0858437567949295, "global_step": 28864, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03322013095021248, "train_loss_bc": 0.03322013095021248, "train_loss_llm": 0.0, "grad_norm": 0.03810612112283707, "global_step": 28865, "epoch": 801, "lr": 0.001} +{"train_loss": 0.027867989614605904, "train_loss_bc": 0.027867989614605904, "train_loss_llm": 0.0, "grad_norm": 0.05779340863227844, "global_step": 28866, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03259928151965141, "train_loss_bc": 0.03259928151965141, "train_loss_llm": 0.0, "grad_norm": 0.08355826139450073, "global_step": 28867, "epoch": 801, "lr": 0.001} +{"train_loss": 0.023383483290672302, "train_loss_bc": 0.023383483290672302, "train_loss_llm": 0.0, "grad_norm": 0.09149307012557983, "global_step": 28868, "epoch": 801, "lr": 0.001} +{"train_loss": 0.030627820640802383, "train_loss_bc": 0.030627820640802383, "train_loss_llm": 0.0, "grad_norm": 0.041054487228393555, "global_step": 28869, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03272562474012375, "train_loss_bc": 0.03272562474012375, "train_loss_llm": 0.0, "grad_norm": 0.06272947788238525, "global_step": 28870, "epoch": 801, "lr": 0.001} +{"train_loss": 0.03014953186114629, "train_loss_bc": 0.027920028194785118, "train_loss_llm": 0.0, "grad_norm": 0.08029323816299438, "global_step": 28871, "epoch": 801, "lr": 0.001, "val_loss": 0.09893171489238739} +{"train_loss": 0.03217606619000435, "train_loss_bc": 0.03217606619000435, "train_loss_llm": 0.0, "grad_norm": 0.09785369783639908, "global_step": 28872, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03384249284863472, "train_loss_bc": 0.03384249284863472, "train_loss_llm": 0.0, "grad_norm": 0.03795098140835762, "global_step": 28873, "epoch": 802, "lr": 0.001} +{"train_loss": 0.028616247698664665, "train_loss_bc": 0.02858749032020569, "train_loss_llm": 0.002875775098800659, "grad_norm": 0.05296110361814499, "global_step": 28874, "epoch": 802, "lr": 0.001} +{"train_loss": 0.030027160421013832, "train_loss_bc": 0.030027160421013832, "train_loss_llm": 0.0, "grad_norm": 0.07101359218358994, "global_step": 28875, "epoch": 802, "lr": 0.001} +{"train_loss": 0.0342562198638916, "train_loss_bc": 0.0342562198638916, "train_loss_llm": 0.0, "grad_norm": 0.08468884974718094, "global_step": 28876, "epoch": 802, "lr": 0.001} +{"train_loss": 0.033345453441143036, "train_loss_bc": 0.033345453441143036, "train_loss_llm": 0.0, "grad_norm": 0.03614579141139984, "global_step": 28877, "epoch": 802, "lr": 0.001} +{"train_loss": 0.02985803782939911, "train_loss_bc": 0.02985803782939911, "train_loss_llm": 0.0, "grad_norm": 0.051019102334976196, "global_step": 28878, "epoch": 802, "lr": 0.001} +{"train_loss": 0.026743818074464798, "train_loss_bc": 0.026743818074464798, "train_loss_llm": 0.0, "grad_norm": 0.058740634471178055, "global_step": 28879, "epoch": 802, "lr": 0.001} +{"train_loss": 0.02874823845922947, "train_loss_bc": 0.028720976784825325, "train_loss_llm": 0.0027261674404144287, "grad_norm": 0.07066842168569565, "global_step": 28880, "epoch": 802, "lr": 0.001} +{"train_loss": 0.02504475601017475, "train_loss_bc": 0.02504475601017475, "train_loss_llm": 0.0, "grad_norm": 0.02656712755560875, "global_step": 28881, "epoch": 802, "lr": 0.001} +{"train_loss": 0.028544452041387558, "train_loss_bc": 0.028544452041387558, "train_loss_llm": 0.0, "grad_norm": 0.06290710717439651, "global_step": 28882, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03176969662308693, "train_loss_bc": 0.031750574707984924, "train_loss_llm": 0.0019123554229736328, "grad_norm": 0.07586667686700821, "global_step": 28883, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03046226315200329, "train_loss_bc": 0.03046226315200329, "train_loss_llm": 0.0, "grad_norm": 0.08363914489746094, "global_step": 28884, "epoch": 802, "lr": 0.001} +{"train_loss": 0.02466120384633541, "train_loss_bc": 0.02466045320034027, "train_loss_llm": 7.50422477722168e-05, "grad_norm": 0.034671004861593246, "global_step": 28885, "epoch": 802, "lr": 0.001} +{"train_loss": 0.026161547750234604, "train_loss_bc": 0.0261496901512146, "train_loss_llm": 0.001185774803161621, "grad_norm": 0.05929861590266228, "global_step": 28886, "epoch": 802, "lr": 0.001} +{"train_loss": 0.028155814856290817, "train_loss_bc": 0.028155814856290817, "train_loss_llm": 0.0, "grad_norm": 0.07603519409894943, "global_step": 28887, "epoch": 802, "lr": 0.001} +{"train_loss": 0.0320734828710556, "train_loss_bc": 0.0320734828710556, "train_loss_llm": 0.0, "grad_norm": 0.08795159310102463, "global_step": 28888, "epoch": 802, "lr": 0.001} +{"train_loss": 0.02852555923163891, "train_loss_bc": 0.02852555923163891, "train_loss_llm": 0.0, "grad_norm": 0.026018546894192696, "global_step": 28889, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03222621604800224, "train_loss_bc": 0.03222621604800224, "train_loss_llm": 0.0, "grad_norm": 0.0468718558549881, "global_step": 28890, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03462880477309227, "train_loss_bc": 0.034625329077243805, "train_loss_llm": 0.0003477036952972412, "grad_norm": 0.0633627325296402, "global_step": 28891, "epoch": 802, "lr": 0.001} +{"train_loss": 0.028821803629398346, "train_loss_bc": 0.028821803629398346, "train_loss_llm": 0.0, "grad_norm": 0.08030001819133759, "global_step": 28892, "epoch": 802, "lr": 0.001} +{"train_loss": 0.02698652818799019, "train_loss_bc": 0.02698652818799019, "train_loss_llm": 0.0, "grad_norm": 0.028119603171944618, "global_step": 28893, "epoch": 802, "lr": 0.001} +{"train_loss": 0.034479402005672455, "train_loss_bc": 0.034479402005672455, "train_loss_llm": 0.0, "grad_norm": 0.06258574873209, "global_step": 28894, "epoch": 802, "lr": 0.001} +{"train_loss": 0.02419966086745262, "train_loss_bc": 0.024197131395339966, "train_loss_llm": 0.0002528727054595947, "grad_norm": 0.08857574313879013, "global_step": 28895, "epoch": 802, "lr": 0.001} +{"train_loss": 0.028424717485904694, "train_loss_bc": 0.028424717485904694, "train_loss_llm": 0.0, "grad_norm": 0.10222770273685455, "global_step": 28896, "epoch": 802, "lr": 0.001} +{"train_loss": 0.02877108007669449, "train_loss_bc": 0.02877108007669449, "train_loss_llm": 0.0, "grad_norm": 0.040171753615140915, "global_step": 28897, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03230298310518265, "train_loss_bc": 0.03230298310518265, "train_loss_llm": 0.0, "grad_norm": 0.06022508069872856, "global_step": 28898, "epoch": 802, "lr": 0.001} +{"train_loss": 0.02891811728477478, "train_loss_bc": 0.028913384303450584, "train_loss_llm": 0.00047326087951660156, "grad_norm": 0.07639114558696747, "global_step": 28899, "epoch": 802, "lr": 0.001} +{"train_loss": 0.02622622810304165, "train_loss_bc": 0.02622622810304165, "train_loss_llm": 0.0, "grad_norm": 0.09018395096063614, "global_step": 28900, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03407926857471466, "train_loss_bc": 0.034068841487169266, "train_loss_llm": 0.0010428428649902344, "grad_norm": 0.033207785338163376, "global_step": 28901, "epoch": 802, "lr": 0.001} +{"train_loss": 0.027819762006402016, "train_loss_bc": 0.027819762006402016, "train_loss_llm": 0.0, "grad_norm": 0.05479247495532036, "global_step": 28902, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03304637596011162, "train_loss_bc": 0.03304637596011162, "train_loss_llm": 0.0, "grad_norm": 0.07202962040901184, "global_step": 28903, "epoch": 802, "lr": 0.001} +{"train_loss": 0.030586998909711838, "train_loss_bc": 0.030586998909711838, "train_loss_llm": 0.0, "grad_norm": 0.09116031974554062, "global_step": 28904, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03764631971716881, "train_loss_bc": 0.03764631971716881, "train_loss_llm": 0.0, "grad_norm": 0.03648556023836136, "global_step": 28905, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03453056886792183, "train_loss_bc": 0.03453056886792183, "train_loss_llm": 0.0, "grad_norm": 0.06379463523626328, "global_step": 28906, "epoch": 802, "lr": 0.001} +{"train_loss": 0.03016193930266632, "train_loss_bc": 0.02912246808409691, "train_loss_llm": 0.0, "grad_norm": 0.08047772943973541, "global_step": 28907, "epoch": 802, "lr": 0.001, "val_loss": 0.09661656618118286} +{"train_loss": 0.03830603137612343, "train_loss_bc": 0.03830603137612343, "train_loss_llm": 0.0, "grad_norm": 0.10119754076004028, "global_step": 28908, "epoch": 803, "lr": 0.001} +{"train_loss": 0.02813304401934147, "train_loss_bc": 0.02813304401934147, "train_loss_llm": 0.0, "grad_norm": 0.02960054576396942, "global_step": 28909, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03046267107129097, "train_loss_bc": 0.030452901497483253, "train_loss_llm": 0.0009770095348358154, "grad_norm": 0.046587035059928894, "global_step": 28910, "epoch": 803, "lr": 0.001} +{"train_loss": 0.02966105379164219, "train_loss_bc": 0.02966105379164219, "train_loss_llm": 0.0, "grad_norm": 0.06342969834804535, "global_step": 28911, "epoch": 803, "lr": 0.001} +{"train_loss": 0.024456163868308067, "train_loss_bc": 0.024456163868308067, "train_loss_llm": 0.0, "grad_norm": 0.08885031193494797, "global_step": 28912, "epoch": 803, "lr": 0.001} +{"train_loss": 0.02985352650284767, "train_loss_bc": 0.02985352650284767, "train_loss_llm": 0.0, "grad_norm": 0.03866403177380562, "global_step": 28913, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03036053106188774, "train_loss_bc": 0.030350642278790474, "train_loss_llm": 0.0009888410568237305, "grad_norm": 0.05936242267489433, "global_step": 28914, "epoch": 803, "lr": 0.001} +{"train_loss": 0.02275441586971283, "train_loss_bc": 0.02275441586971283, "train_loss_llm": 0.0, "grad_norm": 0.08116794377565384, "global_step": 28915, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03465903922915459, "train_loss_bc": 0.03465903922915459, "train_loss_llm": 0.0, "grad_norm": 0.10867077857255936, "global_step": 28916, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03268178552389145, "train_loss_bc": 0.03268178552389145, "train_loss_llm": 0.0, "grad_norm": 0.044885795563459396, "global_step": 28917, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03008173778653145, "train_loss_bc": 0.03008173778653145, "train_loss_llm": 0.0, "grad_norm": 0.06571771204471588, "global_step": 28918, "epoch": 803, "lr": 0.001} +{"train_loss": 0.028448650613427162, "train_loss_bc": 0.028448650613427162, "train_loss_llm": 0.0, "grad_norm": 0.0812634825706482, "global_step": 28919, "epoch": 803, "lr": 0.001} +{"train_loss": 0.030914926901459694, "train_loss_bc": 0.030914926901459694, "train_loss_llm": 0.0, "grad_norm": 0.09753219038248062, "global_step": 28920, "epoch": 803, "lr": 0.001} +{"train_loss": 0.026352545246481895, "train_loss_bc": 0.026352545246481895, "train_loss_llm": 0.0, "grad_norm": 0.030807046219706535, "global_step": 28921, "epoch": 803, "lr": 0.001} +{"train_loss": 0.030305320397019386, "train_loss_bc": 0.030305320397019386, "train_loss_llm": 0.0, "grad_norm": 0.05046163871884346, "global_step": 28922, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03386978060007095, "train_loss_bc": 0.03386978060007095, "train_loss_llm": 0.0, "grad_norm": 0.07471834868192673, "global_step": 28923, "epoch": 803, "lr": 0.001} +{"train_loss": 0.027170825749635696, "train_loss_bc": 0.027170825749635696, "train_loss_llm": 0.0, "grad_norm": 0.09664952009916306, "global_step": 28924, "epoch": 803, "lr": 0.001} +{"train_loss": 0.02606913447380066, "train_loss_bc": 0.02606913447380066, "train_loss_llm": 0.0, "grad_norm": 0.0376104898750782, "global_step": 28925, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03136128932237625, "train_loss_bc": 0.03136128932237625, "train_loss_llm": 0.0, "grad_norm": 0.06874067336320877, "global_step": 28926, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03766828030347824, "train_loss_bc": 0.037656739354133606, "train_loss_llm": 0.0011541545391082764, "grad_norm": 0.08204354345798492, "global_step": 28927, "epoch": 803, "lr": 0.001} +{"train_loss": 0.022689903154969215, "train_loss_bc": 0.0226878859102726, "train_loss_llm": 0.00020176172256469727, "grad_norm": 0.09894521534442902, "global_step": 28928, "epoch": 803, "lr": 0.001} +{"train_loss": 0.029791666194796562, "train_loss_bc": 0.029771018773317337, "train_loss_llm": 0.0020647644996643066, "grad_norm": 0.039805807173252106, "global_step": 28929, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03257538750767708, "train_loss_bc": 0.03257538750767708, "train_loss_llm": 0.0, "grad_norm": 0.056636322289705276, "global_step": 28930, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03314296156167984, "train_loss_bc": 0.03314296156167984, "train_loss_llm": 0.0, "grad_norm": 0.08761397749185562, "global_step": 28931, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03068363666534424, "train_loss_bc": 0.030681222677230835, "train_loss_llm": 0.00024145841598510742, "grad_norm": 0.10446981340646744, "global_step": 28932, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03314932435750961, "train_loss_bc": 0.03314932435750961, "train_loss_llm": 0.0, "grad_norm": 0.03483455255627632, "global_step": 28933, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03178756311535835, "train_loss_bc": 0.03178756311535835, "train_loss_llm": 0.0, "grad_norm": 0.05795914679765701, "global_step": 28934, "epoch": 803, "lr": 0.001} +{"train_loss": 0.030409108847379684, "train_loss_bc": 0.030409108847379684, "train_loss_llm": 0.0, "grad_norm": 0.07476270943880081, "global_step": 28935, "epoch": 803, "lr": 0.001} +{"train_loss": 0.0322687029838562, "train_loss_bc": 0.03225058317184448, "train_loss_llm": 0.0018119513988494873, "grad_norm": 0.09487488865852356, "global_step": 28936, "epoch": 803, "lr": 0.001} +{"train_loss": 0.02802734635770321, "train_loss_bc": 0.02802734635770321, "train_loss_llm": 0.0, "grad_norm": 0.03978220000863075, "global_step": 28937, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03089383989572525, "train_loss_bc": 0.03089383989572525, "train_loss_llm": 0.0, "grad_norm": 0.06170065701007843, "global_step": 28938, "epoch": 803, "lr": 0.001} +{"train_loss": 0.02447284199297428, "train_loss_bc": 0.024472016841173172, "train_loss_llm": 8.243322372436523e-05, "grad_norm": 0.07346317917108536, "global_step": 28939, "epoch": 803, "lr": 0.001} +{"train_loss": 0.02737078256905079, "train_loss_bc": 0.02737078256905079, "train_loss_llm": 0.0, "grad_norm": 0.0952010303735733, "global_step": 28940, "epoch": 803, "lr": 0.001} +{"train_loss": 0.03181103616952896, "train_loss_bc": 0.03181103616952896, "train_loss_llm": 0.0, "grad_norm": 0.036089252680540085, "global_step": 28941, "epoch": 803, "lr": 0.001} +{"train_loss": 0.033834151923656464, "train_loss_bc": 0.033834151923656464, "train_loss_llm": 0.0, "grad_norm": 0.07476083934307098, "global_step": 28942, "epoch": 803, "lr": 0.001} +{"train_loss": 0.0304196464518706, "train_loss_bc": 0.038621801882982254, "train_loss_llm": 0.0006464719772338867, "grad_norm": 0.10794872045516968, "global_step": 28943, "epoch": 803, "lr": 0.001, "val_loss": 0.09564678370952606} +{"train_loss": 0.03293972462415695, "train_loss_bc": 0.03293972462415695, "train_loss_llm": 0.0, "grad_norm": 0.12100066244602203, "global_step": 28944, "epoch": 804, "lr": 0.001} +{"train_loss": 0.025393467396497726, "train_loss_bc": 0.025393467396497726, "train_loss_llm": 0.0, "grad_norm": 0.02997012622654438, "global_step": 28945, "epoch": 804, "lr": 0.001} +{"train_loss": 0.02868770621716976, "train_loss_bc": 0.028676912188529968, "train_loss_llm": 0.0010794401168823242, "grad_norm": 0.04525795951485634, "global_step": 28946, "epoch": 804, "lr": 0.001} +{"train_loss": 0.03050471469759941, "train_loss_bc": 0.030497388914227486, "train_loss_llm": 0.0007325410842895508, "grad_norm": 0.06809335201978683, "global_step": 28947, "epoch": 804, "lr": 0.001} +{"train_loss": 0.027789641171693802, "train_loss_bc": 0.027789641171693802, "train_loss_llm": 0.0, "grad_norm": 0.08654698729515076, "global_step": 28948, "epoch": 804, "lr": 0.001} +{"train_loss": 0.03333078324794769, "train_loss_bc": 0.03333078324794769, "train_loss_llm": 0.0, "grad_norm": 0.04678817838430405, "global_step": 28949, "epoch": 804, "lr": 0.001} +{"train_loss": 0.031363096088171005, "train_loss_bc": 0.031363096088171005, "train_loss_llm": 0.0, "grad_norm": 0.06640839576721191, "global_step": 28950, "epoch": 804, "lr": 0.001} +{"train_loss": 0.023026101291179657, "train_loss_bc": 0.023026101291179657, "train_loss_llm": 0.0, "grad_norm": 0.07837168127298355, "global_step": 28951, "epoch": 804, "lr": 0.001} +{"train_loss": 0.026427702978253365, "train_loss_bc": 0.026427702978253365, "train_loss_llm": 0.0, "grad_norm": 0.09550593048334122, "global_step": 28952, "epoch": 804, "lr": 0.001} +{"train_loss": 0.029759828001260757, "train_loss_bc": 0.029759828001260757, "train_loss_llm": 0.0, "grad_norm": 0.03579328581690788, "global_step": 28953, "epoch": 804, "lr": 0.001} +{"train_loss": 0.03347110003232956, "train_loss_bc": 0.03347110003232956, "train_loss_llm": 0.0, "grad_norm": 0.055942222476005554, "global_step": 28954, "epoch": 804, "lr": 0.001} +{"train_loss": 0.03053111396729946, "train_loss_bc": 0.03053111396729946, "train_loss_llm": 0.0, "grad_norm": 0.079243503510952, "global_step": 28955, "epoch": 804, "lr": 0.001} +{"train_loss": 0.03093714453279972, "train_loss_bc": 0.03093714453279972, "train_loss_llm": 0.0, "grad_norm": 0.09162402898073196, "global_step": 28956, "epoch": 804, "lr": 0.001} +{"train_loss": 0.033959634602069855, "train_loss_bc": 0.03395666182041168, "train_loss_llm": 0.0002973973751068115, "grad_norm": 0.0475040003657341, "global_step": 28957, "epoch": 804, "lr": 0.001} +{"train_loss": 0.03015044331550598, "train_loss_bc": 0.03015044331550598, "train_loss_llm": 0.0, "grad_norm": 0.07453387975692749, "global_step": 28958, "epoch": 804, "lr": 0.001} +{"train_loss": 0.02804194763302803, "train_loss_bc": 0.02804194763302803, "train_loss_llm": 0.0, "grad_norm": 0.10646998137235641, "global_step": 28959, "epoch": 804, "lr": 0.001} +{"train_loss": 0.032024968415498734, "train_loss_bc": 0.032024968415498734, "train_loss_llm": 0.0, "grad_norm": 0.12735596299171448, "global_step": 28960, "epoch": 804, "lr": 0.001} +{"train_loss": 0.031666599214076996, "train_loss_bc": 0.031666599214076996, "train_loss_llm": 0.0, "grad_norm": 0.037289418280124664, "global_step": 28961, "epoch": 804, "lr": 0.001} +{"train_loss": 0.03234117850661278, "train_loss_bc": 0.03234117850661278, "train_loss_llm": 0.0, "grad_norm": 0.06445776671171188, "global_step": 28962, "epoch": 804, "lr": 0.001} +{"train_loss": 0.032580431550741196, "train_loss_bc": 0.032580431550741196, "train_loss_llm": 0.0, "grad_norm": 0.08284971117973328, "global_step": 28963, "epoch": 804, "lr": 0.001} +{"train_loss": 0.03172113746404648, "train_loss_bc": 0.03172113746404648, "train_loss_llm": 0.0, "grad_norm": 0.09433779865503311, "global_step": 28964, "epoch": 804, "lr": 0.001} +{"train_loss": 0.029324378818273544, "train_loss_bc": 0.029324378818273544, "train_loss_llm": 0.0, "grad_norm": 0.03955843672156334, "global_step": 28965, "epoch": 804, "lr": 0.001} +{"train_loss": 0.026299184188246727, "train_loss_bc": 0.026299184188246727, "train_loss_llm": 0.0, "grad_norm": 0.05954980477690697, "global_step": 28966, "epoch": 804, "lr": 0.001} +{"train_loss": 0.028726227581501007, "train_loss_bc": 0.028726227581501007, "train_loss_llm": 0.0, "grad_norm": 0.06968367099761963, "global_step": 28967, "epoch": 804, "lr": 0.001} +{"train_loss": 0.030451951548457146, "train_loss_bc": 0.030445586889982224, "train_loss_llm": 0.0006365180015563965, "grad_norm": 0.0985633134841919, "global_step": 28968, "epoch": 804, "lr": 0.001} +{"train_loss": 0.029838480055332184, "train_loss_bc": 0.029838480055332184, "train_loss_llm": 0.0, "grad_norm": 0.03465060517191887, "global_step": 28969, "epoch": 804, "lr": 0.001} +{"train_loss": 0.027462879195809364, "train_loss_bc": 0.027462879195809364, "train_loss_llm": 0.0, "grad_norm": 0.058014173060655594, "global_step": 28970, "epoch": 804, "lr": 0.001} +{"train_loss": 0.03675863891839981, "train_loss_bc": 0.03675863891839981, "train_loss_llm": 0.0, "grad_norm": 0.07048048079013824, "global_step": 28971, "epoch": 804, "lr": 0.001} +{"train_loss": 0.02894737757742405, "train_loss_bc": 0.02891656383872032, "train_loss_llm": 0.003081381320953369, "grad_norm": 0.09764783829450607, "global_step": 28972, "epoch": 804, "lr": 0.001} +{"train_loss": 0.033426471054553986, "train_loss_bc": 0.033426471054553986, "train_loss_llm": 0.0, "grad_norm": 0.03127603232860565, "global_step": 28973, "epoch": 804, "lr": 0.001} +{"train_loss": 0.02997877448797226, "train_loss_bc": 0.02997877448797226, "train_loss_llm": 0.0, "grad_norm": 0.058209504932165146, "global_step": 28974, "epoch": 804, "lr": 0.001} +{"train_loss": 0.030850881710648537, "train_loss_bc": 0.030850881710648537, "train_loss_llm": 0.0, "grad_norm": 0.07665306329727173, "global_step": 28975, "epoch": 804, "lr": 0.001} +{"train_loss": 0.03071651980280876, "train_loss_bc": 0.03071651980280876, "train_loss_llm": 0.0, "grad_norm": 0.09436102956533432, "global_step": 28976, "epoch": 804, "lr": 0.001} +{"train_loss": 0.030420638620853424, "train_loss_bc": 0.030417051166296005, "train_loss_llm": 0.0003586709499359131, "grad_norm": 0.03979979455471039, "global_step": 28977, "epoch": 804, "lr": 0.001} +{"train_loss": 0.02433432824909687, "train_loss_bc": 0.02433432824909687, "train_loss_llm": 0.0, "grad_norm": 0.050701528787612915, "global_step": 28978, "epoch": 804, "lr": 0.001} +{"train_loss": 0.030342929686109226, "train_loss_bc": 0.03813933953642845, "train_loss_llm": 0.002093404531478882, "grad_norm": 0.07167862355709076, "global_step": 28979, "epoch": 804, "lr": 0.001, "val_loss": 0.09481199830770493} +{"train_loss": 0.02923816628754139, "train_loss_bc": 0.02923816628754139, "train_loss_llm": 0.0, "grad_norm": 0.09334206581115723, "global_step": 28980, "epoch": 805, "lr": 0.001} +{"train_loss": 0.028668612241744995, "train_loss_bc": 0.028668612241744995, "train_loss_llm": 0.0, "grad_norm": 0.0328698456287384, "global_step": 28981, "epoch": 805, "lr": 0.001} +{"train_loss": 0.03515777736902237, "train_loss_bc": 0.03515777736902237, "train_loss_llm": 0.0, "grad_norm": 0.0520889014005661, "global_step": 28982, "epoch": 805, "lr": 0.001} +{"train_loss": 0.03212302550673485, "train_loss_bc": 0.03212302550673485, "train_loss_llm": 0.0, "grad_norm": 0.07506702095270157, "global_step": 28983, "epoch": 805, "lr": 0.001} +{"train_loss": 0.026022907346487045, "train_loss_bc": 0.026020115241408348, "train_loss_llm": 0.0002792477607727051, "grad_norm": 0.09439576417207718, "global_step": 28984, "epoch": 805, "lr": 0.001} +{"train_loss": 0.030912959948182106, "train_loss_bc": 0.030894838273525238, "train_loss_llm": 0.0018121898174285889, "grad_norm": 0.03336502984166145, "global_step": 28985, "epoch": 805, "lr": 0.001} +{"train_loss": 0.032446518540382385, "train_loss_bc": 0.03242505341768265, "train_loss_llm": 0.002146512269973755, "grad_norm": 0.05490648001432419, "global_step": 28986, "epoch": 805, "lr": 0.001} +{"train_loss": 0.0272621251642704, "train_loss_bc": 0.0272621251642704, "train_loss_llm": 0.0, "grad_norm": 0.0843612402677536, "global_step": 28987, "epoch": 805, "lr": 0.001} +{"train_loss": 0.02614019066095352, "train_loss_bc": 0.02614019066095352, "train_loss_llm": 0.0, "grad_norm": 0.09414486587047577, "global_step": 28988, "epoch": 805, "lr": 0.001} +{"train_loss": 0.029865724965929985, "train_loss_bc": 0.029865724965929985, "train_loss_llm": 0.0, "grad_norm": 0.03487047180533409, "global_step": 28989, "epoch": 805, "lr": 0.001} +{"train_loss": 0.032466400414705276, "train_loss_bc": 0.032466400414705276, "train_loss_llm": 0.0, "grad_norm": 0.05453808978199959, "global_step": 28990, "epoch": 805, "lr": 0.001} +{"train_loss": 0.032164398580789566, "train_loss_bc": 0.032164398580789566, "train_loss_llm": 0.0, "grad_norm": 0.07234039157629013, "global_step": 28991, "epoch": 805, "lr": 0.001} +{"train_loss": 0.03414948284626007, "train_loss_bc": 0.03414948284626007, "train_loss_llm": 0.0, "grad_norm": 0.09386470913887024, "global_step": 28992, "epoch": 805, "lr": 0.001} +{"train_loss": 0.030240479856729507, "train_loss_bc": 0.030240479856729507, "train_loss_llm": 0.0, "grad_norm": 0.03562449291348457, "global_step": 28993, "epoch": 805, "lr": 0.001} +{"train_loss": 0.0241832435131073, "train_loss_bc": 0.024182740598917007, "train_loss_llm": 5.027651786804199e-05, "grad_norm": 0.0521538071334362, "global_step": 28994, "epoch": 805, "lr": 0.001} +{"train_loss": 0.029968207702040672, "train_loss_bc": 0.029968207702040672, "train_loss_llm": 0.0, "grad_norm": 0.06656055152416229, "global_step": 28995, "epoch": 805, "lr": 0.001} +{"train_loss": 0.041187115013599396, "train_loss_bc": 0.041187115013599396, "train_loss_llm": 0.0, "grad_norm": 0.09387395530939102, "global_step": 28996, "epoch": 805, "lr": 0.001} +{"train_loss": 0.03182920068502426, "train_loss_bc": 0.03182920068502426, "train_loss_llm": 0.0, "grad_norm": 0.03680984303355217, "global_step": 28997, "epoch": 805, "lr": 0.001} +{"train_loss": 0.030969571322202682, "train_loss_bc": 0.03095894865691662, "train_loss_llm": 0.0010621845722198486, "grad_norm": 0.06727851927280426, "global_step": 28998, "epoch": 805, "lr": 0.001} +{"train_loss": 0.031791966408491135, "train_loss_bc": 0.03178303688764572, "train_loss_llm": 0.0008930563926696777, "grad_norm": 0.08562200516462326, "global_step": 28999, "epoch": 805, "lr": 0.001} +{"train_loss": 0.031143825501203537, "train_loss_bc": 0.031143825501203537, "train_loss_llm": 0.0, "grad_norm": 0.10517680644989014, "global_step": 29000, "epoch": 805, "lr": 0.001} +{"train_loss": 0.027758413925766945, "train_loss_bc": 0.027758413925766945, "train_loss_llm": 0.0, "grad_norm": 0.03574318066239357, "global_step": 29001, "epoch": 805, "lr": 0.001} +{"train_loss": 0.030993126332759857, "train_loss_bc": 0.030993126332759857, "train_loss_llm": 0.0, "grad_norm": 0.043379224836826324, "global_step": 29002, "epoch": 805, "lr": 0.001} +{"train_loss": 0.027906911447644234, "train_loss_bc": 0.027905911207199097, "train_loss_llm": 0.00010007619857788086, "grad_norm": 0.059665776789188385, "global_step": 29003, "epoch": 805, "lr": 0.001} +{"train_loss": 0.028560783714056015, "train_loss_bc": 0.028560783714056015, "train_loss_llm": 0.0, "grad_norm": 0.07523371279239655, "global_step": 29004, "epoch": 805, "lr": 0.001} +{"train_loss": 0.030513081699609756, "train_loss_bc": 0.030513081699609756, "train_loss_llm": 0.0, "grad_norm": 0.04152891784906387, "global_step": 29005, "epoch": 805, "lr": 0.001} +{"train_loss": 0.030192367732524872, "train_loss_bc": 0.030192367732524872, "train_loss_llm": 0.0, "grad_norm": 0.06736543029546738, "global_step": 29006, "epoch": 805, "lr": 0.001} +{"train_loss": 0.02800232544541359, "train_loss_bc": 0.02800232544541359, "train_loss_llm": 0.0, "grad_norm": 0.09870229661464691, "global_step": 29007, "epoch": 805, "lr": 0.001} +{"train_loss": 0.025708744302392006, "train_loss_bc": 0.025708744302392006, "train_loss_llm": 0.0, "grad_norm": 0.12426861375570297, "global_step": 29008, "epoch": 805, "lr": 0.001} +{"train_loss": 0.03401719406247139, "train_loss_bc": 0.03399590030312538, "train_loss_llm": 0.002129286527633667, "grad_norm": 0.03900972008705139, "global_step": 29009, "epoch": 805, "lr": 0.001} +{"train_loss": 0.027860207483172417, "train_loss_bc": 0.02785595878958702, "train_loss_llm": 0.0004248619079589844, "grad_norm": 0.058900754898786545, "global_step": 29010, "epoch": 805, "lr": 0.001} +{"train_loss": 0.0267866849899292, "train_loss_bc": 0.02678157016634941, "train_loss_llm": 0.00051155686378479, "grad_norm": 0.07121865451335907, "global_step": 29011, "epoch": 805, "lr": 0.001} +{"train_loss": 0.030461832880973816, "train_loss_bc": 0.030461832880973816, "train_loss_llm": 0.0, "grad_norm": 0.09585169702768326, "global_step": 29012, "epoch": 805, "lr": 0.001} +{"train_loss": 0.02720889262855053, "train_loss_bc": 0.02720889262855053, "train_loss_llm": 0.0, "grad_norm": 0.02696516178548336, "global_step": 29013, "epoch": 805, "lr": 0.001} +{"train_loss": 0.03370825946331024, "train_loss_bc": 0.03370825946331024, "train_loss_llm": 0.0, "grad_norm": 0.049094971269369125, "global_step": 29014, "epoch": 805, "lr": 0.001} +{"train_loss": 0.030294997462381918, "train_loss_bc": 0.033009182661771774, "train_loss_llm": 0.0, "grad_norm": 0.07257196307182312, "global_step": 29015, "epoch": 805, "lr": 0.001, "train/cumulative_reward": 92.16301706756803, "train/mean_score": 11.81551672304611, "train/success_rate": 0.2, "test/cumulative_reward": 47.11271189384809, "test/mean_score": 7.562699339818447, "test/success_rate": 0.2, "val_loss": 0.09705249965190887, "train_action_mse_error": 0.032824303954839706} +{"train_loss": 0.027549609541893005, "train_loss_bc": 0.027549609541893005, "train_loss_llm": 0.0, "grad_norm": 0.08381927758455276, "global_step": 29016, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03276506066322327, "train_loss_bc": 0.03276506066322327, "train_loss_llm": 0.0, "grad_norm": 0.033664342015981674, "global_step": 29017, "epoch": 806, "lr": 0.001} +{"train_loss": 0.027626730501651764, "train_loss_bc": 0.027626730501651764, "train_loss_llm": 0.0, "grad_norm": 0.04823563992977142, "global_step": 29018, "epoch": 806, "lr": 0.001} +{"train_loss": 0.028759168460965157, "train_loss_bc": 0.028759168460965157, "train_loss_llm": 0.0, "grad_norm": 0.06137178838253021, "global_step": 29019, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03019999898970127, "train_loss_bc": 0.030194345861673355, "train_loss_llm": 0.0005653500556945801, "grad_norm": 0.0735345333814621, "global_step": 29020, "epoch": 806, "lr": 0.001} +{"train_loss": 0.028583949431777, "train_loss_bc": 0.028583949431777, "train_loss_llm": 0.0, "grad_norm": 0.049631472676992416, "global_step": 29021, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03256884217262268, "train_loss_bc": 0.03256884217262268, "train_loss_llm": 0.0, "grad_norm": 0.08609499037265778, "global_step": 29022, "epoch": 806, "lr": 0.001} +{"train_loss": 0.035847678780555725, "train_loss_bc": 0.035847678780555725, "train_loss_llm": 0.0, "grad_norm": 0.11037372052669525, "global_step": 29023, "epoch": 806, "lr": 0.001} +{"train_loss": 0.023515945300459862, "train_loss_bc": 0.023515945300459862, "train_loss_llm": 0.0, "grad_norm": 0.13211435079574585, "global_step": 29024, "epoch": 806, "lr": 0.001} +{"train_loss": 0.02644408494234085, "train_loss_bc": 0.02644408494234085, "train_loss_llm": 0.0, "grad_norm": 0.03030947968363762, "global_step": 29025, "epoch": 806, "lr": 0.001} +{"train_loss": 0.027041738852858543, "train_loss_bc": 0.027041738852858543, "train_loss_llm": 0.0, "grad_norm": 0.04915809631347656, "global_step": 29026, "epoch": 806, "lr": 0.001} +{"train_loss": 0.030302150174975395, "train_loss_bc": 0.03029295988380909, "train_loss_llm": 0.0009189844131469727, "grad_norm": 0.07377726584672928, "global_step": 29027, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03005104884505272, "train_loss_bc": 0.0300509762018919, "train_loss_llm": 7.18235969543457e-06, "grad_norm": 0.09838790446519852, "global_step": 29028, "epoch": 806, "lr": 0.001} +{"train_loss": 0.028842315077781677, "train_loss_bc": 0.028842315077781677, "train_loss_llm": 0.0, "grad_norm": 0.02722104825079441, "global_step": 29029, "epoch": 806, "lr": 0.001} +{"train_loss": 0.02651423029601574, "train_loss_bc": 0.02651051990687847, "train_loss_llm": 0.00037103891372680664, "grad_norm": 0.047119852155447006, "global_step": 29030, "epoch": 806, "lr": 0.001} +{"train_loss": 0.025837469846010208, "train_loss_bc": 0.025837469846010208, "train_loss_llm": 0.0, "grad_norm": 0.05596489459276199, "global_step": 29031, "epoch": 806, "lr": 0.001} +{"train_loss": 0.028461599722504616, "train_loss_bc": 0.028461599722504616, "train_loss_llm": 0.0, "grad_norm": 0.0672491118311882, "global_step": 29032, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03374316170811653, "train_loss_bc": 0.03374316170811653, "train_loss_llm": 0.0, "grad_norm": 0.04721866175532341, "global_step": 29033, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03025992400944233, "train_loss_bc": 0.030258918181061745, "train_loss_llm": 0.00010058283805847168, "grad_norm": 0.05952661857008934, "global_step": 29034, "epoch": 806, "lr": 0.001} +{"train_loss": 0.030850190669298172, "train_loss_bc": 0.030850190669298172, "train_loss_llm": 0.0, "grad_norm": 0.08289381861686707, "global_step": 29035, "epoch": 806, "lr": 0.001} +{"train_loss": 0.029643258079886436, "train_loss_bc": 0.029643258079886436, "train_loss_llm": 0.0, "grad_norm": 0.1047765463590622, "global_step": 29036, "epoch": 806, "lr": 0.001} +{"train_loss": 0.026189183816313744, "train_loss_bc": 0.026189183816313744, "train_loss_llm": 0.0, "grad_norm": 0.032509852200746536, "global_step": 29037, "epoch": 806, "lr": 0.001} +{"train_loss": 0.028726650401949883, "train_loss_bc": 0.028726650401949883, "train_loss_llm": 0.0, "grad_norm": 0.06486749649047852, "global_step": 29038, "epoch": 806, "lr": 0.001} +{"train_loss": 0.030223602429032326, "train_loss_bc": 0.030223602429032326, "train_loss_llm": 0.0, "grad_norm": 0.08219827711582184, "global_step": 29039, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03362628072500229, "train_loss_bc": 0.03362628072500229, "train_loss_llm": 0.0, "grad_norm": 0.10742849111557007, "global_step": 29040, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03186731040477753, "train_loss_bc": 0.03186726197600365, "train_loss_llm": 4.738569259643555e-06, "grad_norm": 0.032664865255355835, "global_step": 29041, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03037094883620739, "train_loss_bc": 0.03037094883620739, "train_loss_llm": 0.0, "grad_norm": 0.044073186814785004, "global_step": 29042, "epoch": 806, "lr": 0.001} +{"train_loss": 0.030338026583194733, "train_loss_bc": 0.030338026583194733, "train_loss_llm": 0.0, "grad_norm": 0.07410590350627899, "global_step": 29043, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03303932771086693, "train_loss_bc": 0.03303932771086693, "train_loss_llm": 0.0, "grad_norm": 0.09835808724164963, "global_step": 29044, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03275775909423828, "train_loss_bc": 0.03275775909423828, "train_loss_llm": 0.0, "grad_norm": 0.042069144546985626, "global_step": 29045, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03769766166806221, "train_loss_bc": 0.03769766166806221, "train_loss_llm": 0.0, "grad_norm": 0.06538930535316467, "global_step": 29046, "epoch": 806, "lr": 0.001} +{"train_loss": 0.030351126566529274, "train_loss_bc": 0.030351126566529274, "train_loss_llm": 0.0, "grad_norm": 0.07646091282367706, "global_step": 29047, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03085620515048504, "train_loss_bc": 0.03085620515048504, "train_loss_llm": 0.0, "grad_norm": 0.09413061290979385, "global_step": 29048, "epoch": 806, "lr": 0.001} +{"train_loss": 0.029094979166984558, "train_loss_bc": 0.029094979166984558, "train_loss_llm": 0.0, "grad_norm": 0.035250164568424225, "global_step": 29049, "epoch": 806, "lr": 0.001} +{"train_loss": 0.03637278825044632, "train_loss_bc": 0.03637148439884186, "train_loss_llm": 0.0001303553581237793, "grad_norm": 0.0593377910554409, "global_step": 29050, "epoch": 806, "lr": 0.001} +{"train_loss": 0.030424689667092428, "train_loss_bc": 0.038324546068906784, "train_loss_llm": 0.004427522420883179, "grad_norm": 0.10251995921134949, "global_step": 29051, "epoch": 806, "lr": 0.001, "val_loss": 0.09851769357919693} +{"train_loss": 0.021947694942355156, "train_loss_bc": 0.021947694942355156, "train_loss_llm": 0.0, "grad_norm": 0.11625608801841736, "global_step": 29052, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03375284746289253, "train_loss_bc": 0.03375284746289253, "train_loss_llm": 0.0, "grad_norm": 0.037079621106386185, "global_step": 29053, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03187670558691025, "train_loss_bc": 0.03187670558691025, "train_loss_llm": 0.0, "grad_norm": 0.06131245568394661, "global_step": 29054, "epoch": 807, "lr": 0.001} +{"train_loss": 0.034904491156339645, "train_loss_bc": 0.034904491156339645, "train_loss_llm": 0.0, "grad_norm": 0.07550833374261856, "global_step": 29055, "epoch": 807, "lr": 0.001} +{"train_loss": 0.032936953008174896, "train_loss_bc": 0.032936953008174896, "train_loss_llm": 0.0, "grad_norm": 0.10006210207939148, "global_step": 29056, "epoch": 807, "lr": 0.001} +{"train_loss": 0.034023165702819824, "train_loss_bc": 0.034023165702819824, "train_loss_llm": 0.0, "grad_norm": 0.03954572230577469, "global_step": 29057, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03364795446395874, "train_loss_bc": 0.03364795446395874, "train_loss_llm": 0.0, "grad_norm": 0.06559926271438599, "global_step": 29058, "epoch": 807, "lr": 0.001} +{"train_loss": 0.029538864269852638, "train_loss_bc": 0.029538864269852638, "train_loss_llm": 0.0, "grad_norm": 0.07822702080011368, "global_step": 29059, "epoch": 807, "lr": 0.001} +{"train_loss": 0.02995011769235134, "train_loss_bc": 0.02995011769235134, "train_loss_llm": 0.0, "grad_norm": 0.1019933745265007, "global_step": 29060, "epoch": 807, "lr": 0.001} +{"train_loss": 0.028752682730555534, "train_loss_bc": 0.02872091718018055, "train_loss_llm": 0.0031765997409820557, "grad_norm": 0.04001303389668465, "global_step": 29061, "epoch": 807, "lr": 0.001} +{"train_loss": 0.027196921408176422, "train_loss_bc": 0.027196921408176422, "train_loss_llm": 0.0, "grad_norm": 0.06111402437090874, "global_step": 29062, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03439956530928612, "train_loss_bc": 0.03439556807279587, "train_loss_llm": 0.0003997981548309326, "grad_norm": 0.09241707623004913, "global_step": 29063, "epoch": 807, "lr": 0.001} +{"train_loss": 0.0381602980196476, "train_loss_bc": 0.0381602980196476, "train_loss_llm": 0.0, "grad_norm": 0.1391189843416214, "global_step": 29064, "epoch": 807, "lr": 0.001} +{"train_loss": 0.027786236256361008, "train_loss_bc": 0.027786236256361008, "train_loss_llm": 0.0, "grad_norm": 0.035656873136758804, "global_step": 29065, "epoch": 807, "lr": 0.001} +{"train_loss": 0.027177458629012108, "train_loss_bc": 0.027177458629012108, "train_loss_llm": 0.0, "grad_norm": 0.06438581645488739, "global_step": 29066, "epoch": 807, "lr": 0.001} +{"train_loss": 0.031898900866508484, "train_loss_bc": 0.03189001604914665, "train_loss_llm": 0.0008885562419891357, "grad_norm": 0.0983717292547226, "global_step": 29067, "epoch": 807, "lr": 0.001} +{"train_loss": 0.035451456904411316, "train_loss_bc": 0.035451456904411316, "train_loss_llm": 0.0, "grad_norm": 0.12445317953824997, "global_step": 29068, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03273378312587738, "train_loss_bc": 0.03273378312587738, "train_loss_llm": 0.0, "grad_norm": 0.03812280669808388, "global_step": 29069, "epoch": 807, "lr": 0.001} +{"train_loss": 0.01924862153828144, "train_loss_bc": 0.01924862153828144, "train_loss_llm": 0.0, "grad_norm": 0.054131314158439636, "global_step": 29070, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03200262412428856, "train_loss_bc": 0.031992681324481964, "train_loss_llm": 0.0009941458702087402, "grad_norm": 0.08531515300273895, "global_step": 29071, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03185967728495598, "train_loss_bc": 0.03185967728495598, "train_loss_llm": 0.0, "grad_norm": 0.11162634938955307, "global_step": 29072, "epoch": 807, "lr": 0.001} +{"train_loss": 0.02777751348912716, "train_loss_bc": 0.02777751348912716, "train_loss_llm": 0.0, "grad_norm": 0.03940845653414726, "global_step": 29073, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03067047707736492, "train_loss_bc": 0.03067047707736492, "train_loss_llm": 0.0, "grad_norm": 0.05097360163927078, "global_step": 29074, "epoch": 807, "lr": 0.001} +{"train_loss": 0.030228380113840103, "train_loss_bc": 0.030228380113840103, "train_loss_llm": 0.0, "grad_norm": 0.07251617312431335, "global_step": 29075, "epoch": 807, "lr": 0.001} +{"train_loss": 0.022553514689207077, "train_loss_bc": 0.02254616655409336, "train_loss_llm": 0.0007347762584686279, "grad_norm": 0.08338603377342224, "global_step": 29076, "epoch": 807, "lr": 0.001} +{"train_loss": 0.028032395988702774, "train_loss_bc": 0.028032395988702774, "train_loss_llm": 0.0, "grad_norm": 0.029841160401701927, "global_step": 29077, "epoch": 807, "lr": 0.001} +{"train_loss": 0.02667338214814663, "train_loss_bc": 0.02667338214814663, "train_loss_llm": 0.0, "grad_norm": 0.04065459966659546, "global_step": 29078, "epoch": 807, "lr": 0.001} +{"train_loss": 0.032475683838129044, "train_loss_bc": 0.03246685117483139, "train_loss_llm": 0.0008833408355712891, "grad_norm": 0.0708557739853859, "global_step": 29079, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03073570691049099, "train_loss_bc": 0.03073570691049099, "train_loss_llm": 0.0, "grad_norm": 0.08346706628799438, "global_step": 29080, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03192639350891113, "train_loss_bc": 0.03192639350891113, "train_loss_llm": 0.0, "grad_norm": 0.04317464306950569, "global_step": 29081, "epoch": 807, "lr": 0.001} +{"train_loss": 0.02947068028151989, "train_loss_bc": 0.02947068028151989, "train_loss_llm": 0.0, "grad_norm": 0.06796276569366455, "global_step": 29082, "epoch": 807, "lr": 0.001} +{"train_loss": 0.02952379174530506, "train_loss_bc": 0.02952379174530506, "train_loss_llm": 0.0, "grad_norm": 0.07729710638523102, "global_step": 29083, "epoch": 807, "lr": 0.001} +{"train_loss": 0.032337307929992676, "train_loss_bc": 0.03232704848051071, "train_loss_llm": 0.0010258853435516357, "grad_norm": 0.09869008511304855, "global_step": 29084, "epoch": 807, "lr": 0.001} +{"train_loss": 0.03130514919757843, "train_loss_bc": 0.03130514919757843, "train_loss_llm": 0.0, "grad_norm": 0.041271455585956573, "global_step": 29085, "epoch": 807, "lr": 0.001} +{"train_loss": 0.030285146087408066, "train_loss_bc": 0.030285146087408066, "train_loss_llm": 0.0, "grad_norm": 0.0653020367026329, "global_step": 29086, "epoch": 807, "lr": 0.001} +{"train_loss": 0.030320421430385776, "train_loss_bc": 0.028290094807744026, "train_loss_llm": 0.00025337934494018555, "grad_norm": 0.10447555780410767, "global_step": 29087, "epoch": 807, "lr": 0.001, "val_loss": 0.09759277105331421} +{"train_loss": 0.031588949263095856, "train_loss_bc": 0.031588949263095856, "train_loss_llm": 0.0, "grad_norm": 0.13463354110717773, "global_step": 29088, "epoch": 808, "lr": 0.001} +{"train_loss": 0.03102693520486355, "train_loss_bc": 0.03102693520486355, "train_loss_llm": 0.0, "grad_norm": 0.039293836802244186, "global_step": 29089, "epoch": 808, "lr": 0.001} +{"train_loss": 0.02884032018482685, "train_loss_bc": 0.02884032018482685, "train_loss_llm": 0.0, "grad_norm": 0.06529582291841507, "global_step": 29090, "epoch": 808, "lr": 0.001} +{"train_loss": 0.0367463119328022, "train_loss_bc": 0.036723989993333817, "train_loss_llm": 0.0022320151329040527, "grad_norm": 0.08817365765571594, "global_step": 29091, "epoch": 808, "lr": 0.001} +{"train_loss": 0.026785777881741524, "train_loss_bc": 0.026778317987918854, "train_loss_llm": 0.0007459521293640137, "grad_norm": 0.10050196945667267, "global_step": 29092, "epoch": 808, "lr": 0.001} +{"train_loss": 0.03432246670126915, "train_loss_bc": 0.03430183231830597, "train_loss_llm": 0.002063274383544922, "grad_norm": 0.040136825293302536, "global_step": 29093, "epoch": 808, "lr": 0.001} +{"train_loss": 0.02661341056227684, "train_loss_bc": 0.02661341056227684, "train_loss_llm": 0.0, "grad_norm": 0.05270747095346451, "global_step": 29094, "epoch": 808, "lr": 0.001} +{"train_loss": 0.023765958845615387, "train_loss_bc": 0.023765362799167633, "train_loss_llm": 5.951523780822754e-05, "grad_norm": 0.05881919339299202, "global_step": 29095, "epoch": 808, "lr": 0.001} +{"train_loss": 0.034098975360393524, "train_loss_bc": 0.034098975360393524, "train_loss_llm": 0.0, "grad_norm": 0.08143144845962524, "global_step": 29096, "epoch": 808, "lr": 0.001} +{"train_loss": 0.033676281571388245, "train_loss_bc": 0.033676281571388245, "train_loss_llm": 0.0, "grad_norm": 0.03751241788268089, "global_step": 29097, "epoch": 808, "lr": 0.001} +{"train_loss": 0.03033139929175377, "train_loss_bc": 0.03033139929175377, "train_loss_llm": 0.0, "grad_norm": 0.05869891867041588, "global_step": 29098, "epoch": 808, "lr": 0.001} +{"train_loss": 0.031078550964593887, "train_loss_bc": 0.031078550964593887, "train_loss_llm": 0.0, "grad_norm": 0.07346664369106293, "global_step": 29099, "epoch": 808, "lr": 0.001} +{"train_loss": 0.029451951384544373, "train_loss_bc": 0.029451951384544373, "train_loss_llm": 0.0, "grad_norm": 0.08630476146936417, "global_step": 29100, "epoch": 808, "lr": 0.001} +{"train_loss": 0.026687372475862503, "train_loss_bc": 0.026683377102017403, "train_loss_llm": 0.00039952993392944336, "grad_norm": 0.03896249830722809, "global_step": 29101, "epoch": 808, "lr": 0.001} +{"train_loss": 0.025521693751215935, "train_loss_bc": 0.025521693751215935, "train_loss_llm": 0.0, "grad_norm": 0.05886409431695938, "global_step": 29102, "epoch": 808, "lr": 0.001} +{"train_loss": 0.03356395661830902, "train_loss_bc": 0.03356395661830902, "train_loss_llm": 0.0, "grad_norm": 0.09470104426145554, "global_step": 29103, "epoch": 808, "lr": 0.001} +{"train_loss": 0.02799132466316223, "train_loss_bc": 0.02799132466316223, "train_loss_llm": 0.0, "grad_norm": 0.11066136509180069, "global_step": 29104, "epoch": 808, "lr": 0.001} +{"train_loss": 0.034699901938438416, "train_loss_bc": 0.03468574956059456, "train_loss_llm": 0.0014151334762573242, "grad_norm": 0.04124132916331291, "global_step": 29105, "epoch": 808, "lr": 0.001} +{"train_loss": 0.029660718515515327, "train_loss_bc": 0.029660718515515327, "train_loss_llm": 0.0, "grad_norm": 0.0719236433506012, "global_step": 29106, "epoch": 808, "lr": 0.001} +{"train_loss": 0.03501816838979721, "train_loss_bc": 0.03501051664352417, "train_loss_llm": 0.0007652044296264648, "grad_norm": 0.10913559049367905, "global_step": 29107, "epoch": 808, "lr": 0.001} +{"train_loss": 0.02770864963531494, "train_loss_bc": 0.02770864963531494, "train_loss_llm": 0.0, "grad_norm": 0.13651837408542633, "global_step": 29108, "epoch": 808, "lr": 0.001} +{"train_loss": 0.032562606036663055, "train_loss_bc": 0.032562606036663055, "train_loss_llm": 0.0, "grad_norm": 0.039647672325372696, "global_step": 29109, "epoch": 808, "lr": 0.001} +{"train_loss": 0.030957765877246857, "train_loss_bc": 0.03094670921564102, "train_loss_llm": 0.0011057555675506592, "grad_norm": 0.06094661355018616, "global_step": 29110, "epoch": 808, "lr": 0.001} +{"train_loss": 0.026306219398975372, "train_loss_bc": 0.026305504143238068, "train_loss_llm": 7.149577140808105e-05, "grad_norm": 0.0787547156214714, "global_step": 29111, "epoch": 808, "lr": 0.001} +{"train_loss": 0.03171755000948906, "train_loss_bc": 0.03171723335981369, "train_loss_llm": 3.1501054763793945e-05, "grad_norm": 0.10661247372627258, "global_step": 29112, "epoch": 808, "lr": 0.001} +{"train_loss": 0.03568161651492119, "train_loss_bc": 0.03568161651492119, "train_loss_llm": 0.0, "grad_norm": 0.044532932341098785, "global_step": 29113, "epoch": 808, "lr": 0.001} +{"train_loss": 0.032036807388067245, "train_loss_bc": 0.032036807388067245, "train_loss_llm": 0.0, "grad_norm": 0.054029107093811035, "global_step": 29114, "epoch": 808, "lr": 0.001} +{"train_loss": 0.02286473661661148, "train_loss_bc": 0.02286473661661148, "train_loss_llm": 0.0, "grad_norm": 0.07009638100862503, "global_step": 29115, "epoch": 808, "lr": 0.001} +{"train_loss": 0.026351457461714745, "train_loss_bc": 0.026351457461714745, "train_loss_llm": 0.0, "grad_norm": 0.08346421271562576, "global_step": 29116, "epoch": 808, "lr": 0.001} +{"train_loss": 0.029692580923438072, "train_loss_bc": 0.029692580923438072, "train_loss_llm": 0.0, "grad_norm": 0.03170505538582802, "global_step": 29117, "epoch": 808, "lr": 0.001} +{"train_loss": 0.030675994232296944, "train_loss_bc": 0.030675994232296944, "train_loss_llm": 0.0, "grad_norm": 0.049520093947649, "global_step": 29118, "epoch": 808, "lr": 0.001} +{"train_loss": 0.029481666162610054, "train_loss_bc": 0.029481666162610054, "train_loss_llm": 0.0, "grad_norm": 0.06196560338139534, "global_step": 29119, "epoch": 808, "lr": 0.001} +{"train_loss": 0.03001566417515278, "train_loss_bc": 0.03001566417515278, "train_loss_llm": 0.0, "grad_norm": 0.07342023402452469, "global_step": 29120, "epoch": 808, "lr": 0.001} +{"train_loss": 0.031041858717799187, "train_loss_bc": 0.031041858717799187, "train_loss_llm": 0.0, "grad_norm": 0.03867471590638161, "global_step": 29121, "epoch": 808, "lr": 0.001} +{"train_loss": 0.03044605627655983, "train_loss_bc": 0.030445560812950134, "train_loss_llm": 4.947185516357422e-05, "grad_norm": 0.062146708369255066, "global_step": 29122, "epoch": 808, "lr": 0.001} +{"train_loss": 0.030308532683799665, "train_loss_bc": 0.032095521688461304, "train_loss_llm": 0.0, "grad_norm": 0.099305659532547, "global_step": 29123, "epoch": 808, "lr": 0.001, "val_loss": 0.09846685826778412} +{"train_loss": 0.027311867102980614, "train_loss_bc": 0.027311867102980614, "train_loss_llm": 0.0, "grad_norm": 0.1149330586194992, "global_step": 29124, "epoch": 809, "lr": 0.001} +{"train_loss": 0.03736131265759468, "train_loss_bc": 0.03736131265759468, "train_loss_llm": 0.0, "grad_norm": 0.04589856043457985, "global_step": 29125, "epoch": 809, "lr": 0.001} +{"train_loss": 0.030456392094492912, "train_loss_bc": 0.030456392094492912, "train_loss_llm": 0.0, "grad_norm": 0.08902014791965485, "global_step": 29126, "epoch": 809, "lr": 0.001} +{"train_loss": 0.03527271747589111, "train_loss_bc": 0.03527271747589111, "train_loss_llm": 0.0, "grad_norm": 0.1316923052072525, "global_step": 29127, "epoch": 809, "lr": 0.001} +{"train_loss": 0.032904643565416336, "train_loss_bc": 0.03290165215730667, "train_loss_llm": 0.00029921531677246094, "grad_norm": 0.14300602674484253, "global_step": 29128, "epoch": 809, "lr": 0.001} +{"train_loss": 0.029390664771199226, "train_loss_bc": 0.029390664771199226, "train_loss_llm": 0.0, "grad_norm": 0.03748079016804695, "global_step": 29129, "epoch": 809, "lr": 0.001} +{"train_loss": 0.029526574537158012, "train_loss_bc": 0.029518423601984978, "train_loss_llm": 0.0008150339126586914, "grad_norm": 0.06484673172235489, "global_step": 29130, "epoch": 809, "lr": 0.001} +{"train_loss": 0.029137998819351196, "train_loss_bc": 0.029137998819351196, "train_loss_llm": 0.0, "grad_norm": 0.07339563220739365, "global_step": 29131, "epoch": 809, "lr": 0.001} +{"train_loss": 0.03139444813132286, "train_loss_bc": 0.03139444813132286, "train_loss_llm": 0.0, "grad_norm": 0.10749853402376175, "global_step": 29132, "epoch": 809, "lr": 0.001} +{"train_loss": 0.02906867116689682, "train_loss_bc": 0.029067199677228928, "train_loss_llm": 0.00014713406562805176, "grad_norm": 0.03171616420149803, "global_step": 29133, "epoch": 809, "lr": 0.001} +{"train_loss": 0.02806207723915577, "train_loss_bc": 0.02806088887155056, "train_loss_llm": 0.00011882185935974121, "grad_norm": 0.05069420859217644, "global_step": 29134, "epoch": 809, "lr": 0.001} +{"train_loss": 0.025453917682170868, "train_loss_bc": 0.025453917682170868, "train_loss_llm": 0.0, "grad_norm": 0.06890981644392014, "global_step": 29135, "epoch": 809, "lr": 0.001} +{"train_loss": 0.02769562602043152, "train_loss_bc": 0.02769562602043152, "train_loss_llm": 0.0, "grad_norm": 0.06884108483791351, "global_step": 29136, "epoch": 809, "lr": 0.001} +{"train_loss": 0.02458495832979679, "train_loss_bc": 0.024580059573054314, "train_loss_llm": 0.0004898607730865479, "grad_norm": 0.03522351756691933, "global_step": 29137, "epoch": 809, "lr": 0.001} +{"train_loss": 0.033216092735528946, "train_loss_bc": 0.03319769352674484, "train_loss_llm": 0.0018398761749267578, "grad_norm": 0.0661638155579567, "global_step": 29138, "epoch": 809, "lr": 0.001} +{"train_loss": 0.03246845677495003, "train_loss_bc": 0.03245384618639946, "train_loss_llm": 0.0014609694480895996, "grad_norm": 0.07833144813776016, "global_step": 29139, "epoch": 809, "lr": 0.001} +{"train_loss": 0.03196602314710617, "train_loss_bc": 0.03196602314710617, "train_loss_llm": 0.0, "grad_norm": 0.09887056797742844, "global_step": 29140, "epoch": 809, "lr": 0.001} +{"train_loss": 0.034426409751176834, "train_loss_bc": 0.034426409751176834, "train_loss_llm": 0.0, "grad_norm": 0.0384160578250885, "global_step": 29141, "epoch": 809, "lr": 0.001} +{"train_loss": 0.028256742283701897, "train_loss_bc": 0.028256742283701897, "train_loss_llm": 0.0, "grad_norm": 0.05851110443472862, "global_step": 29142, "epoch": 809, "lr": 0.001} +{"train_loss": 0.03241179510951042, "train_loss_bc": 0.03241179510951042, "train_loss_llm": 0.0, "grad_norm": 0.09539075195789337, "global_step": 29143, "epoch": 809, "lr": 0.001} +{"train_loss": 0.029981927946209908, "train_loss_bc": 0.029981927946209908, "train_loss_llm": 0.0, "grad_norm": 0.12170492857694626, "global_step": 29144, "epoch": 809, "lr": 0.001} +{"train_loss": 0.03425029665231705, "train_loss_bc": 0.03425029665231705, "train_loss_llm": 0.0, "grad_norm": 0.04273873195052147, "global_step": 29145, "epoch": 809, "lr": 0.001} +{"train_loss": 0.031325943768024445, "train_loss_bc": 0.031325943768024445, "train_loss_llm": 0.0, "grad_norm": 0.058447614312171936, "global_step": 29146, "epoch": 809, "lr": 0.001} +{"train_loss": 0.026126962155103683, "train_loss_bc": 0.026126962155103683, "train_loss_llm": 0.0, "grad_norm": 0.09043533354997635, "global_step": 29147, "epoch": 809, "lr": 0.001} +{"train_loss": 0.026802146807312965, "train_loss_bc": 0.026799073442816734, "train_loss_llm": 0.00030726194381713867, "grad_norm": 0.10248545557260513, "global_step": 29148, "epoch": 809, "lr": 0.001} +{"train_loss": 0.028793880715966225, "train_loss_bc": 0.028793880715966225, "train_loss_llm": 0.0, "grad_norm": 0.030813151970505714, "global_step": 29149, "epoch": 809, "lr": 0.001} +{"train_loss": 0.02683577872812748, "train_loss_bc": 0.02683577872812748, "train_loss_llm": 0.0, "grad_norm": 0.06886371225118637, "global_step": 29150, "epoch": 809, "lr": 0.001} +{"train_loss": 0.03100455366075039, "train_loss_bc": 0.030981576070189476, "train_loss_llm": 0.0022976696491241455, "grad_norm": 0.08642470836639404, "global_step": 29151, "epoch": 809, "lr": 0.001} +{"train_loss": 0.026197830215096474, "train_loss_bc": 0.026194820180535316, "train_loss_llm": 0.00030094385147094727, "grad_norm": 0.10030420124530792, "global_step": 29152, "epoch": 809, "lr": 0.001} +{"train_loss": 0.030578533187508583, "train_loss_bc": 0.030578533187508583, "train_loss_llm": 0.0, "grad_norm": 0.044852688908576965, "global_step": 29153, "epoch": 809, "lr": 0.001} +{"train_loss": 0.02849571779370308, "train_loss_bc": 0.02849571779370308, "train_loss_llm": 0.0, "grad_norm": 0.06957480311393738, "global_step": 29154, "epoch": 809, "lr": 0.001} +{"train_loss": 0.035099443048238754, "train_loss_bc": 0.035099443048238754, "train_loss_llm": 0.0, "grad_norm": 0.09360018372535706, "global_step": 29155, "epoch": 809, "lr": 0.001} +{"train_loss": 0.030294615775346756, "train_loss_bc": 0.030294615775346756, "train_loss_llm": 0.0, "grad_norm": 0.10369314253330231, "global_step": 29156, "epoch": 809, "lr": 0.001} +{"train_loss": 0.03564143925905228, "train_loss_bc": 0.0356321856379509, "train_loss_llm": 0.0009254217147827148, "grad_norm": 0.04771488904953003, "global_step": 29157, "epoch": 809, "lr": 0.001} +{"train_loss": 0.030947869643568993, "train_loss_bc": 0.030947869643568993, "train_loss_llm": 0.0, "grad_norm": 0.05552351847290993, "global_step": 29158, "epoch": 809, "lr": 0.001} +{"train_loss": 0.030213447546379432, "train_loss_bc": 0.024939782917499542, "train_loss_llm": 0.0, "grad_norm": 0.0768369510769844, "global_step": 29159, "epoch": 809, "lr": 0.001, "val_loss": 0.09681850671768188} +{"train_loss": 0.03354622796177864, "train_loss_bc": 0.033544860780239105, "train_loss_llm": 0.0001366734504699707, "grad_norm": 0.11007803678512573, "global_step": 29160, "epoch": 810, "lr": 0.001} +{"train_loss": 0.02885862998664379, "train_loss_bc": 0.028849119320511818, "train_loss_llm": 0.0009510219097137451, "grad_norm": 0.037065159529447556, "global_step": 29161, "epoch": 810, "lr": 0.001} +{"train_loss": 0.03281074017286301, "train_loss_bc": 0.03280618414282799, "train_loss_llm": 0.00045546889305114746, "grad_norm": 0.05093497782945633, "global_step": 29162, "epoch": 810, "lr": 0.001} +{"train_loss": 0.029483357444405556, "train_loss_bc": 0.029483357444405556, "train_loss_llm": 0.0, "grad_norm": 0.059420306235551834, "global_step": 29163, "epoch": 810, "lr": 0.001} +{"train_loss": 0.021436156705021858, "train_loss_bc": 0.021436156705021858, "train_loss_llm": 0.0, "grad_norm": 0.07223296910524368, "global_step": 29164, "epoch": 810, "lr": 0.001} +{"train_loss": 0.030112462118268013, "train_loss_bc": 0.030112462118268013, "train_loss_llm": 0.0, "grad_norm": 0.03645462542772293, "global_step": 29165, "epoch": 810, "lr": 0.001} +{"train_loss": 0.03630421310663223, "train_loss_bc": 0.03630421310663223, "train_loss_llm": 0.0, "grad_norm": 0.06264282763004303, "global_step": 29166, "epoch": 810, "lr": 0.001} +{"train_loss": 0.029391154646873474, "train_loss_bc": 0.029391154646873474, "train_loss_llm": 0.0, "grad_norm": 0.07760879397392273, "global_step": 29167, "epoch": 810, "lr": 0.001} +{"train_loss": 0.03055179864168167, "train_loss_bc": 0.03055179864168167, "train_loss_llm": 0.0, "grad_norm": 0.1080155074596405, "global_step": 29168, "epoch": 810, "lr": 0.001} +{"train_loss": 0.03347834572196007, "train_loss_bc": 0.03347834572196007, "train_loss_llm": 0.0, "grad_norm": 0.04308449104428291, "global_step": 29169, "epoch": 810, "lr": 0.001} +{"train_loss": 0.02936670184135437, "train_loss_bc": 0.02936670184135437, "train_loss_llm": 0.0, "grad_norm": 0.0688711479306221, "global_step": 29170, "epoch": 810, "lr": 0.001} +{"train_loss": 0.029157454147934914, "train_loss_bc": 0.029157454147934914, "train_loss_llm": 0.0, "grad_norm": 0.08328618109226227, "global_step": 29171, "epoch": 810, "lr": 0.001} +{"train_loss": 0.036151740700006485, "train_loss_bc": 0.03613511845469475, "train_loss_llm": 0.001662224531173706, "grad_norm": 0.10220769047737122, "global_step": 29172, "epoch": 810, "lr": 0.001} +{"train_loss": 0.03157614916563034, "train_loss_bc": 0.031561363488435745, "train_loss_llm": 0.001478433609008789, "grad_norm": 0.05134556069970131, "global_step": 29173, "epoch": 810, "lr": 0.001} +{"train_loss": 0.02694830298423767, "train_loss_bc": 0.02694830298423767, "train_loss_llm": 0.0, "grad_norm": 0.07170262932777405, "global_step": 29174, "epoch": 810, "lr": 0.001} +{"train_loss": 0.03377435356378555, "train_loss_bc": 0.03377435356378555, "train_loss_llm": 0.0, "grad_norm": 0.0977010577917099, "global_step": 29175, "epoch": 810, "lr": 0.001} +{"train_loss": 0.03542475402355194, "train_loss_bc": 0.035409439355134964, "train_loss_llm": 0.0015313327312469482, "grad_norm": 0.1244516596198082, "global_step": 29176, "epoch": 810, "lr": 0.001} +{"train_loss": 0.02762511558830738, "train_loss_bc": 0.02761552482843399, "train_loss_llm": 0.0009590983390808105, "grad_norm": 0.0323413610458374, "global_step": 29177, "epoch": 810, "lr": 0.001} +{"train_loss": 0.03069530427455902, "train_loss_bc": 0.03069387748837471, "train_loss_llm": 0.00014275312423706055, "grad_norm": 0.06313323229551315, "global_step": 29178, "epoch": 810, "lr": 0.001} +{"train_loss": 0.030093155801296234, "train_loss_bc": 0.030093155801296234, "train_loss_llm": 0.0, "grad_norm": 0.07366301119327545, "global_step": 29179, "epoch": 810, "lr": 0.001} +{"train_loss": 0.026386035606265068, "train_loss_bc": 0.026386035606265068, "train_loss_llm": 0.0, "grad_norm": 0.09804719686508179, "global_step": 29180, "epoch": 810, "lr": 0.001} +{"train_loss": 0.03329378738999367, "train_loss_bc": 0.03328637406229973, "train_loss_llm": 0.0007413923740386963, "grad_norm": 0.031158261001110077, "global_step": 29181, "epoch": 810, "lr": 0.001} +{"train_loss": 0.024338185787200928, "train_loss_bc": 0.024333007633686066, "train_loss_llm": 0.0005178749561309814, "grad_norm": 0.06154513731598854, "global_step": 29182, "epoch": 810, "lr": 0.001} +{"train_loss": 0.031879179179668427, "train_loss_bc": 0.031879179179668427, "train_loss_llm": 0.0, "grad_norm": 0.08017458021640778, "global_step": 29183, "epoch": 810, "lr": 0.001} +{"train_loss": 0.029312893748283386, "train_loss_bc": 0.029312893748283386, "train_loss_llm": 0.0, "grad_norm": 0.09933845698833466, "global_step": 29184, "epoch": 810, "lr": 0.001} +{"train_loss": 0.034499574452638626, "train_loss_bc": 0.034499574452638626, "train_loss_llm": 0.0, "grad_norm": 0.036233723163604736, "global_step": 29185, "epoch": 810, "lr": 0.001} +{"train_loss": 0.029758336022496223, "train_loss_bc": 0.029747694730758667, "train_loss_llm": 0.001064211130142212, "grad_norm": 0.06009170040488243, "global_step": 29186, "epoch": 810, "lr": 0.001} +{"train_loss": 0.026989497244358063, "train_loss_bc": 0.026989497244358063, "train_loss_llm": 0.0, "grad_norm": 0.07411595433950424, "global_step": 29187, "epoch": 810, "lr": 0.001} +{"train_loss": 0.02946343645453453, "train_loss_bc": 0.029449615627527237, "train_loss_llm": 0.0013821721076965332, "grad_norm": 0.09903609752655029, "global_step": 29188, "epoch": 810, "lr": 0.001} +{"train_loss": 0.0293903611600399, "train_loss_bc": 0.0293903611600399, "train_loss_llm": 0.0, "grad_norm": 0.031557176262140274, "global_step": 29189, "epoch": 810, "lr": 0.001} +{"train_loss": 0.030577069148421288, "train_loss_bc": 0.030577069148421288, "train_loss_llm": 0.0, "grad_norm": 0.06301283091306686, "global_step": 29190, "epoch": 810, "lr": 0.001} +{"train_loss": 0.02941926009953022, "train_loss_bc": 0.02941926009953022, "train_loss_llm": 0.0, "grad_norm": 0.08480913192033768, "global_step": 29191, "epoch": 810, "lr": 0.001} +{"train_loss": 0.030577844008803368, "train_loss_bc": 0.03057122603058815, "train_loss_llm": 0.0006617605686187744, "grad_norm": 0.11324271559715271, "global_step": 29192, "epoch": 810, "lr": 0.001} +{"train_loss": 0.0331512987613678, "train_loss_bc": 0.0331512987613678, "train_loss_llm": 0.0, "grad_norm": 0.03509809076786041, "global_step": 29193, "epoch": 810, "lr": 0.001} +{"train_loss": 0.023268025368452072, "train_loss_bc": 0.023265212774276733, "train_loss_llm": 0.00028133392333984375, "grad_norm": 0.0643438771367073, "global_step": 29194, "epoch": 810, "lr": 0.001} +{"train_loss": 0.030323466648244195, "train_loss_bc": 0.03255389630794525, "train_loss_llm": 0.0, "grad_norm": 0.1104356124997139, "global_step": 29195, "epoch": 810, "lr": 0.001, "train/cumulative_reward": 97.9439459532875, "train/mean_score": 15.74732683324667, "train/success_rate": 0.4, "test/cumulative_reward": 27.487395067775907, "test/mean_score": 7.025795615960452, "test/success_rate": 0.18, "val_loss": 0.09592099487781525, "train_action_mse_error": 0.033384039998054504} +{"train_loss": 0.033358823508024216, "train_loss_bc": 0.033358823508024216, "train_loss_llm": 0.0, "grad_norm": 0.1430891901254654, "global_step": 29196, "epoch": 811, "lr": 0.001} +{"train_loss": 0.029055049642920494, "train_loss_bc": 0.029055049642920494, "train_loss_llm": 0.0, "grad_norm": 0.04361589625477791, "global_step": 29197, "epoch": 811, "lr": 0.001} +{"train_loss": 0.028652483597397804, "train_loss_bc": 0.028644898906350136, "train_loss_llm": 0.0007584989070892334, "grad_norm": 0.08142361044883728, "global_step": 29198, "epoch": 811, "lr": 0.001} +{"train_loss": 0.028580352663993835, "train_loss_bc": 0.028580352663993835, "train_loss_llm": 0.0, "grad_norm": 0.10675700008869171, "global_step": 29199, "epoch": 811, "lr": 0.001} +{"train_loss": 0.025878874585032463, "train_loss_bc": 0.025862056761980057, "train_loss_llm": 0.0016818046569824219, "grad_norm": 0.13047976791858673, "global_step": 29200, "epoch": 811, "lr": 0.001} +{"train_loss": 0.036178261041641235, "train_loss_bc": 0.036178261041641235, "train_loss_llm": 0.0, "grad_norm": 0.04853121191263199, "global_step": 29201, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03162049129605293, "train_loss_bc": 0.03162049129605293, "train_loss_llm": 0.0, "grad_norm": 0.06416940689086914, "global_step": 29202, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03158955276012421, "train_loss_bc": 0.03158955276012421, "train_loss_llm": 0.0, "grad_norm": 0.0764230266213417, "global_step": 29203, "epoch": 811, "lr": 0.001} +{"train_loss": 0.030234523117542267, "train_loss_bc": 0.030228322371840477, "train_loss_llm": 0.0006201565265655518, "grad_norm": 0.08579998463392258, "global_step": 29204, "epoch": 811, "lr": 0.001} +{"train_loss": 0.025579581037163734, "train_loss_bc": 0.025579581037163734, "train_loss_llm": 0.0, "grad_norm": 0.030955951660871506, "global_step": 29205, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03283914178609848, "train_loss_bc": 0.03283914178609848, "train_loss_llm": 0.0, "grad_norm": 0.05162480100989342, "global_step": 29206, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03123556077480316, "train_loss_bc": 0.03123556077480316, "train_loss_llm": 0.0, "grad_norm": 0.0713425874710083, "global_step": 29207, "epoch": 811, "lr": 0.001} +{"train_loss": 0.022785266861319542, "train_loss_bc": 0.022785266861319542, "train_loss_llm": 0.0, "grad_norm": 0.09413085132837296, "global_step": 29208, "epoch": 811, "lr": 0.001} +{"train_loss": 0.0377105250954628, "train_loss_bc": 0.0377105250954628, "train_loss_llm": 0.0, "grad_norm": 0.03747216984629631, "global_step": 29209, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03683345392346382, "train_loss_bc": 0.03683345392346382, "train_loss_llm": 0.0, "grad_norm": 0.058959200978279114, "global_step": 29210, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03267364203929901, "train_loss_bc": 0.03267364203929901, "train_loss_llm": 0.0, "grad_norm": 0.08116352558135986, "global_step": 29211, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03612137958407402, "train_loss_bc": 0.03612137958407402, "train_loss_llm": 0.0, "grad_norm": 0.11103077232837677, "global_step": 29212, "epoch": 811, "lr": 0.001} +{"train_loss": 0.026206405833363533, "train_loss_bc": 0.026206405833363533, "train_loss_llm": 0.0, "grad_norm": 0.0326089970767498, "global_step": 29213, "epoch": 811, "lr": 0.001} +{"train_loss": 0.027418682351708412, "train_loss_bc": 0.027418682351708412, "train_loss_llm": 0.0, "grad_norm": 0.0629039853811264, "global_step": 29214, "epoch": 811, "lr": 0.001} +{"train_loss": 0.029953105375170708, "train_loss_bc": 0.029948171228170395, "train_loss_llm": 0.0004933774471282959, "grad_norm": 0.08471883833408356, "global_step": 29215, "epoch": 811, "lr": 0.001} +{"train_loss": 0.02962312661111355, "train_loss_bc": 0.02962312661111355, "train_loss_llm": 0.0, "grad_norm": 0.11994974315166473, "global_step": 29216, "epoch": 811, "lr": 0.001} +{"train_loss": 0.026811573654413223, "train_loss_bc": 0.026811573654413223, "train_loss_llm": 0.0, "grad_norm": 0.0400225855410099, "global_step": 29217, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03359038382768631, "train_loss_bc": 0.03359038382768631, "train_loss_llm": 0.0, "grad_norm": 0.06537014991044998, "global_step": 29218, "epoch": 811, "lr": 0.001} +{"train_loss": 0.029735127463936806, "train_loss_bc": 0.029720796272158623, "train_loss_llm": 0.0014331936836242676, "grad_norm": 0.10026591271162033, "global_step": 29219, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03622022271156311, "train_loss_bc": 0.03622022271156311, "train_loss_llm": 0.0, "grad_norm": 0.13346098363399506, "global_step": 29220, "epoch": 811, "lr": 0.001} +{"train_loss": 0.025300100445747375, "train_loss_bc": 0.025288177654147148, "train_loss_llm": 0.0011923015117645264, "grad_norm": 0.03574676811695099, "global_step": 29221, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03026735410094261, "train_loss_bc": 0.030265824869275093, "train_loss_llm": 0.00015288591384887695, "grad_norm": 0.05838337168097496, "global_step": 29222, "epoch": 811, "lr": 0.001} +{"train_loss": 0.034094616770744324, "train_loss_bc": 0.034094616770744324, "train_loss_llm": 0.0, "grad_norm": 0.08029787242412567, "global_step": 29223, "epoch": 811, "lr": 0.001} +{"train_loss": 0.028370028361678123, "train_loss_bc": 0.028370028361678123, "train_loss_llm": 0.0, "grad_norm": 0.10611677169799805, "global_step": 29224, "epoch": 811, "lr": 0.001} +{"train_loss": 0.02550249919295311, "train_loss_bc": 0.02549188956618309, "train_loss_llm": 0.00106087327003479, "grad_norm": 0.03301433101296425, "global_step": 29225, "epoch": 811, "lr": 0.001} +{"train_loss": 0.033475361764431, "train_loss_bc": 0.033475361764431, "train_loss_llm": 0.0, "grad_norm": 0.04821479693055153, "global_step": 29226, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03106554038822651, "train_loss_bc": 0.03106554038822651, "train_loss_llm": 0.0, "grad_norm": 0.062297072261571884, "global_step": 29227, "epoch": 811, "lr": 0.001} +{"train_loss": 0.029131516814231873, "train_loss_bc": 0.029131516814231873, "train_loss_llm": 0.0, "grad_norm": 0.07508181035518646, "global_step": 29228, "epoch": 811, "lr": 0.001} +{"train_loss": 0.024821383878588676, "train_loss_bc": 0.024808667600154877, "train_loss_llm": 0.0012715458869934082, "grad_norm": 0.032396797090768814, "global_step": 29229, "epoch": 811, "lr": 0.001} +{"train_loss": 0.03290595859289169, "train_loss_bc": 0.03288280963897705, "train_loss_llm": 0.0023148655891418457, "grad_norm": 0.059718817472457886, "global_step": 29230, "epoch": 811, "lr": 0.001} +{"train_loss": 0.030463375720298953, "train_loss_bc": 0.031238479539752007, "train_loss_llm": 0.002309352159500122, "grad_norm": 0.08190880715847015, "global_step": 29231, "epoch": 811, "lr": 0.001, "val_loss": 0.09709177911281586} +{"train_loss": 0.03107193484902382, "train_loss_bc": 0.03107193484902382, "train_loss_llm": 0.0, "grad_norm": 0.1084902286529541, "global_step": 29232, "epoch": 812, "lr": 0.001} +{"train_loss": 0.03831185773015022, "train_loss_bc": 0.03827351704239845, "train_loss_llm": 0.0038340985774993896, "grad_norm": 0.05267470329999924, "global_step": 29233, "epoch": 812, "lr": 0.001} +{"train_loss": 0.02904634177684784, "train_loss_bc": 0.029041560366749763, "train_loss_llm": 0.0004781782627105713, "grad_norm": 0.07780032604932785, "global_step": 29234, "epoch": 812, "lr": 0.001} +{"train_loss": 0.02776719443500042, "train_loss_bc": 0.02776719443500042, "train_loss_llm": 0.0, "grad_norm": 0.09813575446605682, "global_step": 29235, "epoch": 812, "lr": 0.001} +{"train_loss": 0.028816035017371178, "train_loss_bc": 0.028816035017371178, "train_loss_llm": 0.0, "grad_norm": 0.11469924449920654, "global_step": 29236, "epoch": 812, "lr": 0.001} +{"train_loss": 0.03236302360892296, "train_loss_bc": 0.03236302360892296, "train_loss_llm": 0.0, "grad_norm": 0.03892168775200844, "global_step": 29237, "epoch": 812, "lr": 0.001} +{"train_loss": 0.023205608129501343, "train_loss_bc": 0.023205608129501343, "train_loss_llm": 0.0, "grad_norm": 0.06538791954517365, "global_step": 29238, "epoch": 812, "lr": 0.001} +{"train_loss": 0.032809048891067505, "train_loss_bc": 0.032809048891067505, "train_loss_llm": 0.0, "grad_norm": 0.0901665985584259, "global_step": 29239, "epoch": 812, "lr": 0.001} +{"train_loss": 0.031202057376503944, "train_loss_bc": 0.031202057376503944, "train_loss_llm": 0.0, "grad_norm": 0.13510209321975708, "global_step": 29240, "epoch": 812, "lr": 0.001} +{"train_loss": 0.03156118839979172, "train_loss_bc": 0.03156062215566635, "train_loss_llm": 5.6743621826171875e-05, "grad_norm": 0.03773412108421326, "global_step": 29241, "epoch": 812, "lr": 0.001} +{"train_loss": 0.03276689723134041, "train_loss_bc": 0.03276689723134041, "train_loss_llm": 0.0, "grad_norm": 0.07512591779232025, "global_step": 29242, "epoch": 812, "lr": 0.001} +{"train_loss": 0.03064076043665409, "train_loss_bc": 0.030622612684965134, "train_loss_llm": 0.0018146932125091553, "grad_norm": 0.09555546194314957, "global_step": 29243, "epoch": 812, "lr": 0.001} +{"train_loss": 0.023745616897940636, "train_loss_bc": 0.023745616897940636, "train_loss_llm": 0.0, "grad_norm": 0.12224438041448593, "global_step": 29244, "epoch": 812, "lr": 0.001} +{"train_loss": 0.028607767075300217, "train_loss_bc": 0.028607767075300217, "train_loss_llm": 0.0, "grad_norm": 0.03512980788946152, "global_step": 29245, "epoch": 812, "lr": 0.001} +{"train_loss": 0.026905689388513565, "train_loss_bc": 0.026905689388513565, "train_loss_llm": 0.0, "grad_norm": 0.04937926307320595, "global_step": 29246, "epoch": 812, "lr": 0.001} +{"train_loss": 0.028229571878910065, "train_loss_bc": 0.028229571878910065, "train_loss_llm": 0.0, "grad_norm": 0.06553517282009125, "global_step": 29247, "epoch": 812, "lr": 0.001} +{"train_loss": 0.032619740813970566, "train_loss_bc": 0.032617658376693726, "train_loss_llm": 0.00020840764045715332, "grad_norm": 0.09471356123685837, "global_step": 29248, "epoch": 812, "lr": 0.001} +{"train_loss": 0.025953639298677444, "train_loss_bc": 0.025953639298677444, "train_loss_llm": 0.0, "grad_norm": 0.03522932156920433, "global_step": 29249, "epoch": 812, "lr": 0.001} +{"train_loss": 0.03520822152495384, "train_loss_bc": 0.03520822152495384, "train_loss_llm": 0.0, "grad_norm": 0.06193504482507706, "global_step": 29250, "epoch": 812, "lr": 0.001} +{"train_loss": 0.028350919485092163, "train_loss_bc": 0.028350919485092163, "train_loss_llm": 0.0, "grad_norm": 0.07718770951032639, "global_step": 29251, "epoch": 812, "lr": 0.001} +{"train_loss": 0.031835176050662994, "train_loss_bc": 0.031835176050662994, "train_loss_llm": 0.0, "grad_norm": 0.09477008879184723, "global_step": 29252, "epoch": 812, "lr": 0.001} +{"train_loss": 0.030678821727633476, "train_loss_bc": 0.030678821727633476, "train_loss_llm": 0.0, "grad_norm": 0.04533102363348007, "global_step": 29253, "epoch": 812, "lr": 0.001} +{"train_loss": 0.033536896109580994, "train_loss_bc": 0.033536896109580994, "train_loss_llm": 0.0, "grad_norm": 0.06490109115839005, "global_step": 29254, "epoch": 812, "lr": 0.001} +{"train_loss": 0.027035245671868324, "train_loss_bc": 0.027035245671868324, "train_loss_llm": 0.0, "grad_norm": 0.09300362318754196, "global_step": 29255, "epoch": 812, "lr": 0.001} +{"train_loss": 0.03315840661525726, "train_loss_bc": 0.03315840661525726, "train_loss_llm": 0.0, "grad_norm": 0.1301184594631195, "global_step": 29256, "epoch": 812, "lr": 0.001} +{"train_loss": 0.03196483105421066, "train_loss_bc": 0.03195714205503464, "train_loss_llm": 0.0007689297199249268, "grad_norm": 0.03836425766348839, "global_step": 29257, "epoch": 812, "lr": 0.001} +{"train_loss": 0.02930481918156147, "train_loss_bc": 0.02930481918156147, "train_loss_llm": 0.0, "grad_norm": 0.06093332916498184, "global_step": 29258, "epoch": 812, "lr": 0.001} +{"train_loss": 0.036374542862176895, "train_loss_bc": 0.03637368232011795, "train_loss_llm": 8.612871170043945e-05, "grad_norm": 0.09709338843822479, "global_step": 29259, "epoch": 812, "lr": 0.001} +{"train_loss": 0.03377770632505417, "train_loss_bc": 0.03377770632505417, "train_loss_llm": 0.0, "grad_norm": 0.12135385721921921, "global_step": 29260, "epoch": 812, "lr": 0.001} +{"train_loss": 0.03384333848953247, "train_loss_bc": 0.03384333848953247, "train_loss_llm": 0.0, "grad_norm": 0.034139327704906464, "global_step": 29261, "epoch": 812, "lr": 0.001} +{"train_loss": 0.027100209146738052, "train_loss_bc": 0.027100209146738052, "train_loss_llm": 0.0, "grad_norm": 0.05794890597462654, "global_step": 29262, "epoch": 812, "lr": 0.001} +{"train_loss": 0.028432175517082214, "train_loss_bc": 0.028432175517082214, "train_loss_llm": 0.0, "grad_norm": 0.08746230602264404, "global_step": 29263, "epoch": 812, "lr": 0.001} +{"train_loss": 0.030759789049625397, "train_loss_bc": 0.030759789049625397, "train_loss_llm": 0.0, "grad_norm": 0.10294614732265472, "global_step": 29264, "epoch": 812, "lr": 0.001} +{"train_loss": 0.029160425066947937, "train_loss_bc": 0.029160425066947937, "train_loss_llm": 0.0, "grad_norm": 0.03737735003232956, "global_step": 29265, "epoch": 812, "lr": 0.001} +{"train_loss": 0.028493495658040047, "train_loss_bc": 0.028493495658040047, "train_loss_llm": 0.0, "grad_norm": 0.06448061764240265, "global_step": 29266, "epoch": 812, "lr": 0.001} +{"train_loss": 0.030594345854802266, "train_loss_bc": 0.03675590455532074, "train_loss_llm": 0.00015544891357421875, "grad_norm": 0.10362675040960312, "global_step": 29267, "epoch": 812, "lr": 0.001, "val_loss": 0.09662178158760071} +{"train_loss": 0.02851119078695774, "train_loss_bc": 0.02851119078695774, "train_loss_llm": 0.0, "grad_norm": 0.11093951761722565, "global_step": 29268, "epoch": 813, "lr": 0.001} +{"train_loss": 0.030695276334881783, "train_loss_bc": 0.030695276334881783, "train_loss_llm": 0.0, "grad_norm": 0.03844166174530983, "global_step": 29269, "epoch": 813, "lr": 0.001} +{"train_loss": 0.033086325973272324, "train_loss_bc": 0.033086325973272324, "train_loss_llm": 0.0, "grad_norm": 0.0706654042005539, "global_step": 29270, "epoch": 813, "lr": 0.001} +{"train_loss": 0.02760213427245617, "train_loss_bc": 0.02760213427245617, "train_loss_llm": 0.0, "grad_norm": 0.0992969274520874, "global_step": 29271, "epoch": 813, "lr": 0.001} +{"train_loss": 0.02802177704870701, "train_loss_bc": 0.02802177704870701, "train_loss_llm": 0.0, "grad_norm": 0.11624413728713989, "global_step": 29272, "epoch": 813, "lr": 0.001} +{"train_loss": 0.026260539889335632, "train_loss_bc": 0.026260539889335632, "train_loss_llm": 0.0, "grad_norm": 0.036121003329753876, "global_step": 29273, "epoch": 813, "lr": 0.001} +{"train_loss": 0.03458729758858681, "train_loss_bc": 0.03458729758858681, "train_loss_llm": 0.0, "grad_norm": 0.07064316421747208, "global_step": 29274, "epoch": 813, "lr": 0.001} +{"train_loss": 0.03384672477841377, "train_loss_bc": 0.033842843025922775, "train_loss_llm": 0.00038808584213256836, "grad_norm": 0.10951767861843109, "global_step": 29275, "epoch": 813, "lr": 0.001} +{"train_loss": 0.030543077737092972, "train_loss_bc": 0.030543077737092972, "train_loss_llm": 0.0, "grad_norm": 0.13164615631103516, "global_step": 29276, "epoch": 813, "lr": 0.001} +{"train_loss": 0.029267622157931328, "train_loss_bc": 0.029265012592077255, "train_loss_llm": 0.00026094913482666016, "grad_norm": 0.03702298551797867, "global_step": 29277, "epoch": 813, "lr": 0.001} +{"train_loss": 0.03262043744325638, "train_loss_bc": 0.032603953033685684, "train_loss_llm": 0.0016482770442962646, "grad_norm": 0.0580386221408844, "global_step": 29278, "epoch": 813, "lr": 0.001} +{"train_loss": 0.028112830594182014, "train_loss_bc": 0.028112830594182014, "train_loss_llm": 0.0, "grad_norm": 0.08200375735759735, "global_step": 29279, "epoch": 813, "lr": 0.001} +{"train_loss": 0.034459538757801056, "train_loss_bc": 0.034459538757801056, "train_loss_llm": 0.0, "grad_norm": 0.09790436923503876, "global_step": 29280, "epoch": 813, "lr": 0.001} +{"train_loss": 0.026156466454267502, "train_loss_bc": 0.026156466454267502, "train_loss_llm": 0.0, "grad_norm": 0.03242835775017738, "global_step": 29281, "epoch": 813, "lr": 0.001} +{"train_loss": 0.034896980971097946, "train_loss_bc": 0.034896980971097946, "train_loss_llm": 0.0, "grad_norm": 0.052764859050512314, "global_step": 29282, "epoch": 813, "lr": 0.001} +{"train_loss": 0.030670376494526863, "train_loss_bc": 0.030670376494526863, "train_loss_llm": 0.0, "grad_norm": 0.07758735120296478, "global_step": 29283, "epoch": 813, "lr": 0.001} +{"train_loss": 0.03294295072555542, "train_loss_bc": 0.03294295072555542, "train_loss_llm": 0.0, "grad_norm": 0.09773080050945282, "global_step": 29284, "epoch": 813, "lr": 0.001} +{"train_loss": 0.02673366665840149, "train_loss_bc": 0.02673366665840149, "train_loss_llm": 0.0, "grad_norm": 0.04616605490446091, "global_step": 29285, "epoch": 813, "lr": 0.001} +{"train_loss": 0.03188344091176987, "train_loss_bc": 0.03188344091176987, "train_loss_llm": 0.0, "grad_norm": 0.07140829414129257, "global_step": 29286, "epoch": 813, "lr": 0.001} +{"train_loss": 0.026126621291041374, "train_loss_bc": 0.026109354570508003, "train_loss_llm": 0.001726597547531128, "grad_norm": 0.0996050015091896, "global_step": 29287, "epoch": 813, "lr": 0.001} +{"train_loss": 0.02768627740442753, "train_loss_bc": 0.02768627740442753, "train_loss_llm": 0.0, "grad_norm": 0.1295873522758484, "global_step": 29288, "epoch": 813, "lr": 0.001} +{"train_loss": 0.0342123880982399, "train_loss_bc": 0.0342123880982399, "train_loss_llm": 0.0, "grad_norm": 0.05901462212204933, "global_step": 29289, "epoch": 813, "lr": 0.001} +{"train_loss": 0.030539846047759056, "train_loss_bc": 0.030539846047759056, "train_loss_llm": 0.0, "grad_norm": 0.08552296459674835, "global_step": 29290, "epoch": 813, "lr": 0.001} +{"train_loss": 0.030008910223841667, "train_loss_bc": 0.030008910223841667, "train_loss_llm": 0.0, "grad_norm": 0.10436464101076126, "global_step": 29291, "epoch": 813, "lr": 0.001} +{"train_loss": 0.029370326548814774, "train_loss_bc": 0.029370326548814774, "train_loss_llm": 0.0, "grad_norm": 0.12254904955625534, "global_step": 29292, "epoch": 813, "lr": 0.001} +{"train_loss": 0.02994655817747116, "train_loss_bc": 0.02992420829832554, "train_loss_llm": 0.0022349655628204346, "grad_norm": 0.037767354398965836, "global_step": 29293, "epoch": 813, "lr": 0.001} +{"train_loss": 0.026541227474808693, "train_loss_bc": 0.02653459832072258, "train_loss_llm": 0.0006628930568695068, "grad_norm": 0.06216247379779816, "global_step": 29294, "epoch": 813, "lr": 0.001} +{"train_loss": 0.031081002205610275, "train_loss_bc": 0.031077666208148003, "train_loss_llm": 0.00033351778984069824, "grad_norm": 0.08192302286624908, "global_step": 29295, "epoch": 813, "lr": 0.001} +{"train_loss": 0.033022720366716385, "train_loss_bc": 0.033004291355609894, "train_loss_llm": 0.001842796802520752, "grad_norm": 0.10663963109254837, "global_step": 29296, "epoch": 813, "lr": 0.001} +{"train_loss": 0.03224605321884155, "train_loss_bc": 0.03224605321884155, "train_loss_llm": 0.0, "grad_norm": 0.035018190741539, "global_step": 29297, "epoch": 813, "lr": 0.001} +{"train_loss": 0.032375823706388474, "train_loss_bc": 0.03237244859337807, "train_loss_llm": 0.0003375113010406494, "grad_norm": 0.05251266434788704, "global_step": 29298, "epoch": 813, "lr": 0.001} +{"train_loss": 0.0269139613956213, "train_loss_bc": 0.0269139613956213, "train_loss_llm": 0.0, "grad_norm": 0.0656731128692627, "global_step": 29299, "epoch": 813, "lr": 0.001} +{"train_loss": 0.030847789719700813, "train_loss_bc": 0.030847789719700813, "train_loss_llm": 0.0, "grad_norm": 0.08882035315036774, "global_step": 29300, "epoch": 813, "lr": 0.001} +{"train_loss": 0.030436018481850624, "train_loss_bc": 0.030436018481850624, "train_loss_llm": 0.0, "grad_norm": 0.040430810302495956, "global_step": 29301, "epoch": 813, "lr": 0.001} +{"train_loss": 0.03209999203681946, "train_loss_bc": 0.03209999203681946, "train_loss_llm": 0.0, "grad_norm": 0.06591576337814331, "global_step": 29302, "epoch": 813, "lr": 0.001} +{"train_loss": 0.030416083780841693, "train_loss_bc": 0.030624844133853912, "train_loss_llm": 0.0, "grad_norm": 0.09565368294715881, "global_step": 29303, "epoch": 813, "lr": 0.001, "val_loss": 0.09774106740951538} +{"train_loss": 0.027161957696080208, "train_loss_bc": 0.027155321091413498, "train_loss_llm": 0.0006637275218963623, "grad_norm": 0.11392819136381149, "global_step": 29304, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03348230570554733, "train_loss_bc": 0.03348230570554733, "train_loss_llm": 0.0, "grad_norm": 0.04502540081739426, "global_step": 29305, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03220796212553978, "train_loss_bc": 0.03220796212553978, "train_loss_llm": 0.0, "grad_norm": 0.08625064045190811, "global_step": 29306, "epoch": 814, "lr": 0.001} +{"train_loss": 0.02345459908246994, "train_loss_bc": 0.023429585620760918, "train_loss_llm": 0.002501368522644043, "grad_norm": 0.11160275340080261, "global_step": 29307, "epoch": 814, "lr": 0.001} +{"train_loss": 0.02991366758942604, "train_loss_bc": 0.02991366758942604, "train_loss_llm": 0.0, "grad_norm": 0.13846750557422638, "global_step": 29308, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03066040389239788, "train_loss_bc": 0.03066040389239788, "train_loss_llm": 0.0, "grad_norm": 0.03744295984506607, "global_step": 29309, "epoch": 814, "lr": 0.001} +{"train_loss": 0.02905876748263836, "train_loss_bc": 0.02905876748263836, "train_loss_llm": 0.0, "grad_norm": 0.06368576735258102, "global_step": 29310, "epoch": 814, "lr": 0.001} +{"train_loss": 0.02721482329070568, "train_loss_bc": 0.02721257135272026, "train_loss_llm": 0.00022518634796142578, "grad_norm": 0.08834666013717651, "global_step": 29311, "epoch": 814, "lr": 0.001} +{"train_loss": 0.029737424105405807, "train_loss_bc": 0.029737424105405807, "train_loss_llm": 0.0, "grad_norm": 0.11299588531255722, "global_step": 29312, "epoch": 814, "lr": 0.001} +{"train_loss": 0.02872510440647602, "train_loss_bc": 0.02872510440647602, "train_loss_llm": 0.0, "grad_norm": 0.04427122324705124, "global_step": 29313, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03169158473610878, "train_loss_bc": 0.03169158473610878, "train_loss_llm": 0.0, "grad_norm": 0.07095572352409363, "global_step": 29314, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03371948003768921, "train_loss_bc": 0.03371948003768921, "train_loss_llm": 0.0, "grad_norm": 0.10555681586265564, "global_step": 29315, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03133692219853401, "train_loss_bc": 0.03132067620754242, "train_loss_llm": 0.0016247034072875977, "grad_norm": 0.1330496221780777, "global_step": 29316, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03429631516337395, "train_loss_bc": 0.03429631516337395, "train_loss_llm": 0.0, "grad_norm": 0.04397358372807503, "global_step": 29317, "epoch": 814, "lr": 0.001} +{"train_loss": 0.030700650066137314, "train_loss_bc": 0.030700650066137314, "train_loss_llm": 0.0, "grad_norm": 0.06534335762262344, "global_step": 29318, "epoch": 814, "lr": 0.001} +{"train_loss": 0.033274609595537186, "train_loss_bc": 0.033274609595537186, "train_loss_llm": 0.0, "grad_norm": 0.08263560384511948, "global_step": 29319, "epoch": 814, "lr": 0.001} +{"train_loss": 0.02596728503704071, "train_loss_bc": 0.02596599981188774, "train_loss_llm": 0.00012859702110290527, "grad_norm": 0.0892818421125412, "global_step": 29320, "epoch": 814, "lr": 0.001} +{"train_loss": 0.02755935676395893, "train_loss_bc": 0.02755935676395893, "train_loss_llm": 0.0, "grad_norm": 0.0314527302980423, "global_step": 29321, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03134696185588837, "train_loss_bc": 0.03134696185588837, "train_loss_llm": 0.0, "grad_norm": 0.058626335114240646, "global_step": 29322, "epoch": 814, "lr": 0.001} +{"train_loss": 0.018901720643043518, "train_loss_bc": 0.01888510026037693, "train_loss_llm": 0.0016620457172393799, "grad_norm": 0.06519734114408493, "global_step": 29323, "epoch": 814, "lr": 0.001} +{"train_loss": 0.029941672459244728, "train_loss_bc": 0.029941672459244728, "train_loss_llm": 0.0, "grad_norm": 0.08516543358564377, "global_step": 29324, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03506888821721077, "train_loss_bc": 0.035067178308963776, "train_loss_llm": 0.0001710653305053711, "grad_norm": 0.04388962686061859, "global_step": 29325, "epoch": 814, "lr": 0.001} +{"train_loss": 0.02985268086194992, "train_loss_bc": 0.029824376106262207, "train_loss_llm": 0.002830415964126587, "grad_norm": 0.07321517169475555, "global_step": 29326, "epoch": 814, "lr": 0.001} +{"train_loss": 0.032629743218421936, "train_loss_bc": 0.03261102735996246, "train_loss_llm": 0.001871645450592041, "grad_norm": 0.08950741589069366, "global_step": 29327, "epoch": 814, "lr": 0.001} +{"train_loss": 0.030542975291609764, "train_loss_bc": 0.030542975291609764, "train_loss_llm": 0.0, "grad_norm": 0.10124674439430237, "global_step": 29328, "epoch": 814, "lr": 0.001} +{"train_loss": 0.028417512774467468, "train_loss_bc": 0.028417512774467468, "train_loss_llm": 0.0, "grad_norm": 0.03815764561295509, "global_step": 29329, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03530314192175865, "train_loss_bc": 0.03530314192175865, "train_loss_llm": 0.0, "grad_norm": 0.0624205619096756, "global_step": 29330, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03040527179837227, "train_loss_bc": 0.03040527179837227, "train_loss_llm": 0.0, "grad_norm": 0.09150077402591705, "global_step": 29331, "epoch": 814, "lr": 0.001} +{"train_loss": 0.032697368413209915, "train_loss_bc": 0.032697368413209915, "train_loss_llm": 0.0, "grad_norm": 0.12142430990934372, "global_step": 29332, "epoch": 814, "lr": 0.001} +{"train_loss": 0.030697081238031387, "train_loss_bc": 0.030697081238031387, "train_loss_llm": 0.0, "grad_norm": 0.03610961139202118, "global_step": 29333, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03221312537789345, "train_loss_bc": 0.03221312537789345, "train_loss_llm": 0.0, "grad_norm": 0.053903691470623016, "global_step": 29334, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03278150409460068, "train_loss_bc": 0.03278150409460068, "train_loss_llm": 0.0, "grad_norm": 0.0854019746184349, "global_step": 29335, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03156466409564018, "train_loss_bc": 0.03156466409564018, "train_loss_llm": 0.0, "grad_norm": 0.10459858924150467, "global_step": 29336, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03339190408587456, "train_loss_bc": 0.03339190408587456, "train_loss_llm": 0.0, "grad_norm": 0.0377507284283638, "global_step": 29337, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03466944769024849, "train_loss_bc": 0.03466944769024849, "train_loss_llm": 0.0, "grad_norm": 0.057163599878549576, "global_step": 29338, "epoch": 814, "lr": 0.001} +{"train_loss": 0.03054012456494901, "train_loss_bc": 0.0288508553057909, "train_loss_llm": 0.00047454237937927246, "grad_norm": 0.07769758254289627, "global_step": 29339, "epoch": 814, "lr": 0.001, "val_loss": 0.09894700348377228} +{"train_loss": 0.02854035794734955, "train_loss_bc": 0.028520340099930763, "train_loss_llm": 0.0020017623901367188, "grad_norm": 0.10458274185657501, "global_step": 29340, "epoch": 815, "lr": 0.001} +{"train_loss": 0.031699176877737045, "train_loss_bc": 0.031699176877737045, "train_loss_llm": 0.0, "grad_norm": 0.03140613064169884, "global_step": 29341, "epoch": 815, "lr": 0.001} +{"train_loss": 0.028856080025434494, "train_loss_bc": 0.028856080025434494, "train_loss_llm": 0.0, "grad_norm": 0.05349388346076012, "global_step": 29342, "epoch": 815, "lr": 0.001} +{"train_loss": 0.028636150062084198, "train_loss_bc": 0.028628280386328697, "train_loss_llm": 0.0007870197296142578, "grad_norm": 0.07189136743545532, "global_step": 29343, "epoch": 815, "lr": 0.001} +{"train_loss": 0.0327119305729866, "train_loss_bc": 0.0327119305729866, "train_loss_llm": 0.0, "grad_norm": 0.10252104699611664, "global_step": 29344, "epoch": 815, "lr": 0.001} +{"train_loss": 0.03196306526660919, "train_loss_bc": 0.03196306526660919, "train_loss_llm": 0.0, "grad_norm": 0.03933512419462204, "global_step": 29345, "epoch": 815, "lr": 0.001} +{"train_loss": 0.028316231444478035, "train_loss_bc": 0.028316231444478035, "train_loss_llm": 0.0, "grad_norm": 0.052216120064258575, "global_step": 29346, "epoch": 815, "lr": 0.001} +{"train_loss": 0.03019401617348194, "train_loss_bc": 0.03019401617348194, "train_loss_llm": 0.0, "grad_norm": 0.07747521996498108, "global_step": 29347, "epoch": 815, "lr": 0.001} +{"train_loss": 0.03327168896794319, "train_loss_bc": 0.03325827419757843, "train_loss_llm": 0.001341402530670166, "grad_norm": 0.11437824368476868, "global_step": 29348, "epoch": 815, "lr": 0.001} +{"train_loss": 0.022318238392472267, "train_loss_bc": 0.02231476828455925, "train_loss_llm": 0.0003470778465270996, "grad_norm": 0.03714314475655556, "global_step": 29349, "epoch": 815, "lr": 0.001} +{"train_loss": 0.030986472964286804, "train_loss_bc": 0.030986472964286804, "train_loss_llm": 0.0, "grad_norm": 0.07145923376083374, "global_step": 29350, "epoch": 815, "lr": 0.001} +{"train_loss": 0.03114052303135395, "train_loss_bc": 0.03114052303135395, "train_loss_llm": 0.0, "grad_norm": 0.0947151929140091, "global_step": 29351, "epoch": 815, "lr": 0.001} +{"train_loss": 0.026921074837446213, "train_loss_bc": 0.026916326954960823, "train_loss_llm": 0.0004748702049255371, "grad_norm": 0.1155669242143631, "global_step": 29352, "epoch": 815, "lr": 0.001} +{"train_loss": 0.027161195874214172, "train_loss_bc": 0.027143370360136032, "train_loss_llm": 0.0017825067043304443, "grad_norm": 0.03867694362998009, "global_step": 29353, "epoch": 815, "lr": 0.001} +{"train_loss": 0.029878444969654083, "train_loss_bc": 0.029878444969654083, "train_loss_llm": 0.0, "grad_norm": 0.06427519768476486, "global_step": 29354, "epoch": 815, "lr": 0.001} +{"train_loss": 0.027315884828567505, "train_loss_bc": 0.027315884828567505, "train_loss_llm": 0.0, "grad_norm": 0.08349438011646271, "global_step": 29355, "epoch": 815, "lr": 0.001} +{"train_loss": 0.032310351729393005, "train_loss_bc": 0.032310351729393005, "train_loss_llm": 0.0, "grad_norm": 0.09424372017383575, "global_step": 29356, "epoch": 815, "lr": 0.001} +{"train_loss": 0.03619518131017685, "train_loss_bc": 0.03619518131017685, "train_loss_llm": 0.0, "grad_norm": 0.03459310531616211, "global_step": 29357, "epoch": 815, "lr": 0.001} +{"train_loss": 0.02873372845351696, "train_loss_bc": 0.02873372845351696, "train_loss_llm": 0.0, "grad_norm": 0.05855725705623627, "global_step": 29358, "epoch": 815, "lr": 0.001} +{"train_loss": 0.03602949529886246, "train_loss_bc": 0.03602949529886246, "train_loss_llm": 0.0, "grad_norm": 0.08209740370512009, "global_step": 29359, "epoch": 815, "lr": 0.001} +{"train_loss": 0.029061241075396538, "train_loss_bc": 0.029061241075396538, "train_loss_llm": 0.0, "grad_norm": 0.09700392931699753, "global_step": 29360, "epoch": 815, "lr": 0.001} +{"train_loss": 0.030964205041527748, "train_loss_bc": 0.030964205041527748, "train_loss_llm": 0.0, "grad_norm": 0.036912113428115845, "global_step": 29361, "epoch": 815, "lr": 0.001} +{"train_loss": 0.0365765281021595, "train_loss_bc": 0.0365765281021595, "train_loss_llm": 0.0, "grad_norm": 0.06619308143854141, "global_step": 29362, "epoch": 815, "lr": 0.001} +{"train_loss": 0.0292122233659029, "train_loss_bc": 0.0292122233659029, "train_loss_llm": 0.0, "grad_norm": 0.07851111143827438, "global_step": 29363, "epoch": 815, "lr": 0.001} +{"train_loss": 0.027205759659409523, "train_loss_bc": 0.0271906740963459, "train_loss_llm": 0.001508563756942749, "grad_norm": 0.09107988327741623, "global_step": 29364, "epoch": 815, "lr": 0.001} +{"train_loss": 0.03169951215386391, "train_loss_bc": 0.03169951215386391, "train_loss_llm": 0.0, "grad_norm": 0.03490997850894928, "global_step": 29365, "epoch": 815, "lr": 0.001} +{"train_loss": 0.030976369976997375, "train_loss_bc": 0.030964359641075134, "train_loss_llm": 0.0012010633945465088, "grad_norm": 0.054828841239213943, "global_step": 29366, "epoch": 815, "lr": 0.001} +{"train_loss": 0.030664829537272453, "train_loss_bc": 0.030664829537272453, "train_loss_llm": 0.0, "grad_norm": 0.07679371535778046, "global_step": 29367, "epoch": 815, "lr": 0.001} +{"train_loss": 0.03589911386370659, "train_loss_bc": 0.03589911386370659, "train_loss_llm": 0.0, "grad_norm": 0.10479248315095901, "global_step": 29368, "epoch": 815, "lr": 0.001} +{"train_loss": 0.02752954140305519, "train_loss_bc": 0.02751830220222473, "train_loss_llm": 0.001123875379562378, "grad_norm": 0.03438815847039223, "global_step": 29369, "epoch": 815, "lr": 0.001} +{"train_loss": 0.032034412026405334, "train_loss_bc": 0.032034412026405334, "train_loss_llm": 0.0, "grad_norm": 0.06029653176665306, "global_step": 29370, "epoch": 815, "lr": 0.001} +{"train_loss": 0.027452947571873665, "train_loss_bc": 0.027452947571873665, "train_loss_llm": 0.0, "grad_norm": 0.07983633130788803, "global_step": 29371, "epoch": 815, "lr": 0.001} +{"train_loss": 0.029261218383908272, "train_loss_bc": 0.029261218383908272, "train_loss_llm": 0.0, "grad_norm": 0.10273739695549011, "global_step": 29372, "epoch": 815, "lr": 0.001} +{"train_loss": 0.030040903016924858, "train_loss_bc": 0.030025657266378403, "train_loss_llm": 0.0015246570110321045, "grad_norm": 0.034385066479444504, "global_step": 29373, "epoch": 815, "lr": 0.001} +{"train_loss": 0.027295755222439766, "train_loss_bc": 0.027295755222439766, "train_loss_llm": 0.0, "grad_norm": 0.047795891761779785, "global_step": 29374, "epoch": 815, "lr": 0.001} +{"train_loss": 0.03036458883434534, "train_loss_bc": 0.03406662866473198, "train_loss_llm": 0.0004718303680419922, "grad_norm": 0.08575648069381714, "global_step": 29375, "epoch": 815, "lr": 0.001, "train/cumulative_reward": 30.6805227440449, "train/mean_score": 13.609116709215167, "train/success_rate": 0.4, "test/cumulative_reward": 37.082648407451536, "test/mean_score": 6.445052279508921, "test/success_rate": 0.14, "val_loss": 0.09761505573987961, "train_action_mse_error": 0.03328398987650871} +{"train_loss": 0.03094596415758133, "train_loss_bc": 0.03094596415758133, "train_loss_llm": 0.0, "grad_norm": 0.10355724394321442, "global_step": 29376, "epoch": 816, "lr": 0.001} +{"train_loss": 0.029418261721730232, "train_loss_bc": 0.029418261721730232, "train_loss_llm": 0.0, "grad_norm": 0.0360407680273056, "global_step": 29377, "epoch": 816, "lr": 0.001} +{"train_loss": 0.029362168163061142, "train_loss_bc": 0.029362168163061142, "train_loss_llm": 0.0, "grad_norm": 0.06908097863197327, "global_step": 29378, "epoch": 816, "lr": 0.001} +{"train_loss": 0.030373672023415565, "train_loss_bc": 0.0303712859749794, "train_loss_llm": 0.00023868680000305176, "grad_norm": 0.0940885916352272, "global_step": 29379, "epoch": 816, "lr": 0.001} +{"train_loss": 0.03100539930164814, "train_loss_bc": 0.03100539930164814, "train_loss_llm": 0.0, "grad_norm": 0.11343482881784439, "global_step": 29380, "epoch": 816, "lr": 0.001} +{"train_loss": 0.030301759019494057, "train_loss_bc": 0.030301759019494057, "train_loss_llm": 0.0, "grad_norm": 0.03683635592460632, "global_step": 29381, "epoch": 816, "lr": 0.001} +{"train_loss": 0.03372476249933243, "train_loss_bc": 0.03372476249933243, "train_loss_llm": 0.0, "grad_norm": 0.05214304476976395, "global_step": 29382, "epoch": 816, "lr": 0.001} +{"train_loss": 0.029527777805924416, "train_loss_bc": 0.029520977288484573, "train_loss_llm": 0.0006799995899200439, "grad_norm": 0.06766892224550247, "global_step": 29383, "epoch": 816, "lr": 0.001} +{"train_loss": 0.03380659595131874, "train_loss_bc": 0.03380659595131874, "train_loss_llm": 0.0, "grad_norm": 0.0814957544207573, "global_step": 29384, "epoch": 816, "lr": 0.001} +{"train_loss": 0.03017221949994564, "train_loss_bc": 0.03015291504561901, "train_loss_llm": 0.001930534839630127, "grad_norm": 0.03397296369075775, "global_step": 29385, "epoch": 816, "lr": 0.001} +{"train_loss": 0.028663793578743935, "train_loss_bc": 0.028663793578743935, "train_loss_llm": 0.0, "grad_norm": 0.05355251207947731, "global_step": 29386, "epoch": 816, "lr": 0.001} +{"train_loss": 0.025820089504122734, "train_loss_bc": 0.025820089504122734, "train_loss_llm": 0.0, "grad_norm": 0.07372600585222244, "global_step": 29387, "epoch": 816, "lr": 0.001} +{"train_loss": 0.028201978653669357, "train_loss_bc": 0.028201978653669357, "train_loss_llm": 0.0, "grad_norm": 0.09280296415090561, "global_step": 29388, "epoch": 816, "lr": 0.001} +{"train_loss": 0.03376004844903946, "train_loss_bc": 0.03376004844903946, "train_loss_llm": 0.0, "grad_norm": 0.043869879096746445, "global_step": 29389, "epoch": 816, "lr": 0.001} +{"train_loss": 0.03246884420514107, "train_loss_bc": 0.03245517238974571, "train_loss_llm": 0.0013671517372131348, "grad_norm": 0.0749027356505394, "global_step": 29390, "epoch": 816, "lr": 0.001} +{"train_loss": 0.032379861921072006, "train_loss_bc": 0.032379861921072006, "train_loss_llm": 0.0, "grad_norm": 0.09084457904100418, "global_step": 29391, "epoch": 816, "lr": 0.001} +{"train_loss": 0.028938695788383484, "train_loss_bc": 0.028938695788383484, "train_loss_llm": 0.0, "grad_norm": 0.10261338204145432, "global_step": 29392, "epoch": 816, "lr": 0.001} +{"train_loss": 0.029112551361322403, "train_loss_bc": 0.029110396280884743, "train_loss_llm": 0.00021544098854064941, "grad_norm": 0.03662881627678871, "global_step": 29393, "epoch": 816, "lr": 0.001} +{"train_loss": 0.026963459327816963, "train_loss_bc": 0.026963459327816963, "train_loss_llm": 0.0, "grad_norm": 0.05459941551089287, "global_step": 29394, "epoch": 816, "lr": 0.001} +{"train_loss": 0.03302537277340889, "train_loss_bc": 0.03302537277340889, "train_loss_llm": 0.0, "grad_norm": 0.07011695951223373, "global_step": 29395, "epoch": 816, "lr": 0.001} +{"train_loss": 0.030851997435092926, "train_loss_bc": 0.030851997435092926, "train_loss_llm": 0.0, "grad_norm": 0.09285354614257812, "global_step": 29396, "epoch": 816, "lr": 0.001} +{"train_loss": 0.027061864733695984, "train_loss_bc": 0.027061864733695984, "train_loss_llm": 0.0, "grad_norm": 0.037354130297899246, "global_step": 29397, "epoch": 816, "lr": 0.001} +{"train_loss": 0.0295256320387125, "train_loss_bc": 0.0295256320387125, "train_loss_llm": 0.0, "grad_norm": 0.052303168922662735, "global_step": 29398, "epoch": 816, "lr": 0.001} +{"train_loss": 0.034574635326862335, "train_loss_bc": 0.034574635326862335, "train_loss_llm": 0.0, "grad_norm": 0.06579817831516266, "global_step": 29399, "epoch": 816, "lr": 0.001} +{"train_loss": 0.0294809490442276, "train_loss_bc": 0.0294809490442276, "train_loss_llm": 0.0, "grad_norm": 0.08773431181907654, "global_step": 29400, "epoch": 816, "lr": 0.001} +{"train_loss": 0.029026268050074577, "train_loss_bc": 0.029006633907556534, "train_loss_llm": 0.0019633471965789795, "grad_norm": 0.03480209410190582, "global_step": 29401, "epoch": 816, "lr": 0.001} +{"train_loss": 0.02969369664788246, "train_loss_bc": 0.02968594804406166, "train_loss_llm": 0.0007748603820800781, "grad_norm": 0.0509398952126503, "global_step": 29402, "epoch": 816, "lr": 0.001} +{"train_loss": 0.03185523673892021, "train_loss_bc": 0.03185523673892021, "train_loss_llm": 0.0, "grad_norm": 0.07378362119197845, "global_step": 29403, "epoch": 816, "lr": 0.001} +{"train_loss": 0.032119106501340866, "train_loss_bc": 0.032119106501340866, "train_loss_llm": 0.0, "grad_norm": 0.08929280936717987, "global_step": 29404, "epoch": 816, "lr": 0.001} +{"train_loss": 0.029581066220998764, "train_loss_bc": 0.029555262997746468, "train_loss_llm": 0.00258028507232666, "grad_norm": 0.0358210988342762, "global_step": 29405, "epoch": 816, "lr": 0.001} +{"train_loss": 0.022889209911227226, "train_loss_bc": 0.022882157936692238, "train_loss_llm": 0.0007051527500152588, "grad_norm": 0.049708329141139984, "global_step": 29406, "epoch": 816, "lr": 0.001} +{"train_loss": 0.020214462652802467, "train_loss_bc": 0.02020244672894478, "train_loss_llm": 0.001201540231704712, "grad_norm": 0.06755200028419495, "global_step": 29407, "epoch": 816, "lr": 0.001} +{"train_loss": 0.032903846353292465, "train_loss_bc": 0.032903846353292465, "train_loss_llm": 0.0, "grad_norm": 0.06995124369859695, "global_step": 29408, "epoch": 816, "lr": 0.001} +{"train_loss": 0.03428063169121742, "train_loss_bc": 0.03428063169121742, "train_loss_llm": 0.0, "grad_norm": 0.03946293890476227, "global_step": 29409, "epoch": 816, "lr": 0.001} +{"train_loss": 0.030556079000234604, "train_loss_bc": 0.030556079000234604, "train_loss_llm": 0.0, "grad_norm": 0.051523607224226, "global_step": 29410, "epoch": 816, "lr": 0.001} +{"train_loss": 0.030033271449307602, "train_loss_bc": 0.02857859618961811, "train_loss_llm": 0.0031217336654663086, "grad_norm": 0.08185192197561264, "global_step": 29411, "epoch": 816, "lr": 0.001, "val_loss": 0.09834598004817963} +{"train_loss": 0.0315825492143631, "train_loss_bc": 0.0315825492143631, "train_loss_llm": 0.0, "grad_norm": 0.10158751159906387, "global_step": 29412, "epoch": 817, "lr": 0.001} +{"train_loss": 0.036496035754680634, "train_loss_bc": 0.036496035754680634, "train_loss_llm": 0.0, "grad_norm": 0.05043698847293854, "global_step": 29413, "epoch": 817, "lr": 0.001} +{"train_loss": 0.02958969958126545, "train_loss_bc": 0.029583606868982315, "train_loss_llm": 0.0006093382835388184, "grad_norm": 0.06617304682731628, "global_step": 29414, "epoch": 817, "lr": 0.001} +{"train_loss": 0.0306632649153471, "train_loss_bc": 0.030652878805994987, "train_loss_llm": 0.0010385513305664062, "grad_norm": 0.09098194539546967, "global_step": 29415, "epoch": 817, "lr": 0.001} +{"train_loss": 0.02680128999054432, "train_loss_bc": 0.02680128999054432, "train_loss_llm": 0.0, "grad_norm": 0.10507231950759888, "global_step": 29416, "epoch": 817, "lr": 0.001} +{"train_loss": 0.02869730070233345, "train_loss_bc": 0.02869730070233345, "train_loss_llm": 0.0, "grad_norm": 0.04097124561667442, "global_step": 29417, "epoch": 817, "lr": 0.001} +{"train_loss": 0.028459616005420685, "train_loss_bc": 0.028459616005420685, "train_loss_llm": 0.0, "grad_norm": 0.0468834787607193, "global_step": 29418, "epoch": 817, "lr": 0.001} +{"train_loss": 0.0366288460791111, "train_loss_bc": 0.0366288460791111, "train_loss_llm": 0.0, "grad_norm": 0.07592669129371643, "global_step": 29419, "epoch": 817, "lr": 0.001} +{"train_loss": 0.03285130113363266, "train_loss_bc": 0.032847627997398376, "train_loss_llm": 0.0003673732280731201, "grad_norm": 0.0978197306394577, "global_step": 29420, "epoch": 817, "lr": 0.001} +{"train_loss": 0.030541792511940002, "train_loss_bc": 0.030519872903823853, "train_loss_llm": 0.002191990613937378, "grad_norm": 0.037900205701589584, "global_step": 29421, "epoch": 817, "lr": 0.001} +{"train_loss": 0.03209318593144417, "train_loss_bc": 0.03209318593144417, "train_loss_llm": 0.0, "grad_norm": 0.06592822074890137, "global_step": 29422, "epoch": 817, "lr": 0.001} +{"train_loss": 0.026666680350899696, "train_loss_bc": 0.02666589990258217, "train_loss_llm": 7.811188697814941e-05, "grad_norm": 0.08093533664941788, "global_step": 29423, "epoch": 817, "lr": 0.001} +{"train_loss": 0.026262691244482994, "train_loss_bc": 0.026262691244482994, "train_loss_llm": 0.0, "grad_norm": 0.10627296566963196, "global_step": 29424, "epoch": 817, "lr": 0.001} +{"train_loss": 0.02856810949742794, "train_loss_bc": 0.02856810949742794, "train_loss_llm": 0.0, "grad_norm": 0.03337389975786209, "global_step": 29425, "epoch": 817, "lr": 0.001} +{"train_loss": 0.0318186916410923, "train_loss_bc": 0.031808700412511826, "train_loss_llm": 0.0009991228580474854, "grad_norm": 0.060290805995464325, "global_step": 29426, "epoch": 817, "lr": 0.001} +{"train_loss": 0.03205966576933861, "train_loss_bc": 0.03205377608537674, "train_loss_llm": 0.0005888938903808594, "grad_norm": 0.0958886444568634, "global_step": 29427, "epoch": 817, "lr": 0.001} +{"train_loss": 0.029996449127793312, "train_loss_bc": 0.029996449127793312, "train_loss_llm": 0.0, "grad_norm": 0.11561088263988495, "global_step": 29428, "epoch": 817, "lr": 0.001} +{"train_loss": 0.027905823662877083, "train_loss_bc": 0.027890976518392563, "train_loss_llm": 0.001484692096710205, "grad_norm": 0.03268406167626381, "global_step": 29429, "epoch": 817, "lr": 0.001} +{"train_loss": 0.031713105738162994, "train_loss_bc": 0.031713105738162994, "train_loss_llm": 0.0, "grad_norm": 0.04898008704185486, "global_step": 29430, "epoch": 817, "lr": 0.001} +{"train_loss": 0.03421013429760933, "train_loss_bc": 0.03421013429760933, "train_loss_llm": 0.0, "grad_norm": 0.06319540739059448, "global_step": 29431, "epoch": 817, "lr": 0.001} +{"train_loss": 0.03043142892420292, "train_loss_bc": 0.03043142892420292, "train_loss_llm": 0.0, "grad_norm": 0.09460774809122086, "global_step": 29432, "epoch": 817, "lr": 0.001} +{"train_loss": 0.028220122680068016, "train_loss_bc": 0.028220122680068016, "train_loss_llm": 0.0, "grad_norm": 0.030287358909845352, "global_step": 29433, "epoch": 817, "lr": 0.001} +{"train_loss": 0.030301237478852272, "train_loss_bc": 0.030301237478852272, "train_loss_llm": 0.0, "grad_norm": 0.04872701317071915, "global_step": 29434, "epoch": 817, "lr": 0.001} +{"train_loss": 0.03090752102434635, "train_loss_bc": 0.03090752102434635, "train_loss_llm": 0.0, "grad_norm": 0.0712958425283432, "global_step": 29435, "epoch": 817, "lr": 0.001} +{"train_loss": 0.026912741363048553, "train_loss_bc": 0.026903774589300156, "train_loss_llm": 0.0008966326713562012, "grad_norm": 0.0841611996293068, "global_step": 29436, "epoch": 817, "lr": 0.001} +{"train_loss": 0.028652368113398552, "train_loss_bc": 0.028652368113398552, "train_loss_llm": 0.0, "grad_norm": 0.029101911932229996, "global_step": 29437, "epoch": 817, "lr": 0.001} +{"train_loss": 0.03265846520662308, "train_loss_bc": 0.03265846520662308, "train_loss_llm": 0.0, "grad_norm": 0.0541381798684597, "global_step": 29438, "epoch": 817, "lr": 0.001} +{"train_loss": 0.026406019926071167, "train_loss_bc": 0.026400433853268623, "train_loss_llm": 0.0005586743354797363, "grad_norm": 0.062052927911281586, "global_step": 29439, "epoch": 817, "lr": 0.001} +{"train_loss": 0.026943419128656387, "train_loss_bc": 0.026943419128656387, "train_loss_llm": 0.0, "grad_norm": 0.07495060563087463, "global_step": 29440, "epoch": 817, "lr": 0.001} +{"train_loss": 0.02689484879374504, "train_loss_bc": 0.02689484879374504, "train_loss_llm": 0.0, "grad_norm": 0.02928304672241211, "global_step": 29441, "epoch": 817, "lr": 0.001} +{"train_loss": 0.028823893517255783, "train_loss_bc": 0.028823893517255783, "train_loss_llm": 0.0, "grad_norm": 0.047757141292095184, "global_step": 29442, "epoch": 817, "lr": 0.001} +{"train_loss": 0.030156316235661507, "train_loss_bc": 0.030156316235661507, "train_loss_llm": 0.0, "grad_norm": 0.0672239139676094, "global_step": 29443, "epoch": 817, "lr": 0.001} +{"train_loss": 0.031112346798181534, "train_loss_bc": 0.031101085245609283, "train_loss_llm": 0.0011262297630310059, "grad_norm": 0.09739446640014648, "global_step": 29444, "epoch": 817, "lr": 0.001} +{"train_loss": 0.029559576883912086, "train_loss_bc": 0.029559576883912086, "train_loss_llm": 0.0, "grad_norm": 0.03680996969342232, "global_step": 29445, "epoch": 817, "lr": 0.001} +{"train_loss": 0.03139067813754082, "train_loss_bc": 0.03139067813754082, "train_loss_llm": 0.0, "grad_norm": 0.06309448927640915, "global_step": 29446, "epoch": 817, "lr": 0.001} +{"train_loss": 0.03015291861568888, "train_loss_bc": 0.032527852803468704, "train_loss_llm": 0.0, "grad_norm": 0.08776286244392395, "global_step": 29447, "epoch": 817, "lr": 0.001, "val_loss": 0.09829036891460419} +{"train_loss": 0.023214083164930344, "train_loss_bc": 0.023214083164930344, "train_loss_llm": 0.0, "grad_norm": 0.10599440336227417, "global_step": 29448, "epoch": 818, "lr": 0.001} +{"train_loss": 0.03111073561012745, "train_loss_bc": 0.031099416315555573, "train_loss_llm": 0.0011319220066070557, "grad_norm": 0.030215900391340256, "global_step": 29449, "epoch": 818, "lr": 0.001} +{"train_loss": 0.03172868490219116, "train_loss_bc": 0.0317075252532959, "train_loss_llm": 0.0021158158779144287, "grad_norm": 0.05339187756180763, "global_step": 29450, "epoch": 818, "lr": 0.001} +{"train_loss": 0.02942788228392601, "train_loss_bc": 0.02942788228392601, "train_loss_llm": 0.0, "grad_norm": 0.06415552645921707, "global_step": 29451, "epoch": 818, "lr": 0.001} +{"train_loss": 0.028335461392998695, "train_loss_bc": 0.028335461392998695, "train_loss_llm": 0.0, "grad_norm": 0.08385223895311356, "global_step": 29452, "epoch": 818, "lr": 0.001} +{"train_loss": 0.028762517496943474, "train_loss_bc": 0.028762517496943474, "train_loss_llm": 0.0, "grad_norm": 0.030932608991861343, "global_step": 29453, "epoch": 818, "lr": 0.001} +{"train_loss": 0.03039584867656231, "train_loss_bc": 0.03039584867656231, "train_loss_llm": 0.0, "grad_norm": 0.05522099882364273, "global_step": 29454, "epoch": 818, "lr": 0.001} +{"train_loss": 0.029944786801934242, "train_loss_bc": 0.029944786801934242, "train_loss_llm": 0.0, "grad_norm": 0.0674787163734436, "global_step": 29455, "epoch": 818, "lr": 0.001} +{"train_loss": 0.02781042642891407, "train_loss_bc": 0.02780969999730587, "train_loss_llm": 7.259845733642578e-05, "grad_norm": 0.0856764167547226, "global_step": 29456, "epoch": 818, "lr": 0.001} +{"train_loss": 0.028952958062291145, "train_loss_bc": 0.028952958062291145, "train_loss_llm": 0.0, "grad_norm": 0.03139537572860718, "global_step": 29457, "epoch": 818, "lr": 0.001} +{"train_loss": 0.025058386847376823, "train_loss_bc": 0.025058386847376823, "train_loss_llm": 0.0, "grad_norm": 0.05119850113987923, "global_step": 29458, "epoch": 818, "lr": 0.001} +{"train_loss": 0.027211913838982582, "train_loss_bc": 0.027182690799236298, "train_loss_llm": 0.0029223263263702393, "grad_norm": 0.06825575977563858, "global_step": 29459, "epoch": 818, "lr": 0.001} +{"train_loss": 0.03225915506482124, "train_loss_bc": 0.03225915506482124, "train_loss_llm": 0.0, "grad_norm": 0.08388067036867142, "global_step": 29460, "epoch": 818, "lr": 0.001} +{"train_loss": 0.0270108450204134, "train_loss_bc": 0.0270108450204134, "train_loss_llm": 0.0, "grad_norm": 0.031365059316158295, "global_step": 29461, "epoch": 818, "lr": 0.001} +{"train_loss": 0.03323198854923248, "train_loss_bc": 0.03323198854923248, "train_loss_llm": 0.0, "grad_norm": 0.05427265912294388, "global_step": 29462, "epoch": 818, "lr": 0.001} +{"train_loss": 0.029008187353610992, "train_loss_bc": 0.029008187353610992, "train_loss_llm": 0.0, "grad_norm": 0.06698528677225113, "global_step": 29463, "epoch": 818, "lr": 0.001} +{"train_loss": 0.027449866756796837, "train_loss_bc": 0.027449866756796837, "train_loss_llm": 0.0, "grad_norm": 0.08429175615310669, "global_step": 29464, "epoch": 818, "lr": 0.001} +{"train_loss": 0.029712745919823647, "train_loss_bc": 0.029712745919823647, "train_loss_llm": 0.0, "grad_norm": 0.03975316509604454, "global_step": 29465, "epoch": 818, "lr": 0.001} +{"train_loss": 0.04120137169957161, "train_loss_bc": 0.04120137169957161, "train_loss_llm": 0.0, "grad_norm": 0.07433007657527924, "global_step": 29466, "epoch": 818, "lr": 0.001} +{"train_loss": 0.028734475374221802, "train_loss_bc": 0.028734475374221802, "train_loss_llm": 0.0, "grad_norm": 0.09432466328144073, "global_step": 29467, "epoch": 818, "lr": 0.001} +{"train_loss": 0.03215645253658295, "train_loss_bc": 0.032154690474271774, "train_loss_llm": 0.00017631053924560547, "grad_norm": 0.11094772815704346, "global_step": 29468, "epoch": 818, "lr": 0.001} +{"train_loss": 0.02362252213060856, "train_loss_bc": 0.02362252213060856, "train_loss_llm": 0.0, "grad_norm": 0.03992502763867378, "global_step": 29469, "epoch": 818, "lr": 0.001} +{"train_loss": 0.0275573767721653, "train_loss_bc": 0.0275573767721653, "train_loss_llm": 0.0, "grad_norm": 0.05825076624751091, "global_step": 29470, "epoch": 818, "lr": 0.001} +{"train_loss": 0.03043278120458126, "train_loss_bc": 0.030428124591708183, "train_loss_llm": 0.00046569108963012695, "grad_norm": 0.08029413968324661, "global_step": 29471, "epoch": 818, "lr": 0.001} +{"train_loss": 0.02989189699292183, "train_loss_bc": 0.02989189699292183, "train_loss_llm": 0.0, "grad_norm": 0.09949517995119095, "global_step": 29472, "epoch": 818, "lr": 0.001} +{"train_loss": 0.037033457309007645, "train_loss_bc": 0.037033457309007645, "train_loss_llm": 0.0, "grad_norm": 0.038910482078790665, "global_step": 29473, "epoch": 818, "lr": 0.001} +{"train_loss": 0.03175622969865799, "train_loss_bc": 0.03175622969865799, "train_loss_llm": 0.0, "grad_norm": 0.06311551481485367, "global_step": 29474, "epoch": 818, "lr": 0.001} +{"train_loss": 0.029255541041493416, "train_loss_bc": 0.029248209670186043, "train_loss_llm": 0.0007330775260925293, "grad_norm": 0.08660563826560974, "global_step": 29475, "epoch": 818, "lr": 0.001} +{"train_loss": 0.030565833672881126, "train_loss_bc": 0.03055095113813877, "train_loss_llm": 0.001488327980041504, "grad_norm": 0.10907942056655884, "global_step": 29476, "epoch": 818, "lr": 0.001} +{"train_loss": 0.0307028628885746, "train_loss_bc": 0.0307028628885746, "train_loss_llm": 0.0, "grad_norm": 0.03198234736919403, "global_step": 29477, "epoch": 818, "lr": 0.001} +{"train_loss": 0.032548077404499054, "train_loss_bc": 0.032548077404499054, "train_loss_llm": 0.0, "grad_norm": 0.05293288454413414, "global_step": 29478, "epoch": 818, "lr": 0.001} +{"train_loss": 0.03367884084582329, "train_loss_bc": 0.03366735205054283, "train_loss_llm": 0.0011488497257232666, "grad_norm": 0.07494020462036133, "global_step": 29479, "epoch": 818, "lr": 0.001} +{"train_loss": 0.028348740190267563, "train_loss_bc": 0.028348740190267563, "train_loss_llm": 0.0, "grad_norm": 0.08795148134231567, "global_step": 29480, "epoch": 818, "lr": 0.001} +{"train_loss": 0.03466787189245224, "train_loss_bc": 0.03466787189245224, "train_loss_llm": 0.0, "grad_norm": 0.03308739885687828, "global_step": 29481, "epoch": 818, "lr": 0.001} +{"train_loss": 0.028716357424855232, "train_loss_bc": 0.028716357424855232, "train_loss_llm": 0.0, "grad_norm": 0.05889405682682991, "global_step": 29482, "epoch": 818, "lr": 0.001} +{"train_loss": 0.030000818324171834, "train_loss_bc": 0.028532296419143677, "train_loss_llm": 0.0, "grad_norm": 0.08238014578819275, "global_step": 29483, "epoch": 818, "lr": 0.001, "val_loss": 0.09832911193370819} +{"train_loss": 0.030359042808413506, "train_loss_bc": 0.030359042808413506, "train_loss_llm": 0.0, "grad_norm": 0.09978961199522018, "global_step": 29484, "epoch": 819, "lr": 0.001} +{"train_loss": 0.03664703667163849, "train_loss_bc": 0.03664703667163849, "train_loss_llm": 0.0, "grad_norm": 0.039102330803871155, "global_step": 29485, "epoch": 819, "lr": 0.001} +{"train_loss": 0.027263058349490166, "train_loss_bc": 0.027258865535259247, "train_loss_llm": 0.00041934847831726074, "grad_norm": 0.0683826431632042, "global_step": 29486, "epoch": 819, "lr": 0.001} +{"train_loss": 0.03222109377384186, "train_loss_bc": 0.03222000226378441, "train_loss_llm": 0.00010922551155090332, "grad_norm": 0.08788581192493439, "global_step": 29487, "epoch": 819, "lr": 0.001} +{"train_loss": 0.02977035939693451, "train_loss_bc": 0.02977035939693451, "train_loss_llm": 0.0, "grad_norm": 0.10737734287977219, "global_step": 29488, "epoch": 819, "lr": 0.001} +{"train_loss": 0.0266486294567585, "train_loss_bc": 0.0266486294567585, "train_loss_llm": 0.0, "grad_norm": 0.04004598408937454, "global_step": 29489, "epoch": 819, "lr": 0.001} +{"train_loss": 0.029873782768845558, "train_loss_bc": 0.029869649559259415, "train_loss_llm": 0.000413358211517334, "grad_norm": 0.0605129636824131, "global_step": 29490, "epoch": 819, "lr": 0.001} +{"train_loss": 0.03318283334374428, "train_loss_bc": 0.03317388519644737, "train_loss_llm": 0.0008949041366577148, "grad_norm": 0.08526971191167831, "global_step": 29491, "epoch": 819, "lr": 0.001} +{"train_loss": 0.034435342997312546, "train_loss_bc": 0.034435342997312546, "train_loss_llm": 0.0, "grad_norm": 0.09140607714653015, "global_step": 29492, "epoch": 819, "lr": 0.001} +{"train_loss": 0.03074873425066471, "train_loss_bc": 0.03074873425066471, "train_loss_llm": 0.0, "grad_norm": 0.03543459624052048, "global_step": 29493, "epoch": 819, "lr": 0.001} +{"train_loss": 0.028214411810040474, "train_loss_bc": 0.02819332852959633, "train_loss_llm": 0.002108335494995117, "grad_norm": 0.05167504772543907, "global_step": 29494, "epoch": 819, "lr": 0.001} +{"train_loss": 0.025976676493883133, "train_loss_bc": 0.025976676493883133, "train_loss_llm": 0.0, "grad_norm": 0.07694687694311142, "global_step": 29495, "epoch": 819, "lr": 0.001} +{"train_loss": 0.03238499537110329, "train_loss_bc": 0.03238499537110329, "train_loss_llm": 0.0, "grad_norm": 0.1041756272315979, "global_step": 29496, "epoch": 819, "lr": 0.001} +{"train_loss": 0.034212589263916016, "train_loss_bc": 0.034212589263916016, "train_loss_llm": 0.0, "grad_norm": 0.03363589569926262, "global_step": 29497, "epoch": 819, "lr": 0.001} +{"train_loss": 0.027453141286969185, "train_loss_bc": 0.027453141286969185, "train_loss_llm": 0.0, "grad_norm": 0.06451022624969482, "global_step": 29498, "epoch": 819, "lr": 0.001} +{"train_loss": 0.036824893206357956, "train_loss_bc": 0.036824893206357956, "train_loss_llm": 0.0, "grad_norm": 0.09924677759408951, "global_step": 29499, "epoch": 819, "lr": 0.001} +{"train_loss": 0.027193356305360794, "train_loss_bc": 0.02718260884284973, "train_loss_llm": 0.001074671745300293, "grad_norm": 0.12028927356004715, "global_step": 29500, "epoch": 819, "lr": 0.001} +{"train_loss": 0.03313346952199936, "train_loss_bc": 0.03313346952199936, "train_loss_llm": 0.0, "grad_norm": 0.04037947580218315, "global_step": 29501, "epoch": 819, "lr": 0.001} +{"train_loss": 0.0267086960375309, "train_loss_bc": 0.0267086960375309, "train_loss_llm": 0.0, "grad_norm": 0.06913270056247711, "global_step": 29502, "epoch": 819, "lr": 0.001} +{"train_loss": 0.031331196427345276, "train_loss_bc": 0.031331196427345276, "train_loss_llm": 0.0, "grad_norm": 0.08990482240915298, "global_step": 29503, "epoch": 819, "lr": 0.001} +{"train_loss": 0.03791311755776405, "train_loss_bc": 0.03791311755776405, "train_loss_llm": 0.0, "grad_norm": 0.12233609706163406, "global_step": 29504, "epoch": 819, "lr": 0.001} +{"train_loss": 0.024256378412246704, "train_loss_bc": 0.024256378412246704, "train_loss_llm": 0.0, "grad_norm": 0.032175201922655106, "global_step": 29505, "epoch": 819, "lr": 0.001} +{"train_loss": 0.023704392835497856, "train_loss_bc": 0.023704392835497856, "train_loss_llm": 0.0, "grad_norm": 0.04872668907046318, "global_step": 29506, "epoch": 819, "lr": 0.001} +{"train_loss": 0.028836021199822426, "train_loss_bc": 0.028836021199822426, "train_loss_llm": 0.0, "grad_norm": 0.060861632227897644, "global_step": 29507, "epoch": 819, "lr": 0.001} +{"train_loss": 0.028786923736333847, "train_loss_bc": 0.028786923736333847, "train_loss_llm": 0.0, "grad_norm": 0.07841970026493073, "global_step": 29508, "epoch": 819, "lr": 0.001} +{"train_loss": 0.036580901592969894, "train_loss_bc": 0.036580901592969894, "train_loss_llm": 0.0, "grad_norm": 0.03474215045571327, "global_step": 29509, "epoch": 819, "lr": 0.001} +{"train_loss": 0.02841489017009735, "train_loss_bc": 0.02841489017009735, "train_loss_llm": 0.0, "grad_norm": 0.0463317334651947, "global_step": 29510, "epoch": 819, "lr": 0.001} +{"train_loss": 0.02655905671417713, "train_loss_bc": 0.02655905671417713, "train_loss_llm": 0.0, "grad_norm": 0.06454109400510788, "global_step": 29511, "epoch": 819, "lr": 0.001} +{"train_loss": 0.02616112120449543, "train_loss_bc": 0.02615009993314743, "train_loss_llm": 0.0011021196842193604, "grad_norm": 0.08253635466098785, "global_step": 29512, "epoch": 819, "lr": 0.001} +{"train_loss": 0.03301108255982399, "train_loss_bc": 0.033004794269800186, "train_loss_llm": 0.0006286799907684326, "grad_norm": 0.03319514915347099, "global_step": 29513, "epoch": 819, "lr": 0.001} +{"train_loss": 0.030485549941658974, "train_loss_bc": 0.030485549941658974, "train_loss_llm": 0.0, "grad_norm": 0.059126898646354675, "global_step": 29514, "epoch": 819, "lr": 0.001} +{"train_loss": 0.02892843261361122, "train_loss_bc": 0.02892843261361122, "train_loss_llm": 0.0, "grad_norm": 0.08051440864801407, "global_step": 29515, "epoch": 819, "lr": 0.001} +{"train_loss": 0.024449525400996208, "train_loss_bc": 0.024449525400996208, "train_loss_llm": 0.0, "grad_norm": 0.09427085518836975, "global_step": 29516, "epoch": 819, "lr": 0.001} +{"train_loss": 0.02885718084871769, "train_loss_bc": 0.02885718084871769, "train_loss_llm": 0.0, "grad_norm": 0.034787051379680634, "global_step": 29517, "epoch": 819, "lr": 0.001} +{"train_loss": 0.0300501249730587, "train_loss_bc": 0.030046237632632256, "train_loss_llm": 0.00038868188858032227, "grad_norm": 0.07124944031238556, "global_step": 29518, "epoch": 819, "lr": 0.001} +{"train_loss": 0.030255848593595955, "train_loss_bc": 0.03763251006603241, "train_loss_llm": 0.0, "grad_norm": 0.10310452431440353, "global_step": 29519, "epoch": 819, "lr": 0.001, "val_loss": 0.09670056402683258} +{"train_loss": 0.03141433000564575, "train_loss_bc": 0.03141433000564575, "train_loss_llm": 0.0, "grad_norm": 0.11315152794122696, "global_step": 29520, "epoch": 820, "lr": 0.001} +{"train_loss": 0.029343396425247192, "train_loss_bc": 0.029343396425247192, "train_loss_llm": 0.0, "grad_norm": 0.033882372081279755, "global_step": 29521, "epoch": 820, "lr": 0.001} +{"train_loss": 0.030839599668979645, "train_loss_bc": 0.030813390389084816, "train_loss_llm": 0.002620875835418701, "grad_norm": 0.05135047435760498, "global_step": 29522, "epoch": 820, "lr": 0.001} +{"train_loss": 0.03074902668595314, "train_loss_bc": 0.030746959149837494, "train_loss_llm": 0.00020682811737060547, "grad_norm": 0.07300541549921036, "global_step": 29523, "epoch": 820, "lr": 0.001} +{"train_loss": 0.03440644219517708, "train_loss_bc": 0.03440644219517708, "train_loss_llm": 0.0, "grad_norm": 0.10432912409305573, "global_step": 29524, "epoch": 820, "lr": 0.001} +{"train_loss": 0.028311537578701973, "train_loss_bc": 0.028311537578701973, "train_loss_llm": 0.0, "grad_norm": 0.034175772219896317, "global_step": 29525, "epoch": 820, "lr": 0.001} +{"train_loss": 0.028243504464626312, "train_loss_bc": 0.028238903731107712, "train_loss_llm": 0.00046011805534362793, "grad_norm": 0.049566615372896194, "global_step": 29526, "epoch": 820, "lr": 0.001} +{"train_loss": 0.030349252745509148, "train_loss_bc": 0.030349252745509148, "train_loss_llm": 0.0, "grad_norm": 0.06384817510843277, "global_step": 29527, "epoch": 820, "lr": 0.001} +{"train_loss": 0.029890574514865875, "train_loss_bc": 0.029881011694669724, "train_loss_llm": 0.0009563267230987549, "grad_norm": 0.07961951196193695, "global_step": 29528, "epoch": 820, "lr": 0.001} +{"train_loss": 0.025981677696108818, "train_loss_bc": 0.025981677696108818, "train_loss_llm": 0.0, "grad_norm": 0.04268636927008629, "global_step": 29529, "epoch": 820, "lr": 0.001} +{"train_loss": 0.026326192542910576, "train_loss_bc": 0.026326192542910576, "train_loss_llm": 0.0, "grad_norm": 0.05783725157380104, "global_step": 29530, "epoch": 820, "lr": 0.001} +{"train_loss": 0.03633761778473854, "train_loss_bc": 0.03633691743016243, "train_loss_llm": 7.012486457824707e-05, "grad_norm": 0.08503060042858124, "global_step": 29531, "epoch": 820, "lr": 0.001} +{"train_loss": 0.02810843475162983, "train_loss_bc": 0.02810843475162983, "train_loss_llm": 0.0, "grad_norm": 0.09714603424072266, "global_step": 29532, "epoch": 820, "lr": 0.001} +{"train_loss": 0.028883589431643486, "train_loss_bc": 0.028883589431643486, "train_loss_llm": 0.0, "grad_norm": 0.03996775671839714, "global_step": 29533, "epoch": 820, "lr": 0.001} +{"train_loss": 0.0314919538795948, "train_loss_bc": 0.0314919538795948, "train_loss_llm": 0.0, "grad_norm": 0.06719451397657394, "global_step": 29534, "epoch": 820, "lr": 0.001} +{"train_loss": 0.024652795866131783, "train_loss_bc": 0.024652795866131783, "train_loss_llm": 0.0, "grad_norm": 0.08013206720352173, "global_step": 29535, "epoch": 820, "lr": 0.001} +{"train_loss": 0.026549188420176506, "train_loss_bc": 0.026532309129834175, "train_loss_llm": 0.0016880035400390625, "grad_norm": 0.10311254858970642, "global_step": 29536, "epoch": 820, "lr": 0.001} +{"train_loss": 0.0254830215126276, "train_loss_bc": 0.0254830215126276, "train_loss_llm": 0.0, "grad_norm": 0.03093036636710167, "global_step": 29537, "epoch": 820, "lr": 0.001} +{"train_loss": 0.030598793178796768, "train_loss_bc": 0.03056970052421093, "train_loss_llm": 0.0029091835021972656, "grad_norm": 0.05525125935673714, "global_step": 29538, "epoch": 820, "lr": 0.001} +{"train_loss": 0.031038857996463776, "train_loss_bc": 0.031038857996463776, "train_loss_llm": 0.0, "grad_norm": 0.07351862639188766, "global_step": 29539, "epoch": 820, "lr": 0.001} +{"train_loss": 0.028876828029751778, "train_loss_bc": 0.02887384220957756, "train_loss_llm": 0.00029852986335754395, "grad_norm": 0.08929013460874557, "global_step": 29540, "epoch": 820, "lr": 0.001} +{"train_loss": 0.0295723844319582, "train_loss_bc": 0.0295723844319582, "train_loss_llm": 0.0, "grad_norm": 0.030390707775950432, "global_step": 29541, "epoch": 820, "lr": 0.001} +{"train_loss": 0.035940688103437424, "train_loss_bc": 0.035940688103437424, "train_loss_llm": 0.0, "grad_norm": 0.05683640390634537, "global_step": 29542, "epoch": 820, "lr": 0.001} +{"train_loss": 0.02798735536634922, "train_loss_bc": 0.02798735536634922, "train_loss_llm": 0.0, "grad_norm": 0.08066968619823456, "global_step": 29543, "epoch": 820, "lr": 0.001} +{"train_loss": 0.0340367928147316, "train_loss_bc": 0.0340367928147316, "train_loss_llm": 0.0, "grad_norm": 0.10513492673635483, "global_step": 29544, "epoch": 820, "lr": 0.001} +{"train_loss": 0.030942020937800407, "train_loss_bc": 0.030942020937800407, "train_loss_llm": 0.0, "grad_norm": 0.04466208070516586, "global_step": 29545, "epoch": 820, "lr": 0.001} +{"train_loss": 0.03151025250554085, "train_loss_bc": 0.03151025250554085, "train_loss_llm": 0.0, "grad_norm": 0.0603947751224041, "global_step": 29546, "epoch": 820, "lr": 0.001} +{"train_loss": 0.028236856684088707, "train_loss_bc": 0.028236856684088707, "train_loss_llm": 0.0, "grad_norm": 0.07490924000740051, "global_step": 29547, "epoch": 820, "lr": 0.001} +{"train_loss": 0.03081243857741356, "train_loss_bc": 0.03081243857741356, "train_loss_llm": 0.0, "grad_norm": 0.09282460808753967, "global_step": 29548, "epoch": 820, "lr": 0.001} +{"train_loss": 0.025525078177452087, "train_loss_bc": 0.025518042966723442, "train_loss_llm": 0.0007035732269287109, "grad_norm": 0.032949283719062805, "global_step": 29549, "epoch": 820, "lr": 0.001} +{"train_loss": 0.0348161980509758, "train_loss_bc": 0.0348161980509758, "train_loss_llm": 0.0, "grad_norm": 0.06464895606040955, "global_step": 29550, "epoch": 820, "lr": 0.001} +{"train_loss": 0.030546968802809715, "train_loss_bc": 0.030546968802809715, "train_loss_llm": 0.0, "grad_norm": 0.07782156765460968, "global_step": 29551, "epoch": 820, "lr": 0.001} +{"train_loss": 0.030716702342033386, "train_loss_bc": 0.030716702342033386, "train_loss_llm": 0.0, "grad_norm": 0.0943039059638977, "global_step": 29552, "epoch": 820, "lr": 0.001} +{"train_loss": 0.03031899221241474, "train_loss_bc": 0.03031899221241474, "train_loss_llm": 0.0, "grad_norm": 0.044663503766059875, "global_step": 29553, "epoch": 820, "lr": 0.001} +{"train_loss": 0.03525897115468979, "train_loss_bc": 0.03525897115468979, "train_loss_llm": 0.0, "grad_norm": 0.05897923558950424, "global_step": 29554, "epoch": 820, "lr": 0.001} +{"train_loss": 0.03012445367251833, "train_loss_bc": 0.03038201667368412, "train_loss_llm": 0.0, "grad_norm": 0.087978295981884, "global_step": 29555, "epoch": 820, "lr": 0.001, "train/cumulative_reward": 63.5738289815806, "train/mean_score": 15.8229009404223, "train/success_rate": 0.4, "test/cumulative_reward": 37.46451679473081, "test/mean_score": 9.633746734870876, "test/success_rate": 0.26, "val_loss": 0.09916549175977707, "train_action_mse_error": 0.03345974162220955} +{"train_loss": 0.0317586250603199, "train_loss_bc": 0.0317586250603199, "train_loss_llm": 0.0, "grad_norm": 0.11354534327983856, "global_step": 29556, "epoch": 821, "lr": 0.001} +{"train_loss": 0.03186953440308571, "train_loss_bc": 0.031853266060352325, "train_loss_llm": 0.0016268491744995117, "grad_norm": 0.03703565523028374, "global_step": 29557, "epoch": 821, "lr": 0.001} +{"train_loss": 0.02603766694664955, "train_loss_bc": 0.02603766694664955, "train_loss_llm": 0.0, "grad_norm": 0.05483829602599144, "global_step": 29558, "epoch": 821, "lr": 0.001} +{"train_loss": 0.033366959542036057, "train_loss_bc": 0.03336428850889206, "train_loss_llm": 0.00026726722717285156, "grad_norm": 0.07697974890470505, "global_step": 29559, "epoch": 821, "lr": 0.001} +{"train_loss": 0.03219690918922424, "train_loss_bc": 0.03219690918922424, "train_loss_llm": 0.0, "grad_norm": 0.09538440406322479, "global_step": 29560, "epoch": 821, "lr": 0.001} +{"train_loss": 0.03310566022992134, "train_loss_bc": 0.03310566022992134, "train_loss_llm": 0.0, "grad_norm": 0.04433530941605568, "global_step": 29561, "epoch": 821, "lr": 0.001} +{"train_loss": 0.030251706019043922, "train_loss_bc": 0.030246024951338768, "train_loss_llm": 0.0005680322647094727, "grad_norm": 0.05788874626159668, "global_step": 29562, "epoch": 821, "lr": 0.001} +{"train_loss": 0.028563542291522026, "train_loss_bc": 0.028563542291522026, "train_loss_llm": 0.0, "grad_norm": 0.08323997259140015, "global_step": 29563, "epoch": 821, "lr": 0.001} +{"train_loss": 0.028515635058283806, "train_loss_bc": 0.028515635058283806, "train_loss_llm": 0.0, "grad_norm": 0.0954117551445961, "global_step": 29564, "epoch": 821, "lr": 0.001} +{"train_loss": 0.03269708901643753, "train_loss_bc": 0.03269708901643753, "train_loss_llm": 0.0, "grad_norm": 0.03809274733066559, "global_step": 29565, "epoch": 821, "lr": 0.001} +{"train_loss": 0.0351690873503685, "train_loss_bc": 0.0351690873503685, "train_loss_llm": 0.0, "grad_norm": 0.06677943468093872, "global_step": 29566, "epoch": 821, "lr": 0.001} +{"train_loss": 0.03170473501086235, "train_loss_bc": 0.03169805556535721, "train_loss_llm": 0.0006679892539978027, "grad_norm": 0.08039448410272598, "global_step": 29567, "epoch": 821, "lr": 0.001} +{"train_loss": 0.032226406037807465, "train_loss_bc": 0.032226406037807465, "train_loss_llm": 0.0, "grad_norm": 0.10443396866321564, "global_step": 29568, "epoch": 821, "lr": 0.001} +{"train_loss": 0.027459420263767242, "train_loss_bc": 0.027459420263767242, "train_loss_llm": 0.0, "grad_norm": 0.03559509664773941, "global_step": 29569, "epoch": 821, "lr": 0.001} +{"train_loss": 0.029248256236314774, "train_loss_bc": 0.029248256236314774, "train_loss_llm": 0.0, "grad_norm": 0.054008565843105316, "global_step": 29570, "epoch": 821, "lr": 0.001} +{"train_loss": 0.03119674324989319, "train_loss_bc": 0.03119674324989319, "train_loss_llm": 0.0, "grad_norm": 0.08357507735490799, "global_step": 29571, "epoch": 821, "lr": 0.001} +{"train_loss": 0.03426465019583702, "train_loss_bc": 0.03426465019583702, "train_loss_llm": 0.0, "grad_norm": 0.10378136485815048, "global_step": 29572, "epoch": 821, "lr": 0.001} +{"train_loss": 0.0309232659637928, "train_loss_bc": 0.0309232659637928, "train_loss_llm": 0.0, "grad_norm": 0.042069949209690094, "global_step": 29573, "epoch": 821, "lr": 0.001} +{"train_loss": 0.035347528755664825, "train_loss_bc": 0.035347528755664825, "train_loss_llm": 0.0, "grad_norm": 0.06612616032361984, "global_step": 29574, "epoch": 821, "lr": 0.001} +{"train_loss": 0.029130855575203896, "train_loss_bc": 0.029121333733201027, "train_loss_llm": 0.0009521543979644775, "grad_norm": 0.09349334985017776, "global_step": 29575, "epoch": 821, "lr": 0.001} +{"train_loss": 0.026719719171524048, "train_loss_bc": 0.026719719171524048, "train_loss_llm": 0.0, "grad_norm": 0.11539990454912186, "global_step": 29576, "epoch": 821, "lr": 0.001} +{"train_loss": 0.0280605535954237, "train_loss_bc": 0.0280605535954237, "train_loss_llm": 0.0, "grad_norm": 0.03429204598069191, "global_step": 29577, "epoch": 821, "lr": 0.001} +{"train_loss": 0.028145326301455498, "train_loss_bc": 0.028145326301455498, "train_loss_llm": 0.0, "grad_norm": 0.05010301247239113, "global_step": 29578, "epoch": 821, "lr": 0.001} +{"train_loss": 0.029569203034043312, "train_loss_bc": 0.029558643698692322, "train_loss_llm": 0.001055985689163208, "grad_norm": 0.07268504798412323, "global_step": 29579, "epoch": 821, "lr": 0.001} +{"train_loss": 0.03368984907865524, "train_loss_bc": 0.03368984907865524, "train_loss_llm": 0.0, "grad_norm": 0.08946637809276581, "global_step": 29580, "epoch": 821, "lr": 0.001} +{"train_loss": 0.023888643831014633, "train_loss_bc": 0.023874910548329353, "train_loss_llm": 0.001373380422592163, "grad_norm": 0.02981247566640377, "global_step": 29581, "epoch": 821, "lr": 0.001} +{"train_loss": 0.028041724115610123, "train_loss_bc": 0.028041724115610123, "train_loss_llm": 0.0, "grad_norm": 0.060321103781461716, "global_step": 29582, "epoch": 821, "lr": 0.001} +{"train_loss": 0.029350977391004562, "train_loss_bc": 0.029350977391004562, "train_loss_llm": 0.0, "grad_norm": 0.08070341497659683, "global_step": 29583, "epoch": 821, "lr": 0.001} +{"train_loss": 0.03026902675628662, "train_loss_bc": 0.03026902675628662, "train_loss_llm": 0.0, "grad_norm": 0.1070791482925415, "global_step": 29584, "epoch": 821, "lr": 0.001} +{"train_loss": 0.030540894716978073, "train_loss_bc": 0.030540894716978073, "train_loss_llm": 0.0, "grad_norm": 0.03589009866118431, "global_step": 29585, "epoch": 821, "lr": 0.001} +{"train_loss": 0.03215593472123146, "train_loss_bc": 0.03214424103498459, "train_loss_llm": 0.0011692047119140625, "grad_norm": 0.05970202758908272, "global_step": 29586, "epoch": 821, "lr": 0.001} +{"train_loss": 0.025276781991124153, "train_loss_bc": 0.025276781991124153, "train_loss_llm": 0.0, "grad_norm": 0.08462469279766083, "global_step": 29587, "epoch": 821, "lr": 0.001} +{"train_loss": 0.0308812465518713, "train_loss_bc": 0.030864177271723747, "train_loss_llm": 0.001706838607788086, "grad_norm": 0.10708118230104446, "global_step": 29588, "epoch": 821, "lr": 0.001} +{"train_loss": 0.027448024600744247, "train_loss_bc": 0.027448024600744247, "train_loss_llm": 0.0, "grad_norm": 0.04091569408774376, "global_step": 29589, "epoch": 821, "lr": 0.001} +{"train_loss": 0.027240898460149765, "train_loss_bc": 0.027240898460149765, "train_loss_llm": 0.0, "grad_norm": 0.07034724950790405, "global_step": 29590, "epoch": 821, "lr": 0.001} +{"train_loss": 0.030247027209649485, "train_loss_bc": 0.032566431909799576, "train_loss_llm": 0.0013466477394104004, "grad_norm": 0.0950893685221672, "global_step": 29591, "epoch": 821, "lr": 0.001, "val_loss": 0.10010853409767151} +{"train_loss": 0.028835726901888847, "train_loss_bc": 0.028835726901888847, "train_loss_llm": 0.0, "grad_norm": 0.10967791080474854, "global_step": 29592, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03290531784296036, "train_loss_bc": 0.03290531784296036, "train_loss_llm": 0.0, "grad_norm": 0.03827041760087013, "global_step": 29593, "epoch": 822, "lr": 0.001} +{"train_loss": 0.02459666319191456, "train_loss_bc": 0.02459418587386608, "train_loss_llm": 0.00024765729904174805, "grad_norm": 0.05456525832414627, "global_step": 29594, "epoch": 822, "lr": 0.001} +{"train_loss": 0.030908072367310524, "train_loss_bc": 0.030908001586794853, "train_loss_llm": 7.12275505065918e-06, "grad_norm": 0.07618885487318039, "global_step": 29595, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03461729735136032, "train_loss_bc": 0.03458910807967186, "train_loss_llm": 0.002819061279296875, "grad_norm": 0.10310391336679459, "global_step": 29596, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03493166342377663, "train_loss_bc": 0.03492921590805054, "train_loss_llm": 0.0002449154853820801, "grad_norm": 0.04467792809009552, "global_step": 29597, "epoch": 822, "lr": 0.001} +{"train_loss": 0.030093425884842873, "train_loss_bc": 0.030093425884842873, "train_loss_llm": 0.0, "grad_norm": 0.06533753126859665, "global_step": 29598, "epoch": 822, "lr": 0.001} +{"train_loss": 0.025939688086509705, "train_loss_bc": 0.025939688086509705, "train_loss_llm": 0.0, "grad_norm": 0.08658231794834137, "global_step": 29599, "epoch": 822, "lr": 0.001} +{"train_loss": 0.02986784465610981, "train_loss_bc": 0.02986784465610981, "train_loss_llm": 0.0, "grad_norm": 0.1043296605348587, "global_step": 29600, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03143208101391792, "train_loss_bc": 0.03143208101391792, "train_loss_llm": 0.0, "grad_norm": 0.038854118436574936, "global_step": 29601, "epoch": 822, "lr": 0.001} +{"train_loss": 0.02226933464407921, "train_loss_bc": 0.02226933464407921, "train_loss_llm": 0.0, "grad_norm": 0.058280594646930695, "global_step": 29602, "epoch": 822, "lr": 0.001} +{"train_loss": 0.026507824659347534, "train_loss_bc": 0.026507824659347534, "train_loss_llm": 0.0, "grad_norm": 0.07348192483186722, "global_step": 29603, "epoch": 822, "lr": 0.001} +{"train_loss": 0.04189547151327133, "train_loss_bc": 0.04189547151327133, "train_loss_llm": 0.0, "grad_norm": 0.0935136154294014, "global_step": 29604, "epoch": 822, "lr": 0.001} +{"train_loss": 0.027297677472233772, "train_loss_bc": 0.027297677472233772, "train_loss_llm": 0.0, "grad_norm": 0.03361405432224274, "global_step": 29605, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03151693940162659, "train_loss_bc": 0.03151693940162659, "train_loss_llm": 0.0, "grad_norm": 0.05885618180036545, "global_step": 29606, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03679775446653366, "train_loss_bc": 0.03679775446653366, "train_loss_llm": 0.0, "grad_norm": 0.09492142498493195, "global_step": 29607, "epoch": 822, "lr": 0.001} +{"train_loss": 0.028560880571603775, "train_loss_bc": 0.028560880571603775, "train_loss_llm": 0.0, "grad_norm": 0.12411704659461975, "global_step": 29608, "epoch": 822, "lr": 0.001} +{"train_loss": 0.02621649205684662, "train_loss_bc": 0.02621649205684662, "train_loss_llm": 0.0, "grad_norm": 0.03390519320964813, "global_step": 29609, "epoch": 822, "lr": 0.001} +{"train_loss": 0.027717430144548416, "train_loss_bc": 0.027717430144548416, "train_loss_llm": 0.0, "grad_norm": 0.053098853677511215, "global_step": 29610, "epoch": 822, "lr": 0.001} +{"train_loss": 0.033010080456733704, "train_loss_bc": 0.03300878405570984, "train_loss_llm": 0.00012952089309692383, "grad_norm": 0.08248258382081985, "global_step": 29611, "epoch": 822, "lr": 0.001} +{"train_loss": 0.037125278264284134, "train_loss_bc": 0.037125278264284134, "train_loss_llm": 0.0, "grad_norm": 0.10404133796691895, "global_step": 29612, "epoch": 822, "lr": 0.001} +{"train_loss": 0.030334748327732086, "train_loss_bc": 0.030334748327732086, "train_loss_llm": 0.0, "grad_norm": 0.031670812517404556, "global_step": 29613, "epoch": 822, "lr": 0.001} +{"train_loss": 0.019595319405198097, "train_loss_bc": 0.019589373841881752, "train_loss_llm": 0.0005944669246673584, "grad_norm": 0.046661607921123505, "global_step": 29614, "epoch": 822, "lr": 0.001} +{"train_loss": 0.029224032536149025, "train_loss_bc": 0.029224032536149025, "train_loss_llm": 0.0, "grad_norm": 0.06724588572978973, "global_step": 29615, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03145340085029602, "train_loss_bc": 0.03145075589418411, "train_loss_llm": 0.0002645254135131836, "grad_norm": 0.09734993427991867, "global_step": 29616, "epoch": 822, "lr": 0.001} +{"train_loss": 0.02463481016457081, "train_loss_bc": 0.02463481016457081, "train_loss_llm": 0.0, "grad_norm": 0.03377833217382431, "global_step": 29617, "epoch": 822, "lr": 0.001} +{"train_loss": 0.022495564073324203, "train_loss_bc": 0.022493969649076462, "train_loss_llm": 0.0001595020294189453, "grad_norm": 0.054680243134498596, "global_step": 29618, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03300962224602699, "train_loss_bc": 0.03300962224602699, "train_loss_llm": 0.0, "grad_norm": 0.07552564889192581, "global_step": 29619, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03709949180483818, "train_loss_bc": 0.03709949180483818, "train_loss_llm": 0.0, "grad_norm": 0.09165430068969727, "global_step": 29620, "epoch": 822, "lr": 0.001} +{"train_loss": 0.027064252644777298, "train_loss_bc": 0.027064252644777298, "train_loss_llm": 0.0, "grad_norm": 0.032440558075904846, "global_step": 29621, "epoch": 822, "lr": 0.001} +{"train_loss": 0.031032614409923553, "train_loss_bc": 0.031014591455459595, "train_loss_llm": 0.0018022656440734863, "grad_norm": 0.05494547262787819, "global_step": 29622, "epoch": 822, "lr": 0.001} +{"train_loss": 0.02756420150399208, "train_loss_bc": 0.02756420150399208, "train_loss_llm": 0.0, "grad_norm": 0.07151586562395096, "global_step": 29623, "epoch": 822, "lr": 0.001} +{"train_loss": 0.037192851305007935, "train_loss_bc": 0.037192851305007935, "train_loss_llm": 0.0, "grad_norm": 0.101177878677845, "global_step": 29624, "epoch": 822, "lr": 0.001} +{"train_loss": 0.029038403183221817, "train_loss_bc": 0.0290333591401577, "train_loss_llm": 0.0005044043064117432, "grad_norm": 0.03510938212275505, "global_step": 29625, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03088456392288208, "train_loss_bc": 0.03088456392288208, "train_loss_llm": 0.0, "grad_norm": 0.05405518040060997, "global_step": 29626, "epoch": 822, "lr": 0.001} +{"train_loss": 0.03017844068300393, "train_loss_bc": 0.03275704383850098, "train_loss_llm": 0.0, "grad_norm": 0.08307870477437973, "global_step": 29627, "epoch": 822, "lr": 0.001, "val_loss": 0.09862459450960159} +{"train_loss": 0.029021160677075386, "train_loss_bc": 0.029021160677075386, "train_loss_llm": 0.0, "grad_norm": 0.09697654098272324, "global_step": 29628, "epoch": 823, "lr": 0.001} +{"train_loss": 0.02849769778549671, "train_loss_bc": 0.02849769778549671, "train_loss_llm": 0.0, "grad_norm": 0.03316720947623253, "global_step": 29629, "epoch": 823, "lr": 0.001} +{"train_loss": 0.032888013869524, "train_loss_bc": 0.032888013869524, "train_loss_llm": 0.0, "grad_norm": 0.05874833092093468, "global_step": 29630, "epoch": 823, "lr": 0.001} +{"train_loss": 0.021403273567557335, "train_loss_bc": 0.021403273567557335, "train_loss_llm": 0.0, "grad_norm": 0.06932960450649261, "global_step": 29631, "epoch": 823, "lr": 0.001} +{"train_loss": 0.029578309506177902, "train_loss_bc": 0.029556408524513245, "train_loss_llm": 0.002190113067626953, "grad_norm": 0.08041947335004807, "global_step": 29632, "epoch": 823, "lr": 0.001} +{"train_loss": 0.030789319425821304, "train_loss_bc": 0.030789319425821304, "train_loss_llm": 0.0, "grad_norm": 0.03485286235809326, "global_step": 29633, "epoch": 823, "lr": 0.001} +{"train_loss": 0.034817133098840714, "train_loss_bc": 0.034817133098840714, "train_loss_llm": 0.0, "grad_norm": 0.06564461439847946, "global_step": 29634, "epoch": 823, "lr": 0.001} +{"train_loss": 0.026259049773216248, "train_loss_bc": 0.026242531836032867, "train_loss_llm": 0.001651763916015625, "grad_norm": 0.08706396073102951, "global_step": 29635, "epoch": 823, "lr": 0.001} +{"train_loss": 0.025154460221529007, "train_loss_bc": 0.025154460221529007, "train_loss_llm": 0.0, "grad_norm": 0.09741900861263275, "global_step": 29636, "epoch": 823, "lr": 0.001} +{"train_loss": 0.030663805082440376, "train_loss_bc": 0.030663805082440376, "train_loss_llm": 0.0, "grad_norm": 0.041548244655132294, "global_step": 29637, "epoch": 823, "lr": 0.001} +{"train_loss": 0.029228391125798225, "train_loss_bc": 0.029228391125798225, "train_loss_llm": 0.0, "grad_norm": 0.052668243646621704, "global_step": 29638, "epoch": 823, "lr": 0.001} +{"train_loss": 0.03447745740413666, "train_loss_bc": 0.03447745740413666, "train_loss_llm": 0.0, "grad_norm": 0.07531145960092545, "global_step": 29639, "epoch": 823, "lr": 0.001} +{"train_loss": 0.029349179938435555, "train_loss_bc": 0.029349179938435555, "train_loss_llm": 0.0, "grad_norm": 0.08371016383171082, "global_step": 29640, "epoch": 823, "lr": 0.001} +{"train_loss": 0.02807728946208954, "train_loss_bc": 0.02807728946208954, "train_loss_llm": 0.0, "grad_norm": 0.03188040107488632, "global_step": 29641, "epoch": 823, "lr": 0.001} +{"train_loss": 0.02943773940205574, "train_loss_bc": 0.029437044635415077, "train_loss_llm": 6.952881813049316e-05, "grad_norm": 0.04716766998171806, "global_step": 29642, "epoch": 823, "lr": 0.001} +{"train_loss": 0.028544878587126732, "train_loss_bc": 0.028544878587126732, "train_loss_llm": 0.0, "grad_norm": 0.06774869561195374, "global_step": 29643, "epoch": 823, "lr": 0.001} +{"train_loss": 0.028675658628344536, "train_loss_bc": 0.028675658628344536, "train_loss_llm": 0.0, "grad_norm": 0.08275296539068222, "global_step": 29644, "epoch": 823, "lr": 0.001} +{"train_loss": 0.029317030683159828, "train_loss_bc": 0.029317030683159828, "train_loss_llm": 0.0, "grad_norm": 0.042007774114608765, "global_step": 29645, "epoch": 823, "lr": 0.001} +{"train_loss": 0.02972528338432312, "train_loss_bc": 0.029720108956098557, "train_loss_llm": 0.0005174577236175537, "grad_norm": 0.07020625472068787, "global_step": 29646, "epoch": 823, "lr": 0.001} +{"train_loss": 0.03126441687345505, "train_loss_bc": 0.03126441687345505, "train_loss_llm": 0.0, "grad_norm": 0.09581597149372101, "global_step": 29647, "epoch": 823, "lr": 0.001} +{"train_loss": 0.030931178480386734, "train_loss_bc": 0.030931178480386734, "train_loss_llm": 0.0, "grad_norm": 0.11126561462879181, "global_step": 29648, "epoch": 823, "lr": 0.001} +{"train_loss": 0.027052372694015503, "train_loss_bc": 0.027052372694015503, "train_loss_llm": 0.0, "grad_norm": 0.034274447709321976, "global_step": 29649, "epoch": 823, "lr": 0.001} +{"train_loss": 0.03220662474632263, "train_loss_bc": 0.03220662474632263, "train_loss_llm": 0.0, "grad_norm": 0.05606014281511307, "global_step": 29650, "epoch": 823, "lr": 0.001} +{"train_loss": 0.02942011132836342, "train_loss_bc": 0.02942011132836342, "train_loss_llm": 0.0, "grad_norm": 0.0642913207411766, "global_step": 29651, "epoch": 823, "lr": 0.001} +{"train_loss": 0.028674770146608353, "train_loss_bc": 0.028674770146608353, "train_loss_llm": 0.0, "grad_norm": 0.0752471536397934, "global_step": 29652, "epoch": 823, "lr": 0.001} +{"train_loss": 0.03799425810575485, "train_loss_bc": 0.03799425810575485, "train_loss_llm": 0.0, "grad_norm": 0.03694011643528938, "global_step": 29653, "epoch": 823, "lr": 0.001} +{"train_loss": 0.031454943120479584, "train_loss_bc": 0.03144892305135727, "train_loss_llm": 0.0006018280982971191, "grad_norm": 0.06195203959941864, "global_step": 29654, "epoch": 823, "lr": 0.001} +{"train_loss": 0.02880333736538887, "train_loss_bc": 0.02879980579018593, "train_loss_llm": 0.00035312771797180176, "grad_norm": 0.07881810516119003, "global_step": 29655, "epoch": 823, "lr": 0.001} +{"train_loss": 0.030781731009483337, "train_loss_bc": 0.030781731009483337, "train_loss_llm": 0.0, "grad_norm": 0.09600501507520676, "global_step": 29656, "epoch": 823, "lr": 0.001} +{"train_loss": 0.027851808816194534, "train_loss_bc": 0.027851808816194534, "train_loss_llm": 0.0, "grad_norm": 0.03633878752589226, "global_step": 29657, "epoch": 823, "lr": 0.001} +{"train_loss": 0.03439543396234512, "train_loss_bc": 0.03439543396234512, "train_loss_llm": 0.0, "grad_norm": 0.06080694869160652, "global_step": 29658, "epoch": 823, "lr": 0.001} +{"train_loss": 0.02803017944097519, "train_loss_bc": 0.028013620525598526, "train_loss_llm": 0.001655876636505127, "grad_norm": 0.07400696724653244, "global_step": 29659, "epoch": 823, "lr": 0.001} +{"train_loss": 0.030447330325841904, "train_loss_bc": 0.030447330325841904, "train_loss_llm": 0.0, "grad_norm": 0.08813713490962982, "global_step": 29660, "epoch": 823, "lr": 0.001} +{"train_loss": 0.030863281339406967, "train_loss_bc": 0.030863281339406967, "train_loss_llm": 0.0, "grad_norm": 0.03717869147658348, "global_step": 29661, "epoch": 823, "lr": 0.001} +{"train_loss": 0.03329678624868393, "train_loss_bc": 0.03329678624868393, "train_loss_llm": 0.0, "grad_norm": 0.06542450934648514, "global_step": 29662, "epoch": 823, "lr": 0.001} +{"train_loss": 0.029951263788259692, "train_loss_bc": 0.02887107990682125, "train_loss_llm": 7.203221321105957e-05, "grad_norm": 0.08870162069797516, "global_step": 29663, "epoch": 823, "lr": 0.001, "val_loss": 0.09738575667142868} +{"train_loss": 0.027022285386919975, "train_loss_bc": 0.027018185704946518, "train_loss_llm": 0.00040990114212036133, "grad_norm": 0.10382743179798126, "global_step": 29664, "epoch": 824, "lr": 0.001} +{"train_loss": 0.028318582102656364, "train_loss_bc": 0.028318582102656364, "train_loss_llm": 0.0, "grad_norm": 0.031272344291210175, "global_step": 29665, "epoch": 824, "lr": 0.001} +{"train_loss": 0.02473253570497036, "train_loss_bc": 0.02473253570497036, "train_loss_llm": 0.0, "grad_norm": 0.05407951772212982, "global_step": 29666, "epoch": 824, "lr": 0.001} +{"train_loss": 0.02762158401310444, "train_loss_bc": 0.02762158401310444, "train_loss_llm": 0.0, "grad_norm": 0.06373997032642365, "global_step": 29667, "epoch": 824, "lr": 0.001} +{"train_loss": 0.03132138401269913, "train_loss_bc": 0.03132138401269913, "train_loss_llm": 0.0, "grad_norm": 0.07761523872613907, "global_step": 29668, "epoch": 824, "lr": 0.001} +{"train_loss": 0.02907365933060646, "train_loss_bc": 0.02907365933060646, "train_loss_llm": 0.0, "grad_norm": 0.037791959941387177, "global_step": 29669, "epoch": 824, "lr": 0.001} +{"train_loss": 0.031268272548913956, "train_loss_bc": 0.031250208616256714, "train_loss_llm": 0.0018062889575958252, "grad_norm": 0.04973755404353142, "global_step": 29670, "epoch": 824, "lr": 0.001} +{"train_loss": 0.028348753228783607, "train_loss_bc": 0.028334900736808777, "train_loss_llm": 0.0013852417469024658, "grad_norm": 0.07724498957395554, "global_step": 29671, "epoch": 824, "lr": 0.001} +{"train_loss": 0.02806953340768814, "train_loss_bc": 0.02806953340768814, "train_loss_llm": 0.0, "grad_norm": 0.0974728912115097, "global_step": 29672, "epoch": 824, "lr": 0.001} +{"train_loss": 0.025734946131706238, "train_loss_bc": 0.025733714923262596, "train_loss_llm": 0.00012320280075073242, "grad_norm": 0.03765412047505379, "global_step": 29673, "epoch": 824, "lr": 0.001} +{"train_loss": 0.03301374614238739, "train_loss_bc": 0.033007971942424774, "train_loss_llm": 0.0005773603916168213, "grad_norm": 0.0712488666176796, "global_step": 29674, "epoch": 824, "lr": 0.001} +{"train_loss": 0.031868964433670044, "train_loss_bc": 0.031868964433670044, "train_loss_llm": 0.0, "grad_norm": 0.09383542090654373, "global_step": 29675, "epoch": 824, "lr": 0.001} +{"train_loss": 0.0332852341234684, "train_loss_bc": 0.0332852341234684, "train_loss_llm": 0.0, "grad_norm": 0.10560835152864456, "global_step": 29676, "epoch": 824, "lr": 0.001} +{"train_loss": 0.031477320939302444, "train_loss_bc": 0.031477320939302444, "train_loss_llm": 0.0, "grad_norm": 0.03125506266951561, "global_step": 29677, "epoch": 824, "lr": 0.001} +{"train_loss": 0.035716231912374496, "train_loss_bc": 0.035716231912374496, "train_loss_llm": 0.0, "grad_norm": 0.054294656962156296, "global_step": 29678, "epoch": 824, "lr": 0.001} +{"train_loss": 0.02527042292058468, "train_loss_bc": 0.025267796590924263, "train_loss_llm": 0.0002626776695251465, "grad_norm": 0.0761571004986763, "global_step": 29679, "epoch": 824, "lr": 0.001} +{"train_loss": 0.03342878818511963, "train_loss_bc": 0.03342878818511963, "train_loss_llm": 0.0, "grad_norm": 0.0990760400891304, "global_step": 29680, "epoch": 824, "lr": 0.001} +{"train_loss": 0.027646878734230995, "train_loss_bc": 0.027646878734230995, "train_loss_llm": 0.0, "grad_norm": 0.03692929074168205, "global_step": 29681, "epoch": 824, "lr": 0.001} +{"train_loss": 0.03090633638203144, "train_loss_bc": 0.03090633638203144, "train_loss_llm": 0.0, "grad_norm": 0.06256501376628876, "global_step": 29682, "epoch": 824, "lr": 0.001} +{"train_loss": 0.030221933498978615, "train_loss_bc": 0.03021000325679779, "train_loss_llm": 0.0011929571628570557, "grad_norm": 0.08695225417613983, "global_step": 29683, "epoch": 824, "lr": 0.001} +{"train_loss": 0.03436897322535515, "train_loss_bc": 0.03436897322535515, "train_loss_llm": 0.0, "grad_norm": 0.10444364696741104, "global_step": 29684, "epoch": 824, "lr": 0.001} +{"train_loss": 0.028669992461800575, "train_loss_bc": 0.028669992461800575, "train_loss_llm": 0.0, "grad_norm": 0.027793006971478462, "global_step": 29685, "epoch": 824, "lr": 0.001} +{"train_loss": 0.025371083989739418, "train_loss_bc": 0.025371083989739418, "train_loss_llm": 0.0, "grad_norm": 0.04290478304028511, "global_step": 29686, "epoch": 824, "lr": 0.001} +{"train_loss": 0.03767605498433113, "train_loss_bc": 0.03767605498433113, "train_loss_llm": 0.0, "grad_norm": 0.058774564415216446, "global_step": 29687, "epoch": 824, "lr": 0.001} +{"train_loss": 0.03515056148171425, "train_loss_bc": 0.03515056148171425, "train_loss_llm": 0.0, "grad_norm": 0.07186409085988998, "global_step": 29688, "epoch": 824, "lr": 0.001} +{"train_loss": 0.02158777229487896, "train_loss_bc": 0.02158777229487896, "train_loss_llm": 0.0, "grad_norm": 0.030417965725064278, "global_step": 29689, "epoch": 824, "lr": 0.001} +{"train_loss": 0.026017623022198677, "train_loss_bc": 0.026000162586569786, "train_loss_llm": 0.0017461180686950684, "grad_norm": 0.042362626641988754, "global_step": 29690, "epoch": 824, "lr": 0.001} +{"train_loss": 0.02684139274060726, "train_loss_bc": 0.02684139274060726, "train_loss_llm": 0.0, "grad_norm": 0.06057687848806381, "global_step": 29691, "epoch": 824, "lr": 0.001} +{"train_loss": 0.02556486427783966, "train_loss_bc": 0.025556664913892746, "train_loss_llm": 0.0008198916912078857, "grad_norm": 0.07790720462799072, "global_step": 29692, "epoch": 824, "lr": 0.001} +{"train_loss": 0.040353406220674515, "train_loss_bc": 0.040353406220674515, "train_loss_llm": 0.0, "grad_norm": 0.04440848156809807, "global_step": 29693, "epoch": 824, "lr": 0.001} +{"train_loss": 0.0287643875926733, "train_loss_bc": 0.0287643875926733, "train_loss_llm": 0.0, "grad_norm": 0.07387705892324448, "global_step": 29694, "epoch": 824, "lr": 0.001} +{"train_loss": 0.032405510544776917, "train_loss_bc": 0.03240297734737396, "train_loss_llm": 0.00025337934494018555, "grad_norm": 0.11075129359960556, "global_step": 29695, "epoch": 824, "lr": 0.001} +{"train_loss": 0.028002668172121048, "train_loss_bc": 0.028002668172121048, "train_loss_llm": 0.0, "grad_norm": 0.1313934326171875, "global_step": 29696, "epoch": 824, "lr": 0.001} +{"train_loss": 0.03237086534500122, "train_loss_bc": 0.03237086534500122, "train_loss_llm": 0.0, "grad_norm": 0.05082190781831741, "global_step": 29697, "epoch": 824, "lr": 0.001} +{"train_loss": 0.033901263028383255, "train_loss_bc": 0.033901263028383255, "train_loss_llm": 0.0, "grad_norm": 0.085567407310009, "global_step": 29698, "epoch": 824, "lr": 0.001} +{"train_loss": 0.03026471199053857, "train_loss_bc": 0.03810903802514076, "train_loss_llm": 0.002678096294403076, "grad_norm": 0.1366325169801712, "global_step": 29699, "epoch": 824, "lr": 0.001, "val_loss": 0.10025487840175629} +{"train_loss": 0.0322701521217823, "train_loss_bc": 0.0322701521217823, "train_loss_llm": 0.0, "grad_norm": 0.1594703197479248, "global_step": 29700, "epoch": 825, "lr": 0.001} +{"train_loss": 0.025817399844527245, "train_loss_bc": 0.025817399844527245, "train_loss_llm": 0.0, "grad_norm": 0.037869520485401154, "global_step": 29701, "epoch": 825, "lr": 0.001} +{"train_loss": 0.031182507053017616, "train_loss_bc": 0.031182507053017616, "train_loss_llm": 0.0, "grad_norm": 0.05931973084807396, "global_step": 29702, "epoch": 825, "lr": 0.001} +{"train_loss": 0.033688984811306, "train_loss_bc": 0.033688984811306, "train_loss_llm": 0.0, "grad_norm": 0.06904130429029465, "global_step": 29703, "epoch": 825, "lr": 0.001} +{"train_loss": 0.02911866083741188, "train_loss_bc": 0.02911866083741188, "train_loss_llm": 0.0, "grad_norm": 0.08552847057580948, "global_step": 29704, "epoch": 825, "lr": 0.001} +{"train_loss": 0.02700110711157322, "train_loss_bc": 0.026999332010746002, "train_loss_llm": 0.00017747282981872559, "grad_norm": 0.036280736327171326, "global_step": 29705, "epoch": 825, "lr": 0.001} +{"train_loss": 0.030507726594805717, "train_loss_bc": 0.030500443652272224, "train_loss_llm": 0.0007282495498657227, "grad_norm": 0.05144990608096123, "global_step": 29706, "epoch": 825, "lr": 0.001} +{"train_loss": 0.030966661870479584, "train_loss_bc": 0.030966661870479584, "train_loss_llm": 0.0, "grad_norm": 0.05998434126377106, "global_step": 29707, "epoch": 825, "lr": 0.001} +{"train_loss": 0.02818070724606514, "train_loss_bc": 0.02818070724606514, "train_loss_llm": 0.0, "grad_norm": 0.07245273143053055, "global_step": 29708, "epoch": 825, "lr": 0.001} +{"train_loss": 0.027370266616344452, "train_loss_bc": 0.02736675553023815, "train_loss_llm": 0.00035119056701660156, "grad_norm": 0.03783291205763817, "global_step": 29709, "epoch": 825, "lr": 0.001} +{"train_loss": 0.028545742854475975, "train_loss_bc": 0.02852366864681244, "train_loss_llm": 0.0022073984146118164, "grad_norm": 0.06361187249422073, "global_step": 29710, "epoch": 825, "lr": 0.001} +{"train_loss": 0.028958415612578392, "train_loss_bc": 0.028958415612578392, "train_loss_llm": 0.0, "grad_norm": 0.0879606157541275, "global_step": 29711, "epoch": 825, "lr": 0.001} +{"train_loss": 0.03165629506111145, "train_loss_bc": 0.03165629506111145, "train_loss_llm": 0.0, "grad_norm": 0.11038368195295334, "global_step": 29712, "epoch": 825, "lr": 0.001} +{"train_loss": 0.02637631632387638, "train_loss_bc": 0.02637631632387638, "train_loss_llm": 0.0, "grad_norm": 0.033666666597127914, "global_step": 29713, "epoch": 825, "lr": 0.001} +{"train_loss": 0.027018865570425987, "train_loss_bc": 0.027016982436180115, "train_loss_llm": 0.0001882314682006836, "grad_norm": 0.05407261848449707, "global_step": 29714, "epoch": 825, "lr": 0.001} +{"train_loss": 0.03034554235637188, "train_loss_bc": 0.030343279242515564, "train_loss_llm": 0.0002263784408569336, "grad_norm": 0.08310647308826447, "global_step": 29715, "epoch": 825, "lr": 0.001} +{"train_loss": 0.03421877324581146, "train_loss_bc": 0.03421877324581146, "train_loss_llm": 0.0, "grad_norm": 0.11256983876228333, "global_step": 29716, "epoch": 825, "lr": 0.001} +{"train_loss": 0.02685571275651455, "train_loss_bc": 0.02685571275651455, "train_loss_llm": 0.0, "grad_norm": 0.04137205705046654, "global_step": 29717, "epoch": 825, "lr": 0.001} +{"train_loss": 0.033537279814481735, "train_loss_bc": 0.033537279814481735, "train_loss_llm": 0.0, "grad_norm": 0.06357705593109131, "global_step": 29718, "epoch": 825, "lr": 0.001} +{"train_loss": 0.029284650459885597, "train_loss_bc": 0.029284650459885597, "train_loss_llm": 0.0, "grad_norm": 0.08379118144512177, "global_step": 29719, "epoch": 825, "lr": 0.001} +{"train_loss": 0.027091186493635178, "train_loss_bc": 0.027091186493635178, "train_loss_llm": 0.0, "grad_norm": 0.09966365247964859, "global_step": 29720, "epoch": 825, "lr": 0.001} +{"train_loss": 0.030516551807522774, "train_loss_bc": 0.030516551807522774, "train_loss_llm": 0.0, "grad_norm": 0.04559679701924324, "global_step": 29721, "epoch": 825, "lr": 0.001} +{"train_loss": 0.03273333981633186, "train_loss_bc": 0.03271801397204399, "train_loss_llm": 0.0015327632427215576, "grad_norm": 0.06359168142080307, "global_step": 29722, "epoch": 825, "lr": 0.001} +{"train_loss": 0.02790897712111473, "train_loss_bc": 0.02790897712111473, "train_loss_llm": 0.0, "grad_norm": 0.08526654541492462, "global_step": 29723, "epoch": 825, "lr": 0.001} +{"train_loss": 0.033455103635787964, "train_loss_bc": 0.033455103635787964, "train_loss_llm": 0.0, "grad_norm": 0.11964511126279831, "global_step": 29724, "epoch": 825, "lr": 0.001} +{"train_loss": 0.02764267660677433, "train_loss_bc": 0.027634767815470695, "train_loss_llm": 0.0007908642292022705, "grad_norm": 0.03988536074757576, "global_step": 29725, "epoch": 825, "lr": 0.001} +{"train_loss": 0.03177454322576523, "train_loss_bc": 0.03177454322576523, "train_loss_llm": 0.0, "grad_norm": 0.0668240338563919, "global_step": 29726, "epoch": 825, "lr": 0.001} +{"train_loss": 0.03383670747280121, "train_loss_bc": 0.03381931781768799, "train_loss_llm": 0.0017390847206115723, "grad_norm": 0.07594984024763107, "global_step": 29727, "epoch": 825, "lr": 0.001} +{"train_loss": 0.030957818031311035, "train_loss_bc": 0.030957818031311035, "train_loss_llm": 0.0, "grad_norm": 0.08897961676120758, "global_step": 29728, "epoch": 825, "lr": 0.001} +{"train_loss": 0.03530866652727127, "train_loss_bc": 0.03530866652727127, "train_loss_llm": 0.0, "grad_norm": 0.030278554186224937, "global_step": 29729, "epoch": 825, "lr": 0.001} +{"train_loss": 0.024944065138697624, "train_loss_bc": 0.024944065138697624, "train_loss_llm": 0.0, "grad_norm": 0.0422762855887413, "global_step": 29730, "epoch": 825, "lr": 0.001} +{"train_loss": 0.02848765254020691, "train_loss_bc": 0.02848765254020691, "train_loss_llm": 0.0, "grad_norm": 0.06547723710536957, "global_step": 29731, "epoch": 825, "lr": 0.001} +{"train_loss": 0.032426148653030396, "train_loss_bc": 0.032426148653030396, "train_loss_llm": 0.0, "grad_norm": 0.07527679204940796, "global_step": 29732, "epoch": 825, "lr": 0.001} +{"train_loss": 0.028251510113477707, "train_loss_bc": 0.028251510113477707, "train_loss_llm": 0.0, "grad_norm": 0.04569951444864273, "global_step": 29733, "epoch": 825, "lr": 0.001} +{"train_loss": 0.03265855833888054, "train_loss_bc": 0.03265855833888054, "train_loss_llm": 0.0, "grad_norm": 0.06854327023029327, "global_step": 29734, "epoch": 825, "lr": 0.001} +{"train_loss": 0.03011802201055818, "train_loss_bc": 0.033343203365802765, "train_loss_llm": 0.001031637191772461, "grad_norm": 0.08048944920301437, "global_step": 29735, "epoch": 825, "lr": 0.001, "train/cumulative_reward": 95.57350926966396, "train/mean_score": 19.660689829671433, "train/success_rate": 0.5, "test/cumulative_reward": 28.544605457775138, "test/mean_score": 6.791278287275455, "test/success_rate": 0.16, "val_loss": 0.09952083230018616, "train_action_mse_error": 0.0332801453769207} +{"train_loss": 0.027482345700263977, "train_loss_bc": 0.027482345700263977, "train_loss_llm": 0.0, "grad_norm": 0.09749514609575272, "global_step": 29736, "epoch": 826, "lr": 0.001} +{"train_loss": 0.02705666981637478, "train_loss_bc": 0.02705666981637478, "train_loss_llm": 0.0, "grad_norm": 0.03679385781288147, "global_step": 29737, "epoch": 826, "lr": 0.001} +{"train_loss": 0.025563325732946396, "train_loss_bc": 0.025560321286320686, "train_loss_llm": 0.00030046701431274414, "grad_norm": 0.05810193717479706, "global_step": 29738, "epoch": 826, "lr": 0.001} +{"train_loss": 0.03130127117037773, "train_loss_bc": 0.03130127117037773, "train_loss_llm": 0.0, "grad_norm": 0.07727330923080444, "global_step": 29739, "epoch": 826, "lr": 0.001} +{"train_loss": 0.028873279690742493, "train_loss_bc": 0.028873279690742493, "train_loss_llm": 0.0, "grad_norm": 0.09412648528814316, "global_step": 29740, "epoch": 826, "lr": 0.001} +{"train_loss": 0.022376179695129395, "train_loss_bc": 0.022376179695129395, "train_loss_llm": 0.0, "grad_norm": 0.02896980755031109, "global_step": 29741, "epoch": 826, "lr": 0.001} +{"train_loss": 0.03011898137629032, "train_loss_bc": 0.03011014312505722, "train_loss_llm": 0.0008838772773742676, "grad_norm": 0.05147441104054451, "global_step": 29742, "epoch": 826, "lr": 0.001} +{"train_loss": 0.033141691237688065, "train_loss_bc": 0.033141691237688065, "train_loss_llm": 0.0, "grad_norm": 0.07163199037313461, "global_step": 29743, "epoch": 826, "lr": 0.001} +{"train_loss": 0.03366135060787201, "train_loss_bc": 0.03366135060787201, "train_loss_llm": 0.0, "grad_norm": 0.08686640858650208, "global_step": 29744, "epoch": 826, "lr": 0.001} +{"train_loss": 0.03240898624062538, "train_loss_bc": 0.03239802271127701, "train_loss_llm": 0.001096189022064209, "grad_norm": 0.03607049584388733, "global_step": 29745, "epoch": 826, "lr": 0.001} +{"train_loss": 0.02730540931224823, "train_loss_bc": 0.02730540931224823, "train_loss_llm": 0.0, "grad_norm": 0.05166035518050194, "global_step": 29746, "epoch": 826, "lr": 0.001} +{"train_loss": 0.03283839672803879, "train_loss_bc": 0.03283839672803879, "train_loss_llm": 0.0, "grad_norm": 0.06888262927532196, "global_step": 29747, "epoch": 826, "lr": 0.001} +{"train_loss": 0.029484841972589493, "train_loss_bc": 0.029484841972589493, "train_loss_llm": 0.0, "grad_norm": 0.09127165377140045, "global_step": 29748, "epoch": 826, "lr": 0.001} +{"train_loss": 0.03246409446001053, "train_loss_bc": 0.03246409446001053, "train_loss_llm": 0.0, "grad_norm": 0.03581111505627632, "global_step": 29749, "epoch": 826, "lr": 0.001} +{"train_loss": 0.027277154847979546, "train_loss_bc": 0.027277154847979546, "train_loss_llm": 0.0, "grad_norm": 0.05161863565444946, "global_step": 29750, "epoch": 826, "lr": 0.001} +{"train_loss": 0.02826540172100067, "train_loss_bc": 0.02826540172100067, "train_loss_llm": 0.0, "grad_norm": 0.06841029226779938, "global_step": 29751, "epoch": 826, "lr": 0.001} +{"train_loss": 0.03521892428398132, "train_loss_bc": 0.03521095588803291, "train_loss_llm": 0.000796884298324585, "grad_norm": 0.09647900611162186, "global_step": 29752, "epoch": 826, "lr": 0.001} +{"train_loss": 0.03313311189413071, "train_loss_bc": 0.03313311189413071, "train_loss_llm": 0.0, "grad_norm": 0.04943056032061577, "global_step": 29753, "epoch": 826, "lr": 0.001} +{"train_loss": 0.027806149795651436, "train_loss_bc": 0.027806149795651436, "train_loss_llm": 0.0, "grad_norm": 0.07387972623109818, "global_step": 29754, "epoch": 826, "lr": 0.001} +{"train_loss": 0.0357879102230072, "train_loss_bc": 0.0357879102230072, "train_loss_llm": 0.0, "grad_norm": 0.08249086141586304, "global_step": 29755, "epoch": 826, "lr": 0.001} +{"train_loss": 0.025222724303603172, "train_loss_bc": 0.025222724303603172, "train_loss_llm": 0.0, "grad_norm": 0.09313809126615524, "global_step": 29756, "epoch": 826, "lr": 0.001} +{"train_loss": 0.034913938492536545, "train_loss_bc": 0.034913938492536545, "train_loss_llm": 0.0, "grad_norm": 0.03840441256761551, "global_step": 29757, "epoch": 826, "lr": 0.001} +{"train_loss": 0.0288422629237175, "train_loss_bc": 0.028831005096435547, "train_loss_llm": 0.0011258423328399658, "grad_norm": 0.05872097611427307, "global_step": 29758, "epoch": 826, "lr": 0.001} +{"train_loss": 0.0325622633099556, "train_loss_bc": 0.0325622633099556, "train_loss_llm": 0.0, "grad_norm": 0.07576930522918701, "global_step": 29759, "epoch": 826, "lr": 0.001} +{"train_loss": 0.03202548995614052, "train_loss_bc": 0.03202548995614052, "train_loss_llm": 0.0, "grad_norm": 0.0940798670053482, "global_step": 29760, "epoch": 826, "lr": 0.001} +{"train_loss": 0.029641790315508842, "train_loss_bc": 0.029641790315508842, "train_loss_llm": 0.0, "grad_norm": 0.04319029301404953, "global_step": 29761, "epoch": 826, "lr": 0.001} +{"train_loss": 0.03397917002439499, "train_loss_bc": 0.03397337719798088, "train_loss_llm": 0.0005793273448944092, "grad_norm": 0.06345319747924805, "global_step": 29762, "epoch": 826, "lr": 0.001} +{"train_loss": 0.02281384915113449, "train_loss_bc": 0.02281384915113449, "train_loss_llm": 0.0, "grad_norm": 0.08250164985656738, "global_step": 29763, "epoch": 826, "lr": 0.001} +{"train_loss": 0.027956491336226463, "train_loss_bc": 0.027956491336226463, "train_loss_llm": 0.0, "grad_norm": 0.09990060329437256, "global_step": 29764, "epoch": 826, "lr": 0.001} +{"train_loss": 0.02843416854739189, "train_loss_bc": 0.02843416854739189, "train_loss_llm": 0.0, "grad_norm": 0.033719103783369064, "global_step": 29765, "epoch": 826, "lr": 0.001} +{"train_loss": 0.032281383872032166, "train_loss_bc": 0.032281383872032166, "train_loss_llm": 0.0, "grad_norm": 0.05136199668049812, "global_step": 29766, "epoch": 826, "lr": 0.001} +{"train_loss": 0.030187608674168587, "train_loss_bc": 0.030187608674168587, "train_loss_llm": 0.0, "grad_norm": 0.07866894453763962, "global_step": 29767, "epoch": 826, "lr": 0.001} +{"train_loss": 0.033364783972501755, "train_loss_bc": 0.03334873542189598, "train_loss_llm": 0.001604706048965454, "grad_norm": 0.09594273567199707, "global_step": 29768, "epoch": 826, "lr": 0.001} +{"train_loss": 0.02642827481031418, "train_loss_bc": 0.02642827481031418, "train_loss_llm": 0.0, "grad_norm": 0.02957337163388729, "global_step": 29769, "epoch": 826, "lr": 0.001} +{"train_loss": 0.033396072685718536, "train_loss_bc": 0.0333942174911499, "train_loss_llm": 0.00018540024757385254, "grad_norm": 0.052287906408309937, "global_step": 29770, "epoch": 826, "lr": 0.001} +{"train_loss": 0.030034548551258113, "train_loss_bc": 0.02762799896299839, "train_loss_llm": 0.0, "grad_norm": 0.07373835891485214, "global_step": 29771, "epoch": 826, "lr": 0.001, "val_loss": 0.09752790629863739} +{"train_loss": 0.025287114083766937, "train_loss_bc": 0.025287114083766937, "train_loss_llm": 0.0, "grad_norm": 0.08010073006153107, "global_step": 29772, "epoch": 827, "lr": 0.001} +{"train_loss": 0.030581234022974968, "train_loss_bc": 0.030581234022974968, "train_loss_llm": 0.0, "grad_norm": 0.033484216779470444, "global_step": 29773, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03405962139368057, "train_loss_bc": 0.03405962139368057, "train_loss_llm": 0.0, "grad_norm": 0.052822183817625046, "global_step": 29774, "epoch": 827, "lr": 0.001} +{"train_loss": 0.030865442007780075, "train_loss_bc": 0.030865442007780075, "train_loss_llm": 0.0, "grad_norm": 0.07763040065765381, "global_step": 29775, "epoch": 827, "lr": 0.001} +{"train_loss": 0.02724338322877884, "train_loss_bc": 0.02724338322877884, "train_loss_llm": 0.0, "grad_norm": 0.09657137840986252, "global_step": 29776, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03551879897713661, "train_loss_bc": 0.03551879897713661, "train_loss_llm": 0.0, "grad_norm": 0.04472794011235237, "global_step": 29777, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03432951867580414, "train_loss_bc": 0.03432951867580414, "train_loss_llm": 0.0, "grad_norm": 0.07281412184238434, "global_step": 29778, "epoch": 827, "lr": 0.001} +{"train_loss": 0.029050735756754875, "train_loss_bc": 0.029042929410934448, "train_loss_llm": 0.0007806122303009033, "grad_norm": 0.09057297557592392, "global_step": 29779, "epoch": 827, "lr": 0.001} +{"train_loss": 0.034175194799900055, "train_loss_bc": 0.034175194799900055, "train_loss_llm": 0.0, "grad_norm": 0.11367492377758026, "global_step": 29780, "epoch": 827, "lr": 0.001} +{"train_loss": 0.028772983700037003, "train_loss_bc": 0.028772983700037003, "train_loss_llm": 0.0, "grad_norm": 0.0343371145427227, "global_step": 29781, "epoch": 827, "lr": 0.001} +{"train_loss": 0.02806345373392105, "train_loss_bc": 0.028063291683793068, "train_loss_llm": 1.615285873413086e-05, "grad_norm": 0.06547977030277252, "global_step": 29782, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03175201267004013, "train_loss_bc": 0.03175201267004013, "train_loss_llm": 0.0, "grad_norm": 0.08782245218753815, "global_step": 29783, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03468940779566765, "train_loss_bc": 0.03468940779566765, "train_loss_llm": 0.0, "grad_norm": 0.10873492062091827, "global_step": 29784, "epoch": 827, "lr": 0.001} +{"train_loss": 0.029978064820170403, "train_loss_bc": 0.029965823516249657, "train_loss_llm": 0.0012240707874298096, "grad_norm": 0.051246821880340576, "global_step": 29785, "epoch": 827, "lr": 0.001} +{"train_loss": 0.02399114891886711, "train_loss_bc": 0.02399114891886711, "train_loss_llm": 0.0, "grad_norm": 0.0686885342001915, "global_step": 29786, "epoch": 827, "lr": 0.001} +{"train_loss": 0.028895754367113113, "train_loss_bc": 0.028895754367113113, "train_loss_llm": 0.0, "grad_norm": 0.09878487139940262, "global_step": 29787, "epoch": 827, "lr": 0.001} +{"train_loss": 0.02819194830954075, "train_loss_bc": 0.02819194830954075, "train_loss_llm": 0.0, "grad_norm": 0.12248910218477249, "global_step": 29788, "epoch": 827, "lr": 0.001} +{"train_loss": 0.027409587055444717, "train_loss_bc": 0.027409587055444717, "train_loss_llm": 0.0, "grad_norm": 0.028353307396173477, "global_step": 29789, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03335823863744736, "train_loss_bc": 0.03335823863744736, "train_loss_llm": 0.0, "grad_norm": 0.05282428488135338, "global_step": 29790, "epoch": 827, "lr": 0.001} +{"train_loss": 0.02885325439274311, "train_loss_bc": 0.02885325439274311, "train_loss_llm": 0.0, "grad_norm": 0.05976821854710579, "global_step": 29791, "epoch": 827, "lr": 0.001} +{"train_loss": 0.024767985567450523, "train_loss_bc": 0.024767890572547913, "train_loss_llm": 9.47713851928711e-06, "grad_norm": 0.07553426176309586, "global_step": 29792, "epoch": 827, "lr": 0.001} +{"train_loss": 0.02785150334239006, "train_loss_bc": 0.027841893956065178, "train_loss_llm": 0.00096091628074646, "grad_norm": 0.032995354384183884, "global_step": 29793, "epoch": 827, "lr": 0.001} +{"train_loss": 0.026340946555137634, "train_loss_bc": 0.026340946555137634, "train_loss_llm": 0.0, "grad_norm": 0.059414491057395935, "global_step": 29794, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03382919728755951, "train_loss_bc": 0.03382919728755951, "train_loss_llm": 0.0, "grad_norm": 0.0816647931933403, "global_step": 29795, "epoch": 827, "lr": 0.001} +{"train_loss": 0.027221040800213814, "train_loss_bc": 0.027221040800213814, "train_loss_llm": 0.0, "grad_norm": 0.09234034270048141, "global_step": 29796, "epoch": 827, "lr": 0.001} +{"train_loss": 0.0284698735922575, "train_loss_bc": 0.0284698735922575, "train_loss_llm": 0.0, "grad_norm": 0.04056228697299957, "global_step": 29797, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03162052482366562, "train_loss_bc": 0.03162052482366562, "train_loss_llm": 0.0, "grad_norm": 0.06195014715194702, "global_step": 29798, "epoch": 827, "lr": 0.001} +{"train_loss": 0.031655967235565186, "train_loss_bc": 0.03165058046579361, "train_loss_llm": 0.0005387663841247559, "grad_norm": 0.07677935063838959, "global_step": 29799, "epoch": 827, "lr": 0.001} +{"train_loss": 0.030680470168590546, "train_loss_bc": 0.03067774698138237, "train_loss_llm": 0.00027239322662353516, "grad_norm": 0.0971691682934761, "global_step": 29800, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03712720051407814, "train_loss_bc": 0.03712720051407814, "train_loss_llm": 0.0, "grad_norm": 0.043211162090301514, "global_step": 29801, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03304358199238777, "train_loss_bc": 0.03304358199238777, "train_loss_llm": 0.0, "grad_norm": 0.059072624891996384, "global_step": 29802, "epoch": 827, "lr": 0.001} +{"train_loss": 0.02574250102043152, "train_loss_bc": 0.02574250102043152, "train_loss_llm": 0.0, "grad_norm": 0.07572904229164124, "global_step": 29803, "epoch": 827, "lr": 0.001} +{"train_loss": 0.03398812562227249, "train_loss_bc": 0.03398812562227249, "train_loss_llm": 0.0, "grad_norm": 0.09432529658079147, "global_step": 29804, "epoch": 827, "lr": 0.001} +{"train_loss": 0.030739104375243187, "train_loss_bc": 0.030727356672286987, "train_loss_llm": 0.0011746883392333984, "grad_norm": 0.03717588260769844, "global_step": 29805, "epoch": 827, "lr": 0.001} +{"train_loss": 0.0306819137185812, "train_loss_bc": 0.0306819137185812, "train_loss_llm": 0.0, "grad_norm": 0.05976815149188042, "global_step": 29806, "epoch": 827, "lr": 0.001} +{"train_loss": 0.030158156632549234, "train_loss_bc": 0.02686680480837822, "train_loss_llm": 0.0, "grad_norm": 0.09867412596940994, "global_step": 29807, "epoch": 827, "lr": 0.001, "val_loss": 0.09942770004272461} +{"train_loss": 0.03797714784741402, "train_loss_bc": 0.03796970099210739, "train_loss_llm": 0.0007445812225341797, "grad_norm": 0.13392546772956848, "global_step": 29808, "epoch": 828, "lr": 0.001} +{"train_loss": 0.02294335886836052, "train_loss_bc": 0.02294033020734787, "train_loss_llm": 0.00030285120010375977, "grad_norm": 0.031733736395835876, "global_step": 29809, "epoch": 828, "lr": 0.001} +{"train_loss": 0.029614098370075226, "train_loss_bc": 0.029614098370075226, "train_loss_llm": 0.0, "grad_norm": 0.05203855410218239, "global_step": 29810, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03010014072060585, "train_loss_bc": 0.03010014072060585, "train_loss_llm": 0.0, "grad_norm": 0.07942868769168854, "global_step": 29811, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03561914712190628, "train_loss_bc": 0.03559989854693413, "train_loss_llm": 0.001924753189086914, "grad_norm": 0.1058240607380867, "global_step": 29812, "epoch": 828, "lr": 0.001} +{"train_loss": 0.029181569814682007, "train_loss_bc": 0.029181569814682007, "train_loss_llm": 0.0, "grad_norm": 0.033395878970623016, "global_step": 29813, "epoch": 828, "lr": 0.001} +{"train_loss": 0.027991104871034622, "train_loss_bc": 0.027991104871034622, "train_loss_llm": 0.0, "grad_norm": 0.04392651841044426, "global_step": 29814, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03075678087770939, "train_loss_bc": 0.03073841705918312, "train_loss_llm": 0.0018363893032073975, "grad_norm": 0.07101549208164215, "global_step": 29815, "epoch": 828, "lr": 0.001} +{"train_loss": 0.029943391680717468, "train_loss_bc": 0.029943391680717468, "train_loss_llm": 0.0, "grad_norm": 0.08602401614189148, "global_step": 29816, "epoch": 828, "lr": 0.001} +{"train_loss": 0.033241309225559235, "train_loss_bc": 0.033241309225559235, "train_loss_llm": 0.0, "grad_norm": 0.04161446541547775, "global_step": 29817, "epoch": 828, "lr": 0.001} +{"train_loss": 0.028041047975420952, "train_loss_bc": 0.028041047975420952, "train_loss_llm": 0.0, "grad_norm": 0.06152433156967163, "global_step": 29818, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03329229727387428, "train_loss_bc": 0.03328295797109604, "train_loss_llm": 0.0009339749813079834, "grad_norm": 0.07732906937599182, "global_step": 29819, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03237126022577286, "train_loss_bc": 0.03237126022577286, "train_loss_llm": 0.0, "grad_norm": 0.09953755140304565, "global_step": 29820, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03239249810576439, "train_loss_bc": 0.03238478675484657, "train_loss_llm": 0.0007710754871368408, "grad_norm": 0.033959534019231796, "global_step": 29821, "epoch": 828, "lr": 0.001} +{"train_loss": 0.026659836992621422, "train_loss_bc": 0.026659836992621422, "train_loss_llm": 0.0, "grad_norm": 0.051716774702072144, "global_step": 29822, "epoch": 828, "lr": 0.001} +{"train_loss": 0.029268961399793625, "train_loss_bc": 0.029268961399793625, "train_loss_llm": 0.0, "grad_norm": 0.07138435542583466, "global_step": 29823, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03147987276315689, "train_loss_bc": 0.03147987276315689, "train_loss_llm": 0.0, "grad_norm": 0.0941084623336792, "global_step": 29824, "epoch": 828, "lr": 0.001} +{"train_loss": 0.027945615351200104, "train_loss_bc": 0.027945615351200104, "train_loss_llm": 0.0, "grad_norm": 0.03828246518969536, "global_step": 29825, "epoch": 828, "lr": 0.001} +{"train_loss": 0.022852571681141853, "train_loss_bc": 0.022852571681141853, "train_loss_llm": 0.0, "grad_norm": 0.05902281031012535, "global_step": 29826, "epoch": 828, "lr": 0.001} +{"train_loss": 0.029426993802189827, "train_loss_bc": 0.029394522309303284, "train_loss_llm": 0.0032471418380737305, "grad_norm": 0.06728411465883255, "global_step": 29827, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03191941976547241, "train_loss_bc": 0.03191941976547241, "train_loss_llm": 0.0, "grad_norm": 0.09578491747379303, "global_step": 29828, "epoch": 828, "lr": 0.001} +{"train_loss": 0.030125297605991364, "train_loss_bc": 0.030114764347672462, "train_loss_llm": 0.0010533630847930908, "grad_norm": 0.03966909646987915, "global_step": 29829, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03542080149054527, "train_loss_bc": 0.03541312366724014, "train_loss_llm": 0.0007677972316741943, "grad_norm": 0.08016681671142578, "global_step": 29830, "epoch": 828, "lr": 0.001} +{"train_loss": 0.032327800989151, "train_loss_bc": 0.032327800989151, "train_loss_llm": 0.0, "grad_norm": 0.10569281876087189, "global_step": 29831, "epoch": 828, "lr": 0.001} +{"train_loss": 0.02904774434864521, "train_loss_bc": 0.02904774434864521, "train_loss_llm": 0.0, "grad_norm": 0.1306028664112091, "global_step": 29832, "epoch": 828, "lr": 0.001} +{"train_loss": 0.027095668017864227, "train_loss_bc": 0.02708945795893669, "train_loss_llm": 0.0006210505962371826, "grad_norm": 0.029758865013718605, "global_step": 29833, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03193524852395058, "train_loss_bc": 0.03193524852395058, "train_loss_llm": 0.0, "grad_norm": 0.06893780827522278, "global_step": 29834, "epoch": 828, "lr": 0.001} +{"train_loss": 0.030574852600693703, "train_loss_bc": 0.030572377145290375, "train_loss_llm": 0.00024762749671936035, "grad_norm": 0.104835145175457, "global_step": 29835, "epoch": 828, "lr": 0.001} +{"train_loss": 0.03055819496512413, "train_loss_bc": 0.03055819496512413, "train_loss_llm": 0.0, "grad_norm": 0.1320495307445526, "global_step": 29836, "epoch": 828, "lr": 0.001} +{"train_loss": 0.035220079123973846, "train_loss_bc": 0.035220079123973846, "train_loss_llm": 0.0, "grad_norm": 0.04255170747637749, "global_step": 29837, "epoch": 828, "lr": 0.001} +{"train_loss": 0.022211462259292603, "train_loss_bc": 0.022211462259292603, "train_loss_llm": 0.0, "grad_norm": 0.06259234249591827, "global_step": 29838, "epoch": 828, "lr": 0.001} +{"train_loss": 0.02878182753920555, "train_loss_bc": 0.02878182753920555, "train_loss_llm": 0.0, "grad_norm": 0.08100000768899918, "global_step": 29839, "epoch": 828, "lr": 0.001} +{"train_loss": 0.02820865996181965, "train_loss_bc": 0.028193173930048943, "train_loss_llm": 0.0015485882759094238, "grad_norm": 0.10023127496242523, "global_step": 29840, "epoch": 828, "lr": 0.001} +{"train_loss": 0.030558835715055466, "train_loss_bc": 0.030558835715055466, "train_loss_llm": 0.0, "grad_norm": 0.03780251741409302, "global_step": 29841, "epoch": 828, "lr": 0.001} +{"train_loss": 0.035119105130434036, "train_loss_bc": 0.035119105130434036, "train_loss_llm": 0.0, "grad_norm": 0.06669885665178299, "global_step": 29842, "epoch": 828, "lr": 0.001} +{"train_loss": 0.030312501049290102, "train_loss_bc": 0.031046034768223763, "train_loss_llm": 0.0, "grad_norm": 0.08512691408395767, "global_step": 29843, "epoch": 828, "lr": 0.001, "val_loss": 0.09819687157869339} +{"train_loss": 0.03173305466771126, "train_loss_bc": 0.03173305466771126, "train_loss_llm": 0.0, "grad_norm": 0.11729912459850311, "global_step": 29844, "epoch": 829, "lr": 0.001} +{"train_loss": 0.029480597004294395, "train_loss_bc": 0.029480597004294395, "train_loss_llm": 0.0, "grad_norm": 0.038823164999485016, "global_step": 29845, "epoch": 829, "lr": 0.001} +{"train_loss": 0.027067061513662338, "train_loss_bc": 0.027067061513662338, "train_loss_llm": 0.0, "grad_norm": 0.05106654763221741, "global_step": 29846, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03177846595644951, "train_loss_bc": 0.03177846595644951, "train_loss_llm": 0.0, "grad_norm": 0.06892029941082001, "global_step": 29847, "epoch": 829, "lr": 0.001} +{"train_loss": 0.0268443264067173, "train_loss_bc": 0.0268443264067173, "train_loss_llm": 0.0, "grad_norm": 0.08343242108821869, "global_step": 29848, "epoch": 829, "lr": 0.001} +{"train_loss": 0.034919623285532, "train_loss_bc": 0.0349164754152298, "train_loss_llm": 0.0003148019313812256, "grad_norm": 0.03904164955019951, "global_step": 29849, "epoch": 829, "lr": 0.001} +{"train_loss": 0.029668154194951057, "train_loss_bc": 0.029668154194951057, "train_loss_llm": 0.0, "grad_norm": 0.061552952975034714, "global_step": 29850, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03220941871404648, "train_loss_bc": 0.03220410272479057, "train_loss_llm": 0.0005314648151397705, "grad_norm": 0.0828954428434372, "global_step": 29851, "epoch": 829, "lr": 0.001} +{"train_loss": 0.02624768577516079, "train_loss_bc": 0.02624768577516079, "train_loss_llm": 0.0, "grad_norm": 0.10681042820215225, "global_step": 29852, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03245958313345909, "train_loss_bc": 0.03243215009570122, "train_loss_llm": 0.0027431249618530273, "grad_norm": 0.043834537267684937, "global_step": 29853, "epoch": 829, "lr": 0.001} +{"train_loss": 0.0307577233761549, "train_loss_bc": 0.0307577233761549, "train_loss_llm": 0.0, "grad_norm": 0.06515617668628693, "global_step": 29854, "epoch": 829, "lr": 0.001} +{"train_loss": 0.029995357617735863, "train_loss_bc": 0.029985254630446434, "train_loss_llm": 0.0010103583335876465, "grad_norm": 0.09310752898454666, "global_step": 29855, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03469523414969444, "train_loss_bc": 0.03469523414969444, "train_loss_llm": 0.0, "grad_norm": 0.11869997531175613, "global_step": 29856, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03196702152490616, "train_loss_bc": 0.03196702152490616, "train_loss_llm": 0.0, "grad_norm": 0.030981777235865593, "global_step": 29857, "epoch": 829, "lr": 0.001} +{"train_loss": 0.030375147238373756, "train_loss_bc": 0.030375147238373756, "train_loss_llm": 0.0, "grad_norm": 0.05307498574256897, "global_step": 29858, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03397050127387047, "train_loss_bc": 0.03397050127387047, "train_loss_llm": 0.0, "grad_norm": 0.07881691306829453, "global_step": 29859, "epoch": 829, "lr": 0.001} +{"train_loss": 0.02430298924446106, "train_loss_bc": 0.02430298924446106, "train_loss_llm": 0.0, "grad_norm": 0.09122682362794876, "global_step": 29860, "epoch": 829, "lr": 0.001} +{"train_loss": 0.031279198825359344, "train_loss_bc": 0.031279198825359344, "train_loss_llm": 0.0, "grad_norm": 0.034091390669345856, "global_step": 29861, "epoch": 829, "lr": 0.001} +{"train_loss": 0.02429976314306259, "train_loss_bc": 0.02429976314306259, "train_loss_llm": 0.0, "grad_norm": 0.05661169812083244, "global_step": 29862, "epoch": 829, "lr": 0.001} +{"train_loss": 0.025932129472494125, "train_loss_bc": 0.025932129472494125, "train_loss_llm": 0.0, "grad_norm": 0.08428733795881271, "global_step": 29863, "epoch": 829, "lr": 0.001} +{"train_loss": 0.030361996963620186, "train_loss_bc": 0.030361996963620186, "train_loss_llm": 0.0, "grad_norm": 0.10052995383739471, "global_step": 29864, "epoch": 829, "lr": 0.001} +{"train_loss": 0.029700571671128273, "train_loss_bc": 0.029700571671128273, "train_loss_llm": 0.0, "grad_norm": 0.041051287204027176, "global_step": 29865, "epoch": 829, "lr": 0.001} +{"train_loss": 0.023977739736437798, "train_loss_bc": 0.023949138820171356, "train_loss_llm": 0.0028600096702575684, "grad_norm": 0.054563794285058975, "global_step": 29866, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03755844011902809, "train_loss_bc": 0.03755844011902809, "train_loss_llm": 0.0, "grad_norm": 0.07371806353330612, "global_step": 29867, "epoch": 829, "lr": 0.001} +{"train_loss": 0.02596733719110489, "train_loss_bc": 0.02596733719110489, "train_loss_llm": 0.0, "grad_norm": 0.0898967981338501, "global_step": 29868, "epoch": 829, "lr": 0.001} +{"train_loss": 0.030711166560649872, "train_loss_bc": 0.030703797936439514, "train_loss_llm": 0.0007367730140686035, "grad_norm": 0.04567066952586174, "global_step": 29869, "epoch": 829, "lr": 0.001} +{"train_loss": 0.02689521387219429, "train_loss_bc": 0.02689521387219429, "train_loss_llm": 0.0, "grad_norm": 0.06631296128034592, "global_step": 29870, "epoch": 829, "lr": 0.001} +{"train_loss": 0.028712091967463493, "train_loss_bc": 0.02870921604335308, "train_loss_llm": 0.00028768181800842285, "grad_norm": 0.08913533389568329, "global_step": 29871, "epoch": 829, "lr": 0.001} +{"train_loss": 0.023356415331363678, "train_loss_bc": 0.023356415331363678, "train_loss_llm": 0.0, "grad_norm": 0.10829705744981766, "global_step": 29872, "epoch": 829, "lr": 0.001} +{"train_loss": 0.029539605602622032, "train_loss_bc": 0.029526887461543083, "train_loss_llm": 0.001271754503250122, "grad_norm": 0.03240743651986122, "global_step": 29873, "epoch": 829, "lr": 0.001} +{"train_loss": 0.031666066497564316, "train_loss_bc": 0.031666066497564316, "train_loss_llm": 0.0, "grad_norm": 0.061859991401433945, "global_step": 29874, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03494562581181526, "train_loss_bc": 0.03494562581181526, "train_loss_llm": 0.0, "grad_norm": 0.09577514231204987, "global_step": 29875, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03492759168148041, "train_loss_bc": 0.03492759168148041, "train_loss_llm": 0.0, "grad_norm": 0.11752073466777802, "global_step": 29876, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03181357681751251, "train_loss_bc": 0.031809817999601364, "train_loss_llm": 0.00037598609924316406, "grad_norm": 0.03798476234078407, "global_step": 29877, "epoch": 829, "lr": 0.001} +{"train_loss": 0.03669669106602669, "train_loss_bc": 0.03669669106602669, "train_loss_llm": 0.0, "grad_norm": 0.07153186202049255, "global_step": 29878, "epoch": 829, "lr": 0.001} +{"train_loss": 0.030208231105158728, "train_loss_bc": 0.030683152377605438, "train_loss_llm": 0.0, "grad_norm": 0.08983200788497925, "global_step": 29879, "epoch": 829, "lr": 0.001, "val_loss": 0.0998641699552536} +{"train_loss": 0.02965298853814602, "train_loss_bc": 0.02965298853814602, "train_loss_llm": 0.0, "grad_norm": 0.10916609317064285, "global_step": 29880, "epoch": 830, "lr": 0.001} +{"train_loss": 0.029436811804771423, "train_loss_bc": 0.029436811804771423, "train_loss_llm": 0.0, "grad_norm": 0.03634331002831459, "global_step": 29881, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03457950800657272, "train_loss_bc": 0.03457950800657272, "train_loss_llm": 0.0, "grad_norm": 0.06367479264736176, "global_step": 29882, "epoch": 830, "lr": 0.001} +{"train_loss": 0.02414519526064396, "train_loss_bc": 0.02414519526064396, "train_loss_llm": 0.0, "grad_norm": 0.0883907899260521, "global_step": 29883, "epoch": 830, "lr": 0.001} +{"train_loss": 0.02898251824080944, "train_loss_bc": 0.028977015987038612, "train_loss_llm": 0.0005502700805664062, "grad_norm": 0.11030664294958115, "global_step": 29884, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03196698799729347, "train_loss_bc": 0.031962715089321136, "train_loss_llm": 0.0004272758960723877, "grad_norm": 0.045896727591753006, "global_step": 29885, "epoch": 830, "lr": 0.001} +{"train_loss": 0.032645851373672485, "train_loss_bc": 0.032645851373672485, "train_loss_llm": 0.0, "grad_norm": 0.07990901172161102, "global_step": 29886, "epoch": 830, "lr": 0.001} +{"train_loss": 0.034542299807071686, "train_loss_bc": 0.034542299807071686, "train_loss_llm": 0.0, "grad_norm": 0.09685249626636505, "global_step": 29887, "epoch": 830, "lr": 0.001} +{"train_loss": 0.02794792316854, "train_loss_bc": 0.027921784669160843, "train_loss_llm": 0.0026138722896575928, "grad_norm": 0.1167583018541336, "global_step": 29888, "epoch": 830, "lr": 0.001} +{"train_loss": 0.02616880275309086, "train_loss_bc": 0.02616880275309086, "train_loss_llm": 0.0, "grad_norm": 0.0327799953520298, "global_step": 29889, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03594215214252472, "train_loss_bc": 0.03594215214252472, "train_loss_llm": 0.0, "grad_norm": 0.047541163861751556, "global_step": 29890, "epoch": 830, "lr": 0.001} +{"train_loss": 0.024927306920289993, "train_loss_bc": 0.024927306920289993, "train_loss_llm": 0.0, "grad_norm": 0.07013475894927979, "global_step": 29891, "epoch": 830, "lr": 0.001} +{"train_loss": 0.030978886410593987, "train_loss_bc": 0.030962888151407242, "train_loss_llm": 0.0015997886657714844, "grad_norm": 0.08463915437459946, "global_step": 29892, "epoch": 830, "lr": 0.001} +{"train_loss": 0.029143262654542923, "train_loss_bc": 0.029143262654542923, "train_loss_llm": 0.0, "grad_norm": 0.039463452994823456, "global_step": 29893, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03297920525074005, "train_loss_bc": 0.03297748044133186, "train_loss_llm": 0.00017258524894714355, "grad_norm": 0.06842595338821411, "global_step": 29894, "epoch": 830, "lr": 0.001} +{"train_loss": 0.027547350153326988, "train_loss_bc": 0.027547350153326988, "train_loss_llm": 0.0, "grad_norm": 0.08530458062887192, "global_step": 29895, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03243708610534668, "train_loss_bc": 0.03243708610534668, "train_loss_llm": 0.0, "grad_norm": 0.10223110020160675, "global_step": 29896, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03214779496192932, "train_loss_bc": 0.03214779496192932, "train_loss_llm": 0.0, "grad_norm": 0.041583333164453506, "global_step": 29897, "epoch": 830, "lr": 0.001} +{"train_loss": 0.02487989515066147, "train_loss_bc": 0.024867340922355652, "train_loss_llm": 0.001255422830581665, "grad_norm": 0.04793146997690201, "global_step": 29898, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03478242829442024, "train_loss_bc": 0.03478242829442024, "train_loss_llm": 0.0, "grad_norm": 0.06599551439285278, "global_step": 29899, "epoch": 830, "lr": 0.001} +{"train_loss": 0.028661228716373444, "train_loss_bc": 0.028661228716373444, "train_loss_llm": 0.0, "grad_norm": 0.07600920647382736, "global_step": 29900, "epoch": 830, "lr": 0.001} +{"train_loss": 0.029895741492509842, "train_loss_bc": 0.029895741492509842, "train_loss_llm": 0.0, "grad_norm": 0.033017221838235855, "global_step": 29901, "epoch": 830, "lr": 0.001} +{"train_loss": 0.032749321311712265, "train_loss_bc": 0.032749321311712265, "train_loss_llm": 0.0, "grad_norm": 0.0492066890001297, "global_step": 29902, "epoch": 830, "lr": 0.001} +{"train_loss": 0.02651996538043022, "train_loss_bc": 0.026502292603254318, "train_loss_llm": 0.0017673373222351074, "grad_norm": 0.06497971713542938, "global_step": 29903, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03123939409852028, "train_loss_bc": 0.03123939409852028, "train_loss_llm": 0.0, "grad_norm": 0.08698055893182755, "global_step": 29904, "epoch": 830, "lr": 0.001} +{"train_loss": 0.031057700514793396, "train_loss_bc": 0.031057700514793396, "train_loss_llm": 0.0, "grad_norm": 0.03322489932179451, "global_step": 29905, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03194890171289444, "train_loss_bc": 0.03194890171289444, "train_loss_llm": 0.0, "grad_norm": 0.06255842745304108, "global_step": 29906, "epoch": 830, "lr": 0.001} +{"train_loss": 0.02854539453983307, "train_loss_bc": 0.02854539453983307, "train_loss_llm": 0.0, "grad_norm": 0.08855006098747253, "global_step": 29907, "epoch": 830, "lr": 0.001} +{"train_loss": 0.026514511555433273, "train_loss_bc": 0.026514511555433273, "train_loss_llm": 0.0, "grad_norm": 0.11999479681253433, "global_step": 29908, "epoch": 830, "lr": 0.001} +{"train_loss": 0.035246968269348145, "train_loss_bc": 0.035246968269348145, "train_loss_llm": 0.0, "grad_norm": 0.03786233067512512, "global_step": 29909, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03191332891583443, "train_loss_bc": 0.03191332891583443, "train_loss_llm": 0.0, "grad_norm": 0.05849633365869522, "global_step": 29910, "epoch": 830, "lr": 0.001} +{"train_loss": 0.026339925825595856, "train_loss_bc": 0.026339925825595856, "train_loss_llm": 0.0, "grad_norm": 0.08328346908092499, "global_step": 29911, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03730538859963417, "train_loss_bc": 0.03730538859963417, "train_loss_llm": 0.0, "grad_norm": 0.10459943115711212, "global_step": 29912, "epoch": 830, "lr": 0.001} +{"train_loss": 0.0283762626349926, "train_loss_bc": 0.0283762626349926, "train_loss_llm": 0.0, "grad_norm": 0.036795441061258316, "global_step": 29913, "epoch": 830, "lr": 0.001} +{"train_loss": 0.023524001240730286, "train_loss_bc": 0.023524001240730286, "train_loss_llm": 0.0, "grad_norm": 0.053549572825431824, "global_step": 29914, "epoch": 830, "lr": 0.001} +{"train_loss": 0.03009896668502026, "train_loss_bc": 0.027890510857105255, "train_loss_llm": 0.0, "grad_norm": 0.06746076792478561, "global_step": 29915, "epoch": 830, "lr": 0.001, "train/cumulative_reward": 61.26035825576164, "train/mean_score": 11.716130540224874, "train/success_rate": 0.3, "test/cumulative_reward": 47.7851325324269, "test/mean_score": 7.43500208711654, "test/success_rate": 0.16, "val_loss": 0.09973926842212677, "train_action_mse_error": 0.032759249210357666} +{"train_loss": 0.027916042134165764, "train_loss_bc": 0.027916042134165764, "train_loss_llm": 0.0, "grad_norm": 0.07680246233940125, "global_step": 29916, "epoch": 831, "lr": 0.001} +{"train_loss": 0.029314078390598297, "train_loss_bc": 0.02931145578622818, "train_loss_llm": 0.00026226043701171875, "grad_norm": 0.037302788347005844, "global_step": 29917, "epoch": 831, "lr": 0.001} +{"train_loss": 0.03151240199804306, "train_loss_bc": 0.03151093050837517, "train_loss_llm": 0.00014707446098327637, "grad_norm": 0.057271767407655716, "global_step": 29918, "epoch": 831, "lr": 0.001} +{"train_loss": 0.03435294330120087, "train_loss_bc": 0.03435294330120087, "train_loss_llm": 0.0, "grad_norm": 0.07722130417823792, "global_step": 29919, "epoch": 831, "lr": 0.001} +{"train_loss": 0.02620675228536129, "train_loss_bc": 0.02620675228536129, "train_loss_llm": 0.0, "grad_norm": 0.09295012056827545, "global_step": 29920, "epoch": 831, "lr": 0.001} +{"train_loss": 0.03049778938293457, "train_loss_bc": 0.03049778938293457, "train_loss_llm": 0.0, "grad_norm": 0.03322090953588486, "global_step": 29921, "epoch": 831, "lr": 0.001} +{"train_loss": 0.02628207765519619, "train_loss_bc": 0.02628207765519619, "train_loss_llm": 0.0, "grad_norm": 0.054176222532987595, "global_step": 29922, "epoch": 831, "lr": 0.001} +{"train_loss": 0.031563472002744675, "train_loss_bc": 0.0315571129322052, "train_loss_llm": 0.0006360709667205811, "grad_norm": 0.07650420814752579, "global_step": 29923, "epoch": 831, "lr": 0.001} +{"train_loss": 0.026266297325491905, "train_loss_bc": 0.026266297325491905, "train_loss_llm": 0.0, "grad_norm": 0.09212883561849594, "global_step": 29924, "epoch": 831, "lr": 0.001} +{"train_loss": 0.031628385186195374, "train_loss_bc": 0.031628385186195374, "train_loss_llm": 0.0, "grad_norm": 0.03648876026272774, "global_step": 29925, "epoch": 831, "lr": 0.001} +{"train_loss": 0.027700429782271385, "train_loss_bc": 0.027700429782271385, "train_loss_llm": 0.0, "grad_norm": 0.05938010290265083, "global_step": 29926, "epoch": 831, "lr": 0.001} +{"train_loss": 0.031035486608743668, "train_loss_bc": 0.031035486608743668, "train_loss_llm": 0.0, "grad_norm": 0.08484378457069397, "global_step": 29927, "epoch": 831, "lr": 0.001} +{"train_loss": 0.030370602384209633, "train_loss_bc": 0.030370602384209633, "train_loss_llm": 0.0, "grad_norm": 0.09416709840297699, "global_step": 29928, "epoch": 831, "lr": 0.001} +{"train_loss": 0.031809788197278976, "train_loss_bc": 0.031809788197278976, "train_loss_llm": 0.0, "grad_norm": 0.03571672737598419, "global_step": 29929, "epoch": 831, "lr": 0.001} +{"train_loss": 0.03343065455555916, "train_loss_bc": 0.0334232859313488, "train_loss_llm": 0.0007368624210357666, "grad_norm": 0.06849776208400726, "global_step": 29930, "epoch": 831, "lr": 0.001} +{"train_loss": 0.028997093439102173, "train_loss_bc": 0.028997093439102173, "train_loss_llm": 0.0, "grad_norm": 0.10127631574869156, "global_step": 29931, "epoch": 831, "lr": 0.001} +{"train_loss": 0.02809229865670204, "train_loss_bc": 0.02809229865670204, "train_loss_llm": 0.0, "grad_norm": 0.11679746210575104, "global_step": 29932, "epoch": 831, "lr": 0.001} +{"train_loss": 0.030611664056777954, "train_loss_bc": 0.030604248866438866, "train_loss_llm": 0.0007416009902954102, "grad_norm": 0.04493492469191551, "global_step": 29933, "epoch": 831, "lr": 0.001} +{"train_loss": 0.030571140348911285, "train_loss_bc": 0.030571140348911285, "train_loss_llm": 0.0, "grad_norm": 0.05609028413891792, "global_step": 29934, "epoch": 831, "lr": 0.001} +{"train_loss": 0.024889983236789703, "train_loss_bc": 0.024889983236789703, "train_loss_llm": 0.0, "grad_norm": 0.0655960887670517, "global_step": 29935, "epoch": 831, "lr": 0.001} +{"train_loss": 0.031065210700035095, "train_loss_bc": 0.031061803922057152, "train_loss_llm": 0.0003407299518585205, "grad_norm": 0.08406181633472443, "global_step": 29936, "epoch": 831, "lr": 0.001} +{"train_loss": 0.024424469098448753, "train_loss_bc": 0.024423930794000626, "train_loss_llm": 5.385279655456543e-05, "grad_norm": 0.03383632376790047, "global_step": 29937, "epoch": 831, "lr": 0.001} +{"train_loss": 0.02940969169139862, "train_loss_bc": 0.02940366230905056, "train_loss_llm": 0.0006029307842254639, "grad_norm": 0.06698986142873764, "global_step": 29938, "epoch": 831, "lr": 0.001} +{"train_loss": 0.030304357409477234, "train_loss_bc": 0.030304357409477234, "train_loss_llm": 0.0, "grad_norm": 0.08723261207342148, "global_step": 29939, "epoch": 831, "lr": 0.001} +{"train_loss": 0.028442082926630974, "train_loss_bc": 0.02843453362584114, "train_loss_llm": 0.0007548928260803223, "grad_norm": 0.09875886142253876, "global_step": 29940, "epoch": 831, "lr": 0.001} +{"train_loss": 0.02928503043949604, "train_loss_bc": 0.02928503043949604, "train_loss_llm": 0.0, "grad_norm": 0.03243377059698105, "global_step": 29941, "epoch": 831, "lr": 0.001} +{"train_loss": 0.03230055794119835, "train_loss_bc": 0.03230055794119835, "train_loss_llm": 0.0, "grad_norm": 0.05603506788611412, "global_step": 29942, "epoch": 831, "lr": 0.001} +{"train_loss": 0.02905896306037903, "train_loss_bc": 0.0290533397346735, "train_loss_llm": 0.0005623996257781982, "grad_norm": 0.09367541968822479, "global_step": 29943, "epoch": 831, "lr": 0.001} +{"train_loss": 0.03167026489973068, "train_loss_bc": 0.03167026489973068, "train_loss_llm": 0.0, "grad_norm": 0.12335560470819473, "global_step": 29944, "epoch": 831, "lr": 0.001} +{"train_loss": 0.031275033950805664, "train_loss_bc": 0.031275033950805664, "train_loss_llm": 0.0, "grad_norm": 0.04501160606741905, "global_step": 29945, "epoch": 831, "lr": 0.001} +{"train_loss": 0.03574167191982269, "train_loss_bc": 0.03574167191982269, "train_loss_llm": 0.0, "grad_norm": 0.07935775071382523, "global_step": 29946, "epoch": 831, "lr": 0.001} +{"train_loss": 0.03636303171515465, "train_loss_bc": 0.03636303171515465, "train_loss_llm": 0.0, "grad_norm": 0.09186995774507523, "global_step": 29947, "epoch": 831, "lr": 0.001} +{"train_loss": 0.031084338203072548, "train_loss_bc": 0.031084338203072548, "train_loss_llm": 0.0, "grad_norm": 0.10466106981039047, "global_step": 29948, "epoch": 831, "lr": 0.001} +{"train_loss": 0.03179527074098587, "train_loss_bc": 0.03179527074098587, "train_loss_llm": 0.0, "grad_norm": 0.036414287984371185, "global_step": 29949, "epoch": 831, "lr": 0.001} +{"train_loss": 0.025807389989495277, "train_loss_bc": 0.025807389989495277, "train_loss_llm": 0.0, "grad_norm": 0.05859699472784996, "global_step": 29950, "epoch": 831, "lr": 0.001} +{"train_loss": 0.029973133311917383, "train_loss_bc": 0.027956053614616394, "train_loss_llm": 0.0, "grad_norm": 0.09028010815382004, "global_step": 29951, "epoch": 831, "lr": 0.001, "val_loss": 0.09945308417081833} +{"train_loss": 0.02403675764799118, "train_loss_bc": 0.02403675764799118, "train_loss_llm": 0.0, "grad_norm": 0.10822027176618576, "global_step": 29952, "epoch": 832, "lr": 0.001} +{"train_loss": 0.031183045357465744, "train_loss_bc": 0.031174881383776665, "train_loss_llm": 0.0008164644241333008, "grad_norm": 0.03668377175927162, "global_step": 29953, "epoch": 832, "lr": 0.001} +{"train_loss": 0.025617675855755806, "train_loss_bc": 0.025617675855755806, "train_loss_llm": 0.0, "grad_norm": 0.06014097109436989, "global_step": 29954, "epoch": 832, "lr": 0.001} +{"train_loss": 0.03039187751710415, "train_loss_bc": 0.03039187751710415, "train_loss_llm": 0.0, "grad_norm": 0.07710493355989456, "global_step": 29955, "epoch": 832, "lr": 0.001} +{"train_loss": 0.032945070415735245, "train_loss_bc": 0.032945070415735245, "train_loss_llm": 0.0, "grad_norm": 0.09192956238985062, "global_step": 29956, "epoch": 832, "lr": 0.001} +{"train_loss": 0.02584681287407875, "train_loss_bc": 0.025843512266874313, "train_loss_llm": 0.0003300905227661133, "grad_norm": 0.03160813823342323, "global_step": 29957, "epoch": 832, "lr": 0.001} +{"train_loss": 0.030846042558550835, "train_loss_bc": 0.030825840309262276, "train_loss_llm": 0.0020202696323394775, "grad_norm": 0.05921649932861328, "global_step": 29958, "epoch": 832, "lr": 0.001} +{"train_loss": 0.032289404422044754, "train_loss_bc": 0.0322863943874836, "train_loss_llm": 0.0003008842468261719, "grad_norm": 0.07232654094696045, "global_step": 29959, "epoch": 832, "lr": 0.001} +{"train_loss": 0.02515958994626999, "train_loss_bc": 0.02515958994626999, "train_loss_llm": 0.0, "grad_norm": 0.09349066019058228, "global_step": 29960, "epoch": 832, "lr": 0.001} +{"train_loss": 0.026799026876688004, "train_loss_bc": 0.026799026876688004, "train_loss_llm": 0.0, "grad_norm": 0.03721483051776886, "global_step": 29961, "epoch": 832, "lr": 0.001} +{"train_loss": 0.024681521579623222, "train_loss_bc": 0.024665730074048042, "train_loss_llm": 0.0015791654586791992, "grad_norm": 0.060458533465862274, "global_step": 29962, "epoch": 832, "lr": 0.001} +{"train_loss": 0.029075225815176964, "train_loss_bc": 0.029075225815176964, "train_loss_llm": 0.0, "grad_norm": 0.08122754842042923, "global_step": 29963, "epoch": 832, "lr": 0.001} +{"train_loss": 0.030329415574669838, "train_loss_bc": 0.030329415574669838, "train_loss_llm": 0.0, "grad_norm": 0.10282260179519653, "global_step": 29964, "epoch": 832, "lr": 0.001} +{"train_loss": 0.033376455307006836, "train_loss_bc": 0.033376455307006836, "train_loss_llm": 0.0, "grad_norm": 0.039300691336393356, "global_step": 29965, "epoch": 832, "lr": 0.001} +{"train_loss": 0.0369936004281044, "train_loss_bc": 0.0369936004281044, "train_loss_llm": 0.0, "grad_norm": 0.07329323887825012, "global_step": 29966, "epoch": 832, "lr": 0.001} +{"train_loss": 0.027342628687620163, "train_loss_bc": 0.027340587228536606, "train_loss_llm": 0.0002041161060333252, "grad_norm": 0.10019652545452118, "global_step": 29967, "epoch": 832, "lr": 0.001} +{"train_loss": 0.033733345568180084, "train_loss_bc": 0.033731717616319656, "train_loss_llm": 0.0001626908779144287, "grad_norm": 0.12305603921413422, "global_step": 29968, "epoch": 832, "lr": 0.001} +{"train_loss": 0.028316009789705276, "train_loss_bc": 0.028316009789705276, "train_loss_llm": 0.0, "grad_norm": 0.03484935313463211, "global_step": 29969, "epoch": 832, "lr": 0.001} +{"train_loss": 0.028138119727373123, "train_loss_bc": 0.028138119727373123, "train_loss_llm": 0.0, "grad_norm": 0.06093289703130722, "global_step": 29970, "epoch": 832, "lr": 0.001} +{"train_loss": 0.0345202311873436, "train_loss_bc": 0.0345202311873436, "train_loss_llm": 0.0, "grad_norm": 0.08380943536758423, "global_step": 29971, "epoch": 832, "lr": 0.001} +{"train_loss": 0.030273767188191414, "train_loss_bc": 0.03023710660636425, "train_loss_llm": 0.0036660730838775635, "grad_norm": 0.1042269691824913, "global_step": 29972, "epoch": 832, "lr": 0.001} +{"train_loss": 0.03243480250239372, "train_loss_bc": 0.03243480250239372, "train_loss_llm": 0.0, "grad_norm": 0.03646554425358772, "global_step": 29973, "epoch": 832, "lr": 0.001} +{"train_loss": 0.03348764777183533, "train_loss_bc": 0.03346902132034302, "train_loss_llm": 0.0018626153469085693, "grad_norm": 0.058407559990882874, "global_step": 29974, "epoch": 832, "lr": 0.001} +{"train_loss": 0.029373539611697197, "train_loss_bc": 0.029369687661528587, "train_loss_llm": 0.0003851354122161865, "grad_norm": 0.07832148671150208, "global_step": 29975, "epoch": 832, "lr": 0.001} +{"train_loss": 0.03132638335227966, "train_loss_bc": 0.03132638335227966, "train_loss_llm": 0.0, "grad_norm": 0.08557279407978058, "global_step": 29976, "epoch": 832, "lr": 0.001} +{"train_loss": 0.029703497886657715, "train_loss_bc": 0.029684636741876602, "train_loss_llm": 0.0018860995769500732, "grad_norm": 0.039880167692899704, "global_step": 29977, "epoch": 832, "lr": 0.001} +{"train_loss": 0.03244081884622574, "train_loss_bc": 0.03244081884622574, "train_loss_llm": 0.0, "grad_norm": 0.05651380494236946, "global_step": 29978, "epoch": 832, "lr": 0.001} +{"train_loss": 0.0310178454965353, "train_loss_bc": 0.0310178454965353, "train_loss_llm": 0.0, "grad_norm": 0.07012949883937836, "global_step": 29979, "epoch": 832, "lr": 0.001} +{"train_loss": 0.034018680453300476, "train_loss_bc": 0.034018680453300476, "train_loss_llm": 0.0, "grad_norm": 0.0891120582818985, "global_step": 29980, "epoch": 832, "lr": 0.001} +{"train_loss": 0.026633556932210922, "train_loss_bc": 0.0266097579151392, "train_loss_llm": 0.002379864454269409, "grad_norm": 0.03934247046709061, "global_step": 29981, "epoch": 832, "lr": 0.001} +{"train_loss": 0.028162380680441856, "train_loss_bc": 0.028157245367765427, "train_loss_llm": 0.0005135834217071533, "grad_norm": 0.07092425227165222, "global_step": 29982, "epoch": 832, "lr": 0.001} +{"train_loss": 0.028410423547029495, "train_loss_bc": 0.028410423547029495, "train_loss_llm": 0.0, "grad_norm": 0.09960965812206268, "global_step": 29983, "epoch": 832, "lr": 0.001} +{"train_loss": 0.025665536522865295, "train_loss_bc": 0.025665536522865295, "train_loss_llm": 0.0, "grad_norm": 0.1185656487941742, "global_step": 29984, "epoch": 832, "lr": 0.001} +{"train_loss": 0.03257519751787186, "train_loss_bc": 0.03257519751787186, "train_loss_llm": 0.0, "grad_norm": 0.0336468406021595, "global_step": 29985, "epoch": 832, "lr": 0.001} +{"train_loss": 0.03486157953739166, "train_loss_bc": 0.03486157953739166, "train_loss_llm": 0.0, "grad_norm": 0.059618666768074036, "global_step": 29986, "epoch": 832, "lr": 0.001} +{"train_loss": 0.030183773901727464, "train_loss_bc": 0.03460834547877312, "train_loss_llm": 0.0, "grad_norm": 0.08150601387023926, "global_step": 29987, "epoch": 832, "lr": 0.001, "val_loss": 0.09902576357126236} +{"train_loss": 0.02786695770919323, "train_loss_bc": 0.02786695770919323, "train_loss_llm": 0.0, "grad_norm": 0.09977728128433228, "global_step": 29988, "epoch": 833, "lr": 0.001} +{"train_loss": 0.02768821455538273, "train_loss_bc": 0.02768821455538273, "train_loss_llm": 0.0, "grad_norm": 0.039243441075086594, "global_step": 29989, "epoch": 833, "lr": 0.001} +{"train_loss": 0.029253114014863968, "train_loss_bc": 0.02923104353249073, "train_loss_llm": 0.002207040786743164, "grad_norm": 0.06914947181940079, "global_step": 29990, "epoch": 833, "lr": 0.001} +{"train_loss": 0.03359144553542137, "train_loss_bc": 0.03359144553542137, "train_loss_llm": 0.0, "grad_norm": 0.08705105632543564, "global_step": 29991, "epoch": 833, "lr": 0.001} +{"train_loss": 0.02457919903099537, "train_loss_bc": 0.02457919903099537, "train_loss_llm": 0.0, "grad_norm": 0.10416656732559204, "global_step": 29992, "epoch": 833, "lr": 0.001} +{"train_loss": 0.03517124429345131, "train_loss_bc": 0.03517124429345131, "train_loss_llm": 0.0, "grad_norm": 0.03402845561504364, "global_step": 29993, "epoch": 833, "lr": 0.001} +{"train_loss": 0.030151769518852234, "train_loss_bc": 0.030144013464450836, "train_loss_llm": 0.0007756650447845459, "grad_norm": 0.057668495923280716, "global_step": 29994, "epoch": 833, "lr": 0.001} +{"train_loss": 0.02844475395977497, "train_loss_bc": 0.02844475395977497, "train_loss_llm": 0.0, "grad_norm": 0.07498882710933685, "global_step": 29995, "epoch": 833, "lr": 0.001} +{"train_loss": 0.03272026777267456, "train_loss_bc": 0.03272026777267456, "train_loss_llm": 0.0, "grad_norm": 0.08952811360359192, "global_step": 29996, "epoch": 833, "lr": 0.001} +{"train_loss": 0.03197377547621727, "train_loss_bc": 0.03197377547621727, "train_loss_llm": 0.0, "grad_norm": 0.03637973219156265, "global_step": 29997, "epoch": 833, "lr": 0.001} +{"train_loss": 0.028724191710352898, "train_loss_bc": 0.028717147186398506, "train_loss_llm": 0.0007044076919555664, "grad_norm": 0.06151127442717552, "global_step": 29998, "epoch": 833, "lr": 0.001} +{"train_loss": 0.02937525324523449, "train_loss_bc": 0.02937525324523449, "train_loss_llm": 0.0, "grad_norm": 0.07320790737867355, "global_step": 29999, "epoch": 833, "lr": 0.001} +{"train_loss": 0.029958639293909073, "train_loss_bc": 0.029958639293909073, "train_loss_llm": 0.0, "grad_norm": 0.08975116163492203, "global_step": 30000, "epoch": 833, "lr": 0.001} +{"train_loss": 0.03371422737836838, "train_loss_bc": 0.03371422737836838, "train_loss_llm": 0.0, "grad_norm": 0.038399554789066315, "global_step": 30001, "epoch": 833, "lr": 0.001} +{"train_loss": 0.030513960868120193, "train_loss_bc": 0.030513960868120193, "train_loss_llm": 0.0, "grad_norm": 0.061459898948669434, "global_step": 30002, "epoch": 833, "lr": 0.001} +{"train_loss": 0.031659845262765884, "train_loss_bc": 0.03165573626756668, "train_loss_llm": 0.00041094422340393066, "grad_norm": 0.0840429961681366, "global_step": 30003, "epoch": 833, "lr": 0.001} +{"train_loss": 0.02668747305870056, "train_loss_bc": 0.02668747305870056, "train_loss_llm": 0.0, "grad_norm": 0.11052526533603668, "global_step": 30004, "epoch": 833, "lr": 0.001} +{"train_loss": 0.02855750173330307, "train_loss_bc": 0.02854551188647747, "train_loss_llm": 0.0011990368366241455, "grad_norm": 0.04769950360059738, "global_step": 30005, "epoch": 833, "lr": 0.001} +{"train_loss": 0.029354527592658997, "train_loss_bc": 0.02934979274868965, "train_loss_llm": 0.00047343969345092773, "grad_norm": 0.06904486566781998, "global_step": 30006, "epoch": 833, "lr": 0.001} +{"train_loss": 0.028876597061753273, "train_loss_bc": 0.028876597061753273, "train_loss_llm": 0.0, "grad_norm": 0.09680850803852081, "global_step": 30007, "epoch": 833, "lr": 0.001} +{"train_loss": 0.029609765857458115, "train_loss_bc": 0.029609765857458115, "train_loss_llm": 0.0, "grad_norm": 0.11088058352470398, "global_step": 30008, "epoch": 833, "lr": 0.001} +{"train_loss": 0.033469829708337784, "train_loss_bc": 0.033469829708337784, "train_loss_llm": 0.0, "grad_norm": 0.03595379367470741, "global_step": 30009, "epoch": 833, "lr": 0.001} +{"train_loss": 0.02799244411289692, "train_loss_bc": 0.027987536042928696, "train_loss_llm": 0.0004908740520477295, "grad_norm": 0.05632634460926056, "global_step": 30010, "epoch": 833, "lr": 0.001} +{"train_loss": 0.029808470979332924, "train_loss_bc": 0.029808470979332924, "train_loss_llm": 0.0, "grad_norm": 0.07118218392133713, "global_step": 30011, "epoch": 833, "lr": 0.001} +{"train_loss": 0.0287642665207386, "train_loss_bc": 0.028757354244589806, "train_loss_llm": 0.0006912350654602051, "grad_norm": 0.08760564029216766, "global_step": 30012, "epoch": 833, "lr": 0.001} +{"train_loss": 0.03342997655272484, "train_loss_bc": 0.03342997655272484, "train_loss_llm": 0.0, "grad_norm": 0.03629626706242561, "global_step": 30013, "epoch": 833, "lr": 0.001} +{"train_loss": 0.03402284160256386, "train_loss_bc": 0.034022700041532516, "train_loss_llm": 1.4007091522216797e-05, "grad_norm": 0.05867946520447731, "global_step": 30014, "epoch": 833, "lr": 0.001} +{"train_loss": 0.03213568404316902, "train_loss_bc": 0.03213568404316902, "train_loss_llm": 0.0, "grad_norm": 0.0830688625574112, "global_step": 30015, "epoch": 833, "lr": 0.001} +{"train_loss": 0.030779225751757622, "train_loss_bc": 0.030779225751757622, "train_loss_llm": 0.0, "grad_norm": 0.10018710047006607, "global_step": 30016, "epoch": 833, "lr": 0.001} +{"train_loss": 0.0286228246986866, "train_loss_bc": 0.0286228246986866, "train_loss_llm": 0.0, "grad_norm": 0.03842925280332565, "global_step": 30017, "epoch": 833, "lr": 0.001} +{"train_loss": 0.031139353290200233, "train_loss_bc": 0.031139353290200233, "train_loss_llm": 0.0, "grad_norm": 0.052170466631650925, "global_step": 30018, "epoch": 833, "lr": 0.001} +{"train_loss": 0.024324938654899597, "train_loss_bc": 0.024324938654899597, "train_loss_llm": 0.0, "grad_norm": 0.06871700286865234, "global_step": 30019, "epoch": 833, "lr": 0.001} +{"train_loss": 0.032818641513586044, "train_loss_bc": 0.032818641513586044, "train_loss_llm": 0.0, "grad_norm": 0.09335958957672119, "global_step": 30020, "epoch": 833, "lr": 0.001} +{"train_loss": 0.02968735620379448, "train_loss_bc": 0.029683159664273262, "train_loss_llm": 0.00041961669921875, "grad_norm": 0.042263444513082504, "global_step": 30021, "epoch": 833, "lr": 0.001} +{"train_loss": 0.02860507369041443, "train_loss_bc": 0.02860507369041443, "train_loss_llm": 0.0, "grad_norm": 0.0603439100086689, "global_step": 30022, "epoch": 833, "lr": 0.001} +{"train_loss": 0.030050334417157702, "train_loss_bc": 0.027738386765122414, "train_loss_llm": 0.0, "grad_norm": 0.08730628341436386, "global_step": 30023, "epoch": 833, "lr": 0.001, "val_loss": 0.0967903882265091} +{"train_loss": 0.024676024913787842, "train_loss_bc": 0.024662110954523087, "train_loss_llm": 0.001391381025314331, "grad_norm": 0.10368378460407257, "global_step": 30024, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03234080597758293, "train_loss_bc": 0.03234080597758293, "train_loss_llm": 0.0, "grad_norm": 0.04031318053603172, "global_step": 30025, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03265535458922386, "train_loss_bc": 0.0326542854309082, "train_loss_llm": 0.00010696053504943848, "grad_norm": 0.06993956863880157, "global_step": 30026, "epoch": 834, "lr": 0.001} +{"train_loss": 0.024028930813074112, "train_loss_bc": 0.024028930813074112, "train_loss_llm": 0.0, "grad_norm": 0.0929887667298317, "global_step": 30027, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03131239861249924, "train_loss_bc": 0.03130657225847244, "train_loss_llm": 0.0005825459957122803, "grad_norm": 0.11870380491018295, "global_step": 30028, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03155112639069557, "train_loss_bc": 0.03155112639069557, "train_loss_llm": 0.0, "grad_norm": 0.0332595519721508, "global_step": 30029, "epoch": 834, "lr": 0.001} +{"train_loss": 0.024489404633641243, "train_loss_bc": 0.024489404633641243, "train_loss_llm": 0.0, "grad_norm": 0.04704833775758743, "global_step": 30030, "epoch": 834, "lr": 0.001} +{"train_loss": 0.02866186760365963, "train_loss_bc": 0.02866186760365963, "train_loss_llm": 0.0, "grad_norm": 0.07687995582818985, "global_step": 30031, "epoch": 834, "lr": 0.001} +{"train_loss": 0.029930341988801956, "train_loss_bc": 0.029930341988801956, "train_loss_llm": 0.0, "grad_norm": 0.09137023240327835, "global_step": 30032, "epoch": 834, "lr": 0.001} +{"train_loss": 0.029077572748064995, "train_loss_bc": 0.029077572748064995, "train_loss_llm": 0.0, "grad_norm": 0.05083221197128296, "global_step": 30033, "epoch": 834, "lr": 0.001} +{"train_loss": 0.028177138417959213, "train_loss_bc": 0.028169069439172745, "train_loss_llm": 0.0008068680763244629, "grad_norm": 0.06125675141811371, "global_step": 30034, "epoch": 834, "lr": 0.001} +{"train_loss": 0.0247032530605793, "train_loss_bc": 0.0247032530605793, "train_loss_llm": 0.0, "grad_norm": 0.08104099333286285, "global_step": 30035, "epoch": 834, "lr": 0.001} +{"train_loss": 0.027445580810308456, "train_loss_bc": 0.027445580810308456, "train_loss_llm": 0.0, "grad_norm": 0.09840735793113708, "global_step": 30036, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03494397550821304, "train_loss_bc": 0.03494397550821304, "train_loss_llm": 0.0, "grad_norm": 0.04303227365016937, "global_step": 30037, "epoch": 834, "lr": 0.001} +{"train_loss": 0.028002310544252396, "train_loss_bc": 0.028002310544252396, "train_loss_llm": 0.0, "grad_norm": 0.06595594435930252, "global_step": 30038, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03402730077505112, "train_loss_bc": 0.03402730077505112, "train_loss_llm": 0.0, "grad_norm": 0.08257897198200226, "global_step": 30039, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03477182611823082, "train_loss_bc": 0.034765467047691345, "train_loss_llm": 0.0006358623504638672, "grad_norm": 0.10190936177968979, "global_step": 30040, "epoch": 834, "lr": 0.001} +{"train_loss": 0.032332390546798706, "train_loss_bc": 0.032332390546798706, "train_loss_llm": 0.0, "grad_norm": 0.050318729132413864, "global_step": 30041, "epoch": 834, "lr": 0.001} +{"train_loss": 0.027350259944796562, "train_loss_bc": 0.027347829192876816, "train_loss_llm": 0.00024309754371643066, "grad_norm": 0.07886062562465668, "global_step": 30042, "epoch": 834, "lr": 0.001} +{"train_loss": 0.031096650287508965, "train_loss_bc": 0.031096650287508965, "train_loss_llm": 0.0, "grad_norm": 0.09354881942272186, "global_step": 30043, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03685412555932999, "train_loss_bc": 0.03685412555932999, "train_loss_llm": 0.0, "grad_norm": 0.13360387086868286, "global_step": 30044, "epoch": 834, "lr": 0.001} +{"train_loss": 0.028211399912834167, "train_loss_bc": 0.028211399912834167, "train_loss_llm": 0.0, "grad_norm": 0.030925793573260307, "global_step": 30045, "epoch": 834, "lr": 0.001} +{"train_loss": 0.026420418173074722, "train_loss_bc": 0.026370085775852203, "train_loss_llm": 0.005033314228057861, "grad_norm": 0.05047699064016342, "global_step": 30046, "epoch": 834, "lr": 0.001} +{"train_loss": 0.027797091752290726, "train_loss_bc": 0.027797091752290726, "train_loss_llm": 0.0, "grad_norm": 0.07315289974212646, "global_step": 30047, "epoch": 834, "lr": 0.001} +{"train_loss": 0.034212708473205566, "train_loss_bc": 0.034212708473205566, "train_loss_llm": 0.0, "grad_norm": 0.09714844077825546, "global_step": 30048, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03203338757157326, "train_loss_bc": 0.03203338757157326, "train_loss_llm": 0.0, "grad_norm": 0.029773002490401268, "global_step": 30049, "epoch": 834, "lr": 0.001} +{"train_loss": 0.030796047300100327, "train_loss_bc": 0.030796047300100327, "train_loss_llm": 0.0, "grad_norm": 0.047449491918087006, "global_step": 30050, "epoch": 834, "lr": 0.001} +{"train_loss": 0.028911007568240166, "train_loss_bc": 0.028911007568240166, "train_loss_llm": 0.0, "grad_norm": 0.07760544866323471, "global_step": 30051, "epoch": 834, "lr": 0.001} +{"train_loss": 0.02771318517625332, "train_loss_bc": 0.02771318517625332, "train_loss_llm": 0.0, "grad_norm": 0.09179811924695969, "global_step": 30052, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03304425626993179, "train_loss_bc": 0.03304425626993179, "train_loss_llm": 0.0, "grad_norm": 0.030099447816610336, "global_step": 30053, "epoch": 834, "lr": 0.001} +{"train_loss": 0.027567770332098007, "train_loss_bc": 0.027567770332098007, "train_loss_llm": 0.0, "grad_norm": 0.05980803444981575, "global_step": 30054, "epoch": 834, "lr": 0.001} +{"train_loss": 0.029692871496081352, "train_loss_bc": 0.029678290709853172, "train_loss_llm": 0.00145798921585083, "grad_norm": 0.08041254431009293, "global_step": 30055, "epoch": 834, "lr": 0.001} +{"train_loss": 0.02831719070672989, "train_loss_bc": 0.02831719070672989, "train_loss_llm": 0.0, "grad_norm": 0.09996040910482407, "global_step": 30056, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03438762202858925, "train_loss_bc": 0.03437682241201401, "train_loss_llm": 0.0010800063610076904, "grad_norm": 0.03016548976302147, "global_step": 30057, "epoch": 834, "lr": 0.001} +{"train_loss": 0.026518508791923523, "train_loss_bc": 0.026518508791923523, "train_loss_llm": 0.0, "grad_norm": 0.048566751182079315, "global_step": 30058, "epoch": 834, "lr": 0.001} +{"train_loss": 0.03017012609375848, "train_loss_bc": 0.04207243397831917, "train_loss_llm": 0.0, "grad_norm": 0.06866667419672012, "global_step": 30059, "epoch": 834, "lr": 0.001, "val_loss": 0.09801483899354935} +{"train_loss": 0.031143831089138985, "train_loss_bc": 0.031143831089138985, "train_loss_llm": 0.0, "grad_norm": 0.08955281972885132, "global_step": 30060, "epoch": 835, "lr": 0.001} +{"train_loss": 0.02744324505329132, "train_loss_bc": 0.02744324505329132, "train_loss_llm": 0.0, "grad_norm": 0.031171675771474838, "global_step": 30061, "epoch": 835, "lr": 0.001} +{"train_loss": 0.031518638134002686, "train_loss_bc": 0.031518638134002686, "train_loss_llm": 0.0, "grad_norm": 0.051351360976696014, "global_step": 30062, "epoch": 835, "lr": 0.001} +{"train_loss": 0.030438072979450226, "train_loss_bc": 0.030433425679802895, "train_loss_llm": 0.000464707612991333, "grad_norm": 0.07632376998662949, "global_step": 30063, "epoch": 835, "lr": 0.001} +{"train_loss": 0.03592851385474205, "train_loss_bc": 0.03591512143611908, "train_loss_llm": 0.0013391673564910889, "grad_norm": 0.10539762675762177, "global_step": 30064, "epoch": 835, "lr": 0.001} +{"train_loss": 0.03224946931004524, "train_loss_bc": 0.03224946931004524, "train_loss_llm": 0.0, "grad_norm": 0.03683988377451897, "global_step": 30065, "epoch": 835, "lr": 0.001} +{"train_loss": 0.022786879912018776, "train_loss_bc": 0.022785890847444534, "train_loss_llm": 9.897351264953613e-05, "grad_norm": 0.05539638176560402, "global_step": 30066, "epoch": 835, "lr": 0.001} +{"train_loss": 0.032088227570056915, "train_loss_bc": 0.032088227570056915, "train_loss_llm": 0.0, "grad_norm": 0.06936044991016388, "global_step": 30067, "epoch": 835, "lr": 0.001} +{"train_loss": 0.030612781643867493, "train_loss_bc": 0.030599497258663177, "train_loss_llm": 0.0013283491134643555, "grad_norm": 0.0936167761683464, "global_step": 30068, "epoch": 835, "lr": 0.001} +{"train_loss": 0.02402782067656517, "train_loss_bc": 0.02402638830244541, "train_loss_llm": 0.00014325976371765137, "grad_norm": 0.027576640248298645, "global_step": 30069, "epoch": 835, "lr": 0.001} +{"train_loss": 0.02530517429113388, "train_loss_bc": 0.02530517429113388, "train_loss_llm": 0.0, "grad_norm": 0.0396830216050148, "global_step": 30070, "epoch": 835, "lr": 0.001} +{"train_loss": 0.037118688225746155, "train_loss_bc": 0.03711775690317154, "train_loss_llm": 9.313225746154785e-05, "grad_norm": 0.06043306365609169, "global_step": 30071, "epoch": 835, "lr": 0.001} +{"train_loss": 0.03180747106671333, "train_loss_bc": 0.03180747106671333, "train_loss_llm": 0.0, "grad_norm": 0.08273939788341522, "global_step": 30072, "epoch": 835, "lr": 0.001} +{"train_loss": 0.034801363945007324, "train_loss_bc": 0.034801363945007324, "train_loss_llm": 0.0, "grad_norm": 0.033023905009031296, "global_step": 30073, "epoch": 835, "lr": 0.001} +{"train_loss": 0.026891225948929787, "train_loss_bc": 0.026890454813838005, "train_loss_llm": 7.706880569458008e-05, "grad_norm": 0.05193896219134331, "global_step": 30074, "epoch": 835, "lr": 0.001} +{"train_loss": 0.03044816665351391, "train_loss_bc": 0.03044816665351391, "train_loss_llm": 0.0, "grad_norm": 0.06393449753522873, "global_step": 30075, "epoch": 835, "lr": 0.001} +{"train_loss": 0.029423577710986137, "train_loss_bc": 0.029410449787974358, "train_loss_llm": 0.0013127028942108154, "grad_norm": 0.08560440689325333, "global_step": 30076, "epoch": 835, "lr": 0.001} +{"train_loss": 0.03180431202054024, "train_loss_bc": 0.03180431202054024, "train_loss_llm": 0.0, "grad_norm": 0.035394806414842606, "global_step": 30077, "epoch": 835, "lr": 0.001} +{"train_loss": 0.03241546452045441, "train_loss_bc": 0.03241546452045441, "train_loss_llm": 0.0, "grad_norm": 0.05738252028822899, "global_step": 30078, "epoch": 835, "lr": 0.001} +{"train_loss": 0.030749570578336716, "train_loss_bc": 0.030749570578336716, "train_loss_llm": 0.0, "grad_norm": 0.07642511278390884, "global_step": 30079, "epoch": 835, "lr": 0.001} +{"train_loss": 0.029044369235634804, "train_loss_bc": 0.029043545946478844, "train_loss_llm": 8.228421211242676e-05, "grad_norm": 0.09925694018602371, "global_step": 30080, "epoch": 835, "lr": 0.001} +{"train_loss": 0.02896779589354992, "train_loss_bc": 0.02896779589354992, "train_loss_llm": 0.0, "grad_norm": 0.040506407618522644, "global_step": 30081, "epoch": 835, "lr": 0.001} +{"train_loss": 0.027256878092885017, "train_loss_bc": 0.027256878092885017, "train_loss_llm": 0.0, "grad_norm": 0.06444138288497925, "global_step": 30082, "epoch": 835, "lr": 0.001} +{"train_loss": 0.03244185075163841, "train_loss_bc": 0.03244185075163841, "train_loss_llm": 0.0, "grad_norm": 0.09227283298969269, "global_step": 30083, "epoch": 835, "lr": 0.001} +{"train_loss": 0.03340287506580353, "train_loss_bc": 0.03340287506580353, "train_loss_llm": 0.0, "grad_norm": 0.11909986287355423, "global_step": 30084, "epoch": 835, "lr": 0.001} +{"train_loss": 0.029055749997496605, "train_loss_bc": 0.029055749997496605, "train_loss_llm": 0.0, "grad_norm": 0.04061591997742653, "global_step": 30085, "epoch": 835, "lr": 0.001} +{"train_loss": 0.027491683140397072, "train_loss_bc": 0.0274598840624094, "train_loss_llm": 0.0031798481941223145, "grad_norm": 0.06833993643522263, "global_step": 30086, "epoch": 835, "lr": 0.001} +{"train_loss": 0.02861805632710457, "train_loss_bc": 0.02859930694103241, "train_loss_llm": 0.0018750131130218506, "grad_norm": 0.08790833503007889, "global_step": 30087, "epoch": 835, "lr": 0.001} +{"train_loss": 0.03332117199897766, "train_loss_bc": 0.03332117199897766, "train_loss_llm": 0.0, "grad_norm": 0.1190488263964653, "global_step": 30088, "epoch": 835, "lr": 0.001} +{"train_loss": 0.02944468893110752, "train_loss_bc": 0.02944468893110752, "train_loss_llm": 0.0, "grad_norm": 0.034008562564849854, "global_step": 30089, "epoch": 835, "lr": 0.001} +{"train_loss": 0.027851713821291924, "train_loss_bc": 0.027851713821291924, "train_loss_llm": 0.0, "grad_norm": 0.044958919286727905, "global_step": 30090, "epoch": 835, "lr": 0.001} +{"train_loss": 0.030863041058182716, "train_loss_bc": 0.030863041058182716, "train_loss_llm": 0.0, "grad_norm": 0.06668946146965027, "global_step": 30091, "epoch": 835, "lr": 0.001} +{"train_loss": 0.0285488311201334, "train_loss_bc": 0.0285488311201334, "train_loss_llm": 0.0, "grad_norm": 0.07761219143867493, "global_step": 30092, "epoch": 835, "lr": 0.001} +{"train_loss": 0.02489517442882061, "train_loss_bc": 0.02486998401582241, "train_loss_llm": 0.002518951892852783, "grad_norm": 0.034270912408828735, "global_step": 30093, "epoch": 835, "lr": 0.001} +{"train_loss": 0.03128063678741455, "train_loss_bc": 0.03128063678741455, "train_loss_llm": 0.0, "grad_norm": 0.05758797749876976, "global_step": 30094, "epoch": 835, "lr": 0.001} +{"train_loss": 0.030017966145856514, "train_loss_bc": 0.029159769415855408, "train_loss_llm": 0.0, "grad_norm": 0.06761881709098816, "global_step": 30095, "epoch": 835, "lr": 0.001, "train/cumulative_reward": 107.19482127626956, "train/mean_score": 11.645014412116996, "train/success_rate": 0.1, "test/cumulative_reward": 35.237879551401534, "test/mean_score": 9.653499387794477, "test/success_rate": 0.24, "val_loss": 0.09848333150148392, "train_action_mse_error": 0.032341115176677704} +{"train_loss": 0.03439556807279587, "train_loss_bc": 0.034390006214380264, "train_loss_llm": 0.0005561709403991699, "grad_norm": 0.07780206203460693, "global_step": 30096, "epoch": 836, "lr": 0.001} +{"train_loss": 0.0384070947766304, "train_loss_bc": 0.0384070947766304, "train_loss_llm": 0.0, "grad_norm": 0.051933322101831436, "global_step": 30097, "epoch": 836, "lr": 0.001} +{"train_loss": 0.027760852128267288, "train_loss_bc": 0.027760852128267288, "train_loss_llm": 0.0, "grad_norm": 0.06665337085723877, "global_step": 30098, "epoch": 836, "lr": 0.001} +{"train_loss": 0.030909543856978416, "train_loss_bc": 0.03090357966721058, "train_loss_llm": 0.0005964040756225586, "grad_norm": 0.07819900661706924, "global_step": 30099, "epoch": 836, "lr": 0.001} +{"train_loss": 0.029361369088292122, "train_loss_bc": 0.029343649744987488, "train_loss_llm": 0.0017719268798828125, "grad_norm": 0.09530177712440491, "global_step": 30100, "epoch": 836, "lr": 0.001} +{"train_loss": 0.03213426098227501, "train_loss_bc": 0.03213426098227501, "train_loss_llm": 0.0, "grad_norm": 0.048136766999959946, "global_step": 30101, "epoch": 836, "lr": 0.001} +{"train_loss": 0.029807642102241516, "train_loss_bc": 0.029807642102241516, "train_loss_llm": 0.0, "grad_norm": 0.07045233994722366, "global_step": 30102, "epoch": 836, "lr": 0.001} +{"train_loss": 0.02715475670993328, "train_loss_bc": 0.02715475670993328, "train_loss_llm": 0.0, "grad_norm": 0.08250579982995987, "global_step": 30103, "epoch": 836, "lr": 0.001} +{"train_loss": 0.030972329899668694, "train_loss_bc": 0.030971111729741096, "train_loss_llm": 0.00012180209159851074, "grad_norm": 0.1134677305817604, "global_step": 30104, "epoch": 836, "lr": 0.001} +{"train_loss": 0.027533307671546936, "train_loss_bc": 0.027533307671546936, "train_loss_llm": 0.0, "grad_norm": 0.04075240343809128, "global_step": 30105, "epoch": 836, "lr": 0.001} +{"train_loss": 0.02395818568766117, "train_loss_bc": 0.02395818568766117, "train_loss_llm": 0.0, "grad_norm": 0.07525080442428589, "global_step": 30106, "epoch": 836, "lr": 0.001} +{"train_loss": 0.030291171744465828, "train_loss_bc": 0.030285455286502838, "train_loss_llm": 0.0005716383457183838, "grad_norm": 0.10135770589113235, "global_step": 30107, "epoch": 836, "lr": 0.001} +{"train_loss": 0.02832855097949505, "train_loss_bc": 0.02832855097949505, "train_loss_llm": 0.0, "grad_norm": 0.11573836952447891, "global_step": 30108, "epoch": 836, "lr": 0.001} +{"train_loss": 0.03262903168797493, "train_loss_bc": 0.03262197971343994, "train_loss_llm": 0.0007052421569824219, "grad_norm": 0.03311026468873024, "global_step": 30109, "epoch": 836, "lr": 0.001} +{"train_loss": 0.04125199466943741, "train_loss_bc": 0.04125199466943741, "train_loss_llm": 0.0, "grad_norm": 0.0626058578491211, "global_step": 30110, "epoch": 836, "lr": 0.001} +{"train_loss": 0.024761749431490898, "train_loss_bc": 0.02474888227880001, "train_loss_llm": 0.001286625862121582, "grad_norm": 0.07872320711612701, "global_step": 30111, "epoch": 836, "lr": 0.001} +{"train_loss": 0.027661116793751717, "train_loss_bc": 0.027658678591251373, "train_loss_llm": 0.00024375319480895996, "grad_norm": 0.0980101227760315, "global_step": 30112, "epoch": 836, "lr": 0.001} +{"train_loss": 0.02509397082030773, "train_loss_bc": 0.02509397082030773, "train_loss_llm": 0.0, "grad_norm": 0.03099379874765873, "global_step": 30113, "epoch": 836, "lr": 0.001} +{"train_loss": 0.02798818238079548, "train_loss_bc": 0.02798818238079548, "train_loss_llm": 0.0, "grad_norm": 0.0558452270925045, "global_step": 30114, "epoch": 836, "lr": 0.001} +{"train_loss": 0.0297798253595829, "train_loss_bc": 0.029760871082544327, "train_loss_llm": 0.001895517110824585, "grad_norm": 0.07948069274425507, "global_step": 30115, "epoch": 836, "lr": 0.001} +{"train_loss": 0.027834180742502213, "train_loss_bc": 0.027834180742502213, "train_loss_llm": 0.0, "grad_norm": 0.09267466515302658, "global_step": 30116, "epoch": 836, "lr": 0.001} +{"train_loss": 0.030490387231111526, "train_loss_bc": 0.030490387231111526, "train_loss_llm": 0.0, "grad_norm": 0.03556172177195549, "global_step": 30117, "epoch": 836, "lr": 0.001} +{"train_loss": 0.03168007731437683, "train_loss_bc": 0.03168007731437683, "train_loss_llm": 0.0, "grad_norm": 0.06367047131061554, "global_step": 30118, "epoch": 836, "lr": 0.001} +{"train_loss": 0.031986139714717865, "train_loss_bc": 0.031974487006664276, "train_loss_llm": 0.0011654198169708252, "grad_norm": 0.08968708664178848, "global_step": 30119, "epoch": 836, "lr": 0.001} +{"train_loss": 0.033052317798137665, "train_loss_bc": 0.033052317798137665, "train_loss_llm": 0.0, "grad_norm": 0.09463942795991898, "global_step": 30120, "epoch": 836, "lr": 0.001} +{"train_loss": 0.02690298669040203, "train_loss_bc": 0.02690298669040203, "train_loss_llm": 0.0, "grad_norm": 0.044522155076265335, "global_step": 30121, "epoch": 836, "lr": 0.001} +{"train_loss": 0.02668565697968006, "train_loss_bc": 0.02668565697968006, "train_loss_llm": 0.0, "grad_norm": 0.0584629625082016, "global_step": 30122, "epoch": 836, "lr": 0.001} +{"train_loss": 0.026061946526169777, "train_loss_bc": 0.026052944362163544, "train_loss_llm": 0.0009002983570098877, "grad_norm": 0.0775662511587143, "global_step": 30123, "epoch": 836, "lr": 0.001} +{"train_loss": 0.02852332778275013, "train_loss_bc": 0.02852332778275013, "train_loss_llm": 0.0, "grad_norm": 0.10725263506174088, "global_step": 30124, "epoch": 836, "lr": 0.001} +{"train_loss": 0.02992224507033825, "train_loss_bc": 0.02992224507033825, "train_loss_llm": 0.0, "grad_norm": 0.03829986974596977, "global_step": 30125, "epoch": 836, "lr": 0.001} +{"train_loss": 0.031956396996974945, "train_loss_bc": 0.031956396996974945, "train_loss_llm": 0.0, "grad_norm": 0.05952715501189232, "global_step": 30126, "epoch": 836, "lr": 0.001} +{"train_loss": 0.028584977611899376, "train_loss_bc": 0.028584977611899376, "train_loss_llm": 0.0, "grad_norm": 0.07928391546010971, "global_step": 30127, "epoch": 836, "lr": 0.001} +{"train_loss": 0.03166021406650543, "train_loss_bc": 0.031650111079216, "train_loss_llm": 0.0010103285312652588, "grad_norm": 0.10084246844053268, "global_step": 30128, "epoch": 836, "lr": 0.001} +{"train_loss": 0.03227829188108444, "train_loss_bc": 0.03227829188108444, "train_loss_llm": 0.0, "grad_norm": 0.03610100969672203, "global_step": 30129, "epoch": 836, "lr": 0.001} +{"train_loss": 0.032435089349746704, "train_loss_bc": 0.032435089349746704, "train_loss_llm": 0.0, "grad_norm": 0.059152111411094666, "global_step": 30130, "epoch": 836, "lr": 0.001} +{"train_loss": 0.03001967046616806, "train_loss_bc": 0.030461255460977554, "train_loss_llm": 0.0012140274047851562, "grad_norm": 0.09880837798118591, "global_step": 30131, "epoch": 836, "lr": 0.001, "val_loss": 0.10260277986526489} +{"train_loss": 0.028630582615733147, "train_loss_bc": 0.028622489422559738, "train_loss_llm": 0.0008092522621154785, "grad_norm": 0.12182065844535828, "global_step": 30132, "epoch": 837, "lr": 0.001} +{"train_loss": 0.032139942049980164, "train_loss_bc": 0.032139942049980164, "train_loss_llm": 0.0, "grad_norm": 0.04001632332801819, "global_step": 30133, "epoch": 837, "lr": 0.001} +{"train_loss": 0.0307319276034832, "train_loss_bc": 0.030697476118803024, "train_loss_llm": 0.0034452080726623535, "grad_norm": 0.06316898763179779, "global_step": 30134, "epoch": 837, "lr": 0.001} +{"train_loss": 0.026139361783862114, "train_loss_bc": 0.026113437488675117, "train_loss_llm": 0.0025924742221832275, "grad_norm": 0.09246966987848282, "global_step": 30135, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03094801865518093, "train_loss_bc": 0.03094801865518093, "train_loss_llm": 0.0, "grad_norm": 0.1273089349269867, "global_step": 30136, "epoch": 837, "lr": 0.001} +{"train_loss": 0.035048454999923706, "train_loss_bc": 0.035048454999923706, "train_loss_llm": 0.0, "grad_norm": 0.03940288722515106, "global_step": 30137, "epoch": 837, "lr": 0.001} +{"train_loss": 0.031047360971570015, "train_loss_bc": 0.031047360971570015, "train_loss_llm": 0.0, "grad_norm": 0.06111389398574829, "global_step": 30138, "epoch": 837, "lr": 0.001} +{"train_loss": 0.030727719888091087, "train_loss_bc": 0.030727719888091087, "train_loss_llm": 0.0, "grad_norm": 0.0854884535074234, "global_step": 30139, "epoch": 837, "lr": 0.001} +{"train_loss": 0.02687695249915123, "train_loss_bc": 0.02687695249915123, "train_loss_llm": 0.0, "grad_norm": 0.11009005457162857, "global_step": 30140, "epoch": 837, "lr": 0.001} +{"train_loss": 0.02812330052256584, "train_loss_bc": 0.028116783127188683, "train_loss_llm": 0.0006516575813293457, "grad_norm": 0.03740015625953674, "global_step": 30141, "epoch": 837, "lr": 0.001} +{"train_loss": 0.030953381210565567, "train_loss_bc": 0.03095139190554619, "train_loss_llm": 0.0001989305019378662, "grad_norm": 0.051170118153095245, "global_step": 30142, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03176865726709366, "train_loss_bc": 0.03176865726709366, "train_loss_llm": 0.0, "grad_norm": 0.062202926725149155, "global_step": 30143, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03144485875964165, "train_loss_bc": 0.03144485875964165, "train_loss_llm": 0.0, "grad_norm": 0.07888615131378174, "global_step": 30144, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03117816522717476, "train_loss_bc": 0.03117816522717476, "train_loss_llm": 0.0, "grad_norm": 0.034189336001873016, "global_step": 30145, "epoch": 837, "lr": 0.001} +{"train_loss": 0.029165498912334442, "train_loss_bc": 0.02915588952600956, "train_loss_llm": 0.0009608864784240723, "grad_norm": 0.05928805470466614, "global_step": 30146, "epoch": 837, "lr": 0.001} +{"train_loss": 0.029620544984936714, "train_loss_bc": 0.029620544984936714, "train_loss_llm": 0.0, "grad_norm": 0.07996536791324615, "global_step": 30147, "epoch": 837, "lr": 0.001} +{"train_loss": 0.02631782367825508, "train_loss_bc": 0.02631782367825508, "train_loss_llm": 0.0, "grad_norm": 0.10629267245531082, "global_step": 30148, "epoch": 837, "lr": 0.001} +{"train_loss": 0.031293004751205444, "train_loss_bc": 0.031293004751205444, "train_loss_llm": 0.0, "grad_norm": 0.033876799046993256, "global_step": 30149, "epoch": 837, "lr": 0.001} +{"train_loss": 0.024487890303134918, "train_loss_bc": 0.024487890303134918, "train_loss_llm": 0.0, "grad_norm": 0.05642775073647499, "global_step": 30150, "epoch": 837, "lr": 0.001} +{"train_loss": 0.02718673087656498, "train_loss_bc": 0.02718673087656498, "train_loss_llm": 0.0, "grad_norm": 0.07542164623737335, "global_step": 30151, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03538232296705246, "train_loss_bc": 0.03538232296705246, "train_loss_llm": 0.0, "grad_norm": 0.09926152974367142, "global_step": 30152, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03375044837594032, "train_loss_bc": 0.03375044837594032, "train_loss_llm": 0.0, "grad_norm": 0.045345913618803024, "global_step": 30153, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03520893305540085, "train_loss_bc": 0.03519890457391739, "train_loss_llm": 0.0010027885437011719, "grad_norm": 0.08671800792217255, "global_step": 30154, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03034108690917492, "train_loss_bc": 0.03034108690917492, "train_loss_llm": 0.0, "grad_norm": 0.10854534059762955, "global_step": 30155, "epoch": 837, "lr": 0.001} +{"train_loss": 0.031826622784137726, "train_loss_bc": 0.031813960522413254, "train_loss_llm": 0.0012663602828979492, "grad_norm": 0.1441626250743866, "global_step": 30156, "epoch": 837, "lr": 0.001} +{"train_loss": 0.032712507992982864, "train_loss_bc": 0.032712507992982864, "train_loss_llm": 0.0, "grad_norm": 0.0409381277859211, "global_step": 30157, "epoch": 837, "lr": 0.001} +{"train_loss": 0.025551848113536835, "train_loss_bc": 0.025551848113536835, "train_loss_llm": 0.0, "grad_norm": 0.0661524310708046, "global_step": 30158, "epoch": 837, "lr": 0.001} +{"train_loss": 0.025390442460775375, "train_loss_bc": 0.025390442460775375, "train_loss_llm": 0.0, "grad_norm": 0.08420107513666153, "global_step": 30159, "epoch": 837, "lr": 0.001} +{"train_loss": 0.029891714453697205, "train_loss_bc": 0.029889218509197235, "train_loss_llm": 0.00024953484535217285, "grad_norm": 0.10938998311758041, "global_step": 30160, "epoch": 837, "lr": 0.001} +{"train_loss": 0.030738575384020805, "train_loss_bc": 0.030726192519068718, "train_loss_llm": 0.0012382268905639648, "grad_norm": 0.038393620401620865, "global_step": 30161, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03621722757816315, "train_loss_bc": 0.03621722757816315, "train_loss_llm": 0.0, "grad_norm": 0.06797990947961807, "global_step": 30162, "epoch": 837, "lr": 0.001} +{"train_loss": 0.02697514370083809, "train_loss_bc": 0.026957696303725243, "train_loss_llm": 0.001744776964187622, "grad_norm": 0.08063356578350067, "global_step": 30163, "epoch": 837, "lr": 0.001} +{"train_loss": 0.02858235500752926, "train_loss_bc": 0.02858235500752926, "train_loss_llm": 0.0, "grad_norm": 0.10059385001659393, "global_step": 30164, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03392575681209564, "train_loss_bc": 0.03392575681209564, "train_loss_llm": 0.0, "grad_norm": 0.04622567072510719, "global_step": 30165, "epoch": 837, "lr": 0.001} +{"train_loss": 0.03009047731757164, "train_loss_bc": 0.03009047731757164, "train_loss_llm": 0.0, "grad_norm": 0.06979676336050034, "global_step": 30166, "epoch": 837, "lr": 0.001} +{"train_loss": 0.0303432529585229, "train_loss_bc": 0.03194146603345871, "train_loss_llm": 0.0, "grad_norm": 0.0918903574347496, "global_step": 30167, "epoch": 837, "lr": 0.001, "val_loss": 0.09946122765541077} +{"train_loss": 0.027117297053337097, "train_loss_bc": 0.027117297053337097, "train_loss_llm": 0.0, "grad_norm": 0.1230648085474968, "global_step": 30168, "epoch": 838, "lr": 0.001} +{"train_loss": 0.027002258226275444, "train_loss_bc": 0.027002258226275444, "train_loss_llm": 0.0, "grad_norm": 0.036030612885951996, "global_step": 30169, "epoch": 838, "lr": 0.001} +{"train_loss": 0.031077025458216667, "train_loss_bc": 0.031076805666089058, "train_loss_llm": 2.2023916244506836e-05, "grad_norm": 0.0573919303715229, "global_step": 30170, "epoch": 838, "lr": 0.001} +{"train_loss": 0.029927527531981468, "train_loss_bc": 0.029918469488620758, "train_loss_llm": 0.0009057223796844482, "grad_norm": 0.07626699656248093, "global_step": 30171, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03425663709640503, "train_loss_bc": 0.03425663709640503, "train_loss_llm": 0.0, "grad_norm": 0.10503538697957993, "global_step": 30172, "epoch": 838, "lr": 0.001} +{"train_loss": 0.026446126401424408, "train_loss_bc": 0.026441600173711777, "train_loss_llm": 0.0004526078701019287, "grad_norm": 0.04519888758659363, "global_step": 30173, "epoch": 838, "lr": 0.001} +{"train_loss": 0.0271036047488451, "train_loss_bc": 0.0271036047488451, "train_loss_llm": 0.0, "grad_norm": 0.07589764893054962, "global_step": 30174, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03162446990609169, "train_loss_bc": 0.03162446990609169, "train_loss_llm": 0.0, "grad_norm": 0.09550099074840546, "global_step": 30175, "epoch": 838, "lr": 0.001} +{"train_loss": 0.031886015087366104, "train_loss_bc": 0.031886015087366104, "train_loss_llm": 0.0, "grad_norm": 0.11049128323793411, "global_step": 30176, "epoch": 838, "lr": 0.001} +{"train_loss": 0.029424283653497696, "train_loss_bc": 0.029424283653497696, "train_loss_llm": 0.0, "grad_norm": 0.03367174044251442, "global_step": 30177, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03412135690450668, "train_loss_bc": 0.03412135690450668, "train_loss_llm": 0.0, "grad_norm": 0.05001423880457878, "global_step": 30178, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03453605994582176, "train_loss_bc": 0.03453605994582176, "train_loss_llm": 0.0, "grad_norm": 0.06962477415800095, "global_step": 30179, "epoch": 838, "lr": 0.001} +{"train_loss": 0.028440967202186584, "train_loss_bc": 0.028440967202186584, "train_loss_llm": 0.0, "grad_norm": 0.08782617002725601, "global_step": 30180, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03648464381694794, "train_loss_bc": 0.03648464381694794, "train_loss_llm": 0.0, "grad_norm": 0.04961339756846428, "global_step": 30181, "epoch": 838, "lr": 0.001} +{"train_loss": 0.02860092557966709, "train_loss_bc": 0.02860092557966709, "train_loss_llm": 0.0, "grad_norm": 0.07231173664331436, "global_step": 30182, "epoch": 838, "lr": 0.001} +{"train_loss": 0.02961568348109722, "train_loss_bc": 0.02961568348109722, "train_loss_llm": 0.0, "grad_norm": 0.09068714082241058, "global_step": 30183, "epoch": 838, "lr": 0.001} +{"train_loss": 0.02945457585155964, "train_loss_bc": 0.02945457585155964, "train_loss_llm": 0.0, "grad_norm": 0.11303821951150894, "global_step": 30184, "epoch": 838, "lr": 0.001} +{"train_loss": 0.024665817618370056, "train_loss_bc": 0.024665817618370056, "train_loss_llm": 0.0, "grad_norm": 0.030541257932782173, "global_step": 30185, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03022105060517788, "train_loss_bc": 0.03022105060517788, "train_loss_llm": 0.0, "grad_norm": 0.047102756798267365, "global_step": 30186, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03084494173526764, "train_loss_bc": 0.030841104686260223, "train_loss_llm": 0.00038376450538635254, "grad_norm": 0.07064329087734222, "global_step": 30187, "epoch": 838, "lr": 0.001} +{"train_loss": 0.030915891751646996, "train_loss_bc": 0.030915891751646996, "train_loss_llm": 0.0, "grad_norm": 0.09392791986465454, "global_step": 30188, "epoch": 838, "lr": 0.001} +{"train_loss": 0.029987554997205734, "train_loss_bc": 0.02993236482143402, "train_loss_llm": 0.005519092082977295, "grad_norm": 0.032093197107315063, "global_step": 30189, "epoch": 838, "lr": 0.001} +{"train_loss": 0.02347262017428875, "train_loss_bc": 0.02347262017428875, "train_loss_llm": 0.0, "grad_norm": 0.04924445599317551, "global_step": 30190, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03401395305991173, "train_loss_bc": 0.03399905562400818, "train_loss_llm": 0.0014898180961608887, "grad_norm": 0.07837370038032532, "global_step": 30191, "epoch": 838, "lr": 0.001} +{"train_loss": 0.027164017781615257, "train_loss_bc": 0.027164017781615257, "train_loss_llm": 0.0, "grad_norm": 0.1015501394867897, "global_step": 30192, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03254294395446777, "train_loss_bc": 0.03254294395446777, "train_loss_llm": 0.0, "grad_norm": 0.036765262484550476, "global_step": 30193, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03500241041183472, "train_loss_bc": 0.03500241041183472, "train_loss_llm": 0.0, "grad_norm": 0.05194615200161934, "global_step": 30194, "epoch": 838, "lr": 0.001} +{"train_loss": 0.028931980952620506, "train_loss_bc": 0.02891852892935276, "train_loss_llm": 0.0013451874256134033, "grad_norm": 0.06825193017721176, "global_step": 30195, "epoch": 838, "lr": 0.001} +{"train_loss": 0.026416148990392685, "train_loss_bc": 0.026416148990392685, "train_loss_llm": 0.0, "grad_norm": 0.08757568895816803, "global_step": 30196, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03226812556385994, "train_loss_bc": 0.03226812556385994, "train_loss_llm": 0.0, "grad_norm": 0.03390419855713844, "global_step": 30197, "epoch": 838, "lr": 0.001} +{"train_loss": 0.02800283581018448, "train_loss_bc": 0.02800283581018448, "train_loss_llm": 0.0, "grad_norm": 0.05500026047229767, "global_step": 30198, "epoch": 838, "lr": 0.001} +{"train_loss": 0.0349692665040493, "train_loss_bc": 0.034956321120262146, "train_loss_llm": 0.0012945830821990967, "grad_norm": 0.08441096544265747, "global_step": 30199, "epoch": 838, "lr": 0.001} +{"train_loss": 0.027830664068460464, "train_loss_bc": 0.027830664068460464, "train_loss_llm": 0.0, "grad_norm": 0.09721250087022781, "global_step": 30200, "epoch": 838, "lr": 0.001} +{"train_loss": 0.024638524278998375, "train_loss_bc": 0.024638524278998375, "train_loss_llm": 0.0, "grad_norm": 0.03798559308052063, "global_step": 30201, "epoch": 838, "lr": 0.001} +{"train_loss": 0.029896462336182594, "train_loss_bc": 0.02989450842142105, "train_loss_llm": 0.00019532442092895508, "grad_norm": 0.0591110922396183, "global_step": 30202, "epoch": 838, "lr": 0.001} +{"train_loss": 0.03010450883044137, "train_loss_bc": 0.034857649356126785, "train_loss_llm": 0.0, "grad_norm": 0.08650745451450348, "global_step": 30203, "epoch": 838, "lr": 0.001, "val_loss": 0.09701566398143768} +{"train_loss": 0.030152706429362297, "train_loss_bc": 0.03013806790113449, "train_loss_llm": 0.001463770866394043, "grad_norm": 0.10303860902786255, "global_step": 30204, "epoch": 839, "lr": 0.001} +{"train_loss": 0.030971642583608627, "train_loss_bc": 0.030969299376010895, "train_loss_llm": 0.00023430585861206055, "grad_norm": 0.049203723669052124, "global_step": 30205, "epoch": 839, "lr": 0.001} +{"train_loss": 0.02655983902513981, "train_loss_bc": 0.02655983902513981, "train_loss_llm": 0.0, "grad_norm": 0.06479719281196594, "global_step": 30206, "epoch": 839, "lr": 0.001} +{"train_loss": 0.02718658559024334, "train_loss_bc": 0.02718658559024334, "train_loss_llm": 0.0, "grad_norm": 0.08605718612670898, "global_step": 30207, "epoch": 839, "lr": 0.001} +{"train_loss": 0.03556843847036362, "train_loss_bc": 0.03556843847036362, "train_loss_llm": 0.0, "grad_norm": 0.11881981045007706, "global_step": 30208, "epoch": 839, "lr": 0.001} +{"train_loss": 0.02868727594614029, "train_loss_bc": 0.02868727594614029, "train_loss_llm": 0.0, "grad_norm": 0.040768466889858246, "global_step": 30209, "epoch": 839, "lr": 0.001} +{"train_loss": 0.020811308175325394, "train_loss_bc": 0.020804008468985558, "train_loss_llm": 0.000729978084564209, "grad_norm": 0.0561630018055439, "global_step": 30210, "epoch": 839, "lr": 0.001} +{"train_loss": 0.02907983586192131, "train_loss_bc": 0.02907983586192131, "train_loss_llm": 0.0, "grad_norm": 0.08488569408655167, "global_step": 30211, "epoch": 839, "lr": 0.001} +{"train_loss": 0.038231682032346725, "train_loss_bc": 0.03820732608437538, "train_loss_llm": 0.0024356842041015625, "grad_norm": 0.10537640750408173, "global_step": 30212, "epoch": 839, "lr": 0.001} +{"train_loss": 0.03141902759671211, "train_loss_bc": 0.03141902759671211, "train_loss_llm": 0.0, "grad_norm": 0.03540300577878952, "global_step": 30213, "epoch": 839, "lr": 0.001} +{"train_loss": 0.022947488352656364, "train_loss_bc": 0.02292710728943348, "train_loss_llm": 0.0020381510257720947, "grad_norm": 0.04896702244877815, "global_step": 30214, "epoch": 839, "lr": 0.001} +{"train_loss": 0.029616637155413628, "train_loss_bc": 0.029616637155413628, "train_loss_llm": 0.0, "grad_norm": 0.07118159532546997, "global_step": 30215, "epoch": 839, "lr": 0.001} +{"train_loss": 0.026261985301971436, "train_loss_bc": 0.026261985301971436, "train_loss_llm": 0.0, "grad_norm": 0.0898459181189537, "global_step": 30216, "epoch": 839, "lr": 0.001} +{"train_loss": 0.028387796133756638, "train_loss_bc": 0.028387796133756638, "train_loss_llm": 0.0, "grad_norm": 0.029145102947950363, "global_step": 30217, "epoch": 839, "lr": 0.001} +{"train_loss": 0.03240082412958145, "train_loss_bc": 0.03239717334508896, "train_loss_llm": 0.0003650784492492676, "grad_norm": 0.05862932279706001, "global_step": 30218, "epoch": 839, "lr": 0.001} +{"train_loss": 0.031929198652505875, "train_loss_bc": 0.031929198652505875, "train_loss_llm": 0.0, "grad_norm": 0.0846923366189003, "global_step": 30219, "epoch": 839, "lr": 0.001} +{"train_loss": 0.024079443886876106, "train_loss_bc": 0.024068212136626244, "train_loss_llm": 0.001123189926147461, "grad_norm": 0.09410437196493149, "global_step": 30220, "epoch": 839, "lr": 0.001} +{"train_loss": 0.03021925315260887, "train_loss_bc": 0.03021925315260887, "train_loss_llm": 0.0, "grad_norm": 0.04141873121261597, "global_step": 30221, "epoch": 839, "lr": 0.001} +{"train_loss": 0.029454359784722328, "train_loss_bc": 0.029454359784722328, "train_loss_llm": 0.0, "grad_norm": 0.060606930404901505, "global_step": 30222, "epoch": 839, "lr": 0.001} +{"train_loss": 0.028729543089866638, "train_loss_bc": 0.028729543089866638, "train_loss_llm": 0.0, "grad_norm": 0.08814583718776703, "global_step": 30223, "epoch": 839, "lr": 0.001} +{"train_loss": 0.028571324422955513, "train_loss_bc": 0.028571324422955513, "train_loss_llm": 0.0, "grad_norm": 0.10042226314544678, "global_step": 30224, "epoch": 839, "lr": 0.001} +{"train_loss": 0.03192412480711937, "train_loss_bc": 0.03192165121436119, "train_loss_llm": 0.0002472996711730957, "grad_norm": 0.03254430741071701, "global_step": 30225, "epoch": 839, "lr": 0.001} +{"train_loss": 0.029138555750250816, "train_loss_bc": 0.029138555750250816, "train_loss_llm": 0.0, "grad_norm": 0.055610209703445435, "global_step": 30226, "epoch": 839, "lr": 0.001} +{"train_loss": 0.031020792201161385, "train_loss_bc": 0.03100135549902916, "train_loss_llm": 0.0019435882568359375, "grad_norm": 0.08988664299249649, "global_step": 30227, "epoch": 839, "lr": 0.001} +{"train_loss": 0.027058281004428864, "train_loss_bc": 0.027058281004428864, "train_loss_llm": 0.0, "grad_norm": 0.10263747721910477, "global_step": 30228, "epoch": 839, "lr": 0.001} +{"train_loss": 0.031067825853824615, "train_loss_bc": 0.03106067329645157, "train_loss_llm": 0.0007151663303375244, "grad_norm": 0.03438013792037964, "global_step": 30229, "epoch": 839, "lr": 0.001} +{"train_loss": 0.028475547209382057, "train_loss_bc": 0.028475547209382057, "train_loss_llm": 0.0, "grad_norm": 0.052433643490076065, "global_step": 30230, "epoch": 839, "lr": 0.001} +{"train_loss": 0.030808091163635254, "train_loss_bc": 0.030808091163635254, "train_loss_llm": 0.0, "grad_norm": 0.0696367546916008, "global_step": 30231, "epoch": 839, "lr": 0.001} +{"train_loss": 0.033513277769088745, "train_loss_bc": 0.033513277769088745, "train_loss_llm": 0.0, "grad_norm": 0.088144451379776, "global_step": 30232, "epoch": 839, "lr": 0.001} +{"train_loss": 0.034620463848114014, "train_loss_bc": 0.034620463848114014, "train_loss_llm": 0.0, "grad_norm": 0.03759381175041199, "global_step": 30233, "epoch": 839, "lr": 0.001} +{"train_loss": 0.028258293867111206, "train_loss_bc": 0.028258293867111206, "train_loss_llm": 0.0, "grad_norm": 0.04611429572105408, "global_step": 30234, "epoch": 839, "lr": 0.001} +{"train_loss": 0.03457800671458244, "train_loss_bc": 0.03457800671458244, "train_loss_llm": 0.0, "grad_norm": 0.06791319698095322, "global_step": 30235, "epoch": 839, "lr": 0.001} +{"train_loss": 0.03477143868803978, "train_loss_bc": 0.034760698676109314, "train_loss_llm": 0.0010741651058197021, "grad_norm": 0.080656036734581, "global_step": 30236, "epoch": 839, "lr": 0.001} +{"train_loss": 0.031021732836961746, "train_loss_bc": 0.031021732836961746, "train_loss_llm": 0.0, "grad_norm": 0.04174516350030899, "global_step": 30237, "epoch": 839, "lr": 0.001} +{"train_loss": 0.02954338677227497, "train_loss_bc": 0.02954338677227497, "train_loss_llm": 0.0, "grad_norm": 0.06807958334684372, "global_step": 30238, "epoch": 839, "lr": 0.001} +{"train_loss": 0.029903499668257102, "train_loss_bc": 0.02945997379720211, "train_loss_llm": 0.0, "grad_norm": 0.09875035285949707, "global_step": 30239, "epoch": 839, "lr": 0.001, "val_loss": 0.0992739349603653} +{"train_loss": 0.02863135002553463, "train_loss_bc": 0.02863135002553463, "train_loss_llm": 0.0, "grad_norm": 0.12224200367927551, "global_step": 30240, "epoch": 840, "lr": 0.001} +{"train_loss": 0.027895832434296608, "train_loss_bc": 0.027895832434296608, "train_loss_llm": 0.0, "grad_norm": 0.038031384348869324, "global_step": 30241, "epoch": 840, "lr": 0.001} +{"train_loss": 0.02997075952589512, "train_loss_bc": 0.02997075952589512, "train_loss_llm": 0.0, "grad_norm": 0.05721825733780861, "global_step": 30242, "epoch": 840, "lr": 0.001} +{"train_loss": 0.03174614533782005, "train_loss_bc": 0.03174614533782005, "train_loss_llm": 0.0, "grad_norm": 0.07348527014255524, "global_step": 30243, "epoch": 840, "lr": 0.001} +{"train_loss": 0.03303034231066704, "train_loss_bc": 0.03303034231066704, "train_loss_llm": 0.0, "grad_norm": 0.08428606390953064, "global_step": 30244, "epoch": 840, "lr": 0.001} +{"train_loss": 0.025673843920230865, "train_loss_bc": 0.02565629594027996, "train_loss_llm": 0.0017547607421875, "grad_norm": 0.030699877068400383, "global_step": 30245, "epoch": 840, "lr": 0.001} +{"train_loss": 0.03592207282781601, "train_loss_bc": 0.03592207282781601, "train_loss_llm": 0.0, "grad_norm": 0.05335791036486626, "global_step": 30246, "epoch": 840, "lr": 0.001} +{"train_loss": 0.02801491878926754, "train_loss_bc": 0.02800951525568962, "train_loss_llm": 0.0005403459072113037, "grad_norm": 0.07194425165653229, "global_step": 30247, "epoch": 840, "lr": 0.001} +{"train_loss": 0.024437107145786285, "train_loss_bc": 0.024425460025668144, "train_loss_llm": 0.0011647045612335205, "grad_norm": 0.09288883209228516, "global_step": 30248, "epoch": 840, "lr": 0.001} +{"train_loss": 0.037108831107616425, "train_loss_bc": 0.037108831107616425, "train_loss_llm": 0.0, "grad_norm": 0.0401444137096405, "global_step": 30249, "epoch": 840, "lr": 0.001} +{"train_loss": 0.026878856122493744, "train_loss_bc": 0.026856880635023117, "train_loss_llm": 0.002197474241256714, "grad_norm": 0.06402713060379028, "global_step": 30250, "epoch": 840, "lr": 0.001} +{"train_loss": 0.02643483690917492, "train_loss_bc": 0.026419904083013535, "train_loss_llm": 0.0014932453632354736, "grad_norm": 0.07976625859737396, "global_step": 30251, "epoch": 840, "lr": 0.001} +{"train_loss": 0.031243862584233284, "train_loss_bc": 0.031243862584233284, "train_loss_llm": 0.0, "grad_norm": 0.10845894366502762, "global_step": 30252, "epoch": 840, "lr": 0.001} +{"train_loss": 0.032356034964323044, "train_loss_bc": 0.032356034964323044, "train_loss_llm": 0.0, "grad_norm": 0.0348343588411808, "global_step": 30253, "epoch": 840, "lr": 0.001} +{"train_loss": 0.02952970378100872, "train_loss_bc": 0.02952970378100872, "train_loss_llm": 0.0, "grad_norm": 0.05553562939167023, "global_step": 30254, "epoch": 840, "lr": 0.001} +{"train_loss": 0.028066502884030342, "train_loss_bc": 0.02806420437991619, "train_loss_llm": 0.00022977590560913086, "grad_norm": 0.07393412292003632, "global_step": 30255, "epoch": 840, "lr": 0.001} +{"train_loss": 0.032634615898132324, "train_loss_bc": 0.032634615898132324, "train_loss_llm": 0.0, "grad_norm": 0.0852552056312561, "global_step": 30256, "epoch": 840, "lr": 0.001} +{"train_loss": 0.03343822807073593, "train_loss_bc": 0.033429667353630066, "train_loss_llm": 0.0008559525012969971, "grad_norm": 0.04814077168703079, "global_step": 30257, "epoch": 840, "lr": 0.001} +{"train_loss": 0.027428116649389267, "train_loss_bc": 0.027428116649389267, "train_loss_llm": 0.0, "grad_norm": 0.07776318490505219, "global_step": 30258, "epoch": 840, "lr": 0.001} +{"train_loss": 0.026295874267816544, "train_loss_bc": 0.026295874267816544, "train_loss_llm": 0.0, "grad_norm": 0.09839193522930145, "global_step": 30259, "epoch": 840, "lr": 0.001} +{"train_loss": 0.030546315014362335, "train_loss_bc": 0.030546315014362335, "train_loss_llm": 0.0, "grad_norm": 0.13008524477481842, "global_step": 30260, "epoch": 840, "lr": 0.001} +{"train_loss": 0.02790532261133194, "train_loss_bc": 0.02790532261133194, "train_loss_llm": 0.0, "grad_norm": 0.04050757363438606, "global_step": 30261, "epoch": 840, "lr": 0.001} +{"train_loss": 0.02753853239119053, "train_loss_bc": 0.02753853239119053, "train_loss_llm": 0.0, "grad_norm": 0.05508292093873024, "global_step": 30262, "epoch": 840, "lr": 0.001} +{"train_loss": 0.028914617374539375, "train_loss_bc": 0.028914617374539375, "train_loss_llm": 0.0, "grad_norm": 0.07217144966125488, "global_step": 30263, "epoch": 840, "lr": 0.001} +{"train_loss": 0.034585848450660706, "train_loss_bc": 0.034585848450660706, "train_loss_llm": 0.0, "grad_norm": 0.10218577831983566, "global_step": 30264, "epoch": 840, "lr": 0.001} +{"train_loss": 0.02984887920320034, "train_loss_bc": 0.029842425137758255, "train_loss_llm": 0.0006454586982727051, "grad_norm": 0.04039045050740242, "global_step": 30265, "epoch": 840, "lr": 0.001} +{"train_loss": 0.03006684221327305, "train_loss_bc": 0.03006684221327305, "train_loss_llm": 0.0, "grad_norm": 0.05748773738741875, "global_step": 30266, "epoch": 840, "lr": 0.001} +{"train_loss": 0.03108169510960579, "train_loss_bc": 0.03108169510960579, "train_loss_llm": 0.0, "grad_norm": 0.08508908748626709, "global_step": 30267, "epoch": 840, "lr": 0.001} +{"train_loss": 0.0311716478317976, "train_loss_bc": 0.0311716478317976, "train_loss_llm": 0.0, "grad_norm": 0.10947012901306152, "global_step": 30268, "epoch": 840, "lr": 0.001} +{"train_loss": 0.03221878409385681, "train_loss_bc": 0.03221878409385681, "train_loss_llm": 0.0, "grad_norm": 0.047640349715948105, "global_step": 30269, "epoch": 840, "lr": 0.001} +{"train_loss": 0.026294274255633354, "train_loss_bc": 0.026294274255633354, "train_loss_llm": 0.0, "grad_norm": 0.08058971911668777, "global_step": 30270, "epoch": 840, "lr": 0.001} +{"train_loss": 0.031224414706230164, "train_loss_bc": 0.031224414706230164, "train_loss_llm": 0.0, "grad_norm": 0.09939338266849518, "global_step": 30271, "epoch": 840, "lr": 0.001} +{"train_loss": 0.03218737989664078, "train_loss_bc": 0.03218737989664078, "train_loss_llm": 0.0, "grad_norm": 0.12796247005462646, "global_step": 30272, "epoch": 840, "lr": 0.001} +{"train_loss": 0.029710819944739342, "train_loss_bc": 0.029710819944739342, "train_loss_llm": 0.0, "grad_norm": 0.044458962976932526, "global_step": 30273, "epoch": 840, "lr": 0.001} +{"train_loss": 0.034238796681165695, "train_loss_bc": 0.034238796681165695, "train_loss_llm": 0.0, "grad_norm": 0.07897192239761353, "global_step": 30274, "epoch": 840, "lr": 0.001} +{"train_loss": 0.0300204088497493, "train_loss_bc": 0.02646271325647831, "train_loss_llm": 0.0, "grad_norm": 0.10459008812904358, "global_step": 30275, "epoch": 840, "lr": 0.001, "train/cumulative_reward": 50.370362288301756, "train/mean_score": 3.5401176469142484, "train/success_rate": 0.0, "test/cumulative_reward": 37.22084597008438, "test/mean_score": 8.474046565036161, "test/success_rate": 0.22, "val_loss": 0.1000027135014534, "train_action_mse_error": 0.03328634798526764} +{"train_loss": 0.031617142260074615, "train_loss_bc": 0.031617142260074615, "train_loss_llm": 0.0, "grad_norm": 0.13889232277870178, "global_step": 30276, "epoch": 841, "lr": 0.001} +{"train_loss": 0.03025679476559162, "train_loss_bc": 0.030249180272221565, "train_loss_llm": 0.0007614493370056152, "grad_norm": 0.03575065732002258, "global_step": 30277, "epoch": 841, "lr": 0.001} +{"train_loss": 0.030989382416009903, "train_loss_bc": 0.030989382416009903, "train_loss_llm": 0.0, "grad_norm": 0.05446549877524376, "global_step": 30278, "epoch": 841, "lr": 0.001} +{"train_loss": 0.03087131306529045, "train_loss_bc": 0.03087131306529045, "train_loss_llm": 0.0, "grad_norm": 0.0707617923617363, "global_step": 30279, "epoch": 841, "lr": 0.001} +{"train_loss": 0.024906069040298462, "train_loss_bc": 0.024906069040298462, "train_loss_llm": 0.0, "grad_norm": 0.08786880970001221, "global_step": 30280, "epoch": 841, "lr": 0.001} +{"train_loss": 0.029386034235358238, "train_loss_bc": 0.029377533122897148, "train_loss_llm": 0.0008501410484313965, "grad_norm": 0.03486550971865654, "global_step": 30281, "epoch": 841, "lr": 0.001} +{"train_loss": 0.031494151800870895, "train_loss_bc": 0.031494151800870895, "train_loss_llm": 0.0, "grad_norm": 0.05287168547511101, "global_step": 30282, "epoch": 841, "lr": 0.001} +{"train_loss": 0.02620839886367321, "train_loss_bc": 0.02620839886367321, "train_loss_llm": 0.0, "grad_norm": 0.06885059922933578, "global_step": 30283, "epoch": 841, "lr": 0.001} +{"train_loss": 0.030792031437158585, "train_loss_bc": 0.030784504488110542, "train_loss_llm": 0.0007527172565460205, "grad_norm": 0.07904775440692902, "global_step": 30284, "epoch": 841, "lr": 0.001} +{"train_loss": 0.03140626102685928, "train_loss_bc": 0.03140626102685928, "train_loss_llm": 0.0, "grad_norm": 0.03493829071521759, "global_step": 30285, "epoch": 841, "lr": 0.001} +{"train_loss": 0.034896280616521835, "train_loss_bc": 0.03489428013563156, "train_loss_llm": 0.00020009279251098633, "grad_norm": 0.04888846352696419, "global_step": 30286, "epoch": 841, "lr": 0.001} +{"train_loss": 0.0350814126431942, "train_loss_bc": 0.0350814126431942, "train_loss_llm": 0.0, "grad_norm": 0.07284127175807953, "global_step": 30287, "epoch": 841, "lr": 0.001} +{"train_loss": 0.026310447603464127, "train_loss_bc": 0.026310447603464127, "train_loss_llm": 0.0, "grad_norm": 0.08926742523908615, "global_step": 30288, "epoch": 841, "lr": 0.001} +{"train_loss": 0.032611068338155746, "train_loss_bc": 0.03260726481676102, "train_loss_llm": 0.0003803074359893799, "grad_norm": 0.0526030957698822, "global_step": 30289, "epoch": 841, "lr": 0.001} +{"train_loss": 0.03021964803338051, "train_loss_bc": 0.03021964803338051, "train_loss_llm": 0.0, "grad_norm": 0.09176908433437347, "global_step": 30290, "epoch": 841, "lr": 0.001} +{"train_loss": 0.03190851956605911, "train_loss_bc": 0.03190851956605911, "train_loss_llm": 0.0, "grad_norm": 0.11038810759782791, "global_step": 30291, "epoch": 841, "lr": 0.001} +{"train_loss": 0.031249158084392548, "train_loss_bc": 0.031249158084392548, "train_loss_llm": 0.0, "grad_norm": 0.15041428804397583, "global_step": 30292, "epoch": 841, "lr": 0.001} +{"train_loss": 0.026614660397171974, "train_loss_bc": 0.026604212820529938, "train_loss_llm": 0.0010446906089782715, "grad_norm": 0.041418738663196564, "global_step": 30293, "epoch": 841, "lr": 0.001} +{"train_loss": 0.031495075672864914, "train_loss_bc": 0.031495075672864914, "train_loss_llm": 0.0, "grad_norm": 0.07879336178302765, "global_step": 30294, "epoch": 841, "lr": 0.001} +{"train_loss": 0.02717553824186325, "train_loss_bc": 0.02717553824186325, "train_loss_llm": 0.0, "grad_norm": 0.10997021198272705, "global_step": 30295, "epoch": 841, "lr": 0.001} +{"train_loss": 0.02577901817858219, "train_loss_bc": 0.02577901817858219, "train_loss_llm": 0.0, "grad_norm": 0.12426875531673431, "global_step": 30296, "epoch": 841, "lr": 0.001} +{"train_loss": 0.02680964209139347, "train_loss_bc": 0.026796801015734673, "train_loss_llm": 0.0012841224670410156, "grad_norm": 0.031948916614055634, "global_step": 30297, "epoch": 841, "lr": 0.001} +{"train_loss": 0.027780719101428986, "train_loss_bc": 0.027780719101428986, "train_loss_llm": 0.0, "grad_norm": 0.049381472170352936, "global_step": 30298, "epoch": 841, "lr": 0.001} +{"train_loss": 0.03793635591864586, "train_loss_bc": 0.03793635591864586, "train_loss_llm": 0.0, "grad_norm": 0.06980134546756744, "global_step": 30299, "epoch": 841, "lr": 0.001} +{"train_loss": 0.026077883318066597, "train_loss_bc": 0.026077883318066597, "train_loss_llm": 0.0, "grad_norm": 0.08767103403806686, "global_step": 30300, "epoch": 841, "lr": 0.001} +{"train_loss": 0.030191941186785698, "train_loss_bc": 0.030191941186785698, "train_loss_llm": 0.0, "grad_norm": 0.03325086832046509, "global_step": 30301, "epoch": 841, "lr": 0.001} +{"train_loss": 0.02607775665819645, "train_loss_bc": 0.02607775665819645, "train_loss_llm": 0.0, "grad_norm": 0.04632626846432686, "global_step": 30302, "epoch": 841, "lr": 0.001} +{"train_loss": 0.03064402937889099, "train_loss_bc": 0.03064402937889099, "train_loss_llm": 0.0, "grad_norm": 0.07199138402938843, "global_step": 30303, "epoch": 841, "lr": 0.001} +{"train_loss": 0.02642756700515747, "train_loss_bc": 0.02642756700515747, "train_loss_llm": 0.0, "grad_norm": 0.08530878275632858, "global_step": 30304, "epoch": 841, "lr": 0.001} +{"train_loss": 0.02639091946184635, "train_loss_bc": 0.02639091946184635, "train_loss_llm": 0.0, "grad_norm": 0.03356502577662468, "global_step": 30305, "epoch": 841, "lr": 0.001} +{"train_loss": 0.02803235687315464, "train_loss_bc": 0.02803235687315464, "train_loss_llm": 0.0, "grad_norm": 0.060516346246004105, "global_step": 30306, "epoch": 841, "lr": 0.001} +{"train_loss": 0.033220324665308, "train_loss_bc": 0.033206939697265625, "train_loss_llm": 0.0013384521007537842, "grad_norm": 0.0922345444560051, "global_step": 30307, "epoch": 841, "lr": 0.001} +{"train_loss": 0.03627853840589523, "train_loss_bc": 0.03627853840589523, "train_loss_llm": 0.0, "grad_norm": 0.113373301923275, "global_step": 30308, "epoch": 841, "lr": 0.001} +{"train_loss": 0.031155012547969818, "train_loss_bc": 0.031155012547969818, "train_loss_llm": 0.0, "grad_norm": 0.03817185014486313, "global_step": 30309, "epoch": 841, "lr": 0.001} +{"train_loss": 0.03464870899915695, "train_loss_bc": 0.03464193642139435, "train_loss_llm": 0.0006771981716156006, "grad_norm": 0.06367425620555878, "global_step": 30310, "epoch": 841, "lr": 0.001} +{"train_loss": 0.03007560668306218, "train_loss_bc": 0.029781678691506386, "train_loss_llm": 0.0, "grad_norm": 0.10536375641822815, "global_step": 30311, "epoch": 841, "lr": 0.001, "val_loss": 0.09818142652511597} +{"train_loss": 0.03309341147542, "train_loss_bc": 0.03309341147542, "train_loss_llm": 0.0, "grad_norm": 0.13481827080249786, "global_step": 30312, "epoch": 842, "lr": 0.001} +{"train_loss": 0.03056819550693035, "train_loss_bc": 0.03056819550693035, "train_loss_llm": 0.0, "grad_norm": 0.03819427266716957, "global_step": 30313, "epoch": 842, "lr": 0.001} +{"train_loss": 0.029862025752663612, "train_loss_bc": 0.02985256165266037, "train_loss_llm": 0.0009464025497436523, "grad_norm": 0.06162850186228752, "global_step": 30314, "epoch": 842, "lr": 0.001} +{"train_loss": 0.028888588771224022, "train_loss_bc": 0.028888588771224022, "train_loss_llm": 0.0, "grad_norm": 0.093772754073143, "global_step": 30315, "epoch": 842, "lr": 0.001} +{"train_loss": 0.03175576776266098, "train_loss_bc": 0.03175576776266098, "train_loss_llm": 0.0, "grad_norm": 0.11501246690750122, "global_step": 30316, "epoch": 842, "lr": 0.001} +{"train_loss": 0.0309772789478302, "train_loss_bc": 0.0309772789478302, "train_loss_llm": 0.0, "grad_norm": 0.036795686930418015, "global_step": 30317, "epoch": 842, "lr": 0.001} +{"train_loss": 0.023034438490867615, "train_loss_bc": 0.023034438490867615, "train_loss_llm": 0.0, "grad_norm": 0.05579644814133644, "global_step": 30318, "epoch": 842, "lr": 0.001} +{"train_loss": 0.03423555940389633, "train_loss_bc": 0.03423555940389633, "train_loss_llm": 0.0, "grad_norm": 0.0800788626074791, "global_step": 30319, "epoch": 842, "lr": 0.001} +{"train_loss": 0.02642080932855606, "train_loss_bc": 0.02642080932855606, "train_loss_llm": 0.0, "grad_norm": 0.09697216004133224, "global_step": 30320, "epoch": 842, "lr": 0.001} +{"train_loss": 0.034735556691884995, "train_loss_bc": 0.034735556691884995, "train_loss_llm": 0.0, "grad_norm": 0.05076931044459343, "global_step": 30321, "epoch": 842, "lr": 0.001} +{"train_loss": 0.02688632160425186, "train_loss_bc": 0.02688632160425186, "train_loss_llm": 0.0, "grad_norm": 0.06488312035799026, "global_step": 30322, "epoch": 842, "lr": 0.001} +{"train_loss": 0.025422317907214165, "train_loss_bc": 0.025422317907214165, "train_loss_llm": 0.0, "grad_norm": 0.08523736894130707, "global_step": 30323, "epoch": 842, "lr": 0.001} +{"train_loss": 0.0320955254137516, "train_loss_bc": 0.03208184987306595, "train_loss_llm": 0.0013674795627593994, "grad_norm": 0.10131987929344177, "global_step": 30324, "epoch": 842, "lr": 0.001} +{"train_loss": 0.030375506728887558, "train_loss_bc": 0.030375506728887558, "train_loss_llm": 0.0, "grad_norm": 0.030689451843500137, "global_step": 30325, "epoch": 842, "lr": 0.001} +{"train_loss": 0.029460767284035683, "train_loss_bc": 0.02945329062640667, "train_loss_llm": 0.0007477402687072754, "grad_norm": 0.05699000880122185, "global_step": 30326, "epoch": 842, "lr": 0.001} +{"train_loss": 0.02590434066951275, "train_loss_bc": 0.02590434066951275, "train_loss_llm": 0.0, "grad_norm": 0.07042957842350006, "global_step": 30327, "epoch": 842, "lr": 0.001} +{"train_loss": 0.02490166574716568, "train_loss_bc": 0.02490166574716568, "train_loss_llm": 0.0, "grad_norm": 0.08631271868944168, "global_step": 30328, "epoch": 842, "lr": 0.001} +{"train_loss": 0.02690204232931137, "train_loss_bc": 0.02690204232931137, "train_loss_llm": 0.0, "grad_norm": 0.030844412744045258, "global_step": 30329, "epoch": 842, "lr": 0.001} +{"train_loss": 0.02869882807135582, "train_loss_bc": 0.02867938205599785, "train_loss_llm": 0.0019446909427642822, "grad_norm": 0.057069599628448486, "global_step": 30330, "epoch": 842, "lr": 0.001} +{"train_loss": 0.027801144868135452, "train_loss_bc": 0.027801144868135452, "train_loss_llm": 0.0, "grad_norm": 0.07238531112670898, "global_step": 30331, "epoch": 842, "lr": 0.001} +{"train_loss": 0.031420089304447174, "train_loss_bc": 0.031420089304447174, "train_loss_llm": 0.0, "grad_norm": 0.0905437022447586, "global_step": 30332, "epoch": 842, "lr": 0.001} +{"train_loss": 0.02902238443493843, "train_loss_bc": 0.02902238443493843, "train_loss_llm": 0.0, "grad_norm": 0.033500466495752335, "global_step": 30333, "epoch": 842, "lr": 0.001} +{"train_loss": 0.032604385167360306, "train_loss_bc": 0.03258547931909561, "train_loss_llm": 0.001890718936920166, "grad_norm": 0.06403211504220963, "global_step": 30334, "epoch": 842, "lr": 0.001} +{"train_loss": 0.03296709805727005, "train_loss_bc": 0.03296303376555443, "train_loss_llm": 0.000406414270401001, "grad_norm": 0.07772207260131836, "global_step": 30335, "epoch": 842, "lr": 0.001} +{"train_loss": 0.02974284067749977, "train_loss_bc": 0.029718216508626938, "train_loss_llm": 0.00246235728263855, "grad_norm": 0.0897812694311142, "global_step": 30336, "epoch": 842, "lr": 0.001} +{"train_loss": 0.022859662771224976, "train_loss_bc": 0.022859662771224976, "train_loss_llm": 0.0, "grad_norm": 0.03811638057231903, "global_step": 30337, "epoch": 842, "lr": 0.001} +{"train_loss": 0.03459576517343521, "train_loss_bc": 0.03459036350250244, "train_loss_llm": 0.0005399882793426514, "grad_norm": 0.06874526292085648, "global_step": 30338, "epoch": 842, "lr": 0.001} +{"train_loss": 0.03474078327417374, "train_loss_bc": 0.03474078327417374, "train_loss_llm": 0.0, "grad_norm": 0.0909838080406189, "global_step": 30339, "epoch": 842, "lr": 0.001} +{"train_loss": 0.02936217375099659, "train_loss_bc": 0.029357748106122017, "train_loss_llm": 0.0004425942897796631, "grad_norm": 0.10955802351236343, "global_step": 30340, "epoch": 842, "lr": 0.001} +{"train_loss": 0.026576386764645576, "train_loss_bc": 0.026576386764645576, "train_loss_llm": 0.0, "grad_norm": 0.027774419635534286, "global_step": 30341, "epoch": 842, "lr": 0.001} +{"train_loss": 0.031321026384830475, "train_loss_bc": 0.031321026384830475, "train_loss_llm": 0.0, "grad_norm": 0.047336868941783905, "global_step": 30342, "epoch": 842, "lr": 0.001} +{"train_loss": 0.03701125085353851, "train_loss_bc": 0.03701125085353851, "train_loss_llm": 0.0, "grad_norm": 0.07028781622648239, "global_step": 30343, "epoch": 842, "lr": 0.001} +{"train_loss": 0.030070388689637184, "train_loss_bc": 0.03006613813340664, "train_loss_llm": 0.00042513012886047363, "grad_norm": 0.09095205366611481, "global_step": 30344, "epoch": 842, "lr": 0.001} +{"train_loss": 0.028484292328357697, "train_loss_bc": 0.028484292328357697, "train_loss_llm": 0.0, "grad_norm": 0.03485485538840294, "global_step": 30345, "epoch": 842, "lr": 0.001} +{"train_loss": 0.029602954164147377, "train_loss_bc": 0.029602954164147377, "train_loss_llm": 0.0, "grad_norm": 0.0529390387237072, "global_step": 30346, "epoch": 842, "lr": 0.001} +{"train_loss": 0.029893805790278647, "train_loss_bc": 0.03377543389797211, "train_loss_llm": 0.0, "grad_norm": 0.06886139512062073, "global_step": 30347, "epoch": 842, "lr": 0.001, "val_loss": 0.09818552434444427} +{"train_loss": 0.027450066059827805, "train_loss_bc": 0.027432216331362724, "train_loss_llm": 0.0017849206924438477, "grad_norm": 0.08600190281867981, "global_step": 30348, "epoch": 843, "lr": 0.001} +{"train_loss": 0.031187493354082108, "train_loss_bc": 0.031187493354082108, "train_loss_llm": 0.0, "grad_norm": 0.026939719915390015, "global_step": 30349, "epoch": 843, "lr": 0.001} +{"train_loss": 0.026205839589238167, "train_loss_bc": 0.026205839589238167, "train_loss_llm": 0.0, "grad_norm": 0.04371289536356926, "global_step": 30350, "epoch": 843, "lr": 0.001} +{"train_loss": 0.030120063573122025, "train_loss_bc": 0.030118459835648537, "train_loss_llm": 0.00016039609909057617, "grad_norm": 0.06004969775676727, "global_step": 30351, "epoch": 843, "lr": 0.001} +{"train_loss": 0.02765013836324215, "train_loss_bc": 0.02765013836324215, "train_loss_llm": 0.0, "grad_norm": 0.08149290084838867, "global_step": 30352, "epoch": 843, "lr": 0.001} +{"train_loss": 0.02375577576458454, "train_loss_bc": 0.023751992732286453, "train_loss_llm": 0.0003783106803894043, "grad_norm": 0.045896489173173904, "global_step": 30353, "epoch": 843, "lr": 0.001} +{"train_loss": 0.02999073825776577, "train_loss_bc": 0.02999073825776577, "train_loss_llm": 0.0, "grad_norm": 0.06737197190523148, "global_step": 30354, "epoch": 843, "lr": 0.001} +{"train_loss": 0.03011656180024147, "train_loss_bc": 0.03011656180024147, "train_loss_llm": 0.0, "grad_norm": 0.09656383097171783, "global_step": 30355, "epoch": 843, "lr": 0.001} +{"train_loss": 0.03111959435045719, "train_loss_bc": 0.03111959435045719, "train_loss_llm": 0.0, "grad_norm": 0.12132265418767929, "global_step": 30356, "epoch": 843, "lr": 0.001} +{"train_loss": 0.02825981192290783, "train_loss_bc": 0.02825981192290783, "train_loss_llm": 0.0, "grad_norm": 0.039453379809856415, "global_step": 30357, "epoch": 843, "lr": 0.001} +{"train_loss": 0.02906169928610325, "train_loss_bc": 0.029048603028059006, "train_loss_llm": 0.0013096630573272705, "grad_norm": 0.05410226806998253, "global_step": 30358, "epoch": 843, "lr": 0.001} +{"train_loss": 0.031499288976192474, "train_loss_bc": 0.031499288976192474, "train_loss_llm": 0.0, "grad_norm": 0.0699000284075737, "global_step": 30359, "epoch": 843, "lr": 0.001} +{"train_loss": 0.02749388851225376, "train_loss_bc": 0.02749388851225376, "train_loss_llm": 0.0, "grad_norm": 0.0945027619600296, "global_step": 30360, "epoch": 843, "lr": 0.001} +{"train_loss": 0.03169526532292366, "train_loss_bc": 0.03169526532292366, "train_loss_llm": 0.0, "grad_norm": 0.03673984110355377, "global_step": 30361, "epoch": 843, "lr": 0.001} +{"train_loss": 0.03184133023023605, "train_loss_bc": 0.03184133023023605, "train_loss_llm": 0.0, "grad_norm": 0.06635687500238419, "global_step": 30362, "epoch": 843, "lr": 0.001} +{"train_loss": 0.029306188225746155, "train_loss_bc": 0.029306188225746155, "train_loss_llm": 0.0, "grad_norm": 0.09155559539794922, "global_step": 30363, "epoch": 843, "lr": 0.001} +{"train_loss": 0.033005040138959885, "train_loss_bc": 0.033005040138959885, "train_loss_llm": 0.0, "grad_norm": 0.11587244272232056, "global_step": 30364, "epoch": 843, "lr": 0.001} +{"train_loss": 0.038743823766708374, "train_loss_bc": 0.038743823766708374, "train_loss_llm": 0.0, "grad_norm": 0.042527440935373306, "global_step": 30365, "epoch": 843, "lr": 0.001} +{"train_loss": 0.02815268561244011, "train_loss_bc": 0.02815268561244011, "train_loss_llm": 0.0, "grad_norm": 0.059882380068302155, "global_step": 30366, "epoch": 843, "lr": 0.001} +{"train_loss": 0.03198549523949623, "train_loss_bc": 0.03197186440229416, "train_loss_llm": 0.0013630986213684082, "grad_norm": 0.07961690425872803, "global_step": 30367, "epoch": 843, "lr": 0.001} +{"train_loss": 0.025642873719334602, "train_loss_bc": 0.025642873719334602, "train_loss_llm": 0.0, "grad_norm": 0.10424381494522095, "global_step": 30368, "epoch": 843, "lr": 0.001} +{"train_loss": 0.025910869240760803, "train_loss_bc": 0.025905728340148926, "train_loss_llm": 0.0005141198635101318, "grad_norm": 0.040090229362249374, "global_step": 30369, "epoch": 843, "lr": 0.001} +{"train_loss": 0.02921896241605282, "train_loss_bc": 0.02921896241605282, "train_loss_llm": 0.0, "grad_norm": 0.06321758776903152, "global_step": 30370, "epoch": 843, "lr": 0.001} +{"train_loss": 0.03416582942008972, "train_loss_bc": 0.03414192050695419, "train_loss_llm": 0.0023907721042633057, "grad_norm": 0.09315116703510284, "global_step": 30371, "epoch": 843, "lr": 0.001} +{"train_loss": 0.03191124275326729, "train_loss_bc": 0.03191124275326729, "train_loss_llm": 0.0, "grad_norm": 0.11435109376907349, "global_step": 30372, "epoch": 843, "lr": 0.001} +{"train_loss": 0.031405020505189896, "train_loss_bc": 0.031405020505189896, "train_loss_llm": 0.0, "grad_norm": 0.03242821246385574, "global_step": 30373, "epoch": 843, "lr": 0.001} +{"train_loss": 0.029203370213508606, "train_loss_bc": 0.029203370213508606, "train_loss_llm": 0.0, "grad_norm": 0.056042324751615524, "global_step": 30374, "epoch": 843, "lr": 0.001} +{"train_loss": 0.03295181319117546, "train_loss_bc": 0.03295181319117546, "train_loss_llm": 0.0, "grad_norm": 0.0731082558631897, "global_step": 30375, "epoch": 843, "lr": 0.001} +{"train_loss": 0.030729617923498154, "train_loss_bc": 0.030729617923498154, "train_loss_llm": 0.0, "grad_norm": 0.0850638672709465, "global_step": 30376, "epoch": 843, "lr": 0.001} +{"train_loss": 0.027542918920516968, "train_loss_bc": 0.027522582560777664, "train_loss_llm": 0.0020336806774139404, "grad_norm": 0.03373490646481514, "global_step": 30377, "epoch": 843, "lr": 0.001} +{"train_loss": 0.028882458806037903, "train_loss_bc": 0.028874782845377922, "train_loss_llm": 0.0007675290107727051, "grad_norm": 0.06048179417848587, "global_step": 30378, "epoch": 843, "lr": 0.001} +{"train_loss": 0.03451332077383995, "train_loss_bc": 0.034507907927036285, "train_loss_llm": 0.0005413591861724854, "grad_norm": 0.08915452659130096, "global_step": 30379, "epoch": 843, "lr": 0.001} +{"train_loss": 0.032148491591215134, "train_loss_bc": 0.032148491591215134, "train_loss_llm": 0.0, "grad_norm": 0.10924018174409866, "global_step": 30380, "epoch": 843, "lr": 0.001} +{"train_loss": 0.026620937511324883, "train_loss_bc": 0.026620937511324883, "train_loss_llm": 0.0, "grad_norm": 0.0322183258831501, "global_step": 30381, "epoch": 843, "lr": 0.001} +{"train_loss": 0.03012729249894619, "train_loss_bc": 0.03012729249894619, "train_loss_llm": 0.0, "grad_norm": 0.05498381331562996, "global_step": 30382, "epoch": 843, "lr": 0.001} +{"train_loss": 0.02990679970631997, "train_loss_bc": 0.027026599273085594, "train_loss_llm": 0.00023820996284484863, "grad_norm": 0.07718761265277863, "global_step": 30383, "epoch": 843, "lr": 0.001, "val_loss": 0.09963919967412949} +{"train_loss": 0.02715385891497135, "train_loss_bc": 0.02715385891497135, "train_loss_llm": 0.0, "grad_norm": 0.0952916070818901, "global_step": 30384, "epoch": 844, "lr": 0.001} +{"train_loss": 0.030479948967695236, "train_loss_bc": 0.030464205890893936, "train_loss_llm": 0.001574397087097168, "grad_norm": 0.038684334605932236, "global_step": 30385, "epoch": 844, "lr": 0.001} +{"train_loss": 0.02982807345688343, "train_loss_bc": 0.029817068949341774, "train_loss_llm": 0.0011004805564880371, "grad_norm": 0.05866294354200363, "global_step": 30386, "epoch": 844, "lr": 0.001} +{"train_loss": 0.03395305946469307, "train_loss_bc": 0.03395305946469307, "train_loss_llm": 0.0, "grad_norm": 0.08362176269292831, "global_step": 30387, "epoch": 844, "lr": 0.001} +{"train_loss": 0.025972584262490273, "train_loss_bc": 0.025970343500375748, "train_loss_llm": 0.00022408366203308105, "grad_norm": 0.09914122521877289, "global_step": 30388, "epoch": 844, "lr": 0.001} +{"train_loss": 0.03212040290236473, "train_loss_bc": 0.03212040290236473, "train_loss_llm": 0.0, "grad_norm": 0.03563995286822319, "global_step": 30389, "epoch": 844, "lr": 0.001} +{"train_loss": 0.03407984599471092, "train_loss_bc": 0.03407984599471092, "train_loss_llm": 0.0, "grad_norm": 0.05604828894138336, "global_step": 30390, "epoch": 844, "lr": 0.001} +{"train_loss": 0.03174963593482971, "train_loss_bc": 0.03174963593482971, "train_loss_llm": 0.0, "grad_norm": 0.07650555670261383, "global_step": 30391, "epoch": 844, "lr": 0.001} +{"train_loss": 0.03256921097636223, "train_loss_bc": 0.03256921097636223, "train_loss_llm": 0.0, "grad_norm": 0.09877519309520721, "global_step": 30392, "epoch": 844, "lr": 0.001} +{"train_loss": 0.02862781286239624, "train_loss_bc": 0.02862781286239624, "train_loss_llm": 0.0, "grad_norm": 0.03535115718841553, "global_step": 30393, "epoch": 844, "lr": 0.001} +{"train_loss": 0.028648434206843376, "train_loss_bc": 0.028648434206843376, "train_loss_llm": 0.0, "grad_norm": 0.0625145435333252, "global_step": 30394, "epoch": 844, "lr": 0.001} +{"train_loss": 0.029493024572730064, "train_loss_bc": 0.029486283659934998, "train_loss_llm": 0.0006740689277648926, "grad_norm": 0.08352614939212799, "global_step": 30395, "epoch": 844, "lr": 0.001} +{"train_loss": 0.0342182032763958, "train_loss_bc": 0.0342182032763958, "train_loss_llm": 0.0, "grad_norm": 0.09283844381570816, "global_step": 30396, "epoch": 844, "lr": 0.001} +{"train_loss": 0.033766597509384155, "train_loss_bc": 0.033766597509384155, "train_loss_llm": 0.0, "grad_norm": 0.04429832100868225, "global_step": 30397, "epoch": 844, "lr": 0.001} +{"train_loss": 0.030147608369588852, "train_loss_bc": 0.030145281925797462, "train_loss_llm": 0.0002327263355255127, "grad_norm": 0.0636214017868042, "global_step": 30398, "epoch": 844, "lr": 0.001} +{"train_loss": 0.0286570955067873, "train_loss_bc": 0.028653182089328766, "train_loss_llm": 0.00039130449295043945, "grad_norm": 0.08394618332386017, "global_step": 30399, "epoch": 844, "lr": 0.001} +{"train_loss": 0.02565162256360054, "train_loss_bc": 0.02565162256360054, "train_loss_llm": 0.0, "grad_norm": 0.10416613519191742, "global_step": 30400, "epoch": 844, "lr": 0.001} +{"train_loss": 0.03057962842285633, "train_loss_bc": 0.03056975081562996, "train_loss_llm": 0.0009877681732177734, "grad_norm": 0.03331045061349869, "global_step": 30401, "epoch": 844, "lr": 0.001} +{"train_loss": 0.030080126598477364, "train_loss_bc": 0.030077017843723297, "train_loss_llm": 0.0003109574317932129, "grad_norm": 0.05862311273813248, "global_step": 30402, "epoch": 844, "lr": 0.001} +{"train_loss": 0.03022623062133789, "train_loss_bc": 0.03022623062133789, "train_loss_llm": 0.0, "grad_norm": 0.06688188016414642, "global_step": 30403, "epoch": 844, "lr": 0.001} +{"train_loss": 0.027281425893306732, "train_loss_bc": 0.027281425893306732, "train_loss_llm": 0.0, "grad_norm": 0.08833316713571548, "global_step": 30404, "epoch": 844, "lr": 0.001} +{"train_loss": 0.028819236904382706, "train_loss_bc": 0.028799427673220634, "train_loss_llm": 0.0019808411598205566, "grad_norm": 0.03612418845295906, "global_step": 30405, "epoch": 844, "lr": 0.001} +{"train_loss": 0.029103737324476242, "train_loss_bc": 0.029103737324476242, "train_loss_llm": 0.0, "grad_norm": 0.057530369609594345, "global_step": 30406, "epoch": 844, "lr": 0.001} +{"train_loss": 0.032589107751846313, "train_loss_bc": 0.03258382901549339, "train_loss_llm": 0.0005278289318084717, "grad_norm": 0.0785357877612114, "global_step": 30407, "epoch": 844, "lr": 0.001} +{"train_loss": 0.02977467142045498, "train_loss_bc": 0.02977467142045498, "train_loss_llm": 0.0, "grad_norm": 0.09104812145233154, "global_step": 30408, "epoch": 844, "lr": 0.001} +{"train_loss": 0.027393072843551636, "train_loss_bc": 0.027393072843551636, "train_loss_llm": 0.0, "grad_norm": 0.028926940634846687, "global_step": 30409, "epoch": 844, "lr": 0.001} +{"train_loss": 0.03054877370595932, "train_loss_bc": 0.03054877370595932, "train_loss_llm": 0.0, "grad_norm": 0.05123404413461685, "global_step": 30410, "epoch": 844, "lr": 0.001} +{"train_loss": 0.029310990124940872, "train_loss_bc": 0.029310990124940872, "train_loss_llm": 0.0, "grad_norm": 0.07895693928003311, "global_step": 30411, "epoch": 844, "lr": 0.001} +{"train_loss": 0.035546574741601944, "train_loss_bc": 0.035546574741601944, "train_loss_llm": 0.0, "grad_norm": 0.09851357340812683, "global_step": 30412, "epoch": 844, "lr": 0.001} +{"train_loss": 0.029574355110526085, "train_loss_bc": 0.029574355110526085, "train_loss_llm": 0.0, "grad_norm": 0.042608391493558884, "global_step": 30413, "epoch": 844, "lr": 0.001} +{"train_loss": 0.02735448069870472, "train_loss_bc": 0.02735448069870472, "train_loss_llm": 0.0, "grad_norm": 0.06567138433456421, "global_step": 30414, "epoch": 844, "lr": 0.001} +{"train_loss": 0.025551198050379753, "train_loss_bc": 0.025551198050379753, "train_loss_llm": 0.0, "grad_norm": 0.0816015750169754, "global_step": 30415, "epoch": 844, "lr": 0.001} +{"train_loss": 0.02907971665263176, "train_loss_bc": 0.02907971665263176, "train_loss_llm": 0.0, "grad_norm": 0.09508483111858368, "global_step": 30416, "epoch": 844, "lr": 0.001} +{"train_loss": 0.029466260224580765, "train_loss_bc": 0.029466260224580765, "train_loss_llm": 0.0, "grad_norm": 0.03846026957035065, "global_step": 30417, "epoch": 844, "lr": 0.001} +{"train_loss": 0.026508620008826256, "train_loss_bc": 0.026505718007683754, "train_loss_llm": 0.00029027462005615234, "grad_norm": 0.05762369558215141, "global_step": 30418, "epoch": 844, "lr": 0.001} +{"train_loss": 0.029946859253363475, "train_loss_bc": 0.03218170627951622, "train_loss_llm": 0.0, "grad_norm": 0.0886056125164032, "global_step": 30419, "epoch": 844, "lr": 0.001, "val_loss": 0.10342574119567871} +{"train_loss": 0.027688032016158104, "train_loss_bc": 0.027688032016158104, "train_loss_llm": 0.0, "grad_norm": 0.10871819406747818, "global_step": 30420, "epoch": 845, "lr": 0.001} +{"train_loss": 0.02770494855940342, "train_loss_bc": 0.027698693796992302, "train_loss_llm": 0.0006255209445953369, "grad_norm": 0.04360383376479149, "global_step": 30421, "epoch": 845, "lr": 0.001} +{"train_loss": 0.03175918385386467, "train_loss_bc": 0.031756769865751266, "train_loss_llm": 0.00024121999740600586, "grad_norm": 0.07023873925209045, "global_step": 30422, "epoch": 845, "lr": 0.001} +{"train_loss": 0.028091011568903923, "train_loss_bc": 0.028091011568903923, "train_loss_llm": 0.0, "grad_norm": 0.08572237193584442, "global_step": 30423, "epoch": 845, "lr": 0.001} +{"train_loss": 0.033761367201805115, "train_loss_bc": 0.033761367201805115, "train_loss_llm": 0.0, "grad_norm": 0.10162951052188873, "global_step": 30424, "epoch": 845, "lr": 0.001} +{"train_loss": 0.03004338964819908, "train_loss_bc": 0.03004338964819908, "train_loss_llm": 0.0, "grad_norm": 0.03334910050034523, "global_step": 30425, "epoch": 845, "lr": 0.001} +{"train_loss": 0.0310373492538929, "train_loss_bc": 0.0310373492538929, "train_loss_llm": 0.0, "grad_norm": 0.056949928402900696, "global_step": 30426, "epoch": 845, "lr": 0.001} +{"train_loss": 0.027555087581276894, "train_loss_bc": 0.027555087581276894, "train_loss_llm": 0.0, "grad_norm": 0.07285717129707336, "global_step": 30427, "epoch": 845, "lr": 0.001} +{"train_loss": 0.03199099004268646, "train_loss_bc": 0.03199099004268646, "train_loss_llm": 0.0, "grad_norm": 0.0920584574341774, "global_step": 30428, "epoch": 845, "lr": 0.001} +{"train_loss": 0.03139572963118553, "train_loss_bc": 0.03139572963118553, "train_loss_llm": 0.0, "grad_norm": 0.03856557980179787, "global_step": 30429, "epoch": 845, "lr": 0.001} +{"train_loss": 0.03230303153395653, "train_loss_bc": 0.03230303153395653, "train_loss_llm": 0.0, "grad_norm": 0.05915544927120209, "global_step": 30430, "epoch": 845, "lr": 0.001} +{"train_loss": 0.030298572033643723, "train_loss_bc": 0.030298572033643723, "train_loss_llm": 0.0, "grad_norm": 0.07439853996038437, "global_step": 30431, "epoch": 845, "lr": 0.001} +{"train_loss": 0.036218974739313126, "train_loss_bc": 0.03620294854044914, "train_loss_llm": 0.0016025006771087646, "grad_norm": 0.09508214890956879, "global_step": 30432, "epoch": 845, "lr": 0.001} +{"train_loss": 0.02563272789120674, "train_loss_bc": 0.02563272789120674, "train_loss_llm": 0.0, "grad_norm": 0.03677281737327576, "global_step": 30433, "epoch": 845, "lr": 0.001} +{"train_loss": 0.0330316498875618, "train_loss_bc": 0.0330316498875618, "train_loss_llm": 0.0, "grad_norm": 0.062331054359674454, "global_step": 30434, "epoch": 845, "lr": 0.001} +{"train_loss": 0.031767819076776505, "train_loss_bc": 0.03173186630010605, "train_loss_llm": 0.0035951435565948486, "grad_norm": 0.09347573667764664, "global_step": 30435, "epoch": 845, "lr": 0.001} +{"train_loss": 0.028225306421518326, "train_loss_bc": 0.028225306421518326, "train_loss_llm": 0.0, "grad_norm": 0.10715164989233017, "global_step": 30436, "epoch": 845, "lr": 0.001} +{"train_loss": 0.026978813111782074, "train_loss_bc": 0.026978813111782074, "train_loss_llm": 0.0, "grad_norm": 0.02916567400097847, "global_step": 30437, "epoch": 845, "lr": 0.001} +{"train_loss": 0.02862042747437954, "train_loss_bc": 0.02862042747437954, "train_loss_llm": 0.0, "grad_norm": 0.05643385648727417, "global_step": 30438, "epoch": 845, "lr": 0.001} +{"train_loss": 0.032658591866493225, "train_loss_bc": 0.032658591866493225, "train_loss_llm": 0.0, "grad_norm": 0.08172879368066788, "global_step": 30439, "epoch": 845, "lr": 0.001} +{"train_loss": 0.027172449976205826, "train_loss_bc": 0.02716601826250553, "train_loss_llm": 0.0006431937217712402, "grad_norm": 0.10746119916439056, "global_step": 30440, "epoch": 845, "lr": 0.001} +{"train_loss": 0.03219212591648102, "train_loss_bc": 0.03219212591648102, "train_loss_llm": 0.0, "grad_norm": 0.03208983689546585, "global_step": 30441, "epoch": 845, "lr": 0.001} +{"train_loss": 0.026998370885849, "train_loss_bc": 0.026998370885849, "train_loss_llm": 0.0, "grad_norm": 0.04876154661178589, "global_step": 30442, "epoch": 845, "lr": 0.001} +{"train_loss": 0.027857063338160515, "train_loss_bc": 0.027857063338160515, "train_loss_llm": 0.0, "grad_norm": 0.07784001529216766, "global_step": 30443, "epoch": 845, "lr": 0.001} +{"train_loss": 0.025394562631845474, "train_loss_bc": 0.0253920778632164, "train_loss_llm": 0.0002484321594238281, "grad_norm": 0.08779594302177429, "global_step": 30444, "epoch": 845, "lr": 0.001} +{"train_loss": 0.028886348009109497, "train_loss_bc": 0.028886348009109497, "train_loss_llm": 0.0, "grad_norm": 0.03718428686261177, "global_step": 30445, "epoch": 845, "lr": 0.001} +{"train_loss": 0.024697985500097275, "train_loss_bc": 0.024682093411684036, "train_loss_llm": 0.0015891492366790771, "grad_norm": 0.054123733192682266, "global_step": 30446, "epoch": 845, "lr": 0.001} +{"train_loss": 0.03251686692237854, "train_loss_bc": 0.03249737247824669, "train_loss_llm": 0.0019494593143463135, "grad_norm": 0.07420095801353455, "global_step": 30447, "epoch": 845, "lr": 0.001} +{"train_loss": 0.03086928278207779, "train_loss_bc": 0.03086928278207779, "train_loss_llm": 0.0, "grad_norm": 0.09321697801351547, "global_step": 30448, "epoch": 845, "lr": 0.001} +{"train_loss": 0.02617991529405117, "train_loss_bc": 0.02617991529405117, "train_loss_llm": 0.0, "grad_norm": 0.0333087332546711, "global_step": 30449, "epoch": 845, "lr": 0.001} +{"train_loss": 0.031334973871707916, "train_loss_bc": 0.031334973871707916, "train_loss_llm": 0.0, "grad_norm": 0.055086635053157806, "global_step": 30450, "epoch": 845, "lr": 0.001} +{"train_loss": 0.031200218945741653, "train_loss_bc": 0.031191708520054817, "train_loss_llm": 0.000850975513458252, "grad_norm": 0.0800434947013855, "global_step": 30451, "epoch": 845, "lr": 0.001} +{"train_loss": 0.029133619740605354, "train_loss_bc": 0.029133619740605354, "train_loss_llm": 0.0, "grad_norm": 0.08988810330629349, "global_step": 30452, "epoch": 845, "lr": 0.001} +{"train_loss": 0.03204728290438652, "train_loss_bc": 0.03204728290438652, "train_loss_llm": 0.0, "grad_norm": 0.047485921531915665, "global_step": 30453, "epoch": 845, "lr": 0.001} +{"train_loss": 0.033256638795137405, "train_loss_bc": 0.033256638795137405, "train_loss_llm": 0.0, "grad_norm": 0.07020557671785355, "global_step": 30454, "epoch": 845, "lr": 0.001} +{"train_loss": 0.029885350209143426, "train_loss_bc": 0.029371898621320724, "train_loss_llm": 0.0, "grad_norm": 0.0842144712805748, "global_step": 30455, "epoch": 845, "lr": 0.001, "train/cumulative_reward": 23.542977954368922, "train/mean_score": 6.498292734219056, "train/success_rate": 0.2, "test/cumulative_reward": 35.429239567227214, "test/mean_score": 8.444888178702566, "test/success_rate": 0.2, "val_loss": 0.10027331113815308, "train_action_mse_error": 0.033701468259096146} +{"train_loss": 0.03373211249709129, "train_loss_bc": 0.03373211249709129, "train_loss_llm": 0.0, "grad_norm": 0.10252567380666733, "global_step": 30456, "epoch": 846, "lr": 0.001} +{"train_loss": 0.026336656883358955, "train_loss_bc": 0.026332251727581024, "train_loss_llm": 0.0004405677318572998, "grad_norm": 0.03589673340320587, "global_step": 30457, "epoch": 846, "lr": 0.001} +{"train_loss": 0.02567427232861519, "train_loss_bc": 0.02567427232861519, "train_loss_llm": 0.0, "grad_norm": 0.058717142790555954, "global_step": 30458, "epoch": 846, "lr": 0.001} +{"train_loss": 0.023724909871816635, "train_loss_bc": 0.023715494200587273, "train_loss_llm": 0.000941544771194458, "grad_norm": 0.07651816308498383, "global_step": 30459, "epoch": 846, "lr": 0.001} +{"train_loss": 0.03010326810181141, "train_loss_bc": 0.03010326810181141, "train_loss_llm": 0.0, "grad_norm": 0.09545168280601501, "global_step": 30460, "epoch": 846, "lr": 0.001} +{"train_loss": 0.02731342427432537, "train_loss_bc": 0.027297260239720345, "train_loss_llm": 0.0016164183616638184, "grad_norm": 0.029668070375919342, "global_step": 30461, "epoch": 846, "lr": 0.001} +{"train_loss": 0.025231802836060524, "train_loss_bc": 0.02522977441549301, "train_loss_llm": 0.0002028048038482666, "grad_norm": 0.0497761033475399, "global_step": 30462, "epoch": 846, "lr": 0.001} +{"train_loss": 0.031398706138134, "train_loss_bc": 0.031398706138134, "train_loss_llm": 0.0, "grad_norm": 0.05769163370132446, "global_step": 30463, "epoch": 846, "lr": 0.001} +{"train_loss": 0.030483262613415718, "train_loss_bc": 0.030483262613415718, "train_loss_llm": 0.0, "grad_norm": 0.07936406135559082, "global_step": 30464, "epoch": 846, "lr": 0.001} +{"train_loss": 0.033222269266843796, "train_loss_bc": 0.033222269266843796, "train_loss_llm": 0.0, "grad_norm": 0.04757724702358246, "global_step": 30465, "epoch": 846, "lr": 0.001} +{"train_loss": 0.02662620320916176, "train_loss_bc": 0.02662620320916176, "train_loss_llm": 0.0, "grad_norm": 0.07055051624774933, "global_step": 30466, "epoch": 846, "lr": 0.001} +{"train_loss": 0.03331955149769783, "train_loss_bc": 0.03331955149769783, "train_loss_llm": 0.0, "grad_norm": 0.09493710845708847, "global_step": 30467, "epoch": 846, "lr": 0.001} +{"train_loss": 0.028578445315361023, "train_loss_bc": 0.028578445315361023, "train_loss_llm": 0.0, "grad_norm": 0.11152651906013489, "global_step": 30468, "epoch": 846, "lr": 0.001} +{"train_loss": 0.02574295550584793, "train_loss_bc": 0.02574295550584793, "train_loss_llm": 0.0, "grad_norm": 0.040001679211854935, "global_step": 30469, "epoch": 846, "lr": 0.001} +{"train_loss": 0.035361696034669876, "train_loss_bc": 0.035361696034669876, "train_loss_llm": 0.0, "grad_norm": 0.0710412934422493, "global_step": 30470, "epoch": 846, "lr": 0.001} +{"train_loss": 0.03288068249821663, "train_loss_bc": 0.03288068249821663, "train_loss_llm": 0.0, "grad_norm": 0.09235517680644989, "global_step": 30471, "epoch": 846, "lr": 0.001} +{"train_loss": 0.03054969385266304, "train_loss_bc": 0.03054969385266304, "train_loss_llm": 0.0, "grad_norm": 0.11711320281028748, "global_step": 30472, "epoch": 846, "lr": 0.001} +{"train_loss": 0.028739457949995995, "train_loss_bc": 0.02871612459421158, "train_loss_llm": 0.002333313226699829, "grad_norm": 0.03563779592514038, "global_step": 30473, "epoch": 846, "lr": 0.001} +{"train_loss": 0.02887582778930664, "train_loss_bc": 0.02887582778930664, "train_loss_llm": 0.0, "grad_norm": 0.053103309124708176, "global_step": 30474, "epoch": 846, "lr": 0.001} +{"train_loss": 0.03185238689184189, "train_loss_bc": 0.03184924274682999, "train_loss_llm": 0.0003142356872558594, "grad_norm": 0.06849205493927002, "global_step": 30475, "epoch": 846, "lr": 0.001} +{"train_loss": 0.03106580302119255, "train_loss_bc": 0.03106580302119255, "train_loss_llm": 0.0, "grad_norm": 0.0943543091416359, "global_step": 30476, "epoch": 846, "lr": 0.001} +{"train_loss": 0.03002993017435074, "train_loss_bc": 0.03002993017435074, "train_loss_llm": 0.0, "grad_norm": 0.03338736295700073, "global_step": 30477, "epoch": 846, "lr": 0.001} +{"train_loss": 0.030611883848905563, "train_loss_bc": 0.03060438111424446, "train_loss_llm": 0.0007502138614654541, "grad_norm": 0.05324969068169594, "global_step": 30478, "epoch": 846, "lr": 0.001} +{"train_loss": 0.032149650156497955, "train_loss_bc": 0.03212859481573105, "train_loss_llm": 0.0021053552627563477, "grad_norm": 0.07666181027889252, "global_step": 30479, "epoch": 846, "lr": 0.001} +{"train_loss": 0.028988730162382126, "train_loss_bc": 0.028987901285290718, "train_loss_llm": 8.285045623779297e-05, "grad_norm": 0.08305681496858597, "global_step": 30480, "epoch": 846, "lr": 0.001} +{"train_loss": 0.029595380648970604, "train_loss_bc": 0.029589110985398293, "train_loss_llm": 0.0006270110607147217, "grad_norm": 0.032214485108852386, "global_step": 30481, "epoch": 846, "lr": 0.001} +{"train_loss": 0.030641552060842514, "train_loss_bc": 0.030641552060842514, "train_loss_llm": 0.0, "grad_norm": 0.0526755154132843, "global_step": 30482, "epoch": 846, "lr": 0.001} +{"train_loss": 0.03083055652678013, "train_loss_bc": 0.03083055652678013, "train_loss_llm": 0.0, "grad_norm": 0.0752207413315773, "global_step": 30483, "epoch": 846, "lr": 0.001} +{"train_loss": 0.03779354691505432, "train_loss_bc": 0.03779354691505432, "train_loss_llm": 0.0, "grad_norm": 0.11115247011184692, "global_step": 30484, "epoch": 846, "lr": 0.001} +{"train_loss": 0.032356854528188705, "train_loss_bc": 0.03235256299376488, "train_loss_llm": 0.000429004430770874, "grad_norm": 0.037310149520635605, "global_step": 30485, "epoch": 846, "lr": 0.001} +{"train_loss": 0.02538914792239666, "train_loss_bc": 0.025382792577147484, "train_loss_llm": 0.0006355345249176025, "grad_norm": 0.0685715451836586, "global_step": 30486, "epoch": 846, "lr": 0.001} +{"train_loss": 0.029342137277126312, "train_loss_bc": 0.02933862805366516, "train_loss_llm": 0.0003508925437927246, "grad_norm": 0.08441722393035889, "global_step": 30487, "epoch": 846, "lr": 0.001} +{"train_loss": 0.026930131018161774, "train_loss_bc": 0.026930131018161774, "train_loss_llm": 0.0, "grad_norm": 0.10153834521770477, "global_step": 30488, "epoch": 846, "lr": 0.001} +{"train_loss": 0.030302461236715317, "train_loss_bc": 0.030302461236715317, "train_loss_llm": 0.0, "grad_norm": 0.030435653403401375, "global_step": 30489, "epoch": 846, "lr": 0.001} +{"train_loss": 0.03142423555254936, "train_loss_bc": 0.03142423555254936, "train_loss_llm": 0.0, "grad_norm": 0.0549049936234951, "global_step": 30490, "epoch": 846, "lr": 0.001} +{"train_loss": 0.02991078830220633, "train_loss_bc": 0.029588792473077774, "train_loss_llm": 0.0, "grad_norm": 0.08652132749557495, "global_step": 30491, "epoch": 846, "lr": 0.001, "val_loss": 0.10060249269008636} +{"train_loss": 0.029167359694838524, "train_loss_bc": 0.029167359694838524, "train_loss_llm": 0.0, "grad_norm": 0.10804799199104309, "global_step": 30492, "epoch": 847, "lr": 0.001} +{"train_loss": 0.0289717186242342, "train_loss_bc": 0.0289717186242342, "train_loss_llm": 0.0, "grad_norm": 0.03406643122434616, "global_step": 30493, "epoch": 847, "lr": 0.001} +{"train_loss": 0.032467205077409744, "train_loss_bc": 0.032467205077409744, "train_loss_llm": 0.0, "grad_norm": 0.05195589363574982, "global_step": 30494, "epoch": 847, "lr": 0.001} +{"train_loss": 0.02951103262603283, "train_loss_bc": 0.02951103262603283, "train_loss_llm": 0.0, "grad_norm": 0.08324958384037018, "global_step": 30495, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03203897178173065, "train_loss_bc": 0.03203897178173065, "train_loss_llm": 0.0, "grad_norm": 0.10443884879350662, "global_step": 30496, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03133850917220116, "train_loss_bc": 0.03133850917220116, "train_loss_llm": 0.0, "grad_norm": 0.04351123422384262, "global_step": 30497, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03102215752005577, "train_loss_bc": 0.03102215752005577, "train_loss_llm": 0.0, "grad_norm": 0.064701147377491, "global_step": 30498, "epoch": 847, "lr": 0.001} +{"train_loss": 0.028983578085899353, "train_loss_bc": 0.028983578085899353, "train_loss_llm": 0.0, "grad_norm": 0.09162905067205429, "global_step": 30499, "epoch": 847, "lr": 0.001} +{"train_loss": 0.02622329816222191, "train_loss_bc": 0.02622329816222191, "train_loss_llm": 0.0, "grad_norm": 0.11352495104074478, "global_step": 30500, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03019176423549652, "train_loss_bc": 0.03019176423549652, "train_loss_llm": 0.0, "grad_norm": 0.0399000458419323, "global_step": 30501, "epoch": 847, "lr": 0.001} +{"train_loss": 0.02520373836159706, "train_loss_bc": 0.02520373836159706, "train_loss_llm": 0.0, "grad_norm": 0.05967578664422035, "global_step": 30502, "epoch": 847, "lr": 0.001} +{"train_loss": 0.02326352708041668, "train_loss_bc": 0.02324487455189228, "train_loss_llm": 0.0018653273582458496, "grad_norm": 0.0776129886507988, "global_step": 30503, "epoch": 847, "lr": 0.001} +{"train_loss": 0.030219923704862595, "train_loss_bc": 0.030219923704862595, "train_loss_llm": 0.0, "grad_norm": 0.09316575527191162, "global_step": 30504, "epoch": 847, "lr": 0.001} +{"train_loss": 0.030191944912075996, "train_loss_bc": 0.030190056189894676, "train_loss_llm": 0.0001888275146484375, "grad_norm": 0.032253969460725784, "global_step": 30505, "epoch": 847, "lr": 0.001} +{"train_loss": 0.0318678617477417, "train_loss_bc": 0.0318678617477417, "train_loss_llm": 0.0, "grad_norm": 0.052100010216236115, "global_step": 30506, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03296732157468796, "train_loss_bc": 0.03296732157468796, "train_loss_llm": 0.0, "grad_norm": 0.07566222548484802, "global_step": 30507, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03011261485517025, "train_loss_bc": 0.030103476718068123, "train_loss_llm": 0.0009138882160186768, "grad_norm": 0.101005919277668, "global_step": 30508, "epoch": 847, "lr": 0.001} +{"train_loss": 0.029954848811030388, "train_loss_bc": 0.02995232492685318, "train_loss_llm": 0.0002524256706237793, "grad_norm": 0.03463439270853996, "global_step": 30509, "epoch": 847, "lr": 0.001} +{"train_loss": 0.032207585871219635, "train_loss_bc": 0.032207585871219635, "train_loss_llm": 0.0, "grad_norm": 0.060349926352500916, "global_step": 30510, "epoch": 847, "lr": 0.001} +{"train_loss": 0.0291307270526886, "train_loss_bc": 0.029125170782208443, "train_loss_llm": 0.0005556344985961914, "grad_norm": 0.07474259287118912, "global_step": 30511, "epoch": 847, "lr": 0.001} +{"train_loss": 0.030434880405664444, "train_loss_bc": 0.030428167432546616, "train_loss_llm": 0.0006712675094604492, "grad_norm": 0.0928896814584732, "global_step": 30512, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03426132723689079, "train_loss_bc": 0.03425430878996849, "train_loss_llm": 0.000701904296875, "grad_norm": 0.036156222224235535, "global_step": 30513, "epoch": 847, "lr": 0.001} +{"train_loss": 0.026208700612187386, "train_loss_bc": 0.026208700612187386, "train_loss_llm": 0.0, "grad_norm": 0.04385353624820709, "global_step": 30514, "epoch": 847, "lr": 0.001} +{"train_loss": 0.033826641738414764, "train_loss_bc": 0.033826641738414764, "train_loss_llm": 0.0, "grad_norm": 0.06745555996894836, "global_step": 30515, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03561583161354065, "train_loss_bc": 0.03561412915587425, "train_loss_llm": 0.00017029047012329102, "grad_norm": 0.09302187711000443, "global_step": 30516, "epoch": 847, "lr": 0.001} +{"train_loss": 0.028165234252810478, "train_loss_bc": 0.028165234252810478, "train_loss_llm": 0.0, "grad_norm": 0.04257313534617424, "global_step": 30517, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03268590569496155, "train_loss_bc": 0.03266056999564171, "train_loss_llm": 0.0025336146354675293, "grad_norm": 0.0641564428806305, "global_step": 30518, "epoch": 847, "lr": 0.001} +{"train_loss": 0.021948738023638725, "train_loss_bc": 0.02193594165146351, "train_loss_llm": 0.0012796223163604736, "grad_norm": 0.08139481395483017, "global_step": 30519, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03162721171975136, "train_loss_bc": 0.03162721171975136, "train_loss_llm": 0.0, "grad_norm": 0.11608511954545975, "global_step": 30520, "epoch": 847, "lr": 0.001} +{"train_loss": 0.028809605166316032, "train_loss_bc": 0.028809605166316032, "train_loss_llm": 0.0, "grad_norm": 0.03412889689207077, "global_step": 30521, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03455648943781853, "train_loss_bc": 0.03455648943781853, "train_loss_llm": 0.0, "grad_norm": 0.06178610026836395, "global_step": 30522, "epoch": 847, "lr": 0.001} +{"train_loss": 0.03141936659812927, "train_loss_bc": 0.03141936659812927, "train_loss_llm": 0.0, "grad_norm": 0.0794677734375, "global_step": 30523, "epoch": 847, "lr": 0.001} +{"train_loss": 0.02903115563094616, "train_loss_bc": 0.02903115563094616, "train_loss_llm": 0.0, "grad_norm": 0.10047813504934311, "global_step": 30524, "epoch": 847, "lr": 0.001} +{"train_loss": 0.031446464359760284, "train_loss_bc": 0.031446464359760284, "train_loss_llm": 0.0, "grad_norm": 0.054679326713085175, "global_step": 30525, "epoch": 847, "lr": 0.001} +{"train_loss": 0.024876972660422325, "train_loss_bc": 0.024876972660422325, "train_loss_llm": 0.0, "grad_norm": 0.06522473692893982, "global_step": 30526, "epoch": 847, "lr": 0.001} +{"train_loss": 0.029836938354290195, "train_loss_bc": 0.02417699806392193, "train_loss_llm": 0.0002568662166595459, "grad_norm": 0.07585135847330093, "global_step": 30527, "epoch": 847, "lr": 0.001, "val_loss": 0.0992780327796936} +{"train_loss": 0.028661029413342476, "train_loss_bc": 0.028661029413342476, "train_loss_llm": 0.0, "grad_norm": 0.0945519506931305, "global_step": 30528, "epoch": 848, "lr": 0.001} +{"train_loss": 0.028203550726175308, "train_loss_bc": 0.028203550726175308, "train_loss_llm": 0.0, "grad_norm": 0.038522131741046906, "global_step": 30529, "epoch": 848, "lr": 0.001} +{"train_loss": 0.029319187626242638, "train_loss_bc": 0.029319187626242638, "train_loss_llm": 0.0, "grad_norm": 0.06034976243972778, "global_step": 30530, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03011692687869072, "train_loss_bc": 0.0301092267036438, "train_loss_llm": 0.0007700026035308838, "grad_norm": 0.08551371842622757, "global_step": 30531, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03134237602353096, "train_loss_bc": 0.03134237602353096, "train_loss_llm": 0.0, "grad_norm": 0.11156515032052994, "global_step": 30532, "epoch": 848, "lr": 0.001} +{"train_loss": 0.026498958468437195, "train_loss_bc": 0.026498958468437195, "train_loss_llm": 0.0, "grad_norm": 0.034816451370716095, "global_step": 30533, "epoch": 848, "lr": 0.001} +{"train_loss": 0.02950415387749672, "train_loss_bc": 0.02950415387749672, "train_loss_llm": 0.0, "grad_norm": 0.06374803930521011, "global_step": 30534, "epoch": 848, "lr": 0.001} +{"train_loss": 0.02864525094628334, "train_loss_bc": 0.02864525094628334, "train_loss_llm": 0.0, "grad_norm": 0.08362022787332535, "global_step": 30535, "epoch": 848, "lr": 0.001} +{"train_loss": 0.029697880148887634, "train_loss_bc": 0.029693640768527985, "train_loss_llm": 0.0004239380359649658, "grad_norm": 0.10458098351955414, "global_step": 30536, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03034478984773159, "train_loss_bc": 0.03034478984773159, "train_loss_llm": 0.0, "grad_norm": 0.04292698949575424, "global_step": 30537, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03363334387540817, "train_loss_bc": 0.03363334387540817, "train_loss_llm": 0.0, "grad_norm": 0.07113895565271378, "global_step": 30538, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03090984933078289, "train_loss_bc": 0.030905280262231827, "train_loss_llm": 0.0004569590091705322, "grad_norm": 0.09411368519067764, "global_step": 30539, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03057159297168255, "train_loss_bc": 0.03057159297168255, "train_loss_llm": 0.0, "grad_norm": 0.10955818742513657, "global_step": 30540, "epoch": 848, "lr": 0.001} +{"train_loss": 0.024565601721405983, "train_loss_bc": 0.024565601721405983, "train_loss_llm": 0.0, "grad_norm": 0.033357731997966766, "global_step": 30541, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03280932828783989, "train_loss_bc": 0.03280932828783989, "train_loss_llm": 0.0, "grad_norm": 0.0527692511677742, "global_step": 30542, "epoch": 848, "lr": 0.001} +{"train_loss": 0.024972476065158844, "train_loss_bc": 0.024964304640889168, "train_loss_llm": 0.0008170604705810547, "grad_norm": 0.07314798980951309, "global_step": 30543, "epoch": 848, "lr": 0.001} +{"train_loss": 0.029168276116251945, "train_loss_bc": 0.029168276116251945, "train_loss_llm": 0.0, "grad_norm": 0.08551359176635742, "global_step": 30544, "epoch": 848, "lr": 0.001} +{"train_loss": 0.0370611846446991, "train_loss_bc": 0.0370611846446991, "train_loss_llm": 0.0, "grad_norm": 0.04067596048116684, "global_step": 30545, "epoch": 848, "lr": 0.001} +{"train_loss": 0.030812138691544533, "train_loss_bc": 0.030812138691544533, "train_loss_llm": 0.0, "grad_norm": 0.06466174125671387, "global_step": 30546, "epoch": 848, "lr": 0.001} +{"train_loss": 0.02704066038131714, "train_loss_bc": 0.02704066038131714, "train_loss_llm": 0.0, "grad_norm": 0.07968243956565857, "global_step": 30547, "epoch": 848, "lr": 0.001} +{"train_loss": 0.02886764146387577, "train_loss_bc": 0.028863120824098587, "train_loss_llm": 0.0004519820213317871, "grad_norm": 0.09736664593219757, "global_step": 30548, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03050348348915577, "train_loss_bc": 0.03050348348915577, "train_loss_llm": 0.0, "grad_norm": 0.03725126013159752, "global_step": 30549, "epoch": 848, "lr": 0.001} +{"train_loss": 0.027152221649885178, "train_loss_bc": 0.027152221649885178, "train_loss_llm": 0.0, "grad_norm": 0.05557069182395935, "global_step": 30550, "epoch": 848, "lr": 0.001} +{"train_loss": 0.030915547162294388, "train_loss_bc": 0.030915547162294388, "train_loss_llm": 0.0, "grad_norm": 0.06462861597537994, "global_step": 30551, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03183119371533394, "train_loss_bc": 0.03183119371533394, "train_loss_llm": 0.0, "grad_norm": 0.08784462511539459, "global_step": 30552, "epoch": 848, "lr": 0.001} +{"train_loss": 0.030019206926226616, "train_loss_bc": 0.030014170333743095, "train_loss_llm": 0.0005035996437072754, "grad_norm": 0.03632756695151329, "global_step": 30553, "epoch": 848, "lr": 0.001} +{"train_loss": 0.029507873579859734, "train_loss_bc": 0.029507873579859734, "train_loss_llm": 0.0, "grad_norm": 0.06333214044570923, "global_step": 30554, "epoch": 848, "lr": 0.001} +{"train_loss": 0.028629500418901443, "train_loss_bc": 0.028612753376364708, "train_loss_llm": 0.0016747117042541504, "grad_norm": 0.08029862493276596, "global_step": 30555, "epoch": 848, "lr": 0.001} +{"train_loss": 0.0341646745800972, "train_loss_bc": 0.034161314368247986, "train_loss_llm": 0.00033605098724365234, "grad_norm": 0.1158781349658966, "global_step": 30556, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03125327080488205, "train_loss_bc": 0.03125327080488205, "train_loss_llm": 0.0, "grad_norm": 0.03190279379487038, "global_step": 30557, "epoch": 848, "lr": 0.001} +{"train_loss": 0.02973286435008049, "train_loss_bc": 0.02973286435008049, "train_loss_llm": 0.0, "grad_norm": 0.04711049422621727, "global_step": 30558, "epoch": 848, "lr": 0.001} +{"train_loss": 0.03153602406382561, "train_loss_bc": 0.03153602406382561, "train_loss_llm": 0.0, "grad_norm": 0.07024102658033371, "global_step": 30559, "epoch": 848, "lr": 0.001} +{"train_loss": 0.027899056673049927, "train_loss_bc": 0.027899056673049927, "train_loss_llm": 0.0, "grad_norm": 0.09418820589780807, "global_step": 30560, "epoch": 848, "lr": 0.001} +{"train_loss": 0.031829144805669785, "train_loss_bc": 0.03181717172265053, "train_loss_llm": 0.0011974871158599854, "grad_norm": 0.03627840057015419, "global_step": 30561, "epoch": 848, "lr": 0.001} +{"train_loss": 0.02832816168665886, "train_loss_bc": 0.028326142579317093, "train_loss_llm": 0.00020197033882141113, "grad_norm": 0.07078688591718674, "global_step": 30562, "epoch": 848, "lr": 0.001} +{"train_loss": 0.02985625569191244, "train_loss_bc": 0.028760595247149467, "train_loss_llm": 0.001618891954421997, "grad_norm": 0.09761454910039902, "global_step": 30563, "epoch": 848, "lr": 0.001, "val_loss": 0.10122083872556686} +{"train_loss": 0.02561456337571144, "train_loss_bc": 0.025613678619265556, "train_loss_llm": 8.851289749145508e-05, "grad_norm": 0.1062491238117218, "global_step": 30564, "epoch": 849, "lr": 0.001} +{"train_loss": 0.029224827885627747, "train_loss_bc": 0.02919766679406166, "train_loss_llm": 0.0027160346508026123, "grad_norm": 0.03653937205672264, "global_step": 30565, "epoch": 849, "lr": 0.001} +{"train_loss": 0.02951282076537609, "train_loss_bc": 0.02951282076537609, "train_loss_llm": 0.0, "grad_norm": 0.05999455973505974, "global_step": 30566, "epoch": 849, "lr": 0.001} +{"train_loss": 0.0226645115762949, "train_loss_bc": 0.0226645115762949, "train_loss_llm": 0.0, "grad_norm": 0.08234643191099167, "global_step": 30567, "epoch": 849, "lr": 0.001} +{"train_loss": 0.030554408207535744, "train_loss_bc": 0.030554408207535744, "train_loss_llm": 0.0, "grad_norm": 0.10453024506568909, "global_step": 30568, "epoch": 849, "lr": 0.001} +{"train_loss": 0.029245823621749878, "train_loss_bc": 0.029245823621749878, "train_loss_llm": 0.0, "grad_norm": 0.032793860882520676, "global_step": 30569, "epoch": 849, "lr": 0.001} +{"train_loss": 0.026517676189541817, "train_loss_bc": 0.02651633508503437, "train_loss_llm": 0.0001341402530670166, "grad_norm": 0.045270126312971115, "global_step": 30570, "epoch": 849, "lr": 0.001} +{"train_loss": 0.029880018904805183, "train_loss_bc": 0.029869796708226204, "train_loss_llm": 0.0010223090648651123, "grad_norm": 0.07023058086633682, "global_step": 30571, "epoch": 849, "lr": 0.001} +{"train_loss": 0.0306922048330307, "train_loss_bc": 0.0306922048330307, "train_loss_llm": 0.0, "grad_norm": 0.09470625221729279, "global_step": 30572, "epoch": 849, "lr": 0.001} +{"train_loss": 0.029390383511781693, "train_loss_bc": 0.029385413974523544, "train_loss_llm": 0.0004970431327819824, "grad_norm": 0.04345836117863655, "global_step": 30573, "epoch": 849, "lr": 0.001} +{"train_loss": 0.029388422146439552, "train_loss_bc": 0.029388422146439552, "train_loss_llm": 0.0, "grad_norm": 0.06523310393095016, "global_step": 30574, "epoch": 849, "lr": 0.001} +{"train_loss": 0.028845854103565216, "train_loss_bc": 0.028845854103565216, "train_loss_llm": 0.0, "grad_norm": 0.08600851893424988, "global_step": 30575, "epoch": 849, "lr": 0.001} +{"train_loss": 0.02893488109111786, "train_loss_bc": 0.028920680284500122, "train_loss_llm": 0.001420140266418457, "grad_norm": 0.11357749998569489, "global_step": 30576, "epoch": 849, "lr": 0.001} +{"train_loss": 0.035735033452510834, "train_loss_bc": 0.035735033452510834, "train_loss_llm": 0.0, "grad_norm": 0.045321352779865265, "global_step": 30577, "epoch": 849, "lr": 0.001} +{"train_loss": 0.026401899755001068, "train_loss_bc": 0.026395458728075027, "train_loss_llm": 0.0006440281867980957, "grad_norm": 0.0724787786602974, "global_step": 30578, "epoch": 849, "lr": 0.001} +{"train_loss": 0.03007606416940689, "train_loss_bc": 0.03007606416940689, "train_loss_llm": 0.0, "grad_norm": 0.09888830035924911, "global_step": 30579, "epoch": 849, "lr": 0.001} +{"train_loss": 0.034265145659446716, "train_loss_bc": 0.034265145659446716, "train_loss_llm": 0.0, "grad_norm": 0.12392475455999374, "global_step": 30580, "epoch": 849, "lr": 0.001} +{"train_loss": 0.02949199639260769, "train_loss_bc": 0.029486795887351036, "train_loss_llm": 0.0005200207233428955, "grad_norm": 0.039479609578847885, "global_step": 30581, "epoch": 849, "lr": 0.001} +{"train_loss": 0.028050236403942108, "train_loss_bc": 0.028046561405062675, "train_loss_llm": 0.0003674924373626709, "grad_norm": 0.049919143319129944, "global_step": 30582, "epoch": 849, "lr": 0.001} +{"train_loss": 0.03222232684493065, "train_loss_bc": 0.03222232684493065, "train_loss_llm": 0.0, "grad_norm": 0.06411543488502502, "global_step": 30583, "epoch": 849, "lr": 0.001} +{"train_loss": 0.032437290996313095, "train_loss_bc": 0.032437290996313095, "train_loss_llm": 0.0, "grad_norm": 0.08827722072601318, "global_step": 30584, "epoch": 849, "lr": 0.001} +{"train_loss": 0.03381047397851944, "train_loss_bc": 0.03380953520536423, "train_loss_llm": 9.384751319885254e-05, "grad_norm": 0.03420284762978554, "global_step": 30585, "epoch": 849, "lr": 0.001} +{"train_loss": 0.03501080349087715, "train_loss_bc": 0.0350053571164608, "train_loss_llm": 0.0005447864532470703, "grad_norm": 0.06460878252983093, "global_step": 30586, "epoch": 849, "lr": 0.001} +{"train_loss": 0.034577660262584686, "train_loss_bc": 0.03457207232713699, "train_loss_llm": 0.0005587339401245117, "grad_norm": 0.08009964972734451, "global_step": 30587, "epoch": 849, "lr": 0.001} +{"train_loss": 0.028308363631367683, "train_loss_bc": 0.028308363631367683, "train_loss_llm": 0.0, "grad_norm": 0.09574097394943237, "global_step": 30588, "epoch": 849, "lr": 0.001} +{"train_loss": 0.029327919706702232, "train_loss_bc": 0.02932352013885975, "train_loss_llm": 0.0004400014877319336, "grad_norm": 0.03371983394026756, "global_step": 30589, "epoch": 849, "lr": 0.001} +{"train_loss": 0.03349369764328003, "train_loss_bc": 0.03347734734416008, "train_loss_llm": 0.0016349554061889648, "grad_norm": 0.0626492127776146, "global_step": 30590, "epoch": 849, "lr": 0.001} +{"train_loss": 0.026139885187149048, "train_loss_bc": 0.026139885187149048, "train_loss_llm": 0.0, "grad_norm": 0.07792673259973526, "global_step": 30591, "epoch": 849, "lr": 0.001} +{"train_loss": 0.02793698199093342, "train_loss_bc": 0.02793698199093342, "train_loss_llm": 0.0, "grad_norm": 0.09788757562637329, "global_step": 30592, "epoch": 849, "lr": 0.001} +{"train_loss": 0.02804737165570259, "train_loss_bc": 0.02804737165570259, "train_loss_llm": 0.0, "grad_norm": 0.03991040214896202, "global_step": 30593, "epoch": 849, "lr": 0.001} +{"train_loss": 0.03272823989391327, "train_loss_bc": 0.03272823989391327, "train_loss_llm": 0.0, "grad_norm": 0.05677714943885803, "global_step": 30594, "epoch": 849, "lr": 0.001} +{"train_loss": 0.028891291469335556, "train_loss_bc": 0.028891291469335556, "train_loss_llm": 0.0, "grad_norm": 0.08822213858366013, "global_step": 30595, "epoch": 849, "lr": 0.001} +{"train_loss": 0.03270737826824188, "train_loss_bc": 0.03270737826824188, "train_loss_llm": 0.0, "grad_norm": 0.10962560772895813, "global_step": 30596, "epoch": 849, "lr": 0.001} +{"train_loss": 0.028386810794472694, "train_loss_bc": 0.028381068259477615, "train_loss_llm": 0.0005742907524108887, "grad_norm": 0.04021432250738144, "global_step": 30597, "epoch": 849, "lr": 0.001} +{"train_loss": 0.026279525831341743, "train_loss_bc": 0.02627306990325451, "train_loss_llm": 0.0006456673145294189, "grad_norm": 0.05787177011370659, "global_step": 30598, "epoch": 849, "lr": 0.001} +{"train_loss": 0.029763894052141242, "train_loss_bc": 0.026707392185926437, "train_loss_llm": 0.0, "grad_norm": 0.08318531513214111, "global_step": 30599, "epoch": 849, "lr": 0.001, "val_loss": 0.09940697997808456} +{"train_loss": 0.02939264103770256, "train_loss_bc": 0.02939264103770256, "train_loss_llm": 0.0, "grad_norm": 0.09439931809902191, "global_step": 30600, "epoch": 850, "lr": 0.001} +{"train_loss": 0.0287818294018507, "train_loss_bc": 0.0287818294018507, "train_loss_llm": 0.0, "grad_norm": 0.036532990634441376, "global_step": 30601, "epoch": 850, "lr": 0.001} +{"train_loss": 0.035068631172180176, "train_loss_bc": 0.035068631172180176, "train_loss_llm": 0.0, "grad_norm": 0.06810933351516724, "global_step": 30602, "epoch": 850, "lr": 0.001} +{"train_loss": 0.027005529031157494, "train_loss_bc": 0.02700052410364151, "train_loss_llm": 0.000500410795211792, "grad_norm": 0.09177900105714798, "global_step": 30603, "epoch": 850, "lr": 0.001} +{"train_loss": 0.03269578143954277, "train_loss_bc": 0.03269578143954277, "train_loss_llm": 0.0, "grad_norm": 0.1095929965376854, "global_step": 30604, "epoch": 850, "lr": 0.001} +{"train_loss": 0.03388359025120735, "train_loss_bc": 0.03388359025120735, "train_loss_llm": 0.0, "grad_norm": 0.04274296760559082, "global_step": 30605, "epoch": 850, "lr": 0.001} +{"train_loss": 0.029613345861434937, "train_loss_bc": 0.029613345861434937, "train_loss_llm": 0.0, "grad_norm": 0.05338168516755104, "global_step": 30606, "epoch": 850, "lr": 0.001} +{"train_loss": 0.022639075294137, "train_loss_bc": 0.022639075294137, "train_loss_llm": 0.0, "grad_norm": 0.06618321686983109, "global_step": 30607, "epoch": 850, "lr": 0.001} +{"train_loss": 0.028136875480413437, "train_loss_bc": 0.028136875480413437, "train_loss_llm": 0.0, "grad_norm": 0.07509774714708328, "global_step": 30608, "epoch": 850, "lr": 0.001} +{"train_loss": 0.03341877460479736, "train_loss_bc": 0.03341877460479736, "train_loss_llm": 0.0, "grad_norm": 0.04798248037695885, "global_step": 30609, "epoch": 850, "lr": 0.001} +{"train_loss": 0.02854606695473194, "train_loss_bc": 0.02854606695473194, "train_loss_llm": 0.0, "grad_norm": 0.07016099989414215, "global_step": 30610, "epoch": 850, "lr": 0.001} +{"train_loss": 0.02810855023562908, "train_loss_bc": 0.02810855023562908, "train_loss_llm": 0.0, "grad_norm": 0.08785472065210342, "global_step": 30611, "epoch": 850, "lr": 0.001} +{"train_loss": 0.025497760623693466, "train_loss_bc": 0.025497760623693466, "train_loss_llm": 0.0, "grad_norm": 0.11277419328689575, "global_step": 30612, "epoch": 850, "lr": 0.001} +{"train_loss": 0.02864074893295765, "train_loss_bc": 0.02864074893295765, "train_loss_llm": 0.0, "grad_norm": 0.03895087540149689, "global_step": 30613, "epoch": 850, "lr": 0.001} +{"train_loss": 0.029935766011476517, "train_loss_bc": 0.029935766011476517, "train_loss_llm": 0.0, "grad_norm": 0.0791812539100647, "global_step": 30614, "epoch": 850, "lr": 0.001} +{"train_loss": 0.031330861151218414, "train_loss_bc": 0.031330861151218414, "train_loss_llm": 0.0, "grad_norm": 0.10039286315441132, "global_step": 30615, "epoch": 850, "lr": 0.001} +{"train_loss": 0.0328470878303051, "train_loss_bc": 0.032836638391017914, "train_loss_llm": 0.0010450482368469238, "grad_norm": 0.13172562420368195, "global_step": 30616, "epoch": 850, "lr": 0.001} +{"train_loss": 0.02976279892027378, "train_loss_bc": 0.02976279892027378, "train_loss_llm": 0.0, "grad_norm": 0.04074231907725334, "global_step": 30617, "epoch": 850, "lr": 0.001} +{"train_loss": 0.02583024837076664, "train_loss_bc": 0.02583024837076664, "train_loss_llm": 0.0, "grad_norm": 0.06238902732729912, "global_step": 30618, "epoch": 850, "lr": 0.001} +{"train_loss": 0.028794819489121437, "train_loss_bc": 0.028794819489121437, "train_loss_llm": 0.0, "grad_norm": 0.08269500732421875, "global_step": 30619, "epoch": 850, "lr": 0.001} +{"train_loss": 0.030100839212536812, "train_loss_bc": 0.030100839212536812, "train_loss_llm": 0.0, "grad_norm": 0.10791627317667007, "global_step": 30620, "epoch": 850, "lr": 0.001} +{"train_loss": 0.022934556007385254, "train_loss_bc": 0.022934556007385254, "train_loss_llm": 0.0, "grad_norm": 0.03530479222536087, "global_step": 30621, "epoch": 850, "lr": 0.001} +{"train_loss": 0.030836213380098343, "train_loss_bc": 0.03083006665110588, "train_loss_llm": 0.0006146132946014404, "grad_norm": 0.061201900243759155, "global_step": 30622, "epoch": 850, "lr": 0.001} +{"train_loss": 0.03164229542016983, "train_loss_bc": 0.03164229542016983, "train_loss_llm": 0.0, "grad_norm": 0.09443259239196777, "global_step": 30623, "epoch": 850, "lr": 0.001} +{"train_loss": 0.02802196890115738, "train_loss_bc": 0.02802196890115738, "train_loss_llm": 0.0, "grad_norm": 0.11596981436014175, "global_step": 30624, "epoch": 850, "lr": 0.001} +{"train_loss": 0.03878999128937721, "train_loss_bc": 0.03878999128937721, "train_loss_llm": 0.0, "grad_norm": 0.047190532088279724, "global_step": 30625, "epoch": 850, "lr": 0.001} +{"train_loss": 0.029329238459467888, "train_loss_bc": 0.029329238459467888, "train_loss_llm": 0.0, "grad_norm": 0.07803085446357727, "global_step": 30626, "epoch": 850, "lr": 0.001} +{"train_loss": 0.022577011957764626, "train_loss_bc": 0.022577011957764626, "train_loss_llm": 0.0, "grad_norm": 0.10485847294330597, "global_step": 30627, "epoch": 850, "lr": 0.001} +{"train_loss": 0.03036792390048504, "train_loss_bc": 0.03036152757704258, "train_loss_llm": 0.0006395578384399414, "grad_norm": 0.13608330488204956, "global_step": 30628, "epoch": 850, "lr": 0.001} +{"train_loss": 0.02843148447573185, "train_loss_bc": 0.02842854894697666, "train_loss_llm": 0.0002935826778411865, "grad_norm": 0.03238983452320099, "global_step": 30629, "epoch": 850, "lr": 0.001} +{"train_loss": 0.031040621921420097, "train_loss_bc": 0.031040621921420097, "train_loss_llm": 0.0, "grad_norm": 0.0597073957324028, "global_step": 30630, "epoch": 850, "lr": 0.001} +{"train_loss": 0.03203562647104263, "train_loss_bc": 0.03203562647104263, "train_loss_llm": 0.0, "grad_norm": 0.09305160492658615, "global_step": 30631, "epoch": 850, "lr": 0.001} +{"train_loss": 0.03549816831946373, "train_loss_bc": 0.03549816831946373, "train_loss_llm": 0.0, "grad_norm": 0.11823777109384537, "global_step": 30632, "epoch": 850, "lr": 0.001} +{"train_loss": 0.032973840832710266, "train_loss_bc": 0.03296877443790436, "train_loss_llm": 0.0005066990852355957, "grad_norm": 0.041370924562215805, "global_step": 30633, "epoch": 850, "lr": 0.001} +{"train_loss": 0.029429612681269646, "train_loss_bc": 0.029394546523690224, "train_loss_llm": 0.003506690263748169, "grad_norm": 0.06716769188642502, "global_step": 30634, "epoch": 850, "lr": 0.001} +{"train_loss": 0.0299721233960655, "train_loss_bc": 0.03505225107073784, "train_loss_llm": 0.00040155649185180664, "grad_norm": 0.08070845901966095, "global_step": 30635, "epoch": 850, "lr": 0.001, "train/cumulative_reward": 99.43089128843489, "train/mean_score": 15.802298901412962, "train/success_rate": 0.4, "test/cumulative_reward": 36.069552983683494, "test/mean_score": 8.415317501326307, "test/success_rate": 0.22, "val_loss": 0.0986536517739296, "train_action_mse_error": 0.03309233859181404} +{"train_loss": 0.028483405709266663, "train_loss_bc": 0.028481293469667435, "train_loss_llm": 0.00021126866340637207, "grad_norm": 0.09779251366853714, "global_step": 30636, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03370418772101402, "train_loss_bc": 0.03370418772101402, "train_loss_llm": 0.0, "grad_norm": 0.039085157215595245, "global_step": 30637, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03140590339899063, "train_loss_bc": 0.03140590339899063, "train_loss_llm": 0.0, "grad_norm": 0.054094333201646805, "global_step": 30638, "epoch": 851, "lr": 0.001} +{"train_loss": 0.026190489530563354, "train_loss_bc": 0.026181504130363464, "train_loss_llm": 0.0008985996246337891, "grad_norm": 0.06688259541988373, "global_step": 30639, "epoch": 851, "lr": 0.001} +{"train_loss": 0.02359413541853428, "train_loss_bc": 0.023587144911289215, "train_loss_llm": 0.0006990730762481689, "grad_norm": 0.08359826356172562, "global_step": 30640, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03305084630846977, "train_loss_bc": 0.03305084630846977, "train_loss_llm": 0.0, "grad_norm": 0.04037215933203697, "global_step": 30641, "epoch": 851, "lr": 0.001} +{"train_loss": 0.02973245643079281, "train_loss_bc": 0.029721464961767197, "train_loss_llm": 0.0010991096496582031, "grad_norm": 0.06981980800628662, "global_step": 30642, "epoch": 851, "lr": 0.001} +{"train_loss": 0.02655157819390297, "train_loss_bc": 0.02655157819390297, "train_loss_llm": 0.0, "grad_norm": 0.09130837768316269, "global_step": 30643, "epoch": 851, "lr": 0.001} +{"train_loss": 0.029927801340818405, "train_loss_bc": 0.029916243627667427, "train_loss_llm": 0.0011557340621948242, "grad_norm": 0.11486855894327164, "global_step": 30644, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03251577913761139, "train_loss_bc": 0.0324956476688385, "train_loss_llm": 0.0020131468772888184, "grad_norm": 0.04965151101350784, "global_step": 30645, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03330206125974655, "train_loss_bc": 0.03330206125974655, "train_loss_llm": 0.0, "grad_norm": 0.09033216536045074, "global_step": 30646, "epoch": 851, "lr": 0.001} +{"train_loss": 0.02661813795566559, "train_loss_bc": 0.02661813795566559, "train_loss_llm": 0.0, "grad_norm": 0.11339820921421051, "global_step": 30647, "epoch": 851, "lr": 0.001} +{"train_loss": 0.027637960389256477, "train_loss_bc": 0.027637960389256477, "train_loss_llm": 0.0, "grad_norm": 0.13881628215312958, "global_step": 30648, "epoch": 851, "lr": 0.001} +{"train_loss": 0.029263008385896683, "train_loss_bc": 0.029263008385896683, "train_loss_llm": 0.0, "grad_norm": 0.029594939202070236, "global_step": 30649, "epoch": 851, "lr": 0.001} +{"train_loss": 0.028369544073939323, "train_loss_bc": 0.028361953794956207, "train_loss_llm": 0.0007590651512145996, "grad_norm": 0.05021033063530922, "global_step": 30650, "epoch": 851, "lr": 0.001} +{"train_loss": 0.026479925960302353, "train_loss_bc": 0.026458023115992546, "train_loss_llm": 0.0021902620792388916, "grad_norm": 0.05880484730005264, "global_step": 30651, "epoch": 851, "lr": 0.001} +{"train_loss": 0.035321611911058426, "train_loss_bc": 0.035321611911058426, "train_loss_llm": 0.0, "grad_norm": 0.07866419851779938, "global_step": 30652, "epoch": 851, "lr": 0.001} +{"train_loss": 0.036524321883916855, "train_loss_bc": 0.03652427718043327, "train_loss_llm": 4.500150680541992e-06, "grad_norm": 0.040641020983457565, "global_step": 30653, "epoch": 851, "lr": 0.001} +{"train_loss": 0.029487216845154762, "train_loss_bc": 0.029487216845154762, "train_loss_llm": 0.0, "grad_norm": 0.06047682836651802, "global_step": 30654, "epoch": 851, "lr": 0.001} +{"train_loss": 0.025650106370449066, "train_loss_bc": 0.025650106370449066, "train_loss_llm": 0.0, "grad_norm": 0.0829746350646019, "global_step": 30655, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03289968892931938, "train_loss_bc": 0.03289968892931938, "train_loss_llm": 0.0, "grad_norm": 0.09598740190267563, "global_step": 30656, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03206345811486244, "train_loss_bc": 0.03206345811486244, "train_loss_llm": 0.0, "grad_norm": 0.03823232278227806, "global_step": 30657, "epoch": 851, "lr": 0.001} +{"train_loss": 0.027968812733888626, "train_loss_bc": 0.027968812733888626, "train_loss_llm": 0.0, "grad_norm": 0.06256101280450821, "global_step": 30658, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03213333711028099, "train_loss_bc": 0.03213333711028099, "train_loss_llm": 0.0, "grad_norm": 0.08609950542449951, "global_step": 30659, "epoch": 851, "lr": 0.001} +{"train_loss": 0.02747245691716671, "train_loss_bc": 0.02747245691716671, "train_loss_llm": 0.0, "grad_norm": 0.10979154706001282, "global_step": 30660, "epoch": 851, "lr": 0.001} +{"train_loss": 0.028719423338770866, "train_loss_bc": 0.028719423338770866, "train_loss_llm": 0.0, "grad_norm": 0.04040921851992607, "global_step": 30661, "epoch": 851, "lr": 0.001} +{"train_loss": 0.027884729206562042, "train_loss_bc": 0.027884729206562042, "train_loss_llm": 0.0, "grad_norm": 0.07122156023979187, "global_step": 30662, "epoch": 851, "lr": 0.001} +{"train_loss": 0.026897020637989044, "train_loss_bc": 0.026897020637989044, "train_loss_llm": 0.0, "grad_norm": 0.10805374383926392, "global_step": 30663, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03328893706202507, "train_loss_bc": 0.03328721970319748, "train_loss_llm": 0.00017175078392028809, "grad_norm": 0.1330445557832718, "global_step": 30664, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03239654749631882, "train_loss_bc": 0.03239654749631882, "train_loss_llm": 0.0, "grad_norm": 0.043385401368141174, "global_step": 30665, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03183791786432266, "train_loss_bc": 0.03183791786432266, "train_loss_llm": 0.0, "grad_norm": 0.05914980545639992, "global_step": 30666, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03057895414531231, "train_loss_bc": 0.03057895414531231, "train_loss_llm": 0.0, "grad_norm": 0.09801860898733139, "global_step": 30667, "epoch": 851, "lr": 0.001} +{"train_loss": 0.03328295052051544, "train_loss_bc": 0.033280570060014725, "train_loss_llm": 0.00023809075355529785, "grad_norm": 0.13098612427711487, "global_step": 30668, "epoch": 851, "lr": 0.001} +{"train_loss": 0.027550185099244118, "train_loss_bc": 0.027534395456314087, "train_loss_llm": 0.0015789568424224854, "grad_norm": 0.0368875227868557, "global_step": 30669, "epoch": 851, "lr": 0.001} +{"train_loss": 0.029239652678370476, "train_loss_bc": 0.029239652678370476, "train_loss_llm": 0.0, "grad_norm": 0.05828488990664482, "global_step": 30670, "epoch": 851, "lr": 0.001} +{"train_loss": 0.029971579658902354, "train_loss_bc": 0.030952317640185356, "train_loss_llm": 0.0, "grad_norm": 0.07534625381231308, "global_step": 30671, "epoch": 851, "lr": 0.001, "val_loss": 0.10029149055480957} +{"train_loss": 0.023913592100143433, "train_loss_bc": 0.023897651582956314, "train_loss_llm": 0.0015940666198730469, "grad_norm": 0.08038470149040222, "global_step": 30672, "epoch": 852, "lr": 0.001} +{"train_loss": 0.02986796572804451, "train_loss_bc": 0.02986796572804451, "train_loss_llm": 0.0, "grad_norm": 0.037639979273080826, "global_step": 30673, "epoch": 852, "lr": 0.001} +{"train_loss": 0.025310203433036804, "train_loss_bc": 0.025308415293693542, "train_loss_llm": 0.00017878413200378418, "grad_norm": 0.05784238502383232, "global_step": 30674, "epoch": 852, "lr": 0.001} +{"train_loss": 0.0356920026242733, "train_loss_bc": 0.0356920026242733, "train_loss_llm": 0.0, "grad_norm": 0.08829768747091293, "global_step": 30675, "epoch": 852, "lr": 0.001} +{"train_loss": 0.0295040812343359, "train_loss_bc": 0.029493313282728195, "train_loss_llm": 0.0010767877101898193, "grad_norm": 0.12263171374797821, "global_step": 30676, "epoch": 852, "lr": 0.001} +{"train_loss": 0.028746571391820908, "train_loss_bc": 0.028746571391820908, "train_loss_llm": 0.0, "grad_norm": 0.04464932158589363, "global_step": 30677, "epoch": 852, "lr": 0.001} +{"train_loss": 0.030286755412817, "train_loss_bc": 0.030286755412817, "train_loss_llm": 0.0, "grad_norm": 0.06704699993133545, "global_step": 30678, "epoch": 852, "lr": 0.001} +{"train_loss": 0.029054446145892143, "train_loss_bc": 0.029054446145892143, "train_loss_llm": 0.0, "grad_norm": 0.08305344730615616, "global_step": 30679, "epoch": 852, "lr": 0.001} +{"train_loss": 0.03386695683002472, "train_loss_bc": 0.03386695683002472, "train_loss_llm": 0.0, "grad_norm": 0.11481482535600662, "global_step": 30680, "epoch": 852, "lr": 0.001} +{"train_loss": 0.031331561505794525, "train_loss_bc": 0.031331561505794525, "train_loss_llm": 0.0, "grad_norm": 0.04185336455702782, "global_step": 30681, "epoch": 852, "lr": 0.001} +{"train_loss": 0.031248293817043304, "train_loss_bc": 0.03123035840690136, "train_loss_llm": 0.0017936229705810547, "grad_norm": 0.07233545184135437, "global_step": 30682, "epoch": 852, "lr": 0.001} +{"train_loss": 0.02675778977572918, "train_loss_bc": 0.02675778977572918, "train_loss_llm": 0.0, "grad_norm": 0.09359461069107056, "global_step": 30683, "epoch": 852, "lr": 0.001} +{"train_loss": 0.026256462559103966, "train_loss_bc": 0.026256462559103966, "train_loss_llm": 0.0, "grad_norm": 0.11301705986261368, "global_step": 30684, "epoch": 852, "lr": 0.001} +{"train_loss": 0.030577264726161957, "train_loss_bc": 0.030566010624170303, "train_loss_llm": 0.0011253654956817627, "grad_norm": 0.032418377697467804, "global_step": 30685, "epoch": 852, "lr": 0.001} +{"train_loss": 0.035506170243024826, "train_loss_bc": 0.035506170243024826, "train_loss_llm": 0.0, "grad_norm": 0.06292188167572021, "global_step": 30686, "epoch": 852, "lr": 0.001} +{"train_loss": 0.03765452280640602, "train_loss_bc": 0.03765452280640602, "train_loss_llm": 0.0, "grad_norm": 0.09728467464447021, "global_step": 30687, "epoch": 852, "lr": 0.001} +{"train_loss": 0.033446379005908966, "train_loss_bc": 0.03344307839870453, "train_loss_llm": 0.0003298819065093994, "grad_norm": 0.11082541942596436, "global_step": 30688, "epoch": 852, "lr": 0.001} +{"train_loss": 0.027949638664722443, "train_loss_bc": 0.027949638664722443, "train_loss_llm": 0.0, "grad_norm": 0.03020133636891842, "global_step": 30689, "epoch": 852, "lr": 0.001} +{"train_loss": 0.031272243708372116, "train_loss_bc": 0.031272243708372116, "train_loss_llm": 0.0, "grad_norm": 0.060359444469213486, "global_step": 30690, "epoch": 852, "lr": 0.001} +{"train_loss": 0.027070093899965286, "train_loss_bc": 0.027068138122558594, "train_loss_llm": 0.00019562244415283203, "grad_norm": 0.06994910538196564, "global_step": 30691, "epoch": 852, "lr": 0.001} +{"train_loss": 0.02312077395617962, "train_loss_bc": 0.02312077395617962, "train_loss_llm": 0.0, "grad_norm": 0.09049423038959503, "global_step": 30692, "epoch": 852, "lr": 0.001} +{"train_loss": 0.028294900432229042, "train_loss_bc": 0.028294900432229042, "train_loss_llm": 0.0, "grad_norm": 0.04170878604054451, "global_step": 30693, "epoch": 852, "lr": 0.001} +{"train_loss": 0.03117809258401394, "train_loss_bc": 0.031164374202489853, "train_loss_llm": 0.001371830701828003, "grad_norm": 0.08058386296033859, "global_step": 30694, "epoch": 852, "lr": 0.001} +{"train_loss": 0.03322140499949455, "train_loss_bc": 0.03322140499949455, "train_loss_llm": 0.0, "grad_norm": 0.09665057808160782, "global_step": 30695, "epoch": 852, "lr": 0.001} +{"train_loss": 0.0281720869243145, "train_loss_bc": 0.028171401470899582, "train_loss_llm": 6.860494613647461e-05, "grad_norm": 0.14298157393932343, "global_step": 30696, "epoch": 852, "lr": 0.001} +{"train_loss": 0.026094533503055573, "train_loss_bc": 0.026094533503055573, "train_loss_llm": 0.0, "grad_norm": 0.03748112544417381, "global_step": 30697, "epoch": 852, "lr": 0.001} +{"train_loss": 0.034760817885398865, "train_loss_bc": 0.03475933521986008, "train_loss_llm": 0.0001481771469116211, "grad_norm": 0.08105213940143585, "global_step": 30698, "epoch": 852, "lr": 0.001} +{"train_loss": 0.03068343922495842, "train_loss_bc": 0.03068343922495842, "train_loss_llm": 0.0, "grad_norm": 0.12795734405517578, "global_step": 30699, "epoch": 852, "lr": 0.001} +{"train_loss": 0.02921110950410366, "train_loss_bc": 0.02921110950410366, "train_loss_llm": 0.0, "grad_norm": 0.15472543239593506, "global_step": 30700, "epoch": 852, "lr": 0.001} +{"train_loss": 0.03111639805138111, "train_loss_bc": 0.03111639805138111, "train_loss_llm": 0.0, "grad_norm": 0.02774970233440399, "global_step": 30701, "epoch": 852, "lr": 0.001} +{"train_loss": 0.032095782458782196, "train_loss_bc": 0.032095782458782196, "train_loss_llm": 0.0, "grad_norm": 0.04698880761861801, "global_step": 30702, "epoch": 852, "lr": 0.001} +{"train_loss": 0.03171292319893837, "train_loss_bc": 0.03171292319893837, "train_loss_llm": 0.0, "grad_norm": 0.08012895286083221, "global_step": 30703, "epoch": 852, "lr": 0.001} +{"train_loss": 0.02610877901315689, "train_loss_bc": 0.02610877901315689, "train_loss_llm": 0.0, "grad_norm": 0.08964314311742783, "global_step": 30704, "epoch": 852, "lr": 0.001} +{"train_loss": 0.027387021109461784, "train_loss_bc": 0.027378469705581665, "train_loss_llm": 0.0008551180362701416, "grad_norm": 0.03958595544099808, "global_step": 30705, "epoch": 852, "lr": 0.001} +{"train_loss": 0.02843029424548149, "train_loss_bc": 0.028425026684999466, "train_loss_llm": 0.0005267858505249023, "grad_norm": 0.052501168102025986, "global_step": 30706, "epoch": 852, "lr": 0.001} +{"train_loss": 0.029912912752479315, "train_loss_bc": 0.029963506385684013, "train_loss_llm": 0.0, "grad_norm": 0.08340234309434891, "global_step": 30707, "epoch": 852, "lr": 0.001, "val_loss": 0.10180089622735977} +{"train_loss": 0.032866451889276505, "train_loss_bc": 0.03286062553524971, "train_loss_llm": 0.0005827844142913818, "grad_norm": 0.10682737082242966, "global_step": 30708, "epoch": 853, "lr": 0.001} +{"train_loss": 0.02679661102592945, "train_loss_bc": 0.02679661102592945, "train_loss_llm": 0.0, "grad_norm": 0.0357908271253109, "global_step": 30709, "epoch": 853, "lr": 0.001} +{"train_loss": 0.029339458793401718, "train_loss_bc": 0.029339458793401718, "train_loss_llm": 0.0, "grad_norm": 0.07150940597057343, "global_step": 30710, "epoch": 853, "lr": 0.001} +{"train_loss": 0.033863380551338196, "train_loss_bc": 0.033863380551338196, "train_loss_llm": 0.0, "grad_norm": 0.10190054029226303, "global_step": 30711, "epoch": 853, "lr": 0.001} +{"train_loss": 0.02297637239098549, "train_loss_bc": 0.02297637239098549, "train_loss_llm": 0.0, "grad_norm": 0.11691685020923615, "global_step": 30712, "epoch": 853, "lr": 0.001} +{"train_loss": 0.03030453808605671, "train_loss_bc": 0.03030453808605671, "train_loss_llm": 0.0, "grad_norm": 0.03220568597316742, "global_step": 30713, "epoch": 853, "lr": 0.001} +{"train_loss": 0.02428392693400383, "train_loss_bc": 0.02428392693400383, "train_loss_llm": 0.0, "grad_norm": 0.05825931578874588, "global_step": 30714, "epoch": 853, "lr": 0.001} +{"train_loss": 0.027164384722709656, "train_loss_bc": 0.027164384722709656, "train_loss_llm": 0.0, "grad_norm": 0.0814814567565918, "global_step": 30715, "epoch": 853, "lr": 0.001} +{"train_loss": 0.02740436978638172, "train_loss_bc": 0.02740436978638172, "train_loss_llm": 0.0, "grad_norm": 0.11725857853889465, "global_step": 30716, "epoch": 853, "lr": 0.001} +{"train_loss": 0.03626594692468643, "train_loss_bc": 0.03626594692468643, "train_loss_llm": 0.0, "grad_norm": 0.03971327841281891, "global_step": 30717, "epoch": 853, "lr": 0.001} +{"train_loss": 0.03191303461790085, "train_loss_bc": 0.03191303461790085, "train_loss_llm": 0.0, "grad_norm": 0.053809016942977905, "global_step": 30718, "epoch": 853, "lr": 0.001} +{"train_loss": 0.027239615097641945, "train_loss_bc": 0.027238450944423676, "train_loss_llm": 0.0001163482666015625, "grad_norm": 0.07035320997238159, "global_step": 30719, "epoch": 853, "lr": 0.001} +{"train_loss": 0.031307660043239594, "train_loss_bc": 0.031307660043239594, "train_loss_llm": 0.0, "grad_norm": 0.09404467046260834, "global_step": 30720, "epoch": 853, "lr": 0.001} +{"train_loss": 0.0355430468916893, "train_loss_bc": 0.0355430468916893, "train_loss_llm": 0.0, "grad_norm": 0.04286355525255203, "global_step": 30721, "epoch": 853, "lr": 0.001} +{"train_loss": 0.03160687908530235, "train_loss_bc": 0.03160687908530235, "train_loss_llm": 0.0, "grad_norm": 0.05652615427970886, "global_step": 30722, "epoch": 853, "lr": 0.001} +{"train_loss": 0.031239047646522522, "train_loss_bc": 0.031239047646522522, "train_loss_llm": 0.0, "grad_norm": 0.07259638607501984, "global_step": 30723, "epoch": 853, "lr": 0.001} +{"train_loss": 0.03282932564616203, "train_loss_bc": 0.03282932564616203, "train_loss_llm": 0.0, "grad_norm": 0.09532491117715836, "global_step": 30724, "epoch": 853, "lr": 0.001} +{"train_loss": 0.025084108114242554, "train_loss_bc": 0.025084108114242554, "train_loss_llm": 0.0, "grad_norm": 0.026074374094605446, "global_step": 30725, "epoch": 853, "lr": 0.001} +{"train_loss": 0.03064606711268425, "train_loss_bc": 0.03064606711268425, "train_loss_llm": 0.0, "grad_norm": 0.04866226017475128, "global_step": 30726, "epoch": 853, "lr": 0.001} +{"train_loss": 0.027299651876091957, "train_loss_bc": 0.027299651876091957, "train_loss_llm": 0.0, "grad_norm": 0.0755615383386612, "global_step": 30727, "epoch": 853, "lr": 0.001} +{"train_loss": 0.027264608070254326, "train_loss_bc": 0.027264608070254326, "train_loss_llm": 0.0, "grad_norm": 0.10060715675354004, "global_step": 30728, "epoch": 853, "lr": 0.001} +{"train_loss": 0.03479953855276108, "train_loss_bc": 0.03479813411831856, "train_loss_llm": 0.00014036893844604492, "grad_norm": 0.047780636698007584, "global_step": 30729, "epoch": 853, "lr": 0.001} +{"train_loss": 0.032824113965034485, "train_loss_bc": 0.032824113965034485, "train_loss_llm": 0.0, "grad_norm": 0.09160984307527542, "global_step": 30730, "epoch": 853, "lr": 0.001} +{"train_loss": 0.03523712232708931, "train_loss_bc": 0.03523712232708931, "train_loss_llm": 0.0, "grad_norm": 0.11812539398670197, "global_step": 30731, "epoch": 853, "lr": 0.001} +{"train_loss": 0.028999218717217445, "train_loss_bc": 0.028999218717217445, "train_loss_llm": 0.0, "grad_norm": 0.16010889410972595, "global_step": 30732, "epoch": 853, "lr": 0.001} +{"train_loss": 0.028786152601242065, "train_loss_bc": 0.028786152601242065, "train_loss_llm": 0.0, "grad_norm": 0.03670591861009598, "global_step": 30733, "epoch": 853, "lr": 0.001} +{"train_loss": 0.02440762147307396, "train_loss_bc": 0.024392277002334595, "train_loss_llm": 0.0015345215797424316, "grad_norm": 0.0668194517493248, "global_step": 30734, "epoch": 853, "lr": 0.001} +{"train_loss": 0.029163390398025513, "train_loss_bc": 0.029158324003219604, "train_loss_llm": 0.0005065500736236572, "grad_norm": 0.08390134572982788, "global_step": 30735, "epoch": 853, "lr": 0.001} +{"train_loss": 0.025647006928920746, "train_loss_bc": 0.025647006928920746, "train_loss_llm": 0.0, "grad_norm": 0.1074860543012619, "global_step": 30736, "epoch": 853, "lr": 0.001} +{"train_loss": 0.03503568843007088, "train_loss_bc": 0.035032015293836594, "train_loss_llm": 0.0003673732280731201, "grad_norm": 0.035341814160346985, "global_step": 30737, "epoch": 853, "lr": 0.001} +{"train_loss": 0.029593372717499733, "train_loss_bc": 0.029579922556877136, "train_loss_llm": 0.0013449490070343018, "grad_norm": 0.06388352811336517, "global_step": 30738, "epoch": 853, "lr": 0.001} +{"train_loss": 0.025743652135133743, "train_loss_bc": 0.025743652135133743, "train_loss_llm": 0.0, "grad_norm": 0.08470503240823746, "global_step": 30739, "epoch": 853, "lr": 0.001} +{"train_loss": 0.029747968539595604, "train_loss_bc": 0.029728855937719345, "train_loss_llm": 0.001911252737045288, "grad_norm": 0.10455527901649475, "global_step": 30740, "epoch": 853, "lr": 0.001} +{"train_loss": 0.030784007161855698, "train_loss_bc": 0.030784007161855698, "train_loss_llm": 0.0, "grad_norm": 0.03553363308310509, "global_step": 30741, "epoch": 853, "lr": 0.001} +{"train_loss": 0.033103473484516144, "train_loss_bc": 0.033103473484516144, "train_loss_llm": 0.0, "grad_norm": 0.05853980779647827, "global_step": 30742, "epoch": 853, "lr": 0.001} +{"train_loss": 0.029796550360818703, "train_loss_bc": 0.025565000250935555, "train_loss_llm": 0.0, "grad_norm": 0.0700262114405632, "global_step": 30743, "epoch": 853, "lr": 0.001, "val_loss": 0.10075333714485168} +{"train_loss": 0.028829270973801613, "train_loss_bc": 0.028829270973801613, "train_loss_llm": 0.0, "grad_norm": 0.08232244104146957, "global_step": 30744, "epoch": 854, "lr": 0.001} +{"train_loss": 0.02597978338599205, "train_loss_bc": 0.02597978338599205, "train_loss_llm": 0.0, "grad_norm": 0.03652820736169815, "global_step": 30745, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03283189982175827, "train_loss_bc": 0.03283189982175827, "train_loss_llm": 0.0, "grad_norm": 0.0775069147348404, "global_step": 30746, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03269021585583687, "train_loss_bc": 0.03269021585583687, "train_loss_llm": 0.0, "grad_norm": 0.10633701086044312, "global_step": 30747, "epoch": 854, "lr": 0.001} +{"train_loss": 0.029023215174674988, "train_loss_bc": 0.029023215174674988, "train_loss_llm": 0.0, "grad_norm": 0.13431860506534576, "global_step": 30748, "epoch": 854, "lr": 0.001} +{"train_loss": 0.02761751227080822, "train_loss_bc": 0.02761751227080822, "train_loss_llm": 0.0, "grad_norm": 0.05141782388091087, "global_step": 30749, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03556330129504204, "train_loss_bc": 0.035555291920900345, "train_loss_llm": 0.0008011162281036377, "grad_norm": 0.09523266553878784, "global_step": 30750, "epoch": 854, "lr": 0.001} +{"train_loss": 0.028254013508558273, "train_loss_bc": 0.028254013508558273, "train_loss_llm": 0.0, "grad_norm": 0.1213020384311676, "global_step": 30751, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03088308684527874, "train_loss_bc": 0.03088308684527874, "train_loss_llm": 0.0, "grad_norm": 0.16015228629112244, "global_step": 30752, "epoch": 854, "lr": 0.001} +{"train_loss": 0.027225809171795845, "train_loss_bc": 0.02722499705851078, "train_loss_llm": 8.127093315124512e-05, "grad_norm": 0.03444109857082367, "global_step": 30753, "epoch": 854, "lr": 0.001} +{"train_loss": 0.02634648233652115, "train_loss_bc": 0.02634648233652115, "train_loss_llm": 0.0, "grad_norm": 0.05692341551184654, "global_step": 30754, "epoch": 854, "lr": 0.001} +{"train_loss": 0.02224535495042801, "train_loss_bc": 0.022241637110710144, "train_loss_llm": 0.00037172436714172363, "grad_norm": 0.07454030215740204, "global_step": 30755, "epoch": 854, "lr": 0.001} +{"train_loss": 0.02840627357363701, "train_loss_bc": 0.02840411104261875, "train_loss_llm": 0.00021630525588989258, "grad_norm": 0.09895962476730347, "global_step": 30756, "epoch": 854, "lr": 0.001} +{"train_loss": 0.029154127463698387, "train_loss_bc": 0.029154127463698387, "train_loss_llm": 0.0, "grad_norm": 0.03477925434708595, "global_step": 30757, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03056025318801403, "train_loss_bc": 0.030554739758372307, "train_loss_llm": 0.0005512833595275879, "grad_norm": 0.061726849526166916, "global_step": 30758, "epoch": 854, "lr": 0.001} +{"train_loss": 0.032636888325214386, "train_loss_bc": 0.032636888325214386, "train_loss_llm": 0.0, "grad_norm": 0.07871994376182556, "global_step": 30759, "epoch": 854, "lr": 0.001} +{"train_loss": 0.035322271287441254, "train_loss_bc": 0.035322271287441254, "train_loss_llm": 0.0, "grad_norm": 0.10164979100227356, "global_step": 30760, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03355483338236809, "train_loss_bc": 0.03355483338236809, "train_loss_llm": 0.0, "grad_norm": 0.05736484378576279, "global_step": 30761, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03181662783026695, "train_loss_bc": 0.03180938959121704, "train_loss_llm": 0.0007239878177642822, "grad_norm": 0.08959570527076721, "global_step": 30762, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03245603293180466, "train_loss_bc": 0.03245603293180466, "train_loss_llm": 0.0, "grad_norm": 0.11614492535591125, "global_step": 30763, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03267008066177368, "train_loss_bc": 0.03267008066177368, "train_loss_llm": 0.0, "grad_norm": 0.14523832499980927, "global_step": 30764, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03151867538690567, "train_loss_bc": 0.03151867538690567, "train_loss_llm": 0.0, "grad_norm": 0.045094989240169525, "global_step": 30765, "epoch": 854, "lr": 0.001} +{"train_loss": 0.031333077698946, "train_loss_bc": 0.031333077698946, "train_loss_llm": 0.0, "grad_norm": 0.06877348572015762, "global_step": 30766, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03611648455262184, "train_loss_bc": 0.03611648455262184, "train_loss_llm": 0.0, "grad_norm": 0.08201761543750763, "global_step": 30767, "epoch": 854, "lr": 0.001} +{"train_loss": 0.021557962521910667, "train_loss_bc": 0.021553300321102142, "train_loss_llm": 0.00046628713607788086, "grad_norm": 0.11492066085338593, "global_step": 30768, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03898685425519943, "train_loss_bc": 0.03898685425519943, "train_loss_llm": 0.0, "grad_norm": 0.0421864315867424, "global_step": 30769, "epoch": 854, "lr": 0.001} +{"train_loss": 0.028427956625819206, "train_loss_bc": 0.028413210064172745, "train_loss_llm": 0.0014746487140655518, "grad_norm": 0.06511644273996353, "global_step": 30770, "epoch": 854, "lr": 0.001} +{"train_loss": 0.026378881186246872, "train_loss_bc": 0.026378881186246872, "train_loss_llm": 0.0, "grad_norm": 0.08981458842754364, "global_step": 30771, "epoch": 854, "lr": 0.001} +{"train_loss": 0.02784222736954689, "train_loss_bc": 0.02784174680709839, "train_loss_llm": 4.8041343688964844e-05, "grad_norm": 0.11633242666721344, "global_step": 30772, "epoch": 854, "lr": 0.001} +{"train_loss": 0.02981247939169407, "train_loss_bc": 0.029810000211000443, "train_loss_llm": 0.0002478957176208496, "grad_norm": 0.0381292849779129, "global_step": 30773, "epoch": 854, "lr": 0.001} +{"train_loss": 0.02911989763379097, "train_loss_bc": 0.02911989763379097, "train_loss_llm": 0.0, "grad_norm": 0.06750410795211792, "global_step": 30774, "epoch": 854, "lr": 0.001} +{"train_loss": 0.02477874420583248, "train_loss_bc": 0.02477874420583248, "train_loss_llm": 0.0, "grad_norm": 0.0955204889178276, "global_step": 30775, "epoch": 854, "lr": 0.001} +{"train_loss": 0.030847618356347084, "train_loss_bc": 0.030847618356347084, "train_loss_llm": 0.0, "grad_norm": 0.11030841618776321, "global_step": 30776, "epoch": 854, "lr": 0.001} +{"train_loss": 0.02938099391758442, "train_loss_bc": 0.02938099391758442, "train_loss_llm": 0.0, "grad_norm": 0.043671075254678726, "global_step": 30777, "epoch": 854, "lr": 0.001} +{"train_loss": 0.03159863129258156, "train_loss_bc": 0.03159863129258156, "train_loss_llm": 0.0, "grad_norm": 0.06458687782287598, "global_step": 30778, "epoch": 854, "lr": 0.001} +{"train_loss": 0.030070001983808145, "train_loss_bc": 0.03075225278735161, "train_loss_llm": 0.0, "grad_norm": 0.10013701766729355, "global_step": 30779, "epoch": 854, "lr": 0.001, "val_loss": 0.09831951558589935} +{"train_loss": 0.030110150575637817, "train_loss_bc": 0.030110150575637817, "train_loss_llm": 0.0, "grad_norm": 0.12070522457361221, "global_step": 30780, "epoch": 855, "lr": 0.001} +{"train_loss": 0.02668774500489235, "train_loss_bc": 0.02668488211929798, "train_loss_llm": 0.00028631091117858887, "grad_norm": 0.038762860000133514, "global_step": 30781, "epoch": 855, "lr": 0.001} +{"train_loss": 0.03271924704313278, "train_loss_bc": 0.03271924704313278, "train_loss_llm": 0.0, "grad_norm": 0.07997649908065796, "global_step": 30782, "epoch": 855, "lr": 0.001} +{"train_loss": 0.031195132061839104, "train_loss_bc": 0.031191039830446243, "train_loss_llm": 0.00040924549102783203, "grad_norm": 0.09752390533685684, "global_step": 30783, "epoch": 855, "lr": 0.001} +{"train_loss": 0.029835205525159836, "train_loss_bc": 0.029835205525159836, "train_loss_llm": 0.0, "grad_norm": 0.11932285130023956, "global_step": 30784, "epoch": 855, "lr": 0.001} +{"train_loss": 0.02644054964184761, "train_loss_bc": 0.02644054964184761, "train_loss_llm": 0.0, "grad_norm": 0.03135053068399429, "global_step": 30785, "epoch": 855, "lr": 0.001} +{"train_loss": 0.028034212067723274, "train_loss_bc": 0.028034212067723274, "train_loss_llm": 0.0, "grad_norm": 0.05524627864360809, "global_step": 30786, "epoch": 855, "lr": 0.001} +{"train_loss": 0.02765118144452572, "train_loss_bc": 0.02765118144452572, "train_loss_llm": 0.0, "grad_norm": 0.05631246790289879, "global_step": 30787, "epoch": 855, "lr": 0.001} +{"train_loss": 0.028952142223715782, "train_loss_bc": 0.028952142223715782, "train_loss_llm": 0.0, "grad_norm": 0.0821285992860794, "global_step": 30788, "epoch": 855, "lr": 0.001} +{"train_loss": 0.028461720794439316, "train_loss_bc": 0.028461720794439316, "train_loss_llm": 0.0, "grad_norm": 0.03611074760556221, "global_step": 30789, "epoch": 855, "lr": 0.001} +{"train_loss": 0.030291376635432243, "train_loss_bc": 0.030285069718956947, "train_loss_llm": 0.0006306767463684082, "grad_norm": 0.06379477679729462, "global_step": 30790, "epoch": 855, "lr": 0.001} +{"train_loss": 0.029805662110447884, "train_loss_bc": 0.029805662110447884, "train_loss_llm": 0.0, "grad_norm": 0.09963823109865189, "global_step": 30791, "epoch": 855, "lr": 0.001} +{"train_loss": 0.027428001165390015, "train_loss_bc": 0.027428001165390015, "train_loss_llm": 0.0, "grad_norm": 0.11889349669218063, "global_step": 30792, "epoch": 855, "lr": 0.001} +{"train_loss": 0.03034285083413124, "train_loss_bc": 0.030342673882842064, "train_loss_llm": 1.767277717590332e-05, "grad_norm": 0.03842826187610626, "global_step": 30793, "epoch": 855, "lr": 0.001} +{"train_loss": 0.025223897770047188, "train_loss_bc": 0.025223897770047188, "train_loss_llm": 0.0, "grad_norm": 0.0556570366024971, "global_step": 30794, "epoch": 855, "lr": 0.001} +{"train_loss": 0.03193530812859535, "train_loss_bc": 0.03193530812859535, "train_loss_llm": 0.0, "grad_norm": 0.08436309546232224, "global_step": 30795, "epoch": 855, "lr": 0.001} +{"train_loss": 0.03291740640997887, "train_loss_bc": 0.03291740640997887, "train_loss_llm": 0.0, "grad_norm": 0.11262210458517075, "global_step": 30796, "epoch": 855, "lr": 0.001} +{"train_loss": 0.030474485829472542, "train_loss_bc": 0.030474485829472542, "train_loss_llm": 0.0, "grad_norm": 0.04533271864056587, "global_step": 30797, "epoch": 855, "lr": 0.001} +{"train_loss": 0.027342524379491806, "train_loss_bc": 0.027342524379491806, "train_loss_llm": 0.0, "grad_norm": 0.07562850415706635, "global_step": 30798, "epoch": 855, "lr": 0.001} +{"train_loss": 0.037333108484745026, "train_loss_bc": 0.037333108484745026, "train_loss_llm": 0.0, "grad_norm": 0.11173845827579498, "global_step": 30799, "epoch": 855, "lr": 0.001} +{"train_loss": 0.028044991195201874, "train_loss_bc": 0.02804172970354557, "train_loss_llm": 0.0003261566162109375, "grad_norm": 0.14990027248859406, "global_step": 30800, "epoch": 855, "lr": 0.001} +{"train_loss": 0.02950075827538967, "train_loss_bc": 0.02949460782110691, "train_loss_llm": 0.0006150305271148682, "grad_norm": 0.03274163976311684, "global_step": 30801, "epoch": 855, "lr": 0.001} +{"train_loss": 0.03369303047657013, "train_loss_bc": 0.03369303047657013, "train_loss_llm": 0.0, "grad_norm": 0.04596586897969246, "global_step": 30802, "epoch": 855, "lr": 0.001} +{"train_loss": 0.031480997800827026, "train_loss_bc": 0.031469788402318954, "train_loss_llm": 0.0011207759380340576, "grad_norm": 0.06689104437828064, "global_step": 30803, "epoch": 855, "lr": 0.001} +{"train_loss": 0.03292021155357361, "train_loss_bc": 0.03292021155357361, "train_loss_llm": 0.0, "grad_norm": 0.09118103981018066, "global_step": 30804, "epoch": 855, "lr": 0.001} +{"train_loss": 0.03100217692553997, "train_loss_bc": 0.030978180468082428, "train_loss_llm": 0.0023997128009796143, "grad_norm": 0.04065003618597984, "global_step": 30805, "epoch": 855, "lr": 0.001} +{"train_loss": 0.029620137065649033, "train_loss_bc": 0.029620137065649033, "train_loss_llm": 0.0, "grad_norm": 0.056627966463565826, "global_step": 30806, "epoch": 855, "lr": 0.001} +{"train_loss": 0.029886748641729355, "train_loss_bc": 0.029886748641729355, "train_loss_llm": 0.0, "grad_norm": 0.08076345175504684, "global_step": 30807, "epoch": 855, "lr": 0.001} +{"train_loss": 0.028517810627818108, "train_loss_bc": 0.028517810627818108, "train_loss_llm": 0.0, "grad_norm": 0.08566486090421677, "global_step": 30808, "epoch": 855, "lr": 0.001} +{"train_loss": 0.026682870462536812, "train_loss_bc": 0.026682870462536812, "train_loss_llm": 0.0, "grad_norm": 0.03618016466498375, "global_step": 30809, "epoch": 855, "lr": 0.001} +{"train_loss": 0.029507379978895187, "train_loss_bc": 0.029507379978895187, "train_loss_llm": 0.0, "grad_norm": 0.06372248381376266, "global_step": 30810, "epoch": 855, "lr": 0.001} +{"train_loss": 0.03269343823194504, "train_loss_bc": 0.03269343823194504, "train_loss_llm": 0.0, "grad_norm": 0.08438673615455627, "global_step": 30811, "epoch": 855, "lr": 0.001} +{"train_loss": 0.03327888622879982, "train_loss_bc": 0.033272452652454376, "train_loss_llm": 0.0006432533264160156, "grad_norm": 0.1078975647687912, "global_step": 30812, "epoch": 855, "lr": 0.001} +{"train_loss": 0.02494291216135025, "train_loss_bc": 0.02494291216135025, "train_loss_llm": 0.0, "grad_norm": 0.02927416004240513, "global_step": 30813, "epoch": 855, "lr": 0.001} +{"train_loss": 0.031531184911727905, "train_loss_bc": 0.031531184911727905, "train_loss_llm": 0.0, "grad_norm": 0.057363081723451614, "global_step": 30814, "epoch": 855, "lr": 0.001} +{"train_loss": 0.029943848132259317, "train_loss_bc": 0.03147854655981064, "train_loss_llm": 0.0014544427394866943, "grad_norm": 0.10608573257923126, "global_step": 30815, "epoch": 855, "lr": 0.001, "train/cumulative_reward": 55.61136263853909, "train/mean_score": 9.538362936196581, "train/success_rate": 0.2, "test/cumulative_reward": 36.66773855082761, "test/mean_score": 8.815741321578736, "test/success_rate": 0.24, "val_loss": 0.10182148218154907, "train_action_mse_error": 0.033069342374801636} +{"train_loss": 0.032003190368413925, "train_loss_bc": 0.032003190368413925, "train_loss_llm": 0.0, "grad_norm": 0.12755198776721954, "global_step": 30816, "epoch": 856, "lr": 0.001} +{"train_loss": 0.026571374386548996, "train_loss_bc": 0.026571374386548996, "train_loss_llm": 0.0, "grad_norm": 0.03888730704784393, "global_step": 30817, "epoch": 856, "lr": 0.001} +{"train_loss": 0.032383713871240616, "train_loss_bc": 0.032383713871240616, "train_loss_llm": 0.0, "grad_norm": 0.06109979376196861, "global_step": 30818, "epoch": 856, "lr": 0.001} +{"train_loss": 0.02922246791422367, "train_loss_bc": 0.02922246791422367, "train_loss_llm": 0.0, "grad_norm": 0.07390579581260681, "global_step": 30819, "epoch": 856, "lr": 0.001} +{"train_loss": 0.028492242097854614, "train_loss_bc": 0.028492242097854614, "train_loss_llm": 0.0, "grad_norm": 0.0899014100432396, "global_step": 30820, "epoch": 856, "lr": 0.001} +{"train_loss": 0.02941150777041912, "train_loss_bc": 0.02941150777041912, "train_loss_llm": 0.0, "grad_norm": 0.03429709002375603, "global_step": 30821, "epoch": 856, "lr": 0.001} +{"train_loss": 0.026352491229772568, "train_loss_bc": 0.026352491229772568, "train_loss_llm": 0.0, "grad_norm": 0.04951731115579605, "global_step": 30822, "epoch": 856, "lr": 0.001} +{"train_loss": 0.02997187152504921, "train_loss_bc": 0.02997187152504921, "train_loss_llm": 0.0, "grad_norm": 0.07642220705747604, "global_step": 30823, "epoch": 856, "lr": 0.001} +{"train_loss": 0.030576925724744797, "train_loss_bc": 0.030576925724744797, "train_loss_llm": 0.0, "grad_norm": 0.09837530553340912, "global_step": 30824, "epoch": 856, "lr": 0.001} +{"train_loss": 0.029156869277358055, "train_loss_bc": 0.029156869277358055, "train_loss_llm": 0.0, "grad_norm": 0.03280021622776985, "global_step": 30825, "epoch": 856, "lr": 0.001} +{"train_loss": 0.029990606009960175, "train_loss_bc": 0.029990606009960175, "train_loss_llm": 0.0, "grad_norm": 0.050529010593891144, "global_step": 30826, "epoch": 856, "lr": 0.001} +{"train_loss": 0.034801896661520004, "train_loss_bc": 0.034801896661520004, "train_loss_llm": 0.0, "grad_norm": 0.08007267862558365, "global_step": 30827, "epoch": 856, "lr": 0.001} +{"train_loss": 0.03256175294518471, "train_loss_bc": 0.03256175294518471, "train_loss_llm": 0.0, "grad_norm": 0.10780428349971771, "global_step": 30828, "epoch": 856, "lr": 0.001} +{"train_loss": 0.02354297786951065, "train_loss_bc": 0.02354297786951065, "train_loss_llm": 0.0, "grad_norm": 0.03529668226838112, "global_step": 30829, "epoch": 856, "lr": 0.001} +{"train_loss": 0.029467279091477394, "train_loss_bc": 0.029467279091477394, "train_loss_llm": 0.0, "grad_norm": 0.05365467444062233, "global_step": 30830, "epoch": 856, "lr": 0.001} +{"train_loss": 0.03065253421664238, "train_loss_bc": 0.03065253421664238, "train_loss_llm": 0.0, "grad_norm": 0.0853850394487381, "global_step": 30831, "epoch": 856, "lr": 0.001} +{"train_loss": 0.02834111638367176, "train_loss_bc": 0.02834111638367176, "train_loss_llm": 0.0, "grad_norm": 0.10043179243803024, "global_step": 30832, "epoch": 856, "lr": 0.001} +{"train_loss": 0.028075681999325752, "train_loss_bc": 0.028075681999325752, "train_loss_llm": 0.0, "grad_norm": 0.0378512479364872, "global_step": 30833, "epoch": 856, "lr": 0.001} +{"train_loss": 0.0299161858856678, "train_loss_bc": 0.029910657554864883, "train_loss_llm": 0.000552833080291748, "grad_norm": 0.06321028620004654, "global_step": 30834, "epoch": 856, "lr": 0.001} +{"train_loss": 0.033020347356796265, "train_loss_bc": 0.03301007300615311, "train_loss_llm": 0.0010274648666381836, "grad_norm": 0.07604023069143295, "global_step": 30835, "epoch": 856, "lr": 0.001} +{"train_loss": 0.027683107182383537, "train_loss_bc": 0.027677148580551147, "train_loss_llm": 0.000595778226852417, "grad_norm": 0.08615526556968689, "global_step": 30836, "epoch": 856, "lr": 0.001} +{"train_loss": 0.028082307428121567, "train_loss_bc": 0.028082307428121567, "train_loss_llm": 0.0, "grad_norm": 0.04823768138885498, "global_step": 30837, "epoch": 856, "lr": 0.001} +{"train_loss": 0.031080245971679688, "train_loss_bc": 0.031080245971679688, "train_loss_llm": 0.0, "grad_norm": 0.06901347637176514, "global_step": 30838, "epoch": 856, "lr": 0.001} +{"train_loss": 0.029696006327867508, "train_loss_bc": 0.029696006327867508, "train_loss_llm": 0.0, "grad_norm": 0.10052250325679779, "global_step": 30839, "epoch": 856, "lr": 0.001} +{"train_loss": 0.03356991708278656, "train_loss_bc": 0.03356991708278656, "train_loss_llm": 0.0, "grad_norm": 0.13656263053417206, "global_step": 30840, "epoch": 856, "lr": 0.001} +{"train_loss": 0.028665270656347275, "train_loss_bc": 0.028665270656347275, "train_loss_llm": 0.0, "grad_norm": 0.03775825351476669, "global_step": 30841, "epoch": 856, "lr": 0.001} +{"train_loss": 0.034380849450826645, "train_loss_bc": 0.03436781093478203, "train_loss_llm": 0.00130385160446167, "grad_norm": 0.04583807289600372, "global_step": 30842, "epoch": 856, "lr": 0.001} +{"train_loss": 0.03413693234324455, "train_loss_bc": 0.03413693234324455, "train_loss_llm": 0.0, "grad_norm": 0.05914485827088356, "global_step": 30843, "epoch": 856, "lr": 0.001} +{"train_loss": 0.030079703778028488, "train_loss_bc": 0.030079703778028488, "train_loss_llm": 0.0, "grad_norm": 0.06656491011381149, "global_step": 30844, "epoch": 856, "lr": 0.001} +{"train_loss": 0.02567843720316887, "train_loss_bc": 0.02567843720316887, "train_loss_llm": 0.0, "grad_norm": 0.031546976417303085, "global_step": 30845, "epoch": 856, "lr": 0.001} +{"train_loss": 0.03162667900323868, "train_loss_bc": 0.03162667900323868, "train_loss_llm": 0.0, "grad_norm": 0.04877395182847977, "global_step": 30846, "epoch": 856, "lr": 0.001} +{"train_loss": 0.027465837076306343, "train_loss_bc": 0.027453767135739326, "train_loss_llm": 0.001206904649734497, "grad_norm": 0.06870479136705399, "global_step": 30847, "epoch": 856, "lr": 0.001} +{"train_loss": 0.030631568282842636, "train_loss_bc": 0.030631568282842636, "train_loss_llm": 0.0, "grad_norm": 0.09507719427347183, "global_step": 30848, "epoch": 856, "lr": 0.001} +{"train_loss": 0.0293145589530468, "train_loss_bc": 0.0293145589530468, "train_loss_llm": 0.0, "grad_norm": 0.037689920514822006, "global_step": 30849, "epoch": 856, "lr": 0.001} +{"train_loss": 0.026177892461419106, "train_loss_bc": 0.026177892461419106, "train_loss_llm": 0.0, "grad_norm": 0.055900316685438156, "global_step": 30850, "epoch": 856, "lr": 0.001} +{"train_loss": 0.02965494431555271, "train_loss_bc": 0.024782462045550346, "train_loss_llm": 0.0013187825679779053, "grad_norm": 0.07578772306442261, "global_step": 30851, "epoch": 856, "lr": 0.001, "val_loss": 0.10072961449623108} +{"train_loss": 0.02396550588309765, "train_loss_bc": 0.023955121636390686, "train_loss_llm": 0.0010384619235992432, "grad_norm": 0.08714163303375244, "global_step": 30852, "epoch": 857, "lr": 0.001} +{"train_loss": 0.029991785064339638, "train_loss_bc": 0.029991785064339638, "train_loss_llm": 0.0, "grad_norm": 0.03500373288989067, "global_step": 30853, "epoch": 857, "lr": 0.001} +{"train_loss": 0.03977848216891289, "train_loss_bc": 0.03977848216891289, "train_loss_llm": 0.0, "grad_norm": 0.07820192724466324, "global_step": 30854, "epoch": 857, "lr": 0.001} +{"train_loss": 0.02971290424466133, "train_loss_bc": 0.02971290424466133, "train_loss_llm": 0.0, "grad_norm": 0.09510117769241333, "global_step": 30855, "epoch": 857, "lr": 0.001} +{"train_loss": 0.03186459466814995, "train_loss_bc": 0.03186459466814995, "train_loss_llm": 0.0, "grad_norm": 0.12074097990989685, "global_step": 30856, "epoch": 857, "lr": 0.001} +{"train_loss": 0.027442889288067818, "train_loss_bc": 0.027442889288067818, "train_loss_llm": 0.0, "grad_norm": 0.0410461351275444, "global_step": 30857, "epoch": 857, "lr": 0.001} +{"train_loss": 0.03193692862987518, "train_loss_bc": 0.03193692862987518, "train_loss_llm": 0.0, "grad_norm": 0.05694573372602463, "global_step": 30858, "epoch": 857, "lr": 0.001} +{"train_loss": 0.023729218170046806, "train_loss_bc": 0.023718176409602165, "train_loss_llm": 0.0011042654514312744, "grad_norm": 0.08450596779584885, "global_step": 30859, "epoch": 857, "lr": 0.001} +{"train_loss": 0.028250275179743767, "train_loss_bc": 0.028250275179743767, "train_loss_llm": 0.0, "grad_norm": 0.09220053255558014, "global_step": 30860, "epoch": 857, "lr": 0.001} +{"train_loss": 0.029560165479779243, "train_loss_bc": 0.029560165479779243, "train_loss_llm": 0.0, "grad_norm": 0.03528855741024017, "global_step": 30861, "epoch": 857, "lr": 0.001} +{"train_loss": 0.027880314737558365, "train_loss_bc": 0.027880314737558365, "train_loss_llm": 0.0, "grad_norm": 0.05795939639210701, "global_step": 30862, "epoch": 857, "lr": 0.001} +{"train_loss": 0.028250358998775482, "train_loss_bc": 0.0282497089356184, "train_loss_llm": 6.496906280517578e-05, "grad_norm": 0.08598816394805908, "global_step": 30863, "epoch": 857, "lr": 0.001} +{"train_loss": 0.02829015627503395, "train_loss_bc": 0.02828914113342762, "train_loss_llm": 0.00010144710540771484, "grad_norm": 0.11131871491670609, "global_step": 30864, "epoch": 857, "lr": 0.001} +{"train_loss": 0.02811160497367382, "train_loss_bc": 0.028082534670829773, "train_loss_llm": 0.002907007932662964, "grad_norm": 0.02834618277847767, "global_step": 30865, "epoch": 857, "lr": 0.001} +{"train_loss": 0.029861770570278168, "train_loss_bc": 0.029861770570278168, "train_loss_llm": 0.0, "grad_norm": 0.04929487034678459, "global_step": 30866, "epoch": 857, "lr": 0.001} +{"train_loss": 0.029814977198839188, "train_loss_bc": 0.029814977198839188, "train_loss_llm": 0.0, "grad_norm": 0.06402119994163513, "global_step": 30867, "epoch": 857, "lr": 0.001} +{"train_loss": 0.027344267815351486, "train_loss_bc": 0.027344267815351486, "train_loss_llm": 0.0, "grad_norm": 0.08014536648988724, "global_step": 30868, "epoch": 857, "lr": 0.001} +{"train_loss": 0.033278994262218475, "train_loss_bc": 0.033278994262218475, "train_loss_llm": 0.0, "grad_norm": 0.039827410131692886, "global_step": 30869, "epoch": 857, "lr": 0.001} +{"train_loss": 0.029726270586252213, "train_loss_bc": 0.02972530759871006, "train_loss_llm": 9.626150131225586e-05, "grad_norm": 0.06938890367746353, "global_step": 30870, "epoch": 857, "lr": 0.001} +{"train_loss": 0.03221885487437248, "train_loss_bc": 0.03221885487437248, "train_loss_llm": 0.0, "grad_norm": 0.09386391192674637, "global_step": 30871, "epoch": 857, "lr": 0.001} +{"train_loss": 0.026848770678043365, "train_loss_bc": 0.02682928740978241, "train_loss_llm": 0.0019482672214508057, "grad_norm": 0.11456768214702606, "global_step": 30872, "epoch": 857, "lr": 0.001} +{"train_loss": 0.03829528018832207, "train_loss_bc": 0.03829528018832207, "train_loss_llm": 0.0, "grad_norm": 0.05586247146129608, "global_step": 30873, "epoch": 857, "lr": 0.001} +{"train_loss": 0.029728930443525314, "train_loss_bc": 0.029728930443525314, "train_loss_llm": 0.0, "grad_norm": 0.07524974644184113, "global_step": 30874, "epoch": 857, "lr": 0.001} +{"train_loss": 0.03231792896986008, "train_loss_bc": 0.03231792896986008, "train_loss_llm": 0.0, "grad_norm": 0.09980770945549011, "global_step": 30875, "epoch": 857, "lr": 0.001} +{"train_loss": 0.02603267878293991, "train_loss_bc": 0.02603267878293991, "train_loss_llm": 0.0, "grad_norm": 0.1148129254579544, "global_step": 30876, "epoch": 857, "lr": 0.001} +{"train_loss": 0.03067166917026043, "train_loss_bc": 0.03067166917026043, "train_loss_llm": 0.0, "grad_norm": 0.0399649515748024, "global_step": 30877, "epoch": 857, "lr": 0.001} +{"train_loss": 0.02999357506632805, "train_loss_bc": 0.02999357506632805, "train_loss_llm": 0.0, "grad_norm": 0.05374065414071083, "global_step": 30878, "epoch": 857, "lr": 0.001} +{"train_loss": 0.030204756185412407, "train_loss_bc": 0.030204756185412407, "train_loss_llm": 0.0, "grad_norm": 0.07187224924564362, "global_step": 30879, "epoch": 857, "lr": 0.001} +{"train_loss": 0.02860313467681408, "train_loss_bc": 0.02860313467681408, "train_loss_llm": 0.0, "grad_norm": 0.09333962947130203, "global_step": 30880, "epoch": 857, "lr": 0.001} +{"train_loss": 0.027437308803200722, "train_loss_bc": 0.027437308803200722, "train_loss_llm": 0.0, "grad_norm": 0.03465491533279419, "global_step": 30881, "epoch": 857, "lr": 0.001} +{"train_loss": 0.035794567316770554, "train_loss_bc": 0.035794567316770554, "train_loss_llm": 0.0, "grad_norm": 0.06557128578424454, "global_step": 30882, "epoch": 857, "lr": 0.001} +{"train_loss": 0.02569502778351307, "train_loss_bc": 0.025682227686047554, "train_loss_llm": 0.0012799501419067383, "grad_norm": 0.09307189285755157, "global_step": 30883, "epoch": 857, "lr": 0.001} +{"train_loss": 0.025893962010741234, "train_loss_bc": 0.025893962010741234, "train_loss_llm": 0.0, "grad_norm": 0.10805975645780563, "global_step": 30884, "epoch": 857, "lr": 0.001} +{"train_loss": 0.03287551552057266, "train_loss_bc": 0.03285997360944748, "train_loss_llm": 0.0015540122985839844, "grad_norm": 0.03865859657526016, "global_step": 30885, "epoch": 857, "lr": 0.001} +{"train_loss": 0.029913650825619698, "train_loss_bc": 0.029904760420322418, "train_loss_llm": 0.000889122486114502, "grad_norm": 0.055513136088848114, "global_step": 30886, "epoch": 857, "lr": 0.001} +{"train_loss": 0.029827850353386667, "train_loss_bc": 0.032485537230968475, "train_loss_llm": 0.0, "grad_norm": 0.07989779859781265, "global_step": 30887, "epoch": 857, "lr": 0.001, "val_loss": 0.10018917173147202} +{"train_loss": 0.030764000490307808, "train_loss_bc": 0.030764000490307808, "train_loss_llm": 0.0, "grad_norm": 0.09692920744419098, "global_step": 30888, "epoch": 858, "lr": 0.001} +{"train_loss": 0.02848968654870987, "train_loss_bc": 0.02848968654870987, "train_loss_llm": 0.0, "grad_norm": 0.0319070890545845, "global_step": 30889, "epoch": 858, "lr": 0.001} +{"train_loss": 0.03370223194360733, "train_loss_bc": 0.03370223194360733, "train_loss_llm": 0.0, "grad_norm": 0.06390408426523209, "global_step": 30890, "epoch": 858, "lr": 0.001} +{"train_loss": 0.029031334444880486, "train_loss_bc": 0.02901027724146843, "train_loss_llm": 0.0021057426929473877, "grad_norm": 0.08407445251941681, "global_step": 30891, "epoch": 858, "lr": 0.001} +{"train_loss": 0.027580607682466507, "train_loss_bc": 0.027580607682466507, "train_loss_llm": 0.0, "grad_norm": 0.1057688519358635, "global_step": 30892, "epoch": 858, "lr": 0.001} +{"train_loss": 0.031515028327703476, "train_loss_bc": 0.031515028327703476, "train_loss_llm": 0.0, "grad_norm": 0.037376224994659424, "global_step": 30893, "epoch": 858, "lr": 0.001} +{"train_loss": 0.025806209072470665, "train_loss_bc": 0.025806209072470665, "train_loss_llm": 0.0, "grad_norm": 0.05186041072010994, "global_step": 30894, "epoch": 858, "lr": 0.001} +{"train_loss": 0.02976168505847454, "train_loss_bc": 0.02976168505847454, "train_loss_llm": 0.0, "grad_norm": 0.06212053447961807, "global_step": 30895, "epoch": 858, "lr": 0.001} +{"train_loss": 0.02740894816815853, "train_loss_bc": 0.02740894816815853, "train_loss_llm": 0.0, "grad_norm": 0.07740576565265656, "global_step": 30896, "epoch": 858, "lr": 0.001} +{"train_loss": 0.023744164034724236, "train_loss_bc": 0.023744164034724236, "train_loss_llm": 0.0, "grad_norm": 0.03549434244632721, "global_step": 30897, "epoch": 858, "lr": 0.001} +{"train_loss": 0.03199795261025429, "train_loss_bc": 0.03199795261025429, "train_loss_llm": 0.0, "grad_norm": 0.05068429186940193, "global_step": 30898, "epoch": 858, "lr": 0.001} +{"train_loss": 0.02779623493552208, "train_loss_bc": 0.02779623493552208, "train_loss_llm": 0.0, "grad_norm": 0.07142292708158493, "global_step": 30899, "epoch": 858, "lr": 0.001} +{"train_loss": 0.028519010171294212, "train_loss_bc": 0.02851409651339054, "train_loss_llm": 0.0004913508892059326, "grad_norm": 0.07865019142627716, "global_step": 30900, "epoch": 858, "lr": 0.001} +{"train_loss": 0.024225812405347824, "train_loss_bc": 0.024225812405347824, "train_loss_llm": 0.0, "grad_norm": 0.032681796699762344, "global_step": 30901, "epoch": 858, "lr": 0.001} +{"train_loss": 0.03337138146162033, "train_loss_bc": 0.03337138146162033, "train_loss_llm": 0.0, "grad_norm": 0.06636665761470795, "global_step": 30902, "epoch": 858, "lr": 0.001} +{"train_loss": 0.031648650765419006, "train_loss_bc": 0.031648650765419006, "train_loss_llm": 0.0, "grad_norm": 0.09095042943954468, "global_step": 30903, "epoch": 858, "lr": 0.001} +{"train_loss": 0.03791683167219162, "train_loss_bc": 0.03791683167219162, "train_loss_llm": 0.0, "grad_norm": 0.12099311500787735, "global_step": 30904, "epoch": 858, "lr": 0.001} +{"train_loss": 0.028843499720096588, "train_loss_bc": 0.028843499720096588, "train_loss_llm": 0.0, "grad_norm": 0.03622770309448242, "global_step": 30905, "epoch": 858, "lr": 0.001} +{"train_loss": 0.02741563878953457, "train_loss_bc": 0.02741296775639057, "train_loss_llm": 0.0002671778202056885, "grad_norm": 0.05663697421550751, "global_step": 30906, "epoch": 858, "lr": 0.001} +{"train_loss": 0.027894776314496994, "train_loss_bc": 0.027894776314496994, "train_loss_llm": 0.0, "grad_norm": 0.07574660331010818, "global_step": 30907, "epoch": 858, "lr": 0.001} +{"train_loss": 0.024889644235372543, "train_loss_bc": 0.024881409481167793, "train_loss_llm": 0.0008234977722167969, "grad_norm": 0.09654895961284637, "global_step": 30908, "epoch": 858, "lr": 0.001} +{"train_loss": 0.02815569005906582, "train_loss_bc": 0.02815569005906582, "train_loss_llm": 0.0, "grad_norm": 0.038739364594221115, "global_step": 30909, "epoch": 858, "lr": 0.001} +{"train_loss": 0.03806959465146065, "train_loss_bc": 0.038051001727581024, "train_loss_llm": 0.0018591582775115967, "grad_norm": 0.07343839108943939, "global_step": 30910, "epoch": 858, "lr": 0.001} +{"train_loss": 0.03409324958920479, "train_loss_bc": 0.03409251570701599, "train_loss_llm": 7.322430610656738e-05, "grad_norm": 0.10293617099523544, "global_step": 30911, "epoch": 858, "lr": 0.001} +{"train_loss": 0.03603190556168556, "train_loss_bc": 0.03603190556168556, "train_loss_llm": 0.0, "grad_norm": 0.144878089427948, "global_step": 30912, "epoch": 858, "lr": 0.001} +{"train_loss": 0.02562805637717247, "train_loss_bc": 0.02562805637717247, "train_loss_llm": 0.0, "grad_norm": 0.037484265863895416, "global_step": 30913, "epoch": 858, "lr": 0.001} +{"train_loss": 0.03140171244740486, "train_loss_bc": 0.03140171244740486, "train_loss_llm": 0.0, "grad_norm": 0.06192305311560631, "global_step": 30914, "epoch": 858, "lr": 0.001} +{"train_loss": 0.030956164002418518, "train_loss_bc": 0.030953340232372284, "train_loss_llm": 0.0002823472023010254, "grad_norm": 0.08093246072530746, "global_step": 30915, "epoch": 858, "lr": 0.001} +{"train_loss": 0.03240569308400154, "train_loss_bc": 0.03240569308400154, "train_loss_llm": 0.0, "grad_norm": 0.09969418495893478, "global_step": 30916, "epoch": 858, "lr": 0.001} +{"train_loss": 0.030753446742892265, "train_loss_bc": 0.030753446742892265, "train_loss_llm": 0.0, "grad_norm": 0.04339239001274109, "global_step": 30917, "epoch": 858, "lr": 0.001} +{"train_loss": 0.034016337245702744, "train_loss_bc": 0.034016337245702744, "train_loss_llm": 0.0, "grad_norm": 0.07561032474040985, "global_step": 30918, "epoch": 858, "lr": 0.001} +{"train_loss": 0.032565828412771225, "train_loss_bc": 0.032565828412771225, "train_loss_llm": 0.0, "grad_norm": 0.09810730069875717, "global_step": 30919, "epoch": 858, "lr": 0.001} +{"train_loss": 0.02859993651509285, "train_loss_bc": 0.02859993651509285, "train_loss_llm": 0.0, "grad_norm": 0.11972317099571228, "global_step": 30920, "epoch": 858, "lr": 0.001} +{"train_loss": 0.027224760502576828, "train_loss_bc": 0.027224760502576828, "train_loss_llm": 0.0, "grad_norm": 0.04794079065322876, "global_step": 30921, "epoch": 858, "lr": 0.001} +{"train_loss": 0.02436138316988945, "train_loss_bc": 0.02436138316988945, "train_loss_llm": 0.0, "grad_norm": 0.07231063395738602, "global_step": 30922, "epoch": 858, "lr": 0.001} +{"train_loss": 0.029832375184115436, "train_loss_bc": 0.027378419414162636, "train_loss_llm": 0.0, "grad_norm": 0.08870852738618851, "global_step": 30923, "epoch": 858, "lr": 0.001, "val_loss": 0.101834237575531} +{"train_loss": 0.027824383229017258, "train_loss_bc": 0.027824383229017258, "train_loss_llm": 0.0, "grad_norm": 0.10832925885915756, "global_step": 30924, "epoch": 859, "lr": 0.001} +{"train_loss": 0.028890229761600494, "train_loss_bc": 0.028890229761600494, "train_loss_llm": 0.0, "grad_norm": 0.03780807554721832, "global_step": 30925, "epoch": 859, "lr": 0.001} +{"train_loss": 0.03004714660346508, "train_loss_bc": 0.03004714660346508, "train_loss_llm": 0.0, "grad_norm": 0.06372927129268646, "global_step": 30926, "epoch": 859, "lr": 0.001} +{"train_loss": 0.02953513152897358, "train_loss_bc": 0.029531484469771385, "train_loss_llm": 0.0003647804260253906, "grad_norm": 0.09121059626340866, "global_step": 30927, "epoch": 859, "lr": 0.001} +{"train_loss": 0.026775075122714043, "train_loss_bc": 0.0267743282020092, "train_loss_llm": 7.477402687072754e-05, "grad_norm": 0.12279263138771057, "global_step": 30928, "epoch": 859, "lr": 0.001} +{"train_loss": 0.02912602759897709, "train_loss_bc": 0.029120394960045815, "train_loss_llm": 0.0005632936954498291, "grad_norm": 0.04396836832165718, "global_step": 30929, "epoch": 859, "lr": 0.001} +{"train_loss": 0.027677025645971298, "train_loss_bc": 0.027677025645971298, "train_loss_llm": 0.0, "grad_norm": 0.06843332946300507, "global_step": 30930, "epoch": 859, "lr": 0.001} +{"train_loss": 0.03052210994064808, "train_loss_bc": 0.03052210994064808, "train_loss_llm": 0.0, "grad_norm": 0.08122532814741135, "global_step": 30931, "epoch": 859, "lr": 0.001} +{"train_loss": 0.031198590993881226, "train_loss_bc": 0.031198590993881226, "train_loss_llm": 0.0, "grad_norm": 0.1005021184682846, "global_step": 30932, "epoch": 859, "lr": 0.001} +{"train_loss": 0.027290701866149902, "train_loss_bc": 0.027286743745207787, "train_loss_llm": 0.00039574503898620605, "grad_norm": 0.033964041620492935, "global_step": 30933, "epoch": 859, "lr": 0.001} +{"train_loss": 0.026554083451628685, "train_loss_bc": 0.026554083451628685, "train_loss_llm": 0.0, "grad_norm": 0.049813445657491684, "global_step": 30934, "epoch": 859, "lr": 0.001} +{"train_loss": 0.031351834535598755, "train_loss_bc": 0.031351834535598755, "train_loss_llm": 0.0, "grad_norm": 0.07366136461496353, "global_step": 30935, "epoch": 859, "lr": 0.001} +{"train_loss": 0.028894100338220596, "train_loss_bc": 0.028890391811728477, "train_loss_llm": 0.0003707706928253174, "grad_norm": 0.08978577703237534, "global_step": 30936, "epoch": 859, "lr": 0.001} +{"train_loss": 0.02854318916797638, "train_loss_bc": 0.02854318916797638, "train_loss_llm": 0.0, "grad_norm": 0.039997201412916183, "global_step": 30937, "epoch": 859, "lr": 0.001} +{"train_loss": 0.03505687788128853, "train_loss_bc": 0.035056814551353455, "train_loss_llm": 6.377696990966797e-06, "grad_norm": 0.06552845239639282, "global_step": 30938, "epoch": 859, "lr": 0.001} +{"train_loss": 0.03003542497754097, "train_loss_bc": 0.03003542497754097, "train_loss_llm": 0.0, "grad_norm": 0.08910606056451797, "global_step": 30939, "epoch": 859, "lr": 0.001} +{"train_loss": 0.0302286297082901, "train_loss_bc": 0.0302286297082901, "train_loss_llm": 0.0, "grad_norm": 0.12339213490486145, "global_step": 30940, "epoch": 859, "lr": 0.001} +{"train_loss": 0.028531935065984726, "train_loss_bc": 0.028531935065984726, "train_loss_llm": 0.0, "grad_norm": 0.037008464336395264, "global_step": 30941, "epoch": 859, "lr": 0.001} +{"train_loss": 0.03152086213231087, "train_loss_bc": 0.03152086213231087, "train_loss_llm": 0.0, "grad_norm": 0.07279228419065475, "global_step": 30942, "epoch": 859, "lr": 0.001} +{"train_loss": 0.032239027321338654, "train_loss_bc": 0.032239027321338654, "train_loss_llm": 0.0, "grad_norm": 0.10452505946159363, "global_step": 30943, "epoch": 859, "lr": 0.001} +{"train_loss": 0.033015552908182144, "train_loss_bc": 0.03300952538847923, "train_loss_llm": 0.0006028413772583008, "grad_norm": 0.12658856809139252, "global_step": 30944, "epoch": 859, "lr": 0.001} +{"train_loss": 0.027179883792996407, "train_loss_bc": 0.027166074141860008, "train_loss_llm": 0.0013809502124786377, "grad_norm": 0.04144272953271866, "global_step": 30945, "epoch": 859, "lr": 0.001} +{"train_loss": 0.03124801255762577, "train_loss_bc": 0.03124801255762577, "train_loss_llm": 0.0, "grad_norm": 0.068742536008358, "global_step": 30946, "epoch": 859, "lr": 0.001} +{"train_loss": 0.03378034755587578, "train_loss_bc": 0.03378034755587578, "train_loss_llm": 0.0, "grad_norm": 0.09094858169555664, "global_step": 30947, "epoch": 859, "lr": 0.001} +{"train_loss": 0.035437677055597305, "train_loss_bc": 0.035437677055597305, "train_loss_llm": 0.0, "grad_norm": 0.12972098588943481, "global_step": 30948, "epoch": 859, "lr": 0.001} +{"train_loss": 0.02664572186768055, "train_loss_bc": 0.02664572186768055, "train_loss_llm": 0.0, "grad_norm": 0.034003596752882004, "global_step": 30949, "epoch": 859, "lr": 0.001} +{"train_loss": 0.03200623020529747, "train_loss_bc": 0.03200623020529747, "train_loss_llm": 0.0, "grad_norm": 0.05700942873954773, "global_step": 30950, "epoch": 859, "lr": 0.001} +{"train_loss": 0.030525511130690575, "train_loss_bc": 0.030525511130690575, "train_loss_llm": 0.0, "grad_norm": 0.0802740529179573, "global_step": 30951, "epoch": 859, "lr": 0.001} +{"train_loss": 0.03395123407244682, "train_loss_bc": 0.03395123407244682, "train_loss_llm": 0.0, "grad_norm": 0.11139093339443207, "global_step": 30952, "epoch": 859, "lr": 0.001} +{"train_loss": 0.0268907081335783, "train_loss_bc": 0.02687620371580124, "train_loss_llm": 0.0014504194259643555, "grad_norm": 0.03631476312875748, "global_step": 30953, "epoch": 859, "lr": 0.001} +{"train_loss": 0.026249926537275314, "train_loss_bc": 0.02622971311211586, "train_loss_llm": 0.0020213425159454346, "grad_norm": 0.05831495672464371, "global_step": 30954, "epoch": 859, "lr": 0.001} +{"train_loss": 0.030661366879940033, "train_loss_bc": 0.030661366879940033, "train_loss_llm": 0.0, "grad_norm": 0.07500392943620682, "global_step": 30955, "epoch": 859, "lr": 0.001} +{"train_loss": 0.026864392682909966, "train_loss_bc": 0.026864392682909966, "train_loss_llm": 0.0, "grad_norm": 0.09201617538928986, "global_step": 30956, "epoch": 859, "lr": 0.001} +{"train_loss": 0.02927764318883419, "train_loss_bc": 0.02927764318883419, "train_loss_llm": 0.0, "grad_norm": 0.03577537089586258, "global_step": 30957, "epoch": 859, "lr": 0.001} +{"train_loss": 0.029253531247377396, "train_loss_bc": 0.029253531247377396, "train_loss_llm": 0.0, "grad_norm": 0.050193075090646744, "global_step": 30958, "epoch": 859, "lr": 0.001} +{"train_loss": 0.029822644622375567, "train_loss_bc": 0.028772754594683647, "train_loss_llm": 0.0012325644493103027, "grad_norm": 0.05590631067752838, "global_step": 30959, "epoch": 859, "lr": 0.001, "val_loss": 0.1015549749135971} +{"train_loss": 0.027471056208014488, "train_loss_bc": 0.027471056208014488, "train_loss_llm": 0.0, "grad_norm": 0.07065064460039139, "global_step": 30960, "epoch": 860, "lr": 0.001} +{"train_loss": 0.029187146574258804, "train_loss_bc": 0.029186373576521873, "train_loss_llm": 7.721781730651855e-05, "grad_norm": 0.036231543868780136, "global_step": 30961, "epoch": 860, "lr": 0.001} +{"train_loss": 0.02612965553998947, "train_loss_bc": 0.02612965553998947, "train_loss_llm": 0.0, "grad_norm": 0.059452470391988754, "global_step": 30962, "epoch": 860, "lr": 0.001} +{"train_loss": 0.030291013419628143, "train_loss_bc": 0.030291013419628143, "train_loss_llm": 0.0, "grad_norm": 0.0877176821231842, "global_step": 30963, "epoch": 860, "lr": 0.001} +{"train_loss": 0.030697718262672424, "train_loss_bc": 0.030655238777399063, "train_loss_llm": 0.004247874021530151, "grad_norm": 0.10568942874670029, "global_step": 30964, "epoch": 860, "lr": 0.001} +{"train_loss": 0.03229022026062012, "train_loss_bc": 0.03227798640727997, "train_loss_llm": 0.0012233555316925049, "grad_norm": 0.04127627611160278, "global_step": 30965, "epoch": 860, "lr": 0.001} +{"train_loss": 0.02702987752854824, "train_loss_bc": 0.02702987752854824, "train_loss_llm": 0.0, "grad_norm": 0.06318799406290054, "global_step": 30966, "epoch": 860, "lr": 0.001} +{"train_loss": 0.036772023886442184, "train_loss_bc": 0.036772023886442184, "train_loss_llm": 0.0, "grad_norm": 0.09294843673706055, "global_step": 30967, "epoch": 860, "lr": 0.001} +{"train_loss": 0.02618345618247986, "train_loss_bc": 0.026165351271629333, "train_loss_llm": 0.0018104016780853271, "grad_norm": 0.11583535373210907, "global_step": 30968, "epoch": 860, "lr": 0.001} +{"train_loss": 0.025838924571871758, "train_loss_bc": 0.025838924571871758, "train_loss_llm": 0.0, "grad_norm": 0.04823526367545128, "global_step": 30969, "epoch": 860, "lr": 0.001} +{"train_loss": 0.03275496885180473, "train_loss_bc": 0.03275496885180473, "train_loss_llm": 0.0, "grad_norm": 0.08286014199256897, "global_step": 30970, "epoch": 860, "lr": 0.001} +{"train_loss": 0.03450068086385727, "train_loss_bc": 0.03450068086385727, "train_loss_llm": 0.0, "grad_norm": 0.11152959614992142, "global_step": 30971, "epoch": 860, "lr": 0.001} +{"train_loss": 0.029944635927677155, "train_loss_bc": 0.029944635927677155, "train_loss_llm": 0.0, "grad_norm": 0.14392641186714172, "global_step": 30972, "epoch": 860, "lr": 0.001} +{"train_loss": 0.03255712240934372, "train_loss_bc": 0.03255712240934372, "train_loss_llm": 0.0, "grad_norm": 0.04201703146100044, "global_step": 30973, "epoch": 860, "lr": 0.001} +{"train_loss": 0.033380795270204544, "train_loss_bc": 0.033380795270204544, "train_loss_llm": 0.0, "grad_norm": 0.05938761308789253, "global_step": 30974, "epoch": 860, "lr": 0.001} +{"train_loss": 0.02884567342698574, "train_loss_bc": 0.02884567342698574, "train_loss_llm": 0.0, "grad_norm": 0.08349310606718063, "global_step": 30975, "epoch": 860, "lr": 0.001} +{"train_loss": 0.024194756522774696, "train_loss_bc": 0.024194756522774696, "train_loss_llm": 0.0, "grad_norm": 0.09703817218542099, "global_step": 30976, "epoch": 860, "lr": 0.001} +{"train_loss": 0.028125880286097527, "train_loss_bc": 0.028125880286097527, "train_loss_llm": 0.0, "grad_norm": 0.03110966831445694, "global_step": 30977, "epoch": 860, "lr": 0.001} +{"train_loss": 0.029670266434550285, "train_loss_bc": 0.029670266434550285, "train_loss_llm": 0.0, "grad_norm": 0.050454508513212204, "global_step": 30978, "epoch": 860, "lr": 0.001} +{"train_loss": 0.029017850756645203, "train_loss_bc": 0.029017850756645203, "train_loss_llm": 0.0, "grad_norm": 0.07338887453079224, "global_step": 30979, "epoch": 860, "lr": 0.001} +{"train_loss": 0.02785434015095234, "train_loss_bc": 0.02785434015095234, "train_loss_llm": 0.0, "grad_norm": 0.08713853359222412, "global_step": 30980, "epoch": 860, "lr": 0.001} +{"train_loss": 0.028470179066061974, "train_loss_bc": 0.028470179066061974, "train_loss_llm": 0.0, "grad_norm": 0.03442356362938881, "global_step": 30981, "epoch": 860, "lr": 0.001} +{"train_loss": 0.03502527251839638, "train_loss_bc": 0.03502527251839638, "train_loss_llm": 0.0, "grad_norm": 0.05464037135243416, "global_step": 30982, "epoch": 860, "lr": 0.001} +{"train_loss": 0.033026985824108124, "train_loss_bc": 0.033026985824108124, "train_loss_llm": 0.0, "grad_norm": 0.07507052272558212, "global_step": 30983, "epoch": 860, "lr": 0.001} +{"train_loss": 0.028939027339220047, "train_loss_bc": 0.028928088024258614, "train_loss_llm": 0.0010940134525299072, "grad_norm": 0.08744709193706512, "global_step": 30984, "epoch": 860, "lr": 0.001} +{"train_loss": 0.029873697087168694, "train_loss_bc": 0.029873697087168694, "train_loss_llm": 0.0, "grad_norm": 0.03823113441467285, "global_step": 30985, "epoch": 860, "lr": 0.001} +{"train_loss": 0.028425166383385658, "train_loss_bc": 0.028425166383385658, "train_loss_llm": 0.0, "grad_norm": 0.061004240065813065, "global_step": 30986, "epoch": 860, "lr": 0.001} +{"train_loss": 0.02821165695786476, "train_loss_bc": 0.02818985842168331, "train_loss_llm": 0.0021798908710479736, "grad_norm": 0.08424821496009827, "global_step": 30987, "epoch": 860, "lr": 0.001} +{"train_loss": 0.03409566730260849, "train_loss_bc": 0.03409443795681, "train_loss_llm": 0.0001227855682373047, "grad_norm": 0.11187028884887695, "global_step": 30988, "epoch": 860, "lr": 0.001} +{"train_loss": 0.02995407208800316, "train_loss_bc": 0.02995407208800316, "train_loss_llm": 0.0, "grad_norm": 0.041232410818338394, "global_step": 30989, "epoch": 860, "lr": 0.001} +{"train_loss": 0.03320242837071419, "train_loss_bc": 0.03320242837071419, "train_loss_llm": 0.0, "grad_norm": 0.0669296458363533, "global_step": 30990, "epoch": 860, "lr": 0.001} +{"train_loss": 0.025695940479636192, "train_loss_bc": 0.025695940479636192, "train_loss_llm": 0.0, "grad_norm": 0.09423216432332993, "global_step": 30991, "epoch": 860, "lr": 0.001} +{"train_loss": 0.031735535711050034, "train_loss_bc": 0.031735535711050034, "train_loss_llm": 0.0, "grad_norm": 0.12182074040174484, "global_step": 30992, "epoch": 860, "lr": 0.001} +{"train_loss": 0.030275704339146614, "train_loss_bc": 0.030275704339146614, "train_loss_llm": 0.0, "grad_norm": 0.038430411368608475, "global_step": 30993, "epoch": 860, "lr": 0.001} +{"train_loss": 0.02733621746301651, "train_loss_bc": 0.02733621746301651, "train_loss_llm": 0.0, "grad_norm": 0.07140520960092545, "global_step": 30994, "epoch": 860, "lr": 0.001} +{"train_loss": 0.02985557008327709, "train_loss_bc": 0.027794908732175827, "train_loss_llm": 0.0, "grad_norm": 0.09016310423612595, "global_step": 30995, "epoch": 860, "lr": 0.001, "train/cumulative_reward": 133.0902308164811, "train/mean_score": 12.748034063990826, "train/success_rate": 0.2, "test/cumulative_reward": 22.322162516503052, "test/mean_score": 5.794107922949368, "test/success_rate": 0.14, "val_loss": 0.10242179036140442, "train_action_mse_error": 0.0329463854432106} +{"train_loss": 0.02818141132593155, "train_loss_bc": 0.02818141132593155, "train_loss_llm": 0.0, "grad_norm": 0.11293341964483261, "global_step": 30996, "epoch": 861, "lr": 0.001} +{"train_loss": 0.030391747131943703, "train_loss_bc": 0.030391747131943703, "train_loss_llm": 0.0, "grad_norm": 0.046720996499061584, "global_step": 30997, "epoch": 861, "lr": 0.001} +{"train_loss": 0.026642439886927605, "train_loss_bc": 0.026642439886927605, "train_loss_llm": 0.0, "grad_norm": 0.0635734498500824, "global_step": 30998, "epoch": 861, "lr": 0.001} +{"train_loss": 0.025597024708986282, "train_loss_bc": 0.025597024708986282, "train_loss_llm": 0.0, "grad_norm": 0.08847948163747787, "global_step": 30999, "epoch": 861, "lr": 0.001} +{"train_loss": 0.03280456364154816, "train_loss_bc": 0.03280456364154816, "train_loss_llm": 0.0, "grad_norm": 0.1154751405119896, "global_step": 31000, "epoch": 861, "lr": 0.001} +{"train_loss": 0.027928048744797707, "train_loss_bc": 0.02790782041847706, "train_loss_llm": 0.0020228028297424316, "grad_norm": 0.03760894387960434, "global_step": 31001, "epoch": 861, "lr": 0.001} +{"train_loss": 0.029975853860378265, "train_loss_bc": 0.029975853860378265, "train_loss_llm": 0.0, "grad_norm": 0.04576638713479042, "global_step": 31002, "epoch": 861, "lr": 0.001} +{"train_loss": 0.03014749474823475, "train_loss_bc": 0.03014749474823475, "train_loss_llm": 0.0, "grad_norm": 0.06469687819480896, "global_step": 31003, "epoch": 861, "lr": 0.001} +{"train_loss": 0.029452938586473465, "train_loss_bc": 0.029436815530061722, "train_loss_llm": 0.0016122162342071533, "grad_norm": 0.09851188212633133, "global_step": 31004, "epoch": 861, "lr": 0.001} +{"train_loss": 0.029330439865589142, "train_loss_bc": 0.02930338867008686, "train_loss_llm": 0.002705127000808716, "grad_norm": 0.034195415675640106, "global_step": 31005, "epoch": 861, "lr": 0.001} +{"train_loss": 0.030322687700390816, "train_loss_bc": 0.030322687700390816, "train_loss_llm": 0.0, "grad_norm": 0.06723449379205704, "global_step": 31006, "epoch": 861, "lr": 0.001} +{"train_loss": 0.025532297790050507, "train_loss_bc": 0.025532297790050507, "train_loss_llm": 0.0, "grad_norm": 0.09191706776618958, "global_step": 31007, "epoch": 861, "lr": 0.001} +{"train_loss": 0.024901101365685463, "train_loss_bc": 0.024901101365685463, "train_loss_llm": 0.0, "grad_norm": 0.10405348241329193, "global_step": 31008, "epoch": 861, "lr": 0.001} +{"train_loss": 0.03294520825147629, "train_loss_bc": 0.03293849155306816, "train_loss_llm": 0.00067177414894104, "grad_norm": 0.0420062355697155, "global_step": 31009, "epoch": 861, "lr": 0.001} +{"train_loss": 0.03324475884437561, "train_loss_bc": 0.03324475884437561, "train_loss_llm": 0.0, "grad_norm": 0.06951474398374557, "global_step": 31010, "epoch": 861, "lr": 0.001} +{"train_loss": 0.026419108733534813, "train_loss_bc": 0.026419108733534813, "train_loss_llm": 0.0, "grad_norm": 0.09749235212802887, "global_step": 31011, "epoch": 861, "lr": 0.001} +{"train_loss": 0.028039397671818733, "train_loss_bc": 0.028039397671818733, "train_loss_llm": 0.0, "grad_norm": 0.12673406302928925, "global_step": 31012, "epoch": 861, "lr": 0.001} +{"train_loss": 0.027989324182271957, "train_loss_bc": 0.02798847109079361, "train_loss_llm": 8.535385131835938e-05, "grad_norm": 0.033797767013311386, "global_step": 31013, "epoch": 861, "lr": 0.001} +{"train_loss": 0.03500882163643837, "train_loss_bc": 0.035006675869226456, "train_loss_llm": 0.00021460652351379395, "grad_norm": 0.07049856334924698, "global_step": 31014, "epoch": 861, "lr": 0.001} +{"train_loss": 0.0332506000995636, "train_loss_bc": 0.0332506000995636, "train_loss_llm": 0.0, "grad_norm": 0.09984885156154633, "global_step": 31015, "epoch": 861, "lr": 0.001} +{"train_loss": 0.03222674876451492, "train_loss_bc": 0.03222592920064926, "train_loss_llm": 8.20457935333252e-05, "grad_norm": 0.12395491451025009, "global_step": 31016, "epoch": 861, "lr": 0.001} +{"train_loss": 0.033520590513944626, "train_loss_bc": 0.033520590513944626, "train_loss_llm": 0.0, "grad_norm": 0.04708270728588104, "global_step": 31017, "epoch": 861, "lr": 0.001} +{"train_loss": 0.03240926191210747, "train_loss_bc": 0.03240926191210747, "train_loss_llm": 0.0, "grad_norm": 0.08940736949443817, "global_step": 31018, "epoch": 861, "lr": 0.001} +{"train_loss": 0.028736216947436333, "train_loss_bc": 0.028719443827867508, "train_loss_llm": 0.0016772449016571045, "grad_norm": 0.1223800852894783, "global_step": 31019, "epoch": 861, "lr": 0.001} +{"train_loss": 0.028779180720448494, "train_loss_bc": 0.028779180720448494, "train_loss_llm": 0.0, "grad_norm": 0.15068595111370087, "global_step": 31020, "epoch": 861, "lr": 0.001} +{"train_loss": 0.02632344327867031, "train_loss_bc": 0.02631630375981331, "train_loss_llm": 0.0007140040397644043, "grad_norm": 0.033908627927303314, "global_step": 31021, "epoch": 861, "lr": 0.001} +{"train_loss": 0.030405322089791298, "train_loss_bc": 0.03039908967912197, "train_loss_llm": 0.0006231963634490967, "grad_norm": 0.056185755878686905, "global_step": 31022, "epoch": 861, "lr": 0.001} +{"train_loss": 0.03441601246595383, "train_loss_bc": 0.03441601246595383, "train_loss_llm": 0.0, "grad_norm": 0.0909733846783638, "global_step": 31023, "epoch": 861, "lr": 0.001} +{"train_loss": 0.03263181075453758, "train_loss_bc": 0.03263181075453758, "train_loss_llm": 0.0, "grad_norm": 0.1146402582526207, "global_step": 31024, "epoch": 861, "lr": 0.001} +{"train_loss": 0.03276883810758591, "train_loss_bc": 0.03276883810758591, "train_loss_llm": 0.0, "grad_norm": 0.051802005618810654, "global_step": 31025, "epoch": 861, "lr": 0.001} +{"train_loss": 0.033581286668777466, "train_loss_bc": 0.03356800973415375, "train_loss_llm": 0.0013275742530822754, "grad_norm": 0.08730155229568481, "global_step": 31026, "epoch": 861, "lr": 0.001} +{"train_loss": 0.02805602177977562, "train_loss_bc": 0.02805602177977562, "train_loss_llm": 0.0, "grad_norm": 0.10743795335292816, "global_step": 31027, "epoch": 861, "lr": 0.001} +{"train_loss": 0.030992472544312477, "train_loss_bc": 0.030988313257694244, "train_loss_llm": 0.0004158616065979004, "grad_norm": 0.13592733442783356, "global_step": 31028, "epoch": 861, "lr": 0.001} +{"train_loss": 0.024781564250588417, "train_loss_bc": 0.024781564250588417, "train_loss_llm": 0.0, "grad_norm": 0.03713899105787277, "global_step": 31029, "epoch": 861, "lr": 0.001} +{"train_loss": 0.027423053979873657, "train_loss_bc": 0.027423053979873657, "train_loss_llm": 0.0, "grad_norm": 0.06670928746461868, "global_step": 31030, "epoch": 861, "lr": 0.001} +{"train_loss": 0.030064253136515617, "train_loss_bc": 0.037115201354026794, "train_loss_llm": 0.00408172607421875, "grad_norm": 0.10728908330202103, "global_step": 31031, "epoch": 861, "lr": 0.001, "val_loss": 0.10035919398069382} +{"train_loss": 0.03298456594347954, "train_loss_bc": 0.03298456594347954, "train_loss_llm": 0.0, "grad_norm": 0.12812761962413788, "global_step": 31032, "epoch": 862, "lr": 0.001} +{"train_loss": 0.029808461666107178, "train_loss_bc": 0.029808461666107178, "train_loss_llm": 0.0, "grad_norm": 0.03910878300666809, "global_step": 31033, "epoch": 862, "lr": 0.001} +{"train_loss": 0.02736184000968933, "train_loss_bc": 0.02736184000968933, "train_loss_llm": 0.0, "grad_norm": 0.05514764040708542, "global_step": 31034, "epoch": 862, "lr": 0.001} +{"train_loss": 0.031037328764796257, "train_loss_bc": 0.031037328764796257, "train_loss_llm": 0.0, "grad_norm": 0.08309619873762131, "global_step": 31035, "epoch": 862, "lr": 0.001} +{"train_loss": 0.03292322903871536, "train_loss_bc": 0.03292322903871536, "train_loss_llm": 0.0, "grad_norm": 0.1112174242734909, "global_step": 31036, "epoch": 862, "lr": 0.001} +{"train_loss": 0.03145971894264221, "train_loss_bc": 0.03145971894264221, "train_loss_llm": 0.0, "grad_norm": 0.04425295814871788, "global_step": 31037, "epoch": 862, "lr": 0.001} +{"train_loss": 0.025604363530874252, "train_loss_bc": 0.025604363530874252, "train_loss_llm": 0.0, "grad_norm": 0.08243291079998016, "global_step": 31038, "epoch": 862, "lr": 0.001} +{"train_loss": 0.032737113535404205, "train_loss_bc": 0.03273195028305054, "train_loss_llm": 0.0005164146423339844, "grad_norm": 0.111624576151371, "global_step": 31039, "epoch": 862, "lr": 0.001} +{"train_loss": 0.031237006187438965, "train_loss_bc": 0.031231725588440895, "train_loss_llm": 0.0005279779434204102, "grad_norm": 0.1412695199251175, "global_step": 31040, "epoch": 862, "lr": 0.001} +{"train_loss": 0.03404907509684563, "train_loss_bc": 0.03404907509684563, "train_loss_llm": 0.0, "grad_norm": 0.04558005928993225, "global_step": 31041, "epoch": 862, "lr": 0.001} +{"train_loss": 0.028231734409928322, "train_loss_bc": 0.028231734409928322, "train_loss_llm": 0.0, "grad_norm": 0.08040620386600494, "global_step": 31042, "epoch": 862, "lr": 0.001} +{"train_loss": 0.029607636854052544, "train_loss_bc": 0.029607636854052544, "train_loss_llm": 0.0, "grad_norm": 0.1171000525355339, "global_step": 31043, "epoch": 862, "lr": 0.001} +{"train_loss": 0.027863409370183945, "train_loss_bc": 0.027863409370183945, "train_loss_llm": 0.0, "grad_norm": 0.14229364693164825, "global_step": 31044, "epoch": 862, "lr": 0.001} +{"train_loss": 0.032099153846502304, "train_loss_bc": 0.032099153846502304, "train_loss_llm": 0.0, "grad_norm": 0.0407511442899704, "global_step": 31045, "epoch": 862, "lr": 0.001} +{"train_loss": 0.02718360535800457, "train_loss_bc": 0.02718360535800457, "train_loss_llm": 0.0, "grad_norm": 0.07657017558813095, "global_step": 31046, "epoch": 862, "lr": 0.001} +{"train_loss": 0.028494177386164665, "train_loss_bc": 0.028494177386164665, "train_loss_llm": 0.0, "grad_norm": 0.10084184259176254, "global_step": 31047, "epoch": 862, "lr": 0.001} +{"train_loss": 0.029048219323158264, "train_loss_bc": 0.029048219323158264, "train_loss_llm": 0.0, "grad_norm": 0.13084489107131958, "global_step": 31048, "epoch": 862, "lr": 0.001} +{"train_loss": 0.03021681122481823, "train_loss_bc": 0.03021681122481823, "train_loss_llm": 0.0, "grad_norm": 0.03847365453839302, "global_step": 31049, "epoch": 862, "lr": 0.001} +{"train_loss": 0.03233052045106888, "train_loss_bc": 0.03233052045106888, "train_loss_llm": 0.0, "grad_norm": 0.06612271815538406, "global_step": 31050, "epoch": 862, "lr": 0.001} +{"train_loss": 0.029624011367559433, "train_loss_bc": 0.029624011367559433, "train_loss_llm": 0.0, "grad_norm": 0.09687209129333496, "global_step": 31051, "epoch": 862, "lr": 0.001} +{"train_loss": 0.02530953288078308, "train_loss_bc": 0.02530953288078308, "train_loss_llm": 0.0, "grad_norm": 0.1074509248137474, "global_step": 31052, "epoch": 862, "lr": 0.001} +{"train_loss": 0.029080186039209366, "train_loss_bc": 0.029080186039209366, "train_loss_llm": 0.0, "grad_norm": 0.0415484718978405, "global_step": 31053, "epoch": 862, "lr": 0.001} +{"train_loss": 0.02956816554069519, "train_loss_bc": 0.029550746083259583, "train_loss_llm": 0.0017419755458831787, "grad_norm": 0.08126764744520187, "global_step": 31054, "epoch": 862, "lr": 0.001} +{"train_loss": 0.02620844542980194, "train_loss_bc": 0.02620844542980194, "train_loss_llm": 0.0, "grad_norm": 0.08947043120861053, "global_step": 31055, "epoch": 862, "lr": 0.001} +{"train_loss": 0.030702920630574226, "train_loss_bc": 0.030702920630574226, "train_loss_llm": 0.0, "grad_norm": 0.11048170924186707, "global_step": 31056, "epoch": 862, "lr": 0.001} +{"train_loss": 0.02734837308526039, "train_loss_bc": 0.02734837308526039, "train_loss_llm": 0.0, "grad_norm": 0.031228749081492424, "global_step": 31057, "epoch": 862, "lr": 0.001} +{"train_loss": 0.032809093594551086, "train_loss_bc": 0.03280030936002731, "train_loss_llm": 0.0008784830570220947, "grad_norm": 0.061405736953020096, "global_step": 31058, "epoch": 862, "lr": 0.001} +{"train_loss": 0.029835548251867294, "train_loss_bc": 0.029835548251867294, "train_loss_llm": 0.0, "grad_norm": 0.08300131559371948, "global_step": 31059, "epoch": 862, "lr": 0.001} +{"train_loss": 0.024978483095765114, "train_loss_bc": 0.024978483095765114, "train_loss_llm": 0.0, "grad_norm": 0.10496348887681961, "global_step": 31060, "epoch": 862, "lr": 0.001} +{"train_loss": 0.03621029108762741, "train_loss_bc": 0.03621029108762741, "train_loss_llm": 0.0, "grad_norm": 0.050558969378471375, "global_step": 31061, "epoch": 862, "lr": 0.001} +{"train_loss": 0.03177567571401596, "train_loss_bc": 0.03177567571401596, "train_loss_llm": 0.0, "grad_norm": 0.08249211311340332, "global_step": 31062, "epoch": 862, "lr": 0.001} +{"train_loss": 0.034916870296001434, "train_loss_bc": 0.03489495441317558, "train_loss_llm": 0.0021916627883911133, "grad_norm": 0.11039622128009796, "global_step": 31063, "epoch": 862, "lr": 0.001} +{"train_loss": 0.027921954169869423, "train_loss_bc": 0.0279057789593935, "train_loss_llm": 0.001617521047592163, "grad_norm": 0.12641437351703644, "global_step": 31064, "epoch": 862, "lr": 0.001} +{"train_loss": 0.02840331196784973, "train_loss_bc": 0.02840331196784973, "train_loss_llm": 0.0, "grad_norm": 0.042053647339344025, "global_step": 31065, "epoch": 862, "lr": 0.001} +{"train_loss": 0.0373646579682827, "train_loss_bc": 0.0373646579682827, "train_loss_llm": 0.0, "grad_norm": 0.08073648810386658, "global_step": 31066, "epoch": 862, "lr": 0.001} +{"train_loss": 0.030189964558101363, "train_loss_bc": 0.030473874881863594, "train_loss_llm": 0.002935647964477539, "grad_norm": 0.10342420637607574, "global_step": 31067, "epoch": 862, "lr": 0.001, "val_loss": 0.10304731130599976} +{"train_loss": 0.03367135673761368, "train_loss_bc": 0.03364569693803787, "train_loss_llm": 0.002565830945968628, "grad_norm": 0.1458347588777542, "global_step": 31068, "epoch": 863, "lr": 0.001} +{"train_loss": 0.02767101861536503, "train_loss_bc": 0.027665603905916214, "train_loss_llm": 0.0005414187908172607, "grad_norm": 0.04559064283967018, "global_step": 31069, "epoch": 863, "lr": 0.001} +{"train_loss": 0.03397886082530022, "train_loss_bc": 0.03397886082530022, "train_loss_llm": 0.0, "grad_norm": 0.07597994804382324, "global_step": 31070, "epoch": 863, "lr": 0.001} +{"train_loss": 0.028646226972341537, "train_loss_bc": 0.028646226972341537, "train_loss_llm": 0.0, "grad_norm": 0.09867949038743973, "global_step": 31071, "epoch": 863, "lr": 0.001} +{"train_loss": 0.027874423190951347, "train_loss_bc": 0.027874423190951347, "train_loss_llm": 0.0, "grad_norm": 0.12737324833869934, "global_step": 31072, "epoch": 863, "lr": 0.001} +{"train_loss": 0.026859000325202942, "train_loss_bc": 0.026847369968891144, "train_loss_llm": 0.0011630654335021973, "grad_norm": 0.036183394491672516, "global_step": 31073, "epoch": 863, "lr": 0.001} +{"train_loss": 0.033408332616090775, "train_loss_bc": 0.033408332616090775, "train_loss_llm": 0.0, "grad_norm": 0.05498393997550011, "global_step": 31074, "epoch": 863, "lr": 0.001} +{"train_loss": 0.032274167984724045, "train_loss_bc": 0.032271645963191986, "train_loss_llm": 0.00025203824043273926, "grad_norm": 0.07761623710393906, "global_step": 31075, "epoch": 863, "lr": 0.001} +{"train_loss": 0.033159151673316956, "train_loss_bc": 0.033159151673316956, "train_loss_llm": 0.0, "grad_norm": 0.08377259969711304, "global_step": 31076, "epoch": 863, "lr": 0.001} +{"train_loss": 0.02283165231347084, "train_loss_bc": 0.02283165231347084, "train_loss_llm": 0.0, "grad_norm": 0.026765257120132446, "global_step": 31077, "epoch": 863, "lr": 0.001} +{"train_loss": 0.031511202454566956, "train_loss_bc": 0.031511202454566956, "train_loss_llm": 0.0, "grad_norm": 0.046993888914585114, "global_step": 31078, "epoch": 863, "lr": 0.001} +{"train_loss": 0.030649330466985703, "train_loss_bc": 0.030649330466985703, "train_loss_llm": 0.0, "grad_norm": 0.07818999141454697, "global_step": 31079, "epoch": 863, "lr": 0.001} +{"train_loss": 0.02614244818687439, "train_loss_bc": 0.02614244818687439, "train_loss_llm": 0.0, "grad_norm": 0.11616069078445435, "global_step": 31080, "epoch": 863, "lr": 0.001} +{"train_loss": 0.029405610635876656, "train_loss_bc": 0.029405610635876656, "train_loss_llm": 0.0, "grad_norm": 0.03739723935723305, "global_step": 31081, "epoch": 863, "lr": 0.001} +{"train_loss": 0.02277941070497036, "train_loss_bc": 0.02277941070497036, "train_loss_llm": 0.0, "grad_norm": 0.0733984038233757, "global_step": 31082, "epoch": 863, "lr": 0.001} +{"train_loss": 0.035346467047929764, "train_loss_bc": 0.035346467047929764, "train_loss_llm": 0.0, "grad_norm": 0.11332210898399353, "global_step": 31083, "epoch": 863, "lr": 0.001} +{"train_loss": 0.03394463285803795, "train_loss_bc": 0.03394463285803795, "train_loss_llm": 0.0, "grad_norm": 0.15758128464221954, "global_step": 31084, "epoch": 863, "lr": 0.001} +{"train_loss": 0.033413805067539215, "train_loss_bc": 0.033413805067539215, "train_loss_llm": 0.0, "grad_norm": 0.05090426653623581, "global_step": 31085, "epoch": 863, "lr": 0.001} +{"train_loss": 0.0339607410132885, "train_loss_bc": 0.0339607410132885, "train_loss_llm": 0.0, "grad_norm": 0.07000823318958282, "global_step": 31086, "epoch": 863, "lr": 0.001} +{"train_loss": 0.033191315829753876, "train_loss_bc": 0.033191315829753876, "train_loss_llm": 0.0, "grad_norm": 0.10642384737730026, "global_step": 31087, "epoch": 863, "lr": 0.001} +{"train_loss": 0.029922081157565117, "train_loss_bc": 0.029922081157565117, "train_loss_llm": 0.0, "grad_norm": 0.13963568210601807, "global_step": 31088, "epoch": 863, "lr": 0.001} +{"train_loss": 0.029576651751995087, "train_loss_bc": 0.029570357874035835, "train_loss_llm": 0.000629425048828125, "grad_norm": 0.039565883576869965, "global_step": 31089, "epoch": 863, "lr": 0.001} +{"train_loss": 0.02684655785560608, "train_loss_bc": 0.02684655785560608, "train_loss_llm": 0.0, "grad_norm": 0.0626084953546524, "global_step": 31090, "epoch": 863, "lr": 0.001} +{"train_loss": 0.0276719331741333, "train_loss_bc": 0.02766009233891964, "train_loss_llm": 0.0011840760707855225, "grad_norm": 0.08133859932422638, "global_step": 31091, "epoch": 863, "lr": 0.001} +{"train_loss": 0.02979784645140171, "train_loss_bc": 0.02979784645140171, "train_loss_llm": 0.0, "grad_norm": 0.09962460398674011, "global_step": 31092, "epoch": 863, "lr": 0.001} +{"train_loss": 0.0261678546667099, "train_loss_bc": 0.0261678546667099, "train_loss_llm": 0.0, "grad_norm": 0.030974620953202248, "global_step": 31093, "epoch": 863, "lr": 0.001} +{"train_loss": 0.027983011677861214, "train_loss_bc": 0.027983011677861214, "train_loss_llm": 0.0, "grad_norm": 0.05340864509344101, "global_step": 31094, "epoch": 863, "lr": 0.001} +{"train_loss": 0.03045032173395157, "train_loss_bc": 0.03045032173395157, "train_loss_llm": 0.0, "grad_norm": 0.07162807881832123, "global_step": 31095, "epoch": 863, "lr": 0.001} +{"train_loss": 0.03118317760527134, "train_loss_bc": 0.031168930232524872, "train_loss_llm": 0.0014247298240661621, "grad_norm": 0.09302831441164017, "global_step": 31096, "epoch": 863, "lr": 0.001} +{"train_loss": 0.03385312855243683, "train_loss_bc": 0.03385312855243683, "train_loss_llm": 0.0, "grad_norm": 0.04946356266736984, "global_step": 31097, "epoch": 863, "lr": 0.001} +{"train_loss": 0.03404960036277771, "train_loss_bc": 0.03404960036277771, "train_loss_llm": 0.0, "grad_norm": 0.06279794126749039, "global_step": 31098, "epoch": 863, "lr": 0.001} +{"train_loss": 0.026963727548718452, "train_loss_bc": 0.026963727548718452, "train_loss_llm": 0.0, "grad_norm": 0.06959731876850128, "global_step": 31099, "epoch": 863, "lr": 0.001} +{"train_loss": 0.02713814005255699, "train_loss_bc": 0.027127671986818314, "train_loss_llm": 0.0010468065738677979, "grad_norm": 0.08723262697458267, "global_step": 31100, "epoch": 863, "lr": 0.001} +{"train_loss": 0.025382965803146362, "train_loss_bc": 0.025382965803146362, "train_loss_llm": 0.0, "grad_norm": 0.04274230822920799, "global_step": 31101, "epoch": 863, "lr": 0.001} +{"train_loss": 0.031653355807065964, "train_loss_bc": 0.031648579984903336, "train_loss_llm": 0.0004774034023284912, "grad_norm": 0.07074242830276489, "global_step": 31102, "epoch": 863, "lr": 0.001} +{"train_loss": 0.029926660160223644, "train_loss_bc": 0.027995986863970757, "train_loss_llm": 0.00042700767517089844, "grad_norm": 0.0904603824019432, "global_step": 31103, "epoch": 863, "lr": 0.001, "val_loss": 0.10026226192712784} +{"train_loss": 0.02878844179213047, "train_loss_bc": 0.028786832466721535, "train_loss_llm": 0.00016099214553833008, "grad_norm": 0.12637484073638916, "global_step": 31104, "epoch": 864, "lr": 0.001} +{"train_loss": 0.0337616391479969, "train_loss_bc": 0.0337514691054821, "train_loss_llm": 0.0010171234607696533, "grad_norm": 0.04522738978266716, "global_step": 31105, "epoch": 864, "lr": 0.001} +{"train_loss": 0.02911384589970112, "train_loss_bc": 0.02911384589970112, "train_loss_llm": 0.0, "grad_norm": 0.08013670146465302, "global_step": 31106, "epoch": 864, "lr": 0.001} +{"train_loss": 0.03309941291809082, "train_loss_bc": 0.03309941291809082, "train_loss_llm": 0.0, "grad_norm": 0.12656863033771515, "global_step": 31107, "epoch": 864, "lr": 0.001} +{"train_loss": 0.0330510213971138, "train_loss_bc": 0.0330510213971138, "train_loss_llm": 0.0, "grad_norm": 0.15325985848903656, "global_step": 31108, "epoch": 864, "lr": 0.001} +{"train_loss": 0.028865613043308258, "train_loss_bc": 0.028847353532910347, "train_loss_llm": 0.0018258988857269287, "grad_norm": 0.041088055819272995, "global_step": 31109, "epoch": 864, "lr": 0.001} +{"train_loss": 0.027537155896425247, "train_loss_bc": 0.027537155896425247, "train_loss_llm": 0.0, "grad_norm": 0.06250326335430145, "global_step": 31110, "epoch": 864, "lr": 0.001} +{"train_loss": 0.02207517810165882, "train_loss_bc": 0.02207517810165882, "train_loss_llm": 0.0, "grad_norm": 0.07898703962564468, "global_step": 31111, "epoch": 864, "lr": 0.001} +{"train_loss": 0.02592751570045948, "train_loss_bc": 0.02592751570045948, "train_loss_llm": 0.0, "grad_norm": 0.09669501334428787, "global_step": 31112, "epoch": 864, "lr": 0.001} +{"train_loss": 0.028462713584303856, "train_loss_bc": 0.028462713584303856, "train_loss_llm": 0.0, "grad_norm": 0.0353640653192997, "global_step": 31113, "epoch": 864, "lr": 0.001} +{"train_loss": 0.028273262083530426, "train_loss_bc": 0.028273262083530426, "train_loss_llm": 0.0, "grad_norm": 0.040093742311000824, "global_step": 31114, "epoch": 864, "lr": 0.001} +{"train_loss": 0.03150824084877968, "train_loss_bc": 0.03150824084877968, "train_loss_llm": 0.0, "grad_norm": 0.053588733077049255, "global_step": 31115, "epoch": 864, "lr": 0.001} +{"train_loss": 0.029020031914114952, "train_loss_bc": 0.029020031914114952, "train_loss_llm": 0.0, "grad_norm": 0.0618637353181839, "global_step": 31116, "epoch": 864, "lr": 0.001} +{"train_loss": 0.02957255393266678, "train_loss_bc": 0.02957255393266678, "train_loss_llm": 0.0, "grad_norm": 0.039430439472198486, "global_step": 31117, "epoch": 864, "lr": 0.001} +{"train_loss": 0.03372511267662048, "train_loss_bc": 0.03372511267662048, "train_loss_llm": 0.0, "grad_norm": 0.05796654522418976, "global_step": 31118, "epoch": 864, "lr": 0.001} +{"train_loss": 0.02383464388549328, "train_loss_bc": 0.023821434006094933, "train_loss_llm": 0.001320958137512207, "grad_norm": 0.07862428575754166, "global_step": 31119, "epoch": 864, "lr": 0.001} +{"train_loss": 0.03008098341524601, "train_loss_bc": 0.03008098341524601, "train_loss_llm": 0.0, "grad_norm": 0.084669329226017, "global_step": 31120, "epoch": 864, "lr": 0.001} +{"train_loss": 0.02928180992603302, "train_loss_bc": 0.02928180992603302, "train_loss_llm": 0.0, "grad_norm": 0.04777742922306061, "global_step": 31121, "epoch": 864, "lr": 0.001} +{"train_loss": 0.033990513533353806, "train_loss_bc": 0.033990513533353806, "train_loss_llm": 0.0, "grad_norm": 0.08113138377666473, "global_step": 31122, "epoch": 864, "lr": 0.001} +{"train_loss": 0.034009773284196854, "train_loss_bc": 0.034009773284196854, "train_loss_llm": 0.0, "grad_norm": 0.10172358900308609, "global_step": 31123, "epoch": 864, "lr": 0.001} +{"train_loss": 0.02410443127155304, "train_loss_bc": 0.02410443127155304, "train_loss_llm": 0.0, "grad_norm": 0.10901220142841339, "global_step": 31124, "epoch": 864, "lr": 0.001} +{"train_loss": 0.02540314570069313, "train_loss_bc": 0.025394253432750702, "train_loss_llm": 0.0008892416954040527, "grad_norm": 0.037214383482933044, "global_step": 31125, "epoch": 864, "lr": 0.001} +{"train_loss": 0.03256404772400856, "train_loss_bc": 0.03256404772400856, "train_loss_llm": 0.0, "grad_norm": 0.07574982196092606, "global_step": 31126, "epoch": 864, "lr": 0.001} +{"train_loss": 0.025175390765070915, "train_loss_bc": 0.025175390765070915, "train_loss_llm": 0.0, "grad_norm": 0.090031199157238, "global_step": 31127, "epoch": 864, "lr": 0.001} +{"train_loss": 0.03427846357226372, "train_loss_bc": 0.03427846357226372, "train_loss_llm": 0.0, "grad_norm": 0.12570780515670776, "global_step": 31128, "epoch": 864, "lr": 0.001} +{"train_loss": 0.03144245594739914, "train_loss_bc": 0.03144245594739914, "train_loss_llm": 0.0, "grad_norm": 0.03223413974046707, "global_step": 31129, "epoch": 864, "lr": 0.001} +{"train_loss": 0.03104785829782486, "train_loss_bc": 0.03104785829782486, "train_loss_llm": 0.0, "grad_norm": 0.051984626799821854, "global_step": 31130, "epoch": 864, "lr": 0.001} +{"train_loss": 0.02738272212445736, "train_loss_bc": 0.02738272212445736, "train_loss_llm": 0.0, "grad_norm": 0.07429622858762741, "global_step": 31131, "epoch": 864, "lr": 0.001} +{"train_loss": 0.031497109681367874, "train_loss_bc": 0.031497109681367874, "train_loss_llm": 0.0, "grad_norm": 0.09721533209085464, "global_step": 31132, "epoch": 864, "lr": 0.001} +{"train_loss": 0.029161958023905754, "train_loss_bc": 0.029161958023905754, "train_loss_llm": 0.0, "grad_norm": 0.03806206211447716, "global_step": 31133, "epoch": 864, "lr": 0.001} +{"train_loss": 0.028586333617568016, "train_loss_bc": 0.02858532778918743, "train_loss_llm": 0.00010064244270324707, "grad_norm": 0.06266681104898453, "global_step": 31134, "epoch": 864, "lr": 0.001} +{"train_loss": 0.030865075066685677, "train_loss_bc": 0.03086205944418907, "train_loss_llm": 0.0003014802932739258, "grad_norm": 0.07924618571996689, "global_step": 31135, "epoch": 864, "lr": 0.001} +{"train_loss": 0.03596318140625954, "train_loss_bc": 0.03596318140625954, "train_loss_llm": 0.0, "grad_norm": 0.0974784567952156, "global_step": 31136, "epoch": 864, "lr": 0.001} +{"train_loss": 0.03211267665028572, "train_loss_bc": 0.03211267665028572, "train_loss_llm": 0.0, "grad_norm": 0.041818637400865555, "global_step": 31137, "epoch": 864, "lr": 0.001} +{"train_loss": 0.030460486188530922, "train_loss_bc": 0.030460486188530922, "train_loss_llm": 0.0, "grad_norm": 0.07291891425848007, "global_step": 31138, "epoch": 864, "lr": 0.001} +{"train_loss": 0.029732768889516592, "train_loss_bc": 0.026343949139118195, "train_loss_llm": 0.001093149185180664, "grad_norm": 0.08161226660013199, "global_step": 31139, "epoch": 864, "lr": 0.001, "val_loss": 0.10181043297052383} +{"train_loss": 0.02814103104174137, "train_loss_bc": 0.028137702494859695, "train_loss_llm": 0.00033283233642578125, "grad_norm": 0.11158689856529236, "global_step": 31140, "epoch": 865, "lr": 0.001} +{"train_loss": 0.03327150642871857, "train_loss_bc": 0.03327150642871857, "train_loss_llm": 0.0, "grad_norm": 0.030305160209536552, "global_step": 31141, "epoch": 865, "lr": 0.001} +{"train_loss": 0.030440155416727066, "train_loss_bc": 0.030440155416727066, "train_loss_llm": 0.0, "grad_norm": 0.046335089951753616, "global_step": 31142, "epoch": 865, "lr": 0.001} +{"train_loss": 0.019214021041989326, "train_loss_bc": 0.01921175606548786, "train_loss_llm": 0.00022646784782409668, "grad_norm": 0.05651486665010452, "global_step": 31143, "epoch": 865, "lr": 0.001} +{"train_loss": 0.024948645383119583, "train_loss_bc": 0.024948645383119583, "train_loss_llm": 0.0, "grad_norm": 0.06565126031637192, "global_step": 31144, "epoch": 865, "lr": 0.001} +{"train_loss": 0.02819235250353813, "train_loss_bc": 0.028168460354208946, "train_loss_llm": 0.0023891329765319824, "grad_norm": 0.033052049577236176, "global_step": 31145, "epoch": 865, "lr": 0.001} +{"train_loss": 0.02966064028441906, "train_loss_bc": 0.02966064028441906, "train_loss_llm": 0.0, "grad_norm": 0.06435616314411163, "global_step": 31146, "epoch": 865, "lr": 0.001} +{"train_loss": 0.030292412266135216, "train_loss_bc": 0.030288347974419594, "train_loss_llm": 0.0004063546657562256, "grad_norm": 0.09447240084409714, "global_step": 31147, "epoch": 865, "lr": 0.001} +{"train_loss": 0.027983350679278374, "train_loss_bc": 0.027965523302555084, "train_loss_llm": 0.0017828047275543213, "grad_norm": 0.12802867591381073, "global_step": 31148, "epoch": 865, "lr": 0.001} +{"train_loss": 0.030391434207558632, "train_loss_bc": 0.030391434207558632, "train_loss_llm": 0.0, "grad_norm": 0.03595895692706108, "global_step": 31149, "epoch": 865, "lr": 0.001} +{"train_loss": 0.02878531627357006, "train_loss_bc": 0.02878531627357006, "train_loss_llm": 0.0, "grad_norm": 0.05440151318907738, "global_step": 31150, "epoch": 865, "lr": 0.001} +{"train_loss": 0.025389578193426132, "train_loss_bc": 0.025389578193426132, "train_loss_llm": 0.0, "grad_norm": 0.06735312193632126, "global_step": 31151, "epoch": 865, "lr": 0.001} +{"train_loss": 0.030217716470360756, "train_loss_bc": 0.030217716470360756, "train_loss_llm": 0.0, "grad_norm": 0.08173500001430511, "global_step": 31152, "epoch": 865, "lr": 0.001} +{"train_loss": 0.03150731325149536, "train_loss_bc": 0.031506456434726715, "train_loss_llm": 8.58306884765625e-05, "grad_norm": 0.03673950582742691, "global_step": 31153, "epoch": 865, "lr": 0.001} +{"train_loss": 0.029835836961865425, "train_loss_bc": 0.029817108064889908, "train_loss_llm": 0.0018728375434875488, "grad_norm": 0.04729723185300827, "global_step": 31154, "epoch": 865, "lr": 0.001} +{"train_loss": 0.027294347062706947, "train_loss_bc": 0.027294347062706947, "train_loss_llm": 0.0, "grad_norm": 0.0653160959482193, "global_step": 31155, "epoch": 865, "lr": 0.001} +{"train_loss": 0.02885911986231804, "train_loss_bc": 0.02885911986231804, "train_loss_llm": 0.0, "grad_norm": 0.07997304946184158, "global_step": 31156, "epoch": 865, "lr": 0.001} +{"train_loss": 0.030844755470752716, "train_loss_bc": 0.030844755470752716, "train_loss_llm": 0.0, "grad_norm": 0.03453902527689934, "global_step": 31157, "epoch": 865, "lr": 0.001} +{"train_loss": 0.030478104948997498, "train_loss_bc": 0.030478104948997498, "train_loss_llm": 0.0, "grad_norm": 0.05835506319999695, "global_step": 31158, "epoch": 865, "lr": 0.001} +{"train_loss": 0.03114207834005356, "train_loss_bc": 0.031123824417591095, "train_loss_llm": 0.0018254220485687256, "grad_norm": 0.08480137586593628, "global_step": 31159, "epoch": 865, "lr": 0.001} +{"train_loss": 0.028093136847019196, "train_loss_bc": 0.028093136847019196, "train_loss_llm": 0.0, "grad_norm": 0.10105915367603302, "global_step": 31160, "epoch": 865, "lr": 0.001} +{"train_loss": 0.03710891678929329, "train_loss_bc": 0.03710891678929329, "train_loss_llm": 0.0, "grad_norm": 0.037373144179582596, "global_step": 31161, "epoch": 865, "lr": 0.001} +{"train_loss": 0.030847860500216484, "train_loss_bc": 0.030847860500216484, "train_loss_llm": 0.0, "grad_norm": 0.06780945509672165, "global_step": 31162, "epoch": 865, "lr": 0.001} +{"train_loss": 0.028535017743706703, "train_loss_bc": 0.028535017743706703, "train_loss_llm": 0.0, "grad_norm": 0.09140722453594208, "global_step": 31163, "epoch": 865, "lr": 0.001} +{"train_loss": 0.028339026495814323, "train_loss_bc": 0.028339026495814323, "train_loss_llm": 0.0, "grad_norm": 0.10788968950510025, "global_step": 31164, "epoch": 865, "lr": 0.001} +{"train_loss": 0.02819119021296501, "train_loss_bc": 0.02819119021296501, "train_loss_llm": 0.0, "grad_norm": 0.03551126644015312, "global_step": 31165, "epoch": 865, "lr": 0.001} +{"train_loss": 0.025795673951506615, "train_loss_bc": 0.02578900009393692, "train_loss_llm": 0.0006673336029052734, "grad_norm": 0.05314966291189194, "global_step": 31166, "epoch": 865, "lr": 0.001} +{"train_loss": 0.030066359788179398, "train_loss_bc": 0.030066359788179398, "train_loss_llm": 0.0, "grad_norm": 0.08957869559526443, "global_step": 31167, "epoch": 865, "lr": 0.001} +{"train_loss": 0.03344932571053505, "train_loss_bc": 0.03344932571053505, "train_loss_llm": 0.0, "grad_norm": 0.11208395659923553, "global_step": 31168, "epoch": 865, "lr": 0.001} +{"train_loss": 0.028499675914645195, "train_loss_bc": 0.028499675914645195, "train_loss_llm": 0.0, "grad_norm": 0.03442954272031784, "global_step": 31169, "epoch": 865, "lr": 0.001} +{"train_loss": 0.02790561504662037, "train_loss_bc": 0.02790561504662037, "train_loss_llm": 0.0, "grad_norm": 0.04874999076128006, "global_step": 31170, "epoch": 865, "lr": 0.001} +{"train_loss": 0.030603311955928802, "train_loss_bc": 0.030603311955928802, "train_loss_llm": 0.0, "grad_norm": 0.0720461905002594, "global_step": 31171, "epoch": 865, "lr": 0.001} +{"train_loss": 0.03154163807630539, "train_loss_bc": 0.03154163807630539, "train_loss_llm": 0.0, "grad_norm": 0.10015334188938141, "global_step": 31172, "epoch": 865, "lr": 0.001} +{"train_loss": 0.034189797937870026, "train_loss_bc": 0.034189797937870026, "train_loss_llm": 0.0, "grad_norm": 0.03515373170375824, "global_step": 31173, "epoch": 865, "lr": 0.001} +{"train_loss": 0.03299352899193764, "train_loss_bc": 0.03299352899193764, "train_loss_llm": 0.0, "grad_norm": 0.06393884122371674, "global_step": 31174, "epoch": 865, "lr": 0.001} +{"train_loss": 0.029612980886465974, "train_loss_bc": 0.03301751986145973, "train_loss_llm": 0.0, "grad_norm": 0.09261290729045868, "global_step": 31175, "epoch": 865, "lr": 0.001, "train/cumulative_reward": 103.1649622944506, "train/mean_score": 15.82648747602948, "train/success_rate": 0.4, "test/cumulative_reward": 32.83768779201077, "test/mean_score": 6.340101292786766, "test/success_rate": 0.16, "val_loss": 0.1008036658167839, "train_action_mse_error": 0.03233017399907112} +{"train_loss": 0.03215464949607849, "train_loss_bc": 0.032152608036994934, "train_loss_llm": 0.0002041459083557129, "grad_norm": 0.11443395167589188, "global_step": 31176, "epoch": 866, "lr": 0.001} +{"train_loss": 0.030323538929224014, "train_loss_bc": 0.030318070203065872, "train_loss_llm": 0.0005468130111694336, "grad_norm": 0.04277580603957176, "global_step": 31177, "epoch": 866, "lr": 0.001} +{"train_loss": 0.02614758536219597, "train_loss_bc": 0.02614758536219597, "train_loss_llm": 0.0, "grad_norm": 0.07497958093881607, "global_step": 31178, "epoch": 866, "lr": 0.001} +{"train_loss": 0.03661717474460602, "train_loss_bc": 0.03661186620593071, "train_loss_llm": 0.0005308091640472412, "grad_norm": 0.11311398446559906, "global_step": 31179, "epoch": 866, "lr": 0.001} +{"train_loss": 0.033068761229515076, "train_loss_bc": 0.033068761229515076, "train_loss_llm": 0.0, "grad_norm": 0.1392788290977478, "global_step": 31180, "epoch": 866, "lr": 0.001} +{"train_loss": 0.029592428356409073, "train_loss_bc": 0.029592428356409073, "train_loss_llm": 0.0, "grad_norm": 0.032128892838954926, "global_step": 31181, "epoch": 866, "lr": 0.001} +{"train_loss": 0.0241712536662817, "train_loss_bc": 0.0241712536662817, "train_loss_llm": 0.0, "grad_norm": 0.0528079979121685, "global_step": 31182, "epoch": 866, "lr": 0.001} +{"train_loss": 0.027981890365481377, "train_loss_bc": 0.027981890365481377, "train_loss_llm": 0.0, "grad_norm": 0.06456518173217773, "global_step": 31183, "epoch": 866, "lr": 0.001} +{"train_loss": 0.02717866189777851, "train_loss_bc": 0.02717866189777851, "train_loss_llm": 0.0, "grad_norm": 0.08177381753921509, "global_step": 31184, "epoch": 866, "lr": 0.001} +{"train_loss": 0.031058594584465027, "train_loss_bc": 0.031058594584465027, "train_loss_llm": 0.0, "grad_norm": 0.034363746643066406, "global_step": 31185, "epoch": 866, "lr": 0.001} +{"train_loss": 0.027939949184656143, "train_loss_bc": 0.027939949184656143, "train_loss_llm": 0.0, "grad_norm": 0.05034704506397247, "global_step": 31186, "epoch": 866, "lr": 0.001} +{"train_loss": 0.029252108186483383, "train_loss_bc": 0.029246343299746513, "train_loss_llm": 0.0005764365196228027, "grad_norm": 0.0611264668405056, "global_step": 31187, "epoch": 866, "lr": 0.001} +{"train_loss": 0.02510855160653591, "train_loss_bc": 0.025101635605096817, "train_loss_llm": 0.0006916224956512451, "grad_norm": 0.07529166340827942, "global_step": 31188, "epoch": 866, "lr": 0.001} +{"train_loss": 0.03193216770887375, "train_loss_bc": 0.03193216770887375, "train_loss_llm": 0.0, "grad_norm": 0.033256709575653076, "global_step": 31189, "epoch": 866, "lr": 0.001} +{"train_loss": 0.035958338528871536, "train_loss_bc": 0.035958338528871536, "train_loss_llm": 0.0, "grad_norm": 0.06271149218082428, "global_step": 31190, "epoch": 866, "lr": 0.001} +{"train_loss": 0.027175435796380043, "train_loss_bc": 0.027175435796380043, "train_loss_llm": 0.0, "grad_norm": 0.08952178806066513, "global_step": 31191, "epoch": 866, "lr": 0.001} +{"train_loss": 0.031210776418447495, "train_loss_bc": 0.031210776418447495, "train_loss_llm": 0.0, "grad_norm": 0.11871151626110077, "global_step": 31192, "epoch": 866, "lr": 0.001} +{"train_loss": 0.02713383175432682, "train_loss_bc": 0.02712131477892399, "train_loss_llm": 0.0012517869472503662, "grad_norm": 0.03401389345526695, "global_step": 31193, "epoch": 866, "lr": 0.001} +{"train_loss": 0.029678571969270706, "train_loss_bc": 0.029678571969270706, "train_loss_llm": 0.0, "grad_norm": 0.048541806638240814, "global_step": 31194, "epoch": 866, "lr": 0.001} +{"train_loss": 0.03333788737654686, "train_loss_bc": 0.03333788737654686, "train_loss_llm": 0.0, "grad_norm": 0.06286069005727768, "global_step": 31195, "epoch": 866, "lr": 0.001} +{"train_loss": 0.031531479209661484, "train_loss_bc": 0.03152386471629143, "train_loss_llm": 0.0007615983486175537, "grad_norm": 0.09086108207702637, "global_step": 31196, "epoch": 866, "lr": 0.001} +{"train_loss": 0.029564648866653442, "train_loss_bc": 0.029564648866653442, "train_loss_llm": 0.0, "grad_norm": 0.036676183342933655, "global_step": 31197, "epoch": 866, "lr": 0.001} +{"train_loss": 0.030801022425293922, "train_loss_bc": 0.0307860616594553, "train_loss_llm": 0.0014961659908294678, "grad_norm": 0.06651496887207031, "global_step": 31198, "epoch": 866, "lr": 0.001} +{"train_loss": 0.02825470268726349, "train_loss_bc": 0.02825470268726349, "train_loss_llm": 0.0, "grad_norm": 0.08514337241649628, "global_step": 31199, "epoch": 866, "lr": 0.001} +{"train_loss": 0.03127821162343025, "train_loss_bc": 0.03126644343137741, "train_loss_llm": 0.0011766552925109863, "grad_norm": 0.08775694668292999, "global_step": 31200, "epoch": 866, "lr": 0.001} +{"train_loss": 0.027569923549890518, "train_loss_bc": 0.027569923549890518, "train_loss_llm": 0.0, "grad_norm": 0.0385710634291172, "global_step": 31201, "epoch": 866, "lr": 0.001} +{"train_loss": 0.032545022666454315, "train_loss_bc": 0.032545022666454315, "train_loss_llm": 0.0, "grad_norm": 0.05232137814164162, "global_step": 31202, "epoch": 866, "lr": 0.001} +{"train_loss": 0.025613419711589813, "train_loss_bc": 0.025613419711589813, "train_loss_llm": 0.0, "grad_norm": 0.05982678383588791, "global_step": 31203, "epoch": 866, "lr": 0.001} +{"train_loss": 0.03329954296350479, "train_loss_bc": 0.03329954296350479, "train_loss_llm": 0.0, "grad_norm": 0.07345636188983917, "global_step": 31204, "epoch": 866, "lr": 0.001} +{"train_loss": 0.023204538971185684, "train_loss_bc": 0.023204538971185684, "train_loss_llm": 0.0, "grad_norm": 0.02883744239807129, "global_step": 31205, "epoch": 866, "lr": 0.001} +{"train_loss": 0.02932656556367874, "train_loss_bc": 0.02932656556367874, "train_loss_llm": 0.0, "grad_norm": 0.04796970635652542, "global_step": 31206, "epoch": 866, "lr": 0.001} +{"train_loss": 0.03043964132666588, "train_loss_bc": 0.03043964132666588, "train_loss_llm": 0.0, "grad_norm": 0.06502766907215118, "global_step": 31207, "epoch": 866, "lr": 0.001} +{"train_loss": 0.027745041996240616, "train_loss_bc": 0.027728935703635216, "train_loss_llm": 0.0016106963157653809, "grad_norm": 0.07995665073394775, "global_step": 31208, "epoch": 866, "lr": 0.001} +{"train_loss": 0.02445395663380623, "train_loss_bc": 0.02445395663380623, "train_loss_llm": 0.0, "grad_norm": 0.025452030822634697, "global_step": 31209, "epoch": 866, "lr": 0.001} +{"train_loss": 0.0321955569088459, "train_loss_bc": 0.0321955569088459, "train_loss_llm": 0.0, "grad_norm": 0.05183342471718788, "global_step": 31210, "epoch": 866, "lr": 0.001} +{"train_loss": 0.029535680491891172, "train_loss_bc": 0.02843906544148922, "train_loss_llm": 0.0, "grad_norm": 0.07403349131345749, "global_step": 31211, "epoch": 866, "lr": 0.001, "val_loss": 0.1017996221780777} +{"train_loss": 0.022551823407411575, "train_loss_bc": 0.022551823407411575, "train_loss_llm": 0.0, "grad_norm": 0.08511233329772949, "global_step": 31212, "epoch": 867, "lr": 0.001} +{"train_loss": 0.033517565578222275, "train_loss_bc": 0.033504776656627655, "train_loss_llm": 0.0012788176536560059, "grad_norm": 0.040281664580106735, "global_step": 31213, "epoch": 867, "lr": 0.001} +{"train_loss": 0.034834545105695724, "train_loss_bc": 0.034834545105695724, "train_loss_llm": 0.0, "grad_norm": 0.06345250457525253, "global_step": 31214, "epoch": 867, "lr": 0.001} +{"train_loss": 0.030046213418245316, "train_loss_bc": 0.030046213418245316, "train_loss_llm": 0.0, "grad_norm": 0.08084505796432495, "global_step": 31215, "epoch": 867, "lr": 0.001} +{"train_loss": 0.022661803290247917, "train_loss_bc": 0.02265544794499874, "train_loss_llm": 0.0006355047225952148, "grad_norm": 0.09537088125944138, "global_step": 31216, "epoch": 867, "lr": 0.001} +{"train_loss": 0.024776555597782135, "train_loss_bc": 0.024776555597782135, "train_loss_llm": 0.0, "grad_norm": 0.027843110263347626, "global_step": 31217, "epoch": 867, "lr": 0.001} +{"train_loss": 0.030183939263224602, "train_loss_bc": 0.03016890399158001, "train_loss_llm": 0.0015036165714263916, "grad_norm": 0.04414377361536026, "global_step": 31218, "epoch": 867, "lr": 0.001} +{"train_loss": 0.024732835590839386, "train_loss_bc": 0.024732835590839386, "train_loss_llm": 0.0, "grad_norm": 0.059224724769592285, "global_step": 31219, "epoch": 867, "lr": 0.001} +{"train_loss": 0.030498743057250977, "train_loss_bc": 0.030498743057250977, "train_loss_llm": 0.0, "grad_norm": 0.0769864171743393, "global_step": 31220, "epoch": 867, "lr": 0.001} +{"train_loss": 0.03404010832309723, "train_loss_bc": 0.03404010832309723, "train_loss_llm": 0.0, "grad_norm": 0.03505561873316765, "global_step": 31221, "epoch": 867, "lr": 0.001} +{"train_loss": 0.02953627146780491, "train_loss_bc": 0.02953627146780491, "train_loss_llm": 0.0, "grad_norm": 0.0509367436170578, "global_step": 31222, "epoch": 867, "lr": 0.001} +{"train_loss": 0.03351333737373352, "train_loss_bc": 0.033510562032461166, "train_loss_llm": 0.00027751922607421875, "grad_norm": 0.06534619629383087, "global_step": 31223, "epoch": 867, "lr": 0.001} +{"train_loss": 0.031965989619493484, "train_loss_bc": 0.031965989619493484, "train_loss_llm": 0.0, "grad_norm": 0.08491158485412598, "global_step": 31224, "epoch": 867, "lr": 0.001} +{"train_loss": 0.02922813966870308, "train_loss_bc": 0.02922813966870308, "train_loss_llm": 0.0, "grad_norm": 0.04034952446818352, "global_step": 31225, "epoch": 867, "lr": 0.001} +{"train_loss": 0.030509786680340767, "train_loss_bc": 0.030509786680340767, "train_loss_llm": 0.0, "grad_norm": 0.06325763463973999, "global_step": 31226, "epoch": 867, "lr": 0.001} +{"train_loss": 0.025012539699673653, "train_loss_bc": 0.025012539699673653, "train_loss_llm": 0.0, "grad_norm": 0.08495894074440002, "global_step": 31227, "epoch": 867, "lr": 0.001} +{"train_loss": 0.03117223270237446, "train_loss_bc": 0.03117223270237446, "train_loss_llm": 0.0, "grad_norm": 0.09783339500427246, "global_step": 31228, "epoch": 867, "lr": 0.001} +{"train_loss": 0.03606485202908516, "train_loss_bc": 0.03606485202908516, "train_loss_llm": 0.0, "grad_norm": 0.05484709143638611, "global_step": 31229, "epoch": 867, "lr": 0.001} +{"train_loss": 0.026296250522136688, "train_loss_bc": 0.026296250522136688, "train_loss_llm": 0.0, "grad_norm": 0.07415115088224411, "global_step": 31230, "epoch": 867, "lr": 0.001} +{"train_loss": 0.031905241310596466, "train_loss_bc": 0.031905241310596466, "train_loss_llm": 0.0, "grad_norm": 0.09392762929201126, "global_step": 31231, "epoch": 867, "lr": 0.001} +{"train_loss": 0.03230532631278038, "train_loss_bc": 0.032285358756780624, "train_loss_llm": 0.0019968152046203613, "grad_norm": 0.11142553389072418, "global_step": 31232, "epoch": 867, "lr": 0.001} +{"train_loss": 0.029774989932775497, "train_loss_bc": 0.02977336384356022, "train_loss_llm": 0.00016254186630249023, "grad_norm": 0.04738987237215042, "global_step": 31233, "epoch": 867, "lr": 0.001} +{"train_loss": 0.028456248342990875, "train_loss_bc": 0.028456248342990875, "train_loss_llm": 0.0, "grad_norm": 0.07142285257577896, "global_step": 31234, "epoch": 867, "lr": 0.001} +{"train_loss": 0.02555156499147415, "train_loss_bc": 0.02555156499147415, "train_loss_llm": 0.0, "grad_norm": 0.08127441257238388, "global_step": 31235, "epoch": 867, "lr": 0.001} +{"train_loss": 0.029667088761925697, "train_loss_bc": 0.029667088761925697, "train_loss_llm": 0.0, "grad_norm": 0.09878971427679062, "global_step": 31236, "epoch": 867, "lr": 0.001} +{"train_loss": 0.025017401203513145, "train_loss_bc": 0.025017401203513145, "train_loss_llm": 0.0, "grad_norm": 0.034938033670186996, "global_step": 31237, "epoch": 867, "lr": 0.001} +{"train_loss": 0.025687461718916893, "train_loss_bc": 0.025687461718916893, "train_loss_llm": 0.0, "grad_norm": 0.050930559635162354, "global_step": 31238, "epoch": 867, "lr": 0.001} +{"train_loss": 0.031922921538352966, "train_loss_bc": 0.03189403936266899, "train_loss_llm": 0.002888202667236328, "grad_norm": 0.06853288412094116, "global_step": 31239, "epoch": 867, "lr": 0.001} +{"train_loss": 0.030250970274209976, "train_loss_bc": 0.030250970274209976, "train_loss_llm": 0.0, "grad_norm": 0.08384238183498383, "global_step": 31240, "epoch": 867, "lr": 0.001} +{"train_loss": 0.02950751595199108, "train_loss_bc": 0.02950751595199108, "train_loss_llm": 0.0, "grad_norm": 0.04330813139677048, "global_step": 31241, "epoch": 867, "lr": 0.001} +{"train_loss": 0.03192844241857529, "train_loss_bc": 0.03192844241857529, "train_loss_llm": 0.0, "grad_norm": 0.06300996989011765, "global_step": 31242, "epoch": 867, "lr": 0.001} +{"train_loss": 0.027887310832738876, "train_loss_bc": 0.027887310832738876, "train_loss_llm": 0.0, "grad_norm": 0.07376130670309067, "global_step": 31243, "epoch": 867, "lr": 0.001} +{"train_loss": 0.029771972447633743, "train_loss_bc": 0.02975665032863617, "train_loss_llm": 0.0015322864055633545, "grad_norm": 0.08551613986492157, "global_step": 31244, "epoch": 867, "lr": 0.001} +{"train_loss": 0.028591690585017204, "train_loss_bc": 0.028591690585017204, "train_loss_llm": 0.0, "grad_norm": 0.032129429280757904, "global_step": 31245, "epoch": 867, "lr": 0.001} +{"train_loss": 0.03121730498969555, "train_loss_bc": 0.031209871172904968, "train_loss_llm": 0.0007434487342834473, "grad_norm": 0.060915954411029816, "global_step": 31246, "epoch": 867, "lr": 0.001} +{"train_loss": 0.029449049124701157, "train_loss_bc": 0.029552536085247993, "train_loss_llm": 0.0026247501373291016, "grad_norm": 0.08127479255199432, "global_step": 31247, "epoch": 867, "lr": 0.001, "val_loss": 0.10100464522838593} +{"train_loss": 0.02649439312517643, "train_loss_bc": 0.02649439312517643, "train_loss_llm": 0.0, "grad_norm": 0.09995859861373901, "global_step": 31248, "epoch": 868, "lr": 0.001} +{"train_loss": 0.02470068633556366, "train_loss_bc": 0.024698209017515182, "train_loss_llm": 0.00024774670600891113, "grad_norm": 0.028195062652230263, "global_step": 31249, "epoch": 868, "lr": 0.001} +{"train_loss": 0.030279526486992836, "train_loss_bc": 0.030279526486992836, "train_loss_llm": 0.0, "grad_norm": 0.05209772661328316, "global_step": 31250, "epoch": 868, "lr": 0.001} +{"train_loss": 0.031340256333351135, "train_loss_bc": 0.03132166713476181, "train_loss_llm": 0.0018590092658996582, "grad_norm": 0.06922955811023712, "global_step": 31251, "epoch": 868, "lr": 0.001} +{"train_loss": 0.02962786890566349, "train_loss_bc": 0.02962786890566349, "train_loss_llm": 0.0, "grad_norm": 0.09717562049627304, "global_step": 31252, "epoch": 868, "lr": 0.001} +{"train_loss": 0.024437610059976578, "train_loss_bc": 0.024437610059976578, "train_loss_llm": 0.0, "grad_norm": 0.03359841927886009, "global_step": 31253, "epoch": 868, "lr": 0.001} +{"train_loss": 0.02799762599170208, "train_loss_bc": 0.02799762599170208, "train_loss_llm": 0.0, "grad_norm": 0.0525040328502655, "global_step": 31254, "epoch": 868, "lr": 0.001} +{"train_loss": 0.024361157789826393, "train_loss_bc": 0.024359947070479393, "train_loss_llm": 0.00012108683586120605, "grad_norm": 0.07079626619815826, "global_step": 31255, "epoch": 868, "lr": 0.001} +{"train_loss": 0.02942385897040367, "train_loss_bc": 0.02942385897040367, "train_loss_llm": 0.0, "grad_norm": 0.08869189023971558, "global_step": 31256, "epoch": 868, "lr": 0.001} +{"train_loss": 0.02586178109049797, "train_loss_bc": 0.02586178109049797, "train_loss_llm": 0.0, "grad_norm": 0.034507423639297485, "global_step": 31257, "epoch": 868, "lr": 0.001} +{"train_loss": 0.032766059041023254, "train_loss_bc": 0.032766059041023254, "train_loss_llm": 0.0, "grad_norm": 0.054069001227617264, "global_step": 31258, "epoch": 868, "lr": 0.001} +{"train_loss": 0.030948465690016747, "train_loss_bc": 0.030948465690016747, "train_loss_llm": 0.0, "grad_norm": 0.06578663736581802, "global_step": 31259, "epoch": 868, "lr": 0.001} +{"train_loss": 0.027106741443276405, "train_loss_bc": 0.027106741443276405, "train_loss_llm": 0.0, "grad_norm": 0.07746298611164093, "global_step": 31260, "epoch": 868, "lr": 0.001} +{"train_loss": 0.030130604282021523, "train_loss_bc": 0.030122321099042892, "train_loss_llm": 0.0008282959461212158, "grad_norm": 0.04428429156541824, "global_step": 31261, "epoch": 868, "lr": 0.001} +{"train_loss": 0.02918294444680214, "train_loss_bc": 0.02917877770960331, "train_loss_llm": 0.00041672587394714355, "grad_norm": 0.05721927806735039, "global_step": 31262, "epoch": 868, "lr": 0.001} +{"train_loss": 0.033484846353530884, "train_loss_bc": 0.033484846353530884, "train_loss_llm": 0.0, "grad_norm": 0.07169676572084427, "global_step": 31263, "epoch": 868, "lr": 0.001} +{"train_loss": 0.03482051193714142, "train_loss_bc": 0.03482051193714142, "train_loss_llm": 0.0, "grad_norm": 0.08345510810613632, "global_step": 31264, "epoch": 868, "lr": 0.001} +{"train_loss": 0.027491901069879532, "train_loss_bc": 0.027491901069879532, "train_loss_llm": 0.0, "grad_norm": 0.032945118844509125, "global_step": 31265, "epoch": 868, "lr": 0.001} +{"train_loss": 0.030897442251443863, "train_loss_bc": 0.030897442251443863, "train_loss_llm": 0.0, "grad_norm": 0.06926457583904266, "global_step": 31266, "epoch": 868, "lr": 0.001} +{"train_loss": 0.028277801349759102, "train_loss_bc": 0.028277801349759102, "train_loss_llm": 0.0, "grad_norm": 0.08700747787952423, "global_step": 31267, "epoch": 868, "lr": 0.001} +{"train_loss": 0.030683796852827072, "train_loss_bc": 0.030683796852827072, "train_loss_llm": 0.0, "grad_norm": 0.10244270414113998, "global_step": 31268, "epoch": 868, "lr": 0.001} +{"train_loss": 0.02825373224914074, "train_loss_bc": 0.02825373224914074, "train_loss_llm": 0.0, "grad_norm": 0.037593092769384384, "global_step": 31269, "epoch": 868, "lr": 0.001} +{"train_loss": 0.033807024359703064, "train_loss_bc": 0.033807024359703064, "train_loss_llm": 0.0, "grad_norm": 0.07415035367012024, "global_step": 31270, "epoch": 868, "lr": 0.001} +{"train_loss": 0.023686259984970093, "train_loss_bc": 0.023686259984970093, "train_loss_llm": 0.0, "grad_norm": 0.08509936928749084, "global_step": 31271, "epoch": 868, "lr": 0.001} +{"train_loss": 0.030197693035006523, "train_loss_bc": 0.030197693035006523, "train_loss_llm": 0.0, "grad_norm": 0.11258675903081894, "global_step": 31272, "epoch": 868, "lr": 0.001} +{"train_loss": 0.02716745249927044, "train_loss_bc": 0.027157040312886238, "train_loss_llm": 0.0010412335395812988, "grad_norm": 0.03162391856312752, "global_step": 31273, "epoch": 868, "lr": 0.001} +{"train_loss": 0.024825984612107277, "train_loss_bc": 0.024825984612107277, "train_loss_llm": 0.0, "grad_norm": 0.05208328366279602, "global_step": 31274, "epoch": 868, "lr": 0.001} +{"train_loss": 0.027804162353277206, "train_loss_bc": 0.027804162353277206, "train_loss_llm": 0.0, "grad_norm": 0.0767093077301979, "global_step": 31275, "epoch": 868, "lr": 0.001} +{"train_loss": 0.02839486673474312, "train_loss_bc": 0.028393078595399857, "train_loss_llm": 0.00017878413200378418, "grad_norm": 0.0995224341750145, "global_step": 31276, "epoch": 868, "lr": 0.001} +{"train_loss": 0.030087394639849663, "train_loss_bc": 0.030073203146457672, "train_loss_llm": 0.0014190971851348877, "grad_norm": 0.03533122316002846, "global_step": 31277, "epoch": 868, "lr": 0.001} +{"train_loss": 0.032552868127822876, "train_loss_bc": 0.032552868127822876, "train_loss_llm": 0.0, "grad_norm": 0.06347382068634033, "global_step": 31278, "epoch": 868, "lr": 0.001} +{"train_loss": 0.03428160026669502, "train_loss_bc": 0.03425222635269165, "train_loss_llm": 0.002937495708465576, "grad_norm": 0.07469817250967026, "global_step": 31279, "epoch": 868, "lr": 0.001} +{"train_loss": 0.0293488297611475, "train_loss_bc": 0.0293488297611475, "train_loss_llm": 0.0, "grad_norm": 0.09642846882343292, "global_step": 31280, "epoch": 868, "lr": 0.001} +{"train_loss": 0.038010869175195694, "train_loss_bc": 0.038010869175195694, "train_loss_llm": 0.0, "grad_norm": 0.04985259100794792, "global_step": 31281, "epoch": 868, "lr": 0.001} +{"train_loss": 0.02875625714659691, "train_loss_bc": 0.028742259368300438, "train_loss_llm": 0.0013997554779052734, "grad_norm": 0.07345860451459885, "global_step": 31282, "epoch": 868, "lr": 0.001} +{"train_loss": 0.029539450744373932, "train_loss_bc": 0.035929352045059204, "train_loss_llm": 0.0, "grad_norm": 0.08886220306158066, "global_step": 31283, "epoch": 868, "lr": 0.001, "val_loss": 0.10176577419042587} +{"train_loss": 0.028710665181279182, "train_loss_bc": 0.028700347989797592, "train_loss_llm": 0.0010316669940948486, "grad_norm": 0.11111345142126083, "global_step": 31284, "epoch": 869, "lr": 0.001} +{"train_loss": 0.028903903439641, "train_loss_bc": 0.028903903439641, "train_loss_llm": 0.0, "grad_norm": 0.040488142520189285, "global_step": 31285, "epoch": 869, "lr": 0.001} +{"train_loss": 0.027455264702439308, "train_loss_bc": 0.027455264702439308, "train_loss_llm": 0.0, "grad_norm": 0.05883423238992691, "global_step": 31286, "epoch": 869, "lr": 0.001} +{"train_loss": 0.0292659979313612, "train_loss_bc": 0.02926359325647354, "train_loss_llm": 0.00024041533470153809, "grad_norm": 0.0659104660153389, "global_step": 31287, "epoch": 869, "lr": 0.001} +{"train_loss": 0.021450484171509743, "train_loss_bc": 0.02144809067249298, "train_loss_llm": 0.00023943185806274414, "grad_norm": 0.07433488219976425, "global_step": 31288, "epoch": 869, "lr": 0.001} +{"train_loss": 0.030057311058044434, "train_loss_bc": 0.030057311058044434, "train_loss_llm": 0.0, "grad_norm": 0.03700092434883118, "global_step": 31289, "epoch": 869, "lr": 0.001} +{"train_loss": 0.032372165471315384, "train_loss_bc": 0.032372165471315384, "train_loss_llm": 0.0, "grad_norm": 0.052771300077438354, "global_step": 31290, "epoch": 869, "lr": 0.001} +{"train_loss": 0.031184231862425804, "train_loss_bc": 0.031184231862425804, "train_loss_llm": 0.0, "grad_norm": 0.07484367489814758, "global_step": 31291, "epoch": 869, "lr": 0.001} +{"train_loss": 0.02834664098918438, "train_loss_bc": 0.02834664098918438, "train_loss_llm": 0.0, "grad_norm": 0.09581829607486725, "global_step": 31292, "epoch": 869, "lr": 0.001} +{"train_loss": 0.03258513659238815, "train_loss_bc": 0.03258513659238815, "train_loss_llm": 0.0, "grad_norm": 0.04864257201552391, "global_step": 31293, "epoch": 869, "lr": 0.001} +{"train_loss": 0.039055999368429184, "train_loss_bc": 0.039055999368429184, "train_loss_llm": 0.0, "grad_norm": 0.09726715832948685, "global_step": 31294, "epoch": 869, "lr": 0.001} +{"train_loss": 0.0305728018283844, "train_loss_bc": 0.0305728018283844, "train_loss_llm": 0.0, "grad_norm": 0.1264292150735855, "global_step": 31295, "epoch": 869, "lr": 0.001} +{"train_loss": 0.030529256910085678, "train_loss_bc": 0.030529256910085678, "train_loss_llm": 0.0, "grad_norm": 0.15085206925868988, "global_step": 31296, "epoch": 869, "lr": 0.001} +{"train_loss": 0.027311772108078003, "train_loss_bc": 0.02730800025165081, "train_loss_llm": 0.0003771781921386719, "grad_norm": 0.026435591280460358, "global_step": 31297, "epoch": 869, "lr": 0.001} +{"train_loss": 0.02543901652097702, "train_loss_bc": 0.02543901652097702, "train_loss_llm": 0.0, "grad_norm": 0.041867032647132874, "global_step": 31298, "epoch": 869, "lr": 0.001} +{"train_loss": 0.02839394472539425, "train_loss_bc": 0.02839394472539425, "train_loss_llm": 0.0, "grad_norm": 0.05973346531391144, "global_step": 31299, "epoch": 869, "lr": 0.001} +{"train_loss": 0.030475284904241562, "train_loss_bc": 0.030475284904241562, "train_loss_llm": 0.0, "grad_norm": 0.0870492160320282, "global_step": 31300, "epoch": 869, "lr": 0.001} +{"train_loss": 0.030844232067465782, "train_loss_bc": 0.030844153836369514, "train_loss_llm": 7.867813110351562e-06, "grad_norm": 0.03362246975302696, "global_step": 31301, "epoch": 869, "lr": 0.001} +{"train_loss": 0.0361197255551815, "train_loss_bc": 0.03611420840024948, "train_loss_llm": 0.0005516707897186279, "grad_norm": 0.04977836459875107, "global_step": 31302, "epoch": 869, "lr": 0.001} +{"train_loss": 0.02979166805744171, "train_loss_bc": 0.02979166805744171, "train_loss_llm": 0.0, "grad_norm": 0.0593574196100235, "global_step": 31303, "epoch": 869, "lr": 0.001} +{"train_loss": 0.033481840044260025, "train_loss_bc": 0.03346733748912811, "train_loss_llm": 0.0014503896236419678, "grad_norm": 0.08182919025421143, "global_step": 31304, "epoch": 869, "lr": 0.001} +{"train_loss": 0.029708117246627808, "train_loss_bc": 0.02970173768699169, "train_loss_llm": 0.0006379783153533936, "grad_norm": 0.0314408577978611, "global_step": 31305, "epoch": 869, "lr": 0.001} +{"train_loss": 0.028669089078903198, "train_loss_bc": 0.02866189181804657, "train_loss_llm": 0.0007196962833404541, "grad_norm": 0.04758615419268608, "global_step": 31306, "epoch": 869, "lr": 0.001} +{"train_loss": 0.02954852394759655, "train_loss_bc": 0.02954852394759655, "train_loss_llm": 0.0, "grad_norm": 0.05511298403143883, "global_step": 31307, "epoch": 869, "lr": 0.001} +{"train_loss": 0.02728046104311943, "train_loss_bc": 0.02728046104311943, "train_loss_llm": 0.0, "grad_norm": 0.07179815322160721, "global_step": 31308, "epoch": 869, "lr": 0.001} +{"train_loss": 0.03279103711247444, "train_loss_bc": 0.03279103711247444, "train_loss_llm": 0.0, "grad_norm": 0.04269592463970184, "global_step": 31309, "epoch": 869, "lr": 0.001} +{"train_loss": 0.023823540657758713, "train_loss_bc": 0.02381390891969204, "train_loss_llm": 0.0009631216526031494, "grad_norm": 0.061209049075841904, "global_step": 31310, "epoch": 869, "lr": 0.001} +{"train_loss": 0.034037306904792786, "train_loss_bc": 0.034037306904792786, "train_loss_llm": 0.0, "grad_norm": 0.08045777678489685, "global_step": 31311, "epoch": 869, "lr": 0.001} +{"train_loss": 0.026678217574954033, "train_loss_bc": 0.026678217574954033, "train_loss_llm": 0.0, "grad_norm": 0.10730390250682831, "global_step": 31312, "epoch": 869, "lr": 0.001} +{"train_loss": 0.028453566133975983, "train_loss_bc": 0.028453566133975983, "train_loss_llm": 0.0, "grad_norm": 0.03724178671836853, "global_step": 31313, "epoch": 869, "lr": 0.001} +{"train_loss": 0.029636437073349953, "train_loss_bc": 0.029636437073349953, "train_loss_llm": 0.0, "grad_norm": 0.059124141931533813, "global_step": 31314, "epoch": 869, "lr": 0.001} +{"train_loss": 0.0248417891561985, "train_loss_bc": 0.0248417891561985, "train_loss_llm": 0.0, "grad_norm": 0.08386873453855515, "global_step": 31315, "epoch": 869, "lr": 0.001} +{"train_loss": 0.027145497500896454, "train_loss_bc": 0.027145497500896454, "train_loss_llm": 0.0, "grad_norm": 0.10949811339378357, "global_step": 31316, "epoch": 869, "lr": 0.001} +{"train_loss": 0.027942970395088196, "train_loss_bc": 0.027942970395088196, "train_loss_llm": 0.0, "grad_norm": 0.03559741750359535, "global_step": 31317, "epoch": 869, "lr": 0.001} +{"train_loss": 0.035468559712171555, "train_loss_bc": 0.03545286878943443, "train_loss_llm": 0.001569211483001709, "grad_norm": 0.06593959033489227, "global_step": 31318, "epoch": 869, "lr": 0.001} +{"train_loss": 0.02969795812128319, "train_loss_bc": 0.030754035338759422, "train_loss_llm": 0.0, "grad_norm": 0.09028004109859467, "global_step": 31319, "epoch": 869, "lr": 0.001, "val_loss": 0.09904490411281586} +{"train_loss": 0.029146263375878334, "train_loss_bc": 0.029146263375878334, "train_loss_llm": 0.0, "grad_norm": 0.1134464219212532, "global_step": 31320, "epoch": 870, "lr": 0.001} +{"train_loss": 0.03320285677909851, "train_loss_bc": 0.03320285677909851, "train_loss_llm": 0.0, "grad_norm": 0.03728286921977997, "global_step": 31321, "epoch": 870, "lr": 0.001} +{"train_loss": 0.028690926730632782, "train_loss_bc": 0.02868516556918621, "train_loss_llm": 0.0005761981010437012, "grad_norm": 0.05380873382091522, "global_step": 31322, "epoch": 870, "lr": 0.001} +{"train_loss": 0.030938338488340378, "train_loss_bc": 0.03091934323310852, "train_loss_llm": 0.0018995404243469238, "grad_norm": 0.06494373083114624, "global_step": 31323, "epoch": 870, "lr": 0.001} +{"train_loss": 0.027079612016677856, "train_loss_bc": 0.027079612016677856, "train_loss_llm": 0.0, "grad_norm": 0.08108500391244888, "global_step": 31324, "epoch": 870, "lr": 0.001} +{"train_loss": 0.031389497220516205, "train_loss_bc": 0.031389497220516205, "train_loss_llm": 0.0, "grad_norm": 0.048734039068222046, "global_step": 31325, "epoch": 870, "lr": 0.001} +{"train_loss": 0.03033456578850746, "train_loss_bc": 0.03033456578850746, "train_loss_llm": 0.0, "grad_norm": 0.05215708538889885, "global_step": 31326, "epoch": 870, "lr": 0.001} +{"train_loss": 0.027588319033384323, "train_loss_bc": 0.027585070580244064, "train_loss_llm": 0.0003247857093811035, "grad_norm": 0.06861957907676697, "global_step": 31327, "epoch": 870, "lr": 0.001} +{"train_loss": 0.03411601856350899, "train_loss_bc": 0.03411601856350899, "train_loss_llm": 0.0, "grad_norm": 0.08049298822879791, "global_step": 31328, "epoch": 870, "lr": 0.001} +{"train_loss": 0.024636467918753624, "train_loss_bc": 0.024636467918753624, "train_loss_llm": 0.0, "grad_norm": 0.032629434019327164, "global_step": 31329, "epoch": 870, "lr": 0.001} +{"train_loss": 0.026389572769403458, "train_loss_bc": 0.026389572769403458, "train_loss_llm": 0.0, "grad_norm": 0.05208352953195572, "global_step": 31330, "epoch": 870, "lr": 0.001} +{"train_loss": 0.027357788756489754, "train_loss_bc": 0.027356283739209175, "train_loss_llm": 0.00015056133270263672, "grad_norm": 0.06797943264245987, "global_step": 31331, "epoch": 870, "lr": 0.001} +{"train_loss": 0.025464151054620743, "train_loss_bc": 0.025464151054620743, "train_loss_llm": 0.0, "grad_norm": 0.08650044351816177, "global_step": 31332, "epoch": 870, "lr": 0.001} +{"train_loss": 0.02912941202521324, "train_loss_bc": 0.02912941202521324, "train_loss_llm": 0.0, "grad_norm": 0.057871345430612564, "global_step": 31333, "epoch": 870, "lr": 0.001} +{"train_loss": 0.03244795277714729, "train_loss_bc": 0.03244795277714729, "train_loss_llm": 0.0, "grad_norm": 0.08305763453245163, "global_step": 31334, "epoch": 870, "lr": 0.001} +{"train_loss": 0.030402952805161476, "train_loss_bc": 0.030402952805161476, "train_loss_llm": 0.0, "grad_norm": 0.10337673872709274, "global_step": 31335, "epoch": 870, "lr": 0.001} +{"train_loss": 0.031643494963645935, "train_loss_bc": 0.031643494963645935, "train_loss_llm": 0.0, "grad_norm": 0.11937876045703888, "global_step": 31336, "epoch": 870, "lr": 0.001} +{"train_loss": 0.035646311938762665, "train_loss_bc": 0.035646311938762665, "train_loss_llm": 0.0, "grad_norm": 0.038537561893463135, "global_step": 31337, "epoch": 870, "lr": 0.001} +{"train_loss": 0.02839076891541481, "train_loss_bc": 0.02839076891541481, "train_loss_llm": 0.0, "grad_norm": 0.06648501753807068, "global_step": 31338, "epoch": 870, "lr": 0.001} +{"train_loss": 0.030207226052880287, "train_loss_bc": 0.030207226052880287, "train_loss_llm": 0.0, "grad_norm": 0.0969267189502716, "global_step": 31339, "epoch": 870, "lr": 0.001} +{"train_loss": 0.027203891426324844, "train_loss_bc": 0.027203891426324844, "train_loss_llm": 0.0, "grad_norm": 0.12591923773288727, "global_step": 31340, "epoch": 870, "lr": 0.001} +{"train_loss": 0.03610850125551224, "train_loss_bc": 0.036078792065382004, "train_loss_llm": 0.002970755100250244, "grad_norm": 0.04323412850499153, "global_step": 31341, "epoch": 870, "lr": 0.001} +{"train_loss": 0.032905660569667816, "train_loss_bc": 0.032905660569667816, "train_loss_llm": 0.0, "grad_norm": 0.07458202540874481, "global_step": 31342, "epoch": 870, "lr": 0.001} +{"train_loss": 0.030307136476039886, "train_loss_bc": 0.030307136476039886, "train_loss_llm": 0.0, "grad_norm": 0.09779495000839233, "global_step": 31343, "epoch": 870, "lr": 0.001} +{"train_loss": 0.028344718739390373, "train_loss_bc": 0.028344718739390373, "train_loss_llm": 0.0, "grad_norm": 0.11172313988208771, "global_step": 31344, "epoch": 870, "lr": 0.001} +{"train_loss": 0.02621729113161564, "train_loss_bc": 0.02621729113161564, "train_loss_llm": 0.0, "grad_norm": 0.035067833960056305, "global_step": 31345, "epoch": 870, "lr": 0.001} +{"train_loss": 0.02776011824607849, "train_loss_bc": 0.02775934524834156, "train_loss_llm": 7.724761962890625e-05, "grad_norm": 0.05238701403141022, "global_step": 31346, "epoch": 870, "lr": 0.001} +{"train_loss": 0.027487432584166527, "train_loss_bc": 0.027487432584166527, "train_loss_llm": 0.0, "grad_norm": 0.06642065197229385, "global_step": 31347, "epoch": 870, "lr": 0.001} +{"train_loss": 0.025550564751029015, "train_loss_bc": 0.025528058409690857, "train_loss_llm": 0.002250581979751587, "grad_norm": 0.07764456421136856, "global_step": 31348, "epoch": 870, "lr": 0.001} +{"train_loss": 0.027798201888799667, "train_loss_bc": 0.027798201888799667, "train_loss_llm": 0.0, "grad_norm": 0.03584214672446251, "global_step": 31349, "epoch": 870, "lr": 0.001} +{"train_loss": 0.03155941888689995, "train_loss_bc": 0.03155941888689995, "train_loss_llm": 0.0, "grad_norm": 0.061340417712926865, "global_step": 31350, "epoch": 870, "lr": 0.001} +{"train_loss": 0.03016088716685772, "train_loss_bc": 0.03016088716685772, "train_loss_llm": 0.0, "grad_norm": 0.08977620303630829, "global_step": 31351, "epoch": 870, "lr": 0.001} +{"train_loss": 0.030886806547641754, "train_loss_bc": 0.030886806547641754, "train_loss_llm": 0.0, "grad_norm": 0.10632666200399399, "global_step": 31352, "epoch": 870, "lr": 0.001} +{"train_loss": 0.03069131262600422, "train_loss_bc": 0.03069131262600422, "train_loss_llm": 0.0, "grad_norm": 0.033077534288167953, "global_step": 31353, "epoch": 870, "lr": 0.001} +{"train_loss": 0.030925797298550606, "train_loss_bc": 0.03091578371822834, "train_loss_llm": 0.0010013282299041748, "grad_norm": 0.054280586540699005, "global_step": 31354, "epoch": 870, "lr": 0.001} +{"train_loss": 0.029547357155630987, "train_loss_bc": 0.025594620034098625, "train_loss_llm": 0.0, "grad_norm": 0.07566652446985245, "global_step": 31355, "epoch": 870, "lr": 0.001, "train/cumulative_reward": 121.1307937042762, "train/mean_score": 19.840940880686453, "train/success_rate": 0.5, "test/cumulative_reward": 29.22602010837838, "test/mean_score": 8.428747910634623, "test/success_rate": 0.22, "val_loss": 0.10035483539104462, "train_action_mse_error": 0.03253908455371857} +{"train_loss": 0.03246442601084709, "train_loss_bc": 0.03246442601084709, "train_loss_llm": 0.0, "grad_norm": 0.1007051095366478, "global_step": 31356, "epoch": 871, "lr": 0.001} +{"train_loss": 0.029427114874124527, "train_loss_bc": 0.029427114874124527, "train_loss_llm": 0.0, "grad_norm": 0.0386999286711216, "global_step": 31357, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03684386610984802, "train_loss_bc": 0.03684386610984802, "train_loss_llm": 0.0, "grad_norm": 0.08027875423431396, "global_step": 31358, "epoch": 871, "lr": 0.001} +{"train_loss": 0.02193991281092167, "train_loss_bc": 0.021936098113656044, "train_loss_llm": 0.0003814995288848877, "grad_norm": 0.1006280928850174, "global_step": 31359, "epoch": 871, "lr": 0.001} +{"train_loss": 0.02951163984835148, "train_loss_bc": 0.02951163984835148, "train_loss_llm": 0.0, "grad_norm": 0.13535405695438385, "global_step": 31360, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03178063780069351, "train_loss_bc": 0.03177187219262123, "train_loss_llm": 0.0008764863014221191, "grad_norm": 0.033881474286317825, "global_step": 31361, "epoch": 871, "lr": 0.001} +{"train_loss": 0.02832118235528469, "train_loss_bc": 0.02832118235528469, "train_loss_llm": 0.0, "grad_norm": 0.05780982971191406, "global_step": 31362, "epoch": 871, "lr": 0.001} +{"train_loss": 0.02642548270523548, "train_loss_bc": 0.026423940435051918, "train_loss_llm": 0.00015431642532348633, "grad_norm": 0.08270981907844543, "global_step": 31363, "epoch": 871, "lr": 0.001} +{"train_loss": 0.036539144814014435, "train_loss_bc": 0.036539144814014435, "train_loss_llm": 0.0, "grad_norm": 0.12145645171403885, "global_step": 31364, "epoch": 871, "lr": 0.001} +{"train_loss": 0.028186578303575516, "train_loss_bc": 0.028186578303575516, "train_loss_llm": 0.0, "grad_norm": 0.027518106624484062, "global_step": 31365, "epoch": 871, "lr": 0.001} +{"train_loss": 0.027460183948278427, "train_loss_bc": 0.027454806491732597, "train_loss_llm": 0.0005377531051635742, "grad_norm": 0.05004497617483139, "global_step": 31366, "epoch": 871, "lr": 0.001} +{"train_loss": 0.02991245873272419, "train_loss_bc": 0.02991245873272419, "train_loss_llm": 0.0, "grad_norm": 0.07359447330236435, "global_step": 31367, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03500870242714882, "train_loss_bc": 0.03500870242714882, "train_loss_llm": 0.0, "grad_norm": 0.10011906176805496, "global_step": 31368, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03345271572470665, "train_loss_bc": 0.033449381589889526, "train_loss_llm": 0.00033354759216308594, "grad_norm": 0.04255068302154541, "global_step": 31369, "epoch": 871, "lr": 0.001} +{"train_loss": 0.02540094219148159, "train_loss_bc": 0.025393249467015266, "train_loss_llm": 0.0007693171501159668, "grad_norm": 0.054955460131168365, "global_step": 31370, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03292202576994896, "train_loss_bc": 0.03292202576994896, "train_loss_llm": 0.0, "grad_norm": 0.0803852304816246, "global_step": 31371, "epoch": 871, "lr": 0.001} +{"train_loss": 0.027990851551294327, "train_loss_bc": 0.027990851551294327, "train_loss_llm": 0.0, "grad_norm": 0.09218518435955048, "global_step": 31372, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03330196440219879, "train_loss_bc": 0.03330196440219879, "train_loss_llm": 0.0, "grad_norm": 0.03622293844819069, "global_step": 31373, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03202040120959282, "train_loss_bc": 0.03202040120959282, "train_loss_llm": 0.0, "grad_norm": 0.06124861165881157, "global_step": 31374, "epoch": 871, "lr": 0.001} +{"train_loss": 0.0260878074914217, "train_loss_bc": 0.0260878074914217, "train_loss_llm": 0.0, "grad_norm": 0.08036287873983383, "global_step": 31375, "epoch": 871, "lr": 0.001} +{"train_loss": 0.027925653383135796, "train_loss_bc": 0.027925653383135796, "train_loss_llm": 0.0, "grad_norm": 0.09744114428758621, "global_step": 31376, "epoch": 871, "lr": 0.001} +{"train_loss": 0.025384902954101562, "train_loss_bc": 0.02537326142191887, "train_loss_llm": 0.0011641085147857666, "grad_norm": 0.030529260635375977, "global_step": 31377, "epoch": 871, "lr": 0.001} +{"train_loss": 0.028154445812106133, "train_loss_bc": 0.028147397562861443, "train_loss_llm": 0.0007048845291137695, "grad_norm": 0.049650516360998154, "global_step": 31378, "epoch": 871, "lr": 0.001} +{"train_loss": 0.02904627099633217, "train_loss_bc": 0.02904627099633217, "train_loss_llm": 0.0, "grad_norm": 0.07799714803695679, "global_step": 31379, "epoch": 871, "lr": 0.001} +{"train_loss": 0.027933305129408836, "train_loss_bc": 0.027933305129408836, "train_loss_llm": 0.0, "grad_norm": 0.09606357663869858, "global_step": 31380, "epoch": 871, "lr": 0.001} +{"train_loss": 0.029543645679950714, "train_loss_bc": 0.029543645679950714, "train_loss_llm": 0.0, "grad_norm": 0.039554573595523834, "global_step": 31381, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03251822292804718, "train_loss_bc": 0.03249780088663101, "train_loss_llm": 0.0020422637462615967, "grad_norm": 0.07092194259166718, "global_step": 31382, "epoch": 871, "lr": 0.001} +{"train_loss": 0.029187645763158798, "train_loss_bc": 0.02918647602200508, "train_loss_llm": 0.0001169741153717041, "grad_norm": 0.08870024234056473, "global_step": 31383, "epoch": 871, "lr": 0.001} +{"train_loss": 0.029661113396286964, "train_loss_bc": 0.029661113396286964, "train_loss_llm": 0.0, "grad_norm": 0.11848364025354385, "global_step": 31384, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03541094809770584, "train_loss_bc": 0.03541094809770584, "train_loss_llm": 0.0, "grad_norm": 0.04595119506120682, "global_step": 31385, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03041219711303711, "train_loss_bc": 0.030405642464756966, "train_loss_llm": 0.0006554126739501953, "grad_norm": 0.06414077430963516, "global_step": 31386, "epoch": 871, "lr": 0.001} +{"train_loss": 0.030216146260499954, "train_loss_bc": 0.030216146260499954, "train_loss_llm": 0.0, "grad_norm": 0.08698859810829163, "global_step": 31387, "epoch": 871, "lr": 0.001} +{"train_loss": 0.02542446181178093, "train_loss_bc": 0.02542446181178093, "train_loss_llm": 0.0, "grad_norm": 0.09933020174503326, "global_step": 31388, "epoch": 871, "lr": 0.001} +{"train_loss": 0.026686064898967743, "train_loss_bc": 0.026672575622797012, "train_loss_llm": 0.001348942518234253, "grad_norm": 0.031618326902389526, "global_step": 31389, "epoch": 871, "lr": 0.001} +{"train_loss": 0.03063710778951645, "train_loss_bc": 0.03063710778951645, "train_loss_llm": 0.0, "grad_norm": 0.06028064712882042, "global_step": 31390, "epoch": 871, "lr": 0.001} +{"train_loss": 0.029822448165052466, "train_loss_bc": 0.030467962846159935, "train_loss_llm": 0.0, "grad_norm": 0.09346569329500198, "global_step": 31391, "epoch": 871, "lr": 0.001, "val_loss": 0.10251912474632263} +{"train_loss": 0.02992478758096695, "train_loss_bc": 0.02992362715303898, "train_loss_llm": 0.00011599063873291016, "grad_norm": 0.11347288638353348, "global_step": 31392, "epoch": 872, "lr": 0.001} +{"train_loss": 0.029825812205672264, "train_loss_bc": 0.029825812205672264, "train_loss_llm": 0.0, "grad_norm": 0.03890248015522957, "global_step": 31393, "epoch": 872, "lr": 0.001} +{"train_loss": 0.0293201245367527, "train_loss_bc": 0.0293201245367527, "train_loss_llm": 0.0, "grad_norm": 0.05299461632966995, "global_step": 31394, "epoch": 872, "lr": 0.001} +{"train_loss": 0.02625281922519207, "train_loss_bc": 0.02625281922519207, "train_loss_llm": 0.0, "grad_norm": 0.07177764177322388, "global_step": 31395, "epoch": 872, "lr": 0.001} +{"train_loss": 0.026850219815969467, "train_loss_bc": 0.02684183605015278, "train_loss_llm": 0.0008383989334106445, "grad_norm": 0.09024545550346375, "global_step": 31396, "epoch": 872, "lr": 0.001} +{"train_loss": 0.02468869835138321, "train_loss_bc": 0.02468869835138321, "train_loss_llm": 0.0, "grad_norm": 0.03290349990129471, "global_step": 31397, "epoch": 872, "lr": 0.001} +{"train_loss": 0.02920086495578289, "train_loss_bc": 0.02920086495578289, "train_loss_llm": 0.0, "grad_norm": 0.05180378258228302, "global_step": 31398, "epoch": 872, "lr": 0.001} +{"train_loss": 0.034396637231111526, "train_loss_bc": 0.034396637231111526, "train_loss_llm": 0.0, "grad_norm": 0.07350903004407883, "global_step": 31399, "epoch": 872, "lr": 0.001} +{"train_loss": 0.03230561688542366, "train_loss_bc": 0.03230561688542366, "train_loss_llm": 0.0, "grad_norm": 0.09228289872407913, "global_step": 31400, "epoch": 872, "lr": 0.001} +{"train_loss": 0.029288161545991898, "train_loss_bc": 0.029288161545991898, "train_loss_llm": 0.0, "grad_norm": 0.05152323469519615, "global_step": 31401, "epoch": 872, "lr": 0.001} +{"train_loss": 0.02912961319088936, "train_loss_bc": 0.02912086807191372, "train_loss_llm": 0.0008744895458221436, "grad_norm": 0.0714067816734314, "global_step": 31402, "epoch": 872, "lr": 0.001} +{"train_loss": 0.028796039521694183, "train_loss_bc": 0.028793344274163246, "train_loss_llm": 0.0002695024013519287, "grad_norm": 0.0872056782245636, "global_step": 31403, "epoch": 872, "lr": 0.001} +{"train_loss": 0.026866551488637924, "train_loss_bc": 0.026866551488637924, "train_loss_llm": 0.0, "grad_norm": 0.11069180071353912, "global_step": 31404, "epoch": 872, "lr": 0.001} +{"train_loss": 0.03226728364825249, "train_loss_bc": 0.03226728364825249, "train_loss_llm": 0.0, "grad_norm": 0.037072621285915375, "global_step": 31405, "epoch": 872, "lr": 0.001} +{"train_loss": 0.030524395406246185, "train_loss_bc": 0.030524395406246185, "train_loss_llm": 0.0, "grad_norm": 0.06552023440599442, "global_step": 31406, "epoch": 872, "lr": 0.001} +{"train_loss": 0.02556665986776352, "train_loss_bc": 0.02554948627948761, "train_loss_llm": 0.0017174184322357178, "grad_norm": 0.08833426982164383, "global_step": 31407, "epoch": 872, "lr": 0.001} +{"train_loss": 0.033093616366386414, "train_loss_bc": 0.033093616366386414, "train_loss_llm": 0.0, "grad_norm": 0.09606152027845383, "global_step": 31408, "epoch": 872, "lr": 0.001} +{"train_loss": 0.03480803593993187, "train_loss_bc": 0.034806106239557266, "train_loss_llm": 0.00019294023513793945, "grad_norm": 0.04396191984415054, "global_step": 31409, "epoch": 872, "lr": 0.001} +{"train_loss": 0.030786262825131416, "train_loss_bc": 0.0307825468480587, "train_loss_llm": 0.00037160515785217285, "grad_norm": 0.06364505738019943, "global_step": 31410, "epoch": 872, "lr": 0.001} +{"train_loss": 0.0238120686262846, "train_loss_bc": 0.0238120686262846, "train_loss_llm": 0.0, "grad_norm": 0.07824414223432541, "global_step": 31411, "epoch": 872, "lr": 0.001} +{"train_loss": 0.03122292459011078, "train_loss_bc": 0.03122292459011078, "train_loss_llm": 0.0, "grad_norm": 0.09692250192165375, "global_step": 31412, "epoch": 872, "lr": 0.001} +{"train_loss": 0.02907797135412693, "train_loss_bc": 0.02907361462712288, "train_loss_llm": 0.00043570995330810547, "grad_norm": 0.031039685010910034, "global_step": 31413, "epoch": 872, "lr": 0.001} +{"train_loss": 0.035858623683452606, "train_loss_bc": 0.03583850339055061, "train_loss_llm": 0.0020119845867156982, "grad_norm": 0.05685487017035484, "global_step": 31414, "epoch": 872, "lr": 0.001} +{"train_loss": 0.030338600277900696, "train_loss_bc": 0.030329905450344086, "train_loss_llm": 0.0008695423603057861, "grad_norm": 0.07089541852474213, "global_step": 31415, "epoch": 872, "lr": 0.001} +{"train_loss": 0.03006747178733349, "train_loss_bc": 0.0300611425191164, "train_loss_llm": 0.000632941722869873, "grad_norm": 0.09834693372249603, "global_step": 31416, "epoch": 872, "lr": 0.001} +{"train_loss": 0.02862967550754547, "train_loss_bc": 0.02862967550754547, "train_loss_llm": 0.0, "grad_norm": 0.038156911730766296, "global_step": 31417, "epoch": 872, "lr": 0.001} +{"train_loss": 0.02753806672990322, "train_loss_bc": 0.02753806672990322, "train_loss_llm": 0.0, "grad_norm": 0.05239395424723625, "global_step": 31418, "epoch": 872, "lr": 0.001} +{"train_loss": 0.025769954547286034, "train_loss_bc": 0.025769954547286034, "train_loss_llm": 0.0, "grad_norm": 0.06949663907289505, "global_step": 31419, "epoch": 872, "lr": 0.001} +{"train_loss": 0.028932254761457443, "train_loss_bc": 0.028932254761457443, "train_loss_llm": 0.0, "grad_norm": 0.08426955342292786, "global_step": 31420, "epoch": 872, "lr": 0.001} +{"train_loss": 0.028450455516576767, "train_loss_bc": 0.028450455516576767, "train_loss_llm": 0.0, "grad_norm": 0.0376453623175621, "global_step": 31421, "epoch": 872, "lr": 0.001} +{"train_loss": 0.02904590032994747, "train_loss_bc": 0.02904590032994747, "train_loss_llm": 0.0, "grad_norm": 0.06390024721622467, "global_step": 31422, "epoch": 872, "lr": 0.001} +{"train_loss": 0.03277656435966492, "train_loss_bc": 0.03276875987648964, "train_loss_llm": 0.0007804334163665771, "grad_norm": 0.09221666306257248, "global_step": 31423, "epoch": 872, "lr": 0.001} +{"train_loss": 0.03043263591825962, "train_loss_bc": 0.03043263591825962, "train_loss_llm": 0.0, "grad_norm": 0.11539245396852493, "global_step": 31424, "epoch": 872, "lr": 0.001} +{"train_loss": 0.030078524723649025, "train_loss_bc": 0.030078524723649025, "train_loss_llm": 0.0, "grad_norm": 0.04057902470231056, "global_step": 31425, "epoch": 872, "lr": 0.001} +{"train_loss": 0.031280141323804855, "train_loss_bc": 0.03127837926149368, "train_loss_llm": 0.00017631053924560547, "grad_norm": 0.06445767730474472, "global_step": 31426, "epoch": 872, "lr": 0.001} +{"train_loss": 0.029561297761069402, "train_loss_bc": 0.027002684772014618, "train_loss_llm": 0.0, "grad_norm": 0.09171030670404434, "global_step": 31427, "epoch": 872, "lr": 0.001, "val_loss": 0.10290399193763733} +{"train_loss": 0.030762283131480217, "train_loss_bc": 0.030753839761018753, "train_loss_llm": 0.0008442699909210205, "grad_norm": 0.1080181673169136, "global_step": 31428, "epoch": 873, "lr": 0.001} +{"train_loss": 0.02746228687465191, "train_loss_bc": 0.02746228687465191, "train_loss_llm": 0.0, "grad_norm": 0.03050648793578148, "global_step": 31429, "epoch": 873, "lr": 0.001} +{"train_loss": 0.02846303954720497, "train_loss_bc": 0.028461087495088577, "train_loss_llm": 0.0001952052116394043, "grad_norm": 0.05706063285470009, "global_step": 31430, "epoch": 873, "lr": 0.001} +{"train_loss": 0.025539133697748184, "train_loss_bc": 0.025539133697748184, "train_loss_llm": 0.0, "grad_norm": 0.06459900736808777, "global_step": 31431, "epoch": 873, "lr": 0.001} +{"train_loss": 0.03514964133501053, "train_loss_bc": 0.03514964133501053, "train_loss_llm": 0.0, "grad_norm": 0.09140872955322266, "global_step": 31432, "epoch": 873, "lr": 0.001} +{"train_loss": 0.02969278208911419, "train_loss_bc": 0.02969278208911419, "train_loss_llm": 0.0, "grad_norm": 0.030404740944504738, "global_step": 31433, "epoch": 873, "lr": 0.001} +{"train_loss": 0.03695647045969963, "train_loss_bc": 0.03695647045969963, "train_loss_llm": 0.0, "grad_norm": 0.057676706463098526, "global_step": 31434, "epoch": 873, "lr": 0.001} +{"train_loss": 0.028795938938856125, "train_loss_bc": 0.028784973546862602, "train_loss_llm": 0.001096576452255249, "grad_norm": 0.0760759711265564, "global_step": 31435, "epoch": 873, "lr": 0.001} +{"train_loss": 0.02727619558572769, "train_loss_bc": 0.02727619558572769, "train_loss_llm": 0.0, "grad_norm": 0.08545539528131485, "global_step": 31436, "epoch": 873, "lr": 0.001} +{"train_loss": 0.03391638025641441, "train_loss_bc": 0.03391638025641441, "train_loss_llm": 0.0, "grad_norm": 0.042460426688194275, "global_step": 31437, "epoch": 873, "lr": 0.001} +{"train_loss": 0.030079852789640427, "train_loss_bc": 0.03007439710199833, "train_loss_llm": 0.0005455613136291504, "grad_norm": 0.05824202299118042, "global_step": 31438, "epoch": 873, "lr": 0.001} +{"train_loss": 0.025287220254540443, "train_loss_bc": 0.025287220254540443, "train_loss_llm": 0.0, "grad_norm": 0.06753959506750107, "global_step": 31439, "epoch": 873, "lr": 0.001} +{"train_loss": 0.025008443742990494, "train_loss_bc": 0.02499799244105816, "train_loss_llm": 0.0010451972484588623, "grad_norm": 0.07971469312906265, "global_step": 31440, "epoch": 873, "lr": 0.001} +{"train_loss": 0.029894933104515076, "train_loss_bc": 0.029894933104515076, "train_loss_llm": 0.0, "grad_norm": 0.03728122636675835, "global_step": 31441, "epoch": 873, "lr": 0.001} +{"train_loss": 0.03633720800280571, "train_loss_bc": 0.03633720800280571, "train_loss_llm": 0.0, "grad_norm": 0.06752151250839233, "global_step": 31442, "epoch": 873, "lr": 0.001} +{"train_loss": 0.02665902115404606, "train_loss_bc": 0.026653246954083443, "train_loss_llm": 0.0005774796009063721, "grad_norm": 0.08221916854381561, "global_step": 31443, "epoch": 873, "lr": 0.001} +{"train_loss": 0.027706561610102654, "train_loss_bc": 0.027706561610102654, "train_loss_llm": 0.0, "grad_norm": 0.0953800156712532, "global_step": 31444, "epoch": 873, "lr": 0.001} +{"train_loss": 0.025928016752004623, "train_loss_bc": 0.02591763064265251, "train_loss_llm": 0.0010385215282440186, "grad_norm": 0.03883647918701172, "global_step": 31445, "epoch": 873, "lr": 0.001} +{"train_loss": 0.029894595965743065, "train_loss_bc": 0.029894595965743065, "train_loss_llm": 0.0, "grad_norm": 0.07019916921854019, "global_step": 31446, "epoch": 873, "lr": 0.001} +{"train_loss": 0.031598735600709915, "train_loss_bc": 0.031598735600709915, "train_loss_llm": 0.0, "grad_norm": 0.103364497423172, "global_step": 31447, "epoch": 873, "lr": 0.001} +{"train_loss": 0.030013656243681908, "train_loss_bc": 0.030001141130924225, "train_loss_llm": 0.0012514293193817139, "grad_norm": 0.1257188618183136, "global_step": 31448, "epoch": 873, "lr": 0.001} +{"train_loss": 0.02896018885076046, "train_loss_bc": 0.02896018885076046, "train_loss_llm": 0.0, "grad_norm": 0.0310510266572237, "global_step": 31449, "epoch": 873, "lr": 0.001} +{"train_loss": 0.027351293712854385, "train_loss_bc": 0.027351293712854385, "train_loss_llm": 0.0, "grad_norm": 0.05312204360961914, "global_step": 31450, "epoch": 873, "lr": 0.001} +{"train_loss": 0.029901983216404915, "train_loss_bc": 0.029901983216404915, "train_loss_llm": 0.0, "grad_norm": 0.07158408313989639, "global_step": 31451, "epoch": 873, "lr": 0.001} +{"train_loss": 0.02762588858604431, "train_loss_bc": 0.027619870379567146, "train_loss_llm": 0.0006017684936523438, "grad_norm": 0.09099620580673218, "global_step": 31452, "epoch": 873, "lr": 0.001} +{"train_loss": 0.025704607367515564, "train_loss_bc": 0.02569696307182312, "train_loss_llm": 0.0007644593715667725, "grad_norm": 0.029471011832356453, "global_step": 31453, "epoch": 873, "lr": 0.001} +{"train_loss": 0.023509811609983444, "train_loss_bc": 0.023491445928812027, "train_loss_llm": 0.0018366575241088867, "grad_norm": 0.04724644869565964, "global_step": 31454, "epoch": 873, "lr": 0.001} +{"train_loss": 0.031307075172662735, "train_loss_bc": 0.031307075172662735, "train_loss_llm": 0.0, "grad_norm": 0.06299764662981033, "global_step": 31455, "epoch": 873, "lr": 0.001} +{"train_loss": 0.029392825439572334, "train_loss_bc": 0.029392825439572334, "train_loss_llm": 0.0, "grad_norm": 0.07515944540500641, "global_step": 31456, "epoch": 873, "lr": 0.001} +{"train_loss": 0.0273294597864151, "train_loss_bc": 0.0273294597864151, "train_loss_llm": 0.0, "grad_norm": 0.029275285080075264, "global_step": 31457, "epoch": 873, "lr": 0.001} +{"train_loss": 0.03313576057553291, "train_loss_bc": 0.03310351073741913, "train_loss_llm": 0.003224879503250122, "grad_norm": 0.04763634502887726, "global_step": 31458, "epoch": 873, "lr": 0.001} +{"train_loss": 0.03941825404763222, "train_loss_bc": 0.03941825404763222, "train_loss_llm": 0.0, "grad_norm": 0.06714458018541336, "global_step": 31459, "epoch": 873, "lr": 0.001} +{"train_loss": 0.028670432046055794, "train_loss_bc": 0.028670432046055794, "train_loss_llm": 0.0, "grad_norm": 0.08509396761655807, "global_step": 31460, "epoch": 873, "lr": 0.001} +{"train_loss": 0.03362929821014404, "train_loss_bc": 0.033622968941926956, "train_loss_llm": 0.0006329119205474854, "grad_norm": 0.03529127314686775, "global_step": 31461, "epoch": 873, "lr": 0.001} +{"train_loss": 0.02444501593708992, "train_loss_bc": 0.024430716410279274, "train_loss_llm": 0.001429915428161621, "grad_norm": 0.05319368839263916, "global_step": 31462, "epoch": 873, "lr": 0.001} +{"train_loss": 0.029382784747415118, "train_loss_bc": 0.024975959211587906, "train_loss_llm": 0.0, "grad_norm": 0.07386744022369385, "global_step": 31463, "epoch": 873, "lr": 0.001, "val_loss": 0.10059177875518799} +{"train_loss": 0.02544332854449749, "train_loss_bc": 0.02544332854449749, "train_loss_llm": 0.0, "grad_norm": 0.0959659293293953, "global_step": 31464, "epoch": 874, "lr": 0.001} +{"train_loss": 0.024000247940421104, "train_loss_bc": 0.024000247940421104, "train_loss_llm": 0.0, "grad_norm": 0.03182707354426384, "global_step": 31465, "epoch": 874, "lr": 0.001} +{"train_loss": 0.03180252015590668, "train_loss_bc": 0.03180252015590668, "train_loss_llm": 0.0, "grad_norm": 0.05220849812030792, "global_step": 31466, "epoch": 874, "lr": 0.001} +{"train_loss": 0.03363285958766937, "train_loss_bc": 0.03363285958766937, "train_loss_llm": 0.0, "grad_norm": 0.08430169522762299, "global_step": 31467, "epoch": 874, "lr": 0.001} +{"train_loss": 0.026817113161087036, "train_loss_bc": 0.026817113161087036, "train_loss_llm": 0.0, "grad_norm": 0.10628298670053482, "global_step": 31468, "epoch": 874, "lr": 0.001} +{"train_loss": 0.028211968019604683, "train_loss_bc": 0.028211968019604683, "train_loss_llm": 0.0, "grad_norm": 0.03649064525961876, "global_step": 31469, "epoch": 874, "lr": 0.001} +{"train_loss": 0.026627033948898315, "train_loss_bc": 0.02662699669599533, "train_loss_llm": 3.635883331298828e-06, "grad_norm": 0.0528661273419857, "global_step": 31470, "epoch": 874, "lr": 0.001} +{"train_loss": 0.028141574934124947, "train_loss_bc": 0.028141574934124947, "train_loss_llm": 0.0, "grad_norm": 0.0725734531879425, "global_step": 31471, "epoch": 874, "lr": 0.001} +{"train_loss": 0.027846822515130043, "train_loss_bc": 0.027846822515130043, "train_loss_llm": 0.0, "grad_norm": 0.08527086675167084, "global_step": 31472, "epoch": 874, "lr": 0.001} +{"train_loss": 0.029933355748653412, "train_loss_bc": 0.029933355748653412, "train_loss_llm": 0.0, "grad_norm": 0.04012451320886612, "global_step": 31473, "epoch": 874, "lr": 0.001} +{"train_loss": 0.03457169979810715, "train_loss_bc": 0.03457169979810715, "train_loss_llm": 0.0, "grad_norm": 0.06212190166115761, "global_step": 31474, "epoch": 874, "lr": 0.001} +{"train_loss": 0.023762159049510956, "train_loss_bc": 0.023762159049510956, "train_loss_llm": 0.0, "grad_norm": 0.0757325142621994, "global_step": 31475, "epoch": 874, "lr": 0.001} +{"train_loss": 0.0308869406580925, "train_loss_bc": 0.030878592282533646, "train_loss_llm": 0.0008348822593688965, "grad_norm": 0.08618012815713882, "global_step": 31476, "epoch": 874, "lr": 0.001} +{"train_loss": 0.028208769857883453, "train_loss_bc": 0.028208769857883453, "train_loss_llm": 0.0, "grad_norm": 0.03421427309513092, "global_step": 31477, "epoch": 874, "lr": 0.001} +{"train_loss": 0.031245823949575424, "train_loss_bc": 0.031245823949575424, "train_loss_llm": 0.0, "grad_norm": 0.06129887327551842, "global_step": 31478, "epoch": 874, "lr": 0.001} +{"train_loss": 0.024662809446454048, "train_loss_bc": 0.024662809446454048, "train_loss_llm": 0.0, "grad_norm": 0.07804736495018005, "global_step": 31479, "epoch": 874, "lr": 0.001} +{"train_loss": 0.027638837695121765, "train_loss_bc": 0.02763097546994686, "train_loss_llm": 0.0007861554622650146, "grad_norm": 0.09907808899879456, "global_step": 31480, "epoch": 874, "lr": 0.001} +{"train_loss": 0.03516456484794617, "train_loss_bc": 0.03516456484794617, "train_loss_llm": 0.0, "grad_norm": 0.04123176634311676, "global_step": 31481, "epoch": 874, "lr": 0.001} +{"train_loss": 0.031080104410648346, "train_loss_bc": 0.031074227765202522, "train_loss_llm": 0.0005877017974853516, "grad_norm": 0.06805958598852158, "global_step": 31482, "epoch": 874, "lr": 0.001} +{"train_loss": 0.03299862891435623, "train_loss_bc": 0.03299862891435623, "train_loss_llm": 0.0, "grad_norm": 0.0996919795870781, "global_step": 31483, "epoch": 874, "lr": 0.001} +{"train_loss": 0.030198346823453903, "train_loss_bc": 0.030198346823453903, "train_loss_llm": 0.0, "grad_norm": 0.1274113804101944, "global_step": 31484, "epoch": 874, "lr": 0.001} +{"train_loss": 0.031907521188259125, "train_loss_bc": 0.031907521188259125, "train_loss_llm": 0.0, "grad_norm": 0.041753336787223816, "global_step": 31485, "epoch": 874, "lr": 0.001} +{"train_loss": 0.028996454551815987, "train_loss_bc": 0.028996454551815987, "train_loss_llm": 0.0, "grad_norm": 0.06743182986974716, "global_step": 31486, "epoch": 874, "lr": 0.001} +{"train_loss": 0.02802450954914093, "train_loss_bc": 0.0280243419110775, "train_loss_llm": 1.671910285949707e-05, "grad_norm": 0.09430092573165894, "global_step": 31487, "epoch": 874, "lr": 0.001} +{"train_loss": 0.031175313517451286, "train_loss_bc": 0.031175313517451286, "train_loss_llm": 0.0, "grad_norm": 0.12653550505638123, "global_step": 31488, "epoch": 874, "lr": 0.001} +{"train_loss": 0.03030485287308693, "train_loss_bc": 0.03030485287308693, "train_loss_llm": 0.0, "grad_norm": 0.03451154753565788, "global_step": 31489, "epoch": 874, "lr": 0.001} +{"train_loss": 0.031594417989254, "train_loss_bc": 0.031594417989254, "train_loss_llm": 0.0, "grad_norm": 0.05314863100647926, "global_step": 31490, "epoch": 874, "lr": 0.001} +{"train_loss": 0.028254911303520203, "train_loss_bc": 0.028254911303520203, "train_loss_llm": 0.0, "grad_norm": 0.07608654350042343, "global_step": 31491, "epoch": 874, "lr": 0.001} +{"train_loss": 0.032345835119485855, "train_loss_bc": 0.03234293311834335, "train_loss_llm": 0.00029030442237854004, "grad_norm": 0.10019061714410782, "global_step": 31492, "epoch": 874, "lr": 0.001} +{"train_loss": 0.027783531695604324, "train_loss_bc": 0.027783531695604324, "train_loss_llm": 0.0, "grad_norm": 0.03125401958823204, "global_step": 31493, "epoch": 874, "lr": 0.001} +{"train_loss": 0.030782336369156837, "train_loss_bc": 0.030782336369156837, "train_loss_llm": 0.0, "grad_norm": 0.054308824241161346, "global_step": 31494, "epoch": 874, "lr": 0.001} +{"train_loss": 0.03233679383993149, "train_loss_bc": 0.03233679383993149, "train_loss_llm": 0.0, "grad_norm": 0.06501521170139313, "global_step": 31495, "epoch": 874, "lr": 0.001} +{"train_loss": 0.025488704442977905, "train_loss_bc": 0.025488704442977905, "train_loss_llm": 0.0, "grad_norm": 0.07776322960853577, "global_step": 31496, "epoch": 874, "lr": 0.001} +{"train_loss": 0.03361666202545166, "train_loss_bc": 0.03361666202545166, "train_loss_llm": 0.0, "grad_norm": 0.0428316630423069, "global_step": 31497, "epoch": 874, "lr": 0.001} +{"train_loss": 0.027159295976161957, "train_loss_bc": 0.027153749018907547, "train_loss_llm": 0.0005546510219573975, "grad_norm": 0.058527328073978424, "global_step": 31498, "epoch": 874, "lr": 0.001} +{"train_loss": 0.029566570754266448, "train_loss_bc": 0.031748849898576736, "train_loss_llm": 0.00010481476783752441, "grad_norm": 0.08272933214902878, "global_step": 31499, "epoch": 874, "lr": 0.001, "val_loss": 0.10139437019824982} +{"train_loss": 0.030667435377836227, "train_loss_bc": 0.030667435377836227, "train_loss_llm": 0.0, "grad_norm": 0.09673723578453064, "global_step": 31500, "epoch": 875, "lr": 0.001} +{"train_loss": 0.03404464200139046, "train_loss_bc": 0.03404464200139046, "train_loss_llm": 0.0, "grad_norm": 0.03750775009393692, "global_step": 31501, "epoch": 875, "lr": 0.001} +{"train_loss": 0.034630514681339264, "train_loss_bc": 0.034630514681339264, "train_loss_llm": 0.0, "grad_norm": 0.06775673478841782, "global_step": 31502, "epoch": 875, "lr": 0.001} +{"train_loss": 0.023423854261636734, "train_loss_bc": 0.023423854261636734, "train_loss_llm": 0.0, "grad_norm": 0.0833667740225792, "global_step": 31503, "epoch": 875, "lr": 0.001} +{"train_loss": 0.025563398376107216, "train_loss_bc": 0.025563398376107216, "train_loss_llm": 0.0, "grad_norm": 0.10512414574623108, "global_step": 31504, "epoch": 875, "lr": 0.001} +{"train_loss": 0.029605185613036156, "train_loss_bc": 0.029605185613036156, "train_loss_llm": 0.0, "grad_norm": 0.03556437790393829, "global_step": 31505, "epoch": 875, "lr": 0.001} +{"train_loss": 0.02984917163848877, "train_loss_bc": 0.02984917163848877, "train_loss_llm": 0.0, "grad_norm": 0.058060068637132645, "global_step": 31506, "epoch": 875, "lr": 0.001} +{"train_loss": 0.02552974410355091, "train_loss_bc": 0.02552574686706066, "train_loss_llm": 0.00039967894554138184, "grad_norm": 0.07021094113588333, "global_step": 31507, "epoch": 875, "lr": 0.001} +{"train_loss": 0.026910966262221336, "train_loss_bc": 0.026910966262221336, "train_loss_llm": 0.0, "grad_norm": 0.09005942195653915, "global_step": 31508, "epoch": 875, "lr": 0.001} +{"train_loss": 0.02780083194375038, "train_loss_bc": 0.02780083194375038, "train_loss_llm": 0.0, "grad_norm": 0.03728390857577324, "global_step": 31509, "epoch": 875, "lr": 0.001} +{"train_loss": 0.03036273643374443, "train_loss_bc": 0.03036273643374443, "train_loss_llm": 0.0, "grad_norm": 0.06492016464471817, "global_step": 31510, "epoch": 875, "lr": 0.001} +{"train_loss": 0.02782319486141205, "train_loss_bc": 0.02782319486141205, "train_loss_llm": 0.0, "grad_norm": 0.09050891548395157, "global_step": 31511, "epoch": 875, "lr": 0.001} +{"train_loss": 0.029091572389006615, "train_loss_bc": 0.029091572389006615, "train_loss_llm": 0.0, "grad_norm": 0.11202415078878403, "global_step": 31512, "epoch": 875, "lr": 0.001} +{"train_loss": 0.030432086437940598, "train_loss_bc": 0.030414940789341927, "train_loss_llm": 0.0017146468162536621, "grad_norm": 0.03572255000472069, "global_step": 31513, "epoch": 875, "lr": 0.001} +{"train_loss": 0.027828237041831017, "train_loss_bc": 0.027828237041831017, "train_loss_llm": 0.0, "grad_norm": 0.04588347673416138, "global_step": 31514, "epoch": 875, "lr": 0.001} +{"train_loss": 0.029315799474716187, "train_loss_bc": 0.029310598969459534, "train_loss_llm": 0.0005200505256652832, "grad_norm": 0.07076045870780945, "global_step": 31515, "epoch": 875, "lr": 0.001} +{"train_loss": 0.027064047753810883, "train_loss_bc": 0.02706107124686241, "train_loss_llm": 0.0002976953983306885, "grad_norm": 0.09802056849002838, "global_step": 31516, "epoch": 875, "lr": 0.001} +{"train_loss": 0.02791181206703186, "train_loss_bc": 0.02791181206703186, "train_loss_llm": 0.0, "grad_norm": 0.03144480288028717, "global_step": 31517, "epoch": 875, "lr": 0.001} +{"train_loss": 0.0231609046459198, "train_loss_bc": 0.023148847743868828, "train_loss_llm": 0.0012056529521942139, "grad_norm": 0.05043984577059746, "global_step": 31518, "epoch": 875, "lr": 0.001} +{"train_loss": 0.02864355593919754, "train_loss_bc": 0.02863762527704239, "train_loss_llm": 0.0005930662155151367, "grad_norm": 0.06724537163972855, "global_step": 31519, "epoch": 875, "lr": 0.001} +{"train_loss": 0.0212501622736454, "train_loss_bc": 0.0212501622736454, "train_loss_llm": 0.0, "grad_norm": 0.08506058901548386, "global_step": 31520, "epoch": 875, "lr": 0.001} +{"train_loss": 0.030035819858312607, "train_loss_bc": 0.030010050162672997, "train_loss_llm": 0.002576977014541626, "grad_norm": 0.03688009828329086, "global_step": 31521, "epoch": 875, "lr": 0.001} +{"train_loss": 0.03248443827033043, "train_loss_bc": 0.03248443827033043, "train_loss_llm": 0.0, "grad_norm": 0.05580471083521843, "global_step": 31522, "epoch": 875, "lr": 0.001} +{"train_loss": 0.027509214356541634, "train_loss_bc": 0.027509214356541634, "train_loss_llm": 0.0, "grad_norm": 0.08019138872623444, "global_step": 31523, "epoch": 875, "lr": 0.001} +{"train_loss": 0.03538835048675537, "train_loss_bc": 0.03538835048675537, "train_loss_llm": 0.0, "grad_norm": 0.11009575426578522, "global_step": 31524, "epoch": 875, "lr": 0.001} +{"train_loss": 0.029059510678052902, "train_loss_bc": 0.029059510678052902, "train_loss_llm": 0.0, "grad_norm": 0.037657901644706726, "global_step": 31525, "epoch": 875, "lr": 0.001} +{"train_loss": 0.03444606438279152, "train_loss_bc": 0.03440896049141884, "train_loss_llm": 0.0037104785442352295, "grad_norm": 0.06369568407535553, "global_step": 31526, "epoch": 875, "lr": 0.001} +{"train_loss": 0.03701892867684364, "train_loss_bc": 0.03701892867684364, "train_loss_llm": 0.0, "grad_norm": 0.10243348777294159, "global_step": 31527, "epoch": 875, "lr": 0.001} +{"train_loss": 0.02727152593433857, "train_loss_bc": 0.02727152593433857, "train_loss_llm": 0.0, "grad_norm": 0.10921551287174225, "global_step": 31528, "epoch": 875, "lr": 0.001} +{"train_loss": 0.029264960438013077, "train_loss_bc": 0.029264960438013077, "train_loss_llm": 0.0, "grad_norm": 0.036164820194244385, "global_step": 31529, "epoch": 875, "lr": 0.001} +{"train_loss": 0.03345811367034912, "train_loss_bc": 0.03345811367034912, "train_loss_llm": 0.0, "grad_norm": 0.053312622010707855, "global_step": 31530, "epoch": 875, "lr": 0.001} +{"train_loss": 0.02703048847615719, "train_loss_bc": 0.027017559856176376, "train_loss_llm": 0.0012928545475006104, "grad_norm": 0.07105804234743118, "global_step": 31531, "epoch": 875, "lr": 0.001} +{"train_loss": 0.03352558612823486, "train_loss_bc": 0.03352558612823486, "train_loss_llm": 0.0, "grad_norm": 0.08647619187831879, "global_step": 31532, "epoch": 875, "lr": 0.001} +{"train_loss": 0.033176396042108536, "train_loss_bc": 0.03317522257566452, "train_loss_llm": 0.00011739134788513184, "grad_norm": 0.04360390454530716, "global_step": 31533, "epoch": 875, "lr": 0.001} +{"train_loss": 0.03480233624577522, "train_loss_bc": 0.03480233624577522, "train_loss_llm": 0.0, "grad_norm": 0.08338663727045059, "global_step": 31534, "epoch": 875, "lr": 0.001} +{"train_loss": 0.02955869347270992, "train_loss_bc": 0.02873137779533863, "train_loss_llm": 0.0, "grad_norm": 0.11047576367855072, "global_step": 31535, "epoch": 875, "lr": 0.001, "train/cumulative_reward": 108.3315596339806, "train/mean_score": 12.8177844982486, "train/success_rate": 0.3, "test/cumulative_reward": 39.25928118583451, "test/mean_score": 7.410907929854264, "test/success_rate": 0.18, "val_loss": 0.10301586240530014, "train_action_mse_error": 0.032829370349645615} +{"train_loss": 0.036246318370103836, "train_loss_bc": 0.036221571266651154, "train_loss_llm": 0.002474844455718994, "grad_norm": 0.14273671805858612, "global_step": 31536, "epoch": 876, "lr": 0.001} +{"train_loss": 0.027907203882932663, "train_loss_bc": 0.027907203882932663, "train_loss_llm": 0.0, "grad_norm": 0.032802194356918335, "global_step": 31537, "epoch": 876, "lr": 0.001} +{"train_loss": 0.0315755158662796, "train_loss_bc": 0.0315755158662796, "train_loss_llm": 0.0, "grad_norm": 0.054142437875270844, "global_step": 31538, "epoch": 876, "lr": 0.001} +{"train_loss": 0.024458961561322212, "train_loss_bc": 0.024458961561322212, "train_loss_llm": 0.0, "grad_norm": 0.07207752764225006, "global_step": 31539, "epoch": 876, "lr": 0.001} +{"train_loss": 0.027708806097507477, "train_loss_bc": 0.027708806097507477, "train_loss_llm": 0.0, "grad_norm": 0.0968308299779892, "global_step": 31540, "epoch": 876, "lr": 0.001} +{"train_loss": 0.028693296015262604, "train_loss_bc": 0.028693296015262604, "train_loss_llm": 0.0, "grad_norm": 0.03423575684428215, "global_step": 31541, "epoch": 876, "lr": 0.001} +{"train_loss": 0.02704850398004055, "train_loss_bc": 0.02702401950955391, "train_loss_llm": 0.0024484992027282715, "grad_norm": 0.04790756478905678, "global_step": 31542, "epoch": 876, "lr": 0.001} +{"train_loss": 0.031520526856184006, "train_loss_bc": 0.031520526856184006, "train_loss_llm": 0.0, "grad_norm": 0.07419522106647491, "global_step": 31543, "epoch": 876, "lr": 0.001} +{"train_loss": 0.03300100192427635, "train_loss_bc": 0.03300100192427635, "train_loss_llm": 0.0, "grad_norm": 0.09179676324129105, "global_step": 31544, "epoch": 876, "lr": 0.001} +{"train_loss": 0.030357448384165764, "train_loss_bc": 0.030357448384165764, "train_loss_llm": 0.0, "grad_norm": 0.03487326204776764, "global_step": 31545, "epoch": 876, "lr": 0.001} +{"train_loss": 0.026552928611636162, "train_loss_bc": 0.026548903435468674, "train_loss_llm": 0.0004024803638458252, "grad_norm": 0.05994100868701935, "global_step": 31546, "epoch": 876, "lr": 0.001} +{"train_loss": 0.0324554517865181, "train_loss_bc": 0.0324554517865181, "train_loss_llm": 0.0, "grad_norm": 0.08412852138280869, "global_step": 31547, "epoch": 876, "lr": 0.001} +{"train_loss": 0.02978101372718811, "train_loss_bc": 0.02978101372718811, "train_loss_llm": 0.0, "grad_norm": 0.1003364771604538, "global_step": 31548, "epoch": 876, "lr": 0.001} +{"train_loss": 0.026880450546741486, "train_loss_bc": 0.026873815804719925, "train_loss_llm": 0.0006634294986724854, "grad_norm": 0.036074668169021606, "global_step": 31549, "epoch": 876, "lr": 0.001} +{"train_loss": 0.031215697526931763, "train_loss_bc": 0.031215697526931763, "train_loss_llm": 0.0, "grad_norm": 0.06317272037267685, "global_step": 31550, "epoch": 876, "lr": 0.001} +{"train_loss": 0.034103378653526306, "train_loss_bc": 0.034103378653526306, "train_loss_llm": 0.0, "grad_norm": 0.09206260740756989, "global_step": 31551, "epoch": 876, "lr": 0.001} +{"train_loss": 0.027922719717025757, "train_loss_bc": 0.027922719717025757, "train_loss_llm": 0.0, "grad_norm": 0.11392103880643845, "global_step": 31552, "epoch": 876, "lr": 0.001} +{"train_loss": 0.02725246548652649, "train_loss_bc": 0.02725246548652649, "train_loss_llm": 0.0, "grad_norm": 0.03430989012122154, "global_step": 31553, "epoch": 876, "lr": 0.001} +{"train_loss": 0.034974053502082825, "train_loss_bc": 0.034974053502082825, "train_loss_llm": 0.0, "grad_norm": 0.06846761703491211, "global_step": 31554, "epoch": 876, "lr": 0.001} +{"train_loss": 0.030370092019438744, "train_loss_bc": 0.030370092019438744, "train_loss_llm": 0.0, "grad_norm": 0.09353168308734894, "global_step": 31555, "epoch": 876, "lr": 0.001} +{"train_loss": 0.021921910345554352, "train_loss_bc": 0.021921910345554352, "train_loss_llm": 0.0, "grad_norm": 0.11455637961626053, "global_step": 31556, "epoch": 876, "lr": 0.001} +{"train_loss": 0.027240591123700142, "train_loss_bc": 0.027240591123700142, "train_loss_llm": 0.0, "grad_norm": 0.035209085792303085, "global_step": 31557, "epoch": 876, "lr": 0.001} +{"train_loss": 0.02882806770503521, "train_loss_bc": 0.02882806770503521, "train_loss_llm": 0.0, "grad_norm": 0.05905013531446457, "global_step": 31558, "epoch": 876, "lr": 0.001} +{"train_loss": 0.02688603661954403, "train_loss_bc": 0.02687915973365307, "train_loss_llm": 0.0006877779960632324, "grad_norm": 0.07824741303920746, "global_step": 31559, "epoch": 876, "lr": 0.001} +{"train_loss": 0.029945077374577522, "train_loss_bc": 0.029945077374577522, "train_loss_llm": 0.0, "grad_norm": 0.0929289236664772, "global_step": 31560, "epoch": 876, "lr": 0.001} +{"train_loss": 0.03040461055934429, "train_loss_bc": 0.03040461055934429, "train_loss_llm": 0.0, "grad_norm": 0.04246004670858383, "global_step": 31561, "epoch": 876, "lr": 0.001} +{"train_loss": 0.029995517805218697, "train_loss_bc": 0.029995517805218697, "train_loss_llm": 0.0, "grad_norm": 0.0680929645895958, "global_step": 31562, "epoch": 876, "lr": 0.001} +{"train_loss": 0.02983715385198593, "train_loss_bc": 0.029830453917384148, "train_loss_llm": 0.0006699860095977783, "grad_norm": 0.08298906683921814, "global_step": 31563, "epoch": 876, "lr": 0.001} +{"train_loss": 0.027469472959637642, "train_loss_bc": 0.027469472959637642, "train_loss_llm": 0.0, "grad_norm": 0.098792664706707, "global_step": 31564, "epoch": 876, "lr": 0.001} +{"train_loss": 0.027982112020254135, "train_loss_bc": 0.02798108384013176, "train_loss_llm": 0.00010272860527038574, "grad_norm": 0.04063360393047333, "global_step": 31565, "epoch": 876, "lr": 0.001} +{"train_loss": 0.03745945170521736, "train_loss_bc": 0.03745945170521736, "train_loss_llm": 0.0, "grad_norm": 0.07838205248117447, "global_step": 31566, "epoch": 876, "lr": 0.001} +{"train_loss": 0.03009823150932789, "train_loss_bc": 0.03009823150932789, "train_loss_llm": 0.0, "grad_norm": 0.10374805331230164, "global_step": 31567, "epoch": 876, "lr": 0.001} +{"train_loss": 0.02695935219526291, "train_loss_bc": 0.026950836181640625, "train_loss_llm": 0.0008515417575836182, "grad_norm": 0.12608934938907623, "global_step": 31568, "epoch": 876, "lr": 0.001} +{"train_loss": 0.03501680865883827, "train_loss_bc": 0.03501680865883827, "train_loss_llm": 0.0, "grad_norm": 0.041290219873189926, "global_step": 31569, "epoch": 876, "lr": 0.001} +{"train_loss": 0.03339133411645889, "train_loss_bc": 0.03337883949279785, "train_loss_llm": 0.0012495815753936768, "grad_norm": 0.07214616984128952, "global_step": 31570, "epoch": 876, "lr": 0.001} +{"train_loss": 0.02969655389380124, "train_loss_bc": 0.025614377111196518, "train_loss_llm": 0.0, "grad_norm": 0.11458837240934372, "global_step": 31571, "epoch": 876, "lr": 0.001, "val_loss": 0.10470335930585861} +{"train_loss": 0.029491910710930824, "train_loss_bc": 0.029474705457687378, "train_loss_llm": 0.001720517873764038, "grad_norm": 0.1421535164117813, "global_step": 31572, "epoch": 877, "lr": 0.001} +{"train_loss": 0.030726540833711624, "train_loss_bc": 0.030722109600901604, "train_loss_llm": 0.0004431605339050293, "grad_norm": 0.03717833012342453, "global_step": 31573, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03304576501250267, "train_loss_bc": 0.033039938658475876, "train_loss_llm": 0.0005828142166137695, "grad_norm": 0.05753512680530548, "global_step": 31574, "epoch": 877, "lr": 0.001} +{"train_loss": 0.031088637188076973, "train_loss_bc": 0.03107631579041481, "train_loss_llm": 0.0012321174144744873, "grad_norm": 0.10652051120996475, "global_step": 31575, "epoch": 877, "lr": 0.001} +{"train_loss": 0.028521185740828514, "train_loss_bc": 0.028521185740828514, "train_loss_llm": 0.0, "grad_norm": 0.13856227695941925, "global_step": 31576, "epoch": 877, "lr": 0.001} +{"train_loss": 0.023159323260188103, "train_loss_bc": 0.023159323260188103, "train_loss_llm": 0.0, "grad_norm": 0.03543506935238838, "global_step": 31577, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03038609027862549, "train_loss_bc": 0.03038609027862549, "train_loss_llm": 0.0, "grad_norm": 0.06974566727876663, "global_step": 31578, "epoch": 877, "lr": 0.001} +{"train_loss": 0.030762208625674248, "train_loss_bc": 0.030762208625674248, "train_loss_llm": 0.0, "grad_norm": 0.07906981557607651, "global_step": 31579, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03440682217478752, "train_loss_bc": 0.03440595418214798, "train_loss_llm": 8.675456047058105e-05, "grad_norm": 0.10026232153177261, "global_step": 31580, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03063073195517063, "train_loss_bc": 0.03063073195517063, "train_loss_llm": 0.0, "grad_norm": 0.04193190485239029, "global_step": 31581, "epoch": 877, "lr": 0.001} +{"train_loss": 0.031071435660123825, "train_loss_bc": 0.031071435660123825, "train_loss_llm": 0.0, "grad_norm": 0.06505191326141357, "global_step": 31582, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03253267705440521, "train_loss_bc": 0.03253267705440521, "train_loss_llm": 0.0, "grad_norm": 0.0979396402835846, "global_step": 31583, "epoch": 877, "lr": 0.001} +{"train_loss": 0.025275442749261856, "train_loss_bc": 0.025275442749261856, "train_loss_llm": 0.0, "grad_norm": 0.11770831048488617, "global_step": 31584, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03542076796293259, "train_loss_bc": 0.03542076796293259, "train_loss_llm": 0.0, "grad_norm": 0.038551148027181625, "global_step": 31585, "epoch": 877, "lr": 0.001} +{"train_loss": 0.027671128511428833, "train_loss_bc": 0.027671128511428833, "train_loss_llm": 0.0, "grad_norm": 0.0647980272769928, "global_step": 31586, "epoch": 877, "lr": 0.001} +{"train_loss": 0.027333304286003113, "train_loss_bc": 0.027333304286003113, "train_loss_llm": 0.0, "grad_norm": 0.08733497560024261, "global_step": 31587, "epoch": 877, "lr": 0.001} +{"train_loss": 0.021991917863488197, "train_loss_bc": 0.021991917863488197, "train_loss_llm": 0.0, "grad_norm": 0.10464154928922653, "global_step": 31588, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03675170987844467, "train_loss_bc": 0.03675170987844467, "train_loss_llm": 0.0, "grad_norm": 0.037155911326408386, "global_step": 31589, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03048085793852806, "train_loss_bc": 0.03048085793852806, "train_loss_llm": 0.0, "grad_norm": 0.05653495714068413, "global_step": 31590, "epoch": 877, "lr": 0.001} +{"train_loss": 0.026787206530570984, "train_loss_bc": 0.026787206530570984, "train_loss_llm": 0.0, "grad_norm": 0.09094437956809998, "global_step": 31591, "epoch": 877, "lr": 0.001} +{"train_loss": 0.023648468777537346, "train_loss_bc": 0.023638948798179626, "train_loss_llm": 0.0009520649909973145, "grad_norm": 0.11324681341648102, "global_step": 31592, "epoch": 877, "lr": 0.001} +{"train_loss": 0.02880432829260826, "train_loss_bc": 0.02880432829260826, "train_loss_llm": 0.0, "grad_norm": 0.03746885433793068, "global_step": 31593, "epoch": 877, "lr": 0.001} +{"train_loss": 0.022048261016607285, "train_loss_bc": 0.022048261016607285, "train_loss_llm": 0.0, "grad_norm": 0.05762467533349991, "global_step": 31594, "epoch": 877, "lr": 0.001} +{"train_loss": 0.0303427055478096, "train_loss_bc": 0.0303427055478096, "train_loss_llm": 0.0, "grad_norm": 0.08802396804094315, "global_step": 31595, "epoch": 877, "lr": 0.001} +{"train_loss": 0.025441093370318413, "train_loss_bc": 0.025441093370318413, "train_loss_llm": 0.0, "grad_norm": 0.10887768119573593, "global_step": 31596, "epoch": 877, "lr": 0.001} +{"train_loss": 0.029489556327462196, "train_loss_bc": 0.029489556327462196, "train_loss_llm": 0.0, "grad_norm": 0.043961212038993835, "global_step": 31597, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03182421624660492, "train_loss_bc": 0.03182421624660492, "train_loss_llm": 0.0, "grad_norm": 0.0573742613196373, "global_step": 31598, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03534475341439247, "train_loss_bc": 0.03534475341439247, "train_loss_llm": 0.0, "grad_norm": 0.09896176308393478, "global_step": 31599, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03199497237801552, "train_loss_bc": 0.03199497237801552, "train_loss_llm": 0.0, "grad_norm": 0.120457723736763, "global_step": 31600, "epoch": 877, "lr": 0.001} +{"train_loss": 0.027825724333524704, "train_loss_bc": 0.02781064808368683, "train_loss_llm": 0.0015075504779815674, "grad_norm": 0.037029776722192764, "global_step": 31601, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03239869698882103, "train_loss_bc": 0.03239869698882103, "train_loss_llm": 0.0, "grad_norm": 0.06447771191596985, "global_step": 31602, "epoch": 877, "lr": 0.001} +{"train_loss": 0.03152283653616905, "train_loss_bc": 0.03152283653616905, "train_loss_llm": 0.0, "grad_norm": 0.09739291667938232, "global_step": 31603, "epoch": 877, "lr": 0.001} +{"train_loss": 0.030101975426077843, "train_loss_bc": 0.030088093131780624, "train_loss_llm": 0.0013882815837860107, "grad_norm": 0.11879022419452667, "global_step": 31604, "epoch": 877, "lr": 0.001} +{"train_loss": 0.030322546139359474, "train_loss_bc": 0.030322546139359474, "train_loss_llm": 0.0, "grad_norm": 0.03823722153902054, "global_step": 31605, "epoch": 877, "lr": 0.001} +{"train_loss": 0.02983837202191353, "train_loss_bc": 0.02983837202191353, "train_loss_llm": 0.0, "grad_norm": 0.06925985962152481, "global_step": 31606, "epoch": 877, "lr": 0.001} +{"train_loss": 0.029749704628354974, "train_loss_bc": 0.03250519558787346, "train_loss_llm": 0.0, "grad_norm": 0.10279478877782822, "global_step": 31607, "epoch": 877, "lr": 0.001, "val_loss": 0.10144928842782974} +{"train_loss": 0.032110851258039474, "train_loss_bc": 0.0320851095020771, "train_loss_llm": 0.0025741159915924072, "grad_norm": 0.12546893954277039, "global_step": 31608, "epoch": 878, "lr": 0.001} +{"train_loss": 0.028899844735860825, "train_loss_bc": 0.028892936185002327, "train_loss_llm": 0.000690847635269165, "grad_norm": 0.03080529160797596, "global_step": 31609, "epoch": 878, "lr": 0.001} +{"train_loss": 0.02463698759675026, "train_loss_bc": 0.02462686039507389, "train_loss_llm": 0.0010127723217010498, "grad_norm": 0.05664198473095894, "global_step": 31610, "epoch": 878, "lr": 0.001} +{"train_loss": 0.032138075679540634, "train_loss_bc": 0.032138075679540634, "train_loss_llm": 0.0, "grad_norm": 0.06998874992132187, "global_step": 31611, "epoch": 878, "lr": 0.001} +{"train_loss": 0.023619694635272026, "train_loss_bc": 0.023619694635272026, "train_loss_llm": 0.0, "grad_norm": 0.10963527858257294, "global_step": 31612, "epoch": 878, "lr": 0.001} +{"train_loss": 0.03331826627254486, "train_loss_bc": 0.03331826627254486, "train_loss_llm": 0.0, "grad_norm": 0.04233371466398239, "global_step": 31613, "epoch": 878, "lr": 0.001} +{"train_loss": 0.02577492967247963, "train_loss_bc": 0.02577492967247963, "train_loss_llm": 0.0, "grad_norm": 0.06904783844947815, "global_step": 31614, "epoch": 878, "lr": 0.001} +{"train_loss": 0.03074684366583824, "train_loss_bc": 0.03074684366583824, "train_loss_llm": 0.0, "grad_norm": 0.08704409003257751, "global_step": 31615, "epoch": 878, "lr": 0.001} +{"train_loss": 0.03248785436153412, "train_loss_bc": 0.03247585892677307, "train_loss_llm": 0.0011993646621704102, "grad_norm": 0.10259845107793808, "global_step": 31616, "epoch": 878, "lr": 0.001} +{"train_loss": 0.03182180970907211, "train_loss_bc": 0.03182180970907211, "train_loss_llm": 0.0, "grad_norm": 0.03909606114029884, "global_step": 31617, "epoch": 878, "lr": 0.001} +{"train_loss": 0.028911421075463295, "train_loss_bc": 0.028911421075463295, "train_loss_llm": 0.0, "grad_norm": 0.06617865711450577, "global_step": 31618, "epoch": 878, "lr": 0.001} +{"train_loss": 0.027971329167485237, "train_loss_bc": 0.027971329167485237, "train_loss_llm": 0.0, "grad_norm": 0.0854138731956482, "global_step": 31619, "epoch": 878, "lr": 0.001} +{"train_loss": 0.03317082300782204, "train_loss_bc": 0.033156268298625946, "train_loss_llm": 0.0014556050300598145, "grad_norm": 0.11940017342567444, "global_step": 31620, "epoch": 878, "lr": 0.001} +{"train_loss": 0.03298788145184517, "train_loss_bc": 0.03298788145184517, "train_loss_llm": 0.0, "grad_norm": 0.0575796402990818, "global_step": 31621, "epoch": 878, "lr": 0.001} +{"train_loss": 0.02583502233028412, "train_loss_bc": 0.02583502233028412, "train_loss_llm": 0.0, "grad_norm": 0.08194003254175186, "global_step": 31622, "epoch": 878, "lr": 0.001} +{"train_loss": 0.030358847230672836, "train_loss_bc": 0.030355341732501984, "train_loss_llm": 0.00035053491592407227, "grad_norm": 0.10419680178165436, "global_step": 31623, "epoch": 878, "lr": 0.001} +{"train_loss": 0.031040702015161514, "train_loss_bc": 0.031040702015161514, "train_loss_llm": 0.0, "grad_norm": 0.12627099454402924, "global_step": 31624, "epoch": 878, "lr": 0.001} +{"train_loss": 0.027664801105856895, "train_loss_bc": 0.027664801105856895, "train_loss_llm": 0.0, "grad_norm": 0.03993577882647514, "global_step": 31625, "epoch": 878, "lr": 0.001} +{"train_loss": 0.031893208622932434, "train_loss_bc": 0.031893208622932434, "train_loss_llm": 0.0, "grad_norm": 0.06898169219493866, "global_step": 31626, "epoch": 878, "lr": 0.001} +{"train_loss": 0.033466048538684845, "train_loss_bc": 0.033466048538684845, "train_loss_llm": 0.0, "grad_norm": 0.100724957883358, "global_step": 31627, "epoch": 878, "lr": 0.001} +{"train_loss": 0.02780204825103283, "train_loss_bc": 0.02780204825103283, "train_loss_llm": 0.0, "grad_norm": 0.1380639672279358, "global_step": 31628, "epoch": 878, "lr": 0.001} +{"train_loss": 0.03407497704029083, "train_loss_bc": 0.03407497704029083, "train_loss_llm": 0.0, "grad_norm": 0.0501570962369442, "global_step": 31629, "epoch": 878, "lr": 0.001} +{"train_loss": 0.03549621254205704, "train_loss_bc": 0.03549621254205704, "train_loss_llm": 0.0, "grad_norm": 0.06960227340459824, "global_step": 31630, "epoch": 878, "lr": 0.001} +{"train_loss": 0.03225504234433174, "train_loss_bc": 0.0322442501783371, "train_loss_llm": 0.0010793805122375488, "grad_norm": 0.10390148311853409, "global_step": 31631, "epoch": 878, "lr": 0.001} +{"train_loss": 0.0317680761218071, "train_loss_bc": 0.03176186978816986, "train_loss_llm": 0.0006204545497894287, "grad_norm": 0.1217210441827774, "global_step": 31632, "epoch": 878, "lr": 0.001} +{"train_loss": 0.026140527799725533, "train_loss_bc": 0.026127975434064865, "train_loss_llm": 0.0012551546096801758, "grad_norm": 0.0379459522664547, "global_step": 31633, "epoch": 878, "lr": 0.001} +{"train_loss": 0.02571389265358448, "train_loss_bc": 0.02571389265358448, "train_loss_llm": 0.0, "grad_norm": 0.05292520299553871, "global_step": 31634, "epoch": 878, "lr": 0.001} +{"train_loss": 0.02306031435728073, "train_loss_bc": 0.02306031435728073, "train_loss_llm": 0.0, "grad_norm": 0.07663673907518387, "global_step": 31635, "epoch": 878, "lr": 0.001} +{"train_loss": 0.02796206995844841, "train_loss_bc": 0.02796206995844841, "train_loss_llm": 0.0, "grad_norm": 0.0944838598370552, "global_step": 31636, "epoch": 878, "lr": 0.001} +{"train_loss": 0.02783951535820961, "train_loss_bc": 0.027796249836683273, "train_loss_llm": 0.004326492547988892, "grad_norm": 0.03570105880498886, "global_step": 31637, "epoch": 878, "lr": 0.001} +{"train_loss": 0.034836895763874054, "train_loss_bc": 0.034836895763874054, "train_loss_llm": 0.0, "grad_norm": 0.05752312391996384, "global_step": 31638, "epoch": 878, "lr": 0.001} +{"train_loss": 0.02997368387877941, "train_loss_bc": 0.02995980903506279, "train_loss_llm": 0.001387566328048706, "grad_norm": 0.09780668467283249, "global_step": 31639, "epoch": 878, "lr": 0.001} +{"train_loss": 0.028799768537282944, "train_loss_bc": 0.028799768537282944, "train_loss_llm": 0.0, "grad_norm": 0.12459293007850647, "global_step": 31640, "epoch": 878, "lr": 0.001} +{"train_loss": 0.03135306388139725, "train_loss_bc": 0.03135306388139725, "train_loss_llm": 0.0, "grad_norm": 0.04197979345917702, "global_step": 31641, "epoch": 878, "lr": 0.001} +{"train_loss": 0.028665976598858833, "train_loss_bc": 0.028665976598858833, "train_loss_llm": 0.0, "grad_norm": 0.0640832856297493, "global_step": 31642, "epoch": 878, "lr": 0.001} +{"train_loss": 0.029782921282781497, "train_loss_bc": 0.027587858960032463, "train_loss_llm": 0.0, "grad_norm": 0.0976906269788742, "global_step": 31643, "epoch": 878, "lr": 0.001, "val_loss": 0.10386791080236435} +{"train_loss": 0.02619345113635063, "train_loss_bc": 0.02619345113635063, "train_loss_llm": 0.0, "grad_norm": 0.11801186203956604, "global_step": 31644, "epoch": 879, "lr": 0.001} +{"train_loss": 0.029104657471179962, "train_loss_bc": 0.029104657471179962, "train_loss_llm": 0.0, "grad_norm": 0.036935482174158096, "global_step": 31645, "epoch": 879, "lr": 0.001} +{"train_loss": 0.03041055053472519, "train_loss_bc": 0.03041055053472519, "train_loss_llm": 0.0, "grad_norm": 0.054936036467552185, "global_step": 31646, "epoch": 879, "lr": 0.001} +{"train_loss": 0.03302013501524925, "train_loss_bc": 0.03302013501524925, "train_loss_llm": 0.0, "grad_norm": 0.07829054445028305, "global_step": 31647, "epoch": 879, "lr": 0.001} +{"train_loss": 0.02725890278816223, "train_loss_bc": 0.02725890278816223, "train_loss_llm": 0.0, "grad_norm": 0.09766136854887009, "global_step": 31648, "epoch": 879, "lr": 0.001} +{"train_loss": 0.026889802888035774, "train_loss_bc": 0.026889802888035774, "train_loss_llm": 0.0, "grad_norm": 0.032743651419878006, "global_step": 31649, "epoch": 879, "lr": 0.001} +{"train_loss": 0.029506366699934006, "train_loss_bc": 0.029506366699934006, "train_loss_llm": 0.0, "grad_norm": 0.05951711907982826, "global_step": 31650, "epoch": 879, "lr": 0.001} +{"train_loss": 0.03324968367815018, "train_loss_bc": 0.03324968367815018, "train_loss_llm": 0.0, "grad_norm": 0.08747566491365433, "global_step": 31651, "epoch": 879, "lr": 0.001} +{"train_loss": 0.03173735737800598, "train_loss_bc": 0.03173735737800598, "train_loss_llm": 0.0, "grad_norm": 0.12007463723421097, "global_step": 31652, "epoch": 879, "lr": 0.001} +{"train_loss": 0.03349306434392929, "train_loss_bc": 0.03349306434392929, "train_loss_llm": 0.0, "grad_norm": 0.03885054215788841, "global_step": 31653, "epoch": 879, "lr": 0.001} +{"train_loss": 0.028609834611415863, "train_loss_bc": 0.028581932187080383, "train_loss_llm": 0.002790212631225586, "grad_norm": 0.051286064088344574, "global_step": 31654, "epoch": 879, "lr": 0.001} +{"train_loss": 0.028716571629047394, "train_loss_bc": 0.028716571629047394, "train_loss_llm": 0.0, "grad_norm": 0.055175069719552994, "global_step": 31655, "epoch": 879, "lr": 0.001} +{"train_loss": 0.02521967515349388, "train_loss_bc": 0.025199173018336296, "train_loss_llm": 0.0020501315593719482, "grad_norm": 0.06142478808760643, "global_step": 31656, "epoch": 879, "lr": 0.001} +{"train_loss": 0.03192707151174545, "train_loss_bc": 0.0319044254720211, "train_loss_llm": 0.0022645890712738037, "grad_norm": 0.0486731119453907, "global_step": 31657, "epoch": 879, "lr": 0.001} +{"train_loss": 0.03638399764895439, "train_loss_bc": 0.03638399764895439, "train_loss_llm": 0.0, "grad_norm": 0.09024946391582489, "global_step": 31658, "epoch": 879, "lr": 0.001} +{"train_loss": 0.02442251518368721, "train_loss_bc": 0.02442251518368721, "train_loss_llm": 0.0, "grad_norm": 0.11133625358343124, "global_step": 31659, "epoch": 879, "lr": 0.001} +{"train_loss": 0.031193915754556656, "train_loss_bc": 0.031193915754556656, "train_loss_llm": 0.0, "grad_norm": 0.13110674917697906, "global_step": 31660, "epoch": 879, "lr": 0.001} +{"train_loss": 0.03552699461579323, "train_loss_bc": 0.03552117198705673, "train_loss_llm": 0.0005823373794555664, "grad_norm": 0.05603458359837532, "global_step": 31661, "epoch": 879, "lr": 0.001} +{"train_loss": 0.030264852568507195, "train_loss_bc": 0.030264852568507195, "train_loss_llm": 0.0, "grad_norm": 0.10770566761493683, "global_step": 31662, "epoch": 879, "lr": 0.001} +{"train_loss": 0.028447099030017853, "train_loss_bc": 0.028441794216632843, "train_loss_llm": 0.0005304217338562012, "grad_norm": 0.14373779296875, "global_step": 31663, "epoch": 879, "lr": 0.001} +{"train_loss": 0.030443686991930008, "train_loss_bc": 0.030443686991930008, "train_loss_llm": 0.0, "grad_norm": 0.1624610871076584, "global_step": 31664, "epoch": 879, "lr": 0.001} +{"train_loss": 0.02336811274290085, "train_loss_bc": 0.02336811274290085, "train_loss_llm": 0.0, "grad_norm": 0.03329205513000488, "global_step": 31665, "epoch": 879, "lr": 0.001} +{"train_loss": 0.02692600153386593, "train_loss_bc": 0.02692600153386593, "train_loss_llm": 0.0, "grad_norm": 0.06263165175914764, "global_step": 31666, "epoch": 879, "lr": 0.001} +{"train_loss": 0.03362287953495979, "train_loss_bc": 0.03362287953495979, "train_loss_llm": 0.0, "grad_norm": 0.09186410158872604, "global_step": 31667, "epoch": 879, "lr": 0.001} +{"train_loss": 0.02985752560198307, "train_loss_bc": 0.02985752560198307, "train_loss_llm": 0.0, "grad_norm": 0.12028952687978745, "global_step": 31668, "epoch": 879, "lr": 0.001} +{"train_loss": 0.031638678163290024, "train_loss_bc": 0.031638678163290024, "train_loss_llm": 0.0, "grad_norm": 0.04455290362238884, "global_step": 31669, "epoch": 879, "lr": 0.001} +{"train_loss": 0.028675122186541557, "train_loss_bc": 0.028675122186541557, "train_loss_llm": 0.0, "grad_norm": 0.07047615200281143, "global_step": 31670, "epoch": 879, "lr": 0.001} +{"train_loss": 0.028265642002224922, "train_loss_bc": 0.02823968417942524, "train_loss_llm": 0.0025958120822906494, "grad_norm": 0.0844322144985199, "global_step": 31671, "epoch": 879, "lr": 0.001} +{"train_loss": 0.030523959547281265, "train_loss_bc": 0.030523959547281265, "train_loss_llm": 0.0, "grad_norm": 0.1054564043879509, "global_step": 31672, "epoch": 879, "lr": 0.001} +{"train_loss": 0.03018639236688614, "train_loss_bc": 0.03018639236688614, "train_loss_llm": 0.0, "grad_norm": 0.056821178644895554, "global_step": 31673, "epoch": 879, "lr": 0.001} +{"train_loss": 0.028773443773388863, "train_loss_bc": 0.028773443773388863, "train_loss_llm": 0.0, "grad_norm": 0.08352405577898026, "global_step": 31674, "epoch": 879, "lr": 0.001} +{"train_loss": 0.030880361795425415, "train_loss_bc": 0.0308692567050457, "train_loss_llm": 0.0011105537414550781, "grad_norm": 0.10501517355442047, "global_step": 31675, "epoch": 879, "lr": 0.001} +{"train_loss": 0.030407683923840523, "train_loss_bc": 0.030407683923840523, "train_loss_llm": 0.0, "grad_norm": 0.1289668083190918, "global_step": 31676, "epoch": 879, "lr": 0.001} +{"train_loss": 0.029398920014500618, "train_loss_bc": 0.029398920014500618, "train_loss_llm": 0.0, "grad_norm": 0.04253244400024414, "global_step": 31677, "epoch": 879, "lr": 0.001} +{"train_loss": 0.025821030139923096, "train_loss_bc": 0.025821030139923096, "train_loss_llm": 0.0, "grad_norm": 0.06440573930740356, "global_step": 31678, "epoch": 879, "lr": 0.001} +{"train_loss": 0.029717472640590534, "train_loss_bc": 0.02946307510137558, "train_loss_llm": 0.0, "grad_norm": 0.09287567436695099, "global_step": 31679, "epoch": 879, "lr": 0.001, "val_loss": 0.10095257312059402} +{"train_loss": 0.03077119216322899, "train_loss_bc": 0.03077119216322899, "train_loss_llm": 0.0, "grad_norm": 0.11142826825380325, "global_step": 31680, "epoch": 880, "lr": 0.001} +{"train_loss": 0.027909575030207634, "train_loss_bc": 0.027887921780347824, "train_loss_llm": 0.0021653473377227783, "grad_norm": 0.03532037138938904, "global_step": 31681, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03010462038218975, "train_loss_bc": 0.03010462038218975, "train_loss_llm": 0.0, "grad_norm": 0.07386372238397598, "global_step": 31682, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03137154504656792, "train_loss_bc": 0.03137154504656792, "train_loss_llm": 0.0, "grad_norm": 0.10325752943754196, "global_step": 31683, "epoch": 880, "lr": 0.001} +{"train_loss": 0.02782468870282173, "train_loss_bc": 0.02782468870282173, "train_loss_llm": 0.0, "grad_norm": 0.13691741228103638, "global_step": 31684, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03222598508000374, "train_loss_bc": 0.03222598508000374, "train_loss_llm": 0.0, "grad_norm": 0.04443638026714325, "global_step": 31685, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03124460019171238, "train_loss_bc": 0.03124460019171238, "train_loss_llm": 0.0, "grad_norm": 0.07188738882541656, "global_step": 31686, "epoch": 880, "lr": 0.001} +{"train_loss": 0.029381202533841133, "train_loss_bc": 0.029381202533841133, "train_loss_llm": 0.0, "grad_norm": 0.09769539535045624, "global_step": 31687, "epoch": 880, "lr": 0.001} +{"train_loss": 0.025618676096200943, "train_loss_bc": 0.025618676096200943, "train_loss_llm": 0.0, "grad_norm": 0.11562135070562363, "global_step": 31688, "epoch": 880, "lr": 0.001} +{"train_loss": 0.026562295854091644, "train_loss_bc": 0.026562295854091644, "train_loss_llm": 0.0, "grad_norm": 0.03551453724503517, "global_step": 31689, "epoch": 880, "lr": 0.001} +{"train_loss": 0.02625216916203499, "train_loss_bc": 0.02625216916203499, "train_loss_llm": 0.0, "grad_norm": 0.06528705358505249, "global_step": 31690, "epoch": 880, "lr": 0.001} +{"train_loss": 0.028400419279932976, "train_loss_bc": 0.028400419279932976, "train_loss_llm": 0.0, "grad_norm": 0.09780848771333694, "global_step": 31691, "epoch": 880, "lr": 0.001} +{"train_loss": 0.02999940887093544, "train_loss_bc": 0.02999940887093544, "train_loss_llm": 0.0, "grad_norm": 0.11781364679336548, "global_step": 31692, "epoch": 880, "lr": 0.001} +{"train_loss": 0.033740706741809845, "train_loss_bc": 0.033740706741809845, "train_loss_llm": 0.0, "grad_norm": 0.04964301735162735, "global_step": 31693, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03400294855237007, "train_loss_bc": 0.03397746384143829, "train_loss_llm": 0.0025486350059509277, "grad_norm": 0.0768287256360054, "global_step": 31694, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03000209853053093, "train_loss_bc": 0.03000209853053093, "train_loss_llm": 0.0, "grad_norm": 0.11389318108558655, "global_step": 31695, "epoch": 880, "lr": 0.001} +{"train_loss": 0.025770217180252075, "train_loss_bc": 0.025770217180252075, "train_loss_llm": 0.0, "grad_norm": 0.1293565183877945, "global_step": 31696, "epoch": 880, "lr": 0.001} +{"train_loss": 0.026931943371891975, "train_loss_bc": 0.026924310252070427, "train_loss_llm": 0.0007632672786712646, "grad_norm": 0.04603245481848717, "global_step": 31697, "epoch": 880, "lr": 0.001} +{"train_loss": 0.026570754125714302, "train_loss_bc": 0.026570754125714302, "train_loss_llm": 0.0, "grad_norm": 0.06938732415437698, "global_step": 31698, "epoch": 880, "lr": 0.001} +{"train_loss": 0.027637075632810593, "train_loss_bc": 0.027637075632810593, "train_loss_llm": 0.0, "grad_norm": 0.08539637178182602, "global_step": 31699, "epoch": 880, "lr": 0.001} +{"train_loss": 0.028354478999972343, "train_loss_bc": 0.028354478999972343, "train_loss_llm": 0.0, "grad_norm": 0.10874328762292862, "global_step": 31700, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03074008785188198, "train_loss_bc": 0.0307319238781929, "train_loss_llm": 0.00081634521484375, "grad_norm": 0.04683453217148781, "global_step": 31701, "epoch": 880, "lr": 0.001} +{"train_loss": 0.026549870148301125, "train_loss_bc": 0.02654254622757435, "train_loss_llm": 0.0007323920726776123, "grad_norm": 0.07143338769674301, "global_step": 31702, "epoch": 880, "lr": 0.001} +{"train_loss": 0.030288729816675186, "train_loss_bc": 0.030288729816675186, "train_loss_llm": 0.0, "grad_norm": 0.0957317054271698, "global_step": 31703, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03372086584568024, "train_loss_bc": 0.03371596336364746, "train_loss_llm": 0.0004903674125671387, "grad_norm": 0.12064219266176224, "global_step": 31704, "epoch": 880, "lr": 0.001} +{"train_loss": 0.02732393890619278, "train_loss_bc": 0.02732393890619278, "train_loss_llm": 0.0, "grad_norm": 0.031780097633600235, "global_step": 31705, "epoch": 880, "lr": 0.001} +{"train_loss": 0.030456166714429855, "train_loss_bc": 0.030448123812675476, "train_loss_llm": 0.0008043348789215088, "grad_norm": 0.0506637841463089, "global_step": 31706, "epoch": 880, "lr": 0.001} +{"train_loss": 0.02766783908009529, "train_loss_bc": 0.02766783908009529, "train_loss_llm": 0.0, "grad_norm": 0.06875927746295929, "global_step": 31707, "epoch": 880, "lr": 0.001} +{"train_loss": 0.026848547160625458, "train_loss_bc": 0.026848547160625458, "train_loss_llm": 0.0, "grad_norm": 0.08829424530267715, "global_step": 31708, "epoch": 880, "lr": 0.001} +{"train_loss": 0.032577402889728546, "train_loss_bc": 0.03256279602646828, "train_loss_llm": 0.0014606118202209473, "grad_norm": 0.03998522460460663, "global_step": 31709, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03004823997616768, "train_loss_bc": 0.03004823997616768, "train_loss_llm": 0.0, "grad_norm": 0.05964995175600052, "global_step": 31710, "epoch": 880, "lr": 0.001} +{"train_loss": 0.034133121371269226, "train_loss_bc": 0.03411067649722099, "train_loss_llm": 0.002244502305984497, "grad_norm": 0.0865263119339943, "global_step": 31711, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03159468248486519, "train_loss_bc": 0.03159468248486519, "train_loss_llm": 0.0, "grad_norm": 0.11540985852479935, "global_step": 31712, "epoch": 880, "lr": 0.001} +{"train_loss": 0.033189255744218826, "train_loss_bc": 0.033189255744218826, "train_loss_llm": 0.0, "grad_norm": 0.04164186865091324, "global_step": 31713, "epoch": 880, "lr": 0.001} +{"train_loss": 0.03497586026787758, "train_loss_bc": 0.03497586026787758, "train_loss_llm": 0.0, "grad_norm": 0.0675801932811737, "global_step": 31714, "epoch": 880, "lr": 0.001} +{"train_loss": 0.029754720214340422, "train_loss_bc": 0.030378717929124832, "train_loss_llm": 0.0, "grad_norm": 0.09797873347997665, "global_step": 31715, "epoch": 880, "lr": 0.001, "train/cumulative_reward": 59.52090519945457, "train/mean_score": 14.80687510612006, "train/success_rate": 0.4, "test/cumulative_reward": 25.73052861138607, "test/mean_score": 8.642034855291893, "test/success_rate": 0.24, "val_loss": 0.10333504527807236, "train_action_mse_error": 0.03269732743501663} +{"train_loss": 0.028410160914063454, "train_loss_bc": 0.028410160914063454, "train_loss_llm": 0.0, "grad_norm": 0.13124988973140717, "global_step": 31716, "epoch": 881, "lr": 0.001} +{"train_loss": 0.03204668313264847, "train_loss_bc": 0.03204668313264847, "train_loss_llm": 0.0, "grad_norm": 0.03778735548257828, "global_step": 31717, "epoch": 881, "lr": 0.001} +{"train_loss": 0.031080419197678566, "train_loss_bc": 0.031080419197678566, "train_loss_llm": 0.0, "grad_norm": 0.061734721064567566, "global_step": 31718, "epoch": 881, "lr": 0.001} +{"train_loss": 0.024444952607154846, "train_loss_bc": 0.024435900151729584, "train_loss_llm": 0.0009052157402038574, "grad_norm": 0.07039856165647507, "global_step": 31719, "epoch": 881, "lr": 0.001} +{"train_loss": 0.028379347175359726, "train_loss_bc": 0.028379347175359726, "train_loss_llm": 0.0, "grad_norm": 0.08871939778327942, "global_step": 31720, "epoch": 881, "lr": 0.001} +{"train_loss": 0.03387248516082764, "train_loss_bc": 0.03387248516082764, "train_loss_llm": 0.0, "grad_norm": 0.030955608934164047, "global_step": 31721, "epoch": 881, "lr": 0.001} +{"train_loss": 0.03066468983888626, "train_loss_bc": 0.0306551493704319, "train_loss_llm": 0.0009540915489196777, "grad_norm": 0.05693972483277321, "global_step": 31722, "epoch": 881, "lr": 0.001} +{"train_loss": 0.02405443601310253, "train_loss_bc": 0.02404976449906826, "train_loss_llm": 0.0004671812057495117, "grad_norm": 0.07813979685306549, "global_step": 31723, "epoch": 881, "lr": 0.001} +{"train_loss": 0.03329809382557869, "train_loss_bc": 0.03329510614275932, "train_loss_llm": 0.0002988278865814209, "grad_norm": 0.11521139740943909, "global_step": 31724, "epoch": 881, "lr": 0.001} +{"train_loss": 0.022412218153476715, "train_loss_bc": 0.022412218153476715, "train_loss_llm": 0.0, "grad_norm": 0.03212285786867142, "global_step": 31725, "epoch": 881, "lr": 0.001} +{"train_loss": 0.03163912147283554, "train_loss_bc": 0.03163912147283554, "train_loss_llm": 0.0, "grad_norm": 0.06076480820775032, "global_step": 31726, "epoch": 881, "lr": 0.001} +{"train_loss": 0.025900108739733696, "train_loss_bc": 0.025900108739733696, "train_loss_llm": 0.0, "grad_norm": 0.08242138475179672, "global_step": 31727, "epoch": 881, "lr": 0.001} +{"train_loss": 0.022687247022986412, "train_loss_bc": 0.022684186697006226, "train_loss_llm": 0.00030601024627685547, "grad_norm": 0.1023545190691948, "global_step": 31728, "epoch": 881, "lr": 0.001} +{"train_loss": 0.02661457657814026, "train_loss_bc": 0.02661457657814026, "train_loss_llm": 0.0, "grad_norm": 0.04103623330593109, "global_step": 31729, "epoch": 881, "lr": 0.001} +{"train_loss": 0.02993674762547016, "train_loss_bc": 0.02993674762547016, "train_loss_llm": 0.0, "grad_norm": 0.07404810190200806, "global_step": 31730, "epoch": 881, "lr": 0.001} +{"train_loss": 0.03609253838658333, "train_loss_bc": 0.03609253838658333, "train_loss_llm": 0.0, "grad_norm": 0.09244488924741745, "global_step": 31731, "epoch": 881, "lr": 0.001} +{"train_loss": 0.03586893156170845, "train_loss_bc": 0.03586893156170845, "train_loss_llm": 0.0, "grad_norm": 0.12019971758127213, "global_step": 31732, "epoch": 881, "lr": 0.001} +{"train_loss": 0.0353408120572567, "train_loss_bc": 0.0353408120572567, "train_loss_llm": 0.0, "grad_norm": 0.052813030779361725, "global_step": 31733, "epoch": 881, "lr": 0.001} +{"train_loss": 0.036058373749256134, "train_loss_bc": 0.036058373749256134, "train_loss_llm": 0.0, "grad_norm": 0.0839555412530899, "global_step": 31734, "epoch": 881, "lr": 0.001} +{"train_loss": 0.029321564361453056, "train_loss_bc": 0.029321564361453056, "train_loss_llm": 0.0, "grad_norm": 0.10601204633712769, "global_step": 31735, "epoch": 881, "lr": 0.001} +{"train_loss": 0.024609819054603577, "train_loss_bc": 0.024596363306045532, "train_loss_llm": 0.0013456344604492188, "grad_norm": 0.12478205561637878, "global_step": 31736, "epoch": 881, "lr": 0.001} +{"train_loss": 0.025556132197380066, "train_loss_bc": 0.025556132197380066, "train_loss_llm": 0.0, "grad_norm": 0.03826100006699562, "global_step": 31737, "epoch": 881, "lr": 0.001} +{"train_loss": 0.035592928528785706, "train_loss_bc": 0.03557657450437546, "train_loss_llm": 0.0016352832317352295, "grad_norm": 0.05556504428386688, "global_step": 31738, "epoch": 881, "lr": 0.001} +{"train_loss": 0.028403902426362038, "train_loss_bc": 0.02840213105082512, "train_loss_llm": 0.00017705559730529785, "grad_norm": 0.07851707935333252, "global_step": 31739, "epoch": 881, "lr": 0.001} +{"train_loss": 0.029582828283309937, "train_loss_bc": 0.029582828283309937, "train_loss_llm": 0.0, "grad_norm": 0.09722555428743362, "global_step": 31740, "epoch": 881, "lr": 0.001} +{"train_loss": 0.02758842147886753, "train_loss_bc": 0.02758190408349037, "train_loss_llm": 0.0006517171859741211, "grad_norm": 0.029270362108945847, "global_step": 31741, "epoch": 881, "lr": 0.001} +{"train_loss": 0.027600768953561783, "train_loss_bc": 0.027600573375821114, "train_loss_llm": 1.963973045349121e-05, "grad_norm": 0.05395926535129547, "global_step": 31742, "epoch": 881, "lr": 0.001} +{"train_loss": 0.02780289202928543, "train_loss_bc": 0.027794331312179565, "train_loss_llm": 0.0008560121059417725, "grad_norm": 0.06419669836759567, "global_step": 31743, "epoch": 881, "lr": 0.001} +{"train_loss": 0.03168129920959473, "train_loss_bc": 0.03168129920959473, "train_loss_llm": 0.0, "grad_norm": 0.08929294347763062, "global_step": 31744, "epoch": 881, "lr": 0.001} +{"train_loss": 0.027193108573555946, "train_loss_bc": 0.027186619117856026, "train_loss_llm": 0.0006489455699920654, "grad_norm": 0.03341911733150482, "global_step": 31745, "epoch": 881, "lr": 0.001} +{"train_loss": 0.02774658612906933, "train_loss_bc": 0.0277264304459095, "train_loss_llm": 0.0020155012607574463, "grad_norm": 0.04865404963493347, "global_step": 31746, "epoch": 881, "lr": 0.001} +{"train_loss": 0.028886107727885246, "train_loss_bc": 0.028880828991532326, "train_loss_llm": 0.0005279481410980225, "grad_norm": 0.06485354155302048, "global_step": 31747, "epoch": 881, "lr": 0.001} +{"train_loss": 0.03517491742968559, "train_loss_bc": 0.03517491742968559, "train_loss_llm": 0.0, "grad_norm": 0.08468703925609589, "global_step": 31748, "epoch": 881, "lr": 0.001} +{"train_loss": 0.029285183176398277, "train_loss_bc": 0.02925274334847927, "train_loss_llm": 0.0032440125942230225, "grad_norm": 0.04870513081550598, "global_step": 31749, "epoch": 881, "lr": 0.001} +{"train_loss": 0.02940310724079609, "train_loss_bc": 0.02940310724079609, "train_loss_llm": 0.0, "grad_norm": 0.07415270805358887, "global_step": 31750, "epoch": 881, "lr": 0.001} +{"train_loss": 0.02946354831672377, "train_loss_bc": 0.026456229388713837, "train_loss_llm": 0.0, "grad_norm": 0.09420399367809296, "global_step": 31751, "epoch": 881, "lr": 0.001, "val_loss": 0.10195263475179672} +{"train_loss": 0.030457431450486183, "train_loss_bc": 0.030457431450486183, "train_loss_llm": 0.0, "grad_norm": 0.11979015171527863, "global_step": 31752, "epoch": 882, "lr": 0.001} +{"train_loss": 0.03068551793694496, "train_loss_bc": 0.03068551793694496, "train_loss_llm": 0.0, "grad_norm": 0.04032960906624794, "global_step": 31753, "epoch": 882, "lr": 0.001} +{"train_loss": 0.028819909319281578, "train_loss_bc": 0.028819909319281578, "train_loss_llm": 0.0, "grad_norm": 0.06735754013061523, "global_step": 31754, "epoch": 882, "lr": 0.001} +{"train_loss": 0.03177459165453911, "train_loss_bc": 0.03177459165453911, "train_loss_llm": 0.0, "grad_norm": 0.0914238691329956, "global_step": 31755, "epoch": 882, "lr": 0.001} +{"train_loss": 0.030217843130230904, "train_loss_bc": 0.030217843130230904, "train_loss_llm": 0.0, "grad_norm": 0.10617885738611221, "global_step": 31756, "epoch": 882, "lr": 0.001} +{"train_loss": 0.025032471865415573, "train_loss_bc": 0.025032471865415573, "train_loss_llm": 0.0, "grad_norm": 0.04515102878212929, "global_step": 31757, "epoch": 882, "lr": 0.001} +{"train_loss": 0.027448397129774094, "train_loss_bc": 0.02743990160524845, "train_loss_llm": 0.0008495151996612549, "grad_norm": 0.08117791265249252, "global_step": 31758, "epoch": 882, "lr": 0.001} +{"train_loss": 0.02651147171854973, "train_loss_bc": 0.02651147171854973, "train_loss_llm": 0.0, "grad_norm": 0.11779191344976425, "global_step": 31759, "epoch": 882, "lr": 0.001} +{"train_loss": 0.029470128938555717, "train_loss_bc": 0.029470128938555717, "train_loss_llm": 0.0, "grad_norm": 0.13862919807434082, "global_step": 31760, "epoch": 882, "lr": 0.001} +{"train_loss": 0.04085586220026016, "train_loss_bc": 0.04085586220026016, "train_loss_llm": 0.0, "grad_norm": 0.0428120382130146, "global_step": 31761, "epoch": 882, "lr": 0.001} +{"train_loss": 0.031062662601470947, "train_loss_bc": 0.031059768050909042, "train_loss_llm": 0.00028952956199645996, "grad_norm": 0.06825344264507294, "global_step": 31762, "epoch": 882, "lr": 0.001} +{"train_loss": 0.026746777817606926, "train_loss_bc": 0.026746777817606926, "train_loss_llm": 0.0, "grad_norm": 0.0902896448969841, "global_step": 31763, "epoch": 882, "lr": 0.001} +{"train_loss": 0.029900889843702316, "train_loss_bc": 0.029900889843702316, "train_loss_llm": 0.0, "grad_norm": 0.12070563435554504, "global_step": 31764, "epoch": 882, "lr": 0.001} +{"train_loss": 0.028518101200461388, "train_loss_bc": 0.028518101200461388, "train_loss_llm": 0.0, "grad_norm": 0.03759108856320381, "global_step": 31765, "epoch": 882, "lr": 0.001} +{"train_loss": 0.02992808260023594, "train_loss_bc": 0.02992808260023594, "train_loss_llm": 0.0, "grad_norm": 0.06003837287425995, "global_step": 31766, "epoch": 882, "lr": 0.001} +{"train_loss": 0.027059607207775116, "train_loss_bc": 0.027059607207775116, "train_loss_llm": 0.0, "grad_norm": 0.07981493324041367, "global_step": 31767, "epoch": 882, "lr": 0.001} +{"train_loss": 0.033828526735305786, "train_loss_bc": 0.03381257876753807, "train_loss_llm": 0.0015948712825775146, "grad_norm": 0.1025335043668747, "global_step": 31768, "epoch": 882, "lr": 0.001} +{"train_loss": 0.0297582745552063, "train_loss_bc": 0.0297582745552063, "train_loss_llm": 0.0, "grad_norm": 0.033930882811546326, "global_step": 31769, "epoch": 882, "lr": 0.001} +{"train_loss": 0.034414708614349365, "train_loss_bc": 0.034414708614349365, "train_loss_llm": 0.0, "grad_norm": 0.05692819505929947, "global_step": 31770, "epoch": 882, "lr": 0.001} +{"train_loss": 0.027667135000228882, "train_loss_bc": 0.027667135000228882, "train_loss_llm": 0.0, "grad_norm": 0.0787787064909935, "global_step": 31771, "epoch": 882, "lr": 0.001} +{"train_loss": 0.02590351551771164, "train_loss_bc": 0.02590351551771164, "train_loss_llm": 0.0, "grad_norm": 0.08768503367900848, "global_step": 31772, "epoch": 882, "lr": 0.001} +{"train_loss": 0.03189246356487274, "train_loss_bc": 0.031891223043203354, "train_loss_llm": 0.0001240372657775879, "grad_norm": 0.04965969920158386, "global_step": 31773, "epoch": 882, "lr": 0.001} +{"train_loss": 0.029180215671658516, "train_loss_bc": 0.029175177216529846, "train_loss_llm": 0.0005038678646087646, "grad_norm": 0.07728078216314316, "global_step": 31774, "epoch": 882, "lr": 0.001} +{"train_loss": 0.026798903942108154, "train_loss_bc": 0.02679237723350525, "train_loss_llm": 0.0006527304649353027, "grad_norm": 0.09980380535125732, "global_step": 31775, "epoch": 882, "lr": 0.001} +{"train_loss": 0.027999265119433403, "train_loss_bc": 0.027988271787762642, "train_loss_llm": 0.0010992884635925293, "grad_norm": 0.13524936139583588, "global_step": 31776, "epoch": 882, "lr": 0.001} +{"train_loss": 0.024599218741059303, "train_loss_bc": 0.024599218741059303, "train_loss_llm": 0.0, "grad_norm": 0.030615849420428276, "global_step": 31777, "epoch": 882, "lr": 0.001} +{"train_loss": 0.030421830713748932, "train_loss_bc": 0.030416756868362427, "train_loss_llm": 0.0005072951316833496, "grad_norm": 0.04765687882900238, "global_step": 31778, "epoch": 882, "lr": 0.001} +{"train_loss": 0.026654280722141266, "train_loss_bc": 0.026654280722141266, "train_loss_llm": 0.0, "grad_norm": 0.06613335758447647, "global_step": 31779, "epoch": 882, "lr": 0.001} +{"train_loss": 0.029475126415491104, "train_loss_bc": 0.029475126415491104, "train_loss_llm": 0.0, "grad_norm": 0.07707757502794266, "global_step": 31780, "epoch": 882, "lr": 0.001} +{"train_loss": 0.02624867856502533, "train_loss_bc": 0.026244375854730606, "train_loss_llm": 0.00043019652366638184, "grad_norm": 0.031466417014598846, "global_step": 31781, "epoch": 882, "lr": 0.001} +{"train_loss": 0.03268935903906822, "train_loss_bc": 0.03268418833613396, "train_loss_llm": 0.000517040491104126, "grad_norm": 0.05299123376607895, "global_step": 31782, "epoch": 882, "lr": 0.001} +{"train_loss": 0.04050394147634506, "train_loss_bc": 0.040493737906217575, "train_loss_llm": 0.001020282506942749, "grad_norm": 0.07130397856235504, "global_step": 31783, "epoch": 882, "lr": 0.001} +{"train_loss": 0.027882885187864304, "train_loss_bc": 0.027882885187864304, "train_loss_llm": 0.0, "grad_norm": 0.07690742611885071, "global_step": 31784, "epoch": 882, "lr": 0.001} +{"train_loss": 0.026766113936901093, "train_loss_bc": 0.026766113936901093, "train_loss_llm": 0.0, "grad_norm": 0.03994961827993393, "global_step": 31785, "epoch": 882, "lr": 0.001} +{"train_loss": 0.02222658134996891, "train_loss_bc": 0.02222658134996891, "train_loss_llm": 0.0, "grad_norm": 0.04627355933189392, "global_step": 31786, "epoch": 882, "lr": 0.001} +{"train_loss": 0.0295835939856867, "train_loss_bc": 0.03560861200094223, "train_loss_llm": 0.0, "grad_norm": 0.0800282433629036, "global_step": 31787, "epoch": 882, "lr": 0.001, "val_loss": 0.1010824516415596} +{"train_loss": 0.03006651997566223, "train_loss_bc": 0.03006651997566223, "train_loss_llm": 0.0, "grad_norm": 0.09986092150211334, "global_step": 31788, "epoch": 883, "lr": 0.001} +{"train_loss": 0.030886070802807808, "train_loss_bc": 0.03087809309363365, "train_loss_llm": 0.0007976889610290527, "grad_norm": 0.04490736126899719, "global_step": 31789, "epoch": 883, "lr": 0.001} +{"train_loss": 0.029967881739139557, "train_loss_bc": 0.029967881739139557, "train_loss_llm": 0.0, "grad_norm": 0.05827116221189499, "global_step": 31790, "epoch": 883, "lr": 0.001} +{"train_loss": 0.029808176681399345, "train_loss_bc": 0.029803799465298653, "train_loss_llm": 0.00043779611587524414, "grad_norm": 0.0751967802643776, "global_step": 31791, "epoch": 883, "lr": 0.001} +{"train_loss": 0.029702078551054, "train_loss_bc": 0.029702078551054, "train_loss_llm": 0.0, "grad_norm": 0.09487113356590271, "global_step": 31792, "epoch": 883, "lr": 0.001} +{"train_loss": 0.02832397073507309, "train_loss_bc": 0.02831641584634781, "train_loss_llm": 0.0007555484771728516, "grad_norm": 0.03347315639257431, "global_step": 31793, "epoch": 883, "lr": 0.001} +{"train_loss": 0.02560480497777462, "train_loss_bc": 0.025583328679203987, "train_loss_llm": 0.0021477043628692627, "grad_norm": 0.05335669219493866, "global_step": 31794, "epoch": 883, "lr": 0.001} +{"train_loss": 0.029995210468769073, "train_loss_bc": 0.02996901050209999, "train_loss_llm": 0.0026199817657470703, "grad_norm": 0.07478691637516022, "global_step": 31795, "epoch": 883, "lr": 0.001} +{"train_loss": 0.029810110107064247, "train_loss_bc": 0.029810110107064247, "train_loss_llm": 0.0, "grad_norm": 0.09724991023540497, "global_step": 31796, "epoch": 883, "lr": 0.001} +{"train_loss": 0.025068236514925957, "train_loss_bc": 0.025068236514925957, "train_loss_llm": 0.0, "grad_norm": 0.031202049925923347, "global_step": 31797, "epoch": 883, "lr": 0.001} +{"train_loss": 0.03030882030725479, "train_loss_bc": 0.03030882030725479, "train_loss_llm": 0.0, "grad_norm": 0.05708213150501251, "global_step": 31798, "epoch": 883, "lr": 0.001} +{"train_loss": 0.024562306702136993, "train_loss_bc": 0.024562306702136993, "train_loss_llm": 0.0, "grad_norm": 0.08691444993019104, "global_step": 31799, "epoch": 883, "lr": 0.001} +{"train_loss": 0.028303807601332664, "train_loss_bc": 0.028303807601332664, "train_loss_llm": 0.0, "grad_norm": 0.09698662161827087, "global_step": 31800, "epoch": 883, "lr": 0.001} +{"train_loss": 0.0287258792668581, "train_loss_bc": 0.0287258792668581, "train_loss_llm": 0.0, "grad_norm": 0.03719274327158928, "global_step": 31801, "epoch": 883, "lr": 0.001} +{"train_loss": 0.02325008250772953, "train_loss_bc": 0.023234983906149864, "train_loss_llm": 0.0015098154544830322, "grad_norm": 0.05634910613298416, "global_step": 31802, "epoch": 883, "lr": 0.001} +{"train_loss": 0.028612881898880005, "train_loss_bc": 0.028612881898880005, "train_loss_llm": 0.0, "grad_norm": 0.07205566018819809, "global_step": 31803, "epoch": 883, "lr": 0.001} +{"train_loss": 0.03223447501659393, "train_loss_bc": 0.032208118587732315, "train_loss_llm": 0.002635776996612549, "grad_norm": 0.08668604493141174, "global_step": 31804, "epoch": 883, "lr": 0.001} +{"train_loss": 0.03708953410387039, "train_loss_bc": 0.03708953410387039, "train_loss_llm": 0.0, "grad_norm": 0.043819885700941086, "global_step": 31805, "epoch": 883, "lr": 0.001} +{"train_loss": 0.02633069083094597, "train_loss_bc": 0.02633069083094597, "train_loss_llm": 0.0, "grad_norm": 0.05656535550951958, "global_step": 31806, "epoch": 883, "lr": 0.001} +{"train_loss": 0.02727586217224598, "train_loss_bc": 0.02727586217224598, "train_loss_llm": 0.0, "grad_norm": 0.07647614181041718, "global_step": 31807, "epoch": 883, "lr": 0.001} +{"train_loss": 0.03197425976395607, "train_loss_bc": 0.03197425976395607, "train_loss_llm": 0.0, "grad_norm": 0.10247177630662918, "global_step": 31808, "epoch": 883, "lr": 0.001} +{"train_loss": 0.026163896545767784, "train_loss_bc": 0.026163896545767784, "train_loss_llm": 0.0, "grad_norm": 0.037602394819259644, "global_step": 31809, "epoch": 883, "lr": 0.001} +{"train_loss": 0.02611910179257393, "train_loss_bc": 0.02611910179257393, "train_loss_llm": 0.0, "grad_norm": 0.06604573130607605, "global_step": 31810, "epoch": 883, "lr": 0.001} +{"train_loss": 0.03468389809131622, "train_loss_bc": 0.03468389809131622, "train_loss_llm": 0.0, "grad_norm": 0.09173072874546051, "global_step": 31811, "epoch": 883, "lr": 0.001} +{"train_loss": 0.033267926424741745, "train_loss_bc": 0.033267926424741745, "train_loss_llm": 0.0, "grad_norm": 0.1172165498137474, "global_step": 31812, "epoch": 883, "lr": 0.001} +{"train_loss": 0.027905425056815147, "train_loss_bc": 0.027905425056815147, "train_loss_llm": 0.0, "grad_norm": 0.0432453453540802, "global_step": 31813, "epoch": 883, "lr": 0.001} +{"train_loss": 0.028431668877601624, "train_loss_bc": 0.028420735150575638, "train_loss_llm": 0.0010934174060821533, "grad_norm": 0.06363785266876221, "global_step": 31814, "epoch": 883, "lr": 0.001} +{"train_loss": 0.02957034297287464, "train_loss_bc": 0.02957034297287464, "train_loss_llm": 0.0, "grad_norm": 0.08562393486499786, "global_step": 31815, "epoch": 883, "lr": 0.001} +{"train_loss": 0.02718598209321499, "train_loss_bc": 0.02718598209321499, "train_loss_llm": 0.0, "grad_norm": 0.09921301901340485, "global_step": 31816, "epoch": 883, "lr": 0.001} +{"train_loss": 0.033999063074588776, "train_loss_bc": 0.033999063074588776, "train_loss_llm": 0.0, "grad_norm": 0.0426749512553215, "global_step": 31817, "epoch": 883, "lr": 0.001} +{"train_loss": 0.035787295550107956, "train_loss_bc": 0.035787295550107956, "train_loss_llm": 0.0, "grad_norm": 0.07258389890193939, "global_step": 31818, "epoch": 883, "lr": 0.001} +{"train_loss": 0.03227777034044266, "train_loss_bc": 0.03227777034044266, "train_loss_llm": 0.0, "grad_norm": 0.0967257022857666, "global_step": 31819, "epoch": 883, "lr": 0.001} +{"train_loss": 0.03239699453115463, "train_loss_bc": 0.03238646313548088, "train_loss_llm": 0.0010530352592468262, "grad_norm": 0.11274342238903046, "global_step": 31820, "epoch": 883, "lr": 0.001} +{"train_loss": 0.02450580522418022, "train_loss_bc": 0.02450580522418022, "train_loss_llm": 0.0, "grad_norm": 0.02828339859843254, "global_step": 31821, "epoch": 883, "lr": 0.001} +{"train_loss": 0.02858296036720276, "train_loss_bc": 0.02858296036720276, "train_loss_llm": 0.0, "grad_norm": 0.05451292172074318, "global_step": 31822, "epoch": 883, "lr": 0.001} +{"train_loss": 0.029424735862347815, "train_loss_bc": 0.03051069937646389, "train_loss_llm": 0.0, "grad_norm": 0.08355388045310974, "global_step": 31823, "epoch": 883, "lr": 0.001, "val_loss": 0.10426761209964752} +{"train_loss": 0.025514358654618263, "train_loss_bc": 0.02549467608332634, "train_loss_llm": 0.001968204975128174, "grad_norm": 0.09858399629592896, "global_step": 31824, "epoch": 884, "lr": 0.001} +{"train_loss": 0.022542279213666916, "train_loss_bc": 0.022542279213666916, "train_loss_llm": 0.0, "grad_norm": 0.03971967101097107, "global_step": 31825, "epoch": 884, "lr": 0.001} +{"train_loss": 0.027009297162294388, "train_loss_bc": 0.0270076971501112, "train_loss_llm": 0.00015997886657714844, "grad_norm": 0.07758847624063492, "global_step": 31826, "epoch": 884, "lr": 0.001} +{"train_loss": 0.03383859992027283, "train_loss_bc": 0.03383859992027283, "train_loss_llm": 0.0, "grad_norm": 0.10619084537029266, "global_step": 31827, "epoch": 884, "lr": 0.001} +{"train_loss": 0.03211120888590813, "train_loss_bc": 0.03210403025150299, "train_loss_llm": 0.0007177889347076416, "grad_norm": 0.12821383774280548, "global_step": 31828, "epoch": 884, "lr": 0.001} +{"train_loss": 0.03386843577027321, "train_loss_bc": 0.03385646641254425, "train_loss_llm": 0.001196831464767456, "grad_norm": 0.0376163087785244, "global_step": 31829, "epoch": 884, "lr": 0.001} +{"train_loss": 0.023660777136683464, "train_loss_bc": 0.023660777136683464, "train_loss_llm": 0.0, "grad_norm": 0.05593954399228096, "global_step": 31830, "epoch": 884, "lr": 0.001} +{"train_loss": 0.0319385752081871, "train_loss_bc": 0.031934890896081924, "train_loss_llm": 0.0003685951232910156, "grad_norm": 0.07438451051712036, "global_step": 31831, "epoch": 884, "lr": 0.001} +{"train_loss": 0.02280031330883503, "train_loss_bc": 0.02280031330883503, "train_loss_llm": 0.0, "grad_norm": 0.09180013835430145, "global_step": 31832, "epoch": 884, "lr": 0.001} +{"train_loss": 0.02775154635310173, "train_loss_bc": 0.02775154635310173, "train_loss_llm": 0.0, "grad_norm": 0.04201313853263855, "global_step": 31833, "epoch": 884, "lr": 0.001} +{"train_loss": 0.03558819741010666, "train_loss_bc": 0.03558819741010666, "train_loss_llm": 0.0, "grad_norm": 0.07394877821207047, "global_step": 31834, "epoch": 884, "lr": 0.001} +{"train_loss": 0.029612908139824867, "train_loss_bc": 0.029612908139824867, "train_loss_llm": 0.0, "grad_norm": 0.08094091713428497, "global_step": 31835, "epoch": 884, "lr": 0.001} +{"train_loss": 0.027869651094079018, "train_loss_bc": 0.027869651094079018, "train_loss_llm": 0.0, "grad_norm": 0.10981341451406479, "global_step": 31836, "epoch": 884, "lr": 0.001} +{"train_loss": 0.02762746438384056, "train_loss_bc": 0.02762746438384056, "train_loss_llm": 0.0, "grad_norm": 0.041428279131650925, "global_step": 31837, "epoch": 884, "lr": 0.001} +{"train_loss": 0.03164517134428024, "train_loss_bc": 0.03164517134428024, "train_loss_llm": 0.0, "grad_norm": 0.06297150999307632, "global_step": 31838, "epoch": 884, "lr": 0.001} +{"train_loss": 0.039682500064373016, "train_loss_bc": 0.039682500064373016, "train_loss_llm": 0.0, "grad_norm": 0.09706045687198639, "global_step": 31839, "epoch": 884, "lr": 0.001} +{"train_loss": 0.02496320940554142, "train_loss_bc": 0.02496320940554142, "train_loss_llm": 0.0, "grad_norm": 0.12194058299064636, "global_step": 31840, "epoch": 884, "lr": 0.001} +{"train_loss": 0.026602985337376595, "train_loss_bc": 0.026573775336146355, "train_loss_llm": 0.0029209554195404053, "grad_norm": 0.037937749177217484, "global_step": 31841, "epoch": 884, "lr": 0.001} +{"train_loss": 0.03466547653079033, "train_loss_bc": 0.03465895727276802, "train_loss_llm": 0.0006518363952636719, "grad_norm": 0.06826668232679367, "global_step": 31842, "epoch": 884, "lr": 0.001} +{"train_loss": 0.03199394419789314, "train_loss_bc": 0.03199394419789314, "train_loss_llm": 0.0, "grad_norm": 0.09609030187129974, "global_step": 31843, "epoch": 884, "lr": 0.001} +{"train_loss": 0.031684380024671555, "train_loss_bc": 0.031684380024671555, "train_loss_llm": 0.0, "grad_norm": 0.11369063705205917, "global_step": 31844, "epoch": 884, "lr": 0.001} +{"train_loss": 0.02832856774330139, "train_loss_bc": 0.02832856774330139, "train_loss_llm": 0.0, "grad_norm": 0.027899041771888733, "global_step": 31845, "epoch": 884, "lr": 0.001} +{"train_loss": 0.030999820679426193, "train_loss_bc": 0.030999820679426193, "train_loss_llm": 0.0, "grad_norm": 0.05159302055835724, "global_step": 31846, "epoch": 884, "lr": 0.001} +{"train_loss": 0.026125621050596237, "train_loss_bc": 0.02612142264842987, "train_loss_llm": 0.00041988492012023926, "grad_norm": 0.0652933195233345, "global_step": 31847, "epoch": 884, "lr": 0.001} +{"train_loss": 0.02799820713698864, "train_loss_bc": 0.027984123677015305, "train_loss_llm": 0.001408308744430542, "grad_norm": 0.07628270983695984, "global_step": 31848, "epoch": 884, "lr": 0.001} +{"train_loss": 0.026135597378015518, "train_loss_bc": 0.026135597378015518, "train_loss_llm": 0.0, "grad_norm": 0.03649405017495155, "global_step": 31849, "epoch": 884, "lr": 0.001} +{"train_loss": 0.030711382627487183, "train_loss_bc": 0.030711382627487183, "train_loss_llm": 0.0, "grad_norm": 0.06029804050922394, "global_step": 31850, "epoch": 884, "lr": 0.001} +{"train_loss": 0.03222909942269325, "train_loss_bc": 0.03222909942269325, "train_loss_llm": 0.0, "grad_norm": 0.07431907206773758, "global_step": 31851, "epoch": 884, "lr": 0.001} +{"train_loss": 0.029043493792414665, "train_loss_bc": 0.029043493792414665, "train_loss_llm": 0.0, "grad_norm": 0.09771925956010818, "global_step": 31852, "epoch": 884, "lr": 0.001} +{"train_loss": 0.0222672987729311, "train_loss_bc": 0.022259097546339035, "train_loss_llm": 0.0008200407028198242, "grad_norm": 0.03934525325894356, "global_step": 31853, "epoch": 884, "lr": 0.001} +{"train_loss": 0.02772684581577778, "train_loss_bc": 0.02772684581577778, "train_loss_llm": 0.0, "grad_norm": 0.0748596116900444, "global_step": 31854, "epoch": 884, "lr": 0.001} +{"train_loss": 0.03448759391903877, "train_loss_bc": 0.03448759391903877, "train_loss_llm": 0.0, "grad_norm": 0.11540046334266663, "global_step": 31855, "epoch": 884, "lr": 0.001} +{"train_loss": 0.03120245225727558, "train_loss_bc": 0.03120245225727558, "train_loss_llm": 0.0, "grad_norm": 0.14551056921482086, "global_step": 31856, "epoch": 884, "lr": 0.001} +{"train_loss": 0.030116600915789604, "train_loss_bc": 0.030116600915789604, "train_loss_llm": 0.0, "grad_norm": 0.05269339308142662, "global_step": 31857, "epoch": 884, "lr": 0.001} +{"train_loss": 0.029345715418457985, "train_loss_bc": 0.029345715418457985, "train_loss_llm": 0.0, "grad_norm": 0.08560685068368912, "global_step": 31858, "epoch": 884, "lr": 0.001} +{"train_loss": 0.029542055187953845, "train_loss_bc": 0.03382416442036629, "train_loss_llm": 2.4437904357910156e-05, "grad_norm": 0.11346296966075897, "global_step": 31859, "epoch": 884, "lr": 0.001, "val_loss": 0.10481927543878555} +{"train_loss": 0.031051816418766975, "train_loss_bc": 0.031051816418766975, "train_loss_llm": 0.0, "grad_norm": 0.14759379625320435, "global_step": 31860, "epoch": 885, "lr": 0.001} +{"train_loss": 0.024617670103907585, "train_loss_bc": 0.024617670103907585, "train_loss_llm": 0.0, "grad_norm": 0.03332962095737457, "global_step": 31861, "epoch": 885, "lr": 0.001} +{"train_loss": 0.02435564249753952, "train_loss_bc": 0.02435564249753952, "train_loss_llm": 0.0, "grad_norm": 0.05905317887663841, "global_step": 31862, "epoch": 885, "lr": 0.001} +{"train_loss": 0.028072470799088478, "train_loss_bc": 0.028067436069250107, "train_loss_llm": 0.0005034804344177246, "grad_norm": 0.08128296583890915, "global_step": 31863, "epoch": 885, "lr": 0.001} +{"train_loss": 0.027994224801659584, "train_loss_bc": 0.027992302551865578, "train_loss_llm": 0.00019228458404541016, "grad_norm": 0.09847480803728104, "global_step": 31864, "epoch": 885, "lr": 0.001} +{"train_loss": 0.02267724648118019, "train_loss_bc": 0.02267724648118019, "train_loss_llm": 0.0, "grad_norm": 0.026505641639232635, "global_step": 31865, "epoch": 885, "lr": 0.001} +{"train_loss": 0.029100915417075157, "train_loss_bc": 0.029100915417075157, "train_loss_llm": 0.0, "grad_norm": 0.05083708465099335, "global_step": 31866, "epoch": 885, "lr": 0.001} +{"train_loss": 0.027518242597579956, "train_loss_bc": 0.027518242597579956, "train_loss_llm": 0.0, "grad_norm": 0.07031135261058807, "global_step": 31867, "epoch": 885, "lr": 0.001} +{"train_loss": 0.029549047350883484, "train_loss_bc": 0.029549047350883484, "train_loss_llm": 0.0, "grad_norm": 0.0858474150300026, "global_step": 31868, "epoch": 885, "lr": 0.001} +{"train_loss": 0.030725302174687386, "train_loss_bc": 0.030725302174687386, "train_loss_llm": 0.0, "grad_norm": 0.03602328151464462, "global_step": 31869, "epoch": 885, "lr": 0.001} +{"train_loss": 0.034276943653821945, "train_loss_bc": 0.034276943653821945, "train_loss_llm": 0.0, "grad_norm": 0.07270244508981705, "global_step": 31870, "epoch": 885, "lr": 0.001} +{"train_loss": 0.03163441643118858, "train_loss_bc": 0.03163441643118858, "train_loss_llm": 0.0, "grad_norm": 0.10051047056913376, "global_step": 31871, "epoch": 885, "lr": 0.001} +{"train_loss": 0.030586037784814835, "train_loss_bc": 0.030576452612876892, "train_loss_llm": 0.000958561897277832, "grad_norm": 0.1262516975402832, "global_step": 31872, "epoch": 885, "lr": 0.001} +{"train_loss": 0.033390503376722336, "train_loss_bc": 0.033375080674886703, "train_loss_llm": 0.0015422403812408447, "grad_norm": 0.05387508124113083, "global_step": 31873, "epoch": 885, "lr": 0.001} +{"train_loss": 0.029034744948148727, "train_loss_bc": 0.02901187166571617, "train_loss_llm": 0.00228726863861084, "grad_norm": 0.08670977503061295, "global_step": 31874, "epoch": 885, "lr": 0.001} +{"train_loss": 0.03585464879870415, "train_loss_bc": 0.03585464879870415, "train_loss_llm": 0.0, "grad_norm": 0.11698044836521149, "global_step": 31875, "epoch": 885, "lr": 0.001} +{"train_loss": 0.03256450593471527, "train_loss_bc": 0.03256450593471527, "train_loss_llm": 0.0, "grad_norm": 0.1386859118938446, "global_step": 31876, "epoch": 885, "lr": 0.001} +{"train_loss": 0.027133356779813766, "train_loss_bc": 0.027132397517561913, "train_loss_llm": 9.59932804107666e-05, "grad_norm": 0.03920605406165123, "global_step": 31877, "epoch": 885, "lr": 0.001} +{"train_loss": 0.030552485957741737, "train_loss_bc": 0.030541885644197464, "train_loss_llm": 0.0010599792003631592, "grad_norm": 0.07048661261796951, "global_step": 31878, "epoch": 885, "lr": 0.001} +{"train_loss": 0.03198431804776192, "train_loss_bc": 0.03198431804776192, "train_loss_llm": 0.0, "grad_norm": 0.09974294155836105, "global_step": 31879, "epoch": 885, "lr": 0.001} +{"train_loss": 0.02772577479481697, "train_loss_bc": 0.02771390788257122, "train_loss_llm": 0.0011866092681884766, "grad_norm": 0.14848996698856354, "global_step": 31880, "epoch": 885, "lr": 0.001} +{"train_loss": 0.03171369805932045, "train_loss_bc": 0.03171369805932045, "train_loss_llm": 0.0, "grad_norm": 0.03874875232577324, "global_step": 31881, "epoch": 885, "lr": 0.001} +{"train_loss": 0.029026072472333908, "train_loss_bc": 0.029026072472333908, "train_loss_llm": 0.0, "grad_norm": 0.06987600028514862, "global_step": 31882, "epoch": 885, "lr": 0.001} +{"train_loss": 0.029148997738957405, "train_loss_bc": 0.029148997738957405, "train_loss_llm": 0.0, "grad_norm": 0.09943358600139618, "global_step": 31883, "epoch": 885, "lr": 0.001} +{"train_loss": 0.0271702129393816, "train_loss_bc": 0.02716316096484661, "train_loss_llm": 0.0007051527500152588, "grad_norm": 0.11760671436786652, "global_step": 31884, "epoch": 885, "lr": 0.001} +{"train_loss": 0.03426988795399666, "train_loss_bc": 0.03426988795399666, "train_loss_llm": 0.0, "grad_norm": 0.03756464272737503, "global_step": 31885, "epoch": 885, "lr": 0.001} +{"train_loss": 0.03186710923910141, "train_loss_bc": 0.03186710923910141, "train_loss_llm": 0.0, "grad_norm": 0.06589726358652115, "global_step": 31886, "epoch": 885, "lr": 0.001} +{"train_loss": 0.021119752898812294, "train_loss_bc": 0.021119752898812294, "train_loss_llm": 0.0, "grad_norm": 0.07501529157161713, "global_step": 31887, "epoch": 885, "lr": 0.001} +{"train_loss": 0.029217014089226723, "train_loss_bc": 0.029217014089226723, "train_loss_llm": 0.0, "grad_norm": 0.08485817909240723, "global_step": 31888, "epoch": 885, "lr": 0.001} +{"train_loss": 0.031862638890743256, "train_loss_bc": 0.031862638890743256, "train_loss_llm": 0.0, "grad_norm": 0.03922129049897194, "global_step": 31889, "epoch": 885, "lr": 0.001} +{"train_loss": 0.024902448058128357, "train_loss_bc": 0.024902448058128357, "train_loss_llm": 0.0, "grad_norm": 0.05044887959957123, "global_step": 31890, "epoch": 885, "lr": 0.001} +{"train_loss": 0.034462351351976395, "train_loss_bc": 0.034447889775037766, "train_loss_llm": 0.0014461278915405273, "grad_norm": 0.07620672136545181, "global_step": 31891, "epoch": 885, "lr": 0.001} +{"train_loss": 0.032807186245918274, "train_loss_bc": 0.03278748691082001, "train_loss_llm": 0.0019698143005371094, "grad_norm": 0.10680625587701797, "global_step": 31892, "epoch": 885, "lr": 0.001} +{"train_loss": 0.03221319615840912, "train_loss_bc": 0.03221319615840912, "train_loss_llm": 0.0, "grad_norm": 0.04362873360514641, "global_step": 31893, "epoch": 885, "lr": 0.001} +{"train_loss": 0.025536416098475456, "train_loss_bc": 0.025536416098475456, "train_loss_llm": 0.0, "grad_norm": 0.06962890177965164, "global_step": 31894, "epoch": 885, "lr": 0.001} +{"train_loss": 0.02965448931273487, "train_loss_bc": 0.03182457387447357, "train_loss_llm": 0.0019742250442504883, "grad_norm": 0.097859688103199, "global_step": 31895, "epoch": 885, "lr": 0.001, "train/cumulative_reward": 76.4329254609402, "train/mean_score": 20.898355838208115, "train/success_rate": 0.6, "test/cumulative_reward": 37.63317358370602, "test/mean_score": 9.600618585785979, "test/success_rate": 0.28, "val_loss": 0.10344213992357254, "train_action_mse_error": 0.03260520100593567} +{"train_loss": 0.031987160444259644, "train_loss_bc": 0.03198675066232681, "train_loss_llm": 4.079937934875488e-05, "grad_norm": 0.1227250024676323, "global_step": 31896, "epoch": 886, "lr": 0.001} +{"train_loss": 0.02140798233449459, "train_loss_bc": 0.02140798233449459, "train_loss_llm": 0.0, "grad_norm": 0.031652603298425674, "global_step": 31897, "epoch": 886, "lr": 0.001} +{"train_loss": 0.033567581325769424, "train_loss_bc": 0.033567581325769424, "train_loss_llm": 0.0, "grad_norm": 0.06260070949792862, "global_step": 31898, "epoch": 886, "lr": 0.001} +{"train_loss": 0.026753036305308342, "train_loss_bc": 0.026753036305308342, "train_loss_llm": 0.0, "grad_norm": 0.0725020244717598, "global_step": 31899, "epoch": 886, "lr": 0.001} +{"train_loss": 0.0263485349714756, "train_loss_bc": 0.0263485349714756, "train_loss_llm": 0.0, "grad_norm": 0.08965574949979782, "global_step": 31900, "epoch": 886, "lr": 0.001} +{"train_loss": 0.028367647901177406, "train_loss_bc": 0.028367646038532257, "train_loss_llm": 1.7881393432617188e-07, "grad_norm": 0.036252591758966446, "global_step": 31901, "epoch": 886, "lr": 0.001} +{"train_loss": 0.03180881217122078, "train_loss_bc": 0.031806617975234985, "train_loss_llm": 0.0002193450927734375, "grad_norm": 0.05273079499602318, "global_step": 31902, "epoch": 886, "lr": 0.001} +{"train_loss": 0.029281290248036385, "train_loss_bc": 0.029281290248036385, "train_loss_llm": 0.0, "grad_norm": 0.07753517478704453, "global_step": 31903, "epoch": 886, "lr": 0.001} +{"train_loss": 0.030128829181194305, "train_loss_bc": 0.030128829181194305, "train_loss_llm": 0.0, "grad_norm": 0.10340988636016846, "global_step": 31904, "epoch": 886, "lr": 0.001} +{"train_loss": 0.025678696110844612, "train_loss_bc": 0.025678696110844612, "train_loss_llm": 0.0, "grad_norm": 0.028024138882756233, "global_step": 31905, "epoch": 886, "lr": 0.001} +{"train_loss": 0.027505183592438698, "train_loss_bc": 0.02749202959239483, "train_loss_llm": 0.001315474510192871, "grad_norm": 0.04837601259350777, "global_step": 31906, "epoch": 886, "lr": 0.001} +{"train_loss": 0.030118972063064575, "train_loss_bc": 0.030118972063064575, "train_loss_llm": 0.0, "grad_norm": 0.062318626791238785, "global_step": 31907, "epoch": 886, "lr": 0.001} +{"train_loss": 0.02559594064950943, "train_loss_bc": 0.025593360885977745, "train_loss_llm": 0.0002580583095550537, "grad_norm": 0.08622770011425018, "global_step": 31908, "epoch": 886, "lr": 0.001} +{"train_loss": 0.027715027332305908, "train_loss_bc": 0.027715027332305908, "train_loss_llm": 0.0, "grad_norm": 0.029835497960448265, "global_step": 31909, "epoch": 886, "lr": 0.001} +{"train_loss": 0.03068065270781517, "train_loss_bc": 0.03068065270781517, "train_loss_llm": 0.0, "grad_norm": 0.04130683094263077, "global_step": 31910, "epoch": 886, "lr": 0.001} +{"train_loss": 0.030361272394657135, "train_loss_bc": 0.030361272394657135, "train_loss_llm": 0.0, "grad_norm": 0.05585477501153946, "global_step": 31911, "epoch": 886, "lr": 0.001} +{"train_loss": 0.027935678139328957, "train_loss_bc": 0.027922211214900017, "train_loss_llm": 0.001346677541732788, "grad_norm": 0.07022247463464737, "global_step": 31912, "epoch": 886, "lr": 0.001} +{"train_loss": 0.031947147101163864, "train_loss_bc": 0.031947147101163864, "train_loss_llm": 0.0, "grad_norm": 0.03635484352707863, "global_step": 31913, "epoch": 886, "lr": 0.001} +{"train_loss": 0.031273096799850464, "train_loss_bc": 0.031273096799850464, "train_loss_llm": 0.0, "grad_norm": 0.05942029505968094, "global_step": 31914, "epoch": 886, "lr": 0.001} +{"train_loss": 0.023888632655143738, "train_loss_bc": 0.023888632655143738, "train_loss_llm": 0.0, "grad_norm": 0.08613777160644531, "global_step": 31915, "epoch": 886, "lr": 0.001} +{"train_loss": 0.03174393251538277, "train_loss_bc": 0.03174393251538277, "train_loss_llm": 0.0, "grad_norm": 0.09390374273061752, "global_step": 31916, "epoch": 886, "lr": 0.001} +{"train_loss": 0.03363139182329178, "train_loss_bc": 0.03363139182329178, "train_loss_llm": 0.0, "grad_norm": 0.03433318808674812, "global_step": 31917, "epoch": 886, "lr": 0.001} +{"train_loss": 0.03408065810799599, "train_loss_bc": 0.03408065810799599, "train_loss_llm": 0.0, "grad_norm": 0.06175202131271362, "global_step": 31918, "epoch": 886, "lr": 0.001} +{"train_loss": 0.027808362618088722, "train_loss_bc": 0.027808362618088722, "train_loss_llm": 0.0, "grad_norm": 0.07849093526601791, "global_step": 31919, "epoch": 886, "lr": 0.001} +{"train_loss": 0.03635242581367493, "train_loss_bc": 0.03634258359670639, "train_loss_llm": 0.0009844005107879639, "grad_norm": 0.0997234433889389, "global_step": 31920, "epoch": 886, "lr": 0.001} +{"train_loss": 0.024305658414959908, "train_loss_bc": 0.024305658414959908, "train_loss_llm": 0.0, "grad_norm": 0.03243781998753548, "global_step": 31921, "epoch": 886, "lr": 0.001} +{"train_loss": 0.028279133141040802, "train_loss_bc": 0.028279133141040802, "train_loss_llm": 0.0, "grad_norm": 0.04649365320801735, "global_step": 31922, "epoch": 886, "lr": 0.001} +{"train_loss": 0.031051943078637123, "train_loss_bc": 0.031048942357301712, "train_loss_llm": 0.0003001093864440918, "grad_norm": 0.06308232247829437, "global_step": 31923, "epoch": 886, "lr": 0.001} +{"train_loss": 0.027600767090916634, "train_loss_bc": 0.027600767090916634, "train_loss_llm": 0.0, "grad_norm": 0.08015922456979752, "global_step": 31924, "epoch": 886, "lr": 0.001} +{"train_loss": 0.031164441257715225, "train_loss_bc": 0.031164441257715225, "train_loss_llm": 0.0, "grad_norm": 0.03755447641015053, "global_step": 31925, "epoch": 886, "lr": 0.001} +{"train_loss": 0.02896147593855858, "train_loss_bc": 0.028960997238755226, "train_loss_llm": 4.789233207702637e-05, "grad_norm": 0.062195029109716415, "global_step": 31926, "epoch": 886, "lr": 0.001} +{"train_loss": 0.031111624091863632, "train_loss_bc": 0.03110559470951557, "train_loss_llm": 0.0006029605865478516, "grad_norm": 0.08682279288768768, "global_step": 31927, "epoch": 886, "lr": 0.001} +{"train_loss": 0.03129607066512108, "train_loss_bc": 0.03129607066512108, "train_loss_llm": 0.0, "grad_norm": 0.09412481635808945, "global_step": 31928, "epoch": 886, "lr": 0.001} +{"train_loss": 0.030816052109003067, "train_loss_bc": 0.030803445726633072, "train_loss_llm": 0.0012607276439666748, "grad_norm": 0.04391160607337952, "global_step": 31929, "epoch": 886, "lr": 0.001} +{"train_loss": 0.02185346558690071, "train_loss_bc": 0.021842708811163902, "train_loss_llm": 0.001075655221939087, "grad_norm": 0.06633572280406952, "global_step": 31930, "epoch": 886, "lr": 0.001} +{"train_loss": 0.029333563883685403, "train_loss_bc": 0.033599723130464554, "train_loss_llm": 0.0, "grad_norm": 0.07960649579763412, "global_step": 31931, "epoch": 886, "lr": 0.001, "val_loss": 0.1053939238190651} +{"train_loss": 0.02546517923474312, "train_loss_bc": 0.02546517923474312, "train_loss_llm": 0.0, "grad_norm": 0.09561334550380707, "global_step": 31932, "epoch": 887, "lr": 0.001} +{"train_loss": 0.0296406801789999, "train_loss_bc": 0.0296406801789999, "train_loss_llm": 0.0, "grad_norm": 0.03437516838312149, "global_step": 31933, "epoch": 887, "lr": 0.001} +{"train_loss": 0.030998943373560905, "train_loss_bc": 0.030998943373560905, "train_loss_llm": 0.0, "grad_norm": 0.06584972888231277, "global_step": 31934, "epoch": 887, "lr": 0.001} +{"train_loss": 0.02654399164021015, "train_loss_bc": 0.026541506871581078, "train_loss_llm": 0.0002484321594238281, "grad_norm": 0.07191099971532822, "global_step": 31935, "epoch": 887, "lr": 0.001} +{"train_loss": 0.03288564458489418, "train_loss_bc": 0.03288564458489418, "train_loss_llm": 0.0, "grad_norm": 0.1076442077755928, "global_step": 31936, "epoch": 887, "lr": 0.001} +{"train_loss": 0.027419138699769974, "train_loss_bc": 0.027419138699769974, "train_loss_llm": 0.0, "grad_norm": 0.03343048319220543, "global_step": 31937, "epoch": 887, "lr": 0.001} +{"train_loss": 0.03866032510995865, "train_loss_bc": 0.03864838555455208, "train_loss_llm": 0.0011939406394958496, "grad_norm": 0.0680290162563324, "global_step": 31938, "epoch": 887, "lr": 0.001} +{"train_loss": 0.023098625242710114, "train_loss_bc": 0.023098625242710114, "train_loss_llm": 0.0, "grad_norm": 0.07866062223911285, "global_step": 31939, "epoch": 887, "lr": 0.001} +{"train_loss": 0.02549806609749794, "train_loss_bc": 0.02549806609749794, "train_loss_llm": 0.0, "grad_norm": 0.09063123166561127, "global_step": 31940, "epoch": 887, "lr": 0.001} +{"train_loss": 0.029532242566347122, "train_loss_bc": 0.029532242566347122, "train_loss_llm": 0.0, "grad_norm": 0.025770045816898346, "global_step": 31941, "epoch": 887, "lr": 0.001} +{"train_loss": 0.03315594792366028, "train_loss_bc": 0.03314266726374626, "train_loss_llm": 0.001328200101852417, "grad_norm": 0.04317602515220642, "global_step": 31942, "epoch": 887, "lr": 0.001} +{"train_loss": 0.03032715804874897, "train_loss_bc": 0.03032715804874897, "train_loss_llm": 0.0, "grad_norm": 0.06172969937324524, "global_step": 31943, "epoch": 887, "lr": 0.001} +{"train_loss": 0.028449799865484238, "train_loss_bc": 0.028449799865484238, "train_loss_llm": 0.0, "grad_norm": 0.07153607159852982, "global_step": 31944, "epoch": 887, "lr": 0.001} +{"train_loss": 0.031512677669525146, "train_loss_bc": 0.031512677669525146, "train_loss_llm": 0.0, "grad_norm": 0.05248711258172989, "global_step": 31945, "epoch": 887, "lr": 0.001} +{"train_loss": 0.03502262383699417, "train_loss_bc": 0.03501548990607262, "train_loss_llm": 0.0007132589817047119, "grad_norm": 0.06442013382911682, "global_step": 31946, "epoch": 887, "lr": 0.001} +{"train_loss": 0.02555963769555092, "train_loss_bc": 0.02555963769555092, "train_loss_llm": 0.0, "grad_norm": 0.09003710001707077, "global_step": 31947, "epoch": 887, "lr": 0.001} +{"train_loss": 0.03249156475067139, "train_loss_bc": 0.03249156475067139, "train_loss_llm": 0.0, "grad_norm": 0.11336901783943176, "global_step": 31948, "epoch": 887, "lr": 0.001} +{"train_loss": 0.030047114938497543, "train_loss_bc": 0.030047114938497543, "train_loss_llm": 0.0, "grad_norm": 0.03770076110959053, "global_step": 31949, "epoch": 887, "lr": 0.001} +{"train_loss": 0.029785558581352234, "train_loss_bc": 0.029768118634819984, "train_loss_llm": 0.001744002103805542, "grad_norm": 0.06077664718031883, "global_step": 31950, "epoch": 887, "lr": 0.001} +{"train_loss": 0.028217773884534836, "train_loss_bc": 0.028217773884534836, "train_loss_llm": 0.0, "grad_norm": 0.08709697425365448, "global_step": 31951, "epoch": 887, "lr": 0.001} +{"train_loss": 0.03133455291390419, "train_loss_bc": 0.03133455291390419, "train_loss_llm": 0.0, "grad_norm": 0.11012130230665207, "global_step": 31952, "epoch": 887, "lr": 0.001} +{"train_loss": 0.03041333705186844, "train_loss_bc": 0.03041333705186844, "train_loss_llm": 0.0, "grad_norm": 0.04001510888338089, "global_step": 31953, "epoch": 887, "lr": 0.001} +{"train_loss": 0.031208157539367676, "train_loss_bc": 0.031208157539367676, "train_loss_llm": 0.0, "grad_norm": 0.060796067118644714, "global_step": 31954, "epoch": 887, "lr": 0.001} +{"train_loss": 0.0265299454331398, "train_loss_bc": 0.0265299454331398, "train_loss_llm": 0.0, "grad_norm": 0.08216153830289841, "global_step": 31955, "epoch": 887, "lr": 0.001} +{"train_loss": 0.03199099004268646, "train_loss_bc": 0.0319819301366806, "train_loss_llm": 0.000906139612197876, "grad_norm": 0.10483847558498383, "global_step": 31956, "epoch": 887, "lr": 0.001} +{"train_loss": 0.030165251344442368, "train_loss_bc": 0.030165251344442368, "train_loss_llm": 0.0, "grad_norm": 0.034521061927080154, "global_step": 31957, "epoch": 887, "lr": 0.001} +{"train_loss": 0.025469398126006126, "train_loss_bc": 0.025469398126006126, "train_loss_llm": 0.0, "grad_norm": 0.06412255764007568, "global_step": 31958, "epoch": 887, "lr": 0.001} +{"train_loss": 0.02913770079612732, "train_loss_bc": 0.02912839874625206, "train_loss_llm": 0.0009301304817199707, "grad_norm": 0.10111704468727112, "global_step": 31959, "epoch": 887, "lr": 0.001} +{"train_loss": 0.02739294245839119, "train_loss_bc": 0.027374455705285072, "train_loss_llm": 0.001848667860031128, "grad_norm": 0.11790977418422699, "global_step": 31960, "epoch": 887, "lr": 0.001} +{"train_loss": 0.026971403509378433, "train_loss_bc": 0.026969628408551216, "train_loss_llm": 0.00017753243446350098, "grad_norm": 0.03125365078449249, "global_step": 31961, "epoch": 887, "lr": 0.001} +{"train_loss": 0.030419282615184784, "train_loss_bc": 0.030407976359128952, "train_loss_llm": 0.0011305809020996094, "grad_norm": 0.05616718903183937, "global_step": 31962, "epoch": 887, "lr": 0.001} +{"train_loss": 0.031112918630242348, "train_loss_bc": 0.031112918630242348, "train_loss_llm": 0.0, "grad_norm": 0.08079972863197327, "global_step": 31963, "epoch": 887, "lr": 0.001} +{"train_loss": 0.026069579645991325, "train_loss_bc": 0.026055723428726196, "train_loss_llm": 0.0013856589794158936, "grad_norm": 0.09038547426462173, "global_step": 31964, "epoch": 887, "lr": 0.001} +{"train_loss": 0.025387316942214966, "train_loss_bc": 0.025387316942214966, "train_loss_llm": 0.0, "grad_norm": 0.034049972891807556, "global_step": 31965, "epoch": 887, "lr": 0.001} +{"train_loss": 0.027417626231908798, "train_loss_bc": 0.027417626231908798, "train_loss_llm": 0.0, "grad_norm": 0.057152289897203445, "global_step": 31966, "epoch": 887, "lr": 0.001} +{"train_loss": 0.02935283998441365, "train_loss_bc": 0.03136914223432541, "train_loss_llm": 0.0, "grad_norm": 0.095042884349823, "global_step": 31967, "epoch": 887, "lr": 0.001, "val_loss": 0.10287704318761826} +{"train_loss": 0.034582655876874924, "train_loss_bc": 0.034582655876874924, "train_loss_llm": 0.0, "grad_norm": 0.12601490318775177, "global_step": 31968, "epoch": 888, "lr": 0.001} +{"train_loss": 0.031378865242004395, "train_loss_bc": 0.03136321157217026, "train_loss_llm": 0.0015654563903808594, "grad_norm": 0.04696396365761757, "global_step": 31969, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03121199458837509, "train_loss_bc": 0.03121199458837509, "train_loss_llm": 0.0, "grad_norm": 0.0664622038602829, "global_step": 31970, "epoch": 888, "lr": 0.001} +{"train_loss": 0.027066806331276894, "train_loss_bc": 0.027066083624958992, "train_loss_llm": 7.224082946777344e-05, "grad_norm": 0.07846269756555557, "global_step": 31971, "epoch": 888, "lr": 0.001} +{"train_loss": 0.029811356216669083, "train_loss_bc": 0.029811356216669083, "train_loss_llm": 0.0, "grad_norm": 0.1003718301653862, "global_step": 31972, "epoch": 888, "lr": 0.001} +{"train_loss": 0.02718052640557289, "train_loss_bc": 0.027180174365639687, "train_loss_llm": 3.528594970703125e-05, "grad_norm": 0.03601547330617905, "global_step": 31973, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03022412769496441, "train_loss_bc": 0.03022412769496441, "train_loss_llm": 0.0, "grad_norm": 0.06312984228134155, "global_step": 31974, "epoch": 888, "lr": 0.001} +{"train_loss": 0.030126692727208138, "train_loss_bc": 0.030126692727208138, "train_loss_llm": 0.0, "grad_norm": 0.08456197381019592, "global_step": 31975, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03421524167060852, "train_loss_bc": 0.03421524167060852, "train_loss_llm": 0.0, "grad_norm": 0.10716129839420319, "global_step": 31976, "epoch": 888, "lr": 0.001} +{"train_loss": 0.027876652777194977, "train_loss_bc": 0.027876652777194977, "train_loss_llm": 0.0, "grad_norm": 0.0455465130507946, "global_step": 31977, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03148489072918892, "train_loss_bc": 0.03147083520889282, "train_loss_llm": 0.001405477523803711, "grad_norm": 0.06555534154176712, "global_step": 31978, "epoch": 888, "lr": 0.001} +{"train_loss": 0.029782038182020187, "train_loss_bc": 0.029782038182020187, "train_loss_llm": 0.0, "grad_norm": 0.08661144226789474, "global_step": 31979, "epoch": 888, "lr": 0.001} +{"train_loss": 0.026965651661157608, "train_loss_bc": 0.026965651661157608, "train_loss_llm": 0.0, "grad_norm": 0.12293853610754013, "global_step": 31980, "epoch": 888, "lr": 0.001} +{"train_loss": 0.035317156463861465, "train_loss_bc": 0.03530336543917656, "train_loss_llm": 0.0013792216777801514, "grad_norm": 0.045253507792949677, "global_step": 31981, "epoch": 888, "lr": 0.001} +{"train_loss": 0.026505911722779274, "train_loss_bc": 0.026500267907977104, "train_loss_llm": 0.0005644559860229492, "grad_norm": 0.061369333416223526, "global_step": 31982, "epoch": 888, "lr": 0.001} +{"train_loss": 0.025857670232653618, "train_loss_bc": 0.02585718035697937, "train_loss_llm": 4.902482032775879e-05, "grad_norm": 0.0808556005358696, "global_step": 31983, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03336148336529732, "train_loss_bc": 0.03336148336529732, "train_loss_llm": 0.0, "grad_norm": 0.09234049916267395, "global_step": 31984, "epoch": 888, "lr": 0.001} +{"train_loss": 0.025188300758600235, "train_loss_bc": 0.025188300758600235, "train_loss_llm": 0.0, "grad_norm": 0.03363664448261261, "global_step": 31985, "epoch": 888, "lr": 0.001} +{"train_loss": 0.023044057190418243, "train_loss_bc": 0.023044057190418243, "train_loss_llm": 0.0, "grad_norm": 0.05734715610742569, "global_step": 31986, "epoch": 888, "lr": 0.001} +{"train_loss": 0.023852359503507614, "train_loss_bc": 0.023852359503507614, "train_loss_llm": 0.0, "grad_norm": 0.061431240290403366, "global_step": 31987, "epoch": 888, "lr": 0.001} +{"train_loss": 0.026233762502670288, "train_loss_bc": 0.026233762502670288, "train_loss_llm": 0.0, "grad_norm": 0.0757768303155899, "global_step": 31988, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03191547095775604, "train_loss_bc": 0.03191547095775604, "train_loss_llm": 0.0, "grad_norm": 0.03653237596154213, "global_step": 31989, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03202902898192406, "train_loss_bc": 0.03202902898192406, "train_loss_llm": 0.0, "grad_norm": 0.05433660000562668, "global_step": 31990, "epoch": 888, "lr": 0.001} +{"train_loss": 0.022582653909921646, "train_loss_bc": 0.022577758878469467, "train_loss_llm": 0.0004895031452178955, "grad_norm": 0.06953874230384827, "global_step": 31991, "epoch": 888, "lr": 0.001} +{"train_loss": 0.02936311438679695, "train_loss_bc": 0.02936311438679695, "train_loss_llm": 0.0, "grad_norm": 0.08668942749500275, "global_step": 31992, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03097270056605339, "train_loss_bc": 0.03097270056605339, "train_loss_llm": 0.0, "grad_norm": 0.04146962612867355, "global_step": 31993, "epoch": 888, "lr": 0.001} +{"train_loss": 0.030384400859475136, "train_loss_bc": 0.030384400859475136, "train_loss_llm": 0.0, "grad_norm": 0.0723898857831955, "global_step": 31994, "epoch": 888, "lr": 0.001} +{"train_loss": 0.026934323832392693, "train_loss_bc": 0.026934323832392693, "train_loss_llm": 0.0, "grad_norm": 0.08729944378137589, "global_step": 31995, "epoch": 888, "lr": 0.001} +{"train_loss": 0.028702354058623314, "train_loss_bc": 0.02869061380624771, "train_loss_llm": 0.001173943281173706, "grad_norm": 0.105666883289814, "global_step": 31996, "epoch": 888, "lr": 0.001} +{"train_loss": 0.031147338449954987, "train_loss_bc": 0.031147338449954987, "train_loss_llm": 0.0, "grad_norm": 0.040567755699157715, "global_step": 31997, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03135388717055321, "train_loss_bc": 0.03135388717055321, "train_loss_llm": 0.0, "grad_norm": 0.06744860112667084, "global_step": 31998, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03444772586226463, "train_loss_bc": 0.03444772586226463, "train_loss_llm": 0.0, "grad_norm": 0.07218708097934723, "global_step": 31999, "epoch": 888, "lr": 0.001} +{"train_loss": 0.028888465836644173, "train_loss_bc": 0.028888465836644173, "train_loss_llm": 0.0, "grad_norm": 0.10090484470129013, "global_step": 32000, "epoch": 888, "lr": 0.001} +{"train_loss": 0.027462007477879524, "train_loss_bc": 0.027462007477879524, "train_loss_llm": 0.0, "grad_norm": 0.03292946144938469, "global_step": 32001, "epoch": 888, "lr": 0.001} +{"train_loss": 0.03064095787703991, "train_loss_bc": 0.03064095787703991, "train_loss_llm": 0.0, "grad_norm": 0.0504765547811985, "global_step": 32002, "epoch": 888, "lr": 0.001} +{"train_loss": 0.02929911907348368, "train_loss_bc": 0.02667565457522869, "train_loss_llm": 0.0, "grad_norm": 0.0757165253162384, "global_step": 32003, "epoch": 888, "lr": 0.001, "val_loss": 0.10259708762168884} +{"train_loss": 0.030614104121923447, "train_loss_bc": 0.030614104121923447, "train_loss_llm": 0.0, "grad_norm": 0.07987547665834427, "global_step": 32004, "epoch": 889, "lr": 0.001} +{"train_loss": 0.028332293033599854, "train_loss_bc": 0.02833089418709278, "train_loss_llm": 0.0001398622989654541, "grad_norm": 0.03281041607260704, "global_step": 32005, "epoch": 889, "lr": 0.001} +{"train_loss": 0.02697291597723961, "train_loss_bc": 0.02697291597723961, "train_loss_llm": 0.0, "grad_norm": 0.04484284669160843, "global_step": 32006, "epoch": 889, "lr": 0.001} +{"train_loss": 0.03387156501412392, "train_loss_bc": 0.03387156501412392, "train_loss_llm": 0.0, "grad_norm": 0.06264370679855347, "global_step": 32007, "epoch": 889, "lr": 0.001} +{"train_loss": 0.02899043634533882, "train_loss_bc": 0.02899043634533882, "train_loss_llm": 0.0, "grad_norm": 0.07469896227121353, "global_step": 32008, "epoch": 889, "lr": 0.001} +{"train_loss": 0.0316791757941246, "train_loss_bc": 0.031674597412347794, "train_loss_llm": 0.0004578828811645508, "grad_norm": 0.036744631826877594, "global_step": 32009, "epoch": 889, "lr": 0.001} +{"train_loss": 0.030755721032619476, "train_loss_bc": 0.030755721032619476, "train_loss_llm": 0.0, "grad_norm": 0.06117739900946617, "global_step": 32010, "epoch": 889, "lr": 0.001} +{"train_loss": 0.029801668599247932, "train_loss_bc": 0.029801668599247932, "train_loss_llm": 0.0, "grad_norm": 0.08532630652189255, "global_step": 32011, "epoch": 889, "lr": 0.001} +{"train_loss": 0.03164628893136978, "train_loss_bc": 0.03164628893136978, "train_loss_llm": 0.0, "grad_norm": 0.10768946260213852, "global_step": 32012, "epoch": 889, "lr": 0.001} +{"train_loss": 0.0253713708370924, "train_loss_bc": 0.025369184091687202, "train_loss_llm": 0.00021860003471374512, "grad_norm": 0.03780604153871536, "global_step": 32013, "epoch": 889, "lr": 0.001} +{"train_loss": 0.03440377116203308, "train_loss_bc": 0.03440377116203308, "train_loss_llm": 0.0, "grad_norm": 0.06676647812128067, "global_step": 32014, "epoch": 889, "lr": 0.001} +{"train_loss": 0.029959022998809814, "train_loss_bc": 0.029959022998809814, "train_loss_llm": 0.0, "grad_norm": 0.09605985879898071, "global_step": 32015, "epoch": 889, "lr": 0.001} +{"train_loss": 0.030189622193574905, "train_loss_bc": 0.030184656381607056, "train_loss_llm": 0.000496596097946167, "grad_norm": 0.1148688942193985, "global_step": 32016, "epoch": 889, "lr": 0.001} +{"train_loss": 0.025786232203245163, "train_loss_bc": 0.025786232203245163, "train_loss_llm": 0.0, "grad_norm": 0.03777081146836281, "global_step": 32017, "epoch": 889, "lr": 0.001} +{"train_loss": 0.02973012439906597, "train_loss_bc": 0.02973012439906597, "train_loss_llm": 0.0, "grad_norm": 0.07088857889175415, "global_step": 32018, "epoch": 889, "lr": 0.001} +{"train_loss": 0.028496548533439636, "train_loss_bc": 0.028494341298937798, "train_loss_llm": 0.0002206563949584961, "grad_norm": 0.0866248831152916, "global_step": 32019, "epoch": 889, "lr": 0.001} +{"train_loss": 0.029267137870192528, "train_loss_bc": 0.029267137870192528, "train_loss_llm": 0.0, "grad_norm": 0.11787020415067673, "global_step": 32020, "epoch": 889, "lr": 0.001} +{"train_loss": 0.03217122331261635, "train_loss_bc": 0.03217122331261635, "train_loss_llm": 0.0, "grad_norm": 0.034245915710926056, "global_step": 32021, "epoch": 889, "lr": 0.001} +{"train_loss": 0.02744392678141594, "train_loss_bc": 0.02744392678141594, "train_loss_llm": 0.0, "grad_norm": 0.05359649285674095, "global_step": 32022, "epoch": 889, "lr": 0.001} +{"train_loss": 0.02502567693591118, "train_loss_bc": 0.025023430585861206, "train_loss_llm": 0.00022456049919128418, "grad_norm": 0.06337694078683853, "global_step": 32023, "epoch": 889, "lr": 0.001} +{"train_loss": 0.028711313381791115, "train_loss_bc": 0.028698593378067017, "train_loss_llm": 0.001271963119506836, "grad_norm": 0.07360971719026566, "global_step": 32024, "epoch": 889, "lr": 0.001} +{"train_loss": 0.026976071298122406, "train_loss_bc": 0.026976071298122406, "train_loss_llm": 0.0, "grad_norm": 0.041480712592601776, "global_step": 32025, "epoch": 889, "lr": 0.001} +{"train_loss": 0.033065225929021835, "train_loss_bc": 0.03306029364466667, "train_loss_llm": 0.0004933476448059082, "grad_norm": 0.07453805208206177, "global_step": 32026, "epoch": 889, "lr": 0.001} +{"train_loss": 0.026152102276682854, "train_loss_bc": 0.026152102276682854, "train_loss_llm": 0.0, "grad_norm": 0.09647975116968155, "global_step": 32027, "epoch": 889, "lr": 0.001} +{"train_loss": 0.027426544576883316, "train_loss_bc": 0.027426544576883316, "train_loss_llm": 0.0, "grad_norm": 0.11482756584882736, "global_step": 32028, "epoch": 889, "lr": 0.001} +{"train_loss": 0.03162432089447975, "train_loss_bc": 0.03162432089447975, "train_loss_llm": 0.0, "grad_norm": 0.047537874430418015, "global_step": 32029, "epoch": 889, "lr": 0.001} +{"train_loss": 0.027605321258306503, "train_loss_bc": 0.02759224735200405, "train_loss_llm": 0.0013074278831481934, "grad_norm": 0.06736526638269424, "global_step": 32030, "epoch": 889, "lr": 0.001} +{"train_loss": 0.025105832144618034, "train_loss_bc": 0.025105832144618034, "train_loss_llm": 0.0, "grad_norm": 0.0818345919251442, "global_step": 32031, "epoch": 889, "lr": 0.001} +{"train_loss": 0.030442556366324425, "train_loss_bc": 0.030429329723119736, "train_loss_llm": 0.001322627067565918, "grad_norm": 0.10573558509349823, "global_step": 32032, "epoch": 889, "lr": 0.001} +{"train_loss": 0.031908221542835236, "train_loss_bc": 0.03188473358750343, "train_loss_llm": 0.0023488998413085938, "grad_norm": 0.05199778825044632, "global_step": 32033, "epoch": 889, "lr": 0.001} +{"train_loss": 0.033745281398296356, "train_loss_bc": 0.033745281398296356, "train_loss_llm": 0.0, "grad_norm": 0.07078182697296143, "global_step": 32034, "epoch": 889, "lr": 0.001} +{"train_loss": 0.027278676629066467, "train_loss_bc": 0.027278676629066467, "train_loss_llm": 0.0, "grad_norm": 0.10976321995258331, "global_step": 32035, "epoch": 889, "lr": 0.001} +{"train_loss": 0.029025889933109283, "train_loss_bc": 0.029025889933109283, "train_loss_llm": 0.0, "grad_norm": 0.14347653090953827, "global_step": 32036, "epoch": 889, "lr": 0.001} +{"train_loss": 0.028507325798273087, "train_loss_bc": 0.028507325798273087, "train_loss_llm": 0.0, "grad_norm": 0.042945925146341324, "global_step": 32037, "epoch": 889, "lr": 0.001} +{"train_loss": 0.030667155981063843, "train_loss_bc": 0.030667155981063843, "train_loss_llm": 0.0, "grad_norm": 0.06820381432771683, "global_step": 32038, "epoch": 889, "lr": 0.001} +{"train_loss": 0.02947216336097982, "train_loss_bc": 0.03224194422364235, "train_loss_llm": 0.0005269646644592285, "grad_norm": 0.09282124042510986, "global_step": 32039, "epoch": 889, "lr": 0.001, "val_loss": 0.10238251090049744} +{"train_loss": 0.02883443608880043, "train_loss_bc": 0.02883443608880043, "train_loss_llm": 0.0, "grad_norm": 0.11291446536779404, "global_step": 32040, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03119800239801407, "train_loss_bc": 0.03119800239801407, "train_loss_llm": 0.0, "grad_norm": 0.03668099269270897, "global_step": 32041, "epoch": 890, "lr": 0.001} +{"train_loss": 0.026856396347284317, "train_loss_bc": 0.026856396347284317, "train_loss_llm": 0.0, "grad_norm": 0.05369238182902336, "global_step": 32042, "epoch": 890, "lr": 0.001} +{"train_loss": 0.028785454109311104, "train_loss_bc": 0.028785454109311104, "train_loss_llm": 0.0, "grad_norm": 0.06688545644283295, "global_step": 32043, "epoch": 890, "lr": 0.001} +{"train_loss": 0.038035716861486435, "train_loss_bc": 0.0380140021443367, "train_loss_llm": 0.0021713674068450928, "grad_norm": 0.08703434467315674, "global_step": 32044, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03039107285439968, "train_loss_bc": 0.03038909286260605, "train_loss_llm": 0.00019803643226623535, "grad_norm": 0.03642473369836807, "global_step": 32045, "epoch": 890, "lr": 0.001} +{"train_loss": 0.029421525076031685, "train_loss_bc": 0.029421525076031685, "train_loss_llm": 0.0, "grad_norm": 0.0503210686147213, "global_step": 32046, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03328152373433113, "train_loss_bc": 0.03328152373433113, "train_loss_llm": 0.0, "grad_norm": 0.06938748061656952, "global_step": 32047, "epoch": 890, "lr": 0.001} +{"train_loss": 0.030274853110313416, "train_loss_bc": 0.030274853110313416, "train_loss_llm": 0.0, "grad_norm": 0.08725428581237793, "global_step": 32048, "epoch": 890, "lr": 0.001} +{"train_loss": 0.026151318103075027, "train_loss_bc": 0.026151318103075027, "train_loss_llm": 0.0, "grad_norm": 0.035008661448955536, "global_step": 32049, "epoch": 890, "lr": 0.001} +{"train_loss": 0.024424148723483086, "train_loss_bc": 0.024424148723483086, "train_loss_llm": 0.0, "grad_norm": 0.060121744871139526, "global_step": 32050, "epoch": 890, "lr": 0.001} +{"train_loss": 0.030198341235518456, "train_loss_bc": 0.030198341235518456, "train_loss_llm": 0.0, "grad_norm": 0.08459152281284332, "global_step": 32051, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03242994844913483, "train_loss_bc": 0.03242333233356476, "train_loss_llm": 0.0006614327430725098, "grad_norm": 0.10454883426427841, "global_step": 32052, "epoch": 890, "lr": 0.001} +{"train_loss": 0.021703170612454414, "train_loss_bc": 0.02169639803469181, "train_loss_llm": 0.0006773173809051514, "grad_norm": 0.0334409698843956, "global_step": 32053, "epoch": 890, "lr": 0.001} +{"train_loss": 0.02878444269299507, "train_loss_bc": 0.02878444269299507, "train_loss_llm": 0.0, "grad_norm": 0.04911470785737038, "global_step": 32054, "epoch": 890, "lr": 0.001} +{"train_loss": 0.026107292622327805, "train_loss_bc": 0.026107292622327805, "train_loss_llm": 0.0, "grad_norm": 0.0895259901881218, "global_step": 32055, "epoch": 890, "lr": 0.001} +{"train_loss": 0.023310571908950806, "train_loss_bc": 0.023310571908950806, "train_loss_llm": 0.0, "grad_norm": 0.0994485393166542, "global_step": 32056, "epoch": 890, "lr": 0.001} +{"train_loss": 0.029843240976333618, "train_loss_bc": 0.029843240976333618, "train_loss_llm": 0.0, "grad_norm": 0.03911584988236427, "global_step": 32057, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03111322596669197, "train_loss_bc": 0.031108368188142776, "train_loss_llm": 0.0004858672618865967, "grad_norm": 0.06049985811114311, "global_step": 32058, "epoch": 890, "lr": 0.001} +{"train_loss": 0.029574500396847725, "train_loss_bc": 0.029574500396847725, "train_loss_llm": 0.0, "grad_norm": 0.08489295840263367, "global_step": 32059, "epoch": 890, "lr": 0.001} +{"train_loss": 0.02681579813361168, "train_loss_bc": 0.02681579813361168, "train_loss_llm": 0.0, "grad_norm": 0.11203958839178085, "global_step": 32060, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03297442942857742, "train_loss_bc": 0.032964203506708145, "train_loss_llm": 0.0010225474834442139, "grad_norm": 0.03782709687948227, "global_step": 32061, "epoch": 890, "lr": 0.001} +{"train_loss": 0.028261398896574974, "train_loss_bc": 0.028258373960852623, "train_loss_llm": 0.0003025233745574951, "grad_norm": 0.05567103251814842, "global_step": 32062, "epoch": 890, "lr": 0.001} +{"train_loss": 0.027618957683444023, "train_loss_bc": 0.027618957683444023, "train_loss_llm": 0.0, "grad_norm": 0.07573837041854858, "global_step": 32063, "epoch": 890, "lr": 0.001} +{"train_loss": 0.02886897325515747, "train_loss_bc": 0.02886897325515747, "train_loss_llm": 0.0, "grad_norm": 0.09012025594711304, "global_step": 32064, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03176083043217659, "train_loss_bc": 0.03176083043217659, "train_loss_llm": 0.0, "grad_norm": 0.033461183309555054, "global_step": 32065, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03086789697408676, "train_loss_bc": 0.03086789697408676, "train_loss_llm": 0.0, "grad_norm": 0.052791766822338104, "global_step": 32066, "epoch": 890, "lr": 0.001} +{"train_loss": 0.02623898722231388, "train_loss_bc": 0.026237405836582184, "train_loss_llm": 0.00015816092491149902, "grad_norm": 0.06444758921861649, "global_step": 32067, "epoch": 890, "lr": 0.001} +{"train_loss": 0.033273935317993164, "train_loss_bc": 0.033260516822338104, "train_loss_llm": 0.0013418197631835938, "grad_norm": 0.09148633480072021, "global_step": 32068, "epoch": 890, "lr": 0.001} +{"train_loss": 0.02680269628763199, "train_loss_bc": 0.02680269628763199, "train_loss_llm": 0.0, "grad_norm": 0.02900596708059311, "global_step": 32069, "epoch": 890, "lr": 0.001} +{"train_loss": 0.028992582112550735, "train_loss_bc": 0.028988385573029518, "train_loss_llm": 0.0004196465015411377, "grad_norm": 0.05897083878517151, "global_step": 32070, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03455395624041557, "train_loss_bc": 0.034553926438093185, "train_loss_llm": 2.9206275939941406e-06, "grad_norm": 0.07429791241884232, "global_step": 32071, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03291440010070801, "train_loss_bc": 0.03291440010070801, "train_loss_llm": 0.0, "grad_norm": 0.08679335564374924, "global_step": 32072, "epoch": 890, "lr": 0.001} +{"train_loss": 0.03179274871945381, "train_loss_bc": 0.03179274871945381, "train_loss_llm": 0.0, "grad_norm": 0.04532329738140106, "global_step": 32073, "epoch": 890, "lr": 0.001} +{"train_loss": 0.025592399761080742, "train_loss_bc": 0.025592399761080742, "train_loss_llm": 0.0, "grad_norm": 0.06824730336666107, "global_step": 32074, "epoch": 890, "lr": 0.001} +{"train_loss": 0.02943012330474125, "train_loss_bc": 0.03143252059817314, "train_loss_llm": 0.00027447938919067383, "grad_norm": 0.09237483143806458, "global_step": 32075, "epoch": 890, "lr": 0.001, "train/cumulative_reward": 55.43085753822802, "train/mean_score": 8.705037498217933, "train/success_rate": 0.2, "test/cumulative_reward": 42.04709920565052, "test/mean_score": 7.979381017993854, "test/success_rate": 0.2, "val_loss": 0.10425782203674316, "train_action_mse_error": 0.03208416327834129} +{"train_loss": 0.033538103103637695, "train_loss_bc": 0.03351648151874542, "train_loss_llm": 0.0021623075008392334, "grad_norm": 0.11265432834625244, "global_step": 32076, "epoch": 891, "lr": 0.001} +{"train_loss": 0.026865070685744286, "train_loss_bc": 0.026865070685744286, "train_loss_llm": 0.0, "grad_norm": 0.03176133707165718, "global_step": 32077, "epoch": 891, "lr": 0.001} +{"train_loss": 0.031997039914131165, "train_loss_bc": 0.031997039914131165, "train_loss_llm": 0.0, "grad_norm": 0.05632062628865242, "global_step": 32078, "epoch": 891, "lr": 0.001} +{"train_loss": 0.03374755010008812, "train_loss_bc": 0.03373610973358154, "train_loss_llm": 0.0011439025402069092, "grad_norm": 0.07850142568349838, "global_step": 32079, "epoch": 891, "lr": 0.001} +{"train_loss": 0.0256978590041399, "train_loss_bc": 0.0256978590041399, "train_loss_llm": 0.0, "grad_norm": 0.09252011030912399, "global_step": 32080, "epoch": 891, "lr": 0.001} +{"train_loss": 0.025342624634504318, "train_loss_bc": 0.02534257248044014, "train_loss_llm": 5.125999450683594e-06, "grad_norm": 0.031106185168027878, "global_step": 32081, "epoch": 891, "lr": 0.001} +{"train_loss": 0.03342937305569649, "train_loss_bc": 0.03341374546289444, "train_loss_llm": 0.0015626847743988037, "grad_norm": 0.05744745582342148, "global_step": 32082, "epoch": 891, "lr": 0.001} +{"train_loss": 0.026965556666254997, "train_loss_bc": 0.026965556666254997, "train_loss_llm": 0.0, "grad_norm": 0.06927613914012909, "global_step": 32083, "epoch": 891, "lr": 0.001} +{"train_loss": 0.021614184603095055, "train_loss_bc": 0.021608078852295876, "train_loss_llm": 0.000610649585723877, "grad_norm": 0.08096189796924591, "global_step": 32084, "epoch": 891, "lr": 0.001} +{"train_loss": 0.027051148936152458, "train_loss_bc": 0.027051148936152458, "train_loss_llm": 0.0, "grad_norm": 0.0327288992702961, "global_step": 32085, "epoch": 891, "lr": 0.001} +{"train_loss": 0.03133406117558479, "train_loss_bc": 0.03133406117558479, "train_loss_llm": 0.0, "grad_norm": 0.059005580842494965, "global_step": 32086, "epoch": 891, "lr": 0.001} +{"train_loss": 0.027669940143823624, "train_loss_bc": 0.027669940143823624, "train_loss_llm": 0.0, "grad_norm": 0.08318424969911575, "global_step": 32087, "epoch": 891, "lr": 0.001} +{"train_loss": 0.026882760226726532, "train_loss_bc": 0.026882760226726532, "train_loss_llm": 0.0, "grad_norm": 0.09795618802309036, "global_step": 32088, "epoch": 891, "lr": 0.001} +{"train_loss": 0.028027402237057686, "train_loss_bc": 0.028027402237057686, "train_loss_llm": 0.0, "grad_norm": 0.03625697270035744, "global_step": 32089, "epoch": 891, "lr": 0.001} +{"train_loss": 0.03323667496442795, "train_loss_bc": 0.03323667496442795, "train_loss_llm": 0.0, "grad_norm": 0.05461020767688751, "global_step": 32090, "epoch": 891, "lr": 0.001} +{"train_loss": 0.027216631919145584, "train_loss_bc": 0.027216631919145584, "train_loss_llm": 0.0, "grad_norm": 0.07714474946260452, "global_step": 32091, "epoch": 891, "lr": 0.001} +{"train_loss": 0.02987102046608925, "train_loss_bc": 0.02987102046608925, "train_loss_llm": 0.0, "grad_norm": 0.0883852168917656, "global_step": 32092, "epoch": 891, "lr": 0.001} +{"train_loss": 0.028987614437937737, "train_loss_bc": 0.028987614437937737, "train_loss_llm": 0.0, "grad_norm": 0.045114435255527496, "global_step": 32093, "epoch": 891, "lr": 0.001} +{"train_loss": 0.032494064420461655, "train_loss_bc": 0.03247521445155144, "train_loss_llm": 0.0018851757049560547, "grad_norm": 0.0762820765376091, "global_step": 32094, "epoch": 891, "lr": 0.001} +{"train_loss": 0.03168247640132904, "train_loss_bc": 0.03168247640132904, "train_loss_llm": 0.0, "grad_norm": 0.10069098323583603, "global_step": 32095, "epoch": 891, "lr": 0.001} +{"train_loss": 0.03137325495481491, "train_loss_bc": 0.03137080743908882, "train_loss_llm": 0.0002449154853820801, "grad_norm": 0.13046681880950928, "global_step": 32096, "epoch": 891, "lr": 0.001} +{"train_loss": 0.03236439824104309, "train_loss_bc": 0.03236439824104309, "train_loss_llm": 0.0, "grad_norm": 0.040288813412189484, "global_step": 32097, "epoch": 891, "lr": 0.001} +{"train_loss": 0.029465284198522568, "train_loss_bc": 0.029465284198522568, "train_loss_llm": 0.0, "grad_norm": 0.05825112760066986, "global_step": 32098, "epoch": 891, "lr": 0.001} +{"train_loss": 0.02612057887017727, "train_loss_bc": 0.02612057887017727, "train_loss_llm": 0.0, "grad_norm": 0.07013846933841705, "global_step": 32099, "epoch": 891, "lr": 0.001} +{"train_loss": 0.027468031272292137, "train_loss_bc": 0.027468031272292137, "train_loss_llm": 0.0, "grad_norm": 0.07843048870563507, "global_step": 32100, "epoch": 891, "lr": 0.001} +{"train_loss": 0.02813580632209778, "train_loss_bc": 0.02813580632209778, "train_loss_llm": 0.0, "grad_norm": 0.031597480177879333, "global_step": 32101, "epoch": 891, "lr": 0.001} +{"train_loss": 0.026743633672595024, "train_loss_bc": 0.026743633672595024, "train_loss_llm": 0.0, "grad_norm": 0.05018456652760506, "global_step": 32102, "epoch": 891, "lr": 0.001} +{"train_loss": 0.03120981529355049, "train_loss_bc": 0.03120981529355049, "train_loss_llm": 0.0, "grad_norm": 0.0680985152721405, "global_step": 32103, "epoch": 891, "lr": 0.001} +{"train_loss": 0.031608305871486664, "train_loss_bc": 0.031608305871486664, "train_loss_llm": 0.0, "grad_norm": 0.09418226033449173, "global_step": 32104, "epoch": 891, "lr": 0.001} +{"train_loss": 0.025002596899867058, "train_loss_bc": 0.025002596899867058, "train_loss_llm": 0.0, "grad_norm": 0.030864886939525604, "global_step": 32105, "epoch": 891, "lr": 0.001} +{"train_loss": 0.026005590334534645, "train_loss_bc": 0.026002125814557076, "train_loss_llm": 0.0003465414047241211, "grad_norm": 0.04620453342795372, "global_step": 32106, "epoch": 891, "lr": 0.001} +{"train_loss": 0.027240654453635216, "train_loss_bc": 0.027240654453635216, "train_loss_llm": 0.0, "grad_norm": 0.06576444953680038, "global_step": 32107, "epoch": 891, "lr": 0.001} +{"train_loss": 0.0348348468542099, "train_loss_bc": 0.034827884286642075, "train_loss_llm": 0.000696331262588501, "grad_norm": 0.08649546653032303, "global_step": 32108, "epoch": 891, "lr": 0.001} +{"train_loss": 0.032492831349372864, "train_loss_bc": 0.032492831349372864, "train_loss_llm": 0.0, "grad_norm": 0.03519774600863457, "global_step": 32109, "epoch": 891, "lr": 0.001} +{"train_loss": 0.0301382876932621, "train_loss_bc": 0.0301382876932621, "train_loss_llm": 0.0, "grad_norm": 0.06292473524808884, "global_step": 32110, "epoch": 891, "lr": 0.001} +{"train_loss": 0.029225292667332623, "train_loss_bc": 0.028255462646484375, "train_loss_llm": 0.0, "grad_norm": 0.0814920961856842, "global_step": 32111, "epoch": 891, "lr": 0.001, "val_loss": 0.10129439830780029} +{"train_loss": 0.02806725539267063, "train_loss_bc": 0.02806725539267063, "train_loss_llm": 0.0, "grad_norm": 0.10446159541606903, "global_step": 32112, "epoch": 892, "lr": 0.001} +{"train_loss": 0.028435617685317993, "train_loss_bc": 0.02843138948082924, "train_loss_llm": 0.0004228353500366211, "grad_norm": 0.04162395000457764, "global_step": 32113, "epoch": 892, "lr": 0.001} +{"train_loss": 0.027819868177175522, "train_loss_bc": 0.027819868177175522, "train_loss_llm": 0.0, "grad_norm": 0.06061439588665962, "global_step": 32114, "epoch": 892, "lr": 0.001} +{"train_loss": 0.02841678261756897, "train_loss_bc": 0.02841678261756897, "train_loss_llm": 0.0, "grad_norm": 0.0681554451584816, "global_step": 32115, "epoch": 892, "lr": 0.001} +{"train_loss": 0.03207462280988693, "train_loss_bc": 0.03207462280988693, "train_loss_llm": 0.0, "grad_norm": 0.09215795993804932, "global_step": 32116, "epoch": 892, "lr": 0.001} +{"train_loss": 0.020398501306772232, "train_loss_bc": 0.020398501306772232, "train_loss_llm": 0.0, "grad_norm": 0.032099444419145584, "global_step": 32117, "epoch": 892, "lr": 0.001} +{"train_loss": 0.030835313722491264, "train_loss_bc": 0.030821608379483223, "train_loss_llm": 0.0013705790042877197, "grad_norm": 0.05110526457428932, "global_step": 32118, "epoch": 892, "lr": 0.001} +{"train_loss": 0.02665284276008606, "train_loss_bc": 0.02665284276008606, "train_loss_llm": 0.0, "grad_norm": 0.06929736584424973, "global_step": 32119, "epoch": 892, "lr": 0.001} +{"train_loss": 0.030308648943901062, "train_loss_bc": 0.030308648943901062, "train_loss_llm": 0.0, "grad_norm": 0.09294169396162033, "global_step": 32120, "epoch": 892, "lr": 0.001} +{"train_loss": 0.03367667272686958, "train_loss_bc": 0.03367667272686958, "train_loss_llm": 0.0, "grad_norm": 0.03946137800812721, "global_step": 32121, "epoch": 892, "lr": 0.001} +{"train_loss": 0.031269535422325134, "train_loss_bc": 0.03126025199890137, "train_loss_llm": 0.0009281933307647705, "grad_norm": 0.05997069552540779, "global_step": 32122, "epoch": 892, "lr": 0.001} +{"train_loss": 0.027791092172265053, "train_loss_bc": 0.027791092172265053, "train_loss_llm": 0.0, "grad_norm": 0.0816086083650589, "global_step": 32123, "epoch": 892, "lr": 0.001} +{"train_loss": 0.02840997464954853, "train_loss_bc": 0.028403330594301224, "train_loss_llm": 0.0006644725799560547, "grad_norm": 0.08911602944135666, "global_step": 32124, "epoch": 892, "lr": 0.001} +{"train_loss": 0.027236077934503555, "train_loss_bc": 0.027236077934503555, "train_loss_llm": 0.0, "grad_norm": 0.03458934649825096, "global_step": 32125, "epoch": 892, "lr": 0.001} +{"train_loss": 0.030697986483573914, "train_loss_bc": 0.030697986483573914, "train_loss_llm": 0.0, "grad_norm": 0.050021279603242874, "global_step": 32126, "epoch": 892, "lr": 0.001} +{"train_loss": 0.03375174477696419, "train_loss_bc": 0.03375174477696419, "train_loss_llm": 0.0, "grad_norm": 0.0650436282157898, "global_step": 32127, "epoch": 892, "lr": 0.001} +{"train_loss": 0.028025710955262184, "train_loss_bc": 0.028025710955262184, "train_loss_llm": 0.0, "grad_norm": 0.08403908461332321, "global_step": 32128, "epoch": 892, "lr": 0.001} +{"train_loss": 0.02952953800559044, "train_loss_bc": 0.029522161930799484, "train_loss_llm": 0.000737607479095459, "grad_norm": 0.032511405646800995, "global_step": 32129, "epoch": 892, "lr": 0.001} +{"train_loss": 0.03184373304247856, "train_loss_bc": 0.03184076026082039, "train_loss_llm": 0.00029724836349487305, "grad_norm": 0.052584871649742126, "global_step": 32130, "epoch": 892, "lr": 0.001} +{"train_loss": 0.031467583030462265, "train_loss_bc": 0.031467583030462265, "train_loss_llm": 0.0, "grad_norm": 0.06458601355552673, "global_step": 32131, "epoch": 892, "lr": 0.001} +{"train_loss": 0.023684505373239517, "train_loss_bc": 0.02368219383060932, "train_loss_llm": 0.00023114681243896484, "grad_norm": 0.07719322293996811, "global_step": 32132, "epoch": 892, "lr": 0.001} +{"train_loss": 0.032626498490571976, "train_loss_bc": 0.032626498490571976, "train_loss_llm": 0.0, "grad_norm": 0.037502724677324295, "global_step": 32133, "epoch": 892, "lr": 0.001} +{"train_loss": 0.030952436849474907, "train_loss_bc": 0.030952436849474907, "train_loss_llm": 0.0, "grad_norm": 0.060540325939655304, "global_step": 32134, "epoch": 892, "lr": 0.001} +{"train_loss": 0.028951896354556084, "train_loss_bc": 0.028951896354556084, "train_loss_llm": 0.0, "grad_norm": 0.06941717118024826, "global_step": 32135, "epoch": 892, "lr": 0.001} +{"train_loss": 0.023327382281422615, "train_loss_bc": 0.023327382281422615, "train_loss_llm": 0.0, "grad_norm": 0.09236879646778107, "global_step": 32136, "epoch": 892, "lr": 0.001} +{"train_loss": 0.024530308321118355, "train_loss_bc": 0.024530308321118355, "train_loss_llm": 0.0, "grad_norm": 0.03522396460175514, "global_step": 32137, "epoch": 892, "lr": 0.001} +{"train_loss": 0.031180365011096, "train_loss_bc": 0.031180365011096, "train_loss_llm": 0.0, "grad_norm": 0.05512193962931633, "global_step": 32138, "epoch": 892, "lr": 0.001} +{"train_loss": 0.036822788417339325, "train_loss_bc": 0.036822788417339325, "train_loss_llm": 0.0, "grad_norm": 0.08326004445552826, "global_step": 32139, "epoch": 892, "lr": 0.001} +{"train_loss": 0.03161083534359932, "train_loss_bc": 0.03161083534359932, "train_loss_llm": 0.0, "grad_norm": 0.10097876936197281, "global_step": 32140, "epoch": 892, "lr": 0.001} +{"train_loss": 0.02320084534585476, "train_loss_bc": 0.02320084534585476, "train_loss_llm": 0.0, "grad_norm": 0.03307810053229332, "global_step": 32141, "epoch": 892, "lr": 0.001} +{"train_loss": 0.026741180568933487, "train_loss_bc": 0.026741180568933487, "train_loss_llm": 0.0, "grad_norm": 0.053406041115522385, "global_step": 32142, "epoch": 892, "lr": 0.001} +{"train_loss": 0.03145585581660271, "train_loss_bc": 0.03145039826631546, "train_loss_llm": 0.000545799732208252, "grad_norm": 0.07750024646520615, "global_step": 32143, "epoch": 892, "lr": 0.001} +{"train_loss": 0.02933333069086075, "train_loss_bc": 0.02933333069086075, "train_loss_llm": 0.0, "grad_norm": 0.09357359260320663, "global_step": 32144, "epoch": 892, "lr": 0.001} +{"train_loss": 0.03157111257314682, "train_loss_bc": 0.03157111257314682, "train_loss_llm": 0.0, "grad_norm": 0.04360498860478401, "global_step": 32145, "epoch": 892, "lr": 0.001} +{"train_loss": 0.027172086760401726, "train_loss_bc": 0.027172086760401726, "train_loss_llm": 0.0, "grad_norm": 0.07856298983097076, "global_step": 32146, "epoch": 892, "lr": 0.001} +{"train_loss": 0.029303400673800044, "train_loss_bc": 0.035051893442869186, "train_loss_llm": 0.0, "grad_norm": 0.09769695997238159, "global_step": 32147, "epoch": 892, "lr": 0.001, "val_loss": 0.10557832568883896} +{"train_loss": 0.029294369742274284, "train_loss_bc": 0.029294369742274284, "train_loss_llm": 0.0, "grad_norm": 0.11267007887363434, "global_step": 32148, "epoch": 893, "lr": 0.001} +{"train_loss": 0.03121514432132244, "train_loss_bc": 0.03121514432132244, "train_loss_llm": 0.0, "grad_norm": 0.04044361785054207, "global_step": 32149, "epoch": 893, "lr": 0.001} +{"train_loss": 0.028983481228351593, "train_loss_bc": 0.028983481228351593, "train_loss_llm": 0.0, "grad_norm": 0.06367228925228119, "global_step": 32150, "epoch": 893, "lr": 0.001} +{"train_loss": 0.03085620142519474, "train_loss_bc": 0.03085620142519474, "train_loss_llm": 0.0, "grad_norm": 0.08115489035844803, "global_step": 32151, "epoch": 893, "lr": 0.001} +{"train_loss": 0.027266686782240868, "train_loss_bc": 0.027266686782240868, "train_loss_llm": 0.0, "grad_norm": 0.10348565876483917, "global_step": 32152, "epoch": 893, "lr": 0.001} +{"train_loss": 0.034614548087120056, "train_loss_bc": 0.034614548087120056, "train_loss_llm": 0.0, "grad_norm": 0.05011045187711716, "global_step": 32153, "epoch": 893, "lr": 0.001} +{"train_loss": 0.02801918238401413, "train_loss_bc": 0.028010226786136627, "train_loss_llm": 0.0008956193923950195, "grad_norm": 0.0807407796382904, "global_step": 32154, "epoch": 893, "lr": 0.001} +{"train_loss": 0.030644632875919342, "train_loss_bc": 0.030644632875919342, "train_loss_llm": 0.0, "grad_norm": 0.10667213797569275, "global_step": 32155, "epoch": 893, "lr": 0.001} +{"train_loss": 0.029779352247714996, "train_loss_bc": 0.02977827750146389, "train_loss_llm": 0.00010749697685241699, "grad_norm": 0.12631483376026154, "global_step": 32156, "epoch": 893, "lr": 0.001} +{"train_loss": 0.0272942204028368, "train_loss_bc": 0.0272942204028368, "train_loss_llm": 0.0, "grad_norm": 0.028208965435624123, "global_step": 32157, "epoch": 893, "lr": 0.001} +{"train_loss": 0.03007751703262329, "train_loss_bc": 0.03006984293460846, "train_loss_llm": 0.0007673203945159912, "grad_norm": 0.04618271067738533, "global_step": 32158, "epoch": 893, "lr": 0.001} +{"train_loss": 0.033773548901081085, "train_loss_bc": 0.03375651314854622, "train_loss_llm": 0.0017035305500030518, "grad_norm": 0.06626932322978973, "global_step": 32159, "epoch": 893, "lr": 0.001} +{"train_loss": 0.028387365862727165, "train_loss_bc": 0.02838164195418358, "train_loss_llm": 0.0005723834037780762, "grad_norm": 0.08025180548429489, "global_step": 32160, "epoch": 893, "lr": 0.001} +{"train_loss": 0.030659141018986702, "train_loss_bc": 0.030659141018986702, "train_loss_llm": 0.0, "grad_norm": 0.03795551136136055, "global_step": 32161, "epoch": 893, "lr": 0.001} +{"train_loss": 0.02952582575380802, "train_loss_bc": 0.02952582575380802, "train_loss_llm": 0.0, "grad_norm": 0.07776803523302078, "global_step": 32162, "epoch": 893, "lr": 0.001} +{"train_loss": 0.036323998123407364, "train_loss_bc": 0.036323998123407364, "train_loss_llm": 0.0, "grad_norm": 0.10922910273075104, "global_step": 32163, "epoch": 893, "lr": 0.001} +{"train_loss": 0.03172491863369942, "train_loss_bc": 0.03171376511454582, "train_loss_llm": 0.0011153817176818848, "grad_norm": 0.1330142468214035, "global_step": 32164, "epoch": 893, "lr": 0.001} +{"train_loss": 0.02799791283905506, "train_loss_bc": 0.02799791283905506, "train_loss_llm": 0.0, "grad_norm": 0.04744529351592064, "global_step": 32165, "epoch": 893, "lr": 0.001} +{"train_loss": 0.029082540422677994, "train_loss_bc": 0.029082540422677994, "train_loss_llm": 0.0, "grad_norm": 0.06925187259912491, "global_step": 32166, "epoch": 893, "lr": 0.001} +{"train_loss": 0.030354227870702744, "train_loss_bc": 0.030354227870702744, "train_loss_llm": 0.0, "grad_norm": 0.09685181081295013, "global_step": 32167, "epoch": 893, "lr": 0.001} +{"train_loss": 0.026774266734719276, "train_loss_bc": 0.026774266734719276, "train_loss_llm": 0.0, "grad_norm": 0.11943709850311279, "global_step": 32168, "epoch": 893, "lr": 0.001} +{"train_loss": 0.02966931462287903, "train_loss_bc": 0.02966931462287903, "train_loss_llm": 0.0, "grad_norm": 0.033423490822315216, "global_step": 32169, "epoch": 893, "lr": 0.001} +{"train_loss": 0.027440696954727173, "train_loss_bc": 0.027424989268183708, "train_loss_llm": 0.0015707612037658691, "grad_norm": 0.0602433905005455, "global_step": 32170, "epoch": 893, "lr": 0.001} +{"train_loss": 0.02298068441450596, "train_loss_bc": 0.02298068441450596, "train_loss_llm": 0.0, "grad_norm": 0.07930375635623932, "global_step": 32171, "epoch": 893, "lr": 0.001} +{"train_loss": 0.027265368029475212, "train_loss_bc": 0.027265368029475212, "train_loss_llm": 0.0, "grad_norm": 0.10146181285381317, "global_step": 32172, "epoch": 893, "lr": 0.001} +{"train_loss": 0.029965899884700775, "train_loss_bc": 0.02996397390961647, "train_loss_llm": 0.0001926422119140625, "grad_norm": 0.036959320306777954, "global_step": 32173, "epoch": 893, "lr": 0.001} +{"train_loss": 0.02712041325867176, "train_loss_bc": 0.02712041325867176, "train_loss_llm": 0.0, "grad_norm": 0.05601336807012558, "global_step": 32174, "epoch": 893, "lr": 0.001} +{"train_loss": 0.027913037687540054, "train_loss_bc": 0.027913037687540054, "train_loss_llm": 0.0, "grad_norm": 0.07846759259700775, "global_step": 32175, "epoch": 893, "lr": 0.001} +{"train_loss": 0.025324173271656036, "train_loss_bc": 0.025324173271656036, "train_loss_llm": 0.0, "grad_norm": 0.09531264752149582, "global_step": 32176, "epoch": 893, "lr": 0.001} +{"train_loss": 0.024742264300584793, "train_loss_bc": 0.02474178932607174, "train_loss_llm": 4.747509956359863e-05, "grad_norm": 0.031954411417245865, "global_step": 32177, "epoch": 893, "lr": 0.001} +{"train_loss": 0.029099207371473312, "train_loss_bc": 0.029099207371473312, "train_loss_llm": 0.0, "grad_norm": 0.05216295272111893, "global_step": 32178, "epoch": 893, "lr": 0.001} +{"train_loss": 0.03212305158376694, "train_loss_bc": 0.032108280807733536, "train_loss_llm": 0.0014770925045013428, "grad_norm": 0.06379374116659164, "global_step": 32179, "epoch": 893, "lr": 0.001} +{"train_loss": 0.02398926578462124, "train_loss_bc": 0.02398926578462124, "train_loss_llm": 0.0, "grad_norm": 0.07696773111820221, "global_step": 32180, "epoch": 893, "lr": 0.001} +{"train_loss": 0.032178204506635666, "train_loss_bc": 0.03216687962412834, "train_loss_llm": 0.0011323094367980957, "grad_norm": 0.03465647995471954, "global_step": 32181, "epoch": 893, "lr": 0.001} +{"train_loss": 0.026945631951093674, "train_loss_bc": 0.026925576850771904, "train_loss_llm": 0.002005547285079956, "grad_norm": 0.05801369622349739, "global_step": 32182, "epoch": 893, "lr": 0.001} +{"train_loss": 0.029427934231029615, "train_loss_bc": 0.0399993360042572, "train_loss_llm": 0.0, "grad_norm": 0.08739756792783737, "global_step": 32183, "epoch": 893, "lr": 0.001, "val_loss": 0.104583241045475} +{"train_loss": 0.032342586666345596, "train_loss_bc": 0.03233000636100769, "train_loss_llm": 0.0012580454349517822, "grad_norm": 0.11159610003232956, "global_step": 32184, "epoch": 894, "lr": 0.001} +{"train_loss": 0.028684543445706367, "train_loss_bc": 0.028684543445706367, "train_loss_llm": 0.0, "grad_norm": 0.051400359719991684, "global_step": 32185, "epoch": 894, "lr": 0.001} +{"train_loss": 0.025755396112799644, "train_loss_bc": 0.025755396112799644, "train_loss_llm": 0.0, "grad_norm": 0.07906295359134674, "global_step": 32186, "epoch": 894, "lr": 0.001} +{"train_loss": 0.029383083805441856, "train_loss_bc": 0.029383083805441856, "train_loss_llm": 0.0, "grad_norm": 0.10788247734308243, "global_step": 32187, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03018827550113201, "train_loss_bc": 0.03018827550113201, "train_loss_llm": 0.0, "grad_norm": 0.15934140980243683, "global_step": 32188, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03071773611009121, "train_loss_bc": 0.03071773611009121, "train_loss_llm": 0.0, "grad_norm": 0.037284813821315765, "global_step": 32189, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03654681146144867, "train_loss_bc": 0.03654681146144867, "train_loss_llm": 0.0, "grad_norm": 0.06503623723983765, "global_step": 32190, "epoch": 894, "lr": 0.001} +{"train_loss": 0.025071771815419197, "train_loss_bc": 0.025071771815419197, "train_loss_llm": 0.0, "grad_norm": 0.08903282880783081, "global_step": 32191, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03006911091506481, "train_loss_bc": 0.03006541170179844, "train_loss_llm": 0.0003698766231536865, "grad_norm": 0.11453786492347717, "global_step": 32192, "epoch": 894, "lr": 0.001} +{"train_loss": 0.026894675567746162, "train_loss_bc": 0.026887090876698494, "train_loss_llm": 0.000758439302444458, "grad_norm": 0.03300073370337486, "global_step": 32193, "epoch": 894, "lr": 0.001} +{"train_loss": 0.029705028980970383, "train_loss_bc": 0.029705028980970383, "train_loss_llm": 0.0, "grad_norm": 0.059460047632455826, "global_step": 32194, "epoch": 894, "lr": 0.001} +{"train_loss": 0.031873542815446854, "train_loss_bc": 0.03184601664543152, "train_loss_llm": 0.002752542495727539, "grad_norm": 0.08680421113967896, "global_step": 32195, "epoch": 894, "lr": 0.001} +{"train_loss": 0.030157875269651413, "train_loss_bc": 0.030157875269651413, "train_loss_llm": 0.0, "grad_norm": 0.10360848903656006, "global_step": 32196, "epoch": 894, "lr": 0.001} +{"train_loss": 0.029388364404439926, "train_loss_bc": 0.029388364404439926, "train_loss_llm": 0.0, "grad_norm": 0.03362030163407326, "global_step": 32197, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03244391828775406, "train_loss_bc": 0.03244391828775406, "train_loss_llm": 0.0, "grad_norm": 0.05138948932290077, "global_step": 32198, "epoch": 894, "lr": 0.001} +{"train_loss": 0.028254065662622452, "train_loss_bc": 0.028254065662622452, "train_loss_llm": 0.0, "grad_norm": 0.0705372616648674, "global_step": 32199, "epoch": 894, "lr": 0.001} +{"train_loss": 0.02464967779815197, "train_loss_bc": 0.02464967779815197, "train_loss_llm": 0.0, "grad_norm": 0.08923225849866867, "global_step": 32200, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03302657604217529, "train_loss_bc": 0.03302657604217529, "train_loss_llm": 0.0, "grad_norm": 0.044027309864759445, "global_step": 32201, "epoch": 894, "lr": 0.001} +{"train_loss": 0.023763785138726234, "train_loss_bc": 0.023763785138726234, "train_loss_llm": 0.0, "grad_norm": 0.07636450976133347, "global_step": 32202, "epoch": 894, "lr": 0.001} +{"train_loss": 0.026169197633862495, "train_loss_bc": 0.026169197633862495, "train_loss_llm": 0.0, "grad_norm": 0.08990810811519623, "global_step": 32203, "epoch": 894, "lr": 0.001} +{"train_loss": 0.026944439858198166, "train_loss_bc": 0.026944439858198166, "train_loss_llm": 0.0, "grad_norm": 0.10533984750509262, "global_step": 32204, "epoch": 894, "lr": 0.001} +{"train_loss": 0.036392904818058014, "train_loss_bc": 0.036392904818058014, "train_loss_llm": 0.0, "grad_norm": 0.05198046565055847, "global_step": 32205, "epoch": 894, "lr": 0.001} +{"train_loss": 0.025233596563339233, "train_loss_bc": 0.025228191167116165, "train_loss_llm": 0.0005405843257904053, "grad_norm": 0.07155709713697433, "global_step": 32206, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03273376449942589, "train_loss_bc": 0.03273376449942589, "train_loss_llm": 0.0, "grad_norm": 0.11677242070436478, "global_step": 32207, "epoch": 894, "lr": 0.001} +{"train_loss": 0.02886023186147213, "train_loss_bc": 0.02886023186147213, "train_loss_llm": 0.0, "grad_norm": 0.14561599493026733, "global_step": 32208, "epoch": 894, "lr": 0.001} +{"train_loss": 0.031481433659791946, "train_loss_bc": 0.031481433659791946, "train_loss_llm": 0.0, "grad_norm": 0.04378791153430939, "global_step": 32209, "epoch": 894, "lr": 0.001} +{"train_loss": 0.026052573695778847, "train_loss_bc": 0.026052573695778847, "train_loss_llm": 0.0, "grad_norm": 0.06239793449640274, "global_step": 32210, "epoch": 894, "lr": 0.001} +{"train_loss": 0.029527835547924042, "train_loss_bc": 0.029517503455281258, "train_loss_llm": 0.001033186912536621, "grad_norm": 0.09463466703891754, "global_step": 32211, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03181130439043045, "train_loss_bc": 0.03181130439043045, "train_loss_llm": 0.0, "grad_norm": 0.12217175960540771, "global_step": 32212, "epoch": 894, "lr": 0.001} +{"train_loss": 0.0310685932636261, "train_loss_bc": 0.0310685932636261, "train_loss_llm": 0.0, "grad_norm": 0.04127439856529236, "global_step": 32213, "epoch": 894, "lr": 0.001} +{"train_loss": 0.026669545099139214, "train_loss_bc": 0.026648085564374924, "train_loss_llm": 0.0021459460258483887, "grad_norm": 0.06444390118122101, "global_step": 32214, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03129231184720993, "train_loss_bc": 0.03129231184720993, "train_loss_llm": 0.0, "grad_norm": 0.08582302182912827, "global_step": 32215, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03027280420064926, "train_loss_bc": 0.03027280420064926, "train_loss_llm": 0.0, "grad_norm": 0.10783255100250244, "global_step": 32216, "epoch": 894, "lr": 0.001} +{"train_loss": 0.03679141029715538, "train_loss_bc": 0.03677935525774956, "train_loss_llm": 0.001205533742904663, "grad_norm": 0.04897584393620491, "global_step": 32217, "epoch": 894, "lr": 0.001} +{"train_loss": 0.02344360016286373, "train_loss_bc": 0.023440778255462646, "train_loss_llm": 0.0002822279930114746, "grad_norm": 0.07999135553836823, "global_step": 32218, "epoch": 894, "lr": 0.001} +{"train_loss": 0.029472610395815637, "train_loss_bc": 0.02733282558619976, "train_loss_llm": 0.0018776357173919678, "grad_norm": 0.11092352867126465, "global_step": 32219, "epoch": 894, "lr": 0.001, "val_loss": 0.1056443601846695} +{"train_loss": 0.03173597902059555, "train_loss_bc": 0.03173597902059555, "train_loss_llm": 0.0, "grad_norm": 0.13915805518627167, "global_step": 32220, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02524775080382824, "train_loss_bc": 0.02524775080382824, "train_loss_llm": 0.0, "grad_norm": 0.03726838901638985, "global_step": 32221, "epoch": 895, "lr": 0.001} +{"train_loss": 0.028465675190091133, "train_loss_bc": 0.0284634530544281, "train_loss_llm": 0.00022214651107788086, "grad_norm": 0.06527522206306458, "global_step": 32222, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02963682822883129, "train_loss_bc": 0.029626091942191124, "train_loss_llm": 0.0010736286640167236, "grad_norm": 0.07712625712156296, "global_step": 32223, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02706155553460121, "train_loss_bc": 0.02706155553460121, "train_loss_llm": 0.0, "grad_norm": 0.10124988853931427, "global_step": 32224, "epoch": 895, "lr": 0.001} +{"train_loss": 0.0310409814119339, "train_loss_bc": 0.0310409814119339, "train_loss_llm": 0.0, "grad_norm": 0.04407474398612976, "global_step": 32225, "epoch": 895, "lr": 0.001} +{"train_loss": 0.032238464802503586, "train_loss_bc": 0.032238464802503586, "train_loss_llm": 0.0, "grad_norm": 0.06963279843330383, "global_step": 32226, "epoch": 895, "lr": 0.001} +{"train_loss": 0.0289431381970644, "train_loss_bc": 0.028935974463820457, "train_loss_llm": 0.0007164478302001953, "grad_norm": 0.08787281066179276, "global_step": 32227, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02892671898007393, "train_loss_bc": 0.02892671898007393, "train_loss_llm": 0.0, "grad_norm": 0.0945076197385788, "global_step": 32228, "epoch": 895, "lr": 0.001} +{"train_loss": 0.029385898262262344, "train_loss_bc": 0.029367225244641304, "train_loss_llm": 0.0018673539161682129, "grad_norm": 0.0293015968054533, "global_step": 32229, "epoch": 895, "lr": 0.001} +{"train_loss": 0.031621478497982025, "train_loss_bc": 0.031606853008270264, "train_loss_llm": 0.0014626383781433105, "grad_norm": 0.04849245026707649, "global_step": 32230, "epoch": 895, "lr": 0.001} +{"train_loss": 0.028270011767745018, "train_loss_bc": 0.028270011767745018, "train_loss_llm": 0.0, "grad_norm": 0.07121601700782776, "global_step": 32231, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02524459734559059, "train_loss_bc": 0.02524459734559059, "train_loss_llm": 0.0, "grad_norm": 0.09149911254644394, "global_step": 32232, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02929125353693962, "train_loss_bc": 0.02929125353693962, "train_loss_llm": 0.0, "grad_norm": 0.03448883816599846, "global_step": 32233, "epoch": 895, "lr": 0.001} +{"train_loss": 0.025013336911797523, "train_loss_bc": 0.025013336911797523, "train_loss_llm": 0.0, "grad_norm": 0.05162544548511505, "global_step": 32234, "epoch": 895, "lr": 0.001} +{"train_loss": 0.026541592553257942, "train_loss_bc": 0.026541592553257942, "train_loss_llm": 0.0, "grad_norm": 0.07023601233959198, "global_step": 32235, "epoch": 895, "lr": 0.001} +{"train_loss": 0.027542829513549805, "train_loss_bc": 0.027528731152415276, "train_loss_llm": 0.0014098584651947021, "grad_norm": 0.09074226021766663, "global_step": 32236, "epoch": 895, "lr": 0.001} +{"train_loss": 0.023539111018180847, "train_loss_bc": 0.023539111018180847, "train_loss_llm": 0.0, "grad_norm": 0.038891103118658066, "global_step": 32237, "epoch": 895, "lr": 0.001} +{"train_loss": 0.03438376635313034, "train_loss_bc": 0.03438376635313034, "train_loss_llm": 0.0, "grad_norm": 0.0681166797876358, "global_step": 32238, "epoch": 895, "lr": 0.001} +{"train_loss": 0.03244722634553909, "train_loss_bc": 0.03244722634553909, "train_loss_llm": 0.0, "grad_norm": 0.10513294488191605, "global_step": 32239, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02298736944794655, "train_loss_bc": 0.022987104952335358, "train_loss_llm": 2.6434659957885742e-05, "grad_norm": 0.1330195516347885, "global_step": 32240, "epoch": 895, "lr": 0.001} +{"train_loss": 0.030757972970604897, "train_loss_bc": 0.030756987631320953, "train_loss_llm": 9.85860824584961e-05, "grad_norm": 0.0374181903898716, "global_step": 32241, "epoch": 895, "lr": 0.001} +{"train_loss": 0.03136317804455757, "train_loss_bc": 0.03135424107313156, "train_loss_llm": 0.0008937716484069824, "grad_norm": 0.06943028420209885, "global_step": 32242, "epoch": 895, "lr": 0.001} +{"train_loss": 0.030136996880173683, "train_loss_bc": 0.030136996880173683, "train_loss_llm": 0.0, "grad_norm": 0.08783350884914398, "global_step": 32243, "epoch": 895, "lr": 0.001} +{"train_loss": 0.03096652217209339, "train_loss_bc": 0.03096652217209339, "train_loss_llm": 0.0, "grad_norm": 0.10430990159511566, "global_step": 32244, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02613086998462677, "train_loss_bc": 0.02613086998462677, "train_loss_llm": 0.0, "grad_norm": 0.02929973416030407, "global_step": 32245, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02941128984093666, "train_loss_bc": 0.02941128984093666, "train_loss_llm": 0.0, "grad_norm": 0.04496395215392113, "global_step": 32246, "epoch": 895, "lr": 0.001} +{"train_loss": 0.028238166123628616, "train_loss_bc": 0.028238166123628616, "train_loss_llm": 0.0, "grad_norm": 0.06527326256036758, "global_step": 32247, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02946234494447708, "train_loss_bc": 0.02946234494447708, "train_loss_llm": 0.0, "grad_norm": 0.08538700640201569, "global_step": 32248, "epoch": 895, "lr": 0.001} +{"train_loss": 0.03448232263326645, "train_loss_bc": 0.03448232263326645, "train_loss_llm": 0.0, "grad_norm": 0.04128680005669594, "global_step": 32249, "epoch": 895, "lr": 0.001} +{"train_loss": 0.029851660132408142, "train_loss_bc": 0.029851660132408142, "train_loss_llm": 0.0, "grad_norm": 0.0558541938662529, "global_step": 32250, "epoch": 895, "lr": 0.001} +{"train_loss": 0.029843490570783615, "train_loss_bc": 0.029843490570783615, "train_loss_llm": 0.0, "grad_norm": 0.07850676774978638, "global_step": 32251, "epoch": 895, "lr": 0.001} +{"train_loss": 0.028386518359184265, "train_loss_bc": 0.028386518359184265, "train_loss_llm": 0.0, "grad_norm": 0.09778724610805511, "global_step": 32252, "epoch": 895, "lr": 0.001} +{"train_loss": 0.03696455433964729, "train_loss_bc": 0.03696455433964729, "train_loss_llm": 0.0, "grad_norm": 0.04853362962603569, "global_step": 32253, "epoch": 895, "lr": 0.001} +{"train_loss": 0.029557891190052032, "train_loss_bc": 0.029557891190052032, "train_loss_llm": 0.0, "grad_norm": 0.05731567367911339, "global_step": 32254, "epoch": 895, "lr": 0.001} +{"train_loss": 0.02937122279157241, "train_loss_bc": 0.0322406068444252, "train_loss_llm": 0.0004070699214935303, "grad_norm": 0.07757072150707245, "global_step": 32255, "epoch": 895, "lr": 0.001, "train/cumulative_reward": 69.04784897849007, "train/mean_score": 17.746030309293474, "train/success_rate": 0.5, "test/cumulative_reward": 31.75703095136556, "test/mean_score": 6.970425674283846, "test/success_rate": 0.18, "val_loss": 0.10520455241203308, "train_action_mse_error": 0.032099224627017975} +{"train_loss": 0.029322287067770958, "train_loss_bc": 0.029313430190086365, "train_loss_llm": 0.0008856356143951416, "grad_norm": 0.10229430347681046, "global_step": 32256, "epoch": 896, "lr": 0.001} +{"train_loss": 0.02770432084798813, "train_loss_bc": 0.027703609317541122, "train_loss_llm": 7.107853889465332e-05, "grad_norm": 0.039020806550979614, "global_step": 32257, "epoch": 896, "lr": 0.001} +{"train_loss": 0.026152823120355606, "train_loss_bc": 0.026152823120355606, "train_loss_llm": 0.0, "grad_norm": 0.05904700979590416, "global_step": 32258, "epoch": 896, "lr": 0.001} +{"train_loss": 0.028310127556324005, "train_loss_bc": 0.028299223631620407, "train_loss_llm": 0.0010904371738433838, "grad_norm": 0.0807451382279396, "global_step": 32259, "epoch": 896, "lr": 0.001} +{"train_loss": 0.034294795244932175, "train_loss_bc": 0.034294795244932175, "train_loss_llm": 0.0, "grad_norm": 0.09370081126689911, "global_step": 32260, "epoch": 896, "lr": 0.001} +{"train_loss": 0.026624232530593872, "train_loss_bc": 0.026624232530593872, "train_loss_llm": 0.0, "grad_norm": 0.033532824367284775, "global_step": 32261, "epoch": 896, "lr": 0.001} +{"train_loss": 0.026065489277243614, "train_loss_bc": 0.026065489277243614, "train_loss_llm": 0.0, "grad_norm": 0.05117466300725937, "global_step": 32262, "epoch": 896, "lr": 0.001} +{"train_loss": 0.031102880835533142, "train_loss_bc": 0.031102880835533142, "train_loss_llm": 0.0, "grad_norm": 0.0734514370560646, "global_step": 32263, "epoch": 896, "lr": 0.001} +{"train_loss": 0.02590947225689888, "train_loss_bc": 0.02590947225689888, "train_loss_llm": 0.0, "grad_norm": 0.08809582144021988, "global_step": 32264, "epoch": 896, "lr": 0.001} +{"train_loss": 0.029774287715554237, "train_loss_bc": 0.029774287715554237, "train_loss_llm": 0.0, "grad_norm": 0.0340341255068779, "global_step": 32265, "epoch": 896, "lr": 0.001} +{"train_loss": 0.02950010448694229, "train_loss_bc": 0.02950010448694229, "train_loss_llm": 0.0, "grad_norm": 0.06223512813448906, "global_step": 32266, "epoch": 896, "lr": 0.001} +{"train_loss": 0.02817157655954361, "train_loss_bc": 0.02817157655954361, "train_loss_llm": 0.0, "grad_norm": 0.08759426325559616, "global_step": 32267, "epoch": 896, "lr": 0.001} +{"train_loss": 0.029402468353509903, "train_loss_bc": 0.029402468353509903, "train_loss_llm": 0.0, "grad_norm": 0.11959120631217957, "global_step": 32268, "epoch": 896, "lr": 0.001} +{"train_loss": 0.027840344235301018, "train_loss_bc": 0.027829831466078758, "train_loss_llm": 0.0010512471199035645, "grad_norm": 0.0321352556347847, "global_step": 32269, "epoch": 896, "lr": 0.001} +{"train_loss": 0.028061693534255028, "train_loss_bc": 0.028061693534255028, "train_loss_llm": 0.0, "grad_norm": 0.05653923749923706, "global_step": 32270, "epoch": 896, "lr": 0.001} +{"train_loss": 0.03251892328262329, "train_loss_bc": 0.03251611813902855, "train_loss_llm": 0.0002804994583129883, "grad_norm": 0.06996849924325943, "global_step": 32271, "epoch": 896, "lr": 0.001} +{"train_loss": 0.032048676162958145, "train_loss_bc": 0.032048676162958145, "train_loss_llm": 0.0, "grad_norm": 0.09775076061487198, "global_step": 32272, "epoch": 896, "lr": 0.001} +{"train_loss": 0.03198668360710144, "train_loss_bc": 0.03198668360710144, "train_loss_llm": 0.0, "grad_norm": 0.040823955088853836, "global_step": 32273, "epoch": 896, "lr": 0.001} +{"train_loss": 0.027178123593330383, "train_loss_bc": 0.027178123593330383, "train_loss_llm": 0.0, "grad_norm": 0.0611000619828701, "global_step": 32274, "epoch": 896, "lr": 0.001} +{"train_loss": 0.03301703557372093, "train_loss_bc": 0.03301209583878517, "train_loss_llm": 0.0004940032958984375, "grad_norm": 0.07672769576311111, "global_step": 32275, "epoch": 896, "lr": 0.001} +{"train_loss": 0.027089791372418404, "train_loss_bc": 0.027089791372418404, "train_loss_llm": 0.0, "grad_norm": 0.0874411016702652, "global_step": 32276, "epoch": 896, "lr": 0.001} +{"train_loss": 0.03364207595586777, "train_loss_bc": 0.03363511711359024, "train_loss_llm": 0.0006959438323974609, "grad_norm": 0.04511900618672371, "global_step": 32277, "epoch": 896, "lr": 0.001} +{"train_loss": 0.033193912357091904, "train_loss_bc": 0.033193912357091904, "train_loss_llm": 0.0, "grad_norm": 0.0698520839214325, "global_step": 32278, "epoch": 896, "lr": 0.001} +{"train_loss": 0.024925261735916138, "train_loss_bc": 0.024925261735916138, "train_loss_llm": 0.0, "grad_norm": 0.09939435869455338, "global_step": 32279, "epoch": 896, "lr": 0.001} +{"train_loss": 0.02507363073527813, "train_loss_bc": 0.025073103606700897, "train_loss_llm": 5.27799129486084e-05, "grad_norm": 0.11514035612344742, "global_step": 32280, "epoch": 896, "lr": 0.001} +{"train_loss": 0.03339051082730293, "train_loss_bc": 0.03339051082730293, "train_loss_llm": 0.0, "grad_norm": 0.039796132594347, "global_step": 32281, "epoch": 896, "lr": 0.001} +{"train_loss": 0.026154272258281708, "train_loss_bc": 0.026154272258281708, "train_loss_llm": 0.0, "grad_norm": 0.05725662037730217, "global_step": 32282, "epoch": 896, "lr": 0.001} +{"train_loss": 0.028067801147699356, "train_loss_bc": 0.028067801147699356, "train_loss_llm": 0.0, "grad_norm": 0.07698683440685272, "global_step": 32283, "epoch": 896, "lr": 0.001} +{"train_loss": 0.03312096744775772, "train_loss_bc": 0.03312096744775772, "train_loss_llm": 0.0, "grad_norm": 0.09564841538667679, "global_step": 32284, "epoch": 896, "lr": 0.001} +{"train_loss": 0.02603726275265217, "train_loss_bc": 0.026018276810646057, "train_loss_llm": 0.0018986165523529053, "grad_norm": 0.036317262798547745, "global_step": 32285, "epoch": 896, "lr": 0.001} +{"train_loss": 0.030684879049658775, "train_loss_bc": 0.030674831941723824, "train_loss_llm": 0.001004636287689209, "grad_norm": 0.06131608784198761, "global_step": 32286, "epoch": 896, "lr": 0.001} +{"train_loss": 0.02812526747584343, "train_loss_bc": 0.02812526747584343, "train_loss_llm": 0.0, "grad_norm": 0.08078573644161224, "global_step": 32287, "epoch": 896, "lr": 0.001} +{"train_loss": 0.030204080045223236, "train_loss_bc": 0.030204080045223236, "train_loss_llm": 0.0, "grad_norm": 0.11460306495428085, "global_step": 32288, "epoch": 896, "lr": 0.001} +{"train_loss": 0.03336779400706291, "train_loss_bc": 0.03336779400706291, "train_loss_llm": 0.0, "grad_norm": 0.035628609359264374, "global_step": 32289, "epoch": 896, "lr": 0.001} +{"train_loss": 0.028492484241724014, "train_loss_bc": 0.028492484241724014, "train_loss_llm": 0.0, "grad_norm": 0.048916641622781754, "global_step": 32290, "epoch": 896, "lr": 0.001} +{"train_loss": 0.029271644954052236, "train_loss_bc": 0.027222881093621254, "train_loss_llm": 0.0, "grad_norm": 0.06345631182193756, "global_step": 32291, "epoch": 896, "lr": 0.001, "val_loss": 0.10339473187923431} +{"train_loss": 0.026832398027181625, "train_loss_bc": 0.026832398027181625, "train_loss_llm": 0.0, "grad_norm": 0.07676180452108383, "global_step": 32292, "epoch": 897, "lr": 0.001} +{"train_loss": 0.02480117231607437, "train_loss_bc": 0.02480117231607437, "train_loss_llm": 0.0, "grad_norm": 0.032055050134658813, "global_step": 32293, "epoch": 897, "lr": 0.001} +{"train_loss": 0.02695704996585846, "train_loss_bc": 0.02695704996585846, "train_loss_llm": 0.0, "grad_norm": 0.04739582911133766, "global_step": 32294, "epoch": 897, "lr": 0.001} +{"train_loss": 0.025229718536138535, "train_loss_bc": 0.025229718536138535, "train_loss_llm": 0.0, "grad_norm": 0.0715695470571518, "global_step": 32295, "epoch": 897, "lr": 0.001} +{"train_loss": 0.03338777273893356, "train_loss_bc": 0.03338777273893356, "train_loss_llm": 0.0, "grad_norm": 0.0977431908249855, "global_step": 32296, "epoch": 897, "lr": 0.001} +{"train_loss": 0.031931545585393906, "train_loss_bc": 0.031931545585393906, "train_loss_llm": 0.0, "grad_norm": 0.03729899600148201, "global_step": 32297, "epoch": 897, "lr": 0.001} +{"train_loss": 0.02439970336854458, "train_loss_bc": 0.02439970336854458, "train_loss_llm": 0.0, "grad_norm": 0.055856067687273026, "global_step": 32298, "epoch": 897, "lr": 0.001} +{"train_loss": 0.02509172074496746, "train_loss_bc": 0.02509172074496746, "train_loss_llm": 0.0, "grad_norm": 0.08250643312931061, "global_step": 32299, "epoch": 897, "lr": 0.001} +{"train_loss": 0.03342657536268234, "train_loss_bc": 0.03342455253005028, "train_loss_llm": 0.0002022087574005127, "grad_norm": 0.10641486942768097, "global_step": 32300, "epoch": 897, "lr": 0.001} +{"train_loss": 0.03494282811880112, "train_loss_bc": 0.03494282811880112, "train_loss_llm": 0.0, "grad_norm": 0.04169609397649765, "global_step": 32301, "epoch": 897, "lr": 0.001} +{"train_loss": 0.03096606768667698, "train_loss_bc": 0.03096606768667698, "train_loss_llm": 0.0, "grad_norm": 0.07370937615633011, "global_step": 32302, "epoch": 897, "lr": 0.001} +{"train_loss": 0.027312729507684708, "train_loss_bc": 0.027312729507684708, "train_loss_llm": 0.0, "grad_norm": 0.10630673915147781, "global_step": 32303, "epoch": 897, "lr": 0.001} +{"train_loss": 0.023808393627405167, "train_loss_bc": 0.023808393627405167, "train_loss_llm": 0.0, "grad_norm": 0.11862596869468689, "global_step": 32304, "epoch": 897, "lr": 0.001} +{"train_loss": 0.032887790352106094, "train_loss_bc": 0.03287002444267273, "train_loss_llm": 0.001776576042175293, "grad_norm": 0.04475899040699005, "global_step": 32305, "epoch": 897, "lr": 0.001} +{"train_loss": 0.029578765854239464, "train_loss_bc": 0.029578765854239464, "train_loss_llm": 0.0, "grad_norm": 0.06127413362264633, "global_step": 32306, "epoch": 897, "lr": 0.001} +{"train_loss": 0.0319867879152298, "train_loss_bc": 0.031986355781555176, "train_loss_llm": 4.32431697845459e-05, "grad_norm": 0.08165019005537033, "global_step": 32307, "epoch": 897, "lr": 0.001} +{"train_loss": 0.024873265996575356, "train_loss_bc": 0.024873265996575356, "train_loss_llm": 0.0, "grad_norm": 0.0943552553653717, "global_step": 32308, "epoch": 897, "lr": 0.001} +{"train_loss": 0.029303237795829773, "train_loss_bc": 0.029303237795829773, "train_loss_llm": 0.0, "grad_norm": 0.036554448306560516, "global_step": 32309, "epoch": 897, "lr": 0.001} +{"train_loss": 0.03158797696232796, "train_loss_bc": 0.03158797696232796, "train_loss_llm": 0.0, "grad_norm": 0.054132428020238876, "global_step": 32310, "epoch": 897, "lr": 0.001} +{"train_loss": 0.027577467262744904, "train_loss_bc": 0.027576033025979996, "train_loss_llm": 0.00014343857765197754, "grad_norm": 0.06931400299072266, "global_step": 32311, "epoch": 897, "lr": 0.001} +{"train_loss": 0.032568566501140594, "train_loss_bc": 0.03255819156765938, "train_loss_llm": 0.0010373890399932861, "grad_norm": 0.09551751613616943, "global_step": 32312, "epoch": 897, "lr": 0.001} +{"train_loss": 0.025144454091787338, "train_loss_bc": 0.025144454091787338, "train_loss_llm": 0.0, "grad_norm": 0.04268129914999008, "global_step": 32313, "epoch": 897, "lr": 0.001} +{"train_loss": 0.027249524369835854, "train_loss_bc": 0.027249524369835854, "train_loss_llm": 0.0, "grad_norm": 0.05995430424809456, "global_step": 32314, "epoch": 897, "lr": 0.001} +{"train_loss": 0.028416989371180534, "train_loss_bc": 0.028416989371180534, "train_loss_llm": 0.0, "grad_norm": 0.08707961440086365, "global_step": 32315, "epoch": 897, "lr": 0.001} +{"train_loss": 0.02847384847700596, "train_loss_bc": 0.02847384847700596, "train_loss_llm": 0.0, "grad_norm": 0.10901124030351639, "global_step": 32316, "epoch": 897, "lr": 0.001} +{"train_loss": 0.031953174620866776, "train_loss_bc": 0.031953174620866776, "train_loss_llm": 0.0, "grad_norm": 0.03399943932890892, "global_step": 32317, "epoch": 897, "lr": 0.001} +{"train_loss": 0.02980153262615204, "train_loss_bc": 0.02980153262615204, "train_loss_llm": 0.0, "grad_norm": 0.0547097846865654, "global_step": 32318, "epoch": 897, "lr": 0.001} +{"train_loss": 0.0309393722563982, "train_loss_bc": 0.0309393722563982, "train_loss_llm": 0.0, "grad_norm": 0.08279217779636383, "global_step": 32319, "epoch": 897, "lr": 0.001} +{"train_loss": 0.03631584718823433, "train_loss_bc": 0.036308154463768005, "train_loss_llm": 0.0007692873477935791, "grad_norm": 0.10400306433439255, "global_step": 32320, "epoch": 897, "lr": 0.001} +{"train_loss": 0.03580894693732262, "train_loss_bc": 0.03579927608370781, "train_loss_llm": 0.0009672045707702637, "grad_norm": 0.039163101464509964, "global_step": 32321, "epoch": 897, "lr": 0.001} +{"train_loss": 0.02736026607453823, "train_loss_bc": 0.02736026607453823, "train_loss_llm": 0.0, "grad_norm": 0.059735264629125595, "global_step": 32322, "epoch": 897, "lr": 0.001} +{"train_loss": 0.0313827283680439, "train_loss_bc": 0.0313827283680439, "train_loss_llm": 0.0, "grad_norm": 0.08260414749383926, "global_step": 32323, "epoch": 897, "lr": 0.001} +{"train_loss": 0.02749156951904297, "train_loss_bc": 0.02749156951904297, "train_loss_llm": 0.0, "grad_norm": 0.09766261279582977, "global_step": 32324, "epoch": 897, "lr": 0.001} +{"train_loss": 0.02463083527982235, "train_loss_bc": 0.02462061122059822, "train_loss_llm": 0.001022428274154663, "grad_norm": 0.03845331817865372, "global_step": 32325, "epoch": 897, "lr": 0.001} +{"train_loss": 0.032616909593343735, "train_loss_bc": 0.032616909593343735, "train_loss_llm": 0.0, "grad_norm": 0.06405095010995865, "global_step": 32326, "epoch": 897, "lr": 0.001} +{"train_loss": 0.0292709577932126, "train_loss_bc": 0.026716947555541992, "train_loss_llm": 0.0, "grad_norm": 0.08221153169870377, "global_step": 32327, "epoch": 897, "lr": 0.001, "val_loss": 0.10437573492527008} +{"train_loss": 0.03179942071437836, "train_loss_bc": 0.03179942071437836, "train_loss_llm": 0.0, "grad_norm": 0.10755156725645065, "global_step": 32328, "epoch": 898, "lr": 0.001} +{"train_loss": 0.025004897266626358, "train_loss_bc": 0.025004897266626358, "train_loss_llm": 0.0, "grad_norm": 0.03911714628338814, "global_step": 32329, "epoch": 898, "lr": 0.001} +{"train_loss": 0.030562521889805794, "train_loss_bc": 0.03054417483508587, "train_loss_llm": 0.0018346607685089111, "grad_norm": 0.06488417088985443, "global_step": 32330, "epoch": 898, "lr": 0.001} +{"train_loss": 0.028039969503879547, "train_loss_bc": 0.028039969503879547, "train_loss_llm": 0.0, "grad_norm": 0.09105635434389114, "global_step": 32331, "epoch": 898, "lr": 0.001} +{"train_loss": 0.028610076755285263, "train_loss_bc": 0.028610076755285263, "train_loss_llm": 0.0, "grad_norm": 0.11842750757932663, "global_step": 32332, "epoch": 898, "lr": 0.001} +{"train_loss": 0.026239829137921333, "train_loss_bc": 0.026239829137921333, "train_loss_llm": 0.0, "grad_norm": 0.030184486880898476, "global_step": 32333, "epoch": 898, "lr": 0.001} +{"train_loss": 0.02545071206986904, "train_loss_bc": 0.02545071206986904, "train_loss_llm": 0.0, "grad_norm": 0.05036121606826782, "global_step": 32334, "epoch": 898, "lr": 0.001} +{"train_loss": 0.025655211880803108, "train_loss_bc": 0.025655211880803108, "train_loss_llm": 0.0, "grad_norm": 0.06043345108628273, "global_step": 32335, "epoch": 898, "lr": 0.001} +{"train_loss": 0.03445987030863762, "train_loss_bc": 0.03445987030863762, "train_loss_llm": 0.0, "grad_norm": 0.09047162532806396, "global_step": 32336, "epoch": 898, "lr": 0.001} +{"train_loss": 0.026549555361270905, "train_loss_bc": 0.026549555361270905, "train_loss_llm": 0.0, "grad_norm": 0.02869039960205555, "global_step": 32337, "epoch": 898, "lr": 0.001} +{"train_loss": 0.031032640486955643, "train_loss_bc": 0.03102761320769787, "train_loss_llm": 0.0005027949810028076, "grad_norm": 0.04548269882798195, "global_step": 32338, "epoch": 898, "lr": 0.001} +{"train_loss": 0.028682202100753784, "train_loss_bc": 0.028682202100753784, "train_loss_llm": 0.0, "grad_norm": 0.062207892537117004, "global_step": 32339, "epoch": 898, "lr": 0.001} +{"train_loss": 0.03126414865255356, "train_loss_bc": 0.031245537102222443, "train_loss_llm": 0.0018612444400787354, "grad_norm": 0.07836192101240158, "global_step": 32340, "epoch": 898, "lr": 0.001} +{"train_loss": 0.03456906974315643, "train_loss_bc": 0.03456906974315643, "train_loss_llm": 0.0, "grad_norm": 0.04392631724476814, "global_step": 32341, "epoch": 898, "lr": 0.001} +{"train_loss": 0.03334122896194458, "train_loss_bc": 0.03334122896194458, "train_loss_llm": 0.0, "grad_norm": 0.05966990441083908, "global_step": 32342, "epoch": 898, "lr": 0.001} +{"train_loss": 0.028935778886079788, "train_loss_bc": 0.028935778886079788, "train_loss_llm": 0.0, "grad_norm": 0.07013954222202301, "global_step": 32343, "epoch": 898, "lr": 0.001} +{"train_loss": 0.02771463803946972, "train_loss_bc": 0.02769860066473484, "train_loss_llm": 0.0016036629676818848, "grad_norm": 0.09001372009515762, "global_step": 32344, "epoch": 898, "lr": 0.001} +{"train_loss": 0.03324897959828377, "train_loss_bc": 0.03324897959828377, "train_loss_llm": 0.0, "grad_norm": 0.03373359516263008, "global_step": 32345, "epoch": 898, "lr": 0.001} +{"train_loss": 0.027364052832126617, "train_loss_bc": 0.027364052832126617, "train_loss_llm": 0.0, "grad_norm": 0.05461196228861809, "global_step": 32346, "epoch": 898, "lr": 0.001} +{"train_loss": 0.02749490737915039, "train_loss_bc": 0.02749253436923027, "train_loss_llm": 0.00023725628852844238, "grad_norm": 0.08053033798933029, "global_step": 32347, "epoch": 898, "lr": 0.001} +{"train_loss": 0.02544272318482399, "train_loss_bc": 0.02544272318482399, "train_loss_llm": 0.0, "grad_norm": 0.09353123605251312, "global_step": 32348, "epoch": 898, "lr": 0.001} +{"train_loss": 0.0235478263348341, "train_loss_bc": 0.0235478263348341, "train_loss_llm": 0.0, "grad_norm": 0.03168591111898422, "global_step": 32349, "epoch": 898, "lr": 0.001} +{"train_loss": 0.030683204531669617, "train_loss_bc": 0.030683204531669617, "train_loss_llm": 0.0, "grad_norm": 0.0596570186316967, "global_step": 32350, "epoch": 898, "lr": 0.001} +{"train_loss": 0.03629365563392639, "train_loss_bc": 0.03628285229206085, "train_loss_llm": 0.0010802149772644043, "grad_norm": 0.0778534859418869, "global_step": 32351, "epoch": 898, "lr": 0.001} +{"train_loss": 0.030218064785003662, "train_loss_bc": 0.030218064785003662, "train_loss_llm": 0.0, "grad_norm": 0.09185721725225449, "global_step": 32352, "epoch": 898, "lr": 0.001} +{"train_loss": 0.020621713250875473, "train_loss_bc": 0.020621713250875473, "train_loss_llm": 0.0, "grad_norm": 0.0290995966643095, "global_step": 32353, "epoch": 898, "lr": 0.001} +{"train_loss": 0.029646052047610283, "train_loss_bc": 0.029641784727573395, "train_loss_llm": 0.0004267096519470215, "grad_norm": 0.04676371440291405, "global_step": 32354, "epoch": 898, "lr": 0.001} +{"train_loss": 0.033142827451229095, "train_loss_bc": 0.033128105103969574, "train_loss_llm": 0.0014721155166625977, "grad_norm": 0.0673699676990509, "global_step": 32355, "epoch": 898, "lr": 0.001} +{"train_loss": 0.03492925316095352, "train_loss_bc": 0.03491732478141785, "train_loss_llm": 0.001192927360534668, "grad_norm": 0.08577068150043488, "global_step": 32356, "epoch": 898, "lr": 0.001} +{"train_loss": 0.027685940265655518, "train_loss_bc": 0.02768264338374138, "train_loss_llm": 0.00032970309257507324, "grad_norm": 0.0339244082570076, "global_step": 32357, "epoch": 898, "lr": 0.001} +{"train_loss": 0.029894648119807243, "train_loss_bc": 0.029894648119807243, "train_loss_llm": 0.0, "grad_norm": 0.06829870492219925, "global_step": 32358, "epoch": 898, "lr": 0.001} +{"train_loss": 0.0293649323284626, "train_loss_bc": 0.0293649323284626, "train_loss_llm": 0.0, "grad_norm": 0.08675246685743332, "global_step": 32359, "epoch": 898, "lr": 0.001} +{"train_loss": 0.029736921191215515, "train_loss_bc": 0.029736921191215515, "train_loss_llm": 0.0, "grad_norm": 0.11081956326961517, "global_step": 32360, "epoch": 898, "lr": 0.001} +{"train_loss": 0.029515156522393227, "train_loss_bc": 0.029515156522393227, "train_loss_llm": 0.0, "grad_norm": 0.03401714935898781, "global_step": 32361, "epoch": 898, "lr": 0.001} +{"train_loss": 0.027593117207288742, "train_loss_bc": 0.027593117207288742, "train_loss_llm": 0.0, "grad_norm": 0.059532634913921356, "global_step": 32362, "epoch": 898, "lr": 0.001} +{"train_loss": 0.02921211015846994, "train_loss_bc": 0.027300218120217323, "train_loss_llm": 0.0, "grad_norm": 0.08556359261274338, "global_step": 32363, "epoch": 898, "lr": 0.001, "val_loss": 0.10236135870218277} +{"train_loss": 0.031101979315280914, "train_loss_bc": 0.031101979315280914, "train_loss_llm": 0.0, "grad_norm": 0.10597311705350876, "global_step": 32364, "epoch": 899, "lr": 0.001} +{"train_loss": 0.02993960492312908, "train_loss_bc": 0.02993960492312908, "train_loss_llm": 0.0, "grad_norm": 0.026863988488912582, "global_step": 32365, "epoch": 899, "lr": 0.001} +{"train_loss": 0.03128771483898163, "train_loss_bc": 0.03128771483898163, "train_loss_llm": 0.0, "grad_norm": 0.061932820826768875, "global_step": 32366, "epoch": 899, "lr": 0.001} +{"train_loss": 0.027490176260471344, "train_loss_bc": 0.027490176260471344, "train_loss_llm": 0.0, "grad_norm": 0.07203882932662964, "global_step": 32367, "epoch": 899, "lr": 0.001} +{"train_loss": 0.03079313598573208, "train_loss_bc": 0.03079313598573208, "train_loss_llm": 0.0, "grad_norm": 0.0758509486913681, "global_step": 32368, "epoch": 899, "lr": 0.001} +{"train_loss": 0.02826392836868763, "train_loss_bc": 0.028240764513611794, "train_loss_llm": 0.0023164451122283936, "grad_norm": 0.034453198313713074, "global_step": 32369, "epoch": 899, "lr": 0.001} +{"train_loss": 0.031683649867773056, "train_loss_bc": 0.031683649867773056, "train_loss_llm": 0.0, "grad_norm": 0.05623351410031319, "global_step": 32370, "epoch": 899, "lr": 0.001} +{"train_loss": 0.029597995802760124, "train_loss_bc": 0.029576141387224197, "train_loss_llm": 0.0021854639053344727, "grad_norm": 0.08352290093898773, "global_step": 32371, "epoch": 899, "lr": 0.001} +{"train_loss": 0.028848618268966675, "train_loss_bc": 0.028828207403421402, "train_loss_llm": 0.002041071653366089, "grad_norm": 0.10363724827766418, "global_step": 32372, "epoch": 899, "lr": 0.001} +{"train_loss": 0.03119092993438244, "train_loss_bc": 0.03119092993438244, "train_loss_llm": 0.0, "grad_norm": 0.04272313416004181, "global_step": 32373, "epoch": 899, "lr": 0.001} +{"train_loss": 0.02464945986866951, "train_loss_bc": 0.02463468909263611, "train_loss_llm": 0.0014771521091461182, "grad_norm": 0.057456180453300476, "global_step": 32374, "epoch": 899, "lr": 0.001} +{"train_loss": 0.024964220821857452, "train_loss_bc": 0.024964220821857452, "train_loss_llm": 0.0, "grad_norm": 0.06828083097934723, "global_step": 32375, "epoch": 899, "lr": 0.001} +{"train_loss": 0.026741590350866318, "train_loss_bc": 0.026730259880423546, "train_loss_llm": 0.001132965087890625, "grad_norm": 0.08814211189746857, "global_step": 32376, "epoch": 899, "lr": 0.001} +{"train_loss": 0.02512742578983307, "train_loss_bc": 0.02512742578983307, "train_loss_llm": 0.0, "grad_norm": 0.044793419539928436, "global_step": 32377, "epoch": 899, "lr": 0.001} +{"train_loss": 0.028272973373532295, "train_loss_bc": 0.028267886489629745, "train_loss_llm": 0.0005087554454803467, "grad_norm": 0.0744660496711731, "global_step": 32378, "epoch": 899, "lr": 0.001} +{"train_loss": 0.02415168099105358, "train_loss_bc": 0.02415168099105358, "train_loss_llm": 0.0, "grad_norm": 0.10261867940425873, "global_step": 32379, "epoch": 899, "lr": 0.001} +{"train_loss": 0.034388598054647446, "train_loss_bc": 0.034388598054647446, "train_loss_llm": 0.0, "grad_norm": 0.13580523431301117, "global_step": 32380, "epoch": 899, "lr": 0.001} +{"train_loss": 0.029468132182955742, "train_loss_bc": 0.029458995908498764, "train_loss_llm": 0.0009135603904724121, "grad_norm": 0.03595472127199173, "global_step": 32381, "epoch": 899, "lr": 0.001} +{"train_loss": 0.023430360481142998, "train_loss_bc": 0.023430360481142998, "train_loss_llm": 0.0, "grad_norm": 0.05047588422894478, "global_step": 32382, "epoch": 899, "lr": 0.001} +{"train_loss": 0.02861500345170498, "train_loss_bc": 0.02861500345170498, "train_loss_llm": 0.0, "grad_norm": 0.0735793337225914, "global_step": 32383, "epoch": 899, "lr": 0.001} +{"train_loss": 0.03385603055357933, "train_loss_bc": 0.03385603055357933, "train_loss_llm": 0.0, "grad_norm": 0.10374631732702255, "global_step": 32384, "epoch": 899, "lr": 0.001} +{"train_loss": 0.0325625017285347, "train_loss_bc": 0.0325625017285347, "train_loss_llm": 0.0, "grad_norm": 0.03797038644552231, "global_step": 32385, "epoch": 899, "lr": 0.001} +{"train_loss": 0.035285696387290955, "train_loss_bc": 0.035285696387290955, "train_loss_llm": 0.0, "grad_norm": 0.05632040277123451, "global_step": 32386, "epoch": 899, "lr": 0.001} +{"train_loss": 0.032414864748716354, "train_loss_bc": 0.032414864748716354, "train_loss_llm": 0.0, "grad_norm": 0.07740882784128189, "global_step": 32387, "epoch": 899, "lr": 0.001} +{"train_loss": 0.028633449226617813, "train_loss_bc": 0.028627678751945496, "train_loss_llm": 0.0005769729614257812, "grad_norm": 0.09670979529619217, "global_step": 32388, "epoch": 899, "lr": 0.001} +{"train_loss": 0.0285419300198555, "train_loss_bc": 0.0285419300198555, "train_loss_llm": 0.0, "grad_norm": 0.031670957803726196, "global_step": 32389, "epoch": 899, "lr": 0.001} +{"train_loss": 0.027775712311267853, "train_loss_bc": 0.027773570269346237, "train_loss_llm": 0.0002142488956451416, "grad_norm": 0.05911384895443916, "global_step": 32390, "epoch": 899, "lr": 0.001} +{"train_loss": 0.02542390115559101, "train_loss_bc": 0.0254118200391531, "train_loss_llm": 0.0012080967426300049, "grad_norm": 0.06937780976295471, "global_step": 32391, "epoch": 899, "lr": 0.001} +{"train_loss": 0.030852481722831726, "train_loss_bc": 0.030852481722831726, "train_loss_llm": 0.0, "grad_norm": 0.09424169361591339, "global_step": 32392, "epoch": 899, "lr": 0.001} +{"train_loss": 0.02653888612985611, "train_loss_bc": 0.026534870266914368, "train_loss_llm": 0.00040149688720703125, "grad_norm": 0.04172736033797264, "global_step": 32393, "epoch": 899, "lr": 0.001} +{"train_loss": 0.03049500659108162, "train_loss_bc": 0.03049500659108162, "train_loss_llm": 0.0, "grad_norm": 0.06469786912202835, "global_step": 32394, "epoch": 899, "lr": 0.001} +{"train_loss": 0.02717159502208233, "train_loss_bc": 0.02717159502208233, "train_loss_llm": 0.0, "grad_norm": 0.0886135920882225, "global_step": 32395, "epoch": 899, "lr": 0.001} +{"train_loss": 0.025969933718442917, "train_loss_bc": 0.025967463850975037, "train_loss_llm": 0.00024694204330444336, "grad_norm": 0.11199106276035309, "global_step": 32396, "epoch": 899, "lr": 0.001} +{"train_loss": 0.034967441111803055, "train_loss_bc": 0.034967441111803055, "train_loss_llm": 0.0, "grad_norm": 0.0363624282181263, "global_step": 32397, "epoch": 899, "lr": 0.001} +{"train_loss": 0.03090064600110054, "train_loss_bc": 0.03090064600110054, "train_loss_llm": 0.0, "grad_norm": 0.06252318620681763, "global_step": 32398, "epoch": 899, "lr": 0.001} +{"train_loss": 0.029281519042948883, "train_loss_bc": 0.03273431956768036, "train_loss_llm": 0.00031110644340515137, "grad_norm": 0.07968590408563614, "global_step": 32399, "epoch": 899, "lr": 0.001, "val_loss": 0.10381074249744415} +{"train_loss": 0.034782975912094116, "train_loss_bc": 0.034782975912094116, "train_loss_llm": 0.0, "grad_norm": 0.10480136424303055, "global_step": 32400, "epoch": 900, "lr": 0.001} +{"train_loss": 0.034975532442331314, "train_loss_bc": 0.03494792431592941, "train_loss_llm": 0.0027607083320617676, "grad_norm": 0.04974761977791786, "global_step": 32401, "epoch": 900, "lr": 0.001} +{"train_loss": 0.027811337262392044, "train_loss_bc": 0.027811337262392044, "train_loss_llm": 0.0, "grad_norm": 0.06989340484142303, "global_step": 32402, "epoch": 900, "lr": 0.001} +{"train_loss": 0.029940076172351837, "train_loss_bc": 0.029931524768471718, "train_loss_llm": 0.0008551478385925293, "grad_norm": 0.09295273572206497, "global_step": 32403, "epoch": 900, "lr": 0.001} +{"train_loss": 0.030797403305768967, "train_loss_bc": 0.030797403305768967, "train_loss_llm": 0.0, "grad_norm": 0.11164343357086182, "global_step": 32404, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03230611979961395, "train_loss_bc": 0.03230611979961395, "train_loss_llm": 0.0, "grad_norm": 0.037215590476989746, "global_step": 32405, "epoch": 900, "lr": 0.001} +{"train_loss": 0.030072413384914398, "train_loss_bc": 0.030072413384914398, "train_loss_llm": 0.0, "grad_norm": 0.0680980384349823, "global_step": 32406, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03178395330905914, "train_loss_bc": 0.03178395330905914, "train_loss_llm": 0.0, "grad_norm": 0.08066160976886749, "global_step": 32407, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03420555591583252, "train_loss_bc": 0.03420555591583252, "train_loss_llm": 0.0, "grad_norm": 0.09694783389568329, "global_step": 32408, "epoch": 900, "lr": 0.001} +{"train_loss": 0.027795976027846336, "train_loss_bc": 0.02779301069676876, "train_loss_llm": 0.00029662251472473145, "grad_norm": 0.03443167731165886, "global_step": 32409, "epoch": 900, "lr": 0.001} +{"train_loss": 0.02569308504462242, "train_loss_bc": 0.02569308504462242, "train_loss_llm": 0.0, "grad_norm": 0.049837011843919754, "global_step": 32410, "epoch": 900, "lr": 0.001} +{"train_loss": 0.025987161323428154, "train_loss_bc": 0.02597750909626484, "train_loss_llm": 0.0009651780128479004, "grad_norm": 0.06696219742298126, "global_step": 32411, "epoch": 900, "lr": 0.001} +{"train_loss": 0.02592051587998867, "train_loss_bc": 0.02592051587998867, "train_loss_llm": 0.0, "grad_norm": 0.08638178557157516, "global_step": 32412, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03244168311357498, "train_loss_bc": 0.03244168311357498, "train_loss_llm": 0.0, "grad_norm": 0.04969153553247452, "global_step": 32413, "epoch": 900, "lr": 0.001} +{"train_loss": 0.02805270068347454, "train_loss_bc": 0.02804667130112648, "train_loss_llm": 0.0006029605865478516, "grad_norm": 0.07291275262832642, "global_step": 32414, "epoch": 900, "lr": 0.001} +{"train_loss": 0.028916725888848305, "train_loss_bc": 0.028916725888848305, "train_loss_llm": 0.0, "grad_norm": 0.10943523794412613, "global_step": 32415, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03349517285823822, "train_loss_bc": 0.03349517285823822, "train_loss_llm": 0.0, "grad_norm": 0.13922937214374542, "global_step": 32416, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03649510070681572, "train_loss_bc": 0.03649510070681572, "train_loss_llm": 0.0, "grad_norm": 0.04919901117682457, "global_step": 32417, "epoch": 900, "lr": 0.001} +{"train_loss": 0.026895780116319656, "train_loss_bc": 0.026895780116319656, "train_loss_llm": 0.0, "grad_norm": 0.05985986813902855, "global_step": 32418, "epoch": 900, "lr": 0.001} +{"train_loss": 0.031052494421601295, "train_loss_bc": 0.031052494421601295, "train_loss_llm": 0.0, "grad_norm": 0.08944329619407654, "global_step": 32419, "epoch": 900, "lr": 0.001} +{"train_loss": 0.025843678042292595, "train_loss_bc": 0.025843678042292595, "train_loss_llm": 0.0, "grad_norm": 0.11086193472146988, "global_step": 32420, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03151853010058403, "train_loss_bc": 0.03151853010058403, "train_loss_llm": 0.0, "grad_norm": 0.04488394036889076, "global_step": 32421, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03305693343281746, "train_loss_bc": 0.03304031491279602, "train_loss_llm": 0.001661747694015503, "grad_norm": 0.06470474600791931, "global_step": 32422, "epoch": 900, "lr": 0.001} +{"train_loss": 0.02954527735710144, "train_loss_bc": 0.02954527735710144, "train_loss_llm": 0.0, "grad_norm": 0.08831390738487244, "global_step": 32423, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03088688664138317, "train_loss_bc": 0.03088688664138317, "train_loss_llm": 0.0, "grad_norm": 0.11832929402589798, "global_step": 32424, "epoch": 900, "lr": 0.001} +{"train_loss": 0.02364964783191681, "train_loss_bc": 0.02363542467355728, "train_loss_llm": 0.0014222264289855957, "grad_norm": 0.039396192878484726, "global_step": 32425, "epoch": 900, "lr": 0.001} +{"train_loss": 0.028717201203107834, "train_loss_bc": 0.028717201203107834, "train_loss_llm": 0.0, "grad_norm": 0.07234479486942291, "global_step": 32426, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03055676631629467, "train_loss_bc": 0.03054950013756752, "train_loss_llm": 0.0007266402244567871, "grad_norm": 0.09103529155254364, "global_step": 32427, "epoch": 900, "lr": 0.001} +{"train_loss": 0.026136113330721855, "train_loss_bc": 0.026136113330721855, "train_loss_llm": 0.0, "grad_norm": 0.11195248365402222, "global_step": 32428, "epoch": 900, "lr": 0.001} +{"train_loss": 0.023359550163149834, "train_loss_bc": 0.023356709629297256, "train_loss_llm": 0.00028398633003234863, "grad_norm": 0.031276971101760864, "global_step": 32429, "epoch": 900, "lr": 0.001} +{"train_loss": 0.026205306872725487, "train_loss_bc": 0.026205306872725487, "train_loss_llm": 0.0, "grad_norm": 0.05621112510561943, "global_step": 32430, "epoch": 900, "lr": 0.001} +{"train_loss": 0.031279899179935455, "train_loss_bc": 0.031279899179935455, "train_loss_llm": 0.0, "grad_norm": 0.07524357736110687, "global_step": 32431, "epoch": 900, "lr": 0.001} +{"train_loss": 0.025029053911566734, "train_loss_bc": 0.025027912110090256, "train_loss_llm": 0.00011414289474487305, "grad_norm": 0.08821474760770798, "global_step": 32432, "epoch": 900, "lr": 0.001} +{"train_loss": 0.03070944920182228, "train_loss_bc": 0.03070944920182228, "train_loss_llm": 0.0, "grad_norm": 0.03653079643845558, "global_step": 32433, "epoch": 900, "lr": 0.001} +{"train_loss": 0.024285467341542244, "train_loss_bc": 0.024285467341542244, "train_loss_llm": 0.0, "grad_norm": 0.05975251644849777, "global_step": 32434, "epoch": 900, "lr": 0.001} +{"train_loss": 0.029227802633411355, "train_loss_bc": 0.021995652467012405, "train_loss_llm": 0.00037175416946411133, "grad_norm": 0.07359723001718521, "global_step": 32435, "epoch": 900, "lr": 0.001, "train/cumulative_reward": 75.53918622316262, "train/mean_score": 6.4550071566989375, "train/success_rate": 0.1, "test/cumulative_reward": 24.386765315623087, "test/mean_score": 8.392857800427024, "test/success_rate": 0.24, "val_loss": 0.10253472626209259, "train_action_mse_error": 0.03328002616763115} +{"train_loss": 0.0247991681098938, "train_loss_bc": 0.0247991681098938, "train_loss_llm": 0.0, "grad_norm": 0.0874032974243164, "global_step": 32436, "epoch": 901, "lr": 0.001} +{"train_loss": 0.03573219105601311, "train_loss_bc": 0.03573219105601311, "train_loss_llm": 0.0, "grad_norm": 0.05569039657711983, "global_step": 32437, "epoch": 901, "lr": 0.001} +{"train_loss": 0.028930455446243286, "train_loss_bc": 0.028930455446243286, "train_loss_llm": 0.0, "grad_norm": 0.09513271600008011, "global_step": 32438, "epoch": 901, "lr": 0.001} +{"train_loss": 0.028605643659830093, "train_loss_bc": 0.028605643659830093, "train_loss_llm": 0.0, "grad_norm": 0.13845627009868622, "global_step": 32439, "epoch": 901, "lr": 0.001} +{"train_loss": 0.024362917989492416, "train_loss_bc": 0.02435767836868763, "train_loss_llm": 0.0005238950252532959, "grad_norm": 0.17025402188301086, "global_step": 32440, "epoch": 901, "lr": 0.001} +{"train_loss": 0.031369127333164215, "train_loss_bc": 0.03136097639799118, "train_loss_llm": 0.0008150041103363037, "grad_norm": 0.04805414378643036, "global_step": 32441, "epoch": 901, "lr": 0.001} +{"train_loss": 0.02782687358558178, "train_loss_bc": 0.027815822511911392, "train_loss_llm": 0.0011050701141357422, "grad_norm": 0.07655399292707443, "global_step": 32442, "epoch": 901, "lr": 0.001} +{"train_loss": 0.02394503355026245, "train_loss_bc": 0.02394503355026245, "train_loss_llm": 0.0, "grad_norm": 0.09474600106477737, "global_step": 32443, "epoch": 901, "lr": 0.001} +{"train_loss": 0.03043530136346817, "train_loss_bc": 0.03043530136346817, "train_loss_llm": 0.0, "grad_norm": 0.1091732531785965, "global_step": 32444, "epoch": 901, "lr": 0.001} +{"train_loss": 0.028734328225255013, "train_loss_bc": 0.028734328225255013, "train_loss_llm": 0.0, "grad_norm": 0.04419597610831261, "global_step": 32445, "epoch": 901, "lr": 0.001} +{"train_loss": 0.02988649718463421, "train_loss_bc": 0.029865900054574013, "train_loss_llm": 0.002059757709503174, "grad_norm": 0.06509605050086975, "global_step": 32446, "epoch": 901, "lr": 0.001} +{"train_loss": 0.03168129175901413, "train_loss_bc": 0.03168129175901413, "train_loss_llm": 0.0, "grad_norm": 0.08717035502195358, "global_step": 32447, "epoch": 901, "lr": 0.001} +{"train_loss": 0.03343287482857704, "train_loss_bc": 0.033427074551582336, "train_loss_llm": 0.0005799531936645508, "grad_norm": 0.11663663387298584, "global_step": 32448, "epoch": 901, "lr": 0.001} +{"train_loss": 0.025656020268797874, "train_loss_bc": 0.02564769797027111, "train_loss_llm": 0.0008322596549987793, "grad_norm": 0.03275607153773308, "global_step": 32449, "epoch": 901, "lr": 0.001} +{"train_loss": 0.0297692958265543, "train_loss_bc": 0.0297692958265543, "train_loss_llm": 0.0, "grad_norm": 0.05684651806950569, "global_step": 32450, "epoch": 901, "lr": 0.001} +{"train_loss": 0.03305339813232422, "train_loss_bc": 0.03305141627788544, "train_loss_llm": 0.0001983344554901123, "grad_norm": 0.08085981011390686, "global_step": 32451, "epoch": 901, "lr": 0.001} +{"train_loss": 0.03184553608298302, "train_loss_bc": 0.03184553608298302, "train_loss_llm": 0.0, "grad_norm": 0.10431636869907379, "global_step": 32452, "epoch": 901, "lr": 0.001} +{"train_loss": 0.03231450915336609, "train_loss_bc": 0.032308515161275864, "train_loss_llm": 0.0005994439125061035, "grad_norm": 0.04046441987156868, "global_step": 32453, "epoch": 901, "lr": 0.001} +{"train_loss": 0.035492412745952606, "train_loss_bc": 0.035492412745952606, "train_loss_llm": 0.0, "grad_norm": 0.07379928231239319, "global_step": 32454, "epoch": 901, "lr": 0.001} +{"train_loss": 0.023137914016842842, "train_loss_bc": 0.023135235533118248, "train_loss_llm": 0.00026786327362060547, "grad_norm": 0.08822459727525711, "global_step": 32455, "epoch": 901, "lr": 0.001} +{"train_loss": 0.032836757600307465, "train_loss_bc": 0.032836757600307465, "train_loss_llm": 0.0, "grad_norm": 0.11255653202533722, "global_step": 32456, "epoch": 901, "lr": 0.001} +{"train_loss": 0.0300376508384943, "train_loss_bc": 0.0300376508384943, "train_loss_llm": 0.0, "grad_norm": 0.05710626393556595, "global_step": 32457, "epoch": 901, "lr": 0.001} +{"train_loss": 0.03498289734125137, "train_loss_bc": 0.03498289734125137, "train_loss_llm": 0.0, "grad_norm": 0.0905003696680069, "global_step": 32458, "epoch": 901, "lr": 0.001} +{"train_loss": 0.030640436336398125, "train_loss_bc": 0.030640436336398125, "train_loss_llm": 0.0, "grad_norm": 0.13156653940677643, "global_step": 32459, "epoch": 901, "lr": 0.001} +{"train_loss": 0.030796516686677933, "train_loss_bc": 0.030796516686677933, "train_loss_llm": 0.0, "grad_norm": 0.16436755657196045, "global_step": 32460, "epoch": 901, "lr": 0.001} +{"train_loss": 0.03134738281369209, "train_loss_bc": 0.03134738281369209, "train_loss_llm": 0.0, "grad_norm": 0.05168948695063591, "global_step": 32461, "epoch": 901, "lr": 0.001} +{"train_loss": 0.03094351477921009, "train_loss_bc": 0.03094351477921009, "train_loss_llm": 0.0, "grad_norm": 0.08293280750513077, "global_step": 32462, "epoch": 901, "lr": 0.001} +{"train_loss": 0.02685816027224064, "train_loss_bc": 0.02685816027224064, "train_loss_llm": 0.0, "grad_norm": 0.11666292697191238, "global_step": 32463, "epoch": 901, "lr": 0.001} +{"train_loss": 0.02881343476474285, "train_loss_bc": 0.02881343476474285, "train_loss_llm": 0.0, "grad_norm": 0.14140281081199646, "global_step": 32464, "epoch": 901, "lr": 0.001} +{"train_loss": 0.021199330687522888, "train_loss_bc": 0.02119792066514492, "train_loss_llm": 0.00014096498489379883, "grad_norm": 0.029051123186945915, "global_step": 32465, "epoch": 901, "lr": 0.001} +{"train_loss": 0.028521336615085602, "train_loss_bc": 0.028511732816696167, "train_loss_llm": 0.0009603202342987061, "grad_norm": 0.057932838797569275, "global_step": 32466, "epoch": 901, "lr": 0.001} +{"train_loss": 0.0335838757455349, "train_loss_bc": 0.0335838757455349, "train_loss_llm": 0.0, "grad_norm": 0.08899953216314316, "global_step": 32467, "epoch": 901, "lr": 0.001} +{"train_loss": 0.02910427376627922, "train_loss_bc": 0.02910427376627922, "train_loss_llm": 0.0, "grad_norm": 0.11057582497596741, "global_step": 32468, "epoch": 901, "lr": 0.001} +{"train_loss": 0.028986023738980293, "train_loss_bc": 0.028986023738980293, "train_loss_llm": 0.0, "grad_norm": 0.03680902719497681, "global_step": 32469, "epoch": 901, "lr": 0.001} +{"train_loss": 0.030250288546085358, "train_loss_bc": 0.030250288546085358, "train_loss_llm": 0.0, "grad_norm": 0.06349369883537292, "global_step": 32470, "epoch": 901, "lr": 0.001} +{"train_loss": 0.029715108788675733, "train_loss_bc": 0.02983124554157257, "train_loss_llm": 0.0, "grad_norm": 0.09791149199008942, "global_step": 32471, "epoch": 901, "lr": 0.001, "val_loss": 0.10466014593839645} +{"train_loss": 0.034251365810632706, "train_loss_bc": 0.034251365810632706, "train_loss_llm": 0.0, "grad_norm": 0.12223520129919052, "global_step": 32472, "epoch": 902, "lr": 0.001} +{"train_loss": 0.023845838382840157, "train_loss_bc": 0.023845838382840157, "train_loss_llm": 0.0, "grad_norm": 0.03629729524254799, "global_step": 32473, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03213363513350487, "train_loss_bc": 0.03213363513350487, "train_loss_llm": 0.0, "grad_norm": 0.06984668225049973, "global_step": 32474, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03090539202094078, "train_loss_bc": 0.03090539202094078, "train_loss_llm": 0.0, "grad_norm": 0.10116558521986008, "global_step": 32475, "epoch": 902, "lr": 0.001} +{"train_loss": 0.027603749185800552, "train_loss_bc": 0.027592336758971214, "train_loss_llm": 0.001141279935836792, "grad_norm": 0.11952410638332367, "global_step": 32476, "epoch": 902, "lr": 0.001} +{"train_loss": 0.02649865671992302, "train_loss_bc": 0.02648971416056156, "train_loss_llm": 0.0008942782878875732, "grad_norm": 0.03920060768723488, "global_step": 32477, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03157466650009155, "train_loss_bc": 0.03157466650009155, "train_loss_llm": 0.0, "grad_norm": 0.07956648617982864, "global_step": 32478, "epoch": 902, "lr": 0.001} +{"train_loss": 0.025856055319309235, "train_loss_bc": 0.025856055319309235, "train_loss_llm": 0.0, "grad_norm": 0.10510393232107162, "global_step": 32479, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03415656089782715, "train_loss_bc": 0.03415656089782715, "train_loss_llm": 0.0, "grad_norm": 0.1423477679491043, "global_step": 32480, "epoch": 902, "lr": 0.001} +{"train_loss": 0.02533925324678421, "train_loss_bc": 0.02533925324678421, "train_loss_llm": 0.0, "grad_norm": 0.050568368285894394, "global_step": 32481, "epoch": 902, "lr": 0.001} +{"train_loss": 0.02848133072257042, "train_loss_bc": 0.02848133072257042, "train_loss_llm": 0.0, "grad_norm": 0.07174131274223328, "global_step": 32482, "epoch": 902, "lr": 0.001} +{"train_loss": 0.031210877001285553, "train_loss_bc": 0.031210877001285553, "train_loss_llm": 0.0, "grad_norm": 0.09199444204568863, "global_step": 32483, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03170112520456314, "train_loss_bc": 0.03167193755507469, "train_loss_llm": 0.002918928861618042, "grad_norm": 0.11986062675714493, "global_step": 32484, "epoch": 902, "lr": 0.001} +{"train_loss": 0.0286884605884552, "train_loss_bc": 0.0286884605884552, "train_loss_llm": 0.0, "grad_norm": 0.0463717095553875, "global_step": 32485, "epoch": 902, "lr": 0.001} +{"train_loss": 0.02366490662097931, "train_loss_bc": 0.02366490662097931, "train_loss_llm": 0.0, "grad_norm": 0.06750357151031494, "global_step": 32486, "epoch": 902, "lr": 0.001} +{"train_loss": 0.02965778484940529, "train_loss_bc": 0.02965562976896763, "train_loss_llm": 0.00021544098854064941, "grad_norm": 0.09247834980487823, "global_step": 32487, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03792461007833481, "train_loss_bc": 0.03792461007833481, "train_loss_llm": 0.0, "grad_norm": 0.13455142080783844, "global_step": 32488, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03125558793544769, "train_loss_bc": 0.03125558793544769, "train_loss_llm": 0.0, "grad_norm": 0.04158566892147064, "global_step": 32489, "epoch": 902, "lr": 0.001} +{"train_loss": 0.02330203540623188, "train_loss_bc": 0.023285675793886185, "train_loss_llm": 0.0016359388828277588, "grad_norm": 0.05746172368526459, "global_step": 32490, "epoch": 902, "lr": 0.001} +{"train_loss": 0.026070883497595787, "train_loss_bc": 0.026070883497595787, "train_loss_llm": 0.0, "grad_norm": 0.06802435219287872, "global_step": 32491, "epoch": 902, "lr": 0.001} +{"train_loss": 0.030244292691349983, "train_loss_bc": 0.030244292691349983, "train_loss_llm": 0.0, "grad_norm": 0.08678552508354187, "global_step": 32492, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03142385557293892, "train_loss_bc": 0.03142280876636505, "train_loss_llm": 0.00010460615158081055, "grad_norm": 0.038912322372198105, "global_step": 32493, "epoch": 902, "lr": 0.001} +{"train_loss": 0.026461495086550713, "train_loss_bc": 0.026461495086550713, "train_loss_llm": 0.0, "grad_norm": 0.060525961220264435, "global_step": 32494, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03582058101892471, "train_loss_bc": 0.03582058101892471, "train_loss_llm": 0.0, "grad_norm": 0.09454531967639923, "global_step": 32495, "epoch": 902, "lr": 0.001} +{"train_loss": 0.026890967041254044, "train_loss_bc": 0.026890967041254044, "train_loss_llm": 0.0, "grad_norm": 0.11809638887643814, "global_step": 32496, "epoch": 902, "lr": 0.001} +{"train_loss": 0.024892522022128105, "train_loss_bc": 0.024889186024665833, "train_loss_llm": 0.00033357739448547363, "grad_norm": 0.03903425857424736, "global_step": 32497, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03360963612794876, "train_loss_bc": 0.03360963612794876, "train_loss_llm": 0.0, "grad_norm": 0.06944254785776138, "global_step": 32498, "epoch": 902, "lr": 0.001} +{"train_loss": 0.02774007059633732, "train_loss_bc": 0.02774007059633732, "train_loss_llm": 0.0, "grad_norm": 0.08828911185264587, "global_step": 32499, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03082778863608837, "train_loss_bc": 0.03082778863608837, "train_loss_llm": 0.0, "grad_norm": 0.11492601037025452, "global_step": 32500, "epoch": 902, "lr": 0.001} +{"train_loss": 0.030519306659698486, "train_loss_bc": 0.03051784262061119, "train_loss_llm": 0.00014635920524597168, "grad_norm": 0.04302487522363663, "global_step": 32501, "epoch": 902, "lr": 0.001} +{"train_loss": 0.02387223392724991, "train_loss_bc": 0.02387223392724991, "train_loss_llm": 0.0, "grad_norm": 0.05921206995844841, "global_step": 32502, "epoch": 902, "lr": 0.001} +{"train_loss": 0.030719228088855743, "train_loss_bc": 0.030715884640812874, "train_loss_llm": 0.0003342926502227783, "grad_norm": 0.09599494189023972, "global_step": 32503, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03391404449939728, "train_loss_bc": 0.03391404449939728, "train_loss_llm": 0.0, "grad_norm": 0.12519599497318268, "global_step": 32504, "epoch": 902, "lr": 0.001} +{"train_loss": 0.037470802664756775, "train_loss_bc": 0.037470802664756775, "train_loss_llm": 0.0, "grad_norm": 0.06439193338155746, "global_step": 32505, "epoch": 902, "lr": 0.001} +{"train_loss": 0.03451839089393616, "train_loss_bc": 0.03451839089393616, "train_loss_llm": 0.0, "grad_norm": 0.10334992408752441, "global_step": 32506, "epoch": 902, "lr": 0.001} +{"train_loss": 0.029759518129544124, "train_loss_bc": 0.02829466201364994, "train_loss_llm": 0.0, "grad_norm": 0.14524319767951965, "global_step": 32507, "epoch": 902, "lr": 0.001, "val_loss": 0.10619032382965088} +{"train_loss": 0.03345515578985214, "train_loss_bc": 0.033446408808231354, "train_loss_llm": 0.000874638557434082, "grad_norm": 0.1851276457309723, "global_step": 32508, "epoch": 903, "lr": 0.001} +{"train_loss": 0.030546700581908226, "train_loss_bc": 0.030546700581908226, "train_loss_llm": 0.0, "grad_norm": 0.034219447523355484, "global_step": 32509, "epoch": 903, "lr": 0.001} +{"train_loss": 0.025248004123568535, "train_loss_bc": 0.025248004123568535, "train_loss_llm": 0.0, "grad_norm": 0.05492978170514107, "global_step": 32510, "epoch": 903, "lr": 0.001} +{"train_loss": 0.025499649345874786, "train_loss_bc": 0.025499649345874786, "train_loss_llm": 0.0, "grad_norm": 0.0813542902469635, "global_step": 32511, "epoch": 903, "lr": 0.001} +{"train_loss": 0.02872633934020996, "train_loss_bc": 0.02872633934020996, "train_loss_llm": 0.0, "grad_norm": 0.10603949427604675, "global_step": 32512, "epoch": 903, "lr": 0.001} +{"train_loss": 0.02766370214521885, "train_loss_bc": 0.02766370214521885, "train_loss_llm": 0.0, "grad_norm": 0.03274710103869438, "global_step": 32513, "epoch": 903, "lr": 0.001} +{"train_loss": 0.030956575646996498, "train_loss_bc": 0.030956575646996498, "train_loss_llm": 0.0, "grad_norm": 0.057192835956811905, "global_step": 32514, "epoch": 903, "lr": 0.001} +{"train_loss": 0.031171834096312523, "train_loss_bc": 0.031171834096312523, "train_loss_llm": 0.0, "grad_norm": 0.0802333727478981, "global_step": 32515, "epoch": 903, "lr": 0.001} +{"train_loss": 0.02860984206199646, "train_loss_bc": 0.02860984206199646, "train_loss_llm": 0.0, "grad_norm": 0.0950501337647438, "global_step": 32516, "epoch": 903, "lr": 0.001} +{"train_loss": 0.027312908321619034, "train_loss_bc": 0.027312908321619034, "train_loss_llm": 0.0, "grad_norm": 0.04105205461382866, "global_step": 32517, "epoch": 903, "lr": 0.001} +{"train_loss": 0.03100127913057804, "train_loss_bc": 0.03100127913057804, "train_loss_llm": 0.0, "grad_norm": 0.0729474425315857, "global_step": 32518, "epoch": 903, "lr": 0.001} +{"train_loss": 0.034532152116298676, "train_loss_bc": 0.034532152116298676, "train_loss_llm": 0.0, "grad_norm": 0.09074273705482483, "global_step": 32519, "epoch": 903, "lr": 0.001} +{"train_loss": 0.03318651393055916, "train_loss_bc": 0.03318651393055916, "train_loss_llm": 0.0, "grad_norm": 0.12730495631694794, "global_step": 32520, "epoch": 903, "lr": 0.001} +{"train_loss": 0.02564726211130619, "train_loss_bc": 0.02562040090560913, "train_loss_llm": 0.002686113119125366, "grad_norm": 0.04613228887319565, "global_step": 32521, "epoch": 903, "lr": 0.001} +{"train_loss": 0.026296965777873993, "train_loss_bc": 0.026296965777873993, "train_loss_llm": 0.0, "grad_norm": 0.07152746617794037, "global_step": 32522, "epoch": 903, "lr": 0.001} +{"train_loss": 0.03385637700557709, "train_loss_bc": 0.03385637700557709, "train_loss_llm": 0.0, "grad_norm": 0.10417583584785461, "global_step": 32523, "epoch": 903, "lr": 0.001} +{"train_loss": 0.02852417342364788, "train_loss_bc": 0.02852417342364788, "train_loss_llm": 0.0, "grad_norm": 0.1414066106081009, "global_step": 32524, "epoch": 903, "lr": 0.001} +{"train_loss": 0.030317801982164383, "train_loss_bc": 0.030304627493023872, "train_loss_llm": 0.0013173818588256836, "grad_norm": 0.048169516026973724, "global_step": 32525, "epoch": 903, "lr": 0.001} +{"train_loss": 0.03430362790822983, "train_loss_bc": 0.03430362790822983, "train_loss_llm": 0.0, "grad_norm": 0.08219431340694427, "global_step": 32526, "epoch": 903, "lr": 0.001} +{"train_loss": 0.02816746197640896, "train_loss_bc": 0.02816436067223549, "train_loss_llm": 0.0003101527690887451, "grad_norm": 0.11304862052202225, "global_step": 32527, "epoch": 903, "lr": 0.001} +{"train_loss": 0.030374880880117416, "train_loss_bc": 0.030374880880117416, "train_loss_llm": 0.0, "grad_norm": 0.1388280838727951, "global_step": 32528, "epoch": 903, "lr": 0.001} +{"train_loss": 0.028177013620734215, "train_loss_bc": 0.028177013620734215, "train_loss_llm": 0.0, "grad_norm": 0.04382419213652611, "global_step": 32529, "epoch": 903, "lr": 0.001} +{"train_loss": 0.03017737716436386, "train_loss_bc": 0.03017737716436386, "train_loss_llm": 0.0, "grad_norm": 0.07321922481060028, "global_step": 32530, "epoch": 903, "lr": 0.001} +{"train_loss": 0.02709467150270939, "train_loss_bc": 0.02709467150270939, "train_loss_llm": 0.0, "grad_norm": 0.08184871077537537, "global_step": 32531, "epoch": 903, "lr": 0.001} +{"train_loss": 0.030103936791419983, "train_loss_bc": 0.030093912035226822, "train_loss_llm": 0.0010025203227996826, "grad_norm": 0.11383672058582306, "global_step": 32532, "epoch": 903, "lr": 0.001} +{"train_loss": 0.029499152675271034, "train_loss_bc": 0.029499152675271034, "train_loss_llm": 0.0, "grad_norm": 0.03744003176689148, "global_step": 32533, "epoch": 903, "lr": 0.001} +{"train_loss": 0.030825035646557808, "train_loss_bc": 0.03082289919257164, "train_loss_llm": 0.0002135634422302246, "grad_norm": 0.06360919028520584, "global_step": 32534, "epoch": 903, "lr": 0.001} +{"train_loss": 0.03249005600810051, "train_loss_bc": 0.03249005600810051, "train_loss_llm": 0.0, "grad_norm": 0.09225510060787201, "global_step": 32535, "epoch": 903, "lr": 0.001} +{"train_loss": 0.03197571635246277, "train_loss_bc": 0.03197571635246277, "train_loss_llm": 0.0, "grad_norm": 0.10946668684482574, "global_step": 32536, "epoch": 903, "lr": 0.001} +{"train_loss": 0.020534202456474304, "train_loss_bc": 0.020534202456474304, "train_loss_llm": 0.0, "grad_norm": 0.028780309483408928, "global_step": 32537, "epoch": 903, "lr": 0.001} +{"train_loss": 0.028248755261301994, "train_loss_bc": 0.028248755261301994, "train_loss_llm": 0.0, "grad_norm": 0.04440852254629135, "global_step": 32538, "epoch": 903, "lr": 0.001} +{"train_loss": 0.03281993418931961, "train_loss_bc": 0.03281993418931961, "train_loss_llm": 0.0, "grad_norm": 0.0694759339094162, "global_step": 32539, "epoch": 903, "lr": 0.001} +{"train_loss": 0.030664613470435143, "train_loss_bc": 0.030662493780255318, "train_loss_llm": 0.00021198391914367676, "grad_norm": 0.089643195271492, "global_step": 32540, "epoch": 903, "lr": 0.001} +{"train_loss": 0.027730878442525864, "train_loss_bc": 0.027730878442525864, "train_loss_llm": 0.0, "grad_norm": 0.03815707564353943, "global_step": 32541, "epoch": 903, "lr": 0.001} +{"train_loss": 0.030648082494735718, "train_loss_bc": 0.030648082494735718, "train_loss_llm": 0.0, "grad_norm": 0.06598210334777832, "global_step": 32542, "epoch": 903, "lr": 0.001} +{"train_loss": 0.02950902283191681, "train_loss_bc": 0.025922318920493126, "train_loss_llm": 0.0013870000839233398, "grad_norm": 0.07591383904218674, "global_step": 32543, "epoch": 903, "lr": 0.001, "val_loss": 0.1030801460146904} +{"train_loss": 0.03317217156291008, "train_loss_bc": 0.03317217156291008, "train_loss_llm": 0.0, "grad_norm": 0.1149442121386528, "global_step": 32544, "epoch": 904, "lr": 0.001} +{"train_loss": 0.03365462273359299, "train_loss_bc": 0.03365462273359299, "train_loss_llm": 0.0, "grad_norm": 0.034982528537511826, "global_step": 32545, "epoch": 904, "lr": 0.001} +{"train_loss": 0.02668186090886593, "train_loss_bc": 0.02668186090886593, "train_loss_llm": 0.0, "grad_norm": 0.05027846246957779, "global_step": 32546, "epoch": 904, "lr": 0.001} +{"train_loss": 0.031639937311410904, "train_loss_bc": 0.03163745999336243, "train_loss_llm": 0.0002478659152984619, "grad_norm": 0.08330654352903366, "global_step": 32547, "epoch": 904, "lr": 0.001} +{"train_loss": 0.028616588562726974, "train_loss_bc": 0.028616588562726974, "train_loss_llm": 0.0, "grad_norm": 0.09985699504613876, "global_step": 32548, "epoch": 904, "lr": 0.001} +{"train_loss": 0.0221729539334774, "train_loss_bc": 0.0221534613519907, "train_loss_llm": 0.0019493401050567627, "grad_norm": 0.03252105042338371, "global_step": 32549, "epoch": 904, "lr": 0.001} +{"train_loss": 0.02755078673362732, "train_loss_bc": 0.02755078673362732, "train_loss_llm": 0.0, "grad_norm": 0.05712398514151573, "global_step": 32550, "epoch": 904, "lr": 0.001} +{"train_loss": 0.02944139577448368, "train_loss_bc": 0.02944139577448368, "train_loss_llm": 0.0, "grad_norm": 0.08518471568822861, "global_step": 32551, "epoch": 904, "lr": 0.001} +{"train_loss": 0.029794543981552124, "train_loss_bc": 0.02977318875491619, "train_loss_llm": 0.0021354854106903076, "grad_norm": 0.10852082818746567, "global_step": 32552, "epoch": 904, "lr": 0.001} +{"train_loss": 0.03560763970017433, "train_loss_bc": 0.035600293427705765, "train_loss_llm": 0.0007347762584686279, "grad_norm": 0.04659077897667885, "global_step": 32553, "epoch": 904, "lr": 0.001} +{"train_loss": 0.03042621538043022, "train_loss_bc": 0.03042621538043022, "train_loss_llm": 0.0, "grad_norm": 0.06141418218612671, "global_step": 32554, "epoch": 904, "lr": 0.001} +{"train_loss": 0.02581016719341278, "train_loss_bc": 0.02579282596707344, "train_loss_llm": 0.0017340481281280518, "grad_norm": 0.07419268041849136, "global_step": 32555, "epoch": 904, "lr": 0.001} +{"train_loss": 0.027684198692440987, "train_loss_bc": 0.027684198692440987, "train_loss_llm": 0.0, "grad_norm": 0.08191950619220734, "global_step": 32556, "epoch": 904, "lr": 0.001} +{"train_loss": 0.031069952994585037, "train_loss_bc": 0.031063254922628403, "train_loss_llm": 0.0006697773933410645, "grad_norm": 0.03619905933737755, "global_step": 32557, "epoch": 904, "lr": 0.001} +{"train_loss": 0.03303363174200058, "train_loss_bc": 0.03303363174200058, "train_loss_llm": 0.0, "grad_norm": 0.06379451602697372, "global_step": 32558, "epoch": 904, "lr": 0.001} +{"train_loss": 0.024506546556949615, "train_loss_bc": 0.02449941448867321, "train_loss_llm": 0.0007132589817047119, "grad_norm": 0.08132325857877731, "global_step": 32559, "epoch": 904, "lr": 0.001} +{"train_loss": 0.029974816367030144, "train_loss_bc": 0.029974816367030144, "train_loss_llm": 0.0, "grad_norm": 0.10288015753030777, "global_step": 32560, "epoch": 904, "lr": 0.001} +{"train_loss": 0.0284099280834198, "train_loss_bc": 0.0284099280834198, "train_loss_llm": 0.0, "grad_norm": 0.03266691789031029, "global_step": 32561, "epoch": 904, "lr": 0.001} +{"train_loss": 0.0289137102663517, "train_loss_bc": 0.0289137102663517, "train_loss_llm": 0.0, "grad_norm": 0.04857216030359268, "global_step": 32562, "epoch": 904, "lr": 0.001} +{"train_loss": 0.026667838916182518, "train_loss_bc": 0.026667838916182518, "train_loss_llm": 0.0, "grad_norm": 0.06818995624780655, "global_step": 32563, "epoch": 904, "lr": 0.001} +{"train_loss": 0.027320897206664085, "train_loss_bc": 0.027316344901919365, "train_loss_llm": 0.0004552304744720459, "grad_norm": 0.08594594895839691, "global_step": 32564, "epoch": 904, "lr": 0.001} +{"train_loss": 0.029651811346411705, "train_loss_bc": 0.029651811346411705, "train_loss_llm": 0.0, "grad_norm": 0.03839162737131119, "global_step": 32565, "epoch": 904, "lr": 0.001} +{"train_loss": 0.026463450863957405, "train_loss_bc": 0.02645915001630783, "train_loss_llm": 0.00043007731437683105, "grad_norm": 0.06585478782653809, "global_step": 32566, "epoch": 904, "lr": 0.001} +{"train_loss": 0.028366994112730026, "train_loss_bc": 0.028366994112730026, "train_loss_llm": 0.0, "grad_norm": 0.0794093981385231, "global_step": 32567, "epoch": 904, "lr": 0.001} +{"train_loss": 0.02824161946773529, "train_loss_bc": 0.02824161946773529, "train_loss_llm": 0.0, "grad_norm": 0.09999212622642517, "global_step": 32568, "epoch": 904, "lr": 0.001} +{"train_loss": 0.03453216701745987, "train_loss_bc": 0.034513264894485474, "train_loss_llm": 0.0018903613090515137, "grad_norm": 0.04418014734983444, "global_step": 32569, "epoch": 904, "lr": 0.001} +{"train_loss": 0.03270038962364197, "train_loss_bc": 0.03270038962364197, "train_loss_llm": 0.0, "grad_norm": 0.0632210522890091, "global_step": 32570, "epoch": 904, "lr": 0.001} +{"train_loss": 0.0262041836977005, "train_loss_bc": 0.0262041836977005, "train_loss_llm": 0.0, "grad_norm": 0.08017539978027344, "global_step": 32571, "epoch": 904, "lr": 0.001} +{"train_loss": 0.03243577480316162, "train_loss_bc": 0.03243577480316162, "train_loss_llm": 0.0, "grad_norm": 0.09708486497402191, "global_step": 32572, "epoch": 904, "lr": 0.001} +{"train_loss": 0.029632093384861946, "train_loss_bc": 0.029632093384861946, "train_loss_llm": 0.0, "grad_norm": 0.04767045006155968, "global_step": 32573, "epoch": 904, "lr": 0.001} +{"train_loss": 0.026114599779248238, "train_loss_bc": 0.026114599779248238, "train_loss_llm": 0.0, "grad_norm": 0.07049737870693207, "global_step": 32574, "epoch": 904, "lr": 0.001} +{"train_loss": 0.03207673877477646, "train_loss_bc": 0.032061025500297546, "train_loss_llm": 0.0015712082386016846, "grad_norm": 0.09484906494617462, "global_step": 32575, "epoch": 904, "lr": 0.001} +{"train_loss": 0.02762828767299652, "train_loss_bc": 0.02762828767299652, "train_loss_llm": 0.0, "grad_norm": 0.11354852467775345, "global_step": 32576, "epoch": 904, "lr": 0.001} +{"train_loss": 0.025562670081853867, "train_loss_bc": 0.025562670081853867, "train_loss_llm": 0.0, "grad_norm": 0.03858495503664017, "global_step": 32577, "epoch": 904, "lr": 0.001} +{"train_loss": 0.03381361439824104, "train_loss_bc": 0.03381361439824104, "train_loss_llm": 0.0, "grad_norm": 0.06097495183348656, "global_step": 32578, "epoch": 904, "lr": 0.001} +{"train_loss": 0.02932015113118622, "train_loss_bc": 0.029950641095638275, "train_loss_llm": 0.0, "grad_norm": 0.08238104730844498, "global_step": 32579, "epoch": 904, "lr": 0.001, "val_loss": 0.10426292568445206} +{"train_loss": 0.025926509872078896, "train_loss_bc": 0.025914466008543968, "train_loss_llm": 0.0012044310569763184, "grad_norm": 0.09485486894845963, "global_step": 32580, "epoch": 905, "lr": 0.001} +{"train_loss": 0.027514159679412842, "train_loss_bc": 0.027514159679412842, "train_loss_llm": 0.0, "grad_norm": 0.03927190601825714, "global_step": 32581, "epoch": 905, "lr": 0.001} +{"train_loss": 0.02698189951479435, "train_loss_bc": 0.02698189951479435, "train_loss_llm": 0.0, "grad_norm": 0.057686496526002884, "global_step": 32582, "epoch": 905, "lr": 0.001} +{"train_loss": 0.03384791314601898, "train_loss_bc": 0.03384791314601898, "train_loss_llm": 0.0, "grad_norm": 0.08441086858510971, "global_step": 32583, "epoch": 905, "lr": 0.001} +{"train_loss": 0.03166291490197182, "train_loss_bc": 0.031660858541727066, "train_loss_llm": 0.00020569562911987305, "grad_norm": 0.1193934828042984, "global_step": 32584, "epoch": 905, "lr": 0.001} +{"train_loss": 0.030391711741685867, "train_loss_bc": 0.030391711741685867, "train_loss_llm": 0.0, "grad_norm": 0.05369843915104866, "global_step": 32585, "epoch": 905, "lr": 0.001} +{"train_loss": 0.03164065629243851, "train_loss_bc": 0.03164065629243851, "train_loss_llm": 0.0, "grad_norm": 0.08117223531007767, "global_step": 32586, "epoch": 905, "lr": 0.001} +{"train_loss": 0.028331708163022995, "train_loss_bc": 0.02833167463541031, "train_loss_llm": 3.3676624298095703e-06, "grad_norm": 0.10536646097898483, "global_step": 32587, "epoch": 905, "lr": 0.001} +{"train_loss": 0.029767051339149475, "train_loss_bc": 0.029765063896775246, "train_loss_llm": 0.00019872188568115234, "grad_norm": 0.13350112736225128, "global_step": 32588, "epoch": 905, "lr": 0.001} +{"train_loss": 0.0325508676469326, "train_loss_bc": 0.0325508676469326, "train_loss_llm": 0.0, "grad_norm": 0.05537812039256096, "global_step": 32589, "epoch": 905, "lr": 0.001} +{"train_loss": 0.032890450209379196, "train_loss_bc": 0.032890450209379196, "train_loss_llm": 0.0, "grad_norm": 0.10612110793590546, "global_step": 32590, "epoch": 905, "lr": 0.001} +{"train_loss": 0.02611153945326805, "train_loss_bc": 0.02611153945326805, "train_loss_llm": 0.0, "grad_norm": 0.11783163994550705, "global_step": 32591, "epoch": 905, "lr": 0.001} +{"train_loss": 0.03118850849568844, "train_loss_bc": 0.031169869005680084, "train_loss_llm": 0.0018638670444488525, "grad_norm": 0.1425584852695465, "global_step": 32592, "epoch": 905, "lr": 0.001} +{"train_loss": 0.029487859457731247, "train_loss_bc": 0.029487859457731247, "train_loss_llm": 0.0, "grad_norm": 0.03389156982302666, "global_step": 32593, "epoch": 905, "lr": 0.001} +{"train_loss": 0.03218410164117813, "train_loss_bc": 0.03218410164117813, "train_loss_llm": 0.0, "grad_norm": 0.0512276329100132, "global_step": 32594, "epoch": 905, "lr": 0.001} +{"train_loss": 0.03112165257334709, "train_loss_bc": 0.03112165257334709, "train_loss_llm": 0.0, "grad_norm": 0.06914818286895752, "global_step": 32595, "epoch": 905, "lr": 0.001} +{"train_loss": 0.025347957387566566, "train_loss_bc": 0.025347957387566566, "train_loss_llm": 0.0, "grad_norm": 0.09417729824781418, "global_step": 32596, "epoch": 905, "lr": 0.001} +{"train_loss": 0.02377375029027462, "train_loss_bc": 0.02377375029027462, "train_loss_llm": 0.0, "grad_norm": 0.03207344934344292, "global_step": 32597, "epoch": 905, "lr": 0.001} +{"train_loss": 0.027459993958473206, "train_loss_bc": 0.027459993958473206, "train_loss_llm": 0.0, "grad_norm": 0.053587656468153, "global_step": 32598, "epoch": 905, "lr": 0.001} +{"train_loss": 0.02857942320406437, "train_loss_bc": 0.028572073206305504, "train_loss_llm": 0.0007350146770477295, "grad_norm": 0.07472754269838333, "global_step": 32599, "epoch": 905, "lr": 0.001} +{"train_loss": 0.03396590054035187, "train_loss_bc": 0.03396590054035187, "train_loss_llm": 0.0, "grad_norm": 0.09989135712385178, "global_step": 32600, "epoch": 905, "lr": 0.001} +{"train_loss": 0.028743665665388107, "train_loss_bc": 0.028743665665388107, "train_loss_llm": 0.0, "grad_norm": 0.038596261292696, "global_step": 32601, "epoch": 905, "lr": 0.001} +{"train_loss": 0.030300945043563843, "train_loss_bc": 0.030300945043563843, "train_loss_llm": 0.0, "grad_norm": 0.0621604397892952, "global_step": 32602, "epoch": 905, "lr": 0.001} +{"train_loss": 0.03127378970384598, "train_loss_bc": 0.03127378970384598, "train_loss_llm": 0.0, "grad_norm": 0.08223740756511688, "global_step": 32603, "epoch": 905, "lr": 0.001} +{"train_loss": 0.02761298231780529, "train_loss_bc": 0.02761298231780529, "train_loss_llm": 0.0, "grad_norm": 0.11482745409011841, "global_step": 32604, "epoch": 905, "lr": 0.001} +{"train_loss": 0.028940308839082718, "train_loss_bc": 0.028940308839082718, "train_loss_llm": 0.0, "grad_norm": 0.04080452769994736, "global_step": 32605, "epoch": 905, "lr": 0.001} +{"train_loss": 0.029012801125645638, "train_loss_bc": 0.029006140306591988, "train_loss_llm": 0.0006661415100097656, "grad_norm": 0.06602021306753159, "global_step": 32606, "epoch": 905, "lr": 0.001} +{"train_loss": 0.03466445952653885, "train_loss_bc": 0.03466445952653885, "train_loss_llm": 0.0, "grad_norm": 0.10100533813238144, "global_step": 32607, "epoch": 905, "lr": 0.001} +{"train_loss": 0.01955309882760048, "train_loss_bc": 0.01955309882760048, "train_loss_llm": 0.0, "grad_norm": 0.1143568754196167, "global_step": 32608, "epoch": 905, "lr": 0.001} +{"train_loss": 0.0293266661465168, "train_loss_bc": 0.0293266661465168, "train_loss_llm": 0.0, "grad_norm": 0.03942689672112465, "global_step": 32609, "epoch": 905, "lr": 0.001} +{"train_loss": 0.029602807015180588, "train_loss_bc": 0.029589157551527023, "train_loss_llm": 0.0013649165630340576, "grad_norm": 0.06107411906123161, "global_step": 32610, "epoch": 905, "lr": 0.001} +{"train_loss": 0.029835399240255356, "train_loss_bc": 0.029829276725649834, "train_loss_llm": 0.0006122887134552002, "grad_norm": 0.08130773156881332, "global_step": 32611, "epoch": 905, "lr": 0.001} +{"train_loss": 0.031113820150494576, "train_loss_bc": 0.031113820150494576, "train_loss_llm": 0.0, "grad_norm": 0.10287410020828247, "global_step": 32612, "epoch": 905, "lr": 0.001} +{"train_loss": 0.029220161959528923, "train_loss_bc": 0.029220161959528923, "train_loss_llm": 0.0, "grad_norm": 0.04541058838367462, "global_step": 32613, "epoch": 905, "lr": 0.001} +{"train_loss": 0.03089291788637638, "train_loss_bc": 0.03089291788637638, "train_loss_llm": 0.0, "grad_norm": 0.07737772911787033, "global_step": 32614, "epoch": 905, "lr": 0.001} +{"train_loss": 0.029577098217689328, "train_loss_bc": 0.03394872322678566, "train_loss_llm": 0.0006461441516876221, "grad_norm": 0.10953566431999207, "global_step": 32615, "epoch": 905, "lr": 0.001, "train/cumulative_reward": 78.04010761328905, "train/mean_score": 13.82865536232543, "train/success_rate": 0.3, "test/cumulative_reward": 31.552040733976618, "test/mean_score": 8.36152175109119, "test/success_rate": 0.22, "val_loss": 0.10520896315574646, "train_action_mse_error": 0.03286708518862724} +{"train_loss": 0.0248564463108778, "train_loss_bc": 0.0248564463108778, "train_loss_llm": 0.0, "grad_norm": 0.1352071911096573, "global_step": 32616, "epoch": 906, "lr": 0.001} +{"train_loss": 0.027634618803858757, "train_loss_bc": 0.027634618803858757, "train_loss_llm": 0.0, "grad_norm": 0.03704705461859703, "global_step": 32617, "epoch": 906, "lr": 0.001} +{"train_loss": 0.03527703136205673, "train_loss_bc": 0.03527703136205673, "train_loss_llm": 0.0, "grad_norm": 0.07985509932041168, "global_step": 32618, "epoch": 906, "lr": 0.001} +{"train_loss": 0.02475983276963234, "train_loss_bc": 0.024745509028434753, "train_loss_llm": 0.0014323890209197998, "grad_norm": 0.09530329704284668, "global_step": 32619, "epoch": 906, "lr": 0.001} +{"train_loss": 0.029843226075172424, "train_loss_bc": 0.029843226075172424, "train_loss_llm": 0.0, "grad_norm": 0.12349546700716019, "global_step": 32620, "epoch": 906, "lr": 0.001} +{"train_loss": 0.031093429774045944, "train_loss_bc": 0.03108029067516327, "train_loss_llm": 0.0013139843940734863, "grad_norm": 0.039416611194610596, "global_step": 32621, "epoch": 906, "lr": 0.001} +{"train_loss": 0.03145047649741173, "train_loss_bc": 0.03145047649741173, "train_loss_llm": 0.0, "grad_norm": 0.06836602091789246, "global_step": 32622, "epoch": 906, "lr": 0.001} +{"train_loss": 0.028135141357779503, "train_loss_bc": 0.028135141357779503, "train_loss_llm": 0.0, "grad_norm": 0.0887603610754013, "global_step": 32623, "epoch": 906, "lr": 0.001} +{"train_loss": 0.02627711370587349, "train_loss_bc": 0.02626512199640274, "train_loss_llm": 0.0011992156505584717, "grad_norm": 0.11164631694555283, "global_step": 32624, "epoch": 906, "lr": 0.001} +{"train_loss": 0.02519901469349861, "train_loss_bc": 0.02519901469349861, "train_loss_llm": 0.0, "grad_norm": 0.03407513350248337, "global_step": 32625, "epoch": 906, "lr": 0.001} +{"train_loss": 0.028015639632940292, "train_loss_bc": 0.028015639632940292, "train_loss_llm": 0.0, "grad_norm": 0.05000604689121246, "global_step": 32626, "epoch": 906, "lr": 0.001} +{"train_loss": 0.02868836000561714, "train_loss_bc": 0.02868836000561714, "train_loss_llm": 0.0, "grad_norm": 0.06872375309467316, "global_step": 32627, "epoch": 906, "lr": 0.001} +{"train_loss": 0.03145517036318779, "train_loss_bc": 0.03145517036318779, "train_loss_llm": 0.0, "grad_norm": 0.08335933834314346, "global_step": 32628, "epoch": 906, "lr": 0.001} +{"train_loss": 0.03160130977630615, "train_loss_bc": 0.03159620612859726, "train_loss_llm": 0.0005102157592773438, "grad_norm": 0.03552541136741638, "global_step": 32629, "epoch": 906, "lr": 0.001} +{"train_loss": 0.031024517491459846, "train_loss_bc": 0.031018298119306564, "train_loss_llm": 0.0006218552589416504, "grad_norm": 0.05463942885398865, "global_step": 32630, "epoch": 906, "lr": 0.001} +{"train_loss": 0.029595699161291122, "train_loss_bc": 0.02958935685455799, "train_loss_llm": 0.0006342232227325439, "grad_norm": 0.07771673798561096, "global_step": 32631, "epoch": 906, "lr": 0.001} +{"train_loss": 0.027843311429023743, "train_loss_bc": 0.0278378464281559, "train_loss_llm": 0.000546574592590332, "grad_norm": 0.10086245834827423, "global_step": 32632, "epoch": 906, "lr": 0.001} +{"train_loss": 0.026825470849871635, "train_loss_bc": 0.026825470849871635, "train_loss_llm": 0.0, "grad_norm": 0.034662459045648575, "global_step": 32633, "epoch": 906, "lr": 0.001} +{"train_loss": 0.033884067088365555, "train_loss_bc": 0.033884067088365555, "train_loss_llm": 0.0, "grad_norm": 0.04993550106883049, "global_step": 32634, "epoch": 906, "lr": 0.001} +{"train_loss": 0.029343998059630394, "train_loss_bc": 0.029343998059630394, "train_loss_llm": 0.0, "grad_norm": 0.06295508146286011, "global_step": 32635, "epoch": 906, "lr": 0.001} +{"train_loss": 0.030772414058446884, "train_loss_bc": 0.030772414058446884, "train_loss_llm": 0.0, "grad_norm": 0.09739396721124649, "global_step": 32636, "epoch": 906, "lr": 0.001} +{"train_loss": 0.031647298485040665, "train_loss_bc": 0.031647298485040665, "train_loss_llm": 0.0, "grad_norm": 0.04389181360602379, "global_step": 32637, "epoch": 906, "lr": 0.001} +{"train_loss": 0.024707922711968422, "train_loss_bc": 0.024691855534911156, "train_loss_llm": 0.0016067028045654297, "grad_norm": 0.0612073577940464, "global_step": 32638, "epoch": 906, "lr": 0.001} +{"train_loss": 0.026125263422727585, "train_loss_bc": 0.026125263422727585, "train_loss_llm": 0.0, "grad_norm": 0.09332002699375153, "global_step": 32639, "epoch": 906, "lr": 0.001} +{"train_loss": 0.031978197395801544, "train_loss_bc": 0.031978197395801544, "train_loss_llm": 0.0, "grad_norm": 0.11203345656394958, "global_step": 32640, "epoch": 906, "lr": 0.001} +{"train_loss": 0.024610474705696106, "train_loss_bc": 0.024610474705696106, "train_loss_llm": 0.0, "grad_norm": 0.0485217347741127, "global_step": 32641, "epoch": 906, "lr": 0.001} +{"train_loss": 0.027369290590286255, "train_loss_bc": 0.027369290590286255, "train_loss_llm": 0.0, "grad_norm": 0.07139883935451508, "global_step": 32642, "epoch": 906, "lr": 0.001} +{"train_loss": 0.03275985270738602, "train_loss_bc": 0.03275985270738602, "train_loss_llm": 0.0, "grad_norm": 0.0955095887184143, "global_step": 32643, "epoch": 906, "lr": 0.001} +{"train_loss": 0.03201223537325859, "train_loss_bc": 0.032007329165935516, "train_loss_llm": 0.0004905462265014648, "grad_norm": 0.12225297838449478, "global_step": 32644, "epoch": 906, "lr": 0.001} +{"train_loss": 0.03613598644733429, "train_loss_bc": 0.03613598644733429, "train_loss_llm": 0.0, "grad_norm": 0.0419723317027092, "global_step": 32645, "epoch": 906, "lr": 0.001} +{"train_loss": 0.029625611379742622, "train_loss_bc": 0.029623815789818764, "train_loss_llm": 0.00017946958541870117, "grad_norm": 0.05933661386370659, "global_step": 32646, "epoch": 906, "lr": 0.001} +{"train_loss": 0.03242814168334007, "train_loss_bc": 0.03242814168334007, "train_loss_llm": 0.0, "grad_norm": 0.07850489765405655, "global_step": 32647, "epoch": 906, "lr": 0.001} +{"train_loss": 0.027640772983431816, "train_loss_bc": 0.027640772983431816, "train_loss_llm": 0.0, "grad_norm": 0.10110072791576385, "global_step": 32648, "epoch": 906, "lr": 0.001} +{"train_loss": 0.026431338861584663, "train_loss_bc": 0.026431338861584663, "train_loss_llm": 0.0, "grad_norm": 0.035012226551771164, "global_step": 32649, "epoch": 906, "lr": 0.001} +{"train_loss": 0.028866615146398544, "train_loss_bc": 0.028866615146398544, "train_loss_llm": 0.0, "grad_norm": 0.05081624910235405, "global_step": 32650, "epoch": 906, "lr": 0.001} +{"train_loss": 0.029302428010851145, "train_loss_bc": 0.028972117230296135, "train_loss_llm": 0.0, "grad_norm": 0.07625159621238708, "global_step": 32651, "epoch": 906, "lr": 0.001, "val_loss": 0.10549730807542801} +{"train_loss": 0.028462117537856102, "train_loss_bc": 0.028462117537856102, "train_loss_llm": 0.0, "grad_norm": 0.10351064056158066, "global_step": 32652, "epoch": 907, "lr": 0.001} +{"train_loss": 0.030767453834414482, "train_loss_bc": 0.030767453834414482, "train_loss_llm": 0.0, "grad_norm": 0.03572346642613411, "global_step": 32653, "epoch": 907, "lr": 0.001} +{"train_loss": 0.02604442648589611, "train_loss_bc": 0.026038911193609238, "train_loss_llm": 0.0005514919757843018, "grad_norm": 0.050798580050468445, "global_step": 32654, "epoch": 907, "lr": 0.001} +{"train_loss": 0.028446439653635025, "train_loss_bc": 0.028445733711123466, "train_loss_llm": 7.05718994140625e-05, "grad_norm": 0.06217474117875099, "global_step": 32655, "epoch": 907, "lr": 0.001} +{"train_loss": 0.025053920224308968, "train_loss_bc": 0.025053920224308968, "train_loss_llm": 0.0, "grad_norm": 0.07995039969682693, "global_step": 32656, "epoch": 907, "lr": 0.001} +{"train_loss": 0.027118615806102753, "train_loss_bc": 0.027118615806102753, "train_loss_llm": 0.0, "grad_norm": 0.0365254171192646, "global_step": 32657, "epoch": 907, "lr": 0.001} +{"train_loss": 0.03233904018998146, "train_loss_bc": 0.03232768550515175, "train_loss_llm": 0.0011355578899383545, "grad_norm": 0.06028592959046364, "global_step": 32658, "epoch": 907, "lr": 0.001} +{"train_loss": 0.024623185396194458, "train_loss_bc": 0.024623185396194458, "train_loss_llm": 0.0, "grad_norm": 0.08211953192949295, "global_step": 32659, "epoch": 907, "lr": 0.001} +{"train_loss": 0.03059988096356392, "train_loss_bc": 0.03059988096356392, "train_loss_llm": 0.0, "grad_norm": 0.10403518378734589, "global_step": 32660, "epoch": 907, "lr": 0.001} +{"train_loss": 0.030442291870713234, "train_loss_bc": 0.030442291870713234, "train_loss_llm": 0.0, "grad_norm": 0.03693748265504837, "global_step": 32661, "epoch": 907, "lr": 0.001} +{"train_loss": 0.03299516439437866, "train_loss_bc": 0.03299516439437866, "train_loss_llm": 0.0, "grad_norm": 0.060830071568489075, "global_step": 32662, "epoch": 907, "lr": 0.001} +{"train_loss": 0.030286718159914017, "train_loss_bc": 0.030286718159914017, "train_loss_llm": 0.0, "grad_norm": 0.08777324855327606, "global_step": 32663, "epoch": 907, "lr": 0.001} +{"train_loss": 0.027110768482089043, "train_loss_bc": 0.027091415598988533, "train_loss_llm": 0.0019353032112121582, "grad_norm": 0.10898850113153458, "global_step": 32664, "epoch": 907, "lr": 0.001} +{"train_loss": 0.027902429923415184, "train_loss_bc": 0.027902429923415184, "train_loss_llm": 0.0, "grad_norm": 0.035760506987571716, "global_step": 32665, "epoch": 907, "lr": 0.001} +{"train_loss": 0.028805015608668327, "train_loss_bc": 0.028799301013350487, "train_loss_llm": 0.0005714893341064453, "grad_norm": 0.05776076018810272, "global_step": 32666, "epoch": 907, "lr": 0.001} +{"train_loss": 0.03455497324466705, "train_loss_bc": 0.03455497324466705, "train_loss_llm": 0.0, "grad_norm": 0.07900090515613556, "global_step": 32667, "epoch": 907, "lr": 0.001} +{"train_loss": 0.024510888382792473, "train_loss_bc": 0.02449045516550541, "train_loss_llm": 0.0020432770252227783, "grad_norm": 0.0986395925283432, "global_step": 32668, "epoch": 907, "lr": 0.001} +{"train_loss": 0.02778838761150837, "train_loss_bc": 0.02778838761150837, "train_loss_llm": 0.0, "grad_norm": 0.03703628107905388, "global_step": 32669, "epoch": 907, "lr": 0.001} +{"train_loss": 0.025994637981057167, "train_loss_bc": 0.025994637981057167, "train_loss_llm": 0.0, "grad_norm": 0.04800419881939888, "global_step": 32670, "epoch": 907, "lr": 0.001} +{"train_loss": 0.029972245916724205, "train_loss_bc": 0.029955491423606873, "train_loss_llm": 0.0016753673553466797, "grad_norm": 0.059488944709300995, "global_step": 32671, "epoch": 907, "lr": 0.001} +{"train_loss": 0.028622517362236977, "train_loss_bc": 0.028622517362236977, "train_loss_llm": 0.0, "grad_norm": 0.07995235919952393, "global_step": 32672, "epoch": 907, "lr": 0.001} +{"train_loss": 0.023476988077163696, "train_loss_bc": 0.023476988077163696, "train_loss_llm": 0.0, "grad_norm": 0.03684045374393463, "global_step": 32673, "epoch": 907, "lr": 0.001} +{"train_loss": 0.03805813938379288, "train_loss_bc": 0.03805813938379288, "train_loss_llm": 0.0, "grad_norm": 0.053167473524808884, "global_step": 32674, "epoch": 907, "lr": 0.001} +{"train_loss": 0.027425680309534073, "train_loss_bc": 0.027425680309534073, "train_loss_llm": 0.0, "grad_norm": 0.06255286186933517, "global_step": 32675, "epoch": 907, "lr": 0.001} +{"train_loss": 0.034371983259916306, "train_loss_bc": 0.034371983259916306, "train_loss_llm": 0.0, "grad_norm": 0.08567686378955841, "global_step": 32676, "epoch": 907, "lr": 0.001} +{"train_loss": 0.023531096056103706, "train_loss_bc": 0.023531096056103706, "train_loss_llm": 0.0, "grad_norm": 0.02934357151389122, "global_step": 32677, "epoch": 907, "lr": 0.001} +{"train_loss": 0.025051651522517204, "train_loss_bc": 0.025051651522517204, "train_loss_llm": 0.0, "grad_norm": 0.05423324927687645, "global_step": 32678, "epoch": 907, "lr": 0.001} +{"train_loss": 0.03531521186232567, "train_loss_bc": 0.03531521186232567, "train_loss_llm": 0.0, "grad_norm": 0.07396001368761063, "global_step": 32679, "epoch": 907, "lr": 0.001} +{"train_loss": 0.027933523058891296, "train_loss_bc": 0.027933523058891296, "train_loss_llm": 0.0, "grad_norm": 0.08923409134149551, "global_step": 32680, "epoch": 907, "lr": 0.001} +{"train_loss": 0.03184126317501068, "train_loss_bc": 0.03184126317501068, "train_loss_llm": 0.0, "grad_norm": 0.042912304401397705, "global_step": 32681, "epoch": 907, "lr": 0.001} +{"train_loss": 0.031585656106472015, "train_loss_bc": 0.03157690167427063, "train_loss_llm": 0.0008755326271057129, "grad_norm": 0.0573866069316864, "global_step": 32682, "epoch": 907, "lr": 0.001} +{"train_loss": 0.03151119872927666, "train_loss_bc": 0.03151119872927666, "train_loss_llm": 0.0, "grad_norm": 0.07510179281234741, "global_step": 32683, "epoch": 907, "lr": 0.001} +{"train_loss": 0.03253059834241867, "train_loss_bc": 0.03253059834241867, "train_loss_llm": 0.0, "grad_norm": 0.10511195659637451, "global_step": 32684, "epoch": 907, "lr": 0.001} +{"train_loss": 0.0288883987814188, "train_loss_bc": 0.0288883987814188, "train_loss_llm": 0.0, "grad_norm": 0.0376884751021862, "global_step": 32685, "epoch": 907, "lr": 0.001} +{"train_loss": 0.0275975801050663, "train_loss_bc": 0.0275975801050663, "train_loss_llm": 0.0, "grad_norm": 0.05397611856460571, "global_step": 32686, "epoch": 907, "lr": 0.001} +{"train_loss": 0.029227778160323698, "train_loss_bc": 0.030636044219136238, "train_loss_llm": 0.00038811564445495605, "grad_norm": 0.07954014092683792, "global_step": 32687, "epoch": 907, "lr": 0.001, "val_loss": 0.10365606099367142} +{"train_loss": 0.029560023918747902, "train_loss_bc": 0.029560023918747902, "train_loss_llm": 0.0, "grad_norm": 0.09001417458057404, "global_step": 32688, "epoch": 908, "lr": 0.001} +{"train_loss": 0.028114868327975273, "train_loss_bc": 0.028114868327975273, "train_loss_llm": 0.0, "grad_norm": 0.026741817593574524, "global_step": 32689, "epoch": 908, "lr": 0.001} +{"train_loss": 0.031553711742162704, "train_loss_bc": 0.031553711742162704, "train_loss_llm": 0.0, "grad_norm": 0.056462544947862625, "global_step": 32690, "epoch": 908, "lr": 0.001} +{"train_loss": 0.02801971696317196, "train_loss_bc": 0.02801971696317196, "train_loss_llm": 0.0, "grad_norm": 0.072026826441288, "global_step": 32691, "epoch": 908, "lr": 0.001} +{"train_loss": 0.02551605924963951, "train_loss_bc": 0.02551605924963951, "train_loss_llm": 0.0, "grad_norm": 0.09402906894683838, "global_step": 32692, "epoch": 908, "lr": 0.001} +{"train_loss": 0.023073984310030937, "train_loss_bc": 0.023073984310030937, "train_loss_llm": 0.0, "grad_norm": 0.03715836629271507, "global_step": 32693, "epoch": 908, "lr": 0.001} +{"train_loss": 0.030007002875208855, "train_loss_bc": 0.030007002875208855, "train_loss_llm": 0.0, "grad_norm": 0.046366333961486816, "global_step": 32694, "epoch": 908, "lr": 0.001} +{"train_loss": 0.02675583027303219, "train_loss_bc": 0.02674715593457222, "train_loss_llm": 0.0008675158023834229, "grad_norm": 0.0644841343164444, "global_step": 32695, "epoch": 908, "lr": 0.001} +{"train_loss": 0.033179786056280136, "train_loss_bc": 0.033179786056280136, "train_loss_llm": 0.0, "grad_norm": 0.08568242937326431, "global_step": 32696, "epoch": 908, "lr": 0.001} +{"train_loss": 0.028574416413903236, "train_loss_bc": 0.028574416413903236, "train_loss_llm": 0.0, "grad_norm": 0.03833658993244171, "global_step": 32697, "epoch": 908, "lr": 0.001} +{"train_loss": 0.029579535126686096, "train_loss_bc": 0.029579535126686096, "train_loss_llm": 0.0, "grad_norm": 0.05996980145573616, "global_step": 32698, "epoch": 908, "lr": 0.001} +{"train_loss": 0.028881635516881943, "train_loss_bc": 0.028879482299089432, "train_loss_llm": 0.00021535158157348633, "grad_norm": 0.07975944131612778, "global_step": 32699, "epoch": 908, "lr": 0.001} +{"train_loss": 0.029270555824041367, "train_loss_bc": 0.029270555824041367, "train_loss_llm": 0.0, "grad_norm": 0.10446840524673462, "global_step": 32700, "epoch": 908, "lr": 0.001} +{"train_loss": 0.033692214637994766, "train_loss_bc": 0.033684905618429184, "train_loss_llm": 0.0007307529449462891, "grad_norm": 0.04135842248797417, "global_step": 32701, "epoch": 908, "lr": 0.001} +{"train_loss": 0.03222280740737915, "train_loss_bc": 0.03222280740737915, "train_loss_llm": 0.0, "grad_norm": 0.08486644178628922, "global_step": 32702, "epoch": 908, "lr": 0.001} +{"train_loss": 0.022477757185697556, "train_loss_bc": 0.022477757185697556, "train_loss_llm": 0.0, "grad_norm": 0.09082364290952682, "global_step": 32703, "epoch": 908, "lr": 0.001} +{"train_loss": 0.030721908435225487, "train_loss_bc": 0.030721908435225487, "train_loss_llm": 0.0, "grad_norm": 0.10635031014680862, "global_step": 32704, "epoch": 908, "lr": 0.001} +{"train_loss": 0.028557436540722847, "train_loss_bc": 0.028557436540722847, "train_loss_llm": 0.0, "grad_norm": 0.03287293016910553, "global_step": 32705, "epoch": 908, "lr": 0.001} +{"train_loss": 0.027187688276171684, "train_loss_bc": 0.027187688276171684, "train_loss_llm": 0.0, "grad_norm": 0.051435261964797974, "global_step": 32706, "epoch": 908, "lr": 0.001} +{"train_loss": 0.03068505972623825, "train_loss_bc": 0.030676761642098427, "train_loss_llm": 0.000829845666885376, "grad_norm": 0.06621430814266205, "global_step": 32707, "epoch": 908, "lr": 0.001} +{"train_loss": 0.028005000203847885, "train_loss_bc": 0.028005000203847885, "train_loss_llm": 0.0, "grad_norm": 0.09042370319366455, "global_step": 32708, "epoch": 908, "lr": 0.001} +{"train_loss": 0.03265900909900665, "train_loss_bc": 0.03265900909900665, "train_loss_llm": 0.0, "grad_norm": 0.039054177701473236, "global_step": 32709, "epoch": 908, "lr": 0.001} +{"train_loss": 0.027321139350533485, "train_loss_bc": 0.027308575809001923, "train_loss_llm": 0.0012563467025756836, "grad_norm": 0.06254158914089203, "global_step": 32710, "epoch": 908, "lr": 0.001} +{"train_loss": 0.031617533415555954, "train_loss_bc": 0.031617533415555954, "train_loss_llm": 0.0, "grad_norm": 0.0789555013179779, "global_step": 32711, "epoch": 908, "lr": 0.001} +{"train_loss": 0.030156999826431274, "train_loss_bc": 0.030156999826431274, "train_loss_llm": 0.0, "grad_norm": 0.10263881087303162, "global_step": 32712, "epoch": 908, "lr": 0.001} +{"train_loss": 0.029173118993639946, "train_loss_bc": 0.029169294983148575, "train_loss_llm": 0.00038245320320129395, "grad_norm": 0.0326780304312706, "global_step": 32713, "epoch": 908, "lr": 0.001} +{"train_loss": 0.029118476435542107, "train_loss_bc": 0.029118476435542107, "train_loss_llm": 0.0, "grad_norm": 0.049410589039325714, "global_step": 32714, "epoch": 908, "lr": 0.001} +{"train_loss": 0.030234510079026222, "train_loss_bc": 0.030225271359086037, "train_loss_llm": 0.0009239614009857178, "grad_norm": 0.071114182472229, "global_step": 32715, "epoch": 908, "lr": 0.001} +{"train_loss": 0.02973615750670433, "train_loss_bc": 0.02970842644572258, "train_loss_llm": 0.0027731657028198242, "grad_norm": 0.09351957589387894, "global_step": 32716, "epoch": 908, "lr": 0.001} +{"train_loss": 0.03134598210453987, "train_loss_bc": 0.03134598210453987, "train_loss_llm": 0.0, "grad_norm": 0.0448116697371006, "global_step": 32717, "epoch": 908, "lr": 0.001} +{"train_loss": 0.02862110361456871, "train_loss_bc": 0.02861984632909298, "train_loss_llm": 0.00012570619583129883, "grad_norm": 0.06741104274988174, "global_step": 32718, "epoch": 908, "lr": 0.001} +{"train_loss": 0.02611328847706318, "train_loss_bc": 0.02611328847706318, "train_loss_llm": 0.0, "grad_norm": 0.07788532227277756, "global_step": 32719, "epoch": 908, "lr": 0.001} +{"train_loss": 0.03419313579797745, "train_loss_bc": 0.03419313579797745, "train_loss_llm": 0.0, "grad_norm": 0.09588479995727539, "global_step": 32720, "epoch": 908, "lr": 0.001} +{"train_loss": 0.030240802094340324, "train_loss_bc": 0.030239686369895935, "train_loss_llm": 0.00011160969734191895, "grad_norm": 0.0526026114821434, "global_step": 32721, "epoch": 908, "lr": 0.001} +{"train_loss": 0.026222964748740196, "train_loss_bc": 0.026222964748740196, "train_loss_llm": 0.0, "grad_norm": 0.07891654968261719, "global_step": 32722, "epoch": 908, "lr": 0.001} +{"train_loss": 0.02923029733614789, "train_loss_bc": 0.029899483546614647, "train_loss_llm": 0.0, "grad_norm": 0.10092422366142273, "global_step": 32723, "epoch": 908, "lr": 0.001, "val_loss": 0.10461843013763428} +{"train_loss": 0.030160116031765938, "train_loss_bc": 0.030160116031765938, "train_loss_llm": 0.0, "grad_norm": 0.11482442915439606, "global_step": 32724, "epoch": 909, "lr": 0.001} +{"train_loss": 0.028268609195947647, "train_loss_bc": 0.028260771185159683, "train_loss_llm": 0.0007838904857635498, "grad_norm": 0.04157334566116333, "global_step": 32725, "epoch": 909, "lr": 0.001} +{"train_loss": 0.027167469263076782, "train_loss_bc": 0.027167469263076782, "train_loss_llm": 0.0, "grad_norm": 0.06982056051492691, "global_step": 32726, "epoch": 909, "lr": 0.001} +{"train_loss": 0.03303220495581627, "train_loss_bc": 0.03303220495581627, "train_loss_llm": 0.0, "grad_norm": 0.10670674592256546, "global_step": 32727, "epoch": 909, "lr": 0.001} +{"train_loss": 0.025421807542443275, "train_loss_bc": 0.025421807542443275, "train_loss_llm": 0.0, "grad_norm": 0.12644395232200623, "global_step": 32728, "epoch": 909, "lr": 0.001} +{"train_loss": 0.025489071384072304, "train_loss_bc": 0.025479234755039215, "train_loss_llm": 0.0009837150573730469, "grad_norm": 0.02790755219757557, "global_step": 32729, "epoch": 909, "lr": 0.001} +{"train_loss": 0.03202030807733536, "train_loss_bc": 0.03202030807733536, "train_loss_llm": 0.0, "grad_norm": 0.04785554111003876, "global_step": 32730, "epoch": 909, "lr": 0.001} +{"train_loss": 0.02317998744547367, "train_loss_bc": 0.02317998744547367, "train_loss_llm": 0.0, "grad_norm": 0.06067753583192825, "global_step": 32731, "epoch": 909, "lr": 0.001} +{"train_loss": 0.026452191174030304, "train_loss_bc": 0.026447705924510956, "train_loss_llm": 0.00044852495193481445, "grad_norm": 0.06965027749538422, "global_step": 32732, "epoch": 909, "lr": 0.001} +{"train_loss": 0.03070060722529888, "train_loss_bc": 0.03070060722529888, "train_loss_llm": 0.0, "grad_norm": 0.0328778438270092, "global_step": 32733, "epoch": 909, "lr": 0.001} +{"train_loss": 0.02554047293961048, "train_loss_bc": 0.02554047293961048, "train_loss_llm": 0.0, "grad_norm": 0.04658244922757149, "global_step": 32734, "epoch": 909, "lr": 0.001} +{"train_loss": 0.029232723638415337, "train_loss_bc": 0.029232723638415337, "train_loss_llm": 0.0, "grad_norm": 0.06747356057167053, "global_step": 32735, "epoch": 909, "lr": 0.001} +{"train_loss": 0.026490261778235435, "train_loss_bc": 0.026490261778235435, "train_loss_llm": 0.0, "grad_norm": 0.09484193474054337, "global_step": 32736, "epoch": 909, "lr": 0.001} +{"train_loss": 0.02692337892949581, "train_loss_bc": 0.02692337892949581, "train_loss_llm": 0.0, "grad_norm": 0.036098312586545944, "global_step": 32737, "epoch": 909, "lr": 0.001} +{"train_loss": 0.03575197234749794, "train_loss_bc": 0.03575197234749794, "train_loss_llm": 0.0, "grad_norm": 0.0644286647439003, "global_step": 32738, "epoch": 909, "lr": 0.001} +{"train_loss": 0.03226923197507858, "train_loss_bc": 0.03226923197507858, "train_loss_llm": 0.0, "grad_norm": 0.1169954165816307, "global_step": 32739, "epoch": 909, "lr": 0.001} +{"train_loss": 0.03558070585131645, "train_loss_bc": 0.03558070585131645, "train_loss_llm": 0.0, "grad_norm": 0.1368323564529419, "global_step": 32740, "epoch": 909, "lr": 0.001} +{"train_loss": 0.030614711344242096, "train_loss_bc": 0.030614711344242096, "train_loss_llm": 0.0, "grad_norm": 0.04642389342188835, "global_step": 32741, "epoch": 909, "lr": 0.001} +{"train_loss": 0.03133223205804825, "train_loss_bc": 0.03132662922143936, "train_loss_llm": 0.0005603134632110596, "grad_norm": 0.062053341418504715, "global_step": 32742, "epoch": 909, "lr": 0.001} +{"train_loss": 0.02752259001135826, "train_loss_bc": 0.02752259001135826, "train_loss_llm": 0.0, "grad_norm": 0.08251911401748657, "global_step": 32743, "epoch": 909, "lr": 0.001} +{"train_loss": 0.028243357315659523, "train_loss_bc": 0.028243357315659523, "train_loss_llm": 0.0, "grad_norm": 0.09730546921491623, "global_step": 32744, "epoch": 909, "lr": 0.001} +{"train_loss": 0.03379543125629425, "train_loss_bc": 0.03379543125629425, "train_loss_llm": 0.0, "grad_norm": 0.03301538899540901, "global_step": 32745, "epoch": 909, "lr": 0.001} +{"train_loss": 0.029303936287760735, "train_loss_bc": 0.029303936287760735, "train_loss_llm": 0.0, "grad_norm": 0.0568113774061203, "global_step": 32746, "epoch": 909, "lr": 0.001} +{"train_loss": 0.033207040280103683, "train_loss_bc": 0.033207040280103683, "train_loss_llm": 0.0, "grad_norm": 0.0704905167222023, "global_step": 32747, "epoch": 909, "lr": 0.001} +{"train_loss": 0.032888639718294144, "train_loss_bc": 0.03287533298134804, "train_loss_llm": 0.0013307034969329834, "grad_norm": 0.0840253084897995, "global_step": 32748, "epoch": 909, "lr": 0.001} +{"train_loss": 0.030730407685041428, "train_loss_bc": 0.030730407685041428, "train_loss_llm": 0.0, "grad_norm": 0.03614194318652153, "global_step": 32749, "epoch": 909, "lr": 0.001} +{"train_loss": 0.03239348903298378, "train_loss_bc": 0.03239348903298378, "train_loss_llm": 0.0, "grad_norm": 0.06566635519266129, "global_step": 32750, "epoch": 909, "lr": 0.001} +{"train_loss": 0.022900013253092766, "train_loss_bc": 0.022900013253092766, "train_loss_llm": 0.0, "grad_norm": 0.06934289634227753, "global_step": 32751, "epoch": 909, "lr": 0.001} +{"train_loss": 0.028981374576687813, "train_loss_bc": 0.028981374576687813, "train_loss_llm": 0.0, "grad_norm": 0.08114685863256454, "global_step": 32752, "epoch": 909, "lr": 0.001} +{"train_loss": 0.02405666932463646, "train_loss_bc": 0.024056240916252136, "train_loss_llm": 4.291534423828125e-05, "grad_norm": 0.03733902797102928, "global_step": 32753, "epoch": 909, "lr": 0.001} +{"train_loss": 0.028843730688095093, "train_loss_bc": 0.028843730688095093, "train_loss_llm": 0.0, "grad_norm": 0.06647169589996338, "global_step": 32754, "epoch": 909, "lr": 0.001} +{"train_loss": 0.029216788709163666, "train_loss_bc": 0.029215829446911812, "train_loss_llm": 9.587407112121582e-05, "grad_norm": 0.09109077602624893, "global_step": 32755, "epoch": 909, "lr": 0.001} +{"train_loss": 0.027225956320762634, "train_loss_bc": 0.027225956320762634, "train_loss_llm": 0.0, "grad_norm": 0.10753816366195679, "global_step": 32756, "epoch": 909, "lr": 0.001} +{"train_loss": 0.026170866563916206, "train_loss_bc": 0.026170866563916206, "train_loss_llm": 0.0, "grad_norm": 0.0367591492831707, "global_step": 32757, "epoch": 909, "lr": 0.001} +{"train_loss": 0.025874542072415352, "train_loss_bc": 0.025866016745567322, "train_loss_llm": 0.0008525252342224121, "grad_norm": 0.05832645297050476, "global_step": 32758, "epoch": 909, "lr": 0.001} +{"train_loss": 0.02920291546939148, "train_loss_bc": 0.03432206064462662, "train_loss_llm": 0.0, "grad_norm": 0.08372684568166733, "global_step": 32759, "epoch": 909, "lr": 0.001, "val_loss": 0.10407847166061401} +{"train_loss": 0.02864666096866131, "train_loss_bc": 0.02864666096866131, "train_loss_llm": 0.0, "grad_norm": 0.09850259870290756, "global_step": 32760, "epoch": 910, "lr": 0.001} +{"train_loss": 0.02966771461069584, "train_loss_bc": 0.02966771461069584, "train_loss_llm": 0.0, "grad_norm": 0.03909113630652428, "global_step": 32761, "epoch": 910, "lr": 0.001} +{"train_loss": 0.0295957513153553, "train_loss_bc": 0.0295957513153553, "train_loss_llm": 0.0, "grad_norm": 0.06844160705804825, "global_step": 32762, "epoch": 910, "lr": 0.001} +{"train_loss": 0.027494926005601883, "train_loss_bc": 0.027494926005601883, "train_loss_llm": 0.0, "grad_norm": 0.09355335682630539, "global_step": 32763, "epoch": 910, "lr": 0.001} +{"train_loss": 0.033066317439079285, "train_loss_bc": 0.033066317439079285, "train_loss_llm": 0.0, "grad_norm": 0.1291789710521698, "global_step": 32764, "epoch": 910, "lr": 0.001} +{"train_loss": 0.02968783490359783, "train_loss_bc": 0.029657363891601562, "train_loss_llm": 0.003047078847885132, "grad_norm": 0.0314960703253746, "global_step": 32765, "epoch": 910, "lr": 0.001} +{"train_loss": 0.029126960784196854, "train_loss_bc": 0.029126960784196854, "train_loss_llm": 0.0, "grad_norm": 0.055349163711071014, "global_step": 32766, "epoch": 910, "lr": 0.001} +{"train_loss": 0.025493305176496506, "train_loss_bc": 0.025493305176496506, "train_loss_llm": 0.0, "grad_norm": 0.07229270786046982, "global_step": 32767, "epoch": 910, "lr": 0.001} +{"train_loss": 0.03167891502380371, "train_loss_bc": 0.03167891502380371, "train_loss_llm": 0.0, "grad_norm": 0.089076928794384, "global_step": 32768, "epoch": 910, "lr": 0.001} +{"train_loss": 0.028502190485596657, "train_loss_bc": 0.028499606996774673, "train_loss_llm": 0.0002582967281341553, "grad_norm": 0.04343041405081749, "global_step": 32769, "epoch": 910, "lr": 0.001} +{"train_loss": 0.03276483714580536, "train_loss_bc": 0.03276483714580536, "train_loss_llm": 0.0, "grad_norm": 0.06974419951438904, "global_step": 32770, "epoch": 910, "lr": 0.001} +{"train_loss": 0.03164244443178177, "train_loss_bc": 0.03164244443178177, "train_loss_llm": 0.0, "grad_norm": 0.10427913814783096, "global_step": 32771, "epoch": 910, "lr": 0.001} +{"train_loss": 0.03274542838335037, "train_loss_bc": 0.03274542838335037, "train_loss_llm": 0.0, "grad_norm": 0.10994154959917068, "global_step": 32772, "epoch": 910, "lr": 0.001} +{"train_loss": 0.02116251550614834, "train_loss_bc": 0.02116103656589985, "train_loss_llm": 0.00014790892601013184, "grad_norm": 0.030028607696294785, "global_step": 32773, "epoch": 910, "lr": 0.001} +{"train_loss": 0.032282762229442596, "train_loss_bc": 0.03226947784423828, "train_loss_llm": 0.00132828950881958, "grad_norm": 0.050771962851285934, "global_step": 32774, "epoch": 910, "lr": 0.001} +{"train_loss": 0.028051327913999557, "train_loss_bc": 0.028044654056429863, "train_loss_llm": 0.0006674230098724365, "grad_norm": 0.06437142193317413, "global_step": 32775, "epoch": 910, "lr": 0.001} +{"train_loss": 0.027604201808571815, "train_loss_bc": 0.027604201808571815, "train_loss_llm": 0.0, "grad_norm": 0.07806646078824997, "global_step": 32776, "epoch": 910, "lr": 0.001} +{"train_loss": 0.029287593439221382, "train_loss_bc": 0.029287593439221382, "train_loss_llm": 0.0, "grad_norm": 0.0476672425866127, "global_step": 32777, "epoch": 910, "lr": 0.001} +{"train_loss": 0.02675880491733551, "train_loss_bc": 0.026750098913908005, "train_loss_llm": 0.0008705258369445801, "grad_norm": 0.07523001730442047, "global_step": 32778, "epoch": 910, "lr": 0.001} +{"train_loss": 0.02600836008787155, "train_loss_bc": 0.02600836008787155, "train_loss_llm": 0.0, "grad_norm": 0.10189466923475266, "global_step": 32779, "epoch": 910, "lr": 0.001} +{"train_loss": 0.0318981297314167, "train_loss_bc": 0.03189276158809662, "train_loss_llm": 0.0005367696285247803, "grad_norm": 0.12249889224767685, "global_step": 32780, "epoch": 910, "lr": 0.001} +{"train_loss": 0.03363276645541191, "train_loss_bc": 0.03363276645541191, "train_loss_llm": 0.0, "grad_norm": 0.03748319670557976, "global_step": 32781, "epoch": 910, "lr": 0.001} +{"train_loss": 0.025308549404144287, "train_loss_bc": 0.025308549404144287, "train_loss_llm": 0.0, "grad_norm": 0.05452731251716614, "global_step": 32782, "epoch": 910, "lr": 0.001} +{"train_loss": 0.02914826199412346, "train_loss_bc": 0.02914826199412346, "train_loss_llm": 0.0, "grad_norm": 0.0791717916727066, "global_step": 32783, "epoch": 910, "lr": 0.001} +{"train_loss": 0.028824618086218834, "train_loss_bc": 0.028824618086218834, "train_loss_llm": 0.0, "grad_norm": 0.1000736653804779, "global_step": 32784, "epoch": 910, "lr": 0.001} +{"train_loss": 0.03138132765889168, "train_loss_bc": 0.03138132765889168, "train_loss_llm": 0.0, "grad_norm": 0.036581236869096756, "global_step": 32785, "epoch": 910, "lr": 0.001} +{"train_loss": 0.025293368846178055, "train_loss_bc": 0.025293368846178055, "train_loss_llm": 0.0, "grad_norm": 0.0506453812122345, "global_step": 32786, "epoch": 910, "lr": 0.001} +{"train_loss": 0.02474066987633705, "train_loss_bc": 0.024726424366235733, "train_loss_llm": 0.001424640417098999, "grad_norm": 0.06959709525108337, "global_step": 32787, "epoch": 910, "lr": 0.001} +{"train_loss": 0.025987491011619568, "train_loss_bc": 0.025987491011619568, "train_loss_llm": 0.0, "grad_norm": 0.08747735619544983, "global_step": 32788, "epoch": 910, "lr": 0.001} +{"train_loss": 0.031869154423475266, "train_loss_bc": 0.031869154423475266, "train_loss_llm": 0.0, "grad_norm": 0.03646568953990936, "global_step": 32789, "epoch": 910, "lr": 0.001} +{"train_loss": 0.03037324920296669, "train_loss_bc": 0.03037324920296669, "train_loss_llm": 0.0, "grad_norm": 0.05363275855779648, "global_step": 32790, "epoch": 910, "lr": 0.001} +{"train_loss": 0.03199153393507004, "train_loss_bc": 0.03199153393507004, "train_loss_llm": 0.0, "grad_norm": 0.07487958669662476, "global_step": 32791, "epoch": 910, "lr": 0.001} +{"train_loss": 0.031142476946115494, "train_loss_bc": 0.031142476946115494, "train_loss_llm": 0.0, "grad_norm": 0.0887344628572464, "global_step": 32792, "epoch": 910, "lr": 0.001} +{"train_loss": 0.027804508805274963, "train_loss_bc": 0.027804508805274963, "train_loss_llm": 0.0, "grad_norm": 0.03664973005652428, "global_step": 32793, "epoch": 910, "lr": 0.001} +{"train_loss": 0.03154534101486206, "train_loss_bc": 0.03154534101486206, "train_loss_llm": 0.0, "grad_norm": 0.05446342006325722, "global_step": 32794, "epoch": 910, "lr": 0.001} +{"train_loss": 0.029069390665325854, "train_loss_bc": 0.024276655167341232, "train_loss_llm": 0.0011109113693237305, "grad_norm": 0.08082664012908936, "global_step": 32795, "epoch": 910, "lr": 0.001, "train/cumulative_reward": 83.37788196228374, "train/mean_score": 14.830524307271807, "train/success_rate": 0.3, "test/cumulative_reward": 35.60964009414618, "test/mean_score": 8.61824542311545, "test/success_rate": 0.22, "val_loss": 0.10581333935260773, "train_action_mse_error": 0.032612815499305725} +{"train_loss": 0.029029930010437965, "train_loss_bc": 0.02901286818087101, "train_loss_llm": 0.0017062127590179443, "grad_norm": 0.09991364926099777, "global_step": 32796, "epoch": 911, "lr": 0.001} +{"train_loss": 0.028890468180179596, "train_loss_bc": 0.028890468180179596, "train_loss_llm": 0.0, "grad_norm": 0.03512635454535484, "global_step": 32797, "epoch": 911, "lr": 0.001} +{"train_loss": 0.026154588907957077, "train_loss_bc": 0.026154588907957077, "train_loss_llm": 0.0, "grad_norm": 0.04605734348297119, "global_step": 32798, "epoch": 911, "lr": 0.001} +{"train_loss": 0.027979185804724693, "train_loss_bc": 0.027979185804724693, "train_loss_llm": 0.0, "grad_norm": 0.05256124213337898, "global_step": 32799, "epoch": 911, "lr": 0.001} +{"train_loss": 0.032148972153663635, "train_loss_bc": 0.03214133903384209, "train_loss_llm": 0.0007633566856384277, "grad_norm": 0.06620234996080399, "global_step": 32800, "epoch": 911, "lr": 0.001} +{"train_loss": 0.032534077763557434, "train_loss_bc": 0.032534077763557434, "train_loss_llm": 0.0, "grad_norm": 0.043230753391981125, "global_step": 32801, "epoch": 911, "lr": 0.001} +{"train_loss": 0.0283344853669405, "train_loss_bc": 0.02831953391432762, "train_loss_llm": 0.0014951825141906738, "grad_norm": 0.0633566752076149, "global_step": 32802, "epoch": 911, "lr": 0.001} +{"train_loss": 0.03349272534251213, "train_loss_bc": 0.03349272534251213, "train_loss_llm": 0.0, "grad_norm": 0.0825020894408226, "global_step": 32803, "epoch": 911, "lr": 0.001} +{"train_loss": 0.025883490219712257, "train_loss_bc": 0.02586686797440052, "train_loss_llm": 0.0016621947288513184, "grad_norm": 0.09960366785526276, "global_step": 32804, "epoch": 911, "lr": 0.001} +{"train_loss": 0.028590572997927666, "train_loss_bc": 0.028590572997927666, "train_loss_llm": 0.0, "grad_norm": 0.03026394732296467, "global_step": 32805, "epoch": 911, "lr": 0.001} +{"train_loss": 0.031064463779330254, "train_loss_bc": 0.031064463779330254, "train_loss_llm": 0.0, "grad_norm": 0.04979533702135086, "global_step": 32806, "epoch": 911, "lr": 0.001} +{"train_loss": 0.028988061472773552, "train_loss_bc": 0.028988061472773552, "train_loss_llm": 0.0, "grad_norm": 0.07051217555999756, "global_step": 32807, "epoch": 911, "lr": 0.001} +{"train_loss": 0.030463427305221558, "train_loss_bc": 0.03045964427292347, "train_loss_llm": 0.0003782212734222412, "grad_norm": 0.09529311209917068, "global_step": 32808, "epoch": 911, "lr": 0.001} +{"train_loss": 0.028273530304431915, "train_loss_bc": 0.028273530304431915, "train_loss_llm": 0.0, "grad_norm": 0.042528800666332245, "global_step": 32809, "epoch": 911, "lr": 0.001} +{"train_loss": 0.025554094463586807, "train_loss_bc": 0.025554094463586807, "train_loss_llm": 0.0, "grad_norm": 0.061820268630981445, "global_step": 32810, "epoch": 911, "lr": 0.001} +{"train_loss": 0.03187207505106926, "train_loss_bc": 0.03187207505106926, "train_loss_llm": 0.0, "grad_norm": 0.08985144644975662, "global_step": 32811, "epoch": 911, "lr": 0.001} +{"train_loss": 0.032375749200582504, "train_loss_bc": 0.032375749200582504, "train_loss_llm": 0.0, "grad_norm": 0.10431895405054092, "global_step": 32812, "epoch": 911, "lr": 0.001} +{"train_loss": 0.03429803252220154, "train_loss_bc": 0.03429803252220154, "train_loss_llm": 0.0, "grad_norm": 0.04256482049822807, "global_step": 32813, "epoch": 911, "lr": 0.001} +{"train_loss": 0.023749612271785736, "train_loss_bc": 0.023747114464640617, "train_loss_llm": 0.0002497732639312744, "grad_norm": 0.06303757429122925, "global_step": 32814, "epoch": 911, "lr": 0.001} +{"train_loss": 0.03681233897805214, "train_loss_bc": 0.03681233897805214, "train_loss_llm": 0.0, "grad_norm": 0.09268253296613693, "global_step": 32815, "epoch": 911, "lr": 0.001} +{"train_loss": 0.026438511908054352, "train_loss_bc": 0.026438511908054352, "train_loss_llm": 0.0, "grad_norm": 0.11084616929292679, "global_step": 32816, "epoch": 911, "lr": 0.001} +{"train_loss": 0.023961883038282394, "train_loss_bc": 0.023961883038282394, "train_loss_llm": 0.0, "grad_norm": 0.027227535843849182, "global_step": 32817, "epoch": 911, "lr": 0.001} +{"train_loss": 0.03185892105102539, "train_loss_bc": 0.03185892105102539, "train_loss_llm": 0.0, "grad_norm": 0.04265245050191879, "global_step": 32818, "epoch": 911, "lr": 0.001} +{"train_loss": 0.024809816852211952, "train_loss_bc": 0.024809816852211952, "train_loss_llm": 0.0, "grad_norm": 0.052469298243522644, "global_step": 32819, "epoch": 911, "lr": 0.001} +{"train_loss": 0.03002668172121048, "train_loss_bc": 0.03002668172121048, "train_loss_llm": 0.0, "grad_norm": 0.0613594725728035, "global_step": 32820, "epoch": 911, "lr": 0.001} +{"train_loss": 0.030733050778508186, "train_loss_bc": 0.030729983001947403, "train_loss_llm": 0.00030681490898132324, "grad_norm": 0.04495588317513466, "global_step": 32821, "epoch": 911, "lr": 0.001} +{"train_loss": 0.02683107927441597, "train_loss_bc": 0.02683107927441597, "train_loss_llm": 0.0, "grad_norm": 0.0514124371111393, "global_step": 32822, "epoch": 911, "lr": 0.001} +{"train_loss": 0.024417847394943237, "train_loss_bc": 0.024417847394943237, "train_loss_llm": 0.0, "grad_norm": 0.06265527009963989, "global_step": 32823, "epoch": 911, "lr": 0.001} +{"train_loss": 0.027440963312983513, "train_loss_bc": 0.027440963312983513, "train_loss_llm": 0.0, "grad_norm": 0.09607572108507156, "global_step": 32824, "epoch": 911, "lr": 0.001} +{"train_loss": 0.0338677279651165, "train_loss_bc": 0.0338677279651165, "train_loss_llm": 0.0, "grad_norm": 0.04649975895881653, "global_step": 32825, "epoch": 911, "lr": 0.001} +{"train_loss": 0.030576365068554878, "train_loss_bc": 0.030576365068554878, "train_loss_llm": 0.0, "grad_norm": 0.07376076281070709, "global_step": 32826, "epoch": 911, "lr": 0.001} +{"train_loss": 0.025095157325267792, "train_loss_bc": 0.025095157325267792, "train_loss_llm": 0.0, "grad_norm": 0.09998951107263565, "global_step": 32827, "epoch": 911, "lr": 0.001} +{"train_loss": 0.032864294946193695, "train_loss_bc": 0.03286375850439072, "train_loss_llm": 5.3495168685913086e-05, "grad_norm": 0.13027791678905487, "global_step": 32828, "epoch": 911, "lr": 0.001} +{"train_loss": 0.029770243912935257, "train_loss_bc": 0.029770243912935257, "train_loss_llm": 0.0, "grad_norm": 0.04158636927604675, "global_step": 32829, "epoch": 911, "lr": 0.001} +{"train_loss": 0.024269280955195427, "train_loss_bc": 0.02426527440547943, "train_loss_llm": 0.0004006922245025635, "grad_norm": 0.06295738369226456, "global_step": 32830, "epoch": 911, "lr": 0.001} +{"train_loss": 0.029121025775869686, "train_loss_bc": 0.028905220329761505, "train_loss_llm": 0.0, "grad_norm": 0.0906507670879364, "global_step": 32831, "epoch": 911, "lr": 0.001, "val_loss": 0.10438162833452225} +{"train_loss": 0.03358425572514534, "train_loss_bc": 0.03358425572514534, "train_loss_llm": 0.0, "grad_norm": 0.10996676236391068, "global_step": 32832, "epoch": 912, "lr": 0.001} +{"train_loss": 0.028737228363752365, "train_loss_bc": 0.028737228363752365, "train_loss_llm": 0.0, "grad_norm": 0.03144275024533272, "global_step": 32833, "epoch": 912, "lr": 0.001} +{"train_loss": 0.02670525386929512, "train_loss_bc": 0.02670525386929512, "train_loss_llm": 0.0, "grad_norm": 0.04643869400024414, "global_step": 32834, "epoch": 912, "lr": 0.001} +{"train_loss": 0.027274779975414276, "train_loss_bc": 0.027274779975414276, "train_loss_llm": 0.0, "grad_norm": 0.06911186128854752, "global_step": 32835, "epoch": 912, "lr": 0.001} +{"train_loss": 0.025556430220603943, "train_loss_bc": 0.025556430220603943, "train_loss_llm": 0.0, "grad_norm": 0.08446811884641647, "global_step": 32836, "epoch": 912, "lr": 0.001} +{"train_loss": 0.030422816053032875, "train_loss_bc": 0.030422816053032875, "train_loss_llm": 0.0, "grad_norm": 0.05010080710053444, "global_step": 32837, "epoch": 912, "lr": 0.001} +{"train_loss": 0.030387908220291138, "train_loss_bc": 0.0303842443972826, "train_loss_llm": 0.00036638975143432617, "grad_norm": 0.06989716738462448, "global_step": 32838, "epoch": 912, "lr": 0.001} +{"train_loss": 0.02540450170636177, "train_loss_bc": 0.02540450170636177, "train_loss_llm": 0.0, "grad_norm": 0.08612146228551865, "global_step": 32839, "epoch": 912, "lr": 0.001} +{"train_loss": 0.03142913803458214, "train_loss_bc": 0.03142913803458214, "train_loss_llm": 0.0, "grad_norm": 0.11029373854398727, "global_step": 32840, "epoch": 912, "lr": 0.001} +{"train_loss": 0.025573549792170525, "train_loss_bc": 0.025573549792170525, "train_loss_llm": 0.0, "grad_norm": 0.040426820516586304, "global_step": 32841, "epoch": 912, "lr": 0.001} +{"train_loss": 0.027978751808404922, "train_loss_bc": 0.027978751808404922, "train_loss_llm": 0.0, "grad_norm": 0.06308189034461975, "global_step": 32842, "epoch": 912, "lr": 0.001} +{"train_loss": 0.028223110362887383, "train_loss_bc": 0.028220877051353455, "train_loss_llm": 0.00022327899932861328, "grad_norm": 0.07000908255577087, "global_step": 32843, "epoch": 912, "lr": 0.001} +{"train_loss": 0.03190385922789574, "train_loss_bc": 0.03190385922789574, "train_loss_llm": 0.0, "grad_norm": 0.0867854431271553, "global_step": 32844, "epoch": 912, "lr": 0.001} +{"train_loss": 0.03215888515114784, "train_loss_bc": 0.03215400129556656, "train_loss_llm": 0.0004883706569671631, "grad_norm": 0.034310098737478256, "global_step": 32845, "epoch": 912, "lr": 0.001} +{"train_loss": 0.03294513747096062, "train_loss_bc": 0.03294513747096062, "train_loss_llm": 0.0, "grad_norm": 0.05198444426059723, "global_step": 32846, "epoch": 912, "lr": 0.001} +{"train_loss": 0.03120548278093338, "train_loss_bc": 0.03119300678372383, "train_loss_llm": 0.0012476742267608643, "grad_norm": 0.07665364444255829, "global_step": 32847, "epoch": 912, "lr": 0.001} +{"train_loss": 0.02785056084394455, "train_loss_bc": 0.02784302644431591, "train_loss_llm": 0.0007534027099609375, "grad_norm": 0.09837833046913147, "global_step": 32848, "epoch": 912, "lr": 0.001} +{"train_loss": 0.028458870947360992, "train_loss_bc": 0.028458870947360992, "train_loss_llm": 0.0, "grad_norm": 0.041715171188116074, "global_step": 32849, "epoch": 912, "lr": 0.001} +{"train_loss": 0.028795672580599785, "train_loss_bc": 0.028795672580599785, "train_loss_llm": 0.0, "grad_norm": 0.06839069724082947, "global_step": 32850, "epoch": 912, "lr": 0.001} +{"train_loss": 0.032419923692941666, "train_loss_bc": 0.03239915892481804, "train_loss_llm": 0.002076566219329834, "grad_norm": 0.09922666847705841, "global_step": 32851, "epoch": 912, "lr": 0.001} +{"train_loss": 0.028065083548426628, "train_loss_bc": 0.028065083548426628, "train_loss_llm": 0.0, "grad_norm": 0.10886181890964508, "global_step": 32852, "epoch": 912, "lr": 0.001} +{"train_loss": 0.02757311798632145, "train_loss_bc": 0.02757311798632145, "train_loss_llm": 0.0, "grad_norm": 0.043479908257722855, "global_step": 32853, "epoch": 912, "lr": 0.001} +{"train_loss": 0.02535816840827465, "train_loss_bc": 0.025357436388731003, "train_loss_llm": 7.325410842895508e-05, "grad_norm": 0.055295608937740326, "global_step": 32854, "epoch": 912, "lr": 0.001} +{"train_loss": 0.02848261035978794, "train_loss_bc": 0.02848261035978794, "train_loss_llm": 0.0, "grad_norm": 0.07219456136226654, "global_step": 32855, "epoch": 912, "lr": 0.001} +{"train_loss": 0.029338885098695755, "train_loss_bc": 0.029338885098695755, "train_loss_llm": 0.0, "grad_norm": 0.09527278691530228, "global_step": 32856, "epoch": 912, "lr": 0.001} +{"train_loss": 0.03248116746544838, "train_loss_bc": 0.03247194364666939, "train_loss_llm": 0.0009222328662872314, "grad_norm": 0.04804500564932823, "global_step": 32857, "epoch": 912, "lr": 0.001} +{"train_loss": 0.0309038907289505, "train_loss_bc": 0.030900293961167336, "train_loss_llm": 0.00035965442657470703, "grad_norm": 0.05830960348248482, "global_step": 32858, "epoch": 912, "lr": 0.001} +{"train_loss": 0.032100945711135864, "train_loss_bc": 0.032100945711135864, "train_loss_llm": 0.0, "grad_norm": 0.08088015764951706, "global_step": 32859, "epoch": 912, "lr": 0.001} +{"train_loss": 0.03134265914559364, "train_loss_bc": 0.03134265914559364, "train_loss_llm": 0.0, "grad_norm": 0.09099936485290527, "global_step": 32860, "epoch": 912, "lr": 0.001} +{"train_loss": 0.025657549500465393, "train_loss_bc": 0.02565552480518818, "train_loss_llm": 0.00020247697830200195, "grad_norm": 0.034825053066015244, "global_step": 32861, "epoch": 912, "lr": 0.001} +{"train_loss": 0.027157241478562355, "train_loss_bc": 0.02714112401008606, "train_loss_llm": 0.0016117691993713379, "grad_norm": 0.05353385582566261, "global_step": 32862, "epoch": 912, "lr": 0.001} +{"train_loss": 0.029401876032352448, "train_loss_bc": 0.029401876032352448, "train_loss_llm": 0.0, "grad_norm": 0.06971824914216995, "global_step": 32863, "epoch": 912, "lr": 0.001} +{"train_loss": 0.024304687976837158, "train_loss_bc": 0.024304687976837158, "train_loss_llm": 0.0, "grad_norm": 0.08962937444448471, "global_step": 32864, "epoch": 912, "lr": 0.001} +{"train_loss": 0.03262573853135109, "train_loss_bc": 0.03262472152709961, "train_loss_llm": 0.00010180473327636719, "grad_norm": 0.038522325456142426, "global_step": 32865, "epoch": 912, "lr": 0.001} +{"train_loss": 0.025545759126544, "train_loss_bc": 0.025545759126544, "train_loss_llm": 0.0, "grad_norm": 0.054214444011449814, "global_step": 32866, "epoch": 912, "lr": 0.001} +{"train_loss": 0.029044502466503117, "train_loss_bc": 0.028229253366589546, "train_loss_llm": 0.0017337501049041748, "grad_norm": 0.0829492062330246, "global_step": 32867, "epoch": 912, "lr": 0.001, "val_loss": 0.10556069016456604} +{"train_loss": 0.030450863763689995, "train_loss_bc": 0.030450863763689995, "train_loss_llm": 0.0, "grad_norm": 0.10046915709972382, "global_step": 32868, "epoch": 913, "lr": 0.001} +{"train_loss": 0.03348640352487564, "train_loss_bc": 0.03348640352487564, "train_loss_llm": 0.0, "grad_norm": 0.0449124351143837, "global_step": 32869, "epoch": 913, "lr": 0.001} +{"train_loss": 0.024786951020359993, "train_loss_bc": 0.024786951020359993, "train_loss_llm": 0.0, "grad_norm": 0.06997658312320709, "global_step": 32870, "epoch": 913, "lr": 0.001} +{"train_loss": 0.02782590128481388, "train_loss_bc": 0.027820033952593803, "train_loss_llm": 0.0005867481231689453, "grad_norm": 0.09840704500675201, "global_step": 32871, "epoch": 913, "lr": 0.001} +{"train_loss": 0.03479874134063721, "train_loss_bc": 0.03479874134063721, "train_loss_llm": 0.0, "grad_norm": 0.1304815709590912, "global_step": 32872, "epoch": 913, "lr": 0.001} +{"train_loss": 0.024732796475291252, "train_loss_bc": 0.024732796475291252, "train_loss_llm": 0.0, "grad_norm": 0.032721005380153656, "global_step": 32873, "epoch": 913, "lr": 0.001} +{"train_loss": 0.032555557787418365, "train_loss_bc": 0.032555557787418365, "train_loss_llm": 0.0, "grad_norm": 0.06540302187204361, "global_step": 32874, "epoch": 913, "lr": 0.001} +{"train_loss": 0.026683365926146507, "train_loss_bc": 0.026683365926146507, "train_loss_llm": 0.0, "grad_norm": 0.08325577527284622, "global_step": 32875, "epoch": 913, "lr": 0.001} +{"train_loss": 0.030889244750142097, "train_loss_bc": 0.030887747183442116, "train_loss_llm": 0.00014975666999816895, "grad_norm": 0.10136531293392181, "global_step": 32876, "epoch": 913, "lr": 0.001} +{"train_loss": 0.031116152182221413, "train_loss_bc": 0.031116152182221413, "train_loss_llm": 0.0, "grad_norm": 0.04499242827296257, "global_step": 32877, "epoch": 913, "lr": 0.001} +{"train_loss": 0.02506880834698677, "train_loss_bc": 0.02506880834698677, "train_loss_llm": 0.0, "grad_norm": 0.058627232909202576, "global_step": 32878, "epoch": 913, "lr": 0.001} +{"train_loss": 0.02706732787191868, "train_loss_bc": 0.027066286653280258, "train_loss_llm": 0.00010412931442260742, "grad_norm": 0.07803235203027725, "global_step": 32879, "epoch": 913, "lr": 0.001} +{"train_loss": 0.032110512256622314, "train_loss_bc": 0.03211021423339844, "train_loss_llm": 2.995133399963379e-05, "grad_norm": 0.09357114136219025, "global_step": 32880, "epoch": 913, "lr": 0.001} +{"train_loss": 0.02793358452618122, "train_loss_bc": 0.02793358452618122, "train_loss_llm": 0.0, "grad_norm": 0.03913944959640503, "global_step": 32881, "epoch": 913, "lr": 0.001} +{"train_loss": 0.027083605527877808, "train_loss_bc": 0.027083605527877808, "train_loss_llm": 0.0, "grad_norm": 0.05958578735589981, "global_step": 32882, "epoch": 913, "lr": 0.001} +{"train_loss": 0.029178930446505547, "train_loss_bc": 0.029169846326112747, "train_loss_llm": 0.0009083747863769531, "grad_norm": 0.08185631781816483, "global_step": 32883, "epoch": 913, "lr": 0.001} +{"train_loss": 0.032343026250600815, "train_loss_bc": 0.032342344522476196, "train_loss_llm": 6.80088996887207e-05, "grad_norm": 0.11864493042230606, "global_step": 32884, "epoch": 913, "lr": 0.001} +{"train_loss": 0.026702390983700752, "train_loss_bc": 0.026699123904109, "train_loss_llm": 0.0003267824649810791, "grad_norm": 0.03479514271020889, "global_step": 32885, "epoch": 913, "lr": 0.001} +{"train_loss": 0.03417632356286049, "train_loss_bc": 0.03417632356286049, "train_loss_llm": 0.0, "grad_norm": 0.0538753978908062, "global_step": 32886, "epoch": 913, "lr": 0.001} +{"train_loss": 0.023163674399256706, "train_loss_bc": 0.023163674399256706, "train_loss_llm": 0.0, "grad_norm": 0.06526271998882294, "global_step": 32887, "epoch": 913, "lr": 0.001} +{"train_loss": 0.029191620647907257, "train_loss_bc": 0.029191292822360992, "train_loss_llm": 3.272294998168945e-05, "grad_norm": 0.08617518842220306, "global_step": 32888, "epoch": 913, "lr": 0.001} +{"train_loss": 0.029986824840307236, "train_loss_bc": 0.02998550608754158, "train_loss_llm": 0.00013181567192077637, "grad_norm": 0.034235045313835144, "global_step": 32889, "epoch": 913, "lr": 0.001} +{"train_loss": 0.030675208196043968, "train_loss_bc": 0.030675208196043968, "train_loss_llm": 0.0, "grad_norm": 0.06645878404378891, "global_step": 32890, "epoch": 913, "lr": 0.001} +{"train_loss": 0.026944337412714958, "train_loss_bc": 0.026944337412714958, "train_loss_llm": 0.0, "grad_norm": 0.08418526500463486, "global_step": 32891, "epoch": 913, "lr": 0.001} +{"train_loss": 0.027517694979906082, "train_loss_bc": 0.027517694979906082, "train_loss_llm": 0.0, "grad_norm": 0.09917552769184113, "global_step": 32892, "epoch": 913, "lr": 0.001} +{"train_loss": 0.022258702665567398, "train_loss_bc": 0.022257059812545776, "train_loss_llm": 0.00016427040100097656, "grad_norm": 0.033232856541872025, "global_step": 32893, "epoch": 913, "lr": 0.001} +{"train_loss": 0.03167281672358513, "train_loss_bc": 0.03167281672358513, "train_loss_llm": 0.0, "grad_norm": 0.051464539021253586, "global_step": 32894, "epoch": 913, "lr": 0.001} +{"train_loss": 0.031093360856175423, "train_loss_bc": 0.031093360856175423, "train_loss_llm": 0.0, "grad_norm": 0.07462622970342636, "global_step": 32895, "epoch": 913, "lr": 0.001} +{"train_loss": 0.03693896532058716, "train_loss_bc": 0.03693624585866928, "train_loss_llm": 0.00027185678482055664, "grad_norm": 0.10368096083402634, "global_step": 32896, "epoch": 913, "lr": 0.001} +{"train_loss": 0.02796909026801586, "train_loss_bc": 0.02796909026801586, "train_loss_llm": 0.0, "grad_norm": 0.03345252946019173, "global_step": 32897, "epoch": 913, "lr": 0.001} +{"train_loss": 0.03156771883368492, "train_loss_bc": 0.0315653495490551, "train_loss_llm": 0.00023677945137023926, "grad_norm": 0.05158490315079689, "global_step": 32898, "epoch": 913, "lr": 0.001} +{"train_loss": 0.026909034699201584, "train_loss_bc": 0.026909034699201584, "train_loss_llm": 0.0, "grad_norm": 0.0732615664601326, "global_step": 32899, "epoch": 913, "lr": 0.001} +{"train_loss": 0.02797292172908783, "train_loss_bc": 0.02797292172908783, "train_loss_llm": 0.0, "grad_norm": 0.09043245762586594, "global_step": 32900, "epoch": 913, "lr": 0.001} +{"train_loss": 0.02317088656127453, "train_loss_bc": 0.023165568709373474, "train_loss_llm": 0.0005317330360412598, "grad_norm": 0.03390869125723839, "global_step": 32901, "epoch": 913, "lr": 0.001} +{"train_loss": 0.030169593170285225, "train_loss_bc": 0.030151505023241043, "train_loss_llm": 0.0018088817596435547, "grad_norm": 0.0494731105864048, "global_step": 32902, "epoch": 913, "lr": 0.001} +{"train_loss": 0.029185909467438858, "train_loss_bc": 0.03449980169534683, "train_loss_llm": 0.0, "grad_norm": 0.08152339607477188, "global_step": 32903, "epoch": 913, "lr": 0.001, "val_loss": 0.10743704438209534} +{"train_loss": 0.02818484418094158, "train_loss_bc": 0.02818484418094158, "train_loss_llm": 0.0, "grad_norm": 0.10030332952737808, "global_step": 32904, "epoch": 914, "lr": 0.001} +{"train_loss": 0.02944386564195156, "train_loss_bc": 0.02944386564195156, "train_loss_llm": 0.0, "grad_norm": 0.05128990486264229, "global_step": 32905, "epoch": 914, "lr": 0.001} +{"train_loss": 0.031132137402892113, "train_loss_bc": 0.031132137402892113, "train_loss_llm": 0.0, "grad_norm": 0.08111374825239182, "global_step": 32906, "epoch": 914, "lr": 0.001} +{"train_loss": 0.029268480837345123, "train_loss_bc": 0.029268480837345123, "train_loss_llm": 0.0, "grad_norm": 0.10865005850791931, "global_step": 32907, "epoch": 914, "lr": 0.001} +{"train_loss": 0.030196521431207657, "train_loss_bc": 0.030196521431207657, "train_loss_llm": 0.0, "grad_norm": 0.13347555696964264, "global_step": 32908, "epoch": 914, "lr": 0.001} +{"train_loss": 0.031760744750499725, "train_loss_bc": 0.031760744750499725, "train_loss_llm": 0.0, "grad_norm": 0.03752089664340019, "global_step": 32909, "epoch": 914, "lr": 0.001} +{"train_loss": 0.028028368949890137, "train_loss_bc": 0.028018224984407425, "train_loss_llm": 0.0010144710540771484, "grad_norm": 0.06152227520942688, "global_step": 32910, "epoch": 914, "lr": 0.001} +{"train_loss": 0.0239653829485178, "train_loss_bc": 0.0239653829485178, "train_loss_llm": 0.0, "grad_norm": 0.08008122444152832, "global_step": 32911, "epoch": 914, "lr": 0.001} +{"train_loss": 0.027004826813936234, "train_loss_bc": 0.027004826813936234, "train_loss_llm": 0.0, "grad_norm": 0.1043180599808693, "global_step": 32912, "epoch": 914, "lr": 0.001} +{"train_loss": 0.029626693576574326, "train_loss_bc": 0.029620615765452385, "train_loss_llm": 0.0006077289581298828, "grad_norm": 0.032098591327667236, "global_step": 32913, "epoch": 914, "lr": 0.001} +{"train_loss": 0.023176100105047226, "train_loss_bc": 0.02317088283598423, "train_loss_llm": 0.0005217790603637695, "grad_norm": 0.051990266889333725, "global_step": 32914, "epoch": 914, "lr": 0.001} +{"train_loss": 0.029702726751565933, "train_loss_bc": 0.029702726751565933, "train_loss_llm": 0.0, "grad_norm": 0.06936100125312805, "global_step": 32915, "epoch": 914, "lr": 0.001} +{"train_loss": 0.033465638756752014, "train_loss_bc": 0.033465638756752014, "train_loss_llm": 0.0, "grad_norm": 0.0793437510728836, "global_step": 32916, "epoch": 914, "lr": 0.001} +{"train_loss": 0.02886856533586979, "train_loss_bc": 0.02886856533586979, "train_loss_llm": 0.0, "grad_norm": 0.03128276392817497, "global_step": 32917, "epoch": 914, "lr": 0.001} +{"train_loss": 0.02983623370528221, "train_loss_bc": 0.02983623370528221, "train_loss_llm": 0.0, "grad_norm": 0.05849491059780121, "global_step": 32918, "epoch": 914, "lr": 0.001} +{"train_loss": 0.03219149261713028, "train_loss_bc": 0.03219149261713028, "train_loss_llm": 0.0, "grad_norm": 0.08023560047149658, "global_step": 32919, "epoch": 914, "lr": 0.001} +{"train_loss": 0.021563058719038963, "train_loss_bc": 0.021563058719038963, "train_loss_llm": 0.0, "grad_norm": 0.09719521552324295, "global_step": 32920, "epoch": 914, "lr": 0.001} +{"train_loss": 0.029949266463518143, "train_loss_bc": 0.02994917333126068, "train_loss_llm": 9.357929229736328e-06, "grad_norm": 0.03839249908924103, "global_step": 32921, "epoch": 914, "lr": 0.001} +{"train_loss": 0.02964942343533039, "train_loss_bc": 0.02964942343533039, "train_loss_llm": 0.0, "grad_norm": 0.06597626209259033, "global_step": 32922, "epoch": 914, "lr": 0.001} +{"train_loss": 0.023574484512209892, "train_loss_bc": 0.023574484512209892, "train_loss_llm": 0.0, "grad_norm": 0.08485149592161179, "global_step": 32923, "epoch": 914, "lr": 0.001} +{"train_loss": 0.03762665018439293, "train_loss_bc": 0.03762665018439293, "train_loss_llm": 0.0, "grad_norm": 0.13548950850963593, "global_step": 32924, "epoch": 914, "lr": 0.001} +{"train_loss": 0.030408617109060287, "train_loss_bc": 0.030408617109060287, "train_loss_llm": 0.0, "grad_norm": 0.039990898221731186, "global_step": 32925, "epoch": 914, "lr": 0.001} +{"train_loss": 0.030033204704523087, "train_loss_bc": 0.030033204704523087, "train_loss_llm": 0.0, "grad_norm": 0.06325224041938782, "global_step": 32926, "epoch": 914, "lr": 0.001} +{"train_loss": 0.02706853859126568, "train_loss_bc": 0.02705787494778633, "train_loss_llm": 0.0010663866996765137, "grad_norm": 0.09250643849372864, "global_step": 32927, "epoch": 914, "lr": 0.001} +{"train_loss": 0.035171136260032654, "train_loss_bc": 0.035171136260032654, "train_loss_llm": 0.0, "grad_norm": 0.12997925281524658, "global_step": 32928, "epoch": 914, "lr": 0.001} +{"train_loss": 0.033807653933763504, "train_loss_bc": 0.033807653933763504, "train_loss_llm": 0.0, "grad_norm": 0.04241352900862694, "global_step": 32929, "epoch": 914, "lr": 0.001} +{"train_loss": 0.026267310604453087, "train_loss_bc": 0.026267310604453087, "train_loss_llm": 0.0, "grad_norm": 0.06094501167535782, "global_step": 32930, "epoch": 914, "lr": 0.001} +{"train_loss": 0.026546411216259003, "train_loss_bc": 0.026540927588939667, "train_loss_llm": 0.0005484223365783691, "grad_norm": 0.08291672170162201, "global_step": 32931, "epoch": 914, "lr": 0.001} +{"train_loss": 0.028502386063337326, "train_loss_bc": 0.028502386063337326, "train_loss_llm": 0.0, "grad_norm": 0.10595603287220001, "global_step": 32932, "epoch": 914, "lr": 0.001} +{"train_loss": 0.02574911154806614, "train_loss_bc": 0.025748057290911674, "train_loss_llm": 0.00010538101196289062, "grad_norm": 0.035651665180921555, "global_step": 32933, "epoch": 914, "lr": 0.001} +{"train_loss": 0.03261455520987511, "train_loss_bc": 0.03261455520987511, "train_loss_llm": 0.0, "grad_norm": 0.05783374235033989, "global_step": 32934, "epoch": 914, "lr": 0.001} +{"train_loss": 0.029365122318267822, "train_loss_bc": 0.029365122318267822, "train_loss_llm": 0.0, "grad_norm": 0.0815613642334938, "global_step": 32935, "epoch": 914, "lr": 0.001} +{"train_loss": 0.03455236554145813, "train_loss_bc": 0.034538935869932175, "train_loss_llm": 0.0013428032398223877, "grad_norm": 0.10984834283590317, "global_step": 32936, "epoch": 914, "lr": 0.001} +{"train_loss": 0.02603602223098278, "train_loss_bc": 0.02603602223098278, "train_loss_llm": 0.0, "grad_norm": 0.04915876314043999, "global_step": 32937, "epoch": 914, "lr": 0.001} +{"train_loss": 0.031561221927404404, "train_loss_bc": 0.031561221927404404, "train_loss_llm": 0.0, "grad_norm": 0.08494248241186142, "global_step": 32938, "epoch": 914, "lr": 0.001} +{"train_loss": 0.02927746090831028, "train_loss_bc": 0.028089428320527077, "train_loss_llm": 0.0, "grad_norm": 0.1100352481007576, "global_step": 32939, "epoch": 914, "lr": 0.001, "val_loss": 0.10536499321460724} +{"train_loss": 0.0313715934753418, "train_loss_bc": 0.03134134039282799, "train_loss_llm": 0.0030252039432525635, "grad_norm": 0.15255962312221527, "global_step": 32940, "epoch": 915, "lr": 0.001} +{"train_loss": 0.030126269906759262, "train_loss_bc": 0.030126269906759262, "train_loss_llm": 0.0, "grad_norm": 0.041036054491996765, "global_step": 32941, "epoch": 915, "lr": 0.001} +{"train_loss": 0.02590588666498661, "train_loss_bc": 0.025901084765791893, "train_loss_llm": 0.0004801154136657715, "grad_norm": 0.06733198463916779, "global_step": 32942, "epoch": 915, "lr": 0.001} +{"train_loss": 0.029763169586658478, "train_loss_bc": 0.029763169586658478, "train_loss_llm": 0.0, "grad_norm": 0.08542507886886597, "global_step": 32943, "epoch": 915, "lr": 0.001} +{"train_loss": 0.03210138902068138, "train_loss_bc": 0.03210138902068138, "train_loss_llm": 0.0, "grad_norm": 0.1152220070362091, "global_step": 32944, "epoch": 915, "lr": 0.001} +{"train_loss": 0.03173643723130226, "train_loss_bc": 0.03173177316784859, "train_loss_llm": 0.0004664957523345947, "grad_norm": 0.03897351026535034, "global_step": 32945, "epoch": 915, "lr": 0.001} +{"train_loss": 0.029390007257461548, "train_loss_bc": 0.029390007257461548, "train_loss_llm": 0.0, "grad_norm": 0.05909739434719086, "global_step": 32946, "epoch": 915, "lr": 0.001} +{"train_loss": 0.02771315537393093, "train_loss_bc": 0.02771129086613655, "train_loss_llm": 0.00018641352653503418, "grad_norm": 0.08380362391471863, "global_step": 32947, "epoch": 915, "lr": 0.001} +{"train_loss": 0.03139723092317581, "train_loss_bc": 0.03139723092317581, "train_loss_llm": 0.0, "grad_norm": 0.12131921947002411, "global_step": 32948, "epoch": 915, "lr": 0.001} +{"train_loss": 0.030108768492937088, "train_loss_bc": 0.030108768492937088, "train_loss_llm": 0.0, "grad_norm": 0.041844483464956284, "global_step": 32949, "epoch": 915, "lr": 0.001} +{"train_loss": 0.02781137451529503, "train_loss_bc": 0.02781137451529503, "train_loss_llm": 0.0, "grad_norm": 0.07503271102905273, "global_step": 32950, "epoch": 915, "lr": 0.001} +{"train_loss": 0.02969558723270893, "train_loss_bc": 0.02969558723270893, "train_loss_llm": 0.0, "grad_norm": 0.10282409191131592, "global_step": 32951, "epoch": 915, "lr": 0.001} +{"train_loss": 0.02484617382287979, "train_loss_bc": 0.02484617382287979, "train_loss_llm": 0.0, "grad_norm": 0.13649718463420868, "global_step": 32952, "epoch": 915, "lr": 0.001} +{"train_loss": 0.03297232463955879, "train_loss_bc": 0.03296329826116562, "train_loss_llm": 0.0009025931358337402, "grad_norm": 0.0406457893550396, "global_step": 32953, "epoch": 915, "lr": 0.001} +{"train_loss": 0.026119038462638855, "train_loss_bc": 0.026119038462638855, "train_loss_llm": 0.0, "grad_norm": 0.05170818790793419, "global_step": 32954, "epoch": 915, "lr": 0.001} +{"train_loss": 0.031119341030716896, "train_loss_bc": 0.031098736450076103, "train_loss_llm": 0.002060443162918091, "grad_norm": 0.08441971242427826, "global_step": 32955, "epoch": 915, "lr": 0.001} +{"train_loss": 0.027014443650841713, "train_loss_bc": 0.027014443650841713, "train_loss_llm": 0.0, "grad_norm": 0.10399562120437622, "global_step": 32956, "epoch": 915, "lr": 0.001} +{"train_loss": 0.03179823234677315, "train_loss_bc": 0.03179823234677315, "train_loss_llm": 0.0, "grad_norm": 0.037093907594680786, "global_step": 32957, "epoch": 915, "lr": 0.001} +{"train_loss": 0.028476538136601448, "train_loss_bc": 0.028476538136601448, "train_loss_llm": 0.0, "grad_norm": 0.06515941023826599, "global_step": 32958, "epoch": 915, "lr": 0.001} +{"train_loss": 0.026189902797341347, "train_loss_bc": 0.02617720328271389, "train_loss_llm": 0.001269906759262085, "grad_norm": 0.08886732906103134, "global_step": 32959, "epoch": 915, "lr": 0.001} +{"train_loss": 0.02732710912823677, "train_loss_bc": 0.02732710912823677, "train_loss_llm": 0.0, "grad_norm": 0.11634086072444916, "global_step": 32960, "epoch": 915, "lr": 0.001} +{"train_loss": 0.025174874812364578, "train_loss_bc": 0.025165479630231857, "train_loss_llm": 0.000939488410949707, "grad_norm": 0.03754256293177605, "global_step": 32961, "epoch": 915, "lr": 0.001} +{"train_loss": 0.032648902386426926, "train_loss_bc": 0.032640233635902405, "train_loss_llm": 0.0008669793605804443, "grad_norm": 0.06380880624055862, "global_step": 32962, "epoch": 915, "lr": 0.001} +{"train_loss": 0.03836242854595184, "train_loss_bc": 0.03836242854595184, "train_loss_llm": 0.0, "grad_norm": 0.1068975031375885, "global_step": 32963, "epoch": 915, "lr": 0.001} +{"train_loss": 0.029570315033197403, "train_loss_bc": 0.029562588781118393, "train_loss_llm": 0.0007725358009338379, "grad_norm": 0.13549022376537323, "global_step": 32964, "epoch": 915, "lr": 0.001} +{"train_loss": 0.02739584818482399, "train_loss_bc": 0.02739584818482399, "train_loss_llm": 0.0, "grad_norm": 0.05333520099520683, "global_step": 32965, "epoch": 915, "lr": 0.001} +{"train_loss": 0.024949364364147186, "train_loss_bc": 0.024949364364147186, "train_loss_llm": 0.0, "grad_norm": 0.06457991153001785, "global_step": 32966, "epoch": 915, "lr": 0.001} +{"train_loss": 0.03410083428025246, "train_loss_bc": 0.03409865126013756, "train_loss_llm": 0.00021842122077941895, "grad_norm": 0.09939927607774734, "global_step": 32967, "epoch": 915, "lr": 0.001} +{"train_loss": 0.02591712400317192, "train_loss_bc": 0.02591712400317192, "train_loss_llm": 0.0, "grad_norm": 0.11677064746618271, "global_step": 32968, "epoch": 915, "lr": 0.001} +{"train_loss": 0.028276747092604637, "train_loss_bc": 0.028276747092604637, "train_loss_llm": 0.0, "grad_norm": 0.03306141495704651, "global_step": 32969, "epoch": 915, "lr": 0.001} +{"train_loss": 0.0318831168115139, "train_loss_bc": 0.0318831168115139, "train_loss_llm": 0.0, "grad_norm": 0.07860348373651505, "global_step": 32970, "epoch": 915, "lr": 0.001} +{"train_loss": 0.023261841386556625, "train_loss_bc": 0.023259351029992104, "train_loss_llm": 0.00024899840354919434, "grad_norm": 0.11874299496412277, "global_step": 32971, "epoch": 915, "lr": 0.001} +{"train_loss": 0.034312471747398376, "train_loss_bc": 0.034312471747398376, "train_loss_llm": 0.0, "grad_norm": 0.147095188498497, "global_step": 32972, "epoch": 915, "lr": 0.001} +{"train_loss": 0.03118056245148182, "train_loss_bc": 0.03118056245148182, "train_loss_llm": 0.0, "grad_norm": 0.04240603744983673, "global_step": 32973, "epoch": 915, "lr": 0.001} +{"train_loss": 0.03139761462807655, "train_loss_bc": 0.03139761462807655, "train_loss_llm": 0.0, "grad_norm": 0.07696231454610825, "global_step": 32974, "epoch": 915, "lr": 0.001} +{"train_loss": 0.02961620134818885, "train_loss_bc": 0.03476722911000252, "train_loss_llm": 0.0, "grad_norm": 0.12319954484701157, "global_step": 32975, "epoch": 915, "lr": 0.001, "train/cumulative_reward": 69.81599338132348, "train/mean_score": 12.824773411141944, "train/success_rate": 0.3, "test/cumulative_reward": 38.08157482172441, "test/mean_score": 7.17897389205299, "test/success_rate": 0.18, "val_loss": 0.10662178695201874, "train_action_mse_error": 0.03309651091694832} +{"train_loss": 0.03351380676031113, "train_loss_bc": 0.033495109528303146, "train_loss_llm": 0.0018697679042816162, "grad_norm": 0.16253578662872314, "global_step": 32976, "epoch": 916, "lr": 0.001} +{"train_loss": 0.02878466434776783, "train_loss_bc": 0.02878466434776783, "train_loss_llm": 0.0, "grad_norm": 0.04159766063094139, "global_step": 32977, "epoch": 916, "lr": 0.001} +{"train_loss": 0.026364117860794067, "train_loss_bc": 0.026364117860794067, "train_loss_llm": 0.0, "grad_norm": 0.06276193261146545, "global_step": 32978, "epoch": 916, "lr": 0.001} +{"train_loss": 0.028647687286138535, "train_loss_bc": 0.028647687286138535, "train_loss_llm": 0.0, "grad_norm": 0.08977361023426056, "global_step": 32979, "epoch": 916, "lr": 0.001} +{"train_loss": 0.027721820399165154, "train_loss_bc": 0.02772132121026516, "train_loss_llm": 4.9948692321777344e-05, "grad_norm": 0.10969626903533936, "global_step": 32980, "epoch": 916, "lr": 0.001} +{"train_loss": 0.028479503467679024, "train_loss_bc": 0.028479503467679024, "train_loss_llm": 0.0, "grad_norm": 0.04000696912407875, "global_step": 32981, "epoch": 916, "lr": 0.001} +{"train_loss": 0.03125372901558876, "train_loss_bc": 0.03125372901558876, "train_loss_llm": 0.0, "grad_norm": 0.05294225737452507, "global_step": 32982, "epoch": 916, "lr": 0.001} +{"train_loss": 0.026625288650393486, "train_loss_bc": 0.026625288650393486, "train_loss_llm": 0.0, "grad_norm": 0.07113733887672424, "global_step": 32983, "epoch": 916, "lr": 0.001} +{"train_loss": 0.03362586349248886, "train_loss_bc": 0.03362586349248886, "train_loss_llm": 0.0, "grad_norm": 0.0891123116016388, "global_step": 32984, "epoch": 916, "lr": 0.001} +{"train_loss": 0.031947001814842224, "train_loss_bc": 0.031947001814842224, "train_loss_llm": 0.0, "grad_norm": 0.05279522389173508, "global_step": 32985, "epoch": 916, "lr": 0.001} +{"train_loss": 0.027056224644184113, "train_loss_bc": 0.027054540812969208, "train_loss_llm": 0.00016838312149047852, "grad_norm": 0.0830739215016365, "global_step": 32986, "epoch": 916, "lr": 0.001} +{"train_loss": 0.033104147762060165, "train_loss_bc": 0.033086054027080536, "train_loss_llm": 0.001809239387512207, "grad_norm": 0.10382349044084549, "global_step": 32987, "epoch": 916, "lr": 0.001} +{"train_loss": 0.03073117509484291, "train_loss_bc": 0.03070654906332493, "train_loss_llm": 0.0024626851081848145, "grad_norm": 0.1228344738483429, "global_step": 32988, "epoch": 916, "lr": 0.001} +{"train_loss": 0.028515838086605072, "train_loss_bc": 0.028512896969914436, "train_loss_llm": 0.00029405951499938965, "grad_norm": 0.05374814569950104, "global_step": 32989, "epoch": 916, "lr": 0.001} +{"train_loss": 0.027531921863555908, "train_loss_bc": 0.027531921863555908, "train_loss_llm": 0.0, "grad_norm": 0.07870525121688843, "global_step": 32990, "epoch": 916, "lr": 0.001} +{"train_loss": 0.03078276850283146, "train_loss_bc": 0.03078276850283146, "train_loss_llm": 0.0, "grad_norm": 0.11180538684129715, "global_step": 32991, "epoch": 916, "lr": 0.001} +{"train_loss": 0.029496464878320694, "train_loss_bc": 0.029496464878320694, "train_loss_llm": 0.0, "grad_norm": 0.13205751776695251, "global_step": 32992, "epoch": 916, "lr": 0.001} +{"train_loss": 0.030651431530714035, "train_loss_bc": 0.030651431530714035, "train_loss_llm": 0.0, "grad_norm": 0.040201202034950256, "global_step": 32993, "epoch": 916, "lr": 0.001} +{"train_loss": 0.0291262399405241, "train_loss_bc": 0.029110822826623917, "train_loss_llm": 0.0015417635440826416, "grad_norm": 0.05897637456655502, "global_step": 32994, "epoch": 916, "lr": 0.001} +{"train_loss": 0.02932087332010269, "train_loss_bc": 0.02932087332010269, "train_loss_llm": 0.0, "grad_norm": 0.07593915611505508, "global_step": 32995, "epoch": 916, "lr": 0.001} +{"train_loss": 0.02562585100531578, "train_loss_bc": 0.02562585100531578, "train_loss_llm": 0.0, "grad_norm": 0.09835920482873917, "global_step": 32996, "epoch": 916, "lr": 0.001} +{"train_loss": 0.032784633338451385, "train_loss_bc": 0.032784633338451385, "train_loss_llm": 0.0, "grad_norm": 0.04160558804869652, "global_step": 32997, "epoch": 916, "lr": 0.001} +{"train_loss": 0.02812427654862404, "train_loss_bc": 0.028123559430241585, "train_loss_llm": 7.170438766479492e-05, "grad_norm": 0.06170402094721794, "global_step": 32998, "epoch": 916, "lr": 0.001} +{"train_loss": 0.03153659775853157, "train_loss_bc": 0.03153659775853157, "train_loss_llm": 0.0, "grad_norm": 0.07888910174369812, "global_step": 32999, "epoch": 916, "lr": 0.001} +{"train_loss": 0.02760401740670204, "train_loss_bc": 0.02760401740670204, "train_loss_llm": 0.0, "grad_norm": 0.0873897522687912, "global_step": 33000, "epoch": 916, "lr": 0.001} +{"train_loss": 0.023644879460334778, "train_loss_bc": 0.023644879460334778, "train_loss_llm": 0.0, "grad_norm": 0.03035632148385048, "global_step": 33001, "epoch": 916, "lr": 0.001} +{"train_loss": 0.02502020262181759, "train_loss_bc": 0.025003239512443542, "train_loss_llm": 0.0016962289810180664, "grad_norm": 0.055994778871536255, "global_step": 33002, "epoch": 916, "lr": 0.001} +{"train_loss": 0.030423622578382492, "train_loss_bc": 0.030423622578382492, "train_loss_llm": 0.0, "grad_norm": 0.08597536385059357, "global_step": 33003, "epoch": 916, "lr": 0.001} +{"train_loss": 0.028849463909864426, "train_loss_bc": 0.028849463909864426, "train_loss_llm": 0.0, "grad_norm": 0.10908085107803345, "global_step": 33004, "epoch": 916, "lr": 0.001} +{"train_loss": 0.028481651097536087, "train_loss_bc": 0.028481651097536087, "train_loss_llm": 0.0, "grad_norm": 0.04219001159071922, "global_step": 33005, "epoch": 916, "lr": 0.001} +{"train_loss": 0.03123071976006031, "train_loss_bc": 0.03123071976006031, "train_loss_llm": 0.0, "grad_norm": 0.06366654485464096, "global_step": 33006, "epoch": 916, "lr": 0.001} +{"train_loss": 0.028662770986557007, "train_loss_bc": 0.028662770986557007, "train_loss_llm": 0.0, "grad_norm": 0.08214788883924484, "global_step": 33007, "epoch": 916, "lr": 0.001} +{"train_loss": 0.033605486154556274, "train_loss_bc": 0.033605486154556274, "train_loss_llm": 0.0, "grad_norm": 0.11416692286729813, "global_step": 33008, "epoch": 916, "lr": 0.001} +{"train_loss": 0.026686064898967743, "train_loss_bc": 0.026686064898967743, "train_loss_llm": 0.0, "grad_norm": 0.036220818758010864, "global_step": 33009, "epoch": 916, "lr": 0.001} +{"train_loss": 0.02854267507791519, "train_loss_bc": 0.02854267507791519, "train_loss_llm": 0.0, "grad_norm": 0.0568566769361496, "global_step": 33010, "epoch": 916, "lr": 0.001} +{"train_loss": 0.029320956725213263, "train_loss_bc": 0.03145096078515053, "train_loss_llm": 0.0, "grad_norm": 0.08739657700061798, "global_step": 33011, "epoch": 916, "lr": 0.001, "val_loss": 0.10519043356180191} +{"train_loss": 0.02588704600930214, "train_loss_bc": 0.02588704600930214, "train_loss_llm": 0.0, "grad_norm": 0.10522443056106567, "global_step": 33012, "epoch": 917, "lr": 0.001} +{"train_loss": 0.027876809239387512, "train_loss_bc": 0.027876809239387512, "train_loss_llm": 0.0, "grad_norm": 0.034765031188726425, "global_step": 33013, "epoch": 917, "lr": 0.001} +{"train_loss": 0.0319853238761425, "train_loss_bc": 0.0319853238761425, "train_loss_llm": 0.0, "grad_norm": 0.057019468396902084, "global_step": 33014, "epoch": 917, "lr": 0.001} +{"train_loss": 0.02872743085026741, "train_loss_bc": 0.028723297640681267, "train_loss_llm": 0.0004132390022277832, "grad_norm": 0.08124051243066788, "global_step": 33015, "epoch": 917, "lr": 0.001} +{"train_loss": 0.028938010334968567, "train_loss_bc": 0.028918512165546417, "train_loss_llm": 0.0019498467445373535, "grad_norm": 0.0919799879193306, "global_step": 33016, "epoch": 917, "lr": 0.001} +{"train_loss": 0.028743799775838852, "train_loss_bc": 0.028743799775838852, "train_loss_llm": 0.0, "grad_norm": 0.03044993244111538, "global_step": 33017, "epoch": 917, "lr": 0.001} +{"train_loss": 0.03052753396332264, "train_loss_bc": 0.03052753396332264, "train_loss_llm": 0.0, "grad_norm": 0.050107646733522415, "global_step": 33018, "epoch": 917, "lr": 0.001} +{"train_loss": 0.030340535566210747, "train_loss_bc": 0.030340535566210747, "train_loss_llm": 0.0, "grad_norm": 0.06733213365077972, "global_step": 33019, "epoch": 917, "lr": 0.001} +{"train_loss": 0.029624368995428085, "train_loss_bc": 0.029624368995428085, "train_loss_llm": 0.0, "grad_norm": 0.07848833501338959, "global_step": 33020, "epoch": 917, "lr": 0.001} +{"train_loss": 0.02749810367822647, "train_loss_bc": 0.02749810367822647, "train_loss_llm": 0.0, "grad_norm": 0.029659034684300423, "global_step": 33021, "epoch": 917, "lr": 0.001} +{"train_loss": 0.02402622438967228, "train_loss_bc": 0.02402622438967228, "train_loss_llm": 0.0, "grad_norm": 0.04780968278646469, "global_step": 33022, "epoch": 917, "lr": 0.001} +{"train_loss": 0.02593380957841873, "train_loss_bc": 0.025912946090102196, "train_loss_llm": 0.0020863711833953857, "grad_norm": 0.07165181636810303, "global_step": 33023, "epoch": 917, "lr": 0.001} +{"train_loss": 0.03326382488012314, "train_loss_bc": 0.03325623273849487, "train_loss_llm": 0.0007590353488922119, "grad_norm": 0.10448706150054932, "global_step": 33024, "epoch": 917, "lr": 0.001} +{"train_loss": 0.03061911277472973, "train_loss_bc": 0.03061911277472973, "train_loss_llm": 0.0, "grad_norm": 0.05152387544512749, "global_step": 33025, "epoch": 917, "lr": 0.001} +{"train_loss": 0.032313261181116104, "train_loss_bc": 0.03230631723999977, "train_loss_llm": 0.0006945133209228516, "grad_norm": 0.07033754885196686, "global_step": 33026, "epoch": 917, "lr": 0.001} +{"train_loss": 0.03000955656170845, "train_loss_bc": 0.03000955656170845, "train_loss_llm": 0.0, "grad_norm": 0.08875271677970886, "global_step": 33027, "epoch": 917, "lr": 0.001} +{"train_loss": 0.025804979726672173, "train_loss_bc": 0.025804979726672173, "train_loss_llm": 0.0, "grad_norm": 0.10955607891082764, "global_step": 33028, "epoch": 917, "lr": 0.001} +{"train_loss": 0.030413562431931496, "train_loss_bc": 0.030413562431931496, "train_loss_llm": 0.0, "grad_norm": 0.04033113270998001, "global_step": 33029, "epoch": 917, "lr": 0.001} +{"train_loss": 0.03761559724807739, "train_loss_bc": 0.037599653005599976, "train_loss_llm": 0.0015944838523864746, "grad_norm": 0.07603660225868225, "global_step": 33030, "epoch": 917, "lr": 0.001} +{"train_loss": 0.025683898478746414, "train_loss_bc": 0.025674043223261833, "train_loss_llm": 0.000985562801361084, "grad_norm": 0.08609431982040405, "global_step": 33031, "epoch": 917, "lr": 0.001} +{"train_loss": 0.023611217737197876, "train_loss_bc": 0.023611217737197876, "train_loss_llm": 0.0, "grad_norm": 0.10939379036426544, "global_step": 33032, "epoch": 917, "lr": 0.001} +{"train_loss": 0.025253526866436005, "train_loss_bc": 0.025253526866436005, "train_loss_llm": 0.0, "grad_norm": 0.035553913563489914, "global_step": 33033, "epoch": 917, "lr": 0.001} +{"train_loss": 0.023375775665044785, "train_loss_bc": 0.023372018709778786, "train_loss_llm": 0.0003757774829864502, "grad_norm": 0.05314364284276962, "global_step": 33034, "epoch": 917, "lr": 0.001} +{"train_loss": 0.030739231035113335, "train_loss_bc": 0.030739231035113335, "train_loss_llm": 0.0, "grad_norm": 0.06196172535419464, "global_step": 33035, "epoch": 917, "lr": 0.001} +{"train_loss": 0.02758442796766758, "train_loss_bc": 0.02758442796766758, "train_loss_llm": 0.0, "grad_norm": 0.08220455795526505, "global_step": 33036, "epoch": 917, "lr": 0.001} +{"train_loss": 0.029099809005856514, "train_loss_bc": 0.029095865786075592, "train_loss_llm": 0.000394284725189209, "grad_norm": 0.03400406986474991, "global_step": 33037, "epoch": 917, "lr": 0.001} +{"train_loss": 0.025079406797885895, "train_loss_bc": 0.025079406797885895, "train_loss_llm": 0.0, "grad_norm": 0.05227018892765045, "global_step": 33038, "epoch": 917, "lr": 0.001} +{"train_loss": 0.03235917165875435, "train_loss_bc": 0.03235917165875435, "train_loss_llm": 0.0, "grad_norm": 0.06720621138811111, "global_step": 33039, "epoch": 917, "lr": 0.001} +{"train_loss": 0.02880411222577095, "train_loss_bc": 0.02880411222577095, "train_loss_llm": 0.0, "grad_norm": 0.09000016003847122, "global_step": 33040, "epoch": 917, "lr": 0.001} +{"train_loss": 0.028646493330597878, "train_loss_bc": 0.02863544039428234, "train_loss_llm": 0.0011052489280700684, "grad_norm": 0.03589196130633354, "global_step": 33041, "epoch": 917, "lr": 0.001} +{"train_loss": 0.031460680067539215, "train_loss_bc": 0.031460680067539215, "train_loss_llm": 0.0, "grad_norm": 0.06699001789093018, "global_step": 33042, "epoch": 917, "lr": 0.001} +{"train_loss": 0.03503420203924179, "train_loss_bc": 0.03502236306667328, "train_loss_llm": 0.001183927059173584, "grad_norm": 0.09925001114606857, "global_step": 33043, "epoch": 917, "lr": 0.001} +{"train_loss": 0.02982250042259693, "train_loss_bc": 0.02982250042259693, "train_loss_llm": 0.0, "grad_norm": 0.10911490768194199, "global_step": 33044, "epoch": 917, "lr": 0.001} +{"train_loss": 0.03010132908821106, "train_loss_bc": 0.03010132908821106, "train_loss_llm": 0.0, "grad_norm": 0.03720670938491821, "global_step": 33045, "epoch": 917, "lr": 0.001} +{"train_loss": 0.030269330367445946, "train_loss_bc": 0.030269330367445946, "train_loss_llm": 0.0, "grad_norm": 0.05874446779489517, "global_step": 33046, "epoch": 917, "lr": 0.001} +{"train_loss": 0.029101892167495355, "train_loss_bc": 0.0305902361869812, "train_loss_llm": 0.0003877580165863037, "grad_norm": 0.08627587556838989, "global_step": 33047, "epoch": 917, "lr": 0.001, "val_loss": 0.10472233593463898} +{"train_loss": 0.027048218995332718, "train_loss_bc": 0.027048218995332718, "train_loss_llm": 0.0, "grad_norm": 0.10973243415355682, "global_step": 33048, "epoch": 918, "lr": 0.001} +{"train_loss": 0.023938126862049103, "train_loss_bc": 0.023938126862049103, "train_loss_llm": 0.0, "grad_norm": 0.029934348538517952, "global_step": 33049, "epoch": 918, "lr": 0.001} +{"train_loss": 0.026989063248038292, "train_loss_bc": 0.026989063248038292, "train_loss_llm": 0.0, "grad_norm": 0.04383844509720802, "global_step": 33050, "epoch": 918, "lr": 0.001} +{"train_loss": 0.033516205847263336, "train_loss_bc": 0.033516205847263336, "train_loss_llm": 0.0, "grad_norm": 0.06598573923110962, "global_step": 33051, "epoch": 918, "lr": 0.001} +{"train_loss": 0.031725507229566574, "train_loss_bc": 0.031725507229566574, "train_loss_llm": 0.0, "grad_norm": 0.08674046397209167, "global_step": 33052, "epoch": 918, "lr": 0.001} +{"train_loss": 0.02962716855108738, "train_loss_bc": 0.029614955186843872, "train_loss_llm": 0.001221299171447754, "grad_norm": 0.04612577334046364, "global_step": 33053, "epoch": 918, "lr": 0.001} +{"train_loss": 0.023874560371041298, "train_loss_bc": 0.023870816454291344, "train_loss_llm": 0.0003743171691894531, "grad_norm": 0.06753154844045639, "global_step": 33054, "epoch": 918, "lr": 0.001} +{"train_loss": 0.027343232184648514, "train_loss_bc": 0.027343232184648514, "train_loss_llm": 0.0, "grad_norm": 0.08232362568378448, "global_step": 33055, "epoch": 918, "lr": 0.001} +{"train_loss": 0.02959207445383072, "train_loss_bc": 0.02959207445383072, "train_loss_llm": 0.0, "grad_norm": 0.10413163900375366, "global_step": 33056, "epoch": 918, "lr": 0.001} +{"train_loss": 0.03362753987312317, "train_loss_bc": 0.03362753987312317, "train_loss_llm": 0.0, "grad_norm": 0.04075266420841217, "global_step": 33057, "epoch": 918, "lr": 0.001} +{"train_loss": 0.03030223958194256, "train_loss_bc": 0.03030223958194256, "train_loss_llm": 0.0, "grad_norm": 0.05478578433394432, "global_step": 33058, "epoch": 918, "lr": 0.001} +{"train_loss": 0.02967078983783722, "train_loss_bc": 0.02967078983783722, "train_loss_llm": 0.0, "grad_norm": 0.06531944870948792, "global_step": 33059, "epoch": 918, "lr": 0.001} +{"train_loss": 0.03263523429632187, "train_loss_bc": 0.03263523429632187, "train_loss_llm": 0.0, "grad_norm": 0.08597197383642197, "global_step": 33060, "epoch": 918, "lr": 0.001} +{"train_loss": 0.028758812695741653, "train_loss_bc": 0.028750834986567497, "train_loss_llm": 0.0007976889610290527, "grad_norm": 0.041973136365413666, "global_step": 33061, "epoch": 918, "lr": 0.001} +{"train_loss": 0.0344485342502594, "train_loss_bc": 0.034440990537405014, "train_loss_llm": 0.0007544755935668945, "grad_norm": 0.06799811124801636, "global_step": 33062, "epoch": 918, "lr": 0.001} +{"train_loss": 0.029433846473693848, "train_loss_bc": 0.029433846473693848, "train_loss_llm": 0.0, "grad_norm": 0.08221729844808578, "global_step": 33063, "epoch": 918, "lr": 0.001} +{"train_loss": 0.023357074707746506, "train_loss_bc": 0.023350339382886887, "train_loss_llm": 0.0006735026836395264, "grad_norm": 0.0990527868270874, "global_step": 33064, "epoch": 918, "lr": 0.001} +{"train_loss": 0.028070157393813133, "train_loss_bc": 0.028070157393813133, "train_loss_llm": 0.0, "grad_norm": 0.03716430068016052, "global_step": 33065, "epoch": 918, "lr": 0.001} +{"train_loss": 0.026855526491999626, "train_loss_bc": 0.02684047818183899, "train_loss_llm": 0.0015048682689666748, "grad_norm": 0.06633223593235016, "global_step": 33066, "epoch": 918, "lr": 0.001} +{"train_loss": 0.03414789214730263, "train_loss_bc": 0.03414789214730263, "train_loss_llm": 0.0, "grad_norm": 0.09577847272157669, "global_step": 33067, "epoch": 918, "lr": 0.001} +{"train_loss": 0.033314142376184464, "train_loss_bc": 0.03329258784651756, "train_loss_llm": 0.002155601978302002, "grad_norm": 0.11740738898515701, "global_step": 33068, "epoch": 918, "lr": 0.001} +{"train_loss": 0.029471006244421005, "train_loss_bc": 0.029471006244421005, "train_loss_llm": 0.0, "grad_norm": 0.0374920479953289, "global_step": 33069, "epoch": 918, "lr": 0.001} +{"train_loss": 0.027988368645310402, "train_loss_bc": 0.027988368645310402, "train_loss_llm": 0.0, "grad_norm": 0.057331208139657974, "global_step": 33070, "epoch": 918, "lr": 0.001} +{"train_loss": 0.03290879726409912, "train_loss_bc": 0.03290879726409912, "train_loss_llm": 0.0, "grad_norm": 0.08196484297513962, "global_step": 33071, "epoch": 918, "lr": 0.001} +{"train_loss": 0.02986309677362442, "train_loss_bc": 0.02986309677362442, "train_loss_llm": 0.0, "grad_norm": 0.10501554608345032, "global_step": 33072, "epoch": 918, "lr": 0.001} +{"train_loss": 0.026546332985162735, "train_loss_bc": 0.02654317393898964, "train_loss_llm": 0.000315934419631958, "grad_norm": 0.03062790259718895, "global_step": 33073, "epoch": 918, "lr": 0.001} +{"train_loss": 0.030602315440773964, "train_loss_bc": 0.030602315440773964, "train_loss_llm": 0.0, "grad_norm": 0.0543818324804306, "global_step": 33074, "epoch": 918, "lr": 0.001} +{"train_loss": 0.027233541011810303, "train_loss_bc": 0.027233541011810303, "train_loss_llm": 0.0, "grad_norm": 0.06134512275457382, "global_step": 33075, "epoch": 918, "lr": 0.001} +{"train_loss": 0.028193948790431023, "train_loss_bc": 0.028193948790431023, "train_loss_llm": 0.0, "grad_norm": 0.07798491418361664, "global_step": 33076, "epoch": 918, "lr": 0.001} +{"train_loss": 0.02507735788822174, "train_loss_bc": 0.025061216205358505, "train_loss_llm": 0.0016141831874847412, "grad_norm": 0.035111408680677414, "global_step": 33077, "epoch": 918, "lr": 0.001} +{"train_loss": 0.027511464431881905, "train_loss_bc": 0.027506159618496895, "train_loss_llm": 0.0005305111408233643, "grad_norm": 0.05636610463261604, "global_step": 33078, "epoch": 918, "lr": 0.001} +{"train_loss": 0.02860356494784355, "train_loss_bc": 0.02860356494784355, "train_loss_llm": 0.0, "grad_norm": 0.06899809092283249, "global_step": 33079, "epoch": 918, "lr": 0.001} +{"train_loss": 0.02940201573073864, "train_loss_bc": 0.02940201573073864, "train_loss_llm": 0.0, "grad_norm": 0.08947198837995529, "global_step": 33080, "epoch": 918, "lr": 0.001} +{"train_loss": 0.02773762308061123, "train_loss_bc": 0.02773762308061123, "train_loss_llm": 0.0, "grad_norm": 0.038644153624773026, "global_step": 33081, "epoch": 918, "lr": 0.001} +{"train_loss": 0.025890862569212914, "train_loss_bc": 0.025890862569212914, "train_loss_llm": 0.0, "grad_norm": 0.053093746304512024, "global_step": 33082, "epoch": 918, "lr": 0.001} +{"train_loss": 0.02896098042320874, "train_loss_bc": 0.027289051562547684, "train_loss_llm": 0.0, "grad_norm": 0.08161703497171402, "global_step": 33083, "epoch": 918, "lr": 0.001, "val_loss": 0.1056424081325531} +{"train_loss": 0.025919482111930847, "train_loss_bc": 0.025919482111930847, "train_loss_llm": 0.0, "grad_norm": 0.09418924897909164, "global_step": 33084, "epoch": 919, "lr": 0.001} +{"train_loss": 0.03256082162261009, "train_loss_bc": 0.03256082162261009, "train_loss_llm": 0.0, "grad_norm": 0.04099394753575325, "global_step": 33085, "epoch": 919, "lr": 0.001} +{"train_loss": 0.027069449424743652, "train_loss_bc": 0.027069449424743652, "train_loss_llm": 0.0, "grad_norm": 0.060401346534490585, "global_step": 33086, "epoch": 919, "lr": 0.001} +{"train_loss": 0.027020201086997986, "train_loss_bc": 0.027020201086997986, "train_loss_llm": 0.0, "grad_norm": 0.07893256098031998, "global_step": 33087, "epoch": 919, "lr": 0.001} +{"train_loss": 0.02757161483168602, "train_loss_bc": 0.02756771259009838, "train_loss_llm": 0.0003902018070220947, "grad_norm": 0.10111472755670547, "global_step": 33088, "epoch": 919, "lr": 0.001} +{"train_loss": 0.03652595728635788, "train_loss_bc": 0.03652595728635788, "train_loss_llm": 0.0, "grad_norm": 0.05487331375479698, "global_step": 33089, "epoch": 919, "lr": 0.001} +{"train_loss": 0.02939186990261078, "train_loss_bc": 0.02939186990261078, "train_loss_llm": 0.0, "grad_norm": 0.0686250627040863, "global_step": 33090, "epoch": 919, "lr": 0.001} +{"train_loss": 0.03473208099603653, "train_loss_bc": 0.03473208099603653, "train_loss_llm": 0.0, "grad_norm": 0.09415297210216522, "global_step": 33091, "epoch": 919, "lr": 0.001} +{"train_loss": 0.02945360355079174, "train_loss_bc": 0.02945360355079174, "train_loss_llm": 0.0, "grad_norm": 0.12672829627990723, "global_step": 33092, "epoch": 919, "lr": 0.001} +{"train_loss": 0.03471318259835243, "train_loss_bc": 0.03471318259835243, "train_loss_llm": 0.0, "grad_norm": 0.044038109481334686, "global_step": 33093, "epoch": 919, "lr": 0.001} +{"train_loss": 0.031173039227724075, "train_loss_bc": 0.031173039227724075, "train_loss_llm": 0.0, "grad_norm": 0.06568019837141037, "global_step": 33094, "epoch": 919, "lr": 0.001} +{"train_loss": 0.028586454689502716, "train_loss_bc": 0.028576001524925232, "train_loss_llm": 0.0010453462600708008, "grad_norm": 0.08177684992551804, "global_step": 33095, "epoch": 919, "lr": 0.001} +{"train_loss": 0.025941265746951103, "train_loss_bc": 0.025941265746951103, "train_loss_llm": 0.0, "grad_norm": 0.1135919839143753, "global_step": 33096, "epoch": 919, "lr": 0.001} +{"train_loss": 0.025629794225096703, "train_loss_bc": 0.025629794225096703, "train_loss_llm": 0.0, "grad_norm": 0.0369189977645874, "global_step": 33097, "epoch": 919, "lr": 0.001} +{"train_loss": 0.03134625777602196, "train_loss_bc": 0.03134625777602196, "train_loss_llm": 0.0, "grad_norm": 0.06294795870780945, "global_step": 33098, "epoch": 919, "lr": 0.001} +{"train_loss": 0.029972413554787636, "train_loss_bc": 0.029972413554787636, "train_loss_llm": 0.0, "grad_norm": 0.084344282746315, "global_step": 33099, "epoch": 919, "lr": 0.001} +{"train_loss": 0.02912078984081745, "train_loss_bc": 0.02912078984081745, "train_loss_llm": 0.0, "grad_norm": 0.10260020941495895, "global_step": 33100, "epoch": 919, "lr": 0.001} +{"train_loss": 0.03280124440789223, "train_loss_bc": 0.03280124440789223, "train_loss_llm": 0.0, "grad_norm": 0.03780292719602585, "global_step": 33101, "epoch": 919, "lr": 0.001} +{"train_loss": 0.025463037192821503, "train_loss_bc": 0.025457600131630898, "train_loss_llm": 0.0005436241626739502, "grad_norm": 0.061156436800956726, "global_step": 33102, "epoch": 919, "lr": 0.001} +{"train_loss": 0.031910691410303116, "train_loss_bc": 0.031910691410303116, "train_loss_llm": 0.0, "grad_norm": 0.07540835440158844, "global_step": 33103, "epoch": 919, "lr": 0.001} +{"train_loss": 0.02874753624200821, "train_loss_bc": 0.02874753624200821, "train_loss_llm": 0.0, "grad_norm": 0.0935557410120964, "global_step": 33104, "epoch": 919, "lr": 0.001} +{"train_loss": 0.03527309000492096, "train_loss_bc": 0.03527309000492096, "train_loss_llm": 0.0, "grad_norm": 0.0409567765891552, "global_step": 33105, "epoch": 919, "lr": 0.001} +{"train_loss": 0.030561190098524094, "train_loss_bc": 0.030561190098524094, "train_loss_llm": 0.0, "grad_norm": 0.060632143169641495, "global_step": 33106, "epoch": 919, "lr": 0.001} +{"train_loss": 0.025138987228274345, "train_loss_bc": 0.025130275636911392, "train_loss_llm": 0.0008711516857147217, "grad_norm": 0.07080163061618805, "global_step": 33107, "epoch": 919, "lr": 0.001} +{"train_loss": 0.028420045971870422, "train_loss_bc": 0.028420045971870422, "train_loss_llm": 0.0, "grad_norm": 0.08820165693759918, "global_step": 33108, "epoch": 919, "lr": 0.001} +{"train_loss": 0.02892855554819107, "train_loss_bc": 0.028925929218530655, "train_loss_llm": 0.0002626180648803711, "grad_norm": 0.04564937949180603, "global_step": 33109, "epoch": 919, "lr": 0.001} +{"train_loss": 0.033226750791072845, "train_loss_bc": 0.03321347385644913, "train_loss_llm": 0.0013275146484375, "grad_norm": 0.06529828161001205, "global_step": 33110, "epoch": 919, "lr": 0.001} +{"train_loss": 0.031480707228183746, "train_loss_bc": 0.031480707228183746, "train_loss_llm": 0.0, "grad_norm": 0.09428725391626358, "global_step": 33111, "epoch": 919, "lr": 0.001} +{"train_loss": 0.02455778792500496, "train_loss_bc": 0.02455778792500496, "train_loss_llm": 0.0, "grad_norm": 0.12078849971294403, "global_step": 33112, "epoch": 919, "lr": 0.001} +{"train_loss": 0.029301518574357033, "train_loss_bc": 0.029301518574357033, "train_loss_llm": 0.0, "grad_norm": 0.041618283838033676, "global_step": 33113, "epoch": 919, "lr": 0.001} +{"train_loss": 0.02592100016772747, "train_loss_bc": 0.02592100016772747, "train_loss_llm": 0.0, "grad_norm": 0.06470237672328949, "global_step": 33114, "epoch": 919, "lr": 0.001} +{"train_loss": 0.03265862911939621, "train_loss_bc": 0.032654110342264175, "train_loss_llm": 0.0004519522190093994, "grad_norm": 0.08727188408374786, "global_step": 33115, "epoch": 919, "lr": 0.001} +{"train_loss": 0.016553403809666634, "train_loss_bc": 0.016540663316845894, "train_loss_llm": 0.001274019479751587, "grad_norm": 0.09766099601984024, "global_step": 33116, "epoch": 919, "lr": 0.001} +{"train_loss": 0.025881506502628326, "train_loss_bc": 0.025881506502628326, "train_loss_llm": 0.0, "grad_norm": 0.03439905121922493, "global_step": 33117, "epoch": 919, "lr": 0.001} +{"train_loss": 0.02893349900841713, "train_loss_bc": 0.028928307816386223, "train_loss_llm": 0.00051918625831604, "grad_norm": 0.060291994363069534, "global_step": 33118, "epoch": 919, "lr": 0.001} +{"train_loss": 0.029107515803641744, "train_loss_bc": 0.02535705454647541, "train_loss_llm": 0.002605438232421875, "grad_norm": 0.0727042481303215, "global_step": 33119, "epoch": 919, "lr": 0.001, "val_loss": 0.10585484653711319} +{"train_loss": 0.028603995218873024, "train_loss_bc": 0.028603995218873024, "train_loss_llm": 0.0, "grad_norm": 0.09079395979642868, "global_step": 33120, "epoch": 920, "lr": 0.001} +{"train_loss": 0.030945055186748505, "train_loss_bc": 0.030945055186748505, "train_loss_llm": 0.0, "grad_norm": 0.03748786449432373, "global_step": 33121, "epoch": 920, "lr": 0.001} +{"train_loss": 0.034543588757514954, "train_loss_bc": 0.034543588757514954, "train_loss_llm": 0.0, "grad_norm": 0.06980915367603302, "global_step": 33122, "epoch": 920, "lr": 0.001} +{"train_loss": 0.026323555037379265, "train_loss_bc": 0.026323555037379265, "train_loss_llm": 0.0, "grad_norm": 0.08154740929603577, "global_step": 33123, "epoch": 920, "lr": 0.001} +{"train_loss": 0.024091245606541634, "train_loss_bc": 0.024068178609013557, "train_loss_llm": 0.002306729555130005, "grad_norm": 0.10491358488798141, "global_step": 33124, "epoch": 920, "lr": 0.001} +{"train_loss": 0.027389122173190117, "train_loss_bc": 0.027389122173190117, "train_loss_llm": 0.0, "grad_norm": 0.038788001984357834, "global_step": 33125, "epoch": 920, "lr": 0.001} +{"train_loss": 0.030030325055122375, "train_loss_bc": 0.030020855367183685, "train_loss_llm": 0.0009470283985137939, "grad_norm": 0.05437937006354332, "global_step": 33126, "epoch": 920, "lr": 0.001} +{"train_loss": 0.028108621016144753, "train_loss_bc": 0.028108621016144753, "train_loss_llm": 0.0, "grad_norm": 0.06946801394224167, "global_step": 33127, "epoch": 920, "lr": 0.001} +{"train_loss": 0.03468087688088417, "train_loss_bc": 0.034663937985897064, "train_loss_llm": 0.0016939640045166016, "grad_norm": 0.08483349531888962, "global_step": 33128, "epoch": 920, "lr": 0.001} +{"train_loss": 0.024555504322052002, "train_loss_bc": 0.024555504322052002, "train_loss_llm": 0.0, "grad_norm": 0.0288382638245821, "global_step": 33129, "epoch": 920, "lr": 0.001} +{"train_loss": 0.02735850028693676, "train_loss_bc": 0.02735850028693676, "train_loss_llm": 0.0, "grad_norm": 0.04728791490197182, "global_step": 33130, "epoch": 920, "lr": 0.001} +{"train_loss": 0.030305970460176468, "train_loss_bc": 0.030285581946372986, "train_loss_llm": 0.002038925886154175, "grad_norm": 0.06548579037189484, "global_step": 33131, "epoch": 920, "lr": 0.001} +{"train_loss": 0.03072202391922474, "train_loss_bc": 0.03072202391922474, "train_loss_llm": 0.0, "grad_norm": 0.09267023205757141, "global_step": 33132, "epoch": 920, "lr": 0.001} +{"train_loss": 0.03130534663796425, "train_loss_bc": 0.03130534663796425, "train_loss_llm": 0.0, "grad_norm": 0.04509168118238449, "global_step": 33133, "epoch": 920, "lr": 0.001} +{"train_loss": 0.027929436415433884, "train_loss_bc": 0.027911890298128128, "train_loss_llm": 0.0017546415328979492, "grad_norm": 0.07023408263921738, "global_step": 33134, "epoch": 920, "lr": 0.001} +{"train_loss": 0.028450757265090942, "train_loss_bc": 0.028448333963751793, "train_loss_llm": 0.00024241209030151367, "grad_norm": 0.09157437086105347, "global_step": 33135, "epoch": 920, "lr": 0.001} +{"train_loss": 0.030005985870957375, "train_loss_bc": 0.030005985870957375, "train_loss_llm": 0.0, "grad_norm": 0.11852510273456573, "global_step": 33136, "epoch": 920, "lr": 0.001} +{"train_loss": 0.02666180022060871, "train_loss_bc": 0.02666180022060871, "train_loss_llm": 0.0, "grad_norm": 0.03930092975497246, "global_step": 33137, "epoch": 920, "lr": 0.001} +{"train_loss": 0.031936075538396835, "train_loss_bc": 0.031936075538396835, "train_loss_llm": 0.0, "grad_norm": 0.061086177825927734, "global_step": 33138, "epoch": 920, "lr": 0.001} +{"train_loss": 0.02945340983569622, "train_loss_bc": 0.02945340983569622, "train_loss_llm": 0.0, "grad_norm": 0.0834469273686409, "global_step": 33139, "epoch": 920, "lr": 0.001} +{"train_loss": 0.026824457570910454, "train_loss_bc": 0.02681851200759411, "train_loss_llm": 0.0005946159362792969, "grad_norm": 0.09779457002878189, "global_step": 33140, "epoch": 920, "lr": 0.001} +{"train_loss": 0.025742381811141968, "train_loss_bc": 0.025742381811141968, "train_loss_llm": 0.0, "grad_norm": 0.03600870445370674, "global_step": 33141, "epoch": 920, "lr": 0.001} +{"train_loss": 0.035457905381917953, "train_loss_bc": 0.035457905381917953, "train_loss_llm": 0.0, "grad_norm": 0.05912789702415466, "global_step": 33142, "epoch": 920, "lr": 0.001} +{"train_loss": 0.032515209168195724, "train_loss_bc": 0.03251132741570473, "train_loss_llm": 0.00038823485374450684, "grad_norm": 0.07585914433002472, "global_step": 33143, "epoch": 920, "lr": 0.001} +{"train_loss": 0.029290569946169853, "train_loss_bc": 0.029290569946169853, "train_loss_llm": 0.0, "grad_norm": 0.1012105792760849, "global_step": 33144, "epoch": 920, "lr": 0.001} +{"train_loss": 0.03102106787264347, "train_loss_bc": 0.03102106787264347, "train_loss_llm": 0.0, "grad_norm": 0.05277545377612114, "global_step": 33145, "epoch": 920, "lr": 0.001} +{"train_loss": 0.02235095016658306, "train_loss_bc": 0.022341288626194, "train_loss_llm": 0.0009661018848419189, "grad_norm": 0.07466629892587662, "global_step": 33146, "epoch": 920, "lr": 0.001} +{"train_loss": 0.03405404090881348, "train_loss_bc": 0.03405160456895828, "train_loss_llm": 0.0002435147762298584, "grad_norm": 0.09882228076457977, "global_step": 33147, "epoch": 920, "lr": 0.001} +{"train_loss": 0.023613430559635162, "train_loss_bc": 0.023611104115843773, "train_loss_llm": 0.00023260712623596191, "grad_norm": 0.11949818581342697, "global_step": 33148, "epoch": 920, "lr": 0.001} +{"train_loss": 0.02954249083995819, "train_loss_bc": 0.02954249083995819, "train_loss_llm": 0.0, "grad_norm": 0.0353519469499588, "global_step": 33149, "epoch": 920, "lr": 0.001} +{"train_loss": 0.03042013570666313, "train_loss_bc": 0.03042013570666313, "train_loss_llm": 0.0, "grad_norm": 0.052433378994464874, "global_step": 33150, "epoch": 920, "lr": 0.001} +{"train_loss": 0.023077480494976044, "train_loss_bc": 0.023077480494976044, "train_loss_llm": 0.0, "grad_norm": 0.0731649100780487, "global_step": 33151, "epoch": 920, "lr": 0.001} +{"train_loss": 0.024334793910384178, "train_loss_bc": 0.024333426728844643, "train_loss_llm": 0.0001367032527923584, "grad_norm": 0.08518695086240768, "global_step": 33152, "epoch": 920, "lr": 0.001} +{"train_loss": 0.034397706389427185, "train_loss_bc": 0.03439576178789139, "train_loss_llm": 0.00019463896751403809, "grad_norm": 0.04222942143678665, "global_step": 33153, "epoch": 920, "lr": 0.001} +{"train_loss": 0.025447584688663483, "train_loss_bc": 0.025447584688663483, "train_loss_llm": 0.0, "grad_norm": 0.06557761877775192, "global_step": 33154, "epoch": 920, "lr": 0.001} +{"train_loss": 0.029106750929107267, "train_loss_bc": 0.036351632326841354, "train_loss_llm": 0.0, "grad_norm": 0.12722036242485046, "global_step": 33155, "epoch": 920, "lr": 0.001, "train/cumulative_reward": 86.49340840687647, "train/mean_score": 17.835057634525782, "train/success_rate": 0.5, "test/cumulative_reward": 23.633884868928625, "test/mean_score": 8.645716469450472, "test/success_rate": 0.24, "val_loss": 0.10371418297290802, "train_action_mse_error": 0.031701523810625076} +{"train_loss": 0.028151698410511017, "train_loss_bc": 0.028151698410511017, "train_loss_llm": 0.0, "grad_norm": 0.16555145382881165, "global_step": 33156, "epoch": 921, "lr": 0.001} +{"train_loss": 0.03027907945215702, "train_loss_bc": 0.03027907945215702, "train_loss_llm": 0.0, "grad_norm": 0.04760125279426575, "global_step": 33157, "epoch": 921, "lr": 0.001} +{"train_loss": 0.024126628413796425, "train_loss_bc": 0.024126628413796425, "train_loss_llm": 0.0, "grad_norm": 0.07984267175197601, "global_step": 33158, "epoch": 921, "lr": 0.001} +{"train_loss": 0.028513925150036812, "train_loss_bc": 0.028513925150036812, "train_loss_llm": 0.0, "grad_norm": 0.09168662875890732, "global_step": 33159, "epoch": 921, "lr": 0.001} +{"train_loss": 0.028934597969055176, "train_loss_bc": 0.028918592259287834, "train_loss_llm": 0.0016005933284759521, "grad_norm": 0.1148775964975357, "global_step": 33160, "epoch": 921, "lr": 0.001} +{"train_loss": 0.028858758509159088, "train_loss_bc": 0.028858758509159088, "train_loss_llm": 0.0, "grad_norm": 0.031481433659791946, "global_step": 33161, "epoch": 921, "lr": 0.001} +{"train_loss": 0.03412654623389244, "train_loss_bc": 0.03412654623389244, "train_loss_llm": 0.0, "grad_norm": 0.07144411653280258, "global_step": 33162, "epoch": 921, "lr": 0.001} +{"train_loss": 0.026115354150533676, "train_loss_bc": 0.026115354150533676, "train_loss_llm": 0.0, "grad_norm": 0.10653789341449738, "global_step": 33163, "epoch": 921, "lr": 0.001} +{"train_loss": 0.026946712285280228, "train_loss_bc": 0.026946712285280228, "train_loss_llm": 0.0, "grad_norm": 0.12452508509159088, "global_step": 33164, "epoch": 921, "lr": 0.001} +{"train_loss": 0.029157819226384163, "train_loss_bc": 0.02915637567639351, "train_loss_llm": 0.00014442205429077148, "grad_norm": 0.03242568299174309, "global_step": 33165, "epoch": 921, "lr": 0.001} +{"train_loss": 0.031035730615258217, "train_loss_bc": 0.031035730615258217, "train_loss_llm": 0.0, "grad_norm": 0.06595608592033386, "global_step": 33166, "epoch": 921, "lr": 0.001} +{"train_loss": 0.028209688141942024, "train_loss_bc": 0.028205793350934982, "train_loss_llm": 0.00038948655128479004, "grad_norm": 0.0800633504986763, "global_step": 33167, "epoch": 921, "lr": 0.001} +{"train_loss": 0.027103496715426445, "train_loss_bc": 0.02710094302892685, "train_loss_llm": 0.00025528669357299805, "grad_norm": 0.10155647248029709, "global_step": 33168, "epoch": 921, "lr": 0.001} +{"train_loss": 0.025987282395362854, "train_loss_bc": 0.025987282395362854, "train_loss_llm": 0.0, "grad_norm": 0.03765898942947388, "global_step": 33169, "epoch": 921, "lr": 0.001} +{"train_loss": 0.03193792700767517, "train_loss_bc": 0.03193792700767517, "train_loss_llm": 0.0, "grad_norm": 0.06141800433397293, "global_step": 33170, "epoch": 921, "lr": 0.001} +{"train_loss": 0.027583137154579163, "train_loss_bc": 0.027583137154579163, "train_loss_llm": 0.0, "grad_norm": 0.10430712252855301, "global_step": 33171, "epoch": 921, "lr": 0.001} +{"train_loss": 0.026362648233771324, "train_loss_bc": 0.026362648233771324, "train_loss_llm": 0.0, "grad_norm": 0.12285900115966797, "global_step": 33172, "epoch": 921, "lr": 0.001} +{"train_loss": 0.031305961310863495, "train_loss_bc": 0.031305961310863495, "train_loss_llm": 0.0, "grad_norm": 0.05010923743247986, "global_step": 33173, "epoch": 921, "lr": 0.001} +{"train_loss": 0.028445929288864136, "train_loss_bc": 0.028445929288864136, "train_loss_llm": 0.0, "grad_norm": 0.06368999183177948, "global_step": 33174, "epoch": 921, "lr": 0.001} +{"train_loss": 0.02664906717836857, "train_loss_bc": 0.026628440245985985, "train_loss_llm": 0.0020627081394195557, "grad_norm": 0.07824575155973434, "global_step": 33175, "epoch": 921, "lr": 0.001} +{"train_loss": 0.03135483339428902, "train_loss_bc": 0.03135483339428902, "train_loss_llm": 0.0, "grad_norm": 0.09780655801296234, "global_step": 33176, "epoch": 921, "lr": 0.001} +{"train_loss": 0.025648651644587517, "train_loss_bc": 0.025648651644587517, "train_loss_llm": 0.0, "grad_norm": 0.03572302311658859, "global_step": 33177, "epoch": 921, "lr": 0.001} +{"train_loss": 0.03492557257413864, "train_loss_bc": 0.03492557257413864, "train_loss_llm": 0.0, "grad_norm": 0.05967956781387329, "global_step": 33178, "epoch": 921, "lr": 0.001} +{"train_loss": 0.026112502440810204, "train_loss_bc": 0.026112502440810204, "train_loss_llm": 0.0, "grad_norm": 0.06603939086198807, "global_step": 33179, "epoch": 921, "lr": 0.001} +{"train_loss": 0.026863107457756996, "train_loss_bc": 0.026863107457756996, "train_loss_llm": 0.0, "grad_norm": 0.08703741431236267, "global_step": 33180, "epoch": 921, "lr": 0.001} +{"train_loss": 0.028311965987086296, "train_loss_bc": 0.028311965987086296, "train_loss_llm": 0.0, "grad_norm": 0.04445074498653412, "global_step": 33181, "epoch": 921, "lr": 0.001} +{"train_loss": 0.026705799624323845, "train_loss_bc": 0.026705799624323845, "train_loss_llm": 0.0, "grad_norm": 0.07794734835624695, "global_step": 33182, "epoch": 921, "lr": 0.001} +{"train_loss": 0.030685313045978546, "train_loss_bc": 0.030671264976263046, "train_loss_llm": 0.0014048218727111816, "grad_norm": 0.10271867364645004, "global_step": 33183, "epoch": 921, "lr": 0.001} +{"train_loss": 0.031884752213954926, "train_loss_bc": 0.031884752213954926, "train_loss_llm": 0.0, "grad_norm": 0.13124479353427887, "global_step": 33184, "epoch": 921, "lr": 0.001} +{"train_loss": 0.029562557116150856, "train_loss_bc": 0.02955876849591732, "train_loss_llm": 0.0003787875175476074, "grad_norm": 0.04998287558555603, "global_step": 33185, "epoch": 921, "lr": 0.001} +{"train_loss": 0.02775956131517887, "train_loss_bc": 0.027758220210671425, "train_loss_llm": 0.00013405084609985352, "grad_norm": 0.07120732963085175, "global_step": 33186, "epoch": 921, "lr": 0.001} +{"train_loss": 0.03361237049102783, "train_loss_bc": 0.03361237049102783, "train_loss_llm": 0.0, "grad_norm": 0.08453170955181122, "global_step": 33187, "epoch": 921, "lr": 0.001} +{"train_loss": 0.02590049058198929, "train_loss_bc": 0.02589126117527485, "train_loss_llm": 0.0009230077266693115, "grad_norm": 0.11999457329511642, "global_step": 33188, "epoch": 921, "lr": 0.001} +{"train_loss": 0.036330025643110275, "train_loss_bc": 0.036330025643110275, "train_loss_llm": 0.0, "grad_norm": 0.046768732368946075, "global_step": 33189, "epoch": 921, "lr": 0.001} +{"train_loss": 0.031102662906050682, "train_loss_bc": 0.031102662906050682, "train_loss_llm": 0.0, "grad_norm": 0.07411756366491318, "global_step": 33190, "epoch": 921, "lr": 0.001} +{"train_loss": 0.02924118537662758, "train_loss_bc": 0.03609051927924156, "train_loss_llm": 0.0, "grad_norm": 0.1068548783659935, "global_step": 33191, "epoch": 921, "lr": 0.001, "val_loss": 0.10490205883979797} +{"train_loss": 0.02873736061155796, "train_loss_bc": 0.02873736061155796, "train_loss_llm": 0.0, "grad_norm": 0.13422326743602753, "global_step": 33192, "epoch": 922, "lr": 0.001} +{"train_loss": 0.030120577663183212, "train_loss_bc": 0.030120577663183212, "train_loss_llm": 0.0, "grad_norm": 0.03644559904932976, "global_step": 33193, "epoch": 922, "lr": 0.001} +{"train_loss": 0.02633087709546089, "train_loss_bc": 0.02633087709546089, "train_loss_llm": 0.0, "grad_norm": 0.047954730689525604, "global_step": 33194, "epoch": 922, "lr": 0.001} +{"train_loss": 0.031436000019311905, "train_loss_bc": 0.031436000019311905, "train_loss_llm": 0.0, "grad_norm": 0.06704266369342804, "global_step": 33195, "epoch": 922, "lr": 0.001} +{"train_loss": 0.028267953544855118, "train_loss_bc": 0.028267953544855118, "train_loss_llm": 0.0, "grad_norm": 0.10487350821495056, "global_step": 33196, "epoch": 922, "lr": 0.001} +{"train_loss": 0.032049596309661865, "train_loss_bc": 0.032049596309661865, "train_loss_llm": 0.0, "grad_norm": 0.04072553291916847, "global_step": 33197, "epoch": 922, "lr": 0.001} +{"train_loss": 0.029376687481999397, "train_loss_bc": 0.029376687481999397, "train_loss_llm": 0.0, "grad_norm": 0.05719596892595291, "global_step": 33198, "epoch": 922, "lr": 0.001} +{"train_loss": 0.024154238402843475, "train_loss_bc": 0.024154238402843475, "train_loss_llm": 0.0, "grad_norm": 0.07688553631305695, "global_step": 33199, "epoch": 922, "lr": 0.001} +{"train_loss": 0.028005897998809814, "train_loss_bc": 0.028005897998809814, "train_loss_llm": 0.0, "grad_norm": 0.09217251092195511, "global_step": 33200, "epoch": 922, "lr": 0.001} +{"train_loss": 0.027540268376469612, "train_loss_bc": 0.027539778500795364, "train_loss_llm": 4.89652156829834e-05, "grad_norm": 0.03903651610016823, "global_step": 33201, "epoch": 922, "lr": 0.001} +{"train_loss": 0.02942916378378868, "train_loss_bc": 0.02942916378378868, "train_loss_llm": 0.0, "grad_norm": 0.08271343261003494, "global_step": 33202, "epoch": 922, "lr": 0.001} +{"train_loss": 0.02722540870308876, "train_loss_bc": 0.027221519500017166, "train_loss_llm": 0.00038892030715942383, "grad_norm": 0.11415529251098633, "global_step": 33203, "epoch": 922, "lr": 0.001} +{"train_loss": 0.027228282764554024, "train_loss_bc": 0.027228282764554024, "train_loss_llm": 0.0, "grad_norm": 0.1640540063381195, "global_step": 33204, "epoch": 922, "lr": 0.001} +{"train_loss": 0.031562451273202896, "train_loss_bc": 0.031562451273202896, "train_loss_llm": 0.0, "grad_norm": 0.04496929049491882, "global_step": 33205, "epoch": 922, "lr": 0.001} +{"train_loss": 0.026933811604976654, "train_loss_bc": 0.026930583640933037, "train_loss_llm": 0.0003228485584259033, "grad_norm": 0.06826809048652649, "global_step": 33206, "epoch": 922, "lr": 0.001} +{"train_loss": 0.02477789856493473, "train_loss_bc": 0.02477789856493473, "train_loss_llm": 0.0, "grad_norm": 0.09773443639278412, "global_step": 33207, "epoch": 922, "lr": 0.001} +{"train_loss": 0.029680868610739708, "train_loss_bc": 0.029680868610739708, "train_loss_llm": 0.0, "grad_norm": 0.12081513553857803, "global_step": 33208, "epoch": 922, "lr": 0.001} +{"train_loss": 0.030564261600375175, "train_loss_bc": 0.030564261600375175, "train_loss_llm": 0.0, "grad_norm": 0.04793016240000725, "global_step": 33209, "epoch": 922, "lr": 0.001} +{"train_loss": 0.02587256021797657, "train_loss_bc": 0.02585347555577755, "train_loss_llm": 0.0019085109233856201, "grad_norm": 0.07634665071964264, "global_step": 33210, "epoch": 922, "lr": 0.001} +{"train_loss": 0.027752678841352463, "train_loss_bc": 0.027752678841352463, "train_loss_llm": 0.0, "grad_norm": 0.09582348167896271, "global_step": 33211, "epoch": 922, "lr": 0.001} +{"train_loss": 0.027267903089523315, "train_loss_bc": 0.027265677228569984, "train_loss_llm": 0.00022262334823608398, "grad_norm": 0.10198378562927246, "global_step": 33212, "epoch": 922, "lr": 0.001} +{"train_loss": 0.029846126213669777, "train_loss_bc": 0.029846126213669777, "train_loss_llm": 0.0, "grad_norm": 0.041593603789806366, "global_step": 33213, "epoch": 922, "lr": 0.001} +{"train_loss": 0.03157825022935867, "train_loss_bc": 0.03157825022935867, "train_loss_llm": 0.0, "grad_norm": 0.06907549500465393, "global_step": 33214, "epoch": 922, "lr": 0.001} +{"train_loss": 0.031816575676202774, "train_loss_bc": 0.031813085079193115, "train_loss_llm": 0.0003489553928375244, "grad_norm": 0.09091152995824814, "global_step": 33215, "epoch": 922, "lr": 0.001} +{"train_loss": 0.028064025565981865, "train_loss_bc": 0.028064025565981865, "train_loss_llm": 0.0, "grad_norm": 0.11121030896902084, "global_step": 33216, "epoch": 922, "lr": 0.001} +{"train_loss": 0.025464169681072235, "train_loss_bc": 0.025464169681072235, "train_loss_llm": 0.0, "grad_norm": 0.037014078348875046, "global_step": 33217, "epoch": 922, "lr": 0.001} +{"train_loss": 0.03299824893474579, "train_loss_bc": 0.03299824893474579, "train_loss_llm": 0.0, "grad_norm": 0.06393145024776459, "global_step": 33218, "epoch": 922, "lr": 0.001} +{"train_loss": 0.033705949783325195, "train_loss_bc": 0.033705949783325195, "train_loss_llm": 0.0, "grad_norm": 0.09555552899837494, "global_step": 33219, "epoch": 922, "lr": 0.001} +{"train_loss": 0.033272452652454376, "train_loss_bc": 0.033272452652454376, "train_loss_llm": 0.0, "grad_norm": 0.10902394354343414, "global_step": 33220, "epoch": 922, "lr": 0.001} +{"train_loss": 0.02758183516561985, "train_loss_bc": 0.02758183516561985, "train_loss_llm": 0.0, "grad_norm": 0.03557656705379486, "global_step": 33221, "epoch": 922, "lr": 0.001} +{"train_loss": 0.024457894265651703, "train_loss_bc": 0.024456312879920006, "train_loss_llm": 0.00015819072723388672, "grad_norm": 0.051751893013715744, "global_step": 33222, "epoch": 922, "lr": 0.001} +{"train_loss": 0.0352020263671875, "train_loss_bc": 0.0352020263671875, "train_loss_llm": 0.0, "grad_norm": 0.07093258202075958, "global_step": 33223, "epoch": 922, "lr": 0.001} +{"train_loss": 0.02760045975446701, "train_loss_bc": 0.02757325768470764, "train_loss_llm": 0.0027202069759368896, "grad_norm": 0.08745983988046646, "global_step": 33224, "epoch": 922, "lr": 0.001} +{"train_loss": 0.033107686787843704, "train_loss_bc": 0.033107686787843704, "train_loss_llm": 0.0, "grad_norm": 0.04174204543232918, "global_step": 33225, "epoch": 922, "lr": 0.001} +{"train_loss": 0.03172934055328369, "train_loss_bc": 0.03172934055328369, "train_loss_llm": 0.0, "grad_norm": 0.061332035809755325, "global_step": 33226, "epoch": 922, "lr": 0.001} +{"train_loss": 0.029275135085400607, "train_loss_bc": 0.03316507488489151, "train_loss_llm": 0.0, "grad_norm": 0.09821563959121704, "global_step": 33227, "epoch": 922, "lr": 0.001, "val_loss": 0.10632824897766113} +{"train_loss": 0.024889878928661346, "train_loss_bc": 0.024889878928661346, "train_loss_llm": 0.0, "grad_norm": 0.12086255848407745, "global_step": 33228, "epoch": 923, "lr": 0.001} +{"train_loss": 0.03503952547907829, "train_loss_bc": 0.03502270579338074, "train_loss_llm": 0.0016821026802062988, "grad_norm": 0.0393223911523819, "global_step": 33229, "epoch": 923, "lr": 0.001} +{"train_loss": 0.02778557687997818, "train_loss_bc": 0.02778557687997818, "train_loss_llm": 0.0, "grad_norm": 0.06449723243713379, "global_step": 33230, "epoch": 923, "lr": 0.001} +{"train_loss": 0.031757522374391556, "train_loss_bc": 0.0317433662712574, "train_loss_llm": 0.0014156103134155273, "grad_norm": 0.08065975457429886, "global_step": 33231, "epoch": 923, "lr": 0.001} +{"train_loss": 0.036219991743564606, "train_loss_bc": 0.036219991743564606, "train_loss_llm": 0.0, "grad_norm": 0.10324518382549286, "global_step": 33232, "epoch": 923, "lr": 0.001} +{"train_loss": 0.025469740852713585, "train_loss_bc": 0.025469740852713585, "train_loss_llm": 0.0, "grad_norm": 0.033760685473680496, "global_step": 33233, "epoch": 923, "lr": 0.001} +{"train_loss": 0.027437902987003326, "train_loss_bc": 0.027437902987003326, "train_loss_llm": 0.0, "grad_norm": 0.06345584988594055, "global_step": 33234, "epoch": 923, "lr": 0.001} +{"train_loss": 0.03342143073678017, "train_loss_bc": 0.03342143073678017, "train_loss_llm": 0.0, "grad_norm": 0.08410376310348511, "global_step": 33235, "epoch": 923, "lr": 0.001} +{"train_loss": 0.026362845674157143, "train_loss_bc": 0.026362845674157143, "train_loss_llm": 0.0, "grad_norm": 0.09718023985624313, "global_step": 33236, "epoch": 923, "lr": 0.001} +{"train_loss": 0.031748443841934204, "train_loss_bc": 0.031748443841934204, "train_loss_llm": 0.0, "grad_norm": 0.050469499081373215, "global_step": 33237, "epoch": 923, "lr": 0.001} +{"train_loss": 0.02355705015361309, "train_loss_bc": 0.02355705015361309, "train_loss_llm": 0.0, "grad_norm": 0.06409846246242523, "global_step": 33238, "epoch": 923, "lr": 0.001} +{"train_loss": 0.029096504673361778, "train_loss_bc": 0.029096504673361778, "train_loss_llm": 0.0, "grad_norm": 0.08369488269090652, "global_step": 33239, "epoch": 923, "lr": 0.001} +{"train_loss": 0.028309598565101624, "train_loss_bc": 0.028306394815444946, "train_loss_llm": 0.0003203451633453369, "grad_norm": 0.10334809124469757, "global_step": 33240, "epoch": 923, "lr": 0.001} +{"train_loss": 0.026344027370214462, "train_loss_bc": 0.026344027370214462, "train_loss_llm": 0.0, "grad_norm": 0.03312569484114647, "global_step": 33241, "epoch": 923, "lr": 0.001} +{"train_loss": 0.03257998079061508, "train_loss_bc": 0.03257998079061508, "train_loss_llm": 0.0, "grad_norm": 0.046981487423181534, "global_step": 33242, "epoch": 923, "lr": 0.001} +{"train_loss": 0.026467926800251007, "train_loss_bc": 0.026464201509952545, "train_loss_llm": 0.0003725886344909668, "grad_norm": 0.06466685980558395, "global_step": 33243, "epoch": 923, "lr": 0.001} +{"train_loss": 0.031507715582847595, "train_loss_bc": 0.03150033578276634, "train_loss_llm": 0.0007380545139312744, "grad_norm": 0.08188342303037643, "global_step": 33244, "epoch": 923, "lr": 0.001} +{"train_loss": 0.03471935912966728, "train_loss_bc": 0.03471645712852478, "train_loss_llm": 0.00029018521308898926, "grad_norm": 0.044974349439144135, "global_step": 33245, "epoch": 923, "lr": 0.001} +{"train_loss": 0.027670105919241905, "train_loss_bc": 0.027670105919241905, "train_loss_llm": 0.0, "grad_norm": 0.07209084182977676, "global_step": 33246, "epoch": 923, "lr": 0.001} +{"train_loss": 0.02550547569990158, "train_loss_bc": 0.02550547569990158, "train_loss_llm": 0.0, "grad_norm": 0.0882541611790657, "global_step": 33247, "epoch": 923, "lr": 0.001} +{"train_loss": 0.0315619632601738, "train_loss_bc": 0.0315619632601738, "train_loss_llm": 0.0, "grad_norm": 0.11053527146577835, "global_step": 33248, "epoch": 923, "lr": 0.001} +{"train_loss": 0.02803957648575306, "train_loss_bc": 0.02803957648575306, "train_loss_llm": 0.0, "grad_norm": 0.035582285374403, "global_step": 33249, "epoch": 923, "lr": 0.001} +{"train_loss": 0.030797172337770462, "train_loss_bc": 0.030797172337770462, "train_loss_llm": 0.0, "grad_norm": 0.07644173502922058, "global_step": 33250, "epoch": 923, "lr": 0.001} +{"train_loss": 0.03331107646226883, "train_loss_bc": 0.03329600393772125, "train_loss_llm": 0.001507401466369629, "grad_norm": 0.0997346043586731, "global_step": 33251, "epoch": 923, "lr": 0.001} +{"train_loss": 0.03326159715652466, "train_loss_bc": 0.03324367105960846, "train_loss_llm": 0.0017927587032318115, "grad_norm": 0.10896701365709305, "global_step": 33252, "epoch": 923, "lr": 0.001} +{"train_loss": 0.027727147564291954, "train_loss_bc": 0.02771904692053795, "train_loss_llm": 0.0008100271224975586, "grad_norm": 0.03737538307905197, "global_step": 33253, "epoch": 923, "lr": 0.001} +{"train_loss": 0.026261037215590477, "train_loss_bc": 0.026261037215590477, "train_loss_llm": 0.0, "grad_norm": 0.0656948909163475, "global_step": 33254, "epoch": 923, "lr": 0.001} +{"train_loss": 0.027713825926184654, "train_loss_bc": 0.02769806794822216, "train_loss_llm": 0.001575767993927002, "grad_norm": 0.07884104549884796, "global_step": 33255, "epoch": 923, "lr": 0.001} +{"train_loss": 0.032661255449056625, "train_loss_bc": 0.032642919570207596, "train_loss_llm": 0.0018336474895477295, "grad_norm": 0.10099352151155472, "global_step": 33256, "epoch": 923, "lr": 0.001} +{"train_loss": 0.030648289248347282, "train_loss_bc": 0.030648289248347282, "train_loss_llm": 0.0, "grad_norm": 0.04627155885100365, "global_step": 33257, "epoch": 923, "lr": 0.001} +{"train_loss": 0.02613547444343567, "train_loss_bc": 0.02613547444343567, "train_loss_llm": 0.0, "grad_norm": 0.06623101979494095, "global_step": 33258, "epoch": 923, "lr": 0.001} +{"train_loss": 0.02665896899998188, "train_loss_bc": 0.02665531635284424, "train_loss_llm": 0.00036534667015075684, "grad_norm": 0.10175526887178421, "global_step": 33259, "epoch": 923, "lr": 0.001} +{"train_loss": 0.02621852606534958, "train_loss_bc": 0.02621852606534958, "train_loss_llm": 0.0, "grad_norm": 0.12339124828577042, "global_step": 33260, "epoch": 923, "lr": 0.001} +{"train_loss": 0.02642509527504444, "train_loss_bc": 0.02642509527504444, "train_loss_llm": 0.0, "grad_norm": 0.03840021789073944, "global_step": 33261, "epoch": 923, "lr": 0.001} +{"train_loss": 0.02618473581969738, "train_loss_bc": 0.02618473581969738, "train_loss_llm": 0.0, "grad_norm": 0.05390900745987892, "global_step": 33262, "epoch": 923, "lr": 0.001} +{"train_loss": 0.029084932576451037, "train_loss_bc": 0.02756122685968876, "train_loss_llm": 0.0, "grad_norm": 0.05873351916670799, "global_step": 33263, "epoch": 923, "lr": 0.001, "val_loss": 0.10594107955694199} +{"train_loss": 0.03063398413360119, "train_loss_bc": 0.03063398413360119, "train_loss_llm": 0.0, "grad_norm": 0.07236310839653015, "global_step": 33264, "epoch": 924, "lr": 0.001} +{"train_loss": 0.026638653129339218, "train_loss_bc": 0.026638653129339218, "train_loss_llm": 0.0, "grad_norm": 0.03543524071574211, "global_step": 33265, "epoch": 924, "lr": 0.001} +{"train_loss": 0.027382507920265198, "train_loss_bc": 0.027382507920265198, "train_loss_llm": 0.0, "grad_norm": 0.07022763788700104, "global_step": 33266, "epoch": 924, "lr": 0.001} +{"train_loss": 0.02703528292477131, "train_loss_bc": 0.02703528292477131, "train_loss_llm": 0.0, "grad_norm": 0.09725379198789597, "global_step": 33267, "epoch": 924, "lr": 0.001} +{"train_loss": 0.028654467314481735, "train_loss_bc": 0.028654467314481735, "train_loss_llm": 0.0, "grad_norm": 0.11876939237117767, "global_step": 33268, "epoch": 924, "lr": 0.001} +{"train_loss": 0.027708427980542183, "train_loss_bc": 0.027708427980542183, "train_loss_llm": 0.0, "grad_norm": 0.05151427909731865, "global_step": 33269, "epoch": 924, "lr": 0.001} +{"train_loss": 0.02958131954073906, "train_loss_bc": 0.02957666479051113, "train_loss_llm": 0.0004655122756958008, "grad_norm": 0.07096779346466064, "global_step": 33270, "epoch": 924, "lr": 0.001} +{"train_loss": 0.030981391668319702, "train_loss_bc": 0.030981391668319702, "train_loss_llm": 0.0, "grad_norm": 0.10997269302606583, "global_step": 33271, "epoch": 924, "lr": 0.001} +{"train_loss": 0.026956237852573395, "train_loss_bc": 0.026956237852573395, "train_loss_llm": 0.0, "grad_norm": 0.12891888618469238, "global_step": 33272, "epoch": 924, "lr": 0.001} +{"train_loss": 0.0254775732755661, "train_loss_bc": 0.0254775732755661, "train_loss_llm": 0.0, "grad_norm": 0.03265555575489998, "global_step": 33273, "epoch": 924, "lr": 0.001} +{"train_loss": 0.030665738508105278, "train_loss_bc": 0.030655445531010628, "train_loss_llm": 0.001029282808303833, "grad_norm": 0.07656075805425644, "global_step": 33274, "epoch": 924, "lr": 0.001} +{"train_loss": 0.023958532139658928, "train_loss_bc": 0.023953678086400032, "train_loss_llm": 0.00048542022705078125, "grad_norm": 0.09382506459951401, "global_step": 33275, "epoch": 924, "lr": 0.001} +{"train_loss": 0.03660629689693451, "train_loss_bc": 0.03660629689693451, "train_loss_llm": 0.0, "grad_norm": 0.1350513994693756, "global_step": 33276, "epoch": 924, "lr": 0.001} +{"train_loss": 0.025028662756085396, "train_loss_bc": 0.025028662756085396, "train_loss_llm": 0.0, "grad_norm": 0.038982901722192764, "global_step": 33277, "epoch": 924, "lr": 0.001} +{"train_loss": 0.03314686566591263, "train_loss_bc": 0.03314686566591263, "train_loss_llm": 0.0, "grad_norm": 0.06869994103908539, "global_step": 33278, "epoch": 924, "lr": 0.001} +{"train_loss": 0.03293757140636444, "train_loss_bc": 0.03293757140636444, "train_loss_llm": 0.0, "grad_norm": 0.08201204240322113, "global_step": 33279, "epoch": 924, "lr": 0.001} +{"train_loss": 0.030135342851281166, "train_loss_bc": 0.030135342851281166, "train_loss_llm": 0.0, "grad_norm": 0.0988549143075943, "global_step": 33280, "epoch": 924, "lr": 0.001} +{"train_loss": 0.025982189923524857, "train_loss_bc": 0.025982189923524857, "train_loss_llm": 0.0, "grad_norm": 0.04299985244870186, "global_step": 33281, "epoch": 924, "lr": 0.001} +{"train_loss": 0.03411107137799263, "train_loss_bc": 0.03411107137799263, "train_loss_llm": 0.0, "grad_norm": 0.06527076661586761, "global_step": 33282, "epoch": 924, "lr": 0.001} +{"train_loss": 0.031139541417360306, "train_loss_bc": 0.031139541417360306, "train_loss_llm": 0.0, "grad_norm": 0.08243712782859802, "global_step": 33283, "epoch": 924, "lr": 0.001} +{"train_loss": 0.031215444207191467, "train_loss_bc": 0.031215444207191467, "train_loss_llm": 0.0, "grad_norm": 0.11005093902349472, "global_step": 33284, "epoch": 924, "lr": 0.001} +{"train_loss": 0.03190566971898079, "train_loss_bc": 0.03190566971898079, "train_loss_llm": 0.0, "grad_norm": 0.038856733590364456, "global_step": 33285, "epoch": 924, "lr": 0.001} +{"train_loss": 0.03747091069817543, "train_loss_bc": 0.03747091069817543, "train_loss_llm": 0.0, "grad_norm": 0.07263518869876862, "global_step": 33286, "epoch": 924, "lr": 0.001} +{"train_loss": 0.026166437193751335, "train_loss_bc": 0.026166437193751335, "train_loss_llm": 0.0, "grad_norm": 0.1055668368935585, "global_step": 33287, "epoch": 924, "lr": 0.001} +{"train_loss": 0.03210160881280899, "train_loss_bc": 0.03209824115037918, "train_loss_llm": 0.00033664703369140625, "grad_norm": 0.15347057580947876, "global_step": 33288, "epoch": 924, "lr": 0.001} +{"train_loss": 0.029102660715579987, "train_loss_bc": 0.029101412743330002, "train_loss_llm": 0.00012472271919250488, "grad_norm": 0.04041459411382675, "global_step": 33289, "epoch": 924, "lr": 0.001} +{"train_loss": 0.030330969020724297, "train_loss_bc": 0.030330969020724297, "train_loss_llm": 0.0, "grad_norm": 0.07722948491573334, "global_step": 33290, "epoch": 924, "lr": 0.001} +{"train_loss": 0.028398150578141212, "train_loss_bc": 0.028398150578141212, "train_loss_llm": 0.0, "grad_norm": 0.1125180795788765, "global_step": 33291, "epoch": 924, "lr": 0.001} +{"train_loss": 0.02434474229812622, "train_loss_bc": 0.02434474229812622, "train_loss_llm": 0.0, "grad_norm": 0.13316847383975983, "global_step": 33292, "epoch": 924, "lr": 0.001} +{"train_loss": 0.026876181364059448, "train_loss_bc": 0.026876181364059448, "train_loss_llm": 0.0, "grad_norm": 0.0423785075545311, "global_step": 33293, "epoch": 924, "lr": 0.001} +{"train_loss": 0.03272848203778267, "train_loss_bc": 0.03272848203778267, "train_loss_llm": 0.0, "grad_norm": 0.08160887658596039, "global_step": 33294, "epoch": 924, "lr": 0.001} +{"train_loss": 0.026886016130447388, "train_loss_bc": 0.026886016130447388, "train_loss_llm": 0.0, "grad_norm": 0.09180112928152084, "global_step": 33295, "epoch": 924, "lr": 0.001} +{"train_loss": 0.026315661147236824, "train_loss_bc": 0.026315661147236824, "train_loss_llm": 0.0, "grad_norm": 0.12357772886753082, "global_step": 33296, "epoch": 924, "lr": 0.001} +{"train_loss": 0.024777257815003395, "train_loss_bc": 0.024766283109784126, "train_loss_llm": 0.0010975003242492676, "grad_norm": 0.0301193967461586, "global_step": 33297, "epoch": 924, "lr": 0.001} +{"train_loss": 0.026430735364556313, "train_loss_bc": 0.026430735364556313, "train_loss_llm": 0.0, "grad_norm": 0.048640161752700806, "global_step": 33298, "epoch": 924, "lr": 0.001} +{"train_loss": 0.029277924499991868, "train_loss_bc": 0.03417055681347847, "train_loss_llm": 0.0022138655185699463, "grad_norm": 0.07446467876434326, "global_step": 33299, "epoch": 924, "lr": 0.001, "val_loss": 0.10485482215881348} +{"train_loss": 0.02733948640525341, "train_loss_bc": 0.02733800932765007, "train_loss_llm": 0.00014778971672058105, "grad_norm": 0.09907636791467667, "global_step": 33300, "epoch": 925, "lr": 0.001} +{"train_loss": 0.03071032278239727, "train_loss_bc": 0.03071032278239727, "train_loss_llm": 0.0, "grad_norm": 0.03622885420918465, "global_step": 33301, "epoch": 925, "lr": 0.001} +{"train_loss": 0.02895752340555191, "train_loss_bc": 0.02895752340555191, "train_loss_llm": 0.0, "grad_norm": 0.061539363116025925, "global_step": 33302, "epoch": 925, "lr": 0.001} +{"train_loss": 0.030817946419119835, "train_loss_bc": 0.03081686981022358, "train_loss_llm": 0.00010773539543151855, "grad_norm": 0.09821789711713791, "global_step": 33303, "epoch": 925, "lr": 0.001} +{"train_loss": 0.030059684067964554, "train_loss_bc": 0.030059684067964554, "train_loss_llm": 0.0, "grad_norm": 0.14233289659023285, "global_step": 33304, "epoch": 925, "lr": 0.001} +{"train_loss": 0.03330014646053314, "train_loss_bc": 0.03326980397105217, "train_loss_llm": 0.003034234046936035, "grad_norm": 0.04653041437268257, "global_step": 33305, "epoch": 925, "lr": 0.001} +{"train_loss": 0.031101543456315994, "train_loss_bc": 0.031101543456315994, "train_loss_llm": 0.0, "grad_norm": 0.07726511359214783, "global_step": 33306, "epoch": 925, "lr": 0.001} +{"train_loss": 0.02976987510919571, "train_loss_bc": 0.02976987510919571, "train_loss_llm": 0.0, "grad_norm": 0.10347424447536469, "global_step": 33307, "epoch": 925, "lr": 0.001} +{"train_loss": 0.024797363206744194, "train_loss_bc": 0.02479240670800209, "train_loss_llm": 0.0004957318305969238, "grad_norm": 0.12956523895263672, "global_step": 33308, "epoch": 925, "lr": 0.001} +{"train_loss": 0.03424063324928284, "train_loss_bc": 0.03424063324928284, "train_loss_llm": 0.0, "grad_norm": 0.041335660964250565, "global_step": 33309, "epoch": 925, "lr": 0.001} +{"train_loss": 0.030910354107618332, "train_loss_bc": 0.030910354107618332, "train_loss_llm": 0.0, "grad_norm": 0.07811781018972397, "global_step": 33310, "epoch": 925, "lr": 0.001} +{"train_loss": 0.029631057754158974, "train_loss_bc": 0.029631057754158974, "train_loss_llm": 0.0, "grad_norm": 0.11842846870422363, "global_step": 33311, "epoch": 925, "lr": 0.001} +{"train_loss": 0.03077700547873974, "train_loss_bc": 0.03077700547873974, "train_loss_llm": 0.0, "grad_norm": 0.1572791039943695, "global_step": 33312, "epoch": 925, "lr": 0.001} +{"train_loss": 0.028551964089274406, "train_loss_bc": 0.028551964089274406, "train_loss_llm": 0.0, "grad_norm": 0.04185884818434715, "global_step": 33313, "epoch": 925, "lr": 0.001} +{"train_loss": 0.027278341352939606, "train_loss_bc": 0.027278341352939606, "train_loss_llm": 0.0, "grad_norm": 0.05360928550362587, "global_step": 33314, "epoch": 925, "lr": 0.001} +{"train_loss": 0.03286002203822136, "train_loss_bc": 0.03286002203822136, "train_loss_llm": 0.0, "grad_norm": 0.07908730953931808, "global_step": 33315, "epoch": 925, "lr": 0.001} +{"train_loss": 0.02300761453807354, "train_loss_bc": 0.02300761453807354, "train_loss_llm": 0.0, "grad_norm": 0.0949782133102417, "global_step": 33316, "epoch": 925, "lr": 0.001} +{"train_loss": 0.024619298055768013, "train_loss_bc": 0.024619298055768013, "train_loss_llm": 0.0, "grad_norm": 0.02711164951324463, "global_step": 33317, "epoch": 925, "lr": 0.001} +{"train_loss": 0.031076567247509956, "train_loss_bc": 0.031076567247509956, "train_loss_llm": 0.0, "grad_norm": 0.051658764481544495, "global_step": 33318, "epoch": 925, "lr": 0.001} +{"train_loss": 0.029483703896403313, "train_loss_bc": 0.029482819139957428, "train_loss_llm": 8.848309516906738e-05, "grad_norm": 0.06714723259210587, "global_step": 33319, "epoch": 925, "lr": 0.001} +{"train_loss": 0.029094701632857323, "train_loss_bc": 0.029094701632857323, "train_loss_llm": 0.0, "grad_norm": 0.08390193432569504, "global_step": 33320, "epoch": 925, "lr": 0.001} +{"train_loss": 0.03167557343840599, "train_loss_bc": 0.03167107328772545, "train_loss_llm": 0.00044983625411987305, "grad_norm": 0.039782650768756866, "global_step": 33321, "epoch": 925, "lr": 0.001} +{"train_loss": 0.03133443742990494, "train_loss_bc": 0.031332891434431076, "train_loss_llm": 0.00015461444854736328, "grad_norm": 0.07221914827823639, "global_step": 33322, "epoch": 925, "lr": 0.001} +{"train_loss": 0.027884913608431816, "train_loss_bc": 0.027884913608431816, "train_loss_llm": 0.0, "grad_norm": 0.10235965251922607, "global_step": 33323, "epoch": 925, "lr": 0.001} +{"train_loss": 0.02957659773528576, "train_loss_bc": 0.02957659773528576, "train_loss_llm": 0.0, "grad_norm": 0.13531549274921417, "global_step": 33324, "epoch": 925, "lr": 0.001} +{"train_loss": 0.02817275933921337, "train_loss_bc": 0.02817275933921337, "train_loss_llm": 0.0, "grad_norm": 0.06754305958747864, "global_step": 33325, "epoch": 925, "lr": 0.001} +{"train_loss": 0.02952735498547554, "train_loss_bc": 0.02952735498547554, "train_loss_llm": 0.0, "grad_norm": 0.11408878862857819, "global_step": 33326, "epoch": 925, "lr": 0.001} +{"train_loss": 0.03211423009634018, "train_loss_bc": 0.03211423009634018, "train_loss_llm": 0.0, "grad_norm": 0.16986799240112305, "global_step": 33327, "epoch": 925, "lr": 0.001} +{"train_loss": 0.029375219717621803, "train_loss_bc": 0.029367972165346146, "train_loss_llm": 0.0007247030735015869, "grad_norm": 0.2019701600074768, "global_step": 33328, "epoch": 925, "lr": 0.001} +{"train_loss": 0.026017850264906883, "train_loss_bc": 0.026017850264906883, "train_loss_llm": 0.0, "grad_norm": 0.045161277055740356, "global_step": 33329, "epoch": 925, "lr": 0.001} +{"train_loss": 0.030764054507017136, "train_loss_bc": 0.030764054507017136, "train_loss_llm": 0.0, "grad_norm": 0.09746797382831573, "global_step": 33330, "epoch": 925, "lr": 0.001} +{"train_loss": 0.030637696385383606, "train_loss_bc": 0.030637696385383606, "train_loss_llm": 0.0, "grad_norm": 0.13743428885936737, "global_step": 33331, "epoch": 925, "lr": 0.001} +{"train_loss": 0.031213514506816864, "train_loss_bc": 0.031205903738737106, "train_loss_llm": 0.0007610619068145752, "grad_norm": 0.15477773547172546, "global_step": 33332, "epoch": 925, "lr": 0.001} +{"train_loss": 0.030058221891522408, "train_loss_bc": 0.030058221891522408, "train_loss_llm": 0.0, "grad_norm": 0.04278448969125748, "global_step": 33333, "epoch": 925, "lr": 0.001} +{"train_loss": 0.028348645195364952, "train_loss_bc": 0.02834426239132881, "train_loss_llm": 0.00043836236000061035, "grad_norm": 0.06689123809337616, "global_step": 33334, "epoch": 925, "lr": 0.001} +{"train_loss": 0.02941785069803397, "train_loss_bc": 0.023946082219481468, "train_loss_llm": 0.0010319650173187256, "grad_norm": 0.07843062281608582, "global_step": 33335, "epoch": 925, "lr": 0.001, "train/cumulative_reward": 87.90669444497956, "train/mean_score": 18.747181658555853, "train/success_rate": 0.5, "test/cumulative_reward": 40.19841374190449, "test/mean_score": 6.8279309065054585, "test/success_rate": 0.16, "val_loss": 0.10480210185050964, "train_action_mse_error": 0.03166940063238144} +{"train_loss": 0.03062564693391323, "train_loss_bc": 0.03062564693391323, "train_loss_llm": 0.0, "grad_norm": 0.09857342392206192, "global_step": 33336, "epoch": 926, "lr": 0.001} +{"train_loss": 0.026486508548259735, "train_loss_bc": 0.026478752493858337, "train_loss_llm": 0.0007755756378173828, "grad_norm": 0.031669266521930695, "global_step": 33337, "epoch": 926, "lr": 0.001} +{"train_loss": 0.02986598201096058, "train_loss_bc": 0.02986598201096058, "train_loss_llm": 0.0, "grad_norm": 0.0657438412308693, "global_step": 33338, "epoch": 926, "lr": 0.001} +{"train_loss": 0.029965808615088463, "train_loss_bc": 0.029964759945869446, "train_loss_llm": 0.00010481476783752441, "grad_norm": 0.09280543774366379, "global_step": 33339, "epoch": 926, "lr": 0.001} +{"train_loss": 0.030878230929374695, "train_loss_bc": 0.030878230929374695, "train_loss_llm": 0.0, "grad_norm": 0.13100150227546692, "global_step": 33340, "epoch": 926, "lr": 0.001} +{"train_loss": 0.024800052866339684, "train_loss_bc": 0.024795016273856163, "train_loss_llm": 0.0005037486553192139, "grad_norm": 0.051010552793741226, "global_step": 33341, "epoch": 926, "lr": 0.001} +{"train_loss": 0.030825719237327576, "train_loss_bc": 0.030825719237327576, "train_loss_llm": 0.0, "grad_norm": 0.0734129548072815, "global_step": 33342, "epoch": 926, "lr": 0.001} +{"train_loss": 0.030105017125606537, "train_loss_bc": 0.030105017125606537, "train_loss_llm": 0.0, "grad_norm": 0.10952490568161011, "global_step": 33343, "epoch": 926, "lr": 0.001} +{"train_loss": 0.03165901452302933, "train_loss_bc": 0.03165901452302933, "train_loss_llm": 0.0, "grad_norm": 0.14606407284736633, "global_step": 33344, "epoch": 926, "lr": 0.001} +{"train_loss": 0.02766941301524639, "train_loss_bc": 0.02766941301524639, "train_loss_llm": 0.0, "grad_norm": 0.03718952462077141, "global_step": 33345, "epoch": 926, "lr": 0.001} +{"train_loss": 0.030529510229825974, "train_loss_bc": 0.030529510229825974, "train_loss_llm": 0.0, "grad_norm": 0.08812783658504486, "global_step": 33346, "epoch": 926, "lr": 0.001} +{"train_loss": 0.026854392141103745, "train_loss_bc": 0.026854392141103745, "train_loss_llm": 0.0, "grad_norm": 0.1312423050403595, "global_step": 33347, "epoch": 926, "lr": 0.001} +{"train_loss": 0.029059547930955887, "train_loss_bc": 0.029059547930955887, "train_loss_llm": 0.0, "grad_norm": 0.16055534780025482, "global_step": 33348, "epoch": 926, "lr": 0.001} +{"train_loss": 0.028714561834931374, "train_loss_bc": 0.028714561834931374, "train_loss_llm": 0.0, "grad_norm": 0.036766424775123596, "global_step": 33349, "epoch": 926, "lr": 0.001} +{"train_loss": 0.02681799791753292, "train_loss_bc": 0.02681799791753292, "train_loss_llm": 0.0, "grad_norm": 0.06069696322083473, "global_step": 33350, "epoch": 926, "lr": 0.001} +{"train_loss": 0.033364951610565186, "train_loss_bc": 0.033360015600919724, "train_loss_llm": 0.0004936456680297852, "grad_norm": 0.09834171831607819, "global_step": 33351, "epoch": 926, "lr": 0.001} +{"train_loss": 0.026599455624818802, "train_loss_bc": 0.02659608982503414, "train_loss_llm": 0.00033661723136901855, "grad_norm": 0.121373251080513, "global_step": 33352, "epoch": 926, "lr": 0.001} +{"train_loss": 0.02618866041302681, "train_loss_bc": 0.02618866041302681, "train_loss_llm": 0.0, "grad_norm": 0.03509100526571274, "global_step": 33353, "epoch": 926, "lr": 0.001} +{"train_loss": 0.028728283941745758, "train_loss_bc": 0.028728283941745758, "train_loss_llm": 0.0, "grad_norm": 0.06657972931861877, "global_step": 33354, "epoch": 926, "lr": 0.001} +{"train_loss": 0.02851123921573162, "train_loss_bc": 0.02851123921573162, "train_loss_llm": 0.0, "grad_norm": 0.08488032966852188, "global_step": 33355, "epoch": 926, "lr": 0.001} +{"train_loss": 0.030175035819411278, "train_loss_bc": 0.030166035518050194, "train_loss_llm": 0.0009000897407531738, "grad_norm": 0.10562318563461304, "global_step": 33356, "epoch": 926, "lr": 0.001} +{"train_loss": 0.03269317373633385, "train_loss_bc": 0.03269317373633385, "train_loss_llm": 0.0, "grad_norm": 0.04904709383845329, "global_step": 33357, "epoch": 926, "lr": 0.001} +{"train_loss": 0.025260617956519127, "train_loss_bc": 0.025260617956519127, "train_loss_llm": 0.0, "grad_norm": 0.088568776845932, "global_step": 33358, "epoch": 926, "lr": 0.001} +{"train_loss": 0.027100104838609695, "train_loss_bc": 0.027100104838609695, "train_loss_llm": 0.0, "grad_norm": 0.13568566739559174, "global_step": 33359, "epoch": 926, "lr": 0.001} +{"train_loss": 0.030906803905963898, "train_loss_bc": 0.03090253472328186, "train_loss_llm": 0.00042685866355895996, "grad_norm": 0.18126890063285828, "global_step": 33360, "epoch": 926, "lr": 0.001} +{"train_loss": 0.029431667178869247, "train_loss_bc": 0.029431667178869247, "train_loss_llm": 0.0, "grad_norm": 0.04024403542280197, "global_step": 33361, "epoch": 926, "lr": 0.001} +{"train_loss": 0.042834073305130005, "train_loss_bc": 0.04282865673303604, "train_loss_llm": 0.0005417466163635254, "grad_norm": 0.10106997936964035, "global_step": 33362, "epoch": 926, "lr": 0.001} +{"train_loss": 0.02834981493651867, "train_loss_bc": 0.02834981493651867, "train_loss_llm": 0.0, "grad_norm": 0.11468149721622467, "global_step": 33363, "epoch": 926, "lr": 0.001} +{"train_loss": 0.030394747853279114, "train_loss_bc": 0.030382052063941956, "train_loss_llm": 0.001269608736038208, "grad_norm": 0.1493421345949173, "global_step": 33364, "epoch": 926, "lr": 0.001} +{"train_loss": 0.026514725759625435, "train_loss_bc": 0.026502078399062157, "train_loss_llm": 0.0012646615505218506, "grad_norm": 0.036961521953344345, "global_step": 33365, "epoch": 926, "lr": 0.001} +{"train_loss": 0.02715604193508625, "train_loss_bc": 0.02715604193508625, "train_loss_llm": 0.0, "grad_norm": 0.0657954066991806, "global_step": 33366, "epoch": 926, "lr": 0.001} +{"train_loss": 0.031630147248506546, "train_loss_bc": 0.031630147248506546, "train_loss_llm": 0.0, "grad_norm": 0.08836249262094498, "global_step": 33367, "epoch": 926, "lr": 0.001} +{"train_loss": 0.02740822173655033, "train_loss_bc": 0.02740822173655033, "train_loss_llm": 0.0, "grad_norm": 0.11379209905862808, "global_step": 33368, "epoch": 926, "lr": 0.001} +{"train_loss": 0.02861321158707142, "train_loss_bc": 0.02861321158707142, "train_loss_llm": 0.0, "grad_norm": 0.0452585443854332, "global_step": 33369, "epoch": 926, "lr": 0.001} +{"train_loss": 0.0357326939702034, "train_loss_bc": 0.0357326939702034, "train_loss_llm": 0.0, "grad_norm": 0.07355336099863052, "global_step": 33370, "epoch": 926, "lr": 0.001} +{"train_loss": 0.029560419368661113, "train_loss_bc": 0.03172402083873749, "train_loss_llm": 0.0, "grad_norm": 0.09978687018156052, "global_step": 33371, "epoch": 926, "lr": 0.001, "val_loss": 0.10769683867692947} +{"train_loss": 0.029123593121767044, "train_loss_bc": 0.029123593121767044, "train_loss_llm": 0.0, "grad_norm": 0.12387700378894806, "global_step": 33372, "epoch": 927, "lr": 0.001} +{"train_loss": 0.032133057713508606, "train_loss_bc": 0.032133057713508606, "train_loss_llm": 0.0, "grad_norm": 0.05050962045788765, "global_step": 33373, "epoch": 927, "lr": 0.001} +{"train_loss": 0.02799282595515251, "train_loss_bc": 0.02798621356487274, "train_loss_llm": 0.0006611943244934082, "grad_norm": 0.07420355081558228, "global_step": 33374, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03366134315729141, "train_loss_bc": 0.03366012126207352, "train_loss_llm": 0.0001220405101776123, "grad_norm": 0.10685952752828598, "global_step": 33375, "epoch": 927, "lr": 0.001} +{"train_loss": 0.027880918234586716, "train_loss_bc": 0.027880918234586716, "train_loss_llm": 0.0, "grad_norm": 0.13801106810569763, "global_step": 33376, "epoch": 927, "lr": 0.001} +{"train_loss": 0.028748556971549988, "train_loss_bc": 0.028748556971549988, "train_loss_llm": 0.0, "grad_norm": 0.040397435426712036, "global_step": 33377, "epoch": 927, "lr": 0.001} +{"train_loss": 0.02904089353978634, "train_loss_bc": 0.02904089353978634, "train_loss_llm": 0.0, "grad_norm": 0.07081553339958191, "global_step": 33378, "epoch": 927, "lr": 0.001} +{"train_loss": 0.029045240953564644, "train_loss_bc": 0.029045240953564644, "train_loss_llm": 0.0, "grad_norm": 0.11567975580692291, "global_step": 33379, "epoch": 927, "lr": 0.001} +{"train_loss": 0.032132286578416824, "train_loss_bc": 0.032130464911460876, "train_loss_llm": 0.00018215179443359375, "grad_norm": 0.15347690880298615, "global_step": 33380, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03168246150016785, "train_loss_bc": 0.031676214188337326, "train_loss_llm": 0.0006246864795684814, "grad_norm": 0.052152812480926514, "global_step": 33381, "epoch": 927, "lr": 0.001} +{"train_loss": 0.027877554297447205, "train_loss_bc": 0.027877554297447205, "train_loss_llm": 0.0, "grad_norm": 0.0796126276254654, "global_step": 33382, "epoch": 927, "lr": 0.001} +{"train_loss": 0.033861350268125534, "train_loss_bc": 0.033861350268125534, "train_loss_llm": 0.0, "grad_norm": 0.12658123672008514, "global_step": 33383, "epoch": 927, "lr": 0.001} +{"train_loss": 0.024326978251338005, "train_loss_bc": 0.024326978251338005, "train_loss_llm": 0.0, "grad_norm": 0.15864317119121552, "global_step": 33384, "epoch": 927, "lr": 0.001} +{"train_loss": 0.029132364317774773, "train_loss_bc": 0.029132364317774773, "train_loss_llm": 0.0, "grad_norm": 0.036895424127578735, "global_step": 33385, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03266475722193718, "train_loss_bc": 0.03266475722193718, "train_loss_llm": 0.0, "grad_norm": 0.0712825357913971, "global_step": 33386, "epoch": 927, "lr": 0.001} +{"train_loss": 0.024944884702563286, "train_loss_bc": 0.024944884702563286, "train_loss_llm": 0.0, "grad_norm": 0.09571342915296555, "global_step": 33387, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03023637644946575, "train_loss_bc": 0.03023376315832138, "train_loss_llm": 0.0002612471580505371, "grad_norm": 0.1218223124742508, "global_step": 33388, "epoch": 927, "lr": 0.001} +{"train_loss": 0.026391830295324326, "train_loss_bc": 0.026391830295324326, "train_loss_llm": 0.0, "grad_norm": 0.04572354629635811, "global_step": 33389, "epoch": 927, "lr": 0.001} +{"train_loss": 0.029520981013774872, "train_loss_bc": 0.029512692242860794, "train_loss_llm": 0.0008288323879241943, "grad_norm": 0.06649219244718552, "global_step": 33390, "epoch": 927, "lr": 0.001} +{"train_loss": 0.02984088845551014, "train_loss_bc": 0.02984088845551014, "train_loss_llm": 0.0, "grad_norm": 0.0911182090640068, "global_step": 33391, "epoch": 927, "lr": 0.001} +{"train_loss": 0.027725478634238243, "train_loss_bc": 0.027725478634238243, "train_loss_llm": 0.0, "grad_norm": 0.10084687173366547, "global_step": 33392, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03008497692644596, "train_loss_bc": 0.03008497692644596, "train_loss_llm": 0.0, "grad_norm": 0.04687019810080528, "global_step": 33393, "epoch": 927, "lr": 0.001} +{"train_loss": 0.028885003179311752, "train_loss_bc": 0.028885003179311752, "train_loss_llm": 0.0, "grad_norm": 0.07328727841377258, "global_step": 33394, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03139542043209076, "train_loss_bc": 0.03139542043209076, "train_loss_llm": 0.0, "grad_norm": 0.11261975765228271, "global_step": 33395, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03440229222178459, "train_loss_bc": 0.03440229222178459, "train_loss_llm": 0.0, "grad_norm": 0.14322887361049652, "global_step": 33396, "epoch": 927, "lr": 0.001} +{"train_loss": 0.027123408392071724, "train_loss_bc": 0.027123408392071724, "train_loss_llm": 0.0, "grad_norm": 0.052706703543663025, "global_step": 33397, "epoch": 927, "lr": 0.001} +{"train_loss": 0.027886107563972473, "train_loss_bc": 0.027863746508955956, "train_loss_llm": 0.0022360384464263916, "grad_norm": 0.09000670164823532, "global_step": 33398, "epoch": 927, "lr": 0.001} +{"train_loss": 0.024275457486510277, "train_loss_bc": 0.024262018501758575, "train_loss_llm": 0.001343846321105957, "grad_norm": 0.12948234379291534, "global_step": 33399, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03356166556477547, "train_loss_bc": 0.03356166556477547, "train_loss_llm": 0.0, "grad_norm": 0.1563451588153839, "global_step": 33400, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03192963823676109, "train_loss_bc": 0.03191472217440605, "train_loss_llm": 0.0014916062355041504, "grad_norm": 0.057849615812301636, "global_step": 33401, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03327934816479683, "train_loss_bc": 0.03327934816479683, "train_loss_llm": 0.0, "grad_norm": 0.11589686572551727, "global_step": 33402, "epoch": 927, "lr": 0.001} +{"train_loss": 0.029776956886053085, "train_loss_bc": 0.029762431979179382, "train_loss_llm": 0.0014525651931762695, "grad_norm": 0.16704164445400238, "global_step": 33403, "epoch": 927, "lr": 0.001} +{"train_loss": 0.03251776844263077, "train_loss_bc": 0.03250935673713684, "train_loss_llm": 0.0008411109447479248, "grad_norm": 0.21131722629070282, "global_step": 33404, "epoch": 927, "lr": 0.001} +{"train_loss": 0.027355192229151726, "train_loss_bc": 0.027355192229151726, "train_loss_llm": 0.0, "grad_norm": 0.028703222051262856, "global_step": 33405, "epoch": 927, "lr": 0.001} +{"train_loss": 0.032220982015132904, "train_loss_bc": 0.032220982015132904, "train_loss_llm": 0.0, "grad_norm": 0.059078775346279144, "global_step": 33406, "epoch": 927, "lr": 0.001} +{"train_loss": 0.02977203174183766, "train_loss_bc": 0.029134303331375122, "train_loss_llm": 0.0, "grad_norm": 0.08066966384649277, "global_step": 33407, "epoch": 927, "lr": 0.001, "val_loss": 0.10644451528787613} +{"train_loss": 0.030931957066059113, "train_loss_bc": 0.030931957066059113, "train_loss_llm": 0.0, "grad_norm": 0.10490768402814865, "global_step": 33408, "epoch": 928, "lr": 0.001} +{"train_loss": 0.02762662060558796, "train_loss_bc": 0.02762662060558796, "train_loss_llm": 0.0, "grad_norm": 0.03417742997407913, "global_step": 33409, "epoch": 928, "lr": 0.001} +{"train_loss": 0.02740045078098774, "train_loss_bc": 0.02740045078098774, "train_loss_llm": 0.0, "grad_norm": 0.06970111280679703, "global_step": 33410, "epoch": 928, "lr": 0.001} +{"train_loss": 0.02923543192446232, "train_loss_bc": 0.02923543192446232, "train_loss_llm": 0.0, "grad_norm": 0.08108927309513092, "global_step": 33411, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03061326965689659, "train_loss_bc": 0.030611813068389893, "train_loss_llm": 0.00014570355415344238, "grad_norm": 0.10227925330400467, "global_step": 33412, "epoch": 928, "lr": 0.001} +{"train_loss": 0.02747063711285591, "train_loss_bc": 0.02747063711285591, "train_loss_llm": 0.0, "grad_norm": 0.03960958495736122, "global_step": 33413, "epoch": 928, "lr": 0.001} +{"train_loss": 0.02659522369503975, "train_loss_bc": 0.02659522369503975, "train_loss_llm": 0.0, "grad_norm": 0.07497652620077133, "global_step": 33414, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03228865936398506, "train_loss_bc": 0.032286010682582855, "train_loss_llm": 0.00026470422744750977, "grad_norm": 0.12006718665361404, "global_step": 33415, "epoch": 928, "lr": 0.001} +{"train_loss": 0.032060593366622925, "train_loss_bc": 0.032060593366622925, "train_loss_llm": 0.0, "grad_norm": 0.16920973360538483, "global_step": 33416, "epoch": 928, "lr": 0.001} +{"train_loss": 0.026919720694422722, "train_loss_bc": 0.026919720694422722, "train_loss_llm": 0.0, "grad_norm": 0.059082549065351486, "global_step": 33417, "epoch": 928, "lr": 0.001} +{"train_loss": 0.02885851263999939, "train_loss_bc": 0.02885851263999939, "train_loss_llm": 0.0, "grad_norm": 0.08661495894193649, "global_step": 33418, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03310832381248474, "train_loss_bc": 0.03310832381248474, "train_loss_llm": 0.0, "grad_norm": 0.11293158680200577, "global_step": 33419, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03127625212073326, "train_loss_bc": 0.03125885874032974, "train_loss_llm": 0.0017394721508026123, "grad_norm": 0.15507522225379944, "global_step": 33420, "epoch": 928, "lr": 0.001} +{"train_loss": 0.027947748079895973, "train_loss_bc": 0.027947748079895973, "train_loss_llm": 0.0, "grad_norm": 0.03625446930527687, "global_step": 33421, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03200706094503403, "train_loss_bc": 0.03200706094503403, "train_loss_llm": 0.0, "grad_norm": 0.05634787678718567, "global_step": 33422, "epoch": 928, "lr": 0.001} +{"train_loss": 0.029940631240606308, "train_loss_bc": 0.029940631240606308, "train_loss_llm": 0.0, "grad_norm": 0.08171997219324112, "global_step": 33423, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03107285499572754, "train_loss_bc": 0.03107285499572754, "train_loss_llm": 0.0, "grad_norm": 0.11583263427019119, "global_step": 33424, "epoch": 928, "lr": 0.001} +{"train_loss": 0.02935454621911049, "train_loss_bc": 0.029351383447647095, "train_loss_llm": 0.00031635165214538574, "grad_norm": 0.04387323558330536, "global_step": 33425, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03046189248561859, "train_loss_bc": 0.03046189248561859, "train_loss_llm": 0.0, "grad_norm": 0.0724373310804367, "global_step": 33426, "epoch": 928, "lr": 0.001} +{"train_loss": 0.02556336298584938, "train_loss_bc": 0.025556573644280434, "train_loss_llm": 0.00067901611328125, "grad_norm": 0.09998950362205505, "global_step": 33427, "epoch": 928, "lr": 0.001} +{"train_loss": 0.02335045486688614, "train_loss_bc": 0.023347878828644753, "train_loss_llm": 0.0002575516700744629, "grad_norm": 0.12801040709018707, "global_step": 33428, "epoch": 928, "lr": 0.001} +{"train_loss": 0.026465540751814842, "train_loss_bc": 0.026465540751814842, "train_loss_llm": 0.0, "grad_norm": 0.04007155820727348, "global_step": 33429, "epoch": 928, "lr": 0.001} +{"train_loss": 0.030803322792053223, "train_loss_bc": 0.030803322792053223, "train_loss_llm": 0.0, "grad_norm": 0.05620963126420975, "global_step": 33430, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03325466439127922, "train_loss_bc": 0.03325466439127922, "train_loss_llm": 0.0, "grad_norm": 0.08457737416028976, "global_step": 33431, "epoch": 928, "lr": 0.001} +{"train_loss": 0.026294955983757973, "train_loss_bc": 0.026294955983757973, "train_loss_llm": 0.0, "grad_norm": 0.10225948691368103, "global_step": 33432, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03458331152796745, "train_loss_bc": 0.03458331152796745, "train_loss_llm": 0.0, "grad_norm": 0.03733155131340027, "global_step": 33433, "epoch": 928, "lr": 0.001} +{"train_loss": 0.025185123085975647, "train_loss_bc": 0.025175193324685097, "train_loss_llm": 0.0009930431842803955, "grad_norm": 0.055729497224092484, "global_step": 33434, "epoch": 928, "lr": 0.001} +{"train_loss": 0.017988895997405052, "train_loss_bc": 0.017988895997405052, "train_loss_llm": 0.0, "grad_norm": 0.07652601599693298, "global_step": 33435, "epoch": 928, "lr": 0.001} +{"train_loss": 0.028934521600604057, "train_loss_bc": 0.028934521600604057, "train_loss_llm": 0.0, "grad_norm": 0.10882832854986191, "global_step": 33436, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03020154871046543, "train_loss_bc": 0.03020154871046543, "train_loss_llm": 0.0, "grad_norm": 0.044717103242874146, "global_step": 33437, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03576704487204552, "train_loss_bc": 0.03576704487204552, "train_loss_llm": 0.0, "grad_norm": 0.08389227092266083, "global_step": 33438, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03188742697238922, "train_loss_bc": 0.03188742697238922, "train_loss_llm": 0.0, "grad_norm": 0.11709627509117126, "global_step": 33439, "epoch": 928, "lr": 0.001} +{"train_loss": 0.027575291693210602, "train_loss_bc": 0.027575291693210602, "train_loss_llm": 0.0, "grad_norm": 0.14938423037528992, "global_step": 33440, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03818265721201897, "train_loss_bc": 0.03818265721201897, "train_loss_llm": 0.0, "grad_norm": 0.05965150147676468, "global_step": 33441, "epoch": 928, "lr": 0.001} +{"train_loss": 0.03305552154779434, "train_loss_bc": 0.03305552154779434, "train_loss_llm": 0.0, "grad_norm": 0.09930914640426636, "global_step": 33442, "epoch": 928, "lr": 0.001} +{"train_loss": 0.029480181427465543, "train_loss_bc": 0.026997080072760582, "train_loss_llm": 0.0025419294834136963, "grad_norm": 0.13294436037540436, "global_step": 33443, "epoch": 928, "lr": 0.001, "val_loss": 0.10704606026411057} +{"train_loss": 0.027501778677105904, "train_loss_bc": 0.02750147320330143, "train_loss_llm": 3.057718276977539e-05, "grad_norm": 0.15937328338623047, "global_step": 33444, "epoch": 929, "lr": 0.001} +{"train_loss": 0.02904493920505047, "train_loss_bc": 0.02904493920505047, "train_loss_llm": 0.0, "grad_norm": 0.03630693629384041, "global_step": 33445, "epoch": 929, "lr": 0.001} +{"train_loss": 0.02355651557445526, "train_loss_bc": 0.02355651557445526, "train_loss_llm": 0.0, "grad_norm": 0.05324357748031616, "global_step": 33446, "epoch": 929, "lr": 0.001} +{"train_loss": 0.027052314952015877, "train_loss_bc": 0.027052314952015877, "train_loss_llm": 0.0, "grad_norm": 0.07715524733066559, "global_step": 33447, "epoch": 929, "lr": 0.001} +{"train_loss": 0.02961302362382412, "train_loss_bc": 0.02961302362382412, "train_loss_llm": 0.0, "grad_norm": 0.09810607135295868, "global_step": 33448, "epoch": 929, "lr": 0.001} +{"train_loss": 0.029948461800813675, "train_loss_bc": 0.029948461800813675, "train_loss_llm": 0.0, "grad_norm": 0.03720923513174057, "global_step": 33449, "epoch": 929, "lr": 0.001} +{"train_loss": 0.0328352153301239, "train_loss_bc": 0.0328352153301239, "train_loss_llm": 0.0, "grad_norm": 0.06206964701414108, "global_step": 33450, "epoch": 929, "lr": 0.001} +{"train_loss": 0.029564151540398598, "train_loss_bc": 0.029554249718785286, "train_loss_llm": 0.0009902119636535645, "grad_norm": 0.07523441314697266, "global_step": 33451, "epoch": 929, "lr": 0.001} +{"train_loss": 0.03168395906686783, "train_loss_bc": 0.03168395906686783, "train_loss_llm": 0.0, "grad_norm": 0.10572340339422226, "global_step": 33452, "epoch": 929, "lr": 0.001} +{"train_loss": 0.027710219845175743, "train_loss_bc": 0.027710219845175743, "train_loss_llm": 0.0, "grad_norm": 0.04183702543377876, "global_step": 33453, "epoch": 929, "lr": 0.001} +{"train_loss": 0.03198162466287613, "train_loss_bc": 0.03198162466287613, "train_loss_llm": 0.0, "grad_norm": 0.05775035545229912, "global_step": 33454, "epoch": 929, "lr": 0.001} +{"train_loss": 0.030853373929858208, "train_loss_bc": 0.030841665342450142, "train_loss_llm": 0.0011708736419677734, "grad_norm": 0.09038664400577545, "global_step": 33455, "epoch": 929, "lr": 0.001} +{"train_loss": 0.023659491911530495, "train_loss_bc": 0.023659491911530495, "train_loss_llm": 0.0, "grad_norm": 0.10529366135597229, "global_step": 33456, "epoch": 929, "lr": 0.001} +{"train_loss": 0.03345556557178497, "train_loss_bc": 0.03345378488302231, "train_loss_llm": 0.0001780986785888672, "grad_norm": 0.053298357874155045, "global_step": 33457, "epoch": 929, "lr": 0.001} +{"train_loss": 0.030699584633111954, "train_loss_bc": 0.030699584633111954, "train_loss_llm": 0.0, "grad_norm": 0.0874268114566803, "global_step": 33458, "epoch": 929, "lr": 0.001} +{"train_loss": 0.031099244952201843, "train_loss_bc": 0.031099244952201843, "train_loss_llm": 0.0, "grad_norm": 0.11530273407697678, "global_step": 33459, "epoch": 929, "lr": 0.001} +{"train_loss": 0.03140858933329582, "train_loss_bc": 0.03139369189739227, "train_loss_llm": 0.001489788293838501, "grad_norm": 0.14259566366672516, "global_step": 33460, "epoch": 929, "lr": 0.001} +{"train_loss": 0.028449008241295815, "train_loss_bc": 0.028449008241295815, "train_loss_llm": 0.0, "grad_norm": 0.04631338268518448, "global_step": 33461, "epoch": 929, "lr": 0.001} +{"train_loss": 0.0283720213919878, "train_loss_bc": 0.0283720213919878, "train_loss_llm": 0.0, "grad_norm": 0.09014817327260971, "global_step": 33462, "epoch": 929, "lr": 0.001} +{"train_loss": 0.025198902934789658, "train_loss_bc": 0.025198902934789658, "train_loss_llm": 0.0, "grad_norm": 0.11313967406749725, "global_step": 33463, "epoch": 929, "lr": 0.001} +{"train_loss": 0.0291968435049057, "train_loss_bc": 0.0291968435049057, "train_loss_llm": 0.0, "grad_norm": 0.14667077362537384, "global_step": 33464, "epoch": 929, "lr": 0.001} +{"train_loss": 0.031164946034550667, "train_loss_bc": 0.031164946034550667, "train_loss_llm": 0.0, "grad_norm": 0.05069959536194801, "global_step": 33465, "epoch": 929, "lr": 0.001} +{"train_loss": 0.02409970946609974, "train_loss_bc": 0.02409970946609974, "train_loss_llm": 0.0, "grad_norm": 0.07472120970487595, "global_step": 33466, "epoch": 929, "lr": 0.001} +{"train_loss": 0.023628942668437958, "train_loss_bc": 0.023628942668437958, "train_loss_llm": 0.0, "grad_norm": 0.09466717392206192, "global_step": 33467, "epoch": 929, "lr": 0.001} +{"train_loss": 0.028094589710235596, "train_loss_bc": 0.028094589710235596, "train_loss_llm": 0.0, "grad_norm": 0.11999819427728653, "global_step": 33468, "epoch": 929, "lr": 0.001} +{"train_loss": 0.026874877512454987, "train_loss_bc": 0.026874877512454987, "train_loss_llm": 0.0, "grad_norm": 0.04463823139667511, "global_step": 33469, "epoch": 929, "lr": 0.001} +{"train_loss": 0.03295008838176727, "train_loss_bc": 0.03295008838176727, "train_loss_llm": 0.0, "grad_norm": 0.06577074527740479, "global_step": 33470, "epoch": 929, "lr": 0.001} +{"train_loss": 0.026801394298672676, "train_loss_bc": 0.026793936267495155, "train_loss_llm": 0.0007458329200744629, "grad_norm": 0.08567077666521072, "global_step": 33471, "epoch": 929, "lr": 0.001} +{"train_loss": 0.028164971619844437, "train_loss_bc": 0.028164971619844437, "train_loss_llm": 0.0, "grad_norm": 0.10860883444547653, "global_step": 33472, "epoch": 929, "lr": 0.001} +{"train_loss": 0.0319066159427166, "train_loss_bc": 0.03190436214208603, "train_loss_llm": 0.00022548437118530273, "grad_norm": 0.05037137120962143, "global_step": 33473, "epoch": 929, "lr": 0.001} +{"train_loss": 0.03875454142689705, "train_loss_bc": 0.03875454142689705, "train_loss_llm": 0.0, "grad_norm": 0.08626547455787659, "global_step": 33474, "epoch": 929, "lr": 0.001} +{"train_loss": 0.027949171140789986, "train_loss_bc": 0.027949171140789986, "train_loss_llm": 0.0, "grad_norm": 0.10425211489200592, "global_step": 33475, "epoch": 929, "lr": 0.001} +{"train_loss": 0.028500203043222427, "train_loss_bc": 0.028500203043222427, "train_loss_llm": 0.0, "grad_norm": 0.12939482927322388, "global_step": 33476, "epoch": 929, "lr": 0.001} +{"train_loss": 0.03497423976659775, "train_loss_bc": 0.03497333079576492, "train_loss_llm": 9.086728096008301e-05, "grad_norm": 0.04336429387331009, "global_step": 33477, "epoch": 929, "lr": 0.001} +{"train_loss": 0.027773907408118248, "train_loss_bc": 0.027773907408118248, "train_loss_llm": 0.0, "grad_norm": 0.06448467075824738, "global_step": 33478, "epoch": 929, "lr": 0.001} +{"train_loss": 0.029218640592363145, "train_loss_bc": 0.027348032221198082, "train_loss_llm": 0.0, "grad_norm": 0.08690119534730911, "global_step": 33479, "epoch": 929, "lr": 0.001, "val_loss": 0.10602221637964249} +{"train_loss": 0.024679772555828094, "train_loss_bc": 0.024679772555828094, "train_loss_llm": 0.0, "grad_norm": 0.10395646840333939, "global_step": 33480, "epoch": 930, "lr": 0.001} +{"train_loss": 0.03051331825554371, "train_loss_bc": 0.03051331825554371, "train_loss_llm": 0.0, "grad_norm": 0.043609485030174255, "global_step": 33481, "epoch": 930, "lr": 0.001} +{"train_loss": 0.022527122870087624, "train_loss_bc": 0.022509479895234108, "train_loss_llm": 0.0017642974853515625, "grad_norm": 0.06147557497024536, "global_step": 33482, "epoch": 930, "lr": 0.001} +{"train_loss": 0.03326347470283508, "train_loss_bc": 0.03326347470283508, "train_loss_llm": 0.0, "grad_norm": 0.08896327763795853, "global_step": 33483, "epoch": 930, "lr": 0.001} +{"train_loss": 0.029516376554965973, "train_loss_bc": 0.029516376554965973, "train_loss_llm": 0.0, "grad_norm": 0.11271476000547409, "global_step": 33484, "epoch": 930, "lr": 0.001} +{"train_loss": 0.03082296811044216, "train_loss_bc": 0.03082296811044216, "train_loss_llm": 0.0, "grad_norm": 0.042861584573984146, "global_step": 33485, "epoch": 930, "lr": 0.001} +{"train_loss": 0.02941022254526615, "train_loss_bc": 0.02941022254526615, "train_loss_llm": 0.0, "grad_norm": 0.05519557371735573, "global_step": 33486, "epoch": 930, "lr": 0.001} +{"train_loss": 0.029539084061980247, "train_loss_bc": 0.029539084061980247, "train_loss_llm": 0.0, "grad_norm": 0.0849800631403923, "global_step": 33487, "epoch": 930, "lr": 0.001} +{"train_loss": 0.029462696984410286, "train_loss_bc": 0.02945539727807045, "train_loss_llm": 0.0007300376892089844, "grad_norm": 0.10960129648447037, "global_step": 33488, "epoch": 930, "lr": 0.001} +{"train_loss": 0.031860802322626114, "train_loss_bc": 0.031860802322626114, "train_loss_llm": 0.0, "grad_norm": 0.047524236142635345, "global_step": 33489, "epoch": 930, "lr": 0.001} +{"train_loss": 0.02759883552789688, "train_loss_bc": 0.02759883552789688, "train_loss_llm": 0.0, "grad_norm": 0.0653878003358841, "global_step": 33490, "epoch": 930, "lr": 0.001} +{"train_loss": 0.02766554057598114, "train_loss_bc": 0.02766554057598114, "train_loss_llm": 0.0, "grad_norm": 0.08068045973777771, "global_step": 33491, "epoch": 930, "lr": 0.001} +{"train_loss": 0.02562929131090641, "train_loss_bc": 0.02562929131090641, "train_loss_llm": 0.0, "grad_norm": 0.08631919324398041, "global_step": 33492, "epoch": 930, "lr": 0.001} +{"train_loss": 0.028089353814721107, "train_loss_bc": 0.028086742386221886, "train_loss_llm": 0.0002611875534057617, "grad_norm": 0.036695510149002075, "global_step": 33493, "epoch": 930, "lr": 0.001} +{"train_loss": 0.03165286034345627, "train_loss_bc": 0.03165286034345627, "train_loss_llm": 0.0, "grad_norm": 0.04812874644994736, "global_step": 33494, "epoch": 930, "lr": 0.001} +{"train_loss": 0.024513056501746178, "train_loss_bc": 0.024504244327545166, "train_loss_llm": 0.0008812844753265381, "grad_norm": 0.0655921921133995, "global_step": 33495, "epoch": 930, "lr": 0.001} +{"train_loss": 0.027479058131575584, "train_loss_bc": 0.027479058131575584, "train_loss_llm": 0.0, "grad_norm": 0.08898228406906128, "global_step": 33496, "epoch": 930, "lr": 0.001} +{"train_loss": 0.028771089389920235, "train_loss_bc": 0.028771089389920235, "train_loss_llm": 0.0, "grad_norm": 0.03271142393350601, "global_step": 33497, "epoch": 930, "lr": 0.001} +{"train_loss": 0.028111159801483154, "train_loss_bc": 0.028111159801483154, "train_loss_llm": 0.0, "grad_norm": 0.059105679392814636, "global_step": 33498, "epoch": 930, "lr": 0.001} +{"train_loss": 0.030411293730139732, "train_loss_bc": 0.030403567478060722, "train_loss_llm": 0.0007725358009338379, "grad_norm": 0.07523591071367264, "global_step": 33499, "epoch": 930, "lr": 0.001} +{"train_loss": 0.03138720616698265, "train_loss_bc": 0.03138720616698265, "train_loss_llm": 0.0, "grad_norm": 0.09992919117212296, "global_step": 33500, "epoch": 930, "lr": 0.001} +{"train_loss": 0.02729499340057373, "train_loss_bc": 0.027280984446406364, "train_loss_llm": 0.0014008581638336182, "grad_norm": 0.04043569043278694, "global_step": 33501, "epoch": 930, "lr": 0.001} +{"train_loss": 0.024931475520133972, "train_loss_bc": 0.024916721507906914, "train_loss_llm": 0.0014753937721252441, "grad_norm": 0.05553781986236572, "global_step": 33502, "epoch": 930, "lr": 0.001} +{"train_loss": 0.029425932094454765, "train_loss_bc": 0.029425932094454765, "train_loss_llm": 0.0, "grad_norm": 0.08989901840686798, "global_step": 33503, "epoch": 930, "lr": 0.001} +{"train_loss": 0.02972571738064289, "train_loss_bc": 0.02972571738064289, "train_loss_llm": 0.0, "grad_norm": 0.11673250794410706, "global_step": 33504, "epoch": 930, "lr": 0.001} +{"train_loss": 0.033915042877197266, "train_loss_bc": 0.033915042877197266, "train_loss_llm": 0.0, "grad_norm": 0.04141277074813843, "global_step": 33505, "epoch": 930, "lr": 0.001} +{"train_loss": 0.028287025168538094, "train_loss_bc": 0.028287025168538094, "train_loss_llm": 0.0, "grad_norm": 0.0666663721203804, "global_step": 33506, "epoch": 930, "lr": 0.001} +{"train_loss": 0.029708847403526306, "train_loss_bc": 0.029708847403526306, "train_loss_llm": 0.0, "grad_norm": 0.08637413382530212, "global_step": 33507, "epoch": 930, "lr": 0.001} +{"train_loss": 0.026577524840831757, "train_loss_bc": 0.026577524840831757, "train_loss_llm": 0.0, "grad_norm": 0.107867032289505, "global_step": 33508, "epoch": 930, "lr": 0.001} +{"train_loss": 0.035714101046323776, "train_loss_bc": 0.03570817783474922, "train_loss_llm": 0.000592350959777832, "grad_norm": 0.05110781267285347, "global_step": 33509, "epoch": 930, "lr": 0.001} +{"train_loss": 0.03393034264445305, "train_loss_bc": 0.033926092088222504, "train_loss_llm": 0.00042501091957092285, "grad_norm": 0.08982145041227341, "global_step": 33510, "epoch": 930, "lr": 0.001} +{"train_loss": 0.03167854622006416, "train_loss_bc": 0.03167854622006416, "train_loss_llm": 0.0, "grad_norm": 0.10373445600271225, "global_step": 33511, "epoch": 930, "lr": 0.001} +{"train_loss": 0.028373075649142265, "train_loss_bc": 0.028373075649142265, "train_loss_llm": 0.0, "grad_norm": 0.11806502938270569, "global_step": 33512, "epoch": 930, "lr": 0.001} +{"train_loss": 0.028720349073410034, "train_loss_bc": 0.028720349073410034, "train_loss_llm": 0.0, "grad_norm": 0.03671661764383316, "global_step": 33513, "epoch": 930, "lr": 0.001} +{"train_loss": 0.021639224141836166, "train_loss_bc": 0.021615441888570786, "train_loss_llm": 0.002378225326538086, "grad_norm": 0.05204710364341736, "global_step": 33514, "epoch": 930, "lr": 0.001} +{"train_loss": 0.029040011370347604, "train_loss_bc": 0.03261362761259079, "train_loss_llm": 0.0, "grad_norm": 0.07462497055530548, "global_step": 33515, "epoch": 930, "lr": 0.001, "train/cumulative_reward": 74.55649228206434, "train/mean_score": 16.83507995322808, "train/success_rate": 0.4, "test/cumulative_reward": 50.93194344440968, "test/mean_score": 8.182308186142633, "test/success_rate": 0.2, "val_loss": 0.10562796890735626, "train_action_mse_error": 0.0317860022187233} +{"train_loss": 0.027904599905014038, "train_loss_bc": 0.027904599905014038, "train_loss_llm": 0.0, "grad_norm": 0.07829492539167404, "global_step": 33516, "epoch": 931, "lr": 0.001} +{"train_loss": 0.029633568599820137, "train_loss_bc": 0.029633568599820137, "train_loss_llm": 0.0, "grad_norm": 0.032837286591529846, "global_step": 33517, "epoch": 931, "lr": 0.001} +{"train_loss": 0.035950981080532074, "train_loss_bc": 0.035950981080532074, "train_loss_llm": 0.0, "grad_norm": 0.05373086780309677, "global_step": 33518, "epoch": 931, "lr": 0.001} +{"train_loss": 0.034947771579027176, "train_loss_bc": 0.034947771579027176, "train_loss_llm": 0.0, "grad_norm": 0.0671350285410881, "global_step": 33519, "epoch": 931, "lr": 0.001} +{"train_loss": 0.029953747987747192, "train_loss_bc": 0.029923100024461746, "train_loss_llm": 0.0030647218227386475, "grad_norm": 0.08168394863605499, "global_step": 33520, "epoch": 931, "lr": 0.001} +{"train_loss": 0.027281280606985092, "train_loss_bc": 0.027281280606985092, "train_loss_llm": 0.0, "grad_norm": 0.04227996617555618, "global_step": 33521, "epoch": 931, "lr": 0.001} +{"train_loss": 0.02858029492199421, "train_loss_bc": 0.02858029492199421, "train_loss_llm": 0.0, "grad_norm": 0.05941256508231163, "global_step": 33522, "epoch": 931, "lr": 0.001} +{"train_loss": 0.030161214992403984, "train_loss_bc": 0.030161214992403984, "train_loss_llm": 0.0, "grad_norm": 0.07881014794111252, "global_step": 33523, "epoch": 931, "lr": 0.001} +{"train_loss": 0.03660960868000984, "train_loss_bc": 0.03660960868000984, "train_loss_llm": 0.0, "grad_norm": 0.10562536120414734, "global_step": 33524, "epoch": 931, "lr": 0.001} +{"train_loss": 0.031339190900325775, "train_loss_bc": 0.031339190900325775, "train_loss_llm": 0.0, "grad_norm": 0.04893561452627182, "global_step": 33525, "epoch": 931, "lr": 0.001} +{"train_loss": 0.026508308947086334, "train_loss_bc": 0.026508308947086334, "train_loss_llm": 0.0, "grad_norm": 0.06793041527271271, "global_step": 33526, "epoch": 931, "lr": 0.001} +{"train_loss": 0.027027063071727753, "train_loss_bc": 0.027021102607250214, "train_loss_llm": 0.0005960762500762939, "grad_norm": 0.08696004748344421, "global_step": 33527, "epoch": 931, "lr": 0.001} +{"train_loss": 0.025858554989099503, "train_loss_bc": 0.025858554989099503, "train_loss_llm": 0.0, "grad_norm": 0.11354648321866989, "global_step": 33528, "epoch": 931, "lr": 0.001} +{"train_loss": 0.031920596957206726, "train_loss_bc": 0.03190767019987106, "train_loss_llm": 0.0012925565242767334, "grad_norm": 0.039799243211746216, "global_step": 33529, "epoch": 931, "lr": 0.001} +{"train_loss": 0.03131112456321716, "train_loss_bc": 0.03131112456321716, "train_loss_llm": 0.0, "grad_norm": 0.05793694779276848, "global_step": 33530, "epoch": 931, "lr": 0.001} +{"train_loss": 0.02751464955508709, "train_loss_bc": 0.02751464955508709, "train_loss_llm": 0.0, "grad_norm": 0.0764038935303688, "global_step": 33531, "epoch": 931, "lr": 0.001} +{"train_loss": 0.03148782253265381, "train_loss_bc": 0.03148782253265381, "train_loss_llm": 0.0, "grad_norm": 0.08694860339164734, "global_step": 33532, "epoch": 931, "lr": 0.001} +{"train_loss": 0.0264793299138546, "train_loss_bc": 0.0264793299138546, "train_loss_llm": 0.0, "grad_norm": 0.03433545306324959, "global_step": 33533, "epoch": 931, "lr": 0.001} +{"train_loss": 0.0268167182803154, "train_loss_bc": 0.0268167182803154, "train_loss_llm": 0.0, "grad_norm": 0.048906359821558, "global_step": 33534, "epoch": 931, "lr": 0.001} +{"train_loss": 0.03198230266571045, "train_loss_bc": 0.03198230266571045, "train_loss_llm": 0.0, "grad_norm": 0.06970232725143433, "global_step": 33535, "epoch": 931, "lr": 0.001} +{"train_loss": 0.02874777838587761, "train_loss_bc": 0.02874777838587761, "train_loss_llm": 0.0, "grad_norm": 0.08429103344678879, "global_step": 33536, "epoch": 931, "lr": 0.001} +{"train_loss": 0.025380674749612808, "train_loss_bc": 0.025380674749612808, "train_loss_llm": 0.0, "grad_norm": 0.03867565095424652, "global_step": 33537, "epoch": 931, "lr": 0.001} +{"train_loss": 0.030284211039543152, "train_loss_bc": 0.030284211039543152, "train_loss_llm": 0.0, "grad_norm": 0.05968603119254112, "global_step": 33538, "epoch": 931, "lr": 0.001} +{"train_loss": 0.029207950457930565, "train_loss_bc": 0.02920129895210266, "train_loss_llm": 0.0006652176380157471, "grad_norm": 0.07343940436840057, "global_step": 33539, "epoch": 931, "lr": 0.001} +{"train_loss": 0.026939786970615387, "train_loss_bc": 0.02693745866417885, "train_loss_llm": 0.00023287534713745117, "grad_norm": 0.0939062312245369, "global_step": 33540, "epoch": 931, "lr": 0.001} +{"train_loss": 0.02771052159368992, "train_loss_bc": 0.02770565077662468, "train_loss_llm": 0.0004871487617492676, "grad_norm": 0.03915375471115112, "global_step": 33541, "epoch": 931, "lr": 0.001} +{"train_loss": 0.026609113439917564, "train_loss_bc": 0.026608942076563835, "train_loss_llm": 1.71661376953125e-05, "grad_norm": 0.059829529374837875, "global_step": 33542, "epoch": 931, "lr": 0.001} +{"train_loss": 0.024733204394578934, "train_loss_bc": 0.024733204394578934, "train_loss_llm": 0.0, "grad_norm": 0.08008631318807602, "global_step": 33543, "epoch": 931, "lr": 0.001} +{"train_loss": 0.0247041042894125, "train_loss_bc": 0.0247041042894125, "train_loss_llm": 0.0, "grad_norm": 0.09399871528148651, "global_step": 33544, "epoch": 931, "lr": 0.001} +{"train_loss": 0.031169671565294266, "train_loss_bc": 0.031169671565294266, "train_loss_llm": 0.0, "grad_norm": 0.03892241418361664, "global_step": 33545, "epoch": 931, "lr": 0.001} +{"train_loss": 0.024016929790377617, "train_loss_bc": 0.024016929790377617, "train_loss_llm": 0.0, "grad_norm": 0.06533218175172806, "global_step": 33546, "epoch": 931, "lr": 0.001} +{"train_loss": 0.02755281887948513, "train_loss_bc": 0.027548525482416153, "train_loss_llm": 0.00042933225631713867, "grad_norm": 0.08838512748479843, "global_step": 33547, "epoch": 931, "lr": 0.001} +{"train_loss": 0.024267323315143585, "train_loss_bc": 0.024265840649604797, "train_loss_llm": 0.00014826655387878418, "grad_norm": 0.10582023859024048, "global_step": 33548, "epoch": 931, "lr": 0.001} +{"train_loss": 0.029458630830049515, "train_loss_bc": 0.029458630830049515, "train_loss_llm": 0.0, "grad_norm": 0.051849618554115295, "global_step": 33549, "epoch": 931, "lr": 0.001} +{"train_loss": 0.030187154188752174, "train_loss_bc": 0.030180523172020912, "train_loss_llm": 0.0006631016731262207, "grad_norm": 0.06986618787050247, "global_step": 33550, "epoch": 931, "lr": 0.001} +{"train_loss": 0.028872090133113995, "train_loss_bc": 0.02915254980325699, "train_loss_llm": 0.0004089474678039551, "grad_norm": 0.09872311353683472, "global_step": 33551, "epoch": 931, "lr": 0.001, "val_loss": 0.10519115626811981} +{"train_loss": 0.028002943843603134, "train_loss_bc": 0.028002943843603134, "train_loss_llm": 0.0, "grad_norm": 0.1061842292547226, "global_step": 33552, "epoch": 932, "lr": 0.001} +{"train_loss": 0.03137858957052231, "train_loss_bc": 0.03137858957052231, "train_loss_llm": 0.0, "grad_norm": 0.04323126748204231, "global_step": 33553, "epoch": 932, "lr": 0.001} +{"train_loss": 0.02645757980644703, "train_loss_bc": 0.02644697204232216, "train_loss_llm": 0.0010607540607452393, "grad_norm": 0.05687999352812767, "global_step": 33554, "epoch": 932, "lr": 0.001} +{"train_loss": 0.02628784254193306, "train_loss_bc": 0.02628784254193306, "train_loss_llm": 0.0, "grad_norm": 0.07309451699256897, "global_step": 33555, "epoch": 932, "lr": 0.001} +{"train_loss": 0.02621770091354847, "train_loss_bc": 0.02621770091354847, "train_loss_llm": 0.0, "grad_norm": 0.08877017349004745, "global_step": 33556, "epoch": 932, "lr": 0.001} +{"train_loss": 0.030458103865385056, "train_loss_bc": 0.030451858416199684, "train_loss_llm": 0.0006246268749237061, "grad_norm": 0.03784212842583656, "global_step": 33557, "epoch": 932, "lr": 0.001} +{"train_loss": 0.02313585765659809, "train_loss_bc": 0.02313585765659809, "train_loss_llm": 0.0, "grad_norm": 0.05950194224715233, "global_step": 33558, "epoch": 932, "lr": 0.001} +{"train_loss": 0.030614372342824936, "train_loss_bc": 0.030614269897341728, "train_loss_llm": 1.0162591934204102e-05, "grad_norm": 0.0787559226155281, "global_step": 33559, "epoch": 932, "lr": 0.001} +{"train_loss": 0.03129660338163376, "train_loss_bc": 0.03129660338163376, "train_loss_llm": 0.0, "grad_norm": 0.10817773640155792, "global_step": 33560, "epoch": 932, "lr": 0.001} +{"train_loss": 0.029306946322321892, "train_loss_bc": 0.029303794726729393, "train_loss_llm": 0.00031512975692749023, "grad_norm": 0.035339031368494034, "global_step": 33561, "epoch": 932, "lr": 0.001} +{"train_loss": 0.029182717204093933, "train_loss_bc": 0.029182717204093933, "train_loss_llm": 0.0, "grad_norm": 0.05579570680856705, "global_step": 33562, "epoch": 932, "lr": 0.001} +{"train_loss": 0.03447660431265831, "train_loss_bc": 0.03447660431265831, "train_loss_llm": 0.0, "grad_norm": 0.0824173092842102, "global_step": 33563, "epoch": 932, "lr": 0.001} +{"train_loss": 0.028065340593457222, "train_loss_bc": 0.028065340593457222, "train_loss_llm": 0.0, "grad_norm": 0.10661227256059647, "global_step": 33564, "epoch": 932, "lr": 0.001} +{"train_loss": 0.0320703350007534, "train_loss_bc": 0.03205632045865059, "train_loss_llm": 0.0014016330242156982, "grad_norm": 0.04178536310791969, "global_step": 33565, "epoch": 932, "lr": 0.001} +{"train_loss": 0.023286251351237297, "train_loss_bc": 0.02327636256814003, "train_loss_llm": 0.0009889304637908936, "grad_norm": 0.06029167398810387, "global_step": 33566, "epoch": 932, "lr": 0.001} +{"train_loss": 0.03322489187121391, "train_loss_bc": 0.03322431072592735, "train_loss_llm": 5.8025121688842773e-05, "grad_norm": 0.08206818252801895, "global_step": 33567, "epoch": 932, "lr": 0.001} +{"train_loss": 0.02679743431508541, "train_loss_bc": 0.02679743431508541, "train_loss_llm": 0.0, "grad_norm": 0.1008898988366127, "global_step": 33568, "epoch": 932, "lr": 0.001} +{"train_loss": 0.030811196193099022, "train_loss_bc": 0.030811196193099022, "train_loss_llm": 0.0, "grad_norm": 0.05049464479088783, "global_step": 33569, "epoch": 932, "lr": 0.001} +{"train_loss": 0.03185752034187317, "train_loss_bc": 0.03185752034187317, "train_loss_llm": 0.0, "grad_norm": 0.0672874003648758, "global_step": 33570, "epoch": 932, "lr": 0.001} +{"train_loss": 0.028530841693282127, "train_loss_bc": 0.028530841693282127, "train_loss_llm": 0.0, "grad_norm": 0.08672886341810226, "global_step": 33571, "epoch": 932, "lr": 0.001} +{"train_loss": 0.031368356198072433, "train_loss_bc": 0.031368356198072433, "train_loss_llm": 0.0, "grad_norm": 0.11350643634796143, "global_step": 33572, "epoch": 932, "lr": 0.001} +{"train_loss": 0.022336620837450027, "train_loss_bc": 0.02231527306139469, "train_loss_llm": 0.0021347999572753906, "grad_norm": 0.02852770872414112, "global_step": 33573, "epoch": 932, "lr": 0.001} +{"train_loss": 0.029680058360099792, "train_loss_bc": 0.029680058360099792, "train_loss_llm": 0.0, "grad_norm": 0.04936075955629349, "global_step": 33574, "epoch": 932, "lr": 0.001} +{"train_loss": 0.027402926236391068, "train_loss_bc": 0.027402926236391068, "train_loss_llm": 0.0, "grad_norm": 0.08179938048124313, "global_step": 33575, "epoch": 932, "lr": 0.001} +{"train_loss": 0.031886987388134, "train_loss_bc": 0.031886987388134, "train_loss_llm": 0.0, "grad_norm": 0.10599182546138763, "global_step": 33576, "epoch": 932, "lr": 0.001} +{"train_loss": 0.02698742412030697, "train_loss_bc": 0.02698742412030697, "train_loss_llm": 0.0, "grad_norm": 0.03932566940784454, "global_step": 33577, "epoch": 932, "lr": 0.001} +{"train_loss": 0.026410197839140892, "train_loss_bc": 0.026410197839140892, "train_loss_llm": 0.0, "grad_norm": 0.05247853323817253, "global_step": 33578, "epoch": 932, "lr": 0.001} +{"train_loss": 0.027639204636216164, "train_loss_bc": 0.027639204636216164, "train_loss_llm": 0.0, "grad_norm": 0.06793161481618881, "global_step": 33579, "epoch": 932, "lr": 0.001} +{"train_loss": 0.02443741261959076, "train_loss_bc": 0.02443741261959076, "train_loss_llm": 0.0, "grad_norm": 0.0820888876914978, "global_step": 33580, "epoch": 932, "lr": 0.001} +{"train_loss": 0.028259510174393654, "train_loss_bc": 0.028259510174393654, "train_loss_llm": 0.0, "grad_norm": 0.031061045825481415, "global_step": 33581, "epoch": 932, "lr": 0.001} +{"train_loss": 0.026676909998059273, "train_loss_bc": 0.026676909998059273, "train_loss_llm": 0.0, "grad_norm": 0.05197444185614586, "global_step": 33582, "epoch": 932, "lr": 0.001} +{"train_loss": 0.032885901629924774, "train_loss_bc": 0.032885901629924774, "train_loss_llm": 0.0, "grad_norm": 0.07084173709154129, "global_step": 33583, "epoch": 932, "lr": 0.001} +{"train_loss": 0.034594252705574036, "train_loss_bc": 0.034594252705574036, "train_loss_llm": 0.0, "grad_norm": 0.09981369227170944, "global_step": 33584, "epoch": 932, "lr": 0.001} +{"train_loss": 0.03204670920968056, "train_loss_bc": 0.03204670920968056, "train_loss_llm": 0.0, "grad_norm": 0.03398244082927704, "global_step": 33585, "epoch": 932, "lr": 0.001} +{"train_loss": 0.02753753401339054, "train_loss_bc": 0.02753753401339054, "train_loss_llm": 0.0, "grad_norm": 0.05452948436141014, "global_step": 33586, "epoch": 932, "lr": 0.001} +{"train_loss": 0.028996730999400217, "train_loss_bc": 0.03227263689041138, "train_loss_llm": 0.0, "grad_norm": 0.07206858694553375, "global_step": 33587, "epoch": 932, "lr": 0.001, "val_loss": 0.1057761162519455} +{"train_loss": 0.02276640199124813, "train_loss_bc": 0.022755863144993782, "train_loss_llm": 0.0010538101196289062, "grad_norm": 0.09651853889226913, "global_step": 33588, "epoch": 933, "lr": 0.001} +{"train_loss": 0.02926786243915558, "train_loss_bc": 0.02926786243915558, "train_loss_llm": 0.0, "grad_norm": 0.034318942576646805, "global_step": 33589, "epoch": 933, "lr": 0.001} +{"train_loss": 0.031290940940380096, "train_loss_bc": 0.031290940940380096, "train_loss_llm": 0.0, "grad_norm": 0.052512358874082565, "global_step": 33590, "epoch": 933, "lr": 0.001} +{"train_loss": 0.03336719051003456, "train_loss_bc": 0.03336719051003456, "train_loss_llm": 0.0, "grad_norm": 0.07413797080516815, "global_step": 33591, "epoch": 933, "lr": 0.001} +{"train_loss": 0.029644666239619255, "train_loss_bc": 0.029644666239619255, "train_loss_llm": 0.0, "grad_norm": 0.09660106152296066, "global_step": 33592, "epoch": 933, "lr": 0.001} +{"train_loss": 0.02956472337245941, "train_loss_bc": 0.02956472337245941, "train_loss_llm": 0.0, "grad_norm": 0.04288364201784134, "global_step": 33593, "epoch": 933, "lr": 0.001} +{"train_loss": 0.024295106530189514, "train_loss_bc": 0.024295106530189514, "train_loss_llm": 0.0, "grad_norm": 0.06527827680110931, "global_step": 33594, "epoch": 933, "lr": 0.001} +{"train_loss": 0.026519598439335823, "train_loss_bc": 0.026519598439335823, "train_loss_llm": 0.0, "grad_norm": 0.08287149667739868, "global_step": 33595, "epoch": 933, "lr": 0.001} +{"train_loss": 0.02547306753695011, "train_loss_bc": 0.02547306753695011, "train_loss_llm": 0.0, "grad_norm": 0.11141795665025711, "global_step": 33596, "epoch": 933, "lr": 0.001} +{"train_loss": 0.027263900265097618, "train_loss_bc": 0.027263900265097618, "train_loss_llm": 0.0, "grad_norm": 0.05249559506773949, "global_step": 33597, "epoch": 933, "lr": 0.001} +{"train_loss": 0.03172998130321503, "train_loss_bc": 0.031727708876132965, "train_loss_llm": 0.00022706389427185059, "grad_norm": 0.0800534039735794, "global_step": 33598, "epoch": 933, "lr": 0.001} +{"train_loss": 0.031590528786182404, "train_loss_bc": 0.031590528786182404, "train_loss_llm": 0.0, "grad_norm": 0.10554952174425125, "global_step": 33599, "epoch": 933, "lr": 0.001} +{"train_loss": 0.03301241621375084, "train_loss_bc": 0.03301241621375084, "train_loss_llm": 0.0, "grad_norm": 0.1389755755662918, "global_step": 33600, "epoch": 933, "lr": 0.001} +{"train_loss": 0.025602277368307114, "train_loss_bc": 0.025599749758839607, "train_loss_llm": 0.00025269389152526855, "grad_norm": 0.03922212868928909, "global_step": 33601, "epoch": 933, "lr": 0.001} +{"train_loss": 0.027400275692343712, "train_loss_bc": 0.027400275692343712, "train_loss_llm": 0.0, "grad_norm": 0.06307078152894974, "global_step": 33602, "epoch": 933, "lr": 0.001} +{"train_loss": 0.034230709075927734, "train_loss_bc": 0.034230709075927734, "train_loss_llm": 0.0, "grad_norm": 0.079367496073246, "global_step": 33603, "epoch": 933, "lr": 0.001} +{"train_loss": 0.030052246525883675, "train_loss_bc": 0.030052246525883675, "train_loss_llm": 0.0, "grad_norm": 0.09034202247858047, "global_step": 33604, "epoch": 933, "lr": 0.001} +{"train_loss": 0.029854541644454002, "train_loss_bc": 0.029854541644454002, "train_loss_llm": 0.0, "grad_norm": 0.03642256557941437, "global_step": 33605, "epoch": 933, "lr": 0.001} +{"train_loss": 0.03140939772129059, "train_loss_bc": 0.03140939772129059, "train_loss_llm": 0.0, "grad_norm": 0.07648592442274094, "global_step": 33606, "epoch": 933, "lr": 0.001} +{"train_loss": 0.025470146909356117, "train_loss_bc": 0.025470146909356117, "train_loss_llm": 0.0, "grad_norm": 0.10895491391420364, "global_step": 33607, "epoch": 933, "lr": 0.001} +{"train_loss": 0.032659851014614105, "train_loss_bc": 0.032659851014614105, "train_loss_llm": 0.0, "grad_norm": 0.14055383205413818, "global_step": 33608, "epoch": 933, "lr": 0.001} +{"train_loss": 0.03254082426428795, "train_loss_bc": 0.03254082426428795, "train_loss_llm": 0.0, "grad_norm": 0.04230891540646553, "global_step": 33609, "epoch": 933, "lr": 0.001} +{"train_loss": 0.028704293072223663, "train_loss_bc": 0.02869628369808197, "train_loss_llm": 0.0008010268211364746, "grad_norm": 0.07409074902534485, "global_step": 33610, "epoch": 933, "lr": 0.001} +{"train_loss": 0.029719660058617592, "train_loss_bc": 0.029713373631238937, "train_loss_llm": 0.0006285905838012695, "grad_norm": 0.10751863569021225, "global_step": 33611, "epoch": 933, "lr": 0.001} +{"train_loss": 0.02994547039270401, "train_loss_bc": 0.02994547039270401, "train_loss_llm": 0.0, "grad_norm": 0.130787655711174, "global_step": 33612, "epoch": 933, "lr": 0.001} +{"train_loss": 0.026392793282866478, "train_loss_bc": 0.026392793282866478, "train_loss_llm": 0.0, "grad_norm": 0.03634602576494217, "global_step": 33613, "epoch": 933, "lr": 0.001} +{"train_loss": 0.028573933988809586, "train_loss_bc": 0.028573933988809586, "train_loss_llm": 0.0, "grad_norm": 0.05858370289206505, "global_step": 33614, "epoch": 933, "lr": 0.001} +{"train_loss": 0.029024112969636917, "train_loss_bc": 0.029024112969636917, "train_loss_llm": 0.0, "grad_norm": 0.07331110537052155, "global_step": 33615, "epoch": 933, "lr": 0.001} +{"train_loss": 0.030502738431096077, "train_loss_bc": 0.030502738431096077, "train_loss_llm": 0.0, "grad_norm": 0.09122706204652786, "global_step": 33616, "epoch": 933, "lr": 0.001} +{"train_loss": 0.027026686817407608, "train_loss_bc": 0.027026686817407608, "train_loss_llm": 0.0, "grad_norm": 0.03540220111608505, "global_step": 33617, "epoch": 933, "lr": 0.001} +{"train_loss": 0.026947367936372757, "train_loss_bc": 0.026947367936372757, "train_loss_llm": 0.0, "grad_norm": 0.05707535892724991, "global_step": 33618, "epoch": 933, "lr": 0.001} +{"train_loss": 0.02404431253671646, "train_loss_bc": 0.02404431253671646, "train_loss_llm": 0.0, "grad_norm": 0.07736194878816605, "global_step": 33619, "epoch": 933, "lr": 0.001} +{"train_loss": 0.030731746926903725, "train_loss_bc": 0.030731746926903725, "train_loss_llm": 0.0, "grad_norm": 0.09565404802560806, "global_step": 33620, "epoch": 933, "lr": 0.001} +{"train_loss": 0.027054451406002045, "train_loss_bc": 0.027039533481001854, "train_loss_llm": 0.0014918744564056396, "grad_norm": 0.03732709214091301, "global_step": 33621, "epoch": 933, "lr": 0.001} +{"train_loss": 0.028509730473160744, "train_loss_bc": 0.028498128056526184, "train_loss_llm": 0.0011602044105529785, "grad_norm": 0.055147141218185425, "global_step": 33622, "epoch": 933, "lr": 0.001} +{"train_loss": 0.028942646717445716, "train_loss_bc": 0.029742253944277763, "train_loss_llm": 0.0009074807167053223, "grad_norm": 0.06696690618991852, "global_step": 33623, "epoch": 933, "lr": 0.001, "val_loss": 0.10431652516126633} +{"train_loss": 0.024847520515322685, "train_loss_bc": 0.024847520515322685, "train_loss_llm": 0.0, "grad_norm": 0.07516178488731384, "global_step": 33624, "epoch": 934, "lr": 0.001} +{"train_loss": 0.025288071483373642, "train_loss_bc": 0.02526484988629818, "train_loss_llm": 0.0023221969604492188, "grad_norm": 0.030667714774608612, "global_step": 33625, "epoch": 934, "lr": 0.001} +{"train_loss": 0.02690158039331436, "train_loss_bc": 0.02690158039331436, "train_loss_llm": 0.0, "grad_norm": 0.044461023062467575, "global_step": 33626, "epoch": 934, "lr": 0.001} +{"train_loss": 0.025948720052838326, "train_loss_bc": 0.025948720052838326, "train_loss_llm": 0.0, "grad_norm": 0.0602535754442215, "global_step": 33627, "epoch": 934, "lr": 0.001} +{"train_loss": 0.026035143062472343, "train_loss_bc": 0.026035143062472343, "train_loss_llm": 0.0, "grad_norm": 0.06772072613239288, "global_step": 33628, "epoch": 934, "lr": 0.001} +{"train_loss": 0.02874583564698696, "train_loss_bc": 0.02874583564698696, "train_loss_llm": 0.0, "grad_norm": 0.04603920131921768, "global_step": 33629, "epoch": 934, "lr": 0.001} +{"train_loss": 0.02812511846423149, "train_loss_bc": 0.028103889897465706, "train_loss_llm": 0.0021227896213531494, "grad_norm": 0.07975154370069504, "global_step": 33630, "epoch": 934, "lr": 0.001} +{"train_loss": 0.023244811221957207, "train_loss_bc": 0.023244811221957207, "train_loss_llm": 0.0, "grad_norm": 0.11023197323083878, "global_step": 33631, "epoch": 934, "lr": 0.001} +{"train_loss": 0.026646804064512253, "train_loss_bc": 0.02664373070001602, "train_loss_llm": 0.00030732154846191406, "grad_norm": 0.14586994051933289, "global_step": 33632, "epoch": 934, "lr": 0.001} +{"train_loss": 0.03045635297894478, "train_loss_bc": 0.03045635297894478, "train_loss_llm": 0.0, "grad_norm": 0.03740096837282181, "global_step": 33633, "epoch": 934, "lr": 0.001} +{"train_loss": 0.02646813914179802, "train_loss_bc": 0.026467153802514076, "train_loss_llm": 9.849667549133301e-05, "grad_norm": 0.058971893042325974, "global_step": 33634, "epoch": 934, "lr": 0.001} +{"train_loss": 0.02710578590631485, "train_loss_bc": 0.02710578590631485, "train_loss_llm": 0.0, "grad_norm": 0.09088850766420364, "global_step": 33635, "epoch": 934, "lr": 0.001} +{"train_loss": 0.024988088756799698, "train_loss_bc": 0.024988088756799698, "train_loss_llm": 0.0, "grad_norm": 0.10331632196903229, "global_step": 33636, "epoch": 934, "lr": 0.001} +{"train_loss": 0.030543051660060883, "train_loss_bc": 0.030543051660060883, "train_loss_llm": 0.0, "grad_norm": 0.041113488376140594, "global_step": 33637, "epoch": 934, "lr": 0.001} +{"train_loss": 0.032034505158662796, "train_loss_bc": 0.032034505158662796, "train_loss_llm": 0.0, "grad_norm": 0.07048842310905457, "global_step": 33638, "epoch": 934, "lr": 0.001} +{"train_loss": 0.024869907647371292, "train_loss_bc": 0.024869907647371292, "train_loss_llm": 0.0, "grad_norm": 0.0866335779428482, "global_step": 33639, "epoch": 934, "lr": 0.001} +{"train_loss": 0.028765805065631866, "train_loss_bc": 0.028765805065631866, "train_loss_llm": 0.0, "grad_norm": 0.11455968767404556, "global_step": 33640, "epoch": 934, "lr": 0.001} +{"train_loss": 0.029879877343773842, "train_loss_bc": 0.029879877343773842, "train_loss_llm": 0.0, "grad_norm": 0.03789999708533287, "global_step": 33641, "epoch": 934, "lr": 0.001} +{"train_loss": 0.03176320716738701, "train_loss_bc": 0.03176320716738701, "train_loss_llm": 0.0, "grad_norm": 0.061864737421274185, "global_step": 33642, "epoch": 934, "lr": 0.001} +{"train_loss": 0.03421667590737343, "train_loss_bc": 0.03421667590737343, "train_loss_llm": 0.0, "grad_norm": 0.0885230153799057, "global_step": 33643, "epoch": 934, "lr": 0.001} +{"train_loss": 0.028427377343177795, "train_loss_bc": 0.028412234038114548, "train_loss_llm": 0.0015142858028411865, "grad_norm": 0.10895173996686935, "global_step": 33644, "epoch": 934, "lr": 0.001} +{"train_loss": 0.02753457799553871, "train_loss_bc": 0.02753457799553871, "train_loss_llm": 0.0, "grad_norm": 0.04205228388309479, "global_step": 33645, "epoch": 934, "lr": 0.001} +{"train_loss": 0.03203822672367096, "train_loss_bc": 0.03203822672367096, "train_loss_llm": 0.0, "grad_norm": 0.058601897209882736, "global_step": 33646, "epoch": 934, "lr": 0.001} +{"train_loss": 0.026560021564364433, "train_loss_bc": 0.026560021564364433, "train_loss_llm": 0.0, "grad_norm": 0.07029196619987488, "global_step": 33647, "epoch": 934, "lr": 0.001} +{"train_loss": 0.03224235773086548, "train_loss_bc": 0.03224235773086548, "train_loss_llm": 0.0, "grad_norm": 0.09738161414861679, "global_step": 33648, "epoch": 934, "lr": 0.001} +{"train_loss": 0.023914773017168045, "train_loss_bc": 0.023900380358099937, "train_loss_llm": 0.0014392435550689697, "grad_norm": 0.031452521681785583, "global_step": 33649, "epoch": 934, "lr": 0.001} +{"train_loss": 0.027899397537112236, "train_loss_bc": 0.02789364382624626, "train_loss_llm": 0.0005753040313720703, "grad_norm": 0.05813046544790268, "global_step": 33650, "epoch": 934, "lr": 0.001} +{"train_loss": 0.030683116987347603, "train_loss_bc": 0.030682329088449478, "train_loss_llm": 7.870793342590332e-05, "grad_norm": 0.10065039247274399, "global_step": 33651, "epoch": 934, "lr": 0.001} +{"train_loss": 0.030684692785143852, "train_loss_bc": 0.030684692785143852, "train_loss_llm": 0.0, "grad_norm": 0.13447551429271698, "global_step": 33652, "epoch": 934, "lr": 0.001} +{"train_loss": 0.028932221233844757, "train_loss_bc": 0.028932221233844757, "train_loss_llm": 0.0, "grad_norm": 0.047784652560949326, "global_step": 33653, "epoch": 934, "lr": 0.001} +{"train_loss": 0.03257090225815773, "train_loss_bc": 0.032566823065280914, "train_loss_llm": 0.00040778517723083496, "grad_norm": 0.0723600834608078, "global_step": 33654, "epoch": 934, "lr": 0.001} +{"train_loss": 0.03761346638202667, "train_loss_bc": 0.03761301189661026, "train_loss_llm": 4.5627355575561523e-05, "grad_norm": 0.099920354783535, "global_step": 33655, "epoch": 934, "lr": 0.001} +{"train_loss": 0.03382822126150131, "train_loss_bc": 0.03382822126150131, "train_loss_llm": 0.0, "grad_norm": 0.11965920776128769, "global_step": 33656, "epoch": 934, "lr": 0.001} +{"train_loss": 0.03044871799647808, "train_loss_bc": 0.03044871799647808, "train_loss_llm": 0.0, "grad_norm": 0.03997090831398964, "global_step": 33657, "epoch": 934, "lr": 0.001} +{"train_loss": 0.031945664435625076, "train_loss_bc": 0.031945664435625076, "train_loss_llm": 0.0, "grad_norm": 0.06576599925756454, "global_step": 33658, "epoch": 934, "lr": 0.001} +{"train_loss": 0.028980724772231445, "train_loss_bc": 0.03107842430472374, "train_loss_llm": 0.00289309024810791, "grad_norm": 0.10110371559858322, "global_step": 33659, "epoch": 934, "lr": 0.001, "val_loss": 0.10681809484958649} +{"train_loss": 0.029880840331315994, "train_loss_bc": 0.029880840331315994, "train_loss_llm": 0.0, "grad_norm": 0.13382427394390106, "global_step": 33660, "epoch": 935, "lr": 0.001} +{"train_loss": 0.027869822457432747, "train_loss_bc": 0.027869822457432747, "train_loss_llm": 0.0, "grad_norm": 0.03731415048241615, "global_step": 33661, "epoch": 935, "lr": 0.001} +{"train_loss": 0.031216207891702652, "train_loss_bc": 0.031216207891702652, "train_loss_llm": 0.0, "grad_norm": 0.05232568085193634, "global_step": 33662, "epoch": 935, "lr": 0.001} +{"train_loss": 0.029380260035395622, "train_loss_bc": 0.029380260035395622, "train_loss_llm": 0.0, "grad_norm": 0.06504509598016739, "global_step": 33663, "epoch": 935, "lr": 0.001} +{"train_loss": 0.025863323360681534, "train_loss_bc": 0.025863323360681534, "train_loss_llm": 0.0, "grad_norm": 0.09279023855924606, "global_step": 33664, "epoch": 935, "lr": 0.001} +{"train_loss": 0.028521714732050896, "train_loss_bc": 0.028521714732050896, "train_loss_llm": 0.0, "grad_norm": 0.035984106361866, "global_step": 33665, "epoch": 935, "lr": 0.001} +{"train_loss": 0.024953693151474, "train_loss_bc": 0.024953693151474, "train_loss_llm": 0.0, "grad_norm": 0.05699668452143669, "global_step": 33666, "epoch": 935, "lr": 0.001} +{"train_loss": 0.02516276203095913, "train_loss_bc": 0.02516276203095913, "train_loss_llm": 0.0, "grad_norm": 0.07017017155885696, "global_step": 33667, "epoch": 935, "lr": 0.001} +{"train_loss": 0.02932250127196312, "train_loss_bc": 0.02932250127196312, "train_loss_llm": 0.0, "grad_norm": 0.08417415618896484, "global_step": 33668, "epoch": 935, "lr": 0.001} +{"train_loss": 0.030364353209733963, "train_loss_bc": 0.030352089554071426, "train_loss_llm": 0.0012263953685760498, "grad_norm": 0.047430917620658875, "global_step": 33669, "epoch": 935, "lr": 0.001} +{"train_loss": 0.022094588726758957, "train_loss_bc": 0.022094588726758957, "train_loss_llm": 0.0, "grad_norm": 0.07041938602924347, "global_step": 33670, "epoch": 935, "lr": 0.001} +{"train_loss": 0.03328677639365196, "train_loss_bc": 0.03328677639365196, "train_loss_llm": 0.0, "grad_norm": 0.09344910085201263, "global_step": 33671, "epoch": 935, "lr": 0.001} +{"train_loss": 0.026180997490882874, "train_loss_bc": 0.02617896907031536, "train_loss_llm": 0.0002028942108154297, "grad_norm": 0.10812335461378098, "global_step": 33672, "epoch": 935, "lr": 0.001} +{"train_loss": 0.02864970453083515, "train_loss_bc": 0.02864970453083515, "train_loss_llm": 0.0, "grad_norm": 0.0394165962934494, "global_step": 33673, "epoch": 935, "lr": 0.001} +{"train_loss": 0.02382753975689411, "train_loss_bc": 0.0238203015178442, "train_loss_llm": 0.0007238984107971191, "grad_norm": 0.05675258859992027, "global_step": 33674, "epoch": 935, "lr": 0.001} +{"train_loss": 0.02820281870663166, "train_loss_bc": 0.02820281870663166, "train_loss_llm": 0.0, "grad_norm": 0.07069972157478333, "global_step": 33675, "epoch": 935, "lr": 0.001} +{"train_loss": 0.026875099167227745, "train_loss_bc": 0.026875099167227745, "train_loss_llm": 0.0, "grad_norm": 0.08142650127410889, "global_step": 33676, "epoch": 935, "lr": 0.001} +{"train_loss": 0.037546731531620026, "train_loss_bc": 0.0375368669629097, "train_loss_llm": 0.000986337661743164, "grad_norm": 0.05452879145741463, "global_step": 33677, "epoch": 935, "lr": 0.001} +{"train_loss": 0.023194532841444016, "train_loss_bc": 0.023194532841444016, "train_loss_llm": 0.0, "grad_norm": 0.0636960119009018, "global_step": 33678, "epoch": 935, "lr": 0.001} +{"train_loss": 0.03624388575553894, "train_loss_bc": 0.03624388575553894, "train_loss_llm": 0.0, "grad_norm": 0.08557935059070587, "global_step": 33679, "epoch": 935, "lr": 0.001} +{"train_loss": 0.028850603848695755, "train_loss_bc": 0.028850603848695755, "train_loss_llm": 0.0, "grad_norm": 0.09403261542320251, "global_step": 33680, "epoch": 935, "lr": 0.001} +{"train_loss": 0.023621855303645134, "train_loss_bc": 0.023621855303645134, "train_loss_llm": 0.0, "grad_norm": 0.03475717455148697, "global_step": 33681, "epoch": 935, "lr": 0.001} +{"train_loss": 0.030429523438215256, "train_loss_bc": 0.03042754903435707, "train_loss_llm": 0.00019744038581848145, "grad_norm": 0.06166434660553932, "global_step": 33682, "epoch": 935, "lr": 0.001} +{"train_loss": 0.029484823346138, "train_loss_bc": 0.029484823346138, "train_loss_llm": 0.0, "grad_norm": 0.07865837216377258, "global_step": 33683, "epoch": 935, "lr": 0.001} +{"train_loss": 0.028052616864442825, "train_loss_bc": 0.028052616864442825, "train_loss_llm": 0.0, "grad_norm": 0.10074824839830399, "global_step": 33684, "epoch": 935, "lr": 0.001} +{"train_loss": 0.0263685192912817, "train_loss_bc": 0.0263685192912817, "train_loss_llm": 0.0, "grad_norm": 0.03646111488342285, "global_step": 33685, "epoch": 935, "lr": 0.001} +{"train_loss": 0.025009553879499435, "train_loss_bc": 0.025009553879499435, "train_loss_llm": 0.0, "grad_norm": 0.0583750419318676, "global_step": 33686, "epoch": 935, "lr": 0.001} +{"train_loss": 0.030370693653821945, "train_loss_bc": 0.030360950157046318, "train_loss_llm": 0.0009742677211761475, "grad_norm": 0.07728905230760574, "global_step": 33687, "epoch": 935, "lr": 0.001} +{"train_loss": 0.03555913642048836, "train_loss_bc": 0.03555913642048836, "train_loss_llm": 0.0, "grad_norm": 0.11374247074127197, "global_step": 33688, "epoch": 935, "lr": 0.001} +{"train_loss": 0.03522627428174019, "train_loss_bc": 0.03521619364619255, "train_loss_llm": 0.0010080039501190186, "grad_norm": 0.03493417426943779, "global_step": 33689, "epoch": 935, "lr": 0.001} +{"train_loss": 0.03595276176929474, "train_loss_bc": 0.03595276176929474, "train_loss_llm": 0.0, "grad_norm": 0.06643389910459518, "global_step": 33690, "epoch": 935, "lr": 0.001} +{"train_loss": 0.02857338637113571, "train_loss_bc": 0.028563693165779114, "train_loss_llm": 0.0009693503379821777, "grad_norm": 0.07266128808259964, "global_step": 33691, "epoch": 935, "lr": 0.001} +{"train_loss": 0.02802698127925396, "train_loss_bc": 0.02802698127925396, "train_loss_llm": 0.0, "grad_norm": 0.07851799577474594, "global_step": 33692, "epoch": 935, "lr": 0.001} +{"train_loss": 0.030750250443816185, "train_loss_bc": 0.030749868601560593, "train_loss_llm": 3.826618194580078e-05, "grad_norm": 0.04030035808682442, "global_step": 33693, "epoch": 935, "lr": 0.001} +{"train_loss": 0.024559110403060913, "train_loss_bc": 0.024559110403060913, "train_loss_llm": 0.0, "grad_norm": 0.06227324157953262, "global_step": 33694, "epoch": 935, "lr": 0.001} +{"train_loss": 0.028824658650490973, "train_loss_bc": 0.028213467448949814, "train_loss_llm": 0.0, "grad_norm": 0.10097987949848175, "global_step": 33695, "epoch": 935, "lr": 0.001, "train/cumulative_reward": 46.44934033067305, "train/mean_score": 17.817759017689383, "train/success_rate": 0.5, "test/cumulative_reward": 43.2631812685581, "test/mean_score": 8.013796941616548, "test/success_rate": 0.2, "val_loss": 0.10715509951114655, "train_action_mse_error": 0.032282959669828415} +{"train_loss": 0.02292647399008274, "train_loss_bc": 0.022926002740859985, "train_loss_llm": 4.70578670501709e-05, "grad_norm": 0.10546227544546127, "global_step": 33696, "epoch": 936, "lr": 0.001} +{"train_loss": 0.030430389568209648, "train_loss_bc": 0.030430389568209648, "train_loss_llm": 0.0, "grad_norm": 0.0515088327229023, "global_step": 33697, "epoch": 936, "lr": 0.001} +{"train_loss": 0.02990412339568138, "train_loss_bc": 0.02990412339568138, "train_loss_llm": 0.0, "grad_norm": 0.07499176263809204, "global_step": 33698, "epoch": 936, "lr": 0.001} +{"train_loss": 0.03571499511599541, "train_loss_bc": 0.03570627421140671, "train_loss_llm": 0.000872194766998291, "grad_norm": 0.11166681349277496, "global_step": 33699, "epoch": 936, "lr": 0.001} +{"train_loss": 0.031902872025966644, "train_loss_bc": 0.03190145641565323, "train_loss_llm": 0.00014162063598632812, "grad_norm": 0.1334504783153534, "global_step": 33700, "epoch": 936, "lr": 0.001} +{"train_loss": 0.028169486671686172, "train_loss_bc": 0.028169486671686172, "train_loss_llm": 0.0, "grad_norm": 0.03970622643828392, "global_step": 33701, "epoch": 936, "lr": 0.001} +{"train_loss": 0.040276601910591125, "train_loss_bc": 0.040276601910591125, "train_loss_llm": 0.0, "grad_norm": 0.060509487986564636, "global_step": 33702, "epoch": 936, "lr": 0.001} +{"train_loss": 0.027940886095166206, "train_loss_bc": 0.027940886095166206, "train_loss_llm": 0.0, "grad_norm": 0.08682326972484589, "global_step": 33703, "epoch": 936, "lr": 0.001} +{"train_loss": 0.027179054915905, "train_loss_bc": 0.02716820314526558, "train_loss_llm": 0.0010852217674255371, "grad_norm": 0.10756499320268631, "global_step": 33704, "epoch": 936, "lr": 0.001} +{"train_loss": 0.03273788467049599, "train_loss_bc": 0.03273788467049599, "train_loss_llm": 0.0, "grad_norm": 0.05309886112809181, "global_step": 33705, "epoch": 936, "lr": 0.001} +{"train_loss": 0.033133309334516525, "train_loss_bc": 0.033133309334516525, "train_loss_llm": 0.0, "grad_norm": 0.08933895081281662, "global_step": 33706, "epoch": 936, "lr": 0.001} +{"train_loss": 0.025044243782758713, "train_loss_bc": 0.02503974549472332, "train_loss_llm": 0.00044974684715270996, "grad_norm": 0.11337297409772873, "global_step": 33707, "epoch": 936, "lr": 0.001} +{"train_loss": 0.028832783922553062, "train_loss_bc": 0.02883068099617958, "train_loss_llm": 0.00021022558212280273, "grad_norm": 0.14297625422477722, "global_step": 33708, "epoch": 936, "lr": 0.001} +{"train_loss": 0.025646323338150978, "train_loss_bc": 0.025631681084632874, "train_loss_llm": 0.001464158296585083, "grad_norm": 0.0448184572160244, "global_step": 33709, "epoch": 936, "lr": 0.001} +{"train_loss": 0.034372564405202866, "train_loss_bc": 0.034349530935287476, "train_loss_llm": 0.0023033320903778076, "grad_norm": 0.06536999344825745, "global_step": 33710, "epoch": 936, "lr": 0.001} +{"train_loss": 0.02575145661830902, "train_loss_bc": 0.02575145661830902, "train_loss_llm": 0.0, "grad_norm": 0.10090841352939606, "global_step": 33711, "epoch": 936, "lr": 0.001} +{"train_loss": 0.029512982815504074, "train_loss_bc": 0.029512982815504074, "train_loss_llm": 0.0, "grad_norm": 0.13475696742534637, "global_step": 33712, "epoch": 936, "lr": 0.001} +{"train_loss": 0.021990396082401276, "train_loss_bc": 0.021990396082401276, "train_loss_llm": 0.0, "grad_norm": 0.029009709134697914, "global_step": 33713, "epoch": 936, "lr": 0.001} +{"train_loss": 0.024695314466953278, "train_loss_bc": 0.024695314466953278, "train_loss_llm": 0.0, "grad_norm": 0.04726720601320267, "global_step": 33714, "epoch": 936, "lr": 0.001} +{"train_loss": 0.027883978560566902, "train_loss_bc": 0.027883978560566902, "train_loss_llm": 0.0, "grad_norm": 0.0554826594889164, "global_step": 33715, "epoch": 936, "lr": 0.001} +{"train_loss": 0.03412413224577904, "train_loss_bc": 0.03412413224577904, "train_loss_llm": 0.0, "grad_norm": 0.0742698609828949, "global_step": 33716, "epoch": 936, "lr": 0.001} +{"train_loss": 0.023685995489358902, "train_loss_bc": 0.023676956072449684, "train_loss_llm": 0.0009039342403411865, "grad_norm": 0.029588252305984497, "global_step": 33717, "epoch": 936, "lr": 0.001} +{"train_loss": 0.02461748570203781, "train_loss_bc": 0.02461748570203781, "train_loss_llm": 0.0, "grad_norm": 0.041206344962120056, "global_step": 33718, "epoch": 936, "lr": 0.001} +{"train_loss": 0.02540063112974167, "train_loss_bc": 0.02540063112974167, "train_loss_llm": 0.0, "grad_norm": 0.05915836989879608, "global_step": 33719, "epoch": 936, "lr": 0.001} +{"train_loss": 0.02629774995148182, "train_loss_bc": 0.02629774995148182, "train_loss_llm": 0.0, "grad_norm": 0.07106555253267288, "global_step": 33720, "epoch": 936, "lr": 0.001} +{"train_loss": 0.02993079647421837, "train_loss_bc": 0.02993079647421837, "train_loss_llm": 0.0, "grad_norm": 0.04294849559664726, "global_step": 33721, "epoch": 936, "lr": 0.001} +{"train_loss": 0.02894044853746891, "train_loss_bc": 0.02894044853746891, "train_loss_llm": 0.0, "grad_norm": 0.056224022060632706, "global_step": 33722, "epoch": 936, "lr": 0.001} +{"train_loss": 0.034931980073451996, "train_loss_bc": 0.034931980073451996, "train_loss_llm": 0.0, "grad_norm": 0.09205538034439087, "global_step": 33723, "epoch": 936, "lr": 0.001} +{"train_loss": 0.03038526512682438, "train_loss_bc": 0.03038526512682438, "train_loss_llm": 0.0, "grad_norm": 0.12560883164405823, "global_step": 33724, "epoch": 936, "lr": 0.001} +{"train_loss": 0.029230134561657906, "train_loss_bc": 0.029230134561657906, "train_loss_llm": 0.0, "grad_norm": 0.04656941816210747, "global_step": 33725, "epoch": 936, "lr": 0.001} +{"train_loss": 0.025533081963658333, "train_loss_bc": 0.025533081963658333, "train_loss_llm": 0.0, "grad_norm": 0.07027183473110199, "global_step": 33726, "epoch": 936, "lr": 0.001} +{"train_loss": 0.02953694760799408, "train_loss_bc": 0.02953694760799408, "train_loss_llm": 0.0, "grad_norm": 0.09558959305286407, "global_step": 33727, "epoch": 936, "lr": 0.001} +{"train_loss": 0.03364025428891182, "train_loss_bc": 0.03364025428891182, "train_loss_llm": 0.0, "grad_norm": 0.13685500621795654, "global_step": 33728, "epoch": 936, "lr": 0.001} +{"train_loss": 0.026484815403819084, "train_loss_bc": 0.026483235880732536, "train_loss_llm": 0.00015798211097717285, "grad_norm": 0.04309912398457527, "global_step": 33729, "epoch": 936, "lr": 0.001} +{"train_loss": 0.02875271812081337, "train_loss_bc": 0.02875271812081337, "train_loss_llm": 0.0, "grad_norm": 0.06678522378206253, "global_step": 33730, "epoch": 936, "lr": 0.001} +{"train_loss": 0.029045241522706218, "train_loss_bc": 0.03009013645350933, "train_loss_llm": 0.0, "grad_norm": 0.08733642101287842, "global_step": 33731, "epoch": 936, "lr": 0.001, "val_loss": 0.1051652729511261} +{"train_loss": 0.028113560751080513, "train_loss_bc": 0.028113560751080513, "train_loss_llm": 0.0, "grad_norm": 0.10994768887758255, "global_step": 33732, "epoch": 937, "lr": 0.001} +{"train_loss": 0.024835316464304924, "train_loss_bc": 0.024835316464304924, "train_loss_llm": 0.0, "grad_norm": 0.03963238745927811, "global_step": 33733, "epoch": 937, "lr": 0.001} +{"train_loss": 0.03015229105949402, "train_loss_bc": 0.030140945687890053, "train_loss_llm": 0.0011345744132995605, "grad_norm": 0.06592760235071182, "global_step": 33734, "epoch": 937, "lr": 0.001} +{"train_loss": 0.02415933646261692, "train_loss_bc": 0.02415933646261692, "train_loss_llm": 0.0, "grad_norm": 0.0788436233997345, "global_step": 33735, "epoch": 937, "lr": 0.001} +{"train_loss": 0.029710199683904648, "train_loss_bc": 0.029710199683904648, "train_loss_llm": 0.0, "grad_norm": 0.10218776017427444, "global_step": 33736, "epoch": 937, "lr": 0.001} +{"train_loss": 0.03039451129734516, "train_loss_bc": 0.03039451129734516, "train_loss_llm": 0.0, "grad_norm": 0.04147926717996597, "global_step": 33737, "epoch": 937, "lr": 0.001} +{"train_loss": 0.024973755702376366, "train_loss_bc": 0.024973755702376366, "train_loss_llm": 0.0, "grad_norm": 0.04988723620772362, "global_step": 33738, "epoch": 937, "lr": 0.001} +{"train_loss": 0.03187636286020279, "train_loss_bc": 0.03187636286020279, "train_loss_llm": 0.0, "grad_norm": 0.05898987129330635, "global_step": 33739, "epoch": 937, "lr": 0.001} +{"train_loss": 0.022877071052789688, "train_loss_bc": 0.022875672206282616, "train_loss_llm": 0.0001399517059326172, "grad_norm": 0.07175540924072266, "global_step": 33740, "epoch": 937, "lr": 0.001} +{"train_loss": 0.02516811341047287, "train_loss_bc": 0.025164544582366943, "train_loss_llm": 0.0003567934036254883, "grad_norm": 0.03514334559440613, "global_step": 33741, "epoch": 937, "lr": 0.001} +{"train_loss": 0.022573698312044144, "train_loss_bc": 0.022573698312044144, "train_loss_llm": 0.0, "grad_norm": 0.050896208733320236, "global_step": 33742, "epoch": 937, "lr": 0.001} +{"train_loss": 0.02928568609058857, "train_loss_bc": 0.02928568609058857, "train_loss_llm": 0.0, "grad_norm": 0.07274547219276428, "global_step": 33743, "epoch": 937, "lr": 0.001} +{"train_loss": 0.03254026174545288, "train_loss_bc": 0.03254026174545288, "train_loss_llm": 0.0, "grad_norm": 0.08972057700157166, "global_step": 33744, "epoch": 937, "lr": 0.001} +{"train_loss": 0.030638590455055237, "train_loss_bc": 0.03061617538332939, "train_loss_llm": 0.002241581678390503, "grad_norm": 0.05437864363193512, "global_step": 33745, "epoch": 937, "lr": 0.001} +{"train_loss": 0.030659977346658707, "train_loss_bc": 0.030659977346658707, "train_loss_llm": 0.0, "grad_norm": 0.09803259372711182, "global_step": 33746, "epoch": 937, "lr": 0.001} +{"train_loss": 0.03298063948750496, "train_loss_bc": 0.03298063948750496, "train_loss_llm": 0.0, "grad_norm": 0.1257903128862381, "global_step": 33747, "epoch": 937, "lr": 0.001} +{"train_loss": 0.027037393301725388, "train_loss_bc": 0.02703522890806198, "train_loss_llm": 0.00021642446517944336, "grad_norm": 0.16199231147766113, "global_step": 33748, "epoch": 937, "lr": 0.001} +{"train_loss": 0.03391764312982559, "train_loss_bc": 0.033910103142261505, "train_loss_llm": 0.0007539987564086914, "grad_norm": 0.05893576890230179, "global_step": 33749, "epoch": 937, "lr": 0.001} +{"train_loss": 0.03458702564239502, "train_loss_bc": 0.03458702564239502, "train_loss_llm": 0.0, "grad_norm": 0.08533256500959396, "global_step": 33750, "epoch": 937, "lr": 0.001} +{"train_loss": 0.03464966267347336, "train_loss_bc": 0.03464966267347336, "train_loss_llm": 0.0, "grad_norm": 0.1132846549153328, "global_step": 33751, "epoch": 937, "lr": 0.001} +{"train_loss": 0.02580905519425869, "train_loss_bc": 0.02580905519425869, "train_loss_llm": 0.0, "grad_norm": 0.15105828642845154, "global_step": 33752, "epoch": 937, "lr": 0.001} +{"train_loss": 0.029232503846287727, "train_loss_bc": 0.029232503846287727, "train_loss_llm": 0.0, "grad_norm": 0.04282999783754349, "global_step": 33753, "epoch": 937, "lr": 0.001} +{"train_loss": 0.024147067219018936, "train_loss_bc": 0.024147067219018936, "train_loss_llm": 0.0, "grad_norm": 0.049339719116687775, "global_step": 33754, "epoch": 937, "lr": 0.001} +{"train_loss": 0.029024383053183556, "train_loss_bc": 0.029024383053183556, "train_loss_llm": 0.0, "grad_norm": 0.06978429108858109, "global_step": 33755, "epoch": 937, "lr": 0.001} +{"train_loss": 0.0316823311150074, "train_loss_bc": 0.0316823311150074, "train_loss_llm": 0.0, "grad_norm": 0.0834730938076973, "global_step": 33756, "epoch": 937, "lr": 0.001} +{"train_loss": 0.027946751564741135, "train_loss_bc": 0.027946751564741135, "train_loss_llm": 0.0, "grad_norm": 0.043046172708272934, "global_step": 33757, "epoch": 937, "lr": 0.001} +{"train_loss": 0.031304147094488144, "train_loss_bc": 0.031304147094488144, "train_loss_llm": 0.0, "grad_norm": 0.06137058511376381, "global_step": 33758, "epoch": 937, "lr": 0.001} +{"train_loss": 0.028827771544456482, "train_loss_bc": 0.028827771544456482, "train_loss_llm": 0.0, "grad_norm": 0.07561316341161728, "global_step": 33759, "epoch": 937, "lr": 0.001} +{"train_loss": 0.028771890327334404, "train_loss_bc": 0.028771890327334404, "train_loss_llm": 0.0, "grad_norm": 0.09838949143886566, "global_step": 33760, "epoch": 937, "lr": 0.001} +{"train_loss": 0.032493334263563156, "train_loss_bc": 0.032492052763700485, "train_loss_llm": 0.00012803077697753906, "grad_norm": 0.045818813145160675, "global_step": 33761, "epoch": 937, "lr": 0.001} +{"train_loss": 0.031285397708415985, "train_loss_bc": 0.031285397708415985, "train_loss_llm": 0.0, "grad_norm": 0.0772276297211647, "global_step": 33762, "epoch": 937, "lr": 0.001} +{"train_loss": 0.027775969356298447, "train_loss_bc": 0.027775969356298447, "train_loss_llm": 0.0, "grad_norm": 0.10355748981237411, "global_step": 33763, "epoch": 937, "lr": 0.001} +{"train_loss": 0.021398402750492096, "train_loss_bc": 0.021398402750492096, "train_loss_llm": 0.0, "grad_norm": 0.1281229853630066, "global_step": 33764, "epoch": 937, "lr": 0.001} +{"train_loss": 0.031652942299842834, "train_loss_bc": 0.031652942299842834, "train_loss_llm": 0.0, "grad_norm": 0.04030851647257805, "global_step": 33765, "epoch": 937, "lr": 0.001} +{"train_loss": 0.03081240877509117, "train_loss_bc": 0.030810169875621796, "train_loss_llm": 0.00022390484809875488, "grad_norm": 0.07469706982374191, "global_step": 33766, "epoch": 937, "lr": 0.001} +{"train_loss": 0.02910020916412274, "train_loss_bc": 0.03431207686662674, "train_loss_llm": 0.0, "grad_norm": 0.1002059131860733, "global_step": 33767, "epoch": 937, "lr": 0.001, "val_loss": 0.10644148290157318} +{"train_loss": 0.03170672804117203, "train_loss_bc": 0.03170672804117203, "train_loss_llm": 0.0, "grad_norm": 0.12810897827148438, "global_step": 33768, "epoch": 938, "lr": 0.001} +{"train_loss": 0.02773739956319332, "train_loss_bc": 0.02773739956319332, "train_loss_llm": 0.0, "grad_norm": 0.042118847370147705, "global_step": 33769, "epoch": 938, "lr": 0.001} +{"train_loss": 0.025414690375328064, "train_loss_bc": 0.025414690375328064, "train_loss_llm": 0.0, "grad_norm": 0.057334840297698975, "global_step": 33770, "epoch": 938, "lr": 0.001} +{"train_loss": 0.02725307270884514, "train_loss_bc": 0.02725307270884514, "train_loss_llm": 0.0, "grad_norm": 0.09821134805679321, "global_step": 33771, "epoch": 938, "lr": 0.001} +{"train_loss": 0.03237216919660568, "train_loss_bc": 0.03237216919660568, "train_loss_llm": 0.0, "grad_norm": 0.12126484513282776, "global_step": 33772, "epoch": 938, "lr": 0.001} +{"train_loss": 0.030960815027356148, "train_loss_bc": 0.030960815027356148, "train_loss_llm": 0.0, "grad_norm": 0.039985980838537216, "global_step": 33773, "epoch": 938, "lr": 0.001} +{"train_loss": 0.03323149308562279, "train_loss_bc": 0.03323149308562279, "train_loss_llm": 0.0, "grad_norm": 0.06947076320648193, "global_step": 33774, "epoch": 938, "lr": 0.001} +{"train_loss": 0.03465905040502548, "train_loss_bc": 0.034650739282369614, "train_loss_llm": 0.0008309781551361084, "grad_norm": 0.09979638457298279, "global_step": 33775, "epoch": 938, "lr": 0.001} +{"train_loss": 0.02415904961526394, "train_loss_bc": 0.024153515696525574, "train_loss_llm": 0.0005533993244171143, "grad_norm": 0.12341614812612534, "global_step": 33776, "epoch": 938, "lr": 0.001} +{"train_loss": 0.03454750403761864, "train_loss_bc": 0.03454750403761864, "train_loss_llm": 0.0, "grad_norm": 0.05123014375567436, "global_step": 33777, "epoch": 938, "lr": 0.001} +{"train_loss": 0.025968527421355247, "train_loss_bc": 0.025968527421355247, "train_loss_llm": 0.0, "grad_norm": 0.07116858661174774, "global_step": 33778, "epoch": 938, "lr": 0.001} +{"train_loss": 0.02824975736439228, "train_loss_bc": 0.02824975736439228, "train_loss_llm": 0.0, "grad_norm": 0.08731462061405182, "global_step": 33779, "epoch": 938, "lr": 0.001} +{"train_loss": 0.030610578134655952, "train_loss_bc": 0.030610578134655952, "train_loss_llm": 0.0, "grad_norm": 0.11697906255722046, "global_step": 33780, "epoch": 938, "lr": 0.001} +{"train_loss": 0.024458222091197968, "train_loss_bc": 0.024458222091197968, "train_loss_llm": 0.0, "grad_norm": 0.03196282312273979, "global_step": 33781, "epoch": 938, "lr": 0.001} +{"train_loss": 0.027452675625681877, "train_loss_bc": 0.02745109610259533, "train_loss_llm": 0.00015789270401000977, "grad_norm": 0.03929998353123665, "global_step": 33782, "epoch": 938, "lr": 0.001} +{"train_loss": 0.026381835341453552, "train_loss_bc": 0.026381835341453552, "train_loss_llm": 0.0, "grad_norm": 0.06412584334611893, "global_step": 33783, "epoch": 938, "lr": 0.001} +{"train_loss": 0.025430357083678246, "train_loss_bc": 0.025430357083678246, "train_loss_llm": 0.0, "grad_norm": 0.08072800189256668, "global_step": 33784, "epoch": 938, "lr": 0.001} +{"train_loss": 0.03116084635257721, "train_loss_bc": 0.03116084635257721, "train_loss_llm": 0.0, "grad_norm": 0.040414273738861084, "global_step": 33785, "epoch": 938, "lr": 0.001} +{"train_loss": 0.03274918347597122, "train_loss_bc": 0.03274918347597122, "train_loss_llm": 0.0, "grad_norm": 0.055522069334983826, "global_step": 33786, "epoch": 938, "lr": 0.001} +{"train_loss": 0.03447030112147331, "train_loss_bc": 0.03443749621510506, "train_loss_llm": 0.0032806098461151123, "grad_norm": 0.07685493677854538, "global_step": 33787, "epoch": 938, "lr": 0.001} +{"train_loss": 0.027224931865930557, "train_loss_bc": 0.027224931865930557, "train_loss_llm": 0.0, "grad_norm": 0.08888783305883408, "global_step": 33788, "epoch": 938, "lr": 0.001} +{"train_loss": 0.03367799147963524, "train_loss_bc": 0.03367799147963524, "train_loss_llm": 0.0, "grad_norm": 0.056382909417152405, "global_step": 33789, "epoch": 938, "lr": 0.001} +{"train_loss": 0.03119276836514473, "train_loss_bc": 0.031173914670944214, "train_loss_llm": 0.0018853247165679932, "grad_norm": 0.09009135514497757, "global_step": 33790, "epoch": 938, "lr": 0.001} +{"train_loss": 0.026429293677210808, "train_loss_bc": 0.026429293677210808, "train_loss_llm": 0.0, "grad_norm": 0.11955185979604721, "global_step": 33791, "epoch": 938, "lr": 0.001} +{"train_loss": 0.0294721107929945, "train_loss_bc": 0.0294721107929945, "train_loss_llm": 0.0, "grad_norm": 0.1481645554304123, "global_step": 33792, "epoch": 938, "lr": 0.001} +{"train_loss": 0.02634720504283905, "train_loss_bc": 0.02634720504283905, "train_loss_llm": 0.0, "grad_norm": 0.0416763611137867, "global_step": 33793, "epoch": 938, "lr": 0.001} +{"train_loss": 0.030167143791913986, "train_loss_bc": 0.030161773785948753, "train_loss_llm": 0.0005370080471038818, "grad_norm": 0.0880873054265976, "global_step": 33794, "epoch": 938, "lr": 0.001} +{"train_loss": 0.02665666677057743, "train_loss_bc": 0.02665666677057743, "train_loss_llm": 0.0, "grad_norm": 0.11554716527462006, "global_step": 33795, "epoch": 938, "lr": 0.001} +{"train_loss": 0.030799129977822304, "train_loss_bc": 0.03079175390303135, "train_loss_llm": 0.0007376670837402344, "grad_norm": 0.15389743447303772, "global_step": 33796, "epoch": 938, "lr": 0.001} +{"train_loss": 0.025286369025707245, "train_loss_bc": 0.025286369025707245, "train_loss_llm": 0.0, "grad_norm": 0.04524819552898407, "global_step": 33797, "epoch": 938, "lr": 0.001} +{"train_loss": 0.02638704515993595, "train_loss_bc": 0.02638704515993595, "train_loss_llm": 0.0, "grad_norm": 0.07167152315378189, "global_step": 33798, "epoch": 938, "lr": 0.001} +{"train_loss": 0.027834588661789894, "train_loss_bc": 0.027829643338918686, "train_loss_llm": 0.0004945099353790283, "grad_norm": 0.0961356833577156, "global_step": 33799, "epoch": 938, "lr": 0.001} +{"train_loss": 0.031394898891448975, "train_loss_bc": 0.031379010528326035, "train_loss_llm": 0.0015888214111328125, "grad_norm": 0.11592015624046326, "global_step": 33800, "epoch": 938, "lr": 0.001} +{"train_loss": 0.025057395920157433, "train_loss_bc": 0.025057395920157433, "train_loss_llm": 0.0, "grad_norm": 0.034907445311546326, "global_step": 33801, "epoch": 938, "lr": 0.001} +{"train_loss": 0.030938703566789627, "train_loss_bc": 0.030926361680030823, "train_loss_llm": 0.0012341141700744629, "grad_norm": 0.05395819619297981, "global_step": 33802, "epoch": 938, "lr": 0.001} +{"train_loss": 0.029135678791337542, "train_loss_bc": 0.03104393742978573, "train_loss_llm": 0.0, "grad_norm": 0.07994228601455688, "global_step": 33803, "epoch": 938, "lr": 0.001, "val_loss": 0.106337770819664} +{"train_loss": 0.028077639639377594, "train_loss_bc": 0.028077639639377594, "train_loss_llm": 0.0, "grad_norm": 0.0893576443195343, "global_step": 33804, "epoch": 939, "lr": 0.001} +{"train_loss": 0.030016817152500153, "train_loss_bc": 0.030016817152500153, "train_loss_llm": 0.0, "grad_norm": 0.0434439517557621, "global_step": 33805, "epoch": 939, "lr": 0.001} +{"train_loss": 0.029438400641083717, "train_loss_bc": 0.02943243458867073, "train_loss_llm": 0.0005965232849121094, "grad_norm": 0.08644142746925354, "global_step": 33806, "epoch": 939, "lr": 0.001} +{"train_loss": 0.029854226857423782, "train_loss_bc": 0.029854226857423782, "train_loss_llm": 0.0, "grad_norm": 0.10094834864139557, "global_step": 33807, "epoch": 939, "lr": 0.001} +{"train_loss": 0.029191279783844948, "train_loss_bc": 0.029191279783844948, "train_loss_llm": 0.0, "grad_norm": 0.12141207605600357, "global_step": 33808, "epoch": 939, "lr": 0.001} +{"train_loss": 0.02492317371070385, "train_loss_bc": 0.02492317371070385, "train_loss_llm": 0.0, "grad_norm": 0.04013120383024216, "global_step": 33809, "epoch": 939, "lr": 0.001} +{"train_loss": 0.02861793152987957, "train_loss_bc": 0.02861793152987957, "train_loss_llm": 0.0, "grad_norm": 0.05748933553695679, "global_step": 33810, "epoch": 939, "lr": 0.001} +{"train_loss": 0.029053283855319023, "train_loss_bc": 0.029053283855319023, "train_loss_llm": 0.0, "grad_norm": 0.0865134671330452, "global_step": 33811, "epoch": 939, "lr": 0.001} +{"train_loss": 0.027494722977280617, "train_loss_bc": 0.02748963050544262, "train_loss_llm": 0.0005092322826385498, "grad_norm": 0.12284746766090393, "global_step": 33812, "epoch": 939, "lr": 0.001} +{"train_loss": 0.03564782068133354, "train_loss_bc": 0.03564782068133354, "train_loss_llm": 0.0, "grad_norm": 0.04417143389582634, "global_step": 33813, "epoch": 939, "lr": 0.001} +{"train_loss": 0.032683536410331726, "train_loss_bc": 0.032683536410331726, "train_loss_llm": 0.0, "grad_norm": 0.07632548362016678, "global_step": 33814, "epoch": 939, "lr": 0.001} +{"train_loss": 0.03630635142326355, "train_loss_bc": 0.03630635142326355, "train_loss_llm": 0.0, "grad_norm": 0.10990136116743088, "global_step": 33815, "epoch": 939, "lr": 0.001} +{"train_loss": 0.02225053310394287, "train_loss_bc": 0.022245384752750397, "train_loss_llm": 0.0005148649215698242, "grad_norm": 0.12634111940860748, "global_step": 33816, "epoch": 939, "lr": 0.001} +{"train_loss": 0.02667064219713211, "train_loss_bc": 0.02664552628993988, "train_loss_llm": 0.0025115907192230225, "grad_norm": 0.0401662141084671, "global_step": 33817, "epoch": 939, "lr": 0.001} +{"train_loss": 0.02543557994067669, "train_loss_bc": 0.02543557994067669, "train_loss_llm": 0.0, "grad_norm": 0.057321593165397644, "global_step": 33818, "epoch": 939, "lr": 0.001} +{"train_loss": 0.028404565528035164, "train_loss_bc": 0.028404565528035164, "train_loss_llm": 0.0, "grad_norm": 0.0795685425400734, "global_step": 33819, "epoch": 939, "lr": 0.001} +{"train_loss": 0.029986398294568062, "train_loss_bc": 0.029982473701238632, "train_loss_llm": 0.00039249658584594727, "grad_norm": 0.11082490533590317, "global_step": 33820, "epoch": 939, "lr": 0.001} +{"train_loss": 0.03278610110282898, "train_loss_bc": 0.03278610110282898, "train_loss_llm": 0.0, "grad_norm": 0.0318528488278389, "global_step": 33821, "epoch": 939, "lr": 0.001} +{"train_loss": 0.026339828968048096, "train_loss_bc": 0.026339828968048096, "train_loss_llm": 0.0, "grad_norm": 0.05286417901515961, "global_step": 33822, "epoch": 939, "lr": 0.001} +{"train_loss": 0.025187311694025993, "train_loss_bc": 0.025187311694025993, "train_loss_llm": 0.0, "grad_norm": 0.06046255677938461, "global_step": 33823, "epoch": 939, "lr": 0.001} +{"train_loss": 0.026341769844293594, "train_loss_bc": 0.026341769844293594, "train_loss_llm": 0.0, "grad_norm": 0.08666659146547318, "global_step": 33824, "epoch": 939, "lr": 0.001} +{"train_loss": 0.029518231749534607, "train_loss_bc": 0.029518231749534607, "train_loss_llm": 0.0, "grad_norm": 0.036311689764261246, "global_step": 33825, "epoch": 939, "lr": 0.001} +{"train_loss": 0.028653183951973915, "train_loss_bc": 0.02863124944269657, "train_loss_llm": 0.0021935105323791504, "grad_norm": 0.06060047075152397, "global_step": 33826, "epoch": 939, "lr": 0.001} +{"train_loss": 0.022443050518631935, "train_loss_bc": 0.022443050518631935, "train_loss_llm": 0.0, "grad_norm": 0.07049492746591568, "global_step": 33827, "epoch": 939, "lr": 0.001} +{"train_loss": 0.0257157851010561, "train_loss_bc": 0.025693286210298538, "train_loss_llm": 0.002249807119369507, "grad_norm": 0.08831844478845596, "global_step": 33828, "epoch": 939, "lr": 0.001} +{"train_loss": 0.030200328677892685, "train_loss_bc": 0.030195336788892746, "train_loss_llm": 0.0004992187023162842, "grad_norm": 0.043707188218832016, "global_step": 33829, "epoch": 939, "lr": 0.001} +{"train_loss": 0.0324542336165905, "train_loss_bc": 0.0324542336165905, "train_loss_llm": 0.0, "grad_norm": 0.0775001049041748, "global_step": 33830, "epoch": 939, "lr": 0.001} +{"train_loss": 0.03158386051654816, "train_loss_bc": 0.03158386051654816, "train_loss_llm": 0.0, "grad_norm": 0.10269469022750854, "global_step": 33831, "epoch": 939, "lr": 0.001} +{"train_loss": 0.026448212563991547, "train_loss_bc": 0.026446761563420296, "train_loss_llm": 0.0001450181007385254, "grad_norm": 0.1256115734577179, "global_step": 33832, "epoch": 939, "lr": 0.001} +{"train_loss": 0.030686259269714355, "train_loss_bc": 0.030686259269714355, "train_loss_llm": 0.0, "grad_norm": 0.04536266624927521, "global_step": 33833, "epoch": 939, "lr": 0.001} +{"train_loss": 0.02948799356818199, "train_loss_bc": 0.02948799356818199, "train_loss_llm": 0.0, "grad_norm": 0.06766928732395172, "global_step": 33834, "epoch": 939, "lr": 0.001} +{"train_loss": 0.03200679272413254, "train_loss_bc": 0.03200679272413254, "train_loss_llm": 0.0, "grad_norm": 0.09720583260059357, "global_step": 33835, "epoch": 939, "lr": 0.001} +{"train_loss": 0.03188147395849228, "train_loss_bc": 0.03188147395849228, "train_loss_llm": 0.0, "grad_norm": 0.12260843068361282, "global_step": 33836, "epoch": 939, "lr": 0.001} +{"train_loss": 0.03278597816824913, "train_loss_bc": 0.03278597816824913, "train_loss_llm": 0.0, "grad_norm": 0.03866429626941681, "global_step": 33837, "epoch": 939, "lr": 0.001} +{"train_loss": 0.02793741784989834, "train_loss_bc": 0.027927566319704056, "train_loss_llm": 0.0009851157665252686, "grad_norm": 0.06179894879460335, "global_step": 33838, "epoch": 939, "lr": 0.001} +{"train_loss": 0.029026458505541086, "train_loss_bc": 0.028441788628697395, "train_loss_llm": 0.0, "grad_norm": 0.07940814644098282, "global_step": 33839, "epoch": 939, "lr": 0.001, "val_loss": 0.10692855715751648} +{"train_loss": 0.030393850058317184, "train_loss_bc": 0.030393850058317184, "train_loss_llm": 0.0, "grad_norm": 0.11380273848772049, "global_step": 33840, "epoch": 940, "lr": 0.001} +{"train_loss": 0.02587125077843666, "train_loss_bc": 0.02587125077843666, "train_loss_llm": 0.0, "grad_norm": 0.023869939148426056, "global_step": 33841, "epoch": 940, "lr": 0.001} +{"train_loss": 0.03168071433901787, "train_loss_bc": 0.03168071433901787, "train_loss_llm": 0.0, "grad_norm": 0.047429732978343964, "global_step": 33842, "epoch": 940, "lr": 0.001} +{"train_loss": 0.03055422753095627, "train_loss_bc": 0.03055422753095627, "train_loss_llm": 0.0, "grad_norm": 0.06823202222585678, "global_step": 33843, "epoch": 940, "lr": 0.001} +{"train_loss": 0.030414624139666557, "train_loss_bc": 0.030414624139666557, "train_loss_llm": 0.0, "grad_norm": 0.08991474658250809, "global_step": 33844, "epoch": 940, "lr": 0.001} +{"train_loss": 0.03205325827002525, "train_loss_bc": 0.03204241022467613, "train_loss_llm": 0.0010847747325897217, "grad_norm": 0.0408809669315815, "global_step": 33845, "epoch": 940, "lr": 0.001} +{"train_loss": 0.028444742783904076, "train_loss_bc": 0.028444742783904076, "train_loss_llm": 0.0, "grad_norm": 0.07351921498775482, "global_step": 33846, "epoch": 940, "lr": 0.001} +{"train_loss": 0.03350040316581726, "train_loss_bc": 0.03350040316581726, "train_loss_llm": 0.0, "grad_norm": 0.10011489689350128, "global_step": 33847, "epoch": 940, "lr": 0.001} +{"train_loss": 0.0319824293255806, "train_loss_bc": 0.0319824293255806, "train_loss_llm": 0.0, "grad_norm": 0.12789395451545715, "global_step": 33848, "epoch": 940, "lr": 0.001} +{"train_loss": 0.0211179181933403, "train_loss_bc": 0.0211179181933403, "train_loss_llm": 0.0, "grad_norm": 0.04369332268834114, "global_step": 33849, "epoch": 940, "lr": 0.001} +{"train_loss": 0.03452195227146149, "train_loss_bc": 0.03452195227146149, "train_loss_llm": 0.0, "grad_norm": 0.07593493163585663, "global_step": 33850, "epoch": 940, "lr": 0.001} +{"train_loss": 0.028526103124022484, "train_loss_bc": 0.028526103124022484, "train_loss_llm": 0.0, "grad_norm": 0.11598920077085495, "global_step": 33851, "epoch": 940, "lr": 0.001} +{"train_loss": 0.021938305348157883, "train_loss_bc": 0.021938305348157883, "train_loss_llm": 0.0, "grad_norm": 0.1330745816230774, "global_step": 33852, "epoch": 940, "lr": 0.001} +{"train_loss": 0.0351824127137661, "train_loss_bc": 0.0351824127137661, "train_loss_llm": 0.0, "grad_norm": 0.047499481588602066, "global_step": 33853, "epoch": 940, "lr": 0.001} +{"train_loss": 0.029331553727388382, "train_loss_bc": 0.029331553727388382, "train_loss_llm": 0.0, "grad_norm": 0.06751570850610733, "global_step": 33854, "epoch": 940, "lr": 0.001} +{"train_loss": 0.02884167619049549, "train_loss_bc": 0.02884167619049549, "train_loss_llm": 0.0, "grad_norm": 0.09002014994621277, "global_step": 33855, "epoch": 940, "lr": 0.001} +{"train_loss": 0.03043823316693306, "train_loss_bc": 0.03043823316693306, "train_loss_llm": 0.0, "grad_norm": 0.12411842495203018, "global_step": 33856, "epoch": 940, "lr": 0.001} +{"train_loss": 0.028742976486682892, "train_loss_bc": 0.028742976486682892, "train_loss_llm": 0.0, "grad_norm": 0.04087366536259651, "global_step": 33857, "epoch": 940, "lr": 0.001} +{"train_loss": 0.03144953027367592, "train_loss_bc": 0.03144953027367592, "train_loss_llm": 0.0, "grad_norm": 0.060055237263441086, "global_step": 33858, "epoch": 940, "lr": 0.001} +{"train_loss": 0.025736866518855095, "train_loss_bc": 0.025734176859259605, "train_loss_llm": 0.0002690255641937256, "grad_norm": 0.07981657236814499, "global_step": 33859, "epoch": 940, "lr": 0.001} +{"train_loss": 0.02977188676595688, "train_loss_bc": 0.02977188676595688, "train_loss_llm": 0.0, "grad_norm": 0.09755726903676987, "global_step": 33860, "epoch": 940, "lr": 0.001} +{"train_loss": 0.028440047055482864, "train_loss_bc": 0.028440047055482864, "train_loss_llm": 0.0, "grad_norm": 0.04035225138068199, "global_step": 33861, "epoch": 940, "lr": 0.001} +{"train_loss": 0.0300084687769413, "train_loss_bc": 0.0300084687769413, "train_loss_llm": 0.0, "grad_norm": 0.06925107538700104, "global_step": 33862, "epoch": 940, "lr": 0.001} +{"train_loss": 0.027353448793292046, "train_loss_bc": 0.027338920161128044, "train_loss_llm": 0.001452803611755371, "grad_norm": 0.09264592081308365, "global_step": 33863, "epoch": 940, "lr": 0.001} +{"train_loss": 0.026283320039510727, "train_loss_bc": 0.026257075369358063, "train_loss_llm": 0.0026244521141052246, "grad_norm": 0.10741635411977768, "global_step": 33864, "epoch": 940, "lr": 0.001} +{"train_loss": 0.026851661503314972, "train_loss_bc": 0.026850715279579163, "train_loss_llm": 9.462237358093262e-05, "grad_norm": 0.04573791101574898, "global_step": 33865, "epoch": 940, "lr": 0.001} +{"train_loss": 0.027638304978609085, "train_loss_bc": 0.027638304978609085, "train_loss_llm": 0.0, "grad_norm": 0.07363725453615189, "global_step": 33866, "epoch": 940, "lr": 0.001} +{"train_loss": 0.027081262320280075, "train_loss_bc": 0.027079977095127106, "train_loss_llm": 0.0001285076141357422, "grad_norm": 0.09143874794244766, "global_step": 33867, "epoch": 940, "lr": 0.001} +{"train_loss": 0.02752661518752575, "train_loss_bc": 0.02752661518752575, "train_loss_llm": 0.0, "grad_norm": 0.12578411400318146, "global_step": 33868, "epoch": 940, "lr": 0.001} +{"train_loss": 0.028002692386507988, "train_loss_bc": 0.028002692386507988, "train_loss_llm": 0.0, "grad_norm": 0.05039416626095772, "global_step": 33869, "epoch": 940, "lr": 0.001} +{"train_loss": 0.02704141102731228, "train_loss_bc": 0.02704141102731228, "train_loss_llm": 0.0, "grad_norm": 0.09829964488744736, "global_step": 33870, "epoch": 940, "lr": 0.001} +{"train_loss": 0.03441357612609863, "train_loss_bc": 0.03441357612609863, "train_loss_llm": 0.0, "grad_norm": 0.14300018548965454, "global_step": 33871, "epoch": 940, "lr": 0.001} +{"train_loss": 0.027939507737755775, "train_loss_bc": 0.027939507737755775, "train_loss_llm": 0.0, "grad_norm": 0.17410780489444733, "global_step": 33872, "epoch": 940, "lr": 0.001} +{"train_loss": 0.031224319711327553, "train_loss_bc": 0.031224319711327553, "train_loss_llm": 0.0, "grad_norm": 0.04640752822160721, "global_step": 33873, "epoch": 940, "lr": 0.001} +{"train_loss": 0.029996110126376152, "train_loss_bc": 0.029996110126376152, "train_loss_llm": 0.0, "grad_norm": 0.07197026163339615, "global_step": 33874, "epoch": 940, "lr": 0.001} +{"train_loss": 0.02902345002318422, "train_loss_bc": 0.024548539891839027, "train_loss_llm": 0.0, "grad_norm": 0.09123323857784271, "global_step": 33875, "epoch": 940, "lr": 0.001, "train/cumulative_reward": 64.62034214521364, "train/mean_score": 10.714831752711731, "train/success_rate": 0.2, "test/cumulative_reward": 39.142430494337326, "test/mean_score": 6.391824757583149, "test/success_rate": 0.12, "val_loss": 0.107463039457798, "train_action_mse_error": 0.032209448516368866} +{"train_loss": 0.024384979158639908, "train_loss_bc": 0.024380220100283623, "train_loss_llm": 0.00047597289085388184, "grad_norm": 0.10040953010320663, "global_step": 33876, "epoch": 941, "lr": 0.001} +{"train_loss": 0.027767546474933624, "train_loss_bc": 0.027767546474933624, "train_loss_llm": 0.0, "grad_norm": 0.03343084454536438, "global_step": 33877, "epoch": 941, "lr": 0.001} +{"train_loss": 0.023945258930325508, "train_loss_bc": 0.023938816040754318, "train_loss_llm": 0.0006442070007324219, "grad_norm": 0.052813414484262466, "global_step": 33878, "epoch": 941, "lr": 0.001} +{"train_loss": 0.026594430208206177, "train_loss_bc": 0.026594430208206177, "train_loss_llm": 0.0, "grad_norm": 0.07190494984388351, "global_step": 33879, "epoch": 941, "lr": 0.001} +{"train_loss": 0.028704462572932243, "train_loss_bc": 0.028704462572932243, "train_loss_llm": 0.0, "grad_norm": 0.07702476531267166, "global_step": 33880, "epoch": 941, "lr": 0.001} +{"train_loss": 0.031237099319696426, "train_loss_bc": 0.031237099319696426, "train_loss_llm": 0.0, "grad_norm": 0.03982070833444595, "global_step": 33881, "epoch": 941, "lr": 0.001} +{"train_loss": 0.02614728920161724, "train_loss_bc": 0.026145897805690765, "train_loss_llm": 0.00013905763626098633, "grad_norm": 0.07298644632101059, "global_step": 33882, "epoch": 941, "lr": 0.001} +{"train_loss": 0.02435280941426754, "train_loss_bc": 0.02435280941426754, "train_loss_llm": 0.0, "grad_norm": 0.09384331852197647, "global_step": 33883, "epoch": 941, "lr": 0.001} +{"train_loss": 0.02986050955951214, "train_loss_bc": 0.02985602617263794, "train_loss_llm": 0.00044840574264526367, "grad_norm": 0.10369537770748138, "global_step": 33884, "epoch": 941, "lr": 0.001} +{"train_loss": 0.030433524399995804, "train_loss_bc": 0.030417263507843018, "train_loss_llm": 0.0016260743141174316, "grad_norm": 0.05162225663661957, "global_step": 33885, "epoch": 941, "lr": 0.001} +{"train_loss": 0.026289304718375206, "train_loss_bc": 0.026289304718375206, "train_loss_llm": 0.0, "grad_norm": 0.08329775929450989, "global_step": 33886, "epoch": 941, "lr": 0.001} +{"train_loss": 0.0332663431763649, "train_loss_bc": 0.0332663431763649, "train_loss_llm": 0.0, "grad_norm": 0.13352830708026886, "global_step": 33887, "epoch": 941, "lr": 0.001} +{"train_loss": 0.036453794687986374, "train_loss_bc": 0.036453794687986374, "train_loss_llm": 0.0, "grad_norm": 0.17355096340179443, "global_step": 33888, "epoch": 941, "lr": 0.001} +{"train_loss": 0.03003375604748726, "train_loss_bc": 0.03003375604748726, "train_loss_llm": 0.0, "grad_norm": 0.05115663260221481, "global_step": 33889, "epoch": 941, "lr": 0.001} +{"train_loss": 0.02871205471456051, "train_loss_bc": 0.02871205471456051, "train_loss_llm": 0.0, "grad_norm": 0.08752215653657913, "global_step": 33890, "epoch": 941, "lr": 0.001} +{"train_loss": 0.02661341428756714, "train_loss_bc": 0.026609469205141068, "train_loss_llm": 0.00039446353912353516, "grad_norm": 0.11527624726295471, "global_step": 33891, "epoch": 941, "lr": 0.001} +{"train_loss": 0.021289866417646408, "train_loss_bc": 0.0212873388081789, "train_loss_llm": 0.00025284290313720703, "grad_norm": 0.142498180270195, "global_step": 33892, "epoch": 941, "lr": 0.001} +{"train_loss": 0.022633718326687813, "train_loss_bc": 0.022633718326687813, "train_loss_llm": 0.0, "grad_norm": 0.029305502772331238, "global_step": 33893, "epoch": 941, "lr": 0.001} +{"train_loss": 0.030070386826992035, "train_loss_bc": 0.030070386826992035, "train_loss_llm": 0.0, "grad_norm": 0.049460671842098236, "global_step": 33894, "epoch": 941, "lr": 0.001} +{"train_loss": 0.02668806165456772, "train_loss_bc": 0.02668806165456772, "train_loss_llm": 0.0, "grad_norm": 0.07779056578874588, "global_step": 33895, "epoch": 941, "lr": 0.001} +{"train_loss": 0.028440721333026886, "train_loss_bc": 0.028440721333026886, "train_loss_llm": 0.0, "grad_norm": 0.10082260519266129, "global_step": 33896, "epoch": 941, "lr": 0.001} +{"train_loss": 0.031109824776649475, "train_loss_bc": 0.031109824776649475, "train_loss_llm": 0.0, "grad_norm": 0.03561814874410629, "global_step": 33897, "epoch": 941, "lr": 0.001} +{"train_loss": 0.03336765244603157, "train_loss_bc": 0.03336765244603157, "train_loss_llm": 0.0, "grad_norm": 0.062008898705244064, "global_step": 33898, "epoch": 941, "lr": 0.001} +{"train_loss": 0.029459483921527863, "train_loss_bc": 0.029459483921527863, "train_loss_llm": 0.0, "grad_norm": 0.08820046484470367, "global_step": 33899, "epoch": 941, "lr": 0.001} +{"train_loss": 0.029008042067289352, "train_loss_bc": 0.02899542823433876, "train_loss_llm": 0.001261383295059204, "grad_norm": 0.10357967019081116, "global_step": 33900, "epoch": 941, "lr": 0.001} +{"train_loss": 0.03565668314695358, "train_loss_bc": 0.03565668314695358, "train_loss_llm": 0.0, "grad_norm": 0.06440076231956482, "global_step": 33901, "epoch": 941, "lr": 0.001} +{"train_loss": 0.025107761844992638, "train_loss_bc": 0.025107761844992638, "train_loss_llm": 0.0, "grad_norm": 0.09205111116170883, "global_step": 33902, "epoch": 941, "lr": 0.001} +{"train_loss": 0.030740197747945786, "train_loss_bc": 0.030740197747945786, "train_loss_llm": 0.0, "grad_norm": 0.12292945384979248, "global_step": 33903, "epoch": 941, "lr": 0.001} +{"train_loss": 0.028086582198739052, "train_loss_bc": 0.028086582198739052, "train_loss_llm": 0.0, "grad_norm": 0.1506093144416809, "global_step": 33904, "epoch": 941, "lr": 0.001} +{"train_loss": 0.030433595180511475, "train_loss_bc": 0.030433595180511475, "train_loss_llm": 0.0, "grad_norm": 0.055142998695373535, "global_step": 33905, "epoch": 941, "lr": 0.001} +{"train_loss": 0.031577564775943756, "train_loss_bc": 0.031571224331855774, "train_loss_llm": 0.0006341040134429932, "grad_norm": 0.09811308979988098, "global_step": 33906, "epoch": 941, "lr": 0.001} +{"train_loss": 0.02741074189543724, "train_loss_bc": 0.02741074189543724, "train_loss_llm": 0.0, "grad_norm": 0.11917315423488617, "global_step": 33907, "epoch": 941, "lr": 0.001} +{"train_loss": 0.03175785019993782, "train_loss_bc": 0.03175202012062073, "train_loss_llm": 0.0005829334259033203, "grad_norm": 0.1606478989124298, "global_step": 33908, "epoch": 941, "lr": 0.001} +{"train_loss": 0.029557015746831894, "train_loss_bc": 0.029557015746831894, "train_loss_llm": 0.0, "grad_norm": 0.04376227408647537, "global_step": 33909, "epoch": 941, "lr": 0.001} +{"train_loss": 0.031869471073150635, "train_loss_bc": 0.031869471073150635, "train_loss_llm": 0.0, "grad_norm": 0.07432835549116135, "global_step": 33910, "epoch": 941, "lr": 0.001} +{"train_loss": 0.029097737951411143, "train_loss_bc": 0.03843348100781441, "train_loss_llm": 0.002328813076019287, "grad_norm": 0.09704392403364182, "global_step": 33911, "epoch": 941, "lr": 0.001, "val_loss": 0.10742535442113876} +{"train_loss": 0.02954588271677494, "train_loss_bc": 0.02954588271677494, "train_loss_llm": 0.0, "grad_norm": 0.11214056611061096, "global_step": 33912, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02802661992609501, "train_loss_bc": 0.028018631041049957, "train_loss_llm": 0.0007988810539245605, "grad_norm": 0.042117245495319366, "global_step": 33913, "epoch": 942, "lr": 0.001} +{"train_loss": 0.027734512463212013, "train_loss_bc": 0.027726154774427414, "train_loss_llm": 0.0008357763290405273, "grad_norm": 0.05966435745358467, "global_step": 33914, "epoch": 942, "lr": 0.001} +{"train_loss": 0.028418613597750664, "train_loss_bc": 0.028418613597750664, "train_loss_llm": 0.0, "grad_norm": 0.09487404674291611, "global_step": 33915, "epoch": 942, "lr": 0.001} +{"train_loss": 0.03439316898584366, "train_loss_bc": 0.03439316898584366, "train_loss_llm": 0.0, "grad_norm": 0.10763564705848694, "global_step": 33916, "epoch": 942, "lr": 0.001} +{"train_loss": 0.031207265332341194, "train_loss_bc": 0.031207265332341194, "train_loss_llm": 0.0, "grad_norm": 0.03368044272065163, "global_step": 33917, "epoch": 942, "lr": 0.001} +{"train_loss": 0.03249736502766609, "train_loss_bc": 0.03249736502766609, "train_loss_llm": 0.0, "grad_norm": 0.06543285399675369, "global_step": 33918, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02975558303296566, "train_loss_bc": 0.029747748747467995, "train_loss_llm": 0.0007834136486053467, "grad_norm": 0.08234492689371109, "global_step": 33919, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02914498932659626, "train_loss_bc": 0.029134323820471764, "train_loss_llm": 0.0010665655136108398, "grad_norm": 0.09999807924032211, "global_step": 33920, "epoch": 942, "lr": 0.001} +{"train_loss": 0.024188503623008728, "train_loss_bc": 0.024188503623008728, "train_loss_llm": 0.0, "grad_norm": 0.03527344763278961, "global_step": 33921, "epoch": 942, "lr": 0.001} +{"train_loss": 0.028403956443071365, "train_loss_bc": 0.028403956443071365, "train_loss_llm": 0.0, "grad_norm": 0.055443957448005676, "global_step": 33922, "epoch": 942, "lr": 0.001} +{"train_loss": 0.03313495218753815, "train_loss_bc": 0.033133044838905334, "train_loss_llm": 0.00019082427024841309, "grad_norm": 0.08760605752468109, "global_step": 33923, "epoch": 942, "lr": 0.001} +{"train_loss": 0.03330881521105766, "train_loss_bc": 0.033282551914453506, "train_loss_llm": 0.002626270055770874, "grad_norm": 0.11084374040365219, "global_step": 33924, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02735491283237934, "train_loss_bc": 0.02735491283237934, "train_loss_llm": 0.0, "grad_norm": 0.03376172110438347, "global_step": 33925, "epoch": 942, "lr": 0.001} +{"train_loss": 0.024388426914811134, "train_loss_bc": 0.024378301575779915, "train_loss_llm": 0.0010124742984771729, "grad_norm": 0.047574106603860855, "global_step": 33926, "epoch": 942, "lr": 0.001} +{"train_loss": 0.031341325491666794, "train_loss_bc": 0.031341325491666794, "train_loss_llm": 0.0, "grad_norm": 0.08176466822624207, "global_step": 33927, "epoch": 942, "lr": 0.001} +{"train_loss": 0.028423119336366653, "train_loss_bc": 0.028423119336366653, "train_loss_llm": 0.0, "grad_norm": 0.10059254616498947, "global_step": 33928, "epoch": 942, "lr": 0.001} +{"train_loss": 0.029367344453930855, "train_loss_bc": 0.029367344453930855, "train_loss_llm": 0.0, "grad_norm": 0.04052111133933067, "global_step": 33929, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02887004055082798, "train_loss_bc": 0.028863895684480667, "train_loss_llm": 0.0006144940853118896, "grad_norm": 0.0686292052268982, "global_step": 33930, "epoch": 942, "lr": 0.001} +{"train_loss": 0.026011517271399498, "train_loss_bc": 0.026011517271399498, "train_loss_llm": 0.0, "grad_norm": 0.08724545687437057, "global_step": 33931, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02941752038896084, "train_loss_bc": 0.02941463142633438, "train_loss_llm": 0.00028887391090393066, "grad_norm": 0.11960184574127197, "global_step": 33932, "epoch": 942, "lr": 0.001} +{"train_loss": 0.029882848262786865, "train_loss_bc": 0.029882848262786865, "train_loss_llm": 0.0, "grad_norm": 0.039357349276542664, "global_step": 33933, "epoch": 942, "lr": 0.001} +{"train_loss": 0.027657683938741684, "train_loss_bc": 0.027657683938741684, "train_loss_llm": 0.0, "grad_norm": 0.05471694841980934, "global_step": 33934, "epoch": 942, "lr": 0.001} +{"train_loss": 0.03394732624292374, "train_loss_bc": 0.03394732624292374, "train_loss_llm": 0.0, "grad_norm": 0.08339597284793854, "global_step": 33935, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02825000509619713, "train_loss_bc": 0.02825000509619713, "train_loss_llm": 0.0, "grad_norm": 0.1005907878279686, "global_step": 33936, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02970530278980732, "train_loss_bc": 0.02970518358051777, "train_loss_llm": 1.1920928955078125e-05, "grad_norm": 0.040475815534591675, "global_step": 33937, "epoch": 942, "lr": 0.001} +{"train_loss": 0.027022816240787506, "train_loss_bc": 0.027022816240787506, "train_loss_llm": 0.0, "grad_norm": 0.06465288996696472, "global_step": 33938, "epoch": 942, "lr": 0.001} +{"train_loss": 0.026160087436437607, "train_loss_bc": 0.026160087436437607, "train_loss_llm": 0.0, "grad_norm": 0.09384404867887497, "global_step": 33939, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02604568563401699, "train_loss_bc": 0.02604568563401699, "train_loss_llm": 0.0, "grad_norm": 0.11112473160028458, "global_step": 33940, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02782701887190342, "train_loss_bc": 0.02782701887190342, "train_loss_llm": 0.0, "grad_norm": 0.03365664556622505, "global_step": 33941, "epoch": 942, "lr": 0.001} +{"train_loss": 0.03252546116709709, "train_loss_bc": 0.03252546116709709, "train_loss_llm": 0.0, "grad_norm": 0.05537940189242363, "global_step": 33942, "epoch": 942, "lr": 0.001} +{"train_loss": 0.022202972322702408, "train_loss_bc": 0.022202972322702408, "train_loss_llm": 0.0, "grad_norm": 0.06743225455284119, "global_step": 33943, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02703637070953846, "train_loss_bc": 0.02703637070953846, "train_loss_llm": 0.0, "grad_norm": 0.08107098191976547, "global_step": 33944, "epoch": 942, "lr": 0.001} +{"train_loss": 0.03057064488530159, "train_loss_bc": 0.03056907467544079, "train_loss_llm": 0.000157088041305542, "grad_norm": 0.03471154347062111, "global_step": 33945, "epoch": 942, "lr": 0.001} +{"train_loss": 0.03154151514172554, "train_loss_bc": 0.03153586387634277, "train_loss_llm": 0.0005651414394378662, "grad_norm": 0.06124581769108772, "global_step": 33946, "epoch": 942, "lr": 0.001} +{"train_loss": 0.02884594045786394, "train_loss_bc": 0.023126978427171707, "train_loss_llm": 0.00167044997215271, "grad_norm": 0.08000684529542923, "global_step": 33947, "epoch": 942, "lr": 0.001, "val_loss": 0.10725751519203186} +{"train_loss": 0.026406951248645782, "train_loss_bc": 0.026393597945570946, "train_loss_llm": 0.0013352632522583008, "grad_norm": 0.09624245762825012, "global_step": 33948, "epoch": 943, "lr": 0.001} +{"train_loss": 0.021657641977071762, "train_loss_bc": 0.021657641977071762, "train_loss_llm": 0.0, "grad_norm": 0.03636524826288223, "global_step": 33949, "epoch": 943, "lr": 0.001} +{"train_loss": 0.029308149591088295, "train_loss_bc": 0.029303420335054398, "train_loss_llm": 0.0004729628562927246, "grad_norm": 0.04755767434835434, "global_step": 33950, "epoch": 943, "lr": 0.001} +{"train_loss": 0.0327414944767952, "train_loss_bc": 0.03272969275712967, "train_loss_llm": 0.0011802613735198975, "grad_norm": 0.06728143244981766, "global_step": 33951, "epoch": 943, "lr": 0.001} +{"train_loss": 0.027766183018684387, "train_loss_bc": 0.027766183018684387, "train_loss_llm": 0.0, "grad_norm": 0.08430686593055725, "global_step": 33952, "epoch": 943, "lr": 0.001} +{"train_loss": 0.022467052564024925, "train_loss_bc": 0.022460270673036575, "train_loss_llm": 0.0006781220436096191, "grad_norm": 0.027737583965063095, "global_step": 33953, "epoch": 943, "lr": 0.001} +{"train_loss": 0.029788078740239143, "train_loss_bc": 0.02978472411632538, "train_loss_llm": 0.00033542513847351074, "grad_norm": 0.05039624497294426, "global_step": 33954, "epoch": 943, "lr": 0.001} +{"train_loss": 0.02474658004939556, "train_loss_bc": 0.024744603782892227, "train_loss_llm": 0.00019761919975280762, "grad_norm": 0.06556326895952225, "global_step": 33955, "epoch": 943, "lr": 0.001} +{"train_loss": 0.024608636274933815, "train_loss_bc": 0.024608636274933815, "train_loss_llm": 0.0, "grad_norm": 0.08910912275314331, "global_step": 33956, "epoch": 943, "lr": 0.001} +{"train_loss": 0.03161542862653732, "train_loss_bc": 0.03161542862653732, "train_loss_llm": 0.0, "grad_norm": 0.0396040640771389, "global_step": 33957, "epoch": 943, "lr": 0.001} +{"train_loss": 0.026897301897406578, "train_loss_bc": 0.02689090371131897, "train_loss_llm": 0.0006398260593414307, "grad_norm": 0.055485907942056656, "global_step": 33958, "epoch": 943, "lr": 0.001} +{"train_loss": 0.031889528036117554, "train_loss_bc": 0.031889528036117554, "train_loss_llm": 0.0, "grad_norm": 0.08099500834941864, "global_step": 33959, "epoch": 943, "lr": 0.001} +{"train_loss": 0.027458785101771355, "train_loss_bc": 0.027458785101771355, "train_loss_llm": 0.0, "grad_norm": 0.09800101816654205, "global_step": 33960, "epoch": 943, "lr": 0.001} +{"train_loss": 0.0300447940826416, "train_loss_bc": 0.0300447940826416, "train_loss_llm": 0.0, "grad_norm": 0.04404440522193909, "global_step": 33961, "epoch": 943, "lr": 0.001} +{"train_loss": 0.02881682850420475, "train_loss_bc": 0.02881682850420475, "train_loss_llm": 0.0, "grad_norm": 0.06416262686252594, "global_step": 33962, "epoch": 943, "lr": 0.001} +{"train_loss": 0.026946116238832474, "train_loss_bc": 0.026935463771224022, "train_loss_llm": 0.0010652542114257812, "grad_norm": 0.08012475073337555, "global_step": 33963, "epoch": 943, "lr": 0.001} +{"train_loss": 0.02763958089053631, "train_loss_bc": 0.02763958089053631, "train_loss_llm": 0.0, "grad_norm": 0.09285169839859009, "global_step": 33964, "epoch": 943, "lr": 0.001} +{"train_loss": 0.030916867777705193, "train_loss_bc": 0.030916867777705193, "train_loss_llm": 0.0, "grad_norm": 0.042777471244335175, "global_step": 33965, "epoch": 943, "lr": 0.001} +{"train_loss": 0.028473634272813797, "train_loss_bc": 0.028473634272813797, "train_loss_llm": 0.0, "grad_norm": 0.06323857605457306, "global_step": 33966, "epoch": 943, "lr": 0.001} +{"train_loss": 0.033097244799137115, "train_loss_bc": 0.033097244799137115, "train_loss_llm": 0.0, "grad_norm": 0.09867826849222183, "global_step": 33967, "epoch": 943, "lr": 0.001} +{"train_loss": 0.028337175026535988, "train_loss_bc": 0.028337175026535988, "train_loss_llm": 0.0, "grad_norm": 0.12414155155420303, "global_step": 33968, "epoch": 943, "lr": 0.001} +{"train_loss": 0.029404837638139725, "train_loss_bc": 0.029404837638139725, "train_loss_llm": 0.0, "grad_norm": 0.04962391406297684, "global_step": 33969, "epoch": 943, "lr": 0.001} +{"train_loss": 0.028187772259116173, "train_loss_bc": 0.028187772259116173, "train_loss_llm": 0.0, "grad_norm": 0.0703800618648529, "global_step": 33970, "epoch": 943, "lr": 0.001} +{"train_loss": 0.02863975055515766, "train_loss_bc": 0.02863975055515766, "train_loss_llm": 0.0, "grad_norm": 0.08243738114833832, "global_step": 33971, "epoch": 943, "lr": 0.001} +{"train_loss": 0.029383603483438492, "train_loss_bc": 0.029383603483438492, "train_loss_llm": 0.0, "grad_norm": 0.09529662877321243, "global_step": 33972, "epoch": 943, "lr": 0.001} +{"train_loss": 0.02889191173017025, "train_loss_bc": 0.028870858252048492, "train_loss_llm": 0.0021052956581115723, "grad_norm": 0.035210151225328445, "global_step": 33973, "epoch": 943, "lr": 0.001} +{"train_loss": 0.02756345644593239, "train_loss_bc": 0.02756345644593239, "train_loss_llm": 0.0, "grad_norm": 0.049891311675310135, "global_step": 33974, "epoch": 943, "lr": 0.001} +{"train_loss": 0.03172314912080765, "train_loss_bc": 0.03172314912080765, "train_loss_llm": 0.0, "grad_norm": 0.06634297221899033, "global_step": 33975, "epoch": 943, "lr": 0.001} +{"train_loss": 0.034227024763822556, "train_loss_bc": 0.034227024763822556, "train_loss_llm": 0.0, "grad_norm": 0.09164391458034515, "global_step": 33976, "epoch": 943, "lr": 0.001} +{"train_loss": 0.02992440201342106, "train_loss_bc": 0.02992440201342106, "train_loss_llm": 0.0, "grad_norm": 0.035076986998319626, "global_step": 33977, "epoch": 943, "lr": 0.001} +{"train_loss": 0.027992943301796913, "train_loss_bc": 0.027992943301796913, "train_loss_llm": 0.0, "grad_norm": 0.06038306653499603, "global_step": 33978, "epoch": 943, "lr": 0.001} +{"train_loss": 0.03471382334828377, "train_loss_bc": 0.03471382334828377, "train_loss_llm": 0.0, "grad_norm": 0.08545955270528793, "global_step": 33979, "epoch": 943, "lr": 0.001} +{"train_loss": 0.028916126117110252, "train_loss_bc": 0.028913969174027443, "train_loss_llm": 0.00021564960479736328, "grad_norm": 0.11328700184822083, "global_step": 33980, "epoch": 943, "lr": 0.001} +{"train_loss": 0.03227728605270386, "train_loss_bc": 0.03227034956216812, "train_loss_llm": 0.0006936490535736084, "grad_norm": 0.05161767825484276, "global_step": 33981, "epoch": 943, "lr": 0.001} +{"train_loss": 0.0264795683324337, "train_loss_bc": 0.0264795683324337, "train_loss_llm": 0.0, "grad_norm": 0.07064566016197205, "global_step": 33982, "epoch": 943, "lr": 0.001} +{"train_loss": 0.02891329199903541, "train_loss_bc": 0.030928803607821465, "train_loss_llm": 0.0, "grad_norm": 0.10850360244512558, "global_step": 33983, "epoch": 943, "lr": 0.001, "val_loss": 0.1091184914112091} +{"train_loss": 0.03293400630354881, "train_loss_bc": 0.03293400630354881, "train_loss_llm": 0.0, "grad_norm": 0.12474632263183594, "global_step": 33984, "epoch": 944, "lr": 0.001} +{"train_loss": 0.027518611401319504, "train_loss_bc": 0.027518611401319504, "train_loss_llm": 0.0, "grad_norm": 0.040989238768815994, "global_step": 33985, "epoch": 944, "lr": 0.001} +{"train_loss": 0.025629883632063866, "train_loss_bc": 0.02562774159014225, "train_loss_llm": 0.00021415948867797852, "grad_norm": 0.06679482012987137, "global_step": 33986, "epoch": 944, "lr": 0.001} +{"train_loss": 0.03062940388917923, "train_loss_bc": 0.03062940388917923, "train_loss_llm": 0.0, "grad_norm": 0.08847668021917343, "global_step": 33987, "epoch": 944, "lr": 0.001} +{"train_loss": 0.031200390309095383, "train_loss_bc": 0.03119540400803089, "train_loss_llm": 0.0004985630512237549, "grad_norm": 0.09792773425579071, "global_step": 33988, "epoch": 944, "lr": 0.001} +{"train_loss": 0.02927435375750065, "train_loss_bc": 0.02926824800670147, "train_loss_llm": 0.0006105899810791016, "grad_norm": 0.03729310631752014, "global_step": 33989, "epoch": 944, "lr": 0.001} +{"train_loss": 0.028418589383363724, "train_loss_bc": 0.028418589383363724, "train_loss_llm": 0.0, "grad_norm": 0.06379152089357376, "global_step": 33990, "epoch": 944, "lr": 0.001} +{"train_loss": 0.028791140764951706, "train_loss_bc": 0.028791140764951706, "train_loss_llm": 0.0, "grad_norm": 0.08248013257980347, "global_step": 33991, "epoch": 944, "lr": 0.001} +{"train_loss": 0.022475574165582657, "train_loss_bc": 0.022475574165582657, "train_loss_llm": 0.0, "grad_norm": 0.09626201540231705, "global_step": 33992, "epoch": 944, "lr": 0.001} +{"train_loss": 0.025241907685995102, "train_loss_bc": 0.025241907685995102, "train_loss_llm": 0.0, "grad_norm": 0.027955371886491776, "global_step": 33993, "epoch": 944, "lr": 0.001} +{"train_loss": 0.029113922268152237, "train_loss_bc": 0.029113922268152237, "train_loss_llm": 0.0, "grad_norm": 0.04890015348792076, "global_step": 33994, "epoch": 944, "lr": 0.001} +{"train_loss": 0.025682689622044563, "train_loss_bc": 0.025682689622044563, "train_loss_llm": 0.0, "grad_norm": 0.06608754396438599, "global_step": 33995, "epoch": 944, "lr": 0.001} +{"train_loss": 0.028679674491286278, "train_loss_bc": 0.028679674491286278, "train_loss_llm": 0.0, "grad_norm": 0.08075207471847534, "global_step": 33996, "epoch": 944, "lr": 0.001} +{"train_loss": 0.02621929720044136, "train_loss_bc": 0.02621929720044136, "train_loss_llm": 0.0, "grad_norm": 0.03626219183206558, "global_step": 33997, "epoch": 944, "lr": 0.001} +{"train_loss": 0.029454920440912247, "train_loss_bc": 0.029454920440912247, "train_loss_llm": 0.0, "grad_norm": 0.05891219526529312, "global_step": 33998, "epoch": 944, "lr": 0.001} +{"train_loss": 0.030888745561242104, "train_loss_bc": 0.030879179015755653, "train_loss_llm": 0.0009566545486450195, "grad_norm": 0.09257739037275314, "global_step": 33999, "epoch": 944, "lr": 0.001} +{"train_loss": 0.033257175236940384, "train_loss_bc": 0.033257175236940384, "train_loss_llm": 0.0, "grad_norm": 0.1145344227552414, "global_step": 34000, "epoch": 944, "lr": 0.001} +{"train_loss": 0.02731679007411003, "train_loss_bc": 0.02731679007411003, "train_loss_llm": 0.0, "grad_norm": 0.03707386553287506, "global_step": 34001, "epoch": 944, "lr": 0.001} +{"train_loss": 0.03377567604184151, "train_loss_bc": 0.03377567604184151, "train_loss_llm": 0.0, "grad_norm": 0.07625226676464081, "global_step": 34002, "epoch": 944, "lr": 0.001} +{"train_loss": 0.026684973388910294, "train_loss_bc": 0.026684973388910294, "train_loss_llm": 0.0, "grad_norm": 0.10272176563739777, "global_step": 34003, "epoch": 944, "lr": 0.001} +{"train_loss": 0.028826376423239708, "train_loss_bc": 0.028826376423239708, "train_loss_llm": 0.0, "grad_norm": 0.12159434705972672, "global_step": 34004, "epoch": 944, "lr": 0.001} +{"train_loss": 0.03324798494577408, "train_loss_bc": 0.03324798494577408, "train_loss_llm": 0.0, "grad_norm": 0.04855917766690254, "global_step": 34005, "epoch": 944, "lr": 0.001} +{"train_loss": 0.028201481327414513, "train_loss_bc": 0.028201481327414513, "train_loss_llm": 0.0, "grad_norm": 0.05732221156358719, "global_step": 34006, "epoch": 944, "lr": 0.001} +{"train_loss": 0.031112167984247208, "train_loss_bc": 0.031111998483538628, "train_loss_llm": 1.6957521438598633e-05, "grad_norm": 0.07757976651191711, "global_step": 34007, "epoch": 944, "lr": 0.001} +{"train_loss": 0.028230782598257065, "train_loss_bc": 0.028230782598257065, "train_loss_llm": 0.0, "grad_norm": 0.10121060907840729, "global_step": 34008, "epoch": 944, "lr": 0.001} +{"train_loss": 0.023820646107196808, "train_loss_bc": 0.023820646107196808, "train_loss_llm": 0.0, "grad_norm": 0.02977079525589943, "global_step": 34009, "epoch": 944, "lr": 0.001} +{"train_loss": 0.030612695962190628, "train_loss_bc": 0.030612695962190628, "train_loss_llm": 0.0, "grad_norm": 0.045630522072315216, "global_step": 34010, "epoch": 944, "lr": 0.001} +{"train_loss": 0.031938210129737854, "train_loss_bc": 0.031938210129737854, "train_loss_llm": 0.0, "grad_norm": 0.05382131040096283, "global_step": 34011, "epoch": 944, "lr": 0.001} +{"train_loss": 0.02702508494257927, "train_loss_bc": 0.02702508494257927, "train_loss_llm": 0.0, "grad_norm": 0.07577158510684967, "global_step": 34012, "epoch": 944, "lr": 0.001} +{"train_loss": 0.03172430023550987, "train_loss_bc": 0.03172430023550987, "train_loss_llm": 0.0, "grad_norm": 0.036429814994335175, "global_step": 34013, "epoch": 944, "lr": 0.001} +{"train_loss": 0.025832703337073326, "train_loss_bc": 0.025827206671237946, "train_loss_llm": 0.0005497336387634277, "grad_norm": 0.05302944779396057, "global_step": 34014, "epoch": 944, "lr": 0.001} +{"train_loss": 0.028679639101028442, "train_loss_bc": 0.028679639101028442, "train_loss_llm": 0.0, "grad_norm": 0.07165849953889847, "global_step": 34015, "epoch": 944, "lr": 0.001} +{"train_loss": 0.030065057799220085, "train_loss_bc": 0.030065057799220085, "train_loss_llm": 0.0, "grad_norm": 0.08696015179157257, "global_step": 34016, "epoch": 944, "lr": 0.001} +{"train_loss": 0.025047332048416138, "train_loss_bc": 0.025047332048416138, "train_loss_llm": 0.0, "grad_norm": 0.03155423700809479, "global_step": 34017, "epoch": 944, "lr": 0.001} +{"train_loss": 0.030110551044344902, "train_loss_bc": 0.030110551044344902, "train_loss_llm": 0.0, "grad_norm": 0.06129140406847, "global_step": 34018, "epoch": 944, "lr": 0.001} +{"train_loss": 0.028821165207773447, "train_loss_bc": 0.029880112037062645, "train_loss_llm": 0.001909494400024414, "grad_norm": 0.0930250957608223, "global_step": 34019, "epoch": 944, "lr": 0.001, "val_loss": 0.10938569158315659} +{"train_loss": 0.032578129321336746, "train_loss_bc": 0.032578129321336746, "train_loss_llm": 0.0, "grad_norm": 0.11213724315166473, "global_step": 34020, "epoch": 945, "lr": 0.001} +{"train_loss": 0.0300704687833786, "train_loss_bc": 0.0300704687833786, "train_loss_llm": 0.0, "grad_norm": 0.03171123191714287, "global_step": 34021, "epoch": 945, "lr": 0.001} +{"train_loss": 0.0324375182390213, "train_loss_bc": 0.0324375182390213, "train_loss_llm": 0.0, "grad_norm": 0.059052590280771255, "global_step": 34022, "epoch": 945, "lr": 0.001} +{"train_loss": 0.02598194219172001, "train_loss_bc": 0.025957779958844185, "train_loss_llm": 0.002416163682937622, "grad_norm": 0.07740350812673569, "global_step": 34023, "epoch": 945, "lr": 0.001} +{"train_loss": 0.025486191734671593, "train_loss_bc": 0.025486191734671593, "train_loss_llm": 0.0, "grad_norm": 0.09349116683006287, "global_step": 34024, "epoch": 945, "lr": 0.001} +{"train_loss": 0.03078433871269226, "train_loss_bc": 0.03078433871269226, "train_loss_llm": 0.0, "grad_norm": 0.040162164717912674, "global_step": 34025, "epoch": 945, "lr": 0.001} +{"train_loss": 0.028016820549964905, "train_loss_bc": 0.028016820549964905, "train_loss_llm": 0.0, "grad_norm": 0.05795698240399361, "global_step": 34026, "epoch": 945, "lr": 0.001} +{"train_loss": 0.0325976200401783, "train_loss_bc": 0.0325976200401783, "train_loss_llm": 0.0, "grad_norm": 0.08550366759300232, "global_step": 34027, "epoch": 945, "lr": 0.001} +{"train_loss": 0.031119970604777336, "train_loss_bc": 0.031115585938096046, "train_loss_llm": 0.00043842196464538574, "grad_norm": 0.09407293796539307, "global_step": 34028, "epoch": 945, "lr": 0.001} +{"train_loss": 0.031717751175165176, "train_loss_bc": 0.031717751175165176, "train_loss_llm": 0.0, "grad_norm": 0.040855877101421356, "global_step": 34029, "epoch": 945, "lr": 0.001} +{"train_loss": 0.029162734746932983, "train_loss_bc": 0.029162734746932983, "train_loss_llm": 0.0, "grad_norm": 0.04575694724917412, "global_step": 34030, "epoch": 945, "lr": 0.001} +{"train_loss": 0.02333245612680912, "train_loss_bc": 0.02333245612680912, "train_loss_llm": 0.0, "grad_norm": 0.06406071782112122, "global_step": 34031, "epoch": 945, "lr": 0.001} +{"train_loss": 0.026567822322249413, "train_loss_bc": 0.026567822322249413, "train_loss_llm": 0.0, "grad_norm": 0.0846361368894577, "global_step": 34032, "epoch": 945, "lr": 0.001} +{"train_loss": 0.03071737103164196, "train_loss_bc": 0.03071737103164196, "train_loss_llm": 0.0, "grad_norm": 0.038368675857782364, "global_step": 34033, "epoch": 945, "lr": 0.001} +{"train_loss": 0.02931712009012699, "train_loss_bc": 0.02931712009012699, "train_loss_llm": 0.0, "grad_norm": 0.05810100957751274, "global_step": 34034, "epoch": 945, "lr": 0.001} +{"train_loss": 0.028435785323381424, "train_loss_bc": 0.028435785323381424, "train_loss_llm": 0.0, "grad_norm": 0.06771168857812881, "global_step": 34035, "epoch": 945, "lr": 0.001} +{"train_loss": 0.024717897176742554, "train_loss_bc": 0.024717897176742554, "train_loss_llm": 0.0, "grad_norm": 0.08712849020957947, "global_step": 34036, "epoch": 945, "lr": 0.001} +{"train_loss": 0.027566751465201378, "train_loss_bc": 0.027558034285902977, "train_loss_llm": 0.0008717179298400879, "grad_norm": 0.03290703520178795, "global_step": 34037, "epoch": 945, "lr": 0.001} +{"train_loss": 0.024869108572602272, "train_loss_bc": 0.024869108572602272, "train_loss_llm": 0.0, "grad_norm": 0.059664998203516006, "global_step": 34038, "epoch": 945, "lr": 0.001} +{"train_loss": 0.029360173270106316, "train_loss_bc": 0.029360173270106316, "train_loss_llm": 0.0, "grad_norm": 0.08597596734762192, "global_step": 34039, "epoch": 945, "lr": 0.001} +{"train_loss": 0.027609676122665405, "train_loss_bc": 0.027596665546298027, "train_loss_llm": 0.0013011395931243896, "grad_norm": 0.11795788258314133, "global_step": 34040, "epoch": 945, "lr": 0.001} +{"train_loss": 0.032433975487947464, "train_loss_bc": 0.032433975487947464, "train_loss_llm": 0.0, "grad_norm": 0.040397871285676956, "global_step": 34041, "epoch": 945, "lr": 0.001} +{"train_loss": 0.031600724905729294, "train_loss_bc": 0.031600724905729294, "train_loss_llm": 0.0, "grad_norm": 0.05508354306221008, "global_step": 34042, "epoch": 945, "lr": 0.001} +{"train_loss": 0.02856394276022911, "train_loss_bc": 0.02856394276022911, "train_loss_llm": 0.0, "grad_norm": 0.06741531193256378, "global_step": 34043, "epoch": 945, "lr": 0.001} +{"train_loss": 0.03086315281689167, "train_loss_bc": 0.03086225688457489, "train_loss_llm": 8.952617645263672e-05, "grad_norm": 0.08460020273923874, "global_step": 34044, "epoch": 945, "lr": 0.001} +{"train_loss": 0.026615915820002556, "train_loss_bc": 0.026615915820002556, "train_loss_llm": 0.0, "grad_norm": 0.03744935616850853, "global_step": 34045, "epoch": 945, "lr": 0.001} +{"train_loss": 0.02646738849580288, "train_loss_bc": 0.026459163054823875, "train_loss_llm": 0.0008225440979003906, "grad_norm": 0.0651959553360939, "global_step": 34046, "epoch": 945, "lr": 0.001} +{"train_loss": 0.03208435699343681, "train_loss_bc": 0.03208435699343681, "train_loss_llm": 0.0, "grad_norm": 0.08511096239089966, "global_step": 34047, "epoch": 945, "lr": 0.001} +{"train_loss": 0.029593294486403465, "train_loss_bc": 0.029592517763376236, "train_loss_llm": 7.769465446472168e-05, "grad_norm": 0.09832064062356949, "global_step": 34048, "epoch": 945, "lr": 0.001} +{"train_loss": 0.023633206263184547, "train_loss_bc": 0.023633206263184547, "train_loss_llm": 0.0, "grad_norm": 0.030415954068303108, "global_step": 34049, "epoch": 945, "lr": 0.001} +{"train_loss": 0.02812967821955681, "train_loss_bc": 0.02812967821955681, "train_loss_llm": 0.0, "grad_norm": 0.05121626704931259, "global_step": 34050, "epoch": 945, "lr": 0.001} +{"train_loss": 0.028416994959115982, "train_loss_bc": 0.028416994959115982, "train_loss_llm": 0.0, "grad_norm": 0.07316574454307556, "global_step": 34051, "epoch": 945, "lr": 0.001} +{"train_loss": 0.026863086968660355, "train_loss_bc": 0.026863086968660355, "train_loss_llm": 0.0, "grad_norm": 0.08494740724563599, "global_step": 34052, "epoch": 945, "lr": 0.001} +{"train_loss": 0.02848200500011444, "train_loss_bc": 0.02847805805504322, "train_loss_llm": 0.00039464235305786133, "grad_norm": 0.03160550445318222, "global_step": 34053, "epoch": 945, "lr": 0.001} +{"train_loss": 0.028329230844974518, "train_loss_bc": 0.028324082493782043, "train_loss_llm": 0.0005148351192474365, "grad_norm": 0.05570145323872566, "global_step": 34054, "epoch": 945, "lr": 0.001} +{"train_loss": 0.02864938938162393, "train_loss_bc": 0.026853416115045547, "train_loss_llm": 0.0, "grad_norm": 0.08295594155788422, "global_step": 34055, "epoch": 945, "lr": 0.001, "train/cumulative_reward": 49.345483408009315, "train/mean_score": 10.715021798598292, "train/success_rate": 0.3, "test/cumulative_reward": 39.307152235421626, "test/mean_score": 7.404158862670759, "test/success_rate": 0.18, "val_loss": 0.10474741458892822, "train_action_mse_error": 0.03166431933641434} +{"train_loss": 0.02875489369034767, "train_loss_bc": 0.028753474354743958, "train_loss_llm": 0.00014194846153259277, "grad_norm": 0.10862117260694504, "global_step": 34056, "epoch": 946, "lr": 0.001} +{"train_loss": 0.03726287558674812, "train_loss_bc": 0.037262458354234695, "train_loss_llm": 4.1604042053222656e-05, "grad_norm": 0.047528911381959915, "global_step": 34057, "epoch": 946, "lr": 0.001} +{"train_loss": 0.030093315988779068, "train_loss_bc": 0.030093315988779068, "train_loss_llm": 0.0, "grad_norm": 0.06880632042884827, "global_step": 34058, "epoch": 946, "lr": 0.001} +{"train_loss": 0.02891245111823082, "train_loss_bc": 0.028911592438817024, "train_loss_llm": 8.589029312133789e-05, "grad_norm": 0.08292720466852188, "global_step": 34059, "epoch": 946, "lr": 0.001} +{"train_loss": 0.03166791796684265, "train_loss_bc": 0.03166791796684265, "train_loss_llm": 0.0, "grad_norm": 0.11254532635211945, "global_step": 34060, "epoch": 946, "lr": 0.001} +{"train_loss": 0.0253443643450737, "train_loss_bc": 0.02532293274998665, "train_loss_llm": 0.0021431446075439453, "grad_norm": 0.03333187848329544, "global_step": 34061, "epoch": 946, "lr": 0.001} +{"train_loss": 0.02780933864414692, "train_loss_bc": 0.02780933864414692, "train_loss_llm": 0.0, "grad_norm": 0.052970580756664276, "global_step": 34062, "epoch": 946, "lr": 0.001} +{"train_loss": 0.028467390686273575, "train_loss_bc": 0.028467390686273575, "train_loss_llm": 0.0, "grad_norm": 0.07060188055038452, "global_step": 34063, "epoch": 946, "lr": 0.001} +{"train_loss": 0.02937028370797634, "train_loss_bc": 0.02937028370797634, "train_loss_llm": 0.0, "grad_norm": 0.10000266134738922, "global_step": 34064, "epoch": 946, "lr": 0.001} +{"train_loss": 0.031104810535907745, "train_loss_bc": 0.031104810535907745, "train_loss_llm": 0.0, "grad_norm": 0.050115905702114105, "global_step": 34065, "epoch": 946, "lr": 0.001} +{"train_loss": 0.03164687007665634, "train_loss_bc": 0.031646374613046646, "train_loss_llm": 4.9442052841186523e-05, "grad_norm": 0.08217141032218933, "global_step": 34066, "epoch": 946, "lr": 0.001} +{"train_loss": 0.026607351377606392, "train_loss_bc": 0.026607351377606392, "train_loss_llm": 0.0, "grad_norm": 0.11083336174488068, "global_step": 34067, "epoch": 946, "lr": 0.001} +{"train_loss": 0.029386717826128006, "train_loss_bc": 0.029385821893811226, "train_loss_llm": 8.952617645263672e-05, "grad_norm": 0.1292259693145752, "global_step": 34068, "epoch": 946, "lr": 0.001} +{"train_loss": 0.02868916280567646, "train_loss_bc": 0.02868916280567646, "train_loss_llm": 0.0, "grad_norm": 0.04486025497317314, "global_step": 34069, "epoch": 946, "lr": 0.001} +{"train_loss": 0.025731602683663368, "train_loss_bc": 0.025731602683663368, "train_loss_llm": 0.0, "grad_norm": 0.05385609716176987, "global_step": 34070, "epoch": 946, "lr": 0.001} +{"train_loss": 0.0370953269302845, "train_loss_bc": 0.0370953269302845, "train_loss_llm": 0.0, "grad_norm": 0.08383827656507492, "global_step": 34071, "epoch": 946, "lr": 0.001} +{"train_loss": 0.03140202909708023, "train_loss_bc": 0.03140202909708023, "train_loss_llm": 0.0, "grad_norm": 0.11527953296899796, "global_step": 34072, "epoch": 946, "lr": 0.001} +{"train_loss": 0.027272414416074753, "train_loss_bc": 0.0272489245980978, "train_loss_llm": 0.0023490488529205322, "grad_norm": 0.03575579449534416, "global_step": 34073, "epoch": 946, "lr": 0.001} +{"train_loss": 0.02659596875309944, "train_loss_bc": 0.026594998314976692, "train_loss_llm": 9.706616401672363e-05, "grad_norm": 0.05416741967201233, "global_step": 34074, "epoch": 946, "lr": 0.001} +{"train_loss": 0.025363808497786522, "train_loss_bc": 0.025363808497786522, "train_loss_llm": 0.0, "grad_norm": 0.06843017786741257, "global_step": 34075, "epoch": 946, "lr": 0.001} +{"train_loss": 0.024400250986218452, "train_loss_bc": 0.024400250986218452, "train_loss_llm": 0.0, "grad_norm": 0.08427949249744415, "global_step": 34076, "epoch": 946, "lr": 0.001} +{"train_loss": 0.029515733942389488, "train_loss_bc": 0.029515733942389488, "train_loss_llm": 0.0, "grad_norm": 0.037578288465738297, "global_step": 34077, "epoch": 946, "lr": 0.001} +{"train_loss": 0.0324745774269104, "train_loss_bc": 0.0324745774269104, "train_loss_llm": 0.0, "grad_norm": 0.07168646901845932, "global_step": 34078, "epoch": 946, "lr": 0.001} +{"train_loss": 0.02972531132400036, "train_loss_bc": 0.02972531132400036, "train_loss_llm": 0.0, "grad_norm": 0.10079894959926605, "global_step": 34079, "epoch": 946, "lr": 0.001} +{"train_loss": 0.030120419338345528, "train_loss_bc": 0.03010530024766922, "train_loss_llm": 0.001511991024017334, "grad_norm": 0.12677405774593353, "global_step": 34080, "epoch": 946, "lr": 0.001} +{"train_loss": 0.027111342176795006, "train_loss_bc": 0.027096226811408997, "train_loss_llm": 0.0015115141868591309, "grad_norm": 0.03680334985256195, "global_step": 34081, "epoch": 946, "lr": 0.001} +{"train_loss": 0.02606945112347603, "train_loss_bc": 0.02606945112347603, "train_loss_llm": 0.0, "grad_norm": 0.06717842817306519, "global_step": 34082, "epoch": 946, "lr": 0.001} +{"train_loss": 0.02804124727845192, "train_loss_bc": 0.02804124727845192, "train_loss_llm": 0.0, "grad_norm": 0.09922719746828079, "global_step": 34083, "epoch": 946, "lr": 0.001} +{"train_loss": 0.03231751173734665, "train_loss_bc": 0.03231751173734665, "train_loss_llm": 0.0, "grad_norm": 0.1297193318605423, "global_step": 34084, "epoch": 946, "lr": 0.001} +{"train_loss": 0.029958320781588554, "train_loss_bc": 0.029958320781588554, "train_loss_llm": 0.0, "grad_norm": 0.0444796048104763, "global_step": 34085, "epoch": 946, "lr": 0.001} +{"train_loss": 0.025509856641292572, "train_loss_bc": 0.025501638650894165, "train_loss_llm": 0.0008218884468078613, "grad_norm": 0.06746316701173782, "global_step": 34086, "epoch": 946, "lr": 0.001} +{"train_loss": 0.029581038281321526, "train_loss_bc": 0.029581038281321526, "train_loss_llm": 0.0, "grad_norm": 0.08988307416439056, "global_step": 34087, "epoch": 946, "lr": 0.001} +{"train_loss": 0.028485067188739777, "train_loss_bc": 0.028485067188739777, "train_loss_llm": 0.0, "grad_norm": 0.12033634632825851, "global_step": 34088, "epoch": 946, "lr": 0.001} +{"train_loss": 0.02192050591111183, "train_loss_bc": 0.02192050591111183, "train_loss_llm": 0.0, "grad_norm": 0.0342496819794178, "global_step": 34089, "epoch": 946, "lr": 0.001} +{"train_loss": 0.024825088679790497, "train_loss_bc": 0.024825088679790497, "train_loss_llm": 0.0, "grad_norm": 0.060367126017808914, "global_step": 34090, "epoch": 946, "lr": 0.001} +{"train_loss": 0.028973050467256043, "train_loss_bc": 0.03439519926905632, "train_loss_llm": 0.0, "grad_norm": 0.09557053446769714, "global_step": 34091, "epoch": 946, "lr": 0.001, "val_loss": 0.10881351679563522} +{"train_loss": 0.026443008333444595, "train_loss_bc": 0.026443008333444595, "train_loss_llm": 0.0, "grad_norm": 0.11572223156690598, "global_step": 34092, "epoch": 947, "lr": 0.001} +{"train_loss": 0.029764147475361824, "train_loss_bc": 0.029743749648332596, "train_loss_llm": 0.002039790153503418, "grad_norm": 0.0424017608165741, "global_step": 34093, "epoch": 947, "lr": 0.001} +{"train_loss": 0.03217788413167, "train_loss_bc": 0.03217788413167, "train_loss_llm": 0.0, "grad_norm": 0.06153932586312294, "global_step": 34094, "epoch": 947, "lr": 0.001} +{"train_loss": 0.03265731781721115, "train_loss_bc": 0.03265731781721115, "train_loss_llm": 0.0, "grad_norm": 0.09334012120962143, "global_step": 34095, "epoch": 947, "lr": 0.001} +{"train_loss": 0.035111721605062485, "train_loss_bc": 0.03510212153196335, "train_loss_llm": 0.000959932804107666, "grad_norm": 0.10006670653820038, "global_step": 34096, "epoch": 947, "lr": 0.001} +{"train_loss": 0.028079748153686523, "train_loss_bc": 0.028079748153686523, "train_loss_llm": 0.0, "grad_norm": 0.048914942890405655, "global_step": 34097, "epoch": 947, "lr": 0.001} +{"train_loss": 0.02853257767856121, "train_loss_bc": 0.02852797321975231, "train_loss_llm": 0.0004604756832122803, "grad_norm": 0.0652410238981247, "global_step": 34098, "epoch": 947, "lr": 0.001} +{"train_loss": 0.031225256621837616, "train_loss_bc": 0.031225256621837616, "train_loss_llm": 0.0, "grad_norm": 0.09674085676670074, "global_step": 34099, "epoch": 947, "lr": 0.001} +{"train_loss": 0.027273904532194138, "train_loss_bc": 0.027257312089204788, "train_loss_llm": 0.001659303903579712, "grad_norm": 0.12989750504493713, "global_step": 34100, "epoch": 947, "lr": 0.001} +{"train_loss": 0.026468763127923012, "train_loss_bc": 0.026468763127923012, "train_loss_llm": 0.0, "grad_norm": 0.03911115601658821, "global_step": 34101, "epoch": 947, "lr": 0.001} +{"train_loss": 0.030352197587490082, "train_loss_bc": 0.030352197587490082, "train_loss_llm": 0.0, "grad_norm": 0.053912632167339325, "global_step": 34102, "epoch": 947, "lr": 0.001} +{"train_loss": 0.03045671619474888, "train_loss_bc": 0.03045671619474888, "train_loss_llm": 0.0, "grad_norm": 0.07304834574460983, "global_step": 34103, "epoch": 947, "lr": 0.001} +{"train_loss": 0.02671530656516552, "train_loss_bc": 0.02671530656516552, "train_loss_llm": 0.0, "grad_norm": 0.0841149091720581, "global_step": 34104, "epoch": 947, "lr": 0.001} +{"train_loss": 0.030592244118452072, "train_loss_bc": 0.030592244118452072, "train_loss_llm": 0.0, "grad_norm": 0.029682781547307968, "global_step": 34105, "epoch": 947, "lr": 0.001} +{"train_loss": 0.02535931020975113, "train_loss_bc": 0.02535931020975113, "train_loss_llm": 0.0, "grad_norm": 0.04515179246664047, "global_step": 34106, "epoch": 947, "lr": 0.001} +{"train_loss": 0.03038167767226696, "train_loss_bc": 0.03038167767226696, "train_loss_llm": 0.0, "grad_norm": 0.06837613880634308, "global_step": 34107, "epoch": 947, "lr": 0.001} +{"train_loss": 0.03115100786089897, "train_loss_bc": 0.031137794256210327, "train_loss_llm": 0.0013213753700256348, "grad_norm": 0.08961133658885956, "global_step": 34108, "epoch": 947, "lr": 0.001} +{"train_loss": 0.025327859446406364, "train_loss_bc": 0.025327859446406364, "train_loss_llm": 0.0, "grad_norm": 0.035586271435022354, "global_step": 34109, "epoch": 947, "lr": 0.001} +{"train_loss": 0.02935662306845188, "train_loss_bc": 0.02935662306845188, "train_loss_llm": 0.0, "grad_norm": 0.05884939059615135, "global_step": 34110, "epoch": 947, "lr": 0.001} +{"train_loss": 0.029039138928055763, "train_loss_bc": 0.02903272770345211, "train_loss_llm": 0.0006411373615264893, "grad_norm": 0.09296310693025589, "global_step": 34111, "epoch": 947, "lr": 0.001} +{"train_loss": 0.032749682664871216, "train_loss_bc": 0.032749682664871216, "train_loss_llm": 0.0, "grad_norm": 0.11682391166687012, "global_step": 34112, "epoch": 947, "lr": 0.001} +{"train_loss": 0.028379714116454124, "train_loss_bc": 0.028375577181577682, "train_loss_llm": 0.00041371583938598633, "grad_norm": 0.03726038709282875, "global_step": 34113, "epoch": 947, "lr": 0.001} +{"train_loss": 0.03127732500433922, "train_loss_bc": 0.03127732500433922, "train_loss_llm": 0.0, "grad_norm": 0.0669017806649208, "global_step": 34114, "epoch": 947, "lr": 0.001} +{"train_loss": 0.029154572635889053, "train_loss_bc": 0.029154572635889053, "train_loss_llm": 0.0, "grad_norm": 0.09213917702436447, "global_step": 34115, "epoch": 947, "lr": 0.001} +{"train_loss": 0.02791658043861389, "train_loss_bc": 0.02791658043861389, "train_loss_llm": 0.0, "grad_norm": 0.12024800479412079, "global_step": 34116, "epoch": 947, "lr": 0.001} +{"train_loss": 0.03140129894018173, "train_loss_bc": 0.03140129894018173, "train_loss_llm": 0.0, "grad_norm": 0.043110575526952744, "global_step": 34117, "epoch": 947, "lr": 0.001} +{"train_loss": 0.02962525375187397, "train_loss_bc": 0.0296182781457901, "train_loss_llm": 0.0006974935531616211, "grad_norm": 0.07242017239332199, "global_step": 34118, "epoch": 947, "lr": 0.001} +{"train_loss": 0.026468520984053612, "train_loss_bc": 0.026468520984053612, "train_loss_llm": 0.0, "grad_norm": 0.08892331272363663, "global_step": 34119, "epoch": 947, "lr": 0.001} +{"train_loss": 0.023379657417535782, "train_loss_bc": 0.023379657417535782, "train_loss_llm": 0.0, "grad_norm": 0.09833370894193649, "global_step": 34120, "epoch": 947, "lr": 0.001} +{"train_loss": 0.03453996405005455, "train_loss_bc": 0.03453996405005455, "train_loss_llm": 0.0, "grad_norm": 0.03760935366153717, "global_step": 34121, "epoch": 947, "lr": 0.001} +{"train_loss": 0.025727026164531708, "train_loss_bc": 0.02572546899318695, "train_loss_llm": 0.000155717134475708, "grad_norm": 0.057273007929325104, "global_step": 34122, "epoch": 947, "lr": 0.001} +{"train_loss": 0.019315404817461967, "train_loss_bc": 0.019312111660838127, "train_loss_llm": 0.0003293156623840332, "grad_norm": 0.06932983547449112, "global_step": 34123, "epoch": 947, "lr": 0.001} +{"train_loss": 0.02341477759182453, "train_loss_bc": 0.02341477759182453, "train_loss_llm": 0.0, "grad_norm": 0.07965672016143799, "global_step": 34124, "epoch": 947, "lr": 0.001} +{"train_loss": 0.0351291261613369, "train_loss_bc": 0.0351291261613369, "train_loss_llm": 0.0, "grad_norm": 0.03786993399262428, "global_step": 34125, "epoch": 947, "lr": 0.001} +{"train_loss": 0.027411306276917458, "train_loss_bc": 0.027388988062739372, "train_loss_llm": 0.0022318363189697266, "grad_norm": 0.05439724400639534, "global_step": 34126, "epoch": 947, "lr": 0.001} +{"train_loss": 0.028906228486448526, "train_loss_bc": 0.028267603367567062, "train_loss_llm": 0.0, "grad_norm": 0.0854940265417099, "global_step": 34127, "epoch": 947, "lr": 0.001, "val_loss": 0.10690423846244812} +{"train_loss": 0.03161503002047539, "train_loss_bc": 0.03161503002047539, "train_loss_llm": 0.0, "grad_norm": 0.09783942997455597, "global_step": 34128, "epoch": 948, "lr": 0.001} +{"train_loss": 0.024183101952075958, "train_loss_bc": 0.024178678169846535, "train_loss_llm": 0.0004424452781677246, "grad_norm": 0.034946661442518234, "global_step": 34129, "epoch": 948, "lr": 0.001} +{"train_loss": 0.03554946929216385, "train_loss_bc": 0.03554946929216385, "train_loss_llm": 0.0, "grad_norm": 0.06265077739953995, "global_step": 34130, "epoch": 948, "lr": 0.001} +{"train_loss": 0.02915355935692787, "train_loss_bc": 0.029147153720259666, "train_loss_llm": 0.00064048171043396, "grad_norm": 0.08314772695302963, "global_step": 34131, "epoch": 948, "lr": 0.001} +{"train_loss": 0.0312880203127861, "train_loss_bc": 0.0312880203127861, "train_loss_llm": 0.0, "grad_norm": 0.09586282819509506, "global_step": 34132, "epoch": 948, "lr": 0.001} +{"train_loss": 0.028823591768741608, "train_loss_bc": 0.028823591768741608, "train_loss_llm": 0.0, "grad_norm": 0.03641078993678093, "global_step": 34133, "epoch": 948, "lr": 0.001} +{"train_loss": 0.028669828549027443, "train_loss_bc": 0.028669828549027443, "train_loss_llm": 0.0, "grad_norm": 0.055400557816028595, "global_step": 34134, "epoch": 948, "lr": 0.001} +{"train_loss": 0.027784552425146103, "train_loss_bc": 0.027784552425146103, "train_loss_llm": 0.0, "grad_norm": 0.07223422080278397, "global_step": 34135, "epoch": 948, "lr": 0.001} +{"train_loss": 0.030232958495616913, "train_loss_bc": 0.030232958495616913, "train_loss_llm": 0.0, "grad_norm": 0.08830543607473373, "global_step": 34136, "epoch": 948, "lr": 0.001} +{"train_loss": 0.026608576998114586, "train_loss_bc": 0.026608576998114586, "train_loss_llm": 0.0, "grad_norm": 0.03955623507499695, "global_step": 34137, "epoch": 948, "lr": 0.001} +{"train_loss": 0.02498004585504532, "train_loss_bc": 0.02498004585504532, "train_loss_llm": 0.0, "grad_norm": 0.07388273626565933, "global_step": 34138, "epoch": 948, "lr": 0.001} +{"train_loss": 0.028133658692240715, "train_loss_bc": 0.028133658692240715, "train_loss_llm": 0.0, "grad_norm": 0.11435665935277939, "global_step": 34139, "epoch": 948, "lr": 0.001} +{"train_loss": 0.02988266758620739, "train_loss_bc": 0.02988266758620739, "train_loss_llm": 0.0, "grad_norm": 0.14345671236515045, "global_step": 34140, "epoch": 948, "lr": 0.001} +{"train_loss": 0.033636145293712616, "train_loss_bc": 0.03361963853240013, "train_loss_llm": 0.0016506612300872803, "grad_norm": 0.03595368564128876, "global_step": 34141, "epoch": 948, "lr": 0.001} +{"train_loss": 0.029672881588339806, "train_loss_bc": 0.029672881588339806, "train_loss_llm": 0.0, "grad_norm": 0.05783030763268471, "global_step": 34142, "epoch": 948, "lr": 0.001} +{"train_loss": 0.027570098638534546, "train_loss_bc": 0.027555176988244057, "train_loss_llm": 0.0014921128749847412, "grad_norm": 0.08718404173851013, "global_step": 34143, "epoch": 948, "lr": 0.001} +{"train_loss": 0.02771769091486931, "train_loss_bc": 0.02771177887916565, "train_loss_llm": 0.0005911588668823242, "grad_norm": 0.11901793628931046, "global_step": 34144, "epoch": 948, "lr": 0.001} +{"train_loss": 0.025433044880628586, "train_loss_bc": 0.025433044880628586, "train_loss_llm": 0.0, "grad_norm": 0.034556794911623, "global_step": 34145, "epoch": 948, "lr": 0.001} +{"train_loss": 0.03195185214281082, "train_loss_bc": 0.03195185214281082, "train_loss_llm": 0.0, "grad_norm": 0.05500837042927742, "global_step": 34146, "epoch": 948, "lr": 0.001} +{"train_loss": 0.02339324913918972, "train_loss_bc": 0.02339324913918972, "train_loss_llm": 0.0, "grad_norm": 0.08002879470586777, "global_step": 34147, "epoch": 948, "lr": 0.001} +{"train_loss": 0.02879457175731659, "train_loss_bc": 0.02879457175731659, "train_loss_llm": 0.0, "grad_norm": 0.10355307161808014, "global_step": 34148, "epoch": 948, "lr": 0.001} +{"train_loss": 0.025749368593096733, "train_loss_bc": 0.025749368593096733, "train_loss_llm": 0.0, "grad_norm": 0.027154112234711647, "global_step": 34149, "epoch": 948, "lr": 0.001} +{"train_loss": 0.02571236714720726, "train_loss_bc": 0.02571236714720726, "train_loss_llm": 0.0, "grad_norm": 0.054168760776519775, "global_step": 34150, "epoch": 948, "lr": 0.001} +{"train_loss": 0.02965100295841694, "train_loss_bc": 0.02965100295841694, "train_loss_llm": 0.0, "grad_norm": 0.07423758506774902, "global_step": 34151, "epoch": 948, "lr": 0.001} +{"train_loss": 0.0249983798712492, "train_loss_bc": 0.0249983798712492, "train_loss_llm": 0.0, "grad_norm": 0.0898333489894867, "global_step": 34152, "epoch": 948, "lr": 0.001} +{"train_loss": 0.0352737195789814, "train_loss_bc": 0.0352737195789814, "train_loss_llm": 0.0, "grad_norm": 0.08120882511138916, "global_step": 34153, "epoch": 948, "lr": 0.001} +{"train_loss": 0.02900729328393936, "train_loss_bc": 0.02900729328393936, "train_loss_llm": 0.0, "grad_norm": 0.10595842450857162, "global_step": 34154, "epoch": 948, "lr": 0.001} +{"train_loss": 0.025331709533929825, "train_loss_bc": 0.025331709533929825, "train_loss_llm": 0.0, "grad_norm": 0.13871058821678162, "global_step": 34155, "epoch": 948, "lr": 0.001} +{"train_loss": 0.029428379610180855, "train_loss_bc": 0.029428379610180855, "train_loss_llm": 0.0, "grad_norm": 0.16751162707805634, "global_step": 34156, "epoch": 948, "lr": 0.001} +{"train_loss": 0.028086209669709206, "train_loss_bc": 0.028086209669709206, "train_loss_llm": 0.0, "grad_norm": 0.044566649943590164, "global_step": 34157, "epoch": 948, "lr": 0.001} +{"train_loss": 0.028938649222254753, "train_loss_bc": 0.028938649222254753, "train_loss_llm": 0.0, "grad_norm": 0.07655905187129974, "global_step": 34158, "epoch": 948, "lr": 0.001} +{"train_loss": 0.02933315560221672, "train_loss_bc": 0.02933315560221672, "train_loss_llm": 0.0, "grad_norm": 0.09313607960939407, "global_step": 34159, "epoch": 948, "lr": 0.001} +{"train_loss": 0.027002839371562004, "train_loss_bc": 0.027002839371562004, "train_loss_llm": 0.0, "grad_norm": 0.11899501830339432, "global_step": 34160, "epoch": 948, "lr": 0.001} +{"train_loss": 0.03394493833184242, "train_loss_bc": 0.03394493833184242, "train_loss_llm": 0.0, "grad_norm": 0.038162969052791595, "global_step": 34161, "epoch": 948, "lr": 0.001} +{"train_loss": 0.033585324883461, "train_loss_bc": 0.03357701376080513, "train_loss_llm": 0.0008310079574584961, "grad_norm": 0.07601740211248398, "global_step": 34162, "epoch": 948, "lr": 0.001} +{"train_loss": 0.028867607781042654, "train_loss_bc": 0.028115946799516678, "train_loss_llm": 0.0, "grad_norm": 0.09257897734642029, "global_step": 34163, "epoch": 948, "lr": 0.001, "val_loss": 0.10708291083574295} +{"train_loss": 0.029247749596834183, "train_loss_bc": 0.029247749596834183, "train_loss_llm": 0.0, "grad_norm": 0.12584030628204346, "global_step": 34164, "epoch": 949, "lr": 0.001} +{"train_loss": 0.02526063285768032, "train_loss_bc": 0.02526063285768032, "train_loss_llm": 0.0, "grad_norm": 0.033866655081510544, "global_step": 34165, "epoch": 949, "lr": 0.001} +{"train_loss": 0.027722816914319992, "train_loss_bc": 0.027722816914319992, "train_loss_llm": 0.0, "grad_norm": 0.05826915055513382, "global_step": 34166, "epoch": 949, "lr": 0.001} +{"train_loss": 0.025817278772592545, "train_loss_bc": 0.025812430307269096, "train_loss_llm": 0.00048485398292541504, "grad_norm": 0.06836114078760147, "global_step": 34167, "epoch": 949, "lr": 0.001} +{"train_loss": 0.0334407277405262, "train_loss_bc": 0.0334407277405262, "train_loss_llm": 0.0, "grad_norm": 0.08202085644006729, "global_step": 34168, "epoch": 949, "lr": 0.001} +{"train_loss": 0.02961886115372181, "train_loss_bc": 0.029600579291582108, "train_loss_llm": 0.0018281638622283936, "grad_norm": 0.05271267890930176, "global_step": 34169, "epoch": 949, "lr": 0.001} +{"train_loss": 0.0264347642660141, "train_loss_bc": 0.0264347642660141, "train_loss_llm": 0.0, "grad_norm": 0.08574818074703217, "global_step": 34170, "epoch": 949, "lr": 0.001} +{"train_loss": 0.027983106672763824, "train_loss_bc": 0.027983106672763824, "train_loss_llm": 0.0, "grad_norm": 0.12441960722208023, "global_step": 34171, "epoch": 949, "lr": 0.001} +{"train_loss": 0.028053944930434227, "train_loss_bc": 0.028053944930434227, "train_loss_llm": 0.0, "grad_norm": 0.1515069305896759, "global_step": 34172, "epoch": 949, "lr": 0.001} +{"train_loss": 0.02849404886364937, "train_loss_bc": 0.028489623218774796, "train_loss_llm": 0.0004425644874572754, "grad_norm": 0.04905803129076958, "global_step": 34173, "epoch": 949, "lr": 0.001} +{"train_loss": 0.025853168219327927, "train_loss_bc": 0.025853168219327927, "train_loss_llm": 0.0, "grad_norm": 0.09026629477739334, "global_step": 34174, "epoch": 949, "lr": 0.001} +{"train_loss": 0.027857035398483276, "train_loss_bc": 0.027857035398483276, "train_loss_llm": 0.0, "grad_norm": 0.11396344006061554, "global_step": 34175, "epoch": 949, "lr": 0.001} +{"train_loss": 0.03137703984975815, "train_loss_bc": 0.03136877343058586, "train_loss_llm": 0.0008265078067779541, "grad_norm": 0.16869805753231049, "global_step": 34176, "epoch": 949, "lr": 0.001} +{"train_loss": 0.030049782246351242, "train_loss_bc": 0.030049782246351242, "train_loss_llm": 0.0, "grad_norm": 0.041649531573057175, "global_step": 34177, "epoch": 949, "lr": 0.001} +{"train_loss": 0.026809943839907646, "train_loss_bc": 0.026808420196175575, "train_loss_llm": 0.00015237927436828613, "grad_norm": 0.06867583096027374, "global_step": 34178, "epoch": 949, "lr": 0.001} +{"train_loss": 0.026259057223796844, "train_loss_bc": 0.02625630795955658, "train_loss_llm": 0.00027492642402648926, "grad_norm": 0.10321588814258575, "global_step": 34179, "epoch": 949, "lr": 0.001} +{"train_loss": 0.03135758265852928, "train_loss_bc": 0.03135758265852928, "train_loss_llm": 0.0, "grad_norm": 0.14615322649478912, "global_step": 34180, "epoch": 949, "lr": 0.001} +{"train_loss": 0.027560362592339516, "train_loss_bc": 0.027554452419281006, "train_loss_llm": 0.0005910396575927734, "grad_norm": 0.052446480840444565, "global_step": 34181, "epoch": 949, "lr": 0.001} +{"train_loss": 0.036130696535110474, "train_loss_bc": 0.03611169010400772, "train_loss_llm": 0.0019005835056304932, "grad_norm": 0.07689133286476135, "global_step": 34182, "epoch": 949, "lr": 0.001} +{"train_loss": 0.0383179672062397, "train_loss_bc": 0.0383179672062397, "train_loss_llm": 0.0, "grad_norm": 0.09878642857074738, "global_step": 34183, "epoch": 949, "lr": 0.001} +{"train_loss": 0.027020316570997238, "train_loss_bc": 0.027020316570997238, "train_loss_llm": 0.0, "grad_norm": 0.123297318816185, "global_step": 34184, "epoch": 949, "lr": 0.001} +{"train_loss": 0.02512333169579506, "train_loss_bc": 0.02512281946837902, "train_loss_llm": 5.128979682922363e-05, "grad_norm": 0.04094770550727844, "global_step": 34185, "epoch": 949, "lr": 0.001} +{"train_loss": 0.030825918540358543, "train_loss_bc": 0.030825918540358543, "train_loss_llm": 0.0, "grad_norm": 0.06347808986902237, "global_step": 34186, "epoch": 949, "lr": 0.001} +{"train_loss": 0.02914990298449993, "train_loss_bc": 0.02914990298449993, "train_loss_llm": 0.0, "grad_norm": 0.09519761800765991, "global_step": 34187, "epoch": 949, "lr": 0.001} +{"train_loss": 0.031773462891578674, "train_loss_bc": 0.031773462891578674, "train_loss_llm": 0.0, "grad_norm": 0.14275072515010834, "global_step": 34188, "epoch": 949, "lr": 0.001} +{"train_loss": 0.02570842206478119, "train_loss_bc": 0.02570842206478119, "train_loss_llm": 0.0, "grad_norm": 0.05018338933587074, "global_step": 34189, "epoch": 949, "lr": 0.001} +{"train_loss": 0.02896251529455185, "train_loss_bc": 0.02895878255367279, "train_loss_llm": 0.0003733634948730469, "grad_norm": 0.09535244852304459, "global_step": 34190, "epoch": 949, "lr": 0.001} +{"train_loss": 0.023853998631238937, "train_loss_bc": 0.023853998631238937, "train_loss_llm": 0.0, "grad_norm": 0.12643414735794067, "global_step": 34191, "epoch": 949, "lr": 0.001} +{"train_loss": 0.03430606797337532, "train_loss_bc": 0.03430606797337532, "train_loss_llm": 0.0, "grad_norm": 0.1586209535598755, "global_step": 34192, "epoch": 949, "lr": 0.001} +{"train_loss": 0.03184710815548897, "train_loss_bc": 0.03184710815548897, "train_loss_llm": 0.0, "grad_norm": 0.05785606801509857, "global_step": 34193, "epoch": 949, "lr": 0.001} +{"train_loss": 0.02954130806028843, "train_loss_bc": 0.02954130806028843, "train_loss_llm": 0.0, "grad_norm": 0.09549156576395035, "global_step": 34194, "epoch": 949, "lr": 0.001} +{"train_loss": 0.030286405235528946, "train_loss_bc": 0.030286405235528946, "train_loss_llm": 0.0, "grad_norm": 0.10879433155059814, "global_step": 34195, "epoch": 949, "lr": 0.001} +{"train_loss": 0.02889651246368885, "train_loss_bc": 0.02889651246368885, "train_loss_llm": 0.0, "grad_norm": 0.1343696266412735, "global_step": 34196, "epoch": 949, "lr": 0.001} +{"train_loss": 0.02927798591554165, "train_loss_bc": 0.02927798591554165, "train_loss_llm": 0.0, "grad_norm": 0.04540199041366577, "global_step": 34197, "epoch": 949, "lr": 0.001} +{"train_loss": 0.029834086075425148, "train_loss_bc": 0.029834086075425148, "train_loss_llm": 0.0, "grad_norm": 0.0744122713804245, "global_step": 34198, "epoch": 949, "lr": 0.001} +{"train_loss": 0.029161820602085855, "train_loss_bc": 0.029756508767604828, "train_loss_llm": 0.0015122294425964355, "grad_norm": 0.09956801682710648, "global_step": 34199, "epoch": 949, "lr": 0.001, "val_loss": 0.10683003067970276} +{"train_loss": 0.028426049277186394, "train_loss_bc": 0.028426049277186394, "train_loss_llm": 0.0, "grad_norm": 0.1253247708082199, "global_step": 34200, "epoch": 950, "lr": 0.001} +{"train_loss": 0.030355822294950485, "train_loss_bc": 0.030355822294950485, "train_loss_llm": 0.0, "grad_norm": 0.04344656318426132, "global_step": 34201, "epoch": 950, "lr": 0.001} +{"train_loss": 0.03021235018968582, "train_loss_bc": 0.03021235018968582, "train_loss_llm": 0.0, "grad_norm": 0.06896433979272842, "global_step": 34202, "epoch": 950, "lr": 0.001} +{"train_loss": 0.021180016919970512, "train_loss_bc": 0.021169155836105347, "train_loss_llm": 0.0010860264301300049, "grad_norm": 0.08236744999885559, "global_step": 34203, "epoch": 950, "lr": 0.001} +{"train_loss": 0.0320076160132885, "train_loss_bc": 0.0320076160132885, "train_loss_llm": 0.0, "grad_norm": 0.11856775730848312, "global_step": 34204, "epoch": 950, "lr": 0.001} +{"train_loss": 0.026680897921323776, "train_loss_bc": 0.026660919189453125, "train_loss_llm": 0.0019979476928710938, "grad_norm": 0.03486282005906105, "global_step": 34205, "epoch": 950, "lr": 0.001} +{"train_loss": 0.02655566670000553, "train_loss_bc": 0.02655566670000553, "train_loss_llm": 0.0, "grad_norm": 0.05366579815745354, "global_step": 34206, "epoch": 950, "lr": 0.001} +{"train_loss": 0.029197391122579575, "train_loss_bc": 0.029194720089435577, "train_loss_llm": 0.0002671182155609131, "grad_norm": 0.0783485695719719, "global_step": 34207, "epoch": 950, "lr": 0.001} +{"train_loss": 0.03118707239627838, "train_loss_bc": 0.03118707239627838, "train_loss_llm": 0.0, "grad_norm": 0.11768925935029984, "global_step": 34208, "epoch": 950, "lr": 0.001} +{"train_loss": 0.03112276829779148, "train_loss_bc": 0.03111540712416172, "train_loss_llm": 0.0007360577583312988, "grad_norm": 0.049956001341342926, "global_step": 34209, "epoch": 950, "lr": 0.001} +{"train_loss": 0.03259844332933426, "train_loss_bc": 0.03259844332933426, "train_loss_llm": 0.0, "grad_norm": 0.09016342461109161, "global_step": 34210, "epoch": 950, "lr": 0.001} +{"train_loss": 0.026884589344263077, "train_loss_bc": 0.026884589344263077, "train_loss_llm": 0.0, "grad_norm": 0.1228269636631012, "global_step": 34211, "epoch": 950, "lr": 0.001} +{"train_loss": 0.023317845538258553, "train_loss_bc": 0.023317845538258553, "train_loss_llm": 0.0, "grad_norm": 0.1471187174320221, "global_step": 34212, "epoch": 950, "lr": 0.001} +{"train_loss": 0.024172121658921242, "train_loss_bc": 0.02417045086622238, "train_loss_llm": 0.00016710162162780762, "grad_norm": 0.039914850145578384, "global_step": 34213, "epoch": 950, "lr": 0.001} +{"train_loss": 0.03266613185405731, "train_loss_bc": 0.03266613185405731, "train_loss_llm": 0.0, "grad_norm": 0.06393764913082123, "global_step": 34214, "epoch": 950, "lr": 0.001} +{"train_loss": 0.025844303891062737, "train_loss_bc": 0.025844303891062737, "train_loss_llm": 0.0, "grad_norm": 0.09227260947227478, "global_step": 34215, "epoch": 950, "lr": 0.001} +{"train_loss": 0.029308468103408813, "train_loss_bc": 0.029289882630109787, "train_loss_llm": 0.0018585622310638428, "grad_norm": 0.12425656616687775, "global_step": 34216, "epoch": 950, "lr": 0.001} +{"train_loss": 0.028154000639915466, "train_loss_bc": 0.028154000639915466, "train_loss_llm": 0.0, "grad_norm": 0.0356503427028656, "global_step": 34217, "epoch": 950, "lr": 0.001} +{"train_loss": 0.026019232347607613, "train_loss_bc": 0.026019232347607613, "train_loss_llm": 0.0, "grad_norm": 0.05253208801150322, "global_step": 34218, "epoch": 950, "lr": 0.001} +{"train_loss": 0.030171066522598267, "train_loss_bc": 0.030171066522598267, "train_loss_llm": 0.0, "grad_norm": 0.06295739114284515, "global_step": 34219, "epoch": 950, "lr": 0.001} +{"train_loss": 0.02755868434906006, "train_loss_bc": 0.027553949505090714, "train_loss_llm": 0.00047340989112854004, "grad_norm": 0.07028646767139435, "global_step": 34220, "epoch": 950, "lr": 0.001} +{"train_loss": 0.029808910563588142, "train_loss_bc": 0.029808245599269867, "train_loss_llm": 6.657838821411133e-05, "grad_norm": 0.04613787680864334, "global_step": 34221, "epoch": 950, "lr": 0.001} +{"train_loss": 0.027618268504738808, "train_loss_bc": 0.027618268504738808, "train_loss_llm": 0.0, "grad_norm": 0.0718594565987587, "global_step": 34222, "epoch": 950, "lr": 0.001} +{"train_loss": 0.02638065442442894, "train_loss_bc": 0.02638065442442894, "train_loss_llm": 0.0, "grad_norm": 0.08795463293790817, "global_step": 34223, "epoch": 950, "lr": 0.001} +{"train_loss": 0.030404450371861458, "train_loss_bc": 0.030404450371861458, "train_loss_llm": 0.0, "grad_norm": 0.10961019992828369, "global_step": 34224, "epoch": 950, "lr": 0.001} +{"train_loss": 0.02680296264588833, "train_loss_bc": 0.02680296264588833, "train_loss_llm": 0.0, "grad_norm": 0.03693193569779396, "global_step": 34225, "epoch": 950, "lr": 0.001} +{"train_loss": 0.029849626123905182, "train_loss_bc": 0.029849626123905182, "train_loss_llm": 0.0, "grad_norm": 0.06535326689481735, "global_step": 34226, "epoch": 950, "lr": 0.001} +{"train_loss": 0.0325179249048233, "train_loss_bc": 0.03249497711658478, "train_loss_llm": 0.002294868230819702, "grad_norm": 0.08302896469831467, "global_step": 34227, "epoch": 950, "lr": 0.001} +{"train_loss": 0.035635121166706085, "train_loss_bc": 0.035624854266643524, "train_loss_llm": 0.0010268688201904297, "grad_norm": 0.11737983673810959, "global_step": 34228, "epoch": 950, "lr": 0.001} +{"train_loss": 0.03462228924036026, "train_loss_bc": 0.03462228924036026, "train_loss_llm": 0.0, "grad_norm": 0.04234917089343071, "global_step": 34229, "epoch": 950, "lr": 0.001} +{"train_loss": 0.027295630425214767, "train_loss_bc": 0.027295630425214767, "train_loss_llm": 0.0, "grad_norm": 0.06834393739700317, "global_step": 34230, "epoch": 950, "lr": 0.001} +{"train_loss": 0.02914360910654068, "train_loss_bc": 0.02914360910654068, "train_loss_llm": 0.0, "grad_norm": 0.09907668083906174, "global_step": 34231, "epoch": 950, "lr": 0.001} +{"train_loss": 0.028407275676727295, "train_loss_bc": 0.028407275676727295, "train_loss_llm": 0.0, "grad_norm": 0.11086080223321915, "global_step": 34232, "epoch": 950, "lr": 0.001} +{"train_loss": 0.026421809569001198, "train_loss_bc": 0.026414792984724045, "train_loss_llm": 0.0007016956806182861, "grad_norm": 0.04133507236838341, "global_step": 34233, "epoch": 950, "lr": 0.001} +{"train_loss": 0.03479161486029625, "train_loss_bc": 0.03479161486029625, "train_loss_llm": 0.0, "grad_norm": 0.06162741780281067, "global_step": 34234, "epoch": 950, "lr": 0.001} +{"train_loss": 0.028767097430924576, "train_loss_bc": 0.022294821217656136, "train_loss_llm": 0.0, "grad_norm": 0.0765635222196579, "global_step": 34235, "epoch": 950, "lr": 0.001, "train/cumulative_reward": 59.468943292586026, "train/mean_score": 16.819244298286048, "train/success_rate": 0.4, "test/cumulative_reward": 34.88704023237121, "test/mean_score": 7.9944681453022755, "test/success_rate": 0.2, "val_loss": 0.10872706770896912, "train_action_mse_error": 0.03174879029393196} +{"train_loss": 0.025805246084928513, "train_loss_bc": 0.02578408271074295, "train_loss_llm": 0.0021163225173950195, "grad_norm": 0.08953874558210373, "global_step": 34236, "epoch": 951, "lr": 0.001} +{"train_loss": 0.03511165827512741, "train_loss_bc": 0.03511165827512741, "train_loss_llm": 0.0, "grad_norm": 0.03490885719656944, "global_step": 34237, "epoch": 951, "lr": 0.001} +{"train_loss": 0.029041249305009842, "train_loss_bc": 0.02903548628091812, "train_loss_llm": 0.0005762279033660889, "grad_norm": 0.05402800813317299, "global_step": 34238, "epoch": 951, "lr": 0.001} +{"train_loss": 0.030505623668432236, "train_loss_bc": 0.03049519658088684, "train_loss_llm": 0.0010426342487335205, "grad_norm": 0.0730808824300766, "global_step": 34239, "epoch": 951, "lr": 0.001} +{"train_loss": 0.03421816974878311, "train_loss_bc": 0.03421816974878311, "train_loss_llm": 0.0, "grad_norm": 0.08598688989877701, "global_step": 34240, "epoch": 951, "lr": 0.001} +{"train_loss": 0.02491200715303421, "train_loss_bc": 0.024908334016799927, "train_loss_llm": 0.00036722421646118164, "grad_norm": 0.031200623139739037, "global_step": 34241, "epoch": 951, "lr": 0.001} +{"train_loss": 0.03200418874621391, "train_loss_bc": 0.03200418874621391, "train_loss_llm": 0.0, "grad_norm": 0.053628988564014435, "global_step": 34242, "epoch": 951, "lr": 0.001} +{"train_loss": 0.03132742643356323, "train_loss_bc": 0.03132742643356323, "train_loss_llm": 0.0, "grad_norm": 0.07630278915166855, "global_step": 34243, "epoch": 951, "lr": 0.001} +{"train_loss": 0.025377782061696053, "train_loss_bc": 0.025363711640238762, "train_loss_llm": 0.0014070570468902588, "grad_norm": 0.08849220722913742, "global_step": 34244, "epoch": 951, "lr": 0.001} +{"train_loss": 0.03434380888938904, "train_loss_bc": 0.03434380888938904, "train_loss_llm": 0.0, "grad_norm": 0.056557733565568924, "global_step": 34245, "epoch": 951, "lr": 0.001} +{"train_loss": 0.02953139878809452, "train_loss_bc": 0.02953139878809452, "train_loss_llm": 0.0, "grad_norm": 0.08307360857725143, "global_step": 34246, "epoch": 951, "lr": 0.001} +{"train_loss": 0.0272264014929533, "train_loss_bc": 0.0272264014929533, "train_loss_llm": 0.0, "grad_norm": 0.11156971007585526, "global_step": 34247, "epoch": 951, "lr": 0.001} +{"train_loss": 0.024111170321702957, "train_loss_bc": 0.024105602875351906, "train_loss_llm": 0.0005567073822021484, "grad_norm": 0.11607002466917038, "global_step": 34248, "epoch": 951, "lr": 0.001} +{"train_loss": 0.029219044372439384, "train_loss_bc": 0.029218511655926704, "train_loss_llm": 5.328655242919922e-05, "grad_norm": 0.03520834818482399, "global_step": 34249, "epoch": 951, "lr": 0.001} +{"train_loss": 0.026801278814673424, "train_loss_bc": 0.02679142728447914, "train_loss_llm": 0.000985175371170044, "grad_norm": 0.05158254876732826, "global_step": 34250, "epoch": 951, "lr": 0.001} +{"train_loss": 0.031018095090985298, "train_loss_bc": 0.031018095090985298, "train_loss_llm": 0.0, "grad_norm": 0.07404561340808868, "global_step": 34251, "epoch": 951, "lr": 0.001} +{"train_loss": 0.023581789806485176, "train_loss_bc": 0.023581789806485176, "train_loss_llm": 0.0, "grad_norm": 0.08754831552505493, "global_step": 34252, "epoch": 951, "lr": 0.001} +{"train_loss": 0.02826625667512417, "train_loss_bc": 0.02826625667512417, "train_loss_llm": 0.0, "grad_norm": 0.038762472569942474, "global_step": 34253, "epoch": 951, "lr": 0.001} +{"train_loss": 0.032864395529031754, "train_loss_bc": 0.032864395529031754, "train_loss_llm": 0.0, "grad_norm": 0.06096683815121651, "global_step": 34254, "epoch": 951, "lr": 0.001} +{"train_loss": 0.027950383722782135, "train_loss_bc": 0.027950383722782135, "train_loss_llm": 0.0, "grad_norm": 0.07962505519390106, "global_step": 34255, "epoch": 951, "lr": 0.001} +{"train_loss": 0.02593405917286873, "train_loss_bc": 0.02593405917286873, "train_loss_llm": 0.0, "grad_norm": 0.09679166227579117, "global_step": 34256, "epoch": 951, "lr": 0.001} +{"train_loss": 0.025054020807147026, "train_loss_bc": 0.025054020807147026, "train_loss_llm": 0.0, "grad_norm": 0.030690673738718033, "global_step": 34257, "epoch": 951, "lr": 0.001} +{"train_loss": 0.023274485021829605, "train_loss_bc": 0.023274485021829605, "train_loss_llm": 0.0, "grad_norm": 0.04847435653209686, "global_step": 34258, "epoch": 951, "lr": 0.001} +{"train_loss": 0.03357812017202377, "train_loss_bc": 0.033574655652046204, "train_loss_llm": 0.0003464221954345703, "grad_norm": 0.07160160690546036, "global_step": 34259, "epoch": 951, "lr": 0.001} +{"train_loss": 0.025788497179746628, "train_loss_bc": 0.025788497179746628, "train_loss_llm": 0.0, "grad_norm": 0.08965499699115753, "global_step": 34260, "epoch": 951, "lr": 0.001} +{"train_loss": 0.03154272958636284, "train_loss_bc": 0.03154272958636284, "train_loss_llm": 0.0, "grad_norm": 0.037030767649412155, "global_step": 34261, "epoch": 951, "lr": 0.001} +{"train_loss": 0.025899549946188927, "train_loss_bc": 0.025896664708852768, "train_loss_llm": 0.0002885162830352783, "grad_norm": 0.06347232311964035, "global_step": 34262, "epoch": 951, "lr": 0.001} +{"train_loss": 0.026902303099632263, "train_loss_bc": 0.026902303099632263, "train_loss_llm": 0.0, "grad_norm": 0.07204372435808182, "global_step": 34263, "epoch": 951, "lr": 0.001} +{"train_loss": 0.03261131793260574, "train_loss_bc": 0.03261131793260574, "train_loss_llm": 0.0, "grad_norm": 0.09281256049871445, "global_step": 34264, "epoch": 951, "lr": 0.001} +{"train_loss": 0.03027552179992199, "train_loss_bc": 0.03027552179992199, "train_loss_llm": 0.0, "grad_norm": 0.039183929562568665, "global_step": 34265, "epoch": 951, "lr": 0.001} +{"train_loss": 0.023361945524811745, "train_loss_bc": 0.023361945524811745, "train_loss_llm": 0.0, "grad_norm": 0.0587100014090538, "global_step": 34266, "epoch": 951, "lr": 0.001} +{"train_loss": 0.025283368304371834, "train_loss_bc": 0.025283368304371834, "train_loss_llm": 0.0, "grad_norm": 0.07039520144462585, "global_step": 34267, "epoch": 951, "lr": 0.001} +{"train_loss": 0.022965727373957634, "train_loss_bc": 0.022965287789702415, "train_loss_llm": 4.398822784423828e-05, "grad_norm": 0.08762989938259125, "global_step": 34268, "epoch": 951, "lr": 0.001} +{"train_loss": 0.03287886828184128, "train_loss_bc": 0.03287886828184128, "train_loss_llm": 0.0, "grad_norm": 0.04665878787636757, "global_step": 34269, "epoch": 951, "lr": 0.001} +{"train_loss": 0.029835648834705353, "train_loss_bc": 0.029835648834705353, "train_loss_llm": 0.0, "grad_norm": 0.062428414821624756, "global_step": 34270, "epoch": 951, "lr": 0.001} +{"train_loss": 0.02853466984298494, "train_loss_bc": 0.028812354430556297, "train_loss_llm": 0.003222256898880005, "grad_norm": 0.08002866804599762, "global_step": 34271, "epoch": 951, "lr": 0.001, "val_loss": 0.10855318605899811} +{"train_loss": 0.030367929488420486, "train_loss_bc": 0.03036600910127163, "train_loss_llm": 0.0001919865608215332, "grad_norm": 0.09913909435272217, "global_step": 34272, "epoch": 952, "lr": 0.001} +{"train_loss": 0.03284238278865814, "train_loss_bc": 0.03283437713980675, "train_loss_llm": 0.000800400972366333, "grad_norm": 0.035651180893182755, "global_step": 34273, "epoch": 952, "lr": 0.001} +{"train_loss": 0.031502559781074524, "train_loss_bc": 0.03149649500846863, "train_loss_llm": 0.000606536865234375, "grad_norm": 0.058143533766269684, "global_step": 34274, "epoch": 952, "lr": 0.001} +{"train_loss": 0.026700126007199287, "train_loss_bc": 0.026696160435676575, "train_loss_llm": 0.0003966391086578369, "grad_norm": 0.0762803703546524, "global_step": 34275, "epoch": 952, "lr": 0.001} +{"train_loss": 0.024817686527967453, "train_loss_bc": 0.024817686527967453, "train_loss_llm": 0.0, "grad_norm": 0.09059739112854004, "global_step": 34276, "epoch": 952, "lr": 0.001} +{"train_loss": 0.025240978226065636, "train_loss_bc": 0.025240978226065636, "train_loss_llm": 0.0, "grad_norm": 0.03597129508852959, "global_step": 34277, "epoch": 952, "lr": 0.001} +{"train_loss": 0.025504356250166893, "train_loss_bc": 0.025504356250166893, "train_loss_llm": 0.0, "grad_norm": 0.047813404351472855, "global_step": 34278, "epoch": 952, "lr": 0.001} +{"train_loss": 0.029878102242946625, "train_loss_bc": 0.029878102242946625, "train_loss_llm": 0.0, "grad_norm": 0.06521978974342346, "global_step": 34279, "epoch": 952, "lr": 0.001} +{"train_loss": 0.02982250601053238, "train_loss_bc": 0.02982250601053238, "train_loss_llm": 0.0, "grad_norm": 0.07532931119203568, "global_step": 34280, "epoch": 952, "lr": 0.001} +{"train_loss": 0.030699988827109337, "train_loss_bc": 0.030681615695357323, "train_loss_llm": 0.0018373429775238037, "grad_norm": 0.04157569631934166, "global_step": 34281, "epoch": 952, "lr": 0.001} +{"train_loss": 0.03160993754863739, "train_loss_bc": 0.03160993754863739, "train_loss_llm": 0.0, "grad_norm": 0.07220970094203949, "global_step": 34282, "epoch": 952, "lr": 0.001} +{"train_loss": 0.03642936795949936, "train_loss_bc": 0.036422718316316605, "train_loss_llm": 0.000664830207824707, "grad_norm": 0.09630375355482101, "global_step": 34283, "epoch": 952, "lr": 0.001} +{"train_loss": 0.02813584730029106, "train_loss_bc": 0.02813584730029106, "train_loss_llm": 0.0, "grad_norm": 0.12322553247213364, "global_step": 34284, "epoch": 952, "lr": 0.001} +{"train_loss": 0.030326638370752335, "train_loss_bc": 0.030326638370752335, "train_loss_llm": 0.0, "grad_norm": 0.037490274757146835, "global_step": 34285, "epoch": 952, "lr": 0.001} +{"train_loss": 0.031125223264098167, "train_loss_bc": 0.031125223264098167, "train_loss_llm": 0.0, "grad_norm": 0.060342103242874146, "global_step": 34286, "epoch": 952, "lr": 0.001} +{"train_loss": 0.02465604618191719, "train_loss_bc": 0.02465604618191719, "train_loss_llm": 0.0, "grad_norm": 0.07838323712348938, "global_step": 34287, "epoch": 952, "lr": 0.001} +{"train_loss": 0.02497148886322975, "train_loss_bc": 0.02497148886322975, "train_loss_llm": 0.0, "grad_norm": 0.09630241990089417, "global_step": 34288, "epoch": 952, "lr": 0.001} +{"train_loss": 0.022777024656534195, "train_loss_bc": 0.022776812314987183, "train_loss_llm": 2.1219253540039062e-05, "grad_norm": 0.033538177609443665, "global_step": 34289, "epoch": 952, "lr": 0.001} +{"train_loss": 0.02249963767826557, "train_loss_bc": 0.02249963767826557, "train_loss_llm": 0.0, "grad_norm": 0.05040770769119263, "global_step": 34290, "epoch": 952, "lr": 0.001} +{"train_loss": 0.026998110115528107, "train_loss_bc": 0.026998110115528107, "train_loss_llm": 0.0, "grad_norm": 0.06873002648353577, "global_step": 34291, "epoch": 952, "lr": 0.001} +{"train_loss": 0.028769632801413536, "train_loss_bc": 0.028769632801413536, "train_loss_llm": 0.0, "grad_norm": 0.08502154052257538, "global_step": 34292, "epoch": 952, "lr": 0.001} +{"train_loss": 0.03148319572210312, "train_loss_bc": 0.03148319572210312, "train_loss_llm": 0.0, "grad_norm": 0.044235557317733765, "global_step": 34293, "epoch": 952, "lr": 0.001} +{"train_loss": 0.02440943568944931, "train_loss_bc": 0.02440069243311882, "train_loss_llm": 0.000874251127243042, "grad_norm": 0.05809473246335983, "global_step": 34294, "epoch": 952, "lr": 0.001} +{"train_loss": 0.02941429428756237, "train_loss_bc": 0.02941429428756237, "train_loss_llm": 0.0, "grad_norm": 0.06883858144283295, "global_step": 34295, "epoch": 952, "lr": 0.001} +{"train_loss": 0.027896219864487648, "train_loss_bc": 0.027896219864487648, "train_loss_llm": 0.0, "grad_norm": 0.08224542438983917, "global_step": 34296, "epoch": 952, "lr": 0.001} +{"train_loss": 0.030703522264957428, "train_loss_bc": 0.030703522264957428, "train_loss_llm": 0.0, "grad_norm": 0.03678543493151665, "global_step": 34297, "epoch": 952, "lr": 0.001} +{"train_loss": 0.025674287229776382, "train_loss_bc": 0.025674287229776382, "train_loss_llm": 0.0, "grad_norm": 0.047214679419994354, "global_step": 34298, "epoch": 952, "lr": 0.001} +{"train_loss": 0.025637611746788025, "train_loss_bc": 0.025634167715907097, "train_loss_llm": 0.00034433603286743164, "grad_norm": 0.06119463965296745, "global_step": 34299, "epoch": 952, "lr": 0.001} +{"train_loss": 0.027883097529411316, "train_loss_bc": 0.027883097529411316, "train_loss_llm": 0.0, "grad_norm": 0.08179327100515366, "global_step": 34300, "epoch": 952, "lr": 0.001} +{"train_loss": 0.038733288645744324, "train_loss_bc": 0.038733288645744324, "train_loss_llm": 0.0, "grad_norm": 0.04268311336636543, "global_step": 34301, "epoch": 952, "lr": 0.001} +{"train_loss": 0.02848167158663273, "train_loss_bc": 0.02848167158663273, "train_loss_llm": 0.0, "grad_norm": 0.06397935748100281, "global_step": 34302, "epoch": 952, "lr": 0.001} +{"train_loss": 0.02943008579313755, "train_loss_bc": 0.02941312827169895, "train_loss_llm": 0.0016957223415374756, "grad_norm": 0.08878348767757416, "global_step": 34303, "epoch": 952, "lr": 0.001} +{"train_loss": 0.02045333757996559, "train_loss_bc": 0.02045333757996559, "train_loss_llm": 0.0, "grad_norm": 0.10813116282224655, "global_step": 34304, "epoch": 952, "lr": 0.001} +{"train_loss": 0.031957875937223434, "train_loss_bc": 0.03194970637559891, "train_loss_llm": 0.0008167922496795654, "grad_norm": 0.037868138402700424, "global_step": 34305, "epoch": 952, "lr": 0.001} +{"train_loss": 0.034140508621931076, "train_loss_bc": 0.034140508621931076, "train_loss_llm": 0.0, "grad_norm": 0.05945422127842903, "global_step": 34306, "epoch": 952, "lr": 0.001} +{"train_loss": 0.028534843327684536, "train_loss_bc": 0.02527599409222603, "train_loss_llm": 0.0004363059997558594, "grad_norm": 0.08395738899707794, "global_step": 34307, "epoch": 952, "lr": 0.001, "val_loss": 0.1081143245100975} +{"train_loss": 0.028909333050251007, "train_loss_bc": 0.028909333050251007, "train_loss_llm": 0.0, "grad_norm": 0.09539246559143066, "global_step": 34308, "epoch": 953, "lr": 0.001} +{"train_loss": 0.02869355119764805, "train_loss_bc": 0.02869355119764805, "train_loss_llm": 0.0, "grad_norm": 0.032302193343639374, "global_step": 34309, "epoch": 953, "lr": 0.001} +{"train_loss": 0.0270191952586174, "train_loss_bc": 0.027013203129172325, "train_loss_llm": 0.000599205493927002, "grad_norm": 0.054020803421735764, "global_step": 34310, "epoch": 953, "lr": 0.001} +{"train_loss": 0.02319406345486641, "train_loss_bc": 0.02319406345486641, "train_loss_llm": 0.0, "grad_norm": 0.0727776363492012, "global_step": 34311, "epoch": 953, "lr": 0.001} +{"train_loss": 0.025352297350764275, "train_loss_bc": 0.025352297350764275, "train_loss_llm": 0.0, "grad_norm": 0.08381243050098419, "global_step": 34312, "epoch": 953, "lr": 0.001} +{"train_loss": 0.03324984759092331, "train_loss_bc": 0.03324984759092331, "train_loss_llm": 0.0, "grad_norm": 0.04173463582992554, "global_step": 34313, "epoch": 953, "lr": 0.001} +{"train_loss": 0.023668326437473297, "train_loss_bc": 0.023668326437473297, "train_loss_llm": 0.0, "grad_norm": 0.05637026205658913, "global_step": 34314, "epoch": 953, "lr": 0.001} +{"train_loss": 0.023242706432938576, "train_loss_bc": 0.023230135440826416, "train_loss_llm": 0.001257181167602539, "grad_norm": 0.06443512439727783, "global_step": 34315, "epoch": 953, "lr": 0.001} +{"train_loss": 0.03658844530582428, "train_loss_bc": 0.03658844530582428, "train_loss_llm": 0.0, "grad_norm": 0.07554598897695541, "global_step": 34316, "epoch": 953, "lr": 0.001} +{"train_loss": 0.03347472846508026, "train_loss_bc": 0.03347472846508026, "train_loss_llm": 0.0, "grad_norm": 0.061688944697380066, "global_step": 34317, "epoch": 953, "lr": 0.001} +{"train_loss": 0.023427970707416534, "train_loss_bc": 0.023427970707416534, "train_loss_llm": 0.0, "grad_norm": 0.0742797777056694, "global_step": 34318, "epoch": 953, "lr": 0.001} +{"train_loss": 0.03128166124224663, "train_loss_bc": 0.031266964972019196, "train_loss_llm": 0.0014696121215820312, "grad_norm": 0.1009407639503479, "global_step": 34319, "epoch": 953, "lr": 0.001} +{"train_loss": 0.025029970332980156, "train_loss_bc": 0.025029970332980156, "train_loss_llm": 0.0, "grad_norm": 0.11098858714103699, "global_step": 34320, "epoch": 953, "lr": 0.001} +{"train_loss": 0.030147165060043335, "train_loss_bc": 0.030147165060043335, "train_loss_llm": 0.0, "grad_norm": 0.03592752292752266, "global_step": 34321, "epoch": 953, "lr": 0.001} +{"train_loss": 0.03154982253909111, "train_loss_bc": 0.03154982253909111, "train_loss_llm": 0.0, "grad_norm": 0.06114642322063446, "global_step": 34322, "epoch": 953, "lr": 0.001} +{"train_loss": 0.03251003101468086, "train_loss_bc": 0.032507918775081635, "train_loss_llm": 0.00021129846572875977, "grad_norm": 0.08403758704662323, "global_step": 34323, "epoch": 953, "lr": 0.001} +{"train_loss": 0.027114391326904297, "train_loss_bc": 0.027114391326904297, "train_loss_llm": 0.0, "grad_norm": 0.11070925742387772, "global_step": 34324, "epoch": 953, "lr": 0.001} +{"train_loss": 0.02800864540040493, "train_loss_bc": 0.02800864540040493, "train_loss_llm": 0.0, "grad_norm": 0.046007100492715836, "global_step": 34325, "epoch": 953, "lr": 0.001} +{"train_loss": 0.026066673919558525, "train_loss_bc": 0.026066673919558525, "train_loss_llm": 0.0, "grad_norm": 0.06624022871255875, "global_step": 34326, "epoch": 953, "lr": 0.001} +{"train_loss": 0.026074901223182678, "train_loss_bc": 0.026074901223182678, "train_loss_llm": 0.0, "grad_norm": 0.08835389465093613, "global_step": 34327, "epoch": 953, "lr": 0.001} +{"train_loss": 0.028366845101118088, "train_loss_bc": 0.028366845101118088, "train_loss_llm": 0.0, "grad_norm": 0.10034292936325073, "global_step": 34328, "epoch": 953, "lr": 0.001} +{"train_loss": 0.02935284562408924, "train_loss_bc": 0.029336124658584595, "train_loss_llm": 0.0016721785068511963, "grad_norm": 0.03185024484992027, "global_step": 34329, "epoch": 953, "lr": 0.001} +{"train_loss": 0.025088012218475342, "train_loss_bc": 0.025088012218475342, "train_loss_llm": 0.0, "grad_norm": 0.04369882866740227, "global_step": 34330, "epoch": 953, "lr": 0.001} +{"train_loss": 0.03381888568401337, "train_loss_bc": 0.03381888568401337, "train_loss_llm": 0.0, "grad_norm": 0.06894361972808838, "global_step": 34331, "epoch": 953, "lr": 0.001} +{"train_loss": 0.02524944208562374, "train_loss_bc": 0.02524944208562374, "train_loss_llm": 0.0, "grad_norm": 0.09179925918579102, "global_step": 34332, "epoch": 953, "lr": 0.001} +{"train_loss": 0.027532340958714485, "train_loss_bc": 0.027532340958714485, "train_loss_llm": 0.0, "grad_norm": 0.04246455058455467, "global_step": 34333, "epoch": 953, "lr": 0.001} +{"train_loss": 0.032811831682920456, "train_loss_bc": 0.03280975669622421, "train_loss_llm": 0.00020760297775268555, "grad_norm": 0.07159700989723206, "global_step": 34334, "epoch": 953, "lr": 0.001} +{"train_loss": 0.028923749923706055, "train_loss_bc": 0.028923749923706055, "train_loss_llm": 0.0, "grad_norm": 0.09067513048648834, "global_step": 34335, "epoch": 953, "lr": 0.001} +{"train_loss": 0.028705626726150513, "train_loss_bc": 0.028705626726150513, "train_loss_llm": 0.0, "grad_norm": 0.11590936779975891, "global_step": 34336, "epoch": 953, "lr": 0.001} +{"train_loss": 0.029088610783219337, "train_loss_bc": 0.029088610783219337, "train_loss_llm": 0.0, "grad_norm": 0.040486857295036316, "global_step": 34337, "epoch": 953, "lr": 0.001} +{"train_loss": 0.025772033259272575, "train_loss_bc": 0.02574986033141613, "train_loss_llm": 0.002217233180999756, "grad_norm": 0.05294252932071686, "global_step": 34338, "epoch": 953, "lr": 0.001} +{"train_loss": 0.025345906615257263, "train_loss_bc": 0.025345906615257263, "train_loss_llm": 0.0, "grad_norm": 0.07116249203681946, "global_step": 34339, "epoch": 953, "lr": 0.001} +{"train_loss": 0.029840366914868355, "train_loss_bc": 0.029840366914868355, "train_loss_llm": 0.0, "grad_norm": 0.08258950710296631, "global_step": 34340, "epoch": 953, "lr": 0.001} +{"train_loss": 0.029410474002361298, "train_loss_bc": 0.029390225186944008, "train_loss_llm": 0.002024918794631958, "grad_norm": 0.03981923311948776, "global_step": 34341, "epoch": 953, "lr": 0.001} +{"train_loss": 0.03300793468952179, "train_loss_bc": 0.03300793468952179, "train_loss_llm": 0.0, "grad_norm": 0.053552333265542984, "global_step": 34342, "epoch": 953, "lr": 0.001} +{"train_loss": 0.028479793678141303, "train_loss_bc": 0.028353938832879066, "train_loss_llm": 0.0, "grad_norm": 0.08168258517980576, "global_step": 34343, "epoch": 953, "lr": 0.001, "val_loss": 0.10892049223184586} +{"train_loss": 0.03216715157032013, "train_loss_bc": 0.03216715157032013, "train_loss_llm": 0.0, "grad_norm": 0.1081443652510643, "global_step": 34344, "epoch": 954, "lr": 0.001} +{"train_loss": 0.02429891936480999, "train_loss_bc": 0.02429891936480999, "train_loss_llm": 0.0, "grad_norm": 0.032270386815071106, "global_step": 34345, "epoch": 954, "lr": 0.001} +{"train_loss": 0.021513182669878006, "train_loss_bc": 0.021513182669878006, "train_loss_llm": 0.0, "grad_norm": 0.04387999325990677, "global_step": 34346, "epoch": 954, "lr": 0.001} +{"train_loss": 0.033593274652957916, "train_loss_bc": 0.033593274652957916, "train_loss_llm": 0.0, "grad_norm": 0.06807707250118256, "global_step": 34347, "epoch": 954, "lr": 0.001} +{"train_loss": 0.02366635389626026, "train_loss_bc": 0.023659303784370422, "train_loss_llm": 0.0007050931453704834, "grad_norm": 0.09083850681781769, "global_step": 34348, "epoch": 954, "lr": 0.001} +{"train_loss": 0.030688906088471413, "train_loss_bc": 0.030688906088471413, "train_loss_llm": 0.0, "grad_norm": 0.040484074503183365, "global_step": 34349, "epoch": 954, "lr": 0.001} +{"train_loss": 0.02692786604166031, "train_loss_bc": 0.02692786604166031, "train_loss_llm": 0.0, "grad_norm": 0.04833085089921951, "global_step": 34350, "epoch": 954, "lr": 0.001} +{"train_loss": 0.030126111581921577, "train_loss_bc": 0.030126111581921577, "train_loss_llm": 0.0, "grad_norm": 0.07486172020435333, "global_step": 34351, "epoch": 954, "lr": 0.001} +{"train_loss": 0.025553595274686813, "train_loss_bc": 0.025547625496983528, "train_loss_llm": 0.0005970001220703125, "grad_norm": 0.09144612401723862, "global_step": 34352, "epoch": 954, "lr": 0.001} +{"train_loss": 0.033436745405197144, "train_loss_bc": 0.033436745405197144, "train_loss_llm": 0.0, "grad_norm": 0.035062532871961594, "global_step": 34353, "epoch": 954, "lr": 0.001} +{"train_loss": 0.03317897394299507, "train_loss_bc": 0.03317897394299507, "train_loss_llm": 0.0, "grad_norm": 0.05187856778502464, "global_step": 34354, "epoch": 954, "lr": 0.001} +{"train_loss": 0.026819976046681404, "train_loss_bc": 0.026819976046681404, "train_loss_llm": 0.0, "grad_norm": 0.07937537878751755, "global_step": 34355, "epoch": 954, "lr": 0.001} +{"train_loss": 0.02763773687183857, "train_loss_bc": 0.027631517499685287, "train_loss_llm": 0.0006218850612640381, "grad_norm": 0.08780761808156967, "global_step": 34356, "epoch": 954, "lr": 0.001} +{"train_loss": 0.028941510245203972, "train_loss_bc": 0.028935572132468224, "train_loss_llm": 0.0005937814712524414, "grad_norm": 0.040890228003263474, "global_step": 34357, "epoch": 954, "lr": 0.001} +{"train_loss": 0.02579973265528679, "train_loss_bc": 0.02579973265528679, "train_loss_llm": 0.0, "grad_norm": 0.05643463134765625, "global_step": 34358, "epoch": 954, "lr": 0.001} +{"train_loss": 0.024471884593367577, "train_loss_bc": 0.024471884593367577, "train_loss_llm": 0.0, "grad_norm": 0.07767000794410706, "global_step": 34359, "epoch": 954, "lr": 0.001} +{"train_loss": 0.028870604932308197, "train_loss_bc": 0.028867756947875023, "train_loss_llm": 0.0002848505973815918, "grad_norm": 0.1013658195734024, "global_step": 34360, "epoch": 954, "lr": 0.001} +{"train_loss": 0.026360513642430305, "train_loss_bc": 0.026360513642430305, "train_loss_llm": 0.0, "grad_norm": 0.03675365075469017, "global_step": 34361, "epoch": 954, "lr": 0.001} +{"train_loss": 0.025689009577035904, "train_loss_bc": 0.025689009577035904, "train_loss_llm": 0.0, "grad_norm": 0.055004626512527466, "global_step": 34362, "epoch": 954, "lr": 0.001} +{"train_loss": 0.029551228508353233, "train_loss_bc": 0.029551228508353233, "train_loss_llm": 0.0, "grad_norm": 0.07456017285585403, "global_step": 34363, "epoch": 954, "lr": 0.001} +{"train_loss": 0.03147416561841965, "train_loss_bc": 0.03147416561841965, "train_loss_llm": 0.0, "grad_norm": 0.09762880951166153, "global_step": 34364, "epoch": 954, "lr": 0.001} +{"train_loss": 0.023638833314180374, "train_loss_bc": 0.023638833314180374, "train_loss_llm": 0.0, "grad_norm": 0.027400176972150803, "global_step": 34365, "epoch": 954, "lr": 0.001} +{"train_loss": 0.033315520733594894, "train_loss_bc": 0.03330487012863159, "train_loss_llm": 0.001064985990524292, "grad_norm": 0.05141480267047882, "global_step": 34366, "epoch": 954, "lr": 0.001} +{"train_loss": 0.03229201212525368, "train_loss_bc": 0.03228791058063507, "train_loss_llm": 0.00041031837463378906, "grad_norm": 0.07227006554603577, "global_step": 34367, "epoch": 954, "lr": 0.001} +{"train_loss": 0.02426806464791298, "train_loss_bc": 0.02426806464791298, "train_loss_llm": 0.0, "grad_norm": 0.09149453788995743, "global_step": 34368, "epoch": 954, "lr": 0.001} +{"train_loss": 0.029361996799707413, "train_loss_bc": 0.029351022094488144, "train_loss_llm": 0.0010974705219268799, "grad_norm": 0.039294902235269547, "global_step": 34369, "epoch": 954, "lr": 0.001} +{"train_loss": 0.03307680785655975, "train_loss_bc": 0.03307680785655975, "train_loss_llm": 0.0, "grad_norm": 0.06604968756437302, "global_step": 34370, "epoch": 954, "lr": 0.001} +{"train_loss": 0.029710717499256134, "train_loss_bc": 0.029710717499256134, "train_loss_llm": 0.0, "grad_norm": 0.09939766675233841, "global_step": 34371, "epoch": 954, "lr": 0.001} +{"train_loss": 0.029566556215286255, "train_loss_bc": 0.029566556215286255, "train_loss_llm": 0.0, "grad_norm": 0.12469004094600677, "global_step": 34372, "epoch": 954, "lr": 0.001} +{"train_loss": 0.03123566322028637, "train_loss_bc": 0.03123566322028637, "train_loss_llm": 0.0, "grad_norm": 0.033311549574136734, "global_step": 34373, "epoch": 954, "lr": 0.001} +{"train_loss": 0.03240259736776352, "train_loss_bc": 0.03240259736776352, "train_loss_llm": 0.0, "grad_norm": 0.05666634067893028, "global_step": 34374, "epoch": 954, "lr": 0.001} +{"train_loss": 0.029248541221022606, "train_loss_bc": 0.029248541221022606, "train_loss_llm": 0.0, "grad_norm": 0.07906176894903183, "global_step": 34375, "epoch": 954, "lr": 0.001} +{"train_loss": 0.029868783429265022, "train_loss_bc": 0.029847819358110428, "train_loss_llm": 0.0020964443683624268, "grad_norm": 0.09897563606500626, "global_step": 34376, "epoch": 954, "lr": 0.001} +{"train_loss": 0.025908999145030975, "train_loss_bc": 0.02589281089603901, "train_loss_llm": 0.0016187727451324463, "grad_norm": 0.03941657394170761, "global_step": 34377, "epoch": 954, "lr": 0.001} +{"train_loss": 0.0246012844145298, "train_loss_bc": 0.0246012844145298, "train_loss_llm": 0.0, "grad_norm": 0.053692638874053955, "global_step": 34378, "epoch": 954, "lr": 0.001} +{"train_loss": 0.028648843316154346, "train_loss_bc": 0.03209034726023674, "train_loss_llm": 0.00041922926902770996, "grad_norm": 0.08464241027832031, "global_step": 34379, "epoch": 954, "lr": 0.001, "val_loss": 0.10720787942409515} +{"train_loss": 0.030324483290314674, "train_loss_bc": 0.030324483290314674, "train_loss_llm": 0.0, "grad_norm": 0.10540676862001419, "global_step": 34380, "epoch": 955, "lr": 0.001} +{"train_loss": 0.028562869876623154, "train_loss_bc": 0.028562869876623154, "train_loss_llm": 0.0, "grad_norm": 0.039815083146095276, "global_step": 34381, "epoch": 955, "lr": 0.001} +{"train_loss": 0.031745120882987976, "train_loss_bc": 0.03172261267900467, "train_loss_llm": 0.0022509992122650146, "grad_norm": 0.06458712369203568, "global_step": 34382, "epoch": 955, "lr": 0.001} +{"train_loss": 0.027410032227635384, "train_loss_bc": 0.027410032227635384, "train_loss_llm": 0.0, "grad_norm": 0.10138808190822601, "global_step": 34383, "epoch": 955, "lr": 0.001} +{"train_loss": 0.023200586438179016, "train_loss_bc": 0.02319657802581787, "train_loss_llm": 0.00040081143379211426, "grad_norm": 0.1227869987487793, "global_step": 34384, "epoch": 955, "lr": 0.001} +{"train_loss": 0.030020492151379585, "train_loss_bc": 0.030020492151379585, "train_loss_llm": 0.0, "grad_norm": 0.05328013375401497, "global_step": 34385, "epoch": 955, "lr": 0.001} +{"train_loss": 0.03008098155260086, "train_loss_bc": 0.030070757493376732, "train_loss_llm": 0.0010223686695098877, "grad_norm": 0.07840684801340103, "global_step": 34386, "epoch": 955, "lr": 0.001} +{"train_loss": 0.02709236368536949, "train_loss_bc": 0.02709236368536949, "train_loss_llm": 0.0, "grad_norm": 0.09910053759813309, "global_step": 34387, "epoch": 955, "lr": 0.001} +{"train_loss": 0.02502426691353321, "train_loss_bc": 0.02500767447054386, "train_loss_llm": 0.0016592741012573242, "grad_norm": 0.12827469408512115, "global_step": 34388, "epoch": 955, "lr": 0.001} +{"train_loss": 0.029112758114933968, "train_loss_bc": 0.029112758114933968, "train_loss_llm": 0.0, "grad_norm": 0.039466727524995804, "global_step": 34389, "epoch": 955, "lr": 0.001} +{"train_loss": 0.022948140278458595, "train_loss_bc": 0.022948140278458595, "train_loss_llm": 0.0, "grad_norm": 0.05747826769948006, "global_step": 34390, "epoch": 955, "lr": 0.001} +{"train_loss": 0.031126119196414948, "train_loss_bc": 0.031104544177651405, "train_loss_llm": 0.00215756893157959, "grad_norm": 0.0862806960940361, "global_step": 34391, "epoch": 955, "lr": 0.001} +{"train_loss": 0.030351370573043823, "train_loss_bc": 0.030351370573043823, "train_loss_llm": 0.0, "grad_norm": 0.09970035403966904, "global_step": 34392, "epoch": 955, "lr": 0.001} +{"train_loss": 0.036034002900123596, "train_loss_bc": 0.036034002900123596, "train_loss_llm": 0.0, "grad_norm": 0.04625684767961502, "global_step": 34393, "epoch": 955, "lr": 0.001} +{"train_loss": 0.029523231089115143, "train_loss_bc": 0.029523231089115143, "train_loss_llm": 0.0, "grad_norm": 0.060247279703617096, "global_step": 34394, "epoch": 955, "lr": 0.001} +{"train_loss": 0.022622305899858475, "train_loss_bc": 0.022622305899858475, "train_loss_llm": 0.0, "grad_norm": 0.07389229536056519, "global_step": 34395, "epoch": 955, "lr": 0.001} +{"train_loss": 0.03090623952448368, "train_loss_bc": 0.030901726335287094, "train_loss_llm": 0.0004513859748840332, "grad_norm": 0.09376129508018494, "global_step": 34396, "epoch": 955, "lr": 0.001} +{"train_loss": 0.03294181078672409, "train_loss_bc": 0.03294181078672409, "train_loss_llm": 0.0, "grad_norm": 0.04823415353894234, "global_step": 34397, "epoch": 955, "lr": 0.001} +{"train_loss": 0.03279530256986618, "train_loss_bc": 0.03279530256986618, "train_loss_llm": 0.0, "grad_norm": 0.07431968301534653, "global_step": 34398, "epoch": 955, "lr": 0.001} +{"train_loss": 0.027921125292778015, "train_loss_bc": 0.027921125292778015, "train_loss_llm": 0.0, "grad_norm": 0.09722656011581421, "global_step": 34399, "epoch": 955, "lr": 0.001} +{"train_loss": 0.026941347867250443, "train_loss_bc": 0.026941347867250443, "train_loss_llm": 0.0, "grad_norm": 0.11720606684684753, "global_step": 34400, "epoch": 955, "lr": 0.001} +{"train_loss": 0.026939289644360542, "train_loss_bc": 0.026939289644360542, "train_loss_llm": 0.0, "grad_norm": 0.03662728890776634, "global_step": 34401, "epoch": 955, "lr": 0.001} +{"train_loss": 0.03005914017558098, "train_loss_bc": 0.03005914017558098, "train_loss_llm": 0.0, "grad_norm": 0.058377090841531754, "global_step": 34402, "epoch": 955, "lr": 0.001} +{"train_loss": 0.02944040298461914, "train_loss_bc": 0.02944040298461914, "train_loss_llm": 0.0, "grad_norm": 0.08161434531211853, "global_step": 34403, "epoch": 955, "lr": 0.001} +{"train_loss": 0.02913607284426689, "train_loss_bc": 0.029117509722709656, "train_loss_llm": 0.0018563270568847656, "grad_norm": 0.09727421402931213, "global_step": 34404, "epoch": 955, "lr": 0.001} +{"train_loss": 0.023446261882781982, "train_loss_bc": 0.023446261882781982, "train_loss_llm": 0.0, "grad_norm": 0.03732422739267349, "global_step": 34405, "epoch": 955, "lr": 0.001} +{"train_loss": 0.03478290140628815, "train_loss_bc": 0.03478290140628815, "train_loss_llm": 0.0, "grad_norm": 0.07293040305376053, "global_step": 34406, "epoch": 955, "lr": 0.001} +{"train_loss": 0.03290990740060806, "train_loss_bc": 0.03290990740060806, "train_loss_llm": 0.0, "grad_norm": 0.10287296026945114, "global_step": 34407, "epoch": 955, "lr": 0.001} +{"train_loss": 0.029257573187351227, "train_loss_bc": 0.029257573187351227, "train_loss_llm": 0.0, "grad_norm": 0.12174486368894577, "global_step": 34408, "epoch": 955, "lr": 0.001} +{"train_loss": 0.027262117713689804, "train_loss_bc": 0.027262117713689804, "train_loss_llm": 0.0, "grad_norm": 0.033409666270017624, "global_step": 34409, "epoch": 955, "lr": 0.001} +{"train_loss": 0.031160062178969383, "train_loss_bc": 0.031157605350017548, "train_loss_llm": 0.00024574995040893555, "grad_norm": 0.07175513356924057, "global_step": 34410, "epoch": 955, "lr": 0.001} +{"train_loss": 0.026803793385624886, "train_loss_bc": 0.026803793385624886, "train_loss_llm": 0.0, "grad_norm": 0.09102431684732437, "global_step": 34411, "epoch": 955, "lr": 0.001} +{"train_loss": 0.027391431853175163, "train_loss_bc": 0.027391431853175163, "train_loss_llm": 0.0, "grad_norm": 0.11426413059234619, "global_step": 34412, "epoch": 955, "lr": 0.001} +{"train_loss": 0.02537749707698822, "train_loss_bc": 0.02537749707698822, "train_loss_llm": 0.0, "grad_norm": 0.0412740595638752, "global_step": 34413, "epoch": 955, "lr": 0.001} +{"train_loss": 0.02683631330728531, "train_loss_bc": 0.026832710951566696, "train_loss_llm": 0.00036016106605529785, "grad_norm": 0.07897716015577316, "global_step": 34414, "epoch": 955, "lr": 0.001} +{"train_loss": 0.02881764004834824, "train_loss_bc": 0.029947325587272644, "train_loss_llm": 0.0, "grad_norm": 0.10296907275915146, "global_step": 34415, "epoch": 955, "lr": 0.001, "train/cumulative_reward": 44.332461898253214, "train/mean_score": 11.552211927068617, "train/success_rate": 0.3, "test/cumulative_reward": 39.22057480923128, "test/mean_score": 6.3417486729119945, "test/success_rate": 0.14, "val_loss": 0.11078101396560669, "train_action_mse_error": 0.03231194242835045} +{"train_loss": 0.026902316138148308, "train_loss_bc": 0.026902316138148308, "train_loss_llm": 0.0, "grad_norm": 0.12751634418964386, "global_step": 34416, "epoch": 956, "lr": 0.001} +{"train_loss": 0.027967091649770737, "train_loss_bc": 0.02796253375709057, "train_loss_llm": 0.0004558563232421875, "grad_norm": 0.0334034338593483, "global_step": 34417, "epoch": 956, "lr": 0.001} +{"train_loss": 0.02176143415272236, "train_loss_bc": 0.02176143415272236, "train_loss_llm": 0.0, "grad_norm": 0.05474993214011192, "global_step": 34418, "epoch": 956, "lr": 0.001} +{"train_loss": 0.028959065675735474, "train_loss_bc": 0.028959065675735474, "train_loss_llm": 0.0, "grad_norm": 0.08269248157739639, "global_step": 34419, "epoch": 956, "lr": 0.001} +{"train_loss": 0.03073062188923359, "train_loss_bc": 0.03073062188923359, "train_loss_llm": 0.0, "grad_norm": 0.10324759036302567, "global_step": 34420, "epoch": 956, "lr": 0.001} +{"train_loss": 0.02692568302154541, "train_loss_bc": 0.02692568302154541, "train_loss_llm": 0.0, "grad_norm": 0.043764617294073105, "global_step": 34421, "epoch": 956, "lr": 0.001} +{"train_loss": 0.03082919307053089, "train_loss_bc": 0.03082919307053089, "train_loss_llm": 0.0, "grad_norm": 0.07107029855251312, "global_step": 34422, "epoch": 956, "lr": 0.001} +{"train_loss": 0.03233259916305542, "train_loss_bc": 0.03233259916305542, "train_loss_llm": 0.0, "grad_norm": 0.10076352208852768, "global_step": 34423, "epoch": 956, "lr": 0.001} +{"train_loss": 0.024613801389932632, "train_loss_bc": 0.024613801389932632, "train_loss_llm": 0.0, "grad_norm": 0.11390525102615356, "global_step": 34424, "epoch": 956, "lr": 0.001} +{"train_loss": 0.03156166151165962, "train_loss_bc": 0.03156166151165962, "train_loss_llm": 0.0, "grad_norm": 0.05071372911334038, "global_step": 34425, "epoch": 956, "lr": 0.001} +{"train_loss": 0.02970968559384346, "train_loss_bc": 0.02970777079463005, "train_loss_llm": 0.00019150972366333008, "grad_norm": 0.06488464027643204, "global_step": 34426, "epoch": 956, "lr": 0.001} +{"train_loss": 0.02984451688826084, "train_loss_bc": 0.02984451688826084, "train_loss_llm": 0.0, "grad_norm": 0.08010056614875793, "global_step": 34427, "epoch": 956, "lr": 0.001} +{"train_loss": 0.028743073344230652, "train_loss_bc": 0.028743073344230652, "train_loss_llm": 0.0, "grad_norm": 0.10736332833766937, "global_step": 34428, "epoch": 956, "lr": 0.001} +{"train_loss": 0.0284968763589859, "train_loss_bc": 0.028447018936276436, "train_loss_llm": 0.0049856603145599365, "grad_norm": 0.03294885903596878, "global_step": 34429, "epoch": 956, "lr": 0.001} +{"train_loss": 0.02963515929877758, "train_loss_bc": 0.02963515929877758, "train_loss_llm": 0.0, "grad_norm": 0.05195547640323639, "global_step": 34430, "epoch": 956, "lr": 0.001} +{"train_loss": 0.028461072593927383, "train_loss_bc": 0.028461072593927383, "train_loss_llm": 0.0, "grad_norm": 0.06999951601028442, "global_step": 34431, "epoch": 956, "lr": 0.001} +{"train_loss": 0.026963263750076294, "train_loss_bc": 0.026963263750076294, "train_loss_llm": 0.0, "grad_norm": 0.09135385602712631, "global_step": 34432, "epoch": 956, "lr": 0.001} +{"train_loss": 0.03174367919564247, "train_loss_bc": 0.03174367919564247, "train_loss_llm": 0.0, "grad_norm": 0.04885737597942352, "global_step": 34433, "epoch": 956, "lr": 0.001} +{"train_loss": 0.02983006462454796, "train_loss_bc": 0.029811235144734383, "train_loss_llm": 0.001883000135421753, "grad_norm": 0.08464603871107101, "global_step": 34434, "epoch": 956, "lr": 0.001} +{"train_loss": 0.03234458714723587, "train_loss_bc": 0.03234458714723587, "train_loss_llm": 0.0, "grad_norm": 0.10788095742464066, "global_step": 34435, "epoch": 956, "lr": 0.001} +{"train_loss": 0.030937567353248596, "train_loss_bc": 0.030928561463952065, "train_loss_llm": 0.000900566577911377, "grad_norm": 0.13635371625423431, "global_step": 34436, "epoch": 956, "lr": 0.001} +{"train_loss": 0.027846839278936386, "train_loss_bc": 0.027846839278936386, "train_loss_llm": 0.0, "grad_norm": 0.035586945712566376, "global_step": 34437, "epoch": 956, "lr": 0.001} +{"train_loss": 0.024099035188555717, "train_loss_bc": 0.024099035188555717, "train_loss_llm": 0.0, "grad_norm": 0.05425143614411354, "global_step": 34438, "epoch": 956, "lr": 0.001} +{"train_loss": 0.0331251285970211, "train_loss_bc": 0.0331251285970211, "train_loss_llm": 0.0, "grad_norm": 0.08763141185045242, "global_step": 34439, "epoch": 956, "lr": 0.001} +{"train_loss": 0.028588658198714256, "train_loss_bc": 0.028588658198714256, "train_loss_llm": 0.0, "grad_norm": 0.11089671403169632, "global_step": 34440, "epoch": 956, "lr": 0.001} +{"train_loss": 0.03012336604297161, "train_loss_bc": 0.030117718502879143, "train_loss_llm": 0.0005647838115692139, "grad_norm": 0.044182125478982925, "global_step": 34441, "epoch": 956, "lr": 0.001} +{"train_loss": 0.02679898589849472, "train_loss_bc": 0.02679898589849472, "train_loss_llm": 0.0, "grad_norm": 0.05441361665725708, "global_step": 34442, "epoch": 956, "lr": 0.001} +{"train_loss": 0.028804978355765343, "train_loss_bc": 0.028804978355765343, "train_loss_llm": 0.0, "grad_norm": 0.07291264086961746, "global_step": 34443, "epoch": 956, "lr": 0.001} +{"train_loss": 0.029893431812524796, "train_loss_bc": 0.029891237616539, "train_loss_llm": 0.00021943449974060059, "grad_norm": 0.09866686910390854, "global_step": 34444, "epoch": 956, "lr": 0.001} +{"train_loss": 0.023970989510416985, "train_loss_bc": 0.023970989510416985, "train_loss_llm": 0.0, "grad_norm": 0.037563953548669815, "global_step": 34445, "epoch": 956, "lr": 0.001} +{"train_loss": 0.03331700339913368, "train_loss_bc": 0.03331700339913368, "train_loss_llm": 0.0, "grad_norm": 0.05905310437083244, "global_step": 34446, "epoch": 956, "lr": 0.001} +{"train_loss": 0.026625635102391243, "train_loss_bc": 0.026620138436555862, "train_loss_llm": 0.0005495846271514893, "grad_norm": 0.08189032226800919, "global_step": 34447, "epoch": 956, "lr": 0.001} +{"train_loss": 0.027130961418151855, "train_loss_bc": 0.027119383215904236, "train_loss_llm": 0.0011577904224395752, "grad_norm": 0.10242079198360443, "global_step": 34448, "epoch": 956, "lr": 0.001} +{"train_loss": 0.028491199016571045, "train_loss_bc": 0.028491199016571045, "train_loss_llm": 0.0, "grad_norm": 0.03398065268993378, "global_step": 34449, "epoch": 956, "lr": 0.001} +{"train_loss": 0.03239902853965759, "train_loss_bc": 0.03239902853965759, "train_loss_llm": 0.0, "grad_norm": 0.06339161098003387, "global_step": 34450, "epoch": 956, "lr": 0.001} +{"train_loss": 0.028755314285970397, "train_loss_bc": 0.024683060124516487, "train_loss_llm": 0.0, "grad_norm": 0.08870479464530945, "global_step": 34451, "epoch": 956, "lr": 0.001, "val_loss": 0.10848797112703323} +{"train_loss": 0.027085455134510994, "train_loss_bc": 0.027077587321400642, "train_loss_llm": 0.0007868409156799316, "grad_norm": 0.12470785528421402, "global_step": 34452, "epoch": 957, "lr": 0.001} +{"train_loss": 0.026099853217601776, "train_loss_bc": 0.026098741218447685, "train_loss_llm": 0.00011113286018371582, "grad_norm": 0.04423960670828819, "global_step": 34453, "epoch": 957, "lr": 0.001} +{"train_loss": 0.026811905205249786, "train_loss_bc": 0.026811905205249786, "train_loss_llm": 0.0, "grad_norm": 0.0646485909819603, "global_step": 34454, "epoch": 957, "lr": 0.001} +{"train_loss": 0.034060172736644745, "train_loss_bc": 0.034060172736644745, "train_loss_llm": 0.0, "grad_norm": 0.09219050407409668, "global_step": 34455, "epoch": 957, "lr": 0.001} +{"train_loss": 0.03140627220273018, "train_loss_bc": 0.03140627220273018, "train_loss_llm": 0.0, "grad_norm": 0.10654719918966293, "global_step": 34456, "epoch": 957, "lr": 0.001} +{"train_loss": 0.0287522803992033, "train_loss_bc": 0.028731491416692734, "train_loss_llm": 0.002078831195831299, "grad_norm": 0.028179163113236427, "global_step": 34457, "epoch": 957, "lr": 0.001} +{"train_loss": 0.028556829318404198, "train_loss_bc": 0.028556829318404198, "train_loss_llm": 0.0, "grad_norm": 0.05023263394832611, "global_step": 34458, "epoch": 957, "lr": 0.001} +{"train_loss": 0.026837937533855438, "train_loss_bc": 0.026837937533855438, "train_loss_llm": 0.0, "grad_norm": 0.06965569406747818, "global_step": 34459, "epoch": 957, "lr": 0.001} +{"train_loss": 0.026371408253908157, "train_loss_bc": 0.026371408253908157, "train_loss_llm": 0.0, "grad_norm": 0.08182947337627411, "global_step": 34460, "epoch": 957, "lr": 0.001} +{"train_loss": 0.02978000044822693, "train_loss_bc": 0.02978000044822693, "train_loss_llm": 0.0, "grad_norm": 0.04453280568122864, "global_step": 34461, "epoch": 957, "lr": 0.001} +{"train_loss": 0.03452744707465172, "train_loss_bc": 0.03452744707465172, "train_loss_llm": 0.0, "grad_norm": 0.05850333720445633, "global_step": 34462, "epoch": 957, "lr": 0.001} +{"train_loss": 0.031498249620199203, "train_loss_bc": 0.031498249620199203, "train_loss_llm": 0.0, "grad_norm": 0.0853324830532074, "global_step": 34463, "epoch": 957, "lr": 0.001} +{"train_loss": 0.028689533472061157, "train_loss_bc": 0.028689533472061157, "train_loss_llm": 0.0, "grad_norm": 0.10363411158323288, "global_step": 34464, "epoch": 957, "lr": 0.001} +{"train_loss": 0.027059616521000862, "train_loss_bc": 0.027059616521000862, "train_loss_llm": 0.0, "grad_norm": 0.038703545928001404, "global_step": 34465, "epoch": 957, "lr": 0.001} +{"train_loss": 0.027477575466036797, "train_loss_bc": 0.027477575466036797, "train_loss_llm": 0.0, "grad_norm": 0.059874359518289566, "global_step": 34466, "epoch": 957, "lr": 0.001} +{"train_loss": 0.03310307115316391, "train_loss_bc": 0.03310307115316391, "train_loss_llm": 0.0, "grad_norm": 0.08934547007083893, "global_step": 34467, "epoch": 957, "lr": 0.001} +{"train_loss": 0.030379900708794594, "train_loss_bc": 0.030379900708794594, "train_loss_llm": 0.0, "grad_norm": 0.09976640343666077, "global_step": 34468, "epoch": 957, "lr": 0.001} +{"train_loss": 0.028659014031291008, "train_loss_bc": 0.028659014031291008, "train_loss_llm": 0.0, "grad_norm": 0.03491232916712761, "global_step": 34469, "epoch": 957, "lr": 0.001} +{"train_loss": 0.03184068948030472, "train_loss_bc": 0.03184068948030472, "train_loss_llm": 0.0, "grad_norm": 0.07293701171875, "global_step": 34470, "epoch": 957, "lr": 0.001} +{"train_loss": 0.026472482830286026, "train_loss_bc": 0.026472482830286026, "train_loss_llm": 0.0, "grad_norm": 0.09487690031528473, "global_step": 34471, "epoch": 957, "lr": 0.001} +{"train_loss": 0.029172003269195557, "train_loss_bc": 0.029172003269195557, "train_loss_llm": 0.0, "grad_norm": 0.11335283517837524, "global_step": 34472, "epoch": 957, "lr": 0.001} +{"train_loss": 0.029170140624046326, "train_loss_bc": 0.029170140624046326, "train_loss_llm": 0.0, "grad_norm": 0.04133247584104538, "global_step": 34473, "epoch": 957, "lr": 0.001} +{"train_loss": 0.03426004573702812, "train_loss_bc": 0.03426004573702812, "train_loss_llm": 0.0, "grad_norm": 0.06686752289533615, "global_step": 34474, "epoch": 957, "lr": 0.001} +{"train_loss": 0.024657471105456352, "train_loss_bc": 0.024630511179566383, "train_loss_llm": 0.002696007490158081, "grad_norm": 0.08587200939655304, "global_step": 34475, "epoch": 957, "lr": 0.001} +{"train_loss": 0.021581748500466347, "train_loss_bc": 0.021573981270194054, "train_loss_llm": 0.0007766783237457275, "grad_norm": 0.10240478813648224, "global_step": 34476, "epoch": 957, "lr": 0.001} +{"train_loss": 0.028759131208062172, "train_loss_bc": 0.028759131208062172, "train_loss_llm": 0.0, "grad_norm": 0.03146819770336151, "global_step": 34477, "epoch": 957, "lr": 0.001} +{"train_loss": 0.026305751875042915, "train_loss_bc": 0.026305751875042915, "train_loss_llm": 0.0, "grad_norm": 0.05762757733464241, "global_step": 34478, "epoch": 957, "lr": 0.001} +{"train_loss": 0.030789528042078018, "train_loss_bc": 0.0307781882584095, "train_loss_llm": 0.0011339783668518066, "grad_norm": 0.08626774698495865, "global_step": 34479, "epoch": 957, "lr": 0.001} +{"train_loss": 0.02936098352074623, "train_loss_bc": 0.02936098352074623, "train_loss_llm": 0.0, "grad_norm": 0.11214260011911392, "global_step": 34480, "epoch": 957, "lr": 0.001} +{"train_loss": 0.02779700979590416, "train_loss_bc": 0.02779700979590416, "train_loss_llm": 0.0, "grad_norm": 0.03567590191960335, "global_step": 34481, "epoch": 957, "lr": 0.001} +{"train_loss": 0.026721125468611717, "train_loss_bc": 0.026721125468611717, "train_loss_llm": 0.0, "grad_norm": 0.05135621502995491, "global_step": 34482, "epoch": 957, "lr": 0.001} +{"train_loss": 0.029348747804760933, "train_loss_bc": 0.029348747804760933, "train_loss_llm": 0.0, "grad_norm": 0.0717434287071228, "global_step": 34483, "epoch": 957, "lr": 0.001} +{"train_loss": 0.030606482177972794, "train_loss_bc": 0.03059893101453781, "train_loss_llm": 0.0007551014423370361, "grad_norm": 0.09455640614032745, "global_step": 34484, "epoch": 957, "lr": 0.001} +{"train_loss": 0.026928424835205078, "train_loss_bc": 0.026928424835205078, "train_loss_llm": 0.0, "grad_norm": 0.04169309884309769, "global_step": 34485, "epoch": 957, "lr": 0.001} +{"train_loss": 0.029959172010421753, "train_loss_bc": 0.029959172010421753, "train_loss_llm": 0.0, "grad_norm": 0.08022421598434448, "global_step": 34486, "epoch": 957, "lr": 0.001} +{"train_loss": 0.0288116246358388, "train_loss_bc": 0.02633102610707283, "train_loss_llm": 0.0, "grad_norm": 0.0927153080701828, "global_step": 34487, "epoch": 957, "lr": 0.001, "val_loss": 0.10744652152061462} +{"train_loss": 0.030966371297836304, "train_loss_bc": 0.030966371297836304, "train_loss_llm": 0.0, "grad_norm": 0.10533744841814041, "global_step": 34488, "epoch": 958, "lr": 0.001} +{"train_loss": 0.0263841412961483, "train_loss_bc": 0.026367543265223503, "train_loss_llm": 0.0016597211360931396, "grad_norm": 0.032979220151901245, "global_step": 34489, "epoch": 958, "lr": 0.001} +{"train_loss": 0.032023265957832336, "train_loss_bc": 0.032023265957832336, "train_loss_llm": 0.0, "grad_norm": 0.049915194511413574, "global_step": 34490, "epoch": 958, "lr": 0.001} +{"train_loss": 0.02475251816213131, "train_loss_bc": 0.02475251816213131, "train_loss_llm": 0.0, "grad_norm": 0.06610307097434998, "global_step": 34491, "epoch": 958, "lr": 0.001} +{"train_loss": 0.03101891465485096, "train_loss_bc": 0.03101891465485096, "train_loss_llm": 0.0, "grad_norm": 0.08585256338119507, "global_step": 34492, "epoch": 958, "lr": 0.001} +{"train_loss": 0.031009623780846596, "train_loss_bc": 0.031009623780846596, "train_loss_llm": 0.0, "grad_norm": 0.04272175952792168, "global_step": 34493, "epoch": 958, "lr": 0.001} +{"train_loss": 0.03189832344651222, "train_loss_bc": 0.03189832344651222, "train_loss_llm": 0.0, "grad_norm": 0.07003585249185562, "global_step": 34494, "epoch": 958, "lr": 0.001} +{"train_loss": 0.029438018798828125, "train_loss_bc": 0.029426824301481247, "train_loss_llm": 0.001119375228881836, "grad_norm": 0.10113643109798431, "global_step": 34495, "epoch": 958, "lr": 0.001} +{"train_loss": 0.02941666729748249, "train_loss_bc": 0.02941666729748249, "train_loss_llm": 0.0, "grad_norm": 0.12871307134628296, "global_step": 34496, "epoch": 958, "lr": 0.001} +{"train_loss": 0.031001709401607513, "train_loss_bc": 0.03099467232823372, "train_loss_llm": 0.0007037818431854248, "grad_norm": 0.04384268447756767, "global_step": 34497, "epoch": 958, "lr": 0.001} +{"train_loss": 0.027044879272580147, "train_loss_bc": 0.027035897597670555, "train_loss_llm": 0.0008980929851531982, "grad_norm": 0.06010491028428078, "global_step": 34498, "epoch": 958, "lr": 0.001} +{"train_loss": 0.03021823987364769, "train_loss_bc": 0.030184632167220116, "train_loss_llm": 0.0033606886863708496, "grad_norm": 0.08183298259973526, "global_step": 34499, "epoch": 958, "lr": 0.001} +{"train_loss": 0.03682367876172066, "train_loss_bc": 0.03682367876172066, "train_loss_llm": 0.0, "grad_norm": 0.09876515716314316, "global_step": 34500, "epoch": 958, "lr": 0.001} +{"train_loss": 0.02552822045981884, "train_loss_bc": 0.02552822045981884, "train_loss_llm": 0.0, "grad_norm": 0.030674871057271957, "global_step": 34501, "epoch": 958, "lr": 0.001} +{"train_loss": 0.02617049403488636, "train_loss_bc": 0.02617049403488636, "train_loss_llm": 0.0, "grad_norm": 0.04728657752275467, "global_step": 34502, "epoch": 958, "lr": 0.001} +{"train_loss": 0.0302606038749218, "train_loss_bc": 0.0302606038749218, "train_loss_llm": 0.0, "grad_norm": 0.07402141392230988, "global_step": 34503, "epoch": 958, "lr": 0.001} +{"train_loss": 0.024486737325787544, "train_loss_bc": 0.024486737325787544, "train_loss_llm": 0.0, "grad_norm": 0.08805316686630249, "global_step": 34504, "epoch": 958, "lr": 0.001} +{"train_loss": 0.025144126266241074, "train_loss_bc": 0.025144126266241074, "train_loss_llm": 0.0, "grad_norm": 0.03309984505176544, "global_step": 34505, "epoch": 958, "lr": 0.001} +{"train_loss": 0.034275006502866745, "train_loss_bc": 0.034275006502866745, "train_loss_llm": 0.0, "grad_norm": 0.058803047984838486, "global_step": 34506, "epoch": 958, "lr": 0.001} +{"train_loss": 0.02927231229841709, "train_loss_bc": 0.02927231229841709, "train_loss_llm": 0.0, "grad_norm": 0.07502526044845581, "global_step": 34507, "epoch": 958, "lr": 0.001} +{"train_loss": 0.027323797345161438, "train_loss_bc": 0.027323797345161438, "train_loss_llm": 0.0, "grad_norm": 0.09466562420129776, "global_step": 34508, "epoch": 958, "lr": 0.001} +{"train_loss": 0.026081768795847893, "train_loss_bc": 0.026081768795847893, "train_loss_llm": 0.0, "grad_norm": 0.045781008899211884, "global_step": 34509, "epoch": 958, "lr": 0.001} +{"train_loss": 0.027665169909596443, "train_loss_bc": 0.027665169909596443, "train_loss_llm": 0.0, "grad_norm": 0.061621665954589844, "global_step": 34510, "epoch": 958, "lr": 0.001} +{"train_loss": 0.024872157722711563, "train_loss_bc": 0.024872157722711563, "train_loss_llm": 0.0, "grad_norm": 0.08003165572881699, "global_step": 34511, "epoch": 958, "lr": 0.001} +{"train_loss": 0.024945244193077087, "train_loss_bc": 0.024931499734520912, "train_loss_llm": 0.0013743937015533447, "grad_norm": 0.09183026105165482, "global_step": 34512, "epoch": 958, "lr": 0.001} +{"train_loss": 0.022725103422999382, "train_loss_bc": 0.022725103422999382, "train_loss_llm": 0.0, "grad_norm": 0.03878882899880409, "global_step": 34513, "epoch": 958, "lr": 0.001} +{"train_loss": 0.03175825998187065, "train_loss_bc": 0.03175825998187065, "train_loss_llm": 0.0, "grad_norm": 0.06523959338665009, "global_step": 34514, "epoch": 958, "lr": 0.001} +{"train_loss": 0.032843660563230515, "train_loss_bc": 0.032843660563230515, "train_loss_llm": 0.0, "grad_norm": 0.08545590192079544, "global_step": 34515, "epoch": 958, "lr": 0.001} +{"train_loss": 0.033547043800354004, "train_loss_bc": 0.033547043800354004, "train_loss_llm": 0.0, "grad_norm": 0.10904912650585175, "global_step": 34516, "epoch": 958, "lr": 0.001} +{"train_loss": 0.028028005734086037, "train_loss_bc": 0.028023287653923035, "train_loss_llm": 0.0004718303680419922, "grad_norm": 0.04094404727220535, "global_step": 34517, "epoch": 958, "lr": 0.001} +{"train_loss": 0.024642813950777054, "train_loss_bc": 0.024624837562441826, "train_loss_llm": 0.0017975866794586182, "grad_norm": 0.06385967135429382, "global_step": 34518, "epoch": 958, "lr": 0.001} +{"train_loss": 0.02863973006606102, "train_loss_bc": 0.028636323288083076, "train_loss_llm": 0.0003406107425689697, "grad_norm": 0.10207465291023254, "global_step": 34519, "epoch": 958, "lr": 0.001} +{"train_loss": 0.029851723462343216, "train_loss_bc": 0.029851723462343216, "train_loss_llm": 0.0, "grad_norm": 0.11792709678411484, "global_step": 34520, "epoch": 958, "lr": 0.001} +{"train_loss": 0.02813199907541275, "train_loss_bc": 0.028126658871769905, "train_loss_llm": 0.0005339980125427246, "grad_norm": 0.03618817403912544, "global_step": 34521, "epoch": 958, "lr": 0.001} +{"train_loss": 0.02878786437213421, "train_loss_bc": 0.02878747135400772, "train_loss_llm": 3.927946090698242e-05, "grad_norm": 0.06753002852201462, "global_step": 34522, "epoch": 958, "lr": 0.001} +{"train_loss": 0.028616541634417243, "train_loss_bc": 0.023217303678393364, "train_loss_llm": 0.0, "grad_norm": 0.07874050736427307, "global_step": 34523, "epoch": 958, "lr": 0.001, "val_loss": 0.10892623662948608} +{"train_loss": 0.02549663931131363, "train_loss_bc": 0.02549663931131363, "train_loss_llm": 0.0, "grad_norm": 0.09956203401088715, "global_step": 34524, "epoch": 959, "lr": 0.001} +{"train_loss": 0.028881696984171867, "train_loss_bc": 0.028881696984171867, "train_loss_llm": 0.0, "grad_norm": 0.03697454556822777, "global_step": 34525, "epoch": 959, "lr": 0.001} +{"train_loss": 0.02923356182873249, "train_loss_bc": 0.02923356182873249, "train_loss_llm": 0.0, "grad_norm": 0.05728231370449066, "global_step": 34526, "epoch": 959, "lr": 0.001} +{"train_loss": 0.027279039844870567, "train_loss_bc": 0.027279039844870567, "train_loss_llm": 0.0, "grad_norm": 0.07503661513328552, "global_step": 34527, "epoch": 959, "lr": 0.001} +{"train_loss": 0.028629256412386894, "train_loss_bc": 0.028629256412386894, "train_loss_llm": 0.0, "grad_norm": 0.10174603015184402, "global_step": 34528, "epoch": 959, "lr": 0.001} +{"train_loss": 0.03232555091381073, "train_loss_bc": 0.03232555091381073, "train_loss_llm": 0.0, "grad_norm": 0.04044722765684128, "global_step": 34529, "epoch": 959, "lr": 0.001} +{"train_loss": 0.028696006163954735, "train_loss_bc": 0.028696006163954735, "train_loss_llm": 0.0, "grad_norm": 0.0673379972577095, "global_step": 34530, "epoch": 959, "lr": 0.001} +{"train_loss": 0.03016146644949913, "train_loss_bc": 0.03015349805355072, "train_loss_llm": 0.0007968246936798096, "grad_norm": 0.09035249054431915, "global_step": 34531, "epoch": 959, "lr": 0.001} +{"train_loss": 0.029938964173197746, "train_loss_bc": 0.029938964173197746, "train_loss_llm": 0.0, "grad_norm": 0.10993543267250061, "global_step": 34532, "epoch": 959, "lr": 0.001} +{"train_loss": 0.03269105404615402, "train_loss_bc": 0.03269105404615402, "train_loss_llm": 0.0, "grad_norm": 0.043597348034381866, "global_step": 34533, "epoch": 959, "lr": 0.001} +{"train_loss": 0.021414868533611298, "train_loss_bc": 0.021414868533611298, "train_loss_llm": 0.0, "grad_norm": 0.07103502750396729, "global_step": 34534, "epoch": 959, "lr": 0.001} +{"train_loss": 0.028101643547415733, "train_loss_bc": 0.028101643547415733, "train_loss_llm": 0.0, "grad_norm": 0.08493337035179138, "global_step": 34535, "epoch": 959, "lr": 0.001} +{"train_loss": 0.02804948203265667, "train_loss_bc": 0.02804948203265667, "train_loss_llm": 0.0, "grad_norm": 0.11904194205999374, "global_step": 34536, "epoch": 959, "lr": 0.001} +{"train_loss": 0.034034363925457, "train_loss_bc": 0.034034363925457, "train_loss_llm": 0.0, "grad_norm": 0.0408037044107914, "global_step": 34537, "epoch": 959, "lr": 0.001} +{"train_loss": 0.028730489313602448, "train_loss_bc": 0.028730489313602448, "train_loss_llm": 0.0, "grad_norm": 0.06458310037851334, "global_step": 34538, "epoch": 959, "lr": 0.001} +{"train_loss": 0.022543128579854965, "train_loss_bc": 0.022543128579854965, "train_loss_llm": 0.0, "grad_norm": 0.08046935498714447, "global_step": 34539, "epoch": 959, "lr": 0.001} +{"train_loss": 0.03205302357673645, "train_loss_bc": 0.03205302357673645, "train_loss_llm": 0.0, "grad_norm": 0.10097731649875641, "global_step": 34540, "epoch": 959, "lr": 0.001} +{"train_loss": 0.02630803734064102, "train_loss_bc": 0.02630803734064102, "train_loss_llm": 0.0, "grad_norm": 0.03386123850941658, "global_step": 34541, "epoch": 959, "lr": 0.001} +{"train_loss": 0.029220720753073692, "train_loss_bc": 0.029220720753073692, "train_loss_llm": 0.0, "grad_norm": 0.04422910138964653, "global_step": 34542, "epoch": 959, "lr": 0.001} +{"train_loss": 0.022125041112303734, "train_loss_bc": 0.022125041112303734, "train_loss_llm": 0.0, "grad_norm": 0.05014089122414589, "global_step": 34543, "epoch": 959, "lr": 0.001} +{"train_loss": 0.03034338355064392, "train_loss_bc": 0.03034338355064392, "train_loss_llm": 0.0, "grad_norm": 0.06778927147388458, "global_step": 34544, "epoch": 959, "lr": 0.001} +{"train_loss": 0.02775692753493786, "train_loss_bc": 0.02775692753493786, "train_loss_llm": 0.0, "grad_norm": 0.0434674471616745, "global_step": 34545, "epoch": 959, "lr": 0.001} +{"train_loss": 0.03057502582669258, "train_loss_bc": 0.03057502582669258, "train_loss_llm": 0.0, "grad_norm": 0.06302481889724731, "global_step": 34546, "epoch": 959, "lr": 0.001} +{"train_loss": 0.03355250880122185, "train_loss_bc": 0.03354287147521973, "train_loss_llm": 0.0009637773036956787, "grad_norm": 0.09826046973466873, "global_step": 34547, "epoch": 959, "lr": 0.001} +{"train_loss": 0.030360836535692215, "train_loss_bc": 0.030360836535692215, "train_loss_llm": 0.0, "grad_norm": 0.11314643919467926, "global_step": 34548, "epoch": 959, "lr": 0.001} +{"train_loss": 0.029025813564658165, "train_loss_bc": 0.029025813564658165, "train_loss_llm": 0.0, "grad_norm": 0.043073125183582306, "global_step": 34549, "epoch": 959, "lr": 0.001} +{"train_loss": 0.025073548778891563, "train_loss_bc": 0.025073548778891563, "train_loss_llm": 0.0, "grad_norm": 0.05916445702314377, "global_step": 34550, "epoch": 959, "lr": 0.001} +{"train_loss": 0.029741613194346428, "train_loss_bc": 0.02973909117281437, "train_loss_llm": 0.00025221705436706543, "grad_norm": 0.08266205340623856, "global_step": 34551, "epoch": 959, "lr": 0.001} +{"train_loss": 0.029248150065541267, "train_loss_bc": 0.029248150065541267, "train_loss_llm": 0.0, "grad_norm": 0.10159515589475632, "global_step": 34552, "epoch": 959, "lr": 0.001} +{"train_loss": 0.029805125668644905, "train_loss_bc": 0.029805125668644905, "train_loss_llm": 0.0, "grad_norm": 0.03950582817196846, "global_step": 34553, "epoch": 959, "lr": 0.001} +{"train_loss": 0.02593112364411354, "train_loss_bc": 0.02592081017792225, "train_loss_llm": 0.0010313689708709717, "grad_norm": 0.056317947804927826, "global_step": 34554, "epoch": 959, "lr": 0.001} +{"train_loss": 0.033320311456918716, "train_loss_bc": 0.033320311456918716, "train_loss_llm": 0.0, "grad_norm": 0.08423469960689545, "global_step": 34555, "epoch": 959, "lr": 0.001} +{"train_loss": 0.025867247954010963, "train_loss_bc": 0.025864096358418465, "train_loss_llm": 0.00031515955924987793, "grad_norm": 0.09800846129655838, "global_step": 34556, "epoch": 959, "lr": 0.001} +{"train_loss": 0.027779169380664825, "train_loss_bc": 0.02777373231947422, "train_loss_llm": 0.0005437135696411133, "grad_norm": 0.036460794508457184, "global_step": 34557, "epoch": 959, "lr": 0.001} +{"train_loss": 0.028181759640574455, "train_loss_bc": 0.0281605813652277, "train_loss_llm": 0.002117753028869629, "grad_norm": 0.06543945521116257, "global_step": 34558, "epoch": 959, "lr": 0.001} +{"train_loss": 0.028620359539571736, "train_loss_bc": 0.027856362983584404, "train_loss_llm": 0.0, "grad_norm": 0.07207193970680237, "global_step": 34559, "epoch": 959, "lr": 0.001, "val_loss": 0.10857106745243073} +{"train_loss": 0.028632689267396927, "train_loss_bc": 0.028632689267396927, "train_loss_llm": 0.0, "grad_norm": 0.09172116965055466, "global_step": 34560, "epoch": 960, "lr": 0.001} +{"train_loss": 0.028081299737095833, "train_loss_bc": 0.028081299737095833, "train_loss_llm": 0.0, "grad_norm": 0.033158112317323685, "global_step": 34561, "epoch": 960, "lr": 0.001} +{"train_loss": 0.026873106136918068, "train_loss_bc": 0.026872258633375168, "train_loss_llm": 8.466839790344238e-05, "grad_norm": 0.05476366728544235, "global_step": 34562, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02279784344136715, "train_loss_bc": 0.02279784344136715, "train_loss_llm": 0.0, "grad_norm": 0.07187548279762268, "global_step": 34563, "epoch": 960, "lr": 0.001} +{"train_loss": 0.024066191166639328, "train_loss_bc": 0.024066191166639328, "train_loss_llm": 0.0, "grad_norm": 0.08468739688396454, "global_step": 34564, "epoch": 960, "lr": 0.001} +{"train_loss": 0.034318964928388596, "train_loss_bc": 0.034318964928388596, "train_loss_llm": 0.0, "grad_norm": 0.044291771948337555, "global_step": 34565, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02953336387872696, "train_loss_bc": 0.02953336387872696, "train_loss_llm": 0.0, "grad_norm": 0.06526336818933487, "global_step": 34566, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02983129397034645, "train_loss_bc": 0.02982242777943611, "train_loss_llm": 0.0008866190910339355, "grad_norm": 0.08459490537643433, "global_step": 34567, "epoch": 960, "lr": 0.001} +{"train_loss": 0.028100647032260895, "train_loss_bc": 0.028100647032260895, "train_loss_llm": 0.0, "grad_norm": 0.10634565353393555, "global_step": 34568, "epoch": 960, "lr": 0.001} +{"train_loss": 0.032776057720184326, "train_loss_bc": 0.0327669195830822, "train_loss_llm": 0.0009136795997619629, "grad_norm": 0.04554710164666176, "global_step": 34569, "epoch": 960, "lr": 0.001} +{"train_loss": 0.025844762101769447, "train_loss_bc": 0.025844762101769447, "train_loss_llm": 0.0, "grad_norm": 0.06573719531297684, "global_step": 34570, "epoch": 960, "lr": 0.001} +{"train_loss": 0.028070960193872452, "train_loss_bc": 0.028070960193872452, "train_loss_llm": 0.0, "grad_norm": 0.08394377678632736, "global_step": 34571, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02768855169415474, "train_loss_bc": 0.02768855169415474, "train_loss_llm": 0.0, "grad_norm": 0.10339049249887466, "global_step": 34572, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02870722860097885, "train_loss_bc": 0.02869527041912079, "train_loss_llm": 0.0011958777904510498, "grad_norm": 0.039824049919843674, "global_step": 34573, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02760152518749237, "train_loss_bc": 0.02760152518749237, "train_loss_llm": 0.0, "grad_norm": 0.07655917853116989, "global_step": 34574, "epoch": 960, "lr": 0.001} +{"train_loss": 0.024986987933516502, "train_loss_bc": 0.024986987933516502, "train_loss_llm": 0.0, "grad_norm": 0.09043236076831818, "global_step": 34575, "epoch": 960, "lr": 0.001} +{"train_loss": 0.030248038470745087, "train_loss_bc": 0.030216321349143982, "train_loss_llm": 0.003171771764755249, "grad_norm": 0.11649879813194275, "global_step": 34576, "epoch": 960, "lr": 0.001} +{"train_loss": 0.03270192816853523, "train_loss_bc": 0.03269459679722786, "train_loss_llm": 0.0007331371307373047, "grad_norm": 0.042243700474500656, "global_step": 34577, "epoch": 960, "lr": 0.001} +{"train_loss": 0.03746308013796806, "train_loss_bc": 0.03746308013796806, "train_loss_llm": 0.0, "grad_norm": 0.07263725250959396, "global_step": 34578, "epoch": 960, "lr": 0.001} +{"train_loss": 0.036896634846925735, "train_loss_bc": 0.036896634846925735, "train_loss_llm": 0.0, "grad_norm": 0.10063357651233673, "global_step": 34579, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02825463004410267, "train_loss_bc": 0.02825463004410267, "train_loss_llm": 0.0, "grad_norm": 0.1217643991112709, "global_step": 34580, "epoch": 960, "lr": 0.001} +{"train_loss": 0.029367225244641304, "train_loss_bc": 0.029367225244641304, "train_loss_llm": 0.0, "grad_norm": 0.04296015202999115, "global_step": 34581, "epoch": 960, "lr": 0.001} +{"train_loss": 0.024042941629886627, "train_loss_bc": 0.024042941629886627, "train_loss_llm": 0.0, "grad_norm": 0.06759151816368103, "global_step": 34582, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02468668669462204, "train_loss_bc": 0.02468668669462204, "train_loss_llm": 0.0, "grad_norm": 0.0813174843788147, "global_step": 34583, "epoch": 960, "lr": 0.001} +{"train_loss": 0.032262858003377914, "train_loss_bc": 0.032262858003377914, "train_loss_llm": 0.0, "grad_norm": 0.09882116317749023, "global_step": 34584, "epoch": 960, "lr": 0.001} +{"train_loss": 0.027620216831564903, "train_loss_bc": 0.027603931725025177, "train_loss_llm": 0.0016285479068756104, "grad_norm": 0.03509316220879555, "global_step": 34585, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02899918705224991, "train_loss_bc": 0.02899918705224991, "train_loss_llm": 0.0, "grad_norm": 0.05504243075847626, "global_step": 34586, "epoch": 960, "lr": 0.001} +{"train_loss": 0.032512348145246506, "train_loss_bc": 0.032506298273801804, "train_loss_llm": 0.000605165958404541, "grad_norm": 0.07343067973852158, "global_step": 34587, "epoch": 960, "lr": 0.001} +{"train_loss": 0.026193607598543167, "train_loss_bc": 0.026193607598543167, "train_loss_llm": 0.0, "grad_norm": 0.0951184555888176, "global_step": 34588, "epoch": 960, "lr": 0.001} +{"train_loss": 0.024045495316386223, "train_loss_bc": 0.024045495316386223, "train_loss_llm": 0.0, "grad_norm": 0.03216450288891792, "global_step": 34589, "epoch": 960, "lr": 0.001} +{"train_loss": 0.027362577617168427, "train_loss_bc": 0.027362577617168427, "train_loss_llm": 0.0, "grad_norm": 0.05835740640759468, "global_step": 34590, "epoch": 960, "lr": 0.001} +{"train_loss": 0.03011060319840908, "train_loss_bc": 0.03011060319840908, "train_loss_llm": 0.0, "grad_norm": 0.08479973673820496, "global_step": 34591, "epoch": 960, "lr": 0.001} +{"train_loss": 0.028506582602858543, "train_loss_bc": 0.028501130640506744, "train_loss_llm": 0.0005452632904052734, "grad_norm": 0.11261538416147232, "global_step": 34592, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02933000959455967, "train_loss_bc": 0.02933000959455967, "train_loss_llm": 0.0, "grad_norm": 0.04410901665687561, "global_step": 34593, "epoch": 960, "lr": 0.001} +{"train_loss": 0.028461193665862083, "train_loss_bc": 0.0284546110779047, "train_loss_llm": 0.000658184289932251, "grad_norm": 0.08014462888240814, "global_step": 34594, "epoch": 960, "lr": 0.001} +{"train_loss": 0.02873240892464916, "train_loss_bc": 0.02738940343260765, "train_loss_llm": 0.0, "grad_norm": 0.11902565509080887, "global_step": 34595, "epoch": 960, "lr": 0.001, "train/cumulative_reward": 73.60781131182401, "train/mean_score": 15.82911897985592, "train/success_rate": 0.4, "test/cumulative_reward": 32.603975133465376, "test/mean_score": 7.6100750383588665, "test/success_rate": 0.18, "val_loss": 0.10983090102672577, "train_action_mse_error": 0.03196178376674652} +{"train_loss": 0.03265568986535072, "train_loss_bc": 0.03265568986535072, "train_loss_llm": 0.0, "grad_norm": 0.1473674774169922, "global_step": 34596, "epoch": 961, "lr": 0.001} +{"train_loss": 0.026640936732292175, "train_loss_bc": 0.02663206309080124, "train_loss_llm": 0.0008874237537384033, "grad_norm": 0.033192411065101624, "global_step": 34597, "epoch": 961, "lr": 0.001} +{"train_loss": 0.028437510132789612, "train_loss_bc": 0.028432589024305344, "train_loss_llm": 0.000492095947265625, "grad_norm": 0.06374496221542358, "global_step": 34598, "epoch": 961, "lr": 0.001} +{"train_loss": 0.024776363745331764, "train_loss_bc": 0.024776363745331764, "train_loss_llm": 0.0, "grad_norm": 0.0800151526927948, "global_step": 34599, "epoch": 961, "lr": 0.001} +{"train_loss": 0.029262421652674675, "train_loss_bc": 0.029262421652674675, "train_loss_llm": 0.0, "grad_norm": 0.10252799093723297, "global_step": 34600, "epoch": 961, "lr": 0.001} +{"train_loss": 0.03116818517446518, "train_loss_bc": 0.031159240752458572, "train_loss_llm": 0.0008944869041442871, "grad_norm": 0.04528328776359558, "global_step": 34601, "epoch": 961, "lr": 0.001} +{"train_loss": 0.028029775246977806, "train_loss_bc": 0.028015054762363434, "train_loss_llm": 0.0014719963073730469, "grad_norm": 0.07513394951820374, "global_step": 34602, "epoch": 961, "lr": 0.001} +{"train_loss": 0.03304450586438179, "train_loss_bc": 0.03304450586438179, "train_loss_llm": 0.0, "grad_norm": 0.10119885951280594, "global_step": 34603, "epoch": 961, "lr": 0.001} +{"train_loss": 0.026795150712132454, "train_loss_bc": 0.02678983472287655, "train_loss_llm": 0.0005316436290740967, "grad_norm": 0.1254899799823761, "global_step": 34604, "epoch": 961, "lr": 0.001} +{"train_loss": 0.022485172376036644, "train_loss_bc": 0.022485172376036644, "train_loss_llm": 0.0, "grad_norm": 0.030250098556280136, "global_step": 34605, "epoch": 961, "lr": 0.001} +{"train_loss": 0.024250702932476997, "train_loss_bc": 0.024250702932476997, "train_loss_llm": 0.0, "grad_norm": 0.05388052761554718, "global_step": 34606, "epoch": 961, "lr": 0.001} +{"train_loss": 0.032539092004299164, "train_loss_bc": 0.032539092004299164, "train_loss_llm": 0.0, "grad_norm": 0.08198956400156021, "global_step": 34607, "epoch": 961, "lr": 0.001} +{"train_loss": 0.02646668627858162, "train_loss_bc": 0.026445429772138596, "train_loss_llm": 0.0021255910396575928, "grad_norm": 0.10659237205982208, "global_step": 34608, "epoch": 961, "lr": 0.001} +{"train_loss": 0.028041686862707138, "train_loss_bc": 0.02803371101617813, "train_loss_llm": 0.0007975101470947266, "grad_norm": 0.043479397892951965, "global_step": 34609, "epoch": 961, "lr": 0.001} +{"train_loss": 0.025325240567326546, "train_loss_bc": 0.025325240567326546, "train_loss_llm": 0.0, "grad_norm": 0.06662651151418686, "global_step": 34610, "epoch": 961, "lr": 0.001} +{"train_loss": 0.03243381902575493, "train_loss_bc": 0.03238203376531601, "train_loss_llm": 0.005178511142730713, "grad_norm": 0.09500403702259064, "global_step": 34611, "epoch": 961, "lr": 0.001} +{"train_loss": 0.028963332995772362, "train_loss_bc": 0.028963332995772362, "train_loss_llm": 0.0, "grad_norm": 0.11681400239467621, "global_step": 34612, "epoch": 961, "lr": 0.001} +{"train_loss": 0.03101453185081482, "train_loss_bc": 0.03101453185081482, "train_loss_llm": 0.0, "grad_norm": 0.039864927530288696, "global_step": 34613, "epoch": 961, "lr": 0.001} +{"train_loss": 0.02684260532259941, "train_loss_bc": 0.02684260532259941, "train_loss_llm": 0.0, "grad_norm": 0.06889935582876205, "global_step": 34614, "epoch": 961, "lr": 0.001} +{"train_loss": 0.029132070019841194, "train_loss_bc": 0.02913188934326172, "train_loss_llm": 1.8030405044555664e-05, "grad_norm": 0.08371705561876297, "global_step": 34615, "epoch": 961, "lr": 0.001} +{"train_loss": 0.027379244565963745, "train_loss_bc": 0.02737225778400898, "train_loss_llm": 0.0006986558437347412, "grad_norm": 0.10410360246896744, "global_step": 34616, "epoch": 961, "lr": 0.001} +{"train_loss": 0.02175593376159668, "train_loss_bc": 0.02175593376159668, "train_loss_llm": 0.0, "grad_norm": 0.03444485366344452, "global_step": 34617, "epoch": 961, "lr": 0.001} +{"train_loss": 0.029070062562823296, "train_loss_bc": 0.029049163684248924, "train_loss_llm": 0.0020898282527923584, "grad_norm": 0.05662941560149193, "global_step": 34618, "epoch": 961, "lr": 0.001} +{"train_loss": 0.026954887434840202, "train_loss_bc": 0.026954887434840202, "train_loss_llm": 0.0, "grad_norm": 0.07221812009811401, "global_step": 34619, "epoch": 961, "lr": 0.001} +{"train_loss": 0.028700079768896103, "train_loss_bc": 0.028685344383120537, "train_loss_llm": 0.0014735758304595947, "grad_norm": 0.10001517087221146, "global_step": 34620, "epoch": 961, "lr": 0.001} +{"train_loss": 0.028022877871990204, "train_loss_bc": 0.028012465685606003, "train_loss_llm": 0.0010411441326141357, "grad_norm": 0.03648686781525612, "global_step": 34621, "epoch": 961, "lr": 0.001} +{"train_loss": 0.028255345299839973, "train_loss_bc": 0.028255345299839973, "train_loss_llm": 0.0, "grad_norm": 0.06606990098953247, "global_step": 34622, "epoch": 961, "lr": 0.001} +{"train_loss": 0.03282570093870163, "train_loss_bc": 0.03282570093870163, "train_loss_llm": 0.0, "grad_norm": 0.09911224246025085, "global_step": 34623, "epoch": 961, "lr": 0.001} +{"train_loss": 0.03094937652349472, "train_loss_bc": 0.03094937652349472, "train_loss_llm": 0.0, "grad_norm": 0.12828853726387024, "global_step": 34624, "epoch": 961, "lr": 0.001} +{"train_loss": 0.029183542355895042, "train_loss_bc": 0.029183542355895042, "train_loss_llm": 0.0, "grad_norm": 0.04632771015167236, "global_step": 34625, "epoch": 961, "lr": 0.001} +{"train_loss": 0.03057948872447014, "train_loss_bc": 0.03057948872447014, "train_loss_llm": 0.0, "grad_norm": 0.08507679402828217, "global_step": 34626, "epoch": 961, "lr": 0.001} +{"train_loss": 0.03321497514843941, "train_loss_bc": 0.03321497514843941, "train_loss_llm": 0.0, "grad_norm": 0.1306341141462326, "global_step": 34627, "epoch": 961, "lr": 0.001} +{"train_loss": 0.03305730223655701, "train_loss_bc": 0.03305730223655701, "train_loss_llm": 0.0, "grad_norm": 0.1599336564540863, "global_step": 34628, "epoch": 961, "lr": 0.001} +{"train_loss": 0.03141172230243683, "train_loss_bc": 0.03141172230243683, "train_loss_llm": 0.0, "grad_norm": 0.03851866349577904, "global_step": 34629, "epoch": 961, "lr": 0.001} +{"train_loss": 0.029391082003712654, "train_loss_bc": 0.029391082003712654, "train_loss_llm": 0.0, "grad_norm": 0.06080532446503639, "global_step": 34630, "epoch": 961, "lr": 0.001} +{"train_loss": 0.028905129784511194, "train_loss_bc": 0.0315275713801384, "train_loss_llm": 0.0, "grad_norm": 0.0878806933760643, "global_step": 34631, "epoch": 961, "lr": 0.001, "val_loss": 0.10854848474264145} +{"train_loss": 0.026565637439489365, "train_loss_bc": 0.026559412479400635, "train_loss_llm": 0.0006224513053894043, "grad_norm": 0.11442526429891586, "global_step": 34632, "epoch": 962, "lr": 0.001} +{"train_loss": 0.02570885233581066, "train_loss_bc": 0.02570885233581066, "train_loss_llm": 0.0, "grad_norm": 0.03889409080147743, "global_step": 34633, "epoch": 962, "lr": 0.001} +{"train_loss": 0.03080689162015915, "train_loss_bc": 0.03080689162015915, "train_loss_llm": 0.0, "grad_norm": 0.06231687590479851, "global_step": 34634, "epoch": 962, "lr": 0.001} +{"train_loss": 0.027609147131443024, "train_loss_bc": 0.027609147131443024, "train_loss_llm": 0.0, "grad_norm": 0.08578237146139145, "global_step": 34635, "epoch": 962, "lr": 0.001} +{"train_loss": 0.025196915492415428, "train_loss_bc": 0.025196915492415428, "train_loss_llm": 0.0, "grad_norm": 0.11075736582279205, "global_step": 34636, "epoch": 962, "lr": 0.001} +{"train_loss": 0.024492161348462105, "train_loss_bc": 0.024492161348462105, "train_loss_llm": 0.0, "grad_norm": 0.03766777738928795, "global_step": 34637, "epoch": 962, "lr": 0.001} +{"train_loss": 0.02731918729841709, "train_loss_bc": 0.02729910984635353, "train_loss_llm": 0.00200769305229187, "grad_norm": 0.05698737874627113, "global_step": 34638, "epoch": 962, "lr": 0.001} +{"train_loss": 0.026083119213581085, "train_loss_bc": 0.026083119213581085, "train_loss_llm": 0.0, "grad_norm": 0.08841439336538315, "global_step": 34639, "epoch": 962, "lr": 0.001} +{"train_loss": 0.03193055838346481, "train_loss_bc": 0.03193055838346481, "train_loss_llm": 0.0, "grad_norm": 0.12426985800266266, "global_step": 34640, "epoch": 962, "lr": 0.001} +{"train_loss": 0.03153655305504799, "train_loss_bc": 0.03152966499328613, "train_loss_llm": 0.0006889700889587402, "grad_norm": 0.045956071466207504, "global_step": 34641, "epoch": 962, "lr": 0.001} +{"train_loss": 0.02544180303812027, "train_loss_bc": 0.02544180303812027, "train_loss_llm": 0.0, "grad_norm": 0.09125828742980957, "global_step": 34642, "epoch": 962, "lr": 0.001} +{"train_loss": 0.024926524609327316, "train_loss_bc": 0.02491440810263157, "train_loss_llm": 0.0012115836143493652, "grad_norm": 0.13274869322776794, "global_step": 34643, "epoch": 962, "lr": 0.001} +{"train_loss": 0.027170971035957336, "train_loss_bc": 0.027170971035957336, "train_loss_llm": 0.0, "grad_norm": 0.14884719252586365, "global_step": 34644, "epoch": 962, "lr": 0.001} +{"train_loss": 0.029110776260495186, "train_loss_bc": 0.029110776260495186, "train_loss_llm": 0.0, "grad_norm": 0.04316595569252968, "global_step": 34645, "epoch": 962, "lr": 0.001} +{"train_loss": 0.02892359159886837, "train_loss_bc": 0.02892359159886837, "train_loss_llm": 0.0, "grad_norm": 0.05895204842090607, "global_step": 34646, "epoch": 962, "lr": 0.001} +{"train_loss": 0.03273962810635567, "train_loss_bc": 0.03273962810635567, "train_loss_llm": 0.0, "grad_norm": 0.08308908343315125, "global_step": 34647, "epoch": 962, "lr": 0.001} +{"train_loss": 0.03293849155306816, "train_loss_bc": 0.03293849155306816, "train_loss_llm": 0.0, "grad_norm": 0.10704328119754791, "global_step": 34648, "epoch": 962, "lr": 0.001} +{"train_loss": 0.03273150697350502, "train_loss_bc": 0.03273150697350502, "train_loss_llm": 0.0, "grad_norm": 0.04351939260959625, "global_step": 34649, "epoch": 962, "lr": 0.001} +{"train_loss": 0.029829585924744606, "train_loss_bc": 0.029812626540660858, "train_loss_llm": 0.0016959011554718018, "grad_norm": 0.06015678495168686, "global_step": 34650, "epoch": 962, "lr": 0.001} +{"train_loss": 0.030833957716822624, "train_loss_bc": 0.030833957716822624, "train_loss_llm": 0.0, "grad_norm": 0.09094978868961334, "global_step": 34651, "epoch": 962, "lr": 0.001} +{"train_loss": 0.030719703063368797, "train_loss_bc": 0.030719703063368797, "train_loss_llm": 0.0, "grad_norm": 0.1128847673535347, "global_step": 34652, "epoch": 962, "lr": 0.001} +{"train_loss": 0.026685072109103203, "train_loss_bc": 0.026685072109103203, "train_loss_llm": 0.0, "grad_norm": 0.044106729328632355, "global_step": 34653, "epoch": 962, "lr": 0.001} +{"train_loss": 0.027624746784567833, "train_loss_bc": 0.027624746784567833, "train_loss_llm": 0.0, "grad_norm": 0.07565519213676453, "global_step": 34654, "epoch": 962, "lr": 0.001} +{"train_loss": 0.03393787890672684, "train_loss_bc": 0.03393787890672684, "train_loss_llm": 0.0, "grad_norm": 0.09704709053039551, "global_step": 34655, "epoch": 962, "lr": 0.001} +{"train_loss": 0.028956597670912743, "train_loss_bc": 0.028956597670912743, "train_loss_llm": 0.0, "grad_norm": 0.11301668733358383, "global_step": 34656, "epoch": 962, "lr": 0.001} +{"train_loss": 0.026699602603912354, "train_loss_bc": 0.026699602603912354, "train_loss_llm": 0.0, "grad_norm": 0.047177545726299286, "global_step": 34657, "epoch": 962, "lr": 0.001} +{"train_loss": 0.02684030681848526, "train_loss_bc": 0.02684030681848526, "train_loss_llm": 0.0, "grad_norm": 0.08898865431547165, "global_step": 34658, "epoch": 962, "lr": 0.001} +{"train_loss": 0.025615932419896126, "train_loss_bc": 0.025615932419896126, "train_loss_llm": 0.0, "grad_norm": 0.1380757838487625, "global_step": 34659, "epoch": 962, "lr": 0.001} +{"train_loss": 0.028059793636202812, "train_loss_bc": 0.028059793636202812, "train_loss_llm": 0.0, "grad_norm": 0.16388508677482605, "global_step": 34660, "epoch": 962, "lr": 0.001} +{"train_loss": 0.030916061252355576, "train_loss_bc": 0.030916061252355576, "train_loss_llm": 0.0, "grad_norm": 0.04760872200131416, "global_step": 34661, "epoch": 962, "lr": 0.001} +{"train_loss": 0.033126093447208405, "train_loss_bc": 0.033126093447208405, "train_loss_llm": 0.0, "grad_norm": 0.08359848707914352, "global_step": 34662, "epoch": 962, "lr": 0.001} +{"train_loss": 0.02970242127776146, "train_loss_bc": 0.029695730656385422, "train_loss_llm": 0.0006690025329589844, "grad_norm": 0.11779762804508209, "global_step": 34663, "epoch": 962, "lr": 0.001} +{"train_loss": 0.03230743110179901, "train_loss_bc": 0.03230743110179901, "train_loss_llm": 0.0, "grad_norm": 0.15198783576488495, "global_step": 34664, "epoch": 962, "lr": 0.001} +{"train_loss": 0.03416568040847778, "train_loss_bc": 0.03415908291935921, "train_loss_llm": 0.0006597638130187988, "grad_norm": 0.044369056820869446, "global_step": 34665, "epoch": 962, "lr": 0.001} +{"train_loss": 0.02662460319697857, "train_loss_bc": 0.026608625426888466, "train_loss_llm": 0.0015977025032043457, "grad_norm": 0.06065516546368599, "global_step": 34666, "epoch": 962, "lr": 0.001} +{"train_loss": 0.028976898692134354, "train_loss_bc": 0.029290568083524704, "train_loss_llm": 0.0, "grad_norm": 0.08151628077030182, "global_step": 34667, "epoch": 962, "lr": 0.001, "val_loss": 0.10974283516407013} +{"train_loss": 0.024472080171108246, "train_loss_bc": 0.024472080171108246, "train_loss_llm": 0.0, "grad_norm": 0.10622362792491913, "global_step": 34668, "epoch": 963, "lr": 0.001} +{"train_loss": 0.029823215678334236, "train_loss_bc": 0.029823215678334236, "train_loss_llm": 0.0, "grad_norm": 0.04400107264518738, "global_step": 34669, "epoch": 963, "lr": 0.001} +{"train_loss": 0.025018751621246338, "train_loss_bc": 0.025018751621246338, "train_loss_llm": 0.0, "grad_norm": 0.07077690213918686, "global_step": 34670, "epoch": 963, "lr": 0.001} +{"train_loss": 0.028219474479556084, "train_loss_bc": 0.02821713499724865, "train_loss_llm": 0.0002339482307434082, "grad_norm": 0.08415844291448593, "global_step": 34671, "epoch": 963, "lr": 0.001} +{"train_loss": 0.026176096871495247, "train_loss_bc": 0.026176096871495247, "train_loss_llm": 0.0, "grad_norm": 0.10872398316860199, "global_step": 34672, "epoch": 963, "lr": 0.001} +{"train_loss": 0.03373849764466286, "train_loss_bc": 0.03372284024953842, "train_loss_llm": 0.0015658140182495117, "grad_norm": 0.04520377144217491, "global_step": 34673, "epoch": 963, "lr": 0.001} +{"train_loss": 0.02954792231321335, "train_loss_bc": 0.029547831043601036, "train_loss_llm": 9.149312973022461e-06, "grad_norm": 0.06696499139070511, "global_step": 34674, "epoch": 963, "lr": 0.001} +{"train_loss": 0.03348533436655998, "train_loss_bc": 0.03348533436655998, "train_loss_llm": 0.0, "grad_norm": 0.09593954682350159, "global_step": 34675, "epoch": 963, "lr": 0.001} +{"train_loss": 0.02765512466430664, "train_loss_bc": 0.027617810294032097, "train_loss_llm": 0.0037314891815185547, "grad_norm": 0.12122814357280731, "global_step": 34676, "epoch": 963, "lr": 0.001} +{"train_loss": 0.027539007365703583, "train_loss_bc": 0.027535589411854744, "train_loss_llm": 0.00034180283546447754, "grad_norm": 0.043554700911045074, "global_step": 34677, "epoch": 963, "lr": 0.001} +{"train_loss": 0.03275986388325691, "train_loss_bc": 0.032753683626651764, "train_loss_llm": 0.0006180405616760254, "grad_norm": 0.0809214636683464, "global_step": 34678, "epoch": 963, "lr": 0.001} +{"train_loss": 0.027088504284620285, "train_loss_bc": 0.027088504284620285, "train_loss_llm": 0.0, "grad_norm": 0.11628201603889465, "global_step": 34679, "epoch": 963, "lr": 0.001} +{"train_loss": 0.03540517017245293, "train_loss_bc": 0.03540517017245293, "train_loss_llm": 0.0, "grad_norm": 0.14369097352027893, "global_step": 34680, "epoch": 963, "lr": 0.001} +{"train_loss": 0.028593743219971657, "train_loss_bc": 0.028593743219971657, "train_loss_llm": 0.0, "grad_norm": 0.04419465363025665, "global_step": 34681, "epoch": 963, "lr": 0.001} +{"train_loss": 0.028249520808458328, "train_loss_bc": 0.028249520808458328, "train_loss_llm": 0.0, "grad_norm": 0.0680205225944519, "global_step": 34682, "epoch": 963, "lr": 0.001} +{"train_loss": 0.027570998296141624, "train_loss_bc": 0.027570998296141624, "train_loss_llm": 0.0, "grad_norm": 0.09729806333780289, "global_step": 34683, "epoch": 963, "lr": 0.001} +{"train_loss": 0.02396441251039505, "train_loss_bc": 0.023950684815645218, "train_loss_llm": 0.0013727247714996338, "grad_norm": 0.11839914321899414, "global_step": 34684, "epoch": 963, "lr": 0.001} +{"train_loss": 0.02603572979569435, "train_loss_bc": 0.02603572979569435, "train_loss_llm": 0.0, "grad_norm": 0.0314144566655159, "global_step": 34685, "epoch": 963, "lr": 0.001} +{"train_loss": 0.02286318875849247, "train_loss_bc": 0.02286318875849247, "train_loss_llm": 0.0, "grad_norm": 0.05487046018242836, "global_step": 34686, "epoch": 963, "lr": 0.001} +{"train_loss": 0.031041324138641357, "train_loss_bc": 0.031035488471388817, "train_loss_llm": 0.0005835294723510742, "grad_norm": 0.08079280704259872, "global_step": 34687, "epoch": 963, "lr": 0.001} +{"train_loss": 0.03083568997681141, "train_loss_bc": 0.03083568997681141, "train_loss_llm": 0.0, "grad_norm": 0.09521913528442383, "global_step": 34688, "epoch": 963, "lr": 0.001} +{"train_loss": 0.03649738058447838, "train_loss_bc": 0.03649738058447838, "train_loss_llm": 0.0, "grad_norm": 0.05263228341937065, "global_step": 34689, "epoch": 963, "lr": 0.001} +{"train_loss": 0.025344878435134888, "train_loss_bc": 0.025344878435134888, "train_loss_llm": 0.0, "grad_norm": 0.08656790852546692, "global_step": 34690, "epoch": 963, "lr": 0.001} +{"train_loss": 0.035370517522096634, "train_loss_bc": 0.035370517522096634, "train_loss_llm": 0.0, "grad_norm": 0.13542689383029938, "global_step": 34691, "epoch": 963, "lr": 0.001} +{"train_loss": 0.023506173864006996, "train_loss_bc": 0.023506173864006996, "train_loss_llm": 0.0, "grad_norm": 0.14892350137233734, "global_step": 34692, "epoch": 963, "lr": 0.001} +{"train_loss": 0.030067356303334236, "train_loss_bc": 0.030063362792134285, "train_loss_llm": 0.0003994107246398926, "grad_norm": 0.04752814769744873, "global_step": 34693, "epoch": 963, "lr": 0.001} +{"train_loss": 0.02601531147956848, "train_loss_bc": 0.02601531147956848, "train_loss_llm": 0.0, "grad_norm": 0.08714042603969574, "global_step": 34694, "epoch": 963, "lr": 0.001} +{"train_loss": 0.030063975602388382, "train_loss_bc": 0.030063975602388382, "train_loss_llm": 0.0, "grad_norm": 0.1275244951248169, "global_step": 34695, "epoch": 963, "lr": 0.001} +{"train_loss": 0.031727518886327744, "train_loss_bc": 0.031727518886327744, "train_loss_llm": 0.0, "grad_norm": 0.15524151921272278, "global_step": 34696, "epoch": 963, "lr": 0.001} +{"train_loss": 0.024348262697458267, "train_loss_bc": 0.024348262697458267, "train_loss_llm": 0.0, "grad_norm": 0.033908672630786896, "global_step": 34697, "epoch": 963, "lr": 0.001} +{"train_loss": 0.02798367664217949, "train_loss_bc": 0.027977900579571724, "train_loss_llm": 0.0005775988101959229, "grad_norm": 0.05944373086094856, "global_step": 34698, "epoch": 963, "lr": 0.001} +{"train_loss": 0.032849524170160294, "train_loss_bc": 0.032849524170160294, "train_loss_llm": 0.0, "grad_norm": 0.08422933518886566, "global_step": 34699, "epoch": 963, "lr": 0.001} +{"train_loss": 0.029259327799081802, "train_loss_bc": 0.029259327799081802, "train_loss_llm": 0.0, "grad_norm": 0.11200480163097382, "global_step": 34700, "epoch": 963, "lr": 0.001} +{"train_loss": 0.028099199756979942, "train_loss_bc": 0.028081422671675682, "train_loss_llm": 0.0017776191234588623, "grad_norm": 0.038576435297727585, "global_step": 34701, "epoch": 963, "lr": 0.001} +{"train_loss": 0.031830959022045135, "train_loss_bc": 0.031830959022045135, "train_loss_llm": 0.0, "grad_norm": 0.05877135321497917, "global_step": 34702, "epoch": 963, "lr": 0.001} +{"train_loss": 0.02897400398635202, "train_loss_bc": 0.030016429722309113, "train_loss_llm": 0.0, "grad_norm": 0.09100382030010223, "global_step": 34703, "epoch": 963, "lr": 0.001, "val_loss": 0.10857053101062775} +{"train_loss": 0.029325634241104126, "train_loss_bc": 0.02931908145546913, "train_loss_llm": 0.0006553232669830322, "grad_norm": 0.11160100251436234, "global_step": 34704, "epoch": 964, "lr": 0.001} +{"train_loss": 0.028526803478598595, "train_loss_bc": 0.028526803478598595, "train_loss_llm": 0.0, "grad_norm": 0.04762895405292511, "global_step": 34705, "epoch": 964, "lr": 0.001} +{"train_loss": 0.03236820176243782, "train_loss_bc": 0.03236820176243782, "train_loss_llm": 0.0, "grad_norm": 0.0846681296825409, "global_step": 34706, "epoch": 964, "lr": 0.001} +{"train_loss": 0.027920348569750786, "train_loss_bc": 0.02791600301861763, "train_loss_llm": 0.00043460726737976074, "grad_norm": 0.12491844594478607, "global_step": 34707, "epoch": 964, "lr": 0.001} +{"train_loss": 0.02765451744198799, "train_loss_bc": 0.02765451744198799, "train_loss_llm": 0.0, "grad_norm": 0.14914600551128387, "global_step": 34708, "epoch": 964, "lr": 0.001} +{"train_loss": 0.024685993790626526, "train_loss_bc": 0.024685993790626526, "train_loss_llm": 0.0, "grad_norm": 0.03943442925810814, "global_step": 34709, "epoch": 964, "lr": 0.001} +{"train_loss": 0.030681690201163292, "train_loss_bc": 0.03065466694533825, "train_loss_llm": 0.002702385187149048, "grad_norm": 0.07561048120260239, "global_step": 34710, "epoch": 964, "lr": 0.001} +{"train_loss": 0.02983071655035019, "train_loss_bc": 0.02983071655035019, "train_loss_llm": 0.0, "grad_norm": 0.10669447481632233, "global_step": 34711, "epoch": 964, "lr": 0.001} +{"train_loss": 0.03360431641340256, "train_loss_bc": 0.03360431641340256, "train_loss_llm": 0.0, "grad_norm": 0.1337091475725174, "global_step": 34712, "epoch": 964, "lr": 0.001} +{"train_loss": 0.021957620978355408, "train_loss_bc": 0.0219427477568388, "train_loss_llm": 0.0014872550964355469, "grad_norm": 0.04091836139559746, "global_step": 34713, "epoch": 964, "lr": 0.001} +{"train_loss": 0.026709621772170067, "train_loss_bc": 0.026709621772170067, "train_loss_llm": 0.0, "grad_norm": 0.0634215697646141, "global_step": 34714, "epoch": 964, "lr": 0.001} +{"train_loss": 0.026442617177963257, "train_loss_bc": 0.026442617177963257, "train_loss_llm": 0.0, "grad_norm": 0.08961039781570435, "global_step": 34715, "epoch": 964, "lr": 0.001} +{"train_loss": 0.02600998990237713, "train_loss_bc": 0.02600998990237713, "train_loss_llm": 0.0, "grad_norm": 0.12743139266967773, "global_step": 34716, "epoch": 964, "lr": 0.001} +{"train_loss": 0.025796696543693542, "train_loss_bc": 0.025796696543693542, "train_loss_llm": 0.0, "grad_norm": 0.03302663564682007, "global_step": 34717, "epoch": 964, "lr": 0.001} +{"train_loss": 0.027128254994750023, "train_loss_bc": 0.027118314057588577, "train_loss_llm": 0.0009941458702087402, "grad_norm": 0.05574638396501541, "global_step": 34718, "epoch": 964, "lr": 0.001} +{"train_loss": 0.03384576365351677, "train_loss_bc": 0.03384576365351677, "train_loss_llm": 0.0, "grad_norm": 0.07532767206430435, "global_step": 34719, "epoch": 964, "lr": 0.001} +{"train_loss": 0.025036126375198364, "train_loss_bc": 0.025036126375198364, "train_loss_llm": 0.0, "grad_norm": 0.0901787132024765, "global_step": 34720, "epoch": 964, "lr": 0.001} +{"train_loss": 0.02743067406117916, "train_loss_bc": 0.027428334578871727, "train_loss_llm": 0.0002339780330657959, "grad_norm": 0.0401148796081543, "global_step": 34721, "epoch": 964, "lr": 0.001} +{"train_loss": 0.03348105028271675, "train_loss_bc": 0.03348105028271675, "train_loss_llm": 0.0, "grad_norm": 0.057932935655117035, "global_step": 34722, "epoch": 964, "lr": 0.001} +{"train_loss": 0.026475675404071808, "train_loss_bc": 0.026475675404071808, "train_loss_llm": 0.0, "grad_norm": 0.07390900701284409, "global_step": 34723, "epoch": 964, "lr": 0.001} +{"train_loss": 0.03365996479988098, "train_loss_bc": 0.03365720435976982, "train_loss_llm": 0.0002759397029876709, "grad_norm": 0.08624514192342758, "global_step": 34724, "epoch": 964, "lr": 0.001} +{"train_loss": 0.03110835887491703, "train_loss_bc": 0.03110835887491703, "train_loss_llm": 0.0, "grad_norm": 0.05128087103366852, "global_step": 34725, "epoch": 964, "lr": 0.001} +{"train_loss": 0.03100578300654888, "train_loss_bc": 0.03100578300654888, "train_loss_llm": 0.0, "grad_norm": 0.09490467607975006, "global_step": 34726, "epoch": 964, "lr": 0.001} +{"train_loss": 0.031157147139310837, "train_loss_bc": 0.031149037182331085, "train_loss_llm": 0.0008109807968139648, "grad_norm": 0.12099559605121613, "global_step": 34727, "epoch": 964, "lr": 0.001} +{"train_loss": 0.026066329330205917, "train_loss_bc": 0.026066329330205917, "train_loss_llm": 0.0, "grad_norm": 0.13898152112960815, "global_step": 34728, "epoch": 964, "lr": 0.001} +{"train_loss": 0.023899173364043236, "train_loss_bc": 0.023899173364043236, "train_loss_llm": 0.0, "grad_norm": 0.03609369322657585, "global_step": 34729, "epoch": 964, "lr": 0.001} +{"train_loss": 0.030674723908305168, "train_loss_bc": 0.030674723908305168, "train_loss_llm": 0.0, "grad_norm": 0.06705182790756226, "global_step": 34730, "epoch": 964, "lr": 0.001} +{"train_loss": 0.0256094541400671, "train_loss_bc": 0.025603458285331726, "train_loss_llm": 0.0005995035171508789, "grad_norm": 0.08455724269151688, "global_step": 34731, "epoch": 964, "lr": 0.001} +{"train_loss": 0.026947209611535072, "train_loss_bc": 0.026947209611535072, "train_loss_llm": 0.0, "grad_norm": 0.1094081923365593, "global_step": 34732, "epoch": 964, "lr": 0.001} +{"train_loss": 0.02624676562845707, "train_loss_bc": 0.02624676562845707, "train_loss_llm": 0.0, "grad_norm": 0.033161718398332596, "global_step": 34733, "epoch": 964, "lr": 0.001} +{"train_loss": 0.0277131088078022, "train_loss_bc": 0.0277131088078022, "train_loss_llm": 0.0, "grad_norm": 0.06716223806142807, "global_step": 34734, "epoch": 964, "lr": 0.001} +{"train_loss": 0.02685759961605072, "train_loss_bc": 0.02685759961605072, "train_loss_llm": 0.0, "grad_norm": 0.09015249460935593, "global_step": 34735, "epoch": 964, "lr": 0.001} +{"train_loss": 0.03495192900300026, "train_loss_bc": 0.03494679182767868, "train_loss_llm": 0.0005137026309967041, "grad_norm": 0.13110439479351044, "global_step": 34736, "epoch": 964, "lr": 0.001} +{"train_loss": 0.03078976273536682, "train_loss_bc": 0.030784834176301956, "train_loss_llm": 0.0004929006099700928, "grad_norm": 0.03969820588827133, "global_step": 34737, "epoch": 964, "lr": 0.001} +{"train_loss": 0.03795994445681572, "train_loss_bc": 0.03795994445681572, "train_loss_llm": 0.0, "grad_norm": 0.05326489359140396, "global_step": 34738, "epoch": 964, "lr": 0.001} +{"train_loss": 0.02886589037047492, "train_loss_bc": 0.029612485319375992, "train_loss_llm": 0.0, "grad_norm": 0.0863908901810646, "global_step": 34739, "epoch": 964, "lr": 0.001, "val_loss": 0.10974914580583572} +{"train_loss": 0.03233790025115013, "train_loss_bc": 0.03233790025115013, "train_loss_llm": 0.0, "grad_norm": 0.11308884620666504, "global_step": 34740, "epoch": 965, "lr": 0.001} +{"train_loss": 0.029642947018146515, "train_loss_bc": 0.029642947018146515, "train_loss_llm": 0.0, "grad_norm": 0.037084490060806274, "global_step": 34741, "epoch": 965, "lr": 0.001} +{"train_loss": 0.03324892371892929, "train_loss_bc": 0.03324892371892929, "train_loss_llm": 0.0, "grad_norm": 0.054992254823446274, "global_step": 34742, "epoch": 965, "lr": 0.001} +{"train_loss": 0.025953518226742744, "train_loss_bc": 0.025953518226742744, "train_loss_llm": 0.0, "grad_norm": 0.07243282347917557, "global_step": 34743, "epoch": 965, "lr": 0.001} +{"train_loss": 0.031107982620596886, "train_loss_bc": 0.031082509085536003, "train_loss_llm": 0.0025472640991210938, "grad_norm": 0.09135282784700394, "global_step": 34744, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02737981639802456, "train_loss_bc": 0.02737981639802456, "train_loss_llm": 0.0, "grad_norm": 0.04215086251497269, "global_step": 34745, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02999824658036232, "train_loss_bc": 0.029986672103405, "train_loss_llm": 0.001157522201538086, "grad_norm": 0.06671134382486343, "global_step": 34746, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02950025536119938, "train_loss_bc": 0.02950025536119938, "train_loss_llm": 0.0, "grad_norm": 0.1011541411280632, "global_step": 34747, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02718392014503479, "train_loss_bc": 0.027172308415174484, "train_loss_llm": 0.0011611878871917725, "grad_norm": 0.1287251114845276, "global_step": 34748, "epoch": 965, "lr": 0.001} +{"train_loss": 0.028172800317406654, "train_loss_bc": 0.028170054778456688, "train_loss_llm": 0.00027447938919067383, "grad_norm": 0.043470531702041626, "global_step": 34749, "epoch": 965, "lr": 0.001} +{"train_loss": 0.0317668579518795, "train_loss_bc": 0.03176603838801384, "train_loss_llm": 8.20159912109375e-05, "grad_norm": 0.08187849074602127, "global_step": 34750, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02473728358745575, "train_loss_bc": 0.02473728358745575, "train_loss_llm": 0.0, "grad_norm": 0.09529414027929306, "global_step": 34751, "epoch": 965, "lr": 0.001} +{"train_loss": 0.027067091315984726, "train_loss_bc": 0.027067091315984726, "train_loss_llm": 0.0, "grad_norm": 0.13588131964206696, "global_step": 34752, "epoch": 965, "lr": 0.001} +{"train_loss": 0.025594636797904968, "train_loss_bc": 0.025581130757927895, "train_loss_llm": 0.0013505220413208008, "grad_norm": 0.04333603382110596, "global_step": 34753, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02945706807076931, "train_loss_bc": 0.02945706807076931, "train_loss_llm": 0.0, "grad_norm": 0.07223648577928543, "global_step": 34754, "epoch": 965, "lr": 0.001} +{"train_loss": 0.0324133075773716, "train_loss_bc": 0.03240976482629776, "train_loss_llm": 0.00035440921783447266, "grad_norm": 0.1060243770480156, "global_step": 34755, "epoch": 965, "lr": 0.001} +{"train_loss": 0.03111335076391697, "train_loss_bc": 0.03109508752822876, "train_loss_llm": 0.0018264055252075195, "grad_norm": 0.13443580269813538, "global_step": 34756, "epoch": 965, "lr": 0.001} +{"train_loss": 0.030720505863428116, "train_loss_bc": 0.03071952797472477, "train_loss_llm": 9.78708267211914e-05, "grad_norm": 0.04650687053799629, "global_step": 34757, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02702467143535614, "train_loss_bc": 0.02702467143535614, "train_loss_llm": 0.0, "grad_norm": 0.06146196275949478, "global_step": 34758, "epoch": 965, "lr": 0.001} +{"train_loss": 0.028030093759298325, "train_loss_bc": 0.028030093759298325, "train_loss_llm": 0.0, "grad_norm": 0.07549732178449631, "global_step": 34759, "epoch": 965, "lr": 0.001} +{"train_loss": 0.029625222086906433, "train_loss_bc": 0.02961636893451214, "train_loss_llm": 0.00088539719581604, "grad_norm": 0.10214164853096008, "global_step": 34760, "epoch": 965, "lr": 0.001} +{"train_loss": 0.024577070027589798, "train_loss_bc": 0.024577070027589798, "train_loss_llm": 0.0, "grad_norm": 0.03295578435063362, "global_step": 34761, "epoch": 965, "lr": 0.001} +{"train_loss": 0.030585894361138344, "train_loss_bc": 0.030585894361138344, "train_loss_llm": 0.0, "grad_norm": 0.04826125502586365, "global_step": 34762, "epoch": 965, "lr": 0.001} +{"train_loss": 0.031322181224823, "train_loss_bc": 0.031322181224823, "train_loss_llm": 0.0, "grad_norm": 0.06912768632173538, "global_step": 34763, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02828371711075306, "train_loss_bc": 0.02828371711075306, "train_loss_llm": 0.0, "grad_norm": 0.08947107195854187, "global_step": 34764, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02842879667878151, "train_loss_bc": 0.028428714722394943, "train_loss_llm": 8.195638656616211e-06, "grad_norm": 0.040672220289707184, "global_step": 34765, "epoch": 965, "lr": 0.001} +{"train_loss": 0.03150609880685806, "train_loss_bc": 0.03150609880685806, "train_loss_llm": 0.0, "grad_norm": 0.07470407336950302, "global_step": 34766, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02915922924876213, "train_loss_bc": 0.02915922924876213, "train_loss_llm": 0.0, "grad_norm": 0.10137947648763657, "global_step": 34767, "epoch": 965, "lr": 0.001} +{"train_loss": 0.022601107135415077, "train_loss_bc": 0.022601107135415077, "train_loss_llm": 0.0, "grad_norm": 0.12921756505966187, "global_step": 34768, "epoch": 965, "lr": 0.001} +{"train_loss": 0.02781238593161106, "train_loss_bc": 0.02780601941049099, "train_loss_llm": 0.000636667013168335, "grad_norm": 0.03199049085378647, "global_step": 34769, "epoch": 965, "lr": 0.001} +{"train_loss": 0.027203135192394257, "train_loss_bc": 0.027203135192394257, "train_loss_llm": 0.0, "grad_norm": 0.061430610716342926, "global_step": 34770, "epoch": 965, "lr": 0.001} +{"train_loss": 0.0290021151304245, "train_loss_bc": 0.0290021151304245, "train_loss_llm": 0.0, "grad_norm": 0.09636962413787842, "global_step": 34771, "epoch": 965, "lr": 0.001} +{"train_loss": 0.028049292042851448, "train_loss_bc": 0.028049292042851448, "train_loss_llm": 0.0, "grad_norm": 0.12481769919395447, "global_step": 34772, "epoch": 965, "lr": 0.001} +{"train_loss": 0.03030231222510338, "train_loss_bc": 0.03030231222510338, "train_loss_llm": 0.0, "grad_norm": 0.043766070157289505, "global_step": 34773, "epoch": 965, "lr": 0.001} +{"train_loss": 0.029210874810814857, "train_loss_bc": 0.029210874810814857, "train_loss_llm": 0.0, "grad_norm": 0.07294416427612305, "global_step": 34774, "epoch": 965, "lr": 0.001} +{"train_loss": 0.029021629494511418, "train_loss_bc": 0.034658052027225494, "train_loss_llm": 0.0, "grad_norm": 0.11697136610746384, "global_step": 34775, "epoch": 965, "lr": 0.001, "train/cumulative_reward": 59.81875058539239, "train/mean_score": 15.731638911832116, "train/success_rate": 0.4, "test/cumulative_reward": 28.34520187504748, "test/mean_score": 5.350780130492215, "test/success_rate": 0.12, "val_loss": 0.11075031757354736, "train_action_mse_error": 0.03176604583859444} +{"train_loss": 0.03334316983819008, "train_loss_bc": 0.03334316983819008, "train_loss_llm": 0.0, "grad_norm": 0.13909482955932617, "global_step": 34776, "epoch": 966, "lr": 0.001} +{"train_loss": 0.029149537906050682, "train_loss_bc": 0.029149537906050682, "train_loss_llm": 0.0, "grad_norm": 0.03462046757340431, "global_step": 34777, "epoch": 966, "lr": 0.001} +{"train_loss": 0.027846014127135277, "train_loss_bc": 0.027846014127135277, "train_loss_llm": 0.0, "grad_norm": 0.06518207490444183, "global_step": 34778, "epoch": 966, "lr": 0.001} +{"train_loss": 0.02721242606639862, "train_loss_bc": 0.02721242606639862, "train_loss_llm": 0.0, "grad_norm": 0.08870191127061844, "global_step": 34779, "epoch": 966, "lr": 0.001} +{"train_loss": 0.03191531077027321, "train_loss_bc": 0.03191531077027321, "train_loss_llm": 0.0, "grad_norm": 0.10694602131843567, "global_step": 34780, "epoch": 966, "lr": 0.001} +{"train_loss": 0.027860309928655624, "train_loss_bc": 0.027860309928655624, "train_loss_llm": 0.0, "grad_norm": 0.035489171743392944, "global_step": 34781, "epoch": 966, "lr": 0.001} +{"train_loss": 0.03189806640148163, "train_loss_bc": 0.03189806640148163, "train_loss_llm": 0.0, "grad_norm": 0.05846301466226578, "global_step": 34782, "epoch": 966, "lr": 0.001} +{"train_loss": 0.028892386704683304, "train_loss_bc": 0.028892386704683304, "train_loss_llm": 0.0, "grad_norm": 0.08447609841823578, "global_step": 34783, "epoch": 966, "lr": 0.001} +{"train_loss": 0.022561563178896904, "train_loss_bc": 0.022561563178896904, "train_loss_llm": 0.0, "grad_norm": 0.10480000078678131, "global_step": 34784, "epoch": 966, "lr": 0.001} +{"train_loss": 0.028316915035247803, "train_loss_bc": 0.028316915035247803, "train_loss_llm": 0.0, "grad_norm": 0.027620503678917885, "global_step": 34785, "epoch": 966, "lr": 0.001} +{"train_loss": 0.0259822029620409, "train_loss_bc": 0.0259822029620409, "train_loss_llm": 0.0, "grad_norm": 0.0453791618347168, "global_step": 34786, "epoch": 966, "lr": 0.001} +{"train_loss": 0.030493993312120438, "train_loss_bc": 0.030493993312120438, "train_loss_llm": 0.0, "grad_norm": 0.08014104515314102, "global_step": 34787, "epoch": 966, "lr": 0.001} +{"train_loss": 0.02679337188601494, "train_loss_bc": 0.026790019124746323, "train_loss_llm": 0.00033530592918395996, "grad_norm": 0.09728862345218658, "global_step": 34788, "epoch": 966, "lr": 0.001} +{"train_loss": 0.03394734114408493, "train_loss_bc": 0.03394734114408493, "train_loss_llm": 0.0, "grad_norm": 0.05413644015789032, "global_step": 34789, "epoch": 966, "lr": 0.001} +{"train_loss": 0.03226260468363762, "train_loss_bc": 0.03226260468363762, "train_loss_llm": 0.0, "grad_norm": 0.07573885470628738, "global_step": 34790, "epoch": 966, "lr": 0.001} +{"train_loss": 0.022872766479849815, "train_loss_bc": 0.022872766479849815, "train_loss_llm": 0.0, "grad_norm": 0.11269064992666245, "global_step": 34791, "epoch": 966, "lr": 0.001} +{"train_loss": 0.03270573168992996, "train_loss_bc": 0.03270573168992996, "train_loss_llm": 0.0, "grad_norm": 0.1446784883737564, "global_step": 34792, "epoch": 966, "lr": 0.001} +{"train_loss": 0.02826535515487194, "train_loss_bc": 0.02826535515487194, "train_loss_llm": 0.0, "grad_norm": 0.04209206625819206, "global_step": 34793, "epoch": 966, "lr": 0.001} +{"train_loss": 0.029109913855791092, "train_loss_bc": 0.029109913855791092, "train_loss_llm": 0.0, "grad_norm": 0.07539507001638412, "global_step": 34794, "epoch": 966, "lr": 0.001} +{"train_loss": 0.028677979484200478, "train_loss_bc": 0.028677979484200478, "train_loss_llm": 0.0, "grad_norm": 0.11639042943716049, "global_step": 34795, "epoch": 966, "lr": 0.001} +{"train_loss": 0.024101227521896362, "train_loss_bc": 0.024101227521896362, "train_loss_llm": 0.0, "grad_norm": 0.13429687917232513, "global_step": 34796, "epoch": 966, "lr": 0.001} +{"train_loss": 0.029769914224743843, "train_loss_bc": 0.029769914224743843, "train_loss_llm": 0.0, "grad_norm": 0.038951028138399124, "global_step": 34797, "epoch": 966, "lr": 0.001} +{"train_loss": 0.023479482159018517, "train_loss_bc": 0.023479482159018517, "train_loss_llm": 0.0, "grad_norm": 0.06238007918000221, "global_step": 34798, "epoch": 966, "lr": 0.001} +{"train_loss": 0.025181829929351807, "train_loss_bc": 0.02517850324511528, "train_loss_llm": 0.0003326833248138428, "grad_norm": 0.09149350970983505, "global_step": 34799, "epoch": 966, "lr": 0.001} +{"train_loss": 0.028308482840657234, "train_loss_bc": 0.02829449065029621, "train_loss_llm": 0.0013991892337799072, "grad_norm": 0.11956784874200821, "global_step": 34800, "epoch": 966, "lr": 0.001} +{"train_loss": 0.032597873359918594, "train_loss_bc": 0.032597873359918594, "train_loss_llm": 0.0, "grad_norm": 0.038725126534700394, "global_step": 34801, "epoch": 966, "lr": 0.001} +{"train_loss": 0.036805566400289536, "train_loss_bc": 0.03679661825299263, "train_loss_llm": 0.0008949935436248779, "grad_norm": 0.06265321373939514, "global_step": 34802, "epoch": 966, "lr": 0.001} +{"train_loss": 0.028523346409201622, "train_loss_bc": 0.028523346409201622, "train_loss_llm": 0.0, "grad_norm": 0.08843646943569183, "global_step": 34803, "epoch": 966, "lr": 0.001} +{"train_loss": 0.029992451891303062, "train_loss_bc": 0.02998906373977661, "train_loss_llm": 0.0003388822078704834, "grad_norm": 0.10865787416696548, "global_step": 34804, "epoch": 966, "lr": 0.001} +{"train_loss": 0.027123650535941124, "train_loss_bc": 0.027123650535941124, "train_loss_llm": 0.0, "grad_norm": 0.04557471722364426, "global_step": 34805, "epoch": 966, "lr": 0.001} +{"train_loss": 0.027103692293167114, "train_loss_bc": 0.0270895566791296, "train_loss_llm": 0.001413583755493164, "grad_norm": 0.06329536437988281, "global_step": 34806, "epoch": 966, "lr": 0.001} +{"train_loss": 0.027720008045434952, "train_loss_bc": 0.027720008045434952, "train_loss_llm": 0.0, "grad_norm": 0.08559389412403107, "global_step": 34807, "epoch": 966, "lr": 0.001} +{"train_loss": 0.03480246290564537, "train_loss_bc": 0.03480246290564537, "train_loss_llm": 0.0, "grad_norm": 0.10789964348077774, "global_step": 34808, "epoch": 966, "lr": 0.001} +{"train_loss": 0.024153493344783783, "train_loss_bc": 0.024148259311914444, "train_loss_llm": 0.0005233287811279297, "grad_norm": 0.03713679686188698, "global_step": 34809, "epoch": 966, "lr": 0.001} +{"train_loss": 0.03104352578520775, "train_loss_bc": 0.031021341681480408, "train_loss_llm": 0.002218395471572876, "grad_norm": 0.05348288267850876, "global_step": 34810, "epoch": 966, "lr": 0.001} +{"train_loss": 0.028927969384110637, "train_loss_bc": 0.03059292957186699, "train_loss_llm": 0.0, "grad_norm": 0.06961836665868759, "global_step": 34811, "epoch": 966, "lr": 0.001, "val_loss": 0.1081734374165535} +{"train_loss": 0.02730463817715645, "train_loss_bc": 0.027288421988487244, "train_loss_llm": 0.0016216039657592773, "grad_norm": 0.09504184871912003, "global_step": 34812, "epoch": 967, "lr": 0.001} +{"train_loss": 0.028795138001441956, "train_loss_bc": 0.028795138001441956, "train_loss_llm": 0.0, "grad_norm": 0.037281692028045654, "global_step": 34813, "epoch": 967, "lr": 0.001} +{"train_loss": 0.031360842287540436, "train_loss_bc": 0.031360842287540436, "train_loss_llm": 0.0, "grad_norm": 0.06396767497062683, "global_step": 34814, "epoch": 967, "lr": 0.001} +{"train_loss": 0.026785513386130333, "train_loss_bc": 0.026785513386130333, "train_loss_llm": 0.0, "grad_norm": 0.08298227190971375, "global_step": 34815, "epoch": 967, "lr": 0.001} +{"train_loss": 0.031036818400025368, "train_loss_bc": 0.031036818400025368, "train_loss_llm": 0.0, "grad_norm": 0.11143729835748672, "global_step": 34816, "epoch": 967, "lr": 0.001} +{"train_loss": 0.032234784215688705, "train_loss_bc": 0.032234784215688705, "train_loss_llm": 0.0, "grad_norm": 0.04654334858059883, "global_step": 34817, "epoch": 967, "lr": 0.001} +{"train_loss": 0.02644938789308071, "train_loss_bc": 0.02644938789308071, "train_loss_llm": 0.0, "grad_norm": 0.08113348484039307, "global_step": 34818, "epoch": 967, "lr": 0.001} +{"train_loss": 0.028601905331015587, "train_loss_bc": 0.028601905331015587, "train_loss_llm": 0.0, "grad_norm": 0.11430317908525467, "global_step": 34819, "epoch": 967, "lr": 0.001} +{"train_loss": 0.028624510392546654, "train_loss_bc": 0.028624029830098152, "train_loss_llm": 4.801154136657715e-05, "grad_norm": 0.15963104367256165, "global_step": 34820, "epoch": 967, "lr": 0.001} +{"train_loss": 0.028210081160068512, "train_loss_bc": 0.028210081160068512, "train_loss_llm": 0.0, "grad_norm": 0.03983449190855026, "global_step": 34821, "epoch": 967, "lr": 0.001} +{"train_loss": 0.025545921176671982, "train_loss_bc": 0.025545921176671982, "train_loss_llm": 0.0, "grad_norm": 0.06841617077589035, "global_step": 34822, "epoch": 967, "lr": 0.001} +{"train_loss": 0.027103492990136147, "train_loss_bc": 0.027103492990136147, "train_loss_llm": 0.0, "grad_norm": 0.08923453092575073, "global_step": 34823, "epoch": 967, "lr": 0.001} +{"train_loss": 0.02670464478433132, "train_loss_bc": 0.02670464478433132, "train_loss_llm": 0.0, "grad_norm": 0.10691042989492416, "global_step": 34824, "epoch": 967, "lr": 0.001} +{"train_loss": 0.028735686093568802, "train_loss_bc": 0.028735686093568802, "train_loss_llm": 0.0, "grad_norm": 0.03703536093235016, "global_step": 34825, "epoch": 967, "lr": 0.001} +{"train_loss": 0.035113293677568436, "train_loss_bc": 0.035113293677568436, "train_loss_llm": 0.0, "grad_norm": 0.07945208251476288, "global_step": 34826, "epoch": 967, "lr": 0.001} +{"train_loss": 0.028113173320889473, "train_loss_bc": 0.028113173320889473, "train_loss_llm": 0.0, "grad_norm": 0.10243932157754898, "global_step": 34827, "epoch": 967, "lr": 0.001} +{"train_loss": 0.0271629486232996, "train_loss_bc": 0.0271629486232996, "train_loss_llm": 0.0, "grad_norm": 0.13010238111019135, "global_step": 34828, "epoch": 967, "lr": 0.001} +{"train_loss": 0.02755947969853878, "train_loss_bc": 0.02755947969853878, "train_loss_llm": 0.0, "grad_norm": 0.04312029480934143, "global_step": 34829, "epoch": 967, "lr": 0.001} +{"train_loss": 0.029911119490861893, "train_loss_bc": 0.029911119490861893, "train_loss_llm": 0.0, "grad_norm": 0.06671524792909622, "global_step": 34830, "epoch": 967, "lr": 0.001} +{"train_loss": 0.02702130563557148, "train_loss_bc": 0.02702130563557148, "train_loss_llm": 0.0, "grad_norm": 0.10017760843038559, "global_step": 34831, "epoch": 967, "lr": 0.001} +{"train_loss": 0.028335215523838997, "train_loss_bc": 0.028335215523838997, "train_loss_llm": 0.0, "grad_norm": 0.12665988504886627, "global_step": 34832, "epoch": 967, "lr": 0.001} +{"train_loss": 0.03276021033525467, "train_loss_bc": 0.03276021033525467, "train_loss_llm": 0.0, "grad_norm": 0.04574176296591759, "global_step": 34833, "epoch": 967, "lr": 0.001} +{"train_loss": 0.026615288108587265, "train_loss_bc": 0.026615288108587265, "train_loss_llm": 0.0, "grad_norm": 0.0734650269150734, "global_step": 34834, "epoch": 967, "lr": 0.001} +{"train_loss": 0.03177344799041748, "train_loss_bc": 0.03177344799041748, "train_loss_llm": 0.0, "grad_norm": 0.09901904314756393, "global_step": 34835, "epoch": 967, "lr": 0.001} +{"train_loss": 0.030959894880652428, "train_loss_bc": 0.03095204383134842, "train_loss_llm": 0.0007850527763366699, "grad_norm": 0.14303123950958252, "global_step": 34836, "epoch": 967, "lr": 0.001} +{"train_loss": 0.032748058438301086, "train_loss_bc": 0.032747503370046616, "train_loss_llm": 5.5670738220214844e-05, "grad_norm": 0.03801145404577255, "global_step": 34837, "epoch": 967, "lr": 0.001} +{"train_loss": 0.029784923419356346, "train_loss_bc": 0.029769616201519966, "train_loss_llm": 0.0015307366847991943, "grad_norm": 0.06244193762540817, "global_step": 34838, "epoch": 967, "lr": 0.001} +{"train_loss": 0.026175864040851593, "train_loss_bc": 0.026175864040851593, "train_loss_llm": 0.0, "grad_norm": 0.08175360411405563, "global_step": 34839, "epoch": 967, "lr": 0.001} +{"train_loss": 0.02889224886894226, "train_loss_bc": 0.02889224886894226, "train_loss_llm": 0.0, "grad_norm": 0.10340503603219986, "global_step": 34840, "epoch": 967, "lr": 0.001} +{"train_loss": 0.02986355870962143, "train_loss_bc": 0.02986355870962143, "train_loss_llm": 0.0, "grad_norm": 0.041126690804958344, "global_step": 34841, "epoch": 967, "lr": 0.001} +{"train_loss": 0.03133441507816315, "train_loss_bc": 0.03133441507816315, "train_loss_llm": 0.0, "grad_norm": 0.07060505449771881, "global_step": 34842, "epoch": 967, "lr": 0.001} +{"train_loss": 0.029857398942112923, "train_loss_bc": 0.029842529445886612, "train_loss_llm": 0.0014868974685668945, "grad_norm": 0.10241762548685074, "global_step": 34843, "epoch": 967, "lr": 0.001} +{"train_loss": 0.025004876777529716, "train_loss_bc": 0.025004876777529716, "train_loss_llm": 0.0, "grad_norm": 0.13246022164821625, "global_step": 34844, "epoch": 967, "lr": 0.001} +{"train_loss": 0.0317595899105072, "train_loss_bc": 0.0317595899105072, "train_loss_llm": 0.0, "grad_norm": 0.053583789616823196, "global_step": 34845, "epoch": 967, "lr": 0.001} +{"train_loss": 0.031160905957221985, "train_loss_bc": 0.031150903552770615, "train_loss_llm": 0.00100022554397583, "grad_norm": 0.07745710760354996, "global_step": 34846, "epoch": 967, "lr": 0.001} +{"train_loss": 0.02908595646214154, "train_loss_bc": 0.0276938509196043, "train_loss_llm": 0.0, "grad_norm": 0.10457815229892731, "global_step": 34847, "epoch": 967, "lr": 0.001, "val_loss": 0.10944877564907074} +{"train_loss": 0.026412224397063255, "train_loss_bc": 0.026396963745355606, "train_loss_llm": 0.0015261173248291016, "grad_norm": 0.12832042574882507, "global_step": 34848, "epoch": 968, "lr": 0.001} +{"train_loss": 0.03123038448393345, "train_loss_bc": 0.03123038448393345, "train_loss_llm": 0.0, "grad_norm": 0.045052286237478256, "global_step": 34849, "epoch": 968, "lr": 0.001} +{"train_loss": 0.03577188774943352, "train_loss_bc": 0.03577188774943352, "train_loss_llm": 0.0, "grad_norm": 0.09269377589225769, "global_step": 34850, "epoch": 968, "lr": 0.001} +{"train_loss": 0.026161132380366325, "train_loss_bc": 0.026161132380366325, "train_loss_llm": 0.0, "grad_norm": 0.11046715080738068, "global_step": 34851, "epoch": 968, "lr": 0.001} +{"train_loss": 0.024982884526252747, "train_loss_bc": 0.024979207664728165, "train_loss_llm": 0.00036770105361938477, "grad_norm": 0.14296098053455353, "global_step": 34852, "epoch": 968, "lr": 0.001} +{"train_loss": 0.027375051751732826, "train_loss_bc": 0.027375051751732826, "train_loss_llm": 0.0, "grad_norm": 0.03210259601473808, "global_step": 34853, "epoch": 968, "lr": 0.001} +{"train_loss": 0.02549925446510315, "train_loss_bc": 0.025484511628746986, "train_loss_llm": 0.0014743208885192871, "grad_norm": 0.05917035788297653, "global_step": 34854, "epoch": 968, "lr": 0.001} +{"train_loss": 0.027197927236557007, "train_loss_bc": 0.027197927236557007, "train_loss_llm": 0.0, "grad_norm": 0.07857881486415863, "global_step": 34855, "epoch": 968, "lr": 0.001} +{"train_loss": 0.029313260689377785, "train_loss_bc": 0.029313260689377785, "train_loss_llm": 0.0, "grad_norm": 0.11693564057350159, "global_step": 34856, "epoch": 968, "lr": 0.001} +{"train_loss": 0.03404829651117325, "train_loss_bc": 0.03404829651117325, "train_loss_llm": 0.0, "grad_norm": 0.03876898065209389, "global_step": 34857, "epoch": 968, "lr": 0.001} +{"train_loss": 0.03218071907758713, "train_loss_bc": 0.03218071907758713, "train_loss_llm": 0.0, "grad_norm": 0.06662625074386597, "global_step": 34858, "epoch": 968, "lr": 0.001} +{"train_loss": 0.026968656107783318, "train_loss_bc": 0.026968656107783318, "train_loss_llm": 0.0, "grad_norm": 0.08945098519325256, "global_step": 34859, "epoch": 968, "lr": 0.001} +{"train_loss": 0.024306315928697586, "train_loss_bc": 0.024300076067447662, "train_loss_llm": 0.0006240010261535645, "grad_norm": 0.11014577001333237, "global_step": 34860, "epoch": 968, "lr": 0.001} +{"train_loss": 0.02860904671251774, "train_loss_bc": 0.02860904671251774, "train_loss_llm": 0.0, "grad_norm": 0.0442064069211483, "global_step": 34861, "epoch": 968, "lr": 0.001} +{"train_loss": 0.033631302416324615, "train_loss_bc": 0.03362079709768295, "train_loss_llm": 0.0010505616664886475, "grad_norm": 0.06974972039461136, "global_step": 34862, "epoch": 968, "lr": 0.001} +{"train_loss": 0.03287851810455322, "train_loss_bc": 0.03286197781562805, "train_loss_llm": 0.0016539692878723145, "grad_norm": 0.10140661150217056, "global_step": 34863, "epoch": 968, "lr": 0.001} +{"train_loss": 0.029526403173804283, "train_loss_bc": 0.029526403173804283, "train_loss_llm": 0.0, "grad_norm": 0.1321231871843338, "global_step": 34864, "epoch": 968, "lr": 0.001} +{"train_loss": 0.027468914166092873, "train_loss_bc": 0.027468914166092873, "train_loss_llm": 0.0, "grad_norm": 0.0413331612944603, "global_step": 34865, "epoch": 968, "lr": 0.001} +{"train_loss": 0.027658337727189064, "train_loss_bc": 0.027658337727189064, "train_loss_llm": 0.0, "grad_norm": 0.07408725470304489, "global_step": 34866, "epoch": 968, "lr": 0.001} +{"train_loss": 0.030964113771915436, "train_loss_bc": 0.030964113771915436, "train_loss_llm": 0.0, "grad_norm": 0.10707515478134155, "global_step": 34867, "epoch": 968, "lr": 0.001} +{"train_loss": 0.03077242523431778, "train_loss_bc": 0.03077242523431778, "train_loss_llm": 0.0, "grad_norm": 0.13059546053409576, "global_step": 34868, "epoch": 968, "lr": 0.001} +{"train_loss": 0.028660442680120468, "train_loss_bc": 0.028660442680120468, "train_loss_llm": 0.0, "grad_norm": 0.03512328490614891, "global_step": 34869, "epoch": 968, "lr": 0.001} +{"train_loss": 0.031417395919561386, "train_loss_bc": 0.031417395919561386, "train_loss_llm": 0.0, "grad_norm": 0.05391733720898628, "global_step": 34870, "epoch": 968, "lr": 0.001} +{"train_loss": 0.03293522074818611, "train_loss_bc": 0.03293522074818611, "train_loss_llm": 0.0, "grad_norm": 0.07542739063501358, "global_step": 34871, "epoch": 968, "lr": 0.001} +{"train_loss": 0.02344861999154091, "train_loss_bc": 0.02344861999154091, "train_loss_llm": 0.0, "grad_norm": 0.09106225520372391, "global_step": 34872, "epoch": 968, "lr": 0.001} +{"train_loss": 0.030934471637010574, "train_loss_bc": 0.03093119151890278, "train_loss_llm": 0.0003280937671661377, "grad_norm": 0.038900598883628845, "global_step": 34873, "epoch": 968, "lr": 0.001} +{"train_loss": 0.03275172412395477, "train_loss_bc": 0.03275172412395477, "train_loss_llm": 0.0, "grad_norm": 0.06651487201452255, "global_step": 34874, "epoch": 968, "lr": 0.001} +{"train_loss": 0.030388684943318367, "train_loss_bc": 0.030388684943318367, "train_loss_llm": 0.0, "grad_norm": 0.08513146638870239, "global_step": 34875, "epoch": 968, "lr": 0.001} +{"train_loss": 0.026759346947073936, "train_loss_bc": 0.026746870949864388, "train_loss_llm": 0.0012475848197937012, "grad_norm": 0.1056755855679512, "global_step": 34876, "epoch": 968, "lr": 0.001} +{"train_loss": 0.02321946620941162, "train_loss_bc": 0.02321946620941162, "train_loss_llm": 0.0, "grad_norm": 0.031026331707835197, "global_step": 34877, "epoch": 968, "lr": 0.001} +{"train_loss": 0.026439815759658813, "train_loss_bc": 0.026435481384396553, "train_loss_llm": 0.00043335556983947754, "grad_norm": 0.04029487445950508, "global_step": 34878, "epoch": 968, "lr": 0.001} +{"train_loss": 0.024111106991767883, "train_loss_bc": 0.02410879172384739, "train_loss_llm": 0.00023159384727478027, "grad_norm": 0.058998070657253265, "global_step": 34879, "epoch": 968, "lr": 0.001} +{"train_loss": 0.02912868745625019, "train_loss_bc": 0.02912868745625019, "train_loss_llm": 0.0, "grad_norm": 0.0809079259634018, "global_step": 34880, "epoch": 968, "lr": 0.001} +{"train_loss": 0.0264690313488245, "train_loss_bc": 0.0264690313488245, "train_loss_llm": 0.0, "grad_norm": 0.03887707740068436, "global_step": 34881, "epoch": 968, "lr": 0.001} +{"train_loss": 0.030974455177783966, "train_loss_bc": 0.030974455177783966, "train_loss_llm": 0.0, "grad_norm": 0.06284727156162262, "global_step": 34882, "epoch": 968, "lr": 0.001} +{"train_loss": 0.02890477345014612, "train_loss_bc": 0.029976317659020424, "train_loss_llm": 0.0, "grad_norm": 0.08811145275831223, "global_step": 34883, "epoch": 968, "lr": 0.001, "val_loss": 0.10978489369153976} +{"train_loss": 0.027622072026133537, "train_loss_bc": 0.027622072026133537, "train_loss_llm": 0.0, "grad_norm": 0.12221897393465042, "global_step": 34884, "epoch": 969, "lr": 0.001} +{"train_loss": 0.03002117946743965, "train_loss_bc": 0.03002117946743965, "train_loss_llm": 0.0, "grad_norm": 0.036319904029369354, "global_step": 34885, "epoch": 969, "lr": 0.001} +{"train_loss": 0.030866531655192375, "train_loss_bc": 0.030851513147354126, "train_loss_llm": 0.0015017688274383545, "grad_norm": 0.06452827900648117, "global_step": 34886, "epoch": 969, "lr": 0.001} +{"train_loss": 0.03127150982618332, "train_loss_bc": 0.03125440329313278, "train_loss_llm": 0.0017107129096984863, "grad_norm": 0.09672223776578903, "global_step": 34887, "epoch": 969, "lr": 0.001} +{"train_loss": 0.02965756319463253, "train_loss_bc": 0.02965756319463253, "train_loss_llm": 0.0, "grad_norm": 0.13551749289035797, "global_step": 34888, "epoch": 969, "lr": 0.001} +{"train_loss": 0.029341191053390503, "train_loss_bc": 0.029341191053390503, "train_loss_llm": 0.0, "grad_norm": 0.04562944546341896, "global_step": 34889, "epoch": 969, "lr": 0.001} +{"train_loss": 0.026493951678276062, "train_loss_bc": 0.026493951678276062, "train_loss_llm": 0.0, "grad_norm": 0.06932235509157181, "global_step": 34890, "epoch": 969, "lr": 0.001} +{"train_loss": 0.029090911149978638, "train_loss_bc": 0.029090911149978638, "train_loss_llm": 0.0, "grad_norm": 0.0885365903377533, "global_step": 34891, "epoch": 969, "lr": 0.001} +{"train_loss": 0.031547144055366516, "train_loss_bc": 0.03153721243143082, "train_loss_llm": 0.0009930133819580078, "grad_norm": 0.10718279331922531, "global_step": 34892, "epoch": 969, "lr": 0.001} +{"train_loss": 0.02482583560049534, "train_loss_bc": 0.02481498382985592, "train_loss_llm": 0.0010852515697479248, "grad_norm": 0.030678139999508858, "global_step": 34893, "epoch": 969, "lr": 0.001} +{"train_loss": 0.03105895034968853, "train_loss_bc": 0.03104708343744278, "train_loss_llm": 0.0011866986751556396, "grad_norm": 0.070391446352005, "global_step": 34894, "epoch": 969, "lr": 0.001} +{"train_loss": 0.029153205454349518, "train_loss_bc": 0.029153205454349518, "train_loss_llm": 0.0, "grad_norm": 0.09184403717517853, "global_step": 34895, "epoch": 969, "lr": 0.001} +{"train_loss": 0.027653083205223083, "train_loss_bc": 0.027652697637677193, "train_loss_llm": 3.8564205169677734e-05, "grad_norm": 0.1143706813454628, "global_step": 34896, "epoch": 969, "lr": 0.001} +{"train_loss": 0.032198548316955566, "train_loss_bc": 0.032198548316955566, "train_loss_llm": 0.0, "grad_norm": 0.038964662700891495, "global_step": 34897, "epoch": 969, "lr": 0.001} +{"train_loss": 0.027264060452580452, "train_loss_bc": 0.027264060452580452, "train_loss_llm": 0.0, "grad_norm": 0.0617850236594677, "global_step": 34898, "epoch": 969, "lr": 0.001} +{"train_loss": 0.02628237195312977, "train_loss_bc": 0.02628237195312977, "train_loss_llm": 0.0, "grad_norm": 0.08775920420885086, "global_step": 34899, "epoch": 969, "lr": 0.001} +{"train_loss": 0.03219887241721153, "train_loss_bc": 0.03219887241721153, "train_loss_llm": 0.0, "grad_norm": 0.12401019036769867, "global_step": 34900, "epoch": 969, "lr": 0.001} +{"train_loss": 0.029450487345457077, "train_loss_bc": 0.029450487345457077, "train_loss_llm": 0.0, "grad_norm": 0.03795408457517624, "global_step": 34901, "epoch": 969, "lr": 0.001} +{"train_loss": 0.024906985461711884, "train_loss_bc": 0.024906985461711884, "train_loss_llm": 0.0, "grad_norm": 0.057692158967256546, "global_step": 34902, "epoch": 969, "lr": 0.001} +{"train_loss": 0.028421252965927124, "train_loss_bc": 0.028421252965927124, "train_loss_llm": 0.0, "grad_norm": 0.07343863695859909, "global_step": 34903, "epoch": 969, "lr": 0.001} +{"train_loss": 0.026440350338816643, "train_loss_bc": 0.026439853012561798, "train_loss_llm": 4.979968070983887e-05, "grad_norm": 0.0854761078953743, "global_step": 34904, "epoch": 969, "lr": 0.001} +{"train_loss": 0.024834010750055313, "train_loss_bc": 0.024834010750055313, "train_loss_llm": 0.0, "grad_norm": 0.04081384837627411, "global_step": 34905, "epoch": 969, "lr": 0.001} +{"train_loss": 0.025851139798760414, "train_loss_bc": 0.025846052914857864, "train_loss_llm": 0.0005086958408355713, "grad_norm": 0.06073629856109619, "global_step": 34906, "epoch": 969, "lr": 0.001} +{"train_loss": 0.03230203315615654, "train_loss_bc": 0.03230203315615654, "train_loss_llm": 0.0, "grad_norm": 0.08294127881526947, "global_step": 34907, "epoch": 969, "lr": 0.001} +{"train_loss": 0.029973087832331657, "train_loss_bc": 0.029973087832331657, "train_loss_llm": 0.0, "grad_norm": 0.10913398116827011, "global_step": 34908, "epoch": 969, "lr": 0.001} +{"train_loss": 0.028884638100862503, "train_loss_bc": 0.028884638100862503, "train_loss_llm": 0.0, "grad_norm": 0.03472290188074112, "global_step": 34909, "epoch": 969, "lr": 0.001} +{"train_loss": 0.025767769664525986, "train_loss_bc": 0.025767769664525986, "train_loss_llm": 0.0, "grad_norm": 0.06694280356168747, "global_step": 34910, "epoch": 969, "lr": 0.001} +{"train_loss": 0.03003082051873207, "train_loss_bc": 0.03003082051873207, "train_loss_llm": 0.0, "grad_norm": 0.0794086903333664, "global_step": 34911, "epoch": 969, "lr": 0.001} +{"train_loss": 0.028772829100489616, "train_loss_bc": 0.028772829100489616, "train_loss_llm": 0.0, "grad_norm": 0.1026199534535408, "global_step": 34912, "epoch": 969, "lr": 0.001} +{"train_loss": 0.026869045570492744, "train_loss_bc": 0.02682601474225521, "train_loss_llm": 0.004303157329559326, "grad_norm": 0.03521468862891197, "global_step": 34913, "epoch": 969, "lr": 0.001} +{"train_loss": 0.03223847225308418, "train_loss_bc": 0.0322294719517231, "train_loss_llm": 0.0009000897407531738, "grad_norm": 0.06451541930437088, "global_step": 34914, "epoch": 969, "lr": 0.001} +{"train_loss": 0.03208677098155022, "train_loss_bc": 0.032081492245197296, "train_loss_llm": 0.0005280077457427979, "grad_norm": 0.08711861819028854, "global_step": 34915, "epoch": 969, "lr": 0.001} +{"train_loss": 0.025845088064670563, "train_loss_bc": 0.02584175579249859, "train_loss_llm": 0.0003331899642944336, "grad_norm": 0.10157755017280579, "global_step": 34916, "epoch": 969, "lr": 0.001} +{"train_loss": 0.02802180126309395, "train_loss_bc": 0.02802180126309395, "train_loss_llm": 0.0, "grad_norm": 0.042112674564123154, "global_step": 34917, "epoch": 969, "lr": 0.001} +{"train_loss": 0.032976534217596054, "train_loss_bc": 0.03296960890293121, "train_loss_llm": 0.0006924271583557129, "grad_norm": 0.06133998557925224, "global_step": 34918, "epoch": 969, "lr": 0.001} +{"train_loss": 0.02883542303202881, "train_loss_bc": 0.027855129912495613, "train_loss_llm": 0.0, "grad_norm": 0.08899826556444168, "global_step": 34919, "epoch": 969, "lr": 0.001, "val_loss": 0.1105140745639801} +{"train_loss": 0.023306945338845253, "train_loss_bc": 0.023306945338845253, "train_loss_llm": 0.0, "grad_norm": 0.1068439856171608, "global_step": 34920, "epoch": 970, "lr": 0.001} +{"train_loss": 0.032393794506788254, "train_loss_bc": 0.032393794506788254, "train_loss_llm": 0.0, "grad_norm": 0.041898004710674286, "global_step": 34921, "epoch": 970, "lr": 0.001} +{"train_loss": 0.02787703089416027, "train_loss_bc": 0.02787703089416027, "train_loss_llm": 0.0, "grad_norm": 0.062297310680150986, "global_step": 34922, "epoch": 970, "lr": 0.001} +{"train_loss": 0.025831641629338264, "train_loss_bc": 0.025827918201684952, "train_loss_llm": 0.00037235021591186523, "grad_norm": 0.08169972151517868, "global_step": 34923, "epoch": 970, "lr": 0.001} +{"train_loss": 0.0303486455231905, "train_loss_bc": 0.03033721260726452, "train_loss_llm": 0.001143336296081543, "grad_norm": 0.09344416111707687, "global_step": 34924, "epoch": 970, "lr": 0.001} +{"train_loss": 0.028293030336499214, "train_loss_bc": 0.028293030336499214, "train_loss_llm": 0.0, "grad_norm": 0.03952288255095482, "global_step": 34925, "epoch": 970, "lr": 0.001} +{"train_loss": 0.027461370453238487, "train_loss_bc": 0.027448350563645363, "train_loss_llm": 0.0013019740581512451, "grad_norm": 0.07367033511400223, "global_step": 34926, "epoch": 970, "lr": 0.001} +{"train_loss": 0.026155726984143257, "train_loss_bc": 0.026155726984143257, "train_loss_llm": 0.0, "grad_norm": 0.08321405202150345, "global_step": 34927, "epoch": 970, "lr": 0.001} +{"train_loss": 0.02613835036754608, "train_loss_bc": 0.026125047355890274, "train_loss_llm": 0.0013303160667419434, "grad_norm": 0.09658821672201157, "global_step": 34928, "epoch": 970, "lr": 0.001} +{"train_loss": 0.029874084517359734, "train_loss_bc": 0.029874084517359734, "train_loss_llm": 0.0, "grad_norm": 0.044851530343294144, "global_step": 34929, "epoch": 970, "lr": 0.001} +{"train_loss": 0.027940617874264717, "train_loss_bc": 0.027940617874264717, "train_loss_llm": 0.0, "grad_norm": 0.06396440416574478, "global_step": 34930, "epoch": 970, "lr": 0.001} +{"train_loss": 0.034342825412750244, "train_loss_bc": 0.034342825412750244, "train_loss_llm": 0.0, "grad_norm": 0.10384976863861084, "global_step": 34931, "epoch": 970, "lr": 0.001} +{"train_loss": 0.03496190533041954, "train_loss_bc": 0.034955017268657684, "train_loss_llm": 0.0006888508796691895, "grad_norm": 0.1394764930009842, "global_step": 34932, "epoch": 970, "lr": 0.001} +{"train_loss": 0.021885564550757408, "train_loss_bc": 0.021875804290175438, "train_loss_llm": 0.0009759366512298584, "grad_norm": 0.03219226747751236, "global_step": 34933, "epoch": 970, "lr": 0.001} +{"train_loss": 0.025087138637900352, "train_loss_bc": 0.0250803604722023, "train_loss_llm": 0.0006778836250305176, "grad_norm": 0.05481081083416939, "global_step": 34934, "epoch": 970, "lr": 0.001} +{"train_loss": 0.029780864715576172, "train_loss_bc": 0.029779933393001556, "train_loss_llm": 9.310245513916016e-05, "grad_norm": 0.07580077648162842, "global_step": 34935, "epoch": 970, "lr": 0.001} +{"train_loss": 0.030220672488212585, "train_loss_bc": 0.030220672488212585, "train_loss_llm": 0.0, "grad_norm": 0.10004854947328568, "global_step": 34936, "epoch": 970, "lr": 0.001} +{"train_loss": 0.02848106436431408, "train_loss_bc": 0.02848106436431408, "train_loss_llm": 0.0, "grad_norm": 0.048160430043935776, "global_step": 34937, "epoch": 970, "lr": 0.001} +{"train_loss": 0.033224619925022125, "train_loss_bc": 0.033224619925022125, "train_loss_llm": 0.0, "grad_norm": 0.07386644184589386, "global_step": 34938, "epoch": 970, "lr": 0.001} +{"train_loss": 0.026656297966837883, "train_loss_bc": 0.0266468096524477, "train_loss_llm": 0.0009488463401794434, "grad_norm": 0.08723825961351395, "global_step": 34939, "epoch": 970, "lr": 0.001} +{"train_loss": 0.03466347977519035, "train_loss_bc": 0.03466347977519035, "train_loss_llm": 0.0, "grad_norm": 0.10579835623502731, "global_step": 34940, "epoch": 970, "lr": 0.001} +{"train_loss": 0.028344521299004555, "train_loss_bc": 0.028344521299004555, "train_loss_llm": 0.0, "grad_norm": 0.033891674131155014, "global_step": 34941, "epoch": 970, "lr": 0.001} +{"train_loss": 0.035419728606939316, "train_loss_bc": 0.035419728606939316, "train_loss_llm": 0.0, "grad_norm": 0.06717938929796219, "global_step": 34942, "epoch": 970, "lr": 0.001} +{"train_loss": 0.026307575404644012, "train_loss_bc": 0.026307575404644012, "train_loss_llm": 0.0, "grad_norm": 0.09321276843547821, "global_step": 34943, "epoch": 970, "lr": 0.001} +{"train_loss": 0.02682812698185444, "train_loss_bc": 0.02682812698185444, "train_loss_llm": 0.0, "grad_norm": 0.11640235036611557, "global_step": 34944, "epoch": 970, "lr": 0.001} +{"train_loss": 0.032880883663892746, "train_loss_bc": 0.032880883663892746, "train_loss_llm": 0.0, "grad_norm": 0.03977474570274353, "global_step": 34945, "epoch": 970, "lr": 0.001} +{"train_loss": 0.030033448711037636, "train_loss_bc": 0.030033448711037636, "train_loss_llm": 0.0, "grad_norm": 0.06933199614286423, "global_step": 34946, "epoch": 970, "lr": 0.001} +{"train_loss": 0.027091385796666145, "train_loss_bc": 0.027091385796666145, "train_loss_llm": 0.0, "grad_norm": 0.10590428858995438, "global_step": 34947, "epoch": 970, "lr": 0.001} +{"train_loss": 0.029062559828162193, "train_loss_bc": 0.029049666598439217, "train_loss_llm": 0.001289278268814087, "grad_norm": 0.12585730850696564, "global_step": 34948, "epoch": 970, "lr": 0.001} +{"train_loss": 0.025799252092838287, "train_loss_bc": 0.025799252092838287, "train_loss_llm": 0.0, "grad_norm": 0.03423049673438072, "global_step": 34949, "epoch": 970, "lr": 0.001} +{"train_loss": 0.02981538511812687, "train_loss_bc": 0.02981538511812687, "train_loss_llm": 0.0, "grad_norm": 0.05984918773174286, "global_step": 34950, "epoch": 970, "lr": 0.001} +{"train_loss": 0.030359141528606415, "train_loss_bc": 0.030359141528606415, "train_loss_llm": 0.0, "grad_norm": 0.0803716853260994, "global_step": 34951, "epoch": 970, "lr": 0.001} +{"train_loss": 0.03384723141789436, "train_loss_bc": 0.03384723141789436, "train_loss_llm": 0.0, "grad_norm": 0.11313942819833755, "global_step": 34952, "epoch": 970, "lr": 0.001} +{"train_loss": 0.026110081002116203, "train_loss_bc": 0.026110081002116203, "train_loss_llm": 0.0, "grad_norm": 0.03394037485122681, "global_step": 34953, "epoch": 970, "lr": 0.001} +{"train_loss": 0.02146575227379799, "train_loss_bc": 0.02146575227379799, "train_loss_llm": 0.0, "grad_norm": 0.05259808152914047, "global_step": 34954, "epoch": 970, "lr": 0.001} +{"train_loss": 0.02881391455108921, "train_loss_bc": 0.028997329995036125, "train_loss_llm": 0.0012848079204559326, "grad_norm": 0.08374353498220444, "global_step": 34955, "epoch": 970, "lr": 0.001, "train/cumulative_reward": 89.22013330660144, "train/mean_score": 15.819668229136742, "train/success_rate": 0.4, "test/cumulative_reward": 40.247509522213804, "test/mean_score": 9.856770164118661, "test/success_rate": 0.26, "val_loss": 0.1089187040925026, "train_action_mse_error": 0.031869687139987946} +{"train_loss": 0.03150191530585289, "train_loss_bc": 0.03150191530585289, "train_loss_llm": 0.0, "grad_norm": 0.10401654243469238, "global_step": 34956, "epoch": 971, "lr": 0.001} +{"train_loss": 0.03444990515708923, "train_loss_bc": 0.03444990515708923, "train_loss_llm": 0.0, "grad_norm": 0.045933451503515244, "global_step": 34957, "epoch": 971, "lr": 0.001} +{"train_loss": 0.029630884528160095, "train_loss_bc": 0.029630884528160095, "train_loss_llm": 0.0, "grad_norm": 0.0658956915140152, "global_step": 34958, "epoch": 971, "lr": 0.001} +{"train_loss": 0.029182059690356255, "train_loss_bc": 0.029182059690356255, "train_loss_llm": 0.0, "grad_norm": 0.0999375730752945, "global_step": 34959, "epoch": 971, "lr": 0.001} +{"train_loss": 0.0264408141374588, "train_loss_bc": 0.0264408141374588, "train_loss_llm": 0.0, "grad_norm": 0.11784806847572327, "global_step": 34960, "epoch": 971, "lr": 0.001} +{"train_loss": 0.029718101024627686, "train_loss_bc": 0.029718101024627686, "train_loss_llm": 0.0, "grad_norm": 0.03434036299586296, "global_step": 34961, "epoch": 971, "lr": 0.001} +{"train_loss": 0.023238223046064377, "train_loss_bc": 0.023238223046064377, "train_loss_llm": 0.0, "grad_norm": 0.04437021166086197, "global_step": 34962, "epoch": 971, "lr": 0.001} +{"train_loss": 0.031256817281246185, "train_loss_bc": 0.031245889142155647, "train_loss_llm": 0.0010927319526672363, "grad_norm": 0.06696449965238571, "global_step": 34963, "epoch": 971, "lr": 0.001} +{"train_loss": 0.026240399107336998, "train_loss_bc": 0.026240399107336998, "train_loss_llm": 0.0, "grad_norm": 0.08445370197296143, "global_step": 34964, "epoch": 971, "lr": 0.001} +{"train_loss": 0.027815179899334908, "train_loss_bc": 0.027815179899334908, "train_loss_llm": 0.0, "grad_norm": 0.03605690225958824, "global_step": 34965, "epoch": 971, "lr": 0.001} +{"train_loss": 0.027166243642568588, "train_loss_bc": 0.027166243642568588, "train_loss_llm": 0.0, "grad_norm": 0.0552244670689106, "global_step": 34966, "epoch": 971, "lr": 0.001} +{"train_loss": 0.02410157211124897, "train_loss_bc": 0.02410157211124897, "train_loss_llm": 0.0, "grad_norm": 0.08186299353837967, "global_step": 34967, "epoch": 971, "lr": 0.001} +{"train_loss": 0.031254883855581284, "train_loss_bc": 0.031254883855581284, "train_loss_llm": 0.0, "grad_norm": 0.1243649274110794, "global_step": 34968, "epoch": 971, "lr": 0.001} +{"train_loss": 0.03012383170425892, "train_loss_bc": 0.03012383170425892, "train_loss_llm": 0.0, "grad_norm": 0.05680945888161659, "global_step": 34969, "epoch": 971, "lr": 0.001} +{"train_loss": 0.025881191715598106, "train_loss_bc": 0.025881191715598106, "train_loss_llm": 0.0, "grad_norm": 0.0785660445690155, "global_step": 34970, "epoch": 971, "lr": 0.001} +{"train_loss": 0.023194972425699234, "train_loss_bc": 0.023194972425699234, "train_loss_llm": 0.0, "grad_norm": 0.10100162029266357, "global_step": 34971, "epoch": 971, "lr": 0.001} +{"train_loss": 0.02639039047062397, "train_loss_bc": 0.026367561891674995, "train_loss_llm": 0.002282947301864624, "grad_norm": 0.12070894986391068, "global_step": 34972, "epoch": 971, "lr": 0.001} +{"train_loss": 0.029403774067759514, "train_loss_bc": 0.02938893809914589, "train_loss_llm": 0.0014836788177490234, "grad_norm": 0.04275284707546234, "global_step": 34973, "epoch": 971, "lr": 0.001} +{"train_loss": 0.03248574584722519, "train_loss_bc": 0.03247971460223198, "train_loss_llm": 0.00060310959815979, "grad_norm": 0.07514321058988571, "global_step": 34974, "epoch": 971, "lr": 0.001} +{"train_loss": 0.0329752191901207, "train_loss_bc": 0.0329752191901207, "train_loss_llm": 0.0, "grad_norm": 0.08240218460559845, "global_step": 34975, "epoch": 971, "lr": 0.001} +{"train_loss": 0.030598964542150497, "train_loss_bc": 0.030598143115639687, "train_loss_llm": 8.213520050048828e-05, "grad_norm": 0.10127934068441391, "global_step": 34976, "epoch": 971, "lr": 0.001} +{"train_loss": 0.021267713978886604, "train_loss_bc": 0.021259047091007233, "train_loss_llm": 0.000866621732711792, "grad_norm": 0.042890626937150955, "global_step": 34977, "epoch": 971, "lr": 0.001} +{"train_loss": 0.027333606034517288, "train_loss_bc": 0.027333606034517288, "train_loss_llm": 0.0, "grad_norm": 0.07374851405620575, "global_step": 34978, "epoch": 971, "lr": 0.001} +{"train_loss": 0.032347217202186584, "train_loss_bc": 0.032347217202186584, "train_loss_llm": 0.0, "grad_norm": 0.09300830960273743, "global_step": 34979, "epoch": 971, "lr": 0.001} +{"train_loss": 0.03887372463941574, "train_loss_bc": 0.03887372463941574, "train_loss_llm": 0.0, "grad_norm": 0.13392001390457153, "global_step": 34980, "epoch": 971, "lr": 0.001} +{"train_loss": 0.032804012298583984, "train_loss_bc": 0.032796360552310944, "train_loss_llm": 0.0007651746273040771, "grad_norm": 0.04462772235274315, "global_step": 34981, "epoch": 971, "lr": 0.001} +{"train_loss": 0.023434218019247055, "train_loss_bc": 0.023431630805134773, "train_loss_llm": 0.0002587735652923584, "grad_norm": 0.05128978192806244, "global_step": 34982, "epoch": 971, "lr": 0.001} +{"train_loss": 0.03231848403811455, "train_loss_bc": 0.03231848403811455, "train_loss_llm": 0.0, "grad_norm": 0.0713130533695221, "global_step": 34983, "epoch": 971, "lr": 0.001} +{"train_loss": 0.03506556153297424, "train_loss_bc": 0.03506556153297424, "train_loss_llm": 0.0, "grad_norm": 0.0952005684375763, "global_step": 34984, "epoch": 971, "lr": 0.001} +{"train_loss": 0.02520865760743618, "train_loss_bc": 0.02520865760743618, "train_loss_llm": 0.0, "grad_norm": 0.051067233085632324, "global_step": 34985, "epoch": 971, "lr": 0.001} +{"train_loss": 0.026873189955949783, "train_loss_bc": 0.026873189955949783, "train_loss_llm": 0.0, "grad_norm": 0.08943179249763489, "global_step": 34986, "epoch": 971, "lr": 0.001} +{"train_loss": 0.029561396688222885, "train_loss_bc": 0.029539313167333603, "train_loss_llm": 0.0022084414958953857, "grad_norm": 0.1128666028380394, "global_step": 34987, "epoch": 971, "lr": 0.001} +{"train_loss": 0.025164984166622162, "train_loss_bc": 0.025164984166622162, "train_loss_llm": 0.0, "grad_norm": 0.13818268477916718, "global_step": 34988, "epoch": 971, "lr": 0.001} +{"train_loss": 0.030500397086143494, "train_loss_bc": 0.03050003945827484, "train_loss_llm": 3.5822391510009766e-05, "grad_norm": 0.049195658415555954, "global_step": 34989, "epoch": 971, "lr": 0.001} +{"train_loss": 0.021712573245167732, "train_loss_bc": 0.02170732617378235, "train_loss_llm": 0.0005246400833129883, "grad_norm": 0.06424827128648758, "global_step": 34990, "epoch": 971, "lr": 0.001} +{"train_loss": 0.028735559278478224, "train_loss_bc": 0.02895445004105568, "train_loss_llm": 0.0008859634399414062, "grad_norm": 0.08839309215545654, "global_step": 34991, "epoch": 971, "lr": 0.001, "val_loss": 0.11005333065986633} +{"train_loss": 0.0234947819262743, "train_loss_bc": 0.0234947819262743, "train_loss_llm": 0.0, "grad_norm": 0.11640845984220505, "global_step": 34992, "epoch": 972, "lr": 0.001} +{"train_loss": 0.027690455317497253, "train_loss_bc": 0.02768447995185852, "train_loss_llm": 0.0005974769592285156, "grad_norm": 0.03362450376152992, "global_step": 34993, "epoch": 972, "lr": 0.001} +{"train_loss": 0.029052341356873512, "train_loss_bc": 0.029052341356873512, "train_loss_llm": 0.0, "grad_norm": 0.05872993543744087, "global_step": 34994, "epoch": 972, "lr": 0.001} +{"train_loss": 0.023834025487303734, "train_loss_bc": 0.02382950857281685, "train_loss_llm": 0.00045174360275268555, "grad_norm": 0.07513108849525452, "global_step": 34995, "epoch": 972, "lr": 0.001} +{"train_loss": 0.02794209122657776, "train_loss_bc": 0.02794209122657776, "train_loss_llm": 0.0, "grad_norm": 0.09000915288925171, "global_step": 34996, "epoch": 972, "lr": 0.001} +{"train_loss": 0.027391446754336357, "train_loss_bc": 0.027391446754336357, "train_loss_llm": 0.0, "grad_norm": 0.04938916862010956, "global_step": 34997, "epoch": 972, "lr": 0.001} +{"train_loss": 0.028283601626753807, "train_loss_bc": 0.028283601626753807, "train_loss_llm": 0.0, "grad_norm": 0.08645380288362503, "global_step": 34998, "epoch": 972, "lr": 0.001} +{"train_loss": 0.031233981251716614, "train_loss_bc": 0.031233981251716614, "train_loss_llm": 0.0, "grad_norm": 0.11592976003885269, "global_step": 34999, "epoch": 972, "lr": 0.001} +{"train_loss": 0.0281998123973608, "train_loss_bc": 0.02819746546447277, "train_loss_llm": 0.00023478269577026367, "grad_norm": 0.13306954503059387, "global_step": 35000, "epoch": 972, "lr": 0.001} +{"train_loss": 0.029226170852780342, "train_loss_bc": 0.029215998947620392, "train_loss_llm": 0.0010171234607696533, "grad_norm": 0.034641120582818985, "global_step": 35001, "epoch": 972, "lr": 0.001} +{"train_loss": 0.031518615782260895, "train_loss_bc": 0.031518615782260895, "train_loss_llm": 0.0, "grad_norm": 0.054123494774103165, "global_step": 35002, "epoch": 972, "lr": 0.001} +{"train_loss": 0.03220629319548607, "train_loss_bc": 0.032204583287239075, "train_loss_llm": 0.0001710355281829834, "grad_norm": 0.07421965152025223, "global_step": 35003, "epoch": 972, "lr": 0.001} +{"train_loss": 0.028461819514632225, "train_loss_bc": 0.0284515880048275, "train_loss_llm": 0.0010231733322143555, "grad_norm": 0.10125748068094254, "global_step": 35004, "epoch": 972, "lr": 0.001} +{"train_loss": 0.02657664567232132, "train_loss_bc": 0.02657664567232132, "train_loss_llm": 0.0, "grad_norm": 0.03735252097249031, "global_step": 35005, "epoch": 972, "lr": 0.001} +{"train_loss": 0.02589454874396324, "train_loss_bc": 0.02589454874396324, "train_loss_llm": 0.0, "grad_norm": 0.05377676710486412, "global_step": 35006, "epoch": 972, "lr": 0.001} +{"train_loss": 0.026525387540459633, "train_loss_bc": 0.026525387540459633, "train_loss_llm": 0.0, "grad_norm": 0.07207932323217392, "global_step": 35007, "epoch": 972, "lr": 0.001} +{"train_loss": 0.028546323999762535, "train_loss_bc": 0.028546323999762535, "train_loss_llm": 0.0, "grad_norm": 0.09932789951562881, "global_step": 35008, "epoch": 972, "lr": 0.001} +{"train_loss": 0.027288757264614105, "train_loss_bc": 0.027288757264614105, "train_loss_llm": 0.0, "grad_norm": 0.04440227150917053, "global_step": 35009, "epoch": 972, "lr": 0.001} +{"train_loss": 0.02950839325785637, "train_loss_bc": 0.029502565041184425, "train_loss_llm": 0.0005829036235809326, "grad_norm": 0.07188694924116135, "global_step": 35010, "epoch": 972, "lr": 0.001} +{"train_loss": 0.025088565424084663, "train_loss_bc": 0.025088565424084663, "train_loss_llm": 0.0, "grad_norm": 0.09404686838388443, "global_step": 35011, "epoch": 972, "lr": 0.001} +{"train_loss": 0.03273281082510948, "train_loss_bc": 0.03273281082510948, "train_loss_llm": 0.0, "grad_norm": 0.11722776293754578, "global_step": 35012, "epoch": 972, "lr": 0.001} +{"train_loss": 0.03221141919493675, "train_loss_bc": 0.03220849856734276, "train_loss_llm": 0.00029221177101135254, "grad_norm": 0.04226851090788841, "global_step": 35013, "epoch": 972, "lr": 0.001} +{"train_loss": 0.03044999949634075, "train_loss_bc": 0.03044999949634075, "train_loss_llm": 0.0, "grad_norm": 0.06635358929634094, "global_step": 35014, "epoch": 972, "lr": 0.001} +{"train_loss": 0.028661364689469337, "train_loss_bc": 0.028661364689469337, "train_loss_llm": 0.0, "grad_norm": 0.08546388894319534, "global_step": 35015, "epoch": 972, "lr": 0.001} +{"train_loss": 0.028827333822846413, "train_loss_bc": 0.028821958228945732, "train_loss_llm": 0.000537574291229248, "grad_norm": 0.10659078508615494, "global_step": 35016, "epoch": 972, "lr": 0.001} +{"train_loss": 0.029032817110419273, "train_loss_bc": 0.02902253158390522, "train_loss_llm": 0.001028507947921753, "grad_norm": 0.03514506667852402, "global_step": 35017, "epoch": 972, "lr": 0.001} +{"train_loss": 0.027844654396176338, "train_loss_bc": 0.027839571237564087, "train_loss_llm": 0.0005082488059997559, "grad_norm": 0.05247221887111664, "global_step": 35018, "epoch": 972, "lr": 0.001} +{"train_loss": 0.030255448073148727, "train_loss_bc": 0.03024369478225708, "train_loss_llm": 0.0011753737926483154, "grad_norm": 0.0792233869433403, "global_step": 35019, "epoch": 972, "lr": 0.001} +{"train_loss": 0.027756467461586, "train_loss_bc": 0.027756467461586, "train_loss_llm": 0.0, "grad_norm": 0.1080944836139679, "global_step": 35020, "epoch": 972, "lr": 0.001} +{"train_loss": 0.02829890325665474, "train_loss_bc": 0.02829890325665474, "train_loss_llm": 0.0, "grad_norm": 0.053167954087257385, "global_step": 35021, "epoch": 972, "lr": 0.001} +{"train_loss": 0.026939038187265396, "train_loss_bc": 0.02691964991390705, "train_loss_llm": 0.0019388198852539062, "grad_norm": 0.07785437256097794, "global_step": 35022, "epoch": 972, "lr": 0.001} +{"train_loss": 0.0339188426733017, "train_loss_bc": 0.0339188426733017, "train_loss_llm": 0.0, "grad_norm": 0.12075639516115189, "global_step": 35023, "epoch": 972, "lr": 0.001} +{"train_loss": 0.028140783309936523, "train_loss_bc": 0.028140783309936523, "train_loss_llm": 0.0, "grad_norm": 0.1327231377363205, "global_step": 35024, "epoch": 972, "lr": 0.001} +{"train_loss": 0.026761706918478012, "train_loss_bc": 0.026761706918478012, "train_loss_llm": 0.0, "grad_norm": 0.03370233252644539, "global_step": 35025, "epoch": 972, "lr": 0.001} +{"train_loss": 0.029491063207387924, "train_loss_bc": 0.029491063207387924, "train_loss_llm": 0.0, "grad_norm": 0.051084816455841064, "global_step": 35026, "epoch": 972, "lr": 0.001} +{"train_loss": 0.028667929478817515, "train_loss_bc": 0.03275875002145767, "train_loss_llm": 0.0, "grad_norm": 0.07157627493143082, "global_step": 35027, "epoch": 972, "lr": 0.001, "val_loss": 0.10967250913381577} +{"train_loss": 0.02936074510216713, "train_loss_bc": 0.02936074510216713, "train_loss_llm": 0.0, "grad_norm": 0.07891006022691727, "global_step": 35028, "epoch": 973, "lr": 0.001} +{"train_loss": 0.02744976058602333, "train_loss_bc": 0.02744976058602333, "train_loss_llm": 0.0, "grad_norm": 0.040736570954322815, "global_step": 35029, "epoch": 973, "lr": 0.001} +{"train_loss": 0.0269089974462986, "train_loss_bc": 0.026892906054854393, "train_loss_llm": 0.0016090571880340576, "grad_norm": 0.05321253836154938, "global_step": 35030, "epoch": 973, "lr": 0.001} +{"train_loss": 0.03357890248298645, "train_loss_bc": 0.03357447683811188, "train_loss_llm": 0.0004425048828125, "grad_norm": 0.07927238941192627, "global_step": 35031, "epoch": 973, "lr": 0.001} +{"train_loss": 0.030658096075057983, "train_loss_bc": 0.030658096075057983, "train_loss_llm": 0.0, "grad_norm": 0.10564807057380676, "global_step": 35032, "epoch": 973, "lr": 0.001} +{"train_loss": 0.03132165968418121, "train_loss_bc": 0.03132165968418121, "train_loss_llm": 0.0, "grad_norm": 0.047497544437646866, "global_step": 35033, "epoch": 973, "lr": 0.001} +{"train_loss": 0.0275078434497118, "train_loss_bc": 0.0275078434497118, "train_loss_llm": 0.0, "grad_norm": 0.08270338177680969, "global_step": 35034, "epoch": 973, "lr": 0.001} +{"train_loss": 0.024914031848311424, "train_loss_bc": 0.024914031848311424, "train_loss_llm": 0.0, "grad_norm": 0.11111663281917572, "global_step": 35035, "epoch": 973, "lr": 0.001} +{"train_loss": 0.03380762040615082, "train_loss_bc": 0.03380762040615082, "train_loss_llm": 0.0, "grad_norm": 0.13659515976905823, "global_step": 35036, "epoch": 973, "lr": 0.001} +{"train_loss": 0.02801423892378807, "train_loss_bc": 0.02801423892378807, "train_loss_llm": 0.0, "grad_norm": 0.03985307738184929, "global_step": 35037, "epoch": 973, "lr": 0.001} +{"train_loss": 0.023489687591791153, "train_loss_bc": 0.023489687591791153, "train_loss_llm": 0.0, "grad_norm": 0.07269581407308578, "global_step": 35038, "epoch": 973, "lr": 0.001} +{"train_loss": 0.02947942167520523, "train_loss_bc": 0.02947942167520523, "train_loss_llm": 0.0, "grad_norm": 0.1063380166888237, "global_step": 35039, "epoch": 973, "lr": 0.001} +{"train_loss": 0.03126825392246246, "train_loss_bc": 0.03126825392246246, "train_loss_llm": 0.0, "grad_norm": 0.12934373319149017, "global_step": 35040, "epoch": 973, "lr": 0.001} +{"train_loss": 0.03625977784395218, "train_loss_bc": 0.03625977784395218, "train_loss_llm": 0.0, "grad_norm": 0.06155771389603615, "global_step": 35041, "epoch": 973, "lr": 0.001} +{"train_loss": 0.03342181444168091, "train_loss_bc": 0.03341113030910492, "train_loss_llm": 0.0010683536529541016, "grad_norm": 0.08116650581359863, "global_step": 35042, "epoch": 973, "lr": 0.001} +{"train_loss": 0.02990870177745819, "train_loss_bc": 0.02990870177745819, "train_loss_llm": 0.0, "grad_norm": 0.10246962308883667, "global_step": 35043, "epoch": 973, "lr": 0.001} +{"train_loss": 0.027654683217406273, "train_loss_bc": 0.0276443213224411, "train_loss_llm": 0.001036137342453003, "grad_norm": 0.12136030942201614, "global_step": 35044, "epoch": 973, "lr": 0.001} +{"train_loss": 0.027964239940047264, "train_loss_bc": 0.027964239940047264, "train_loss_llm": 0.0, "grad_norm": 0.04413753002882004, "global_step": 35045, "epoch": 973, "lr": 0.001} +{"train_loss": 0.0314071960747242, "train_loss_bc": 0.0314071960747242, "train_loss_llm": 0.0, "grad_norm": 0.06577492505311966, "global_step": 35046, "epoch": 973, "lr": 0.001} +{"train_loss": 0.025733308866620064, "train_loss_bc": 0.02573201060295105, "train_loss_llm": 0.00012990832328796387, "grad_norm": 0.07962432503700256, "global_step": 35047, "epoch": 973, "lr": 0.001} +{"train_loss": 0.03224055469036102, "train_loss_bc": 0.03224055469036102, "train_loss_llm": 0.0, "grad_norm": 0.09826622903347015, "global_step": 35048, "epoch": 973, "lr": 0.001} +{"train_loss": 0.027328211814165115, "train_loss_bc": 0.027328211814165115, "train_loss_llm": 0.0, "grad_norm": 0.03515113517642021, "global_step": 35049, "epoch": 973, "lr": 0.001} +{"train_loss": 0.030932823196053505, "train_loss_bc": 0.030928507447242737, "train_loss_llm": 0.0004316568374633789, "grad_norm": 0.06623759865760803, "global_step": 35050, "epoch": 973, "lr": 0.001} +{"train_loss": 0.028543541207909584, "train_loss_bc": 0.028543541207909584, "train_loss_llm": 0.0, "grad_norm": 0.08355829119682312, "global_step": 35051, "epoch": 973, "lr": 0.001} +{"train_loss": 0.028621412813663483, "train_loss_bc": 0.028621412813663483, "train_loss_llm": 0.0, "grad_norm": 0.11119986325502396, "global_step": 35052, "epoch": 973, "lr": 0.001} +{"train_loss": 0.02776619978249073, "train_loss_bc": 0.02776440419256687, "train_loss_llm": 0.00017952919006347656, "grad_norm": 0.04904424399137497, "global_step": 35053, "epoch": 973, "lr": 0.001} +{"train_loss": 0.030452346429228783, "train_loss_bc": 0.03044205717742443, "train_loss_llm": 0.0010289251804351807, "grad_norm": 0.07506871968507767, "global_step": 35054, "epoch": 973, "lr": 0.001} +{"train_loss": 0.026910101994872093, "train_loss_bc": 0.026910101994872093, "train_loss_llm": 0.0, "grad_norm": 0.09436757862567902, "global_step": 35055, "epoch": 973, "lr": 0.001} +{"train_loss": 0.02891196310520172, "train_loss_bc": 0.028908561915159225, "train_loss_llm": 0.0003401339054107666, "grad_norm": 0.11510634422302246, "global_step": 35056, "epoch": 973, "lr": 0.001} +{"train_loss": 0.024051601067185402, "train_loss_bc": 0.024051601067185402, "train_loss_llm": 0.0, "grad_norm": 0.03894136846065521, "global_step": 35057, "epoch": 973, "lr": 0.001} +{"train_loss": 0.026386596262454987, "train_loss_bc": 0.026386596262454987, "train_loss_llm": 0.0, "grad_norm": 0.07026703655719757, "global_step": 35058, "epoch": 973, "lr": 0.001} +{"train_loss": 0.029208704829216003, "train_loss_bc": 0.029207322746515274, "train_loss_llm": 0.00013816356658935547, "grad_norm": 0.09968520700931549, "global_step": 35059, "epoch": 973, "lr": 0.001} +{"train_loss": 0.02707449160516262, "train_loss_bc": 0.02706405520439148, "train_loss_llm": 0.001043558120727539, "grad_norm": 0.11687180399894714, "global_step": 35060, "epoch": 973, "lr": 0.001} +{"train_loss": 0.02569611743092537, "train_loss_bc": 0.02569611743092537, "train_loss_llm": 0.0, "grad_norm": 0.038212232291698456, "global_step": 35061, "epoch": 973, "lr": 0.001} +{"train_loss": 0.025851109996438026, "train_loss_bc": 0.025851109996438026, "train_loss_llm": 0.0, "grad_norm": 0.06178620830178261, "global_step": 35062, "epoch": 973, "lr": 0.001} +{"train_loss": 0.028610858445366222, "train_loss_bc": 0.019906146451830864, "train_loss_llm": 0.0, "grad_norm": 0.0750076100230217, "global_step": 35063, "epoch": 973, "lr": 0.001, "val_loss": 0.11000944674015045} +{"train_loss": 0.02453400008380413, "train_loss_bc": 0.024533452466130257, "train_loss_llm": 5.4717063903808594e-05, "grad_norm": 0.09150548279285431, "global_step": 35064, "epoch": 974, "lr": 0.001} +{"train_loss": 0.031661637127399445, "train_loss_bc": 0.031661637127399445, "train_loss_llm": 0.0, "grad_norm": 0.041176632046699524, "global_step": 35065, "epoch": 974, "lr": 0.001} +{"train_loss": 0.03009038418531418, "train_loss_bc": 0.030070148408412933, "train_loss_llm": 0.0020234882831573486, "grad_norm": 0.04848021641373634, "global_step": 35066, "epoch": 974, "lr": 0.001} +{"train_loss": 0.03148811683058739, "train_loss_bc": 0.03148811683058739, "train_loss_llm": 0.0, "grad_norm": 0.06899785250425339, "global_step": 35067, "epoch": 974, "lr": 0.001} +{"train_loss": 0.01967819593846798, "train_loss_bc": 0.01967819593846798, "train_loss_llm": 0.0, "grad_norm": 0.08796332031488419, "global_step": 35068, "epoch": 974, "lr": 0.001} +{"train_loss": 0.026643382385373116, "train_loss_bc": 0.026643382385373116, "train_loss_llm": 0.0, "grad_norm": 0.029607180505990982, "global_step": 35069, "epoch": 974, "lr": 0.001} +{"train_loss": 0.028101401403546333, "train_loss_bc": 0.028101401403546333, "train_loss_llm": 0.0, "grad_norm": 0.05019843950867653, "global_step": 35070, "epoch": 974, "lr": 0.001} +{"train_loss": 0.027365941554307938, "train_loss_bc": 0.027365941554307938, "train_loss_llm": 0.0, "grad_norm": 0.07053340971469879, "global_step": 35071, "epoch": 974, "lr": 0.001} +{"train_loss": 0.028510551899671555, "train_loss_bc": 0.028492018580436707, "train_loss_llm": 0.001853346824645996, "grad_norm": 0.09270676225423813, "global_step": 35072, "epoch": 974, "lr": 0.001} +{"train_loss": 0.020496627315878868, "train_loss_bc": 0.020496627315878868, "train_loss_llm": 0.0, "grad_norm": 0.03243548050522804, "global_step": 35073, "epoch": 974, "lr": 0.001} +{"train_loss": 0.02587767317891121, "train_loss_bc": 0.02587767317891121, "train_loss_llm": 0.0, "grad_norm": 0.06073055416345596, "global_step": 35074, "epoch": 974, "lr": 0.001} +{"train_loss": 0.027220051735639572, "train_loss_bc": 0.027220051735639572, "train_loss_llm": 0.0, "grad_norm": 0.08332037180662155, "global_step": 35075, "epoch": 974, "lr": 0.001} +{"train_loss": 0.024441882967948914, "train_loss_bc": 0.024441882967948914, "train_loss_llm": 0.0, "grad_norm": 0.10727450251579285, "global_step": 35076, "epoch": 974, "lr": 0.001} +{"train_loss": 0.028567463159561157, "train_loss_bc": 0.028567463159561157, "train_loss_llm": 0.0, "grad_norm": 0.03820372745394707, "global_step": 35077, "epoch": 974, "lr": 0.001} +{"train_loss": 0.028219005092978477, "train_loss_bc": 0.028219005092978477, "train_loss_llm": 0.0, "grad_norm": 0.06031404063105583, "global_step": 35078, "epoch": 974, "lr": 0.001} +{"train_loss": 0.03753449022769928, "train_loss_bc": 0.03753449022769928, "train_loss_llm": 0.0, "grad_norm": 0.10628566145896912, "global_step": 35079, "epoch": 974, "lr": 0.001} +{"train_loss": 0.026812588796019554, "train_loss_bc": 0.026812588796019554, "train_loss_llm": 0.0, "grad_norm": 0.1319122612476349, "global_step": 35080, "epoch": 974, "lr": 0.001} +{"train_loss": 0.03345803543925285, "train_loss_bc": 0.03345803543925285, "train_loss_llm": 0.0, "grad_norm": 0.03720217943191528, "global_step": 35081, "epoch": 974, "lr": 0.001} +{"train_loss": 0.026295166462659836, "train_loss_bc": 0.026295166462659836, "train_loss_llm": 0.0, "grad_norm": 0.052649013698101044, "global_step": 35082, "epoch": 974, "lr": 0.001} +{"train_loss": 0.03219587355852127, "train_loss_bc": 0.03219587355852127, "train_loss_llm": 0.0, "grad_norm": 0.07684383541345596, "global_step": 35083, "epoch": 974, "lr": 0.001} +{"train_loss": 0.03240518271923065, "train_loss_bc": 0.03240518271923065, "train_loss_llm": 0.0, "grad_norm": 0.11455458402633667, "global_step": 35084, "epoch": 974, "lr": 0.001} +{"train_loss": 0.025872891768813133, "train_loss_bc": 0.025872891768813133, "train_loss_llm": 0.0, "grad_norm": 0.038125015795230865, "global_step": 35085, "epoch": 974, "lr": 0.001} +{"train_loss": 0.030423682183027267, "train_loss_bc": 0.030423682183027267, "train_loss_llm": 0.0, "grad_norm": 0.05987450107932091, "global_step": 35086, "epoch": 974, "lr": 0.001} +{"train_loss": 0.029927322641015053, "train_loss_bc": 0.029927322641015053, "train_loss_llm": 0.0, "grad_norm": 0.08672091364860535, "global_step": 35087, "epoch": 974, "lr": 0.001} +{"train_loss": 0.030454907566308975, "train_loss_bc": 0.030454907566308975, "train_loss_llm": 0.0, "grad_norm": 0.10175982117652893, "global_step": 35088, "epoch": 974, "lr": 0.001} +{"train_loss": 0.02548985183238983, "train_loss_bc": 0.02548985183238983, "train_loss_llm": 0.0, "grad_norm": 0.04021202400326729, "global_step": 35089, "epoch": 974, "lr": 0.001} +{"train_loss": 0.02446695789694786, "train_loss_bc": 0.0244571752846241, "train_loss_llm": 0.0009781718254089355, "grad_norm": 0.0539669543504715, "global_step": 35090, "epoch": 974, "lr": 0.001} +{"train_loss": 0.028235146775841713, "train_loss_bc": 0.02823045663535595, "train_loss_llm": 0.00046899914741516113, "grad_norm": 0.07693952322006226, "global_step": 35091, "epoch": 974, "lr": 0.001} +{"train_loss": 0.028569897636771202, "train_loss_bc": 0.028569897636771202, "train_loss_llm": 0.0, "grad_norm": 0.09062420576810837, "global_step": 35092, "epoch": 974, "lr": 0.001} +{"train_loss": 0.034977179020643234, "train_loss_bc": 0.034977179020643234, "train_loss_llm": 0.0, "grad_norm": 0.04469360038638115, "global_step": 35093, "epoch": 974, "lr": 0.001} +{"train_loss": 0.028254883363842964, "train_loss_bc": 0.028254883363842964, "train_loss_llm": 0.0, "grad_norm": 0.06831269711256027, "global_step": 35094, "epoch": 974, "lr": 0.001} +{"train_loss": 0.0303086768835783, "train_loss_bc": 0.030289823189377785, "train_loss_llm": 0.0018853247165679932, "grad_norm": 0.10985933989286423, "global_step": 35095, "epoch": 974, "lr": 0.001} +{"train_loss": 0.03186555206775665, "train_loss_bc": 0.03186555206775665, "train_loss_llm": 0.0, "grad_norm": 0.13365620374679565, "global_step": 35096, "epoch": 974, "lr": 0.001} +{"train_loss": 0.02717517502605915, "train_loss_bc": 0.02717072330415249, "train_loss_llm": 0.0004451274871826172, "grad_norm": 0.04410100355744362, "global_step": 35097, "epoch": 974, "lr": 0.001} +{"train_loss": 0.028841737657785416, "train_loss_bc": 0.028841737657785416, "train_loss_llm": 0.0, "grad_norm": 0.07049589604139328, "global_step": 35098, "epoch": 974, "lr": 0.001} +{"train_loss": 0.02860548502455155, "train_loss_bc": 0.03333594650030136, "train_loss_llm": 0.0, "grad_norm": 0.10728293657302856, "global_step": 35099, "epoch": 974, "lr": 0.001, "val_loss": 0.11117459088563919} +{"train_loss": 0.022419508546590805, "train_loss_bc": 0.022419508546590805, "train_loss_llm": 0.0, "grad_norm": 0.1250966191291809, "global_step": 35100, "epoch": 975, "lr": 0.001} +{"train_loss": 0.024625128135085106, "train_loss_bc": 0.024625128135085106, "train_loss_llm": 0.0, "grad_norm": 0.03987134248018265, "global_step": 35101, "epoch": 975, "lr": 0.001} +{"train_loss": 0.028147168457508087, "train_loss_bc": 0.028146961703896523, "train_loss_llm": 2.059340476989746e-05, "grad_norm": 0.07442297786474228, "global_step": 35102, "epoch": 975, "lr": 0.001} +{"train_loss": 0.023267949000000954, "train_loss_bc": 0.023267949000000954, "train_loss_llm": 0.0, "grad_norm": 0.08527599275112152, "global_step": 35103, "epoch": 975, "lr": 0.001} +{"train_loss": 0.02849842794239521, "train_loss_bc": 0.028466131538152695, "train_loss_llm": 0.0032297074794769287, "grad_norm": 0.10872220247983932, "global_step": 35104, "epoch": 975, "lr": 0.001} +{"train_loss": 0.026221631094813347, "train_loss_bc": 0.026195397600531578, "train_loss_llm": 0.0026232898235321045, "grad_norm": 0.034401316195726395, "global_step": 35105, "epoch": 975, "lr": 0.001} +{"train_loss": 0.030894892290234566, "train_loss_bc": 0.030894234776496887, "train_loss_llm": 6.568431854248047e-05, "grad_norm": 0.059443049132823944, "global_step": 35106, "epoch": 975, "lr": 0.001} +{"train_loss": 0.027611656114459038, "train_loss_bc": 0.027611656114459038, "train_loss_llm": 0.0, "grad_norm": 0.08248662203550339, "global_step": 35107, "epoch": 975, "lr": 0.001} +{"train_loss": 0.02643408440053463, "train_loss_bc": 0.02643408440053463, "train_loss_llm": 0.0, "grad_norm": 0.10907676815986633, "global_step": 35108, "epoch": 975, "lr": 0.001} +{"train_loss": 0.03301311656832695, "train_loss_bc": 0.03301311656832695, "train_loss_llm": 0.0, "grad_norm": 0.05749168246984482, "global_step": 35109, "epoch": 975, "lr": 0.001} +{"train_loss": 0.032175857573747635, "train_loss_bc": 0.032175857573747635, "train_loss_llm": 0.0, "grad_norm": 0.08831343054771423, "global_step": 35110, "epoch": 975, "lr": 0.001} +{"train_loss": 0.03118366375565529, "train_loss_bc": 0.031179651618003845, "train_loss_llm": 0.0004011690616607666, "grad_norm": 0.11025833338499069, "global_step": 35111, "epoch": 975, "lr": 0.001} +{"train_loss": 0.03047148510813713, "train_loss_bc": 0.03047148510813713, "train_loss_llm": 0.0, "grad_norm": 0.13514290750026703, "global_step": 35112, "epoch": 975, "lr": 0.001} +{"train_loss": 0.03238258138298988, "train_loss_bc": 0.03237789124250412, "train_loss_llm": 0.00046887993812561035, "grad_norm": 0.046573854982852936, "global_step": 35113, "epoch": 975, "lr": 0.001} +{"train_loss": 0.03614499419927597, "train_loss_bc": 0.03614438697695732, "train_loss_llm": 6.079673767089844e-05, "grad_norm": 0.0865732729434967, "global_step": 35114, "epoch": 975, "lr": 0.001} +{"train_loss": 0.026770927011966705, "train_loss_bc": 0.026770927011966705, "train_loss_llm": 0.0, "grad_norm": 0.09994576126337051, "global_step": 35115, "epoch": 975, "lr": 0.001} +{"train_loss": 0.02219056524336338, "train_loss_bc": 0.02219056524336338, "train_loss_llm": 0.0, "grad_norm": 0.11006834357976913, "global_step": 35116, "epoch": 975, "lr": 0.001} +{"train_loss": 0.029972286894917488, "train_loss_bc": 0.029972286894917488, "train_loss_llm": 0.0, "grad_norm": 0.04936665669083595, "global_step": 35117, "epoch": 975, "lr": 0.001} +{"train_loss": 0.026011325418949127, "train_loss_bc": 0.026011325418949127, "train_loss_llm": 0.0, "grad_norm": 0.07322492450475693, "global_step": 35118, "epoch": 975, "lr": 0.001} +{"train_loss": 0.024150662124156952, "train_loss_bc": 0.024139493703842163, "train_loss_llm": 0.0011167824268341064, "grad_norm": 0.09284693002700806, "global_step": 35119, "epoch": 975, "lr": 0.001} +{"train_loss": 0.028903767466545105, "train_loss_bc": 0.028903767466545105, "train_loss_llm": 0.0, "grad_norm": 0.1144300177693367, "global_step": 35120, "epoch": 975, "lr": 0.001} +{"train_loss": 0.0323532372713089, "train_loss_bc": 0.0323532372713089, "train_loss_llm": 0.0, "grad_norm": 0.04371073469519615, "global_step": 35121, "epoch": 975, "lr": 0.001} +{"train_loss": 0.03082374669611454, "train_loss_bc": 0.03082374669611454, "train_loss_llm": 0.0, "grad_norm": 0.06554615497589111, "global_step": 35122, "epoch": 975, "lr": 0.001} +{"train_loss": 0.028886955231428146, "train_loss_bc": 0.028886955231428146, "train_loss_llm": 0.0, "grad_norm": 0.07151273638010025, "global_step": 35123, "epoch": 975, "lr": 0.001} +{"train_loss": 0.029093235731124878, "train_loss_bc": 0.029093235731124878, "train_loss_llm": 0.0, "grad_norm": 0.09084758907556534, "global_step": 35124, "epoch": 975, "lr": 0.001} +{"train_loss": 0.034818846732378006, "train_loss_bc": 0.034818846732378006, "train_loss_llm": 0.0, "grad_norm": 0.05264170467853546, "global_step": 35125, "epoch": 975, "lr": 0.001} +{"train_loss": 0.02281847968697548, "train_loss_bc": 0.02281847968697548, "train_loss_llm": 0.0, "grad_norm": 0.07921040058135986, "global_step": 35126, "epoch": 975, "lr": 0.001} +{"train_loss": 0.03500540181994438, "train_loss_bc": 0.03500540181994438, "train_loss_llm": 0.0, "grad_norm": 0.11215797811746597, "global_step": 35127, "epoch": 975, "lr": 0.001} +{"train_loss": 0.02480248734354973, "train_loss_bc": 0.02480248734354973, "train_loss_llm": 0.0, "grad_norm": 0.1386629194021225, "global_step": 35128, "epoch": 975, "lr": 0.001} +{"train_loss": 0.032188378274440765, "train_loss_bc": 0.032188378274440765, "train_loss_llm": 0.0, "grad_norm": 0.04320048913359642, "global_step": 35129, "epoch": 975, "lr": 0.001} +{"train_loss": 0.030545733869075775, "train_loss_bc": 0.030545733869075775, "train_loss_llm": 0.0, "grad_norm": 0.06719287484884262, "global_step": 35130, "epoch": 975, "lr": 0.001} +{"train_loss": 0.032245419919490814, "train_loss_bc": 0.032245419919490814, "train_loss_llm": 0.0, "grad_norm": 0.08860646933317184, "global_step": 35131, "epoch": 975, "lr": 0.001} +{"train_loss": 0.026364387944340706, "train_loss_bc": 0.026364387944340706, "train_loss_llm": 0.0, "grad_norm": 0.1260979175567627, "global_step": 35132, "epoch": 975, "lr": 0.001} +{"train_loss": 0.0288478285074234, "train_loss_bc": 0.02884766273200512, "train_loss_llm": 1.665949821472168e-05, "grad_norm": 0.04346144571900368, "global_step": 35133, "epoch": 975, "lr": 0.001} +{"train_loss": 0.022238614037632942, "train_loss_bc": 0.022235922515392303, "train_loss_llm": 0.00026917457580566406, "grad_norm": 0.0627637431025505, "global_step": 35134, "epoch": 975, "lr": 0.001} +{"train_loss": 0.028628762087060347, "train_loss_bc": 0.02811100333929062, "train_loss_llm": 0.0, "grad_norm": 0.09238862991333008, "global_step": 35135, "epoch": 975, "lr": 0.001, "train/cumulative_reward": 75.49692294827041, "train/mean_score": 17.730820336715205, "train/success_rate": 0.4, "test/cumulative_reward": 24.68715546735914, "test/mean_score": 7.404697578647511, "test/success_rate": 0.18, "val_loss": 0.11086119711399078, "train_action_mse_error": 0.031271032989025116} +{"train_loss": 0.023798402398824692, "train_loss_bc": 0.02377515472471714, "train_loss_llm": 0.0023247599601745605, "grad_norm": 0.10880931466817856, "global_step": 35136, "epoch": 976, "lr": 0.001} +{"train_loss": 0.028315339237451553, "train_loss_bc": 0.028315339237451553, "train_loss_llm": 0.0, "grad_norm": 0.03416630998253822, "global_step": 35137, "epoch": 976, "lr": 0.001} +{"train_loss": 0.02604314684867859, "train_loss_bc": 0.02604314684867859, "train_loss_llm": 0.0, "grad_norm": 0.04680198058485985, "global_step": 35138, "epoch": 976, "lr": 0.001} +{"train_loss": 0.03031950816512108, "train_loss_bc": 0.03031591698527336, "train_loss_llm": 0.0003592073917388916, "grad_norm": 0.06880752742290497, "global_step": 35139, "epoch": 976, "lr": 0.001} +{"train_loss": 0.03427371382713318, "train_loss_bc": 0.03427371382713318, "train_loss_llm": 0.0, "grad_norm": 0.08913412690162659, "global_step": 35140, "epoch": 976, "lr": 0.001} +{"train_loss": 0.031015103682875633, "train_loss_bc": 0.031015103682875633, "train_loss_llm": 0.0, "grad_norm": 0.03809763863682747, "global_step": 35141, "epoch": 976, "lr": 0.001} +{"train_loss": 0.02898133359849453, "train_loss_bc": 0.02898133359849453, "train_loss_llm": 0.0, "grad_norm": 0.07068728655576706, "global_step": 35142, "epoch": 976, "lr": 0.001} +{"train_loss": 0.029362693428993225, "train_loss_bc": 0.029362693428993225, "train_loss_llm": 0.0, "grad_norm": 0.09122151136398315, "global_step": 35143, "epoch": 976, "lr": 0.001} +{"train_loss": 0.03221884369850159, "train_loss_bc": 0.032211799174547195, "train_loss_llm": 0.0007045865058898926, "grad_norm": 0.10749031603336334, "global_step": 35144, "epoch": 976, "lr": 0.001} +{"train_loss": 0.032863467931747437, "train_loss_bc": 0.032863467931747437, "train_loss_llm": 0.0, "grad_norm": 0.047785934060811996, "global_step": 35145, "epoch": 976, "lr": 0.001} +{"train_loss": 0.02720743976533413, "train_loss_bc": 0.02720743976533413, "train_loss_llm": 0.0, "grad_norm": 0.06422143429517746, "global_step": 35146, "epoch": 976, "lr": 0.001} +{"train_loss": 0.029284613206982613, "train_loss_bc": 0.029284613206982613, "train_loss_llm": 0.0, "grad_norm": 0.08427543193101883, "global_step": 35147, "epoch": 976, "lr": 0.001} +{"train_loss": 0.027840506285429, "train_loss_bc": 0.02783478982746601, "train_loss_llm": 0.0005716681480407715, "grad_norm": 0.10199320316314697, "global_step": 35148, "epoch": 976, "lr": 0.001} +{"train_loss": 0.03384571149945259, "train_loss_bc": 0.03384571149945259, "train_loss_llm": 0.0, "grad_norm": 0.04010545089840889, "global_step": 35149, "epoch": 976, "lr": 0.001} +{"train_loss": 0.02968539297580719, "train_loss_bc": 0.0296771302819252, "train_loss_llm": 0.0008263587951660156, "grad_norm": 0.06285833567380905, "global_step": 35150, "epoch": 976, "lr": 0.001} +{"train_loss": 0.027868319302797318, "train_loss_bc": 0.027868319302797318, "train_loss_llm": 0.0, "grad_norm": 0.08945979177951813, "global_step": 35151, "epoch": 976, "lr": 0.001} +{"train_loss": 0.025166243314743042, "train_loss_bc": 0.025166243314743042, "train_loss_llm": 0.0, "grad_norm": 0.10710521042346954, "global_step": 35152, "epoch": 976, "lr": 0.001} +{"train_loss": 0.029162142425775528, "train_loss_bc": 0.029162142425775528, "train_loss_llm": 0.0, "grad_norm": 0.04004565253853798, "global_step": 35153, "epoch": 976, "lr": 0.001} +{"train_loss": 0.030995609238743782, "train_loss_bc": 0.030995609238743782, "train_loss_llm": 0.0, "grad_norm": 0.06850630789995193, "global_step": 35154, "epoch": 976, "lr": 0.001} +{"train_loss": 0.029987214133143425, "train_loss_bc": 0.029987214133143425, "train_loss_llm": 0.0, "grad_norm": 0.08517234772443771, "global_step": 35155, "epoch": 976, "lr": 0.001} +{"train_loss": 0.026040932163596153, "train_loss_bc": 0.026040932163596153, "train_loss_llm": 0.0, "grad_norm": 0.09377958625555038, "global_step": 35156, "epoch": 976, "lr": 0.001} +{"train_loss": 0.028469817712903023, "train_loss_bc": 0.028469817712903023, "train_loss_llm": 0.0, "grad_norm": 0.035117555409669876, "global_step": 35157, "epoch": 976, "lr": 0.001} +{"train_loss": 0.03083874098956585, "train_loss_bc": 0.030838320031762123, "train_loss_llm": 4.217028617858887e-05, "grad_norm": 0.05850648880004883, "global_step": 35158, "epoch": 976, "lr": 0.001} +{"train_loss": 0.02696281298995018, "train_loss_bc": 0.02696281298995018, "train_loss_llm": 0.0, "grad_norm": 0.08272693306207657, "global_step": 35159, "epoch": 976, "lr": 0.001} +{"train_loss": 0.02592753991484642, "train_loss_bc": 0.025927461683750153, "train_loss_llm": 7.808208465576172e-06, "grad_norm": 0.09614186733961105, "global_step": 35160, "epoch": 976, "lr": 0.001} +{"train_loss": 0.0267326757311821, "train_loss_bc": 0.026724081486463547, "train_loss_llm": 0.0008594691753387451, "grad_norm": 0.036284711211919785, "global_step": 35161, "epoch": 976, "lr": 0.001} +{"train_loss": 0.029571468010544777, "train_loss_bc": 0.029571468010544777, "train_loss_llm": 0.0, "grad_norm": 0.0626174807548523, "global_step": 35162, "epoch": 976, "lr": 0.001} +{"train_loss": 0.026798710227012634, "train_loss_bc": 0.026798710227012634, "train_loss_llm": 0.0, "grad_norm": 0.0765451118350029, "global_step": 35163, "epoch": 976, "lr": 0.001} +{"train_loss": 0.025276102125644684, "train_loss_bc": 0.025276102125644684, "train_loss_llm": 0.0, "grad_norm": 0.09668506681919098, "global_step": 35164, "epoch": 976, "lr": 0.001} +{"train_loss": 0.027631141245365143, "train_loss_bc": 0.027631141245365143, "train_loss_llm": 0.0, "grad_norm": 0.044125430285930634, "global_step": 35165, "epoch": 976, "lr": 0.001} +{"train_loss": 0.025866687297821045, "train_loss_bc": 0.025866687297821045, "train_loss_llm": 0.0, "grad_norm": 0.0690745860338211, "global_step": 35166, "epoch": 976, "lr": 0.001} +{"train_loss": 0.0241963192820549, "train_loss_bc": 0.02419108897447586, "train_loss_llm": 0.0005229413509368896, "grad_norm": 0.08523336052894592, "global_step": 35167, "epoch": 976, "lr": 0.001} +{"train_loss": 0.02807758003473282, "train_loss_bc": 0.02807758003473282, "train_loss_llm": 0.0, "grad_norm": 0.1141628846526146, "global_step": 35168, "epoch": 976, "lr": 0.001} +{"train_loss": 0.031162139028310776, "train_loss_bc": 0.031162139028310776, "train_loss_llm": 0.0, "grad_norm": 0.03359255939722061, "global_step": 35169, "epoch": 976, "lr": 0.001} +{"train_loss": 0.024153707548975945, "train_loss_bc": 0.024153707548975945, "train_loss_llm": 0.0, "grad_norm": 0.06417546421289444, "global_step": 35170, "epoch": 976, "lr": 0.001} +{"train_loss": 0.028399899136275053, "train_loss_bc": 0.02645524963736534, "train_loss_llm": 0.0, "grad_norm": 0.07035017013549805, "global_step": 35171, "epoch": 976, "lr": 0.001, "val_loss": 0.11023776233196259} +{"train_loss": 0.023743823170661926, "train_loss_bc": 0.02373458258807659, "train_loss_llm": 0.0009240508079528809, "grad_norm": 0.09488475322723389, "global_step": 35172, "epoch": 977, "lr": 0.001} +{"train_loss": 0.029607076197862625, "train_loss_bc": 0.029607076197862625, "train_loss_llm": 0.0, "grad_norm": 0.035604920238256454, "global_step": 35173, "epoch": 977, "lr": 0.001} +{"train_loss": 0.029844798147678375, "train_loss_bc": 0.029844798147678375, "train_loss_llm": 0.0, "grad_norm": 0.06743405014276505, "global_step": 35174, "epoch": 977, "lr": 0.001} +{"train_loss": 0.02722194977104664, "train_loss_bc": 0.02722194977104664, "train_loss_llm": 0.0, "grad_norm": 0.10228622704744339, "global_step": 35175, "epoch": 977, "lr": 0.001} +{"train_loss": 0.033835966140031815, "train_loss_bc": 0.033835966140031815, "train_loss_llm": 0.0, "grad_norm": 0.12297292798757553, "global_step": 35176, "epoch": 977, "lr": 0.001} +{"train_loss": 0.029807226732373238, "train_loss_bc": 0.029807226732373238, "train_loss_llm": 0.0, "grad_norm": 0.045071519911289215, "global_step": 35177, "epoch": 977, "lr": 0.001} +{"train_loss": 0.027801841497421265, "train_loss_bc": 0.027801841497421265, "train_loss_llm": 0.0, "grad_norm": 0.06281085312366486, "global_step": 35178, "epoch": 977, "lr": 0.001} +{"train_loss": 0.027688568457961082, "train_loss_bc": 0.027688568457961082, "train_loss_llm": 0.0, "grad_norm": 0.07851763069629669, "global_step": 35179, "epoch": 977, "lr": 0.001} +{"train_loss": 0.025562381371855736, "train_loss_bc": 0.025562381371855736, "train_loss_llm": 0.0, "grad_norm": 0.08429024368524551, "global_step": 35180, "epoch": 977, "lr": 0.001} +{"train_loss": 0.02735012397170067, "train_loss_bc": 0.027346672490239143, "train_loss_llm": 0.0003452301025390625, "grad_norm": 0.030891485512256622, "global_step": 35181, "epoch": 977, "lr": 0.001} +{"train_loss": 0.027126969769597054, "train_loss_bc": 0.027126969769597054, "train_loss_llm": 0.0, "grad_norm": 0.04286758974194527, "global_step": 35182, "epoch": 977, "lr": 0.001} +{"train_loss": 0.027943242341279984, "train_loss_bc": 0.027943242341279984, "train_loss_llm": 0.0, "grad_norm": 0.05481775477528572, "global_step": 35183, "epoch": 977, "lr": 0.001} +{"train_loss": 0.03532881289720535, "train_loss_bc": 0.03532881289720535, "train_loss_llm": 0.0, "grad_norm": 0.07039861381053925, "global_step": 35184, "epoch": 977, "lr": 0.001} +{"train_loss": 0.03096463903784752, "train_loss_bc": 0.030937280505895615, "train_loss_llm": 0.0027358531951904297, "grad_norm": 0.04036501422524452, "global_step": 35185, "epoch": 977, "lr": 0.001} +{"train_loss": 0.02616233192384243, "train_loss_bc": 0.02616233192384243, "train_loss_llm": 0.0, "grad_norm": 0.052237384021282196, "global_step": 35186, "epoch": 977, "lr": 0.001} +{"train_loss": 0.028033819049596786, "train_loss_bc": 0.028033819049596786, "train_loss_llm": 0.0, "grad_norm": 0.059342160820961, "global_step": 35187, "epoch": 977, "lr": 0.001} +{"train_loss": 0.024595975875854492, "train_loss_bc": 0.024595975875854492, "train_loss_llm": 0.0, "grad_norm": 0.07176664471626282, "global_step": 35188, "epoch": 977, "lr": 0.001} +{"train_loss": 0.027885520830750465, "train_loss_bc": 0.02787918411195278, "train_loss_llm": 0.0006336569786071777, "grad_norm": 0.03244918957352638, "global_step": 35189, "epoch": 977, "lr": 0.001} +{"train_loss": 0.02807757258415222, "train_loss_bc": 0.02807757258415222, "train_loss_llm": 0.0, "grad_norm": 0.050970759242773056, "global_step": 35190, "epoch": 977, "lr": 0.001} +{"train_loss": 0.02773837000131607, "train_loss_bc": 0.02773837000131607, "train_loss_llm": 0.0, "grad_norm": 0.07070725411176682, "global_step": 35191, "epoch": 977, "lr": 0.001} +{"train_loss": 0.024476628750562668, "train_loss_bc": 0.024476628750562668, "train_loss_llm": 0.0, "grad_norm": 0.09278836101293564, "global_step": 35192, "epoch": 977, "lr": 0.001} +{"train_loss": 0.02566603757441044, "train_loss_bc": 0.025665851309895515, "train_loss_llm": 1.8596649169921875e-05, "grad_norm": 0.0320880189538002, "global_step": 35193, "epoch": 977, "lr": 0.001} +{"train_loss": 0.02639799565076828, "train_loss_bc": 0.02639799565076828, "train_loss_llm": 0.0, "grad_norm": 0.05713431537151337, "global_step": 35194, "epoch": 977, "lr": 0.001} +{"train_loss": 0.02467498555779457, "train_loss_bc": 0.024674508720636368, "train_loss_llm": 4.76837158203125e-05, "grad_norm": 0.06690331548452377, "global_step": 35195, "epoch": 977, "lr": 0.001} +{"train_loss": 0.029640890657901764, "train_loss_bc": 0.029640890657901764, "train_loss_llm": 0.0, "grad_norm": 0.08552142977714539, "global_step": 35196, "epoch": 977, "lr": 0.001} +{"train_loss": 0.03010270930826664, "train_loss_bc": 0.03010270930826664, "train_loss_llm": 0.0, "grad_norm": 0.039107587188482285, "global_step": 35197, "epoch": 977, "lr": 0.001} +{"train_loss": 0.025111975148320198, "train_loss_bc": 0.025111975148320198, "train_loss_llm": 0.0, "grad_norm": 0.05230370908975601, "global_step": 35198, "epoch": 977, "lr": 0.001} +{"train_loss": 0.034545138478279114, "train_loss_bc": 0.03453638032078743, "train_loss_llm": 0.0008758306503295898, "grad_norm": 0.0717930719256401, "global_step": 35199, "epoch": 977, "lr": 0.001} +{"train_loss": 0.029201924800872803, "train_loss_bc": 0.02919432520866394, "train_loss_llm": 0.0007598996162414551, "grad_norm": 0.08938369154930115, "global_step": 35200, "epoch": 977, "lr": 0.001} +{"train_loss": 0.02640857733786106, "train_loss_bc": 0.02640489488840103, "train_loss_llm": 0.0003682374954223633, "grad_norm": 0.045516759157180786, "global_step": 35201, "epoch": 977, "lr": 0.001} +{"train_loss": 0.025188449770212173, "train_loss_bc": 0.025188449770212173, "train_loss_llm": 0.0, "grad_norm": 0.0596974641084671, "global_step": 35202, "epoch": 977, "lr": 0.001} +{"train_loss": 0.0349082425236702, "train_loss_bc": 0.0349082425236702, "train_loss_llm": 0.0, "grad_norm": 0.08018811047077179, "global_step": 35203, "epoch": 977, "lr": 0.001} +{"train_loss": 0.03056996315717697, "train_loss_bc": 0.03056996315717697, "train_loss_llm": 0.0, "grad_norm": 0.10298503190279007, "global_step": 35204, "epoch": 977, "lr": 0.001} +{"train_loss": 0.027173906564712524, "train_loss_bc": 0.027173906564712524, "train_loss_llm": 0.0, "grad_norm": 0.04197952151298523, "global_step": 35205, "epoch": 977, "lr": 0.001} +{"train_loss": 0.031050845980644226, "train_loss_bc": 0.031050845980644226, "train_loss_llm": 0.0, "grad_norm": 0.0615917444229126, "global_step": 35206, "epoch": 977, "lr": 0.001} +{"train_loss": 0.028374673591719732, "train_loss_bc": 0.030048968270421028, "train_loss_llm": 0.0, "grad_norm": 0.0855555534362793, "global_step": 35207, "epoch": 977, "lr": 0.001, "val_loss": 0.11060830950737} +{"train_loss": 0.030408771708607674, "train_loss_bc": 0.030408771708607674, "train_loss_llm": 0.0, "grad_norm": 0.0988517478108406, "global_step": 35208, "epoch": 978, "lr": 0.001} +{"train_loss": 0.026912705972790718, "train_loss_bc": 0.026912705972790718, "train_loss_llm": 0.0, "grad_norm": 0.030340369790792465, "global_step": 35209, "epoch": 978, "lr": 0.001} +{"train_loss": 0.027649996802210808, "train_loss_bc": 0.027649996802210808, "train_loss_llm": 0.0, "grad_norm": 0.051930926740169525, "global_step": 35210, "epoch": 978, "lr": 0.001} +{"train_loss": 0.03050098940730095, "train_loss_bc": 0.030496744439005852, "train_loss_llm": 0.0004245340824127197, "grad_norm": 0.0712384283542633, "global_step": 35211, "epoch": 978, "lr": 0.001} +{"train_loss": 0.02846257947385311, "train_loss_bc": 0.02846134640276432, "train_loss_llm": 0.0001233220100402832, "grad_norm": 0.09405054897069931, "global_step": 35212, "epoch": 978, "lr": 0.001} +{"train_loss": 0.022571679204702377, "train_loss_bc": 0.022567756474018097, "train_loss_llm": 0.0003922581672668457, "grad_norm": 0.03143269941210747, "global_step": 35213, "epoch": 978, "lr": 0.001} +{"train_loss": 0.027470223605632782, "train_loss_bc": 0.027462659403681755, "train_loss_llm": 0.0007565021514892578, "grad_norm": 0.05360342562198639, "global_step": 35214, "epoch": 978, "lr": 0.001} +{"train_loss": 0.031493041664361954, "train_loss_bc": 0.03148913010954857, "train_loss_llm": 0.00039130449295043945, "grad_norm": 0.0794723629951477, "global_step": 35215, "epoch": 978, "lr": 0.001} +{"train_loss": 0.026356831192970276, "train_loss_bc": 0.026356831192970276, "train_loss_llm": 0.0, "grad_norm": 0.10011053830385208, "global_step": 35216, "epoch": 978, "lr": 0.001} +{"train_loss": 0.02457614056766033, "train_loss_bc": 0.02457614056766033, "train_loss_llm": 0.0, "grad_norm": 0.03710925206542015, "global_step": 35217, "epoch": 978, "lr": 0.001} +{"train_loss": 0.025990383699536324, "train_loss_bc": 0.025986425578594208, "train_loss_llm": 0.00039577484130859375, "grad_norm": 0.05991024151444435, "global_step": 35218, "epoch": 978, "lr": 0.001} +{"train_loss": 0.031103787943720818, "train_loss_bc": 0.031103787943720818, "train_loss_llm": 0.0, "grad_norm": 0.0819561630487442, "global_step": 35219, "epoch": 978, "lr": 0.001} +{"train_loss": 0.02909836173057556, "train_loss_bc": 0.02909836173057556, "train_loss_llm": 0.0, "grad_norm": 0.10600993037223816, "global_step": 35220, "epoch": 978, "lr": 0.001} +{"train_loss": 0.02707792818546295, "train_loss_bc": 0.027054017409682274, "train_loss_llm": 0.0023910701274871826, "grad_norm": 0.028182292357087135, "global_step": 35221, "epoch": 978, "lr": 0.001} +{"train_loss": 0.0282118059694767, "train_loss_bc": 0.0282118059694767, "train_loss_llm": 0.0, "grad_norm": 0.03899652138352394, "global_step": 35222, "epoch": 978, "lr": 0.001} +{"train_loss": 0.031578078866004944, "train_loss_bc": 0.031578078866004944, "train_loss_llm": 0.0, "grad_norm": 0.05724881589412689, "global_step": 35223, "epoch": 978, "lr": 0.001} +{"train_loss": 0.032357316464185715, "train_loss_bc": 0.03235124796628952, "train_loss_llm": 0.000606924295425415, "grad_norm": 0.07283724099397659, "global_step": 35224, "epoch": 978, "lr": 0.001} +{"train_loss": 0.03047783486545086, "train_loss_bc": 0.03047783486545086, "train_loss_llm": 0.0, "grad_norm": 0.04312494024634361, "global_step": 35225, "epoch": 978, "lr": 0.001} +{"train_loss": 0.025558702647686005, "train_loss_bc": 0.025558702647686005, "train_loss_llm": 0.0, "grad_norm": 0.06297271698713303, "global_step": 35226, "epoch": 978, "lr": 0.001} +{"train_loss": 0.028441233560442924, "train_loss_bc": 0.028431441634893417, "train_loss_llm": 0.0009792149066925049, "grad_norm": 0.07795587927103043, "global_step": 35227, "epoch": 978, "lr": 0.001} +{"train_loss": 0.026894627138972282, "train_loss_bc": 0.026894627138972282, "train_loss_llm": 0.0, "grad_norm": 0.09843621402978897, "global_step": 35228, "epoch": 978, "lr": 0.001} +{"train_loss": 0.02614443562924862, "train_loss_bc": 0.02614443562924862, "train_loss_llm": 0.0, "grad_norm": 0.03779271990060806, "global_step": 35229, "epoch": 978, "lr": 0.001} +{"train_loss": 0.028826221823692322, "train_loss_bc": 0.028826221823692322, "train_loss_llm": 0.0, "grad_norm": 0.06581480801105499, "global_step": 35230, "epoch": 978, "lr": 0.001} +{"train_loss": 0.0281197689473629, "train_loss_bc": 0.028110843151807785, "train_loss_llm": 0.0008926093578338623, "grad_norm": 0.08441225439310074, "global_step": 35231, "epoch": 978, "lr": 0.001} +{"train_loss": 0.027699027210474014, "train_loss_bc": 0.027699027210474014, "train_loss_llm": 0.0, "grad_norm": 0.11094342172145844, "global_step": 35232, "epoch": 978, "lr": 0.001} +{"train_loss": 0.03268660232424736, "train_loss_bc": 0.03268660232424736, "train_loss_llm": 0.0, "grad_norm": 0.041610054671764374, "global_step": 35233, "epoch": 978, "lr": 0.001} +{"train_loss": 0.024486642330884933, "train_loss_bc": 0.024463042616844177, "train_loss_llm": 0.002360016107559204, "grad_norm": 0.059343673288822174, "global_step": 35234, "epoch": 978, "lr": 0.001} +{"train_loss": 0.024066565558314323, "train_loss_bc": 0.024066565558314323, "train_loss_llm": 0.0, "grad_norm": 0.06875622272491455, "global_step": 35235, "epoch": 978, "lr": 0.001} +{"train_loss": 0.027211513370275497, "train_loss_bc": 0.027211513370275497, "train_loss_llm": 0.0, "grad_norm": 0.09161640703678131, "global_step": 35236, "epoch": 978, "lr": 0.001} +{"train_loss": 0.0261223167181015, "train_loss_bc": 0.026120416820049286, "train_loss_llm": 0.00018990039825439453, "grad_norm": 0.0442793034017086, "global_step": 35237, "epoch": 978, "lr": 0.001} +{"train_loss": 0.03050369769334793, "train_loss_bc": 0.03050369769334793, "train_loss_llm": 0.0, "grad_norm": 0.06644540280103683, "global_step": 35238, "epoch": 978, "lr": 0.001} +{"train_loss": 0.027240857481956482, "train_loss_bc": 0.027240857481956482, "train_loss_llm": 0.0, "grad_norm": 0.09319136291742325, "global_step": 35239, "epoch": 978, "lr": 0.001} +{"train_loss": 0.03497511148452759, "train_loss_bc": 0.03497511148452759, "train_loss_llm": 0.0, "grad_norm": 0.1067189946770668, "global_step": 35240, "epoch": 978, "lr": 0.001} +{"train_loss": 0.031364791095256805, "train_loss_bc": 0.031364791095256805, "train_loss_llm": 0.0, "grad_norm": 0.04204603284597397, "global_step": 35241, "epoch": 978, "lr": 0.001} +{"train_loss": 0.03166937083005905, "train_loss_bc": 0.03166937083005905, "train_loss_llm": 0.0, "grad_norm": 0.05810660496354103, "global_step": 35242, "epoch": 978, "lr": 0.001} +{"train_loss": 0.028323743544105027, "train_loss_bc": 0.025333218276500702, "train_loss_llm": 0.0011606812477111816, "grad_norm": 0.07038430869579315, "global_step": 35243, "epoch": 978, "lr": 0.001, "val_loss": 0.11099394410848618} +{"train_loss": 0.02615363337099552, "train_loss_bc": 0.02615363337099552, "train_loss_llm": 0.0, "grad_norm": 0.0824626162648201, "global_step": 35244, "epoch": 979, "lr": 0.001} +{"train_loss": 0.030759790912270546, "train_loss_bc": 0.030759790912270546, "train_loss_llm": 0.0, "grad_norm": 0.03616752102971077, "global_step": 35245, "epoch": 979, "lr": 0.001} +{"train_loss": 0.029875753447413445, "train_loss_bc": 0.029875753447413445, "train_loss_llm": 0.0, "grad_norm": 0.04917522147297859, "global_step": 35246, "epoch": 979, "lr": 0.001} +{"train_loss": 0.030711179599165916, "train_loss_bc": 0.03070121631026268, "train_loss_llm": 0.000996321439743042, "grad_norm": 0.07946145534515381, "global_step": 35247, "epoch": 979, "lr": 0.001} +{"train_loss": 0.030148889869451523, "train_loss_bc": 0.03013741783797741, "train_loss_llm": 0.0011472105979919434, "grad_norm": 0.09531896561384201, "global_step": 35248, "epoch": 979, "lr": 0.001} +{"train_loss": 0.026684509590268135, "train_loss_bc": 0.026684509590268135, "train_loss_llm": 0.0, "grad_norm": 0.04096929728984833, "global_step": 35249, "epoch": 979, "lr": 0.001} +{"train_loss": 0.02326042391359806, "train_loss_bc": 0.023254085332155228, "train_loss_llm": 0.0006337761878967285, "grad_norm": 0.06549946963787079, "global_step": 35250, "epoch": 979, "lr": 0.001} +{"train_loss": 0.0295866746455431, "train_loss_bc": 0.0295866746455431, "train_loss_llm": 0.0, "grad_norm": 0.08567503839731216, "global_step": 35251, "epoch": 979, "lr": 0.001} +{"train_loss": 0.027867324650287628, "train_loss_bc": 0.027867324650287628, "train_loss_llm": 0.0, "grad_norm": 0.11517839133739471, "global_step": 35252, "epoch": 979, "lr": 0.001} +{"train_loss": 0.02401556633412838, "train_loss_bc": 0.02401556633412838, "train_loss_llm": 0.0, "grad_norm": 0.03483881801366806, "global_step": 35253, "epoch": 979, "lr": 0.001} +{"train_loss": 0.03212662413716316, "train_loss_bc": 0.03212662413716316, "train_loss_llm": 0.0, "grad_norm": 0.06013769283890724, "global_step": 35254, "epoch": 979, "lr": 0.001} +{"train_loss": 0.02795465663075447, "train_loss_bc": 0.027938928455114365, "train_loss_llm": 0.0015727579593658447, "grad_norm": 0.07485439628362656, "global_step": 35255, "epoch": 979, "lr": 0.001} +{"train_loss": 0.029633654281497, "train_loss_bc": 0.029633654281497, "train_loss_llm": 0.0, "grad_norm": 0.08915991336107254, "global_step": 35256, "epoch": 979, "lr": 0.001} +{"train_loss": 0.024968475103378296, "train_loss_bc": 0.02494952827692032, "train_loss_llm": 0.0018947422504425049, "grad_norm": 0.035421229898929596, "global_step": 35257, "epoch": 979, "lr": 0.001} +{"train_loss": 0.026065832003951073, "train_loss_bc": 0.026065832003951073, "train_loss_llm": 0.0, "grad_norm": 0.06599113345146179, "global_step": 35258, "epoch": 979, "lr": 0.001} +{"train_loss": 0.026892201974987984, "train_loss_bc": 0.026892201974987984, "train_loss_llm": 0.0, "grad_norm": 0.08371138572692871, "global_step": 35259, "epoch": 979, "lr": 0.001} +{"train_loss": 0.0318315215408802, "train_loss_bc": 0.031818658113479614, "train_loss_llm": 0.0012862086296081543, "grad_norm": 0.10366087406873703, "global_step": 35260, "epoch": 979, "lr": 0.001} +{"train_loss": 0.026395242661237717, "train_loss_bc": 0.026395242661237717, "train_loss_llm": 0.0, "grad_norm": 0.031322650611400604, "global_step": 35261, "epoch": 979, "lr": 0.001} +{"train_loss": 0.02580481953918934, "train_loss_bc": 0.02580481953918934, "train_loss_llm": 0.0, "grad_norm": 0.05476371571421623, "global_step": 35262, "epoch": 979, "lr": 0.001} +{"train_loss": 0.030387816950678825, "train_loss_bc": 0.030387816950678825, "train_loss_llm": 0.0, "grad_norm": 0.08179771900177002, "global_step": 35263, "epoch": 979, "lr": 0.001} +{"train_loss": 0.024483269080519676, "train_loss_bc": 0.024483269080519676, "train_loss_llm": 0.0, "grad_norm": 0.10270300507545471, "global_step": 35264, "epoch": 979, "lr": 0.001} +{"train_loss": 0.025828324258327484, "train_loss_bc": 0.025828324258327484, "train_loss_llm": 0.0, "grad_norm": 0.03553680703043938, "global_step": 35265, "epoch": 979, "lr": 0.001} +{"train_loss": 0.027710864320397377, "train_loss_bc": 0.027710864320397377, "train_loss_llm": 0.0, "grad_norm": 0.06370443850755692, "global_step": 35266, "epoch": 979, "lr": 0.001} +{"train_loss": 0.025560198351740837, "train_loss_bc": 0.025560198351740837, "train_loss_llm": 0.0, "grad_norm": 0.08185023069381714, "global_step": 35267, "epoch": 979, "lr": 0.001} +{"train_loss": 0.026836436241865158, "train_loss_bc": 0.02681896835565567, "train_loss_llm": 0.001746833324432373, "grad_norm": 0.107204370200634, "global_step": 35268, "epoch": 979, "lr": 0.001} +{"train_loss": 0.0318506695330143, "train_loss_bc": 0.03184274211525917, "train_loss_llm": 0.0007925629615783691, "grad_norm": 0.043926775455474854, "global_step": 35269, "epoch": 979, "lr": 0.001} +{"train_loss": 0.030483851209282875, "train_loss_bc": 0.03047941066324711, "train_loss_llm": 0.00044414401054382324, "grad_norm": 0.06564469635486603, "global_step": 35270, "epoch": 979, "lr": 0.001} +{"train_loss": 0.0279970895498991, "train_loss_bc": 0.027993028983473778, "train_loss_llm": 0.0004061460494995117, "grad_norm": 0.09478604793548584, "global_step": 35271, "epoch": 979, "lr": 0.001} +{"train_loss": 0.033343397080898285, "train_loss_bc": 0.033343397080898285, "train_loss_llm": 0.0, "grad_norm": 0.11767204105854034, "global_step": 35272, "epoch": 979, "lr": 0.001} +{"train_loss": 0.031067118048667908, "train_loss_bc": 0.031067118048667908, "train_loss_llm": 0.0, "grad_norm": 0.041218552738428116, "global_step": 35273, "epoch": 979, "lr": 0.001} +{"train_loss": 0.02880653366446495, "train_loss_bc": 0.02880653366446495, "train_loss_llm": 0.0, "grad_norm": 0.06051833555102348, "global_step": 35274, "epoch": 979, "lr": 0.001} +{"train_loss": 0.03387191891670227, "train_loss_bc": 0.03387191891670227, "train_loss_llm": 0.0, "grad_norm": 0.08213797956705093, "global_step": 35275, "epoch": 979, "lr": 0.001} +{"train_loss": 0.030517306178808212, "train_loss_bc": 0.030517306178808212, "train_loss_llm": 0.0, "grad_norm": 0.09489578008651733, "global_step": 35276, "epoch": 979, "lr": 0.001} +{"train_loss": 0.022863557562232018, "train_loss_bc": 0.022863557562232018, "train_loss_llm": 0.0, "grad_norm": 0.030939839780330658, "global_step": 35277, "epoch": 979, "lr": 0.001} +{"train_loss": 0.03318654000759125, "train_loss_bc": 0.03318654000759125, "train_loss_llm": 0.0, "grad_norm": 0.05369070917367935, "global_step": 35278, "epoch": 979, "lr": 0.001} +{"train_loss": 0.02839451629875435, "train_loss_bc": 0.02666904591023922, "train_loss_llm": 0.0001876354217529297, "grad_norm": 0.07511400431394577, "global_step": 35279, "epoch": 979, "lr": 0.001, "val_loss": 0.10992974787950516} +{"train_loss": 0.03194074332714081, "train_loss_bc": 0.03192539140582085, "train_loss_llm": 0.0015350580215454102, "grad_norm": 0.09238658845424652, "global_step": 35280, "epoch": 980, "lr": 0.001} +{"train_loss": 0.02898464910686016, "train_loss_bc": 0.02898464910686016, "train_loss_llm": 0.0, "grad_norm": 0.04597717151045799, "global_step": 35281, "epoch": 980, "lr": 0.001} +{"train_loss": 0.03205686807632446, "train_loss_bc": 0.03204451873898506, "train_loss_llm": 0.0012348294258117676, "grad_norm": 0.05866935849189758, "global_step": 35282, "epoch": 980, "lr": 0.001} +{"train_loss": 0.02771003358066082, "train_loss_bc": 0.02769543044269085, "train_loss_llm": 0.001460343599319458, "grad_norm": 0.0824267566204071, "global_step": 35283, "epoch": 980, "lr": 0.001} +{"train_loss": 0.025134937837719917, "train_loss_bc": 0.025134937837719917, "train_loss_llm": 0.0, "grad_norm": 0.09980577230453491, "global_step": 35284, "epoch": 980, "lr": 0.001} +{"train_loss": 0.029162896797060966, "train_loss_bc": 0.029162896797060966, "train_loss_llm": 0.0, "grad_norm": 0.05055604502558708, "global_step": 35285, "epoch": 980, "lr": 0.001} +{"train_loss": 0.034612059593200684, "train_loss_bc": 0.034612059593200684, "train_loss_llm": 0.0, "grad_norm": 0.09095728397369385, "global_step": 35286, "epoch": 980, "lr": 0.001} +{"train_loss": 0.03261256590485573, "train_loss_bc": 0.03261256590485573, "train_loss_llm": 0.0, "grad_norm": 0.12021990865468979, "global_step": 35287, "epoch": 980, "lr": 0.001} +{"train_loss": 0.030795292928814888, "train_loss_bc": 0.030795292928814888, "train_loss_llm": 0.0, "grad_norm": 0.16090214252471924, "global_step": 35288, "epoch": 980, "lr": 0.001} +{"train_loss": 0.030837276950478554, "train_loss_bc": 0.030837276950478554, "train_loss_llm": 0.0, "grad_norm": 0.041846636682748795, "global_step": 35289, "epoch": 980, "lr": 0.001} +{"train_loss": 0.03034895472228527, "train_loss_bc": 0.03034895472228527, "train_loss_llm": 0.0, "grad_norm": 0.07575257867574692, "global_step": 35290, "epoch": 980, "lr": 0.001} +{"train_loss": 0.031022625043988228, "train_loss_bc": 0.031022625043988228, "train_loss_llm": 0.0, "grad_norm": 0.09988686442375183, "global_step": 35291, "epoch": 980, "lr": 0.001} +{"train_loss": 0.028591280803084373, "train_loss_bc": 0.028591280803084373, "train_loss_llm": 0.0, "grad_norm": 0.12986186146736145, "global_step": 35292, "epoch": 980, "lr": 0.001} +{"train_loss": 0.021873952820897102, "train_loss_bc": 0.021873952820897102, "train_loss_llm": 0.0, "grad_norm": 0.02817203663289547, "global_step": 35293, "epoch": 980, "lr": 0.001} +{"train_loss": 0.030083246529102325, "train_loss_bc": 0.030083246529102325, "train_loss_llm": 0.0, "grad_norm": 0.05123055353760719, "global_step": 35294, "epoch": 980, "lr": 0.001} +{"train_loss": 0.026033690199255943, "train_loss_bc": 0.026033690199255943, "train_loss_llm": 0.0, "grad_norm": 0.066642627120018, "global_step": 35295, "epoch": 980, "lr": 0.001} +{"train_loss": 0.024875501170754433, "train_loss_bc": 0.024857403710484505, "train_loss_llm": 0.0018097758293151855, "grad_norm": 0.08029912412166595, "global_step": 35296, "epoch": 980, "lr": 0.001} +{"train_loss": 0.027651261538267136, "train_loss_bc": 0.027651261538267136, "train_loss_llm": 0.0, "grad_norm": 0.03832339122891426, "global_step": 35297, "epoch": 980, "lr": 0.001} +{"train_loss": 0.020190555602312088, "train_loss_bc": 0.020190555602312088, "train_loss_llm": 0.0, "grad_norm": 0.05448470264673233, "global_step": 35298, "epoch": 980, "lr": 0.001} +{"train_loss": 0.028802186250686646, "train_loss_bc": 0.028802186250686646, "train_loss_llm": 0.0, "grad_norm": 0.07762905955314636, "global_step": 35299, "epoch": 980, "lr": 0.001} +{"train_loss": 0.028511447831988335, "train_loss_bc": 0.028511447831988335, "train_loss_llm": 0.0, "grad_norm": 0.08989937603473663, "global_step": 35300, "epoch": 980, "lr": 0.001} +{"train_loss": 0.032395776361227036, "train_loss_bc": 0.032395776361227036, "train_loss_llm": 0.0, "grad_norm": 0.050641197711229324, "global_step": 35301, "epoch": 980, "lr": 0.001} +{"train_loss": 0.0264251958578825, "train_loss_bc": 0.0264251958578825, "train_loss_llm": 0.0, "grad_norm": 0.07631231844425201, "global_step": 35302, "epoch": 980, "lr": 0.001} +{"train_loss": 0.03015151247382164, "train_loss_bc": 0.03015151247382164, "train_loss_llm": 0.0, "grad_norm": 0.10307701677083969, "global_step": 35303, "epoch": 980, "lr": 0.001} +{"train_loss": 0.024703901261091232, "train_loss_bc": 0.024703901261091232, "train_loss_llm": 0.0, "grad_norm": 0.1206916943192482, "global_step": 35304, "epoch": 980, "lr": 0.001} +{"train_loss": 0.029252752661705017, "train_loss_bc": 0.029252752661705017, "train_loss_llm": 0.0, "grad_norm": 0.04407569766044617, "global_step": 35305, "epoch": 980, "lr": 0.001} +{"train_loss": 0.027475181967020035, "train_loss_bc": 0.027475181967020035, "train_loss_llm": 0.0, "grad_norm": 0.0693577378988266, "global_step": 35306, "epoch": 980, "lr": 0.001} +{"train_loss": 0.032002732157707214, "train_loss_bc": 0.032002732157707214, "train_loss_llm": 0.0, "grad_norm": 0.09149099886417389, "global_step": 35307, "epoch": 980, "lr": 0.001} +{"train_loss": 0.02715558186173439, "train_loss_bc": 0.02715558186173439, "train_loss_llm": 0.0, "grad_norm": 0.11326595395803452, "global_step": 35308, "epoch": 980, "lr": 0.001} +{"train_loss": 0.029959985986351967, "train_loss_bc": 0.029959985986351967, "train_loss_llm": 0.0, "grad_norm": 0.04162268340587616, "global_step": 35309, "epoch": 980, "lr": 0.001} +{"train_loss": 0.02372301183640957, "train_loss_bc": 0.023715728893876076, "train_loss_llm": 0.0007283687591552734, "grad_norm": 0.07698282599449158, "global_step": 35310, "epoch": 980, "lr": 0.001} +{"train_loss": 0.02473009191453457, "train_loss_bc": 0.02473009191453457, "train_loss_llm": 0.0, "grad_norm": 0.09785774350166321, "global_step": 35311, "epoch": 980, "lr": 0.001} +{"train_loss": 0.03238082677125931, "train_loss_bc": 0.03235829621553421, "train_loss_llm": 0.0022530555725097656, "grad_norm": 0.12357787787914276, "global_step": 35312, "epoch": 980, "lr": 0.001} +{"train_loss": 0.03209467604756355, "train_loss_bc": 0.03208736702799797, "train_loss_llm": 0.0007309615612030029, "grad_norm": 0.0495019294321537, "global_step": 35313, "epoch": 980, "lr": 0.001} +{"train_loss": 0.030453018844127655, "train_loss_bc": 0.03044292703270912, "train_loss_llm": 0.0010092556476593018, "grad_norm": 0.08823894709348679, "global_step": 35314, "epoch": 980, "lr": 0.001} +{"train_loss": 0.028722073313676648, "train_loss_bc": 0.0292583666741848, "train_loss_llm": 0.0, "grad_norm": 0.10995783656835556, "global_step": 35315, "epoch": 980, "lr": 0.001, "train/cumulative_reward": 46.948310866046334, "train/mean_score": 10.69728375769391, "train/success_rate": 0.3, "test/cumulative_reward": 44.00738086567429, "test/mean_score": 7.358517254246502, "test/success_rate": 0.18, "val_loss": 0.10988905280828476, "train_action_mse_error": 0.03176404908299446} +{"train_loss": 0.03247721493244171, "train_loss_bc": 0.03247721493244171, "train_loss_llm": 0.0, "grad_norm": 0.12756162881851196, "global_step": 35316, "epoch": 981, "lr": 0.001} +{"train_loss": 0.028987320140004158, "train_loss_bc": 0.02896740660071373, "train_loss_llm": 0.00199127197265625, "grad_norm": 0.0417732335627079, "global_step": 35317, "epoch": 981, "lr": 0.001} +{"train_loss": 0.02852146327495575, "train_loss_bc": 0.02852146327495575, "train_loss_llm": 0.0, "grad_norm": 0.06899580359458923, "global_step": 35318, "epoch": 981, "lr": 0.001} +{"train_loss": 0.02725471369922161, "train_loss_bc": 0.027237746864557266, "train_loss_llm": 0.0016967058181762695, "grad_norm": 0.09036068618297577, "global_step": 35319, "epoch": 981, "lr": 0.001} +{"train_loss": 0.030379237607121468, "train_loss_bc": 0.030379237607121468, "train_loss_llm": 0.0, "grad_norm": 0.10783399641513824, "global_step": 35320, "epoch": 981, "lr": 0.001} +{"train_loss": 0.03049461916089058, "train_loss_bc": 0.03049461916089058, "train_loss_llm": 0.0, "grad_norm": 0.05085408687591553, "global_step": 35321, "epoch": 981, "lr": 0.001} +{"train_loss": 0.027324937283992767, "train_loss_bc": 0.027324937283992767, "train_loss_llm": 0.0, "grad_norm": 0.07751411199569702, "global_step": 35322, "epoch": 981, "lr": 0.001} +{"train_loss": 0.025941329076886177, "train_loss_bc": 0.025941329076886177, "train_loss_llm": 0.0, "grad_norm": 0.09657792747020721, "global_step": 35323, "epoch": 981, "lr": 0.001} +{"train_loss": 0.027021843940019608, "train_loss_bc": 0.02700887992978096, "train_loss_llm": 0.0012964308261871338, "grad_norm": 0.11839962750673294, "global_step": 35324, "epoch": 981, "lr": 0.001} +{"train_loss": 0.024472149088978767, "train_loss_bc": 0.024466685950756073, "train_loss_llm": 0.0005463063716888428, "grad_norm": 0.038864895701408386, "global_step": 35325, "epoch": 981, "lr": 0.001} +{"train_loss": 0.026415996253490448, "train_loss_bc": 0.026415996253490448, "train_loss_llm": 0.0, "grad_norm": 0.04952239617705345, "global_step": 35326, "epoch": 981, "lr": 0.001} +{"train_loss": 0.03156496584415436, "train_loss_bc": 0.03156496584415436, "train_loss_llm": 0.0, "grad_norm": 0.07671910524368286, "global_step": 35327, "epoch": 981, "lr": 0.001} +{"train_loss": 0.030980804935097694, "train_loss_bc": 0.030980804935097694, "train_loss_llm": 0.0, "grad_norm": 0.10886666923761368, "global_step": 35328, "epoch": 981, "lr": 0.001} +{"train_loss": 0.029247058555483818, "train_loss_bc": 0.029247058555483818, "train_loss_llm": 0.0, "grad_norm": 0.03505896031856537, "global_step": 35329, "epoch": 981, "lr": 0.001} +{"train_loss": 0.023372622206807137, "train_loss_bc": 0.023372622206807137, "train_loss_llm": 0.0, "grad_norm": 0.06723596155643463, "global_step": 35330, "epoch": 981, "lr": 0.001} +{"train_loss": 0.028369786217808723, "train_loss_bc": 0.028369786217808723, "train_loss_llm": 0.0, "grad_norm": 0.09147170186042786, "global_step": 35331, "epoch": 981, "lr": 0.001} +{"train_loss": 0.030725782737135887, "train_loss_bc": 0.03071587160229683, "train_loss_llm": 0.0009910762310028076, "grad_norm": 0.11446675658226013, "global_step": 35332, "epoch": 981, "lr": 0.001} +{"train_loss": 0.02688918262720108, "train_loss_bc": 0.02688918262720108, "train_loss_llm": 0.0, "grad_norm": 0.037587180733680725, "global_step": 35333, "epoch": 981, "lr": 0.001} +{"train_loss": 0.02745729498565197, "train_loss_bc": 0.027452360838651657, "train_loss_llm": 0.0004933476448059082, "grad_norm": 0.06309942156076431, "global_step": 35334, "epoch": 981, "lr": 0.001} +{"train_loss": 0.03400593623518944, "train_loss_bc": 0.0340007059276104, "train_loss_llm": 0.0005230307579040527, "grad_norm": 0.10033669322729111, "global_step": 35335, "epoch": 981, "lr": 0.001} +{"train_loss": 0.03136987239122391, "train_loss_bc": 0.03136987239122391, "train_loss_llm": 0.0, "grad_norm": 0.11879722774028778, "global_step": 35336, "epoch": 981, "lr": 0.001} +{"train_loss": 0.025226673111319542, "train_loss_bc": 0.025226673111319542, "train_loss_llm": 0.0, "grad_norm": 0.04607716575264931, "global_step": 35337, "epoch": 981, "lr": 0.001} +{"train_loss": 0.028708383440971375, "train_loss_bc": 0.028708383440971375, "train_loss_llm": 0.0, "grad_norm": 0.08781720697879791, "global_step": 35338, "epoch": 981, "lr": 0.001} +{"train_loss": 0.029080694541335106, "train_loss_bc": 0.029080694541335106, "train_loss_llm": 0.0, "grad_norm": 0.10181883722543716, "global_step": 35339, "epoch": 981, "lr": 0.001} +{"train_loss": 0.028718994930386543, "train_loss_bc": 0.02869541011750698, "train_loss_llm": 0.0023585259914398193, "grad_norm": 0.11214029043912888, "global_step": 35340, "epoch": 981, "lr": 0.001} +{"train_loss": 0.026919053867459297, "train_loss_bc": 0.026903554797172546, "train_loss_llm": 0.0015499591827392578, "grad_norm": 0.05644645541906357, "global_step": 35341, "epoch": 981, "lr": 0.001} +{"train_loss": 0.03069581463932991, "train_loss_bc": 0.03069581463932991, "train_loss_llm": 0.0, "grad_norm": 0.07750458270311356, "global_step": 35342, "epoch": 981, "lr": 0.001} +{"train_loss": 0.024144701659679413, "train_loss_bc": 0.024144701659679413, "train_loss_llm": 0.0, "grad_norm": 0.10346966236829758, "global_step": 35343, "epoch": 981, "lr": 0.001} +{"train_loss": 0.030595703050494194, "train_loss_bc": 0.03057371824979782, "train_loss_llm": 0.0021984875202178955, "grad_norm": 0.13241520524024963, "global_step": 35344, "epoch": 981, "lr": 0.001} +{"train_loss": 0.030717145651578903, "train_loss_bc": 0.030716385692358017, "train_loss_llm": 7.599592208862305e-05, "grad_norm": 0.04177026450634003, "global_step": 35345, "epoch": 981, "lr": 0.001} +{"train_loss": 0.03299902379512787, "train_loss_bc": 0.03299902379512787, "train_loss_llm": 0.0, "grad_norm": 0.07626263797283173, "global_step": 35346, "epoch": 981, "lr": 0.001} +{"train_loss": 0.024701906368136406, "train_loss_bc": 0.024697832763195038, "train_loss_llm": 0.0004073679447174072, "grad_norm": 0.08990587294101715, "global_step": 35347, "epoch": 981, "lr": 0.001} +{"train_loss": 0.024590404704213142, "train_loss_bc": 0.024589572101831436, "train_loss_llm": 8.317828178405762e-05, "grad_norm": 0.10289456695318222, "global_step": 35348, "epoch": 981, "lr": 0.001} +{"train_loss": 0.028453316539525986, "train_loss_bc": 0.028444601222872734, "train_loss_llm": 0.0008715689182281494, "grad_norm": 0.030850162729620934, "global_step": 35349, "epoch": 981, "lr": 0.001} +{"train_loss": 0.03181851655244827, "train_loss_bc": 0.03181495890021324, "train_loss_llm": 0.0003559291362762451, "grad_norm": 0.045306187123060226, "global_step": 35350, "epoch": 981, "lr": 0.001} +{"train_loss": 0.0285327456270655, "train_loss_bc": 0.02653437852859497, "train_loss_llm": 0.0, "grad_norm": 0.08030033111572266, "global_step": 35351, "epoch": 981, "lr": 0.001, "val_loss": 0.1102471649646759} +{"train_loss": 0.026333417743444443, "train_loss_bc": 0.026333417743444443, "train_loss_llm": 0.0, "grad_norm": 0.09660165011882782, "global_step": 35352, "epoch": 982, "lr": 0.001} +{"train_loss": 0.027898499742150307, "train_loss_bc": 0.027898499742150307, "train_loss_llm": 0.0, "grad_norm": 0.044593390077352524, "global_step": 35353, "epoch": 982, "lr": 0.001} +{"train_loss": 0.031114572659134865, "train_loss_bc": 0.031114572659134865, "train_loss_llm": 0.0, "grad_norm": 0.06287121027708054, "global_step": 35354, "epoch": 982, "lr": 0.001} +{"train_loss": 0.028546668589115143, "train_loss_bc": 0.028546668589115143, "train_loss_llm": 0.0, "grad_norm": 0.0831824541091919, "global_step": 35355, "epoch": 982, "lr": 0.001} +{"train_loss": 0.03033275157213211, "train_loss_bc": 0.030315745621919632, "train_loss_llm": 0.0017006099224090576, "grad_norm": 0.09993726760149002, "global_step": 35356, "epoch": 982, "lr": 0.001} +{"train_loss": 0.02600383758544922, "train_loss_bc": 0.02600383758544922, "train_loss_llm": 0.0, "grad_norm": 0.03528409078717232, "global_step": 35357, "epoch": 982, "lr": 0.001} +{"train_loss": 0.026763951405882835, "train_loss_bc": 0.026758942753076553, "train_loss_llm": 0.0005009174346923828, "grad_norm": 0.06241179630160332, "global_step": 35358, "epoch": 982, "lr": 0.001} +{"train_loss": 0.02932709828019142, "train_loss_bc": 0.02930845133960247, "train_loss_llm": 0.0018647611141204834, "grad_norm": 0.08766327798366547, "global_step": 35359, "epoch": 982, "lr": 0.001} +{"train_loss": 0.025225376710295677, "train_loss_bc": 0.025225376710295677, "train_loss_llm": 0.0, "grad_norm": 0.1084069013595581, "global_step": 35360, "epoch": 982, "lr": 0.001} +{"train_loss": 0.027573516592383385, "train_loss_bc": 0.027573516592383385, "train_loss_llm": 0.0, "grad_norm": 0.05306454375386238, "global_step": 35361, "epoch": 982, "lr": 0.001} +{"train_loss": 0.03517981991171837, "train_loss_bc": 0.03516370803117752, "train_loss_llm": 0.001611262559890747, "grad_norm": 0.08248932659626007, "global_step": 35362, "epoch": 982, "lr": 0.001} +{"train_loss": 0.03380362689495087, "train_loss_bc": 0.03380362689495087, "train_loss_llm": 0.0, "grad_norm": 0.134496808052063, "global_step": 35363, "epoch": 982, "lr": 0.001} +{"train_loss": 0.024151429533958435, "train_loss_bc": 0.024151429533958435, "train_loss_llm": 0.0, "grad_norm": 0.15974313020706177, "global_step": 35364, "epoch": 982, "lr": 0.001} +{"train_loss": 0.029117386788129807, "train_loss_bc": 0.029117386788129807, "train_loss_llm": 0.0, "grad_norm": 0.04172859340906143, "global_step": 35365, "epoch": 982, "lr": 0.001} +{"train_loss": 0.03336631879210472, "train_loss_bc": 0.03336631879210472, "train_loss_llm": 0.0, "grad_norm": 0.07781427353620529, "global_step": 35366, "epoch": 982, "lr": 0.001} +{"train_loss": 0.035993244498968124, "train_loss_bc": 0.035993244498968124, "train_loss_llm": 0.0, "grad_norm": 0.10153514891862869, "global_step": 35367, "epoch": 982, "lr": 0.001} +{"train_loss": 0.033711276948451996, "train_loss_bc": 0.033711276948451996, "train_loss_llm": 0.0, "grad_norm": 0.1373906135559082, "global_step": 35368, "epoch": 982, "lr": 0.001} +{"train_loss": 0.03245130553841591, "train_loss_bc": 0.03245130553841591, "train_loss_llm": 0.0, "grad_norm": 0.0420290008187294, "global_step": 35369, "epoch": 982, "lr": 0.001} +{"train_loss": 0.025376686826348305, "train_loss_bc": 0.025376686826348305, "train_loss_llm": 0.0, "grad_norm": 0.05086493492126465, "global_step": 35370, "epoch": 982, "lr": 0.001} +{"train_loss": 0.030487682670354843, "train_loss_bc": 0.030487682670354843, "train_loss_llm": 0.0, "grad_norm": 0.07545208185911179, "global_step": 35371, "epoch": 982, "lr": 0.001} +{"train_loss": 0.024707144126296043, "train_loss_bc": 0.024707144126296043, "train_loss_llm": 0.0, "grad_norm": 0.08794938027858734, "global_step": 35372, "epoch": 982, "lr": 0.001} +{"train_loss": 0.025302056223154068, "train_loss_bc": 0.025302056223154068, "train_loss_llm": 0.0, "grad_norm": 0.033798493444919586, "global_step": 35373, "epoch": 982, "lr": 0.001} +{"train_loss": 0.03043307177722454, "train_loss_bc": 0.03043307177722454, "train_loss_llm": 0.0, "grad_norm": 0.06347818672657013, "global_step": 35374, "epoch": 982, "lr": 0.001} +{"train_loss": 0.023916194215416908, "train_loss_bc": 0.023894958198070526, "train_loss_llm": 0.0021235644817352295, "grad_norm": 0.08539305627346039, "global_step": 35375, "epoch": 982, "lr": 0.001} +{"train_loss": 0.026140417903661728, "train_loss_bc": 0.026140417903661728, "train_loss_llm": 0.0, "grad_norm": 0.11387111991643906, "global_step": 35376, "epoch": 982, "lr": 0.001} +{"train_loss": 0.029944630339741707, "train_loss_bc": 0.029944630339741707, "train_loss_llm": 0.0, "grad_norm": 0.04863002523779869, "global_step": 35377, "epoch": 982, "lr": 0.001} +{"train_loss": 0.023378238081932068, "train_loss_bc": 0.023378238081932068, "train_loss_llm": 0.0, "grad_norm": 0.06474100053310394, "global_step": 35378, "epoch": 982, "lr": 0.001} +{"train_loss": 0.02867663837969303, "train_loss_bc": 0.02867663837969303, "train_loss_llm": 0.0, "grad_norm": 0.11496292799711227, "global_step": 35379, "epoch": 982, "lr": 0.001} +{"train_loss": 0.02997957356274128, "train_loss_bc": 0.02997957356274128, "train_loss_llm": 0.0, "grad_norm": 0.13569030165672302, "global_step": 35380, "epoch": 982, "lr": 0.001} +{"train_loss": 0.03130662068724632, "train_loss_bc": 0.03130662068724632, "train_loss_llm": 0.0, "grad_norm": 0.047287728637456894, "global_step": 35381, "epoch": 982, "lr": 0.001} +{"train_loss": 0.02745979093015194, "train_loss_bc": 0.027456611394882202, "train_loss_llm": 0.0003179311752319336, "grad_norm": 0.08096131682395935, "global_step": 35382, "epoch": 982, "lr": 0.001} +{"train_loss": 0.02784557081758976, "train_loss_bc": 0.02784557081758976, "train_loss_llm": 0.0, "grad_norm": 0.1204739585518837, "global_step": 35383, "epoch": 982, "lr": 0.001} +{"train_loss": 0.028322884812951088, "train_loss_bc": 0.02831730619072914, "train_loss_llm": 0.0005577802658081055, "grad_norm": 0.14087659120559692, "global_step": 35384, "epoch": 982, "lr": 0.001} +{"train_loss": 0.030289307236671448, "train_loss_bc": 0.030275307595729828, "train_loss_llm": 0.0013999640941619873, "grad_norm": 0.04201200604438782, "global_step": 35385, "epoch": 982, "lr": 0.001} +{"train_loss": 0.025936996564269066, "train_loss_bc": 0.02591833844780922, "train_loss_llm": 0.0018658936023712158, "grad_norm": 0.059905532747507095, "global_step": 35386, "epoch": 982, "lr": 0.001} +{"train_loss": 0.02854831682311164, "train_loss_bc": 0.021341800689697266, "train_loss_llm": 0.0, "grad_norm": 0.08520203083753586, "global_step": 35387, "epoch": 982, "lr": 0.001, "val_loss": 0.11018352210521698} +{"train_loss": 0.026240168139338493, "train_loss_bc": 0.026240168139338493, "train_loss_llm": 0.0, "grad_norm": 0.104705810546875, "global_step": 35388, "epoch": 983, "lr": 0.001} +{"train_loss": 0.025023529306054115, "train_loss_bc": 0.025023529306054115, "train_loss_llm": 0.0, "grad_norm": 0.03435478359460831, "global_step": 35389, "epoch": 983, "lr": 0.001} +{"train_loss": 0.02562253177165985, "train_loss_bc": 0.02562253177165985, "train_loss_llm": 0.0, "grad_norm": 0.05148307606577873, "global_step": 35390, "epoch": 983, "lr": 0.001} +{"train_loss": 0.03151720017194748, "train_loss_bc": 0.03151720017194748, "train_loss_llm": 0.0, "grad_norm": 0.07858627289533615, "global_step": 35391, "epoch": 983, "lr": 0.001} +{"train_loss": 0.02520904317498207, "train_loss_bc": 0.02520904317498207, "train_loss_llm": 0.0, "grad_norm": 0.09987620264291763, "global_step": 35392, "epoch": 983, "lr": 0.001} +{"train_loss": 0.028177693486213684, "train_loss_bc": 0.028177693486213684, "train_loss_llm": 0.0, "grad_norm": 0.03477863594889641, "global_step": 35393, "epoch": 983, "lr": 0.001} +{"train_loss": 0.028711944818496704, "train_loss_bc": 0.028711944818496704, "train_loss_llm": 0.0, "grad_norm": 0.06504976749420166, "global_step": 35394, "epoch": 983, "lr": 0.001} +{"train_loss": 0.03171267732977867, "train_loss_bc": 0.03171267732977867, "train_loss_llm": 0.0, "grad_norm": 0.0897580161690712, "global_step": 35395, "epoch": 983, "lr": 0.001} +{"train_loss": 0.025813555344939232, "train_loss_bc": 0.025813555344939232, "train_loss_llm": 0.0, "grad_norm": 0.11564046144485474, "global_step": 35396, "epoch": 983, "lr": 0.001} +{"train_loss": 0.03229237347841263, "train_loss_bc": 0.03229237347841263, "train_loss_llm": 0.0, "grad_norm": 0.04635849967598915, "global_step": 35397, "epoch": 983, "lr": 0.001} +{"train_loss": 0.03278733417391777, "train_loss_bc": 0.03275682032108307, "train_loss_llm": 0.0030514299869537354, "grad_norm": 0.08009110391139984, "global_step": 35398, "epoch": 983, "lr": 0.001} +{"train_loss": 0.030943971127271652, "train_loss_bc": 0.030943971127271652, "train_loss_llm": 0.0, "grad_norm": 0.11494365334510803, "global_step": 35399, "epoch": 983, "lr": 0.001} +{"train_loss": 0.033540740609169006, "train_loss_bc": 0.03352958709001541, "train_loss_llm": 0.001115262508392334, "grad_norm": 0.13084544241428375, "global_step": 35400, "epoch": 983, "lr": 0.001} +{"train_loss": 0.02462305687367916, "train_loss_bc": 0.02462305687367916, "train_loss_llm": 0.0, "grad_norm": 0.031982433050870895, "global_step": 35401, "epoch": 983, "lr": 0.001} +{"train_loss": 0.031799398362636566, "train_loss_bc": 0.031799398362636566, "train_loss_llm": 0.0, "grad_norm": 0.06532230973243713, "global_step": 35402, "epoch": 983, "lr": 0.001} +{"train_loss": 0.02873728796839714, "train_loss_bc": 0.02873728796839714, "train_loss_llm": 0.0, "grad_norm": 0.09814456105232239, "global_step": 35403, "epoch": 983, "lr": 0.001} +{"train_loss": 0.030821865424513817, "train_loss_bc": 0.030815014615654945, "train_loss_llm": 0.0006850957870483398, "grad_norm": 0.1302206814289093, "global_step": 35404, "epoch": 983, "lr": 0.001} +{"train_loss": 0.02425401099026203, "train_loss_bc": 0.02423839643597603, "train_loss_llm": 0.0015614628791809082, "grad_norm": 0.03167600929737091, "global_step": 35405, "epoch": 983, "lr": 0.001} +{"train_loss": 0.023760898038744926, "train_loss_bc": 0.023756973445415497, "train_loss_llm": 0.0003924071788787842, "grad_norm": 0.046540677547454834, "global_step": 35406, "epoch": 983, "lr": 0.001} +{"train_loss": 0.028015023097395897, "train_loss_bc": 0.028015023097395897, "train_loss_llm": 0.0, "grad_norm": 0.06997433304786682, "global_step": 35407, "epoch": 983, "lr": 0.001} +{"train_loss": 0.02865304984152317, "train_loss_bc": 0.02865304984152317, "train_loss_llm": 0.0, "grad_norm": 0.09669329971075058, "global_step": 35408, "epoch": 983, "lr": 0.001} +{"train_loss": 0.022398527711629868, "train_loss_bc": 0.022398527711629868, "train_loss_llm": 0.0, "grad_norm": 0.03204759210348129, "global_step": 35409, "epoch": 983, "lr": 0.001} +{"train_loss": 0.02755529247224331, "train_loss_bc": 0.02753458544611931, "train_loss_llm": 0.0020707547664642334, "grad_norm": 0.05808467045426369, "global_step": 35410, "epoch": 983, "lr": 0.001} +{"train_loss": 0.026012714952230453, "train_loss_bc": 0.026012714952230453, "train_loss_llm": 0.0, "grad_norm": 0.08857329934835434, "global_step": 35411, "epoch": 983, "lr": 0.001} +{"train_loss": 0.02363186702132225, "train_loss_bc": 0.023608539253473282, "train_loss_llm": 0.002332717180252075, "grad_norm": 0.10959777235984802, "global_step": 35412, "epoch": 983, "lr": 0.001} +{"train_loss": 0.034496139734983444, "train_loss_bc": 0.03449207544326782, "train_loss_llm": 0.00040656328201293945, "grad_norm": 0.04267481341958046, "global_step": 35413, "epoch": 983, "lr": 0.001} +{"train_loss": 0.029752448201179504, "train_loss_bc": 0.029743105173110962, "train_loss_llm": 0.0009342432022094727, "grad_norm": 0.069145068526268, "global_step": 35414, "epoch": 983, "lr": 0.001} +{"train_loss": 0.02734261006116867, "train_loss_bc": 0.02734261006116867, "train_loss_llm": 0.0, "grad_norm": 0.10741978883743286, "global_step": 35415, "epoch": 983, "lr": 0.001} +{"train_loss": 0.0354963019490242, "train_loss_bc": 0.0354963019490242, "train_loss_llm": 0.0, "grad_norm": 0.13012464344501495, "global_step": 35416, "epoch": 983, "lr": 0.001} +{"train_loss": 0.026288585737347603, "train_loss_bc": 0.026288585737347603, "train_loss_llm": 0.0, "grad_norm": 0.03781301900744438, "global_step": 35417, "epoch": 983, "lr": 0.001} +{"train_loss": 0.025495756417512894, "train_loss_bc": 0.02548098750412464, "train_loss_llm": 0.0014769434928894043, "grad_norm": 0.05711950361728668, "global_step": 35418, "epoch": 983, "lr": 0.001} +{"train_loss": 0.03388243541121483, "train_loss_bc": 0.03388243541121483, "train_loss_llm": 0.0, "grad_norm": 0.09515240043401718, "global_step": 35419, "epoch": 983, "lr": 0.001} +{"train_loss": 0.026730384677648544, "train_loss_bc": 0.026730384677648544, "train_loss_llm": 0.0, "grad_norm": 0.13361141085624695, "global_step": 35420, "epoch": 983, "lr": 0.001} +{"train_loss": 0.030683910474181175, "train_loss_bc": 0.03068387135863304, "train_loss_llm": 3.9637088775634766e-06, "grad_norm": 0.037563204765319824, "global_step": 35421, "epoch": 983, "lr": 0.001} +{"train_loss": 0.0295058973133564, "train_loss_bc": 0.0295058973133564, "train_loss_llm": 0.0, "grad_norm": 0.06995738297700882, "global_step": 35422, "epoch": 983, "lr": 0.001} +{"train_loss": 0.02860416906575362, "train_loss_bc": 0.03221273422241211, "train_loss_llm": 0.000712662935256958, "grad_norm": 0.10910111665725708, "global_step": 35423, "epoch": 983, "lr": 0.001, "val_loss": 0.11120737344026566} +{"train_loss": 0.02942158468067646, "train_loss_bc": 0.02942158468067646, "train_loss_llm": 0.0, "grad_norm": 0.11629770696163177, "global_step": 35424, "epoch": 984, "lr": 0.001} +{"train_loss": 0.02203253284096718, "train_loss_bc": 0.02203253284096718, "train_loss_llm": 0.0, "grad_norm": 0.028912771493196487, "global_step": 35425, "epoch": 984, "lr": 0.001} +{"train_loss": 0.022982586175203323, "train_loss_bc": 0.022982586175203323, "train_loss_llm": 0.0, "grad_norm": 0.05030782148241997, "global_step": 35426, "epoch": 984, "lr": 0.001} +{"train_loss": 0.030770808458328247, "train_loss_bc": 0.030770808458328247, "train_loss_llm": 0.0, "grad_norm": 0.07253389805555344, "global_step": 35427, "epoch": 984, "lr": 0.001} +{"train_loss": 0.02839071862399578, "train_loss_bc": 0.02839071862399578, "train_loss_llm": 0.0, "grad_norm": 0.08994390815496445, "global_step": 35428, "epoch": 984, "lr": 0.001} +{"train_loss": 0.026475340127944946, "train_loss_bc": 0.026475340127944946, "train_loss_llm": 0.0, "grad_norm": 0.037320856004953384, "global_step": 35429, "epoch": 984, "lr": 0.001} +{"train_loss": 0.02647959068417549, "train_loss_bc": 0.02647959068417549, "train_loss_llm": 0.0, "grad_norm": 0.05610731616616249, "global_step": 35430, "epoch": 984, "lr": 0.001} +{"train_loss": 0.023301050066947937, "train_loss_bc": 0.023301050066947937, "train_loss_llm": 0.0, "grad_norm": 0.07622848451137543, "global_step": 35431, "epoch": 984, "lr": 0.001} +{"train_loss": 0.029640154913067818, "train_loss_bc": 0.029640154913067818, "train_loss_llm": 0.0, "grad_norm": 0.09803900122642517, "global_step": 35432, "epoch": 984, "lr": 0.001} +{"train_loss": 0.028935033828020096, "train_loss_bc": 0.028935033828020096, "train_loss_llm": 0.0, "grad_norm": 0.044144157320261, "global_step": 35433, "epoch": 984, "lr": 0.001} +{"train_loss": 0.030836746096611023, "train_loss_bc": 0.030836746096611023, "train_loss_llm": 0.0, "grad_norm": 0.06889665871858597, "global_step": 35434, "epoch": 984, "lr": 0.001} +{"train_loss": 0.029269034042954445, "train_loss_bc": 0.029269034042954445, "train_loss_llm": 0.0, "grad_norm": 0.09151385724544525, "global_step": 35435, "epoch": 984, "lr": 0.001} +{"train_loss": 0.02708176150918007, "train_loss_bc": 0.02708176150918007, "train_loss_llm": 0.0, "grad_norm": 0.11841271072626114, "global_step": 35436, "epoch": 984, "lr": 0.001} +{"train_loss": 0.02602069266140461, "train_loss_bc": 0.026012787595391273, "train_loss_llm": 0.0007904171943664551, "grad_norm": 0.04655877873301506, "global_step": 35437, "epoch": 984, "lr": 0.001} +{"train_loss": 0.030653882771730423, "train_loss_bc": 0.030653882771730423, "train_loss_llm": 0.0, "grad_norm": 0.07626334577798843, "global_step": 35438, "epoch": 984, "lr": 0.001} +{"train_loss": 0.031344763934612274, "train_loss_bc": 0.031344763934612274, "train_loss_llm": 0.0, "grad_norm": 0.08769986778497696, "global_step": 35439, "epoch": 984, "lr": 0.001} +{"train_loss": 0.03039216436445713, "train_loss_bc": 0.03039216436445713, "train_loss_llm": 0.0, "grad_norm": 0.124314084649086, "global_step": 35440, "epoch": 984, "lr": 0.001} +{"train_loss": 0.02870314195752144, "train_loss_bc": 0.02870314195752144, "train_loss_llm": 0.0, "grad_norm": 0.029274096712470055, "global_step": 35441, "epoch": 984, "lr": 0.001} +{"train_loss": 0.029350727796554565, "train_loss_bc": 0.029350727796554565, "train_loss_llm": 0.0, "grad_norm": 0.04774409160017967, "global_step": 35442, "epoch": 984, "lr": 0.001} +{"train_loss": 0.031126948073506355, "train_loss_bc": 0.031126948073506355, "train_loss_llm": 0.0, "grad_norm": 0.058880437165498734, "global_step": 35443, "epoch": 984, "lr": 0.001} +{"train_loss": 0.029649069532752037, "train_loss_bc": 0.029649069532752037, "train_loss_llm": 0.0, "grad_norm": 0.07750651240348816, "global_step": 35444, "epoch": 984, "lr": 0.001} +{"train_loss": 0.029300326481461525, "train_loss_bc": 0.029300326481461525, "train_loss_llm": 0.0, "grad_norm": 0.04183313250541687, "global_step": 35445, "epoch": 984, "lr": 0.001} +{"train_loss": 0.0255719181150198, "train_loss_bc": 0.0255719181150198, "train_loss_llm": 0.0, "grad_norm": 0.06496194005012512, "global_step": 35446, "epoch": 984, "lr": 0.001} +{"train_loss": 0.027479566633701324, "train_loss_bc": 0.027479566633701324, "train_loss_llm": 0.0, "grad_norm": 0.09261736273765564, "global_step": 35447, "epoch": 984, "lr": 0.001} +{"train_loss": 0.026572108268737793, "train_loss_bc": 0.026565104722976685, "train_loss_llm": 0.0007003247737884521, "grad_norm": 0.10356243699789047, "global_step": 35448, "epoch": 984, "lr": 0.001} +{"train_loss": 0.02876308746635914, "train_loss_bc": 0.02876308746635914, "train_loss_llm": 0.0, "grad_norm": 0.03551650047302246, "global_step": 35449, "epoch": 984, "lr": 0.001} +{"train_loss": 0.030526116490364075, "train_loss_bc": 0.030522648245096207, "train_loss_llm": 0.00034680962562561035, "grad_norm": 0.060875583440065384, "global_step": 35450, "epoch": 984, "lr": 0.001} +{"train_loss": 0.024628115817904472, "train_loss_bc": 0.024622447788715363, "train_loss_llm": 0.0005668103694915771, "grad_norm": 0.07171906530857086, "global_step": 35451, "epoch": 984, "lr": 0.001} +{"train_loss": 0.02621307596564293, "train_loss_bc": 0.02621307596564293, "train_loss_llm": 0.0, "grad_norm": 0.08790109306573868, "global_step": 35452, "epoch": 984, "lr": 0.001} +{"train_loss": 0.03702295944094658, "train_loss_bc": 0.03702295944094658, "train_loss_llm": 0.0, "grad_norm": 0.04176797717809677, "global_step": 35453, "epoch": 984, "lr": 0.001} +{"train_loss": 0.031106948852539062, "train_loss_bc": 0.031106948852539062, "train_loss_llm": 0.0, "grad_norm": 0.08400101214647293, "global_step": 35454, "epoch": 984, "lr": 0.001} +{"train_loss": 0.031073957681655884, "train_loss_bc": 0.031073957681655884, "train_loss_llm": 0.0, "grad_norm": 0.09495525062084198, "global_step": 35455, "epoch": 984, "lr": 0.001} +{"train_loss": 0.021652191877365112, "train_loss_bc": 0.021652191877365112, "train_loss_llm": 0.0, "grad_norm": 0.11618955433368683, "global_step": 35456, "epoch": 984, "lr": 0.001} +{"train_loss": 0.030831556767225266, "train_loss_bc": 0.030831556767225266, "train_loss_llm": 0.0, "grad_norm": 0.04276489093899727, "global_step": 35457, "epoch": 984, "lr": 0.001} +{"train_loss": 0.028347481042146683, "train_loss_bc": 0.028334500268101692, "train_loss_llm": 0.0012980997562408447, "grad_norm": 0.07735269516706467, "global_step": 35458, "epoch": 984, "lr": 0.001} +{"train_loss": 0.0284548695716593, "train_loss_bc": 0.03242756053805351, "train_loss_llm": 0.0, "grad_norm": 0.10680398344993591, "global_step": 35459, "epoch": 984, "lr": 0.001, "val_loss": 0.11290960013866425} +{"train_loss": 0.024075044319033623, "train_loss_bc": 0.024075044319033623, "train_loss_llm": 0.0, "grad_norm": 0.1273641139268875, "global_step": 35460, "epoch": 985, "lr": 0.001} +{"train_loss": 0.02733430825173855, "train_loss_bc": 0.027329808101058006, "train_loss_llm": 0.0004500150680541992, "grad_norm": 0.037616174668073654, "global_step": 35461, "epoch": 985, "lr": 0.001} +{"train_loss": 0.029296530410647392, "train_loss_bc": 0.029295047745108604, "train_loss_llm": 0.00014826655387878418, "grad_norm": 0.0690230131149292, "global_step": 35462, "epoch": 985, "lr": 0.001} +{"train_loss": 0.022950895130634308, "train_loss_bc": 0.022950895130634308, "train_loss_llm": 0.0, "grad_norm": 0.08384684473276138, "global_step": 35463, "epoch": 985, "lr": 0.001} +{"train_loss": 0.022999748587608337, "train_loss_bc": 0.022999748587608337, "train_loss_llm": 0.0, "grad_norm": 0.09534105658531189, "global_step": 35464, "epoch": 985, "lr": 0.001} +{"train_loss": 0.027488216757774353, "train_loss_bc": 0.02748742513358593, "train_loss_llm": 7.918477058410645e-05, "grad_norm": 0.04092176631093025, "global_step": 35465, "epoch": 985, "lr": 0.001} +{"train_loss": 0.033969696611166, "train_loss_bc": 0.033938512206077576, "train_loss_llm": 0.003118276596069336, "grad_norm": 0.06234517693519592, "global_step": 35466, "epoch": 985, "lr": 0.001} +{"train_loss": 0.029881546273827553, "train_loss_bc": 0.029881546273827553, "train_loss_llm": 0.0, "grad_norm": 0.08903605490922928, "global_step": 35467, "epoch": 985, "lr": 0.001} +{"train_loss": 0.02545902691781521, "train_loss_bc": 0.02545902691781521, "train_loss_llm": 0.0, "grad_norm": 0.12526053190231323, "global_step": 35468, "epoch": 985, "lr": 0.001} +{"train_loss": 0.025223372504115105, "train_loss_bc": 0.025223372504115105, "train_loss_llm": 0.0, "grad_norm": 0.03567655757069588, "global_step": 35469, "epoch": 985, "lr": 0.001} +{"train_loss": 0.02687579207122326, "train_loss_bc": 0.02687579207122326, "train_loss_llm": 0.0, "grad_norm": 0.06345195323228836, "global_step": 35470, "epoch": 985, "lr": 0.001} +{"train_loss": 0.03093215450644493, "train_loss_bc": 0.030910085886716843, "train_loss_llm": 0.0022068917751312256, "grad_norm": 0.08834186941385269, "global_step": 35471, "epoch": 985, "lr": 0.001} +{"train_loss": 0.029385821893811226, "train_loss_bc": 0.029385821893811226, "train_loss_llm": 0.0, "grad_norm": 0.11493084579706192, "global_step": 35472, "epoch": 985, "lr": 0.001} +{"train_loss": 0.026185601949691772, "train_loss_bc": 0.026185601949691772, "train_loss_llm": 0.0, "grad_norm": 0.035386256873607635, "global_step": 35473, "epoch": 985, "lr": 0.001} +{"train_loss": 0.03017023205757141, "train_loss_bc": 0.030157849192619324, "train_loss_llm": 0.0012382566928863525, "grad_norm": 0.05544089153409004, "global_step": 35474, "epoch": 985, "lr": 0.001} +{"train_loss": 0.0261056087911129, "train_loss_bc": 0.0261056087911129, "train_loss_llm": 0.0, "grad_norm": 0.0880374014377594, "global_step": 35475, "epoch": 985, "lr": 0.001} +{"train_loss": 0.027767570689320564, "train_loss_bc": 0.027767570689320564, "train_loss_llm": 0.0, "grad_norm": 0.10356900095939636, "global_step": 35476, "epoch": 985, "lr": 0.001} +{"train_loss": 0.03102315217256546, "train_loss_bc": 0.03102315217256546, "train_loss_llm": 0.0, "grad_norm": 0.040628764778375626, "global_step": 35477, "epoch": 985, "lr": 0.001} +{"train_loss": 0.02646590583026409, "train_loss_bc": 0.02646590583026409, "train_loss_llm": 0.0, "grad_norm": 0.06224752962589264, "global_step": 35478, "epoch": 985, "lr": 0.001} +{"train_loss": 0.029054025188088417, "train_loss_bc": 0.029054025188088417, "train_loss_llm": 0.0, "grad_norm": 0.08077897131443024, "global_step": 35479, "epoch": 985, "lr": 0.001} +{"train_loss": 0.0296940878033638, "train_loss_bc": 0.0296940878033638, "train_loss_llm": 0.0, "grad_norm": 0.10558350384235382, "global_step": 35480, "epoch": 985, "lr": 0.001} +{"train_loss": 0.03065062314271927, "train_loss_bc": 0.03065062314271927, "train_loss_llm": 0.0, "grad_norm": 0.0435258187353611, "global_step": 35481, "epoch": 985, "lr": 0.001} +{"train_loss": 0.0334138497710228, "train_loss_bc": 0.033407580107450485, "train_loss_llm": 0.000626981258392334, "grad_norm": 0.06721679866313934, "global_step": 35482, "epoch": 985, "lr": 0.001} +{"train_loss": 0.03074958547949791, "train_loss_bc": 0.03074958547949791, "train_loss_llm": 0.0, "grad_norm": 0.09003547579050064, "global_step": 35483, "epoch": 985, "lr": 0.001} +{"train_loss": 0.031925301998853683, "train_loss_bc": 0.031925301998853683, "train_loss_llm": 0.0, "grad_norm": 0.11250254511833191, "global_step": 35484, "epoch": 985, "lr": 0.001} +{"train_loss": 0.020752573385834694, "train_loss_bc": 0.020748883485794067, "train_loss_llm": 0.00036901235580444336, "grad_norm": 0.03334598243236542, "global_step": 35485, "epoch": 985, "lr": 0.001} +{"train_loss": 0.026726601645350456, "train_loss_bc": 0.026720181107521057, "train_loss_llm": 0.0006420016288757324, "grad_norm": 0.05516836792230606, "global_step": 35486, "epoch": 985, "lr": 0.001} +{"train_loss": 0.03106127306818962, "train_loss_bc": 0.03106127306818962, "train_loss_llm": 0.0, "grad_norm": 0.07331801950931549, "global_step": 35487, "epoch": 985, "lr": 0.001} +{"train_loss": 0.031097833067178726, "train_loss_bc": 0.03109746053814888, "train_loss_llm": 3.7223100662231445e-05, "grad_norm": 0.09083028137683868, "global_step": 35488, "epoch": 985, "lr": 0.001} +{"train_loss": 0.026640523225069046, "train_loss_bc": 0.026640523225069046, "train_loss_llm": 0.0, "grad_norm": 0.041514962911605835, "global_step": 35489, "epoch": 985, "lr": 0.001} +{"train_loss": 0.033361922949552536, "train_loss_bc": 0.033361922949552536, "train_loss_llm": 0.0, "grad_norm": 0.08066646754741669, "global_step": 35490, "epoch": 985, "lr": 0.001} +{"train_loss": 0.03244268521666527, "train_loss_bc": 0.032437968999147415, "train_loss_llm": 0.0004716813564300537, "grad_norm": 0.10241203010082245, "global_step": 35491, "epoch": 985, "lr": 0.001} +{"train_loss": 0.03189472854137421, "train_loss_bc": 0.03189472854137421, "train_loss_llm": 0.0, "grad_norm": 0.1198468804359436, "global_step": 35492, "epoch": 985, "lr": 0.001} +{"train_loss": 0.030096476897597313, "train_loss_bc": 0.030096476897597313, "train_loss_llm": 0.0, "grad_norm": 0.03550495579838753, "global_step": 35493, "epoch": 985, "lr": 0.001} +{"train_loss": 0.027145449072122574, "train_loss_bc": 0.027135109528899193, "train_loss_llm": 0.0010339319705963135, "grad_norm": 0.055900514125823975, "global_step": 35494, "epoch": 985, "lr": 0.001} +{"train_loss": 0.02857536610422863, "train_loss_bc": 0.028415413573384285, "train_loss_llm": 0.0, "grad_norm": 0.07910031825304031, "global_step": 35495, "epoch": 985, "lr": 0.001, "train/cumulative_reward": 83.67172305016852, "train/mean_score": 18.845639704609418, "train/success_rate": 0.5, "test/cumulative_reward": 23.819404992065873, "test/mean_score": 6.792369462372986, "test/success_rate": 0.16, "val_loss": 0.11051014065742493, "train_action_mse_error": 0.030993249267339706} +{"train_loss": 0.026702551171183586, "train_loss_bc": 0.026698686182498932, "train_loss_llm": 0.0003865361213684082, "grad_norm": 0.09335725009441376, "global_step": 35496, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02223852276802063, "train_loss_bc": 0.02223852276802063, "train_loss_llm": 0.0, "grad_norm": 0.033721085637807846, "global_step": 35497, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02463373728096485, "train_loss_bc": 0.02463373728096485, "train_loss_llm": 0.0, "grad_norm": 0.05853135511279106, "global_step": 35498, "epoch": 986, "lr": 0.001} +{"train_loss": 0.0300084687769413, "train_loss_bc": 0.0300084687769413, "train_loss_llm": 0.0, "grad_norm": 0.0930304303765297, "global_step": 35499, "epoch": 986, "lr": 0.001} +{"train_loss": 0.03096139430999756, "train_loss_bc": 0.03096139430999756, "train_loss_llm": 0.0, "grad_norm": 0.1082141324877739, "global_step": 35500, "epoch": 986, "lr": 0.001} +{"train_loss": 0.026623468846082687, "train_loss_bc": 0.026613285765051842, "train_loss_llm": 0.0010183453559875488, "grad_norm": 0.03051004186272621, "global_step": 35501, "epoch": 986, "lr": 0.001} +{"train_loss": 0.026500718668103218, "train_loss_bc": 0.026494638994336128, "train_loss_llm": 0.0006080567836761475, "grad_norm": 0.05048253759741783, "global_step": 35502, "epoch": 986, "lr": 0.001} +{"train_loss": 0.029339810833334923, "train_loss_bc": 0.029339810833334923, "train_loss_llm": 0.0, "grad_norm": 0.07337582856416702, "global_step": 35503, "epoch": 986, "lr": 0.001} +{"train_loss": 0.023328538984060287, "train_loss_bc": 0.023321256041526794, "train_loss_llm": 0.000728219747543335, "grad_norm": 0.09026408195495605, "global_step": 35504, "epoch": 986, "lr": 0.001} +{"train_loss": 0.03586972877383232, "train_loss_bc": 0.03586972877383232, "train_loss_llm": 0.0, "grad_norm": 0.052338965237140656, "global_step": 35505, "epoch": 986, "lr": 0.001} +{"train_loss": 0.03509761765599251, "train_loss_bc": 0.03509761765599251, "train_loss_llm": 0.0, "grad_norm": 0.10566285252571106, "global_step": 35506, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02835591323673725, "train_loss_bc": 0.02835591323673725, "train_loss_llm": 0.0, "grad_norm": 0.12892714142799377, "global_step": 35507, "epoch": 986, "lr": 0.001} +{"train_loss": 0.027379779145121574, "train_loss_bc": 0.027379779145121574, "train_loss_llm": 0.0, "grad_norm": 0.1402067095041275, "global_step": 35508, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02479020319879055, "train_loss_bc": 0.024786794558167458, "train_loss_llm": 0.0003408491611480713, "grad_norm": 0.029846567660570145, "global_step": 35509, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02195691131055355, "train_loss_bc": 0.021953346207737923, "train_loss_llm": 0.0003565847873687744, "grad_norm": 0.04797765240073204, "global_step": 35510, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02973981760442257, "train_loss_bc": 0.02973981760442257, "train_loss_llm": 0.0, "grad_norm": 0.06977658718824387, "global_step": 35511, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02767678163945675, "train_loss_bc": 0.02767678163945675, "train_loss_llm": 0.0, "grad_norm": 0.09348951280117035, "global_step": 35512, "epoch": 986, "lr": 0.001} +{"train_loss": 0.03500012680888176, "train_loss_bc": 0.03498687967658043, "train_loss_llm": 0.0013248622417449951, "grad_norm": 0.048374470323324203, "global_step": 35513, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02656751126050949, "train_loss_bc": 0.02656751126050949, "train_loss_llm": 0.0, "grad_norm": 0.09055756032466888, "global_step": 35514, "epoch": 986, "lr": 0.001} +{"train_loss": 0.020654555410146713, "train_loss_bc": 0.020654555410146713, "train_loss_llm": 0.0, "grad_norm": 0.10994754731655121, "global_step": 35515, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02589423954486847, "train_loss_bc": 0.02589423954486847, "train_loss_llm": 0.0, "grad_norm": 0.1534169316291809, "global_step": 35516, "epoch": 986, "lr": 0.001} +{"train_loss": 0.026946138590574265, "train_loss_bc": 0.026946138590574265, "train_loss_llm": 0.0, "grad_norm": 0.04745376855134964, "global_step": 35517, "epoch": 986, "lr": 0.001} +{"train_loss": 0.032260358333587646, "train_loss_bc": 0.032260358333587646, "train_loss_llm": 0.0, "grad_norm": 0.08449118584394455, "global_step": 35518, "epoch": 986, "lr": 0.001} +{"train_loss": 0.0312676802277565, "train_loss_bc": 0.0312676802277565, "train_loss_llm": 0.0, "grad_norm": 0.11036314815282822, "global_step": 35519, "epoch": 986, "lr": 0.001} +{"train_loss": 0.031443916261196136, "train_loss_bc": 0.031443916261196136, "train_loss_llm": 0.0, "grad_norm": 0.13187381625175476, "global_step": 35520, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02673514559864998, "train_loss_bc": 0.02673514559864998, "train_loss_llm": 0.0, "grad_norm": 0.047867439687252045, "global_step": 35521, "epoch": 986, "lr": 0.001} +{"train_loss": 0.030542541295289993, "train_loss_bc": 0.030542541295289993, "train_loss_llm": 0.0, "grad_norm": 0.08786635845899582, "global_step": 35522, "epoch": 986, "lr": 0.001} +{"train_loss": 0.028188010677695274, "train_loss_bc": 0.028188010677695274, "train_loss_llm": 0.0, "grad_norm": 0.10814593732357025, "global_step": 35523, "epoch": 986, "lr": 0.001} +{"train_loss": 0.031100450083613396, "train_loss_bc": 0.031087571755051613, "train_loss_llm": 0.0012878179550170898, "grad_norm": 0.14081843197345734, "global_step": 35524, "epoch": 986, "lr": 0.001} +{"train_loss": 0.03528735786676407, "train_loss_bc": 0.03528735786676407, "train_loss_llm": 0.0, "grad_norm": 0.05109700560569763, "global_step": 35525, "epoch": 986, "lr": 0.001} +{"train_loss": 0.033474378287792206, "train_loss_bc": 0.033474378287792206, "train_loss_llm": 0.0, "grad_norm": 0.07685531675815582, "global_step": 35526, "epoch": 986, "lr": 0.001} +{"train_loss": 0.029738113284111023, "train_loss_bc": 0.029738113284111023, "train_loss_llm": 0.0, "grad_norm": 0.10452613979578018, "global_step": 35527, "epoch": 986, "lr": 0.001} +{"train_loss": 0.028111444786190987, "train_loss_bc": 0.028099099174141884, "train_loss_llm": 0.0012344717979431152, "grad_norm": 0.12719911336898804, "global_step": 35528, "epoch": 986, "lr": 0.001} +{"train_loss": 0.030943213030695915, "train_loss_bc": 0.030943213030695915, "train_loss_llm": 0.0, "grad_norm": 0.04231574386358261, "global_step": 35529, "epoch": 986, "lr": 0.001} +{"train_loss": 0.029291531071066856, "train_loss_bc": 0.029291531071066856, "train_loss_llm": 0.0, "grad_norm": 0.0654786229133606, "global_step": 35530, "epoch": 986, "lr": 0.001} +{"train_loss": 0.02881904262014561, "train_loss_bc": 0.03282985836267471, "train_loss_llm": 0.0004999339580535889, "grad_norm": 0.0882197842001915, "global_step": 35531, "epoch": 986, "lr": 0.001, "val_loss": 0.11038844287395477} +{"train_loss": 0.030129950493574142, "train_loss_bc": 0.030129950493574142, "train_loss_llm": 0.0, "grad_norm": 0.11381896585226059, "global_step": 35532, "epoch": 987, "lr": 0.001} +{"train_loss": 0.029478177428245544, "train_loss_bc": 0.029478177428245544, "train_loss_llm": 0.0, "grad_norm": 0.045296426862478256, "global_step": 35533, "epoch": 987, "lr": 0.001} +{"train_loss": 0.026998933404684067, "train_loss_bc": 0.026998933404684067, "train_loss_llm": 0.0, "grad_norm": 0.07321926951408386, "global_step": 35534, "epoch": 987, "lr": 0.001} +{"train_loss": 0.0301637202501297, "train_loss_bc": 0.0301637202501297, "train_loss_llm": 0.0, "grad_norm": 0.09454336762428284, "global_step": 35535, "epoch": 987, "lr": 0.001} +{"train_loss": 0.03419535607099533, "train_loss_bc": 0.034178778529167175, "train_loss_llm": 0.0016579031944274902, "grad_norm": 0.12854881584644318, "global_step": 35536, "epoch": 987, "lr": 0.001} +{"train_loss": 0.020591028034687042, "train_loss_bc": 0.020591028034687042, "train_loss_llm": 0.0, "grad_norm": 0.029787296429276466, "global_step": 35537, "epoch": 987, "lr": 0.001} +{"train_loss": 0.0283551886677742, "train_loss_bc": 0.0283551886677742, "train_loss_llm": 0.0, "grad_norm": 0.06089583411812782, "global_step": 35538, "epoch": 987, "lr": 0.001} +{"train_loss": 0.02881392277777195, "train_loss_bc": 0.02881392277777195, "train_loss_llm": 0.0, "grad_norm": 0.10064166784286499, "global_step": 35539, "epoch": 987, "lr": 0.001} +{"train_loss": 0.02744922786951065, "train_loss_bc": 0.02744922786951065, "train_loss_llm": 0.0, "grad_norm": 0.1106463372707367, "global_step": 35540, "epoch": 987, "lr": 0.001} +{"train_loss": 0.030784107744693756, "train_loss_bc": 0.030784107744693756, "train_loss_llm": 0.0, "grad_norm": 0.05120537057518959, "global_step": 35541, "epoch": 987, "lr": 0.001} +{"train_loss": 0.02674904279410839, "train_loss_bc": 0.02674904279410839, "train_loss_llm": 0.0, "grad_norm": 0.07873843610286713, "global_step": 35542, "epoch": 987, "lr": 0.001} +{"train_loss": 0.026752416044473648, "train_loss_bc": 0.026750482618808746, "train_loss_llm": 0.00019338726997375488, "grad_norm": 0.10885696858167648, "global_step": 35543, "epoch": 987, "lr": 0.001} +{"train_loss": 0.027372200042009354, "train_loss_bc": 0.027372200042009354, "train_loss_llm": 0.0, "grad_norm": 0.1488710641860962, "global_step": 35544, "epoch": 987, "lr": 0.001} +{"train_loss": 0.02574194222688675, "train_loss_bc": 0.02574194222688675, "train_loss_llm": 0.0, "grad_norm": 0.038622643798589706, "global_step": 35545, "epoch": 987, "lr": 0.001} +{"train_loss": 0.029949620366096497, "train_loss_bc": 0.029926722869277, "train_loss_llm": 0.0022898316383361816, "grad_norm": 0.06309886276721954, "global_step": 35546, "epoch": 987, "lr": 0.001} +{"train_loss": 0.02919342927634716, "train_loss_bc": 0.02919342927634716, "train_loss_llm": 0.0, "grad_norm": 0.08479716628789902, "global_step": 35547, "epoch": 987, "lr": 0.001} +{"train_loss": 0.028129376471042633, "train_loss_bc": 0.028129376471042633, "train_loss_llm": 0.0, "grad_norm": 0.10883977264165878, "global_step": 35548, "epoch": 987, "lr": 0.001} +{"train_loss": 0.026007281616330147, "train_loss_bc": 0.026007281616330147, "train_loss_llm": 0.0, "grad_norm": 0.042874254286289215, "global_step": 35549, "epoch": 987, "lr": 0.001} +{"train_loss": 0.029211251065135002, "train_loss_bc": 0.029201211407780647, "train_loss_llm": 0.001004040241241455, "grad_norm": 0.0637817308306694, "global_step": 35550, "epoch": 987, "lr": 0.001} +{"train_loss": 0.02695445343852043, "train_loss_bc": 0.02695445343852043, "train_loss_llm": 0.0, "grad_norm": 0.08644045889377594, "global_step": 35551, "epoch": 987, "lr": 0.001} +{"train_loss": 0.028239566832780838, "train_loss_bc": 0.028239566832780838, "train_loss_llm": 0.0, "grad_norm": 0.10279270261526108, "global_step": 35552, "epoch": 987, "lr": 0.001} +{"train_loss": 0.032651111483573914, "train_loss_bc": 0.032651111483573914, "train_loss_llm": 0.0, "grad_norm": 0.0479201003909111, "global_step": 35553, "epoch": 987, "lr": 0.001} +{"train_loss": 0.02663600817322731, "train_loss_bc": 0.02663600817322731, "train_loss_llm": 0.0, "grad_norm": 0.08366970717906952, "global_step": 35554, "epoch": 987, "lr": 0.001} +{"train_loss": 0.030391979962587357, "train_loss_bc": 0.030391979962587357, "train_loss_llm": 0.0, "grad_norm": 0.11871939897537231, "global_step": 35555, "epoch": 987, "lr": 0.001} +{"train_loss": 0.029549608007073402, "train_loss_bc": 0.029549608007073402, "train_loss_llm": 0.0, "grad_norm": 0.157282754778862, "global_step": 35556, "epoch": 987, "lr": 0.001} +{"train_loss": 0.028549376875162125, "train_loss_bc": 0.028546907007694244, "train_loss_llm": 0.00024694204330444336, "grad_norm": 0.051722537726163864, "global_step": 35557, "epoch": 987, "lr": 0.001} +{"train_loss": 0.02812262624502182, "train_loss_bc": 0.028117742389440536, "train_loss_llm": 0.0004883408546447754, "grad_norm": 0.07406191527843475, "global_step": 35558, "epoch": 987, "lr": 0.001} +{"train_loss": 0.028861073777079582, "train_loss_bc": 0.028861073777079582, "train_loss_llm": 0.0, "grad_norm": 0.0858408659696579, "global_step": 35559, "epoch": 987, "lr": 0.001} +{"train_loss": 0.0369001105427742, "train_loss_bc": 0.0369001105427742, "train_loss_llm": 0.0, "grad_norm": 0.12061215192079544, "global_step": 35560, "epoch": 987, "lr": 0.001} +{"train_loss": 0.028998930007219315, "train_loss_bc": 0.028998930007219315, "train_loss_llm": 0.0, "grad_norm": 0.04281100630760193, "global_step": 35561, "epoch": 987, "lr": 0.001} +{"train_loss": 0.03371109440922737, "train_loss_bc": 0.033709567040205, "train_loss_llm": 0.00015270709991455078, "grad_norm": 0.07659707218408585, "global_step": 35562, "epoch": 987, "lr": 0.001} +{"train_loss": 0.027825456112623215, "train_loss_bc": 0.027825456112623215, "train_loss_llm": 0.0, "grad_norm": 0.10634405165910721, "global_step": 35563, "epoch": 987, "lr": 0.001} +{"train_loss": 0.030434533953666687, "train_loss_bc": 0.030434533953666687, "train_loss_llm": 0.0, "grad_norm": 0.13652893900871277, "global_step": 35564, "epoch": 987, "lr": 0.001} +{"train_loss": 0.02333812415599823, "train_loss_bc": 0.02333812415599823, "train_loss_llm": 0.0, "grad_norm": 0.032859958708286285, "global_step": 35565, "epoch": 987, "lr": 0.001} +{"train_loss": 0.0321064256131649, "train_loss_bc": 0.0321064256131649, "train_loss_llm": 0.0, "grad_norm": 0.06835922598838806, "global_step": 35566, "epoch": 987, "lr": 0.001} +{"train_loss": 0.02864550447298421, "train_loss_bc": 0.021896004676818848, "train_loss_llm": 0.0005504786968231201, "grad_norm": 0.09494806081056595, "global_step": 35567, "epoch": 987, "lr": 0.001, "val_loss": 0.11173036694526672} +{"train_loss": 0.02725129947066307, "train_loss_bc": 0.02725129947066307, "train_loss_llm": 0.0, "grad_norm": 0.11809869110584259, "global_step": 35568, "epoch": 988, "lr": 0.001} +{"train_loss": 0.02711932733654976, "train_loss_bc": 0.02711932733654976, "train_loss_llm": 0.0, "grad_norm": 0.03629318252205849, "global_step": 35569, "epoch": 988, "lr": 0.001} +{"train_loss": 0.02792731113731861, "train_loss_bc": 0.02792731113731861, "train_loss_llm": 0.0, "grad_norm": 0.057306401431560516, "global_step": 35570, "epoch": 988, "lr": 0.001} +{"train_loss": 0.027527248486876488, "train_loss_bc": 0.027527248486876488, "train_loss_llm": 0.0, "grad_norm": 0.07538451999425888, "global_step": 35571, "epoch": 988, "lr": 0.001} +{"train_loss": 0.027355356141924858, "train_loss_bc": 0.027355356141924858, "train_loss_llm": 0.0, "grad_norm": 0.10291308164596558, "global_step": 35572, "epoch": 988, "lr": 0.001} +{"train_loss": 0.034165963530540466, "train_loss_bc": 0.034165963530540466, "train_loss_llm": 0.0, "grad_norm": 0.050684526562690735, "global_step": 35573, "epoch": 988, "lr": 0.001} +{"train_loss": 0.024794980883598328, "train_loss_bc": 0.024794571101665497, "train_loss_llm": 4.106760025024414e-05, "grad_norm": 0.06800729781389236, "global_step": 35574, "epoch": 988, "lr": 0.001} +{"train_loss": 0.02863474190235138, "train_loss_bc": 0.02863474190235138, "train_loss_llm": 0.0, "grad_norm": 0.08726140856742859, "global_step": 35575, "epoch": 988, "lr": 0.001} +{"train_loss": 0.02753824181854725, "train_loss_bc": 0.02753824181854725, "train_loss_llm": 0.0, "grad_norm": 0.11294388025999069, "global_step": 35576, "epoch": 988, "lr": 0.001} +{"train_loss": 0.029328810051083565, "train_loss_bc": 0.029328810051083565, "train_loss_llm": 0.0, "grad_norm": 0.04605371132493019, "global_step": 35577, "epoch": 988, "lr": 0.001} +{"train_loss": 0.037744808942079544, "train_loss_bc": 0.037744808942079544, "train_loss_llm": 0.0, "grad_norm": 0.07252298295497894, "global_step": 35578, "epoch": 988, "lr": 0.001} +{"train_loss": 0.02521364577114582, "train_loss_bc": 0.02521364577114582, "train_loss_llm": 0.0, "grad_norm": 0.10200284421443939, "global_step": 35579, "epoch": 988, "lr": 0.001} +{"train_loss": 0.030910281464457512, "train_loss_bc": 0.030910281464457512, "train_loss_llm": 0.0, "grad_norm": 0.12262611836194992, "global_step": 35580, "epoch": 988, "lr": 0.001} +{"train_loss": 0.02927827648818493, "train_loss_bc": 0.02927827648818493, "train_loss_llm": 0.0, "grad_norm": 0.04077506437897682, "global_step": 35581, "epoch": 988, "lr": 0.001} +{"train_loss": 0.02717650681734085, "train_loss_bc": 0.02717650681734085, "train_loss_llm": 0.0, "grad_norm": 0.05854896456003189, "global_step": 35582, "epoch": 988, "lr": 0.001} +{"train_loss": 0.02813217043876648, "train_loss_bc": 0.02813217043876648, "train_loss_llm": 0.0, "grad_norm": 0.08559481054544449, "global_step": 35583, "epoch": 988, "lr": 0.001} +{"train_loss": 0.030926356092095375, "train_loss_bc": 0.030926356092095375, "train_loss_llm": 0.0, "grad_norm": 0.12951286137104034, "global_step": 35584, "epoch": 988, "lr": 0.001} +{"train_loss": 0.030977586284279823, "train_loss_bc": 0.030977586284279823, "train_loss_llm": 0.0, "grad_norm": 0.038327429443597794, "global_step": 35585, "epoch": 988, "lr": 0.001} +{"train_loss": 0.027407824993133545, "train_loss_bc": 0.02739265188574791, "train_loss_llm": 0.0015173554420471191, "grad_norm": 0.05721541866660118, "global_step": 35586, "epoch": 988, "lr": 0.001} +{"train_loss": 0.036611251533031464, "train_loss_bc": 0.036604978144168854, "train_loss_llm": 0.0006274282932281494, "grad_norm": 0.08673526346683502, "global_step": 35587, "epoch": 988, "lr": 0.001} +{"train_loss": 0.02855098992586136, "train_loss_bc": 0.02853172831237316, "train_loss_llm": 0.0019261837005615234, "grad_norm": 0.10910500586032867, "global_step": 35588, "epoch": 988, "lr": 0.001} +{"train_loss": 0.03177322447299957, "train_loss_bc": 0.03177322447299957, "train_loss_llm": 0.0, "grad_norm": 0.039575349539518356, "global_step": 35589, "epoch": 988, "lr": 0.001} +{"train_loss": 0.024862045422196388, "train_loss_bc": 0.024862045422196388, "train_loss_llm": 0.0, "grad_norm": 0.05193274840712547, "global_step": 35590, "epoch": 988, "lr": 0.001} +{"train_loss": 0.022690407931804657, "train_loss_bc": 0.0226883664727211, "train_loss_llm": 0.0002041161060333252, "grad_norm": 0.06820434331893921, "global_step": 35591, "epoch": 988, "lr": 0.001} +{"train_loss": 0.024653075262904167, "train_loss_bc": 0.024653075262904167, "train_loss_llm": 0.0, "grad_norm": 0.0825524777173996, "global_step": 35592, "epoch": 988, "lr": 0.001} +{"train_loss": 0.025568434968590736, "train_loss_bc": 0.025568434968590736, "train_loss_llm": 0.0, "grad_norm": 0.039312273263931274, "global_step": 35593, "epoch": 988, "lr": 0.001} +{"train_loss": 0.024060944095253944, "train_loss_bc": 0.02405635640025139, "train_loss_llm": 0.00045883655548095703, "grad_norm": 0.06499136984348297, "global_step": 35594, "epoch": 988, "lr": 0.001} +{"train_loss": 0.03121579810976982, "train_loss_bc": 0.03121579810976982, "train_loss_llm": 0.0, "grad_norm": 0.08535780757665634, "global_step": 35595, "epoch": 988, "lr": 0.001} +{"train_loss": 0.02877971902489662, "train_loss_bc": 0.02877971902489662, "train_loss_llm": 0.0, "grad_norm": 0.12341303378343582, "global_step": 35596, "epoch": 988, "lr": 0.001} +{"train_loss": 0.0329989418387413, "train_loss_bc": 0.0329989418387413, "train_loss_llm": 0.0, "grad_norm": 0.04017205536365509, "global_step": 35597, "epoch": 988, "lr": 0.001} +{"train_loss": 0.030498353764414787, "train_loss_bc": 0.030483288690447807, "train_loss_llm": 0.0015065670013427734, "grad_norm": 0.06511259824037552, "global_step": 35598, "epoch": 988, "lr": 0.001} +{"train_loss": 0.029333064332604408, "train_loss_bc": 0.029333064332604408, "train_loss_llm": 0.0, "grad_norm": 0.09810115396976471, "global_step": 35599, "epoch": 988, "lr": 0.001} +{"train_loss": 0.027647672221064568, "train_loss_bc": 0.027647672221064568, "train_loss_llm": 0.0, "grad_norm": 0.12329161912202835, "global_step": 35600, "epoch": 988, "lr": 0.001} +{"train_loss": 0.024864623323082924, "train_loss_bc": 0.024864623323082924, "train_loss_llm": 0.0, "grad_norm": 0.03908722847700119, "global_step": 35601, "epoch": 988, "lr": 0.001} +{"train_loss": 0.026121122762560844, "train_loss_bc": 0.026121122762560844, "train_loss_llm": 0.0, "grad_norm": 0.06509827822446823, "global_step": 35602, "epoch": 988, "lr": 0.001} +{"train_loss": 0.028729143500741985, "train_loss_bc": 0.034608758985996246, "train_loss_llm": 0.0, "grad_norm": 0.09252016246318817, "global_step": 35603, "epoch": 988, "lr": 0.001, "val_loss": 0.11080199480056763} +{"train_loss": 0.031437549740076065, "train_loss_bc": 0.031437549740076065, "train_loss_llm": 0.0, "grad_norm": 0.12446451932191849, "global_step": 35604, "epoch": 989, "lr": 0.001} +{"train_loss": 0.026068255305290222, "train_loss_bc": 0.026059113442897797, "train_loss_llm": 0.0009141862392425537, "grad_norm": 0.036140117794275284, "global_step": 35605, "epoch": 989, "lr": 0.001} +{"train_loss": 0.030898921191692352, "train_loss_bc": 0.030891288071870804, "train_loss_llm": 0.000763237476348877, "grad_norm": 0.06814991682767868, "global_step": 35606, "epoch": 989, "lr": 0.001} +{"train_loss": 0.03034248761832714, "train_loss_bc": 0.03034248761832714, "train_loss_llm": 0.0, "grad_norm": 0.09550022333860397, "global_step": 35607, "epoch": 989, "lr": 0.001} +{"train_loss": 0.026220660656690598, "train_loss_bc": 0.026220660656690598, "train_loss_llm": 0.0, "grad_norm": 0.11029580235481262, "global_step": 35608, "epoch": 989, "lr": 0.001} +{"train_loss": 0.0318974032998085, "train_loss_bc": 0.0318974032998085, "train_loss_llm": 0.0, "grad_norm": 0.05128661170601845, "global_step": 35609, "epoch": 989, "lr": 0.001} +{"train_loss": 0.028140680864453316, "train_loss_bc": 0.028131743893027306, "train_loss_llm": 0.0008936226367950439, "grad_norm": 0.0677114799618721, "global_step": 35610, "epoch": 989, "lr": 0.001} +{"train_loss": 0.024425631389021873, "train_loss_bc": 0.024425631389021873, "train_loss_llm": 0.0, "grad_norm": 0.08450201898813248, "global_step": 35611, "epoch": 989, "lr": 0.001} +{"train_loss": 0.02720104157924652, "train_loss_bc": 0.027200112119317055, "train_loss_llm": 9.301304817199707e-05, "grad_norm": 0.11119088530540466, "global_step": 35612, "epoch": 989, "lr": 0.001} +{"train_loss": 0.029880119487643242, "train_loss_bc": 0.029858291149139404, "train_loss_llm": 0.002182871103286743, "grad_norm": 0.04520399123430252, "global_step": 35613, "epoch": 989, "lr": 0.001} +{"train_loss": 0.029180938377976418, "train_loss_bc": 0.029180938377976418, "train_loss_llm": 0.0, "grad_norm": 0.06561128795146942, "global_step": 35614, "epoch": 989, "lr": 0.001} +{"train_loss": 0.032700855284929276, "train_loss_bc": 0.032700855284929276, "train_loss_llm": 0.0, "grad_norm": 0.087814562022686, "global_step": 35615, "epoch": 989, "lr": 0.001} +{"train_loss": 0.027843432500958443, "train_loss_bc": 0.027843432500958443, "train_loss_llm": 0.0, "grad_norm": 0.1260288655757904, "global_step": 35616, "epoch": 989, "lr": 0.001} +{"train_loss": 0.028028370812535286, "train_loss_bc": 0.028028370812535286, "train_loss_llm": 0.0, "grad_norm": 0.043393075466156006, "global_step": 35617, "epoch": 989, "lr": 0.001} +{"train_loss": 0.03089742176234722, "train_loss_bc": 0.03088427148759365, "train_loss_llm": 0.0013150572776794434, "grad_norm": 0.07082494348287582, "global_step": 35618, "epoch": 989, "lr": 0.001} +{"train_loss": 0.026357140392065048, "train_loss_bc": 0.026354119181632996, "train_loss_llm": 0.000302046537399292, "grad_norm": 0.0863363966345787, "global_step": 35619, "epoch": 989, "lr": 0.001} +{"train_loss": 0.031953129917383194, "train_loss_bc": 0.031952954828739166, "train_loss_llm": 1.7344951629638672e-05, "grad_norm": 0.1208612471818924, "global_step": 35620, "epoch": 989, "lr": 0.001} +{"train_loss": 0.02389121614396572, "train_loss_bc": 0.02389121614396572, "train_loss_llm": 0.0, "grad_norm": 0.04038868099451065, "global_step": 35621, "epoch": 989, "lr": 0.001} +{"train_loss": 0.029186325147747993, "train_loss_bc": 0.029186325147747993, "train_loss_llm": 0.0, "grad_norm": 0.06748282164335251, "global_step": 35622, "epoch": 989, "lr": 0.001} +{"train_loss": 0.029756618663668633, "train_loss_bc": 0.029756618663668633, "train_loss_llm": 0.0, "grad_norm": 0.08686277270317078, "global_step": 35623, "epoch": 989, "lr": 0.001} +{"train_loss": 0.028178421780467033, "train_loss_bc": 0.028178071603178978, "train_loss_llm": 3.510713577270508e-05, "grad_norm": 0.11773131042718887, "global_step": 35624, "epoch": 989, "lr": 0.001} +{"train_loss": 0.02835046499967575, "train_loss_bc": 0.028341462835669518, "train_loss_llm": 0.0009002089500427246, "grad_norm": 0.04298925772309303, "global_step": 35625, "epoch": 989, "lr": 0.001} +{"train_loss": 0.031377725303173065, "train_loss_bc": 0.031377725303173065, "train_loss_llm": 0.0, "grad_norm": 0.07065985351800919, "global_step": 35626, "epoch": 989, "lr": 0.001} +{"train_loss": 0.02974316105246544, "train_loss_bc": 0.02974316105246544, "train_loss_llm": 0.0, "grad_norm": 0.09958451241254807, "global_step": 35627, "epoch": 989, "lr": 0.001} +{"train_loss": 0.035301972180604935, "train_loss_bc": 0.035301972180604935, "train_loss_llm": 0.0, "grad_norm": 0.12059002369642258, "global_step": 35628, "epoch": 989, "lr": 0.001} +{"train_loss": 0.03131173178553581, "train_loss_bc": 0.03131173178553581, "train_loss_llm": 0.0, "grad_norm": 0.03127795457839966, "global_step": 35629, "epoch": 989, "lr": 0.001} +{"train_loss": 0.0245280209928751, "train_loss_bc": 0.02452644519507885, "train_loss_llm": 0.0001576244831085205, "grad_norm": 0.05452350154519081, "global_step": 35630, "epoch": 989, "lr": 0.001} +{"train_loss": 0.02817525900900364, "train_loss_bc": 0.02817525900900364, "train_loss_llm": 0.0, "grad_norm": 0.07182298600673676, "global_step": 35631, "epoch": 989, "lr": 0.001} +{"train_loss": 0.02512652799487114, "train_loss_bc": 0.02512652799487114, "train_loss_llm": 0.0, "grad_norm": 0.09462852776050568, "global_step": 35632, "epoch": 989, "lr": 0.001} +{"train_loss": 0.027983736246824265, "train_loss_bc": 0.027983736246824265, "train_loss_llm": 0.0, "grad_norm": 0.03561801835894585, "global_step": 35633, "epoch": 989, "lr": 0.001} +{"train_loss": 0.028140557929873466, "train_loss_bc": 0.028140557929873466, "train_loss_llm": 0.0, "grad_norm": 0.06352273374795914, "global_step": 35634, "epoch": 989, "lr": 0.001} +{"train_loss": 0.023621195927262306, "train_loss_bc": 0.023621195927262306, "train_loss_llm": 0.0, "grad_norm": 0.08411503583192825, "global_step": 35635, "epoch": 989, "lr": 0.001} +{"train_loss": 0.02790513075888157, "train_loss_bc": 0.02790513075888157, "train_loss_llm": 0.0, "grad_norm": 0.1263580620288849, "global_step": 35636, "epoch": 989, "lr": 0.001} +{"train_loss": 0.02817448601126671, "train_loss_bc": 0.02817448601126671, "train_loss_llm": 0.0, "grad_norm": 0.04509279504418373, "global_step": 35637, "epoch": 989, "lr": 0.001} +{"train_loss": 0.029259245842695236, "train_loss_bc": 0.029259245842695236, "train_loss_llm": 0.0, "grad_norm": 0.07968293875455856, "global_step": 35638, "epoch": 989, "lr": 0.001} +{"train_loss": 0.02853908312196533, "train_loss_bc": 0.02392115443944931, "train_loss_llm": 0.0, "grad_norm": 0.09382403641939163, "global_step": 35639, "epoch": 989, "lr": 0.001, "val_loss": 0.11290611326694489} +{"train_loss": 0.028139889240264893, "train_loss_bc": 0.028139889240264893, "train_loss_llm": 0.0, "grad_norm": 0.12360253185033798, "global_step": 35640, "epoch": 990, "lr": 0.001} +{"train_loss": 0.028207609429955482, "train_loss_bc": 0.028207609429955482, "train_loss_llm": 0.0, "grad_norm": 0.041128817945718765, "global_step": 35641, "epoch": 990, "lr": 0.001} +{"train_loss": 0.03168518468737602, "train_loss_bc": 0.03168518468737602, "train_loss_llm": 0.0, "grad_norm": 0.06059851497411728, "global_step": 35642, "epoch": 990, "lr": 0.001} +{"train_loss": 0.024659622460603714, "train_loss_bc": 0.024659622460603714, "train_loss_llm": 0.0, "grad_norm": 0.07726714760065079, "global_step": 35643, "epoch": 990, "lr": 0.001} +{"train_loss": 0.03513507544994354, "train_loss_bc": 0.03511364758014679, "train_loss_llm": 0.002142786979675293, "grad_norm": 0.10060323774814606, "global_step": 35644, "epoch": 990, "lr": 0.001} +{"train_loss": 0.023097101598978043, "train_loss_bc": 0.023097101598978043, "train_loss_llm": 0.0, "grad_norm": 0.026301829144358635, "global_step": 35645, "epoch": 990, "lr": 0.001} +{"train_loss": 0.026023956015706062, "train_loss_bc": 0.026023956015706062, "train_loss_llm": 0.0, "grad_norm": 0.049758732318878174, "global_step": 35646, "epoch": 990, "lr": 0.001} +{"train_loss": 0.034151315689086914, "train_loss_bc": 0.034151315689086914, "train_loss_llm": 0.0, "grad_norm": 0.08311043679714203, "global_step": 35647, "epoch": 990, "lr": 0.001} +{"train_loss": 0.026589518412947655, "train_loss_bc": 0.026584157720208168, "train_loss_llm": 0.000536113977432251, "grad_norm": 0.09653089940547943, "global_step": 35648, "epoch": 990, "lr": 0.001} +{"train_loss": 0.027659321203827858, "train_loss_bc": 0.027659321203827858, "train_loss_llm": 0.0, "grad_norm": 0.0427730567753315, "global_step": 35649, "epoch": 990, "lr": 0.001} +{"train_loss": 0.0346406027674675, "train_loss_bc": 0.0346406027674675, "train_loss_llm": 0.0, "grad_norm": 0.07805946469306946, "global_step": 35650, "epoch": 990, "lr": 0.001} +{"train_loss": 0.03165333345532417, "train_loss_bc": 0.03165333345532417, "train_loss_llm": 0.0, "grad_norm": 0.09705550223588943, "global_step": 35651, "epoch": 990, "lr": 0.001} +{"train_loss": 0.027954600751399994, "train_loss_bc": 0.027949556708335876, "train_loss_llm": 0.0005044639110565186, "grad_norm": 0.11974027007818222, "global_step": 35652, "epoch": 990, "lr": 0.001} +{"train_loss": 0.02677874267101288, "train_loss_bc": 0.02677874267101288, "train_loss_llm": 0.0, "grad_norm": 0.054545022547245026, "global_step": 35653, "epoch": 990, "lr": 0.001} +{"train_loss": 0.029413597658276558, "train_loss_bc": 0.029401490464806557, "train_loss_llm": 0.0012107789516448975, "grad_norm": 0.07623817771673203, "global_step": 35654, "epoch": 990, "lr": 0.001} +{"train_loss": 0.027071256190538406, "train_loss_bc": 0.027071256190538406, "train_loss_llm": 0.0, "grad_norm": 0.10783898830413818, "global_step": 35655, "epoch": 990, "lr": 0.001} +{"train_loss": 0.0286659374833107, "train_loss_bc": 0.0286659374833107, "train_loss_llm": 0.0, "grad_norm": 0.14108099043369293, "global_step": 35656, "epoch": 990, "lr": 0.001} +{"train_loss": 0.03078954666852951, "train_loss_bc": 0.03078954666852951, "train_loss_llm": 0.0, "grad_norm": 0.04012589529156685, "global_step": 35657, "epoch": 990, "lr": 0.001} +{"train_loss": 0.023555712774395943, "train_loss_bc": 0.023547906428575516, "train_loss_llm": 0.0007806718349456787, "grad_norm": 0.05966660752892494, "global_step": 35658, "epoch": 990, "lr": 0.001} +{"train_loss": 0.026365159079432487, "train_loss_bc": 0.026365159079432487, "train_loss_llm": 0.0, "grad_norm": 0.09883303195238113, "global_step": 35659, "epoch": 990, "lr": 0.001} +{"train_loss": 0.029620438814163208, "train_loss_bc": 0.029620438814163208, "train_loss_llm": 0.0, "grad_norm": 0.12697212398052216, "global_step": 35660, "epoch": 990, "lr": 0.001} +{"train_loss": 0.02936462126672268, "train_loss_bc": 0.0293467715382576, "train_loss_llm": 0.001784980297088623, "grad_norm": 0.03741709887981415, "global_step": 35661, "epoch": 990, "lr": 0.001} +{"train_loss": 0.024999679997563362, "train_loss_bc": 0.024999679997563362, "train_loss_llm": 0.0, "grad_norm": 0.05403934419155121, "global_step": 35662, "epoch": 990, "lr": 0.001} +{"train_loss": 0.025371234863996506, "train_loss_bc": 0.025371234863996506, "train_loss_llm": 0.0, "grad_norm": 0.07501521706581116, "global_step": 35663, "epoch": 990, "lr": 0.001} +{"train_loss": 0.02872435376048088, "train_loss_bc": 0.02872435376048088, "train_loss_llm": 0.0, "grad_norm": 0.09141801297664642, "global_step": 35664, "epoch": 990, "lr": 0.001} +{"train_loss": 0.033294372260570526, "train_loss_bc": 0.033294372260570526, "train_loss_llm": 0.0, "grad_norm": 0.03999604284763336, "global_step": 35665, "epoch": 990, "lr": 0.001} +{"train_loss": 0.03095778450369835, "train_loss_bc": 0.03095778450369835, "train_loss_llm": 0.0, "grad_norm": 0.06550296396017075, "global_step": 35666, "epoch": 990, "lr": 0.001} +{"train_loss": 0.025412647053599358, "train_loss_bc": 0.025412647053599358, "train_loss_llm": 0.0, "grad_norm": 0.08888500928878784, "global_step": 35667, "epoch": 990, "lr": 0.001} +{"train_loss": 0.02785271592438221, "train_loss_bc": 0.02785271592438221, "train_loss_llm": 0.0, "grad_norm": 0.1148170679807663, "global_step": 35668, "epoch": 990, "lr": 0.001} +{"train_loss": 0.026953844353556633, "train_loss_bc": 0.026953844353556633, "train_loss_llm": 0.0, "grad_norm": 0.03681343421339989, "global_step": 35669, "epoch": 990, "lr": 0.001} +{"train_loss": 0.025701984763145447, "train_loss_bc": 0.02568531036376953, "train_loss_llm": 0.0016674697399139404, "grad_norm": 0.059033483266830444, "global_step": 35670, "epoch": 990, "lr": 0.001} +{"train_loss": 0.0343623049557209, "train_loss_bc": 0.0343623049557209, "train_loss_llm": 0.0, "grad_norm": 0.08255670219659805, "global_step": 35671, "epoch": 990, "lr": 0.001} +{"train_loss": 0.02568729594349861, "train_loss_bc": 0.02568729594349861, "train_loss_llm": 0.0, "grad_norm": 0.10350176692008972, "global_step": 35672, "epoch": 990, "lr": 0.001} +{"train_loss": 0.026057101786136627, "train_loss_bc": 0.026057101786136627, "train_loss_llm": 0.0, "grad_norm": 0.042276859283447266, "global_step": 35673, "epoch": 990, "lr": 0.001} +{"train_loss": 0.032678473740816116, "train_loss_bc": 0.032678473740816116, "train_loss_llm": 0.0, "grad_norm": 0.07538194209337234, "global_step": 35674, "epoch": 990, "lr": 0.001} +{"train_loss": 0.028651789555119142, "train_loss_bc": 0.03218848630785942, "train_loss_llm": 0.0, "grad_norm": 0.0917600616812706, "global_step": 35675, "epoch": 990, "lr": 0.001, "train/cumulative_reward": 34.3693644397549, "train/mean_score": 7.35386375483358, "train/success_rate": 0.2, "test/cumulative_reward": 30.477877752080012, "test/mean_score": 8.16090167992796, "test/success_rate": 0.22, "val_loss": 0.11161056160926819, "train_action_mse_error": 0.032170869410037994} +{"train_loss": 0.030590973794460297, "train_loss_bc": 0.030590973794460297, "train_loss_llm": 0.0, "grad_norm": 0.12981148064136505, "global_step": 35676, "epoch": 991, "lr": 0.001} +{"train_loss": 0.02428942732512951, "train_loss_bc": 0.02428942732512951, "train_loss_llm": 0.0, "grad_norm": 0.032733552157878876, "global_step": 35677, "epoch": 991, "lr": 0.001} +{"train_loss": 0.032367199659347534, "train_loss_bc": 0.03236255794763565, "train_loss_llm": 0.0004641413688659668, "grad_norm": 0.052462752908468246, "global_step": 35678, "epoch": 991, "lr": 0.001} +{"train_loss": 0.025259502232074738, "train_loss_bc": 0.025238044559955597, "train_loss_llm": 0.0021456778049468994, "grad_norm": 0.07157614827156067, "global_step": 35679, "epoch": 991, "lr": 0.001} +{"train_loss": 0.027141528204083443, "train_loss_bc": 0.0271272212266922, "train_loss_llm": 0.0014306902885437012, "grad_norm": 0.10199771821498871, "global_step": 35680, "epoch": 991, "lr": 0.001} +{"train_loss": 0.029633954167366028, "train_loss_bc": 0.029633954167366028, "train_loss_llm": 0.0, "grad_norm": 0.04750582203269005, "global_step": 35681, "epoch": 991, "lr": 0.001} +{"train_loss": 0.031039215624332428, "train_loss_bc": 0.031039215624332428, "train_loss_llm": 0.0, "grad_norm": 0.0711035206913948, "global_step": 35682, "epoch": 991, "lr": 0.001} +{"train_loss": 0.02411593683063984, "train_loss_bc": 0.02411503717303276, "train_loss_llm": 8.991360664367676e-05, "grad_norm": 0.09354924410581589, "global_step": 35683, "epoch": 991, "lr": 0.001} +{"train_loss": 0.02875259518623352, "train_loss_bc": 0.02875259518623352, "train_loss_llm": 0.0, "grad_norm": 0.1263282597064972, "global_step": 35684, "epoch": 991, "lr": 0.001} +{"train_loss": 0.033025555312633514, "train_loss_bc": 0.0330144464969635, "train_loss_llm": 0.0011109709739685059, "grad_norm": 0.04904432222247124, "global_step": 35685, "epoch": 991, "lr": 0.001} +{"train_loss": 0.025630144402384758, "train_loss_bc": 0.0256139375269413, "train_loss_llm": 0.0016207396984100342, "grad_norm": 0.06869950145483017, "global_step": 35686, "epoch": 991, "lr": 0.001} +{"train_loss": 0.027792368084192276, "train_loss_bc": 0.027792368084192276, "train_loss_llm": 0.0, "grad_norm": 0.0878659263253212, "global_step": 35687, "epoch": 991, "lr": 0.001} +{"train_loss": 0.03189690783619881, "train_loss_bc": 0.03189690783619881, "train_loss_llm": 0.0, "grad_norm": 0.11545459181070328, "global_step": 35688, "epoch": 991, "lr": 0.001} +{"train_loss": 0.03109288029372692, "train_loss_bc": 0.03109288029372692, "train_loss_llm": 0.0, "grad_norm": 0.04739456623792648, "global_step": 35689, "epoch": 991, "lr": 0.001} +{"train_loss": 0.03146833926439285, "train_loss_bc": 0.03146833926439285, "train_loss_llm": 0.0, "grad_norm": 0.07413668185472488, "global_step": 35690, "epoch": 991, "lr": 0.001} +{"train_loss": 0.026367945596575737, "train_loss_bc": 0.026367945596575737, "train_loss_llm": 0.0, "grad_norm": 0.10229624062776566, "global_step": 35691, "epoch": 991, "lr": 0.001} +{"train_loss": 0.027660850435495377, "train_loss_bc": 0.027660850435495377, "train_loss_llm": 0.0, "grad_norm": 0.12759116291999817, "global_step": 35692, "epoch": 991, "lr": 0.001} +{"train_loss": 0.02944585494697094, "train_loss_bc": 0.02944585494697094, "train_loss_llm": 0.0, "grad_norm": 0.0411917120218277, "global_step": 35693, "epoch": 991, "lr": 0.001} +{"train_loss": 0.03513345494866371, "train_loss_bc": 0.03513345494866371, "train_loss_llm": 0.0, "grad_norm": 0.07483376562595367, "global_step": 35694, "epoch": 991, "lr": 0.001} +{"train_loss": 0.034510817378759384, "train_loss_bc": 0.03450736030936241, "train_loss_llm": 0.0003457069396972656, "grad_norm": 0.11890134960412979, "global_step": 35695, "epoch": 991, "lr": 0.001} +{"train_loss": 0.02551230415701866, "train_loss_bc": 0.02551230415701866, "train_loss_llm": 0.0, "grad_norm": 0.14517459273338318, "global_step": 35696, "epoch": 991, "lr": 0.001} +{"train_loss": 0.025645235553383827, "train_loss_bc": 0.025645235553383827, "train_loss_llm": 0.0, "grad_norm": 0.03897911682724953, "global_step": 35697, "epoch": 991, "lr": 0.001} +{"train_loss": 0.02196446806192398, "train_loss_bc": 0.02196446806192398, "train_loss_llm": 0.0, "grad_norm": 0.06128864362835884, "global_step": 35698, "epoch": 991, "lr": 0.001} +{"train_loss": 0.02676762267947197, "train_loss_bc": 0.026761500164866447, "train_loss_llm": 0.0006123185157775879, "grad_norm": 0.09092637151479721, "global_step": 35699, "epoch": 991, "lr": 0.001} +{"train_loss": 0.03193189576268196, "train_loss_bc": 0.03190823271870613, "train_loss_llm": 0.0023664534091949463, "grad_norm": 0.11885145306587219, "global_step": 35700, "epoch": 991, "lr": 0.001} +{"train_loss": 0.030116043984889984, "train_loss_bc": 0.030104681849479675, "train_loss_llm": 0.0011363029479980469, "grad_norm": 0.048445526510477066, "global_step": 35701, "epoch": 991, "lr": 0.001} +{"train_loss": 0.029912564903497696, "train_loss_bc": 0.029901370406150818, "train_loss_llm": 0.001119375228881836, "grad_norm": 0.0891539603471756, "global_step": 35702, "epoch": 991, "lr": 0.001} +{"train_loss": 0.02748289704322815, "train_loss_bc": 0.02748289704322815, "train_loss_llm": 0.0, "grad_norm": 0.11908122897148132, "global_step": 35703, "epoch": 991, "lr": 0.001} +{"train_loss": 0.027661046013236046, "train_loss_bc": 0.027661046013236046, "train_loss_llm": 0.0, "grad_norm": 0.1546563357114792, "global_step": 35704, "epoch": 991, "lr": 0.001} +{"train_loss": 0.024246621876955032, "train_loss_bc": 0.02424028515815735, "train_loss_llm": 0.0006337165832519531, "grad_norm": 0.032996613532304764, "global_step": 35705, "epoch": 991, "lr": 0.001} +{"train_loss": 0.02811506576836109, "train_loss_bc": 0.02811506576836109, "train_loss_llm": 0.0, "grad_norm": 0.06706839054822922, "global_step": 35706, "epoch": 991, "lr": 0.001} +{"train_loss": 0.034122005105018616, "train_loss_bc": 0.03410869836807251, "train_loss_llm": 0.0013308525085449219, "grad_norm": 0.0989825576543808, "global_step": 35707, "epoch": 991, "lr": 0.001} +{"train_loss": 0.028741130605340004, "train_loss_bc": 0.028741130605340004, "train_loss_llm": 0.0, "grad_norm": 0.1208747923374176, "global_step": 35708, "epoch": 991, "lr": 0.001} +{"train_loss": 0.03084855154156685, "train_loss_bc": 0.030840128660202026, "train_loss_llm": 0.0008423030376434326, "grad_norm": 0.046834997832775116, "global_step": 35709, "epoch": 991, "lr": 0.001} +{"train_loss": 0.029654717072844505, "train_loss_bc": 0.029654717072844505, "train_loss_llm": 0.0, "grad_norm": 0.08488117158412933, "global_step": 35710, "epoch": 991, "lr": 0.001} +{"train_loss": 0.028830890078097582, "train_loss_bc": 0.02797442115843296, "train_loss_llm": 0.0, "grad_norm": 0.11464790254831314, "global_step": 35711, "epoch": 991, "lr": 0.001, "val_loss": 0.11134542524814606} +{"train_loss": 0.035418134182691574, "train_loss_bc": 0.03541722521185875, "train_loss_llm": 9.077787399291992e-05, "grad_norm": 0.14427824318408966, "global_step": 35712, "epoch": 992, "lr": 0.001} +{"train_loss": 0.028998345136642456, "train_loss_bc": 0.028998345136642456, "train_loss_llm": 0.0, "grad_norm": 0.05161086469888687, "global_step": 35713, "epoch": 992, "lr": 0.001} +{"train_loss": 0.025666510686278343, "train_loss_bc": 0.025666510686278343, "train_loss_llm": 0.0, "grad_norm": 0.08066015690565109, "global_step": 35714, "epoch": 992, "lr": 0.001} +{"train_loss": 0.02389703318476677, "train_loss_bc": 0.02389703318476677, "train_loss_llm": 0.0, "grad_norm": 0.11073268949985504, "global_step": 35715, "epoch": 992, "lr": 0.001} +{"train_loss": 0.0278437789529562, "train_loss_bc": 0.0278437789529562, "train_loss_llm": 0.0, "grad_norm": 0.1462450623512268, "global_step": 35716, "epoch": 992, "lr": 0.001} +{"train_loss": 0.03339378163218498, "train_loss_bc": 0.03339378163218498, "train_loss_llm": 0.0, "grad_norm": 0.052751459181308746, "global_step": 35717, "epoch": 992, "lr": 0.001} +{"train_loss": 0.032482605427503586, "train_loss_bc": 0.032482605427503586, "train_loss_llm": 0.0, "grad_norm": 0.09673260897397995, "global_step": 35718, "epoch": 992, "lr": 0.001} +{"train_loss": 0.026959281414747238, "train_loss_bc": 0.026959281414747238, "train_loss_llm": 0.0, "grad_norm": 0.12038025259971619, "global_step": 35719, "epoch": 992, "lr": 0.001} +{"train_loss": 0.03048192337155342, "train_loss_bc": 0.03048192337155342, "train_loss_llm": 0.0, "grad_norm": 0.14646729826927185, "global_step": 35720, "epoch": 992, "lr": 0.001} +{"train_loss": 0.02782801166176796, "train_loss_bc": 0.02782801166176796, "train_loss_llm": 0.0, "grad_norm": 0.03547292575240135, "global_step": 35721, "epoch": 992, "lr": 0.001} +{"train_loss": 0.024190271273255348, "train_loss_bc": 0.024190271273255348, "train_loss_llm": 0.0, "grad_norm": 0.061518993228673935, "global_step": 35722, "epoch": 992, "lr": 0.001} +{"train_loss": 0.026065552607178688, "train_loss_bc": 0.02606271393597126, "train_loss_llm": 0.00028383731842041016, "grad_norm": 0.07852496951818466, "global_step": 35723, "epoch": 992, "lr": 0.001} +{"train_loss": 0.02920996956527233, "train_loss_bc": 0.02920996956527233, "train_loss_llm": 0.0, "grad_norm": 0.10484641045331955, "global_step": 35724, "epoch": 992, "lr": 0.001} +{"train_loss": 0.021972861140966415, "train_loss_bc": 0.021972861140966415, "train_loss_llm": 0.0, "grad_norm": 0.029657751321792603, "global_step": 35725, "epoch": 992, "lr": 0.001} +{"train_loss": 0.0255440641194582, "train_loss_bc": 0.0255440641194582, "train_loss_llm": 0.0, "grad_norm": 0.05003657564520836, "global_step": 35726, "epoch": 992, "lr": 0.001} +{"train_loss": 0.029777558520436287, "train_loss_bc": 0.0297686904668808, "train_loss_llm": 0.0008867383003234863, "grad_norm": 0.07503173500299454, "global_step": 35727, "epoch": 992, "lr": 0.001} +{"train_loss": 0.027402043342590332, "train_loss_bc": 0.027402043342590332, "train_loss_llm": 0.0, "grad_norm": 0.09665097296237946, "global_step": 35728, "epoch": 992, "lr": 0.001} +{"train_loss": 0.027449052780866623, "train_loss_bc": 0.027435416355729103, "train_loss_llm": 0.0013636946678161621, "grad_norm": 0.04739239439368248, "global_step": 35729, "epoch": 992, "lr": 0.001} +{"train_loss": 0.02958199381828308, "train_loss_bc": 0.02958199381828308, "train_loss_llm": 0.0, "grad_norm": 0.0843413844704628, "global_step": 35730, "epoch": 992, "lr": 0.001} +{"train_loss": 0.027349036186933517, "train_loss_bc": 0.027311643585562706, "train_loss_llm": 0.0037392377853393555, "grad_norm": 0.11429891735315323, "global_step": 35731, "epoch": 992, "lr": 0.001} +{"train_loss": 0.02940392680466175, "train_loss_bc": 0.02940392680466175, "train_loss_llm": 0.0, "grad_norm": 0.13512542843818665, "global_step": 35732, "epoch": 992, "lr": 0.001} +{"train_loss": 0.03290680795907974, "train_loss_bc": 0.03290680795907974, "train_loss_llm": 0.0, "grad_norm": 0.0726267620921135, "global_step": 35733, "epoch": 992, "lr": 0.001} +{"train_loss": 0.0288484375923872, "train_loss_bc": 0.02883261814713478, "train_loss_llm": 0.0015819370746612549, "grad_norm": 0.09196554869413376, "global_step": 35734, "epoch": 992, "lr": 0.001} +{"train_loss": 0.033713798969984055, "train_loss_bc": 0.033713798969984055, "train_loss_llm": 0.0, "grad_norm": 0.14482226967811584, "global_step": 35735, "epoch": 992, "lr": 0.001} +{"train_loss": 0.025291888043284416, "train_loss_bc": 0.025291888043284416, "train_loss_llm": 0.0, "grad_norm": 0.17953376471996307, "global_step": 35736, "epoch": 992, "lr": 0.001} +{"train_loss": 0.028890762478113174, "train_loss_bc": 0.02888776920735836, "train_loss_llm": 0.0002993941307067871, "grad_norm": 0.03181910887360573, "global_step": 35737, "epoch": 992, "lr": 0.001} +{"train_loss": 0.025457559153437614, "train_loss_bc": 0.025457559153437614, "train_loss_llm": 0.0, "grad_norm": 0.05275740846991539, "global_step": 35738, "epoch": 992, "lr": 0.001} +{"train_loss": 0.03238663449883461, "train_loss_bc": 0.03238663449883461, "train_loss_llm": 0.0, "grad_norm": 0.07635948061943054, "global_step": 35739, "epoch": 992, "lr": 0.001} +{"train_loss": 0.031889658421278, "train_loss_bc": 0.031889658421278, "train_loss_llm": 0.0, "grad_norm": 0.10402631759643555, "global_step": 35740, "epoch": 992, "lr": 0.001} +{"train_loss": 0.030776794999837875, "train_loss_bc": 0.030776794999837875, "train_loss_llm": 0.0, "grad_norm": 0.034948837012052536, "global_step": 35741, "epoch": 992, "lr": 0.001} +{"train_loss": 0.02727251872420311, "train_loss_bc": 0.02726341038942337, "train_loss_llm": 0.0009109079837799072, "grad_norm": 0.0542435459792614, "global_step": 35742, "epoch": 992, "lr": 0.001} +{"train_loss": 0.02399861253798008, "train_loss_bc": 0.02399861253798008, "train_loss_llm": 0.0, "grad_norm": 0.07470640540122986, "global_step": 35743, "epoch": 992, "lr": 0.001} +{"train_loss": 0.0299164280295372, "train_loss_bc": 0.0299164280295372, "train_loss_llm": 0.0, "grad_norm": 0.09016447514295578, "global_step": 35744, "epoch": 992, "lr": 0.001} +{"train_loss": 0.03341208025813103, "train_loss_bc": 0.03341208025813103, "train_loss_llm": 0.0, "grad_norm": 0.04769815132021904, "global_step": 35745, "epoch": 992, "lr": 0.001} +{"train_loss": 0.029059631749987602, "train_loss_bc": 0.029033638536930084, "train_loss_llm": 0.002599388360977173, "grad_norm": 0.07033038884401321, "global_step": 35746, "epoch": 992, "lr": 0.001} +{"train_loss": 0.028784122990651265, "train_loss_bc": 0.031491078436374664, "train_loss_llm": 0.0, "grad_norm": 0.08282747119665146, "global_step": 35747, "epoch": 992, "lr": 0.001, "val_loss": 0.11106928437948227} +{"train_loss": 0.030291955918073654, "train_loss_bc": 0.030291955918073654, "train_loss_llm": 0.0, "grad_norm": 0.10178115218877792, "global_step": 35748, "epoch": 993, "lr": 0.001} +{"train_loss": 0.025377709418535233, "train_loss_bc": 0.025377709418535233, "train_loss_llm": 0.0, "grad_norm": 0.03703425079584122, "global_step": 35749, "epoch": 993, "lr": 0.001} +{"train_loss": 0.025412343442440033, "train_loss_bc": 0.025412343442440033, "train_loss_llm": 0.0, "grad_norm": 0.06026780605316162, "global_step": 35750, "epoch": 993, "lr": 0.001} +{"train_loss": 0.0277598537504673, "train_loss_bc": 0.027751538902521133, "train_loss_llm": 0.0008315443992614746, "grad_norm": 0.09154181182384491, "global_step": 35751, "epoch": 993, "lr": 0.001} +{"train_loss": 0.03275865316390991, "train_loss_bc": 0.03273893520236015, "train_loss_llm": 0.001971900463104248, "grad_norm": 0.12943629920482635, "global_step": 35752, "epoch": 993, "lr": 0.001} +{"train_loss": 0.0291148591786623, "train_loss_bc": 0.0291148591786623, "train_loss_llm": 0.0, "grad_norm": 0.03528042137622833, "global_step": 35753, "epoch": 993, "lr": 0.001} +{"train_loss": 0.028863290324807167, "train_loss_bc": 0.028863290324807167, "train_loss_llm": 0.0, "grad_norm": 0.07116962969303131, "global_step": 35754, "epoch": 993, "lr": 0.001} +{"train_loss": 0.025288475677371025, "train_loss_bc": 0.025280069559812546, "train_loss_llm": 0.000840604305267334, "grad_norm": 0.0967140942811966, "global_step": 35755, "epoch": 993, "lr": 0.001} +{"train_loss": 0.028687719255685806, "train_loss_bc": 0.028687719255685806, "train_loss_llm": 0.0, "grad_norm": 0.13221977651119232, "global_step": 35756, "epoch": 993, "lr": 0.001} +{"train_loss": 0.022888189181685448, "train_loss_bc": 0.022886531427502632, "train_loss_llm": 0.00016582012176513672, "grad_norm": 0.027950728312134743, "global_step": 35757, "epoch": 993, "lr": 0.001} +{"train_loss": 0.03132658079266548, "train_loss_bc": 0.03132658079266548, "train_loss_llm": 0.0, "grad_norm": 0.06340868026018143, "global_step": 35758, "epoch": 993, "lr": 0.001} +{"train_loss": 0.034575916826725006, "train_loss_bc": 0.034575916826725006, "train_loss_llm": 0.0, "grad_norm": 0.08606076240539551, "global_step": 35759, "epoch": 993, "lr": 0.001} +{"train_loss": 0.024191085249185562, "train_loss_bc": 0.024191085249185562, "train_loss_llm": 0.0, "grad_norm": 0.09829950332641602, "global_step": 35760, "epoch": 993, "lr": 0.001} +{"train_loss": 0.028041716665029526, "train_loss_bc": 0.028041716665029526, "train_loss_llm": 0.0, "grad_norm": 0.0327121801674366, "global_step": 35761, "epoch": 993, "lr": 0.001} +{"train_loss": 0.03129846602678299, "train_loss_bc": 0.03129846602678299, "train_loss_llm": 0.0, "grad_norm": 0.04738001525402069, "global_step": 35762, "epoch": 993, "lr": 0.001} +{"train_loss": 0.0285454373806715, "train_loss_bc": 0.0285454373806715, "train_loss_llm": 0.0, "grad_norm": 0.07051671296358109, "global_step": 35763, "epoch": 993, "lr": 0.001} +{"train_loss": 0.028244737535715103, "train_loss_bc": 0.028244737535715103, "train_loss_llm": 0.0, "grad_norm": 0.09345900267362595, "global_step": 35764, "epoch": 993, "lr": 0.001} +{"train_loss": 0.02835085615515709, "train_loss_bc": 0.028328238055109978, "train_loss_llm": 0.0022618770599365234, "grad_norm": 0.034830860793590546, "global_step": 35765, "epoch": 993, "lr": 0.001} +{"train_loss": 0.02791309915482998, "train_loss_bc": 0.02791309915482998, "train_loss_llm": 0.0, "grad_norm": 0.058484841138124466, "global_step": 35766, "epoch": 993, "lr": 0.001} +{"train_loss": 0.024438507854938507, "train_loss_bc": 0.02442234754562378, "train_loss_llm": 0.0016160905361175537, "grad_norm": 0.08077261596918106, "global_step": 35767, "epoch": 993, "lr": 0.001} +{"train_loss": 0.025091255083680153, "train_loss_bc": 0.025062892585992813, "train_loss_llm": 0.002836287021636963, "grad_norm": 0.10218608379364014, "global_step": 35768, "epoch": 993, "lr": 0.001} +{"train_loss": 0.029622986912727356, "train_loss_bc": 0.029622986912727356, "train_loss_llm": 0.0, "grad_norm": 0.04131801053881645, "global_step": 35769, "epoch": 993, "lr": 0.001} +{"train_loss": 0.023281436413526535, "train_loss_bc": 0.023281436413526535, "train_loss_llm": 0.0, "grad_norm": 0.06111976504325867, "global_step": 35770, "epoch": 993, "lr": 0.001} +{"train_loss": 0.025386998429894447, "train_loss_bc": 0.025386998429894447, "train_loss_llm": 0.0, "grad_norm": 0.0884532704949379, "global_step": 35771, "epoch": 993, "lr": 0.001} +{"train_loss": 0.03220194950699806, "train_loss_bc": 0.03220194950699806, "train_loss_llm": 0.0, "grad_norm": 0.11177432537078857, "global_step": 35772, "epoch": 993, "lr": 0.001} +{"train_loss": 0.0274953693151474, "train_loss_bc": 0.0274953693151474, "train_loss_llm": 0.0, "grad_norm": 0.03606102243065834, "global_step": 35773, "epoch": 993, "lr": 0.001} +{"train_loss": 0.02571789175271988, "train_loss_bc": 0.025711899623274803, "train_loss_llm": 0.0005991458892822266, "grad_norm": 0.04943910613656044, "global_step": 35774, "epoch": 993, "lr": 0.001} +{"train_loss": 0.030123479664325714, "train_loss_bc": 0.030120525509119034, "train_loss_llm": 0.00029543042182922363, "grad_norm": 0.0731196328997612, "global_step": 35775, "epoch": 993, "lr": 0.001} +{"train_loss": 0.02811146341264248, "train_loss_bc": 0.02811146341264248, "train_loss_llm": 0.0, "grad_norm": 0.0902523323893547, "global_step": 35776, "epoch": 993, "lr": 0.001} +{"train_loss": 0.030574306845664978, "train_loss_bc": 0.030574306845664978, "train_loss_llm": 0.0, "grad_norm": 0.04681858420372009, "global_step": 35777, "epoch": 993, "lr": 0.001} +{"train_loss": 0.030749067664146423, "train_loss_bc": 0.030749067664146423, "train_loss_llm": 0.0, "grad_norm": 0.09194624423980713, "global_step": 35778, "epoch": 993, "lr": 0.001} +{"train_loss": 0.029031485319137573, "train_loss_bc": 0.029031485319137573, "train_loss_llm": 0.0, "grad_norm": 0.10409589856863022, "global_step": 35779, "epoch": 993, "lr": 0.001} +{"train_loss": 0.03247562795877457, "train_loss_bc": 0.0324721597135067, "train_loss_llm": 0.00034683942794799805, "grad_norm": 0.13306006789207458, "global_step": 35780, "epoch": 993, "lr": 0.001} +{"train_loss": 0.023901166394352913, "train_loss_bc": 0.023901166394352913, "train_loss_llm": 0.0, "grad_norm": 0.03575945645570755, "global_step": 35781, "epoch": 993, "lr": 0.001} +{"train_loss": 0.03290816396474838, "train_loss_bc": 0.03290816396474838, "train_loss_llm": 0.0, "grad_norm": 0.058011457324028015, "global_step": 35782, "epoch": 993, "lr": 0.001} +{"train_loss": 0.028536681913667254, "train_loss_bc": 0.037224479019641876, "train_loss_llm": 0.00539669394493103, "grad_norm": 0.0881904736161232, "global_step": 35783, "epoch": 993, "lr": 0.001, "val_loss": 0.11320044845342636} +{"train_loss": 0.028327701613307, "train_loss_bc": 0.028327701613307, "train_loss_llm": 0.0, "grad_norm": 0.11653617769479752, "global_step": 35784, "epoch": 994, "lr": 0.001} +{"train_loss": 0.028250163421034813, "train_loss_bc": 0.02824411913752556, "train_loss_llm": 0.0006044507026672363, "grad_norm": 0.03457583859562874, "global_step": 35785, "epoch": 994, "lr": 0.001} +{"train_loss": 0.029631948098540306, "train_loss_bc": 0.029631948098540306, "train_loss_llm": 0.0, "grad_norm": 0.07130339741706848, "global_step": 35786, "epoch": 994, "lr": 0.001} +{"train_loss": 0.02936471253633499, "train_loss_bc": 0.029354875907301903, "train_loss_llm": 0.0009836852550506592, "grad_norm": 0.09821703284978867, "global_step": 35787, "epoch": 994, "lr": 0.001} +{"train_loss": 0.027615470811724663, "train_loss_bc": 0.027615470811724663, "train_loss_llm": 0.0, "grad_norm": 0.11472099274396896, "global_step": 35788, "epoch": 994, "lr": 0.001} +{"train_loss": 0.033960968255996704, "train_loss_bc": 0.033960968255996704, "train_loss_llm": 0.0, "grad_norm": 0.0390704907476902, "global_step": 35789, "epoch": 994, "lr": 0.001} +{"train_loss": 0.02209610491991043, "train_loss_bc": 0.02209610491991043, "train_loss_llm": 0.0, "grad_norm": 0.05913297459483147, "global_step": 35790, "epoch": 994, "lr": 0.001} +{"train_loss": 0.026779336854815483, "train_loss_bc": 0.026773249730467796, "train_loss_llm": 0.0006087422370910645, "grad_norm": 0.07869064062833786, "global_step": 35791, "epoch": 994, "lr": 0.001} +{"train_loss": 0.030801625922322273, "train_loss_bc": 0.030801625922322273, "train_loss_llm": 0.0, "grad_norm": 0.10141298919916153, "global_step": 35792, "epoch": 994, "lr": 0.001} +{"train_loss": 0.02904651127755642, "train_loss_bc": 0.029040560126304626, "train_loss_llm": 0.0005951225757598877, "grad_norm": 0.04547696188092232, "global_step": 35793, "epoch": 994, "lr": 0.001} +{"train_loss": 0.028214558959007263, "train_loss_bc": 0.028214558959007263, "train_loss_llm": 0.0, "grad_norm": 0.06537745147943497, "global_step": 35794, "epoch": 994, "lr": 0.001} +{"train_loss": 0.02932823821902275, "train_loss_bc": 0.02932823821902275, "train_loss_llm": 0.0, "grad_norm": 0.09795470535755157, "global_step": 35795, "epoch": 994, "lr": 0.001} +{"train_loss": 0.03261232376098633, "train_loss_bc": 0.03261232376098633, "train_loss_llm": 0.0, "grad_norm": 0.12043043226003647, "global_step": 35796, "epoch": 994, "lr": 0.001} +{"train_loss": 0.02245955727994442, "train_loss_bc": 0.02245108038187027, "train_loss_llm": 0.0008476376533508301, "grad_norm": 0.028355883434414864, "global_step": 35797, "epoch": 994, "lr": 0.001} +{"train_loss": 0.03095981292426586, "train_loss_bc": 0.03095981292426586, "train_loss_llm": 0.0, "grad_norm": 0.06034203618764877, "global_step": 35798, "epoch": 994, "lr": 0.001} +{"train_loss": 0.026399172842502594, "train_loss_bc": 0.026399172842502594, "train_loss_llm": 0.0, "grad_norm": 0.07935220003128052, "global_step": 35799, "epoch": 994, "lr": 0.001} +{"train_loss": 0.0307163093239069, "train_loss_bc": 0.0307163093239069, "train_loss_llm": 0.0, "grad_norm": 0.09650006890296936, "global_step": 35800, "epoch": 994, "lr": 0.001} +{"train_loss": 0.03068975731730461, "train_loss_bc": 0.03068975731730461, "train_loss_llm": 0.0, "grad_norm": 0.043139416724443436, "global_step": 35801, "epoch": 994, "lr": 0.001} +{"train_loss": 0.021739257499575615, "train_loss_bc": 0.021727683022618294, "train_loss_llm": 0.0011574923992156982, "grad_norm": 0.06528165936470032, "global_step": 35802, "epoch": 994, "lr": 0.001} +{"train_loss": 0.02472670190036297, "train_loss_bc": 0.024720951914787292, "train_loss_llm": 0.000574946403503418, "grad_norm": 0.07624707370996475, "global_step": 35803, "epoch": 994, "lr": 0.001} +{"train_loss": 0.025391679257154465, "train_loss_bc": 0.025391679257154465, "train_loss_llm": 0.0, "grad_norm": 0.08695165812969208, "global_step": 35804, "epoch": 994, "lr": 0.001} +{"train_loss": 0.03292204067111015, "train_loss_bc": 0.03292204067111015, "train_loss_llm": 0.0, "grad_norm": 0.049765750765800476, "global_step": 35805, "epoch": 994, "lr": 0.001} +{"train_loss": 0.024461016058921814, "train_loss_bc": 0.024461016058921814, "train_loss_llm": 0.0, "grad_norm": 0.06540004909038544, "global_step": 35806, "epoch": 994, "lr": 0.001} +{"train_loss": 0.03112073801457882, "train_loss_bc": 0.031113239005208015, "train_loss_llm": 0.0007498264312744141, "grad_norm": 0.08851751685142517, "global_step": 35807, "epoch": 994, "lr": 0.001} +{"train_loss": 0.030666884034872055, "train_loss_bc": 0.030666884034872055, "train_loss_llm": 0.0, "grad_norm": 0.11471952497959137, "global_step": 35808, "epoch": 994, "lr": 0.001} +{"train_loss": 0.022487027570605278, "train_loss_bc": 0.022487027570605278, "train_loss_llm": 0.0, "grad_norm": 0.03735945373773575, "global_step": 35809, "epoch": 994, "lr": 0.001} +{"train_loss": 0.02839142270386219, "train_loss_bc": 0.02839142270386219, "train_loss_llm": 0.0, "grad_norm": 0.04834238439798355, "global_step": 35810, "epoch": 994, "lr": 0.001} +{"train_loss": 0.027200942859053612, "train_loss_bc": 0.027200942859053612, "train_loss_llm": 0.0, "grad_norm": 0.07145515829324722, "global_step": 35811, "epoch": 994, "lr": 0.001} +{"train_loss": 0.028273360803723335, "train_loss_bc": 0.02825615182518959, "train_loss_llm": 0.0017208755016326904, "grad_norm": 0.08889409899711609, "global_step": 35812, "epoch": 994, "lr": 0.001} +{"train_loss": 0.028050091117620468, "train_loss_bc": 0.028050091117620468, "train_loss_llm": 0.0, "grad_norm": 0.0377378836274147, "global_step": 35813, "epoch": 994, "lr": 0.001} +{"train_loss": 0.02605089545249939, "train_loss_bc": 0.02605089545249939, "train_loss_llm": 0.0, "grad_norm": 0.06780410557985306, "global_step": 35814, "epoch": 994, "lr": 0.001} +{"train_loss": 0.03626938909292221, "train_loss_bc": 0.03626938909292221, "train_loss_llm": 0.0, "grad_norm": 0.11212436109781265, "global_step": 35815, "epoch": 994, "lr": 0.001} +{"train_loss": 0.02877604402601719, "train_loss_bc": 0.02877604402601719, "train_loss_llm": 0.0, "grad_norm": 0.12830489873886108, "global_step": 35816, "epoch": 994, "lr": 0.001} +{"train_loss": 0.03326639160513878, "train_loss_bc": 0.03326639160513878, "train_loss_llm": 0.0, "grad_norm": 0.05631370469927788, "global_step": 35817, "epoch": 994, "lr": 0.001} +{"train_loss": 0.032110605388879776, "train_loss_bc": 0.032110605388879776, "train_loss_llm": 0.0, "grad_norm": 0.079481340944767, "global_step": 35818, "epoch": 994, "lr": 0.001} +{"train_loss": 0.028401052320583, "train_loss_bc": 0.023270590230822563, "train_loss_llm": 0.0008530914783477783, "grad_norm": 0.10282263904809952, "global_step": 35819, "epoch": 994, "lr": 0.001, "val_loss": 0.1119808703660965} +{"train_loss": 0.02680802345275879, "train_loss_bc": 0.026795316487550735, "train_loss_llm": 0.001270681619644165, "grad_norm": 0.12167415767908096, "global_step": 35820, "epoch": 995, "lr": 0.001} +{"train_loss": 0.0323713980615139, "train_loss_bc": 0.0323713980615139, "train_loss_llm": 0.0, "grad_norm": 0.0457860492169857, "global_step": 35821, "epoch": 995, "lr": 0.001} +{"train_loss": 0.028211256489157677, "train_loss_bc": 0.028211256489157677, "train_loss_llm": 0.0, "grad_norm": 0.07382596284151077, "global_step": 35822, "epoch": 995, "lr": 0.001} +{"train_loss": 0.02953719161450863, "train_loss_bc": 0.02953719161450863, "train_loss_llm": 0.0, "grad_norm": 0.09661056846380234, "global_step": 35823, "epoch": 995, "lr": 0.001} +{"train_loss": 0.027547389268875122, "train_loss_bc": 0.02753477357327938, "train_loss_llm": 0.0012616515159606934, "grad_norm": 0.12290889769792557, "global_step": 35824, "epoch": 995, "lr": 0.001} +{"train_loss": 0.032370880246162415, "train_loss_bc": 0.032370880246162415, "train_loss_llm": 0.0, "grad_norm": 0.036568984389305115, "global_step": 35825, "epoch": 995, "lr": 0.001} +{"train_loss": 0.030457260087132454, "train_loss_bc": 0.030457260087132454, "train_loss_llm": 0.0, "grad_norm": 0.05062124878168106, "global_step": 35826, "epoch": 995, "lr": 0.001} +{"train_loss": 0.02623523212969303, "train_loss_bc": 0.02623523212969303, "train_loss_llm": 0.0, "grad_norm": 0.06864633411169052, "global_step": 35827, "epoch": 995, "lr": 0.001} +{"train_loss": 0.033972080796957016, "train_loss_bc": 0.033972080796957016, "train_loss_llm": 0.0, "grad_norm": 0.08653206378221512, "global_step": 35828, "epoch": 995, "lr": 0.001} +{"train_loss": 0.03090289793908596, "train_loss_bc": 0.03090289793908596, "train_loss_llm": 0.0, "grad_norm": 0.037993449717760086, "global_step": 35829, "epoch": 995, "lr": 0.001} +{"train_loss": 0.024768834933638573, "train_loss_bc": 0.024766966700553894, "train_loss_llm": 0.0001868605613708496, "grad_norm": 0.04869721457362175, "global_step": 35830, "epoch": 995, "lr": 0.001} +{"train_loss": 0.025062158703804016, "train_loss_bc": 0.025062158703804016, "train_loss_llm": 0.0, "grad_norm": 0.06741146743297577, "global_step": 35831, "epoch": 995, "lr": 0.001} +{"train_loss": 0.030500464141368866, "train_loss_bc": 0.030500464141368866, "train_loss_llm": 0.0, "grad_norm": 0.0846848413348198, "global_step": 35832, "epoch": 995, "lr": 0.001} +{"train_loss": 0.029525799676775932, "train_loss_bc": 0.029525799676775932, "train_loss_llm": 0.0, "grad_norm": 0.03355559706687927, "global_step": 35833, "epoch": 995, "lr": 0.001} +{"train_loss": 0.022291136905550957, "train_loss_bc": 0.022291136905550957, "train_loss_llm": 0.0, "grad_norm": 0.05779606103897095, "global_step": 35834, "epoch": 995, "lr": 0.001} +{"train_loss": 0.028505900874733925, "train_loss_bc": 0.028505900874733925, "train_loss_llm": 0.0, "grad_norm": 0.08284880965948105, "global_step": 35835, "epoch": 995, "lr": 0.001} +{"train_loss": 0.028794962912797928, "train_loss_bc": 0.028794962912797928, "train_loss_llm": 0.0, "grad_norm": 0.10038383305072784, "global_step": 35836, "epoch": 995, "lr": 0.001} +{"train_loss": 0.028012530878186226, "train_loss_bc": 0.028012530878186226, "train_loss_llm": 0.0, "grad_norm": 0.04164275527000427, "global_step": 35837, "epoch": 995, "lr": 0.001} +{"train_loss": 0.030519232153892517, "train_loss_bc": 0.030519232153892517, "train_loss_llm": 0.0, "grad_norm": 0.06442799419164658, "global_step": 35838, "epoch": 995, "lr": 0.001} +{"train_loss": 0.03184139356017113, "train_loss_bc": 0.03184139356017113, "train_loss_llm": 0.0, "grad_norm": 0.08618155121803284, "global_step": 35839, "epoch": 995, "lr": 0.001} +{"train_loss": 0.028822705149650574, "train_loss_bc": 0.028822705149650574, "train_loss_llm": 0.0, "grad_norm": 0.10967826098203659, "global_step": 35840, "epoch": 995, "lr": 0.001} +{"train_loss": 0.025731630623340607, "train_loss_bc": 0.025731630623340607, "train_loss_llm": 0.0, "grad_norm": 0.04038989543914795, "global_step": 35841, "epoch": 995, "lr": 0.001} +{"train_loss": 0.02835407666862011, "train_loss_bc": 0.02835407666862011, "train_loss_llm": 0.0, "grad_norm": 0.06888016313314438, "global_step": 35842, "epoch": 995, "lr": 0.001} +{"train_loss": 0.029835371300578117, "train_loss_bc": 0.029820863157510757, "train_loss_llm": 0.0014507770538330078, "grad_norm": 0.09525076299905777, "global_step": 35843, "epoch": 995, "lr": 0.001} +{"train_loss": 0.026450594887137413, "train_loss_bc": 0.026450594887137413, "train_loss_llm": 0.0, "grad_norm": 0.12698043882846832, "global_step": 35844, "epoch": 995, "lr": 0.001} +{"train_loss": 0.028081366792321205, "train_loss_bc": 0.028081366792321205, "train_loss_llm": 0.0, "grad_norm": 0.03984145447611809, "global_step": 35845, "epoch": 995, "lr": 0.001} +{"train_loss": 0.027725251391530037, "train_loss_bc": 0.027725251391530037, "train_loss_llm": 0.0, "grad_norm": 0.053308986127376556, "global_step": 35846, "epoch": 995, "lr": 0.001} +{"train_loss": 0.02938169427216053, "train_loss_bc": 0.02938169427216053, "train_loss_llm": 0.0, "grad_norm": 0.08295042812824249, "global_step": 35847, "epoch": 995, "lr": 0.001} +{"train_loss": 0.025480525568127632, "train_loss_bc": 0.025480525568127632, "train_loss_llm": 0.0, "grad_norm": 0.09732282161712646, "global_step": 35848, "epoch": 995, "lr": 0.001} +{"train_loss": 0.03235634043812752, "train_loss_bc": 0.03235634043812752, "train_loss_llm": 0.0, "grad_norm": 0.044099632650613785, "global_step": 35849, "epoch": 995, "lr": 0.001} +{"train_loss": 0.027701493352651596, "train_loss_bc": 0.027701493352651596, "train_loss_llm": 0.0, "grad_norm": 0.05602802708745003, "global_step": 35850, "epoch": 995, "lr": 0.001} +{"train_loss": 0.022909050807356834, "train_loss_bc": 0.02290632203221321, "train_loss_llm": 0.00027292966842651367, "grad_norm": 0.08436618745326996, "global_step": 35851, "epoch": 995, "lr": 0.001} +{"train_loss": 0.02718198299407959, "train_loss_bc": 0.02718198299407959, "train_loss_llm": 0.0, "grad_norm": 0.09850770980119705, "global_step": 35852, "epoch": 995, "lr": 0.001} +{"train_loss": 0.031478747725486755, "train_loss_bc": 0.031478747725486755, "train_loss_llm": 0.0, "grad_norm": 0.055378325283527374, "global_step": 35853, "epoch": 995, "lr": 0.001} +{"train_loss": 0.026158325374126434, "train_loss_bc": 0.026158325374126434, "train_loss_llm": 0.0, "grad_norm": 0.07698976993560791, "global_step": 35854, "epoch": 995, "lr": 0.001} +{"train_loss": 0.028237283126347594, "train_loss_bc": 0.0206570103764534, "train_loss_llm": 0.0, "grad_norm": 0.10151594132184982, "global_step": 35855, "epoch": 995, "lr": 0.001, "train/cumulative_reward": 52.40677500406647, "train/mean_score": 17.82440465404907, "train/success_rate": 0.5, "test/cumulative_reward": 33.704996382572716, "test/mean_score": 7.178650150542108, "test/success_rate": 0.16, "val_loss": 0.1114034429192543, "train_action_mse_error": 0.032071854919195175} +{"train_loss": 0.029265115037560463, "train_loss_bc": 0.02926037833094597, "train_loss_llm": 0.0004736185073852539, "grad_norm": 0.12229906022548676, "global_step": 35856, "epoch": 996, "lr": 0.001} +{"train_loss": 0.02867940254509449, "train_loss_bc": 0.02867940254509449, "train_loss_llm": 0.0, "grad_norm": 0.036376941949129105, "global_step": 35857, "epoch": 996, "lr": 0.001} +{"train_loss": 0.027008581906557083, "train_loss_bc": 0.027008581906557083, "train_loss_llm": 0.0, "grad_norm": 0.0675387978553772, "global_step": 35858, "epoch": 996, "lr": 0.001} +{"train_loss": 0.025208357721567154, "train_loss_bc": 0.025208357721567154, "train_loss_llm": 0.0, "grad_norm": 0.08469084650278091, "global_step": 35859, "epoch": 996, "lr": 0.001} +{"train_loss": 0.02096938155591488, "train_loss_bc": 0.020964832976460457, "train_loss_llm": 0.00045481324195861816, "grad_norm": 0.10385633260011673, "global_step": 35860, "epoch": 996, "lr": 0.001} +{"train_loss": 0.026232019066810608, "train_loss_bc": 0.026232019066810608, "train_loss_llm": 0.0, "grad_norm": 0.03584262356162071, "global_step": 35861, "epoch": 996, "lr": 0.001} +{"train_loss": 0.028531398624181747, "train_loss_bc": 0.028531398624181747, "train_loss_llm": 0.0, "grad_norm": 0.06348118185997009, "global_step": 35862, "epoch": 996, "lr": 0.001} +{"train_loss": 0.02857263572514057, "train_loss_bc": 0.02857263572514057, "train_loss_llm": 0.0, "grad_norm": 0.07983853667974472, "global_step": 35863, "epoch": 996, "lr": 0.001} +{"train_loss": 0.029087934643030167, "train_loss_bc": 0.029087934643030167, "train_loss_llm": 0.0, "grad_norm": 0.10040606558322906, "global_step": 35864, "epoch": 996, "lr": 0.001} +{"train_loss": 0.0323069766163826, "train_loss_bc": 0.0323069766163826, "train_loss_llm": 0.0, "grad_norm": 0.04659510403871536, "global_step": 35865, "epoch": 996, "lr": 0.001} +{"train_loss": 0.030238736420869827, "train_loss_bc": 0.030238736420869827, "train_loss_llm": 0.0, "grad_norm": 0.0725550428032875, "global_step": 35866, "epoch": 996, "lr": 0.001} +{"train_loss": 0.035681404173374176, "train_loss_bc": 0.03567759692668915, "train_loss_llm": 0.0003808140754699707, "grad_norm": 0.09658833593130112, "global_step": 35867, "epoch": 996, "lr": 0.001} +{"train_loss": 0.027537113055586815, "train_loss_bc": 0.027533728629350662, "train_loss_llm": 0.00033852458000183105, "grad_norm": 0.11251504719257355, "global_step": 35868, "epoch": 996, "lr": 0.001} +{"train_loss": 0.029947733506560326, "train_loss_bc": 0.02993900142610073, "train_loss_llm": 0.0008732080459594727, "grad_norm": 0.040004290640354156, "global_step": 35869, "epoch": 996, "lr": 0.001} +{"train_loss": 0.02949639782309532, "train_loss_bc": 0.02949639782309532, "train_loss_llm": 0.0, "grad_norm": 0.054214805364608765, "global_step": 35870, "epoch": 996, "lr": 0.001} +{"train_loss": 0.02520390972495079, "train_loss_bc": 0.02520390972495079, "train_loss_llm": 0.0, "grad_norm": 0.07697759568691254, "global_step": 35871, "epoch": 996, "lr": 0.001} +{"train_loss": 0.02963739074766636, "train_loss_bc": 0.02963739074766636, "train_loss_llm": 0.0, "grad_norm": 0.10372886061668396, "global_step": 35872, "epoch": 996, "lr": 0.001} +{"train_loss": 0.028103958815336227, "train_loss_bc": 0.028103113174438477, "train_loss_llm": 8.463859558105469e-05, "grad_norm": 0.04366579279303551, "global_step": 35873, "epoch": 996, "lr": 0.001} +{"train_loss": 0.02917363867163658, "train_loss_bc": 0.02917363867163658, "train_loss_llm": 0.0, "grad_norm": 0.06862732768058777, "global_step": 35874, "epoch": 996, "lr": 0.001} +{"train_loss": 0.0313534289598465, "train_loss_bc": 0.031348153948783875, "train_loss_llm": 0.0005274415016174316, "grad_norm": 0.08826399594545364, "global_step": 35875, "epoch": 996, "lr": 0.001} +{"train_loss": 0.0247453972697258, "train_loss_bc": 0.024739159271121025, "train_loss_llm": 0.0006237328052520752, "grad_norm": 0.10259079933166504, "global_step": 35876, "epoch": 996, "lr": 0.001} +{"train_loss": 0.030828136950731277, "train_loss_bc": 0.030824126675724983, "train_loss_llm": 0.0004010796546936035, "grad_norm": 0.04539939761161804, "global_step": 35877, "epoch": 996, "lr": 0.001} +{"train_loss": 0.027972344309091568, "train_loss_bc": 0.027972344309091568, "train_loss_llm": 0.0, "grad_norm": 0.08082029968500137, "global_step": 35878, "epoch": 996, "lr": 0.001} +{"train_loss": 0.02982250228524208, "train_loss_bc": 0.02982250228524208, "train_loss_llm": 0.0, "grad_norm": 0.12178154289722443, "global_step": 35879, "epoch": 996, "lr": 0.001} +{"train_loss": 0.027584506198763847, "train_loss_bc": 0.027584506198763847, "train_loss_llm": 0.0, "grad_norm": 0.14256832003593445, "global_step": 35880, "epoch": 996, "lr": 0.001} +{"train_loss": 0.03205990791320801, "train_loss_bc": 0.03205990791320801, "train_loss_llm": 0.0, "grad_norm": 0.04804206266999245, "global_step": 35881, "epoch": 996, "lr": 0.001} +{"train_loss": 0.024754414334893227, "train_loss_bc": 0.024754414334893227, "train_loss_llm": 0.0, "grad_norm": 0.07946040481328964, "global_step": 35882, "epoch": 996, "lr": 0.001} +{"train_loss": 0.028419796377420425, "train_loss_bc": 0.028408406302332878, "train_loss_llm": 0.0011390447616577148, "grad_norm": 0.09569840133190155, "global_step": 35883, "epoch": 996, "lr": 0.001} +{"train_loss": 0.031767066568136215, "train_loss_bc": 0.031767066568136215, "train_loss_llm": 0.0, "grad_norm": 0.1321904957294464, "global_step": 35884, "epoch": 996, "lr": 0.001} +{"train_loss": 0.02413884550333023, "train_loss_bc": 0.02413884550333023, "train_loss_llm": 0.0, "grad_norm": 0.040449101477861404, "global_step": 35885, "epoch": 996, "lr": 0.001} +{"train_loss": 0.026733966544270515, "train_loss_bc": 0.026733966544270515, "train_loss_llm": 0.0, "grad_norm": 0.07373040169477463, "global_step": 35886, "epoch": 996, "lr": 0.001} +{"train_loss": 0.02615523152053356, "train_loss_bc": 0.02615523152053356, "train_loss_llm": 0.0, "grad_norm": 0.0946163460612297, "global_step": 35887, "epoch": 996, "lr": 0.001} +{"train_loss": 0.029165077954530716, "train_loss_bc": 0.029165077954530716, "train_loss_llm": 0.0, "grad_norm": 0.1095791682600975, "global_step": 35888, "epoch": 996, "lr": 0.001} +{"train_loss": 0.031567517668008804, "train_loss_bc": 0.031550291925668716, "train_loss_llm": 0.0017225444316864014, "grad_norm": 0.04874636232852936, "global_step": 35889, "epoch": 996, "lr": 0.001} +{"train_loss": 0.029002483934164047, "train_loss_bc": 0.028994418680667877, "train_loss_llm": 0.0008064806461334229, "grad_norm": 0.07326284795999527, "global_step": 35890, "epoch": 996, "lr": 0.001} +{"train_loss": 0.028435619547963142, "train_loss_bc": 0.02672959305346012, "train_loss_llm": 0.0, "grad_norm": 0.1061277762055397, "global_step": 35891, "epoch": 996, "lr": 0.001, "val_loss": 0.11133401840925217} +{"train_loss": 0.027383288368582726, "train_loss_bc": 0.02737618423998356, "train_loss_llm": 0.0007103979587554932, "grad_norm": 0.1280050128698349, "global_step": 35892, "epoch": 997, "lr": 0.001} +{"train_loss": 0.03296554833650589, "train_loss_bc": 0.03296554833650589, "train_loss_llm": 0.0, "grad_norm": 0.05323446914553642, "global_step": 35893, "epoch": 997, "lr": 0.001} +{"train_loss": 0.030040185898542404, "train_loss_bc": 0.030040185898542404, "train_loss_llm": 0.0, "grad_norm": 0.08218130469322205, "global_step": 35894, "epoch": 997, "lr": 0.001} +{"train_loss": 0.031087292358279228, "train_loss_bc": 0.031087292358279228, "train_loss_llm": 0.0, "grad_norm": 0.10900472104549408, "global_step": 35895, "epoch": 997, "lr": 0.001} +{"train_loss": 0.028830457478761673, "train_loss_bc": 0.028813984245061874, "train_loss_llm": 0.001647353172302246, "grad_norm": 0.1380959302186966, "global_step": 35896, "epoch": 997, "lr": 0.001} +{"train_loss": 0.029174692928791046, "train_loss_bc": 0.02916613593697548, "train_loss_llm": 0.0008556842803955078, "grad_norm": 0.04775264859199524, "global_step": 35897, "epoch": 997, "lr": 0.001} +{"train_loss": 0.02764257602393627, "train_loss_bc": 0.02764257602393627, "train_loss_llm": 0.0, "grad_norm": 0.07352641969919205, "global_step": 35898, "epoch": 997, "lr": 0.001} +{"train_loss": 0.023491214960813522, "train_loss_bc": 0.02348843216896057, "train_loss_llm": 0.0002783536911010742, "grad_norm": 0.08575353026390076, "global_step": 35899, "epoch": 997, "lr": 0.001} +{"train_loss": 0.03260906785726547, "train_loss_bc": 0.03260551393032074, "train_loss_llm": 0.0003555119037628174, "grad_norm": 0.10596290230751038, "global_step": 35900, "epoch": 997, "lr": 0.001} +{"train_loss": 0.029108868911862373, "train_loss_bc": 0.029108868911862373, "train_loss_llm": 0.0, "grad_norm": 0.03766653314232826, "global_step": 35901, "epoch": 997, "lr": 0.001} +{"train_loss": 0.026889340952038765, "train_loss_bc": 0.026889340952038765, "train_loss_llm": 0.0, "grad_norm": 0.06507641822099686, "global_step": 35902, "epoch": 997, "lr": 0.001} +{"train_loss": 0.03337589651346207, "train_loss_bc": 0.03337589651346207, "train_loss_llm": 0.0, "grad_norm": 0.08892285078763962, "global_step": 35903, "epoch": 997, "lr": 0.001} +{"train_loss": 0.023874394595623016, "train_loss_bc": 0.023874394595623016, "train_loss_llm": 0.0, "grad_norm": 0.10840316861867905, "global_step": 35904, "epoch": 997, "lr": 0.001} +{"train_loss": 0.03007582761347294, "train_loss_bc": 0.03006763570010662, "train_loss_llm": 0.0008192062377929688, "grad_norm": 0.04178285971283913, "global_step": 35905, "epoch": 997, "lr": 0.001} +{"train_loss": 0.027713032439351082, "train_loss_bc": 0.027713032439351082, "train_loss_llm": 0.0, "grad_norm": 0.07092281430959702, "global_step": 35906, "epoch": 997, "lr": 0.001} +{"train_loss": 0.027887556701898575, "train_loss_bc": 0.027885127812623978, "train_loss_llm": 0.0002428293228149414, "grad_norm": 0.09567656368017197, "global_step": 35907, "epoch": 997, "lr": 0.001} +{"train_loss": 0.029854843392968178, "train_loss_bc": 0.02985442243516445, "train_loss_llm": 4.208087921142578e-05, "grad_norm": 0.12915532290935516, "global_step": 35908, "epoch": 997, "lr": 0.001} +{"train_loss": 0.029048318043351173, "train_loss_bc": 0.029048318043351173, "train_loss_llm": 0.0, "grad_norm": 0.04417288303375244, "global_step": 35909, "epoch": 997, "lr": 0.001} +{"train_loss": 0.030973443761467934, "train_loss_bc": 0.030969638377428055, "train_loss_llm": 0.00038057565689086914, "grad_norm": 0.08320071548223495, "global_step": 35910, "epoch": 997, "lr": 0.001} +{"train_loss": 0.030266933143138885, "train_loss_bc": 0.030263006687164307, "train_loss_llm": 0.00039258599281311035, "grad_norm": 0.10847549140453339, "global_step": 35911, "epoch": 997, "lr": 0.001} +{"train_loss": 0.024968570098280907, "train_loss_bc": 0.024968570098280907, "train_loss_llm": 0.0, "grad_norm": 0.12287071347236633, "global_step": 35912, "epoch": 997, "lr": 0.001} +{"train_loss": 0.02923412434756756, "train_loss_bc": 0.02923412434756756, "train_loss_llm": 0.0, "grad_norm": 0.03692643716931343, "global_step": 35913, "epoch": 997, "lr": 0.001} +{"train_loss": 0.03159854933619499, "train_loss_bc": 0.03159854933619499, "train_loss_llm": 0.0, "grad_norm": 0.0561603419482708, "global_step": 35914, "epoch": 997, "lr": 0.001} +{"train_loss": 0.026850096881389618, "train_loss_bc": 0.026850096881389618, "train_loss_llm": 0.0, "grad_norm": 0.07247644662857056, "global_step": 35915, "epoch": 997, "lr": 0.001} +{"train_loss": 0.029376424849033356, "train_loss_bc": 0.029371393844485283, "train_loss_llm": 0.0005031228065490723, "grad_norm": 0.10249063372612, "global_step": 35916, "epoch": 997, "lr": 0.001} +{"train_loss": 0.02724180929362774, "train_loss_bc": 0.02724180929362774, "train_loss_llm": 0.0, "grad_norm": 0.03374848887324333, "global_step": 35917, "epoch": 997, "lr": 0.001} +{"train_loss": 0.028110703453421593, "train_loss_bc": 0.028110703453421593, "train_loss_llm": 0.0, "grad_norm": 0.04624711349606514, "global_step": 35918, "epoch": 997, "lr": 0.001} +{"train_loss": 0.027965130284428596, "train_loss_bc": 0.027965130284428596, "train_loss_llm": 0.0, "grad_norm": 0.06392346322536469, "global_step": 35919, "epoch": 997, "lr": 0.001} +{"train_loss": 0.023070888593792915, "train_loss_bc": 0.023070888593792915, "train_loss_llm": 0.0, "grad_norm": 0.07372922450304031, "global_step": 35920, "epoch": 997, "lr": 0.001} +{"train_loss": 0.02695378288626671, "train_loss_bc": 0.02695378288626671, "train_loss_llm": 0.0, "grad_norm": 0.047768570482730865, "global_step": 35921, "epoch": 997, "lr": 0.001} +{"train_loss": 0.023684948682785034, "train_loss_bc": 0.023684948682785034, "train_loss_llm": 0.0, "grad_norm": 0.07719133794307709, "global_step": 35922, "epoch": 997, "lr": 0.001} +{"train_loss": 0.03188609331846237, "train_loss_bc": 0.03188609331846237, "train_loss_llm": 0.0, "grad_norm": 0.10792554914951324, "global_step": 35923, "epoch": 997, "lr": 0.001} +{"train_loss": 0.02784670889377594, "train_loss_bc": 0.02784670889377594, "train_loss_llm": 0.0, "grad_norm": 0.1376928985118866, "global_step": 35924, "epoch": 997, "lr": 0.001} +{"train_loss": 0.03095979616045952, "train_loss_bc": 0.03095979616045952, "train_loss_llm": 0.0, "grad_norm": 0.05003714561462402, "global_step": 35925, "epoch": 997, "lr": 0.001} +{"train_loss": 0.0257052481174469, "train_loss_bc": 0.0257052481174469, "train_loss_llm": 0.0, "grad_norm": 0.07722373306751251, "global_step": 35926, "epoch": 997, "lr": 0.001} +{"train_loss": 0.028502497925526567, "train_loss_bc": 0.02834426984190941, "train_loss_llm": 0.0, "grad_norm": 0.1147422268986702, "global_step": 35927, "epoch": 997, "lr": 0.001, "val_loss": 0.11298423260450363} +{"train_loss": 0.028224050998687744, "train_loss_bc": 0.028218097984790802, "train_loss_llm": 0.0005953311920166016, "grad_norm": 0.15226668119430542, "global_step": 35928, "epoch": 998, "lr": 0.001} +{"train_loss": 0.03168525546789169, "train_loss_bc": 0.03168525546789169, "train_loss_llm": 0.0, "grad_norm": 0.04085066169500351, "global_step": 35929, "epoch": 998, "lr": 0.001} +{"train_loss": 0.024038029834628105, "train_loss_bc": 0.024038029834628105, "train_loss_llm": 0.0, "grad_norm": 0.07639192044734955, "global_step": 35930, "epoch": 998, "lr": 0.001} +{"train_loss": 0.026533251628279686, "train_loss_bc": 0.026533251628279686, "train_loss_llm": 0.0, "grad_norm": 0.09995946288108826, "global_step": 35931, "epoch": 998, "lr": 0.001} +{"train_loss": 0.027193240821361542, "train_loss_bc": 0.027193240821361542, "train_loss_llm": 0.0, "grad_norm": 0.13146550953388214, "global_step": 35932, "epoch": 998, "lr": 0.001} +{"train_loss": 0.028780700638890266, "train_loss_bc": 0.028780700638890266, "train_loss_llm": 0.0, "grad_norm": 0.03543076664209366, "global_step": 35933, "epoch": 998, "lr": 0.001} +{"train_loss": 0.023094162344932556, "train_loss_bc": 0.023094162344932556, "train_loss_llm": 0.0, "grad_norm": 0.06450658291578293, "global_step": 35934, "epoch": 998, "lr": 0.001} +{"train_loss": 0.02711278386414051, "train_loss_bc": 0.02711278386414051, "train_loss_llm": 0.0, "grad_norm": 0.08031097054481506, "global_step": 35935, "epoch": 998, "lr": 0.001} +{"train_loss": 0.030932540073990822, "train_loss_bc": 0.030932540073990822, "train_loss_llm": 0.0, "grad_norm": 0.10347855091094971, "global_step": 35936, "epoch": 998, "lr": 0.001} +{"train_loss": 0.02696937322616577, "train_loss_bc": 0.02696937322616577, "train_loss_llm": 0.0, "grad_norm": 0.0404481515288353, "global_step": 35937, "epoch": 998, "lr": 0.001} +{"train_loss": 0.02803436480462551, "train_loss_bc": 0.02803436480462551, "train_loss_llm": 0.0, "grad_norm": 0.07022713869810104, "global_step": 35938, "epoch": 998, "lr": 0.001} +{"train_loss": 0.028511907905340195, "train_loss_bc": 0.02851053699851036, "train_loss_llm": 0.00013709068298339844, "grad_norm": 0.09646390378475189, "global_step": 35939, "epoch": 998, "lr": 0.001} +{"train_loss": 0.027934258803725243, "train_loss_bc": 0.027934258803725243, "train_loss_llm": 0.0, "grad_norm": 0.11425277590751648, "global_step": 35940, "epoch": 998, "lr": 0.001} +{"train_loss": 0.02714075706899166, "train_loss_bc": 0.02714075706899166, "train_loss_llm": 0.0, "grad_norm": 0.038887135684490204, "global_step": 35941, "epoch": 998, "lr": 0.001} +{"train_loss": 0.028226377442479134, "train_loss_bc": 0.028226377442479134, "train_loss_llm": 0.0, "grad_norm": 0.06847404688596725, "global_step": 35942, "epoch": 998, "lr": 0.001} +{"train_loss": 0.033251166343688965, "train_loss_bc": 0.033251166343688965, "train_loss_llm": 0.0, "grad_norm": 0.1101200208067894, "global_step": 35943, "epoch": 998, "lr": 0.001} +{"train_loss": 0.031556449830532074, "train_loss_bc": 0.031556449830532074, "train_loss_llm": 0.0, "grad_norm": 0.12579581141471863, "global_step": 35944, "epoch": 998, "lr": 0.001} +{"train_loss": 0.026824090629816055, "train_loss_bc": 0.02681640349328518, "train_loss_llm": 0.0007687807083129883, "grad_norm": 0.04966655746102333, "global_step": 35945, "epoch": 998, "lr": 0.001} +{"train_loss": 0.024875570088624954, "train_loss_bc": 0.024875570088624954, "train_loss_llm": 0.0, "grad_norm": 0.07822209596633911, "global_step": 35946, "epoch": 998, "lr": 0.001} +{"train_loss": 0.026706963777542114, "train_loss_bc": 0.026706963777542114, "train_loss_llm": 0.0, "grad_norm": 0.0847107395529747, "global_step": 35947, "epoch": 998, "lr": 0.001} +{"train_loss": 0.029865706339478493, "train_loss_bc": 0.02986270748078823, "train_loss_llm": 0.00029984116554260254, "grad_norm": 0.10789083689451218, "global_step": 35948, "epoch": 998, "lr": 0.001} +{"train_loss": 0.03400634229183197, "train_loss_bc": 0.03400634229183197, "train_loss_llm": 0.0, "grad_norm": 0.04215095564723015, "global_step": 35949, "epoch": 998, "lr": 0.001} +{"train_loss": 0.030235454440116882, "train_loss_bc": 0.030235454440116882, "train_loss_llm": 0.0, "grad_norm": 0.06351511925458908, "global_step": 35950, "epoch": 998, "lr": 0.001} +{"train_loss": 0.03153092414140701, "train_loss_bc": 0.03152100369334221, "train_loss_llm": 0.0009920895099639893, "grad_norm": 0.09162287414073944, "global_step": 35951, "epoch": 998, "lr": 0.001} +{"train_loss": 0.02602298930287361, "train_loss_bc": 0.02602298930287361, "train_loss_llm": 0.0, "grad_norm": 0.11245790123939514, "global_step": 35952, "epoch": 998, "lr": 0.001} +{"train_loss": 0.02215055748820305, "train_loss_bc": 0.02215055748820305, "train_loss_llm": 0.0, "grad_norm": 0.029927212744951248, "global_step": 35953, "epoch": 998, "lr": 0.001} +{"train_loss": 0.029369434341788292, "train_loss_bc": 0.029369434341788292, "train_loss_llm": 0.0, "grad_norm": 0.057690639048814774, "global_step": 35954, "epoch": 998, "lr": 0.001} +{"train_loss": 0.02781362645328045, "train_loss_bc": 0.027793463319540024, "train_loss_llm": 0.0020163357257843018, "grad_norm": 0.08589711040258408, "global_step": 35955, "epoch": 998, "lr": 0.001} +{"train_loss": 0.035160135477781296, "train_loss_bc": 0.035160135477781296, "train_loss_llm": 0.0, "grad_norm": 0.11079765856266022, "global_step": 35956, "epoch": 998, "lr": 0.001} +{"train_loss": 0.026148777455091476, "train_loss_bc": 0.026144586503505707, "train_loss_llm": 0.0004190504550933838, "grad_norm": 0.03655127063393593, "global_step": 35957, "epoch": 998, "lr": 0.001} +{"train_loss": 0.03178800269961357, "train_loss_bc": 0.03178800269961357, "train_loss_llm": 0.0, "grad_norm": 0.06645244359970093, "global_step": 35958, "epoch": 998, "lr": 0.001} +{"train_loss": 0.027414772659540176, "train_loss_bc": 0.027414772659540176, "train_loss_llm": 0.0, "grad_norm": 0.08507342636585236, "global_step": 35959, "epoch": 998, "lr": 0.001} +{"train_loss": 0.025506194680929184, "train_loss_bc": 0.02550610527396202, "train_loss_llm": 8.970499038696289e-06, "grad_norm": 0.11068269610404968, "global_step": 35960, "epoch": 998, "lr": 0.001} +{"train_loss": 0.03062477521598339, "train_loss_bc": 0.03059394843876362, "train_loss_llm": 0.0030826330184936523, "grad_norm": 0.03566085919737816, "global_step": 35961, "epoch": 998, "lr": 0.001} +{"train_loss": 0.029399823397397995, "train_loss_bc": 0.029399823397397995, "train_loss_llm": 0.0, "grad_norm": 0.055072441697120667, "global_step": 35962, "epoch": 998, "lr": 0.001} +{"train_loss": 0.028449843896345958, "train_loss_bc": 0.0295315682888031, "train_loss_llm": 0.0, "grad_norm": 0.07997240871191025, "global_step": 35963, "epoch": 998, "lr": 0.001, "val_loss": 0.11384508013725281} +{"train_loss": 0.02603396587073803, "train_loss_bc": 0.0260318536311388, "train_loss_llm": 0.00021120905876159668, "grad_norm": 0.09620457887649536, "global_step": 35964, "epoch": 999, "lr": 0.001} +{"train_loss": 0.03019614703953266, "train_loss_bc": 0.03019614703953266, "train_loss_llm": 0.0, "grad_norm": 0.0369158498942852, "global_step": 35965, "epoch": 999, "lr": 0.001} +{"train_loss": 0.02083124779164791, "train_loss_bc": 0.020826732739806175, "train_loss_llm": 0.00045159459114074707, "grad_norm": 0.05715273693203926, "global_step": 35966, "epoch": 999, "lr": 0.001} +{"train_loss": 0.026805609464645386, "train_loss_bc": 0.02679680474102497, "train_loss_llm": 0.0008804798126220703, "grad_norm": 0.0695536732673645, "global_step": 35967, "epoch": 999, "lr": 0.001} +{"train_loss": 0.026320666074752808, "train_loss_bc": 0.026320666074752808, "train_loss_llm": 0.0, "grad_norm": 0.0924433097243309, "global_step": 35968, "epoch": 999, "lr": 0.001} +{"train_loss": 0.03151261806488037, "train_loss_bc": 0.03151261806488037, "train_loss_llm": 0.0, "grad_norm": 0.04030171036720276, "global_step": 35969, "epoch": 999, "lr": 0.001} +{"train_loss": 0.02742304652929306, "train_loss_bc": 0.027416499331593513, "train_loss_llm": 0.0006546378135681152, "grad_norm": 0.061239343136548996, "global_step": 35970, "epoch": 999, "lr": 0.001} +{"train_loss": 0.03022219054400921, "train_loss_bc": 0.03022219054400921, "train_loss_llm": 0.0, "grad_norm": 0.08692310750484467, "global_step": 35971, "epoch": 999, "lr": 0.001} +{"train_loss": 0.028253525495529175, "train_loss_bc": 0.028253525495529175, "train_loss_llm": 0.0, "grad_norm": 0.10917855054140091, "global_step": 35972, "epoch": 999, "lr": 0.001} +{"train_loss": 0.02683599852025509, "train_loss_bc": 0.02683599852025509, "train_loss_llm": 0.0, "grad_norm": 0.034532591700553894, "global_step": 35973, "epoch": 999, "lr": 0.001} +{"train_loss": 0.027476999908685684, "train_loss_bc": 0.027476999908685684, "train_loss_llm": 0.0, "grad_norm": 0.058252327144145966, "global_step": 35974, "epoch": 999, "lr": 0.001} +{"train_loss": 0.026704540476202965, "train_loss_bc": 0.026704540476202965, "train_loss_llm": 0.0, "grad_norm": 0.08734460175037384, "global_step": 35975, "epoch": 999, "lr": 0.001} +{"train_loss": 0.029354732483625412, "train_loss_bc": 0.029354732483625412, "train_loss_llm": 0.0, "grad_norm": 0.10759095847606659, "global_step": 35976, "epoch": 999, "lr": 0.001} +{"train_loss": 0.02796677127480507, "train_loss_bc": 0.02795458398759365, "train_loss_llm": 0.0012187063694000244, "grad_norm": 0.03932775557041168, "global_step": 35977, "epoch": 999, "lr": 0.001} +{"train_loss": 0.024421505630016327, "train_loss_bc": 0.024421505630016327, "train_loss_llm": 0.0, "grad_norm": 0.06344179809093475, "global_step": 35978, "epoch": 999, "lr": 0.001} +{"train_loss": 0.030598822981119156, "train_loss_bc": 0.030598822981119156, "train_loss_llm": 0.0, "grad_norm": 0.08332942426204681, "global_step": 35979, "epoch": 999, "lr": 0.001} +{"train_loss": 0.02838701382279396, "train_loss_bc": 0.02838088758289814, "train_loss_llm": 0.0006125867366790771, "grad_norm": 0.10235096514225006, "global_step": 35980, "epoch": 999, "lr": 0.001} +{"train_loss": 0.025169413536787033, "train_loss_bc": 0.025168729946017265, "train_loss_llm": 6.839632987976074e-05, "grad_norm": 0.03772352263331413, "global_step": 35981, "epoch": 999, "lr": 0.001} +{"train_loss": 0.024423768743872643, "train_loss_bc": 0.024423768743872643, "train_loss_llm": 0.0, "grad_norm": 0.04880454018712044, "global_step": 35982, "epoch": 999, "lr": 0.001} +{"train_loss": 0.02994891256093979, "train_loss_bc": 0.02994891256093979, "train_loss_llm": 0.0, "grad_norm": 0.05658667907118797, "global_step": 35983, "epoch": 999, "lr": 0.001} +{"train_loss": 0.031026901677250862, "train_loss_bc": 0.031026901677250862, "train_loss_llm": 0.0, "grad_norm": 0.07737089693546295, "global_step": 35984, "epoch": 999, "lr": 0.001} +{"train_loss": 0.024877069517970085, "train_loss_bc": 0.024877069517970085, "train_loss_llm": 0.0, "grad_norm": 0.03526029735803604, "global_step": 35985, "epoch": 999, "lr": 0.001} +{"train_loss": 0.0335545651614666, "train_loss_bc": 0.0335545651614666, "train_loss_llm": 0.0, "grad_norm": 0.055501628667116165, "global_step": 35986, "epoch": 999, "lr": 0.001} +{"train_loss": 0.030688827857375145, "train_loss_bc": 0.030688827857375145, "train_loss_llm": 0.0, "grad_norm": 0.08783449232578278, "global_step": 35987, "epoch": 999, "lr": 0.001} +{"train_loss": 0.0300153736025095, "train_loss_bc": 0.030012086033821106, "train_loss_llm": 0.0003288090229034424, "grad_norm": 0.12324608117341995, "global_step": 35988, "epoch": 999, "lr": 0.001} +{"train_loss": 0.030890192836523056, "train_loss_bc": 0.030885867774486542, "train_loss_llm": 0.000432431697845459, "grad_norm": 0.047800641506910324, "global_step": 35989, "epoch": 999, "lr": 0.001} +{"train_loss": 0.030510609969496727, "train_loss_bc": 0.030510609969496727, "train_loss_llm": 0.0, "grad_norm": 0.07058871537446976, "global_step": 35990, "epoch": 999, "lr": 0.001} +{"train_loss": 0.031246110796928406, "train_loss_bc": 0.031246110796928406, "train_loss_llm": 0.0, "grad_norm": 0.09152740240097046, "global_step": 35991, "epoch": 999, "lr": 0.001} +{"train_loss": 0.027135828509926796, "train_loss_bc": 0.027135828509926796, "train_loss_llm": 0.0, "grad_norm": 0.11159467697143555, "global_step": 35992, "epoch": 999, "lr": 0.001} +{"train_loss": 0.028505956754088402, "train_loss_bc": 0.028490127995610237, "train_loss_llm": 0.0015829503536224365, "grad_norm": 0.04214181751012802, "global_step": 35993, "epoch": 999, "lr": 0.001} +{"train_loss": 0.023323295637965202, "train_loss_bc": 0.023323295637965202, "train_loss_llm": 0.0, "grad_norm": 0.05941754952073097, "global_step": 35994, "epoch": 999, "lr": 0.001} +{"train_loss": 0.028308210894465446, "train_loss_bc": 0.028308210894465446, "train_loss_llm": 0.0, "grad_norm": 0.08420946449041367, "global_step": 35995, "epoch": 999, "lr": 0.001} +{"train_loss": 0.027378346771001816, "train_loss_bc": 0.027378346771001816, "train_loss_llm": 0.0, "grad_norm": 0.1090436726808548, "global_step": 35996, "epoch": 999, "lr": 0.001} +{"train_loss": 0.02940574288368225, "train_loss_bc": 0.02937452867627144, "train_loss_llm": 0.0031213462352752686, "grad_norm": 0.030686303973197937, "global_step": 35997, "epoch": 999, "lr": 0.001} +{"train_loss": 0.03059062547981739, "train_loss_bc": 0.030586963519454002, "train_loss_llm": 0.0003661215305328369, "grad_norm": 0.053277529776096344, "global_step": 35998, "epoch": 999, "lr": 0.001} +{"train_loss": 0.028243827788780134, "train_loss_bc": 0.030432645231485367, "train_loss_llm": 0.0, "grad_norm": 0.0851111188530922, "global_step": 35999, "epoch": 999, "lr": 0.001, "val_loss": 0.1130925863981247} +{"train_loss": 0.022915663197636604, "train_loss_bc": 0.022914506494998932, "train_loss_llm": 0.0001157522201538086, "grad_norm": 0.10129846632480621, "global_step": 36000, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.028067506849765778, "train_loss_bc": 0.028067506849765778, "train_loss_llm": 0.0, "grad_norm": 0.04441632330417633, "global_step": 36001, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.03262927010655403, "train_loss_bc": 0.03262927010655403, "train_loss_llm": 0.0, "grad_norm": 0.06857011467218399, "global_step": 36002, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.02347499504685402, "train_loss_bc": 0.02347499504685402, "train_loss_llm": 0.0, "grad_norm": 0.08706793934106827, "global_step": 36003, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.03461264818906784, "train_loss_bc": 0.03461071103811264, "train_loss_llm": 0.00019380450248718262, "grad_norm": 0.1048448234796524, "global_step": 36004, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.028760414570569992, "train_loss_bc": 0.028760414570569992, "train_loss_llm": 0.0, "grad_norm": 0.0462430901825428, "global_step": 36005, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.021843280643224716, "train_loss_bc": 0.021843280643224716, "train_loss_llm": 0.0, "grad_norm": 0.05636743828654289, "global_step": 36006, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.031223172321915627, "train_loss_bc": 0.031223172321915627, "train_loss_llm": 0.0, "grad_norm": 0.0762638971209526, "global_step": 36007, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.029995791614055634, "train_loss_bc": 0.029995791614055634, "train_loss_llm": 0.0, "grad_norm": 0.09350350499153137, "global_step": 36008, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.027448948472738266, "train_loss_bc": 0.027436433359980583, "train_loss_llm": 0.0012514889240264893, "grad_norm": 0.03145981207489967, "global_step": 36009, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.029725447297096252, "train_loss_bc": 0.02967975288629532, "train_loss_llm": 0.004569500684738159, "grad_norm": 0.04689101129770279, "global_step": 36010, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.030294684693217278, "train_loss_bc": 0.030294684693217278, "train_loss_llm": 0.0, "grad_norm": 0.06639363616704941, "global_step": 36011, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.02335014007985592, "train_loss_bc": 0.02335014007985592, "train_loss_llm": 0.0, "grad_norm": 0.0810014009475708, "global_step": 36012, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.029523134231567383, "train_loss_bc": 0.02950618788599968, "train_loss_llm": 0.0016946494579315186, "grad_norm": 0.03799524903297424, "global_step": 36013, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.025909194722771645, "train_loss_bc": 0.025899944826960564, "train_loss_llm": 0.0009249448776245117, "grad_norm": 0.05701250582933426, "global_step": 36014, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.027647020295262337, "train_loss_bc": 0.027647020295262337, "train_loss_llm": 0.0, "grad_norm": 0.08011013269424438, "global_step": 36015, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.02798040769994259, "train_loss_bc": 0.02798040769994259, "train_loss_llm": 0.0, "grad_norm": 0.10261666774749756, "global_step": 36016, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.028054941445589066, "train_loss_bc": 0.028054941445589066, "train_loss_llm": 0.0, "grad_norm": 0.03793593496084213, "global_step": 36017, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.025972140952944756, "train_loss_bc": 0.025972140952944756, "train_loss_llm": 0.0, "grad_norm": 0.06148034334182739, "global_step": 36018, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.028411302715539932, "train_loss_bc": 0.028411302715539932, "train_loss_llm": 0.0, "grad_norm": 0.09120258688926697, "global_step": 36019, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.02323869615793228, "train_loss_bc": 0.02323869615793228, "train_loss_llm": 0.0, "grad_norm": 0.09971295297145844, "global_step": 36020, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.026926007121801376, "train_loss_bc": 0.02690826542675495, "train_loss_llm": 0.0017742514610290527, "grad_norm": 0.0544767826795578, "global_step": 36021, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.0308843906968832, "train_loss_bc": 0.0308843906968832, "train_loss_llm": 0.0, "grad_norm": 0.08113867789506912, "global_step": 36022, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.029307451099157333, "train_loss_bc": 0.029307451099157333, "train_loss_llm": 0.0, "grad_norm": 0.09564003348350525, "global_step": 36023, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.0286120418459177, "train_loss_bc": 0.028605088591575623, "train_loss_llm": 0.0006953775882720947, "grad_norm": 0.11806569993495941, "global_step": 36024, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.028372392058372498, "train_loss_bc": 0.028372392058372498, "train_loss_llm": 0.0, "grad_norm": 0.043235860764980316, "global_step": 36025, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.02641252428293228, "train_loss_bc": 0.02641252428293228, "train_loss_llm": 0.0, "grad_norm": 0.06835737079381943, "global_step": 36026, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.032943498343229294, "train_loss_bc": 0.032943498343229294, "train_loss_llm": 0.0, "grad_norm": 0.08958547562360764, "global_step": 36027, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.024372240528464317, "train_loss_bc": 0.02437107264995575, "train_loss_llm": 0.00011676549911499023, "grad_norm": 0.10264307260513306, "global_step": 36028, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.026694899424910545, "train_loss_bc": 0.026694895699620247, "train_loss_llm": 2.980232238769531e-07, "grad_norm": 0.0423826165497303, "global_step": 36029, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.029990294948220253, "train_loss_bc": 0.029990294948220253, "train_loss_llm": 0.0, "grad_norm": 0.07298055291175842, "global_step": 36030, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.028227513656020164, "train_loss_bc": 0.028227513656020164, "train_loss_llm": 0.0, "grad_norm": 0.104871965944767, "global_step": 36031, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.03150227293372154, "train_loss_bc": 0.03150227293372154, "train_loss_llm": 0.0, "grad_norm": 0.13822759687900543, "global_step": 36032, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.033783406019210815, "train_loss_bc": 0.033783406019210815, "train_loss_llm": 0.0, "grad_norm": 0.04559703543782234, "global_step": 36033, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.02795550972223282, "train_loss_bc": 0.02795550972223282, "train_loss_llm": 0.0, "grad_norm": 0.07497616112232208, "global_step": 36034, "epoch": 1000, "lr": 0.001} +{"train_loss": 0.02831302945398622, "train_loss_bc": 0.032205816358327866, "train_loss_llm": 0.0, "grad_norm": 0.10948453843593597, "global_step": 36035, "epoch": 1000, "lr": 0.001, "train/cumulative_reward": 55.695824149928185, "train/mean_score": 4.7880583257433305, "train/success_rate": 0.0, "test/cumulative_reward": 50.17946204304098, "test/mean_score": 7.625797571066719, "test/success_rate": 0.18, "val_loss": 0.11186444759368896, "train_action_mse_error": 0.031062040477991104} diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log new file mode 100644 index 0000000000000000000000000000000000000000..fea00fd36dada64701c87dad154b5e52068fa223 --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/train.log @@ -0,0 +1,13 @@ +[2026-04-10 14:21:57,276][numexpr.utils][INFO] - Note: detected 144 virtual cores but NumExpr set to maximum of 64, check "NUMEXPR_MAX_THREADS" environment variable. +[2026-04-10 14:21:57,276][numexpr.utils][INFO] - Note: NumExpr detected 144 cores but "NUMEXPR_MAX_THREADS" not set, so enforcing safe limit of 16. +[2026-04-10 14:21:57,276][numexpr.utils][INFO] - NumExpr defaulting to 16 threads. +[2026-04-10 14:22:00,623][datasets][INFO] - PyTorch version 2.2.2 available. +[2026-04-10 14:22:00,624][datasets][INFO] - TensorFlow version 2.15.1 available. +[2026-04-10 14:22:00,624][datasets][INFO] - JAX version 0.4.30 available. +[2026-04-10 14:22:04,555][root][INFO] - running build_ext +[2026-04-10 14:22:04,557][root][INFO] - building 'mujoco_py.cymj' extension +[2026-04-10 14:22:04,557][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o -fopenmp -w +[2026-04-10 14:22:46,366][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -Wno-unused-result -Wsign-compare -DNDEBUG -O2 -Wall -fPIC -O2 -isystem /venv/llm-bc/include -I/venv/llm-bc/include -fPIC -O2 -isystem /venv/llm-bc/include -fPIC -I/venv/llm-bc/lib/python3.9/site-packages/mujoco_py -I/root/.mujoco/mujoco210/include -I/venv/llm-bc/lib/python3.9/site-packages/numpy/core/include -I/venv/llm-bc/include/python3.9 -c /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.c -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -fopenmp -w +[2026-04-10 14:22:46,459][root][INFO] - gcc -pthread -B /venv/llm-bc/compiler_compat -shared -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib -L/venv/llm-bc/lib -Wl,-rpath,/venv/llm-bc/lib -Wl,-rpath-link,/venv/llm-bc/lib -L/venv/llm-bc/lib /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/cymj.o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/temp.linux-x86_64-cpython-39/venv/llm-bc/lib/python3.9/site-packages/mujoco_py/gl/osmesashim.o -L/root/.mujoco/mujoco210/bin -Wl,--enable-new-dtags,-rpath,/root/.mujoco/mujoco210/bin -lmujoco210 -lglewosmesa -lOSMesa -lGL -o /venv/llm-bc/lib/python3.9/site-packages/mujoco_py/generated/_pyxbld_2.1.2.14_39_linuxcpuextensionbuilder/lib.linux-x86_64-cpython-39/mujoco_py/cymj.cpython-39-x86_64-linux-gnu.so -fopenmp +[2026-04-10 14:22:49,485][absl][INFO] - MUJOCO_GL=osmesa, attempting to import specified OpenGL backend. +[2026-04-10 14:22:49,495][absl][INFO] - MuJoCo library version is: 2.3.7 diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_142250-821w1msn/files/config.yaml b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_142250-821w1msn/files/config.yaml new file mode 100644 index 0000000000000000000000000000000000000000..0e4b41171ecec1defa457b174a12cc741d0d6250 --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_142250-821w1msn/files/config.yaml @@ -0,0 +1,271 @@ +_target_: + value: llmbc.workspace.train_llmbc_lowdim_workspace.TrainLLMBCLowdimWorkspace +_wandb: + value: + cli_version: 0.18.6 + m: [] + python_version: 3.9.25 + t: + "1": + - 1 + - 2 + - 3 + - 5 + - 11 + - 12 + - 41 + - 49 + - 50 + - 51 + - 53 + - 55 + - 71 + - 83 + - 95 + - 98 + - 100 + - 105 + "2": + - 1 + - 2 + - 3 + - 5 + - 11 + - 12 + - 41 + - 49 + - 50 + - 51 + - 53 + - 55 + - 71 + - 83 + - 95 + - 98 + - 100 + - 105 + "3": + - 13 + - 15 + - 16 + - 23 + - 55 + - 61 + "4": 3.9.25 + "5": 0.18.6 + "6": 4.47.1 + "8": + - 5 + "12": 0.18.6 + "13": linux-x86_64 +action_dim: + value: 30 +checkpoint: + value: + save_last_ckpt: true + save_last_snapshot: false + topk: + format_str: epoch={epoch:04d}-test_success_rate={test_success_rate:.3f}.ckpt + k: 5 + mode: max + monitor_key: test_success_rate +dataloader: + value: + batch_size: 32 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true +exp_name: + value: default +horizon: + value: 1 +llm: + value: + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + finetune_mode: orig + hydra: + job: + override_dirname: HuggingFaceTB/SmolLM2-135M-Instruct + run: + dir: data/outputs/2026.04.10/14.21.55_HuggingFaceTB/SmolLM2-135M-Instruct + llm_mode: ete-finetuned + lora_config: + bias: none + lora_alpha: 64 + lora_dropout: 0.05 + r: 32 + task_type: CAUSAL_LM + max_length: 100 + model_name: SmolLM2-135M-Instruct + name: HuggingFaceTB/SmolLM2-135M-Instruct + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + use_joint_mlp_projector: true + use_quantization: false +llm_do_sample: + value: false +llm_orig_expert_feedback: + value: true +logging: + value: + group: null + id: null + mode: online + name: 2026.04.10-14.21.56_train_llmbc_lowdim_adroit-hand-relocate-v1 + project: adroit-hand-relocate-v1-training-mix + resume: true + tags: + - train_llmbc_lowdim + - adroit-hand-relocate-v1 + - default +model_name: + value: HuggingFaceTB/SmolLM2-135M-Instruct +multi_run: + value: + run_dir: data/outputs/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1 + wandb_name_base: 2026.04.10-14.21.56_train_llmbc_lowdim_adroit-hand-relocate-v1 +n_action_steps: + value: 1 +n_latency_steps: + value: 0 +n_obs_steps: + value: 1 +name: + value: train_llmbc_lowdim +obs_dim: + value: 39 +optimizer: + value: + _target_: torch.optim.AdamW + betas: + - 0.95 + - 0.999 + eps: 1e-08 + lr: 0.001 + weight_decay: 1e-06 +output_dir: + value: /workspace/LLM-BC/data/outputs/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1 +past_action_visible: + value: false +policy: + value: + _target_: llmbc.policy.llmbc_lowdim_policy.LLMBCLowdimPolicy + action_dim: 30 + horizon: 1 + llm_discriminator: + _target_: llmbc.discriminator.llm_ce_discriminator.LLMCEDiscriminator + llm_translator: + _target_: llmbc.translator.llm_translator.LLMTranslator + action_dim: 30 + cfg: + causal_lm_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaForCausalLM + checkpoint: data/outputs/2026.04.03/12.10.45_train_llm_lowdim_adroit-hand-relocate-v1/HuggingFaceTB/SmolLM2-135M-Instruct-finetuned-adroit-hand-relocate-v1/checkpoint-2340 + config_target: llmbc.model.llm.llama_lowdim_model.LowdimLlamaConfig + finetune_mode: orig + hydra: + job: + override_dirname: HuggingFaceTB/SmolLM2-135M-Instruct + run: + dir: data/outputs/2026.04.10/14.21.55_HuggingFaceTB/SmolLM2-135M-Instruct + llm_mode: ete-finetuned + lora_config: + bias: none + lora_alpha: 64 + lora_dropout: 0.05 + r: 32 + task_type: CAUSAL_LM + max_length: 100 + model_name: SmolLM2-135M-Instruct + name: HuggingFaceTB/SmolLM2-135M-Instruct + prompter: + _target_: llmbc.translator.prompter.smollm2_prompter.SmolLM2Prompter + use_joint_mlp_projector: true + use_joint_mlp_projector: true + use_quantization: false + horizon: 1 + n_action_steps: 1 + n_obs_steps: 1 + obs_dim: 39 + task_id: adroit-hand-relocate-v1 + loss_bc_weight: 1 + loss_llm_weight: 0.01 + model: + _target_: llmbc.model.policy.policy_mlp.PolicyMLP + activation: relu + hidden_size: + - 256 + - 256 + input_size: 39 + n_action_steps: 1 + n_obs_steps: 1 + output_size: 30 + n_action_steps: 1 + n_obs_steps: 1 + normalize_llm_loss: true + obs_dim: 39 +task: + value: + action_dim: 30 + dataset: + _target_: llmbc.dataset.adroit_lowdim_dataset.AdroitHandLowdimDataset + data_path: datasets/adroit-hand-relocate-v1-mix(csp100).pt + data_path2: datasets/adroit-hand-relocate-v1.pt + dummy_normalizer: true + horizon: 1 + obs_eef_target: true + pad_after: 0 + pad_before: 0 + use_manual_normalizer: false + val_ratio: 0.1 + env_runner: + _target_: llmbc.env_runner.adroit_lowdim_runner.AdroitHandLowdimRunner + discount: 0.99 + env_name: llf-adroit-adroit-hand-relocate-v1 + feedback_type: + - hp + - hn + - fp + instruction_type: b + max_steps: 150 + n_action_steps: 1 + n_envs: 10 + n_obs_steps: 1 + n_test: 50 + n_train: 10 + visual: false + instructor: + _target_: llmbc.translator.instructor.adroit_instructor.adroit_hand_relocate_v1_instructor.AdroitHandRelocateV1Instructor + name: adroit-hand-relocate-v1 + obs_dim: 39 +task_name: + value: adroit-hand-relocate-v1 +training: + value: + checkpoint_every: 5 + debug: false + device: cuda:0 + grad_norm_clip: 0.5 + gradient_accumulate_every: 4 + lr_scheduler: constant + lr_warmup_steps: 10 + max_train_steps: null + max_val_steps: null + num_epochs: 1001 + resume: false + rollout_every: 5 + sample_every: 5 + sample_max_batch: 128 + seed: 42 + tqdm_interval_sec: 1 + val_every: 1 +val_dataloader: + value: + batch_size: 32 + num_workers: 0 + persistent_workers: false + pin_memory: false + shuffle: true diff --git a/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_142250-821w1msn/files/output.log b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_142250-821w1msn/files/output.log new file mode 100644 index 0000000000000000000000000000000000000000..7a489d1d5a701c82a916d4b4effa4af46cd0dc2e --- /dev/null +++ b/2026.04.10/14.21.55_train_llmbc_lowdim_adroit-hand-relocate-v1/wandb/run-20260410_142250-821w1msn/files/output.log @@ -0,0 +1,3 @@ +Eval AdroitHandLowdimRunner 1/6: 0%| | 0/150 [00:00